Shrinkage Priors on Complex-Valued Circular- Symmetric Autoregressive Processes

We investigate shrinkage priors on power spectral densities for complex-valued circular-symmetric autoregressive processes. construct predictive densities, which asymptotically dominate (i) the Bayesian density based Jeffreys prior and (ii) estimative with maximum likelihood estimator, where Kullback-Leibler divergence from true to a is adopted as risk. Furthermore, we propose general constructions of objective Kähler parameter spaces by utilizing positive continuous eigenfunction Laplace-Beltrami operator negative eigenvalue. present numerical experiments stationary model order 1.