Re: [computer-go] False Eyes

[Mark Boon <tesujisoftware@xxxxxxxxx>]

On 7/22/05, Christoph Vielhauer <c_vielhauer@xxxxxxxxxxxxxxxxx> wrote:
> Hi Mark, 
>  
> actually I was trying to find a rule, which does not refer to the control of
> the "diagonals" on the board. Instead, I intended to use the relation of
> clusters to regions only (without further reference to the position on the
> board).
>  
> If I understand the "four headed dragon" correctly, there are 4 clusters
> (say 1 -4, clockwise starting at the top) and 4 regions (say a-d, clockwise
> starting at the top-left). 
>  
> The region a (for example) has two clusters (1 and 4) and therefore might be
> a false eye. But 1 and 4 are also linked  via 1 - b - 2 - c - 3 - d - 4 and
> this link does not use region a. Thus, a is a real eye and so are the other
> regions. 
>  
> So I'd say, that my definiton does work in the case of the four-headed
> dragon (but  indirect (recursive) links between clusters are necessary). Am
> I missing something?
>  

Well, you were not clear about what you mean by 'linked' then in your
original post. I suppose you could make some recursive definition that
would work, but it's not at all straightforward. And it would still
only work under perfect conditions. Slightly imperfect conditions
break your definition and will make it impossible to say for sure if
something has eyes or not. Think about two separate eyes connected in
an imperfect way, like a false eye or a bamboo-joint. Not even
mentioning the possibility of ko.

You may not like using the diagonals for some pure abstract reason,
but I can assure you it gives a more accurate result in anything
you'll be able to implement within a reasonable time.
_______________________________________________
computer-go mailing list
computer-go@xxxxxxxxxxxxxxxxx
http://www.computer-go.org/mailman/listinfo/computer-go/