Author: Howard Exner
Date: 12:35:21 11/12/98
Go up one level in this thread
On November 12, 1998 at 12:56:05, Ed Schröder wrote: >>I was noticing during testing of positions that I was getting different >>times even though my settings were identical. See if someone can duplicate >>this... set combination Off, AntiGM to Off and run the Louguet position #10 >>in infinite thinking mode. Record the time. Now set AntiGM to smart and >>record the time. They should be the same. Now set AntiGM to Strong, record the >>time. Finally set AntiGm to Smart and now the time is the same as the one just >>previously tested (ie:AG=Strong). It seems as if the Smart setting is just >>remembering the setting it had previously. Should not the Smart setting >>reproduce the same time? > >Your observations are correct. Smart has its own mysterious ways. That's >all I want to say about it. So to have equal times/moves/scores use: >off/active/strong. Does this mean that if Rebel 10 were to play Anand again and Anand played the same moves, Rebel would play differently? Also will test suite results or Analysis of games always be varied? I thought the the Smart setting allowed Rebel to choose one of the 3 AntiGM options based on Rebel's assesment of what the position suggests. Is the Smart setting just choosing one of the 3 at random then? Which of the 3 settings do you suggest for Rebel in the absence of using the Smart Option? For example the PlayStyle options are understandable in how they affect mobility - I usually like Active or Agressive as just a personal preference. It will be fun to test the different AntiGM options. From your testing did one choice do significantly better on test suites? > >>I further experimented and found that when I change the thinking time >>from infinite to a tournament control, say 40 in 6 hours, that all seems to be >>working fine using AntiGM = Smart. Rebel now is choosing from among the 3 >>options as the position dictates. Can others confirm this also? If true >>should we not test EPD suites in infinite thinking mode but use a tournament >>time control instead? The other possibility is that I am doing something wrong. > >There are no differences because of changing the levels. > >>One last question on AntiGM. For endings it looks as if the Off,Active,Strong >>and Smart options do not have any effect. Is this the case? This would make >>sense to me as endings require their own unique approach. > >You are a real expert. You are right. In certain endings anti-GM does not have >any effect. Well seen, my compliments. > >- Ed -
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.