COMMENT ⊗   VALID 00002 PAGES
C REC  PAGE   DESCRIPTION
C00001 00001
C00002 00002	ideas[e86,jmc]
C00004 ENDMK
C⊗;
ideas[e86,jmc]

1986 Aug 12

Sterility and partial evaluation

	Shmuel Safra and Ehud Shapiro have a paper "Meta Interpreters for
Real" which Shapiro gave me at the Logic Programming meeting at
Imperial College.  It is a strong advertisement for partial
evaluation.  What is the relation of that to "strength reduction"?
The question for now is whether there is a relation between
partial evaluation and general reasoning of the kind used in the
sterile container.

Circumscription vs. logic prgrams

	In certain cases, circumscription and logic programs give
the same result, and logic programs give it computationally.  However,
a circumscriptive theory consists of sentences with the usual meanings
of the connectives.  The preferred models of the theory are determined
by the sentences, and sometimes by a circumscription policy.  In
logic programs, as pointed out in Apt, Blair and Walker "Towards a
Theory of Declarative Knowledge", logic programs with negation, even
if properly stratified don't admit all minimal models.  Their example
is 

p ← ¬q

which is logically just  p ∨ q  and therefore symmetric.  However,
the model {p,¬q} is acceptable and the model {¬p,q}  is not.