Author: Uri Blass
Date: 10:51:21 11/15/02
Go up one level in this thread
On November 15, 2002 at 13:22:45, martin fierz wrote: >On November 15, 2002 at 12:41:19, Eduard Nemeth wrote: > >>On November 14, 2002 at 22:09:42, martin fierz wrote: >> >>>i was looking at eduard nemeth's page on computer chess and stumbled across the >>>following famous position: >>> >>>[D]1k6/8/8/8/8/5p1p/PPP5/6K1 w - - 0 1 >>> >>>white is a pawn up but will lose the game. i first saw this position about 20 >>>years ago when our coach showed it to us as kids - he let us choose which side >>>we wanted to play, and we always wanted white of course, being materialistic :-) >>>anyway, eduard says on his page that as long as he cared to look, fritz thought >>>white was winning. i let fritz7 run over night on my laptop, with the result >>>pasted below: it takes fritz over 10 hours on my old P3 450MHz to see that white >>>is worse, and after 24 hours the score is not yet showing that white is lost. >>>any engines do better on this? >>> >>>aloha >>> martin >>> >>>Analysis by Fritz 7: >>> >>>1.a4 Kc7 2.b4 Kc6 3.c3 Kd5 4.a5 h2+ 5.Kxh2 Kc6 6.Kg3 >>> +- (3.91) Depth: 10/18 00:00:00 58kN >>>1.a4 Kc7 2.b4 Kd6 3.c4 Kc6 4.a5 >>> +- (4.06) Depth: 11/17 00:00:00 73kN >>>1.a4-- >>> +- (3.78) Depth: 12/18 00:00:00 113kN >>>1.a4 Kc7 2.b4 Kc6 3.b5+ Kb6 4.c4 f2+ 5.Kxf2 h2 6.Kg2 h1Q+ 7.Kxh1 >>> +- (3.53) Depth: 12/18 00:00:00 128kN >>>1.a4! >>> +- (3.81) Depth: 13/20 00:00:01 191kN >>>1.a4! >>> +- (4.09) Depth: 13/20 00:00:01 207kN >>>1.a4 Kc7 2.b4 Kc6 3.a5 Kb5 4.c3 >>> +- (4.13) Depth: 13/20 00:00:01 227kN >>>1.a4 Kc7 2.b4 Kd6 3.a5 Kc6 4.c4 >>> +- (4.34) Depth: 14/21 00:00:02 344kN >>>1.a4 Kc7 2.a5 Kd6 3.b4 Kd5 4.a6 Kc6 5.b5+ >>> +- (4.13) Depth: 15/26 00:00:03 596kN >>>1.a4-- >>> +- (3.84) Depth: 16/27 00:00:05 1040kN >>>1.a4 >>> +- (3.84) Depth: 16/27 00:00:06 1083kN >>>1.a4-- >>> +- (3.56) Depth: 17/27 00:00:08 1464kN >>>1.a4 Kb7 2.a5 Ka6 3.b4 Kb5 >>> +- (3.50) Depth: 17/28 00:00:10 1671kN >>>1.a4-- >>> +- (3.22) Depth: 18/30 00:00:12 2366kN >>>1.a4 >>> +- (3.22) Depth: 18/30 00:00:13 2471kN >>>1.a4 Kb7 2.a5 Ka6 3.b4 Kb5 >>> +- (3.22) Depth: 19/32 00:00:17 3585kN >>>1.a4-- >>> +- (2.94) Depth: 20/33 00:00:22 5058kN >>>1.a4 Kb7 2.a5 Ka6 3.b4 Kb5 >>> +- (1.81) Depth: 20/33 00:00:26 6126kN >>>1.a4 Kb7 2.a5 Ka6 3.b4 Kb5 >>> +- (1.66) Depth: 21/36 00:00:38 9266kN >>>1.a4 Kb7 2.a5 Ka6 3.b4 Kb5 >>> +- (1.66) Depth: 22/39 00:01:02 16085kN >>>1.a4-- >>> ± (1.38) Depth: 23/40 00:01:34 24793kN >>>1.a4 Kb7 2.a5 Ka6 3.b4 Kb5 >>> ± (1.28) Depth: 23/42 00:02:56 51884kN >>>1.a4 Kb7 2.a5 Ka6 3.b4 Kb5 >>> ± (1.25) Depth: 24/42 00:04:15 74045kN >>>1.a4 Kb7 2.a5 Ka6 3.b4 Kb5 >>> ± (1.25) Depth: 25/44 00:06:46 116088kN >>>1.a4 Kc7 2.a5 Kb7 3.b3 Ka6 >>> ± (1.25) Depth: 26/47 00:19:01 320992kN >>>1.a4-- >>> ± (0.97) Depth: 27/49 00:38:23 644062kN >>>1.a4 Ka7 2.a5 Ka6 3.b4 Kb5 4.c3 Ka6 5.c4 Ka7 >>> µ (-1.22) Depth: 27/51 10:23:09 13057728kN >>>1.a4-- >>> -+ (-1.50) Depth: 28/52 10:42:25 13390390kN >>> >>>(Fierz, Honolulu 14.11.2002) >> >> >>See here: >> >>http://www.beepworld.de/members37/computerschach/tests.htm >> >>Eduard > >hi eduard, > >as i wrote, i found this position on your page and still decided to ask here, as >there is not much info on your page :-) >e.g. you did not wait to see how long fritz would take to see black was winning. > >anyway, my next question is obvious: why can fritz7, arguably one of the best >programs around, not see something in many hours that others can see in a few >seconds?? > >aloha > martin I think that it is obvious Different pruning rules. I believe that fritz's pruning rules are not good for this pawn endgame. I read that Shredder does even worse and see a clear win for white because it use null move pruning for this pawn endgame(I guess that it believe that zugzwang is impossible when there are a lot of passed pawns and in this case it is simply wrong) 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.