CGAL users discussion list

[Monique Teillaud <>]

Le 25/11/11 13:06, Joao Dinis a écrit :
> Hi there,
>
> I've been comparing the running times of computing the Delaunay
> triangulation of points on the surface of a unit radius sphere.

> Further (extensive) tests led me to find out that it is possible to
> greatly reduce run-times by adding, not one, but a lot of points near
> the origin.

> Is this behavior to be expected?

Hi,

I did not look at your timings carefully, but still, yes, this is to be 
expected: you remove most degeneracies by doing this. Hence you remove 
most arithmetic filter failures, hence most calls to exact arithmetic.

Best,

--
Monique Teillaud
INRIA Sophia Antipolis - Méditerranée
http://www.inria.fr/sophia/members/Monique.Teillaud/