CGAL users discussion list

[Bernd Gärtner <>]

Dear Jérémie,

thanks for reporting this. I think I know where the problem is (the 
underlying quadratic programming solver cycles). For now, I cannot offer 
any fix, except the one you suggest: since cycling is a very instable 
phenomenon, it usually goes away when you change the input order. Of 
course, we will look into a general fix.

All the best,
Bernd.

On 5/9/13 12:07 AM, jdumas wrote:
> Hi.
>
> I think I've put a finger on an unexpected behavior in the package
> Polytope_distance_d. It happens when the second polygon is reduced to a
> single point, and only with certain coordinates. The issue is that the
> solver simply hangs indefinitely instead of computing the right answer. The
> interesting thing is that if I switch P and Q then the solver find the
> optimal solution immediatly.
>
> Here is a link to the minimal example code (adapted from the examples
> shipped with CGAL):
> http://pastebin.com/viCkT8Xw
>
>
> My version of CGAL is 4.2, and gcc is 4.7.2.
>
> Best,
> Jérémie.
>
>
>
> --
> View this message in context: 
> http://cgal-discuss.949826.n4.nabble.com/Bug-report-Polytope-distance-d-package-tp4657397.html
> Sent from the cgal-discuss mailing list archive at Nabble.com.