ARtransPars: Function to parametrize a stationary AR process

In dlm: Bayesian and Likelihood Analysis of Dynamic Linear Models

Description

The function maps a vector of length p to the vector of autoregressive coefficients of a stationary AR(p) process. It can be used to parametrize a stationary AR(p) process

Usage

ARtransPars(raw)

Arguments

raw	a vector of length p

Details

The function first maps each element of raw to (0,1) using tanh. The numbers obtained are treated as the first partial autocorrelations of a stationary AR(p) process and the vector of the corresponding autoregressive coefficients is computed and returned.

Value

The vector of autoregressive coefficients of a stationary AR(p) process corresponding to the parameters in raw.

Author(s)

Giovanni Petris, [e-mail]

References

Jones, 1987. Randomly choosing parameters from the stationarity and invertibility region of autoregressive-moving average models. Applied Statistics, 36.

Examples

(ar <- ARtransPars(rnorm(5)))
all( Mod(polyroot(c(1,-ar))) > 1 ) # TRUE

Example output

[1] -2.0249046 -0.9750569  1.1852448  1.9619655  0.8326637
[1] TRUE

dlm documentation built on May 2, 2019, 4:58 p.m.