Author: Vincent Diepeveen
Date: 17:12:38 08/21/02
Go up one level in this thread
On August 21, 2002 at 19:24:19, Uri Blass wrote: As a human i agree. i searched with a version with no dangerous extensions obviously. latest experimental one. can see it from the output even. just 2 k7 cpu's that's all. But if i go make a few moves i see a possible win is anything but easy. For sure it is deep and needing EGTBs (which i haven't turned on in this version of course as my windows gets too slow then and machne requires a reboot). My program is just evaluating it as i want it to evaluate it. The stupid rook on g2 however seems to assure white from losing here. Probably i misevaluate the thing there as being not too bad instead of being worth a pawn. However it is still far from easy to proof this to be a win if it is anyway. I will not draw conclusions about it. It is a typical computer position. Why do you care you played nxh4 here? To proof white to be not drawing is very deep. If diep would have a similar position like this in a game, the rook of white would be better placed. White has at least a draw then. Perhaps even a win. So i don't care basically despite the position being interesting for seeing the *entire* win for black. That's pretty deep however. I'm not going to modify my evaluation for a computer position. Best regards, Vincent >In the game horizon did not play Kg8 but played Kh6 that seems to be a winning >move. >Finding it or not finding it like avoiding Nxh4 in the previous diagram is a >question of evaluation. > >Crafty has no problem to do it because it evaluates black as better. >Deep Fritz can find Kh6 but only with a very small score. > >When I analyzed the game by going backward I could convince yace that black has >almost 1 pawn advantage. > >Movei 0.0.72h - Horizon 3.2 >[D]8/6rk/8/2p2Q2/2P1P2q/1P1P2p1/P5R1/3b2K1 b - - 0 1 > >Analysis by Deep Fritz: > >48...Kh7-g8-- > = (0.03) Depth: 1/3 00:00:00 >48...Kh7-g8-- 49.Qf5xc5 > ² (0.31) Depth: 1/6 00:00:00 >48...Kh7-g8 49.Qf5xc5 > ² (0.31) Depth: 2/8 00:00:00 >48...Kh7-g8 49.Qf5xc5 Bd1-f3 50.Qc5-d5+ Kg8-h7 > = (0.25) Depth: 3/11 00:00:00 >48...Rg7-g6! > = (0.22) Depth: 3/11 00:00:00 1kN >48...Rg7-g6! 49.Qf5xc5 > = (0.19) Depth: 3/11 00:00:00 1kN >48...Rg7-g6-- > ² (0.50) Depth: 4/10 00:00:00 1kN >48...Rg7-g6-- > ² (0.50) Depth: 4/14 00:00:00 2kN >48...Kh7-g8! > ² (0.47) Depth: 4/14 00:00:00 2kN >48...Kh7-g8! 49.Qf5xc5 > ² (0.44) Depth: 4/14 00:00:00 3kN >48...Kh7-g8 49.Qf5xc5 Rg7-h7 50.Qc5-c8+ Kg8-f7 51.Qc8-d7+ Kf7-g8 > ² (0.47) Depth: 5/14 00:00:00 7kN >48...Rg7-g6! > ² (0.44) Depth: 5/14 00:00:00 9kN >48...Rg7-g6! 49.Qf5xc5 > ² (0.41) Depth: 5/14 00:00:00 9kN >48...Rg7-g6 49.Qf5xc5 Bd1-f3 50.Qc5-f5 Bf3xg2 51.Qf5-d7+ Kh7-g8 52.Kg1xg2 > ² (0.34) Depth: 6/15 00:00:00 17kN >48...Rg7-g6 49.Qf5xc5 Bd1-f3 50.Qc5-f5 Bf3xg2 51.Qf5-f7+ Rg6-g7 52.Qf7-f5+ >Rg7-g6 53.Qf5-f7+ Rg6-g7 54.Qf7-f5+ > = (0.00) Depth: 7/19 00:00:00 51kN >48...Rg7-g6 49.Qf5-f7+ Rg6-g7 50.Qf7-f5+ > = (0.00) Depth: 8/22 00:00:00 204kN >48...Rg7-g6 49.Qf5-f7+ Rg6-g7 50.Qf7-f5+ > = (0.00) Depth: 9/23 00:00:00 352kN >48...Rg7-g6 49.Qf5-f7+ Rg6-g7 50.Qf7-f5+ > = (0.00) Depth: 10/25 00:00:01 1084kN >48...Rg7-g6 49.Qf5-f7+ Rg6-g7 50.Qf7-f5+ > = (0.00) Depth: 11/27 00:00:03 2110kN >48...Rg7-g6 49.Qf5-f7+ Rg6-g7 50.Qf7-f5+ > = (0.00) Depth: 12/31 00:00:07 4322kN >48...Rg7-g6 49.Qf5-f7+ Rg6-g7 50.Qf7-f5+ > = (0.00) Depth: 13/33 00:00:15 9679kN >48...Kh7-h6! > = (0.12) Depth: 13/35 00:00:31 19533kN >48...Kh7-h6! 49.Qf5-e6+ Rg7-g6 50.Qe6-e5 Kh6-h7 51.Qe5-c7+ Rg6-g7 52.Qc7xc5 >Rg7-f7 53.Qc5-e3 Bd1-f3 54.Qe3-d2 > = (-0.19) Depth: 13/35 00:01:04 38122kN >48...Kh7-h6 49.Qf5-e6+ Rg7-g6 50.Qe6-e5 Kh6-h7 51.Qe5-f5 Qh4-g5 52.Qf5xg5 Rg6xg5 >53.d3-d4 Bd1-f3 54.d4xc5 > = (-0.16) Depth: 14/37 00:01:45 65729kN >48...Kh7-h6 49.Qf5-e6+ Rg7-g6 50.Qe6-e5 Kh6-h7 51.Qe5-c7+ Rg6-g7 52.Qc7-e5 >Bd1-h5 53.a2-a3 > = (-0.16) Depth: 15/39 00:03:01 113863kN > >(Blass, Tel-aviv 22.08.2002) > > > >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.