Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: 8/8/8/6Np/2N5/Pk6/6K1/8 w - a3

Author: Robert Hyatt

Date: 18:29:45 02/12/03

Go up one level in this thread


On February 12, 2003 at 15:37:54, Dieter Buerssner wrote:

>On February 12, 2003 at 11:55:12, Robert Hyatt wrote:
>
>>>8/8/8/6Np/2N5/Pk6/6K1/8 w - a3
>>>
>>To me the a4 idea was pretty simple.  a4.  Ka4 is a lost KNN vs KP ending.  [...]
>>So
>>a4 Kxa4 is the
>>only hope and that is lost in a ridiculous number of moves.  But to the
>>computer, a mate is
>>a mate.  :)
>
>As we previously discussed, the computer might regret this choice. It is not
>clear yet, that he will be able to mate the opponent (within chess rules) after
>a4 Kxa4. Actually, I doubt, that any currently available chess programs will win
>this after a4 Kxa4 vs. good opposition (many possible opponents - my engine with
>default setup included - will resign wrongly. But in a "serious" tournament, I
>would set it up, to not resign ...)
>
>Regards,
>Dieter


The PV that Crafty (and Yace) give needs to include two things.

1.  optimal play for the shortest mate.

2.  sub-optimal play that tries to challenge the authenticity of the mate
via the 50-move rule.  IE the losing side tries to avoid pawn pushes, or
captures, and tries to prevent the winning side from doing the same, to bring
the 50 move rule into the picture.

we are all doing 1 right now, which is OK.  And that means that for the
given position, I believe Crafty and Yace would end up drawing if I counted
the pawn pushes correctly.  Doing 2 is a lot of work, obviously, and impractical
in the search.  And once you are in the ending, it is too late as you are
already committed to that ending, whether it is really winnable or not.

We get incorrect data from the tables, based on the rules of chess we have to
use.  That's a pain.



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.