Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Pawn endgame test suite

Author: Peter McKenzie

Date: 12:24:57 10/07/00

Go up one level in this thread


On October 07, 2000 at 14:49:16, martin fierz wrote:

>On October 07, 2000 at 10:41:27, Vincent Diepeveen wrote:
>
>>On October 07, 2000 at 04:31:09, Peter McKenzie wrote:
>>
>>>I've been browsing 'Comprehensive Chess Endings, Pawn Endings' by Averbakh &
>>>Maiselis, picking out a few interesting looking positions to test my program
>>>with.  Here is the little test suite I came up with:
>>>
>>>8/1p4kP/5pP1/3p4/8/4P3/7K/8 w - - bm e4; id "CCE4 479";
>>>8/1pp5/3k3p/PP6/2P2K2/8/8/8 b - - bm Kd7; id "CCE 491a";
>>>2k2K2/8/pp6/2p5/2P5/PP6/8/8 w - - bm a4; id "CCE4 530";
>>>8/pp3p2/8/6kp/8/3K1PP1/PP6/8 b - - bm f5; id "CCE4 608";
>>>8/pp2k1pp/2p5/2P1p3/2P1P2P/6P1/P7/2K5 b - - bm g5; id "CCE4 679";
>>>8/1p6/p1p5/P1Pp2pp/1P1P1p1k/5P1P/6PK/8 w - - bm g3 g4; id "CCE 680";
>>>8/1k6/p4p2/2p2P2/p1P2P2/2P5/P1K5/8 w - - bm Kc1; id "CCE 765";
>>
>>all peanuts except #7 kc1 under 1 minute single cpu,
>
>>i have to checkout Kc1 position 7. i'm at 40 ply now, but it doesn't
>>see how to win with Kc1. all moves are +1.96
>
>position 7 looks wrong to me. i analysed it a bit - it's one of those
>'corresponding squares' studies - the author's intention is this: there
>are 3 special squares on the board: a3, e4 and h6. whenever the white
>king appears on one of these, the black king MUST be on a5, d6 and g7,
>respectively. you call a3 and a5 'corresponding squares' (CS). the point of
>all of these studies is this: once you have identified the primary CS,
>you can go on and identify secondary CS - for instance f3 corresponds to e7,
>because from f3 the white king threatens to move both to e4 and to h5
>very quickly, so black must be ready for this. b2 corresponds to b6, and
>so on. to win you must find a square for which the other side has no
>corresponding square.
>i see the author's intention in this study, but i suppose he forgot something.
>the reasoning seems to be that black must do something about his f6-pawn,
>but black has a stalemate trick with ...a3,Kb6,Ka5,Ka4 and a5. so white
>cannot go after the f6 pawn and the position is a draw. i'm pretty sure
>that the main line in peter's book goes
>1.Kc1 Kc7 2.Kd1 Kd7 3.Ke1 Kc7 4.Kf2 but here 4 ...a3 is a draw.
>(but of course it's rather dangerous to critizice averbakh :-) )

Well spotted, I think you are correct!
The study is by Zinar, 1983.

>
>for a position with corresponding squares which really works and is hard
>for computers, check my chess puzzle page, www.fierz.ch/pontresina.htm,
>position number 2. the solution is Kb6 or Kc6 followed by Kc7 - fritz
>doesnt really see it's winning in this position, just a small +1 or so.
>
>cheers
>  martin



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.