[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
Re: computer-go: [Question] Alpha-Beta Pruning & Null Move
Tristan,
Your description is a good description of what null move pruning is
all about. Slightly more precise is to say, "if I am not happy
playing 2 moves in a row, I'm not going to look at anything at all
from this position."
The tricky part is defining "happy." Fortunately, "happy" has a very
precise definition. It simply means that given 2 moves in a row, can
I achieve a score good enough to achieve at least my expectation? If
you can, you are "happy." In a losing position, you can still be
"happy" given this definition because you expectations are
appropriately lowered! Your "expectation" is the "alpha" part of
"alpha/beta pruning" or put another way your proven lower bound.
Don
From: "Tristan Cazenave" <cazenave@xxxxxxxxxxxxxxxxx>
References: <200104211200.OAA31828@xxxxxxxxxxxxxxxxx>
Date: Sat, 21 Apr 2001 23:30:34 +0200
MIME-Version: 1.0
Content-Transfer-Encoding: 7bit
X-Priority: 3
X-MSMail-Priority: Normal
X-Mailer: Microsoft Outlook Express 5.00.2919.6700
X-MimeOLE: Produced By Microsoft MimeOLE V5.00.2919.6700
Sender: owner-computer-go@xxxxxxxxxxxxxxxxx
Precedence: bulk
Reply-To: computer-go@xxxxxxxxxxxxxxxxx
Content-Type: text/plain;
charset="iso-8859-1"
Content-Length: 641
> If I've understood correctly null-move will be useful mainly to prune the moves that are worst
than pass.
> This really looks interesting in chess, but in go it's hard to have this kind of moves.
> And being usually of the type
> "don't put your-self in atari"
> are easily avoided by more heuristical means.
In my opinion, Null move is rather: it is useless to try to live with one move,
if I cannot live by playing two moves of the same color in a row.
In my experiments, Null move enables to speed up by a factor of two
capturing problems search.
Any other one has measured the usefulness of Null Move
in their programs ?
Tristan.