A globally convergent matricial algorithm for multivariate spectral estimation

TL;DR

This paper introduces a globally convergent matricial Newton-type algorithm for multivariate spectral estimation, demonstrating its effectiveness through simulations and offering an alternative to traditional identification methods.

Contribution

The paper presents a novel globally convergent algorithm for multivariate spectral estimation, with proven convergence and practical validation via simulations.

Findings

Algorithm converges globally in spectral estimation tasks.

Effective alternative to PEM and N4SID with short data records.

Simulation results confirm practical applicability.

Abstract

In this paper, we first describe a matricial Newton-type algorithm designed to solve the multivariable spectrum approximation problem. We then prove its global convergence. Finally, we apply this approximation procedure to multivariate spectral estimation, and test its effectiveness through simulation. Simulation shows that, in the case of short observation records, this method may provide a valid alternative to standard multivariable identification techniques such as MATLAB's PEM and MATLAB's N4SID.