Author: Uri Blass
Date: 13:13:21 08/26/01
Go up one level in this thread
On August 26, 2001 at 14:56:24, Eduard Nemeth wrote: >On August 26, 2001 at 14:40:33, Roy Eassa wrote: > >>[D] 8/2r1k2p/p3p1p1/3r1p2/8/1P3PP1/P2RK2P/3R4 b - - >> >>I reached this position as white recently, and after the game I used Gambit >>Tiger 2.0 to help analyze. What occurred in the game was: >> >>1... Rcd7(?) 2. Rxd5 Rxd5?? 3. Rxd5 exd5 >> >>after which I believe white should win (which I did). I believe it's very bad >>for black to trade off both pairs of rooks, and best to avoid allowing white to >>exchange even one pair of rooks here because of isolating black's pawn. >> >>It appears that Gambit Tiger 2 on my Athlon 1.2 GHz does not really understand >>that after move white's move 3 above, white should win. And it believes the >>position is essentially equal after white's move 2. >> >>Here is its analysis after black's move 1: >> >>00:00:00.9 0.22 9 168541 h3 Kf6 Rxd5 exd5 Rc1 Ke6 Kd3 f4 Rc6+ Rd6 Rxd6+ Kxd6 >>00:00:00.3 0.16 9 184134 h3 Rxd2+ Rxd2 Rxd2+ Kxd2 e5 Ke3 Ke6 g4 fxg4 fxg4 h5 a3 >>hxg4 hxg4 >>00:00:01.9 0.12 10 445643 h3 Rxd2+ Rxd2 Rc7 Kd1 e5 Rc2 Rxc2 Kxc2 e4 fxe4 fxe4 >>Kc3 >>00:00:01.4 0.16 10 710893 Rxd5 Rxd5 Rc1 Kd6 Rc8 Rc5 Rxc5 Kxc5 Kd3 Kb4 h4 Ka3 >>00:00:02.8 0.16 11 905910 Rxd5 Rxd5 Rc1 Kd6 Rc8 Rc5 Rxc5 Kxc5 Kd3 Kd5 g4 e5 gxf5 >>gxf5 >>00:00:03.0 0.16 12 1519152 Rxd5 Rxd5 Rxd5 exd5 b4 Ke6 Ke3 Ke5 f4+ >>00:00:05.5 0.16 13 2286175 Rxd5 Rxd5 Rxd5 exd5 b4 Ke6 Ke3 Ke5 f4+ Kd6 a4 Ke6 Kd4 >>h6 b5 axb5 axb5 >>00:00:09.6 0.18 14 3833264 Rxd5 Rxd5 Rxd5 exd5 b4 Ke6 Ke3 h6 f4 g5 a3 gxf4+ Kxf4 >>d4 Kf3 d3 Ke3 >>00:00:18.1 0.20 15 7566502 Rxd5 Rxd5 Rxd5 exd5 b4 Kd6 Ke3 g5 f4 gxf4+ Kxf4 d4 a3 >>h5 Kf3 d3 Ke3 >> >> >>but after 2. Rxd5, suddenly Tiger prefers black to recapture with the pawn >>rather than the rook, and its evaluation becomes worse for black. >> >>Non-Gambit Tiger behaves very similarly. >> >>Is the lack of endgame knowledge in me or the program, or is it something else? >> >>(Using 96MB hash; Not using table bases.) > >This is NOT (!!) only position who Tiger have on "blackout" to pawn endgame. >This is real! > >here an other position (from a game Eduard - Tiger): > >[D]3k3r/1r4pp/p3Rp2/8/p7/2P5/P4PPP/R5K1 b - - 0 27 > >Tiger played now Re8??? and after Rd1+ game was lost for Tiger! > >(German: es gibt noch viel zu tun Christophe - vor allem im Bauernendspiel= >pawn endgame!) > >Eduard New position 3k3r/1r4pp/p3Rp2/8/p7/2P5/P4PPP/R5K1 b - - 0 1 Analysis by Gambit Tiger 2.0: 1...Kd7 2.Rxa6 Rc8 3.Rd1+ Ke8 4.Rxa4 Rxc3 5.Ra8+ Ke7 6.Rh8 ² (0.38) Depth: 9 00:00:00 107kN 1...Kd7 2.Rxa6 Rc8 3.Rd1+ Ke8 4.Rxa4 Rxc3 5.Ra8+ Ke7 6.a4 Rc2 ² (0.50) Depth: 9 00:00:00 133kN 1...a5 2.Re4 a3 3.Ra4 Re8 4.Rxa3 Re2 5.Rxa5 Rbb2 ² (0.32) Depth: 9 00:00:01 410kN 1...a5 2.Re4 a3 3.Ra4 Kc7 4.Rxa3 Rhb8 5.Kf1 Rb2 6.Rxa5 ² (0.48) Depth: 10 00:00:02 604kN 1...Re8 2.Rxa6 Re4 3.Rd1+ Kc7 4.Kf1 Rc4 5.Rd3 Rb1+ 6.Ke2 Ra1 7.Ra7+ Kc6 ² (0.36) Depth: 10 00:00:03 744kN 1...Re8 2.Rxa6 Re4 3.Ra8+ Kc7 4.Rg8 Kc6 5.Rc8+ Rc7 6.Rh8 g6 7.Rf8 Kd5 ² (0.30) Depth: 11 00:00:03 944kN 1...Re8 2.Rxa6 Re4 3.Ra8+ Kc7 4.Kf1 Rc4 5.Rg8 Kc6 6.Re1 ² (0.28) Depth: 12 00:00:05 1482kN 1...Re8 2.Rxa6 Re4 3.Kf1 Rc4 4.Ra8+ Kc7 5.Rg8 ² (0.32) Depth: 13 00:00:12 3068kN 1...Re8 2.Rxa6 Re4 3.Kf1 Ke8 4.Rd1 Rb2 5.Ra7 Rxa2 6.Rxg7 Rf4 7.f3 Rc4 8.Rd6 Rxc3 ² (0.32) Depth: 14 00:00:26 6611kN 1...Re8 2.Rxa6 Re4 3.Kf1 Ke8 4.Rd1 Rb2 5.Ra8+ Kf7 6.Ra7+ Kg6 7.f3 Rc4 8.Rdd7 Rxa2 9.Rxg7+ Kf5 10.Rxh7 ² (0.28) Depth: 15 00:00:56 13969kN 1...Re8 2.Rxa6 Re4 3.Kf1 Ke8 4.a3 Kf7 5.Re1 Rc4 6.Re3 Rb3 7.Ra7+ Kg6 8.Rg3+ Kf5 9.Ke2 Rxa3 10.Raxg7 ² (0.36) Depth: 16 00:02:15 33441kN 1...Re8 2.Rxa6 Re4 3.Kf1 Ke8 4.a3 Kf7 5.Re1 Rc4 6.Re3 Rb3 7.Ra7+ Kg6 8.Rg3+ Kf5 9.Ke2 Rxa3 10.Raxg7 h5 ² (0.36) Depth: 17 00:05:43 85093kN 1...Re8 2.Rxa6 Re4 3.Kf1 Ke8 4.a3 Kf7 5.Re1 Rc4 6.Re3 Rb3 7.Ra7+ Kg6 8.Rg3+ Kf5 9.Ke2 Rxa3 10.Raxg7 h5 ² (0.36) Depth: 18 00:12:15 182227kN (Blass, Tel-aviv 26.08.2001) Note that Crafty18.10 also does not see the problem and the main line is also Re8 Rxa6 New position 3k3r/1r4pp/p3Rp2/8/p7/2P5/P4PPP/R5K1 b - - 0 1 Analysis by Crafty 18.10: 1...Rb2 2.Rxa6 = (0.17) Depth: 1/5 00:00:00 1...Re8 2.Rxa6 Re4 3.Rc1 ² (0.56) Depth: 4/8 00:00:00 1...Re8 2.Rxa6 Re4 3.Rc1 ² (0.56) Depth: 4/8 00:00:00 1...Re8 2.Rxa6 Re4 3.Rc1 ² (0.56) Depth: 4/8 00:00:00 1...Re8 2.Rxa6 Re4 3.Rc1 ² (0.56) Depth: 4/8 00:00:00 1...Re8 2.Rxa6 Re4 3.Rc1 ² (0.56) Depth: 4/8 00:00:00 1...Re8 2.Rxa6 Re4 3.Rc1 ² (0.56) Depth: 4/8 00:00:00 1...Re8 2.Rxa6 Re4 3.Rc1 ² (0.56) Depth: 4/8 00:00:00 1...a5 2.Rae1 Rb2 3.Re7 ² (0.38) Depth: 4/8 00:00:00 1...a5 2.Ra6 Re8 3.Rd1+ Ke7 4.Re1+ Kd7 5.Rxe8 Kxe8 6.Rxa5 ² (0.44) Depth: 5/11 00:00:00 1...a5 2.Ra6 Ke7 3.Re1+ Kd7 4.Rxa5 Rc8 ± (0.83) Depth: 6/13 00:00:00 30kN 1...Re8 2.Rxa6 Re4 3.f3 Re3 4.Rxa4 Rxc3 ± (0.71) Depth: 6/13 00:00:00 42kN 1...Re8 2.Rxa6 Re4 3.f3 Rc4 4.Rd1+ Ke7 5.Re1+ Kd7 ± (0.82) Depth: 7/13 00:00:00 80kN 1...Ra7 2.Rd1+ Kc8 3.Rc6+ Kb8 4.Rc4 Re8 5.Rxa4 Kc7 ± (0.80) Depth: 7/13 00:00:00 179kN 1...Ra7 2.Rd1+ Kc8 3.Rc6+ Kb8 4.Rc4 Rc8 5.Rb4+ Kc7 6.Rxa4 Re8 ± (0.80) Depth: 8/18 00:00:01 326kN 1...Re8 2.Rxa6 Re4 3.Rc1 Rc4 4.Rd6+ Ke7 5.Rd4 Rxd4 6.cxd4 ± (0.74) Depth: 8/18 00:00:01 404kN 1...Re8 2.Rxa6 Re4 3.f3 Rc4 4.Ra8+ Ke7 5.Re1+ Kf7 6.Kf2 Rb2+ 7.Re2 ± (0.74) Depth: 9/20 00:00:01 698kN 1...Re8 2.Rxa6 Re4 3.f3 Rc4 4.Ra8+ Kc7 5.Re1 Rxc3 6.Re7+ Kb6 7.Rxb7+ Kxb7 8.Rxa4 ± (0.75) Depth: 10/21 00:00:02 1070kN 1...Re8 2.Rxa6 Re4 3.f3 Rc4 4.Re1 Kd7 5.Kf2 Rxc3 6.Rd1+ Ke7 7.Re1+ Kd7 8.Rxa4 ± (0.86) Depth: 11/24 00:00:06 2868kN 1...Ra7 2.Rd1+ Kc8 3.Rc6+ Kb8 4.Rc4 Rc8 5.Rb4+ Rb7 6.Rxa4 Rxc3 7.Rxa6 Rc2 8.a4 ± (0.81) Depth: 11/24 00:00:23 11775kN 1...Ra7 2.Kf1 a5 3.Re4 a3 4.Rb1 Kd7 5.Rb3 f5 6.Re5 Re8 7.Rxf5 ± (0.95) Depth: 12/26 00:01:02 31656kN 1...Re8 2.Rxa6 Re4 3.Rd1+ Ke7 4.f3 Rc4 5.Re1+ Kd7 6.Re3 f5 7.Kf2 Rb2+ 8.Re2 Rxe2+ 9.Kxe2 Rxc3 10.Rxa4 ± (0.75) Depth: 12/26 00:01:09 35623kN 1...Re8 2.Rxa6 Re4 3.f3 Rc4 4.Re1 Kd7 ± (0.80) Depth: 13/29 00:01:23 42373kN 1...Re8 2.Rxa6 Re4 3.f3 Rc4 4.Re1 Rb2 5.Ra7 Rxc3 6.Rxg7 h5 7.a3 Rxa3 8.Ree7 ± (0.79) Depth: 14/29 00:02:00 62118kN 1...Re8 2.Rxa6 Re4 3.f3 Rc4 4.Rd1+ Ke7 5.Rd2 h5 6.Kf2 Kf7 7.Re2 g5 8.Ra8 Rxc3 9.Rxa4 ± (0.76) Depth: 15/31 00:04:34 141173kN (Blass, Tel-aviv 26.08.2001) Note that I am not surprised from the fact that chess programs fail to see that trading all the rooks is bad for black. Black has 2 advanced pawns in the a file and it is known that chess programs like advanced pawns so they do not see a clear advantage for white in the endgame and believe that winning a pawn is better than trading all the rooks. Uri
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.