Synchronization Invariance Under Network Structural Transformations

TL;DR

This paper introduces network transformations that preserve synchronization behavior in complex systems of oscillators, enabling the mapping between different network structures while maintaining collective dynamics.

Contribution

It proposes a novel information-theoretic method to transform network weights, preserving synchronization, and reveals the differing informational requirements for mapping homogeneous and heterogeneous networks.

Findings

Heterogeneous networks can be mapped to homogeneous ones with local information.

Mapping from homogeneous to heterogeneous networks requires higher-order information.

The formalism offers new insights into handling uncertainty in network connectivity measurements.

Abstract

Synchronization processes are ubiquitous despite the many connectivity patterns that complex systems can show. Usually, the emergence of synchrony is a macroscopic observable, however, the microscopic details of the system, as e.g. the underlying network of interactions, is many times partially or totally unknown. We already know that different interaction structures can give rise to a common functionality, understood as a common macroscopic observable. Building upon this fact, here, we propose network transformations that keep the collective behavior of a large system of Kuramoto oscillators functionally invariant. We derive a method based on information theory principles, that allows us to adjust the weights of the structural interactions to map random homogeneous -in degree- networks into random heterogeneous networks and vice-versa, keeping synchronization values invariant. The results of the proposed transformations reveal an interesting principle; heterogeneous networks can be mapped to homogeneous ones with local information, but the reverse process needs to exploit higher-order information. The formalism provides new analytical insight to tackle real complex scenarios when dealing with uncertainty in the measurements of the underlying connectivity structure.