Re: [Bug-gnubg] standard error in cubefull 2ply rollouts

[Achim Mueller]

* Jim Segrave <address@hidden> [070609 18:16]:
> On Sat 09 Jun 2007 (08:35 +0200), Achim Mueller wrote:
> > Hi all,
> > 
> > can someone please explain me the following numbers:
> > 
> > 
> > posID: a9skAAzHeTsAAA matchID: MBkgAlAACAAA
> > 
> > Cube analysis
> > Rollout cubeless equity  +0.566 (Money:  +0.566)
> > 
> > Cubeful equities:
> > 1. No double            +1.010
> > 2. Double, take         +1.079  ( +0.070)
> > 3. Double, pass         +1.000  ( -0.010)
> > Proper cube action: Too good to double, pass (12.3%)
> > Rollout details:
> > Centered 1-cube:
> >   0.779 0.016 0.000 - 0.221 0.008 0.000 CL  +0.566 CF  +1.010
> >  [0.019 0.001 0.000 - 0.019 0.000 0.000 CL   0.014 CF   0.442]
> > Player ace owns 2-cube:
> >   0.783 0.015 0.000 - 0.217 0.007 0.000 CL  +1.198 CF  +1.079
> >  [0.001 0.001 0.000 - 0.001 0.000 0.000 CL   0.004 CF   0.005]
> > Full cubeful rollout with var.redn.
> > 1296 games, Mersenne Twister dice gen. with seed -715772175 and
> > quasi-random dice
> > Play: world class 2-ply cubeful prune [world class]
> > keep the first 0 0-ply moves and up to 8 more moves within equity 0.16
> > Skip pruning for 1-ply moves.
> > Cube: 2-ply cubeful prune [world class]
> > 
> > Why is the standard error for Centered 1-cube (cubefull) at 0.442?
> 
> Can you send me the .sgf file for this rollout. It contains the
> interim values from the rollout.

Unfortunately I only have the html file, but I probably can reproduce
it. Did you check using matchID and posID for your own? It shlod give
the same results.

It happened on a SuSE-10.1 with the snapshot from 04.06.2006.

Ciao

Achim