Testposition mishandled opening

[Eelco de Groot]

Black has all ignored principles of normal opening play and went for a premature attack on the King with just the lone Queen as an army. The game is already more or less over after move 18... Nd5 19.Bc1! but Ancalagon could have settled for only a slight disadvantage, if he had found the move 18... Nxd4! For a Quad and tactical monsters probably a matter of seconds but this was on a single processor a bit too much to ask. Ancalagon can not find it yet in reasonable time to play it in a 12 minute + 6 seconds testgame but I am happy with the PV after Nxd4 that includes 20. Qc5 b6!

[d]r1br2k1/pp3p1p/2n2np1/8/3PNq2/P2B4/1BQ2PPP/R3R1K1 b - -

Engine: Ancalagon 2.0 Beta 006j settings Build 61 (Athlon 2009 MHz, 64 MB)
by Tord Romstad, Eelco de Groot

2.00 0:00 +0.13 18...Ng4 19.g3 (737) 16

2.00 0:00 +0.35 18...Nxd4 19.Nxf6+ Qxf6 (822) 17

3.00 0:00 +0.45 18...Nxd4 19.Nxf6+ Qxf6 20.Bxd4 Qxd4 (1.722) 37

3.00 0:00 +0.64 18...Ng4 19.h3 Nh2 (2.311) 37

4.00 0:00 +2.05 18...Ng4 19.g3 Qf3 20.Qc4 (4.535) 73

5.00 0:00 +2.13 18...Ng4 19.g3 Qf3 20.Qc4 Be6 (6.875) 88

6.00 0:00 -0.09 18...Ng4 19.g3 Qh6 20.f3 Ne3 21.Nf6+ Kh8 (12.269) 131

7.00 0:00 +0.27 18...Ng4 19.g3 Qh6 20.f3 Ne3 21.Nf6+ Kh8
22.Qd2 (24.240) 193

8.00 0:00 -1.54 18...Ng4 19.g3 Qf5 20.h3 Nh6 21.d5 Ne5
22.Nf6+ Qxf6 23.Bxe5 (40.032) 256

8.00 0:00 -0.64 18...Nd5 19.Bc1 Qg4 20.h3 Qh5 21.Qb2 Bf5
22.Qxb7 Nxd4 (159.359) 329

9.00 0:00 -0.60 18...Nd5 19.Bc1 Qg4 20.h3 Qh5 21.Bb2 Bf5
22.Rac1 Bxe4 23.Rxe4 (185.434) 339

10.00 0:00 -1.15 18...Nd5 19.Bc1 Qg4 20.f3 Qh5 21.Bg5 Re8
22.Nf6+ Nxf6 23.Bxf6 Bf5 24.Rac1 Rxe1+
25.Rxe1 Bxd3 26.Qxd3 (288.160) 376

11.01 0:02 -1.33 18...Nd5 19.Bc1 Nxd4 20.Qa4 Nb6
21.Qd1 Qc7 22.Bg5 Nd5 23.Bxd8 Qxd8
24.Rc1 (805.000) 399

12.01 0:04 -1.17 18...Nd5 19.Bc1 Nxd4 20.Qa4 Nb6
21.Qd1 Qb8 22.Bg5 Bf5 23.Bxd8 Qxd8
24.Nc5 Bxd3 25.Qxd3 (1.678.062) 416

13.01 0:09 -1.31 18...Nd5 19.Bc1 Nxd4 20.Qa4 Nb6
21.Qd1 Qb8 22.Bg5 Bf5 23.Bxd8 Qxd8
24.Nc5 Qc7 25.Bxf5 Qxc5 (4.000.525) 427

14.01 0:20 -1.68 18...Nd5 19.Bc1 Nxd4 20.Qa4 Nb6
21.Qd1 Qc7 22.Bg5 Nd5 23.Qa4 Nf5
24.Bxd8 Qxd8 25.Rac1 Bd7 (8.792.268) 433

15.01 1:01 -1.86 18...Nd5 19.Bc1 Nxd4 20.Qa4 Nb6
21.Qd1 Qc7 22.Bg5 Nd5 23.Rc1 Qb6
24.Bxd8 Qxd8 25.Nd6 Qf8 26.Re8 (27.034.363) 440

15.02 1:37 -1.68 18...Nxd4 19.Bxd4 Rxd4 20.Nxf6+ Qxf6
21.Re8+ Kg7 22.Qc5 h5 23.Bf1 Rd2
24.Rc1 b6 25.Qf8+ Kh7 26.Qg8+ Kh6
27.Rexc8 Rxc8 28.Qxc8 Qxf2+ 29.Kh1 Kg7
30.Qa6 Kg8 31.Qxa7 Kg7 (44.041.884) 451

16.01 6:38 -0.39 18...Nxd4 19.Qc1 Nh5 20.Qc5 b6
21.Qb4 Nc6 22.Qa4 Ne5 23.Nf6+ Qxf6
24.Bxe5 Qe7 25.Rad1 Bf5 26.Bxf5 gxf5
27.Rxd8+ Rxd8 28.Qb3 (181.765.738) 455

17.01 10:12 -0.39 18...Nxd4 19.Qc1 Nh5 20.Qc5 b6
21.Qb4 Nc6 22.Qa4 Ne5 23.Nf6+ Qxf6
24.Bxe5 Qe7 25.Rad1 Bf5 26.Bxf5 gxf5
27.Qc6 f6 28.Qc4+ Qf7 (276.171.999) 451

best move: Nc6xd4 time: 22:07.500 min n/s: 439.902 nodes: 583.950.000


(By the way also a minor promotion to a Knight in this game, that seems fairly common these days)

[Event "12 Minutes/Game + 6 Seconds/Move"]
[Site "Engine Match"]
[Date "2008.06.24"]
[Round "1"]
[White "Naum 3.1"]
[Black "Ancalagon 2.0 Beta 006i settings Build 52"]
[ECO "E49"]
[Result "1-0"]

1. d4 {book 0s} Nf6 {book 0s} 2. c4 {book 0s} e6 {book 0s}
3. Nc3 {book 0s} Bb4 {book 0s} 4. e3 {book 0s} O-O {book
0s} 5. Bd3 {book 0s} d5 {book 0s} 6. a3 {book 0s} Bxc3+
{book 0s} 7. bxc3 {book 0s} dxc4 {book 0s} 8. Bxc4 {book
0s} c5 {book 0s} 9. Ne2 {book 0s} Qc7 {book 0s} 10. Bd3
{book 0s} e5 {book 0s} 11. O-O {book 0s} Nc6 {book 0s}
12. Bb2 {book 0s} Rd8 {book 0s} 13. Ng3 {-0.02/12 23s} cxd4
{+0.01/13 26s} 14. cxd4 {-0.10/13 15s} exd4 {+0.17/13 21s}
15. Qc2 {-0.04/14 33s} Qe5 {0.00/12 21s} 16. Rfe1 {+0.30/12
22s (Rae1)} g6 {-0.29/13 35s} 17. exd4 {+0.38/13 18s} Qf4
{-0.68/13 22s} 18. Ne4 {+0.86/14 28s} Nd5 {-0.76/13 19s}
(18... Nxd4 19. Qc1 Nh5 20. Qc5 $14 {-0.11 20...b6 21.Qb4
Nc6 22.Qa4 Ne5 23.Nf6+ Qxf6 24.Bxe5 Qe7 25.Bb2 Qd7 26.Qb3
Bb7 27.Qc3 f6 28.Rad1 Qc6 29.Qxc6 Bxc6 -0.92 20...Ne6
21.Qc4 Qc7 22.Rac1 Qxc4 23.Bxc4 Nef4 24.Nf6+ Nxf6 25.Bxf6
Be6 26.Bxd8 Rxd8 27.Kf1 Kg7 28.Bxe6 Nxe6 29.Ke2 Nd4+ 30.Ke3
Kf6 31.Ke4 Ancalagon}) 19. Bc1 {+1.10/14 1:21m (Qc5)} Qc7
{-2.15/14 1:21m} 20. Bg5 {+1.11/14 24s} Bf5 {-1.96/13 27s}
21. Bxd8 {+1.31/13 19s} Rxd8 {-1.86/14 25s} 22. Rab1
{+1.35/14 29s} Qe7 {-2.05/13 47s} 23. Qc5 {+1.46/15 35s
(Qb2)} Qxc5 {-1.54/14 29s} 24. dxc5 {+1.47/17 17s} Bxe4
{-1.70/14 15s} 25. Bxe4 {+1.53/18 17s} Nc3 {-1.84/16 20s}
26. Rbc1 {+1.53/17 14s} Nxe4 {-1.84/15 15s} 27. Rxe4
{+1.45/4 0s} Rd7 {-1.84/15 21s} 28. g3 {+1.58/16 24s (Kf1)}
f5 {-1.70/15 21s} 29. Ree1 {+1.72/16 35s} Rc7 {-1.82/15
34s} 30. Kg2 {+1.75/17 33s} Kf7 {-1.88/15 27s} 31. Red1
{+1.70/17 25s (Rc4)} Nb8 {-1.78/14 13s} 32. Rc4 {+1.73/18
30s (Rd4)} Na6 {-1.76/15 32s} 33. Ra4 {+1.65/16 16s (c6)}
Ke6 {-1.82/15 26s} 34. Re1+ {+1.72/16 21s (Rc1)} Kd5
{-1.43/15 13s} 35. Ra5 {+1.73/16 18s (Rc1)} Rxc5 {-1.88/15
42s} 36. Rxc5+ {+1.75/14 11s} Nxc5 {-2.21/17 13s} 37. Re7
{+1.75/15 12s} h6 {-2.56/16 13s} 38. Rh7 {+2.66/15 16s
(h4)} g5 {-2.74/16 10s} 39. Rxh6 {+2.76/15 14s (h4)} g4
{-3.21/16 26s} 40. f3 {+2.83/15 11s (Rh7)} Nd3 {-3.35/17
19s} 41. fxg4 {+3.01/16 19s} fxg4 {-3.35/17 14s} 42. Rh7
{+2.97/16 13s} Kc6 {-3.56/17 16s} 43. Re7 {+3.15/16 14s} a5
{-3.29/17 13s} 44. Re4 {+3.36/16 11s} Kd5 {-3.60/17 14s}
45. Rxg4 {+4.06/16 19s} b5 {-3.39/17 12s} 46. Kf1 {+4.10/16
12s (Rg5+)} Ne5 {-3.88/16 10s} 47. Rf4 {+4.10/17 43s (Rg8)}
Ng6 {-4.21/16 9s} 48. Rf6 {+4.26/17 53s (Rf5+)} Ne7
{-4.70/16 11s} 49. Ra6 {+4.26/16 13s (h4)} Nc6 {-5.50/17
8s} 50. Rxc6 {+8.23/16 11s (h4)} Kxc6 {-2.66/16 0s} 51. Ke2
{+8.34/18 5s} Kd6 {-3.45/22 12s} 52. h4 {+8.98/19 11s
(Kd3)} Ke6 {-3.64/23 12s} 53. Ke3 {+10.54/20 13s (g4)} Ke5
{-3.74/21 10s} 54. h5 {+11.24/20 11s (g4)} b4 {-3.60/21
11s} 55. axb4 {+M25/15 2s} axb4 {-74.80/21 9s} 56. g4
{+M18/1 0s (Kd3)} b3 {-74.80/22 9s} 57. Kd2 {+M17/1 0s} b2
{-74.80/22 11s} 58. Kc2 {+M16/1 0s} b1=N {minor promotion}{-74.80/21 11s}
59. Kxb1 {+M15/1 0s} Kf6 {-M16/26 7s} 60. Kc2 {+M14/1 0s}
Kg5 {-M14/24 0s} 61. Kd3 {+M13/1 0s} Kh6 {-M13/23 0s}
62. Ke4 {+M12/1 0s} Kg5 {-M12/20 1s} 63. Ke5 {+M11/1 0s}
Kxg4 {-M11/17 0s} 64. h6 {+M9/1 0s} Kf3 {-M9/15 0s} 65. h7
{+M8/1 0s} Ke3 {-M8/13 0s} 66. h8=Q {+M7/1 0s} Kd3 {-M7/11
0s} 67. Qc8 {+M6/1 0s (Qd8+)} Ke3 {-M5/11 0s} 68. Qc3+
{+M5/1 0s} Ke2 {-M4/8 0s} 69. Kf4 {+M4/1 0s} Kd1 {-M3/7 0s}
70. Qb2 {+M3/1 0s} Ke1 {-M2/6 0s} 71. Ke3 {+M2/1 0s} Kd1
{-M1/6 0s} 72. Qb1# {+M1/1 0s} 1-0

[Eelco de Groot]

HIARCS 11.2 does not see 18... Nxd4 very quickly either and also plays Nd5 at first.

r1br2k1/pp3p1p/2n2np1/8/3PNq2/P2B4/1BQ2PPP/R3R1K1 b - -

Engine: HIARCS 11.2 SP (64 MB)
by Mark John Uniacke

8/26 0:00 -0.80 18...Ng4 19.g3 Qf5 (168.664) 262

9/26 0:01 -1.07-- 18...Ng4 (264.045) 263

9/26 0:01 -1.36 18...Ng4 19.g3 Qf5 20.h3 Nh2 21.Kxh2 Qxh3+
22.Kg1 Nxd4 23.Bxd4 (280.732) 263

9/26 0:01 -1.35 18...Nd5 (392.308) 261

9/26 0:01 -1.01 18...Nd5 19.Qc5 Qh4 20.Bc1 h6 21.Rb1 a5
22.Nd6 (450.104) 261

10/26 0:02 -0.85 18...Nd5 19.Qc5 Qh4 20.Bc1 h6 21.Rb1 Rb8
22.Nd6 Qxd4 23.Qxd4 Nxd4 (782.069) 266

11/26 0:06 -0.85 18...Nd5 19.Bc1 Nxd4 20.Qc4 Qe5
21.Bg5 b5 (1.671.368) 271

12/29 0:09 -1.10-- 18...Nd5 (2.731.552) 282

12/29 0:11 -1.82 18...Nd5 19.Bc1 Nxd4 20.Qa4 Nb6
21.Qd1 Qb8 22.Nf6+ Kh8 23.Qd2 Nf5 (3.116.612) 282

13/33 0:32 -1.71 18...Nd5 19.Bc1 Nxd4 20.Qa4 Nb6
21.Qd1 Qc7 22.Bg5 Be6 23.Rc1 Qb8
24.Bxd8 Qxd8 25.Be2 Bb3 (8.855.908) 275

14/35 0:55 -1.95 18...Nd5 19.Bc1 Qc7 20.Bg5 Be6
21.Bxd8 Rxd8 22.Rab1 (15.407.542) 277

15/38 2:08 -1.90 18...Nd5 19.Bc1 Qc7 20.Bg5 Be6
21.Qd2 Rdc8 22.Rac1 Kh8 (34.918.468) 272

16/41 4:29 -2.13 18...Nd5 19.Bc1 Qc7 20.Bg5 Bf5
21.Bxd8 Rxd8 22.Rab1 Kg7 23.Ng5 f6
24.Nf3 Bg4 (72.612.927) 269

16/47 7:41 -2.12 18...Nxd4 (131.200.629) 284

16/47 10:21 -1.67 18...Nxd4 19.Qc5 Ng4 20.Bxd4 Qxh2+
21.Kf1 Bf5 22.Rad1 Qh1+ 23.Ke2 Qxg2
24.Rg1 Qh2 (179.077.304) 288

17/47 23:48 -1.42 18...Nxd4 19.Qc5 Ng4 20.Bxd4 Qxh2+
21.Kf1 Bf5 22.Qc3 Qh1+ 23.Ke2 Qxg2
24.Bh8 f6 25.Rg1 Bxe4 (413.649.354) 289

18/48 48:42 -1.17++ 18...Nxd4 (854.180.704) 292

18/48 72:59 -1.07 18...Nxd4 19.Qc4 Be6 20.Qb4 a5
21.Qb6 Nxe4 22.Rxe4 Qd6 (1.276.438.154) 291

19/52 138:01 -0.97 18...Nxd4 19.Qc5 Ng4 20.Bxd4 Qxh2+
21.Kf1 Bf5 22.Rad1 Qh1+ 23.Ke2 Qxg2
24.Rg1 Qh2 25.Rh1 Qf4 26.f3 b6
27.Qc4 Bxe4 28.Bxe4 Rac8 29.Qd3 Nf6
30.Qe3 Nh5 (2.465.115.782) 297

best move: Nc6xd4 time: 168:24.719 min n/s: 293.420 nodes: 2.964.920.900


Output from Naum 3.1 with the same limited amount of hash, Naum is fastest of the three engines:
r1br2k1/pp3p1p/2n2np1/8/3PNq2/P2B4/1BQ2PPP/R3R1K1 b - -

Engine: Naum 3.1 (64 MB)
by Aleksandar Naumov

9/22 0:00 -0.63 18...Nxe4 19.Rxe4 Qd6 20.d5 Bf5
21.Qc3 f6 22.dxc6 Qxd3 (477.775) 848

10/26 0:01 -1.70 18...Nxe4 19.Rxe4 Qd6 20.d5 f6
21.dxc6 Qxd3 22.c7 Rd7 23.Qxd3 (881.399) 881

10/33 0:01 -0.99 18...Nxd4 19.Qc5 Ng4 20.Bxd4 Qxh2+
21.Kf1 Be6 22.Rad1 Bb3 23.Rc1 (1.349.426) 880

11/32 0:02 -0.93 18...Nxd4 19.Qc5 Ng4 20.Bxd4 Qxh2+
21.Kf1 Be6 22.Rad1 Rac8 23.Qxa7 Qh1+ (2.435.072) 870

11/32 0:03 -0.88 18...Nd5 19.Qc5 Qc7 20.Bc1 b6 21.Qc4 Be6
22.Bg5 Re8 23.Rac1 Rac8 (3.364.780) 854

12/33 0:06 -0.88 18...Nd5 19.Qc5 Qc7 20.Bc1 b6 21.Qc4 Be6
22.Bg5 Re8 23.Rac1 Rac8 24.h3 (5.282.835) 851

13/28 0:09 -0.96 18...Nd5 19.Qc5 Qc7 20.Bc1 b6 21.Qc4 Be6
22.Bg5 Re8 23.Rac1 Rac8 24.h3 Qd7 (8.266.592) 863

14/34 0:29 -0.84 18...Nd5 19.Qc5 Qc7 20.Bc1 b6 21.Qc4 Be6
22.Bg5 Re8 23.Rac1 Rac8 24.Qb5 f5
25.Nf6+ (25.122.835) 862

15/33 0:56 -1.08 18...Nd5 19.Bc1 Nxd4 20.Qa4 Nb6
21.Qb4 Nc6 22.Qb2 Na4 23.Qc2 Qc7
24.Bg5 Be6 25.Bxd8 Rxd8 (49.551.576) 884

15/38 1:04 -0.62 18...Nxd4 19.Qc5 Ng4 20.Bxd4 Qxh2+
21.Kf1 Bf5 22.Rad1 Qh1+ 23.Ke2 Qxg2
24.Rg1 Qh2 25.Qb4 Qf4 (58.445.985) 901

16/40 1:20 -0.54 18...Nxd4 19.Qc5 Ng4 20.Bxd4 Qxh2+
21.Kf1 Bf5 22.Rad1 Qh1+ 23.Ke2 Qxg2
24.Rg1 Qh2 25.Rh1 Qf4 26.f3 (74.132.822) 925

17/37 1:47 -0.57 18...Nxd4 19.Qc5 Ng4 20.Bxd4 Qxh2+
21.Kf1 Bf5 22.Rad1 Qh1+ 23.Ke2 Qxg2
24.Rg1 Qh2 25.Rh1 Qf4 26.f3 h5 (103.381.017) 957

18/39 2:48 -0.50 18...Nxd4 19.Qc5 Ng4 20.Bxd4 Qxh2+
21.Kf1 Bf5 22.Rad1 Qh1+ 23.Ke2 Qxg2
24.Rg1 Qh2 25.Rh1 Qf4 26.f3 h5
27.Rh4 (167.072.286) 990

19/47 5:44 -0.38 18...Nxd4 19.Qc5 Ng4 20.Bxd4 Qxh2+
21.Kf1 Bf5 22.Rad1 Qh1+ 23.Ke2 Qxg2
24.Rg1 Qh2 25.Rh1 Qf4 26.f3 h5
27.Rh4 b6 (349.755.545) 1016

20/44 10:57 -0.45 18...Nxd4 19.Qc1 Nh5 20.Qc5 b6
21.Qb4 Nc6 22.Qc3 Nd4 23.Rad1 Bg4
24.Rd2 Bf3 25.gxf3 Nxf3+ 26.Kf1 Nxd2+
27.Nxd2 Qf6 28.Ba6 (675.104.918) 1026

best move: Nc6xd4 time: 13:29.578 min n/s: 1.026.018 nodes:
779.466.645

Look at the differences in nodes per second, Naum is a fast searcher!

In case Tord is looking for ideas to improve his code or is not yet sure what version of Glaurung should play in the Chess960 play-off at ICC, it is tomorrow I believe, should be an interesting tournament, a small report about most recent change in Mobility counting in Ancalagon:

In Ancalagon Mobility is still at 90, where in Glaurung it is at 256 (0x100 hexadecimal). I do think it is too high but maybe it depends on the other parameters somewhat how much. Experimenting with "safe mobility" by not counting squares that are attacked by the enemy. I think I'll keep that in although in some positions it makes a large difference in eval values.

This is the code for Queen Mobility, & NOT attacked_by (them ) does an AND thankfully making use of Tord's bitboardcode containing the squares attacked by the opponent. count_1s counts the bits, it can be replaced by count_1s_max15, I don't know exactly what the max15 stood for but I think this is the indentical function (Edit: _max15 probably limits the maximum count to 15 that would make sense, but I have not yet checked how that works):

Code: Select all

    // Mobility
    int mob = count_1s(b & ~p.pieces_of_color(us) & ~ei.attacked_by(them));
    ei.mgMobility += Sign[us] * MidgameQueenMobilityBonus[mob];
    ei.egMobility += Sign[us] * EndgameQueenMobilityBonus[mob];
  }

Eelco

[kgburcham]

[d] r1br2k1/pp3p1p/2n2np1/8/3PNq2/P2B4/1BQ2PPP/R3R1K1 b - - 0 1

Rybka 2.3.2a mp x64
11.00 0:01 +0.35 1...Nxd4 2.Qc5 Ng4 3.Bxd4 Qxh2+ 4.Kf1 Bf5 5.Rad1 Qh1+ 6.Ke2 Qxg2 7.Rg1 Qh2 (455.288) 397
12.01 0:01 +0.28 1...Nxd4 2.Qc5 Ng4 3.Bxd4 Qxh2+ 4.Kf1 Bf5 5.Rad1 Qh1+ 6.Ke2 Qxg2 7.Rg1 Qh2 8.Rh1 (768.283) 412
13.01 0:03 +0.23 1...Nxd4 2.Qc5 Ng4 3.Bxd4 Qxh2+ 4.Kf1 Bf5 5.Rad1 Qh1+ 6.Ke2 Qxg2 7.Rg1 Qh2 8.Rh1 (1.371.788) 425
14.01 0:04 +0.22 1...Nxd4 2.Qc5 Ng4 3.Bxd4 Qxh2+ 4.Kf1 Bf5 5.Rad1 Qh1+ 6.Ke2 Qxg2 7.Rg1 Qh2 8.Rh1 (1.942.527) 441
15.01 0:06 +0.20 1...Nxd4 2.Qc5 Ng4 3.Bxd4 Qxh2+ 4.Kf1 Bf5 5.Rad1 Qh1+ 6.Ke2 Qxg2 7.Rg1 Qh2 8.Rh1 (2.754.044) 422
16.01 0:12 +0.19 1...Nxd4 2.Qc5 Ng4 3.Bxd4 Qxh2+ 4.Kf1 Bf5 5.Rad1 Qh1+ 6.Ke2 Qxg2 7.Rg1 Qh2 8.Rh1 (4.812.365) 410
17.01 0:45 +0.30 1...Nxd4 2.Qc1 Qf5 3.Qd2 Qe5 4.Rad1 Bg4 5.f3 Bxf3 6.Nxf6+ Qxf6 7.Qf2 Qg5 8.Bxd4 (18.863.493) 422
18.01 1:05 +0.21 1...Nxd4 2.Qc1 Nh5 3.Qc5 b6 4.Qb4 Nc6 5.Qa4 Ne5 6.Nf6+ Qxf6 7.Bxe5 Qg5 8.h4 (28.137.784) 436
19.01 1:21 +0.35 1...Nxd4 2.Qc1 Nh5 3.Qc5 b6 4.Qb4 Nc6 5.Qc3 Nd4 6.Qb4 Nc6 7.Qc3 Nd4 8.Qb4 (34.345.145) 431
20.01 2:21 +0.48 1...Nxd4 2.Qc1 Nh5 3.Qc5 b6 4.Qb4 Nc6 5.Qc3 Nd4 6.Qb4 Nc6 7.Qc3 Nd4 8.Qb4 (57.849.265) 419
best move: Nc6xd4 time: 3:19.609 min n/s: 389.538 nodes: 75.838.259

Deep Hiarcs Paderborn 2007:
1...Nxd4
+- (1.95) Depth: 16/45 00:01:23 121mN
1...Nxd4 2.Qc5 Ng4 3.Bxd4 Qxh2+ 4.Kf1 Bf5 5.Qc3 Rac8 6.Qb2 Qh1+ 7.Ke2 Qxg2 8.Rg1 Qh3 9.Rg3
+- (1.90) Depth: 16/46 00:01:40 147mN
1...Nxd4
+- (1.65) Depth: 17/46 00:03:28 309mN
1...Nxd4 2.Qc5 Ng4 3.Bxd4 Qxh2+ 4.Kf1 Bf5 5.Rad1 Qh1+ 6.Ke2 Qxg2 7.Rg1 Qh2 8.Rh1 Qf4 9.f3 Rd7 10.Rh4 Re8 11.fxg4 Bxe4
+/- (1.21) Depth: 17/46 00:04:21 394mN

[swami]

Bright @ Q6600.

Time:38 Seconds


New game
r1br2k1/pp3p1p/2n2np1/8/3PNq2/P2B4/1BQ2PPP/R3R1K1 b - - 0 1

Analysis by bright-0.3a:

1...Nxe4 2.Rxe4
+/- (0.81) Depth: 1/5 00:00:00
1...Nxd4 2.Nxf6+ Qxf6
=/+ (-0.27) Depth: 1/5 00:00:00
1...Nxd4 2.Nxf6+ Qxf6
=/+ (-0.27) Depth: 2/9 00:00:00
1...Nxd4 2.Nxf6+ Qxf6 3.Qc7 Kg7
= (-0.16) Depth: 3/16 00:00:00
1...Nxd4 2.Nxf6+ Qxf6 3.Qc7 Kg7
= (-0.16) Depth: 4/16 00:00:00
1...Nxd4 2.Nxf6+ Qxf6 3.Qc7 Kg7 4.Rac1 b6
= (0.21) Depth: 5/16 00:00:00
1...Ng4 2.g3 Qh6 3.f3 Ne3 4.Nf6+ Kh8
= (-0.22) Depth: 5/16 00:00:00
1...Ng4 2.h3 Qh2+ 3.Kf1 Qh1+ 4.Ke2 Qxg2 5.hxg4 Nxd4+ 6.Bxd4 Bxg4+ 7.Kd2 Rxd4 8.Nf6+ Kh8 9.Nxg4 Qxg4
= (0.18) Depth: 6/23 00:00:00 54kN
1...Ng4 2.h3 Qh2+ 3.Kf1 Qh1+ 4.Ke2 Qxg2 5.hxg4 Nxd4+ 6.Bxd4 Bxg4+ 7.Kd2 Rxd4 8.Nf6+ Kh8 9.Nxg4 Qxg4
= (0.18) Depth: 7/24 00:00:00 127kN
1...Ng4 2.g3 Qh6 3.f3 Ne3 4.Qd2 Nf5 5.Nf6+ Kg7 6.Qxh6+ Nxh6
+/- (1.13) Depth: 8/29 00:00:01 354kN
1...Nh5 2.Qa4 Qc7 3.Nc5 b6 4.Be4 Rb8 5.Nd3 Bd7
+/- (1.02) Depth: 8/29 00:00:01 835kN
1...Nxe4 2.Rxe4 Qg5 3.f4 Qh5 4.Qc3 Qd5 5.Rae1 Bf5
+/- (0.99) Depth: 8/29 00:00:01 1389kN
1...Nd5 2.Rac1 Be6 3.Nc5 Nxd4 4.Bxd4 Qxd4 5.Nxe6 fxe6 6.Rxe6
+/- (0.93) Depth: 8/29 00:00:01 1874kN
1...Nd5 2.Qc5 Be6 3.Bc1 Qc7 4.Bg5 Nde7 5.Nf6+ Kg7 6.Ng4 Bxg4 7.Rxe7
+/- (1.04) Depth: 9/29 00:00:01 2682kN
1...Nxd4 2.Qc5 Ng4 3.Bxd4 Qxh2+ 4.Kf1 Qh1+ 5.Ke2 Qxg2 6.Rab1 h6 7.Nd6 b6
+/- (0.96) Depth: 9/30 00:00:01 3258kN
1...Nxd4 2.Qc5 Ng4 3.Bxd4 Qxh2+ 4.Kf1 Qh1+ 5.Ke2 Qxg2 6.Rab1 h6 7.Qc3 Re8 8.Kd1 Bf5
+/- (1.10) Depth: 10/39 00:00:01 4518kN
1...Nd5 2.Rab1 Qh4 3.Qb3 Na5 4.Qb5 Nc6 5.Bc4 a6 6.Qb3 Qh5
+/- (0.99) Depth: 10/39 00:00:02 5619kN
1...Nd5 2.Rab1 Qh4 3.Qb3 Na5 4.Qb5 Nc6 5.Bc4 a6 6.Qb3 Qh5
+/- (0.99) Depth: 11/39 00:00:02 8068kN
1...Nd5 2.Qc5 Be6 3.Bc1 Qc7 4.Bg5 Rd7 5.Rac1 Re8 6.Kh1 Bf5 7.Rc2 Bg4
+/- (0.91) Depth: 12/39 00:00:04 26035kN
1...Nd5 2.Qc5 Be6 3.Bc1 Qc7 4.Bg5 Rd7 5.Rac1 Re8 6.Kh1 Kg7 7.Bc4 a6
+/- (1.01) Depth: 13/39 00:00:07 38384kN
1...Nd5 2.Bc1 Nxd4 3.Qa4 Nb6 4.Bxf4 Nxa4 5.Bg5 Be6 6.Bxd8 Rxd8 7.Rab1 f5 8.Ng5 Bd5
+/- (1.29) Depth: 14/41 00:00:21 116mN
1...Nxd4 2.Qc5 Ng4 3.Bxd4 Qxh2+ 4.Kf1 Bf5 5.Rad1 b6 6.Qc6 Rac8 7.Nf6+ Kf8 8.Qb7 Rxd4 9.Bxf5 Qh1+ 10.Ke2
+/- (1.26) Depth: 14/49 00:00:38 213mN
1...Nxd4 2.Qc5 Ng4 3.Bxd4 Qxh2+ 4.Kf1 Bf5 5.Rad1 b6 6.Qb4 Qh1+ 7.Ke2 Qxg2 8.Rg1 Qh2 9.Rh1 Qf4 10.Be3 Qe5
+/- (0.78) Depth: 15/55 00:01:06 381mN
1...Nxd4 2.Qc4 Be6 3.Qb4 Ng4 4.Bxd4 Qxh2+ 5.Kf1 Rac8 6.Rad1 b5 7.Bxb5 Rc2 8.Rd2 Qh1+ 9.Ke2 Qxg2 10.Bd3 a5 11.Qa4 Bb3 12.Qxb3 Rxd2+ 13.Kxd2 Rxd4
+/- (0.87) Depth: 16/55 00:02:06 738mN
1...Nxd4 2.Qa4 Ng4 3.Bxd4 Qxh2+ 4.Kf1 Bf5 5.Rad1 Qh1+ 6.Ke2 Qxg2 7.Rg1 Qh2 8.Rh1 Qf4 9.f3 Ne5 10.Bxe5 Qxe5 11.Qb3 b6
+/- (0.95) Depth: 17/69 00:03:43 1300mN

[Eelco de Groot]

Thanks guys for your input! I'm still using the position as a testposition, in fact much as the only testposition together with a move from the Rybka forum, Kasimdzhanov - McShane Kings Indian Defence where Luke McShane played 22... Bh4!


[d]2b2rk1/r2nq3/p1Qp1bp1/2pPpp2/2P1P1P1/P1N1B3/6PP/1R2RBK1 b - - 0 22

http://rybkaforum.net/cgi-bin/rybkaforu ... l?tid=4250


Maybe this is a bit limited testing material with all the weird changes I tried...

Ancalagon has now a modified search that in theory should be so inefficient that even plain alpha-beta, without any null-move pruning, should be running rings around it That is, if the search would be acting the way I thought it would but clearly I have not understood the first principles of alpha-beta or understand still nothing of what Tord's search is doing, or most of the change is bypassed in some way... I'm not going into details, I just would like to get this to work a little better and try to understand for myself what is really going on.

It could be that with only one testposition I'm seeing fata morganas and the search really is not working at all and just found the right move by accident. But I hope not!


[d]r1br2k1/pp3p1p/2n2np1/8/3PNq2/P2B4/1BQ2PPP/R3R1K1 b - -

Engine: Ancalagon 2.0 Beta 006j settings Build 69 (Athlon 2009 MHz, 64 MB)
by Tord Romstad, Eelco de Groot

9.00 0:00 -1.00 18...Nd5 19.Qc5 Qh4 20.g3 Qh5
21.Qxd5 Qxd5 22.Nf6+ Kg7 23.Nxd5 Rxd5 (320.824) 373

10.01 0:01 -1.05 18...Nd5 19.Qc5 Qh4 20.Nd6 Nf4
21.Bc4 Ne6 22.g3 Nxc5 23.gxh4 Ne6 (632.098) 396

11.01 0:03 -1.82 18...Nd5 19.Bc1 Nxd4 20.Qa4 Nb6
21.Qd1 Qe5 22.Bg5 Qc7 23.Bxd8 Qxd8
24.Rc1 (1.340.543) 414

12.01 0:05 -1.49 18...Nd5 19.Bc1 Nxd4 20.Qa4 Nb6
21.Qd1 Qc7 22.Nf6+ Kh8 23.Qd2 Qc6
24.Qf4 Bf5 (2.460.653) 427

13.01 0:10 -2.17 18...Nd5 19.Bc1 Qg4 20.h3 Qh5 21.Bg5 Bf5
22.Bxd8 Rxd8 23.Qb2 Bxe4 24.Bxe4 b6
25.Qd2 Rc8 (4.556.101) 437

14.01 0:33 -2.43 18...Nd5 19.Bc1 Qc7 20.Bg5 Bf5
21.Bxd8 Rxd8 22.Rab1 Bc8 23.Qc5 Bf5
24.Qxd5 Rxd5 25.Nf6+ Kf8 26.Re8+ Kg7
27.Nxd5 Bxd3 28.Nxc7 Bxb1 (15.116.042) 454

14.02 0:51 -0.58 18...Nxd4 19.Bxd4 Rxd4 20.Nxf6+ Qxf6
21.Re8+ Kg7 22.Rae1 Rd8 23.Rxd8 Qxd8
24.Qb2+ Qf6 25.Qb5 Bf5 26.Qxb7 Qc3
27.Re7 Qa1+ 28.Bf1 Qf6 (23.936.430) 462

15.01 2:13 0.00 18...Nxd4 19.Qc1 Nh5 20.Qxf4 Nxf4
21.Nf6+ Kg7 22.Re8 Be6 23.Bxd4 Nxd3
24.Ng8+ Rxd4 25.Rxa8 f6 26.Ne7 Kf7
27.Ng8 Kg7 28.Ne7 (64.167.490) 481

16.01 3:58 -0.60 18...Nxd4 19.Qc1 Nh5 20.Qc5 b6
21.Qb4 Nc6 22.Qc4 Ne5 23.Nf6+ Qxf6
24.Bxe5 Qe6 25.Bc7 Qxc4 26.Bxc4 Rf8
27.Re7 Bf5 (113.065.023) 473

17.01 8:53 -0.52 18...Nxd4 19.Qc1 Nh5 20.Qc5 Ne6
21.Qc2 Qc7 22.g4 Qxc2 23.Bxc2 Nhf4
24.Nf6+ Kh8 25.Rad1 Rxd1 26.Rxd1 Nh3+
27.Kf1 Nhg5 28.Nd5+ Kg8 29.Be5 (252.695.159) 473

18.01 24:07 -0.62 18...Nxd4 19.Qc1 Nh5 20.Qc5 Ne6
21.Qc2 Qc7 22.g4 Qxc2 23.Bxc2 Nhf4
24.Nf6+ Kh8 25.Rad1 Rxd1 26.Rxd1 Nh3+
27.Kf1 b6 28.Ne4+ Kg8 29.Nf6+ Kh8
30.Nd5+ Kg8 (685.315.081) 473

19.01 76:57 -0.84 18...Nxd4 19.Qc1 Nh5 20.Qc5 Ne6
21.Qc2 Qc7 22.g4 Qxc2 23.Bxc2 Nhf4
24.Nf6+ Kg7 25.Rad1 Rxd1 26.Rxd1 Nh3+
27.Kg2 Nhf4+ 28.Kg3 Kf8 29.Nxh7+ Kg8
30.Nf6+ Kf8 31.h4 (2.157.584.508) 467

best move: Nc6xd4 time: 80:11.078 min n/s: 466.403 nodes: 2.243.880.000

It is a bit of a pity that in the PV after 20.Qc5 the move 20... b6 is lost, I thought that was the best move, not 20... Ne6, it could be that the PV is deteriorating a bit deeper on when 18... Nxd4 is clearly found as best, or it could be a side effect or bug, I really expected this code to crash repeatedly, a bit the way Bruce Moreland described in his first experiments with search instability by too small windows in the search, "hellacious crashes"