autofragment.algorithms.seeding

Seeding strategies for k-means initialization.

Provides pluggable strategies that generate initial centroids to guide k-means clustering toward physically meaningful molecular partitions.

Each strategy takes an (N, 3) array of molecular centroids and the desired number of clusters, returning an (n_clusters, 3) array of seed positions.

Functions

compute_seeds(centroids, n_clusters, ...)	Compute initial centroid seeds for k-means.
get_strategy(name)	Look up a seeding strategy by name.