On Tue, 14 Sep 2010, Brian Sheppard wrote:
You're mixing the two pondering strategy. In this strategy, you always
have a match, so you really get 1 + 1/BF.
Ah, a common misunderstanding of alpha-beta.
Alpha-beta search only produces *moves* along the principal continuation.
In other variations you only get *upper bounds*.
So you have to match the opponent's move to get a predicted move.
I won't even discuss this assertion. But its immediate consequence would
be that the first strategy gives 1 + P[match]/BF, and the second 1 +
P[match].
If you need a match anyhow, then whatever the branching factor and match
probability, of course you take the strategy needing only the match.
It's then the algorithm that is responsible, and absolutely not the
specifics of a game.
Jonas
_______________________________________________
Computer-go mailing list
[email protected]
http://dvandva.org/cgi-bin/mailman/listinfo/computer-go
