In this paper we study the convergence properties of the power series algor
ithm, which gives a general method to determine (functions of) stationary d
istributions of Markov chains. We show that normalization plays a crucial r
ole and that the convergence can be improved by introducing some minor chan
ges in the algorithm. We illustrate this with several numerical examples. (
C) 2000 Elsevier Science B.V. All rights reserved.