CGAL users discussion list

["Sebastien Loriot (GeometryFactory)" <>]

As Mariette indicated, what is observed in your case is due to the fact
that your inputs are on a regular grid (thus a degenerate input).

Consider the case of a triangulation in 2D of four points on a circle. 
The circumcenter of the two triangles is the same.

Sebastien.


randooom wrote:
> Following a minimal(-ish) example, which mostly resembles my actual code.
>
> I hope it is clear as is, but if not please feel free to ask for
> clarification.
>
> Compiled with gcc version 4.6.0. with following flags:
>
> g++ -lCGAL -lCGAL_Core -lmpfr -lgmp  -frounding-math -W -Wall -Wextra
> -Wshadow -Wfloat-equal -Werror -g -O0 Test.cpp -o Test
>
> Now the code:
>
> ///////////////////////////////////////////////////////////////////Test.cpp////////////////
>
> #include <iostream>
> #include <vector>
>
> #include
> &lt;CGAL/Exact_predicates_exact_constructions_kernel_with_sqrt.h&gt;
> #include &lt;CGAL/Regular_triangulation_3.h&gt;
> #include &lt;CGAL/Regular_triangulation_euclidean_traits_3.h&gt;
>
> int main()
> {
>     typedef CGAL::Exact_predicates_exact_constructions_kernel_with_sqrt
> Kernel;
>     typedef CGAL::Regular_triangulation_euclidean_traits_3<Kernel>
> RegularTriangulationTraits;
>     typedef CGAL::Regular_triangulation_3<RegularTriangulationTraits>
> Regular_Triangulation;
>
>     typedef Regular_Triangulation::Finite_facets_iterator
> Finite_facets_iterator;
>     typedef RegularTriangulationTraits::Weighted_point_3 Weighted_point;
>     typedef RegularTriangulationTraits::Bare_point Point;
>     typedef Kernel::FT FT;
>
>     typedef std::vector<Weighted_point> WeightedPointContainer;
>
>
>     WeightedPointContainer weightedPoints;
>
>     /////////////////////
>     // This part basically takes one point and continues it in 26 space
> directions by offset.
>     // So each point input generates 27 points in space.
>     FT x(0.5);
>     FT y(0.5);
>     FT z(0.5);
>     FT weight(1);
>     FT offset(1);
>     
>     const int a[3] = {-1 , 0, 1};
>     
>     for(int iX = 0; iX < 3; ++iX)
>     {
>         for(int iY = 0; iY < 3; ++iY)
>         {
>             for(int iZ = 0; iZ < 3; ++iZ)
>             {
>                 weightedPoints.push_back(Weighted_point(Point(x + a[iX] *
> offset,
>                                                               y + a[iY] *
> offset,
>                                                               z + a[iZ] *
> offset),
>                                                         weight));
>             }
>         }
>     }
>     /////////////////////
>
>     /////////////////////
>     // Following the triangulation of the afore generated points and
>     // checking the dual edges of each finite_facet.
>     Regular_Triangulation triangulation(weightedPoints.begin(),
> weightedPoints.end());
>
>     assert(triangulation.is_valid());
>
>     for(Finite_facets_iterator itFinFacet =
> triangulation.finite_facets_begin();
>         itFinFacet != triangulation.finite_facets_end(); ++ itFinFacet)
>     {
>         typedef Kernel::Segment_3 Segment;
>         CGAL::Object o = triangulation.dual(*itFinFacet);
>         const Segment * pSegment = CGAL::object_cast<Segment>(&o);
>         
>         if(!pSegment)
>         {
>             // check only the Segments.
>             continue;
>         }
>         
>         assert(pSegment->source() != pSegment->target());
>     }
>     /////////////////////
>     
>     return 0;
> }
> ///////////////////////////////////////////////////////////////////Test.cpp
>
> Code aborts with "[...]Assertion `pSegment->source() != pSegment->target()'
> failed."
>
> --
> View this message in context: 
> http://cgal-discuss.949826.n4.nabble.com/Why-does-Regular-triangulation-3-dual-Facet-f-sometimes-return-a-edge-with-identical-source-and-targ-tp3625888p3627638.html
> Sent from the cgal-discuss mailing list archive at Nabble.com.