Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Bug in Nimzo 7.32 engine?

Author: Dann Corbit

Date: 22:31:19 09/23/99

Go up one level in this thread


On September 24, 1999 at 00:45:45, Steve Maughan wrote:
>I gave Nimzo 7.32 the following test position
>
>r2b1qrk/ppp3pp/3p1p2/4pPPQ/4P2P/3PB3/PPP5/1K4RR w - - 0 1
>
>It's mate in 4 via 1.Qxh7+ Kxh7 2.g6+ Kh8 3.Rg5 fxg5 4.hxg5 mate.  The position
>is simple for todays programs to solve.
>
>However giving Nimzo the white pieces it immediately announces mate in 2!!  Now
>I can accept that it is reasonable to announce mate in more than 4, but not
>less.  Also after playing 2.g6 it announces mate in 1, and then after Rg5 and
>hxg5 it claims mate in 0.
>
>I assume that this is a minor bug in the engine.  It really should be sorted out
>as it means that Nimzo cannot be trusted to give shortest mates.
As a rule, chess engines cannot be trusted to give shortest mates.  That job
goes to mate finders.

It is clearly a bug in the engine.  It also casts aspersions on test results
with that engine if Nimzo announces a mate and the other engine does not see it.

I suggest that you notify the manufacturer.

Here is the output of the mate finder Chest:
H:\foobart\sub\Release>chest < a
CHEST version 3.15, 28-Dec-1998
Options =
Input file: STDIN
Reading job:
W:  Kb1 Qh5 Rg1 Rh1 Be3 Pa2 Pb2 Pc2 Pd3 Pe4 Pf5 Pg5 Ph4 (13)
B:  Kh8 Qf8 Ra8 Rg8 Bd8 Pa7 Pb7 Pc7 Pd6 Pe5 Pf6 Pg7 Ph7 (13)
FEN: r2b1qrk/ppp3pp/3p1p2/4pPPQ/4P2P/3PB3/PPP5/1K4RR w - -
analysing (mate in 4 moves):
Solution (in 4 moves):
        Qh5 : h7
refu  1: Qh7:+  Kh7:   [  2-]
solu         6: Kh7:   [  3+]
refu  2: g6     h6     [  3-]
solu         1: Qf7    [  1+]
solu         2: Rb8    [  1+]
solu         3: Rc8    [  1+]
refu  3: gf6    Qf7    [  3-]
refu  4: d4     Qf7    [  3-]
refu  5: c3     Qf7    [  3-]
refu  6: c4     Rc8    [  3-]
solu         4: fg5    [  2+]
refu  7: b3     Qf7    [  3-]
refu  8: b4     Qf7    [  3-]
refu  9: a3     Qf7    [  3-]
refu 10: a4     Qf7    [  3-]
refu 11: Bf4    Qf7    [  3-]
refu 12: Bd2    Qf7    [  3-]
refu 13: Bc1    Qf7    [  3-]
refu 14: Bf2    Qf7    [  3-]
refu 15: Bd4    Qf7    [  3-]
refu 16: Bc5    Qf7    [  3-]
refu 17: Bb6    Qf7    [  3-]
refu 18: Ba7:   Qf7    [  3-]
refu 19: Rh2    Qf7    [  3-]
refu 20: Rh3    Qf7    [  3-]
refu 21: Rf1    Qf7    [  3-]
refu 22: Re1    Qf7    [  3-]
refu 23: Rd1    Qf7    [  3-]
refu 24: Rc1    Qf7    [  3-]
refu 25: Rg2    Qf7    [  3-]
refu 26: Rg3    Qf7    [  3-]
refu 27: Rg4    Qf7    [  3-]
refu 28: Qh6    Qf7    [  3-]
refu 29: Qg4    Qf7    [  3-]
refu 30: Qf3    Qf7    [  3-]
refu 31: Qe2    Qf7    [  3-]
refu 32: Qd1    Qf7    [  3-]
refu 33: Qg6    Qf7    [  3-]
refu 34: Qf7    Qf7:   [  3-]
refu 35: Qe8    Qf7    [  3-]
refu 36: Kc1    Rb8    [  3-]
solu         5: fg5    [  2+]
refu 37: Ka1    Qf7    [  3-]
Time (user) = 1.00 sec
Reading job:

BTW, Crafty also finds it.
PFGA: EPD record: 1
middle-game phase
              clearing hash tables
              time surplus   0.00  time limit 600:00 (600:00)
         nss  depth   time  score   variation (1)
White(1): exit

                5     0.18  Mat04   1. Qxh7+ Kxh7 2. g6+ Kh8 3. Rg5 fxg5
                                    4. hxg5#
                5->   0.18  Mat04   1. Qxh7+ Kxh7 2. g6+ Kh8 3. Rg5 fxg5
                                    4. hxg5#
                6     0.20  Mat04   1. Qxh7+ Kxh7 2. g6+ Kh8 3. Rg5 fxg5
                                    4. hxg5#
                6->   0.22  Mat04   1. Qxh7+ Kxh7 2. g6+ Kh8 3. Rg5 fxg5
                                    4. hxg5#
                7     0.23  Mat04   1. Qxh7+ Kxh7 2. g6+ Kh8 3. Rg5 fxg5
                                    4. hxg5#
                7->   0.26  Mat04   1. Qxh7+ Kxh7 2. g6+ Kh8 3. Rg5 fxg5
                                    4. hxg5#
              time=0.26  cpu=100%  mat=0  n=41489  fh=94%  nps=159573
              ext-> checks=831 recaps=239 pawns=149 1rep=288 thrt:80
              predicted=0  nodes=41489  evals=6575
              endgame tablebase-> probes done=0  successful=0



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.