Author: Robert Hyatt
Date: 15:32:33 03/01/00
Go up one level in this thread
On March 01, 2000 at 15:55:39, Daniel Clausen wrote: >Hi > >[snip] >:It is possible to get a good estimate for the relevant probability by dividing >:the number of games when ignoring the EP square caused problems by the number >:of total relevant games > >Um.. why (the hell ^H^H^H^H^H^H^H^H :) would you want more data about this EP- >issue? We all agree it *can* lead to problems, we (prolly) all agree it can be >fixed in a rather trivial way. Why not "case closed"? I dunno about you, but my >engine is full of non-obvious bugs - I'm happy to fix such an obvious one with >almost no effort. > >Ok, here's an analogy: For a root position we know that if we choose move A, >the opponent gets a draw given best play. For move B from the root position >we find out that there's a move for the opponent which wins our queen. I suggest >we stop searching move B. Deal? :) > >Kind regards, > -sargon :) We also got led astray by mentioning castling in the same post. Castling isn't such a problem, since one side can only castle at most one time in any path. But EP can happen _many_ times in one path, and can continue to happen throughout the game... They have different probabilities of affecting the root move.
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.