On Tue, Sep 28, 2010 at 1:09 AM, Petr Baudis <[email protected]> wrote: > On Tue, Sep 28, 2010 at 07:14:46AM +0900, Hideki Kato wrote: >> steve uurtamo: >> <[email protected]>: >> >a draw system would be pretty interesting. >> >> GPW Open tournament, a night event at Game Programming Workshop, uses >> 7.0 pt komi and allows draw since last year. >> GPW-10: <http://sig-gi.c.u-tokyo.ac.jp/gpw/2010/index-e.html> > > Are there any publications (scientific papers or mailing list posts or > whatever) on adapting MCTS to jigo results? So far, I have only heard > that it is surprisingly annoying problem to tackle, though I have to > admit I did not think about it much myself.
Not sure of any publications, but I don't think it's difficult or annoying. Steenvreter uses 3 possible values for the simulation results (win +1, draw 0, loss -1). Simply ignoring that the distribution is not perfectly binomial seems to work fine. Erik _______________________________________________ Computer-go mailing list [email protected] http://dvandva.org/cgi-bin/mailman/listinfo/computer-go
