RE: [computer-go] Chains and liberties

["Mark Boon" <tesuji@xxxxxxxxx>]

> -----Original Message-----
> From: computer-go-bounces@xxxxxxxxxxxxxxxxx
> [mailto:computer-go-bounces@xxxxxxxxxxxxxxxxx]On Behalf Of Yonik Seeley
> Sent: Saturday, November 13, 2004 16:46
> To: computer-go
> Subject: Re: [computer-go] Chains and liberties
>
>
> Sounds interesting Mark!
>
> Don't focus too much on the O(1) vs O(n) thing though, since we aren't
> after major scalability... we know the upper bound on everything, and
> it's not that big (19x19).  For example, an O(n) solution may be faster
> than the O(1) solution for the common case.

I know that theoretically O(1) is in the same class as O(361), but it can
make a big difference in performance. We have to be pragmatic and just see
which solution gives the best performance.

> I think it might be best to try different solutions and see which
> is fastest
> for the avegage game (either random games, or a collection of real
> games).

A full game or set of games is not necessarily a good test-set. A tactical
reader encounters different circumstances on average than what you see in a
full game. Some thought may have to be given what *would* be considered a
good test-set. Probably a mixture of things.

> Maybe an interface should be standardized, and then we can see
> who can come up with the fastest implementation!
>

I'll think about that, but I know I don't want to implement more than one.

_______________________________________________
computer-go mailing list
computer-go@xxxxxxxxxxxxxxxxx
http://www.computer-go.org/mailman/listinfo/computer-go/