Hard testposition from TCEC

Discussion of anything and everything relating to chess playing software and machines.

Moderators: hgm, Harvey Williamson, bob

Forum rules
This textbox is used to restore diagrams posted with the [d] tag before the upgrade.
Jouni
Posts: 1949
Joined: Wed Mar 08, 2006 7:15 pm

Hard testposition from TCEC

Post by Jouni » Sat Jan 19, 2019 1:56 pm



SF played 22.g5 and draw (Lc0 to superfinal) , but 22.c5 seems to be winning:

22. c5 Rf6 {4} 23. Qg8+ {2} Nf8 {13} 24. c6 {41} Bxc6 {10} 25. Be5 {6} Rd6 {7}
(25... Bd6 26. Bxd6 {12} Rxd6 {3} 27. Bf3 {6} Bxf3 {3} 28. Re1+ {2} Qe7 {3} 29.
Rxe7+ {1}) 26. Bxd6 {7} Bxd6 {6} * 1 - 0 ??

Too hard for Houdini and Komodo, only SF seems to understand with slow PC.
Jouni

User avatar
Eduard
Posts: 160
Joined: Fri Oct 26, 2018 10:58 pm
Location: Germany
Full name: Eduard Nemeth
Contact:

Re: Hard testposition from TCEC

Post by Eduard » Sat Jan 19, 2019 4:27 pm

Seems to win. But Leela could defend better with 21...Nf8, right?

User avatar
Eduard
Posts: 160
Joined: Fri Oct 26, 2018 10:58 pm
Location: Germany
Full name: Eduard Nemeth
Contact:

Re: Hard testposition from TCEC

Post by Eduard » Sat Jan 19, 2019 4:35 pm

Full pgn:

[Event "TCEC Season 14 - Division P"]
[Site "http://tcec.chessdom.com"]
[Date "2019.01.16"]
[Round "42.4"]
[White "Stockfish 181224"]
[Black "LCZero v20rc2-32194"]
[Result "1/2-1/2"]
[TimeControl "5400+10"]
[BlackElo "3297"]
[ECO "B90"]
[GameDuration "03:27:30"]
[GameEndTime "2019-01-16T17:08:51.839 W. Europe Standard Time"]
[GameStartTime "2019-01-16T13:41:20.929 W. Europe Standard Time"]
[Opening "Sicilian"]
[PlyCount "185"]
[Termination "adjudication"]
[TerminationDetails "TCEC draw rule"]
[Variation "Najdorf, Byrne (English) attack"]
[WhiteElo "3612"]

1. e4 c5 2. Nf3 d6 3. d4 cxd4 4. Nxd4 Nf6 5. Nc3 a6 6. Be3 e6 7. Qf3 Nbd7 8.
Be2 Qc7 9. a4 b6 10. g4 d5 11. exd5 Bb7 12. Bf4 Qc8 13. O-O Bc5 14. Qe3 Nxd5
15. Nxd5 Bxd5 16. b4 Bxb4 17. c4 Bb7 18. Nxe6 fxe6 19. Qxe6+ Be7 20. Bd6 Qd8
21. Rfd1 Rf8 22. g5 Rf7 23. Bh5 g6 24. Bg4 Kf8 25. Rab1 a5 26. Bg3 Ba3 27. Bd6+
Bxd6 28. Qxd6+ Qe7 29. Qxe7+ Rxe7 30. Bxd7 Bf3 31. Rd6 Rd8 32. Rf6+ Kg7 33. Bc6
Bxc6 34. Rxc6 Rf7 35. Rbxb6 Rd2 36. Rf6 Rxf6 37. Rxf6 Ra2 38. Kg2 Rxa4 39. Rc6
Rb4 40. f3 a4 41. h3 a3 42. Ra6 Rxc4 43. Rxa3 Rc7 44. Ra5 h6 45. h4 hxg5 46.
hxg5 Rb7 47. Kg3 Rb4 48. f4 Rc4 49. Ra7+ Kg8 50. Ra6 Kg7 51. Ra7+ Kg8 52. Ra6
Kg7 53. Rf6 Ra4 54. Kf2 Rc4 55. Ke3 Ra4 56. Rc6 Rb4 57. Rc7+ Kg8 58. Rc6 Kg7
59. Rc7+ Kg8 60. Rc8+ Kg7 61. Rc6 Ra4 62. Rc7+ Kg8 63. Rc8+ Kg7 64. Rc7+ Kg8
65. Rc6 Kg7 66. Rd6 Ra5 67. Rd7+ Kg8 68. Rd6 Kg7 69. Rd7+ Kg8 70. Rb7 Ra8 71.
Rb6 Kg7 72. Re6 Ra5 73. Re7+ Kf8 74. Rc7 Kg8 75. Rc6 Kg7 76. Rd6 Ra1 77. Rd7+
Kg8 78. Kf3 Ra5 79. Rd8+ Kg7 80. Rd7+ Kg8 81. Rb7 Ra1 82. Re7 Rc1 83. Re4 Kg7
84. Ra4 Rc7 85. Ra6 Rb7 86. Ra3 Rb5 87. Ra7+ Kg8 88. Ra4 Kg7 89. Ke4 Rb7 90.
Kd4 Rf7 91. Kc3 Rf5 92. Kd3 Rf7 93. Re4 1/2-1/2

zullil
Posts: 5133
Joined: Mon Jan 08, 2007 11:31 pm
Location: PA USA
Full name: Louis Zulli

Re: Hard testposition from TCEC

Post by zullil » Sat Jan 19, 2019 4:47 pm

Eduard wrote:
Sat Jan 19, 2019 4:35 pm
Full pgn:


zullil
Posts: 5133
Joined: Mon Jan 08, 2007 11:31 pm
Location: PA USA
Full name: Louis Zulli

Re: Hard testposition from TCEC

Post by zullil » Sat Jan 19, 2019 4:50 pm

Jouni wrote:
Sat Jan 19, 2019 1:56 pm


SF played 22.g5 and draw (Lc0 to superfinal) , but 22.c5 seems to be winning:

22. c5 Rf6 {4} 23. Qg8+ {2} Nf8 {13} 24. c6 {41} Bxc6 {10} 25. Be5 {6} Rd6 {7}
(25... Bd6 26. Bxd6 {12} Rxd6 {3} 27. Bf3 {6} Bxf3 {3} 28. Re1+ {2} Qe7 {3} 29.
Rxe7+ {1}) 26. Bxd6 {7} Bxd6 {6} * 1 - 0 ??

Too hard for Houdini and Komodo, only SF seems to understand with slow PC.
If I recall, "Bluefish" (Stockfish-dev, 176 threads, 7-man Syzygy) found 22. c5 during the game, but TCEC-fish didn't.

User avatar
Eduard
Posts: 160
Joined: Fri Oct 26, 2018 10:58 pm
Location: Germany
Full name: Eduard Nemeth
Contact:

Re: Hard testposition from TCEC

Post by Eduard » Sat Jan 19, 2019 5:17 pm

I am using Komodo 11.3.1 x64 on my Smartphone for my Analysis (with Droidfish and 6-men Syzygy).

Settings:
King Safety=110
Selektivity=70
Reduction=30
Dynamism=131

Rest default.

Komodo 11.3.1:
[32] >=0.57 22. c5 Rf6 23. Qg8+ *

User avatar
Eduard
Posts: 160
Joined: Fri Oct 26, 2018 10:58 pm
Location: Germany
Full name: Eduard Nemeth
Contact:

Re: Hard testposition from TCEC

Post by Eduard » Sat Jan 19, 2019 6:01 pm

Btw, I am using this Komodo Settings for other analyzes too. Of course with Contemt=0 (hash 128 MB).

User avatar
Eelco de Groot
Posts: 4043
Joined: Sun Mar 12, 2006 1:40 am
Location: Groningen

Re: Hard testposition from TCEC

Post by Eelco de Groot » Sat Jan 19, 2019 8:22 pm

Jouni wrote:
Sat Jan 19, 2019 1:56 pm


SF played 22.g5 and draw (Lc0 to superfinal) , but 22.c5 seems to be winning:

22. c5 Rf6 {4} 23. Qg8+ {2} Nf8 {13} 24. c6 {41} Bxc6 {10} 25. Be5 {6} Rd6 {7}
(25... Bd6 26. Bxd6 {12} Rxd6 {3} 27. Bf3 {6 (diagram)}
Is it very hard? Kaissa does not see the win:

[D]r2qknQ1/6pp/ppbr4/8/P5P1/5B2/5P1P/R2R2K1 b q -

Engine: Kaissa IV (512 MB)
by T. Romstad, M. Costalba, J. Kiiski, G. Linscott

35 10:19 +0.69 27...Rxd1+ 28.Rxd1 Bxf3 29.Re1+ Kd7
30.Qf7+ Kd6 31.Qf4+ Kd7 32.Qxf3 Kc7
33.Qf4+ Kc6 34.Qc4+ Kb7 35.Qf7+ Nd7
36.Re7 Kc7 37.Qf4+ Kc6 38.Qe4+ Kc5
39.Qf5+ Kc6 40.Qe6+ Kc7 (1.246.715.823) 2013

35 10:19 +1.75 27...Qe7 28.Re1 Bxf3 29.Rxe7+ Kxe7
30.Qxg7+ Ke8 31.Re1+ Re6 32.Rxe6+ Nxe6
33.Qc3 Bb7 34.Qf6 Bd5 35.Qe5 Rc8
36.f4 Rc1+ 37.Kf2 Rc2+ 38.Kg3 Rg2+
39.Kh3 Kf8 40.Qh8+ Ke7 (1.246.715.823) 2013

35 10:19 +2.05 27...Bxf3 28.Re1+ Qe7 29.Rxe7+ Kxe7
30.Qxg7+ Ke8 31.Re1+ Re6 32.Rxe6+ Nxe6
33.Qc3 Bb7 34.Qf6 Nc5 35.Qh8+ Ke7
36.Qxh7+ Ke6 37.Qg6+ Kd5 38.Qxb6 Kc4
39.Qc7 Re8 40.h4 Re1+ (1.246.715.823) 2013
27... Bxf3 {3} 28. Re1+ {2} Qe7 {3} 29.
Rxe7+ {1}) 26. Bxd6 {7} Bxd6 {6} * 1 - 0 ??

Too hard for Houdini and Komodo, only SF seems to understand with slow PC.
Debugging is twice as hard as writing the code in the first
place. Therefore, if you write the code as cleverly as possible, you
are, by definition, not smart enough to debug it.
-- Brian W. Kernighan

User avatar
Eelco de Groot
Posts: 4043
Joined: Sun Mar 12, 2006 1:40 am
Location: Groningen

Re: Hard testposition from TCEC

Post by Eelco de Groot » Sun Jan 20, 2019 6:14 am

I get this a little further in that line:



Code: Select all

Engine: Kaissa IV (512 MB)
by T. Romstad, M. Costalba, J. Kiiski, G. Linscott

35     16:27  +0.52    34...Ne6 35.Qe5 Kc6 36.h4 b5 37.f5 Nc5 
                       38.Qf6+ Kd7 39.Qg7+ Kd6 40.Qxh7 bxa4 
                       41.Qh6+ Ke5 42.Qf6+ Ke4 43.Qa1 Nd3 
                       44.f6 Ne5 45.Qxa4+ Kf5 46.Qc2+ Ke6 
                       47.f7 Rf8 (2.130.765.620) 2157 

35     16:27  +0.75    34...b5 35.f5 Kc7 36.Qd4 Kd6 37.Qb4+ Kc6 
                       38.Qc3+ Kd6 39.h4 Nd7 40.h5 Re8 
                       41.axb5 axb5 42.Qa3+ Nc5 43.g6 hxg6 
                       44.Qg3+ Kc6 45.fxg6 b4 46.g7 Bg8 
                       47.Qg6+ Re6 (2.130.765.620) 2157 

35     16:27  +1.77    34...Ng6 35.Qd4 Ne7 36.f5 Rg8 37.h4 Kd6 
                       38.Qf6+ Kd7 39.Qe5 Bf7 40.f6 Ng6 
                       41.Qe4 Kd6 42.Qd4+ Kc6 43.Qc3+ Kd7 
                       44.Qd2+ Kc6 45.a5 bxa5 46.Qxa5 Bc4 
                       47.Kf2 Re8 (2.130.765.620) 2157 

35     16:27  +1.91    34...Re8 35.h4 Kc7 36.Qg7+ Kc6 
                       37.Qh6+ Kc7 38.h5 Be4 39.Qg7+ Nd7 
                       40.g6 Bxg6 41.hxg6 hxg6 42.Qxg6 Rf8 
                       43.Kg2 Rf6 44.Qc2+ Kd8 45.Kf3 b5 
                       46.Qe4 Kc7 47.Qd4 Rc6 (2.130.765.620) 2157 
____________________________________________________

36     18:30  +0.71    34...Ne6 35.Qe5 Kc6 36.h4 b5 37.f5 Nc5 
                       38.Qf6+ Kd7 39.Qg7+ Kd6 40.Qxh7 bxa4 
                       41.Qh6+ Ke5 42.Qf6+ Ke4 43.Qa1 Nd3 
                       44.f6 Ne5 45.Qxa4+ Kf5 46.Qc2+ Ke6 
                       47.Qb2 Re8 (2.388.548.928) 2150 

36     18:30  +0.79    34...b5 35.f5 Kc7 36.Qc3+ Kd6 37.h4 Nd7 
                       38.h5 Re8 39.axb5 axb5 40.Qg3+ Ne5 
                       41.Qa3+ Kc6 42.g6 hxg6 43.fxg6 Rh8 
                       44.Qh3 Kd6 45.g7 Ra8 46.Qg3 Ra1+ 
                       47.Kf2 Ra2+ (2.388.548.928) 2150 

36     18:30  +1.25    34...Re8 35.f5 Kc7 36.Qc3+ Bc6 37.h4 Re4 
                       38.Qg7+ Nd7 39.Qxh7 Bd5 40.g6 Kd6 
                       41.Qh8 Rg4+ 42.Kf2 Rf4+ 43.Ke2 Rxf5 
                       44.g7 b5 45.axb5 axb5 46.g8Q Bxg8 
                       47.Qxg8 Rd5 (2.388.548.928) 2150 

36     18:30  +1.64    34...Ng6 35.Qd4 Ne7 36.f5 Rg8 37.g6 Kc6 
                       38.Qf6+ Kd7 39.g7 Nc6 40.Kf2 Re8 
                       41.Qc3 Re4 42.Qd3 Rf4+ 43.Ke1 Nb4 
                       44.Qxd5+ Nxd5 45.g8Q Rxf5 46.Qxh7+ Ke6 
                       47.Qb7 Rf4 (2.388.548.928) 2150 
Debugging is twice as hard as writing the code in the first
place. Therefore, if you write the code as cleverly as possible, you
are, by definition, not smart enough to debug it.
-- Brian W. Kernighan

User avatar
Eelco de Groot
Posts: 4043
Joined: Sun Mar 12, 2006 1:40 am
Location: Groningen

Re: Hard testposition from TCEC

Post by Eelco de Groot » Sun Jan 20, 2019 7:18 pm

Is the difference I see just because I'm not using tablebases?


Code: Select all

Engine: Kaissa IV (512 MB)
by T. Romstad, M. Costalba, J. Kiiski, G. Linscott

57     1385:49+1.05    22.c5 Rf6 23.Qg8+ Nf8 24.c6 Bxc6 
                       25.Be5 Bd6 26.Bxd6 Rxd6 27.Bf3 Rxd1+ 
                       28.Rxd1 Bxf3 29.Re1+ Qe7 30.Rxe7+ Kxe7 
                       31.Qxg7+ Ke6 32.g5 Bd5 33.f4 Kd6 
                       34.f5 b5 35.Qf6+ (167.594.721.391) 2015 

57     1385:49+0.94    22.g5 Rf7 23.Bh5 g6 24.Bg4 Kf8 
                       25.Rab1 a5 26.h3 Qe8 27.Qxd7 Qxd7 
                       28.Bxd7 Bf3 29.Bxe7+ Kxe7 30.Rd3 Rd8 
                       31.Bf5 Rxd3 32.Bxd3 Rf4 33.Rxb6 Rd4 
                       34.Bf1 Rd1 35.Rb3 (167.594.721.391) 2015 

57     1385:490.00     22.Rac1 Rf6 23.Qg8+ Rf8 (167.594.721.391) 2015



Code: Select all

Engine: Kaissa IV (512 MB)
by T. Romstad, M. Costalba, J. Kiiski, G. Linscott

49     438:48 +0.56    34...Ne6 35.Qe5 Kc6 36.h4 b5 37.f5 Nc5 
                       38.Qf6+ Kd7 39.axb5 axb5 40.h5 b4 
                       41.g6 h6 42.Qd4 Kc6 43.f6 b3 44.f7 Nd7 
                       45.Qc3+ Kd6 46.Qc1 Ra2 47.g7 Bxf7 (56.168.186.704) 2133 

48     438:48 +0.89    34...b5 35.f5 Kc7 36.h4 Nd7 37.Qc3+ Kd6 
                       38.Qg3+ Kc6 39.h5 b4 40.h6 Rg8 41.g6 b3 
                       42.a5 Nc5 43.Qe5 Be4 44.Kh2 hxg6 
                       45.h7 Rf8 46.fxg6 Bxg6 47.h8R Rxh8+ (56.168.186.704) 2133 

48     438:48 +1.16    34...Kc7 35.h4 Nd7 36.Qf5 Bg8 
                       37.Qc2+ Kd8 38.f5 Rb8 39.f6 b5 
                       40.Qf5 b4 41.f7 Bxf7 42.Qxf7 b3 
                       43.Qxh7 b2 44.Qb1 Ne5 45.Kg2 Ng6 
                       46.Kg3 Rb3+ 47.Kh2 Ne5 (56.168.186.704) 2133 

48     438:48 +1.67    34...Ng6 35.Qd4 Kc6 36.f5 Nh8 
                       37.Qf6+ Kc5 38.Qe7+ Kd4 39.Qg7+ Ke3 
                       40.Qe5+ Be4 41.Qg3+ Kd4 42.Qd6+ Bd5 
                       43.Qf4+ Kc5 44.Qc7+ Kd4 45.Qxh7 Nf7 
                       46.Qh4+ Ke3 47.Qg3+ Ke4 (56.168.186.704) 2133
Debugging is twice as hard as writing the code in the first
place. Therefore, if you write the code as cleverly as possible, you
are, by definition, not smart enough to debug it.
-- Brian W. Kernighan

Post Reply