Author: Vincent Diepeveen
Date: 15:05:15 08/09/01
Go up one level in this thread
On August 09, 2001 at 15:58:31, Tom King wrote: >Following recent discussions re: using R=2 for null move vs. using R=2 near the >tips, and R=3 nearer the root, I ran a little test. >Here's the results of running Francesca R=2 vs Francesca R=2/3 (running for 12 >hours, playing blitz, on my little Celeron 500 machine). >Francesca R=2 : 48.2 % >Francesca R=2/3 : 51.8 % Please also compare with R=3 everywhere. I did extensive tests with 3 combinations a) R=3 b) R=3 and near tips R=2 (last 4 ply) c) first nullmove R=3 second and further nullmove: R=2 the last i abandonned quick as you need to store a bit more in hashtable to do it and i can afford a bit but i founded it a waste of time to try. i experimented most between a and b. And with a i search usually a ply deeper than with b, but i still am doubting which is better. Some experiments indicated that at auto232 player i score better with b. Opponents were the commercial engines. I still find it hard to conclude anything from play against the same engine. I bet R=3 comes out as best when you do that with francesca. >109 games were played. > >A small sample, but as I alluded to in earlier postings, R=2/3 might be >*slightly* better for Francesca. Loads more games would be needed to verify this >with any certainty, of course.. > >Rgds, >Tom
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.