python - How does sklearn.cluster.KMeans handle an init ndarray parameter with missing centroids (available centroids less than n_clusters)? -

in python sklearn kmeans (see documentation), wondering happens internally when passing ndarray of shape (n, n_features) init parameter, when n<n_clusters

1. does drop given centroids , starts kmeans++ initialization default choice init parameter ? (pdf paper kmeans++) (how kmeans++ work)
2. does consider given centroids , fill accordingly remaining centroids using kmeans++ ?
3. does consider given centroids , fill remaining centroids using random values ?

i didn't expect method returns no warning in case. that's why need know how manages this.

if give mismatching init adjust number of clusters, can see source. not documented , consider bug. i'll propose fix it.