auto_VARMA: computes autocovariances of SVARMA usng frequency domain

In jlivsey/sigex: Ecce Signum: Multivariate Signal Extraction

computes autocovariances of SVARMA usng frequency domain

Description

Backgroud: function computes autocovariances of SVARMA (p,q,ps,qs) from lag zero to maxlag, with array inputs phi and theta. SVARMA equation: (1 - phi[1]B ... - phi[p]B^p) (1 - Phi[1]B^s ... - Phi[ps]B^s*ps) X_t = (1 + theta[1]B ... + theta[q]B^q) (1 + Theta[1]B^s ... + Theta[qs]B^s*qs) WN_t.

Usage

auto_VARMA(param, p, q, ps, qs, season, grid, maxlag)

Arguments

param	- matrix of dimension m x (p+q+ps+qs+1), equals [ phi | theta | phiseas | thetaseas | sigma ] phi: block matrix of dimension N x N*p of VAR coefficients theta: block matrix of dimension N x N*q of VMA coefficients phiseas: block matrix of dimension N x N*ps of SVAR coefficients thetaseas: block matrix of dimension N x N*qs of SVMA coefficients sigma: N x N covariance matrix of white noise
p	- AR order
q	- MA order
ps	- seasonal AR order
qs	- seasonal MA order
season	- period (e.g. monthly = 12)
grid	- Riemann mesh size
maxlag	- maximum autocovariance lag needed

Value

autocovariances at lags 0 through maxlag, as array of dimension m x m x (maxlag+1)

jlivsey/sigex documentation built on March 20, 2024, 3:17 a.m.