Re: computer-go: Perfect-play score on small boards

[Erik van der Werf <E.vanderWerf@xxxxxxxxxxxxx>]

Francois Grieu wrote:
> 
> Erik van der Werf wrote:
> 
> >  5x5 is solved by computer.
> 
> Any reference? 

I wrote a paper on it which is accepted for publication in the ICGA
journal. (June issue). If you want I can send you the latest draft.

> Under which rule? 

Mostly Chinese rules, however I also experimented a bit with SSK.

> What is the score for the best line?

Opening move (3,3): B+25
Opening move (3,2): B+3
Opening move (2,2): W+1
Opening move on edge: W+25


> 
> So far the best I got is the paper by Shinichi Sei and Toshiaki Kawashima:
> "A Solution of Go on 4x4 Board by Game Tree Search Program"
> <http://homepage1.nifty.com/Ike/katsunari/publications_e.html>
> which concludes the game is a draw on 4x4, under a rule set
> which makes the game a draw when "the arrangement of stones, player
> to move, position of Ko, and whether the move immediately before is
> pass or not" repeats.
> This rule set IMHO strongly bias the game towards a draw, when more
> usual rule set (i.e prohibiting repetition with some proper definition
> thereof) could give a different outcome.

4x4 + Japanese scoring is a draw because no points are counted in seki.

4x4 + Chinese scoring gives B+2

Erik