perm filename CONV[1,LMM] blob sn#031684
filedate 1973-03-27 generic text, type C, neo UTF8
COMMENT ⊗ VALID 00002 PAGES
C REC PAGE DESCRIPTION
C00002 00002 (... I would like to discuss some things related ...)
(... I would like to discuss some things related ...)
TO CYCGEN.. BUT DO NOT FEEL SATISFIED
WITH WHAT I CAN DO WITH BOB OR BRUCE -- WHILE I FEEL MORE OF A RAPPORT
WITH YOU WHILE DISCUSSING (OWING TO MY TASTE FOR SOMEWHAT RIGROUS AND
FORMAL WAYS OF TREATING IDEAS). I AM LOOKING FORWARD TO GETTING
SOME KIND OF FREQUENT INTERACTION WITH YYOU. CARE TO COMMENT?
; I'VE BEEN ON A STRANGE SCHEDULE LATELY-- UP AT 8PM, START WORK AT
; MIDNIGHT. IF YOU'LL BE IN TOMORROW MORNING (BEFORE NOON) I'LL
; BE AROUND THE TRAILER THEN. (COMMENT ON AN IMMEDIATE SCALE)
I HAVE AN APPT AT 10 AM AND PERHAPS I WILL COMEIN BUT CANNOT REALLY TELL.
; BGB SUGGESTED (OR MAYBE I JUST THOUGHT) IT WOULD BE A GOOD IDEA FOR
; YOU, ME AND JL TO GET TOGETHER, RE C10H10 AND RELATED PROBLEMS
I AGRREE, HOWEVER WOULD LIKE TO IVE PRIO TO THE BIG.MESS PAPER COMMENTS
AND LABELLING PAPER FROM JL. --
I AM GETTING THE FEELING THAT THE EXPANSION OF THE CATALOG (AND/OR)
THE LAYING IN OF DOUBLE BONDS ON CARBOCYCLES IS A DISTRACTION FROM
MORE IMMEDIATE AND DOABLE AND (DO-WORTHY) PROBLEMS LIKE --
GETTING A SIMPLE GOODLIST/BADLIST AND INTEGRATION OF THE GRAPH-MATCH
AND BRINGING TOGETHER OF SEVERAL CONSTRAINS ALREADY IMPLEMENTED ,
LIKE ACAYCLICSONLY ONLYONESUPEATOM PURERINGS ETC. SHOUDL GET DONE.
IF THEY GET OUT OF THE WAY -- I COULD PURGE MIND AND TRY TO GIVE A
FRESH APPROACH TO THE DEEPER PROBLEMS OF MONITOR AND DB AND AROMATICITY
well, the thing about the C10H10 is that it points up an impaortant
deficiency in the theory/ and thus, perhaps might really need to
be worked out before BIIG.MESS is really publishable. (i.e., well, if
it really can't do a lot of interesting problems ...
THERE ARE TWO WAYS OF LOOKING AT THE PROBLEM OF C10H10
ONE IS TO LOOK UPON IT AS POINTING TO A GREAT DEFICIENCY IN THEORY
THE OTHER IS TO SIMPLY CONSIDER THAT THE CATALOG IS VERY INCOMPLETE
EVEN FOR EVEN MODERATE PROBLEMS
HOWEVER THE PHILOSOPHY OF PRESENT BIG.MESS IS FINE AND CAN AND WILL
WORK FOR SEVERAL SMALLER PROBLEMS OF INTEREST -- LIKE THE URINE ANALYSIS
the point is that the CATALOG is NOT simply computable by
J Lederburgs algorithm
SEVERAL PROBLEMS OF INTEREST LIKE THE URINE ANAL AND JAN SIMEK TYPE OF
PROBLEMS TOO. SO AS FAR AS THAT GOES THE THEORY IS CORRECT AND WORKS.
BUT IT HAS (NOT A DEFIECIENCY) DRAWBACK THAT THE CATALOG NEEDS ALL
KINDS OF WEIRD LARGE STRUCS THAT ARE HARD TO COMPUTE SYSTEMATICLY,
WITH OUT PRESENT KNOWLEDGE. BUT IN OUR DISCUSSIONS ON THAT DAY AND IN
MY OWN THINKING I BELIEVE THAT WE ARE BEGINNING TO
GET SOMEWHERE. AND THE PAPER IS WORTH PRESENTING IN ITSELF.
it doesn`t seem complete, however , without some sort of CATALOG generator.
POSSIBLY, IT ONLY RREPRESENTS A VIABLE TRANSFORMER FROM CATALOG TO
STRUCTURES, AND IS COMPLETE ONLY TO TTHE EXTENT THE CATALOG IS COMPLETE.
well, enough, I am ready to concede already.
WOULD LIKE TO HEAR YOUR COMMENTS (REPLY IF YOU PREFER) ABOUT GETTING
SOME OF THE DIRTY WORK OVER WITH SO THAT YOU AND I WILL FEEL JUSTIFIED
SPENDING TIME (RESEARCHING) TALKING ABOUT THE BIGGER ISSUES.
as I see it, tasks are:
(0) this NIH demo
(1) goodlist , with simple embedding of superatoms into larger rings
(simple in that it will only work in some cases, specifically when
there is only one of a given superatom, and problems about
symmetry induced by part of the real struc being the same as
the superatom ignored.
(2) Badlist, trying to apply the match inside, rather than just
as a final filter
(3) other constraints, like ONLYRINGS , getting a monitor
to know what function to call when itIs given that kind
(4) laying in double bonds, dots, handleing aromaticity in
a smart way
(5) a general purpose CATALOG generator
(6) (just thought I might add this one to the list) A smart
"estimator" that has knowledge of how to estimate how
many structures will come out of a given generation.
can you think of any more?
AFTER (1) THRU (3) I WOULD ADD "DOCUMENTATION". FOR IF, AGAIN,
WE ENTER INTO THE REASEARCH MODE ON THE PROBLEMS OF (4) THRU (6)
THE DOCUMENTATION OF WHAT WE ALREADY HAVE WILL TEND TO GET BY PASSED
AND THAT WILL BE LOUSY TO HAVE TO GET BACK LATER TO DO THAT-- IT IS
ALREADY AT THAT STAGE OF BEING A LOUSY TASK.
well, the sooner the better. I think an argument can be made for putting
off even (1)-(3) for another weeks or four. I might suggest that
we both work on documentation now. Do you think you can do some
more of the files other that STRUCTURE? If so, let us discuss
WE HAVE JUST ABOUT ONE WHOLE MONTH FOR THE "NATO" DEMO -- THE SECOND
MONTH BEING SET ASIDE FOR TRANSFER FROM HERE OR CAMPUS TO NIH.
AND THE GENERATOR (AS IT IS) WITHOUT CONSTRAINTS WILL SURELY NOT FORM
ANY KIND OF DEMONSTRATION. AT THE SAME TIME DOCUMENTATION (AT LEAST AT
THE USER LEVEL) IS NEEDED FOR THE DEMO. HOW BAOUT IF TOOK THEM IN THAT
what in which order
THE ORDER : SOME CONSTRAINGS (1) THRU (3) : USER LEVEL DOCUMENTATION
: DOCUMENTATION AT TOP-LEVEL
of (1)-(3), (3) has rolled off my screen and I forget what
it was; I'd like to work on (1), and if you would like to work on (2)
... since you did some of the graph matching things already
(3) WAS UNIFYING THE CONSTRAINTS INTO A BUNCH OF FLAGS AT THE USER
WE SHOULD DISCUSS HOW WE WANT TO DO THAT
If you and I can get together a documentation of the constraints
that we have I think we can see things more clearly-- Why don't
I work up a write-up of things like:
If you want Rings only, call RINGS
If you have a skeleton and want to put radicals on it, call PERMARDS,
... i.e., in terms of what one wants to do --->> what to call
as well as (instead of) this function does that.
IF YOU WANT ONLY ONE SUPERATOMS CALL ? ... IF YOU WANT ACYCLICS CALL?
EVEN I DO NOT KNOW .
call ME!! 854-69600
AS I SAID EARLIER I HAVE A 1000 APPT. I BETTER SIGN OFF. YOU LINKED IN
JUST AS I WAS ABOUT TO CLOSE SHOP; WILL YOU PLEASE SPOOL THE TWO FILES
THAT I DID FTP.
i saw KETONE-...
ok, will do
SEE YOU TOMORROW IF PROVIDENCE HAS IT SO!
and if you get up
I BETTER. BYE FOR NOW.