A factorial covariance structure model for space-time multivariate stochastic processes (Q1283297)

A factorial covariance structure model is one in which the (observable) normal vector \({\mathbf y}\) has \(E({\mathbf y})= \mathbf{0}\), and \(E({\mathbf {yy}}')= \Sigma_1 \otimes \Sigma_2 \otimes \Sigma_3= \Sigma\). The authors illustrate the use of this structure with two examples of spacetime stochastic processes, namely: (1) STAR (spacetime autoregressive) models, and (2) evolution models governed by systems of differential equations depending on an unknown parameter. In this paper the authors consider procedures for estimating the corresponding Gaussian likelihood function.