Pith. sign in

REVIEW 3 major objections 3 minor 60 references

Large Deviations for Markovian Graphon Processes and Associated Dynamical Systems on Networks

T0 review · 3 major / 3 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 This paper proves large deviation principles for rapidly changing dense random graphs and for the particle systems they drive, with explicit rate functions.

desk verdict Serious new LDPs for time-averaged dynamic graphons with inhomogeneous rates; the weak-topology proofs hold up, but the cut-norm theorem rests on an omitted lemma and the abstract overpromises. read the letter →

arxiv 2506.08333 v2 pith:MKUW3SZ3 submitted 2025-06-10 math.PR math.CO

Shankar Bhamidi , Amarjit Budhiraja , Souvik Ray This is my paper
classification math.PRmath.CO MSC 60K3505C80
keywords time-evolvingnetworkslargedeviationsgraphonprocessescut-normtopologyMarkovchainsonedgesinteractingparticlesystemsdenserandomgraphsvariationalrepresentation
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 analyzes dense random directed graphs whose edges flip independently at rates a(n)$β^{{n,±}}$_{i,j}(t) that grow with n and depend on time and on the two endpoints. Because the network changes rapidly over the observation window, the right state descriptor is the graphon time-averaged over the window, and the paper proves laws of large numbers and large deviation principles for both the raw graphon process and its time average, with speed a(n)n², in the weak topology and in the cut-norm topology. The rate functions are explicit: for the time-averaged graphon in the time-homogeneous case, the cost of producing an average graphon f is $\int_{[0,1]^2}(\sqrt{\gamma^+}(1-f)-\sqrt{\gamma^-}f)^2$. These results also yield a large deviation principle for node-level dynamical systems driven by the fast-evolving network, quantifying deviations from the mean-field averaging limit.

What carries the argument

The argument is carried by three tools. (1) A Poisson-random-measure representation of the edge-flip chains turns Laplace asymptotics into the variational formula $-1/(a(n)n^2)\log E e^{-a(n)n^2\Psi(H^n)}=\inf_{\Phi^n}E[\Psi(\bar H^n)+\frac{1}{n^2}\sum_{i,j}\int_0^1(\ell(\varphi^{n,+}_{i,j})\beta^{n,+}_{i,j}+\ell(\varphi^{n,-}_{i,j})\beta^{n,-}_{i,j})ds]$, with $\ell(x)=x\log x-x+1$, taken from [16,17]. (2) Tightness of the associated random measures $\Lambda^n$ on control fields, edge states, and space-time gives weak-limit objects whose balance equation $(1-u)\beta^+v_+=u\beta^-v_-$ reduces the rate function to the square-root form through the identity $\inf_{a_+,a_-:a_+\beta^+(1-h)=a_-\beta^-h}[\ell(a_+)\beta^+(1-h)+\ell(a_-)\beta^-h]=(\sqrt{\beta^+}(1-h)-\sqrt{\beta^-}h)^2$. (3) The lift from weak to cut-norm topology block-approximates the rate kernels, uses a finite permutation set from [25], and covers graphons by a finite family of block functions via a regularity-type lemma [20], while the LDP lower bound is built from an explicitly tilted measure $\nu^{n,\phi}$ whose entropy rate equals $J(\phi)$.

What would settle it

A direct computational falsifier: for time-homogeneous rates, simulate the $n$-vertex edge-flip chains for a range of $n$, estimate $(a(n)n^2)^{-1}\log P(M^n\in B_\square(f,\varepsilon))$ for several target graphons $f$, and compare with $\int(\sqrt{\gamma^+}(1-f)-\sqrt{\gamma^-}f)^2$; if the finite-$n$ rates extrapolate to a different functional, or if the cut-norm topology yields different exponential rates than the weak topology when Assumption 2.2(ii) is dropped, the central claim fails.

Watch

Extended reading notes

Core claim

The central claim is that the Markovian edge-flip model of Definition 2.3, viewed through graphon representations, satisfies a large deviation principle with an explicit, tractable rate function in the fast-flip regime $a(n)\to\infty$. Specifically, under Assumption 2.2(i), the graphon process $H^n$ satisfies an LDP in the weak topology of $L^2([0,1]^3)$ with speed $a(n)n^2$ and rate function $J(\beta^+,\beta^-)(\phi)=\int_{[0,1]^3}(\sqrt{\beta^+_s}(1-\phi_s)-\sqrt{\beta^-_s}\phi_s)^2$, and the time-averaged graphon $M^n$ satisfies an LDP in the weak topology with rate function $I(\beta^+,\beta^-)(f)=\inf_{\phi:\int\phi=f}J(\phi)$. Under the stronger Assumption 2.2, the same LDP holds in the cut-norm quotient space $\widehat{S}_0$ with rate $\widehat I(\widehat f)=\inf_{f\in\widehat f}I(f)$; for time-homogeneous rates this collapses to the product-form integral $\int(\sqrt{\gamma^+}(1-f)-\sqrt{\gamma^-}f)^2dxdy$. The paper also proves that the raw graphon path without window averaging admits no nontrivial path-space LDP, and it establishes an LDP for interacting particle systems on the evolving network whose rate function minimizes $J$ plus the initial-data rate subject to the limiting continuum equation.

Load-bearing premise

The load-bearing premise is Assumption 2.2(ii): the block rate kernels must converge to the limiting kernels uniformly in time, in the spatial $L^1$ sense; if that uniform-in-time control fails, the cut-norm LDP upper bound in Theorem 3.11 collapses.

Editorial extensions

If this is right

  • In the time-homogeneous case the rate function for the time-averaged graphon is simply the spatial integral of a pointwise square, so rare-event probabilities factor over vertex pairs and can be evaluated or optimized pointwise.
  • The cut-norm LDP transfers, via the contraction principle, to LDPs for all homomorphism densities $t(\widehat{M}^n,F)$ of the averaged network, since cut convergence is equivalent to convergence of these densities for every simple graph $F$.
  • For particle systems on the fast network, an averaging-principle LLN holds and deviations from it are exponentially controlled by the variational formula $I^{solution}(v)=\inf_{(\phi,z)\in C(v)}[J(\phi)+I^{initial}(z)]$; this gives a quantitative handle on rare noise-induced patterns in excitable network models.
  • The nonexistence of a path-space LDP for the un-averaged process clarifies the observational regime: time-window averaging is not a convenience but the correct state descriptor in the fast-flip limit.

Reading between the lines

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

  • One extension the paper leaves implicit is an LDP for the empirical measures $\mu^n=(1/s)\int_0^s\delta_{H^n_u}du$; such an LDP would immediately give an LDP for time-integrated motif counts $T_n(F)=\int_0^1 t(\widehat H^n_s,F)ds$ by contraction, and the ingredients developed here (exponential tightness, explicit $J$) suggest the rate should be the pathwise integral of $J$.
  • A testable consequence of the rate function's square-root form is that the optimal way to force an average $f\ne w^*$ is to make the controlled transition rate imbalance $\sqrt{\beta^+}(1-f)=\sqrt{\beta^-}f$ hold at each space-time point; this cost geometry could be checked in small-$n$ experiments by tilting the local rates and measuring the resulting occupancy profile.
  • The cut-norm LDP's dependence on Assumption 2.2(ii) can be probed numerically: choose kernels that converge in $L^1([0,1]^3)$ but oscillate in time within each block (so the uniform-in-time condition fails), and test whether the finite-$n$ exponential rates still converge to $\widehat I$ in $\widehat S_0$; the paper's proof suggests they may not.
  • For the dynamical-system LDP, the paper works in the weak topology on $L^2([0,1])$; a strong-topology version would need a cut-norm LDP for the raw graphon path, a gap the authors flag. A concrete route is to prove such a path LDP in the integrated sense $\frac1s\int_0^s H^n_udu$ mentioned in Remark 3.15.
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.

Referee Report

3 major / 3 minor

Summary. The paper studies large deviations for rapidly evolving Markovian directed random graph processes whose edge-flip rates are a(n) times time- and space-dependent kernels β^{n,±}. The main objects are the graphon process H^n and its time average M^n. The authors prove a law of large numbers (Theorems 3.2, 3.7, 3.8) and large deviation principles in the weak topology (Theorems 3.3 and 3.5) with speed a(n)n² and explicit rate functions J(β⁺,β⁻) and I(β⁺,β⁻). Under the stronger Assumption 2.2, they prove an LDP in the cut-norm quotient space (Theorem 3.11) with rate function given as the lower-semicontinuous envelope of the infimum of I over equivalence classes. They also establish an LDP for interacting particle systems driven by the time-varying network (Theorem 3.21). The proofs of the weak-topology results are detailed, combining concentration estimates, a stochastic-control variational representation, and a tilting construction for the lower bound.

Significance. If the results are fully correct, this is a substantial contribution to the large-deviation theory of dynamic dense random graphs. The weak-topology LDPs are genuinely new in allowing time-inhomogeneous, spatially non-uniform rates and a speed a(n)n², and the explicit rate functions are clean and tractable. The lower-bound construction in Section 5.7 is carefully executed and the rate-function identities in Lemma 5.4 are proved in full. The paper also provides a useful lifting framework from weak to cut-norm topology that goes beyond the static settings of [20,25]. However, two load-bearing parts of the cut-norm program are not proved in the manuscript: Lemma 6.3 is delegated to a 'straightforward adaption' of a static result, and the lower semi-continuity underlying Remark 3.14 is deferred to a supplement. The abstract also contains a claim about a negative path-space LDP that does not appear in the body.

major comments (3)
  1. [Section 6.2.2, Lemma 6.3] Lemma 6.3 is the key step in the cut-norm upper bound that replaces an arbitrary relabeling σ_n ∈ S_n by one of finitely many τ ∈ T uniformly in n, for a time-inhomogeneous block-rate process. The proof is omitted as 'a straightforward adaption of [25, Lemma 3.3]'. This is not a minor technicality: [25] is a static setting, whereas here the rates β^{k}±(s) depend on time, and the adaptation must control the pullback of the time-dependent rates under arbitrary permutations and produce a finite net independent of n and of the rate variation in s. Since the upper-bound lifting in Section 6.2 collapses if Lemma 6.3 fails, the authors should provide the full argument or a detailed proof sketch that addresses the time-inhomogeneity explicitly.
  2. [Remark 3.14 and Section 6.3] Remark 3.14 defers to an online supplement the proof that bI(β⁺,β⁻)(bf) = inf_{f∈bf} I(β⁺,β⁻)(f) in the time-inhomogeneous case. Without this proof, the cut-norm rate function in Theorem 3.11 is only the lower-semicontinuous envelope of that infimum, not the explicit formula that the abstract and introduction advertise. The time-homogeneous case is handled in Corollary 3.12, but the general statement promised in the abstract 'rate functions admit explicit and tractable representations' is not established in the manuscript for the cut-norm LDP. The authors should either include the proof of the lower semi-continuity or clearly qualify the abstract and introduction to state which rate functions are explicit.
  3. [Abstract and Remark 3.15] The abstract states that 'without such local averaging, the rapidly oscillating graphon process does not satisfy a nontrivial path-space LDP', but no such statement or proof appears anywhere in the body. Remark 3.15 instead says that LDPs in a suitable path space are left to future work. This discrepancy is confusing and should be fixed: either prove the claimed negative result, state it as a conjecture, or delete it from the abstract. As written, the abstract asserts a result that the paper does not contain.
minor comments (3)
  1. [Section 2 and Section 3.2] The symbol S is used both for the set of time-homogeneous kernels in Section 2 and for the group of measure-preserving bijections in Section 3.2. This overloading is confusing; one of the two should be renamed (e.g., use G for the group).
  2. [Section 6.2.3] The phrase 'with ε replaced by ε2' appears to intend ε² (or a square), but the notation is ambiguous. Please clarify the scaling of ε throughout the proof.
  3. [Remark 3.14] The sentence 'In order to avoid making the paper longer we have chosen not to present this result here' is unusual for a main theorem's proof; if the proof is not included, it should at least be summarized in an appendix.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: the main LDPs are derived from model primitives via external variational representations, and the cited lifting lemmas are independent, not self-referential.

full rationale

The paper's central claims—Theorems 3.3, 3.5, and 3.11—are derived from the Markovian edge-flip model with prescribed rate kernels beta^{n,+-}. The rate functions J and I are explicit functionals of the limiting kernels beta^{+-}, with no fitted parameters and no quantity defined in terms of the target result. The upper bound uses the stochastic-control variational representation cited from [16,17]; this is an external, established theorem, not a restatement of the paper's conclusions, and the same representation is applied to the original process rather than to a modified process engineered to reproduce the rate function. The lower bound constructs tilted measures whose relative entropy is computed directly and shown to converge to J, an independent verification rather than a definitional shortcut. The cut-norm lifting follows the Chatterjee–Varadhan strategy via [20,25], and the invoked Lemma 6.3 is said to adapt an external lemma from [25]; even though its proof is omitted, an omitted proof is a completeness gap, not circularity. The homogeneity simplification in Corollary 3.6 uses concavity of sqrt(x(1-x)) and an inequality that is verified in the text; it is not an ansatz smuggled in by citation. No fitted-input-called-prediction pattern appears, and the few self-citations (e.g., [8,14,16,17]) are either background references or standard external tools with stated assumptions that do not include the target LDP. The unsupported abstract claim about the absence of a nontrivial path-space LDP is a verifiability issue, not a circular one. The analysis is therefore self-contained with respect to circularity concerns.

Assumptions & free parameters 0 free parameters · 8 assumptions · 0 invented entities

The results are conditional on the stated Assumptions 2.1, 2.2, 3.17, 3.20 about the rate kernels and interaction functions. No free parameters are fitted and no new entities are postulated. The main tools are standard weak-convergence and variational-representation methods from [16,17] and graphon and regularity techniques from [20,42].

assumptions (8)
  • domain assumption Assumption 2.1: block rate kernels beta^{n,+-} converge in L1 to beta+-, are uniformly bounded below by c_beta > 0, and are left-continuous and bounded in t.
    Standing assumption controlling the rate model; used throughout Sections 4-6.
  • domain assumption Assumption 2.2(i): beta^{n,+-} and beta+- lie in L^{1+eta} and converge in L^{1+eta}.
    Used to prove the LDP in the weak topology of Theorem 3.3 and 3.5.
  • domain assumption Assumption 2.2(ii): sup_s || beta^{n,+-}_s - beta+-_s ||_{L1([0,1]^2)} tends to 0.
    Needed for the cut-norm LDP in Theorem 3.11.
  • domain assumption Assumption 3.17: intrinsic dynamics F and interaction D are bounded and Lipschitz in the L2 path.
    Ensures well-posedness of the continuum equation (3.5).
  • domain assumption Assumption 3.20: the approximations F^n,D^n converge to F,D in a uniform-in-compact sense and F,D are weakly continuous.
    Used in the contraction-principle argument for the dynamical system LDP.
  • standard math Weak convergence method and variational representation of Budhiraja-Dupuis [16,17].
    Tool for the LDP upper bound via controlled processes.
  • standard math Chatterjee-Varadhan approach and Szemeredi regularity lemma for lifting weak-topology LDP to cut-norm topology [20,25].
    Basis of Section 6.
  • standard math Graphon limit theory: equivalence of cut metric and homomorphism densities [42].
    Used for equivalence classes and the functional t(bh,F).

how reviews work

0 comments
Cite this review

Pith. "Pith review of Large Deviations for Markovian Graphon Processes and Associated Dynamical Systems on Networks." pith.science (2026). https://pith.science/paper/MKUW3SZ3

@misc{pith2026250608333,
  author       = {Pith},
  title        = {Pith review of: Large Deviations for Markovian Graphon Processes and Associated Dynamical Systems on Networks},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/MKUW3SZ3}},
  note         = {Machine review of arXiv:2506.08333}
}
abstract

We consider temporal models of rapidly evolving Markovian networks whose edge-formation and dissolution rates are determined by time-dependent spatial kernels. Equivalently, these may be viewed as Markovian networks with $O(1)$ jump rates observed over long time horizons. In this regime, paths of graphon-valued processes obtained by averaging over suitable moving time windows provide natural state descriptors. Under appropriate conditions on the jump-rate kernels, we establish laws of large numbers and large deviation principles for these window-averaged paths, both in the weak topology and in the cut metric. We also show that, without such local averaging, the rapidly oscillating graphon process does not satisfy a nontrivial path-space LDP. The resulting rate functions admit explicit and tractable representations, distinct from those arising in static random graph models and finite-horizon dynamic graph models. We further analyze the associated variational problems in several examples and apply the graphon LDP to node-valent dynamical systems driven by the evolving network.

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

60 extracted references · 60 canonical work pages

  1. [25]

    Dhara and S

    S. Dhara and S. Sen, Large deviation for uniform graphs with given degrees, The Annals of Applied Probability 32(3) (2022), 2327–2353

    work page 2022

  2. [1]

    Albert and A.-L

    R. Albert and A.-L. Barab ´asi, Statistical mechanics of complex networks, Reviews of modern physics 74 (2002), no. 1, 47

    work page 2002

  3. [2]

    Neural Fields and Noise-Induced Patterns in Neurons on Large Disordered Networks

    D. Avitabile and J. MacLaurin, Neural fields and noise-induced patterns in neurons on large disordered networks, arXiv preprint arXiv:2408.12540 (2024)

    work page Pith review arXiv 2024

  4. [3]

    Baladron, D

    J. Baladron, D. Fasoli, O. Faugeras, and J. Touboul, Mean-field description and propagation of chaos in networks of Hodgkin–Huxley and FitzHugh–Nagumo neurons, The Journal of Mathematical Neuroscience2 (2012), no. 1, 1–50

    work page 2012

  5. [4]

    S Bassett and O

    D. S Bassett and O. Sporns, Network neuroscience, Nature neuroscience20 (2017), no. 3, 353–364

    work page 2017

  6. [5]

    Bayraktar, S

    E. Bayraktar, S. Chakraborty, and R. Wu, Graphon mean field systems, The Annals of Applied Probability33 (2023), no. 5, 3587–3619

    work page 2023

  7. [6]

    Berner, T

    R. Berner, T. Gross, C. Kuehn, J. Kurths, and S. Yanchuk,Adaptive dynamical networks, Physics Reports1031 (2023), 1–59

    work page 2023

  8. [7]

    G. Bet, F. Coppini, and F. R. Nardi,Weakly interacting oscillators on dense random graphs, Journal of Applied Probability 61 (2024), no. 1, 255–278

    work page 2024

Show all 60 references
  1. [8]

    Bhamidi, A

    S. Bhamidi, A. Budhiraja, and R. Wu, Weakly interacting particle systems on inhomogeneous random graphs, Stochastic Processes and their Applications129 (2019), no. 6, 2174–2206

    2019

  2. [9]

    Boccaletti, P

    S. Boccaletti, P. De Lellis, C. Del Genio, K. Alfaro-Bittner, R. Criado, S. Jalan, and M. Romance, The structure and dynamics of networks with higher order interactions, Physics Reports1018 (2023), 1–64

    2023

  3. [10]

    Borgs, J

    C. Borgs, J. Chayes, L. Lova ´sz, V. T. S´os, and K. Vesztergombi, Counting graph homomorphisms, Topics in discrete mathematics. Algorithms and Combinatorics26 (2006), 315–371

    2006

  4. [11]

    Borgs, J

    C. Borgs, J. Chayes, L. Lova ´sz, V. T. S´os, and K. Vesztergombi, Convergent sequences of dense graphs. i. subgraph fre- quencies, metric properties and testing, Advances in Mathematics219(6) (2008), 1801–1851

    2008

  5. [12]

    Borgs, J

    C. Borgs, J. Chayes, L. Lova´sz, V. T. S´os, and K. Vesztergombi,Convergent sequences of dense graphs. ii. multiway cuts and statistical physics, Annals of Mathematics(2)176(1) (2012), 151–219

    2012

  6. [13]

    Boucheron, G

    S. Boucheron, G. Lugosi, and P. Massart, Concentration inequalities : A nonasymptotic theory of independence, Oxford University Press, 2013

    2013

  7. [14]

    Braunsteins, F

    P. Braunsteins, F. D. Hollander, and M. Mandjes,A sample-path large deviation principle for dynamic Erd˝os-R´enyi random graphs, The Annals of Applied Probability33(4) (2023), 3278–3320

    2023

  8. [15]

    C Bressloff and J

    P. C Bressloff and J. M Newby, Path integrals and large deviations in stochastic hybrid systems, Physical Review E 89 (2014), no. 4, 042701

    2014

  9. [16]

    Budhiraja and P

    A. Budhiraja and P. Dupuis,Analysis and Approximation of Rare Events : Representations and Weak Convergence Methods, Springer, New York, USA, 2019

    2019

  10. [17]

    Budhiraja, P

    A. Budhiraja, P. Dupuis, and V. Maroulas, Variational representations for continuous time processes, Annales de l’IHP Probabilit´es et statistiques47 (2011), no. 3, 725–747

    2011

  11. [18]

    Chatterjee, Large deviations for random graphs, Springer International Publishing, ´Ecole d’ ´Et´e de probabilit´es de saint-flour xlv - 2015 edition, 2017

    S. Chatterjee, Large deviations for random graphs, Springer International Publishing, ´Ecole d’ ´Et´e de probabilit´es de saint-flour xlv - 2015 edition, 2017

    2015

  12. [19]

    Chatterjee and P

    S. Chatterjee and P. Diaconis, Estimating and understanding exponential random graph models, The Annals of Statistics 41 (2013), no. 5, 2428–2461

    2013

  13. [20]

    Chatterjee and S

    S. Chatterjee and S. R. S. Varadhan, The large deviation principle for the Erd˝os-R´enyi random graph, European Journal of Combinatorics 32(7) (2011), 1000–1017

    2011

  14. [21]

    Chiba and G

    H. Chiba and G. S Medvedev, The mean field analysis of the kuramoto model on graphs i. the mean field equation and transition point formulas, Discrete and Continuous Dynamical Systems39 (2019), no. 1, 131–155

    2019

  15. [22]

    Dembo and L

    A. Dembo and L. Lubetzky, A large deviation principle for the Erd˝os-R´enyi uniform random graph, Electronic Commu- nications in Probability 23 (2018), 1–13

    2018

  16. [23]

    Dembo and O

    A. Dembo and O. Zeitouni, Large deviations techniques and applications, Vol. 38, Springer Science & Business Media, 2009

    2009

  17. [24]

    den Hollander,Large deviations, Fields Institute Monographs, vol

    F. den Hollander,Large deviations, Fields Institute Monographs, vol. 14, American Mathematical Society, Providence, RI, 2000. MR1739680

    2000

  18. [26]

    Dupuis and G

    P. Dupuis and G. S. Medvedev, The large deviation principle for interacting dynamical systems on random graphs, Com- munications in mathematical physica390 (2022), 545–575

    2022

  19. [27]

    Durrett, Random graph dynamics, Cambridge Series in Statistical and Probabilistic Mathematics, vol

    R. Durrett, Random graph dynamics, Cambridge Series in Statistical and Probabilistic Mathematics, vol. 20, Cam- bridge University Press, Cambridge, 2007. MR2271734 LDP FOR GRAPHON PROCESSES AND ASSOCIATED DYNAMICAL SYSTEMS 55

    2007

  20. [28]

    Eswaran, C

    D. Eswaran, C. Faloutsos, S. Guha, and N. Mishra,Spotlight: Detecting anomalies in streaming graphs, Proceedings of the 24-th ACM SIGKDD international conference on knowledge discovery & data mining, 2018, pp. 1378–1386

    2018

  21. [29]

    Faugeras, J

    O. Faugeras, J. Touboul, and B. Cessac,A constructive mean-field analysis of multi-population neural networks with random synaptic weights and stochastic inputs, Frontiers in Computational Neuroscience3 (2009), no. 1

    2009

  22. [30]

    Hanneke, W

    S. Hanneke, W. Fu, and E. P. Xing,Discrete temporal models of social networks, Electronic Journal of Statistics4 (2010), 585–605

    2010

  23. [31]

    P. W. Holland and S. Leinhardt, A dynamic model for social networks , Journal of Mathematical Sociology 5 (1977), 5–20

    1977

  24. [32]

    Holme and J

    P. Holme and J. Saram¨aki, Temporal networks, Physics reports519 (2012), no. 3, 97–125

    2012

  25. [33]

    Janson, Graphons, cut norm and distance, couplings and rearrangements, New York Journal of Mathematics Mono- graphs, Volume 4, 2013

    S. Janson, Graphons, cut norm and distance, couplings and rearrangements, New York Journal of Mathematics Mono- graphs, Volume 4, 2013

    2013

  26. [34]

    Jiang, P

    M. Jiang, P. Cui, A. Beutel, C. Faloutsos, and S. Yang,Catching synchronized behaviors in large networks: A graph mining approach, ACM Transactions on Knowledge Discovery from Data (TKDD)10 (2016), no. 4, 1–27

    2016

  27. [35]

    V. N. Kolokoltsov, Nonlinear Markov processes and kinetic equations, Vol. 182, Cambridge University Press, 2010

    2010

  28. [36]

    Kovanen, M

    L. Kovanen, M. Karsai, K. Kaski, J. Kert´esz, and J. Saram¨aki, Temporal motifs, Temporal networks (2013), 119–133

    2013

  29. [37]

    Lacker, K

    D. Lacker, K. Ramanan, and R. Wu, Locally interacting diffusions as markov random fields on path space , Stochastic Processes and their Applications140 (2021), 81–114

    2021

  30. [38]

    Lacker, K

    D. Lacker, K. Ramanan, and R. Wu, Local weak convergence for sparse networks of interacting processes, The Annals of Applied Probability 33 (2023), no. 2, 843–888

    2023

  31. [39]

    K. H. Lee, L. Xue, and D. R. Hunter,Model-based clustering of time-evolving networks through temporal exponential-family random graph models, Journal of Multivariate Analysis175 (2020), 104540

    2020

  32. [40]

    A. Li, S. P Cornelius, Y-Y Liu, L. Wang, and A-L Barab´asi, The fundamental advantages of temporal networks, Science 358 (2017), no. 6366, 1042–1046

    2017

  33. [41]

    Lov ´asz and B

    L. Lov ´asz and B. Szegedy,Limits of dense graph sequences, Journal of Combinatorial Theory, Series B96 (2006), 933– 957

    2006

  34. [42]

    Lov ´asz, Large networks and graph Limits, American Mathematical Society, USA, 2012

    L. Lov ´asz, Large networks and graph Limits, American Mathematical Society, USA, 2012

    2012

  35. [43]

    Markering, The large deviation principle for inhomogeneous Erd˝os-R´enyi random graphs, Journal of Theoretical Prob- ability 36 (2023), 711–727

    M. Markering, The large deviation principle for inhomogeneous Erd˝os-R´enyi random graphs, Journal of Theoretical Prob- ability 36 (2023), 711–727

    2023

  36. [44]

    S Medvedev, The nonlinear heat equation on dense graphs and graph limits, SIAM Journal on Mathematical Analysis 46 (2014), no

    G. S Medvedev, The nonlinear heat equation on dense graphs and graph limits, SIAM Journal on Mathematical Analysis 46 (2014), no. 4, 2743–2766

    2014

  37. [45]

    S Medvedev, The nonlinear heat equation on w-random graphs, Archive for Rational Mechanics and Analysis 212 (2014), no

    G. S Medvedev, The nonlinear heat equation on w-random graphs, Archive for Rational Mechanics and Analysis 212 (2014), no. 3, 781–803

    2014

  38. [46]

    S Medvedev, The continuum limit of the kuramoto model on sparse random graphs, Communications in mathematical sciences 17 (2019), no

    G. S Medvedev, The continuum limit of the kuramoto model on sparse random graphs, Communications in mathematical sciences 17 (2019), no. 4, 883–898

    2019

  39. [47]

    Newman, Networks: an introduction, Oxford University Press, 2010

    M. Newman, Networks: an introduction, Oxford University Press, 2010

    2010

  40. [48]

    M. E. J. Newman, The structure and function of complex networks, SIAM review (2003), 167–256

    2003

  41. [49]

    Paranjape, A

    A. Paranjape, A. R Benson, and J. Leskovec,Motifs in temporal networks, Proceedings of the tenth ACM international conference on web search and data mining, 2017, pp. 601–610

    2017

  42. [50]

    M. I. Roberts and B. S ¸eng¨ul, Exceptional times of the critical dynamical Erd ˝os–R´enyi graph, The Annals of Applied Probability 28 (2018), no. 4, 2275 –2308

    2018

  43. [51]

    G. L. Robins and P. E. Pattinson,Random graph models for temporal processes in social networks, Journal of Mathematical Sociology 25 (2001), 5–41

    2001

  44. [52]

    G. L. Robins and P. E. Pattinson, Interdependencies and social processes: Generalized dependence structures , Models and Methods in Social Network Analysis, 2005, pp. 192–214

    2005

  45. [53]

    T. A. B. Snijders,Statistical methods for network dynamics, Proceedings of the XLIII Scientific Meeting, Italian Statistical Society, 2006, pp. 281–296

    2006

  46. [54]

    Sznitman, Topics in propagation of chaos, Ecole d’Et´e de Probabilit´es de Saint-Flour XIX—1989, 1991, pp

    A-S. Sznitman, Topics in propagation of chaos, Ecole d’Et´e de Probabilit´es de Saint-Flour XIX—1989, 1991, pp. 165– 251

    1989

  47. [55]

    van der Hofstad, Random graphs and complex networks

    R. van der Hofstad, Random graphs and complex networks. Vol. 1 , Cambridge Series in Statistical and Probabilistic Mathematics, vol. [43], Cambridge University Press, Cambridge, 2017. MR3617364

    2017

  48. [56]

    Van Der Hofstad, Random graphs and complex networks, Vol

    R. Van Der Hofstad, Random graphs and complex networks, Vol. 2, Cambridge university press, 2024

    2024

  49. [57]

    S. R. S. Varadhan,Large deviations, Courant Lecture Notes in Mathematics, vol. 27, Courant Institute of Mathematical Sciences, New York; American Mathematical Society, Providence, RI, 2016. MR3561097

    2016

  50. [58]

    von Brecht, T

    J. von Brecht, T. Kolokolnikov, A. L Bertozzi, and H. Sun,Swarming on random graphs, Journal of Statistical Physics 151 (2013), no. 1-2, 150–173

    2013

  51. [59]

    H von Brecht, B

    J. H von Brecht, B. Sudakov, and A. L Bertozzi,Swarming on random graphs ii, Journal of Statistical Physics158 (2015), no. 3, 699–734

    2015

  52. [60]

    Zhang, C

    X. Zhang, C. Moore, and M. E. J. Newman,Random graph models for dynamic networks, The European Physics Journal B 90 (2017), no. 10, 200

    2017

Pith tools

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