On Opinion Dynamics in Heterogeneous Networks

TL;DR

This paper analyzes a complex opinion dynamics model with heterogeneous confidence ranges, classifies agent interactions, and provides conditions for convergence to steady states, enhancing understanding of opinion evolution in diverse networks.

Contribution

It introduces novel sufficient conditions for convergence and classifies agents based on their interaction topology in the Hegselmann-Krause model.

Findings

Convergence to steady states is guaranteed under certain conditions.

The model's topology remains constant over infinite time under these conditions.

Leader groups influence the convergence rate and direction.

Abstract

This paper studies the opinion dynamics model recently introduced by Hegselmann and Krause: each agent in a group maintains a real number describing its opinion; and each agent updates its opinion by averaging all other opinions that are within some given confidence range. The confidence ranges are distinct for each agent. This heterogeneity and state-dependent topology leads to poorly-understood complex dynamic behavior. We classify the agents via their interconnection topology and, accordingly, compute the equilibria of the system. We conjecture that any trajectory of this model eventually converges to a steady state under fixed topology. To establish this conjecture, we derive two novel sufficient conditions: both conditions guarantee convergence and constant topology for infinite time, while one condition also guarantees monotonicity of the convergence. In the evolution under fixed topology for infinite time, we define leader groups that determine the followers' rate and direction of convergence.