Hello, David Doshay wrote: > On 3, Feb 2007, at 2:51 AM, Sylvain Gelly wrote: > >> the speed of the best simulation policy in MoGo is 0.6 * the speed >> of the uniform random one. > > I think that this is very good. You give up less than a factor of 2 > from uniform random and you probably get better than a factor of 2 in > the % of relevant moves.
Is there any known (by theory or tests) function of how much a increase in the strength of the simulation policy increases the strength of the MC/UCT Program as a whole? Imagine you're using a almost random playout that has a strength, if playing alone, of say 10 ELO (just a number, I don't know how strong it really is). Using some thousand MC simulations with that 10 ELO playout and some other clever tricks like UCT, and you get a program that plays with 1500 ELO. Now what if you change the playout to one that has a strength, if playing alone, of 20 ELO, 50 ELO or 100 ELO. How whould that affect the strength of your overall programm, if you run the same number of MCs? Any known numbers on that? [I understand that a non-random playout would decrease the speed, so couln't run as many simulations as before in the same time, but let's imagine we get it for free.] eph _______________________________________________ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/
