[Top][All Lists]

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

[Bug-gnubg] Re: race net: 0ply and 1ply

From: Joseph Heled
Subject: [Bug-gnubg] Re: race net: 0ply and 1ply
Date: Wed, 27 Nov 2002 08:12:42 +1300
User-agent: Mozilla/5.0 (X11; U; SunOS sun4u; en-US; rv:1.1) Gecko/20020829

Moore, Dave wrote:
The discontinuities at odd plys for gnubg have always seemed more than a bit
strange to me.  I'm a much better computer scientist than backgammon player,
but something just seems off there.  Alas, I have no spare time (only spare
CPU cycles) to back this up with contributions to the source code!

I agree. I was as baffled as you, but now I have part of the answer. I am trying to write this down, but it is not easy explaining it (my writing skills are up to the task)

Would this net perform even better using the new movefilter code being added
to gnubg, which might let it "skip" over 1-ply pruning, thereby helping
2-ply even more?

This is the whole idea behind the movefilter. 2ply is stronger now, of that I am sure.

To answer the actual question, I'd be fine with a better 0-ply to play
against, and a better 2-ply for my analyses.


-----Original Message-----
From: Joseph Heled [mailto:address@hidden
Sent: Tuesday, November 26, 2002 1:56 PM
To: gnubg; Gary Wong; Øystein O Johansen; Joern Thyssen; Achim Mueller;
montgomery; max-d; Jim Segrave; Nardy Pillards; Ian Shaw; Steve J.
Baedke; Tore Saanum; MSquonk; Moore, Dave; Albert Silver; "Pål Schrader"
Subject: race net: 0ply and 1ply

I now have a race net whose 0ply move error (as given by the race benchmark which you kindly helped to roll) is 62% of the current net. The 2ply error is 90% (i.e. 10% better). Alas, 1ply error is twice as large!

This is a somewhat bizarre result. I think I know in part why and will try to solve this, but it seems that improving the 1ply will degrade the 0ply.

This raises a theoretical question. would YOU rather use a net with better 0ply and 2ply, or a lesser one with a better 1ply?

Personally, I never use 1ply anymore. I use 0ply for quick, 2ply for more quality, but I would like your opinion.


reply via email to

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