LinearGaussianSSM.stationary_covariance#

classmethod LinearGaussianSSM.stationary_covariance(transition, selection, state_cov)[source]#

Solve the discrete Lyapunov equation for the stationary state covariance.

Parameters:

  • transition (ArrayLike) – Time-invariant transition matrix T (must be stable).

  • selection (ArrayLike) – Selection matrix R.

  • state_cov (ArrayLike) – State disturbance covariance Q.

Return type:

NDArray[double]

Returns:

The stationary covariance P solving P = T P T' + R Q R'.

Raises:

NumericalError – If T is not stable (has an eigenvalue on or outside the unit circle).