Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: A test position for chess programs

Author: John Merlino

Date: 10:52:22 12/29/00

Go up one level in this thread


On December 29, 2000 at 12:00:41, Uri Blass wrote:

>This position is similiar to the position that crafty had no problem to solve
>
>[D]R7/8/6k1/4n3/5p2/5P2/5r1p/7K b - - 0 1
>
>The test is to avoid Nxf3 that is leading to a draw.
>Another test is to see the draw after Nxf3.
>
>I am interested to know if there is a program that can see the draw after Nxf3
>by search.
>I am not interested in the result of littlegoliath because I know that this
>program can evaluate as draws positions that are not draws because it assumes
>that there is a perpetual check if it cannot prove that there is no perpetual
>check.
>
>
>
>Did somebody try to do a program not for playing games but only to find if there
>is a perpetual check?
>The program can use selective search for white when it considers only checks for
>white and replies for check for black.
>
>It needs to evaluate only positions when white cannot do a check or positions
>when there is a repetition.
>
>It can evaluate repetitions as a draw and stalemates as a draw.
>
>I believe that a program like this can prove that there is a perpetual check
>because the number of relevant positions that it needs is small enough to search
>to the end of the checks.
>
>Did somebody try to build this program and how much time does it need to find
>the pereptual check in the position after Nxf3?
>
>Uri

Chessmaster 8000, default values, sees Rxf3 immediately and sticks with it, with
an eval of 5.64.

However, it does not spot that Nxf3 is a draw after 6 minutes of searching.
Although the PV has at least 15 consecutive checks by the rook, the eval is
still -4.73.

jm



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.