Mean-field Concentration of Opinion Dynamics in Random Graphs

TL;DR

This paper analyzes how the stable opinions in a social network modeled by opinion dynamics concentrate around a mean-field approximation when the network is an Erdős-Rényi random graph, especially as the network size grows.

Contribution

It proves that the expected stable opinions in random graphs concentrate around the mean-field solution under certain conditions, extending understanding of opinion dynamics in uncertain networks.

Findings

Expected stable opinions concentrate around the mean-field solution as network size increases.

Concentration holds under the ll_{0}-norm for both directed and undirected Erd
51s-Re9nyi graphs.

The mean-field approximation applies to general analytic matrix functions, with norm extensions under certain degree conditions.

Abstract

Opinion and belief dynamics are a central topic in the study of social interactions through dynamical systems. In this work, we study a model where, at each discrete time, all the agents update their opinion as an average of their intrinsic opinion and the opinion of their neighbors. While it is well-known how to compute the stable opinion state for a given network, studying the dynamics becomes challenging when the network is uncertain. Motivated by the task of finding optimal policies by a decision-maker that aims to incorporate the opinion of the agents, we address the question of how well the stable opinions can be approximated when the underlying network is random. We consider Erd\H{o}s-R\'enyi random graphs to model the uncertain network. Under the connectivity regime and an assumption of minimal stubbornness, we show the expected value of the stable opinion $\mathbf{E}(x(G,\infty))$ concentrates, as the size of the network grows, around the stable opinion $\bar{x}(\infty)$ obtained by considering a mean-field dynamical system, i.e., averaging over the possible network realizations. For both the directed and undirected graph model, the concentration holds under the $\ell_{\infty}$-norm to measure the gap between $\mathbf{E}(x(G,\infty))$ and $\bar{x}(\infty)$. We deduce this result by studying a mean-field approximation of general analytic matrix functions. The approximation result for the directed graph model also holds for any $\ell_{\rho}$-norm with $\rho\in (1,\infty)$, under a slightly enhanced expected average degree.