Author: Tony Werten
Date: 05:29:30 02/05/03
Go up one level in this thread
On February 05, 2003 at 06:17:20, Vladimir Medvedev wrote:
>My engine uses the following extensions:
>
>* mate threat extension ( = 32/16 ply )
>* single move extension ( = 24/16 ply )
>* pawn on 7th extension ( = 16/16 ply )
>* check extension ( = 12/16 ply )
>* recapture extension ( = 12/16 ply )
>
>At each node only one (max) extension can be applied.
>I tried two possible variants of extensions rule:
>
>Variant 1):
>AlphaBeta(...)
>{
> ...
> if( remain_depth < 4*HALFMOVE ){
>
> // analyze extensions conditions and apply extensions here
> // this is done for nodes near qsearch margin
>
> }
> ...
>}
>
>Variant 2):
>AlphaBeta(...)
>{
> ...
> if( current_ply < 2*current_iteration ){
>
> // analyze extensions conditions and apply extensions here
> // this is done for nodes not very far from root
>
> }
> ...
>}
>
>//
>
>While reading sources of other engines I found that 2nd variant is more common
>(enable extensions at nodes not very deep). But my engine seems to perform
>better with the first one! (faster solving tactical problems and analyzing). As
>to strength, I did not found significant difference in these two versions
>matches.
>
>Dear GURUs, what can you advice? What is the more suitable place in the tree to
>extend - near root or near leaves?
The general idea is that for mate threats variant 1 is used, for the other
variant 2. Special in this case with the mate threat extension being so high ( 2
ply ) you should limit it to the last ply.
Tony
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.