Author: Peter Fendrich
Date: 05:25:25 05/30/01
Go up one level in this thread
On May 29, 2001 at 21:53:23, Fernando Villegas wrote: >Hi Peter: >How about mobility as a principle to take always into account? A closed position >is one where mayor pieces lose room to manouver. From that simple fact some also >simple rules follows: >a)to penalize in a degree pawns chains located in the same colour squares than >the bishop IF just one bishop stay in the board. >b)to penalize moves where, if not a clear an sustantial advantage is won inside >the horizon of search, the piece moved lose more than a certain proportion of >his previous mobility. >Attention to the fact that those rules should apply only if nothing compensate >the losing of mobility. Shpuld be as, let us say, one of the last rules to be >seen after everything else did not give nothing. >I presume that with only those rules in operation Terra would had avoided that >awfully ugly and losing position >Fernando Hi, Thanks! I will take a look at it. As you know it's not easy to write an efficient evaluation function. In fact it would be very easy to make Terra to never end up in such a positions but that could hurt the overall performance if not done right. Currenlty, Terra gives some bonus for the white pawnchain, severe punish for whites King position (without pawn shelter), some overweight for its own bishop compared to the white (See below). All stupid in this position but useful in a more open ones. a) I do that (for the bishop's own pawns) with increased weight in the endgame and as you can see in this specific position black has the 'strong' bishop in this respect. Terra evaluates it's bishop slightly stronger than whites despite its h8-position which is somewhat penalised! b) I'm not sure how to implement that, especially the exceptions that you give. Regards, Peter
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.