Author: Steffen Jakob
Date: 06:24:27 12/28/01
Go up one level in this thread
On December 28, 2001 at 09:17:00, Mark Young wrote: >On December 28, 2001 at 09:10:27, Steffen Jakob wrote: > >>Hi! >> >>Today Hossa played a blitz game with Tinker at ICC where a KBPKP endgame was on >>the board: >> >>[D]8/8/p4b1P/3K4/8/8/8/1k6 b >> >>Hossa evaluates this as a win for black, because the black pawn has a black >>promotion square and there is no white pawn on the b file in front of the black >>pawn. Unfortunately this position is a draw. Can you define a rule for such a >>position class ("search will do it" doesn't count; at least the big score I gave >>for this position costed 1/2 point in that game)? I think it should be a >>combination of this: >>1. white has an advanced white passer >>2. the black king isnt able to protect the black pawn >>3. the distance between the pawns is big >Would it not be better for Hossa to use EGTB for this type position? Then Hossa >will solve this position without error. Yes it would solve this position, but 1. I want my engine to be able to play this correctly even if the user (so far only me) doesnt have all 5 men tables installed. 2. I dont probe tablebases in qsearch 3. This is a pattern which might occur in positions with more than 5 men like this: [D] 8/8/p4b1P/3K4/8/8/7P/1k6 b Greetings, Steffen.
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.