porepy.utils.sort_points module

Functions to sort points and edges belonging to geometric objects.

sort_multiple_point_pairs(lines)

Function to sort multiple pairs of points to form circular chains.

The routine contains essentially the same functionality as sort_point_pairs, but stripped down to the special case of circular chains. Differently to sort_point_pairs, this variant sorts an arbitrary amount of independent point pairs. The chains are restricted by the assumption that each contains equally many line segments. Finally, this routine uses numba.

Parameters:

lines (np.ndarray) – Array of size 2 * num_chains x num_lines_per_chain, containing node indices. For each pair of two rows, each column represents a line segment connectng the two nodes in the two entries of this column.

Returns:

Sorted version of lines, where for each chain, the collection

of l

Return type:

np.ndarray

:raises ImportError ifine segments has been potentially flipped` and :py:class:`sorted.:

sort_point_pairs(lines, check_circular=True, is_circular=True)

Sort pairs of numbers to form a chain.

The target application is to sort lines, defined by their start end endpoints, so that they form a continuous polyline.

The algorithm is brute-force, using a double for-loop. This can surely be improved.

Parameters:

• lines (np.ndarray, 2xn) – the line pairs. If lines has more than 2 rows, we assume that the points are stored in the first two rows.

• check_circular (bool) – Verify that the sorted polyline form a circle.

• is_circular (bool) – if the lines form a closed set. Default is True.

Returns:

sorted line pairs. If lines had more than 2 rows,

the extra are sorted accordingly.

np.ndarray, n: Sorted column indices, so that

sorted_lines = lines[:, sort_ind], modulo flipping of rows in individual columns

Return type:

np.ndarray, 2xn

sort_point_plane(pts, centre, normal=None, tol=1e-5)

Sort the points which lie on a plane.

The algorithm assumes a star-shaped disposition of the points with respect the centre.

Parameters:

• pts (ndarray) – np.ndarray, 3xn, the points.

• centre (ndarray) – np.ndarray, 3x1, the face centre.

• normal (ndarray | None) – (optional) the normal of the plane, otherwise three points are required.

• tol (float) – Absolute tolerance used to identify active (non-constant) dimensions.

Returns:

np.array, 1xn, sorted point ids.

Return type:

map_pts

sort_triangle_edges(t)

Sort a set of triangles so that no edges occur twice with the same ordering.

For a planar triangulation, this will end up with all the triangles being ordered CW or CCW. In cases where the triangulated surface(s) do not share a common plane, methods based on geometry are at best cumbersome. This approach should work also in those cases.

Parameters:

t (np.ndarray, 3 x n_tri) – Triangulation to have vertexes ordered.

Returns:

With the vertexes ordered.

Return type:

np.ndarray, 3 x n_tri

Example

>>> t = np.array([[0, 1, 2], [1, 2, 3]]).T
>>> sort_triangle_edges(t)
np.array([[0, 2], [1, 1], [2, 3]])