Author: Uri Blass
Date: 10:59:00 08/13/05
Go up one level in this thread
On August 13, 2005 at 13:32:06, Steve Maughan wrote: >Uri, > >>... and I think that his ideas about evaluation are clearly better than the >>ideas that I had before reading fruit. > >Out of interest which of Fruit's ideas are you speaking about? > >Steve I will give 2 examples: 1)King attack I never thought about the idea to evaluate attack near the king based on the value of the attacker(Queen 4,Rook 2 Bishop 1 knight 1). In movei I simply used the number of squares near the king that were attacked and even in this case I decided to limit it to 3 squares in the rank above the king because I did not want to have bigger maximal evaluation when the king controls more squares. Of course it can be improved. Maybe the idea is not original but I learned it from fruit and I plan to use something similiar but hopefully better. 2)giving a small bonus for rook that is in the same file as the opponent king and small bonus for rook that is at distance of one file from the opponent king(the bonus is based on the stage of the game and is 0 in the endgame). I also did not think about that idea. It is still not implemented in Movei but I plan to implement it when I work again on Movei. I am sure that I am going to find good ideas how to evaluate backward pawns and candidate pawns in fruit. Movei does not evaluate backward pawns because the last time that I tried I could not see improvement. I even did not try to evaluate candidate pawns but I plan to learn exactly how fruit evaluate them. I plan to emulate a lot of fruit evaluation in order to do Movei better but I have no promise when it will happen. Uri
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.