structuretoolkit.analyse.spatial.cluster_by_steinhardt

structuretoolkit.analyse.spatial.cluster_by_steinhardt(positions: ndarray, neigh, l_values: list[int], q_eps: float, var_ratio: float, min_samples: int) → ndarray

Clusters candidate positions via Steinhardt parameters and the variance in distances to host atoms.

The cluster that has the lowest variance is returned, i.e. those positions that have the most “regular” coordination polyhedra.

Parameters:

• positions (array) – candidate positions

• neigh (Neighbor) – neighborhood information of the candidate positions

• l_values (list of int) – which steinhardt parameters to use for clustering

• q_eps (float) – maximum intercluster distance in steinhardt parameters for DBSCAN clustering

• var_ratio (float) – multiplier to make steinhardt’s and distance variance numerically comparable

• min_samples (int) – minimum size of clusters

Returns:

Positions of the most likely interstitial sites

Return type:

array