perm filename S85[JNK,JMC] blob
sn#801133 filedate 1985-08-05 generic text, type C, neo UTF8
COMMENT ⊗ VALID 00392 PAGES
C REC PAGE DESCRIPTION
C00001 00001
C00052 00002
C00056 00003 ∂02-Apr-85 0727 @SU-SCORE.ARPA:coraki!pratt@Navajo move to engineering
C00061 00004 ∂02-Apr-85 0818 BERGMAN@SU-SCORE.ARPA RA appointments
C00062 00005 ∂02-Apr-85 0858 EMMA@SU-CSLI.ARPA talk by albert meyer (mit)
C00065 00006 ∂02-Apr-85 1153 @SU-SCORE.ARPA:cheriton@Pescadero Re: move to engineering
C00067 00007 ∂02-Apr-85 1620 BLOCK@SU-CSLI.ARPA
C00068 00008 ∂02-Apr-85 1749 @SU-SCORE.ARPA:WELCH@Ames-VMSB SIGBIG
C00072 00009 ∂02-Apr-85 1830 ullman@diablo meeting
C00074 00010 ∂02-Apr-85 2021 BRESNAN@SU-CSLI.ARPA Linguistics Drinks Resume Fridays
C00076 00011 ∂03-Apr-85 0009 @SU-SCORE.ARPA:KARP@SUMEX-AIM.ARPA CSD TGIF reminder
C00078 00012 ∂03-Apr-85 0034 @SU-SCORE.ARPA:ARK*@SU-AI.ARPA CS Colloq, April 9: Cynthia Dwork, On the Sequential Nature of Unification
C00081 00013 ∂03-Apr-85 0846 PIERCE@SU-SCORE.ARPA My name
C00082 00014 ∂03-Apr-85 1217 LIBRARY@SU-SCORE.ARPA Journal of Automated Reasoning--And Other New Journals In The Math/CS Library
C00084 00015 ∂03-Apr-85 1436 EMMA@SU-CSLI.ARPA Linguistics Drinks Resume Fridays
C00086 00016 ∂03-Apr-85 1519 LIBRARY@SU-SCORE.ARPA 1985 Chapel Hill Conference on VLSI
C00088 00017 ∂03-Apr-85 1556 LIBRARY@SU-SCORE.ARPA SOCRATES--Searching Techniques
C00091 00018 ∂03-Apr-85 1605 chertok%ucbcogsci@Berkeley UCB Cognitive Science Seminar--April 9
C00095 00019 ∂03-Apr-85 1713 AUGARTEN@SU-SCORE.ARPA CS 100: History of Computers
C00097 00020 ∂03-Apr-85 1718 EMMA@SU-CSLI.ARPA Newsletter Apr. 4, No. 23
C00111 00021 ∂04-Apr-85 1104 CLAY@SU-CSLI.ARPA System Security
C00112 00022 ∂04-Apr-85 1132 MACKINLAY@SUMEX-AIM.ARPA SILVER is back on the net
C00113 00023 ∂04-Apr-85 1202 @SU-CSLI.ARPA:sugai.pa@Xerox.ARPA RRR Meeting Reminder
C00116 00024 ∂04-Apr-85 1334 NILSSON@SU-SCORE.ARPA Tuesday Lunch
C00117 00025 ∂04-Apr-85 1711 BSCOTT@SU-SCORE.ARPA Down Payment Assistance Program
C00120 00026 ∂05-Apr-85 1016 @SU-CSLI.ARPA:PCOHEN@SRI-AI.ARPA Tues, the 9th
C00121 00027 ∂05-Apr-85 1025 @SU-CSLI.ARPA:PCOHEN@SRI-AI.ARPA correction
C00122 00028 ∂05-Apr-85 1308 LIBRARY@SU-SCORE.ARPA Math/CS Library New Books
C00124 00029 ∂05-Apr-85 1349 Jskud@diablo Re: another paper
C00125 00030 ∂05-Apr-85 1401 @SU-CSLI.ARPA:brown@Kestrel Re: Linguistics Drinks Resume Fridays
C00127 00031 ∂05-Apr-85 1508 BLOCK@SU-CSLI.ARPA Linguistics for non-linguists
C00129 00032 ∂05-Apr-85 1541 BSCOTT@SU-SCORE.ARPA Closing today at 4:00
C00130 00033 ∂05-Apr-85 1553 EMMA@SU-CSLI.ARPA Linguistics for non-linguists
C00132 00034 ∂06-Apr-85 @SU-SCORE.ARPA:udi@wisc-rsch.arpa 06-Apr-85 JMC Toronto Day at Wesleyan
C00137 00035 ∂07-Apr-85 2252 @SU-SCORE.ARPA:udi@wisc-rsch.arpa Conference Schedule
C00154 00036 ∂08-Apr-85 0048 RESTIVO@SU-SCORE.ARPA PROLOG Digest V3 #16
C00183 00037 ∂08-Apr-85 0914 EMMA@SU-CSLI.ARPA Mailing lists
C00185 00038 ∂08-Apr-85 1019 @SU-CSLI.ARPA:BrianSmith.pa@Xerox.ARPA RRR Meeting, 4/9
C00186 00039 ∂08-Apr-85 1100 BRODER@SU-SCORE.ARPA First spring AFLB
C00191 00040 ∂08-Apr-85 1133 @SU-CSLI.ARPA:PENTLAND@SRI-AI.ARPA Talk by Fanya Montalvo
C00194 00041 ∂08-Apr-85 1724 REGES@SU-SCORE.ARPA Questions and Answers about SU-SUSHI
C00200 00042 ∂09-Apr-85 0029 @SU-SCORE.ARPA:FEIGENBAUM@SUMEX-AIM.ARPA Bob Engelmore returns
C00202 00043 ∂09-Apr-85 0044 RESTIVO@SU-SCORE.ARPA PROLOG Digest V3 #17
C00214 00044 ∂09-Apr-85 1010 JOHN@SU-CSLI.ARPA talk on tacit knowledge by Martin Davies
C00215 00045 ∂09-Apr-85 1112 FORSTER@SU-CSLI.ARPA Phil. Dept. Colloquium
C00216 00046 ∂09-Apr-85 1208 @SU-CSLI.ARPA:PCOHEN@SRI-AI.ARPA reminder
C00217 00047 ∂09-Apr-85 1352 WRIGHT@SUMEX-AIM.ARPA siglunch
C00218 00048 ∂09-Apr-85 1526 TULEY@SU-SCORE.ARPA room reservations
C00219 00049 ∂09-Apr-85 1715 CLT Seminar in Logic and Foundations of Mathematics
C00221 00050 ∂10-Apr-85 0022 @SU-SCORE.ARPA:ARK*@SU-AI.ARPA CS Colloq, April 16: Vaughan Pratt, Graphics for User Interfaces
C00223 00051 ∂10-Apr-85 0955 ullman@diablo meeting
C00224 00052 ∂10-Apr-85 1632 chertok%ucbcogsci@Berkeley UCB Cognitive Science Seminar--April 16
C00229 00053 ∂10-Apr-85 1740 EMMA@SU-CSLI.ARPA Newsletter Apr. 11, No. 24
C00249 00054 ∂10-Apr-85 1757 EMMA@SU-CSLI.ARPA Newsletter
C00251 00055 ∂10-Apr-85 2039 ullman@diablo Week of 4/22
C00252 00056 ∂10-Apr-85 2246 @SU-SCORE.ARPA:udi@wisc-rsch.arpa Preliminary Program for PODC 85
C00260 00057 ∂10-Apr-85 2258 @SU-SCORE.ARPA:udi@wisc-rsch.arpa STOC directions
C00267 00058 ∂11-Apr-85 0924 EMMA@SU-CSLI.ARPA Colloquium Cancelled
C00268 00059 ∂11-Apr-85 1335 JOHN@SU-CSLI.ARPA tgif
C00269 00060 ∂11-Apr-85 1356 HEDGES@SU-SCORE.ARPA Visit of Steven Zucker
C00271 00061 ∂11-Apr-85 1411 HEDGES@SU-SCORE.ARPA Visit of Steven Zucker
C00273 00062 ∂11-Apr-85 1453 PIERCE@SU-SCORE.ARPA Quad Dedication
C00275 00063 ∂11-Apr-85 1507 PIERCE@SU-SCORE.ARPA Quad Dedication
C00276 00064 ∂11-Apr-85 1639 TOM@SU-SCORE.ARPA Re: Quad Dedication
C00277 00065 ∂11-Apr-85 1926 @SU-SCORE.ARPA:reid@Glacier Quad dedication, tea dancing, and spouses
C00279 00066 ∂11-Apr-85 2314 @SU-SCORE.ARPA:WIEDERHOLD@SUMEX-AIM.ARPA Re: Quad Dedication
C00281 00067 ∂12-Apr-85 0056 RESTIVO@SU-SCORE.ARPA PROLOG Digest V3 #18
C00309 00068 ∂12-Apr-85 0837 JOHN@SU-CSLI.ARPA Lunch at Ventura
C00313 00069 ∂12-Apr-85 0948 MARJORIE@SU-CSLI.ARPA system shutdown
C00314 00070 ∂12-Apr-85 1246 @SU-CSLI.ARPA:BrianSmith.pa@Xerox.ARPA No meeting 4/16 see you 4/23
C00316 00071 ∂12-Apr-85 1427 @SU-SCORE.ARPA:WASHINGTON@SUMEX-AIM.ARPA POTLUCK
C00319 00072 ∂12-Apr-85 1508 EMMA@SU-CSLI.ARPA tgif
C00321 00073 ∂13-Apr-85 1334 @SU-SCORE.ARPA:udi@wisc-rsch.arpa Conference on Theoretical Aspects of Reasoning About Knowledge
C00327 00074 ∂14-Apr-85 1127 BARWISE@SU-CSLI.ARPA Job needed for college student
C00329 00075 ∂14-Apr-85 2030 WASOW@SU-CSLI.ARPA Benefit Concert
C00332 00076 ∂15-Apr-85 0810 EMMA@SU-CSLI.ARPA Reminder: Meyer talk Mon, 4:15, @SRI
C00334 00077 ∂15-Apr-85 0820 EMMA@SU-CSLI.ARPA Conference on Theoretical Aspects of Reasoning About Knowledge
C00340 00078 ∂15-Apr-85 0910 LIBRARY@SU-SCORE.ARPA Math/CS Library New Books
C00342 00079 ∂15-Apr-85 0948 JOHN@SU-CSLI.ARPA reminder
C00350 00080 ∂15-Apr-85 1003 EMMA@SU-CSLI.ARPA [Janet Fodor <FODOR@SU-CSLI.ARPA>: Soc. for Phil & Psych meeting]
C00352 00081 ∂15-Apr-85 1012 JOHN@SU-CSLI.ARPA lunch
C00353 00082 ∂15-Apr-85 1121 HEDGES@SU-SCORE.ARPA Faculty Lunch
C00354 00083 ∂15-Apr-85 1135 PHILOSOPHY@SU-CSLI.ARPA Tanner Lectures
C00355 00084 ∂15-Apr-85 1339 BRODER@SU-SCORE.ARPA Speaker needed
C00356 00085 ∂15-Apr-85 1358 @SU-SCORE.ARPA:udi@wisc-rsch.arpa Reminder
C00361 00086 ∂15-Apr-85 1539 WASOW@SU-CSLI.ARPA summer RAships
C00362 00087 ∂15-Apr-85 1608 WRIGHT@SUMEX-AIM.ARPA Siglunch Friday, April 19
C00365 00088 ∂15-Apr-85 2011 BRODER@SU-SCORE.ARPA Next AFLB talk
C00370 00089 ∂15-Apr-85 2154 BLOCK@SU-CSLI.ARPA SYNTAX PANEL
C00372 00090 ∂16-Apr-85 0800 JOHN@SU-CSLI.ARPA [John Perry <JOHN@SU-CSLI.ARPA>: talk on tacit knowledge by Martin Davies]
C00374 00091 ∂16-Apr-85 0926 BARWISE@SU-CSLI.ARPA Warning
C00375 00092 ∂16-Apr-85 1053 LIBRARY@SU-SCORE.ARPA SOCRATES: Searching Techniques RE: Hutchinson's message concerning premature abandonment of the card catalog
C00379 00093 ∂16-Apr-85 1356 LIBRARY@SU-SCORE.ARPA SOCRATES: Searching Techniques--the AT index in the Command Mode
C00382 00094 ∂16-Apr-85 1436 HEDGES@SU-SCORE.ARPA Zucker Abstract
C00385 00095 ∂16-Apr-85 1443 ULLMAN@SU-SCORE.ARPA Summer Workshop
C00386 00096 ∂16-Apr-85 1520 YAMARONE@SU-CSLI.ARPA SANDWICHES AT CSLI
C00389 00097 ∂16-Apr-85 1706 ullman@diablo next meeting
C00391 00098 ∂16-Apr-85 2348 POLLARD@SU-CSLI.ARPA lost receipts
C00392 00099 ∂17-Apr-85 0740 RESTIVO@SU-SCORE.ARPA PROLOG Digest V3 #19
C00407 00100 ∂17-Apr-85 0807 EMMA@SU-CSLI.ARPA Program:Meeting of Society for Philosophy and Psychology
C00437 00101 ∂17-Apr-85 1540 TAJNAI@SU-SCORE.ARPA Fortune 500/Bay Area Companies
C00440 00102 ∂17-Apr-85 1550 ullman@diablo
C00446 00103 ∂17-Apr-85 1555 LIBRARY@SU-SCORE.ARPA Mathematical People--New Book in the Math/CS Library
C00448 00104 ∂17-Apr-85 1556 LIBRARY@SU-SCORE.ARPA New Books in the Math/CS Library
C00450 00105 ∂17-Apr-85 1622 VARDI@SU-CSLI.ARPA Comments
C00455 00106 ∂17-Apr-85 1700 @SU-SCORE.ARPA:ARK@SU-AI.ARPA CS Colloq, April 23: Steven Zucker, From Contours to Waterfalls: Converging Constraints in Early Vision
C00459 00107 ∂17-Apr-85 1751 @SU-SCORE.ARPA:udi@wisc-rsch.arpa
C00465 00108 ∂17-Apr-85 1800 EMMA@SU-CSLI.ARPA Newsletter Apr. 18, No. 25
C00480 00109 ∂17-Apr-85 1946 JF@SU-SCORE.ARPA Coffee and Other Drugs
C00485 00110 ∂17-Apr-85 2347 @SU-SCORE.ARPA:SCHOEN@SUMEX-AIM.ARPA Re: Coffee and Other Drugs
C00487 00111 ∂18-Apr-85 0944 @SU-SCORE.ARPA:WASHINGTON@SUMEX-AIM.ARPA POTLUCK (last chance!)
C00490 00112 ∂18-Apr-85 1032 @SU-SCORE.ARPA:ullman@diablo CIS request for proposals
C00491 00113 ∂18-Apr-85 1207 BRAD@SU-CSLI.ARPA Odd Thursday lecture cancelled
C00492 00114 ∂18-Apr-85 1233 @SU-SCORE.ARPA:GELMAN@SUMEX-AIM.ARPA Coffee / N.J.Turnpike
C00495 00115 ∂18-Apr-85 1237 YAMARONE@SU-CSLI.ARPA SANDWICH UPDATE
C00498 00116 ∂18-Apr-85 1452 chertok%ucbcogsci@Berkeley UCB Cognitive Science Seminar--April 23
C00504 00117 ∂18-Apr-85 1458 MARJORIE@SU-CSLI.ARPA Consultant schedule
C00505 00118 ∂18-Apr-85 1506 FRANK@SU-CSLI.ARPA Re: Consultant schedule
C00506 00119 ∂18-Apr-85 1515 LIBRARY@SU-SCORE.ARPA SOCRATES: Searching Techniques- Browse is available in the Lookup Mode
C00508 00120 ∂18-Apr-85 1555 @SU-SCORE.ARPA:WASHINGTON@SUMEX-AIM.ARPA POTLUCK (you blew it!)
C00510 00121 ∂18-Apr-85 1702 chertok%ucbcogsci@Berkeley Fodor talk at Stanford
C00513 00122 ∂18-Apr-85 1712 KEMMER@SU-CSLI.ARPA fodor talk
C00515 00123 ∂18-Apr-85 2231 @SU-SCORE.ARPA:WIEDERHOLD@SUMEX-AIM.ARPA Re: Coffee / N.J.Turnpike
C00517 00124 ∂18-Apr-85 2305 @SU-SCORE.ARPA:ARK@SU-AI.ARPA Coffee
C00519 00125 ∂19-Apr-85 0013 @SU-SCORE.ARPA:reid@Glacier Re: POTLUCK (you blew it!)
C00521 00126 ∂19-Apr-85 0959 JF@SU-SCORE.ARPA Moderate Length Report on Coffee Crusade
C00524 00127 ∂19-Apr-85 1016 HEDGES@SU-SCORE.ARPA Tuesday Lunch
C00525 00128 ∂19-Apr-85 1059 ABRAIDO@SU-SCORE.ARPA Coffee
C00526 00129 ∂19-Apr-85 1248 @SU-SCORE.ARPA:vardi@diablo Roomate for STOC
C00527 00130 ∂19-Apr-85 1300 EMMA@SU-CSLI.ARPA fodor talk
C00530 00131 ∂19-Apr-85 1418 HEDGES@SU-SCORE.ARPA Zucker Visit
C00532 00132 ∂19-Apr-85 1616 @SU-SCORE.ARPA:fagin.sjrlvm1%ibm-sj.csnet@csnet-relay.arpa Super-saver to STOC
C00534 00133 ∂22-Apr-85 1433 JUTTA@SU-SCORE.ARPA Alliant Computer Systems presentation
C00535 00134 ∂22-Apr-85 1853 @SU-CSLI.ARPA:PCOHEN@SRI-AI.ARPA Next meeting
C00536 00135 ∂22-Apr-85 2325 BRODER@SU-SCORE.ARPA Next AFLB talk
C00540 00136 ∂23-Apr-85 1043 STUCKY@SU-CSLI.ARPA A Reminder
C00541 00137 ∂23-Apr-85 1123 @SU-CSLI.ARPA:BrianSmith.pa@Xerox.ARPA RRR: Fernando Pereira Today
C00543 00138 ∂23-Apr-85 1149 BARWISE@SU-CSLI.ARPA CSLI Summer Meeting
C00545 00139 ∂23-Apr-85 1326 WRIGHT@SUMEX-AIM.ARPA SIGLUNCH Friday, April 26
C00548 00140 ∂23-Apr-85 2200 @SU-SCORE.ARPA:udi@wisc-rsch.arpa Toronto Day
C00553 00141 ∂23-Apr-85 2218 ullman@diablo meeting
C00554 00142 ∂24-Apr-85 0858 CLAY@SU-CSLI.ARPA South Asian fonts
C00556 00143 ∂24-Apr-85 1308 YAMARONE@SU-CSLI.ARPA HUNGRY YET? SANDWICH?
C00558 00144 ∂24-Apr-85 1736 chertok%ucbcogsci@Berkeley UCB Cognitive Science Seminar--April 30
C00563 00145 ∂24-Apr-85 1742 EMMA@SU-CSLI.ARPA Newsletter Apr. 25, No. 26
C00577 00146 ∂25-Apr-85 0942 EMMA@SU-CSLI.ARPA logic seminar
C00579 00147 ∂25-Apr-85 0946 CLT Seminar in Logic and Foundations of Mathematics
C00580 00148 ∂25-Apr-85 1048 avg@diablo Implementation follow-up
C00581 00149 ∂25-Apr-85 1441 HEDGES@SU-SCORE.ARPA Sr. Faculty Meeting
C00583 00150 ∂25-Apr-85 1703 ullman@diablo CPROLOG
C00588 00151 ∂25-Apr-85 1704 ullman@diablo points of agreement
C00590 00152 ∂26-Apr-85 0830 @SU-SCORE.ARPA:udi@wisc-rsch.arpa Are you looking for a room mate for the STOC???
C00592 00153 ∂26-Apr-85 0946 EMMA@SU-CSLI.ARPA Psych. Dept. Friday Cognitive Seminar
C00596 00154 ∂28-Apr-85 0504 @SU-SCORE.ARPA:udi@wisc-rsch.arpa Are you looking for a room mate for the STOC???
C00598 00155 ∂28-Apr-85 1807 ullman@diablo NAIL! Prototype
C00604 00156 ∂28-Apr-85 1817 ullman@diablo A Thought on System Outline
C00607 00157 ∂28-Apr-85 1821 ullman@diablo Next meeting
C00609 00158 ∂28-Apr-85 2139 cheriton@Pescadero Re: NAIL! Prototype
C00611 00159 ∂29-Apr-85 0924 @SU-CSLI.ARPA:BrianSmith.pa@Xerox.ARPA RRR: John Perry Tomorrow
C00612 00160 ∂29-Apr-85 1057 roy@diablo Re: system outline.
C00617 00161 ∂29-Apr-85 1116 CLAY@SU-CSLI.ARPA Turing use today
C00619 00162 ∂29-Apr-85 1609 ullman@diablo MRS
C00622 00163 ∂29-Apr-85 1743 INGRID@SU-CSLI.ARPA Meeting tomorrow
C00623 00164 ∂29-Apr-85 2355 @SU-CSLI.ARPA:PCOHEN@SRI-AI.ARPA Next meeting of NL4 (Language and Action)
C00625 00165 ∂29-Apr-85 2355 @SU-CSLI.ARPA:PCOHEN@SRI-AI.ARPA Next meeting of NL4 (Language and Action)
C00627 00166 ∂29-Apr-85 2355 @SU-CSLI.ARPA:PCOHEN@SRI-AI.ARPA Next meeting of NL4 (Language and Action)
C00629 00167 ∂30-Apr-85 0038 BRESNAN@SU-CSLI.ARPA housekeeper
C00630 00168 ∂30-Apr-85 0920 @SU-CSLI.ARPA:PCOHEN@SRI-AI.ARPA Time for NL4 meeting
C00631 00169 ∂30-Apr-85 0920 @SU-CSLI.ARPA:PCOHEN@SRI-AI.ARPA Time for NL4 meeting
C00632 00170 ∂30-Apr-85 0920 @SU-CSLI.ARPA:PCOHEN@SRI-AI.ARPA Time for NL4 meeting
C00633 00171 ∂30-Apr-85 1221 @SU-SCORE.ARPA:JF@SU-SUSHI.ARPA may 17 bats
C00636 00172 ∂30-Apr-85 1225 PAPA@SU-SCORE.ARPA STOC forms
C00637 00173 ∂30-Apr-85 1413 WRIGHT@SUMEX-AIM.ARPA SIGLUNCH Friday, May 3
C00640 00174 ∂30-Apr-85 1520 @SU-SCORE.ARPA:JF@SU-SUSHI.ARPA thanks
C00641 00175 ∂30-Apr-85 1533 @SU-SCORE.ARPA:WELCH@Ames-VMSB SIGBIG
C00645 00176 ∂30-Apr-85 1637 ullman@diablo meetings
C00646 00177 ∂30-Apr-85 2010 JOHN@SU-CSLI.ARPA Housing Wanted
C00647 00178 ∂01-May-85 0043 RESTIVO@SU-SCORE.ARPA PROLOG Digest V3 #20
C00662 00179 ∂01-May-85 0852 EMMA@SU-CSLI.ARPA Newsletter
C00663 00180 ∂01-May-85 0858 EMMA@SU-CSLI.ARPA Tuesday teas
C00664 00181 ∂01-May-85 1647 EMMA@SU-CSLI.ARPA Newsletter May 2, No. 27
C00677 00182 ∂01-May-85 1759 BARWISE@SU-CSLI.ARPA The fallacy of the homomorphism of content
C00679 00183 ∂02-May-85 0906 HANRAHAN@SU-SCORE.ARPA SENIOR-FACULTY MEETING
C00680 00184 ∂02-May-85 1126 BLOCK@SU-CSLI.ARPA homomorphism of content
C00682 00185 ∂02-May-85 1138 @SU-CSLI.ARPA:ISRAEL@SRI-AI.ARPA Re: The fallacy of the homomorphism of content
C00694 00186 ∂02-May-85 1254 YAMARONE@SU-CSLI.ARPA EXTRA SANDWICHES
C00696 00187 ∂02-May-85 2104 chertok%ucbcogsci@Berkeley Stanford Linguistics Colloquium
C00701 00188 ∂03-May-85 1215 EMMA@SU-CSLI.ARPA ling colloq
C00705 00189 ∂03-May-85 1648 avg@diablo Wumpus and NAIL
C00725 00190 ∂03-May-85 1754 ullman@diablo Re: Wumpus and NAIL
C00728 00191 ∂03-May-85 2021 @SU-CSLI.ARPA:PCOHEN@SRI-AI.ARPA
C00730 00192 ∂03-May-85 2021 @SU-CSLI.ARPA:PCOHEN@SRI-AI.ARPA
C00732 00193 ∂03-May-85 2021 @SU-CSLI.ARPA:PCOHEN@SRI-AI.ARPA
C00734 00194 ∂03-May-85 2021 TREITEL@SUMEX-AIM.ARPA Re: Wumpus and NAIL
C00736 00195 ∂04-May-85 0219 @SU-SCORE.ARPA:JF@SU-SUSHI.ARPA One more BATS abstract
C00737 00196 ∂04-May-85 1004 @SU-SCORE.ARPA:udi@wisc-rsch.arpa crusade (long message)
C00758 00197 ∂04-May-85 1023 MMACKEN@SU-CSLI.ARPA grammars & learning
C00759 00198 ∂05-May-85 1104 BARWISE@SU-CSLI.ARPA Messages to Cooper v Engdahl
C00760 00199 ∂06-May-85 0037 RESTIVO@SU-SCORE.ARPA PROLOG Digest V3 #21
C00771 00200 ∂06-May-85 0854 LIBRARY@SU-SCORE.ARPA The Way of Mathematics and Mathematicians a Historical-Philosophical Study
C00773 00201 ∂06-May-85 0905 LIBRARY@SU-SCORE.ARPA New Books in the Math/CS Library
C00775 00202 ∂06-May-85 0923 GOLUB@SU-SCORE.ARPA Praise for Andy Yao
C00776 00203 ∂06-May-85 0941 @SU-SCORE.ARPA:JJW@SU-AI.ARPA Terminals available
C00777 00204 ∂06-May-85 0952 CHEADLE@SU-SCORE.ARPA Ph.D. admissions results
C00786 00205 ∂06-May-85 1004 ISRAEL@SU-CSLI.ARPA Tuesday mtg.
C00787 00206 ∂06-May-85 1034 @SU-SCORE.ARPA:MDIXON@SU-SUSHI.ARPA Spring Picnic: May 31
C00789 00207 ∂06-May-85 1100 avg@diablo Re: Wumpus and NAIL
C00790 00208 ∂06-May-85 1103 avg@diablo Re: Wumpus files
C00791 00209 ∂06-May-85 1210 NILSSON@SU-SCORE.ARPA Tuesday Lunch
C00793 00210 ∂06-May-85 1458 avg@diablo Same Generation and NC
C00797 00211 ∂06-May-85 1540 FISCHER@SU-CSLI.ARPA Chronicles
C00798 00212 ∂06-May-85 2114 @SU-CSLI.ARPA:PCOHEN@SRI-AI.ARPA Last reminder
C00799 00213 ∂06-May-85 2206 ullman@diablo next meeting
C00801 00214 ∂07-May-85 0055 RESTIVO@SU-SCORE.ARPA PROLOG Digest V3 #22
C00837 00215 ∂07-May-85 1143 BLOCK@SU-CSLI.ARPA benevolence
C00842 00216 ∂07-May-85 1623 HEDGES@SU-SCORE.ARPA ACM Doctoral Dissertation Award Program
C00843 00217 ∂07-May-85 1718 PIERCE@SU-SCORE.ARPA Quad Dedication
C00845 00218 ∂07-May-85 1748 JODY@SU-CSLI.ARPA The Burrito Bandit
C00847 00219 ∂07-May-85 1803 CLT Seminar in Logic and Foundations of Mathematics
C00848 00220 ∂07-May-85 1943 JOHN@SU-CSLI.ARPA Good courses at Berkeley
C00852 00221 ∂08-May-85 0513 allegra!seismo!mcvax!ukc!snow!req@diablo cprolog speed
C00854 00222 ∂08-May-85 0728 @SU-SCORE.ARPA:ARK@SU-AI.ARPA CS Colloq, May 14: Simon Lam, Observing Protocol Systems with Low-Resolution Snapshots
C00857 00223 ∂08-May-85 0825 YAMARONE@SU-CSLI.ARPA [Ingrid Deiwiks - 497-3084 <INGRID@SU-CSLI.ARPA>: Sandwiches]
C00860 00224 ∂08-May-85 1047 JOHN@SU-CSLI.ARPA next week's session
C00862 00225 ∂08-May-85 1136 WRIGHT@SUMEX-AIM.ARPA SIGlunch: April 10
C00863 00226 ∂08-May-85 1321 HEDGES@SU-SCORE.ARPA Robotics Candidate
C00864 00227 ∂08-May-85 1409 WRIGHT@SUMEX-AIM.ARPA SIGlunch ooops
C00865 00228 ∂08-May-85 1504 PETERS@SU-CSLI.ARPA Hausser Talk and Demo
C00867 00229 ∂08-May-85 1519 @SU-CSLI.ARPA:BMOORE@SRI-AI.ARPA Your comments
C00871 00230 ∂08-May-85 1540 LIBRARY@SU-SCORE.ARPA SOCRATES: Searching Techniques--The Browse Command An Update
C00876 00231 ∂08-May-85 1748 EMMA@SU-CSLI.ARPA Newsletter May 9, No. 28
C00888 00232 ∂08-May-85 2236 GOLUB@SU-SCORE.ARPA housing wanted
C00890 00233 ∂09-May-85 0939 NILSSON@SU-SCORE.ARPA Research Support for Entering PhD Students
C00894 00234 ∂09-May-85 1036 @SU-SCORE.ARPA:ullman@diablo Re: Research Support for Entering PhD Students
C00895 00235 ∂09-May-85 1048 JOHN@SU-CSLI.ARPA [John Perry <JOHN@SU-CSLI.ARPA>: next week's session]
C00897 00236 ∂09-May-85 1142 @SU-CSLI.ARPA:PCOHEN@SRI-AI.ARPA NL4 Talk Next Tues
C00898 00237 ∂09-May-85 1152 PATASHNIK@SU-SCORE.ARPA aflb today
C00899 00238 ∂09-May-85 1400 BRODER@SU-SCORE.ARPA Late abstracts
C00903 00239 ∂09-May-85 1837 @SU-SCORE.ARPA:JF@SU-SUSHI.ARPA One more BATS abstract for 5/17 meeting
C00907 00240 ∂09-May-85 2231 @SU-SCORE.ARPA:coraki!pratt@Navajo Re: Research Support for Entering PhD Students
C00909 00241 ∂09-May-85 2348 @SU-SCORE.ARPA:udi@wisc-rsch.arpa ACM mailing
C00911 00242 ∂10-May-85 0911 YAMARONE@SU-CSLI.ARPA SANDWICH UPDATE
C00913 00243 ∂10-May-85 0958 TAJNAI@SU-SCORE.ARPA Going to Sweden?
C00914 00244 ∂10-May-85 1017 MARJORIE@SU-CSLI.ARPA Thefts
C00915 00245 ∂10-May-85 1029 @SU-SCORE.ARPA:cheriton@Pescadero Re: Research Support for Entering PhD Students
C00917 00246 ∂10-May-85 1220 @SU-CSLI.ARPA:halvorsen.pa@Xerox.ARPA Reminder: Today at 2:30 pm: Carlota S. Smith: New Aspects of
C00919 00247 ∂10-May-85 1315 PAPA@SU-SCORE.ARPA Re: Same Generation and NC
C00920 00248 ∂10-May-85 1322 @SU-SCORE.ARPA:udi@wisc-rsch.arpa Referee reports
C00922 00249 ∂10-May-85 1343 ullman@diablo Re: Same Generation and NC
C00925 00250 ∂10-May-85 1617 NILSSON@SU-SCORE.ARPA Prize
C00929 00251 ∂11-May-85 2024 SARAIYA@SUMEX-AIM.ARPA Release 6 Systems
C00930 00252 ∂11-May-85 2155 ullman@diablo Hot Off the Wire
C00944 00253 ∂12-May-85 1440 BRODER@SU-SCORE.ARPA No AFLB this week
C00945 00254 ∂13-May-85 0832 EMMA@SU-CSLI.ARPA talk by shankar, weds, 4:15, in EJ232, on Mechanical Proofs in Metamathematics
C00948 00255 ∂13-May-85 0844 HEDGES@SU-SCORE.ARPA Faculty Lunch
C00949 00256 ∂13-May-85 0858 HEDGES@SU-SCORE.ARPA Visit of Peter Will
C00950 00257 ∂13-May-85 0949 NILSSON@SU-SCORE.ARPA New CSD Curric Committee
C00955 00258 ∂13-May-85 1102 ULLMAN@SU-SCORE.ARPA Undergraduate program
C00958 00259 ∂13-May-85 1125 vardi@diablo Re: Same Generation and NC
C00959 00260 ∂13-May-85 1332 @SU-CSLI.ARPA:PCOHEN@SRI-AI.ARPA Meeting tomorrow, the 14th
C00960 00261 ∂13-May-85 1343 EMMA@SU-CSLI.ARPA This Thursday's Colloquium
C00964 00262 ∂13-May-85 1444 ullman@diablo next meetings
C00965 00263 ∂13-May-85 1631 EMMA@SU-CSLI.ARPA Hausser talk
C00967 00264 ∂13-May-85 1651 @SU-CSLI.ARPA:BrianSmith.pa@Xerox.ARPA John Perry tomorrow (May 14)
C00969 00265 ∂13-May-85 1810 vardi@diablo Nothing new under the sun
C00970 00266 ∂13-May-85 1848 @SU-SCORE.ARPA:JF@SU-SUSHI.ARPA bats schedule
C00972 00267 ∂13-May-85 1904 @SU-SCORE.ARPA:JF@SU-SUSHI.ARPA final abstract
C00976 00268 ∂13-May-85 2050 ullman@diablo Re: Nothing new under the sun
C00977 00269 ∂13-May-85 2231 DAVIES@SUMEX-AIM.ARPA Architectures meeting -- 5/15/85, 9 am
C00979 00270 ∂13-May-85 2331 SCHOEN@SUMEX-AIM.ARPA Rel 6 progress
C00981 00271 ∂14-May-85 0739 @SU-CSLI.ARPA:RCOOPER@SRI-AI.ARPA job in Sweden
C00983 00272 ∂14-May-85 0909 MACKAYE@SU-CSLI.ARPA STANFORD OUT OF SOUTH AFRICA
C00985 00273 ∂14-May-85 0947 BCM Seminar
C00988 00274 ∂14-May-85 0956 @SU-CSLI.ARPA:BCM@SU-AI.ARPA Seminar
C00992 00275 ∂14-May-85 1034 EMMA@SU-CSLI.ARPA Modems
C00994 00276 ∂14-May-85 1208 YAN@SUMEX-AIM.ARPA release 6 at site HPP
C00995 00277 ∂14-May-85 1231 YAMARONE@SU-CSLI.ARPA SANDWICH REMINDER
C00997 00278 ∂14-May-85 1248 WRIGHT@SUMEX-AIM.ARPA SIGLUNCH FRIDAY MAY 17
C01000 00279 ∂14-May-85 1712 LIBRARY@SU-SCORE.ARPA Math/CS Library New Books
C01002 00280 ∂14-May-85 1930 avg@diablo nonlinear recursion example
C01007 00281 ∂14-May-85 1956 YAN@SUMEX-AIM.ARPA Resource Management Presentations
C01010 00282 ∂14-May-85 1957 YAN@SUMEX-AIM.ARPA Abstract: Initial Placement of Agents
C01014 00283 ∂14-May-85 2042 @SU-SCORE.ARPA:JF@SU-SUSHI.ARPA bats abstracts
C01027 00284 ∂15-May-85 1230 @SU-SCORE.ARPA:JF@SU-SUSHI.ARPA Re: bats
C01030 00285 ∂15-May-85 1234 DIKRAN@SU-CSLI.ARPA Celebration
C01031 00286 ∂15-May-85 1233 ROSENBLOOM@SUMEX-AIM.ARPA Special Siglunch
C01033 00287 ∂15-May-85 1235 JODY@SU-CSLI.ARPA Burrito Bandito's visit
C01035 00288 ∂15-May-85 1251 @SU-SCORE.ARPA:ARK@SU-AI.ARPA CS Colloq, May 21: Paul Vitanyi, Distributed Match-Making in Computer Networks
C01039 00289 ∂15-May-85 1252 avg@diablo nonlinear recursion followup
C01041 00290 ∂15-May-85 1304 @SU-SCORE.ARPA:ROSENBLOOM@SUMEX-AIM.ARPA Special Siglunch
C01043 00291 ∂15-May-85 1452 ullman@diablo Re: nonlinear recursion followup
C01047 00292 ∂15-May-85 1721 EMMA@SU-CSLI.ARPA Newsletter May 16, No. 29
C01062 00293 ∂15-May-85 1730 maier%oregon-grad.csnet@csnet-relay.ARPA Prolog as a grammar
C01066 00294 ∂01-Jun-85 0120 @SU-SCORE.ARPA:ARK@SU-AI.ARPA CS Colloq, June 4: Daniel Lehmann, Modal Temporal Logics: a survey of recent results
C01069 00295 ∂16-May-86 1019 @SU-SCORE.ARPA:JF@SU-SUSHI.ARPA Talk Monday, 5/20/85 on Optimal VLSI Sorting Circuits
C01071 00296 ∂16-May-85 1855 @SU-SCORE.ARPA:JF@SU-SUSHI.ARPA BATS final reminder
C01073 00297 ∂16-May-85 1903 BLOCK@SU-CSLI.ARPA Situation Semantics
C01075 00298 ∂16-May-85 1916 BERG@SU-SCORE.ARPA Peterson's Guide
C01077 00299 ∂16-May-85 1917 NILSSON@SU-SCORE.ARPA Stanford Standings
C01079 00300 ∂17-May-85 0946 PIERCE@SU-SCORE.ARPA Visitor Space
C01081 00301 ∂17-May-85 1158 LIBRARY@SU-SCORE.ARPA New Books In The Math/CS Library
C01083 00302 ∂17-May-85 1200 EMMA@SU-CSLI.ARPA Situation Semantics
C01085 00303 ∂17-May-85 1248 @SU-SCORE.ARPA:udi@wisc-rsch.arpa A Thank You
C01087 00304 ∂17-May-85 1301 @SU-SCORE.ARPA:udi@wisc-rsch.arpa PODC 85 General Information
C01098 00305 ∂17-May-85 1348 LIBRARY@SU-SCORE.ARPA Socrates Update newsletter-Free, Request To Be Placed On Mailing List
C01100 00306 ∂17-May-85 1402 @SU-CSLI.ARPA:PCOHEN@SRI-AI.ARPA [Phil Cohen <PCOHEN@SRI-AI.ARPA>: Seminar -- Rational Interaction. Jeff Rosenschein]
C01104 00307 ∂17-May-85 1433 HEDGES@SU-SCORE.ARPA Faculty Lunch
C01105 00308 ∂18-May-85 1213 @SU-SCORE.ARPA:coraki!pratt@Navajo input on fac. lunch
C01115 00309 ∂18-May-85 1918 @SU-SCORE.ARPA:golub@Navajo AI Fellowships
C01119 00310 ∂19-May-85 1848 @SU-SCORE.ARPA:JF@SU-SUSHI.ARPA Talk Monday by A. Siegel on Optimal VLSI Sorters
C01122 00311 ∂20-May-85 0037 RESTIVO@SU-SCORE.ARPA PROLOG Digest V3 #23
C01132 00312 ∂20-May-85 0758 NILSSON@SU-SCORE.ARPA Tuesday lunch
C01134 00313 ∂20-May-85 1029 TULEY@SU-SCORE.ARPA Palo Alto Phone books
C01135 00314 ∂20-May-85 1120 BARWISE@SU-CSLI.ARPA Reminder
C01136 00315 ∂20-May-85 1126 EMMA@SU-CSLI.ARPA Celebration reminder
C01137 00316 ∂20-May-85 1138 INGRID@SU-CSLI.ARPA Word Processing
C01138 00317 ∂20-May-85 1413 MARJORIE@SU-CSLI.ARPA Macintosh software upgrade
C01139 00318 ∂20-May-85 1449 REGES@SU-SCORE.ARPA MacFest Reminder
C01141 00319 ∂20-May-85 1456 TAJNAI@SU-SCORE.ARPA Re: MacFest Reminder
C01142 00320 ∂20-May-85 1613 DAVIES@SUMEX-AIM.ARPA Architecture meeting
C01144 00321 ∂20-May-85 2104 @SUMEX-AIM.ARPA:YAN@SU-SCORE.ARPA New abstract: Dynamic load balancing
C01149 00322 ∂20-May-85 2255 @SU-SCORE.ARPA:WIEDERHOLD@SUMEX-AIM.ARPA Re: MacFest Reminder
C01151 00323 ∂20-May-85 2330 @SU-SCORE.ARPA:FEIGENBAUM@SUMEX-AIM.ARPA strategic computing project funding
C01153 00324 ∂21-May-85 0122 BRODER@SU-SCORE.ARPA Speaker needed this week
C01154 00325 ∂21-May-85 1008 WRIGHT@SUMEX-AIM.ARPA SIGLunch: Friday, May 24, 1985
C01157 00326 ∂21-May-85 1048 FORSTER@SU-CSLI.ARPA Phil. Dept. Colloquium
C01158 00327 ∂21-May-85 1220 BRODER@SU-SCORE.ARPA Forsythe Award
C01160 00328 ∂21-May-85 1349 YAN@SUMEX-AIM.ARPA 3602, 3674 and 3605
C01161 00329 ∂21-May-85 1625 BSCOTT@SU-SCORE.ARPA Velo-Bind Machine
C01163 00330 ∂21-May-85 2044 SHAHN@SUMEX-AIM.ARPA 6.0
C01164 00331 ∂21-May-85 2125 ullman@diablo meeting
C01165 00332 ∂22-May-85 0015 @SU-SCORE.ARPA:ARK@SU-AI.ARPA CS Colloq, May 28: Matthew Ginsberg, Counterfactual Implication
C01168 00333 ∂22-May-85 1042 @SU-SCORE.ARPA:M.SUSAN@SU-SIERRA.ARPA FLEX/32 presentation - today!
C01170 00334 ∂22-May-85 1356 JODY@SU-CSLI.ARPA Burrito Bandito
C01172 00335 ∂22-May-85 1542 LIBRARY@SU-SCORE.ARPA New Books In The Math/CS Library
C01175 00336 ∂22-May-85 1621 BETSY@SU-CSLI.ARPA NSF Visits
C01180 00337 ∂22-May-85 1929 EMMA@SU-CSLI.ARPA Newsletter May 23, No. 30
C01196 00338 ∂22-May-85 1931 @SU-SCORE.ARPA:berglund@Pescadero Systems Qual Proposal
C01200 00339 ∂22-May-85 1933 @SU-SCORE.ARPA:pratt@cicero Systems Qual Proposal
C01202 00340 ∂22-May-85 2028 @SU-SCORE.ARPA:ullman@diablo Re: Systems Qual Proposal
C01204 00341 ∂22-May-85 2142 @SU-SCORE.ARPA:reid@Glacier Re: Systems Qual Proposal
C01207 00342 ∂23-May-85 0020 YM@SU-AI.ARPA Special seminar
C01211 00343 ∂23-May-85 0044 RESTIVO@SU-SCORE.ARPA PROLOG Digest V3 #24
C01237 00344 ∂23-May-85 0223 BRODER@SU-SCORE.ARPA AFLB today
C01241 00345 ∂23-May-85 0825 EMMA@SU-CSLI.ARPA Special seminar
C01245 00346 ∂23-May-85 1523 SELLS@SU-CSLI.ARPA end of quarter/year party
C01247 00347 ∂23-May-85 1539 @SU-SCORE.ARPA:EENGELMORE@SUMEX-AIM.ARPA Showing of KNOWLEDGE ENGINEERING Movie - 5/28
C01251 00348 ∂23-May-85 1612 SCHOLZ@SU-SCORE.ARPA 2nd semiannual postcomp 7k
C01253 00349 ∂24-May-85 0538 RESTIVO@SU-SCORE.ARPA PROLOG Digest V3 #25
C01277 00350 ∂24-May-85 0900 ullman@diablo Intermediate Language Proposal
C01284 00351 ∂24-May-85 1125 FORD@SU-CSLI.ARPA language processing workshop
C01286 00352 ∂24-May-85 1130 PIERCE@SU-SCORE.ARPA Quad Phone
C01287 00353 ∂24-May-85 1155 EMMA@SU-CSLI.ARPA Copying machine
C01288 00354 ∂24-May-85 1202 BLOCK@SU-CSLI.ARPA Situation Semantics
C01290 00355 ∂24-May-85 1254 EMMA@SU-CSLI.ARPA Situation Semantics
C01292 00356 ∂24-May-85 1421 BRODER@SU-SCORE.ARPA Next AFLB talk
C01294 00357 ∂24-May-85 1541 TAJNAI@SU-SCORE.ARPA 83/84 Industrial Affiliates Report
C01296 00358 ∂24-May-85 1719 REGES@SU-SCORE.ARPA affiliated students
C01300 00359 ∂24-May-85 2104 @SU-SCORE.ARPA:berglund@Pescadero The Quals and My Advisor
C01302 00360 ∂25-May-85 1525 BSCOTT@SU-SCORE.ARPA Velo-Bind
C01304 00361 ∂27-May-85 1812 NILSSON@SU-SCORE.ARPA Lunch
C01306 00362 ∂28-May-85 0931 YAN@SUMEX-AIM.ARPA Abstract for this Wednesday's talk at 9:30
C01309 00363 ∂28-May-85 1012 HANRAHAN@SU-SCORE.ARPA HEAT
C01310 00364 ∂28-May-85 1354 NILSSON@SU-SCORE.ARPA meetings
C01312 00365 ∂28-May-85 1727 LIBRARY@SU-SCORE.ARPA Socrates: Automatic Renewal of Personal Accounts in August
C01314 00366 ∂28-May-85 1737 BRODER@SU-SCORE.ARPA Next AFLB talk
C01317 00367 ∂29-May-85 0015 @SU-SCORE.ARPA:ARK@SU-AI.ARPA CS Colloq, June 4: Daniel Lehmann, Mutual Exclusion
C01321 00368 ∂29-May-85 1252 ullman@diablo today's plan
C01323 00369 ∂29-May-85 1250 LIBRARY@SU-SCORE.ARPA New Books In the Math/CS Library
C01325 00370 ∂29-May-85 1256 MARJORIE@SU-CSLI.ARPA System problems
C01327 00371 ∂29-May-85 1304 WRIGHT@SUMEX-AIM.ARPA Siglunch Friday, May 31
C01328 00372 ∂29-May-85 1312 BERGMAN@SU-SCORE.ARPA Summer RA's
C01329 00373 ∂29-May-85 1359 HEDGES@SU-SCORE.ARPA General Faculty Meeting
C01330 00374 ∂29-May-85 1404 HEDGES@SU-SCORE.ARPA Sr. Faculty Meeting
C01331 00375 ∂30-May-85 0631 REGES@SU-SCORE.ARPA Black Friday Meeting
C01333 00376 ∂30-May-85 0645 BLOCK@SU-CSLI.ARPA Next Meeting
C01336 00377 ∂30-May-85 0651 EMMA@SU-CSLI.ARPA Newsletter May 30, No. 31
C01347 00378 ∂30-May-85 0854 PIERCE@SU-SCORE.ARPA Proposed Space Committee
C01349 00379 ∂30-May-85 0926 @SU-SCORE.ARPA:udi@wisc-rsch.arpa classification question
C01352 00380 ∂30-May-85 1141 PIERCE@SU-SCORE.ARPA Space
C01353 00381 ∂30-May-85 1632 PIERCE@SU-SCORE.ARPA Visitor Space
C01354 00382 ∂30-May-85 2304 @SU-SCORE.ARPA:cheriton@Pescadero Re: Proposed Space Committee
C01357 00383 ∂31-May-85 0900 NILSSON@SU-SCORE.ARPA Re: Proposed Space Committee
C01359 00384 ∂31-May-85 0939 @SU-SCORE.ARPA:coraki!pratt@Navajo Re: Proposed Space Committee
C01364 00385 ∂31-May-85 1134 NILSSON@SU-SCORE.ARPA Visiting Committee
C01367 00386 ∂31-May-85 1248 @SU-SCORE.ARPA:TW@SU-AI.ARPA Visiting committee
C01369 00387 ∂31-May-85 1259 avg@diablo NC, P-completeness, and extension joins
C01371 00388 ∂31-May-85 1344 FORD@SU-CSLI.ARPA Language Processing Workshop
C01374 00389 ∂31-May-85 1410 PIERCE@SU-SCORE.ARPA House for rent in Palo Alto
C01375 00390 ∂31-May-85 1635 BLOCK@SU-CSLI.ARPA CONNECTIONISM
C01377 00391 ∂31-May-85 1641 ullman@diablo Re: NC, P-completeness, and extension joins
C01378 00392 ∂31-May-85 1746 PARIS@MIT-XX.ARPA P
C01383 ENDMK
C⊗;
∂02-Apr-85 0727 @SU-SCORE.ARPA:coraki!pratt@Navajo move to engineering
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 2 Apr 85 07:27:29 PST
Received: from Navajo ([36.8.0.48].#Internet) by SU-SCORE.ARPA with TCP; Tue 2 Apr 85 07:21:08-PST
Received: by coraki.uucp (1.1/SMI-1.2)
id AA00966; Tue, 2 Apr 85 07:17:53 pst
Date: Tue, 2 Apr 85 07:17:53 pst
From: coraki!pratt@Navajo (Vaughan Pratt)
Message-Id: <8504021517.AA00966@coraki.uucp>
To: faculty@su-score.ARPA
Subject: move to engineering
From: John McCarthy <JMC@SU-AI.ARPA>
At Berkeley there was enormous fuss according to my
informant there.
Fuss or no fuss, they had to do it. The computer scientists were split
between two buildings quarter of a mile apart because some of them
wanted to belong to Liberal Arts and some to Engineering. As a
Berkeley graduate student (69-70) prior to the merge I found it
inconvenient to travel between the two buildings. As an occasional
visitor to Berkeley after the merge I am convinced that they are better
off. In particular, had they stayed apart the Computer Science
Department would not have been able to have Manuel Blum for chairman -
Manuel was in EECS. Interestingly, one side effect of the merge was to
bring Velvel Kahan out of the cold (of the Math. department) into the
fold (in office at least, I don't know whether he actually changed his
official affiliation). Had the merge taken place before Berkeley's
Math. department froze out Steve Cook, Steve might have transferred to
CSD and Berkeley would now have a Turing award winner.
Unless there ends up being a single
department that teaches computer science courses, the fact that there
are two departments in the same school won't make much difference.
For one thing (the one that affects me the most), what is keeping some
people out of the Computer Science Department is not that they don't
like the department but that they like the engineering school. Ever
since arriving at Stanford I have found it a royal pain to be cut off
from the systems faculty. It is great to have David Cheriton and
Keith Lantz here in Jacks, but they made this move at the cost of
cutting themselves off from the rest of the systems faculty. Jeff
Ullman, Leo Guibas, Manolis Katevenis, and I have offices in CIS,
allowing us better contact with the other systems people (the CIS-Jacks
distance is quite enough to result in almost total isolation of the two
groups), but I for one find this quite inconvenient, not to mention
wasteful, and would much prefer to be able to associate with both
systems people and other CSD people from the convenience of a single
office.
The combination of moving the department to Engineering (to attract the
engineering faculty to computer science) and to the new building (for
its space, and for its proximity to CIS and other engineering-related
groups) is almost certainly necessary to deal with this. A bigger
concern is whether these two steps suffice.
-v
∂02-Apr-85 0818 BERGMAN@SU-SCORE.ARPA RA appointments
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 2 Apr 85 08:17:52 PST
Date: Tue 2 Apr 85 08:16:08-PST
From: Sharon Bergman <BERGMAN@SU-SCORE.ARPA>
Subject: RA appointments
To: faculty@SU-SCORE.ARPA
cc: sharon@SU-SCORE.ARPA
If you will be supporting any students Spring quarter and have not notified
Lynn Hassler or myself, please let us know as soon as possible so that we
can set up their appointments. Thank you.
Sharon Bergman
-------
∂02-Apr-85 0858 EMMA@SU-CSLI.ARPA talk by albert meyer (mit)
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 2 Apr 85 08:58:13 PST
Return-Path: <WALDINGER@SRI-AI.ARPA>
Received: from SRI-AI.ARPA by SU-CSLI.ARPA with TCP; Mon 1 Apr 85 17:13:06-PST
Date: Mon 1 Apr 85 17:04:23-PST
From: WALDINGER@SRI-AI.ARPA
Subject: talk by albert meyer (mit)
To: AIC-Associates: ;,
CSL: ;, friends@SU-CSLI.ARPA, bboard@SRI-AI.ARPA, su-bboards@SU-AI.ARPA
cc: meyer@MIT-MC.ARPA
ReSent-Date: Tue 2 Apr 85 08:53:52-PST
ReSent-From: Emma Pease <Emma@SU-CSLI.ARPA>
ReSent-To: friends@SU-CSLI.ARPA
when: mon, april 15, 4:15pm
where: csl conference room (el381)
building e, sri, on ravenswood av. opposite pine street, menlo park
coffee: 3:45 pm in waldinger's office (ek292)
REASONING ABOUT BLOCK STRUCTURED VARIABLES: WHAT MAKES THE FREE-LIST FREE?
Albert R. Meyer
Abstract: Let Q be an identifier denoting a "global" (i.e., unknown) procedure
which takes a procedure parameter. In many familiar programming languages (say
Pascal or ALGOL, but not C) the block
BEGIN
INT X:=0;
PROCEDURE P();
x:=x+2;
END(*of P*);
Q(P);
IF EVEN?(x) THEN RUN←FOREVER
ELSE...FI
END
always runs forever. The reason is that the "new" variable x is initialized to
0, and the global procedure Q, which has no independent access to x, is only
given the ability to increment x by 2, so if and when Q(P) terminates, it can
only have added 2 to 0 a finite number of times, thereby leaving x even.
None of the logical systems in the literature for reasoning about procedures is
powerful enough to prove this fact. Indeed, this fact is not even true
according to the standard models of storage allocation. In this talk, we
discuss some of the theoretical difficulties in reasoning about local storage.
-------
∂02-Apr-85 1153 @SU-SCORE.ARPA:cheriton@Pescadero Re: move to engineering
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 2 Apr 85 11:53:26 PST
Received: from Pescadero by SU-SCORE.ARPA with TCP; Tue 2 Apr 85 11:11:45-PST
Date: Tue, 2 Apr 85 11:11:11 pst
From: David Cheriton <cheriton@Pescadero>
Subject: Re: move to engineering
To: coraki!pratt@Navajo, faculty@SU-Score
I agree with JMC. We need to get more for this move to make it worthwhile,
like the "living and breathing" portion of CSL.
Seems to me that Vaughan would be half satisfied if CSL and CSD shared
a building. I dont know of any evidence that suggests being in the
engineering school is an advantage from a hiring standpoint, other than
Gibbons famous 6 slots. Certainly, I am happy in H&S and much happier being
in MJH than I was in ERL, but that is for a variety of reasons.
Personally, I think there is a multiple of dangers for this department
in moving to engineering and we need to be well protected if and when
we move.
∂02-Apr-85 1620 BLOCK@SU-CSLI.ARPA
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 2 Apr 85 16:19:55 PST
Date: Tue 2 Apr 85 16:12:26-PST
From: Ned Block <BLOCK@SU-CSLI.ARPA>
To: rrr@SU-CSLI.ARPA
cc: susi@SU-CSLI.ARPA
Susi has some copies of the paper on functional role semantics on which
my Thursday talk will be based.
-------
∂02-Apr-85 1749 @SU-SCORE.ARPA:WELCH@Ames-VMSB SIGBIG
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 2 Apr 85 17:49:24 PST
Received: from Ames-VMSB.ARPA by SU-SCORE.ARPA with TCP; Tue 2 Apr 85 17:40:41-PST
Date: 2 Apr 1985 1727-PST
From: WELCH at Ames-VMSB
Subject: SIGBIG
To: SUPER at SU-SCORE.ARPA.ARPA
Reply-To: WELCH@Ames-VMSB
ASSOCIATION FOR COMPUTING MACHINERY
San Francisco Golden Gate Chapter
"SIGBIG" Special Interest Committee
For Large High Speed Computers
Meetings on the first Wednesday of each month at 7:30 PM. Speakers
who can give insights to various aspects of SUPERCOMPUTING are
featured each month.
For information contact Mary Fowler, Chairperson (415) 694-6515
or Mike Austin, Publ. Chair (415) 423-8446
Next meeting:
Wednesday, April 3,1985, 7:30 PM
Speaker: Joseph Oliger/Stanford University
Subject: Relating the Structures of Problems, Algorithms
and Machines
Location: Stanford University
Room 252 Margaret Jacks Hall
This building is at the very end of Palm Drive
the continuation of University onto campus.
Abstract:
We consider simulations in mathematical physics using highly
parallel machines with distributed memory. The problem of
constructing "balanced" machines (not severly I/O or compute
bound) leads us to approximate conservation relations which
mimic the physics. A density and momentum of information are
defined and the objective of minimizing communication
requirements leads to minimum flux algorithms and hardware
configurations. Optimal directions of problem dissection can
be identified using this model. After introducing an
associated entropy for the process, estimates of desirable
granularity and limitations on attainable efficiency of
parallel machines can be obtained.
---------------------------------------------------------------
Tape-recordings of most of the previous may be
obtained in exchange for a tape cassette or $5.00 by contacting:
Mary Fowler (415)694-6515
Arpanet: mer.sigbig@ames-vmsb
Future Meetings:
May 1,1985 Alvin Despain/U.C. Berkeley The project AQUARIUS
------
∂02-Apr-85 1830 ullman@diablo meeting
Received: from SU-AIMVAX.ARPA by SU-AI.ARPA with TCP; 2 Apr 85 18:30:35 PST
Date: Tue, 2 Apr 85 18:29:17 pst
From: Jeff Ullman <ullman@diablo>
Subject: meeting
To: nail@diablo
Remember the meeting is (just this week) Wednesday 1PM in 252.
On subsequent weeks we meet in 301 at 11AM.
The schedule of speakers and topics as far as I know it is:
Date Speaker Topic
4/3 Roy Naish on mu-Prolog
4/10 Vardi Vassilou on interfacing Prolog/DBMS
4/17 Van Gelder Kannelakis on partition semantics
I have a number of articles that I would like somebody to
present in the future:
Lozinskii "Inference by generating and structuring of deductive DB's"
Sellis & Shapiro "Optimization of Extended DB Query languages"
Sciore & Warren "Towards an integrated DB←Prolog system"
I also have a copy of a survey "Foundations of Logic Programming"
by J. W. Lloyd, that is probably not suitable for presentation,
but that people might like to read.
∂02-Apr-85 2021 BRESNAN@SU-CSLI.ARPA Linguistics Drinks Resume Fridays
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 2 Apr 85 20:13:22 PST
Date: Tue 2 Apr 85 20:09:51-PST
From: Joan Bresnan <BRESNAN@SU-CSLI.ARPA>
Subject: Linguistics Drinks Resume Fridays
To: researchers@SU-CSLI.ARPA, RAS@SU-CSLI.ARPA, visitors@SU-CSLI.ARPA,
folks@SU-CSLI.ARPA, friends@SU-CSLI.ARPA
To Linguists of All Persuasions & Affiliations--
Starting this Friday, April 5, on the full moon of the month
drinks will resume in the Lounge of Linguistics, Building 100, at
4:30 p.m. Everyone with any affinal, agnatal, intellectual, or
disinterested connection to linguistics is welcome. You may also
wish to come unconnected, simply to observe real linguists, who
are known for their charisma, panache, and eloquence under drink,
interact ("interact": a California euphemism for . . . what we'll
be doing).
Your hostess will be a noted theoretical linguist (someone has
noted everything, Berkeley).
-------
∂03-Apr-85 0009 @SU-SCORE.ARPA:KARP@SUMEX-AIM.ARPA CSD TGIF reminder
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 3 Apr 85 00:09:02 PST
Received: from SUMEX-AIM.ARPA by SU-SCORE.ARPA with TCP; Wed 3 Apr 85 00:00:41-PST
Date: Wed 3 Apr 85 00:00:30-PST
From: Peter Karp <KARP@SUMEX-AIM.ARPA>
Subject: CSD TGIF reminder
To: su-bboard@SUMEX-AIM.ARPA, csd@SU-SCORE.ARPA
Our first TGIF of the quarter will be held Friday at 4:30. As
the weather has been getting nicer and the days have been getting
longer lately, it's likely we will soon be moving the TGIF outside
in front of Jacks. Presumably it is unnecessary to repeat Karp's
TGIF-locating algorithm (is this on the comp reading list?).
-------
∂03-Apr-85 0034 @SU-SCORE.ARPA:ARK*@SU-AI.ARPA CS Colloq, April 9: Cynthia Dwork, On the Sequential Nature of Unification
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 3 Apr 85 00:34:06 PST
Received: from SU-AI.ARPA by SU-SCORE.ARPA with TCP; Wed 3 Apr 85 00:01:07-PST
Date: 3 Apr 85 0000 PST
From: Arthur Keller <ARK@SU-AI.ARPA>
Subject: CS Colloq, April 9: Cynthia Dwork, On the Sequential Nature of Unification
To: All-Colloq@SU-SCORE.ARPA, Faculty@SU-SCORE.ARPA,
Dwork@MIT-XX.ARPA
CC: ARK@SU-AI.ARPA
CS Colloquium, April 9, 4:15pm, Terman Auditorium
ON THE SEQUENTIAL NATURE OF UNIFICATION
Cynthia Dwork
Massachusetts Institute of Technology
Unification of terms is a crucial step in resolution theorem proving, with
applications to a variety of symbolic computation problems. It will be
shown that the general problem is log-space complete for P, even if
inifinite substitutions are allowed. Thus a fast parallel algorithm for
unification is unlikely. More positively, term matching, an important
subcase of unification, will be shown to have a parallel algorithm
requiring a number of processors polynomial in n, the size of the input,
and running in time poly-logarithmic in n.
This talk assumes no familiarity with unification or its applications.
----
Cookies, thanks to Ginger, and possibly juice will probably be served in
the 3rd floor lounge of MJH at 3:45pm. See you there.
[This reminder is now expired.]
∂03-Apr-85 0846 PIERCE@SU-SCORE.ARPA My name
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 3 Apr 85 08:46:05 PST
Date: Wed 3 Apr 85 08:37:40-PST
From: Lee Pierce <PIERCE@SU-SCORE.ARPA>
Subject: My name
To: CSD@SU-SCORE.ARPA
Guess it's time to announce--I was married on March 23 and my new name
is Pierce. If you forget and send a message to Berman, however, I
will still receive it. Thanks.
Lee Pierce
-------
∂03-Apr-85 1217 LIBRARY@SU-SCORE.ARPA Journal of Automated Reasoning--And Other New Journals In The Math/CS Library
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 3 Apr 85 12:15:36 PST
Date: Wed 3 Apr 85 12:08:37-PST
From: C.S./Math Library <LIBRARY@SU-SCORE.ARPA>
Subject: Journal of Automated Reasoning--And Other New Journals In The Math/CS Library
To: su-bboards@SU-SCORE.ARPA
cc: faculty@SU-SCORE.ARPA
For those of you who have missed passed messages, the following is a list
of some of our more recent subscription purchases:
Abacus
Computers and Artificial Intelligence (Czechoslovakia)
Expert Systems: the international journal of knowledge engineering
ICOT Journal Digest: Fifth Generation Computer Systems
Integration: the VLSI Journal
Journal of Automated Reasoning
Journal of Logic Programming
Journal of Parallel and Distributed Computing
Journal of Pascal, Ada & Modula-2
Macworld
Tech Journal; for IBM Personal Computer Users
Technology and Sciences of Information; cover to cover translation of
Technique et Science Informatiques
As you hear of new journals that you would like to suggest for purchase,
please send me a message Library@SCORE to the attention of Harry Llull.
HL
-------
∂03-Apr-85 1436 EMMA@SU-CSLI.ARPA Linguistics Drinks Resume Fridays
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 3 Apr 85 14:35:57 PST
Mail-From: BRESNAN created at 2-Apr-85 20:09:51
Date: Tue 2 Apr 85 20:09:51-PST
From: Joan Bresnan <BRESNAN@SU-CSLI.ARPA>
Subject: Linguistics Drinks Resume Fridays
To: researchers@SU-CSLI.ARPA, RAS@SU-CSLI.ARPA, visitors@SU-CSLI.ARPA,
folks@SU-CSLI.ARPA, friends@SU-CSLI.ARPA
ReSent-Date: Wed 3 Apr 85 14:30:15-PST
ReSent-From: Emma Pease <Emma@SU-CSLI.ARPA>
ReSent-To: friends@SU-CSLI.ARPA
To Linguists of All Persuasions & Affiliations--
Starting this Friday, April 5, on the full moon of the month
drinks will resume in the Lounge of Linguistics, Building 100, at
4:30 p.m. Everyone with any affinal, agnatal, intellectual, or
disinterested connection to linguistics is welcome. You may also
wish to come unconnected, simply to observe real linguists, who
are known for their charisma, panache, and eloquence under drink,
interact ("interact": a California euphemism for . . . what we'll
be doing).
Your hostess will be a noted theoretical linguist (someone has
noted everything, Berkeley).
-------
∂03-Apr-85 1519 LIBRARY@SU-SCORE.ARPA 1985 Chapel Hill Conference on VLSI
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 3 Apr 85 15:07:07 PST
Date: Wed 3 Apr 85 14:51:42-PST
From: C.S./Math Library <LIBRARY@SU-SCORE.ARPA>
Subject: 1985 Chapel Hill Conference on VLSI
To: su-bboard@SU-SCORE.ARPA
cc: faculty@SU-SCORE.ARPA
The 7th annual conference on advanced research in VLSI will be held at Chapel
Hill, North Carolina, May 15-17. It is sponsored by NSF and the Dept. of
Computer Science, University of North Carolina. Reservations at the Granville
Towers should be made through Bobette Eckland, Conference Registration, Dept.
of Computer Science, New West Hall 035A, UNC, Chapel Hill, NC 27514.
Program committee includes Bryant-CMU, George Cox-Intel, Henry Fuchs-UNC,
Glasser-MIT, John Hayes-Michigan, Kung-CMU, Ousterhout-Berkeley, Penfield-MIT,
Preparata-Illinois, Rosenberg-Duke, Seitz-Caltech, Szymanski-Murray hill,
Tredennick-T.J. Watson.
HL
-------
∂03-Apr-85 1556 LIBRARY@SU-SCORE.ARPA SOCRATES--Searching Techniques
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 3 Apr 85 15:42:47 PST
Date: Wed 3 Apr 85 15:31:05-PST
From: C.S./Math Library <LIBRARY@SU-SCORE.ARPA>
Subject: SOCRATES--Searching Techniques
To: su-bboards@SU-SCORE.ARPA
cc: faculty@SU-SCORE.ARPA,
: ;
SOCRATES includes all the materials located in the Meyer Undergraduate Library.
Because of this, when searching books published before 1973, you may come up
with Meyer as the only location. However this does not mean that other
copies are not in the science branches or Green. In particular, if the title
is a mathematical or computer sciences title it is most probably also located
in the Math/CS Library.
You may have noticed that in some cases citations for the same title are not
clustered together. There can be a separate record for Math/CS, Meyer,
and/or Green. In addition, Meyer data appears to display first before
the more substantial collections found in the branches and Green. Therefore
it is sometimes necessary to look at all records to make sure you haven't
missed a more recent edition located in a branch as opposed to an out of
date edition located in Meyer or another part of the libraries.
Harry Llull
-------
∂03-Apr-85 1605 chertok%ucbcogsci@Berkeley UCB Cognitive Science Seminar--April 9
Received: from UCB-VAX.ARPA by SU-AI.ARPA with TCP; 3 Apr 85 15:30:12 PST
Received: from ucbcogsci.ARPA by UCB-VAX.ARPA (4.24/4.42)
id AA19704; Wed, 3 Apr 85 15:27:57 pst
Received: by ucbcogsci.ARPA (4.24/4.38)
id AA28347; Wed, 3 Apr 85 15:29:21 pst
Date: Wed, 3 Apr 85 15:29:21 pst
From: chertok%ucbcogsci@Berkeley (Paula Chertok)
Message-Id: <8504032329.AA28347@ucbcogsci.ARPA>
To: cogsci-friends%ucbcogsci@Berkeley
Subject: UCB Cognitive Science Seminar--April 9
BERKELEY COGNITIVE SCIENCE PROGRAM
Cognitive Science Seminar -- IDS 237B
TIME: Tuesday, April 9, 11 - 12:30
PLACE: 240 Bechtel Engineering Center
(followed by)
DISCUSSION: 12:30 - 1:30 in 200 Building T-4
SPEAKER: Jerry Fodor, Philosophy Department, MIT and
UC Berkeley
TITLE: ``Modules, Frames, Fridgeons, Sleeping
Dogs and The Music of The Spheres''
This paper continues the discussion of the frame problem
from ←λT←λh←λe ←λM←λo←λd←λu←λl←λa←λr←λi←λt←λy ←λo←λf ←λM←λi←λn←λd. It's argued that outbreaks of the
frame problem are characteristic of the study of unencapsulated -
hence nonmodular - cognitive faculties, and that this explains
why the effects of the problem are felt much more strongly when
we consider cognition and problem solving than when the topic is
perceptual integration. Since unencapsulation is a leading
characteristic of any process of rational nondemonstrative infer-
ence, it is argued that the solution of the frame problem is not
dissociable from the general problem of understanding such
processes. This rather gloomy assessment is then compared with
views current in AI according to which resort to `sleeping dog'
strategies has already made it possible to circumvent the frame
problem. I argue that the proposed solution begs the problem.
-------------------------------------------------------------
UPCOMING TALKS
April 16: Mark Johnson, Philosophy Department, Southern
Illinois University
April 23: David Dowty, Center for Advanced Study in the
Behavioral Sciences
April 30: Herbert H. Clark, Psychology Department, Stan-
ford University
-------------------------------------------------------------
∂03-Apr-85 1713 AUGARTEN@SU-SCORE.ARPA CS 100: History of Computers
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 3 Apr 85 17:13:22 PST
Date: Wed 3 Apr 85 17:03:06-PST
From: Stan Augarten <AUGARTEN@SU-SCORE.ARPA>
Subject: CS 100: History of Computers
To: csd@SU-SCORE.ARPA
I'm teaching a course on the history of computers--a very interesting course
based on my book Bit by Bit: An Illustrated History of Computers. The course
has three goals: to give students a general overview of the history of
computers, to give them some practice at historical research, and to give them
an understanding of where they, and their prospective profession, stand in
the grand continuum of time. So far only five students--all blessed with
outstanding taste and intelligence--have registered for the class. But I'd
love to have more students, and that's why I've sent you this message.
If you're a professor or an instructor, please tell your students about this
class; if you're a student, please stop by my classroom. The class meets
every Monday and Wednesday afternoons between 3:15 and 4:30 in room 320 of
ERL. For more information, call me at 325-3411 between 9 a.m. and 5 p.m.
Thanks, Stan Augarten.
-------
∂03-Apr-85 1718 EMMA@SU-CSLI.ARPA Newsletter Apr. 4, No. 23
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 3 Apr 85 17:18:13 PST
Date: Wed 3 Apr 85 16:26:36-PST
From: Emma Pease <Emma@SU-CSLI.ARPA>
Subject: Newsletter Apr. 4, No. 23
To: friends@SU-CSLI.ARPA
Tel: 497-3479
C S L I N E W S L E T T E R
←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←
April 4, 1985 Stanford Vol. 2, No. 23
←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←
A weekly publication of The Center for the Study of Language and
Information, Ventura Hall, Stanford University, Stanford, CA 94305
←←←←←←←←←←←←
CSLI ACTIVITIES FOR *THIS* THURSDAY, April 4, 1985
12 noon TINLunch
Ventura Hall ``Types, Translations, and Prepositions''
Conference Room by Mark Gawron, New York University
Discussion will be led by Mark Gawron
2:15 p.m. CSLI Seminar
Redwood Hall ``Manipulating Models in Syllogistic Reasoning''
Room G-19 Marilyn Ford, CSLI
Tom Wasow will lead the discussion
3:30 p.m. Tea
Ventura Hall
4:15 p.m. CSLI Colloquium
Redwood Hall ``Two Cheers for Functional Role Semantics''
Room G-19 Ned Block, Massachusetts Institute of Technology
←←←←←←←←←←←
CSLI ACTIVITIES FOR *NEXT* THURSDAY, April 11, 1985
12 noon TINLunch
Ventura Hall ``Semantics for Natural Language: Metaphysics
Conference Room for the Simple-minded?''
Chris Menzel, CSLI
(Abstract on page 2)
2:15 p.m. CSLI Seminar
Redwood Hall ``What if the World Were Really Quite Simple?''
Room G-19 Alex Pentland, CSLI
Discussion leader to be announced
(Abstract on page 2)
3:30 p.m. Tea
Ventura Hall
4:15 p.m. CSLI Colloquium
Redwood Hall ``A Formal Theory of Innate Linguistic Knowledge''
Room G-19 Janet Dean Fodor, University of Connecticut and CSLI
(Abstract on page 3)
!
Page 2 CSLI Newsletter April 4, 1985
←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←
ABSTRACT OF NEXT WEEK'S TINLUNCH
``Semantics for Natural Language: Metaphysics for the Simple-minded?''
What, exactly, is the connection between semantics and metaphysics?
A semantical theory gives an account of the meaning of certain
expressions in natural language, and, intuitively, the meaning of an
expression has to do with the connection between the expression (or an
utterance of it) and the world. Thus, a simple-minded view might be
that (as far as it goes) a correct semantical theory ipso facto yields
the sober metaphysical truth about what there is.
To the contrary, implicit in much work in semantics is the idea
that all we should expect of a good theory is that it be, in Keenan's
terms, descriptively adequate: it should provide a theoretical
structure which preserves our judgments of logical truth and
entailment, never mind the question of the literal metaphysical
details of the structure (e.g., that the denotations of singular terms
are complex sets of sets rather than individuals).
For next week's TINlunch I will provide a framework for discussion
by laying out the simple-minded view and its chief rival in somewhat
more detail. Being rather simple-minded myself, I'll attempt to
defend a reasonable version of the former. As grist for both
philosophical mills I will draw upon recent work in intensional logic,
Montague grammar, generalized quantifiers, the semantics of plurals,
and situation semantics. --Chris Menzel
←←←←←←←←←←←←
ABSTRACT OF NEXT WEEK'S SEMINAR
``What if the World Were Really Quite Simple?''
One of the major stumbling blocks for efforts in AI has been the
apparent overwhelming complexity of the natural world; for instance,
when an AI program tries to decide on a course of action (or the
meaning of a sentence) it is often defeated by the incredible number
of alternatives to consider. Results such as those of Tversky,
however, argue that people are able to use characteristics of the
current situation to somehow "index" directly into the two or three
most likely alternatives, so that deductive reasoning per se plays a
relatively minor role.
How could people accomplish such indexing? One possibility is that
the structure of our environment is really quite a bit simpler that it
appears on the surface, and that people are able to use this structure
to constrain their reasoning much more tightly than is done in current
AI research.
Is it possible that the world is really relatively simple? In
forming a scientific theory we may trade the size and complexity of
description against the amount of error. Because modern scientific
endeavors have placed great emphasis on increasingly accurate
description, very little effort has gone toward discovering a grain
size of description at which the world may be relatively simply
described while still maintaining a useful level of accuracy.
I will argue that such a simple description of the world is
plausible, discuss progress in discovering such a descriptive
vocabulary, and comment on how knowledge of such a vocabulary might
have a profound impact on AI and psychology. --Alex Pentland
!
Page 3 CSLI Newsletter April 4, 1985
←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←
ABSTRACT OF NEXT WEEK'S COLLOQUIUM
``A Formal Theory of Innate Linguistic Knowledge''
I assume that an infant is innately provided with some sort of
representational medium in which to record what he observes about his
target language. It has occasionally been suggested that the formal
properties of this mental metalanguage could be the source of
universal properties of natural languages. This differs from the
standard ( = substantive) approach, which assumes in addition that
certain statements of this metalanguage are innately tagged as true.
I propose to take the formal approach seriously. The way to do so
seems to be to try for a theory which accounts for ALL universals in
this way, i.e., solely on the basis of what can and cannot be
expressed in the metalanguage. The attempt is very informative, even
if ultimately it fails.
Success is certainly not guaranteed, for the formal theory
overthrows many familiar assumptions. For instance, it can be shown to
be incompatible (on standard assumptions about children and their
linguistic input) with the existence of any constraints on rule
application or on derivational representations. All the work of
distinguishing well-formed from ill-formed sentences must be done by
rules only. Constraints can determine the shape of the rules, but
cannot tidy up after them if they overgenerate.
It is easiest to see how to set about formulating grammars of this
kind within the framework of GPSG, and it is encouraging that a number
of universals do fall out as consequences of the GPSG formalism. But
there are problems too. Syntactic features, in particular, create
headaches for learnability. --Janet Fodor
[Note to attendees of the Berkeley Cognitive Science Seminars -- this
is the same as the paper presented there on 3/19.]
←←←←←←←←←←←←
NEW CSLI REPORT
Report No. 14C, ``Aspectual Classes in Situation Semantics'' by
Robin Cooper, has just been published. This analysis of certain tenses
of English, using the theory of situation semantics, may be obtained
by writing to Dikran@SU-CSLI or Dikran Karagueuzian, CSLI, Ventura
Hall, Stanford, CA 94305.
-------
∂04-Apr-85 1104 CLAY@SU-CSLI.ARPA System Security
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 4 Apr 85 11:02:47 PST
Date: Thu 4 Apr 85 10:57:35-PST
From: Clay Andres <CLAY@SU-CSLI.ARPA>
Subject: System Security
To: folks@SU-CSLI.ARPA
Yesterday afternoon, someone tried to gain illegal access to Turing. This
is a serious federal offense. If anyone receives messages asking for help
logging into the system, they should let the computer group know immediatly.
Clay
-------
∂04-Apr-85 1132 MACKINLAY@SUMEX-AIM.ARPA SILVER is back on the net
Received: from SUMEX-AIM.ARPA by SU-AI.ARPA with TCP; 4 Apr 85 11:31:20 PST
Date: Thu 4 Apr 85 11:24:23-PST
From: Jock Mackinlay <MACKINLAY@SUMEX-AIM.ARPA>
Subject: SILVER is back on the net
To: ksl-lisp-machines@SUMEX-AIM.ARPA
See subject. - Jock
-------
∂04-Apr-85 1202 @SU-CSLI.ARPA:sugai.pa@Xerox.ARPA RRR Meeting Reminder
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 4 Apr 85 12:01:56 PST
Received: from Xerox.ARPA by SU-CSLI.ARPA with TCP; Thu 4 Apr 85 11:56:08-PST
Received: from Semillon.ms by ArpaGateway.ms ; 04 APR 85 11:57:55 PST
Date: 4 Apr 85 11:57 PST
From: sugai.pa@XEROX.ARPA
Subject: RRR Meeting Reminder
To: RRR@SU-CSLI.ARPA
cc: sugai.pa@XEROX.ARPA
Please note that the previously scheduled representation group meeting
has been incorporated into the RRR seminar.
----------------------------
Spring Quarter Seminar
The Role of Representation in Computation and Reasoning
Time: Tuesday 2:00 -- 4:00 (tea will be provided)
First meeting: April 9, 1985
Place: The trailer seminar room (in front of Ventura)
For: RRR (F1, F2, F4, and C3 groups, and the representation
group)
Topic: 1. Rosenschein and Pereira, on the representational
nature (if any) of situated automata
In order to leave room for plenty of discussion, the format will be to
start with a 45-minute presentation, followed by a short break for tea,
followed by a one-hour discussion. Please note that the previously
scheduled representation group meeting has been incorporated into the
RRR seminar.
See you all on April 9.
∂04-Apr-85 1334 NILSSON@SU-SCORE.ARPA Tuesday Lunch
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 4 Apr 85 13:33:35 PST
Date: Thu 4 Apr 85 13:28:13-PST
From: Nils Nilsson <NILSSON@SU-SCORE.ARPA>
Subject: Tuesday Lunch
To: faculty@SU-SCORE.ARPA
Tuesday faculty lunches will resume next Tuesday, April 9. Although
I will be away, Jeff Ullman has agreed to moderate, introduce any
guests, etc. Discussion topic(s) will undoubtedly surface. On
April 16 we will be talking informally about teaching load and will
hear from the committee looking into these matters. -Nils
-------
∂04-Apr-85 1711 BSCOTT@SU-SCORE.ARPA Down Payment Assistance Program
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 4 Apr 85 17:10:50 PST
Date: Thu 4 Apr 85 16:39:32-PST
From: Betty Scott <BSCOTT@SU-SCORE.ARPA>
Subject: Down Payment Assistance Program
To: Faculty@SU-SCORE.ARPA
cc: BScott@SU-SCORE.ARPA
Nils has received from Dean Wessells a memorandum concerning a Down Payment
Assistance Program which has been recently approved by the Board of Trustees.
I have the memorandum, with attachments, in my office if you are interested
in the program. This is a merely a summary for your information:
This program will allow many faculty with only 10% of a required down payment
to purchase homes on campus or in the Stanford vicinity. Special supplementary
salary will be provided to faculty should the rate of interest on the Down
Payment Assistance Program (DAP) loan be greater than 10%. This supplementary
pay is unrelated to the Housing Allowance Plan (HAP) which is a separate salary
supplement for new faculty when a home is purchased. Eligibility for DAP loans
includes current faculty who have been renting, but never able to buy. Note
the program is available only for purchases, and not for refinancing, and only
to those who do not currently own a home within a fifteen mile radius of Hoover
Tower or currently have a COIN loan.
It is expected that this new program will be significant when recruiting new
faculty members.
Betty
-------
∂05-Apr-85 1016 @SU-CSLI.ARPA:PCOHEN@SRI-AI.ARPA Tues, the 9th
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 5 Apr 85 10:16:14 PST
Received: from SRI-AI.ARPA by SU-CSLI.ARPA with TCP; Fri 5 Apr 85 10:12:28-PST
Date: Fri 5 Apr 85 10:13:35-PST
From: Phil Cohen <PCOHEN@SRI-AI.ARPA>
Subject: Tues, the 9th
To: NL4: ;
We will have a talk after all. Stephen Neale of the Linguistics Dept. will
speak about: "Utterance Situations", a reconciliation of Gricean
theory with Situation Semantics. Time: 12:30 - 2pm, Ventura Seminar
room.
See you there,
Phil
-------
∂05-Apr-85 1025 @SU-CSLI.ARPA:PCOHEN@SRI-AI.ARPA correction
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 5 Apr 85 10:21:40 PST
Received: from SRI-AI.ARPA by SU-CSLI.ARPA with TCP; Fri 5 Apr 85 10:18:45-PST
Date: Fri 5 Apr 85 10:20:02-PST
From: Phil Cohen <PCOHEN@SRI-AI.ARPA>
Subject: correction
To: NL4: ;
Stephen Neale is in the Philosophy Dept.
Phil
-------
∂05-Apr-85 1308 LIBRARY@SU-SCORE.ARPA Math/CS Library New Books
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 5 Apr 85 13:08:42 PST
Date: Fri 5 Apr 85 13:05:51-PST
From: C.S./Math Library <LIBRARY@SU-SCORE.ARPA>
Subject: Math/CS Library New Books
To: su-bboards@SU-SCORE.ARPA
cc: : ;
Graphs and Applications. Proceedings of the First Colorado Symposium on
Graph Theory. edited by Harary and Maybee. QA166.C616 1982
Expert Systems; Principles and Case Studies. edited by Richard Forsyth
QA76.9.E96E96 1984
Development and Maintenance of Large Software Systems. by Pizzarello.
QA76.6.P59 1984
Dynamo User's Manual. 6th edition. by Pugh. HF5548.5.D2P8 1983
IEEE Standard C/Atlas Test Language. IEEE Std. 716-1982. TK7878.I34 1982
IEEE Standard C/Atlas Syntax. IEEE Std. 717-1982. TK7878.I345 1982
Pascal by Ledin. QA76.73.P2L43 1982
Understanding the MacIntosh Computer. by Dayton. QA76.8.M3W44 1984
Single-chip Microcomputers. edited by Lister. QA76.5.S5538 1984
Harry Llull
-------
∂05-Apr-85 1349 Jskud@diablo Re: another paper
Received: from SU-AIMVAX.ARPA by SU-AI.ARPA with TCP; 5 Apr 85 13:49:33 PST
Date: Fri, 5 Apr 85 13:48:08 pst
From: Joseph P. Skudlarek <Jskud@diablo>
Subject: Re: another paper
To: nail@diablo, ullman@diablo
I'm interestedin presenting the "Optimization of Extended
DB Query Languages" paper. Talk with you later today.
∂05-Apr-85 1401 @SU-CSLI.ARPA:brown@Kestrel Re: Linguistics Drinks Resume Fridays
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 5 Apr 85 14:01:48 PST
Received: from Kestrel.ARPA by SU-CSLI.ARPA with TCP; Fri 5 Apr 85 13:54:36-PST
Received: by Kestrel.ARPA (4.12/4.7)
id AA00264; Fri, 5 Apr 85 11:24:16 pst
Date: Fri, 5 Apr 85 11:24:16 pst
From: brown@Kestrel (Tom Brown)
Message-Id: <8504051924.AA00264@Kestrel.ARPA>
To: BRESNAN@SU-CSLI.ARPA, RAS@SU-CSLI.ARPA, folks@SU-CSLI.ARPA,
friends@SU-CSLI.ARPA, researchers@SU-CSLI.ARPA, visitors@SU-CSLI.ARPA
Subject: Re: Linguistics Drinks Resume Fridays
Are you sure about the R status of this message? --X
∂05-Apr-85 1508 BLOCK@SU-CSLI.ARPA Linguistics for non-linguists
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 5 Apr 85 15:08:12 PST
Date: Fri 5 Apr 85 15:02:29-PST
From: Ned Block <BLOCK@SU-CSLI.ARPA>
Subject: Linguistics for non-linguists
To: folks@SU-CSLI.ARPA, friends@SU-CSLI.ARPA, su-bboards@SU-CSLI.ARPA
cc: emma@SU-CSLI.ARPA
PANEL DISCUSSION ON SYNTACTIC THEORIES
The lecturers in our series on syntactic theories for non-linguists
are back, this time in a panel discussion. Joan Bresnan, Geoff
Pullum, and Peter Sells will take questions from the audience (no
initial presentations, so come with questions). This time, linguists
ARE allowed, but they are asked to stick to matters that non-linguists
will have a chance of understanding. If you think of questions in
advance, send them to the panelists so that they can think about them
(BRESNAN@SU-CSLI,PULLUM@SU-CSLI,SELLS@SU-CSLI).
Tuesday, April 16
1:30
Redwood G-19
-------
∂05-Apr-85 1541 BSCOTT@SU-SCORE.ARPA Closing today at 4:00
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 5 Apr 85 15:41:31 PST
Date: Fri 5 Apr 85 15:17:36-PST
From: Betty Scott <BSCOTT@SU-SCORE.ARPA>
Subject: Closing today at 4:00
To: CSD@SU-SCORE.ARPA
cc: BScott@SU-SCORE.ARPA
This is to let everyone know that administrative offices in MJH will close at
4:00 p.m. today.
Betty
-------
∂05-Apr-85 1553 EMMA@SU-CSLI.ARPA Linguistics for non-linguists
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 5 Apr 85 15:52:40 PST
Mail-From: BLOCK created at 5-Apr-85 15:02:29
Date: Fri 5 Apr 85 15:02:29-PST
From: Ned Block <BLOCK@SU-CSLI.ARPA>
Subject: Linguistics for non-linguists
To: folks@SU-CSLI.ARPA, friends@SU-CSLI.ARPA, su-bboards@SU-CSLI.ARPA
cc: emma@SU-CSLI.ARPA
ReSent-Date: Fri 5 Apr 85 15:43:11-PST
ReSent-From: Emma Pease <Emma@SU-CSLI.ARPA>
ReSent-To: friends@SU-CSLI.ARPA
PANEL DISCUSSION ON SYNTACTIC THEORIES
The lecturers in our series on syntactic theories for non-linguists
are back, this time in a panel discussion. Joan Bresnan, Geoff
Pullum, and Peter Sells will take questions from the audience (no
initial presentations, so come with questions). This time, linguists
ARE allowed, but they are asked to stick to matters that non-linguists
will have a chance of understanding. If you think of questions in
advance, send them to the panelists so that they can think about them
(BRESNAN@SU-CSLI,PULLUM@SU-CSLI,SELLS@SU-CSLI).
Tuesday, April 16
1:30
Redwood G-19
-------
∂06-Apr-85 @SU-SCORE.ARPA:udi@wisc-rsch.arpa 06-Apr-85 JMC Toronto Day at Wesleyan
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 6 Apr 85 08:48:55 PST
Received: from wisc-rsch.arpa by SU-SCORE.ARPA with TCP; Sat 6 Apr 85 08:47:57-PST
Received: by wisc-rsch.arpa; Sat, 6 Apr 85 10:36:44 cst
Received: from yale.arpa by wisc-rsch.arpa; Sat, 6 Apr 85 03:52:51 cst
Received: by YALE-BULLDOG.YALE.ARPA; 5 Apr 85 16:37:53 EST (Fri)
Message-Id: <8504052137.AA18134@YALE-BULLDOG.YALE.ARPA>
Received: from YALE-RING by YALE-RES via CHAOS; Fri, 5 Apr 85 16:38:20 EST
Subject: Toronto Day at Wesleyan
Date: Fri, 5 Apr 85 16:38:23 EST
From: Neil Immerman <Immerman@YALE.ARPA>
To: udi@WISC-RSCH
Sender: udi@wisc-rsch.arpa
Resent-From: udi@wisc-rsch.arpa (Udi Manber)
Resent-To: theory:;@wisc-rsch.arpa
announcing . . . . . TORONTO DAY AT WESLEYAN
Saturday May 4
Talks and discussions led by a contingent of computer scientists
from the University of Toronto.
9:30 am Steve Cook
A Taxonomy of Problems with Fast Parallel Algorihms
10:30 am Coffee Break
11:00am Mike Luby
A Simple Parallel Algorithm for the Maximal Independent Set Problem
12:00 Noon Lunch in Woodhead Lounge (184 Science Center)
2:00 pm Allan Borodin
Parallel Algebraic Complexity
3:00 pm Coffee Break
3:30 pm Charles Rackoff
Some Definitions and Issues in the Theory of Cryptography
4:30 pm Problem Session, Math Lounge (Sixth Floor)
Beer, Soda and Munchies
6:30 pm Buffet Dinner
[All talks will be held in Room 58 Science Center.]
On Friday, May 3, Steve Cook will give a talk at 4 P.M. in
150 Science Center intended for a general audience: Can Computers
Routinely Discover Mathematical Proofs? This talk will be followed
by a reception in the Woodhead Lounge at 5 P.M. and then dinner at a
local restaurant at 6:30 P.M. All Toronto Day participants are welcome
to attend. On Sunday there will be informal activities, including a
morning hike, before heading for Providence (STOC), 75 miles east.
Assistance with local transportation and travel to STOC provided
on request. Lodging available at a local motel; sleeping bag space
provided by hosts.
Toronto Day is hosted by Alan Cobham, Dan Dougherty, Sorin Istrail,
Susan Landau, and Carol Wood, and is funded in part by The Sloan
Foundation, Proctor and Gamble Co., and the Wesleyan University Department
of Mathematics.
For further information contact the Wesleyan Mathematics
Department at (203)347-9411 Ext.2398 or Carol Wood at Ext.2648
(Bitnet address WOODatWESLYN).
* * * * * * * * * * * * * * * * * * * * * * * * * * * * * *
-------
jmc - Shall we organize a car pool?
∂07-Apr-85 2252 @SU-SCORE.ARPA:udi@wisc-rsch.arpa Conference Schedule
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 7 Apr 85 22:52:45 PST
Received: from wisc-rsch.arpa by SU-SCORE.ARPA with TCP; Sun 7 Apr 85 22:47:14-PST
Received: by wisc-rsch.arpa; Mon, 8 Apr 85 00:29:58 cst
Received: from wisc-crys.arpa by wisc-rsch.arpa; Sun, 7 Apr 85 20:07:45 cst
Message-Id: <8504080207.AA16556@wisc-crys.arpa>
Received: from csnet-relay.arpa by wisc-crys.arpa; Sun, 7 Apr 85 20:07:21 cst
Received: from kansas-state by csnet-relay.csnet id ad07201; 7 Apr 85 21:08 EST
Received: by ksuvax1.KSU id AA26276; Fri, 5 Apr 85 16:32:34 cst
Date: Fri, 5 Apr 85 16:32:34 cst
From: Austin Melton <austin%kansas-state.csnet@csnet-relay.arpa>
To: theory@wisc-crys.ARPA
Subject: Conference Schedule
Sender: udi@wisc-rsch.arpa
Resent-From: udi@wisc-rsch.arpa (Udi Manber)
Resent-To: theory:;@wisc-rsch.arpa
ADVANCED PROGRAM
CONFERENCE ON THE MATHEMATICAL FOUNDATIONS OF PROGRAMMING SEMANTICS
Union Building Sponsored By:
Kansas State University Iowa State University
Manhattan, Kansas Kansas State University
April 11-12, 1985 The University of Kansas
The University of Nebraska
Wichita State University
-----------------------------------------------------------------------
WEDNESDAY APRIL 10
-----------------------------------------------------------------------
7:30 PM Registration and Social
Ramada Inn, Room 622
(Ramada Inn is located directly south
of the Union Building)
-----------------------------------------------------------------------
THURSDAY APRIL 11
-----------------------------------------------------------------------
9:00 - 10:00 Registration and Coffee
Big 8 Room Union, Second floor
10:00 - 10:10 Opening Remarks Elizabeth Unger, Conference Co-chairperson
Welcome: Owen Koeppe, Provost
Kansas State University
10:10 - 11:00 Keynote Speaker: Dana J. Scott, University Professor
of Computer Science and Mathematical Logic
Carnegie-Mellon University
Introduction by: Adrian Tang
The University of Kansas
11:05 - 12:30 SESSION 1
Chair: Roy Keller
University of Nebraska
Paper 1: S. Kamal Abdali & David S. Wise
Tektronix, Inc. & Oregon State Univ-Indiana
University
"Standard Semantics for Algol-style Block
Structures"
Paper 2: Michael Mislove
Tulane University
"Detecting Local Finite Breadth in
Continuous Lattices and Semilattices"
Paper 3: David Benson & Jerzy Tiuryn
Washington State Univ. & Univ. of
Warsaw, Poland
"Fixed-point in Process Algebras with
Internal Actions"
12:30 - 1:30 Lunch (Cottonwood Room, Union)
1:30 - 3:10 SESSION 2 (Big 8 Room)
Chair: David Gustafson
Kansas State University
Invited Speaker: Horst Herrlich
University of Bremen, West Germany
"Universal Topology"
Paper 4: Paul C. Gilmore
University of British Columbia, Canada
"Natural Deduction Based Set Theories"
Paper 5: Ana Pasztor
Carnegie-Mellon University
"On the Variety Concept for w-Continuous
Algebras: A General Approach"
3:10 - 3:30 Break
3:30 - 5:10 SESSION 3 (Big 8 Room)
Chair: Robert Wherritt
Wichita State University
Invited Speaker: Carl Gunter
Carnegie-Mellon University
"A Survey of Categories of Domains"
Paper 6: Jon Shultis
University of Colorado
"What is a Model? A Consumer's Perspective
on Semantic Theory"
Paper 7: Jimmie D. Lawson
Louisiana State University
"Quasicontinuous Posets"
6:30 - 8:00 Banquet (Flinthills Room, Union)
Welcome: William Stamey, Dean
College of Arts and Science
Address by: Dana Scott
Carnegie-Mellon University
8:00 - 10:00 Research Discussion (Room 206, Union)
---------------------------------------------------------------------
FRIDAY APRIL 12
---------------------------------------------------------------------
8:00 - 8:30 Coffee
8:30 - 10:25 SESSION 4 (Room 207, Union)
Chair: Diana Palenz
Wichita State University
Paper 8: Jurgen Koslowski
Kansas State University
"Continuous Categories"
Paper 9: Austin Melton & David Schmidt
Kansas State University Iowa State
University
"A Topological Framework for cpos Lacking
Bottom Elements"
Paper 10: Mike Main
University of Colorado
"Free Constructions of Powerdomains"
Paper 11: P. Venugopalan
University of Connecticut
"Union Compatible Countable Subset Systems"
8:30 - 10:25 SESSION 5 (Room 212, Union)
Chair: Maria Wilson
Kansas State University
Paper 12: Naphtali Rishe
University of California, Santa Barbara
"On Denotational Semantics of Data Bases"
Paper 13: D.E. Stevenson
Clemson University
"Encapsulating Semantics: An Approach to
Unifying Semantic Concepts"
Paper 14: Teresa A. Thomas & Donald F. Stanat
University of North Carolina at Chapel Hill
"An FP Domain with Streams"
Paper 15: Maria Zamfir & David Martin
University of California, Los Angeles
"On the Syntax and Semantics of Concurrent
Computing"
10:25 - 10:45 Break
10:45 - 12:10 SESSION 6 (Room 212, Union)
Chair: David Schmidt
Iowa State University
Invited Speaker: Stephen Brookes
Carnegie-Mellon University
"A Fully Abstract Semantics and a Proof
System for an Algol-like Language with
with Sharing"
Research Question & Answer Session
12:10 - 1:15 Lunch (Cottonwood Room)
1:15 - 2:55 SESSION 7 (Room 212, Union)
Chair: Robert Neufeld
Wichita State University
Invited Speaker: George Strecker
Kansas State University
"Cartesian Closed Topological Hulls as
Injective Hulls"
Paper 16: Stuart Margolis
University of Nebraska
"Recent Developments in the Theory of
Rational Languages and Rational Paths"
Paper 17: Jonathan Smith
Iowa State University
"Model Theory, Partial Orders, and Digital
Geometry"
2:55 - 3:15 Break
3:15 - 5:00 SESSION 8 (Room 212, Union)
Chair: Tsutomu Kamimura
The University of Kansas
Invited Speaker: Adrian Tang
The University of Kansas
Paper 18: Ernie Manes
University of Massachusetts at Amherst
"Boolean Structure and Iteration in a
Partially-additive Category"
Paper 19: P. Curien
University of Paris
"Categorical Combinatoric Logic"
6:30 - 10:00 Social: Melton's Home
1743 Fairchild
∂08-Apr-85 0048 RESTIVO@SU-SCORE.ARPA PROLOG Digest V3 #16
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 8 Apr 85 00:47:47 PST
Date: Sunday, April 7, 1985 7:51PM
From: Chuck Restivo (The Moderator) <PROLOG-REQUEST@SU-SCORE.ARPA>
Reply-to: PROLOG@SU-SCORE.ARPA
US-Mail: P.O. Box 4584, Stanford CA 94305
Phone: (415) 326-5550
Subject: PROLOG Digest V3 #16
To: PROLOG@SU-SCORE.ARPA
PROLOG Digest Monday, 8 Apr 1985 Volume 3 : Issue 16
Today's Topics:
Puzzles - Maps & C-Prolog
Implementations - Bugs & Cases & Cuts & Determinism,
& Snips & Denotational Semantics
----------------------------------------------------------------------
Date: Wed, 27 Mar 85 11:04:31 -0200
From: Ehud Shapiro <Udi%Wisdom.bitnet@WISCVM.ARPA>
Subject: Not the semantics of Concurrent Prolog
I know, I know I should be doing semantics of Concurrent
Prolog, but instead I wrote a Prolog program for map
colouring. It seems that the program doesn't do any work.
The magic is, as usual, in the combination of unification
and non-determinism.
The program defines the relation colour(Map,Colours),
between a map and a list of colours, which is true if Map
is legally coloured using Colours.
A map is represented using adjacency-lists, where with
each country we associate its colour, and the list of
colours of its neighbours. If the program is used in
generate-mode, i.e. to colour an uncoloured map, these
colours would be uninstantiated variables. The program
would then compute all possible colourings. For example,
the uncoloured map
←←←←←←←←←←
| a |
|←←←←←←←←|
|b |c |d |
|←←|←←|←←|
|e |f |
|←←←|←←←←|
is represented by the term:
map( [country(a,A,[B,C,D]),
country(b,B,[A,C,E]),
country(c,C,[A,B,D,E,F]),
country(d,D,[A,B,F]),
country(e,E,[B,C,F]),
country(f,F,[C,D,E])
]
).
And here is the program:
colour←map([Country|Map],Colours) :-
colour←country(Country,Colours),
colour←map(Map,Colours).
colour←map([],←Colours).
colour←country(country(←Name,C,AdjacentCs),Colours) :-
remove(C,Colours,Colours1),
subset(AdjacentCs,Colours1).
Which uses a couple of utilities:
subset([C|Cs],Colours) :-
remove(C,Colours,←),
subset(Cs,Colours).
subset([],←Colours).
remove(C,[C|Cs],Cs).
remove(C,[C1|Cs],[C1|Cs1]) :-
remove(C,Cs,Cs1).
To test the program, use the following code:
test(Map) :-
map(Map),
colours(Colours),
colour←map(Map,Colours).
colours([red,green,blue,white]).
By the way, the running time of the algorithm is
exponential in the size of the map.
-- Ehud Shapiro
------------------------------
Date: Thu, 28 Mar 85 00:23:12 -0200
From: Ehud Shapiro <Udi%Wisdom.bitnet@WISCVM.ARPA>
Subject: Addendum
p.s. An exercise, for all you program-provers out there:
Prove that colour←map(Map,Colours) terminates succesfully,
if Maprepresents a planar graph, and Colours contains four
distinct colours.
Hint: assume that remove/3 and subset/2 are defined
correctly, just concentrate on proving the first three
axioms...
------------------------------
Date: Tue, 2 Apr 85 19:24:06 pst
From: Allen VanGelder <AVG@Diablo>
Subject: Riddle: How did I do this in C Prolog?
In the following script I did what it looks like I did.
cpl gives me C Prolog top level. write/1 has not been
redefined; the write you see actually wrote "thisIsX(6)";
p.pl has no output statements.
RIDDLE: What kind of rule(s) in p.pl account for this
behavior?
Script started on Tue Apr 2 18:50:19 1985
% cpl
C Prolog version 1.4e.edai
yes
| ?- ['p.pl'].
p.pl consulted 84 bytes 0.016667 sec.
yes
| ?- p(X).
X = 6 ;
no
| ?- p(X), write(thisIsX(X)), X=6.
thisIsX(6)
no
| ?- halt.
% Prolog execution halted
% ↑D
script done on Tue Apr 2 18:53:33 1985
For those not fluent in C Prolog syntax, I
loaded "p.pl", then issued goal "p(X)" and
got the solution "X=6". The semi-colon asks
for more solutions; there were none. Then I
issued the more complex goal, which printed
"thisIsX(6)" and failed.
------------------------------
Date: Thu, 28 Mar 85 15:54:12 pst
From: Newton@CIT-Vax (Mike Newton)
Subject: CProlog bug (revisited)
Hi,
There was an error in my posting on the C-Prolog Problem,
instead of
f(A) :- (true, ! ; B=0), A =< 7.
the test case should have been:
f(A) :- (true, ! ; A=0), A =< 7.
↑------ i goofed here.
Here is a transcript of the session using the distributed
C-Prolog 1.5:
Script started on Thu Mar 28 10:09:24 1985
((cit-vax:1)) pd
/usr/src/cit/cprolog ~
((cit-vax:2)) ls
AIDigest STRIPSTANFORD cpl1x4a fprolog
README coling cpl1x5 stanford
((cit-vax:3)) cd cpl1x5
((cit-vax:4)) ./prolog
C-Prolog version 1.5
| ?- [user].
| f(A) :- (true, ! ; A = 0) , A =< 7.
user consulted 128 bytes 0.0500008 sec.
yes
| ?- f(3).
! Error in arithmetic expression: ! is not a number
no
| ?
[ Prolog execution halted ]
((cit-vax:5)
script done on Thu Mar 28 10:10:16 1985
The problem first surfaced in a grammar rule that was
being used in our compiler. I simplified it to the
above test case... The fix that i posted in the earlier
newsletter has not yet caused any noticeable problems, but..
Regards,
-- Mike
[818-356-6771]
------------------------------
Date: 29 Mar 85 14:05 PST
From: Kahn.pa@XEROX.ARPA
Subject: A Case for Cases
The case for Cases is mostly a mater of taste. I think
a Prolog clause should be meaningful on its own. Yet
this often leads to redundant computation. Consider
the common way of writing a predicate to compute the
maximum of two numbers.
max(X,Y,X) :- X > Y, !.
max(X,Y,Y).
The second clause is pretty bizarre. Using Cases one
can package up the two clauses and get some clearer (and
just as efficient code). I admit that in this example and
those that come to mind If would do as well. (Richard
is right that its hard to find examples where the
non-determinism of the test is useful). Cases has a
neater semantics than "If though".
I sometimes think that max should be written simply
as follows
max(X,Y,X) :- X > Y.
max(X,Y,Y) :- not (X > Y).
and let the compiler compile out the redundant test.
The problem with this is that it can be pretty verbose
if there are many cases since one needs to explictly
write the negation of the previous tests.
------------------------------
Date: Wed, 27 Mar 85 18:13:23 pst
From: Peter Ludemann <Ludemann%UBC.Csnet@csnet-relay.arpa>
Subject: Cuts, Determinism.
I agree that "soft" cuts don't seem to have much use.
I think that tests by their nature shouldn't instantiate
any variables, so backtracking over them won't produce
anything new. However, the semantics of soft cuts are
nicer: it is straightforward to rewrite clauses containing
soft cuts into a form which contains no cuts at all.
As to non-determinism of append. What I meant is this.
Suppose we have a clause:
g(A, B, C) :- append(A, B, C), g2(A, B, C).
Unless we know that at least two of the three variables
are ground, we don't know if the call TO append is
deterministic. For example, if we called "g(X, Y,
[1,2,3])", the call to append is clearly
non-deterministic and we cannot do the last call
optimisation with "g2". So, in the absence of mode
declarations and some moderately complicated compile-time
analysis, we can only tell at run time that a given call
is deterministic. This is easy (and reasonably efficient)
to determine at run-time - if a goal is "done with"
(backtracking over it will only result in failure), it can
be removed from the stack. The last call optimisation is
applicable if the backtrack clause is the parent clause.
Warren's SRI Tech Note 309 talks about goal stacking in
the last few pages and he makes the point that tail
recursion optimization (which I prefer the call last call
optimization) is applicable at every procedure call -
"one simply discards the calling goal if it is later than
the last choice point". I think that my (run-time) method
achieves that.
------------------------------
Date: Wed, 27 Mar 85 22:12:56 pst
From: Prolog@CIT-Vax (Paul Prolog)
Subject: Cuts and Snips
We have an implementation of prolog at Caltech
with a new concept we are calling snips, a concept
conceived by Jim Kajiya. This is somewhat like the
"soft" cut described in the last couple of digests.
Our point is to allow more localized control of
backtracking within a clause. Snips allow a backtrack
frame to be created within the scope of a clause,
but then, unlike a cut, allow only the backtrack
points the user wishes, to be removed. It would work
like this. If one had a clause
a :- b,c,[!,d,e,f,!],g
where the [! marks the beginning of a snip region,
and !] marks the end of a snip region, backtracking
would be normal for the b and c clauses. Backtracking
would continue normally into the snip region, but once
the !] was passed, any backtrack points created by
d, e, or f would be deleted from the backtrack stack.
Then, if g failed, any alternatives of c or b would be
tried. If before the !] were executed, backtracking
took execution to the left of the snip region (left
of [!), nothing special would happen. Cut regions can
be nested as deeply as the user wishes.
Snips are useful when one wants to have backtracking
procedures, but also wants to use the same procedure
in deterministic programs. For instance, in a pattern
matcher I am writing to handle associativity and
commutivity, I need to take a list and take elements
out, while keeping a list of the non-selected
elements. This is written in a clause as
..... ,
append(Begin,[TestElement|End],OriginalList),
append(Begin,End,RestList),
.....
(Of course, there may be a better way to write this,
but I wrote it this way). Now the second append
doesn't ever need to backtrack, since it is not
being used for search. However, having two append
predicates, one with no backtrack point, and one
without, would be silly. A cut may not be appropriate,
since the procedure that this clause fragment is a
part of may need to backtrack. A singlet clause
with a cut could be used, but program comprehension
would be better if all logical parts of a clause
could stick together, as in the example above.
So, rewriting it as
..... ,
append(Begin,[TestElement|End],OriginalList),
[!,append(Begin,End,RestList),!],
.....
would remove the problem.
We feel that snips are a valid addition to
standard prolog. Though there are ways of getting
around the need for them, they are useful in their
own right. (A more detailed article may be coming).
Any comments?
-Keith Hughes
hughes@cit-vax
PS. The other members of our group are Mike Newton
(newton@cit-vax) and Jim Kajiya (kajiya@cit-20).
------------------------------
Date: Thu, 28 Mar 85 21:06:54 pst
From: Paul Voda <Voda%ubc.csnet@csnet-relay.arpa>
Subject: Denot. semantics versus the Symbiosis approach.
This is a note on the relation between the
denotational and "logical" semantics of logic
programming languages. Denotational semantics deals
with the meaning and control at the same time. Thus
for a sequential disjunction the formula
"P(5) or 5 = 5" will denote the undefined element if
the computation of "P(5)" does not terminate.
Viewed as a formula of a predicate calculus, i.e.
ignoring the control and concentrating only on the
meaning, the formula is true. This combination of
the meaning and control in the denotational semantics
is not too bad when the control is just the normal
reduction (lazy evaluation) of the lambda calculus.
As the control gets more complicated (explicit
parallelism, input-output annotations, cuts and commits,
etc.) the meaning functions of the denotational
semantics must necessarily grow more and more complicated
(in the case of parallelism there is even no satisfactory
solution).
In my Symbiosis paper I essentially took the step taken
by Hoare when he separated the partial correctness from
the termination. Divide and impera is always a good
methodological principle but in the case of logic (or
functional) languages the gains are quite significant.
First of all, programs of these languages are formulas resp.
terms of a first order logic theory. The meaning of programs
(corresponding to the partial correctness) is obtained
simply from the standard model of the theory. The reasoning
about the programs can utilize the full deductive power
of first order theories (quantifiers, induction). The
termination is separated from the "logical" meaning.
Now, computations are proofs and in the Symbiosis paper
I show on five different programming languages how to set up
a deductively restricted subtheory of the meaning theory
where the computations exactly correspond to the derivations
in the subtheories. This takes care of the operational
semantics and the problem of termination is reduced to
the proof-theoretical question of existence of a proof.
The questions of existence of proofs are much harder
then the quite straightforward questions of meaning.
But I do not share the gloomy view of Uday
(Prolog Digest 3#15) that we cannot do better than "run-it-
and-see". First of all, we are dealing with deductively
restricted subtheories. Thus it is not too difficult to
arithmetize the predicate of the provability (i.e the
provability in the operational subtheory) in the meaning
theory. One can then proceed and prove a couple of
sufficient conditions for the termination. The conditions
are generally of the form "if .... terminates then
←←←← ....←←← also terminates". These theorems can
be then used to reason about the termination.
----Paul Voda
------------------------------
Date: Tue 2 Apr 85 13:46:32-PST
From: Joseph A. Goguen <GOGUEN@SRI-AI.ARPA>
Subject: logic programming semantics
I have been following the discussions of the semantics
of Prolog and Concurrent Prolog in the Digest with a
rising mixture of fascination and despair. If the
designers of a language have trouble explaining (and
perhaps even understanding) some constructs of their
language, then perhaps there is something wrong with
those constructs. I hasten to add that I find the
read-only annotation idea very appealing, and I hope
that somehow its semantics can be made as simple as I
once thought it might be.
There is a point that I would like to add to the
discussion: Having a denotational semantics for a
language does not make it more respectable. There are
many nice things about denotational semantics, besides
its mathematical elegance and the way it handles
potentially infinite control structures; one of these
is that it serves as a pretty accurate measure of
just how horrible a programming language is; in particular,
you can see how many levels of continuations are needed,
which constructs need the extra complications, how large
the definition is (e.g., how many hundreds or thousands
of semantic equations), etc. Of course, you can give a
denotational semantics for almost anything if you really
want to (e.g., Ada). But I claim that if a language
*needs* a denotational semantics in order to explain its
constructs, then that language is too complicated. The
good thing about a real logic programming language is
exactly that it *does* *not* *need* such a semantics:
a program means exactly what it says. It is a symptom
of the constructs of a language being too operational that
one is driven to denotational semantics to explain them.
So I guess what I'm saying is I am afraid that Concurrent
Prolog is getting us further away from the original
ideals of logic programming, closer to PL/I and its cousins.
In my opinion, what we need is more research on how to do
practical programming with pure logic programming
languages; this may mean enriching the logic, or doing some
things in a metalogic, or whatever. But making Prolog less
and less pure does not seem to me the right way to go.
In case you wonder what I mean by a "logic programming
language", here is a general definition: a language
whose programs consist of sentences in a well understood
(and reasonably simple) logical system, with operational
semanticsgiven by deduction in that logical system. In
particular, "well understood" should include a completeness
theorem for deduction. Notice that both "relational" or
"Horn clause" programming (i.e., what is usually called
"logic programming") and functional programming are special
cases, with logics respectively first order Horn clause
logic, and equational logic (in fact, usually higher order
equational logic).
I would also impose another requirement, that every program
has an initial model. This provides a foundation for
database manipulations, since you know exactly what is true
-- namely, what can be proved from the axioms -- and
everything else is false (this is Occam's famous razor); in
fact, the "closed world" that the program is talking about
is exactly the initial model. (For those not familiar with
this terminology, the initial model is actually
characterized (for the usual logics) uniquely up to
isomorphism by the property that only what is provable is
true; it is closely related to the Herbrand universe.
It has very recently been proven (by Tarlecki, and by Mahr
and Makowsky) that the largest sublanguage of first order
logic such that all sets of sentences have initial models
is Horn clause logic. This result extends to many-sorted
logic and/or logic with equality. So that gives us a
good idea of just how far we can go with logic programming
and still provide the programming with a good idea of what
his program is about (namely, whatever is in the initial model).
------------------------------
End of PROLOG Digest
********************
∂08-Apr-85 0914 EMMA@SU-CSLI.ARPA Mailing lists
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 8 Apr 85 09:14:05 PST
Date: Mon 8 Apr 85 09:11:24-PST
From: Emma Pease <Emma@SU-CSLI.ARPA>
Subject: Mailing lists
To: folks@SU-CSLI.ARPA
cc: bboard@SU-CSLI.ARPA
Tel: 497-3479
Several reminders about the mailing lists.
1) The friends mailing list is open to all people with an interest in
the RESEARCH done at CSLI; hence items on it should be of interest to
those people. The friends mailing list is also monitored (loosely).
2) The CSLI bboard is on the friends list so do NOT send to both the
bboard and friends.
3) The folks mailing list is for those directly involved in CSLI as employees
or researchers. It overlaps with the friends mailing list; hence messages
should be sent either to folks OR friends but not both (those who are on
folks but not on friends generally read friends mail on the bboard).
4) Bboard is open to any message with good taste.
-Emma Pease
ps. I hope everyone enjoyed the lovely weekend we just had.
-------
∂08-Apr-85 1019 @SU-CSLI.ARPA:BrianSmith.pa@Xerox.ARPA RRR Meeting, 4/9
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 8 Apr 85 10:19:33 PST
Received: from Xerox.ARPA by SU-CSLI.ARPA with TCP; Mon 8 Apr 85 10:16:16-PST
Received: from Semillon.ms by ArpaGateway.ms ; 08 APR 85 10:16:06 PST
Date: 8 Apr 85 10:15 PST
From: BrianSmith.pa@Xerox.ARPA
Subject: RRR Meeting, 4/9
To: RRR@SU-CSLI.ARPA
cc: Sugai.pa@Xerox.ARPA
Due to conflicts in scheduling times for the Trailer Seminar Room, the
time for tomorrow's (4/9/85) RRR meeting has been changed to 2:15 p.m. -
4:15 p.m.
∂08-Apr-85 1100 BRODER@SU-SCORE.ARPA First spring AFLB
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 8 Apr 85 11:00:11 PST
Date: Mon 8 Apr 85 10:48:32-PST
From: Andrei Broder <Broder@SU-SCORE.ARPA>
Subject: First spring AFLB
To: aflb.all@SU-SCORE.ARPA
cc: tuley@SU-SCORE.ARPA
4/11/85 - Dr. Peter van Emde Boas, (IBM-San Jose and Univ. of Amsterdam)
"Two results on tables"
We consider the problem of storing arbitrary k-element subsets A of an
u-element Universe U . Elements can be stored in a table of s slots
where every slot can contain an arbitrary member of the universe U.
Andy Yao has investigated the maximal size of the universe for which
it is possible to test membership in A in a single probe in the table,
assuming that s = k and that only elements in A are stored. He
obtained the upper bound u <= 2k-2 . If we relax the condition that
only elements in A are stored in the table this bound is not valid.
For example k=4 and u=16 is solved by storing a bit table
representation in the table. Our first result shows that this way of
cheating against Yao's model is essentially the only possibility: if a
single probe strategy exists at all for some pair of sizes u and k
there is sufficient space in the table to encode a bit table
representation of arbitrary subsets of U .
Our second result deals with the subject of perfect hashing into a
hash table of linear size. The existence of perfect hash functions
which achieve this goal was shown by Fredman, Komlos & Szemeredi in
their 1982 FOCS paper. We investigate the space requirements of these
perfect hash functions. The space required by the functions as
described in their 1982 paper equals O(k.log(u)) . This was improved
by the authors together with Kurt Melhorn to O(k.log(k) + loglog(u)) .
Mehlhorn also proved an O(k + loglog(u)) lowerbound. In our 1984 STOC
paper we showed that space O(k + loglog(u)) for a perfect hash
function can be achieved at the price of a linear time evaluation
algorithm for the hash function - the O(k.log(k) + loglog(u)) bound is
achieved for a hash function which can be evaluated in linear time.
In our talk we show that O(1) evaluation time can be obtained for a
perfect hash function requiring space O(k.loglog(k) + loglog(u)) .
These two results were obtained jointly with Christiaan T.M. Jacobs ,
a student at the Univ. of Amsterdam.
***** Time and place: April 11, 12:30 pm in MJ352 (Bldg. 460) ******
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
Regular AFLB meetings are on Thursdays, at 12:30pm, in MJ352 (Bldg.
460).
If you have a topic you would like to talk about in the AFLB seminar
please let me know. (Electronic mail: broder@su-score.arpa, phones:
(415) 853-2118 and (415) 948-9172). Contributions are wanted and
welcome. Not all time slots for this academic year have been filled
so far.
For more information about future and past AFLB meetings and topics
you might want to look at the file [SCORE]<broder>aflb.bboard .
- Andrei Broder
-------
∂08-Apr-85 1133 @SU-CSLI.ARPA:PENTLAND@SRI-AI.ARPA Talk by Fanya Montalvo
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 8 Apr 85 11:33:30 PST
Received: from SRI-AI.ARPA by SU-CSLI.ARPA with TCP; Mon 8 Apr 85 11:29:15-PST
Date: Mon 8 Apr 85 11:19:34-PST
From: PENTLAND@SRI-AI.ARPA
Subject: Talk by Fanya Montalvo
To: folks@SU-CSLI.ARPA
Area P1 Talk --
WHERE: SRI Int'l Room EK242 (conference room)
WHEN: Tues April 9 (tommorrow) at 2:30 (sorry for the short notice)
DIAGRAM UNDERSTANDING:
THE INTERSECTION OF COMPUTER GRAPHICS AND COMPUTER VISION
Fanya S. Montalvo
MIT, Artificial Intelligence Laboratory
ABSTRACT
A problem common to Computer Vision and Computer Graphics is
identified. It deals with the representation, acquisition, and
validation of symbolic descriptions for visual properties. The
utility of treating this area as one is explained in terms of
providing the facility for diagrammatic conversations with systems. I
call this area "Diagram Understanding", which is analogous to Natural
Language Understanding. The recognition and generation of visual
objects are two sides of the same symbolic coin. A paradigm for the
discovery of higher-level visual properties is introduced, and its
application to Computer Vision and Computer Graphics described. The
notion of denotation is introduced in this context. It is the map
between linguistic symbols and visual properties. A method is
outlined for associating symbolic descriptions with visual properties
in such a way that human subjects can be brought into the loop in
order to validate (or specify) the denotation map. Secondly, a way of
discovering a natural set of visual primitives is introduced.
-------
∂08-Apr-85 1724 REGES@SU-SCORE.ARPA Questions and Answers about SU-SUSHI
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 8 Apr 85 17:23:36 PST
Date: Mon 8 Apr 85 17:10:24-PST
From: Stuart Reges <REGES@SU-SCORE.ARPA>
Subject: Questions and Answers about SU-SUSHI
To: csd@SU-SCORE.ARPA
cc: su-bboards@SU-SCORE.ARPA
Office: Margaret Jacks 210, 497-9798
I have been asked many questions about SUSHI. I'm tired of answering
individually, so I'll answer in bulk:
Q: Who can get a SUSHI account?
A: Stanford students who are either: a CS major; a CS TA or instructor;
an affiliated student (e.g., in CSL or MIS).
Q: Who has SUSHI accounts already?
A: Everyone that I could find in our records as a CS student or TA. I
have sent electronic mail to those people informing them of their
account names and passwords.
Q: What if a person deserves a SUSHI account but didn't get it?
A: If you are a CS student or TA and didn't get an account, let me
know. I'll try to remedy that as soon as possible. If you are an
affiliated student, you should talk to a faculty member to get
sponsorship. I will develop a faculty sponsorship form in the next
few days that I will distribute to faculty. Please don't talk to me
before the sponsorship form has been made up. Probably a week from
now I'll start opening accounts for such students.
Q: What can SUSHI be used for?
A: We aren't making strict usage rules right now. SUSHI is supposed to
be an academic machine. You can use it to do coursework that you
would normally do at LOTS. You can use it to do unsupported
research. You can use it to have fun. The only thing we are
prohibiting right now is sponsored research--your Principal
Investigator should give you an account on another machine for this.
Q: What if a person wants to change his account name or set up a mail
forwarding entry?
A: Please wait a week until Bill Westfield and I have the system in
fairly stable condition. Then we will entertain requests like
this. We will post announcements as system messages on SUSHI.
Q: How stable is SUSHI?
A: The machine is still going through birth pains. There is no file
backup yet. But we believe the machine will be very stable quickly.
Len and company are working on it.
Q: What will happen to SCORE and other accounts?
A: The Department is sponsoring many student accounts on SCORE. These
accounts will eventually go away. I hope to close them all on May
1st, depending upon how quickly SUSHI stabalizes. I will give at
least two weeks notice. When I do close the accounts on SCORE I
will create mail forwarding entries that forward mail to SUSHI and I
will update all of the departmental mailing lists that I control to
send mail to the SUSHI accounts rather than the SCORE accounts. I
will not close accounts not sponsored by the Department. The
sponsors of those accounts will have to make a decision about that.
Q: What about PEDIT and LOOKUP?
A: We are going to bring up a version of PEDIT on SUSHI. Every PEDIT
entry will have a "home system." You will only be allowed to change
your people entry on your home system. At night the two databases
will communicate in order to update entries. LOOKUP will work the
same on both systems, except for the daily updates. We expect that
eventually students will use PEDIT on SUSHI rather than SCORE. More
details will be announced as they are worked out.
-------
∂09-Apr-85 0029 @SU-SCORE.ARPA:FEIGENBAUM@SUMEX-AIM.ARPA Bob Engelmore returns
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 9 Apr 85 00:27:06 PST
Received: from SUMEX-AIM.ARPA by SU-SCORE.ARPA with TCP; Tue 9 Apr 85 00:23:52-PST
Date: Tue 9 Apr 85 00:20:28-PST
From: Edward Feigenbaum <FEIGENBAUM@SUMEX-AIM.ARPA>
Subject: Bob Engelmore returns
To: faculty@SU-SCORE.ARPA, ksl@SUMEX-AIM.ARPA, welch-road@SUMEX-AIM.ARPA
cc: feigenbaum@SUMEX-AIM.ARPA, engelmore@SUMEX-AIM.ARPA
Today Bob Engelmore rejoined the HPP, and we're very glad to have him back.
He takes on the new job of Executive Director of the HPP, in which role he
will both do and manage research at the HPP. Bob will have the responsibility,
and the necessary authority, to insure that the affairs of the HPP run
smoothly.
My own title will be Research Director (and PI of the grants and contracts of
the HPP).
Bob's return is good news for all of us who value his scientific talents,
his collaboration on projects, his unending good humor and cheerfulness, and
his warm friendship.
Ed Feigenbaum
-------
∂09-Apr-85 0044 RESTIVO@SU-SCORE.ARPA PROLOG Digest V3 #17
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 9 Apr 85 00:43:50 PST
Date: Monday, April 8, 1985 8:16PM
From: Chuck Restivo (The Moderator) <PROLOG-REQUEST@SU-SCORE.ARPA>
Reply-to: PROLOG@SU-SCORE.ARPA
US-Mail: P.O. Box 4584, Stanford CA 94305
Phone: (415) 326-5550
Subject: PROLOG Digest V3 #17
To: PROLOG@SU-SCORE.ARPA
PROLOG Digest Tuesday, 9 Apr 1985 Volume 3 : Issue 17
Today's Topics:
Implementations - Searching & C-Prolog & Cuts,
& Denotational Semantics
----------------------------------------------------------------------
Date: Sun, 7 Apr 85 18:41:41 pst
From: (Paul Hilfinger) Hilfingr%ucbrenoir@Berkeley
Subject: Searching for best
Suppose that p/1 is defined so that p(X) unifies, in
turn, to all terms in some finite set. Suppose that
if X is instantiated to a term s.t. p(X), then
cost(X,N) instantiates N to some non-negative integer
cost measure of the term X. Define the predicate best/1
such that best(X) unifies to all terms such that p(X)
where X has minimal cost. I'm looking for the most
elegant and efficient general solution that does NOT
use assert, retract, findall, bagof, or things of that
ilk (cut is OK.) I do have a solution, but I'm sure
that it isn't optimal.
-- Paul Hilfinger
------------------------------
Date: 7 Apr 1985 23:10:46 PST
From: Mike Newton <Newton@CIT-20.ARPA>
Subject: C-Prolog on an IBM
About 10 (?) issues ago i sent in an article describing
the changes to C-Prolog 1.5 to get it running on an IBM
mainframe. Anybody that has done this change is requested
to run the following Prolog 'programs':
?- 4.1 < 2.
?- 4 + 7 < 2.
If either cause a problem, send me a note for a long (~ two
page) fix.
[This bug only shows up as a result of some C compilers.]
-- Mike
Caltech 256-80
Pasadena CA 91125
818 356 6771
------------------------------
Date: Fri, 5 Apr 85 17:21 EST
From: Mark Beutnagel <Beutnagel%upenn.csnet@csnet-relay.arpa>
Subject: Another C-Prolog Port
I have ported CProlog v1.5.EdCAAD from VAX to a M68000
based HP workstation,the HP 9000 Series 200. It is running
under HP-UX v2.1, which is basically System III Unix plus
extras from 4.1 and 4.2 BSD.
Only minor changes were required, mainly in the makefile
and in parms.c, to use the correct #defines and startup
file. A record of these changes can be made available to
interested parties.
All seems to be running well with the exception of the
following message from the save predicate: "Not a
typewriter". In spite of this and [execution aborted]
from Prolog, the reloaded files work (so far).
Apparently save tries to do some terminal IO to the file
and Unix objects, but I haven't found the culprit code yet.
If anyone has seen or fixed this problem, suggestions
would be gratefully recieved.
Regarding Mike Newton's bug:
[ f(Y) :- (true,! ; X=0),Y=<7. ]
this version exhibits no symptoms even though pl/init uses
$call rather than
$hidden←call to define (A;B).
-- Mark Beutnagel
------------------------------
Date: 3 Apr 85 01:19:37 GMT
From: (David Powers)decvax!mulga!elecvax.oz!DavidP@Berkeley
Subject: "soft" cuts
Coming in on a response to a query I missed concerning
"soft" cuts:
I added such a construct to UNSW PROLOG in 1982 - it is
a very simple change, related closely to traditional cut.
I used the symbol "#", hash, and the term "half cut".
An example of its use compared with alternative ways of
achieving the same end is given in Chapter 5, pp19-28 of
UNSW DCS Technical Report 8313, D.M.W.Powers and G.B.McMahon,
"A Compendium of Interesting PROLOG programs" (December 1983).
The advantage of half cut is that it allows obtaining all
solutions, if any exist, and execution of an alternative
strategy or error routine otherwise. I have not found any
markedly different application. My original application
was natural language parsing/learning.
This half cut cannot be simulated with Horn + cut
(without assert/bagof/...), but the other half - commit to
current track within clause but not to current clause -
can be simulated with cut alone.
I am happy to look out the specific mods to UNSW PROLOG for
those who request them.
-- David M. W. Powers
------------------------------
Date: Mon 8 Apr 85 19:33:03-MST
From: Uday Reddy <U-REDDY@UTAH-20.ARPA>
Subject: Logic Programming Semantics
I agree with Joseph Goguen in principle. But, I think
his proposal is a bit too idealistic.
First of all, my notion of denotation semantics is very
broad. I would consider any specification of semantics
that is (1) compositional and (2) complete to be
denotational semantics. It does'nt have to be written
in the Oxford style, or deal with infinite structures.
Denotations don't also have to be functions. Functions
just turn out to be the most convenient denotations most
of the time.
When Josephs says "the good thing about a real logic
programming language is exactly that it does not need
such a [denotational] semantics", I take it to mean that
we already know its denotational semantics. Well, this
does not hold most of the time. For instance, pure Horn
clauses don't satisfy this, as they work in the
{true, unsepcified} domain as opposed to the good old
logic we know, which works in a {true, false} domain.
Are Horn clauses then "well understood"?
I see nothing wrong with Pure Prolog, for instance,
which uses sequential "and" and "or" operations in a
3-point {true, unspecified, fail} domain. Even though
its semantics may not have been "well understood" in
classical logic, the pragmatics of computing make it
necessary and it is simple enough.
Pragmatics of computing require us to move away from or
add to what we import from other disciplines such as
logic or mathematics. This is not necessarily wrong. I
can't see anything other than the subjective notion of
"simplicity" to guide us in deciding what is good or
bad.
Pragmatics of computing is also what makes languages
which work in initial models (the models which precisely
describe the inferrable equality) impractical. Testing
for equality of two terms t and u requires m.n
comparisons where m and n are the number of reduction
steps for t and u respectively. On the other hand,
continuous equality in the least model (which is what is
used in functional languages) requires only 1
comparison. So, if continuity requires learning a good
amount of computer science's own (not logicians')
theory, it is certainly worth it.
The dangerous attitude in logic programming community is
the step-motherly treatment given to "control". As if
control does not affect the meanings of programs or as
if we don't need to understand it. This is what I am
objecting to. Suppose someone said FORTRAN was a
language of well understood arithmetic expressions.
Everything else in it was just "control"!
-- Uday Reddy
------------------------------
End of PROLOG Digest
********************
∂09-Apr-85 1010 JOHN@SU-CSLI.ARPA talk on tacit knowledge by Martin Davies
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 9 Apr 85 10:10:02 PST
Date: Tue 9 Apr 85 10:05:48-PST
From: John Perry <JOHN@SU-CSLI.ARPA>
Subject: talk on tacit knowledge by Martin Davies
To: folks@SU-CSLI.ARPA
What: Talk and Discussion
Topic: Tacit Knowledge: Subdoxasticity and Modularity
Who: Martin Davies, Birkbeck College, London
When: 10:15 (a.m.), Tuesday, April 16
Where: Ventura Seminar Room (or other Ventura location
if complications arise)
-------
∂09-Apr-85 1112 FORSTER@SU-CSLI.ARPA Phil. Dept. Colloquium
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 9 Apr 85 11:11:58 PST
Date: Tue 9 Apr 85 10:58:11-PST
From: Eckart Forster <FORSTER@SU-CSLI.ARPA>
Subject: Phil. Dept. Colloquium
To: folks@SU-CSLI.ARPA
PHILOSOPHY DEPARTMENT COLLOQUIUM
Speaker: Edwin McCann (USC)
Title: "Cartesian Selves and Lockean Substances"
Time: Friday, April 12, 3:15
Place: Philosophy Seminar Room 90-92Q
-------
∂09-Apr-85 1208 @SU-CSLI.ARPA:PCOHEN@SRI-AI.ARPA reminder
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 9 Apr 85 12:08:09 PST
Received: from SRI-AI.ARPA by SU-CSLI.ARPA with TCP; Tue 9 Apr 85 12:05:13-PST
Date: Tue 9 Apr 85 12:04:47-PST
From: Phil Cohen <PCOHEN@SRI-AI.ARPA>
Subject: reminder
To: NL4: ;
Stephen Neale will talk about "Utterance Situations" today at 12:45
Phil
-------
∂09-Apr-85 1352 WRIGHT@SUMEX-AIM.ARPA siglunch
Received: from SUMEX-AIM.ARPA by SU-AI.ARPA with TCP; 9 Apr 85 13:52:28 PST
Date: Tue 9 Apr 85 13:38:57-PST
From: Carol Wright <WRIGHT@SUMEX-AIM.ARPA>
Subject: siglunch
To: Siglunch@SUMEX-AIM.ARPA
Folks,
There will be no Siglunch this Friday.
Carol
-------
∂09-Apr-85 1526 TULEY@SU-SCORE.ARPA room reservations
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 9 Apr 85 15:22:02 PST
Date: Tue 9 Apr 85 15:00:59-PST
From: Kimberly Tuley <TULEY@SU-SCORE.ARPA>
Subject: room reservations
To: csd@SU-SCORE.ARPA
This is a reminder to everyone:
Please reserve the seminar rooms through Kim at the reception
desk - the signout sheet on doors of 301, 352, 252 are only a rough
schedule put up on Monday mornings ... not where you reserve the room.
thanks very much, Kim
-------
∂09-Apr-85 1715 CLT Seminar in Logic and Foundations of Mathematics
To: "@DIS.DIS[1,CLT]"@SU-AI.ARPA
Speaker: Prof. Jon Barwise
Title: On the model theory of shared information
Time: Tuesday, April 16, 1985 at 4:15-5:30 P.M.
Place: Room 381-T, Math Corner, Stanford
Abstract:
The traditional model-theoretic approach to the problem of
shared understanding (public information, common knowledge, mutual
belief) has been through an iterated hierarchy of attitude reports (c
knows that b knows ... that c knows that P), mirroring the iterated
hierarchy in set theory and higher-order model theory. In this talk I
want to show that Aczel's work on non-wellfounded sets gives us a new
tool for a "direct" model-theoretic approach through situations. I
will go on to state some approximation theorems that show to what
extent the hierarchy approach does and does not add up, in the limit,
to the direct approach. The results raise a number of interesting
model-theoretic questions that only arise in the context of
non-wellfounded sets.
S. Feferman
∂10-Apr-85 0022 @SU-SCORE.ARPA:ARK*@SU-AI.ARPA CS Colloq, April 16: Vaughan Pratt, Graphics for User Interfaces
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 10 Apr 85 00:22:32 PST
Received: from SU-AI.ARPA by SU-SCORE.ARPA with TCP; Wed 10 Apr 85 00:01:15-PST
Date: 10 Apr 85 0000 PST
From: Arthur Keller <ARK@SU-AI.ARPA>
Subject: CS Colloq, April 16: Vaughan Pratt, Graphics for User Interfaces
To: All-Colloq@SU-SCORE.ARPA, Faculty@SU-SCORE.ARPA
CC: ARK@SU-AI.ARPA
CS Colloquium, April 16, 4:15pm, Terman Auditorium
GRAPHICS FOR USER INTERFACES
Vaughan Pratt
Computer Science Dept.
Stanford University
Responsive user interfaces need high performance high quality graphics
support. I shall describe recent research at Sun addressing the
graphics needs of text, splines, and windows.
----
Cookies, thanks to Ginger, and possibly juice will probably be served in
the 3rd floor lounge of MJH at 3:45pm. See you there.
[This reminder is now expired.]
∂10-Apr-85 0955 ullman@diablo meeting
Received: from SU-AIMVAX.ARPA by SU-AI.ARPA with TCP; 10 Apr 85 09:51:21 PST
Date: Wed, 10 Apr 85 09:50:23 pst
From: Jeff Ullman <ullman@diablo>
Subject: meeting
To: nail@diablo
Don't forget we meet 11AM today in 301.
Moshe is covering the Vassilou et al. paper.
(tableau optimization applied to Prolog)
∂10-Apr-85 1632 chertok%ucbcogsci@Berkeley UCB Cognitive Science Seminar--April 16
Received: from UCB-VAX.ARPA by SU-AI.ARPA with TCP; 10 Apr 85 16:31:49 PST
Received: from ucbcogsci.ARPA by UCB-VAX.ARPA (4.24/4.45)
id AA25273; Wed, 10 Apr 85 16:29:25 pst
Received: by ucbcogsci.ARPA (4.24/4.38)
id AA17956; Wed, 10 Apr 85 16:31:59 pst
Date: Wed, 10 Apr 85 16:31:59 pst
From: chertok%ucbcogsci@Berkeley (Paula Chertok)
Message-Id: <8504110031.AA17956@ucbcogsci.ARPA>
To: cogsci-friends%ucbcogsci@Berkeley
Subject: UCB Cognitive Science Seminar--April 16
BERKELEY COGNITIVE SCIENCE PROGRAM
Cognitive Science Seminar -- IDS 237B
TIME: Tuesday, April 16, 11 - 12:30
PLACE: 240 Bechtel Engineering Center
(followed by)
DISCUSSION: 12:30 - 1:30 in 200 Building T-4
SPEAKER: Mark Johnson, Philosophy Department, Southern
Illinois University
TITLE: ``The Bodily Basis of Meaning and Imagination''
The idea that human rationality is an abstract, disembodied,
formal structure is deeply rooted in the Western Philosophical
tradition and is manifested most recently in model-theoretic and
Davidsonian semantics. According to this view, meaning is an
abstract relation between symbolic representations (either words
or mental representations) and objective (mind-independent) real-
ity. Meaning is thus a matter of objective senses and has nothing
to do with how human beings understand their experience. And
rationality is a rule-governed manipulation of the symbols that
express meaning. In this whole picture nothing is said about the
role of bodily experience, either in the emergence of meaning or
in our reasoning about our world.
But it is a fact that we humans ←λd←λo have bodies, and it would
be rather strange if this fact didn't have some important bearing
on what we experience as meaningful and how we make sense of our
world in a rational fashion. I suggest that there are recurrent
preconceptual structures in our bodily interactions with our
environment that are the basis for human meaning. These are
structures of our perceptual activity and bodily movements that
are metaphorically extended to structure more abstract, non-
physical domains. So I am claiming that our more `abstract' rea-
soning is grounded in a concrete reasoning via metaphorical con-
nections. My argument is based on an analysis of the experience
and meaning of ←λb←λa←λl←λa←λn←λc←λe.
-------------------------------------------------------------
UPCOMING TALKS
April 23: David Dowty, Center for Advanced Study in the Behavioral
Sciences
April 30: Herbert H. Clark, Psychology Department, Stanford Univ.
-------------------------------------------------------------
∂10-Apr-85 1740 EMMA@SU-CSLI.ARPA Newsletter Apr. 11, No. 24
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 10 Apr 85 17:39:52 PST
Date: Wed 10 Apr 85 17:26:31-PST
From: Emma Pease <Emma@SU-CSLI.ARPA>
Subject: Newsletter Apr. 11, No. 24
To: friends@SU-CSLI.ARPA
Tel: 497-3479
C S L I N E W S L E T T E R
←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←
April 11, 1985 Stanford Vol. 2, No. 24
←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←
A weekly publication of The Center for the Study of Language and
Information, Ventura Hall, Stanford University, Stanford, CA 94305
←←←←←←←←←←←←
CSLI ACTIVITIES FOR *THIS* THURSDAY, April 11, 1985
12 noon TINLunch
Ventura Hall ``Semantics for Natural Language: Metaphysics
Conference Room for the Simple-minded?''
Chris Menzel, CSLI
2:15 p.m. CSLI Seminar
Redwood Hall ``What if the World Were Really Quite Simple?''
Room G-19 Alex Pentland, CSLI
Discussion led by Jerry Hobbs, SRI International
3:30 p.m. Tea
Ventura Hall
4:15 p.m. CSLI Colloquium
Redwood Hall ``A Formal Theory of Innate Linguistic Knowledge''
Room G-19 Janet Dean Fodor, University of Connecticut and CSLI
←←←←←←←←←←←
CSLI ACTIVITIES FOR *NEXT* THURSDAY, April 18, 1985
12 noon TINLunch
Ventura Hall A. P. Martinich's ``A Theory for Metaphor''
Conference Room Discussion led by Paul Schacht
(Abstract on page 2)
2:15 p.m. CSLI Seminar
Redwood Hall Title to be announced
Room G-19 Brian Smith, Xerox PARC and CSLI
Discussion led by Stan Rosenschein
3:30 p.m. Tea
Ventura Hall
4:15 p.m. CSLI Colloquium
Redwood Hall ``Two Examiners Marked Six Papers: Interpretations
Room G-19 of Numerically Quantified Sentences''
Martin Davies, Birkbeck College, U. of London
NEW DIRECTOR
Jon Barwise, CSLI's first Director, stepped down on April 1 in
order to devote more time to research. John Perry, the newly endowed
Henry Waldgrave Stuart professor of philosophy at Stanford, succeeds
him.
!
Page 2 CSLI Newsletter April 11, 1985
←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←
ABSTRACT OF NEXT WEEK'S TINLUNCH
Much work has been done on the concept of metaphor, but most of
this work does not place metaphor within a general theory of language
or language use. So argues A. P. Martinich in his article ``A Theory
for Metaphor.'' Martinich attempts to explain metaphor in terms of
Grice's theory of conversation, maintaining that metaphor is
pragmatically rather than semantically based and that, while ``there
is a sense in which the sentence used metaphorically has a
metaphorical meaning, this meaning is itself a consequence of the
mechanisms that give rise to the metaphor and are not what makes the
metaphor possible.'' We will use Martinich's assertions as a point of
departure-- as it were--for a general discussion of metaphor.
←←←←←←←←←←←← --Paul Schacht
ABSTRACT OF NEXT WEEK'S COLLOQUIUM
``Two Examiners Marked Six Papers''
Interpretations of numerically quantified sentences
Numerically quantified sentences, such as
(1) Two examiners marked six scripts
admit of several different readings. In ``Ambiguity and Quantification'',
Ruth Kempson and Annabel Cormack proposed four interpretations to be
derived from a ``single semantic representation''. I begin with a brief
exposition of their proposal, and raise several questions about it.
My main aim is to present an alternative semantic proposal. After
a brief glance at the distributive reading of sentences with just one
numerical quantifier, I move to the group or collective reading. Here
I rely on work by Barry Taylor on articulated predication. This is
related to Adam Morton's multigrade relations, and Richard Grandy's
anadic logic.
Iterated deployment of the semantic resources used for the
distributive and collective readings of very simple sentences
provides, in principle, for eight readings of a sentence like (1).
But some of the readings turn out to be equivalent, and the pattern of
equivalences varies with different choices of binary predicate in
place of ``marked''. After comparison of these readings with those
proposed by Kempson and Cormack, a branching quantifier representation
is proposed for the so-called complete group interpretation. I
conclude with some reflections on the questions raised at the outset.
←←←←←←←←←←←← --Martin Davies
LOGIC SEMINAR
``On the Model Theory of Shared Information''
Jon Barwise, CSLI
April 16, at 4:15, Room 381 T (Math Corner)
The traditional model-theoretic approach to the problem of shared
understanding (public information, common knowledge, mutual belief)
has been through an iterated hierarchy of attitude reports (c knows
that b knows ... that c knows that P), mirroring the iterated
hierarchy in set theory and higher-order model theory. In this talk I
want to show that Aczel's work on non-wellfounded sets gives us a new
tool for a ``direct'' model-theoretic approach through situations. I
will go on to state some approximation theorems that show to what
extent the hierarchy approach does and does not add up, in the limit,
to the direct approach. The results raise a number of interesting
model-theoretic questions that only arise in the context of
non-wellfounded sets.
!
Page 3 CSLI Newsletter April 11, 1985
←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←
PSYCHOLOGY DEPARTMENT SEMINAR
``Morphological & Prosodic Cues in the Learning
of a Miniature Phrase-Structure Language''
Richard Meier, Stanford
April 12, 3:15pm, Jordan Hall, Rm. 100
I will claim that the input to language learning is a grouped and
structured sequence of words and that learning operates most
successfully on such structures, and not on mere word strings. After
briefly reviewing evidence for such groupings in natural language,
this claim will be supported by three experiments in artificial
language learning. These experiments allow rigorous control of the
input to the learner. Prior work had argued that, in such
experiments, adult subjects can learn complex syntactic rules only
with extensive semantic mediation. In the current experiments,
subjects fully learned complex aspects of syntax if they viewed, or
heard, sentences (paired with an uninformative semantics) containing
one of three grouping cues for constituent structure: prosody,
function words, or agreement suffixes on the words within a
constituent. Absent such cues, subjects learned only limited aspects
of syntax. These results suggest that, in natural languages, such
grouping cues may subserve syntax learning.
←←←←←←←←←←←←
CSLI SEMINAR
``Tacit Knowledge: Subdoxasticity and Modularity''
Martin Davies, Birkbeck College, London
10:15, Tuesday, April 16, Ventura Seminar Room
←←←←←←←←←←←←
CONFERENCE ON EVOLUTION AND INFORMATION
A conference on Evolution and Information with major support from
CSLI will be held at Stanford this April 19-21. The specific focus of
the conference will be on the use of optimality models both in biology
and in the human sciences. Papers will be contributed to the
conference by biologists, philosophers, psychologists, and
anthropologists. Apart from addressing problems and limitations of
optimality models within biology, an important aim of the conference
will be to explore the relevance of biological results, either
factually or methodologically, to other areas of inquiry.
Contributors will be asked to give a brief summary of their papers
at the conference sessions but papers will not be read. For further
information about the conference contact John Dupre, Philosophy,
Stanford University (415-497-2587, Dupre@Turing).
←←←←←←←←←←←←
PHILOSOPHY COURSE
The seminar ``Nonexistent Objects and the Semantics of Fiction''
will now be meeting regularly on Tuesdays from 12:30 - 2:15 in the
Ventura Trailers Conference Room. The course, though listed in the
Philosophy Department, will satisfy requirements for the formal
systems major. --Ed Zalta
!
Page 4 CSLI Newsletter April 11, 1985
←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←
WORKSHOP ON FINITE STATE MORPHOLOGY
CSLI, Stanford July 29-30, 1985
In the area of syntax there has been for a long time a connection
between linguistics and computer science. Mathematical and
computational issues are often raised in connection with certain kinds
of syntactic problems. Some concepts, such as unification, that have
their origins in computer science have been added to the linguistic
vocabulary as a result of this interaction.
In phonology and morphology, the situation has so far been
different. In this domain, descriptive and theoretical work for the
most part has proceeded without parallel mathematical and
computational effort. It appears that this situation is about to
change. There is a great deal of new activity in computational
morphology that stems from yet unpublished work by Martin Kay and
Ronald Kaplan on implementing phonological rules as finite state
transducers. Because the use of finite state devices is the central
idea that characterizes this approach, it seems appropriate to talk
about FINITE STATE MORPHOLOGY. One major piece of work in this line
of research is Kimmo Koskenniemi's recent dissertation on Two-level
Morphology. There are many features in current phonology that are
missing from implementations of Finite State Morphology that have been
built so far. It has not been shown that all relevant phenomena can
be handled in a satisfactory way by finite state means.
Given this state of affairs, the stage is set for useful exchanges
between theoretical and descriptive phonologists, computer scientists,
and linguists who are working on computational morphology. We are
planning a workshop on Finite State Morphology in Palo Alto under the
auspices of CSLI. The dates for the workshop are July 29-30. Among
the topics that we expect to discuss are the following:
- Points of friction between Finite State approaches to phonology
and linguistic theory; phenomena that present fundamental problems.
- New ideas within this framework.
- Descriptive work on particular languages.
- Representation of rules as transducers, compilation.
- Mathematical properties of rule systems. Send comments and
inquiries to Lauri Karttunen (LAURI@SU-CSLI.ARPA).
←←←←←←←←←←←←
PANEL DISCUSSION ON SYNTACTIC THEORIES
1:30, Tuesday, April 16, Redwood G-19
The lecturers in our series on syntactic theories for non-linguists
are back, this time in a panel discussion. Joan Bresnan, Geoff
Pullum, and Peter Sells will take questions from the audience (no
initial presentations, so come with questions). This time, linguists
ARE allowed, but they are asked to stick to matters that non-linguists
will have a chance of understanding. If you think of questions in
advance, send them to the panelists so that they can think about them
(BRESNAN@SU-CSLI,PULLUM@SU-CSLI,SELLS@SU-CSLI).
-------
∂10-Apr-85 1757 EMMA@SU-CSLI.ARPA Newsletter
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 10 Apr 85 17:56:55 PST
Date: Wed 10 Apr 85 17:54:07-PST
From: Emma Pease <Emma@SU-CSLI.ARPA>
Subject: Newsletter
To: folks@SU-CSLI.ARPA
Tel: 497-3479
A reminder about deadlines.
1) The deadline for the newsletter is noon of the Wednesday before it
comes out i.e. if you want something to appear in April 11's edition, the
article must be in by noon, April 10.
2) Information about Thursday's activities appear in the previous Thursday's
newsletter i.e., Thursday abstracts and titles must be received a week and
a day before they happen.
3) IF you can't make the noon deadline, send me a message before noon telling
me WHEN I can expect the information and then send it before 3 that afternoon.
Many thanks,
Emma
ps. I prefer to receive announcements online, if possible.
-------
∂10-Apr-85 2039 ullman@diablo Week of 4/22
Received: from SU-AIMVAX.ARPA by SU-AI.ARPA with TCP; 10 Apr 85 20:39:17 PST
Date: Wed, 10 Apr 85 20:38:46 pst
From: Jeff Ullman <ullman@diablo>
Subject: Week of 4/22
To: nail@diablo, wiederhold@sumex
David Maier will talk to the group on FRIDAY 4/26 at 11AM in 252 MJH.
I will then give the CS345 seminar at 3:15PM in 352MJH.
We will have our usual 11AM Wednesday 4/24 meeting in 301 MJH,
with no speaker, but a discussion of how NAIL! ought to be designed.
∂10-Apr-85 2246 @SU-SCORE.ARPA:udi@wisc-rsch.arpa Preliminary Program for PODC 85
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 10 Apr 85 22:46:15 PST
Received: from wisc-rsch.arpa by SU-SCORE.ARPA with TCP; Wed 10 Apr 85 22:41:43-PST
Received: by wisc-rsch.arpa; Thu, 11 Apr 85 00:25:58 cst
Message-Id: <8504110547.AA28889@wisc-rsch.arpa>
Received: from csnet-relay.arpa by wisc-rsch.arpa; Wed, 10 Apr 85 23:47:17 cst
Received: from ibm-sj by csnet-relay.csnet id ak00218; 10 Apr 85 19:32 EST
Date: Tue, 9 Apr 85 15:02:51 PST
From: Ray Strong <strong%ibm-sj.csnet@csnet-relay.arpa>
To: udi@wisc-rsch.ARPA
Subject: Preliminary Program for PODC 85
Sender: udi@wisc-rsch.arpa
Resent-From: udi@wisc-rsch.arpa (Udi Manber)
Resent-To: theory:;@wisc-rsch.arpa
Tentative Program
ACM Symposium on Principles of Distributed Computing
August 5-7, 1985
Minacki, Ontario, CANADA
======================================================================
Monday, 5 August
Invited Speaker
9:00 am to 10:00 am Butler Lampson
Session I Logic Chair: Amir Pnueli
10:30 am to 12:30 pm
Symmetry and Similarity, Johnson and Schneider
On the Analysis of Cooperation and Antagonism in
Networks of Communicating Processes, Kanellakis and Smolka
Characterization of Safety and Liveness Properties
in Temporal Logic, Sistla
A Model and Proof System for Asynchronous
Networks, Jonnson
Session II Byzantine Chair: Cynthia Dwork
2:00 pm to 3:30 pm
Shifting Scenarios: Easy Impossibility Proofs for Distributed
Consensus Problems, Fischer, Lynch, and Merritt
Optimal Clock Synchronization, Srikanth and Toueg
Fast Distributed Agreement, Toueg, Perry, and Srikanth
Session III Database Chair: Paul Leach
4:00 pm to 5:30 pm
Comparing How Atomicity Mechanisms Support Replication,
Herlihy
Site-Optimal Termination Protocols for a Distributed Database
under Networking Partitioning, Cheung and Kameda
Distributed Version Management for Read-Only Actions,
Weihl
======================================================================
Tuesday, 6 August
Session IV Coins Chair: Joe Halpern
9:00 am to 10:30 am
A Provably Secure Polynomial Approximation Scheme for the
Distributed Lottery Problem, Broder
Fast Asynchronous Byzantine Agreement, Ben-Or
Simple Constant-Time Consensus Protocols in Realistic Failure
Models, Chor, Merritt, and Shmoys
Session V Leaders Chair: Nicola Santoro
10:50 am to 12:30 pm
A Modular Technique for the Design of Efficient Distributed
Leader Finding Algorithms, Korach, Kutten, and Moran
Improvements in the Time Complexity of Two Message-Optimal
Election Algorithms, Gafni
Time and Message Bounds for Election in Synchronous and
Asynchronous Complete Networks, Afek and Gafni
Computing on an Anonymous Ring, Attiya, Snir, and Warmuth
Session VI Knowledge Chair: Allen Emerson
2:00 pm to 3:30 pm
How Processes Learn, Chandy and Misra
Cheating Husbands and Other Stories: A Case Study of Knowledge,
Action, and Communication, Dolev, Halpern, and Moses
A Formal Model of Knowledge, Action, and Communication in
Distributed Systems, Fagin and Halpern
Session VII Systems Chair: Mani Chandy
4:00 pm to 5:30 pm
Inexact Agreement: Accuracy, Precision, and Graceful Degradation,
Mahaney and Schneider
Towards a Universal Directory Service, Lantz, Edighoffer, and
Hitson
Distributed Match-Making for for Processes in Computer Networks,
Mullender and Vitanyi
======================================================================
Wednesday, 7 August
Session VIII Graphs Chair Bob Galleger
9:00 am to 10:30 am
Communication-Time Trade-offs in Network Synchronization,
Awerbuch
The Optimality of Distributive Constructions of Minimum Weight and
Degree Restricted Spanning Trees in a Complete Network of
Processors, Korach, Moran, and Zaks
Terminating Iterative Solutions of Simultaneous Equations
in Distributed Message Passing Systems, Szymanski, Shi,
and Prywes
Session IX Parallel Chair: Ray Strong
11:00 am to 12:30 pm
Parallel Arithmetic with Concurrent Writes, Itai
A Simple, Efficient Asynchronous Parallel Algorithm for Maximization,
Greenberg, Lobachevsky, and Odlyzko
Choice Coordination with Limited Failure, Bar-Noy, Ben-Or, and Dolev
======================================================================
∂10-Apr-85 2258 @SU-SCORE.ARPA:udi@wisc-rsch.arpa STOC directions
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 10 Apr 85 22:58:21 PST
Received: from wisc-rsch.arpa by SU-SCORE.ARPA with TCP; Wed 10 Apr 85 22:43:11-PST
Received: by wisc-rsch.arpa; Thu, 11 Apr 85 00:27:13 cst
Received: from wisc-crys.arpa by wisc-rsch.arpa; Wed, 10 Apr 85 17:19:58 cst
Received: from csnet-relay.arpa by wisc-crys.arpa; Wed, 10 Apr 85 17:19:41 cst
Received: from brown by csnet-relay.csnet id ak04412; 10 Apr 85 18:14 EST
Message-Id: <8504102030.AA27229@nancy.CS.Brown.CSNet>
Date: 10 Apr 85 (Wed) 15:30:25 EST
From: Jeff Vitter <jsv%brown.csnet@csnet-relay.arpa>
To: theory@wisc-crys.ARPA
Subject: STOC directions
Sender: udi@wisc-rsch.arpa
Resent-From: udi@wisc-rsch.arpa (Udi Manber)
Resent-To: theory:;@wisc-rsch.arpa
By popular demand, here are driving directions and general tidbits for STOC:
The conference will be held at the Biltmore Hotel on Kennedy Plaza in downtown
Providence. Parking is available in a parking lot building behind the
Biltmore, on Westminster.
To get to the Biltmore. . .
FROM THE NORTH (e.g., from Boston): Take I-95 south and exit at Exit 21
(Broadway) at downtown Providence. When you get off the ramp, take a left
at the traffic light (Atwell's Ave.) and cross over the interstate.
Continue another block to the end of the street; there's a light there and a
Gulf gas station on your right. Take a left. The road makes a quick
90-degree veer to the right. Then execute *.
*: Immediately after the veer, take a left at the light. Then proceed to the
end of the street (second light) and take a right. The Biltmore will be on
your right. To get to the parking lot, take another right at the far corner
of the Biltmore, then go down a block.
FROM THE SOUTH (e.g., from Providence's Green Airport): Take I-95 north and
exit at Exit 21 (Broadway) near downtown. When you get off the ramp, take a
right at the light (Broadway). After you pass thru a light, the road will
veer 90-degrees to the right. Then execute *.
FROM THE WEST (e.g., from Hartford): Come in on Temporary I-195 east and go
all the way to the end until you're forced to exit. Take a right at the light
at the top of the offramp (Dean St.). Take a left at the next light you come
to (Atwell's Ave.). This will take you through part of the Italian restaurant
district. Wave hi to Guido as you pass Marcello's Rest. Eventually you will
cross over I-95 and shortly thereafter the street ends at a light. Take a
left. The road will veer 90-degrees to the right. Then execute *.
FROM THE EAST (e.g., from Newport, where the banquet is): Take I-195 west.
At the end, you're forced to merge onto either I-95 north or south. Go north.
Take the next exit (Exit 21 Broadway) and follow the rest of the directions
for "FROM THE SOUTH."
There's a lot of construction going on downtown; it's in the midst of being
totally overhauled. I'll send an update if some detours get put up that mess
up these directions.
To get to Brown University (a "must" see), proceed out of the front of the
Biltmore (in the direction you're facing when you leave thru the front
entrance). The road goes thru a couple lights, passes thru Suicide Circle,
then heads up College Hill, crossing Benefit Street on the way. Brown is at
the top of the hill, in the section of the city called the East Side. This is
within walking distance from the Biltmore. If you want to go to the von Bulow
trial, the State Courthouse is to the right on Benefit Street.
The Annual Festival of Historic Homes will be Friday, May 3 - Sunday, May 5.
There are candlelight tours Friday and Saturday eves. in the Benefit Street
area, and a tour on Saturday afternoon in a different area along Benefit.
On Sunday, there's a brunch and tour of the Broadway area. It's well worth it.
Call the Providence Preservation Society at (401) 831-7440 if interested.
Tickets usually go fast!
∂11-Apr-85 0924 EMMA@SU-CSLI.ARPA Colloquium Cancelled
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 11 Apr 85 09:24:30 PST
Date: Thu 11 Apr 85 09:21:02-PST
From: Emma Pease <Emma@SU-CSLI.ARPA>
Subject: Colloquium Cancelled
To: friends@SU-CSLI.ARPA
Tel: 497-3479
THIS THURSDAY'S COLLOQUIUM
``A Formal Theory of Innate Linguistic Knowledge''
Janet Fodor
Has been postponed because of illness (laryngitis). The new time will
be announced later.
-------
∂11-Apr-85 1335 JOHN@SU-CSLI.ARPA tgif
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 11 Apr 85 13:35:06 PST
Date: Thu 11 Apr 85 13:31:07-PST
From: John Perry <JOHN@SU-CSLI.ARPA>
Subject: tgif
To: folks@SU-CSLI.ARPA
T G I F
Join us in a farewell party for Eric.
TIME: Friday, 12 April, 4:30 p.m.*
PLACE: Ventura Trailers Deck
*Philosophers: join us after the colloquium.
-------
∂11-Apr-85 1356 HEDGES@SU-SCORE.ARPA Visit of Steven Zucker
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 11 Apr 85 13:56:25 PST
Date: Thu 11 Apr 85 13:42:56-PST
From: Karen Hedges <HEDGES@SU-SCORE.ARPA>
Subject: Visit of Steven Zucker
To: Faculty@SU-SCORE.ARPA
NOTICE: Steven Zucker (Robotics Candidate) will be visiting on April 23.
I would like to schedule times for him to meet with Faculty whom
might be available on that day. There are currently openings
Please let me know at your earliest convenience what times
would be best to meet with him.
Thanks. --Karen
-------
∂11-Apr-85 1411 HEDGES@SU-SCORE.ARPA Visit of Steven Zucker
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 11 Apr 85 14:11:45 PST
Date: Thu 11 Apr 85 13:46:49-PST
From: Karen Hedges <HEDGES@SU-SCORE.ARPA>
Subject: Visit of Steven Zucker
To: Faculty@SU-SCORE.ARPA
Please Disregard Previous Message - Not Complete!
NOTICE: Steven Zucker (Robotics Candidate) will be visiting on April 23.
I would like to schedule times for him to meet with Faculty whom
might be available on that day. There are currently openings from
10:00 to Noon.
Please let me know at your earliest convenience what times would be
best to meet with him.
Thanks.
Karen
-------
∂11-Apr-85 1453 PIERCE@SU-SCORE.ARPA Quad Dedication
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 11 Apr 85 14:53:02 PST
Date: Thu 11 Apr 85 14:49:44-PST
From: Lee Pierce <PIERCE@SU-SCORE.ARPA>
Subject: Quad Dedication
To: faculty@SU-SCORE.ARPA, staff@SU-SCORE.ARPA
I have just received word from the Development Office that an invitation
is extended to Stanford faculty and staff to attend the Quad Dedication
to be held Friday, May 17, 1985, from 4-6pm. There will be a ceremony
and a Tea Dance (Ernie Heckshire's orchestra) and refreshments.
I need to let them know who will be attending so please get back to me
as soon as you can either yes or no. Thanks.
Lee
-------
∂11-Apr-85 1507 PIERCE@SU-SCORE.ARPA Quad Dedication
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 11 Apr 85 15:07:35 PST
Date: Thu 11 Apr 85 15:03:31-PST
From: Lee Pierce <PIERCE@SU-SCORE.ARPA>
Subject: Quad Dedication
To: faculty@SU-SCORE.ARPA, staff@SU-SCORE.ARPA
I forgot to mention--sorry, spouses cannot be invited to Quad Dedication.
Lee
-------
∂11-Apr-85 1639 TOM@SU-SCORE.ARPA Re: Quad Dedication
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 11 Apr 85 16:38:58 PST
Date: Thu 11 Apr 85 16:12:23-PST
From: Thomas Dienstbier <TOM@SU-SCORE.ARPA>
Subject: Re: Quad Dedication
To: PIERCE@SU-SCORE.ARPA
cc: faculty@SU-SCORE.ARPA, staff@SU-SCORE.ARPA, TOM@SU-SCORE.ARPA
In-Reply-To: Message from "Lee Pierce <PIERCE@SU-SCORE.ARPA>" of Thu 11 Apr 85 14:50:33-PST
put me down for myself and my wife joyce who is also a stanford
employee..
thanks
tom
-------
∂11-Apr-85 1926 @SU-SCORE.ARPA:reid@Glacier Quad dedication, tea dancing, and spouses
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 11 Apr 85 19:24:55 PST
Received: from Glacier by SU-SCORE.ARPA with TCP; Thu 11 Apr 85 19:19:41-PST
Date: 11 Apr 1985 1919-PST (Thursday)
From: Brian Reid <reid@Glacier>
To: Pierce@Score
Cc: Faculty@Score, Staff@Score
Subject: Quad dedication, tea dancing, and spouses
I find it difficult to reconcile these two statements:
------- Forwarded Message
From: Lee Pierce <PIERCE@SU-SCORE.ARPA>
Subject: Quad Dedication
.... Quad Dedication to be held Friday, May 17, 1985, from 4-6pm.
There will be a ceremony and a Tea Dance (Ernie Heckshire's
orchestra)...
From: Lee Pierce <PIERCE@SU-SCORE.ARPA>
Subject: Quad Dedication
I forgot to mention--sorry, spouses cannot be invited to Quad
Dedication.
Lee
------- End of Forwarded Messages
Does this mean that we are to Tea Dance with our colleagues, or perhaps
to pick up partners from among the faculty and staff?
∂11-Apr-85 2314 @SU-SCORE.ARPA:WIEDERHOLD@SUMEX-AIM.ARPA Re: Quad Dedication
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 11 Apr 85 23:12:15 PST
Received: from SUMEX-AIM.ARPA by SU-SCORE.ARPA with TCP; Thu 11 Apr 85 23:07:45-PST
Date: Thu 11 Apr 85 23:07:58-PST
From: Gio Wiederhold <WIEDERHOLD@SUMEX-AIM.ARPA>
Subject: Re: Quad Dedication
To: TOM@SU-SCORE.ARPA
cc: PIERCE@SU-SCORE.ARPA, faculty@SU-SCORE.ARPA, staff@SU-SCORE.ARPA
In-Reply-To: Message from "Thomas Dienstbier <TOM@SU-SCORE.ARPA>" of Thu 11 Apr 85 16:53:30-PST
That's ok for some us, but I think it is very uncivil to have a dance and
not permit spouses. Downright immoral ... . Gio
-------
∂12-Apr-85 0056 RESTIVO@SU-SCORE.ARPA PROLOG Digest V3 #18
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 12 Apr 85 00:56:23 PST
Date: Thursday, April 11, 1985 6:10PM
From: Chuck Restivo (The Moderator) <PROLOG-REQUEST@SU-SCORE.ARPA>
Reply-to: PROLOG@SU-SCORE.ARPA
US-Mail: P.O. Box 4584, Stanford CA 94305
Phone: (415) 326-5550
Subject: PROLOG Digest V3 #18
To: PROLOG@SU-SCORE.ARPA
PROLOG Digest Friday, 12 Apr 1985 Volume 3 : Issue 18
Today's Topics:
Puzzles - Maps,
Implementations - Searching & Denotational Semantics,
& Cuts & Snips & Control & C-Prolog Ports,
LP-Library - Proceedings
----------------------------------------------------------------------
Date: Tue, 9 Apr 85 21:44:49 pst
From: Allen VanGelder <AVG@Diablo>
Subject: Shapiro's coloring program
Shapiro's coloring program can be vastly improved in
efficiency by the not-not trick in "subset". The
modified rules are shown below, but first I am showing
how to fix "map" to correct a typo in his original
message.
The not-not in "subset" avoids blindly instantiating the
colors of adjacent countries. Someone who can profile
might be interested in reporting performance (after the
typo is fixed) of both methods. Two run-types should be
compared: (1) time to first solution; (2) time of all 120
solutions. A space report would also be interesting.
The number of solutions can be reduced to 30 by adding
":- B=blue" to the clause for "map".
It is disturbing that one such "non-logical" trick can
deliver huge performance gains over correct "pure Prolog"
programs.
/*
←←←←←←←←←←
| a |
|←←←←←←←←|
|b |c |d |
|←←|←←|←←|
|e |f |
|←←←|←←←←|
is represented by the term:
*/
map( [country(a,A,[B,C,D]),
country(b,B,[A,C,E]),
country(c,C,[A,B,D,E,F]),
country(d,D,[A,C,F]), /* <---- fix */
country(e,E,[B,C,F]),
country(f,F,[C,D,E])
]
).
subset([C|Cs],Colours) :-
\+ \+ remove(C,Colours,←), /* <------- changed */
subset(Cs,Colours).
subset([],←Colours).
------------------------------
Date: Tue, 9 Apr 85 10:18:34 bst
From: William Clocksin <wfc%cl.cam@ucl-cs.arpa>
Subject: Map coloring∞
Ehud Shapiro's map colouring program reminds me of a
program I wrote a few months ago. It performs greedy
gate assignment of digital circuits. The representation
of circuits is similar to the representation of countries,
except that I use Prolog clause syntax instead of lists.
So, the simple half-subtractor looks like:
half←sub(In1,In2,Diff,Borr) :-
xor(In1,In2,Diff),
not(In1,T),
and(In2,T,Borr).
The only difference between this and the countries is that
this can be nested to arbitrary depth. For example, the
logic functions (actually relations) shown can be defined
in terms of CMOS transistors, if you like. Anyway, the
main part of greedy gate assignment is about three lines
long, and looks similar to Shapiro's map colourer.
Another program, only slightly more complicated, finds
subcircuits isomorphic to some given ones, and rewrites
them into other subcircuits. The experience, shared
with Shapiro's map colourer, is that variables are very
convenient for expressing adjacencies.
------------------------------
Date: Wed, 10 Apr 85 07:09:38 EST
From: cugini@NBS-VMS
Subject: Finding Best Solution
In any situation where no more is known about p(X),
other than that it generates all appropriate X's,
in particular, where the order of generation is
unrelated to the cost, then clearly (?) the best
algorithm is proportional to the size of the set {p(X)}.
A simple linear procedural solution is:
low-cost := infinity
set-of-best := null
for all X s.t. P(X)
if cost(X) < low-cost then
set-of-best := {X}
low-cost := cost(X)
else
if cost(X) = low-cost then
set-of-best := set-of-best + {X}
endif
endif
end for
Is there any motivation for doing this in Prolog other
than fun?
A related issue is the usefulness of such a generic
solution. As mentioned, I don't see how it could be
better than linear relative to the size of the candidate
set, {p(X)}, in the absence of other information about
the way the solutions are ordered or constructed. Notice
that in the case of the shortest-path problem for graphs,
this approach quickly becomes unrealistic - it means
generating *all* paths between the two specified vertices,
and choosing the shortest of these. The number of paths
is (I think) something like: avg-degree-of-graph
** number-of-vertices. The realistic algorithm involves
building a kind of balanced tree outward from one of the
nodes (balanced in the sense that all the reached nodes
are at least as close to the origin node as any of the
unreached, i.e. all partial paths built are optimal) until
the other is reached. This ensures that the first path
found is an optimal one - the other sub-optimal paths are
never explicitly generated and examined. This algorithm
runs, I think in O(number-of-vertices ** 2).
The point here is that for "large" search spaces, one
would almost surely need/want to take advantage of
domain-specific knowledge in order to prune the search.
-- John Cugini
------------------------------
Date: Tue, 9 Apr 85 13:12:11 pst
From: Allen VanGelder <AVG@Diablo>
Subject: Searching
One straightforward solution to Hilfinger's problem is:
best(X) :- p(X), cost(X, N), \+ (p(Y), cost(Y, M), M < N).
This satisfies his criteria, I guess, but is certainly not
very efficient compared to:
best(X) :-
bagof(c(Y, M), (p(Y), cost(Y, M), Cs),
mincost(Cs, N),
member(c(X, N), Cs).
where mincost has an obvious definition and succeeds
exactly once, and member has the usual definition.
The stigma attached to bagof and findall is misguided in my
opinion. There are many common operations that are known
not to be expressible AT ALL by first-order operations.
Counting is the canonical example. (*) Others, like this
example are an order of magnitude more efficient with
setof, bagof, or findall. Why use a grossly inefficient
method like the first rule given?
* Actually it is possible to count how many times p(X)
succeeds by a tortuous first-order program by using the
fact that Prolog supplies a built-in predicate that orders
all terms. Don't ask me for details; look in the
literature. Immerman in STOC 82 is a starting place.
------------------------------
Date: Tue 9 Apr 85 17:23:32-MST
From: Uday Reddy <U-REDDY@UTAH-20.ARPA>
Subject: Semantics - Initial models
Oops! I should have qualified my statement about
combinatorial complexity of equality in initial
model-based languages. That is necessary only if the
implementation wants to deal with nontermination. Most
implementations of initial model-based languages get
around the problem by assuming that the programs are
strongly terminating (all reduction sequences are
finite). The implementations then disagree with the
initial model for the nonterminating case. Note that
if-then-else has to be a lazy function and so violates
the assumption of strong termination. Special treatment
is required to deal with conditional equations. The
domain-theoretic least model corresponds to an efficient
implementation even for the nonterminating case.
We are really comparing theories with different
strengths here. Initial models explain well general
equational programs of first order. Domain theory is
used for fixed point equations of the form
f = t(f).
These can be added syntactic sugar to get constructor-
based equational languages. But, more general equations
(e.g. associativity that Goguen pointed earlier) cannot
be dealt with. On the positive side, domain theory can
easily deal with lazy and higher order functions.
Evidently, there are good research problems here in
bringing the two theories together.
Let me also add that this has no relevance to Horn
clause languages without equality (e.g. Prolog) for
which initial models and least models are the same.
-- Uday Reddy
------------------------------
Date: 3 Apr 85 01:19:37 GMT
From: (David Powers) decvax!mulga!elecvax.oz!DavidP@Berkeley
Subject: "soft" cuts
Coming in on a response to a query I missed concerning
"soft" cuts:
I added such a construct to UNSW PROLOG in 1982 - it is a
very simple change, related closely to traditional cut. I
used the symbol "#", hash, and the term "half cut". An
example of its use compared with alternative ways of
achieving the same end is given in Chapter 5, pp19-28 of
UNSW DCS Technical Report 8313, D.M.W.Powers and G.B.McMahon,
"A Compendium of Interesting PROLOG programs" (December 1983).
The advantage of half cut is that it allows obtaining all
solutions, if any exist, and execution of an alternative
strategy or error routine otherwise. I have not found any
markedly different application. My original application was
natural language parsing/learning.
This half cut cannot be simulated with Horn + cut
(without assert/bagof/...), but the other half -
commit to current track within clause but not to
current clause - can be simulated with cut alone.
I am happy to look out the specific mods to UNSW
PROLOG for those who request them.
-- David M. W. Powers
------------------------------
Date: 10 Apr 1985 11:57-EST
From: Saumya Debray <debray%suny-sb.csnet@csnet-relay.arpa>
Subject: snips; cut on the WPE; static control of backtracking
Keith Hughes reported a construct, "snip", which
permits more flexible cutting. We've used something
similar in our implementation of D H D Warren's
Abstract Prolog Machine last summer. Our idea
(originating with David S. Warren here at Stony
Brook) is to introduce system built-in predicates,
savecp(X), cutto(X) and cutout(X). These are
low-level primitives unavailable to the user (we don't
trust him, as you can see :-)), and are introduced by
our compiler. They work as follows: savecp(X) stores
the address of the current choice point in X; cutto(X)
makes the current choice point to be the same as that
stored in X; and cutout(X) zaps the choice point in X
to nil.
With this, Hughes' snipped clause
p :- q1, q2, [!, q3, q4, !], q5.
is equivalent to the clause
p :- q1, q2, savecp(X), q3, q4, cutto(X), q5.
where X is a variable that doesn't occur anywhere
except in the savecp/cutto pair. An ordinary cut is
translated as follows:
p(X←1, ..., X←n) :- q1, !, q2.
is transformed to
p(X←1, ..., X←n) :-
savecp(X←n+1), p1(X←1, ..., X←n, X←n+1).
p1(X←1, ..., X←n, X←n+1) :- q1, cutto(X←n+1), q2.
where p1 is a new predicate, and X←n+1 a new variable.
Our register allocation algorithms eliminate
practically all the overhead that might be expected
with the extra call.
Things get a little more interesting when these
primitives are used for static pruning of Prolog's
search tree (runtime intelligent backtracking incurs
such tremendous overhead that it often isn't worth
it). The idea is to isolate independent goals via
compile-time analysis and use our primitives to
control backtracking: e.g. with
p(X,Y,Z) :- q(X), r(Y), s(X,Z).
if it can be shown that the goals r and s will never
share a variable at runtime, then this can be
transformed by the compiler to
p(X,Y,Z) :- q(X), savecp(V1), r(Y),
((savecp(V2), s(X,Z), cutout(V2)) ;
(cutto(V1), fail)
).
Here, the goal "savecp(V2)" stores the choice point
for the disjunction (';'). If execution succeeds past
"s(X,Z)", then the "cutout(V2)" zaps the choice point
for the disjunction; then, if execution ever
backtracks to this point, the other disjunctive branch
("cutto(V1),fail") is not executed, and execution can
backtrack to the goal "r(Y)". On the other hand, if
the goal "s(X,Z)" fails, then the other branch of the
disjunction is executed, the current choice point reset
to V1 (which is before the call to r), and the "fail"
then causes execution to backtrack to "q(X)", rather
than to "r(Y)". This is similar to D.H.D. Warren's
independent-subgoal-processing in Chat-80, except that
the analysis is a lot more complicated because of the
possibility of partially instantiated structures.
There are logically correct programs which do not
terminate with Prolog's naive backtracking strategy,
as the following example (due to Jieh Hsiang) shows:
between(X,Y,Z) :-
int(X), int(Y), int(Z), gt(X,Y), gt(Z,X).
int(0).
int(X+1) :- int(X).
gt(X,Y) :- X1 is X, Y1 is Y, X1 > Y1.
:- between(0,0+1,Z).
However, the goals "int(Z)" and "gt(X,Y)" in the
definition of between/3 are independent, and, when
transformed as above, behaves correctly.
In general, our strategy does not detect all possible
intelligent backtrack points. However, the fact that
the analysis is all done statically means that the
runtime overhead is practically nonexistent.
- Saumya Debray
SUNY at Stony Brook
------------------------------
Date: 11 Apr 85 23:37:18 +1000 (Thu)
From: decvax!mulga!mungunni.oz!lee@Berkeley
Subject: Control
I think this is rather unfair. If you change a goto
in a FORTRAN program, the chances are, you will get
a different answer out of it. In (pure) PROLOG, if
you change the computation rule, the answer(s) remain
the same, since SLD resolution is complete(*). There
really should be more distinction between forms of
"control" which can affect answers (like cut) and those
which cant (like wait declarations and freeze). Uday's
points are only valid for the former category.
(*) Of course, many people have pointed out that PROLOG
is *not* complete, since it uses a depth first search
strategy. However, it does have a weaker form of
completeness:
if the query evaluation terminates then all solutions
are guaranteed to have been found.
This is all that is needed in most applications. Its
important that your logic programming system can tell
you when it has found all solutions (if any) to your
query, rather than looping indefinitely. No system
can prevent all infinite loops, so we must rely on people
to write programs which terminate and so the weaker form
of completeness is OK.
-- Lee Naish
------------------------------
Date: Mon, 8 Apr 85 15:14 EST
From: Mark Beutnagel <Beutnagel%upenn.csnet@csnet-relay.arpa>
Subject: C-Prolog Port Notes
C-Prolog port notes
University of Pennsylvania
HP 9000 Series 200 workstation
HP-UX 2.1 operating system
Mark Beutnagel, Spring 1985
1. Version 1.5 copied from Vax (unix) to HP via Kermit;
the C-Prolog source directory and the pl subdirectory
are needed.
2. Include file sys/types.h contained a typedef Unsigned
which conflicted with the macro Unsigned used by
C-Prolog. A clean copy of types.h was placed in the
C-Prolog directory and sysbits.c (~line 35) was changed
to use the local types.h.
3. Changes to the makefile:
a. -Dunix added compile flags
b. float type = IEEE
c. EOF = null entry (not AsDEC10)
4. Check location and name of the startup file; change
in the makefile (or in parms.c) if not suitable.
5. The error message "Not a typewriter" appears when
states are 'saved', but restored states >>seem<< to work
anyway. This could be a bug in HP-UX. It probably
results from some piece of ioctl wanting a screen but
sent to the state file.
------------------------------
Date: 11 Apr 85 10:03:32 +1000 (Thu)
From: Isaac Balbin <munnari!mungunni.oz!isaac@seismo.ARPA>
Subject: C-Prolog Ports
In case anyone is interested I have done ports of 1.4d
(plus extra fixes from Richard O'Keefe) - I am unsure how
this differs with 1.5 - to
(1) a Perkin Elmer 3240 running (almost) Unix 4.2BSD,
(2) an ELXSI System 6400 (64 bit super-computer)
this is a SysV port of Unix to a multi processor, which
during the Beta Test period ran only ~ 2.3 times the speed
of a Vax 11/780 (4.2BSD).
(3) A Pyramid 90x which ran at ~ 1.3 times the Vax 11/780.
Note my timings were done using the standard (but
controversial) nrev. The fixes usually had to do with bit
twiddling and masking. Some nice tricks involving double
casts were used. I expect they will at least also be
necessary for 1.5. If anyone wants details please mail me.
As for trouble with the 68000 Port regarding the 'not a
typewriter' message, I vaguely remember the same problem
with 1.4d running on 4.2 and solving it in sysbits.c by
changing the 'fancy' IsaTty assignment back to
IsaTty = isatty(fileno(stdin));
because isaTty *does* work (at least now, with pipes etc).
PS: If anyone at Edcaad is listening could you send us a
tape of 1.5?
-- Isaac
------------------------------
Date: 11 Apr 85 1225 PST
From: Yoni Malachi <YM@SU-AI.ARPA>
Subject: Lisp conferences proceedings
[from SIGACT News]
ACM SIGPLAN has republished the conference proceeding of
previous Lisp conferences.
Price
Order No. Members Others
1980 Lisp Conf. 552800 $15 $21
1982 Lisp Conf. 552820 $18 $26
1984 Lisp Conf. 552840 $20 $27
Ordering address (prepaid)
ACM Order Dept.
P.O.Box 64145
Baltimore, MD 21264
------------------------------
End of PROLOG Digest
********************
∂12-Apr-85 0837 JOHN@SU-CSLI.ARPA Lunch at Ventura
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 12 Apr 85 08:36:48 PST
Date: Fri 12 Apr 85 08:34:48-PST
From: John Perry <JOHN@SU-CSLI.ARPA>
Subject: Lunch at Ventura
To: Folks@SU-CSLI.ARPA
Starting today, you can order sandwiches for lunch at Ventura.
Requests should be sent to
Csli-lunch
by 10:30 a.m. each day. You should specify what kind of sandwich
you want, among,
turkey, roast beef, ham, vegetarian, chicken
you may specify additional desired aspects of your sandwich, but
until we have experimented with this for a bit, we won't know if
it will make any difference. Please bear with us.
Also, until we get going, substitutions may be necessary.
There will be a few extra sandwiches available for last minute
lunchers.
Sandwiches cost $2.50.
A five-sandwich meal ticket will be available (either today
or before too long) for $12.
Cokes etc. will be available at the machine.
You can also add a piece of fruit (apples, oranges, anything else
in season) to your order, for $.25. Not included in meal ticket.
You can add suggestions for what we should do to your order, too.
----
What is the point of this?
It seems that the lunch hour might be a good time for people to make
it to ventura for casual meetings and situated conversation.
We are exploring the idea of encouraging the scheduling of project
meetings between 11-2 next year, so that people can combine their
lunch hour with their time at Ventura, and the popularity of lunch
will help us guage the reaction to this. (Direct comments also
welcome). (One point is to lessen conflicts with departmental
seminars, which tend to be later in the afternoon).
We want to provide an alternative to the faculty club for official
and semi-official lunches.
Its sunny and nice here.
------------------
Volleyball.
Would you like to see a (inexpensive) volleyball court at Ventura?
Would you play? would it decrease from situated conversation or
add to it? Send your comments to Jamie.
------------
Reminder:
Lunch requests to
csli-lunch
by 10:30 each day. Sandwiches will be there about noon.
--------
ps: Not networkers, those with down computers, and their ilk, can
order by phone:
497-0628
-------
∂12-Apr-85 0948 MARJORIE@SU-CSLI.ARPA system shutdown
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 12 Apr 85 09:48:51 PST
Date: Fri 12 Apr 85 09:47:08-PST
From: Marjorie Maxwell <MARJORIE@SU-CSLI.ARPA>
Subject: system shutdown
To: Folks@SU-CSLI.ARPA
There will be a short shutdown of systems this morning (between 10-11:00) due
to work on airconditioning in Pine Hall.
-------
∂12-Apr-85 1246 @SU-CSLI.ARPA:BrianSmith.pa@Xerox.ARPA No meeting 4/16; see you 4/23
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 12 Apr 85 12:46:35 PST
Received: from Xerox.ARPA by SU-CSLI.ARPA with TCP; Fri 12 Apr 85 12:44:12-PST
Received: from Semillon.ms by ArpaGateway.ms ; 12 APR 85 12:40:38 PST
Date: 12 Apr 85 12:23 PST
From: BrianSmith.pa@Xerox.ARPA
Subject: No meeting 4/16; see you 4/23
To: RRR@SU-CSLI.ARPA
cc: BrianSmith.pa@Xerox.ARPA
For two reasons (conflict with the one-time linguistics debate, and
because Fernando had a prior commitment) we won't have an rrr meeting
next Tuesday (4/16). Fernando will present his account of
representation within situated automata the next week, on Tuesday 4/23.
See you all then.
Brian
∂12-Apr-85 1427 @SU-SCORE.ARPA:WASHINGTON@SUMEX-AIM.ARPA POTLUCK
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 12 Apr 85 14:27:39 PST
Received: from SUMEX-AIM.ARPA by SU-SCORE.ARPA with TCP; Fri 12 Apr 85 14:15:53-PST
Date: Fri 12 Apr 85 13:51:30-PST
From: Rich Washington <washington@SUMEX-AIM.ARPA>
Subject: POTLUCK
To: csd@SU-SCORE.ARPA
Have you been looking for an opportunity to unveil your secret
recipe for Herhsey bars in garlic sauce? Now's your chance:
SON OF CSD POTLUCK, coming up a mere 2 weeks from now, on April 26.
Nils Nilsson, having not learned from his previous experience, is
offering us the use of his house again. This means, however, that
once again we'll have to limit the number of people who can come.
So... If you want to escape from the glow of the video screen for an
evening, let us know soon if you'll be coming, and whether you'll be
bringing any guests. We're working on a first come, first serve basis
(although large bribes will be considered).
We'd like to balance the food contributions, so please tell us which
of the following you plan to bring:
Salad
Main Dish
Dessert
Drink
If the distribution looks too lopsided, we'll start twisting arms, but
we're hoping this will work. We'll provide plates, cups, and plasticware.
So, to summarize:
WHAT: CSD Potluck
WHEN: Friday, April 26, 6:30 PM
WHO: CSD faculty, staff, students, and their guests
WHERE: Nils Nilsson's house
HOW: directions to Nil's house will be provided
WHY: why not?
RSVP to this account (washington@sumex), or talk to Rich Washington
or Roger Crew (yes Roger, you're running it too).
-------
∂12-Apr-85 1508 EMMA@SU-CSLI.ARPA tgif
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 12 Apr 85 15:08:39 PST
Mail-From: JOHN created at 11-Apr-85 13:31:07
Date: Thu 11 Apr 85 13:31:07-PST
From: John Perry <JOHN@SU-CSLI.ARPA>
Subject: tgif
To: folks@SU-CSLI.ARPA
ReSent-Date: Fri 12 Apr 85 15:03:29-PST
ReSent-From: Emma Pease <Emma@SU-CSLI.ARPA>
ReSent-To: folks@SU-CSLI.ARPA, bboard@SU-CSLI.ARPA
T G I F
Join us in a farewell party for Eric.
TIME: Friday, 12 April, 4:30 p.m.*
PLACE: Ventura Trailers Deck
*Philosophers: join us after the colloquium.
-------
∂13-Apr-85 1334 @SU-SCORE.ARPA:udi@wisc-rsch.arpa Conference on Theoretical Aspects of Reasoning About Knowledge
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 13 Apr 85 13:34:36 PST
Received: from wisc-rsch.arpa by SU-SCORE.ARPA with TCP; Sat 13 Apr 85 13:31:58-PST
Received: by wisc-rsch.arpa; Sat, 13 Apr 85 15:19:51 cst
Received: from wisc-crys.arpa by wisc-rsch.arpa; Fri, 12 Apr 85 23:21:15 cst
Message-Id: <8504130520.AA20352@wisc-crys.arpa>
Received: from csnet-relay.arpa by wisc-crys.arpa; Fri, 12 Apr 85 23:20:58 cst
Received: from ibm-sj by csnet-relay.csnet id a009239; 13 Apr 85 0:17 EST
Date: Fri, 12 Apr 85 19:24:07 PST
From: Joe Halpern <halpern%ibm-sj.csnet@csnet-relay.arpa>
To: theory@wisc-crys.ARPA
Subject: Conference on Theoretical Aspects of Reasoning About Knowledge
Cc: friends@su-csli.ARPA
Sender: udi@wisc-rsch.arpa
Resent-From: udi@wisc-rsch.arpa (Udi Manber)
Resent-To: theory:;@wisc-rsch.arpa
CONFERENCE ON THEORETICAL ASPECTS OF REASONING ABOUT KNOWLEDGE:
FIRST CALL FOR PAPERS
A conference on Theoretical Aspects of Reasoning About Knowledge
will be held Mar. 19-22, 1986, at the Asilomar Conference Center in
Monterey. While traditionally research in this area was mainly done by
philosophers, recently it has been shown to be of great
relevance to computer science, especially in such areas as artificial
intelligence, distributed systems, database systems,
and cryptography. There has also been interest in the area among
linguists and economists. The aim
of this conference is to bring together
researchers from these various disciplines
with the intent of furthering our theoretical understanding of
reasoning about knowledge.
Topics of interest include, but are not limited to:
* Semantic models for knowledge and belief
* Resource-bounded knowledge (appropriate for modelling reasoners with
limited reasoning power and reasoning about cryptographic protocols)
* Using knowledge to specify and reason about distributed systems
* Semantic models of knowledge acquisition and learning
* Nonmonotonic reasoning
Please send 8 copies of a detailed abstract
not exceeding 10 double-spaced typewritten pages in length
(not a full paper),
by September 15, 1985, to the program chair:
Dr. J. Halpern
IBM Research, K51/281
5600 Cottle Rd.
San Jose, CA 95193
The abstract should include a clear description of the problem being
addressed, comparisons with extant work, and a section on major
original contributions of this work. The abstract must provide
sufficient detail for the program committee to make a decision.
Papers will be chosen on the basis of scientific merit, originality,
and appropriateness for this conference.
Authors will be notified of acceptance by Nov. 1, 1985. Accepted
papers typed on special pages will be due at the above address
by Dec. 15, 1985.
The program committee members are:
M. Fischer, Yale
J. Halpern, IBM San Jose
H. Levesque, University of Toronto
R. Moore, SRI
R. Parikh, CUNY/Brooklyn College
R. Stalnaker, Cornell
R. Thomason, Pittsburg
M. Vardi, Stanford/CSLI
We hope to allow enough time between the talks during the conference
for private discussions and small group meetings. In order to
ensure that the conference remains relatively small, attendance will
be limited to invited participants and
authors of accepted papers.
Support for the conference has been received from IBM and AAAI;
an application for further support is pending at ONR.
∂14-Apr-85 1127 BARWISE@SU-CSLI.ARPA Job needed for college student
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 14 Apr 85 11:27:09 PST
Date: Sun 14 Apr 85 11:25:01-PST
From: Jon Barwise <BARWISE@SU-CSLI.ARPA>
Subject: Job needed for college student
To: BBoard@SU-CSLI.ARPA
cc: Folks@SU-CSLI.ARPA
My older daughter, Melanie, would like to find a job for the summer
around here. She will be here from mid May to Sept 1, if she can find
a job. She is one semester from her BA in art education from the
University of Wisconsin - Madison, and has is just finishing her
student teaching. Ideally, she would like a job that has to do with
art and/or education, so that she can use it on her resume. For
example, she would like to work in a nursery school, though junior
high is her specialty. But if you can cope with junior high kids, you
can cope with anything. If you know of any leads, please let me know.
Thanks, Jon
-------
∂14-Apr-85 2030 WASOW@SU-CSLI.ARPA Benefit Concert
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 14 Apr 85 20:30:29 PST
Date: Sun 14 Apr 85 20:28:07-PST
From: Tom Wasow <WASOW@SU-CSLI.ARPA>
Subject: Benefit Concert
To: Folks@SU-CSLI.ARPA
cc: Bboard@SU-CSLI.ARPA
Anyone interested in beautiful chamber music, delicious desserts, and
supporting a worthy cause simultaneously should be interested in the
following announcement:
HAYDN PIANO SONATA IN E-FLAT MAJOR
SCHUMANN SONG CYCLE: FRAUENLIEBEN UND LEBEN
BRAHMS STRING QUARTET OP. 26 IN A MAJOR
played by
Susan Gross Bratman, piano
Sharon Sheehan, mezzo soprano
Manzanita Quartet:
Marmee Eddy, violin; Stanley Evans, viola;
Tita McCall, cello; Edwin Good, piano
Coffee, tea, and wine, plus a selection of delectable desserts
prepared by Judith Wasow (with a little help from her family) will be
provided after the concert.
You can be part of this never-to-be-forgotten event for a contribution
of merely $10, all of which will be donated to the PTA of Juana
Briones Elementary School, a public school in Palo Alto.
Time: April 19, 1985, 8pm
Place: 914 Matadero Court, Palo Alto (the home of Sharon & Tom Wagner)
For tickets call Dionys Briggs at 856-7783.
-------
∂15-Apr-85 0810 EMMA@SU-CSLI.ARPA Reminder: Meyer talk Mon, 4:15, @SRI
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 15 Apr 85 08:10:34 PST
Return-Path: <WALDINGER@SRI-AI.ARPA>
Received: from SRI-AI.ARPA by SU-CSLI.ARPA with TCP; Fri 12 Apr 85 17:13:14-PST
Date: Fri 12 Apr 85 17:05:32-PST
From: WALDINGER@SRI-AI.ARPA
Subject: Reminder: Meyer talk Mon, 4:15, @SRI
To: AIC-Associates: ;,
CSL: ;, bboard@SRI-AI.ARPA, su-bboards@SU-AI.ARPA, friends@SU-CSLI.ARPA
cc: Meyer@MIT-MC.ARPA
ReSent-Date: Mon 15 Apr 85 08:06:01-PST
ReSent-From: Emma Pease <Emma@SU-CSLI.ARPA>
ReSent-To: friends@SU-CSLI.ARPA
on: Reasoning About Block Structured Variables:
What Makes the Free List Free?
in: EL381 (the Computer Science Lab Conference Room)
Building E, SRI International, on Ravenswood
Avenue, opposite the Pine Street Intersection, Menlo Park
at: 4:15pm, this Monday (not last Monday), 4/15
(note that time=date)
coffee: in Waldinger office (ek292) at 3:45pm
(come on time, it goes fast)
The speaker is Albert Meyer
-------
∂15-Apr-85 0820 EMMA@SU-CSLI.ARPA Conference on Theoretical Aspects of Reasoning About Knowledge
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 15 Apr 85 08:20:15 PST
Return-Path: <halpern.sjrlvm1%ibm-sj.csnet@csnet-relay.arpa>
Received: from csnet-relay by SU-CSLI.ARPA with TCP; Fri 12 Apr 85 21:20:10-PST
Received: from ibm-sj by csnet-relay.csnet id a009239; 13 Apr 85 0:17 EST
Date: Fri, 12 Apr 85 19:24:07 PST
From: Joe Halpern <halpern%ibm-sj.csnet@csnet-relay.arpa>
To: theory@wisc-crys.ARPA
Subject: Conference on Theoretical Aspects of Reasoning About Knowledge
CC: arpanet-bboards@mit-mc.ARPA, ailist-request@sri-ai.ARPA,
friends@su-csli.ARPA
ReSent-Date: Mon 15 Apr 85 08:06:42-PST
ReSent-From: Emma Pease <Emma@SU-CSLI.ARPA>
ReSent-To: friends@SU-CSLI.ARPA
CONFERENCE ON THEORETICAL ASPECTS OF REASONING ABOUT KNOWLEDGE:
FIRST CALL FOR PAPERS
A conference on Theoretical Aspects of Reasoning About Knowledge
will be held Mar. 19-22, 1986, at the Asilomar Conference Center in
Monterey. While traditionally research in this area was mainly done by
philosophers, recently it has been shown to be of great
relevance to computer science, especially in such areas as artificial
intelligence, distributed systems, database systems,
and cryptography. There has also been interest in the area among
linguists and economists. The aim
of this conference is to bring together
researchers from these various disciplines
with the intent of furthering our theoretical understanding of
reasoning about knowledge.
Topics of interest include, but are not limited to:
* Semantic models for knowledge and belief
* Resource-bounded knowledge (appropriate for modelling reasoners with
limited reasoning power and reasoning about cryptographic protocols)
* Using knowledge to specify and reason about distributed systems
* Semantic models of knowledge acquisition and learning
* Nonmonotonic reasoning
Please send 8 copies of a detailed abstract
not exceeding 10 double-spaced typewritten pages in length
(not a full paper),
by September 15, 1985, to the program chair:
Dr. J. Halpern
IBM Research, K51/281
5600 Cottle Rd.
San Jose, CA 95193
The abstract should include a clear description of the problem being
addressed, comparisons with extant work, and a section on major
original contributions of this work. The abstract must provide
sufficient detail for the program committee to make a decision.
Papers will be chosen on the basis of scientific merit, originality,
and appropriateness for this conference.
Authors will be notified of acceptance by Nov. 1, 1985. Accepted
papers typed on special pages will be due at the above address
by Dec. 15, 1985.
The program committee members are:
M. Fischer, Yale
J. Halpern, IBM San Jose
H. Levesque, University of Toronto
R. Moore, SRI
R. Parikh, CUNY/Brooklyn College
R. Stalnaker, Cornell
R. Thomason, Pittsburg
M. Vardi, Stanford/CSLI
We hope to allow enough time between the talks during the conference
for private discussions and small group meetings. In order to
ensure that the conference remains relatively small, attendance will
be limited to invited participants and
authors of accepted papers.
Support for the conference has been received from IBM and AAAI;
an application for further support is pending at ONR.
∂15-Apr-85 0910 LIBRARY@SU-SCORE.ARPA Math/CS Library New Books
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 15 Apr 85 09:10:31 PST
Date: Mon 15 Apr 85 09:07:09-PST
From: C.S./Math Library <LIBRARY@SU-SCORE.ARPA>
Subject: Math/CS Library New Books
To: su-bboards@SU-SCORE.ARPA
cc: faculty@SU-SCORE.ARPA
Computational Linguistics. A special issue of Computers and Mathematics With
Applications. Guest Editor. Nick Cercone P98.C612 1983 c.2
VAX Fortran by Middlebrooks. QA76.8.V37M53 1984
Elementary Cryptanalysis: a Mathematical Approach by Sinkov. Z104.S47 1966
The C Programmer's Handbook. by Hogan QA76.73.C15H64 1984
Computer Chess by Welsh GV1456.W44 1984
Datenanalyse: Mit Statistischen Methoden und Computerprogrammen by Siegmund
Brandt. QA276.B6844 1981.
Harry Llull
-------
∂15-Apr-85 0948 JOHN@SU-CSLI.ARPA reminder
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 15 Apr 85 09:47:52 PST
Date: Mon 15 Apr 85 09:43:03-PST
From: John Perry <JOHN@SU-CSLI.ARPA>
Subject: reminder
To: folks@SU-CSLI.ARPA
12-Mar-85 16:45:36-PST,513;000000000001
Mail-From: INGRID created at 12-Mar-85 16:45:30
Date: Tue 12 Mar 85 16:45:30-PST
From: Ingrid Deiwiks - 497-3084 <INGRID@SU-CSLI.ARPA>
Subject: Lunch Meeting
To: Betsy@SU-CSLI.ARPA, Pentland@SRI-AI.ARPA
cc: John@SU-CSLI.ARPA, Sandy@SU-CSLI.ARPA
John will be out of town Monday and Tuesday, so lunch could be Thursday,
March 21, 12:30. I will make a reservation at the Faculty Club.
(John: there is an executive committee meeting from 11-12, that's why I've
set the lunch for 12:30).
Ingrid
-------
12-Mar-85 16:46:58-PST,352;000000000001
Mail-From: INGRID created at 12-Mar-85 16:46:53
Date: Tue 12 Mar 85 16:46:52-PST
From: Ingrid Deiwiks - 497-3084 <INGRID@SU-CSLI.ARPA>
Subject: Lunch Meeting
To: Sag@Psych, Wasow@SU-CSLI.ARPA, Betsy@SU-CSLI.ARPA
cc: John@SU-CSLI.ARPA, Sandy@SU-CSLI.ARPA
This has been fixed for Wednesday, March 20, 12:30 at the Faculty Club.
Ingrid
-------
19-Mar-85 11:40:48-PST,329;000000000001
Mail-From: SAG created at 19-Mar-85 11:40:42
Date: Tue 19 Mar 85 11:40:41-PST
From: Ivan Sag <SAG@SU-CSLI.ARPA>
Subject: lunch
To: ingrid@SU-CSLI.ARPA
cc: betsy@SU-CSLI.ARPA, john@SU-CSLI.ARPA
Ingrid,
Could you add one more person to our lunch reservation for tomorow.
Martin Kay will join us.
Thanks,
Ivan
-------
20-Mar-85 14:18:22-PST,1105;000000000011
Mail-From: INGRID created at 20-Mar-85 14:18:20
Date: Wed 20 Mar 85 14:18:19-PST
From: Ingrid Deiwiks - 497-3084 <INGRID@SU-CSLI.ARPA>
Subject: Meeting with Betsy, Ivan, Tom, Martin
To: John@SU-CSLI.ARPA
cc: Sandy@SU-CSLI.ARPA
I only heard today that you were going to be out of town until tonight.
So, Ivan, Tom and Martin had lunch on their own to pre-discuss what they
want to discuss with you. They would like to have a discussion with you
before you go to Washington, though, as Tom thinks that what Ivan and Martin
have to say might be useful in Washington. How about a lunch meeting next
Thursday?
Also a reminder that there'll be another lunch meeting tomorrow, Thursday,
with Sandy Pentland and Betsy, 12:30 at the Faculty Club.
Luisa Ponta, the woman who is going to translate your and Jon's book, arrived
last week. She would like to have a meeting with you and Jon, so that she
can get started soon. We have also been unable to find her a desk anywhere
around CSLI. Would it be possible at all for her to use one of the desks
in your office?
Ingrid
-------
12-Apr-85 08:34:48-PST,2222;000000000001
Date: Fri 12 Apr 85 08:34:47-PST
From: John Perry <JOHN@SU-CSLI.ARPA>
Subject: Lunch at Ventura
To: Folks@SU-CSLI.ARPA
Starting today, you can order sandwiches for lunch at Ventura.
Requests should be sent to
Csli-lunch
by 10:30 a.m. each day. You should specify what kind of sandwich
you want, among,
turkey, roast beef, ham, vegetarian, chicken
you may specify additional desired aspects of your sandwich, but
until we have experimented with this for a bit, we won't know if
it will make any difference. Please bear with us.
Also, until we get going, substitutions may be necessary.
There will be a few extra sandwiches available for last minute
lunchers.
Sandwiches cost $2.50.
A five-sandwich meal ticket will be available (either today
or before too long) for $12.
Cokes etc. will be available at the machine.
You can also add a piece of fruit (apples, oranges, anything else
in season) to your order, for $.25. Not included in meal ticket.
You can add suggestions for what we should do to your order, too.
----
What is the point of this?
It seems that the lunch hour might be a good time for people to make
it to ventura for casual meetings and situated conversation.
We are exploring the idea of encouraging the scheduling of project
meetings between 11-2 next year, so that people can combine their
lunch hour with their time at Ventura, and the popularity of lunch
will help us guage the reaction to this. (Direct comments also
welcome). (One point is to lessen conflicts with departmental
seminars, which tend to be later in the afternoon).
We want to provide an alternative to the faculty club for official
and semi-official lunches.
Its sunny and nice here.
------------------
Volleyball.
Would you like to see a (inexpensive) volleyball court at Ventura?
Would you play? would it decrease from situated conversation or
add to it? Send your comments to Jamie.
------------
Reminder:
Lunch requests to
csli-lunch
by 10:30 each day. Sandwiches will be there about noon.
--------
ps: Not networkers, those with down computers, and their ilk, can
order by phone:
497-0628
-------
-------
∂15-Apr-85 1003 EMMA@SU-CSLI.ARPA [Janet Fodor <FODOR@SU-CSLI.ARPA>: Soc. for Phil & Psych meeting]
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 15 Apr 85 10:03:41 PST
Date: Mon 15 Apr 85 09:59:08-PST
From: Emma Pease <Emma@SU-CSLI.ARPA>
Subject: [Janet Fodor <FODOR@SU-CSLI.ARPA>: Soc. for Phil & Psych meeting]
To: folks@SU-CSLI.ARPA
Reply-To: fodor@csli
Tel: 497-3479
This year's meeting of the Society for Philosophy and Psychology will
be held at the University of Toronto, May 15 - May 18. The program is
posted on the CSLI board in the corridor outside the kitchen.
Abstracts of the symposia are available at the front desk. For further
information write to Steva Harnad, bellcore!princeton!mind!srh
-------
∂15-Apr-85 1012 JOHN@SU-CSLI.ARPA lunch
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 15 Apr 85 10:12:21 PST
Date: Mon 15 Apr 85 10:07:42-PST
From: John Perry <JOHN@SU-CSLI.ARPA>
Subject: lunch
To: folks@SU-CSLI.ARPA
Due to a computer age mistake, my reminder about lunch was appended
to a file that had some odds and ends about previous luches.
Please ignore and forgive.
-------
∂15-Apr-85 1121 HEDGES@SU-SCORE.ARPA Faculty Lunch
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 15 Apr 85 11:21:42 PST
Date: Mon 15 Apr 85 11:05:44-PST
From: Karen Hedges <HEDGES@SU-SCORE.ARPA>
Subject: Faculty Lunch
To: faculty@SU-SCORE.ARPA
REMINDER: The Faculty Lunch will take place tomorrow, April 16 @ 12:15 in
room 146.
The topic for discussion will be the "CSD Teaching Load".
-------
∂15-Apr-85 1135 PHILOSOPHY@SU-CSLI.ARPA Tanner Lectures
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 15 Apr 85 11:35:03 PST
Date: Mon 15 Apr 85 11:30:50-PST
From: Eve Wasmer <PHILOSOPHY@SU-CSLI.ARPA>
Subject: Tanner Lectures
To: Folks@SU-CSLI.ARPA
This year's Tanner Lectures in Human Values will be given by Michael Slote
from the University of Marylandon Wednesday, April 17 and Monday, April 22.
The lectures will be in Jordan Hall, Building 420, room 041 at 8:00 p.m.
The subject of the lectures will be "Moderation, Rationality, and Virtue.
-------
∂15-Apr-85 1339 BRODER@SU-SCORE.ARPA Speaker needed
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 15 Apr 85 13:38:48 PST
Date: Mon 15 Apr 85 13:03:49-PST
From: Andrei Broder <Broder@SU-SCORE.ARPA>
Subject: Speaker needed
To: aflb.all@SU-SCORE.ARPA
.... for Apr. 25. Help !!! - Andrei
-------
∂15-Apr-85 1358 @SU-SCORE.ARPA:udi@wisc-rsch.arpa Reminder
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 15 Apr 85 13:58:25 PST
Received: from wisc-rsch.arpa by SU-SCORE.ARPA with TCP; Mon 15 Apr 85 13:13:03-PST
Received: by wisc-rsch.arpa; Mon, 15 Apr 85 14:55:54 cst
Received: from wisc-crys.arpa by wisc-rsch.arpa; Mon, 15 Apr 85 13:34:58 cst
Message-Id: <8504151934.AA17837@wisc-crys.arpa>
Received: from COLUMBIA-20.ARPA by wisc-crys.arpa; Mon, 15 Apr 85 13:34:44 cst
Date: Mon 15 Apr 85 14:34:06-EST
From: Zvi Galil <GALIL@COLUMBIA-20.ARPA>
Subject: Reminder
To: theory@WISC-CRYS.ARPA
Sender: udi@wisc-rsch.arpa
Resent-From: udi@wisc-rsch.arpa (Udi Manber)
Resent-To: theory:;@wisc-rsch.arpa
This is a message I sent a month ago.
This week is THE LAST week for preregistring to STOC.
Students as well as nonstudents are asked to register this week.
See you soon. Zvi
Dear SIGACT member,
I am happy to report to you that SIGACT already has 6 institutional
sponsors. The money that we received will be spent in helping students
who attend STOC. This STOC
1.We will return to the students part of their registration fees.
2.We will either give students one free lunch or sell them
substantially discounted banquet tickets.
The level of support will depend on the number of institutional
sponsors (with luck we may have new sponsors), and on the number
of students attending the conference. Late in April we will
decide on the specific details of 1 and 2 above. Therefore, it is
important that all those who plan to attend register as soon as
possible.
You are going to receive the brochure with the program for STOC85 any
day. If you plan to attend the conference, I urge you to register
right away. Students who do not register before the April 22
deadline may not be able to get 1 and 2 above. (This is a kind
of student late registration fee..) Nonstudents are asked to
register soon because there are upper bounds on the capacities
of the lunch room and the banquet room.
In the business meeting we will discuss the possible ways to spend the
(institutional sponsors') money in the future.
One suggestion is to have a best
student paper award. Another is to have stipends for partial
reimbursement of travel costs. This option will become possible only if
we have several new sponsors. If you have other ideas, please let me
know. Also, I'd appreciate suggestions for new sponsors.
See you in Providence, Zvi Galil
PS. Please spread the word about this message to those that do not
have an access to theory-net.
-------
∂15-Apr-85 1539 WASOW@SU-CSLI.ARPA summer RAships
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 15 Apr 85 15:39:27 PST
Date: Mon 15 Apr 85 15:30:54-PST
From: Tom Wasow <WASOW@SU-CSLI.ARPA>
Subject: summer RAships
To: Folks@SU-CSLI.ARPA
Decisions on summer RAship applications have been delayed, due to the
large influx of late applications. Final decisions WILL be made, however,
over the coming weekend, and announced on Monday, April 22. Any materials
to be included in applications must reach me by Friday, April 19.
Tom
-------
∂15-Apr-85 1608 WRIGHT@SUMEX-AIM.ARPA Siglunch Friday, April 19
Received: from SUMEX-AIM.ARPA by SU-AI.ARPA with TCP; 15 Apr 85 16:08:47 PST
Date: Mon 15 Apr 85 16:03:46-PST
From: Carol Wright <WRIGHT@SUMEX-AIM.ARPA>
Subject: Siglunch Friday, April 19
To: Siglunch@SUMEX-AIM.ARPA
SIGLUNCH
DATE: Friday, April 19, 1985
LOCATION: Chemistry Gazebo, between Physical & Organic
Chemistry
TIME: 12:05
SPEAKER: Shigeki Goto
Musashino Electrical Communication Laboratory,
NTT, Japan
Visitor SU Computer Science Department
TITLE: Japanese Lisp Machine ELIS and it's Language TAO
NTT is a Japanese telephone company. Electrical Communication
Laboratories are often called "Japanese Bell Labs". At the Feburary
15 Siglunch, Professor Feigenbaum spoke about NTT's new Lisp machine
being ten times faster than Symbolics 3600. This talk is a follow
up on his talk, and will cover the following points:
1) A brief overview of NTT, and Research activities at Electrical
Communication Laboratories.
2) Lisp Machine ELIS: It is safe to say that ELIS is at least six times
faster than Symbolics 3600. Because the interpreter is implemented
fully by microcode, interpreted code runs faster than most dedicated
machines.
3) A dialect of Lisp: TAO
Although TAO looks like Zetalisp, it takes in features of Prolog and
Smalltalk. Users can program by selecting and mixing the programming
paradigms.
-------
∂15-Apr-85 2011 BRODER@SU-SCORE.ARPA Next AFLB talk
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 15 Apr 85 20:11:42 PST
Date: Mon 15 Apr 85 20:08:25-PST
From: Andrei Broder <Broder@SU-SCORE.ARPA>
Subject: Next AFLB talk
To: aflb.all@SU-SCORE.ARPA
cc: tuley@SU-SCORE.ARPA, su-bboards@SU-SCORE.ARPA
4-18-85 - Prof. Albert Mayer (MIT):
"Complexity of program flow-analysis for strictness:
application of a fundamental theorem of denotational semantics"
Call-by-value strategy specifies that evaluation of the following
functional expression would not terminate:
LETREC
f(x,y,z) = IF x<=0 THEN y ELSE f(x-1,z,y) FI
AND
g(z) = g(z)+1
IN
f(2*2,1,g(0))
END
The source of the trouble is the divergent argument g(0). In
contrast, call-by-need strategy postpones evaluation of g(0) until it
is needed in evaluating the body of f -- which it isn't -- and
ultimately terminates with the value 1. A function is "operationally
STRICT" in its kth argument if its call-by-NEED application to some
arguments fails to terminate whenever evaluation of the actual kth
argument fails to terminate. It is OK to evaluate operationally
strict arguments at "apply time" according to call-by-value strategy,
even when call-by-need semantics is specified. The f above is NOT
operationally strict in its third or second arguments, but is in its
first.
Call-by-need yields a mathematically more attractive semantics, but
call-by-value is generally more efficient. This motivates the
question of analyzing declarations to determine which arguments are
strict. We discuss the possibility of carrying out an abstract
"strictness flow-analysis" of functional programs, pointing out
undecidability and complexity results. The investigation provides a
case study of how denotational semantics yields an algorithmic
solution to an operationally specified program optimization problem.
In the finitely typed case without any interpreted functions
(including conditional), the problem is decidable but of iterated
exponential complexity. Strictness analysis for first-order
declarations (like f above) turns out to be complete in deterministic
exponential time.
***** Time and place: April 18, 12:30 pm in MJ352 (Bldg. 460) ******
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
Regular AFLB meetings are on Thursdays, at 12:30pm, in MJ352 (Bldg.
460).
If you have a topic you would like to talk about in the AFLB seminar
please let me know. (Electronic mail: broder@su-score.arpa, phones:
(415) 853-2118 and (415) 948-9172). Contributions are wanted and
welcome. Not all time slots for this academic year have been filled
so far.
For more information about future and past AFLB meetings and topics
you might want to look at the file [SCORE]<broder>aflb.bboard .
- Andrei Broder
-------
∂15-Apr-85 2154 BLOCK@SU-CSLI.ARPA SYNTAX PANEL
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 15 Apr 85 21:54:32 PST
Date: Mon 15 Apr 85 21:51:47-PST
From: Ned Block <BLOCK@SU-CSLI.ARPA>
Subject: SYNTAX PANEL
To: folks@SU-CSLI.ARPA, bboard@SU-CSLI.ARPA
TODAY
PANEL DISCUSSION ON SYNTACTIC THEORIES
Tuesday, April 16
1:30
Redwood G-19
The lecturers in our series on syntactic theories for non-linguists
are back, this time in a panel discussion. Joan Bresnan, Geoff
Pullum, and Peter Sells will take questions from the audience (no
initial presentations, so come with questions). This time, linguists
ARE allowed, but they are asked to stick to matters that non-linguists
will have a chance of understanding. If you think of questions in
advance, send them to the panelists so that they can think about them
(BRESNAN@SU-CSLI,PULLUM@SU-CSLI,SELLS@SU-CSLI).
-------
∂16-Apr-85 0800 JOHN@SU-CSLI.ARPA [John Perry <JOHN@SU-CSLI.ARPA>: talk on tacit knowledge by Martin Davies]
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 16 Apr 85 08:00:45 PST
Date: Tue 16 Apr 85 07:56:39-PST
From: John Perry <JOHN@SU-CSLI.ARPA>
Subject: [John Perry <JOHN@SU-CSLI.ARPA>: talk on tacit knowledge by Martin Davies]
To: folks@SU-CSLI.ARPA
Reminder (and notice that the meeting will start about 10:20 instead
of 10:15):
---------------
Date: Tue 9 Apr 85 10:05:48-PST
From: John Perry <JOHN@SU-CSLI.ARPA>
Subject: talk on tacit knowledge by Martin Davies
To: folks@SU-CSLI.ARPA
What: Talk and Discussion
Topic: Tacit Knowledge: Subdoxasticity and Modularity
Who: Martin Davies, Birkbeck College, London
When: 10:15 (a.m.), Tuesday, April 16
Where: Ventura Seminar Room (or other Ventura location
if complications arise)
-------
-------
∂16-Apr-85 0926 BARWISE@SU-CSLI.ARPA Warning
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 16 Apr 85 09:26:02 PST
Date: Tue 16 Apr 85 09:22:39-PST
From: Jon Barwise <BARWISE@SU-CSLI.ARPA>
Subject: Warning
To: Folks@SU-CSLI.ARPA
The talk I am giving today in the logic seminar on shared information
will not be very technical by mathematical standards, but it will
presuppose familiarity with things like the basic theory of ordinals
and with definitions by transfinite recursion. I will be giving a
more expository talk on the same material to the CSLI seminar in late
May.
-------
∂16-Apr-85 1053 LIBRARY@SU-SCORE.ARPA SOCRATES: Searching Techniques RE: Hutchinson's message concerning premature abandonment of the card catalog
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 16 Apr 85 10:53:41 PST
Date: Tue 16 Apr 85 10:47:40-PST
From: C.S./Math Library <LIBRARY@SU-SCORE.ARPA>
Subject: SOCRATES: Searching Techniques RE: Hutchinson's message concerning premature abandonment of the card catalog
To: su-bboard@SU-SCORE.ARPA, faculty@SU-SCORE.ARPA
cc: r.rdh%lots-c%LOTS-C@SU-SCORE.ARPA
Roland raises some important concerns and I will attempt to explain some of
them. In regards to BROWSE, there is a browse command. Unfortunately there
is no documentation on this command. I have sent out some messages about it
and have explained it to patrons of the Math/CS Library. I don't believe
the libraries have been advertising this capability over the past year. In
fact the SOCRATES UPDATE that Roland was referring to explains that there
will be an article in the future about browse. However if you go into the
command mode you can say BROWSE C QA76.# and retrieve in call number order
titles of books on similar topics. This will retrieve not only what is in
the Math/CS Library on computer science but will include books held in all
the libraries.
The question about selecting files is a good one. IF YOU ARE SEARCHING IN THE
COMMAND MODE I WOULD SUGGEST AT THIS POINT TO ALWAYS SELECT A FILE: ie books,
serials, scores etc. The type of search in headings that Roland refers to
seems to only work in the Lookup mode when you indicate that you want to
search an author (A) and a title (T). This seems very confusing to me also.
I'm not sure what Roland is referring to when he says there are not cross
references and that you can not get all the books in the database that have
been written by a single author. I need more information concerning the
problem.
Although I am interested in hearing from anyone concerning problems with
SOCRATES, I am most interested in hearing if those who would be considered
primary users of the Math/CS Library (math, cs, or, stat, ee, ees, csli, cis
etc) would let me know if you are having similar problems searching SOCRATES.
It is also important to realize even when we are no longer filing cards in the
catalog that everyone will need to consult both the online catalog and the
manual catalog if they wish to do a comprehensive search covering a period
of time that dates back before 1973.
Harry Llull
-------
∂16-Apr-85 1356 LIBRARY@SU-SCORE.ARPA SOCRATES: Searching Techniques--the AT index in the Command Mode
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 16 Apr 85 13:56:35 PST
Date: Tue 16 Apr 85 13:38:07-PST
From: C.S./Math Library <LIBRARY@SU-SCORE.ARPA>
Subject: SOCRATES: Searching Techniques--the AT index in the Command Mode
To: su-bboards@SU-SCORE.ARPA
cc: faculty@SU-SCORE.ARPA, r.rdh%lots-c%LOTS-C@SU-SCORE.ARPA,
: ;
Earlier I said that in the command mode you should select a file (ie books)
if you wanted to do an author and title search. However there is an index
(the AT index) that will do the same thing in the headings file. Instead
of inserting the boolean operator yourself--fin a Knuth and t Art-- you
can search by entering fin at Knuth/Art. It is important to become familiar
with the indexes available on SOCRATES. Anyone using SOCRATES should also
have the single green sheet titled A Reference Guide to Socrates. We have
some in the Math/CS Library. On this geen sheet under command mode in box
number 9 you will see the various indexes listed. While online, you can
enter SHO INDEXES and you will receive a list of the indexes available.
You can also do HELP INDEXES for more explanations.
Harry LLull
-------
∂16-Apr-85 1436 HEDGES@SU-SCORE.ARPA Zucker Abstract
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 16 Apr 85 14:36:30 PST
Date: Tue 16 Apr 85 14:12:24-PST
From: Karen Hedges <HEDGES@SU-SCORE.ARPA>
Subject: Zucker Abstract
To: Faculty@SU-SCORE.ARPA
Following is the abstract of Steven Zucker (Robotics Candidate) who will be
visiting and presenting a colloquium on April 23.
From Contours to Waterfalls:
Converging Constraints in Early Vision
Steven W. Zucker
Computer Vision and Robotics Laboratory
Department of Electrical Engineering
McGill University
Montreal, Quebec, Canada
Early vision involves the recovery of those aspects of physical scene
structure that can be inferred from image intensity arrays. As such it is
an inverse problem, the solution to which requires additional constraint.
Many diverse sources of constraint exist, however, from mathematics to
biology, and putting them together is one of the most exciting aspects of
current research in computational vision. We shall illustrate how these
constraints are converging for a number of inverse problems involving
matching and minimization, and shall concentrate on orientation selection,
or the inference of fields of tangents from images, of say, bounding
contours and waterfalls. We show that two related matching problems arise,
one for 1-dimensional contours and another for 2-dimensional flows, and
explore some of their engineering, psychophysical, and physiological
consequences.
-------
∂16-Apr-85 1443 ULLMAN@SU-SCORE.ARPA Summer Workshop
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 16 Apr 85 14:42:54 PST
Date: Tue 16 Apr 85 14:15:52-PST
From: Jeffrey D. Ullman <ULLMAN@SU-SCORE.ARPA>
Subject: Summer Workshop
To: super@SU-SCORE.ARPA
I have some announcements of a summer workshop on Parallel Architectures
and Algorithms to be held in Washington DC.
The dates are 7/15-8/2/85, and attendence pays $1000/week.
It is limited to 20 people, who must be US citizens, and invitation
is by application by May 1.
Any interest?
-------
∂16-Apr-85 1520 YAMARONE@SU-CSLI.ARPA SANDWICHES AT CSLI
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 16 Apr 85 15:19:37 PST
Date: Tue 16 Apr 85 15:13:27-PST
From: Tom Yamarone <YAMARONE@SU-CSLI.ARPA>
Subject: SANDWICHES AT CSLI
To: FOLKS@SU-CSLI.ARPA
THE SANDWICH SERVICE IN A NUTSHELL FOR THE RECENTLY INFORMED:
SEND ORDERS FOR SANDWICHES, FRUIT(APPLES,BANANAS , ORANGES) & MILK (NO FRUIT
JUICES, YET) TO:
(USER:) LUNCH @CSLI
BEFORE 10:30 AM EACH WEEKDAY.
LATE ORDERS WON'T BE TAKEN, BUT THERE WILL BE SEVERAL SANDWICHES AVAILABLE
TO THOSE WHO HAVE SECOND THOUGHTS AT 12:05 PM..
MORE DATA:
THE SANDWICHES ARE MADE FRESH EACH DAY AND DELIVERED TO NUTS AND MUD AT TERMAN.
I GET THEM THERE AND THE VARIETY IS AS FOLLOWS:
ROAST BEEF ON ONION ROLL.
TURKEY ON ONION ROLL.
HAM ON RYE. $2.50
HAM AND SWISS ON RYE.
CHICKEN or TUNA SALAD ON WHOLE WHEAT.
VEGETARIAN(AVACADO &CHEESE)ON WHOLE WHEAT.
THEY'RE GARNISHED WITH THE APPROPRIATE INGREDIENTS..
FRUIT(APPLES ,BANANAS, ORANGES) .25
MILK (HOMOGENIZED, LOWFAT, NONFAT, CHOCOLATE) .25
YOU CAN ORDER BY COMPUTER OR BY PHONE: 49 7-0628 (ASK FOR SUSI)
PAY WHEN YOU PICK UP. EXACT CHANGE IS A PLUS!!!!!!
SEND ALL COMMENTS , ORDERS, COMPLIMENTS TO "LUNCH" @ CSLI..
CSLI SANDWICHES, INC. IS A NON-PROFIT ORGANIZATION.
-------
∂16-Apr-85 1706 ullman@diablo next meeting
Received: from SU-AIMVAX.ARPA by SU-AI.ARPA with TCP; 16 Apr 85 17:06:04 PST
Date: Tue, 16 Apr 85 17:02:40 pst
From: Jeff Ullman <ullman@diablo>
Subject: next meeting
To: nail@diablo
Allen is speaking 11AM tomorrow (4/17) on the Partition Semantics
paper by Cosmadakis and Kanellakis.
The next meeting, 4/24, will be devoted to a discussion of architectures
for a real "NAIL!" system.
Also, on Friday 4/26, 11AM in 252 MJH, Dave Maier will give a special
seminar on his crazy ideas about logical DB's.
Then at 3:15 the same day, I am going to give an introduction
to "capture rules" in the CS345 seminar.
We have a speaker for 5/8: Joe Skudlarek on the *QUEL language.
We need speakers for 5/1, 5/15, ...
Some of the papers I'd like volunteered are:
Sciore/Warren, Towards an integrated DB/Prolog system
Lozinskii, Inference by generating and structuring deductive DBs
Walker: The Syllog Language
Lloyd and Topor: A basis for deductive DB's
I'd also like to know if anything *you* have seen interests you
enough to present it to the group.
Oh yes, copies of "The Real Stone Soup" are now available.
Mrs. Napier has copies.
---Jeff Ullman
∂16-Apr-85 2348 POLLARD@SU-CSLI.ARPA lost receipts
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 16 Apr 85 23:47:09 PST
Date: Tue 16 Apr 85 23:44:23-PST
From: Carl Pollard <POLLARD@SU-CSLI.ARPA>
Subject: lost receipts
To: folks@SU-CSLI.ARPA
I have lost an envelope containing a bundle of receipts and a list of
itemized expenses. HELP! I need these things to be reimbursed for a trip.
If you find them, would you please return them to my mailbox in Trailer A?
Thanks --
Carl Pollard
-------
∂17-Apr-85 0740 RESTIVO@SU-SCORE.ARPA PROLOG Digest V3 #19
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 17 Apr 85 07:40:35 PST
Date: Tuesday, April 16, 1985 5:42AM
From: Chuck Restivo (The Moderator) <PROLOG-REQUEST@SU-SCORE.ARPA>
Reply-to: PROLOG@SU-SCORE.ARPA
US-Mail: P.O. Box 4584, Stanford CA 94305
Phone: (415) 326-5550
Subject: PROLOG Digest V3 #19
To: PROLOG@SU-SCORE.ARPA
PROLOG Digest Wednesday, 17 Apr 1985 Volume 3 : Issue 19
Today's Topics:
Query - Redundant Comparisons,
Implementations - C-Prolog & Optimal Logics & Semantics
----------------------------------------------------------------------
Date: Fri 12 Apr 85 08:43:28-PST
From: PEREIRA@SRI-AI.ARPA
Subject: C-Prolog cloning
It is becoming more and more difficult to keep track
of all the versions of C-Prolog that people mention
on this Digest. In particular, I was surprised to
read that there is anything in sysbits.c in the
version 1.5.edcaad that checks for some file
descriptor being a terminal. You see, I *wrote*
1.5.edcaad, and my copy of the source has no such
thing... Maybe this is a novel manifestation of
self-modifying code :-).
-- Fernando
PS. EdCAAD is not on this net. To get the latest
EdCAAD C-Prolog, write to them using paper mail.
wgr
------------------------------
Date: Sat, 13 Apr 85 02:01:07 est
From: "Dennis R. Bahler" <DRB%Virginia@csnet-relay.arpa>
Subject: Advice for redundant comparisons
I have a problem which could use the advice of an expert
Prolog programmer, which I am not.
I have defined a predicate closure/2, defined over the
power set of an arbitrary set, and wish to test its
instantiations for NONconformance to an axiom, call it
c5, which states that
CL(X int Y) = CL(X) int CL(Y)
(where CL(X) means closure(X), and 'X int Y' means X
intersect Y).
What I have now is:
c5:- closure(X,CX), closure(Y,CY), not(eqset(X,Y)),
intersection(X,Y,IXY), closure(IXY,CIXY),
intersection(CX,CY,ICXCY),
not(eqset(CIXY,ICXCY)),
printerror(5), etc....
with the utility functions defined in reasonable ways.
My problem is that the default execution of Prolog will
test all PERMUTATIONS of pairs of X and Y, i.e.,
X = [a,b] Y = [b,c] and then
X = [b,c] Y = [a,b]
etc.
and thus does exactly twice as much work as necessary.
I would like advice about how to efficiently test non-redundant pairs
of clauses, something on the order of:
for (i=0; i<MAX; i++)
for (j=i+1; j<=MAX; j++)
{ clause[i] <relop> clause[j] }
-- Dennis Bahler
------------------------------
Date: Fri 12 Apr 85 13:22:12-PST
From: Joseph A. Goguen <GOGUEN@SRI-AI.ARPA>
Subject: Optimal Logics for Logic Programming
Several people have asked me for references for the
results about largest logics having initial models.
Here they are:
A. Tarlecki, "Abstract Algebraic Institutions which
Strongly Admit Initial Semantics", University of Edinburgh,
Computer Science Department, report number CSR-165-84, 1984.
B. Mahr and J.A. Makowsky, "Characterizing Specification
Languages which Admit Initial Semantics", Theoretical
Computer Science, volume 31, pages 49-60, 1984.
B. Mahr and J.A. Makowsky, "An Axiomatic Approach to
Semantics of Specification Languages", Proceedings, 6th GI
Conference on Theoretical Computer Science, Dortmund,
Springer-Verlag Lecture Notes in Computer Science, Volume
145, 1983.
J.A. Makowsky, "Why Horn Formulas Matterin Computer Science:
Initial Structures and Generic Examples", Department of
Computer Science, Technion, Haifa, Israel, 1985.
The first report is the most general, covering cases such
as strongly typed first order equational logic with
subsorts. The next three cover only the the Horn clause
cases, and the last (and first) correct some errors in the
second and third reports.
A careful statement of the results showing that first order
Horn clause logic with equality admits initial models in
exactly the same way as equational logic (and thus also
admits user defined generic abstract data types in the same
way as the special case of equational logic) can be found
in our paper on Eqlog in Jnl. Logic Programming, vol.1,
no.2, Theorem 1; of course the result also applies to the
special case of Horn clause logic without equality, which
is (pure) Prolog, where the initial model coincides with the
traditional minimal Herbrand model. Proofs will be found
in the revised and extended version of our JLP paper, to
appear in the book edited by de Groot and Lindstrom
entitled "Functional and Logic Programming".
------------------------------
Date: Sun 14 Apr 85 22:13:34-PST
From: Joseph A. Goguen <GOGUEN@SRI-AI.ARPA>
Subject: Semantics of Logic Programming
I agree with Uday Reddy that control is a rather
neglected grandchild in logic programming. However,
Bob Kowalski's famous slogan "Programming = Logic +
Control" expresses a clean factorization into two
components, so presumably they can be studied and
mastered separately. In fact, control has been
studied and mastered to a considerably lesser extent;
but perhaps we are about to see some real progress
in this area. In particular, I hope that some of
the various annotations being suggested can be given
a clean semantics that doesn't violate the logic
component, i.e., that only tells how to solve the
same problem faster.
I don't understand where Reddy got the idea that Horn
clause logic is {true, undefined}-valued. Every book
on logic or logic programming I ever saw takes the
absolutely traditional view that sentences (in this
case, ground Horn clauses) are either true or false
(the trick of representing them as functions having
values doesn't come up unless you want to translate
clauses into, say, equational logic, as Dershowitz
does); Lloyd's new book is a lovely exposition of
pure logic programming. It may be that Reddy has
a way to use {true, undefined} as a *domain* for
a denotational semantics of Horn clause programs
-- but that is an entirely different kettle of fish,
and I claim a messier one. The basic idea of Horn
clause programming semantics (as in the famous 1976
van Emden and Kowalski JACM paper) is that Horn clause
programs give the answers that are true of the initial
model (of course, they call it the minimal Herbrand
model). So (pure) Prolog is an initial model-based
language. Theorem 1 of our Jnl. Logic Programming
paper explains this (if you work through the math).
Perhaps it's worth giving a bit more detail here. The
first point is that pure Horn clause programming doesn't
involve any equations, so the minimal Herbrand model
contains just terms, rather than equivalence classes of
terms, which makes everything simpler (Eqlog is our
generalization for the case where there are equations);
in particular, equality is just syntactic equality, as
in Prolog. If you add equations, the only difference is
that you compare reduced forms to check equality; but this
is no extra work, since (presumably) you already have the
reduced forms as a result of doing the computations that
you wanted done anyway. This is because term reduction is
actually the (very powerful and efficient, by the way) form
of computation that is specified by the equations -- i.e.,
logic programming with equations has operational semantics
given by term reduction. All this should be familiar from
our OBJ language and also work of O'Donnell.
Non-termination is an interesting issue, both for Prolog
and for other initial model-based languages; we think that
we have a good solution based on the theory of continuous
algebras, in which initial algebras contain infinite
computation trees, but that is too complicated to explain
here, except perhaps to say that it seems to be a fairly
straightforward generalization of the usual theory. By
the way, Prolog II is (in effect) taking a similar line.
Regarding higher-order logic, I agree with David Warren
that we don't really need it in logic programming. Yes,
it is elegant and powerful. But it is hard to read, hard
to write, hard to compute with, and generally altogether
more complicated than first order logic. Even so, we don't
need denotational semantics to explain it, since there is
lots of nice theory in logic, including higher order
equational logic (which will do higher order functional
programming for us, such as Backus' FP) and higher order
model theory; we can even have higher order abstract data
types (due to work by Parsaye and Poigne).
I'd like to repeat what I see as the major moral to be
drawn from much of the discussion in the Digest: we need
more research on making pure logic programming powerful
enough so that we can do practical programming in it
easily and naturally. Good topics include control,
input-output, states (so we can get rid of assert and
retract), combining functional and logic programming,
termination issues (including lazy evlauation and streams),
polymorphic typing, and (if you don't agree that Eqlog has
already solved them) generic modules and subtypes (as for
multiple inheritance).
------------------------------
End of PROLOG Digest
********************
∂17-Apr-85 0807 EMMA@SU-CSLI.ARPA Program:Meeting of Society for Philosophy and Psychology
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 17 Apr 85 08:06:53 PST
Mail-From: BLOCK created at 16-Apr-85 15:46:50
Date: Tue 16 Apr 85 15:46:50-PST
From: Ned Block <BLOCK@SU-CSLI.ARPA>
Subject: Program:Meeting of Society for Philosophy and Psychology
To: su-bboards@SU-CSLI.ARPA
cc: friends@SU-CSLI.ARPA
ReSent-Date: Wed 17 Apr 85 08:02:02-PST
ReSent-From: Emma Pease <Emma@SU-CSLI.ARPA>
ReSent-To: friends@SU-CSLI.ARPA
PROGRAM AND ABSTRACTS FOR MEETING OF SOCIETY FOR PHILOSOPHY AND
PSYCHOLOGY
University of Toronto Wednesday May 15 - Saturday May 18, 1985
For information about the program [note that there may still be
room for some discussants or speakers], the usenet address for
the Program Chairman, Stevan Harnad, is:
bellcore!princeton!mind!srh
or write to: Stevan Harnad, Behavioral & Brain Sciences, 20 Nas-
sau Street, Suite 240, Princeton NJ 08540
For information about local arrangements, write to: David Olson,
McLuhan Center, University of Toronto, Toronto, Ontario, CANADA
M5S 1A1
For information about the Society and attendance, write to: Owen
Flanagan, Secretary/Treasurer, Society for Philosophy & Psycholo-
gy, Philosophy Department, Wellesley College, Wellesley, MA 02181
Workshop (2 full sessions)
Ia & Ib. Artificial Intelligence Versus Neural Modeling in
Psychological Theory
Participants include: D. Ballard, J. Barnden, P. Churchland,
P.C. Dodwell, J. Feldman, A. Goldman, S. Grossberg, S.J. Hanson,
P. Kitcher, W. Lycan, A. Newell, R. Schank, W. Seager.
Symposia (11)
II. Category Formation
Participants include: S. Harnad, G. Hirst, M. Lipton, G.
Matthews, R. Jackendoff, N. Macmillan, R. Millikan, R. Schank.
III. Unconscious Processing
Participants include: T. Carr, P. Kolers, A. Marcel, P. Merikle,
W. Savage, A. Treisman.
IV. Memory and Consciousness
Participants include: K. Bowers, M. Moscovitch, D. Schacter, A.
Marcel, R. Lockhart, E. Tulving.
V. New Directions in Evolutionary Theory
Participants include: E. Balon, O. Flanagan, A. Jensen, A. Rapo-
port, A. Rosenberg, M. Ruse, E. Sober, W. Shields.
VI. Paradoxical Neurological Syndromes
Participants include: O. Flanagan, M. Gazzaniga, A. Kertesz, A.
Marcel, R. Puccetti, O.Sacks.
VII. The Empirical Status of Psychoanalytic Theory
Participants include: M. Eagle, E. Erwin, A. Grunbaum, P. Kar-
vath, J. Masling, B. von Eckardt, R. Woolfolk.
VIII. The Scientific Status of Parapsychological Research
Participants include: J. Alcock, K. Emmett, R. Hyman, C. Honor-
ton, R.L. Morris, M. Truzzi.
IX. The Reality of the "G" (General) Factor in the Measurement
and Modeling of Intelligence
Participants include: D. Detterman, P. Hertzberg, A. Jensen, W.
Rozeboom, R. Traub.
X. The Ascription of Knowledge States to Children: Seeing,
Believing and Knowing
Participants include: D. Olson & J. Astington, J. Perner & H.
Wimmer, M. Taylor & J. Flavell, F. Dretske, S. Kuczaj.
XI. Psychology, Pictures and Drawing
Participants include: J. Caron-Prague, S. Dennis, J. Kennedy, D.
Pariser, S. Wilcox, J. Willats, S. Brison, W. Savage.
XII. Interpretation Versus Explanation in Cognitive and Social
Theory
Participants include: R. DeSousa, A. Grunbaum, S. Harnad, R. Ni-
choloson, A. Rosenberg, E. Sullivan, R. Woolfolk.
Contributed Paper Sessions (4):
XIII. Perception and Cognition (chairman: C. Normore)
To What Extent Do Beliefs Affect Apparent Motion (M. Dawson, R.
Wright) (discussant: P. Kolers)
Images, Pictures and Percepts (D. Reisberg, D. Chambers) (discus-
sant: W. Savage)
What the First Words Tell Us About Meaning and Cognition (A. Gop-
nik) (discussant: G. Matthews)
XIV. Induction and Information (chairman: R. Cohen)
Beyond Holism: Induction in the Context of Problem-Solving (P.
Thagard, K. Holyoak) (discussant: C.F. Schmidt)
The Semantic of Pragmatics (M.A. Gluck, J.E. Corter) (discussant:
D. H. Helman)
About Promises (J. Astington) (discussant: L. Forgerson)
XV. Evolution of Cognitive and Social Structures (chairman: C.
Olsen)
Is Decision Theory Reducible to Evolutionary Biology? (W.E. Coop-
er) (discussant: F. Wilson)
Human Nature, Love and Morality: The Possibility of Altruism (L.
Thomas) (discussant: N. Mrosovsky)
On How to Get Rid of the Craftsman (B. Dahlbom) (discussant: K.
Norwich)
XVI. Inferences About the Mind (chairman: J. Poland)
The Puzzle of Split-Brain Phenomena (S.C. Bringsjord) (discus-
sant: R. Puccetti)
The Mark of the Mental (R. Puccetti) discussant: L. Alanen
Natural Teleology (S. Silvers) (discussant: J. Barnden)
SYMPOSIUM ABSTRACTS FOLLOW:
I. ARTIFICIAL INTELLIGENCE VERSUS NEURAL MODELING IN PSYCHOLOGI-
CAL THEORY
The issues will be discussed at two levels, a practical one (P)
and a foundational one (F). At the practical level the following
two questions will be considered: (P1) Is psychological theory-
building more successful with or without constraints from neuros-
cientific evidence and neuroscientific considerations? (P2) Are
the current differences between models that are neurally motivat-
ed (which tend to be statistical, connectionistic, and lately,
parallel) and models that are not neurally motivated (which tend
to be symbol/sentence manipulative) fundamental differences, and
is one approach more promising than the other?
At the foundational level the questions will be: (F1) What are
the data that psychological theory should account for (behavioral
performance? cognitive competence? real-time topography and exe-
cution? neural activity?)? (F2) Is a successful functional theory
of higher cognitive performance and competence necessarily
"implementation-independent" (i.e., independent of the architec-
ture of the mechanism that embodies it)? Tne issues will be dis-
cussed in the context of actual current work in modeling.
II. CATEGORY FORMATION
Categorization is a fundamental human activity. It is involved in
everything from operant discrimination to perceptual recognition
to naming to describing. Five different approaches to categori-
zation now exist more or less in parallel: (1) The nativist ap-
proach, which holds that there are few, if any, nontrivial induc-
tive categories, and hence that most categories are preformed
[see Symposium V]; (2) the statistical pattern recognition and
multidimensional scaling approach, which computer-models category
formation probabilistically; (3) the artificial intelligence ap-
proach, which models categorization with symbol-manipulation
rules; (4) the natural category approach, which investigates
categorization through reaction time studies and typicality judg-
ments and developmentally; (5) the categorical perception ap-
proach, which investigates categorization through discrimination
and identification studies. These approaches will be presented
and the interaction will aim at a synthesis.
III. UNCONSCIOUS PROCESSING
It is undeniable that most cerebral information processing is un-
conscious. Not only are vegetative functions such as posture and
respiration (as well as automatized, overlearned skills) uncons-
ciously controlled by the brain, but even the basic processes
underlying higher cognitive activity are unavailable to conscious
introspection: No one knows "how" he actually adds two and two,
retrieves a name, recognizes a face. This is what makes cognitive
modeling a nontrivial enterprise. But apart from these basic cog-
nitive processes (about which our ignorance is sufficient to
demonstrate that that they are not conscious), there are some
kinds of processes that are at least normally accompanied by some
awareness of their occurrence. These include the detection,
discrimination and identification of verbal and perceptual in-
puts. New data indicate that even these activities may sometimes
occur without introspective awareness of their occurrence. This
new look at "subliminal perception" and related phenomena in a
contemporary psychophysical, information processing framework
will examine the evidence, methodological criteria and theoreti-
cal interpretations of the newer findings. [See also Symposium
VI.)
IV. MEMORY AND CONSCIOUSNESS
The symposium will examine the distinction between memory (the
consequence of some experience) and remembering (the awareness of
past events), which involves consciousness of a past experience.
The distinction involves the relation between mental processes
that reasonably decribe the performance of intelligent systems
(whether animals, people or machines), that is, "subpersonal"
cognitive psychology, and the intentional mental activities and
states of conscious human adults: "intentional psychology."
V. NEW DIRECTIONS IN EVOLUTIONARY THEORY
Among the current developments in evolutionary theory and their
implications for psychology that will be discussed are: (1) The
"new preformationism," arising chiefly from develomental biology,
according to which there are substantial structural constraints
on the variation on which selection can operate; this implies
that there are structures and functions that cannot be regarded
as having been shaped by random variation and selection by conse-
quences but rather as having arisen from boundary conditions on
biological structures. The issue is particularlly relevant to
questions about the origins of cognitive and linguistic struc-
tures [Symposium II]. (2) Current sociobiological theory has be-
come concerned with cognitive questions, in particular, the ex-
istence of "cognitive primitives" on which selection would
operate in a way that is analogous to its effects on traits coded
by genes: Is this "gene-culture co-evolution" and its new unit,
the "culturgen" just overinclusive curve-fitting or is there a
real empirical phenomenon here? (3) In general, are the kinds of
assumptions and inclusive-fitness calculations that characterize
sociobiological theorizing (and that have been critically re-
ferred to as "just-so stories") a reasonable explanatory handicap
or signs of taking the wrong theoretical direction? In particu-
lar, when is a conscious, cognitive explanation of a behavior
[Symposium III] preferable to an unconscious, fitness-related
one?
VI. PARADOXICAL NEUROLOLOGICAL STATES
This symposium will consider neurological states that (based on
their symptoms and inferences from their symptoms) are very hard
to imagine "being in." These include: (1) "blindsight," i.e., the
loss of all conscious visual experience, but with the retention
of "visual" information (e.g., object location); (2) the anosag-
nosias and attentional disorders, i.e., the apparent unawareness
and denial of dramatic neurological deficits such as loss of
large portions of the visual field or of body sensation; (3)
deconnection phenomena such as alexia without agraphia (intact
vision with the loss of all ability to read but the retention of
the ability to write) or the split-brain patient's ability to
match but inability to name out-of-sight objects grasped with the
left hand; (4) various memory disorders such as the ability to
acquire cognitive information and skills with complete inability
to remember the episodes in which they were acquired [cf. Sympo-
sium IV]; (5) confabulations arising from these paradoxical
states (i.e., the unusual way patients rationalize having these
deficits). The clinical phenomenology of these paradoxical states
will be decsribed and then they will be discussed in terms of
current philosophical, psychological and neurological theories of
cognition and consciousness.
VII. THE EMPIRICAL STATUS OF PSYCHOANALYTIC THEORY
The empirical status of psychoanalytic theory will be considered
in terms of the following questions: (1) Is psychoanalytic theory
testable? (2) If so, how much of it is testable, and, in particu-
lar, what parts? (3) How is it testable (clinically? experimen-
tally? epidemiologically?)? (4) How much of psychoanalytic theory
has actually been tested in these ways, and was the theory sup-
ported by the evidence? (5) Are future tests of psychoanalytic
theory likely to yield outcomes that support the theory, and is
this theory the best one to use to guide future research? (6) Is
the proportion of psychoanalytic theory that is testable compar-
able to the proportions of other scientific theories that are te-
stable, or is evidence disproportionately remote from or ir-
relevant to psychoanalytic theory? (7) Is testability irrelevant
to some kinds of theoretical understanding? (8) Is psychoanalytic
theory based on adequate views of conscious and unconscious
processes and explanation? These questions will be discussed by
clinicians, experimentalists and methodologists of science.
VIII. THE SCIENTIFIC STATUS OF PARAPSYCHOLOGICAL RESEARCH
In parapsychology there appears to be a chronic polarization of
rival views in a way that only occurs occasionally and briefly at
the frontiers of other kinds of scientific research. The polari-
zation consists of those who accept the validity of the reported
phenomena and of the theoretical framework accounting for them
and those who do not. The following questions will be considered:
(1) Is the polarization merely a prejudice, or are there objec-
tive characteristics that set this field of research apart? (2)
Are there special problems with furnishing replicable positive
evidence in this area? (3) Are there logical problems with the
theoretical framework in which the research is undertaken? (4)
Are there statistical problems with the data-analysis and the
underlying assumptions? (5) Is there any possibility of resolu-
tion, or will the field always continue to split among believers
and nonbelievers, and if the latter, (6) what does that imply
about the scientific validity of this domain of inquiry? These
questions will be discussed, in the context of representative
current experimental work in parapsychological research, by
parapsychologists, skeptics and (as yet) uncommitted methodolo-
gists.
IX. THE REALITY OF THE "G" (GENERAL) FACTOR IN THE MEASUREMENT
AND MODELING OF INTELLIGENCE
When intelligence tests are factor-analyzed (i.e., the structure
of their correlations with one another is reduced to a small
number of underlying variables), one general, overall factor al-
ways emerges, along with a number of special factors peculiar to
some groups of tests and not others. The general ("g") factor has
been interpreted as a unitary measure of general intelligence.
Some have challenged the reality of "g" on the grounds that indi-
vidual test items (and indeed entire tests) are so constructed as
to correlate with one another, and hence the overall positive
correlation factor is built in; moreover, it is argued that it is
fallacious to think in terms of an underlying, one-dimensional
unitary intelligence. Others have argued that "g" is an empirical
finding after all, because even tests constructed and validated
to measure the special abilities (e.g., verbal versus spatial
skills) have high "g" loadings, and indeed the more discriminat-
ing tests (the ones that are more sensitive to and predictive of
individual differences) tend to have the higher "g" loadings. The
technical and conceptual problems of measuring, validating and
modeling human cognitive capacities will be discussed in the con-
text of the interpretation of "g."
X. THE ASCRIPTION OF KNOWLEDGE STATES TO CHILDREN: SEEING,
BELIEVING AND KNOWING
Considerable discussion in cognitive science surrounds the issue
of the ascription of beliefs to animals, machines and young chil-
dren. Opinions range from that of Davidson, who argues that one
cannot have beliefs unless one has a concept of belief, to that
of Searle, who argues that "only someone in the grip of a philo-
sophical theory would deny that dogs and children have beliefs."
Recent research on children's ascription of beliefs to others and
to themselves in the interpretation of visual events may cast
some light on this question.
XI. PSYCHOLOGY, PICTURES AND DRAWING
The past decade has seen considerable interest in theory of dep-
iction and allied theories of drawing. Current theories are
technically well constructed, significant in themselves and, in
addition, have important implications for neighboring areas of
psychology. Yet they are often distinct in the assumptions they
make about perception, communication and the environment. The
present symposium draws together philosophers, educators and
psychologists who have developed theories about pictures, percep-
tion and drawing. Assumptions will be reviewed and implications
will be discussed.
XII. INTERPRETATION VERSUS EXPLANATION IN COGNITIVE AND SOCIAL
THEORY
The following questions will be considered: (1) What is an expla-
nation, and is "scientific" explanation an atypical case or a
paradigmatic one? (2) What is the role of testability and falsi-
fiability in explanation? (3) What is the role of considerations
of satisfyingness, coherence, elegance and other subjective cri-
teria in explanation? (4) Are there different explanatory metho-
dologies in the natural sciences and ther "human" sciences? (5)
Is there an objective way to choose among rival interpretations?
(Should there be? Is there one in the case of rival scientific
theories?) (6) Is there anything objective to replace the outmod-
ed "positivistic" stereotype? Pro and antihermeneuticists will
participate and the discussion will focus on the role of in-
terpretation in psychological and social scientific theory.
PROVISIONAL TIMETABLE
Wed am: VII vs. XIII (parallel sessions)
Wed pm: III vs. XIV
Wed eve: VI
Thurs am: II
Thurs pm: Ia
Thurs eve: Ib
Fri am: IV vs. XVI
Fri pm: II vs XV
Fri eve: (presidential address and business meeting)
Sat am: V vs. XI
Sat pm: VIII vs X
Sat eve: XII
SEND COMMENTS TO STEVAN HARNAD,
srh%mind%princeton%bellcore@ucb-vax
-------
∂17-Apr-85 1540 TAJNAI@SU-SCORE.ARPA Fortune 500/Bay Area Companies
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 17 Apr 85 15:33:47 PST
Date: Wed 17 Apr 85 14:05:40-PST
From: Carolyn Tajnai <TAJNAI@SU-SCORE.ARPA>
Subject: Fortune 500/Bay Area Companies
To: Faculty@SU-SCORE.ARPA
cc: : ;
There are 19 Bay Area firms on the new Fortune 500 list.
Only 3 belong to the Forum: H-P, Intel, and Tandem
There are 7 other companies who have been invited to join -- 2 once belonged,
but dropped membership.
11. (9 in 1983) $26.8 billion - CHEVRON was invited to join in
1982 and 1984.
225. (261 in 1983) $1.7 billion - NATIONAL SEMICONDUCTOR belonged from
7/79 - 6/80; invited to rejoin in 1982.
234. (299 in 1983) - $1.5 billion - APPLE belonged from 10/82 through
9/83; dropped when they went through a "belt tightening" period.
317. (317 in 1983) $928 million - VARIAN ASSOCIATES - invited
to join 1984, 1985. Allen Bennett, Vice President, said in January
there was a 90% chance Varian would join the Forum in 1985.
357. (350 in 1983) $779.4 million - AMDAHL belonged from 1/78 through
12/81. Invited to rejoin in 1982 and 1984.
397. (397 in 1983) $659 million - ROLM was invited to join in 1982 and 1984.
414. (new on list) $583 million - ADVANCED MICRO DEVICES invited
to join in 1983.
If you have contacts with any of these companies, and wish assistance
in recruiting them, please let me know.
Carolyn
-------
∂17-Apr-85 1550 ullman@diablo
Received: from SU-AIMVAX.ARPA by SU-AI.ARPA with TCP; 17 Apr 85 15:50:36 PST
Date: Wed, 17 Apr 85 14:39:06 pst
From: Jeff Ullman <ullman@diablo>
To: nail@diablo
These are some random notes designed to provide a framework
for thinking about a "NAIL!" system.
1. We need some interesting paradigms.
How about a CAD system, where cells are described by
defining predicate symbols ala Prolog?
"Queries" might be: what are the design rule violations in...?
How about a system to analyze parsed programs for style?
What other problems might naturally make use of both recursion
and function symbols?
2. We ought to think in terms of the DBMS dichotomy DDL vs. DML,
i.e., data definition needs a language different from data manipulation
(querying). However, where do the rules go?
Part of their function is DDL-like: defining views, security;
part is DML like: asserting a rule is like entering a tuple into a
relation. I prefer not to deal with assertion and retraction of rules
during the processing of a query, so maybe the DML is trivial
(=goals), the rules correspond to the usual DML and the structure
of the database to the DDL. If so, what do you want in the
way of structure information for relations? FDs? Partition dependencies?
Can any of that info. be used in query optimization?
3. What is the rule language? Something like prolog, only "purer."
There are some uglinesses of Prolog that we can probably improve
upon. What about the role of negation? Do we support just "cons"
or arbitrary function symbols? How about allowing arbitrary, but
regarding f(a1,...,an) as shorthand for "f"|a1|...|an (note quotes
around f only. Clean way to handle comparison atoms, e.g., a<=b?
4. Are we getting close to the point where we know all the
"capture rules" we might want to try? How do we design the
system to provide hooks to insert more rules?
How does declared information, e.g., FDs, get used? Or can it be?
5. Big Issue: Do we construct the system to work top-down
or bottom-up, as far as capturing nodes is concerned?
How does our choice interact with the fact that we don't
know all the capture rules at the outset?
6. Big Issue: How do we respond to the user when we cannot
figure out how to capture the query node? In general,
what feedback and interaction should we be using?
7. Just today I got word that IBM is promising to provide us
with non-product PC's that run Prolog interfaced to System/R.
Unfortunately, the time frame isn't specified.
Should we plan on Prolog as the "intermediate language"
into which queries are translated?
Prolog calls to System/R can be simulated temporarily (or even
permanently) by calls to AWK and JOIN, with the relations
returned by these calls translated into lists of Prolog statements.
8. What are the reasonable ways to write a system that
processes queries without having to write the underlying DBMS?
(I'll leave it to you whether (7) gets included under "reasonable ways."
9. What should be the system implementation language?
Obviously we could agree on anything from C or Modula II, to Lisp.
There is a reason to struggle with Prolog: if we force ourselves
to do so, we'll get a feel for its strengths and weaknesses that
will be important in our own work.
∂17-Apr-85 1555 LIBRARY@SU-SCORE.ARPA Mathematical People--New Book in the Math/CS Library
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 17 Apr 85 15:54:59 PST
Date: Wed 17 Apr 85 12:04:19-PST
From: C.S./Math Library <LIBRARY@SU-SCORE.ARPA>
Subject: Mathematical People--New Book in the Math/CS Library
To: su-bboards@SU-SCORE.ARPA
cc: faculty@SU-SCORE.ARPA, veinott@SU-SIERRA.ARPA, cottle@SU-SIERRA.ARPA
Mathematical People by Albers and Alexanderson (QA28.M37 1985) has arrived
in the Math/CS Library and is on the new books shelf for sign-ups. This
is the book that includes interviews with Don Knuth, Persi Diaconis, and
George Polya all of Stanford. Interviews of the following people are
also included:
Garrett Birkhoff, David Blackwell, Shing-shen Chern, John Horton Conway,
H. S. M. Coxeter, Paul Erdos, Martin Gardner, Ronald L. Graham, Paul R.Halmos,
Peter J. Hilton, John Kemeny, Morris Kline, Benoit Mandelbrot, Henry O.
Pollak, Mina Rees, Constance Reid, Herbert Robbins, Raymond Smullyan,
Olga Taussky-Todd, Albert W. Tucker, Stanislaw M. Ulam, and Reminiscences
of Solomon Lefschetz by Albert W. Tucker.
Harry Llull
-------
∂17-Apr-85 1556 LIBRARY@SU-SCORE.ARPA New Books in the Math/CS Library
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 17 Apr 85 15:56:17 PST
Date: Wed 17 Apr 85 12:13:06-PST
From: C.S./Math Library <LIBRARY@SU-SCORE.ARPA>
Subject: New Books in the Math/CS Library
To: su-bboard@SU-SCORE.ARPA
cc: faculty@SU-SCORE.ARPA
Human-Computer Interaction. Proceedings of the first USA-Japan conference
on Human-Computer Interaction, Hawaii, 1984. edited by Salvendy
QA76.9.I58U17 1984
Calculating Machines Recent and Prospective Developments and their impact
on Mathematical Physics (1947) and Calculating Instruments and Machines(1949)
by Douglas R. Hartee. Volume VI in the Charles Babbage Institute Reprint
Series for the History of Computing. QA85.H3 1984
Inside Concurrent CO/M; a guide for users. by Cortesi QA76.6.C6648 1984
Total Poker by Spanier. GV1251.S6 1979
Mass Storage Systems. IEEE Computer Society. TK7895.M4I33 1984
Microprocessor Operating Systems. by John Zarrella. 3 volumes.
QA76.6.M486 v. 1, 2, 3.
HLlull
-------
∂17-Apr-85 1622 VARDI@SU-CSLI.ARPA Comments
Received: from [36.9.0.46] by SU-AI.ARPA with TCP; 17 Apr 85 16:22:01 PST
Date: Wed 17 Apr 85 16:17:37-PST
From: Moshe Y. Vardi <VARDI@SU-CSLI.ARPA>
Subject: Comments
To: barwise@SU-CSLI.ARPA
cc: finterest@SU-CSLI.ARPA, yom@SU-AI.ARPA, fagin.ibm-sj@CSNET-RELAY.ARPA
1. Why do people stop at omega when they talk about shared information?
I think the reason is that in Kripke structures [] distributes over /\,
that is [](p/\q) is equivalent to []p/\[]q. Furthermore, [] also distributes
over infinite conjunction. Thus the hierarchy collapses at the omega-th level.
In your semantics, and in other semantics (such as Montague Semantics),
S does not distributes over conjunction. Thus you cannot really stop at
any prescribed ordinal.
2. The issue of circularity arises also in possible worlds semantics.
There a world consists of a truth assignment + a set of possible worlds.
This is captured not in a set-theoretic way, but in a graph-theoretic way,
in Kripke structures. That is, rather than use AAFAA, set theory is left
intact, and worlds are defined in a graph-theoretic way.
Ron Fagin and myself have been troubled by a feeling that is closely related
to your now abandoned intuition that objects ought to be defined in a
well-founded way. Our internal semantics essentially replaces worlds
by well-founded approximations. Standard modal logic formulas talk about
worlds only implicitly and they have bounded depth. Therefore, omega-
approximations are sufficient. When formulas talk about mutual knowledge,
you need omega↑2 levels, but there is a way to collapse it to omega levels.
When you have infinitary formulas, you need all levels of the approximation.
3. The mu-calculus is a formalism for non-well-founded formulas, whose
basic idea is that the semantics of such formulas is defined via fixpoints.
Computer scientists have studied have studied propositional modal mu-calculus.
In such a formalism it is easy to express mutual knowledge. If KaP (KbP)
stands for "a (b) knows P", then nu-X (KaP/\KbP/\X), where nu is the greatest
fixpoint operator, expresses that P is the mutual knowledge of a and b.
Mu-calculus formulas do not have an apriori depth. Nevertheless, for any
particular Kripke structure the fixpoints converges at a certain ordinal.
Thus if you take all levels of well-founded approximations, that should be
sufficient. It seems to be that only when you have an explicit language,
you cannot well-founded approximations.
4. Interestingly, non-well-founded sets arise also in databases: records
can point to themselves. The formalism suggested there seems closely
related to Aczel's.
Moshe
-------
∂17-Apr-85 1700 @SU-SCORE.ARPA:ARK@SU-AI.ARPA CS Colloq, April 23: Steven Zucker, From Contours to Waterfalls: Converging Constraints in Early Vision
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 17 Apr 85 17:00:20 PST
Received: from SU-AI.ARPA by SU-SCORE.ARPA with TCP; Wed 17 Apr 85 16:56:08-PST
Date: 17 Apr 85 1655 PST
From: Arthur Keller <ARK@SU-AI.ARPA>
Subject: CS Colloq, April 23: Steven Zucker, From Contours to Waterfalls: Converging Constraints in Early Vision
To: Faculty@SU-SCORE.ARPA, All-Colloq@SU-SCORE.ARPA
CC: ARK@SU-AI.ARPA
CS Colloquium, April 23, 4:15pm, Terman Auditorium
FROM CONTOURS TO WATERFALLS:
CONVERGING CONSTRAINTS IN EARLY VISION
Steven W. Zucker
Computer Vision and Robotics Laboratory
Department of Electrical Engineering
McGill University
Montreal, Quebec, Canada
Early Vision involves the recovery of those aspects of physical scene
structure that can be inferred from image intensity arrays. As such it is
an inverse problem, the solution to which requires additional constraint.
Many diverse sources of constraint exist, however, from mathematics to
biology, and putting them together is one of the most exciting aspects of
current research in computational vision. We shall illustrate how these
constraints are converging for a number of inverse problems involving
matching and minimization, and shall concentrate on orientation selection,
or the inference of fields of tangents from images of, say, bounding
contours and waterfalls. We show that two related matching problems
arise, one for 1-dimensional contours and another for 2-dimensional flows,
and explore some of their engineering, psychophysical, and physiological
consequences.
----
Cookies, thanks to Ginger, and possibly juice will probably be served in
the 3rd floor lounge of MJH at 3:45pm. See you there.
∂17-Apr-85 1751 @SU-SCORE.ARPA:udi@wisc-rsch.arpa
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 17 Apr 85 17:51:29 PST
Received: from wisc-rsch.arpa by SU-SCORE.ARPA with TCP; Wed 17 Apr 85 17:49:45-PST
Received: by wisc-rsch.arpa; Wed, 17 Apr 85 19:26:41 cst
Received: from wisc-crys.arpa by wisc-rsch.arpa; Tue, 16 Apr 85 21:35:30 cst
Message-Id: <8504170335.AA19430@wisc-crys.arpa>
Received: from csnet-relay.arpa by wisc-crys.arpa; Tue, 16 Apr 85 21:35:09 cst
Received: from iowa-state by csnet-relay.csnet id ab00706; 16 Apr 85 22:32 EST
Received: by isucs1.UUCP (4.12/4.7)
id AA05427; Tue, 16 Apr 85 09:17:14 cst
Date: Tue, 16 Apr 85 09:17:14 cst
From: Alan Selman <selman%isucs1@csnet-relay.arpa>
To: udi@wisc-crys.ARPA
Sender: udi@wisc-rsch.arpa
Resent-From: udi@wisc-rsch.arpa (Udi Manber)
Resent-To: theory:;@wisc-rsch.arpa
←λS←λt←λr←λu←λc←λt←λu←λr←λe ←λi←λn ←λC←λo←λm←λp←λl←λe←λx←λi←λt←λy ←λT←λh←λe←λo←λr←λy ←λC←λo←λn←λf←λe←λr←λe←λn←λc←λe
←λA←λn←λn←λo←λu←λn←λc←λe←λm←λe←λn←λt ←λa←λn←λd ←λC←λa←λl←λl ←λf←λo←λr ←λP←λa←λp←λe←λr←λs
A conference is being planned that will focus on global
aspects of computational complexity theory and structural
properties of both complexity classes and complexity-
bounded reducibilities. This conference will take place
at the University of California, Berkeley, June 2 through
June 5, 1986, the week immediately following the ACM -
STOC Symposium to be held on the Berkeley campus also.
Topics of interest will include, but not be limited to,
the following issues in complexity theory:
Structure of complexity classes
Properties of NP-complete sets
Relations between complexity classes
Resource-bounded reducibilities
Theory of relativizations
Recursion theoretic aspects
Kolmogorov complexity and randomness
Crypto-complexity
Applications of finite model theory
Independence results
Original research papers and technical expository talks
are sought. Authors can anticipate 40 minutes for
presenting research papers and 60 minutes for presenting
expository talks. Send 10 copies of an extended abstract
or full draft paper by ←λN←λo←λv←λe←λm←λb←λe←λr ←λ2←λ9, ←λ1←λ9←λ8←λ5 to
Alan L. Selman
Computer Science Department
Iowa State University
Ames, Iowa 50011, U.S.A.
The Program Committee consists of Ron Book, Juris Hart-
manis, Harry Lewis, Steve Mahaney (co-Chair), Ken McAloon,
Alan Selman (co-Chair), Mike Sipser, Peter van Emde Boas,
and Paul Young. Some members of the program committee will
also present talks, some of which will be technical expo-
sitory talks providing perspective on their current
research programs.
Springer-Verlag has agreed to publish Conference Proceed-
ings as an issue in the Lecture Notes in Computer Science
series. Authors will be notified of acceptance or rejec-
tion by January 31, 1986. Final papers are due March 21,
1986. Final plans for this conference will depend on
funding from the National Science Foundation.
∂17-Apr-85 1800 EMMA@SU-CSLI.ARPA Newsletter Apr. 18, No. 25
Received: from [36.9.0.46] by SU-AI.ARPA with TCP; 17 Apr 85 17:59:51 PST
Date: Wed 17 Apr 85 17:11:18-PST
From: Emma Pease <Emma@SU-CSLI.ARPA>
Subject: Newsletter Apr. 18, No. 25
To: friends@SU-CSLI.ARPA
Tel: 497-3479
C S L I N E W S L E T T E R
←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←
April 18, 1985 Stanford Vol. 2, No. 25
←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←
A weekly publication of The Center for the Study of Language and
Information, Ventura Hall, Stanford University, Stanford, CA 94305
←←←←←←←←←←←←
CSLI ACTIVITIES FOR *THIS* THURSDAY, April 18, 1985
12 noon TINLunch
Ventura Hall A. P. Martinich's ``A Theory for Metaphor''
Conference Room Discussion led by Paul Schacht
2:15 p.m. CSLI Seminar
Redwood Hall ``The Formality of Computation''
Room G-19 Brian Smith, Xerox PARC and CSLI
Discussion led by Stan Rosenschein
(Abstract on page 2)
3:30 p.m. Tea
Ventura Hall
4:15 p.m. CSLI Colloquium
Redwood Hall ``Two Examiners Marked Six Papers: Interpretations
Room G-19 of Numerically Quantified Sentences''
Martin Davies, Birkbeck College, U. of London
←←←←←←←←←←←
CSLI ACTIVITIES FOR *NEXT* THURSDAY, April 25, 1985
12 noon TINLunch
Ventura Hall ``Cell Psychology: An Evolutionary Approach to
Conference Room the Symbol-Matter Problem'' by H. H. Pattee
Discussion led by Ivan Blair, CSLI
(Abstract on page 2)
2:15 p.m. CSLI Seminar
Redwood Hall ``Whither CSLI?''
Room G-19 John Perry, Director, CSLI
(Abstract on page 3)
3:30 p.m. Tea
Ventura Hall
4:15 p.m. CSLI Colloquium
Redwood Hall ``The Representational Basis for Everyday Aesthetic
Room G-19 Experience -- A Motivational Constraint on Learnable
Systems of Knowledge''
Tom Bever, Columbia University and CASBS
(Abstract on page 3)
!
Page 2 CSLI Newsletter April 18, 1985
←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←
ABSTRACT OF THIS WEEK'S SEMINAR
``The Formality of Computation''
Most people would agree that computation is ``formal'' -- as for
example in the claim that computation is ``formal symbol manipulation''.
I will argue, however, that there is no reading of the term ``formal''
under which such a claim is both interesting and true. More
specifically, I will argue that the conditions we have called formality
are really projections into the symbolic or representational domain of
constraints arising from physical embodiment. As a consequence, I
will propose that a proper account of computation -- especially one
designed to mesh with other concerns at CSLI -- should set aside the
notion of formality and pay much more attention to matters of embodiment.
Note that I will NOT claim:
(i) that ``formal'' doesn't mean anything at all (in fact I will
propose several different coherent things it can mean);
(ii) that computation isn't representational (only that adding the
term ``formal'' doesn't buy you anything extra);
(iii) that current computers aren't formal, under at least one
coherent reading of the term (in fact I think they are).
I will also argue that we should distinguish two different reigning
notions of computation: one digital, one representational. Most
current computer systems, I will suggest, are digital, representational,
and formal. My claim is only that these three notions differ in
meaning, and could differ in extension as well. Only the first two
seem likely candidates in terms of which to define the notion(s) of
computation. --Brian Smith
←←←←←←←←←←←←
ABSTRACT OF NEXT WEEK'S TINLUNCH
``Cell Psychology:
An Evolutionary Approach to the Symbol-Matter Problem''
The central problem that Pattee is concerned with is that of the
relation between a symbol and its referent, what he calls the
matter-symbol problem. In other papers, he draws the same basic
distinction in terms of a dynamic (physical) mode and a linguistic
(symbolic) mode of operation of certain systems. Pattee has
consistently argued that the matter-symbol problem occurs at the level
of the single cell (genetic symbol and phenotypic referent, where this
terminology is to be taken seriously and literally, not
metaphorically), and that we should study this comparatively simple
example, using it as a test case for theories about the nature of
representation. One question to which we would like an answer is,
What distinguishes physical interaction from genuine symbol
manipulation?
In this article, Pattee considers both the information processing
approach and ecological realism, arguing that neither can offer any
real insight into the central problem, since each ignores one half of
it. The computational approach typically omits constraints that could
arise from material structures (instantiation in specific hardware)
from consideration and focuses on the more abstract level of programs,
while ecological realism ignores totally the symbolic side of the
problem. Adoption of formal, or logical concepts in the one case, and
physical concepts in the other, prohibits formulation of the problem.
Pattee's own approach, which embraces both the perspectives of symbol
processing and physical dynamics, attempts to relate these and
provides necessary conditions for a system to be dealing in symbols.
!
Page 3 CSLI Newsletter April 18, 1985
←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←
ABSTRACT OF NEXT WEEK'S SEMINAR
``Whither CSLI?''
Since no other forum is conveniently available, I am going to abuse
the seminar format for a practical talk rather than a report of
research.
I will discuss the problems and opportunities facing CSLI, and how
they relate to budget, computing environment, future funding
opportunities and the structure of our research effort. --John Perry
←←←←←←←←←←←←
ABSTRACT OF NEXT WEEK'S COLLOQUIUM
``The Representational Basis for Everyday Aesthetic Experience --
A Motivational Constraint on Learnable Systems of Knowledge''
The structure of everyday aesthetic judgements depends on computations
of mental representations and relations between representations.
Examination of objects of everyday aesthetic preference (e.g., simple
rhythms, shapes, and songs) affords a definition of the aesthetically
satisfying experience: such experiences involve the formation of
incompatible representations and their resolution within the framework
of an overarching representational system. The enjoyment of such
experiences follows from the extent to which they are like solving a
problem during normal cognitive development. Indigenous systems like
language must have formal properties that stimulate aesthetically
satisfying experiences as an immediate motivation for the acquisition of
abstract structures. That is, we learn a multi-levelled representational
structure for language because it is fun. --Tom Bever
←←←←←←←←←←←←
LINGUISTICS DEPARTMENT COLLOQUIUM
``Deterministic Parsing and Subjacency''
Janet Dean Fodor, University of Connecticut and CSLI
Rm. 200-217 (History Corner), Tuesday, April 23, 3:15
Berwick and Weinberg (1984, MIT Press) claim that their
deterministic parser predicts the existence and range of application
of Subjacency, and that this provides simultaneous support for
Subjacency as the correct description of the linguistic facts, and for
deterministic models of the human sentence parsing mechanism.
I argue that the determinism of their parser is irrelevant to their
predictions, and that in any case all four of their predictions are
false.
←←←←←←←←←←←←
AREA P-2 MEETING
``Lexical Phonology and Tone in Temne''
Will Leben, CSLI
Ventura Conference Room, Wednesday, April 24, 4:30
-------
∂17-Apr-85 1946 JF@SU-SCORE.ARPA Coffee and Other Drugs
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 17 Apr 85 19:46:29 PST
Date: Wed 17 Apr 85 19:35:29-PST
From: Joan Feigenbaum <JF@SU-SCORE.ARPA>
Subject: Coffee and Other Drugs
To: csd@SU-SCORE.ARPA
The Bottom Line (so you don't have to read on if you don't care): This
Department should get some Peet's coffee, a grinder, some decent tea, and
some powdered cocoa mix.
Introductory Anecdote: When I was a senior in college, I hung around the
Math Department lounge a lot. One of the junior faculty used to buy, bring,
grind, and drip his own beans. The rest of the facculty, as well as the
graduate students and the undergraduates, drank the muck that the department
provided, which I seem to recall tasted a lot like the muck that this
department provides. One day Professor RealCoffee said to Professor Random
something like, ``I don't know how you can drink that coffee.'' Random's
reply: ``This isn't coffee, it's math juice.''
Well, it seemed funny at the time, but today I had my first cup of department
muck in about one year (during the past year I have been following the
example of role models of yore and bringing my own coffee and brewing it
one cup at a time; today I got impatient, poured a cup of muck, tasted it,
and decided it was time to flame). Anyone who expects us to do serious
research on such deplorable stimulants is simply expecting too much.
I propose that we adopt one of the following plans:
1) Get a grinder and start buying good beans, both regular and decaf, from
Peets. Also buy some tea and powdered cocoa mix. Perhaps some instant soup--
I don't know what the non-addicted among you sip. Provide all this free of
charge to department members and visitors.
2) Same as 1, except don't provide it free of charge. Just charge an
accordingly higher per-month and per-cup fee to make up the difference in
cost between coffee and muck.
I feel quite strongly that 1) is a superior option but can live with 2).
In answer to the unavoidable charge that ``non-coffee drinkers should not have
to subsidize coffee drinkers, as they would be doing if department funds went
for this purpose rather than for something everyone uses'' I say
A) What does ``everyone use''? Colloquium cookies? TGIF? Sushi?
Boise? HOGWASH!!!!! These things are provided, everyone is
welcome to use them, but it's quite apparent (and quite reasonable)
that not everyone does.
B) Xerox PARC, SRI, CIS, ERL, and various other cs-type establishments
around here provide good coffee for employees and visitors. There
haven't been any serious charges of subversive, unAmerican
tendencies or discrimination-on-the-basis-of-drug-nonaddiction
leveled against these places.
C) It's thinking like option (2) that gives us New Jersey Turnpike
drivingconditions and thinking like option (1) that gives
us Highway 280. If **I** am advocating the Californian way of
doing things, it must have some inherent merit.
How about it, coffee drinkers?
-------
∂17-Apr-85 2347 @SU-SCORE.ARPA:SCHOEN@SUMEX-AIM.ARPA Re: Coffee and Other Drugs
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 17 Apr 85 23:46:51 PST
Received: from SUMEX-AIM.ARPA by SU-SCORE.ARPA with TCP; Wed 17 Apr 85 23:32:10-PST
Date: Wed 17 Apr 85 23:22:30-PST
From: Eric Schoen <Schoen@SUMEX-AIM.ARPA>
Subject: Re: Coffee and Other Drugs
To: JF@SU-SCORE.ARPA, csd@SU-SCORE.ARPA
In-Reply-To: Message from "Joan Feigenbaum <JF@SU-SCORE.ARPA>" of Wed 17 Apr 85 19:50:32-PST
Here, here! If cost is a question, perhaps we can find out what the
tab at Welch Rd runs. We don't stock Peet's, but we do stock random
University Stores Hills Bros, plus coffee, tea, hot chocolate, and
Cup-a-Salt.
Eric
-------
∂18-Apr-85 0944 @SU-SCORE.ARPA:WASHINGTON@SUMEX-AIM.ARPA POTLUCK (last chance!)
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 18 Apr 85 09:44:10 PST
Received: from SUMEX-AIM.ARPA by SU-SCORE.ARPA with TCP; Thu 18 Apr 85 09:28:35-PST
Date: Thu 18 Apr 85 09:28:42-PST
From: Rich Washington <washington@SUMEX-AIM.ARPA>
Subject: POTLUCK (last chance!)
To: csd@SU-SCORE.ARPA
tick-tick-tick-tick-tick-tick-tick.....
For those of you who have been procrastinating -- hey, its okay, we do
it, too, but...
TIME'S RUNNING OUT!!!
Nils' house is almost completely packed, but we can squeeze in a few more
bodies at the potluck next Friday. Naturally, everybody else is going to
be responding to this message him/her/it/them/ter/ti/wuoeir/-self, so by
the time you get around to reading this, it will probably be TOO LATE.
So ...
WAKE UP!
Get yourself out of that drug-induced stupor and
WRITE BACK NOW!
If you're still debating what to bring, we're getting a reasonable
balance, but, as usual, main dishes are the least plentiful, so we'd
appreciate any contributions along those lines. We might even feel
inclined to relax our cutoff a bit if we knew we were going to get a
few more main dishes out of it.
Anyhow, hope to hear from you (if you've already responded, you can
ignore this message [wouldn't it have been nice if we had said that
at the beginning?])
-rog crew@score
-rich washington@sumex
---------------
-------
∂18-Apr-85 1032 @SU-SCORE.ARPA:ullman@diablo CIS request for proposals
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 18 Apr 85 10:31:56 PST
Received: from diablo by SU-SCORE.ARPA with TCP; Thu 18 Apr 85 10:26:14-PST
Date: Thu, 18 Apr 85 10:27:00 pst
From: Jeff Ullman <ullman@diablo>
Subject: CIS request for proposals
To: faculty@score
There is a May 15 deadline for "research initiation"
proposals to CIS. They are limited to 50K/year (but there
is no overhead), and you have to be a CIS affiliate to apply.
I have a copy of the "RFP" if you need one.
∂18-Apr-85 1207 BRAD@SU-CSLI.ARPA Odd Thursday lecture cancelled
Received: from [36.9.0.46] by SU-AI.ARPA with TCP; 18 Apr 85 12:07:25 PST
Date: Thu 18 Apr 85 12:04:28-PST
From: Brad Horak <BRAD@SU-CSLI.ARPA>
Subject: Odd Thursday lecture cancelled
To: folks@SU-CSLI.ARPA
Today's CCRMA Odd Thursday lecture has been cancelled. Alan Kay has the flu.
--Brad
-------
∂18-Apr-85 1233 @SU-SCORE.ARPA:GELMAN@SUMEX-AIM.ARPA Coffee / N.J.Turnpike
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 18 Apr 85 12:33:00 PST
Received: from SUMEX-AIM.ARPA by SU-SCORE.ARPA with TCP; Thu 18 Apr 85 12:22:05-PST
Date: Thu 18 Apr 85 12:22:15-PST
From: Andrew Gelman <GELMAN@SUMEX-AIM.ARPA>
Subject: Coffee / N.J.Turnpike
To: csd@SU-SCORE.ARPA
I agree with Joan wholeheartedly on her Coffee Plan #1. A dark French roast
would do nicely, or Kona for the insecticide-conscious.
A note of dissent about Rte. 280 vs. the N.J. Turnpike, however. No scenery
had to be built for 280. The lakes, forests and undulating hills were
installed by a previous Contractor working on an unrelated project. On the
Turnpike, particularly between Exits 11 & 16, the tourist is inundated with
a fabulous display of surreal artworks (with complementary aromas). These were
constructed at great expense, with minute attentions to function as well as
form. South of Newark (ahh, Newark!) is an added attraction; the pike splits
into two, allowing the novices to take the left (autos only), while the
skilled and adventurous driver may hone his/her craft on the truck-ridden
right. This is what you pay for, and I can hardly see how one could object.
Why, the Vince Lombardi and Grover Cleveland Rest Stops are alone worth the
price of admission.
Note: The Turnpike is best viewed in a winter sunset.
Andy
------
-------
∂18-Apr-85 1237 YAMARONE@SU-CSLI.ARPA SANDWICH UPDATE
Received: from [36.9.0.46] by SU-AI.ARPA with TCP; 18 Apr 85 12:37:05 PST
Date: Thu 18 Apr 85 12:34:40-PST
From: Tom Yamarone <YAMARONE@SU-CSLI.ARPA>
Subject: SANDWICH UPDATE
To: folks@SU-CSLI.ARPA
AFTER THE THIRD DAY OF OPERATIONS, EVERYTHING IS GOING WELL AND EVERYONE
SEEMS WELL FED. THERE ARE TWO MORE POINTS OF THE WHOLE MATTER TO CLARIFY.
I DON'T WANT TO BE"RUNNING THE GAMUT"ON ITEMS THAT I'LL PICK UP, BUT
ALONG WITH SANDWICHES, MILK(SPECIFY LOW,NON, WHOLE), AND FRUIT(MOSTLY APPLES
, SOME BANANAS SO FAR) , I WILL PICK UP
YOGURT FOR $0.75.
IT'S DANNON BRAND AND COMES IN MOST COMMON YOGURT FLAVORS..
THE SECOND ITEM IS THE ORDERING DEADLINE AND THE PROBLEM LIES IN THE FACT THAT
I'M OFFICIALLY NOT HERE FROM 10-11 , SO I'D LIKE TO ASK IF THE ORDERS COULD
BE PLACED BEFORE
10:00 AM
-------------
I WILL GET MORE EXTRA FRUIT IN THE FUTURE AND EXTRA SANDWICHES WILL ALWAYS
BE AVAILABLE. BY THE WAY, WE HAVE TWO EXTRA SANDWICHES STILL AVAILABLE
TODAY: TUNA AND CHICKEN SALAD...HUNGRY YET?
THANKS FOR YOUR COOPERATION IN THE CAUSE TO ABATE STARVATION!!!
SANDWICH MAN.
-------
∂18-Apr-85 1452 chertok%ucbcogsci@Berkeley UCB Cognitive Science Seminar--April 23
Received: from UCB-VAX.ARPA by SU-AI.ARPA with TCP; 18 Apr 85 14:52:04 PST
Received: from ucbcogsci.ARPA by UCB-VAX.ARPA (4.24/4.45)
id AA00749; Thu, 18 Apr 85 14:49:17 pst
Received: by ucbcogsci.ARPA (4.24/4.38)
id AA14413; Thu, 18 Apr 85 14:49:18 pst
Date: Thu, 18 Apr 85 14:49:18 pst
From: chertok%ucbcogsci@Berkeley (Paula Chertok)
Message-Id: <8504182249.AA14413@ucbcogsci.ARPA>
To: cogsci-friends%ucbcogsci@Berkeley
Subject: UCB Cognitive Science Seminar--April 23
BERKELEY COGNITIVE SCIENCE PROGRAM
Cognitive Science Seminar -- IDS 237B
TIME: Tuesday, April 23, 11 - 12:30
PLACE: 240 Bechtel Engineering Center
(followed by)
DISCUSSION: 12:30 - 1:30 in 200 Building T-4
SPEAKER: David Dowty, Center for Advanced Study in the
Behavioral Sciences
TITLE: ``TYPE RAISING, FUNCTIONAL COMPOSITION, AND NON-
CONSTITUENT CONJUNCTION''
This talk presents an analysis of English non-constituent
conjunction (e.g. "John gave Mary a book and Susan a record") in
the version of categorial grammar developed by Mark Steedman and
Anthony Ades. This theory, which will be explained in the course
of the talk, employs the operations of functional composition and
type-raising (or function-argument "flip-flop") in its syntax and
semantics, and thereby treats WH-gaps without appeal to "traces"
in its syntax or to variables (and variable Binding) in its
semantics. This theory has two properties of significant poten-
tial interest to cognitive science: it permits sentences to be
parsed, and fully interpreted semantically, incrementally from
left to right, almost word by word (as humans apparently do);
secondly, it has been argued that applicative programming
languages can be more efficiently evaluated by first compiling
them into equivalent expressions using combinators (like func-
tional composition) but excluding bound variables.
A necessary feature of Steedman's theory is that it posits
many "phantom constituents" of English that are unlike any con-
stituents found in traditional linguistic analyses. Though these
phantom constituents normally have no visible effect on the weak
generative capacity of the grammar (other than permitting WH-
constructions without traces), I will argue that such phantom
constituents are in fact exactly what we need to describe so-
called non-constituent conjunction (e.g. the string "Mary a book"
in the example above is a predicted constituent in this theory).
The talk shows how this gives a more adequate account of English
conjunction data than recent proposals in Generalized Phrase
Structure Grammar by Gazdar and associates and by Schachter and
Mordechai.
-------------------------------------------------------------
UPCOMING TALK (final for this semester)
April 30: Herbert H. Clark, Psychology Department, Stan-
ford University
-------------------------------------------------------------
∂18-Apr-85 1458 MARJORIE@SU-CSLI.ARPA Consultant schedule
Received: from [36.9.0.46] by SU-AI.ARPA with TCP; 18 Apr 85 14:58:11 PST
Date: Thu 18 Apr 85 14:48:15-PST
From: Marjorie Maxwell <MARJORIE@SU-CSLI.ARPA>
Subject: Consultant schedule
To: Folks@SU-CSLI.ARPA
With the new Spring quarter underway we have a new consultant schedule
(and also some new , and some not-so-new consultants). To get a copy of the
updated consultant schedule which will tell you who to expect when you come
in or call for assistance (or to look for someone specifically) - do
canon con:schedule.
Marj
-------
∂18-Apr-85 1506 FRANK@SU-CSLI.ARPA Re: Consultant schedule
Received: from [36.9.0.46] by SU-AI.ARPA with TCP; 18 Apr 85 15:06:26 PST
Date: Thu 18 Apr 85 15:00:28-PST
From: Frank Chen <Frank@SU-CSLI.ARPA>
Subject: Re: Consultant schedule
To: MARJORIE@SU-CSLI.ARPA
cc: Folks@SU-CSLI.ARPA
In-Reply-To: Message from "Marjorie Maxwell <MARJORIE@SU-CSLI.ARPA>" of Thu 18 Apr 85 14:51:26-PST
Also-known-as: Franky@SCORE, Frank@SIERRA, F.Frank@LOTS-C
Telephone: (415) 424-8166
HELP CONSULTING-SCHEDULE will also show you the schedule on-line
(to save Canon costs).
Frank
-------
∂18-Apr-85 1515 LIBRARY@SU-SCORE.ARPA SOCRATES: Searching Techniques- Browse is available in the Lookup Mode
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 18 Apr 85 15:15:01 PST
Date: Thu 18 Apr 85 15:10:34-PST
From: C.S./Math Library <LIBRARY@SU-SCORE.ARPA>
Subject: SOCRATES: Searching Techniques- Browse is available in the Lookup Mode
To: su-bboards@SU-SCORE.ARPA
cc: faculty@SU-SCORE.ARPA,
: ;
The Browse command is also available in the lookup mode. If you type
HELP BROWSE you will receive the following information from SOCRATES:
The BROWSE command may be used to scan headings alphabetically. Authors,
titles, organization names, subjects, subject names, and call numbers may
be scanned by the use of the BROWSE command. This command is exceptionally
helpful when looking for variant spellings of author's name.
Type BROWSE at the YOUR RESPONSE: prompt to browse in Lookup mode.
It is also available in Command Mode.
HLlull
-------
∂18-Apr-85 1555 @SU-SCORE.ARPA:WASHINGTON@SUMEX-AIM.ARPA POTLUCK (you blew it!)
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 18 Apr 85 15:52:22 PST
Received: from SUMEX-AIM.ARPA by SU-SCORE.ARPA with TCP; Thu 18 Apr 85 15:35:13-PST
Date: Thu 18 Apr 85 15:27:55-PST
From: Rich Washington <washington@SUMEX-AIM.ARPA>
Subject: POTLUCK (you blew it!)
To: csd@SU-SCORE.ARPA
Okay. You were warned. We have to draw the line somewhere, so....
NO SPACE LEFT AT THE POTLUCK
Our apologies to those who didn't make it. Better luck next time.
For those of you who are going, we'll be sending out incredibly clear
directions and a map as soon as they come back from the design crew.
-------
∂18-Apr-85 1702 chertok%ucbcogsci@Berkeley Fodor talk at Stanford
Received: from UCB-VAX.ARPA by SU-AI.ARPA with TCP; 18 Apr 85 17:02:00 PST
Received: from ucbcogsci.ARPA by UCB-VAX.ARPA (4.24/4.45)
id AA03628; Thu, 18 Apr 85 17:00:45 pst
Received: by ucbcogsci.ARPA (4.24/4.38)
id AA14939; Thu, 18 Apr 85 17:00:49 pst
Date: Thu, 18 Apr 85 17:00:49 pst
From: chertok%ucbcogsci@Berkeley (Paula Chertok)
Message-Id: <8504190100.AA14939@ucbcogsci.ARPA>
To: cogsci-friends%ucbcogsci@Berkeley
Subject: Fodor talk at Stanford
STANFORD LINGUISTICS COLLOQUIUM
Speaker:Janet Dean Fodor
University of Connecticut and CSLI
Date: Tuesday, April 23
Time: 3:15pm
Location: Bldg. 200, Rm.217
(History Corner, Quadrangle)
Reception following.
DETERMINISTIC PARSING AND SUBJACENCY
Berwick and Weinberg (1984, MIT Press) claim that their
deterministic parser predicts the existence and range of application
of Subjacency, and that this provides simultaneous support for
Subjacency as the correct description of the linguistic facts, and for
deterministic models of the human sentence parsing mechanism.
I argue that the determinism of their parser is irrelevant to
their predictions, and that in any case all four of their predictions
are false.
∂18-Apr-85 1712 KEMMER@SU-CSLI.ARPA fodor talk
Received: from [36.9.0.46] by SU-AI.ARPA with TCP; 18 Apr 85 17:11:49 PST
Date: Thu 18 Apr 85 17:07:18-PST
From: Suzanne Kemmer <KEMMER@SU-CSLI.ARPA>
Subject: fodor talk
To: folks@SU-CSLI.ARPA
STANFORD LINGUISTICS COLLOQUIUM
Speaker:Janet Dean Fodor
University of Connecticut and CSLI
Date: Tuesday, April 23
Time: 3:15pm
Location: Bldg. 200, Rm.217
(History Corner, Quadrangle)
Reception following.
DETERMINISTIC PARSING AND SUBJACENCY
Berwick and Weinberg (1984, MIT Press) claim that their
deterministic parser predicts the existence and range of application
of Subjacency, and that this provides simultaneous support for
Subjacency as the correct description of the linguistic facts, and for
deterministic models of the human sentence parsing mechanism.
I argue that the determinism of their parser is irrelevant to
their predictions, and that in any case all four of their predictions
are false.
-------
-------
∂18-Apr-85 2231 @SU-SCORE.ARPA:WIEDERHOLD@SUMEX-AIM.ARPA Re: Coffee / N.J.Turnpike
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 18 Apr 85 22:31:10 PST
Received: from SUMEX-AIM.ARPA by SU-SCORE.ARPA with TCP; Thu 18 Apr 85 22:14:48-PST
Date: Thu 18 Apr 85 22:15:19-PST
From: Gio Wiederhold <WIEDERHOLD@SUMEX-AIM.ARPA>
Subject: Re: Coffee / N.J.Turnpike
To: GELMAN@SUMEX-AIM.ARPA
cc: csd@SU-SCORE.ARPA
In-Reply-To: Message from "Andrew Gelman <GELMAN@SUMEX-AIM.ARPA>" of Thu 18 Apr 85 12:39:16-PST
If we get decent coffee we will be able to navigate by smell through MJH,
as I used to be able to do driving down the NewJersey turnpike. Gio
-------
∂18-Apr-85 2305 @SU-SCORE.ARPA:ARK@SU-AI.ARPA Coffee
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 18 Apr 85 23:05:05 PST
Received: from SU-AI.ARPA by SU-SCORE.ARPA with TCP; Thu 18 Apr 85 22:49:26-PST
Date: 18 Apr 85 2249 PST
From: Arthur Keller <ARK@SU-AI.ARPA>
Subject: Coffee
To: CSD@SU-SCORE.ARPA
As one who never drinks coffee, I have no objection to the department
providing decent coffee, even for free. Regardless of whether the coffee
is free or costs money, I think it is ludicrous not to have a supply of
styrofoam cups available for visitors to use.
I don't think that this discussion should continue by mass mailing to
everyone in the department through the CSD mailing list. Perhaps we could
all send our responses to Joan Feigenbaum who could distribute them
collectively to those who are interested.
Thanks.
Arthur
∂19-Apr-85 0013 @SU-SCORE.ARPA:reid@Glacier Re: POTLUCK (you blew it!)
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 19 Apr 85 00:13:26 PST
Received: from Glacier by SU-SCORE.ARPA with TCP; Thu 18 Apr 85 23:59:25-PST
Date: 18 Apr 1985 2358-PST (Thursday)
From: Brian Reid <reid@Glacier>
To: Rich Washington <washington@SUMEX-AIM.ARPA>
Cc: csd@SU-SCORE.ARPA
Subject: Re: POTLUCK (you blew it!)
In-Reply-To: Rich Washington <washington@SUMEX-AIM.ARPA> /
Thu 18 Apr 85 15:27:55-PST.
Clearly the potluck needs to be held at a bigger place next time.
Anybody who is not going to the CSD potluck is welcome to come to the
CIS picnic, which is the same day, outside the CIS building.
For that matter, anybody who IS going is also welcome to come.
∂19-Apr-85 0959 JF@SU-SCORE.ARPA Moderate Length Report on Coffee Crusade
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 19 Apr 85 09:59:47 PST
Date: Fri 19 Apr 85 09:43:26-PST
From: Joan Feigenbaum <JF@SU-SCORE.ARPA>
Subject: Moderate Length Report on Coffee Crusade
To: csd@SU-SCORE.ARPA
Do not read this if you are not interested in coffee, and DO NOT SEND ME A
COMPLAINT ABOUT HAVING TO READ THIS BECAUSE YOU DON'T HAVE TO!!!!!!!!!!!!!
There is overwhelming support for the idea that CSD, like PARC, CIS, certain
labs at SRI, etc., should provide good quality coffee (as well as styrofoam
cups, tea, and instant cocoa) free of charge for members and guests. There
is also considerable enthusiasm for going ahead and having a RealCoffee Pool
(that is, member-sponsored) if it turns out to be impossible to pay for it with
department funds.
We're investigating how much things cost, in time and hassle as well as money,
right now and will keep everyone informed. There have been several volunteers
to do the purchasing once the thing gets going, and I need people who are
willing to do the feasibility studies right now (as well as more volunteers
to do the purchasing once the thing gets going). If you are interested in
volunteering, please inform jf@score.
Thanks,
Joan
-------
∂19-Apr-85 1016 HEDGES@SU-SCORE.ARPA Tuesday Lunch
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 19 Apr 85 10:16:28 PST
Date: Fri 19 Apr 85 10:06:13-PST
From: Karen Hedges <HEDGES@SU-SCORE.ARPA>
Subject: Tuesday Lunch
To: Faculty@SU-SCORE.ARPA, Bureaucrats@SU-SCORE.ARPA
REMINDER:
The next Faculty Lunch is scheduled for Tuesday, April 23 @ 12:15 pm in
Conference Room 146.
The topic will be Discussion about Computer Forum with Guest, Dr.William
F. Miller of SRI.
-------
∂19-Apr-85 1059 ABRAIDO@SU-SCORE.ARPA Coffee
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 19 Apr 85 10:55:58 PST
Date: Fri 19 Apr 85 10:40:07-PST
From: Leonor M. Abraido <ABRAIDO@SU-SCORE.ARPA>
Subject: Coffee
To: csd@SU-SCORE.ARPA
PLEASE, I beg of you, STOP SENDING THESE MESSAGES TO PEOPLE'S ELECTRONIC
MAIL BOXES!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!!! Bboard is more suited to this
kind of flaming.
Leonor
-------
∂19-Apr-85 1248 @SU-SCORE.ARPA:vardi@diablo Roomate for STOC
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 19 Apr 85 12:47:56 PST
Received: from diablo by SU-SCORE.ARPA with TCP; Fri 19 Apr 85 10:50:00-PST
Date: Fri, 19 Apr 85 10:37:32 pst
From: Moshe Vardi <vardi@diablo>
Subject: Roomate for STOC
To: aflb.all@score, pw@diablo
My supposed roomate for STOC changed his mind and is not going. I already
have a reservation for a double room. Anybody interested in sharing a room?
Moshe
∂19-Apr-85 1300 EMMA@SU-CSLI.ARPA fodor talk
Received: from [36.9.0.46] by SU-AI.ARPA with TCP; 19 Apr 85 13:00:22 PST
Mail-From: KEMMER created at 18-Apr-85 17:06:27
Date: Thu 18 Apr 85 17:06:27-PST
From: Suzanne Kemmer <KEMMER@SU-CSLI.ARPA>
Subject: fodor talk
To: friends@SU-CSLI.ARPA
ReSent-Date: Fri 19 Apr 85 12:56:29-PST
ReSent-From: Emma Pease <Emma@SU-CSLI.ARPA>
ReSent-To: friends@SU-CSLI.ARPA
STANFORD LINGUISTICS COLLOQUIUM
Speaker:Janet Dean Fodor
University of Connecticut and CSLI
Date: Tuesday, April 23
Time: 3:15pm
Location: Bldg. 200, Rm.217
(History Corner, Quadrangle)
Reception following.
DETERMINISTIC PARSING AND SUBJACENCY
Berwick and Weinberg (1984, MIT Press) claim that their
deterministic parser predicts the existence and range of application
of Subjacency, and that this provides simultaneous support for
Subjacency as the correct description of the linguistic facts, and for
deterministic models of the human sentence parsing mechanism.
I argue that the determinism of their parser is irrelevant to
their predictions, and that in any case all four of their predictions
are false.
-------
-------
∂19-Apr-85 1418 HEDGES@SU-SCORE.ARPA Zucker Visit
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 19 Apr 85 14:18:25 PST
Date: Fri 19 Apr 85 14:07:33-PST
From: Karen Hedges <HEDGES@SU-SCORE.ARPA>
Subject: Zucker Visit
To: Faculty@SU-SCORE.ARPA
I have only heard from a few of you! I still have openings between
10:00 to 11:30 to visit with Dr. Zucker on Tuesday, April 23. Sorry,
but his afternoon is filled. He may possibly still be around the morning
of Wednesday, April 24, if that would work better for anyone. Please
let me know.
I realize you all have busy schedules and appreciate your help with this.
Karen
-------
∂19-Apr-85 1616 @SU-SCORE.ARPA:fagin.sjrlvm1%ibm-sj.csnet@csnet-relay.arpa Super-saver to STOC
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 19 Apr 85 16:16:36 PST
Received: from csnet-relay by SU-SCORE.ARPA with TCP; Fri 19 Apr 85 16:13:48-PST
Received: from ibm-sj by csnet-relay.csnet id aa19194; 19 Apr 85 19:07 EST
Date: Fri, 19 Apr 85 15:41:45 PST
From: Ron Fagin <fagin%ibm-sj.csnet@csnet-relay.arpa>
To: aflb.all@su-score.ARPA
Subject: Super-saver to STOC
I bought my supersaver plane ticket for STOC, but just
discovered I won't be able to go
(because I will be moving that week).
Does anyone want to buy it at my (IBM's) cost?
The ticket cost $258 (the full coach fare is over $900).
The flight departs Sat. afternoon, May 4, from San Jose to Providence,
and returns Wednesday, May 8, leaving Providence at 4:55 p.m. for
San Jose.
∂22-Apr-85 1433 JUTTA@SU-SCORE.ARPA Alliant Computer Systems presentation
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 22 Apr 85 14:32:43 PST
Date: Mon 22 Apr 85 14:24:04-PST
From: Jutta McCormick <JUTTA@SU-SCORE.ARPA>
Subject: Alliant Computer Systems presentation
To: super@SU-SCORE.ARPA
Stanford-Phone: (415) 497-0572
Alliant Computer Systems Corporation (formerly known as Dataflow) will
discuss the dataflow machine they are building on Tuesday, 7 May 1985,
at 1:15 p.m., in MJH 252. Interested parties are invited.
--Joe Oliger
-----
-------
∂22-Apr-85 1853 @SU-CSLI.ARPA:PCOHEN@SRI-AI.ARPA Next meeting
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 22 Apr 85 18:49:50 PST
Received: from SRI-AI.ARPA by SU-CSLI.ARPA with TCP; Mon 22 Apr 85 18:47:20-PST
Date: Fri 19 Apr 85 11:35:43-PST
From: Phil Cohen <PCOHEN@SRI-AI.ARPA>
Subject: Next meeting
To: NL4: ;
Our next meeting will be Tues, April 23, at 12:45, in the Ventura
Seminar room. Lauri Carlson will speak about Semantics for Dialogue
Games.
See you there,
Phil
-------
∂22-Apr-85 2325 BRODER@SU-SCORE.ARPA Next AFLB talk
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 22 Apr 85 23:25:00 PST
Date: Mon 22 Apr 85 23:20:04-PST
From: Andrei Broder <Broder@SU-SCORE.ARPA>
Subject: Next AFLB talk
To: aflb.all@SU-SCORE.ARPA
cc: tuley@SU-SCORE.ARPA, su-bboards@SU-SCORE.ARPA
4-25-85 - Dr. J. Reeds (Bell Labs):
"Unique extrapolation of polynomial sequences"
Let a sequence of k-dimensional vectors x sub 0, x1 sub 1, ... (over a
ring A) be determined by a polynomial recurrence of form x sub n = T(x
sub n-1), where T itself is unknown but is known to be a polynomial
map of degree at most d. We show that there is a finite N such that
the entire sequence {x sub n : n >= 0} can be deduced from the first
N+1 terms x sub 0 , x sub 1 , ... , x sub N alone. The number N =
phi(d, k, A) depends on d and k and the ring A but not on T.
Let phi*(d,k) denote the maximum of phi(d,k,A) over all commutative
rings with unit. Then phi*(d,k)<infinity and in particular
phi*(d,1)=d+1 and phi*(1,k)=k+1. In the general case
phi*(d,k)>=binom(k+d, k) and equality does not always hold because
phi*(2,2)>=7. Finally, we show for each k that max{phi(d,k,F): F a
field} is bounded by a polynomial in d.
This is joint work with J. Lagarias.
***** Time and place: April 25, 12:30 pm in MJ352 (Bldg. 460) ******
5-2-85 - Prof. Nishizeki (Tohoku U., Japan):
"Multicommodity flows in planar undirected graphs and shortest paths"
Abstract will be published later.
***** Time and place: May 2, 12:30 pm in MJ352 (Bldg. 460) ******
++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
Regular AFLB meetings are on Thursdays, at 12:30pm, in MJ352 (Bldg.
460).
If you have a topic you would like to talk about in the AFLB seminar
please let me know. (Electronic mail: broder@su-score.arpa, phones:
(415) 853-2118 and (415) 948-9172). Contributions are wanted and
welcome. Not all time slots for this academic year have been filled
so far.
For more information about future and past AFLB meetings and topics
you might want to look at the file [SCORE]<broder>aflb.bboard .
- Andrei Broder
-------
∂23-Apr-85 1043 STUCKY@SU-CSLI.ARPA A Reminder
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 23 Apr 85 10:43:21 PST
Date: Tue 23 Apr 85 10:37:24-PST
From: Susan Stucky <Stucky@SU-CSLI.ARPA>
Subject: A Reminder
To: RRR@SU-CSLI.ARPA
Just a reminder about the RRR meeting: Fernando will be speaking as
planned this afternoon (that's Tuesday, the 23rd) at 2:15, in the
trailer seminar room.
-Susan
-------
∂23-Apr-85 1123 @SU-CSLI.ARPA:BrianSmith.pa@Xerox.ARPA RRR: Fernando Pereira Today
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 23 Apr 85 11:23:03 PST
Received: from Xerox.ARPA by SU-CSLI.ARPA with TCP; Tue 23 Apr 85 11:16:00-PST
Received: from Semillon.ms by ArpaGateway.ms ; 23 APR 85 10:22:11 PST
Date: 23 Apr 85 10:20 PST
From: BrianSmith.pa@Xerox.ARPA
Subject: RRR: Fernando Pereira Today
To: RRR@SU-CSLI.ARPA
cc: BrianSmith.pa@Xerox.ARPA
Remember, Fernando Pereira will talk at today's meeting (Tuesday 4/23 at
2:15 in the Ventura Trailer seminar room).
Approximate subject: Representational Structures in Situated Automata.
See you there.
Brian
∂23-Apr-85 1149 BARWISE@SU-CSLI.ARPA CSLI Summer Meeting
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 23 Apr 85 11:49:30 PST
Date: Tue 23 Apr 85 11:39:38-PST
From: Jon Barwise <BARWISE@SU-CSLI.ARPA>
Subject: CSLI Summer Meeting
To: Folks@SU-CSLI.ARPA
cc: BBoard@SU-CSLI.ARPA, Newsletter@SU-CSLI.ARPA
This is to remind you of the summer meetings in mid July on "Logic,
Language, and Computation." Susi Parker has a list of courses and
talks. Researchers supported by CSLI do not have to pay the
registration fee, but they must register with Ingrid by June 1, so
that we can order the right number of programs, abstracts, coffee,
cookies, etc. For more information, contact Ingrid@su-csli, 497-3084.
-------
∂23-Apr-85 1326 WRIGHT@SUMEX-AIM.ARPA SIGLUNCH Friday, April 26
Received: from SUMEX-AIM.ARPA by SU-AI.ARPA with TCP; 23 Apr 85 13:26:39 PST
Date: Tue 23 Apr 85 13:19:28-PST
From: Carol Wright/Susie Barnes <WRIGHT@SUMEX-AIM.ARPA>
Subject: SIGLUNCH Friday, April 26
To: siglunch@SUMEX-AIM.ARPA
SIGLUNCH
DATE: Friday, April 26, 1985
LOCATION: Chemistry Gazebo, between Physical &
Organic Chemistry
TIME: 12:05
SPEAKER: Johan Dekleer
Member of Research Staff in Qualitative
Physics at Xerox Park
TITLE: An Assumption-Based Truth Maintenance System
This talk presents a new view of problem solving motivated by a
new kind of truth maintenance system. Unlike previous truth maintenance
systems which were based on manipulating justifications, this truth
maintenance system is, in addition, based on manipulating assumption
sets. As a consequence it is possible to work effectively and
efficiently with inconsistent information, context switching is free,
and most backtracking (and all retraction) is avoided. These
capabilities motivate a different kind of problem-solving architecture
in which multiple potential solutions are explored simultaneously. This
architecture is particularly well-suited for tasks where a reasonable
fraction of the potential solutions must be explored.
-------
-------
∂23-Apr-85 2200 @SU-SCORE.ARPA:udi@wisc-rsch.arpa Toronto Day
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 23 Apr 85 22:00:00 PST
Received: from wisc-rsch.arpa by SU-SCORE.ARPA with TCP; Tue 23 Apr 85 21:58:55-PST
Received: by wisc-rsch.arpa; Tue, 23 Apr 85 21:39:56 cst
Received: from yale.arpa by wisc-rsch.arpa; Tue, 23 Apr 85 17:49:13 cst
Received: by YALE-BULLDOG.YALE.ARPA; 23 Apr 85 18:36:34 EST (Tue)
Message-Id: <8504232336.AA04631@YALE-BULLDOG.YALE.ARPA>
Received: from YALE-RING by YALE-RES via CHAOS; Tue, 23 Apr 85 18:37:47 EST
Subject: Toronto Day
Date: Tue, 23 Apr 85 18:37:50 EST
From: Neil Immerman <Immerman@YALE.ARPA>
To: udi@WISC-RSCH
Cc: Immerman@YALE.ARPA
Sender: udi@wisc-rsch.arpa
Resent-From: udi@wisc-rsch.arpa (Udi Manber)
Resent-To: theory:;@wisc-rsch.arpa
announcing . . . . . TORONTO DAY AT WESLEYAN
Saturday May 4
9:30 am Steve Cook
A Taxonomy of Problems with Fast Parallel Algorihms
10:30 am Coffee Break
11:00am Mike Luby
A Simple Parallel Algorithm for the Maximal Independent Set Problem
12:00 Noon Lunch in Woodhead Lounge (184 Science Center)
2:00 pm Allan Borodin
Parallel Algebraic Complexity
3:00 pm Coffee Break
3:30 pm Charles Rackoff
Some Definitions and Issues in the Theory of Cryptography
4:30 pm Problem Session, Math Lounge (Sixth Floor)
Beer, Soda and Munchies
6:30 pm Buffet Dinner
[All talks will be held in Room 58 Science Center.]
On Friday, May 3, Steve Cook will give a talk at 4 P.M. in
150 Science Center intended for a general audience: Can Computers
Routinely Discover Mathematical Proofs? This talk will be followed
by a reception in the Woodhead Lounge at 5 P.M. and then dinner at a
local restaurant at 6:30 P.M. All Toronto Day participants are welcome
to attend. On Sunday there will be informal activities, including a
morning hike, before heading for Providence (STOC), 75 miles east.
Assistance with local transportation and travel to STOC provided
on request. Lodging available at a local motel; sleeping bag space
provided by hosts.
Toronto Day is hosted by Alan Cobham, Dan Dougherty, Sorin Istrail,
Susan Landau, and Carol Wood, and is funded in part by The Sloan
Foundation, Proctor and Gamble Co., and the Wesleyan University Department
of Mathematics.
For further information contact the Wesleyan Mathematics
Department at (203)347-9411 Ext.2398 or Carol Wood at Ext.2648
(Bitnet address WOODatWESLYN).
-------
-------
∂23-Apr-85 2218 ullman@diablo meeting
Received: from SU-AIMVAX.ARPA by SU-AI.ARPA with TCP; 23 Apr 85 22:18:33 PST
Date: Tue, 23 Apr 85 22:16:55 pst
From: Jeff Ullman <ullman@diablo>
Subject: meeting
To: nail@diablo
Meeting Wednesday, 11AM, 301MJH.
Bring your ideas.
∂24-Apr-85 0858 CLAY@SU-CSLI.ARPA South Asian fonts
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 24 Apr 85 08:57:19 PST
Date: Wed 24 Apr 85 08:52:10-PST
From: Clay Andres <CLAY@SU-CSLI.ARPA>
Subject: South Asian fonts
To: folks@SU-CSLI.ARPA
Mark Cutter, a Macintosh Software Developer at Apple, has developed a
fairly extensive series of South Asian fonts. He is interested in making
them available to CSLI and talking to anyone here interested in South
Asian languages. Please send me mail and let me know what specific things
CSLI might like to see or hear about from Mark, and I'll try and set up
something appropriate. At least we'll be getting a copy of the fonts for
our Macintoshes.
Clay
-------
∂24-Apr-85 1308 YAMARONE@SU-CSLI.ARPA HUNGRY YET? SANDWICH?
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 24 Apr 85 13:07:17 PST
Date: Wed 24 Apr 85 13:00:30-PST
From: Tom Yamarone <YAMARONE@SU-CSLI.ARPA>
Subject: HUNGRY YET? SANDWICH?
To: FOLKS@SU-CSLI.ARPA
THERE ARE FIVE DELICIOUS SANDWICHES IN THE REFRIGERATOR SHOULD YOU FEEL
THE PANGS OF HUNGER TUGGING ON YOUR TUMMY!
ROAST BEEF, HAM AND SWISS, SALAMI ON RYE, CHICKEN SALAD, AND TURKEY.
MMMMMMMM, MMMMMMMMMM, GOOD!!!!!!!!!!!!!!!!!!!!
SEE AND PAY SUSI IF YOU GET RAVENOUS AND TAKE ADVANTAGE OF THIS OPPORTUNITY!
BYE, SANDWICH MAN.
-------
∂24-Apr-85 1736 chertok%ucbcogsci@Berkeley UCB Cognitive Science Seminar--April 30
Received: from UCB-VAX.ARPA by SU-AI.ARPA with TCP; 24 Apr 85 17:36:36 PST
Received: from ucbcogsci.ARPA by UCB-VAX.ARPA (4.24/4.46)
id AA09974; Wed, 24 Apr 85 17:33:29 pst
Received: by ucbcogsci.ARPA (4.24/4.38)
id AA07166; Wed, 24 Apr 85 17:34:14 pst
Date: Wed, 24 Apr 85 17:34:14 pst
From: chertok%ucbcogsci@Berkeley (Paula Chertok)
Message-Id: <8504250134.AA07166@ucbcogsci.ARPA>
To: cogsci-friends%ucbcogsci@Berkeley
Subject: UCB Cognitive Science Seminar--April 30
BERKELEY COGNITIVE SCIENCE PROGRAM
Cognitive Science Seminar -- IDS 237B
TIME: Tuesday, April 30, 11 - 12:30
PLACE: 240 Bechtel Engineering Center
(followed by)
DISCUSSION: 12:30 - 1:30 in 200 Building T-4
SPEAKER: Herbert H. Clark, Department of Psychology, Stan-
ford University
TITLE: ``Illocutionary acts, illocutionary perfor-
mances''
From John Austin on, theorists have said a good deal about
what it is to be a question, assertion, promise, or other illocu-
tionary act. But in their characterizations they have generally
assumed a rather strong idealization about how illocutionary acts
are performed. Among other things, they have taken these four
points for granted: (1) An illocutionary act is a preplanned
event. (2) It is performed by the speaker acting alone. (3) The
speaker acts with certain definite intentions about affecting his
addressee. And (4) the speaker discharges these intentions
merely by issuing a sentence (or sentence surrogate) in the right
circumstances. As with any idealization, these assumptions
aren't quite right. Indeed, I will document that illocutionary
acts in conversation are not preplanned events but processes that
the participants may alter midcourse for various purposes, and
that they are accomplished by the speaker and addressees acting
together. Once the traditional assumptions are replaced by more
realistic ones, we are led to quite a different notion of illocu-
tionary act.
The view I will develop is that performing illocutionary
acts in conversation is a collaborative process between speaker
and addressees. One of the goals of these participants is to
establish the mutual belief, roughly by the beginning of each new
contribution, that the addressees have understood the speaker's
meaning well enough for current purposes. The speaker and
addressees have systematic linguistic techniques for reaching
this goal. In support of this view I will report a study by
Deanna Wilkes-Gibbs and myself on how definite references get
made in conversation and another study by Edward F. Schaefer and
myself on what it is, more generally, to make certain contribu-
tions to conversation.
∂24-Apr-85 1742 EMMA@SU-CSLI.ARPA Newsletter Apr. 25, No. 26
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 24 Apr 85 17:42:42 PST
Date: Wed 24 Apr 85 17:10:00-PST
From: Emma Pease <Emma@SU-CSLI.ARPA>
Subject: Newsletter Apr. 25, No. 26
To: friends@SU-CSLI.ARPA
Tel: 497-3479
C S L I N E W S L E T T E R
←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←
April 25, 1985 Stanford Vol. 2, No. 26
←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←
A weekly publication of The Center for the Study of Language and
Information, Ventura Hall, Stanford University, Stanford, CA 94305
←←←←←←←←←←←←
CSLI ACTIVITIES FOR *THIS* THURSDAY, April 25, 1985
12 noon TINLunch
Ventura Hall ``Cell Psychology: An Evolutionary Approach to
Conference Room the Symbol-Matter Problem'' by H. H. Pattee
Ivan Blair, CSLI
2:15 p.m. CSLI Seminar
Redwood Hall ``Whither CSLI?''
Room G-19 John Perry, Director, CSLI
3:30 p.m. Tea
Ventura Hall
4:15 p.m. CSLI Colloquium
Redwood Hall ``The Representational Basis for Everyday Aesthetic
Room G-19 Experience -- A Motivational Constraint on Learnable
Systems of Knowledge''
Tom Bever, Columbia University and CASBS
←←←←←←←←←←←
CSLI ACTIVITIES FOR *NEXT* THURSDAY, May 2, 1985
12 noon TINLunch
Ventura Hall ``Categorizing the Senses of `Take' ''
Conference Room by Peter Norvig
Discussion led by Douglas Edwards
(Abstract on page 2)
2:15 p.m. CSLI Seminar
Redwood Hall ``Property Theory and Second-Order Logic''
Room G-19 Chris Menzel, CSLI
(Abstract on page 2)
3:30 p.m. Tea
Ventura Hall
4:15 p.m. CSLI Colloquium
Redwood Hall ``A Formal Theory of Innate Linguistic Knowledge''
Room G-19 Janet Fodor, University of Connecticut
Originally scheduled for April 11
(Abstract on page 3)
!
Page 2 CSLI Newsletter April 25, 1985
←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←
ABSTRACT OF NEXT WEEK'S TINLUNCH
``Categorizing the Senses of `Take' ''
Polysemy is a perennial problem for semantic analysis of natural
language. Most common words are highly polysemous, and in a way which
is not plausibly construed as simple ambiguity among unrelated senses.
Polysemy also, though less obviously, offers a serious challenge to
efforts at understanding commonsense reasoning: apparently clear
verbal statements of goals, beliefs, and modes of reasoning may
conceal unanalyzed complexity due to systematic ambiguity of the terms
in which they are stated. (There may be an analogy between
commonsense concepts and the ``generic operations'' in object-oriented
programming environments, which apply to heterogeneous data types and
are differently interpreted according to the data type operated upon.)
In the paper under consideration, Norvig attempts to analyze about
130 uses of the verb ``take'' into 15 main senses clustered in a graph
about a single primary sense. He considers syntactic differences
between senses, pragmatic appropriateness of usage (in answering
questions), as well as a wide variety of intuitions and prior theories
about case structure, motivation and intended use of concepts, and
metaphorical and metonymic extension. Understanding the derivation of
the senses of words, in the way that Norvig tries to come to grips
with the derivation of the subsenses of ``take'' from its primary
sense, will be a necessary step toward characterizing commonsense
knowledge of the world. --Douglas Edwards
←←←←←←←←←←←←
ABSTRACT OF NEXT WEEK'S SEMINAR
``Property Theory and Second-Order Logic''
Much recent work in semantics (e.g., Barwise and Perry, Chierchia,
Sells, Zalta) involves an extensive appeal to abstract logical
objects--properties, relations, and propositions. Such objects were
of course not unheard of in semantics prior to this work. What is
noteworthy is the extent to which the conception of these objects
differs from the prevailing conception in formal semantics, viz.,
Montague's. Two ways in which they differ (not necessarily common to
all recent accounts) stand out: first, these abstract objects are
metaphysically primitive, not set theoretic constructions out of
possible worlds and individuals; second, they are untyped--properties
can exemplify each other as well as themselves, relations can fall
within their own field, and so on.
With properties, relations, and propositions playing this more
prominent role in semantics (as well as in philosophy), it is
essential that there be a rigorous mathematical theory of these
objects. The framework for such a theory, I think, is second-order
logic; indeed, I will argue that second-order logic, rightly
understood, just IS the theory of properties, relations, and
propositions. To this end, building primarily on the work of Bealer,
Cocchiarella, and Zalta, I will present a second-order logic that is
provably consistent along with an algebraic intensional semantics
which yields significant insights into the structure of the abstract
ontology of logic and the paradoxes. Time permitting, I will apply the
logic to two issues, one in semantics and the other in the philosophy
of mathematics--specifically, to the analysis of noun phrases
involving terms of order like `fourth' and `last', and the question of
what the (ordinal) numbers are, to which I will give a logicist answer
adumbrated by Russell. --Chris Menzel
!
Page 3 CSLI Newsletter April 25, 1985
←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←
ABSTRACT OF NEXT WEEK'S COLLOQUIUM
``A Formal Theory of Innate Linguistic Knowledge''
I assume that an infant is innately provided with some sort of
representational medium in which to record what he observes about his
target language. It has occasionally been suggested that the formal
properties of this mental metalanguage could be the source of
universal properties of natural languages. This differs from the
standard ( = substantive) approach, which assumes in addition that
certain statements of this metalanguage are innately tagged as true.
I propose to take the formal approach seriously. The way to do so
seems to be to try for a theory which accounts for ALL universals in
this way, i.e., solely on the basis of what can and cannot be
expressed in the metalanguage. The attempt is very informative, even
if ultimately it fails.
Success is certainly not guaranteed, for the formal theory
overthrows many familiar assumptions. For instance, it can be shown to
be incompatible (on standard assumptions about children and their
linguistic input) with the existence of any constraints on rule
application or on derivational representations. All the work of
distinguishing well-formed from ill-formed sentences must be done by
rules only. Constraints can determine the shape of the rules, but
cannot tidy up after them if they overgenerate.
It is easiest to see how to set about formulating grammars of this
kind within the framework of GPSG, and it is encouraging that a number
of universals do fall out as consequences of the GPSG formalism. But
there are problems too. Syntactic features, in particular, create
headaches for learnability. --Janet Fodor
[Note to attendees of the Berkeley Cognitive Science Seminars -- this
is the same as the paper presented there on 3/19.]
←←←←←←←←←←←←
LOGIC, LANGUAGE, AND COMPUTATION MEETINGS
July 8-19, 1985, Stanford University
The Association for Symbolic Logic (ASL) and the Center for the
Study of Language and Information (CSLI) will be combining the CSLI
Summer School (July 8-13) with the ASL Meeting (July 15-19). For
further information and registration forms, write to Ingrid Deiwiks,
CSLI, Ventura Hall, Stanford, CA 94305, or call (415)497-3084 or send
computer mail to Ingrid@su-csli.arpa. The deadline for registering is
June 1, 1985.
-------
∂25-Apr-85 0942 EMMA@SU-CSLI.ARPA logic seminar
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 25 Apr 85 09:42:35 PST
Mail-From: SF created at 25-Apr-85 09:34:24
Date: Thu 25 Apr 85 09:34:24-PST
From: Sol Feferman <SF@SU-CSLI.ARPA>
Subject: logic seminar
To: emma@SU-CSLI.ARPA, clt@SU-AI.ARPA
ReSent-Date: Thu 25 Apr 85 09:38:56-PST
ReSent-From: Emma Pease <Emma@SU-CSLI.ARPA>
ReSent-To: friends@SU-CSLI.ARPA
Seminar in Logic and Foundations of Mathematics
Speaker: Prof. Itala d'Ottaviano, Univ. of Campinas, Brazil,
visiting Stanford and UC Berkeley
Title: Extension model theorems, definability and quantifier-
elimination for some many-valued theories.
Time: Tuesday, April 30, 4:15-5:30 P.M.
Place: Room 381-T, Math. Dept., Bldg. 380, Stanford
S. Feferman
-------
∂25-Apr-85 0946 CLT Seminar in Logic and Foundations of Mathematics
To: "@DIS.DIS[1,CLT]"@SU-AI.ARPA
Speaker: Prof. Itala d'Ottaviano, Univ. of Campinas, Brazil,
visiting Stanford and UC Berkeley
Title: Extension model theorems, definability and quantifier-
elimination for some many-valued theories.
Time: Tuesday, April 30, 4:15-5:30 P.M.
Place: Room 381-T, Math. Dept., Bldg. 380, Stanford
S. Feferman
∂25-Apr-85 1048 avg@diablo Implementation follow-up
Received: from SU-AIMVAX.ARPA by SU-AI.ARPA with TCP; 25 Apr 85 10:47:26 PST
Date: Thu, 25 Apr 85 10:44:13 pst
From: Allen VanGelder <avg@diablo>
Subject: Implementation follow-up
To: nail@diablo
1. If you didn't get my earlier message, sorry, I sent it to a bad list.
You should get it soon.
2. If you don't have a diablo account and do want to look at
C Prolog source, I will try to help out. Get in touch.
Allen
∂25-Apr-85 1441 HEDGES@SU-SCORE.ARPA Sr. Faculty Meeting
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 25 Apr 85 14:41:01 PST
Date: Thu 25 Apr 85 13:49:31-PST
From: Karen Hedges <HEDGES@SU-SCORE.ARPA>
Subject: Sr. Faculty Meeting
To: tenured@SU-SCORE.ARPA
PLEASE NOTE: There will be a Sr. Faculty Meeting on Thursday, May 2 @ 2:30
in Conference Room 252.
One of the main agenda items will be:
Consideration of Appointments of Consulting Professors
Copies of candidates C.V.'s will be given to you for review
before the Thursday meeting.
-------
∂25-Apr-85 1703 ullman@diablo CPROLOG
Received: from SU-AIMVAX.ARPA by SU-AI.ARPA with TCP; 25 Apr 85 17:02:41 PST
Date: Thu, 25 Apr 85 16:54:59 pst
From: Jeff Ullman <ullman@diablo>
Subject: CPROLOG
To: nail@diablo
For those of you who missed it: here is Allan's message.
********************************
Received: from SU-AI.ARPA by diablo with TCP; Thu, 25 Apr 85 10:17:19 pst
Received: from SU-AIMVAX.ARPA by SU-AI.ARPA with TCP; 25 Apr 85 10:15:54 PST
Date: Thu, 25 Apr 85 10:16:43 pst
From: Allen VanGelder <avg@diablo>
Subject: An implementation approach
To: ark@sail, asubramanian@sumex, avg@diablo, cheriton@pescadero, cohn@diablo,
freeman@diablo, hasan@score, jef@sail, karlin@diablo, kuper@diablo,
naughton@diablo, restivo@score, roy@diablo, stumm@pescadero,
treitel@sumex, upfal@diablo, vardi@diablo, waltzman@score
We have C Prolog in source form. This could serve as a base for
implementing nail. To evaluate this, look in /vlsi/cprolog on
diablo, and particularly the READ-ME files in various directories.
Their approach has been to write some low-level C programs that
provide some sort of nucleus, and a fair number of prolog programs
to implement many of Prolog's standard features. THEY intended it
to be extensible, so maybe our best bet is to extend it.
My idea is that we write what we conveniently can in prolog, but when
we identify the need for a procedure that is clumsy or inefficient
in prolog, e.g., something in which assigns and/or arrays are a big
win, we then write an additional built-in procedure in C.
We want reasonable portability; this approach provides it thru C.
A critical consideration for this approach is: how good is the
design of C Prolog? If we are going to adapt it successfully, we
will need to find it well documented, well modularized, and having
a consistent design philosophy. I hope some of you will look over
the files and offer opinions on this question.
Another question is: are we as a group willing to write a fair amount
of software in Prolog? If the answer is no, then we might as well
start at the bottom with yacc/lex. But then, I have to wonder if
we can hope to produce a reasonable quality logic programming system
if we don't do a lot of logic programming ourselves? It will probably
be a lot closer to an enhanced DBMS.
One final point. Our C Prolog is from Edinburgh University, not a
commercial vendor. Therefore, re-distributing a modified version
of it should present no serious problems; we can surely make a
reasonable agreement on that. Also, we may well benefit from their
ongoing efforts. For example, our version has an interpreter, but
no compiler. I think their newest version has a compiler, or they
are close to having one. Under our current license, I think we can
get it by just sending a tape. Chuck Restivo knows people at
Edinburgh, and probably has firmer info in this area.
Allen
∂25-Apr-85 1704 ullman@diablo points of agreement
Received: from SU-AIMVAX.ARPA by SU-AI.ARPA with TCP; 25 Apr 85 17:03:36 PST
Date: Thu, 25 Apr 85 16:57:58 pst
From: Jeff Ullman <ullman@diablo>
Subject: points of agreement
To: nail@diablo
I was very pleased by the meeting yesterday.
It seems that we have several major areas of agreement.
1. The system should be written in some language.
2. The system should be designed to run on some machine.
3. There should be some sort of interface usable by guys wearing hats.
I hope that in the future we can refine this understanding.
By the way, don't forget Dave Maier is speaking 11AM Friday
in 252. The topic is data models as a cross between Prolog
and Smalltalk.
∂26-Apr-85 0830 @SU-SCORE.ARPA:udi@wisc-rsch.arpa Are you looking for a room mate for the STOC???
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 26 Apr 85 08:30:19 PST
Received: from wisc-rsch.arpa by SU-SCORE.ARPA with TCP; Fri 26 Apr 85 08:24:56-PST
Received: by wisc-rsch.arpa; Fri, 26 Apr 85 10:10:45 cst
Received: from wisc-crys.arpa by wisc-rsch.arpa; Fri, 26 Apr 85 04:43:02 cst
Message-Id: <8504261042.AA24648@wisc-crys.arpa>
Received: from csnet-relay.arpa by wisc-crys.arpa; Fri, 26 Apr 85 04:42:52 cst
Received: from ucsc by csnet-relay.csnet id ad05417; 26 Apr 85 5:46 EST
Received: by vger.UCSC (4.12/4.7)
id AA10486; Thu, 25 Apr 85 20:09:20 pst
Date: Thu, 25 Apr 85 20:09:20 pst
From: manfred <@csnet-relay.arpa,@ucsc.CSNET:manfred@ucsc.CSNET>
To: theory@wisc-crys.ARPA
Subject: Are you looking for a room mate for the STOC???
Sender: udi@wisc-rsch.arpa
Resent-From: udi@wisc-rsch.arpa (Udi Manber)
Resent-To: theory:;@wisc-rsch.arpa
I am looking for somebody to match up with!
Please call me right away!:
Manfred Warmuth
UC California
Santa Cruz
office: 408 429-4950
message:408 429-2565
home: 408 423-9091
Greetings from the redwooods
Manfred
∂26-Apr-85 0946 EMMA@SU-CSLI.ARPA Psych. Dept. Friday Cognitive Seminar
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 26 Apr 85 09:46:33 PST
Return-Path: <gluck@SU-PSYCH>
Received: from SU-PSYCH ([36.36.0.202].#Internet) by SU-CSLI.ARPA with TCP; Thu 25 Apr 85 19:08:45-PST
Date: Thu, 25 Apr 85 05:01:59 pst
From: gluck@SU-PSYCH (Mark Gluck)
Subject: Psych. Dept. Friday Cognitive Seminar
To: friends@CSLI, su-bboards@score
ReSent-Date: Fri 26 Apr 85 09:36:03-PST
ReSent-From: Emma Pease <Emma@SU-CSLI.ARPA>
ReSent-To: friends@SU-CSLI.ARPA
A Computational Model of
Skill Aquisition
KURT VAN LEHN (Xerox PARC)
Abstract:
A theory will be presented that describes how people learn certain
procedural skills, such as the written algorithms of arithmetic and
algebra, from multi-lesson curricula. There are two main hypotheses.
(1) Teachers enforce, perhaps unknowingly, certain constraints that
relate the structure of the procedure to the structure of the lesson
sequence, and moreover, students employ these constraints, perhaps
unknowingly, as they induce a procedure from the lesson sequence. (2)
As students follow the procedure they have induced, they employ a
certain kind of meta-level problem solving to free themselves when their
interpretation of the procedure gets stuck. The theory's predictions,
which are generated by a computer model of the putative learning and
problem solving processes, have been tested against error data from
several thousand students. The usual irrefutability of computer
simulations of complex cognition has been avoided by a linguistic style
of argumentation that assigns empirical responsibility to individual
hypotheses.
←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←
April 26th 3:15pm Jordan Hall; Rm. 100
∂28-Apr-85 0504 @SU-SCORE.ARPA:udi@wisc-rsch.arpa Are you looking for a room mate for the STOC???
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 28 Apr 85 05:04:52 PDT
Received: from wisc-rsch.arpa by SU-SCORE.ARPA with TCP; Sun 28 Apr 85 04:59:39-PDT
Received: by wisc-rsch.arpa; Fri, 26 Apr 85 10:10:45 cst
Received: from wisc-crys.arpa by wisc-rsch.arpa; Fri, 26 Apr 85 04:43:02 cst
Message-Id: <8504261042.AA24648@wisc-crys.arpa>
Received: from csnet-relay.arpa by wisc-crys.arpa; Fri, 26 Apr 85 04:42:52 cst
Received: from ucsc by csnet-relay.csnet id ad05417; 26 Apr 85 5:46 EST
Received: by vger.UCSC (4.12/4.7)
id AA10486; Thu, 25 Apr 85 20:09:20 pst
Date: Thu, 25 Apr 85 20:09:20 pst
From: manfred <@csnet-relay.arpa,@ucsc.CSNET:manfred@ucsc.CSNET>
To: theory@wisc-crys.ARPA
Subject: Are you looking for a room mate for the STOC???
Sender: udi@wisc-rsch.arpa
Resent-From: udi@wisc-rsch.arpa (Udi Manber)
Resent-To: theory:;@wisc-rsch.arpa
I am looking for somebody to match up with!
Please call me right away!:
Manfred Warmuth
UC California
Santa Cruz
office: 408 429-4950
message:408 429-2565
home: 408 423-9091
Greetings from the redwooods
Manfred
∂28-Apr-85 1807 ullman@diablo NAIL! Prototype
Received: from SU-AIMVAX.ARPA by SU-AI.ARPA with TCP; 28 Apr 85 18:06:56 PDT
Date: Sun, 28 Apr 85 18:05:38 pdt
From: Jeff Ullman <ullman@diablo>
Subject: NAIL! Prototype
To: nail@diablo
I've been talking with Allen Van G. and Dave Maier, and they've
convinced me that Prolog is actually well suited to implementing
another logic-based language using (a subset of) its own syntax.
One of the key benefits is that Prolog takes its own metasymbols,
like :-, very seriously when reading its own input.
The idea expressed in Allen's message, of using a restricted Prolog
as our input language, sounds feasible.
Probably, the system, written in cprolog, would be 100 times
slower than a system written in C or a similar language.
I'm not all that concerned with the speed of a *prototype* system,
though. There seemed to be some confusion at the Wednesday
meeting about what we were really going to do. The intent of
the proposals that provided funding for the NAIL! project is
clear: to design a database system with a query language
roughly as expressive as Prolog, but with the system doing
as much as possible of the decision regarding how rules are
to be applied.
I think the sponsors (NSF and IBM) would be happy to see good
ideas, even if they digressed from the proposals, but in the
absence of an alternative target of intellectual interest,
let's stick to the original goal.
The alternative model, of mating a knowledge system with a DBMS,
seems to me unlikely to provide an opportunity for scientific
research, although I do not want to imply that the problems are
easy ones. The reason is that, as I understand it, commercial
knowledge systems essentially provide a rule-based language,
and the user has to solve the optimization problems in the code he
writes. To make the KS do the optimization would require a
complete redesign, which is close to what I propose we do.
However, we can learn a great deal about query optimization for
rule-based query languages with a system that does not have
a real DBMS behind it; we do have to remember what the characteristics
of such a system will be, i.e., how it retrieves its data from
secondary storage. Thus, I propose that we either write in
prolog or take another path that makes the coding relatively
easy, at the expense of performance, see what we can learn,
and then, perhaps, we shall try a real system, with real performance.
One other issue that kept surfacing is the need for some
real examples of problems that the system could solve.
I heard: (1) VLSI/CAD (2) scene understanding (3) text formatting
(4) program style criticism. Anybody have other suggestions
for problems that would be facilitated by a system that both
handled large amounts of raw data efficiently and handled
large numbers of rules efficiently?
Undoubtedly, we shall have to try our hand at one or more
of these, although we may get some help from experts in these
areas. Possibly "rule bases" exist in these areas that could
be translated into a restricted Prolog.
By the way, does everyone know that you can mail messages to
nail@diablo to further the discussion.
For example, would someone like to speak for another implementation
language besides Prolog?
∂28-Apr-85 1817 ullman@diablo A Thought on System Outline
Received: from SU-AIMVAX.ARPA by SU-AI.ARPA with TCP; 28 Apr 85 18:17:09 PDT
Date: Sun, 28 Apr 85 18:16:20 pdt
From: Jeff Ullman <ullman@diablo>
Subject: A Thought on System Outline
To: nail@diablo
One of the big worries I have concerning the design of NAIL! is
the problem of dealing with very large rule/goal graphs.
Roughly, the choice seemed to be between a time-consuming top-down
search from the query node and a space-consuming bottom-up
pre-capturing of all possible nodes.
Something that Moshe said got me thinking:
If we look at the rule/goal graph without adornments, then it is
likely to divide into many small strong components.
Why? Because people tend to write modular programs; if the rules
fall into a few giant components, then likely the programmer doesn't
understand what he is doing anyway.
If that is the case, we can establish bottom-up, once and for all,
the dependencies among strong components that tell us which need
to be captured to capture a query node in any component.
Given a query node with particular adornment, we attempt to capture
a strong component at a time. In so doing, we may need to
ask about predecessor components such things as whether a node
therein can be captured with a particular adornment, or whether
an inequality between two argument sizes holds.
At worst, though this analysis will be exponential in the
number of components, which is less than the number of rules+predicate
symbols, and << the rule/goal graph in general.
∂28-Apr-85 1821 ullman@diablo Next meeting
Received: from SU-AIMVAX.ARPA by SU-AI.ARPA with TCP; 28 Apr 85 18:20:53 PDT
Date: Sun, 28 Apr 85 18:19:50 pdt
From: Jeff Ullman <ullman@diablo>
Subject: Next meeting
To: nail@diablo
Nobody seems to want to volunteer to present papers anymore,
although we have a volunteer for 5/8. How about it folks?
Anyway, I'd like to take 5/1 to talk about some of the ideas
that were developed while Dave Maier was here. The thrust
is that (1) we don't know all that much about how to prove one
rule processing method is better than another, and (2) there may
be a general methodology, based on Henschen and Naqvi's idea
(if we can ever figure out what that really is), but applicable
to more general logic than "linear rules."
∂28-Apr-85 2139 cheriton@Pescadero Re: NAIL! Prototype
Received: from SU-AIMVAX.ARPA by SU-AI.ARPA with TCP; 28 Apr 85 21:39:36 PDT
Received: from Pescadero by diablo with TCP; Sun, 28 Apr 85 21:38:43 pdt
Date: Sun, 28 Apr 85 21:37:42 pdt
From: David Cheriton <cheriton@Pescadero>
Subject: Re: NAIL! Prototype
To: nail@diablo, ullman@diablo
Now that you are considering things 100 times as slow as prolog ...
It might be worth having someone take a look at MRS - Genesereth's
creation. I'm not really suggesting it as a better or even appropriate
language since I dont really know it and you have expertise in Prolog.
However, it might be worth knowing how it would compare with Prolog,
if nothing else.
∂29-Apr-85 0924 @SU-CSLI.ARPA:BrianSmith.pa@Xerox.ARPA RRR: John Perry Tomorrow
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 29 Apr 85 09:23:50 PDT
Received: from Xerox.ARPA by SU-CSLI.ARPA with TCP; Mon 29 Apr 85 09:20:12-PDT
Received: from Semillon.ms by ArpaGateway.ms ; 29 APR 85 09:18:09 PDT
Date: 29 Apr 85 09:17 PDT
From: BrianSmith.pa@Xerox.ARPA
Subject: RRR: John Perry Tomorrow
To: RRR@SU-CSLI.ARPA
cc: BrianSmith.pa@Xerox.ARPA
Just a reminder that John Perry will be leading tomorrow's discussion (4/30).
2:15 in the Ventura seminar room, as usual.
Brian
∂29-Apr-85 1057 roy@diablo Re: system outline.
Received: from SU-AIMVAX.ARPA by SU-AI.ARPA with TCP; 29 Apr 85 10:57:39 PDT
Date: Mon, 29 Apr 85 10:55:09 pdt
From: Shaibal Roy <roy@diablo>
Subject: Re: system outline.
To: ullman@diablo
Cc: nail@diablo
Here are copies two messages regarding Jeff's suggestions.
-shaibal
------
From: Shaibal Roy <roy@diablo>
Subject: Re: system outline.
To: ullman@diablo
> Roughly, the choice seemed to be between a time-consuming top-down
> search from the query node and a space-consuming bottom-up
> pre-capturing of all possible nodes.
If you insist that the search never forgets anything,
then the top-down method should requires no more time
than the bottom-up search (amotized time - you try any
one node at most once). The amount of memory required
is linear in #rules+#predicates, and exponential in the
maximum number of variables in a rule. If you're willing
to assume that the latter is bounded, then this doesn't
cost anything more than the rule/goal graph itself.
> If that is the case, we can establish bottom-up, once and for all,
> the dependencies among strong components that tell us which need
> to be captured to capture a query node in any component.
> ....
> At worst, though this analysis will be exponential in the
> number of components, which is less than the number of rules+predicate
> symbols, and << the rule/goal graph in general.
If you do take in account of the adornments, then the amount
of pre-processing indicated above is potentially exponential
in the number of variables in the components. For example, if
you try p↑bb in
p(x,y) :- q(x,z),r(z,y)
you can potentially get two alternative sets of components,
one for {q↑bf,r↑bb}, and the other for {q↑bb,r↑fb}. That
suggests that the analysis is exponetial in both the number
of components, as well as the number of variables in the
components (which brings us back to exponential in length of
the program). Am I missing something? Do you have some way
of avoiding the effect of replication of components when you
do take in account of the adornments?
-shaibal
------
From ullman Mon Apr 29 09:49:11 1985
Date: Mon, 29 Apr 85 09:49:02 pdt
From: Jeff Ullman <ullman@diablo
Subject: Re: system outline.
To: roy@diablo
Your analysis is correct. I'm sort of assuming that in practice
the number of combinations of b/f* that we need to consider
will be small--that was the intent of your programming project,
no?
By the way, should I forward your message to nail, or would you
like to do so?
∂29-Apr-85 1116 CLAY@SU-CSLI.ARPA Turing use today
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 29 Apr 85 11:16:19 PDT
Date: Mon 29 Apr 85 11:09:35-PDT
From: Clay Andres <CLAY@SU-CSLI.ARPA>
Subject: Turing use today
To: folks@SU-CSLI.ARPA
Several people are working on the final stages of the NSF initiative.
It is very important that they have unimpeded access to Turing for all
of today. Therefore, I am asking all other users to make as much disk
space and CPU time available as possible.
Please do not do any CPU intensive work today. Mail and editing are
fine, but do not run Scribe, TeX, or any compilers.
The Public Structure is filling up, so please delete and expunge any
unneeded files from your directories. This will assure that there is
plenty of room for the large files being created by the NSF project.
Your help in this will be greatly appreciated.
Clay
-------
-------
∂29-Apr-85 1609 ullman@diablo MRS
Received: from SU-AIMVAX.ARPA by SU-AI.ARPA with TCP; 29 Apr 85 16:08:53 PDT
Date: Mon, 29 Apr 85 16:00:55 pdt
From: Jeff Ullman <ullman@diablo>
Subject: MRS
To: nail@diablo
I received the following suggestion from Ben Grosof.
Comments are welcome.
***********************************************
Received: from SU-SCORE.ARPA by diablo with TCP; Sun, 28 Apr 85 18:20:06 pdt
Date: Sun 28 Apr 85 18:19:17-PDT
From: Benjamin N. Grosof <GROSOF@SU-SCORE.ARPA>
Subject: Re: NAIL! Prototype
To: ullman@SU-AIMVAX.ARPA
Cc: grosof@SU-SCORE.ARPA
In-Reply-To: Message from "Jeff Ullman <ullman@diablo>" of Sun 28 Apr 85 18:07:23-PDT
Hi Jeff,
I'd like to propose you consider MRS as the implementation language.
Maybe Allen and I could discuss it a bit. MRS seems to me to have the
same advantage you cited for PROLOG, if I understand you, namely that
it will take a subset of itself as input. It has the advantage of
already existing, letting you get up a prototype more quickly, perhaps.
Moreover, it is more general than the characteristic you cited of
most commercial rul-based "knowledge-based" systems, namely it lets one
specify the control via rules and thus in essentially the higher-level
fasion you would be exploiting PROLOG for. Since it is more powerful than
PROLOG in this respect, I think it is worth seriously considering.
Benjamin
-------
∂29-Apr-85 1743 INGRID@SU-CSLI.ARPA Meeting tomorrow
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 29 Apr 85 17:43:19 PDT
Date: Mon 29 Apr 85 17:35:25-PDT
From: Ingrid Deiwiks - 497-3084 <INGRID@SU-CSLI.ARPA>
Subject: Meeting tomorrow
To: RRR@SU-CSLI.ARPA
John's paper for tomorrow's meeting, "Circumstantial Attitudes and
Benevolent Cognition", can be picked up from my office.
Ingrid
-------
∂29-Apr-85 2355 @SU-CSLI.ARPA:PCOHEN@SRI-AI.ARPA Next meeting of NL4 (Language and Action)
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 29 Apr 85 23:55:18 PDT
Received: from SRI-AI.ARPA by SU-CSLI.ARPA with TCP; Mon 29 Apr 85 23:52:15-PDT
Date: Mon 29 Apr 85 23:53:23-PDT
From: Phil Cohen <PCOHEN@SRI-AI.ARPA>
Subject: Next meeting of NL4 (Language and Action)
To: NL4: ;, f4@SU-CSLI.ARPA, finterest@SU-CSLI.ARPA
cc: bmoore@SRI-AI.ARPA
Our next meeting will be Tues, May 7, in the Ventura seminar room.
Bob Moore will speak on:
A POSSIBLE-WORLD ANALYIS OF DESIRE
Belief-Desire-Intention (BDI) models of rational agents are
used in many otherwise disparate research projects within CSLI. While
formal analyses of belief abound, analyses of intention are much less
developed, and analyses of desire are all but non-existent. This talk
attempts to help remedy the situation by presenting a formal analysis
of desire in terms of preferences among possible worlds. The analysis
gives an account of desire that allows for such things as
* conflicting desires
* not desiring all entailments of a desire
* not desiring all possible means to a desired end
-------
∂29-Apr-85 2355 @SU-CSLI.ARPA:PCOHEN@SRI-AI.ARPA Next meeting of NL4 (Language and Action)
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 29 Apr 85 23:55:18 PDT
Received: from SRI-AI.ARPA by SU-CSLI.ARPA with TCP; Mon 29 Apr 85 23:52:15-PDT
Date: Mon 29 Apr 85 23:53:23-PDT
From: Phil Cohen <PCOHEN@SRI-AI.ARPA>
Subject: Next meeting of NL4 (Language and Action)
To: NL4: ;, f4@SU-CSLI.ARPA, finterest@SU-CSLI.ARPA
cc: bmoore@SRI-AI.ARPA
Our next meeting will be Tues, May 7, in the Ventura seminar room.
Bob Moore will speak on:
A POSSIBLE-WORLD ANALYIS OF DESIRE
Belief-Desire-Intention (BDI) models of rational agents are
used in many otherwise disparate research projects within CSLI. While
formal analyses of belief abound, analyses of intention are much less
developed, and analyses of desire are all but non-existent. This talk
attempts to help remedy the situation by presenting a formal analysis
of desire in terms of preferences among possible worlds. The analysis
gives an account of desire that allows for such things as
* conflicting desires
* not desiring all entailments of a desire
* not desiring all possible means to a desired end
-------
∂29-Apr-85 2355 @SU-CSLI.ARPA:PCOHEN@SRI-AI.ARPA Next meeting of NL4 (Language and Action)
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 29 Apr 85 23:55:18 PDT
Received: from SRI-AI.ARPA by SU-CSLI.ARPA with TCP; Mon 29 Apr 85 23:52:15-PDT
Date: Mon 29 Apr 85 23:53:23-PDT
From: Phil Cohen <PCOHEN@SRI-AI.ARPA>
Subject: Next meeting of NL4 (Language and Action)
To: NL4: ;, f4@SU-CSLI.ARPA, finterest@SU-CSLI.ARPA
cc: bmoore@SRI-AI.ARPA
Our next meeting will be Tues, May 7, in the Ventura seminar room.
Bob Moore will speak on:
A POSSIBLE-WORLD ANALYIS OF DESIRE
Belief-Desire-Intention (BDI) models of rational agents are
used in many otherwise disparate research projects within CSLI. While
formal analyses of belief abound, analyses of intention are much less
developed, and analyses of desire are all but non-existent. This talk
attempts to help remedy the situation by presenting a formal analysis
of desire in terms of preferences among possible worlds. The analysis
gives an account of desire that allows for such things as
* conflicting desires
* not desiring all entailments of a desire
* not desiring all possible means to a desired end
-------
∂30-Apr-85 0038 BRESNAN@SU-CSLI.ARPA housekeeper
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 30 Apr 85 00:38:50 PDT
Date: Tue 30 Apr 85 00:36:26-PDT
From: Joan Bresnan <BRESNAN@SU-CSLI.ARPA>
Subject: housekeeper
To: research@SU-CSLI.ARPA
cc: bresnan@SU-CSLI.ARPA
My housekeeper, Francoise Rihon, is looking for extra work, both
for herself and for her friend. If anyone has such work, I can
provide phone numbers and references.
-------
∂30-Apr-85 0920 @SU-CSLI.ARPA:PCOHEN@SRI-AI.ARPA Time for NL4 meeting
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 30 Apr 85 09:20:32 PDT
Received: from SRI-AI.ARPA by SU-CSLI.ARPA with TCP; Tue 30 Apr 85 09:16:07-PDT
Date: Tue 30 Apr 85 09:17:06-PDT
From: Phil Cohen <PCOHEN@SRI-AI.ARPA>
Subject: Time for NL4 meeting
To: NL4: ;, f4@SU-CSLI.ARPA, finterest@SU-CSLI.ARPA
cc: bmoore@SRI-AI.ARPA
Sorry, I forgot to mention the time of the NL4 meeting. 12:45 - 2:15,
Tues. May 7, in the Ventura conf. room. Bob Moore will speak on a
possible worlds analysis of desire. See you there!
Phil
-------
∂30-Apr-85 0920 @SU-CSLI.ARPA:PCOHEN@SRI-AI.ARPA Time for NL4 meeting
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 30 Apr 85 09:20:32 PDT
Received: from SRI-AI.ARPA by SU-CSLI.ARPA with TCP; Tue 30 Apr 85 09:16:07-PDT
Date: Tue 30 Apr 85 09:17:06-PDT
From: Phil Cohen <PCOHEN@SRI-AI.ARPA>
Subject: Time for NL4 meeting
To: NL4: ;, f4@SU-CSLI.ARPA, finterest@SU-CSLI.ARPA
cc: bmoore@SRI-AI.ARPA
Sorry, I forgot to mention the time of the NL4 meeting. 12:45 - 2:15,
Tues. May 7, in the Ventura conf. room. Bob Moore will speak on a
possible worlds analysis of desire. See you there!
Phil
-------
∂30-Apr-85 0920 @SU-CSLI.ARPA:PCOHEN@SRI-AI.ARPA Time for NL4 meeting
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 30 Apr 85 09:20:32 PDT
Received: from SRI-AI.ARPA by SU-CSLI.ARPA with TCP; Tue 30 Apr 85 09:16:07-PDT
Date: Tue 30 Apr 85 09:17:06-PDT
From: Phil Cohen <PCOHEN@SRI-AI.ARPA>
Subject: Time for NL4 meeting
To: NL4: ;, f4@SU-CSLI.ARPA, finterest@SU-CSLI.ARPA
cc: bmoore@SRI-AI.ARPA
Sorry, I forgot to mention the time of the NL4 meeting. 12:45 - 2:15,
Tues. May 7, in the Ventura conf. room. Bob Moore will speak on a
possible worlds analysis of desire. See you there!
Phil
-------
∂30-Apr-85 1221 @SU-SCORE.ARPA:JF@SU-SUSHI.ARPA may 17 bats
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 30 Apr 85 12:21:08 PDT
Received: from SU-SUSHI.ARPA by SU-SCORE.ARPA with TCP; Tue 30 Apr 85 12:15:57-PDT
Date: Tue 30 Apr 85 12:15:29-PDT
From: Joan Feigenbaum <JF@SU-SUSHI.ARPA>
Subject: may 17 bats
To: aflb.su@SU-SCORE.ARPA
Please tell me whether you plan to attend the May 17 bats at stanford
so that I can give an accurate estimate to the caterers. The speakers
will be Gilles Brassard (Montreal via Berkeley), Alan Siegel (NYU),
Andrei Broder (DECSRC), and Ernst Mayr (Stanford).
I have only Siegel's abstract, which I enclose.
From: siegel@NYU-CSD2 (Alan R. Siegel)
To: JF@SU-SUSHI.ARPA
BLower Bounds for Optimal VLSI Sorters
Alan Siegel, New York University
ABSTRACT
We analyze VLSI sorting problems for N numbers in the range [0, M].
The principal portion of this talk will establish the
minimum area necessary for a sorting circuit that can read its input data
R <= N times. Although this problem seems rather peculiar, the analysis
exposes the central information theoretic issues
underlying AT↑2 bounds for sorting numbers in general ranges.
As a consequence, we attain bounds for the AT↑2 complexity of
a standard VLSI (read-once) sorter, for all ranges of M.
(Subsequent joint work with R. Cole has shown that these bounds are
tight for virtually all cases.)
This talk will be mostly combinatorial in nature.
Please reply to me here at jf@sushi
Thanks,
Joan
-------
∂30-Apr-85 1225 PAPA@SU-SCORE.ARPA STOC forms
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 30 Apr 85 12:25:20 PDT
Date: Tue 30 Apr 85 12:22:35-PDT
From: C. Papadimitriou <PAPA@SU-SCORE.ARPA>
Subject: STOC forms
To: aflb.su@SU-SCORE.ARPA
Anybody has the STO registration and accommodation forms handy?
Send me a message, and I will come borrow it to make a copy.
I thought it was not too early for this...
Thanks,
---Christos.
-------
∂30-Apr-85 1413 WRIGHT@SUMEX-AIM.ARPA SIGLUNCH Friday, May 3
Received: from SUMEX-AIM.ARPA by SU-AI.ARPA with TCP; 30 Apr 85 14:13:11 PDT
Date: Tue 30 Apr 85 14:07:24-PDT
From: Carol Wright/Susie Barnes <WRIGHT@SUMEX-AIM.ARPA>
Subject: SIGLUNCH Friday, May 3
To: Siglunch@SUMEX-AIM.ARPA
SIGLUNCH
DATE: Friday, May 3, 1985
LOCATION: Braun Auditorium in Mudd/Chemistry Building
TIME: 12:05-1:05
SPEAKER: Barbara Hayes-Roth
Helix Group, Knowledge Systems Laboratory
TITLE: BB1: An Architecture for Blackboard Systems
BB1 is a domain-independent blackboard architecture. The prototypical
blackboard system includes: functionally independent knowledge sources to
generate solution elements; a structured blackboard to record solution
elements and mediate knowledge source interactions; and a control mechanism
to trigger, schedule, and execute knowledge sources. Building upon this basic
functionality, BB1 also provides capabilities for dynamic control planning,
strategic explanation, and automatic learning of control heuristics. BB1 will
be illustrated with examples from PROTEAN, a system that integrates multiple
constraints to determine the 3-D structures of proteins in solution.
-------
-------
∂30-Apr-85 1520 @SU-SCORE.ARPA:JF@SU-SUSHI.ARPA thanks
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 30 Apr 85 15:19:52 PDT
Received: from SU-SUSHI.ARPA by SU-SCORE.ARPA with TCP; Tue 30 Apr 85 15:16:41-PDT
Date: Tue 30 Apr 85 15:12:57-PDT
From: Joan Feigenbaum <JF@SU-SUSHI.ARPA>
Subject: thanks
To: aflb.su@SU-SCORE.ARPA
I've never had such prompt response to a BATS announcement. Thanks for
being so considerate. And keep those reponses coming!
Joan
-------
∂30-Apr-85 1533 @SU-SCORE.ARPA:WELCH@Ames-VMSB SIGBIG
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 30 Apr 85 15:32:52 PDT
Received: from Ames-VMSB.ARPA by SU-SCORE.ARPA with TCP; Tue 30 Apr 85 15:24:45-PDT
Date: 30 Apr 1985 1458-PDT
From: WELCH at Ames-VMSB
Subject: SIGBIG
To: SUPER at SU-SCORE.ARPA.ARPA
Reply-To: WELCH@Ames-VMSB
ASSOCIATION FOR COMPUTING MACHINERY
San Francisco Golden Gate Chapter
"SIGBIG" Special Interest Committee
For Large High Speed Computers
Meetings on the first Wednesday of each month at 7:30 PM. Speakers
who can give insights to various aspects of SUPERCOMPUTING are
featured each month.
For information contact Mary Fowler, Chairperson (415) 965-6515
or Mike Austin, Publ. Chair (415) 423-8446
Next meeting:
Wednesday, May 1,1985, 7:30 PM
Speaker: Alvin M Despain University of California, Berkeley
Subject: The AQUARIUS PROJECT
Location: Computer Science Lounge
597 Evans Hall
U.C. Berkeley
Abstract:
A new approach to computer orgainzation and architecture is
investigated to achieve a radical improvement in performance.
Parallel Architectures, Very fast circuits; VLSI implementation;
and PROLOG are studied to achieve a breakthrough in problem
solution.
---------------------------------------------------------------
Tape-recordings of most of the previous may be
obtained in exchange for a tape cassette or $5.00 by contacting:
Tom Attwood (415) 965-6551
Future Meetings:
June 5, 1985 Richard Stamm/Convex The CONVEX C-1
July 3, 1985 Bill Joy/Sun Unix and Supercomputers
Please contact either of the following individuals if you are
interested in arranging car-pooling to the meeting in Berkeley.
Mary Fowler at email- mer.fowler@ames-vmsb.arpa or
Eugene Miya at email- ames!eugene@riacs.arpa
------
∂30-Apr-85 1637 ullman@diablo meetings
Received: from SU-AIMVAX.ARPA by SU-AI.ARPA with TCP; 30 Apr 85 16:37:53 PDT
Date: Tue, 30 Apr 85 16:35:15 pdt
From: Jeff Ullman <ullman@diablo>
Subject: meetings
To: nail@diablo
1. Allen and I will talk about crazy ways to handle linear rules
at the usual nail meeting Wednesday 11AM
2. Note that the circumscription seminar is covering the Closed-World
Assumption at 2PM Weds.
∂30-Apr-85 2010 JOHN@SU-CSLI.ARPA Housing Wanted
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 30 Apr 85 20:10:13 PDT
Date: Tue 30 Apr 85 20:06:28-PDT
From: John Perry <JOHN@SU-CSLI.ARPA>
Subject: Housing Wanted
To: folks@SU-CSLI.ARPA
High school French teacher, in Palo Alto June 23-Aug 3.
Needs place to house-sit or rent inexpensively.
Call 714-787-5208, 5-8
202-547-4368 evenings
ask for Tamara
-------
∂01-May-85 0043 RESTIVO@SU-SCORE.ARPA PROLOG Digest V3 #20
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 1 May 85 00:43:02 PDT
Date: Tuesday, April 30, 1985 7:05PM
From: Chuck Restivo (The Moderator) <PROLOG-REQUEST@SU-SCORE.ARPA>
Reply-to: PROLOG@SU-SCORE.ARPA
US-Mail: P.O. Box 4584, Stanford CA 94305
Phone: (415) 326-5550
Subject: PROLOG Digest V3 #20
To: PROLOG@SU-SCORE.ARPA
PROLOG Digest Wednesday, 1 May 1985 Volume 3 : Issue 20
Today's Topics:
Puzzles - AVG's Riddle
Implementation - C-Prolog & Semantics & Control & CP
LP Library - Update
----------------------------------------------------------------------
Date: Wed, 17 Apr 85 18:16:35 pst
From: Allen VanGelder <AVG@Diablo>
Subject: Answer to Allen VanGelder's Riddle
From: Ran Ever-Hadani
Subject: Answer to Allen VanGelder's riddle
p('6').
-- Ran Ever-Hadani
This does NOT work in C Prolog to produce my results.
?- p(X). answers X = '6', and not X = 6. However,
Ran's program does work in DEC-20 Prolog, but only if
you use write instead of writeq. My riddle (see
Digest Vol 3 Issue 16) has the same output if writeq
is used instead of write, but works only on C Prolog,
not DEC-20 Prolog. So this is very much an
implementation dependent riddle.
------------------------------
Date: 23 Apr 85 16:43:06 +1000 (Tue)
From: decvax!mulga!mungunni.oz!Lee@Berkeley
Subject: Bugs in listing
Here's a little test for the listing predicate of
your favourite Prolog system with DEC-10 style
syntax. The test is to read in a clause, write it
out with listing then read it in again and see if
it is the same. Many systems will get syntax
errors when they try to read it back in or just
get it wrong.
=======================consult this=====================
p((a,b)) :-
q(','),
- -r,
s(['+x+',(-),'Fred']),
t('/*', '%'),
(+)-(+)=(.)-(.),
not not v is w,
w(','(a,b,'.')),
a = && .
?- write('TESTING...'), nl,
tell(lstest), listing(p), told,
seeing(S), see(lstest), read(X), see(S),
clause(p(Y), Z),
X=(p(Y):-Z),
write('SUCCESS'), nl.
=========================================================
With C-Prolog 1.4d, the extra parentheses on line 1 are
missing and there is no space (or parenthesis) between
"&&" and ".", so it runs into the end of file.
With UNSW Prolog (the test needed to be modified
slightly) the "-" in the list should have been
parenthesised but wasn't, "." was not quoted or
parenthesised and there was no space after "&&".
With previous versions of MU-Prolog, there were
several problems with lack of quotes, spaces and
parentheses. The current version works, though
it does print some unnecessary parentheses and
spaces (as does C Prolog). If anyone has any other
nasty tests, please post them.
-- Lee Naish
------------------------------
Date: Thu 18 Apr 85 12:11:06-MST
From: Uday Reddy <U-REDDY@UTAH-20.ARPA>
Subject: Semantics
Notwithstanding our differences in the styles and
approaches to semantic specification, I think Joseph
Goguen and I agree on the essentials. My summary of the
whole discussion is the following.
Firstly, programming languages need a semantics that is
compositional, whereby the meaning of composite
constructs is expressed solely in terms of the meanings
of their constituents. This is necessary so that we can
understand predicates or functions defined by the
programs by abstracting over the implementation/
definition details. Whether this semantic specification
is done using conventional model theory or domain theory
or any of their variants or extensions is open to
debate, taste and choice.
Secondly, the operational semantics/proof theory of the
language has to be sound and complete with respect to
the semantics. This should be so, for the semantics to
have any real use in practice, so that what we
understand by the programs is exactly what is computed
by the machine.
The so called "control" mechanisms currently used with
various logic languages create a wide gap between the
proof theory and the semantics. For instance, the van
Emden-Kowalski semantics is too far from sequential
Prolog. We, the users, are then left with a semantics
that is simple and well-understood, but useless in
practice, and a language which can only be understood by
tracing the states of a machine (however abstract the
machine may be).
The first possible solution is to specify clearly the
conditions under which the the operational semantics is
complete with respect to the compositional semantics
(assuming that it is at least sound). For instance,
rewriting languages specify that the implementations are
complete with respect to the initial model, only if the
rewriting system is strongly terminating.
A second solution is to specify the compositional
semantics in a theory that can account for control
aspects (termination, failure and so on). Possibly all
control features can be explained in an adequately
general theory, but the theory may be too complex to be
of any pragmatic relevance. We should then reject
languages based on the complexity of the theory required
for their compositional semantics. This is the solution
I favor.
A third and longer-term solution is to find control
mechanisms that retain completeness, at least under
well-understood sufficient conditions. This is the
solution Joseph favors.
With my experience in the use of logic languages and
their implementations, I am personally skeptical about
the viability of the third solution.
-- Uday Reddy
------------------------------
Date: 19 Apr 1985 1625-CST
From: THRIFT%ti-csl.csnet@csnet-relay.arpa
Subject: Control annotation in CP
There has been much recent discussion [3] on the
'correct' unification definition when read-only
variables are concerned (as whether two read-onlys
can unify, whether a read-only can unify with a
regular variable, etc.) My basic objection to this
particular control annotation is as follows:
Variables may 'become' read-only during execution
(at least in the original definition of unification
with read-onlys [1,2]). This makes understanding
of a program difficult. I basically believe it is
confusing when a control annotation passed around
during execution.
A control annotation (bearing some similarities some
some of the recent suggestions in [3]) is what I call
a 'mode' annotation, somewhat like the mode declaration
for the DEC-10 compiler. There are two symbols '+' and
'-' which may annotate the (surface) arguments of the
head. The annotations describe preconditions before
unification can be attempted: a '+' argument can only
be unified with a nonvar term and a '-' argument can
only be unified with a var term (input and output
restrictions).
An example is the merge definition:
merge([]+,Y,Y).
merge(X,[]+,X).
merge([X1|Xs]+,Y,[X1|Z]) :-
merge(Y,Xs,Z).
merge(X,[Y1|Ys]+,Z) :-
merge(Ys,X,Z).
which implements a fair merge (stable machine).
Suppose for some reason one wanted to specify a
priority merge (that if there are elements on one
stream they are guaranteed to be output before
elements of the other, which have to wait until
until the first stream is 'waiting'). This could
be done as follows:
merge([]+,Y,Y).
merge(X,[]+,X).
merge([X1|Xs]+,Y,[X1|Z]) :-
merge(Xs,Y,Z).
merge(X-,[Y1|Ys]+,Z) :- % The second stream is ouput only
merge(X,Ys,Z). % when the first is 'waiting'
This could not be done as neatly with the read-only
annotation (this depends I guess on which definition
of read-only unification you choose).
With this notation a definition of lt(X,Y) as given
in [1]:
lt(X,Y) :- wait(X,X1),wait(Y,Y1) | X1<Y1.
is simply
lt(X+,Y+) :- X<Y.
'plus' as in [2] is:
plus(X+,Y+,Z) :- Z is X+Y.
plus(X,Y+,Z+) :- X is Z-Y.
plus(X+,Y,Z+) :- Y is Z-X.
I basically feel that this control annotation
1. Makes programs much easier to understand in that
a. the control is associated with the definition of
a process and not its execution
b. the control annotation is never passed around
c. no wait/2 built-in is required
2. Is sufficient for most of what one wanted the original
'?' notation (I believe it is completely sufficient for
practical purposes)
3. Could possibly make for a more efficient implementation.
[1] Shapiro,E.Y.:A Subset of Concurrent Prolog and Its
Interpreter,ICOT Technical Report TR-003 (1983)
[2] Shapiro,E.Y.:Object Oriented Programming in Concurrent
Prolog,New Generation Computing,1 (1983)
[3] Prolog@SU-Score:Prolog Digest Vol3 ...
------------------------------
Date: Tue 30 Apr 85 12:03:21-PDT
From: Chuck Restivo (The Moderator) <RESTIVO@SU-SCORE.ARPA>
Subject: Update
[cwr] Professor Zohar Manna and Richard Waldinger
authored a new book that might be interesting to
to some readers.
"The Logical Basis for Computer Programming"
Volume 1: Deductive Reasoning
ISBN: 0-201-18620-2 Addison Wesley
------------------------------
End of PROLOG Digest
********************
∂01-May-85 0852 EMMA@SU-CSLI.ARPA Newsletter
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 1 May 85 08:52:35 PDT
Date: Wed 1 May 85 08:48:29-PDT
From: Emma Pease <Emma@SU-CSLI.ARPA>
Subject: Newsletter
To: folks@SU-CSLI.ARPA
Tel: 497-3479
I must ask that any articles for tomorrow's newsletter be sent in
as early as possible today. I cannot promise that anything late (after
noon) will be printed.
Many thanks,
Emma
-------
∂01-May-85 0858 EMMA@SU-CSLI.ARPA Tuesday teas
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 1 May 85 08:58:15 PDT
Date: Wed 1 May 85 08:52:25-PDT
From: Emma Pease <Emma@SU-CSLI.ARPA>
Subject: Tuesday teas
To: folks@SU-CSLI.ARPA, consultants@SU-CSLI.ARPA, bboard@SU-CSLI.ARPA
Tel: 497-3479
For various reasons Tuesday teas will start at 3:00 instead of 3:30
during the immediate future.
-Emma
-------
∂01-May-85 1647 EMMA@SU-CSLI.ARPA Newsletter May 2, No. 27
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 1 May 85 16:47:34 PDT
Date: Wed 1 May 85 16:34:55-PDT
From: Emma Pease <Emma@SU-CSLI.ARPA>
Subject: Newsletter May 2, No. 27
To: friends@SU-CSLI.ARPA
Tel: 497-3479
C S L I N E W S L E T T E R
←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←
May 2, 1985 Stanford Vol. 2, No. 27
←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←
A weekly publication of The Center for the Study of Language and
Information, Ventura Hall, Stanford University, Stanford, CA 94305
←←←←←←←←←←←←
CSLI ACTIVITIES FOR *THIS* THURSDAY, May 2, 1985
12 noon TINLunch
Ventura Hall ``Categorizing the Senses of `Take' ''
Conference Room by Peter Norvig
Discussion led by Douglas Edwards
2:15 p.m. CSLI Seminar
Redwood Hall ``Property Theory and Second-Order Logic''
Room G-19 Chris Menzel, CSLI
3:30 p.m. Tea
Ventura Hall
4:15 p.m. CSLI Colloquium
Redwood Hall ``A Formal Theory of Innate Linguistic Knowledge''
Room G-19 Janet Fodor, University of Connecticut
Originally scheduled for April 11
←←←←←←←←←←←
CSLI ACTIVITIES FOR *NEXT* THURSDAY, May 9, 1985
12 noon TINLunch
Ventura Hall ``Scenes and Events''
Conference Room by Steven Neale, Dept. of Linguistics, Stanford
(Abstract on page 2)
2:15 p.m. CSLI Seminar
Redwood Hall ``Approaches to Generalized Quantifiers in
Room G-19 Heim/Kamp Semantics''
Mats Rooth, CSLI
(Abstract on page 2)
3:30 p.m. Tea
Ventura Hall
4:15 p.m. CSLI Colloquium
Redwood Hall ``Reduced Forms of Comparative Clauses''
Room G-19 James D. McCawley, University of Chicago
(Abstract on page 3)
!
Page 2 CSLI Newsletter May 2, 1985
←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←
ABSTRACT OF NEXT WEEK'S TINLUNCH
``Scenes and Events''
In his paper ``The Logic of Perceptual Reports", Jim Higginbotham
presents an alternative to Situation Semantics' treatment of the
semantics of naked-infinitive perceptual reports such as ``John saw
Mary wink''. Drawing on an idea of Davidson's, Higginbotham attempts
to make explicit the implicit quantification over events in
NI-perceptual reports by augmenting the valency of certain verbs with
an extra quantifiable place. In this way, he purports to capture
Barwise's semantic generalizations purely formally at a level of
linguistic representation intimately related to LF in
Government-Binding theory.
In ``Scenes and Events'' Stephen Neale critically evaluates
Higginbotham's proposal, concluding that it fails on both semantic and
syntactic grounds: (i) it neither gives an adequate account of the
semantic facts it was meant to account for nor meshes with the sorts
of syntactic considerations which are supposed to motivate it, (ii) it
fails to confront problems which must be encountered by any purely
formal account of certain classes of semantic facts, and (iii) it
admits of no simple incorporation into the GB framework within which
Higginbotham wishes to embed it.
←←←←←←←←←←←←
ABSTRACT OF NEXT WEEK'S SEMINAR
``Approaches to Generalized Quantifiers in Heim/Kamp Semantics''
The original versions of Heim's file change semantics and Kamp's
discourse representation theory treated the quantificational
determiners ``every'' and ``no''. It has been pointed out that
extensions to other quantifiers are not immediate. One problem is
that the variable corresponding to the head of a quantified NP and the
variables corresponding to indefinites in the NP are given equal
status, although ``many a man who owns a donkey beats it'' and ``many
a donkey which is owned by a man is beaten by him'' appear to have
different truth conditions. Recently, generalized quantifier
treatments for DR theory have been proposed by Klein and others. I
will show how Barwise's parameterized set quantifiers can be
considered a theory of generalized quantifiers for file change
semantics, and consider extensions to plurals. --Mats Rooth
!
Page 3 CSLI Newsletter May 2, 1985
←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←
ABSTRACT OF NEXT WEEK'S COLLOQUIUM
``Reduced Forms of Comparative Clauses''
Russell in 1905 observed that ``than''-clauses within a subordinate
clause can be ambiguous with regard to scope, e.g., (1) is ambiguous
with regard to whether the clause introduced by ``than'' is
semantically part of the complement of ``think.'' However, not all
reduced forms of ``than''-clauses exhibit this ambiguity; i.e., for
example, it is absent from the fully reduced ``than''-clause of (2).
(1) I thought your yacht was longer than it is.
(2) I thought your yacht was longer than her yacht.
If the clause introduced by ``than'' or ``as'' is treated as a
definite description (the x such that your yacht is x much long) and
underlying structures are assumed in which quantified expressions
(including definite descriptions) are sisters of the Ss that serve as
their scopes, the difference in possible interpretations of the
different reduced ``than''-clauses follows from the typology of
deletion transformations that distinguishes pronominal deletions,
which are subject only to the general constraints on where pronouns
can occur in relation to their antecedents, from REDUCTIONS, which
delete all but one constituent of an item and are subject to a
locality condition.
The resulting analysis of fully reduced ``than''-clauses, as in
(2), reveals them in fact to be ambiguous, but in a different way from
(1), and, in conjunction with an analysis in which tenses and
auxiliary verbs are external to their host Ss in underlying structure,
accounts for the 3-way ambiguity of (3).
(3) John has eaten more pizza than Bill.
--James McCawley
←←←←←←←←←←←←
ABSTRACT OF AREA NL-1 MEETING
``New Aspects of Aspect: A Look at Mandarin Chinese''
Carlota S. Smith, University of Texas
Friday, May 10, 2:30 pm, Ventura Hall Conference Room
A study of the aspectual system of Mandarin Chinese tests current
approaches to aspect: the system is considerably more complex than
that of familiar Indo-European languages, with several perfectives and
two imperfectives. Certain features of Chinese are particularly
interesting. One perfective involves an interval that spans beyond
the final endpoint of the situation talked about; it requires a
viewpoint component of aspect separate from situation type. Another
perfective, with reduplication, presents a particular situation type.
It can be accounted for with an aspect-changing lexical rule and
suggests the notion of marked, language-specific situation types. The
imperfectives differentiate the internal structure of statives and
non-statives. Finally, the Aristotelian situation types are realized
in Chinese within the general pattern of the language. Some verbs are
subtly different from their English counterparts, realizing different
situation types in each language. Thus, ``die'' is an Accomplishment
in English and an Achievement in Chinese. (No interpretation of this
point is offered.)
-------
∂01-May-85 1759 BARWISE@SU-CSLI.ARPA The fallacy of the homomorphism of content
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 1 May 85 17:59:23 PDT
Date: Wed 1 May 85 17:53:21-PDT
From: Jon Barwise <BARWISE@SU-CSLI.ARPA>
Subject: The fallacy of the homomorphism of content
To: John@SU-CSLI.ARPA
cc: rrr@SU-CSLI.ARPA
John, it seems to me that your point about the disparity that can come
in between signal and content is very important. One way to think
about it is this. Things like "I am not sitting" and the way its
meaning plus context get to interpretation make the interpretation
very much like a homomorphic image of the signal. Your point with the
turn signal, is that this is very special, that other factors can
contribute structure not in the interpretation. I take it that this
is part of what is behind situated automata. So a reasonable thing to
call the view that finds all structure mirrored in the signal would be
the fallacy of the homomorphism of content.
Jon
-------
∂02-May-85 0906 HANRAHAN@SU-SCORE.ARPA SENIOR-FACULTY MEETING
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 2 May 85 09:06:27 PDT
Date: Thu 2 May 85 09:06:14-PDT
From: Katherine Hanrahan <HANRAHAN@SU-SCORE.ARPA>
Subject: SENIOR-FACULTY MEETING
To: senior-faculty@SU-SCORE.ARPA
Stanford-Phone: (415) 497-2273
This is a reminder of the senior-faculty meeting this afternoon at 2:30
in MJH 252.
-------
∂02-May-85 1126 BLOCK@SU-CSLI.ARPA homomorphism of content
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 2 May 85 11:26:42 PDT
Date: Thu 2 May 85 11:13:34-PDT
From: Ned Block <BLOCK@SU-CSLI.ARPA>
Subject: homomorphism of content
To: barwise@SU-CSLI.ARPA
cc: RRR@SU-CSLI.ARPA
Seems to me there are 2 very different ways an unstructured entity can
carry highly structured content. One is exemplified by the turn
signal example. The other is a case where, eg, location of a
signal in a space has different effects in virtue of the causal
properties of the space. So such a space inside the robot could be
used to carry structured content just because points inthe space give
what is located at them very different causal roles. The latter type
of case has nothing to do with external context. Though I have
understood very little of what Stan and Fernando are up to, I would
have thought it has more to do with the latter than the former.
-------
∂02-May-85 1138 @SU-CSLI.ARPA:ISRAEL@SRI-AI.ARPA Re: The fallacy of the homomorphism of content
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 2 May 85 11:38:42 PDT
Received: from SRI-AI.ARPA by SU-CSLI.ARPA with TCP; Thu 2 May 85 11:18:56-PDT
Date: Thu 2 May 85 11:17:05-PDT
From: David Israel <ISRAEL@SRI-AI.ARPA>
Subject: Re: The fallacy of the homomorphism of content
To: BARWISE@SU-CSLI.ARPA
cc: John@SU-CSLI.ARPA, rrr@SU-CSLI.ARPA, ISRAEL@SRI-AI.ARPA
In-Reply-To: Message from "Jon Barwise <BARWISE@SU-CSLI.ARPA>" of Wed 1 May 85 17:57:55-PDT
Subject: Re: The fallacy of the homomorphism of content
To: BARWISE@SU-CSLI.ARPA
cc: John@SU-CSLI.ARPA, rrr@SU-CSLI.ARPA, ISRAEL@SRI-AI.ARPA
In-Reply-To: Message from "Jon Barwise <BARWISE@SU-CSLI.ARPA>" of Wed 1 May 85 17:57:55-PDT
The fallacy of the HOMOMORPHISM OF CONTENT; I like it, it has a
certain je ne sais KWA; I think it'll be BOFFO.
That said, let me go on a little about the point that John was trying
manfully to make AFTER the tea break -- you know the one that got all
the flak and that distracted everybody from the crucial point about
THE FALLACY OF THE HOMOMORPHISM OF CONTENT; the one (if truth be told)
that I was pressing the poor guy to make. (YUP, Jon, I'm to blame, if
not for the point--and I'm not--then for pressing John to make it.)
The picture is as follows: the total functional (mental) state of the
beastie causes it to move its body as it does. This total state,
howver, has many "aspects", or as John puts in the paper, the
psychology of the beastie involves or supports many different
projections. One yields beliefs and desires; others might yield as
well intentions and plans (that's a sop thrown in Bratman's
direction); another yields a level of contents more tightly connected
to the deliverances (ferom and to) the sensory-motor system. John
spoke of "knowing" vs. believing to mark the distinction between the
belief (desire) projection and this latter. This is terminologically
confusing because (1), it allows for knowing that p without believing
that p (which, as it happens, is fine by me) and (2), more seriously,
it divorces knowing from the requiremnet of veridicality--putting in
its place some waving of hands, though IN THE RIGHT DIRECTION, toward
the delievrances of evolutionrily conditioned mechanisms in normal
circumstances. SO, let's follow CHOMSKY and talk of cognizing. (WHAT
THE HELL.)
So, explanations via content of ACTIONS are via contents yielded by
projections at the level of belief (and desire, and maybe intention
and plan...); explanations via content of BODILY MOVEMENTS are getting
at the projection which yields the contents of the beasties cognizing.
The structure of the alternative possibilities that gives rise to
content in the first place are, in general, different at these two
levels. (Here I'm assuming a bigger picture, like that of B&P or
Stalnaker on the nature of content; there is some structure of
possibilities, of possible states of bits and pieces of the world
--for Stalnaker, of the WHOLE DAMN BALL OF WAX--and this structure
indexes the representing states.) That is, the properties and
relations, even the individuals, whose various possible configurations
yield (or are) the contents at these two levels differ. (Of course,
there may also be sub-levels, e.g. within the visual system, as in the
work of Marr, etc...) So, at the level of belief, the squarrel may
have truck with nuts and dogs, etc (or is it with food and predators),
while at the level of visual cognizing, it may truck rather with
generalized cylinders, fractals, superquads, and Gawd knows what all.
So in the case of the pen in front of John; say John really wants to
grab hold of that pen, that pen which he believes is right in front of
him, so he reaches for it and VOILA, grabs it. (Would that the S.F.
Giants could do do so well, eh John?) John's total mental
(functional) state carried the information about the relative position
of the pen - its distance from John's eyes and its displacement from
the center of his visual field - required to govern the movements of
his arm JUST SO. Moreover, and more crucially, the darn thing works
over time, as John leans back and forth, perhaps moves his head
around, etc., constantly updating cahnges in relative location, depth
and displacement. So crucially significant aspects of the content
cognized are changing over time, but the belief that there's that pen
RIGHT THERE IN FRONT OF HIM can remain constant, throughout alll these
changes. (Watch out for the trickiness of "in front of"; but that's a
long story.)
(NOTE RIGHT AWAY: the intended indeterminateness of "as it does" in talk
of the beasties moving as it does. Every time I move my arm, it is
with some velocity (hence momentum) and acceleration (hence force).
Often that the movement has the velocity and acceleration it does is
not governed by aspects of my mental state, though it is so governed
by aspects of my total bodily state PLUS external circumstances. Even
when the bodily movement, as caused by my total mental state can be
characterized as a QUICK movement of the right arm, the exact velocity
and acceleration should not be thought of as being under sensory-motor
control. The issues here are of "granularity", of "least discriminable
differences", on both the sensory and the motor side; needless to say
they are issues about which I have nothing to say.)
So, note the parallel: here, instead of gravity, think of the kinds of
constraints, regularities, cited by Marr in his work on stereopsis.
These regularities need not be (and surely are not) "represnted"
anywhere; however, given them as (stable, enduring) aspects of the
surrounding environments, the deliverances of the visual system are
assigned content (as part of the projection on the total, dynamically
evolving, functional or mental state of the organism) in terms of the
relative locations from the perceiver of objects, together with
certain spatial properties of and spatial relations among those
objects. (I mean, of course, real external objects.) Just as there
need be (and, again, surely in general isn't) any aspect of the
"signal" corresponding to the relatively stable local value of the
gravitational constant, so too there need be no aspects of the
"signal" corresponding to these relatively stable and enduring
regularities. So, too, also, there need be no aspect of the
belief-content corresponding to those parameters, values of which and
changes in which are and must be cognized to control bodily movement;
e.g. the parameters involved in relative location from the organism,
say distance and angle, and the change from 3cm. to 3.2, or in angle
from theta to theta + a teensy-weensy epsilon.
THE END
P.S. This does raise the issue of the role of the concept of belief;
of the utility, for varying kinds of beaties, of the particular
projection onto mental states of which philosopher have traditionally
been so fond. That is a real honest-to-goodness deep
problem--generated to be sure by the picture John was sketching; but
none the worse for that GIVEN the theoretical interest of that
picture. Anyway, since by my lights it IS a real deep issue, I have
nothing whatsoever to say about it.
-------
∂02-May-85 1254 YAMARONE@SU-CSLI.ARPA EXTRA SANDWICHES
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 2 May 85 12:54:11 PDT
Date: Thu 2 May 85 12:47:38-PDT
From: Tom Yamarone <YAMARONE@SU-CSLI.ARPA>
Subject: EXTRA SANDWICHES
To: FOLKS@SU-CSLI.ARPA
YOU MIGHT HAVE NOTICED THERE WAS A SEVERE SHORTAGE OF EXTRA SANDWICHES THIS
WEEK(IF YOU DIDN'T , DISREGARD THIS NOTICE) IN ACCORDANCE WITH OUR CONSUM-
TIVE PATTERNS AND MONETARY FUND.
HOW TO AVOID GOING HUNGRY?
ORDER YOU SANDWICH IRREGARDLESS OF WHETHER OR NOT YOU'RE CERTAIN YOU'LL WANT
IT. DON'T COUNT ON EXTRAS..IF YOU DON'T WANT IT , IT'S AN EXTRA AND WILL
GET SOLD THE NEXT DAY. (IF YOU TELL ME AT NOON , I CAN SELL IT THAT DAY)
MONDAY -WEDNESDAY THERE WILL BE 1-4 EXTRA SANDWICHES
THURSDAY- 4-6
FRIDAY 1-3.
PLEASE JUST ORDER YOUR SANDWICH IF YOU THINK YOU MIGHT WANT ONE AND DON'T
COUNT ON EXTRAS...
SEND REQUESTS TO "LUNCH" @ TURING
BYE , AND I HOPE NEXT WEEK IS VICTUALLY MORE FULFILLING THAN THIS WAS!!
-------
∂02-May-85 2104 chertok%ucbcogsci@Berkeley Stanford Linguistics Colloquium
Received: from UCB-VAX.ARPA by SU-AI.ARPA with TCP; 2 May 85 21:04:30 PDT
Received: from ucbcogsci.ARPA by UCB-VAX.ARPA (4.24/4.46)
id AA20400; Thu, 2 May 85 21:01:04 pdt
Received: by ucbcogsci.ARPA (4.24/4.38)
id AA05787; Thu, 2 May 85 21:02:14 pdt
Date: Thu, 2 May 85 21:02:14 pdt
From: chertok%ucbcogsci@Berkeley (Paula Chertok)
Message-Id: <8505030402.AA05787@ucbcogsci.ARPA>
To: cogsci-friends%ucbcogsci@Berkeley
Subject: Stanford Linguistics Colloquium
From KEMMER@SU-CSLI.ARPA Thu May 2 16:07:28 1985
Date: Thu 2 May 85 15:58:48-PDT
From: Suzanne Kemmer <KEMMER@SU-CSLI.ARPA>
To: linguists@SU-CSLI.ARPA
Cc: friends@SU-CSLI.ARPA
Resent-Date: Thu 2 May 85 16:01:25-PDT
Resent-From: Suzanne Kemmer <KEMMER@SU-CSLI.ARPA>
Resent-To: chertok@ucbcogsci
Status: R
STANFORD LINGUISTICS COLLOQUIUM
Speaker: Leonard Talmy
University of California, Berkeley
Date: Tuesday, May 7
Time: 3:15pm
Location: Bldg. 200, Rm.217
(History Corner, Quadrangle)
Reception following.
FORCE DYNAMICS IN LANGUAGE AND THOUGHT
The term 'force dynamics' describes a previously neglected semantic
domain, that of how entities interact with respect to force,
including: the exertion of force, resistance to such a force, the
overcoming of such a resistance, etc. For example, contrast the
force-dynamically neutral expression 'John doesn't go out', which
merely reports an objective observation, with 'John can't go out'.
The latter indicates a force-dynamic complex: that John WANTS to go
out (conceivable as a force-like tendency toward that act), that there
is some external force opposing that tendency, and that the latter is
stronger than the former, with a net result of no overt action.
Force dynamics is a major semantic system evident across a range
of language levels. It has direct grammatical representation, for
example in auxiliaries, conjunctions, quantifiers and the particle
system of English. Force dynamic configurations are extensively
represented in the lexicon, in lexical items referring to both
physical force interactions and also psychological and social
interactions. Finally, force dynamic principles can be seen to operate
in discourse that is involved with persuasion, governing connective
particles and, implicitly, the propositional content.
Force dynamics is a major conceptual organizing system,
constituting one of four major "imaging" systems that I have developed
which provide an integrated semantic schematization of a referent
scene. Cognitively, it is a part of "naive (socio-)psychology",
analogous to "naive physics".
-------
∂03-May-85 1215 EMMA@SU-CSLI.ARPA ling colloq
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 3 May 85 12:14:34 PDT
Mail-From: KEMMER created at 2-May-85 15:58:48
Date: Thu 2 May 85 15:58:48-PDT
From: Suzanne Kemmer <KEMMER@SU-CSLI.ARPA>
Subject: ling colloq
To: linguists@SU-CSLI.ARPA
cc: friends@SU-CSLI.ARPA
ReSent-Date: Fri 3 May 85 12:07:48-PDT
ReSent-From: Emma Pease <Emma@SU-CSLI.ARPA>
ReSent-To: friends@SU-CSLI.ARPA
STANFORD LINGUISTICS COLLOQUIUM
Speaker: Leonard Talmy
University of California, Berkeley
Date: Tuesday, May 7
Time: 3:15pm
Location: Bldg. 200, Rm.217
(History Corner, Quadrangle)
Reception following.
FORCE DYNAMICS IN LANGUAGE AND THOUGHT
The term 'force dynamics' describes a previously neglected semantic
domain, that of how entities interact with respect to force,
including: the exertion of force, resistance to such a force, the
overcoming of such a resistance, etc. For example, contrast the
force-dynamically neutral expression 'John doesn't go out', which
merely reports an objective observation, with 'John can't go out'.
The latter indicates a force-dynamic complex: that John WANTS to go
out (conceivable as a force-like tendency toward that act), that there
is some external force opposing that tendency, and that the latter is
stronger than the former, with a net result of no overt action.
Force dynamics is a major semantic system evident across a range
of language levels. It has direct grammatical representation, for
example in auxiliaries, conjunctions, quantifiers and the particle
system of English. Force dynamic configurations are extensively
represented in the lexicon, in lexical items referring to both
physical force interactions and also psychological and social
interactions. Finally, force dynamic principles can be seen to operate
in discourse that is involved with persuasion, governing connective
particles and, implicitly, the propositional content.
Force dynamics is a major conceptual organizing system,
constituting one of four major "imaging" systems that I have developed
which provide an integrated semantic schematization of a referent
scene. Cognitively, it is a part of "naive (socio-)psychology",
analogous to "naive physics".
-------
∂03-May-85 1648 avg@diablo Wumpus and NAIL
Received: from SU-AIMVAX.ARPA by SU-AI.ARPA with TCP; 3 May 85 16:47:56 PDT
Date: Fri, 3 May 85 16:30:54 pdt
From: Allen VanGelder <avg@diablo>
Subject: Wumpus and NAIL
To: nail@diablo
Wumpus and NAIL --- some rather extensive remarks --- Allen Van Gelder
I have been looking over some stuff on the game wumpus that Richard
Treitel sent me, to see if we could develop a NAIL application using it
as a base. For those of you that are unfamiliar with the game, as I
was, here is the help message's summary. I am going to assume you
have enough concept of the game to understand evaluation strategies
for it, so if you plan to read the rest of what I wrote here,
I suggest you think about how you would evaluate your move options
as you read the game description.
Script started on Fri May 3 08:22:34 1985
% lisp
Franz Lisp, Opus 38.79
-> (load 'wumpus)
[load wumpus.l]
t
-> (helpwumpus)
Your mission is to descend into the lair of the deadly
Wumpus and shoot an arrow through his heart. Each cave is
connected to 3 other caves. Fortunately for you, you can smell
the Wumpus from a cave away. Unfortunately, there are pits in
some caves and bats in others. If you fall into a pit, you die.
If you wander into a bat cave, the bats will pick you up and
deposit you elsewhere. Fortunately, you can feel drafts from pits
one cave away, and you can hear bats squeaking one cave away.
The legals commands are:
(helpwumpus) prints this message
(loadrules <f>) load mrs file <f>.
(start) starts a new game
(shoot <c>) shoots an arrow into cave <c>.
(move <c>) moves to the specified cave
(quit) quits the game.
here is the name of the current location.
score is the number of wumpi you have killed.
penalty is your penalty so far.
Penalties are assigned as follows:
1 point for each move
25 points for each arrow you shoot.
25 points each time you are killed.
25 points each time you start a new game.
nil
-> (start)
Error: eval: Undefined function unassert
<1>: (exit)
% ↑D
script done on Fri May 3 08:23:55 1985
As you can see, MRS is needed to actually play the game.
!
The first idea that comes to mind (my mind, anyway) is to let the
raw information you have been given over the course of the game
so far be considered the EDB (extensional database) and let the query be
``what is the best move?'' The NAIL program, or IDB (intentional
database) would evaluate possible moves (calling ``shoot'' a move)
and select the optimal one according to the IDB's ranking (which might
not be theoretically optimal -- the IDB might not be a skillful
wumpus player). I would be content to build and update the EDB
outside of NAIL, at least in the beginning.
Thinking about this quickly brings to light the fragility of pure
Horn clause logic. For example, suppose I HAVE a relation
rank(H, M, R), with meaning ``The rank of moving from H (present
location) to M (new location) is R (an integer).''
How do I formulate bestmove(H, M) in Horn clauses without
recourse to non-logical predicates? I can use the fact that R is
an integer, if that helps. It's reasonable to assume that I have
> and < that work fine when both arguments are bound integers, but that does
not look like enough in this case. Even assume ``3 < X'' is
somehow OK, giving tuples (3, 4), (3, 5), ..., if you want.
Moshe Vardi knows more about the
theory of this than I do, but it is my understanding that it is
impossible to chose the maximum in pure Horn clause logic without
having something more.
So, let's back off from bestmove(H, M) and just try for a program to
compute rank(H, M, R). To further simplify, let us call ``shoot''
a single choice, represented by staying where we are, and forget
about ranking different directions to shoot separately.
It is easy to limit M to legal moves:
legal(M, M).
legal(H, M) :- tunnel(H, M).
To go further, we want to know the schema of the EDB, so let's
assume it is:
Relation Meaning
-------- -------
cave(C) you know that cave C exists.
visited(C) you have visited cave C.
here(H) your present location, a cave.
tunnel(C1, C2) there is a tunnel between caves C1 and
C2 (you have visited at least one of
those caves; tunnel(C2, C1) is also
true).
stench(C) You smelled a stench when in cave C.
draft(C) You felt a draft when in cave C.
squeak(C) You heard squeaks when in cave C.
!
Now certainly we want to be able to express the fact that we did
NOT smell anything when in cave C, i.e., ``visited(C), \+stench(C)'',
as it would be expressed in Prolog. This already exceeds the
capability of pure Horn logic, but we can duck this problem by
``redesigning'' the EDB to tell us when we did not smell something,
as well:
stench(C, A) You smelled a stench when in cave C, if
A=t and did not smell one if A=f.
draft(C, A) You felt a draft when in cave C, if etc.
squeak(C, A) You heard squeaks when in cave C, if etc.
unvisited(C) You know C exists, but have not visited it.
notthere(C) You know C exists, but are not there.
nottunn(C1, C2) No KNOWN tunnel between
distinct, known caves C1 and C2.
Our new design simulates having the closed world assumption available
on the EDB. E.g., assume \+stench(X) is a relation containing every
constant in the EDB except those for which stench(X) holds.
In general, recall that \+ in Prolog is commonly read ``not'', but
really means ``not derivable''. In Prolog the goal \+stench(X)
with X a variable simply fails if there is any instance c for which
stench(c) holds. However, we shall assume (pretend?) that \+stench(X)
would actually succeed as stated above.
Then the rules
stench(C, f) :- visited(C), \+stench(C).
stench(C, t) :- stench(C).
cut out the constants we don't really want to complement over.
Similarly, ``unvisited'' and ``nottunn'' can be defined satisfactorily by
unvisited(X) :- \+visited(X), cave(X).
nottunn(X, Y) :- \+tunnel(X, Y), \+(X=Y), cave(X), cave(Y).
I have purposely put the \+'s first to emphasize that the rules
should not depend on subgoal order. Notice that I had to assume
(X=Y) was an EDB relation that in theory is equal to the relation
having a tuple (c, c) for every constant c in the EDB.
Also, recall that the predicate
``stench'' with two arguments is different from the one with one
argument, just as though they were spelled differently.
!
Now let's see if we are ready to make even a crude pass at ranking.
One reasonable approach is to estimate the probabilities that a cave
contains bats, a pit, or a wumpus, and rank moving to a cave by
using some function of those probabilities. Assume we can compute
fractions. We would like our probability estimates to be correct on
the boundaries, at least. E.g., we want pitprob(C, 0.0) if
we have visited C or if we have visited D and there was no draft in D.
Also, we want pitprob(C, 1.0) if there is a tunnel from D to C, and
we felt a draft in D, and we know pitprob(E, 0.0) for every other
cave E such that tunnel(D, E).
Now I am not ruling out the use of function symbols in the IDB, if
anyone thinks that can help. Still,
it looks to me like we have to use \+ on an IDB predicate to express
pitprob(C, 1.0); at least that is the most natural way:
pitprob(C, 0.0) :- visited(C).
pitprob(C, 0.0) :- tunnel(C, D), draft(D, f).
pitprob(C, 1.0) :- tunnel(C, D), draft(D, t), \+ pitalt(C, D).
pitalt(C, D) :- tunnel(D, E), \+(E=C), \+ pitprob(E, 0.0).
These rules work in Prolog only because it obeys my subgoal order, and
I got that order right. A more subtle point is that the apparent loop
wherein pitprob is reduced to pitalt which is reduced to pitprob, etc.
is not a real loop because of the constants appearing in the arguments.
However, real loops of this sort CAN arise if we allow \+ on IDB
predicates in our language. We are on thin ice here.
!
Still other problems arise with counting and aggregation in general.
For example, I might reasonably
want to compute a ``pit influence'' function pitinf(D, Q) with the
idea that if there is a tunnel(C, D) and a draft in D and there are
K neighbors of D that cannot be eliminated as the source of the draft,
then assume that Q = (K-1)/K is the chance that any one neighbor
is NOT the source of the draft, based on that one piece of data.
We then assume that the chance that C is not
the source of ANY known draft is the product of the influence functions
over C's known neighbors. (I'm not claiming this is really true! I'm
just supposing that we want to do a computation as though it were.)
Notice that this method gives pitprob(C, 1.0) when C is the only
possible source of the draft for some neighbor, making separate
rules for pitprob(C, 1.0) unnecessary.
Designing my rules by stepwise refinement:
pitprob(C, P) :-
unvisited(C),
neighbors(C, Ds),
pitinfall(Ds, QA),
P is (1.0 - QA).
pitinfall([], 1.0).
pitinfall([D | Ds], QA) :-
pitinf(D, Q),
pitinfall(Ds, QB),
QA is QB * Q.
pitinf(D, Q) :- neighbors(D, Es), pitcount(Es, K), Q is (K-1)/K.
pitcount([], 0).
pitcount([E | Es], K) :-
\+ pitprob(E, 0.0),
pitcount(Es, M),
K is M+1.
neighbors(C, Ds) :- setof(D, tunnel(C, D), Ds).
I needed to bail out not only with \+ applied to an IDB predicate, but
also with a non-first-order predicate ``setof''. I was forced to use
setof essentially for arithmetic on separate tuples of the same
(intermediate) relation. In one case I wanted to filter and count.
In the other I wanted to multiply.
Incidentally, I could have incorporated the filtering into ``setof'',
e.g., with ``setof(E, (tunnel(D, E), \+ pitprob(E, 0.0)), Es),''
but I wanted to encapsulate the non-first-order stuff.
For those unfamiliar with Prolog's quite powerful ``setof,''
The subgoal ``setof(T1, SG2, L3)'' makes a list L3 consisting of the
tuples of the form of T1 whose variables satisfy the subgoals SG2.
In our usage, T1 was just one variable and SG2 was one subgoal.
A careful look shows that SG2 has ANOTHER variable besides the
one corresponding to T1. If you want to understand your program,
I advise to to be sure that all variables in SG2 that are not in T1
are bound when you evaluate setof! One robust solution is to add
them to T1 even if you really don't need them there. Following this
advice myself, I could rewrite ``neighbors'' as:
neighbors(C, CDs) :- setof(tuple(C, D), tunnel(C, D), CDs).
and change the rules that use ``neighbors'' to expect a list that
looks something like [tuple(c, w), tuple(c, m), tuple(c, b)] as
an answer, where w, m, and b are c's neighbors. In this more robust
form, there is no interpretation problem if I call ``neighbors''
with both arguments unbound.
!
Coming back to my theme on the fragility of Horn clause logic,
it looks like we can't do much with wumpus move evaluation if we
confine ourselves to pure first order Horn clauses.
If we think this problem is a sample of the kind of problem we would
like to be able to program in NAIL, we need to think serously about
which extensions are important, and whether we can handle them
the way we want to, i.e., without requiring the rule-writer to
specify rule order and subgoal order.
Two extensions that came up immediately were \+ and setof.
The \+ operator has two flavors: EDB and IDB. The EDB flavor
presents fewer problems: we insist that all variables in its scope
are either bound or both local and joinable. This means that
\+ (a(X, Y), b(Y, Z))
is OK if X and Z are bound and Y does not appear elsewhere in the rule.
The system selects an information passing strategy that accomplishes
this requirement. If no strategy is adequate, the system rejects
the rule.
Using \+ on an IDB predicate that is in a ``lower'' strong component
(than the head of the rule is in)
can probably be worked out in a similar manner. The real problems
arise when it is in the same strong component. How about a
beauty like
p(X, Y) :- s(X, W), \+ p(W, Y).
My personal belief is that ``setof'' will be the foundation of the
interface between the function-free EDB and the algorithic IDB.
In fact, I think we may need to implement ``btreeof'' and/or
``rbtreeof'' in order not to limit ourselves to O(n) access time.
That's enough for now.
∂03-May-85 1754 ullman@diablo Re: Wumpus and NAIL
Received: from SU-AIMVAX.ARPA by SU-AI.ARPA with TCP; 3 May 85 17:53:54 PDT
Date: Fri, 3 May 85 17:48:06 pdt
From: Jeff Ullman <ullman@diablo>
Subject: Re: Wumpus and NAIL
To: avg@diablo, nail@diablo
My offhand impression is that Allen is right on.
I never imagined that we wouldn't have negation.
In fact the old NAIL! permits it everyplace, and defers
actually doing the complementation until it has something
to subtract (set difference) from. If so, it does difference
like ISBL, and if there is nothing to subtract from it cries foul
(and runs off the rail, I think).
Setof, i.e., the ability to non-first-normal-form-alize,
also seems essential, for exactly the reason Allen says:
it is the interface between the database and the Turing machine.
About hunting wumpuses:
Didn't Richard T. say an expert system already exists?
Presumably it solves the problems Allen raised, somehow.
Also, why don't we define pit and nopit, interms of hypothetical
DB relations draft, nodraft, visited, and tunnel, as:
nopit(X) :- visited(X).
nopit(X) :- tunnel(X, Y), nodraft(Y).
pit(X) :- tunnel(X, Y),
tunnel(Y, W),
tunnel(Y, Z),
\+(X=W), \+(X=Z), \+(W=Z), %X, W, Z are the three neighbors of Y
nopit(W), nopit(Z),
draft(Y).
I'm not sure this is the only way to infer pits.
Wumpus is, after all, a game of logic, and I *think* this is
the only way to infer pits without falling into them, which case I
omitted. If I am wrong, what other rules are useful?
---Jeff
∂03-May-85 2021 @SU-CSLI.ARPA:PCOHEN@SRI-AI.ARPA
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 3 May 85 20:20:37 PDT
Received: from SRI-AI.ARPA by SU-CSLI.ARPA with TCP; Fri 3 May 85 20:15:29-PDT
Date: Fri 3 May 85 20:18:29-PDT
From: Phil Cohen <PCOHEN@SRI-AI.ARPA>
To: f4@SU-CSLI.ARPA, finterest@SU-CSLI.ARPA,
NL4: ;
Reminder: Bob Moore will speak Tues, the 7th, from 12:45-2:15 in the
Ventura Seminar room. Abstract follows:
Phil
---------------
A POSSIBLE-WORLD ANALYIS OF DESIRE
Belief-Desire-Intention (BDI) models of rational agents are
used in many otherwise disparate research projects within CSLI. While
formal analyses of belief abound, analyses of intention are much less
developed, and analyses of desire are all but non-existent. This talk
attempts to help remedy the situation by presenting a formal analysis
of desire in terms of preferences among possible worlds. The analysis
gives an account of desire that allows for such things as
* conflicting desires
* not desiring all entailments of a desire
* not desiring all possible means to a desired end
-------
∂03-May-85 2021 @SU-CSLI.ARPA:PCOHEN@SRI-AI.ARPA
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 3 May 85 20:20:37 PDT
Received: from SRI-AI.ARPA by SU-CSLI.ARPA with TCP; Fri 3 May 85 20:15:29-PDT
Date: Fri 3 May 85 20:18:29-PDT
From: Phil Cohen <PCOHEN@SRI-AI.ARPA>
To: f4@SU-CSLI.ARPA, finterest@SU-CSLI.ARPA,
NL4: ;
Reminder: Bob Moore will speak Tues, the 7th, from 12:45-2:15 in the
Ventura Seminar room. Abstract follows:
Phil
---------------
A POSSIBLE-WORLD ANALYIS OF DESIRE
Belief-Desire-Intention (BDI) models of rational agents are
used in many otherwise disparate research projects within CSLI. While
formal analyses of belief abound, analyses of intention are much less
developed, and analyses of desire are all but non-existent. This talk
attempts to help remedy the situation by presenting a formal analysis
of desire in terms of preferences among possible worlds. The analysis
gives an account of desire that allows for such things as
* conflicting desires
* not desiring all entailments of a desire
* not desiring all possible means to a desired end
-------
∂03-May-85 2021 @SU-CSLI.ARPA:PCOHEN@SRI-AI.ARPA
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 3 May 85 20:20:37 PDT
Received: from SRI-AI.ARPA by SU-CSLI.ARPA with TCP; Fri 3 May 85 20:15:29-PDT
Date: Fri 3 May 85 20:18:29-PDT
From: Phil Cohen <PCOHEN@SRI-AI.ARPA>
To: f4@SU-CSLI.ARPA, finterest@SU-CSLI.ARPA,
NL4: ;
Reminder: Bob Moore will speak Tues, the 7th, from 12:45-2:15 in the
Ventura Seminar room. Abstract follows:
Phil
---------------
A POSSIBLE-WORLD ANALYIS OF DESIRE
Belief-Desire-Intention (BDI) models of rational agents are
used in many otherwise disparate research projects within CSLI. While
formal analyses of belief abound, analyses of intention are much less
developed, and analyses of desire are all but non-existent. This talk
attempts to help remedy the situation by presenting a formal analysis
of desire in terms of preferences among possible worlds. The analysis
gives an account of desire that allows for such things as
* conflicting desires
* not desiring all entailments of a desire
* not desiring all possible means to a desired end
-------
∂03-May-85 2021 TREITEL@SUMEX-AIM.ARPA Re: Wumpus and NAIL
Received: from SU-AIMVAX.ARPA by SU-AI.ARPA with TCP; 3 May 85 20:21:12 PDT
Received: from SUMEX-AIM.ARPA by diablo with TCP; Fri, 3 May 85 20:17:30 pdt
Date: Fri 3 May 85 20:15:28-PDT
From: Richard Treitel <TREITEL@SUMEX-AIM.ARPA>
Subject: Re: Wumpus and NAIL
To: nail@SU-AIMVAX.ARPA, treitel@SUMEX-AIM.ARPA
In-Reply-To: Message from "Jeff Ullman <ullman@diablo>" of Fri 3 May 85 17:56:27-PDT
There is indeed an MRS program (call it an "expert system" if you want) for
playing Wumpus. I think I sent Allen a copy, but presumably he wanted to work
on the problem from scratch and perhaps see how many of the dirty tricks used
were really necessary, or something. For anybody who is anxious to have sight
of the rule file, it's in [SUMEX]<MRS.MAC.NEW>WUMPUS.MRS and is unprotected so
that you can get at it by anonymous FTP. If there is demand, I can discuss it
when I talk on MRS at the (now + Nth) NAIL! meeting, for some value of N.
- Richard
Oh and yes, MRS does include negation-by-failure and Setof (or Bagof).
-------
∂04-May-85 0219 @SU-SCORE.ARPA:JF@SU-SUSHI.ARPA One more BATS abstract
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 4 May 85 02:19:32 PDT
Received: from SU-SUSHI.ARPA by SU-SCORE.ARPA with TCP; Sat 4 May 85 00:56:31-PDT
Date: Fri 3 May 85 10:03:29-PDT
From: Joan Feigenbaum <JF@SU-SUSHI.ARPA>
Subject: One more BATS abstract
To: aflb.su@SU-SCORE.ARPA
I've received another abstract for the May 17 BATS. It's in
SUSHI:<JF>brassard.abstract
If you want to take a look at this before deciding whether to come, please
do. If you want me to mail you a copy, let me know.
Joan
-------
∂04-May-85 1004 @SU-SCORE.ARPA:udi@wisc-rsch.arpa crusade (long message)
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 4 May 85 10:04:06 PDT
Received: from wisc-rsch.arpa by SU-SCORE.ARPA with TCP; Sat 4 May 85 10:01:41-PDT
Received: by wisc-rsch.arpa; Thu, 2 May 85 21:56:10 cdt
Received: from wisc-crys.arpa by wisc-rsch.arpa; Thu, 2 May 85 06:27:45 cdt
Received: from UCB-VAX.ARPA by wisc-crys.arpa; Thu, 2 May 85 06:27:32 cdt
Received: from ucbernie.ARPA by UCB-VAX.ARPA (4.24/4.46)
id AA01418; Thu, 2 May 85 04:25:05 pdt
Received: by ucbernie.ARPA (4.24/4.46)
id AA07323; Thu, 2 May 85 04:03:58 pdt
Date: Thu, 2 May 85 04:03:58 pdt
From: brassard%ucbernie@Berkeley
Message-Id: <8505021103.AA07323@ucbernie.ARPA>
To: theory@uwisc
Subject: crusade (long message)
Cc: wolinsky%ucbernie@Berkeley
Sender: udi@wisc-rsch.arpa
Resent-From: udi@wisc-rsch.arpa (Udi Manber)
Resent-To: theory:;@wisc-rsch.arpa
Dear friends and colleagues,
I intend to send the following letter to SIGACT/NEWS shortly.
Any feedback/comments would be greatly appreciated.
I will be at STOC next week, or you can send me mail at
brassard@ucbernie (on ARPAnet).
Thank you for your cooperation,
Gilles.
P.S. I apologize to those who will receive this long message twice,
or who received the first draught of this letter a few weeks ago.
←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←
!
CRUSADE FOR A BETTER NOTATION (second draught)
by Gilles Brassard
Computer Science Division
University of California
Berkeley, CA 94720
In his well-known SIGACT/NEWS paper of 1976, Knuth sets forth
the asymptotic notation by which we all now live. He closes his
discussion with: "I propose that members of SIGACT (...) adopt the
big Oh, big Omega and big Theta notations as defined above, unless
a better alternative can be found reasonably soon." Although one
can hardly consider nearly a full decade later as being "reasonably
soon", I would like to share with you my thoughts on the subject.
A few years ago, it occurred to me that O(f(n)) should really
be defined as the SET of all functions eventually upper-bounded by
a constant multiple of f(n) . As I later realized, this was hardly
new; it is in fact precisely what Knuth says in his paper, giving
credit to Rivest for the original idea. Knuth then goes on to say:
"The phenomenon of one-way equalities arises in this connection (...)
The equal sign really means set inclusion (the current author would
add: 'or set membership, in most cases'). My (Knuth's) feeling is
that we should continue to use one-way equalities together with
O-notations, since it has been common practice of thousands of
mathematicians for so many years now, and since we understand the
meaning of our existing notation sufficiently well."
I strongly disagree with the above statement. I do not believe
that a bad and confusing notation should be encouraged when a better
and more natural alternative is available. To give an analogy outside
the realm of Computer Science: should not the United States switch to
the metric system, even though the British system is common practice of
hundreds of millions of Americans and does not perform all that badly?
What is the alternative? Once the asymptotic notation is defined
in terms of sets, it is only natural to manipulate it using set
notation. If "the equal sign really means set inclusion", why not
use set inclusion? I wish to emphasise that this reaches far beyond
a purist's aesthetic preferences. Indeed, it allows one to make
full use of the power of set theory. It allows any number of big Oh's
to be used in complicated set theoretic formulae, with no fear of
confusion or ambiguous meaning. It even allows free mixing of big
Oh's with big Omegas, big Thetas, and any set theoretic and arithmetic
symbols appropriate to convey one's intended meaning. Why refuse this
powerful tool? As a trivial example, one can write:
"Theta(f(n)) = O(f(n)) intersection Omega(f(n))". To be fair, I must
admit that no ambiguity can arise from the usual asymptotic notation
if, when reading, one systematically substitutes set inclusions
(or set memberships, whichever makes sense) for equal signs. Still,
there remains a difficulty with this conventional approach: how does
one actually represent set equality and proper set inclusion?
!
Besides keeping with tradition, the only thing I can say in favour
of using the equal sign to mean set inclusion and set membership is
that it is sometimes easier to typeset. This is particularly true for
electronic mail. Perhaps this was sufficient reason to submit the
equal sign to such bizarre treatment in the late nineteenth century,
when the big Oh notation was invented. However, modern typesetting
facilities result in this consideration being mostly obsolete.
I therefore propose that members of SIGACT adopt the following
definitions in their set theoretic sense, and use them as such, in
accordance with set theory:
O(t(n)) = {f : N -> R* | f(n) <= ct(n) for all n > n0, for some
integer n0 and some positive real constant c};
Omega(t(n)) = {f : N -> R* | f(n) >= ct(n) for all n > n0, for some
integer n0 and some positive real constant c};
Theta(t(n)) = intersection of O(t(n)) and Omega(t(n)),
where N denotes the set of non-negative integers and R* denotes the
set of non-negative reals (I would appreciate any suggestions for a
better symbol to denote the non-negative reals). For some applications,
it is useful to extend this notation to functions taking negative
values or even undefined on some or all of their domain up to n0 .
For instance, one can freely use O(n/log n) without worrying about
the cases n=0 and n=1. It can also be useful to extend this notation
in the natural way to functions of more than one argument.
Arithmetic operations on asymptotic notation can be performed
in a systematic fashion by extending the usual operators to sets of
functions in the obvious way: for instance, X + Y denotes
{f+g | f is in X and g is in Y }, where function addition is
pointwise. Similarly, if f(n) is a function and X is a set of
functions, f(n) + X can be used as an abbreviation for {f(n)} + X .
In order to avoid potential confusion between arithmetic pointwise
subtraction and set difference, one should use the symbol "\" for the
latter (and of course never "+" to mean "union"). Confusion between
pointwise multiplication and Cartesian product can generally be avoided
by the context. Nesting of asymptotic notation can also be given a
systematic definition by using O(X) , where X is a set of functions,
to denote the union of the O(t(n))'s for each t(n) in X.
Perhaps the thought of saying that the time taken by an algorithm
BELONGS TO O(...), or is IN O(...), instead of the usual IS O(...),
seems strange at first. I maintain however that this is a very small
price to pay for the possibility of using the full power of set theory
and its notation. Moreover, people who have adopted my ideas over the
years reported that it rapidly gets to sound quite natural. I have
successfully used the set notation in undergraduate and graduate
algorithm classes, both in Montreal and Berkeley. Most students appear
to grasp the basic ideas with less difficulty than when I previously
used the conventional notation. For one thing, they never have to ask
the eternal question: "but since n = O(n log n), why can't you write
that O(n log n) = n ?". An extensive treatment of this notation can
be found in my forthcoming undergraduate textbook on Algorithmics
(not to be confused with a book on algorithms). This book is
co-authored by my colleague Paul Bratley at the Universite de Montreal.
!
I would like to make a small case for another unconventional aspect
of my definition of O(t(n)) . This notation is usually defined as
the set of functions from the naturals to the reals (not only the
positive reals) such that their absolute value is upper-bounded by some
positive constant multiple of t(n) . My reasons for disqualifying
negative valued functions in this context are twofold. First, it is
necessary to do so in order to obtain the following simple, yet very
useful identity: O(f(n)+g(n)) = O(max(f(n),g(n))) . Second, it
encourages (but does not force) one to be more precise in those rare
circumstances in the analysis of algorithms where negative valued
functions are appropriate. Given that time and space are never
negative, this usually occurs only for lower order terms. As an
example, the worst-case number of comparisons performed by heapsort
is of course in O(n log n), but more precisely in 2n log n - O(n).
Here, explicit use of the minus sign allows to state that not only
the hidden constant in O(n log n) is 2 , but also that 2n log n
is a slight OVERestimate. Should lower order terms oscillate around
zero, or should one not care to be precise, it is always possible to
make good use of "+/-" .
I would also like to comment on my definition of Omega. Some
authors have proposed that this should be an "infinitely often" notion
(thus almost going back to Hardy and Littlewood's original definition).
I consider that there are good arguments in favour as well as against this
idea. My main reason for avoiding the "infinitely often" can of worms
is that it is extremely messy to manipulate mathematically. The tip
of the iceberg is that this makes the notion non-transitive! That is,
even should f(n) be in Omega(g(n)) and g(n) be in Omega(h(n)),
it would not necessarily follow that f(n) is in Omega(h(n)).
CONDITIONAL ASYMPTOTIC NOTATION
Conditional asymptotic notation is a tool that allows substantial
simplifications in the rigorous resolution of many recurrence equations
of the divide-and-conquer variety. This idea can be used regardless of
how you feel about using set theoretic notation. as an example,
consider the following function, arising in the analysis of mergesort:
t(n) = t(floor(n/2)) + t(ceiling(n/2)) + n for all n > 1 ,
and t(0) = 0, t(1) = 1. Solving this equation exactly for all n
is not particularly easy, but it is immediate to find that
t(n) = n (1 + log n), as long as n is an exact power of two.
The conditional asymptotic notation consists in this case of saying
that t(n) is in Theta(n log n | n is a power of two), read: "Theta of
n log n WHEN n is a power of two". Of course, this can be extended
to any predicate P(n), as well as to big Oh and Omega. Now, if one
is really interested in the order of t(n) for every integer n,
the following theorem gives conditional asymptotic notation's
main raison d'etre:
"If f(n) is in Theta(g(n) | n is a power of b) for some
integer b > 1 , if f(n) and g(n) are eventually non-decreasing,
and if g(n) is b-smooth in the sense that g(bn) is in O(g(n)),
then f(n) is in Theta(g(n))."
In our example, it is easy to show that "n log n" is non-decreasing
and 2-smooth, and that t(n), as defined by the recurrence equation,
is non-decreasing. We may therefore conclude that t(n) is in
Theta(n log n).
!
HISTORICAL REMARKS
In the first edition of his Volume 1, Knuth does not mention that
asymptotic notation really defines sets of functions. This can also be
said of Horowitz and Sahni's book on algorithms. Both these sources
freely use one-way equalities. In the second edition of his Volume 1 ,
Knuth replaced 11 lines on page 105 with a different 9 lines, mentioning
the fact that one may regard asymptotic notation as defining sets.
The notation is directly defined as sets in Purdom and Brown's new book
on the analysis of algorithms. Again, however, both these sources
make use of one-way equalities. In her book, Baase defines the big-Oh
notation in the traditional way, but she defines Theta as a set;
she then says: "it is conventional to write 'g = Theta(f)' or
'g is Theta(f)' rather than 'g belongs to Theta(f)' ",
thus coming closer to this proposal than any other source I could
find.
Other authors avoid the one-way equality problem by using
asymptotic notation in prose only, writing things like "the algorithm
takes O(n) time" or "f(n) is O(n)". This can be said in particular
of Aho, Hopcroft and Ullman's book on the design and analysis of
algorithms and of Tarjan's monograph. Finally, Sedgewick avoids the
problem entirely by never mentioning asymptotic notation.
ACKNOWLEDGEMENTS
I would like to thank Manuel Blum and Sara Baase for their
enthusiasm about the use of set theory in this context. Their moral
support helped me write up this proposal. A preliminary version of
this letter was circulated a few weeks ago within a limited circle.
Very useful suggestions for improvements were given by Sara Baase,
Manuel Blum, Russel Brand, Joan Feigenbaum, Richard Karp, Oren
Patashnik and Jeff Wolinsky.
∂04-May-85 1023 MMACKEN@SU-CSLI.ARPA grammars & learning
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 4 May 85 10:23:42 PDT
Date: Sat 4 May 85 10:19:15-PDT
From: Marcy Macken <MMACKEN@SU-CSLI.ARPA>
Subject: grammars & learning
To: folks@SU-CSLI.ARPA
A few people will be getting together this Thursday to continue
the discussion started by Janet Fodor's colloq. General topic:
the relation between the linguist's theory and the learning of
language. May 9th, 10 a.m., Ventura. All interested are welcome.
-------
∂05-May-85 1104 BARWISE@SU-CSLI.ARPA Messages to Cooper v Engdahl
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 5 May 85 11:04:20 PDT
Date: Sun 5 May 85 11:00:24-PDT
From: Jon Barwise <BARWISE@SU-CSLI.ARPA>
Subject: Messages to Cooper v Engdahl
To: Researchers@SU-CSLI.ARPA
I am off to Sweden to a workshop on Tuesday. I just spoke to R and E
and asked them if there was anything they wanted from Palo Alto. They
said they are hungry for ideas from and news of people here, as well
as sun. If there is any message you would like me to take, let me
know.
Jon
-------
∂06-May-85 0037 RESTIVO@SU-SCORE.ARPA PROLOG Digest V3 #21
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 6 May 85 00:37:05 PDT
Date: Sunday, May 5, 1985 6:01PM
From: Chuck Restivo (The Moderator) <PROLOG-REQUEST@SU-SCORE.ARPA>
Reply-to: PROLOG@SU-SCORE.ARPA
US-Mail: P.O. Box 4584, Stanford CA 94305
Phone: (415) 326-5550
Subject: PROLOG Digest V3 #21
To: PROLOG@SU-SCORE.ARPA
PROLOG Digest Monday, 6 May 1985 Volume 3 : Issue 21
Today's Topics:
Announcement - Workshop,
Implementation - Control & CP,
LP Library - Manna Book
----------------------------------------------------------------------
Date: Wed, 1 May 85 19:01:22 -0200
From: Ehud Shapiro <Udi%Wisdom.bitnet@WISCVM.ARPA>
Subject: Workshop
As a welcome party for the summer and for our summer
students, guests, and visitors, we will hold on Tuesday
and Wednesday, June 28 and 29, and informal workshop on
"Logic Programming and Parallel Processing". If you
plan to be in Israel during that time, and would like
to participate and/or contribute, please let me know.
-- Ehud Shapiro
------------------------------
Date: 2 May 1985 0843-CDT
From: THRIFT%TI-CSL.CSNet@csnet-relay.arpa
Subject: Control annotation in CP
In reference to a previous note on control annotation
for CP:
I have been made aware that the 'mode annotation' for
procedure heads (input and output restrictions) was
first made in IC-Prolog. The annotation here is '?' for
input and '↑' for output.
For example (Clark,K.L. "Predicate Logic as a
Computational Formalism")
append([X1|X]?,Y?,[Z1|Z]↑) :- append(X,Y,Z).
restricts the use of this clause to a case where the
first two arguments are input and the third is output.
My apologies for missing this.
Although I realize that the control achieved through the
read-only annotation is greater than that achieved by
the mode annotation, I would like to see a practical
(that is, not 'artificial') example showing this.
------------------------------
Date: 03 May 85 1425 PDT
From: Yoni Malachi <YM@SU-AI.ARPA>
Subject: More details on "The Logical Basis for
THE LOGICAL BASIS FOR COMPUTER PROGRAMMING
Volume 1: Deductive Reasoning
ZOHAR MANNA and RICHARD WALDINGER
Addison-Wesley, ISBN: 0-201-18620-2
This exceptionally clear text, laced with many examples,
provides a most readable introduction to the logical
concepts and techniques underlying computer programming.
Computational notions are explored in a logical realm
independent of any programming language or any machine.
The text is accessible to readers with no background in
mathematics or computer programming, yet it supplies
axiomatization for a rich collection of abstract data
types.
This book provides the intellectual tools for studying
artificial intelligence, software engineering, automatic
programming, database theory, logic programming, and the
theory of computation.
A forthcoming second volume, DEDUCTIVE SYSTEMS, describes
logical techniques for automated theorem proving and its
applications.
The Authors:
ZOHAR MANNA is Professor of Computer Science at Stanford
University; he is the author of the textbook "Mathematical
Theory of Computation."
RICHARD WALDINGER is Staff Scientist in the Artificial
Intelligence Center at SRI International.
TABLE OF CONTENTS:
Part I. Mathematical Logic
Chapter 1: Propositional Logic
Introduction. The Language. The Meaning of a Sentence.
Properties of Sentences. Truth Tables. Semantic Trees.
Proof by Falsification. Valid Sentence Schemata.
Substitution. Extended Interpretation. Equivalence.
Problems.
Chapter 2: Predicate Logic: Basic
Introduction. The Language. The Meaning of a Sentence.
Semantic Rules. Validity. Universal and Existential
Closure. Problems.
Chapter 3: Predicate Logic: Advanced
Valid Sentence Schemata. Equivalence. Safe Substitution.
The Value Property. Valid Schemata with Substitution.
Function Introduction and Elimination.
Problems.
Chapter 4: Special Theories
Definition of a Theory. Augmenting Theories. Relationship
between Theories. Theory of Strict Partial Orderings.
Theory of Equivalence Relations. Problems.
Chapter 5: Theories with Equality
Theory of Equality. Theory of Weak Partial Orderings.
Theory of Associated Relations. Theory of Groups. Theory
of Pairs. Relativized Quantifiers. The Lexicographic
Relation. Problems.
Part II. Theories with Induction
Chapter 6: Nonnegative Integers
Basic Properties. The Addition Function. Multiplication
and Exponentiation. Predecessor and Subtraction.
Decomposition Induction. The Weak Less-than Relation.
The Strict Less-than Relation. Complete Induction.
Quotient and Remainder. Proof of Complete Induction. The
Divides Relation. The Least-Number Principle. Problems.
Chapter 7: Strings
Basic Properties. The Head and Tail Functions. The
Concatenation Function. The Reverse Function. The
Decomposition Induction Principle. The Substring Relation.
The Complete Induction Principle. Nonnegative Integers
and Strings. String Representation of Integers. Problems.
Chapter 8: Trees
Basic Properties. The Left and Right Functions. The
Subtree Relation. Strings and Trees. Problems.
Chapter 9: Lists
Basic Properties. The Head and Tail Functions. Append
and Member. Example: Flatlist. Tree Representation of
Lists. Example: Parsing. Problems.
Chapter 10: Sets
Basic Properties. The Equality Proposition. The Choice
and Rest Functions. The Union and Intersection
Functions. The Deletion and Difference Functions.
The Subset Relation. The Set Constructor. Cardinality.
Singleton Sets. Problems.
Chapter 11: Bags
Basic Properties. The Equal-Multiplicity Relation.
Multiplicity and Equality. The Count Function.
Additional Functions and Relations. Sum, Union,
and Intersection. Problems.
Chapter 12: Tuples
Basic Properties. Nonnegative Integers and Tuples.
Mapping Tuples into Sets and Bags. The Permutation
Relation. The Ordered Relation. The Sort Function.
Recursive Definition of Functions. Problems.
Related Textbooks
Index of Symbols
General Index
------------------------------
End of PROLOG Digest
********************
∂06-May-85 0854 LIBRARY@SU-SCORE.ARPA The Way of Mathematics and Mathematicians; a Historical-Philosophical Study
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 6 May 85 08:54:31 PDT
Date: Mon 6 May 85 08:53:41-PDT
From: C.S./Math Library <LIBRARY@SU-SCORE.ARPA>
Subject: The Way of Mathematics and Mathematicians; a Historical-Philosophical Study
To: su-bboards@SU-SCORE.ARPA
cc: faculty@SU-SCORE.ARPA
Volume 17, 1984, fo the Communications of the Mathematical Institute,
Rijksuniversiteit Utrecht is a 47 page publication by A.F.Monna titled
The Way Of Mathematics And Mathematicians. The contents reads as
follows: Introduction--the aims of this study, subjectivity and
objectivity; The way of mathematics--the evolution of mathematics, a
natural development?, basic discontinuities, case-studies; Mathematics,
a chain of phases; What is mathematics?, historical remarks, objects
and aims, free mental creation; Research Programmes; Notes.
This publication is available in the Math/CS Library under call number
QA1.C683 v. 17
H. Llull
-------
∂06-May-85 0905 LIBRARY@SU-SCORE.ARPA New Books in the Math/CS Library
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 6 May 85 09:04:59 PDT
Date: Mon 6 May 85 09:04:11-PDT
From: C.S./Math Library <LIBRARY@SU-SCORE.ARPA>
Subject: New Books in the Math/CS Library
To: su-bboards@SU-SCORE.ARPA
cc: faculty@SU-SCORE.ARPA
Dataflow Computation by A.P.W. Bohm. QA76.9A73B63 1984
Eurographics Tutorials '83. edited by Paul J. W. ten Hagen T385.E97 1983 Suppl.
Papers On Computational and Cognitive Science: Proceedings of the Conference
on Linguistics in the Humanities and Sciences. edited by E. Battistella
(84-04674 c.2)
Data Compression;Techniques and Applications. by Lynch.QA76.9.D33L96 1985
Graphical Methods for Data Analysis. by Chambers, Cleveland, Kleiner, and
Tukey. QA276.3.G73 1983
COMPSTAT; Solving Statistical Problems by Microcomputer. by Presby
QA276.4.P74 1984
Microprocessors and Intelligence. Proceedings of an ASLIB seminar
May 1979. edited by Anthony. Q334.M5 1979
Microcomputer Interfacing by Stone. TK7868.I58.S76 1983
8086 8088 Assembly Language Programming. by Yeung. QA76.8.I292Y48 1984
How To Copyright Software. by Salone and Elias. KF3024.C6S17 1984
H.Llull
-------
∂06-May-85 0923 GOLUB@SU-SCORE.ARPA Praise for Andy Yao
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 6 May 85 09:20:35 PDT
Date: Mon 6 May 85 09:16:33-PDT
From: Gene Golub <GOLUB@SU-SCORE.ARPA>
Subject: Praise for Andy Yao
To: faculty@SU-SCORE.ARPA
It's nice to see Andy Yao so highly praised in the April 26 issue of
SCIENCE. Congratulations, Andy.
Gene
-------
∂06-May-85 0941 @SU-SCORE.ARPA:JJW@SU-AI.ARPA Terminals available
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 6 May 85 09:41:05 PDT
Received: from SU-AI.ARPA by SU-SCORE.ARPA with TCP; Mon 6 May 85 09:31:31-PDT
Date: 06 May 85 0930 PDT
From: Joe Weening <JJW@SU-AI.ARPA>
Subject: Terminals available
To: CSD@SU-SCORE.ARPA
Reply-To: Tom@SU-SCORE.ARPA
CSD-CF has a batch of new Z29A terminals available for installation in
MJH offices. If you would like one put into your office, send a message
to Tom Dienstbier (Tom@Score).
∂06-May-85 0952 CHEADLE@SU-SCORE.ARPA Ph.D. admissions results
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 6 May 85 09:52:03 PDT
Date: Mon 6 May 85 09:36:03-PDT
From: Victoria Cheadle <CHEADLE@SU-SCORE.ARPA>
Subject: Ph.D. admissions results
To: faculty@SU-SCORE.ARPA, grosz@SRI-AI.ARPA, rosenschein@SRI-AI.ARPA,
berglund@SU-PESCADERO.ARPA, russell@SUMEX-AIM.ARPA
Office: Margaret Jacks 258, 497-1519
Here is the final tally of the Ph.D. admits for Autumn 1985-86.
There were 29 accepts (two of which deferred for 1 year, Max
Hailperin and David Jones), and 12 who declined our offer.
Of the 12 who declined, we know that 5 chose CMU, 2 MIT, and
1 U. of Essex; the remaining 3 we have no information
regarding their final choice.
LAST NAME FIRST NAME INTERESTS UNDERGRAD SCHOOL STATUS VISITED FELLOWSHIP DECISION
Christensen Jens AI ROB Columbia University ACCEPT N
Enderton Eric B. MTC CG UC Berkeley REJECT N NSF UCBerkeley
Enderton Herbert D. AI ROB UC Berkeley REJECT Y NSF CMU
Fernando Timothy P. MTC AI California Institute of Technology REJECT Y CMU
Frants Leonid AI MTC University of Auckland ACCEPT N
Friedman Joseph OS MTC Ben-Gurion University of the Negev ACCEPT N
Glasser Nathaniel E. PSL OS Yale University REJECT N NSF MIT
Goldman Sally A. DB AA Brown University REJECT N NSF
Hailperin Max AI DCS MIT DEFER Y
Healey Glenn E. University of Michigan ACCEPT CSMS Student NSF
Hirsh Haym B. AI CM UCLA ACCEPT MSAI Student
Jacobs Joseph D. AI PSL Northwest Missouri State University ACCEPT Y GE
Jones David G. AI PSL University of Western Ontario DEFER N
Kapur Shyam MTC AA Indian Institute Of Technology Kanpur REJECT N
Kipnis Shlomo AA MTC Hebrew University REJECT N
Kosoresow Andrew P. NA AI Columbia College ACCEPT Y
Lazanas Anthony AI APP National Tech. Univ. Of Athens ACCEPT N
Madhav Neel PSL OS Indian Institute ofTechnology ACCEPT N
Maggs Bruce M. AA MTC MIT REJECT N NSF MIT
Merchant Arif A. AI Indian Institute of Technology ACCEPT MSAI Student
Myers Karen L. NDS AI University of Toronto ACCEPT Y
Nayak Pangal P. VLSI DCS India Institute of Technology ACCEPT N
Niziot Wiestawa K. Warsaw University ACCEPT N
Nowlan Steven J. AI ROB University of Waterloo REJECT Y CMU
Paek Eunok NDS AI Seoul National University ACCEPT N
Pieper Karen L. PSL NDS Rice University ACCEPT Y
Pierce Benjamin C. PSL APP Stanford University REJECT Stanford UG NSF CMU
Plambeck Thane E. MTC CM University of Nebraska ACCEPT N
Plaxton Charles, G AA DCS University of Toronto ACCEPT N
Roberts Paul M. AI MTC St. John's College, Oxford University ACCEPT Y
Rokicki Tomas G. AI Texas A&M University ACCEPT N NSF, Hertz
Saraiya Yatin P. AA CG Swarthmore College ACCEPT Y
Seligman Scott M. PSL CG Princeton University ACCEPT Y NSF, IBM
Singer Stephanie F. AI Yale University ACCEPT Y
Snoeyink Jack S. MTC AA Calvin College ACCEPT Y NSF
Strat Thomas M. AI ROB MIT ACCEPT N
Stubblefield Phillip L. AI MTC University of Puget Sound ACCEPT N
Traugott Jonathan C. AI MTC Stanford University ACCEPT Stanford UG
Williamson Carey L. OS NDS University of Saskatchewan ACCEPT Y
Wobcke Wayne R. AI MTC University of Queenland REJECT N Univerisity of Essex
Yost Gregg R. AI PSL Carnegie-Mellon University REJECT Y NSF CMU
Please feel free to drop by my office to look
at any of their folders.
Victoria
-------
∂06-May-85 1004 ISRAEL@SU-CSLI.ARPA Tuesday mtg.
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 6 May 85 10:04:28 PDT
Date: Mon 6 May 85 09:58:10-PDT
From: David Israel <ISRAEL@SU-CSLI.ARPA>
Subject: Tuesday mtg.
To: rrr@SU-CSLI.ARPA
David Israel will wax poetical on (of all things) Turing Machines.
In the Ventura seminar room, 2:15-4:15
-------
∂06-May-85 1034 @SU-SCORE.ARPA:MDIXON@SU-SUSHI.ARPA Spring Picnic: May 31
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 6 May 85 10:34:30 PDT
Received: from SU-SUSHI.ARPA by SU-SCORE.ARPA with TCP; Mon 6 May 85 10:22:22-PDT
Date: Mon 6 May 85 10:21:28-PDT
From: Mike Dixon <MDIXON@SU-SUSHI.ARPA>
Subject: Spring Picnic: May 31
To: csd@SU-SCORE.ARPA, su-bboards@SU-SUSHI.ARPA
By a relatively overwhelming majority (24 to 10) may 31 has been chosen as the
date of the CSD picnic. Apologies to all those who will be unable to come as a
result.
We need volunteers to buy (with $$ from the department), transport, prepare,
and cook food. Also, people who can find & bring various forms of
entertainment (softball? soccer? croquet?) and, of course, people to clean
up. Volunteer early and volunteer often!
.mike.
-------
∂06-May-85 1100 avg@diablo Re: Wumpus and NAIL
Received: from SU-AIMVAX.ARPA by SU-AI.ARPA with TCP; 6 May 85 11:00:52 PDT
Date: Mon, 6 May 85 10:57:28 pdt
From: Allen VanGelder <avg@diablo>
Subject: Re: Wumpus and NAIL
To: nail@diablo
Jeff's rule for pit(X) works fine if you are willing to limit the rules
to systems of caves with degree three. However, knowledge bases are
probably not so regular, so I wanted to look at the issues raised by
we can enumerate all states of pit-knowledge of X's neighbors and
write an appropriate pitprob program, but this is not the way to
develop expert systems.
∂06-May-85 1103 avg@diablo Re: Wumpus files
Received: from SU-AIMVAX.ARPA by SU-AI.ARPA with TCP; 6 May 85 11:02:41 PDT
Date: Mon, 6 May 85 10:59:57 pdt
From: Allen VanGelder <avg@diablo>
Subject: Re: Wumpus files
To: nail@diablo
For those on diablo, all wumpus files are in ~avg/nail/wumpus*
I forgot to mention this in my previous ramblings.
∂06-May-85 1210 NILSSON@SU-SCORE.ARPA Tuesday Lunch
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 6 May 85 12:09:52 PDT
Date: Mon 6 May 85 12:06:58-PDT
From: Nils Nilsson <NILSSON@SU-SCORE.ARPA>
Subject: Tuesday Lunch
To: faculty@SU-SCORE.ARPA
Tomorrow, May 7, we have our regularly scheduled faculty informal
lunch in MJH 146. Since such an important part of our department is
our students, I have invited the student representatives who normally
attend our lunch to play a special role tomorrow and lead a discussion
about student concerns. The informal setting will allow us to talk
about what's on their minds in a helpful way, and I hope several
folks join us tomorrow. -Nils
-------
∂06-May-85 1458 avg@diablo Same Generation and NC
Received: from SU-AIMVAX.ARPA by SU-AI.ARPA with TCP; 6 May 85 14:57:17 PDT
Date: Mon, 6 May 85 14:45:33 pdt
From: Allen VanGelder <avg@diablo>
Subject: Same Generation and NC
To: nail@diablo
An NC algorithm for the same generation problem:
Given a population of n individuals, named 1..n wlog (ho ho),
and an ACYCLIC parent relation over them, p(X, Y),
find those pairs in the same generation.
Let an(X, Y, K) be an n↑3 bitmap to record the fact that X has
ancestor Y at K generations above. Initialize it for K = 0 and 1
in constant time, using ``only'' n↑3 processors.
Now use ``only'' n↑5 processors to update the an bitmap by the rule:
an(X, Z, I+J) :- an(X, Y, I), an(Y, Z, J).
Assign a separate processor for each (X, Y, Z, I, J) combination.
Clearly all ancestor chains of length K or less are found in log K
steps of this type.
Another set of n↑4 processors reads out sg(X, Y), the same generation
relation, via the rule:
sg(X, Y) :- an(X, Z, K), an(Y, Z, K).
Assign a processor for each (X, Y, Z, K) combination.
REMARKS AND QUESTIONS:
1. The analogy to method 3 of the May 1 nail meeting is clear.
Both the number of generations apart and the acyclicity enter.
2. What happens if the acyclicity restriction is removed?
If p1 and p2 are two primes below n, and p has simple cycles
of those lengths that intercept, the brute force method
described above would need to look at chains of length p1 * p2.
A more sophisticated algorithm would observe that gcd(p1, p2) = 1
and put everyone in both cycles in the same generation. Still this
does not solve the whole problem.
However, a series of intersecting cycles forces the brute force
method out of NC.
3. What happens if there are more than one base rule and one
recursive rule?
4. Can this be generalized to Henschen/Naqvi's class of rules
of the form p :- a, p, b. with or without restrictions on b
and a?
In other words, how close can we come to the statement,
``Linear recursion without function symbols and with bounded
numbers of variables per rule is in NC?''
∂06-May-85 1540 FISCHER@SU-CSLI.ARPA Chronicles
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 6 May 85 15:40:13 PDT
Date: Mon 6 May 85 15:33:13-PDT
From: Susan Fischer <FISCHER@SU-CSLI.ARPA>
Subject: Chronicles
To: folks@SU-CSLI.ARPA
Does anyone have a recent copy of the Chronicles of Higher Education that I could borrow or have?
Thanks.
Susan
-------
∂06-May-85 2114 @SU-CSLI.ARPA:PCOHEN@SRI-AI.ARPA Last reminder
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 6 May 85 21:12:32 PDT
Received: from SRI-AI.ARPA by SU-CSLI.ARPA with TCP; Mon 6 May 85 21:12:37-PDT
Date: Mon 6 May 85 21:10:34-PDT
From: Phil Cohen <PCOHEN@SRI-AI.ARPA>
Subject: Last reminder
To: NL4: ;, f4@SU-CSLI.ARPA, finterest@SU-CSLI.ARPA
The very last. Bob Moore will speak Tues, the 7th of May
(probably "today", when you read this) on a Possible Worlds Semantics
for Desire.
Time: 12:45 - 2:15
Place: Ventura Seminar Room
Phil
-------
∂06-May-85 2206 ullman@diablo next meeting
Received: from SU-AIMVAX.ARPA by SU-AI.ARPA with TCP; 6 May 85 22:06:01 PDT
Date: Mon, 6 May 85 22:05:45 pdt
From: Jeff Ullman <ullman@diablo>
Subject: next meeting
To: jmc@sail, nail@diablo
This Wednesday 5/8, John McCarthy will talk about
control algorithms for logic.
Joe Skudlarek's talk about *QUEL (or is it QUEL*) will be
given next week, 5/15.
∂07-May-85 0055 RESTIVO@SU-SCORE.ARPA PROLOG Digest V3 #22
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 7 May 85 00:54:51 PDT
Date: Monday, May 6, 1985 7:30PM
From: Chuck Restivo (The Moderator) <PROLOG-REQUEST@SU-SCORE.ARPA>
Reply-to: PROLOG@SU-SCORE.ARPA
US-Mail: P.O. Box 4584, Stanford CA 94305
Phone: (415) 326-5550
Subject: PROLOG Digest V3 #22
To: PROLOG@SU-SCORE.ARPA
PROLOG Digest Tuesday, 7 May 1985 Volume 3 : Issue 22
Today's Topics:
Implementation - CP,
LP Library - Abstracts
----------------------------------------------------------------------
Date: Sun 5 May 85 19:31:35-EDT
From: Vijay <Vijay.Saraswat@CMU-CS-C.ARPA>
Subject: The !-annotation revisited. #1]
Some time ago in these channels I had defined a read-only
annotation '!' which was meant to replace Concurrent
Prolog's '?', which I felt had a messy and unjustifiable
semantics.
In the last issue, "THRIFT%ti-csl.csnet@csnet-relay.arpa"
mentions a '+' annotation, but I fail tosee how it is
different from '!'. To make matters clear, here is a
more thorough discussion of '!'. (Note that this is a
substantially revised version of my previous submission.)
In the following, "term" is used to refer to
either an atom or a functional term.
First, assume that unification of a compound term T1 against
a compound term T2 is carried out by checking if T1 and T2
have the same functor and arity and then unifying their
corresponding arguments in PARALLEL.
The '!' annotation can only decorate instances of terms in
the head of a clause. If a Term (variable OR constant OR
compound term) T! occurs in the head of a clause, then
unification T!-V, where V is a variable will suspend and
remain suspended until V has been instantiated (to a
constant or a possibly non-ground compound term), after
which the unification T-V will be attempted. Like the `?',
the `!' is not inherited by embedded terms, that is, it
applies only to the term instance textually indicated in the
program.
However, if '!' annotates a term t1 inside a term t, then it
must also annotate all sub-terms of t which contain t1
(including t). In fact, we will define an embedded
occurrence of a '!' to be shorthand for just such a series
of '!' annotations in the term. (The atom at the head of a
clause is always !-annotated.)
Note: 1. This restriction is necessary to prevent
occurrences of unify(X, f(a!)) which doesn't make sense
because you cannot require a sub-term to be present unless
you also require the superior term to be present.
2. wait/1 in CP seems to achieve the suspension part of '!'
but cannot be used to simulate it in CP-without-? without a
control primtive to sequence goals.
3. A simple lazy sequential O(N↑2) algorithm can implement
!-unification correctly. (See next message in this Digest)
Note:
1. unify(Y!, X!) can never occur.
2. unify(Y, Y!) can, and suspends till Y is instantiated.
3. There is no 'inheritance' of '!' via X!-Y
unifications like there is for X?-Y.
4. The !-annotation can never 'occur' in any goal call at
run-time.
With '!' each CLAUSE decides what is to be INPUT to it.
With '?' each CALL decided what would be input to that call.
If all the clauses have the same pattern of input
specifications, then the '!' annotations could be removed in
favor of a mode-specification for the predicate. Since
nested '!' annotations are allowed within a term, in general
it is not possible to remove '!' annotations in favour of
mode declarations. (Of course, every program annotated with
the Dec-10 Prolog '+' (input) or '?' (dont-know) annotations
can be rewritten using '!' annotations; hence '!' is more
'general'.)
Examples: 1. merge/3. Equivalent mode: merge(+,+,?).
merge([A|X]!, Y, [A|Z]):- merge(X, Y, Z).
merge(X, [A|Y]!, [A|Z]):- merge(X, Y, Z).
2. plus/3. (No single equivalent mode declaration.)
plus(X!, Y!, Z):- Z is X+Y.
plus(X!, Y, Z!):- Y is Z-X.
plus(X, Y!, Z!):- X is Z-Y.
'!' is less powerful than '?' because it cannot be used to
simulate 'var' and hence write a b-merge like program, even
assuming strict-AND fairnes of the scheduler.
Hence I would propose using '!' in conjunction with var/1.
For var/1 I would propose some syntactic sugar: another
annotation (↑, say). Again, the ↑-annotation can occur only
in the clause-head, and annotates just the occurrence of the
Term it is textually adjacent to. The unification Term↑-X
succeeds only if X is a variable and results in X being
unified with Term. If X is not a variable, then Term↑-X
FAILS, though this must be regarded as a control failure.
Term↑-X NEVER suspends.
Note that nested occurrences of ↑ do not make sense.
Examples:
1. "foo(foo↑)↑" is meaningless.
2. "foo(foo↑)" will match any of the following terms:
X, foo(X).
3. "foo(foo↑)!" will match any of the following terms:
foo(X).
I feel I must also sound this note of warning. With the
assumption of weak AND-fairness (which, roughly paraphrased
says that every enabled goal will be reduced in a finite
amount of time), var/1 (or ↑) is enough to write a program
which exhibits unbounded non-determinism. That is, the
program always terminates, but for every natural number 'k'
can produces as output y > k. Here it is:
p(X):- p(X,Stop), stop(Stop).
p(s(X), Stop↑):- p(X,Stop).
p(X, stop).
stop(stop).
(Note: this is not a CP program: its a 'pure' (save for ↑)
Horn logic program.) Without going into details suffice to
say that the semantics of constructs which allow unbounded
non-determinism can cause severe headaches because the limit
of arbitrary chains may not exist. My personal feeling is
that '↑', or its equivalent, should be allowed in any
language which purports to be a CONCURRENT language; of
course as soon as we introduce '!' or '↑' we are not talking
of LOGIC languages any more.
'!' cannot be (because it was not DESIGNED to be) used to
export 'protected' variables. But the whole thesis here is
that clauses should be forced to specify what their input is
and hence the 'embedded channel' problem in Hellerstein and
Shapiro, ISLP, 1984 can be taken care of by !-protecting at
the consume site instead of ?-protecting at the produce
site.
'!' seems to cover the vast majority of uncontroversial uses
of the `?' and is far simpler semantically. For example, it
can be shown that once a goal previously blocked on '!'
becomes unblocked, it remains unblocked through commit time.
No proposal for '?' which does things like unify X with Y?
to make X Y? in all its occurrences can do that.
Finally, for a more formal treatment of the !-annotation,
see next message.
-- Vijay Saraswat.
------------------------------
Date: Sun 5 May 85 15:39:36-EDT
From: Vijay <Vijay.Saraswat@CMU-CS-C.ARPA>
Subject: Semantic of the !-annotation. #2
Here is a more formal discussion of the '!' annotation.
Please see previous post for an informal motivation. It is
taken from my forthcoming tech rep 'Problems with Concurrent
Prolog'.
There are a number of other alternate formal definitions.
This semantics of '!' has been used to give two formal
semantics to Concurrent Prolog programs. Please see
succeeding messages.
-- Vijay A. Saraswat
The !-annotation.
In this section, we will think of terms as trees, that is,
as partial functions from the set of all possible paths
(i.e. finite and infinite sequences of natural numbers) to a
co-domain C +{\bot}, where C is the set of node-labels. If
the function is not defined for a given argument, we will
take its value to be \bot. Then terms are trees over the
co-domain of function symbols and variables.
We can now define ANNOTATIONS to be simply trees over the
domain {tt}+ {\bot}, with the interpretation that the (node
specified by the) path l is annotated by p iff p(l)=tt.
However, we would also like to insist that if a term is
annotated then all its super-terms are also annotated.
Therefore, we require that if p(l)=tt, then so is p(l'), for
all prefixes l' of l. [Ancestor condition]
An annotation p is APPLICABLE to a term t iff for all l,
t(l) is defined when p(l) is.
We would now like to give the semantics of an annotated
term. The annotations serve to restrict the set of terms
with which the annotated term can unify. Consider a term s
annotated by p and a term t. First we would like to express
the notion of a most general unifier for s and t which
ignores all the annotated nodes in s and the corresponding
nodes in t, if they exist.
The p-RESTRICTED MGU of two terms s and t, denoted by
mgu←r(p,s,t), where p is an annotation applicable to s, is
the most general substitution q such that p(l)# tt =>
(q(s))(l) = (q(t))(l).
According to the intuitive meaning of !, we must ensure that
all annotated terms unify against non-variable terms. This
means that when unifying a term s against t, if there is a
path l such that p(l)=tt and s(l)=s' and t(l)=t', we cannot
proceed until and unless t' is instantiated. But if t' is
instantiated, we can remove the annotation on s'. So we can
find mgu-!(p,s,t) by unifying the terms in s which are not
!-protected by p against the corresponding sub-terms of t,
i.e. by computing mgu←r(p,s,t). If this leads to
instantiating a variable in t which has a corresponding
sub-term in s that is !-protected, then we can remove this
annotation and start again. If mgu(s,t) exists, then the
process terminates in success iff there are no more
!-annotations left, and then mgu-!(p,s,t)= mgu(s,t) and in
failure iff all nodes !-annotated by the current p in t are
variables. Hence:
An !-MGU of two terms s, t where p is an annotation
applicable to s, is a substitution q←n such that there
exists a sequence p=p←0, ... p←n=\bot such that for all l.
p←{i+1}(l)=tt <-> p←i=tt & (q←i(s))(l) is a variable, where
q←i=mgu←r(p←i, s, t)
(p←n=\bot means that there are no nodes in p←n which have
the value tt.)
Note that mgu←r(\bot,s,t)=mgu(s,t).
------------------------------
Date: Sun 5 May 85 15:40:22-EDT
From: Vijay <Vijay.Saraswat@CMU-CS-C.ARPA>
Subject: A new commit: & #3
Now that we seem to have recovered from the discussions on
the meaning of '?' in Concurrent Prolog, let me initiate a
discussion on the role of '|'.
In the following, by CP I mean CP with '!' and without '?'.
The reason is that there doesn't exist (as yet) a consistent
definition of '?' and I consider '!' to be a perfectly
useful and simple substitute for '?'.
The semantics of CP can be given by specifying which
SLD-derivations are ADMISSIBLE, given a query and a
CP-program. I contend that the reason CP is not a logic
programming language is because it does not distinguish
failed admissible SLD-derivations from admissible
SLD-refutations. Because of its don't care non-determinism,
committing can only be locally angelic, i.e it chooses
values for its free variables such that its guard executes
successfully, but its body may still fail for the chosen
values. This means:
1. Validity of unannotated axioms is not sufficient for
partial correctness: a given query wil assuredly succeed
(or loop) ONLY if all finite admissible SLD-derivations
are refutations, which is a very strong condition.
2. There can be no notion of negation-as-failure
even with respect to admissible derivations. (i.e. at
best we can hope that negation means that no admissible
derivation for the query is a refutation... even that is not
compatible with '|'.)
3. Many Horn logic axiom definitions cannot be used in CP.
As an illustration, no version of the axioms:
p(X,Y):- t(X,Y).
p(X,Y):- t(X,Z), p(Z,Y).
annotated with CP's annotations can be guaranteed to work
correctly (i.e. compute the transitive closure of p/2) for
an arbitrary (Horn) definition of p/2. (For example,
consider the program:
p(0,1). p(0,2). p(1,3).
I. t(X!, Y!):- p(X,Y)| true.
II. t(X!, Y!):- p(X,Z), t(Z,Y)|true.
The query '?-t(0,3).' may fail because in II, p(X,Z) may
commit to:
p(0,2). )
There are two simple alternate interpretations for 'commit'
which distinguish between successful and unsuccesful
admissible derivations.
1. CP-amb or the '&' annotation: don't-know operator. (This
corresponds to McCarthy's famous "amb" operator in the
context of HLP.) Interpretes "commit" as "make global". It
does not snip off other OR-siblings, but instead continues
to follow them, allowing MULTIPLE commits of OR-siblings.
Each commit is to a different copy of the rest of the
environment. In effect whereas 'a:-g | b.' extends some
admissible refutation of 'g' by an admissible refutation for
'b' to return ONE (selected from possibly many) refutation
for 'a' (all this talk is modulo multiple environments),
'a:-g & b.' extends EVERY admissible refutation of 'g' by
an admissible refutation for 'b' and returns ALL of them as
refutations for 'a', thereby avoiding a local commitment to
one refutation of 'a'. Therefore, an SLD-derivation ends in
failure only if all admissible derivations are finite and
failing, just as for Prolog.
A completely formal description of & is given in my
forthcoming tech rep 'An operational semantics for
Concurrent Prolog'.
2.CP-backtrack or the '\' annotation. Interpretes "commit"
as "make global and freeze other OR-brothers". Here ONE
admissible SLD-derivation is followed until it terminates.
If it terminates in success, nothing is done. Failure
induces backtracking. For partial correctness, the exact
backtracking scheme used is not important (chronological,
dependency-directed), though of course pragmatically it is
quite important, as long as it can be guaranteed that the
system will not terminate in failure as long as even one
admissible SLD-derivation path has not been pursued. That
is, no FINITE SLD-derivation is admissible unless it is a
refutation or else all admissible derivations are finite and
failing.
Lemma: Given a CP program and a query, the set of possible
answers to a query is the same if the '|' is interpreted as
'commit', '&' or '\' uniformly in the program.
That is, | and & are equivalent as far as the success
semantics is concerned.
The relationship between interpreting '|' as commit, '&' and
'\' uniformly in a program is as follows:
[CP]
1.i Execution of a query ALWAYS terminates IFF all
admissible SLD-derivations are finite.
1.ii Execution may terminate in failure even if there
is an admissible refutation.
[CP-backtrack]
2.i As in 1.i.
2.ii Execution terminates in failure ONLY IF no
admissible SLD-derivation is a refutation. If a
query terminates in failure under CP-backtrack, it
will always terminate in failure under CP.
[CP-amb]
3.i Execution of a query always terminates IFF either
there is an admissible SLD-refutation or all
admissible SLD-derivations are finite.
3.ii As in 2.ii.
Caveat:
As in Prolog, allowing implicit search allows the user to
write very inefficient programs. On the other hand, for
whatever it is worth, much of Guy Steele's thesis work on
constraints can be recast in such a logic framework.
I would like to retain '|' much for the same reason that the
'!' is used in Prolog: sometimes it helps to signal
determinate situations and avoid redundant solutions. When
mixing '&' and '|', executing a '|' does not affect any
previous '&' that may have been executed: it just kills off
any remaining OR-siblings.
I leave you with the following example: (& represents the
"amb-commit").
prod(X!,Y!,Z):- Z is X*Y | true.
prod(X!, Y, Z!):- X =/= 0 | Y is Z/X.
prod(X, Y!, Z!):- Y =/= 0 | X is Z/Y.
prod(X, Y, Z!):- less(X,Z), less(Y, Z), Z := X*Y | true.
Z := X!*Y! :- Z is X*Y.
This can "solve" prod(X,X,16) to give X=4 provided that
less/2 is assumed defined as if by the collection of clauses
less(i#,j#):- true & true. for each value of i# and j# such
that i# < j#, so that it serves as a generator.
-- Vijay Saraswat
------------------------------
Date: Sun 5 May 85 15:41:48-EDT
From: Vijay <Vijay.Saraswat@CMU-CS-C.ARPA>
Subject: A meta-interpreter for CP[!,|,&] #4.
Here is a simple meta-interpreter for CP[!,|,&] in
CP[!,|,&]. Note that it is not possible to give as simple
an interpreter for CP[!,|] in CP[!,|]. This interpreter can
be extended to give an interpreter for CP[!,↑,|,&] in
CP[!,↑,|,&].
We define a predicate cp/1 which takes as input a goal and
solves it. The user-program is added in as clauses of the
form:
clause(Head:-Guard | Body):- true & true.
or
clause(Head:- Guard & Body):- true & true.
where:
1. All instances of '!' in Head are replaced by '$', which will
be regarded as a unary post-fix function symbol,
2. Guard and Body are sequences of goals of the form '{g←1, ...,
g←n}'.
In addition we also have the following axioms (which can be
added automatically):
For every functor f/n, (n>= 0) in the user-program, the
axioms:
f(X1,...Xn)! = f(Y1,...Yn)$:- X1 = Y1,...Y1=Yn | true.
f(X1,...Xn)$ = f(Y1,...Yn)!:- X1 = Y1,...Y1=Yn | true.
Then the interpreter is the following:
X=X:- X =\= G$.
unify(X!, Y):- X =Y | true.
execute(G!):- clause((Goal :- Guard | Body)),
unify(Goal, G),
execute←all(Guard) | execute←all(Body).
execute(G!):- clause((Goal :- Guard & Body)),
unify(Goal, G),
execute←all(Guard) & execute←all(Body).
execute←all({One|Rest}!):- true | execute(One), execute←all(Rest).
execute←all( true!):- true | true.
There are two cheats involved in this simple interpreter:
1. The term {g←1, ... g←n} (analogous to Dec-10 Prolog's [g←1,
... g←n]) is actually shorthand for the term g←1@(g←2@...
g←n)))..), where the @/2 operator is ASSOCIATIVE and
COMMUTATIVE. Hence {g←1,... g←n} represents a SET (actually
multi-set) of goals, rather than a list of goals, and when
unified against {X←1,... X←m|X} succeeds if 1 <=m <=n, and
unifies SOME m goals (not necessarily the FIRST m) from g←1
... g←n with X←1 ... X←m and the set of the rest of them
with X. We use sets instead of lists in order to avoid
giving priority to goals that might textually occur first in
the body of a clause; the same effect could be achieved by
using some random selection function. So the first cheat is
assuming that CP has a notion of associative-commutative
unification built in.
2. Second, we cheat in using the binary predicate =/=, which
has a weak form of var/1 built into it. Of course, given
=/=, we can simulate =/=, so we still have a complete
meta-interpreter. I do not know how to avoid using some
such device and still get an elegant program. Possibly by
representing user-program terms in a special fashion, one
may define a var/1 for user-program terms, but it is sure to
be messy.
The reason one cannot write as simple an interpreter for
CP[!,|] is that in CP[!,|] one cannot add the user-program
as a list of clauses to the interpreter as we have done
because then a call to clause(Head:-Guard|Body) would
succeed at most once, selecting some clause/1 clause at
random whereas we would like to select all clauses in
parallel and execute their guards concurrently,i.e. we DON'T
KNOW which clause we want. Hence we are forced to represent
the program explicitly as an argument to the interpreter, as
a list of clauses and that makes any meta-interpreter very
messy. But of course using the & in the body of clause/1
clauses achieves just the desired effect.
-- Vijay A. Saraswat
------------------------------
Date: Sun 5 May 85 15:42:29-EDT
From: Vijay <Vijay.Saraswat@CMU-CS-C.ARPA>
Subject: An operational semantics for CP. #5
Here is the abstract from a forthcoming tech rep "An
operational semantics for Concurrent Prolog".
-- Vijay A. Saraswat
Abstract. In this paper we present a simple
operational semantics for CP[!,|,&] using the notion of
Plotkin-style labelled transition system, which we believe
offers a simple, general framework for issues of concurrency
in Horn logic programming. The space of configurations we
work with naturally reflects the AND-OR process structure of
an executing concurrent logic program. The problems solved
in giving this semantics were the representation and
manipulation of a distributed environment and the effect of
two different kinds of commits: the don't care commit
originally proposed in \ref{Cp} and the don't-know commit
proposed in \ref{problems-cp}, together with their
interactions.
From the transition system, we extract various
`meaning-functions' for a program. We show that with
respect to one of these (the `success-semantics') a
CP[!,&] is the same as the corresponding CP[!, |]
program.
Essentially the transition system is a formalisation of a
derivability predicate in pure Horn logic, i.e. the
semantics can be looked upon as the specification in pure
Horn logic of a high-level interpreter for CP which
represents precisely the sequence of computations possible
for a given query and a given CP program.
------------------------------
Date: Sun 5 May 85 15:46:13-EDT
From: Vijay <Vijay.Saraswat@CMU-CS-C.ARPA>
Subject: A scenario semantics for CP. #6
Here is the abstract from my forthcoming CMU Tech report: "A
scenario semantics for Concurrent Prolog".
-- Vijay A. Saraswat
Abstract: In this paper we present a simple bottom-up
semantics for CP which captures the set of all possible
answer-substitutions for a query, given a CP[!,|,&] program.
The basic idea is to define a transformation analogous to
T←p:I rightarrow I (which defines the semantics of a pure
Horn logic program as lfp(T←p)). Because of CP's
annotations, however, we cannot interprete processes as
predicates over the underlying Herbrand base; we need more
operational information. We define the notion of {it
scenarios}, adapted from the semantics given in cite{brock}
for non-deterministic data-flow languages, and give the
success-semantics of a CP[!,|,&] program as lfp(S←p) of a
transformation S←p:I←s rightarow I←s from sets of scenarios
to sets of scenarios.
------------------------------
End of PROLOG Digest
********************
∂07-May-85 1143 BLOCK@SU-CSLI.ARPA benevolence
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 7 May 85 11:43:15 PDT
Date: Tue 7 May 85 10:03:58-PDT
From: Ned Block <BLOCK@SU-CSLI.ARPA>
Subject: benevolence
To: rrr@SU-CSLI.ARPA
This is a very teeny criticism of John's analysis of benevolence. Recall
that the 2 versions are:
(1)If (i) believing P and desiring Q cause A, then (ii) A Promotes Q,
given P.
(2)Whenever (i)a is in circs C such that Proj (a,S,C,Q,P), then (ii) a
is also in circs C' s.t. some A caused by S is s.t. Prom(a,A,C',Q,P)
Note that the antecedent of (1) but not (2) requires that A obtain.
As a result, benevolence1 fails when action doesn't succeed, whereas
benevolence2 fails, in addition, when you don't get what you want, even
if you wouldn't dream of acting on the wants. John offers (2) as an
improvement on (1). My point is that in so doing, he changes ideas of
benevolence, and that benevolence1 is arguably more interesting than
benevolence2.
The idea of benevolence1 is that there is a correspondence between the
desires that motivate our actions and the effects of our actions when
our beliefs are true (viz., satisfying the desires). The idea of
benevolence2 is: we get what we want. Benevolence in the former sense
obtains--or at least, it is a good idealization for many purposes.
The latter, alas...
Benevolence1 seems necessary for a thriving organism; presumably, some
sort of evolultionary account can be offered for it. The degree of
Benevolence2, by contrast, probably varies from culture to culture,
since it goes down when Rupert Murdoch buys a local paper and suggests
to people a greater variety of things they might want (but can't get).
For this reason, benevolence2 is unlikely to be biologically
important. You can't always get what you want, but if you try real
hard, you just might, you just might, get what you need. Rationality
dictates prior attention to wants that reflect needs. Think of it
from a decision-theoretic point of view. My preference function
weights a huge class of outcomes, most of which are far from the realm
of real possibility. Think of outcomes above a certain level as
"desired". In my case, this class would include meeting my twin on
twin earth. My realization that this isn't possible doesn't harm me.
It could be said that such desires are mere fantasies rather than real
wants, but that way of reducing the distance between benevolence1 and
benevolence2 would be question-begging, since the idea behind it is
that "desires" that aren't likely to motivate action are only
fantasies, and this is a benevolence1-style conception. We need some
notion of desire that is neutral to the issue of what benevolence is,
since the project is to characterize benevolence.
Notice that those who aren't afraid of "narrow" states can retain the
nice proerties of version 1, while avoiding Perry's objection to it by
retaining 1, individuating the entities in 1(i) narrowly, and the
entities in 1(ii) widely.
-------
∂07-May-85 1623 HEDGES@SU-SCORE.ARPA ACM Doctoral Dissertation Award Program
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 7 May 85 16:22:55 PDT
Date: Tue 7 May 85 14:13:28-PDT
From: Karen Hedges <HEDGES@SU-SCORE.ARPA>
Subject: ACM Doctoral Dissertation Award Program
To: Faculty@SU-SCORE.ARPA
ACM has asked if there are any members of our department who would be
interested or willing to assist with the evaluation process for the
ACM Doctoral Dissertation Award Program.
If any faculty are interested, I have the letter from ACM available for
review in my office.
Thanks.
Karen
-------
∂07-May-85 1718 PIERCE@SU-SCORE.ARPA Quad Dedication
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 7 May 85 17:17:56 PDT
Date: Tue 7 May 85 15:20:05-PDT
From: Lee Pierce <PIERCE@SU-SCORE.ARPA>
Subject: Quad Dedication
To: faculty@SU-SCORE.ARPA, staff@SU-SCORE.ARPA
Hi folks,
This is your last chance to let me know if you plan to attend the
famous Quad Dedication on May 17 from 4-6pm. I have kept a careful
list of all who have responded, so you need only respond if you want
to attend and haven't let me know. Don't forget that spouses ARE
invited as well as those who don't work on the Quad. Thanks for your
help.
---Lee
-------
∂07-May-85 1748 JODY@SU-CSLI.ARPA The Burrito Bandit
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 7 May 85 17:46:06 PDT
Date: Tue 7 May 85 17:44:56-PDT
From: Joe Zingheim <JODY@SU-CSLI.ARPA>
Subject: The Burrito Bandit
To: folks@SU-CSLI.ARPA
The Burrito Bandit will provide a change of pace for the lunch menu this
Thursday. There are nine choices, each $2.75 for the regular and $3.50 for
the "especial" which has avacado, cheese, and sour cream added. These
burritos are huge! They weigh around a pound and a half. Orders placed by
10:30 Thursday will be ready for pickup after 11:45. Indicate your choice
and regular or special when ordering.
Carnitas (shredded pork) with rice
Pollo (chicken) with rice
Chile Verde (pork in green sauce) with rice
Chile Colorado (beef in red sauce) wiht rice
Lengua (beef tongue) with with rice
Chile Relleno (green chile stuffed with cheese) and rice
Picadillo (shreaded beef) with beans
Carne Asada (Mexican style steak) with beans
Chorizo (Mexican sausage) with beans
Send mail, as usual, to Lunch@CSLI
-------
∂07-May-85 1803 CLT Seminar in Logic and Foundations of Mathematics
To: "@DIS.DIS[1,CLT]"@SU-AI.ARPA
Speaker: Prof. Craig Smorynski, San Jose State University
Title: Quantified modal logics of provability.
Time: Tuesday, May 14, 4:15-5:30 PM
Place: Room 381-T, 1st floor Math. Bldg., Stanford
S. Feferman
Next Meeting: Volunteers welcome for the meeting of May 28. If you
would like to speak, please let Jon Barwise know (after the 15th) or
David Israel, as I shall be away. SF
∂07-May-85 1943 JOHN@SU-CSLI.ARPA Good courses at Berkeley
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 7 May 85 19:43:21 PDT
Date: Tue 7 May 85 19:42:59-PDT
From: John Perry <JOHN@SU-CSLI.ARPA>
Subject: Good courses at Berkeley
To: folks@SU-CSLI.ARPA
I am interested in learning of extremely good courses at UC Berekely that
are open to freshman. These courses need not have anything to do with
the interests of CSLI. I am just interested in learning which are the very
best taught, most interesting courses a freshman can take, regardless of
area. Short notes identifying such courses will be much appreciated.
Thanks. JP.
-------
∂08-May-85 0513 allegra!seismo!mcvax!ukc!snow!req@diablo cprolog speed
Received: from SU-AIMVAX.ARPA by SU-AI.ARPA with TCP; 8 May 85 05:13:32 PDT
Received: from mcvax.UUCP by seismo.ARPA with UUCP; Wed, 8 May 85 06:10:18 EDT
Received: by mcvax.UUCP; Wed, 8 May 85 09:05:35 -0200 (MET)
Received: by mcvax.UUCP; Wed, 8 May 85 09:05:30 -0200 (MET)
Message-Id: <8505080705.AA25857@mcvax.UUCP>
Received: from ubu by ukc.UUCP id a000202; 8 May 85 4:47 BST
Date: Mon, 6 May 85 17:32:50 GMT
From: allegra!seismo!mcvax!warwick!snow!req@diablo
To: diablo!nail@diablo
Subject: cprolog speed
> Probably, the system, written in cprolog, would be 100 times
> slower than a system written in C or a similar language.
> I'm not all that concerned with the speed of a *prototype* system,
> though. There seemed to be some confusion at the Wednesday
But there is the possibility of a prolog compiler in the near future from a
number of sources. Prolog seems by and large to be reasonably portable across
versions, so this may be feasible.
∂08-May-85 0728 @SU-SCORE.ARPA:ARK@SU-AI.ARPA CS Colloq, May 14: Simon Lam, Observing Protocol Systems with Low-Resolution Snapshots
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 8 May 85 07:27:58 PDT
Received: from SU-AI.ARPA by SU-SCORE.ARPA with TCP; Wed 8 May 85 07:16:59-PDT
Date: 08 May 85 0000 PDT
From: Arthur Keller <ARK@SU-AI.ARPA>
Subject: CS Colloq, May 14: Simon Lam, Observing Protocol Systems with Low-Resolution Snapshots
To: Faculty@SU-SCORE.ARPA, All-Colloq@SU-SCORE.ARPA
CC: ARK@SU-AI.ARPA, lam%muan.UTEXAS@UT-SALLY.ARPA
CS Colloquium, May 14, 4:15pm, Terman Auditorium
OBSERVING AND VERIFYING PROTOCOL SYSTEMS
WITH LOW-RESOLUTION SNAPSHOTS
Simon S. Lam
Dept. of Computer Sciences
University of Texas at Austin
We consider communication protocol systems modeled as networks of
processes that interact via message passing. To prove that a specific
logical property holds for a protocol system, it is often unnecessary to
observe the state of the system in all its details, i.e., the identities
of many process states and many messages can be indistinguishable in our
observation of the system state.
We present a method for constructing smaller systems, from a given
protocol system, by aggregating some process states and messages in the
given system. A system constructed this way has a lower resolution than
that of the original. (We call them image protocol systems.) We present
conditions under which safety and liveness properties of an image protocol
system will also hold in the original system.
----
Cookies and juice will probably be served in the 3rd floor lounge of MJH
at 3:45pm. See you there.
∂08-May-85 0825 YAMARONE@SU-CSLI.ARPA [Ingrid Deiwiks - 497-3084 <INGRID@SU-CSLI.ARPA>: Sandwiches]
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 8 May 85 08:25:47 PDT
Date: Wed 8 May 85 08:24:03-PDT
From: Tom Yamarone <YAMARONE@SU-CSLI.ARPA>
Subject: [Ingrid Deiwiks - 497-3084 <INGRID@SU-CSLI.ARPA>: Sandwiches]
To: folks@SU-CSLI.ARPA
I will be calling in an order at 10:00 this morning if any of you are
interested in ordering lunch from our new vendor.
Please don't wait,
Don't hesitate,
It'll be too late,
And you'll be hungry.
---------------
Mail-From: INGRID created at 7-May-85 15:31:22
Date: Tue 7 May 85 15:31:22-PDT
From: Ingrid Deiwiks - 497-3084 <INGRID@SU-CSLI.ARPA>
Subject: Sandwiches
To: John@SU-CSLI.ARPA
cc: Yamarone@SU-CSLI.ARPA, Jamie@SU-CSLI.ARPA, Betsy@SU-CSLI.ARPA
I have just spoken to Kamel who is the Manager at Stanford Delicatessen
in the Shopping Center. Their prices are as follows
small sandwich $2.15
$2.45 with cheese
large sandwich $2.95
$3.25 with cheese
small box lunch $4.50
large box lunch $5.00
(The box lunch consists of either a small or large sandwich,
a salad, and a desert (choice of brownie, cake, etc.)
There would be choices of bread on all of the above.
In addition, we could have drinks, fruit .... whatever.
If we phone in around 10 or 10:15 in the morning, we can have the
sandwiches made to order.
However, the only discount they can offer is the $5.00 for the large
box lunch (instead of $5.25).
I think this is what we have been looking for, and it would only mean a
small amount of extra work (probably for Tom) to phone in the order
every day. We could pick them up at 11:30.
Ingrid
-------
-------
∂08-May-85 1047 JOHN@SU-CSLI.ARPA next week's session
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 8 May 85 10:47:01 PDT
Date: Wed 8 May 85 10:44:05-PDT
From: John Perry <JOHN@SU-CSLI.ARPA>
Subject: next week's session
To: rrr@SU-CSLI.ARPA
cc: ingrid@SU-CSLI.ARPA
Ingrid will have three (3) additional papers (old) by me that those truly
interested in these things might find worth plowing through, in addition to
more copies of the benevlolent etc. one for thsoe who didn't get one last time.
In order of importance
Perception, action and the structure of believing
A Problem About Continued Belief
Acceptance and Belief
-------
∂08-May-85 1136 WRIGHT@SUMEX-AIM.ARPA SIGlunch: April 10
Received: from SUMEX-AIM.ARPA by SU-AI.ARPA with TCP; 8 May 85 11:36:14 PDT
Date: Wed 8 May 85 11:17:38-PDT
From: Carol Wright/Susie Barnes <WRIGHT@SUMEX-AIM.ARPA>
Subject: SIGlunch: April 10
To: siglunch@SUMEX-AIM.ARPA
Because of a RIPlunch there will be no SIGlunch (Heuristic Programming
Seminar/CS 322) this week.
Thanks, Susie
-------
∂08-May-85 1321 HEDGES@SU-SCORE.ARPA Robotics Candidate
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 8 May 85 13:20:20 PDT
Date: Wed 8 May 85 13:18:53-PDT
From: Karen Hedges <HEDGES@SU-SCORE.ARPA>
Subject: Robotics Candidate
To: Faculty@SU-SCORE.ARPA
Dr. Peter Will will be visiting the Department on Wednesday, May 15. IF
anyone is interested in talking with him there is time available, both
morning and afternoon.
Please let me know what times might be convenient for you to meet and talk
with him.
Thanks.
-------
∂08-May-85 1409 WRIGHT@SUMEX-AIM.ARPA SIGlunch ooops
Received: from SUMEX-AIM.ARPA by SU-AI.ARPA with TCP; 8 May 85 14:08:59 PDT
Date: Wed 8 May 85 13:58:20-PDT
From: Carol Wright/Susie Barnes <WRIGHT@SUMEX-AIM.ARPA>
Subject: SIGlunch ooops
To: SIGlunch@SUMEX-AIM.ARPA
The message sent previously about no Siglunch is for this week,
MAY 10 not APRIL 10. Sorry.
Susie
-------
∂08-May-85 1504 PETERS@SU-CSLI.ARPA Hausser Talk and Demo
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 8 May 85 15:04:13 PDT
Date: Wed 8 May 85 15:02:01-PDT
From: Stanley Peters <PETERS@SU-CSLI.ARPA>
Subject: Hausser Talk and Demo
To: nlinterest@SU-CSLI.ARPA, linguists@SU-CSLI.ARPA, csli↑.pa@XEROX.ARPA,
bboard@SU-CSLI.ARPA
Next Tuesday, Roland Hausser will give a talk and demonstration
of a system of grammar and parsing he has been working on during
his stay at CSLI. The presentation will take place from in the
CSLI trailer Classroom 1 till 1:45 on May 14th.
He will illustrate his system by applying it to German. The
grammar contains 32 linguistic rules covering declarative and
interrogative main clauses, relative clauses embedded to
arbitrary depth or extraposed, variant word-orders, adverbs and
adverbial clauses, discontinuous constituents, coordinate
structures, and other phenomena. The parser employs a bottom-up,
left-associative, data-driven algorithm and is implemented in
Interlisp-D on the Dandelion.
-------
∂08-May-85 1519 @SU-CSLI.ARPA:BMOORE@SRI-AI.ARPA Your comments
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 8 May 85 15:19:19 PDT
Received: from SRI-AI.ARPA by SU-CSLI.ARPA with TCP; Wed 8 May 85 15:19:25-PDT
Date: Wed 8 May 85 15:16:24-PDT
From: Bob Moore <BMOORE@SRI-AI.ARPA>
Subject: Your comments
To: bratman@SU-CSLI.ARPA
cc: NL4@SU-CSLI.ARPA, F4@SU-CSLI.ARPA, bmoore@SRI-AI.ARPA
Michael,
Your comments yesterday were insightful as ever. I think what your
observations show is that the notion of desire my proposal is relevant
to is slightly different than I first thought. It now seems to me
that we need to distinguish "cravings" from "reflective desires". The
proposal I made seems much better suited to the latter. Using
possible worlds the way it does seems to make it a notion of desire
"all things considered." In this role it seems to fit rather well.
Take the case of having $10M versus having $20M. If someone has less
than $10M and values having more money, then both alternatives will
seem desirable. If one has $20M, though, having $10M will seem
undesirable. This is exactly what the analysis predicts should be the
case.
In the chocolate case, though, there may be a <<craving>> for
chocolate that remains, whatever one desires all things considered.
Parallel to the money example, if you are neither eating chocolate nor
losing weight, either might seem desirable (suppose eating chocolate
prevents you from losing weight, but doesn't cause you to gain
weight). If you are losing weight, though, and you value that MORE
than eating chocolate, eating chocolate ought to be undesirable <<all
things considered>>. This seems to be the basis of statements such as
"Although I like (i.e. crave) chocolate, I don't want to eat any
because I want to keep losing weight." This doesn't mean, however,
that you don't still crave chocolate, which might be taken to be a
kind of desire (e.g. "Desire Under the Elms"). The reason this
distinction doesn't naturally arise in the money case is that having
$10M and having $20M BOTH address the same craving, the craving for
wealth.
--Bob
-------
∂08-May-85 1540 LIBRARY@SU-SCORE.ARPA SOCRATES: Searching Techniques--The Browse Command An Update
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 8 May 85 15:40:42 PDT
Date: Wed 8 May 85 15:39:30-PDT
From: C.S./Math Library <LIBRARY@SU-SCORE.ARPA>
Subject: SOCRATES: Searching Techniques--The Browse Command An Update
To: su-bboards@SU-SCORE.ARPA
cc: faculty@SU-SCORE.ARPA
8-May-85 11:17:58-PDT,2650;000000000001
Return-Path: <CN.MCS@Lindy>
Received: from Lindy by SU-SCORE.ARPA with TCP; Wed 8 May 85 11:17:32-PDT
Date: Wed, 8 May 85 11:16:55 PDT
From: <CN.MCS@Forsythe>
To: LIBRARY@SCORE
THE BROWSE COMMAND
The BROWSE command may be used in both the Lookup and
Command modes. Help BROWSE will display the following:
"The BROWSE command may be used to scan headings
alphabetically. Authors, titles, organization names,
subjects, subject names, and call numbers may be
scanned by the use of the BROWSE command. This
command is exceptionally helpful when looking for
variant spellings of author's name." To use the
BROWSE command one should type BROWSE at the "Your
Response:" prompt when in the Lookup mode. In the
Command mode, just type BROWSE followed by the
index name such as S for subject or T for title and
the subject word or title word you wish to BROWSE.
At times the BROWSE command can be the most efficient and
effective way of searching.
When attempting to find material on the Japanese game Go,
one would receive the message "request has no content" if
you searched Find T Go or Find S Go. This is because when
searching the word indexes such as title or subject you
must use at least three characters. However one can
Browse S Go or Browse T Go and immediately see a display
listing material. Browse is also useful when searching a
single word title when the word is also very general and
would retrieve a large number of hits if searched as a
single word. Science and Nature are two journal titles
easily retrieved by using the BROWSE command. Searches
using the names of computer programming languages such
as Ada, Pascal, and Forth are more effective with the
Browse command since searching in other ways would retrieve
a large number of irrelevant hits.
To see how the BROWSE command differs from other commands,
experiment with a topic and try the following searches:
Find T, Find TP, and Browse T--Were there differences in the number
of hits? How did the order in which the hits were displayed differ?
Try your topic again with Find S, Find SP, and Browse S.
Find S will include both titles and topics while Find SP
and Browse S include only topics. Basically the BROWSE
command is a unique form of the various phrase indexes while
also displaying your retrieval in an alphabetical listing
just as you would find in the physical card catalog.
For a unique capability of the BROWSE command, use it to search
on a call number with Browse Call Number. You will retrieve
a listing in shelf order of all the material located in the
particular call number area you searched.
-------
∂08-May-85 1748 EMMA@SU-CSLI.ARPA Newsletter May 9, No. 28
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 8 May 85 17:47:52 PDT
Date: Wed 8 May 85 17:07:29-PDT
From: Emma Pease <Emma@SU-CSLI.ARPA>
Subject: Newsletter May 9, No. 28
To: friends@SU-CSLI.ARPA
Tel: 497-3479
C S L I N E W S L E T T E R
←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←
May 9, 1985 Stanford Vol. 2, No. 28
←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←
A weekly publication of The Center for the Study of Language and
Information, Ventura Hall, Stanford University, Stanford, CA 94305
←←←←←←←←←←←←
CSLI ACTIVITIES FOR *THIS* THURSDAY, May 9, 1985
12 noon TINLunch
Ventura Hall ``Scenes and Events''
Conference Room by Stephen Neale, Dept. of Philosophy, Stanford
Discussion led by Mark Johnson, Dept. of Linguistics
2:15 p.m. CSLI Seminar
Redwood Hall ``Approaches to Generalized Quantifiers in
Room G-19 Heim/Kamp Semantics''
Mats Rooth, CSLI
3:30 p.m. Tea
Ventura Hall
4:15 p.m. CSLI Colloquium
Redwood Hall ``Reduced Forms of Comparative Clauses''
Room G-19 James D. McCawley, University of Chicago
←←←←←←←←←←←
CSLI ACTIVITIES FOR *NEXT* THURSDAY, May 16, 1985
12 noon TINLunch
Ventura Hall ``Combinators, Categorial Grammars, and Parasitic
Conference Room Gaps'' by Mark Steedman, University of Edinburgh
Discussion led by Hans Uszkoreit, CSLI and SRI
(Abstract on page 2)
2:15 p.m. CSLI Seminar
Redwood Hall ``Action Theory for Dialogue Games''
Room G-19 Lauri Carlson, CSLI
Discussion led by Phil Cohen
(Abstract on page 2)
3:30 p.m. Tea
Ventura Hall
4:15 p.m. CSLI Colloquium
Redwood Hall ``Tracking Dogs and the Traces of Speech''
Room G-19 Vicki Hearne, Yale University
(No abstract available)
!
Page 2 CSLI Newsletter May 9, 1985
←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←
ABSTRACT OF NEXT WEEK'S TINLUNCH
``Combinators, Categorial Grammars, and Parasitic Gaps''
In his previous work, Steedman has applied an augmented version of
Categorial Grammar (CG) to discontiguous dependencies such as
Wh-Fronting and cross-serial dependencies in Dutch. This paper
extends the use of functional composition in CG to permit the
generation of parasitic gaps. The operations on functions that are
allowed in the resulting grammar formalism are defined as combinators.
It is argued that the utilization of a certain class of combinators in
syntax, semantics, and morphology leads to a natural and adequate
extension of CG. --Hans Uszkoreit
←←←←←←←←←←←←
ABSTRACT OF NEXT WEEK'S SEMINAR
``Action Theory for Dialogue Games''
I will describe a theory of social action which generalizes the
game theoretical concept of a game in extensive form. To generalize
the game theoretical model, I make use of a number of analogies
between the game theoretical notion of a game and certain varieties of
possible worlds semantics. The key game theoretical notions are
generalized as follows. Game trees are reinterpreted as branching
future temporal logic frames. Preferences and strategies are
represented by choice functions over possible courses of events.
Information sets are generalized into sets of epistemic alternatives.
The result is a definition of a ``game theoretical'' possible worlds
semantics frame. Sentences describing agents' beliefs, wants,
abilities, and plans can be interpreted against such frames. Desired
interdependencies among the different action-related modalities can be
shown to fall out. Different schemata purporting to characterize
rational action can be evaluated in the resulting semantics using
examples from game theory.
The resulting theory of action is intended to be used to partially
formalize the informal dialogue game approach to discourse theory
described in my book, Dialogue Games (Reidel 1983).
←←←←←←←←←←←←
SEMINAR IN LOGIC AND FOUNDATIONS OF MATHEMATICS
``Quantified Modal Logics of Provability''
Prof. Craig Smorynski, San Jose State University
Tuesday, May 14, 4:15-5:30
Math Bldg., Room 381-T, Stanford University
←←←←←←←←←←←←
AREA NL-1 MEETING
``New Aspects of Aspect: A Look at Mandarin Chinese''
Carlota S. Smith, University of Texas
Friday, May 10, 2:30, Ventura Conference Room
(for the abstract see last week's newsletter)
←←←←←←←←←←←←
AREA NL-2 MEETING
There will be an NL-2 meeting on Tuesday, May 14th at 2.15 in
Redwood Hall in which I will present some of the amendments to
Government- Binding Theory proposed in Chomsky's manuscript
``Knowledge of Language: Its Nature, Origins, and Use.'' The
discussion will cover roughly pages 100-285; all are welcome though a
knowledge of current GB would be useful. --Peter Sells
!
Page 3 CSLI Newsletter May 9, 1985
←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←
AREA NL MEETING
There will be a NL meeting on Friday, May 17th at 2.15 in the
Ventura Conference Room to discuss James Higginbotham's paper, ``On
Semantics''. Copies of the paper will be made available at the front
desk in a few days. --Peter Sells
←←←←←←←←←←←←
NEW CSLI LECTURE NOTES
Number 2 in the Lecture Notes series, ``Emotion and Focus'' by
Helen Nissenbaum, has just appeared. The author describes this work as
follows:
``After examining several prominent views on object directedness,
including those of Hume, Kenny, and J. R. S. Wilson, I conclude that
the notion is no longer a viable one. I propose a reconceptualization
of the phenomena that it is seen to cover. The result is a breakdown
of object directedness into a number of independent conceptual units
that I call `aspects of emotional episodes.' I reject the picture of
emotion traditionally forwarded in academic writings, offering another
in its place, one that preserves the complexity and variation
suggested in the common conception of emotion.''
The list price of ``Emotion and Focus'' is $6, with a 25% discount
to the CSLI community. California residents should add 6.5% sales tax.
To obtain a copy, contact David Brown (Brown@CSLI), CSLI, Ventura
Hall, Stanford, CA 94305.
-------
∂08-May-85 2236 GOLUB@SU-SCORE.ARPA housing wanted
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 8 May 85 22:36:20 PDT
Date: Wed 8 May 85 22:35:52-PDT
From: Gene Golub <GOLUB@SU-SCORE.ARPA>
Subject: housing wanted
To: faculty@SU-SCORE.ARPA
cc: su-bboards@SU-SCORE.ARPA
I have a Scottish friend visiting this summer for two weeks in August. The
period is flexible but something like Aug 5- 16 or Aug 12- 23 would be
possible. He and his wife have two small kids. If you would like to
sublet your place then, please let me know. From previous experience, I know
they would be first rate tenants.
Any help would be appreciated.
GENE
-------
∂09-May-85 0939 NILSSON@SU-SCORE.ARPA Research Support for Entering PhD Students
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 9 May 85 09:39:42 PDT
Date: Thu 9 May 85 09:38:04-PDT
From: Nils Nilsson <NILSSON@SU-SCORE.ARPA>
Subject: Research Support for Entering PhD Students
To: faculty@SU-SCORE.ARPA
There seems to be a great deal of sentiment in the department for moving
toward a system in which first-year PhD students are not forced to find
and associate with research projects so soon. I must say that I think
that a system in which first-year students concentrate more on courses,
preparing for the comps, and carefully looking over the various research
possibilities has a lot to recommend it. Installing such a system would
be feasible financially since so many of the entering students (over
half) have fellowships of one sort or another. We would also need the
cooperation of the various research projects in continuing to support
students during their first year even though those students are not
tightly coupled to the projects. We will need to "associate" students
with projects in some way to satisfy auditing/accounting requirements,
but my idea is that we adopt a rather flexible policy of what it means
for a student to be associated with a project. Association ought to
come to mean that the student is busy preparing himself/herself during
the first-year for future research contributions. It should definitely
not mean that the student mainly spends time doing routine chores around
the project nor that the student is locked into any specific research
project during his/her first year. In short, I am proposing
something like a CMU-type model during the first year in which the
research projects adopt a long-run view of what it means to support
a student. Of course, if a student is ready to begin research immediately,
that's fine, but such a decision should be left up to the student.
I'd be interested in faculty (and student representative) comments about
this matter. -Nils
-------
∂09-May-85 1036 @SU-SCORE.ARPA:ullman@diablo Re: Research Support for Entering PhD Students
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 9 May 85 10:36:05 PDT
Received: from diablo by SU-SCORE.ARPA with TCP; Thu 9 May 85 10:33:07-PDT
Date: Thu, 9 May 85 10:32:09 pdt
From: Jeff Ullman <ullman@diablo>
Subject: Re: Research Support for Entering PhD Students
To: NILSSON@SU-Score, faculty@SU-Score
May I suggest that we try, at least for one year, calling
the support "CSD fellowships." Let's see whether that increases
our yield compared to our current promise of a "research assistantship."
∂09-May-85 1048 JOHN@SU-CSLI.ARPA [John Perry <JOHN@SU-CSLI.ARPA>: next week's session]
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 9 May 85 10:48:33 PDT
Date: Thu 9 May 85 10:46:12-PDT
From: John Perry <JOHN@SU-CSLI.ARPA>
Subject: [John Perry <JOHN@SU-CSLI.ARPA>: next week's session]
To: rrr@SU-CSLI.ARPA
If you already got this, my apologies.
---------------
Date: Wed 8 May 85 10:44:04-PDT
From: John Perry <JOHN@SU-CSLI.ARPA>
Subject: next week's session
To: rrr@SU-CSLI.ARPA
cc: ingrid@SU-CSLI.ARPA
Ingrid will have three (3) additional papers (old) by me that those truly
interested in these things might find worth plowing through, in addition to
more copies of the benevlolent etc. one for thsoe who didn't get one last time.
In order of importance
Perception, action and the structure of believing
A Problem About Continued Belief
Acceptance and Belief
-------
-------
∂09-May-85 1142 @SU-CSLI.ARPA:PCOHEN@SRI-AI.ARPA NL4 Talk Next Tues
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 9 May 85 11:42:41 PDT
Received: from SRI-AI.ARPA by SU-CSLI.ARPA with TCP; Thu 9 May 85 11:34:30-PDT
Date: Thu 9 May 85 11:32:30-PDT
From: Phil Cohen <PCOHEN@SRI-AI.ARPA>
Subject: NL4 Talk Next Tues
To: NL4: ;
Next week, the 14th, at 12:45 - 2:15, Lauri Carlson will continue his
discussion of "Semantics for Dialogue Games". Place, Ventura Seminar
room.
Phil
-------
∂09-May-85 1152 PATASHNIK@SU-SCORE.ARPA aflb today
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 9 May 85 11:52:25 PDT
Date: Thu 9 May 85 11:49:21-PDT
From: Oren Patashnik <PATASHNIK@SU-SCORE.ARPA>
Subject: aflb today
To: aflb.su@SU-SCORE.ARPA
Vaughan Pratt is speaking today, usual time, usual place.
-------
∂09-May-85 1400 BRODER@SU-SCORE.ARPA Late abstracts
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 9 May 85 13:59:56 PDT
Date: Thu 9 May 85 13:56:21-PDT
From: Andrei Broder <Broder@SU-SCORE.ARPA>
Subject: Late abstracts
To: aflb.all@SU-SCORE.ARPA
Score was down and my message got lost. Here are the abstracts for
the last two AFLB talks. - Andrei
5-2-85 - Prof. Nishizeki (Tohoku U., Japan):
"Multicommodity flows in planar undirected graphs and shortest paths"
This paper deals with multicommodity flow problems for two classes of
planar undirected graphs. The first class $C sub 12$ consists of
graphs in which each source-sink pair is loacted on one of two
specified face boundaries. The second class $C sub 01$ consists of
graphs in which some of the specified source-sink pairs are located on
a specified face boundary and all the other pairs share a common sink
located on the boundary. We show that the multicommodity flow problem
for a graph in $C sub 12$ (resp. $C sub 01$) can be reduced to the
shortest path problem for an undirected (resp. a directed) graph
obtained from the dual of the original undirected graph.
***** Time and place: May 2, 12:30 pm in MJ352 (Bldg. 460) ******
5/9/85 - Prof. Vaughan Pratt
"Conic splines"
In a recent CSD colloquium I described conic splines from a systems
perspective. In this talk I will focus on the theoretical basis for
this research. The workhorse splines of the graphics community are
polynomial cubics. I will present results showing that for 2D graphics
most of the worthwhile effects possible with these curves are just as
possible with rational quadratics, i.e. conics, whereas the converse
appears unlikely: rational quadratics are easier both to draw and to
project in perspective.
***** Time and place: May 9, 12:30 pm in MJ352 (Bldg. 460) ******
-------
∂09-May-85 1837 @SU-SCORE.ARPA:JF@SU-SUSHI.ARPA One more BATS abstract for 5/17 meeting
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 9 May 85 18:37:23 PDT
Received: from SU-SUSHI.ARPA by SU-SCORE.ARPA with TCP; Thu 9 May 85 18:34:37-PDT
Date: Thu 9 May 85 18:31:16-PDT
From: Joan Feigenbaum <JF@SU-SUSHI.ARPA>
Subject: One more BATS abstract for 5/17 meeting
To: ragde%ucbernie@UCB-VAX.ARPA, yao.pa@XEROX.ARPA, broder@DECWRL.ARPA,
klawe.ibm-sj@CSNET-RELAY.ARPA, allen.ucsc@CSNET-RELAY.ARPA
cc: aflb.su@SU-SCORE.ARPA
I enclose the third abstract for next friday's (May 17) BATS. It is
in the file
SUSHI:<jf>broder.abstract.
The other abstracts I've gotten so far are in
SUSHI:<jf>brassard.abstract
SUSHI:<jf>siegel.abstract
If you plan to attend bats and have not yet told me, please do so.
Joan
(jf@sushi)
Speaker: Andrei Broder, DECSRC
How hard is it to marry at random?
(On the approximation of the permanent)
The main result is that we can construct a fully polynomial (epsilon, delta)
approximation scheme for the number of perfect matchings in a bipartite
graph G if we have a procedure that given a subgraph H of G finds a perfect
matching in H ``almost'' uniformly at random (in terms of the variation
distance) in polynomial time, and conversely, we can construct such a
procedure if we have a fully polynomial (epsilon, delta) approximation
scheme for the number of perfect matchings in each subgraph of G. Thus
approximating the permanent is equivalent with finding ``almost'' random
matchings.
Markov chains that converge to the uniform distribution seem to be suitable
for the implementation of such sampling procedures. We present a Markov
chain that converges to the uniform distribution on the space of perfect
matchings for any given graph. We show that it converges in polynomial time
on complete graphs and conjecture that it converges fast for all dense
enough graphs. The conjecture implies a fully polynomial (epsilon, delta)
approximation scheme for the permanent of 0-1 matrices where at least half
of the entries in every row and every column are 1's. Finally we show that
the exact computation of the permanent of such matrices is still #P-complete.
-------
∂09-May-85 2231 @SU-SCORE.ARPA:coraki!pratt@Navajo Re: Research Support for Entering PhD Students
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 9 May 85 22:31:11 PDT
Received: from Navajo ([36.8.0.48].#Internet) by SU-SCORE.ARPA with TCP; Thu 9 May 85 22:30:31-PDT
Received: by coraki.uucp (1.1/SMI-1.2)
id AA03722; Thu, 9 May 85 22:26:00 pdt
Date: Thu, 9 May 85 22:26:00 pdt
From: coraki!pratt@Navajo (Vaughan Pratt)
Message-Id: <8505100526.AA03722@coraki.uucp>
To: faculty@su-score.ARPA
Subject: Re: Research Support for Entering PhD Students
In-Reply-To: message of Thu, 9 May 85 10:32:09 pdt.
<8505091907.AA03346@coraki.uucp>
From: Jeff Ullman <ullman@diablo.ARPA>
May I suggest that we try, at least for one year, calling
the support "CSD fellowships."
Here's a more elaborate naming scheme calculated to make even more
people happy:
* Let the students call themselves CSD fellows,
* let the sponsors call them research assistants,
* and let the faculty call them coolies.
-v
∂09-May-85 2348 @SU-SCORE.ARPA:udi@wisc-rsch.arpa ACM mailing
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 9 May 85 23:46:36 PDT
Received: from wisc-rsch.arpa by SU-SCORE.ARPA with TCP; Thu 9 May 85 23:45:30-PDT
Received: by wisc-rsch.arpa; Thu, 9 May 85 23:30:15 cdt
Received: from wisc-crys.arpa by wisc-rsch.arpa; Thu, 9 May 85 09:56:42 cdt
Message-Id: <8505091456.AA14648@wisc-crys.arpa>
Received: from COLUMBIA-20.ARPA by wisc-crys.arpa; Thu, 9 May 85 09:56:31 cdt
Date: Thu 9 May 85 10:54:58-EDT
From: Zvi Galil <GALIL@COLUMBIA-20.ARPA>
Subject: ACM mailing
To: theory@WISC-CRYS.ARPA
Sender: udi@wisc-rsch.arpa
Resent-From: udi@wisc-rsch.arpa (Udi Manber)
Resent-To: theory:;@wisc-rsch.arpa
I heard several complaints from people that had not received the
brochure for STOC (at all).
If you have not received it, please let me know.
Please indicate if you are ACM member or SIGACT member or both.
(If you're neither nor, pls don't respond.)
I already know about IBM San Jose.
Send the message to galil@columbia.
-------
∂10-May-85 0911 YAMARONE@SU-CSLI.ARPA SANDWICH UPDATE
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 10 May 85 09:11:15 PDT
Date: Fri 10 May 85 08:31:27-PDT
From: Tom Yamarone <YAMARONE@SU-CSLI.ARPA>
Subject: SANDWICH UPDATE
To: folks@SU-CSLI.ARPA
Last Wednesday, we switched sandwich vendors from the Nuts and Mud
prefabbed to the Stanford Deli in the shopping center. They are
now made to order and there is a wide variety of sandwiches to choose.
All the standard meats, cheeses and extra special ones like avacado,
sausage , meatballs, etc. I don't have a complete menu for today
but feel free to order anything you might desire, and I'll do my best
to get it. All kinds of bread from w.w. to the ryes to a french roll.
Everything on it includes mustard, mayo, lettuce, tomato, pickle.
Happy ordering and please do it before 10:00 am as that is when I
will call in the order to have the sandwiches made for today.
Have a good morning,
Ventura Sandwich Corp.
-------
∂10-May-85 0958 TAJNAI@SU-SCORE.ARPA Going to Sweden?
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 10 May 85 09:58:25 PDT
Date: Fri 10 May 85 09:56:38-PDT
From: Carolyn Tajnai <TAJNAI@SU-SCORE.ARPA>
Subject: Going to Sweden?
To: Faculty@SU-SCORE.ARPA, csl-faculty@SU-SIERRA.ARPA
If anyone is planning a trip to Sweden, and could make the
Forum visit to Ericsson in Stockholm, please let me know.
Carolyn
-------
∂10-May-85 1017 MARJORIE@SU-CSLI.ARPA Thefts
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 10 May 85 10:17:28 PDT
Date: Fri 10 May 85 10:11:03-PDT
From: Marjorie Maxwell <MARJORIE@SU-CSLI.ARPA>
Subject: Thefts
To: Folks@SU-CSLI.ARPA
cc: consultants@SU-CSLI.ARPA
A warning - there have been a number of thefts of small items (Walkmans)
over at Pine Hall recently. If you have any similar items better keep
an eye on them or locked up. A stranger who looks not too out of place may
be the culprit -so beware!
-------
∂10-May-85 1029 @SU-SCORE.ARPA:cheriton@Pescadero Re: Research Support for Entering PhD Students
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 10 May 85 10:29:19 PDT
Received: from Pescadero by SU-SCORE.ARPA with TCP; Fri 10 May 85 10:25:10-PDT
Date: Fri, 10 May 85 10:25:08 pdt
From: David Cheriton <cheriton@Pescadero>
Subject: Re: Research Support for Entering PhD Students
To: faculty@score
This sounds like a great idea providing that another aspect of the CMU model
is also recognized, namely that the senior faculty are providing almost
all of the funds for this pool. In fact, the full CMU model is fairly
paternal, as I understand, with junior faculty also being taken care of by
this umbrella funding. In terms of losing people, we havent had a great
record with hiring junior faculty (in part because of support issues) either.
Certainly, I dont feel that (with my relatively short track record with
DARPA and other funding agencies) that I am given much latitude to provide
support with no return although I will do what I can.
∂10-May-85 1220 @SU-CSLI.ARPA:halvorsen.pa@Xerox.ARPA Reminder: Today at 2:30 pm: Carlota S. Smith: New Aspects of
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 10 May 85 12:20:32 PDT
Received: from Xerox.ARPA by SU-CSLI.ARPA with TCP; Fri 10 May 85 12:18:30-PDT
Received: from Salvador.ms by ArpaGateway.ms ; 10 MAY 85 12:13:39 PDT
Date: 10 May 85 12:12 PDT
From: halvorsen.pa@Xerox.ARPA
Subject: Reminder: Today at 2:30 pm: Carlota S. Smith: New Aspects of
Aspect...
To: NL1@SU-CSLI.ARPA,NLInterest@Su-CSLI.ARPA
New Aspects of Aspect: A Look at Mandarin Chinese
Carlota S. Smith
University of Texas
Date: Friday, May 10 Time: 2:30 pm
Place: Ventura Hall Seminar Room
∂10-May-85 1315 PAPA@SU-SCORE.ARPA Re: Same Generation and NC
Received: from SU-AIMVAX.ARPA by SU-AI.ARPA with TCP; 10 May 85 13:11:45 PDT
Received: from SU-SCORE.ARPA by diablo with TCP; Fri, 10 May 85 13:00:38 pdt
Date: Fri 10 May 85 13:00:48-PDT
From: C. Papadimitriou <PAPA@SU-SCORE.ARPA>
Subject: Re: Same Generation and NC
To: avg@SU-AIMVAX.ARPA
Cc: nail@SU-AIMVAX.ARPA
In-Reply-To: Message from "Allen VanGelder <avg@diablo>" of Mon 6 May 85 18:50:56-PDT
I was told by Paris Kanellakis that ``linear recursion'' is in NC.
I know of no details. That was yesterday.
---Christos.
-------
∂10-May-85 1322 @SU-SCORE.ARPA:udi@wisc-rsch.arpa Referee reports
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 10 May 85 13:22:28 PDT
Received: from wisc-rsch.arpa by SU-SCORE.ARPA with TCP; Fri 10 May 85 13:19:20-PDT
Received: by wisc-rsch.arpa; Fri, 10 May 85 15:00:46 cdt
Date: Fri, 10 May 85 14:50:46 cdt
From: bent@wisc-rsch.arpa
Message-Id: <8505101950.AA22603@wisc-rsch.arpa>
Received: by wisc-rsch.arpa; Fri, 10 May 85 14:50:46 cdt
To: theory@uwvax
Subject: Referee reports
Sender: udi@wisc-rsch.arpa
Resent-From: udi@wisc-rsch.arpa (Udi Manber)
Resent-To: theory:;@wisc-rsch.arpa
My backpack vanished recently; I presume it's gone forever.
Among other valuable things, it contained my stack of papers to referee.
I don't remember what they were, or who sent them to me.
If I owe you a referee report, you'll have to send me a copy of the
paper and start all over again.
- Sam
∂10-May-85 1343 ullman@diablo Re: Same Generation and NC
Received: from SU-AIMVAX.ARPA by SU-AI.ARPA with TCP; 10 May 85 13:43:13 PDT
Date: Fri, 10 May 85 13:31:34 pdt
From: Jeff Ullman <ullman@diablo>
Subject: Re: Same Generation and NC
To: PAPA@SU-Score, avg@diablo
Cc: nail@diablo
There is an easy construction that every linear recursion is in NC,
assuming no function symbols.
Let D be the domain of all symbols mentioned by your relations.
Construct a graph whose nodes correspond to R(a1,...,an), where R
is a recursive predicate and the ai's are in D.
There is an edge from R(a1,...an) to S(b1,...,bm) if there are
tuples for the nonrecursive predicates that let you infer
the latter from the former (it is easy to build the nonrecursive
predicates in parallel, so we only have to consider rules of
the form S :- R, (nonrecursive stuff)).
Solve the problem by taking the transitive closure of the graph in
parallel.
D is no greater than the size of the data, and the graph size
is polynomial in ||D||.
The interesting problem is when a nonlinear recursion is in NC.
Surely not all are, because you can simulate circuit value.
For that matter, when can you convert a nonlinear recursion to
a linear one? For example, hte well known
T(x,y) :- A(x,y).
T(x,y) :- T(x,z), T(z,y).
Can be replaced, using a process like Greibach-normal-form-ization by
T(x,y) :- A(x,y).
T(x,y) :- A(x,z), T(z,y).
---Jeff Ullman
∂10-May-85 1617 NILSSON@SU-SCORE.ARPA Prize
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 10 May 85 16:17:12 PDT
Date: Fri 10 May 85 16:04:41-PDT
From: Nils Nilsson <NILSSON@SU-SCORE.ARPA>
Subject: Prize
To: csd@SU-SCORE.ARPA
Intel Corporation has awarded Stanford an "iPDS system" as a prize for
the winning team in the student programming competition. I now have
it in my office (thanks to the generousity of the student winners!).
My question now is who might like to use this system? Here is a brief
description of it excerpted from its manual:
"The iPDS system supports the design and development of products that
incorporate Intel microprocessors or microcontrollers.
The system and its options aid in both hardware and software development for
products based on many different families of chips, such as the following:
* MCS-51 microcontroller family
* MCS-85 general purpose microprocessor family
* iAPX-88 general purpose microprocessor family
TYPICAL USES:
By incorporating microprocessors from these families, products can range in
complexity from a simple process controller to an advanced microcomputer
system. Software can range from a single-purpose control program to a complex
software system.
The iPDS system is useful at all stages of product design from the initial
idea to customer support after the product is in the field.
MAJOR CHARACTERISTICS:
The development system supports integrated hardware and software development
by assembling or compiling source programs for execution and by emulating
the target microprocessor, the processor used in the product.
The system optionally includes a PROM Programmer for programming EPROMS as
well as E2PROMs to store software in the target processor's memory.
Another feature of the development system is it portability; the basic
system weighs only 29 pounds and has a handle for carrying."
The system is "IBM-PC compatible" (I think) and includes a variety of
common personal computer software--so it could be used as a PC.
If anyone out there could make good use of this machine, please let me
know. -Nils
-------
∂11-May-85 2024 SARAIYA@SUMEX-AIM.ARPA Release 6 Systems
Received: from SUMEX-AIM.ARPA by SU-AI.ARPA with TCP; 11 May 85 20:24:22 PDT
Date: Sat 11 May 85 20:23:49-PDT
From: Nakul P. Saraiya <SARAIYA@SUMEX-AIM.ARPA>
Subject: Release 6 Systems
To: aap@SUMEX-AIM.ARPA
P-Helios, Simple, XCare, and Care are up on Rel. 6.
-Nakul
-------
∂11-May-85 2155 ullman@diablo Hot Off the Wire
Received: from SU-AIMVAX.ARPA by SU-AI.ARPA with TCP; 11 May 85 21:55:14 PDT
Date: Sat, 11 May 85 21:51:18 pdt
From: Jeff Ullman <ullman@diablo>
Subject: Hot Off the Wire
To: nail@diablo
I received the following message, which may be of interest.
*************************************************************
Return-Path: <PARIS@MIT-XX.ARPA>
Received: from MIT-XX.ARPA by SU-SCORE.ARPA with TCP; Fri 10 May 85 14:59:35-PDT
Date: Fri 10 May 85 18:00:31-EDT
From: Paris Kanellakis <PARIS@MIT-XX.ARPA>
Subject: SiRuPs
To: ullman@SU-SCORE.ARPA
Dear Jeff,
Although I am on a different coast I would love to participate in
your sessions on logic programming.
We are also holding some parallel discussions here with Stavros.
I pointed out to Christos,
what looked to me like a simple observation I made about linear recursion
that would justify much of the practical interest in it,
(in fact one of the first folk theorems in an emerging area).
I am enclosing some more observations, which also set up a framework
for thinking about these things
the framework of SiRuPs, sweet and bitter ones.
I wonder how the NAIL group feels about using this New England term.
Best Regards
Paris
------------------------------------------------------------------------
On the classification of SiRuPs
THE SINGLE RULE PROBLEM------------------------------------
THE QUERY LANGUAGE:
Let our query language consist of a single predicate symbol
R of arity m, and variables a,a1,...,b,b1,...,c,c1,...
(i.e. no function symbols, no constants, only one possible
relation in the database).
Let our database consist of a single relation r with m columns.
In our language we write a SINGLE RULE, which is in fact a Horn rule.
Here are some syntactic restrictions on a SINGLE RULE.
Variables appearing on left hand side of the rule are DISTINGUISHED,
the rest of the variables are UNDISTINGUISHED.
We will use a,b,c... for DISTINGUISHED variables only.
If the variables appearing in the first position of R are
from a,a1,..., in the second position of R from b,b1,... etc
then the rule is TYPED, else it is UNTYPED.
TYPED rules, where all DISTINGUISHED variables appear also
on right hand side of rule can be thought of as TABLEAUx
of [Aho, Sagiv, Ullman] with full summary.
With each one of these TABLEAUx we have the natural tableau mapping T;
we are interested in solving equations of the form r=T(r).
Some of these tableau are disguised project-natural←join expressions,
although not all tableau can be expressed in this way.
COMPUTATION: Given the SINGLE RULE and an initial relation r := r0, we apply
the rule to r. This can only result in adding tuples to r.
We wish to construct the smallest fixpoint solution to the SINGLE RULE,
which contains the initial relation r0.
COMPLEXITY: The asymptotic part is the size of r0 (i.e., n), we assume the
single rule is of fixed size (e.g., m is fixed in this analysis).
That is, we examine each SINGLE RULE problem separately.
Intuitively this corresponds to
compiling the rule into a "good" algorithm for all possible r0's.
-----This notion was defined by Moshe Vardi as Data Complexity.------
MAIN PROBLEM: For each SINGLE RULE problem there is an algorithm
to compute the smallest fixpoint containing r0 (of size n), which works
in time polynomial in n. This is the "naive algorithm" of iterating
the application of the rule on the whole relation, n to the power m times.
We would like to identify:
I) The SINGLE RULE problems for which the "naive algorithm" involves
at most k>=1 (where k is a constant) iterations.
II) The SINGLE RULE problems for which the "naive algorithm" involves
at most O(log n) iterations.
III) The SINGLE RULE problems for which the fixpoint can be evaluated
in the parallel class NC (polylog in n parallel time,
polynomial in n amount of hardware).
IV) The SINGLE RULE problems, which are logspace complete in PTIME
and therefore probably not in NC.
I,II,III are sweet SiRuPs and IV are bitter ones
----------------------------------------------------------------------
OBSERVATIONS ON THE SINGLE RULE PROBLEMS
Example 1: R(a,b) :- R(a,c1), R(c1,b) UNTYPED
This rule guarantees closure under transitivity.
Because the size of paths covered each time in the directed
graph, which w.l.o.g. represents relation r, doubles with each iteration,
this rule is in case II above. Obviously it is not in I.
---------------
Example 2: R(a,b) :- R(a,b1), R(a1,b1), R(a1,b) TYPED
corresponding to tableau
a b
----
a b1
a1 b1
a1 b
This rule has the same complexity properties as example 1. It is
the smallest tableau which is not a disguised project-natural←join
expression. It computes the sum in the lattice of partitions
THIS IS ONE RECURSION HANDLED BY PARTITION DEPENDENCIES IN
OUR PODS PAPER.
----------------
Example 3: R(a,b) :- R(b,a) UNTYPED
This rule is closure under symmetry. As all rules without
UNDISTINGUISHED variables it is in case I.
----------------
Example 4: R(a,b) :- R(c1,c1), R(b,a) UNTYPED
This rule is closure under symmetry, provided the initial relation contains
an <x,x> pair. It is in case I; one iteration suffices; it is different
than example 3.
----------------
Example 5: R(a,b) :- R(a,c1), R(c1,a), R(b,a) UNTYPED
This rule is in case III (one must reason about paths here),
using a simple r0 example one can demonstrate that it is not in II.
←←←←←←←←←←←←←←←←
Example 6: R(a,b,c) :- R(a,b,c1), R(a,b,c2), R(c1,c2,c) UNTYPED
This rule codes FD implication (together with an ingenious r0).
For this one can prove it is in IV.
←←←←←←←←←←←←←←←←
Example 7: R(a,b,c) :- R(a,b,c1), R(a1,b,c2), R(a1,b1,c) TYPED
corresponding to tableau
a b c
--------
a b c1
a1 b c2
a1 b1 c
This rule corresponds to the following algebraic equation
r = AB(r) * BC( AB(r) * AC(r) ).
It is not in case I for k=1, (in fact for no fixed k).
---------------------
Example 8: R(a,b,c) :- R(a,b,c1), R(a1,b,c2), R(a1,b1,c), R(a2,b,c) TYPED
corresponding to tableau
a b c
--------
a b c1
a1 b c2
a1 b1 c
a2 b c
This rule corresponds to the following algebraic equation
r = AB(r) * BC(r) * BC( AB(r) * AC(r) ).
Since this expression is equivalent to
r = AB(r) * BC(r) (see next example 9)
it is in case I with k=1 (by a well known idempotence property).
------------------
Example 9: R(a,b,c) :- R(a,b,c1), R(a2,b,c) TYPED
corresponding to tableau
a b c
--------
a b c1
a2 b c
Simplify previous case.
--------------------
Example 10: R(a,b) :- R(a,c1), R(c1,b), R(c1,c2) UNTYPED
This is really a "disguised" form of example 1.
--------------------
Example 11: R(a,b) :- R(a,c1), R(c1,b), R(c1,c2), R(c2,c3) UNTYPED
This computes something different from example 1, so one must
be careful about what constitutes a "disguise", as was that of example 10.
-------------------
OBSERVATION ON SIRUP CASES:
Case I is properly included in case II,
case II is properly included in case III.
Case IV is nonempty.
(proof by previous examples)
--------------------
k-SIRUP OBSERVATION:
If we are given a number k and a SiRuP,
one can decide (using an exponential construction) if and only if
this SiRuP is in case I.
(proof uses simple homomorphism argument)
----------------------
A RELEVANT THEOREM BY SAGIV:
Consider the restricted case of TABLEAU.
The T that belong to case I with k=1 (idempotent tableaux)
are exactly the ones that are disguised expressions of
the form: a join of projections.
The proof of the only if direction is non-trivial.
"On Computing Restricted Projections of Representative Instances"
Y. Sagiv, 4th PODS, pages 171-180, Theorem 3.1.
-------------------
SIMPLE-RULE OBSERVATION:
Consider a single recursively defined predicate R(...) and any number
of non-recursive predicates Q1(...),Q2(...),...
A set of rules is simple if the consequent of each rule is R(...),
and among the antecedents of each rule there is at most one R(...).
Then evaluating the least fixpoint containing the initial database is
in NC (case III). Moreover, one can easily write an equivalent set of
rules in case II.
(This is really what Jeff calls linear recursion, many rules
as opposed to SiRuPs but of this very restricted form).
--------------------------------------
∂12-May-85 1440 BRODER@SU-SCORE.ARPA No AFLB this week
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 12 May 85 14:40:00 PDT
Date: Sun 12 May 85 14:39:46-PDT
From: Andrei Broder <Broder@SU-SCORE.ARPA>
Subject: No AFLB this week
To: aflb.all@SU-SCORE.ARPA
No meeting this Thursday (May 16). Enjoy BATS on May 17. - Andrei
-------
∂13-May-85 0832 EMMA@SU-CSLI.ARPA talk by shankar, weds, 4:15, in EJ232, on Mechanical Proofs in Metamathematics
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 13 May 85 08:32:26 PDT
Return-Path: <WALDINGER@SRI-AI.ARPA>
Received: from SRI-AI.ARPA by SU-CSLI.ARPA with TCP; Fri 10 May 85 17:15:07-PDT
Date: Fri 10 May 85 17:08:40-PDT
From: WALDINGER@SRI-AI.ARPA
Subject: talk by shankar, weds, 4:15, in EJ232, on Mechanical Proofs in Metamathematics
To: AIC-Associates: ;,
CSL: ;, bboard@SRI-AI.ARPA, su-bboards@SU-SCORE.ARPA, friends@SU-CSLI.ARPA
cc: cl.shankar@UTEXAS-20.ARPA
ReSent-Date: Mon 13 May 85 08:27:46-PDT
ReSent-From: Emma Pease <Emma@SU-CSLI.ARPA>
ReSent-To: friends@SU-CSLI.ARPA
That's May 15th.
This is different from Shankar's talk at Stanford.
Abstract below:
MECHANICAL PROOFS IN METAMATHEMATICS
N. Shankar
University of Texas at Austin
Austin, Texas
Metamathematics is a source of many interesting theorems and difficult
proofs. Among these are theorems which express the soundness of
derived inference rules. The talk will describe a formalization of
first-order logic within the Boyer-Moore logic and discuss some
well-known derived inference rules that were proved to be sound using
the Boyer-Moore theorem prover. The most important of these is the
tautology theorem which states that every tautology has a proof. The
proof of the tautology theorem will be discussed in some detail. Such
proofs demonstrate a feasible way to construct sound, efficient, and
extensible proof-checking programs. No familiarity with automated
theorem-proving will be assumed.
Visitors from outside SRI should come to the Engineering building
reception, on Ravenswood Avenue opposite Pine street.
-------
∂13-May-85 0844 HEDGES@SU-SCORE.ARPA Faculty Lunch
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 13 May 85 08:44:44 PDT
Date: Mon 13 May 85 08:36:33-PDT
From: Karen Hedges <HEDGES@SU-SCORE.ARPA>
Subject: Faculty Lunch
To: Faculty@SU-SCORE.ARPA
REMINDER: The Faculty Lunch will be held tomorrow, May 14 at 12:15 in
Conference Room 146. The topic for discussion will be
"CSD Committees--Is the Present System Working?"
-------
∂13-May-85 0858 HEDGES@SU-SCORE.ARPA Visit of Peter Will
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 13 May 85 08:58:17 PDT
Date: Mon 13 May 85 08:50:49-PDT
From: Karen Hedges <HEDGES@SU-SCORE.ARPA>
Subject: Visit of Peter Will
To: CSD@SU-SCORE.ARPA
Peter Will who was scheduled to visit on Wednesday, May 15 with various
department members, has postponed his visit due to illness. His visit
will be rescheduled at a later date.
Karen Hedges
-------
∂13-May-85 0949 NILSSON@SU-SCORE.ARPA New CSD Curric Committee
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 13 May 85 09:48:14 PDT
Date: Mon 13 May 85 09:25:22-PDT
From: Nils Nilsson <NILSSON@SU-SCORE.ARPA>
Subject: New CSD Curric Committee
To: csd@SU-SCORE.ARPA
cc: gibbons@SU-SIERRA.ARPA, white@SU-SIERRA.ARPA, cr.jnr@SU-FORSYTHE.ARPA
Announcement About a New Curriculum Committee
Don Knuth has suggested that the "turn-over" time for the CSD curriculum
committee be in Winter Quarter because that is the time that course
descriptions must be submitted to "Courses and Degrees." Don and his
committee have done an outstanding job, and I wish that I could talk him
into another year, but he will be on sabbatical next year.
I have asked the following people to be on the new curriculum committee
and they have agreed: Jeff Ullman (chair), Leo Guibas, Ernst Mayr, Brian
Reid, Mike Genesereth, Forest Baskett, and Stuart Reges. In addition,
Jeff will consult with the student representatives about student
member(s). The committee will begin functioning immediately and will
continue in its present form until sometime during next Winter Quarter.
At that time, in order to provide continuity, I expect that I will ask
four or five of the present members to form the core (and chair) of the
new committee.
The main task for the committee this year will be to suggest a curriculum
for an undergraduate major in computer science. (Although the CSD has
not yet officially decided to offer such a major, I would like that
decision to be made in the context of having a specific suggestion about a
curriculum.) Besides designing a curriculum, the committee should
estimate the resources needed to staff and equip the major. I would
also be interested in any suggestions the committee might have that
would allow us to begin the major with minimum disruption of the
graduate program and of research activities.
The committee should also serve as liaison between the CSD and other
departments in the University who might want to participate in special
undergraduate programs with a computer orientation. The Math and
Computational Sciences program and the proposed CSD/EE "CSE" program are
examples.
The curriculum committee will continue to have its usual role in
maintaining and improving our entire CS curriculum--both graduate and
undergraduate courses. Suggestions for new courses should be
coordinated through the committee chairman, Jeff Ullman.
I would like the committee to have its proposal for the undergraduate CS
major ready early in Fall Quarter, 1985, so that the entire CSD faculty
will be able to consider it in time for the relevant academic bodies to
approve a 1986 announcement.
-Nils Nilsson
-------
∂13-May-85 1102 ULLMAN@SU-SCORE.ARPA Undergraduate program
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 13 May 85 11:01:45 PDT
Date: Mon 13 May 85 10:44:00-PDT
From: Jeffrey D. Ullman <ULLMAN@SU-SCORE.ARPA>
Subject: Undergraduate program
To: faculty@SU-SCORE.ARPA
The curriculum committee is about to consider the design of an
undergraduate program in Computer Science. A likely outline of such
a program is that CS108ABC will serve as a universal prerequisite
to more advanced courses. Recall the new 108 includes not only the
old 108, but material from CS111 and CS142 (I use the old numbering
scheme throughout).
Beyond 108 we have a choice of creating new courses or adopting
existing courses. To start our discussion, I would like to hear
opinions from those who have taught, or are concerned with,
our "mezzanine" level courses, among which I count
137A, 143, 145, 146, 148, 154, 155, 161, 157A, 161, and 223;
there may be others as well.
Which of these courses can admit undergrads with only CS108
as background? Should we create separate UG courses or
sections with the same content? Should a new UG course in
the same area be created with a modified content? How modified?
Do we need to introduce material not now covered in our offerings?
Should content of several courses be recombined in some way?
Please send me electronic or hardcopy comments on these issues
for any of the courses with which you are familiar.
Other comments are also welcome, and all responses will be
shared with the committee.
---Jeff Ullman
-------
∂13-May-85 1125 vardi@diablo Re: Same Generation and NC
Received: from SU-AIMVAX.ARPA by SU-AI.ARPA with TCP; 13 May 85 11:25:44 PDT
Date: Mon, 13 May 85 11:20:51 pdt
From: Moshe Vardi <vardi@diablo>
Subject: Re: Same Generation and NC
To: PAPA@SU-Score, avg@diablo
Cc: nail@diablo
What is linear recursion?
Moshe
∂13-May-85 1332 @SU-CSLI.ARPA:PCOHEN@SRI-AI.ARPA Meeting tomorrow, the 14th
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 13 May 85 13:31:55 PDT
Received: from SRI-AI.ARPA by SU-CSLI.ARPA with TCP; Mon 13 May 85 13:25:13-PDT
Date: Mon 13 May 85 13:27:20-PDT
From: Phil Cohen <PCOHEN@SRI-AI.ARPA>
Subject: Meeting tomorrow, the 14th
To: NL4: ;
A quick reminder,
Tomorrow, Tues, the 14th, from 12:45 - 2:15, Ventura Seminar room,
Lauri Carlson will continue his discussion of "Semantics for Dialogue
Games". See you there,
Phil
-------
∂13-May-85 1343 EMMA@SU-CSLI.ARPA This Thursday's Colloquium
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 13 May 85 13:43:15 PDT
Mail-From: EMMA created at 13-May-85 13:31:34
Date: Mon 13 May 85 13:31:34-PDT
From: Emma Pease <Emma@SU-CSLI.ARPA>
Subject: This Thursday's Colloquium
To: friends@SU-CSLI.ARPA
Tel: 497-3479
ReSent-Date: Mon 13 May 85 13:32:08-PDT
ReSent-From: Emma Pease <Emma@SU-CSLI.ARPA>
ReSent-To: friends@SU-CSLI.ARPA
From: Richard Jeffrey <JEFFREY@SU-CSLI.ARPA>
Subject: Vicki Hearne
Extended examination of the language by means of which tracking dog handlers
think about their work and actually do work their dogs. What's at stake is an
initial question about whether there is any knowledge that language can
provide that can go outside of the exchanges that are our performance of
knowledge. I use the case of tracking dogs to challenge the skeptic's sense
that the problem of the other is the problem of knowing the other, not in
order foolishly to attempt to bring skepticism to an end but rather to suggest
a new turn skepticism might take. But skepticism must admit that dogs exist
in order to find itself again.
Vicki Hearne is a poet and an animal trainer (dogs, horses). She
teaches at Yale. Has published two books of poetry and a couple of
philosophical articles on our relations with animals (in the Raritan
Review: "How to say `fetch!'" and "A walk with Washowe - How far can
we go?"). She has been much influenced by (you should pardon the
expression) Stanley Cavell. I find her ideas very sympatico, and
especially interesting because of the difference between our
philosophical slants - as well as because her ideas are informed by a
great deal of experience with animals (whereas my relations with our
family cat, even, are correct but distant). Like me, she's in polar
opposition to (say) Donald Davidson in the matter of the
intelligibility of attributions of thoughts (beliefs, desires) to our
furry friends but, unlike me, she has a really nuanced of what these
friends are on about.
--Richard Jeffrey
{Copies of the two articles mentioned above are in the CSLI reading room}
-------
-------
∂13-May-85 1444 ullman@diablo next meetings
Received: from SU-AIMVAX.ARPA by SU-AI.ARPA with TCP; 13 May 85 14:44:09 PDT
Date: Mon, 13 May 85 14:40:14 pdt
From: Jeff Ullman <ullman@diablo>
Subject: next meetings
To: nail@diablo
At long last, Joe Skudlarek will talk about QUEL* on 5/15.
On 5/23 Richard Treitel will talk about MRS as an implementation
route for NAIL!.
Beyond that--we need volunteers.
∂13-May-85 1631 EMMA@SU-CSLI.ARPA Hausser talk
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 13 May 85 16:31:14 PDT
Date: Mon 13 May 85 16:18:01-PDT
From: Emma Pease <Emma@SU-CSLI.ARPA>
Subject: Hausser talk
To: friends@SU-CSLI.ARPA
Tel: 497-3479
Tomorrow (probably today by the time you read this, Roland Hausser
will give a talk and demonstration of a system of grammar and parsing
he has been working on during his stay at CSLI. The presentation will
take place from in the CSLI trailer Classroom 1 till 1:45 on May 14th.
He will illustrate his system by applying it to German. The
grammar contains 32 linguistic rules covering declarative and
interrogative main clauses, relative clauses embedded to
arbitrary depth or extraposed, variant word-orders, adverbs and
adverbial clauses, discontinuous constituents, coordinate
structures, and other phenomena. The parser employs a bottom-up,
left-associative, data-driven algorithm and is implemented in
Interlisp-D on the Dandelion.
-------
-------
∂13-May-85 1651 @SU-CSLI.ARPA:BrianSmith.pa@Xerox.ARPA John Perry tomorrow (May 14)
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 13 May 85 16:51:20 PDT
Received: from Xerox.ARPA by SU-CSLI.ARPA with TCP; Mon 13 May 85 16:44:19-PDT
Received: from Salvador.ms by ArpaGateway.ms ; 13 MAY 85 16:46:57 PDT
Date: 13 May 85 16:46 PDT
From: BrianSmith.pa@Xerox.ARPA
Subject: John Perry tomorrow (May 14)
To: rrr@su-csli.ARPA
cc: BrianSmith.pa@Xerox.ARPA
Reminder: John Perry will be leading the RRR group tomorrow, continuing
the discussion of benevolence and related matters.
John sent out a note with the following additional readings that people
might be interested in (as well as the benevolence one). Available from
Ingrid; listed in order of importance:
Perception, action and the structure of believing
A Problem About Continued Belief
Acceptance and Belief
Usual time and place: 2:15 in the Ventura seminar room.
Brian
∂13-May-85 1810 vardi@diablo Nothing new under the sun
Received: from SU-AIMVAX.ARPA by SU-AI.ARPA with TCP; 13 May 85 18:10:03 PDT
Date: Mon, 13 May 85 18:05:46 pdt
From: Moshe Vardi <vardi@diablo>
Subject: Nothing new under the sun
To: nail@diablo
When Dave Maier was here last year, we observed that linear recursion is in
NC, and raised the question of identifying cases of nonlinear recursion
that are in NC. Eli Upfal and I tried unsuccessfully to crack the problem,
and couldn't come up with any interesting cases. I'm pleased to see the issue
rise from the ashes.
Moshe
∂13-May-85 1848 @SU-SCORE.ARPA:JF@SU-SUSHI.ARPA bats schedule
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 13 May 85 18:48:24 PDT
Received: from SU-SUSHI.ARPA by SU-SCORE.ARPA with TCP; Mon 13 May 85 18:46:09-PDT
Date: Mon 13 May 85 18:44:37-PDT
From: Joan Feigenbaum <JF@SU-SUSHI.ARPA>
Subject: bats schedule
To: ragde%ucbernie@UCB-VAX.ARPA, yao.pa@XEROX.ARPA, broder@DECWRL.ARPA,
klawe.ibm-sj@CSNET-RELAY.ARPA, allen.ucsc@CSNET-RELAY.ARPA
cc: mbern%ucbernie@UCB-VAX.ARPA, aflb.su@SU-SCORE.ARPA
The schedule for bats this friday, May 17 is:
10:00 Ernst Mayr (Stanford) 2-Proc. Scheduling is in NC
11:00 Andrei Broder (DECSRC) How Hard is it to Marry at Random?
12:00 Lunch
1:00 If there are any: open problems and solutions to past open
problems
1:30 Gilles Brassard Generating Random Perms. on the Fly
(U. of Montreal/Berkeley)
2:30 Alan Siegel (NYU) Lower Bounds for Optimal VLSI Sorters
Driving and Parking directions tomorrow.
Joan
-------
∂13-May-85 1904 @SU-SCORE.ARPA:JF@SU-SUSHI.ARPA final abstract
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 13 May 85 19:04:44 PDT
Received: from SU-SUSHI.ARPA by SU-SCORE.ARPA with TCP; Mon 13 May 85 18:54:46-PDT
Date: Mon 13 May 85 18:51:58-PDT
From: Joan Feigenbaum <JF@SU-SUSHI.ARPA>
Subject: final abstract
To: ragde%ucbernie@UCB-VAX.ARPA, yao.pa@XEROX.ARPA, broder@DECWRL.ARPA,
klawe.ibm-sj@CSNET-RELAY.ARPA, allen.ucsc@CSNET-RELAY.ARPA
cc: mbern%ucbernie@UCB-VAX.ARPA, aflb.su@SU-SCORE.ARPA
here is the fourth and final abstract for friday's bats
Ernst Mayr, Stanford:
"Two Processor Scheduling is in NC"
The unit execution time scheduling problem is one of the most fundamental
scheduling problems. It consists of n tasks with precedence constraints among
them which have to be scheduled on m identical processors in such a way that
the precedence constraints are not violated, and the total length of the
schedule is minimized. Every task has to be executed without interruption and
takes unit time.
If the number of processors, m, is part of the problem instance, the problem
is a well-known NP-complete problem. For m=2, efficient algorithms have been
known for some time, the asymptotically best running in linear time.
At a recent BATS (and STOC last week), Kozen/Vazirani↑2 presented an RNC
algorithm for this problem (i.e., a parallel algorithm running in polylog time
on a polynomial number of processors, which uses random coin flips). Their
algorithm is based on the Karp/Upfal/Wigderson algorithm for maximum
cardinality matching which is also in RNC.
We show that an optimal two processor schedule can actually be found by a
deterministic NC algorithm. Our algorithm is based on a fast procedure to
obtain the length of an optimal schedule. However, as it turned out, it is not
at all trivial to obtain an actual optimal schedule, given the length
procedure. This provides another example for an apparent complexity gap
between decision problems in NC and corresponding function problems. Our two
processor scheduling algorithm also solves the maximum cardinality matching
problem for a restricted class of graphs.
This represents joint work with Dave Helmbold.
-------
for stanford people: all four abstracts are in the directory
SUSHI:<JF>
in four separate files named speaker.abstract, where speaker \in {mayr,
broder, brassard, siegel}
-------
∂13-May-85 2050 ullman@diablo Re: Nothing new under the sun
Received: from SU-AIMVAX.ARPA by SU-AI.ARPA with TCP; 13 May 85 20:50:28 PDT
Date: Mon, 13 May 85 20:47:49 pdt
From: Jeff Ullman <ullman@diablo>
Subject: Re: Nothing new under the sun
To: nail@diablo, vardi@diablo
∂13-May-85 2231 DAVIES@SUMEX-AIM.ARPA Architectures meeting -- 5/15/85, 9 am
Received: from SUMEX-AIM.ARPA by SU-AI.ARPA with TCP; 13 May 85 22:31:10 PDT
Date: Mon, 13 May 1985 22:29 PDT
Message-ID: <DAVIES.12110850701.BABYL@Sumex>
From: DAVIES@Sumex
To: AAP@Sumex
Subject: Architectures meeting -- 5/15/85, 9 am
Horace Flatt
IBM Scientific Center
Palo Alto
Parallel Computing: Can you have your cake and eat it too?
A simple model for parallel processing has been developed which shows
under fairly general conditions that limitations exist on the
effectiveness of tying together large numbers of processors to work on a
given problem. Some of the consequences of this model will be discussed
along with numerical illustrations.
∂13-May-85 2331 SCHOEN@SUMEX-AIM.ARPA Rel 6 progress
Received: from SUMEX-AIM.ARPA by SU-AI.ARPA with TCP; 13 May 85 23:31:09 PDT
Date: Mon 13 May 85 23:29:25-PDT
From: Eric Schoen <Schoen@SUMEX-AIM.ARPA>
Subject: Rel 6 progress
To: aap@SUMEX-AIM.ARPA
CZAR and ELINT now appear to be functional under release 6. I observed
a number of unexpected Lisp behaviors:
* FUNCALL of a function defined by DEFUN-METHOD doesn't quite
work. "Free" reference of instance variables does not resolve
properly.
* (FUNCTIONP <symbol> T), which should return T if <symbol> is
a macro or special form, returns NIL instead. I suspect this
was how they decided to prevent FUNCALL of special forms or
macros; it's hard to tell, since the associated source code
is part of the set they're no longer distributing.
So, with appropriate "caveat programmer," I move we consider upgrading
the remaining two machines to release 6.
Eric
-------
∂14-May-85 0739 @SU-CSLI.ARPA:RCOOPER@SRI-AI.ARPA job in Sweden
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 14 May 85 07:39:31 PDT
Received: from SRI-AI.ARPA by SU-CSLI.ARPA with TCP; Tue 14 May 85 07:35:53-PDT
Date: Tue 14 May 85 07:38:58-PDT
From: Robin Cooper <RCOOPER@SRI-AI.ARPA>
Subject: job in Sweden
To: researchers@SU-CSLI.ARPA
Visiting Position in AI
Lund University
The Computer Science department at Lund University, Sweden, is planning to
develop a program in AI. During this planning stage, they would like to
invite a guest professor... They are primarily interested in someone who has
experience in AI work and who could provide inspiration for the program Low
teaching load, time negotiable, possibly Spring semester 1986.
For further details, contact
Christian Collberg
Department of Computer Science,
Box 725
S-220 07 Lund, Sweden
-------
∂14-May-85 0909 MACKAYE@SU-CSLI.ARPA STANFORD OUT OF SOUTH AFRICA
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 14 May 85 09:08:57 PDT
Date: Tue 14 May 85 09:00:05-PDT
From: Susannah Mackaye <MACKAYE@SU-CSLI.ARPA>
Subject: STANFORD OUT OF SOUTH AFRICA
To: folks@SU-CSLI.ARPA
There is a sit-in today (tuesday) from 9 to 12 at Hoover Institute,
where the trustees are meeting to discuss (among other things) the demands
for divestment. Come spend some time and show your support. The trustees
can hear us in their meeting room, come let them hear you. Angela
Davis will be speaking at 6:30 pm at Mandela Free University (formerly the
Inner Quad).
-------
∂14-May-85 0947 BCM Seminar
To: "@DIS.DIS[1,CLT]"@SU-AI.ARPA
∂13-May-85 1102 ARIADNE@SU-SCORE.ARPA Seminar
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 13 May 85 11:01:12 PDT
Date: Mon 13 May 85 10:52:55-PDT
From: Ariadne Johnson <ARIADNE@SU-SCORE.ARPA>
Subject: Seminar
To: su-bboards@SU-SCORE.ARPA
Stanford-Phone: (415) 497-0872
Ben Moszkowski will give the following talk at SRI on Wednesday, at
2:15 in EJ232, and at Stanford on Thursday, at 3:15 in MJ301.
Applications of Interval Temporal Logic
Ben Moszkowski
The Computer Laboratory, University of Cambridge
Corn Exchange Street, Cambridge CB2 3QG, England
What do sequential computer programs, digital circuits and distributed
networks have in common? One answer is time. With this in mind we
present Interval Temporal Logic (ITL), a formalism that augments
standard predicate logic with operators for reasoning about intervals
of time.
Within ITL we have specified and compared digital circuits ranging from
delay elements up to the Am2901 ALU bit slice developed by Advanced
Micro Devices, Inc. Furthermore, due to its expressiveness and
generality, ITL can directly capture various control structures found
in sequential and parallel computations. It has been therefore used to
reason about computer programs and simple message-passing systems.
We describe ITL and then show how it can express various functional and
timing-dependent aspects of hardware and software in a unified manner.
We also discuss Tempura, a prototype programming language based on ITL.
-------
∂14-May-85 0956 @SU-CSLI.ARPA:BCM@SU-AI.ARPA Seminar
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 14 May 85 09:54:54 PDT
Received: from SU-AI.ARPA by SU-CSLI.ARPA with TCP; Tue 14 May 85 09:50:35-PDT
Date: 14 May 85 0947 PDT
From: Ben Moszkowski <BCM@SU-AI.ARPA>
Subject: Seminar
To: "@DIS.DIS[1,CLT]"@SU-AI.ARPA
∂13-May-85 1102 ARIADNE@SU-SCORE.ARPA Seminar
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 13 May 85 11:01:12 PDT
Date: Mon 13 May 85 10:52:55-PDT
From: Ariadne Johnson <ARIADNE@SU-SCORE.ARPA>
Subject: Seminar
To: su-bboards@SU-SCORE.ARPA
Stanford-Phone: (415) 497-0872
Ben Moszkowski will give the following talk at SRI on Wednesday, at
2:15 in EJ232, and at Stanford on Thursday, at 3:15 in MJ301.
Applications of Interval Temporal Logic
Ben Moszkowski
The Computer Laboratory, University of Cambridge
Corn Exchange Street, Cambridge CB2 3QG, England
What do sequential computer programs, digital circuits and distributed
networks have in common? One answer is time. With this in mind we
present Interval Temporal Logic (ITL), a formalism that augments
standard predicate logic with operators for reasoning about intervals
of time.
Within ITL we have specified and compared digital circuits ranging from
delay elements up to the Am2901 ALU bit slice developed by Advanced
Micro Devices, Inc. Furthermore, due to its expressiveness and
generality, ITL can directly capture various control structures found
in sequential and parallel computations. It has been therefore used to
reason about computer programs and simple message-passing systems.
We describe ITL and then show how it can express various functional and
timing-dependent aspects of hardware and software in a unified manner.
We also discuss Tempura, a prototype programming language based on ITL.
-------
∂14-May-85 1034 EMMA@SU-CSLI.ARPA Modems
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 14 May 85 10:33:59 PDT
Date: Tue 14 May 85 10:27:05-PDT
From: Emma Pease <Emma@SU-CSLI.ARPA>
Subject: Modems
To: folks@SU-CSLI.ARPA, consultants@SU-CSLI.ARPA, bboard@SU-CSLI.ARPA
Tel: 497-3479
I am presently doing an inventory of the modems belonging to CSLI.
If you have a CSLI modem, please send me the following information.
1) The serial number (usually found on the bottom of the modem but sometimes
on the back).
AND
2) The Stanford Tag # (The # in the upper, left-hand corner of a green
rimmed, grey tag. It will probably start with the number 7 and contains
5 digits.)
AND
3) The modem's location.
Many thanks for your help,
Emma
-------
∂14-May-85 1208 YAN@SUMEX-AIM.ARPA release 6 at site HPP
Received: from SUMEX-AIM.ARPA by SU-AI.ARPA with TCP; 14 May 85 12:08:39 PDT
Date: Tue 14 May 85 12:04:52-PDT
From: Jerry Yan <YAN@SUMEX-AIM.ARPA>
Subject: release 6 at site HPP
To: aap@SUMEX-AIM.ARPA
cc: henaGER@SUMEX-AIM.ARPA, rosenbLOOM@SUMEX-AIM.ARPA
I will upgrade all the machines to run release 6 this week.
=jerry=
-------
∂14-May-85 1231 YAMARONE@SU-CSLI.ARPA SANDWICH REMINDER
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 14 May 85 12:31:11 PDT
Date: Tue 14 May 85 12:24:13-PDT
From: Tom Yamarone <YAMARONE@SU-CSLI.ARPA>
Subject: SANDWICH REMINDER
To: FOLKS@SU-CSLI.ARPA
WE ARE CURRENTLY GETTING SANDWICHES FROM THE STANFORD DELI IN THE SHOPPING
CTR. THEY ARE HAND-MADE DELIGHTS AND ARE FASHIONED WITH YOUR CHOICE
OF MEATS, CHEESES, BREADS AND ACUTRAMENTS. (IF YOU WANT A MENU, GO
BY THERE OR JUST ORDER WHATEVER YOU MIGHT AT YOUR FAVORITE DELI. REALLY,
IF YOU WANT IT , WE GOT IT.)
THE POINT OF THIS MESSAGE IS TO ENCOURAGE YOU TO HAVE LUNCH AT VENTURA HALL,
AND TO RECAPITULATE THAT ORDERS MUST BE INTO THE "LUNCH" FILE BEFORE 10 AM
THAT DAY.
THERE WILL ALWAYS BE EXTRAS PROPORTIONAL TO THE AMOUNT OF ORDERS,
BUT AS MANY OF YOU KNOW, YOU CAN'T COUNT ON THEM. SO , DO FEEL FREE TO
ENJOY A FINE LUNCH ANY DAY YOU WISH , THANKS TO THE VENTURA SANDWICH CORP..
BYE.
-------
∂14-May-85 1248 WRIGHT@SUMEX-AIM.ARPA SIGLUNCH FRIDAY MAY 17
Received: from SUMEX-AIM.ARPA by SU-AI.ARPA with TCP; 14 May 85 12:48:17 PDT
Date: Tue 14 May 85 12:41:00-PDT
From: Carol Wright/Susie Barnes <WRIGHT@SUMEX-AIM.ARPA>
Subject: SIGLUNCH FRIDAY MAY 17
To: Siglunch@SUMEX-AIM.ARPA
SIGLUNCH
DATE: Friday, May 17, 1985
LOCATION: Braun Audiltorium, Mudd/Chemistry Building
TIME: 12:05
SPEAKER: Johan Dekleer
Xerox
TITLE: Diagnosing Multiple Faults
Diagnostic tasks require determining the differences between a model of
an artifact and the artifact itself. The differences between the
manifested behavior of the artifact and the predicted behavior of the
model guide the search for the difference between the artifact and its
model. The diagnostic procedure presented in this paper reasons from
first principles, inferring the behavior of the composite device from
knowledge of the structure and function of the individual components
comprising the device. The novel contributions of this research are:
Multiple-faults: No presupposition is made about the number of failed
components.
Measurements: Proposes optimal measurements to localize the fault.
Probabilistic: A priori probabilities of component faultedness are taken into
account.
Intermittency: The approach is robust in response intermittent faults.
The system is based on incorporating probabilistic information into an
Assumption-Based Truth Maintenance System.
-------
-------
-------
∂14-May-85 1712 LIBRARY@SU-SCORE.ARPA Math/CS Library New Books
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 14 May 85 17:12:31 PDT
Date: Tue 14 May 85 16:20:12-PDT
From: C.S./Math Library <LIBRARY@SU-SCORE.ARPA>
Subject: Math/CS Library New Books
To: su-bboards@SU-SCORE.ARPA
cc: faculty@SU-SCORE.ARPA
Computer Algebra. Eurocam 82. edited by Calmet. QA155.7.E4.E85 1982
CONLAN Report. by Piloty, Barbacci, Borrione, Dietmeyer, Hill, Skelly.
TK7885.C66 1983.
Introductory Theory of Computer Science. by Krishnamurthy. QA76.K728 1985
Logics For Artificial Intelligence. by Turner. Q335.T87 1984.
Problem Solving With Structured Fortran 77. by Etter. QA76.73.F25E848 1984.
The DEC Dictionary. A Guide to Digital's Technical Terminology. QA76.8.D43D43
1984.
How To Write Papers and Reports About Computer Technology. by Sides.
ISI publication. QA76.165.S48 1984.
HLlull
-------
∂14-May-85 1930 avg@diablo nonlinear recursion example
Received: from SU-AIMVAX.ARPA by SU-AI.ARPA with TCP; 14 May 85 19:30:03 PDT
Date: Tue, 14 May 85 19:23:42 pdt
From: Allen VanGelder <avg@diablo>
Subject: nonlinear recursion example
To: nail@diablo
I proposed the following example of nonlinear recursion, where "e" and "p"
are EDB relations:
a(X, Y) :- e(X, Y).
a(X, Y) :- p(X, Y), a(X, Z), a(Z, Y).
Eli Upfal came up with an EDB family that makes it look like this is not
in NC, namely:
e(0, 1) e(1, 2) e(2, 3) ... e(n-1, n)
p(0, 2) p(0, 3) ... p(0, n)
With this EDB, consider resolution proofs of a(0, n) and you will see that
they have depth n, even considering nonlinear proofs.
However, if p is true everywhere, giving us transitive closure of e,
There is a nonlinear proof of a(0, n) of depth log(n), namely:
a(0, n)
/ | \
/ | \
p(0, n) a(0, n/2) a(n/2, n)
/ | \ / | \
etc. etc.
/ | \ /|\ / | \
p(0, 2) a(0, 1) a(1, 2) ... p(n-2, n) a(n-2, n-1) a(n-1, n)
| | | | |
e(0, 1) e(1, 2) ... e(n-2, n-1) e(n-1, n)
Questions:
1. Is the decision problem "Given an EDB (p and e) and the above rules,
is a(0, n) a theorem?" P-complete? In NC?
2. If P-complete, is the presence of p(X, Y) the critical factor that
makes the nonlinear recursion P-complete?
3. If in NC, is there a way to re-write
the rules as a linear recursion? The rewrite cannot be expected to
preserve a(X, Y) in all models, but it must preserve it in the minimum
model. Alternatively, is there a way to rewrite the rules as a non-linear
recursion with only the recursive predicate in the subgoals, as
is shown below for a class of linear recursions?
Jeff Ullman thinks this is the case, and has sketched his
method to me. Maybe he will send out the details to this list.
4. What are the features that determine when "NC re-writes" exist?
-----------------------------------------------------------------
Suppose we have a linear recursion in the form
s(X, Y) :- c(X, Y).
s(X, Y) :- b(X, Y, V), s(V, W), a(W).
where a, b and c are EDB relations. This is something like
Henschen and Naqvi's "normal form."
One approach to NC solution is to construct the s-implication graph
e(X, Y, V, W) :- b(X, Y, V), a(W).
where we interpret e(X, Y, V, W) to mean "if s(V, W), then s(X, Y)"
or an arc from node <V, W> to node <X, Y> in the s-implication graph.
Now we add the rules
t(X, Y, V, W) :- e(X, Y, V, W).
t(X, Y, V, W) :- t(X, Y, A, B), t(A, B, V, W).
and finally
s(X, Y) :- t(X, Y, V, W), c(V, W).
These rules "demonstrate" the NC computation. e takes one step.
t takes log(something) steps. s takes one step.
"something" n↑(number of vars in second rule for t), where n is the
number of constants in the EDB.
∂14-May-85 1956 YAN@SUMEX-AIM.ARPA Resource Management Presentations
Received: from SUMEX-AIM.ARPA by SU-AI.ARPA with TCP; 14 May 85 19:55:53 PDT
Date: Tue 14 May 85 19:55:56-PDT
From: Jerry Yan <YAN@SUMEX-AIM.ARPA>
Subject: Resource Management Presentations
To: aap@SUMEX-AIM.ARPA
cc: vsingH@SUMEX-AIM.ARPA, mak@SU-SIERRA.ARPA, i.ikabod9%LOTS-A@SUMEX-AIM.ARPA
What?
I will be presenting various issues in resource management in the context of
ELINT/CZAR/CARE.
When?
Three consecutive Wednesday mornings around 11 a.m. - after the usual AAP
meeting.
1. Static Initial Placement (of processes/agents/objects)
2. Dynamic Load Balancing (of processes/agents/objects)
3. Dynamic Reclamation (of processes/agents/objects)
Who is invited?
If you are interested in DISCUSSING about these issues, listening what others
have proposed in the past, and giving me feedback to my suggestions, you are
cordially invited. There will be NO RESULTS! I will only discuss a number of
experiments that I will perform and some ideas on initial algorithms. The
abstract of the talk follows in another message.
Why?
This will be a good opportunity for everyone to understand the problems
involved in resource amangement at various level. Also, it will be a good
opportunity for me to obtain some feedbacks to my ideas and help me
correct/refine them.
So...
see you there!
=jerry=
-------
∂14-May-85 1957 YAN@SUMEX-AIM.ARPA Abstract: Initial Placement of Agents
Received: from SUMEX-AIM.ARPA by SU-AI.ARPA with TCP; 14 May 85 19:57:27 PDT
Date: Tue 14 May 85 19:57:26-PDT
From: Jerry Yan <YAN@SUMEX-AIM.ARPA>
Subject: Abstract: Initial Placement of Agents
To: aap@SUMEX-AIM.ARPA
cc: vsinGH@SUMEX-AIM.ARPA, mak@SU-SIERRA.ARPA, i.ikabod9%LOTS-A@SUMEX-AIM.ARPA
"Where to Put Things When Time Begins"
Abstract
Distributed memory systems present non-uniform access costs. These additional
complications force one to make the extra decision of "where" to put "what".
In the CAOS problem solving paradigm, computations may be broken into smaller
discrete entities called "agents". This series of presentations is concerned
with three issues:
Placement at Load-Time
Dynamic Placement, Load-Balancing and Migration
Remote Access Protocol and Dynamic Reclamation
This first presentaion discuss the foundations of "Overture" - an initial
placement system that compares/suggests mapping of agents to sites. Making
use of programmer supplied information, "Overture" should recommend an
acceptable mapping that minimizes communication but maximizes concurrency. At
this point in time, the details of the algorithms to be used is not finalized
yet. The talk begins with an overview of the resource management problem and
a survey of past work on "process scheduling for multi-processors". ELINT is
then analyzed - suggesting the division of agents into two classes: ephemeral
and permanent. A recursive allocation approach is suggested. A solution
based on graphical techniques will be suggested although an "expert system"
approach is also presented at the end.
If you are confused, good.
Come...
... and be entertained
=jerry=
-------
∂14-May-85 2042 @SU-SCORE.ARPA:JF@SU-SUSHI.ARPA bats abstracts
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 14 May 85 20:42:02 PDT
Received: from SU-SUSHI.ARPA by SU-SCORE.ARPA with TCP; Tue 14 May 85 20:41:01-PDT
Date: Tue 14 May 85 20:40:15-PDT
From: Joan Feigenbaum <JF@SU-SUSHI.ARPA>
Subject: bats abstracts
To: aflb.all@SU-SCORE.ARPA
Andrei has suggested that I send the abstracts for this friday's bats
to everyone. To those who have already received them, I apologize for
the duplicates.
If you need driving and parking directions, please let me know.
Joan
Ernst Mayr, Stanford:
"Two Processor Scheduling is in NC"
The unit execution time scheduling problem is one of the most fundamental
scheduling problems. It consists of n tasks with precedence constraints among
them which have to be scheduled on m identical processors in such a way that
the precedence constraints are not violated, and the total length of the
schedule is minimized. Every task has to be executed without interruption and
takes unit time.
If the number of processors, m, is part of the problem instance, the problem
is a well-known NP-complete problem. For m=2, efficient algorithms have been
known for some time, the asymptotically best running in linear time.
At a recent BATS (and STOC last week), Kozen/Vazirani↑2 presented an RNC
algorithm for this problem (i.e., a parallel algorithm running in polylog time
on a polynomial number of processors, which uses random coin flips). Their
algorithm is based on the Karp/Upfal/Wigderson algorithm for maximum
cardinality matching which is also in RNC.
We show that an optimal two processor schedule can actually be found by a
deterministic NC algorithm. Our algorithm is based on a fast procedure to
obtain the length of an optimal schedule. However, as it turned out, it is not
at all trivial to obtain an actual optimal schedule, given the length
procedure. This provides another example for an apparent complexity gap
between decision problems in NC and corresponding function problems. Our two
processor scheduling algorithm also solves the maximum cardinality matching
problem for a restricted class of graphs.
This represents joint work with Dave Helmbold.
-------
Speaker: Andrei Broder, DECSRC
How hard is it to marry at random?
(On the approximation of the permanent)
The main result is that we can construct a fully polynomial (epsilon, delta)
approximation scheme for the number of perfect matchings in a bipartite
graph G if we have a procedure that given a subgraph H of G finds a perfect
matching in H ``almost'' uniformly at random (in terms of the variation
distance) in polynomial time, and conversely, we can construct such a
procedure if we have a fully polynomial (epsilon, delta) approximation
scheme for the number of perfect matchings in each subgraph of G. Thus
approximating the permanent is equivalent with finding ``almost'' random
matchings.
Markov chains that converge to the uniform distribution seem to be suitable
for the implementation of such sampling procedures. We present a Markov
chain that converges to the uniform distribution on the space of perfect
matchings for any given graph. We show that it converges in polynomial time
on complete graphs and conjecture that it converges fast for all dense
enough graphs. The conjecture implies a fully polynomial (epsilon, delta)
approximation scheme for the permanent of 0-1 matrices where at least half
of the entries in every row and every column are 1's. Finally we show that
the exact computation of the permanent of such matrices is still #P-complete.
How to Build a Random Permutation on the Fly
By Gilles Brassard and Sampath Kannan,
with ideas from Manuel Blum, Steve Rudich and Umesh Vazirani
Let n be some integer and denote {1,2,...,n} by X . Suppose
one needs a random permutation p : X --> X . If the application makes
use of the value of p(i) for each i in X or for a sizeable fraction
of it, we may as well spend the time and space needed to build p once
and for all, and store it in an array. However, if only a few values of
the permutation will be needed by the application, setting up this array
could be a waste of time. A faster initialization becomes
desirable, even if actual requests for the value of the permutation on
given arguments take more time. This problem becomes particularly
interesting when one cannot predict in advance the actual requests or
even their number. Building a random permutation on-line, or "on the fly",
consists of deciding on the value of p(i) only when it is requested for
the first time.
More formally, we wish to entertain two types of requests: init(n)
and p(i) , where the former sets up a new random permutation of the
first n integers and the latter queries for the value of the current
permutation on argument i . To be valid, a stream of requests must
start with some init(n), and any p(i) must query within the domain of
the current permutation. This problem has an on-line and an off-line
version, depending on whether the algorithm is asked to answer the
requests as they come, or whether it is allowed to look at the entire
stream of requests before producing the first answer.
Our purpose is to design new algorithms and data structures for
implementing these two types of requests, with emphasis on the case
where the number of requests is small compared to the size of the
permutations involved. We are interested in worst case, expected case
and amortized case analyses, both with respect to time and space.
For simplicity, we assume here that one can index an array and
select a random integer between two given bounds in constant time.
One implementation allows us to process an arbitrary valid stream in
constant time per request in the worst case, including initializations.
However, this algorithm allocates a block of theta(n) space in order to
process init(n) . It is therefore possible to initialize a random
permutation of the first n integers and query the value of this
permutation on k not necessarily distinct arbitrary arguments in
total time in O(k) in the worst case, as long as omega(n) space is
available. We do not know how to achieve this using less space, even
if k is much smaller than n , thus providing an intriguing example
of algorithm requiring substantially more space than time.
At the other end of the spectrum, if one is willing to spend
more time per request, it is possible to avoid the need for so much
memory. There is a different algorithm that requires O(1) time
and O(log n) initial space investment in the worst case in order to
process init(n) ; subsequent requests for the value of the permutation
on given arguments can then be processed in O(log n / log log n)
expected time and only O(1) amortized additional space. This allows us
to process a length k sequence as above in total expected time
in O(k log n / log log n), even if only O(k) space is available,
as long as k is in omega(log n) .
From: siegel@NYU-CSD2 (Alan R. Siegel)
To: JF@SU-SUSHI.ARPA
Lower Bounds for Optimal VLSI Sorters
Alan Siegel, New York University
ABSTRACT
We analyze VLSI sorting problems for N numbers in the range [0, M].
The principal portion of this talk will establish the
minimum area necessary for a sorting circuit that can read its input data
R <= N times. Although this problem seems rather peculiar, the analysis
exposes the central information theoretic issues
underlying AT↑2 bounds for sorting numbers in general ranges.
As a consequence, we attain bounds for the AT↑2 complexity of
a standard VLSI (read-once) sorter, for all ranges of M.
(Subsequent joint work with R. Cole has shown that these bounds are
tight for virtually all cases.)
This talk will be mostly combinatorial in nature.
-------
∂15-May-85 1230 @SU-SCORE.ARPA:JF@SU-SUSHI.ARPA Re: bats
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 15 May 85 12:17:46 PDT
Received: from SU-SUSHI.ARPA by SU-SCORE.ARPA with TCP; Wed 15 May 85 11:17:59-PDT
Date: Wed 15 May 85 10:57:44-PDT
From: Joan Feigenbaum <JF@SU-SUSHI.ARPA>
Subject: Re: bats
To: yigal@SU-ISL.ARPA
cc: aflb.all@SU-SCORE.ARPA, ragde%ucbernie@UCB-VAX.ARPA, yao.pa@XEROX.ARPA,
broder@DECWRL.ARPA, klawe.ibm-sj@CSNET-RELAY.ARPA,
allen.ucsc@CSNET-RELAY.ARPA, mbern%ucbernie@UCB-VAX.ARPA
In-Reply-To: Message from "yigal@SU-ISL (Yigal Brandman)" of Wed 15 May 85 10:30:42-PDT
Several people have asked me WHERE bats will be held this friday. Apologies
to those who have already received this information.
Friday, May 17, CERAS LGI, Stanford
(That's on the lower floor of the CERAS building. You have to walk down
one flight of stairs if you enter CERAS from the Alvarado Street entrance.)
10 a.m.: Mayr, 2-Proc. Sched. is in NC
11 a.m.: Broder, How hard is it to Marry at Random (Approximating the
Permanent of Dense Matrices)
12 noon: Lunch
1 p.m.: If people wish, open problems
1:30 p.m.: Brassard, Generating Random Perms. on the Fly
2:30 p.m.: Siegel, Lower Bounds on Optimal VLSI Sorters
If you have any questions, contact me at JF@su-sushi or call 4971787 or
4940928
Joan Feigenbaum
-------
∂15-May-85 1234 DIKRAN@SU-CSLI.ARPA Celebration
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 15 May 85 12:33:47 PDT
Date: Wed 15 May 85 09:25:05-PDT
From: Dikran Karagueuzian <DIKRAN@SU-CSLI.ARPA>
Subject: Celebration
To: folks@SU-CSLI.ARPA
You are cordially invited to celebrate, albeit
belatedly, the publication of our Lecture Notes
series. ``Emotion and Focus'' by Helen Nissenbaum
has just been released. A reception will be held in
Ventura at 4:30 in the afternoon of Tuesday, May 21.
-------
∂15-May-85 1233 ROSENBLOOM@SUMEX-AIM.ARPA Special Siglunch
Received: from SUMEX-AIM.ARPA by SU-AI.ARPA with TCP; 15 May 85 12:22:14 PDT
Date: Wed 15 May 85 12:03:09-PDT
From: Paul Rosenbloom <ROSENBLOOM@SUMEX-AIM.ARPA>
Subject: Special Siglunch
To: csd@SU-SCORE.ARPA, siglunch@SUMEX-AIM.ARPA, su-bboards@SUMEX-AIM.ARPA,
msgs@SU-PSYCH.ARPA
There will be a special SIGLunch today (Wednesday, 5/15), 3:30 - 4:30
in Braun Auditorium (the regular SIGLunch will still be held this
Friday). The speaker will be Tom Mitchell of Rutgers. The topic is
"Explanation-based generalizations and the Learning Apprentice
Project".
I apologize if you receive more than one notification, but due to the
last minute arrangements it is better to saturate than to miss. Pass
the word around.
-------
∂15-May-85 1235 JODY@SU-CSLI.ARPA Burrito Bandito's visit
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 15 May 85 12:35:15 PDT
Date: Wed 15 May 85 11:10:04-PDT
From: Joe Zingheim <JODY@SU-CSLI.ARPA>
Subject: Burrito Bandito's visit
To: folks@SU-CSLI.ARPA
cc: consultants@SU-CSLI.ARPA
Thursday's visit from the Burrito Bandito is coming! This time we will add
Tacos for those that want a somewhat lighter Mexican food fix. Tacos are
available in the same conbinations as the beautiful burritos, wrapped in two
soft corn tortillas. The tacos will cost $1.75 including tax. The
burritos are $3.75 for the especial (sour cream, cheese, and avacado added)
and $3.00 for the regular (no increase in price, just added in the tax).
Carnitas (shredded pork) with rice
Pollo (chicken) with rice
Chile Verde (pork in green sauce) with rice
Chile Colorado (beef in red sauce) wiht rice
Lengua (beef tongue) with with rice
Chile Relleno (green chile stuffed with cheese) and rice
Picadillo (shreaded beef) with beans
Carne Asada (Mexican style steak) with beans
Chorizo (Mexican sausage) with beans
Send mail, as usual, to Lunch@CSLI
-------
∂15-May-85 1251 @SU-SCORE.ARPA:ARK@SU-AI.ARPA CS Colloq, May 21: Paul Vitanyi, Distributed Match-Making in Computer Networks
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 15 May 85 12:51:27 PDT
Received: from SU-AI.ARPA by SU-SCORE.ARPA with TCP; Wed 15 May 85 12:13:42-PDT
Date: 15 May 85 1212 PDT
From: Arthur Keller <ARK@SU-AI.ARPA>
Subject: CS Colloq, May 21: Paul Vitanyi, Distributed Match-Making in Computer Networks
To: Faculty@SU-SCORE.ARPA, All-Colloq@SU-SCORE.ARPA
CC: ARK@SU-SCORE.ARPA, mcvax!paulv@SEISMO.ARPA,
TeX82@SRI-AI.ARPA
CS Colloquium, May 21, 4:15pm, Terman Auditorium
DISTRIBUTED MATCH-MAKING IN COMPUTER NETWORKS
Paul M.B. Vitanyi
Centrum voor Wiskunde & Informatica
Kruislaan 413, 1098 SJ Amsterdam, The Netherlands
In the very large multiprocessor systems and, on a grander scale,
computernetworks now emerging, processes are not tied to fixed processors
but run on processors taken from a pool of processors. Processors are
released when a process dies, migrates or when the process crashes. In
distributed operating systems using the service concept, processes can be
clients asking for a service, servers giving a service or both.
Establishing communication between a process asking for a service and a
process giving that service, without cen- tralized control in a
distributed environment with mobile processes, constitutes the problem of
distributed match- making. New algorithms for distributed match-making
are developed and their complexity is investigated in terms of needed
storage, in terms of message passes and in terms of processing needed.
The theoretical limitations of distri- buted match-making are established,
and the techniques are applied to several network topologies. This is
joint work with Sape J. Mullender of the Amoeba distributed operating
system project at CWI.
----
Cookies, thanks to Ginger, and possibly juice will probably be served in
the 3rd floor lounge of MJH at 3:45pm. See you there.
[This reminder is now expired.]
∂15-May-85 1252 avg@diablo nonlinear recursion followup
Received: from SU-AIMVAX.ARPA by SU-AI.ARPA with TCP; 15 May 85 12:52:13 PDT
Date: Wed, 15 May 85 11:02:33 pdt
From: Allen VanGelder <avg@diablo>
Subject: nonlinear recursion followup
To: nail@diablo
Jeff Ullman has switched sides on my previous example. Although
his proposed algorithm handles Eli Upfal's class of EDBs in log(n)
parallel time, he found other classes that look worse.
So now we are studying whether this is P-complete.
This reinforces my conjecture, which loosely phrased is: Qualified
nonlinear recursion is P-complete. The qualification is the
non-recursive p(X, Y) in my example.
∂15-May-85 1304 @SU-SCORE.ARPA:ROSENBLOOM@SUMEX-AIM.ARPA Special Siglunch
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 15 May 85 13:04:01 PDT
Received: from SUMEX-AIM.ARPA by SU-SCORE.ARPA with TCP; Wed 15 May 85 12:16:36-PDT
Date: Wed 15 May 85 12:03:09-PDT
From: Paul Rosenbloom <ROSENBLOOM@SUMEX-AIM.ARPA>
Subject: Special Siglunch
To: csd@SU-SCORE.ARPA, siglunch@SUMEX-AIM.ARPA, su-bboards@SUMEX-AIM.ARPA,
msgs@SU-PSYCH.ARPA
There will be a special SIGLunch today (Wednesday, 5/15), 3:30 - 4:30
in Braun Auditorium (the regular SIGLunch will still be held this
Friday). The speaker will be Tom Mitchell of Rutgers. The topic is
"Explanation-based generalizations and the Learning Apprentice
Project".
I apologize if you receive more than one notification, but due to the
last minute arrangements it is better to saturate than to miss. Pass
the word around.
-------
∂15-May-85 1452 ullman@diablo Re: nonlinear recursion followup
Received: from SU-AIMVAX.ARPA by SU-AI.ARPA with TCP; 15 May 85 14:52:41 PDT
Date: Wed, 15 May 85 13:55:51 pdt
From: Jeff Ullman <ullman@diablo>
Subject: Re: nonlinear recursion followup
To: avg@diablo, nail@diablo
Well I think I can handle Eli's class of EDB's in O(log↑2(n)) parallel
time. I also think that the general problem stated by Allen and Eli:
a(xy) :- e(xy).
a(xy) :- a(xz), a(zy), p(xy).
is P-complete, but that my recursive-doubling algorithm works
in the case that e is an acyclic graph.
Watch this space for further claims and withdrawals of same.
Here's a sketch of my algorithm.
First, define d(x,y), the distance from x to y to be the
number of edges in the shortest legal (according to p) path.
That is, d(x,y)=1 if e(x,y) and if a(x,z), a(z,y), and p(x,y)
are all true, then d(x,y) is at most d(x,z)+d(z,y).
Notice that the shortest legal path in an n-node graph
could have length 2↑n, but if the graph is acyclic,
then n is an upper bound--that makes a difference, I think.
Assuming n is an upper bound on legal path length, we perform
log n *phases*. At phase i, we compute ai, the restriction of a(x,y)
to those pairs x,y at distance at most 2↑i.
Assume during phase i that a is really the relation a←(i-1),
and suppose x,y is a pair at distance <= 2↑i.
Lemma: We can find sequences of nodes x0, x1,...,xk and y0, y1,...,yk
such that x0=x, y0=y, and for all j, either
1. x←(j-1) = xj and a(yj, y←(j-1)) or
2. y←(j-1) = yj and a(x←(j-1), xj)
Also, a(xj, yj).
Proof: The path x->y can be broken into x->z->y, such that
p(x,z) and p(z,y) hold (or one of these is an edge in e).
Only one of the paths x->z and z->y can be longer than 2↑(i-1),
so repeat on the long path until we have broken x->y
into legal intervals that are in relation a=a←(i-1).
Now we do recursive doubling for this phase, so that ai can
be constructed from a←(i-1). Build a graph whose nodes are
pairs of nodes of the original graph (the one denoted by e).
We have arcs:
uv->uw if a(v,w) and p(u,w)
uv->wv if a(w,u) and p(w,v)
Starting with the pairs uv such that a(u,v) is now true, we
find all reachable nodes in log n parallel time, using a transitive
closure algorithm.
The lemma assures us that if ai(x,y) is true, then the node xy
will be accessible in this graph, so we can replace a=a←(i-1)
by a=ai.
∂15-May-85 1721 EMMA@SU-CSLI.ARPA Newsletter May 16, No. 29
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 15 May 85 17:19:52 PDT
Date: Wed 15 May 85 16:59:50-PDT
From: Emma Pease <Emma@SU-CSLI.ARPA>
Subject: Newsletter May 16, No. 29
To: friends@SU-CSLI.ARPA
Tel: 497-3479
C S L I N E W S L E T T E R
←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←
May 16, 1985 Stanford Vol. 2, No. 29
←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←
A weekly publication of The Center for the Study of Language and
Information, Ventura Hall, Stanford University, Stanford, CA 94305
←←←←←←←←←←←←
CSLI ACTIVITIES FOR *THIS* THURSDAY, May 16, 1985
12 noon TINLunch
Ventura Hall ``Combinators, Categorial Grammars, and Parasitic
Conference Room Gaps'' by Mark Steedman, University of Edinburgh
Discussion led by Hans Uszkoreit, CSLI and SRI
2:15 p.m. CSLI Seminar
Redwood Hall ``Action Theory for Dialogue Games''
Room G-19 Lauri Carlson, CSLI
Discussion led by Phil Cohen
3:30 p.m. Tea
Ventura Hall
4:15 p.m. CSLI Colloquium
Redwood Hall ``Tracking Dogs and the Traces of Speech''
Room G-19 Vicki Hearne, Yale University
(Abstract on page 2)
←←←←←←←←←←←
CSLI ACTIVITIES FOR *NEXT* THURSDAY, May 23, 1985
12 noon TINLunch
Ventura Hall ``A Procedural Logic''
Conference Room Michael Georgeff (SRI and CSLI), Amy Lansky (SRI),
and Pierre Bessiere (SRI)
(Abstract on page 2)
2:15 p.m. CSLI Seminar
Redwood Hall ``Representations, Information, and the
Room G-19 Physical World'' by Ivan Blair
Discussion led by Meg Withgott
(Abstract on page 2)
3:30 p.m. Tea
Ventura Hall
4:15 p.m. CSLI Colloquium
Redwood Hall ``Lexical Structure Constraints and Morphological
Room G-19 Parsing''
John McCarthy, AT&T Bell Laboratories
(Abstract on page 3)
!
Page 2 CSLI Newsletter May 16, 1985
←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←
ABSTRACT OF THIS WEEK'S COLLOQUIUM
``Tracking Dogs and the Traces of Speech''
Extended examination of the language by means of which tracking dog
handlers think about their work and actually do work their dogs. What's
at stake is an initial question about whether there is any knowledge
that language can provide that can go outside of the exchanges that
are our performance of knowledge. I use the case of tracking dogs to
challenge the skeptic's sense that the problem of the other is the
problem of knowing the other, not in order foolishly to attempt to
bring skepticism to an end but rather to suggest a new turn skepticism
might take. But skepticism must admit that dogs exist in order to find
itself again. --Vicki Hearne
←←←←←←←←←←←←
ABSTRACT OF NEXT WEEK'S TINLUNCH
``A Procedural Logic''
Much of our commonsense knowledge about the real world is concerned
with the way things are done. This knowledge is often in the form of
`procedures' or `sequences' of actions for achieving particular goals.
In this paper, a formalism is presented for representing such
knowledge based on the notion of `process'. A declarative semantics
for the representation is given, which allows a user to state `facts'
about the effects of doing things in the problem domain of interest.
An operational semantics is also provided, which shows `how' this
knowledge can be used to achieve given goals or to form intentions
regarding their achievement. The formalism also serves as an
executable program specification language suitable for constructing
complex systems. --Michael Georgeff and Amy Lansky
←←←←←←←←←←←←
ABSTRACT OF NEXT WEEK'S SEMINAR
``Representations, Information, and the Physical World''
The notions of representation and information have been much used
in recent cognitive psychology and philosophy of mind, yet much
remains to be done to determine more precisely what is meant by these
notions, particularly in elucidating the basis of their
intentionality. I think that the place to start with an investigation
of these matters is the analysis proposed by Howard Pattee. Pattee
has for a long time wrestled with the question of how symbols are
related to their referents, and has tried to establish some general
principles of the symbol-referent or symbol-matter relation.
I shall attempt to do two things in this presentation. Firstly, I
want to explain as briefly as possible Pattee's view of symbolic
information (information carried by a symbol or string of symbols) and
the relation of symbolic information to the physical world. Secondly,
I shall consider a prominent theory of information -- Dretske's, as
presented in his book, ``Knowledge and the Flow of Information''
(1981), -- in the light of various results about the nature of symbols
and information that emerge from Pattee's analysis. --Ivan Blair
!
Page 3 CSLI Newsletter May 16, 1985
←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←
ABSTRACT OF NEXT WEEK'S COLLOQUIUM
``Lexical Structure Constraints and Morphological Parsing''
A number of formal constraints on lexical structure developed on
the basis of investigations in linguistic theory can be shown to
contribute directly to the design of a morphological parser. These
constraints include both exceptionless principles and well-established
tendencies of lexical entries, drawn from Semitic languages and other
nonconcatenative morphological systems. A constraint like the
Obligatory Contour Principle, which in the instance prohibits Arabic
roots with adjacent identical elements, has a straightforward analogue
in the parser: modulo affixation, identical surface consonants must
derive from the same root consonant. A tendency for Arabic roots to
avoid containing `t' or `y' in initial position can permit parsing
without regard to at least some of the affixational possibilities of
Arabic verbs as well. The observation that reduplicative affixes
constitute well-formed phonological constituents (McCarthy and Prince
1985) suggests ways of bringing reduplication under the purview of
recent results in finite-state parsing. In addition to these, several
other such observations will be discussed. --John McCarthy
←←←←←←←←←←←←
LINGUISTICS DEPARTMENT TALK
``How Panini's Grammar is Composed''
S. D. Joshi, University of Poona
Monday, May 20, 3:15, Bldg. 200, Room 303 (History Corner)
The composition of a grammar consisting of `sutras' which are to
be committed to memory and orally transmitted in recitation poses
peculiar problems for the author: how to achieve the necessary
brevity, how to indicate the connection between the rules, how to mark
off individual rules in the continuous text. We discuss, in a way
intended to be accessible to non-Sanskritists, the formal techniques
by which these problems are dealt with in the ``Astadhyayi'': a
procedure for condensing sequences of partially similar rules,
disjunctivity between general and particular rules, and special
conventions for the use of the conjunction `ca'. Their precise
formulation yields criteria for resolving ambiguities of the `sutra'
text and adds support for the hypothesis of multiple authorship.
←←←←←←←←←←←←
SUMMARY OF PARSER DEMONSTRATION
On Monday, May 14, Roland Hausser gave a talk and demonstration of
a system of grammar and parsing he has been working on during his stay
at CSLI. He illustrated his system by applying it to German. The
grammar contains 32 linguistic rules covering declarative and
interrogative main clauses, relative clauses embedded to arbitrary
depth or extraposed, variant word-orders, adverbs and adverbial
clauses, discontinuous constituents, coordinate structures, and other
phenomena. The parser employs a bottom-up, left-associative,
data-driven algorithm and is implemented in Interlisp-D on the
Dandelion.
←←←←←←←←←←←←
AREA NL-2 MEETING
Monday, May 20, 4:30, Ventura Seminar Room
The discussion of Noam Chomsky's manuscript, ``Knowledge of
Language: Its Nature, Origins, and Use,'' will continue.
-------
∂15-May-85 1730 maier%oregon-grad.csnet@csnet-relay.ARPA Prolog as a grammar
Received: from SU-AIMVAX.ARPA by SU-AI.ARPA with TCP; 15 May 85 17:28:42 PDT
Received: from csnet-relay by diablo with TCP; Wed, 15 May 85 17:23:12 pdt
Received: from oregon-grad by csnet-relay.csnet id ab08999; 15 May 85 20:19 EDT
Received: by ogcvax.ogc.uucp (4.12/OGC←1.1)
id AA26547; Wed, 15 May 85 11:47:18 pdt
Date: Wed, 15 May 85 11:47:18 pdt
From: "Prof. David Maier" <maier%oregon-grad.csnet@csnet-relay.ARPA>
Message-Id: <8505151847.AA26547@ogcvax.ogc.uucp>
To: nail@su-aimvax.ARPA
Subject: Prolog as a grammar
When I was down there recently, Moshe was telling me about a Prolog
evaluation scheme that some Japanese guys had worked out. It looked
very much like computing LR(k) items, with a dot that moves through
the right side of a clause. One of my students (Harry Porter) came in
this morning to talk about a paper by Pereira and Warren called
"Parsing as Deduction." [SRI TR# 295; Proc. 21st ACL Meeting, June 83.]
The idea there is almost identical, except rather than moving the
dot around, they just remove satisfied literals. They look at there
method as an adaptation of chart parsing to the Prolog domain. [Actually,
they define "Earley Deduction" as a variant on "Earley Parsing".] The
gist of the method seems to be to use top-down deduction to generate
rules for bottom-up satisfaction. I haven't looked at it enough to
decide whether it suffers from the "finding all proofs rather than
finding all answers" problem of most top-down methods.
Harry pointed out that Earley deduction always halts for Datalog
(function-free Prolog). The issue came up when Adrian Walker was giving
a talk here. Walker and Brough have a theorem that any top-down, left-to-
right interpreter that only checks its own partially completed proof tree
either will not halt on some programs or will miss some answers.
I'm still intrigued that there might be some syntactic characterization of
"semi-determinism" in Prolog. Can we detect that for any goal, the choice
of the rule to satisfy it will be confirmed or rejected within k deduction
steps, for some fixed k? Something like disjointness of guards in
Concurrent Prolog, except it may extend through several rule applications.
D Maier
∂01-Jun-85 0120 @SU-SCORE.ARPA:ARK@SU-AI.ARPA CS Colloq, June 4: Daniel Lehmann, Modal Temporal Logics: a survey of recent results
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 1 Jun 85 01:20:12 PDT
Received: from SU-AI.ARPA by SU-SCORE.ARPA with TCP; Thu 16 May 85 01:03:43-PDT
Date: 1 Jun 85 0100 PDT
From: Arthur Keller <ARK@SU-AI.ARPA>
Subject: CS Colloq, June 4: Daniel Lehmann, Modal Temporal Logics: a survey of recent results
To: All-Colloq@SU-SCORE.ARPA, Faculty@SU-SCORE.ARPA
CC: Vardi@SU-AIMVAX.ARPA, ARK@SU-SCORE.ARPA
CS Colloquium, June 4, 4:15pm, Terman Auditorium
MODAL TEMPORAL LOGICS: A SURVEY OF RECENT RESULTS
Daniel Lehmann
Hebrew University
(visiting Brandeis University)
In a joint work with S. Shelah, some extensions of the propositional
temporal logic of discrete time were advocated as useful for stating and
proving properties of probabilistic concurrent programs. Deductive
completeness theorems were proved. In a joint work with S. Kraus
corresponding decision procedures were investigated. Recently a system for
describing time and knowledge has been proposed. All those systems can be
characterized as two-dimensional modal logics, i.e. they involve two
essentially orthogonal modalities, one of them being time, that satisfy
some interchange law. The techniques involved in studying such systems
and some open problems will be described.
----
Cookies, thanks to Ginger, and possibly juice will probably be served in
the 3rd floor lounge of MJH at 3:45pm. See you there.
∂16-May-86 1019 @SU-SCORE.ARPA:JF@SU-SUSHI.ARPA Talk Monday, 5/20/85 on Optimal VLSI Sorting Circuits
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 16 May 86 10:19:36 PST
Received: from SU-SUSHI.ARPA by SU-SCORE.ARPA with TCP; Thu 16 May 85 11:16:59-PDT
Date: Thu 16 May 85 11:16:50-PDT
From: Joan Feigenbaum <JF@SU-SUSHI.ARPA>
Subject: Talk Monday, 5/20/85 on Optimal VLSI Sorting Circuits
To: aflb.local@SU-SCORE.ARPA
cc: su-bboards@SU-SUSHI.ARPA
Monday, 5/20/85, 11 a.m. to 12 noon, Margaret Jacks Hall 252
Alan Siegel, NYU
Optimal VLSI Sorting Circuits
(This is a follow-up to Alan's BATS talk Friday afternoon on Lower Bounds
for Optimal VLSI sorters. It's not necessary to have heard Friday's talk
to understand this one.)
-------
∂16-May-85 1855 @SU-SCORE.ARPA:JF@SU-SUSHI.ARPA BATS final reminder
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 16 May 85 18:54:50 PDT
Received: from SU-SUSHI.ARPA by SU-SCORE.ARPA with TCP; Thu 16 May 85 11:35:17-PDT
Date: Thu 16 May 85 11:30:14-PDT
From: Joan Feigenbaum <JF@SU-SUSHI.ARPA>
Subject: BATS final reminder
To: ragde%ucbernie@UCB-VAX.ARPA, yao.pa@XEROX.ARPA, broder@DECWRL.ARPA,
klawe.ibm-sj@CSNET-RELAY.ARPA, allen.ucsc@CSNET-RELAY.ARPA
cc: mbern%ucbernie@UCB-VAX.ARPA, aflb.all@SU-SCORE.ARPA,
su-bboards@SU-SUSHI.ARPA
Tomorrow, Friday, 5/17/85 is BATS in the CERAS LGI at Stanford. First
talk at 10 a.m. If you need any directions or information, contact
JF@su-sushi or call 497-1787 or 494-0928.
Joan Feigenbaum
-------
∂16-May-85 1903 BLOCK@SU-CSLI.ARPA Situation Semantics
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 16 May 85 19:03:19 PDT
Date: Thu 16 May 85 16:07:45-PDT
From: Ned Block <BLOCK@SU-CSLI.ARPA>
Subject: Situation Semantics
To: folks@SU-CSLI.ARPA, friends@SU-CSLI.ARPA, SU-bboards@SU-CSLI.ARPA
SITUATION SEMANTICS MADE EASY
Three lectures by
John Perry
M,W,F, June 3, 5, 7, at 3:15
REDWOOD G-19
The first lecture will be aimed at those who know nothing at all about
situation semantics.
-------
∂16-May-85 1916 BERG@SU-SCORE.ARPA Peterson's Guide
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 16 May 85 19:16:24 PDT
Date: Thu 16 May 85 15:15:10-PDT
From: Kathy Berg <BERG@SU-SCORE.ARPA>
Subject: Peterson's Guide
To: faculty@SU-SCORE.ARPA
cc: berg@SU-SCORE.ARPA
Stanford-Phone: (415) 497-4776
The deadline for submitting our changes to Peterson's Guides
is approaching.
Two weeks ago I distributed a copy of the 1984 listing of
the CSD faculty to each professor,and asked that individual
listings be updated. Please return this sheet to me at your
earliest convenience. If the original memo has been misplaced,
please let me know--I would be happy to supply another copy.
I must send our changes in quite soon. If I do not receive
a response within the next couple of days, I will assume that
no changes in your listing are required.
Thank you for your kind assistance.
Kathy
-------
∂16-May-85 1917 NILSSON@SU-SCORE.ARPA Stanford Standings
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 16 May 85 19:17:20 PDT
Date: Thu 16 May 85 15:46:13-PDT
From: Nils Nilsson <NILSSON@SU-SCORE.ARPA>
Subject: Stanford Standings
To: faculty@SU-SCORE.ARPA
I need to give a presentation to my fellow department heads (in
the Engrg School) next Monday noon. They would like to know a bit
about what computer science is and how it is done at Stanford. While
telling them this, I would also like to tell them about areas I think
we are particularly strong in and areas where we admit we want to
get better. I'd also like to tell them about what other universities
are strong in the various subareas of CS. For example, I would
probably say that UC Berkeley is strong in certain parts of systems.
So, I am soliciting comments from people about our relative
strengths versus other universities in: Numerical Analysis/Scientific
Computing; Analysis of Algorithms; Theory of Computation; Systems.
(I can already cover AI.) No need to stick to the grouping I've used
above--other breakdowns of CS into interest areas would be fine. To
do me any good, I'd need top-of-the-head responses by this Saturday.
Thanks, -Nils
-------
∂17-May-85 0946 PIERCE@SU-SCORE.ARPA Visitor Space
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 17 May 85 09:44:54 PDT
Date: Fri 17 May 85 09:43:41-PDT
From: Lee Pierce <PIERCE@SU-SCORE.ARPA>
Subject: Visitor Space
To: faculty@SU-SCORE.ARPA
I have been assigned the responsibility for visitor space allocations.
Please be sure that both Nils and I know of any visitors you are
expecting, the length and dates of their stay and I will do my best
to see that desk space is made available for them. I need your help
to make your visitor feel welcome. If you know of any space in the
office that I may not be aware of, I would appreciate that information,
too. Thanks for your help.
Lee
-------
∂17-May-85 1158 LIBRARY@SU-SCORE.ARPA New Books In The Math/CS Library
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 17 May 85 11:57:51 PDT
Date: Fri 17 May 85 11:38:21-PDT
From: C.S./Math Library <LIBRARY@SU-SCORE.ARPA>
Subject: New Books In The Math/CS Library
To: su-bboards@SU-SCORE.ARPA
cc: faculty@SU-SCORE.ARPA
On Generalizations Of Adaptive Algorithms And Application Of The Fuzzy Sets
Concept To Pattern Classification. by Wee, W 1967. Q327.W4 1976a
High-Performance Graphics System Architecture; A Methodology For Design And
Evaluation. by Carlbom.T385.C37 1984
Microprocessor Programming And Applications For Scientists And Engineers.
by Smardzewski. QA76.6.S6153 1984
Brain, Mind And Computers by Stanley Jaki. Q335.J261979
A Computer & Communications Network Performance Analysis Primer by Stuck and
Arthurs. QA76.9E94S78 1985
Humanizing Office Automation: the Impact of Ergonomics On Productivity. by
Galitz. HF5548.2G34
Local Area Networks Selection Guidelines. by Fritz, Kaldenbach, and Progar.
TK5105.7.F75 1985
C Programmeer's Library. by Purdum, Leslie and Stegemoller. QA76.73.C15P867
1984.
H. Llull
-------
∂17-May-85 1200 EMMA@SU-CSLI.ARPA Situation Semantics
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 17 May 85 11:59:50 PDT
Mail-From: BLOCK created at 16-May-85 16:07:46
Date: Thu 16 May 85 16:07:45-PDT
From: Ned Block <BLOCK@SU-CSLI.ARPA>
Subject: Situation Semantics
To: folks@SU-CSLI.ARPA, friends@SU-CSLI.ARPA, SU-bboards@SU-CSLI.ARPA
ReSent-Date: Fri 17 May 85 10:49:35-PDT
ReSent-From: Emma Pease <Emma@SU-CSLI.ARPA>
ReSent-To: friends@SU-CSLI.ARPA
SITUATION SEMANTICS MADE EASY
Three lectures by
John Perry
M,W,F, June 3, 5, 7, at 3:15
REDWOOD G-19
The first lecture will be aimed at those who know nothing at all about
situation semantics.
-------
∂17-May-85 1248 @SU-SCORE.ARPA:udi@wisc-rsch.arpa A Thank You
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 17 May 85 12:47:12 PDT
Received: from wisc-rsch.arpa by SU-SCORE.ARPA with TCP; Fri 17 May 85 12:45:10-PDT
Received: by wisc-rsch.arpa; Fri, 17 May 85 14:33:01 cdt
Message-Id: <8505132035.AA05640@wisc-rsch.arpa>
Received: from RUTGERS.ARPA by wisc-rsch.arpa; Mon, 13 May 85 15:35:04 cdt
Date: 13 May 85 16:31:47 EDT
From: Michael <Berman@RUTGERS.ARPA>
Subject: A Thank You
To: theory@WISC-RSCH.ARPA
Sender: udi@wisc-rsch.arpa
Resent-From: udi@wisc-rsch.arpa (Udi Manber)
Resent-To: theory:;@wisc-rsch.arpa
To the organizers of STOC:
As graduate student attendees of this year's STOC, we would like to thank the
organizers who worked to make it more accessible to students. We appreciate
the efforts of all who arranged for the seven corporate sponsors. We are glad
to see that there are many in SIGACT who really do care about the students.
Michael Berman
Martin Carroll
Fritz Henglein
Suzanne Menzel
Rutgers University, New Jersey
-------
∂17-May-85 1301 @SU-SCORE.ARPA:udi@wisc-rsch.arpa PODC 85 General Information
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 17 May 85 13:01:35 PDT
Received: from wisc-rsch.arpa by SU-SCORE.ARPA with TCP; Fri 17 May 85 12:51:22-PDT
Received: by wisc-rsch.arpa; Fri, 17 May 85 14:31:30 cdt
Message-Id: <8505170831.AA11067@wisc-rsch.arpa>
Received: from csnet-relay.arpa by wisc-rsch.arpa; Fri, 17 May 85 03:31:26 cdt
Received: from ibm-sj by csnet-relay.csnet id aa17528; 17 May 85 4:22 EDT
Date: Fri, 17 May 85 00:19:48 PDT
From: Ray Strong <strong%ibm-sj.csnet@csnet-relay.arpa>
To: theory@wisc-rsch.ARPA
Subject: PODC 85 General Information
Sender: udi@wisc-rsch.arpa
Resent-From: udi@wisc-rsch.arpa (Udi Manber)
Resent-To: theory:;@wisc-rsch.arpa
GENERAL INFORMATION
LOCATION
All technical sessions, receptions, the banquet, and lodging will
be held at Minaki Lodge, Minaki, Ontario, Canada.
TRANSPORTATION
"BY ROAD:" Minaki Lodge is located 240 km (150 miles) east of
Winnipeg, Manitoba, on Sand Lake, adjacent to the town of Minaki,
where Highway 596 terminates; 50 km (30 miles) north of Highway
17 (Trans Canada Highway) and Kenora, Ontario.
"BY AIR:" Winnipeg International Airport is the closest major
airport; it is served by the following scheduled airlines. Cana-
da: Air Canada, C.P. Air, Nordair, Pacific Western and Wardair.
U.S.: Air Canada, Frontier, and Northwest Orient. Chartered ser-
vice is available from Minneapolis to Kenora via Thunder Bay.
Bearskin Airlines (800-465-2370 Canada, 204-774-8414 other) has a
scheduled flight from Dryden directly to Minaki Lodge's Airstrip,
connecting with Nordair's flights from Toronto to Dryden. G & B
Aviation (612-944-2500) offers charter flights into Minaki from
Minneapolis. Perimeter Airlines (204-774-1958/786-7031) offers
charter flights from Winnipeg International Airport to Minaki
Lodge. For private aircraft, there is a convenient landing strip
on-site and complimentary tie-down service for float planes at
Minaki Lodge's Marina.
CLIMATE
The temperature in Minaki in early August averages between 24 and
28 degrees Celsius (75-85 Fahrenheit).
SOCIAL EVENTS
The planned social events include; a reception both Sunday and
Monday evenings and a banquet Tuesday evening.
THINGS TO DO
Minaki Lodge, located in the Canadian wilderness, provides an un-
disturbed setting with wildlife, sprawling lakes and waterways,
and tranquil pine forests.
The 80-acre resort caters to outdoor enthusiasts with river
cruises, sailing, waterskiing, canoeing, windsurfing, paddleboat-
ing and swimming in Sand Lake. There is also an indoor pool with
whirlpool, sauna and adjacent sundeck. Guided fishing trips and
northwoods camping excurisions are available.
There is a nine-hole golf course, three tennis courts with night
lighting, archery, volleyball, badminton, children's playground,
library and billiard room.
REGISTRATION DESK
The registration desk will be open from 4:00 P.M. to 9:00 P.M. on
Sunday, August 4, 8:00 A.M. to 6:00 P.M. Monday and Tuesday, and
8:00 A.M. to noon on Wednesday.
OTHERS
The registration fee includes luncheons on Monday and Tuesday,
receptions on Sunday and Monday evenings, the banquet Tuesday
evening, and a copy of the proceedings.
For further information, please call
"Pat Donovan", Department of Computer Science, University of
Waterloo, (519)885-1211, or send electronic mail to
mamalcolm@waterloo.csnet or pjdonovan@waterloo.csnet.
Hotel Registration Form
A block of rooms has been reserved (until July 1,1985) for
conference participants. If you wish to reserve one of these rooms,
please complete the form below and return it to:
Minaki Lodge
A Radisson Resort and
Conference Centre
Minaki, Ontario
Canada P0X lJ0
If you phone Minaki Lodge, (807) 224-4000, mention that you will be
attending the Symposium on Principles of Distributed Computing.
RESERVATIONS MUST BE RECEIVED BY JULY 1, 1985
Accommodations will not be confirmed without a cheque for the 1st
night's deposit, or use your American Express, Diners Club, VISA,
or Mastercard number to guarantee your reservation. You will be
charged for the lst night if reservations are not cancelled 7 days
prior to arrival.
Canadian U.S.
Rate: single occupancy $79.20/night $63.36/night
double occupancy $89.10/night $71.28/night
Prices do not include 5% sales tax or 10% service charge.
Name: ←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←
Affiliation:←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←
Address:←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←
City:←←←←←←←←←←←←←←←←←←←←State/Province:←←←←←←←←←←←←←
Zip/Postal Code:←←←←←←←←←Country:←←←←←←←←←←←←←←←←←←←←
Phone number:(←←←)←←←←←←←←← No.of Persons←←←←←←←←←←←←
Arrival: Time:←←←←←←←←←←←←Date:←←←←←←←←←←←←←←←←←←←
Departure: Time:←←←←←←←←←←←←Date:←←←←←←←←←←←←←←←←←←←
( ) check enclosed ( ) VISA ( ) MASTERCARD
( ) AMERICAN EXPRESS ( ) Diners' Club
Card No.←←←←←←←←←←←Exp.Date←←←←←←←←
Signature←←←←←←←←←←←←←←←←←←←←←←←←←←
ADVANCED REGISTRATION FORM
Please use this form or a facsimile to pre-register. Advanced
registration closes July 29, 1985. Registration after July 29 or at
the conference site is subject to a late fee. Please mail your
completed form with cheque (drawn on a North American bank) or
international money order (in Canadian or U.S. funds) payable to
"Symposium on PODC-85" to:
PODC'85
c/o Pat Donovan
Computer Science Dept.
Univeristy of Waterloo
Waterloo, Ontario
CANADA N2L 3G1
The rates for registration are listed below.
Requests for refunds will be honoured until July 29,1985.
Name: ←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←
Affiliation:←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←
Address:←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←
City:←←←←←←←←←←←←←←←←←←←←State/Province:←←←←←←←←←←←←←
Zip/Postal Code:←←←←←←←←←Country:←←←←←←←←←←←←←←←←←←←←
Phone number:(←←←)←←←←←←←←←
Please circle appropriate fees
CDN U.S.
ACM/SIG MEMBER $175 $133
(Membership #←←←←←←←←←←←←←←)
NON-MEMBER $225 $171
STUDENT $ 75 $ 57
Non-student late fee $ 50 $ 40
Student late fee $ 15 $ 12
Additional Banquet tickets ←←←←X $ 30 $ 23
Total enclosed: $←←←←←←←←←←←←←←U.S. $←←←←←←←←←←←←←CDN
∂17-May-85 1348 LIBRARY@SU-SCORE.ARPA Socrates Update newsletter-Free, Request To Be Placed On Mailing List
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 17 May 85 13:48:37 PDT
Date: Fri 17 May 85 13:23:56-PDT
From: C.S./Math Library <LIBRARY@SU-SCORE.ARPA>
Subject: Socrates Update newsletter-Free, Request To Be Placed On Mailing List
To: su-bboards@SU-SCORE.ARPA
cc: faculty@SU-SCORE.ARPA
Socrates Update is a newsletter about the Stanford online catalog. It is
a single sheet newsletter with information on both sides. Two sheets have
been published, vol. 1 no. 1 April 12 and vol. 1, no. 2 May 2. The purpose
of the newsletter is to inform the Stanford community and users of
Socrates abouth the online catalog. Different searching techniques are
described in each issue. The newsletter is a joint effort of the Stanford
University Libraries and I.T.S. There is no subscription charge for the
Stanford community. To receive your own copy of Socrates Update, send
your request to ec.cat@forsythe. Be sure to give your name and the
physical address to which you would like the material mailed.
Harry Llull
-------
∂17-May-85 1402 @SU-CSLI.ARPA:PCOHEN@SRI-AI.ARPA [Phil Cohen <PCOHEN@SRI-AI.ARPA>: Seminar -- Rational Interaction. Jeff Rosenschein]
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 17 May 85 14:01:39 PDT
Received: from SRI-AI.ARPA by SU-CSLI.ARPA with TCP; Fri 17 May 85 13:51:22-PDT
Date: Fri 17 May 85 13:50:08-PDT
From: Phil Cohen <PCOHEN@SRI-AI.ARPA>
Subject: [Phil Cohen <PCOHEN@SRI-AI.ARPA>: Seminar -- Rational Interaction. Jeff Rosenschein]
To: NL4: ;
A talk of interest,
Phil
---------------
Mail-From: PCOHEN created at 17-May-85 13:47:13
Date: Fri 17 May 85 13:47:12-PDT
From: Phil Cohen <PCOHEN@SRI-AI.ARPA>
Subject: Seminar -- Rational Interaction. Jeff Rosenschein
To: AIC-Staff: ;
cc: bboard@SRI-AI.ARPA
Jeff Rosenschein will speak on Friday, June 7, at 10 am in EJ232
RATIONAL INTERACTION:
COOPERATION AMONG INTELLIGENT AGENTS
Jeffrey S. Rosenschein
Stanford University
The development of intelligent agents presents opportunities to
exploit intelligent cooperation. Before this can occur, however, a
framework must be built for reasoning about interactions. This talk
describes such a framework, and explores strategies of interaction
among intelligent agents.
The formalism that has been developed removes some serious
restrictions that underlie previous research in distributed artificial
intelligence, particularly the assumption that the interacting agents
have identical or non-conflicting goals. The formalism allows each
agent to make various assumptions about both the goals and the
rationality of other agents. In addition, it allows the modeling of
restrictions on communication and the modeling of binding promises
among agents.
This work has been done in conjunction with Matt Ginsberg
and Michael Genesereth.
-------
-------
-------
∂17-May-85 1433 HEDGES@SU-SCORE.ARPA Faculty Lunch
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 17 May 85 14:32:52 PDT
Date: Fri 17 May 85 13:46:42-PDT
From: Karen Hedges <HEDGES@SU-SCORE.ARPA>
Subject: Faculty Lunch
To: Faculty@SU-SCORE.ARPA
cc: Bureaucrats@SU-SCORE.ARPA
REMINDER: The Faculty Lunch will be held on Tuesday, May 21 @ 12:15
in Conference Room 146. The topic for discussion will be
"Should the CSD Require its Students to Take Courses?"
-------
∂18-May-85 1213 @SU-SCORE.ARPA:coraki!pratt@Navajo input on fac. lunch
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 18 May 85 12:13:24 PDT
Received: from Navajo ([36.8.0.48].#Internet) by SU-SCORE.ARPA with TCP; Sat 18 May 85 12:11:10-PDT
Received: by coraki.uucp (1.1/SMI-1.2)
id AA03972; Fri, 17 May 85 23:32:47 pdt
Date: Fri, 17 May 85 23:32:47 pdt
From: coraki!pratt@Navajo (Vaughan Pratt)
Message-Id: <8505180632.AA03972@coraki.uucp>
To: Faculty@SU-SCORE.ARPA
Subject: input on fac. lunch
Unfortunately I won't be able to attend the lunch. However I'd
like to provide my input in advance on the lunch topic,
"Should the CSD Require its Students to Take Courses?"
Rather than simply answering yes or no and giving reasons, I would like
to propose a fundamental change in the progression of students through
the department. They should plan on being out in three years. Taking
four years should be considered a relatively leisurely progression,
five years should be considered downright slow.
In this connection a distinction is often drawn between practical
(implementation-oriented) and theoretical theses. I am somewhat
sympathetic to the idea that implementation is both more
labor-intensive and more exacting (the computer is mindlessly accurate,
unlike thesis readers and referees) than obtaining theoretical
results. On the other hand I am at present willing to concede at most
between six months and a year between a thesis requiring that something
be built and a purely theoretical thesis. This additional time should
be regarded as the "implementation handicap." Thus a good time for a
thesis requiring implementation would be 3.5 to 4 years. Closer study
may suggest a different value - either higher or lower - for this
handicap. The following is for the case of no handicap.
The first year should be spent essentially exclusively on course work;
roughly four courses a term for each of three terms. At the end of the
year the students should be examined to determine who advances, who
requires additional or repeat coursework, and who is just plain out.
This represents a tightening up of the administration of the comp,
which in effect would be given only once a year instead of twice, and
which would allow only two attempts, though with the "second attempt"
only dealing with previously failed subjects rather than the present
all-or-nothing approach.
The courses taken during the first year should contain a mixture of
core courses corresponding roughly to the current comp. syllabus and
more researchy courses and seminars. The latter courses should serve
to get students acquainted with some, but not all, of the department's
areas. Furthermore it should aim to get them thinking in some depth in
either one or two areas, normally two, one only in the case where the
student is absolutely convinced that this is the right area.
At the beginning of the second year the student should choose a
research area and officially begin work in that area. At some point
late in the year the student should be examined in that area to
establish the extent to which he is mastering the area, i.e. the area
qualifying exam.
At the beginning of the third year the student should begin writing.
A skeleton of the thesis should be drawn up early in the year, and
expanded and modified as work proceeds. The gap between completion
of the work destined for the thesis and the final polishing of the
thesis should be at most three or four months.
At any time during years 2 and 3, hopefully pretty rarely in year 3, it
may become apparent that the area is just not working out. The
appropriate action is then *restart without prejudice*; the student is
in the same administrative state as at the start of year 2. A second
restart may be needed; if this happens beyond the student's third year
this should be considered *restart with prejudice*, where the student
is placed on probation and fairly smooth progress is expected
thereafter.
The point of this approach is to move up the time at which students
feel they have a definite goal, without however locking them too
rigidly into the resulting course of affairs. I see too much vague
wandering around at present in this period, with students having
difficulty focussing themselves even well into their third and fourth
years. I think it is better to spend the second year with a definite
goal, even if that goal does not pan out, than to spend that time
daydreaming and hoping for inspiration to strike.
I do not propose unconditionally booting out people who are still
working on their thesis a year or even two years after they should have
graduated according to the above timetable. On the other hand they
should be put under increasing pressure to hurry up and finish.
For such people Black Friday should be considered to be every day of the
year, and alarm bells should be ringing steadily in the administration
as a reminder to everyone that they are late.
I do not believe the timetable I propose is feasible with the present
resources. A major obstacle is the curriculum; it is not sufficiently
tightly organized to mesh properly with my timetable. Nor does it seem
to cover the presumably appropriate material appropriately, inferrable
from the present serious mismatch between the curriculum and the comp.
The biggest objection I could raise to my plan however is that is
counter to the department's present leisurely style. If there were
universal opposition to my proposal on this ground then I would be
inclined to regard this as a vote of confidence for the present system,
which I would then be happy to go along with, since I confess I don't
really know what is the optimal educational strategy for the #1 ranked
CS department in the country. After all, maybe it really does take two
years from entry before a student can sensibly choose an area.
-v
∂18-May-85 1918 @SU-SCORE.ARPA:golub@Navajo AI Fellowships
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 18 May 85 19:18:25 PDT
Received: from Navajo ([36.8.0.48].#Internet) by SU-SCORE.ARPA with TCP; Sat 18 May 85 19:15:08-PDT
Date: 18 May 1985 1915-PDT (Saturday)
From: Gene Golub <golub@Navajo>
To: faculty@score
Cc:
Subject: AI Fellowships
------- Forwarded Message
Return-Path: <@SU-SCORE.ARPA:FOX@AFSC-HQ.ARPA>
Received: from SU-SCORE.ARPA by Navajo with TCP; Wed, 15 May 85 12:10:48 pdt
Received: from AFSC-HQ.ARPA by SU-SCORE.ARPA with TCP; Wed 15 May 85 11:59:21-PDT
Date: Wed 15 May 85 14:57:57-EDT
From: FOX@AFSC-HQ.ARPA
Subject: ARTIFICIAL INTELLIGENCE FELLOWSHIPS
To: na.dis@SU-SCORE.ARPA
Cc: FOX@AFSC-HQ.ARPA
Rome Air Development Center (RADC) invites qualified university
professors to apply for research and technical mamagement fellowhips
in artilficial intelligence R&D. Candidates must demonstrate, either
through formal academic training or extensive research experience,
the ability to conduct or direct basic and applied artificial
intelligence (AI) research in one of the following areas:
Logic Programming Environments
Natural Language Understanding
Planning
Knowledge-based and Expert Systems
Knowledge-based Software Development Aids
Problem Solving and Reasoning Under Uncertainty
Located at the edge of the Adirondack State Park, RADC provides
the opportunity for university researchers (and their students)
to participate in a program of artificial intellegence R&D focused
on challenging real world applications. The RADC AI Laboratory
features expanding hardware facilities, including VAXs, Symbolics
3600s, and LMI LAMBDAs, and is surrounded by the developing AI
community of activity associated with the RADC University AI
Consortium. As both a participant in the DARPA Strategic Computer
Initiative and as a major node on the Arpnet/Milnet, AI researchers
at RADC have both the opportunity and the means to develop and
maintain close associations with colleagues at the major university
centers of AI research.
For further information or to express interest contact
Ray Urtz RADC/CO Griffiss AFB, NY 13441 or
FOWLER@RADC-MULTICS.ARPA
-------
------- End of Forwarded Message
∂19-May-85 1848 @SU-SCORE.ARPA:JF@SU-SUSHI.ARPA Talk Monday by A. Siegel on Optimal VLSI Sorters
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 19 May 85 18:47:32 PDT
Received: from SU-SUSHI.ARPA by SU-SCORE.ARPA with TCP; Sun 19 May 85 18:44:19-PDT
Date: Sun 19 May 85 18:44:04-PDT
From: Joan Feigenbaum <JF@SU-SUSHI.ARPA>
Subject: Talk Monday by A. Siegel on Optimal VLSI Sorters
To: su-bboards@SU-SUSHI.ARPA
cc: aflb.su@SU-SCORE.ARPA
TIME: Monday, May 20, 1985, 11 a.m.
PLACE: Margaret Jacks Hall 252
Constructing Optimal VLSI Sorting Circuits
Alan Siegel
New York University
ABSTRACT
(This is joint work with Richard Cole). We consider
the problem of sorting N numbers in the range 0 to M-1 in
time T. For every value of N and M, and just about any value
of T, we give VLSI circuits that are optimal, i.e. use
minimal area, up to a constant factor. Our work distin-
guishes two types of circuits: perimeter sorters, which have
I/O ports located along the perimeter, and dense sorters,
which can have I/O ports in the interior of the circuit. We
show, for example, that Thompson's original AT↑2 = OMEGA
(N↑2) bound for (dense) sorters of N numbers in [0,N] is
tight for log N log↑* N <= T <= sqrt(N).
The major part of this work consists of building novel
merging networks. One curious aspect of our constructions
is that many of the merging networks rely on non-optimal
encodings of data. Another interesting feature is that in
some cases, different merging networks within the same
sorter are run at differing speeds.
-------
∂20-May-85 0037 RESTIVO@SU-SCORE.ARPA PROLOG Digest V3 #23
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 20 May 85 00:36:54 PDT
Date: Friday, May 17, 1985 7:31PM
From: Chuck Restivo (The Moderator) <PROLOG-REQUEST@SU-SCORE.ARPA>
Reply-to: PROLOG@SU-SCORE.ARPA
US-Mail: P.O. Box 4584, Stanford CA 94305
Phone: (415) 326-5550
Subject: PROLOG Digest V3 #23
To: PROLOG@SU-SCORE.ARPA
PROLOG Digest Monday, 20 May 1985 Volume 3 : Issue 23
Today's Topics:
Query - External Calls & Interlisp/LOOPS
Announcement - InterKibernetik '85
Implementation - If-Then-Else Syntax
LP Library - New Book
----------------------------------------------------------------------
Date: 16 May 1985 11:01-EDT
From: Robert.Coyne@CMU-CS-CAD.ARPA
Subject: External calls from Prolog!
We have a project in the Architecture Dept. at CMU
wherein we are considering implementing a shape
grammar in Prolog. In order to display the results
and interactively check the grammar we would like
to be able to make external calls from Prolog to a
Graphics package we have which is written in C (also
in Pascal).
The version of Prolog we have is C-Prolog, Version 1.3
(Pereira et al) running on a Vax under Unix. There is
also a Tops20 version available at CMU.
Thanks you,
-- Robert F. Coyne
------------------------------
Date: 13 May 85 18:04:49 EDT
From: Louis Steinberg <STEINBERG@RUTGERS.ARPA>
Subject: Prolog for Interlisp/Loops
Can anyone point me to an implementation of Prolog
that will run on a Xerox Lisp machine, i.e. is
implemented in Interlisp-D or Interlisp-D with
LOOPS? I know of Ken Kahn's version but that
unfortunately does not use standard LOOPS. This is
for an educational environment so efficiency is not
essential. Also welcome would be advice on porting
some other, existing version.
Thank you for the help.
-- Lou Steinberg
------------------------------
Date: Sun, 12-May-85 17:50:59 PDT
From: (Neal D. McBurnett) ihnp4!drutx!druny!neal@UCB-Vax.ARPA
Subject: Cybernetics conference in Esperanto: InterKibernetik '85
The World Association of Cybernetics, Computer Science,
and System Theory (Tutmonda Asocio pri Kibernetiko,
Informatiko kajSistemiko or TAKIS) is sponsoring
InterKibernetik '85 in Budapest, Hungary, from July 27
to August 2. It is also being sponsored by associations
in Hungary, France, Italy and Belgium. It will be an
interdisciplinary conference, covering fields as diverse
as
Information Theory,
Programming Methodology,
and Linguistic Cybernetics.
Three working languages will be used: Esperanto, English
and French.
Dr. Stuart Umpleby of Georgetown University is the US
representative on the Program Committee.
I plan to attend, so stay tuned for a summary...
-- Neal McBurnett
------------------------------
Date: Thu, 9 May 85 18:09:15 pdt
From: Peter Ludemann <Ludemann%UBC@csnet-relay.arpa>
Subject: if-then-else syntax in Prolog
For getting rid of cuts, soundly implemented
if-then-else (as in MU-Prolog) is helpful.
However, the syntax leaves something to be
desired. Here's an example (the "split" in
quicksort) where "split(X,L,Low,High)" takes
the list "L" and splits it into two lists
"Low" and "High" with all elements in "Low"
less than or equal to "X" and all those in
"High" greater than "X":
split(←, nil, nil, nil).
split(X, Y.Rest, X.Low, High) :- X =< Y, split(X, Rest, Low, High).
split(X, Y.Rest, Low, X.High) :- X > Y, split(X, Rest, Low, High).
This can be made more efficient with a cut:
split(←, nil, nil, nil).
split(X, Y.Rest, X.Low, High) :- X =< Y, !, split(X, Rest, Low, High).
split(X, Y.Rest, Low, Y.High) :- split(X, Rest, Low, High).
With if-then-else, this ends up as:
split(←, nil, nil, nil).
split(X, Y.Rest, Low, High) :- if X =< Y
then Low = Y.Low2, split(X, Rest, Low2, High)
else High = Y.High2, split(X, Rest, Low, High2)
X = X.
Some clarity has been lost and the result may actually be less
efficient because of the calls to "=".
Does anyone have any thoughts on how to improve this?
------------------------------
Date: Thu, 9-May-85 21:33:09 PDT
From: (Simon MacDonald) munnari!SimonMac@Seismo.ARPA
Subject: New Book
Logic Programming: a Classified Bibliography
by I. Balbin, and K. Lecot.
This exciting area in computer science is growing very
quickly, so quickly in fact, that an overview of its ori-
gins, seminal papers, landmark treatises, pioneering arti-
cles etc., may be forgotten, or difficult to identify and
obtain.
Now, the authors of this comprehensive bibliography of well
over 1,600 entries have brought it all together, classed the
type of each paper or monograph into one of 18 broad
categories. This is somewhat arbitrary, but very useful for
those wishing to cover a specific sub-topic. If the item
covers more than one category, it is entered in each
relevant section.
The bibliography has many other useful features, namely,
it's more than twice the size of the one which appeared in
Journal of Logic Programming, and is very recent. It also
has an author index, and subject index.
Prepublication price of A$17.95 is valid only until June
30th, 1985.
Approx. 330 pages, softcovers,laminated, burstbound.
Orders may be placed ONLY with the publisher or his agents
in U.S.:
Polygonal Publishing House, 210 Broad Street,
Washington, NJ, 07882.
The Publisher is:
WILDGRASS BOOKS, 289A Smith Street, Fitzroy,
Victoria, 3065, AUSTRALIA.
Please note that only prepaid orders will be processed
immediately. Others will attract handling charges, and be
subject to a 20% increase in price after June 30th, 1985.
Release date July 15, 1985.
------------------------------
End of PROLOG Digest
********************
∂20-May-85 0758 NILSSON@SU-SCORE.ARPA Tuesday lunch
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 20 May 85 07:58:50 PDT
Date: Mon 20 May 85 07:58:10-PDT
From: Nils Nilsson <NILSSON@SU-SCORE.ARPA>
Subject: Tuesday lunch
To: faculty@SU-SCORE.ARPA
Tomorrow's faculty lunch topic was announced as "Should our first-year
students be required to take courses?" or something like that. Let's
expand the topic to include the whole matter of what a PhD student's
first year ought to be like. Should we try to give fellowships during
the first year and encourage students to take courses and get to know
what is going on more thoroughly before signing on with a research
project? -Nils
-------
∂20-May-85 1029 TULEY@SU-SCORE.ARPA Palo Alto Phone books
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 20 May 85 10:29:01 PDT
Date: Mon 20 May 85 10:08:33-PDT
From: Kimberly Tuley <TULEY@SU-SCORE.ARPA>
Subject: Palo Alto Phone books
To: csd@SU-SCORE.ARPA
If you haven't alreay......
Please stop by the CS reception desk and pick up your Palo Alto
Phone book(s). Limited Supply will be removed at the end of this week.
thanks,
Kim
-------
∂20-May-85 1120 BARWISE@SU-CSLI.ARPA Reminder
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 20 May 85 11:19:56 PDT
Date: Mon 20 May 85 11:14:14-PDT
From: Jon Barwise <BARWISE@SU-CSLI.ARPA>
Subject: Reminder
To: RRR@SU-CSLI.ARPA
Brian asked me to remind you of tomorrows meeting. He will talk on
"An initial look at a theory of representation"
-------
∂20-May-85 1126 EMMA@SU-CSLI.ARPA Celebration reminder
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 20 May 85 11:26:11 PDT
Date: Mon 20 May 85 11:17:32-PDT
From: Emma Pease <Emma@SU-CSLI.ARPA>
Subject: Celebration reminder
To: folks@SU-CSLI.ARPA
Tel: 497-3479
REMINDER:
You are cordially invited to celebrate, albeit
belatedly, the publication of our Lecture Notes
series. ``Emotion and Focus'' by Helen Nissenbaum
has just been released. A reception will be held in
Ventura at 4:30 in the afternoon of Tuesday, May 21.
-------
∂20-May-85 1138 INGRID@SU-CSLI.ARPA Word Processing
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 20 May 85 11:38:18 PDT
Date: Mon 20 May 85 11:32:21-PDT
From: Ingrid Deiwiks - 497-3084 <INGRID@SU-CSLI.ARPA>
Subject: Word Processing
To: Research@SU-CSLI.ARPA
cc: Ingrid@SU-CSLI.ARPA
Paul Skokowski from Livermore Labs will be a Philosophy graduate student
at Stanford next year. He phoned me this morning to ask if anyone at
CSLI would be interested in doing some high-speed parallel language
processing on a Cray X-MP or Cray 2.
Anyone interested, please call Paul at 422-6407.
Ingrid
-------
∂20-May-85 1413 MARJORIE@SU-CSLI.ARPA Macintosh software upgrade
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 20 May 85 14:13:23 PDT
Date: Mon 20 May 85 14:10:44-PDT
From: Marjorie Maxwell <MARJORIE@SU-CSLI.ARPA>
Subject: Macintosh software upgrade
To: folks@SU-CSLI.ARPA
cc: consultants@SU-CSLI.ARPA
As advertised, Apple has released new versions of MacWrite, MacPaint, and
the System. All of you who have Macintoshes can bring in your System master
disk, and Write/Paint master disk here to me at the Computer Group and we can
do the upgrading for you to the latest version.
Marj
-------
∂20-May-85 1449 REGES@SU-SCORE.ARPA MacFest Reminder
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 20 May 85 14:49:21 PDT
Date: Mon 20 May 85 14:44:19-PDT
From: Stuart Reges <REGES@SU-SCORE.ARPA>
Subject: MacFest Reminder
To: faculty@SU-SCORE.ARPA
Office: Margaret Jacks 210, 497-9798
I thought I'd send out a personal reminder of the MacFest that is going on
today and tomorrow at Tressider. Two of the most interesting events are a
talk by Alan Kay tonight at 7 pm in Terman Auditorium and a talk tomorrow
evening by the Mac Development team (the original group of people who designed
the thing) at Kresge Auditorium at 7 pm. Kay's talk promises to be quite
interesting. I know that he asked for at least 6 different kinds of A/V
equipment and that he plans to talk for somewhere between 2 and 4 hours. CSD
is contributing part of the money to have the talk videotaped, for those of
you who miss the talk. The videotape will be in the Math/CS Library. Other
than these two big talks, there are a lot of activities all day at Tressider
with software vendors, hardware vendors, Mac music, and other MacStuff.
-------
∂20-May-85 1456 TAJNAI@SU-SCORE.ARPA Re: MacFest Reminder
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 20 May 85 14:56:18 PDT
Date: Mon 20 May 85 14:54:41-PDT
From: Carolyn Tajnai <TAJNAI@SU-SCORE.ARPA>
Subject: Re: MacFest Reminder
To: REGES@SU-SCORE.ARPA, faculty@SU-SCORE.ARPA
In-Reply-To: Message from "Stuart Reges <REGES@SU-SCORE.ARPA>" of Mon 20 May 85 14:46:44-PDT
APPLE SHOULD BE IN THE FORUM!!
-------
∂20-May-85 1613 DAVIES@SUMEX-AIM.ARPA Architecture meeting
Received: from SUMEX-AIM.ARPA by SU-AI.ARPA with TCP; 20 May 85 16:13:41 PDT
Date: Mon, 20 May 1985 16:12 PDT
Message-ID: <DAVIES.12112616983.BABYL@Sumex>
From: DAVIES@Sumex
To: AAP@Sumex
CC: i.ikabod9@lotsa, Yeager@Sumex, Mak@Sierra, Davies@Sumex
Subject: Architecture meeting
The circumstances for this week's meeting of the Advanced
Architectures Group are a bit unusual. Officially, there is no
specific topic. Unofficially, Jerry Yan will be speaking about his
work in distributed resource allocation. Jerry's unofficial talk will
begin at 9:30 am. Attendance will be optional, because Jerry feels
that the work is too preliminary and tentative to warrant our usual
"compulsory" attendance.
-- Byron
∂20-May-85 2104 @SUMEX-AIM.ARPA:YAN@SU-SCORE.ARPA New abstract: Dynamic load balancing
Received: from SUMEX-AIM.ARPA by SU-AI.ARPA with TCP; 20 May 85 21:04:08 PDT
Received: from SU-SCORE.ARPA by SUMEX-AIM.ARPA with TCP; Mon 20 May 85 21:00:02-PDT
Date: Mon 20 May 85 21:00:02-PDT
From: Jerry C. Yan <YAN@SU-SCORE.ARPA>
Subject: New abstract: Dynamic load balancing
To: aap@SUMEX-AIM.ARPA
cc: mak@SU-SIERRA.ARPA, i.ikabod9%LOTS-A@SU-SCORE.ARPA, yeager@SUMEX-AIM.ARPA
Ladies and Gentlemen,
I have decided to give my "second" talk first. Firstly, because it is
shorter than the first talk; and secondly, I am still waiting from some
experimental results I am doing with the static allocation problem. Anyway,
here is the abstract for the second talk.
Abstract
Resouce Management in a Distributed Multiprocesing Environment
... running x/czar/care/simple
Prelude and Second Movement: "Divertimento"
A Run-time Allocation System
Given an infinite supply of processing elements connected in any
scalable topology and a distributed concurrent problem solving scheme, the
speed up obtained is still critically dependent on the system's ability to
capure locality in its communications and balance the load evenly among the
processing elements. The more communications, the lower the connectivity of
the network, the more critical is this dependence.
A distributed resourcemanagement system that makes proper use of
i. a description (prediction) of program behavior
ii. machine charcteristics
iii. an intelligent run-time allocation system
iv. and a minimal overhead reclamation scheme
captures reference locality and considers proper trade offs between
communication and concurrency to achieve close-to optimal load-balancing and
maximum speed up.
"Divertimento" the run-time allocation system is responsible for
i. monitoring system load
ii. make placement recommendation at run-time
for dynamically created agents
iii. initiate load balancing actions when appropriate
The talk begins by a survey of five load-balancing proposals. Then,
the problem is decomposed into three sub-problems. The presentation will
proceed how these sub-problems applies in the context of CARE. A set of
experiment plan follows.
This talk is going to be long. Please sleep early tuesday night!!
Looking forward for your comments.
=jerry=
-------
∂20-May-85 2255 @SU-SCORE.ARPA:WIEDERHOLD@SUMEX-AIM.ARPA Re: MacFest Reminder
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 20 May 85 22:55:28 PDT
Received: from SUMEX-AIM.ARPA by SU-SCORE.ARPA with TCP; Mon 20 May 85 22:50:21-PDT
Date: Mon 20 May 85 22:35:09-PDT
From: Gio Wiederhold <WIEDERHOLD@SUMEX-AIM.ARPA>
Subject: Re: MacFest Reminder
To: TAJNAI@SU-SCORE.ARPA
cc: REGES@SU-SCORE.ARPA, faculty@SU-SCORE.ARPA
In-Reply-To: Message from "Carolyn Tajnai <TAJNAI@SU-SCORE.ARPA>" of Mon 20 May 85 14:55:33-PDT
A non-MACFEST Apple question:
For a demo by the speaker this coming friday in CS345 (3 pm) we would
like to have an APPLE 2+ with a colorcard (I could provide a monitor
with a 9-pin jack)
Any ideas?
PS I am doing some consulting with Larry Tesler at APPLE, will
bring it up. This is not their best time.
Gio
-------
∂20-May-85 2330 @SU-SCORE.ARPA:FEIGENBAUM@SUMEX-AIM.ARPA strategic computing project funding
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 20 May 85 23:30:04 PDT
Received: from SUMEX-AIM.ARPA by SU-SCORE.ARPA with TCP; Mon 20 May 85 23:26:15-PDT
Date: Mon 20 May 85 23:25:48-PDT
From: Edward Feigenbaum <FEIGENBAUM@SUMEX-AIM.ARPA>
Subject: strategic computing project funding
To: faculty@SU-SCORE.ARPA
Does anyone besides me have current funding (or expect to have funding
in the next fiscal year) from the DARPA Strategic Computing Program?
The funding is in trouble at the congressional level, and I want to gather
some facts.
Ed
-------
∂21-May-85 0122 BRODER@SU-SCORE.ARPA Speaker needed this week
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 21 May 85 01:22:21 PDT
Date: Tue 21 May 85 01:18:18-PDT
From: Andrei Broder <Broder@SU-SCORE.ARPA>
Subject: Speaker needed this week
To: AFLB.ALL@SU-SCORE.ARPA
Help! Help! Help! - Andrei
-------
∂21-May-85 1008 WRIGHT@SUMEX-AIM.ARPA SIGLunch: Friday, May 24, 1985
Received: from SUMEX-AIM.ARPA by SU-AI.ARPA with TCP; 21 May 85 10:08:22 PDT
Date: Tue 21 May 85 10:03:38-PDT
From: Carol Wright/Susie Barnes <WRIGHT@SUMEX-AIM.ARPA>
Subject: SIGLunch: Friday, May 24, 1985
To: siglunch@SUMEX-AIM.ARPA
SIGLUNCH
DATE: Friday, May 24, 1985
LOCATION: Braun Auditorium, Mudd/Chemistry Bldg.
TIME: 12:05
SPEAKER: Mike Hewett
Knowledge Systems Laboratory
TITLE: MARCK: A Module for Acquiring and Refining
Control Knowledge in BB1
MARCK is the first step in exploiting the capabilities for
knowledge acquisition and learning within the BB1 blackboard
architecture. MARCK automates the role of a BB1 knowledge
engineer by assisting the expert with defining, formulating, and
integrating new control knowledge into a knowledge base during
a problem-solving session.
While running a system in BB1, an expert may take an action
which indicates that some piece of control knowledge is missing.
MARCK uses its knowledge of the control architecture in BB1 to
guide the expert toward identifying the missing knowledge. It
then uses a set of data analysis routines and a library of generic
heuristic forms to postulate some applicable control heuristics.
Although MARCK has no knowledge of the domain, it can generate
domain-dependent control knowledge. MARCK communicates to the
expert through a translator which converts heuristic forms and
LISP code into English sentences.
MARCK is currently being used in the PROTEAN project at KSL
and several examples will be presented where MARCK has formulated
important and useful pieces of control knowledge.
-------
-------
∂21-May-85 1048 FORSTER@SU-CSLI.ARPA Phil. Dept. Colloquium
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 21 May 85 10:48:45 PDT
Date: Tue 21 May 85 10:44:39-PDT
From: Eckart Forster <FORSTER@SU-CSLI.ARPA>
Subject: Phil. Dept. Colloquium
To: folks@SU-CSLI.ARPA
PHILOSOPHY DEPARTMENT COLLOQUIUM
Speaker: Brian Smith (CSLI/Xerox)
Title: "Is Computation Formal?"
Time: Friday, May 24, 3:15
Place: Philosophy Seminar Room 90-92Q
-------
∂21-May-85 1220 BRODER@SU-SCORE.ARPA Forsythe Award
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 21 May 85 12:20:28 PDT
Date: Tue 21 May 85 11:34:06-PDT
From: Andrei Broder <Broder@SU-SCORE.ARPA>
Subject: Forsythe Award
To: CSD@SU-SCORE.ARPA, SU-BBOARDS@SU-SCORE.ARPA
Nominations are invited for the Eleventh George E. Forsythe Memorial
Award. This award, in recognition of *outstanding* student
contributions to the teaching of computer science at Stanford, is given
annually by the Computer Science Department, following the
recommendations of a committee composed of the past winners. (The
current committee is composed of Arthur Keller and Andrei Broder)
The award criteria guidelines require that a recipient exhibit
continued involvment as well as excellent achievement in the field of
teaching. Please include with your nomination an explanation of how
your nominee fulfills these requirements.
Please mail your contribution to broder@score.
Thank you,
The Forsythe Award Committee
-------
∂21-May-85 1349 YAN@SUMEX-AIM.ARPA 3602, 3674 and 3605
Received: from SUMEX-AIM.ARPA by SU-AI.ARPA with TCP; 21 May 85 13:49:32 PDT
Date: Tue 21 May 85 13:48:34-PDT
From: Jerry Yan <YAN@SUMEX-AIM.ARPA>
Subject: 3602, 3674 and 3605
To: aap@SUMEX-AIM.ARPA
cc: rosenbloom@SUMEX-AIM.ARPA, henager@SUMEX-AIM.ARPA
These machines have been tested and ephemeral garbage collection should run
properly
=jerry=
-------
∂21-May-85 1625 BSCOTT@SU-SCORE.ARPA Velo-Bind Machine
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 21 May 85 16:25:11 PDT
Date: Tue 21 May 85 16:10:58-PDT
From: Betty Scott <BSCOTT@SU-SCORE.ARPA>
Subject: Velo-Bind Machine
To: CSD@SU-SCORE.ARPA
cc: BScott@SU-SCORE.ARPA
Our Velo-Bind machine in MJH 321 has died. I really don't know whether or
not its use warrants buying another one, and will appreciate users' opinions
on whether to make such a purchase--the cost is about $1200. Word Graphics
on campus will Velo-bind reports, etc., for us. Is the convenience of having
one here in MJH worth the $1200 expenditure?
Will appreciate your comments.
Betty
-------
∂21-May-85 2044 SHAHN@SUMEX-AIM.ARPA 6.0
Received: from SUMEX-AIM.ARPA by SU-AI.ARPA with TCP; 21 May 85 20:43:21 PDT
Date: Tue 21 May 85 17:44:34-PDT
From: Sam Hahn <SHahn@SUMEX-AIM.ARPA>
Subject: 6.0
To: aap@SUMEX-AIM.ARPA
cc: rosenbloom@SUMEX-AIM.ARPA, henager@SUMEX-AIM.ARPA
3673 is now running release 6.0. -- sam
-------
∂21-May-85 2125 ullman@diablo meeting
Received: from SU-AIMVAX.ARPA by SU-AI.ARPA with TCP; 21 May 85 21:25:36 PDT
Date: Tue, 21 May 85 21:22:12 pdt
From: Jeff Ullman <ullman@diablo>
Subject: meeting
To: nail@diablo
Our meeting is Wednesday 11AM, 301MJH, as usual.
Richard Treitel is going to talk about MRS.
∂22-May-85 0015 @SU-SCORE.ARPA:ARK@SU-AI.ARPA CS Colloq, May 28: Matthew Ginsberg, Counterfactual Implication
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 22 May 85 00:15:08 PDT
Received: from SU-AI.ARPA by SU-SCORE.ARPA with TCP; Wed 22 May 85 00:01:02-PDT
Date: 22 May 85 0000 PDT
From: Arthur Keller <ARK@SU-AI.ARPA>
Subject: CS Colloq, May 28: Matthew Ginsberg, Counterfactual Implication
To: All-Colloq@SU-SCORE.ARPA, Faculty@SU-SCORE.ARPA
CC: ARK@SU-AI.ARPA
CS Colloquium, May 28, 4:15pm, Terman Auditorium
COUNTERFACTUAL IMPLICATION
Matthew Ginsberg
Computer Science Department
Stanford University
Counterfactuals are a form of commonsense non-monotonic inference that
has been of long-term interest to philosophers. In this talk, I
discuss the problem of deriving counterfactual statements from a
predicate calculus database, and present a formal description of this
derivation that allows the encoding of some context-dependent
information in the choice of a sublanguage of the logical language in
which we are working. The construction is formally identical to the
"possible worlds" interpretation due to David Lewis.
A concrete example is given which uses counterfactual implication for
the purpose of diagnosing digital hardware, and the talk concludes
with a discussion of other possible applications of counterfactuals in
artificial intelligence.
----
Cookies, thanks to Ginger, and possibly juice will probably be served in
the 3rd floor lounge of MJH at 3:45pm. See you there.
∂22-May-85 1042 @SU-SCORE.ARPA:M.SUSAN@SU-SIERRA.ARPA FLEX/32 presentation - today!
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 22 May 85 10:41:37 PDT
Received: from SU-SIERRA.ARPA by SU-SCORE.ARPA with TCP; Wed 22 May 85 10:28:59-PDT
Date: Wed 22 May 85 10:06:10-PDT
From: Susan Gere <M.SUSAN@SU-SIERRA.ARPA>
Subject: FLEX/32 presentation - today!
To: csd@SU-SCORE.ARPA, csl-faculty@SU-SIERRA.ARPA, csl-students@SU-SIERRA.ARPA
Visitors from Flexible Computer Corporation will make a presentation about
the FLEX/32 at 11 am Wednesday 5/22 in Terman 156 (and its overflow room).
This is the regularly scheduled session of EE482. Anyone wishing to hear
this presentation is welcome to attend (EE482 students have first dibs on
seats and table space of course).
Steve Lundstrom lundstrom@sierra
-------
∂22-May-85 1356 JODY@SU-CSLI.ARPA Burrito Bandito
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 22 May 85 13:55:22 PDT
Date: Wed 22 May 85 13:51:42-PDT
From: Joe Zingheim <JODY@SU-CSLI.ARPA>
Subject: Burrito Bandito
To: folks@SU-CSLI.ARPA
cc: consultants@SU-CSLI.ARPA
The Burrito Bandito comes again! Burritos are, once again, $3.00 and $3.75
for the regular and especial, and soft tacos are $1.75.
Carnitas (shredded pork) with rice
Pollo (chicken) with rice
Chile Verde (pork in green sauce) with rice
Chile Colorado (beef in red sauce) with rice
Lengua (beef tongue) with with rice
Chile Relleno (green chile stuffed with cheese) and rice
Picadillo (shreaded beef) with beans
Carne Asada (Mexican style steak) with beans
Chorizo (Mexican sausage) with beans
Send mail, as usual, to Lunch@CSLI
-------
∂22-May-85 1542 LIBRARY@SU-SCORE.ARPA New Books In The Math/CS Library
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 22 May 85 15:42:01 PDT
Date: Wed 22 May 85 15:37:12-PDT
From: C.S./Math Library <LIBRARY@SU-SCORE.ARPA>
Subject: New Books In The Math/CS Library
To: su-bboards@SU-SCORE.ARPA
cc: faculty@SU-SCORE.ARPA
STACS 85. 2nd Annual Symposium on Theoretical Aspects of Computer
Science. edited by k. Mehlhorn.QA75.5.S958 1985
Solving Elliptic Problems Using ELLPACK. by Rice and Boisvert. QA377.R531985
Software Validation: Inspection, Testing, Verification, Alternatives.
Proceedings of the Symposium on Software Validation. Sept. 1983
edited by Hausen. QA76.6.S91984 1983
Modeling And Analysis Of Computer Communications Networks. by Hayes
TK5105.5.H39 1984
PDE Software: Modules, Interfaces and Systems. edited by Engquist and Smedsaas.
QA377.I44 1983
The Engineering Of Numerical Software. by Miller. QA297.M527 1984
Security and Privacy. Proceedings of the Joint IBM/University of Newcastle
upon Tyne Seminar. Sept. 1984. ed. by B. Randell. (8512701)
IEEE First Interntional Conference on Office Automation. Proceedings.
(8506749)
Computer Operating Systems; for micros, minis, and mainframes. by D. Barron
QA76.6.B37 1984
System Software: an Introduction to Systems Programming. by L. Beck
QA76.B327 1985
Programming Principles; an introduction. by J. Motil.QA76.6.M69 1984 c.2
H. Llull
-------
∂22-May-85 1621 BETSY@SU-CSLI.ARPA NSF Visits
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 22 May 85 16:21:23 PDT
Date: Wed 22 May 85 16:16:02-PDT
From: Betsy Macken <BETSY@SU-CSLI.ARPA>
Subject: NSF Visits
To: researchers@SU-CSLI.ARPA
cc: administration@SU-CSLI.ARPA
Some of you may not have heard yet about the NSF personnel who will be
visiting CSLI Thursday, May 23, and Friday, May 24. Here is an update
and a request.
On Thursday, Erich Bloch, the Director of NSF, will be visiting CSLI
for about 45 minutes beginning about 2:15. He is coming to Stanford
mainly to visit CIS and speak at the dedication ceremonies for their
building. However, he asked to spend some time at CSLI. In view of
the brevity of his visit, we have prepared a short program involving
only the Director and a few senior researchers.
On Friday David Kingsbury, the Head of the Biological, Behavioral, and
Social Sciences Directorate at NSF, and Charles Brownstein, the
Director of the Division of Information Science and Technology, will
visit CSLI from 9 until 11 AM. They (especially Brownstein) know quite
a bit about the emerging interdisciplinary field of Information,
Computation and Cognition (the name selected by the NSF workshop
participants) from attending the NSF workshop and are conscious of the
need for centers like CSLI. To increase their knowledge of current
research in the field for the purpose of planning NSF directions, they
wish to see what is happening at CSLI and to talk to researchers about
their work.
Brownstein is visiting SRI in the afternoon; his interest in SRI is
wider than CSLI-related activities, but some of our researchers will
be part of that presentation.
We thought the best thing to do for the CSLI visit in the morning was
to have representatives of all our research efforts available and
ready to give 15-minute presentations (and demos if appropriate) about
their work. John, Tom, and I will give Kingsbury and Brownstein an
overview of the work going on here and then take them to talk to the
various research groups. They will probably not be able to see
everyone, but we think it's important for them to get a sense of the
size of the effort here and to have a choice about what they see.
We have contacted some of you to be sure that a representative
group would be on-site. But it would be great to have as many
of you available as would like to be here. If you can be here
and would be willing to give a brief presentation about your
work, would you let me know so I can be aware of all the
choices they would have.
Thanks.
Betsy
-------
∂22-May-85 1929 EMMA@SU-CSLI.ARPA Newsletter May 23, No. 30
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 22 May 85 19:29:20 PDT
Date: Wed 22 May 85 17:10:28-PDT
From: Emma Pease <Emma@SU-CSLI.ARPA>
Subject: Newsletter May 23, No. 30
To: friends@SU-CSLI.ARPA
Tel: 497-3479
C S L I N E W S L E T T E R
←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←
May 23, 1985 Stanford Vol. 2, No. 30
←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←
A weekly publication of The Center for the Study of Language and
Information, Ventura Hall, Stanford University, Stanford, CA 94305
←←←←←←←←←←←←
CSLI ACTIVITIES FOR *THIS* THURSDAY, May 23, 1985
12 noon TINLunch
Ventura Hall ``A Procedural Logic''
Conference Room Michael Georgeff (SRI and CSLI), Amy Lansky (SRI),
and Pierre Bessiere (SRI)
Discussion led by Michael Georgeff and Amy Lansky
2:15 p.m. CSLI Seminar
Redwood Hall ``Representations, Information, and the
Room G-19 Physical World'' by Ivan Blair
Discussion led by Meg Withgott
3:30 p.m. Tea
Ventura Hall
4:15 p.m. CSLI Colloquium
Redwood Hall ``Lexical Structure Constraints and Morphological
Room G-19 Parsing''
John McCarthy, AT&T Bell Laboratories
←←←←←←←←←←←
CSLI ACTIVITIES FOR *NEXT* THURSDAY, May 30, 1985
12 noon TINLunch
Ventura Hall ``Computers and Emotion''
Conference Room Discussion led by Helen Nissenbaum
(Abstract on page 2)
2:15 p.m. CSLI Seminar
Redwood Hall ``On Modelling Shared Understanding''
Room G-19 Jon Barwise, CSLI
(Abstract on page 2)
3:30 p.m. Tea
Ventura Hall
4:15 p.m. CSLI Colloquium
Redwood Hall ``Natural Kinds, Homeostasis, and the Limits of
Room G-19 Essentialism''
Richard Boyd, Prof. of Philosophy, Cornell University
(Abstract on page 2)
!
Page 2 CSLI Newsletter May 23, 1985
←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←
ABSTRACT OF NEXT WEEK'S TINLUNCH
``Computers and Emotions''
Emotion is an integral part of human consciousness. Yet common
practice in AI takes its ideal to be an intelligent, goal-driven agent
entirely devoid of passion. The assumption behind the practice is
that emotionless intellect is possible, and that a purely cognitive
agent is a valid abstraction from the total human individual. The
TINLunch, inspired by a TINLunch held October 13, 1981, titled ``Will
Robots Need Emotions?'', probes the AI assumption. I offer the
following as starting points for the discussion:
- What would it take to have a computer with emotions?
- Why worry about this? A passionless automaton is fully rational
and far better off for not having emotions. (Too bad we humans suffer
this affliction.)
- These are idle speculations. A sufficiently complex robot, that
could truly be said to understand and be goal-driven, will, of
necessity, have emotion, no matter what the intentions of its
creators.
Background readings are excerpts from Hume's ``Of the Passions'',
Jerome Shaffer's ``An Assessment of Emotion'' and TINLunch Outline
``Will Robots Need Emotions?'' by A. Archbold and N. Haas.
--Helen Nissenbaum
←←←←←←←←←←←←
ABSTRACT OF NEXT WEEK'S SEMINAR
``On Modelling Shared Understanding''
There are basically three approaches to understanding shared
understanding -- things like public information, common knowledge, and
mutual belief: the iterated attitude approach, the fixed point
approaches, and the shared environment approach. In this talk I will
discuss ways to model each of these, and show how the resulting models
are related. I will conclude with a brief discussion of applications
to language, deterrence and the Conway paradox. --Jon Barwise
←←←←←←←←←←←←
ABSTRACT OF NEXT WEEK'S COLLOQUIUM
``Natural Kinds, Homeostasis, and the Limits of Essentialism''
An account of naturalistic definitions is offered which applies to
a wide class of natural kinds, properties and relations. It agrees
with current naturalistic accounts in holding that natural kinds,
properties, etc., possess `a posteriori' naturalistic definitions:
that they are defined by ``real essences'' rather than by conventional
``nominal essences''. It departs from the picture suggested by the
definition ``water = H2O'' in that it holds that `a posteriori'
definitions are sometimes provided by partially indeterminate
homeostatic property clusters. In this regard, it resembles ordinary
language property cluster conceptions of definitions, except that the
unity of definitions is held to be causal rather than conceptual.
Several philosophical applications of the proposed account are
offered. In particular, it is argued that realism regarding kinds
with such real essences entails indeterminacy rather than bivalence,
that such real essences support a much narrower range of
counterfactuals than contemporary essentialism might suggest, and that
key philosophical notions like knowledge and reference are among those
which possess homeostatic cluster definitions. Implications for
philosophical methodology are explored. --Richard Boyd
!
Page 3 CSLI Newsletter May 23, 1985
←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←
SUMMARY OF THE RRR MEETING
A new seminar on the Role of Representation in Reasoning (``RRR'')
has been meeting since the beginning of the spring quarter. It
combines a variety of groups that met last spring to discuss similar
issues: F1, F3, F4, C3, and the special representation group that Meg
Withgott organized. The format is to have a one-hour presentation
followed by a full hour of discussion. So far this quarter, Stan
Rosenschein and Fernando Pereira have presented a discussion of
situated automata, including a suggestion as to how to include a
notion of representation in their theory; David Israel led a
discussion reexamining the notion of a Turing machine; and John Perry
led two sessions on his paper ``Circumstantial Attitudes and
Benevolent Cognition''. The RRR group meets at 2:15 each Tuesday in
the Ventura Hall seminar room.
On May 21, Brian Smith presented an overview of a general theory of
representation that he is developing, covering what we normally think
of as models, representations, perhaps in the end leading up even to
language. This theory will play a role in his project of developing
an account of embedded computation. Traditionally, representation
relationships are viewed in one of two ways: as forming a strict
hierarchy, like a hierarchy of meta-languages, with use/mention errors
resulting from a failure to distinguish each level. On the other
hand, other representation relationships, like that between a model
and what it models, are sometimes viewed as so close to an isomorphism
that the representation (model) and what is represented (modelled) are
identified. Brian sketched an account of a continuum of
representation relationships, ranging from strong isomorphisms up
through the complexities of language. He identified some increasingly
strong properties such relationships can have, including: absorption
(when a property or relation in the representation, such as linear
order, is used to represent exactly the same property in what is
represented; objectification, when a property or relationship is
represented by an object (as for example in predicate calculus when a
relation is signified by a predicate letter); what he called
``inexistence'', when an object's presence in the representation
signifies an absence in what is represented (such as traces in
linguistics, or ``eof'' signals in computation); and so on and so
forth.
Next week's RRR discussion will be led by Ned Block.
←←←←←←←←←←←←
SITUATION SEMANTICS MADE EASY
Three lectures by John Perry
Monday, Wednesday, Friday, June 3, 5, 7, at 3:15
Redwood G-19
The first lecture will be aimed at those who know nothing at all
about situation semantics.
←←←←←←←←←←←←
PHILOSOPHY DEPARTMENT COLLOQUIUM
``Is Computation Formal?''
Brian Smith, CSLI and Xerox PARC
Friday, May 24, 3:15, Philosophy Seminar Room, 90:92Q
!
Page 4 CSLI Newsletter May 23, 1985
←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←
NEW CSLI REPORT
Report No. CSLI--85--23, ``Querying Logical Databases'' by Moshe
Vardi, has just been published. This report may be obtained by
writing to David Brown, CSLI, Ventura Hall, Stanford, CA 94305 or
Brown@SU-CSLI.
-------
∂22-May-85 1931 @SU-SCORE.ARPA:berglund@Pescadero Systems Qual Proposal
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 22 May 85 19:30:41 PDT
Received: from Pescadero by SU-SCORE.ARPA with TCP; Wed 22 May 85 17:43:32-PDT
Date: Wed, 22 May 85 17:42:58 pdt
From: Eric Berglund <berglund@Pescadero>
Subject: Systems Qual Proposal
To: faculty@score, phd@score
As a big fan of the comp/qual scheme here (I'm not being sarcastic), I haven't
had too many complaints at all about the way I've been prepared to do research.
However, over the last couple months I've been thinking about my preparation
to do research in distributed systems and been a bit disappointed.
I understand that my education is my own responsibility and that I can fill
in any gaps I'm missing by taking classes or doing reading on my own. But
I had been trusting the faculty, through the comp/qual system, to force me
to prepare well. Even though I've passed the qual, however, I don't feel
well prepared--I feel like I've slipped through the cracks.
I believe the problem is not with my specific set of exams or examiners, but
with the systems qual. (Sorry about this all you AIers, AAers, NAers, and
MTCers.) I believe I would have been much better served if I'd been forced
to know much more about Distributed Systems--including very recent
literature--and less about Compilers and Hardware. But the systems qual asks
students to know a little about several areas of systems, rather than a lot
about one. (Actually, it seems likely that this is true on other quals, too,
though I don't know.) It seems to me that in the ideal world there would
be separate systems qual exams for Hardware, Distributed Systems, Compilers,
Databases, Operating Systems, and Programming Languages instead of the mix
and match we now have. (Maybe D.S. and O.S. would be combined, and Compilers
and Programming Languages, but you get the idea.)
I do understand that many of these areas relate to each other, and that
getting a broad overview is important. I also understand that the ideal
world would have an infinite supply of faculty time for writing different
exams. I am just looking for some scheme which will focus more directly
on preparing for research in a specific area. I guess I would even contend
that such a scheme would lower the average time to graduation, as students
would be forced earlier to narrow their interests to a specific field and
would then learn enough about that field to pick a thesis topic sooner.
--Eric
∂22-May-85 1933 @SU-SCORE.ARPA:pratt@cicero Systems Qual Proposal
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 22 May 85 19:33:16 PDT
Received: from cicero ([36.8.0.14].#Internet) by SU-SCORE.ARPA with TCP; Wed 22 May 85 18:35:44-PDT
Date: Wed, 22 May 85 18:31:49 pdt
From: Vaughan Pratt <pratt@cicero>
To: faculty@score, phd@score
Subject: Systems Qual Proposal
Distributed Systems is hardly alone here. What correlation is there
between any of the area quals and what someone needs to know who is
going to do say a Computer Graphics thesis under Leo Guibas, a Digital
Typography thesis under Chuck Bigelow, or a Logics of Programs thesis
under me. All these fields have gone far beyond the subject matter
currently comprehended by the quals. Worse, some of these fields now
have nearly a decade-old component (closer to two decades for some
of computer graphics) that is old hat to current workers yet that
still hasn't penetrated the qual system.
There's a depressingly large gap for the quals to close to better
serve those of our students working in several hot fields today.
-v
∂22-May-85 2028 @SU-SCORE.ARPA:ullman@diablo Re: Systems Qual Proposal
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 22 May 85 20:28:40 PDT
Received: from diablo by SU-SCORE.ARPA with TCP; Wed 22 May 85 20:25:12-PDT
Date: Wed, 22 May 85 20:24:39 pdt
From: Jeff Ullman <ullman@diablo>
Subject: Re: Systems Qual Proposal
To: faculty@SU-Score, phd@SU-Score, pratt@cicero
There is a lot of merit to the Burglund proposal.
I'm not sorry he had to learn a bit about things outside
distributed systems; he may yet find use for one or two
ideas by and by. However, there is a certain quantitative
difference in the scope of the SYSQUAL compared with other areas.
I'm trying to recall the 11 areas into which the COSERS report
divided the world of computer science, but most of the quals
represent at most one of the 11. AI and NA were chapters, if I
recollect, and AA and MTC each represent a fraction of the
Theory chapter. On the other hand, the SYSQUAL covers chapters on:
OS, Hardware, Programming languages, and Database systems,
i.e., 4 chapters, plus distributed systems which I don't
think was in COSERS (or was there a chapter on networking?)
∂22-May-85 2142 @SU-SCORE.ARPA:reid@Glacier Re: Systems Qual Proposal
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 22 May 85 21:42:31 PDT
Received: from Glacier by SU-SCORE.ARPA with TCP; Wed 22 May 85 21:36:03-PDT
Date: 22 May 1985 2134-PDT (Wednesday)
From: Brian Reid <reid@Glacier>
To: Eric Berglund <berglund@Pescadero>
Cc: faculty@score, phd@score
Subject: Re: Systems Qual Proposal
In-Reply-To: Eric Berglund <berglund@Pescadero> / Wed, 22 May 85 17:42:58 pdt.
I believe that this issue is just an artifact of the Stanford CS
department's being tremendously skewed towards theory and away from
systems. By comparison, the other extreme (CMU) lumps NA, AA, and MTC
together and gives a "Theory" qual, and then has separate Software,
Hardware, and Performance Analysis quals. At least they did when I was
there. (CMU's quals were Software, Hardware, Theory, AI, and Performance
Analysis. I understand they have merged Performance Analysis in with
Theory and now give only 4 quals.)
The right answer to this and many other problems is, and always has
been, the abolition of CSL and the merging of CSL faculty and interest
areas completely into CSD. Then we could think of dividing the systems
area into pieces. Already we are giving systems students the right to
chose any 3 of the 6 subareas of the systems qual (compilers,
databases, operating systems, networks/distributed systems, programming
languages, and something else I am forgetting).
Brian
∂23-May-85 0020 YM@SU-AI.ARPA Special seminar
Received: from SU-AIMVAX.ARPA by SU-AI.ARPA with TCP; 23 May 85 00:20:12 PDT
Received: from SU-AI.ARPA by diablo with TCP; Thu, 23 May 85 00:01:25 pdt
Date: 23 May 85 0000 PDT
From: Yoni Malachi <YM@SU-AI.ARPA>
Subject: Special seminar
To: bayboards@SU-AIMVAX.ARPA, friends@SU-CSLI.ARPA, nail@SU-AIMVAX.ARPA,
prolog@SU-SCORE.ARPA
Cc: tuley@SU-SCORE.ARPA, lindstrom@UTAH-20.ARPA
Thursday 6-6-85, 11am in MJH 352
FUNCTIONAL PROGRAMMING AND THE LOGICAL VARIABLE
Gary Lindstrom
Department of Computer Science
University of Utah
Salt Lake City, Utah 84112
Logic programming offers a variety of computational effects which go beyond
those customarily found in functional programming languages. Among these
effects is the notion of the "logical variable," i.e. a value determined by the
intersection of constraints, rather than by direct binding. We argue that this
concept is "separable" from logic programming, and can sensibly be incorporated
into existing functional languages. Moreover, this extension appears to
significantly widen the range of problems which can efficiently be addressed in
function form, albeit at some loss of conceptual purity. In particular, a form
of side-effects arises under this extension, since a function invocation can
exert constraints on variables shared with other function invocations.
Nevertheless, we demonstrate that determinacy can be retained, even under
parallel execution. The graph reduction language FGL is used for this
demonstration, by being extended to a language FGL+LV permitting formal
parameter expressions, with variables occurring therein bound by unification.
The determinacy argument is based on a novel dataflow-like rendering of
unification. In addition the complete partial order employed in this proof is
unusual in its explicit representation of demand, a necessity given the "benign"
side-effects that arise. An implementation technique is suggested, suitable for
reduction architectures.
∂23-May-85 0044 RESTIVO@SU-SCORE.ARPA PROLOG Digest V3 #24
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 23 May 85 00:43:49 PDT
Date: Wednesday, May 22, 1985 7:56PM
From: Chuck Restivo (The Moderator) <PROLOG-REQUEST@SU-SCORE.ARPA>
Reply-to: PROLOG@SU-SCORE.ARPA
US-Mail: P.O. Box 4584, Stanford CA 94305
Phone: (415) 326-5550
Subject: PROLOG Digest V3 #24
To: PROLOG@SU-SCORE.ARPA
PROLOG Digest Thursday, 23 May 1985 Volume 3 : Issue 24
Today's Topics:
Query - Common Lisp Prologs & External calls,
Implementation - CP,
Announcement - 1985 LP Symposium & Wiezmann Workshop
----------------------------------------------------------------------
Date: Mon 20 May 85 13:07:48-EDT
From: Vijay <Vijay.Saraswat@CMU-CS-C.ARPA>
Subject: Common Lisp Prologs.
I am interested in finding out if there are any
commercially available Prolog programming
environments (compiler/interpreter/debugger/editor)
written in CommonLisp.
Also, I am interested in comments from anyone who has
worked with Symbolics Prolog (I am not referring to Ken
Kahn's implementation) with regard to speed.
Cheers,
-- Vijay.
------------------------------
Date: Mon, 20 May 85 14:06:59 pdt
From: Allen VanGelder <AVG@Diablo>
Subject: External calls
We are in a similar situation at Stanford, and
would appreciate copies of any code that does
external calls.
Thank you.
------------------------------
Date: Sun, 19 May 85 03:22:37 cdt
From: (Raghunath Ramakrishnan) Raghu@UT-sally.ARPA
Subject: The Annotation '%'
This is a continuation of the discussion of the read-only annotation
in CP. While the current definition (disallowing '?' in guards)
eliminates some of the ambiguities in the definition of
'?', it is still difficult to undertand in programs because
it can be passed on at runtime, and further, the unification
is order-dependent.
The '%' annotation is an attempt to provide the same power (i.e. any
programming techniques in CP should be available) but with a semantic
that is static and independent of the order of unification.
The following is a definition and brief description of '%'.
A more complete description can be found in a forthcoming UT
Austin Tech Report. If anyone is interested in a copy,
please send me your address (Mail it to Raghu@UT-Sally.
Responses may be delayed since I'll be absent for about a month).
The clauses in a program are viewed as templates which can be used
to create processes. A call is a specification for a
required process and we attempt to create such a process by
matching the call with some clause head and satisfying the
corresponding guard.
The annotation '%' has the same semantics whether it
annotates a variable in a call or in a clause head. It is
a guarantee that the process in which it appears (it
may appear either in the call which specifies the required
process or a clause head being matched in an attempt to
spawn such a process) will not instantiate it to a
non-variable. In other words, such an instantiation
must be due to some other process (a sibling of the
call or a sibling of an ancestor), and this fact can be
used to make the process wait for input (which may in fact
decide the nature of this process since attempts to match
the call against a clause head may be delayed until the
input is available). These semantics are guaranteed
regardless of the order of unification.
The unification rules are defined by this table:
| HEAD
--|---------------------------------------------------------------
| TERM VAR TERM% VAR%
|
G |TERM Unify(GL,HD) HD = Ref(GL) Fail Fail
O |
A |VAR GL = Ref(HD) HD = Ref(GL) Fail Fail
L |
|TERM% Fail Fail Unify(GL,HD) HD = Ref(GL)
|
|VAR% Fail Fail Suspend HD = Ref(GL)
Note that the '%' annotation is never passed on due to unification.
Also, like CP's read-only annotation, it is not inherited.
'%' annotations in the call are used when the corresponding
arguments in the target clause head are annotated, and thus
ensure that a clause is being used correctly (with respect
to its annotations).
The intended semantics for '%' can be achieved with these
rules only if some syntactic restrictions are met. The
idea behind these restrictions is to prevent this process
from instantiating a %-annotated variable during the process
of matching the call against the head. Compile time
checking can be used to enforce all these restrictions. We
summarise them below:
1. Each variable occurrence in a clause head is unique.
2. A %-annotated variable in a clause head is only used
as a %-annotated variable throughout that clause.
3. All superior terms of a %-annotated term or variable in
a clause head are %-annotated. (Given the rule that only
annotated variables or terms unify, this in
effect requires superior terms of all annotated
terms or variables - in both calls and clause
heads - to be annotated.)
4. Each %-annotated variable on the right hand side of a
clause that does not appear in the clause head is:
a. Used consistently in each call (ie. it is always
%-annotated or always unannotated).
b. Appears in at least one call without being annotated.
If a given program passes these checks, the unification algorithm
described above ensures the intended semantics. If a process
suspends, it is waiting for some other process to instantiate a
%-variable. Incidentally, 4.b above detects one obvious instance
where such a wait could be futile. The problems involved in detecting
such failures are the familiar starvation and deadlock problems of
traditional process models since we have eliminated the side effects
of unification which enable a process to unexpectedly 'feed' itself!
The only other way in which the annotations intrude (in the
sense that they sometimes lead to unexpected situations) is
that a call may fail because no clause head matching the
annotations is found. We view this as being in the same
class of errors as there not being a clause head with the
same functor as the call or the same number of arguments
since the annotations are as much a part of the static
description of a clause as the functor name or number of
arguments.
The Programming Techniques Supported by '%'
The %-annotation supports all the programming techniques
that the read-only annotation in Concurrent Prolog does. It
does not allow us to create protected channel variables
(ie. specify that a process always expects to find a
given argument to be a variable), and it is not intended
to do so. We note that the current definition of the
read-only annotion in CP, as opposed to the original
definition, does not permit this either.
The unification algorithm described here will be
implemented shortly in Prolog, along with a compile-time
checking routine.
Several examples are included in a separate posting. These
discuss the decisions made in the definition of '%', the
syntactic restrictions required to guarantee its semantics,
and the programming techniques it supports.
-- Raghu
------------------------------
Date: Mon, 20 May 85 18:38:04 -0200
From: Ehud Shapiro <Udi%Wisdom@WISCVM.ARPA>
Subject: Announcement
The Second Weizmann Institute Workshop
on
Logic Programming and Parallel Processing
The Weizmann Institute of Science
June 4-5, 1985
(preliminary program)
Tuesday, June 4:
Morning Welcome
The Flat Concurrnt Prolog compiler and system:
A Status Report.
Ehud Shapiro, Weizmann Institute of Science
ICOT's project: a trip report and critical evaluation.
Marc Snir, Hebrew University
Adapting Warren's abstarct machine to Flat Concurrent
Prolog. Avshalom Houri, Weizmann Institute of Science
High Level Characteristics and Specification
of a Logic Computer System.
Toni Kusalik, University of British Columbia
Afternoon
MOS: Unix on a multiprocessor.
Amnon Barak, Hebrew University
Quadtrees in Concurrent Prolog.
Shimon Edelman, Weizmann Institute of Science
TBA
Mats Carlson, Uppsala University
Solution-stream based semantics for Prolog.
Nissim Francez, Technion, and
Amir Pnueli, Weizmann Institute of Science
Concurrent Prolog for VLSI design simulation
David Veinbaum, Tel Aviv University
Wednesday, June 4:
Morning
Meta-interpreters for real, or the magic of
partial evaluation.
Ehud Shapiro, Weizmann Institute of Science
An FCP partial evaluator.
Shmuel Safra, Weizmann Institute of Science
TBA
Ran Ginosar, Technion
DARPA's strategic computing project.
Robert Kahn, DARPA
Afternoon
A Concurrent Prolog based region finding algorithm.
Lisa Hellerstein, UC Berkeley
Compiling Or-parallelism into And-parallelism.
Michael Codish, Weizmann Institute of Science
TBA
Oded Shmueli, Technion
An abstract machine for Ueda's language of
Guarded Horn Clauses.
Jacob Levi, Weizmann Institute of Science
------------------------------
Date: 17 May 1985 11:54:34-PDT
From: Conery%UOregon@csnet-relay.arpa
Subject: 1985 Symposium
-- Preliminary Schedule --
-- 1985 Symposium on Logic Program --
For more information, contact:
John Conery (JC@UOregon.csnet) 503-686-4408
Jacques Cohen (JC@Brandeis.csnet) 617-647-3370
MONDAY, July 15
Tutorials (Two parallel sessions, 9:00 - 4:30)
(Two more tutorials on Friday)
Expert Systems Using Prolog, Bob Kowalski, Imperial College
Parallel Logic Programming, Gary Lindstrom, Univerity of Utah
8:00 p.m. - 10:00 p.m. Early Registration and Reception
TUESDAY, July 16, 1985
8:00 a.m. - 9:00 a.m. Registration
9:00 a.m. - 10:30 a.m. KEYNOTE SPEAKER: Robert Kowalski,
Imperial College London, England
11:00 a.m. - 12:30 p.m. PARALLELISM
Semi-intelligent Backtracking of Prolog
Based on a Static Data Dependency
Analysis, Jung-Herng Chang and
Alvin M. Despain, University of California,
Berkeley
User-defined Parallel Control Strategies,
J. I. Glasgow, M. A. Jenkins, and C. D.
McCrosky,Queen's University, Canada
AND-parallelism with Intelligent
Backtracking for Annotated Logic
Programs, J. Maluszynski, Linkoping
University and P. Dembinski, Chalmbers
University of Technology, Sweden
12:30 p.m. - 2:00 p.m. Lunch
2:00 p.m. - 3:30 p.m. EXTENSIONS
An experiment in Programming with Full
First-Order Logic, Zerkis D.
Umigar and Vijay Pitchumani,
Syracuse University
A Meta-Level Extension of Prolog,
Kenneth A. Bowen, Syracuse
University, and Tobias Weinberg, Digital
Equipment Corporation
Logic Programming Cum Applicative
Programming, Nachum Swjodershowitz and
David Plaisted, University of
Illinois at Urbana Champaign
3:30 p.m. - 4:00 p.m. Coffee Break
4:00 p.m. - 5:30 p.m. LANGUAGE ISSUES
On the Treatment of Cuts in Prolog Source-Level
Tools, R. A. O'Keefe, University of Edinburgh,
United Kingdom
All Solutions Predicates in Prolog, Lee Naish,
University of Melbourne, Australia
Unification-free Execution of Logic Programs,
Jan Komorowski, Aiken Computation Laboratory,
Harvard University and Jan Maluszynski,
Linkoping University, Sweden
Wednesday, July 17, 1985
9:00 a.m. - 10:30 a.m. INVITED SPEAKER
Dr. Herve Gallaire, European Computer Research
Center, Munich, West Germany
10:30 a.m. - 11:00 a.m. Coffee break
11:00 a.m. - 12:30 p.m. CONCURRENT PROLOG
Concurrent Prolog in a Multi-process Environment,
Rosanna Lee and Randy Goebel, University of
Waterloo, Canada
A sequential Implementation of Concurrent Prolog
Based on the Shallow Binding Scheme, Toshihiko
Miyazaki, Akikazu Takeuchi and Takashi Chikayama,
ICOT, Japan
12:30 p.m. - 2:00 p.m. Lunch
2:00 p.m. - 3:30 p.m. SEMANTICS
The Declarative Semantics of Logical Read-only
Variables, G. Levi and C. Palamidessi,
Universita'di Pisa, Italy
Narrowing as the Operational Semantics of Functional
Languages, Uday Reddy, University of Utah
Towards an Algebra for Constructing Logic Programs,
R.A. O'Keefe, University of Edinburgh, United
Kingdom
3:30 p.m. - 4:00 p.m. Coffee Break
4:00 p.m. - 5:30 p.m. IMPLEMENTATION ISSUES
A Microcoded Unifier for Lisp Machine Prolog, Mats
Carlssn, Uppsala University, Sweden
SLOG: A Logic Programming Language Interpreter Based
on Clausal Superposition and Rewriting, Laurent
Fribourg, Laboratoires de Marcoussis, France
A Real Time Garbage Collector for Prolog, Edwin
Pittombils and Maurice Bruynooghe, K.U. Leuven,
Belgium
Evening Banquet Speaker
Dr. Maurice Wilkes, Digital Equipment Corporation
Thursday, July 19, 1985
9:00 a.m. - 10:30 a.m. THEORY
recursive Unsolvability of Determinacy, Solvable
Cases of Determinacy, and Their Applications to
Prolog Optimization, Jajime Sawamura and Taku
Takeshima, International Institute for Advanced
Study of Social Information Science, Japan
Graph-based Logic Programming Interpreters, Jean
Gallier and Stan Raatz, University of Pennsylvania
Surface Deduction: A Uniform Mechanism for Logic
Programming, P. T. Cox and T. Pietrzykowski,
Technical University of Nova Scotia, Canada
10:30 a.m. - 11:00 a.m. Coffee Break
11:00 a.m. - 1:00 p.m. SPECIAL TOPICS
Towards a Programming Environment for Large
Prolog Programs, Jan Chomicki and Naftaly H.
Minsky, Rutgers University
Modular Logic Programming of Compilers, Harald
Ganzinger and Michael Hanus, University Dortmund,
West Germany
An(other) Integration of Logic and Functional
Programming, Amitabh Srivastava, Don Oxley and
Aditya Srivastava, Central Research Laboratories,
Texas Instruments, Inc.
A Technique for Doing Lazy Evaluation in Logic,
Sanjai Narain, Rand Corp.
FRIDAY July 19
Tutorials: (Two concurrent sessions, 9:00 - 4:30)
"Concurrent Logic Programming Techniques",
Ehud Shapiro, Weizmann Institute
"Prolog and Software Engineering",
Susan Gerhart, Wang Institute.
------------------------------
End of PROLOG Digest
********************
∂23-May-85 0223 BRODER@SU-SCORE.ARPA AFLB today
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 23 May 85 02:23:33 PDT
Date: Thu 23 May 85 02:22:27-PDT
From: Andrei Broder <Broder@SU-SCORE.ARPA>
Subject: AFLB today
To: aflb.all@SU-SCORE.ARPA
cc: tuley@SU-SCORE.ARPA, su-bboards@SU-SCORE.ARPA
5/23/85 - Thanasis Hadzilacos (University of Athens):
Multiversion schedulers are now a widely accepted method to enhance
the performance of the concurrency control component of a database. A
typical multiversion concurrency control algorithm keeps several
versions of each entity. When a read request arrives, the algorithm
makes available to it not necessarily the latest version of the entity
(as would an ordinary concurrency control algorithm) but the one which
best serves our ultimate goals: correctness and performance.
Consequently, any multiversion algorithm, besides deciding at each
moment whether to grant, delay, or abort an arriving read or write
step of a transaction (as all concurrency control algorithms do), must
also decide which of the existing versions to present to a read step,
and which, if any, to overwrite. These latter decisions, particular
to the multiversion concurrency control, constitute the main added
complexity of this approach. In this work we introduce a new notion
of multiversion serializability (MVSR) based on conflicts (MVCSR), and
discuss its relation with the well known single version conflict
serializability (CSR). On-line schedulable (OLS) subsets of MVSR were
defined in a paper by Papadimitriou and Kanellakis. We prove here
that it is NP-complete to decide whether a set of schedules is OLS.
We next introduce the concept of maximal OLS sets, and show that no
efficient scheduler can be designed that recognizes maximal subsets of
the MVSR or MVCSR schedules -unless P=NP. Finally we present a
general framework for algorithms based on MVCSR and argue that it is
the most general set of algorithms based on the MVCSR concept that can
be implemented efficiently.
(This is joint work with Christos Papadimitriou.)
***** Time and place: May 23, 12:30 pm in MJ352 (Bldg. 460) ******
-------
∂23-May-85 0825 EMMA@SU-CSLI.ARPA Special seminar
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 23 May 85 08:25:33 PDT
Return-Path: <YM@SU-AI.ARPA>
Received: from SU-AI.ARPA by SU-CSLI.ARPA with TCP; Thu 23 May 85 00:00:14-PDT
Date: 23 May 85 0000 PDT
From: Yoni Malachi <YM@SU-AI.ARPA>
Subject: Special seminar
To: bayboards@SU-AIMVAX.ARPA, friends@SU-CSLI.ARPA,
nail@SU-AIMVAX.ARPA, prolog@SU-SCORE.ARPA
CC: tuley@SU-SCORE.ARPA, lindstrom@UTAH-20.ARPA
ReSent-Date: Thu 23 May 85 08:22:20-PDT
ReSent-From: Emma Pease <Emma@SU-CSLI.ARPA>
ReSent-To: friends@SU-CSLI.ARPA
Thursday 6-6-85, 11am in MJH 352
FUNCTIONAL PROGRAMMING AND THE LOGICAL VARIABLE
Gary Lindstrom
Department of Computer Science
University of Utah
Salt Lake City, Utah 84112
Logic programming offers a variety of computational effects which go beyond
those customarily found in functional programming languages. Among these
effects is the notion of the "logical variable," i.e. a value determined by the
intersection of constraints, rather than by direct binding. We argue that this
concept is "separable" from logic programming, and can sensibly be incorporated
into existing functional languages. Moreover, this extension appears to
significantly widen the range of problems which can efficiently be addressed in
function form, albeit at some loss of conceptual purity. In particular, a form
of side-effects arises under this extension, since a function invocation can
exert constraints on variables shared with other function invocations.
Nevertheless, we demonstrate that determinacy can be retained, even under
parallel execution. The graph reduction language FGL is used for this
demonstration, by being extended to a language FGL+LV permitting formal
parameter expressions, with variables occurring therein bound by unification.
The determinacy argument is based on a novel dataflow-like rendering of
unification. In addition the complete partial order employed in this proof is
unusual in its explicit representation of demand, a necessity given the "benign"
side-effects that arise. An implementation technique is suggested, suitable for
reduction architectures.
∂23-May-85 1523 SELLS@SU-CSLI.ARPA end of quarter/year party
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 23 May 85 15:22:53 PDT
Date: Thu 23 May 85 15:18:07-PDT
From: Peter Sells <Sells@SU-CSLI.ARPA>
Subject: end of quarter/year party
To: folks@SU-CSLI.ARPA
CSLI would like to invite you and yours to a party in the celebration of
the end of the quarter/school-year, to take place from 6.30pm to 1am on
Friday June 14th. The party will be held at the Bechtel International
Center, and we plan to have barbecue and volleyball followed by music and
dancing and the Dead Tongues.
Further details will be sent out soon--CSLI will provide drinks, but we
need people to (a) bring stuff to barbecue (b) bring barbecue fuel etc. (c)
do the barbecuing.
Please send comments, suggestions, offers of help, etc. to me. We hope
everyone will be able to attend.
Peter
-------
∂23-May-85 1539 @SU-SCORE.ARPA:EENGELMORE@SUMEX-AIM.ARPA Showing of KNOWLEDGE ENGINEERING Movie - 5/28
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 23 May 85 15:38:52 PDT
Received: from SUMEX-AIM.ARPA by SU-SCORE.ARPA with TCP; Thu 23 May 85 15:15:41-PDT
Date: Thu 23 May 85 15:12:05-PDT
From: Ellie Engelmore <EENGELMORE@SUMEX-AIM.ARPA>
Subject: Showing of KNOWLEDGE ENGINEERING Movie - 5/28
To: csd@SU-SCORE.ARPA
Tuesday, May 28
5:30pm
Terman Auditorium
At the conclusion of the CS Colloquium (CS300) on Tuesday, May 28,
there will be a showing of a short 16mm movie produced by the
Heuristic Programming Project (now known as the Knowledge Systems
Laboratory) in collaboration with the Stanford Department of
Communications.
KNOWLEDGE ENGINEERING:
Artificial Intelligence Research
at the
Stanford Heuristic Programming Project
This film illustrates the current state of knowledge-based system
research in the laboratory, the transfer of this technology into
diverse applications in society, and the significant research issues
that have to be addressed for the 1980's and 1990's.
It begins with an introduction to the methodology developed at the
laboratory for discovering and working on basic problems of artificial
intelligence. The process of knowledge engineering and the structure
of expert systems are explored.
Several projects are described, including the following:
KBVLSI: An expert system for the design of very large-scale integrated
systems
Protean: An expert system to determine the chemical structure of
proteins in solution
ONCOCIN: A system utilized to help manage patients who are on a
particular chemotherapy protocol
-------
∂23-May-85 1612 SCHOLZ@SU-SCORE.ARPA 2nd semiannual postcomp 7k
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 23 May 85 16:12:22 PDT
Date: Thu 23 May 85 16:04:33-PDT
From: Karen Scholz <SCHOLZ@SU-SCORE.ARPA>
Subject: 2nd semiannual postcomp 7k
To: csd@SU-SCORE.ARPA
THE 2ND SEMIANNUAL POSTCOMP 7K
you know you'll need it. after the exam and before the beer.
pound the pavement. let the quadriceps kick in and shoot for zero
percent cerebral activation level.
bring your nikes and your shorts to the exam on sunday.
we'll meet, dressed out, in front of jax at 5:45pm.
we'll choose a pace which will allow some leaping and dancing along the
way. depending upon who shows up, we may form multiple groups of
cardiovascularly compatible runners.
non-comptakers are definitely invited, provided they can either
imagine or remember what it feels like to finish the comp.
see you there, o fleet of foot and soon-to-be light of heart!
-------
∂24-May-85 0538 RESTIVO@SU-SCORE.ARPA PROLOG Digest V3 #25
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 24 May 85 05:38:41 PDT
Date: Thursday, May 23, 1985 7:55PM
From: Chuck Restivo (The Moderator) <PROLOG-REQUEST@SU-SCORE.ARPA>
Reply-to: PROLOG@SU-SCORE.ARPA
US-Mail: P.O. Box 4584, Stanford CA 94305
Phone: (415) 326-5550
Subject: PROLOG Digest V3 #25
To: PROLOG@SU-SCORE.ARPA
PROLOG Digest Friday, 24 May 1985 Volume 3 : Issue 25
Today's Topics:
Query - Qualitative Reasoning,
Announcement - Seminar (Stanford),
Implementation - '%' Notation,
LP Library - Berkeley PLM Benchmarks
----------------------------------------------------------------------
Date: Mon, 20-May-85 09:52:10 PDT
From: (David Sherman) pesnta!dave@UCB-Vax
Subject: Suggestions needed for tax rules
I am trying to design a system which will apply the rules of
the Income Tax Act (Canada) to a set of facts and transactions
in the area of corporate reorganizations.
This is not a typical "AI and law" problem, because the Income
Tax Act is highly technical and extremely specific about what
rules apply and when. To the extent there is "open texture", or
issues which require legal judgment (e.g., whether an amount
is "reasonable"), I assume the lawyer using the system will
provide the judgment as an input fact. The problems are therefore
quite different from those addressed by McCarty@Rutgers' TAXMAN
program, because the approaches of the Income Tax Act and the
Internal Revenue Code are very different.
The difficulty in programming the system is simply the complexity
of the Income Tax Act. On a given transaction, a large number of
rules have to be examined to determine possible tax effects. Some
of these rules create new facts (e.g., a deemed dividend or a deemed
disposition). The passage of time is very important: steps happen
in a particular sequence, and the "state of the world" at the moment
a step is taken is crucial to determining the tax effects. In the
context of corporate transactions, this state includes such things
as who controls a corporation; who owns what assets; the residence
and status of and relationships among taxpayers; cost bases and
proceeds of disposition; and so on.
(My background to this: I'm a tax lawyer and an experienced C
programmer. I'm doing this work towards an LL.M. thesis.)
I've tried several approaches to this field before. Last year
I did a small version in C which uses an event-driven simulation,
and as it encounters each event calls a function for each rule in
the database to generate new events and tax results. (Incidentally,
if anyone wants a copy of that paper, "Towards a
Comprehensive Computer-Based Problem Solving Model of the
Income Tax Act: A Suggested Approach and Implementation of
Examples from Corporate Reorganizations", let me know.)
One of the problems with the C implementation was designing the
order in which the rules should be applied. For even the small subset
I implemented, this was awkward and difficult.
I came across Prolog at an ICS course on expert systems a few
weeks ago. It looks like the perfect tool for much of what I want
to do. Besides the workshops at the course (taught=sdcsvax!vis!greg),
I've now read Clocksin & Mellish and skimmed through How To Solve
It In Prolog.
The definitional stuff is fine. Using C-Prolog 1.4 on a VAX (not this
machine), I've already coded the Income Tax Act's definitions for
things like "resident", "private corporation", etc.
The problem I have (if you've read this far, thank you!) is how
to deal with ←time←. I suppose the overall model I need to work with
is one of "changing states", where a transaction (e.g., X transfers
property to corporation Y in exchange for shares in Y) changes the
world from state A to state B, and the rules can examine differences
between the states to determine the effects. But then how does that
jive with definitional tests which may need to look back in time
(e.g., a corporation is resident in Canada in a given year if it
meets certain conditions and it was resident in Canada, or carried
on business in Canada, in the previous year. This is a perfect
recursive definition for Prolog, and I've implemented the rule,
but only as a fixed definition, not as part of a "changing
state" system.)?
I guess part of my problem is that I'm not working in an
AI department, or even a Computer Science department, and
so I don't have knowledgeable people to bounce my ideas off.
If you have any suggestions as to approaches I should be taking,
I'd greatly appreciate hearing from you. I'd be happy to send you
a copy of the code I've written so far to explain what I mean.
(You could even learn all about Canadian tax law!)
-- David Sherman
The Law Society of Upper Canada
Osgoode Hall
Toronto, Canada M5H 2N6
(416) 947-3466
------------------------------
Date: 23 May 85 0000 PDT
From: Yoni Malachi <YM@SU-AI.ARPA>
Subject: Special seminar
Thursday 6-6-85, 11am in MJH 352
FUNCTIONAL PROGRAMMING AND THE LOGICAL VARIABLE
Gary Lindstrom
Department of Computer Science
University of Utah
Salt Lake City, Utah 84112
Logic programming offers a variety of computational effects
which go beyond those customarily found in functional
programming languages. Among these effects is the notion of
the "logical variable," i.e. a value determined by the
intersection of constraints, rather than by direct binding.
We argue that this concept is "separable" from logic
programming, and can sensibly be incorporated into existing
functional languages. Moreover, this extension appears to
significantly widen the range of problems which can
efficiently be addressed in function form, albeit at some
loss of conceptual purity. In particular, a form of
side-effects arises under this extension, since a function
invocation can exert constraints on variables shared with
other function invocations. Nevertheless, we demonstrate
that determinacy can be retained, even under parallel
execution. The graph reduction language FGL is used for
this demonstration, by being extended to a language FGL+LV
permitting formal parameter expressions, with variables
occurring therein bound by unification. The determinacy
argument is based on a novel dataflow-like rendering of
unification. In addition the complete partial order
employed in this proof is unusual in its explicit
representation of demand, a necessity given the "benign"
side-effects that arise. An implementation technique is
suggested, suitable for reduction architectures.
------------------------------
Date: Sun, 19 May 85 03:22:53 cdt
From: (Raghu Ramakrishnan) Raghu@UT-sally
Subject: '%' notation
Examples 1 and 2 discuss design decisions.
Example 1
This example discusses why unification with an unannotated
term or variable in the call should be defined to fail.
Suppose it is not so defined.
Call: Equivalence(Y, Y)
Clause: Equivalence(5, X%) <--- ...
Clearly, the two Y's equivalence X to 5, and a
left-to-right unification will succeed with X being
instantiated to 5. However, a right-to-left unification
will instantiate Y to X, with or without preserving the '%'
annotation depending on our definition of '%'.
If both the occurrences of Y are annotated with '%'
unification will suspend until some other process
instantiates Y regardless of the order of unification,
consistently yielding the semantics we desire. []
Example 2
This example emphasizes the fact that no assurance is given
as to which process instantiates the variables in a
%-annotated term.
Call: No←guarantee([X|Y]%, X, Y)
Clause: No←guarantee(Z%, 5, 6) <--- ...
Unification always succeeds, instantiating X to 5 and Y to
6. The only fact guaranteed by the %-annotation, that the
argument corresponding to Z will be instantiated to a term
by some other process, has been satisfied trivially by
virtue of the corresponding argument being a term [X|Y].
This term, moreover, is %-annotated, thus indicating that
this call is to be matched with a clause that expects to
find its first argument instantiated to a term; and since
this is precisely what the given clause expects,
unification succeeds. []
Examples 3 - 5 show why the syntactic restrictions are
necessary.
Example 3
This example shows why variables in the clause head must
be unique.
Call: Sneaky←Equivalence([X|Y%], [X|5%])
Clause: Sneaky←Equivalence(Z, Z) <--- ...
Left-to-right unification will succeed with Y being
instantiated to 5. Right-to-left unification will suspend
until some other process instantiates Y, and will then try
to unify it with 5. Note that %-annotating the two Z's and
the corresponding terms would still yield the same undesirable
behaviour. The problem lies in the fact that the Z's
equivalence Y with 5. It might be argued that this
process will in any case fail if Y is not eventually
instantiated to 5, and there is no harm in letting this
failure be discovered later. The flaw lies in the fact that
there might be several alternative clauses (in other words,
potential process descriptions). The call, as we observed
earlier, is the specification of a needed process, and
this entire exercise of matching with the clause heads and
trying to satisfy the guards is intended to find a process
template (one of the clauses) which meets the
specifications. If, for instance, we assume an empty guard,
the choice is made once the unification of the call with a
clause head succeeds. The above order-dependent behaviour
could thus lead to the call reducing to a process that
fails or one that succeeds. Such a possibility is inherent
in any committed-choice logic programming language, but it
should at least be independent of things like the order of
unification! The fact that this objective is compromised by
some extra-logical features should not be interpreted as
licence to design other extra-logical features with the same
deficiency. []
Example 4
This illustrates why variables in a call must be annotated
consistently.
Call: Fickle(X%, X)
Clause: Fickle(Y%, 5) <--- ...
Left-to-right unification will suspend until some other
process instantiates X but right-to-left unification will
succeed with X and Y instantiated to 5. If both
occurrences of X in the call are %-annotated, this problem
does not arise since unification always suspends until X is
instantiated by another process. []
Example 5
If a term T in a clause head contains a %-annotated subterm
T1, then every sub-term of T including T must be %-annotated.
This is because the call call could equivalence two terms
in the head as in this example.
Call: Machiavelli(Z, Z)
Clause: Machiavelli([5|Y%], [5|6%]) <--- ...
Left-to-right unification will suspend whereas
right-to-left unification will succeed with Y instantiated
to 6. However, if both the terms in the clause head as well
as the two Z occurrences are %-annotated, unification will
always suspend until Z is instantiated by another process,
hopefully to some term of the form [←|U%], where the
underscore stands for an arbitrary term or variable, and U
is either '6' or a variable. At this point, unification
will again suspend if U is a variable. []
The following two examples illustrate the programming power
of '%'.
Example 6
This is an implementation of Quicksort in CP with the
%-annotation. The logic is exactly as in Shapiro's original
program.
quicksort(Unsorted%, Sorted) <--- qsort(Unsorted%, Sorted-[]).
qsort([X|Unsorted]%, Sorted-[]) <---
partition(Unsorted%, X%, Smaller, Larger),
qsort(Smaller%, Sorted-[X|Sorted1]),
qsort(Larger%, Sorted1-[]).
qsort([]%, []-[]).
partition([X|Xs]%, A%, Smaller, [Y|Larger]) <---
A < X, X = Y | partition(Xs%, A%, Smaller, Larger).
partition([X|Xs]%, A%, [Y|Smaller], Larger) <---
A >= X, X = Y | partitioned(Xs%, A%, Smaller, Larger).
partition([]%, ←%, [], []).
While the logic is the same as in Shapiro's program, the
above version has one important difference. Each clause
describes precisely what annotations it expects of its
arguments. This information is available without looking
at any of its calls, and so each clause can be understood
by considering it statically, with the assurance that
nothing unexpected will happen at run-time due to some
annotations that are passed on during execution. []
Example 7
An important feature of CP is the ability to pass partially
instantiated data structures, which can then be filled in
by the receiver and thus provide a response for the sender.
The following example, again adapted from Shapiro's paper,
demonstrates that this feature is retained.
stack(S%) <--- stack(S%, []).
stack([pop(X)|S]%, [Y|Xs]) <--- X = Y | stack(S%, Xs).
stack([push(X)|S]%, Xs) <--- stack(S%, [X|Xs]).
stack([]%, []).
This program receives a stream of push and pop instructions
and services them, putting the response in the variable X.
It initialises itself using the first clause. []
-- Raghu
------------------------------
Date: Wed, 22 May 85 10:47:11 pdt
From: (Tep Dobry) Tep%ucbdali@Berkeley
Subject: The Berkeley PLM Benchmarks
At the Warren Abstract Machine Workshop a few weeks ago
I was asked to publish the set of benchmarks programs I've
been using on my simulator for the Berkeley Prolog
Machine(PLM). I've finally got them all collected together
in Prolog form (CProlog) and have sent them to the Digest.
They're really too big to just publish in the Digest, so
they are being set up in a directory in the PROLOG directory
at SU-SCORE. There are 11 files with a total of 400 lines.
Since our machine is based on compiled Prolog, the top level
queries are also compiled in, generally as the predicate
main/0.
The benchmarks were primarily chosen to exercise all of
the features of the PLM, not for any complexity of program-
ming. About half of them come from Warren's thesis, and the
others we've added here. Our original performance figures
were based on simulations of hand compiled versions of these
benchmarks. We are currently looking for larger, more com-
plex benchmarks to run on the hardware when it is available.
So I'd be interested seeing large benchmarks sent to the
Digest.
-- Tep Dobry
[ the code for these benchmarks is available from the SCORE
PROLOG library under the subdirectory PS:<Prolog.BM> ]
------------------------------
End of PROLOG Digest
********************
∂24-May-85 0900 ullman@diablo Intermediate Language Proposal
Received: from SU-AIMVAX.ARPA by SU-AI.ARPA with TCP; 24 May 85 09:00:27 PDT
Date: Fri, 24 May 85 08:56:14 pdt
From: Jeff Ullman <ullman@diablo>
Subject: Intermediate Language Proposal
To: nail@diablo
The discussion on MRS set me thinking that what we really need
to do is define an intermediate language.
Then we can write in whatever language we choose a translator
from logic to icode, and translate icode into MRS, Prolog,
database commands, or a combination.
The trouble is, I'm not sure what the icode should be.
I'm going to list some operations that I think should be
there, and show some simple algorithms to "substantiate"
simple capture rules, written in this icode.
We're going to have to think carefully about capture rules
and their substantiation algorithms, to make sure we have a
rich enough icode.
DATA ENVIRONMENT
I assume that all variables in the icode take relations as
values. The important special cases of single tuples, unary
relations, and constants (single unary tuples) may be indicated
by special bits in the icode statements.
We assume that there is an associative store for assigning values
to variables and that there is a stack, on which values may be
pushed or popped. Maybe we shall want several stacks.
There is a global flag CHANGE, used to indicate whether a variable
changes its value after an assignment; see (9) below.
TUPLE-AT-A-TIME OPERATIONS
1. A := pi (<list of components>) B
Your usual projection.
2. A := sigma (<condition>) B
Your usual selection.
3. A := car (#i) B
A becomes a copy of B, but each value in component i is replaced by
its head. If the value is not a list structure, the tuple does
not appear in A.
4. A := cdr (#i) B
Similar, for the tail of component i.
5. A := cons (#i, #j) B
A becomes a copy of B, but the ith component of each tuple becomes
the cons of the ith and jth compoenents of the corresponding
tuple in B, and the jth component is not present in A.
6. A := B
B may be a variable or constant.
BINARY OPERATIONS
7. A := join (<correspondence list>) B, C
The equijoin of B and C.
8. A := union B, C
The union.
9. A += B
Equivalent to A := union A, B, but with the side effect of
setting CHANGE to true or false depending on whether
or not the value of A has changed, i.e., whether B is a subset of A.
BRANCHING STATEMENTS
10. goto <label>
Sorry about this folks.
11. if <condition> goto <label>
The conditions can be chosen from:
a. CHANGE
b. empty(X) for variable X
c. emptystack
d. ???
I'm undecided on whether it makes sense to have a call
and return statement. They would make life easier for translation
to icode but harder for execution of icode.
STACK OPERATIONS
12. PUSH(X)
13. A := TOP
14. A := POP
15. STACK := empty
with the obvious meanings
EXAMPLE 1.
Transitive Closure:
T(x, y) :- A(x, y)
T(x, y) :- A(x, z), T(z, y)
Assume we want to capture T↑bf, and we are given bound value b for
the first argument of T.
R := empty relation /* R accumulates answer */
U := {b} /* U is the current "frontier" */
1: T1 := sigma (#1 member of U) A
U := pi (#2) T1
R += U
if CHANGE goto 1
EXAMPLE 2.
Append:
append(A, [], [A|[]]) :-
append(A, [B|X], [B|Y]) :- append(A, X, Y)
Assume we want to capture append↑bbf, and we are given value a
and list s for the first two arguments.
S := s /* S will be taken apart and stacked */
STACK := empty
4: if S = [] goto 1
T1 := car S
PUSH T1
S := cdr S
goto 4
1: T2 := cons a [] /* Here the list s has been stacked in
reverse order. T2 reassembles the list, but
with a at the end */
3: if emptystack goto 2
T3 := POP
T2 := cons T3 T2
goto 3
2: OUTPUT := T2
I admit the general rules for transforming logical rules "like"
the examples is not clear. A good part of the problem is
getting straight how unifications translate into sequences of
icode steps. Even so, implementing capture rules is not
going to be trivial, and I'm open to suggestions regarding
a change in icode style that would make life easier.
∂24-May-85 1125 FORD@SU-CSLI.ARPA language processing workshop
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 24 May 85 11:24:55 PDT
Date: Fri, 24 May 1985 11:20 PDT
Message-ID: <FORD.12113612371.BABYL@SU-CSLI.ARPA>
From: FORD@SU-CSLI.ARPA
To: researchers@csli
Subject: language processing workshop
ANOUNCING A WORKSHOP ON LANGUAGE PROCESSING
sponsored by CSLI and Sloan
Dates: Monday June 10 - Wednesday June 12
Location: CSLI
The Psychology Study Group at CSLI in conjunction with areas P and
NL is planning an open workshop on Language Processing and has invited
several speakers both from the United States and abroard. Participants
will include:
Tom Bever, Herb Clark, Stephen Crain, Gary Dell, Marilyn Ford,
Ken Forster, Don Foss, Willem Levelt, William Marslen-Wilson,
James McClelland, Mark Seidenberg, Dave Swinney, Michael
Tanenhaus and Lorraine Tyler.
We are considering having discussants for the papers. If you are
interested in being a discussant, please contact Marilyn Ford
(ford) or Marcy Macken (mmacken).
We will have a schedule available next week.
∂24-May-85 1130 PIERCE@SU-SCORE.ARPA Quad Phone
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 24 May 85 11:30:38 PDT
Date: Fri 24 May 85 11:20:29-PDT
From: Lee Pierce <PIERCE@SU-SCORE.ARPA>
Subject: Quad Phone
To: csd@SU-SCORE.ARPA
There is a phone in the quad near the Information Booth which the department
pays for. It is intended for use by people who need access to the building
when it is locked. They can use the phone to call and someone can then let
them in.
Lee
-------
∂24-May-85 1155 EMMA@SU-CSLI.ARPA Copying machine
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 24 May 85 11:55:00 PDT
Date: Fri 24 May 85 11:50:04-PDT
From: Emma Pease <Emma@SU-CSLI.ARPA>
Subject: Copying machine
To: bboard@SU-CSLI.ARPA, folks@SU-CSLI.ARPA
Tel: 497-3479
Because we will be switching copying machines in Ventura 7 next
Wednesday, it will be difficult or impossible to do any copying on
that day.
-------
∂24-May-85 1202 BLOCK@SU-CSLI.ARPA Situation Semantics
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 24 May 85 12:01:50 PDT
Date: Fri 24 May 85 11:55:33-PDT
From: Ned Block <BLOCK@SU-CSLI.ARPA>
Subject: Situation Semantics
To: bboard@SU-CSLI.ARPA, su-bboards@SU-CSLI.ARPA, folks@SU-CSLI.ARPA,
friends@SU-CSLI.ARPA
SITUATION SEMANTICS MADE EASY
Three lectures by
John Perry
M,W,M, June 3, 5, 10, at 3:15
REDWOOD G-19
(Note that the lecture formerly scheduled for June 7 has been changed
to June 10, same time and place.) The first lecture will be aimed at
those who know nothing at all about situation semantics.
-------
∂24-May-85 1254 EMMA@SU-CSLI.ARPA Situation Semantics
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 24 May 85 12:54:26 PDT
Mail-From: BLOCK created at 24-May-85 11:55:33
Date: Fri 24 May 85 11:55:33-PDT
From: Ned Block <BLOCK@SU-CSLI.ARPA>
Subject: Situation Semantics
To: bboard@SU-CSLI.ARPA, su-bboards@SU-CSLI.ARPA, folks@SU-CSLI.ARPA,
friends@SU-CSLI.ARPA
ReSent-Date: Fri 24 May 85 12:50:20-PDT
ReSent-From: Emma Pease <Emma@SU-CSLI.ARPA>
ReSent-To: friends@SU-CSLI.ARPA
SITUATION SEMANTICS MADE EASY
Three lectures by
John Perry
M,W,M, June 3, 5, 10, at 3:15
REDWOOD G-19
(Note that the lecture formerly scheduled for June 7 has been changed
to June 10, same time and place.) The first lecture will be aimed at
those who know nothing at all about situation semantics.
-------
∂24-May-85 1421 BRODER@SU-SCORE.ARPA Next AFLB talk
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 24 May 85 14:21:28 PDT
Date: Fri 24 May 85 14:19:25-PDT
From: Andrei Broder <Broder@SU-SCORE.ARPA>
Subject: Next AFLB talk
To: aflb.all@SU-SCORE.ARPA
cc: tuley@SU-SCORE.ARPA
5/30/85 - Daniel Lehmann (Hebrew U. visiting Brandeis U.)
"The temporal logic of probabilistic programs"
In a joint work with S. Shelah, some extensions of the propositional
temporal logic of discrete time were advocated as useful for stating
and proving properties of probabilistic concurrent programs. Deductive
completeness theorems were proved. In a joint work with S. Kraus
corresponding decision procedures were investigated. Recently a system
for describing time and knowledge has been proposed. All those systems
can be characterized as two-dimensional modal logics, i.e. they
involve two essentially orthogonal modalities, one of them being time,
that satisfy some interchange law. The techniques involved in
studying such systems and some open problems will be described.
***** Time and place: May 30, 12:30 pm in MJ352 (Bldg. 460) ******
-------
∂24-May-85 1541 TAJNAI@SU-SCORE.ARPA 83/84 Industrial Affiliates Report
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 24 May 85 15:41:40 PDT
Date: Fri 24 May 85 15:37:24-PDT
From: Carolyn Tajnai <TAJNAI@SU-SCORE.ARPA>
Subject: 83/84 Industrial Affiliates Report
To: faculty@SU-SCORE.ARPA
cc: diaz@SU-SCORE.ARPA
I just received the 83/84 Industrial Affiliates Annual Report.
Contributions were:
Program Members Fee Contributions
------- ------- ----- --------------
GSB 174 $5K $1,487.514
Forum 56 $12.5K(fiche) 757,806
$13K(hard)
Solid State 51 $10K 456,885
Rock Physics 12 $20K 244,000
Energy Modeling 13 $15K 200,000
The rest were less than $200,000. There were 27 programs; with 567
memberships; Total contributions: $5,172,070.
Membership fees range from a low of $1-2K for Earthquake Engineering
to a high of $20K for Rock Physics and Stanford Exploration Project.
Current status of the Forum is 71 members and $723,596.48.
Carolyn
-------
∂24-May-85 1719 REGES@SU-SCORE.ARPA affiliated students
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 24 May 85 17:19:30 PDT
Date: Fri 24 May 85 17:07:30-PDT
From: Stuart Reges <REGES@SU-SCORE.ARPA>
Subject: affiliated students
To: faculty@SU-SCORE.ARPA
Office: Margaret Jacks 210, 497-9798
I have been getting many requests for SUSHI accounts, so I thought that I would
clarify my current policy on who can get an account. The primary population is:
o all students working towards CS degrees
o all student instructors teaching courses paid for by CS
o all TAs working for courses paid for by CS
Notice that this does not include students who are doing a project in a CS
course, nor does it include instructors/TAs in cross-listed courses that are
not paid for by CS. There are two primary exceptions.
First, we recognize that many students might be working towards other degrees
(e.g., EE and MIS) but might be so closely affiliated with the Department that
they are "virtual CS students." Such students can get SUSHI accounts if they
get faculty sponsorship.
Second, some non-CS students help to improve CS courses. They might develop
new software or write a textbook, for example. I am willing to open accounts
for these students as well, if they have a faculty sponsor.
The CS students and TAs already have accounts. For the other students, I am
sending them back to you, the faculty. I'm telling them that if they fit one
of these other two categores, then they can get accounts through faculty
sponsorship. Therefore, they'll soon be asking you instead of me whether or
not they can get an account.
Please don't treat this lightly. An affiliated student should be someone who
is really working closely with you or your research group.
To sponsor a student, send me a message telling me:
o name of student
o reason for sponsorship (affiliated student/course improvement)
o how I can contact the student (e.g., an existing EMAIL address)
You can send the message either physically or electronically.
-------
∂24-May-85 2104 @SU-SCORE.ARPA:berglund@Pescadero The Quals and My Advisor
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 24 May 85 21:04:04 PDT
Received: from Pescadero by SU-SCORE.ARPA with TCP; Fri 24 May 85 21:01:59-PDT
Date: Fri, 24 May 85 21:01:42 pdt
From: Eric Berglund <berglund@Pescadero>
Subject: The Quals and My Advisor
To: faculty@score, phd@score, su-bboards@Pescadero
A friend of mine has pointed out that the feeling of unpreparedness to do
research (even after passing the Computer Science Systems Qual) that I
described in a message to the faculty and Ph.D. students of CSD could
POSSIBLY be misconstrued as a criticism of individuals.
I want to state as clearly as possible that I believe my advisor and all the
members of the systems qual committee having performed their roles with the
utmost competence. I believe that the comp/qual scheme needs to be
reexamined, not the individuals administering it.
Sorry for the confusion, and sorry to send this to such a wide list, but
bad gossip spreads quickly and corrections only ooze.
--Eric
∂25-May-85 1525 BSCOTT@SU-SCORE.ARPA Velo-Bind
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 25 May 85 15:25:29 PDT
Date: Sat 25 May 85 15:14:56-PDT
From: Betty Scott <BSCOTT@SU-SCORE.ARPA>
Subject: Velo-Bind
To: CSD@SU-SCORE.ARPA
cc: BScott@SU-SCORE.ARPA
Many thanks to all of you who responded to my message about whether to replace
the Velo-Bind. It is obvious from the responses that we do need an in-house
binding machine. Several suggestions were made concerning other methods of
binding and other equipment available. We will check on all alternatives
and have the Velo-Bind replaced as soon as possible. The replacement may not
be a Velo-Bind.
Thanks again,
Betty
-------
∂27-May-85 1812 NILSSON@SU-SCORE.ARPA Lunch
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 27 May 85 18:12:22 PDT
Date: Mon 27 May 85 18:11:52-PDT
From: Nils Nilsson <NILSSON@SU-SCORE.ARPA>
Subject: Lunch
To: faculty@SU-SCORE.ARPA
Tuesday's faculty lunch will have as its guest Harry Llull of the
Math/CS Library. We'll have informal discussion about the library
and its future plans.
Our room has been commandeered (for some high purpose,no doubt) on
June 4, so no lunch that day. On June 11, finals week, we'll have
an informal discussion summing up the year and a chance to think
about next year. -Nils
-------
∂28-May-85 0931 YAN@SUMEX-AIM.ARPA Abstract for this Wednesday's talk at 9:30
Received: from SUMEX-AIM.ARPA by SU-AI.ARPA with TCP; 28 May 85 09:31:20 PDT
Date: Tue 28 May 85 09:30:16-PDT
From: Jerry Yan <YAN@SUMEX-AIM.ARPA>
Subject: Abstract for this Wednesday's talk at 9:30
To: aap@SUMEX-AIM.ARPA
cc: mak@SU-SIERRA.ARPA, yeager@SUMEX-AIM.ARPA, i.ikabod9%LOTS-A@SUMEX-AIM.ARPA
Ladies and Gentlemen,
I will continue to give an informal presentation on some ideas on
resouce management in our context. This week, it will be "reclamation".
The presentation will begin at 9:30 (again)
Resource Management
in a Distributed Multiprocessing Environment
... running x/czar/care/simple
Third Movement: "Requiem"
Distributed Asynchronous Agent Reclamation
Key ideas:
1 A scheme is proposed to identify and collect "garbage" agents/
processes "on the fly".
2. Garbage is identified on the basis of:
a. inaccessibility
b. (performing) "uselessness" (tasks)
c. dependencies (killed because "parent" is killed)
3. An analysis of the three established collection algorithms:
reference count
mark/sweep
copy/swap
4. My proposal: a "self-management" scheme
all pointers are bi-directional
list of "aquaintance" for each agent is updated by messages
collection based on the 3 definitions of "garbage" is then
discussed
5. Epilog: a comparison and an experiment plan
So be there!
=j=
-------
∂28-May-85 1012 HANRAHAN@SU-SCORE.ARPA HEAT
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 28 May 85 10:12:03 PDT
Date: Tue 28 May 85 10:05:22-PDT
From: Katherine Hanrahan <HANRAHAN@SU-SCORE.ARPA>
Subject: HEAT
To: csd@SU-SCORE.ARPA
Stanford-Phone: (415) 497-2273
TOMORROW FROM 8 A.M. TO 1 P.M. WE WILL BE WITHOUT HEAT (HOPE FOR A SUNNY,
WARM DAY/ BRING A SWEATER) WORK CENTER SAYS: "SORRY FOR THE INCONVENIENCE"
-------
∂28-May-85 1354 NILSSON@SU-SCORE.ARPA meetings
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 28 May 85 13:54:47 PDT
Date: Tue 28 May 85 13:52:55-PDT
From: Nils Nilsson <NILSSON@SU-SCORE.ARPA>
Subject: meetings
To: faculty@SU-SCORE.ARPA
We have a regularly scheduled senior faculty meeting on Thursday
June 6. We have a regularly scheduled general faculty meeting on
Tuesday June 11 at 2:30. To make life less complicated, let's change
the SENIOR MEETING so that it occurs on TUESDAY JUNE 11 also (right
after the general meeting). Let me know if this causes any problems.
(I have only one agenda item so far for the senior faculty meeting,
namely a request by one of our faculty for a new consulting professorship.
Papers are being distributed.)
Please let me know of any items you would like on the agenda for either
meeting.
-Nils
-------
∂28-May-85 1727 LIBRARY@SU-SCORE.ARPA Socrates: Automatic Renewal of Personal Accounts in August
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 28 May 85 17:27:25 PDT
Date: Tue 28 May 85 17:06:51-PDT
From: C.S./Math Library <LIBRARY@SU-SCORE.ARPA>
Subject: Socrates: Automatic Renewal of Personal Accounts in August
To: su-bboards@SU-SCORE.ARPA
cc: : ;
If you have a Socrates account and are still at Stanford in the fall as
a student, faculty, or staff, your account will be renewed automatically
in August. Any accounts out to students who are not registered in
September will be closed.
The library card catalogs will not be closing before fall quarter.
The latest Socrates Update describes some of the enhancements that
will be made to the Socrates system over the summer.
H.Llull
-------
∂28-May-85 1737 BRODER@SU-SCORE.ARPA Next AFLB talk
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 28 May 85 17:37:25 PDT
Date: Tue 28 May 85 17:11:17-PDT
From: Andrei Broder <Broder@SU-SCORE.ARPA>
Subject: Next AFLB talk
To: aflb.all@SU-SCORE.ARPA
cc: tuley@SU-SCORE.ARPA
5/30/85 - Daniel Lehmann (Hebrew U. visiting Brandeis U.)
"The temporal logic of probabilistic programs"
In a joint work with S. Shelah, some extensions of the propositional
temporal logic of discrete time were advocated as useful for stating
and proving properties of probabilistic concurrent programs. Deductive
completeness theorems were proved. In a joint work with S. Kraus
corresponding decision procedures were investigated. Recently a system
for describing time and knowledge has been proposed. All those systems
can be characterized as two-dimensional modal logics, i.e. they
involve two essentially orthogonal modalities, one of them being time,
that satisfy some interchange law. The techniques involved in
studying such systems and some open problems will be described.
***** Time and place: May 30, 12:30 pm in MJ352 (Bldg. 460) ******
6/6/85 - Profs. D. Knuth and C. Papadimitriou (Stanford)
Solutions to the 1985 Qualifying Examination in Analysis of
Algorithms. (Problem 2 and 3)
Copies of the Exam are available from Phyllis Winkler, in MJH 326.
***** Time and place: June 6, 12:30 pm in MJ352 (Bldg. 460) ******
6/13/85 - Prof. L. Guibas (Stanford)
Solutions to the 1985 Qualifying Examination in Analysis of
Algorithms. (Problem 1)
Copies of the Exam are available from Phyllis Winkler, in MJH 326.
***** Time and place: June 13, 12:30 pm in MJ352 (Bldg. 460) ******
-------
∂29-May-85 0015 @SU-SCORE.ARPA:ARK@SU-AI.ARPA CS Colloq, June 4: Daniel Lehmann, Mutual Exclusion
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 29 May 85 00:15:31 PDT
Received: from SU-AI.ARPA by SU-SCORE.ARPA with TCP; Wed 29 May 85 00:01:06-PDT
Date: 29 May 85 0000 PDT
From: Arthur Keller <ARK@SU-AI.ARPA>
Subject: CS Colloq, June 4: Daniel Lehmann, Mutual Exclusion
To: All-Colloq@SU-SCORE.ARPA, Faculty@SU-SCORE.ARPA
CC: Vardi@SU-AIMVAX.ARPA, ARK@SU-AI.ARPA
*** Please note the changed title and abstract ***
CS Colloquium, June 4, 4:15pm, Terman Auditorium
SYMMETRIC AND ECONOMICAL SOLUTIONS TO THE MUTUAL EXCLUSION PROBLEM
IN A DISTRIBUTED SYSTEM
Daniel Lehmann
Brandeis University
(on sabbatical leave from Hebrew University)
The mutual exclusion problem in a distributed system, in which each
process has a memory of its own, into which it has exclusive write
privileges but from which others may read, is reconsidered. Symmetric
solutions are looked for. It is shown that, though no such solution
may be deterministic, there are probabilistic solutions. Different
solutions are provided for two processes, and then a solution is
proposed for any number of processes. The solutions offered are
amenable to a formal proof of their correctness with a small effort.
The solutions are correct even against a very well informed scheduler,
unlike Rabin's probabilistic solution to the mutual exclusion problem
in a centralized system. Some of the solutions are correct even
against an evil scheduler that knows in advance the results of the
future random draws, in sharp contrast with other algorithms. The
solutions are economical: mutual exclusion between two processes may
be achieved with variables capable of holding four different values,
and mutual exclusion between n processes may be achieved with
variables capable of holding ten different values. All solutions have
been attained by careful reasoning and not by an exhaustive computer
search, they exhibit general principles of design that may be useful
in solving other similar problems. This is joint work with S. Cohen
and A. Pnueli.
----
Cookies, thanks to Ginger, and possibly juice will probably be served in
the 3rd floor lounge of MJH at 3:45pm. See you there.
∂29-May-85 1252 ullman@diablo today's plan
Received: from SU-AIMVAX.ARPA by SU-AI.ARPA with TCP; 29 May 85 12:52:09 PDT
Date: Wed, 29 May 85 09:08:42 pdt
From: Jeff Ullman <ullman@diablo>
Subject: today's plan
To: nail@diablo
We'll have a brief meeting at 11AM.
ELi, Allen, and I will probably occupy the time talking
about P vs. NC issues for logical rules.
I have to prerecord a lecture at noon, but the meeting can
continue. By the way, I'm covering "logical databases"
and "capture rules" in CS245, which meets TuTh 11AM in 320 ERL.
Next week, I'd like to discuss the intermediate code I sent to
the nail list.
∂29-May-85 1250 LIBRARY@SU-SCORE.ARPA New Books In the Math/CS Library
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 29 May 85 12:48:49 PDT
Date: Wed 29 May 85 11:18:12-PDT
From: C.S./Math Library <LIBRARY@SU-SCORE.ARPA>
Subject: New Books In the Math/CS Library
To: su-bboards@SU-SCORE.ARPA
cc: faculty@SU-SCORE.ARPA
LISP Machine Manual. 6th ed. version 99. June 1984. Stallman, Weinreb,
and Moon. (8512726)
Software Process Workshop. IEEE. Feb. 1984. edited by Potts. QA76.6.S636 1984
Data Analysis and Informatics, III. Third International Symposium. Versailles
1983. QA278.I56 1981
Numerical Analysis. 3rd ed. by Burden and Faires. QA297.B84 1985
Procedural Elements for Computer Graphics. by Rogers. T385.R63 1985
Acoustic Signal Processing in the Context of a Perceptual Model. 1980
by Petersen. (8507467)
HL
-------
∂29-May-85 1256 MARJORIE@SU-CSLI.ARPA System problems
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 29 May 85 12:55:31 PDT
Date: Wed 29 May 85 11:03:09-PDT
From: Marjorie Maxwell <MARJORIE@SU-CSLI.ARPA>
Subject: System problems
To: Folks@SU-CSLI.ARPA
cc: consultant@SU-CSLI.ARPA
Clay has asked me to announce the following message concerning the system-
The tips are up and working but cannot connect to Turing. Turing, Whitehead,
and Russell are all up, but can't connect to each other. The dandelions can't
chat to Turing.
If anyone needs a direct terminal please contact me at 9030.
Thank you.
Marjorie
-------
∂29-May-85 1304 WRIGHT@SUMEX-AIM.ARPA Siglunch Friday, May 31
Received: from SUMEX-AIM.ARPA by SU-AI.ARPA with TCP; 29 May 85 13:04:10 PDT
Date: Wed 29 May 85 09:45:28-PDT
From: Carol Wright/Susie Barnes <WRIGHT@SUMEX-AIM.ARPA>
Subject: Siglunch Friday, May 31
To: siglunch@SUMEX-AIM.ARPA
There will be no SigLunch this week because of a RipLunch.
CS 322 (Heuristic Programming Seminar) will meet as usual
(for the RipLunch), in the chemistry gazebo.
-------
∂29-May-85 1312 BERGMAN@SU-SCORE.ARPA Summer RA's
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 29 May 85 13:12:22 PDT
Date: Wed 29 May 85 10:32:44-PDT
From: Sharon Bergman <BERGMAN@SU-SCORE.ARPA>
Subject: Summer RA's
To: faculty@SU-SCORE.ARPA
The time has come to send through the paperwork to set up research
assistantships for the summer quarter. If you will be supporting any
RA's for the summer and have not yet informed us, please let me know
within the next couple of days.
Thanks, Sharon Bergman
-------
∂29-May-85 1359 HEDGES@SU-SCORE.ARPA General Faculty Meeting
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 29 May 85 13:59:23 PDT
Date: Wed 29 May 85 13:57:33-PDT
From: Karen Hedges <HEDGES@SU-SCORE.ARPA>
Subject: General Faculty Meeting
To: Faculty@SU-SCORE.ARPA
PLEASE NOTE: The General Faculty Meeting scheduled for Tuesday, June 11
will be held @ 3:15 pm in Room 146 not 2:30 pm. This is a
time change only.
-------
∂29-May-85 1404 HEDGES@SU-SCORE.ARPA Sr. Faculty Meeting
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 29 May 85 14:04:21 PDT
Date: Wed 29 May 85 14:04:28-PDT
From: Karen Hedges <HEDGES@SU-SCORE.ARPA>
Subject: Sr. Faculty Meeting
To: Tenured@SU-SCORE.ARPA
PLEASE NOTE: The Sr.Faculty Meeting will be held immediately following
the General Faculty Meeting on Tuesday, June 11 in room 146.
-------
∂30-May-85 0631 REGES@SU-SCORE.ARPA Black Friday Meeting
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 29 May 85 16:49:30 PDT
Date: Wed 29 May 85 16:47:58-PDT
From: Stuart Reges <REGES@SU-SCORE.ARPA>
Subject: Black Friday Meeting
To: faculty@SU-SCORE.ARPA
Office: Margaret Jacks 210, 497-9798
Since Nils has cancelled the Senior Faculty Meeting next Thursday, I thought
I would snatch up the time for Black Friday. I have reserved Jacks 252 for
the meeting starting at 2:30. We will only be discussing the students who
received "threatening" Grey Tuesday letters, so I expect the meeting will be
shorter than Grey Tuesday was. Victoria and I will distribute some materials
about the students to be discussed early next week so that you can examine it
before the meeting.
-------
∂30-May-85 0645 BLOCK@SU-CSLI.ARPA Next Meeting
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 29 May 85 17:00:00 PDT
Date: Wed 29 May 85 16:56:45-PDT
From: Ned Block <BLOCK@SU-CSLI.ARPA>
Subject: Next Meeting
To: RRR@SU-CSLI.ARPA
Kurt van Lehn (Xerox) will give us an overview and a critique of
the movement known as connectionism. He hopes we will interrupt within
the first 45 seconds--no problem there, eh!
In the unlikely event that anyone will do any homework, here is a list
of background reading that van Lehn has provided (however, he is not
going to assume that anyone has read anything).
McClelland, J.L. & Rumelhart, D.E. "An interactive activation modle of
context effects in letter perception: Part 1. An account of basic
findings." Psycholgoical Review, Vol 88, No. 5, 1981.
A major paper collection, which provides a good view of the field circa
late 70's is
Hinton, G.E. & Anderson, J.A. Parallel models of associative memory.
Hillsdale, NJ: Erlbaum, 1981.
The most recent issue of Cognitive Science is a special issue on
connectionism. It contains most of the papers presented at last year's
CogSci & AAAI conventions. I would guess that these two collections,
plus the M&R article, represent about 2/3 of the literature in the
field. Surely anyone who reads them all will see most of ideas that
connectionism has to offer.
-------
∂30-May-85 0651 EMMA@SU-CSLI.ARPA Newsletter May 30, No. 31
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 29 May 85 19:55:39 PDT
Date: Wed 29 May 85 17:16:33-PDT
From: Emma Pease <Emma@SU-CSLI.ARPA>
Subject: Newsletter May 30, No. 31
To: friends@SU-CSLI.ARPA
Tel: 497-3479
C S L I N E W S L E T T E R
←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←
May 30, 1985 Stanford Vol. 2, No. 31
←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←
A weekly publication of The Center for the Study of Language and
Information, Ventura Hall, Stanford University, Stanford, CA 94305
←←←←←←←←←←←←
CSLI ACTIVITIES FOR *THIS* THURSDAY, May 30, 1985
12 noon TINLunch
Ventura Hall ``Computers and Emotion''
Conference Room Discussion led by Helen Nissenbaum
2:15 p.m. CSLI Seminar
Redwood Hall ``On Modelling Shared Understanding''
Room G-19 Jon Barwise, CSLI
3:30 p.m. Tea
Ventura Hall
4:15 p.m. CSLI Colloquium
Redwood Hall ``Natural Kinds, Homeostasis, and the Limits of
Room G-19 Essentialism''
Richard Boyd, Prof. of Philosophy, Cornell University
←←←←←←←←←←←
CSLI ACTIVITIES FOR *NEXT* THURSDAY, June 6, 1985
12 noon TINLunch
Ventura Hall ``Speech Act Distinctions in Syntax''
Conference Room by Jerrold Sadock and Arnold Zwicky
Discussion led by Dietmar Zaefferer
(Abstract on page 2)
2:15 p.m. CSLI Seminar
Redwood Hall ``Existential Sentences''
Room G-19 Edit Doron, CSLI
Discussion led by Larry Moss
(Abstract on page 2)
3:30 p.m. Tea
Ventura Hall
4:15 p.m. CSLI Colloquium
Redwood Hall ``An Assumption-Based Truth-Maintenance System''
Room G-19 Johan De Kleer, Xerox PARC, Intelligent Systems Lab.
(Abstract on page 2)
!
Page 2 CSLI Newsletter May 30, 1985
←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←
ABSTRACT OF NEXT WEEK'S TINLUNCH
``Speech Act Distinctions in Syntax''
There is an apparent mismatch between what speech act theorists
claim to be the prototypical speech act types (promises, requests,
etc.) and the speech act types indicated by syntactical means in most
natural languages (assertions, questions, etc.). This paper by Sadock
and Zwicky gives an excellent survey of the sentence types in a sample
of twenty-three languages. It contains a lot of interesting
observations (e.g., ``imperatives have characteristically bare verb
stems'') and raises a lot of interesting questions (e.g., how to
explain the above mentioned fact). It is also a good starting point for
a discussion of the role of typology in the CSLI research program.
--Dietmar Zaefferer
←←←←←←←←←←←←
ABSTRACT OF NEXT WEEK'S SEMINAR
``Existential Sentences''
I will be concerned with the old puzzle concerning existential
sentences: why is sentence 1 odd in a way that sentence 2 is not?
1. There is every man in the garden.
2. There is a man in the garden.
I will discuss a semantic solution, and compare it to proposals by
Barwise and Cooper, Keenan, Higginbotham. --Edit Doron
←←←←←←←←←←←←
ABSTRACT OF NEXT WEEK'S COLLOQUIUM
``An Assumption-Based Truth-Maintenance System''
This paper presents a new view of problem solving motivated by a
new kind of truth maintenance system. Unlike previous truth
maintenance systems which were based on manipulating justifications,
this truth maintenance system is, in addition, based on manipulating
assumption sets. As a consequence it is possible to work effectively
and efficiently with inconsistent information, context switching is
free, and most backtracking (and all retraction) is avoided. These
capabilities motivate a different kind of problem-solving architecture
in which multiple potential solutions are explored simultaneously.
This architecture is particularly well-suited for tasks where a
reasonable fraction of the potential solutions must be explored.
--Johan De Kleer
!
Page 3 CSLI Newsletter May 30, 1985
←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←
WORKSHOP ON LANGUAGE PROCESSING
sponsored by CSLI and Sloan
Monday, June 10 through Wednesday, June 12
Several research groups at CSLI are holding an open, informal
workshop on language processing, with invited speakers from the United
States and abroad. The goal of the workshop is to examine current
psychological issues in language processing at the sentence and
discourse level, with particular focus on the relation of language
processing to language structure and the situations in which language
is used. Participants include:
Tom Bever, Herb Clark, Stephen Crain, Gary Dell, Marilyn Ford, Ken
Forster, Don Foss, Willem Levelt, William Marslen-Wilson, James
McClelland, Mark Seidenberg, Dave Swinney, Michael Tanenhaus, and
Lorraine Tyler.
The schedule is available from Suzanne Parker at the front desk of
CSLI (to receive a copy by mail, contact Sandy McConnell-Riggs
(Sandy@csli)). Copies of papers by the speakers relevant to their
workshop presentations are available in the CSLI Reading Room and the
Psychology Library, Jordan Hall. For further information, contact
members of the organizing committee: Herb Clark, Phil Cohen, Marilyn
Ford, Barbara Grosz, Ron Kaplan, Marcy Macken, Stanley Peters and Ivan
Sag. --Marcy Macken (mmacken@su-csli)
←←←←←←←←←←←←
SITUATION SEMANTICS MADE EASY
Three lectures by John Perry
Monday, Wednesday, Monday, June 3, 5, 10, at 3:15
Redwood G-19
The first lecture will be aimed at those who know nothing at all
about situation semantics. Please note that the day for the third
meeting has been changed from Friday, June 7 to Monday, June 10.
←←←←←←←←←←←←
-------
∂30-May-85 0854 PIERCE@SU-SCORE.ARPA Proposed Space Committee
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 30 May 85 08:54:37 PDT
Date: Thu 30 May 85 08:53:14-PDT
From: Lee Pierce <PIERCE@SU-SCORE.ARPA>
Subject: Proposed Space Committee
To: faculty@SU-SCORE.ARPA, staff@SU-SCORE.ARPA
Since I have been assigned responsibility for allocating visitor space
for the department, I have begun to realize that I'm going to need
some help. Would anyone like to work with me as part of a space
committee? I am hoping that at least one faculty member and one staff
person would be willing and able to spend some time helping to find
solutions to the numerous questions surrounding this issue. Even if
you are not able to be on a committee and have thoughts, ideas,
suggestions, I would appreciate any information you can provide.
Thanks. ---Lee
-------
∂30-May-85 0926 @SU-SCORE.ARPA:udi@wisc-rsch.arpa classification question
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 30 May 85 09:26:41 PDT
Received: from wisc-rsch.arpa by SU-SCORE.ARPA with TCP; Thu 30 May 85 09:21:15-PDT
Received: by wisc-rsch.arpa; Thu, 30 May 85 10:58:30 cdt
Received: from wisc-crys.arpa by wisc-rsch.arpa; Thu, 30 May 85 10:39:57 cdt
Received: from yale.arpa by wisc-crys.arpa; Thu, 30 May 85 10:39:47 cdt
Received: by YALE-BULLDOG.YALE.ARPA; 30 May 85 11:24:05 EDT (Thu)
Message-Id: <8505301524.AA00855@YALE-BULLDOG.YALE.ARPA>
Date: Thu, 30 May 85 11:30:51 EDT
From: Neta Amit <Amit@YALE.ARPA>
Subject: classification question
To: theory@WISC-CRYS
Sender: udi@wisc-rsch.arpa
Resent-From: udi@wisc-rsch.arpa (Udi Manber)
Resent-To: theory:;@wisc-rsch.arpa
I'd like to find out if the following problem is in P (algorithm?),
NP-C/hard, open:
Let U be a set of boolean variables, f:U-->{T,F} a boolean function
with AND/OR connectives only.
For every variable u in U, the number of appearances of u in f is
either one (u or u') or two (u and u'). Example: u1+u2'(u3+u1').
The question is: can we efficiently minimize ("simplify") f, according
to some minimization criterion? I.e. given a positive integer K, is
there a function f1 equivalent to f, with exactly
(a) K literals , or
(b) K terms, (assuming f1 is a sum of products) etc.
Notice that (1) no normal form is assumed and (2) negation is not
allowed, except for negated literals.
Any leads will be appreciated.
--Neta
Amit@Yale.Arpa
-------
∂30-May-85 1141 PIERCE@SU-SCORE.ARPA Space
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 30 May 85 11:41:17 PDT
Date: Thu 30 May 85 11:09:58-PDT
From: Lee Pierce <PIERCE@SU-SCORE.ARPA>
Subject: Space
To: staff@SU-SCORE.ARPA, faculty@SU-SCORE.ARPA, su-bboard@SU-SCORE.ARPA
It is very important that anyone who is planning to vacate their desk
space this summer, or who knows of anyone who is, to let me know as
soon as possible.
Thank you.
---Lee Pierce
-------
∂30-May-85 1632 PIERCE@SU-SCORE.ARPA Visitor Space
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 30 May 85 16:32:10 PDT
Date: Thu 30 May 85 16:29:20-PDT
From: Lee Pierce <PIERCE@SU-SCORE.ARPA>
Subject: Visitor Space
To: faculty@SU-SCORE.ARPA
This is a reminder that the department chairman must approve all requests
for visitors. I have a list of possible visitors and will be sending you
individual messages asking for dates of your visitors and what office
space you have in mind for them. I will be happy to meet with to talk
about your space requirements.
Thanks.
---Lee
-------
∂30-May-85 2304 @SU-SCORE.ARPA:cheriton@Pescadero Re: Proposed Space Committee
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 30 May 85 23:04:08 PDT
Received: from Pescadero by SU-SCORE.ARPA with TCP; Thu 30 May 85 23:01:38-PDT
Date: Thu, 30 May 85 23:01:26 pdt
From: David Cheriton <cheriton@Pescadero>
Subject: Re: Proposed Space Committee
To: PIERCE@SU-Score, faculty@SU-Score, staff@SU-Score
As a suggestion, I think your current efforts to dig up free space for visitors
are a bit of a concern. While it is nice to have visitors, those of us that
are staying around for the summer are here to get work done and putting
visitors whereever there happens to be some free space can be very disruptive.
I am particularly concerned about graduate students sharing offices and
areas with people who will be away (and therefore potentially threatened
with a new neighbor or more for the summer). Perhaps we should solicit
some input from the students on this. Personally, I feel we have more of
a responsibility to maintain a good working environment for the students
that stay around than accomodate visitors.
In addition, some areas are effectively used primarily by one research
group and putting unassociated visitors in the midst of same can be
especially disruptive.
Again, I am not against visitors but I am finding the current push for
visitor space very disconcerting, to say the least.
∂31-May-85 0900 NILSSON@SU-SCORE.ARPA Re: Proposed Space Committee
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 31 May 85 09:00:46 PDT
Date: Fri 31 May 85 08:58:46-PDT
From: Nils Nilsson <NILSSON@SU-SCORE.ARPA>
Subject: Re: Proposed Space Committee
To: cheriton@SU-PESCADERO.ARPA
cc: PIERCE@SU-SCORE.ARPA, faculty@SU-SCORE.ARPA, staff@SU-SCORE.ARPA,
NILSSON@SU-SCORE.ARPA
In-Reply-To: Message from "David Cheriton <cheriton@Pescadero>" of Thu 30 May 85 23:02:21-PDT
There isn't really a "push" for visitor space. What there is a push
for is for some attempt to find out what visitors people have already
invited and to try to find space for them. When the process begins
to run smoothly a person wanting to invite a visitor will first check
with Lee Pierce and me to find out if there is space and to talk about
how important it is to the CSD to have that visitor. -Nils
-------
∂31-May-85 0939 @SU-SCORE.ARPA:coraki!pratt@Navajo Re: Proposed Space Committee
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 31 May 85 09:39:23 PDT
Received: from Navajo ([36.8.0.48].#Internet) by SU-SCORE.ARPA with TCP; Fri 31 May 85 09:35:01-PDT
Received: by coraki.uucp (1.1/SMI-1.2)
id AA00837; Fri, 31 May 85 09:29:57 pdt
Date: Fri, 31 May 85 09:29:57 pdt
From: coraki!pratt@Navajo (Vaughan Pratt)
Message-Id: <8505311629.AA00837@coraki.uucp>
To: David Cheriton <cheriton@Pescadero.ARPA>
Cc: PIERCE@SU-Score.ARPA, faculty@SU-Score.ARPA, staff@SU-Score.ARPA
Subject: Re: Proposed Space Committee
In-Reply-To: message of Thu, 30 May 85 23:01:26 pdt.
<8505310906.AA00423@coraki.uucp>
From: David Cheriton <cheriton@Pescadero.ARPA>
I am particularly concerned about graduate students sharing
offices and areas with people who will be away (and therefore
potentially threatened with a new neighbor or more for the
summer). ... Personally, I feel we have more of a responsibility
to maintain a good working environment for the students that
stay around than accomodate visitors.
Having been a visitor at many places, let me speak with my visitor's
hat on (but hopefully not talking through it).
First a plug for the summer visit. It is difficult to overstate the
value of the summer visit. It broadens the perspective of
*both visitors and hosts* by exposing both to ideas from other places
in greater depth than usually happens via colloquia, conferences, etc.
Now if you think for a moment about the logistics of the summer visit
it will become apparent that it involves something that ideally
is a permutation of the inhabitants of the participating institutions.
Pigeons are shuffled amongst pigeonholes, with the number of pigeons
per pigeon-hole being held as constant as possible. In this scenario
office-swapping of the kind under attack is an unavoidable component
of the visitor system.
I am somewhat sympathetic to the notion that having a strange pigeon
next door for two or three months may be disconcerting to the more
staid and conservative members of our establishment. However I have no
sympathy whatsoever for the concept of protecting any of our students
from visitors. Attempting to shelter them from visitors in this way
makes about as much sense as putting them in a hermetically sealed
bubble to protect them from airborne viruses. In fact less sense,
since at least some of the pigeons that will be roosting here for the
summer will hopefully have a more beneficial impact on their closest
neighbors than your average virus.
The idea of protecting students from visitors is completely backwards.
They - and all of us - need to be exposed to visitors, *especially* those
of us that don't play visitor very often.
A blow against the visitor system is a blow against cross-fertilization
of ideas and a step towards mental stagnation.
-v
∂31-May-85 1134 NILSSON@SU-SCORE.ARPA Visiting Committee
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 31 May 85 11:34:02 PDT
Date: Fri 31 May 85 11:32:37-PDT
From: Nils Nilsson <NILSSON@SU-SCORE.ARPA>
Subject: Visiting Committee
To: faculty@SU-SCORE.ARPA
I have recently had discussions with Dean Jim Gibbons about restarting a
"visiting committee" for computer science at Stanford. The visiting
committee would help the CSD and Stanford evaluate the general health of
computer science at Stanford--our strengths and any potential areas for
improvement. I personally think this is a very important component of
maintaining our high quality; if the visiting committee is chosen
wisely, its constructive criticisms can only help us. Jim Gibbons is
also very interested in forming such a committee. Jim has suggested
some possible names, and so have I. We would be interested in hearing
faculty comments and recommendations. Here are some potential
candidates; we might want to ask five or so of them to serve:
John Hopcroft, Cornell Joel Moses, MIT
Albert Meyer, MIT Jerry Sussman, MIT
Al Aho, ATT Bell Labs Al Borodin, Toronto
Jack Schwarz, NYU Ray Reiter, Toronto
Bob Tarjan, Princeton Drew McDermott, Yale
Steve Cook, Toronto Robert Kahn, DARPA
Dick Karp, Berkeley Bob Sproull, SSA
Forest Baskett, DEC David Paterson, UCBerkeley
If you have opinions about all of this, please let me know. -Nils
-------
∂31-May-85 1248 @SU-SCORE.ARPA:TW@SU-AI.ARPA Visiting committee
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 31 May 85 12:46:35 PDT
Received: from SU-AI.ARPA by SU-SCORE.ARPA with TCP; Fri 31 May 85 12:43:56-PDT
Date: 31 May 85 1241 PDT
From: Terry Winograd <TW@SU-AI.ARPA>
Subject: Visiting committee
To: faculty@SU-SCORE.ARPA
Nils,
Looking at your proposed list I see a significant (and I suspect
conscious) shift in who you are looking for. The old committee was
primarily people with importance in the computer industry (Bob Taylor,
Fred Brooks, etc.) while these are all our peers in the research
community. The argument for the old version is that they could serve as a
weightier voice in making recommendations to the university, while the new
list will be seen as our friends who share our vested interests. On the
other hand, the new list will certainly be able to provide more focussed
and exacting commentary on the content of our research. On the assumption
that we are not in a position of needing additional leverage with the
University (or that it didn't work anyway), it seems that the second
function is more valuable and we should proceed with a reasonable subset
of these. --t
∂31-May-85 1259 avg@diablo NC, P-completeness, and extension joins
Received: from SU-AIMVAX.ARPA by SU-AI.ARPA with TCP; 31 May 85 12:58:58 PDT
Date: Fri, 31 May 85 12:45:32 pdt
From: Allen VanGelder <avg@diablo>
Subject: NC, P-completeness, and extension joins
To: nail@diablo
As I mentioned in the meeting after Jeff left, there is another version
of the nonlinear example that is P-complete.
a(X,Y) :- e(X,Y).
a(X,Y) :- a(X,Z), a(Y,Z), p(X,Y).
I now conjecture that nonlinear recursion is P-complete iff a tertiary
join is involved (assuming NC not = P). I.e., it is in NC if only
extension joins, and P-complete if tertiary join.
The terminolgy needs some refinement to include cases like
r(X) :- r(Y), r(Z), e(X, Y, Z).
There is no extension join starting from r(X) or from r(Y), but if you
start at e(X, Y, Z) there is.
My hand-wave is: Suppose no extension join. Then there exists in the
variable/argument graph a homeomorph of our basic P-complete examples.
Suppose there is an extension join.
Then the fringe of a proof tree is O(n) and can be
halved in each parallel stage.
∂31-May-85 1344 FORD@SU-CSLI.ARPA Language Processing Workshop
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 31 May 85 13:44:25 PDT
Date: Fri, 31 May 1985 13:36 PDT
Message-ID: <FORD.12115472110.BABYL@SU-CSLI.ARPA>
From: FORD@SU-CSLI.ARPA
To: folks@csli
Subject: Language Processing Workshop
ANNOUNCING
WORKSHOP ON LANGUAGE PROCESSING
sponsored by CSLI and Sloan
Dates: Monday June 10 - Wednesday June 12
(reception Sunday evening, June 9)
Location: CSLI
Several research groups at CSLI are holding an open informal
workshop on language processing, with invited speakers both from the
United States and abroad. The goal of the workshop is to examine
current psychological issues in language processing at the sentence
and discourse level, with particular focus on the relation of language
processing to language structure and the situations in which language
is used. Participants include:
Tom Bever, Herb Clark, Stephen Crain, Gary Dell, Marilyn Ford, Ken
Forster, Don Foss, Willem Levelt, William Marslen-Wilson, James
McClelland, Mark Seidenberg, Dave Swinney, Michael Tanenhaus and
Lorraine Tyler.
Schedule available at front desk, Ventura Hall, or by mail on request
from Sandy McConnell-Riggs (Sandy@csli). Copies of papers by the
speakers relevant to their workshop presentations are available in the
CSLI Reading Room and the Psychology Library, Jordan Hall. For
further information, contact members of the organizing committee: Herb
Clark, Phil Cohen, Marilyn Ford, Barbara Grosz, Ron Kaplan, Marcy
Macken, Stanley Peters and Ivan Sag.
-------
∂31-May-85 1410 PIERCE@SU-SCORE.ARPA House for rent in Palo Alto
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 31 May 85 14:10:10 PDT
Date: Fri 31 May 85 14:06:11-PDT
From: Lee Pierce <PIERCE@SU-SCORE.ARPA>
Subject: House for rent in Palo Alto
To: faculty@SU-SCORE.ARPA
There is a house for rent for the summer months--July and August--
in Palo Alto close to campus. 4 bedrooms, 2 baths, study. Membership
in swim club is included. Dog and cat in residence (optional).
Family preferred. $1500 per month negotiable. Contact pierce@score
for details if interested.
-------
∂31-May-85 1635 BLOCK@SU-CSLI.ARPA CONNECTIONISM
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 31 May 85 16:35:33 PDT
Date: Fri 31 May 85 16:30:19-PDT
From: Ned Block <BLOCK@SU-CSLI.ARPA>
Subject: CONNECTIONISM
To: RRR@SU-CSLI.ARPA
Kurt Van Lehn has brought over some introductory reading materials
for our meeting (this Tuesday at 2:15) on connectionism. I will leave
7 copies of 2 items from the Hinton and Anderson collection
(1) the intro by Rumelhart and Norman
(2) the overview chapter by Anderson and Hinton
Also, I will leave (at the front desk) 35 copies of a 3 page section of
the overview chapter "Parallel Hardware and the Symbol Processing
Paradigm". This section appears to summarize the case against the
symbol-manipulation view of mental processes.
-------
∂31-May-85 1641 ullman@diablo Re: NC, P-completeness, and extension joins
Received: from SU-AIMVAX.ARPA by SU-AI.ARPA with TCP; 31 May 85 16:41:34 PDT
Date: Fri, 31 May 85 16:32:20 pdt
From: Jeff Ullman <ullman@diablo>
Subject: Re: NC, P-completeness, and extension joins
To: avg@diablo, nail@diablo
At SIGMOD, Paris Kanellakis showed me a proof of P-completeness
for another class:
a(x) :- e(x)
a(x) :- a(y), a(z), b(x,y), c(x,z)
[I think I got that right--Paris: confirm?]
∂31-May-85 1746 PARIS@MIT-XX.ARPA P
Received: from SU-AIMVAX.ARPA by SU-AI.ARPA with TCP; 31 May 85 17:46:11 PDT
Received: from MIT-XX.ARPA by diablo with TCP; Fri, 31 May 85 17:41:19 pdt
Date: Fri 31 May 85 20:42:08-EDT
From: Paris Kanellakis <PARIS@MIT-XX.ARPA>
Subject: P
To: nail@SU-AIMVAX.ARPA
Consider the following SiRuP:
a(x):-a(y),a(z),b(y,x),c(z,x)
[ the rule a(x):-e(x) is used to initialize the fixpoint computation]
1) Now consider a monotone circuit consisting of alternating layers
of OR and AND gates. On top it has the input 0's and 1's, the first
layer is an AND layer, where each gate has one input.
The AND gates are constrained to indegree at most two (only first layer
is limited to indegree one). The AND gates can have arbitrary outdegree.
Let last layer be a single AND gate whose output we wish to evaluate.
It is immediate that the circuit value here is complete in P.
2) Modify the above circuit by adding new OR gates at each OR layer
so that each OR gate has outdegree one. This is a simple transformation
since the outdegree of AND gates is unbounded.
Thus here the circuit value is also complete in P.
REDUCTION: Given a circuit value problem of the above type,
construct a dag. There is one node in the dag for each AND gate.
For AND gates of the top layer which have 1 inputs put corresponding
nodes in e(.).
Now all that remains is to construct b(y,x) and c(z,x). These will
be represented by arcs between successive layers. The arcs will
be of two flavors: b and c.
Let AND gate A correspond to node x. It has two incoming wires
from OR gates O1, O2, which in turn have incoming wires from
the AND gates of layer immediately above. From the set of AND gates
whose output go into O1 draw b-flavor arcs to x, from the set of
AND gates whose output go into O2 draw c-flavor arcs to x.
-------------------------------
I think that this reduction simulates the SiRuP.
The only surprising thing about it is that it seems a
stronger version of the PATH ACCESSIBILITY problem.
Also F. Bancilhon had first proposed this example as a bad candidate and
christened it BLUE-BLOODED FRENCHMAN.
-------
∂02-Jun-85 1048 @SU-SCORE.ARPA:coraki!pratt@Navajo Visiting Committee
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 2 Jun 85 10:48:27 PDT
Received: from Navajo ([36.8.0.48].#Internet) by SU-SCORE.ARPA with TCP; Sun 2 Jun 85 10:45:57-PDT
Received: by coraki.uucp (1.1/SMI-1.2)
id AA03203; Sun, 2 Jun 85 10:40:29 pdt
Date: Sun, 2 Jun 85 10:40:29 pdt
From: coraki!pratt@Navajo (Vaughan Pratt)
Message-Id: <8506021740.AA03203@coraki.uucp>
To: faculty@score.ARPA
Subject: Visiting Committee
* The main point of a visiting committee is to allow us to see
ourselves as others see us. Recent faculty would seem to be
handicapped for playing the role of "others."
* The weight of opinion of the visiting committee should be visible to
all to whom they speak at Stanford, both inside and outside the
department. Hence there should be a predilection for
"heavy hitters."
* It should include representatives of academia, government, industry,
and the "younger set." It would also be helpful to have
representatives of national laboratories and professional societies.
There should be one or two major prizewinners. I consider having a
representative of women's views important. A department chairman or
university dean would be helpful, as would a lab director.
The following adds some names to Nils' list, and attempts a ranking in
decreasing order of suitability.
Marvin Dennicoff, ONR Government
David Brandon, SRI Professional society, industry
Steve Cook, Toronto Academia, Turing award
Adele Goldberg, Xerox Professional society, women, industry
Robert Kahn, DARPA Government
Barbara Liskov, MIT Academia, women
Herb Simon, CMU Academia, Nobel prize, Turing award
Richard Watson, LLL National laboratory
David Paterson, UCBerkeley Academia, junior
Ralph Gomory, IBM Industry, lab director
Joel Moses, MIT Academia, department chairman
Allen Newell, CMU Academia, Turing award
Al Aho, ATT Bell Labs Industry
Dick Karp, Berkeley Academia
Bob Fano, MIT Academia
Bob Sproull, SSA Industry, academia
Ken Olsen, DEC Industry
Bernie Olsen Industry
Andy van Dam, Brown Academia
Ivan Sutherland Industry
Bert Sutherland Industry
Cuthbert Hurd, Quintus Industry
John Hopcroft, Cornell Academia
Gerry Sussman, MIT Academia, junior
Jack Schwarz, NYU Academia
Albert Meyer, MIT Academia
Ray Reiter, Toronto Academia
Drew McDermott, Yale Academia, junior
Al Borodin, Toronto Academia
Forest Baskett, DEC Academia, recent faculty
Bob Tarjan, Princeton Academia, recent faculty
∂02-Jun-85 1242 @SU-SCORE.ARPA:WIEDERHOLD@SUMEX-AIM.ARPA More hitters
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 2 Jun 85 12:42:41 PDT
Received: from SUMEX-AIM.ARPA by SU-SCORE.ARPA with TCP; Sun 2 Jun 85 12:38:34-PDT
Date: Sun 2 Jun 85 12:38:02-PDT
From: Gio Wiederhold <WIEDERHOLD@SUMEX-AIM.ARPA>
Subject: More hitters
To: faculty@SU-SCORE.ARPA
Another heavy hitter
Josh Lederberg Rockefeller Univ. Acad-president, Nobel, ex Stanford
also
Klaus Wirth, ETH Zurich Academia, ex Stanford, well known in industry
gio
-------
∂03-Jun-85 0917 BLOCK@SU-CSLI.ARPA Situation Semantics Made Easy
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 3 Jun 85 09:17:36 PDT
Date: Mon 3 Jun 85 09:12:24-PDT
From: Ned Block <BLOCK@SU-CSLI.ARPA>
Subject: Situation Semantics Made Easy
To: su-bboards@SU-CSLI.ARPA, folks@SU-CSLI.ARPA, friends@SU-CSLI.ARPA
SITUATION SEMANTICS MADE EASY
First of three lectures by John Perry
is TODAY, Monday, June 3 at 3:15 in Redwood G-19
(#2 will be Wed, #3 next Monday--same time and place)
-------
∂03-Jun-85 0918 BLOCK@SU-CSLI.ARPA Connectionism REMINDER
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 3 Jun 85 09:18:33 PDT
Date: Mon 3 Jun 85 09:13:37-PDT
From: Ned Block <BLOCK@SU-CSLI.ARPA>
Subject: Connectionism REMINDER
To: RRR@SU-CSLI.ARPA
Tuesday at 2:15 (apparently I mistakenly said 3:15 in an earlier message)
-------
∂03-Jun-85 0923 EMMA@SU-CSLI.ARPA Situation Semantics Made Easy
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 3 Jun 85 09:23:21 PDT
Mail-From: BLOCK created at 3-Jun-85 09:12:25
Date: Mon 3 Jun 85 09:12:24-PDT
From: Ned Block <BLOCK@SU-CSLI.ARPA>
Subject: Situation Semantics Made Easy
To: su-bboards@SU-CSLI.ARPA, folks@SU-CSLI.ARPA, friends@SU-CSLI.ARPA
ReSent-Date: Mon 3 Jun 85 09:18:50-PDT
ReSent-From: Emma Pease <Emma@SU-CSLI.ARPA>
ReSent-To: friends@SU-CSLI.ARPA
SITUATION SEMANTICS MADE EASY
First of three lectures by John Perry
is TODAY, Monday, June 3 at 3:15 in Redwood G-19
(#2 will be Wed, #3 next Monday--same time and place)
-------
∂03-Jun-85 1201 RICE@SUMEX-AIM.ARPA Wednesday's Talk. The MXA Blackboard system.
Received: from SUMEX-AIM.ARPA by SU-AI.ARPA with TCP; 3 Jun 85 12:00:53 PDT
Date: Mon 3 Jun 85 11:59:51-PDT
From: Jim Rice <RICE@SUMEX-AIM.ARPA>
Subject: Wednesday's Talk. The MXA Blackboard system.
To: aap@SUMEX-AIM.ARPA, bhayes-roth@SUMEX-AIM.ARPA
The subject for Wednesday's meeting will be something like :-
The MXA Blackboard system.
--------------------------
This is the blackboard system developed by SPL in Britain so if you want
to find out just how strange we all are over there then this is your chance.
I would think that the batting order should be something like :-
REPEAT
I describe the system.
You can't see why we did it that way so start wingeing at me.
UNTIL bored ()
I don't know what you will want to know so we'll probably have to play it a bit
by ear.
ALSO...
As an addendum to the Famous Feigenbaum Friday riplunch it might be of
interest to you to hear what a real Brit thinks of the Alvey and ESPRIT
programmes. It hardly seems worth having a riplunch just for me to moan for
five minutes so I might as well do it on Wednesday if it's of any interest to
you.
It seems pointless even to think about starting at 9:00 so I plan to
start at about 9:30 if that's all right with you.
===================
Rice.
-------
∂03-Jun-85 1924 SELLS@SU-CSLI.ARPA Re: emergency job
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 3 Jun 85 19:24:42 PDT
Date: Mon 3 Jun 85 17:19:42-PDT
From: Peter Sells <Sells@SU-CSLI.ARPA>
Subject: Re: emergency job
To: BRESNAN@SU-CSLI.ARPA
cc: linguists@SU-CSLI.ARPA, researchers@SU-CSLI.ARPA
In-Reply-To: Message from "Joan Bresnan <BRESNAN@SU-CSLI.ARPA>" of Mon 3 Jun 85 17:16:36-PDT
Please ignore Joan's message. A volunteer has forthcome.
P.
-------
∂03-Jun-85 1924 BRESNAN@SU-CSLI.ARPA emergency job
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 3 Jun 85 19:24:29 PDT
Date: Mon 3 Jun 85 17:10:55-PDT
From: Joan Bresnan <BRESNAN@SU-CSLI.ARPA>
Subject: emergency job
To: linguists@SU-CSLI.ARPA, researchers@SU-CSLI.ARPA
cc: bresnan@SU-CSLI.ARPA, l.iida@SU-CSLI.ARPA
We need someone to ink in tree diagram lines and number the
pages of a 250 page dissertation, to be turned in by Wed. at 5:00
p.m. Is anyone willing to undertake this job? If so, please
respond to l.iida@csli.
-------
∂03-Jun-85 1950 avg@diablo Paris' P-complete problem
Received: from SU-AIMVAX.ARPA by SU-AI.ARPA with TCP; 3 Jun 85 19:49:49 PDT
Date: Mon, 3 Jun 85 17:14:16 pdt
From: Allen VanGelder <avg@diablo>
Subject: Paris' P-complete problem
To: nail@diablo
The proof of P-completeness sent in by Paris looks good, but perhaps
can be simplified. The rules were
a(X) :- e(X).
a(X) :- a(Y), b(Y,X), a(Z), c(Z,X).
It is sufficient to consider boolean circuits whose gates have 2 inputs
and arbitrary fanout, except for input gates, which have 0 inputs and
arbitrary fanout. Associate gate x with constants x0 and x1.
Interpret a(x0) to mean gate x outputs a 1. x1 is used only for or-gates.
For and-gate x, say it has inputs y and z. Put in edges b(y0, x0)
and c(z0, x0).
For or-gate x, say it has inputs y and z. Put in e(x1) and
c(x1, x0) and b(y0,x0) and b(z0,x0).
For input-gate x, put in e(x0).
Using induction on the height of the gate,
it is clear that if gate x outputs a 1, then a(x0) is derivable.
For the other direction, use
induction on the number of proof steps in the shortest derivation.
Let gate x have inputs y and z.
By considering each of the limited number of top-down expansions
of goal a(x0), we conclude that if a(x0) is derivable, then gate x
must output a 1; for an and-gate we show that a(y0) and a(z0) must both
have been derived; for an or-gate we show one of them must have been derived;
They are sub-derivations, so are covered by the inductive hypothesis,
which is "If a(x0) is derivable, then gate x outputs a 1."
This construction is similar to the one used by Jeff for the earlier
nonlinear example, and adapted by me to a second nonlinear example.
My conjecture is that it can be generalized to handle "all" P-complete
function-free logic programs, admittedly a bit ambitious.
The overall construction strategy goes something like this.
For an and-gate x, make it reduce to two subproblems, and make each
subproblem reduce to an input plus a trivial subproblem.
In Paris' problem, this would go
a(x0) :- a(x1), b(x1, x0), a(x2), c(x2, x0).
b(x1, x0).
c(x2, x0).
a(x1) :- a(y0), b(y0, x1), a(x11), c(x11, x1).
a(x11) :- e(x11).
e(x11).
c(x11, x1).
b(y0, x1).
a(x2) :- a(z0), b(z0, x2), a(x22), c(x22, x2).
a(x22) :- e(x22).
e(x22).
c(x22, x2).
b(z0, x2).
For an or-gate, make two alternative reductions, each involving one input
and a trivial sub-problem.
a(x0) :- a(y0), b(y0, x0), a(x1), c(x1, x0).
a(x1) :- e(x1).
e(x1).
b(y0, x0).
c(x1, x0).
a(x0) :- a(z0), b(z0, x0), a(x2), c(x2, x0).
a(x2) :- e(x2).
e(x2).
b(z0, x0).
c(x2, x0).
My suggested construction at the beginning of this message was based
on this idea with some eyeball simplifications. Of course we want a
general construction to be completely mechanical, and not use any
cleverness.
∂04-Jun-85 0835 HEDGES@SU-SCORE.ARPA Faculty Lunch
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 4 Jun 85 08:35:03 PDT
Date: Tue 4 Jun 85 08:35:05-PDT
From: Karen Hedges <HEDGES@SU-SCORE.ARPA>
Subject: Faculty Lunch
To: Faculty@SU-SCORE.ARPA
NOTICE: The Faculty Lunch scheduled for today, June 4 has been cancelled.
The next Faculty Lunch will be held on Tuesday, June 11 in room
146 @ 12:15 pm.
-------
∂04-Jun-85 0954 HEDGES@SU-SCORE.ARPA Visit of Peter Will & Jean-Claude Latombe
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 4 Jun 85 09:53:57 PDT
Date: Tue 4 Jun 85 09:53:24-PDT
From: Karen Hedges <HEDGES@SU-SCORE.ARPA>
Subject: Visit of Peter Will & Jean-Claude Latombe
To: Faculty@SU-SCORE.ARPA
Dr. Peter Will will be visiting on June 17. If any faculty are here that
day and would like to meet with him, please let me know your available times.
Jean-Claude Latombe will be visiting on July 9 for part of the day. If any
faculty would like to meet with him, please let me know your available times.
I do not know yet, exactly what time he will be at Stanford other than it
will be July 9.
Thanks.
Karen Hedges
-------
∂04-Jun-85 1041 FORSTER@SU-CSLI.ARPA Phil. Dept. Colloquium
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 4 Jun 85 10:41:05 PDT
Date: Tue 4 Jun 85 10:31:06-PDT
From: Eckart Forster <FORSTER@SU-CSLI.ARPA>
Subject: Phil. Dept. Colloquium
To: folks@SU-CSLI.ARPA
PHILOSOPHY DEPARTMENT COLLOQUIUM
Speaker: Marcia Baron (Univ. of Illinois, Urbana-Champaign;
Visiting at Stanford)
Title: "Kantian Ethics and Supererogation"
Time: Friday, June 7 3:15
Place: Philosophy Seminar Room 90-92Q
-------
∂04-Jun-85 1154 avg@diablo more NC handwaving
Received: from SU-AIMVAX.ARPA by SU-AI.ARPA with TCP; 4 Jun 85 11:54:44 PDT
Date: Tue, 4 Jun 85 11:48:14 pdt
From: Allen VanGelder <avg@diablo>
Subject: more NC handwaving
To: nail@diablo
Following up on my comments about Paris Kanellakis' P-complete
example, it is instructive to see why my so-called general approach
does NOT work on NC examples, such as:
a(X,Y) :- e(X,Y).
a(X,Y) :- a(X,U), p(U,V), a(V,Y).
Suppose I try to simulate an and-gate by interpreting a(x0,x1) to mean
gate x outputs a 1. Using my bull-in-the-china-shop approach,
a(x0,x1) :- a(x0,x2), p(x2,x3), a(x3,x1).
p(x2,x3).
a(x0,x2) :- a(x0,x4), p(x4, y0), a(y0,x2).
a(x0,x4) :- e(x0,x4).
e(x0,x4).
p(x4,y0).
a(y0,x2) :- a(y0,y1), p(y1,x5), a(x5,x2).
p(y1,x5).
a(x5,x2) :- e(x5,x2).
e(x5,x2).
A similar set of tuples to reduce goal a(x3,x1)
to showing a(z0,z1).
Or-gates are simulated by
a(x0,x1) :- a(x0,x4), p(x4, y0), a(y0,x1).
a(x0,x4) :- e(x0,x4).
e(x0,x4).
p(x4,y0).
a(y0,x1) :- a(y0,y1), p(y1,x5), a(x5,x1).
p(y1,x5).
a(x5,x1) :- e(x5,x1).
e(x5,x1).
A similar set of tuples to reduce goal a(x0,x1)
to showing a(z0,z1).
Now it easily follows that if gate x outputs a 1, then a(x0, x1)
is derivable. However, the converse breaks down. Consider the case
where x is an and-gate. The key gap is in the
argument that the ONLY way to successfully reduce goal a(x0,x1) is the
way suggested by the above display. Looking carefully we see that the
most general reduction is
a(x0,x1) :- a(x0,U), p(U,V), a(V,x1).
Therefore the candidates for p(U,V) range over the entire EDB relation for
p(.,.), not over a few tuples associated with gate x!
Informally, we see that the above rules are an NC program because the
fringe of the proof tree for any tuple a(x,y) consists of a chain of
alternating p's and e's, i.e., e(x,u1), p(u1, u2), e(u2,u3), ..., e(u←k, y).
If this chain contains any repeated elements we can just delete the part
in between and get a new shorter proof. But without repeated elements,
it has O(n↑2) length, where there are n constants in the EDB.
With Jeff's souped up forward-chaining algorithm, we can derive all
tuples in O(log(|fringe|) log(n)) parallel time, possibly adding a few powers
of logs, depending on your parallel model.
∂04-Jun-85 1205 @SU-CSLI.ARPA:PCOHEN@SRI-AI.ARPA No meeting today
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 4 Jun 85 12:04:54 PDT
Received: from SRI-AI.ARPA by SU-CSLI.ARPA with TCP; Tue 4 Jun 85 11:54:27-PDT
Date: Tue 4 Jun 85 11:56:05-PDT
From: Phil Cohen <PCOHEN@SRI-AI.ARPA>
Subject: No meeting today
To: NL4: ;
Obviously, we are not meeting today. The next scheduled meeting is
preempted by the Language Processing Workshop. Following that, on the
18th, Herb Clark will speak. Title and abstract to follow
Phil
-------
∂04-Jun-85 1209 FORD@SU-CSLI.ARPA Schedule for Language Processing Workshop
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 4 Jun 85 12:09:45 PDT
Date: Tue, 4 Jun 1985 12:00 PDT
Message-ID: <FORD.12116503360.BABYL@SU-CSLI.ARPA>
From: FORD@SU-CSLI.ARPA
To: folks@SU-CSLI.ARPA
Subject: Schedule for Language Processing Workshop
SCHEDULE FOR
WORKSHOP ON LANGUAGE PROCESSING
sponsored by Sloan and CSLI
June 10-12, Ventura Hall, Stanford
Mon a.m. Jay McClelland. Marilyn Ford.
Discussion leader Fernando Pereira
p.m. Don Foss. Lorraine Tyler.
Discussion leader Joseph Danks
Tue a.m. Willem Levelt. Gary Dell.
Discussion leader TBA
p.m. William Marslen-Wilson. Mark Seidenberg. Michael Tanenhaus.
Discussion leader Ivan Sag
Wed a.m. Tom Bever. Herbert Clark.
Discussion leader Phil Cohen
p.m. Kenneth Forster. Stephen Crain and Janet Fodor
Discussion leader Peter Sells
Detailed schedules are available at the front desk Ventura Hall.
∂04-Jun-85 1255 DAVIES@SUMEX-AIM.ARPA Rice talk postponed
Received: from SUMEX-AIM.ARPA by SU-AI.ARPA with TCP; 4 Jun 85 12:55:31 PDT
Date: Tue, 4 Jun 1985 12:55 PDT
Message-ID: <DAVIES.12116513234.BABYL@Sumex>
From: DAVIES@Sumex
To: AAP@Sumex, BHayes-Roth@Sumex
Subject: Rice talk postponed
Since several people can't make it to the meeting this Wednesday, Jim
Rice's talk on British blackboards (BBBs) has been postponed until
next week at the same time.
-- Byron
∂04-Jun-85 1323 @SU-SCORE.ARPA:udi@wisc-rsch.arpa POPL 86 Call
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 4 Jun 85 13:23:37 PDT
Received: from wisc-rsch.arpa by SU-SCORE.ARPA with TCP; Tue 4 Jun 85 13:22:16-PDT
Received: by wisc-rsch.arpa; Tue, 4 Jun 85 14:51:08 cdt
Received: from wisc-crys.arpa by wisc-rsch.arpa; Tue, 4 Jun 85 14:41:04 cdt
Date: Tue, 4 Jun 85 14:40:58 cdt
From: fischer@wisc-crys.arpa
Message-Id: <8506041940.AA03366@wisc-crys.arpa>
Received: by wisc-crys.arpa; Tue, 4 Jun 85 14:40:58 cdt
To: udi@wisc-crys.arpa
Subject: POPL 86 Call
Sender: udi@wisc-rsch.arpa
Resent-From: udi@wisc-rsch.arpa (Udi Manber)
Resent-To: theory:;@wisc-rsch.arpa
Call for Papers: 13th POPL
The Thirteenth annual ACM SIGACT-SIGPLAN Symposium on
Principles of Programming Languages
St. Petersburg, Florida, January 13-15, 1986
The symposium will address fundamental principles and
important innovations in the design, definition and imple-
mentation of programming languages and systems. We
encourage papers describing work in which an implemented
system embodies an important principle in such a way that
the usefulness of that principle can be better understood.
Our goal is to present "cutting edge" research, which is in
a constant state of flux. The symposium is therefore in no
sense limited to topics included in previous symposia.
Rather, we are eager for papers on important new topics, and
shall not attempt to prescribe particular topics.
All submitted papers will be read by the program com-
mittee.
Charles N. Fischer, University of Wisconsin (Program Chairman)
Robert Henry, University of Washington
Dexter Kozen, IBM Research
Steven Muchnick, Sun Microsystems
Michael O'Donnell, Johns Hopkins University
Gordon D. Plotkin, University of Edinburgh
Fred B. Schneider, Cornell University
Mark Wegman, IBM Research
Peter Weinberger, AT&T Bell Laboratories
Please submit nine copies[1] of a 6- to 10-page summary
of your paper to the program chairman. Summaries must be
typed double-spaced, or typeset 10 on 16. It is important
to include specific results, and specific comparisons with
other work. The committee will consider the relevance,
clarity, originality, significance, and overall quality of
each summary. Mail to:
Charles N. Fischer
Attn: C. Wasilewski
Computer Sciences Department
University of Wisconsin - Madison
Madison, Wisconsin 53706 U.S.A.
Summaries must be received by the program chairman by
August 9, 1985. Authors will be notified of acceptance or
rejection by September 23, 1985. The accepted papers must
be received in camera-ready form by the program chairman at
the above address by October 15, 1985. Authors of accepted
papers will be expected to sign a copyright release form.
Proceedings will be distributed at the symposium and
will be subsequently available for purchase from ACM. The
local arrangements chairman is Edmund Gallizzi, Eckerd Col-
lege, St. Petersburg, Florida 33733.
←←←←←←←←←←←←←←←←←←←←
[1]Persons submitting papers from countries in which ac-
cess to copying machines is difficult or impossible may sub-
mit a single copy.
∂04-Jun-85 1651 ullman@diablo meeting
Received: from SU-AIMVAX.ARPA by SU-AI.ARPA with TCP; 4 Jun 85 16:51:20 PDT
Date: Tue, 4 Jun 85 16:46:59 pdt
From: Jeff Ullman <ullman@diablo>
Subject: meeting
To: nail@diablo
The topic will be intermediate code.
By the way, there seems to be a dearth of volunteers to talk
about papers. I still have a collection of things that
you should know about, including a very interesting paper
by Bancilhon I just received; it deals with the question of
how to show optimality of a rule-evaluation strategy.
∂05-Jun-85 0002 @SUMEX-AIM.ARPA:YM@SU-AI.ARPA Functional Programming and the Logical Variable
Received: from SUMEX-AIM.ARPA by SU-AI.ARPA with TCP; 5 Jun 85 00:02:37 PDT
Received: from SU-AI.ARPA by SUMEX-AIM.ARPA with TCP; Wed 5 Jun 85 00:01:44-PDT
Date: 5 Jun 85 0000 PDT
From: Yoni Malachi <YM@SU-AI.ARPA>
Subject: Functional Programming and the Logical Variable
To: su-bboards@SU-AIMVAX.ARPA, pai@SUMEX-AIM.ARPA,
aic-associates@SRI-AI.ARPA, c1@SU-CSLI.ARPA
Special seminar, tomorrow:
11am on Thursday, 6th June 1985
in Margaret Jacks Hall, Room 352
FUNCTIONAL PROGRAMMING AND THE LOGICAL VARIABLE
Professor Gary Lindstrom
Department of Computer Science
University of Utah
Salt Lake City, Utah 84112
Logic programming offers a variety of computational effects which go beyond
those customarily found in functional programming languages. Among these
effects is the notion of the "logical variable," i.e. a value determined by the
intersection of constraints, rather than by direct binding. We argue that this
concept is "separable" from logic programming, and can sensibly be incorporated
into existing functional languages. Moreover, this extension appears to
significantly widen the range of problems which can efficiently be addressed in
function form, albeit at some loss of conceptual purity. In particular, a form
of side-effects arises under this extension, since a function invocation can
exert constraints on variables shared with other function invocations.
Nevertheless, we demonstrate that determinacy can be retained, even under
parallel execution. The graph reduction language FGL is used for this
demonstration, by being extended to a language FGL+LV permitting formal
parameter expressions, with variables occurring therein bound by unification.
The determinacy argument is based on a novel dataflow-like rendering of
unification. In addition the complete partial order employed in this proof is
unusual in its explicit representation of demand, a necessity given the "benign"
side-effects that arise. An implementation technique is suggested, suitable for
reduction architectures.
∂05-Jun-85 1053 HEDGES@SU-SCORE.ARPA General Faculty Meeting
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 5 Jun 85 10:53:15 PDT
Date: Wed 5 Jun 85 09:00:40-PDT
From: Karen Hedges <HEDGES@SU-SCORE.ARPA>
Subject: General Faculty Meeting
To: Faculty@SU-SCORE.ARPA
REMINDER: If you have any agenda items for the General Faculty Meeting
please send them to me no later than Friday, June 7.
Thanks.
General Faculty Meeting, Tuesday, June 11 @ 3:15 pm in room 146.
Karen Hedges.
-------
∂05-Jun-85 1059 WRIGHT@SUMEX-AIM.ARPA SIGLunch: June 7
Received: from SUMEX-AIM.ARPA by SU-AI.ARPA with TCP; 5 Jun 85 10:59:17 PDT
Date: Wed 5 Jun 85 09:00:25-PDT
From: Carol Wright/Susie Barnes <WRIGHT@SUMEX-AIM.ARPA>
Subject: SIGLunch: June 7
To: siglunch@SUMEX-AIM.ARPA
Because of RIPLunch, there will be no SIGLunch (Heuristic Programming
Seminar/CS 322), this week.
Susie
-------
∂05-Jun-85 1138 CHEADLE@SU-SCORE.ARPA Black Friday Info
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 5 Jun 85 11:36:05 PDT
Date: Wed 5 Jun 85 11:36:03-PDT
From: Victoria Cheadle <CHEADLE@SU-SCORE.ARPA>
Subject: Black Friday Info
To: faculty@SU-SCORE.ARPA
Office: Margaret Jacks 258, 497-1519
I have put a list of your advisees (including all relevant
data) in your Jacks mailboxes this morning. The names
highlighted in yellow are the ones who were expected
to meet some kind of requirement (per their Grey Tuesday
letter) by Black Friday. A report on their progress
will be made at the meeting.
Please go over the information on your advisees and
feel free to let me know if there is any additional
information that should be added/deleted.
See you at the meeting!
Victoria
-------
∂05-Jun-85 1144 BRODER@SU-SCORE.ARPA Last two AFLB talks this quarter
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 5 Jun 85 11:44:38 PDT
Date: Wed 5 Jun 85 11:43:40-PDT
From: Andrei Broder <Broder@SU-SCORE.ARPA>
Subject: Last two AFLB talks this quarter
To: aflb.all@SU-SCORE.ARPA
cc: su-bboards@SU-SCORE.ARPA
6/6/85 - Profs. D. Knuth and C. Papadimitriou (Stanford)
Solutions to the 1985 Qualifying Examination in Analysis of
Algorithms. (Problem 2 and 3)
Copies of the Exam are available from Phyllis Winkler, in MJH 326.
***** Time and place: June 6, 12:30 pm in MJ352 (Bldg. 460) ******
6/13/85 - Prof. L. Guibas (Stanford)
Solutions to the 1985 Qualifying Examination in Analysis of
Algorithms. (Problem 1)
Copies of the Exam are available from Phyllis Winkler, in MJH 326.
***** Time and place: June 13, 12:30 pm in MJ352 (Bldg. 460) ******
-------
∂05-Jun-85 1159 CHEADLE@SU-SCORE.ARPA Forgot to mention...
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 5 Jun 85 11:58:46 PDT
Date: Wed 5 Jun 85 11:51:13-PDT
From: Victoria Cheadle <CHEADLE@SU-SCORE.ARPA>
Subject: Forgot to mention...
To: faculty@SU-SCORE.ARPA
Office: Margaret Jacks 258, 497-1519
Please bring the information sheets on your advisees to the Black
Friday meeting. We will use these copies to review the students.
Thanks,
Victoria
-------
∂05-Jun-85 1216 @SU-SCORE.ARPA:reid@Glacier Re: Black Friday Info
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 5 Jun 85 12:16:36 PDT
Received: from Glacier by SU-SCORE.ARPA with TCP; Wed 5 Jun 85 11:56:29-PDT
Date: 5 Jun 1985 1154-PDT (Wednesday)
From: Brian Reid <reid@Glacier>
To: Victoria Cheadle <CHEADLE@SU-SCORE.ARPA>
Cc: faculty@SU-SCORE.ARPA
Subject: Re: Black Friday Info
In-Reply-To: Victoria Cheadle <CHEADLE@SU-SCORE.ARPA> /
Wed 5 Jun 85 11:36:03-PDT.
I don't have a Jacks mailbox. Should we CIS refugees send over a
courier?
∂05-Jun-85 1404 vardi@diablo Paris' P-complete problem
Received: from SU-AIMVAX.ARPA by SU-AI.ARPA with TCP; 5 Jun 85 14:03:42 PDT
Date: Wed, 5 Jun 85 13:54:16 pdt
From: Moshe Vardi <vardi@diablo>
Subject: Paris' P-complete problem
To: nail@diablo
Here is an even simpler proof, from PATH SYSTEM.
Recall that PATH SYSTEM is the following program:
a(X):-e(X)
a(X):-a(Y), a(Z), d(X,Y,Z)
Without loss of generality we can assume that in the d relation the second and
third arguments (together) functionally determine the first argument. This is
true because we can replaces each tuple (x,y,z) in d by the tuples
(<x,y,z>,y,z) and (x,<x,y,z>,<x,y,z>) (I'm using <x,y,z> as an identifier that
is unique to the tuple (x,y,z).
Now we can encode every tuple d(x,y,z) by tuple b(y,x) and c(z,x). This
shows that the program
a(X):-e(X)
a(X):-a(Y),a(Z),b(Y,X),c(Z,X)
is P-complete.
Moshe
∂05-Jun-85 1445 REGES@SU-SCORE.ARPA Black Friday reminder
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 5 Jun 85 14:45:05 PDT
Date: Wed 5 Jun 85 14:43:35-PDT
From: Stuart Reges <REGES@SU-SCORE.ARPA>
Subject: Black Friday reminder
To: faculty@SU-SCORE.ARPA
Office: Margaret Jacks 210, 497-9798
Remember that Black Friday starts tomorrow afternoon at 2:30 in Jacks 252.
Since Victoria has done all the work of identifying those students who need to
be discussed, I think the meeting can run very quickly. If you can't make it
and have students who are going to be discussed (as indicated by Victoria's
handout to you), please send a message to either Victoria or me to let us know
what should be said at the meeting.
-------
∂05-Jun-85 1614 ullman@diablo Paris' P-completeness result
Received: from SU-AIMVAX.ARPA by SU-AI.ARPA with TCP; 5 Jun 85 16:14:13 PDT
Date: Wed, 5 Jun 85 16:03:49 pdt
From: Jeff Ullman <ullman@diablo>
Subject: Paris' P-completeness result
To: nail@diablo
On AVG's prodding, I took a look at the "standard"
reduction from P to path systems. The idea is on pp. 379-380
of Hopcroft and Ullman '79, although there we are talking
about another result of Cook's: eqivalence of P to
logspace auxiliary PDA's.
The essence is that we take any polytime TM and redesign it
to make sweeps back and forth across its tape, so given time t,
we can compute in logspace the previous time t' at which the
TM last visited its current square.
The elements in the domain are triples [qZt] that say
"at time t, this TM is in state q, scanning symbol Z."
In the rule a(X) :- a(Y), a(Z), b(Y,X), c(Z,X), think of
a(X) as meaning "true fact". Think of b as saying, if Y was true
at the previous time the TM scanned the same square, then X
has the proper tape symbol. Think of c as saying if Z was true
at the previous time, then X has the correct state at this time.
Thus, b([pUt'], [qVt]) whenever delta(p,U) = (?,V.?) and t' is
the previous time the head was at the position it holds at time
t (roughly, t'=t-sqrt(t)).
c([pUt-1], [qVt]) is true whenever delta(p,U) = (q,?,?).
The point made by Allen is that the original proof of path systems
P-completeness could have been expressed in the form Paris stated
his SyRuP.
---Jeff Ullman
∂05-Jun-85 1729 EMMA@SU-CSLI.ARPA Newsletter June 6, No. 32
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 5 Jun 85 17:29:40 PDT
Date: Wed 5 Jun 85 16:59:26-PDT
From: Emma Pease <Emma@SU-CSLI.ARPA>
Subject: Newsletter June 6, No. 32
To: friends@SU-CSLI.ARPA
Tel: 497-3479
C S L I N E W S L E T T E R
←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←
June 6, 1985 Stanford Vol. 2, No. 32
←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←
A weekly publication of The Center for the Study of Language and
Information, Ventura Hall, Stanford University, Stanford, CA 94305
←←←←←←←←←←←←
CSLI ACTIVITIES FOR *THIS* THURSDAY, June 6, 1985
12 noon TINLunch
Ventura Hall ``Speech Act Distinctions in Syntax''
Conference Room by Jerrold Sadock and Arnold Zwicky
Discussion led by Dietmar Zaefferer
2:15 p.m. CSLI Seminar
Redwood Hall ``Existential Sentences''
Room G-19 Edit Doron, CSLI
Discussion led by Larry Moss
3:30 p.m. Tea
Ventura Hall
4:15 p.m. CSLI Colloquium
Redwood Hall ``An Assumption-Based Truth-Maintenance System''
Room G-19 Johan De Kleer, Xerox PARC, Intelligent Systems Lab.
←←←←←←←←←←←
CSLI ACTIVITIES FOR *NEXT* THURSDAY, June 13, 1985
12 noon TINLunch
Ventura Hall ``Type Raising, Functional Composition,
Conference Room and Non-Constituent Conjunction''
David Dowty, Center for the Advanced Study of
the Behavioral Sciences
(Abstract on page 2)
2:15 p.m. CSLI Seminar
Redwood Hall No seminar
Room G-19
3:30 p.m. Tea
Ventura Hall
4:15 p.m. CSLI Colloquium
Redwood Hall No colloquium
Room G-19
←←←←←←←←←←←←←
ANNOUNCEMENT
No seminars or colloquiums are scheduled for June 6 or June 13
because of the University, end-of-quarter break. TINLunch will be
held on these days. Regular activities will resume on June 20.
!
Page 2 CSLI Newsletter June 6, 1985
←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←
ABSTRACT OF NEXT WEEK'S TINLUNCH
``Type Raising, Functional Composition,
and Non-Constituent Conjunction''
Examples of so-called non-constituent conjunction, like the
following, remain an outstanding problem for base-generated syntactic
theories:
John eats beans on Tuesday and rice on Thursday
John gave Mary a book and Susan a record
John went to Chicago on Tuesday and Detroit on Thursday
John painted the chair red and the table blue
It will be shown that such examples are correctly generated by a
syntactic theory that involves (i) a categorial, rather than
phrase-structure grammar, (ii) a generalized syntactic rule of
functional composition (such as has been frequently suggested in the
categorial literature), and (iii) category assignments for verbs and
NPs that are similar to but differ crucially from those suggested by
Mark Steedman and Partee/Rooth.
It will also be pointed out that the theory recently proposed by
Mark Steedman, in which functional composition alone is supposed to
describe ``extraction'' constructions (without appeal to SLASH
features), predicts that non-constituent conjunction constructions
will be subject to the same island constraints as extractions are.
Though this prediction seems to be borne out in some cases, it is
problematic in others. Comparison will therefore be made between
theories, like Steedman's, which treat both conjunction and extraction
by functional composition, and theories which use functional
composition for conjunction and other local dependencies (like
reflexives) but use SLASH features for long-distance dependencies.
--David Dowty
---------
CSLI REPORT
Report No. CSLI-85-24, ``Computationally Relevant Properties of
Natural Languages and Their Grammar'' by Gerald Gazdar and Geoffrey K.
Pullum, has just been published. This report may be obtained by
writing to David Brown, CSLI, Ventura Hall, Stanford, CA 94305 or
Brown@SU-CSLI.
!
Page 3 CSLI Newsletter June 6, 1985
←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←
WORKSHOP ON LANGUAGE PROCESSING
sponsored by CSLI and Sloan
Monday, June 10 through Wednesday, June 12
Several research groups at CSLI are holding an open, informal
workshop on language processing, with invited speakers from the United
States and abroad. The goal of the workshop is to examine current
psychological issues in language processing at the sentence and
discourse level, with particular focus on the relation of language
processing to language structure and the situations in which language
is used. Schedule of speakers follows:
Monday morning Jay McClelland. Marilyn Ford.
afternoon Don Foss. Lorraine Tyler.
Tuesday morning Willem Levelt. Gary Dell.
afternoon William Marslen-Wilson. Mark Seidenberg.
Michael Tanenhaus.
Wednesday morning Tom Bever. Herbert Clark.
afternoon Kenneth Forster. Stephen Crain and Janet Fodor.
Morning sessions begin at 9; afternoon sessions begin at
1:15 (Monday) or 1:30 (Tuesday and Wednesday).
The schedule is available from Suzanne Parker at the front desk of
CSLI (to receive a copy by mail, contact Sandy McConnell-Riggs
(Sandy@csli)). Copies of papers by the speakers relevant to their
workshop presentations are available in the CSLI Reading Room and the
Psychology Library, Jordan Hall. For further information, contact
members of the organizing committee: Herb Clark, Phil Cohen, Marilyn
Ford, Barbara Grosz, Ron Kaplan, Marcy Macken, Stanley Peters and Ivan
Sag. --Marcy Macken (mmacken@su-csli)
-------
∂05-Jun-85 1736 ullman@diablo NAIL-ing things down at last?
Received: from SU-AIMVAX.ARPA by SU-AI.ARPA with TCP; 5 Jun 85 17:36:14 PDT
Date: Wed, 5 Jun 85 17:28:31 pdt
From: Jeff Ullman <ullman@diablo>
Subject: NAIL-ing things down at last?
To: nail@diablo
I would like to formalize the results of our discussion this
morning. First, Mike Bender is going to take the next crack
at defining the ICODE, with an array-based model of storage
and a unified tuple-at-a-time operator. Pun was intended,
because probably that operator will be, in effect,
the application of an (inverse) unifying substitution to a tuple.
I also want to agree with Allen that a better organization
for the compiler (logic+query => ICODE translator) is the following.
I. Assume that the capture rules are ordered, and we prefer to
use CR i rather than j, if i<j.
Possibly, CR's are really groups of equally expensive CR's.
II. Construct strongly connected components in the rule/goal
graph, with no adornments.
III. for each query Q do
for i := 1 to #CR's do begin
try CR i on the strong component of the node for
query Q, with the appropriate adornment;
if success then goto ICODE generator (IV.)
end
The procedure TRY(A↑q, S, i) takes goal node A with adornment q,
SCC S containing A, and "current" capture rule i, as arguments.
Informally, TRY does some mysterious calculation determined by
what CR i is. It may call TRY(B↑p, T, j), where T is a predecessor
SCC of S, B is one of its goal nodes, and p an adornment for B;
such calls may be made many (many, many,...) times. We must have
j<=i, where i is the global "current" capture rule, not necessarily
the argument of TRY.
A key point is that the results of a call to TRY persist.
Thus, the first thing TRY does is look up in a table to see
whether it already knows that TRY(B↑r, T, k) succeeds, where
k <= j, and r is an adornment no more restrictive that p
(i.e., when p has an f, r has an f).
If so, this call succeeds immediately.
Similarly, the table may say that TRY(B↑p, T, j) fails.
If neither holds, a recursive use of TRY occurs, and whatever
is discovered is installed in the table.
Shaibal's programming project may serve as a model for
storing this information efficiently.
IV. Having succeeded in capturing the goal node, we generate
intermediate code. It is important that the table store
not only what arguments to TRY succeeded, but what arguments were used
for recursive calls to TRY on predecessor SCC's of S.
Armed with this information, we know what capture rule
(or rules--since a SCC may have to be captured in several
different contexts) were used to capture each SCC.
A "treewalk" allows us to visit each SCC the appropriate
number of times and generate ICODE corresponding to the CR
at each visit.
∂05-Jun-85 1808 SELLS@SU-CSLI.ARPA party invitation
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 5 Jun 85 18:07:54 PDT
Date: Wed 5 Jun 85 18:02:53-PDT
From: Peter Sells <Sells@SU-CSLI.ARPA>
Subject: party invitation
To: folks@SU-CSLI.ARPA
The 2nd Annual CSLI
**Welcome Summer** Party
CSLI cordially invites you and yours to its
End-of-School-Year Party
at the Bechtel International Center
Friday June 14th.
Barbecue and Volleyball from 6.30 pm
Dancing from 9 pm to 1 am
Live music by DEAD TONGUES
Drinks for all will be provided.
Please bring food for the potluck barbecue:
meat, bread, side-dishes, desserts etc.
(Any comments, suggestions, inquiries to Sells@csli)
-------
∂05-Jun-85 2132 ullman@diablo Linstrom talk
Received: from SU-AIMVAX.ARPA by SU-AI.ARPA with TCP; 5 Jun 85 21:32:10 PDT
Date: Wed, 5 Jun 85 21:28:21 pdt
From: Jeff Ullman <ullman@diablo>
Subject: Linstrom talk
To: nail@diablo
I note that Gary Lindstrom is talking about logic and functional
programming, just 24 hours to the minute after our last NAIL meeting.
However, his talk is in 352 MJH.
∂05-Jun-85 2337 CHEADLE@SU-SCORE.ARPA Black Friday info
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 5 Jun 85 23:37:35 PDT
Date: Wed 5 Jun 85 23:37:22-PDT
From: Victoria Cheadle <CHEADLE@SU-SCORE.ARPA>
Subject: Black Friday info
To: Faculty@SU-SCORE.ARPA
Office: Margaret Jacks 258, 497-1519
Please feel free to send me any additional comments or information
you feel is missing form any of the students' records. I will be
happy to update them before the meeting.
I will also have additional copies of the records at the meeting
if you somehow forget yours.
Victoria
-------
∂06-Jun-85 1028 JODY@SU-CSLI.ARPA Burrito Bandito's visit
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 6 Jun 85 10:27:52 PDT
Date: Thu 6 Jun 85 08:49:36-PDT
From: Joe Zingheim <JODY@SU-CSLI.ARPA>
Subject: Burrito Bandito's visit
To: folks@SU-CSLI.ARPA
cc: consultants@SU-CSLI.ARPA
The Burrito Bandito comes again! Burritos are, once again, $3.00 and $3.75
for the regular and especial, and soft tacos are $1.75.
Carnitas (shredded pork) with rice
Pollo (chicken) with rice
Chile Verde (pork in green sauce) with rice
Chile Colorado (beef in red sauce) with rice
Lengua (beef tongue) with with rice
Chile Relleno (green chile stuffed with cheese) and rice
Picadillo (shreaded beef) with beans
Carne Asada (Mexican style steak) with beans
Chorizo (Mexican sausage) with beans
Send mail, as usual, to Lunch@CSLI
-------
∂06-Jun-85 1027 EMMA@SU-CSLI.ARPA Newsletter Announcement Correction
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 6 Jun 85 10:26:49 PDT
Date: Thu 6 Jun 85 08:13:24-PDT
From: Emma Pease <Emma@SU-CSLI.ARPA>
Subject: Newsletter Announcement Correction
To: friends@SU-CSLI.ARPA
Tel: 497-3479
Should read no Seminars or Colloquiums on June 13 and June 20. There
will be a full schedule of events today. Regular schedule will resume
on June 27.
-Emma
-------
∂06-Jun-85 1040 EMMA@SU-CSLI.ARPA Today's Seminar
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 6 Jun 85 10:40:33 PDT
Date: Thu 6 Jun 85 10:32:09-PDT
From: Emma Pease <Emma@SU-CSLI.ARPA>
Subject: Today's Seminar
To: friends@SU-CSLI.ARPA
Tel: 497-3479
The 2:15 seminar by Edit Doron has been cancelled. It will be
rescheduled for sometime this summer.
-Emma
-------
∂06-Jun-85 1101 SCHAFFER@SU-SCORE.ARPA Mail Folders for Summer Visitors
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 6 Jun 85 10:59:58 PDT
Date: Thu 6 Jun 85 08:23:55-PDT
From: Alejandro Schaffer <SCHAFFER@SU-SCORE.ARPA>
Subject: Mail Folders for Summer Visitors
To: faculty@SU-SCORE.ARPA
I suppose that many of you will be hosting summer visitors who would like
to receive mail here during their stay. If you send me their names I can
prepare mail folders for them. I will give the mail folders to the hosts
who can then put them in the file drawers at the appropriate time.
I will be going away this summer, so if you choose to wait, the
visitors will have to prepare their own folders. Of course, that is
not particularly difficult.
Send replies to schaffer@score
Alejandro Schaffer
-------
∂06-Jun-85 1200 BSCOTT@SU-SCORE.ARPA Senior Research Associates
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 6 Jun 85 12:00:06 PDT
Date: Thu 6 Jun 85 12:00:22-PDT
From: Betty Scott <BSCOTT@SU-SCORE.ARPA>
Subject: Senior Research Associates
To: Tenured@SU-SCORE.ARPA
cc: Hedges@SU-SCORE.ARPA, BScott@SU-SCORE.ARPA
The following persons have been suggested for promotion to Senior Research
Associate:
William Clancey
Peter Friedland
Penny Nii
Nils Nilsson asked me to let you know that these promotions will be taken
up at the senior faculty meeting which will follow the general faculty meeting
on Tuesday, June 11. Each person being recommended should have a faculty
member at the meeting to present the case. If there is no one present to
do this, then the promotion consideration will be delayed until the next
meeting of the senior faculty.
Betty
-------
∂06-Jun-85 1218 vardi@diablo PTIME and NC
Received: from SU-AIMVAX.ARPA by SU-AI.ARPA with TCP; 6 Jun 85 12:18:31 PDT
Date: Thu, 6 Jun 85 12:12:45 pdt
From: Moshe Vardi <vardi@diablo>
Subject: PTIME and NC
To: nail@diablo
Recent message seems to assume that what would push a problem to PTIME is
not the nonlinearity of the rules, but the existence of "supporting" relation
in the rules. Consider the following example:
a(x,y):-e(x,y)
a(x,y):-a(u,v),a(u,x),a(v,y)
Consider the instances where e conatains the following pairs:
(a0,a1) (b0,b1)
(a1,a2) (b1,b2)
. .
. .
(an-1,an) (bn-1,bn)
and (a0,b0).
The program would takes n steps to prove that a(an,bn). Is this problem
P-complete? in NC?
Moshe
∂06-Jun-85 1414 LIBRARY@SU-SCORE.ARPA Math/CS Library New Books
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 6 Jun 85 14:14:33 PDT
Date: Thu 6 Jun 85 14:10:21-PDT
From: C.S./Math Library <LIBRARY@SU-SCORE.ARPA>
Subject: Math/CS Library New Books
To: su-bboards@SU-SCORE.ARPA
cc: faculty@SU-SCORE.ARPA
The Munich Project CIP. Volume 1: The Wide Spectrum Language CIP-L.
by the CIP Language Group. QA76.6.M86 1985 V. 1
Abstraction For Programmers. by Zimmer QA76.6.Z56 1985
Extensional Relationships In Information Retrieval (Project Information
Bridge). Gesellschaft Fur Information Und Dokumentation MBH (GID)
by Wessel. (8507759)
Getting Into VSAM; An Introduction and Technical Reference. by Bouros
QA76.9.V5.B68 1985
Renewing U. S. Mathematicts: Critical Resource for the Future. Report
of the ad hoc committee on Resources for the Mathematical Sciences.
The Commission on Physical Sciences, Mathematics, and Resources
National Research Council. National Academy Press. 1984
H.Llull
-------
∂06-Jun-85 1517 LIBRARY@SU-SCORE.ARPA Lasers in Graphics--Proceedings
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 6 Jun 85 15:17:16 PDT
Date: Thu 6 Jun 85 15:05:00-PDT
From: C.S./Math Library <LIBRARY@SU-SCORE.ARPA>
Subject: Lasers in Graphics--Proceedings
To: su-bboards@SU-SCORE.ARPA
cc: faculty@SU-SCORE.ARPA
Lasers in Graphics -- Electronic Publishing in the 80's conference.
Third annual conference. Two volumes.
This volume and series located in the Engineering Library.
Z249.4.L37 1981
H.Llull
-------
∂06-Jun-85 2032 avg@diablo Moshe's problem is NC (I think)
Received: from SU-AIMVAX.ARPA by SU-AI.ARPA with TCP; 6 Jun 85 20:32:03 PDT
Date: Thu, 6 Jun 85 20:27:33 pdt
From: Allen VanGelder <avg@diablo>
Subject: Moshe's problem is NC (I think)
To: nail@diablo
Regarding Moshe's example where a(X,Y) depends on 3 subgoals of
itself, I expect it to be in NC on this informal basis:
Any derivable tuple a(c1,c2) has a minimum proof tree whose fringe is
polynomial in n, the number of constants in the EDB. The fringe is
a chain of e's, where there are up to 4 kinds of links, arg 1 to arg 1,
arg 1 to arg 2, arg 2 to arg 1, and arg 2 to arg 2.
I claim that a
given tuple e(x,y) can occur at most 4 times in the fringe of the minimum
proof tree. For suppose e(x,y) occurs twice with the same link to its
left neighbor. E.g., if the 1-1 link occurs twice, we have
... e(x,u) e(x,y) ... e(x,v) e(x,y) ...
-----------------
But now the underlined portion of the chain can be removed, and the remaining
chain is the fringe of a derivation of a(c1,c2) also.
So it appears that 4n↑2 bounds the fringe, and Jeff's algorithm runs in
log↑something time. Maybe the constant 4 is too low because certain links
need to be neighbors, but I think the idea is sound.
∂07-Jun-85 1051 vardi@diablo A pure-recursion P-complete program
Received: from SU-AIMVAX.ARPA by SU-AI.ARPA with TCP; 7 Jun 85 10:50:57 PDT
Date: Fri, 7 Jun 85 10:10:42 pdt
From: Moshe Vardi <vardi@diablo>
Subject: A pure-recursion P-complete program
To: nail@diablo
a(x,y,z,w):-e(x,y,z,w)
a(x,y,z,w):-a(y,y1,y2,y3),a(z,z1,z2,z3),a(w,x,y,z)
This program is just PATH SYSTEM. Now we just have to reduce 4-ary relations
to binary relations to show that even with binary relation we can get
P-completeness without supporting relations. This can be done in many ways.
I can think of at least one way, but I won't write it down here, because it
gives rise to rule with more than 60 claues on the LHS. Undoubtedly, somebody
can do better.
Moshe
∂07-Jun-85 1053 vardi@diablo Re: Moshe's problem is NC (I think)
Received: from SU-AIMVAX.ARPA by SU-AI.ARPA with TCP; 7 Jun 85 10:52:56 PDT
Date: Fri, 7 Jun 85 09:25:24 pdt
From: Moshe Vardi <vardi@diablo>
Subject: Re: Moshe's problem is NC (I think)
To: avg@diablo, nail@diablo
Does that mean that you can do any recursion without supporting predicates in
NC?
Moshe
∂07-Jun-85 1126 avg@diablo Re: Moshe's problem is NC (I think)
Received: from SU-AIMVAX.ARPA by SU-AI.ARPA with TCP; 7 Jun 85 11:25:56 PDT
Date: Fri, 7 Jun 85 11:17:41 pdt
From: Allen VanGelder <avg@diablo>
Subject: Re: Moshe's problem is NC (I think)
To: nail@diablo
To: avg@diablo, nail@diablo
Does that mean that you can do any recursion without supporting
predicates in NC?
Moshe
I am not sure. Your example's "weakness" was that every proof could be
reduced to a chain of EDB tuples, which will be the case with binary
predicates and a rule in which each variable appears only twice.
By a "chain" I mean a list of subgoals in
which all occurrences of any given variable are within a bounded
distance of each other in the list, the bound depending on the rules, but
not on the EDB's contents.
For your example the bound was 1.
If you generalize to 3 arguments, or have
some variables appear 3 times, I suspect that P-complete examples exist,
but have not come up with one yet. The important lemmas will be of the
form: a(x0, Y, Z) is derivable only for Y and Z equal to ...
Where "..." is a small set of constants related locally to the simulation
of a boolean circuit, i.e., related to gates adjacent to the gate
associted with constant x0.
This is easier to enforce with p(x0, Y, Z) where p is a supporting
relation (read EDB predicate), because p has no recursive rules.
Here is a candidate for P-completeness with binary predicates and no
"supporting relation" in the recursion. Note that there are two
"paths from X to Y" in the subgoals, X->U->Y and X->V->Y. The intuition
for a reduction from boolean circuits is that when X and Y are constants
associated with a certain and-gate g, then the X->U->Y path should be
derivable if and only if the left input to g is 1, and the X->V->Y
path should be derivable iff the right input is 1.
a(X,Y) :- e(X,Y).
a(X,Y) :- a(X,U), a(Y,U), a(V,X), a(V,Y).
This is clearly equivalent to 3CNF form:
a(X,Y) :- e(X,Y).
a(X,Y) :- b(X,Y), c(X,Y).
b(X,Y) :- a(X,U), a(Y,U).
c(X,Y) :- a(V,X), a(V,Y).
∂07-Jun-85 1256 vardi@diablo Re: Allen's problem is NC
Received: from SU-AIMVAX.ARPA by SU-AI.ARPA with TCP; 7 Jun 85 12:56:10 PDT
Date: Fri, 7 Jun 85 11:57:55 pdt
From: Moshe Vardi <vardi@diablo>
Subject: Re: Allen's problem is NC
To: nail@diablo
I don't have a proof, but I'll bet $10 it's in NC.
Moshe
∂07-Jun-85 1353 vardi@diablo Re: A pure-recursion P-complete program
Received: from SU-AIMVAX.ARPA by SU-AI.ARPA with TCP; 7 Jun 85 13:52:49 PDT
Date: Fri, 7 Jun 85 13:43:17 pdt
From: Moshe Vardi <vardi@diablo>
Subject: Re: A pure-recursion P-complete program
To: nail@diablo
I mistyped my program. Here is how it should be:
a(x,y,z,w):-e(x,y,z,w)
a(x,x,y,z):-a(y,y1,y2,y3),a(z,z1,z2,z3),a(w,x,y,z)
Moshe
∂07-Jun-85 1559 avg@diablo Moshe's pure recursion
Received: from SU-AIMVAX.ARPA by SU-AI.ARPA with TCP; 7 Jun 85 15:59:10 PDT
Date: Fri, 7 Jun 85 15:52:46 pdt
From: Allen VanGelder <avg@diablo>
Subject: Moshe's pure recursion
To: nail@diablo
a(x,y,z,w):-e(x,y,z,w)
a(x,x,y,z):-a(y,y1,y2,y3),a(z,z1,z2,z3),a(w,x,y,z)
Moshe's 4-ary example that "embodies" PATH SYSTEM becomes clearer
if rewritten as
a(W, X, Y, Z) :- e(W, X, Y, Z).
r(X) :- a(X, ←, ←, ←).
a(X, X, Y, Z) :- r(Y), r(Z), a(←, X, Y, Z).
Note the use of underscore for anonymous variables. Each underscore
represents a different variable that occurs nowhere else.
I highly recommend this Prolog convention.
To get path system, put in only e's of the form e(0,x,y,z) where 0
is a constant not corresponding to any node, and x, y, z are nodes
that have a hyper-edge h(x,y,z). Also put in e(x,x,x,x) for starting nodes.
The key lemma is that any a(....)
derived by recursion has equal 1st and 2nd arguments, and so can't
be used as the third subgoal a(W,X,Y,Z) to derive any new tuples
(it would just re-derive itself). Therefore the only useful source
of a(W,X,Y,Z) is the hyper-edges represented by e(0,X,Y,Z).
Moshe, perhaps you can reduce this version to binary predicates
without going to 60+ clauses??
Allen
∂07-Jun-85 1637 vardi@diablo Re: Moshe's pure recursion
Received: from SU-AIMVAX.ARPA by SU-AI.ARPA with TCP; 7 Jun 85 16:36:46 PDT
Date: Fri, 7 Jun 85 16:32:25 pdt
From: Moshe Vardi <vardi@diablo>
Subject: Re: Moshe's pure recursion
To: avg@diablo, nail@diablo
Sorry. Not 60+ clauses, but 60+ literals. Not that that would make Allen
happy. I get it by a general reduction of 4-tuples to 2-tuples, whereby
<a,b,c,d> is replaced by
<a0,a>,<a0,b0>,<b0,b>,<b0,c0>,<c0,c>,<c0,d0>,<d0,d>
<z,a>,<z,b>,<z,c>,<z,d>
<z,z1>,<z1,z2>,<z2,z3>,<z3,z4>,<z4,z5>
There are much more economical encoding of 4-tuples by pairs,
but we need an encoding that can be decoded by Horn formulas.
Still, there cpuld be better encodings, and there could be better
encodings for the particular problems we have in mind.
The only point I was trying to make is that pure recursion can also
be bad, i.e., P-complete.
Moshe
∂08-Jun-85 0039 @SU-SCORE.ARPA:TOB@SU-AI.ARPA
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 8 Jun 85 00:39:14 PDT
Received: from SU-AI.ARPA by SU-SCORE.ARPA with TCP; Sat 8 Jun 85 00:35:41-PDT
Date: 08 Jun 85 0034 PDT
From: Tom Binford <TOB@SU-AI.ARPA>
To: faculty@SU-SCORE.ARPA
Visiting committee
Steve Johnson, Bell Labs
Larry Druffel, Rational Machines
Larry Roberts
∂09-Jun-85 2250 avg@diablo Pure recursion P-complete
Received: from SU-AIMVAX.ARPA by SU-AI.ARPA with TCP; 9 Jun 85 22:50:26 PDT
Date: Sun, 9 Jun 85 22:47:20 pdt
From: Allen VanGelder <avg@diablo>
Subject: Pure recursion P-complete
To: nail@diablo
I offered the following example as a possible P-complete instance of
pure recursion with binary predicates.
a(X,Y) :- e(X,Y).
a(X,Y) :- b(X,Y), c(X,Y).
b(X,Y) :- a(X,U), a(Y,U).
c(X,Y) :- a(V,X), a(V,Y).
Here is my idea for reduction from MCV. We want a(x0, x1) to hold iff gate x
outputs a 1. Let and-gate x have two inputs, gates y and z. Put the
following into the EDB:
e(x1, x2). e(x3, x0).
e(x2, y0). e(x3, z0).
e(x4, x0). e(x5, x1).
e(x4, x2). e(x5, x3).
e(x0, y1). e(x1, z1).
e(x6, x0). e(x7, x1).
e(x6, y0). e(x7, z0).
The tuples in the first column, when supplemented by a(y0,y1), derive
b(x0,x1).
The tuples in the second column, when supplemented by a(z0,z1), derive
c(x0,x1).
Thus a(y0,y1) and a(z0,z1) derive a(x0,x1).
The usual idea for or-gates is used. Make column 2
e(x3,x0).
e(x3,x1).
so that a(y0,y1) alone derives a(x0,x1). Then add stuff for input gate z.
To show that nothing spurious is derived requires an exhausting case
analysis. Not worth it for ten bucks.
∂10-Jun-85 0037 RESTIVO@SU-SCORE.ARPA PROLOG Digest V3 #26
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 10 Jun 85 00:37:25 PDT
Date: Saturday, June 8, 1985 9:00PM
From: Chuck Restivo (The Moderator) <PROLOG-REQUEST@SU-SCORE.ARPA>
Reply-to: PROLOG@SU-SCORE.ARPA
US-Mail: P.O. Box 4584, Stanford CA 94305
Phone: (415) 326-5550
Subject: PROLOG Digest V3 #26
To: PROLOG@SU-SCORE.ARPA
PROLOG Digest Monday, 10 Jun 1985 Volume 3 : Issue 26
Today's Topics:
Query - AI and the CAIS,
Reasoning - Time,
Implementation - Warren Engine & External Calls & Events,
LP Library - 1985 SLP & New LIST
----------------------------------------------------------------------
Date: Thursday, 6 Jun 85 09:30:37 PDT
From: "David M. Meyer" <Meyer%tekchips%tektronix@csnet-relay>
Subject: Warren Prolog Engine
We are interested in the use of "put←variable Xn, Ai" in
Warren's "New Prolog Engine".
Basically, why would one ever need to save a temporary
variable on the heap (we're never going to reference it)?
Why not just create an unbound variable in Ai, that is,
Ai := ref←to(Ai)?
For example, consider the following example:
f(X, Y, Z) :- h(Q), g(Y, Z).
This will generate the following code (before optimization):
allocate
get←variable X1, A1
get←variable Y1, A2
get←variable X3, A3
put←variable X4, A1 % What should be generated here?
call h/1, 1
put←unsafe←value Y1, A1
put←variable X5, A2 % What happens here (i.e., why does
% this need to be on the heap?)
deallocate
execute g/2
Note that where the instruction "put←variable X4, A1" was
generated in the above code, Warren doesn't give us a byte-code,
since the "put←variable Xn, Ai" instruction is only used for
an unbound (temporary) variable in the *last* goal.
Thank you,
- Kent Beck and David Meyer
------------------------------
Date: Sat 1 Jun 85 17:24:03-PDT
From: Fernando Pereira <PEREIRA@SU-CSLI.ARPA>
Subject: Reasoning about time
I've just heard about some work that might be just what you need.
Write to
Prof. R. Kowalski
Dept. of Computing
Imperial College
180 Queen's Gate
London SW7 2BZ
England
asking for his paper on ``event calculus''. I haven't seen
the paper yet, but from a recent conversation with him it
would seem very relevant to your needs.
-- Fernando Pereira
------------------------------
Date: 24 May 85 17:48:32 +1000 (Fri)
From: decvax!mulga!mungunni.oz!Lee@Berkeley
Subject: External calls
The latest version of MU-Prolog allows object files to
be loaded at run time and linked to Prolog predicates.
This version is not yet released, but if people
particularly want the feature, we could give them that
version. Its a pretty awful hack and is unlikely to
be portable but the current version works under 4.2BSD
UN*X on a VAX, Pyramid and Perkin Elmer 3240. I believe
Basser Prolog also has this facility, though I don't know
if it is distributed or how portable it is.
-- Lee Naish
------------------------------
Date: Mon, 3-Jun-85 09:39:10 PDT
From: Vantreeck%Logic.DEC@DECWRL.ARPA
Subject: Event handling in Prolog
I'm not a theorist. From a programming point of view,
the ease of establishing event handlers depends on the
hardware OS support more than some limitation of Prolog.
Adding event handlers to Prolog implementations on VAX/VMS
should be a trivial task. Adding an Prolog predicate to
generate an exception, should also be easy.
Besides control-C, probably the best other example
common of a Prolog event handler, would be a garbage
collection which occurs on the event of a stack running
into a protected page. Establishing a handler for integer
or floating point underflow over flow is easy - provided
that the calculations are done with routines that generate
an exception. I don't know if implementations like QUINTAS
Prolog, which uses their own routines for their short floats,
generate exceptions on underflow/overflow.
If I were to write a concurrent Prolog, I think I
would probably use an event handler to manage the
concurrency. I hope that the newer Prologs that will be
hitting the market in the next year or two of years will
have predicates for establishing and handling events.
-- George Van Treeck
------------------------------
Date: 3 Jun 1985 08:10-PDT
From: JWOLFE@USC-ECLB
Subject: AI and the CAIS
Dear Colleagues:
The Institute for Defense Analysis (IDA) is
investigating the Common APSE Interface Set (CAIS) for Ada
Programming Support Environments (APSE). Although it is a
proposed DoD standard, it is unlike other standards in that it
will evolve as interface needs are identified. My particular
interest is in what needs are generated by the introduction of AI
languages and techniques (e.g. LISP, PROLOG, Knowledge Bases,
Inferencing, etc.) to software engineering environments. This
issue may effect the CAIS at two levels: First, tools may be
written using AI languages and techniques; second, applications
may use AI languages or techniques which may require unique
tools. I am coming to the Ada and AI communities to solicit your
help and advise. I would appreciate hearing from anyone who has
experience in:
* Using AI languages and techniques to develop software
tools.
* Developing tools for the life cycle maintenance of
AI software.
* Anyone else who can contribute.
Since the results of this study will influence the evolution
of the CAIS standard, it is important to have the participation
of government, industry, and acedemia. Please note that since
IDA is a FCRC, issues of proprietary information and
non-disclosure CAN be resolved. Your participation is
needed and appreciated.
-- James Wolfe
Institute for Defense Analysis
------------------------------
Date: Fri, 24 May 85 10:48:44 EDT
From: Doug DeGroot <Degroot.Yktvmv%ibm-sj@csnet-relay>
Subject: Registration/Hotel forms for 1985 SLP
============================================================
1985 Symposium on Logic Programming
July 15-19, 1985
Sheraton Boston Hotel and Towers
Boston, Massachusetts
Advance Registration Form:
←←←←←←←←←←←←←←←←←←←←←←←←←
Mail to and make checks payable to:
1985 Symposium on Logic Programming
c/o IEEE Computer Society
1109 Spring St., Suite 300
Silver Spring, Maryland 20910
USA
deadline for advance registration: July 1, 1985
Name: ←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←
Affiliation: ←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←
Address: ←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←
←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←
City: ←←←←←←←←←←←←←←←←←←←←←←←←← State: ←←←←←←←←←← Zip: ←←←←←←←
Country: ←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←
Tutorials: Advance Late
←←←←←←←←← Registration Registration
IEEE Member: $140 each $170 each
Non-member: $175 each $215 each
Symposium:
←←←←←←←←←
IEEE Member: $185 $215
Non-member: $220 $260
Student*: $50* $50*
Make checks payable to: 1985 Symposium on Logic Programming
Amount enclosed: ←←←←←←←←←←←←←←←←←
No refunds after July 12, 1985
All refunds subject to a $10.00 service charge.
* Student registration does not include the banquet or
cocktail parties.
============================================================
1985 Symposium on Logic Programming
July 15-19, 1985
Sheraton Boston Hotel and Towers
Boston, Massachusetts
Hotel Registration Form:
←←←←←←←←←←←←←←←←←←←←←←←
Mail to: Sheraton Boston Hotel and Towers
attn: Reservations Office
39 Dalton Street
Boston, MA 02199
Rate: $110/night (single or double occupancy)
A block of rooms has been reserved for the participants.
Reservations should be made directly with the hotel before
June 14, 1985.
Name: ←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←
Affiliation: ←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←
Address: ←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←
←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←
City: ←←←←←←←←←←←←←←←←←←←←←←←←← State: ←←←←←←←←←← Zip: ←←←←←←←
Country: ←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←
Arrival Date: ←←←←←←←←←←←←←←←←←←← Time: ←←←←←←←←←a.m./p.m.
Departure Date: ←←←←←←←←←←←←←←←←← Time: ←←←←←←←←←a.m./p.m.
Amount enclosed: ←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←
One night's deposit required if arrival time is after 6 p.m.
------------------------------
Date: Fri, 7 Jun 85 17:27:27 edt
From: Bruce Smith <BTS%UNC@csnet-relay>
Subject: New LIST of PROLOGs
The "LIST of PROLOGs" was started by Randy Harr (then at
CWRU) and myself, largely from the USENET and ARPANET
computer networks. It's grown over the last couple of
years, with help from numerous Prolog folks.
The list emphasizes Prolog systems that are currently
available. Also, I've tried to note which vendors offer
educational discounts. There are a lot of new Prolog
systems being announced, with rumors of still more on the
way. In particular, I expect that this list'll be very
much out of date after this summer's Logic Programming
Symposium and, of course, IJCAI.
Please let me know of any additions or corrections
to the list. Sorry, but I didn't have time to verify all
the information. Some of the entries are also over a year
old and are likely out of date.
-- Bruce T. Smith
[ this file is available from the SCORE:PS:<Prolog>
directory as Prolog.NImplementations -ed ]
------------------------------
End of PROLOG Digest
********************
∂10-Jun-85 1318 @SU-CSLI.ARPA:PCOHEN@SRI-AI.ARPA Next NL4 talk, Lucy Suchman
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 10 Jun 85 13:17:50 PDT
Received: from SRI-AI.ARPA by SU-CSLI.ARPA with TCP; Mon 10 Jun 85 13:11:42-PDT
Date: Mon 10 Jun 85 13:11:50-PDT
From: Phil Cohen <PCOHEN@SRI-AI.ARPA>
Subject: Next NL4 talk, Lucy Suchman
To: NL4: ;
cc: finterest@SU-CSLI.ARPA
Our next NL4 seminar will be June 18, at 12:45, in the Ventura Seminar
room.
Speaker: Lucy Suchman
Intelligent Systems Laboratory
Xerox PARC
Plans and Situated Actions
Researchers in Cognitive Science take the organization and
significance of action to be derived from plans, which are
prerequisite to and prescribe action at whatever level of detail one
might imagine. Mutual intelligibility on this view is a matter of the
recognizability of plans, due to common conventions for the expression
of intent, and common knowledge about typical situations and
appropriate actions. An alternative view, drawn from recent work in
social science, treats plans as derived from situated actions.
Situated actions as such comprise necessarily ad hoc responses to the
actions of others and to the contingencies of particular situations.
Rather than depend upon the reliable recognition of intent, successful
interaction consists in the in situ production of intelligibility,
through mutual access to situation resources, and through the
detection, repair or exploitation of differences in understanding.
As common sense formulations, designed to accomodate the unforseeable
contingencies of situated action, plans are inherently vague. One
approach to human-machine communication has been to improve on the
vagueness of plans, to make them the basis for artifacts able to
interact with their human users. This talk looks at the problem of
human-machine interaction through a case study of people using a machine
designed on the planning model, and intended to be intelligent and
interactive. A conversation analysis of "interactions" between users
and the machine reveals that the machine's insensitivity to particular
circumstances is both a central design resource, and a fundamental
limitation. More generally, problems in Cognitive Science's theorizing
about purposeful action as a basis for machine intelligence can be
understood as an artifact of the substitution of plans for actions, and
representations of the situation of action for action's actual
circumstances.
-------
∂10-Jun-85 1318 @SU-CSLI.ARPA:PCOHEN@SRI-AI.ARPA Next NL4 talk, Lucy Suchman
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 10 Jun 85 13:17:50 PDT
Received: from SRI-AI.ARPA by SU-CSLI.ARPA with TCP; Mon 10 Jun 85 13:11:42-PDT
Date: Mon 10 Jun 85 13:11:50-PDT
From: Phil Cohen <PCOHEN@SRI-AI.ARPA>
Subject: Next NL4 talk, Lucy Suchman
To: NL4: ;
cc: finterest@SU-CSLI.ARPA
Our next NL4 seminar will be June 18, at 12:45, in the Ventura Seminar
room.
Speaker: Lucy Suchman
Intelligent Systems Laboratory
Xerox PARC
Plans and Situated Actions
Researchers in Cognitive Science take the organization and
significance of action to be derived from plans, which are
prerequisite to and prescribe action at whatever level of detail one
might imagine. Mutual intelligibility on this view is a matter of the
recognizability of plans, due to common conventions for the expression
of intent, and common knowledge about typical situations and
appropriate actions. An alternative view, drawn from recent work in
social science, treats plans as derived from situated actions.
Situated actions as such comprise necessarily ad hoc responses to the
actions of others and to the contingencies of particular situations.
Rather than depend upon the reliable recognition of intent, successful
interaction consists in the in situ production of intelligibility,
through mutual access to situation resources, and through the
detection, repair or exploitation of differences in understanding.
As common sense formulations, designed to accomodate the unforseeable
contingencies of situated action, plans are inherently vague. One
approach to human-machine communication has been to improve on the
vagueness of plans, to make them the basis for artifacts able to
interact with their human users. This talk looks at the problem of
human-machine interaction through a case study of people using a machine
designed on the planning model, and intended to be intelligent and
interactive. A conversation analysis of "interactions" between users
and the machine reveals that the machine's insensitivity to particular
circumstances is both a central design resource, and a fundamental
limitation. More generally, problems in Cognitive Science's theorizing
about purposeful action as a basis for machine intelligence can be
understood as an artifact of the substitution of plans for actions, and
representations of the situation of action for action's actual
circumstances.
-------
∂10-Jun-85 1417 @MIT-MC.ARPA:Dan.Offutt@CMU-CS-A.ARPA Request for addition to list
Received: from MIT-MC.ARPA by SU-AI.ARPA with TCP; 10 Jun 85 14:17:38 PDT
Received: from MIT-OZ by MIT-MC via Chaosnet; 10 JUN 85 17:16:22 EDT
Received: from MIT-MC by MIT-OZ via Chaosnet; 10 Jun 85 17:10-EDT
Received: from CMU-CS-A.ARPA by MIT-MC.ARPA 10 Jun 85 17:11:22 EST
Date: Mon, 10 Jun 85 17:10 EDT
From: Dan.Offutt@CMU-CS-A.ARPA
To: PHILOSOPHY-OF-SCIENCE@MIT-MC.ARPA
Subject: Request for addition to list
CC: Dan.Offutt <offutt@CMU-CS-PS3.ARPA>
Message-Id: <10Jun85.171003.DO00@CMU-CS-A.ARPA>
Please add me to the PHILOSOPHY-OF-SCIENCE mailing list.
Daniel Offutt
∂10-Jun-85 1456 PIERCE@SU-SCORE.ARPA Receptionist welcome
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 10 Jun 85 14:56:24 PDT
Date: Mon 10 Jun 85 14:53:59-PDT
From: Lee Pierce <PIERCE@SU-SCORE.ARPA>
Subject: Receptionist welcome
To: staff@SU-SCORE.ARPA, faculty@SU-SCORE.ARPA
Please join me in welcoming Tina Contreras on her first day as receptionist
for the department. Tina comes to us through the SEE program. Nice to
have you with us, Tina!
---Lee
------
-------
∂10-Jun-85 1727 NILSSON@SU-SCORE.ARPA committee assignments
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 10 Jun 85 17:27:34 PDT
Date: Mon 10 Jun 85 17:25:36-PDT
From: Nils Nilsson <NILSSON@SU-SCORE.ARPA>
Subject: committee assignments
To: faculty@SU-SCORE.ARPA
The next msg will be a long one--describing the committee assignments
for 1985-86. I'll have hard copies of it available at the faculty
meeting tomorrow. -Nils
-------
∂10-Jun-85 1731 NILSSON@SU-SCORE.ARPA committee assignments (long msg)
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 10 Jun 85 17:31:42 PDT
Date: Mon 10 Jun 85 17:27:01-PDT
From: Nils Nilsson <NILSSON@SU-SCORE.ARPA>
Subject: committee assignments (long msg)
To: faculty@SU-SCORE.ARPA
Memo
To: CSD and CSL Faculty
Other interested people
From: Nils Nilsson
Subject: 1985-1986 CSD Committee Assignments
Date: June 10, 1985
First, thanks to everyone who worked so hard for the department during
the academic year just ending. As Gene said in his memo last fall,
committee work is "onerous" but "enriching." When everything turns out
as well as it did this year, we can all feel proud. Several people
thought it would be a good idea to work out before the end of spring
quarter the committee assignments for the next academic year. So, here
we are--just in time.
I am assuming that the student bureaucrats, on receipt of this memo,
will begin thinking about assigning students to committees. (If any
committees decide to get off to a running start by meeting before Fall,
please contact the student bureaucrats so they can assign students early.)
On the last row of the chart at the end of this memo I have indicated
how many students are normally on each committee.
As in previous years, we will ask some of our friends from the local
computer science laboratories to serve on the admissions committee. I
am also hoping that some CSL people currently in E.E. will volunteer for
some committee work. As Gene did last year, I will leave that matter to
discussions between CSL people and the committee chairmen.
I have intentionally tried to have some veterans from last year on each
committee in order to encourage a sense of continuity in the policies
and practices of the committees. Since the committee chairmen of some
of the committees put in a great deal of work, I have tried to arrange
that such chairmen serve on only one committee. Some people want to
have the same committee assignments as last year, and I have tried to
arrange that. Most people will be on at most two committees, although
there are some exceptions--none of them unduly unfair I hope. In two
or three cases there are some additional charges to the committees.
Ph.D. Admissions: Yao (chair), Feigenbaum, Floyd, Papadimitriou, Cheriton,
plus two "volunteers" from local labs.
The committee should have some meetings during the fall quarter to
discuss general policies so that folders that are ranked by different
people are done so reasonably consistently. The committee will be
assisted by Victoria Cheadle. The committee should also obtain from
the entire faculty a sense of how many PhD students should be admitted.
Colloquium: I will volunteer to run the Fall Quarter colloquium--if
helped by Ed Feigenbaum. I would like to coordinate the first few weeks
of the quarter with the departmental seminar. The departmental seminar
will let our students hear about the research interests of our faculty,
and we will try to have the colloquium let people hear about the
research being performed by prominent computer scientists in the Palo
Alto area. I need a volunteer (from CSL) for Winter Quarter. Keith
Lantz is in charge during Spring Quarter.
Comprehensive Exams: Pratt (chair), Golub, Genesereth, Lantz, Wiederhold
The committee should coordinate with the Ph.D. Program committee
(described below) in order to implement any changes in policies for the
comprehensive exams that are recommended and approved. The committee
will serve for the entire year, administering both exams. It will be
aided by Victoria Cheadle.
Curriculum: Ullman (chair), Guibas, Genesereth, Mayr, Reid, Baskett
Stuart Reges is an ex officio member.
This committee runs from April to April and has already started work.
In addition to its usual functions, it will be proposing a curriculum
for an undergraduate major in computer science. We can expect to hear
from it in the fall.
Facilities: Cheriton (chair), McCarthy, Rindfleisch
Fellowships: Tajnai (chair), Nilsson
Advisor for First-Year PhD students: Nilsson (Why not?)
Forum: Miller (chair), Tajnai, Winograd
Visiting Industrial Professors: McCarthy
Library and Publications: Buchanan
MSAI: Buchanan (chair), Feigenbaum, Genesereth, Rosenbloom, Binford, Clancey
This committee is also responsible for monitoring the progress of people
as they progress through the MSAI program. Perhaps they will want to
institute something like a grey Tuesday/black Friday procedure. The
committee will be aided by a staff person to be named.
MS Program: Oliger (chair), Floyd, Guibas, Binford, Wiederhold, Hennessy
This committee is also responsible for monitoring the progress of students.
The committee will be aided by a staff person to be named.
Mathematical and Computational Sciences: Mayr (chair), Golub, Buchanan,
Herriot
Ph.D. Program: Winograd (chair), McCarthy, Pratt, Rosenbloom, Reid
This is a new committee. In the Fall, I would like it to re-evaluate
thoroughly our quals and comprehensive exams policies--taking special
account of any stresses that might arise from the increasing breadth of
computer science. It should also look at the "programming project" to
see if its method of administration or grading needs to be changed in
any way. (Extensive notes and suggestions from people about proposed
changes in the Ph.D. program are available.) Any suggestions for major
changes should be submitted to the entire faculty for approval. The
chair of the committee will supervise the grey Tuesday/Black Friday
process for Ph.D. students. The committee will be a source of
inspiration to students who are having trouble finding a thesis topic
(striking a happy medium between paternalistic hand-holding and callous
neglect). The committee will be assisted by Victoria Cheadle.
Volunteers from CSL are welcome to join the committee.
Ad Hoc Committees (such as search, finance, fund-raising, etc.) will
be commissioned as needed.
I hope all of these chores go as smoothly next year as they have this
year. Let me know if any of you have any major problems with the
suggestions in this memo. A chart summarizing the assignments is
attached.
!
1985-1986 CSD Committee Assignments
Key: x = new member F = fall quarter member
X = member who was on W = winter quarter member
this committee last year S = spring quarter member
c = chairman
C = chairman who was on this committee last year
| A | C | C | C | F | F |1st| F |Ind |Lib|MS| MS |Math|PhD |
| d | o | o | u | a | e | Yr| o |Prof|Pub|AI|Prog|Sci |Prog|
| m | m | l | r | c | l |Adv| r | | | | | | |
| i | p | l | r | i | l | | u | | | | | | |
| s | | o | i | l | o | | m | | | | | | |
| | | q | c | | w | | | | | | | | |
-------------------------------------------------------------------------------
Binford, T. | | | | | | | | | | |X | x | | |
-------------------------------------------------------------------------------
Buchanan, B. | | | | | | | | | | C |C | | x | |
-------------------------------------------------------------------------------
Cheriton, D. | X | | | | C | | | | | | | | | |
-------------------------------------------------------------------------------
Clancey, W. | | | | | | | | | | |X | | | |
-------------------------------------------------------------------------------
Feigenbaum, E. | x | | F | | | | | | | |X | | | |
-------------------------------------------------------------------------------
Floyd, R. | x | | | | | | | | | | | X | | |
-------------------------------------------------------------------------------
Genesereth, M. | | x | | X | | | | | | |X | | | |
-------------------------------------------------------------------------------
Golub, G. | | X | | | | | | | | | | | X | |
-------------------------------------------------------------------------------
Guibas, L. | | | | x | | | | | | | | x | | |
[1/2 LWOS '85-'86] | | | | | | | | | | | | | | |
-------------------------------------------------------------------------------
Hennessy, J. | | | | | | | | | | | | x | | |
-------------------------------------------------------------------------------
Herriot, J. | | | | | | | | | | | | | X | |
-------------------------------------------------------------------------------
Katevenis, M. | | | | | | | | | | | | | | |
[LWOS '85-'86 | | | | | | | | | | | | | | |
'86-'87] | | | | | | | | | | | | | | |
-------------------------------------------------------------------------------
Knuth, D. | | | | | | | | | | | | | | |
[Sabbat. '85-'86] | | | | | | | | | | | | | | |
-------------------------------------------------------------------------------
Lantz, K. | | x | S | | | | | | | | | | | |
-------------------------------------------------------------------------------
Manna, Z. | | | | | | | | | | | | | | |
[LWOS '85-'86] | | | | | | | | | | | | | | |
-------------------------------------------------------------------------------
Mayr, E. | | | | | | | | | | | | | | |
[Sabbat. Fall '85 | | | | x | | | | | | | | | C | |
Spring '86] | | | | | | | | | | | | | | |
-------------------------------------------------------------------------------
McCarthy, J. | | | | | X | | | | C | | | | | x |
-------------------------------------------------------------------------------
McCluskey, C. | | | | | | | | | | | | | | |
-------------------------------------------------------------------------------
Miller, W. | | | | | | | | C | | | | | | |
-------------------------------------------------------------------------------
!
| A | C | C | C | F | F |1st| F |Ind |Lib|MS| MS |Math|PhD |
| d | o | o | u | a | e | Yr| o |Prof|Pub|AI|Prog|Sci |Prog|
| m | m | l | r | c | l |Adv| r | | | | | | |
| i | p | l | r | i | l | | u | | | | | | |
| s | | o | i | l | o | | m | | | | | | |
| | | q | c | | w | | | | | | | | |
-------------------------------------------------------------------------------
Nilsson, N. | | | F | | | x | x | | | | | | | |
-------------------------------------------------------------------------------
Oliger, J. | | | | | | | | | | | | C | | |
-------------------------------------------------------------------------------
Papadimitriou, C. | X | | | | | | | | | | | | | |
[Sabbat. Spr. '86] | | | | | | | | | | | | | | |
-------------------------------------------------------------------------------
Pratt, V. | | C | | | | | | | | | | | | x |
-------------------------------------------------------------------------------
Reges, S. | | | | x | | | | | | | | | | |
-------------------------------------------------------------------------------
Reid, B. | | | | x | | | | | | | | | | |
-------------------------------------------------------------------------------
Rindfleisch, T. | | | | | x | | | | | | | | | |
-------------------------------------------------------------------------------
Rosenbloom, P, | | | | | | | | | | | X| | | x |
-------------------------------------------------------------------------------
Tajnai, C. | | | | | | C | | X | | | | | | |
-------------------------------------------------------------------------------
Ullman, J. | | | | C | | | | | | | | | | |
-------------------------------------------------------------------------------
Wiederhold, G. | | x | | | | | | | | | | X | | |
-------------------------------------------------------------------------------
Winograd, T. | | | | | | | | X | | | | | | c |
-------------------------------------------------------------------------------
Yao, A. | c | | | | | | | | | | | | | |
-------------------------------------------------------------------------------
CSL Volunteers | | | W | | | | | | | | | | | |
-------------------------------------------------------------------------------
Industr. Vols. |x x| | | | | | | | | | | | | |
-------------------------------------------------------------------------------
No. of Students: | 2 | 6 | 1 | 3 | 3 | 1 | | 1 | | 1 | 2| 2 | | 2 |
-------------------------------------------------------------------------------
-------
∂10-Jun-85 1934 @SU-CSLI.ARPA:BrianSmith.pa@Xerox.ARPA No RRR meeting 6/10.
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 10 Jun 85 19:34:14 PDT
Received: from Xerox.ARPA by SU-CSLI.ARPA with TCP; Mon 10 Jun 85 19:30:23-PDT
Received: from Semillon.ms by ArpaGateway.ms ; 10 JUN 85 19:08:04 PDT
Date: 10 Jun 85 19:08 PDT
From: BrianSmith.pa@Xerox.ARPA
Subject: No RRR meeting 6/10.
To: RRR@SU-CSLI.ARPA
cc: BrianSmith.pa@Xerox.ARPA
Message-ID: <850610-190804-1194@Xerox>
As previously announced, there will be no meeting tomorrow, June 10.
Our next (and final for the summer) meeting will be next Tuesday, 6/18,
at 2:15, reading Fodor's "Why there Still has to be a Language of
Thought".
Brian
∂11-Jun-85 0902 HEDGES@SU-SCORE.ARPA Faculty Lunch
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 11 Jun 85 09:02:34 PDT
Date: Tue 11 Jun 85 09:01:49-PDT
From: Karen Hedges <HEDGES@SU-SCORE.ARPA>
Subject: Faculty Lunch
To: Faculty@SU-SCORE.ARPA
cc: Bureaucrats@SU-SCORE.ARPA
REMINDER:
FACULTY LUNCH WILL BE HELD TODAY, JUNE 11 @ 12:15 pm in Room 146.
REMEMBER THIS IS THE LAST LUNCH UNTIL SEPTEMBER!! THERE WILL BE NO
SPECIFIC TOPIC FOR DISCUSSION TODAY -- ONLY SANDWICHES AND RANDOM DISCUSSION!
-------
∂11-Jun-85 0907 HEDGES@SU-SCORE.ARPA GENERAL FACULTY MEETING
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 11 Jun 85 09:07:07 PDT
Date: Tue 11 Jun 85 09:03:47-PDT
From: Karen Hedges <HEDGES@SU-SCORE.ARPA>
Subject: GENERAL FACULTY MEETING
To: FACULTY@SU-SCORE.ARPA
cc: BUREAUCRATS@SU-SCORE.ARPA
THE GENERAL FACULTY MEETING WILL BE HELD TODAY, JUNE 11 @ 3:15 pm in ROOM 146.
-------
∂11-Jun-85 1024 ullman@diablo meeting tomorrow
Received: from SU-AIMVAX.ARPA by SU-AI.ARPA with TCP; 11 Jun 85 10:23:53 PDT
Date: Tue, 11 Jun 85 10:17:29 pdt
From: Jeff Ullman <ullman@diablo>
Subject: meeting tomorrow
To: nail@diablo
I have no program for the meeting 6/12.
Several people are away, but if there are enough people
willing to attend, I'll try to prepare a discussion of
Bancilhon's paper. Send mail to me--not the whole nail list--
if you are planning to attend.
∂11-Jun-85 1413 GOLUB@SU-SCORE.ARPA Math Science Library hours
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 11 Jun 85 14:12:43 PDT
Date: Tue 11 Jun 85 14:11:17-PDT
From: Gene Golub <GOLUB@SU-SCORE.ARPA>
Subject: Math Science Library hours
To: cn.dir@SU-FORSYTHE.ARPA
cc: library@SU-SCORE.ARPA, faculty@SU-SCORE.ARPA
For David Weber
David,
I was very disappointed to see that the math sciences library will be closed
on weekends and at 5 pm during the week. This service is inadequate.
During the summer, many of us are actively engaged in the research that
we have postponed during the academic year. We need to obtain books and
papers at a much higher rate than usual. In the past, we have had keys to
the library but we have not been given keys since the new security system
was installed. Our department reluctantly agreed to the new system only
after we were guaranteed that the library would be open sufficiently long
hours.
Therefore, I insist that the schedule of hours be changed as quickly
as possible to the usual service hours. If this cannot be done, I expect
that keys be distributed to the faculty desiring them.
GENE GOLUB
-------
∂11-Jun-85 1633 WRIGHT@SUMEX-AIM.ARPA Summer Siglunch
Received: from SUMEX-AIM.ARPA by SU-AI.ARPA with TCP; 11 Jun 85 16:33:37 PDT
Date: Tue 11 Jun 85 16:28:17-PDT
From: Carol Wright/Susie Barnes <WRIGHT@SUMEX-AIM.ARPA>
Subject: Summer Siglunch
To: siglunch@SUMEX-AIM.ARPA
There will be no siglunch this week. For the summer, siglunch
will be in sporadic mode. The default will be no siglunch,
and an announcement will be sent out when a siglunch is actually
scheduled.
Thanks, Susie
-------
∂11-Jun-85 1833 L.HANSON@SU-CSLI.ARPA Room to Sublet for Summer
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 11 Jun 85 18:32:51 PDT
Date: Tue 11 Jun 85 17:15:58-PDT
From: Kristin Hanson <L.HANSON@SU-CSLI.ARPA>
Subject: Room to Sublet for Summer
To: folks@SU-CSLI.ARPA
Does anyone know anyone who needs to sublet a room for the summer? I have a large, pretty room (with a fireplace) in a house in Palo Alto (with a garden) that I need to sublet from late June until at least mid-August. The house has a living room, kitchen, washer and dryer, parking, etc., and two other occupants, one law student and one graduate student. It's located on Tennyson, between Waverly and Cowper. The rent is negotiable, as I have only a week and a half to sublet it in before I leave. Anyone interested can call me at 325-8132. --Kristin Hanson
-------
∂12-Jun-85 0011 POLLARD@SU-CSLI.ARPA house needed
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 12 Jun 85 00:10:49 PDT
Date: Wed 12 Jun 85 00:06:36-PDT
From: Carl Pollard <POLLARD@SU-CSLI.ARPA>
Subject: house needed
To: folks@SU-CSLI.ARPA
We are looking for an unfurnished three-bedroom house to rent (for
$1250/month or under) in Palo Alto or Menlo Park that will become
available between now and fall. If you know of such a place, please
let me know. Thanks --
Carl Pollard
856-3326
-------
∂12-Jun-85 0849 DAVIES@SUMEX-AIM.ARPA Architectures Meeting
Received: from SUMEX-AIM.ARPA by SU-AI.ARPA with TCP; 12 Jun 85 08:49:12 PDT
Date: Wed 12 Jun 85 08:48:44-PDT
From: Byron Davies <DAVIES@SUMEX-AIM.ARPA>
Subject: Architectures Meeting
To: AAP@SUMEX-AIM.ARPA
Yes, there is an architectures meeting at 9:30 this morning.
Jim Rice will speak about British blackboards and AI in Europe.
-------
∂12-Jun-85 0947 ullman@diablo meeting
Received: from SU-AIMVAX.ARPA by SU-AI.ARPA with TCP; 12 Jun 85 09:47:02 PDT
Date: Wed, 12 Jun 85 09:39:13 pdt
From: Jeff Ullman <ullman@diablo>
Subject: meeting
To: nail@diablo
We'll have an informal meeting on system design.
Allen has agreed to lead a discussion on formalizing the
source language.
I also attach a message from Mike Bender regarding ICODE.
I think it is a well-thought out proposal.
Let me say in advance of your reading it that with regard to
Mike's section II, I never intended that tuple-at-a-time
operations were applied to tuples, but rather to relations.
e.g., selection and projection are TAT operations of relational
algebra. Of course, i suspect that some capture rules will
use relations that the author knows are single tuple relations,
or even unary single-tuple relations (constants, to you).
****************************************************
SOME I-CODE SUGGESTIONS
=======================
Actually, the suggestions I have to make about the i-code are
not that dramatic. What I have tried to do is push Jeff's suggestions
in the direction of greater generality, without (I hope) making it too
difficult to implement.
In the first section, I propose different tuple-at-a-time
operations. Following that I raise the question about whether it
wouldn't be better to replace these operations with binary
(relational) operations. In the third section I propose a new
binary operation to capture differences between relations. In
the fourth section I propose a stack system based on indexes.
Finally, I raise some questions concerning how and when operators can
modify pre-existing objects.
Although there are a number of references here to Jeff's original
document I have tried to explain them briefly, so that if you don't
have Jeff's document you can still follow along. However, it would
probably be easier to understand if you have Jeff's paper.
I. TUPLE-AT-A-TIME OPERATIONS:
The original proposal had a Project, a Select, an Assign, and
three Update operations. I suggest that the Assign and the three
update commands be replaced by the following two operations:
1) An assign command:
A := ASSIGN (<T1,...,Tn>) B
where Ti is a term used to specify/generate the i-th component of
A (so A has n components).
Ti is either a constant or a non-recursive function which
accepts the components of B as arguments.
2) Although it is just a special case of the above, I would suggest
a special operator to create tuples from constants, without
using the Update, Select, or Project operations. Ie:
A := CREATE (c1, ... ,cn)
where ci is a constant and will be the i-th component of A.
II. COMMENT:
Is it really appropriate to have TUPLE-AT-A-TIME operations?
What would be the disadvantage in using relational operations? I
understand that eventually the i-code will have to be converted into
relational operators, so wouldn't it be easier to use relational
operators in the i-code too?
Also, it would probably be "cleaner". (We know that the relational
calculus, which only deals with relations, is theoretically sufficient.)
For instance, how do stack operators work when dealing with multiple
types of objects?
- What gets saved on the Stack? Relations? Tuples? Attributes?
- Since a relation is merely an aggregate of tuples, there is a lot
of room for problems if we start saving different types of objects
on the stack. Eg if we save the relation and then we modify a tuple
in it - what happens?
I realize that I am trying to push us back into the lap of relational
algebra, however it seems to me that Jeff's original proposal was
essentially an extension to relational algebra (ie features had been
added to support increased control/performance).
III. BINARY OPERATIONS:
Clearly the initial set of binary operators could be expanded in
many different ways. One potentially valuable extension is the
MERGE operator (ala Logix):
A := MERGE B,C
Merge works like a Union, except that it creates an additional
attribute that specifies the source of the tuple, eg
0 would indicate that the tuple appeared in both B and C
1 would indicate that it appeared in only the first arg (B)
2 would indicate that it appeared in only the second arg (C)
Clearly this is an extension of the "+=" operator, since CHANGE
could be set whenever no tuple had a "2" in the source attribute.
This operator is also a generalization of a "difference" operator,
because it can be used to compare two relations, to see how they
differ.
This implies that we would also need the capability of directly
setting the CHANGE flag, should we desire to use it still. Ie, we
should have the capability to do:
SET CHANGE
RESET CHANGE
or something similar.
IV. INDEX OPERATIONS:
Although a stack offers the most primitive level of operations, it
does not offer as much strength as we might like. I propose,
instead, that the icode support indexing for every object.
The way indexing would work, it would be possible to refer to any
variable with an index, eg "A[5]". By definition:
"A" is equivalent to "A[0]"
"A[0]" always refers to the LATEST (ie the most current)
version of A. "A[i]" always refers to the version of A
that followed "A[i-1]".
1) We would still need the PUSH operation to save versions of an
object:
PUSH X
PUSH would create a new 0th version of X. The previous i-th
version of X would now become the (i+1)-th version.
2) We would also need a modified POP operation to access versions
on the stack:
Y := POP (I, X)
POP would return the I-th version of X. (POP X would be shorthand
for POP (1,X).)
(Probably the best way to implement indexing would be with three
stacks: one for relations, one for tuples, and one for attribute values.)
FORGET and REMEMBER operations are needed for performance reasons:
3) The FORGET operation would specify when old versions of a relation
should be removed.
FORGET (I, X)
This operation would remove every version of an object except for
the latest I versions.
4) The REMEMBER operation would specify how many versions of an object
are to be kept on the stack. (No versions are ever saved without
a PUSH operation).
REMEMBER (I, X)
This operation would specify that the icode will automatically
save on the stack the latest I versions of an object X. If more
than I versions are created, the oldest ones will be removed from
the stack automatically.
V. OTHER ISSUES:
The other issue is really a question: Can any of the operators
be used to modify an existing object (eg relation, tuple, attribute)?
And if so, which ones?
For example: the expression
A := UNION A, B
probably should be legal.
However, the expression
A := POP
might cause problems if the object A already exists. For example,
what if A is a pre-existing relation and POP returns a tuple or an
attribute value? Or a relation with a different schema? The problems
occur because the way the i-code is designed ANYTHING can be put on
the stack.
The indexed version of POP would partially get around this problem.
Ie, the statement
A := POP (I, A)
would be legal but the statement
A := POP (I, B)
would not be legal unless A and B had the same type (however that is
defined).
This would not solve the problem entirely, though. For instance,
what if T refers to a tuple in relation R. Then the following
commands:
PUSH (R)
ASSIGN (...) T
POP (T)
POP (R, 1)
would rewrite over our modifications to T. Was this our intention?
There is no way to save the semantic link that exists between T and
R. (Ie if T is modified then R should be modified too).
I believe that if we were to eliminate the tuple-at-a-time
operations and on convert them to relational operations, we would
remove some of the "philosophical" problems involved.
(The way I wrote the wording of the ASSIGN operator above it will
permit the expression:
A := ASSIGN (...) A.
If it was decided not to permit this, the wording would have to be
modified.)
∂12-Jun-85 1152 JODY@SU-CSLI.ARPA Burrito Bandito
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 12 Jun 85 11:52:18 PDT
Date: Wed 12 Jun 85 11:31:59-PDT
From: Joe Zingheim <JODY@SU-CSLI.ARPA>
Subject: Burrito Bandito
To: folks@SU-CSLI.ARPA
cc: consultants@SU-CSLI.ARPA
Amigos of the Burrito Bandito will be saddened to hear that their valient
mucacho was captured by the Border Patrol, and may be deported to his
strife-torn homeland. Sources say that the ever-vigilent Border Patrol
may have acted on a tip from Weight Watchers. Appeal for mercy may be sent
to the Border Patrol, in care of Clay@CSLI.
-------
∂12-Jun-85 1349 NILSSON@SU-SCORE.ARPA Wegner Visit
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 12 Jun 85 13:48:53 PDT
Date: Wed 12 Jun 85 13:47:22-PDT
From: Nils Nilsson <NILSSON@SU-SCORE.ARPA>
Subject: Wegner Visit
To: faculty@SU-SCORE.ARPA
Peter Wegner of Brown Univ. will be in the Stanford area between July 8-20.
He has proposed to give a talk here if there is interest. I attach his
proposed abstract. I'll be in and out a lot during that period, so if
anyone is interested in having him give the talk and wants to organize it,
please let me know and I'll have Peter get in touch with you. -Nils
Language Paradigms for Programming in the Large
Peter Wegner, Brown University
Abstract:
We evaluate and compare three programming language paradigms - the
block structure, object-oriented, and distributed programming paradigms -
by examining the nature of their bricks and their mortar.
(1) block-structure paradigm
Algol, Pascal, Pl/I, Ada, Modula
(2) object-oriented programming paradigm
Simula, Smalltalk, Object Pascal, object-oriented database languages
(3) distributed programming paradigm
Argus, NIL-like languages
Paradigms are described by Thomas Kuhn as patterns of thought that serve
as a model for research among a community of researchers. Each of the
programming language paradigms determines a pattern of thought for both
designers and users of programming languages and therefore qualifies as a
paradigm. By focussing attention on the paradigm rather than on the details
of specific languages within the paradigm, we can explore fundamental concepts
common to classes of related programming languages at an appropriate level of
abstraction.
Each paradigm is defined in terms of a necessary and sufficient set of
paradigmatic requirements that may be used to determine if a given language
is an instance of the paradigm. The block structure paradigm is characterized
in terms of visibility and lifetime conditions. The object-oriented paradigm
is characterized in terms of data abstraction, object types, and inheritance.
The distributed paradigm is characterized in terms of message passing,
concurrency, and access control.
The object-oriented and distributed paradigms complement each other.
Several "object-oriented distributed paradigms" are derived by combining
features of object-oriented and distributed paradigms, using an inheritance
hierarchy for language types that parallels inheritance hierarchies in object-
oriented programming. The requirements for a unified "object-oriented
distributed programming language" are examined.
-------
∂12-Jun-85 1736 EMMA@SU-CSLI.ARPA Newsletter June 13, No. 33
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 12 Jun 85 17:36:07 PDT
Date: Wed 12 Jun 85 16:53:43-PDT
From: Emma Pease <Emma@SU-CSLI.ARPA>
Subject: Newsletter June 13, No. 33
To: friends@SU-CSLI.ARPA
Tel: 497-3479
C S L I N E W S L E T T E R
←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←
June 13, 1985 Stanford Vol. 2, No. 33
←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←
A weekly publication of The Center for the Study of Language and
Information, Ventura Hall, Stanford University, Stanford, CA 94305
←←←←←←←←←←←←
CSLI ACTIVITIES FOR *THIS* THURSDAY, June 13, 1985
12 noon TINLunch
Ventura Hall ``Type Raising, Functional Composition,
Conference Room and Non-Constituent Conjunction''
David Dowty, Center for the Advanced Study of
the Behavioral Sciences
2:15 p.m. CSLI Seminar
Redwood Hall No seminar
Room G-19
3:30 p.m. Tea
Ventura Hall
4:15 p.m. CSLI Colloquium
Redwood Hall No colloquium
Room G-19
←←←←←←←←←←←
CSLI ACTIVITIES FOR *NEXT* THURSDAY, June 20, 1985
12 noon TINLunch
Ventura Hall ``How Many Possible Human Languages are There?''
Conference Room by Geoff Pullum, UCSC and CSLI
Discussion led by Gerald Gazdar, CASBS
(Abstract on page 2)
2:15 p.m. CSLI Seminar
Redwood Hall No seminar
Room G-19
3:30 p.m. Tea
Ventura Hall
4:15 p.m. CSLI Colloquium
Redwood Hall No colloquium
Room G-19
←←←←←←←←←←←←←
ANNOUNCEMENT
No seminars or colloquiums are scheduled for June 13 or June 20
because of the University, end-of-quarter break. TINLunch will be
held on these days. Regular activities will resume on June 27.
!
Page 2 CSLI Newsletter June 13, 1985
←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←
ABSTRACT OF NEXT WEEK'S TINLUNCH
``How Many Possible Human Languages are There?''
Linguistic Inquiry 14, 447-467.
Beginning with Chomsky (1980), a number of linguists have claimed
(i) that their favored grammatical framework only allows for the
existence of a finite number of grammars,
(ii) that there only exist a finite number of possible human
languages.
Pullum's paper, published two years ago, argues that claim (i) is
false in every relevant case, and that claim (ii) is uninteresting,
even if true. His paper has been greeted by a deafening silence. How
can this be? Are his arguments so obviously invalid that it would be
cruel and undignified to reply to them? In which case, how did they
get into print? Or are they so obviously valid that those attacked
are ashamed even to allude to the matter? In which case, what are we
to make of their intellectual integrity? --Gerald Gazdar
---------
RRR MEETINGS
The RRR meeting of May 28 was on representationalism and was led by
Ned Block. He mentioned a number of distinctions among representationa-
list views, most importantly the by now familiar (in RRR) distinction
between (1) thoughts have representational parts and (2) thoughts are
sentential. He observed that the arguments usually given for the
latter better support the former, and claimed that evidence for the
latter view is not statable succinctly, but rather is suggested by a
mass of experiments, no small group of which is very convincing.
At the meeting of June 4, Kurt van Lehn gave an exposition and
critique of ``connectionism''. The view's opponents had different
complaints. Van Lehn argued that the successes of the connectionist
movement have had little to do with its controversial architectural
proposals. Jerry Fodor faulted connectionist models for not handling
the kind of thinking involving reasoning from one step to another,
while Brian Smith pointed out that connectionism was committed to an
overly ``iconic'' conception of representation. Pentti Kanerva said
that he was working on a connectionist theory that might be able to
overcome the faults that had been pointed out.
Upcoming meetings: There was NO meeting on June 11 (because of
conflict with the Workshop on Language Processing. The next RRR
meeting will be on June 18; we will discuss Jerry Fodor's ``Why there
STILL has to be a Language of Thought''. Unless there is a change of
plans, the June 18th meeting will be the last for the spring quarter.
We won't meet over the summer; what will happen next year is far from
clear. --Ned Block
---------
CSLI REPORT
Report No. CSLI-85-25, ``An Internal Semantics for Modal Logic:
Preliminary Report'' by Ronald Fagin and Moshe Vardi, has just been
published. This report may be obtained by writing to David Brown,
CSLI, Ventura Hall, Stanford, CA 94305 or Brown@SU-CSLI.
-------
∂12-Jun-85 2121 SNYDER%upenn.csnet@csnet-relay.ARPA please remove SNYDER@UPENN from nail mailing list. thank you.
Received: from SU-AIMVAX.ARPA by SU-AI.ARPA with TCP; 12 Jun 85 21:21:37 PDT
Received: from csnet-relay by diablo with TCP; Wed, 12 Jun 85 21:17:52 pdt
Received: from upenn by csnet-relay.csnet id al00406; 13 Jun 85 0:11 EDT
From: Wayne Snyder <Snyder%upenn.csnet@csnet-relay.ARPA>
Subject: please remove SNYDER@UPENN from nail mailing list. thank you.
To: nail@su-aimvax.ARPA
Date: Wed, 12 Jun 85 12:24 EDT
∂12-Jun-85 2249 ullman@diablo Re: please remove SNYDER@UPENN from nail mailing list. thank you.
Received: from SU-AIMVAX.ARPA by SU-AI.ARPA with TCP; 12 Jun 85 22:49:25 PDT
Date: Wed, 12 Jun 85 22:45:58 pdt
From: Jeff Ullman <ullman@diablo>
Subject: Re: please remove SNYDER@UPENN from nail mailing list. thank you.
To: Snyder%upenn.csnet@csnet-relay.ARPA, nail@diablo
gladly. Who are you, and how did you get on in the first place.
∂13-Jun-85 1131 EMMA@SU-CSLI.ARPA TINLUNCH
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 13 Jun 85 11:31:11 PDT
Return-Path: <HANS@SRI-AI.ARPA>
Received: from SRI-AI.ARPA by SU-CSLI.ARPA with TCP; Thu 13 Jun 85 10:17:33-PDT
Date: Thu 13 Jun 85 10:19:16-PDT
From: Hans Uszkoreit <Hans@SRI-AI.ARPA>
Subject: TINLUNCH
To: friends@SU-CSLI.ARPA
ReSent-Date: Thu 13 Jun 85 11:21:05-PDT
ReSent-From: Emma Pease <Emma@SU-CSLI.ARPA>
ReSent-To: friends@SU-CSLI.ARPA
I would like to remind you of today's TINLUNCH. Since TINLUNCH is a
research-oriented discussion series, it will not be affected by breaks
between quarters, dead weeks, and other academic regeneration periods.
Today, David Dowty (CASBS and Ohio State U.) will discuss his recent
paper:
"Type Raising, Functional Composition,
and Non-Constituent Conjunction"
Next Thursday, a paper by Geoff Pullum (UCSC and CSLI) will be
discussed. Title:
"How Many Possible Human Languages are There?"
Gerald Gazdar (CASBS and Univ. of Sussex) will lead the discussion.
For abstracts of the papers, please refer to last week's and this week's
CSLI newsletters.
Finally, let me also remind you of the excellent CSLI lunch service. If
you want to save a trip to the store (or just be on time for TINLUNCH),
send your sandwich order to LUNCH@SU-CSLI.
Hans Uszkoreit
-------
∂13-Jun-85 1235 YAMARONE@SU-CSLI.ARPA DEATH OF EL BURRITO BANDITO
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 13 Jun 85 12:35:50 PDT
Date: Thu 13 Jun 85 12:24:41-PDT
From: Tom Yamarone <YAMARONE@SU-CSLI.ARPA>
Subject: DEATH OF EL BURRITO BANDITO
To: folks@SU-CSLI.ARPA
cc: CONSULTANT@SU-CSLI.ARPA
Folks, Consultants,Friends,
I have some bad news and some good news.
First, the bad news. The Burrito Bandit is Dead.(sigh)
No more burritos on Thursday. It seems he was ambushed deep in
Redwood City last week (where he purchases the burritos)and
suffered a great loss(nearly 20 pounds)
Now, the good news. The Ventura Sandwich Corp. is alive and well!(Yeah!)
But those of you who consistently await "extras" are very often disappointed.
(sigh) So, may I reiterate that you order whenever you think you "might"
want a sandwich to LUNCH on the mail system. And , more importantly, order
BEFORE 10:00 AM that day. IF this is too early for you late night hackers and
late morning risers, then may I suggest ordering the night before when you
are up and buzzing along.
Sandwiches come in all deli varieties and cost 2.50 .(roast beef and
special orders such as "club sandwiches" are 3.00)
I hope this clears up some of the confusion surrounding our lunch service
lately. So , happy ordering and happy eating....The Ventura Sandwich Corp.
PS: Don't mail order to my account and don't count on extras(especially THURS.)
-------
∂13-Jun-85 1341 INGRID@SU-CSLI.ARPA Reading for Tuesday
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 13 Jun 85 13:40:48 PDT
Date: Thu 13 Jun 85 13:34:47-PDT
From: Ingrid Deiwiks - 497-3084 <INGRID@SU-CSLI.ARPA>
Subject: Reading for Tuesday
To: RRR@SU-CSLI.ARPA
The reading for the next RRR meeting on Tuesday, Jerry Fodor's
"Why There Still Has To Be A Language Of Thought", will be available
from Susi by this afternoon.
Ingrid
-------
∂13-Jun-85 1732 LIBRARY@SU-SCORE.ARPA Do you use or need machine readable data from the Stanford Academic Data Service?
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 13 Jun 85 17:32:03 PDT
Date: Thu 13 Jun 85 17:23:57-PDT
From: C.S./Math Library <LIBRARY@SU-SCORE.ARPA>
Subject: Do you use or need machine readable data from the Stanford Academic Data Service?
To: su-bboards@SU-SCORE.ARPA
cc: faculty@SU-SCORE.ARPA,
: ;
The Stanford Academic Data Service has been supplying machine readable data,
such as that from ICPSR, the census, etc. to the Stanford community for over
ten years. The users of this service have rarely had the opportunity to
comment directly on the operation of this program. ACIS/IRIS, ITS and the
Libraries are currently analyzing the way faculty and graduate students use
machine readable data in research, and we are considering a refinement or
reorganization of the service. We are very interested in obtaining your
opinions regarding the present program and how it might change to meet
your future needs.
We are presently distributing a questionnaire which addresses many of the
issues which are being considered in this process. If you would be
willing to fill out a 2-page questionnaire they are available at
General Reference in the Green Library or by contacting Steve MacLeod
in General Reference in the Green Library (497-1811) Any input
or help you are willing to give will be appreciated.
-------
∂13-Jun-85 1737 ullman@diablo ICODE
Received: from SU-AIMVAX.ARPA by SU-AI.ARPA with TCP; 13 Jun 85 17:35:27 PDT
Date: Thu, 13 Jun 85 17:29:43 pdt
From: Jeff Ullman <ullman@diablo>
Subject: ICODE
To: nail@diablo
An informal discussion of Mike's ICODE proposal by a skeleton
crew consisting of Kuper, Van Gelder, Winslett, and myself
yielded the following proposal, which is a blend of my old
proposal, Mike's and some new issues that had not surfaced
previously. Comments welcome.
*********************************************************
I. ICODE RUN-TIME ENVIRONMENT
Adopt Mike's approach. All variables are implicitly lists of
relations, with A[i] denoting the i-th "version" of relation A.
A[0] is the "most recent" version, and A is shorthand for A[0].
Sometimes we shall want variables that are single tuples or
even constants, i.e., single unary tuples. Fine, just make sure
you assign values of the proper type to the variable in question.
We have two viewpoints regarding the implementation of relations.
First, relations may be small, and they will be represented by linked
lists (say). I.e., a variable is a list of relation "versions,"
a relation is a list of tuples; a tuple is a list of components;
a component is a term, probably represented as a tree structure.
Second, relations may be big, and we have to call upon the underlying
database system to store them. A variable may now be a list of versions,
but each version is an (internal) name of a relation stored by the
DBMS.
We have to design the ICODE so a reasonable decision about which
representation to use can be made. Thus, we'll introduce optional
declarations that are essential if we need to invoke the DBMS.
We'll keep the global CHANGE to detect when an incrementation of
a relation actually adds something, as in my original proposal.
II. OPERATIONS
1. A := (T1,...,Tn) B
This is our only tuple-at-a-time operation, following Mike's suggestion.
Ti is a term that constructs the i-th component of any tuple of A
from the corresponding tuple of B. Operand $j appearing in Ti
represents the j-th component of the tuple.
2. A := CREATE(c1,...,cn)
Creates a relation consisting of one tuple with the constants c1,...,cn
as components. A := c is shorthand for A := CREATE(c).
The global variable CHANGE can be set and reset with this statement,
e.g., CHANGE := 1.
3. A := UNION(B,C)
Obvious meaning.
4. A := JOIN ($i1=$j1,...,$in=$jn) B,C
Takes the equijoin with equality on each of the indicated pair of
components, the i←k-th from B versus the j←k-th from C.
5. A += B
Does A := UNION(A,B), setting CHANGE to 1 if A changes.
I still prefer this to Mike's merge, because while the latter
is clearly more general, my += offers the opportunity of efficient
implementation, while the latter requires A be copied at each step.
6. A := <arith. op>(B,C)
This family of operations almost requires that A, B, and C be
scalars, i.e., single-tuple, unary relations. However, we can
define A := +(B,C) to work on more general B and C, ala APL
if we like (yucch!)
7. A := ANY(B)
Makes A equal to {t}, where t is any tuple in B, and deletes t from B.
This is a new thought. We can use it, for example, when we have a
set S of constants, and we want to evaluate a goal with the first (e.g.)
argument bound to each of the constants in S, in turn, then take the union.
Allen suggests a "for t in B {...}" construct instead,
but I'm nervious, because it means the run-time state needs more than
a "current instruction" counter; it needs a stack of markers to
keep track of where you are.
III. CONTROL FLOW
8. GOTO <label>
Sorry folks
9. IF <condition> GOTO <label>
The conditions I have in mind are:
a) EMPTY(R)
b) A <relop> B
Here <relop> could be any arithmetic operation, or set
membership, its negation, set containment, etc., with the
assumption that A and B have appropriate types in each case.
Extended operators, again like APL, are a possibility, so
we might use A < B, for example to mean that for every tuple
t in A and s in B, each component of t is < the corresponding
component of s. Do we really care about this?
c) EMPTYSTACK(R)
Not to be confused with (a), this tests whether there
are any versions of R in existence.
d) others??
10. The use of for-, while- or other structured control statements
is up for debate. See my remarks under statement (7).
IV. VERSION CONTROL
One convenience is to allow A[MAX] to refer to the highest-indexed
version of A in existence. This would be handy if A were used as
a queue, e.g., for Allen's message-passing algorithm, although we
could always keep a separate variable that counted the number of
versions A currently had.
11. PUSH(X)
X[0] becomes the empty relation, X[1] becomes the old X[0],
X[2] becomes the old X[1], etc.
12. POP(X)
X[0] := X[1]; X[1] := X[2];...
13. A := B[i]
Sets A[0] to be a copy of the i-th version of B.
This corresponds to Mike's POP, although here, no data is destroyed
(it wasn't clear whether POP destroys or not).
14. FORGET(i, X)
Returns to storage X[i], X[i+1],...
V. DECLARATIONS
15. ARITY(R) = n
Declares R to have n attributes. I'm assuming for the moment that
all attributes are of the type variable-length character string.
If, for example, a particular attribute turns out to be a list
of integers, then *should it be stored by the DBMS*,
the integers would each be converted to character strings (of digits),
and characters indicating the "cons" function would be inserted
appropriately.
This matter is under debate.
16. INDEX(R) or INDEX(R; i1,...,in)
The first says create an index for R on complete tuples (if the DBMS
maintains R). We would like this, for example, if R were a huge
relation and we did R += A. The second form calls for an index
on attributes i1,...,in of R, again if R is maintained by the DBMS.
This statement might be important for efficiency, if we new we
were taking selections or joins involving R.
17. REMEMBER(i, X)
Maintain only X[0],...,X[i-1], i.e., automatically execute FORGET(i,X)
whenever X[i] is created by a PUSH.
∂13-Jun-85 2006 avg@diablo ICODE ideas and comments
Received: from SU-AIMVAX.ARPA by SU-AI.ARPA with TCP; 13 Jun 85 20:06:30 PDT
Date: Thu, 13 Jun 85 20:00:18 pdt
From: Allen VanGelder <avg@diablo>
Subject: ICODE ideas and comments
To: nail@diablo
I believe the ICODE should be much closer to logic for two reasons.
1. The underlying program is in logic.
2. We are going to implement in Prolog.
Here are some ideas on revising the Icode in that direction.
This is some of Jeff's text, with my comments indented.
I. ICODE RUN-TIME ENVIRONMENT
Adopt Mike's approach. All variables are implicitly lists of
relations, with A[i] denoting the i-th "version" of relation A.
A[0] is the "most recent" version, and A is shorthand for A[0].
Sometimes we shall want variables that are single tuples or
even constants, i.e., single unary tuples. Fine, just make sure
you assign values of the proper type to the variable in question.
Allow various types of variables, letting context determine
the type.
Maintain a relation version(A, A[i]) where A[i] is an arbitrary
naming sceme; could be p0001, p0002, or anything.
Let the relation have an order on the tuples. Do this
only if you WANT versions. Instead of shifting version numbers
all the time, let the highest number be the most recent, and
remember it.
Maybe we want version(A, A[i], A[i-1]) so we can chain to previous
versions without relying on the tuples being in any order. Two-way
chains are also easy.
We'll keep the global CHANGE to detect when an incrementation of
a relation actually adds something, as in my original proposal.
I expect we'll need CHANGE for each relation, so let's think
in terms of a relation changed(A, V) where V is 0 or 1.
II. OPERATIONS
1. A := (T1,...,Tn) B [was A := ASSIGN (T1,...,Tn) B in the notes]
A(T1,...,Tn) :- B
2. A := CREATE(c1,...,cn)
A(c1,...,cn)
3. A := UNION(B,C)
A :- B; C
4. A := JOIN ($i1=$j1,...,$in=$jn) B,C
A(X1,...,Xr) :- B(Xi1,...Xin), C(Xj1,...,Xjm)
5. A += B
:- abolish(changed(A))
A(X1,...) :- B(X1,...), \+ A(X1,...), mark(A); A(X1,...).
mark(A) :- changed(A); \+ changed(A), assert(changed(A)).
or
:- asserta(changed(A, 0))
A(X1,...) :- B(X1,...), \+ A(X1,...), asserta(changed(A,1)); A(X1,...)
This is the hackiest, because I am not assuming you want a new
version of A. Note that I use ":-" as a prefix for imperative
commands with side-effects, that do not operate on our relational
variables.
6. A := <arith. op>(B,C)
... :- ..., A is <arith. op>(B,C), ...
7. A := ANY(B)
... :- ..., B(A), ...
III. CONTROL FLOW
8. GOTO <label>
Sorry folks
Don't apologize; just scrap it. Let's assume that WHILE is
sufficient for our iteration needs until proven otherwise.
9. IF <condition> GOTO <label>
... :- <condition>, <true-action>; \+ <condition>, <false-action>
or use what Prolog gives us:
<condition> -> <true-action>; <false-action>
The conditions I have in mind are:
a) EMPTY(R)
\+ R(X)
b) A <relop> B
c) EMPTYSTACK(R)
Not to be confused with (a), this tests whether there
are any versions of R in existence.
\+ version(R, ←) or \+ version(R, ←, ←)
d) others??
10. The use of for-, while- or other structured control statements
is up for debate. See my remarks under statement (7).
IV. VERSION CONTROL
11. PUSH(X)
:- asserta(version(X, X[i]))
or
:- version(X, CurrX, ←), asserta(X, X[i], CurrX)
Recall that I am assuming X[i] represents any naming scheme.
12. POP(X)
X[0] := X[1]; X[1] := X[2];...
:- retract(X, ←), version(X, PrevX)
or
:- retract(X, ←, PrevX)
13. A := B[i]
Sets A[0] to be a copy of the i-th version of B.
This corresponds to Mike's POP, although here, no data is destroyed
(it wasn't clear whether POP destroys or not).
:- version(B, B[i]), asserta(version(A, B[i]))
or a fancier one to maintain the chain on A when "version" has
3 arguments.
14. FORGET(i, X)
Returns to storage X[i], X[i+1],...
:- while version(X, ←), retract(version(X, Xver)), abolish(Xver)
to get rid of X completely.
Skip the "while" to get rid of a single version.
A fancier loop gets rid of versions "in a range."
V. DECLARATIONS
15. ARITY(R) = n
:- arity(R, n)
16. INDEX(R) or INDEX(R; i1,...,in)
:- index(R, [i1, ...])
17. REMEMBER(i, X)
:- remember(X, i)
∂13-Jun-85 2105 @SU-CSLI.ARPA:halvorsen.pa@Xerox.ARPA Discussion of Higginbotham's "Definiteness and Predication"
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 13 Jun 85 21:05:37 PDT
Received: from Xerox.ARPA by SU-CSLI.ARPA with TCP; Thu 13 Jun 85 20:58:30-PDT
Received: from Semillon.ms by ArpaGateway.ms ; 13 JUN 85 21:00:25 PDT
Date: 13 Jun 85 21:00 PDT
From: halvorsen.pa@Xerox.ARPA
Subject: Discussion of Higginbotham's "Definiteness and Predication"
To: NL1@Su-CSLI.ARPA
cc: halvorsen.pa@Xerox.ARPA,NLInterest@Su-CSLI.ARPA
Message-ID: <850613-210025-1872@Xerox>
Higginbotham's "Definiteness and Predication" is the topic of
discussion at our meeting Friday. Susi has a master copy if you don't
already have the paper.
Time and Place: 2:15 Ventura Conference Room
∂14-Jun-85 1214 BRODER@SU-SCORE.ARPA Goodbye
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 14 Jun 85 12:14:32 PDT
Date: Fri 14 Jun 85 12:13:48-PDT
From: Andrei Broder <Broder@SU-SCORE.ARPA>
Subject: Goodbye
To: aflb.all@SU-SCORE.ARPA
Friends:
I had a great time coordinating AFLB for the last (almost) three years,
but everything must come to an end.
Oren Patashnik (patashnik@su-sushi.arpa) agreed to take over. I am
sure that he will do a great job.
Many thanks to all of you that, by speaking or listening, helped me
along the way.
Regards,
Andrei
P.S. All the mailing adresses (aflb.all@score, aflb.local etc.) will
stay the same, but all your inquiries about mailing lists, speakers,
schedule, and so on, should be sent to Oren.
-------
∂14-Jun-85 1456 BARWISE@SU-CSLI.ARPA Party Reminder
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 14 Jun 85 14:56:40 PDT
Date: Fri 14 Jun 85 14:50:52-PDT
From: Jon Barwise <BARWISE@SU-CSLI.ARPA>
Subject: Party Reminder
To: Folks@SU-CSLI.ARPA
Old habits die hard. I cannot resist reminding you of the party
tonight, even though circumstances prevent my being there. But I will
be thinking of you all eating and dancing away the night. Have fun.
Jon
-------
∂14-Jun-85 1526 DAVIES@SUMEX-AIM.ARPA Symbolics-to-Explorer porting document
Received: from SUMEX-AIM.ARPA by SU-AI.ARPA with TCP; 14 Jun 85 15:26:24 PDT
Date: Fri, 14 Jun 1985 15:25 PDT
Message-ID: <DAVIES.12119162149.BABYL@Sumex>
From: DAVIES@Sumex
To: AAP@Sumex
Subject: Symbolics-to-Explorer porting document
cc: Davies@Sumex
Copies of "Porting Software from Symbolics 36XX to the Texas
Instruments Explorer" are available on my keyboard table in 1110.
Please take one.
-- Byron
∂14-Jun-85 1530 MARJORIE@SU-CSLI.ARPA Key for bicycle
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 14 Jun 85 15:30:16 PDT
Date: Fri 14 Jun 85 15:24:27-PDT
From: Marjorie Maxwell <MARJORIE@SU-CSLI.ARPA>
Subject: Key for bicycle
To: folks@SU-CSLI.ARPA
cc: consultants@SU-CSLI.ARPA
There was a bicycle lock key found on the deck. It can be picked up in my
office.
Marjorie (H-5 -trailers)
-------
∂14-Jun-85 1850 YM@SU-AI.ARPA New book
Received: from SU-AIMVAX.ARPA by SU-AI.ARPA with TCP; 14 Jun 85 18:49:49 PDT
Received: from SU-AI.ARPA by diablo with TCP; Fri, 14 Jun 85 18:47:14 pdt
Date: 14 Jun 85 1845 PDT
From: Yoni Malachi <YM@SU-AI.ARPA>
Subject: New book
To: nail@SU-AIMVAX.ARPA
LOGIC PROGRAMMING: RELATIONS, FUNCTIONS, AND EQUATIONS
Editors:
Doug DeGroot, IBM Yorktown
Gary Lindstrom, University of Utah
Prentice-Hall, Inc.
Publication date: Summer 1985
Announcement with table of contents is in LOGPRO.TOC[1,YM].
You can FTP it without login.
∂17-Jun-85 0943 mhb@FORD-WDL1 more ICODE comments
Received: from SU-AIMVAX.ARPA by SU-AI.ARPA with TCP; 17 Jun 85 09:41:41 PDT
Received: from FORD-WDL1 by diablo with TCP; Mon, 17 Jun 85 09:32:03 pdt
Return-Path: <>
Date: 14-Jun-85 16:10:15-PDT
From: mhb@FORD-WDL1.ARPA
Subject: more ICODE comments
To: nail@SU-AIMVAX.ARPA
SOME MORE COMMENTS REGARDING ICODE
==================================
Since I got Jeff's and Allen's comments together, I'm sending out some
collective comments. The first part deals with Jeff's proposal, and
the second part deals with Allen's.
Regarding Jeff's latest proposals:
==================================
INDEXES:
Reading Jeff's latest ICODE proposal, the following questions come up
regarding the INDEX declaration:
1) Can a relation have more than one index? If so, does this mean
that indexes have to be named?
2) If we perform a join between two relations, how is the join order
resolved? Clearly it matters which one of the two has an index on
it, but is the ICODE supposed to make this decision implicitly or
should we do it explicitly? (I contend that at the ICODE level we
should be making the decision to use an INDEX explicitly, so we need
the operators to make this specification).
3) Doesn't the term INDEX imply some kind of implementation?
In fact, aren't we really interested in ORDER (a logical concept)
instead of INDEX (a physical concept). I know that this may sound
like a moot point, but I think it is very important that the ICODE not
be bound to any implementation, ie that it preserve its logical
independence.
4) If INDEXES mean what I assume them to mean (eg B-Trees, etc.), then
what happens to relations with INDEXES that get saved on the stack?
Do the Indexes get saved too? Would we want to save the indexes,
assuming that we could? If we don't save them, does this mean that a
relation's declaration changes when it gets put on the stack?
Having raised these questions, I would like to put a possible solution
up for discussion:
a) The ability to access a relation in a given order should not be
a *declaration* in the ICODE, rather it should be the result of an
operation. In other words, it should always be possible to SORT a
relation in any given order, and this ability should not be dependent
on any previous declarations that exist regarding the relation.
b) Therefore, we will need a unary operator to sort a relation:
ORDER (R, x1, ..., xn)
or:
ORDER (R)
depending on whether the relation is to be sorted by some or all of
its components.
c) It should be possible to use an
"(ORDER (R))"
instead of
"R"
wherever R occurs. For instance, when we say
ANY (ORDER (R))
it would imply that we remove the first tuple in R according to the
order defined by the ORDER operator.
d) I think we need two SAVE←ORDER operators that tell the DBMS
to make an order permanent (for most DBMSs this would be equivalent
to creating an index instead of just doing a sort):
SAVE←ORDER (R)
and
SAVE←ORDER←STACK (R).
The former would imply that after being PUSHed and POPed, or after
being referenced with a version number > 0 (eg R[5]), the index would no
longer exist. Whereas the latter would imply that after being PUSHed
and POPed or being referenced with a non-zero version number, the
index will still exist.
The ICODE specifications should not specify how these are
implemented.
e) When it comes to performance, some binary operators (like joins)
may not be symmetrical, ie JOINing A with B may not be the same as
joining B with A. Therefore, the ICODE syntax should specify that
JOIN A,B
means go from A to B. Thus if we had an index on B and did not have
an index on A, it might be more efficient to do:
JOIN A, (ORDER (B, ...))
than to do
JOIN (ORDER (B, ...)), A
and we should be able to control this efficiency.
(I don't feel completely comfortable with this proposal, especially the
way it handles saving indexes on stacks, but I think that at least it
gives us something to start from.)
DUPLICATES:
We will probably need an operator to remove duplicates from a
relation. For instance:
A := UNIQUE B
Unique should permit to write to itself, eg:
A := UNIQUE A.
DATA TYPES:
It seems to me that somewhere along the way we will have to encorporate
other types of variables and not just character strings. For example,
numbers are a very important concept. So much of what we do is
iterative in nature (and *not* recursive) and it would be a pity if
the ICODE was not rich enough to support this type of capability.
ARITHMETIC OPERATORS (ala APL):
I think that at some future time we should explore this in more
depth.
GOTOs:
Don't apologize for them - KEEP THEM. Any language that has
eliminated GOTOs has had to develop a large number of constructs to
use in their stead, eg EXIT, LEAVE, LOOP, WHILE, FOR, etc. I don't
see why we should have to support dozens of different constructs on
the ICODE level. Remember, the users will not be using this
language.
IMPLEMENTATION OF RELATIONS:
Another thought that should be brought up - storing versions of a
relation makes it that much harder to decide how to implement the
storage. For instance, what if a version of a relation has 1000
tuples, 1 is changed, and a new version of the relation is PUSHed.
Do we really want to save all 100 tuples or only the one that has
changed (ala SCCS)?
Regarding Allen's latest proposals:
===================================
LOGICAL SYNTAX:
I think that the real question I have regarding Allen's approach has
to do with our goals: what are we trying to achieve with the ICODE?
It seems to me that if our motive is to develop a low level language
that will permit efficient usage of a database, we are probably going
to do better with Jeff's approach.
For example, in Jeff's version we can support a GOTO, whereas in Allen's
we can't. Personally, I think GOTOs are very important for what we
want to do. If we give up on them we will have to pay a high price
in efficiency, or else we will need another level of code (I-I-CODE?)
that the I-CODE would have to be translated into.
CHANGE:
I think the idea of a CHANGE for each relation will be very useful.
VERSION NUMBERS:
Allen, could you give some examples where it might be advantageous to
use your generalized relation version?
∂17-Jun-85 1003 @SU-CSLI.ARPA:sugai.pa@Xerox.ARPA 6/18, RRR Meeting
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 17 Jun 85 10:01:32 PDT
Received: from Xerox.ARPA by SU-CSLI.ARPA with TCP; Mon 17 Jun 85 09:53:16-PDT
Received: from Semillon.ms by ArpaGateway.ms ; 17 JUN 85 09:53:41 PDT
Date: 17 Jun 85 09:53 PDT
From: sugai.pa@Xerox.ARPA
Subject: 6/18, RRR Meeting
To: RRR@SU-CSLI.ARPA
cc: sugai.pa@Xerox.ARPA
Message-ID: <850617-095341-1446@Xerox>
The last RRR meeting for the summer will be tomorrow (Tuesday, June 18)
at 2:15 p.m. with Jerry Fodor discussing his paper, "Why There Still Has
To Be A Language of Thought."
∂17-Jun-85 1004 @SU-CSLI.ARPA:PCOHEN@SRI-AI.ARPA [Phil Cohen <PCOHEN@SRI-AI.ARPA>: Next NL4 talk, Lucy Suchman]
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 17 Jun 85 10:04:08 PDT
Received: from SRI-AI.ARPA by SU-CSLI.ARPA with TCP; Mon 17 Jun 85 09:55:04-PDT
Date: Mon 17 Jun 85 09:56:26-PDT
From: Phil Cohen <PCOHEN@SRI-AI.ARPA>
Subject: [Phil Cohen <PCOHEN@SRI-AI.ARPA>: Next NL4 talk, Lucy Suchman]
To: finterest@SU-CSLI.ARPA,
NL4: ;
A reminder...
---------------
Mail-From: PCOHEN created at 10-Jun-85 13:11:50
Date: Mon 10 Jun 85 13:11:50-PDT
From: Phil Cohen <PCOHEN@SRI-AI.ARPA>
Subject: Next NL4 talk, Lucy Suchman
To: NL4: ;
cc: finterest@SU-CSLI.ARPA
Our next NL4 seminar will be June 18, at 12:45, in the Ventura Seminar
room.
Speaker: Lucy Suchman
Intelligent Systems Laboratory
Xerox PARC
Plans and Situated Actions
Researchers in Cognitive Science take the organization and
significance of action to be derived from plans, which are
prerequisite to and prescribe action at whatever level of detail one
might imagine. Mutual intelligibility on this view is a matter of the
recognizability of plans, due to common conventions for the expression
of intent, and common knowledge about typical situations and
appropriate actions. An alternative view, drawn from recent work in
social science, treats plans as derived from situated actions.
Situated actions as such comprise necessarily ad hoc responses to the
actions of others and to the contingencies of particular situations.
Rather than depend upon the reliable recognition of intent, successful
interaction consists in the in situ production of intelligibility,
through mutual access to situation resources, and through the
detection, repair or exploitation of differences in understanding.
As common sense formulations, designed to accomodate the unforseeable
contingencies of situated action, plans are inherently vague. One
approach to human-machine communication has been to improve on the
vagueness of plans, to make them the basis for artifacts able to
interact with their human users. This talk looks at the problem of
human-machine interaction through a case study of people using a machine
designed on the planning model, and intended to be intelligent and
interactive. A conversation analysis of "interactions" between users
and the machine reveals that the machine's insensitivity to particular
circumstances is both a central design resource, and a fundamental
limitation. More generally, problems in Cognitive Science's theorizing
about purposeful action as a basis for machine intelligence can be
understood as an artifact of the substitution of plans for actions, and
representations of the situation of action for action's actual
circumstances.
-------
-------
∂17-Jun-85 1004 @SU-CSLI.ARPA:PCOHEN@SRI-AI.ARPA [Phil Cohen <PCOHEN@SRI-AI.ARPA>: Next NL4 talk, Lucy Suchman]
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 17 Jun 85 10:04:08 PDT
Received: from SRI-AI.ARPA by SU-CSLI.ARPA with TCP; Mon 17 Jun 85 09:55:04-PDT
Date: Mon 17 Jun 85 09:56:26-PDT
From: Phil Cohen <PCOHEN@SRI-AI.ARPA>
Subject: [Phil Cohen <PCOHEN@SRI-AI.ARPA>: Next NL4 talk, Lucy Suchman]
To: finterest@SU-CSLI.ARPA,
NL4: ;
A reminder...
---------------
Mail-From: PCOHEN created at 10-Jun-85 13:11:50
Date: Mon 10 Jun 85 13:11:50-PDT
From: Phil Cohen <PCOHEN@SRI-AI.ARPA>
Subject: Next NL4 talk, Lucy Suchman
To: NL4: ;
cc: finterest@SU-CSLI.ARPA
Our next NL4 seminar will be June 18, at 12:45, in the Ventura Seminar
room.
Speaker: Lucy Suchman
Intelligent Systems Laboratory
Xerox PARC
Plans and Situated Actions
Researchers in Cognitive Science take the organization and
significance of action to be derived from plans, which are
prerequisite to and prescribe action at whatever level of detail one
might imagine. Mutual intelligibility on this view is a matter of the
recognizability of plans, due to common conventions for the expression
of intent, and common knowledge about typical situations and
appropriate actions. An alternative view, drawn from recent work in
social science, treats plans as derived from situated actions.
Situated actions as such comprise necessarily ad hoc responses to the
actions of others and to the contingencies of particular situations.
Rather than depend upon the reliable recognition of intent, successful
interaction consists in the in situ production of intelligibility,
through mutual access to situation resources, and through the
detection, repair or exploitation of differences in understanding.
As common sense formulations, designed to accomodate the unforseeable
contingencies of situated action, plans are inherently vague. One
approach to human-machine communication has been to improve on the
vagueness of plans, to make them the basis for artifacts able to
interact with their human users. This talk looks at the problem of
human-machine interaction through a case study of people using a machine
designed on the planning model, and intended to be intelligent and
interactive. A conversation analysis of "interactions" between users
and the machine reveals that the machine's insensitivity to particular
circumstances is both a central design resource, and a fundamental
limitation. More generally, problems in Cognitive Science's theorizing
about purposeful action as a basis for machine intelligence can be
understood as an artifact of the substitution of plans for actions, and
representations of the situation of action for action's actual
circumstances.
-------
-------
∂17-Jun-85 1009 ullman@diablo next meeting
Received: from SU-AIMVAX.ARPA by SU-AI.ARPA with TCP; 17 Jun 85 10:09:35 PDT
Date: Mon, 17 Jun 85 10:01:05 pdt
From: Jeff Ullman <ullman@diablo>
Subject: next meeting
To: nail@diablo
As no one has volunteered to read a paper, we may as well
devote the meeting to further discussion of intermediate code.
See you wednesday?
∂17-Jun-85 1140 WINSLETT@SU-SCORE.ARPA "any" and "goto"
Received: from SU-AIMVAX.ARPA by SU-AI.ARPA with TCP; 17 Jun 85 11:40:17 PDT
Received: from SU-SCORE.ARPA by diablo with TCP; Mon, 17 Jun 85 11:27:53 pdt
Date: Mon 17 Jun 85 11:27:29-PDT
From: Marianne Winslett <WINSLETT@SU-SCORE.ARPA>
Subject: "any" and "goto"
To: nail@SU-AIMVAX.ARPA
The purpose of a generator like "any" is to free the program writer
from figuring out which item to select next in the iteration, and to
shift that decision to someone who presumably is in a better position
to choose, namely, the person who wrote the data structure that is
being explored. The dbms user should not assume that a call to "any"
on a sorted relation would return the first element of the relation.
Generalized constructs like "any" are aesthetically pleasing and
provide independence from the underlying system, and yet a dbms user
would often prefer to process tuples in some particular order.
The loop-controlling "goto" need not be replaced by a hodge-podge of
constructs. A loop/exit-when construct is simple, yet more flexible
than while+for+repeat-until. And even a hodge-podge of "goto"
replacements is easier for a user to read and modify, and easier for a
computer to optimize and verify. However, since "goto" has uses other
than in loops, and since the intermediate code should perhaps be purged
of higher-level constructs, "goto" may be the way to go.
-------
∂17-Jun-85 1246 @MIT-MC.ARPA:snewman.pa@XEROX.ARPA Please add me to PHILOSOPHY-OF-SCIENCE DL
Received: from MIT-MC.ARPA by SU-AI.ARPA with TCP; 17 Jun 85 12:45:54 PDT
Received: from MIT-OZ by MIT-MC.ARPA via Chaosnet; 17 JUN 85 15:45:46 EDT
Received: from MIT-MC by MIT-OZ via Chaosnet; 17 Jun 85 15:41-EDT
Received: from Xerox.ARPA by MIT-MC.ARPA 17 Jun 85 15:42:31 EST
Received: from Salvador.ms by ArpaGateway.ms ; 17 JUN 85 12:40:45 PDT
Date: 17 Jun 85 12:32 PDT
From: snewman.pa@Xerox.ARPA
Subject: Please add me to PHILOSOPHY-OF-SCIENCE DL
To: PHILOSOPHY-OF-SCIENCE@MIT-MC.ARPA
cc: snewman.pa@Xerox.ARPA
Message-ID: <850617-124045-1727@Xerox>
Thanks.
∂17-Jun-85 1511 avg@diablo Re: more ICODE comments
Received: from SU-AIMVAX.ARPA by SU-AI.ARPA with TCP; 17 Jun 85 15:10:30 PDT
Date: Mon, 17 Jun 85 15:02:18 pdt
From: Allen VanGelder <avg@diablo>
Subject: Re: more ICODE comments
To: mhb@FORD-WDL1.ARPA, nail@diablo
In regard to versions, I am suggesting that we don't need additional
language features to do what we want with versions. Why clutter up the
meaning of simple operations by having implicit creation and reference to
versions?
My global concept of Icode is that it add just enough to the user's
logic program to make it work, or work efficiently. This addition is
the embodiment of the capture decisions. E.g., if the Icode were a
Prolog program that works, subject to running out of space, that would
be fine. Let someone else write the better Prolog interpreter/compiler
and let them worry about GOTOs.
Let's get a prototype working and implement a few capture rules, and
then see if we need more than a WHILE for iteration.
∂17-Jun-85 2033 @SUMEX-AIM.ARPA:YAN@SU-SCORE.ARPA Abstract for this Wednesday morning's talk
Received: from SUMEX-AIM.ARPA by SU-AI.ARPA with TCP; 17 Jun 85 20:33:02 PDT
Received: from SU-SCORE.ARPA by SUMEX-AIM.ARPA with TCP; Mon 17 Jun 85 20:32:46-PDT
Date: Mon 17 Jun 85 20:32:45-PDT
From: Jerry C. Yan <YAN@SU-SCORE.ARPA>
Subject: Abstract for this Wednesday morning's talk
To: aap@SUMEX-AIM.ARPA
cc: yeager@SUMEX-AIM.ARPA, mak@SU-SIERRA.ARPA
Ladies and Gentlemen,
This Wednesday at 9:30, I will present (informally again) the
final talk in the series on "Resource Management in a Distributed
Environment running X/CZAR/CARE/SIMPLE".
Recall that resouce management is partitioned into 3 sub-problems:
static/compile time placement
dynamic placement/relocation
dynamic reclamation.
This presentation deals with the first aspect: "Where does one place agents
given some predictions on how they will interact?"
The first idea is to divide the placement into two phases:
clustering and mapping. Clustering is topology independent. It decides
which agents reside on the same site. In the second phase, clusters are
"mapped" to a particular sites - minimizing communication costs between
them.
The second key(new) idea is to define a ratio known as the
"CRITICAL DISTANCE" between any two agents. Loosely speaking:
amount of parallelism between i & j
CRITICAL DISTANCE(i,j) = --------------------------------------
amount of communication between i & j
Two agents should not be placed further apart than the CD.
A table of CDs between all agents are built. Clustering is
done by "collapsing" this table in the first stage. At the second stage,
the table represent constrains on placement - thus guides the
combinatorial search.
See you Wednesday!
=jerry=
-------
∂18-Jun-85 0828 LIBRARY@SU-SCORE.ARPA Math/CS Library Hours For The Summer
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 18 Jun 85 08:28:41 PDT
Date: Tue 18 Jun 85 08:27:14-PDT
From: C.S./Math Library <LIBRARY@SU-SCORE.ARPA>
Subject: Math/CS Library Hours For The Summer
To: faculty@SU-SCORE.ARPA
We have received additional funds for staffing longer hours in the summer.
Our hours will be:
Monday-Thursday 8am-7pm
Friday 8am-6pm
Sunday 1pm-5pm
These service hours are effective as of Tuesday June 18, 1985 and will continue
throughout the summer.
Harry
-------
∂18-Jun-85 0914 LIBRARY@SU-SCORE.ARPA Socrates: Alternate way of telneting when Lindy is down!!!
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 18 Jun 85 09:13:57 PDT
Date: Tue 18 Jun 85 09:08:54-PDT
From: C.S./Math Library <LIBRARY@SU-SCORE.ARPA>
Subject: Socrates: Alternate way of telneting when Lindy is down!!!
To: su-bboards@SU-SCORE.ARPA
cc: faculty@SU-SCORE.ARPA
Last week there were some problems with telneting into Lindy. However there
is an alternate way to login to Forsythe from the network. It uses a
commercial IP Gateway that Bridge communications builds. Here is how you
may login to Forsythe from the net:
Step #1: Login to a Dec-20 or Unix machine (SCORE, Lots ect)
Step #2: Type: telnet 36.9.0.30
Step #3: You should get the "Welcome to ITS...." messages now.
You may use this approach when Lindy is down or even when Lindy is up.
The 36.9.0.30 will eventually be replaced by the word "fob"
(Forsythe-Bridge). This approach is not completely stable.
Now that we have a number of people using Socrates from their own computers
we need an effective way to communicate quickly with users when problems
come up. At this point, I would suggest that if you are having telneting
or other problems with Socrates send me a message Library@SCORE. At that
point, I will announce the problem on the bulletin board and then begin
investigating with ITS and the Library to find out what is happening.
I will continue using the bulleten board to announce updates on the
problems and when they have been cleared.
Harry
-------
∂18-Jun-85 1934 @SU-CSLI.ARPA:LANSKY@SRI-AI.ARPA NEXT MONDAY'S PLANLUNCH
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 18 Jun 85 19:34:42 PDT
Received: from SRI-AI.ARPA by SU-CSLI.ARPA with TCP; Tue 18 Jun 85 19:25:04-PDT
Date: Tue 18 Jun 85 19:27:09-PDT
From: LANSKY@SRI-AI.ARPA
Subject: NEXT MONDAY'S PLANLUNCH
To: aic-associates@SRI-AI.ARPA, folks@SU-CSLI.ARPA, rsimmons@SRI-KL.ARPA
PLANNING AS DEBUGGING
Reid Simmons -- MIT AI Lab / SPAR
11:00 am, Monday, June 24
Room EJ232, SRI International
We are currently building a domain independent planner which can
represent and reason about fairly complex domains. The first part of
the talk will focus on the representations used and the rationale for
choosing them. The planner uses explicit temporal representations,
based on time points and the notion of "histories". It also extends
the traditional precondition/postcondition representation of actions
to include quantification, conditionals and the ability to reason
about cumulative changes.
The second part of the talk will focus on techniques to organize
and control the search for a plan. We view planning as "debugging
a blank sheet of paper". We correct a bug (ie. unachieved goal) by
changing one of the underlying assumptions in the plan which are
responsible for the bug. This problem solving approach combines
backtracking with traditional planning techniques, giving the planner
the potential for finding a solution with much less search. We also
present a simple, but effective, technique for choosing which plan
modification to pursue, based on maintaining a complete goal structure
of the plan.
This planner has been partially implemented and tested on
traditional blocks-world and register-transfer examples. It is
currently being applied to the problem of geologic interpretation and
to diagnosis of chip manufacturing problems.
-------
∂19-Jun-85 1809 EMMA@SU-CSLI.ARPA Newsletter June 20, No. 34
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 19 Jun 85 18:09:18 PDT
Date: Wed 19 Jun 85 17:02:36-PDT
From: Emma Pease <Emma@SU-CSLI.ARPA>
Subject: Newsletter June 20, No. 34
To: friends@SU-CSLI.ARPA
Tel: 497-3479
C S L I N E W S L E T T E R
←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←
June 20, 1985 Stanford Vol. 2, No. 34
←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←
A weekly publication of The Center for the Study of Language and
Information, Ventura Hall, Stanford University, Stanford, CA 94305
←←←←←←←←←←←←
CSLI ACTIVITIES FOR *THIS* THURSDAY, June 20, 1985
12 noon TINLunch
Ventura Hall ``How Many Possible Human Languages are There?''
Conference Room by Geoff Pullum, UCSC and CSLI
Discussion led by Gerald Gazdar, CASBS
2:15 p.m. CSLI Seminar
Redwood Hall No seminar
Room G-19
3:30 p.m. Tea
Ventura Hall
4:15 p.m. CSLI Colloquium
Redwood Hall No colloquium
Room G-19
←←←←←←←←←←←
CSLI ACTIVITIES FOR *NEXT* THURSDAY, June 27, 1985
12 noon TINLunch
Ventura Hall ``The Algebra of Events''
Conference Room by Emmon Bach
Discussion led by Edit Doron
(Abstract on page 2)
2:15 p.m. CSLI Seminar
Redwood Hall ``An Organism and Its Internal Model of the World''
Room G-19 Pentti Kanerva, CSLI
Discussion led by Alex Pentland
(Abstract on page 2)
3:30 p.m. Tea
Ventura Hall
4:15 p.m. CSLI Colloquium
Redwood Hall ``Qualitative Process Theory''
Room G-19 Ken Forbus, University of Illinois, Computer Science
(Abstract on page 3)
←←←←←←←←←←←←←
ANNOUNCEMENT
No seminars or colloquia are scheduled for June 20 because of the
University, end-of-quarter break. TINLunch will be held on that day.
Regular activities will resume on June 27.
!
Page 2 CSLI Newsletter June 20, 1985
←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←
ABSTRACT OF NEXT WEEK'S TINLUNCH
``The Algebra of Events''
``The Algebra of Events'' by Emmon Bach takes as its point of
departure the apparently close parallels between the mass-count
distinction in nominals and the distinction between processes and
events (also known as the activity-accomplishment distinction) in
aspectual classes of verbs. These parallels have been commented on by
a number of scholars over the last decade but never analyzed formally.
What Bach demonstrates here is that the richly-structured
model-theoretic semantics for mass, count, and plural nominals
developed by Godehard Link provides the model for a semantics of verb
aspect (including an ontology of events) that is explicitly parallel
to nominal semantics. After sketching this Link-type semnatics for
events, he shows that the resulting formal analogy between the two
domains leads to the discovery of further properties of both events
and nominals, such as a nominal analogue of the so-called
``imperfective paradox.'' --David Dowty
←←←←←←←←←←←←←
ABSTRACT OF NEXT WEEK'S SEMINAR
``An Organism and Its Internal Model of the World''
There is a glaring disparity in how children and computers learn
things. By and large, children are not instructed explicitly but
learn by observation, imitation, and trial and error. What kind of
computer architecture would allow a machine to learn the way children
do?
In the model I have been studying, an organism is coupled to the
world by its sensors and effectors. The organism's mind-ware consists
of a relatively small focus and a large memory. The sensors feed
information into the focus, the effectors are driven from the focus,
the memory is addressed by the contents of the focus, the contents of
the focus are stored in memory, and the memory feeds information into
the focus. The contents of the focus at a moment account for the
subjective experience of the organism at that moment.
The function of the memory is to store a model of the world for
later reference. The memory is sensitive to similarity in that
approximate retrieval cues can be used to retrieve exact information.
It is dynamic in that the present situation (its encoding) brings to
focus the consequences of similar past situations. The model sheds
light on the frame problem of robotics, and it appears that a robot
built according to this principle would learn by trial and error and
would be able to plan actions and to perform planned sequences of
actions.
Reading: ``Parallel Structures in Human and Computer Memory,''
available from Susi Parker at the Ventura Hall receptionist desk and
on line as <PKANERVA>COGNITIVA.PAPER at SU-CSLI.ARPA.
--Pentti Kanerva
!
Page 3 CSLI Newsletter June 20, 1985
←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←
ABSTRACT OF NEXT WEEK'S COLLOQUIUM
``Qualitative Process Theory''
Things move, collide, flow, bend, stretch, break, cool down, heat up,
and boil. Intuitively we think of the things that cause changes in
physical situations as processes. Qualitative Process Theory defines
simple notions of quantity, function, and process that allow
interesting common-sense inferences to be drawn about dynamical
systems. This talk will describe the basics of the Qualitative
Process Theory, illustrate how it can be used to capture certain
aspects of different models of physical phenomena, and discuss the
claims it makes about causal reasoning. --Ken Forbus
←←←←←←←←←←←←←
KROCH LECTURE SERIES
Sponsored by CSLI and Stanford Linguistics Dept.
Tony Kroch (University of Pennsylvania) will give a series of five
lectures on ``Grammar, Processing, and Linguistic Change'' on June 26
- July 2. The lectures will be held at 3:15 p.m. in the CSLI Seminar Room
(not in Cubberley as previously announced).
1. Grammar and Usage -- (Wednesday, June 26) What can we learn
about linguistic structure from looking at usage data? Importance of
this question for historical linguistics, where the data available is
all usage data. The problem that such data show the influence of all
the factors that influence linguistic patterns. The fallacy of
treating usage data as direct indications of grammatical organization
and the contrary fallacy of ignoring the information revealed by
studies of usage.
2. Some Promising Results -- (Thursday, June 27) When and how
usage patterns reflect grammar. The experimental and observational
evidence for syntactic priming; its use as a probe for linguistic
structure. The problem of what determines overall frequencies of use.
The nature of style shifting. Human beings as trackers of
frequencies.
3. A Mathematical Model of Syntactic Change -- (Friday, June 28)
The characteristic S-shaped profile of linguistic drift. Evidence for
its generality. How it can be modeled mathematically and explained
psycholinguistically. The notion of competition among alternative
forms. The problem of competition in the face of differences in
meaning.
4. A Case Study -- (Monday, July 1) The loss of subject-verb
inversion in English. The nature of the change and its relationship
to the loss of Germanic word order. The rise of periphrastic 'do' and
the substitution of subject-aux inversion for subject-verb inversion
in questions. Parallel developments in French and Portuguese.
5. Processing Effects on Usage Patterns and Their Role in Change
-- (Tuesday, July 2) Where can we look for the active force behind the
change described in the previous lecture? Why it cannot be internal
to the grammar. A solution in constraints on sentence processing. The
limits of processing effects on usage.
Copies of selected papers by Kroch, will be available in the
Greenberg Room and at CSLI. To arrange appointments with Kroch,
please call Sonia Oliva at 7-4284.
!
Page 4 CSLI Newsletter June 20, 1985
←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←
AREA P1 MEETING
``Pixels and Predicates''
Beginning Wednesday, June 26 we will start a discussion series on
visual (graphic) communication: how can we relate predicates to
pixels, and vice versa?
Topics will include:
* What image regularities do we perceive as the primitive elements
of form, the ``visual morphemes'' that convey information?
* How do people organize images into these parts, gain information
about the situation from them, and use them in communication?
* How can we use our knowledge of such matters to design graphic
interfaces to facilitate visual communication?
Those interested in these topics are encouraged to attend, debate
vigorously, and perhaps suggest further topics for discussion. The
first speaker (tentative) is:
``Visual Morphemes in the 3-D World.''
Alex Pentland, CSLI
Wednesday June 26th, 3:00pm, Ventura Hall
People have a strong perceptual notion of the ``parts'' of a 3-D
form; a good understanding of what constitues ``a part'' is critical
to communication about visual data. A theory of parts will be
presented and a 3-D graphics modeling tool based on this theory will
be discussed. --Alex Pentland
←←←←←←←←←←←←←
LOGIC, LANGUAGE, AND COMPUTATION MEETINGS
July 8-19, 1985, Stanford University
The final mailing for those coming to the meetings will be sent out
by June 25.
←←←←←←←←←←←←←
NEW CSLI POSTDOCTORAL STUDENTS
CSLI is sponsoring three, new postdoctoral students for the
academic year 1985-1986. Carol Cleland who arrives July 1 has a Ph.D.
in Philosophy from Brown University and has most recently worked on
developing an expert system. Mark Gawron who will arrive on August 5
has a Ph.D. in Linguistics from the University of California at
Berkeley and has worked at the Department of Artificial Intelligence,
University of Edinburgh and at the Courant Institute of Mathematical
Science in New York. Helene Kirchner who will come in September has a
These d'Etat in Computer Science from the University of Nancy I in
France and has been a researcher at the Centre National de la
Recherche Scientifique.
-------
∂19-Jun-85 1818 SCHMIDT@SUMEX-AIM.ARPA Stanford HPP 3600's
Received: from SUMEX-AIM.ARPA by SU-AI.ARPA with TCP; 19 Jun 85 18:17:42 PDT
Date: Wed 19 Jun 85 16:09:38-PDT
From: Christopher Schmidt <SCHMIDT@SUMEX-AIM.ARPA>
Subject: Stanford HPP 3600's
To: Hostmaster@SRI-NIC.ARPA
cc: Almquist@SUMEX-AIM.ARPA, KSL-Lisp-Machines@SUMEX-AIM.ARPA
The HPP 3600's seem to have mysteriously disappeared from HOSTS.TXT.
In any event, here they are again plus changes plus 2 more machines.
Thanks, --Christopher
HOST : 36.8.0.28 : HPP-3600-1.ARPA,HPP-3600-1,SU-SILVER
: SYMBOLICS-3600
: LISPM
: TCP/TELNET,TCP/FTP,TCP/FINGER,TCP/TIME,UDP/TIME,UDP/TFTP,UDP/FINGER :
HOST : 36.47.0.10 : HPP-3600-2.ARPA,HPP-3600-2
: SYMBOLICS-3600
: LISPM
: TCP/TELNET,TCP/FTP,TCP/FINGER,TCP/TIME,UDP/TIME,UDP/TFTP,UDP/FINGER :
HOST : 36.8.0.18 : HPP-3670-3.ARPA,HPP-3670-3,SU-GOLD
: SYMBOLICS-3600
: LISPM
: TCP/TELNET,TCP/FTP,TCP/FINGER,TCP/TIME,UDP/TIME,UDP/TFTP,UDP/FINGER :
HOST : 36.47.0.12 : HPP-3670-4.ARPA,HPP-3670-4
: SYMBOLICS-3600
: LISPM
: TCP/TELNET,TCP/FTP,TCP/FINGER,TCP/TIME,UDP/TIME,UDP/TFTP,UDP/FINGER :
HOST : 36.47.0.14 : HPP-3600-5.ARPA,HPP-3600-5
: SYMBOLICS-3600
: LISPM
: TCP/TELNET,TCP/FTP,TCP/FINGER,TCP/TIME,UDP/TIME,UDP/TFTP,UDP/FINGER :
HOST : 36.47.0.15 : HPP-3600-6.ARPA,HPP-3600-6
: SYMBOLICS-3600
: LISPM
: TCP/TELNET,TCP/FTP,TCP/FINGER,TCP/TIME,UDP/TIME,UDP/TFTP,UDP/FINGER :
-------
∂20-Jun-85 1211 MCCABE@SUMEX-AIM.ARPA Symbolics Software Catalogue
Received: from SUMEX-AIM.ARPA by SU-AI.ARPA with TCP; 20 Jun 85 12:11:22 PDT
Date: Thu 20 Jun 85 12:09:41-PDT
From: Patti McCabe <MCCABE@SUMEX-AIM.ARPA>
Subject: Symbolics Software Catalogue
To: KSL-Lisp-Machines@SUMEX-AIM.ARPA
Does anyone have any software that runs on a Symbolics machine that
they would like publicized in a catalogue that Symbolics puts out on
software available for their machines?
If you have any (such as MRS) please send me a brief description, the dollar
amount that will be charged for the software and a contact on how to order
the software.
Patti McCabe
-------
∂20-Jun-85 1411 LIBRARY@SU-SCORE.ARPA Math/CS Library New Books
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 20 Jun 85 14:11:02 PDT
Date: Thu 20 Jun 85 14:06:35-PDT
From: C.S./Math Library <LIBRARY@SU-SCORE.ARPA>
Subject: Math/CS Library New Books
To: su-bboards@SU-SCORE.ARPA
cc: faculty@SU-SCORE.ARPA
Advances in Petri Nets 1984. edited by G. Rozenberg. QA267.A3
Petri Nets An Introduction. by Wolfgang Reisig QA267.R4513 1985
Mathematical Foundations of Software Development Proceedings of the
International Joint Conference on Theory and Practice of Software
Development March 1985. Volume 1: Collogquium on Trees in Algebra
and Programming. QA76.76.D47I57 1985 V. 1
Proceedings of the Third Joint Ada Europe/AdaTEC Conference, Brussels
26-28 June 1984. edited by J. Teller
Intelligent Machines: an Introductory Perspective of Artificial Intelligence
and Robotics. by William Gevarter. Q335.G483 1984
Tailoring Software for Multiple Processor Systems by Karsten Schwan
QA76.5.S318 1985
System Design With Ada by Buhr QA76.73.A35B83 1984
Structured Cobol Reference Summary. National Computing Centre (8507651)
Understanding FORTRAN 77 With Structured Problem Solving by Michel
Boillot. QA76.73.F25B643 1984
How To Use The Macintosh. Jerry Willis (8507709)
Harry Llull
-------
∂20-Jun-85 1535 avg@diablo NAIL source language proposal
Received: from SU-AIMVAX.ARPA by SU-AI.ARPA with TCP; 20 Jun 85 15:35:35 PDT
Date: Thu, 20 Jun 85 15:23:26 pdt
From: Allen VanGelder <avg@diablo>
Subject: NAIL source language proposal
To: nail@diablo
NAIL SOURCE LANGUAGE PROPOSAL Allen Van Gelder June 20, 1985
This is a first draft source language proposal for NAIL.
The file is ~avg/nail/source.prop on diablo.
Familiarity with Cprolog is assumed. "Prolog" means Cprolog.
A Nail program is a file containing rules and certain directives.
Rules look like Prolog rules, e.g., p(X, Y) :- a(X), b(Y, [Z | ←]), c(f(Z)).
Operators work as in Prolog.
User-defined operators are defined in directives as in Prolog,
and must be defined before their first use. E.g.,
:- op(400, xfy, :).
defines right-associative operator ':' with the same precedence as '*'.
The other directive that effects the reading of the program is "include":
:- include(F).
causes the contents of F, which must be instantiated to a file name,
to be inserted at that point in the program. This is a Nail directive.
Other Nail directives are part of the language, and are covered later.
If a directive that is not a Nail directive is encountered, it will be
passed to Prolog for execution, at the user's risk, possibly preceded by
a warning message. The anticipated use of such directives is for debugging
Nail, assuming it is implemented in Prolog. If Nail is not implemented in
Prolog, such directives will be implementation-dependent.
Nail will recognize a subset of Prolog's evaluable predicates. The classes
of evaluable (built-in) predicates are:
Input/Output
Arithmetic
Convenience
Extra Control
Meta-Logical
Sets
Term Comparison
Environmental facilities
Nail requires arguments to built-in predicates to be properly bound (or free);
it has a table of proper combinations for use in capture decisions.
INPUT/OUTPUT
To control formating, Nail provides a directive, "portray," which works
analogously to Prolog.
:- portray(emp(X, Y), (write('employee: '), write(X), tab(2), write(Y), nl)).
has the same effect as
portray(emp(X, Y)) :- write('employee: '), write(X), tab(2), write(Y), nl.
would in a Prolog program; i.e., "write(Term)" checks to see if Term or any
subterm of it matches a "portray" declaration.
The only Input/Output predicates to be initially supported are write,
tab(N), and nl. (The user may be able to change the output stream between
queries. This would be part of the query language, not part of a Nail
program.)
ARITHMETIC
As in Prolog. However, "Z is X" where X is an arithmetic expression is
subject to optimization if X is of a sufficiently simple form.
Similarly "X > Y" may be optimized. Here is an example:
p(X, Y) :- q(X, Z), Z > X, r(Z, Y), 3 is Z+Y.
may be executed as "retrieve q(X, Z) where Z > X; retrieve r(Z, Y) where
Z+Y = 3; join on Z." (The subgoal order is immaterial, of course.)
CONVENIENCE
As in Prolog, except that X \= Y will be rewritten as \+(X = Y) and
semi-colon terms will be rewritten with generated predicate names. E.g.,
p(X) :- (q(X, 1); r(X)), s(X).
will become something like
p(X) :- 'pred#7'(X), s(X).
'pred#7'(←1) :- q(←1, 1).
'pred#7'(←1) :- r(←1).
The arguments of the generated term will be just the variables that appear
both within the scope of the semi-colon and outside its scope as well.
The \+ predicate is discussed in the next section.
EXTRA CONTROL
This category is greatly reduced in Nail.
The cut, "!", is a no-op to Nail as far as results go. It may be used
as a guard to help Nail recognize mutually exclusive rules, and take
appropriate optimization actions. However, this will be transparent to
the user, and probably will not be implemented in any early version.
The finite failure predicate \+, commonly read as "not," but more
precisely read as "not provable," is accepted by Nail, but its meaning
needs to be determined carefully. In Prolog the position of \+ in the
list of subgoals is material and often critcal; in Nail, we want it to be
immaterial. Here is an example:
bachelor(X) :- \+married(X, Y), male(X).
In Prolog, if bachelor(X) is called with X uninstantiated, then
married(X, Y) presumably succeeds for SOME (X, Y), so \+ fails; i.e.,
there are no bachelors found. However, bachelor(john) will succeed
when appropriate. Moreover, bachelor(X) will succeed as we intend if
the subgoals are reversed in order, in Prolog.
We need Nail to give a consistent evaluation.
Additional problems remain to be resolved if a rule such as
p(X, Y) :- \+p(Y, Z), q(Z).
appears in a Nail program. If first implementation of Nail will probably not
permit such circular use of \+.
The "embedded if" (P -> Q; R) will be rewritten as a generated predicate G
with rules:
G :- P, !, Q.
G :- \+P, !, R.
If R is not present, "fail" will be substituted for R. This also illustrates
a use of "!" as a guard. However, in Prolog, P is allowed to succeed only
once; in Nail it may succeed any number of times.
The 0-ary "fail" always fails.
The predicates "once," "repeat," and "forall" are not part of Nail.
It would be desirable to have "forall" in some future version.
META-LOGICAL
No meta-logical Prolog predicates are included in Nail in this version.
Type predicates atom, number, integer, and atomic are candidates for a
future version, but with the meaning that the argument should be
restricted to that class in future instantiations, as well as checking
whether it is already so instantiated. For this reason, var and nonvar
are not similar. Also, db←reference is not considered a type.
Limited forms of functor, arg, and =.. may also be implemented in the future.
Here are some candidates for meta-logical directives to be used by Nail:
:- edb(R) a database relation
:- capturable(R, Pattern, CaptureRule)
:- fixed(emp(name(last(W), first(X)), dept(Y), sal(Z)))
This allows structured values like Pascal records to be
present as predicate arguments
without allowing RECURSIVE structures like lists
and trees. Non-recursive structures are as "easy" as
function-free arguments.
Nail will provide a relation evaluable(P) to identify Prolog predicates that
are simply referred to Prolog (with properly instantiated arguments, in
accordance with the Nail-provided capturable(P) clauses). Thus the
imposing array of evaluable predicates listed herein should not be much
of an implementation problem if Nail is done over Prolog. Otherwise,
all bets are off!
SETS
Nail allows bagof. Problems of self-reference need to be solved, as in:
p(X) :- bagof(Y, p(Y), X).
which means "p(X) holds if X is a list of terms Y such that p(Y) holds."
Other questions concerning subgoal order need to be resolved, as was the
case for \+.
The notation Z↑p(Y, Z) for "there exists Z such that p(Y, Z)" is not resolved.
COMPARISON OF TERMS
Only "compare(Op, T1, T2)" is available in Nail. This predicate is
capturable only with all arguments bound, and is subject to optimization,
as were arithmetic predicates.
The sort predicates sort, msort, keysort, and merge also may appear.
Nail requires the first argument (first two arguments for merge) to be
bound; normally the last is free.
ENVIRONMENTAL
Operator declarations with "op" and "statistics" may appear in Nail programs.
OTHER PROLOG CATEGORIES
None of these may appear in Nail programs, but some may be available
interactively, primarily for debugging use.
∂23-Jun-85 1810 @SU-CSLI.ARPA:LANSKY@SRI-AI.ARPA REMINDER: Monday Planlunch
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 23 Jun 85 18:10:12 PDT
Received: from SRI-AI.ARPA by SU-CSLI.ARPA with TCP; Sun 23 Jun 85 18:05:15-PDT
Date: Sun 23 Jun 85 18:05:27-PDT
From: LANSKY@SRI-AI.ARPA
Subject: REMINDER: Monday Planlunch
To: aic-associates@SRI-AI.ARPA, folks@SU-CSLI.ARPA
PLANNING AS DEBUGGING
Reid Simmons -- MIT AI Lab / SPAR
11:00 am, Monday, June 24
Room EJ232, SRI International
We are currently building a domain independent planner which can
represent and reason about fairly complex domains. The first part of
the talk will focus on the representations used and the rationale for
choosing them. The planner uses explicit temporal representations,
based on time points and the notion of "histories". It also extends
the traditional precondition/postcondition representation of actions
to include quantification, conditionals and the ability to reason
about cumulative changes.
The second part of the talk will focus on techniques to organize
and control the search for a plan. We view planning as "debugging
a blank sheet of paper". We correct a bug (ie. unachieved goal) by
changing one of the underlying assumptions in the plan which are
responsible for the bug. This problem solving approach combines
backtracking with traditional planning techniques, giving the planner
the potential for finding a solution with much less search. We also
present a simple, but effective, technique for choosing which plan
modification to pursue, based on maintaining a complete goal structure
of the plan.
This planner has been partially implemented and tested on
traditional blocks-world and register-transfer examples. It is
currently being applied to the problem of geologic interpretation and
to diagnosis of chip manufacturing problems.
-------
∂23-Jun-85 2008 @SU-SCORE.ARPA:udi@wisc-rsch.arpa FFT in parallel
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 23 Jun 85 20:08:16 PDT
Received: from wisc-rsch.arpa by SU-SCORE.ARPA with TCP; Sun 23 Jun 85 20:07:29-PDT
Received: by wisc-rsch.arpa; Sun, 23 Jun 85 21:53:45 cdt
Received: from wisc-crys.arpa by wisc-rsch.arpa; Sun, 23 Jun 85 19:36:27 cdt
Received: from csnet-relay.arpa by wisc-crys.arpa; Sun, 23 Jun 85 19:36:09 cdt
Received: from ubc by csnet-relay.csnet id aa02645; 23 Jun 85 20:36 EDT
Date: Sun, 23 Jun 85 17:17:59 pdt
Received: by ubc.csnet id AA07031; Sun, 23 Jun 85 17:17:59 pdt
From: Gilles Brassard <brassard%iro.udem.cdn%ubc.csnet@csnet-relay.arpa>
To: theory@WISC-CRYS.ARPA
Message-Id: <31:brassard@iro.udem.cdn>
Subject: FFT in parallel
Sender: udi@wisc-rsch.arpa
Resent-From: udi@wisc-rsch.arpa (Udi Manber)
Resent-To: theory:;@wisc-rsch.arpa
I would appreciate any pointers/information on how fast and how
the discrete fourier transform can be computed in parallel
(both from a theoretical and a practical point of view)
Thank you in advance,
Gilles Brassard
brassard%iro.udem.cdn@ubc on CSnet,
brassard%iro.udem.cdn%ubc@csnet-relay on ARPAnet.
∂24-Jun-85 0040 RESTIVO@SU-SCORE.ARPA PROLOG Digest V3 #27
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 24 Jun 85 00:40:03 PDT
Date: Saturday, June 22, 1985 2:04PM
From: Chuck Restivo (The Moderator) <PROLOG-REQUEST@SU-SCORE.ARPA>
Reply-to: PROLOG@SU-SCORE.ARPA
US-Mail: P.O. Box 4584, Stanford CA 94305
Phone: (415) 326-5550
Subject: PROLOG Digest V3 #27
To: PROLOG@SU-SCORE.ARPA
PROLOG Digest Monday, 24 Jun 1985 Volume 3 : Issue 27
Today's Topics:
Query - KBS/DBS Coupling
Implementations - Thanks & Gripe & WPE
----------------------------------------------------------------------
Date: Fri, 14-Jun-85 08:24:23 PDT
From: (IJme Schilstra) oce-rd1!oce-rd2!ysc@Seismo
Subject: KBS/DBS Coupling
I'd like to know about research (as well as commercial
products, if there are) dealing with the coupling of
knowledge base systems (particularly those written in
Prolog) with relational data base systems. (I've access
to INGRES on a VAX 11/750 under UNIX BSD 4.1, but I'm
interested in other DBMS's being used too.) Please mail
me:
(1) Commercial products
(a) Manufacturer
(b) Product description
(c) Application area
(d) Experiences
(2) Research
(a) Project description
(b) Relational DBMS being used
(c) Prolog version being used
(or more general: knowledge
base system implementation)
(d) Application area (if any)
(e) Experiences
Thanks very much,
-- IJme Schilstra
------------------------------
Date: Mon, 10-Jun-85 16:21:10 PDT
From: (Michael O'Hair) tektronix!michaelo@UCB-Vax
Subject: Thanks you
I want to thank all those who took the time to offer
assistance. My main problem stems from making an assumption
about a system library. There was no rules defined for
'append' so the interpreter balked. Once I put in a
set of rules, things work fine. Spoiled by 'C', I guess.
:-)
The next question is: what is the preferred way of
developing a system library and implementing an "includes"
file?
As always, any and all information is appreciated.
Thanks again.
-- Michael O'Hair
------------------------------
Date: Sun 16 Jun 85 16:34:38-PDT
From: Pereira@SRI-AI
Subject: Bugs, more bugs...
If anyone doubted the urgent need for a Prolog standard and
test suite, the increase of messages of the form ``The
Prolog system I got from Dr. Il Logical is behaving
strangely and my students are crying over their terminals''
is a clear indication of the lack of any accepted means to
test the performance and reliability of Prolog systems.
I fear that people may be turned away from Prolog, not
because of its REAL limitations (which are inevitable in
any engineering compromise) but because of contact with
unreliable or slow implementations that do not do justice
to the language. The Prolog community must put its house
in order, by establishing an accepted standard to have a
means to distinguish between non-problems (``Slimylog blows
up without warning when reading a term with more that 27
subterms'') and real problems (``what is the best approach
to add an equational component to Horn clauses'').
The best standards (the only possible ones...?) are those
based on successful current practice. Standard committees
should not engage in design tasks that are better left to
researchers. Instead, they should look in a non-partisan
manner at the actual practice (what systems are most
successful: satisfied customers, big applications, performance,
time in the field; what features are used by programmers,
what do they miss) and build a standard that helps
education and programming by setting a framework to
evaluate existing systems and new extensions. It is a
somewhat unglamorous task, hammering a standard that will not
include all those exciting new ideas that we may be
implementing next week. But it needs to be done.
-- Fernando Pereira
------------------------------
Date: Thu, 13 Jun 85 17:41:46 -0100
From: Burkhard Neidecker <neideck%germany@csnet-relay>
Subject: WPE variable handling
I would like to answer the question roused by David M.
Meyer in Vol3#26 concerning the handling of temporal vari-
ables in the WPE. The example given by him contained
several misconceptions about the handling of variables in
the WPE. We have implemented the WPE in C here at Karlsruhe,
so we have some insight into the concepts that guided War-
ren.
The first misconception is that a variable is no longer
accessed after its last occurrence in a clause. In general
it is still accessible for all subsequent goals (subgoals as
well as continuation) via its reference put into a register
or heap slot:
a) p1 :- p2(T).
b) p2(P) :- p3(...),p4(P),...
Clause a) is just
put←variable X2,A1 % Put T
execute p2/1
The head of b) translates into
get←variable Yn,A1
which internally is simply an assignment. Suppose, the
internal operation of put←variable Xn,Ai would be
Ai := Xn := ref←to(Ai)
as suggested by David Mayer, then Yn in b) would receive a
reference to A1 while executing get←variable Yn,A1. Only
heaven knows what the contents of A1 will be after execution
of all the subgoals of p3. Thus p4 would be called with a
wrong parameter. To avoid such clashes without using a glo-
bal register allocator, you have to use a distinct memory
location for variables instead of a register reference.
The second misconception is a variable classification
different from that proposed by Warren in his SRI TR 309
paper. I assume, the use of Z in the head was a typing error
(otherwise, Z would be permanent), so the clause should
perhaps really have been:
f(X,Y,←) :- h(Q), g(Y,Z).
David Meyer classified Q as temporary, while according to
Warren it should be permanent:
"A temporary variable is a variable that (1) has its
first occurrence in the head or in a structure or in
the last goal, and (2) that does not occur in more than
goal in the body, where the head is counted as part of
the first goal.
The head is considered part of the first goal only in
respect to condition (2)! Q neither occurs first in the
head nor in in a structure, thus it is classified a *per-
manent* variable.
Warrens apparently intricate classification scheme is
driven by two different considerations:
(1) All variables supporting "dataflow" between different
goals in the body must be located in an environment
because the registers might be overwritten by subgoal
execution.
(2) Whenever possible, space is allocated on the local
stack where it can be subject to tail recursion optimi-
zation.
If a variable occurs first in a body goal, you *have* to
allocate a cell for it (see above). If it occurs in a struc-
ture, you will need a global cell anyway, so there is no
extra penalty. In the last goal, you don't have any choice
as on which stack to allocate the cell, since at this point
your environment (if any) will have become deallocated. Thus
you are forced to get a global cell which is exactly what
the put←variable Xn,Ai instruction does.
As long as you are in the middle of the body, you may
allocate your storage from the local stack, retain the allo-
cated cells throughout the very next call and then deallo-
cate them (together with all subgoals which might have
references to that cells) by virtue of the following *call*
or *deallocate* instruction.
Thus the correct code for the given clause (redundant
instructions included) should be:
f(X,Y,←) :- h(Q), g(Y,Z).
allocate
/* get←variable X1,A1 */ % X
get←variable Y1,A2 % Y
/* get←variable X3,A3 */ % ←
put←variable Y2,A1 % Q : allocate
% local cell and
call h/1,2 % keep it for this
% call
put←value Y1,A1 % ***
put←variable X3,A2 % Since
% environment
% becomes
deallocate % deallocated
execute g/2
*** David Mayer used put←unsafe←value
here, which is unnecessary
Upon deterministic return from h/1, the cell for Q can be
safely reclaimed.
-- Burkhard Neidecker
------------------------------
End of PROLOG Digest
********************
∂24-Jun-85 1449 DAVIES@SUMEX-AIM.ARPA No meeting this week, unless ...
Received: from SUMEX-AIM.ARPA by SU-AI.ARPA with TCP; 24 Jun 85 14:49:05 PDT
Date: Mon, 24 Jun 1985 14:46 PDT
Message-ID: <DAVIES.12121776445.BABYL@Sumex>
From: DAVIES@Sumex
To: AAP@Sumex
Subject: No meeting this week, unless ...
cc: Davies@Sumex
Several people won't be around for a meeting this week. Please let me
know if you think there should be a meeting anyway.
Is there any interest in having someone from David Cheriton's group
give a talk over here? They have lots of experience in distributed
computing using a network of 20 Suns.
-- Byron
∂24-Jun-85 2233 @SU-SCORE.ARPA:udi@wisc-rsch.arpa PODC 85 Deadline for Hotel Reservations
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 24 Jun 85 22:33:09 PDT
Received: from wisc-rsch.arpa by SU-SCORE.ARPA with TCP; Mon 24 Jun 85 22:29:26-PDT
Received: by wisc-rsch.arpa; Tue, 25 Jun 85 00:07:22 cdt
Message-Id: <8506250230.AA09205@wisc-rsch.arpa>
Received: from csnet-relay.arpa by wisc-rsch.arpa; Mon, 24 Jun 85 21:30:44 cdt
Received: from ibm-sj by csnet-relay.csnet id ad02338; 24 Jun 85 22:22 EDT
Date: 24 Jun 1985 17:04:41-PDT (Monday)
From: Ray Strong <strong%ibm-sj.csnet@csnet-relay.arpa>
To: udi@wisc-rsch.ARPA
Subject: PODC 85 Deadline for Hotel Reservations
Sender: udi@wisc-rsch.arpa
Resent-From: udi@wisc-rsch.arpa (Udi Manber)
Resent-To: theory:;@wisc-rsch.arpa
HOTEL RESERVATION FOR PODC85
Please make Hotel reservations for the Fourth ACM Symposium
on Principles of Distributed Computing by July 1 (using
the telephone number for Minaki Lodge if necessary).
For further information, please call
"Pat Donovan", Department of Computer Science, University of
Waterloo, (519)885-1211, or send electronic mail to
mamalcolm@waterloo.csnet or pjdonovan@waterloo.csnet.
Hotel Registration Form
A block of rooms has been reserved (until July 1,1985) for
conference participants. If you wish to reserve one of these rooms,
please complete the form below and return it to:
Minaki Lodge
A Radisson Resort and
Conference Centre
Minaki, Ontario
Canada P0X lJ0
If you phone Minaki Lodge, (807) 224-4000, mention that you will be
attending the Symposium on Principles of Distributed Computing.
RESERVATIONS MUST BE RECEIVED BY JULY 1, 1985
Accommodations will not be confirmed without a cheque for the 1st
night's deposit, or use your American Express, Diners Club, VISA,
or Mastercard number to guarantee your reservation. You will be
charged for the lst night if reservations are not cancelled 7 days
prior to arrival.
Canadian U.S.
Rate: single occupancy $79.20/night $63.36/night
double occupancy $89.10/night $71.28/night
Prices do not include 5% sales tax or 10% service charge.
Name: ←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←
Affiliation:←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←
Address:←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←
City:←←←←←←←←←←←←←←←←←←←←State/Province:←←←←←←←←←←←←←
Zip/Postal Code:←←←←←←←←←Country:←←←←←←←←←←←←←←←←←←←←
Phone number:(←←←)←←←←←←←←← No.of Persons←←←←←←←←←←←←
Arrival: Time:←←←←←←←←←←←←Date:←←←←←←←←←←←←←←←←←←←
Departure: Time:←←←←←←←←←←←←Date:←←←←←←←←←←←←←←←←←←←
( ) check enclosed ( ) VISA ( ) MASTERCARD
( ) AMERICAN EXPRESS ( ) Diners' Club
Card No.←←←←←←←←←←←Exp.Date←←←←←←←←
Signature←←←←←←←←←←←←←←←←←←←←←←←←←←
====================================================================
ADVANCED REGISTRATION FORM
Please use this form or a facsimile to pre-register. Advanced
registration closes July 29, 1985. Registration after July 29 or at
the conference site is subject to a late fee. Please mail your
completed form with cheque (drawn on a North American bank) or
international money order (in Canadian or U.S. funds) payable to
"Symposium on PODC-85" to:
PODC'85
c/o Pat Donovan
Computer Science Dept.
Univeristy of Waterloo
Waterloo, Ontario
CANADA N2L 3G1
The rates for registration are listed below.
Requests for refunds will be honoured until July 29,1985.
Name: ←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←
Affiliation:←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←
Address:←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←
City:←←←←←←←←←←←←←←←←←←←←State/Province:←←←←←←←←←←←←←
Zip/Postal Code:←←←←←←←←←Country:←←←←←←←←←←←←←←←←←←←←
Phone number:(←←←)←←←←←←←←←
Please circle appropriate fees
CDN U.S.
ACM/SIG MEMBER $175 $133
(Membership #←←←←←←←←←←←←←←)
NON-MEMBER $225 $171
STUDENT $ 75 $ 57
Non-student late fee $ 50 $ 40
Student late fee $ 15 $ 12
Additional Banquet tickets ←←←←X $ 30 $ 23
Total enclosed: $←←←←←←←←←←←←←←U.S. $←←←←←←←←←←←←←CDN
∂25-Jun-85 0111 RESTIVO@SU-SCORE.ARPA PROLOG Digest V3 #28
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 25 Jun 85 01:11:29 PDT
Date: Sunday, June 23, 1985 10:16AM
From: Chuck Restivo (The Moderator) <PROLOG-REQUEST@SU-SCORE.ARPA>
Reply-to: PROLOG@SU-SCORE.ARPA
US-Mail: P.O. Box 4584, Stanford CA 94305
Phone: (415) 326-5550
Subject: PROLOG Digest V3 #28
To: PROLOG@SU-SCORE.ARPA
PROLOG Digest Tuesday, 25 Jun 1985 Volume 3 : Issue 28
Today's Topics:
Announcement - 1985 Symposium on LP & Call for Papers,
LP Library - Book Review
----------------------------------------------------------------------
Date: Thu, 13 Jun 85 14:59:02 EDT
From: Doug DeGroot <Degroot.yktvmv%ibm-sj@csnet-relay>
Subject: 1985 Symposium on Logic Programming
It has been brought to our attention that we failed to mention
in the SLP 85 registration forms appearing in a previous Digest
issue that registrants should note on their registration forms
which tutorials, if any, they plan on attending. If you have
already mailed in your registration form, please mail a notice
to the IEEE Computer Society in care of Mrs. Gerry Katz indicating
to her your choice of tutorials. If you have not yet registered
but intend to do so, please indicate on your registration form
your choice of tutorials.
Thank you, and our apologies for the inconvenience.
-- Doug DeGroot
------------------------------
Date: Tue, 11 Jun 85 10:04:20 edt
From: (Larry Kerschberg) allegra!usceast!kersch@Berkeley
Subject: Call for Papers
Call for Papers and Participation
FIRST INTERNATIONAL CONFERENCE ON EXPERT DATABASE SYSTEMS
April 1-4, 1986, Charleston, South Carolina
Sponsored by:
--------- --
The Institute of Information Management, Technology and
Policy College of Business Administration,
University of South Carolina
In Cooperation With:
-- ----------- ----
American Association for Artificial Intelligence
Association for Computing Machinery -- SIGMOD,
SIGART and SIGPLAN IEEE Technical Committee on
Data Base Engineering Agence de l'Informatique,
France
Conference Objectives
---------- ----------
The goal of this conference is to explore both the
theoretical and practical issues of Expert Database
Systems. These systems represent the confluence of
R&D activities in Artificial Intelligence, Logic,
and Database Management.
Expert Database Systems will play an ever-increasing
role in scientific, governmental and business
applications by:
o providing intelligent, knowledge-based access to
large shared databases through novel
user-interfaces and natural-language question-answering
facilities,
o endowing database systems with reasoning, planning, and
justification capabilities,
o creating knowledge base management tools and techniques
to support the creation, manipulation, indexing, and
evolution of large knowledge bases, and
o integrating AI & DB functional requirements into new
hardware and software environments for the specification,
prototyping, testing and debugging of knowledge-based
applications.
In order to foster the cross-fertilization of ideas from
AI, Logic, and Databases the Conference will be
composed of tutorial sessions, paper sessions, and panel
discussions.
Topics of Interest
------ -- --------
The Program Committee invites original papers (of
approximately 5000 words) addressing (but not limited to)
the following areas:
Theory of Knowledge Bases (including knowledge
representation, knowledge models, recursive data models,
object-oriented models, knowledge indexing and
transformation),
Knowledge Engineering (including acquisition, maintenance,
learning, knowledge-directed database specification and
design methodologies, and case studies),
Knowledge Base Management (including architectures and
languages, constraint and rule management, metadata
management, and extensible data dictionaries),
Reasoning on Large Data/Knowledge Bases (including inexact
and fuzzy reasoning, non-monotonic reasoning, deductive
databases, logic based query languages, semantic
query optimization, and constraint-directed reasoning),
Natural Language Interaction (including question-answering,
extended responses, cooperative behavior, explanation and
justification),
Intelligent Database Interfaces (including expert system
-- database communication, knowledge gateways,
knowledgeable user agents, browsers, and videotex),
Knowledge-Based Environments (including Decision Support
Systems, CAD/CAM, and VLSI Design),
Organizational Issues (including technology transfer,
procurement of expert database systems, and knowledge
certification).
Please send five (5) copies of papers by September 12,
1985 to:
Larry Kerschberg, Program Chairman
College of Business Administration
University of South Carolina
Columbia, SC, 29208
Program Committee
------- ---------
Hideo Aiso Sham Navathe
Keio University University of Florida
---- ---------- ---------- -- -------
Antonio Albano Erich Neuhold
University of Pisa Technical University of Vienna
---------- -- ---- --------- ---------- -- ------
Robert Balzer S. Ohsuga
USC/Information Sciences Institute University of Tokyo
--- ----------- -------- --------- ---------- -- -----
James Bezdek D. Stott Parker, Jr.
University of South Carolina UCLA and Silogic
---------- -- ----- -------- ---- --- -------
Ronald J. Brachman Alain Pirotte
Schlumberger Palo Alto Research Philips Research Lab, Brussels
------------ ---- ---- -------- ------- -------- --- --------
Michael Brodie Harry Pople
Computer Corporation of America University of Pittsburgh
-------- ----------- -- ------- ---------- -- ----------
Peter Buneman Erik Sandewall
University of Pennsylvania Linkoping University
---------- -- ------------ --------- ----------
Mark Fox Edgar H. Sibley
Robotics Institute, Carnegie-Mellon George Mason University
-------- --------- -------- ------ ------ ----- ----------
Antonio L. Furtado John Miles Smith
PUC -- Rio de Janeiro Computer Corp. of America
--- --- -- ------- -------- ----- -- -------
Herve Gallaire Reid Smith
ECRC, Munich Schlumberger-Doll Research
---- ------ ------------ ---- --------
Georges Gardarin Michael Stonebraker
Univ. of Paris 6 and INRIA UC -- Berkeley
---- -- ----- - --- ----- -- --------
Matthias Jarke Jeffrey Ullman
New York University Stanford University
--- ---- ---------- -------- ----------
Jonathan King Bonnie L. Webber
Teknowledge University of Pennsylvania
----------- ---------- -- ------------
Robert Kowalski Andrew B. Whinston
Imperial College Purdue University
-------- ------- ------ ----------
Jack Minker Gio Wiederhold
University of Maryland Stanford University
---------- -- -------- -------- ----------
Michele Missikoff Carlo Zaniolo
IASI-CNR, Rome MCC Corporation
---- --- ---- --- -----------
John Mylopoulos
University of Toronto
---------- -- -------
Important Dates
-----------------------------------------------
| Submission Deadline: September 12, 1985|
| Acceptance Notification: November 25, 1985 |
| Final Version Due: January 10, 1986 |
| Conference: April 1-4, 1986 |
-----------------------------------------------
Conference proceedings will be available at the conference,
and subsequently will appear in book form.
Conference General Chairman Conference Coordinator
---------- ------- -------- ---------- -----------
Donald A. Marchand Cathie L. Hughes
Institute of Information Management, Technology and Policy
Panel Coordinator Conference Treasurer
----- ----------- ---------- ---------
Arun Sen Libby Shropshier
Dept. of Management Science Institute of Information
College of Business Administration Technology and Policy
Univ. of South Carolina Univ. of South Carolina
Columbia, SC 29208 Columbia, SC 29208
Publicity Chairman Tutorial Chairman
--------- -------- -------- --------
John Weitzel Jonathan King
Dept. of Management Science Teknowledge, Inc.
College of Business Administration 525 University Avenue
Univ. of South Carolina Palo Alto, CA 94301
Columbia, SC 29208
International Representatives
Latin America Europe Far East
----- ------- ------ --- ----
Claudio M.O. Moura Jean-Claude Rault Masahiro Nakazawa
Independent Consultant Agence de l'InformatiqueNihon Digital Equip.
Rua R. Eduardo Guinle 60 Tour Fiat-Cedex 16 Sunlight Bldg.
Botafogo Paris-La Defense 5-29-1, Toyotamakita,
22.260 Rio de Janeiro, RJParis Nerima-ku Tokyo, 176
Brazil France Japan
--------------------------------------------------------------------
Response Card (To be on our mailing list, please detach, fill out,
and mail to address below)
Name Telephone
------------------------------------------- --------
Organization
-----------------------------------------------------
Address
----------------------------------------------------------
City, State,
ZIP, and Country
--------------------------------------------------
Please check all that apply:
I intend to submit a paper.
-----
Subject of paper
----------------------------------------------
---------------------------------------------------------------
I intend to attend the Conference.
-----
I would be interested in tutorial(s) on.
-----
Artificial Intelligence.
-----
Expert Systems.
-----
Database Management.
-----
Logic and Databases.
-----
Not sure I can participate, but please keep me informed.
-----
Cathie Hughes
EDS Conference Coordinator
Institute of Information Management
Technology and Policy
College of Business Administration
University of South Carolina
Columbia, SC 29208
------------------------------
Date: 22 Jun 85 1842 PDT
From: Yoni Malachi <YM@SU-AI.ARPA>
LOGIC PROGRAMMING: RELATIONS, FUNCTIONS, AND EQUATIONS
Doug DeGroot
Gary Lindstrom
Editors
Prentice-Hall, Inc.
Publication date: Summer 1985
June 14, 1985
1. Concept
This book addresses the topical and rapidly developing
areas of logic, functional, and equational programming, with
special emphasis on their relationships and prospects for
fruitful amalgamation. A distinguished set of researchers
have contributed fourteen articles addressing this field
from a wide variety of perspectives. The book will be
approximately 500 pages, published in hard cover form, with
foreword by the editors and combined index.
2. Table of Contents
2.1. Setting the Stage
- Uday Reddy: On the Relationship between Logic and
Functional Languages (34 pp.).
An essential distinction between logic and functional
languages is drawn based on input-output directionality.
Functional languages are directional in that programs
written in them make an explicit commitment about which
quantities are inputs and which are outputs. Logic
programs to do not make such a commitment. However, the
non-directionality makes the operational behavior of
logic programs hard to understand and poses problems in
developing parallel implementations of logic languages.
We present here a notation for introducing
directionality information in logic programs and show
that directional logic programs are equivalent to
first-order functional programs. Finally, we discuss
how the syntax and semantics of functional languages can
be extended to capture the additional expressive power
of logic languages.
- J. Darlington, A.J. Field, and H. Pull: The Unification
of Functional and Logic Languages (34 pp.).
Certain aspects of logic programs give them greater
expressive power than functional programs. In
particular, the presence of logical variables enable
relations to be used in arbitrary modes and allow
components of data structure templates to be bound by
unification. However, functional programs can also be
more expressive than logic programs because of their
cleaner syntax, which is devoid of output variables, and
because of their higher-order capabilities. The run
time behaviour of functional programs is much simpler to
control than that of logic programs, particularly in a
parallel context. Techniques, such as graph reduction
and data flow, have been evolved for the parallel
evaluation of functional languages taking advantage of
their simplicity of execution and it would be
advantageous if these techniques could also be used to
support languages with the extra expressive capability
of logic. This paper discusses the differences between
the two styles and proposes an extended functional
language (HOPE with unification) which provides all the
expressive power of logic programs whilst retaining most
of the underlying functional simplicity. This is
achieved by the introduction of absolute set abstraction
which allows logical variables to be introduced within
set expressions on the right hand sides of
function-defining equations.
We proposes a technique for compiling certain types
of functions defined implicitly within set expressions
into explicit functions. This effectively involves
synthesising function inverses using the processes of
symbolic unification and program transformation. When
this can be achieved, logical variables can be
eliminated altogether and replaced by function
composition.
2.2. Unification and Functional Programming
- Harvey Abramson: A Prological Definition of HASL, a
Purely Functional Language with Unification Based
Conditional Binding Expressions (57 pp.).
We present a definition in Prolog of a new purely
functional (applicative) language HASL (HArvey's Static
Language). HASL is a descendant of Turner's SASL, but,
among other features, introduces a one-way unification
based conditional binding expression, one-way in the
sense that of the two expressions being unified, only
one may contain variables. This one-way unification
based conditional binding construct is used to structure
the design of the compilation of HASL clausal
definitions to combinators which may then be reduced.
The specification of HASL and its reduction machine is
entirely in Prolog, thus providing an executable
specification - implementation - of the language. Since
HASL programs may be considered a syntactic sugaring of
combinator code, we adapt techniques derived from
compiling practice to specify the language and its
reduction machine. The definition is divided into four
parts. The first part defines the lexical structure of
the language by means of a simple Definite Clause
Grammar which relates character strings to "token"
strings. The second part defines the syntactic
structure of the language by means of a more complex
Definite Clause Grammar and relates token strings to a
parse tree. The third part is semantic in nature and
translates the parse tree definitions and expressions to
a variable-free string of combinators and global names.
The fourth part of the definition consists of a set of
Prolog predicates which specify how strings of
combinators and global names are reduced to "values",
i.e. integers, truth values, characters, lists,
functions, fail, and has an operational flavour: one can
think of this fourth part as the definition of a normal
order reduction machine.
- M. Sato and T. Sakurai: QUTE: a Functional Language
Based on Unification (24 pp.).
A new programming language called Qute is introduced.
Qute is a functional programming language which permits
parallel evaluation. While most functional programming
languages use pattern matching as basic variable-value
binding mechanism, Qute uses unification as its binding
mechanism. Since unification is bidirectional, as
opposed to pattern match which is unidirectional, Qute
becomes a more powerful functional programming language
than most of existing functional languages. This
approach enables the natural unification of logic
programming language and functional programming
language. In Qute it is possible to write a program
which is very much like one written in conventional
logic programming language, say, Prolog. At the same
time, it is possible to write a Qute program which looks
like an ML (which is a functional language) program. A
Qute program can be evaluated in parallel
(and-parallelism) and the same result is obtained
irrespective of the particular order of evaluation.
This is guaranteed by the Church-Rosser property enjoyed
by the evaluation algorithm. A completely formal
semantics of Qute is given in this paper.
- P.A. Subrahmanyam and J.-H. You: FUNLOG: a
Computational Model Integrating Logic Programming and
Functional Programming (42 pp.).
Funlog involves a computational model which integrate
functional programming and logic programming. This
model is described, along with evaluation strategies to
support the execution of programs based upon it. A lazy
mechanism, pattern-driven reduction, is developed for
the underlying functional model and cleanly and
naturally achieves reduction by need. The notion of
semantic unification is discussed. Semantic unification
serves as a basis for achieving the desired integration
of functions and logic, and can be used to replace the
conventional unification procedure in logic programming
systems. The resulting model supports computations with
infinite data structures while avoiding the introduction
of complicated control issues at the user level. In
addition, it provides the programmer the flexibility of
choosing between a backtracking free computation
framework and a conventional logic computation
framework, i.e., a nondeterministic one involving
backtracking. The use of this facility is illustrated
via examples. The model can be extended to include the
notion of equality when complete E-unification
algorithms are used.
2.3. Symmetric Combinations
- R. Barbuti, M. Bellia, G. Levi, and M. Martelli:
LEAF: a Language which Integrates Logic, Equations and
Functions (33 pp.).
The paper describes a language which integrates a
declarative language, consisting of Horn clauses and
equational theories with constructors, and a first order
functional language. Both language components will
eventually be directly supported by a hardware machine.
The declarative component permits the definition of both
relations and functions, possibly dealing with infinite
data structures. A formal semantics, coping with the
novel language features, is given in the standard style
(operational, model-theoretic and fixpoint). The
functional (procedural) component is essentially the
functional (deterministic) sublanguage of the
declarative one. It has a lazy evaluation based
parallel interpreter and allows efficient programming of
system software, tools and algorithms. The technique
for integrating these two language components is based
on using the procedural component as the metalanguage of
the declarative component, thus allowing procedural
programs to act on meta-objects such as declarative
theories and (possibly infinite) sets of solutions of
declarative programs. Examples are given of tools for
the declarative component and of integrated
procedural-declarative applications.
- Shimon Cohen: The APPLOG Language (38 pp.).
The virtues of PROLOG and LISP are discussed, and the
conclusion is reached that a mixture of these two
languages is desirable. Toward that end, APPLOG, a
combination of LISP and PROLOG, is described. APPLOG is
embedded within the PROLOG language with the facilities
of PROLOG made available through a simple goal function.
APPLOG is an applicative language, i.e. one whose
primary composition method is the application of
functions to arguments. Variables in APPLOG are
compatible with PROLOG variables, and serve as means for
data transfer between APPLOG and PROLOG. APPLOG
supports lambda and nlambda function definitions and
one-to-one, one-to-many and mixed binding mechanisms, in
the manner of INTERLISP. The main advantage of APPLOG
is the simple integration of LISP and PROLOG into one
powerful language which incorporates, in our judgment,
the best features of both languages. In particular,
APPLOG has the following advantages over traditional
LISP languages: (i) pattern directed invocation; (ii)
call by reference; (iii) an interface to PROLOG as a
database query language; (iv) functions as operators
(infix, prefix and postfix); (v) backtracking, and (vi)
generators. APPLOG includes aggregates and grouping
constructs, and has been extended to a simple relational
database query language similar to Query-By-Example. An
appendix includes a listing of the principal functions
of the APPLOG interpreter.
2.4. Programming with Equality
- Wm. Kornfeld: Equality for Prolog (15 pp.).
An extension of the language Prolog, called
"Prolog-with- Equality", is presented which allows the
inclusion of assertions about equality. When an attempt
is made to unify two terms that do not unify
syntactically, an equality theorem may be used to prove
the two terms equal. If it is possible to prove that
the two terms are equal, the unification succeeds with
the variable bindings introduced by the equality proof.
It is shown that this mechanism significantly improves
the power of Prolog. Sophisticated data abstraction
with the advantages of object-oriented programming
becomes available. Techniques for passing partially
instantiated data are described that extend the
"multi-use" capabilities of the language, improve the
efficiency of some programs, and allow the
implementation of arithmetic relations that are both
general and efficient. The modification to standard
Prolog are simple and straightforward and in addition
the computational overhead for the extra linguistic
power is believed to not be significant. Equality
theorems will probably play an important role in future
logic programming systems.
- Joseph Goguen and Jose Meseguer: EQLOG: Equality,
Types, and Generic Modules for Logic Programming (69 pp).
This paper presents Eqlog, a logic programming
language that unifies (predicate-based) Horn clause
relational programming with (equality-based) functional
programming. The unification seems quite natural, since
it is based on the smallest logic having both Horn
clauses and equality, namely Horn clause logic with
equality. Rather than force functions to be viewed as
predicates, or predicates to be viewed as functions,
Eqlog allows using both functions and predicates as
convenient. Furthermore, Eqlog computes differently
with functions than with predicates: functions are
computed by term rewriting, while queries to predicates
are computed in the usual Prolog-like fashion with
unification and backtracking (but with unification
modulo equations, implemented by narrowing in general,
and by more efficient methods for built-in types). In
fact, narrowing does much more than this, since it
allows solving for the values of logical variables that
occur in equations. This gives Eqlog very powerful
capabilities as a "constraint language", using narrowing
and Prolog-style backtracking to solve constraints over
user-defined abstract data types, and using efficient
built-in algorithms for solving constraints over
built-in types (such as numbers). With the usual
Church-Rosser assumptions, Eqlog's operational semantics
is logically complete. In effect, this paper shows how
to expand the paradigm of logic programming with a
number of features that are prominent in current
programming methodology, without any sacrifice of
logical rigor. Beyond simple functional programming,
Eqlog also provides strong typing, user definable
abstract data types, and generic (i.e., "parameterized")
modules, using methods developed for the specification
language Clear. A very useful refinement is "subsorts,"
which provide polymorphic operators and an inheritance
relation on types of data. We show that our logic
admits "initial models" having properties that
generalize those of minimal Herbrand models for ordinary
Horn clause logic. All of these features are given a
rigorous semantics in terms of the underlying logic, and
illustrated with a number of examples, including the
well-known Missionaries and Cannibals problem and some
natural language processing problems.
- Y. Malachi, Z. Manna and R. Waldinger: TABLOG: a New
Approach to Logic Programming (30 pp.).
TABLOG is a logic programming language based on
first-order predicate logic with equality that combines
relational and functional programming. In addition to
featuring both the advantages of functional notation and
the power of unification as a binding mechanism, TABLOG
also supports a more general subset of standard
first-order logic than do PROLOG and most other logic
programming languages.
The Manna-Waldinger deductive tableau proof system is
employed as an interpreter for TABLOG in the same way
that PROLOG uses an SLD resolution proof system.
Unification is used by TABLOG to match a call with a
line in the program and to bind arguments. The basic
rules of deduction used for computing are nonclausal
resolution and an equality rule that can be viewed as a
generalization of narrowing and paramodulation.
In this article we describe the basic features of
TABLOG and its (implemented) sequential interpreter and
we discuss some of its properties. We give examples of
programs for which TABLOG is better than a functional
language like LISP and others for which it is better
than a relational language like PROLOG.
2.5. Augmented Unification
- Robert G. Bandes (deceased): Constraining-Unification
and the Programming Language Unicorn (14 pp.).
Up to this point, direct implementations of axiomatic
or equational specifications have been limited because
the implementation mechanisms are incapable of capturing
the full semantics of the specifications. The
programming language Unicorn was designed and
implemented with the intention of exploring the full
potential of programming with equations. Unicorn
introduces a new language mechanism, called
constraining-unification. When coupled with semantic
unification, constraining-unification closely models the
semantics of equational specifications thereby allowing
for the implementation of a wider class of
specifications. Unlike the language mechanisms of
rewrite-rule and logic programming,
constraining-unification is free of order dependencies.
The same results are produced regardless of the order in
which the axioms are stated. The use of viewpoints
contributes to the flexibility of the Unicorn language.
Preconditions for partial operations can be specified
without added machinery.
- Ken Kahn: Uniform -- A Language Based upon Unification
which Unfies (much of) Lisp, Prolog, and Act 1 (28 pp.).
Uniform is an AI programming language based upon
augmented unification. It is an attempt to combine, in
a simple coherent framework, the most important features
of Lisp, actor languages such as Act 1 and SmallTalk,
and logic programming languages such as Prolog. Among
the unusual abilities of the language is its ability to
use the same program as a function, an inverse function,
a predicate, a pattern, or a generator. All of these
uses can be performed upon concrete, symbolic, and
partially instantiated data. Uniform features automatic
inheritance from multiple super classes, facilities for
the manipulation of programs, and a unification-oriented
database.
2.6. Semantic Foundations
- Joxan Jaffar, Jean-Louis Lassez and Michael J. Maher:
A Logic Programming Language Scheme (27 pp.).
Numerous extended versions of PROLOG are now
emerging. In order to provide greater versatility and
expressive power, some versions allow functional
programming features; others allow infinite data
structres. However, there is concern that such
languages may have little connection left with logic.
In some instances, various logic frameworks have been
proposed to solve this problem. Nevertheless, the
crucial point has not been addressed: the preservation
of the unique semantic properties of logic programs.
The significance of our effort here is twofold: (1)
There is a natural logic programming language scheme
wherein these properties hold. (2) Formal foundations
for extended versions of traditional PROLOG can be
obtained as instances of this scheme. They
automatically enjoy its properties.
- Gert Smolka: Fresh: A Higher-Order Language with
Unification and Multiple Results (56 pp.).
This paper presents Fresh, a language that integrates
logic programming features into higher-order functional
programming. The language incorporates unification,
multiple results and a collection construct. Many
examples illustrate that these extensions of functional
programming are useful. We define an operational
semantics along the lines of Plotkin's structural
approach. The semantics is of intrinsic interest since
it covers backtracking and the collection construct. To
illustrate the conceptual similarities and differences
between logic and functional programming, we begin with
a purely functional core language and add first
unification and then backtracking. With each addition
we discuss the enhanced eloquence of the language and
the concomitant modifications to the semantics.
------------------------------
End of PROLOG Digest
********************
∂25-Jun-85 1452 SUSI@SU-CSLI.ARPA goodbye from Fodor
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 25 Jun 85 14:52:05 PDT
Date: Tue 25 Jun 85 14:42:06-PDT
From: Susi Parker <SUSI@SU-CSLI.ARPA>
Subject: goodbye from Fodor
To: folks@SU-CSLI.ARPA
Fodors are setting off for points east tomorrow. My thanks to
everyone here, for talking, for helping, for organizing things,
teaching me things, listening, ommenting, objecting, and etc.
With the exception of your beastly flu, I've had a great time
here this spring, and it's your fault that I'm sad that I have
to leave now.
Bet wishes to all,
Janet
P.S. I'll be at: Linguistics Dept.
University of Connecticut U-145
Storrs, Conn.06268
Or on the machine: LINQUIST%UCONNVM.BITNET@BERKELEY
Telephone: 203-486-4229
Come and visit, if you're travelling our way.
-------
∂25-Jun-85 1636 @SU-CSLI.ARPA:PCOHEN@SRI-AI.ARPA when shall we meet?
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 25 Jun 85 16:36:17 PDT
Received: from SRI-AI.ARPA by SU-CSLI.ARPA with TCP; Tue 25 Jun 85 16:27:52-PDT
Date: Tue 25 Jun 85 16:29:25-PDT
From: Phil Cohen <PCOHEN@SRI-AI.ARPA>
Subject: when shall we meet?
To: NL4: ;
For the occasional meetings over the summer, when can you meet?
I can't be avail Mond. or Thurs. AM's. How 'bout you?
Phil
-------
∂26-Jun-85 1013 SCHMIDT@SUMEX-AIM.ARPA chfinger updated
Received: from SUMEX-AIM.ARPA by SU-AI.ARPA with TCP; 26 Jun 85 10:13:00 PDT
Date: Wed 26 Jun 85 10:11:20-PDT
From: Christopher Schmidt <SCHMIDT@SUMEX-AIM.ARPA>
Subject: chfinger updated
To: KSL-Lisp-Machines@SUMEX-AIM.ARPA
cc: Torres@SUMEX-AIM.ARPA
I added the new, MJH, address for HPP-3670-3 (GOLD) to the chfinger
PCL program on the 2060 (sample output below). I was impressed to see all
machines up. I edited the output below to make it readable. Would someone
fix up the namespace for the MJH machines? They still have RPG's name on
them, etc., which makes the unedited output much too long and unreadable.
--Christopher
P.S. Hey, that Iguana's a heavily used machine!
@dec pcl pcl:chfinger
[Command CHFINGER defined]
@chfinger
Schoen Eric Schoen HPP-3600-2 C-104 Whelan Bldg. 1536K
Rice Jim HPP-3670-4 Outside C-104, 1024K
Saraiya M Nakul Pratap Saraiya HPP-3600-5 13:48 Outside C-104, 1024K
- HPP-3600-6 9:46 Outside C-104, 2048K
HENAGER Don Henager SILVER Logic Group - 1024k
LISP-MACHINE GOLD 17:21 MJH 222 1536K
File Server IGUANA 144:52 MJH324 512k
File Server COAX 11:08 MJH360 512k
-------
∂26-Jun-85 1317 NISSENBAUM@SU-CSLI.ARPA Thursday, 2.15, seminars
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 26 Jun 85 13:17:45 PDT
Date: Wed 26 Jun 85 13:08:22-PDT
From: Helen Nissenbaum <NISSENBAUM@SU-CSLI.ARPA>
Subject: Thursday, 2.15, seminars
To: folks@SU-CSLI.ARPA
There will be only two scheduled 2:15, 'in-house', seminars this
quarter:
Pennti Kannerva (Thursday, 6/27)
Edit Doron (date to be announced)
For the rest of the summer, CSLI folk wishing to give, or arrange,
talks should go ahead and schedule them on their own. Contact Emma
Pease about getting these talks announced in the Newsletter.
Regularly scheduled seminars will resume (I trust) in the fall. Have
a good summer.
Helen Nissenbaum
-------
∂26-Jun-85 1743 EMMA@SU-CSLI.ARPA Newsletter June 27, No. 35
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 26 Jun 85 17:43:27 PDT
Date: Wed 26 Jun 85 17:18:01-PDT
From: Emma Pease <Emma@SU-CSLI.ARPA>
Subject: Newsletter June 27, No. 35
To: friends@SU-CSLI.ARPA
Tel: 497-3479
C S L I N E W S L E T T E R
←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←
June 27, 1985 Stanford Vol. 2, No. 35
←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←
A weekly publication of The Center for the Study of Language and
Information, Ventura Hall, Stanford University, Stanford, CA 94305
←←←←←←←←←←←←
CSLI ACTIVITIES FOR *THIS* THURSDAY, June 27, 1985
12 noon TINLunch
Ventura Hall ``The Algebra of Events''
Conference Room by Emmon Bach
Discussion led by Edit Doron
2:15 p.m. CSLI Seminar
Redwood Hall ``An Organism and Its Internal Model of the World''
Room G-19 Pentti Kanerva, CSLI
Discussion led by Alex Pentland
3:30 p.m. Tea
Ventura Hall
4:15 p.m. CSLI Colloquium
Redwood Hall ``Qualitative Process Theory''
Room G-19 Ken Forbus, University of Illinois, Computer Science
←←←←←←←←←←←
ANNOUNCEMENT
Next Thursday, July 4, is a National Holiday and no activities will
take place.
!
Page 2 CSLI Newsletter June 27, 1985
←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←
AREA P1 MEETING: PIXELS AND PREDICATES
``Visual Communication for Severely Impaired Aphasics''
Richard D. Steele, Rehab R&D Center, Palo Alto VA Hospital
Wednesday, July 3, 11:00 a.m., Ventura Hall
The study to be presented concerns the progress of a single,
globally aphasic individual who has been trained on a computerized,
extended version of the visual communication system ``VIC'' originally
developed and tested by Baker (1975), Gardner (1976), and their
colleagues. The VIC system is currently implemented on a Mactinosh XL
computer. The goal has been to produce a device that combines lexical
and grammatical richness with ease of use and practical utility. After
one year of work, results show that:
a) Errors favor telegraphic communications,
b) Prepositions and word order present the greatest difficulty,
c) The patient has learned to comprehend and construct both simple
phrases and complex communications including possessives and conjoined
constructions,
d) Performance is better for reception than for production
performance,
e) The patient readily activates the translation device and
initiates communications in situationally appropriate contexts.
A video tape will be shown of the patient using the computerized VIC
system.
←←←←←←←←←←←←←
CSLI TALK
``A Situational Theory of Analogy''
Todd Davies, Stanford Philosophy Dept.
Ventura Conference Room
Monday, July 1, 1985, 1:15 p.m.
Analogy in logic is generally given the form:
P(A)&Q(A)
and P(B) are premises
---------
therefore Q(B)
can be concluded, where P is a property or set of properties held by
the analogous situation A in common with the present situation B, and
where Q is a property which is initially held to be true of A. The
question is: What justifies the conclusion? Sometimes the conclusion
is clearly bogus, but for other pairs of situations and properties it
seems quite plausible. I will give examples of both intuitively good
and intuitively bad analogies as a way to argue that theories of
analogy hitherto proposed have yet to answer this question, and that
the rationale for analogy which has been assumed for most early work
on analogy in AI -- namely, that the inference is good if and only if
the situations being compared are similar enough -- is inadequate. I
will also point to traditional logic's inadequacies as a formal
language for analogy and develop a theory which incorporates ideas
from (and finds its easiest expression in) the theory of situations of
Barwise and Perry. The theory suggests a general means by which
computers can infer conclusions about problems which have analogues
for which the solution is known, when failing to inspect the analogue
would make such an inference impossible.
The discussion following will be led by David H. Helman, Case
Western Reserve University and CSLI
!
Page 3 CSLI Newsletter June 27, 1985
←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←
CSLI VISITING SCHOLARS
This is a brief summary of the Visiting Scholars who have already
arrived at the Center for the summer.
1) Kimmo Koskenniemi has been here since the first of June and will be
with us until the first of August. He is currently with the
University of Helsinki and is working on morphological analysis with
Martin Kay, Ron Kaplan, and Lauri Karttunen.
2) Dorit Abusch arrived two weeks ago and will stay until the middle
of October. She will then return to Israel to teach at the University
of Tel Aviv. John Perry is her sponsor.
3) Yves Lesperance has been at the Center since June 10 and will leave
at the end of August. He is from the University of Toronto Computer
Science Department, and his sponsor is David Israel. Among other
things he will participate in the CSLI Summer School.
4) David Helman, who is from the Center for Automation and Intelligent
Systems at Case Western Reserve University, came in late May and will
be here through the first week in August.
5) Manfred Pinkal has been here since the first of the month and will
be here through the end of August. He is from the University of
Duesseldorf, and his sponsor is Hans Uszkoreit.
Thus far, we know of at least four more visitors who will be with us
in the next month.
1) Irene Guessarian arrives on June 30 and will stay for approximately
four weeks. She will collaborate on a paper with Jose Meseguer.
Professor Guessarian is from the National Center for Scientific
Research in France.
2) Peter Mosses will arrive during the second week in July and will be
here for a month. Dr. Mosses is currently at Aarhus University in the
Computer Science Department. His sponsor is Joseph Goguen, and his
specialty is denotational semantics.
3) Luis Monteiro will come on the first of July and will be here for
two months. He is from the Universidade Nova de Lisboa, and his
sponsor is Fernando Pereira.
4) Harald Ganzinger of Dortmund University will be here starting July
19 and will stay until the third of August. Professor Ganzinger's
sponsor is Jose Meseguer. --Dave Brown
-------
∂27-Jun-85 1131 EMMA@SU-CSLI.ARPA Ventura Copying machine
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 27 Jun 85 11:31:39 PDT
Date: Thu 27 Jun 85 10:48:34-PDT
From: Emma Pease <Emma@SU-CSLI.ARPA>
Subject: Ventura Copying machine
To: bboard@SU-CSLI.ARPA, folks@SU-CSLI.ARPA
Tel: 497-3479
The next time you walk into Ventura 7 to use the copying machine,
you will notice that you need an auditron. DON'T PANIC, if you don't
have one. The person at the receptionist's desk (Suzy or someone else)
will. You will be asked a few questions (name and purpose of the
copying) for survey purposes only. As usual, personal copies are five
cents.
If you plan to do any copying after 10 p.m. on weekdays and between
5 p.m. and 9 a.m. on weekends, you may make special arrangements with
the receptionist or other staff members.
-------
∂27-Jun-85 1203 EMMA@SU-CSLI.ARPA Ant spraying
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 27 Jun 85 12:03:14 PDT
Date: Thu 27 Jun 85 11:57:02-PDT
From: Emma Pease <Emma@SU-CSLI.ARPA>
Subject: Ant spraying
To: bboard@SU-CSLI.ARPA, folks@SU-CSLI.ARPA
Tel: 497-3479
Those of you who have been suffering from ants, relief is in sight.
The ant man comes tomorrow at 3:30 and will be spraying in the offices
which are having ant problems. The spray is harmless to humans though
it has an unpleasant odor so you may want to leave for that reason.
-------
∂27-Jun-85 1218 EMMA@SU-CSLI.ARPA Ant spraying
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 27 Jun 85 12:18:02 PDT
Date: Thu 27 Jun 85 12:12:34-PDT
From: Emma Pease <Emma@SU-CSLI.ARPA>
Subject: Ant spraying
To: folks@SU-CSLI.ARPA, bboard@SU-CSLI.ARPA
Tel: 497-3479
Offices that have NOT been reported as having ant problems will not be
sprayed.
p.s. Jamie is in charge of organizing the spraying.
-------
∂27-Jun-85 1228 @SU-CSLI.ARPA:LANSKY@SRI-AI.ARPA Next week's PLANLUNCH -- WEDNESDAY July 3
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 27 Jun 85 12:27:50 PDT
Received: from SRI-AI.ARPA by SU-CSLI.ARPA with TCP; Thu 27 Jun 85 12:21:43-PDT
Date: Thu 27 Jun 85 12:21:03-PDT
From: LANSKY@SRI-AI.ARPA
Subject: Next week's PLANLUNCH -- WEDNESDAY July 3
To: aic-associates@SRI-AI.ARPA, folks@SU-CSLI.ARPA
"HOW TO CLEAR A BLOCK"
or
Unsolved Problems in the Blocks World #17
Richard Waldinger -- SRI AI Center
11:00 am, WEDNESDAY, July 3 (notice change in day)
Room EJ232, SRI International
ABSTRACT:
Apparently simple problems in the blocks world get more complicated
when we look at them closely. Take the problem of clearing a block.
In general, it requires forming conditionals and loops and even
strengthening the specifications; no planner has solved it.
We consider how such problems might be approached by bending a
theorem prover a little bit.
-------
∂27-Jun-85 1701 SCHMIDT@SUMEX-AIM.ARPA new SUMEX staff for LispM's
Received: from SUMEX-AIM.ARPA by SU-AI.ARPA with TCP; 27 Jun 85 17:01:43 PDT
Date: Thu 27 Jun 85 17:00:45-PDT
From: Christopher Schmidt <SCHMIDT@SUMEX-AIM.ARPA>
Subject: new SUMEX staff for LispM's
To: KSL-Lisp-Machines@SUMEX-AIM.ARPA
cc: Acuff@SUMEX-AIM.ARPA, Gotelli@SU-SCORE.ARPA
This is to advise you that Richard Acuff has today joined the
SUMEX staff as our lightening rod for LispM affairs. (For the past 3
years I have nominally had that duty, but in the last year Gordon
Foyster and then Jerry Yan, being closer to the front-lines, have been
doing the real work of keeping things going.)
Richard comes from Ohio State where he supported D-machines
and a DECsystem20 and vaxen, but LispMs are new to him, so I hope
you'll give him the same help in coming up to speed therewith that you
gave me when I was new to the D-machines. You'll probably see him
around the LispMs or in his new office in A-1105D.
So dust off those problem-reports/feature-requests I've been
no help with and give him something to do!
--Christopher
P.S. Obviously it will take him a day or two to completely absorb
all those Symbolics manuals, so maybe you should save some of those
reports/requests for next week! <:-)
-------
∂28-Jun-85 0943 CLAY@SU-CSLI.ARPA Welcome Peter King
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 28 Jun 85 09:42:53 PDT
Date: Fri 28 Jun 85 09:37:15-PDT
From: Clay Andres <CLAY@SU-CSLI.ARPA>
Subject: Welcome Peter King
To: folks@SU-CSLI.ARPA
A hardy welcome to Peter King. He will be our system administrator for UNIX
and Tops-20. Peter has worked for LOTS for most of his undergraduate career and
is a senior History major.
Clay
-------
∂28-Jun-85 1110 WRIGHT@SUMEX-AIM.ARPA SIGLunch, Friday, June 28
Received: from SUMEX-AIM.ARPA by SU-AI.ARPA with TCP; 28 Jun 85 10:51:05 PDT
Date: Fri 28 Jun 85 10:43:35-PDT
From: Carol Wright/Susie Barnes <WRIGHT@SUMEX-AIM.ARPA>
Subject: SIGLunch, Friday, June 28
To: siglunch@SUMEX-AIM.ARPA
My apologies if you did not receive this notice the first time it was
sent. Here it is:
SIGLUNCH
DATE: Friday, June 28, 1985
LOCATION: Chemistry Gazebo, between Physical &
Organic Chemistry
TIME: 12:05
SPEAKER: Ken Forbus
Xerox
TITLE: Qualitative Process Theory
Things move, collide, flow, bend, stretch, break, cool down, heat up,
and boil. Intuitively we think of the things that cause changes in
physical situations as processes. Qualitative Process Theory defines
simple notions of quantity, function, and process that allow interesting
common-sense inferences to be drawn about dynamical systems. This talk
will describe the basics of the Qualitative Process Theory, illustrate
how it can be used to capture certain aspects of different models of
physical phenomena and shed light on causal reasoning. Current research
directions and potential applications will also be discussed.
-------
-------
∂28-Jun-85 1141 WRIGHT@SUMEX-AIM.ARPA Siglunch: Friday, June 28
Received: from SUMEX-AIM.ARPA by SU-AI.ARPA with TCP; 28 Jun 85 11:40:48 PDT
Date: Fri 28 Jun 85 11:37:10-PDT
From: Carol Wright/Susie Barnes <WRIGHT@SUMEX-AIM.ARPA>
Subject: Siglunch: Friday, June 28
To: siglunch@SUMEX-AIM.ARPA
My apologies if you did not receive this message the first time
it was sent! Here it is again.
SIGLUNCH
DATE: Friday, June 28, 1985
LOCATION: Chemistry Gazebo, between Physical &
Organic Chemistry
TIME: 12:05
SPEAKER: Ken Forbus
Xerox
TITLE: Qualitative Process Theory
Things move, collide, flow, bend, stretch, break, cool down, heat up,
and boil. Intuitively we think of the things that cause changes in
physical situations as processes. Qualitative Process Theory defines
simple notions of quantity, function, and process that allow interesting
common-sense inferences to be drawn about dynamical systems. This talk
will describe the basics of the Qualitative Process Theory, illustrate
how it can be used to capture certain aspects of different models of
physical phenomena and shed light on causal reasoning. Current research
directions and potential applications will also be discussed.
-------
-------
∂28-Jun-85 1219 @SU-SCORE.ARPA:welch@ames-vmsb.ARPA SIGBIG
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 28 Jun 85 12:19:31 PDT
Received: from ames-vmsb.ARPA by SU-SCORE.ARPA with TCP; Fri 28 Jun 85 12:11:07-PDT
Date: 28 Jun 85 11:15:00 PDT
From: welch@ames-vmsb.ARPA
Subject: SIGBIG
To: super@su-score.arpa
Reply-To: welch@ames-vmsb.ARPA
ASSOCIATION FOR COMPUTING MACHINERY
San Francisco Golden Gate Chapter
"SIGBIG" Special Interest Committee
For Large High Speed Computers
Meetings on the first Wednesday of each month at 7:30 PM. Speakers
who can give insights to various aspects of SUPERCOMPUTING are
featured each month.
Next meeting:
Wednesday, July 3,1985, 7:30 PM
Speaker: Bill Joy/Sun
Subject: Unix and Supercomputers
Location: Wheeler Hall Room 30
University California at Berkeley
---------------------------------------------------------------
Future Meetings:
August 7, 1985 Steve Perrenod/CRI CRAY 2
---------------------------------------------------------------
Tape-recordings of most of the previous may be obtained
in exchange for a tape cassette or $5.00 by contacting:
Mary Fowler (415)261-4058
Supercomputing #192, BOX 2787
Alameda, CA. 94501-0787
For information contact Mary Fowler, Chairperson (415) 261-4058
or Mike Austin, Publ. Chair (415) 423-8446
------
∂28-Jun-85 1312 ullman@diablo goodbye
Received: from SU-AIMVAX.ARPA by SU-AI.ARPA with TCP; 28 Jun 85 13:12:15 PDT
Date: Fri, 28 Jun 85 13:07:04 pdt
From: Jeff Ullman <ullman@diablo>
Subject: goodbye
To: nail@diablo
I'm leaving tomorrow on vacation.
Therefore, I'm not calling any meetings until Wednesday 7/24
at the usual time and place.
At that time, I'd like to "finalize" source and intermediate code.
∂28-Jun-85 1413 @SU-SCORE.ARPA:welch@ames-vmsb.ARPA SIGBIG
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 28 Jun 85 14:13:11 PDT
Received: from ames-vmsb.ARPA by SU-SCORE.ARPA with TCP; Fri 28 Jun 85 14:05:33-PDT
Date: 28 Jun 85 11:15:00 PDT
From: welch@ames-vmsb.ARPA
Subject: SIGBIG
To: super@su-score.arpa
Reply-To: welch@ames-vmsb.ARPA
ASSOCIATION FOR COMPUTING MACHINERY
San Francisco Golden Gate Chapter
"SIGBIG" Special Interest Committee
For Large High Speed Computers
Meetings on the first Wednesday of each month at 7:30 PM. Speakers
who can give insights to various aspects of SUPERCOMPUTING are
featured each month.
Next meeting:
Wednesday, July 3,1985, 7:30 PM
Speaker: Bill Joy/Sun
Subject: Unix and Supercomputers
Location: Wheeler Hall Room 30
University California at Berkeley
---------------------------------------------------------------
Future Meetings:
August 7, 1985 Steve Perrenod/CRI CRAY 2
---------------------------------------------------------------
Tape-recordings of most of the previous may be obtained
in exchange for a tape cassette or $5.00 by contacting:
Mary Fowler (415)261-4058
Supercomputing #192, BOX 2787
Alameda, CA. 94501-0787
For information contact Mary Fowler, Chairperson (415) 261-4058
or Mike Austin, Publ. Chair (415) 423-8446
------
∂28-Jun-85 1511 @SU-SCORE.ARPA:avi.sjrlvm1%ibm-sj.csnet@csnet-relay.arpa BATS at IBM
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 28 Jun 85 15:11:36 PDT
Received: from csnet-relay by SU-SCORE.ARPA with TCP; Fri 28 Jun 85 15:09:58-PDT
Received: from ibm-sj by csnet-relay.csnet id aa08619; 28 Jun 85 18:07 EDT
Date: Fri, 28 Jun 85 14:49:19 PDT
From: Avi Wigderson <avi%ibm-sj.csnet@csnet-relay.arpa>
To: aflb.all@su-score.ARPA
Subject: BATS at IBM
CC: theory-b@ucb-vax.ARPA
Hi
IBM will host the next BATS, which we plan
for Thursday, August 1. There is still time,
but make sure you don't miss it just because
you are in the middle of writing your FOCS papers.
Avi
∂28-Jun-85 1542 @SU-SCORE.ARPA:JF@SU-SUSHI.ARPA August 1
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 28 Jun 85 15:42:45 PDT
Received: from SU-SUSHI.ARPA by SU-SCORE.ARPA with TCP; Fri 28 Jun 85 15:37:41-PDT
Date: Fri 28 Jun 85 15:24:56-PDT
From: Joan Feigenbaum <JF@SU-SUSHI.ARPA>
Subject: August 1
To: aflb.su@SU-SCORE.ARPA
cc: avi%ibm-sj.csnet@CSNET-RELAY.ARPA
Apparently, the next BATS will be at San Jose on August 1. I will be out
of town and hence won't be able to arrange carpools from Stanford. If you
are willing to take on this minor responsibility, please let me know.
More importantly, if you would like to be the Stanford speaker, let me know.
Thanks,
Joan
-------
∂28-Jun-85 1723 JOHN@SU-CSLI.ARPA Course on Heidegger and Artificial Intelligence
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 28 Jun 85 17:23:20 PDT
Date: Fri 28 Jun 85 17:17:50-PDT
From: John Perry <JOHN@SU-CSLI.ARPA>
Subject: Course on Heidegger and Artificial Intelligence
To: folks@SU-CSLI.ARPA
A course of possible interest to some CSLI folks is being offered
this Summer by Dagfinn Follesdal:
Philosophy 133S. Heidegger's Being and Time. Graduate Students
Register for 233S. Lectures and seminar disucssion of Heidegger's
Being and Time, with special attention to his notion of "background"
("horizon"), and its bearing on recent work on artficial intelligence
by Terry Winograd and others.
Hubert Dreyfus of UC Berkeley is planning to participate more or
less regularly, I am told.
Course work, I am told, will not include any writing of programs.
--John Perry
-------
∂28-Jun-85 1724 JOHN@SU-CSLI.ARPA By the way
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 28 Jun 85 17:24:34 PDT
Date: Fri 28 Jun 85 17:18:50-PDT
From: John Perry <JOHN@SU-CSLI.ARPA>
Subject: By the way
To: folks@SU-CSLI.ARPA
By the way, the course on Heidegger meets Tuesdays and Thursdays,
10-12, in the Philosophy Dept. Seminar Room in Building 90.
-------
∂28-Jun-85 2206 BETSY@SU-CSLI.ARPA vacation plans
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 28 Jun 85 22:06:30 PDT
Date: Fri 28 Jun 85 22:02:13-PDT
From: Betsy Macken <BETSY@SU-CSLI.ARPA>
Subject: vacation plans
To: folks@SU-CSLI.ARPA
I'll be back-packing with my family next week, so would you be
sure to let Ingrid, Jamie, Joyce, or Clay know about anything
that can't wait until after July 8? Have a nice fourth of
July and I'll see you on the 8th.
Betsy
-------
∂01-Jul-85 0938 EMMA@SU-CSLI.ARPA Ants II: The Overlooked
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 1 Jul 85 09:38:09 PDT
Date: Mon 1 Jul 85 09:33:40-PDT
From: Emma Pease <Emma@SU-CSLI.ARPA>
Subject: Ants II: The Overlooked
To: bboard@SU-CSLI.ARPA, folks@SU-CSLI.ARPA
Tel: 497-3479
We are interested in knowing if anyone is still having trouble with
ants. Please send messages to Emma since Jamie will be going on
vacation this week.
-Emma
-------
∂01-Jul-85 1027 WEIN@SU-CSLI.ARPA Apartment needed for Visiting Scholar
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 1 Jul 85 10:26:55 PDT
Date: Mon 1 Jul 85 10:17:55-PDT
From: Gina Wein <WEIN@SU-CSLI.ARPA>
Subject: Apartment needed for Visiting Scholar
To: folks@SU-CSLI.ARPA
A visiting scholar from the Netherlands will be arriving in September
'85 with his wife and child for a year and will need a 2 bedroom
apartment. He would like one close to Stanford (naturally). Price
range: $750 - $1,000. If you become aware of anything that might
be suitable, please let me know. Gina
-------
∂02-Jul-85 1556 @SU-CSLI.ARPA:LANSKY@SRI-AI.ARPA planlunch reminder
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 2 Jul 85 15:56:09 PDT
Received: from SRI-AI.ARPA by SU-CSLI.ARPA with TCP; Tue 2 Jul 85 15:44:26-PDT
Date: Tue 2 Jul 85 15:42:32-PDT
From: LANSKY@SRI-AI.ARPA
Subject: planlunch reminder
To: aic-associates@SRI-AI.ARPA, folks@SU-CSLI.ARPA, rsimmons@SRI-KL.ARPA
I have decided to keep planlunch at the same time and location -- it
was not a conflict for too many people. Here, once again, is the abstract:
-------------------
"HOW TO CLEAR A BLOCK"
or
Unsolved Problems in the Blocks World #17
Richard Waldinger -- SRI AI Center
11:00 am, WEDNESDAY, July 3 (notice change in day)
Room EJ232, SRI International
ABSTRACT:
Apparently simple problems in the blocks world get more complicated
when we look at them closely. Take the problem of clearing a block.
In general, it requires forming conditionals and loops and even
strengthening the specifications; no planner has solved it.
We consider how such problems might be approached by bending a
theorem prover a little bit.
-------
∂03-Jul-85 0905 EMMA@SU-CSLI.ARPA Ants III:
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 3 Jul 85 09:05:20 PDT
Date: Wed 3 Jul 85 09:03:12-PDT
From: Emma Pease <Emma@SU-CSLI.ARPA>
Subject: Ants III:
To: folks@SU-CSLI.ARPA, bboard@SU-CSLI.ARPA
Tel: 497-3479
The following places are still reporting ants and will be sprayed
again. Please send me a message by 3 this afternoon, if you wish to
add or delete a place.
Trailer office - A-1
Trudy's office - E-5
Trailer rest rooms, kitchen, and seminar room
Offices G-1 to G-4 and H-1 to H-5 (computing trailer)
Ventura kitchen
Joyce's office
Ventura 24
Helen's office - Casita 41
Casita foyer
Emma
-------
∂03-Jul-85 0924 LB@SU-CSLI.ARPA Visitor Policy for 1985-86
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 3 Jul 85 09:24:32 PDT
Date: Wed 3 Jul 85 09:22:49-PDT
From: John Perry
Subject: Visitor Policy for 1985-86
Sender: LB@SU-CSLI.ARPA
To: Researchers@SU-CSLI.ARPA
cc: Betsy@SU-CSLI.ARPA, Brown@SU-CSLI.ARPA, Emma@SU-CSLI.ARPA,
Ingrid@SU-CSLI.ARPA, Jamie@SU-CSLI.ARPA, Joyce@SU-CSLI.ARPA
Reply-To: John
TEL: (415) 497-9007
Non-casual visitors to CSLI fall roughly into three groups:
(1) "Collaborators"; those whose visit is actively sought by a CSLI
researcher(s), for purposes of collaboration in some CSLI
project, or at any rate to work closely with such researcher(s).
Often travel money and other support is supplied from initiators
or area funds.
(2) "Sojourners"; those who are here mostly of their own volition,
finding it convenient to spend a sabbatical or other leave in
the midst of the resources offered here. Support is limited to
computer accounts and possibly shared office space.
(3) Special cases.
Visitors bring a great deal to CSLI, and independently of that we have
some obligation to share our good fortune. However, there are many
hidden and not so hidden costs. To keep things in balance, I intend
to adopt the following policies, mostly modest revisions of the
policies of the past.
Collaborators will be most welcome, of course. But:
a) We ask that the details of an impending visit, the duration, the
facilities needed, etc., be made available before the visit is
too impending. There are sudden opportunities, of course, but
in general a couple of weeks warning, via the forms available
from Bach-Hong, Elsie, or Jackie, seems reasonable. Of course,
the less the needs of the visitor, the less this matters.
b) We will allocate a limited amount of space at Ventura for
Collaborators; if this space is taken for the period in question,
by the time the details reach us, we may be unable to provide
any.
c) Please discuss the details of any proposed long-term visits
(much more than a month, say) with Betsy well in advance.
As to Sojourners, I think we would all agree that we are better off
with a few that we can treat pretty well, rather than a lot we don't
have time, space, or xerox paper for. So, we will let Sojourner
applications for Visiting Scholar status during a given academic year
(June to June) accumulate (except for special cases) until March 15 of
the previous year. Decisions will then be made on the basis of
projected available space and the credentials and projects of the
visitors. Thus Visiting Scholar status at CSLI will be somewhat more
meaningful, and somewhat less easy to obtain than the same status in
departments at Stanford. We will be nice to everyone, but only
committed to contributing to the space and computational needs of our
own Visiting Scholars. If you receive a request from someone wanting
to be a Sojourner, send it to Ingrid to be added to our applications
file.
Special cases. E.g., people whom we more actively encourage to come,
perhaps even providing some salary support or travel money. These
will be treated as special cases.
-------
∂03-Jul-85 1403 @SU-CSLI.ARPA:LANSKY@SRI-AI.ARPA NEXT WEEK'S PLANLUNCH -- Monday as usual
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 3 Jul 85 14:03:30 PDT
Received: from SRI-AI.ARPA by SU-CSLI.ARPA with TCP; Wed 3 Jul 85 13:52:46-PDT
Date: Wed 3 Jul 85 13:50:01-PDT
From: LANSKY@SRI-AI.ARPA
Subject: NEXT WEEK'S PLANLUNCH -- Monday as usual
To: aic-associates@SRI-AI.ARPA, folks@SU-CSLI.ARPA, rsimmons@SRI-KL.ARPA,
mugs@SUMEX-AIM.ARPA
Special Solutions to the Nonlinear Frame Problem
or
Unsolved Problems in the Blocks World #18
David Chapman
SRI AI Center and MIT AI Lab
11:00 AM, Monday, July 8
Room EJ232, SRI International
ABSTRACT:
The correctness of current nonlinear planners depends on so
restricting their action representation as to make them useless for
real-world planning. If the action representation is extended,
correctness, efficiency, or generality is lost. I will propose
planning techniques that sacrifice generality for correctness and
efficiency.
-------
∂03-Jul-85 1421 @SU-SCORE.ARPA:welch@ames-vmsb.ARPA SIGBIG!CHANGE!
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 3 Jul 85 14:21:31 PDT
Received: from ames-vmsb.ARPA by SU-SCORE.ARPA with TCP; Wed 3 Jul 85 14:02:50-PDT
Date: 3 Jul 85 13:14:00 PDT
From: welch@ames-vmsb.ARPA
Subject: SIGBIG!CHANGE!
To: super@su-score.arpa
Reply-To: welch@ames-vmsb.ARPA
ASSOCIATION FOR COMPUTING MACHINERY
San Francisco Golden Gate Chapter
"SIGBIG" Special Interest Committee
For Large High Speed Computers
NOTICE! INFORMATION PERTAINING TO CHANGE IN MEETING LOCATION !
Next meeting:
Wednesday, July 3,1985, 7:30 PM
Speaker: Bill Joy/Sun
Subject: Unix and Supercomputers
OLD- Location: Wheeler Hall Room 30
NEW- Location: 100 Lewis (On campus)
Orientation: Using East entrance to campus,
building is just past entrance and before kiosk.
University California at Berkeley
---------------------------------------------------------------
NOTICE! INFORMATION PERTAINING TO CHANGE IN MEETING LOCATION !
For information contact Mary Fowler, Chairperson (415) 261-4058
or Mike Austin, Publ. Chair (415) 423-8446
------
∂03-Jul-85 1740 EMMA@SU-CSLI.ARPA Newsletter July 4, No. 36
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 3 Jul 85 17:39:51 PDT
Date: Wed 3 Jul 85 16:58:09-PDT
From: Emma Pease <Emma@SU-CSLI.ARPA>
Subject: Newsletter July 4, No. 36
To: friends@SU-CSLI.ARPA
Tel: 497-3479
C S L I N E W S L E T T E R
←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←
July 4, 1985 Stanford Vol. 2, No. 36
←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←
A weekly publication of The Center for the Study of Language and
Information, Ventura Hall, Stanford University, Stanford, CA 94305
←←←←←←←←←←←←
ANNOUNCEMENT
This Thursday, July 4, is a National Holiday and no activities will
take place. No activities will take place on Thursday, July 11.
←←←←←←←←←←←←
AREA P1 MEETING: PIXELS AND PREDICATES
``The Bitmap as Reality''
Scott Kim, author of ``Inversions''
3:00 pm, Wednesday July 10, Ventura Seminar room
What would a truly graphic computer be like? Not just an iconic
veneer on an underlying textual representation, but rather a system in
which everything would be pictorial. ``Viewpoint'' is a computer
system that demonstrates my approach to this question. In Viewpoint
there is no underlying representation lurking behind the screen --
what you see is what you AND the computer get. As consequences of this
radical viewpoint, text and graphics can be treated homogeneously,
bitmap-oriented and structured drawing systems can be truly
integrated, and other graphic media such as paper and videodisks can
be incorporated in computer-based communication.
←←←←←←←←←←←←
AREA NL2 MEETING: INTERACTIONS OF MORPHOLOGY, SYNTAX, AND DISCOURSE
On June 26, the Working Group on Interactions of Morphology,
Syntax, and Discourse had its first meeting, and Bresnan and Mchombo's
``Agreement and Pronominal Incorporation in Chichewa'' was discussed.
This group will hold closed meetings throughout the Summer, but the
results of the research will be presented in the Fall, and summaries
will be printed in the Newsletter.
SUMMARY OF THE FIRST MEETING
In Bantu languages, there are highly systematic interactions
between word structure, word order, and discourse structure. Although
these interactions are well known among Bantuists, it is not
recognized in general that they pose deep problems for current
linguistic theory. These interactions are problematic because words,
phrases, and discourses are independent systems in their grammatical
form; yet, despite the autonomy of their structural formation, they
are functionally interdependent to a high degree. How then does
functional information flow between word, phrase, and discourse? The
answer to this question suggests a radically different conception of
the organization of linguistic information fromt that which has been
prevalent in generative grammar. --Joan Bresnan
-------
∂05-Jul-85 0852 CLAY@SU-CSLI.ARPA Prolog-20
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 5 Jul 85 08:52:46 PDT
Date: Fri 5 Jul 85 08:50:13-PDT
From: Clay Andres <CLAY@SU-CSLI.ARPA>
Subject: Prolog-20
To: folks@SU-CSLI.ARPA
Does anyone have a Prolog-20 manual that we can copy before monday?
Clay
-------
∂05-Jul-85 1428 SCHOEN@SUMEX-AIM.ARPA Simple, etc changes?
Received: from SUMEX-AIM.ARPA by SU-AI.ARPA with TCP; 5 Jul 85 14:28:08 PDT
Date: Fri 5 Jul 85 14:28:20-PDT
From: Eric Schoen <Schoen@SUMEX-AIM.ARPA>
Subject: Simple, etc changes?
To: aap@SUMEX-AIM.ARPA, acuff@SUMEX-AIM.ARPA
I've been encountering a bizarre problem while doing simulations the last
few days. When I call (GC-IMMEDIATELY) from within the simulation, I
crash the machine with "out of swap space" after about 15 minutes of doing
GC. Is there something going on in the lowest levels of the system that
I'm not aware of (that might conflict with running the collector)? If
we're sure there's nothing we're doing that would kill GC, I'll flame
vociferously to Symbolics.
Thanks,
Eric
-------
∂07-Jul-85 1716 @MIT-MC.ARPA:Laws@SRI-AI.ARPA Articles in AI Magazine
Received: from MIT-MC.ARPA by SU-AI.ARPA with TCP; 7 Jul 85 17:15:48 PDT
Received: from MIT-OZ by MIT-MC.ARPA via Chaosnet; 7 JUL 85 20:03:01 EDT
Received: from MIT-MC.ARPA by MIT-OZ via Chaosnet; 7 Jul 85 20:00-EDT
Received: from SRI-AI.ARPA by MIT-MC.ARPA.ARPA; 7 Jul 85 19:51:25 EDT
Date: Sun 7 Jul 85 16:45:40-PDT
From: Ken Laws <Laws@SRI-AI.ARPA>
Subject: Articles in AI Magazine
To: Phil-Sci@MIT-MC.ARPA
Members of this list may be interested in "A Biologist Looks at Cognitive
AI" by William K. Purves (Yale), AI Magazine, Summer 1985. Dr. Purves
draws some interesting analogies between Biology and AI (concluding that
AI is indeed a science).
Other interesting sections occur in Robert Lindsay's description of AI
work at U. Michigan, particularly the work of Scott, Kochen (both p. 67),
and Thagard (p. 71) on scientific theory formation.
-- Ken Laws
-------
∂07-Jul-85 2131 DAVIES@SUMEX-AIM.ARPA PARSYM -- new mailing list for Parallel Symbolic Computing
Received: from SUMEX-AIM.ARPA by SU-AI.ARPA with TCP; 7 Jul 85 21:31:28 PDT
Date: Sun, 7 Jul 1985 21:32 PDT
Message-ID: <DAVIES.12125258162.BABYL@Sumex>
From: DAVIES@Sumex
To: PAI@Sumex
Subject: PARSYM -- new mailing list for Parallel Symbolic Computing
cc: Davies@Sumex
PARSYM: A Netwide Mailing List for
Parallel Symbolic Computing
The PARSYM mailing list has been started to encourage communication
between individuals and groups involved in PARALLEL SYMBOLIC COMPUTING
(non-numeric computing using multiple processors). The moderator
encourages submissions relating either to parallelism in symbolic
computing or to the use of symbolic computing techniques (AI, objects,
logic programming, expert systems) in parallel computing. All manner
of communication is welcomed: project overviews, research results,
questions, answers, commentary, criticism, humor, opinions,
speculation, historical notes, or any combination thereof, as long as
it relates to the hardware, software, or application of parallel
symbolic computing.
To contribute, send mail to PARSYM@SUMEX (or PARSYM@SUMEX-AIM.ARPA, if
your mailer requires). To be added to the PARSYM distribution list,
or to make other editorial or administrative requests, send mail to
PARSYM-Request@SUMEX. When you are added to the PARSYM distribution
list, I will send you a welcoming message with additional information
about PARSYM and some necessary cautions about copyright and
technology export.
To get the list off the ground, I offer the following set of
discussion topics:
1. Will there be a general-purpose parallel symbolic processor, or
should parallel architectures always be specialized to particular
tasks?
2. The primary languages for sequential symbolic computing are Lisp,
Prolog, and SmallTalk. Which is a better basis for developing a
programming language for parallel computing? Do we need something
fundamentally different?
3. Sequential computing took about 30 years to reach its current
state. Thirty years ago, programming tools were nonexistent:
programmers spent their time cramming programs into a few hundred
memory cells, without programming languages or compilers or
symbolic debuggers. Now, sequential programming is in a highly
developed state: most programmers worry less about the
limitations of their hardware than about managing the
complexity of their applications and of their evolving computer
systems.
Today, parallel programming is where sequential programming was
thirty years ago: to optimize computation and communication,
programmers spend their time manually assigning processes to a few
processors, without benefit of programming languages or compilers
or symbolic debuggers that deal adequately with parallelism.
Will it take 30 years to bring parallel computing up to the current
level of serial computing?
Submissions, queries, and suggestions are equally welcome. Fire away!
PARSYM's Moderator,
Byron Davies (Davies@SUMEX)
∂08-Jul-85 0009 @SU-CSLI.ARPA:LANSKY@SRI-AI.ARPA PLANLUNCH reminder
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 8 Jul 85 00:08:49 PDT
Received: from SRI-AI.ARPA by SU-CSLI.ARPA with TCP; Mon 8 Jul 85 00:00:42-PDT
Date: Mon 8 Jul 85 00:00:34-PDT
From: LANSKY@SRI-AI.ARPA
Subject: PLANLUNCH reminder
To: aic-associates@SRI-AI.ARPA, folks@SU-CSLI.ARPA, mugs@SUMEX-AIM.ARPA
Special Solutions to the Nonlinear Frame Problem
or
Unsolved Problems in the Blocks World #18
David Chapman
SRI AI Center and MIT AI Lab
11:00 AM, Monday, July 8
Room EJ232, SRI International
ABSTRACT:
The correctness of current nonlinear planners depends on so
restricting their action representation as to make them useless for
real-world planning. If the action representation is extended,
correctness, efficiency, or generality is lost. I will propose
planning techniques that sacrifice generality for correctness and
efficiency.
-------
∂08-Jul-85 0038 RESTIVO@SU-SCORE.ARPA PROLOG Digest V3 #29
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 8 Jul 85 00:38:01 PDT
Date: Saturday, July 6, 1985 4:48PM
From: Chuck Restivo (The Moderator) <PROLOG-REQUEST@SU-SCORE.ARPA>
Reply-to: PROLOG@SU-SCORE.ARPA
US-Mail: P.O. Box 4584, Stanford CA 94305
Phone: (415) 326-5550
Subject: PROLOG Digest V3 #29
To: PROLOG@SU-SCORE.ARPA
PROLOG Digest Monday, 8 Jul 1985 Volume 3 : Issue 29
Today's Topics:
Query - Numerical Analysis,
Implementations - A Standard Syntax
----------------------------------------------------------------------
Date: Fri, 5 Jul 85 09:40 EST
From: D E Stevenson <DSteven%Clemson@csnet-relay>
Subject: Numerical analysis
Can anyone point me to serious attempts to do numerical
analysis programming in either logic or functional
programming environments. I know there have been a couple
of papers in Newton quadrature schemes.
Thank you,
-- Steve
------------------------------
Date: Wed, 3-Jul-85 10:01:38 PDT
From: Vantreeck%Logic.DEC@DECWRL.ARPA
Subject: A standard syntax for Prolog
F. Pereira pointed out that it is difficult to convert
between Edinburgh and Micro-Prolog. I think that problems
of conversion between implementations is not due to syntax.
The problem is due to differences in semantics which are
reflected by different syntaxs. Before we can begin
thinking about a standard syntax for Prolog, we need to
standardize the semantics. But implementers can't even agree
on what cut means.
-- George Van Treeck
------------------------------
Date: Wed, 3-Jul-85 06:07:25 PDT
From: (Ray Reeves) decvax!cca!emacs!Ray@UCB-Vax
Subject: Bugs, and more bugs...
I must gently rebuke Fernando for using pejorative
terms like "religous" and "confuse" when I make a
valid point. Micro-Prolog uses one data type
- lists - for aggregates, instead of distinguishing
lists from structures. It is not confused about
that, and if it leads to better meta-programming
it *is* a good excuse for doing so. Of course, when
translating DEC-10 syntax to lisp syntax it cannot
maintain that distinction, but that is only important
if it is required to translate back again. It so
happens that micro-Prolog does now support DEC-10
syntax so it can't be such a big deal.
The points that lisp syntax is unsuitable for a
reference language and that it makes compilation
more difficult seem perfectly valid, although we
might observe that the Symbolics Prolog compiler has
made a pretty good job of it.
-- Ray Reeves,
------------------------------
Date: Sun, 30-Jun-85 10:54:58 PDT
From: (PEREIRA) sri-iu!PEREIRA@Seismo.ARPA
Subject: Bugs, and more bugs...
I am not about to enter in religious discussions over
syntax, but it should be noted that micro-Prolog syntax
*loses* information with respect to Edinburgh syntax,
because it confuses lists with functor application. This
means that Edinburgh Prolog programs cannot in general be
converted to micro-Prolog syntax without major rewriting.
Micro-Prolog syntax cannot thus be used as a standard or
interchange format.
Making meta-level programming easier is not a good excuse to
confuse conceptually distinct datatypes, as well as making
compilation more difficult.
-- Fernando Pereira
------------------------------
Date: Fri, 28-Jun-85 08:05:06 PDT
From: (Ray Reeves) decvax!cca!emacs!ray@UCB-Vax
Subject: Bugs, and more bugs...
I would like to follow up Van Treeck's comments
avoiding, if I can, the patois of perversion.
His point about having a theory of modularisation
is a good one, but he may not be aware that
micro-Prolog has been a modular system since it
was first introduced in 1980.
On the other hand, his endorsement of systems that
modify text behind your back seems capricious.
What possible merit is there in constraining the
expressive power of the ASCII set of characters?
His plea for a more pedantic style of syntax is
misplaced. Any Prolog can be made to support a
special interface, what the Lisp style does is
represent a clause with minimum syntax, and the
fact that it can be entered that way is much
appreciated by those who don't care for typing.
Building spurious noise marks into the syntax has
the unfortunate consequence of preempting those
marks from use elsewhere.
The important point about Lisp syntax is that it
is more suitable for reflective programming. Not
only does it vitiate the need for "univ" but it
enables a term to be represented totally abstractly,
whereas reference to a structure needs some
knowledge of it's form. For example, "((X|Y)|Z)"
in micro-Prolog matches any clause. Meta-programming
in DEC-10 style is a much more clumsy business.
The idea of a "lambda" style syntax is also good, but
I suggest that lambda is a better word than "for←all",
which suggests universal quantification. Variables
that do not first appear in the head are not universally
quantified.
It seems to me that soft cuts under a disjunction are
very important. There is a distinct need for them there,
and the semantics of soft cut in that context does not
seem strange if the disjunction is thought of as a
separate anonymous clause set which has been made local
to share variable scope and generally increase efficiency.
That, after all, is what it is. The practise of treating
cuts under a disjunction as soft seems quite satisfactory.
-- Ray Reeves
------------------------------
Date: Tue, 25 Jun 85 17:12 EDT
From: Tim Finin <Tim%UPenn@csnet-relay>
Subject: Prolog Standards
If there are many implementations of Prolog which try
to adhere to a common standard, there will undoubtedly
be numerous small differences. One aid for coping with
this situation would be a Prolog "read time
conditionalization facility" similar to that found in
Common Lisp. In this note I'd like to ask people to
consider the inclusion of this idea as a standard Prolog
feature.
Common Lisp's #+ and #- readmacros
----------------------------------
In Common Lisp, there is a global variable *features*
which is bound to a list of symbols which describe various
aspects of the current Lisp. For example, its initial
value for the Lisp runnig on our Vax is:
(COMMON COMPILER DEBUGGER EDITOR VAX VMS)
which indicates that this is a COMMON lisp with the
COMPILER, DEBUGGER and EDITOR resident running on a
VAX under VMS. The user is allowed to add or
remove features as he likes.
In Lisp code one can use expressions like
#+(and VAX VMS (not FRANZ)) <s-expression>
The effect is that the Lisp reader will see <s-expression>
only if the features VAX and VMS are in the list and the
feature FRANZ is not. If the test is not satisfied, the
whole expression will be treated as whitespace.
Thus it is typical to see things like:
(defun system (S)
#+VMS (dcl-command S)
#+UNIX (shell-command S))
This has been extremely useful to the Lisp community,
allowing one to write code which is portable over a
number of similar implementations in the generic Common
Lisp family (e.g., Franz, PSL, Maclisp, DECLisp, Zetalisp,
etc.). The fact that this is done by the reader makes it
very powerful, and allows it to even cover differences at
the character stream level.
Can we have this in Prolog?
---------------------------
I believe that we need some thing like this in Prolog. I
am currently faced with a situation in which I would like
to maintain a single system that can run in both C-Prolog
and in Symbolics Prolog. The implementations two are
sufficiently close to imagine this. I think I can do it
if there were such a "read time conditionalization
facility" that is commonly supported by various Prolog
implementations. There are, of course, some problems due
the the nature of the Prolog reader, but I believe that
they can be dealt with.
I would like to get some feedback on this idea and perhaps
some concrete suggestions for its form.
-- Tim
------------------------------
Date: Friday, 28 Jun 1985 06:55:51-PDT
From: Vantreeck%Logic.DEC@decwrl
Subject: Bugs, more bugs...
A standard suite of tests would be nice. Maybe we can
get the DoD to "validate" Prologs also? - Just joking! I
think it would be particulaly nice if there were a standard
suite of tests which could also be used for benchmarking.
What does 20KLIPS mean if every one is using a different
formula to compute the LIPS in a naive reverse?
There are several people in Digital investigating
various implementations of Prolog. They also think that
current implementations of Prolog are severely lacking in
features that help produce reliable and maintainable
programs. But some experts think that there is so much good
work being done on the theoretical issues of how to extend
the language to make it more useful while remaining
strictly first order logic that to put Prolog into concrete
at this time would be not be productive.
For example, M-Prolog has implemented modules. Modular
programming is good thing. But was the modularity built on
a theory or simply for the need for modularity. Recent work
by Ken Bowen and Toby Weinberg provides a mechanism of
adding modules within a first order logic framework. As far
as I know, no current Prolog has implemented modules based
on theoretical work that indicates that it is consistent
with some system of logic.
Should we freeze the syntax and sematics of modules on
a current implementation, like M-Prolog, or wait for some
promising theoretical work to be completed? I think modules
should be implemented whether there's a theoretical
framework to support it or not. But if it appears that a
clean theoretical framework might be available, then it's
worth waiting a few months or even a year for that work to
finish before making a standard for modules.
I don't see any reason, except political, why we can't
come up with a standard syntax. I'm glad there's no
standard syntax, because I think the syntax of current
implementations suck! Variables should be predecared, as in
the proposed Meta-Prolog at Syracuse U., And variables
should not be case sensitive, e.g.,
for←all [x,y]: foo(x,y) if bar(x) and snafu(y).
There is a minor kink in predeclaring variables in a clause
that also asserts a clause containing variables. The kink
is that a standard should exist on how to deal with the
interpretation.
I think the parser should automatically uppercase (or
lowercase) all strings before being converted to atoms,
unless it was a quoted string of characters.
We should move the syntax of the language from
something cryptic (particularly bad are those Prologs with
LISP-like syntax) to something that is more natural to the
uninitiated. The ":-", ",", and ";" of DEC-10 and C-Prolog,
should be replaced with "IF", "AND", and "OR".
We need to standardize cut. In particular, should a
cut on disjuction (or) be a soft cut or hard cut? I vote
for the hard cut. I would rather see an exclusive or,
"XOR", predicate than the else-like predicate, "->", of
C-Prolog.
All Prolog error messages should be in a file(s)
seperate from the rest of the Prolog code. Prologs should
index to the error message, via a standardized error
number, instead of the messages being built into the bodys
of clauses.
DISCLAIMER:
These are my opinions, and are not necessarily the
opinions of my employer, Digital Equipment Corporation.
-- George Van Treeck
------------------------------
End of PROLOG Digest
********************
∂08-Jul-85 1638 EMMA@SU-CSLI.ARPA An Object Model of Information
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 8 Jul 85 16:38:27 PDT
Return-Path: <BEECH@SU-SCORE.ARPA>
Received: from SU-SCORE.ARPA by SU-CSLI.ARPA with TCP; Mon 8 Jul 85 15:57:11-PDT
Date: Mon 8 Jul 85 15:51:21-PDT
From: David Beech <BEECH@SU-SCORE.ARPA>
Subject: An Object Model of Information
To: Emma@SU-CSLI.ARPA
cc: Beech@SU-SCORE.ARPA
ReSent-Date: Mon 8 Jul 85 16:30:56-PDT
ReSent-From: Emma Pease <Emma@SU-CSLI.ARPA>
ReSent-To: friends@SU-CSLI.ARPA
I am giving a seminar this Friday which may be of interest to some
CSLI people - I should welcome some comments from the experts in this
kind of thing! Could you perhaps copy your CSLI distribution list?
The first Database Seminar of the summer quarter will be this Friday,
12th July, at 3:15 in MJH352.
For advance copies of the paper or further information, please contact
Beech@Score or call 497-9118.
TOWARDS AN OBJECT MODEL OF THE REPRESENTATION AND USE OF INFORMATION
David Beech
Stanford CIS and HP Laboratories
Future general-purpose information systems will need to deal with a
wide range of information, and offer flexible access to it, if they
are to appeal to the potential millions of non-specialist users.
For example, they should process pictures and sounds as naturally as
numbers and texts; they should answer questions which require some
deduction from the often incomplete information previously given to
the system; and they should move towards the support of natural
language interfaces, including spoken inputs.
An object-oriented model of the representation and use of information
is proposed, with the necessary generality for the description and
design of such systems. Fundamental concepts including those of
agent, object, type, action, formula, process, transaction, predicator
and generator are introduced. Recursive functions, predicate calculus,
and n-ary relations are brought together in a data abstraction framework,
with an emphasis on intensional definition of concepts and their
instantiation by means of predicators and generators.
-------
∂09-Jul-85 0042 RESTIVO@SU-SCORE.ARPA PROLOG Digest V3 #30
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 9 Jul 85 00:42:25 PDT
Date: Monday, July 8, 1985 7:57AM
From: Chuck Restivo (The Moderator) <PROLOG-REQUEST@SU-SCORE.ARPA>
Reply-to: PROLOG@SU-SCORE.ARPA
US-Mail: P.O. Box 4584, Stanford CA 94305
Phone: (415) 326-5550
Subject: PROLOG Digest V3 #30
To: PROLOG@SU-SCORE.ARPA
PROLOG Digest Tuesday, 9 Jul 1985 Volume 3 : Issue 30
Today's Topics:
Implementation - A Standard Syntax,
Announcement - DB System,
LP Library - PARLOG
----------------------------------------------------------------------
Date: Fri, 28-Jun-85 01:08:22 PDT
From: mcvax!ukc!icdoc!cdsm@Seismo.ARPA
Subject: Investigating implementations of cut
Implementations of CUT
As part of the BSI Prolog standardisation process I
am looking at the way in which cut is implemented in different
Prolog systems. The following program illustrates the
differences which have been found. I would be grateful if you
could run the program on your system and send me the answers;
if you have an interpreter and compiler please execute it for
both systems.
---------------------------------------------------------
/* Tests to distinguish various implementations of cut */
/* Chris Moss, Imperial College, June 1985 */
test1 :- do('Testing that cut is implemented). ', t1).
test2 :- do('Test if cut acts within disjunction', t2).
test3 :- do('Test if it cuts prev. choice within disjunction',t3).
test4 :- do('Test if cut acts when passed as metacall', t4).
test5 :- do('Test if & cut acts within metacall', t5 ).
test6 :- do('Test if cut acts through not', t6).
do(Message,Test) :- w(Message), Test.
do(Message,Test) :- w('Does act').
w(X) :- write(X), nl.
t :- w('Does not act').
t1 :- (true;w('Did not cut alternatives correctly'),fail),
!, w('Succeeds going forwards'), fail.
t1 :- w('Failed to cut goal').
t2 :- (!;w('Fails to cut disjoint alternatives')), fail.
t2 :- t.
t3 :- t3a(X),(!,fail;w('Fails to cut disjunction')).
t3 :- t.
t3a(!).
t3a(X) :- w('Did not cut alternatives'), fail.
t4 :- t3a(X), X, w('Ok going forwards'),fail.
t4 :- t.
t5 :- t5a(X), X, fail.
t5 :- t.
t5a((true,!)).
t6 :- not(not(!)), fail.
t6 :- t.
-------------------------------------------------
From initial tests we have the following results:
Implementation Test 1 2 3 4 5 6
DEC-10 Compiler Y Y Y Y Y Y
Waterloo, MU-Prolog Y Y Y Y Y N
DEC-10 int, C-Prolog Y Y Y N N N
POPLOG Y Y Y I I I
micro, Sigma Y N N Y Y N
where Y means did cut, N means did not cut and I means
that it was trapped as an illegal use and failed.
Note that in all cases 2=3 and 4=5; however in some
implementations this may not be the case. If there
are any other discriminating cases not covered above
I would be glad to hear of them.
------------------------------
Date: 05 Jul 85 13:45:05 +1000 (Fri)
From: John Shepherd <munnari!mungunni.oz!JAS@Seismo>
Subject: New Deductive Database System
A System for Very Large Deductive Databases
using a Superimposed Codeword Indexing Scheme
=====
This note is to announce the (near) availability of a
deductive database system suitable for dealing with very
large databases of Prolog rules. The indexing scheme used
by the system is based on the method of two-level
superimposed codewords as described in [1], which allows
partial match retrieval. Superimposed codeword schemes
provide a very efficient method of retrieving records from
large databases in only a small number of disk accesses.
Further, the access method can be tuned so that the ratio
of "false matches" can be reduced by an arbitrary amount
(with a corresponding increase in storage costs). Unlike
many earlier systems, this system supports the storage and
retrieval of completely general Prolog terms, including
functors and variables, and it is even possible to store
Prolog rules in the database.
The system is in the final stages of development under
Berkeley Unix (4.2BSD) and has already been interfaced to
the MU-Prolog system[2,3]; it will be incorporated into
release 3.2db of MU-Prolog which will be available soon
for Unix and VMS. It is being developed as part of the
Machine Intelligence Project at the University of
Melbourne on a Pyramid 90x which was loaned to the project
by Pyramid Technology in Australia. The figures given
below are taken from the Pyramid running version 2.3.1 of
the OSx operating system (in the Berkeley universe) with
one 400 Mb disk.
Preliminary tests, on a database of mail transfer pathways
through Usenet containing one million facts, have been
very encouraging. To store these facts, which have an
average length of 60 bytes, required just over 80Mb, which
means a storage overhead of about 30%. In the present
system, with a one-million record database indexed on
three attributes, the rate of insertion is six records per
CPU second. The rate of insertion could be significantly
increased if the system were run as a single-user batch-
type system without locking controls. Specifying just two
of the fields (each record contains four fields),
retrieved on average just 3 records for a query which had
only one correct answer. The system can achieve a record
retrieval rate of around 1000 records per second for a
query on highly clustered records, to about 50 records per
second for a query on unclustered records, even on this
large database; for smaller databases, even faster rates
are achievable. A query with complete information,
required on average 1.1 retrievals, and required 4 disk
accesses (excluding overheads from the Unix file system).
This system overcomes some of the limitations of the Unix
file system. For example, it overcomes the limit of
twenty open files per process by caching on Unix file
descriptors, thus allowing several database relations to
be accessed simultaneously. The system also provides data
buffering to reduce the number of file opens and data
reads. The processing time of logic programs such as the
"ancestor" relation can be minimised by this feature.
We would be interested in hearing from other groups who
are developing similar systems. For further information
on this system, contact Dr. K. Ramamohanarao (Rao) or John
Shepherd at the following addresses:
HardMail:
Department of Computer Science
University of Melbourne
Parkville, Victoria, 3052
AUSTRALIA
SoftMail:
UUCP: {seismo,ukc,prlb2}!munnari!jas
{decvax,eagle,pesnta}!mulga!jas (SLOW)
ARPA: munnari!jas@seismo.ARPA
CSNET: jas@munnari.oz
("jas" can be substituted by "rao")
Also, Dr. Rao will be attending the Logic Programming
Symposium in Boston, and would be willing to discuss the
system there.
[1] R.Sacks-Davis and K.Ramamohanarao "A Two Level
Superimposed Coding Scheme for Partial Match
Retrieval", Information Systems, v.8, n.4, 1983
[2] By way of comparison, this system eliminates a number
of restrictions which were associated with the
deductive database system provided with release 3.1
of MU-Prolog. That system implemented the database
manager as a separate process from the Prolog
interpreter, communicating via Unix pipes. The
present system is designed as a library package which
is compiled into the host system; it could be
incorporated fairly easily into most Prolog
interpreters, or, in fact, into any systems that
wished to perform partial match retrieval. The use of
pipes in the old MU-Prolog system placed severe
limitations (because of Unix file descriptor
limitations) on the number of transactions (queries)
which could be active concurrently; the new system
has eliminated this restriction. Finally, this system
lifts the restriction that only ground facts could be
stored in the database, by allowing the storage of
arbitrary Prolog terms (including rules).
[3] L.Naish "MU-Prolog 3.2db Reference Manual", Technical
Report, Department of Computer Science, University of
Melbourne, 1985.
------------------------------
Date: Thu, 27-Jun-85 21:01:28 PDT
From: mcvax!ukc!icdoc!sg@Seismo.ARPA
Subject: PARLOG system for C-Prolog
Since my recent announcement "PARLOG system available",
in which I offered to mail copies of the PARLOG system
to anyone on request, I have been inundated (!) with
enquiries. Owing to the high cost of transatlantic net
mail and its less-than-100% reliability, I have been
advised to post it instead. So here it is.
The version here runs on top of C-Prolog 1.4 - 1.5. Some
changes may be required to run on compatible Prolog systems.
There are 14 files altogether:
parlog hamming.par
parcomp primes.par
parrts adpairs.par
parstats print.par
orrts prolog.par
editor
par
npar
nparrts
Those with extension ".par" are example PARLOG
programs, the others are Prolog programs
comprising the PARLOG system.
Before use, the article should be run through a
shell to separate out the files: "sh <article".
You will need the manual "How to use PARLOG". To
get it, please send me a postal address, and request
any PARLOG papers you require.
[ PARLOG is available from the SCORE: <Prolog> library as
PAROLOG.PL. You'll need to unpack the file. -ed ]
------------------------------
End of PROLOG Digest
********************
∂09-Jul-85 0848 HEDGES@SU-SCORE.ARPA Visit of Jean-Claude Latombe
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 9 Jul 85 08:48:16 PDT
Date: Tue 9 Jul 85 08:46:24-PDT
From: Karen Hedges <HEDGES@SU-SCORE.ARPA>
Subject: Visit of Jean-Claude Latombe
To: Faculty@SU-SCORE.ARPA
If any faculty are interested in meeting with Jean-Claude Latombe (Robotics
Candidate), there are still openings late morning and this afternoon. Also
the Search Committee is having lunch with him at Noon today at the Faculty
Club and there is room for a few who might wish to join them for lunch.
Latombe is here today (July 9).
Please let me know if you would like to talk with him today.
Karen
-------
∂09-Jul-85 0951 @SU-SCORE.ARPA:udi@wisc-rsch.arpa STOC86 Call for papers (a message from Juris Hartmanis)
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 9 Jul 85 09:51:32 PDT
Received: from wisc-rsch.arpa by SU-SCORE.ARPA with TCP; Tue 9 Jul 85 09:30:53-PDT
Received: by wisc-rsch.arpa; Tue, 9 Jul 85 11:04:12 cdt
Received: from wisc-crys.arpa by wisc-rsch.arpa; Sun, 7 Jul 85 19:14:02 cdt
Message-Id: <8507080013.AA16556@wisc-crys.arpa>
Received: from COLUMBIA-20.ARPA by wisc-crys.arpa; Sun, 7 Jul 85 19:13:48 cdt
Date: Sun 7 Jul 85 20:13:50-EDT
From: Zvi Galil <GALIL@COLUMBIA-20.ARPA>
Subject: STOC86 Call for papers (a message from Juris Hartmanis)
To: theory@WISC-CRYS.ARPA
Sender: udi@wisc-rsch.arpa
Resent-From: udi@wisc-rsch.arpa (Udi Manber)
Resent-To: theory:;@wisc-rsch.arpa
CALL FOR PAPERS
1986 ACM SYMPOSIUM ON
THEORY OF COMPUTING
The Eighteenth Annual ACM Symposium on Theory of Computing,
sponsored by the ACM Special Interest Group for Automata and
Computability Theory, will be held in Berkeley, California
on May 28-30, 1986. Papers presenting original research on
theoretical aspects of computer science are being sought.
Typical, but not exclusive, topics include:
o Algorithms and Data Structures
o Theory of Logical Design and Layout
o Computability and Complexity Theory
o Cryptography
o Models of Computation
o Logic and Correctness of Programs
o Parallel and Distributed Computation
o Theory of Data Bases
o Semantics of Programming Languages
o Theory of Formal Languages and Automata
PAPER SUBMITTAL
Authors should send ten copies of a detailed abstract (not a
full paper) by December 2, 1985 to the program committee
chairman:
Juris Hartmanis
Department of Computer Science
405 Upson Hall
Cornell University
Ithaca, NY 14853
The abstract must provide sufficient detail to allow the
program committee to assess the merits of the paper, and
should include appropriate references and comparisons with
extent work. The abstract should be at most ten double-
spaced typed pages (about 2,500 words) in length. Submis-
sions that arrive late or that deviate from the prescribed
format risk rejection without consideration of their merits.
The program committee consists of M.J. Fischer, M.R. Garey,
J. Hartmanis, N. Immerman, M.M. Klawe, S. Micali, R. Parikh,
F.P. Preparata, R. Sedgewick and A. Wigderson.
Authors will be notified of acceptance or rejection by Janu-
ary 31, 1986. A copy of each accepted paper, typed on spe-
cial forms for inclusion in the symposium proceedings, will
be due by March 17, 1986.
MEETING FORMAT
The format of the meeting, including time allocations for
presentations, will be determined by the program committee.
Authors having a preference for a short (10-15 minute) or a
long (20-25 minute) presentation should express it at the
time of submission. Such a preference will not influence
acceptance, and time allocation will not be noted in the
proceedings or affect the space allocation for the paper.
The local arrangements chairman is:
Eugene Lawler
Computer Science Division
591 Evens Hall
University of California
Berkeley, CA 94720
-------
∂09-Jul-85 1458 DAVIES@SUMEX-AIM.ARPA Meeting Wednesday, June 10
Received: from SUMEX-AIM.ARPA by SU-AI.ARPA with TCP; 9 Jul 85 14:58:38 PDT
Date: Tue, 9 Jul 1985 14:56 PDT
Message-ID: <DAVIES.12125710329.BABYL@Sumex>
From: DAVIES@Sumex
To: AAP@Sumex
Subject: Meeting Wednesday, June 10
cc: Davies@Sumex
There will be a meeting this Wednesday at 9:30 am. Penny will report
on the SPL-Insight Workshop she attended in Britain.
-- Byron
∂09-Jul-85 1640 DAVIES@SUMEX-AIM.ARPA Welcome to PARSYM
Received: from SUMEX-AIM.ARPA by SU-AI.ARPA with TCP; 9 Jul 85 16:39:46 PDT
Date: Tue 9 Jul 85 16:35:27-PDT
From: Byron Davies <DAVIES@SUMEX-AIM.ARPA>
Subject: Welcome to PARSYM
To: AAP@SUMEX-AIM.ARPA
I figured that everyone in the Architectures Project would be
interested, so I have added you all to the PARSYM mailing list.
Let me know if you would like to be taken off.
WELCOME
Welcome to PARSYM, the netwide mailing list for discussion of parallel
symbolic computing.
I am the moderator of the PARSYM discussion. I am responsible for
composing the digest from pending submissions, keeping an archive, and
answering administrative requests.
You may submit material for the digest to PARSYM@SUMEX.ARPA.
Administrative requests should be sent to PARSYM-Request@SUMEX.ARPA.
Archival copies of all digests will be kept. Recent issues may be
obtained by anonymous FTP from <PARSYM>archive.txt on SUMEX.
PARSYM is open to discussion of any topic related to parallel symbolic
computing.
Examples of such topics are:
Programming languages for parallel symbolic computing
General-purpose parallel symbolic computing
Special-purpose parallel symbolic computing
Hardware for parallel symbolic computing
Applications which would benefit from parallel symbolic computing
AI on multiprocessor architectures
Problem-solving architectures to exploit parallelism
Control structures in parallel problem-solving systems
Data structures
Memory subsystems
Reviews of existing architectures
Pointers to seminars and articles
Descriptions of ongoing projects
Please observe the cautions and caveats appended below regarding
sender's responsibility, technology export, and copyright.
If you leave your current net address, please notify me. It is
difficult to distinguish an abandoned address from ordinary mailer
trouble, and tracing the source of mailer problems takes a fair amount
of effort.
Parallel symbolic computing depends on communication as well as
processing. So does this mailing list. Please communicate!
-- Byron Davies
!
Cautions and Caveats
To protect contributors and their affiliated organizations as well as
the moderator and his affiliated organizations, I urge you to observe
the following when sending material to the PARSYM mailing list:
1. Open Channel: PARSYM will be an open channel. Contributors will be
responsible for the contents of their messages.
2. Technology Export: The U.S. government is concerned about the
export of technical information. Do not forward a PARSYM message to a
network recipient who is resident abroad (even if the mailbox is at
some U.S. site). If you do so, it is your responsibility. As a
guideline, follow the same rules as would apply to the mailing abroad
of technical material using the U.S. post office. Transmission of
scientific information -- information that would ordinarily be
transmitted at a university seminar or in a scientific journal -- is
PROBABLY (but not definitely) safe. Transmission of detailed
engineering or manufacturing information is more likely to cause
problems.
3. Copyright: Responsibility for protecting the copyrights of
contributors and others rests with the contributors. Except for
"brief excerpts", copyright material should not be distributed via
PARSYM.
-------
∂09-Jul-85 1804 INGRID@SU-CSLI.ARPA Trip to Monterey
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 9 Jul 85 18:04:00 PDT
Date: Tue 9 Jul 85 17:56:54-PDT
From: Ingrid Deiwiks - 497-3084 <INGRID@SU-CSLI.ARPA>
Subject: Trip to Monterey
To: Folks@SU-CSLI.ARPA
In conjunction with the CSLI Summer School and the ASL Meeting, we are
offering a trip to visit the new and exciting Monterey Bay Aquarium on
Sunday, July 14. The bus will leave at 9:00 sharp from Stern Hall on
Escondido Road, take you to the aquarium for a 2-hour tour, then you'll
have a couple of hours to yourself for lunch, etc., and then go on to
the famous 17-Mile-Drive on the Monterey Peninsula. You should be back
at Stanford by 7:00. Tickets are $20 each and include bus fare and
entrance fees for the aquarium and the 17-Mile-Drive. Interested?
Please send a message to Ingrid@SU-CSLI, or call me at 497-3084
(11:00 to 5:00).
Ingrid
-------
∂10-Jul-85 1208 @SU-CSLI.ARPA:BrianSmith.pa@Xerox.ARPA All Mail Lost!
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 10 Jul 85 12:08:11 PDT
Received: from Xerox.ARPA by SU-CSLI.ARPA with TCP; Wed 10 Jul 85 11:54:23-PDT
Received: from Semillon.ms by ArpaGateway.ms ; 10 JUL 85 11:49:13 PDT
Date: 10 Jul 85 10:15 PDT
From: BrianSmith.pa@Xerox.ARPA
Subject: All Mail Lost!
To: Folks@SU-CSLI.ARPA, AllPa↑.pa@Xerox.ARPA
cc: BrianSmith.pa@Xerox.ARPA
Reply-to: BrianSmith.pa@Xerox.ARPA
Message-ID: <850710-114913-1595@Xerox>
Due to a major disk failure, I have unfortunately lost 6 months worth of
mail, including ALL MAIL SENT TO ME SINCE JUNE 18 (none of which I had
yet read, since I was out of town). So: if you sent any crucial
messages during that time, could you please resend them?
Sorry for the trouble, and thanks in advance.
Brian
∂10-Jul-85 1234 @SU-CSLI.ARPA:LANSKY@SRI-AI.ARPA Next Monday's PLANLUNCH
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 10 Jul 85 12:34:02 PDT
Received: from SRI-AI.ARPA by SU-CSLI.ARPA with TCP; Wed 10 Jul 85 12:21:53-PDT
Date: Wed 10 Jul 85 12:22:15-PDT
From: LANSKY@SRI-AI.ARPA
Subject: Next Monday's PLANLUNCH
To: aic-associates@SRI-AI.ARPA, folks@SU-CSLI.ARPA, mugs@SUMEX-AIM.ARPA,
val@SU-AI.ARPA, rsimmons@SRI-KL.ARPA, carnese@SRI-KL.ARPA
Temporal Imagery:
An Approach to Reasoning about Time for Planning and Problem Solving
Thomas Dean
Yale University
11:00 AM, Monday, July 15
Room EJ232, SRI International
ABSTRACT:
Reasoning about time typically involves drawing conclusions on the basis
of incomplete information. Uncertainty arises in the form of ignorance,
indeterminacy, and indecision. Despite the lack of complete information
a problem solver is continually forced to make predictions in order to
pursue hypotheses and plan for the future. Such predictions are
frequently contravened by subsequent evidence. The talk will describe
a computational approach to temporal reasoning that directly confronts
these issues. The approach relies upon a method for keeping track
of the dependency relations among assertions in a temporalized data base.
The resulting computational framework extends the functionality of
reason maintenance systems [Doyle 79] to handle assertions with a temporal
extent. The techniques developed extend the functionality of current
approaches to dealing with time in planning (e.g., [Sacerdoti 77], [Tate
77], [Vere 83], and [Allen 83]). Examples from robot problem solving
will be used to illustrate the techniques.
-------
-------
∂10-Jul-85 1705 EMMA@SU-CSLI.ARPA Newsletter
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 10 Jul 85 17:05:29 PDT
Date: Wed 10 Jul 85 16:53:45-PDT
From: Emma Pease <Emma@SU-CSLI.ARPA>
Subject: Newsletter
To: friends@SU-CSLI.ARPA
Tel: 497-3479
There is no newsletter this week. No activities are scheduled for
either Thursday, July 11 or Thursday, July 18.
-Emma Pease
-------
∂11-Jul-85 0039 RESTIVO@SU-SCORE.ARPA PROLOG Digest V3 #31
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 11 Jul 85 00:38:55 PDT
Date: Wednesday, July 10, 1985 8:10PM
From: Chuck Restivo (The Moderator) <PROLOG-REQUEST@SU-SCORE.ARPA>
Reply-to: PROLOG@SU-SCORE.ARPA
US-Mail: P.O. Box 4584, Stanford CA 94305
Phone: (415) 326-5550
Subject: PROLOG Digest V3 #31
To: PROLOG@SU-SCORE.ARPA
PROLOG Digest Thursday, 11 Jul 1985 Volume 3 : Issue 31
Today's Topics:
Implementations - Cut & Conditionalization,
& A Standard Syntax
----------------------------------------------------------------------
Date: Wed, 10 Jul 85 14:59 EDT
From: Tim Finin <Tim%UPenn@csnet-relay>
Subject: Investigating implementations of cut
I ran Chris Moss's tests on several of the Prologs
we use here with the following results. I should
note that the Symbolics Prolog is still a Beta-test
version and that it is a compile-only system (i.e.
there is no interpreter).
Implementation Test 1 2 3 4 5 6
UNH Prolog Y Y Y Y Y N
8086 Prolog-1 Y Y Y Y Y N
Symbolics Prolog (Beta) Y Y Y I I N
-- Tim
------------------------------
Date: Mon, 8 Jul 85 15:57:14 PDT
From: Deutsch.pa@Xerox.ARPA
Subject: Read-time conditionalization
The result of including read-time conditionalization
in Common Lisp is that it makes structure-based (as
opposed to file-based) environments very difficult
or impossible. I think this is a medium-size disaster.
I would oppose its inclusion in a Prolog standard.
------------------------------
Date: 10 Jul 85 02:26:08 +1000 (Wed)
From: munnari!mungunni.oz!Lee@seismo
Subject: Read Time Conditionalization
Recently I implemented a simple facility like this for
MU-PROLOG (we needed it for the VMS version).
It is written entirely in PROLOG, without any changes
to the reader or consult. One restriction is that,
within the conditionals, you can only have entire
clauses (so a bit of code may need to be duplicated).
The construct is rather like #if, #else and #endif in C.
Here is an example:
?- use←if $database.
?- consult('ext←dbpreds.pl').
% .....
?- use←else.
?- consult('dbpreds.pl').
% .....
?- use←end.
The implementation is below (it could be improved a bit, to
check for some other cases of unmatched uses). If we can get
a portable version of this (shouldn't be hard), and we can
figure out some tests to find what Prolog we are running
(more difficult), then we can have reasonably portable
programs.
==========================================================
?- op(990, fx, use←if).
use←if(X) :-
not(X),
repeat,
read(Y),
nonvar(Y),
( Y = (?- use←else)
;
Y = (?- use←end)
;
eof(Y),
writeln('Unmatched "use"')
;
Y = (?- use←if ←),
use←else,
fail
),
!.
use←if(←).
% gobbles terms up to use←end.
use←else :-
repeat,
read(Y),
nonvar(Y),
( Y = (?- use←end)
;
eof(Y),
writeln('Unmatched "use"')
;
Y = (?- use←if ←),
use←else,
fail
),
!.
use←end.
?- protect([use←if(1), use←else, use←end]).
===================================================
-- Lee Naish
"This fancy Prolog package of yours. How portable is it?"
------------------------------
Date: Mon, 8-Jul-85 12:54:50 PDT
From: Vantreeck%Logic.DEC@DECWRL
Subject: Prolog syntax
I'm glad to see all the responses. I was wondering if
very many read this notes file.
Indeed, the S-expression of Lisp is a powerful syntax to
express both functional and logic programming, e.g., the
language TAO. Other languages, e.g., APL, have put a higher
premium on expression "with a minimum of syntax" than on
ease of reading. It's said that C code can be made very
readable. Unfortunately, too many follow the path least
resistance. I think that the DoD chose a Pascal-like
syntax for Ada, because it tended to encourage self
documenting code.
Suppose a major computer/software company were going to
market a language for which there is no standard. Should
the company make it's implementation such that it targets
those logicians and AI researchers that prefer the economy
of expression? Or should it make its implementation be
usable by logicians and AI researchers by remaining within
the bounds of first order logic and also provide software
engineering features that the more conventional software
shops prefer?
Some large corporations are staffing up with PhDs to lead
the development of expert systems. But many of the
programmers will be in house people from COBOL/FORTRAN
backgrounds who will have to learn a new language. Having
tried to introduce some people to Prolog, I've learned
that most have enough difficulty understanding bactracking,
cut, and unification. And the syntax was no great help
in the learning process. The last thing these people
need to hear about are S-expressions, lambda calculus,
predicate calculus, horn clauses, etc.. Their need is
very pragmatic: "This is what I want to do. How do I
do it in Prolog?"
"for←all" is not a universal quantifier. It is an
existential quantifier, i.e., one per clause. Want an
alternative name? Lambda? Sounds like Greek to me. How
about an existential quantifier that I can understand, like
"exists".
Should cut be implemented in a way that allows one to write
a faster program, or should cut behave in a consistent
manner? I think it's a question of speed versus
consistency. I vote for consistency (the hard cut). When I
say cut, that means that I want to exit the procedure on
backtracking to the cut - no exceptions.
------------------------------
End of PROLOG Digest
********************
∂11-Jul-85 0958 PARSYM-Request@SUMEX-AIM.ARPA PARSYM Digest V1 #1
Received: from SUMEX-AIM.ARPA by SU-AI.ARPA with TCP; 11 Jul 85 09:57:55 PDT
Date: 10 Jul 85 1718-PDT
From: Moderator Byron Davies <PARSYM-REQUEST@SUMEX-AIM.ARPA>
Reply-to: PARSYM@SUMEX-AIM.ARPA
Subject: PARSYM Digest V1 #1
To: PARSYM@SUMEX-AIM.ARPA
PARSYM Digest Wednesday, 10 Jul 1985 Volume 1 : Issue 1
Today's Topics:
Welcome
What is symbolic computing?
Systolic Programming
A current research project on parallel symbolling
----------------------------------------------------------------------
Date: Wed, 10 Jul 1985 14:57 PDT
From: Byron Davies <PARSYM-Request@SUMEX-AIM.ARPA>
Subject: Welcome
Welcome to PARSYM, the netwide mailing list for parallel symbolic
computing.
I'm impressed to see so much interest in the topic. PARSYM-Request
has received over 130 individual requests to be added to the list, as
well as several requests for local redistribution. Many of the
members are researchers in the field of parallel symbolic computing,
so we should have technical content as well as news and views.
Included in this issue are three significant firsts for PARSYM: the
first flame, the first logic programming article, and the first LONG
submission (see if you can tell which is which). I thank the three
contributors for their contributions -- they've made this first issue
possible.
Several new members offered summaries of work going on in parallel
symbolic computing at their institutions. I accept all your offers
and encourage others to make similar contributions.
-- Byron
------------------------------
Date: Mon, 8 Jul 85 09:15:00 pdt
From: coraki!pratt@Navajo (Vaughan Pratt)
Subject: What is symbolic computing?
I suppose with quarter of a century of Lisp experience we should
understand by now what symbolic computing is. Indeed everyone appears
to be quite happy to talk about symbolic computing as though the
concept had the unanimous blessing of the last three centuries of
mathematics.
Yet I have been unable to find in the literature a definition of the
concept that matches the facts.
Discussion based on the premise that the topic is well-defined when it
isn't is apt to run at cross-purposes.
I therefore challenge this list to agree on a definition of symbolic
computing.
Even if this challenge cannot be met (as I expect will be the case), at
least people will have been made aware of the variety of definitions
and will know to make allowance for this variety.
-v
------------------------------
Date: 9 Jul 1985 1448-CDT
From: THRIFT%CSL60%ti-csl.csnet@csnet-relay.arpa
Subject: Systolic Programming
A high-level logic based language for parallel computing
is presented in "Systolic Programming: A Paradigm of
Parallel Processing" E. Shapiro, Proceedings of the
International Conference on Fifth Generation Computer
Systems 1984. The programming language is Concurrent
Prolog augmented with LOGO-like Turtle programs as a
process to processor mapping notation. The target machine
is a set of processors interconnected according to a
discrete torus topology. The result has the following
properties:
"1. It is scalable.
2. It has a simple and regular interconnection pattern.
3. It executes a high-level concurrent programming
language, augmented with a mapping notation.
4. It is the responsibility of the applications programs,
not the architecture or the operating system, to insure
locality of interprocess communication."
PARLOG (Clark and Gregory) is a contender in the logic
programming world for parallel computation, but ,as far
as I'm aware, doesn't have the notation for programmability
of process-to-processor mapping. The Socratic dialog in the
Prologue of Shapiro's paper argues effectively for the
the necessity of the programmer being able to control
explicitly this new dimension: " the definition of process
structures and the mapping of processes to processors."
The Shapiro paper should be read by anyone interested in
the logic programming approach to parallel computing.
------------------------------
Date: Wed, 10 Jul 85 09:55:18 EST
From: munnari!dmtadel.oz!crw@seismo.CSS.GOV (charles watson)
Subject: A current research project on parallel symbolling
A RESEARCH PROPOSAL FOR A
FIFTH GENERATION COMPUTER DESIGN
Charles R. Watson
AIM
My aim is to design and eventually implement a highly
concurrent inference machine to form the basis for powerful
knowledge-based systems in the future.
The final implementation is targetted at the wafer scale
integrated circuit technology that will be available in 1990.
This design will provide a potential increase in logical
inference power of 100,000 times the capability of contemporary
expert systems.
The short term plan is to continue following the
literature, do a comparison of the various inference paradigms,
design an architecture for the most promising paradigm, and then
implement a prototype system. In the longer term (2 to 5 years)
a full scale commercial implementation could be achieved with a
small design team and a moderate capital outlay.
SHORT TERM PLAN
Literature reading and evaluation of inference paradigms
The two main inference mechanisms are LISP (mainly used
in the U.S.) and logic programming (Japan and Europe).
Concurrency can be utilized in LISP by the simultaneous
evaluation of lists of expressions, and in logic programming by
either parallel unification, parallel selection of procedures, or
parallel evaluation of disjunctives in the body of procedures.
Several issues need to be compared for each of these mechanisms.
These include hardware requirements, cost, and performance. The
applicability to expert systems should be determined, i.e. the
ease of programming, or storing knowledge, and the degree of
utilization of available concurrency.
Architecture
In developing the preferred architecture, functional
simulation will be done on the various schemes to predict
performance. A small knowledge-based system will be emulated to
determine the critical performance areas, such as the degree of
utilization of processing elements (PEs). The architecture will
be extensible, that is results from simulation and testing of
small systems (10 PEs) will be easily applied to larger systems.
Design
This will involve a detailed implementation of the
architecture down to the IC mask layout. Timing and logic
simulation of the processing elements (PEs) and memory management
units (MMUs) will be done to verify the correctness of the
implementation. An initial prototype, with about 10 PEs, will be
fabricated and interfaced to an expert system development
environment for performance evaluation.
A CONCURRENT LISP ARCHITECTURE
The basic PE is similar to the microcoded LISP processors
used in the SCHEME chip or the Symbolics 3600 machine. A garbage
collector would be attached to each PE and would work in parallel
with the PE, freeing the used local memory space. The PE would
differ from single processor LISP machines in the evaluation of
the CONS primitive. To evaluate CONS( A, B ) the processor would
first set up a data structure to store the result then spawn a
process on an idle processor to evaluate B, then evaluate A on
its own processor, and wait for the spawned process to finish.
The processor finally returns the completed data structure. More
elaborate schemes exist, which avoid waiting by PEs.
When evaluation is spawned on another PE, a pointer to
the ancestral environment (set of variable bindings) is also
passed to the new PE. Local bindings can then be added to form an
environment tree, in such a way that local bindings can only be
used by descendents.
Several performance issues need to be considered. The
time required to spawn a new process should be small compared to
the time saved by parallel evaluation. Creation of the data
structure, and passing the address to the spawned processor can
be implemented as a single machine instruction using suitable
hardware. If the spawned processes have only one instruction,
then there is no advantage in spawning. This would occur in tail
recursion of a list of atoms. The rate of spawning should also
be effected by the number of neighbouring PEs that are idle, and
to a lesser extent the load on more remote PEs.
The use of SETQ, RPLACA, and RPLACD can cause errors in
concurrent evaluation. Either a functional form of LISP should
be used without these primitives, or a method must be devised to
check for the side effects they cause.
In a large scale implementation (1000 PEs) a
multiprocessor virtual memory system would be employed to provide
fast inexpensive memory access (see the enclosed paper "A
Concurrent Computer Architecture"). It would be necessary to
ensure the locality of reference in knowledge-base applications
to maintain high performance.
BENEFIT TO AUSTRALIA
MITI in Japan is spending $500M in the next 10 years on
their 5th Generation Project. A similar amount is being spent by
the U.S. DARPA for their Super Computer Project. Australia
cannot compete with such capital intensive research, but we must
keep our foot in the door. The project I propose would provide a
vehicle to utilize our local intellect and ingenuity with
moderate cost. Standard interfaces to modern integrated circuit
fabrication make such novel designs economically feasible. There
is already significant Australian research in the theory of Logic
Programming, and the implementation of expert systems. The
proposed architecture would create many new research and
commercial opportunities.
------------------------------
End of PARSYM Digest
********************
∂11-Jul-85 1148 RICHARDSON@SU-SCORE.ARPA Visit of Peter Will
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 11 Jul 85 11:46:01 PDT
Date: Thu 11 Jul 85 11:42:50-PDT
From: Anne Richardson <RICHARDSON@SU-SCORE.ARPA>
Subject: Visit of Peter Will
To: Faculty@SU-SCORE.ARPA
cc: Richardson@SU-SCORE.ARPA
Message-ID: <12126199415.23.RICHARDSON@SU-SCORE.ARPA>
Dr. Peter Will (Robotics Candidate) will be visiting Stanford on Tuesday,
July 23. If you would like to meet and talk with him, please let me
know times you are available.
Thanks.
Anne
-------
∂11-Jul-85 1317 @SU-SCORE.ARPA:avi.sjrlvm1%ibm-sj.csnet@csnet-relay.arpa BATS at IBM
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 11 Jul 85 13:17:47 PDT
Received: from csnet-relay by SU-SCORE.ARPA with TCP; Thu 11 Jul 85 13:14:05-PDT
Received: from ibm-sj by csnet-relay.csnet id ao29038; 11 Jul 85 16:11 EDT
Date: Thu, 11 Jul 85 12:38:18 PDT
From: Avi Wigderson <avi%ibm-sj.csnet@csnet-relay.arpa>
To: aflb.all@su-score.ARPA
Subject: BATS at IBM
Hi
Here is the plan for the coming BATS (Bay Area Theory Seminar)
at IBM. Please spread the word in your institutions.
It will take place on Thursday, August 1, in the research building
in IBM (the usual place). This time the BATS will be devoted to
Randomness, as can be seen from the preliminary titles below.
Joel Spencer, SUNY at Stony Brook
"Probabilistic Methods Made Concrete"
Noga Alon, Tel-Aviv University
"Probabilistic Communication Complexity"
Umesh Vazirani, MSRI at Berkeley
"Random Polynomial Time = Slightly-Random Polynomial Time"
Avi Wigderson, IBM
"Deterministic Simulation of Probabilistic Constant-Depth Circuits"
Abstracts will be sent in about a week.
Hope to see you there,
AVI
∂11-Jul-85 1344 @SU-SCORE.ARPA:JF@SU-SUSHI.ARPA bats
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 11 Jul 85 13:43:46 PDT
Received: from SU-SUSHI.ARPA by SU-SCORE.ARPA with TCP; Thu 11 Jul 85 13:39:26-PDT
Date: Thu 11 Jul 85 13:39:22-PDT
From: Joan Feigenbaum <JF@SU-SUSHI.ARPA>
Subject: bats
To: aflb.su@SU-SCORE.ARPA
As I've already said, I will not be around to arrange carpools to IBM on
August 1. Is there someone out there who would like to do that? Otherwise,
you're on your own.
-------
∂11-Jul-85 1428 ACUFF@SUMEX-AIM.ARPA Common Lisp Implementors Mailing List
Received: from SUMEX-AIM.ARPA by SU-AI.ARPA with TCP; 11 Jul 85 14:28:33 PDT
Date: Thu 11 Jul 85 14:26:39-PDT
From: Richard Acuff <Acuff@SUMEX-AIM.ARPA>
Subject: Common Lisp Implementors Mailing List
To: sumex-bboard@SUMEX-AIM.ARPA, ksl-lisp-machines@SUMEX-AIM.ARPA
Message-ID: <12126229237.55.ACUFF@SUMEX-AIM.ARPA>
There is a bboard for Common Lisp issues now available in
<bboard>common-lisp.txt. This list is used primarily for discussing
implementation issues, and would likely be of minimal interest to
potential Common Lisp users, but it's there if anyone would like
to follow it.
-- Rich
-------
∂11-Jul-85 1517 YAMARONE@SU-CSLI.ARPA NO SANDS.FRI orMON.
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 11 Jul 85 15:17:24 PDT
Date: Thu 11 Jul 85 15:02:34-PDT
From: Tom Yamarone <YAMARONE@SU-CSLI.ARPA>
Subject: NO SANDS.FRI orMON.
To: folks@SU-CSLI.ARPA
Due to a temporary shortage of personnel and a recent decline in sales,
The Ventura Sandwich Corporation will be closed Friday(July 12) and
Monday(July 15). Tell a friend and may we recommend you try the food
services of Tresidder Union or the Medical Center on these days. We
look forward to serving you Tuesday(July 16) and thank you for your
patronage. Most sincerely, The Ventura Sandwich Corporation.
-------
∂11-Jul-85 1749 @SU-SCORE.ARPA:WIEDERHOLD@SUMEX-AIM.ARPA Re: Visit of Peter Will
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 11 Jul 85 17:48:51 PDT
Received: from SUMEX-AIM.ARPA by SU-SCORE.ARPA with TCP; Thu 11 Jul 85 17:46:18-PDT
Date: Thu 11 Jul 85 17:46:45-PDT
From: Gio Wiederhold <WIEDERHOLD@SUMEX-AIM.ARPA>
Subject: Re: Visit of Peter Will
To: RICHARDSON@SU-SCORE.ARPA
cc: Faculty@SU-SCORE.ARPA, ariadne@SUMEX-AIM.ARPA
In-Reply-To: <12126199415.23.RICHARDSON@SU-SCORE.ARPA>
Message-ID: <12126265663.52.WIEDERHOLD@SUMEX-AIM.ARPA>
yes, i would like to meet with him. Ariadne will verify. Gio
-------
∂12-Jul-85 1527 PARSYM-Request@SUMEX-AIM.ARPA PARSYM Digest V1 #2
Received: from SUMEX-AIM.ARPA by SU-AI.ARPA with TCP; 12 Jul 85 15:27:17 PDT
Date: 12 Jul 85 1501-PDT
From: Moderator Byron Davies <PARSYM-REQUEST@SUMEX-AIM.ARPA>
Reply-to: PARSYM@SUMEX-AIM.ARPA
Subject: PARSYM Digest V1 #2
To: PARSYM@SUMEX-AIM.ARPA
PARSYM Digest Friday, 12 Jul 1985 Volume 1 : Issue 2
Today's Topics:
Restructuring LISP
Concurrent LISP systems
----------------------------------------------------------------------
Date: Wed, 10 Jul 85 20:54:17 cdt
From: harrison%uiucdcsb@Uiuc.ARPA (Williams ["Luddy"] Harrison)
Subject: Restructuring LISP
I was pleased to learn of the creation of this mailing list. Here
at the Center for Supercomputer Research at the U of I we are working on
restructuring LISP for interpretation on multiprocessors, such as the RP3
or the CEDAR machine. The PARAFRASE analyzer, which restructures FORTRAN
code, was developed here and has been turning sequential FORTRAN programs
into parallel code with great success for several years now.
We aren't yet prepared to quantify the success of restructuring LISP,
but we have developed a series of transformations that produce remarkable
results on the examples on which they've been tried.
Anyone interested in reading about the work is welcome to send me
mail at
harrison@uiucdcs
(we're on the arpanet). We're in the process of preparing documents for
external distribution on the subject.
I can't say which of the many languages used for symbolic processing
would form the best basis for a language with parallelism made explicit
by the programmer; but I feel that the issue is perhaps more complex than is
frequently appreciated.
Consider the function "flatten" that is used to take an arbitrary
list and create from one that has all of the atoms of the original, but
in a single, one-level list (a list of the atoms as they are encountered
in an inorder search of the s-expr):
(def flatten
(lambda (x)
(cond [(atom x) x]
[(atom (car x))
(cons (car x) (flatten (cdr x)))]
[t (nconc (flatten (car x)) (flatten (cdr x)))]
)))
Who can say what the inherent parallelism of this function is?
If we take the approach of the dataflow folks, and spawn recursive
processes as we descend into the list, we may get Sp=n/lgn (assumming
that the s-expr was a balanced tree); but what happens in this approach
if the expression was already (or nearly) in the flattened form? Our
speedup is O(1), which is not so good.
(Incidentally, the restructuring techniques we are proposing
can transform this function such that it runs in time proportional to
the depth of the deepest atom, i.e., O(1) time on the pre-flattened case,
O(lgn) on the balanced tree case.)
Given a language in which the programmer can express parallelism
directly, how can we write this function such that the parallelism inherent
in the algorithm is explicit?
It should be pointed out that we have found that, despite the
success that program resturcturing has had, there are quite a few
computations whose parallelism seems completely to elude attempts at automatic
detection; most of our experience in this area is with FORTRAN, but it
is becoming clear what sorts of constructs in LISP code will foil attempts
at "parallelization". For this reason, let me put in my bid for a mixed
approach: a language in which parallelsm can be made explicit, but for
which a compiler exists that can pore over the code finding parallelism
that is either obvious and therefore tedious to express, or that requires
such unnatural bending and twisting of the original code that it is too
painful and time-consuming for the programmer to attempt.
Let me add as a supporting argument for "Supercompilers" that many
sophisticated architectural features have been introduced into computers
over the years; many of these are painful for the user to take advantage
of. Consider what it would be like if we all had to manually allocate
registers; likewise, no one argues for a language in which cache management
or paging can be expressed explicitly. We expect these architectural
features to be exploited by the software and hardware of the system. Yet
we do not expect the software of the system to do much to help us in employing
certain parallel features of the machine (although we expect help in
benefitting from the parallelism of overlapped i/o). Is it not asking too
much of a programmer, that he should spend so much effort exploiting an
architecture, when designing an application?
Luddy Harrison
Center for Supercomputer Research and Development
302D Talbot Laboratory
104 South Wright Street
Urbana Illinois 61801-2932
------------------------------
Date: Thu, 11 Jul 85 16:32:24 cdt
From: harrison%uicsrd@Uiuc.ARPA (Luddy Harrison)
Subject: Concurrent LISP systems
Much has been said concerning concurrent LISP systems in which the
primary source of parallelism is the spawning of processes which correspond
to multiple arguments to functions; frankly, I'm skeptical that such a
method could ever achieve an acceptable efficiency or degree of parallelism.
But I'm willing to be convinced otherwise.
Some questions:
1) Is there any empirical evidence that lisp programs that exist can be
submitted to such a system with acceptable results? Most that I've seen
are rife with side effects, and I presume that either synchronization would
be introduced to overcome this or else that such programs would simply not
be allowed. If they are not, is there a technique for converting them into
acceptable form, such that the resulting code is reasonably efficient with
respect to the original, sequential version?
2) One note that I saw in today's digest suggested that setq and rplac* would
not be allowed in one such system; what about properties of atoms?
Using putprop is essentially the same as using setq on a global variable,
so I guess that this would be disallowed also, or possibly would be treated
as a critical section, using synchronization.
I'm concerned about such systems because it seems to me that they treat a
subset of real-life lisp that is too constraining, and that even if one
accepts this subset, that there is little evidence that it is possible to
write general-purpose code in which there are lots of significant tasks to
spawn. It seems to me that in some ways this approach also suffers from the
defects of dataflow architectures: so much communication overhead to amortize
over such a fine grain of parallelism.
Luddy Harrison
Center for Supercomp. R & D
U of I, Urbana
(unix addr: harrison@uiucdcs)
------------------------------
End of PARSYM Digest
********************
∂13-Jul-85 1405 SELLS@SU-CSLI.ARPA ants!!
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 13 Jul 85 14:05:34 PDT
Date: Sat 13 Jul 85 14:01:39-PDT
From: Peter Sells <Sells@SU-CSLI.ARPA>
Subject: ants!!
To: folks@SU-CSLI.ARPA
cc: bboard@SU-CSLI.ARPA
Will everyone using the trailers please STOP leaving food and drink garbage
in the trash cans in the trailers. The terminal that I am currently using
is crawling with ants attracted by the lure of a cheap lunch. A little
consideration here will make conditions more pleasant for all.
Peter Sells
-------
∂13-Jul-85 1953 DAVIES@SUMEX-AIM.ARPA [YM: PARSYM -- new mailing list for Parallel Symbolic Computing ]
Received: from SUMEX-AIM.ARPA by SU-AI.ARPA with TCP; 13 Jul 85 19:53:35 PDT
Date: Sat, 13 Jul 1985 19:52 PDT
Message-ID: <DAVIES.12126812853.BABYL@Sumex>
From: DAVIES@Sumex
To: YM@Sumex, PAI@Sumex
Subject: [YM: PARSYM -- new mailing list for Parallel Symbolic Computing ]
Re:
Date: Saturday, 13 July 1985 17:54-PDT
From: Yoni Malachi <YM at SU-AI.ARPA>
To: DAVIES at SUMEX-AIM.ARPA
Re: PARSYM -- new mailing list for Parallel Symbolic Computing
Is PAI going to be included in the new list automatically?
The note I sent out was an announcement and invitation to subscribe to
PARSYM. I intended to leave the option up to the individuals on the
list. Now that you bring it up, I might as well ask:
Is there anybody on PAI (besides Vineet Singh) who does not want to be
on PARSYM? Would people prefer bboards to be set up on the various
local computers?
-- Byron
∂14-Jul-85 1847 @SU-CSLI.ARPA:LANSKY@SRI-AI.ARPA PLANLUNCH reminder
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 14 Jul 85 18:47:28 PDT
Received: from SRI-AI.ARPA by SU-CSLI.ARPA with TCP; Sun 14 Jul 85 18:42:26-PDT
Date: Sun 14 Jul 85 18:44:42-PDT
From: LANSKY@SRI-AI.ARPA
Subject: PLANLUNCH reminder
To: aic-associates@SRI-AI.ARPA, folks@SU-CSLI.ARPA, mugs@SUMEX-AIM.ARPA,
val@SU-AI.ARPA, rsimmons@SRI-KL.ARPA, carnese@SRI-KL.ARPA
Temporal Imagery:
An Approach to Reasoning about Time for Planning and Problem Solving
Thomas Dean
Yale University
11:00 AM, Monday, July 15
Room EJ232, SRI International
ABSTRACT:
Reasoning about time typically involves drawing conclusions on the basis
of incomplete information. Uncertainty arises in the form of ignorance,
indeterminacy, and indecision. Despite the lack of complete information
a problem solver is continually forced to make predictions in order to
pursue hypotheses and plan for the future. Such predictions are
frequently contravened by subsequent evidence. The talk will describe
a computational approach to temporal reasoning that directly confronts
these issues. The approach relies upon a method for keeping track
of the dependency relations among assertions in a temporalized data base.
The resulting computational framework extends the functionality of
reason maintenance systems [Doyle 79] to handle assertions with a temporal
extent. The techniques developed extend the functionality of current
approaches to dealing with time in planning (e.g., [Sacerdoti 77], [Tate
77], [Vere 83], and [Allen 83]). Examples from robot problem solving
will be used to illustrate the techniques.
-------
-------
∂15-Jul-85 0031 RESTIVO@SU-SCORE.ARPA PROLOG Digest V3 #32
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 15 Jul 85 00:31:30 PDT
Date: Saturday, July 13, 1985 1:04PM
From: Chuck Restivo (The Moderator) <PROLOG-REQUEST@SU-SCORE.ARPA>
Reply-to: PROLOG@SU-SCORE.ARPA
US-Mail: P.O. Box 4584, Stanford CA 94305
Phone: (415) 326-5550
Subject: PROLOG Digest V3 #32
To: PROLOG@SU-SCORE.ARPA
PROLOG Digest Monday, 15 Jul 1985 Volume 3 : Issue 32
Today's Topics:
Query - Building Large Systems,
Implementation - Cut & A Standard Syntax & Conditionalization
----------------------------------------------------------------------
Date: Thu, 11 Jul 85 14:30 EST
From: Robert France <France%vpi@csnet-relay>
Subject: Prolog for Building Large Systems
What have your experiences been with using your favorite
Prolog in large system contruction? What Prolog would
you recommend for building something substantial?
We are currently designing a system here that will apply
techniques of expert systems and knowledge engineering to
the task of information retrieval from very large text
databases. For reasons of software engineering and general
godliness, we would like to take a modular approach to this
project; for reasons of efficency, we may want to code
portions of it in a procedural language (C by concensus);
but for other obvious reasons, we would like to build the
cooperating experts in Prolog.
We currently are using a local lisp-syntax Prolog which,
although a very nice little language for prototyping, is
inadequate for the project in several ways. What we need
is a Prolog that:
-- supports modularity, either within the language, or through
appropriate process-control primitives,
-- has a workable interface with (at least one) procedural
language (preferably C), and/or
-- directly suports large, fast database functions.
If you have any knowledge of or experience with a Prolog
with some ofthese features, please take the time to tell us
about it. We will, of course, summarize the results and
post them back to this Digest.
Thank you all,
-- Robert France
------------------------------
Date: 11 July 85 23:30-EDT
From: BYKAT%UTCVM.BITNET@Berkeley
Subject: Chris Moss Cut-Test results
I have run Chris Moss' tests on our MPROLOG Version1.5
supplied by Logicware and running under VM/CMS on an IBM
4381. Here are the results:
Implementation Test 1 2 3 4 5 6
MProlog Y Y Y Y N N
-- Alex Bykat
------------------------------
Date: Mon, 8-Jul-85 10:51:54 PDT
From: (Frank McCabe) mcvax!ukc!icdoc!fgm@Seismo
Subject: A Standard Syntax
The standard micro-Prolog system includes a package
which 'translates' between the 'standard' (quotes
intentional) Edinburgh syntax and micro-Prolog syntax.
This doesn't do a particularly good job on standard
micro-Prolog although the systems on sigma-Prolog
and Mac-Prolog do a great deal better.
The problems over semantics are very hard to solve; however
not even Edinburgh Prolog systems agree over how to
implement cut! (For proof of this see C.D.Moss's
article on the differences between cut on various systems).
------------------------------
Date: Fri, 12 Jul 85 00:08 EDT
From: Tim Finin <Tim%upenn@csnet-relay>
Subject: Read Time Conditionalization
I'm not sure what Peter Deutsch means by "structure-based
(as opposed to file-based) environments" but it seems to
me that it is neccessary to handle differences at the reader
level. I am assuming that "structure-based" refers to the
ability to have things behave differently in environments
which differ after the reader has done its job.
Actually, what I have in mind really has two independent
parts:
(1) having a standard way one can find out if a particular
Prolog instantiation does or does not have a particular
"feature" (e.g. is C-Prolog running on a VAX under BSD4.2)
(2) having a mechanism to make the reader sensitive to
these features.
One can use (1) to achieve "structure-based" environments,
if I understand the term correctly, with out resorting to
(2). I believe that (2) is neccessary to handle
reader-level differences and useful for several other
reasons (e.g. efficiency).
One problem I forsee with (1) is having some consensus on
what a feature is and agreeing on some standard ones (e.g.
names for common machines and operating systems). This
does not seem to have been a big problem in the Common
Lisp family, however.
-- Tim
------------------------------
Date: 12 Jul 85 9:11:56 PDT
From: Deutsch.pa@Xerox.ARPA
Subject: Read Time Conditionalization
By a "structure-based" environment, I mean one like
Interlisp, where the tools are based on structures held
within the address space rather than on external text
files manipulated by a foreigneditor. An apparent
requirement for such a system (which I believe makes
the construction of tools easier than a file-based
system -- you don't have to be constantly worrying about
manipulating things that have language-relevant semantics
but are represented as character strings) is that the
reader doesn't lose any information that the programmer
cares about. Read-time conditionalization doesn't have
this property, unless it is done in a very arcane manner
using something like hash links to remember the information
that is discarded from the list structure.
Maclisp and Zetalisp (and apparently Common Lisp) all take
the opposite view, that the tools are at least partly file
based and the editor is a text editor. While this is
obviously a workable approach, I suspect it is part of the
reason why it took so long for that world to develop
retrieval tools anywhere near as good as Masterscope, for
example.
------------------------------
End of PROLOG Digest
********************
∂15-Jul-85 0932 @SU-SCORE.ARPA:udi@wisc-rsch.arpa FOCS program & information
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 15 Jul 85 09:32:31 PDT
Received: from wisc-rsch.arpa by SU-SCORE.ARPA with TCP; Mon 15 Jul 85 09:17:47-PDT
Received: by wisc-rsch.arpa; Mon, 15 Jul 85 10:41:22 cdt
Received: from wisc-crys.arpa by wisc-rsch.arpa; Sun, 14 Jul 85 19:50:49 cdt
Message-Id: <8507150049.AA15091@wisc-crys.arpa>
Received: from csnet-relay.arpa by wisc-crys.arpa; Sun, 14 Jul 85 19:49:26 cdt
Received: from uoregon by csnet-relay.csnet id ag18658; 14 Jul 85 20:45 EDT
Received: by uo-vax3. (4.12/1.0.uoregon)
id AA24447; Sat, 13 Jul 85 22:53:56 pdt
Date: Sat, 13 Jul 85 22:53:56 pdt
From: Eugene Luks <luks@uo-vax3.csnet>
Posted-Date: Sat, 13 Jul 85 22:53:56 pdt
To: theory@wisconsin.csnet
Subject: FOCS program & information
Sender: udi@wisc-rsch.arpa
Resent-From: udi@wisc-rsch.arpa (Udi Manber)
Resent-To: theory:;@wisc-rsch.arpa
Following is an advance copy of the program for FOCS. Please pass on
to colleagues who do not have access to TheoryNet. Take particular
note of our arrangement with United Airlines and of the pre-FOCS
excursions.
------------------------------------------------------------------------
26th Annual Symposium
on
Foundations of Computer Science
Portland, Oregon
October 21-23, 1985
The symposium is sponsored by the IEEE Computer Society Technical
Committee for Mathematical Foundations of Computing in cooperation with
the ACM Special Interest Group for Automata and Computability Theory.
------------------------------------------------------------------------
INFORMATION
===========
------------------------------------------------------------------------
PORTLAND
The Coastal Range to the west and the snowcapped Cascades to the east,
the majestic Columbia River and the serene Willamette, recollections of
a pioneer heritage, fountains, forests, sculpture, carrousels,
galleries, and roses, every view of Portland is the right one. You'll
find the 26th FOCS is right in the city, within walking distance of fine
restaurants, shops, theatres, but it is right on the riverbank as well,
near parks, esplanades, and jogging paths.
------------------------------------------------------------------------
LOCATION
The technical session, business meeting, reception and lunches will all
be at the Portland Marriott. The Marriott is located downtown at 1401
S.W. Front Avenue, one block south of the Hawthorne Bridge, on the west
bank of the Willamette River. The Tuesday evening banquet-cruise will
be on Portland's new sternwheeler "Columbia Gorge", built to the
classic specifications of the 19th century `darlings' of these
waterways. The sternwheeler will be berthed within a short walk from
the Marriott.
------------------------------------------------------------------------
CLIMATE
Portland's weather is mild. High temperature in October is likely to be
in the low 60's, low in the 40's. On Portland's precipitation, natives
have observed, "If you can't see Mt. Hood, it's raining, if you can,
it's about to rain." On the other hand, rain, if any, is likely to be
light and intermittent; anyway, in the dry moments, you'll appreciate
rain's legacy in the clean air and lush vegetation.
------------------------------------------------------------------------
GETTING THERE
Served by 13 airlines, Portland International Airport is within a twenty
minute drive of downtown. Transportation to the Marriott is available
through "Airporter" bus, taxi, or Tri-Met bus. Frequent "Airporter"
bus service to downtown hotels leaves every 20 minutes; access is
well-marked; also several terminal phones offer direct dial to the
"Airporter" for information on the next bus; cost: $4.00 (one way).
Taxi fare is about $17 via Freeway and about $14 via Sandy Blvd. (Sandy
Blvd. is shorter route but Freeway is faster). Tri-Met bus service is
the cheapest of all but is longer and requires a transfer; take the 72
bus to Sandy Blvd. and transfer to the 14 bus; buses run every 20
minutes; cost: $.75 .
If you are driving, exit Interstate 5 at "City Center" and follow
signs to "Front Avenue".
------------------------------------------------------------------------
AIR TRAVEL TIPS
United Airlines has been designated as the official carrier of FOCS
1985. As a benefit, you can obtain a 35% discount from United's
unrestricted Day Coach fare OR a 15% discount from the United's Easy
Saver fare (requiring a Saturday night stay). For reservations, and
other information, call United at 1-800-521-4041 (your travel agent may
make the call). Be sure to mention FOCS and to give the FOCS account
number 562T.
The lowest available airfares, through any carrier, are usually of the
super-saver type. Typical restrictions may include a 30 day advanced
purchase and stay over Saturday night (check out our pre-FOCS Sunday
excursions - you'll find a Saturday arrival well worthwhile). Note,
though, that there is limited availability of these fares so don't
expect to get what you want if you call exactly 30 days in advance.
(Suggestion: call NOW).
------------------------------------------------------------------------
TOURIST INFORMATION
Suggestions on what-to-see, what-to-do will be available at the
registration desk. For trip-planning in the great Pacific Northwest,
you may want to call: 1-800-233-3306 (Oregon info.) and 1-800-541-9274
(Washington info.)
------------------------------------------------------------------------
MACHTEY AWARD
The Machtey Award is presented for the most outstanding paper written by
a student or students, as judged by the program committee. It includes
a grant to help defray authors' expenses in attending the FOCS
Symposium. Please consider a donation to the Machtey Award fund to
foster this tradition. You may include the donation with your
registration form. A receipt will be provided for tax purposes.
------------------------------------------------------------------------
ACKNOWLEDGEMENTS
Student luncheons are being subsidized by a generous gift from Tektronix
Laboratories.
A reduction in student fees was made possible through the industrial
affiliates of SIGACT and the Technical Committee who support both STOC
and FOCS.
Local arrangements support is being provided by Tektronix Laboratories
and by the Departments of Computer Science and Electrical Engineering at
Portland State University.
------------------------------------------------------------------------
CONFERENCE INFORMATION
Local Arrangements Chair:
Eugene M. Luks
University of Oregon
Department of Computer and Information Science
Eugene, OR 97405
Telephone: 503-686-4408
CSNet: luks@uoregon uucp: ... tektronix!uoregon!luks
Technical Committee Chair: Technical Committee Secretary:
Ashok K. Chandra David W. Bray
IBM Watson Research Center Clarkson College
PO Box 218 Educational Computing
Yorktown Heights, NY 10598 Potsdam, NY 13676
Program Committee Chair: Symposium Coordinator:
Robert E. Tarjan John C. Cherniovsky
AT&T Bell Laboratories Georgetown University
600 Mountain Avenue Department of Computer Science
Murray Hill, NJ 07974 Washington, D.C. 20057
Program Committee:
Manuel Blum, John Hopcroft, Tom Leighton, Jeff Lagarias,
Michael O'Donnell, Charles Rackoff, Larry Ruzzo, Larry Stockmeyer,
Bob Tarjan, Frances Yao
------------------------------------------------------------------------
SCHEDULE
========
------------------------------------------------------------------------
EXCURSIONS. Sunday, October 20
For those who will have arrived in Portland by Sunday morning, there is
a choice of excursions unique to the Pacific Northwest.
Excursion A. Mount St. Helens scenic flight
Groups leave hotel 9am, 11am, and 1pm; 2 hour round
trip
After more than a century of slumber, Washington's Mt. St. Helens awoke
with a blast on May 18, 1980. The massive eruption hurled ash and
debris over twelve miles skyward. The resulting cloud circled the earth
and in some areas turned day into night as it deposited more than a foot
of ash. From the comfort and safety of the aircraft you will view
firsthand the steaming crater and mudflows of the only active volcano on
the American continent. Photographic opportunities abound. Flight time
is about an hour and fifteen minutes. Weather permitting (full refund
if trip is rained out). Price: $38. Payment must accompany preregis-
tration.
Excursion B. The Columbia River Gorge - afoot
Leave hotel at 8am, return about 3pm
This magnificent area is a lovely blend of gently flowing streams, cas-
cading waterfalls and spectacular cliffs. Autumn brings a new character
to "The Gorge" as the foliage is changing and the air is cool and
crisp. You'll be transported by van to the trailhead at the top of
Larch Mountain. From this point, on a clear day, views include Mt. Hood
and Mt. St. Helens. You will be guided on a downhill hike, along an
enchanting forest trail, to Multnomah Falls. The van will be waiting at
trail's end. Weather is not a deterrent, the Gorge is even more beauti-
ful through a drizzle (rain gear will be provided, if needed). Price:
$25 (includes box lunch). Payment must accompany preregistration.
------------------------------------------------------------------------
RECEPTION. Sunday, October 20, 9pm-midnight
Salon E, Level LL1
------------------------------------------------------------------------
REGISTRATION. Sunday, 6pm-9pm & Monday, 7:30am-noon
Foyer, Level LL1
------------------------------------------------------------------------
SESSION 1. Monday Morning, October 21
Chair: Manuel Blum, Univ. of California, Berkeley
Salon F, Level LL1
8:30 am Separating the Polynomial-Time Hierarchy by Oracles
Andrew Chi-Chih Yao, Stanford Univ.
8:50 am Deterministic Simulation of Probabilistic Constant Depth Cir-
cuits
Avi Wigderson and Miklos Ajtai, IBM Research, San Jose
9:10 am Amplification of Probabilistic Boolean Formulas
Ravi B. Boppana, MIT
9:30 am On Networks of Noisy Gates
Nicholas Pippenger, IBM Research, San Jose
9:50 am How Easy is Local Search?
Christos H. Papadimitriou, Stanford Univ. and NTU Athens, and
David S. Johnson and Mihalis Yannakakis, AT&T Bell Labs
10:10 am Coffee Break - Foyer
10:30 am Identification Is Easier Than Decoding
Joseph Ja'Ja', Univ. of Maryland
10:50 am Three Theorems on Polynomial Degrees of NP-Sets
Klaus Ambos-Spies, Universitat Dortmund
11:10 am Simulating Two Pushdown Stores by One Tape in O(n**1.5) Time
Ming Li, Cornell Univ.
11:30 am Nondeterministic versus Probabilistic Linear Search Algorithms
Friedhelm Meyer auf der Heide, IBM Research, San Jose
11:50 am The Complexity of Facets Resolved
David Wolfe, Cornell Univ., and Christos H. Papadimitriou,
Stanford Univ. and NTU Athens
------------------------------------------------------------------------
12:10 pm Lunch - Salon E
------------------------------------------------------------------------
SESSION 2. Monday Afternoon, October 21
Chair: Frances Yao, Xerox PARC
Salon F
<Beverages provided in Foyer from 3:00 to 4:30>
2:00 pm Using Dual Approximation Algorithms for Scheduling Problems:
Theoretical & Practical Results
Dorit S. Hochbaum, Univ. of California, Berkeley, and David
B. Shmoys, Harvard Univ.
2:20 pm A Scaling Algorithm for Weighted Matching on General Graphs
Harold N. Gabow, Univ. of Colorado
2:40 pm An All Pairs Shortest Path Algorithm with Expected Running
Time O((n**2)log n)
Tadao Takaoka, Ibaraki Univ., and Alistair Moffat, Univ. of
Canterbury, New Zealand
3:00 pm Recognizing Circle Graphs in Polynomial Time
Kenneth J. Supowit and Csaba P. Gabor, Princeton Univ., and
Wen-Lian Hsu, Northwestern Univ.
3:20 pm Why Certain Subgraph Computations Require Only Linear Time
A. L. Wong, M. W. Bern and E. L. Lawler, Univ. of California,
Berkeley
3:40 pm Efficient String Matching in the Presence of Errors
Gad M. Landau, Tel-Aviv Univ.,
and Uzi Vishkin, NYU and Tel-Aviv Univ.
4:00 pm The Least Weight Subsequence Problem
D. S. Hirschberg and L. L. Larmore, Univ. of California,
Irvine
4:20 pm Motion Planning in the Presence of Moving Obstacles
John H. Reif, Harvard Univ., and Micha Sharir, Tel-Aviv Univ.
4:40 pm Visibility-Polygon Search and Euclidean Shortest Paths
Hiroshi Imai, Univ. of Tokyo, Tetsuo Asano, Osaka Electro-
Communication Univ., Takao Asano, Sophia Univ., Tokyo,
Leo Guibas, DECSRC, and John Hershberger, Stanford Univ.
5:00 pm Slimming Down Search Structures: A Functional Approach to
Algorithm Design
B. Chazelle, Brown Univ.
5:20 pm The Complexity of Recognizing Polyhedral Scenes
Christos H. Papadimitriou, Stanford Univ. and NTU Athens, and
Lefteris Kirousis, Univ. of Patras and NTU Athens
------------------------------------------------------------------------
9:00 pm Business Meeting - Salon F
------------------------------------------------------------------------
SESSION 3. Tuesday Morning, October 22
Chair: Tom Leighton, MIT
Salon F
8:30 am Layer Changes in VLSI
Alok Aggarwal, IBM Research, Maria Klawe, IBM Research, San
Jose, David Lichtenstein, Yale Univ., Nathan Linial, Hebrew
Univ., and Avi Wigderson, IBM Research, San Jose
8:50 am Area Penalty for Sublinear Signal Propagation Delay on Chip
Paul M. B. Vitanyi, Centrum voor Wiskunde en Informatica
9:10 am On Information Flow and Sorting: New Upper and Lower Bounds
for VLSI Circuits
Richard Cole and Alan R. Siegel, NYU
9:30 am Solving Tree Problems on a Mesh-Connected Processor Array
Mikhail J. Atallah and Susanne E. Hambrusch, Purdue Univ.
9:50 am Solving Some Graph Problems with Optimal or Near-Optimal
Speedup on Mesh-of-Trees Networks
Ming-Deh A. Huang, IBM Research
10:10 am Coffee Break - Foyer
10:30 am Randomized Routing on Fat-Trees
Charles E. Leisersen and Ronald L. Greenberg, MIT
10:50 am Distributed BFS Algorithms
Robert G. Gallager and Baruch Awerbuch, MIT
11:10 am An Almost Linear Time and O(n log(n) + e) Messages Distributed
Algorithm for Minimum-Weight Spanning Trees
Francis Chin and H. F. Ting, Univ. of Hong Kong
11:30 am Byzantine Agreement in Constant Expected Time (and Trusting No
One)
Paul Feldman and Silvio Micali, MIT
11:50 am Geometrical Realizations of Set Systems and Probabilistic Com-
munication Complexity
N. Alon, MIT, P. Frankl, CNRS, Paris, and V. Rodl, FJFI, CVUT.
------------------------------------------------------------------------
12:10 am Lunch - Salon E
------------------------------------------------------------------------
SESSION 4. Tuesday Afternoon, October 22
Chair: Michael J. O'Donnell, Univ. of Chicago
Salon F
<Beverages provided in Foyer from 3:00 to 4:30>
2:00 pm Robin Hood Hashing
Pedro Celis, Per-Ake Larson and J. Ian Munro, Univ. of Water-
loo
2:20 pm Dynamic Monotone Priorities on Planar Sets
Michael J. Fischer, Yale Univ., and Michael S. Paterson,
Warwick Univ.
2:40 pm Design and Analysis of Dynamic Huffman Coding
Jeffrey Scott Vitter, Brown Univ.
3:00 pm Average Case Lower Bounds on the Construction and Searching of
Partial Orders
Harry G. Mairson, INRIA
3:20 pm On Minima of Functions, Intersection Patterns of Curves, and
Davenport-Schinzel Sequences
Ron Livne and Micha Sharir, Tel-Aviv Univ.
3:40 pm Inferring the Structure of Markov Chain From Its Output
Steven Rudich, Univ. of California, Berkeley
4:00 pm Automatic Verification of Probabilistic Concurrent Finite-
State Programs
Moshe Y. Vardi, Stanford Univ.
4:20 pm Partial Polymorphic Type Inference is Undecidable
Hans-J. Boehm, Rice Univ.
4:40 pm Fixpoint Extensions of First-Order Logic
Yuri Gurevich, Univ. of Michigan, and Saharon Shelah, Hebrew
Univ.
5:00 pm Equivalences and Transformations of Recursive Definitions
B. Courcelle, Universite de Bordeaux
------------------------------------------------------------------------
6:15 pm Banquet & Cruise - "Columbia Gorge" Sternwheeler, Willamette
River
------------------------------------------------------------------------
SESSION 5. Wednesday Morning
Chair: Jeff Lagarias, AT&T Bell Labs.
Salon F
8:30 am A Private Interactive Test of a Boolean Predicate and
Minimum-Knowledge Public-Key Cryptosystems
Stuart Haber, Columbia Univ., Zvi Galil, Columbia Univ. and
Tel Aviv Univ., and Moti Yung, Columbia Univ.
8:50 am A Robust and Verifiable Cryptographically Secure Election
Scheme
Josh D. Cohen and Michael J. Fischer, Yale Univ.
9:10 am Verifiable Secret Sharing and Achieving Simultaneity in the
Presence of Faults
Baruch Awerbuch, Benny Chor, Shafi Goldwasser and Silvio
Micali, MIT
9:30 am The Bit Extraction Problem or t-Resilient Functions
Benny Chor, MIT, Joel Fridmann, Univ. of California, Berkeley,
Oded Goldreich and Johan Hastad, MIT, Steve Rudich and Romana
Smolenski, Univ. of California, Berkeley
9:50 am Collective Coin Flipping, Robust Voting Games and Minima of
Banzhaf Values Nathan Linial and Michael Ben-Or, Hebrew Univ.
10:10 am Coffee Break - Foyer
10:30 am Random Polynomial Time is Equal to Slightly-Random Polynomial
Time
Umesh V. Vazirani, Univ. of California, Berkeley, and Vijay
V. Vazirani, Cornell Univ.
10:50 am Unbiased Bits from Weak Sources of Randomness
Benny Chor and Oded Goldreich, MIT
11:10 am Factoring with Cyclotomic Polynomials
Eric Bach, Univ. of Wisconsin, and Jeffrey Shallit, Univ. of
Chicago
11:30 am Computing with Polynomials Given by Straight-Line Programs II:
Factorization
Erich Kaltofen, Rensselaer Polytechnic Inst.
11:50 am An Application of Simultaneous Approximation in Combinatorial
Optimization
Eva Tardos, Res. Inst. for Telecommunications, Hungary, and
Andras Frank, Eotvos Lorand University, Hungary
------------------------------------------------------------------------
12:10 pm Lunch - Salon E
------------------------------------------------------------------------
SESSION 6. Wednesday Afternoon
Chair: Larry Ruzzo, Univ. of Washington
Salon F
<Beverages provided in Foyer from 3:00 to 4:30>
2:00 pm Ears and Branchings in Parallel
L. Lovasz, Eotvos Lorand University, Hungary
2:20 pm Parallel Computational Geometry
A. Aggarwal, IBM Research, B. Chazelle, Brown Univ., L. Guibas,
DECSRC, and C. O'Dunlaing and C. Yap, NYU
2:40 pm Parallel Tree Contraction and Its Application
Gary L. Miller, USC, and John Reif, Harvard Univ.
3:00 pm Improved Processor Bounds for Algebraic and Combinatorial
Problems in RNC
Zvi Galil, Columbia Univ. and Tel-Aviv Univ., and
Victor Y. Pan, SUNY at Albany
3:20 pm Optimal Parallel Graph Algorithms for Integer Sorting and
Graph Connectivity
John H. Reif, Harvard Univ.
3:40 pm Fast Parallel Computation with Permutation Groups
Eugene M. Luks, Univ. of Oregon, and Pierre McKenzie,
Universite de Montreal
4:00 pm Algebraic Cell Decomposition in NC
Dexter Kozen, IBM Research, and Chee-Keng Yap, NYU
4:20 pm Fast and Efficient Algorithms for Sequential and Parallel
Evaluation of Polynomial Zeros and of Matrix Polynomials
Victor Y. Pan, SUNY at Albany
4:40 pm The Complexity of Parallel Sorting
Friedhelm Meyer auf der Heide and Avi Wigderson, IBM Research,
San Jose
5:00 pm The Complexity of Parallel Computation on Matroids
Richard M. Karp, Univ. of California, Berkeley, Eli Upfal,
Stanford Univ., and Avi Wigderson, IBM Research, San Jose
-----------------------------------------------------------------------
RESERVATIONS - REGISTRATION
===========================
-----------------------------------------------------------------------
=================================================================
cut here
=================================================================
HOTEL RESERVATION FORM
Send completed form to:
FOCS Symposium
Portland Marriott Hotel
1401 S.W. Front Avenue
Portland, OR 97201
Telephone: (503) 226-7600
Accommodations desired:
Single (1 Bed, 1 Person) $56 [ ]
Double (1 Bed, 2 Persons) $66 [ ]
Double Double (2 Beds, 2 Persons) $66 [ ]
Triple (3 Persons) $76 [ ]
Quads (4 Persons) $86 [ ]
Student Rates*
Single $42 [ ]
Double $42 [ ]
Triple $45 [ ]
Quad $48 [ ]
*Limited number of rooms available at special rates for full-time
students.
Name ←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←
Address ←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←
City ←←←←←←←←←←←←←←←←←←←←←←←←← State ←←←←←←←←←←←← Zip ←←←←←←←←←←←
Arrival date/time ←←←←←←←←←←←←← Departure date/time ←←←←←←←←←←←←←
To avoid duplications of reservations, please submit only one
form when sharing accommodations with one or more individuals.
Names of Persons Sharing Accommodations
←←←←←←←←←←←←←←←←←←←←←←←←←←←←←← ←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←
←←←←←←←←←←←←←←←←←←←←←←←←←←←←←← ←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←
Please be sure your reservation reaches the hotel 21 days in
advance to insure your accommodations. Otherwise rooms will
will be provided on a space available basis. For arrival after
6 pm, indicate method of guarantee (or first night's deposit
enclosed [ ] ).
Credit Card ←←←←←←←←←← # ←←←←←←←←←←←←←←←←←←←←←←←← Exp. ←←←←←←←←←
Signature ←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←
Name/Firm ←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←
Address ←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←
City ←←←←←←←←←←←←←←←←←←←←←←←←← State ←←←←←←←←←←←←← Zip ←←←←←←←←←
=================================================================
cut here
=================================================================
ADVANCE REGISTRATION FORM
Registration fee includes techical sessions, a copy of the
proceedings, coffee and beverages during sessions, the Sunday
evening reception and three luncheons. The non-student fees
include the Tuesday evening banquet-cruise. Advance Registration
closes Friday, September 27, 1985. All non-student fees will be
$35 higher after close of advance registration. Requests for
refunds will be honored until 9/27/85.
Fees (Check one) Before 9/27 After 9/27
Member of SIGACT or $125 $160 $←←←←←
IEEE Computer Society
Non-members $160 $195 $←←←←←
Full-time student $ 30 $ 40 $←←←←←
(Be sure to bring verification of student
status to registration).
Machtey Fund (contribution) $←←←←←
Excursions (mutually exclusive) Sunday, October 20
A. Mt. St. Helens Scenic Flight $38 $←←←←←
Guest(s) ←←←←← @ $38 $←←←←←
Please rank choice of times (1-3):
9 am ←←←←← 11 am ←←←←← 1 pm ←←←←←
B. The Columbia River Gorge - afoot $25 $←←←←←
Guest(s) ←←←←← @ $25 $←←←←←
TOTAL ENCLOSED $←←←←←
Make checks payable to FOCS. Please pay in U.S. funds. Send
check and completed form to:
FOCS
Computer & Information Science Dept.
University of Oregon
Eugene, OR 97403
Name ←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←
Affiliation ←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←
Address ←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←
City ←←←←←←←←←←←← State ←←← Zip ←←←←← Phone ←←←←←←←←←←←←←←
Electronic mail address ←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←
Dietary restrictions: ←←←←←← Kosher ←←←←←← Vegetarian
=================================================================
∂15-Jul-85 1049 NILSSON@SU-SCORE.ARPA [James F. Gibbons <GIBBONS@SU-SIERRA.ARPA>: Overhead rate for 86,87,88]
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 15 Jul 85 10:49:11 PDT
Date: Mon 15 Jul 85 10:21:05-PDT
From: Nils Nilsson <NILSSON@SU-SCORE.ARPA>
Subject: [James F. Gibbons <GIBBONS@SU-SIERRA.ARPA>: Overhead rate for 86,87,88]
To: faculty@SU-SCORE.ARPA
Message-ID: <12127233108.23.NILSSON@SU-SCORE.ARPA>
This item will be discussed at a meeting I will attend on Friday, Jul 19.
I'm collecting comments. -Nils
---------------
Return-Path: <GIBBONS@SU-SIERRA.ARPA>
Received: from SU-SIERRA.ARPA by SU-SCORE.ARPA with TCP; Mon 1 Jul 85 22:26:29-PDT
Date: Mon 1 Jul 85 22:24:37-PDT
From: James F. Gibbons <GIBBONS@SU-SIERRA.ARPA>
Subject: Overhead rate for 86,87,88
To: xcomx@SU-SIERRA.ARPA
cc: cloutier@SU-SIERRA.ARPA
Friends:
The University is trying to negotiate an overhead rate for the next
three years. They are trying to operate with a "full recovery" policy,
with some estimates of depreciation, etc. Naturally any new buildings
that are built will increase the overhead rate, due to the increased
depreciation expense.
The proposal is to use an overhead rate schedule of 69% for the
coming year, and 73% for each of the two following years. Actually
74% for the two following years is what will be required for "full
recovery", according to Frank Riddle.
The question is whether this is the straw that will break the
camel's back. My guess is that the amount of money that we raise on
research contracts will stay about the same, so the University will
get a slightly larger fraction of a roughly fixed pie, while the PIs
get a smaller fraction. This will reduce the volume of research, of
course, and possibly also cause people to attempt to reduce their
effective overhead rates by seeking gifts for research that bear no
overhead.
We can argue to keep the overhead rate fixed, but that will require
belt tightening elsewhere.
I would like to hear from each of you as soon as possible as to the
effects you expect given the rolling increase suggested. I know this
will not be a pleasant subject to think about, but be as reasonable as
you can. It would be good if we could fix the rate with the Fed for a
three year period so we don't have to go through this every summer.
The depreciation allowance is spent in renovating buildings, though we
are generating M dollars and spending only N, with M>N. The excess
goes into unpredictable things, including McCullough renovation, a new
course in Western Culture, etc. We could change the entire accounting
scheme and make each school sit on its own bottom, but the changeover
would not be trivial, nor necessarily in the best interests of the
UNiversity overall.
What are your thoughts?
-------
-------
∂15-Jul-85 1443 LIBRARY@SU-SCORE.ARPA New Books In The Math/CS Library
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 15 Jul 85 14:43:34 PDT
Date: Mon 15 Jul 85 14:30:59-PDT
From: C.S./Math Library <LIBRARY@SU-SCORE.ARPA>
Subject: New Books In The Math/CS Library
To: su-bboards@SU-SCORE.ARPA
cc: faculty@SU-SCORE.ARPA
Message-ID: <12127278600.39.LIBRARY@SU-SCORE.ARPA>
Fundamentals Of The Average Case Analysis Of Praticular Algorithms.
by Rainer Kemp. QA76.6.K453 1984
Computer Archictecture. 3rd ed. by Foster and Iberall. QA76.9.A73.F67 1985.
Modula-2; a Software Development Approach by Ford and Wiener. QA76.73.M63
.F67 1985.
Paragon: a Language Using Type Hierarchies for the Specification, Implementa-
tion and Selection of Abstract Data Types. by Sherman QA76.73.P19.S54 1985.
Local Area Networks: An Advanced Course edited by Hutchison, Mariani and
Shepherd. TK5105.7.L64 1985.
Lisp Programming. by I. Danicic. QA76.73.L23D36 1983
Principles of Functional Programming. by Glaser, Hankin, and Till.
QA76.6.G558 1984.
Computational Functional Analysis. by Moore. QA320.M67 1985.
Computer Organization and Programming With an Emphasis on the Personal
Computer. by Gear. 4th ed. QA76.6.G38 1985.
An Introduction to Formal Program Verification by Mili. QA76.6.M5217 1985
Multiple Processor Systems for Real-Time Applications. by Liebowitz and Carson
QA76.9.D5L5 1985.
Advances in Cryptology. Proceedings of Cryto 83. ed. by David Chaum.
QA76.9.A25C79 1983.
Interconnection Networks for Large-Scale Parallel Processing by Siegel.
TK5105.5.S54 1985.
Application Debugging; an MVS abend handbook for COBOL, Assembly, PL/1,
and FORTRAN Programmer. by Binder. QA76.6.B56 1985.
Advances In Office Automation. HF5547.5.A28 1985. v.1. ed. by Quinn.
HF5547.5.A28 1985 v.1
MacPaint; drawing, drafting, design. by Schmieman. T385.S29 1985.
H. Llull
-------
∂15-Jul-85 1518 ACUFF@SUMEX-AIM.ARPA SLUG '85 Notes
Received: from SUMEX-AIM.ARPA by SU-AI.ARPA with TCP; 15 Jul 85 15:17:20 PDT
Date: Mon 15 Jul 85 15:04:32-PDT
From: Richard Acuff <Acuff@SUMEX-AIM.ARPA>
Subject: SLUG '85 Notes
To: ksl-lisp-machines@SUMEX-AIM.ARPA
Message-ID: <12127284707.23.ACUFF@SUMEX-AIM.ARPA>
This is a summary of the Symbolics Lisp Users Group national
meeting in SF, June '85, Rich Cohen, Chairman. Much of the first day
was given to talks from Symbolics representatives.
George Gosselin from software support claims that the base of soft
support people in the field is expanding such that customers should
turn to Field Office Software Support (FOSS) before going to the Home
Office Software Support (HOSS). Several people complained at this
point about being forced to have hardware support to get software
support.
Mike Hilgenberg of Customer Service (hardware maint) claimed that
his group is also getting better. They are hoping to get less than 1
hour call back times on service requests. There was much todo over
the extreme cost of service contracts, but little comment about
reducing it.
There were several future product "announcments", including a low
cost dot-matrix printer, the fabled IFU, another cheap laser printer
to replace the LGP, and Prolog. Release 6.1 is due out in October,
and should include an incremental disk save system so that only
changed pages need to be saved to build a world on top of another, a
new FEP version, more printer support, and a lexical debugger.
There was a *lot* of comment from the user community about the
revised source code policy. The policy classifies source files into
three categories: basic, optional, and restricted. Basic sources are
shipped to all customers. Optional sources can be obtained via
special request to Symbolics. Restricted sources are not available.
Obviously, the users were not pleased with this new restrictive
attitude, though nobody had a good suggestion to keep other vendors
from copying Symbolics source almost verbatim. There was, however,
strong indication that, if need was demonstrated, special deals could
be worked on a case by case basis. Keep this in mind if you need to
frob something that we don't have.
I think that's most of the meat of the meeting. Most everything
else presented is already common knowledge here (eg. Zetalisp is
moving towards Common Lisp).
-- Rich
-------
∂15-Jul-85 1518 DAVIES@SUMEX-AIM.ARPA Paper mentioned in PARSYM
Received: from SUMEX-AIM.ARPA by SU-AI.ARPA with TCP; 15 Jul 85 15:18:25 PDT
Date: Mon, 15 Jul 1985 15:06 PDT
Message-ID: <DAVIES.12127285120.BABYL@Sumex>
From: DAVIES@Sumex
To: AAP@Sumex
Subject: Paper mentioned in PARSYM
cc: Davies@Sumex
In the first PARSYM digest, Phil Thrift of TI strongly recommended a
paper by Ehud Shapiro: "Systolic Progrmaming: A Paradigm of Parallel
Processing". I put a copy of the paper in the PAI library and left
a few additional copies on the table in 1105. Take one if you're
interested.
-- Byron
∂15-Jul-85 1520 ACUFF@SUMEX-AIM.ARPA TI Compact Lisp Machine Talk Summary
Received: from SUMEX-AIM.ARPA by SU-AI.ARPA with TCP; 15 Jul 85 15:20:36 PDT
Date: Mon 15 Jul 85 15:07:53-PDT
From: Richard Acuff <Acuff@SUMEX-AIM.ARPA>
Subject: TI Compact Lisp Machine Talk Summary
To: ksl-lisp-machines@SUMEX-AIM.ARPA
Message-ID: <12127285317.23.ACUFF@SUMEX-AIM.ARPA>
This is a review of the talk given by TI about their Compact Lisp
Machine (CLM). The DARPA sponsered project is called Hummingbird, and
is due to be delivered in June, 1986. TI claims that the project is
on schedule, and will be delivered on time.
In the machine, there are 4 main modules built around a nu-bus--the
simple MIT developed bus used in most MIT style Lisp machines. There
are 16 slots on the bus. The memory module is of fairly standard
high-density design, with 4MB per module. Peripheral interfacing is
via a multibus interface module which maps the multibus address and
data spaces into the nu-bus. Of course, the intent is to allow off
the shelf multibus components to plug into the CLM. The cache is 2
way set associative with 32KB/set, 16b/blk, uses write through, and is
indexed with virtual addresses. There's a 2 set address translation
cache (didn't catch the size). Page size is 1KB. There is some
hardware for GC assist, but I didn't get what it was doing, other than
something to improve cache and address translation performance during
GC; tweaking the sets, I imagine.
The processor module contains the nu-bus interface for the CPU and
cache, and the CPU chip. The chip is to be in a square pin grid
package with several hundred pins. The CPU architecture is a fairly
normal 3 bus system, with additional hardware to aid in fast
extraction/manipulation of tag fields. 60% of the chip area is taken
up by two large scratch areas, and a stack buffer, with the rest given
over to logic and control. The CLM is to have a 16k RAM control
store.
TI hopes to have the CLM running macro code identical to the
Explorer, with the resultant portability. Indeed, the sales people
were talking about swapping Explorer processor boards with CLM modules
to upgrade Explorers.
The initial configuration (as delivered to DARPA) will have an
Explorer and a CLM in one box, with the Explorer used to control the
CLM over a maintenance bus. However, since the CLM chip is targeted
towards embedded systems (rather than development), they spent some
time talking about how it will be ruggedized for use in military
systems.
There was some speculation about future uses for such a chip as in
a "delivery vehicle", in multi-processor configurations, and in truly
personal Lisp machines, but TI was, in general, saying, "Yes, that's
possible, but we have no plans to do it" (while the sales people said,
"Oh, sure we'll have that.", of course).
I have very little more information, but if you have questions, get
in touch and I'll try to help.
-- Rich
-------
∂15-Jul-85 1649 @SUMEX-AIM.ARPA:PEDNAULT@SRI-AI.ARPA PhD Orals -- Plan Synthesis
Received: from SUMEX-AIM.ARPA by SU-AI.ARPA with TCP; 15 Jul 85 16:48:53 PDT
Received: from SRI-AI.ARPA by SUMEX-AIM.ARPA with TCP; Mon 15 Jul 85 16:41:33-PDT
Date: Mon 15 Jul 85 16:42:24-PDT
From: Ed Pednault <PEDNAULT@SRI-AI.ARPA>
Subject: PhD Orals -- Plan Synthesis
To: SIGlunch@SUMEX-AIM.ARPA
PHD ORALS
Toward a Mathematical Theory of Plan Synthesis
Edwin P.D. Pednault
Electrical Engineering
Thursday, July 18, 2:15pm
Margaret Jacks Hall, Room 146
ABSTRACT
Planning problems have the following form: given a set of goals, a set
of allowable actions and a description of the current state of the world,
find a sequence of actions that will transform the world from its current
state to a state in which all of the goals are satisfied. This talk is a
presentation of my thesis research which examines the question of how to
solve planning problems automatically. The question of plan synthesis will
be addressed from a rigorous, mathematical standpoint in contrast to the
informal and highly experimental treatments found in most previous works.
By introducing mathematical rigor, it has been possible to unify many
existing ideas in automatic planning, showing how they arise from first
principles and how they may be applied to solve a much broader class of
problems than had previously been considered. In addition, some entirely
new ideas have been developed and a number of theorems have been proved
that further our understanding of the synthesis problem. The talk will
concentrate on my techniques for plan synthesis with only a brief summary
of the other contributions of my research. A mathematical framework for
studying planning problems will be introduced and a number of theorems
will be presented that form the basis for the synthesis techniques.
These theorems will then be combined with a least-commitment search
strategy to obtain a solution method that unifies and generalizes
means-ends analysis, opportunistic planning, goal protection, goal
regression, constraint formulation/propagation, hierarchical planning
and nonlinear planning.
-------
∂15-Jul-85 1704 ACUFF@SUMEX-AIM.ARPA Name Change for the List
Received: from SUMEX-AIM.ARPA by SU-AI.ARPA with TCP; 15 Jul 85 17:04:36 PDT
Date: Mon 15 Jul 85 17:00:40-PDT
From: Richard Acuff <Acuff@SUMEX-AIM.ARPA>
Subject: Name Change for the List
To: ksl-lisp-machines@SUMEX-AIM.ARPA
Message-ID: <12127305849.23.ACUFF@SUMEX-AIM.ARPA>
From now on, instead of KSL-Lisp-Machines this list will be called
KSL-LispM for naming consistency. This should only affect posters and
people reading the bboard file. Sorry for any inconvenience this
causes.
-- Rich
-------
∂15-Jul-85 2249 PARSYM-Request@SUMEX-AIM.ARPA PARSYM Digest V1 #3
Received: from SUMEX-AIM.ARPA by SU-AI.ARPA with TCP; 15 Jul 85 22:49:40 PDT
Date: 15 Jul 85 1532-PDT
From: Moderator Byron Davies <PARSYM-REQUEST@SUMEX-AIM.ARPA>
Reply-to: PARSYM@SUMEX-AIM.ARPA
Subject: PARSYM Digest V1 #3
To: PARSYM@SUMEX-AIM.ARPA
PARSYM Digest Monday, 15 Jul 1985 Volume 1 : Issue 3
Today's Topics:
Query: Distributed Data Structures
Columbia University Software Job Opening: DADO
Applicative Symbolic Programming Recommended
BBN & MIT Multi-LISP and Parallel Scheme
----------------------------------------------------------------------
Date: Sat 13 Jul 85 19:11:43-EDT
From: Randy Haskins <rh%MIT-EECS@MIT-MC.ARPA>
Subject: Re: PARSYM Digest V1 #2
I work at a small custom software house in Cambridge, MA, and it turns
out that one of the things I have to concern myself with now fits into
the topic of discussion here nicely. We are making a production
system whose main purpose is to parse English text and glean out the
essential information it contains, and present this to a human who
will verify that it is correct. We are using LMI Lambda's to
accomplish this. It turns out that this sort of thing can be broken
down into several tasks, each of which are independent of each other.
The idea is to have the independent tasks performed on different
machines since this allows processor dedication (avoiding the expense
of switching context from one process to another). One of the biggest
problems, however, is sending data objects across a network (CHAOS, to
be exact). Lists are okay, but arrays, and abstractions like
structures and flavors, (Zeta Lisp) are quite a bit more difficult for
a general case. If you know in advance that the slots of your
structure or the instance variables of your flavor (and all its
mixins) are going to be simple readable lisp objects, then it is
pretty simple (I have already written something to do this). If,
however, you have more complicated things (like structures of flavor
instances), it becomes quite a bit more difficult to do this
efficiently, if at all. Have people given much thought to these sorts
of problems? Do people think that global address spaces would be a
solution? If you went to a scheme like global address space, however,
you would have to worry about arbitration when two processes wanted to
modify something, unless you locked everything all the time. I'm
interested to hear people's thoughts about these issues.
Randy Haskins
------------------------------
Forwarded by:
Date: Sun, 14 Jul 85 23:25:17 EDT
From: Steven A. Swernofsky <SASW@MIT-MC.ARPA>
Date: Mon 1 Jul 85 15:56:34-EDT
From: Beverly Dyer <DYER@COLUMBIA-20.ARPA>
To: arpanet-bboards@MIT-MC.ARPA, cu-bboards@COLUMBIA-20.ARPA
Subject: Columbia University Software Job Opening
Columbia University
Department of Computer Science
The DADO project at Columbia University has a full time position available
for a Software Engineer. The DADO project is directed towards design
and development of both hardware and software for the parallel execution
of artificial intelligence programs.
The position requires an undergraduate degree in Computer Science.
We are seeking applicants with extensive experience in LISP, some
background in Artificial Intelligence, particularly in the area of
Expert Systems, experience with and interest in parallelism, and very
good programming skills.
Please respond with a resume to:
Beverly Dyer
457 Computer Science
Columbia University
New York, NY 10027
(212) 280-8109
Arpanet: dyer@columbia-20
Columbia University is an Equal Opportunity/Affirmative Action Employer.
We welcome applications from qualified women and minorities.
------------------------------
Date: Mon, 15 Jul 85 15:34:33 EST
From: munnari!dmtadel.oz!crw@seismo.CSS.GOV (charles watson)
Subject: Applicative Symbolic Programming
Firstly I consider symbolic computing, to be abstract and non-numeric, dealing
with structured objects that match human concepts and images.
Secondly, I'd like to argue the case for side-effect free programming.
I feel that the side-effects of much real-world software are the
result of poor programming practice. John McCarthy's (1959) LISP is
sufficient. All the features added in the name of efficiency have
encumbered their compilers and interpreters. Except for debugging,
side effects have been tolerable in the past. For highly concurrent
systems of the future, side-effects would be catastrophic. I accept
that the cost of re-writing software is a strong reason for downward
compatability to existing non-applicative code, but redesign in a
sound paradigm would take less effort than conversion. The machine
architecture I'm working on will allow setq to avoid re-evaluating
S-exprs but not for explicit register assignment; it will allow
replac* to append non-circular lists, but not to point back to any
object in the ancestral tree. There are those who still believe that
the shortest distance between two pieces of software is a GOTO
statement. In my experience real-world software is easier to develop
and maintain in the structured paradigm. The last project I was
involved in used an applicative hierarchical specification of IC
designs to drive a silicon compiler. The problem would have otherwise
been intractable. Also, processors can be custom designed for
symbolic processing and replicated for the cost of about $10 each
(ignoring the development cost). A system with thousands of these
processors, could cost the same as a Symbolics 3600. Arguments such
as "this side-effect riddled feature gives a 20% performance
improvement" would be irrelevent to the cost-effectiveness of such a
system.
------------------------------
Date: Mon, 15 Jul 85 09:10 EDT
From: Seth Steinberg <sas@BBN-VAX.ARPA>
Subject: BBN & MIT Multi-LISP and Parallel Scheme
A number of us at BBN (Don Allen, Seth Steinberg, Curtis Scott) are
working with a group at MIT (Jim Miller, Bert Halstead et al) on a
parallel implementation of MIT Scheme. At BBN we are doing this for
the Butterfly which is a shared memory system which can contain up to
256 68000 processors connected by a Butterfly switch. At MIT they are
using Halstead's Concert system which is a slightly smaller 68000
shared memory based system.
We have already done some work with Halstead's Multi-LISP which was
written with C and have both studied the behavior of a few algorithms
and experimented with the exotica of a parallel garbage collector.
Multi-LISP runs on both machines. We are now porting the MIT Scheme
system and expect to have a multiple processor version running by the
Butterfly sometime this fall.
The fundamental element of parallelism in both systems (Scheme and
Multi-LISP) is the future which is a combination of a delayed
evaluation data type and parallel execution. When the form:
(future xxx)
is evaluated a process is formed to evaluate the form xxx and a special
LISP value is returned. This value may be assigned or passed as an
argument at any time, but it cannot be "touched" without waiting for
the computation of xxx to complete.
Futures work very well when computing to produce values, but side
effecting must be synchronized by the programmer. The non-symbolic
parallelism people at BBN are attempting to order side effects by the
means of what they call generators in which all side effecting
operations are loosely ordered. It appears that futures are
sufficiently inclusive as to be used to emulate a generator oriented
system.
We are not investigating static analysis, but are assuming that a LISP
macro package could be developed which does the most common parallelism
transforms. We are interested in hearing of other work being done on
LISP (and LISP like) systems which run on parallel processors.
Seth Steinberg
sas -at bbn-vax
------------------------------
End of PARSYM Digest
********************
∂15-Jul-85 2358 @SU-SCORE.ARPA:WIEDERHOLD@SUMEX-AIM.ARPA Re: [James F. Gibbons <GIBBONS@SU-SIERRA.ARPA>: Overhead rate for 86,87,88]
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 15 Jul 85 23:57:57 PDT
Received: from SUMEX-AIM.ARPA by SU-SCORE.ARPA with TCP; Mon 15 Jul 85 23:51:44-PDT
Date: Mon 15 Jul 85 23:52:39-PDT
From: Gio Wiederhold <WIEDERHOLD@SUMEX-AIM.ARPA>
Subject: Re: [James F. Gibbons <GIBBONS@SU-SIERRA.ARPA>: Overhead rate for 86,87,88]
To: NILSSON@SU-SCORE.ARPA
cc: faculty@SU-SCORE.ARPA
In-Reply-To: <12127233108.23.NILSSON@SU-SCORE.ARPA>
Message-ID: <12127380848.12.WIEDERHOLD@SUMEX-AIM.ARPA>
It is obvious that donors of new buildings are donating funds in part to
make it easier for the university to attract research, and not to make it
more difficult. So there is obviously something very wrong in the
funding schemes. Suggestions for a cure should be solicited.
Gio
-------
∂16-Jul-85 1344 SARAIYA@SUMEX-AIM.ARPA Talk on Wednesday re: Distributed Languages
Received: from SUMEX-AIM.ARPA by SU-AI.ARPA with TCP; 16 Jul 85 13:43:40 PDT
Date: Tue 16 Jul 85 13:41:38-PDT
From: Nakul P. Saraiya <SARAIYA@SUMEX-AIM.ARPA>
Subject: Talk on Wednesday re: Distributed Languages
To: aap@SUMEX-AIM.ARPA
Message-ID: <12127531761.13.SARAIYA@SUMEX-AIM.ARPA>
From: David Cheriton <cheriton@Pescadero>
Subject: Talk by Peter Wegner on Language Paradigms
Wed. July 17th at 2:00 pm in MJH 352
(As part of Distributed Systems Group meeting/seminar) Others are welcome.
Language Paradigms for Programming in the Large
Peter Wegner, Brown University
We evaluate and compare three programming language paradigms - the
block structure, object-oriented, and distributed programming paradigms -
by examining the nature of their bricks and their mortar.
(1) block-structure paradigm
Algol, Pascal, Pl/I, Ada, Modula
(2) object-oriented programming paradigm
Simula, Smalltalk, Object Pascal, object-oriented database languages
(3) distributed programming paradigm
Argus, NIL-like languages
Paradigms are described by Thomas Kuhn as patterns of thought that serve
as a model for research among a community of researchers. Each of the
programming language paradigms determines a pattern of thought for both
designers and users of programming languages and therefore qualifies as a
paradigm. By focussing attention on the paradigm rather than on the details
of specific languages within the paradigm, we can explore fundamental concepts
common to classes of related programming languages at an appropriate level of
abstraction.
Each paradigm is defined in terms of a necessary and sufficient set of
paradigmatic requirements that may be used to determine if a given language
is an instance of the paradigm. The block structure paradigm is characterized
in terms of visibility and lifetime conditions. The object-oriented paradigm
is characterized in terms of data abstraction, object types, and inheritance.
The distributed paradigm is characterized in terms of message passing,
concurrency, and access control.
The object-oriented and distributed paradigms complement each other.
Several "object-oriented distributed paradigms" are derived by combining
features of object-oriented and distributed paradigms, using an inheritance
hierarchy for language types that parallels inheritance hierarchies in object-
oriented programming. The requirements for a unified "object-oriented
distributed programming language" are examined.
-------
-------
∂16-Jul-85 1551 DAVIES@SUMEX-AIM.ARPA Meeting tomorrow-- 9:30 am
Received: from SUMEX-AIM.ARPA by SU-AI.ARPA with TCP; 16 Jul 85 15:50:52 PDT
Date: Tue, 16 Jul 1985 15:49 PDT
Message-ID: <DAVIES.12127555105.BABYL@Sumex>
From: DAVIES@Sumex
To: AAP@Sumex
Subject: Meeting tomorrow-- 9:30 am
cc: Davies@Sumex
Byron will talk briefly about implementing QLambda (a shared-memory
programming language) on CARE (a distributed-memory architecture).
Feedback and suggestions will be appreciated.
-- Byron
∂17-Jul-85 1023 EMMA@SU-CSLI.ARPA The Ant Strikes Back
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 17 Jul 85 10:23:34 PDT
Date: Wed 17 Jul 85 10:15:25-PDT
From: Emma Pease <Emma@SU-CSLI.ARPA>
Subject: The Ant Strikes Back
To: bboard@SU-CSLI.ARPA, consultants@SU-CSLI.ARPA, folks@SU-CSLI.ARPA
Tel: 497-3479
Jamie would like to know of anyone who is still having ant problems.
-------
∂17-Jul-85 1428 @SU-CSLI.ARPA:LANSKY@SRI-AI.ARPA NEXT MONDAY'S PLANLUNCH
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 17 Jul 85 14:28:42 PDT
Received: from SRI-AI.ARPA by SU-CSLI.ARPA with TCP; Wed 17 Jul 85 14:12:09-PDT
Date: Wed 17 Jul 85 14:12:17-PDT
From: LANSKY@SRI-AI.ARPA
Subject: NEXT MONDAY'S PLANLUNCH
To: aic-associates@SRI-AI.ARPA, folks@SU-CSLI.ARPA, mugs@SUMEX-AIM.ARPA,
val@SU-AI.ARPA, rsimmons@SRI-KL.ARPA, carnese@SRI-KL.ARPA,
alpert@SU-SUSHI.ARPA, weld@XEROX.ARPA, dmrussell@XEROX.ARPA,
vanLehn@XEROX.ARPA, araya@XEROX.ARPA
Speech Acts and Rationality
Phil Cohen
SRI AI Center
11:00 AM, Monday, July 22
SRI International, Building E, Room EJ232
This talk will describe how a theory of communication can be grounded in a
theory of rational interaction. I will present a formalism,
jointly developed with Hector Levesque, that characterizes how an agent's
beliefs and goals eventually lead to action, and how goals to affect the
beliefs and goals of other agents leads to communication.
Communicative acts will be modelled along the lines of Grice's account
of non-natural meaning. I will show how the speech acts of informing,
requesting, and questioning can be defined (rather than stipulated)
in this framework. Importantly, these definitions will allow one
to distinguish insincere imperatives from true requests, and
exam questions from real questions.
-------
-------
∂17-Jul-85 1608 EMMA@SU-CSLI.ARPA Canon
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 17 Jul 85 16:08:14 PDT
Date: Wed 17 Jul 85 15:54:59-PDT
From: Emma Pease <Emma@SU-CSLI.ARPA>
Subject: Canon
To: bboard@SU-CSLI.ARPA, folks@SU-CSLI.ARPA
Tel: 497-3479
The canon will not be accessible tomorrow (July 18) from 10 am till
1 pm because of repair work on the Kodak machine.
-------
∂17-Jul-85 1616 EMMA@SU-CSLI.ARPA Missing Dissertation
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 17 Jul 85 16:16:24 PDT
Mail-From: MORGAN created at 17-Jul-85 14:50:18
Date: Wed 17 Jul 85 14:50:17-PDT
From: Nannette Morgan <MORGAN@SU-CSLI.ARPA>
Subject: Missing Dissertation
To: Friends@SU-CSLI.ARPA
ReSent-Date: Wed 17 Jul 85 16:04:41-PDT
ReSent-From: Emma Pease <Emma@SU-CSLI.ARPA>
ReSent-To: folks@SU-CSLI.ARPA
Has anyone seen the Linguistics Department's copy of Sharon Veach's
1981 dissertation "Children's Telephone Conversations"? It's been
missing since the beginning of the spring quarter. Please let us
know if you have any knowledge of its whereabouts. Thanks.
Nannette
-------
∂17-Jul-85 1720 PARSYM-Request@SUMEX-AIM.ARPA PARSYM Digest V1 #4
Received: from SUMEX-AIM.ARPA by SU-AI.ARPA with TCP; 17 Jul 85 17:20:40 PDT
Date: 17 Jul 85 1652-PDT
From: Moderator Byron Davies <PARSYM-REQUEST@SUMEX-AIM.ARPA>
Reply-to: PARSYM@SUMEX-AIM.ARPA
Subject: PARSYM Digest V1 #4
To: PARSYM@SUMEX-AIM.ARPA
PARSYM Digest Wednesday, 17 Jul 1985 Volume 1 : Issue 4
Today's Topics:
Distributed Data Structures for Lisp Machines
Definitional Issues
Sending data across a net
What is symbolic computing?
----------------------------------------------------------------------
Date: Tue, 16 Jul 85 15:21 PDT
From: Christopher Garrigues <7thSon@CERRIDWYN.SSF.Symbolics.COM>
Subject: Re: PARSYM Digest V1 #2
Reply-to: 7THSON@RUSSIAN.SPA.Symbolics.COM
Re:
Date: Sat 13 Jul 85 19:11:43-EDT
From: Randy Haskins <rh%MIT-EECS@MIT-MC.ARPA>
[Moderator's paraphrase of query: how do you share complex Lisp
objects between network nodes? Does a global address space help?]
I've got a quick hack that I wrote in Symbolics Common Lisp on the 3600
that allows one to access a flavor object on another machine almost as
if it were on your machine. I say almost, because it has obvious
problems with references to global variables such as *TERMINAL-IO*, but
it will work with networks of objects where the objects may be on any
number of machines within your network. It's written with no attempts
at being efficient and has one minor bug at the moment that I'll fix
whenever I find time to play with my hacks again. (Time? What's that?)
------------------------------
Date: Tue, 16 Jul 85 07:46:47 pdt
From: coraki!pratt@Navajo (Vaughan Pratt)
Subject: Definitional Issues
Subject: PARSYM Digest V1 #3
Date: Mon, 15 Jul 85 15:34:33 EST
From: munnari!dmtadel.oz!crw@seismo.CSS.GOV (charles watson)
Subject: Applicative Symbolic Programming
Firstly I consider symbolic computing, to be abstract and non-numeric,
dealing with structured objects that match human concepts and images.
Excellent! A first attempt (on PARSYM) to define symbolic computing.
Let's try it out.
1. What does "abstract" mean? One definition I like a lot is
"authorized." Computation is abstract when it depends only on what
is authorized by the documentation. Is that what you meant? If so,
why is that in the definition of "symbolic" computing, as opposed to
other kinds of computing? If not, then what do you mean?
2. Lisp has numbers. Does this rule out Lisp as a symbolic language?
3. In PROLOG without equality an inefficient but convenient way to
represent natural numbers is symbolically: 3 = (S (S (S 0))) and so
on. How do you reconcile "symbolic" and "nonnumeric" for such a case?
4. What is an example of a structured object *not* matching human
concepts and images? I find it hard to conceive of or imagine such a
thing. This requirement is surely more appropriate for a definition of
AI computing than symbolic computing.
In short, I see neither neither the rationale for nor the application
of any part of this definition.
Secondly, I'd like to argue the case for side-effect free programming.
...
For highly concurrent systems
of the future, side-effects would be catastrophic.
Speaking as one highly concurrent system trying to side effect another,
I hope I haven't thereby caused a catastrophe. And on behalf of human
society, another highly concurrent system, it would certainly be
interesting, and surely catastrophic in some sense, if we ceased to
have side effects on each other.
Whether you have, or for that matter can identify, side effects is very
dependent on your particular computational paradigm, e.g. the
distinction between functional and imperative programming. As soon as
one starts to explore other paradigms appropriate for concurrency,
e.g. dataflow (of particular interest to me), the concept of
side-effect free programming becomes either irrelevant or meaningless.
About the only sense one could make of it in dataflow would be if one
introduced ESP for processes, i.e. communication by unseen channels.
Sorry not to be more constructive here. For more constructive remarks
in the above spirit see my POPL-83 paper "Five Paradigm Shifts in
Programming Language Design and their Application to Viron, a Dataflow
Programming Language" After having taken off a couple of years helping
out with getting workstations out to people I am just now returning to
academia to continue the design and implementation of Viron, or
something resembling it (in addition to continuing my work on fonts, a
side-effect (hak coff) of my working at Sun). If people would be
interested I'd be happy to make occasional short contributions to this
column expressing the general philosophy behind Viron, which is very
much a parallel and abstract programming language. Since Viron
processes don't have a notion of internal state (how do you define "the
state" of a process consisting of an ocean of ships each loaded with
microprocessors with cycle times measured in nanoseconds, where
"simultaneous" is both physically and practically undefined?) one has
to define "side effect" in a way that does not depend on the notion of
state - in this sense Viron is free of any state-based notion of side
effect. Whether Viron could be called "symbolic" depends on whether we
ever find a workable definition of "symbolic," but it should pass
almost any plausible definition that does not rule out numeric
computation and that does not specify implementation or representation
details.
-v
------------------------------
Date: Tue, 16 Jul 85 08:13:09 pdt
From: coraki!pratt@Navajo (Vaughan Pratt)
Subject: Sending data across a net
Subject: PARSYM Digest V1 #3
Date: Sat 13 Jul 85 19:11:43-EDT
From: Randy Haskins <rh%MIT-EECS@MIT-MC.ARPA>
Subject: Re: PARSYM Digest V1 #2
One of the biggest problems, however, is sending data objects
across a network...
If ... you have ... complicated things (like structures of
flavor instances), it becomes quite a bit more difficult to do this
efficiently, if at all.
This is a great (in both senses) problem. People seem to worry about it
a lot more than they write about it, though.
Do people think that global address spaces
would be a solution?
That's how Apollo does things within one domain - they have a 96-bit
address space. I don't know what their strategy is for passing objects
between domains however - anyone know? Sun's NFS (Network File System)
adopts a quite different strategy: the XDR (eXternal Data
Representation) protocol specifies a representation for data
transmitted by the RPC (Remote Procedure Call) protocol that is
independent of how it is represented internally to the caller and
callee. Two advantages of this are: you don't run out of address
space, and you can tie a heterogeneous collection of computers
together.
If you went to a scheme like global address
space, however, you would have to worry about arbitration when two
processes wanted to modify something, unless you locked everything all
the time.
A beautifully simple protocol for this is the EWP - Exclusive Write
Protocol. By only having one write-permission token, which gets passed
around, you don't have to worry about any other form of locking.
Updates should be atomic, i.e. readers should never be permitted to see
inconsistent (partially updated) databases. A big problem with the
protocol is that the token may get lost, due to its current owner
crashing or hanging on to it unreasonably long.
-v
------------------------------
From: eugene@AMES-NAS.ARPA (Eugene Miya)
Date: 16 Jul 1985 1736-PDT (Tuesday)
Subject: Re: PARSYM Digest V1 #1: RE: What is symbolic computing?
> Date: Mon, 8 Jul 85 09:15:00 pdt
> From: coraki!pratt@Navajo (Vaughan Pratt)
> Subject: What is symbolic computing?
>
> I suppose with quarter of a century of Lisp experience we should
> understand by now what symbolic computing is. Indeed everyone appears
> to be quite happy to talk about symbolic computing as though the
> concept had the unanimous blessing of the last three centuries of
> mathematics.
I work with people who crunch numbers, permit me to play devils advocate
(I do support research on symbolic processing): the only people
who have given their blessing is the LISP community. Let's not
close our eyes to that fact.
> Yet I have been unable to find in the literature a definition of the
> concept that matches the facts.
>
> Discussion based on the premise that the topic is well-defined when it
> isn't is apt to run at cross-purposes.
>
> I therefore challenge this list to agree on a definition of symbolic
> computing.
I realize this is rehashing hallway arguments we have all had:
please define that which is "not" symbolic computing and why we should
make a distinction in these two types of parallel computing. After all
isn't crunching a number the same as manipulating a symbol, and aren't
we possibly creating artificial distinctions of computing types (symbolic
and numeric)?
> Even if this challenge cannot be met (as I expect will be the case), at
> least people will have been made aware of the variety of definitions
> and will know to make allowance for this variety.
>
> -v
I like the idea of academic discussions of this nature. Work can get too
serious at times. I also would like to point out that I just returned from SU
and I have seen copies of Dr. Pratt's TRs on new thoughts for concurrency
models.
--eugene miya
NASA
------------------------------
End of PARSYM Digest
********************
∂17-Jul-85 1740 EMMA@SU-CSLI.ARPA Newsletter July 18, No. 37
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 17 Jul 85 17:40:06 PDT
Date: Wed 17 Jul 85 16:54:11-PDT
From: Emma Pease <Emma@SU-CSLI.ARPA>
Subject: Newsletter July 18, No. 37
To: friends@SU-CSLI.ARPA
Tel: 497-3479
C S L I N E W S L E T T E R
←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←
July 18, 1985 Stanford Vol. 2, No. 37
←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←
A weekly publication of The Center for the Study of Language and
Information, Ventura Hall, Stanford University, Stanford, CA 94305
←←←←←←←←←←←←
CSLI ACTIVITIES FOR *NEXT* THURSDAY, July 25, 1985
12 Noon CSLI Talk
Ventura Hall ``Algebraic Semantics and Logics of Programs''
Conference Room Irene Guessarian, National Center for Scientific
Research, France
(Abstract will be printed next week)
2:15 p.m. CSLI Talk
Ventura Hall ``Abstract Semantic Algebras: Theory and Practice''
Conference Room Peter Mosses, Computer Science Dept., Aarhus University
(Abstract on page 1)
3:30 p.m. Tea
Ventura Hall
←←←←←←←←←←←←
ANNOUNCEMENT
No newsletter was published last week, July 11, since there was
nothing to report. There are no activities this Thursday, July 18.
Please note that Thursday Activities during the summer will generally
consist of one or two talks given by visitors at CSLI and held in the
Ventura Conference room.
←←←←←←←←←←←←
CSLI TALK
``Abstract Semantic Algebras: Theory and Practice''
Thursday, July 25, 2:15 p.m.
It seems possible to improve some of the pragmatic aspects of
denotational semantics by using ``abstract semantic algebras'' (ASAs).
Informally, the elements of ASAs can be considered as ``actions'', and
the operations express fundamental ways of combining actions.
Formally, ASAs are just abstract data types, specified axiomatically
in an algebraic framework.
After introducing ASAs, we consider the foundations and pragmatics
of their specification. We conclude by looking at the semantic
analysis of some of the less trivial constructs of programming
languages. --Peter Mosses
(This talk is the same as the one Peter Mosses will have given on
July 23 at Berkeley.)
!
Page 2 CSLI Newsletter July 18, 1985
←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←
INTERACTIONS OF MORPHOLOGY, SYNTAX, AND DISCOURSE
``Partitive Quantifier Phrases in Serbo-Croatian''
Summary of the Meeting on July 3
Partitive quantifier phrases in Serbo-Croatian pose a number of
problems for a syntactic treatment. Since they consist of a
non-declinable quantifier and a partitive genitive nominal form, it is
in no way obvious which of the constituents acts as phrasal head. I
argue that the quantifier acts as modifier rather than head, by
showing that its behaviour with respect to cliticization and the
possibility of being questioned parallels that of other NP modifiers
in the language. This makes the genitive nominal an obvious candidate
for head, but which leaves unresolved the issue of case-assigner. It
is therefore proposed that partitive genitive is an example of
ungoverned, i.e., semantic case, which itself acts as case-assigner. In
this function the genitive affix acts as an argument-taking predicate
and takes the nominal stem as its argument. This imposes a
doubled-layered f-structure of the partitive nominal, predicting that
its own features can but do not have to be ``visible'' in the broader
syntactic environment. This is exactly what is found as agreement
behaviour of these phrases, which presents supportive evidence for the
proposed analysis. --Draga Zec
←←←←←←←←←←←←
INTERACTIONS OF MORPHOLOGY, SYNTAX, AND DISCOURSE
Summary of the Meeting on July 10
Japanese deverbal nominals show verb-like properties in certain
environments: they assign verbal case and can be modified by an
adverb (`Verbal case' includes nominative, accusative and dative,
i.e., case normally assigned by a verb). These case assignment
phenomena pose a problem for current syntactic theories, which assume
that verbs assign case, while nouns do not. I observe the fact that a
deverbal nominal assigns verbal case only when it is concatenated with
a suffix bearing temporal information, which might be encoded with the
feature [+aspect]. The nominal assigns case when the following two
conditions are satisfied:
(i) The nominal has an argument structure.
(ii) It is concatenated with a suffix which bears an aspectual
feature.
I would like to claim that category membership is not sufficient to
license case assignment, but rather that an aspectual feature is also
necessary. --Masayo Iida
←←←←←←←←←←←←
NEW CSLI COMPUTER DIRECTOR
Richard Cower will start as the new computer director on August 1.
He is presently the Director of Research Computing Facilities for
Columbia University's Department of Computer Science. He has also
worked for both SRI International and Stanford University and is a
Stanford University alumnus.
-------
∂18-Jul-85 0718 FREDERIC@SU-CSLI.ARPA
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 18 Jul 85 07:18:44 PDT
Date: Thu, 18 Jul 1985 07:13 PDT
Message-ID: <FREDERIC.12127985342.BABYL@SU-CSLI.ARPA>
From: FREDERIC@SU-CSLI.ARPA
To: csli-bboard@SU-CSLI.ARPA, folks@SU-CSLI.ARPA,
consultants@SU-CSLI.ARPA
Cc: frederic@SU-CSLI.ARPA
CCRMA's fifth annual Computer Music Festival is happening
today, Thursday 18th July, in Frost amphitheater at 8:15 pm.
Gates open at 7pm. Bring picnic, blankets, pillows, and so on.
Tickets are on sale at tresidder ticket office or at the door.
$6 general and $4 for students and seniors.
I went last year and it was really worth it. Let us see if they
performed miracles with our dandelions !
-frederic.
∂18-Jul-85 0813 BARWISE@SU-CSLI.ARPA ASL Meeting
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 18 Jul 85 08:13:24 PDT
Date: Thu 18 Jul 85 08:09:58-PDT
From: Jon Barwise <BARWISE@SU-CSLI.ARPA>
Subject: ASL Meeting
To: Folks@SU-CSLI.ARPA
Today an tomorrow are the end of the joint CSLI/ASL summer school and
ASL meeting. It is a great success, thanks in large part to the help
of the CSLI staff, computer staff, Jamie, Susi, Dave, Chris, and
especially Ingrid. Over 250 people are very appreciative of what you
have made are are making possible.
This morning there is a symposium on the role of logic in AI, with all
CSLI folk: Stan R, John McC, and David I. This afternoon there will
be a lecture on model theory and resolution in theorem proving, by Ken
Kunen. I have known Ken for 20 years and have never heard him give a
talk that was not very intelligible and interesting. Tomorrow morning
there is a symposium on the uses of types in the study of natural,
computer and mathematical languages (Chierchia, Mcqueen and
Feferman). All are welcome. The lectures on in Jordan 40, Psychology
Building.
Again thanks to everyone who has helped make this a landmark event.
Jon Barwise
-------
∂18-Jul-85 0946 SELLS@SU-CSLI.ARPA pizza for brunch
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 18 Jul 85 09:46:17 PDT
Date: Thu 18 Jul 85 09:41:51-PDT
From: Peter Sells <Sells@SU-CSLI.ARPA>
Subject: pizza for brunch
To: folks@SU-CSLI.ARPA
Anyone who feels like some Ramona's Too pizza is welcome to help themselves
to some--they'll find it in the Ventura kitchen. (A slight miscalculation
last night.) There's even veggie and non-veggie.
Bon Appetit,
Peter
-------
∂18-Jul-85 1353 EMMA@SU-CSLI.ARPA CSLI Talk
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 18 Jul 85 13:53:42 PDT
Date: Thu 18 Jul 85 13:47:14-PDT
From: Emma Pease <Emma@SU-CSLI.ARPA>
Subject: CSLI Talk
To: friends@SU-CSLI.ARPA
cc: bresnan@SU-CSLI.ARPA, sag@SU-CSLI.ARPA, Lauri@SRI-AI.ARPA
Tel: 497-3479
CSLI TALK
``Some Null Subject Constructions in Modern Irish''
Jim McCloskey, University College Dublin
Tuesday, July 23, 1:00 in Ventura Conference Room
The paper discusses a group of related constructions in Modern Irish
which have two characteristics in common. They have subjects which
are phonologically null and which are pleonastic. The paper is
particularly concerned with the interaction between unaccusatives and
a passive construction.
(This talk is sponsored by the the NL2 group)
-------
∂18-Jul-85 1554 DAVIES@SUMEX-AIM.ARPA Next week's Architectures meeting
Received: from SUMEX-AIM.ARPA by SU-AI.ARPA with TCP; 18 Jul 85 15:54:19 PDT
Date: Thu, 18 Jul 1985 15:53 PDT
Message-ID: <DAVIES.12128080066.BABYL@Sumex>
From: DAVIES@Sumex
To: AAP@Sumex
Subject: Next week's Architectures meeting
Since I will be away through Tuesday, Jim Rice has agreed to organize
next Wednesday's meeting. If you'd like to make a presentation or if
you have a suggestion for a discussion topic, please let Jim know.
-- Byron
∂18-Jul-85 1601 @SU-SUSHI.ARPA,@SU-SCORE.ARPA:avi.sjrlvm1%ibm-sj.csnet@csnet-relay.arpa BATS at IBM
Received: from SU-SUSHI.ARPA by SU-AI.ARPA with TCP; 18 Jul 85 16:01:08 PDT
Received: from SU-SCORE.ARPA by SU-SUSHI.ARPA with TCP; Thu 18 Jul 85 15:50:28-PDT
Received: from SU-SUSHI.ARPA by SU-SCORE.ARPA with TCP; Thu 18 Jul 85 15:47:36-PDT
Received: from SU-SCORE.ARPA by SU-SUSHI.ARPA with TCP; Thu 18 Jul 85 15:44:11-PDT
Received: from csnet-relay by SU-SCORE.ARPA with TCP; Thu 18 Jul 85 15:39:38-PDT
Received: from ibm-sj by csnet-relay.csnet id ae11852; 18 Jul 85 18:27 EDT
Date: Thu, 18 Jul 85 14:40:08 PDT
From: Avi Wigderson <avi%ibm-sj.csnet@csnet-relay.arpa>
To: aflb.all@su-score.ARPA
Subject: BATS at IBM
Hi
Here is the complete info about the coming BATS at IBM.
Time: Thursday, August 1, 10 am - 3 pm.
Place: Main Auditorium, Research Building (bldg 28), 5600 Cottle Rd.
San Jose.
To get here, take 101 south, and about 10 miles after it intersects
the 17, get off at the Blossom Hill exit (which is on the right!)
Turn right at the T-junction, which brings you to Blossom hill Rd.
and follow signs to IBM, which involve another 3 right turns.
Then you are on Cottle, and at the second light, which is the main
IBM gate, turn left. Say to the guards you come to attend BATS,
and they will direct you to the research building. You can park in
any prking lot, and the closest is the research visitir lot which
is in front of the building. I will be waiting at the lobby.
If you are late, ask the receptionist to let you into the main auditorium.
Schedule:
10:00 Joel Specer, Suny at Stony Brook
11:00 Noga Alon, BCR
12:00 Lunch
1:00 Umesh Vazirani, MSRI
2:00 Avi Wigderson, IBM
ABSTRACTS
10:00 Joel Spencer, SUNY at Stony Brook
"Probabilistic Methods Made Concrete"
Abstract: Given n sets on n points, there exsists a two-coloring
such that all sets have discrepancy (?#red - #blue?) at most
O( sqrt (n log n) ). This result is shown by a simple probabilistic
argument. We discuss how to implement this argument by a good
(nonprobabilistic) algorithm that computes a coloring.
We also give a proof that there is a coloring for which all
discrepancies are at most O( sqrt (n) ), and discuss why no
correspondig good algorithm is likely to be found.
11:00 Noga Alon, Bell Communications Research
"Geometrical Realization of Set Systems and Probabilistic Communication
Complexity"
Abstract: Let d=d(n) be the minimum d such that fo every sequence of
n subsets F1, F2, ...,Fn of {1,2,...,n}, there exist n points
P1, P2, ...,Pn and n hyperplanes H1, H2, ...,Hn in the d-dimentional
Euclidean space, such that Pj lies on the positive side of Hi iff
j is a member of Fi. Then
n/32 < d < n/2 (1).
This implies that the probabilistic, unbounded error 2-way
complexity of almost all
Boolean functions of 2p variables is between p-5 and
p, thus solving a problem of Yao and another problem of Paturi and Simon.
The proof of (1) combines some known
geometric facts with certain probabilistic
arguments and a theorem of Milnor from real algebraic geometry.
This is joint work with P. Frankl and V. Rodl.
12:00 Lunch
1:00 Umesh Vazirani, MSRI, Berkeley
"Random Polynomial Time is Equal to Slightly-random Polynomail Time".
An important goal of theoretical computer science is to capture
theoretically the notion of a tractable computational problem.
It is currently well-accepted that a problem is tractable if
there is a polynomial time algorithm for it, having the ability
of flipping a fair coin at each step.
This formalization is theoretically appealing, and captures
important problems such as primality testing, which are not
known to be efficiently solvable without randomization.
However, it is beset with a difficulty:
the available sources of randomness such as
Zener diodes, and Geiger counters are imperfect. They do not output
unbiased, independent coin-flips.
Santha and Vazirani introduced a general model for such imperfect sources
of randomness: the slightly-random source.
Determining the usefulness of such a source is of theoretical as well as
practical importance.
Consider the class Slightly-random Polynomial Time
SRp , the class of functions computable in polynomial time using
a slightly-random source. ``Is Rp = SRp?'', where Rp
is the class of problems solvable in polynomial time, using a fair coin.
We give an affirmative answer to this question.
Our result is constructive: given an Rp algorithm for a problem,
we show how to obtain an @SR sub p@ algorithm for it.
.pp
The above result is surprising in view of the following theorem:
One Slightly-random source cannot be used to obtain even a single
random bit. Thus proving SRp = Rp may be a step towards
answering the ``P = Rp ?'' question.
This is joint work with Vijay Vazirani.
2:00 Avi Wigderson, IBM-San Jose
"Deterministic Simulation of Probabilistic Constant-Depth Circuits"
Abstract:
We explicitly construct, for every integer n and epsilon > 0,
a family of functions (psuedo-random bit generators)
f sub <n,epsilon> that take as input (seed) binary strings of length
n sup epsilon and output strings of length n,
with the following property: for a random seed, the pseudo-random
output "looks random" to any polynomial size, constant depth, unbounded
fan-in circuit. Moreover, the functions f sub <n,epsilon>
themselves can be computed by uniform polynomial size, constant depth
circuits.
Some (interrelated) consequences of this result are given below.
1) Deterministic simulation of probabilistic algorithms.
The constant depth analogues of the probabilistic complexity classes
RP and BPP are contained in the deterministic
complexity classes
DSPACE(n sup epsilon ) and DTIME(2 sup <n sup epsilon>)
for any epsilon > 0.
2) Making probabilistic constructions deterministic .
Some probablistic constructions of structures that
elude explicit constructions can be simulated in the above complexity
classes.
3) Approximate counting .
The number of satisfying assignments to a (CNF or DNF) formula,
if not too small, can be arbitrarily approximated in
the above complexity classes.
We also present a result for the special case of depth 2 circuits.
It gives
an interesting algorithm for 3-SAT, whose running time depends
on the fraction of satisfying assignmemts. For example, if the fraction
is fixed, the algorithm finds a satisfying assignment in
polynomial time!
This is joint work with M. Ajtai.
∂18-Jul-85 1637 BERGMAN@SU-SCORE.ARPA 85-86 RA's
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 18 Jul 85 16:37:33 PDT
Date: Thu 18 Jul 85 15:55:30-PDT
From: Sharon Bergman <BERGMAN@SU-SCORE.ARPA>
Subject: 85-86 RA's
To: faculty@SU-SCORE.ARPA
cc: bergman@SU-SCORE.ARPA
Message-ID: <12128080419.26.BERGMAN@SU-SCORE.ARPA>
I would like to begin processing the paperwork for Research Assistants
for the 85-86 academic year. If you plan on supporting any students as
RA's, please send me a list of names with the account numbers to charge
them to. If you don't yet know the account numbers you want to use, please
at least send a list of names. If any students will be supported for less
the full year (Fall, Winter, Spring quarters), please so designate.
Thank you. Sharon Bergman
-------
∂19-Jul-85 1359 CLT mini seminar series
To: "@DIS.DIS[1,CLT]"@SU-AI.ARPA, su-bboards@SU-AI.ARPA
There will be as short series of talks given by visitors to the
computer science department from Japan. The seminars will occur on
Thursdays at 4pm. The announcement for the first seminar is below.
-------------------------------------------------------------------------
Speaker: Prof. Masahiko Sato, University of Tokyo
Title: Typed Logical Calculus
Time: Thursday, July 25, 4:00-5:00 pm
Place: Room 352 Margaret Jacks Hall
(Computer Science Department), Stanford
Abstract
We present a typed formal system QJ which is
intended both as a logical system and a program-
ming system. QJ is a constructive system based on
free intuitionistic logic. QJ is a typed system
where forms play the roles of both formulas and
terms of conventional logical systems.
The logic of QJ is free in the sense that
forms (considered as terms) may fail to denote.
The type structure of QJ is rich enough to include
such data types as integers, lists, trees and
function spaces. A form of QJ, when considered as
a term, becomes a program in the usual sense. As
a programming language, QJ becomes a typed func-
tional language somewhat similar to ML. A form
of QJ, when viewed as a formula, may be used to
specify a program. We can also verify programs in
QJ. By implementing QJ on a computer, we will
have a uniform environment where we can specify,
execute and verify programs.
∂19-Jul-85 1416 @SU-CSLI.ARPA:CLT@SU-AI.ARPA mini seminar series
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 19 Jul 85 14:15:35 PDT
Received: from SU-AI.ARPA by SU-CSLI.ARPA with TCP; Fri 19 Jul 85 14:10:57-PDT
Date: 19 Jul 85 1359 PDT
From: Carolyn Talcott <CLT@SU-AI.ARPA>
Subject: mini seminar series
To: "@DIS.DIS[1,CLT]"@SU-AI.ARPA, su-bboards@SU-AI.ARPA
There will be as short series of talks given by visitors to the
computer science department from Japan. The seminars will occur on
Thursdays at 4pm. The announcement for the first seminar is below.
-------------------------------------------------------------------------
Speaker: Prof. Masahiko Sato, University of Tokyo
Title: Typed Logical Calculus
Time: Thursday, July 25, 4:00-5:00 pm
Place: Room 352 Margaret Jacks Hall
(Computer Science Department), Stanford
Abstract
We present a typed formal system QJ which is
intended both as a logical system and a program-
ming system. QJ is a constructive system based on
free intuitionistic logic. QJ is a typed system
where forms play the roles of both formulas and
terms of conventional logical systems.
The logic of QJ is free in the sense that
forms (considered as terms) may fail to denote.
The type structure of QJ is rich enough to include
such data types as integers, lists, trees and
function spaces. A form of QJ, when considered as
a term, becomes a program in the usual sense. As
a programming language, QJ becomes a typed func-
tional language somewhat similar to ML. A form
of QJ, when viewed as a formula, may be used to
specify a program. We can also verify programs in
QJ. By implementing QJ on a computer, we will
have a uniform environment where we can specify,
execute and verify programs.
∂21-Jul-85 2105 ullman@diablo Hi, I'm back
Received: from SU-AIMVAX.ARPA by SU-AI.ARPA with TCP; 21 Jul 85 21:05:42 PDT
Date: Sun, 21 Jul 85 21:00:20 pdt
From: Jeff Ullman <ullman@diablo>
Subject: Hi, I'm back
To: nail@diablo
1. I note there is a talk by Lloyd on logic programming 10AM
tomorrow (Monday) in 352MJH.
2. While I was away, I received a few papers of possible
interest for presentations at our "reading group" which has
not been doing much reading for lack of volunteers.
One of these is Lee Naish's thesis on (what I would call) sideways
capture rules.
3. If there are no strong objections, could we meet 10AM instead
of 11AM this wednesday? My son is taking his driving test, and
my presence is needed.
At this meeting, I'd like to try to "finalize" the intermediate code,
the "preliminary source language" and the "ultimate source language."
Allen would like to discuss some additional source constructs and
their semantics.
---Jeff
∂22-Jul-85 0010 @SU-CSLI.ARPA:LANSKY@SRI-AI.ARPA PLANLUNCH REMINDER
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 22 Jul 85 00:10:39 PDT
Received: from SRI-AI.ARPA by SU-CSLI.ARPA with TCP; Mon 22 Jul 85 00:02:00-PDT
Date: Mon 22 Jul 85 00:01:45-PDT
From: LANSKY@SRI-AI.ARPA
Subject: PLANLUNCH REMINDER
To: aic-associates@SRI-AI.ARPA, folks@SU-CSLI.ARPA, mugs@SUMEX-AIM.ARPA,
val@SU-AI.ARPA, rsimmons@SRI-KL.ARPA, carnese@SRI-KL.ARPA,
alpert@SU-SUSHI.ARPA, frayman@SUMEX-AIM.ARPA, dmrussell@XEROX.ARPA,
vanLehn@XEROX.ARPA, araya@XEROX.ARPA, frayman@XEROX.ARPA,
suchman@XEROX.ARPA, weld@XEROX.ARPA, mittal@XEROX.ARPA, dekleer@XEROX.ARPA
Speech Acts and Rationality
Phil Cohen
SRI AI Center
11:00 AM, Monday, July 22
SRI International, Building E, Room EJ232
This talk will describe how a theory of communication can be grounded in a
theory of rational interaction. I will present a formalism,
jointly developed with Hector Levesque, that characterizes how an agent's
beliefs and goals eventually lead to action, and how goals to affect the
beliefs and goals of other agents leads to communication.
Communicative acts will be modelled along the lines of Grice's account
of non-natural meaning. I will show how the speech acts of informing,
requesting, and questioning can be defined (rather than stipulated)
in this framework. Importantly, these definitions will allow one
to distinguish insincere imperatives from true requests, and
exam questions from real questions.
-------
-------
∂22-Jul-85 0036 RESTIVO@SU-SCORE.ARPA PROLOG Digest V3 #33
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 22 Jul 85 00:36:02 PDT
Date: Sunday, July 21, 1985 1:28PM
From: Chuck Restivo (The Moderator) <PROLOG-REQUEST@SU-SCORE.ARPA>
Reply-to: PROLOG@SU-SCORE.ARPA
US-Mail: P.O. Box 4584, Stanford CA 94305
Phone: (415) 326-5550
Subject: PROLOG Digest V3 #33
To: PROLOG@SU-SCORE.ARPA
PROLOG Digest Monday, 22 Jul 1985 Volume 3 : Issue 33
Today's Topics:
Query - Image Mode & Readers,
Implementation - Cut,
LP Library - PARLOG Update
----------------------------------------------------------------------
Date: Thu 18 Jul 85 15:00:08-EDT
From: Joe Zachary <Zachary@MIT-XX>
Subject: Image Mode
The DEC-10 Prolog manual explains a hack for
forcing Prolog to output characters in image
mode. Does anyone know if a similar hack
exists for the DECSystem-20? I'd rather my
escape characters didn't get translated to
dollar signs.
-- Joe Zachary
------------------------------
Date: Thu 18 Jul 85 09:19:15-EDT
From: Hilton@RADC-TOPS20
Subject: Implementing readers
I am interested in how different Prolog readers
are implemented, and comparisons of the data
structures produced. Is there a general concencus
on what data structure is best (read best as
"fastest") for supporting the inference mechanism?
Consider the VAX 11/780 as the basic architecture
the Prolog will run on.
I have seen many papers written on implementing the
inference mechanism for Prolog, but no discussions
on readers or efficient data structures for clause
storage. If anyone has any references to papers
that discuss these issues, I would appreciate hearing
from you.
Thanks for thinking about it,
-- Mike Hilton
------------------------------
Date: Mon, 15 Jul 85 17:36:00 bst
From: William Clocksin <WFC%computer-lab.Cambridge@ucl-cs>
Subject: Moss's cut tests.
Running Chris Moss' cut tests on my Prolog-X system
(incremental compiler) gave the following results:
Implementation Test 1 2 3 4 5 6
Prolog-X Y Y Y N N N
Note: this is the same as C-Prolog and DEC-10 interpreter.
I can redesign the system in an hour to answer all
the tests Y if you want.
------------------------------
Date: Thu, 11-Jul-85 18:06:00 PDT
From: (Jeremy West) !uts.am.reading.UUCP!west@Seismo
Subject: PARLOG for C-Prolog - list of built-in predicates
Here is a list of built-in predicates for C-Prolog, v1.5,
for systems which don't provide the 'system' predicate as
required by PARLOG (as per addendum sheet i.e. true if arg
if built-in proceedure). You should change the definition
of sys/1 in file 'parlog' to call built←in/1 instead of
system/1.
I think it's complete, if you find any bugs, please let me know.
-- Jerry
[ this is available as SCORE:<Prolog>PARLOG←BUILTIN.PL -ed ]
------------------------------
Date: Thu, 11-Jul-85 22:22:21 PDT
From: (Jeremy West) uts.am.reading.UUCP!west@Seismo
Subject: PARLOG addendum for C-Prolog versions (again)
The built←in/1 predicate I distributed earlier fails
to treat arithmetic predicates correctly. If they are
not expanded (by use of expand←exprs/2 (sometimes
expanded←exprs/2) they will be arguments to is/1. This
was why I commented them out of my earlier posting. If
expanded, however, they are translated into predicates
of the form $<name>/<arity+1>. These will not be
recognised by built←in/1 as it stands.
[ this file is available as SCORE:<Prolog>PARLOG←ADDENDUM.PL
-ed ]
------------------------------
End of PROLOG Digest
********************
∂22-Jul-85 1001 JF@SU-SUSHI.ARPA bats on August 1
Received: from SU-SUSHI.ARPA by SU-AI.ARPA with TCP; 22 Jul 85 10:01:22 PDT
Date: Mon 22 Jul 85 09:58:45-PDT
From: Joan Feigenbaum <JF@SU-SUSHI.ARPA>
Subject: bats on August 1
To: aflb.su@SU-SUSHI.ARPA
cc: avi%ibm-sj@CSNET-RELAY.ARPA
I will be out of town on August 1; so you are all on your own for BATS.
The talk abstracts and directions for getting to IBM-SJ and to the right
building within IBM-SJ are online in
SUSHI:<JF>BATS.ANNOUNCE
You can pick up a hard copy of that file from an envelope on my office
door at Stanford--Margaret Jacks Hall 325.
Joan
-------
∂22-Jul-85 1346 WEIN@SU-CSLI.ARPA housing needed
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 22 Jul 85 13:46:02 PDT
Date: Mon 22 Jul 85 13:39:20-PDT
From: Gina Wein <WEIN@SU-CSLI.ARPA>
Subject: housing needed
To: folks@SU-CSLI.ARPA
Can anyone help out?
Housing is needed:
FOR: A visiting scholar in Linguistics from U. of Aarhus.
Would like to rent a room for the months of September
and October 1985. Correspond directly to Kim Plunkett,
Hesselvej 3, DK-8240 Risskov, Denmark.
Arnold Zwicky, Visiting Professor in Linguistics, Winter
qtr 85-86. Needs 1-2 bdrm. apt. for 3 months, Jan., Feb.,
March 1986. Correspond directly to Arnold Zwicky,
63 W. Beaumont Rd., Columbus, Ohio 43214.
-------
∂22-Jul-85 1502 EMMA@SU-CSLI.ARPA Ingrid
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 22 Jul 85 15:02:49 PDT
Date: Mon 22 Jul 85 14:50:10-PDT
From: Emma Pease <Emma@SU-CSLI.ARPA>
Subject: Ingrid
To: folks@SU-CSLI.ARPA
Tel: 497-3479
Ingrid is taking a well deserved and well needed 4 week vacation.
Trudy will be taking over for her while she is away.
-------
∂22-Jul-85 2202 PARSYM-Request@SUMEX-AIM.ARPA PARSYM Digest V1 #5
Received: from SUMEX-AIM.ARPA by SU-AI.ARPA with TCP; 22 Jul 85 22:02:18 PDT
Date: 22 Jul 85 1206-PDT
From: Moderator Byron Davies <PARSYM-REQUEST@SUMEX-AIM.ARPA>
Reply-to: PARSYM@SUMEX-AIM.ARPA
Subject: PARSYM Digest V1 #5
To: PARSYM@SUMEX-AIM.ARPA
PARSYM Digest Monday, 22 Jul 1985 Volume 1 : Issue 5
Today's Topics:
New Languages Needed for Parallelism
Apollo address space
What is symbolic computing?
Call for Papers -- Fault-Tolerant Computing
----------------------------------------------------------------------
Date: Wed, 17 Jul 85 17:01:37 EDT
From: Martin Lee Schoffstall <schoff%rpi.csnet@csnet-relay.arpa>
Subject: restructuring LISP
Not to be to simplistic or offensive...........
Are we really buying a whole lot of efficiency from our parallel
machines by putting "warts" on Fortran and LISP that trace their
lineage to the 50's?
For major software investments probably yes, however, wouldn't it be
more reasonable to spend the time on a new language for present and
future developments.
marty
schoff%rpics.csnet@csnet-relay
------------------------------
Subject: Apollo address space...
Date: Fri, 19 Jul 85 09:52:03 EDT
From: Ashwin Ram <Ram@YALE.ARPA>
> Date: Tue, 16 Jul 85 08:13:09 pdt
> From: coraki!pratt@Navajo (Vaughan Pratt)
> Subject: Sending data across a net
>
> Do people think that global address spaces
> would be a solution?
>
> That's how Apollo does things within one domain - they have a 96-bit
> address space.
> -------
That doesn't sound right... a 96-bit address space would imply a 2↑76 Mbyte
or 2↑66 Gbyte memory. (I think there are 2↑76 atoms in the universe, give
or take a few.) Actually the theoretical address space for an Apollo is
32 bits (I checked with our sysop here at Yale), which corresponds to a
4 Gbyte memory, but there seems to be a practical limit on the usable portion
of that of 26 bits (32 Mbytes.)
Ashwin Ram
AI Lab
Yale University
------------------------------
Date: Fri 19 Jul 85 21:25:35-CDT
From: Mayank Prakash <AI.Mayank@MCC.ARPA>
Subject: Re: What is symbolic computing?
> Date: Mon, 8 Jul 85 09:15:00 pdt
> From: coraki!pratt@Navajo (Vaughan Pratt)
> Subject: What is symbolic computing?
>
> I therefore challenge this list to agree on a definition of symbolic
> computing.
>
Here's another attempt at it. First of all, I think that the terms
symbolic computing and numerical computing are different modes of
computing rather than mutually exclusive taxonomical categories.
Then, numerical computing is the mode when the major data elements are
numerical and one is interested in changing the numerical values of
these data elements. That is, both the input to and the output from
the program are mainly numerical. In symbolic computing, one is
interested mainly in manipulating structures. That is, both the input
and the output to the program are structures. Note that in this
definiton one mode of computing does not exclude the other. In fact,
most programs do some of each. It is the predominant activity of the
program that determines it's mode.
One could look at it from a lower level as well. The memory cells in
the computer's data memory (as opposed to the instruction memory)
contain binary values. If they are mostly interpreted as representing
numbers, and the majority of operations that are carried out on them
are numerical, i.e., add, subtract, multiply, shift etc. their values,
then the program is a numerical mode program. If they are mostly
pointers to other cells in memory, and the operations on them are
mainly follow the pointers along, modify them to point to some other
cells etc., then the program is a symbolic mode program.
A characteristic that generally distinguishes the languages for the
two kinds of programs is memory allocation. The languages developed
for numerical processing have mostly static memory allocation schemes.
By that I mean that the data memory is allocated to a procedure (or a
function, or block, whatever you want to call it) upon entry, and
released upon exit. Generally, though not always, the procedure does
not (and in most cases, can not) change its data memory. In contrast,
symbolic processing languages have dynamic memory allocation with
attendant garbage collection. This is necessary for symbolic
computing since in this case one is dealing with structures, which are
essentially pointers pointing at each other in various ways, and since
the main activity here is manipulating these structures, i.e.,
releasing and allocating pointers.
Admittedly these are somewhat vague definitions, but I hope that this
posting will at least spur a discussion on the subject.
- mayank.
------------------------------
[Long message -- last in today's digest -- BD]
Date: Sat, 20 Jul 85 03:27:06 EDT
From: Steven A. Swernofsky <SASW@MIT-MC.ARPA>
Subject: Call for Papers -- Fault-Tolerant Computing
ReSent-from: Arpanet-BBoards-Request at MIT-MC
The organizing committee of the 16th International Symposium on
Fault-tolerant Computing would appreciate if you could broadcast to
the sites on the Arpanet and CSnet the following call for papers:
===========================================================
CALL FOR PAPERS
FTCS16
THE SIXTEENTH INTERNATIONAL
SYMPOSIUM ON FAULT-TOLERANT
COMPUTING
VIENNA, AUSTRIA JULY 1-3, 1986
FTCS is the conference on fault tolerant computing systems. It
encompasses all aspects of specifying, designing, modeling,
implementing, testing, diagnosing and evaluating dependable and fault
tolerant computing systems and their components. In addition to the
established fields of fault tolerance particular emphasis is placed on
papers relating to practical experience with real time systems,
switching systems and transaction systems as well as the application
of artificial intelligence techniques to the solution of problems in
fault tolerance.
Papers relating to the following areas are invited: fault-tolerant
architectures, fault-tolerance in distributed systems and
interconnection networks, artificial intelligence for diagnosis and
maintenance, reliable synchronization, consensus and interprocess
communication in distributed systems, hardware/software tradeoffs in
the design of fault-tolerant systems, design diversity in software and
VLSI, robust programs and data structures, exception handling, recon-
figuration and restart, testing techniques, coverage and test tools
for VLSI components and systems, fault-tolerance aspects of VLSI and
WSI, modeling, verification and experimental evaluation of fault-
tolerant systems, application of fault-tolerance techniques (robotics,
pattern recognition, knowledge based systems etc.), reliability and
safety in real time systems, availability of transaction systems and
electronic switching systems.
Information for Authors:
An abstract of the paper including up to five keywords should be
submitted before October 25, 1985. Submit 6 copies of the paper
(double spaced) before the submission deadline, November 25, 1985.
Papers should be no longer than 5000 words. The first page of each
paper must include the following information: title, the author's
name, affiliations, complete mailing address, telephone number and
electronic mail address where applicable, a maximum 150-words ab-
stract of the paper and up to five keywords (important for the
correct classification of the paper). If there are multiple authors,
please indicate who will present the paper at FTCS-16 if the paper
is accepted. The first page should also indicate that the papers
has been cleared through the author's affiliations. The conference
language is English only.
The general chairman for this symposium is Herman Kopetz, TU Vienna,
Austria. The Program Chairman is Mario Dal Cin, Univ. Tuebingen.
The program committee consists of:
J. A. Abraham (USA) V. K. Agraval (Canada)
T. Anderson (GB) A. Avizienis (USA)
J. Bartlett (USA) W. C. Carter (USA)
F. Cristian (USA) K. E. Grosspietsch (FRG)
J. Hlavicka (CSSR) R. Iyer (USA)
K. H. Kim (USA) G. Le Lann (France)
B. Littlewood (GB) R. Maxion (USA)
D. Morgan (USA) S. Naito (Japan)
B. E. Ossfeldt (Sweden) D. Powell (France)
L. Simoncini (Italy) L. Svobodova (Switzerl.)
Y. Tohma (Japan) K. Trivedi (USA)
U. Voges (FRG) J. Wensley (USA)
Y. W. Yang (China)
------------------------------
End of PARSYM Digest
********************
∂23-Jul-85 0817 EMMA@SU-CSLI.ARPA talk on rplaca wed., 4:15
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 23 Jul 85 08:16:54 PDT
Return-Path: <WALDINGER@SRI-AI.ARPA>
Received: from SRI-AI.ARPA by SU-CSLI.ARPA with TCP; Mon 22 Jul 85 17:15:38-PDT
Date: Mon 22 Jul 85 17:01:31-PDT
From: WALDINGER@SRI-AI.ARPA
Subject: talk on rplaca wed., 4:15
To: AIC-Associates: ;,
CSL: ;, bboard@SRI-AI.ARPA, su-bboards@SU-AI.ARPA, friends@SU-CSLI.ARPA
ReSent-Date: Tue 23 Jul 85 08:13:35-PDT
ReSent-From: Emma Pease <Emma@SU-CSLI.ARPA>
ReSent-To: friends@SU-CSLI.ARPA
when: wednesday, 24 july, 4:15pm
where: aic new conference room ej226 or so
(that's engineering building, sri, 333 ravenswood, menlo park,
opposite pine street intersection)
who and what:
VERIFICATION OF PROCEDURES INVOLVING POINTERS
Dennis de Champeaux
ADAC Laboratory
San Jose, CA. 95138
408/365-2000 x1255
An axiomatization is given for LISP style entities and pointers. This
formalism is employed to give the symbolic execution semantics of a
subset of LISP primitives, embodying the troublemakers RPLACA and
RPLACD. Side effects due to structure sharing and aliasing are
captured as demnstrated by an application.
coffee: in waldinger's office at 3:45 be on time. ek292.
-------
∂23-Jul-85 1432 RICE@SUMEX-AIM.ARPA Wednesday's Architecture Meeting
Received: from SUMEX-AIM.ARPA by SU-AI.ARPA with TCP; 23 Jul 85 14:32:01 PDT
Date: Tue 23 Jul 85 14:29:37-PDT
From: Jim Rice <RICE@SUMEX-AIM.ARPA>
Subject: Wednesday's Architecture Meeting
To: aap@SUMEX-AIM.ARPA
Message-ID: <12129375504.53.RICE@SUMEX-AIM.ARPA>
Well I got stuck with trying to find someone to talk to us all and I've
failed. Thus there will be no meeting tomorrow morning.
As some of you may know the Care system is now running under the new
source code control system that I have written. This is on a somewhat
beta-testish basis but if there is anyone out there, who is interested
to know anything about it, to get a demo, or give me some quick
feedback so that it might be made as useful as possible as quickly as
possible then please do come forward.
For those who are working with the Care sources..
Please, please do not try to circumvent the source code control system.
It should be able to spot the fact that you have anyway but we will
learn nothing and improvements will not happen if you cheat.
Unauthorised file updates are always superseded by the next legal one
anyway so failure to use the system correctly is a quick
way to waste your time.
Rice.
-------
∂24-Jul-85 0926 ullman@diablo meeting
Received: from SU-AIMVAX.ARPA by SU-AI.ARPA with TCP; 24 Jul 85 09:26:44 PDT
Date: Wed, 24 Jul 85 09:18:34 pdt
From: Jeff Ullman <ullman@diablo>
Subject: meeting
To: nail@diablo
Don't forget we meet at 10AM instead of 11AM today.
We'll use 301 MJH; I hope it's available.
∂24-Jul-85 1151 BETSY@SU-CSLI.ARPA year-end report
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 24 Jul 85 11:51:36 PDT
Date: Wed 24 Jul 85 11:49:53-PDT
From: Betsy Macken <BETSY@SU-CSLI.ARPA>
Subject: year-end report
To: researchers@SU-CSLI.ARPA
Our second year report to SDF will be a much smaller version of our
first year report. Susan and Dave are gathering information for the
appendices which will be similar to those in the first year report,
and Tom and I need to write 12 to 15 pages of text. We plan to base
the text on information contained in the proposals you have submitted
and on information in the Examples section of the NSF Workshop Report.
We think that there is sufficient information in these sources for us
to be able to describe both our 1984-85 activities and our future
plans. However, if there is work you particulary want mentioned, will
you send me a paragraph or two about it? Write the paragraphs so that
the SDF Board can understand them. If the information about your work
is available in the sources we plan to use but you particularly want
it mentioned, let me know that too so that it won't get left out in
the process of condensing the information.
We have to have this finished in time for the SDF Board Meeting
on August 12, so will you send me your information as soon as
possible? Early next week would be great.
Thanks.
Betsy
-------
∂24-Jul-85 1201 @SU-CSLI.ARPA:BMOORE@SRI-AI.ARPA PARTY! PARTY! PARTY!
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 24 Jul 85 12:01:40 PDT
Received: from SRI-AI.ARPA by SU-CSLI.ARPA with TCP; Wed 24 Jul 85 11:56:41-PDT
Date: Wed 24 Jul 85 11:55:49-PDT
From: Bob Moore <BMOORE@SRI-AI.ARPA>
Subject: PARTY! PARTY! PARTY!
To: party-list: ;
To celebrate the completion of our remodeling and landscaping, and our
upcomming departure for England, we are having a party. Good food,
good wine, and good conversation will be provided. Time: Sunday,
August 4, 2:00-6:00 PM. Place: 737 DeSoto Drive, Palo Alto.
(Directions provided on request.) If you plan to come, please let us
know, so we know how many people to expect.
Bob and Rita Moore
328-4253
BMOORE@SRI-AI
-------
∂24-Jul-85 1340 @SU-SUSHI.ARPA,@SU-SCORE.ARPA:udi@wisc-rsch.arpa Conference on Reasoning About Knowledge
Received: from SU-SUSHI.ARPA by SU-AI.ARPA with TCP; 24 Jul 85 13:39:56 PDT
Received: from SU-SCORE.ARPA by SU-SUSHI.ARPA with TCP; Wed 24 Jul 85 13:35:27-PDT
Received: from wisc-rsch.arpa by SU-SCORE.ARPA with TCP; Wed 24 Jul 85 13:35:13-PDT
Received: by wisc-rsch.arpa; Wed, 24 Jul 85 14:31:55 cdt
Received: from wisc-crys.arpa by wisc-rsch.arpa; Mon, 22 Jul 85 16:28:57 cdt
Message-Id: <8507222123.AA28985@wisc-crys.arpa>
Received: from csnet-relay.arpa by wisc-crys.arpa; Mon, 22 Jul 85 16:23:50 cdt
Received: from ibm-sj by csnet-relay.csnet id am05001; 22 Jul 85 17:15 EDT
Date: Mon, 22 Jul 85 13:49:45 PDT
From: Joe Halpern <halpern%ibm-sj.csnet@csnet-relay.arpa>
To: theory@WISC-CRYS.ARPA
Subject: Conference on Reasoning About Knowledge
Sender: udi@wisc-rsch.arpa
Resent-From: udi@wisc-rsch.arpa (Udi Manber)
Resent-To: theory:;@wisc-rsch.arpa
CONFERENCE ON THEORETICAL ASPECTS OF REASONING ABOUT KNOWLEDGE:
SECOND (AND LAST!) CALL FOR PAPERS
A conference on Theoretical Aspects of Reasoning About Knowledge
will be held Mar. 19-22, 1986, at the Asilomar Conference Center in
Monterey. While traditionally research in this area was mainly done by
philosophers, recently it has been shown to be of great
relevance to computer science, especially in such areas as artificial
intelligence, distributed systems, database systems,
and cryptography. There has also been interest in the area among
linguists and economists. The aim
of this conference is to bring together
researchers from these various disciplines
with the intent of furthering our theoretical understanding of
reasoning about knowledge.
Topics of interest include, but are not limited to:
* Semantic models for knowledge and belief
* Resource-bounded knowledge (appropriate for modelling reasoners with
limited reasoning power and reasoning about cryptographic protocols)
* Using knowledge to specify and reason about distributed systems
* Semantic models of knowledge acquisition and learning
* Nonmonotonic reasoning
Please send 8 copies of a detailed abstract
not exceeding 10 double-spaced typewritten pages in length
(not a full paper),
by September 15, 1985, to the program chair:
Dr. J. Halpern
IBM Research, K51/281
5600 Cottle Rd.
San Jose, CA 95193
The abstract should include a clear description of the problem being
addressed, comparisons with extant work, and a section on major
original contributions of this work. The abstract must provide
sufficient detail for the program committee to make a decision.
Papers will be chosen on the basis of scientific merit, originality,
and appropriateness for this conference.
Authors will be notified of acceptance by Nov. 1, 1985. Accepted
papers typed on special pages will be due at the above address
by Dec. 15, 1985.
The program committee members are:
M. Fischer, Yale
J. Halpern, IBM San Jose
H. Levesque, University of Toronto
R. Moore, SRI
R. Parikh, CUNY/Brooklyn College
R. Stalnaker, Cornell
R. Thomason, Pittsburg
M. Vardi, Stanford/CSLI
We hope to allow enough time between the talks during the conference
for private discussions and small group meetings. In order to
ensure that the conference remains relatively small, attendance will
be limited to invited participants and
authors of accepted papers. The conference is being held in cooperation
with ACM. Support has been received from IBM and AAAI;
an application for further support is pending at ONR.
∂24-Jul-85 1409 @SU-CSLI.ARPA:LANSKY@SRI-AI.ARPA Next Monday's PLANLUNCH
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 24 Jul 85 14:09:26 PDT
Received: from SRI-AI.ARPA by SU-CSLI.ARPA with TCP; Wed 24 Jul 85 14:00:06-PDT
Date: Wed 24 Jul 85 13:59:10-PDT
From: LANSKY@SRI-AI.ARPA
Subject: Next Monday's PLANLUNCH
To: aic-associates@SRI-AI.ARPA, folks@SU-CSLI.ARPA, mugs@SUMEX-AIM.ARPA,
val@SU-AI.ARPA, rsimmons@SRI-KL.ARPA, carnese@SRI-KL.ARPA,
alpert@SU-SUSHI.ARPA, frayman@SUMEX-AIM.ARPA, dmrussell@XEROX.ARPA,
vanLehn@XEROX.ARPA, araya@XEROX.ARPA, frayman@XEROX.ARPA,
suchman@XEROX.ARPA, weld@XEROX.ARPA, mittal@XEROX.ARPA, dekleer@XEROX.ARPA
Title: Time and Causation from the Standpoint of AI.
Nontitle: The Frame Problem is not.
Yoav Shoham
Yale University, SRI-AI
11:00 AM, Monday, July 29
SRI International, Building E, Room EJ232
Most tasks undertaken by AI researchers involve reasoning about
time in one way or another. In particular, the somewhat ill-defined
areas of planning and naive physics reasoning rely crucially on the
passage of time and the taking place of change. I am aiming at
a general and yet rigorous theory of time and change.
I am primarily interested in a useful notational device. Psychological
plausibility is only an added benefit, and philosophical truth
is something over which I lose little sleep.
The talk is structured as follows:
1. A first-order theory of time, which could be viewed as a generalization
of James Allen's theory.
2. A modal version of the same theory. Several interval-based modal logics
will be presented, along with the few of their theoretical properties
which I am beginning to understand.
3. The theory of causal counterfactuals, a particular theory of change
that relies on part 1. I will demonstrate how this theory appears
to avoid three major problems:
a. The cross-world identification problem for time tokens.
b. The frame problem.
c. A nameless problem encountered in the philosophy literature.
Also, in the spirit of recent "rigorous reconstructions" I will
reformulate Ken Forbus' Qualitative Process theory in terms of
causal counterfactuals.
In the unlikely event of our having time left over I'll discuss
Richard Waldinger's toy car.
-------
-------
∂24-Jul-85 1424 @SU-SUSHI.ARPA,@SU-SCORE.ARPA:udi@wisc-rsch.arpa Conference on Reasoning About Knowledge
Received: from SU-SUSHI.ARPA by SU-AI.ARPA with TCP; 24 Jul 85 14:24:16 PDT
Received: from SU-SCORE.ARPA by SU-SUSHI.ARPA with TCP; Wed 24 Jul 85 13:35:27-PDT
Received: from wisc-rsch.arpa by SU-SCORE.ARPA with TCP; Wed 24 Jul 85 13:35:13-PDT
Received: by wisc-rsch.arpa; Wed, 24 Jul 85 14:31:55 cdt
Received: from wisc-crys.arpa by wisc-rsch.arpa; Mon, 22 Jul 85 16:28:57 cdt
Message-Id: <8507222123.AA28985@wisc-crys.arpa>
Received: from csnet-relay.arpa by wisc-crys.arpa; Mon, 22 Jul 85 16:23:50 cdt
Received: from ibm-sj by csnet-relay.csnet id am05001; 22 Jul 85 17:15 EDT
Date: Mon, 22 Jul 85 13:49:45 PDT
From: Joe Halpern <halpern%ibm-sj.csnet@csnet-relay.arpa>
To: theory@WISC-CRYS.ARPA
Subject: Conference on Reasoning About Knowledge
Sender: udi@wisc-rsch.arpa
Resent-From: udi@wisc-rsch.arpa (Udi Manber)
Resent-To: theory:;@wisc-rsch.arpa
CONFERENCE ON THEORETICAL ASPECTS OF REASONING ABOUT KNOWLEDGE:
SECOND (AND LAST!) CALL FOR PAPERS
A conference on Theoretical Aspects of Reasoning About Knowledge
will be held Mar. 19-22, 1986, at the Asilomar Conference Center in
Monterey. While traditionally research in this area was mainly done by
philosophers, recently it has been shown to be of great
relevance to computer science, especially in such areas as artificial
intelligence, distributed systems, database systems,
and cryptography. There has also been interest in the area among
linguists and economists. The aim
of this conference is to bring together
researchers from these various disciplines
with the intent of furthering our theoretical understanding of
reasoning about knowledge.
Topics of interest include, but are not limited to:
* Semantic models for knowledge and belief
* Resource-bounded knowledge (appropriate for modelling reasoners with
limited reasoning power and reasoning about cryptographic protocols)
* Using knowledge to specify and reason about distributed systems
* Semantic models of knowledge acquisition and learning
* Nonmonotonic reasoning
Please send 8 copies of a detailed abstract
not exceeding 10 double-spaced typewritten pages in length
(not a full paper),
by September 15, 1985, to the program chair:
Dr. J. Halpern
IBM Research, K51/281
5600 Cottle Rd.
San Jose, CA 95193
The abstract should include a clear description of the problem being
addressed, comparisons with extant work, and a section on major
original contributions of this work. The abstract must provide
sufficient detail for the program committee to make a decision.
Papers will be chosen on the basis of scientific merit, originality,
and appropriateness for this conference.
Authors will be notified of acceptance by Nov. 1, 1985. Accepted
papers typed on special pages will be due at the above address
by Dec. 15, 1985.
The program committee members are:
M. Fischer, Yale
J. Halpern, IBM San Jose
H. Levesque, University of Toronto
R. Moore, SRI
R. Parikh, CUNY/Brooklyn College
R. Stalnaker, Cornell
R. Thomason, Pittsburg
M. Vardi, Stanford/CSLI
We hope to allow enough time between the talks during the conference
for private discussions and small group meetings. In order to
ensure that the conference remains relatively small, attendance will
be limited to invited participants and
authors of accepted papers. The conference is being held in cooperation
with ACM. Support has been received from IBM and AAAI;
an application for further support is pending at ONR.
∂24-Jul-85 1653 ullman@diablo
Received: from SU-AIMVAX.ARPA by SU-AI.ARPA with TCP; 24 Jul 85 16:53:16 PDT
Date: Wed, 24 Jul 85 16:21:30 pdt
From: Jeff Ullman <ullman@diablo>
To: nail@diablo
A COUNTERPROPOSAL ON NEGATION SEMANTICS
Let's start by looking at Allen's rule:
ne(Y,Z) ifnot e(Y,Z,W).
and the interpretation of the goal ne(3,Z) as returning
either the empty relation or the relation with one "tuple"
(3,Z). [Refer to p. 4 of the second part of Allen's writup.]
Suppose e is the relation with two tuples (1,1,1) and (3,3,3),
as in Allen's example.
Now suppose we use ne in the rule
a(Z) :- b(Z) & ne(3,Z)
where b is a database relation with three tuples: (1), (2), and (3).
If we take Allen's interpretation that ne(3,Z) is {(3,Z)}
unless forallZ existsW e(3,Z,W) then we find that ne(3,Z) indeed
does have relation {(3,Z)} associated with it, since, for example,
Z=2 has no W that makes e(3,2,W) true.
If we then evaluate b(Z), we get the relation {1,2,3}.
It is not exactly clear how to take the join of relations whose
tuples have variables instead of concrete values, but presuming
that unification is the right thing to do, then
{1,2,3} JOIN {(3,Z)} = {(3,1), (3,2), (3,3)}, and the relation
associated with head a(Z) is thus the projection onto second
components: {1,2,3}.
On the other hand, we could evaluate a(Z) by first evaluating b(Z),
to the set {1,2,3}, of course, and doing sideways information
passing. In that case, we have to evaluate the relations for
ne(3,1), ne(3,2), and ne(3,3). The first two of these return the
tuples (3,1) and (3,2), because for no W is e(3,1,W) or e(3,2,W) true.
However, the third produces the empty relation, since there does
exist W=3 for which e(3,3,W) is true.
Thus this interpretation gives a(Z) = {1,2}.
A BETTER SEMANTICS
I think that the right thing to do is to regard a variable that
appears only in negated terms as existentially quantified by EACH
negated term in which it appears. There are several reason for this choice:
I. It agrees with what NAIL I does. Well who cares about that? Even
I don't. But:
II. It makes sense. For example, what means:
orphan(X) :- person(X) & \+ mother(X,Y) & \+ father(X,Y)
Is an orphan a person with no mother and no father, or a person
whose mother and father are different persons?
III. It agrees with what Prolog does. As Allen pointed out this morning,
\+ p(X,Y) can NEVER provide a binding for X or Y, if those were free
variables. Thus, in the orphan example above, while it is true
that the two occurrences of Y are allocated a single storage location,
it is never possible that the stored value from a call to mother
is available to a call to father or vice-versa.
IV. It has a simple implementation. First, let me beseech the group
to outlaw implementations in which infinite relations, even
in covert forms like {(3,Z)}, are used.
Then, we may assume that all negated goals in a clause body are
evaluated AFTER all nonegated goals, and therefore, every variable
in such a goal is either bound or local (implicitly existentially
quantified). If we have an instance of a goal \+ p(a1,...,an,W1,...,Wm)
we solve the unnegated goal, getting a relation over W1,...,Wm.
If this relation is empty, then the negated goal has the relation
{epsilon}, i.e., the relation containing the empty tuple.
This relation acts as the identity when the relations for the
various goals in the body arre joined.
If the relation for p is nonempty, then the relation for \+ p
is empty, which acts as the zero for joins, of course.
DO WE NEED NEW ADORNMENTS?
I'm getting convinced that perhaps we need an "e" adornment
(exists) in addition to the b and f adornments. The reason is
that in point IV above, we really do not want to capture
p↑{b...bf...f}, i.e, the full relations over the variables;
we just want to know if the projection of that relation onto
the constant components is nonempty. We could express this
desire by the goal node p↑{b...be...e}.
Research question: Are there some useful capture rules that
work for e-adornments but not for f?
∂24-Jul-85 1714 avg@diablo Re: Jeff's counter-proposal
Received: from SU-AIMVAX.ARPA by SU-AI.ARPA with TCP; 24 Jul 85 17:13:57 PDT
Date: Wed, 24 Jul 85 17:07:08 pdt
From: Allen VanGelder <avg@diablo>
Subject: Re: Jeff's counter-proposal
To: nail@diablo, ullman@diablo
I have not looked at all of Jeff's message yet, but in his first
example, he wonders how to join {1, 2, 3} with {(3, Z)} and says
"probably by unification."
No. Unification is definitely wrong and is not what I am proposing.
His example perfectly illustrates WHY it is wrong.
∂24-Jul-85 1747 EMMA@SU-CSLI.ARPA Newsletter July 25, No. 38
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 24 Jul 85 17:47:26 PDT
Date: Wed 24 Jul 85 17:04:56-PDT
From: Emma Pease <Emma@SU-CSLI.ARPA>
Subject: Newsletter July 25, No. 38
To: friends@SU-CSLI.ARPA
Tel: 497-3479
C S L I N E W S L E T T E R
←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←
July 25, 1985 Stanford Vol. 2, No. 38
←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←
A weekly publication of The Center for the Study of Language and
Information, Ventura Hall, Stanford University, Stanford, CA 94305
←←←←←←←←←←←←
CSLI ACTIVITIES FOR *THIS* THURSDAY, July 25, 1985
12 noon CSLI Lunch
Ventura Hall ``Algebraic Semantics and the Logic of Programs''
Conference Room Irene Guessarian, University of Paris VII
(Abstract on page 1)
2:15 p.m. CSLI Talk
Ventura Hall ``Abstract Semantic Algebras: Theory and Practice''
Conference Room Peter Mosses, Computer Science Dept., Aarhus University
3:30 p.m. Tea
Ventura Hall
←←←←←←←←←←←←
CSLI ACTIVITIES FOR *NEXT* THURSDAY, August 1, 1985
12 noon TINLunch
Ventura Hall No TINLunch this week
Conference Room
2:15 p.m. CSLI Talk
Ventura Hall ``Realism and Antirealism in Cognitive Artificial
Conference Room Intelligence''
David H. Helman, Department of Philosophy, Case
Western Reserve University
(Abstract on page 2)
3:30 p.m. Tea
Ventura Hall
←←←←←←←←←←←←
ABSTRACT OF THIS WEEK'S CSLI LUNCH
``Algebraic Semantics and the Logic of Programs''
Thursday, July 25, 12 noon, Ventura Conference Room
We will present the basic ideas of algebraic semantics. The
overall goal is to describe and prove properties of programs in the
nicest possible way. The algebraic way consists in first
characterizing a program by an infinite tree, which is an object in
some free algebra. We then give the semantics of programs using
algebraic tools which are well-known. After that, we can introduce
progressive constraints on the free algebras, in a modular way, in
order to model the properties of programs. In so doing, we will
relate algebraic semantics to logics of programs, mainly equational
logics. Finally, we will show an application to an equational and
complete proof system for the IF-THEN-ELSE. --Irene Guessarian
!
Page 2 CSLI Newsletter July 25, 1985
←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←
ABSTRACT OF NEXT WEEK'S CSLI TALK
``Realism and Antirealism in Cognitive Artificial Intelligence''
Ventura Conference Room, Thursday, August 1, 2:15 pm
In the philosophy of mind, one controversy between realists and
antirealists concerns the semantics of sentences embedded in attitude
reports. Antirealists believe that the interpretation or reference of
a sentence embedded in an attitude report is a psychological state of
the agent who is the subject of the attitude report. Realists believe
that the interpretation or reference of a sentence is a state of the
world and not a state of mind, whether or not the sentence is embedded
in an attitude report.
In this paper, I show how these two semantic analyses may be
associated with different theories of mental representation in
cognitive artificial intelligence. Realists in cognitive artificial
intelligence describe the mind by supposing that agents partially
represent objects' law-like interactions. Antirealism does not,
perhaps, constitute a single well-defined research strategy in
cognitive artificial intelligences. We may, however, certainly count
as antirealists those researchers in cognitive artificial intelligence
who attempt to simulate mental processes by means of procedures which
mirror tenets of associationist psychology. I argue that acurate
computational models of mind must contain elements from both realist
and antirealist research programs. --David Helman
←←←←←←←←←←←←
PIXELS AND PREDICATES
``Pixels `n' Predicates for Menus `n' Mice''
Henry Lieberman, MIT Artificial Intelligence Laboratory
AI Center, Room EK240, SRI International
Wednesday, July 31, 3:00 pm
User interfaces using menu commands, pointing devices, and direct
manipulation of graphical objects constitute a new kind of ``visual
language'' for communicating with computers. What are the basic
building blocks, or ``visual morphemes'' of this new language?
The building blocks supplied by most languages for programming
graphical user interfaces deal mainly with the lowest level of this
visual language: drawing graphical objects such as lines and text,
reading the coordinates of pointing devices. These predicates are too
close to the pixel level to permit rapid construction of modular
interfaces. The next step up is to explicitly represent graphical
representations of manipulable objects, actions the user can perform,
and styles of interaction.
An analogy with interpreters for conventional programming languages
shows how an interpreter can be written for a language of actions
performed by selecting menu commands, operating on arguments obtained
by pointing at graphical objects on the screen.
[This week the meeting is at SRI's AI Center, Room EK240, 333
Ravenswood Ave, Menlo Park which is 1 mile north on El Camino from
Stanford University, turn right on Ravenswood, past first light, go in
driveway on right, ask receptionist for more directions.]
!
Page 3 CSLI Newsletter July 25, 1985
←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←
INTERACTIONS OF MORPHOLOGY, SYNTAX, AND DISCOURSE
``Morphological Structure of Kunparlang Verbs''
Summary of the meeting on July 18
At the meeting, Carolyn Coleman presented the results of some of
her research on the morphological structure of Kunparlang verbs.
Kunparlang verbs are extremely complex morphologically. They
cross- reference Subject and Object functions, incorporate nominal
roots, use `applicative' derivational morphology, carry modal,
directional and aspectual affixes, and inflect for Tense and Mood.
There are two levels of hierarchical morphological structure,
(i) The stem, which carries all morphology having compositional
semantics.
(ii) The lexical base, which caries all semantically idiosyncratic
morphology.
Kunparlang verbs undergo two types of reflexive operation which
have a partially complementary distribution and which have different
semantic effects on the verbs to which they apply. With the first
reflexive operation the reflexive subject is always an Agent; with the
second the reflexive subject is always a Theme. The second reflexive
operation also has incohative and mediopassive readings as well as the
reflexive reading. Both reflexivizing operations are derivations that
apply at the level of the lexical base; given that they have the same
morphological status, there is a problem of how to semantically
characterise them in a manner that will clearly show the semantic
similarities and differences between them. --Carolyn Coleman
←←←←←←←←←←←←
CORRECTION
Rich Cower, the new Computing Director for CSLI, will start August
12 not August 1 as announced last week.
!
Page 4 CSLI Newsletter July 25, 1985
←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←
PROGRAM FOR THE WORKSHOP ON FINITE STATE MORPHOLOGY
Ventura Hall, Trailor Class Room
July 29-30, 1985
For more information please contact Lauri Karttunen (Karttunen@sri-ai)
at (415)859-5082.
Monday, July 29
9:30 Lauri Karttunen ``Issues in Finite State Morphology''
10:30 Ronald Kaplan ``Phonological Rules and Finite-state
Transducers''
11:30 Kimmo Koskenniemi ``Compilation of Automata from
Two-level Rules''
2:00 John Bear ``Implementing Two-level Phonological Rules
Directly''
3:00 Edward Barton ``Complexity of Two-level Morphology''
4:00 Demonstrations by Bear, Karttunen, and Koskenniemi
Tuesday, July 30
9:30 Kenneth Church ``Morphological Stress Decomposition and
Stress Assignment for Speech Synthesis''
10:30 William Poser ``Locality Constraints on Phonological Rules''
11:30 Michael Bateman ``ATEF: A Finite State Model for Morphological
Analysis''
2:00 Mark Johnson ``Acquisition of a Restricted Set of
Phological Rules''
3:00 Martin Kay ``Two-level Morphology with Tiers''
(Discussant: John McCarthy)
4:00 Demonstrations by A. Golding, M. Johnson
-------
∂24-Jul-85 1752 avg@diablo More on Jeff's counter proposal
Received: from SU-AIMVAX.ARPA by SU-AI.ARPA with TCP; 24 Jul 85 17:52:36 PDT
Date: Wed, 24 Jul 85 17:43:51 pdt
From: Allen VanGelder <avg@diablo>
Subject: More on Jeff's counter proposal
To: nail@diablo
Jeff has proposed what amounts to the following rule: If a variable
under a negation does not also appear outside the scope of the "not"
AND in a "positive" literal there, THEN it is locally existentially
quantified.
We need to be careful about several layers of scopes due to nested
"not"s and "bagof"s. We will get into the issue of "scope of variables
within a rule," an issue that supposedly does not exist in
logic programming. Maybe we can't avoid it if we are going to delve
into "not" and "bagof."
While I have no basic objection to this approach, since I know of no
use for the "feature," it is a restriction on the user, not an
added capability. E.g., in Jeff's "orphan" example, he uses Y
for both mother and father -- surely we don't want to encourage this.
Why not use M and F? Alternatively use the underscore, which means
"anonymous variable," in both places.
Jeff Naughton has made a very good suggestion about this:
WHATEVER we do in this bizarre case, we should issue a read-time
warning to the user, telling how we are treating this rule,
since it is likely not what he or she intended.
Incidentally, I DON'T think our policy should be to copy Prolog when
it does inconsistent things, depending on subgoal order and whether
the top goal is bound or free. In this case, we should decide which of the
inconsistent results are correct, and make it always come out that way.
E.g., X=3 should be a solution to ?- p(X) if and only if ?- p(3)
succeeds.
∂24-Jul-85 2220 avg@diablo Even more on Jeff's counter-proposal
Received: from SU-AIMVAX.ARPA by SU-AI.ARPA with TCP; 24 Jul 85 22:20:44 PDT
Date: Wed, 24 Jul 85 22:08:50 pdt
From: Allen VanGelder <avg@diablo>
Subject: Even more on Jeff's counter-proposal
To: nail@diablo
First, those who want to follow this exciting discussion closely can
pick up a copy of the notes to which Jeff referred from either my office,
or from Rosemary Napier, his sec'y.
Jeff's message addresses a few independent issues.
1. What about a negated expression containing some unbound non-local
variables?
2. Suppose a variable is non-local only because it occurs in another
negated expression?
3. Infinite relations are worse than in-laws, and should be outlawed.
(1) His first example pinpointed a fuzzy point in my write-up. What do we
get when we evaluate ne(3, Z)? Recall that we are working with
ne(Y, Z) ifnot e(Y, Z, W).
a(Z) if b(Z) & ne(3,Z).
e(1, 1, 1).
e(3, 3, 3).
b(1).
b(2).
b(3).
I said that if ne(3, Z) is "called," we get the single tuple
(3, Z), as the answer. However, this tuple, having come from an
ifnot, is very WEAK for joining, not very STRONG. What I mean is
that the "Z" in (3, Z) can only join with an unbound variable from
a positive subgoal. That is, we get a non-empty answer for a(Z)
only if there is a rule
b(U).
Whereas b(U) means "b is true for all U, so joins with anything,"
ne(3, Z) means "ne is true for SOME Z, but we don't know any more,
so it does not join with anything except a positive free variable."
Perhaps the right idea is that the call of ne(3, Z) should return the
single tuple (3, "#$%&'), where "#$%&' is a constant that appears no
where else.
It is clear that from this description that it is WRONG to tackle a
negated subgoal with an unbound variable if that variable is unbound because
the subgoals in which it appears have not yet been solved.
The only time that "calling" such a negated subgoal with a unbound variable
is at all reasonable is when that variable is unbound because
the other subgoals SUCCEEDED without binding it.
Our initial implementation may well "punt" on this, and require all
non-local variables to be bound before going after the negated subgoals.
Then the example in my write-up, with the rule
b(1, U).
instead of b(1, 3) will not work. Too bad.
(2) Repeated variables in different negated subgoals is still lacking
a good example, on which to base our intuition. Jeff's orphan example
points up a bug in Prolog more than anything, because it uses the
same variable to refer to two different objects, a mother and a father,
and still works. Maybe we should begin by saying this is simply an
error.
A similar situation exists in bagof. The Prolog manual says that
variables appearing in the first argument of bagof "should not" appear
outside of the bagof. But if they do, and are instantiated before
bagof is "called," their instantiations are used. If they are not
instantiated at the time of the call, they are used unbound in the
evaluation of bagof, giving different results, even if they are later
bound to the same value. I think this "should not" in Prolog should
be elevated to "may not" in Nail.
In other words, to handle the extensions to Horn clauses, we need to
have exceptions to the rule that the scope of a variable is always
the entire rule. Sometimes it is less, when "not" and "bagof" are involved.
To prevent obscure programs and discrepancies between
what the user meant and what Nail read, we should probably make it an
error for the same variable name to be used in two different contexts
or scopes within the same rule (unless one scope contains the other,
of course).
(3) Infinite relations. These are the entry-point to quantified
rules. If we can implement quantification with just a few mechanisms,
then this is desirable. In the future, we can build the "syntactic
sugar" if we think the mechanisms are too tough to use correctly in
their raw form.
On the other hand, implementing Nail without quantified rules will still
be a big step forward. I would like to "leave the door open" with
the understanding that quantified rules are probably low priority.
That is, I would like the design to allow for this step in the future.
∂25-Jul-85 1205 NAUGHTON@SU-SCORE.ARPA more (or less) on negation
Received: from SU-AIMVAX.ARPA by SU-AI.ARPA with TCP; 25 Jul 85 12:04:44 PDT
Received: from SU-SCORE.ARPA by diablo with TCP; Thu, 25 Jul 85 11:56:23 pdt
Date: Thu 25 Jul 85 11:38:19-PDT
From: Jeff Naughton <NAUGHTON@SU-SCORE.ARPA>
Subject: more (or less) on negation
To: nail@SU-AIMVAX.ARPA
Message-Id: <12129868608.27.NAUGHTON@SU-SCORE.ARPA>
I'd like to cast a vote for what Jeff U. is proposing. As I understand it,
any variable appearing under a negation but in no non-negated literal
will be treated as existentially quantified within that negation. If
someone writes such a rule, NAIL should produce a warning that the
offending variable(s) will be treated as existentially quantified
within the negation.
This has the advantage that the implementation and interpretation are
both relatively straightforward. It seems that negation in NAIL
should be regarded as a kind of a filter, something to remove tuples
from a result, rather than something to produce new candidate tuples.
In most cases, if you really want \+ A(X,Y) to generate tuples you
probably should explicitly build a relation NotA, containing the
tuples you want. This negation-as-filter idea is the informal
equivalent of evaluating all positive predicates first to get
bindings, then evaluating negated predicates with any unbound
variables treated as "don't cares."
-------
∂25-Jul-85 1342 SARAIYA@SUMEX-AIM.ARPA 3600 metering
Received: from SUMEX-AIM.ARPA by SU-AI.ARPA with TCP; 25 Jul 85 13:42:23 PDT
Date: Thu 25 Jul 85 13:40:46-PDT
From: Nakul P. Saraiya <SARAIYA@SUMEX-AIM.ARPA>
Subject: 3600 metering
To: KSL-LispM@SUMEX-AIM.ARPA
cc: aap@SUMEX-AIM.ARPA
Message-ID: <12129890900.39.SARAIYA@SUMEX-AIM.ARPA>
The system "DCP-Metering" is an interface to the pc-metering
software on the 36xxs. In addition, it allows one to measure the run-
time of functions and count consing. It resides in 3605:>Metering-Software.
To use : (make-system "DCP-Meter" :noconfirm :silent)
<select>-% gets you a metering window, <function>-square
starts metering after reset, <function>-circle stops metering and
<function>-control-square continues metering. The only documentation
that exists is the source code!
Nakul
-------
∂25-Jul-85 1342 SARAIYA@SUMEX-AIM.ARPA 3600 metering
Received: from SUMEX-AIM.ARPA by SU-AI.ARPA with TCP; 25 Jul 85 13:42:23 PDT
Date: Thu 25 Jul 85 13:40:46-PDT
From: Nakul P. Saraiya <SARAIYA@SUMEX-AIM.ARPA>
Subject: 3600 metering
To: KSL-LispM@SUMEX-AIM.ARPA
cc: aap@SUMEX-AIM.ARPA
Message-ID: <12129890900.39.SARAIYA@SUMEX-AIM.ARPA>
The system "DCP-Metering" is an interface to the pc-metering
software on the 36xxs. In addition, it allows one to measure the run-
time of functions and count consing. It resides in 3605:>Metering-Software.
To use : (make-system "DCP-Meter" :noconfirm :silent)
<select>-% gets you a metering window, <function>-square
starts metering after reset, <function>-circle stops metering and
<function>-control-square continues metering. The only documentation
that exists is the source code!
Nakul
-------
∂25-Jul-85 1527 CER@SU-SCORE.ARPA Follow-up on counterproposal
Received: from SU-AIMVAX.ARPA by SU-AI.ARPA with TCP; 25 Jul 85 15:27:36 PDT
Received: from SU-SCORE.ARPA by diablo with TCP; Thu, 25 Jul 85 15:20:11 pdt
Date: Thu 25 Jul 85 15:18:19-PDT
From: Stefano Ceri <CER@SU-SCORE.ARPA>
Subject: Follow-up on counterproposal
To: nail@SU-AIMVAX.ARPA
Message-Id: <12129908656.12.CER@SU-SCORE.ARPA>
We also cast our vote for Jeff proposal, however with some comments.
1. The assumption of evaluating all negated goals in a clause after evaluating
all positive goals seems too strong. By appropriate capture rules,
a negated goal might be evaluated as soon as all of its variables which
appear in some positive goals are bound.
2. The possibility of getting useful information from negative goals in rules
should not be disregarded. For instance:
equal(X,Y):- X=Y.
p(X,Z):- \+equal(X,Y), rel(X,Z).
This example shows that a negative clause which is independent on the
database can be efficient.
Ceri & Gottlob
-------
∂25-Jul-85 1736 @SU-SUSHI.ARPA:HERSHBERGER@SU-SCORE.ARPA Rides for BATS at San Jose
Received: from SU-SUSHI.ARPA by SU-AI.ARPA with TCP; 25 Jul 85 17:35:51 PDT
Received: from SU-SCORE.ARPA by SU-SUSHI.ARPA with TCP; Thu 25 Jul 85 17:32:38-PDT
Date: Thu 25 Jul 85 17:32:32-PDT
From: John E. Hershberger <HERSHBERGER@SU-SCORE.ARPA>
Subject: Rides for BATS at San Jose
To: aflb.local@SU-SCORE.ARPA
Message-ID: <12129933091.17.HERSHBERGER@SU-SCORE.ARPA>
Since nobody else seems to be doing it, I hereby volunteer to coordinate
rides to BATS next Thursday. If those who are driving and those who need
rides will respond to me, I will at least attempt to verify that demand does
not exceed car capacity. I will also let the riders know who the drivers are
to allow plans to be made before Thursday morning.
I guess the most useful information for space calculations would be (1) the
number of unclaimed seats in drivers' cars, and (2) responses from riders who
haven't already found a driver to go with.
Thanks,
John
-------
∂25-Jul-85 2231 avg@diablo YAPON (Yet another proposal on Negation)
Received: from SU-AIMVAX.ARPA by SU-AI.ARPA with TCP; 25 Jul 85 22:31:00 PDT
Date: Thu, 25 Jul 85 22:23:36 pdt
From: Allen VanGelder <avg@diablo>
Subject: YAPON (Yet another proposal on Negation)
To: nail@diablo
The following example involving variables that appear in 2 negated
subgoals, but no positive subgoals illustrates that considering such
variables as local may not be satisfactory.
a(Y) :- b(Y, Z) & \+ e(Y, Z, W) & \+ h(Z, W).
h(Z, W) :- f(Z) & \+ g(Z, W).
The EDB is b, e, f, and g, as given in the table below.
b(Y, Z). e(Y, Z, W). f(Z). g(Z, W).
--------------------------------------------------
b(5, 1). e(5, 1, 1). f(1). g(1, 2).
e(5, 1, 2). f(2). g(2, 1).
b(6, 1). e(6, 1, 1). g(2, 2).
b(7, 3). e(7, 3, 3). g(3, 2).
b(8, 2).
To the query ?- a(Y) it is clear that the possible answers are among
the set {5, 6, 7, 8}. I claim that the correct evaluation should be
{6, 7, 8}, but not 5. My reasoning is that Y=6, Z=1, W=2 satisfies
b and falsifies both e and h, thus deriving a(6).
Also, Y=7, Z=3, W=2 (or anything but 3) does the same for a(7).
Finally, Y=8, Z=2, W=1 (or 2) does the job for a(8).
However, when Y=5 and Z=1, NO value of W simultaneously falsifies e and h,
even if W is given a "ridiculous" value like mnbvcx.
Notice that Prolog would find only a(8) by treating the two W's as
unrelated variables.
A COMPREHENSIVE SEMANTICS
Interpret a Nail rule R as a first order formula F as follows:
First, express the rule in an AND-OR-NOT-IF tree without quantifiers.
Then insert quantifiers appropriately, as described next.
The variables appearing in the head are free; they are the columns in
the relation to be found. All other variables are existentially
quantified at the lowest scope that includes all mentions of the
variable. For our purposes, we can regard AND and OR as polyadic.
With the quantifiers in place, we can make transformations according
to deMorgan's rules, distributive laws, etc. to arrive at a normal
form.
In common cases we get the result that most of us agree we want,
and this also agrees with Prolog on its better days.
The main bone of contention is treatment of weird things like the above
example, so let's see how that one comes out.
(Repeated for convenience)
a(Y) :- b(Y, Z) & \+ e(Y, Z, W) & \+ h(Z, W).
h(Z, W) :- f(Z) & \+ g(Z, W).
Let's use -] for there-exists and \-/ for for-all, and use curly braces
to exhibit tree structure (instead of parens). In addition, p1, p2, ...
will be generated predicate symbols.
We get for the first rule:
a(Y) IF -]Z {b(Y,Z) AND -]W {NOT e(Y,Z,W) AND NOT h(Z,W)}}.
Use deM's law:
a(Y) IF -]Z {b(Y,Z) AND -]W NOT {e(Y,Z,W) OR h(Z,W)}}.
We can push the W-quantifier thru the NOT but not thru the OR:
a(Y) IF -]Z {b(Y,Z) AND NOT \-/W {e(Y,Z,W) OR h(Z,W)}}.
This does not look any better for evaluation, at least to me.
To try to make progress, let us "substitute for" h(Z,W), since it has
only one rule:
a(Y) IF -]Z {b(Y,Z) AND NOT \-/W {e(Y,Z,W) OR {f(Z) AND NOT g(Z,W)}}}.
Now after distributing the OR, we CAN push the W-quantifier thru the AND,
calling the separate occurrences W1 and W2.
We can also push it thru one OR because only one child contains W:
a(Y) IF -]Z {b(Y,Z) AND
NOT {{\-/W1 {e(Y,Z,W1)} OR f(Z)} AND
\-/W2 {e(Y,Z,W2) OR NOT g(Z,W2)}
}
}.
Courage, the end is near. Push NOT thru AND:
a(Y) IF -]Z {b(Y,Z) AND
{ {NOT \-/W1 {e(Y,Z,W1)} AND NOT f(Z)} OR
-]W2 {NOT e(Y,Z,W2) AND g(Z,W2)}
}
}.
Now, introduce p1(Y,Z,W) and supplemental rules:
a(Y) IF -]Z {b(Y,Z) AND p1(Y,Z)}.
p1(Y,Z) IF {NOT f(Z) AND NOT \-/W1 e(Y,Z,W1)}.
p1(Y,Z) IF -]W2 {g(Z,W2) AND NOT e(Y,Z,W2)}.
Looking closely, we see that the only real problem is the \-/W1.
Assume we know e is a finite EDB relation. Then we can always
choose a "ridiculous" value for W1 in the first p1 rule, so we might as
well rewrite it as
p1(Y,Z) IF NOT f(Z).
Now with this trickery done, we can write the "equivalent" Nail rules:
a(Y) :- b(Y, Z) & p1(Y, Z).
p1(Y, Z) :- \+ f(Z).
p1(Y, Z) :- g(Z, W2), \+ e(Y, Z, W2).
These rules look perfectly tractable to me now.
∂25-Jul-85 2253 PARSYM-Request@SUMEX-AIM.ARPA PARSYM Digest V1 #6
Received: from SUMEX-AIM.ARPA by SU-AI.ARPA with TCP; 25 Jul 85 22:53:00 PDT
Date: 25 Jul 85 2231-PDT
From: Moderator Byron Davies <PARSYM-REQUEST@SUMEX-AIM.ARPA>
Reply-to: PARSYM@SUMEX-AIM.ARPA
Subject: PARSYM Digest V1 #6
To: PARSYM@SUMEX-AIM.ARPA
PARSYM Digest Friday, 26 Jul 1985 Volume 1 : Issue 6
Today's Topics:
Restructuring LISP
What is symbolic computation?
FP Languages and Architectures: Conference Registration
Software position at Columbia University
----------------------------------------------------------------------
Date: Wed, 24 Jul 1985 09:13 PDT
Subject: Restructuring LISP
From: harrison%uicsrd@Uiuc.ARPA (Luddy Harrison)
>Subject: restructuring LISP
>
>Are we really buying a whole lot of efficiency from our parallel
>machines by putting "warts" on Fortran and LISP that trace their
>lineage to the 50's?
I'm afraid I don't understand; what "warts" are you speaking of? Language
extensions? These have nothing to do with restructuring code written in
the language; by restructuring we mean a fully (or largely) automated
process by which a code is "massaged" such that parallelism within it is
detected and made explicit. Indeed, the idea is to avoid the use of
language extensions, and the process is much easier when the language is
not overly large.
>For major software investments probably yes, however, wouldn't it be
>more reasonable to spend the time on a new language for present and
>future developments.
Bring on the new languages! As it turns out, many of the techniques
of restructuring are independent of the language to which they are
applied; the preference for FORTRAN and LISP has been strictly
practical: there happens to be a lot of code written in them, that's
all.
------------------------------
Date: Thu, 25 Jul 85 22:43:43 edt
From: Tom Blenko <blenko@rochester.arpa>
Subject: What is symbolic computation?
Vaughn's question is an interesting one. My proposal is that
numerical computation is performed over a flat domain, while symbolic
computing permits computation over terms which are partially bound or
instantiated. Binding is used in a broad sense here, and subsumes
type declaration and generic instantiation, as well as the
conventional notion of variable binding.
According to this scheme, FORTRAN is almost purely numerical because it
allows variables to be bound only to constants (although a form of
compile-time co-referential binding is possible using EQUIVALENCE).
FORTRAN is not purely numerical because variables are typed. Type
declaration is a form of binding under the notion of binding described
above, although an especially weak one because all type declarations
must be made at compile time. The class of nearly-numerical languages
(those with flat domains plus some support for typed variables) can be
expanded somewhat by including languages with slightly more powerful
type mechanisms, i.e., those which support discriminated unions or
procedure name overloading. For the purposes of discussion, I'd be
willing to refer to all of these as languages which support numerical
computation exclusively. This seems like a reasonable approximation
because their competence as symbolic languages is both weak and well
known.
Languages which permit various forms of abstract data type or generic
procedure definition would be classified as partially-symbolic because
they each support some form of run-time partial binding of variables.
Representative examples in this class are CLU, ADA, and SMALLTALK-80.
This is the class currently of greatest interest to the imperative
language people, and certainly there needs to be more work on what
abstraction and type mechanisms are useful and can be implemented
efficiently.
My two choices to represent (nearly) fully-symbolic languages are
PROLOG and LISP. In PROLOG, the binding mechanism, unification, can
also be viewed as a type restriction mechanism, so that a variable
becomes bound to a grounded term by successive application of type
restrictions to (variable) subterms of intermediate bindings as the
computation proceeds (reference for related work is Hassan Ait-Kaci's
thesis, A New Model of Computation based on a Calculus of Type
Subsumption). This identification of variable typing with more
traditional notions of variable binding is precisely what I propose
permits one to view symbolic computation in a coherent way.
LISP is well-known, and it would be quite a task to persuade some that
it is anything except the ultimate symbolic language. Clearly its
binding mechanism allows the kind of partial binding which occurs
naturally in PROLOG (although, of course, this could be said to be
only one consequence of its excessive permissiveness). Let me mention
two ways in which it differs from PROLOG, however, and might be viewed
as a more powerful symbolic language.
First, it allows variables to be bound as pointers to other variables.
This is undeniably a powerful mechanism, although it makes for more
complicated language semantics. It is also not a particularly good
substitute for what is (arguably) the corresponding mechanism in
PROLOG, specifically the co-referential binding resulting from the
unification of variables. I say the two mechanisms correspond because
the only way to do equivalencing in LISP is for variables A and B to
both point to C, and for C to store the equivalenced value of A and B,
which may be accessed by a dereference followed by an evaluation --
which leads to the second point.
Many LISPs implement what might be termed a first-class recursive call
to the interpreter. INTERLISP's evala, for example, allows any
procedure to call the interpreter recursively on a LISP data structure
with the binding environment of the computation completely specified
in an argument to evala. Intuitively, this is a powerful mechanism
for symbolic computation, and is moreover necessary for the rather
awkward implementation of LISP equivalencing described above, since
dereferencing is indistinguishable from evaluation (although Brian
Smith has succeeded in separating the two in his definition of
3-LISP). The (second-class) recursive call implemented in most
PROLOGs is more restricted because the environment of the calling
procedure is unconditionally inherited by the called procedure
(although proposals for more general approaches have been made). Many
partially- and non-symbolic languages do not provide any recursive
call.
These are admittedly incomplete thoughts, and I'd be interested in any
responses. I have not specifically addressed binding of parameters
across procedure calls -- call-by-value and call-by-reference can be
understood rather easily once the notion of variable creation is
included, I think. I haven't though about the symbolic power of
exotica such as thunks, and I suspect that macro expansion doesn't add
much in the way of symbolic power.
I'd be particularly interested in a coherent exposition of the
relationships between what I've been proposing as the primary
characteristic of symbolic computation (partial binding) and
mechanisms such as pointer creation and dereferencing, and lazy or
eager evaluation. For example, one might interpret PROLOG unification
of variables as a lazy assignment of the source variable to the target
variable, with evaluation of the source variable binding delayed
indefinitely (this is correct because PROLOG variables are
write-once). The obvious way to do dereferencing in PROLOG is through
a "procedure call" or recursive call to the interpreter, except that
the PROLOG interpreter treats bound and unbound variables differently,
so that unbound variables "evaluate" to themselves both during
unification and when used as parameters to recursive function calls
(under what interpretation is this eager evaluation?).
Another question is why one is normally tempted to categorize a
language like C as non-symbolic, since it allows liberal pointer
creation/dereferencing and thereby allows the same binding of
variables to variables as LISP to be performed in a fairly direct
fashion. (Note, however, that no recursive call to the interpreter is
permitted).
I'd be interested in any thoughts or comments the list might have.
Tom
BLENKO@ROCHESTER
------------------------------
Date: Wed 17 Jul 85 16:46:18-EDT
From: Joseph E. Stoy <JES@MIT-XX.ARPA>
Subject: Conference Registration Details
CONFERENCE ON FUNCTIONAL PROGRAMMING LANGUAGES AND ARCHITECTURES
NANCY, FRANCE
16 to 19 September, 1985
REGISTRATION DETAILS
Registration fee: $150 or FF1300.
Students (send copy of Affiliation card): $55 or FF450.
Late Registration (after August 10): for everyone, $175 or FF1500.
Room Reservation: state the number of people and range of prices
(either FF 120/200 or 210/250: the first is 2 stars, and second 3 stars).
To guarantee the room: FF250.
Please make cheques payable to FPCA85.
Write to
Danielle Marchand
Conference FPLCA-85
CRIN BP 239
54506 Vandoeuvre-les-Nancy Cedex
FRANCE
Please send any question to the person in charge of local arrangements:
Pierre Lescanne (pierre@mit-dash.arpa).
PROVISIONAL PROGRAMME
MONDAY
9:30 - 10:30 OPENING SESSION
D A Turner
Miranda: A non-strict functional language with
polymorphic types
Coffee
11:00 - 12:00 PROGRAM TRANSFORMATION
M Welcome S Skedzielewski
Dataflow Graph Optimization in IF1
C Clack S L Peyton Jones
Generating Parallelism From Strictness Analysis
Lunch
14:30 - 15:30 SEMANTICS
M Mauny P-L Curien G Cousineau
The Categorical Abstract Machine
P Bellot
Operations fonctionelles dans les systemes de
Programmation Sans Variables
Coffee
16:00 - 17:00 SYSTEM ARCHITECTURE
S Abramsky R Sykes
The Design and Implementation of an Applicative
Multiprocessor
M J Shute P E Osmon C L Hankin
COBWEB - A combinator reduction architecture
(Reception)
TUESDAY
9:30 - 10:30 LAZY EVALUATION
P Wadler
How to replace failure by a list of successes
J M R Hughes
Lazy Memo-functions
Coffee
11:00 - 12:00 MACHINE ARCHITECTURES
D A Plaisted
An Architecture for Fast Data Movement in the FFP
Machine
J T O'Donnell
An Architecture that Efficiently Updates
Associative Aggregates in Applicative Programming
Languages
Lunch
(Excursion)
WEDNESDAY
9:30 - 10:30 PROGRAM TRANSFORMATION
T Johnsson
Lambda Lifting: Transforming Programs to
Recursive Equations
S K Debray
Optimizing Almost-Tail-Recursive Prolog Programs
Coffee
11:00 - 12:00 HARDWARE ARCHITECTURE DESIGN
M Sheeran
Designing Regular Array Architectures using
Higher Order Functions
M D Ercegovac M Schlag D Patel
vFP: An Environment for Multi-level Specification,
Analysis, and Synthesis of Hardware Algorithms
Lunch
14:30 - 15:30 STORAGE MANAGEMENT
J M R Hughes
A Distributed Garbage Collection Algorithm
D Brownbridge
Cyclic Reference Counting for Combinator Machines
Coffee
16:00 - 17:00 SYSTEM ARCHITECTURE
D Wise
Design for a Multiprocessing Heap with On-board
Reference Counting
M F Young
A Functional Language and Modular Architecture
for Scientific Computing
(Banquet)
THURSDAY
9:30 - 10:30 TYPES; CONSTRUCTIONS
R Nikhil
Practical Polymorphism
P Dybjer
Program Verification in a Logical Theory of
Constructions
Coffee
11:00 - 12:00 COMPILATION AND CODE GENERATION
V J Bush J R Gurd
Transforming Recursive Programs for Execution on
Parallel Machines
L Augustsson
Compiling Pattern Matching
Lunch
14:30 - 15:30 EXECUTION MODELS
B Goldberg P Hudak
Serial Combinators: Optimal Grains of Parallelism
R B Kieburtz
The G Machine: A fast, graph-reduction evaluator
(END OF CONFERENCE)
=====================================================
------------------------------
Date: Wed, 24 Jul 85 03:08:38 EDT
From: Steven A. Swernofsky <SASW@MIT-MC.ARPA>
Subject: Software position opening at Columbia University
[Message originated from DYER@COLUMBIA-20.ARPA, forwarded from
ARPANET-BBOARDS.]
Columbia University
Department of Computer Science
The DADO project at Columbia University has a full time position available
for a Software Engineer/Project Manager. The DADO project is directed
towards design and development of both hardware and software for the
parallel execution of Artificial Intelligence programs.
Software engineering responsibilities are the major part of the position,
and involve several systems and languages.
Project management responsibilities include a variety of tasks such as
supervision of undergraduate and masters students' independent projects,
communications with hardware and software vendors, some proposal writing, etc.
This portion of the position takes only about 1/4 of the time.
We are seeking someone with both good communications skills and good
software skills.
Please respond to:
Beverly Dyer
457 Computer Science
Columbia University
New York, NY 10027
(212) 280-8109
Arpanet: dyer@columbia-20
Columbia University is an Equal Opportunity/Affirmative Action Employer.
We welcome applications from qualified women and minorities.
------------------------------
End of PARSYM Digest
********************
∂26-Jul-85 1011 MARJORIE@SU-CSLI.ARPA tgif
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 26 Jul 85 10:11:01 PDT
Date: Fri 26 Jul 85 09:36:26-PDT
From: Marjorie Maxwell <MARJORIE@SU-CSLI.ARPA>
Subject: tgif
To: folks@SU-CSLI.ARPA
cc: consultants@SU-CSLI.ARPA
Next Friday, August 2, we will be having a TGIF to say goodbye to Frank
Chen who is on his way to UCLA Law School and hello to Peter King our
new programmer and UNIX administrator. For details please contact me.
Hope to see you all there.
Marjorie
-------
∂26-Jul-85 1224 avg@diablo They're holding our seminar!
Received: from SU-AIMVAX.ARPA by SU-AI.ARPA with TCP; 26 Jul 85 12:24:14 PDT
Date: Fri, 26 Jul 85 12:16:43 pdt
From: Allen VanGelder <avg@diablo>
Subject: They're holding our seminar!
To: nail@diablo
I am forwarding this message from a different mailing list in case some of you
are not on it. My apologies to those who got it already.
If we are going to stage a mass invasion, maybe we should let the
organizer know. If you plan to attend, but haven't been going to that
series, I suggest you send me a message. I will give him a head count.
From VAL@SU-AI.ARPA Fri Jul 26 11:23:11 1985
Received: from SU-AI.ARPA by diablo with TCP; Fri, 26 Jul 85 11:23:08 pdt
Date: 26 Jul 85 1119 PDT
From: Vladimir Lifschitz <VAL@SU-AI.ARPA>
Subject: Circumscription Seminar
To: "@CS.DST[1,VAL]"@SU-AI.ARPA
Status: R
An Application of A Query Evaluation Procedure to Computing Circumscription
Teodor C. Przymusinski
University of Texas at El Paso
2:00 PM, Wednesday, July 31
MJH 252 (new place!)
We introduce a minimal model query evaluation procedure (MQEP) which, given
a database DB and a query Q, verifies whether there exists a minimal model of
DB in which Q is satisfied.
In view of the fact that circumscription relative to DB implies a formula F
iff F is satisfied in all minimal models of DB, the procedure MQEP can be used
to evaluate queries in circumscriptive theories. Similarly, since the generalized
closed world assumption (GCWA) as defined by Minker can be viewed as a weak form
of circumscription, our procedure can also be used to evaluate queries under
the GCWA.
When applied to a Horn database, the procedure MQEP coincides with Clark's
query evaluation procedure QEP.
∂26-Jul-85 1336 @SU-CSLI.ARPA:GOGUEN@SRI-CSLA.ARPA Round table on Semantics of Programming Languages
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 26 Jul 85 13:36:12 PDT
Received: from SRI-CSLA.ARPA by SU-CSLI.ARPA with TCP; Fri 26 Jul 85 13:28:03-PDT
Date: Fri 26 Jul 85 13:28:58-PDT
From: Joseph A. Goguen <GOGUEN@SRI-CSLA.ARPA>
Subject: Round table on Semantics of Programming Languages
To: friends@SU-CSLI.ARPA
cc: Ashcroft@SRI-CSLA.ARPA, Boyer@SRI-CSLA.ARPA, Crow@SRI-CSLA.ARPA,
Denning@SRI-CSLA.ARPA, DHare@SRI-CSLA.ARPA, Elspas@SRI-CSLA.ARPA,
Geller@SRI-CSLA.ARPA, Geoff@SRI-CSLA.ARPA, Goguen@SRI-CSLA.ARPA,
Jagan@SRI-CSLA.ARPA, Jamieson@SRI-CSLA.ARPA, JGoldberg@SRI-CSLA.ARPA,
Joan@SRI-CSLA.ARPA, Kautz@SRI-CSLA.ARPA, Ladkin@SRI-CSLA.ARPA,
Lescanne@SRI-CSLA.ARPA, Levitt@SRI-CSLA.ARPA, Linde@SRI-CSLA.ARPA,
Melliar-Smith@SRI-CSLA.ARPA, Meseguer@SRI-CSLA.ARPA, MKL@SRI-CSLA.ARPA,
Morgenstern@SRI-CSLA.ARPA, Moriconi@SRI-CSLA.ARPA, Moser@SRI-CSLA.ARPA,
Neumann@SRI-CSLA.ARPA, Pease@SRI-CSLA.ARPA, Rosanna@SRI-CSLA.ARPA,
Rushby@SRI-CSLA.ARPA, Schwartz@SRI-CSLA.ARPA, Shostak@SRI-CSLA.ARPA,
Smolka@SRI-CSLA.ARPA, Tammy@SRI-CSLA.ARPA, VonHenke@SRI-CSLA.ARPA
Title: Round Table Discussion on Semantics of Programming Languages
Speakers: H. Ganzinger, P. Mosses, J. Meseguer, J. Goguen, J. Barwise(?),
plus strong audience participation
Place: Ventura Hall Seminar Room, CSLI, Stanford
Time: 12 noon, Thursday, August 1, 1985
Abstract:
We are fortunate to have visiting CSLI two experts on the semantics of
programming languages, who have unique and promising new approaches. Rather
than schedule yet another formal lecture, we will have a round table
discussion, featuring short presentions by the speakers, followed by
discussion among the speakers, followed by general discussion in which we
hope the audience will play a very strong role.
-------
∂26-Jul-85 2113 avg@diablo Proposed bagof semantics
Received: from SU-AIMVAX.ARPA by SU-AI.ARPA with TCP; 26 Jul 85 21:13:34 PDT
Date: Fri, 26 Jul 85 21:05:57 pdt
From: Allen VanGelder <avg@diablo>
Subject: Proposed bagof semantics
To: nail@diablo
BAGOF SEMANTICS
Another Wild-eyed Proposal from the Desk of ...
Allen Van Gelder
In the following expression, U represents variables not appearing elsewhere
in the rule, X represents "bound" variables appearing elsewhere, Y
represents unbound variables appearing elsewhere, and Z represents
variables appearing only in this expression.
bagof(t(U), e(U, X, Y, Z), L)
is represented by a relation having tuples pr#1(L, X, Y, Z) such that
L is a list in which each element has the form t(U), where t(U)
is a "solution tuple" of the "virtual rule"
t(U) :- e(U, X, Y, Z).
Note that e() represents a subgoal expression with possibly more than a single
literal, while t(U) represents a term, possibly with structure, whose
variables are U. Often this term is simply one variable. In this
case L is a list of U's bindings.
Also, pr#1 represents some generated predicate name used as a "place holder"
for the relation that represents the bagof.
Essentially, bagof can be captured when its "virtual rule"
formed by the first and second arguments is computable. Since a list
of successes must be formed, the virtual rule must define a finite
constructible relation. Thus certain expressions, principally involving
negated subgoals, will not be possible as bagof arguments.
It is an error for any U variable to appear outside of this "bagof."
Usually there will be no Y variables, on the assumption that it is
advantageous to bind such variables in other subgoals before working
on "bagof." However, they are treated like Z variables if present.
The relation has one tuple for each instantiation of Z (and Y) that
produces solutions. The L for that column is the list of solutions
for that Z. Each element in the list is of the form t(U).
To see how this works, let us look at some examples.
Assume we have a database containing these relations:
emp mgr
-------------------------- ----------
jones toy smith 12 loeb 13
green bath smith 13 smith 11
paul toy green 14 green 13
brown china loeb 15
rand china loeb 12
A SIMPLE EXAMPLE
We have the rules:
empDept(Name, Dept) :- emp(Name, Dept, ←, ←).
q(Dept, L) :- bagof(Name, empDept(Name, Dept), L).
Then the query ?- q(toy, L) has an answer containing one tuple,
which is
q(toy, [jones, paul]).
Alternatively, the query ?- (D, L) would group all employees in their
Depts, one Dept per tuple.
A different rule,
q2(L) :- bagof(Name, empDept(Name, Dept), L).
would also group employees by Dept, but now the value of Dept is lost.
A COMPLEX EXAMPLE
The user supplies the rule
p(X, N, L) :-
mgr(X, MgrSal) &
bagof(hp(Name,Sal), emp(Name,Dept,X,Sal) & Sal > MgrSal, L) &
length(L, Nhps) &
Nhps > N.
The value of L in each tuple will be a list of elements, hp(Name, Sal).
The "virtual rule" is:
hp(Name, Sal) :- emp(Name, Dept, X, Sal) & Sal > MgrSal.
Nail will rewrite the rule with generated predicates, say pr#1 and pr#2, as:
p(X, N, L) :-
mgr(X, MgrSal) &
pr#1(L, Dept, X, MgrSal) &
length(L, Nhps) &
Nhps > N.
pr#1(L, Dept, X, MgrSal) :-
bagof(hp(Name,Sal), pr#2(Name,Dept,X,Sal,MgrSal), L).
pr#2(Name, Dept, X, Sal, MgrSal) :-
emp(Name, Dept, X, Sal) &
Sal > MgrSal.
Nail should conclude that pr#2 ↑ ffffb is capturable. MgrSal must be bound
to satisfy the requirements of ">". It follows that pr#1 ↑ fffb is
capturable. Then p ↑ fbf is capturable, with the second argument again
required to be bound to satisfy ">".
Now let us closely examine how the evaluation of p proceeds.
The query ?- p(X, 1, L) first produces the subgoal pr#1(L, Dept, loeb, 13),
which has the relation with one tuple:
pr#1([hp(brown,15)], china, loeb, 13)
Its first argument has length only 1, so p fails on this try.
The next subgoal is pr#1(L, Dept, smith, 11), which has this relation:
pr#1([hp(jones,12)], toy, smith, 11)
pr#1([hp(green,13)], bath, smith, 11)
There are two tuples, one for each value of Dept. Thus even though
smith manages two employees that have higher salaries, there is no one
list of length > 1, so p fails on both these tuples.
Goal p also fails on the last try, involving subgoal pr#1(L, Dept, green, 13).
To keep things simple for the implementors, Nail does not provide a way
to specify that Dept should be existentially quantified and not an argument
of pr#1. To effect this, it is necessary to explicitly project it out
of emp in a separate rule.
emp1(Name,X,Sal) :- emp(Name,Dept,X,Sal).
p(X, N, L) :-
mgr(X, MgrSal) &
bagof(hp(Name,Sal), emp1(Name,X,Sal) & Sal > MgrSal, L) &
length(L, Nhps) &
Nhps > N.
In this case, there are no Z-variables, and we get the (probably desired)
relation with one tuple:
p(smith, 1, [hp(jones,12), hp(green,13)])
as the solution.
∂26-Jul-85 2226 ullman@diablo source language
Received: from SU-AIMVAX.ARPA by SU-AI.ARPA with TCP; 26 Jul 85 22:26:35 PDT
Date: Fri, 26 Jul 85 22:21:31 pdt
From: Jeff Ullman <ullman@diablo>
Subject: source language
To: nail@diablo
Perhaps a more mundane issue, but shall we use assert and retract
to insert and delete from existing DB relations, or to create new ones.
e.g., given a DB relation E, and rules like
t(X,Y) :- e(X,Y).
t(X,Y) :- t(X,Z) & t(Z,Y).
the "query" assert(t(X,Y)) instantiates the t relation,
and retract(t(0,W)) deletes from t all the pairs with first
component 0.
Well we can use asserta or assertz for assert if Allen insists.
∂26-Jul-85 2257 avg@diablo A simpler bagof
Received: from SU-AIMVAX.ARPA by SU-AI.ARPA with TCP; 26 Jul 85 22:56:54 PDT
Date: Fri, 26 Jul 85 22:51:43 pdt
From: Allen VanGelder <avg@diablo>
Subject: A simpler bagof
To: nail@diablo
BAGOF SEMANTICS, II
A Less Wild-eyed Proposal from the Desk of ...
Allen Van Gelder
In the following expression, U represents variables not appearing elsewhere
in the rule, X represents "bound" variables appearing elsewhere, Y
represents unbound variables appearing elsewhere, and Z represents
variables appearing only in this expression.
bagof(t(U), e(U, X, Y, Z), L)
is represented by a relation having tuples pr#1(L, X, Y) such that
L is a list in which each element has the form t(U), where t(U)
is a "solution tuple" of the "virtual rule"
t(U) :- e(U, X, Y, Z).
Note that e() represents a subgoal expression with possibly more than a single
literal, while t(U) represents a term, possibly with structure, whose
variables are U. Often this term is simply one variable. In this
case L is a list of U's bindings.
Also, pr#1 represents some generated predicate name used as a "place holder"
for the relation that represents the bagof.
Essentially, bagof can be captured when its "virtual rule"
formed by the first and second arguments is computable. Since a list
of successes must be formed, the virtual rule must define a finite
constructible relation. Thus certain expressions, principally involving
negated subgoals, will not be possible as bagof arguments.
It is an error for any U variable to appear outside of this "bagof."
Often there will be no Y variables, on the assumption that it is
advantageous to bind such variables in other subgoals before working
on "bagof." However, when there are Y variables,
the relation has one tuple for each instantiation of Y that
produces solutions. The L for that column is the list of solutions
for that Y and all Z. Each element in the list is of the form t(U).
To see how this works, let us look at some examples.
Assume we have a database containing these relations:
emp mgr
-------------------------- ----------
jones toy smith 12 loeb 13
green bath smith 13 smith 11
paul toy green 14 green 13
brown china loeb 15
rand china loeb 12
A SIMPLE EXAMPLE
We have the rule:
q(Dept, L) :- bagof(Name, emp(Name, Dept, ←, ←), L).
Then the query ?- q(toy, L) has an answer containing one tuple,
which is
q(toy, [jones, paul]).
Alternatively, the query ?- (D, L) would group all employees in their
Depts, one Dept per tuple.
A different rule, using the "name-dropping" predicate, =\, might be:
q2(L) :- =\ Dept & bagof(Name, emp(Name, Dept, ←, ←), L).
would also group employees by Dept, but now the value of Dept is lost.
A COMPLEX EXAMPLE
The user supplies the rule
p(X, N, L) :-
mgr(X, MgrSal) &
bagof(hp(Name,Sal), emp(Name,Dept,X,Sal) & Sal > MgrSal, L) &
length(L, Nhps) &
Nhps > N.
The value of L in each tuple will be a list of elements, hp(Name, Sal).
The "virtual rule" is:
hp(Name, Sal) :- emp(Name, Dept, X, Sal) & Sal > MgrSal.
Nail will rewrite the rule with generated predicates, say pr#1 and pr#2, as:
p(X, N, L) :-
mgr(X, MgrSal) &
pr#1(L, X, MgrSal) &
length(L, Nhps) &
Nhps > N.
pr#1(L, X, MgrSal) :-
bagof(hp(Name,Sal), pr#2(Name,X,Sal,MgrSal), L).
pr#2(Name, X, Sal, MgrSal) :-
emp(Name, Dept, X, Sal) &
Sal > MgrSal.
Nail should conclude that pr#2 ↑ fffb is capturable. MgrSal must be bound
to satisfy the requirements of ">". It follows that pr#1 ↑ ffb is
capturable. Then p ↑ fbf is capturable, with the second argument again
required to be bound to satisfy ">".
Now let us closely examine how the evaluation of p proceeds.
The query ?- p(X, 1, L) first produces the subgoal pr#1(L, loeb, 13),
which has the relation with one tuple:
pr#1([hp(brown,15)], china, loeb, 13)
Its first argument has length only 1, so p fails on this try.
The next subgoal is pr#1(L, smith, 11), which has this relation:
pr#1([hp(jones,12), hp(green,13)], smith, 11)
There is again one tuple, with a list of two elements as its first
argument. Thus p succeeds here, delivering the tuple
p(smith, 1, [hp(jones,12), hp(green,13)])
Goal p also fails on the last try, involving subgoal pr#1(L, Dept, green, 13),
becuase the list has length one.
∂28-Jul-85 2222 @SU-CSLI.ARPA:LANSKY@SRI-AI.ARPA Planlunch Reminder!
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 28 Jul 85 22:22:16 PDT
Received: from SRI-AI.ARPA by SU-CSLI.ARPA with TCP; Sun 28 Jul 85 22:13:24-PDT
Date: Sun 28 Jul 85 22:13:36-PDT
From: LANSKY@SRI-AI.ARPA
Subject: Planlunch Reminder!
To: aic-associates@SRI-AI.ARPA, folks@SU-CSLI.ARPA, mugs@SUMEX-AIM.ARPA,
val@SU-AI.ARPA, rsimmons@SRI-KL.ARPA, carnese@SRI-KL.ARPA,
alpert@SU-SUSHI.ARPA, frayman@SUMEX-AIM.ARPA, dmrussell@XEROX.ARPA,
vanLehn@XEROX.ARPA, araya@XEROX.ARPA, frayman@XEROX.ARPA,
suchman@XEROX.ARPA, weld@XEROX.ARPA, mittal@XEROX.ARPA, dekleer@XEROX.ARPA
Title: Time and Causation from the Standpoint of AI.
Nontitle: The Frame Problem is not.
Yoav Shoham
Yale University, SRI-AI
11:00 AM, Monday, July 29
SRI International, Building E, Room EJ232
Most tasks undertaken by AI researchers involve reasoning about
time in one way or another. In particular, the somewhat ill-defined
areas of planning and naive physics reasoning rely crucially on the
passage of time and the taking place of change. I am aiming at
a general and yet rigorous theory of time and change.
I am primarily interested in a useful notational device. Psychological
plausibility is only an added benefit, and philosophical truth
is something over which I lose little sleep.
The talk is structured as follows:
1. A first-order theory of time, which could be viewed as a generalization
of James Allen's theory.
2. A modal version of the same theory. Several interval-based modal logics
will be presented, along with the few of their theoretical properties
which I am beginning to understand.
3. The theory of causal counterfactuals, a particular theory of change
that relies on part 1 and which may be regarded as a generalization
of the STRIPS formalism. I will demonstrate how this theory appears
to avoid three major problems:
a. The cross-world identification problem for time tokens.
b. The frame problem.
c. A nameless problem encountered in the philosophy literature.
Also, in the spirit of recent "rigorous reconstructions" I will
reformulate Ken Forbus' Qualitative Process theory in terms of
causal counterfactuals.
In the unlikely event of our having time left over I'll discuss
Richard Waldinger's toy car.
-------
-------
∂28-Jul-85 2224 morris@diablo nail implementation: stage 1
Received: from SU-AIMVAX.ARPA by SU-AI.ARPA with TCP; 28 Jul 85 22:24:23 PDT
Date: 28 Jul 1985 2219-PDT (Sunday)
From: Kathy Morris <morris@diablo>
To: nail@diablo
Cc:
Reply-To:
Subject: nail implementation: stage 1
This is the first part of what I'm going to be doing.
The initial steps are to read in rules and predicates, and construct an
unadorned rule-goal graph. Then the strong components and the reduced
graph are extracted from this graph. The strong components and reduced
graph are the inputs to the next stage, which applies the capture
rules.
Input syntax:
This looks very much like prolog (actually the Cprolog
dialect [1]), with subgoals separated by ampersands (`&'), rather than commas.
(This idea is taken from Basser Prolog [2]) This means the comma
operator has fewer meanings than in conventional Prolog. Semicolons
are used for the 'or' operator, as in CProlog.
As the rules are read, some transformations are performed.
Extra predicates are used to remove uses of the semicolon
operator. So a rule like
a(X) :- b(X) & c(X, Y) ; d(X).
becomes:
a(X) :- p←0(X).
p←0(X) :- b(X) & c(X, Y).
p←0(X) :- d(X).
(p←0 is a generated predicate name).
Because we are using `&' rather than `,' for conjunctions, the
difficulties that C-Prolog has with spaces after operators don't occur.
So \+(a & b) is exactly the same structure as \+ (a, b). (In C-Prolog,
\+ (a, b) is an application of the \+ operator to the subgoals a, b,
but \+(a,b) is a compound term with functor \+ and arguments a and b.)
There are some transformations to be performed on rules involving
\+ and bagof, which have not (last time I looked) been finalised.
In particular, they will be transformed so that their argument
is always a single goal, not a conjunction or disjunction of goals. This
will involve adding extra rules.
Output:
This stage produces three predicates:
strongComponent(Predicate, NodeNumber, predicate).
strongComponent(Rule, NodeNumber, rule).
/* NodeNumber refers to a node in the rule-goal graph which is part
* of the strong component containing the rule or predicate appearing
* as the first argument. The third argument is a type tag. NodeNumber
* may or may not correspond to the node number of the first argument
* in the rule-goal graph.
*/
strongComponentNum(NodeNumber).
/* NodeNumber is a node number in the rule-goal graph which is used to
* identify a strong component.
*/
reducedGraph(StrongComponentNumber, InArcs, OutArcs).
/* This specifies the reduced rule-goal graph, with each strong
* component replaced with a vertex. StrongComponentNumber must appear
* as an argument to strongComponentNum. InArcs and OutArcs are lists
* of vertices entering and leaving the given strong component.
*/
An alternative for reducedGraph is simply to specify the arcs
individually, ie
reducedGraph(SCNIn, SCNOut).
/* The reduced rule-goal graph contains an arc from SCNIn to SCNOut. */
As Jeff has pointed out, there is no real need for the reducedGraph
to be passed explicitly -- the information in it can be easily be
obtained from the rule-goal graph.
I haven't yet decided on the exact format of the rule-goal graph.
The goal nodes will probably look something like:
rgGraph(NodeNum, GoalNode, SubGoalNumList, CaptureList).
SubGoalNumList is a list of the subgoal nodes (well, their node
numbers in the graph).
CaptureList is a list of adornments and the best method to use to
capture the node with those adornments, ie a list of:
capture([Var = b, Var2 = f], bottomup).
(The list is analogous to the one returned by read in C-Prolog -- it's
a list of Var = Name terms.)
We need to rank the capture rules in some order based on their relative
`goodness', and consider this when deciding if it's worth trying to
capture the node with different capture rules. For example: if we can
capture the node a↑ff, we're only interested in knowing if we can capture
the node a↑bf if we can do it with a better capture rule than the one we
used to capture a↑ff.
------------
[1] C-Prolog User's Manual (Version 1.4d.edai), Fernando Pereira (ed).
[2] Basser Prolog Users Manual, Kathy Morris and Andrew Taylor, TR243,
Basser Department of Computer Science, Sydney University (1984)
∂29-Jul-85 0035 RESTIVO@SU-SCORE.ARPA PROLOG Digest V3 #34
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 29 Jul 85 00:34:18 PDT
Date: Saturday, July 27, 1985 8:34PM
From: Chuck Restivo (The Moderator) <PROLOG-REQUEST@SU-SCORE.ARPA>
Reply-to: PROLOG@SU-SCORE.ARPA
US-Mail: P.O. Box 4584, Stanford CA 94305
Phone: (415) 326-5550
Subject: PROLOG Digest V3 #34
To: PROLOG@SU-SCORE.ARPA
PROLOG Digest Monday, 29 Jul 1985 Volume 3 : Issue 34
Today's Topics:
Implementation - Standards & Semantics & POPLOG,
LP Library - Shapiro's Debugger
----------------------------------------------------------------------
Date: Fri, 19 Jul 85 09:49:36 bst
From: William Clocksin <wfc%computer-lab.cambridge@ucl-cs>
Subject: Prolog standards
A number of contributors have recently displayed their
preferences about Prolog syntax, mentioning the context
of a Prolog standard. As Prolog has adherents for nearly
every syntactic convention imaginable, it is clear that a
Standard, if drafted, will not be able to satisfy everybody.
Part of the problem stems from how people like to see their
activity with Prolog: ordinary programming? Making
mathematical formalisms? Designing new languages?
The purpose of a standard is not to force people to program
in a particular way, but to determine whether a given
implementation is standard-conforming. Then, people can
offer standard-conforming implementations, and program in a
standard-conforming way. It need not cramp anyone's style
if they think they have no need to program in a standard
conforming way.
I am a member of a committee set up by the NPL (like the
US's NBS) and the BSI (a standards body) to attempt to draft
a Prolog standard. In our meetings we discuss in depth many
topics. Without wishing to denigrate Prolog Digest or its
contributors, I should say that we probably do a better job
of discussion than the slanging matches in Prolog Digest.
Contributions to Prolog Digest on this topic are important
because it give us on the committee. It would be even more
helpful if opinions were backed up by a short list of Pros
and Cons. For example, if you think that ',' should be
replaced by AND, give Pros and Cons of doing this. (Aside:
actually, ',' is not really an 'and' of any sort given the
usual Prolog execution strategy, so this would be an example
of a Con).
One more point. Believe it or not, there is a negligible
chance of having a national standard adopted as an international
standard unless the ISO 7-bit character set is used. This caters
for international variants, for example Swedish, which has three
extra letters. Character codes are reserved for the variants.
In English we use these slots for curly brackets among other
things. In a standard we are not allowed to use these slots!
This is so that non-English users can actually spell their name
by sacrificing curly brackets. And, we all know there are a lot
of non-English Prolog users. I welcome suggestions on a Prolog
syntax which does not use the characters reserved for national
variants. The relevant character code is ISO 646. The reserved
codes are what the English variant uses for square brackets, curly
brackets, backslash, vertical bar, tilde, up-arrow, at-sign, and
accent. Bother!
-- William Clocksin
------------------------------
Date: Sun 21 Jul 85 17:22:15-MDT
From: Uday Reddy <U-REDDY@UTAH-20.ARPA>
Subject: Semantics of equality
We have had some discussion on this topic a few months
ago, in which Joseph Goguen and Paul Voda participated.
In continuation of this discussion, I compare the
conventional semantics of equality with its
domain-theoretic semantics through a simple and elegant
example suggested to me by Seif Haridi.
Consider the equation
f(x,x) = 1
assuming no other equations for f. Now, given an
expression f(e,d) - where e and d are some expressions -
we want to test if f(e,d) = 1.
By equational logic semantics, f(e,d) would be 1, if e
and d can be proved equal from the equations of the
program. If both e and d are ground, and we know in
advance that they have normal forms, then we can reduce
them to their normal forms and check the normal forms
for syntactic equality. But, when they are non-ground,
or when we do not know if they have normal forms (i.e.
when we do not know if their reductions terminate), we
need to reduce or narrow e and d in parallel, and check
pairwise all their intermediate forms of expressions for
syntactic equality. This leads to potential
combinatorial explosion.
By domain-theoretic semantics, on the other hand, normal
forms of expressions are treated as their "values" and
attain semantic significance. Expressions which do not
have normal forms are deemed to have a special value
"unknown" (or "bottom") as their value. Similarly,
expressions which cannot be reduced completely, such as
f(1,2), are also deemed to have "unknown" as their
value. A partial order is then imposed on the domain of
values, by which "unknown" is treated as less defined
than the others.
"unknown" <= v for all normal values v
v <= v for all normal values v
Further, all functions are required to be monotonic with
respect to this partial order.
Coming to the problem of whether f(e,d) = 1, if e does
not have a normal form, then its value would be
"unknown". What can be the value of
f("unknown",v)
where v is some value? It can only be "unknown" or 1.
If it were 1, then, by the monotonicity of f, f(w,v)
would also have to be 1 for all values w. But, f(w,v) =
"unknown" whenever w and v are distinct values. So,
f("unknown",v) has to be "unknown". By a similar
argument f(e,d) has to be "unknown" if d is "unknown".
Even if it seems paradoxical, f("unknown", "unknown")
would also have to be "unknown".
The operational significance of this observation is that
if either e or d does not have a normal form, then
f(e,d) cannot be equal to 1. Hence, it is legal to
completely reduce e and d to their normal forms and then
check the normal forms for syntactic equality. The
combinatorial explosion is thus avoided.
Note that these considerations would not apply for
g(x,y) = 1
In this case, g("unknown", "unknown") can either be
"unknown" or 1 without violating the monotonicity
restriction. The language designer/implementor chooses
between the two choices depending on whether an
innermost or an outermost strategy is used in the
implementation.
As an aside, it is possible to distinguish between
expressions which do not have normal forms, and those
that have illegitimate normal forms, by introducing
another special value "fail" (similar to Prolog's
notion of fail) with the partial order
"unknown" <= "fail"
"unknown" <= v
v <= v
for all normal values v. Now, the function f would be
f("unknown", "unknown) |
f("unknown", "fail") |
f("fail", "unknown") | = "unknown"
f("unknown", v) |
f(v, "unknown") |
f("fail", "fail") |
f(v, "fail") | = "fail"
f("fail", v) |
f(v, w) = "fail" if v and w are distinct
f(v, v) = 1
for all normal values v and w.
-- Uday Reddy
------------------------------
Date: 23 Jul 1985 23:58:11-BST
From: Aaron Sloman <aarons%svgv@ucl-cs>
Subject: POPLOG - A mixed language development system.
Poplog is available on VAX and DEC 8600 computers.
It includes Prolog (compiled to machine code), Common Lisp (large
subset ready now, remainder available early 1986), POP-11
(comparable in power to Common Lisp, but uses a PASCAL-like syntax),
VED an integrated multi-window multi-buffer screen editor, which can
be used for all interactions with programs, operating system
utilities, online help, program libraries, teaching libraries, etc.
VED includes 'compile this procedure' 'compile from here to here'
'splice output into current file' etc.)
Incremental compilers are provided for Prolog, Lisp, and
POP-11. All the languages compile to the same intermediate
POPLOG 'Virtual machine' language, which is then compiled
to machine code. The 'syscompile' facilities make it easy
to add new front end compilers for additional languages,
which all share the same back-end compiler, editor and
environmental facilities. Mixed language facilities allow
sharing of libraries without re-coding and also allow
portions of a program to be written in the language which
is most suitable.
Approximate recent Prolog benchmarks, for naive reverse test,
without mode declarations:
VAX/780 + VMS 4.2 KLIPS
VAX/750 + Unix 4.2 2.4 KLIPS (750+Systime accelerator)
DEC 8600 13.0 KLIPS
SUN2 + Unix 4.2 2.5 KLIPS (also HP 9000/200)
GEC-63 + Unix V approx 6 KLIPS
The Prolog is being substantially re-written, for greater
modularity and improved efficiency. Mode declarations should
be available late 1985, giving substantial speed increase.
POP-11 and Common Lisp include both dynamic and lexical scoping,
a wide range of data-types, strings, arrays, infinite precision
arithmetic, hashed 'properties', etc. (Not yet packages, rationals
or complex numbers.) POP-11 includes a pattern-matcher (one-way
unification) with segment variables and pattern-restrictors.
External←load now allows 'external' modules to be linked in and
unlinked dynamically (e.g. programs written in C, Fortran, Pascal,
etc.). This almost amounts to a 'rapid prototyping' incremental
compiler for such languages.
A considerable number of AI-projects funded by the UK Alvey
Programme in universities and industry now use a mixture of
Prolog and POP-11, within Poplog.
Enquiries:
UK Educational institutions:
Alison Mudd,
Cognitive Studies Programme,
Sussex University,
Brighton, England. 0273 606755
-- Aaron Sloman
------------------------------
Date: Thu 25 Jul 85 14:34:21-CDT
From: Bill Murray <ATP.Murray@UTEXAS-20.ARPA>
Subject: Shapiro's Prolog debugging system
The next message contains the code from Ehud Shapiro's book
Algorithmic Program Debugging for a Prolog debugging system and for a
model inference system which can synthesize small Prolog programs from
examples. Instructions on how to run the scenarios are also included.
The instructions are in the file EXAMPLES..0. Caveats are that the
code runs on PROLOG on the DEC 2060, version number 3.3, but that it
won't necessarily run on other Prologs without some modifications
to account for dialectal differences [e.g. it doesn't run immediately
in Quintus PROLOG].
Regards,
-- Bill
[ the debugger is available as
SCORE:<Prolog>Shapiro←Debugger.pl -ed ]
------------------------------
End of PROLOG Digest
********************
∂29-Jul-85 1149 CLT mini seminar series
To: "@DIS.DIS[1,CLT]"@SU-AI.ARPA, su-bboards@SU-AI.ARPA
Speaker: Shigeki Gotoi
NTT Musashino Electrical Communication Laboratories
Title: Concurrency in Proof Normalization and Logic Programming
Time: Thursday, August 1, 4:00-5:00 pm
Place: Room 252 Margaret Jacks Hall
(Computer Science Department), Stanford
Abstract
Proof normalization manipulates formal proofs. It also
provides a computation mechanism which belongs to the logic
programming family.
Although proof normalization can treat full predicate
calculus, it is less practical than the well-known
programming language, Prolog.
In this talk, we propose a new technique of attaching proofs
to Skolem functions. This technique enables one to nomalize
a proof eagerly; that is, one can get a partial answer
before the proof is totally normalized. This improves the
usability of proof normalization. Partial answers are also
useful in normalizing proofs concurrently. We compare our
method with computation in Concurrent Prolog.
∂29-Jul-85 1200 ullman@diablo meeting
Received: from SU-AIMVAX.ARPA by SU-AI.ARPA with TCP; 29 Jul 85 11:36:41 PDT
Date: Mon, 29 Jul 85 11:26:49 pdt
From: Jeff Ullman <ullman@diablo>
Subject: meeting
To: nail@diablo
Since we seem to have so much to talk about, let's
again start the meeting at 10AM on July 31, in 301 MJH.
The next two Wednesdays, I'm going to be occupied by the WICS
course in the mornings. May we move the meeting back to 2PM
on the 7th and 14th? We actually have a speaker scheduled for
the 14th, by the way.
∂29-Jul-85 1234 @SU-CSLI.ARPA:CLT@SU-AI.ARPA mini seminar series
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 29 Jul 85 12:34:26 PDT
Received: from SU-AI.ARPA by SU-CSLI.ARPA with TCP; Mon 29 Jul 85 12:30:49-PDT
Date: 29 Jul 85 1149 PDT
From: Carolyn Talcott <CLT@SU-AI.ARPA>
Subject: mini seminar series
To: "@DIS.DIS[1,CLT]"@SU-AI.ARPA, su-bboards@SU-AI.ARPA
Speaker: Shigeki Gotoi
NTT Musashino Electrical Communication Laboratories
Title: Concurrency in Proof Normalization and Logic Programming
Time: Thursday, August 1, 4:00-5:00 pm
Place: Room 252 Margaret Jacks Hall
(Computer Science Department), Stanford
Abstract
Proof normalization manipulates formal proofs. It also
provides a computation mechanism which belongs to the logic
programming family.
Although proof normalization can treat full predicate
calculus, it is less practical than the well-known
programming language, Prolog.
In this talk, we propose a new technique of attaching proofs
to Skolem functions. This technique enables one to nomalize
a proof eagerly; that is, one can get a partial answer
before the proof is totally normalized. This improves the
usability of proof normalization. Partial answers are also
useful in normalizing proofs concurrently. We compare our
method with computation in Concurrent Prolog.
∂29-Jul-85 1247 avg@diablo simpler "bagof" followup
Received: from SU-AIMVAX.ARPA by SU-AI.ARPA with TCP; 29 Jul 85 12:30:43 PDT
Date: Mon, 29 Jul 85 12:26:29 pdt
From: Allen VanGelder <avg@diablo>
Subject: simpler "bagof" followup
To: nail@diablo
My second bagof proposal is actually closer to "findall" than "bagof"
in its handling of quantification. If we adopt it, or something like it,
we might as well call it "findall."
∂29-Jul-85 1649 @SU-SCORE.ARPA:welch@ames-vmsb.ARPA SIGBIG
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 29 Jul 85 16:49:21 PDT
Received: from ames-vmsb.ARPA by SU-SCORE.ARPA with TCP; Mon 29 Jul 85 16:43:18-PDT
Date: 29 Jul 85 16:23:00 PDT
From: welch@ames-vmsb.ARPA
Subject: SIGBIG
To: super@su-score.arpa
Reply-To: welch@ames-vmsb.ARPA
ASSOCIATION FOR COMPUTING MACHINERY
San Francisco Golden Gate Chapter
"SIGBIG" Special Interest Committee
For Large High Speed Computers
Meetings on the first Wednesday of each month at 7:30 PM. Speakers
who can give insights to various aspects of SUPERCOMPUTING are
featured each month.
Next meeting:
Wednesday, August 7,1985, 7:30 PM
Speaker: Steve Perrenod/CRI
Subject: CRAY 2
Location: Hewlett Packard Auditorium - Oak Room
Building #48
19447 Pruneridge , Cupertino CA
Travel: I-280 to Wolfe Rd., North one light and
right onto Pruneridge. There will be a
guard at the HP gate capable of directing
arrivals to building 48.
---------------------------------------------------------------
Tape-recordings of most of the previous may be obtained
in exchange for a tape cassette or $5.00 by contacting:
Mary Fowler (415)261-4058
Supercomputing #192, BOX 2787
Alameda, CA. 94501-0787
For information contact Mary Fowler, Chairperson (415) 261-4058
or Mike Austin, Publ. Chair (415) 423-8446
------
∂29-Jul-85 1710 EMMA@SU-CSLI.ARPA [Lauri Karttunen <Lauri@SU-CSLI.ARPA>: Question]
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 29 Jul 85 17:10:08 PDT
Date: Mon 29 Jul 85 17:05:28-PDT
From: Emma Pease <Emma@SU-CSLI.ARPA>
Subject: [Lauri Karttunen <Lauri@SU-CSLI.ARPA>: Question]
To: friends@SU-CSLI.ARPA
Tel: 497-3479
Mail-From: LAURI created at 29-Jul-85 17:04:04
Date: Mon 29 Jul 85 17:04:04-PDT
From: Lauri Karttunen <Lauri@SU-CSLI.ARPA>
Subject: Question
To: emma@SU-CSLI.ARPA
*** REVISED SCHEDULE ***
Program for the Workshop on Finite State Morphology
Ventura Hall, Trailor Class Room
July 29-30, 1985
Tuesday
10:00 Kenneth Church "Morphological Stress Decomposition and
Stress Assignment for Speech Synthesis"
11:00 William Poser "Locality Constraints on Phonological Rules"
1:30 Edward Barton "Complexity of Two-level Morphology"
2:00 Mark Johnson "Acquisition of a Restricted Set of
Phological Rules"
3:30 Michael Bateman "ATEF: A Finite State Model for Morphological
Analysis"
4:30 George Hankamer "Morphological Analysis of Turkish"
(Martin Kay's paper on Arabic was presented on Monday afternoon.)
-------
-------
∂30-Jul-85 0531 PARSYM-Request@SUMEX-AIM.ARPA PARSYM Digest V1 #7
Received: from SUMEX-AIM.ARPA by SU-AI.ARPA with TCP; 30 Jul 85 05:31:38 PDT
Date: 30 Jul 85 0524-PDT
From: Moderator Byron Davies <PARSYM-REQUEST@SUMEX-AIM.ARPA>
Reply-to: PARSYM@SUMEX-AIM.ARPA
Subject: PARSYM Digest V1 #7
To: PARSYM@SUMEX-AIM.ARPA
PARSYM Digest Tuesday, 30 Jul 1985 Volume 1 : Issue 7
Today's Topics:
Symbolic vs Numerical Computing
job implementing functional language for parallel architecure
----------------------------------------------------------------------
Date: Fri, 26 Jul 85 10:04 EDT
From: Seth Steinberg <sas@BBN-VAX.ARPA>
Subject: Symbolic vs Numerical Computing
We had better be careful here. Programs, not languages are either
symbolic or numerical. If I write a numerical matrix inverter in LISP
it is still a numerical program while if I write a MACSYMA-like
symbolic algebra matrix inverter in Fortran it is a symbolic program.
(Never mind why I would want to do the latter).
Numerical programs make the vast majority of their decisions based on
the values of terminal values which are stored in relatively
homogeneous data structures. Symbolic programs make signifigantly more
decisions based on examination of the structure of the data which may
vary more freely. The exact implementation of runtime typed data may
be part of the programming language (LISP or SMALLTALK types), assisted
by the programming language (PASCAL records used with case or C struct
unions), or it can be implemented in spite of the programming language
(FORTRAN, possibly with a package like SLIP or ASSEMBLY language using
the high bits for type and the rest for pointer). [Obviously, some
languages make symbolic programming easier than others].
This definition is far from perfect so I'll propose a test. Try
ranking a list of programs on the numerical<->symbolic scale. For
example (probably not in the right order):
Fast Fourier Transform
Sparse Matrix Multiply
Graph Coloring Algorithm
Simple Expression Compiler
ELIZA
LISP Interpreter
Simple Theorem Prover
Algebraic Integrator
Noun Phrase Parser
Try out your own ordering. Where would you put in things like:
FTP support for the Symbolics 3600
UNIX (kernel or cshell)
MacPaint vs. MacDraw
A Turing Machine
I'd be interested to see how these would be ranked or whether it is
meaningless to perform such rankings.
Seth Steinberg
SAS @ BBN-VAX
------------------------------
Date: Fri 26 Jul 85 11:23:50-PDT
From: Joseph A. Goguen <GOGUEN@SRI-CSLA.ARPA>
Subject: job implementing functional language for parallel architecure
Job Available at SRI International, Computer Science Lab
Title of Position: Computer Scientist
Duties: To assist in design and implementation of a programming environment
for a functional language. To implement the language and the environment on
a highly parallel computer architecture.
Experience: Practical experience in Lisp programming. Practical experience
with large software systems desirable. Knowledge of functional programming
useful.
Education: M.Sc. Computer Science or B.S. plus experience
Salary Range: $30,000 to $45,000
Date Needed: September 1, 1985
Contact: Donna Boyer, Computer Science Lab, SRI International, 333
Ravenswood Ave, Menlo Park Ca 94025
------------------------------
Date: Sat, 27 Jul 85 14:14:44 pdt
From: coraki!pratt@Navajo (Vaughan Pratt)
Date: Fri 19 Jul 85 21:25:35-CDT
From: Mayank Prakash <AI.Mayank@MCC.ARPA>
Subject: Re: What is symbolic computing?
First of all, I think that the terms
symbolic computing and numerical computing are different modes of
computing rather than mutually exclusive taxonomical categories.
Then, numerical computing is the mode when the major data elements are
numerical and one is interested in changing the numerical values of
these data elements. That is, both the input to and the output from
the program are mainly numerical. In symbolic computing, one is
interested mainly in manipulating structures. That is, both the input
and the output to the program are structures. Note that in this
definiton one mode of computing does not exclude the other. In fact,
most programs do some of each. It is the predominant activity of the
program that determines it's mode.
I do a lot of computing with:
* complex numbers
* polynomials over various fields, including the reals
* vector spaces of various dimensions
* linear transformations
* survey maps, involving bearings, lot boundaries (expressed as lists
of line segments), areas, etc.
* outline fonts based on conic splines
Now if these aren't examples of structured data I'm a monkey's uncle.
Yet most of the time spent manipulating these structures goes into
floating point operations. On the one hand this is certainly
consistent with Mayank's observation that "most programs do some of
each." On the other hand I don't see how to apply the test "predominant
activity." Is this determined by the proportion of time spent on
floating point operations? If so then does plugging in a floating point
accelerator convert my program from a numerical to a symbolic one?
Or is it determined by the number of calls to floating point routines
in my programs? Most of my calls are to things like dot and matrix
products.
One could look at it from a lower level as well. The memory cells in
the computer's data memory (as opposed to the instruction memory)
contain binary values. If they are mostly interpreted as representing
numbers, and the majority of operations that are carried out on them
are numerical, i.e., add, subtract, multiply, shift etc. their values,
then the program is a numerical mode program. If they are mostly
pointers to other cells in memory, and the operations on them are
mainly follow the pointers along, modify them to point to some other
cells etc., then the program is a symbolic mode program.
At this level the definition is vulnerable to the compiler. Given a
vector of reals a powerful optimizing compiler may see fit to implement
it either as a linked list (if the optimizer detects operations on the
vector that amount to expanding or contracting the vector) or an array
of contiguous locations (if there is much random access to the array).
How does one classify a program that leaves such decisions to the
compiler?
A characteristic that generally distinguishes the languages for the
two kinds of programs is memory allocation. The languages developed
for numerical processing have mostly static memory allocation schemes.
By that I mean that the data memory is allocated to a procedure (or a
function, or block, whatever you want to call it) upon entry, and
released upon exit. Generally, though not always, the procedure does
not (and in most cases, can not) change its data memory. In contrast,
symbolic processing languages have dynamic memory allocation with
attendant garbage collection. This is necessary for symbolic
computing since in this case one is dealing with structures, which are
essentially pointers pointing at each other in various ways, and since
the main activity here is manipulating these structures, i.e.,
releasing and allocating pointers.
When memory is allocated and released in LIFO order it is very efficient
to allocate it off a stack. When the release order is unpredictable
one resorts to a heap. How does this have anything to do with whether
numeric data types are involved?
-v
------------------------------
End of PARSYM Digest
********************
∂30-Jul-85 1042 DAVIES@SUMEX-AIM.ARPA Davies Open House
Received: from SUMEX-AIM.ARPA by SU-AI.ARPA with TCP; 30 Jul 85 10:41:07 PDT
Date: Tue, 30 Jul 1985 10:32 PDT
Message-ID: <DAVIES.12131167301.BABYL@SUMEX-AIM.ARPA>
From: DAVIES@SUMEX-AIM.ARPA
To: AAP@SUMEX-AIM.ARPA
Subject: Davies Open House
Byron and Vicky Davies will be having an OPEN HOUSE and WINE COOLER
TASTING this Sunday from 1 to 5 pm. Please drop by to help us
initiate our new house.
The address is 59 Walnut Avenue in Atherton. If you're going logical
north on El Camino, Walnut is on the right, about half a mile after
you enter Atherton from Menlo Park. If you pass a traffic light in
Atherton (Fair Oaks/Atherton Ave.), you've gone to far. If you're
proceeding logical south on El Camino, Walnut is on the left shortly
after you cross Atherton Ave.
-- Byron
∂30-Jul-85 1153 MARJORIE@SU-CSLI.ARPA tgif
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 30 Jul 85 11:50:18 PDT
Date: Tue 30 Jul 85 11:42:25-PDT
From: Marjorie Maxwell <MARJORIE@SU-CSLI.ARPA>
Subject: tgif
To: folks@SU-CSLI.ARPA
Update on Tgif for Frank Chen this Friday, August 2 - it will start about
4:00 PM - appetizers and drinks - see you all there to say goodbye to Frank.
-------
∂30-Jul-85 1251 DAVIES@SUMEX-AIM.ARPA No meeting this week
Received: from SUMEX-AIM.ARPA by SU-AI.ARPA with TCP; 30 Jul 85 12:47:49 PDT
Date: Tue, 30 Jul 1985 12:47 PDT
Message-ID: <DAVIES.12131191938.BABYL@SUMEX-AIM.ARPA>
From: DAVIES@SUMEX-AIM.ARPA
To: AAP@SUMEX-AIM.ARPA
Subject: No meeting this week
There will be no architectures meeting this week. Happy hacking!
-- Byron
∂30-Jul-85 1519 avg@diablo Reminder: Query Eval seminar Wed., 2PM
Received: from SU-AIMVAX.ARPA by SU-AI.ARPA with TCP; 30 Jul 85 15:19:27 PDT
Date: Tue, 30 Jul 85 15:11:46 pdt
From: Allen VanGelder <avg@diablo>
Subject: Reminder: Query Eval seminar Wed., 2PM
To: nail@diablo
An Application of A Query Evaluation Procedure to Computing Circumscription
Teodor C. Przymusinski
University of Texas at El Paso
2:00 PM, Wednesday, July 31
MJH 252 (new place!)
We introduce a minimal model query evaluation procedure (MQEP) which, given
a database DB and a query Q, verifies whether there exists a minimal model of
DB in which Q is satisfied.
In view of the fact that circumscription relative to DB implies a formula F
iff F is satisfied in all minimal models of DB, the procedure MQEP can be used
to evaluate queries in circumscriptive theories. Similarly, since the generalized
closed world assumption (GCWA) as defined by Minker can be viewed as a weak form
of circumscription, our procedure can also be used to evaluate queries under
the GCWA.
When applied to a Horn database, the procedure MQEP coincides with Clark's
query evaluation procedure QEP.
∂30-Jul-85 2045 @SU-SUSHI.ARPA:HERSHBERGER@SU-SCORE.ARPA Rides to BATS
Received: from SU-SUSHI.ARPA by SU-AI.ARPA with TCP; 30 Jul 85 20:45:00 PDT
Received: from SU-SCORE.ARPA by SU-SUSHI.ARPA with TCP; Tue 30 Jul 85 18:53:24-PDT
Date: Tue 30 Jul 85 18:53:14-PDT
From: John E. Hershberger <HERSHBERGER@SU-SCORE.ARPA>
Subject: Rides to BATS
To: HERSHBERGER@SU-SCORE.ARPA, ashok@SU-SUSHI.ARPA, avg@SU-AIMVAX.ARPA,
GANGOLLI@SU-SUSHI.ARPA, FOULSER@SU-SCORE.ARPA, roy@SU-AIMVAX.ARPA,
HELMBOLD@SU-SCORE.ARPA, MAYR@SU-SCORE.ARPA, KARLIN@SU-SUSHI.ARPA
cc: aflb.local@SU-SCORE.ARPA
Message-ID: <12131258502.40.HERSHBERGER@SU-SCORE.ARPA>
As usual, there are enough drivers to take care of the riders. So far, the
confirmed drivers are
Ernst Mayr, Dave Helmbold, Shaibal Roy, and John Hershberger.
Barbara Simons, who lives in Palo Alto, also said that she could
take some people down to BATS. The riders that I know of are
Anil Gangolli, Ashok Subramanian, Allen Van Gelder, Anna Karlin,
and Dave Foulser.
Anil and Ashok plan to go with Ernst, so he has one more space. Shaibal has
one more space and needs to leave BATS by 3:15 at the latest. Dave Helmbold
plans to leave from his apartment in Mountain View. I will be driving down but
will be staying there after BATS, so I can only give rides one way. I haven't
talked to Dave Helmbold yet, but perhaps he could bring back anybody who rides
down with me. (Both his and my car hold three people in addition to the
driver.) (OK, Dave?)
Is 8:30 a good time for those of us who are leaving from MJH to meet?
If we should leave earlier or later, somebody please let me/us know;
otherwise, let's plan on meeting then.
John
-------
∂30-Jul-85 2046 @SU-SUSHI.ARPA:HERSHBERGER@SU-SCORE.ARPA 8:45 BATS meeting
Received: from SU-SUSHI.ARPA by SU-AI.ARPA with TCP; 30 Jul 85 20:46:05 PDT
Received: from SU-SCORE.ARPA by SU-SUSHI.ARPA with TCP; Tue 30 Jul 85 19:12:18-PDT
Date: Tue 30 Jul 85 19:12:11-PDT
From: John E. Hershberger <HERSHBERGER@SU-SCORE.ARPA>
Subject: 8:45 BATS meeting
To: aflb.local@SU-SCORE.ARPA
Message-ID: <12131261950.40.HERSHBERGER@SU-SCORE.ARPA>
To BATS-goers:
Ernst suggests that those of us leaving from MJH can leave later than 8:30.
Unless I hear further suggestions, let's plan on meeting about 8:45 and
leaving at or before 9.
John
-------
∂31-Jul-85 1119 MARJORIE@SU-CSLI.ARPA tgif
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 31 Jul 85 11:19:09 PDT
Date: Wed 31 Jul 85 11:07:07-PDT
From: Marjorie Maxwell <MARJORIE@SU-CSLI.ARPA>
Subject: tgif
To: Folks@SU-CSLI.ARPA
If anyone is interested in contributing their culinary talents (appetizers)
to Friday's TGIF that would be great. Please either see (or contact) me
or Susi.
Marjorie
-------
∂31-Jul-85 1226 YAMARONE@SU-CSLI.ARPA SANDWICH UPDATE:NEW PRICES
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 31 Jul 85 12:25:52 PDT
Date: Wed 31 Jul 85 12:20:02-PDT
From: Tom Yamarone <YAMARONE@SU-CSLI.ARPA>
Subject: SANDWICH UPDATE:NEW PRICES
To: folks@SU-CSLI.ARPA
In order to reduce our operating deficit, The Ventura Sandwich Corp. is
announcing a new price structure:
All basic sandwiches will remain the same: 2.50
Any sandwich with cheese extra(e.g. ham +swiss, avo.+ jack, turkey + swiss)
will now be 3.00
A roast beef is always 3.00 and with cheese will be 3.50
Milk will now be .50
Apples are still .25
Thank you for ordering before 10:00 to "Lunch @ Turing"
and we look forward to serving you in the future!
-------
∂31-Jul-85 1606 @SU-CSLI.ARPA:BMOORE@SRI-AI.ARPA party reminder
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 31 Jul 85 16:05:57 PDT
Received: from SRI-AI.ARPA by SU-CSLI.ARPA with TCP; Wed 31 Jul 85 16:00:35-PDT
Date: Wed 31 Jul 85 16:00:21-PDT
From: Bob Moore <BMOORE@SRI-AI.ARPA>
Subject: party reminder
To: party-list: ;
Just a reminder to people who plan to come, but haven't replied to the
first message. We would really like to know that you are coming, so
that we know how many people to expect. Directions will be sent
tomorrow to all those who requested them.
--Bob
---------------
Return-Path: <@SU-CSLI.ARPA:BMOORE@SRI-AI.ARPA>
Received: from SU-CSLI.ARPA by SRI-AI.ARPA with TCP; Wed 24 Jul 85 12:00:56-PDT
Received: from SRI-AI.ARPA by SU-CSLI.ARPA with TCP; Wed 24 Jul 85 11:56:41-PDT
Date: Wed 24 Jul 85 11:55:49-PDT
From: Bob Moore <BMOORE@SRI-AI.ARPA>
Subject: PARTY! PARTY! PARTY!
To: party-list: ;
To celebrate the completion of our remodeling and landscaping, and our
upcomming departure for England, we are having a party. Good food,
good wine, and good conversation will be provided. Time: Sunday,
August 4, 2:00-6:00 PM. Place: 737 DeSoto Drive, Palo Alto.
(Directions provided on request.) If you plan to come, please let us
know, so we know how many people to expect.
Bob and Rita Moore
328-4253
BMOORE@SRI-AI
-------
-------
∂31-Jul-85 1707 EMMA@SU-CSLI.ARPA Newsletter August 1, No 39
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 31 Jul 85 17:07:05 PDT
Date: Wed 31 Jul 85 16:53:17-PDT
From: Emma Pease <Emma@SU-CSLI.ARPA>
Subject: Newsletter August 1, No 39
To: friends@SU-CSLI.ARPA
Tel: 497-3479
C S L I N E W S L E T T E R
←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←
August 1, 1985 Stanford Vol. 2, No. 39
←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←
A weekly publication of The Center for the Study of Language and
Information, Ventura Hall, Stanford University, Stanford, CA 94305
←←←←←←←←←←←←
CSLI ACTIVITIES FOR *THIS* THURSDAY, August 1, 1985
12 noon CSLI Lunch
Ventura Hall ``Round Table Discussion on Semantics of
Conference Room Programming Languages''
(Abstract on page 1)
2:15 p.m. CSLI Talk
Ventura Hall ``Realism and Antirealism in Cognitive Artificial
Conference Room Intelligence''
David H. Helman, Department of Philosophy, Case
Western Reserve University
Discussion led by Ivan Blair
3:30 p.m. Tea
Ventura Hall
←←←←←←←←←←←←
CSLI LUNCH
``Round Table Discussion on Semantics of Programming Languages''
12 noon, Thursday, August 1, Ventura Hall Seminar Room
We are fortunate to have visiting CSLI two experts on the semantics
of programming languages, who have unique and promising new
approaches. Rather than schedule yet another formal lecture, we will
have a round table discussion, featuring short presentions by the
speakers, followed by discussion among the speakers, followed by
general discussion in which we hope the audience will play a very
strong role.
Speakers will include H. Ganzinger, P. Mosses, J. Meseguer, and J.
Goguen plus strong audience participation. --Joseph Goguen
←←←←←←←←←←←←
CSLI TALK
``The Processing of Motives in Intelligent Systems''
Aaron Sloman, University of Sussex
4 p.m., August 6, Tuesday, Ventura Conference Room
←←←←←←←←←←←←
NEW CSLI REPORT
Report No. CSLI-85-27, ``Semantic Automata'' by Johan van Benthem,
has just been published. This report may be obtained by writing to
David Brown, CSLI, Ventura Hall, Stanford, CA 94305 or Brown@su-csli.
!
Page 2 CSLI Newsletter August 1, 1985
←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←←
INTERACTIONS OF MORPHOLOGY, SYNTAX, AND DISCOURSE
Summary of the meeting on Thursday, July 25
Sells, Zaenen, and Zec presented a typology of reflexive
constructions, showing that there is no simple correlation between the
way the reflexive is morphologically realized and the behavior of the
form as a transitive or an intransitive. They defined three notions of
transitivity and showed which combinations can occur in reflexive
forms:
1. Lexical transitivity, testable through the interactions with
lexical rules that behave differently when applied to verbs that take
objects and those that do not (e.g. causativization and impersonal
passive in several languages)
2. C-structure transitivity, the property of having an overt NP or
pronoun as a PS-constituent separated from the verb in the position
normally assigned to OBJECTS; c-structure intransitive then means to
have no PS-constituent in object position
3. Semantic transitivity, the property of being a two-place
predicate; semantic intransitives then, are one-place predicates
including the ones that are 'derived' from two-place predicates by
variable binding.
Reflexive constructions can be not only intransitive (e.g. Finnish)
or transitive (e.g. English or Walpiri) along all these dimensions at
once but the following cases are also found:
a. Lexically intransitive, c-structure transitive and semantically
intransitive (e.g. German and Serbo-Croatian)
b. Lexically transitive, c-structure transitive and semantically
intransitive (e.g. Dutch and Japanese)
c. Lexically transitive, c-structure intransitive and semantically
transitive (e.g. Chichewa).
The combinations they postulate not to exist are the ones involving
a lexically intransitive and a semantically transitive reflexive form.
Another session will be devoted to the presentation of a theory that
captures the generalizations presented, involving some developments in
the format of lexical rules and a sketch of the integration of DRS and
LFG. --Annie Zaenen
-------
∂31-Jul-85 1721 @SU-CSLI.ARPA:LANSKY@SRI-AI.ARPA NEXT MONDAY'S PLANLUNCH
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 31 Jul 85 17:21:10 PDT
Received: from SRI-AI.ARPA by SU-CSLI.ARPA with TCP; Wed 31 Jul 85 17:11:06-PDT
Date: Wed 31 Jul 85 16:55:51-PDT
From: LANSKY@SRI-AI.ARPA
Subject: NEXT MONDAY'S PLANLUNCH
To: aic-associates@SRI-AI.ARPA, folks@SU-CSLI.ARPA, mugs@SUMEX-AIM.ARPA,
val@SU-AI.ARPA, rsimmons@SRI-KL.ARPA, carnese@SRI-KL.ARPA,
alpert@SU-SUSHI.ARPA, frayman@SUMEX-AIM.ARPA, dmrussell@XEROX.ARPA,
vanLehn@XEROX.ARPA, araya@XEROX.ARPA, frayman@XEROX.ARPA,
suchman@XEROX.ARPA, weld@XEROX.ARPA, mittal@XEROX.ARPA, dekleer@XEROX.ARPA
How to Plan an Action When You Don't Know What to Do:
A Logic of Knowledge, Action, and Communication
Leora Morgenstern
New York University, SRI-AIC
11:00 AM, Monday, August 5
SRI International, Building E, Room EJ232
Most AI planners work on the assumption that they have complete knowledge
of their problem domain and situation, so that formulating a plan really
consists of searching through some pre-packaged list of action operators
for an action sequence that achieves some desired goal. Real life planning
rarely works this way, because we usually don't have enough information to
map out a detailed plan of action when we start out. Instead, we initially
draw up a sketchy plan and fill in details as we proceed and gain more
exact information about the world. This talk will present a formalism
that is expressive enough to describe this flexible planning process.
The talk will consist of 5 (hopefully short) parts:
1. Motivation for a flexible logic of knowledge, action, and communication,
2. Discussion of Bob Moore's modal logic of knowledge and action,
its advantages, and its limitations with respect to a robust theory
of planning,
3. A move towards a syntactic theory of knowledge, and a discussion of the
resulting paradoxes (esp. the Knower Paradox),
4. A solution to the Knower Paradox based on Kripke's solution to the
Liar Paradox,
5. A solution to the problem of knowledge preconditions.
-------
-------
∂01-Aug-85 1037 ROSENBLOOM@SUMEX-AIM.ARPA Siglunch this Friday (August 2nd)
Received: from SUMEX-AIM.ARPA by SU-AI.ARPA with TCP; 1 Aug 85 10:37:16 PDT
Date: Thu 1 Aug 85 10:28:51-PDT
From: Paul Rosenbloom <ROSENBLOOM@SUMEX-AIM.ARPA>
Subject: Siglunch this Friday (August 2nd)
To: siglunch@SUMEX-AIM.ARPA
Message-ID: <12131690970.13.ROSENBLOOM@SUMEX-AIM.ARPA>
There will be a SigLunch this Friday August 2 in the Chemistry Gazebo
at 12:05.
---------------------------
There's More to Menu Systems than Meets the Screen
Henry Lieberman
Artificial Intelligence Laboratory
Massachusetts Institute of Technology
Love playing with those fancy menu-based graphical user interfaces,
but afraid to program one yourself for your own application? Do
windows seem opaque to you? Are you scared of mice? Like
what-you-see-is-what-you-get but don't know how to get what you want
to see on the screen?
Everyone agrees using systems like graphical document illustrators,
circuit designers, and iconic file systems is fun, but programming
user interfaces for these systems isn't as much fun as it should be.
Systems like the Lisp Machines, Xerox D-Machines, and Apple Macintosh
provide powerful graphics primitives, but the casual applications
designer is often stymied by the difficulty of mastering the details
of window specification, multiple processes, interpreting mouse input,
etc.
This paper presents a kit called EZWin, which provides many services
common to implementing a wide variety of interfaces, described as
generalized editors for sets of graphical objects. An individual
application is programmed simply by creating objects to represent the
interface itself, each kind of graphical object, and each command. A
unique interaction style is established which is insensitive to
whether commands are chosen before or after their arguments. The
system anticipates the types of arguments needed by commands,
preventing selection mistakes which are a common source of frustrating
errors. Displayed objects are made "mouse-sensitive" only if
selection of the object is appropriate in the current context. The
implementation of a graphical interface for a computer network
simulation is described to illustrate how EZWin works.
-------
∂01-Aug-85 1048 ullman@diablo meetings
Received: from SU-AIMVAX.ARPA by SU-AI.ARPA with TCP; 1 Aug 85 10:48:41 PDT
Date: Thu, 1 Aug 85 10:43:22 pdt
From: Jeff Ullman <ullman@diablo>
Subject: meetings
To: nail@diablo
We meet at 1PM on each of the next two Wednesdays.
On the 7th it's in 301 MJH, and on the 14th in 352 MJH.
Next time, I'd like to talk about "magic sets" again.
∂01-Aug-85 1300 @SU-SUSHI.ARPA,@SU-SCORE.ARPA:udi@wisc-rsch.arpa CALL FOR PAPERS -- ACM symposium on Principles of Database systems
Received: from SU-SUSHI.ARPA by SU-AI.ARPA with TCP; 1 Aug 85 13:00:18 PDT
Received: from SU-SCORE.ARPA by SU-SUSHI.ARPA with TCP; Thu 1 Aug 85 12:54:06-PDT
Received: from wisc-rsch.arpa by SU-SCORE.ARPA with TCP; Thu 1 Aug 85 12:50:00-PDT
Received: by wisc-rsch.arpa; Thu, 1 Aug 85 14:34:17 cdt
Received: from wisc-crys.arpa by wisc-rsch.arpa; Wed, 31 Jul 85 16:12:44 cdt
Received: from cca-unix.ARPA by wisc-crys.arpa; Wed, 31 Jul 85 16:12:34 cdt
Received: by cca-unix.ARPA (4.12/4.7)
id AA03168; Tue, 30 Jul 85 19:29:56 edt
Date: Tue, 30 Jul 85 19:29:56 edt
From: nori@cca-unix
Message-Id: <8507302329.AA03168@cca-unix.ARPA>
To: theory@uwisc
Subject: CALL FOR PAPERS -- ACM symposium on Principles of Database systems
Cc: nori@cca-unix
Sender: udi@wisc-rsch.arpa
Resent-From: udi@wisc-rsch.arpa (Udi Manber)
Resent-To: theory:;@wisc-rsch.arpa
CALL FOR PAPERS
Fifth ACM SIGACT-SIGMOD Symposium on
PRINCIPLES OF DATABASE SYSTEMS
Cambridge, Massachusetts, March 24-26, 1986
The conference will cover new developments in both the theoretical and
practical aspects of database systems. Papers are solicited which
describe original and novel research about the theory, design,
specification, or implementation of database systems and query
languages.
Some suggested, although not exclusive, topics of interest are:
artifical intelligence for databases, concurrency control, database
design, database security, data models, data structures for databases,
dependency theory, distributed databases, file organization, logic for
databases, performance evaluation of database systems, query languages,
and schema design.
You are invited to submit nine (9) copies of a detailed abstract (not a
complete paper) to the program chairman:
Avi Silberschatz
Department of Computer Sciences
The University of Texas at Austin
Austin, TX 78712
(512) 471-4353
ARPANET: avi@utexas-20
Submissions will be evaluated on the basis of significance, originality,
and overall quality. Each abstract should 1) contain enough information
to enable the program committee to identify the main contribution of the
work; 2) explain the importance of the work - its novelty and its
practical or theoretical relevance to database management; and 3)
include comparisons with and references to relevant literature.
Abstracts should be no longer than ten double-spaced pages.
Program Committee:
Francois Bancilhon Christos Papadimitriou
Hector Garcia-Molina Ed Sciore
Jim Gray Avi Silberschatz
Alberto Mendelzon Moshe Vardi
Meral Ozsoyoglu
The deadline for submission of abstracts is October 11, 1985. Authors
will be notified of acceptance or rejection by December 6, 1985. The
accepted papers, typed on special forms, will be due at the above
address by January 10, 1986. All authors of accepted papers will be
expected to sign copyright release forms. Proceedings will be
distributed at the conference, and will be subsequently available for
purchase through ACM.
General Chairman: Local Arrangements Chairman:
Ashok K. Chandra Arvola Chan
IBM T.J. Watson Research Center Computer Corporation of America
P.O. Box 218 4 Cambridge Center
Yorktown Heights, NY 10598 Cambridge, MA 02142.
(914) 945-1752 (617) 492-8860
∂01-Aug-85 1352 ullman@diablo PODS call for papers
Received: from SU-AIMVAX.ARPA by SU-AI.ARPA with TCP; 1 Aug 85 13:52:09 PDT
Date: Thu, 1 Aug 85 13:44:32 pdt
From: Jeff Ullman <ullman@diablo>
Subject: PODS call for papers
To: nail@diablo
CALL FOR PAPERS
Fifth ACM SIGACT-SIGMOD Symposium on
PRINCIPLES OF DATABASE SYSTEMS
Cambridge, Massachusetts, March 24-26, 1986
The conference will cover new developments in both the theoretical and
practical aspects of database systems. Papers are solicited which
describe original and novel research about the theory, design,
specification, or implementation of database systems and query
languages.
Some suggested, although not exclusive, topics of interest are:
artifical intelligence for databases, concurrency control, database
design, database security, data models, data structures for databases,
dependency theory, distributed databases, file organization, logic for
databases, performance evaluation of database systems, query languages,
and schema design.
You are invited to submit nine (9) copies of a detailed abstract (not a
complete paper) to the program chairman:
Avi Silberschatz
Department of Computer Sciences
The University of Texas at Austin
Austin, TX 78712
(512) 471-4353
ARPANET: avi@utexas-20
Submissions will be evaluated on the basis of significance, originality,
and overall quality. Each abstract should 1) contain enough information
to enable the program committee to identify the main contribution of the
work; 2) explain the importance of the work - its novelty and its
practical or theoretical relevance to database management; and 3)
include comparisons with and references to relevant literature.
Abstracts should be no longer than ten double-spaced pages.
Program Committee:
Francois Bancilhon Christos Papadimitriou
Hector Garcia-Molina Ed Sciore
Jim Gray Avi Silberschatz
Alberto Mendelzon Moshe Vardi
Meral Ozsoyoglu
The deadline for submission of abstracts is October 11, 1985. Authors
will be notified of acceptance or rejection by December 6, 1985. The
accepted papers, typed on special forms, will be due at the above
address by January 10, 1986. All authors of accepted papers will be
expected to sign copyright release forms. Proceedings will be
distributed at the conference, and will be subsequently available for
purchase through ACM.
General Chairman: Local Arrangements Chairman:
Ashok K. Chandra Arvola Chan
IBM T.J. Watson Research Center Computer Corporation of America
P.O. Box 218 4 Cambridge Center
Yorktown Heights, NY 10598 Cambridge, MA 02142.
(914) 945-1752 (617) 492-8860
∂01-Aug-85 1802 PARSYM-Request@SUMEX-AIM.ARPA PARSYM Digest V1 #8
Received: from SUMEX-AIM.ARPA by SU-AI.ARPA with TCP; 1 Aug 85 18:01:51 PDT
Date: 1 Aug 85 1755-PDT
From: Moderator Byron Davies <PARSYM-REQUEST@SUMEX-AIM.ARPA>
Reply-to: PARSYM@SUMEX-AIM.ARPA
Subject: PARSYM Digest V1 #8
To: PARSYM@SUMEX-AIM.ARPA
PARSYM Digest Thursday, 1 Aug 1985 Volume 1 : Issue 8
Today's Topics:
Symbolic and numerical computing
Parallel symbolic computing in industry
----------------------------------------------------------------------
Date: Wed, 31 Jul 85 11:59 EDT
From: Guy Steele <gls@THINK-AQUINAS.ARPA>
Subject: Symbolic and numerical computing
Symbolic programs:
* laugh at themselves.
* philosophize.
* till the soil.
* are featherless bipeds.
Here's a more serious attempt: ALL computing applications are symbolic.
All applications rely on processing data organized according to some
structural discipline. This discipline may be trivial or exceedingly
complex. There are typically certain invariants or axioms of the
structure, and the operations on the structure, on which the processing
relies: for example, that a tree is binary and balanced, and insertion
and deletion maintain these invariants.
A particularly large and important class of applications relies heavily
on the axioms for rings and fields, particularly the ring of integers
and the fields of real and complex numbers, and for such applications
much of the computation is done with data structures amnd operations
that are organized so as to obey these axioms (more or less, given the
usual finiteness of the representations). Because these applications
are so important, and the theory is well-understood and agreed-upon,
special hardware accelerators for certain very complicated operations
(such as multiplication) are the norm rather than the exception; but
thes presence or absense of such hardware has, to my mind, little
bearing on the numericalness of the application.
So I propose that an application be considered numerical to the extent
that it relies on data structures obeying the axioms for rings or
fields, however these data structures may be represented as bits. I
would regard a LISP program operating on lists of NIL's as numerical if
it were so organized as to treat these lists primarily as unary
encodings of numbers, using routines to concatenate the lists (addition)
and repeatedly self-concatenate (double, multiply), and so on. (Indeed,
the SCHEME chips that I and others designed to directly interpret LISP
code had no on-chip ALU to speak of, and the chips were tested on
numerical applications using numbers encoded in exactly this way.)
Is a program that relies on a group structure numerical? What if there
is hardware to compute a*b quickly, where * is the group operator?
Suppose the group were SU(16) or GF(16) instead of Z[2↑16]? By the
proposed criterion, any such program might be somewhat numerical,
but less so than one using a ring or field.
--Guy Steele
------------------------------
Date: Thu, 1 Aug 1985 13:44 PDT
From: DAVIES@SUMEX-AIM.ARPA
Subject: Parallel Symbolic Computing in Industry
[Here's a short PARSYM-related piece from the August 1985 issue of Inc.
"Start-up Fever May Return to Computer Industry" focuses on the the
new industrial wave of parallel computing. The Connection Machine
from Thinking Machines is a fine-grained, massively parallel symbolic
processor. As reported in an earlier PARSYM digest, BBN's Butterfly,
although not specifically designed for symbolic computing, supports
Multilisp, a multiprocessing dialect of Lisp. -- BD]
Two companies that recently announced new computers weren't
telling the usual dull tale of bits and bytes. Looked at
historically, their machines actually represent a major departure
in computer design, which could rekindle entrepreneurship in the
industry.
Thinking Machines Corp. and Bolt Beranek and Newman Inc. each
revealed plans for a computer that uses parallel processors. In
the past, major innovations like this one have stimulated waves
of start-ups: The integrated circuit made the minicomputer
business possible; the microprocessor led to the personal
computer industry. `There's a chance for a few Digital
Equipments and Apple Computers to get started,' says Daniel
Hillis, co-founder of Thinking Machines.
Parallel processors are a radically new design. Computers have
always used one processor, which works on one piece of data at a
time, but at great speed. Parallel machines could be much faster
because their thousands of processors work on data
simultaneously.
Almost 60 universities are working on the computers, and a dozen
companies are now building or marketing them. At least six
models are expected in the next year. Most of the initial uses
will be in defense-related fields.
The market could blossom quickly. `Researchers will develop uses
for industry, and the community will come to understand that
parallelism isn't as complicated as the feared,' says Roy
Coppinger, of Intel Scientific Computers, an Intel Corp. unit
that is selling parallel machines. When that happens,
opportunities for software development will abound. `That's
where 90% of the work will go on,' says Hillis.
The new companies are looking over their shoulders at -- of
course -- IBM Corp., which is also developing a parallel
processor. `The start-ups are going to force the large
manufacturers to make changes,' says Sidney Fernbach, a computer
consultant in San Jose, Calif. `As soon as IBM decides to do it,
it will dominate the market.'
Where have we heard that before?
------------------------------
End of PARSYM Digest
********************
∂01-Aug-85 2246 morris@diablo accessing rules
Received: from SU-AIMVAX.ARPA by SU-AI.ARPA with TCP; 1 Aug 85 22:46:12 PDT
Date: 1 Aug 1985 2242-PDT (Thursday)
From: Kathy Morris <morris@diablo>
To: nail@diablo
Subject: accessing rules
Rules will be stored internally in the following format:
rule(File, LineNo, Head, [], Vars, ifRule).
rule(File, LineNo, Head, [SubGoal | ←], Vars, ifRule).
rule(File, LineNo, Head, Neg, Vars, ifNot).
rule(File, LineNo, Head, findall(←,←,←), Vars, ifFind).
(This is assuming that the current←line←number builtin predicate
of CProlog works correctly; otherwise I'll give the predicate name and
the rule number within that predicate.)
The linenumber of a generated predicate is inherited from the rule which
generated it.
They can be accessed using:
nailRule(Template, SubGoalList, Type)
or
getRule(Template, Rule, Vars)
The Template has the form h(←, ..., ←), and will be bound to the head of
the rule. Type is one of: ifRule (for ordinary rules), ifNot, or
ifFind (for findalls/bagofs). SubGoalList is the list of subgoals
in the rule.
getRule gives the rule in a (more or less) human-readable form; Vars is a list
of the variables appearing in the Rule along with their names (see
the read predicate in CProlog).
nailRule gives the rule in a better form for applying capture rules, etc.
There'll also be a predicate to print a rule matching a given template
(so you can print all the rules with a given predicate using backtracking).
I'm open to suggestions as to whether either/both of the predicates
accessing the rules should return the File and LineNo -- they'll
only be useful for debugging, though I need them for reporting errors
in the input.
The usual caveats about accessing internals apply -- you should only
use nailRule and getRule to access rules (anyone have better ideas for
what to call these predicates?).
Any other comments?
Kathy
∂02-Aug-85 0645 @SU-SUSHI.ARPA,@SU-SCORE.ARPA:udi@wisc-rsch.arpa Problem. $100 reward
Received: from SU-SUSHI.ARPA by SU-AI.ARPA with TCP; 2 Aug 85 06:45:27 PDT
Received: from SU-SCORE.ARPA by SU-SUSHI.ARPA with TCP; Fri 2 Aug 85 06:42:11-PDT
Received: from wisc-rsch.arpa by SU-SCORE.ARPA with TCP; Fri 2 Aug 85 06:40:34-PDT
Received: by wisc-rsch.arpa; Fri, 2 Aug 85 08:25:13 cdt
Received: from wisc-crys.arpa by wisc-rsch.arpa; Mon, 29 Jul 85 13:37:53 cdt
Received: from SU-SCORE.ARPA by wisc-crys.arpa; Mon, 29 Jul 85 13:37:45 cdt
Date: Mon 22 Jul 85 22:12:20-PDT
From: Gene Golub <GOLUB@SU-SCORE.ARPA>
Subject: Problem. $100 reward
To: udi@WISC-CRYS.ARPA
Message-Id: <12129197594.39.GOLUB@SU-SCORE.ARPA>
Sender: udi@wisc-rsch.arpa
Resent-From: udi@wisc-rsch.arpa (Udi Manber)
Resent-To: theory:;
The other day we had a seminar by Manfred Trummer of UBC ( soon MIT).
The following problem arose.
We are given 2n distinct elements c(i), i=1,...,2n.
Let b(i,j)= (c(i)-c(j))↑-1 when i not equal j, else b(i,j) = 0.
Let Y be a vector with 2n components.
TRUMMER PROBLEM
Give an algorithm which depends on the form of B for computing
BxY in less than O(n↑2) multiplications. If this is impossible,
show that it cann't be done.
Reward: $100.
Gene Golub
-------
∂02-Aug-85 1531 SAMI@SU-CSLI.ARPA tgif
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 2 Aug 85 15:31:05 PDT
Date: Fri, 2 Aug 1985 15:19 PDT
Message-ID: <SAMI.12132005945.BABYL@SU-CSLI.ARPA>
From: SAMI@SU-CSLI.ARPA
To: Marjorie Maxwell <MARJORIE@SU-CSLI.ARPA>
Cc: consultants@SU-CSLI.ARPA, folks@SU-CSLI.ARPA
Subject: tgif
In-reply-to: Msg of 26 Jul 1985 09:36-PDT from Marjorie Maxwell <MARJORIE>
At what time is the tgif?
∂02-Aug-85 1640 vardi@diablo Seminar on Parallelism in Logic Programs
Received: from SU-AIMVAX.ARPA by SU-AI.ARPA with TCP; 2 Aug 85 16:40:31 PDT
Date: Fri, 2 Aug 85 16:33:25 pdt
From: Moshe Vardi <vardi@diablo>
Subject: Seminar on Parallelism in Logic Programs
To: bboard@csli, bboard@sri-ai, bboard@sri-csl, nail@diablo, su-bboards@diablo
AND/OR PARALLELISM IN LOGIC PROGRAMS
by
Simon Kasif
Department of Computer Science
University of Maryland, College Park
MJH 352
Aug. 14, 1:00pm
ABSTRACT
The separation of logic and control in logic programs
has been shown to allow the programmer to write declara-
tively lucid programs whose execution is determined by the
interpreter. This appealing characteristic of logic program-
ming spurred research directed towards diversifying the
means for controlling the execution of logic programs. In
particular, parallelism in logic programs may be exploited
even when it is impossible to state a priori that two goals
may be executed concurrently, but such an opportunity may be
detected during the course of the execution.
This talk will address the problem of AND/OR parallel-
ism in logic programming. We describe a computational model
for AND/OR parallel execution of logic programs. The model
provides the primitives to describe and analyze parallel
interpreters, emphasizing the data-flow among conjunctive
goals. The effectiveness of our computational model is esta-
blished through its ability to support both old and new com-
munication paradigms for the parallel interpretation of
logic programs.
Several methods to implement AND/OR parallelism in
logic programs are investigated based on notation developed
in the model. The methods are shown to define a spectrum of
communication schemes, ranging from the set intersection
method where communication is eliminated altogether,
through methods based on producers-consumers, where communi-
cation is uni-directional and finally ending at a flexible
bi-directional scheme introduced in the paper, called the
Intelligent Channel.
The primitives that comprise the model are used to syn-
thesize two new parallel interpreters: Disjunctive System
(DS) interpreters and Intelligent Channel Interpreters. The
Intelligent Channel is a scheme we propose to constrain the
combinatorial explosion of active processes, and to elim-
inate the need to maintain a separate binding environment
for every active OR-branch.
∂04-Aug-85 1612 CLT Mini seminar series
To: "@DIS.DIS[1,CLT]"@SU-AI.ARPA, su-bboards@SU-AI.ARPA
Speaker: Susumu Hayashi, Kyoto University
Title: PX : a system extracting LISP programs from proofs
Time: Thursday, August 8, 4:00-5:00 pm
Place: Room 252 Margaret Jacks Hall
(Computer Science Department), Stanford
Abstract
PX is a computer implementation of a constructive theory of pure LISP
based on Feferman's T0. I will show how to represent programs of pure LISP
as proofs in PX.
∂04-Aug-85 1626 @SU-CSLI.ARPA:CLT@SU-AI.ARPA Mini seminar series
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 4 Aug 85 16:25:52 PDT
Received: from SU-AI.ARPA by SU-CSLI.ARPA with TCP; Sun 4 Aug 85 16:20:50-PDT
Date: 04 Aug 85 1612 PDT
From: Carolyn Talcott <CLT@SU-AI.ARPA>
Subject: Mini seminar series
To: "@DIS.DIS[1,CLT]"@SU-AI.ARPA, su-bboards@SU-AI.ARPA
Speaker: Susumu Hayashi, Kyoto University
Title: PX : a system extracting LISP programs from proofs
Time: Thursday, August 8, 4:00-5:00 pm
Place: Room 252 Margaret Jacks Hall
(Computer Science Department), Stanford
Abstract
PX is a computer implementation of a constructive theory of pure LISP
based on Feferman's T0. I will show how to represent programs of pure LISP
as proofs in PX.
∂05-Aug-85 1159 EMMA@SU-CSLI.ARPA Thought for the Day
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 5 Aug 85 11:58:52 PDT
Date: Mon 5 Aug 85 11:52:02-PDT
From: Anonymous
Subject: Thought for the Day
Sender: EMMA@SU-CSLI.ARPA
To: folks@SU-CSLI.ARPA
Reply-To: Request
Tel: 497-3479
Man knows that there are in the soul tints more bewildering, more
numberless, and more nameless than the colours of an autumn forest...
Yet he seriously believes that these things can every one of them, in
all their tones and semi-tones, in all their blends and unions, be
accurately represented by an arbitrary system of grunts and squeals.
He believes that an ordinary civilized stockbroker can produce out of
his own inside noises which denote all the mysteries of memory and all
the agonies of desire.
G. K. Chesterton, "G. F. Watts", p. 55
-------
∂05-Aug-85 1221 YAMARONE@SU-CSLI.ARPA EXTRA SANDWICHES
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 5 Aug 85 12:21:05 PDT
Date: Mon 5 Aug 85 12:13:33-PDT
From: Tom Yamarone <YAMARONE@SU-CSLI.ARPA>
Subject: EXTRA SANDWICHES
To: folks@SU-CSLI.ARPA
TWO:
TURKEY ON LIGHT RYE
AVO + JACK ON SLICED SOURDOUGH
BOTH 2.50
FIRST COME, FIRST SERVED.......
THE VENTURA SANDWICH , CORP.
-------
∂05-Aug-85 1704 @SU-CSLI.ARPA:FRD@SU-AI.ARPA P-n-P Wed Aug 7: Diagram Understanding
Received: from SU-CSLI.ARPA by SU-AI.ARPA with TCP; 5 Aug 85 17:04:49 PDT
Received: from SU-AI.ARPA by SU-CSLI.ARPA with TCP; Mon 5 Aug 85 16:53:13-PDT
Date: 05 Aug 85 1655 PDT
From: Fred Lakin <FRD@SU-AI.ARPA>
Subject: P-n-P Wed Aug 7: Diagram Understanding
To: folks@SU-CSLI.ARPA
∂01-Aug-85 1811 PENTLAND@SRI-AI.ARPA Pixels and Predicates
Received: from SRI-AI.ARPA by SU-AI.ARPA with TCP; 1 Aug 85 18:11:01 PDT
Date: Thu 1 Aug 85 18:10:57-PDT
From: PENTLAND@SRI-AI.ARPA
Subject: Pixels and Predicates
To: P-n-P: ;
PIXELS AND PREDICATES, Area P1, Wednesday Aug 7, 3:00pm,
Ventura Hall, Stanford Campus
DIAGRAM UNDERSTANDING:
THE INTERSECTION OF COMPUTER GRAPHICS AND COMPUTER VISION
Fanya S. Montalvo
MIT, Artificial Intelligence Laboratory
ABSTRACT
A problem common to Computer Vision and Computer Graphics is identified. It
deals with the representation, acquisition, and validation of symbolic
descriptions for visual properties. The utility of treating this area as one
is explained in terms of providing the facility for diagrammatic conversations
with systems. I call this area "Diagram Understanding", which is analogous to
Natural Language Understanding. The recognition and generation of visual
objects are two sides of the same symbolic coin. A paradigm for the discovery
of higher-level visual properties is introduced, and its application to
Computer Vision and Computer Graphics described. The notion of denotation is
introduced in this context. It is the map between linguistic symbols and
visual properties. A method is outlined for associating symbolic descriptions
with visual properties in such a way that human subjects can be brought into
the loop in order to validate (or specify) the denotation map. Secondly, a way
of discovering a natural set of visual primitives is introduced.
-------
∂05-Aug-85 1924 RESTIVO@SU-SCORE.ARPA PROLOG Digest V3 #35
Received: from SU-SCORE.ARPA by SU-AI.ARPA with TCP; 5 Aug 85 19:22:46 PDT
Date: Monday, August 5, 1985 6:27PM
From: Chuck Restivo (The Moderator) <PROLOG-REQUEST@SU-SCORE.ARPA>
Reply-to: PROLOG@SU-SCORE.ARPA
US-Mail: P.O. Box 4584, Stanford CA 94305
Phone: (415) 326-5550
Subject: PROLOG Digest V3 #35
To: PROLOG@SU-SCORE.ARPA
PROLOG Digest Monday, 5 Aug 1985 Volume 3 : Issue 35
Today's Topics:
Editorial - Digest Advertisements,
Announcements - Call for papers & Flat Concurrent Prolog
----------------------------------------------------------------------
Date: Thu 1 Aug 85 20:21:26-PDT
From: Pereira@SRI-AI
Subject: Advertisements on the Prolog digest
I was disturbed to see what amounts to an advertisement to
POPLOG in the latest Prolog Digest. This would be acceptable
if POPLOG were a genuine academic system available to all
just for a handling fee, but this is only the case for UK
academic institutions (as the ad indicates near the end).
Others must acquire POPLOG from a commercial distributor
who sells it on behalf of Sussex University.
I think this use of the digest is unethical, infringes the
ARPANET rules and is unfair to other Prolog suppliers that
cannot or will not avail themselves of such a ``convenient''
form of free advertisement.
I leave you to reach your own conclusions about a software
supplier that uses such means to advertise its products.
-- Fernando Pereira
------------------------------
Date: Mon, 29 Jul 85 19:04:32 -0200
From: Ehud Shapiro <Udi%wisdom.bitnet@WISCVM>
Subject: Call for papers
CALL FOR PAPERS
Third International Conference on Logic Programming
Imperial College of Science and Technology, London, UK
July 14-18, 1986
In cooperation with:
Association for Computing Machinery
British Computer Society
IEEE Computer Society
Japan Society for Software Science and Technology
The conference will consider all aspects of logic programming,
including, but not limited to:
Theory and foundations
Architectures and Implementations
Methodology
Programming Languages and Environments
Applications
Relations to other computation models, programming
languages, and programming methodologies.
Of special interest are papers related to parallel processing,
papers discussing novel applications and applications that
address the unique character of logic programming, and papers
which constitute a contribution to computer science at large.
Papers can be submitted under two categories, short --
up to 2000 words, and long -- up to 6000 words. Submissions
will be considered on the basis of appropriateness, clarity,
originality, significance, and overall quality.
Authors should send eight copies of their manuscript, plus
an extra copy of the abstract, to:
Ehud Shapiro
ICLP Program Chairman
The Weizmann Institute of Science
Rehovot 76100, Israel.
Deadline for submission of papers is December 1, 1985.
Authors will be notified of acceptance or rejection
by February 28, 1986. Camera ready copies are due April
1st, 1986.
General Chairman
Keith Clark
Imperial College of Science and Technology
180 Queen's Gate
London SW7 2BZ, United Kingdom
Local Arrangements and Exhibition Chairman
Richard Ennals
Imperial College of Science and Technology
180 Queen's Gate
London SW7 2BZ, United Kingdom
Program Committee
Martin van Caneghem, University of Marseille, France
Veronica Dahl, Simon Fraser University, Canada
Maarten van Emden, University of Waterloo, Canada
Kazuhiro Fuchi, ICOT, Japan
Koichi Furukawa, ICOT, Japan
Ake Hanssen, Uppsala University, Sweden
Kenneth M. Kahn, Xerox PARC, USA
Peter Koves, Logicware Inc., Canada
Giorgio Levi, University of Pisa, Italy
John Lloyd, University of Melbourne, Australia
Frank G. McCabe, Imperial College, UK
Jack Minker, Maryland University, USA
Fernando Pereira, SRI International, USA
Luis M. Pereira, University of Lisbon, Portugal
Antonio Porto, University of Lisbon, Portugal
Ehud Shapiro, Chairman, Weizmann Institute, Israel
------------------------------
Date: Fri, 26 Jul 85 15:25:56 -0200
From: Ehud Shapiro <Udi%wisdom.bitnet@WISCVM>
Subject: Beta test-sites for our Flat Concurrent Prolog
We will be ready in a month or so to release our Flat
Concurrent Prolog system to beta test sites (FCP is the
And-parallel subset of Concurrent Prolog). The system
consists of an emulator for a Warren-style FCP abstract
machine, written in C, which also includes kernels and a
garbage collector; an FCP compiler, written entirely in
FCP (including the tokenizer, parser, precompiler, encoder,
and assembler); and a basic interactive programming
environmnt that includes a shell, I/O routines, and a
source-level debugger, which are also written in FCP.
The system has an interesting module system, which implements
remote procedure calls via message-passing between module
manager processes. It supports separate compilation and
runtime linking.
The system runs on VAX and on the Sun under Berkeley
Unix 4.2. It runs at about the third of the speed of
Quintus Prolog on the Vax for similar programs, on a
wide range of examples.
Some statistics: compiling the main module of the compiler,
the encoder, which is 8937 bytes and 418 lines of code
long, takes 258 CPU seconds on the VAX/750. The resulting
binary file is 17560 bytes of code long (at present we use
word-encoding, rather then byte encoding for the abstract
machine instructions). The compilation consumes about 1.5
Mbytes of heap memory (without garbage collection).
During this computation, 29000 processes are generated (yes,
twenty nine thousand), and altogether they perform 92000
reductions. During the computation 13000 process suspensions
occur. This gives an effective rate of 350 process reductions
per second, and an avarage of 3 reductions per process.
The system's C code is currently 5270 lines of code long, 2942
for the emulator, 1624 for kernels, and 704 for the garbage
collector. Its FCP code is currently 4529 lines of code long,
2060 for the compiler, 324 for the debugger, 722 for I/O, and
1423 for the rest of the system. Of course the interactive shell
and the compiler share the tokenizer and parser.
The main designers and implementors of the system are Avshalom
Houri and myself. Other contributors include Bill Silverman,
Michael Hirsch, Jim Crammond, Colin Mierowski, Steve Taylor,
Muli Safra, Nir Friedman, and Shimon Cohen. The development of
the system was supported by IBM Poughkeepsie, Data Systems
Division.
The system is still under development. The major avenues of
improvement being investigated at present are extending the module
system to be hierarchical, and to integrate better the debugging
and module concepts; integrating the system with a partial
evaluator; improving the performance by optimizing the emulator
and improving the instruction set (we do not plan at present
rewriting the emulator in assembler or in micro-code); adding
a window system; an independent file-system; and other gadgets.
Longer term research includes full compilation and implementation
on a multiprocessor.
We plan to distribute the system following standard university
basedsoftware distribution practices. Before releasing the
system for the general public, we would like to obtain some
feedback, and improve the system some more. We would like to
deliver it to some selected groups with strong logic programming
or concurrent programming background, who are interested in one
or more of the following:
1. Improve and extend the system in various ways.
2. Use it as a research tool for developing pilot FCP
applications.
3. Use it to teach a course in concurrent logic programming.
(we beleive it is reliable enough even at present for this.
It has just begun to be used for doing course projects in
my Concurrent Prolog Programming course at the Hebrew
University and at the Weizmann Institute, so we will know
better in a few weeks).
Interested parties should contact me, with relevant information,
at:
udi%wisdom.bitnet@WISCVM.arpa
or
...!decvax!humus!wisdom!udi
We hope to sort out the details of the distribution mechanism
by mid-August.
-- Ehud Shapiro
------------------------------
End of PROLOG Digest
********************