After the calculation of the trajectories of the individual measures and the trajectories of pairwise correlations/covariances of the individual measures are obtained, the developmental analysis of the structural features of the covariance matrices of the observed measures is performed by the calculation of Riemannian (geodesic) distances between covariance matrices. Covariance matrices are properly analyzed as tensors in the non-Euclidean space of positive definite symmetric (PDS) matrices. The methods of tensorial analysis enable the analysis of the developmental process of a system characterized by a multivariate data set as a single system, rather than simply as a collection of individual items. Tensorial analysis provides only measures of the structural relations between matrices; it does not indicate which elements of the matrices are responsible for the distances calculated.
