Author: Tord Romstad
Date: 03:00:15 11/29/03
Go up one level in this thread
On November 29, 2003 at 04:01:35, martin fierz wrote: >hi tord, Hi Martin! >interesting post! Thanks! I'm happy that you appreciated it. >in fact, i am already doing one of the many things you suggest: if a king is >under heavy attack (whatever my eval function thinks that is...) i don't >nullmove. If you read my post more carefully, you will see that this was not precisely what I suggested. I avoid nullmove only if my evaluation function detected a mate threat or something almost as serious (a pawn promotion or the capture of a big piece). If the king is under heavy attack but there is no direct mate threat, I do a nullmove search, but with a lower reduction factor than usual. >i'm not so sure whether this helps at all though - i never really >benchmarked it, it seemed like a sensible idea. >on the other hand, if you do nullmove in such positions, you might get a mate >score back and decide to extend based on that mate score, so i have become a bit >uncertain whether it's really a good idea not to nullmove in such positions - >you lose this extra information that you will be mated if you do nothing about >it. You're right, but I try to detect all mate-in-1 threats in my evaluation function (it's not perfect yet, but it's getting better), hence I only need a nullmove search to detect deeper threats. When I search with a lower value of R in positions with significant pressure against the king, this just increases the probability that the null move search will discover a serious threat. Tord
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.