Author: Uri Blass
Date: 03:35:58 07/14/02
Go up one level in this thread
On July 14, 2002 at 04:39:47, Mike Byrne wrote:
>On July 14, 2002 at 02:14:33, Ron Murawski wrote:
>
>>Horizon 3.2 drew a game against Cerebro 1.10 in Leo Dijksman's 3rd Division
>>Ridderkerk tournament. Just for the record, here is the game:
>>
>>[Event "WBEC3_3rdDivision"]
>>[Site "DUAL-P3-933"]
>>[Date "2002.07.12"]
>>[Round "1.3"]
>>[Number "399"]
>>[White "Cerebro 1.10b"]
>>[Black "Horizon 3.2"]
>>[Result "1/2-1/2"]
>>[TimeControl "40/2400"]
>>
>>1. d4 d5 2. c4 c6 3. Nc3 dxc4 4. e3 b5 5. a4 b4 6. Ne4 Nf6 7. Nxf6+ exf6 8.
>>Bxc4 Be7 9. Qc2 Be6 10. Ne2 Bxc4 11. Qxc4 O-O 12. O-O Qd5 13. Qc2 c5 14. e4
>>Qb7 15. d5 f5 16. f3 Nd7 17. Bf4 fxe4 18. fxe4 Bf6 19. g4 h6 20. Kh1 b3 21.
>>Qc1 Rfe8 22. Ng3 c4 23. Bd2 Be5 24. Bf4 Bf6 25. Qxc4 Bxb2 26. Rad1 Rac8 27.
>>Qe2 Rc2 28. Rd2 Rxd2 29. Qxd2 Bf6 30. Be3 Be5 31. Bf4 Bxf4 32. Qxf4 Nf6 33.
>>Qf3 b2 34. Rb1 Rc8 35. Ne2 Rc4 36. g5 Nxe4 37. gxh6 Rxa4 38. hxg7 Nd2 39.
>>Qh5 Rh4 40. Rxb2 Qxb2 41. Qxh4 Qb1+ 42. Ng1 Qe4+ 43. Qxe4 Nxe4 44. Ne2 Kxg7
>>45. Kg2 Kf6 46. h4 Ke5 47. h5 Nf6 48. h6 Kxd5 49. Kf1 Ke4 50. Ke1 Nh7 51.
>>Kd2 Kf5 52. Kc3 Kg6 53. Kb4 Kh5 54. Nd4 Kg4 55. Ka5 Kf4 56. Ka6 Ke4 57. Ne2
>>Ke3 58. Ng3 Kf4 59. Ne2+ Kf3 60. Nd4+ Kf2 61. Nf5 Kf3 62. Nd6 Kf4 63. Nxf7
>>Kf5 64. Kxa7 Ke6 65. Ka6 Kxf7 66. Ka5 Ke8 67. Ka4 Kd8 68. Ka3 Kc8 69. Ka2
>>Kb8 70. Ka1 Ka8 71. Kb1 Kb8 72. Ka1 Ka8 73. Kb1 Kb8 74. Ka1
>>{Three fold repetition} 1/2-1/2
>>
>>
>>I believe that Horizon had a win in hand until this position:
>>
>>[D] 6k1/pq3pP1/8/3P4/r3n3/5Q2/1p2N2P/1R5K b - - 0 38
>>
>>where Horizon played 38...Nd2 with an eval score of more that +6.00.
>>
>>Here's two WinBoard analysis snapshots after 5 minutes running on an AMD 1.8 GHz
>>w/64 MB hash for Horizon and for Crafty.
>>
>>Horizon 3.2 (where +6.70 means Black is ahead)
>>depth=11 15/35 +6.70 Nd2 Qh5 Rh4 Rxb2 Qxb2 Qxh4 Qb1+ Ng1 Qxg1+ Kxg1 Nf3+ Kf2
>>Nodes: 77677043 NPS: 258071
>>Time: 00:05:00.99
>>
>>Crafty 18.15 (where -5.41 means Black is ahead)
>>depth=14 1/35 -5.41 1. ... Nd2 2. Qf5 Rh4 3. Rd1 b1=Q 4. Rxb1 Qxb1+ 5. Qxb1 Nxb1
>>6. d6 Rh6 7. d7 Rd6 8. Ng3 Nd2 9. h4 Kxg7
>>Nodes: 319129163 NPS: 1060758
>>Time: 00:05:00.85
>>
>>It is interesting to note that Crafty 18.15 endorses Horizon's move 38...Nd2 and
>>also scores it as Black ahead big-time.
>>
>>But when I analyzed using Chess Tiger and Fritz, I saw a different assessment.
>>(In both of the following analyses, the more negative the score the better it is
>>for Black.)
>>
>
><snip>
>
>I don't think it's an evaluation issue as much as a search issue - my latest
>beta modified Crafty likes "Nd2" until it sees something bad in the search and
>switches in big way to Qd5 - the Nd2 move drops to a draw score
It is also evaluation issue.
My movei always likes Qd5.
After Nd2 it can see also that black is better but the score is smaller.
and at depth 11 the score after Nd2 goes down to almost 0.00 score.
The score after Nd2 is:
0.01 for white at depth 5
1.25 for black at depth 6,7
1.45 for black at depth 8
2.19 for black at depth 9
3.06 for black at depth 10
0.08 for black at depth 11
after Qd5 I get
1.44 and not 0.01 for black at depth 5
1.39 and not 1.25 for black at depth 6
3.03 and not 1.25 for black at depth 7
3.17 and not 1.45 for black at depth 8
3.33 and not 2.19 for black at depth 9
3.91 and not 3.06 for black at depth 10
Uri
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.