- ott.tools.gaussian_mixture.fit_gmm.initialize(rng, points, point_weights, n_components, n_attempts=50, verbose=False)#
Initialize a GMM via K-means++ with retries on failure.
PRNGKeyArray) – jax.random key
Array) – (n, n_dimensions) array of observations
int) – desired number of components
int) – number of attempts to initialize before failing
bool) – if True, print status information
- Return type
An initial Gaussian mixture model.
ValueError if initialization was unsuccessful after n_attempts attempts. –