pypcga.PCGA.get_eigen_post_cov#

PCGA.get_eigen_post_cov(is_direct_solve: bool | None = None, inflation: float | None = None, n_pc: int | None = None, random_state: int | Generator | RandomState | None = None) → CovViaEigenFactorization[source]#

Return the posterior covariance matrix through an Eigen factorization.

Notes

This is practical for posterior sampling.

Parameters:

  • is_direct_solve (Optional[bool], optional) – _description_, by default None

  • inflation (Optional[float], optional) – Inflation factor used to build the posterior covariance matrix. If None, the random_state used by PCGA is taken. By default None.

  • n_pc (Optional[int], optional) – Number of principal component used for the eigen factorization of the posterior covariance matrix. It can differ from the number of PC used by PCGA. If None, the number of PC used by PCGA is taken. By default None.

  • random_state (Optional[ Union[int, np.random.Generator, np.random.RandomState]]) – If None, the random_state used by PCGA is taken. By default None.

Returns:

Low rank approximation of the posterior covariance matrix.

Return type:

covmats.CovViaEigenFactorization