Re: [Bug-gnubg] Re: more on bearoff databases

[Neil Kaz]

New comments prefaced by (KAZ)
----- Original Message -----
From: "Joern Thyssen" <address@hidden>
To: "Neil Kaz" <address@hidden>
Cc: "GNU Backgammon Bugs" <address@hidden>
Sent: Sunday, October 27, 2002 1:06 PM
Subject: Re: [Bug-gnubg] Re: more on bearoff databases


> On Sun, Oct 27, 2002 at 11:46:09AM -0600, Neil Kaz wrote
> > Embedded comments inside prefaced by (NK)
> > ----- Original Message -----
> > From: "Joern Thyssen" <address@hidden>
> > To: "GNU Backgammon Bugs" <address@hidden>
> > Cc: "Neil Kaz" <address@hidden>
> > Sent: Sunday, October 27, 2002 4:06 AM
> > Subject: more on bearoff databases
> > >
> > > #pts #positions size(unc) estimated size comp
> > > 11    7,726,160       988,948,480 250 MB
> > > 12 17,383,860 2,225,134,080 550 MB
> > > 13 37,442,160 4,792,596,480   1200 MB
> > > 14 77,558,760 9,927,521,280 2500 MB
> > >
> > > So the practiacal limit appears to be the 12-pt or perhaps the 13-pt
> > > databases. So if we want to go further we have to do something else:
> > >
> > (NK) I'd stop at the 13 point since this is where so many BG position
become
> > straight races. Even most midpoint to midpoint contact is essentially a
> > straight race, but we may wish to ignore the database for playing with
any
> > midpoint contact. However, if there are a couple other checkers for both
> > sides in the outboard, the database evaluation of racing chances with
> > midpoint contact should be fine for cube evaluation purposes.
> >
> > I'd stop the database at the 13 point (mid point) and add only the
following
> > expansion which would continue the database further out with 1 or 2
checkers
> > only on points higher than the midpoint. This will be used only in non
> > contact and will not require that many more positions (or will it)
>
> Unfortunately there are 10M positions with 13 chequers on 13 points, and
> 15 positions with 2 chequers on five points, hence a total of 150M
> positions.

 (KAZ) OK I was afraid of something like that so lets just stop at the 13
point !

> > (NK) I don't see how a normal distribution can pick up on wastage
effects
> > or tell GNU how to properly play during the bear in.
>
> I keep the mean and the variance for the roll distribution, so I think
> wastage effects are accounted for.

(KAZ) I think I misunderstood. I trust that this is from a database ? I'm
confused here.

> If you have some bear-in examples that any bots misplay, I can check with
> my bearoff databases?! So far I just have the 10 pt databases.

(KAZ) I use Snowie 1's which goes out to the 11 point so I'll send along a
few examples of GNU's errors in bearin to you.

> 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!
>