Skip to Content.
Sympa Menu

cgal-discuss - Re: [cgal-discuss] efficiency of a O(n^2) algorithm

Subject: CGAL users discussion list

List archive

Re: [cgal-discuss] efficiency of a O(n^2) algorithm


Chronological Thread 
  • From: Ben Supnik <>
  • To:
  • Subject: Re: [cgal-discuss] efficiency of a O(n^2) algorithm
  • Date: Thu, 05 Nov 2009 13:09:49 -0500

Hi,

Marco Aurelio Sterpa wrote:
I've used this definition for numeric type, can the running time depends by this?

typedef CGAL::Exact_predicates_exact_constructions_kernel_with_sqrt Kernel_2;
typedef Kernel_2::FT NT;

Yes and no. That kernel may be unnecessarily slow...you might get more speed with a faster one, but the fundamental O(N^2) of the algorithm isn't going to go away even if you use a faster kernel. I don't think any of the kernels have faster than constant-time operations but I am not sure about this. You may want to try a filtered kernel or NT or something...for my project I simply tried some common ops with a whole range of kernels and picked the fastest one that met my numeric needs (exact ctors) as measured on real data. :)



Archive powered by MHonArc 2.6.16.

Top of Page