Spectral analysis of sample autocovariance matrices of a class of linear time series in moderately high dimensions (Q2405106)

The article performs a spectral analysis for multivariate linear time series within a moderately high-dimensional case (when dimensionality increases slower than sample size). It demonstrates the existence of limiting spectral distributions of the renormalized and symmetrized sample autocovariance matrices under these conditions. The theoretical properties proved in this article are useful for model parametrization and diagnostics, as well as for prediction, within multivariate linear time series.