[Paper Review] Topological determinants of complex networks spectral properties: structural and dynamical effects

Claudio Castellano, Romualdo Pastor‐Satorras|arXiv (Cornell University)|Mar 30, 2017

Complex Network Analysis Techniques4 citations

TL;DR

This paper proposes a physically grounded analytical formula that predicts the largest eigenvalue of a network's adjacency matrix by decomposing the network into two key subgraphs: the hub with its immediate neighbors and the highest K-core subgraph. The method achieves high-accuracy predictions across diverse synthetic and real-world networks, revealing fundamental topological determinants of spectral properties and highlighting qualitative differences in spectral behavior between growing (preferential attachment) and static network models.

ABSTRACT

The largest eigenvalue of a network's adjacency matrix and its associated principal eigenvector are key elements for determining the topological structure and the properties of dynamical processes mediated by it. We present a physically grounded expression relating the value of the largest eigenvalue of any network to the largest eigenvalue of two network subgraphs, considered as isolated: The hub with its immediate neighbors and the densely connected set of nodes with maximum $K$-core index. We validate this formula showing that it predicts with good accuracy the largest eigenvalue of a large set of synthetic and real-world topologies, with no exception. We also present evidence of the consequences of these findings for broad classes of dynamics taking place on the networks. As a byproduct, we reveal that the spectral properties of heterogeneous networks built according to the linear preferential attachment model are qualitatively different from those of their static counterparts.

Motivation & Objective

• To identify the topological determinants governing the largest eigenvalue of complex networks' adjacency matrices.
• To develop a physically grounded analytical expression linking the largest eigenvalue to two isolated subgraph components: the hub neighborhood and the maximal K-core.
• To validate the predictive accuracy of the formula across diverse synthetic and real-world network topologies.
• To explore the dynamical implications of the derived spectral relationships for processes on networks.
• To compare spectral properties of growing networks (preferential attachment) with their static counterparts.

Proposed method

• The method decomposes a network into two subgraphs: the immediate neighborhood of the highest-degree node (hub) and the subgraph with maximum K-core index.
• It derives a formula expressing the largest eigenvalue of the full network as a function of the largest eigenvalues of these two isolated subgraphs.
• The approach relies on spectral graph theory and physical intuition about network structure, particularly the dominance of hubs and dense cores.
• The formula is validated using extensive numerical experiments on synthetic networks (e.g., Erdős–Rényi, scale-free) and real-world networks (e.g., social, infrastructure, biological).
• Dynamical processes such as epidemic spreading and synchronization are analyzed using the derived spectral properties to assess their sensitivity to network topology.
• The method enables comparison between growing (linear preferential attachment) and static network models by analyzing their spectral differences.

Experimental results

Research questions

• RQ1 What topological components most strongly determine the largest eigenvalue of a network's adjacency matrix?
• RQ2 Can the largest eigenvalue be accurately predicted using only the largest eigenvalues of two isolated subgraphs: the hub neighborhood and the highest K-core?
• RQ3 How do the spectral properties of growing networks generated via linear preferential attachment compare to those of their static counterparts?
• RQ4 What are the dynamical consequences of the derived spectral relationships for processes such as epidemic spreading or synchronization on networks?
• RQ5 To what extent does the proposed formula generalize across diverse network topologies, including real-world systems?

Key findings

• The proposed formula predicts the largest eigenvalue of a wide range of synthetic and real-world networks with high accuracy, showing no exceptions in the tested set.
• The largest eigenvalue is predominantly determined by the spectral properties of the hub’s immediate neighborhood and the densest K-core subgraph, confirming their structural dominance.
• The spectral properties of growing networks based on linear preferential attachment are qualitatively different from those of their static counterparts, particularly in the distribution and scaling of the largest eigenvalue.
• The method reveals that the interplay between the hub and the K-core subgraph governs the network’s spectral radius, which in turn influences the stability and dynamics of processes on the network.
• The derived formula enables accurate estimation of the largest eigenvalue without full matrix diagonalization, offering a computationally efficient alternative for large-scale network analysis.
• The results demonstrate that topological features such as degree heterogeneity and core-periphery structure are fundamental to understanding network spectral behavior.