It occurred to me yesterday, though I haven’t coded it, that you can work 
through twice, first for the minimum suitable spanning box, and second for that 
box extended one row up and down and one column left and right. The inner 
regions are those whose size doesn’t change. Probably pretty inefficient, but 
easier to understand?

Mike

Please reply to [email protected].      
Sent from my iPad

> On 12 Dec 2018, at 13:05, Brian Schott <[email protected]> wrote:
> 
> Jimmy,
> 
> Your latest method works great, and so does Raul's.
> I'm a little surprised that both methods avoid the problems I was having
> using the convex hull approach, but I will study them more closely.
> 
> Thanks, to all of you,
> ----------------------------------------------------------------------
> For information about J forums see http://www.jsoftware.com/forums.htm

----------------------------------------------------------------------
For information about J forums see http://www.jsoftware.com/forums.htm

Reply via email to