This topic is now archived and is closed to further replies.


looking for better chess evaluation function

Recommended Posts

i have been working on a chess engine for few monthes by now. it has alpha-beta, PV, Iterative deepening, verified null-prunning. i use, highest victim, lowest attacher move ordering scheme. i recently removed the aspiration windows, because the pruning reached a point where a -infinity to +infinity windows, are better than having to re-search. the engine now reached the point where it can search 7-ply with quiescence in an average of 10 sec. (on my 2.0GHz computer) with nearly 3.5x for each additional ply. thats not a record at all, i know. i dont use bitboards. anyways, here is what i want to do next: make a better evaluation function, mine is too lame, (it knows about doubled, isolated, and passed pawns, and importance of castling) it dont have piece-square tables, nor king safety, and mobility evaluation. and it treats openings, middle-games, and end-games the same. i gave my program to some good chess playing friends, and had the same feedback: its hard to go against it on tactical basis, but you can easily put in bad situations because it neglects development, and strategic moves. i really need a good explantion for the implementaion of such stuff: **how can i tell the computer that its in the opening, middle-game, or end-game stage. a suggested piece-square table for each. **an explantion of terms important to define king safety, and mobility. **better pawn structure evaluation is also important. where can i find such information. any explanations, or links are much apreciated. questions asking for help in evalating functions has been asked before but i found no well-explained answers. please help. [edited by - zaidgs on April 8, 2004 7:07:21 AM]

Share this post

Link to post
Share on other sites