Adaptive Cucker-Smale Networks: Limiting Laplacian Time-Varying Dynamics

Christian Kuehn; Jaeyoung Yoon

REVIEW 3 major objections 4 minor 31 references

Adaptive Cucker-Smale Networks: Limiting Laplacian Time-Varying Dynamics

T0 review · 3 major / 4 minor · reviewed 2026-08-07 · deepseek-v4-flash

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

Pith's one-line read Fast-adapting opinion networks reach consensus under three graph conditions.

desk verdict New model worth reading, but the Fenichel transfer is asserted rather than proved and Theorem 3.10's counting estimate needs correction. read the letter →

arxiv 2506.06131 v1 pith:7DEOKSFM submitted 2025-06-06 math.DS math.CA

Christian Kuehn , Jaeyoung Yoon This is my paper

classification math.DSmath.CA MSC 34D0637N9905C5034E15

keywords adaptiveCucker-SmaleopinionclusteringLaplaciandynamicstemporalgraphsingularlimitboundedconfidenceconsensusflocking

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

This paper proposes an adaptive Cucker-Smale model in which the strength of each connection between agents is itself a state variable that tracks how similar the agents' velocity vectors are. The authors argue that in the fast-adaptation limit these nonlinear dynamics reduce to a linear consensus process on a time-varying graph, and they prove exponential convergence for three classes of temporal graphs: complete symmetric networks, neighbor-connected digraphs, and strongly connected graphs with bounded asymmetry. If these results hold, adaptive networks driven by opinion similarity produce asymptotic flocking rather than permanent disagreement, providing a bridge between nonlinear opinion dynamics and classical Laplacian consensus theory.

What carries the argument

The central object is the time-varying Laplacian matrix $L(t)$ of the temporal graph whose edge weights are $(\kappa / I_i(t)) \psi_{ij}(t) a_{ij}(t)$, where $\psi_{ij}$ is the Hegselmann-Krause detection kernel that cuts off at distance $d$, $I_i$ counts the neighbors within that radius, and $a_{ij} = (v_i \cdot v_j)/(|v_i||v_j|)$ is the cosine similarity of velocities. The argument works by exploiting the spectral structure of this Laplacian: Theorem 3.2 uses a comparison lemma for eigenvalues of Laplacians with element-wise ordered adjacency matrices and the Fiedler value, Theorem 3.6 builds a dichotomy spectrum estimate on the matrix of pairwise squared distances using the neighbor-connected property, and Theorem 3.10 uses a Lyapunov function on the fluctuation vector with a combinatorial bound on cross-term sums that is controlled by the asymmetry constants.

What would settle it

Run the full adaptive system (1.1) with a fixed finite adaptation speed $\varepsilon$ and an initial configuration that satisfies the hypotheses of one of the three theorems, then measure the maximal velocity difference over time. If the velocity difference fails to converge to zero for some finite $\varepsilon$, the singular-limit transfer breaks down and the theorems describe only the limit system, not the adaptive model.

Watch

Extended reading notes

Core claim

The central claim is that opinion clustering without consensus can be captured by an adaptive Cucker-Smale model whose coupling weights are state variables that tend to the cosine similarity of velocities, and that in the singular limit of infinitely fast adaptation the model becomes a linear Laplacian consensus system over a temporal graph. The paper establishes exponential convergence to consensus in three regimes: complete networks with symmetric weights bounded below (Theorem 3.2), neighbor-connected digraphs with positive lower-bounded weights (Theorem 3.6), and strongly connected temporal graphs with near-symmetric weights and a balanced neighbor count (Theorem 3.10). In each case the result is stated as convergence of the states, and the corresponding corollaries transfer the conclusion to the original adaptive model in the fast-adaptation limit, yielding asymptotic flocking with exponential rate.

Load-bearing premise

The paper assumes rather than proves that the adaptive Cucker-Smale model has unique global Carathéodory solutions for the chosen initial data, and then transfers the singular-limit theorems to the full finite-$\varepsilon$ model via Fenichel's theorem without estimating how small $\varepsilon$ must be for that transfer to be valid.

Editorial extensions

If this is right

Under the three graph conditions, all agents' velocities converge to the same limiting velocity exponentially fast, so the system exhibits asymptotic flocking with group formation and velocity alignment.
The complete-network result has an explicit exponential rate depending only on the minimum weight and the number of agents, giving a quantitative consensus bound that holds for all positive lower-bounded symmetric temporal graphs.
The neighbor-connected result shows that directed and asymmetric interaction patterns, including leader-based and three-group structures, are still sufficient for consensus as long as the edge weights stay positive.
The strongly-connected but asymmetric result shows that convergence survives bounded asymmetry in the weights and requires only that each agent has at least about half of the others as neighbors, with a quantitative decay rate $\gamma_m \delta / N$.
For the original adaptive model these theorems provide, via the singular limit, sufficient conditions for fast adaptation to produce clumping of opinions rather than persistent fragmentation.

Reading between the lines

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

An implicit extension is that the methodology transfers beyond the adaptive Cucker-Smale model itself: any network with slowly varying node states and fast adaptation of edge weights should be analyzable by first deriving a singular-limit Laplacian system on a temporal graph and then applying the three sufficiency conditions as consensus certificates.
A concrete testable extension would be to measure the finite-adaptation error by simulating the full system (1.1) for a sequence of increasing $\varepsilon$ and checking that the convergence time and final cluster structure approach the singular-limit predictions; the paper does not provide this comparison.
The neighbour-connected condition suggests that consensus is robust to switching edge sets, so a natural check is whether convergence persists under randomly switching topologies or under adaptive rewiring driven by the pairwise distances, rather than by the velocities as in the paper's adaptive rule.
Because the three theorems only give sufficient conditions, a natural neighbouring question is whether there exist critical thresholds in the asymmetry parameters below which consensus fails, indicated qualitatively by the toy-model separation result for two particles.

Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Desk Editor's Note

read the letter

Short version: this is a serious paper with a genuinely new model, but the advertised bridge from the singular limit to the original adaptive system is asserted, not proved, and Theorem 3.10 has a counting error in the key estimate. The Laplacian results mostly stand; the transfer is the load-bearing overreach.

What is actually new: the combined adaptive Cucker-Smale model with Hegselmann-Krause cutoff and normalization, the singular-limit formulation as Laplacian dynamics on temporal graphs, and the sufficient degree and asymmetry conditions in Theorem 3.10. Theorems 3.2 and 3.6 are essentially correct; 3.2 is a clean spectral comparison, and 3.6's row-dominance plus dichotomy-spectrum argument is credible. The examples and numerics match the claims, and the paper is honest in Section 2.1 that existence and uniqueness of Caratheodory solutions are assumed rather than proved.

Soft spots, in proportion. The Fenichel transfer is the big one. System (1.1) is discontinuous at the HK cutoff and the normalizer I_i, and the fast contraction toward the critical manifold can be reset at every switching event. The paper gives no lower bound on epsilon and no bound on accumulated switching error, so its conclusion that the singular-limit analysis explains the finite-epsilon model is not supported. The three theorems should be read as results about the limit system (3.3) unless this gap is closed. That is fixable, but it is central to the abstract's promise.

Second, Theorem 3.10: the estimate in (3.24) subtracts 2s(Nm-s+1) cross edges from (N-s)NM, but the number of directed edges from the negative set to the positive set is only s(Nm-s+1) (one directed edge per cross undirected edge). The over-subtraction means the count is not an upper bound, so the Lyapunov estimate is not justified as written. The theorem may still be true, but the proof needs repair.

Minor points: Lemma 3.3's convex-hull argument is sketched but plausible; the two-particle toy model is explicitly not overclaimed. The citation pattern looks fine, including the reliance on the authors' own dichotomy-spectrum paper, which is published and used here in a legitimate external way.

Bottom line: worth a serious referee. This is for readers in opinion dynamics and adaptive networks who want a new model and a template for singular-limit Laplacian analysis; it is not a finished treatment of finite-epsilon behavior. The model and the two easier theorems are useful, the flaws are specific and likely fixable, and the numerics give real support. I would not desk-reject; I would ask for a corrected counting argument and either a proof or an explicit scoping of the Fenichel transfer. Send it to review.

Referee Report

3 major / 4 minor

Summary. The paper proposes an adaptive Cucker-Smale model with Hegselmann-Krause-type bounded-confidence coupling (1.1), studies the fast-adaptation singular limit ε→∞, and analyzes the resulting Laplacian dynamics (3.3) over temporal graphs. It proves exponential consensus for three graph classes: complete symmetric graphs with uniformly positive weights (Theorem 3.2), neighbor-connected directed graphs with positive lower-bounded weights (Theorem 3.6), and strongly connected 'almost symmetric' graphs satisfying a quantitative degree/asymmetry condition (Theorem 3.10). Corollaries translate these into asymptotic flocking statements for the singular-limit model, and numerical experiments illustrate the convergence and clustering behavior.

Significance. If the results are made fully rigorous, the paper offers a useful template: it converts a nonlinear adaptive-network model into a linear non-autonomous consensus problem and gives explicit exponential rates rather than only qualitative convergence. The neighbor-connected and almost-symmetric conditions in Theorems 3.6 and 3.10 are concrete and checkable, and the numerical section supports the exponential-decay predictions. The authors are also transparent about the Carathéodory-solution assumption in Section 2.1. However, the advertised bridge from the singular limit back to the original finite-ε model is not proved, and Theorem 3.10's sufficient condition contains counting inconsistencies; these issues currently cap the significance.

major comments (3)

[§1.2, §6, and §2.1] The assertion in §1.2 and §6 that 'by Fenichel's Theorem' the singular-limit analysis transfers to the adaptive model (1.1) is not justified. The vector field of (1.1) is discontinuous at ‖x_i-x_j‖=d, Fenichel's theorem requires smoothness and normal hyperbolicity of a compact critical manifold, no ε lower bound is given, and Section 2.1 explicitly assumes rather than proves existence and uniqueness of global Carathéodory solutions. Since the abstract's advertised conclusion concerns the original adaptive model for fast adaptation, the paper should either prove a switching/geometric-singular-perturbation statement or reformulate the conclusions as conditional on such a transfer.
[§3.3, Eqs. (3.24) and (3.18)] The counting underlying assumption (B2) is inconsistent. In bounding the number of A_n edges inside the negative set, the proof subtracts 2s(Nm-s+1) from (N-s)NM, but the term to subtract is the number of directed edges leaving the negative set toward the positive set, which is only s(Nm-s+1). Moreover, the expanded bracket in (3.18), 3s^2-(NM+2Nm+1)s+NNM, does not match the bracket defined in (3.24), which expands to 3s^2-(NM+2Nm+3)s+NNM. The sufficient condition actually proved is therefore different from the one stated; Theorem 3.10, Corollary 3.11, and the numbers in Table 1 all depend on this condition and must be re-derived.
[Corollary 3.4] The stated condition d0 < d - 2‖v(0)‖_F/am is parameter-inconsistent with system (3.3). In the proof, the complete-graph system is written without the factor κ in (3.11), Theorem 3.2 is applied with decay e^{-amt} instead of e^{-κ am t}, and then the integral contains e^{-κ ms}. Carrying κ through the computation gives the condition d0 < d - 2‖v(0)‖_F/(κ am), so the corollary as stated requires a parameter restriction or a corrected statement.

minor comments (4)

[Theorem 3.6 proof] The step 'one can see that a linear ODE u'=E(t)u has a dichotomy spectrum contained in (-∞,0)' is too terse; the row dominance (3.15) directly implies that the L∞ logarithmic norm of E(t) is at most -2w_m, which would make the comparison argument self-contained.
[Eq. (4.29)] In the displayed lower bound, the factor (1-e^{κt}) should read (1-e^{-κt}); as written the expression diverges and the inequality is not a useful lower bound.
[§1.2 and §2.1] There is a duplicated phrase 'Theorem Theorem 3.6' in the introduction, and the statement in §1.2 that existence is discussed in Section 2.2 should refer to Section 2.1.
[Theorem 3.10 proof] The sentence 'Integrating (3.23) and (3.24)' should be 'Combining' or 'Using', since the displayed inequalities are pointwise differential inequalities rather than integrated bounds.

Circularity Check

0 steps flagged · score 0.0 of 10

No circular derivation: the consensus theorems are proved from explicit graph hypotheses, the Laplacian limit is derived rather than assumed, and the reliance on the authors' prior dichotomy-spectrum work is an external published tool, not a self-referential premise.

full rationale

The three main results are derived, not fitted. Theorem 3.2 proves exponential average consensus from symmetric weights with a uniform positive lower bound via the Laplacian Lyapunov identity and the spectral comparison Lemma 3.1; no rate constant is tuned to the data. Theorem 3.6 proves exponential consensus for neighbor-connected lower-bounded weighted digraphs from a differential inequality for the pairwise error vector, invoking dichotomy spectrum theory from the published article [18] as an external tool, not as a premise that already contains this paper's conclusion. Theorem 3.10 provides a self-contained Lyapunov estimate under assumptions (B), with the constants chosen explicitly in the proof, again without fitting. The corollaries for the singular adaptive Cucker-Smale model (3.3) combine these theorems with Lemma 3.3, which derives a positive lower bound on velocity similarities from the initial data; the bound is an input condition, not an output that is then called a prediction. The numerical simulations are illustrative, not used to fit parameters, and no quantity appearing as a 'prediction' is defined in terms of the data it is supposed to explain. The central caveat is the Fenichel transfer from the singular limit back to the full adaptive model (Section 1.2 and Section 6) and the assumed existence of global Caratheodory solutions (Section 2.1); these are unproved assumptions that affect correctness, but they do not make any theorem true by definition or by self-citation. The singular Laplacian system is a distinct mathematical object derived from the fast-adaptation limit, and no equation in the paper reduces to its own input. The one self-citation, reference [18], is load-bearing in Theorem 3.6, but it is a published external result with stated hypotheses, so it constitutes independent support rather than circularity. Overall, the derivation chain is self-contained and the circularity score is 0.

Assumptions & free parameters 3 free parameters · 4 assumptions · 0 invented entities

The central claim rests on the unproved well-posedness of the discontinuous adaptive model, the unproved applicability of Fenichel's theorem, and the countably-many-switching assumption on temporal graphs. The model introduces adaptive coupling strengths kappa_ij as variables, but these are model states, not invented physical entities. No new particles, forces, or dimensions are proposed.

free parameters (3)

d (interaction radius) = not fitted; simulation values include d=1, d=0.009-0.012, d=0.5-0.65
The interaction radius d is a hyperparameter of the model that determines which agents interact. It is chosen by hand in the numerical experiments and is not derived from data. The asymptotic clustering results treat d as fixed, but the cluster count in Section 5.3 depends on its value.
epsilon (adaptation time constant) = not specified in numerics; singular limit epsilon to infinity
The paper analyzes the singular limit epsilon to infinity but gives no bound on 'sufficiently large' epsilon. In the numerical simulations of the full adaptive model (Section 5.2), the value of epsilon is not reported, so the connection to the singular limit is not demonstrated.
kappa (global coupling strength) = implicitly 1 in most of the analysis
The global coupling strength kappa multiplies the Laplacian terms. It appears in rate bounds, but the paper absorbs it or sets it to 1 without calibration. It is a model parameter rather than a derived quantity.

assumptions (4)

domain assumption There exists a set of initial conditions for which the adaptive CS model (1.1) admits a unique global Caratheodory solution.
Section 2.1 explicitly states that all analysis proceeds under this assumption and that rigorous investigation is left to future work. This is load-bearing because the corollaries for flocking restrict initial data to this set D.
domain assumption Fenichel's geometric singular perturbation theorem applies to the adaptive CS model with the discontinuous Hegselmann-Krause cutoff psi_ij and normalizer I_i.
The conclusion claims 'Due to Fenichel's Theorem, our results for this singular limit transfer to the full model.' Fenichel's theorem requires smooth vector fields, while (1.1) is discontinuous. No smoothness regularization or alternative justification is provided.
domain assumption The edge set E(t) changes at most countably many times over time.
Assumptions (A3) and (B3) in Section 3 restrict the temporal graphs. This ensures Caratheodory solutions for the Laplacian dynamics, but it is an additional structural assumption on the model.
standard math Standard spectral and dichotomy-spectrum results: Courant-Fischer, Laplacian spectral monotonicity, the comparison theorem [23], and exponential dichotomy [13].
These results are used in the proofs of Theorem 3.2 and Theorem 3.6. They are background results from the literature and not specific to this paper.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Adaptive Cucker-Smale Networks: Limiting Laplacian Time-Varying Dynamics." pith.science (2026). https://pith.science/paper/7DEOKSFM

@misc{pith2026250606131,
  author       = {Pith},
  title        = {Pith review of: Adaptive Cucker-Smale Networks: Limiting Laplacian Time-Varying Dynamics},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/7DEOKSFM}},
  note         = {Machine review of arXiv:2506.06131}
}

read the original abstract

Differences in opinion can be seen as distances between individuals, and such differences do not always vanish over time. In this paper, we propose a modeling framework that captures the formation of opinion clusters, based on extensions of the Cucker Smale and Hegselmann Krause models to a combined adaptive (or co-evolutionary) network. Reducing our model to a singular limit of fast adaptation, we mathematically analyze the asymptotic behavior of the resulting Laplacian dynamics over various classes of temporal graphs and use these results to explain the behavior of the original proposed adaptive model for fast adaptation. In particular, our approach provides a general methodology for analyzing linear consensus models over time-varying networks that naturally arise as singular limits in many adaptive network models.

Figures

Figures reproduced from arXiv: 2506.06131 by the authors.

**Figure 1.** Sketch of proof in Lemma 3.3 Corollary 3.4. Suppose that initial data {(x 0 i , v 0 i )}i∈[N] satisfy max i,j∈[N] kx 0 i − x 0 jk ≤ d0, am := min i,j∈[N] v 0 i · v 0 j |v 0 i ||v 0 j | > 0, for a positive constant d0 < d − 2kv˜(0)kF /am, where k · kF denotes the Frobenius norm. Then, the solution {(xi , vi)}i∈[N] to (3.3) exhibits asymptotic flocking. Proof. First, we show that all particles are within a detection r… view at source ↗

**Figure 2.** Examples of neighbor-connected graphs tations of these graphs with a vertex set of size 5 are shown in [PITH_FULL_IMAGE:figures/full_fig_p026_2.png] view at source ↗

**Figure 3.** 4.3 Examples of (B) In this subsection, we provide examples of graphs satisfying the assumption (B). As we discussed in Remark 3.9, we consider a graph G = (V, E,W) where the edge set E is symmetric. The construction process involves satisfying (B1) for the edge set and (B2) for the weights. To begin, we consider the edge set under the constraints of the first assumption. This can be categorized into two cases: • W… view at source ↗

Figures from the paper (7 more)

**Figure 3.** Figure 3: Adjacency matrix representation of a directed neighbor-connected graph, as described in Section 4.2. The graph illustrates the neighbor-connectivity defined in Definition 3.5. (i) Select NM + 1 vertices to form a complete subgraph. (ii) Independently, select Nm + 1 v…

**Figure 4.** Figure 4: Adjacency matrix of the edge set E for a connected graph ([N], E,W) satisfying the neighbor degree constraint (B1), with the weight information W excluded. The graph is connected, and each node connects to a limited number of neighbors. The simulations are conducted in…

**Figure 5.** Figure 5: Diameter evolution over time for various graph typ [PITH_FULL_IMAGE:figures/full_fig_p029_5.png]

**Figure 6.** Figure 6: Evolution over time for Case 6 in Table 2. (a)-(f) sh [PITH_FULL_IMAGE:figures/full_fig_p032_6.png]

**Figure 7.** Figure 7: Evolution over time for Case 6 in Table 3 when [PITH_FULL_IMAGE:figures/full_fig_p033_7.png]

**Figure 8.** Figure 8: (Singular adaptive CS model) Result for different va [PITH_FULL_IMAGE:figures/full_fig_p034_8.png]

**Figure 9.** Figure 9: Asymptotic graph structures for selected values o [PITH_FULL_IMAGE:figures/full_fig_p035_9.png]

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

31 extracted references · 30 canonical work pages

[1]

Altaﬁni and F

C. Altaﬁni and F. Ceragioli. Signed bounded conﬁdence mo dels for opinion dynamics. Automatica, 93:114–125, 2018

work page 2018
[2]

A. V. Andreev, A. A. Badarin, V. A. Maximenko, and A. E. Hra mov. Forecasting macroscopic dynamics in adaptive kuramoto network using re servoir computing. Chaos: An Interdisciplinary Journal of Nonlinear Science , 32(10):103126, 10 2022

work page 2022
[3]

Bellomo, S.-Y

N. Bellomo, S.-Y. Ha, N. Outada, and J. Yoon. On the mathem atical theory of be- havioral swarms emerging collective dynamics. Mathematical Models and Methods in Applied Sciences, 32(14):2927–2959, 2022

work page 2022
[4]

Berner, T

R. Berner, T. Gross, C. Kuehn, J. Kurths, and S. Yanchuk. A daptive dynamical networks. Physics Reports, 1031:1–59, 2023. Adaptive dynamical networks

work page 2023
[5]

Berner, E

R. Berner, E. Sch¨ oll, and S. Yanchuk. Multiclusters in n etworks of adaptively coupled phase oscillators. SIAM Journal on Applied Dynamical Systems , 18(4):2227–2266, 2019

work page 2019
[6]

Berner, S

R. Berner, S. Vock, E. Sch¨ oll, and S. Yanchuk. Desynchro nization transitions in adap- tive networks. Phys. Rev. Lett. , 126:028301, Jan 2021

work page 2021
[7]

V. D. Blondel, J. M. Hendrickx, and J. N. Tsitsiklis. Cont inuous-time average- preserving opinion dynamics with opinion-dependent commu nications. SIAM Journal on Control and Optimization , 48(8):5214–5240, 2010

work page 2010
[8]

Cai and H

K. Cai and H. Ishii. Average consensus on general strongl y connected digraphs. Auto- matica, 48(11):2750–2761, 2012

work page 2012

Show all 31 references

[9]

Ceragioli and P

F. Ceragioli and P. Frasca. Continuous and discontinuou s opinion dynamics with bounded conﬁdence. Nonlinear Analysis: Real World Applications , 13(3):1239–1251, 2012

2012
[10]

J. J. DaCunha. Stability for time varying linear dynami c systems on time scales. Journal of Computational and Applied Mathematics , 176(2):381–410, 2005

2005
[11]

Feketa, A

P. Feketa, A. Schaum, and T. Meurer. Synchronization an d multicluster capabilities of oscillatory networks with adaptive coupling. IEEE Transactions on Automatic Control , 66(7):3084–3096, 2021

2021
[12]

Fenichel

N. Fenichel. Geometric singular perturbation theory f or ordinary diﬀerential equations. Journal of Diﬀerential Equations , 31(1):53–98, 1979

1979
[13]

A. M. Fink. Almost periodic diﬀerential equations , volume 377. Springer, 2006

2006
[14]

J. Gallier. Spectral theory of unsigned and signed grap hs. applications to graph clus- tering: a survey, 2016

2016
[15]

Godsil and G

C. Godsil and G. F. Royle. Algebraic graph theory , volume 207. Springer Science & Business Media, 2001. 36

2001
[16]

R. A. Horn and C. R. Johnson. Matrix Analysis . Cambridge University Press, 1985

1985
[17]

O. H´ ajek. Discontinuous diﬀerential equations, i. Journal of Diﬀerential Equations , 32(2):149–170, 1979

1979
[18]

Jard´ on-Kojakhmetov, C

H. Jard´ on-Kojakhmetov, C. Kuehn, and I. P. Longo. Pers istent synchronization of heterogeneous networks with time-dependent linear diﬀusiv e coupling. SIAM Journal on Applied Dynamical Systems , 23(2):1540–1578, 2024

2024
[19]

D. V. Kasatkin, S. Yanchuk, E. Sch¨ oll, and V. I. Nekorki n. Self-organized emergence of multilayer structure and chimera states in dynamical netwo rks with adaptive couplings. Phys. Rev. E , 96:062211, Dec 2017

2017
[20]

Kuehn et al

C. Kuehn et al. Multiple time scale dynamics , volume 191. Springer, 2015

2015
[21]

Kunegis, S

J. Kunegis, S. Schmidt, A. Lommatzsch, J. Lerner, E. W. D . Luca, and S. Albayrak. Spectral Analysis of Signed Graphs for Clustering, Predicti on and Visualization , pages 559–570
[22]

Y. L. Maistrenko, B. Lysyansky, C. Hauptmann, O. Burylk o, and P. A. Tass. Multi- stability in the kuramoto model with synaptic plasticity. Phys. Rev. E , 75:066207, Jun 2007

2007
[23]

Olech and Z

C. Olech and Z. Opial. Sur une in´ egalit´ e diﬀ´ erentielle. In Annales Polonici Mathematici, volume 7, pages 247–254. Polska Akademia Nauk. Instytut Mat ematyczny PAN, 1960

1960
[24]

Olfati-Saber and R

R. Olfati-Saber and R. Murray. Consensus problems in ne tworks of agents with switch- ing topology and time-delays. IEEE Transactions on Automatic Control , 49(9):1520– 1533, 2004

2004
[25]

Piccoli and F

B. Piccoli and F. Rossi. Generalized solutions to bound ed-conﬁdence models. Mathe- matical Models and Methods in Applied Sciences , 31(06):1237–1276, 2021

2021
[26]

O. V. Popovych, S. Yanchuk, and P. A. Tass. Self-organiz ed noise resistance of os- cillatory neural networks with spike timing-dependent pla sticity. Scientiﬁc Reports , 3(1):2926, Oct 2013

2013
[27]

P¨ otzsche, S

C. P¨ otzsche, S. Siegmund, and F. Wirth. A spectral char acterization of exponential stability for linear time-invariant systems on time scales . Discrete and Continuous Dynamical Systems , 9(5):1223–1241, 2003

2003
[28]

Rainer and U

H. Rainer and U. Krause. Opinion dynamics and bounded co nﬁdence: models, analysis and simulation. 2002

2002
[29]

Thiele, R

M. Thiele, R. Berner, P. A. Tass, E. Sch¨ oll, and S. Yanch uk. Asymmetric adaptivity induces recurrent synchronization in complex networks. Chaos: An Interdisciplinary Journal of Nonlinear Science , 33(2):023123, 02 2023

2023
[30]

J. J. P. Veerman and R. Lyons. A primer on laplacian dynam ics in directed graphs. Nonlinear Phenomena in Complex Systems , 2020. 37

2020
[31]

Zhang and J

H. Zhang and J. Chen. Bipartite consensus of general lin ear multi-agent systems. In 2014 American Control Conference , pages 808–812, 2014. 38

2014

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.

[1] [1]

Altaﬁni and F

C. Altaﬁni and F. Ceragioli. Signed bounded conﬁdence mo dels for opinion dynamics. Automatica, 93:114–125, 2018

work page 2018

[2] [2]

A. V. Andreev, A. A. Badarin, V. A. Maximenko, and A. E. Hra mov. Forecasting macroscopic dynamics in adaptive kuramoto network using re servoir computing. Chaos: An Interdisciplinary Journal of Nonlinear Science , 32(10):103126, 10 2022

work page 2022

[3] [3]

Bellomo, S.-Y

N. Bellomo, S.-Y. Ha, N. Outada, and J. Yoon. On the mathem atical theory of be- havioral swarms emerging collective dynamics. Mathematical Models and Methods in Applied Sciences, 32(14):2927–2959, 2022

work page 2022

[4] [4]

Berner, T

R. Berner, T. Gross, C. Kuehn, J. Kurths, and S. Yanchuk. A daptive dynamical networks. Physics Reports, 1031:1–59, 2023. Adaptive dynamical networks

work page 2023

[5] [5]

Berner, E

R. Berner, E. Sch¨ oll, and S. Yanchuk. Multiclusters in n etworks of adaptively coupled phase oscillators. SIAM Journal on Applied Dynamical Systems , 18(4):2227–2266, 2019

work page 2019

[6] [6]

Berner, S

R. Berner, S. Vock, E. Sch¨ oll, and S. Yanchuk. Desynchro nization transitions in adap- tive networks. Phys. Rev. Lett. , 126:028301, Jan 2021

work page 2021

[7] [7]

V. D. Blondel, J. M. Hendrickx, and J. N. Tsitsiklis. Cont inuous-time average- preserving opinion dynamics with opinion-dependent commu nications. SIAM Journal on Control and Optimization , 48(8):5214–5240, 2010

work page 2010

[8] [8]

Cai and H

K. Cai and H. Ishii. Average consensus on general strongl y connected digraphs. Auto- matica, 48(11):2750–2761, 2012

work page 2012