Re: OSR [Was: Re: [Bug-gnubg] call for cycles (unix/libux only for now)]

[Joseph Heled]

Joern Thyssen wrote:
> 
> > When cube > 1, My cube code estimates the equity by  'a * live + (1-a) * 
> > dead'
> > For contact positions, the best empirical value I found is a = 0.7
> 
> This is identical to gnubg (except that gnubg uses 0.68 for contact,
> 0.60 for BEAROFF1 and BEAROFF2, and some linear interpolation for
> CLASS_RACE).
> 
> > For race, I developed another model
> >
> >   a = M * P * I
> >
> >   M = a magic scaler (0.0606)
> >   P = average number of rolls to race end.
> >       (Take the smaller value computed for the two players.)
> >   I = redouble incentive
> 
> It was something like that I was looking for.
> 
> > P is obtained from a modified pip count. Wastage is accounted for by
> > adding 2.5 points for every checker on the 1 point, 2 for checkers on
> > the 2 point etc., and the total is divided by 8.16
> 
> If 'a' is to be used with a OSR, the average number of rolls from the
> OSR or BEAROFF2 can be used.
> 
> > I is 'G/L', where G (gain) is (equity after redouble - equity at
> > current cube) when he wins, and L is same when he loses. I tried to
> > capture the "eagerness" of the cube holder to redouble.
> 
> So at -4, -5 with a centered cube:
> 

Not exactly. Note the (When cube > 1) at the very beginning. This does not apply
for centered cube

> G = MWC(-2,-5) - MWC(-3,-5) ~= 75% - 66% = 9%
> L = MWC(-4,-4) - MWC(-4,-3) ~= 50% - 41% = 9%
> 
> So I = 1 in this case. Also, for money I = 1.
> 
> At -3, -4 the same calculation gives: I = 1.5
> 
> I think this should be easy to implememt in gnubg as well.
> 

For centered cube, I use  'a * live + (1-a)/2 * (dead-for-player1 +
dead-for-player-2)'

a = 0.65 is the best I know of. 

I tried to improve this generic case for race and failed.

-Joseph