Cluster Synchronization via Graph Laplacian Eigenvectors (2503.18978v1)

Abstract: Almost equitable partitions (AEPs) have been linked to cluster synchronization in oscillatory systems, providing a mathematical framework for understanding collective behavior. Using the spectral properties of AEPs, we can describe this synchronization behavior in terms of the graph Laplacian eigenvectors. Our results also shed light on transient hierarchical clustering as well as multi frequency clustering, and the conditions under which they can occur. Through our analysis we are able to relate dynamical clustering phenomena directly to network symmetry and community detection, joining the structural and functional notions of clustering in complex networks. This bridges a crucial gap between static network topology and emergent dynamic behavior. Additionally, this description of the problem allows us to define a relaxation of an AEP we call a quasi-equitable partition (QEP). Perfect AEPs are rare in real-world networks since most have some degree of irregularity or noise. While relaxing these strict conditions, QEPs are able to maintain many of the useful properties allowing for qualitatively similar clustering behavior as with AEPs. Our findings have important implications for understanding synchronization patterns in real-world networks, from neural circuits to power grids.