Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: When to stop searching captures?

Author: Robert Hyatt

Date: 11:25:47 09/17/98

Go up one level in this thread


On September 17, 1998 at 14:16:48, John Coffey wrote:

>Let me see if I understand this correctly ...
>
>1.  Your basic search will search full width to a certain level.
>
>2.  Your "basic" search includes extensions for "active" positions.  What
>type of criteria do you use for "active" positions?  (This includes check,
>right?)

yes, checks, positions where one side is in check and has only one legal move
at its disposal, pushing an advanced passed pawn, recaptures (similar in idea
to Amir's 1/2 ply for a capture, but more restrictive).


>
>3.  Your "qsearch" looks at captures only beyond that point?  Do you ignore
>any captures, or do you just look at all captures to "infinity?"


I use a static exchange evaluator to toss out captures that appear to lose
material (QxP where the pawn is defended more times than it is attacked, for
example).  I also notice the value of the piece I am about to capture, and if
that isn't enough to bring the score back up to within alpha/beta, I also don't
try the capture (ie if I am a queen down, capturing a pawn is not going to help
at all).

But within that framework, I look at captures as deeply as I can follow them,
which *could* be 30 plies although it is nearly impossible to construct a
position where that might happen...






>
>Thanks in advance,
>
>John Coffey



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.