Re: [computer-go] End Game

[Nick Wedd <nick@xxxxxxxxxxxxx>]

In message <4801.67.38.162.176.1076540882.squirrel@xxxxxxxxxxxxxxxxx>, 
Robin Kramer <robin@xxxxxxxxxxxxxxxxx> writes
> Hello,
>
> Just contemplating the end game a little.  Seems like their are several
> open problems.
>
> 1)Given a finished game determine the score,
> a)finding the outlines
> b)determining the dead stones.
> c)counting the territory
>
> 2)Given a game is it finished,
> a) are the dead stones dead and the living stones alive,
> b) is there any more territory to secure.
>
> Are these seperate problems, or the same problem?
>
> If the first problem was solved, could a machine learning algorithm such
> as explanation based learning(EBL) be used to solve the second one?
What do you mean by a "finished game"?  You might mean
  one which the rules say is finished because both players have passed
  one in which there is no move worth making, so a skilful player would 
pass

Determining which stones are dead, for either definition of "finished 
game", can be very difficult.

Given my second definition of "finished game", then there can't be any 
more territory to secure.  Calculating how much secured territory you 
have then got (given the status of all groups) looks possible, but I am 
not aware of any software which does it perfectly.  The closest I have 
seen is Yoshikawa's ScoreEst function, implemented on KGS.

Nick
--
Nick Wedd    nick@xxxxxxxxxxxxxxxxx
_______________________________________________
computer-go mailing list
computer-go@xxxxxxxxxxxxxxxxx
http://computer-go.org/mailman/listinfo/computer-go