Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Another Fritz 8.0.0.23 engine bug - Oops, corrected diagram

Author: Kurt Utzinger

Date: 15:12:22 05/07/03

Go up one level in this thread


On May 07, 2003 at 17:45:21, Dan Simmons wrote:

>In the diagram, it is White to move and win with the concept of interference in
>this composition.
>
>[D] 8/6b1/5k2/8/P3K1P1/8/8/8 w - - 0 1
>
>My solution involved 1.g5+ Kxg5 2.a5 Bf8 3.Kd5 with a white win.  It didn’t
>occur to me until after I saw that I had missed the correct solution that maybe
>Black doesn’t have to recapture on his first move but can instead try 1…Ke6.
>Well, let’s see what the grandmaster computer has to say.
>
>I set up the position in F8 (ver 23) on a PIII, 1.2Ghz, 64MB hash.  I played
>1.g5.  Fritz immediately selects 1…Kxg5 and never changes for at least 15
>minutes of thinking, declaring the position dead even after 2. a5 Bf8 3. a6 Bc5.
> Fritz gives equal weight (0.00) to 1…Ke6 but selects Kxg5.
>
>Say what?  After 15 minutes, I couldn’t find Fritz8 even considering 3.Kd5 in
>its search on move 1.  So I allowed Fritz to play 1…Kxg5 and I played 2.a5.
>Despite the time control being 20 minutes/move,  Fritz plays 2…Bf8 after seven
>seconds and declares the position a draw, only considering 3.a6 in its search
>window.  I then play 3.Kd5 and Fritz’ evaluation  shows a win for white in 2
>seconds.
>
>I then moved back to the position after 2.a5 and set time for infinite.  Fritz
>only thinks for seven seconds then the engine clock (as opposed to the game
>clock) stops and no further searching is apparently done.  At this point, Fritz
>is still evaluating as a draw with 2…Bf8 3.a6 Bc5.
>
>Thinking that was a freaky bug, I then went back to the original position and
>allowed Fritz to analyze.  It fails to find the correct solution 1.a5 Bf8 2.Kd5
>Bh6 3. g5+ Bxg5 4.Ke4 Bh4 5.Kf3 with a white win.
>
>Fritz selects 1.a5 but then selects the drawing 2.a6? after black’s Bf8.  Per
>the search window, F8 never considers Kd5.  I then advance the solution another
>move but Fritz as white again fails to even consider the simple king move that
>wins:  4.Ke4.
>
>Interestingly, only after Fritz8 is forced to play 2.Kd5 and 4.Ke4 does it see
>the win.  After 4.Ke4, it finds the win immediately in 00:00
>
>Is this a Fritz problem?  I dunno.  In the same GUI, Fritz 5.32 finds the win
>immediately from the original position and has no problem rejecting 1…Kxg5 after
>my chosen 1.g5+?.
>
>Unfortunately, I am on my laptop which has only 8.0.0.23 rather than my desktop
>which also has 8.0.0.8 to see if there is a difference in this bug between the
>versions.
>
>Comments?


     Hi Dann
     Same behaviour with Fritz 8008 without EGTB but of course correct
     solution 1.g5? Ke6! draw with 5-men EGTB. In the above position
     White wins with 1.a5! Bf8 2.Kd5!
     Kurt





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.