Author: Uri Blass
Date: 03:07:23 04/12/04
Go up one level in this thread
On April 11, 2004 at 20:41:29, Steve Maughan wrote: >Uri, > >>I also prune based on history but one of my conditions is that >>faillowcount>128*failhighcount and I do not use number of nodes. >> >>I never tried your way because it seemed more logical for me to compare number >>of failhigh with number of faillow and not number pf failhigh with number of >>nodes. > >Interesting stuffand I plan to experiment with History based pruning in Monarch. > One point that I would make is that number of nodes may be interesting near the >leafs since there may be little search experience with the move and hence the # >of FH v FL may be a spurious statistic. > >Steve If there is a little experience with the move then it seems to me dangerous to prune it and this is the reason that I try FH v FL and not FH vs nodes. If a moves failed high 10 times in 1000000 nodes and was tried 10 times then FH vs nodes may tell you to prune it and I do not think that it is a good idea. Uri
This page took 0.02 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.