Author: Uri Blass
Date: 13:19:26 11/11/01
Go up one level in this thread
On November 11, 2001 at 15:51:55, Sune Larsson wrote: >snip >>> Problem or not - Fritz7 doesn't seem to recognize the mate in 8, but >>> instead shows mate in 13 and this after several minutes thought. >>> The NoMMX version instantly sees the mate in 8. Just queer, or?! >>> >>> Sune >> >> >>Maybe Fritz7 is using a selective search for the stronger side. >>I am interested to know what is the evaluation of Fritz after 1 second. >> >>If it is +9 pawns or something similiar then there is no important problem. >>If it says +1 instead of mate then there is a problem because in a different >>position it may go for another move that gives it +1.2 and this move may lead >>to a draw. >> >>I do not use Fritz as a mate engine because I know from expereince cases when it >>believed that the distance to mate is shorter than the real distance to mate. >> >>Chessmaster in the middlegame or Crafty and yace in the endgame are engines that >>I can trust more for finding mates. >> >> >>Uri > > >Fritz 7 analysis looks like this on a P III 800, 128 MB hash. > > >New position >5Q2/1qr4p/2p2pp1/6k1/2B5/4PP2/5PKP/8 w - - 0 1 > >Analysis by Fritz 7: > >1.f4+ Kh5 2.Be2+ > +- (1.60) Depth: 7/23 00:00:00 151kN >1.f4+! > +- (1.88) Depth: 8/21 00:00:00 207kN >1.f4+! Kh5 2.Be2+ > +- (1.94) Depth: 8/27 00:00:01 443kN >1.f4+! > +- (2.22) Depth: 9/28 00:00:01 625kN >1.f4+! Kh5 2.Be2+ > +- (2.29) Depth: 9/29 00:00:03 1249kN >1.f4+! > +- (2.57) Depth: 10/28 00:00:04 1694kN >1.f4+! Kh5 2.Be2+ > +- (2.60) Depth: 10/30 00:00:05 2086kN >1.f4+! > +- (2.88) Depth: 11/27 00:00:08 3284kN >1.f4+! Kh5 2.Be2+ > +- (2.91) Depth: 11/32 00:00:10 4242kN >1.f4+! > +- (3.19) Depth: 12/33 00:00:18 7967kN >1.f4+! > +- (3.47) Depth: 12/33 00:00:24 10521kN >1.f4+! > +- (4.04) Depth: 12/33 00:00:25 10731kN >1.f4+! Kf5 2.Qc5+ Kg4 3.h3+ Kh4 4.Qf8 Qb5 5.Bxb5 > +- (#13) Depth: 12/38 00:02:57 79417kN > >(Larsson, KIL 11.11.2001) I guess that I know what happens. Fritz fails high and does not solve the fail high and go to analyze other moves. I think that it is usually not a practical probelm because if there was an alternative with a score of +3 in one of the first iteration Fritz could have another window to search and also could find f4+ in the next iteration. The only case when there is a practical problem in games is if Fritz fails high on another move in the last iteration but I think that this case is rare and it is possible(I did not write Fritz so I do not know) that in that case Fritz go back to the original move in order to solve the fail high. 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.