On Thu, Sep 10, 2009 at 09:18:40AM -0400, Jason House wrote: > Somewhat... One could generate a random number (r) and combine it > with the move mask (m) as follows: > black moves = m & r > white moves = m & ~r > > This has the drawback that the number of black and white moves may > not be equal. It can be modified to give an equal number of moves > such as requiring that each subset of n intersections generates the > same number of black and white moves, or doing some kind of looping > until the condition is met. > > For looping, you could keep some subset of the generated moves in > order to guarantee convergence. For example, if there are 14 more > white moves than black moves, keep all generated moves except for 14 > white moves. Then you only have to pick 14 moves in the next loop > iteration.
As a quick test mainly to test the speedup, I have made a simple hack that takes every other intersection (creating an X-pattern) and assigns it a randomized value (S_NONE,S_BLACK,S_WHITE). No check on number of stones is made. Now it thinks black has 46% win probability on empty board with no komi, for a speedup of 1000 playouts/s. So the speedup seems totally disappointing, and the bottleneck is in fact in the latter phases of the game (merging groups and capturing). Hmm, something to dig into. -- Petr "Pasky" Baudis A lot of people have my books on their bookshelves. That's the problem, they need to read them. -- Don Knuth _______________________________________________ computer-go mailing list computer-go@computer-go.org http://www.computer-go.org/mailman/listinfo/computer-go/