[computer-go] Solving Go

[Álvaro Begué]

On 11/7/07, Don Dailey <drdailey at cox.net> wrote:
>
> [...]
> To go beyond 5x5, say 7x7 would require an endgame table with 3**49
> entries or 239299329230617529590083 entries.   This can be reduced by
> about 8x if you remove symmetrically equivalent positions.    This is
> pretty intractable, won't happen any time soon.  Nor is it likely to be
> searched by brute force methods.


As I said before, the configuration on the board doesn't contain enough
information to represent the state of the game; you also need to consider
the prior history of configurations, to properly enforce super-ko. So this
method wouldn't work even if we could handle tables that are that large.

Try to write a program that solves 2x2 go using EGTBs and you'll see what I
am talking about.

Álvaro.
-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://computer-go.org/pipermail/computer-go/attachments/20071107/62adc78f/attachment.htm