Author: Roland Pfister
Date: 02:04:32 05/12/98
Go up one level in this thread
On May 12, 1998 at 03:24:54, Bruce Moreland wrote: > >On May 12, 1998 at 01:17:08, Kai Skibbe wrote: > > >>WKh3, Rh6, Pg3,g2 >>BKc5, Rb6, Ph7 >>White to move > >>Why is this position so difficult for programs ? > >Thanks for this position, it was a lot of fun. > >The reason it is hard is that after the rook exchange, if white just >goes straight for the pawns, it is a draw. This K+P ending is trickier >than it looks. > >1. Rxb6 Kxb6 2. Kh4 is drawn after 2. ... Kc5. > >1. Rxb6 Kxb6 2. Kg5 wins I believe, but 2. ... Kc5 3. Kg5? is drawn You believe? I thought you had a tablebase for that ending. >after 3. ... Kd4. The correct move seems to be 3. Kf4! (box), with a >big fail high. I think everything else draws in this case. > >1. Rxb6 Kxb6 2. g4 also wins after 2. ... Kc5 3. Kg3! (box) Kd4 4. Kf4! >(box). > >In these endings, black gets to hang onto his h7 pawn for a while, so >these programs are choosing 1. Rxh7 because they'd rather be two pawns >up in a R+P ending than one pawn up in a K+P ending. > >If the K+P ending were easier, they'd all go for it. > >In my estimation, this is a judgement call on the part of the program. >I think that tweaking a program so that it solves this might be a >mistake. > >bruce Patzer "finds" Rxb6 after 50 secs on a Ultra 167MHz at iteration 12 but has fail highs and lows with rising evaluation (from 3 to 4). No PV :-( When I entered Rxb6 Kxb6 the evalution stays at 2.x. PV at iteration 22: 2.Kg4 Kc5 3.Kf4 Kd6 4.g4 Kd5 5.g5 Ke6 (2.09) after 15 minutes. What does your tablebase say to my PV? Roland
This page took 0.01 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.