Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Method of Analogies??

Author: Steven J. Edwards

Date: 14:23:42 05/29/98

Go up one level in this thread


On May 29, 1998 at 15:03:04, Bruce Cleaver wrote:

>Whatever happened to the old Russian idea of 'Method of Analogies'
>whereby foolish moves would not be searched until conditions that
>refuted them changed?  This idea was described in David Levy's book "How
>Computers Play Chess".

The Russian idea is essentially the same as what others call a
"causality facility".  See Hans Berliner's PhD dissertation for one
example, and David Wilkins' Paradise implementation for antoher.

The idea is this: given a node N in the search with available moves M1
and M2,
a causal analysis A1 generated from the search of M1 may eliminate the
need to
search M2 if a function CausalThreshold(N, M1, M2, A1) is less than some
limiting value.

As you might imagine, there is some guesswork involved here as otherwise
the causal analysis process would take more resources than the search of
M2.

I have done some work with this.  My thought is that is can be useful
for eliminating some simple piece exchange tactical quiescence searches.
 It can be implemented using bitboards that keep track of which pieces
moved or exchanged
along with bitboards of trajectory squares covering sweep piece
pathways.

-- Steven (sje@mv.mv.com)



This page took 0 seconds to execute

Last modified: Thu, 15 Apr 21 08:11:13 -0700

Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.