Geometry Boost List: Re: [geometry] GSoc Project 2 Clarification

From: Vissarion Fisikopoulos (fisikop_at_[hidden])
Date: 2020-03-16 11:18:38

On Fri, 13 Mar 2020 at 08:19, Aditya Mohan via Geometry
<geometry_at_[hidden]> wrote:
>
> Hi,
>
> With my PRs being (almost) wrapped up, I'd like to ask some doubts about the other project (besides continuing to improve documentation, namely for the arithmetic operations) I'd like to work on as part of GSoc '20 - Implement support for non-cartesian geometries in convex_hull() either by adapting existing algorithm and strategies (preferable) or implementing it as different algorithm.
>
> Assuming a coordinate system - the Polar coordinate system, for this, can we not simply use the boost::geometry::transform function to convert cartesian to polar coordinates and vice versa? In this case we can call this function and implement the existing algorithm convex_hull suited for cartesian coordinates and convert it back to polar coordinates once the convex_hull is created in the cartesian coordinate system?
> I can maybe tweak the transform function to incorporate the conversion of ellipsoidal to cartesian system as well in order to extend its scope.

you can always solve the spherical and ellipsoidal problems by
projection to 2d cartesian and then transform back, but in
Boost.Geometry we have dedicated algorithms for spherical and
ellipsoidal that work directly in the CS space without transformations
and this is what we want from this project for convex hull and
centroid.

Best,
Vissarion

> Thank You
> _______________________________________________
> Geometry mailing list
> Geometry_at_[hidden]
> https://lists.boost.org/mailman/listinfo.cgi/geometry