In othe rparts of the code (cubeful match equity computations via janowski formula) the fact that a double is followed by an automatic redouble is taken into account directly, without the need to go to 1ply. That's why I'd say 0ply should get the d/t equity right: tafter a D/T the redouble is automatic, the position is hence cubeless and it is in the 2s dB: no need to eval (neither for 0ply, nor for 1/2/n ply), if th e2s dB is in use.
2009/9/8 Neil Robins <[email protected]> > Well, it would be good if 0-ply did get it right, but it s only obvious to > me that it should definitely be right after at least a 1-ply lookahead which > sees that after any throw it is a recube for the match and take. >
_______________________________________________ Bug-gnubg mailing list [email protected] http://lists.gnu.org/mailman/listinfo/bug-gnubg
