Stochastic approximation, with applications. (Q4407573)

Many applications lead (from the mathematical point of view) to the problem of finding a root of some function. There exist many common (deterministic) numerical methods to solve this problem in the case when the above mentioned function is observed without errors. In the case when the observations are corrupted by errors, the stochastic approximation methods can be employed. These methods belong to the iterative methods and they were, first, introduced by Robbins and Monro. The asymptotic behaviour of these methods have been investigated (in the literature) mostly under the assumptions that the observed errors can be modelled as a sequence of independent identically distributed random variables and, moreover, that they are independent of the value of the function. The aim of the paper is to deal with the case when the last assumption is not fulfilled. In detail, a new algorithm is suggested (for this case) in the paper. At the end, the algorithm is applied to an adaptive filtering problem. For the entire collection see [Zbl 1005.00025].