[Computer-go] Tsumego Solvers

Detlef Schmicker ds2 at physik.de
Sat Jun 22 22:47:02 PDT 2013


Hi,

oakfoam does not have such a feature yet. We have simelar features for
debugging, e.g. we could ask with which probability does a playout make
a point xy be owned black e.g. (and give an example playout)

But a possible approach would be straight forward I think. Change the
definition of winning to winning at point xy. If you want to take the
moyo effect into account possibly +(usual winning)/100 (100 is only an
example for weighting of the usual win). A problem will be the
definition of the last move made. I think all current strong mc programs
relay to some extend on distance from the last move to narrow the
search...

As I am currently collecting ideas for regression tests, there is a good
chance, that oakfoam will incorperate this kind of solver. At the moment
I was planing only to implement: check probability of owning xy with a
usual move, but the solver approach might give better and more
reproducible results. I probably will give it a try:) Thanks!

Detlef
P.S.: In case you are a C++ developer you might consider joining us:)



Am Samstag, den 22.06.2013, 21:11 -0500 schrieb Mark Goldfain:
> Does anyone have a good tsumego-solving program?  It seems a good, 
> separate question, not necessarily part of a full game-playing program.  
> I'd like to see if I could modify one to seek specific goals.  For 
> example, given a portion of a board, is there a move sequence that 
> successfully invades and lives?  Then, given that one can invade and 
> live, with some initial move giving probability of success alpha, then 
> lowering the threshold slightly to beta, which move gives the 
> largest-point outcome with probability > beta ?
> 
> Example:
> 
>       ++ at ++@@@++          black = @
>       ++ at +++++++          white = O
>       +++ at ++++++
>       ++ at +++++++          white to move
>       ++ at +++O+++
>       +++@@++O++
>       ++++++@@O+
>       ++++++++++
> 
> If this diagram renders well, then the first question is what move for 
> white has the best probability of living?  But the second question, is 
> whether one sequence of moves leads to more damage to black's moyo than 
> other sequences.
> 
> I'd be interested to hear from someone who has either worked on such 
> issues, or has a tool that they feel is helpful for such questions.  I'd 
> also like to hear which programs include this kind of reasoning in their 
> repertoire.
> 
>                           Many thanks,
>                           -- Mark
> 
> _______________________________________________
> Computer-go mailing list
> Computer-go at dvandva.org
> http://dvandva.org/cgi-bin/mailman/listinfo/computer-go
> 





More information about the Computer-go mailing list