Author: Christophe Theron
Date: 10:34:14 09/16/02
Go up one level in this thread
On September 15, 2002 at 15:38:25, Dieter Buerssner wrote:
>The following position is from a game Tiger-Yace on ICS. I watched the game, and
>was happy about it first (you will find the game at the end). At some stage, I
>thought, it looks really bad, and the knight of Yace may be lost. Then, there
>was a trade into an endgame, with B vs. N in closed position. Yace had the
>knight side, and it really seemed superior. I wondered, if it could introdude
>into the white position with his K via th Q-side. I feared, that this may be
>over the search horizont, but finally Yace succeeded, and had a winning
>position. It showed score +8 (in favor of black) in this position:
>
>[D] 8/8/4K1p1/4PpB1/3p1P1P/5n2/4k3/8 b - -
>
>Now it moved Nxg5. After this move the position is a draw. It took quite a few
>move, to find this out ...
>
>I analyzed the pos later, and now and then used different chess engines. At
>various points of the winning line, they all seem to fail, and spoil this to a
>draw.
>
>So, I give several test positions. The test should be, to solve them all
>(otherwise it won't win). Actually each position is "avoid Nxg5". If at any
>point, this is played, the game will be draw.
I give results for Chess Tiger 15.0 running on PIII-733, 96Mb hash, all
positions searched for at least 3 minutes.
>8/8/4K1p1/4PpB1/3p1P1P/5n2/4k3/8 b - - bm d3; c0 "Nxg5 only draws";
Tiger 15 plays it instantly, never considers anything else.
>After d3 Kf7
>
>8/5K2/6p1/4PpB1/5P1P/3p1n2/4k3/8 b - - bm d2; c0 "Nxg5 only draws";
Tiger 15 plays it instantly, never considers anything else.
>After d2 e6
>
>8/5K2/4P1p1/5pB1/5P1P/5n2/3pk3/8 b - - bm d1=Q; c0 "Nxg5 only draws";
Tiger 15 plays it instantly, never considers anything else.
>after d1Q e7
>
>8/4PK2/6p1/5pB1/5P1P/5n2/4k3/3q4 b - - bm Qd7; c0 "Nxg5 only draws";
Tiger 15 at first wants to play Nxg5, realizes that it is not a good idea in 17
seconds, and switches to Qd7 in 24 seconds.
>After Qd7 Kf8, black has a nice winning line: Nd4 e8Q+ Ne6+ Qxe6+ Qxe6: three
>checks in a row. More difficult for engines to win seems Qd7 Bf6 Ke3 Kf8:
>
>5K2/3qP3/5Bp1/5p2/5P1P/4kn2/8/8 b - - bm Kxf4;
OK, here the real problems arise.
Tiger 15 wants to play Qd6 at first and needs 5 minutes to realize that this is
drawing (score 1.62). It plays Kxf4 in 5 minutes 12 seconds.
>There are more ways to win, and what can happen here: The black side does not
>want to give his Q for the pawn, and do many Q moves. In the end, it must give
>the Q anyway, and it may very well end in a draw (at least with fast playing, my
>engined did this stupid thing). So, I think, Kxf4 is the only convincing
>immediate move.
>
>The first position after Nxg5.
>[D] 8/8/4K1p1/4Ppn1/3p1P1P/8/4k3/8 w - -
>
>How does your engine evaluate this position? Are there any engines, that see
>something close to draw value (or even say smaller 5 after some time)?
Tiger 15 says hxg5 is -7.76 in 4 seconds and switches to fxg5 in 9 seconds.
Score is -6.06.
After 10 minutes score is -5.98.
Christophe
>After +8 and draw, my excitement was not so big anymore. Especially, after not
>long time earlier, I saw a similar, but less extreme case:
>http://www.f11.parsimony.net/forum16635/messages/34505.htm
>
>The whole game, the first position shown above was after 87. Kxe6
>
>[Event "ICC 15 3"]
>[Site "Internet Chess Club"]
>[Date "2002.09.14"]
>[Round "-"]
>[White "Abraxis"]
>[Black "Yace"]
>[Result "1/2-1/2"]
>[ICCResult "Game drawn by repetition"]
>[WhiteElo "2653"]
>[BlackElo "2586"]
>[Opening "Sicilian"]
>[ECO "B50"]
>[NIC "SI.01"]
>[Time "19:26:42"]
>[TimeControl "900+3"]
>
>1. e4 c5 2. Nf3 d6 3. c3 Nf6 4. Be2 g6 5. O-O Bg7 6. Bb5+ Bd7 7. Bxd7+ Qxd7
>8. Re1 O-O 9. d4 cxd4 10. cxd4 d5 11. e5 Ne4 12. Nbd2 Nxd2 13. Bxd2 Nc6 14.
>Rc1 Rfc8 15. h3 Nd8 16. Qb3 Ne6 17. Be3 a6 18. Rxc8+ Rxc8 19. Rc1 Rxc1+ 20.
>Bxc1 Qc6 21. Be3 Nd8 22. Qb4 Qc7 23. Bg5 Nc6 24. Qb3 e6 25. a3 Na5 26. Qa4
>b5 27. Qb4 Bf8 28. Qd2 Nb3 29. Qd1 Qc4 30. Be3 Be7 31. Nd2 Qc1 32. Qxc1 Nxc1
>33. b4 Kf8 34. Kf1 h5 35. g3 Ke8 36. Ke1 f6 37. f4 f5 38. Kd1 Nd3 39. Kc2
>Ne1+ 40. Kc3 Ng2 41. Bf2 h4 42. Nb3 hxg3 43. Bxg3 Ne3 44. Nc5 Bxc5 45. bxc5
>Kd7 46. Bh4 Kc6 47. Bd8 Ng2 48. Bg5 a5 49. h4 Ne3 50. Bd8 Nc4 51. Kb3 Kd7
>52. Bg5 Nd2+ 53. Kc3 Nf3 54. Bf6 b4+ 55. axb4 axb4+ 56. Kxb4 Nxd4 57. Bg5
>Ne2 58. Kb5 Kc7 59. c6 Nd4+ 60. Kc5 Nxc6 61. Bf6 Kd7 62. Bg7 Ke8 63. Bh6 Na5
>64. Kd4 Nb3+ 65. Kc3 Nc5 66. Kd4 Ne4 67. Ke3 Kd7 68. Bf8 Kc6 69. Kd4 Kc7 70.
>Ba3 Kb6 71. Kd3 Ka5 72. Bf8 Nf2+ 73. Kc3 Nh3 74. Bh6 Kb5 75. Kd4 Kb4 76. Bg5
>Kb3 77. Kd3 Nf2+ 78. Ke3 Kc2 79. Bd8 Ng4+ 80. Kd4 Kd2 81. Be7 Ne3 82. Bd8
>Ng2 83. Be7 Ke2 84. Bg5 Ne1 85. Kc5 Nf3 86. Kd6 d4 87. Kxe6 Nxg5+ 88. fxg5
>d3 89. Kf7 d2 90. e6 d1=Q 91. e7 Qd5+ 92. Kf8 Qc5 93. Kf7 Qc7 94. Kf8 Qd6
>95. Kf7 Qd7 96. Kf8 Kf3 97. h5 Qd6 98. h6 Kg4 99. h7 Qh2 100. Kg7 Qb8 101.
>Kf7 Qh8 102. e8=Q Qxh7+ 103. Kf6 Qh2 104. Qd7 Qh8+ 105. Kxg6 Qh5+ 106. Kf6
>Qxg5+ 107. Ke5 Qe3+ 108. Kf6 Qc3+ 109. Kg6 Qe5 110. Qd1+ Kf4 111. Qf1+ Ke4
>112. Qb1+ Kf4 113. Qf1+ Ke4 114. Qb1+ Kf4 115. Qf1+ {Game drawn by
>repetition} 1/2-1/2
>
>Regards,
>Dieter
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.