gustaf.utils.arr.close_rows#

gustaf.utils.arr.close_rows(arr, tolerance=None, return_intersection=False, nthreads=None, **_kwargs)[source]#

Similar to unique_rows, but if data type is floats, use this one. Performs radius search using KDTree. Currently uses scipy.spatial.cKDTree.

Parameters:

  • arr ((n, d) array-like)

  • tolerance ((float)) – Defaults to None.

  • return_intersection (bool) – Default is False. Returns intersection. For vertices with singular points, this will take a lot of memory space.

  • nthreads (int) – number of concurrent query. In case of napf, concurrent build as well. Default is taken from settings.NTHREADS

Returns:

  • unique_arrays ((n, d) np.ndarray)

  • unique_ids ((m) np.ndarray)

  • inverse ((n) np.ndarray)

  • overlapping (list(list)) – id of neighbors within the tolerance.