Structural classification of continuous time Markov chains with applications

Abstract

This paper is motivated by examples from stochastic reaction network theory. The Q-matrix of a stochastic reaction network can be derived from the reaction graph, an edge-labelled directed graph encoding the jump vectors of an associated continuous time Markov chain on the invariant space ${\mathbb{N}}_0^d$. An open question is how to decompose the space ${\mathbb{N}}_0^d$ into neutral, trapping, and escaping states, and open and closed communicating classes, and whether this can be done from the reaction graph alone. Such general continuous time Markov chains can be understood as natural generalizations of birth-death processes, incorporating multiple different birth and death mechanisms. We characterize the structure of ${\mathbb{N}}_0^d$ imposed by a general Q-matrix generating continuous time Markov chains with values in ${\mathbb{N}}_0^d$, in terms of the set of jump vectors and their corresponding transition rate functions. Thus the setting is not limited to stochastic reaction networks. Furthermore, we define structural equivalence of two Q-matrices, and provide sufficient conditions for structural equivalence. Examples are abundant in applications. We apply the results to stochastic reaction networks, a Lotka-Volterra model in ecology, the EnvZ-OmpR system in systems biology, and a class of extended branching processes, none of which are birth-death processes.

Keywords:

• Q-matrix
• birth-death processes
• structural equivalence
• stochastic reaction networks
• positive irreducible components
• quasi irreducible components
• extinction
• persistence

Acknowledgments

The work was initiated with most part of it done when the first author was at the University of Copenhagen. The authors thank the editors' and referees' comments which helped improve the presentation of the paper. The authors acknowledge the support from The Erwin Schrödinger Institute (ESI) for the workshop on “Advances in Chemical Reaction Network Theory”.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Additional information

Funding

CX acknowledges the support from TUM University Foundation and the Alexander von Humboldt Foundation. CW acknowledges support from the Novo Nordisk Foundation (Denmark), grant NNF19OC0058354.