Re: Disjoint circle merge NP complete for L^n error?

Thanks! Great idea. I think that works.

Indeed in real problems, the goodness of a result around major cities
is heavily related to the order of iterative merges. We had just
started looking at SA because greediness isn't good enough.

Amazing that I was just talking to a friend about the relationship with
halftoning. For a completely different project, we have been working
on a variation of Hopfield nets (Error-Diffusion Nets) that find local
minima of frequency-weighted L^2 error. Mapping the circle merging
problem into halftoning is something we're looking at.

Best regards,
Gene

.