Author: Howard Exner
Date: 13:53:55 05/12/98
Go up one level in this thread
On May 12, 1998 at 12:23:37, Kai Skibbe wrote: >On May 12, 1998 at 07:05:53, Howard Exner wrote: > >>On May 12, 1998 at 01:17:08, Kai Skibbe wrote: >> >>>The following position is from a game between >>>Rebel 9.0 28 MB HT and Fritz 5.03 64MB HT on >>>a PC with a AMD K6-200. >>> >>>WKh3, Rh6, Pg3,g2 >>>BKc5, Rb6, Ph7 >>>White to move >>> >>>Rebel 9.0 as white was unable to find the winning >>>exchange Rxb6 in 5 minutes. He played Rxh7 and after >>>some moves the game became a draw. >>>I have tested the same position with Shredder 2.0 64MB >>>and Genius 5.0 32 MB. Both programs didn´t see the winning >>>exchange in 10 minutes and played the same move as >>>Rebel (Rxh7). >>>I´m not sure if the game could be a win for white after >>>Rxh7, but after Rxb6 the win should be clear. >>> >>>Does any chessprogram play the move Rxb6 and see the win ? >>>The evaluation should be at least 4 pawn, if the program >>>has understood the position. >>> >>>Why is this position so difficult for programs ? >> >>Both moves, Rxb6 and Rxh7 will win. In the Rxh7 line the >>black king is too far away and will be cut off by white's >>Rook (Rd7). >> >>Could you post the remainder of the game ? I'd like to see >>how Fritz managed to draw this. I've seen Rebel routinely >>win these kinds of positions when the king is cut off. > >Here is the rest of the game after the position >Rebel 9.0 28MB - Fritz 5.03 64MB >59. Rxh7 Kd5 >60. g4 Ke5 Rebel should not let the king approach. After 59 ... Kd5 60. Rh5+ Ke4 61. g4 and then placing the Rook on f5. This plan effectively cuts the king off horizontally which is often more effective than being cut off vertically. >61. g5 Kf5 >62. Kh4 Rb1 >63. Rf7+ Kg6 >64. Rf6+ Kg7 >65. Rd6 Rh1+ >66. Kg3 Rg1 >67. Rb6 Kf7 >68. Kf3 Rc1 >69. Kg4 Rd1 >70. Rc6 Kg7 >71. Ra6 Rc1 >72. g3 Rf1 >73. Ra7+ Kg6 >74. Ra4 Rd1 >75. Ra6+ Rg7 > >At this point I stopped the game. IMO there is no win for white. > >I have tested the position after 59. Rxb6 Kxb6. From this position >all programs play the correct moves ( 60. Kg4 or g4 ). > >BTW CM 5500 and CM5555 find the move Rxb6 instantly, but the evaluation >is only >about +1.8 pawns. So they both have not seen a promoted pawn.
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.