[Top][All Lists]

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

Re: [Bug-gnubg] GNU Backgammon Tutorial - Reduced Evaluation

From: nedc
Subject: Re: [Bug-gnubg] GNU Backgammon Tutorial - Reduced Evaluation
Date: Sat, 09 Oct 2004 14:33:44 -0700

On Sun, 10 Oct 2004 09:58:13 +1300
 Joseph Heled <address@hidden> wrote:
Funny that this comes up. I would not use any "reduced" setting at all. In fact I put reduced code under an ifdef which make it possible to comment it out, since I am working on adding the prune nets to GNUbg. They are generic (i.e. work for any ply), and give better results than any reduced setting).

More later, since I am student now, and this will take a little to get right.


Holger wrote:
On Sat, 9 Oct 2004 16:34:14 +0100 (GMT Daylight Time), Tim Line wrote:

In partnership with a team mate of mine Enrique (who also plays on MSN Zone under nic of BIG_buddie) we produced a GNU Backgammon tutorial - meant as a

Nice effort.

We would be very happy if you were to consider publishing a link to this
from your website.

Shouldn't be a problem. Oystein?
Btw, you could add http://www.gnubg.org/ to your page. ;)

We would also to be very happy to hear any comments that you might have or
any additions that we should add.

I'd not use 33% speed for checker play evaluation, but a full scan of rolls. For cube I've heard the opinion that this may not disturb the result much. The "supremo" setting for cube is exactly the same like "World class" since the only difference is the enlarged move filter. And I'm kind of missing the explanation for the settings.



Many of us use the reduced evaluation in our rollout settings, and find they significantly reduce the time for long rollouts in complex positions that benefit from 2-ply analysis.


reply via email to

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