Jacques Basaldúa: <[email protected]>:
>Hideki Kato wrote:
>
> > | . . . . . . .
> > | . # # . # . .
> > | . O # . # . .
> > | O O O . # # .
> > | # # O O O # .
> > | . # # . . . .
> >  ---------------
>
> > Here, the problem is "filling own eye" at 1-1 for B, right?
>
>Yes, that is the figure I meant, but it is not related with
>nakade, only with filling your own eye.

I don't understand.  Black can kill white by nakade with filling 
its eye first.  Why not related?

> > Self-atari (of a small group) is already allowed by almost all MCTS
> > programs, I believe.  I don't understand how this relates with seki.
>
>The size of the group can be one criterion, but there are others.
>The relation with seki is: if a group that lives in seki plays the
>suicidal self-atari move, it dies and the playouts don't get the 
>evaluation ok. No matter if you avoid the move in the tree or not,
>you must avoid it in the playouts or else the group dies. But that
>is tricky, as you cannot simply forbid self atari moves because
>some self-atari moves are tesuji. Therefore, there is "black magic"
>to decide when and when not to play self atari. The size of the
>group is important, but I guess strong programs must have found better
>ideas.

Agree.  AFAIR, Remi implemented seki in Crazy Stone and told me it's 
a big kludge or similar a few years ago.

Hideki
-- 
Hideki Kato <mailto:[email protected]>
_______________________________________________
Computer-go mailing list
[email protected]
http://dvandva.org/cgi-bin/mailman/listinfo/computer-go

Reply via email to