Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Junior-Crafty hardware user experiment - 19th and final game

Author: Tord Romstad

Date: 02:42:54 12/27/03

Go up one level in this thread


On December 26, 2003 at 18:12:26, Dieter Buerssner wrote:

>On December 26, 2003 at 17:44:36, Tord Romstad wrote:
>
>>The same problem occurs, of course, in all cases when you do some search
>>decision
>>which depends on the path leading to the position.  I'd really like to know how
>>other
>>programmers handle this problem.  At the moment, it is my single most difficult
>>problem in chess programming.
>
>I agree fully with your description of the problem. I cannot think of a good
>solution, besides ignoring it. In the case of extensions, this might cost one
>ply now and then, when you see a tactical problem later.

You are right.  In my experience, the consequences tend to be even more serious
when you use
path-dependent reductions in addition to path-dependent extensions.

>Related is the problem
>with repetitions (and 50 moves rule), which are also not handled correctly by
>the HT.

Yes.  In the past I used to solve this problem by not storing nodes with
repetition
draws or 50 move draws in the transposition table, but this became too expensive
when I changed my search to MTD(f).

Tord



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.