special 7-men tablebase

[bbfreak333]

hi

i play a little correspondece chess and i urgently need a special 7-men tablebase

Rook + 2 pawns vs 2 bishops

can this be generated or downloaded ? thanks very much


regards
gernot

[Codeman]

There are currently no full 7men available, neither for generation nor for download.

The only thing you might want to try is the program Freezer by Eiko Bleicher. Especially if the two pawns are fixed this might do the trick.

regards,
Edmund

[kronsteen]

Problem reduction with two pawns fixed has only to do with pawns of opposite colors – which is not the case there – hasn’t it ?

[Codeman]

Problem reduction with two pawns fixed has only to do with pawns of opposite colors – which is not the case there – hasn’t it ?

right, I missed that. So the only sort of reduction that can be applied is to restrict the bishops to their color-squares and the pawns to their files.

Gernot, can you post the position?

[bbfreak333]

http://img101.imageshack.us/img101/7171/correspond.jpg

black to move. i have white. i try to understand what you mean with the reduction thing
ty so far

[Codeman]

http://img101.imageshack.us/img101/7171/correspond.jpg

black to move. i have white. i try to understand what you mean with the reduction thing
ty so far

regarding the reduction; if you have a 7 men position, the full file without any optimizations would have to store all possible combinations of pieces on the board. That is 64^7 positions, with 1byte per position that would take 4TB to store.
now you have certain information that can simplify the process. Eg. the two bishops can never be on the save color square and the pawns are also very restricted.

[Codeman]

http://img101.imageshack.us/img101/7171/correspond.jpg

black to move. i have white. i try to understand what you mean with the reduction thing
ty so far

The position has still too many possibilities to be tackled with egtbs. Maybe you want to try IDEA or Montecarlo analysis.

[mbourzut]

I started generating 7-man tablebases for the endings rook + 2 pawns vs. 2 minor pieces some time ago, with the constraint of promotions only to queen to avoid having to generate tons of sub-games. This case here with 2 bishops is almost done. The longest DTC is 107. I should be able to post some analysis here in the next few days.

The case with 2 knights is also interesting and arises occasionally in practice, giving people considerable trouble.

[bbfreak333]

i cannot use montecarlo because i only have rybkaengine with chessbase

i dont know how to use IDEA, i find it useless anyway. it takes ages and the moves it suggests are losing in that particular game (correspondence) in 15 moves forced.....

[mbourzut]

I started generating 7-man tablebases for the endings rook + 2 pawns vs. 2 minor pieces some time ago, with the constraint of promotions only to queen to avoid having to generate tons of sub-games. This case here with 2 bishops is almost done. The longest DTC is 107. I should be able to post some analysis here in the next few days.

The case with 2 knights is also interesting and arises occasionally in practice, giving people considerable trouble.

Well, the database is done and Black can draw this ending with any reasonable move.

[mbourzut]

I started generating 7-man tablebases for the endings rook + 2 pawns vs. 2 minor pieces some time ago, with the constraint of promotions only to queen to avoid having to generate tons of sub-games. This case here with 2 bishops is almost done. The longest DTC is 107. I should be able to post some analysis here in the next few days.

The case with 2 knights is also interesting and arises occasionally in practice, giving people considerable trouble.

Well, the database is done and Black can draw this ending with any reasonable move.

Given that this is an ongoing game, I'm not going to post any specific analysis here to help either the attacker or defender. It is interesting that White wins with the pawns on g3 and f4 instead of h3 and g4.

To give some ideas about the complexity of this type of ending, here is one of the longest DTC lines:

[Event "W +107"]
[Date "????.??.??"]
[Site "?"]
[Round "?"]
[White "RPP"]
[Black "BB-11"]
[Result "1-0"]
[SetUp "1"]
[FEN "8/7k/b7/8/1b6/8/3PP3/1K4R1 w - - 0 1"]

1. d3! Bb5 2. Rg5! Bd7 3. Rd5! Be8 4. Re5! Bd7 5. Re4 Bf8 6. Rd4! Be6 7. Rd8!
Bb4 8. Re8! Bf7 9. Rb8! Bd6 10. Rb5 Bh2 11. Kc2! Kg7 12. Rb6! Bg1 13. Rc6 Be3
14. Kc3! Bg6 15. Re6! Bf2 16. e4 Be1+ 17. Kc2 Bf2 18. Kd2 Kh6 19. Rf6 Bd4 20.
Rd6! Bb2 21. Rb6 Bd4 22. Rb4 Bf2 23. Ke2! Bg1 24. Rb1! Bd4 25. Rb5 Kg7 26. Kf3!
Kf6 27. Kf4! Bg1 28. Ra5 Bh2+ 29. Kf3 Bg1 30. Ra1 Bh5+ 31. Kf4! Bh2+ 32. Ke3
Kg5 33. Kd4! Kg4 34. Kc5! Bg6 35. Ra6! Kg5 36. Kd4 Bg1+ 37. Ke5! Bf7 38. Ra8!
Bb6 39. Rc8 Be3 40. Rf8 Bf4+ 41. Kd4 Be6 42. Kc3 Ba2 43. Ra8 Bf7 44. Ra7 Bg6
45. Kd4! Bg3 46. Ra8! Bf2+ 47. Ke5! Bf7 48. Rf8! Bg3+ 49. Kd4 Bf2+ 50. Kc3 Be1+
51. Kc2 Kg6 52. Rb8 Bf2 53. Rb2 Bd4 54. Rb4! Bc5 55. Ra4 Be3 56. Kc3 Kf6 57.
Ra6+! Kg5 58. Ra1! Bg6 59. Rf1! Bh7 60. Rf3 Bf4 61. Rf2 Bb8 62. Rf1 Bg6 63. Kd4
Bf4 64. Kc4 Bf7+ 65. Kc3 Bg6 66. Rg1+ Kf6 67. Rg4 Bg5 68. d4 Bh7 69. Kd3 Bg6
70. Rg3 Bf4 71. Rg1 Bh7 72. Re1 Bg6 73. Kc4 Bf7+ 74. Kc5 Ke7 75. Rf1 Bd6+ 76.
Kc6 Ba3 77. Rf3 Be8+ 78. Kb6 Bc1 79. Kc5 Bh5 80. Rh3 Bg4 81. Rh1 Ba3+ 82. Kc4
Kf6 83. Kd3 Be7 84. Ke3 Bd7 85. Rh6+ Kg7 86. Rb6 Bg5+ 87. Kd3 Kf7 88. e5 Bf5+
89. Kc4 Bd8 90. Rc6 Bh4 91. d5 Kg7 92. Rc7+ Kg6 93. e6 Kf6 94. e7! Kf7 95. d6!
Ke8 96. Kd5! Bd7 97. Rc4 Bg5 98. Re4 Bc1 99. Re2 Bh6 100. Kc5 Bf4 101. Rg2 Be3+
102. Kd5 Kf7 103. Rh2 Be6+ 104. Kc6 Bg4 105. Rg2 Bf3+ 106. Kd7 Bc6+ 107. Kxc6
1-0

[Codeman]

I started generating 7-man tablebases for the endings rook + 2 pawns vs. 2 minor pieces some time ago, with the constraint of promotions only to queen to avoid having to generate tons of sub-games. This case here with 2 bishops is almost done. The longest DTC is 107. I should be able to post some analysis here in the next few days.

The case with 2 knights is also interesting and arises occasionally in practice, giving people considerable trouble.

Well, the database is done and Black can draw this ending with any reasonable move.

Given that this is an ongoing game, I'm not going to post any specific analysis here to help either the attacker or defender. It is interesting that White wins with the pawns on g3 and f4 instead of h3 and g4.

To give some ideas about the complexity of this type of ending, here is one of the longest DTC lines:

[Event "W +107"]
[Date "????.??.??"]
[Site "?"]
[Round "?"]
[White "RPP"]
[Black "BB-11"]
[Result "1-0"]
[SetUp "1"]
[FEN "8/7k/b7/8/1b6/8/3PP3/1K4R1 w - - 0 1"]

1. d3! Bb5 2. Rg5! Bd7 3. Rd5! Be8 4. Re5! Bd7 5. Re4 Bf8 6. Rd4! Be6 7. Rd8!
Bb4 8. Re8! Bf7 9. Rb8! Bd6 10. Rb5 Bh2 11. Kc2! Kg7 12. Rb6! Bg1 13. Rc6 Be3
14. Kc3! Bg6 15. Re6! Bf2 16. e4 Be1+ 17. Kc2 Bf2 18. Kd2 Kh6 19. Rf6 Bd4 20.
Rd6! Bb2 21. Rb6 Bd4 22. Rb4 Bf2 23. Ke2! Bg1 24. Rb1! Bd4 25. Rb5 Kg7 26. Kf3!
Kf6 27. Kf4! Bg1 28. Ra5 Bh2+ 29. Kf3 Bg1 30. Ra1 Bh5+ 31. Kf4! Bh2+ 32. Ke3
Kg5 33. Kd4! Kg4 34. Kc5! Bg6 35. Ra6! Kg5 36. Kd4 Bg1+ 37. Ke5! Bf7 38. Ra8!
Bb6 39. Rc8 Be3 40. Rf8 Bf4+ 41. Kd4 Be6 42. Kc3 Ba2 43. Ra8 Bf7 44. Ra7 Bg6
45. Kd4! Bg3 46. Ra8! Bf2+ 47. Ke5! Bf7 48. Rf8! Bg3+ 49. Kd4 Bf2+ 50. Kc3 Be1+
51. Kc2 Kg6 52. Rb8 Bf2 53. Rb2 Bd4 54. Rb4! Bc5 55. Ra4 Be3 56. Kc3 Kf6 57.
Ra6+! Kg5 58. Ra1! Bg6 59. Rf1! Bh7 60. Rf3 Bf4 61. Rf2 Bb8 62. Rf1 Bg6 63. Kd4
Bf4 64. Kc4 Bf7+ 65. Kc3 Bg6 66. Rg1+ Kf6 67. Rg4 Bg5 68. d4 Bh7 69. Kd3 Bg6
70. Rg3 Bf4 71. Rg1 Bh7 72. Re1 Bg6 73. Kc4 Bf7+ 74. Kc5 Ke7 75. Rf1 Bd6+ 76.
Kc6 Ba3 77. Rf3 Be8+ 78. Kb6 Bc1 79. Kc5 Bh5 80. Rh3 Bg4 81. Rh1 Ba3+ 82. Kc4
Kf6 83. Kd3 Be7 84. Ke3 Bd7 85. Rh6+ Kg7 86. Rb6 Bg5+ 87. Kd3 Kf7 88. e5 Bf5+
89. Kc4 Bd8 90. Rc6 Bh4 91. d5 Kg7 92. Rc7+ Kg6 93. e6 Kf6 94. e7! Kf7 95. d6!
Ke8 96. Kd5! Bd7 97. Rc4 Bg5 98. Re4 Bc1 99. Re2 Bh6 100. Kc5 Bf4 101. Rg2 Be3+
102. Kd5 Kf7 103. Rh2 Be6+ 104. Kc6 Bg4 105. Rg2 Bf3+ 106. Kd7 Bc6+ 107. Kxc6
1-0

Thanks for that line, that was really interesting! I wouldn't have imagined such a complexity in this type of endgame. Anyway its great to see you are back to 7men-generation!

[Les]

I started generating 7-man tablebases for the endings rook + 2 pawns vs. 2 minor pieces some time ago, with the constraint of promotions only to queen to avoid having to generate tons of sub-games. This case here with 2 bishops is almost done. The longest DTC is 107. I should be able to post some analysis here in the next few days.

Hi mbourzut,

Quick question. By any chance have you generated the 5-1 complete set of the 6 piece EGTB? Those are the last tablebases I need for a project I am working on. If you did not generate them is there any chance you can generate them without much trouble? I am not really interested in the actual tables themselves but desperatly need the stats for WIN-DRAW-LOSE like the stat files Eugenes generator use to produce.

Thanks in advance,

Les

The case with 2 knights is also interesting and arises occasionally in practice, giving people considerable trouble.

[jshriver]

http://img101.imageshack.us/img101/7171/correspond.jpg

black to move. i have white. i try to understand what you mean with the reduction thing
ty so far

Sorry to be OT, but there aren't any publically available 7men egtb's, and what little exists hasn't really been updated in several years.

However, I would absolutely love to get into correspondence chess (old snail mail, or even email). How did you get into it, or more importantly find someone to play?

-Josh