Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: passed pawn evaluation.

Author: Omid David Tabibi

Date: 14:07:56 01/12/04

Go up one level in this thread


On January 12, 2004 at 13:05:38, Robert Hyatt wrote:

>On January 12, 2004 at 10:24:59, scott farrell wrote:
>
>>On January 12, 2004 at 05:52:07, Tord Romstad wrote:
>>
>>>On January 11, 2004 at 21:18:30, macaroni wrote:
>>>
>>>>I have recently been fiddling arround with some passed pawn code, however I just
>>>>can't seem to get it working very well. Can you give me an idea of what sort of
>>>>values you give different types (in different stages) of passed pawns?
>>>
>>>Here is a rough description of what I do:
>>>
>>>Start by giving the passed pawn a bonus depending on which rank it occupies.
>>>You then adjust this bonus depending on the placement of the other pieces
>>>on the board.  Some examples:
>>>
>>>Protected passed pawns and passed pawn duos are especially dangerous.
>>>Adjust the bonus up if there is a friendly pawn beside or diagonally
>>>behind the passed pawn.
>>>
>>>Adjust the bonus up if there is a friendly rook behind the pawn, and down if
>>>there is an enemy rook behind it.
>>
>>this is where you need to be careful if you have a pawnhash, you cant hash in
>>other pieces :) you have to put it into the rook/king eval instead.
>>
>>Scott
>
>SSSHHHHHHHHhhhhhhhhhh!!!!!!!!
>
>Everybody needs to debug _that_ problem at least once.  Some of us more
>than once.  :)

Hehe, tell me about it :) When I first implemented pawn hash, I ran a fixed
depth search to make sure the node count with and without pawn hashing is
exactly the same (I do that after every non-eval/search related change). The
number differed in some positions, but was the same in others. After looking at
the code for the nth time, I found out that I check for the presence of a bishop
in g2 when pawn structure is f2-g3-h2 and that affects the pawn_eval score!

>
>
>>
>>>
>>>A passed pawn is usually more dangerous if it is supported by a king.
>>>Increase the bonus if the friendly king is close, especially if it is
>>>somewhere in front of the pawn.  Likewise, you should increase the bonus
>>>if the enemy king is far away.  In rook endgames, it is probably a good
>>>idea to further increase the bonus if the enemy king is cut off by
>>>a rook (I don't do this yet, but I plan to add it in the near future).
>>>
>>>Passed pawns are most dangerous if they are able to advance safely.
>>>If the passed pawn is blocked by an enemy piece, adjust the bonus down.
>>>If it is unblocked, increase the bonus if your static exchange evaluator
>>>decides that it is safe to push the pawn, and reduce the bonus if it is
>>>not safe to push it.
>>>
>>>Tord



This page took 0.07 seconds to execute

Last modified: Thu, 07 Jul 11 08:48:38 -0700

Current Computer Chess Club Forums at Talkchess. This site by Sean Mintz.