Re: [Bug-gnubg] Eval

[Joern Thyssen]

On Sat, Mar 08, 2003 at 04:55:10PM +0000, Joern Thyssen wrote
> On Sat, Mar 08, 2003 at 01:18:45PM -0300, Albert Silver wrote
> > There's something wrong with GNU's eval. Take the following position and
> > look at the respective evaluations at 0, 1, 2, and 3 plies.
> 
> What's wrong? This is obviously a position that neural net doesn't
> grasp, hence the huge difference between odd and even plies. 

Hmm, on linux I get very different results from you! Can anyone confirm
Albert's results?

Jørn

My result (gnubg-devel, CVS March 8 2003):

Cube analysis
0-ply cubeless equity  -0.664
  0.411 0.016 0.000 - 0.589 0.388 0.116
Cubeful equities:
1. No double            -1.000
2. Double, pass         +1.000  ( +2.000)
3. Double, take         -1.729  ( -0.729)
Proper cube action: No double, beaver (26.7%)



Cube analysis
1-ply cubeless equity  -0.683
  0.409 0.012 0.000 - 0.591 0.389 0.124
Cubeful equities:
1. No double            -0.990
2. Double, pass         +1.000  ( +1.990)
3. Double, take         -1.981  ( -0.990)
Proper cube action: No double, beaver (33.2%)



Cube analysis
2-ply cubeless equity  -0.706
  0.399 0.011 0.000 - 0.601 0.393 0.122
Cubeful equities:
1. No double            -0.915
2. Double, pass         +1.000  ( +1.915)
3. Double, take         -1.878  ( -0.962)
Proper cube action: No double, beaver (33.4%)



Cube analysis
3-ply cubeless equity  -0.654
  0.414 0.010 0.000 - 0.586 0.374 0.118
Cubeful equities:
1. No double            -0.973
2. Double, pass         +1.000  ( +1.973)
3. Double, take         -1.945  ( -0.973)
Proper cube action: No double, beaver (33.0%)