CGAL users discussion list

[Peter Hachenberger <>]

Hi Amir,

the extended kernels are nice conceptually, but they are indeed not very fast. I focused

my time on improving the rest of the package as very few people really need the

extended kernels. Let's find out whether you really need them, or whether it suffices

to work with the standard kernels.

The standard kernels do not support halfspaces per se. That is, without the extended kernel

we cannot represent polyhedra bounded by an infinitely large facet. On the other hand, we

support intersections with halfspaces. So, if you want to construct a polyhedron by intersecting

halfspaces and the result has no infinite facets, you probably can work without the extended kernels.

You just have to find out, how you can start. For instance you might want to start with a huge cube

and then intersect it with all your halfspaces. Does that help you?

Peter

-----"Amir Vaxman" <> wrote: -----

To: <>
From: "Amir Vaxman" <>
Date: 03/09/2009 11:30AM
Subject: RE:  [cgal-discuss] An Example of Nef Polyhedra doesn't work correctly

This workaround exists in the documentation of Nef Polyhedra in the CGAL manual. Still, I get very slow result when computing the arrangement of simple full planes (a minute or two for 20 planes). I aggregate every plane individually, but I can do a divide-and-conquer. Is there a possibility to insert all planes at once? The aggregation is a waste of time for this case..

 

Thanks,

Amir.

 

From: Peter Hachenberger [mailto:]
Sent: Wednesday, March 04, 2009 10:13 PM
To:
Subject: Re: [cgal-discuss] An Example of Nef Polyhedra doesn't work correctly

 

Hi Amir,

the workaround does not work because there is some code in Nef_polyhedron_3
that is specialized for Extended_homogeneous and Extended_cartesian. I guess
that the specialized code is not executed when the workaround is used. Try whether
the workaround works as follows:

#include <CGAL/Is_extended_kernel.h>
struct Kernel : public CGAL::Extended_homogeneous<NT> {};
struct Is_extended_kernel<Kernel > {
     typedef Tag_true value_type;
};

Where did you find the workaround? I should comment on the workaround in
the Nef documentation.

Peter

 

----- wrote: -----

To:
From:
Date: 02/28/2009 04:35PM
Subject: [cgal-discuss] An Example of Nef Polyhedra doesn't work correctly

Hello,

I tried to operate the example "point_set_operations" of Nef3 polyhedra. I am
using a VC8(2005) with CGAL 3.4. When I try to run the example verbatim, with
the kernel definition as:

"typedef CGAL::extended_homogeneous<CGAL::gmpz> Kernel", in debug mode, it runs
VERY slowly, up to the point where I couldn't keep on and stopped the debugging
manually. If I use the workaround offered in the documentation:

"struct Kernel : public CGAL::Extended_homogeneous<NT> {};"

the constructors of N1-6 assert that there is no proper construction during
runtime.

I add the aforementioned example file.

Thanks,
Amir.
--
You are currently subscribed to cgal-discuss.
To unsubscribe or access the archives, go to
https://lists-sop.inria.fr/wws/info/cgal-discuss

[attachment "point_set_operations.cpp" removed by Peter Hachenberger]
=