perm filename ACCOMP[W76,JMC] blob
sn#204200 filedate 1976-02-27 generic text, type C, neo UTF8
COMMENT ā VALID 00002 PAGES
C REC PAGE DESCRIPTION
C00001 00001
C00002 00002
C00008 ENDMK
Cā;
ACCOMPLISHMENTS IN FORMAL REASONING
1. Christopher Goad, a student research assistant, has succeeded in
modifying McCarthy's knowledge formalism so that conclusions to the
effect that an individual does not know something can be drawn. In
all previous knowledge formalisms, this has been possible only in
trivial cases, and it is fundamental in making machines reason about
knowledge, because they have to know when their present knowledge is
insufficient to answer a question so they can search for more.
2. John McCarthy has identified a class of problems called Boolean
search problems that admit very fast parallel solution algorithms and
has shown how many of the quasi-static problems encountered in AI can
be transformed into Boolean search form.
3. Zohar Manna has shown how recursive programs often admit more
defined fixed points than the minimal fixed points that have
previously been studied. He calls them optimal fixed point and has
determined many of their properties.
4. The algorithm for proving sentences in monadic predicate calculus
has been debugged and has been incorporated into FOL.
5. Many user oriented improvements to the FOL program have been made
during this period. They include several new quantifier manipulation
commands, new printing routines, machine language changes effecting
overall efficiency, updated versions of the tautology and tauteq
routines. In addition CMU has requested a version of FOL and we are
bringing it up there. This work is being done by Richard Weyhrauch
and Bill Glassmire.
6. Zohar Manna has explored the applicability of a technique of
"intermittent assertions" due to Burstall for verifying programs, has
compared it with other techniques on a number of problems, and has
demonstrated that it is potentially the most practical technique.
GOALS FOR JULY TO OCTOBER 1976
1. A formalism for describing sources of information will be
developed and checked by writing a program that will find telephone
numbers form the telephone number files at a number of computers on
the ARPA net. No changes in the existing files will be required.
This will be a step towards making the computer files all over the
country accessible in a uniform way without requiring that they be
redesigned.
2. Pattern matching has mostly involved matching a pattern to a
symbolic expression. Dave Wilkins is working on matching patterns to
non-symbolic situations, e.g. situations in the real world.
3. The McCarthy-Painter compiler will be proved correct in FOL.
(This has taken longer than was expected).
4. A good way of filing already proved theorems for future use will
be implemented in FOL.
5. There are several projects for building needed features of FOL.
They include: the ability to handle more than one proof at a time, a
facility for stating and checking goals and subgoals, and a new
version of the FOL machinery for using direct observation of the real
world in making deductions
6. Zohar Manna will investigate applying the "intermittent
assertions" technique to the automatic synthesis of programs meeting
given specifications. He is also investigating the class of all
fixed points of recursive programs and looking for practical
applications of his optimal fixed point approach.