Author: Paul Jacobean Sacral
Date: 08:32:45 11/03/05
Go up one level in this thread
On November 02, 2005 at 16:42:44, James T. Walker wrote: >The 6 man tablebases indicate a draw after 58. ...Kxh6. At that point Fritz 8 >also indicates a -2.74 score. I don't see the problem since it ended in a draw >as it should. The problem is, if you analyse positions of that type - and maybe some are less easy to understand, like miniature studies etc. - with an engine and it responds with a 0.00 score, you have a strong hint that it's about a possible perpetual or a stalemate (or the 50 move rule), but not with evals like 2.x or 6.x which leave you "stumbling through the darkness" :) I tried the 5 piece position after 85...Ke6. I found only one engine in my small collection, which finds a 0.00 score quickly without tablebases (I skip some lines from the first second): K7/1r2r3/4k3/8/8/8/3R4/8 w - - 0 1 Analysis by Ruffian 1.0.5: 1.Rd1 -+ (-6.08) Depth: 1/1 00:00:00 1.Rd6+ Ke5 2.Rd5+ Ke4 3.Rd4+ Kf5 4.Rd5+ Kf4 5.Rd4+ Kg5 6.Rd5+ Kg4 7.Rd4+ Kf3 8.Rd3+ Kf2 9.Rc3 Rb4 10.Rd3 Re3 -+ (-5.90) Depth: 11/26 00:00:00 964kN 1.Rd6+ Ke5 2.Rd5+ Ke4 3.Re5+ Rxe5 4.Kxb7 Re6 5.Kc7 Ke5 6.Kd7 Ra6 7.Kc7 Rd6 8.Kb7 Kd4 -+ (-5.87) Depth: 12/29 00:00:01 2169kN 1.Re2+ Kd5 2.Rd2+ Ke5 3.Re2+ Kf5 4.Rf2+ Kg6 5.Rf6+ Kg5 6.Rf5+ Kg4 7.Rf4+ Kh5 8.Rf5+ Kh4 9.Rf4+ Kg3 10.Rf3+ Kg2 11.Rf2+ Kg3 12.Rf3+ -+ (-5.86) Depth: 12/30 00:00:01 2687kN 1.Re2+ -+ (-5.51) Depth: 13/30 00:00:01 3026kN 1.Re2+ Kd5 2.Rd2+ Ke5 3.Re2+ Kf5 4.Rf2+ Kg6 5.Rf6+ Kg5 6.Rf5+ Kg4 7.Rf4+ Kh5 8.Rf5+ Kh4 9.Rf4+ Kg3 10.Rf3+ Kg2 11.Rf2+ Kh1 12.Rf1+ Kh2 13.Rf3 Ra7+ 14.Kb8 Kg2 15.Rb3 Kf2 16.Kc8 Re8# -+ (-5.51) Depth: 13/30 00:00:01 3055kN 1.Re2+ Kd5 2.Rd2+ Ke5 3.Re2+ Kf5 4.Rf2+ Kg6 5.Rf6+ Kg5 6.Rf5+ Kg4 7.Rf4+ Kh5 8.Rf5+ Kh4 9.Rf4+ Kg3 10.Rf3+ Kg2 11.Rf2+ Kg1 12.Rb2 Ra7+ 13.Kb8 Kf1 14.Kc8 Re8# -+ (-5.32) Depth: 14/31 00:00:02 3242kN 1.Re2+ -+ (-4.97) Depth: 15/33 00:00:02 3494kN 1.Re2+ Kd5 2.Rd2+ Ke5 3.Re2+ Kf5 4.Rf2+ Kg6 5.Rf6+ Kg5 6.Rf5+ Kg4 7.Rf4+ Kh5 8.Rf5+ Kh4 9.Rf4+ Kg3 10.Rf3+ Kg2 11.Rf2+ Kg1 12.Rf1+ Kh2 13.Rf2+ Kg1 = (0.00) Depth: 15/33 00:00:02 3683kN 1.Re2+ Kd5 2.Rd2+ Ke5 3.Re2+ Kf5 4.Rf2+ Kg6 5.Rf6+ Kg5 6.Rf5+ Kg4 7.Rf4+ Kh5 8.Rf5+ Kh4 9.Rf4+ Kg3 10.Rf3+ Kg2 11.Rf2+ Kg1 12.Rf1+ Kh2 13.Rf2+ Kg1 = (0.00) Depth: 16/33 00:00:02 3923kN 1.Re2+ Kd5 2.Rd2+ Ke5 3.Re2+ Kf5 4.Rf2+ Kg6 5.Rf6+ Kg5 6.Rf5+ Kg4 7.Rf4+ Kh5 8.Rf5+ Kh4 9.Rf4+ Kg3 10.Rf3+ Kg2 11.Rf2+ Kg1 12.Rf1+ Kh2 13.Rf2+ Kg1 = (0.00) Depth: 17/33 00:00:02 4241kN 2 seconds, P4/3.0 GHz, 128 MB hash. I'm sure that 0.5 points are lost now and then, when an engine which is weak regarding this, misses the draw (win) and goes for another slightly higher evaluated continuation which turns out to be loss (draw) only. So it will matter in computer game results as well, but not often. I would guess, once in 300 games or less. Nevertheless it could decide about the exact ranking on a rating list, as the margins are small. Yours truly Paul J. Sacral
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.