pflm.fpca.utils.get_covariance_matrix#

get_covariance_matrix(raw_cov: ndarray, obs_grid: ndarray) → ndarray[source][source]#

Aggregate raw covariance entries onto a dense symmetric matrix.

This maps the compact raw covariance table returned by get_raw_cov to a dense covariance matrix on the observation grid by summing weighted covariances per unique (t1, t2) pair and normalizing by (sum_w - 1).

Parameters:

raw_covnp.ndarray of shape (M, 5): Columns are (sid, t1, t2, w, cov), typically from get_raw_cov.
obs_gridnp.ndarray of shape (nt,): Sorted unique observation grid values.

Returns:

cov_matrixnp.ndarray of shape (nt, nt): Symmetric covariance matrix aligned to obs_grid.

Raises:

ValueError: If raw_cov does not have at least 5 columns, or obs_grid is empty.