otmol.tl.perturbation_before_gw

otmol.tl.perturbation_before_gw(X_A, X_B, p_list=[1], n_trials=100, scale=0.1)

Find various suboptimal transport plans between clusters of atoms.

When calculating the distance matrix, Gaussian noise is added to the coordinates to generate various suboptimal transport plans. We first do a GW, then do a Kantorovich (ot.emd) on the aligned coordinates from GW.

Parameters

• X_A (numpy.ndarray) – Coordinates of cluster A.

• X_B (numpy.ndarray) – Coordinates of cluster B.

• p_list (list,) – Power of the distance matrix, by default [1].

• n_trials (int, optional) – Number of trials to run, by default 100.

• scale (float, optional) – Standard deviation of the Gaussian noise, by default 0.1.

Returns

list_perms – List of permutations.

Return type

List[numpy.ndarray]