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

RE: [computer-go] SlugGo v.s. Many Faces update




> -----Original Message-----
> From: computer-go-bounces@xxxxxxxxxxxxxxxxx 
> 
> I am very interested.  I have looked around for information 
> on how to do Joseki/Fuseki and I can find nothing that 
> explains it in any technical detail.  Either I'm not looking 
> in the right place or it's one of those little secrets the Go 
> programmers don't share.

No secret.  Many Faces is not very complicated.  I have a joseki database
with about 60,000 moves in it
that I entered by hand from go books (mostly on airplanes during business
trips in the 1980's).  It's
stored as a DAG, really just a tree with a small number of crosslinks.  You
can explore the database using the Joseki Tutor function.

If there is a good joseki move in a corner, I don't consider any other
moves.  The alpha-beta search plays out the most popular sequnces to the end
(perhaps 5 to 20 ply), and does a full board quiescence search. 

I have a full board database created automatically from about 10,000 pro
games and 50,000 strong amateur games.  It's also stored as a tree, with the
board in a canonical orientation.  It stores the number of times each
position was reached and the number of wins for each side.  You can explore
the database using the fuseki tutor function. 

Every game it plays is added to the fuseki database to prevent it from
losing the same way twice.

Database moves are generated, and win/loss ratios have at term in the
evaluation function.

There is a few hundred lines of code to determine the best extensions along
the sides, depending on the strength of the group on either end, and the
location of the stone (3rd ine or 4th line).  The standard extensions are
pretty sinple.  You can ask any strong player to explain them.

Standard extensions are generated, and get an evaluation bonus.

That's it.

David Fotland

> 
> I will probably look at gnugo to see what they do about 
> openings, but I would be interested in what you (or anyone 
> else) has tried.
> - Don
> 
> 
> _______________________________________________
> computer-go mailing list
> computer-go@xxxxxxxxxxxxxxxxx 
> http://www.computer-go.org/mailman/listinfo/computer-go/
> 


_______________________________________________
computer-go mailing list
computer-go@xxxxxxxxxxxxxxxxx
http://www.computer-go.org/mailman/listinfo/computer-go/