Author: Dieter Buerssner
Date: 17:03:25 04/09/05
Go up one level in this thread
On April 09, 2005 at 18:42:04, Uri Blass wrote: >On April 09, 2005 at 18:15:58, Vincent Lejeune wrote: > >>Congratulation Dieter, your engine is the first who report a draw here ! >Yes but it needs many hours when movei need short time to see scores near -0.3 >pawns against itself. > >It use drawing heuristics but it's heuristics helps it only to get the score >closer to draw. > >It may also find 0.00 after many hours but I do not think that it is important >to give it many hours to check it. Uri, we both won't doubt, that such heuristics will help here - and probably even won't hurt in most games. Perhaps they can help in some games (perhaps 1 in 100ß) and hurt in even less gams (say one in 100ß0). I personally do not like the idea for my engine. I prefer my engine to be rather reliable in analysis in general. Also, you use HTs only for move ordering. In that case, having path dependent scores will never be a problem. Without having it investigated very carefully, I fear that path dependent scores (which you will have, when you say: after the fifth check in a row, when my score looks negative, pull the score in the direction of zero) could hurt in very rare occasions. They at least could confuse the user. OTOH not using such heuristics, I fear could not really hurt either. For example in this position. I assume (without having it tested), that Yace will draw it at Blitz time control without any problem (while showing unrealistic score). Of course, there might be positions, where it matters. Regard, Dieter
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.