Skip to Content.
Sympa Menu

cgal-discuss - [cgal-discuss] Inexact kernels for Nef_polyhedron

Subject: CGAL users discussion list

List archive

[cgal-discuss] Inexact kernels for Nef_polyhedron


Chronological Thread 
  • From: Andre Schmeisser <>
  • To:
  • Subject: [cgal-discuss] Inexact kernels for Nef_polyhedron
  • Date: Wed, 09 Oct 2013 17:17:04 +0200
  • Organization: Fraunhofer ITWM

Hello all,

is it possible to use Nef polyhedra with inexact kernels?
After some feedback from here on constructing the polyhedra, I'm
currently using the Exact_predicates_exact_constructions Kernel.

The documentation says in 29.5
"We use the CGAL::Cartesian kernel.
All Cartesian and homogeneous kernels of Cgal are suitable if the number
type parameter follows the usual requirements of being a model of the
CGAL::FieldNumberType concept for the Cartesian kernels [...]"

The example following this, however, doesn't use the stated Cartesian
kernel.
Also, if I'm reading the documentation correctly, double fulfills the
requirement of being a model of FieldNumberType.

I've tried compiling with either Simple_cartesian<double> and
Exact_predicates_inexact_constructions_kernel. Both give different
compiler errors. Is there a way to make it compile, or a different
kernel based on doubles that works?


A little background information:
I'm currently trying to perform boolean operations on a list of
relatively simple convex polyhedra. Most are as simple as
triangle-prisms (pentahedra). After several "difference" operations, the
original polyhedron because non-convex and also slightly more complex,
but not too much. Still, the operation sometimes gets stuck unreasonably
long in the computation.
An operation of "A -= B" took about 1 hour, where B is one of the
mentioned convex polyhedra and A is no more complex than this:
> vertices: 25, edges: 38, facets: 15, sfaces 50, volumes: 2

Another computation, with a complexity of A as
> vertices: 112, edges: 169, facets: 59, sfaces 224, volumes: 2
I had to interrupt after being stuck in the innards of the GMP library
for 10 hours.

Getting a result quickly but with maybe some numerical errors would be
preferable to this.



Best regards,
Andre


--
Dipl.-Inf. Andre Schmeißer
Department of Transport Processes
Fraunhofer Institute for Industrial Mathematics ITWM
Fraunhofer-Platz 1, D-67663 Kaiserslautern
Internet: www.itwm.fraunhofer.de


  • [cgal-discuss] Inexact kernels for Nef_polyhedron, Andre Schmeisser, 10/09/2013

Archive powered by MHonArc 2.6.18.

Top of Page