Author: Rafael Andrist
Date: 06:54:52 05/28/01
Go up one level in this thread
In your postion Uri, Be7 is not a legal move, but if you mean move No 3 of the solution, my program finds this move (with a draw score because I stopped search at some point). The PV does not correlate with the solution, it prefers other moves for black, not seeing they lead to a quicker mate. I hope you can read this output from my program for move No 3. Note that the PV is taken from hash and at low depths the hash PV does not start with the move suggested by the search. 1 ply Ld8-e7 N: 6 Q: 2 H: 3 value: -544 Kb1-a2 2 ply Ld8-e7 N: 25 Q: 10 H: 22 value: -552 Kb1-a2 b5-b4 3 ply Kb1-a2 N: 79 Q: 30 H: 73 value: -548 Kb1-a2 b5-b4 Ka2-b1 4 ply Kb1-a2 N: 138 Q: 45 H: 119 value: -574 Kb1-a2 b5-b4 Ka2-b1 e6-e5 5 ply Kb1-a2 N: 260 Q: 96 H: 221 value: -574 Kb1-a2 b5-b4 Ka2-b1 e6-e5 Kb1-a2 6 ply Kb1-a2 N: 442 Q: 222 H: 354 value: -574 Kb1-a2 b5-b4 Ka2-b1 e6-e5 Kb1-a2 e5-e4 7 ply Kb1-a2 N: 730 Q: 251 H: 542 value: -574 Kb1-a2 b5-b4 Ka2-b1 e6-e5 Kb1-a2 Kb8-a8 Ka2-b1 8 ply Ld8-e7 N: 1836 Q: 729 H: 1065 value: -575 Ld8-e7 b5-b4 Le7-f8 Kb8-a8 Lf8-h6 e6-e5 Lh6-g7 b4-b3 9 ply Ld8-c7+ N: 4308 Q: 1181 H: 2426 value: -577 Ld8-c7+ Kb8-a8 Lc7-d8 b5-b4 Ld8-e7 b4-b3 Le7-f8 e6-e5 Lf8-h6 10 ply Ld8-c7+ N: 5279 Q: 1424 H: 3120 value: -577 Ld8-c7+ Kb8-a8 Lc7-d8 b5-b4 Ld8-e7 b4-b3 Le7-f8 e6-e5 Lf8-h6 e5-e4 11 ply Ld8-c7+ N: 10589 Q: 1680 H: 5663 value: -577 Ld8-c7+ Kb8-a8 Lc7-d8 b5-b4 Ld8-e7 b4-b3 Le7-f8 e6-e5 Lf8-h6 e5-e4 Lh6-g7 12 ply Ld8-c7+ N: 13139 Q: 2300 H: 7454 value: -577 Ld8-c7+ Kb8-a8 Lc7-d8 b5-b4 Ld8-e7 b4-b3 Le7-f8 e6-e5 Lf8-h6 e5-e4 Lh6-g7 h7-h6 13 ply Ld8-c7+ N: 35311 Q: 3021 H: 19532 value: -607 Ld8-c7+ Kb8-a8 Lc7-d8 b5-b4 Ld8-e7 b4-b3 Le7-f8 e6-e5 Lf8-g7 h7-h6 g5xh6 e5-e4 h6-h7 14 ply Ld8-e7 N: 81029 Q: 9946 H: 39653 value: -576 Ld8-e7 b5-b4 Le7-f8 e6-e5 Lf8-h6 Kb8-a8 g5-g6 h7xg6 Lh6xe3 b4-b3 Le3-c5 g6-g5 Lc5-a3 15 ply Ld8-e7 N: 156547 Q: 14148 H: 81130 value: -569 Ld8-e7 b5-b4 Le7-f8 e6-e5 Lf8-h6 Kb8-a8 Lh6-f8 b4-b3 Lf8-h6 Ka8-b8 g5-g6 h7xg6 Lh6-f4 e5-e4 Lf4xe3 16 ply Ld8-e7 N: 221363 Q: 28905 H: 120420 value: -528 Ld8-e7 b5-b4 Le7-f8 e6-e5 Lf8-h6 Kb8-a8 g5-g6 h7xg6 Lh6xe3 b4-b3 Le3-c5 g6-g5 Lc5-a3 17 ply Ld8-e7 N: 501964 Q: 54707 H: 253643 value: 0 Ld8-e7 b5-b4 Le7-f8 e6-e5 Lf8-h6 Kb8-a8 Lh6-f8 b4-b3 Lf8-e7 Ka8-b8 Le7-d8 Kb8-a8 Ld8-c7 e5-e4 Lc7-d8 Ka8-b8 Ld8-e7 On May 27, 2001 at 15:04:42, Uri Blass wrote: >On May 27, 2001 at 12:57:57, Rafael Andrist wrote: > >>Why sould this position be difficult? The move Kb1 is practically forced. My >>program finds Kb1 immediately and at ply 18 it sees that White can at least >>equalize. Its clear that it does not see the mate, because it is too far away. > >The test is not to find Kb1 but to win the game. >Can your program win the game with white. > >I suspect that it is going to do a draw by repetition by 1.Kb1 e3 2.Bd8 Kb8 >3.Bc7+ Ka8 4.Bd8 Kb8 5.Bc7+ Ka8 > >If it cannot avoid 3.Bc7+ or 5.Bc7+ then I do not consider the position solved. > >The hard test is to find Be7 in the following position. > >[D]k1b5/1pBp1p1p/1P1PpP2/1p4P1/8/4p3/1p2P3/bK6 w - - 0 1 > >Be7 is a simple draw but for some reason programs prefer check. > >I believe that the problem can be solved practically if the evaluation function >of the program gives a bigger score for a longer draw. > >It is going to push it to try longer lines that lead to a draw and it can push >it to find the right moves. > >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.