Collective fluctuations in networks of noisy components

TL;DR

This paper analytically investigates how network connectivity influences fluctuations in collective dynamics of noisy components, revealing that fluctuations decay slowly with system size and can remain finite in large directed networks.

Contribution

It provides an analytical framework linking network structure to fluctuation intensity, showing deviations from the central limit theorem in directed and nonlinear networks.

Findings

Fluctuations decrease more slowly than 1/√N in directed networks.

Fluctuations can remain finite even in large systems with global directionality.

Nonlinear systems like coupled oscillators exhibit similar fluctuation behaviors.

Abstract

Collective dynamics result from interactions among noisy dynamical components. Examples include heartbeats, circadian rhythms, and various pattern formations. Because of noise in each component, collective dynamics inevitably involve fluctuations, which may crucially affect functioning of the system. However, the relation between the fluctuations in isolated individual components and those in collective dynamics is unclear. Here we study a linear dynamical system of networked components subjected to independent Gaussian noise and analytically show that the connectivity of networks determines the intensity of fluctuations in the collective dynamics. Remarkably, in general directed networks including scale-free networks, the fluctuations decrease more slowly with the system size than the standard law stated by the central limit theorem. They even remain finite for a large system size when global directionality of the network exists. Moreover, such nontrivial behavior appears even in undirected networks when nonlinear dynamical systems are considered. We demonstrate it with a coupled oscillator system.