Author: Norm Pollock
Date: 18:32:16 07/02/04
Go up one level in this thread
When I used the "optimize" setting (which is the default) in the book setting of each engine's opening book and played a 60 game match between TheKing 3.23 and TheKing 3.33, I found 22 twin games. I used Scid's detection of twins. The key things about the "optimize" setting is that it turns on "Tournament book" and sets "Variety of play" to near its minimum. (results with twins present: v3.33=34.0/60.0::v3.23=26.0/60.0) When I did not use "optimize" and turned "Tournament book" off, and set "Variety of play to its maximum, I was able to run another 60 game match between the same two engines and there were NO twins. (result with no twins: v3.33=37.0/60.0::v3.23=23.0/60.0) In both experiments, "Influence of learn value" and "Learning strength" were at their maximum. Suggestion: If you are running a tournament in the new Fritz8 gui, and 2 engines will face each other more than 2 times, then there is a risk of having twin games. To effectively avoid (or minimize) this risk, do the following in the Book Setting for each opening book: turn OFF "Tournament book" and set "Variety of play" to MAXIMUM.
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.