[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]
RE: computer-go: Perl Module for next move.
Actually, I suppose you can have a two point seki as well. Like this, on a 7x7
board:
. O . O X . X
. O . O X X .
O O O O X X X
X X X X O O O
X X X . O O O
X X X . O O O
X X X X X X X
Probably best to scrap the automatic-loss rule and turn in into a
pass-not-allowed rule.
Pass not allowed when:
Any owned region is bigger than 1 space
Any unowned space is bigger than 2 spaces
In the above game, white would have to add two more stones before being allowed
to pass.
-----Original Message-----
From: Fant, Chris
Sent: Wednesday, June 06, 2001 2:53 PM
To: 'computer-go@xxxxxxxxxxxxxxxxx'
Subject: RE: computer-go: Perl Module for next move.
That's why I said "larger than one point." For seki. Both strings will be
considered alive, simply because they are still on the board, but the NN should
not attempt to remove the opposing group b/c it would be killed if it tried. So
both sides are satisfied to concede that the opposing group is alive. You are
trying to make the NN guess the right move, right? I don't think it will work,
but it won't be b/c of the scoring.
-----Original Message-----
From: Grajdeanu, Adrian [mailto:adrian.grajdeanu@xxxxxxxxxxxxxxxxx]
Sent: Wednesday, June 06, 2001 12:09 PM
To: 'computer-go@xxxxxxxxxxxxxxxxx'
Subject: RE: computer-go: Perl Module for next move.
OOps, here is a problem!
I understand that unsettled territory is whatever is not completely
surrounded by either white or black. In other words, if you can reach it
starting from a white stone and from a black stone, following vertical or
horizontal segments and only stepping in empty slots. Well, this seems to
include neutral territory. I have no problem with that. Yet, what is to be
done in case of seki? Penalizing for not covering neutral territory would be
a mistake. Covering it up would be an even bigger mistake.
Is there a straightforward algorithm to determine seki?
Thanx.
> -----Original Message-----
> From: Fant, Chris [mailto:chris.fant@xxxxxxxxxxxxxxxxx]
> Sent: Tuesday, 05 June, 2001 15:49
> To: computer-go@xxxxxxxxxxxxxxxxx
> Subject: RE: computer-go: Perl Module for next move.
>
>
> Once again, I strongly agree with Heikki on the whole Chinese
> scoring thing.
> The only reason I can see to play Japanese scoring is if you
> are playing against
> an opponent who demands it. But you aren't. Here is a cool
> idea I just thought
> of to avoid the game-is-over-after-10-moves situation: For
> any game where there
> is a region of unsettled territory larger than one point,
> record the game as a
> loss for both sides. I think the NN would quickly learn to
> avoid that type of
> ending.