Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Mate in 30

Author: Michael Neish

Date: 06:50:27 09/08/04

Go up one level in this thread



>I did not read the source code of gnuchess but no chess program does it
>unless the programmer does not care about strength of the program.
>
>Extending checks and one reply by one ply everywhere is going to cause the
>search to explode and I will not be surprised if the program cannot get deep
>because of extending useless lines.
>
>Uri

Thanks.  Well I'm trying to figure out what's going on.  As I said, Gnuchess is
going deep enough -- I tested it -- so for some reason it's failing to go deep
enough along the line that counts.

Well I don't really know, but let's say that at a crucial point in the line
Gnuchess can either play the move that eventually leads to mate or choose
another one that wins Black's Queen several moves later.  Let's say that both
moves at this crucial point are non-captures.  If the move it tries first is the
one that wins the Queen, the node may fail high there and then, and return,
missing the mate altogether (although I suppose it depends on the values of
alpha and beta at that time).  Does this sound plausible to you?

If so, what are the tricks to avoid missing mate in favour of material gain?

Cheers,

Mike.




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.