Petr, Google for "Monte Carlo Tree Search Solver", a very nice 2008 paper by Mark Winands from Maastricht University (or use the link provided by Darren Cook in another thread). It deals with exactly this situation.
Or, for a simpler solution, you can use the original suggestion by Magnus Persson (author of Valkyria) and update the tree with the (exact) result of this node as if it came from a very large number of simulations (I belief Magnus used 10,000 at the time). That reduces the uncertainty (exploration) term and should reduce the number of times it gets chosen. René 2010/10/28 Петр Смолов <[email protected]> > Hello all! > > How do you think what the program should do if the game is over in the Tree > Policy, not in the Default Policy? Do we have to make the program not to > select this node any more (not to call procedure PlaySimulation for this > node)? > > > _______________________________________________ > Computer-go mailing list > [email protected] > http://dvandva.org/cgi-bin/mailman/listinfo/computer-go >
_______________________________________________ Computer-go mailing list [email protected] http://dvandva.org/cgi-bin/mailman/listinfo/computer-go
