CGAL users discussion list

[Sartaj Hundal <>]

One of your envelopes may be empty.

On Tuesday, August 27, 2013, Raghu Kisore N <> wrote:
> Efi Fogel,
>
> What exactly are you trying to do?
>
> In my case I have set of 500 data points which have been broken into 2 groups of 250 points each. A voronoi diagram is created for the first 250 points and the same is done with the other 250 points. Now I am using the function merge_envelope to combine the two.
>
> ok, but first you need to compute the 2 envelopes of the 2 structures, respectively. How do you do that?
>
>   Using the logic provided in the example code “print_cropped_voronoi.cpp” I first obtained cropped version of the voronoi diagrams.
>
> By traversing over these diagrams I converted them into objects of type
>
> Typedef Arrangement_2<Env_plane_traits_3<Traits>, Envelope_pm_dcel<Traits, Traits::Xy_monotone_surface_3> > Arrag_2
>
> Traits in the above  are of type Env_plane_Traits_3<Cartesian<Gmpq>>
>
> Using the above an object of the class Envelope_divide_and_conquer is created
>
> typedef CGAL::Envelope_diagram_on_surface_2<Traits>::Arrangement min_diagram;
> typedef CGAL::Envelope_divide_and_conquer_3<Traits, min_diagram> DivConq;
>
>  
>
> The program crashes when I call DivConq.merge_envelopes();//the three required arguments are of type Arrag_2 defined above
>
>  
>
> After debugging I found the following cause for the crash.
>
> The function merge-envelopes makes use of objectàget_data() [this happens during the àresolve function call line 532 of the function merge_envelopes ] without ever calling set_data()[of course it is eventually called at the end of merge_envelopes when calling update_envelope_surfaces_decision()..line 674 ].
>
>  
>
> [Alternatively is there any other way of converting object of type VD into Arrangement_2?]
>
>  
>
> I meant what CGAL kernel are you using?
>
>  
>
> I am using CGAL-4.2.
>
>>>>>>>>You must use a kernel that supports exact constructions as well as >>>>>>exact predicates, e.g., Exact_predicates_exact_constructions_kernel
>
> The example “print_cropped_voronoi” uses Exact_predicates_inexact_constructions_kernel.
>
> As per your suggestion I used Exact_predicates_exact_constructions_kernel, but it still doesn’t work.
>
>   
>
> Several papers have been published claiming that the class can be used to merge two arrangement’s.
>
> Can you please point them to me.
>
> Ø  “Exact and Efficient Construction of General Voronoi Diagrams in the Plane via Divide and Conquer of Envelopes in Space” Workshop on Robust Shape Operations, INRIA Sophia-Antipolis, September 2007.  Dan Halperin Ophir Setter Micha Sharir.
>
> Ø  “Constructing Two-Dimensional Voronoi Diagrams via Divide-and-Conquer of Envelopes in Space” by Ophir Setter. Tel-Aviv University, May 2009
>
> Regards
>
> Raghu.
>
>  
>
>  
>
> From:  [mailto:] On Behalf Of Efi Fogel
> Sent: Monday, August 12, 2013 5:18 PM
> To: 
> Subject: Re: [cgal-discuss] question reg class Envelope_divide_and_conquer_3
>
>  
>
> On Mon, Aug 12, 2013 at 7:24 AM, Raghu Kisore N <> wrote:
>
>  
>
> What exactly are you trying to do?
>
> In my case I have set of 500 data points which have been broken into 2 groups of 250 points each. A voronoi diagram is created for the first 250 points and the same is done with the other 250 points. Now I am using the function merge_envelope to combine the two.
>
> ok, but first you need to compute the 2 envelopes of the 2 structures, respectively.
>
> How do you do that?
>
> What version are you using?
>
> CGAL-4.2
>
> What kernel are you using?
>
> I am using FC-15 and kernel version is 2.6.38
>
>  
>
> I meant what CGAL kernel are you using?
>
> You must use a kernel that supports exact constructions as well as exact predicates, e.g., Exact_predicates_exact_constructions_kernel
>
>  
>
> I am happy if you could post an example that makes use of this class.
>
>  
>
> I don't have one, and as I've said, it is internal and as such it is not documented.
>
> However, you are absolutely free to try and use it any way you want.
>
>  
>
> Several papers have been published claiming that the class can be used to merge two arrangement’s.
>
>  
>
> Can you please point them to me.
>
>  
>
> Regards
>
> Raghu.
>
>  
>
> From:  [mailto: