Obtaining Structural Network Controllability with Higher-Order Local Dynamics

Marco Peruzzo, Giacomo Baggio, Francesco Ticozzi · 2025

Read Obtaining Structural Network Controllability with Higher-Order Local Dynamics by Marco Peruzzo, Giacomo Baggio, Francesco Ticozzi (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-2504-17417.txt.

Introduction

Marco Peruzzo, Giacomo Baggio, and Francesco Ticozzi wrote a 2025 control theory paper on steering large networks. Their network is made of many small linked units, each with simple first-order behavior. A key question is how few units must change for the whole network to become fully steerable.

The authors swap some units for higher-order ones with richer inner dynamics. They show this can cut the number of changed units by a lot. They also link control of the state in one setup to control of the output in another. Last, they name a class of networks where no mixed units are needed at all. Such networks reach full control with matched higher-order parts.

Catalog record

CREATORMarco Peruzzo, Giacomo Baggio, Francesco Ticozzi
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE72 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2504-17417

Browse the full library · About the Tower of Babel

Full text

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

Obtaining Structural Network Controllability with Higher-Order Local Dynamics Marco Peruzzo, Giacomo Baggio, Francesco Ticozzi Abstract We consider a network of identical, first-order linear systems, and investigate how replacing a subset of the systems composing the network with higher-order ones, either taken to be generic or specifically designed, may affect its controllability. After establishing a correspondence between state controllability in networks of first-order systems with output controllability in networks of higher-order systems, we show that adding higher-order dynamics may require significantly fewer subsystem modifications to achieve structural controllability, when compared to first-order heterogeneous subsystems. Furthermore, we characterize the topology of networks (which we call X-networks) in which the introduction of heterogeneous local dynamics is not necessary for structural output controllability, as the latter can be attained by suitable higher-order subsystems with homogeneous internal dynamics. I Introduction Many natural and engineered dynamical systems can be modeled as networks consisting of a large number of interconnected, simpler dynamical units [ 1 , 2 , 3 ] . In such systems, the interconnection structure significantly influences the collective behavior of the system and the response to external inputs. As a result, understanding how network topology affects dynamics is fundamental to the analysis and control of these systems. Over the past few decades, the analysis of complex network dynamics has gained increasing attention from different scientific communities, ranging from physics and engineering [ 4 , 5 ] to biology and social sciences [ 6 , 7 ] . This growing body of research has deepened our understanding of how the structural properties of networks impact collective phenomena in interconnected systems, with notable implications for synchronization [ 8 ] , information transfer [ 9 ] , and controllability [ 10 ] . In particular, the challenge of efficiently controlling large-scale networks has driven renewed interest in the structural controllability analysis of networks governed by linear time-invariant dynamics [ 11 ] . Structural controllability is a relaxed notion of controllability that depends only on the sparsity patterns of system matrices, rather than their specific numerical values [ 12 ] . For network systems, it highlights the role of the interconnection structure in determining whether the system state can be steered to any desired configuration via appropriate inputs, regardless of the precise values of the edge weights. Although structural controllability has been extensively studied in control theory [ 13 , 14 , 15 ] , recent works have opened up new areas of investigation. These include: developing strategies to efficiently achieve structural controllability [ 11 ] ; understanding how key network properties, such as degree distribution and modularity, influence structural controllability [ 16 , 17 ] ; and analyzing structural controllability in time-varying settings [ 18 , 19 ] or under constraints on edge weights [ 20 , 21 ] . Regarding the first area, much of existing research has focused on developing efficient methods to identify a minimal set of driver nodes [ 11 , 22 , 23 ] , or to implement minimal changes to the network topology [ 24 , 25 , 26 ] to guarantee structural controllability. In this work, we explore an alternative strategy to achieve structural controllability, namely through the introduction of higher-order local dynamics in a subset of network nodes. 1 1 1 We stress that, in this paper, the term ‘higher-order’ refers to the number of state variables describing the local dynamics of network nodes, not to the nature of interactions between nodes as in [ 27 ] . More precisely, we consider networks composed of identical first-order systems and replace the internal dynamics of certain nodes with higher-order ones, while preserving the original configuration of interconnections between nodes. We assume that a generic linear combination of the internal states of each modified node is measured, and we analyze the structural output controllability [ 28 ] of the resulting network system. First, we characterize the topologies of networks which are structurally uncontrollable and can be made structurally (output) controllable by replacing the internal dynamics of some nodes with homogeneous higher-order ones. The existence of such systems, which we name X-networks, already clearly demonstrates the potential of higher-order systems, as heterogeneous local dynamics is needed to achieve structural controllability using first-order systems [ 29 ] . Second, we identify a class of networks for which higher-order homogeneous dynamics cannot lead to structural output controllability, referred to as Y-networks. We then show that, for Y-networks, using heterogeneous higher-order local dynamics provides an advantage in the minimal number of modified nodes, compared to employing first-order heterogeneous dynamics. Combining the results, we have that structural output controllability can be attained by introducing a reduced number of homogeneous and of heterogeneous higher-order dynamics. As a by-product of our analysis, we establish alternative graphical characterizations of structurally uncontrollable network topologies and present conditions for output controllability of linear systems based on the PBH test, which complement the results in [ 30 ] . Finally, we mention that a closely related work to ours is [ 31 ] , where the author provides conditions for the structural controllability of interconnected subsystems with potentially higher-order internal dynamics. However, in [ 31 ] the subsystems are assumed to be single-input single-output, a condition we do not impose in our work. Another related work is [ 32 ] , which considers multi-input multi-output higher-order subsystems, provides conditions for the structural controllability of the network, and discusses how to improve controllability via topology design. We emphasize that, in our work, the network topology remains fixed, and the designer’s degrees of freedom are limited to selecting which first-order subsystems to replace with higher-order ones. Furthermore and more importantly, both works [ 31 , 32 ] do not associate a (scalar) output equation with the nodes of a subsystem, as we do here, and focus on the structural (state) controllability of the global network. The rest of the paper is organized as follows. In Sec. II , we introduce the models of both the original and extended network systems, the latter incorporating higher-order nodes, and formally define the problem addressed in this study. Sec. III presents some instrumental results on the output structural controllability of the extended system. The main findings of the paper, namely the characterization of the controllability advantages provided by higher-order dynamics, are discussed in Sec. IV . Finally, in Sec. V we present some concluding remarks and outline potential directions for future research. Notation. We denote with \mathbb{R}^{n\times m} the set of n\times m matrices with real entries. We let A^{\top} denote the transpose of a matrix A\in\mathbb{R}^{n\times m} and \ker(A) be its kernel. The symbol I_{n} stands for the identity matrix in \mathbb{R}^{n\times n} , we will drop the subscript n when the dimension is clear from the context. Given a set of vectors \{w_{i}\} in \mathbb{R}^{n} , the notation \operatorname{span}\{w_{1},\dots,w_{n}\} represents their linear span. Finally, we use calligraphic letters to denote sets, and for a set \mathcal{Z} , the symbol |\mathcal{Z}| indicate its cardinality. II Models and problem II-A Networks of identical first-order systems We consider a network system composed of n interconnected subsystems with identical (homogeneous) first-order internal dynamics. The interconnections among these subsystems are represented by a directed graph {\rm G}_{0}=(\mathcal{V},{\rm\mathcal{E}}_{v}) , where \mathcal{V} and

Read the complete text