Alternating birth-death processes

Abstract

We consider a continuous time Markov process on $\mathbb{N}_0$ which can be interpreted as generalized alternating birth-death process in a non-autonomous random environment. Depending on the status of the environment the process either increases until the environment changes and the process starts to decrease until the environment changes again, and the process restarts to increase, and so on, or its starts decreasing, reversing its direction due to environmental changes, et cetera. The birth and death rates depend on the state (height, population size) of the birth-death process and the environment's transition rates depend on the state of the birth-death process as well. Moreover, a birth or death event may trigger an immediate change of the environment. Our main result is an explicit expression for the stationary distribution if the system is ergodic, providing ergodicity conditions as well. Removing the reflecting boundary at zero we obtain a two-sided version on $\mathbb{Z}$ of this alternating birth-death process, which for suitable parameter constellations is ergodic as well. We determine the stationary distribution. This two-sided version is a locally inhomogeneous discrete space version of the classical telegraph process. We demonstrate that alternating birth-death processes in a random environment provide a versatile class of models from different areas of applications. Examples from the literature are discussed.