Am 17.06.2008, 06:54 Uhr, schrieb Russ Williams <[EMAIL PROTECTED]>:
On Mon, Jun 16, 2008 at 11:42 PM, WSK <[EMAIL PROTECTED]>
wrote:
Am 16.06.2008, 22:20 Uhr, schrieb Peter Drake <[EMAIL PROTECTED]>:
UCT
The idea of MonteCarlo and UCT contain the elemination of special game
rules
( include experienced moves like ladders ) and used the real ability of
computers: searching and sorting. I am right ?
So, the question has the wrong perspectiv. The clou and baseidea of
MC/UCT
is, that there is no need to implement special rules like ladders or
things
like mojo, sente etc.
MC evaluate only the hole game, when looking who will win, if a special
move
is done.
But the idea of writing a strong game player is that it play the game
well, not that it strictly follow some specific algorithm or
methodology or ideology...
Yes I am on your side. I only want to focus to the consequences. "Old"
Go Engines works in this way (they implement -in short- a long list of
if-then
for every know situation on board).
No pratice without theory.
If someone finds that adding some knowledge of reading ladders, or
joseki, or whatever, makes their program stronger, it would be rather
silly to refuse to add that to their program just because it's not
pure MC/UCT.
Unless their goal is researching MC/UCT instead of making a strong
game player, of course!
You are a little agressive, why? I just read my thoughts on the list. If
you want
a Go engine that plays as strong as modern chess or backgammon computers
it is a loooooong
way...
Peace
WSK
PS
@all
The pseudocode I written right?
_______________________________________________
computer-go mailing list
computer-go@computer-go.org
http://www.computer-go.org/mailman/listinfo/computer-go/
