bug-gnubg
[Top][All Lists]
Advanced

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

RE: [Bug-gnubg] smartest program


From: Ian Shaw
Subject: RE: [Bug-gnubg] smartest program
Date: Fri, 30 Nov 2007 13:54:57 -0000

        From:  Wojciech Zaremba
        
        Does exist in gnu backgammon options for better player than
grandmaster ?? It's not problem for me that program can compute to much
longer time.
        
Grandmaster is 3-ply evaluation. 
 
If you go to the Settings...Players menu, you will see that. Change to
User Defined and select a higher ply.
 
Be warned, 4-ply is very slow on my 2.3 GHz PC. I've never tried a
higher ply.

You can also play with the move filter settings. Accepting more moves at
greater tolerances will find a few moved that gnubg would initially
reject at first evaluation. This will increase the evaluation time at
each ply, but the benefit is usually very small because the basic
evaluations almost always include the best move near the top of the
list.

Maybe you have an incredibly fast computer; perhaps you're trying to do
something specific; most people get by on 0-ply or 2-ply for playing.
3-ply is useful in some situations - its evaluation of holding games
seems to be closer to rollout results than other plies. 4-ply is
generally only useable where the choices are very limited, such as
bearoffs.

I play against 2-ply, sometimes 0-ply if I want a faster game or an
opponent that will make a few errors. I analyse and roll out on 2-ply.

-- Ian





reply via email to

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