[computer-go] Incrementally recognizing regions of emptyspace(Murpheys law of floodfill)

[Mark Boon]

On 28-sep-06, at 04:49, Chrilly wrote:

> The side which is in the global game to move is also in the local  
> tactics to move. Which has the problem, that in fact one can move  
> only in one local fight first. This leads to an overestimation of  
> the right to move. Which is partly intended for selective pruning  
> in the search tree, but which leads also to very nasty horizon  
> effects.
> It seems to be that there is no real correct way.
>

You should look at both, one where the hunter moves first and one  
where the prey moves first and use the average. If there's one eye  
where the hunter moves first but no eye where the prey moves first  
you have half an eye. This way you are only overoptmistic or  
pessimistic in cases where there's overlap, but you're not nearly as  
overoptimistic as when you always assume the same side to move first.

Mark

-------------- next part --------------
An HTML attachment was scrubbed...
URL: http://computer-go.org/pipermail/computer-go/attachments/20060928/02995bd6/attachment.htm