Interacting Jump Processes Preserve Semi-Global Markov Random Fields on Path Space

arXiv2210.09253MaRDI QIDQ6414218FDOQ6414218

Publication date: 17 October 2022

Abstract: Consider a system of interacting particles indexed by the nodes of a graph whose vertices are equipped with marks representing parameters of the model such as the environment or initial data. Each particle takes values in a countable state space and evolves according to a (possibly non-Markovian) continuous-time pure jump process whose jump intensities depend only on its own state (or history) and marks as well as the states (or histories) and marks of particles and edges in its neighborhood in the graph. Under mild conditions on the jump intensities, it is shown that the trajectories of the interacting particle system exhibit a certain local or semi-global Markov random field property whenever the initial condition satisfies the same property. Our results complement recent works that establish the preservation of a local second-order Markov random field property for interacting diffusions. Our proof methodology in the context of jump processes is different, and works directly on infinite graphs, thereby bypassing any limiting arguments. Our results apply to models arising in diverse fields including statistical physics, neuroscience, epidemiology and opinion dynamics, and have direct applications to the study of marginal distributions of interacting particle systems on Cayley trees.

Mathematics Subject Classification ID

Interacting random processes; statistical mechanics type models; percolation theory (60K35) Queueing theory (aspects of probability theory) (60K25) Branching processes (Galton-Watson, birth-and-death, etc.) (60J80) Jump processes on discrete state spaces (60J74)

This page was built for publication: Interacting Jump Processes Preserve Semi-Global Markov Random Fields on Path Space

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q6414218)