_______________________________________________ Bug-gnubg mailing list [email protected] https://lists.gnu.org/mailman/listinfo/bug-gnubg
Philippe, without the inversion, my hand calculation matches the gnubg
output in non-VR rollouts. What I did with the second position (with MET
inversion) I posted was this: I rolled it out for one given seed, first
for 1, then for 2, then for 3 trials and so on. The equity in the output
matched my hand calculation exactly until the first gammon appeared in
the cube statistics (for what it's worth, it was an undoubled gammon for
the 6-away shark, and he seemed to get 48.29% MWC for that when it
really should have been 46.9%. A single win on the 2-cube was handled
correctly, though.)
