Maximum entropy for dynamic processes on networks
Abstract
Dynamic processes on networks are fundamental to understanding modern-day phenomena such as information diffusion and opinion polarization on the internet or epidemics spreading through society. However, such processes are notoriously difficult to study broadly as small changes in initial conditions, the process, or the network can lead to very different evolution trajectories. Here we apply the information-theoretic framework of maximum caliber to study the statistics of such systems analytically, focusing on processes that can be interpreted as driven by interactions between populations of different types of individuals in the network. We verify the dynamics deduced from maximum caliber by using simulations of different processes on different networks, introduce an approximation of the dynamics that significantly simplifies the problem, and show that the approximation can be used to recover well-established models of population dynamics that are typically not thought of as taking place on networks.
Cite
@article{arxiv.2504.04240,
title = {Maximum entropy for dynamic processes on networks},
author = {Noam Abadi and Franco Ruzzenenti},
journal= {arXiv preprint arXiv:2504.04240},
year = {2025}
}