Simple crowd dynamics to generate complex temporal contact networks

TL;DR

This paper explores whether simple 2D particle models can replicate complex temporal contact network properties observed in real human interactions, focusing on contact durations and inter-contact time distributions.

Contribution

It demonstrates that basic crowd particle models can reproduce key dynamical features of empirical contact networks, linking inter-contact durations to first-return processes.

Findings

Inter-contact duration distributions are reproducible by simple models.

The -3/2 exponent is explained by first-return process dynamics.

Models successfully mimic empirical network temporal properties.

Abstract

Empirical contact networks or interaction networks demonstrate peculiar characteristics stemming from the fundamental social, psychological, physical mechanisms governing human interactions. Although these mechanisms are complex, we test whether we are able to reproduce some dynamical properties of these empirical networks from relatively simple models. In this study, we perform simulations for a range of 2D models of particle dynamics, namely the Random Walk, Active Brownian Particles, and Vicsek models, to generate artificial contact networks. We investigate temporal properties of these contact networks: the distributions of contact durations, inter-contact durations and number of contact per pair of particle. We demonstrate that the distribution of inter-contact durations can be recovered by the dynamics of these simple crowd particle models, and show that it is simply related to the well-know first-return process, which explains the -3/2 exponent that is found in both the numerical models and empirical contact networks.