property CFASimulationSmoother.posterior_cov

Posterior covariance of the states conditional on the data


Warning: the matrix computed when accessing this property can be extremely large: it is shaped (nobs * k_states, nobs * k_states). In most cases, it is better to use the posterior_cov_inv_chol_sparse property if possible, which holds in sparse diagonal banded storage the Cholesky factor of the inverse of the posterior covariance matrix.

\[Var[\alpha \mid Y^n ]\]

This posterior covariance matrix is not identical to the smoothed_state_cov attribute produced by the Kalman smoother, because it additionally contains all cross-covariance terms. Instead, smoothed_state_cov contains the (k_states, k_states) block diagonal entries of this posterior covariance matrix.

Last update: Dec 14, 2023