Author: Dave Gomboc
Date: 00:19:50 05/27/99
Go up one level in this thread
On May 27, 1999 at 03:16:58, Dave Gomboc wrote: >On May 27, 1999 at 02:40:10, Peter Kappler wrote: > >> >>Do any of you have any useful positions that you use for debugging your >>hashtables? >> >>I've been trying this one, which I think is pretty well known: >> >>8/k7/3p4/p2P1p2/P2P1P2/8/8/K7 w - - bm Kb1; >> >> >>Supposedly only Kb1 wins, but Grok likes Kb2. After 21 plies, the score jumps >>above +2 as Grok sees the f5 pawn dropping. >> >>Genius 6 and Fritz 5.32 both like Kb1 and give it a winning score (+2.5 to +3.0) >>immediately. >> >>I've seen this problem posted often enough that I trust that Kb1 is the only >>winning move. So not only is Grok not finding the best move, but it seems to be >>misevaluating Kb2. >> >>Does anybody else have problems with this position? >> >>Thanks, >>Peter > >You might (hand-)generate the co-ordinate square map for this position. Then >you will be able to look through the analysis of your program to see where it >screws up from, working back from where white can pick up the f-pawn. > >Dave Argh, sorry about the double post -- unresponsive GUI. :( I wanted to add, if this is the famous Fine position, then Kb2 is also winning, even though it gives away the "twisted diagonal opposition", so to speak. Reference "Computer Chess Compendium" (Levy, ed. 1998) for more details. I think it's discussed in 8.2 (though I wish I could find a copy to confirm this.) Dave
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.