Author: Robert Hyatt
Date: 06:32:48 12/29/99
Go up one level in this thread
On December 28, 1999 at 12:42:31, Randall Shane wrote: >On December 26, 1999 at 12:39:59, Len Eisner wrote: > >> <snip> >> >>It would be helpful if Chessbase could address this limitation programmatically >>in Fritz 6 and the other commercial engines in the same way that Bob Hyatt did >>it in Crafty. Right now, if I want to study the KPP vs. KP ending, I use the >>Crafty engine because it is the only one that will promote the pawn without the >>KQP vs. KP tablebase. > >Dr. Hyatt, > How does Crafty handle missing tablebases to avoid the problem above? > >Thanks! It is actually not hard, although it is also not optimal. If the initial position at the root is a mate-in-N tablebase position, but after trying each legal root move, no mate-in-(N-1) positions were found, it must mean that a promotion move drops into a non-tablebase position. What I do is simply turn tablebase probes off, and do a normal search. This is guaranteed to prefer the promotion if it is safe and if it doesn't stalemate the opponent, which neatly solves the problem.
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.