Author: JW de Kort
Date: 05:15:27 06/01/01
Go up one level in this thread
I have implemented a draw by repetition code once i foud out about this behaviour. My evaluation function does noet return negatieve values in darwn posities. The methode is use is the one John Stanback (hope i have the vcorrect spelling) send to CC a few weeks ago. It is indeed not properly implemented yet. On May 31, 2001 at 23:28:58, James Swafford wrote: >On May 31, 2001 at 08:55:44, JW de Kort wrote: > >Assuming your draw by rep works properly, the only other conclusion >would be that the search revealed that, in the program's estimation, >the draw was more valuable than proceeding. > >This implies your program is scoring won positions with negative values, >or the draw by rep code is buggy. How do you detect draw by rep? > >If this happens when two pawns up, I'd suspect the draw by rep code >was busted. It's pretty tough for the positional evaluation to outweigh >two pawns (not that it doesn't happen...). > > >-- >James > > >>Hi, >> >>My program has by now reached a level were it can play strong enough to leve the >>middle game in a favorable position. e.g. advantage of a bishop or a few pawns. >>Every time this happens the program seems to reach some kind of an equilibrium >>and the position will result in a draw by repetition. Can anybody explain why >>this happens? I would expect the program to try to make use of this advantage. >>This even happens when the position was completely won having an advantage of >>two pawns. >> >> >>Can anybody explain why this is happening? >> >> >>Greetings >> >>Jan Willem
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.