Author: Tim Foden
Date: 07:50:57 09/19/01
Go up one level in this thread
On September 19, 2001 at 08:33:51, Tim Foden wrote:
>On September 19, 2001 at 07:50:10, Heiner Marxen wrote:
>
>>On September 19, 2001 at 03:45:56, Tim Foden wrote:
>>
>>> 6k1/1pp2p2/p2p2q1/2PPb3/4r3/Pr4PK/2R5/2Q2NR1 b - - bm Rh4+; id "ECM.1179";
>>> 7 3.78 +Mate06 2624651 Rh4+ Kxh4 Qh7+ Qh6 Qxh6+ Kg4 f5+ Kxf5 Qh5+ Ke6 Qf7#
>>> {ht}
>>
>>Chest says:
>>6k1/1pp2p2/p2p2q1/2PPb3/4r3/Pr4PK/2R5/2Q2NR1 b - - acn 46263143; acs 558; bm
>>Rh4+; ce 32754; dm 7; id "ECM.1179"; pv Rh4+ Kxh4 Qh7+ Kg5 f6+ Kg4 f5+ Kg5 Kg7
>>g4 Qh6+ Kxf5 Qg6#;
>>
>>I have not yet looked closer at this one, but it appears that one of us has
>>got a bug: if your mate in 6 is correct, Chest should have found it.
>>Otherwise the mate in 6 just is not correct, and you have the bug. :-(
>>Not yet done with depth=8
>
>>Summarizing: we have one disagreement, which implies a bug.
>>I will inspect that later, and tell you, why Chest thinks that your
>>mate-in-6-PV is not correct.
>
>I agree with Chest. I don't believe that your program has a bug... I believe
>that mine does :-/
>
>I have suspected this since we had a similar disagreement about mates for one of
>Leonid's test problems. This was months ago. GLC also has a problem with
>Fine70. Somewhere there is a search/hashing bug in GLC, but I'm buggered if I
>can find it :) (and I have been trying to find it). I will keep looking though.
OK, I think I found the bug with respect to finding mates. GLC now says:
7 3.63 +Mate07 2517274 Rh4+ Kxh4 Qh7+ Kg5 f6+ Kg4 f5+ Kg5 Kg7 cxd6 Qg6+
Kh4 Qg4#
Which agrees much better with Chest.
However, the fix I made only affects CheckMates, so Fine70 still has a bug for
me. :)
Cheers, Tim.
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.