Re: [Bug-gnubg] Bug: MET files not of sufficient accuracy

[Joern Thyssen]

On Fri, Apr 16, 2004 at 07:31:26PM +0200, Joachim Matussek wrote
> Hello,
> 
> some days ago i wrote a script to produce dead and live cube takepoint
> tables depending on several match equity tables. As basis i used the
> MET files that were included with GNUBG. After having developed the
> script and having put the resulting data into an Excel sheet i found
> some strange behaviour. To avoid errors on my side i compared my
> results with those in GNUBG theory window (Analyse/Market window). 
> 
> I wrote a little article with some diagrams, an explanation and some
> conclusions. It is a zipped .pdf file. Its URL is
> 
> http://storm.prohosting.com/gnubg/met/gnubg_mets.zip

Something similar was discussed back in 2002 on the gnubg mailing list
(http://mail.gnu.org/archive/html/bug-gnubg/2002-12/msg00306.html).

BTW, a small correction: the Zadeh table is computer-generated from a
set of parameters. The algorithms are slightly more complicated than the
ones found in mec.c.

I've plans to rewrite mec.c so it can be compiled into gnubg the same
way as the Zadeh table is. So your suggestion was already on my internal
TODO list :-)

Your piece is quite interesting; it might be of value to include it as a
technical appendix in the gnubg manual. This requires that it's
rewritten into texinfo -- is this something you would like to do? (note
that your text will be published under the GNU FDL).

Jørn

pgpJeCOHbBMkw.pgp
Description: PGP signature