Computational Aspects of a Method of Stochastic Approximation

Abstract

A method of stochastic approximation is studied in the framework of the general convergence theory for families of linear polynomial operators of interpolation type. The description of the corresponding computational procedure, in particular, its input parameters, is given. Some optimization problems and aspects of implementation of the algorithm by means of Maple are discussed. It is shown that the algorithm can be applied not only to problems of “pure approximation” in the spaces $L_p$ with $0<p\le +\infty$, but also to problems of signal processing, especially, if one is interested in strong oscillating data or data containing an essential stochastic item.