Spectral fluctuations and crossovers in multilayer network

Himanshu Shekhar 1* , Ashutosh Dheer 1 , Santosh Kumar 1 , N. Sukumar 2 · 2025

Read Spectral fluctuations and crossovers in multilayer network by Himanshu Shekhar 1* , Ashutosh Dheer 1 , Santosh Kumar 1 , N. Sukumar 2 (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2508-12913.txt.

Introduction

Himanshu Shekhar, Ashutosh Dheer, Santosh Kumar, and Sukumar study eigenvalue patterns in layered networks with random matrix tools, in a paper from 2025. Layered networks, like social plus contact layers, have block-structured connection tables. How their spectra shift as layers couple was not fully mapped.

After scaling blocks to equal variance, they compare spectra across coupling strengths. A crossover model for two layers captures the smooth move from two independent spectra to one merged spectrum. They also test the frame on protein atom networks from crystal structures. Universal fluctuation patterns persist across layouts, which supports random matrix tools for real networks.

Catalog record

CREATORHimanshu Shekhar 1* , Ashutosh Dheer 1 , Santosh Kumar 1 , N. Sukumar 2
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE57 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2508-12913

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Spectral fluctuations and crossovers in multilayer network Preprint: APS/123-QED Himanshu Shekhar 1* , Ashutosh Dheer 1 , Santosh Kumar 1 , N. Sukumar 2 Abstract We investigate spectral fluctuations in multilayer networks within the random matrix theory (RMT) framework to characterize universal and non-universal features. The adjacency matrix of a multilayer network exhibits a block structure, with diagonal blocks representing intra-layer connections and off-diagonal blocks encoding inter-layer connections. Applying appropriate scaling factors for these blocks, we equalize variances across inter- and intra-layers, enabling direct comparison of spectral statistics. We analyze eigenvalue spectra across multilayer network configurations with varying inter- and intra-layer connectivities. Introducing a crossover model for bilayer networks, we capture the smooth transition of spectral properties from block-diagonal (two independent GOEs) to single-layer (one GOE) statistics as the relative strength of inter-layer to intra-layer connection varies. Furthermore, we analyze interatomic distance networks derived from protein crystal structures, including 1EWT, 1EWK, and 1UW6, to demonstrate applicability. Our findings reveal that the universality of spectral fluctuations persists across multilayer network architectures and highlight RMT as a robust tool for probing topological and dynamical complexities of real-world networks. I Introduction Complex systems permeate the natural and engineered world, from neural circuits and social dynamics to infrastructural grids and biological assemblies AL2002 ; MN2018 ; SH2001 ; SV2006 ; EE2012 . Understanding such systems requires modeling the interactions among their constituent elements, often using tools from network science. At its core, a network is an abstraction: nodes represent entities, and edges represent interactions or relationships between these entities. This abstraction allows a systematic comparison of otherwise disparate systems. This formalization underpins graph theory, which originated with Euler’s solution Euler1741 to the Seven Bridges of Königsberg problem in 1736. Over time, graph theory has evolved into a powerful mathematical and computational framework for analyzing connectivity, structure, and dynamics in complex systems, driven by applications in sociology, epidemiology, computer science, physics, economics, and molecular biology WF1994 ; Larson2021 ; LH2007 ; BGL2011 ; JDB2017 ; ETYD2018 ; RKD2000 ; BJN2002 ; Barthelemy2011 . Foundational models such as the Erdős-Rényi (ER) random graph, the Watts-Strogatz small-world network, and the Barabási-Albert scale-free network have provided benchmarks for understanding how topology influences robustness, diffusion, synchronization, and percolation ER1960 ; Bollobas1998 ; Gilbert1959 ; Fienberg2012 ; WS1998 ; AB2002 . Mathematically, the structure of a network is encoded in its adjacency matrix, a square matrix whose entries specify the presence or weight of edges. For undirected simple graphs, the adjacency matrix is a symmetric matrix with eigenvalues that are real-valued. Spectral graph theory relates these eigenvalues and their eigenvectors to network properties such as connectivity, community structure, and diffusion dynamics. Beyond such structural measures, the statistical fluctuations of eigenvalues reveal the system’s degree of order or randomness. In this context, random matrix theory (RMT) becomes remarkably useful as a statistical framework for characterizing spectral fluctuations in complex networks. RMT posits that large, complex systems exhibit universal spectral fluctuation statistics determined only by symmetry class, regardless of microscopic details. For adjacency matrices of large random graphs, the nearest-neighbor spacing distribution (NNSD) of eigenvalues follows the Wigner-Dyson distribution BJ2007 ; JB2007 characteristic of the Gaussian ensembles in RMT. The spacing ratio distribution (SRD), which avoids spectral unfolding, captures the same universal features and is particularly robust for empirical and finite-size networks. A vast number of real-world applications of network analysis through RMT have been documented across diverse domains AA2014 ; MSN2022 ; AWJ2022 ; RL2016 . However, many real-world systems are inherently multilayered, where nodes interact through multiple types of connections or participate in various contexts simultaneously. Examples include multimodal transportation systems, multiplex social networks, interconnected infrastructure grids, and biological systems such as interatomic distance networks in proteins. Modeling such systems as single-layer networks can obscure essential structural correlations, differences in layer structure, and the interplay between intra-layer and inter-layer connectivity. In the multilayer network formalism KABG2014 ; BBCG2014 ; DSCK2013 , the adjacency matrix acquires a block structure: diagonal blocks encode intra-layer connections, and off-diagonal blocks encode inter-layer connections. This richer representation allows a more faithful description of real systems but also raises new questions: How do spectral fluctuations behave in multilayered networks? Can they display features akin to the universal signatures predicted by RMT? Spectral analysis of multilayer networks has begun to attract attention CARM2016 ; MCMA2016 ; RJ2022 ; RJ2023 ; GM2025 , particularly in multiplex configurations where the same set of nodes is present in each layer. Yet, a systematic exploration of universality in general multilayer architectures—those with differing sizes, connection probabilities, and intra-/inter-layer structures—remains limited. Moreover, recovering RMT universality in such systems requires careful treatment: without appropriate scaling, block variances may differ substantially, leading to spurious deviations from universal statistics. In this work, we develop a general RMT-based framework for spectral fluctuation analysis in multilayer networks. We treat the most general architectures, including networks with only intra-layer connections, only inter-layer connections, both types, and multiplex structures. To ensure comparability of spectral fluctuations across layers, we introduce scaling factors for diagonal and off-diagonal blocks that normalize variances irrespective of block size or connection probability. Fluctuations are characterized using higher-order spacing ratios, which are sensitive to spectral correlations and avoid the need for unfolding. Our analysis proceeds in two parts. First, we construct ensembles of random multilayer networks with ER connectivity and systematically vary intra-layer and inter-layer connection probabilities to explore the dependence of spectral statistics on connectivity patterns. This is then generalized by introducing a crossover model for bilayer networks, in which the relative strength of intra-layer and inter-layer couplings is tuned to track the transition from two independent GOEs to a single GOE. Second, we apply the framework to empirical multilayer networks derived from protein crystal structures (1EWT, 1EWK, and 1UW6), modeling residues as nodes and spatial proximity as edges. By varying distance thresholds, we probe how changes in intra- and inter-layer connectivity drive transitions between universality classes, using both SRD and cumulative SRD (CSRD) to quantify the behavior. We find that with appropriate scaling, multilayer networks exhibit the universal spectral fluctuations predicted by RMT, even in the presence of structurally distinct layers and varying topologies. These results establish a general framework for quantifying spectral universality and inter-layer coupling in layered systems, with applications from communication and transportation networks to complex biological assemblies. The paper is organized as follows. In section II we define the multilayer network model, its adjacency matrix representation, and the scaling procedure required to recover universal spectral fluctuations. Section III presents the spectral statistics of random multilayer networks for different architectures, including the bilayer crossover model. In section IV we apply the framework to empirical multilayer networks derived from protein crystal structures, and summarize the results and discuss possible extensions in section V . II Preliminaries II.1 Multilayer Network and General Matrix Model A multilayer network can be defined as a function \Phi\equiv\Phi(G,E) of two sets: a set of graphs G and a set E of connections. The set G=\{g_{a}|a=1,2,\ldots,m\} is the set of graphs in each layer, where each graph g_{a}\equiv g_{a}(N_{a},e_{a}) is defined by the set of nodes N_{a} and the set of edges e_{a} . The set E=\{e_{ab};a,b\in\{1,2,\ldots,m\}\} contains connections between nodes; elements of E could denote intra-layer connections

Read the complete text