[Top][All Lists]

[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Re: [Bug-gnubg] long simulation on matches

From: Holger
Subject: Re: [Bug-gnubg] long simulation on matches
Date: Wed, 03 Dec 2003 19:46:05 +0100

At 15:11 03.12.2003, Achim Mueller wrote:
* Martin Janke wrote on 28 Nov 2003:
> On Thu, 2003-11-27 at 16:23, Achim Mueller wrote:
> >  Is anybody able to write a small python script that starts a
> > new match, stores the result of each game (including
> > gammons/backgammons) and the result of the
> > match into a file, and finally starts a new match?
> I could do that. I've already got a script which does something similar.
> Exactly what pieces of information do you need in the log file and which
> format do you prefer for it?

I just checked the script here in the office, worked like a charme.
I will do some minor tests and then start a big session of 5 pointers
woolsey.xml vs. snowie.xml

What do you estimate, how many matches have to be played to get
a significant conclusion on

1. which met is better
2. building a new met bases on the results?

Actually I thought about 2ply (checkerplay/cubedecision). What about
0ply checkerplay and 2ply cube decisions to speed up things and get
more material?

You should contact Albert about this project. IIRC he's (was?) working on almost exactly the same thing with some world class player. Maybe you can combine forces. :)



reply via email to

[Prev in Thread] Current Thread [Next in Thread]