Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: SEE for forward pruning in Q. Search

Author: Robert Hyatt

Date: 17:00:49 08/04/99

Go up one level in this thread


On August 04, 1999 at 16:32:10, Tom King wrote:

>Hi all,
>
>What are the pro's and con's of using a static exchange evaluator (SEE)
>for pruning captures in the Q. search? My program currently *doesn't*
>use a SEE, but if I remember right, some programs like Crafty and Ferret
>use a SEE to prune "losing" captures, i.e they examine a swap sequence on
>a particular square, and if it appears to lose material, they prune that
>capture right out.
>
>Who else is using a SEE in this way? I was playing around with this kind
>of thing at the weekend, and I found that pruning these "losing" captures
>is not without risk - some test problems are solved a ply later than normal
>(for my program, anyhow, and this could be a bug in my SEE). On the other
>hand, search speed is increased, sometimes quite dramatically.
>
>Opinions? Is pruning SEE losers in the Q. search a win?


I find the following:

using SEE to order captures in the q-search, without eliminating any, will
shrink the tree about 10% over using something simple like MVV/LVA.  But the
SEE code will likely cost you more than 10% (unless you are a bitmap program
where this can be done fairly efficiently).

using SEE to eliminate losing captures can speed you up another 50%, or a factor
of two, which is very significant.  And no matter how slow your SEE code is,
that become a 'winner' of an idea.



This page took 0.04 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.