Distributed Saddle-Point Dynamics in Multilayer Networks

Christian D. Rodríguez-Camargo, Andrés F. Urquijo-Rodríguez, and Eduardo Mojica-Nava · 2025

Read Distributed Saddle-Point Dynamics in Multilayer Networks by Christian D. Rodríguez-Camargo, Andrés F. Urquijo-Rodríguez, and Eduardo Mojica-Nava (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-2501-11808.txt.

Introduction

Many real systems, from power grids to social platforms, form networks with several layers that interact. Christian Rodríguez-Camargo and co-authors extend distributed optimization methods from simpler multiplex networks to general multilayer ones. Their tool is a tensor formalism, which writes the whole layered network and its diffusion in one mathematical object.

Unlike multiplex setups, their framework does not force each layer to mirror the same nodes, so layers can differ in size and wiring. They derive a multilayer version of the Laplacian and extend a distributed gradient method on top of it. Tests on sample networks confirm the method converges. The team also explores how uneven layer shapes and cross-layer links change the time a network needs to agree.

Catalog record

CREATORChristian D. Rodríguez-Camargo, Andrés F. Urquijo-Rodríguez, and Eduardo Mojica-Nava
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE55 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2501-11808

Browse the full library · About the Tower of Babel

Full text

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

Distributed Saddle-Point Dynamics in Multilayer Networks Christian D. Rodríguez-Camargo, Andrés F. Urquijo-Rodríguez, and Eduardo Mojica-Nava Abstract Multilayer networks provide a more advanced and comprehensive framework for modeling real-world systems compared to traditional single-layer and multiplex networks. Unlike single-layer models, multilayer networks have multiple interacting layers, each with unique topological features. In this paper, we generalize previously developed results for distributed optimization in multiplex networks to the more general case of multilayer networks by employing a tensor formalism to represent multilayer networks and their tensor-Laplacian diffusion dynamics. Although multiplex networks are a special case of multilayer networks, where each layer has the same number of replica nodes connected one-to-one, this generalized framework removes the need for replica nodes, allowing variability in both topology and number of nodes across layers. This approach provides a fully generalized structure for distributed optimization in multilayer networks and enables more complex interlayer connections. We derive the multilayer combinatorial Laplacian tensor and extend the distributed gradient descent algorithm. We provide a theoretical analysis of the convergence of algorithms. Numerical examples validate our approach, and we explore the impact of heterogeneous layer topologies and complex interlayer dynamics on consensus time, underscoring their implications for real-world multilayer systems. I Introduction The quantitative study of networks has provided a deep understanding of the structure and dynamics of complex systems in various fields [ 1 , 2 ] , such as biology [ 3 ] , neuroscience [ 4 ] , engineering [ 5 , 6 ] , and social sciences [ 7 ] . However, the traditional single-layer network formalism does not capture the full complexity of real-world systems where multiple types of interaction occur simultaneously across distinct layers [ 8 ] . Multilayer networks provide a more holistic framework for representing these interconnected systems, allowing researchers to explore a broader range of dynamics and interactions that are not apparent in single-layer or multiplex network models [ 9 ] . The multilayer network framework has been successfully applied in various fields. For instance, this structure has produced significant and robust results in ecological dynamics [ 10 ] , brain networks [ 11 ] , transport networks [ 12 ] , epidemiology [ 13 ] , molecular biology [ 14 ] , and energy networks [ 15 , 16 ] . Furthermore, multilayer networks reveal dynamical behaviors similar to those of single-layer networks, such as synchronization [ 17 ] , percolation [ 18 ] , and diffusion and spreading processes [ 19 ] . Given the multidimensional nature of these processes, it is worth establishing a tensorial representation for such systems. Tensors generalize scalars, vectors, and matrices [ 20 ] and provide a concise framework for modeling and solving multidimensional problems. Tensors are widely used in all fields, from linear algebra and physics to engineering [ 21 , 22 ] . In [ 23 ] it is developed a tensor-based mathematical formalism that provides a general, robust, and comprehensive framework to represent multilayer networks with all their complexities, offering insights that remain hidden when using traditional adjacency matrices. This formalism has been applied explicitly to study disease localization [ 24 ] , generalization of the concept of assortativity [ 25 ] , spreading [ 26 ] , and eigenvector multicentrality [ 27 ] . Despite recent advances in multilayer networks, progress in generalizing distributed optimization problems from single-layer to multilayer networks has been limited. In conventional distributed optimization, each node typically has access to its own convex function, and the overall objective is defined as the sum of these individual functions [ 28 ] . To solve this, each agent seeks the optimal solution to the global convex function by exchanging information with neighboring nodes and performing local computations [ 29 ] . Various methods have been developed to address this problem, ranging from discrete-time gradient descent approaches [ 29 ] to more recent continuous-time techniques [ 30 , 31 , 32 ] . In particular, from a continuous-time perspective, optimization algorithms have been framed as dynamical systems [ 33 ] , with methods based on saddle-point dynamics [ 34 , 35 ] gaining prominence. I-A Related Work A multilayer network is formed by several interacting networks [ 36 ] , [ 9 ] . The first introduction of multilayer networks is in the context of social science to characterize the different types of social relationships that exist between the nodes of a social network [ 37 ] . Currently, multilayer networks are investigated in many fields, including diffusion dynamics and information spreading [ 38 ] , [ 39 ] , [ 40 ] , epidemic spreading [ 41 ] , neuroscience [ 42 ] , molecular biology [ 43 ] , ecology [ 10 ] , and economy [ 44 ] . Multilayer networks have also been applied effectively across various engineering domains to model and analyze complex systems with multiple interacting components such as power networks, where the interdependence between power grids and communication infrastructures is effectively captured using multilayer network models [ 16 ] . One layer represents the electrical grid, while another represents the communication network that monitors and controls the grid’s operations. This modeling enables the analysis of cascading failures [50], [51], where a disruption in the power grid can impact the communication network and vice versa, aiding in the design of more resilient interdependent infrastructures. Another successful application is urban transportation networks including roads, railways, buses, and metro systems, which can be represented as multilayer networks where each mode of transport forms a layer [ 45 ] . Information propagation is also a relevant application of multilayer networks [ 46 ] . Although there are significant advances in other engineering applications phenomena such as percolation [ 47 ] , there is a lack of developments in the generalization of single-layer network distributed optimization problems to multilayer networks. A specific type of multilayer network is multiplex networks, where there is a one-to-one correspondence between each node across all layers. In other words, the same set of nodes is connected by different types of interactions in each layer. A multiplex network is a set of graphs composed by different layers in which the same set of nodes may be connected to each other by means of links belonging to different classes or types [ 48 ] . Multiplex networks are initially introduced in the context of social networks, where individuals are linked through various social relationships such as friendship, collaboration, and family, or communicate through different channels such as email, mobile phones, and instant messaging [ 49 ] . Over time, the concept of multiplex networks has been extended to encompass a wide range of domains, including transportation systems [ 50 ] , and brain networks [ 51 ] . Multilayer networks represent a generalization of peer-to-peer architectures, where agents are embedded within single-layer communication networks. In peer-to-peer architectures, consensus protocols have garnered significant attention and research interest due to their applications in synchronization, formation control, learning, and the coordination of multi-robot tasks, among others. These systems are inherently distributed. Distributed computational models and coordination mechanisms are based on local communication between agents. Such models are crucial for advancing technologies like automated transportation, smart cities, smart power

Read the complete text