Pith. sign in

REVIEW 4 major objections 5 minor 60 references

Generalized Master Stability of Heterogeneous Delay-Coupled Networks

T0 review · 4 major / 5 minor · reviewed 2026-08-14 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read Synchronization stability of delay-coupled networks reduces to one low-dimensional master stability function over adjacency eigenvalues.

desk verdict Genuinely useful MSF extension for delayed heterogeneous directed networks, but the headline claim as stated is too strong; the exact reduction to a single d* MSF is proven only under restrictive conditions and is false for the LK model. read the letter →

arxiv 2608.10076 v1 pith:HST7J6JD submitted 2026-08-10 nlin.AO cond-mat.dis-nncond-mat.stat-mech

Ana Elisa D. Barioni , Arthur N. Montanari , Adilson E. Motter This is my paper · ORCID
classification nlin.AOcond-mat.dis-nncond-mat.stat-mech MSC 05C8234D0634K20 PACS 05.45.Xt42.55.Px89.75.-k
keywords masterstabilityfunctiondelay-couplednetworkssynchronizationheterogeneousdirectednetworkoptimizationStuart-LandauoscillatorsLang-Kobayashilasers
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

The paper extends the master stability function (MSF) framework, previously limited to degree-homogeneous delay-coupled networks, to weighted, directed, and degree-heterogeneous networks. Its central claim is that synchronization stability of the full M n-dimensional delay-coupled system is decided by a low-dimensional MSF, evaluated at the complex eigenvalues of the adjacency matrix, provided the network satisfies a simultaneous triangularization condition. On this basis the paper shows that time delays shrink the MSF stability region, so degree-homogeneous structures such as all-to-all networks can lose stability and become suboptimal. An optimization algorithm then produces master-slave, nonreciprocal, heterogeneous networks whose eigenvalues sit inside the stability region and can synchronize more stably than homogeneous ones, including in a small-delay regime where delayed coupling outperforms its non-delayed counterpart. The framework is demonstrated on Stuart-Landau oscillators, representing systems near a Hopf bifurcation, and on coupled semiconductor lasers.

What carries the argument

The central object is the generalized master stability function for delay-coupled systems, built on a simultaneous Jordanization-and-triangularization matrix $P$: a similarity transformation that sends the adjacency matrix $A$ to its Jordan form and the diagonal indegree matrix $\Delta$ to a lower-triangular matrix. Acting on the variational equation, it yields a staircase system of $M$ $n$-dimensional mode equations in which each mode depends only on earlier modes; if every mode subsystem $\Xi_j$ has negative maximum transverse Lyapunov exponent, the synchronous state is stable. The paper also uses the characteristic equation $\det(J_1 + J_2 e^{-\lambda_\ell \tau} - \lambda_\ell I_{Mn}) = 0$ for linear delay systems and, under small-delay and identity-coupling conditions, proves that the stability region $\Omega_j$ expands with the diagonal entry $\tilde{\Delta}_{jj}$, reducing the analysis to the smallest indegree $d_*$. The MSF maps each complex eigenvalue $\nu$ of $A$ to the maximum transverse Lyapunov exponent, turning network design into a spectral placement problem with an indegree constraint.

What would settle it

Take one of the optimized Stuart-Landau networks at $\tau=10$ (or a Lang-Kobayashi network at large coupling), compute the residual $P^{-1}\Delta P$ after Jordanizing $A$, then compare the MSF-predicted maximum transverse Lyapunov exponent against a direct numerical solution of the full $Mn$-dimensional variational delay equation; a systematic discrepancy growing with the residual norm would falsify the reduction. A cheaper test: construct a directed network with identical indegrees plus a small strictly upper-triangular perturbation of $\Delta$ and check whether the mode-decoupling prediction holds to first order.

Watch

Extended reading notes

Core claim

The paper's central conclusion is that the synchronization stability of the delay-coupled system is entirely determined by the MSF function $\lambda_{\max}(J'_1, J'_2(\nu))$, which maps the complex eigenvalue $\nu$ of the adjacency matrix $A$ to the maximum transverse Lyapunov exponent computed from the $n\times n$ matrices $J'_1 = D^{(0)}f + d_* D^{(0)}h$ and $J'_2 = \nu D^{(\tau)}h$. When a similarity matrix $P$ simultaneously puts $A$ in Jordan form and the indegree matrix $\Delta$ in lower-triangular form, the variational equation splits into staircase-coupled modes; stability of the full system follows by induction from the stability of the $n$-dimensional mode subsystems. The paper proves this reduction for identical-indegree and master-slave networks and verifies it approximately for optimized heterogeneous networks. Using this reduction, it shows that increasing delay shrinks the stability region $\Omega$, so homogeneous networks with a single longitudinal eigenvalue $\nu_0 = d$ eventually lose synchronization; optimized networks instead place their whole eigenvalue spectrum inside $\Omega$ and are directed, weighted, and heterogeneous, typically master-slave. For small delays, the delayed MSF can be deeper than the non-delayed one in the Lang-Kobayashi laser model, so delaying coupling can strengthen synchronization rather than only weaken it.

Load-bearing premise

The whole reduction rests on the existence of a similarity matrix that simultaneously puts the adjacency matrix in Jordan form and the indegree matrix in lower-triangular form; this is proven for identical-indegree and master-slave networks but only verified approximately for the optimized networks, where the leftover upper-triangular piece has Frobenius norm up to $2.7\times 10^{-2}$.

Editorial extensions

If this is right

  • Stability analysis of a delay-coupled network can be performed by computing the MSF over the complex plane and checking that all eigenvalues of the adjacency matrix lie inside the stability region, reducing an $Mn$-dimensional problem to $n$ dimensions.
  • Degree-homogeneous networks, including all-to-all coupling, become increasingly suboptimal as delay grows; their longitudinal eigenvalue can leave the stability region even when transverse modes remain stable.
  • Deliberately directed, weighted, heterogeneous (master-slave-like) topologies can synchronize more stably than homogeneous ones, and in small-delay laser arrays can outperform the optimal non-delayed configuration.
  • For systems near a Hopf bifurcation, represented by Stuart-Landau oscillators, larger minimal indegree expands the stability region, while delay shrinks it.
  • Network optimization guided by the MSF converges to hierarchical structures with high edge-weight variance, high maximum outdegree and eigenvector centrality, and low clustering.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • If the spectral-placement view is taken seriously, one can invert the logic and design networks by prescribing eigenvalue locations inside the stability region and then finding adjacency matrices with those spectra and indegree bounds; the paper shows existence only indirectly through optimization, so a constructive spectral-synthesis algorithm is a natural next step.
  • The master-slave optima suggest a testable engineering rule: in delayed multi-agent control and photonic arrays, adding a few dominant, nonreciprocal hub connections may improve synchronization tolerance to delay more than increasing overall coupling strength.
  • The Lang-Kobayashi result that small delays deepen the stability region implies delays can be exploited as a resource rather than only a nuisance; a concrete extension would be a delay-tuning experiment on semiconductor arrays sweeping $\tau$ through the predicted 0.05–0.27 ns window and measuring phase coherence.
  • Because the carrier variable in Lang-Kobayashi is uncoupled from other nodes, higher indegree shrinks the stability region; this predicts that any master-slave network with an uncoupled internal variable in its coupling function will invert the usual degree–stability monotonicity.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

4 major / 5 minor

Summary. This Letter proposes a master stability function (MSF) generalization for delay-coupled networks of identical oscillators with weighted, directed, and heterogeneous degree structure. The authors linearize the coupled system around an identical synchronous solution, introduce a similarity transformation P that simultaneously Jordanizes the adjacency matrix A and lower-triangularizes the indegree matrix Δ, and obtain a staircase system of n-dimensional variational modes (Eqs. (6)-(9)). They define per-subsystem stability regions Ω_j and, under an assumed monotonicity in the diagonal entries eΔ_jj, reduce the stability condition to a single MSF λmax(J'_1,J'_2(ν)) with J'_1 = D(0)f + d*D(0)h and J'_2 = νD(τ)h, where d* is the minimal indegree. They apply the framework to Stuart-Landau oscillators and Lang-Kobayashi laser arrays, compute MSF landscapes, and develop a constrained optimization algorithm that minimizes the maximum transverse Lyapunov exponent. The numerical results identify optimal networks with directed master-slave-like, heterogeneous structures and show that, for small delays, delayed coupling can produce a deeper stability region than the non-delayed case. Independent time-series simulations of the LK model are used to test stability predictions for all-to-all networks.

Significance. If the low-dimensional reduction were valid in the stated generality, the paper would be practically important: synchronization stability of a high-dimensional delay-coupled network would reduce to the spectral placement of the adjacency matrix eigenvalues in a low-dimensional stability region, enabling systematic network design. The special-case derivations for identical indegree and master-slave networks are correct, and the independent LK time-series validation that distinguishes predicted stable and unstable regimes is a genuine strength; the availability of code and data is also commendable. However, the central claim as stated is too broad: the paper itself shows that the monotonicity in indegree fails for the LK model, and the general heterogeneous case relies on an approximate simultaneous triangularization without an error bound. The conditional version of the result and a proper validation on optimized heterogeneous networks would make this a solid contribution; in its current form the central conclusion is not established.

major comments (4)
  1. [Main text, 'MSF generalization for delay-coupled systems', Eqs. (6)-(9) and central conclusion] The central conclusion that synchronization stability is 'entirely determined' by λmax(J'_1,J'_2(ν)) with J'_1 = D(0)f + d*D(0)h is not established for general heterogeneous networks. The reduction requires two conditions: an exact simultaneous similarity transformation putting A in Jordan form and Δ in lower-triangular form, and monotone inclusion of the per-subsystem stability regions Ω_j in the diagonal entries eΔ_jj. The paper proves the simultaneous transformation only for identical-indegree and master-slave networks; for the optimized networks in Figs. 2 and 4 the non-lower-triangular part of eΔ is nonzero as reported in SM Eqs. (S4)-(S5). The monotonicity is proven only in the limit τ→0 with D(τ)h = -D(0)h = I_n (SM Sec. SI), and the LK model in Fig. 3(b) displays the opposite monotonicity, with the stability region shrinking as d* increases. The text itself concedes that all subsystems must be analyzed in that case. The unconditional statement should therefore be replaced by a conditional statement: if the regions Ω_j are monotonically ordered in eΔ_jj, then d* suffices; otherwise the intersection Ω = ∩_j Ω_j must be checked for each subsystem Ξ_j.
  2. [Supplemental Material, Sec. SI, Eq. (S7)] The proof that Ω_j expands monotonically with eΔ_jj is incomplete. In Eq. (S7) the authors show that increasing eΔ_jj shifts μ leftward, but with their definition μ = λ_ℓ(1+ντ) the actual characteristic exponents are λ_ℓ = μ/(1+ντ). Therefore the real part of the shift is -δ Re(1/(1+ντ)) rather than -δ; for complex ν with Re(1/(1+ντ)) negative, the shift can be destabilizing. The conclusion that the entire spectrum is 'uniformly shifted leftward' does not follow from the displayed equation. In addition, the main text states that for large delays the monotonicity is 'observed numerically', but no large-delay monotonicity plot is presented; Fig. 3 and Fig. S2 only show small-delay comparisons. The authors should either provide a valid proof of the monotonicity condition, state precisely the conditions under which it holds, or base the general criterion on the intersection Ω = ∩_j Ω_j.
  3. [Supplemental Material, Sec. SI, 'Analysis for the optimal network structures'] The numerical evidence for simultaneous triangularization of the optimized heterogeneous networks is only approximate: the normalized Frobenius norm of the non-lower-triangular part of eΔ is up to 2.7×10^{-2} for the SL model at τ=10 and 2.5×10^{-2} for the LK model (SM Eqs. (S4)-(S5)). Because the mode-decoupling in Eq. (8) is exact only when this part vanishes, the manuscript needs a quantitative perturbation bound connecting this norm to the error in λmax or to a corrected stability condition. Without such a bound, the eigenvalue-placement plots in Fig. 2(c) do not prove that the optimized heterogeneous networks synchronize. The independent LK time-series tests in SM Sec. SIII use all-to-all networks, not the optimized structures of Fig. 4, and therefore do not close this gap.
  4. [Main text, Fig. 2(c) and 'Optimizing networks for synchronization'] The verification that the eigenvalues of optimized networks lie inside the stability region is partly self-referential. The stability region Ω is computed from the MSF using the minimal indegree d* of the same optimized network, and the optimization minimizes λmax evaluated through that same MSF. The agreement therefore confirms consistency of the optimization with its own objective function rather than providing independent confirmation that the full delay-coupled system synchronizes. An independent test for at least one optimized heterogeneous network, such as direct integration of Eq. (1), would substantially strengthen the claim; the current independent time-series evidence is limited to the all-to-all LK configurations in SM Sec. SIII.
minor comments (5)
  1. [Supplemental Material, Sec. SI] The phrase 'approximately transforms D into a lower triangular matrix' appears to refer to the indegree matrix Δ, not to the matrix D; the notation should be corrected.
  2. [Supplemental Material, Sec. SI and main text] The diagonal entries of the transformed indegree matrix are denoted eΔ_jj in the main text and eD_jj in the SM; please use one consistent notation throughout.
  3. [Supplemental Material, Sec. SI, Eqs. (S6)-(S7)] The Taylor expansion e^{λτ} ≈ 1 + λτ in the derivation of Eq. (S7) appears to produce a sign inconsistency in the definition of μ; please check whether μ = λ(1+ντ) or μ = λ(1-ντ) is intended.
  4. [Main text, 'MSF analysis for non-diffusive coupling'] The phrase 'stationary synchronous states of the form E_j(t) = r* e^{iΩ0 t}' is not literally stationary in the original time variable; please clarify that this is a state that is stationary in a rotating frame.
  5. [Fig. 2 caption] Please state the number of nodes M used for the optimized networks in Fig. 2(b), since the eigenvalue placement and the stability conclusions depend on M.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the MSF is derived from the variational equation rather than fitted; the unqualified d* claim is an overreach, not a circular reduction.

full rationale

The derivation chain is self-contained. Equation (2) is the exact linearization of system (1); Eqs. (5)-(9) follow from the stated simultaneous Jordanization/triangularization condition, which is proven for identical-indegree and master-slave networks in SM Sec. SI and numerically checked for optimized networks. The reduction to the minimal-indegree MSF is derived under the explicit small-delay, identity-coupling hypothesis in SM Eq. (S7), and the paper concedes that for the LK model, where D(τ)h is not the identity, all subsystems Ξ_j must be analyzed; this concession shows the unqualified central conclusion overreaches, but an overstated theorem is a correctness risk, not circularity. The LK time-series tests in Fig. S3 are independent predictions from the MSF stability region, not fits to those simulations. The optimized-network eigenvalue-in-region plots are consistency checks because the optimizer minimizes the full MTLE, but the paper does not present them as free predictions, and they do not feed back into the construction of the MSF. Self-citations (Refs. [6, 12, 26, 27, 31, 34, 52]) are background or motivation and are not load-bearing. No equation or fitted parameter reduces to the claimed output by construction.

Assumptions & free parameters 0 free parameters · 3 assumptions · 0 invented entities

The framework introduces no new physical entities. The central reduction relies on the existence of a simultaneous Jordanization and triangularization matrix P, and on monotonicity of the stability region in the indegree. The first is proven only for restricted classes and checked approximately for optimized networks; the second is proven only for small delay and identity Jacobian coupling. The LK example explicitly violates the monotonicity condition, limiting the smallest-indegree formulation.

assumptions (3)
  • domain assumption All nodes are identical and an identical synchronous state x*(t) exists, or is nearly identical with a small coefficient of variation.
    The variational equation (2) linearizes about a common trajectory x*. The SM relaxes this to near-identical states and reports epsilon_CV values; for the SL model at tau=10, epsilon_CV=3e-2, which is small but not negligible.
  • ad hoc to paper There exists a similarity matrix P that simultaneously brings A to Jordan form and Delta to lower triangular form.
    Equation (6) requires this transformation. The SM proves it for identical-indegree and master-slave networks, but for optimized general networks it is validated only approximately, with Frobenius norms of the non-lower-triangular part up to 2.7e-2.
  • domain assumption The stability region expands monotonically with the diagonal entries of the transformed degree matrix, proven only for tau -> 0 with D(tau) h = -D(0) h = I_n.
    This monotonicity is used to reduce the MSF to the smallest indegree d*. The SM proof uses a small-delay Taylor expansion and assumes identity Jacobian coupling; for the LK model the opposite trend is observed, so all subsystems must be analyzed.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Generalized Master Stability of Heterogeneous Delay-Coupled Networks." pith.science (2026). https://pith.science/paper/HST7J6JD

@misc{pith2026260810076,
  author       = {Pith},
  title        = {Pith review of: Generalized Master Stability of Heterogeneous Delay-Coupled Networks},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/HST7J6JD}},
  note         = {Machine review of arXiv:2608.10076}
}
read the original abstract

Time delays are ubiquitous in physical and biological networked systems, playing a fundamental role in the emergence and stability of collective behavior. Yet, existing theoretical methods for synchronization analysis of delay-coupled systems are largely limited to networks with homogeneous degree distributions. Here, we extend the master stability function framework to a broad class of degree-heterogeneous, weighted, and directed delay-coupled networks. The analysis reveals that synchronization can be enhanced in heterogeneous networks compared to homogeneous ones, including all-to-all networks, which are known to be optimal in non-delayed systems. To identify optimal heterogeneous structures, we develop a network optimization method that finds directional, edge-weighted configurations maximizing synchronization stability. Our results show that, in delay-coupled networks, heterogeneity and nonreciprocity are key resources for synchronization.

Figures

Figures reproduced from arXiv: 2608.10076 by the authors.

Figure 1
Figure 1. FIG. 1. Key extensions of the MSF framework to delay [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. FIG. 2. MSF analysis of delay-coupled SL oscillators. (a) MTLE of the synchronous state as a function of the coupling strength [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. FIG. 3. MSF comparison analysis for the SL and LK mod [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figures from the paper (1 more)
Figure 4
Figure 4. Figure 4: FIG. 4. Network optimization to enhance synchronizability. (a) Minimization of the MTLE of the LK model for different initial [PITH_FULL_IMAGE:figures/full_fig_p005_4.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

60 extracted references · 56 canonical work pages

  1. [1]

    the intrinsic dynamics of all nodes are identical, 2) the coupling term admits an invariant synchronization mani- fold, and 3) the coupling matrix is diagonalizable. These assumptions have since been relaxed to account for non- diagonalizable networks [26, 27], cluster synchronization [28], higher-order interactions [29–32], and non-identical oscillators ...

    work page Pith review arXiv 2026

  2. [2]

    P. A. Merolla, J. V. Arthur, R. Alvarez-Icaza, A. S. Cas- sidy, J. Sawada, F. Akopyan, B. L. Jackson, N. Imam, C. Guo, Y. Nakamura,et al., A million spiking-neuron integrated circuit with a scalable communication network and interface, Sci.345, 668 (2014)

    work page 2014

  3. [3]

    C. R. Stoelzel, Y. Bereshpolova, J.-M. Alonso, and H. A. 6 Swadlow, Axonal conduction delays, brain state, and corticogeniculate communication, J. Neurosci.37, 6342 (2017)

    work page 2017

  4. [4]

    M´ esz´ aros, J

    B. M´ esz´ aros, J. C. Knight, and T. Nowotny, Efficient event-based delay learning in spiking neural networks, Nat. Commun.16, 10422 (2025)

    work page 2025

  5. [5]

    Zamora-Munt, C

    J. Zamora-Munt, C. Masoller, J. Garcia-Ojalvo, and R. Roy, Crowd synchrony and quorum sensing in delay- coupled lasers, Phys. Rev. Lett.105, 264101 (2010)

    work page 2010

  6. [6]

    Junges, T

    L. Junges, T. P¨ oschel, and J. A. Gallas, Characteriza- tion of the stability of semiconductor lasers with delayed feedback according to the Lang-Kobayashi model, Eur. Phys. J. D67, 149 (2013)

    work page 2013

  7. [7]

    A. E. D. Barioni, A. N. Montanari, and A. E. Motter, Interpretable disorder-promoted synchronization and co- herence in coupled laser networks, Phys. Rev. Lett.135, 197401 (2025)

    work page 2025

  8. [8]

    L.-L. Ye, N. Vigne, F.-Y. Lin, H. Cao, and Y.-C. Lai, Optimal sparse networks for synchronization of semicon- ductor lasers (2025), arXiv:2511.03205

    arXiv 2025

Show all 60 references
  1. [9]

    T. Li, A. K. Sahu, M. Zaheer, M. Sanjabi, A. Talwalkar, and V. Smith, Federated optimization in heterogeneous networks, MLSys.2, 429 (2020)

    2020

  2. [10]

    W. Liu, L. Chen, and W. Zhang, Decentralized feder- ated learning: Balancing communication and computing costs, IEEE Trans. Signal Inf. Process. over Networks.8, 131 (2022)

    2022

  3. [11]

    Zhang, M

    X. Zhang, M. Hong, and N. Elia, Understanding a class of decentralized and federated optimization algorithms: A multirate feedback control perspective, SIAM J. Optim. 33, 652 (2023)

    2023

  4. [12]

    Tian and C.-L

    Y.-P. Tian and C.-L. Liu, Consensus of multi-agent sys- tems with diverse input and communication delays, IEEE Trans. Automat. Contr.53, 2122 (2008)

    2008

  5. [13]

    A. N. Montanari, A. E. D. Barioni, C. Duan, and A. E. Motter, Optimal flock formation induced by agent het- erogeneity, Nat. Commun.16, 9626 (2025)

    2025

  6. [14]

    Trinh, P

    H. Trinh, P. T. Nam, and T. Fernando, Existence and design of functional observers for time-delay systems with delayed output measurements (2026), arXiv:2603.09395

    2026 arXiv

  7. [15]

    Punetha and L

    N. Punetha and L. Wetzel, Heterogeneity-induced syn- chronization in delay-coupled electronic oscillators, Phys. Rev. E106, L052201 (2022)

    2022

  8. [16]

    Shahal, A

    S. Shahal, A. Wurzberg, I. Sibony, H. Duadi, E. Shni- derman, D. Weymouth, N. Davidson, and M. Fridman, Synchronization of complex human networks, Nat. Com- mun.11, 3854 (2020)

    2020

  9. [17]

    W. Zou, J. Lu, Y. Tang, C. Zhang, and J. Kurths, Control of delay-induced oscillation death by coupling phase in coupled oscillators, Phys. Rev. E84, 066208 (2011)

    2011

  10. [18]

    Biswas, T

    D. Biswas, T. Mandal, and T. Banerjee, Transition among oscillation death, amplitude death, and revival of oscillation in coupled time-delayed systems with dif- fusivity and common environment, Chaos Soliton Fract. 188, 115550 (2024)

    2024

  11. [19]

    S. Kim, S. H. Park, and C. S. Ryu, Multistability in coupled oscillator systems with time delay, Phys. Rev. Lett.79, 2911 (1997)

    1997

  12. [20]

    C. R. S. Williams, F. Sorrentino, T. E. Murphy, and R. Roy, Synchronization states and multistability in a ring of periodic oscillators: Experimentally variable cou- pling delays, Chaos23, 043117 (2013)

    2013

  13. [21]

    Kantner, E

    M. Kantner, E. Sch¨ oll, and S. Yanchuk, Delay-induced patterns in a two-dimensional lattice of coupled oscilla- tors, Sci. Rep.5, 8522 (2015)

    2015

  14. [22]

    G. C. Sethia, A. Sen, and F. M. Atay, Clustered chimera states in delay-coupled oscillator systems, Phys. Rev. Lett.100, 144102 (2008)

    2008

  15. [23]

    J. D. Hart, K. Bansal, T. E. Murphy, and R. Roy, Ex- perimental observation of chimera and cluster states in a minimal globally coupled network, Chaos26, 094801 (2016)

    2016

  16. [24]

    Gjurchinovski, E

    A. Gjurchinovski, E. Sch¨ oll, and A. Zakharova, Control of amplitude chimeras by time delay in oscillator networks, Phys. Rev. E95, 042218 (2017)

    2017

  17. [25]

    L. M. Pecora and T. L. Carroll, Master stability functions for synchronized coupled systems, Phys. Rev. Lett.80, 2109 (1998)

    1998

  18. [26]

    Acharyya, P

    S. Acharyya, P. Pradhan, and C. Meena, A comprehen- sive review on master stability functions in complex net- work dynamics, Rep. Prog. Phys.89, 073901 (2026)

    2026

  19. [27]

    Nishikawa and A

    T. Nishikawa and A. E. Motter, Synchronization is op- timal in nondiagonalizable networks, Phys. Rev. E73, 065106 (2006)

    2006

  20. [28]

    Nishikawa and A

    T. Nishikawa and A. E. Motter, Maximum performance at minimum cost in network synchronization, Phys. D 224, 77 (2006)

    2006

  21. [29]

    L. M. Pecora, F. Sorrentino, A. M. Hagerstrom, T. E. Murphy, and R. Roy, Cluster synchronization and iso- lated desynchronization in complex networks with sym- metries, Nat. Commun.5, 4079 (2014)

    2014

  22. [30]

    Mulas, C

    R. Mulas, C. Kuehn, and J. Jost, Coupled dynamics on hypergraphs: Master stability of steady states and syn- chronization, Phys. Rev. E101, 062313 (2020)

    2020

  23. [31]

    L. V. Gambuzza, F. Di Patti, L. Gallo, S. Lepri, M. Ro- mance, R. Criado, M. Frasca, V. Latora, and S. Boc- caletti, Stability of synchronization in simplicial com- plexes, Nat. Commun.12, 1255 (2021)

    2021

  24. [32]

    Zhang, V

    Y. Zhang, V. Latora, and A. E. Motter, Unified treat- ment of synchronization patterns in generalized networks with higher-order, multilayer, and temporal interactions, Commun. Phys.4, 195 (2021)

    2021

  25. [33]

    Carletti, L

    T. Carletti, L. Giambagli, and G. Bianconi, Global topo- logical synchronization on simplicial and cell complexes, Phys. Rev. Lett.130, 187401 (2023)

    2023

  26. [34]

    J. Sun, E. M. Bollt, and T. Nishikawa, Master stability functions for coupled nearly identical dynamical systems, EPL85, 60011 (2009)

    2009

  27. [35]

    Zhang and A

    Y. Zhang and A. E. Motter, Identical synchronization of nonidentical oscillators: when only birds of different feathers flock together, Nonlinearity31, R1 (2018)

    2018

  28. [36]

    S. Leng, W. Lin, and J. Kurths, Basin stability in delayed dynamics, Sci. Rep.6, 21449 (2016)

    2016

  29. [37]

    B¨ orner, P

    R. B¨ orner, P. Schultz, B.¨Unzelmann, D. Wang, F. Hell- mann, and J. Kurths, Delay master stability of iner- tial oscillator networks, Phys. Rev. Research2, 023409 (2020)

    2020

  30. [38]

    C.-U. Choe, T. Dahms, P. H¨ ovel, and E. Sch¨ oll, Con- trolling synchrony by delay coupling in networks: From in-phase to splay and cluster states, Phys. Rev. E81, 025205 (2010)

    2010

  31. [39]

    Keane, T

    A. Keane, T. Dahms, J. Lehnert, S. A. Suryanarayana, P. H¨ ovel, and E. Sch¨ oll, Synchronisation in networks of delay-coupled type-I excitable systems, Eur. Phys. J. B 85, 407 (2012)

    2012

  32. [40]

    Dahms, J

    T. Dahms, J. Lehnert, and E. Sch¨ oll, Cluster and group synchronization in delay-coupled networks, Phys. Rev. E 7 86, 016202 (2012)

    2012

  33. [41]

    Mahdavi, M

    E. Mahdavi, M. Zarei, P. H¨ ovel, and F. Shahbazi, Time-delayed dynamics in regular Kuramoto networks with inertia: Multistability, traveling waves, chimera states, and transitions to seizure-like activity (2025), arXiv:2512.16640

    2025

  34. [42]

    Bellen and M

    A. Bellen and M. Zennaro,Numerical Methods for Delay Differential Equations(Oxford University Press, 2013)

    2013

  35. [43]

    Richard, Time-delay systems: an overview of some recent advances and open problems, Automatica39, 1667 (2003)

    J.-P. Richard, Time-delay systems: an overview of some recent advances and open problems, Automatica39, 1667 (2003)

    2003

  36. [44]

    [44–47], for details on the theoretical analysis, cou- pling classes, and network optimization method

    See Supplemental Material at [URL], which includes Refs. [44–47], for details on the theoretical analysis, cou- pling classes, and network optimization method

  37. [45]

    Engelborghs, T

    K. Engelborghs, T. Luzyanina, and D. Roose, Numeri- cal bifurcation analysis of delay differential equations, J. Comput. Appl. Math.125, 265 (2000)

    2000

  38. [46]

    Kozyreff, A

    G. Kozyreff, A. G. Vladimirov, and P. Mandel, Dynam- ics of a semiconductor laser array with delayed global coupling, Phys. Rev. E64, 016613 (2001)

    2001

  39. [47]

    B. Liu, Y. Liu, and Y. Braiman, Coherent addition of high power laser diode array with a V-shape external Talbot cavity, Opt. Express.16, 20935 (2008)

    2008

  40. [48]

    B. Liu, Y. Braiman, N. Nair, Y. Lu, Y. Guo, P. Colet, and M. Wardlaw, Nonlinear dynamics and synchronization of an array of single mode laser diodes in external cavity subject to current modulation, Opt. Commun.324, 301 (2014)

    2014

  41. [49]

    Lang and K

    R. Lang and K. Kobayashi, External optical feedback effects on semiconductor injection laser properties, IEEE J. Quantum Electron.16, 347 (1980)

    1980

  42. [50]

    Masoller, Implications of how the linewidth enhance- ment factor is introduced on the Lang and Kobayashi model, IEEE J

    C. Masoller, Implications of how the linewidth enhance- ment factor is introduced on the Lang and Kobayashi model, IEEE J. Quantum Electron.33, 796 (1997)

    1997

  43. [51]

    T. Heil, I. Fischer, and W. Els¨ aßer, Influence of amplitude-phase coupling on the dynamics of semicon- ductor lasers subject to optical feedback, Phys. Rev. A 60, 634 (1999)

    1999

  44. [52]

    Lenstra, B

    D. Lenstra, B. Verbeek, and A. Den Boef, Coherence col- lapse in single-mode semiconductor lasers due to optical feedback, IEEE J. Quantum Electron.21, 674 (2003)

    2003

  45. [53]

    Nishikawa and A

    T. Nishikawa and A. E. Motter, Network synchroniza- tion landscape reveals compensatory structures, quanti- zation, and the positive effect of negative interactions, Proc. Natl. Acad. Sci. U.S.A.107, 10342 (2010)

    2010

  46. [54]

    Mikaberidze and D

    G. Mikaberidze and D. Taylor, Emergent topology of op- timal networks for synchrony, arXiv:2509.18279 (2025)

    2025 arXiv

  47. [55]

    S.-W. Son, B. J. Kim, H. Hong, and H. Jeong, Dynamics and directionality in complex networks, Phys. Rev. Lett. 103, 228702 (2009)

    2009

  48. [56]

    Zeng, S.-W

    A. Zeng, S.-W. Son, C. H. Yeung, Y. Fan, and Z. Di, Enhancing synchronization by directionality in complex networks, Phys. Rev. E83, 045101 (2011)

    2011

  49. [57]

    1 SUPPLEMENTAL MATERIAL Generalized Master Stability of Heterogeneous Delay-Coupled Networks CONTENTS SI

    Custom code used to generate data for this work can be found at our GitHub repository:https://github.com/ aedbarioni/DelayedMSF. 1 SUPPLEMENTAL MATERIAL Generalized Master Stability of Heterogeneous Delay-Coupled Networks CONTENTS SI. Conditions for the MSF of delay-coupled sy...

  50. [58]

    Therefore, for any similarity matrixP(including one that transformsAinto its Jordan form), we have thatP −1∆P=dP −1InP= ∆, which is diagonal and hence lower triangular

    Networks with identical indegrees—In this case, ∆ =dI n. Therefore, for any similarity matrixP(including one that transformsAinto its Jordan form), we have thatP −1∆P=dP −1InP= ∆, which is diagonal and hence lower triangular. Special cases of this class of networks include all...

  51. [59]

    These networks are generally represented by the following lower-triangular adjacency matrix: A=   A11 0

    Master-slave networks with strictly heterogeneous self-coupling—A master-slave network is defined by a directed graph with a single master node and a hierarchical organization among the remaining nodes. These networks are generally represented by the following lower-triangular...

  52. [60]

    weakly diffusive cou- pling,

    Master-slave networks with identical self-coupling—Consider the adjacency matrix (S1) with identical nonzero diagonal elements, which we normalize to one without loss of generality:A ii = 1,∀i. In this case,Ahas repeated eigenvalues. Thus, it may not be diagonalizable, giving ...

Pith tools

Reviewed August 14, 2026 · model on record in the stance chip above.