Author: Robert Hyatt
Date: 17:38:48 05/27/99
Go up one level in this thread
On May 27, 1999 at 19:04:24, Paul Richards wrote: >On May 26, 1999 at 23:24:02, Robert Hyatt wrote: > >>OK.. I read the whole thread, and tried it. It is the piece of knowledge >>that says K + B + RP on wrong color draws if the opposing king can get to the >>queening square before the pawn. That is why Crafty plays Nh6 as black and >>_never_ considers gxh6. If you play Nh6, a 1 ply search rejects gh because it >>evaluates to DRAW. Very simple answer after looking at it.. so forget my >>first answer. That happens, but not in this case... > >What's interesting is that a 1-ply search rejects gxh6 immediately, but >if you set Crafty to analyze for black after gxh6, it searches to 15-16 ply >before seeing that Kh8 leads to the draw. As a matter of fact it awarded >Kh8 a !! all by itself, which I thought was a very nice touch. :) So it >appears that it decides on the draw different ways before and after gxh6, >or that the special knowledge code is invoked at a different point? 1. make sure you don't have a "position.bin" file. That kind of learning data can screw things up if you played with this position a bit. 2. if not, all I get for any K move, from ply 1 on is 0.00... ie Kxh6 is a draw and it doesn't vary. Bet you got 'learned' into this. :) Happens to me frequently when debugging, so much so that my test Makefile does a "rm position.*" every time I compile.
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.