Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Hmm... the mystery remains...

Author: Matthew White

Date: 15:53:01 03/13/03

Go up one level in this thread


On March 13, 2003 at 14:57:32, Albert Bertilsson wrote:

>Now I've done some testing with smaller values...
>
>Using these constants increases the nodes searched by 5%
>#define DOUBLED_PAWNS -1
>#define PAWN_NOT_ISOLATED 1
>#define BOTH_BISHOPS 1
>#define ROOK_HALF_OPEN_RANK 1
>#define PAWN_CLOSE_TO_KING 1
>#define QUEEN_CLOSE_TO_OKING 1
>#define KNIGHT_CLOSE_TO_OKING 1
>
>Using these constants increases the nodes searched by 30%
>#define DOUBLED_PAWNS -5
>#define PAWN_NOT_ISOLATED 4
>#define BOTH_BISHOPS 10
>#define ROOK_HALF_OPEN_RANK 4
>#define PAWN_CLOSE_TO_KING 3
>#define QUEEN_CLOSE_TO_OKING 4
>#define KNIGHT_CLOSE_TO_OKING 5
>
>Obviously anything I do makes the node count go up, and the basic questing
>remains, isn't that wrong?
>
>Or is the problem that I've only done very little evaluation and that this is to
>little and this "confuses" the engine?
>
>/regards Albert
One reason that you will get more searches with this type of knowledge is that
your eval will show ANY rook move to a half-open file as good. It will cause the
search to evaluate more rook moves, as those positions without material
difference will now be searched deeper.

Matt



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.