Author: Dan Homan
Date: 10:07:13 07/16/98
Go up one level in this thread
On July 16, 1998 at 12:08:58, blass uri wrote: > >On July 16, 1998 at 10:35:17, Dan Homan wrote: > >>On July 15, 1998 at 07:35:10, blass uri wrote: >> >>> >>>On July 14, 1998 at 11:59:33, blass uri wrote: >>> >>>>the position >>>>7r/1k2P2P/4p3/1pp5/8/2pp4/1n3PP1/R4K1R >>>>white to play >>>>white should play g4 >>>>but computers like e8=Queen Rxe8 h8=Queen Rxh8 Rxh8 and >>>>I think that black is winning instead of losing >>>> >>>>can a computer program find g4 in a reasonable time >>>>or all of them are stupid and do not understand that the white pawns h7 g6 are >>>>stronger than a rook >>>> >>>>Uri >>>Junior needed near 10 hours to find g4 >>>I think fritz is not better >>>computer programs do not understand that knight and passed pawns are >>>better than a 2 rooks >>> >> >>Here is the output of my amateur program, EXchess. It finds g4 in about >>6 seconds on a slow machine. Although the score stays negative for quite >>some time, it does steadily increase. I have put very little work into the >>passed pawn code, so I am not sure if finding this move is more than just >>dumb luck. >> >> - Dan >> >>White-To-Move[1]: setboard 7r/1k2P2P/4p3/1pp5/8/2pp4/1n3PP1/R4K1R w >>White-To-Move[1]: display >> >> +---+---+---+---+---+---+---+---+ >>8 | | | | | | | || White to move >> +---+---+---+---+---+---+---+---+ >>7 | || | | P | | | P | castle: - >> +---+---+---+---+---+---+---+---+ >>6 | | | | || | | | >> +---+---+---+---+---+---+---+---+ >>5 | ||| | | | | | >> +---+---+---+---+---+---+---+---+ >>4 | | | | | | | | | fifty: 0 >> +---+---+---+---+---+---+---+---+ >>3 | | ||| | | | | Computer time: 30 seconds >> +---+---+---+---+---+---+---+---+ >>2 | || | | | P | P | | >> +---+---+---+---+---+---+---+---+ >>1 | R | | | | | K | | R | >> +---+---+---+---+---+---+---+---+ >> a b c d e f g h >> >>White-To-Move[1]: analyze >>Analysis mode: Enter commands/moves as ready. >> >> 1. ++ 0 2 1. e8=Q >> 1. -556 0 14 1. g3 >> 1. -499 0 15 1. g4 >> 1. -499 0 36 1. g4 >> 2. -- >> 2. -546 0 425 1. g4 b4 >> 2. -540 0 706 1. Re1 Kc6 >> 2. -540 0 763 1. Re1 Kc6 >> 3. ++ 0 859 1. Re1 >> 3. -493 0 1163 1. e8=Q Rxe8 2. h8=Q >> 3. -493 1 2173 1. e8=Q Rxe8 2. h8=Q >> 4. ++ 1 2343 1. e8=Q >> 4. -468 1 2463 1. e8=Q Rxe8 2. Ra7+ Kxa7 >> 4. -453 1 10558 1. Ra3 c2 2. Rc3 Kc6 3. g4 >> 4. -453 1 12178 1. Ra3 c2 2. Rc3 Kc6 3. g4 >> 5. -477 1 17598 1. e8=Q Rxe8 2. Ra8 Rxa8 >> 5. -476 1 24848 1. Rh3 b4 2. e8=Q Rxe8 3. Ra7+ Kxa7 >> 5. -444 2 35082 1. g4 b4 2. e8=Q Rxe8 3. Ra7+ Kxa7 >> 5. -443 2 43613 1. Rh4 b4 2. e8=Q Rxe8 3. Ra7+ Kxa7 >> 5. -442 2 53415 1. Kg1 d2 2. e8=Q Rxe8 3. Ra7+ Kxa7 >> 5. -442 2 58118 1. Kg1 d2 2. e8=Q Rxe8 3. Ra7+ Kxa7 >> 6. -459 3 97429 1. Rc1 d2 2. Ra1 d1=Q+ 3. Rxd1 Nxd1 4. Ke2 Nb2 >> 6. -448 5 156222 1. Rh3 b4 2. Rh4 Kb8 3. e8=Q+ Rxe8 4. Ra8+ Kxa8 >> 6. ++ 6 194502 1. g4 >> 6. -409 7 227112 1. g4 Nc4 2. e8=Q Rxe8 3. h8=Q Rxh8 >> 6. -409 8 289357 1. g4 Nc4 2. e8=Q Rxe8 3. h8=Q Rxh8 >> 7. -406 13 447301 1. g4 b4 2. g5 Kb6 3. e8=Q Rxe8 4. h8=Q Rxh8 >> 7. -406 20 705727 1. g4 b4 2. g5 Kb6 3. e8=Q Rxe8 4. h8=Q Rxh8 >> 8. -383 47 1628965 1. g4 Nc4 2. Rh3 Nd2+ 3. Ke1 c4 4. e8=Q Rxe8 >> 5. h8=Q Rxh8 >> 8. -383 130 4602417 1. g4 Nc4 2. Rh3 Nd2+ 3. Ke1 c4 4. e8=Q Rxe8 >> 5. h8=Q Rxh8 >> 9. ++ 149 5327119 1. g4 >> 9. -- >> 9. -336 773 28096190 1. g4 c2 2. e8=Q Rxe8 3. h8=Q Rxh8 4. Rxh8 Kc6 >> 5. Ra6+ Kd5 6. g5 >> 9. -336 773 28124503 1. g4 c2 2. e8=Q Rxe8 3. h8=Q Rxh8 4. Rxh8 Kc6 >> 5. Ra6+ Kd5 6. g5 >> 10. -315 1196 43034481 1. g4 c2 2. e8=Q Rxe8 3. h8=Q Rxh8 4. Rxh8 b4 >> 5. Ke1 Kc6 >> 10. -315 1868 68186566 1. g4 c2 2. e8=Q Rxe8 3. h8=Q Rxh8 4. Rxh8 b4 >> 5. Ke1 Kc6 >> 11. ++ 2838 102232936 1. g4 >> 11. -259 5603 205077307 1. g4 Nc4 2. Rh3 Nd2+ 3. Kg2 c4 4. Rh5 c2 >> 5. Rg5 b4 6. Rg8 c1=Q 7. Rxc1 Rxh7 >> 11. -259 5785 212303465 1. g4 Nc4 2. Rh3 Nd2+ 3. Kg2 c4 4. Rh5 c2 >> 5. Rg5 b4 6. Rg8 c1=Q 7. Rxc1 Rxh7 >> >> >>>I believe that most of humans with 2000 elo will not miss g4 >>> >>>Uri >how did you get negative evaluation? >if you count material white has the advantage > >my programs has positive evaluation after e8=Q because they evaluate 2 rooks as >better than knight and advanced passed pawns. Black's two passed pawns are very scary (for my program). I have a 4 pawn penalty if the opponent has connected passed pawns that close to queening. Perhaps it is too strong a penalty, but it seems to help over the board... - Dan > >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.