Author: Richard Pijl
Date: 05:37:59 03/07/06
Go up one level in this thread
>>The key here is that the white king can help stop the black pawns, in the >>original position its own pawn block the white's king access to the black pawn. >>You removed it in this position so the defending king is close enough. The Baron >>recognizes this and doesn't play the sac: > >I understand but this is not about distance of the king from the pawns. > >I assumed that the "defending king is too far away" is only about distance of >the king from the pawns. > >I decided not to try to implement too complicated rules in the evaluation >because I may have bugs if I try to do it and I believe that it is possible to >earn more by search improvement so if I spend time on chess programming it is >mainly about search. It is not as complicated as it sounds. I'm really measuring distance, but it depends on side to move. So when it is the defending side to move, I check whether the defending king can move to a square that is within the quadrant of the farthest pawn. With this, I remove all squares that are attacked by the pawnside, or occupied by the own side. This means that I do get some false negatives sometimes (i.e. not spotting that pawns are unstoppable when the king is permanently locked out of the quadrant), but that will be compensated by search anyway. Where you want to spend your time when improving your engine is a matter of taste. I like to fiddle with evaluation, you focus on search. We're both improving so I guess both approaches work. Richard.
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.