CGAL users discussion list

[Matthieu Chavent <>]

I analyzed my set of points and I saw a problem: I had two points which 
were the same in "double" but when I checked if they were equal with 
CGAL (with == ) it  was not the case. In fact, at the beginning I 
calculated some circumcenters of  Delaunay cells and I have two Delaunay 
cells with the same circumcenter.. But CGAL considered the circumcenter 
as two points (points that cause problem for the convex_hull).. So, the 
problem is more a problem of interpretation of CGAL than a problem of 
convex_hull. I would like to keep the 
exact_constructions_inexact_predicates kernel (to have a faster program) 
and finally,I prefer to disable the post conditions..

Thanks everybody for the help!!

     Matthieu.



 a écrit :
> Le Jeu 12 avril 2007 13:50, Matthieu Chavent a écrit :
>   
> > I use the exact_predicates_inexacte_constructions kernel. I though that
> > this kernel was enough to perform what I want: a convex hull of few
> > points. I don't create any other points.. The very strange thing is when
> > I disable the post conditions, I obtain only triangles as facets of my
> > polyhedron so I don't understand why I can have the "strongly_convex"
> > error..
> >     
>
> In fact, the is_strongly_convex_3 post-condition uses constructions. As
> such, it is not robust with exact_predicates_inexact_constructions kernel.
>   

---------------------------

Matthieu Chavent
LORIA
PhD student

tel: ++33(0)3-54-95-86-05

---------------------------