Author: Simon Waters
Date: 16:18:43 11/08/00
Go up one level in this thread
On November 08, 2000 at 18:55:12, Bob Durrett wrote: > >Actually, I was assuming that the GM would have selected an opening repertoire >based partly on his/her past experience, plus knowledge of the general >difficulties such programs face in the opening phase. The main concern regards >preparation from the position [leaf node] that is at the end of the repertoire >line. > >Perhaps the GM should prepare for more than one line, each of which might begin >at that leaf node. Obviously, too many branches could become unmanageable. > >While playing with Fritz, in home analysis, I vaguely recall that the "best line >so far" changes rapidly in the beginning but as time goes by, the line changes >do not occur so often. It is during one of these "long waits" that the exact >amount of time the GM uses in analysis may not be so critical. Sound right? I see what you mean - I think you are more likely to stumble on positions it doesn't understand, than to "defeat it from home". It is possible that a few seconds extra in the opening might alter the state of the program such that it would exclude or include a line in it's analysis resulting in a different move. Some programs are more deterministic than others, I do a little with gnuchess, which doesn't think in the opponents time, and doesn't carry much over from one move to the next in terms of analysis, killers etc. Matched against Crafty, they can be persuaded to play very similar games again and again. Of course Crafty learns and varies slightly game to game, even without any deliberate randomness features.
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.