CGAL users discussion list

[Marc Glisse <>]

On Tue, 3 Nov 2020, Renaud GILLON wrote:

> Is there a technical reason that the construction of barycentric
> coordinates for the Epick kernel is not specified or implemented? There is
> only a dummy placeholder in CGAL 5.1.

Are you talking about Epick_d? When I wrote that kernel, I wrote what I 
needed, plus the minimum to satisfy the Kernel_d concept, and it happens 
that the people who wrote that concept a long, long time ago forgot to 
document a few things, including Barycentric_coordinates_d. So all I 
technically needed was a dummy typedef...

I think it would be good to implement some of those missing pieces for 
real. However I don't think I am likely to find time to do that, unless I 
need to use it myself.

> Is there a problem to implement the transform() method for Epick::Point
> objects ?
>
> The mapping of cartesian to barycentric coordinates could be implemented as
> an affine transform.

You may notice that affine transformations are not implemented either in 
Epick_d.

If I didn't have to follow the existing Kernel_d concept, we could have 
Eigen::VectorXd as point type, and then it would be easy to use Eigen 
directly yourself for such missing functionality. We almost have that 
internally with undocumented interfaces...

> The transform could be stored as Data in the full_cells of a Triangulation.

--
Marc Glisse