Almost sure convergence of recursive density estimators for stationary mixing processes (Q1094772): Difference between revisions

Kernel estimators of the marginal probability density of a multivariate stationary mixing process are considered. Under certain conditions on the kernel, bandwidth parameters and the rate of mixing, strong consistency and the rates of strong convergence of kernel estimators are obtained. The proof depends mainly on the convergence property of mixingales, which is a generalization of the martingale convergence property.