Laplacian Flows in Complex-valued Directed Networks: Analysis, Design, and Consensus
Aditi Saxena, Twinkle Tripathy, Rajasekhar Anguluri · 2025
Read Laplacian Flows in Complex-valued Directed Networks: Analysis, Design, and Consensus by Aditi Saxena, Twinkle Tripathy, Rajasekhar Anguluri (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-2509-04196.txt.
Introduction
Aditi Saxena and two co-authors study agreement in networks whose links carry complex numbers in a 2025 research paper. Each connection weight then holds both a strength and a phase angle, like a push with a twist. Such models suit power distribution grids and other directed systems. The authors ask exactly when flow across the network settles into consensus, meaning all nodes end with the same value, for both tightly and loosely linked directed graphs.
One extra condition, called real dominance, turns out to decide convergence, and it depends on the links' phase angles. Spectral properties of the network's Laplacian matrix explain why. When agreement fails naturally, modified flows are offered that force it. Tests cover invented and real networks, plus diffusion as the mirror process.
Catalog record
| CREATOR | Aditi Saxena, Twinkle Tripathy, Rajasekhar Anguluri |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 64 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2509-04196
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
Laplacian Flows in Complex-valued Directed Networks: Analysis, Design, and Consensus Aditi Saxena, Twinkle Tripathy, Rajasekhar Anguluri Abstract In the interdisciplinary field of network science, a complex-valued network, with edges assigned complex weights, provides a more nuanced representation of relationships by capturing both the magnitude and phase of interactions. Additionally, an important application of this setting arises in distribution power grids. Motivated by the richer framework, we study the necessary and sufficient conditions for achieving consensus in both strongly and weakly connected digraphs. The paper establishes that complex-valued Laplacian flows converge to consensus subject to an additional constraint termed as ‘real dominance’ which relies on the phase angles of the edge weights. Our approach builds on the complex Perron-Frobenius properties to study the spectral properties of the Laplacian and its relation to graphical conditions. Finally, we propose modified flows that guarantee consensus even if the original network does not converge to consensus. Additionally, we explore diffusion in complex-valued networks as a dual process of consensus and simulate our results on synthetic and real-world networks. I Introduction A Laplacian flow describes the state evolution of a vector-valued continuous linear time-invariant homogeneous system, where the state matrix is the negative of a Laplacian associated with an underlying finite-vertex network. The choice of Laplacian is governed by applications including the human brain network; power, water, and gas networks; societal groups; robotics; and transportation. In the literature, important results have been established on network flows based on network and matrix-theoretic conditions to achieve properties, including controllability, observability, asymptotic convergence, and importantly consensus . Nonetheless, the edge-weights in a network are real-valued in many studies. We depart from this custom by studying networks with complex-valued edges; hereafter, complex-valued networks . The central mathematical object in the study of flows is the map e^{-Lt} , where t\geq 0 and L\in\mathbb{C}^{n\times n} is the Laplacian (Section II ). For real-valued, non-negative Laplacians, many properties of this map follow from standard results, including the Perron-Frobenius theory for dominant eigenvalues and the positive semi-definiteness of L . One important property is the eventual exponential positivity (or non-negativity) of e^{-Lt} (see [ 1 , 2 ] ) that allows to study the steady-state behavior of flows using the network reachability and eigen-spectrum of L\in\mathbb{R}^{n\times n} . The desirable properties of real-valued digraphs do not carry over in a straightforward manner to complex-valued Laplacian networks. For example, complex-symmetric Laplacians L=L^{\top}\in\mathbb{C}^{n\times n} may have eigenvalues that are not real or positive (see Ex. ( 1 )), and e^{-Lt} may not exhibit eventual exponential behavior. Thus, classical reachability fails to capture the null-space of L corresponding to its zero eigenvalue, unlike in the real-valued setting. These intriguing properties are intrinsic to complex-valued dynamics and are absent in networks with non-negative real weights. The goal of this work is to formalize these properties and develop suitable generalizations of eventual positivity to the complex domain, enabling analogs of known results for real-valued Laplacians (see [ 3 , 1 ] ). Motivation : Our interest in complex-valued networks is due to two applications: (i) distribution power grids and (ii) opinion dynamics in social networks. Operations in power grids, such as static state estimation, stability, and control, depend on the complex-valued bus admittance matrix. Traditionally, only its imaginary part (which is real-valued) has been used to simplify analysis. However, recent research has highlighted the need to consider the complex-valued admittance matrix in distribution networks for network reconstruction and power-flow problems [ 4 ] . It is worth acknowledging the extensive work on the mathematical theory of multi-port electrical circuits and network synthesis from the past century [ 5 , 6 , 7 ] , where the admittance matrix can have complex values or polynomials. In social networks, specifically concerning opinion dynamics, complex-valued edges allow agents (nodes) to be described beyond simple binary relations like friendship or foes. These relationships can be achieved by expressing edge-weights in the polar form ( re^{\iota\beta} ), where the phase angle \beta\in(-90^{\circ},90^{\circ}) captures the nature and strength of the relationship [ 8 ] . Finally, complex-valued networks are employed to study frustration in digraphs [ 9 ] . Other areas where complex-valued networks play a key role include quantum computation [ 10 , 11 , 12 , 13 ] ; electrical circuits [ 14 ] ; complex-valued neural networks [ 15 ] ; and communication networks [ 16 ] . In image processing, complex values provide an extra rotational information of the visual data [ 17 ] . In signal networks, complex-valued edge-weights encode magnitude and phase transfer characteristics of interferometer networks [ 18 ] . Finally, see [ 19 ] for a survey covering various aspects of complex-valued networks for network science. Literature Review : Consensus problems have been the mainstay in the literature of multi-agent networks [ 20 , 3 , 21 , 22 ] . Applications range from quantum, power, and brain networks to coordination in autonomous vehicles [ 3 , 23 , 2 , 24 ] . At the heart of consensus problem is Laplacian flows. Ref. [ 3 ] and the subsequent work set the stage for studying Laplacian flows by highlighting that non-negative matrices (e.g., M-matrix; Perron and stochastic matrices) arise in modeling both large-scale engineering and communication systems. Further, [ 1 ] provided normality as a sufficient condition for achieving consensus in signed digraphs using the Perron-Frobenius properties of real-valued matrices. Studies on complex-valued networks for analyzing network dynamics, including consensus, is limited within the broader field of network science and virtually absent in control theory. A few exceptions exist. Ref. [ 25 ] studies Hermitian Laplacians associated with complex unit gain graphs and derives spectral bounds. Ref. [ 8 ] extended structural notions such as balance, antibalance, and unbalance to the Hermitian setting, along with random walk dynamics and a spectral clustering method. In Ref. [ 26 ] , complex edge weights were used in a leader–follower framework to model planar formations via rotated lines of sight. Refs. [ 27 ] and [ 28 ] explored singularity conditions for non-Hermitian Laplacians using a generalized notion of balance, where a cycle is positive if the product of its edge weights is real and positive [ 29 ] . Most existing studies focus on undirected or Hermitian networks. Here, we relax those assumptions and consider general non-Hermitian digraphs. While prior works have addressed such cases, they often rely on balance assumptions. In contrast, our analysis applies to both balanced and unbalanced complex-valued networks. This enables us to investigate broader questions: Can non-trivial consensus emerge in unbalanced complex digraphs? Do the conditions align with the real-valued case? And if not, what accounts for the differences? I-A Contributions This research builds on our preliminary studies on complex-valued Laplacian and pseudoinverse flows [ 30 , 31 ] in undirected and weight-balanced digraphs 1 1 1 A digraph is weight-balanced if each node’s in-degree equals its out-degree. , having non-Hermitian Laplacians. For these digraphs, the eigenvector corresponding to the zero eigenvalue being both real and positive analogous to the case of the real-valued Laplacian, allowed us to develop sufficient conditions for consensus. We build upon this prior work to accommodate non weight-balanced digraphs. Moreover, we relax the signed edge weight conditions