This paper builds on earlier work in  on metrics for power spectral densities (PSD) of multivariable time-series. We present an approach to quantify dissimilarities aimed at optimal prediction and smoothing. Divergence measures are constructed based on the degradation of prediction-error and smoothing-error variances. These induce Riemannian metrics which generalize earlier results for scalar PSD's.
|Original language||English (US)|
|Number of pages||6|
|Journal||Proceedings of the IEEE Conference on Decision and Control|
|State||Published - Dec 1 2012|
|Event||51st IEEE Conference on Decision and Control, CDC 2012 - Maui, HI, United States|
Duration: Dec 10 2012 → Dec 13 2012