ssfortran.AugmentedResults

class ssfortran.AugmentedResults(filter, delta, delta_cov, llf, llf_fixed, llf_marginal, smoothed_state, smoothed_state_cov)

Output of the augmented Kalman filter and smoother (DK §5.7).

The diffuse elements \(\delta\) of the initial state (DK eq. 5.2) are treated as unknown constants: the filter runs once with \(\delta = 0\) and carries their effect as extra columns.

Attributes:

filterFilterResults

The filter with \(\delta = 0\).

deltandarray, shape (k,)

Generalized least squares estimate of \(\delta\).

delta_covndarray, shape (k, k)

Its variance.

llffloat

Diffuse log likelihood (DK §7.2.3); equal to that of the exact initial filter.

llf_fixedfloat

Log likelihood with \(\delta\) fixed but unknown (DK §7.2.4).

llf_marginalfloat

Marginal log likelihood (DK §7.2.6; Francke, Koopman and de Vos 2010).

smoothed_statendarray, shape (m, n)

Smoothed state.

smoothed_state_covndarray, shape (m, m, n)

Its variance, including the uncertainty about \(\delta\).