[Computer-go] Monte-Carlo Tree Search in other games

Петр Смолов petr.smolov83 at mail.ru
Thu Oct 28 00:56:27 PDT 2010


Excuse me if I'm posting the question in this forum, but it seems to me that only go programmers know Monte-Carlo well.

How does UCT work in the games where players can make a draw?

After the program plays all simulations, which move should it choose?

(Wins/Visits) + SQRT(ln(...))

or

(Wins+Draw/2)/Visits + SQRT(ln(...))


How do you think which exploration coefficient C is good for such games?



More information about the Computer-go mailing list