Re: [Bug-gnubg] Re: Confidence intervals from rollouts

[Joern Thyssen]

On Thu, Sep 05, 2002 at 12:25:09PM -0700, David Montgomery wrote
> > Actually thrice :-)
> 
> As long as we are counting, we should count Fredrik Dahl, 
> who I believe did it first, for a total of four. :-)
>  
> > Gary's rollout code from rollout.c has:
> > 
> >   for( i = 0; i < cGames; i++ ) {
> >      if( prc->rngRollout != RNG_MANUAL )
> >         InitRNGSeed( rcRollout.nSeed + ( i << 8 ), prc->rngRollout );
> > 
> > Since gnubg doesn't change the seed between rollouts of different moves
> > the same dice sequence will be used for each game.
> 
> This gives duplicate dice, but unless I'm mistaken it
> doesn't do stratified sampling (perfect distribution
> of rolls at every depth).  

You're not mistaken :-) I think it would require quite some work to
implement that.

Jørn

-- 
Joern Thyssen, PhD
Vendsysselgade 3, 3., DK-9000 Aalborg, Denmark
+45 9813 2791 (private) / +45 2077 2689 (mobile) / +45 9633 7036 (work)