Necessary and Sufficient Condition for Nonsingular Fisher Information Matrix in ARMA Models
Applications
2016-11-07 v1 Statistics Theory
Statistics Theory
Abstract
It is demonstrated that a necessary and sufficient condition that the Fisher information matrix of an ARMA model be nonsingular is that the model not be redundant, that is, the autoregressive and moving-average polynomials do not share common roots.
Keywords
Cite
@article{arxiv.1611.01387,
title = {Necessary and Sufficient Condition for Nonsingular Fisher Information Matrix in ARMA Models},
author = {A. Ian McLeod},
journal= {arXiv preprint arXiv:1611.01387},
year = {2016}
}
Comments
5 pages