Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Crafty and Repetitions

Author: Marcus Heidkamp

Date: 00:59:20 07/18/01

Go up one level in this thread


On July 18, 2001 at 01:37:56, Artem Pyatakov wrote:

>I am not sure if Dr. Hyatt is back yet, so if someone else can answer this I
>would appreciate it.
>
>I was looking at the way Crafty handles repetition and came across the following
>comment:
>/*
> ----------------------------------------------------------
>|                                                          |
>|   insert the board into the next slot in the repetition  |
>|   list.  then scan the list.  we look for the case where |
>|   the position has been seen one time before, unless we  |
>|   are at ply 1 or 2, where we must have seen the same    |
>|   position twice prior to this.                          |
>|                                                          |
> ----------------------------------------------------------
>*/
>
>There is also a related comment:
>"note that for a repetition to happen in the first two plies of the tree, the
>position has to be repeated three times, while for plies beyond two, two
>repetitions trigger a draw score."
>
>My question is this: Why does Crafty make this exception for plies 1 and 2? What
>is special about them?
>
>Thank you.
>
>Artem Pyatakov

IN case nobody else answers this first, let me try to explain: According to the
official rules of chess a position has to be repeated three times before a
player can claim a draw. If chess engines would look for threefold repetition,
the trees would grow so large and the third repetition might be pushed beyond
the search horizon (so the engine "thinks" to avoid the draw) that the search
would become inefficient. The "normal" procedure is therefore to look, if the
position has been on the board once, and score this a draw. However, for the
first couple of plies (IE root, and plies one and two) Crafty want's to prevent
scoring a position draw, if it is not a thrird repetition. This avoids the
search to do root moves it thinks to draw, that are officially no draw at all.
In summary: By searching only for twofold repetition below the root you seedup
the search and detect potential draws by repetition faster, you have to be
careful not to violate the rules of chess at the root.

I hope that clariefies things a little.

Marcus



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.