ssfortran.Cycle#

class ssfortran.Cycle(cov='diagonal', damped=True, period_bounds=(2.0, None), at_observations=False)#

Stochastic cycle (DK §3.2.4).

\((c_{t+1}, c^*_{t+1})' = \rho C(\lambda) (c_t, c^*_t)' + w_t\) with \(C(\lambda)\) the rotation by frequency \(\lambda\). Parameters: the disturbance variance, \(\lambda\) and, if damped, \(\rho\).

Parameters:
cov{“diagonal”, “full”, None}

Form of \(Var(w_t)\).

dampedbool

Estimate \(\rho \in (0, 1)\); the cycle is then stationary and starts from its unconditional distribution. Otherwise \(\rho = 1\) and the cycle is diffuse.

period_boundstuple of float

Bounds on the period \(2\pi / \lambda\), in observations; None as upper bound means the sample size.

at_observationsbool

Act on the observations rather than the signals.