Author: ujecrh
Date: 15:11:35 05/23/00
Go up one level in this thread
On May 23, 2000 at 15:55:43, Mogens Larsen wrote: > >I could be very interesting to see, if it's possible to implement opening >knowledge without hurting the general performance of the chess program itself. > >Best wishes... >Mogens Lets take crafty for example. Even if it has changed since the last time I looked at the code (which was at least a year ago), there used to be a kind of phase() function that aimed to set if we were in opening, middle or endgame. With this in mind, we can imagine that specific opening evaluation would not arm at all any other phases of the game because it would then be processed only in the opening. Simple things like keeping track of how many moves a piece have done so far to avoid tempi loss might help the engine. But I may not see all the work to do and it is even possible that some programs already have this kind of evaluations. Anyway, I still do not consider this as very difficult to do and I believe the only reason why many programmers do not even consider it is still that opening books are good enough. Another true argument in favor of data vs knowledge in opening is that, using the book saves a lot of time which allows to do more search in stages that are more critical. Ujecrh
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.