Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: alpha-beta is silly?

Author: Roberto Waldteufel

Date: 06:35:48 06/02/98

Go up one level in this thread





>OK. The Problem is the quiet search. I accept.
>Has anybody tried to learn the quiet search non qiet positions, and is
>the slow down worth?
>
>Werner


One thing that might be worth trying is to include promoting moves of
pawns, possibly even moves of a pawn to the seventh rank as well. I know
this has been implemented in some programs (eg Cray Blitz did this I
believe - probably also Crafty, but you'll have to ask Bob). I once
tried allowing the quiescence search to look at all moves of
"unstoppable" passed pawns, but this proved far too expensive to be
acceptable. It would solve your running pawn problem, though. I think it
might be possible provided that it was only done in a very small
proportion of positions - the trick is how to decide when "passed pawn"
quiescence is appropriate. Just like "check quiescence", this will blow
your search sky high if it is not accompanied by very strict conditions
for its applicaton. Maybe something could be done along the lines of a
smart preprocessing routine to try and determine if running pawns were
likely to be an issue within the search depth about to be done, but this
would be difficult to catch all cases. You could hope to weed out some
of them, though. Just an idea to kick around, I have not implemented
this myself, but it is on my "things to do" list.

Roberto



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.