Author: Robert Hyatt
Date: 03:32:56 08/04/98
Go up one level in this thread
On August 03, 1998 at 22:37:29, blass uri wrote: >I think a better idea is not to use the opening book without evaluating >in this tournament but to give the programs to calculate the best move with >some bias for the book moves(If a move not in the book is better significantly >than the book move the program should not play the book move). > >If it is not possible to do it directly then it is possible to give the computer >to evaluate the position after the book move it wants to play and the evaluation >without the book and if there is a difference of more than 0.7 pawn for the not >book move the computer will not play the book move. >I do not know if using 0.7 is a good idea and what number is optimal. > >Uri I do this. I use "book random 0" in crafty, whichs says to identify the set of book moves, and search *only* those moves with a normal search. This search is for something like 1/10th the normal search time, so book still saves time, but this makes sure crafty doesn't play an outright blunder that drops a pawn or piece (which absolutely does happen in book lines). If all book lines are not so hot (best is <= -1/2 pawn) then I pretend there is no book line and do a normal search on all the moves. If this also returns <= -1/2 I play the best book move. there are other "tricks" too... when pondering, in "tournament mode" I find the set of book moves for the opponent, do a short search on everything else, and ponder this move (which is not a book move). If he makes a book move, I still get a quick response, if he doesn't, I will very likely predict his move and *still* get to ponder a while...
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.