Author: Dann Corbit
Date: 11:30:35 09/28/01
Go up one level in this thread
On September 28, 2001 at 14:27:05, Dann Corbit wrote: >Here is the current test set. >ftp://cap.connx.com/pub/EPD/pawntest.epd > >Now that I have something to calibrate with, I would like to know what sort of >things to tinker with in fixing pawn evaluations. > >Obviously pawn chains are good, isolated pawns are bad. > >From looking at what ExChess does (which has particularly good pawn evaluation) >I see that it --> >1. checks for outside passed pawns, >2. checks to see if the king can catch the pawn, >3. looks to see if the pawn is protected and >4. looks to see if the pawn is connected. > >It also gives a sliding bonus, depending upon the stage of the game. > >The relevant code is attached for your perusal. Actually, no -- it's not. It's found here: ftp://cap.connx.com/pub/chess-engines/new-approach/EXCH402S.ZIP In particular, see score.cpp >Other programs I have examined also check to see if there is an enemy rook >attacking on the file and if there is a friendly rook protecting on the file. > >Pawn trousers are also something nice: >[D]8/8/4k3/5p2/1P1P1p2/2K2P1p/7P/8 w - - id "The King's got his trousers on!"; > >What else should I look for in evaluation of pawns?
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.