CGAL users discussion list

[Kuan <>]

Hi all,

Thank for all the help. It's solved!

best regards

Kuan

 

2007/8/15, Andreas Fabri <>:

Kuan,

I am not sure you understood Matthijs mail correctly.

When you write coordinates to std::cout or std::ofstream
you must increase the precision as otherwise the floating point
numbers get truncated, that is 1.2345567789 becomes 1.2345.

His next point was that when you approximate a cylinder by
a polyhedron, you might miss points which are in the difference
in case the polyhedron is inscribed the cylinder.

In 2D (say the top of the cylinder)


                          *
                      *        *
                   *--------------*
                    |             |
                  * |             | *
                    |             |
                   *--------------*
                      *    q   *
                           *

q is outside the polyhedron (the rectangle), but inside the
cyliner (the circle)

andreas


Kuan wrote:
> I write it not so precise. Just some points, and build a
> delaunay_triangulation_3, in some algorithm i have to find out whether a
> point is in convex hull (crust algorithm).
>
> 2007/8/15, Matthijs Sypkens Smit <
> <mailto:>>:
>
>     On Wednesday 15 August 2007 02:13, Kuan wrote:
>
>      > Hi all,
>      > Now i have a precision problem.
>      > I am using delaunay_triangulation_3 to find out whether a point is in
>      > convex hull.
>      > But because the problem of precision, i usually get a wrong answer.
>      > For example i generate some points on a cylinder by code, and
>     write it
>      > into a file and read it out.
>
>     Did you make sure to write the points out with maximal precision?
>
>      > And then many of the points is defined as not in convex
>     hull.(because
>      > the number is not so precision, you can image that?).
>
>     The points are not in the convex hull of what? A cyclinder-surface
>     represented by triangles?
>
>
>     --
>     Matthijs Sypkens Smit
>     --
>     You are currently subscribed to cgal-discuss.
>     To unsubscribe or access the archives, go to
>     https://lists-sop.inria.fr/wws/info/cgal-discuss
>
>
--
You are currently subscribed to cgal-discuss.
To unsubscribe or access the archives, go to
https://lists-sop.inria.fr/wws/info/cgal-discuss