Software for Multivariate Spectrum Analysis

Abstract

The development of a subprogram library for the examination of the interrelations between frequency components of multivariate time series is discussed. Much of the present work is concerned with the computation of reliable estimates of the transfer function and impulse response, which describe the linear time invariant relationship between stationary time series. The parameters employed are simple extensions of multivariate regression theory and are immediate functions of the discrete Fourier Transform, which may be computed rapidly by means of the fast Fourier transform algorithm. This also allows for an independent analysis to be carried out for each frequency.