Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Chess Programming: how to identify ending phase?

Author: Normand M. Blais

Date: 13:17:55 04/21/01

Go up one level in this thread


On April 21, 2001 at 09:31:00, Robert Hyatt wrote:

>On April 20, 2001 at 15:49:33, Normand M. Blais wrote:
>
>>Hi,
>>
>>I'm using the following:
>>
>>#define ENDING \
>>((pawn_mat[side]+pawn_mat[xside] <= 1200) && \
>>(piece_num[side]+piece_num[xside] <= 5))
>>
>>
>>Assuming a value of 1000 for a pawn and that the kings or the pawns are not
>>counted as a pieces for "piece_num", is the expression above correct?
>>
>>Thank you for your help.
>>
>>NMB
>

I'll follow your advice.

Many thanks.

NMB

>
>You _really_ don't want to do this in an evaluation.  It will give you a
>serious "boundary effect" where the eval has significant changes around some
>boundary in the game.  In this case, around the point where you decide whether
>it is still a middlegame or an endgame.  You want a _smooth_ transition from
>MG to EG, not an abrupt change.  The program will search around that boundary
>and produce some very bizarre results...
>
>I simply phase MG eval terms out as material comes off, and phase EG eval
>terms in at the same time.  But it is gradual rather than an abrupt change
>around some magic material boundary...



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.