Re: [Bug-gnubg] Re: Race database and bearoff DB

[Joern Thyssen]

On Tue, Oct 01, 2002 at 02:30:41PM +0200, Øystein O Johansen wrote
> 
> > I have ported Joseph's OSR code to a local copy of gnubg,
> > but I never got around testing it.
> 
> Which bearoff database do you use for this? The problem with
> the porting is that bearoff databases. I ported this as well
> but I had to use the bearoff database from Joseph, and then
> where was suddenly two bearoff databases for one program.
> 
> (BTW: Should we maybe find a better encoding for the gnubg
> bearoff database? We don't want to store all these zeros.)

If we try to compress it, it'll be much harder to read it. We can easily
compress the current gnubg.bd since it's read into memory and we can
just read it with functions from libz. However, for the new gnubg_os.bd
it'll be more complicated since we can't read it into memory. We have to
use something like fseek to find bearoff equities. Perhaps there is a
libz_fseek... I'll check.

> > Maybe it's possible to make another OS bear-off DB that goes
> > even further if we restrict it to, say, 10 chequers per point.
> 
> Will it be any interest of that?

Many of these positions in the bearoff DB are quite esoteric, e.g., 15
chequers on the 10 point. Maybe the space is better used for more
frequent positions, like 3 chequers on the 13-point plus 5 chequers on
the 6 point etc.

> I've heard of this, but I don't know/understand how it works.
> Joseph? Is this code in fibs2html?

The basic idea is that the cube efficiency has a  simple relation to the
average number of rolls left. The match score is takn into account with
some factors.

> 
Jørn

-- 
Joern Thyssen, PhD
Vendsysselgade 3, 3., DK-9000 Aalborg, Denmark
+45 9813 2791 (private) / +45 2818 0183 (mobile) / +45 9633 7036 (work)
Note: new mobile number!