Estimation methods for a linearly structured covariance matrix

Covariance matrices with a linear structure are widely used in multivariate
analysis. The choice of the most appropriate covariance structure can be
made from a class of possible linear structures. Once we have made the
choice, an important question is how we can estimate the covariance matrix
for a given covariance structure. This article describes methods used to
estimate the structured covariance matrix, and indicates the advantages
and disadvantages of the selected methods.