[Swarm-Modelling] Re: contour graph

[Marcus G. Daniels]

Hi Sam,

> i need a continuous contour sort of graph of densities in each cell 
of the grid...
> a line graph connecting cels of equal densities...

What I'd suggest is to rescale and truncate the densities to integer 
group levels and then subset the points per those integer levels   For 
example, if the densities were normalized between 0.0 and 1.0, then 0 < 
density <= 0.25 could be group 1, 0.25 < density <= 0.5 could be group 
2, 0.5 < density <= 0.75 group 3, and 0.75 < density <= 1 group 4.    
Say the high density points were the ones of interest, then you'd only 
keep the x,y coordinates of those points in group 4.

Now using that list of coordinates of you can then compute a 2d convex 
hull / Delaunay triangulation to get the perimeter, and you can overlay 
contours at different densities by using different level sets.  Here's a 
web page that illustrates some alternatives:

http://www.cse.unsw.edu.au/~lambert/java/3d/hull.html

I see there is some Java source code for that demo available, but with 
some web searches I suspect you'll surely find other implementations.

Marcus