Author: Drexel,Michael
Date: 12:08:44 01/22/04
Go up one level in this thread
On January 22, 2004 at 12:03:35, Vincent Diepeveen wrote: >On January 21, 2004 at 17:10:20, Drexel,Michael wrote: > >>On January 21, 2004 at 17:02:30, Mike Hood wrote: >> >>>Here's a small example of "perfect play" using the KRPP-KR tablebase. It's not >>>the only path that leads to mate, there are a few possible branches along the >>>way, but there are no shorter paths. How many engines playing black without >>>6-piece EGTBs could get a win against an engine playing white with 6-piece >>>EGTBs? Not many, if any. Take a look >>> >>>[D] 3R4/6k1/8/6p1/7p/2r5/7K/8 b - - 0 63 >>> >>>63...Kg6 64. Rh8 Kf5 >>>65. Rf8+ Kg4 66. Rd8 Rc2+ 67. Kg1 h3 68. Rd4+ Kg3 69. Rd3+ Kh4 70. Rd8 g4 71. >>>Rd4 Kg3 72. Rd3+ Kf4 73. Rd4+ Kf3 74. Rd3+ Ke4 75. Rd8 g3 76. Re8+ Kf4 77. Rf8+ >>>Ke5 78. Re8+ Kf6 79. Re1 Kg5 80. Rd1 Rf2 81. Rd5+ Kg4 82. Rd4+ Rf4 83. Rd2 h2+ >>>84. Kg2 Rf2+ 85. Rxf2 gxf2 86. Kxf2 h1=Q 87. Ke3 Qd5 88. Ke2 Kf4 89. Kf1 Qa2 >>>90. Ke1 Ke3 91. Kd1 Qb1# >> >>That is a basic endgame. I would win that in my sleep. >> >>Michael > >Sure, i drew KRPP KR with the KR side 2 times in 2002 in a 120 0 game in my >club. One time opponent had pawns on e2,f3 the other time there were pawns on f >and g. So connected passers. Not f + h, they all know that too well to be a >draw. > >One opponent has a rating of 2220 (Robert Beekman) the other is FM (Pieter >Nieuwenhuis). > >Neither of them was in time trouble... It's still a basic endgame. You can find it in most endgame books. The win is easy if you know the rules for this ending: 1. Support the 2 connected passers with King and Rook 2. Keep your rook active 3. Do not allow the opponents King to safely block those pawns The example position is a dream position for black. > >KRPP KR is a very useful EGTB. I didn't question that. > >Of course guys like Haworth & co are too bad chessplayers to select interesting >positions from such EGTBs, can't blame them. The conclusion is trivial however >that chessprograms do a better job there with connected passers than humans, >because the way to draw it is like a 10 ply combination which even in blitz >happens. However the problems computers have is with for example f + h pawns. > >I'm sure most programs say it's +3.0 or so but in reality it's a simple draw. That's wrong. The theory is rather complicated. See below... [Event "Amsterdam"] [Site "Amsterdam"] [Date "1981.??.??"] [Round "?"] [White "Polugaevsky, Lev"] [Black "Ree, Hans"] [Result "0-1"] [ECO "E11"] [WhiteElo "2620"] [BlackElo "2485"] [PlyCount "148"] [EventDate "1981.05.??"] [Source ""] 1. d4 e6 2. c4 Nf6 3. Nf3 Bb4+ 4. Bd2 Qe7 5. g3 O-O 6. Bg2 Bxd2+ 7. Qxd2 Ne4 8. Qc2 f5 9. O-O d6 10. Nh4 g5 11. Nf3 Nd7 12. Ne1 Nef6 13. Nc3 Ng4 14. Nf3 Qg7 15. e4 f4 16. Nb5 fxg3 17. hxg3 Nb8 18. e5 a6 19. Na3 dxe5 20. Qe4 h5 21. Nxe5 Nxe5 22. Qxe5 Qxe5 23. dxe5 Nd7 24. f4 h4 25. Bh3 hxg3 26. Bxe6+ Kg7 27. Rad1 Nxe5 28. Bxc8 gxf4 29. Bxb7 Rab8 30. Be4 Rxb2 31. Rd5 f3 32. Nc2 g2 33. Rxe5 gxf1=Q+ 34. Kxf1 Rxa2 35. Kf2 Rf4 36. Bd3 Rxc4 37. Bxc4 Rxc2+ 38. Kxf3 Rxc4 39. Ra5 Rc6 40. Ke4 Kf6 41. Kd5 Rd6+ 42. Kc5 Ke6 43. Ra1 Kd7 44. Rh1 Rc6+ 45. Kb4 Kc8 46. Ka5 Rb6 47. Rh5 Kb7 48. Rg5 Rb1 49. Rh5 Rd1 50. Kb4 Kb6 51. Rh6+ c6 52. Rh8 a5+ 53. Kc4 Rc1+ 54. Kb3 Rg1 55. Rh5 Rd1 56. Kc4 Rc1+ 57. Kb3 Rc5 58. Rh8 Rg5 59. Kc4 Rg4+ 60. Kb3 Kb5 61. Rh5+ c5 62. Rh8 a4+ 63. Ka3 Rg3+ 64. Kb2 Kb4 65. Rb8+ Kc4 66. Ra8 Rg2+ 67. Ka3 Kc3 68. Rxa4 c4 69. Ra8 Rg7 70. Ka2 Kc2 71. Ka1 c3 72. Ka2 Rb7 73. Ra6 Rd7 74. Ra8 Kd2 0-1 Michael > >The interesting positions are not the 6 men themselves, but exchanging *into* >them. > >That's why win/draw/loss egtbs are very useful for those who do not have 2 >terabyte of storage space for the 6 men but only a few gigabyte.
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.