levelset: interface intersection

[ksamouchos]

1. Introduce the pixel, triangle and polygon intersection with plane.
2. Allow the intersection of interfaces with smoothed continuous surface. It's mandatory for the conversion of Immerflow to a cut-cell software.

[ksamouchos]

Rebased on current branch #461

[ksamouchos]

Rebased on current branch #461

[ksamouchos]

Rebased on current branch #461

[ksamouchos]

Rebased on current branch #461. Additionally, small modifications were made to enable the return of the intersection point between an interface and the boundary.

[ksamouchos]

Rebased on current master

[ksamouchos]

After the review of @andrea-iob and the discussion I had with @marcocisternino and @andrea-iob I completely changed this branch by reprogramming everything from scratch. New features are introduced and others are removed. The new version is more generic and does not depend exclusively on my work related to the cut-cell method. This branch is responsible for:

1. introducing the pixel, triangle and polygon intersections with planes
2. the intersecting interfaces with smoothed continuous surfaces

The considerable amount of modifications requires a new review to the whole pull-request.

[ksamouchos]

I updated the branch following the review of @andrea-iob . I believe that @edoardolombardi and @roccoarpa should also be added as reviewers, because this pull-request affects their work.

[roccoarpa]

hello @ksamouchos, thank you for the warning, but for me it's difficult to do a proper review since there's lot of stuff going on in this pull (not only the part that interest me). This is my personal opinion: for me it's ok whatever you decide to do, as long as the code works and it's properly tested. I will update my part in mimic eventually. Ty.

[ksamouchos]

hello @ksamouchos, thank you for the warning, but for me it's difficult to do a proper review since there's lot of stuff going on in this pull (not only the part that interest me). This is my personal opinion: for me it's ok whatever you decide to do, as long as the code works and it's properly tested. I will update my part in mimic eventually. Ty.

ok, thank you for your kind response

[andrea-iob]

The changes are fine, however explicitly using the triangulation functions of the CG module would have been better because if these function are improved (for example adding support for concave polygons), also the intersection functions will befit of this improvements for free.

[ksamouchos]

The changes are fine, however explicitly using the triangulation functions of the CG module would have been better because if these function are improved (for example adding support for concave polygons), also the intersection functions will befit of this improvements for free.

The algorithm works fine for convex or concave polygons as it is. The triangulation is just a trick for using the plane-triangle intersection provided by bitpit and it doesn't have any important role in the geometrical construction. That's why it's removed without harming the generality of the algorithm.

[ksamouchos]

@edoardolombardi would you like to have a look before merging this branch?

[ksamouchos]

One last thing. I don't really like the introduction of function "computePolygonNormal" in CG.hpp. The computation of the polygon's normal already exist in module "patchkernel", but it has a different definition. I think that using the already existing function instead of "computePolygonNormal" won't harm the algorithm, but I'm not completely sure. That's why I kept Rocco's definition.

Let me know, if you have any better idea about how to improve this part.