Re: computer-go: Go4++: Passing in tsume searches.

[David Fotland <fotland@xxxxxxxxxxxxxxx>]

My tsume-go searches are very small, typically only about
20 to 50 nodes, and rarely more than 10 ply deep.  I don't
generate passes, since the life/death evaluator recognizes seki
and similar positions without playing out the sequences that
require the pass moves.  I haven't tried any of these sample
positions, but they all look to me like Many Faces would
recognize a leaf node before the passes are required.

The position below is recognized as a leaf, being a seki,
so there is no further reading, and no passes required.
It's not very hard to statically recognize this kind of seki.

David Fotland

At 10:46 PM 1/24/2002 +0000, you wrote:
> > It works if you change the shape a bit, e.g.
> >
> >  9 . . * * * * * . .
> >  8 . * * 0 0 0 * * .
> >  7 . * 0 0 . 0 0 * .
> >  6 . * 0 0 * * 0 * .
> >  5 . * 0 . * * 0 * .
> >  4 . * 0 0 0 . 0 * .
> >  3 . * 0 0 0 0 0 * .
> >  2 . * * * * * * * .
> >  1 . . . . . . . . .
> >    A B C D E F G H
> >
> > Then * cannot almost-fill the area with a nakade shape.
> >
> > Martin
> >
>
> So that's two examples found for having to pass twice...
> anyone construct a 3-pass situation?
>
> I'm praying that I can have some sort of limit on the
> number of passes allowed because some of my tsumego
> searches are getting 80 ply deep and half the defensive
> moves are passes!! Passes are very expensive in tsumego
> searches firstly because they add an extra move to
> look at within any given ply and they also fail to fill
> up space in the search region and so leave more choices
> in the next ply.
>
> Am I the only one who's searches are clogged up with
> passes? How have other people dealt with this?
>
> Mick.
David Fotland