Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: History Heuristic

Author: Sune Fischer

Date: 02:02:02 03/16/04

Go up one level in this thread


On March 16, 2004 at 04:41:40, Renze Steenhuisen wrote:

>
>Hi all!
>
>I did some measurements on my move-ordering, and it seems that the positive
>effect of the History Heuristic wears of with increasing search depth? Is this
>observed by others as well or am I doing something awfully wrong here...

You should recalibrate the history table every so often, I do it about once a
second.

The history table contains information accumulated from the local search space
and is usable as long as you are in that area.
In a global search it has little value, the trees just look too different when
far apart.

If you do not recalibrate the table will grow stale and useless over time.

-S.


>History Heuristic:
>
>    int table[64][64];
>
>    every time the 'best move' was found (either fail-high or all moves
>    searched) I do:
>           table[from][to]+=remaining_depth*remaining_depth
>
>    The non-capture moves are sorted in History Heuristic order.
>
>Cheers!
>
>Renze



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.