So far in ComputeGeometry points are flattened in one list, passed to Face2D (resp. 3D) where they are put back into matrix form of shape (n_vertices, 2)' (resp.3`). This avoids doing unnecessary works and keeps the matrix form.
This PR passes the test of the present repository. It passes 2D coupling tests with CoMMA (with and without anisotropy). However, unfortunately, when running this kind of tests in 3D I got a segmentation fault.
So far in
ComputeGeometry
points are flattened in one list, passed toFace2D
(resp.3D
) where they are put back into matrix form of shape(n_vertices, 2)' (resp.
3`). This avoids doing unnecessary works and keeps the matrix form.This PR passes the test of the present repository. It passes 2D coupling tests with
CoMMA
(with and without anisotropy). However, unfortunately, when running this kind of tests in 3D I got a segmentation fault.