Shirov's Bh3, Houdini solves it

[kgburcham]

[D] 8/8/4kpp1/3p1b2/p6P/2B5/6P1/6K1 b - -


CPU0: GenuineIntel Intel64 Family 6 Model 44 Stepping 2 3373 MHz
CPU1: GenuineIntel Intel64 Family 6 Model 44 Stepping 2 3373 MHz
CPU2: GenuineIntel Intel64 Family 6 Model 44 Stepping 2 3373 MHz
CPU3: GenuineIntel Intel64 Family 6 Model 44 Stepping 2 3373 MHz
CPU4: GenuineIntel Intel64 Family 6 Model 44 Stepping 2 3373 MHz
CPU5: GenuineIntel Intel64 Family 6 Model 44 Stepping 2 3373 MHz
6 processor(s) found, POPCNT available
NUMA configuration with 1 node(s), offset 0
4096 MB Large Page Hash
Nalimov 6 men EGTB available - 999 MB cache

Houdini 2.0c

29/66 1:53 -1.42++ 47...Bh3 (2.150.809.593) 18895
29/66 2:15 -1.58++ 47...Bh3 (2.581.417.283) 19016
29/66 2:56 -1.94++ 47...Bh3 (3.381.598.157) 19144
29/72 4:56 -2.90++ 47...Bh3 (5.819.064.164) 19654

Houdini DEV 3T

29/66 2:33 -1.45++ 1...Bh3 (1.067.275.340) 6930
29/66 2:47 -1.61++ 1...Bh3 (1.168.173.148) 6975
26/66 2:52 -1.97++ 1...Bh3 (1.204.615.635) 6985
26/66 3:05 -2.92++ 1...Bh3 (1.298.645.183) 7010
26/66 3:35 -3.96 1...Bh3 (1.522.155.628) 7063
27/66 3:47 -3.77-- 1...Bh3 2.g3 (1.609.053.638) 7083
27/66 3:50 -3.40-- 1...Bh3 2.g3 (1.630.813.610) 7083
27/67 4:00 -4.12++ 1...Bh3 (1.710.238.541) 7105
27/67 4:02 -4.10 1...Bh3 (1.723.909.906) 7103

[Event "It (cat.21)"]
[Site "Linares (Spain)"]
[Date "1998"]
[Round "10"]
[White "Veselin Topalov"]
[Black "Alexey Shirov"]
[ECO "D85"]
[Result "0-1"]

1. d4 Nf6 2. c4 g6 3. Nc3 d5 4. cxd5 Nxd5 5. e4 Nxc3
6. bxc3 Bg7 7. Bb5+ c6 8. Ba4 O-O 9. Ne2 Nd7 10. O-O e5
11. f3 Qe7 12. Be3 Rd8 13. Qc2 Nb6 14. Bb3 Be6 15. Rad1 Nc4
16. Bc1 b5 17. f4 exd4 18. Nxd4 Bg4 19. Rde1 Qc5 20. Kh1 a5
21. h3 Bd7 22. a4 bxa4 23. Ba2 Be8 24. e5 Nb6 25. f5 Nd5
26. Bd2 Nb4 27. Qxa4 Nxa2 28. Qxa2 Bxe5 29. fxg6 hxg6
30. Bg5 Rd5 31. Re3 Qd6 32. Qe2 Bd7 33. c4 Bxd4 34. cxd5
Bxe3 35. Qxe3 Re8 36. Qc3 Qxd5 37. Bh6 Re5 38. Rf3 Qc5
39. Qa1 Bf5 40. Re3 f6 41. Rxe5 Qxe5 42. Qa2+ Qd5 43. Qxd5+
cxd5 44. Bd2 a4 45. Bc3 Kf7 46. h4 Ke6 47. Kg1 Bh3 48. gxh3
Kf5 49. Kf2 Ke4 50. Bxf6 d4 51. Be7 Kd3 52. Bc5 Kc4 53. Be7
Kb3 0-1

[fern]

I remember this problem very wel, because a chess friend putted it to me in a board, as a challenge, after I told him that even being, myself, just an average player, I was pretty good at solving problems. He did it and in half a minute or so I gave him the bh3 solution.
It was not a matter of calculation, but just intuition. Is the strongest part of my mind, as it fits to a writer.

My best
Fern

[kgburcham]

in half a minute or so I gave him the bh3 solution

right move wrong reason

[fern]

right move, NO reason...

Fern

[Vinvin]

right move, NO reason...

Fern

= no chance to play it in a serious game.

[fern]

In a serious game I wouldn't have ever found that move, but when you are told that there is a solution and you put yourself in "problem mode", all is different and you "play" 500 points above your head.

Fern

[Vinvin]

In a serious game I wouldn't have ever found that move, but when you are told that there is a solution and you put yourself in "problem mode", all is different and you "play" 500 points above your head.

Fern

In fact, it's not "playing chess", it's "solve a position where you know there's unusual things to play" and you search unusual moves ... computers solve in "normal play mode"

[fern]

It is what I am saying...In fact, sometimes, as I am playing a game, I try to stop and tell myself "this position is in problem book, find the best move". Of course it is hard to do, because in a game you have not an unlimited lapse of time to find a move.... clock is ticking....


bh3 regards
Fern

[Vinvin]

Test sets must be upgraded every few years. A good set for a top program should be difficult enough that a good percentage of the positions would take an impractical amount of time to solve by the top programs.

Also, they must be accurate - in past years sets would come out that had all sort of alternate solutions not considered. Also, the key move (or moves) should be the only one that wins or draws, or the only one that avoids a loss. The reason for that is that computers look deeper and deeper and there comes a point that a clever tactic that wins in an already winning position may not be particularly impressive to a computer program. A future program might even see the tactic and choose a different path that wins material later in the game. You should not be penalized for playing a winning move and it may not be easy for a general problem set to determine if a computer see's the tactic if it's not required to win.

On Intel 3 cores @ 3.1 Ghz

FEN: 8/8/4kpp1/3p1b2/p6P/2B5/6P1/6K1 b - - 0 1

Houdini_15_w32:
...
30/56 14:11 2.726.080.589 3.202.000 +1,41 Ff5-e4 Rg1-f2
f6-f5 Fc3-d4 Re6-d7 g2-g3 Rd7-c6 Rf2-e3 Rc6-b5 Fd4-b2 Rb5-c4 Re3-d2
d5-d4 Rd2-e2 d4-d3+ Re2-d2 Rc4-b3 Fb2-d4 a4-a3 Rd2-c1 Rb3-c4 Fd4-e5
a3-a2 Fe5-a1 Fe4-c6 Fa1-b2 Rc4-b3 Fb2-a1 Rb3-b4 Rc1-d2 Rb4-c4 Fa1-f6
Fc6-a4 Ff6-a1 Fa4-c2 Fa1-c3 Rc4-b3 Fc3-d4 Rb3-b4
30/67+ 24:59 4.913.246.933 3.277.000 +1,49 Ff5-h3
30/78+ 1:36:56 21.150.955.106 3.636.000 +5,54 Ff5-h3
30/87 5:12:05 75.495.674.421 4.031.000 +5,37 Ff5-h3
Rg1-f2 Re6-f5 g2-g3 Rf5-e4 Fc3-b2 Re4-d3 Rf2-f3 Fh3-f1 Rf3-f2 Ff1-e2
Rf2-e1 f6-f5 Fb2-a3 Rd3-c2 Fa3-c5 Fe2-g4 Fc5-e3 Fg4-h5 Re1-f2 Rc2-c3
Rf2-e1 d5-d4 Fe3xd4+ Rc3xd4 Re1-d2 a4-a3 Rd2-c2 a3-a2
31/87 5:31:41 80.019.827.765 4.020.000 +5,37 Ff5-h3
Rg1-f2 Re6-f5 g2-g3 Rf5-e4 Fc3-b2 Re4-d3 Rf2-f3 Fh3-f1 Rf3-f2 Ff1-e2
Rf2-g2 Fe2-h5 Rg2-f1 Fh5-g4 Rf1-g1 d5-d4 Rg1-f2 f6-f5 Rf2-f1 Rd3-c2
Fb2xd4 a4-a3 Rf1-f2 a3-a2 Fd4-e5 Rc2-b1 Rf2-e3 a2-a1D Fe5xa1 Rb1xa1
Re3-f2 Ra1-b2 Rf2-e1 Rb2-c3
32/87 6:06:50 87.684.517.651 3.983.000 +5,49 Ff5-h3
Rg1-f2 Re6-f5 g2-g3 Rf5-e4 Fc3-b2 Re4-d3 Rf2-f3 Fh3-f1 Rf3-f2 Ff1-e2
Rf2-g2 Fe2-h5 Rg2-f2 f6-f5 Rf2-g2 Fh5-g4 Rg2-f2 Fg4-e2 Rf2-g1 Fe2-f3
Rg1-f1 Ff3-h5 Rf1-e1 Fh5-e2 Fb2-a3 Fe2-f3 Re1-f2 Ff3-g4 Fa3-b2 Fg4-e2
33/87+ 7:26:21 105.294.949.142 3.931.000 +5,72 Ff5-h3
33/87 7:49:08 110.308.795.417 3.918.000 +5,71 Ff5-h3
Rg1-f2 Re6-f5 g2-g3 Rf5-e4 Fc3-b2 Re4-d3 Rf2-f3 Fh3-f1 Rf3-f2 Ff1-e2
Rf2-g2 Fe2-h5 Rg2-f2 f6-f5 Rf2-f1 Fh5-e2+ Rf1-e1 Fe2-f3 Re1-f1 Ff3-h5
Rf1-f2 Fh5-g4 Rf2-e1 Fg4-f3 Re1-f1

24m59s to find the move but 1h12m more to solve the fail-high.
Robert, is the latest Houdini faster to resolve the fail-high ?


BTW, I prepare a new Hard test set (hardest from hard-CCC, the-nightmare-game, MGV77, 100-brillante Schachzug and hard positions from Rybka forum). I hope to get ready in 1 week.

+1 ,No make that three!

If you want to know in wich kind of mess I am, I stopped adding positions today : http://home.scarlet.be/vincentlejeune/h ... s-2012.pgn
Still to do :
1) Verification for new-unknown
2) make PGN for best lines
3) remove too easy
4) categorize

Probably more than 3 weeks

[mhalstern]

WOW - This brings back memories. Right after Shirov played this, (sometime in the 1990's I believe) I gave M-Chess (Whatever was the latest version at the time) the position and let it analyze for 3 days. It did not find Shirov's Bh3.

I seem to remember the pc was a 486 runninf at 100 mhz.