Computer Chess Club Archives


Search

Terms

Messages

Subject: Re: Fritz endgame CD

Author: José de Jesús García Ruvalcaba

Date: 07:47:46 03/30/99

Go up one level in this thread


On March 30, 1999 at 08:38:23, James T. Walker wrote:

>On March 30, 1999 at 05:24:39, Sylvain Renard wrote:
>
>>>>I found a position when Junior (using tablebases) goes wrong in the winning side
>>>>of KRP vs KR
>>>>6R1/7k/5P2/8/8/8/7r/K7 w - - 0 1
>>>>Fritz's tablebases show the right winning move 1.f7 (8) (I do not understand
>>>>8) but after this move the tablebases show for black Rf2 with draw.
>>  Yes, good work! But this is kind of error is easy to understand
>>because on Fritz CD ROM there are the Ken Thompson's files : Pr_r.g,
>>Rb_r.g, Rn_r.g but the file Rr_r.g is missing. After 1...Rf2, the only
>>winning move is 2.f8R!! all other moves lead to a draw.
>>
>>>At that point white has a "Won" endgame and abandons the CD ROM because it
>>>believes the program can win the game from there.
>>No it is because it does not have the good file on the CD! The file Rr_r.g
>>is missing.
>>
>>>Yes, this is an error in the tablebase.
>>Because it is uncomplete.
>>
>**********
>Hello Sylvain,
>I don't think the missing tablebase should matter !  If your tablebase would
>simply tell your program to promote to a Rook from that point on it is a won
>game and no tablebases are needed.  The error in the tablebase is in not
>promoting to a rook.  When it promotes to a bishop it is drawn.  That is the
>error.  Most any program can win  2 rooks vs 1 rook or KR vs K.
>Jim Walker
>**************
>

	But tablebases do not tell you what to do, they do not contain best moves. I
think any program with missing tablebases (promotion or capture classes of other
files they have) will show strange behavior. In this case the Fritz CD has KRP
vs KR, but not KRR vs KR which is a promotion class of the former.

>
>>>>I am not sure about the result of 2.f8Q, the only question is if black has a
>>>>perpetual check and I did not analyze this position and did not check it with
>>>>KQR vs KR tablebases.
>>If you have Ken Thompson CD N°1 (and for instance ChessBase) you
>>can check it's a draw.
>>
>>>Yes, I believe the 2. f8=Q is a draw because of moves like Ra2+.
>>>Jim Walker
>>It is confirmed both by Ken Thompson database (when complete...)
>>and Edward's tablebases (and human thinking too!).
>>
>>My program Capture uses Edward's tablebases. It plays 1.f7 and
>>announces mate in 20 moves. I have also checked that after
>>1.f7 Rf2 2.f8Q?? it plays 2...Ra2+ and announces a draw.
>>
>>  Best regards,
>>Sylvain Renard



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.