tvarCycle: Time-varying cycle component

In dajmcdon/myFKF: Fast Kalman Filtering and Smoothing

Description

Time-varying cycle component

Usage

tvarCycle(parmlist)

Arguments

parmlist

A list of parameters as generated by parmList.

Details

This function generates state space model matrices related to an a time-varying cyclic component. Such a specification is usually more slowly varying than a seasonal component, say for a yearly cycle with daily data. The model has the form:

y(t) = Z_c * c(t) + extra stuff

c(t+1) = Tt_c * c(t) + R * sig_cycle

In particular, the vector c(t) evolves as

c_1(t+1) = c_1(t) * cos(lam) + c_2(t) * sin(lam) + omega_1

c_2(t+1) = -c_1(t) * sin(lam) + c_2(t) * cos(lam) + omega_2

where omega_1 and omega_2 are iid normal with mean 0 and variance sig2cycle, and lam = 2π/period.

For a particular period (parmlist$period), this function produces the appropriate model matrices above.

Value

A list of matrices with components Z, Tt, Q, and R.

dajmcdon/myFKF documentation built on May 3, 2019, 5:16 p.m.