Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Zugzwang in a Null Move Search?

Author: Robert Hyatt

Date: 16:05:30 12/24/98

Go up one level in this thread


On December 24, 1998 at 11:39:23, KarinsDad wrote:

>A sudden thought occured.
>
>How do you handle zugzwang in a null move search?
>
>I could see how it would be a problem:
>
>Player A = computer
>Player B = opponent
>
>A is examining a move Ply x.
>B can counter with Ply x+1.
>B is given a null move; effectively, but not really Ply x+2.
>Every response by A at Ply x+3 results in the Ply x+2 move having a lower score
>than Ply x+1, hence we ran into zugzwang.
>
>However, how does one handle a more complex case such that zugzwang is not
>detected until Ply x+4 or Ply x+6? Zugzwang is often associated with only a few
>number of moves on the board (although that is not always the case). If the
>zugzwang would not be detected until Ply x+4, it seems that this would defeat
>the entire purpose of using a null move search (i.e. pruning the tree of moves
>that would be bad anyway).
>
>And how does this work when you do not run into zugzwang? Do you check for
>zugzwang by ignoring the results at Ply x+2 where that move results in either a
>piece capture or a piece promotion (and hence the score will jump materially
>towards the move being good as opposed to bad) and only look at the other moves?
>
>This is somewhat unclear and confusing. Can anyone explain it in a clear manner?
>
>Thanks,
>
>KarinsDad

You don't, because that is exactly what breaks the null-move search in the
first place.  Most turn nulls off when material is reduced to the point where
zug can happen.  Nobody uses it in pawn-only endings that I know of since zug
is the most common theme there...



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.