CGAL users discussion list

[Daniel Duque <>]

Hello,

This is actually a crucial question for people who use evolving 
triangulations. They may evolve following some time evolution, or in some 
geometrical procedure (e.g. Lloyd's algorithm for centroidal Voronoi 
tessellations). I am currently using both!

On Friday 11 December 2009 13:31:51 Pedro Machado Manhães de Castro wrote:
> As far as I understood, you are within a static setting since you know the
> next position of all the points, and all your points move.

This is often the case: the new positions are known for all points. Sometimes 
you would move them one at a time (as is Monte Carlo dynamics, which I am not 
currently doing).

> Then, in general, your approach (2) is better (rebuilding).

But, sometimes the new points will be fairly close to the old ones. AFAIK, 
this would mean that perhaps moving each of the points to its new (close-by) 
location could be more efficient. There are two ways to do this: erase the 
old 
point, insert the new one; or, insert then erase. I am currently using the 
(undocumented?) T.move( ) function (in Delaunay 2D triangulations), which 
does 
one of the two (I think the later, but I may be wrong).

Daniel




-- 
http://debin.etsin.upm.es/~daniel