Hypernetworks: cluster synchronisation is a higher-order effect

Sören von der Gracht,Eddie Nijholt,Bob Rink

2023-08-10

Abstract:Many networked systems are governed by non-pairwise interactions between nodes. The resulting higher-order interaction structure can then be encoded by means of a hypernetwork. In this paper we consider dynamical systems on hypernetworks by defining a class of admissible maps for every such hypernetwork. We explain how to classify robust cluster synchronisation patterns on hypernetworks by finding balanced partitions, and we generalize the concept of a graph fibration to the hypernetwork context. We also show that robust synchronisation patterns are only fully determined by polynomial admissible maps of high order. This means that, unlike in dyadic networks, cluster synchronisation on hypernetworks is a higher-order, i.e., nonlinear effect. We give a formula, in terms of the order of the hypernetwork, for the degree of the polynomial admissible maps that determine robust synchronisation patterns. We also demonstrate that this degree is optimal by investigating a class of examples. We conclude by displaying how this effect may cause remarkable synchrony breaking bifurcations that occur at high polynomial degree.

Dynamical Systems

What problem does this paper attempt to address?

The main problem that this paper attempts to solve is to explore how the cluster synchronization phenomenon is determined by nonlinear effects in high - order networks (i.e., hyper - networks). Specifically:

1. **Dynamical systems under high - order interaction structures**: The paper first defines a class of admissible mappings for dynamical systems applicable to hyper - networks and explains how to classify robust cluster synchronization patterns on hyper - networks by finding equilibrium partitions. These concepts are generalized from the concepts of traditional coupled - cell networks, admissible mappings and vector fields, graph fibrations, and quotient networks.

2. **High - order effects of cluster synchronization**: Different from traditional binary networks, in hyper - networks, cluster synchronization is determined by high - order terms (i.e., nonlinear terms) in the equations of motion. The paper provides an exact expression for determining the degree of polynomial admissible mappings of cluster synchronization patterns (Theorem 4.1) and proves that this degree is optimal through examples (Theorem 5.4).

3. **Differences in synchronization patterns**: The paper points out that in classical networks, cluster synchronization is mainly determined by linear terms in the equations of motion; while in hyper - networks, this synchronization pattern is a pure nonlinear effect. This indicates that there are significant differences in some basic characteristics between hyper - network dynamical systems and classical network dynamical systems.

4. **New bifurcation phenomena**: The paper also shows the significant synchronization - breaking bifurcation phenomena that may be caused by this high - order effect, which occur at high polynomial degrees. The author shows this through numerical simulations in Section 5 and indicates that a systematic analysis and study of this type of bifurcation will be carried out in subsequent papers.

Overall, this paper aims to reveal the essence of cluster synchronization as a high - order nonlinear effect and explore its potential impact in practical applications by introducing and analyzing a new framework for hyper - network dynamical systems.