Re: [Bug-gnubg] Rotating rolls in rollout

[Robert-Jan Veldhuizen]

I was wondering whether it would be possible to use the same idea of 2-ply 
33% speed (and other speeds) for rotating the first two rolls when you use 
n*36 trials (MOD 1296), with n=1,2,..,35?

So, a 432 trials rollout would be somewhat similar to a 33% speed 2-ply 
evaluation, with the second rolls picked by some algorithm that spreads the 
equities somehow, like 33% speed does with evaluation?

Is something like that possible, or would it take too much time, or? Any 
ideas on this?


--
Robert-Jan (Zorba on FIBS)