Maximization of the Power of Hypotheses Testing for the s-th Order Markov Chains by Empirical Characteristic Functions

Abstract

The problem of hypotheses testing based on empirical characteristic functions is considered for the s-th order Markov chains. The problem of maximization of the test power is solved in the case of simple hypotheses, in general case with training samples, and in the case of contiguous hypotheses. Numerical results are presented.