About cookies on this site Our websites require some cookies to function properly (required). In addition, other cookies may be used with your consent to analyze site usage, improve the user experience and for advertising. For more information, please review your options. By visiting our website, you agree to our processing of information as described in IBM’sprivacy statement. To provide a smooth navigation, your cookie preferences will be shared across the IBM web domains listed here.
Publication
Communications in Statistics. Part C: Stochastic Models
Paper
Matrix-geometric invariant measures for G/M/1 type Markov chains
Abstract
Necessary and sufficient conditions for an irreducible Markov chain of G/M/1 type to have an invariant measure that is matrix-geometric are given. For example, it is shown that such an invariant measure exists when a(z), the generating function corresponding to transitions in the homogeneous part of the chain, is either an entire function or a rational function. This generalizes a recent result of Latouche, Pearce and Taylor, who showed that a matrix-geometric invariant measure always exists for level-independent quasi-birth-and-death processes. Conditions ensuring the uniqueness of such an invariant measure up to multiplication by a positive constant are also given. Examples of G/M/1 type Markov chains with no matrix-geometric invariant measure and with more than one distinct matrix-geometric invariant measure are presented. As a byproduct of our work, it is shown in the transient case that if det{zI - a(z)} = 0 has a solution in the exterior of the closed unit disk, then the solution of smallest modulus there is real and positive. Copyright © 1998 by Marcel Dekker, Inc.