pith. sign in

arxiv: 2604.20984 · v1 · submitted 2026-04-22 · 🧮 math.DS · math.PR

Graphon Limits of Graph Reaction--Diffusion Equations

Edith J. Zhang , James Scott , Qiang Du This is my paper

Pith reviewed 2026-05-09 22:53 UTC · model grok-4.3

classification 🧮 math.DS math.PR
keywords graphonreaction-diffusioncut normgraph limitsnonlocal equationsstochastic processesconvergencebirth-death processes
0
0 comments X

The pith

Reaction-diffusion equations on graphs converge in L^p to a limiting graphon equation as graphs approach the graphon in cut norm.

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

The paper proves that when a sequence of graphs converges in cut norm to a graphon, the solutions of the graph reaction-diffusion equations on those graphs converge in every L^p norm to the solution of a nonlocal reaction-diffusion equation defined on the graphon. It further shows that a stochastic particle process consisting of random walks and birth-death dynamics on the graphs converges in probability to the same graphon equation. This limit bridges discrete network models to continuous nonlocal PDEs, allowing analysis of large-scale dynamics without solving increasingly large systems of ODEs. A reader would care because many models in chemistry, biology, and social science use graphs that grow large, and the graphon provides a tractable continuum description.

Core claim

For any sequence of graphs converging in cut norm to a graphon, the corresponding graph reaction-diffusion systems converge in L^p for all p in [1, infinity] to the graphon reaction-diffusion equation obtained by replacing the graph adjacency with the graphon integral operator. The same convergence in probability holds for the associated stochastic particle systems that combine diffusion via random walk with birth-death reactions.

What carries the argument

Cut-norm convergence of graphs to a graphon, which replaces the discrete diffusion operator with a nonlocal integral operator against the graphon kernel.

If this is right

  • Dynamics on very large finite graphs can be approximated by solving the single graphon equation instead of the full discrete system.
  • The deterministic graphon equation is the law of large numbers limit for the stochastic particle processes on the approximating graphs.
  • Well-posedness results for the graphon equation transfer to the graph equations for all sufficiently large graphs in the sequence.
  • Pattern formation or traveling waves observed on large networks can be studied by analyzing the corresponding graphon equation.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The same cut-norm limit technique could apply to other evolution equations on graphs, such as those with nonlinear diffusion or higher-order interactions.
  • Numerical schemes that discretize the graphon equation might provide efficient approximations for dynamics on real-world networks sampled from the graphon.
  • Extensions to time-varying graphons or to convergence under weaker notions such as graphon convergence in other metrics could be examined.

Load-bearing premise

The graphs converge to the graphon in cut norm, and the reaction functions satisfy the regularity needed for the equations to be well-posed.

What would settle it

Construct a sequence of graphs converging in cut norm to a graphon for which the graph RD solutions fail to converge in some L^p norm, for example by exhibiting persistent oscillations or blow-up not present in the graphon equation.

Figures

Figures reproduced from arXiv: 2604.20984 by Edith J. Zhang, James Scott, Qiang Du.

Figure 1
Figure 1. Figure 1: Existing results for graph diffusion equation Going beyond pure diffusion equations, [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Reaction–diffusion analogues A more statistical mechanics–based perspective of this problem is seen in [17] in which the RWBD dynamics are called Glauber–Kawasaki dynamics. The density of these dynamics, with a special choice of birth- and death-rates, are shown to converge to the solution of the Allen–Cahn equation, which is a special case of RD equations. In a previous paper [36], we proved that sequence… view at source ↗
Figure 3
Figure 3. Figure 3: is reproduced from our previous paper [36]. 1 2 4 3 (a) The 4-cycle graph.   0 1 0 1 1 0 1 0 0 1 0 1 1 0 1 0   (b) Its associated adjacency matrix. 1 1 2 2 3 3 4 4 (c) Its corresponding graphon [PITH_FULL_IMAGE:figures/full_fig_p006_3.png] view at source ↗
read the original abstract

A graph reaction--diffusion (RD) equation is a system of differential equations that is defined on the nodes of a graph. Consider a sequence of growing graphs that converges in cut norm to a limiting graphon. We show that the solutions of the sequence of graph RD equations converge in $L^p$ norm, for $p \in [1,\infty]$, to the solution of a limiting nonlocal RD equation, which we call a graphon RD equation. Furthermore, we show a large numbers result for a stochastic particle process that consists of a random walk and a birth-death process on graphs. For a sequence of graphs that converge in cut norm to a limiting graphon, the sequence of stochastic processes converges in probability to the solution of the graphon RD equation.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

1 major / 1 minor

Summary. The manuscript proves that if a sequence of graphs G_n converges in cut norm to a graphon W, then the solutions u_n of the associated graph reaction-diffusion equations converge in L^p([0,1]) for all p in [1,∞] to the solution u of the limiting nonlocal graphon RD equation. It further establishes that the empirical measures of a stochastic particle process (random walk plus birth-death) on the graphs converge in probability to the same graphon RD solution.

Significance. If the proofs are complete and the requisite regularity on the reaction terms is supplied, the results supply a rigorous continuum limit for both deterministic and stochastic RD dynamics on large graphs. This is a useful contribution to the theory of graphon limits in dynamical systems, as it handles the nonlinear reaction terms and provides both L^p and probabilistic convergence statements.

major comments (1)
  1. [Abstract and main theorems (e.g., the statement of L^p convergence)] The abstract and main convergence statements do not record explicit hypotheses on the reaction functions (e.g., local Lipschitz continuity, linear growth, or one-sided Lipschitz conditions). Standard Gronwall estimates for continuous dependence in L^p and passage to the limit in the semilinear term require at least one of these; without them the claimed L^p convergence can fail for superlinear reactions that produce blow-up on some graphs but not others. This assumption is load-bearing for both the deterministic and stochastic claims.
minor comments (1)
  1. [Introduction] Notation for the graphon RD equation (the precise form of the nonlocal diffusion operator) should be displayed explicitly in the introduction for immediate comparison with the discrete graph equations.

Simulated Author's Rebuttal

1 responses · 0 unresolved

We thank the referee for the positive assessment of the manuscript and for identifying this important point regarding the presentation of hypotheses. We address the major comment below.

read point-by-point responses

  1. Referee: [Abstract and main theorems (e.g., the statement of L^p convergence)] The abstract and main convergence statements do not record explicit hypotheses on the reaction functions (e.g., local Lipschitz continuity, linear growth, or one-sided Lipschitz conditions). Standard Gronwall estimates for continuous dependence in L^p and passage to the limit in the semilinear term require at least one of these; without them the claimed L^p convergence can fail for superlinear reactions that produce blow-up on some graphs but not others. This assumption is load-bearing for both the deterministic and stochastic claims.

    Authors: We agree that the hypotheses on the reaction functions must be stated explicitly in the abstract and main theorem statements to make the results rigorous and to preclude counterexamples involving blow-up. The manuscript assumes local Lipschitz continuity together with linear growth on the reaction terms (to guarantee global existence, uniqueness, and the applicability of Gronwall estimates in L^p); these conditions appear in the setup of Section 2 and are used throughout the proofs. However, they were not recorded in the abstract or the formal statements of the main theorems. We will revise the abstract, the introduction, and the statements of Theorems 1.1, 1.2, and 1.3 to list these assumptions explicitly. This change will also clarify the scope of both the deterministic L^p convergence and the stochastic large-numbers result. revision: yes

Circularity Check

0 steps flagged

No circularity: standard limit theorem under external graph convergence hypothesis

full rationale

The derivation is a convergence result: given ||G_n - W||_□ → 0, solutions u_n of the graph RD system converge in L^p to the solution u of the graphon RD equation. This relies on the cut-norm convergence assumption (external to the paper's own equations) plus standard well-posedness hypotheses on the reaction terms. No equation or step reduces by construction to a fitted parameter, self-definition, or self-citation chain; the limit passage uses standard Gronwall estimates and integral-operator continuity after the graphon limit is taken. The result is self-contained against external benchmarks and does not rename known patterns or smuggle ansatzes via self-citation.

Axiom & Free-Parameter Ledger

0 free parameters · 2 axioms · 0 invented entities

The central claim rests on the cut-norm convergence assumption for the graph sequence and standard well-posedness conditions for the RD equations; no free parameters or new entities are introduced in the abstract.

axioms (2)
  • domain assumption Sequence of graphs converges in cut norm to a limiting graphon
    Explicitly stated as the hypothesis enabling the limit results.
  • domain assumption Reaction terms permit unique solutions to the graph and graphon RD equations
    Implicit requirement for the convergence statements to make sense.

pith-pipeline@v0.9.0 · 5421 in / 1343 out tokens · 26111 ms · 2026-05-09T22:53:46.158891+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

37 extracted references · 37 canonical work pages

  1. [1]

    Pattern formation on networks with reactions: A continuous-time random-walk approach

    Christopher N Angstmann, Isaac C Donnelly, and Bruce I Henry. “Pattern formation on networks with reactions: A continuous-time random-walk approach”. In:Physical Review E—Statistical, Nonlinear, and Soft Matter Physics87.3 (2013), p. 032804

    work page 2013

  2. [2]

    The University of Wisconsin–Madison, 1987

    Douglas James Blount.Comparison of a stochastic model of a chemical reaction with diffusion and the deterministic model. The University of Wisconsin–Madison, 1987

    work page 1987

  3. [3]

    Metrics for sparse graphs

    B´ ela Bollob´ as and Oliver Riordan. “Metrics for sparse graphs”. In:Surveys in Combinatorics

    work page

  4. [4]

    by Sophie Huczynska, James D

    Ed. by Sophie Huczynska, James D. Mitchell, and Colva M. Roney-Dougal. London Mathematical Society Lecture Note Series. Cambridge, UK: Cambridge University Press, 2009, pp. 211–288.doi:10.1017/CBO9781107325975.009

    work page doi:10.1017/cbo9781107325975.009 2009

  5. [5]

    AnL p theory of sparse graph convergence I: Limits, sparse random graph models, and power law distributions

    Christian Borgs, Jennifer Chayes, Henry Cohn, and Yufei Zhao. “AnL p theory of sparse graph convergence I: Limits, sparse random graph models, and power law distributions”. In: Transactions of the American Mathematical Society372.5 (2019), pp. 3019–3062

    work page 2019

  6. [6]

    AnL p theory of sparse graph convergence II: LD convergence, quotients and right convergence

    Christian Borgs, Jennifer T. Chayes, Henry Cohn, and Yufei Zhao. “AnL p theory of sparse graph convergence II: LD convergence, quotients and right convergence”. In:The Annals of Probability46.1 (2018), pp. 337–396

    work page 2018

  7. [7]

    Pattern formation in random networks using graphons

    Jason Bramburger and Matt Holzer. “Pattern formation in random networks using graphons”. In:SIAM Journal on Mathematical Analysis55.3 (2023), pp. 2150–2185

    work page 2023

  8. [8]

    Nicholas F Britton.Reaction-diffusion equations and their applications to biology.1987

    work page 1987

  9. [9]

    John Wiley & Sons, 2004

    Robert Stephen Cantrell and Chris Cosner.Spatial ecology via reaction-diffusion equations. John Wiley & Sons, 2004

    work page 2004

  10. [10]

    Reaction-diffusion processes and metapopulation models in heterogeneous networks

    Vittoria Colizza, Romualdo Pastor-Satorras, and Alessandro Vespignani. “Reaction-diffusion processes and metapopulation models in heterogeneous networks”. In:Nature Physics3 (2007), pp. 276–282

    work page 2007

  11. [11]

    Reaction-diffusion equations for in- teracting particle systems

    Anna De Masi, Pablo A Ferrari, and Joel L Lebowitz. “Reaction-diffusion equations for in- teracting particle systems”. In:Journal of Statistical Physics44.3 (1986), pp. 589–644

    work page 1986

  12. [12]

    SIAM, 2019

    Qiang Du.Nonlocal Modeling, Analysis, and Computation. SIAM, 2019

    work page 2019

  13. [13]

    Multiscale modeling, homogenization and nonlocal effects: Mathematical and computational issues

    Qiang Du, Bjorn Engquist, and Xiaochuan Tian. “Multiscale modeling, homogenization and nonlocal effects: Mathematical and computational issues”. In:arXiv preprint arXiv:1909.00708 (2019)

    work page arXiv 1909

  14. [14]

    Analysis and approximation of the Ginzburg–Landau model of superconductivity

    Qiang Du, Max D Gunzburger, and Janet S Peterson. “Analysis and approximation of the Ginzburg–Landau model of superconductivity”. In:Siam Review34.1 (1992), pp. 54–81

    work page 1992

  15. [15]

    Maximum bound principles for a class of semilinear parabolic equations and exponential time-differencing schemes

    Qiang Du, Lili Ju, Xiao Li, and Zhonghua Qiao. “Maximum bound principles for a class of semilinear parabolic equations and exponential time-differencing schemes”. In:SIAM Review 63.2 (2021), pp. 317–359. 26 REFERENCES

    work page 2021

  16. [16]

    Cambridge University Press, 2023

    Imad El Bouchairi, Jalal Fadili, Yosra Hafiene, and Abderrahim Elmoataz.Nonlocal Contin- uum Limits ofp-Laplacian Problems on Graphs. Cambridge University Press, 2023

    work page 2023

  17. [17]

    Klaus-Jochen Engel, Rainer Nagel, and Simon Brendle.One-parameter semigroups for linear evolution equations. Vol. 194. Springer, 2000

    work page 2000

  18. [18]

    Motion by mean curvature from Glauber–Kawasaki dynamics

    Tadahisa Funaki and Kenkichi Tsunoda. “Motion by mean curvature from Glauber–Kawasaki dynamics”. In:Journal of Statistical Physics177 (2019), pp. 183–208

    work page 2019

  19. [19]

    Graph-based nonlocal gradient systems and their local limits

    Georg Heinze. “Graph-based nonlocal gradient systems and their local limits”. In: (2024)

    work page 2024

  20. [20]

    Svante Janson.Graphons, Cut Norm and Distance, Couplings and Rearrangements. Vol. 4. NYJM Monographs. SUNY at Albany, 2013

    work page 2013

  21. [21]

    A submartingale type inequality with applications to stochastic evolution equations

    Peter Kotelenez. “A submartingale type inequality with applications to stochastic evolution equations”. In:Stochastics: An International Journal of Probability and Stochastic Processes 8.2 (1982), pp. 139–151

    work page 1982

  22. [22]

    L´ aszl´ o Lov´ asz.Large networks and graph limits. Vol. 60. American Mathematical Society, 2012

    work page 2012

  23. [23]

    Szemer´ edi’s lemma for the analyst

    L´ aszl´ o Lov´ asz and Bal´ azs Szegedy. “Szemer´ edi’s lemma for the analyst”. In:GAFA Geometric And Functional Analysis17 (2007), pp. 252–270

    work page 2007

  24. [24]

    Convergence of the graph Allen–Cahn scheme

    Xiyang Luo and Andrea L Bertozzi. “Convergence of the graph Allen–Cahn scheme”. In: Journal of Statistical Physics167 (2017), pp. 934–958

    work page 2017

  25. [25]

    The nonlinear heat equation on dense graphs and graph limits

    Georgi S Medvedev. “The nonlinear heat equation on dense graphs and graph limits”. In: SIAM Journal on Mathematical Analysis46.4 (2014), pp. 2743–2766

    work page 2014

  26. [26]

    Laplacian matrices of graphs: a survey

    Russell Merris. “Laplacian matrices of graphs: a survey”. In:Linear Algebra and its Applica- tions197 (1994), pp. 143–176

    work page 1994

  27. [27]

    James R Norris.Markov chains. 2. Cambridge University Press, 1998

    work page 1998

  28. [28]

    Springer Science & Business Media, 2013

    Bernt Oksendal.Stochastic differential equations: an introduction with applications. Springer Science & Business Media, 2013

    work page 2013

  29. [29]

    Amnon Pazy.Semigroups of linear operators and applications to partial differential equations. Vol. 44. Springer Science & Business Media, 2012

    work page 2012

  30. [30]

    Random walks on dense graphs and graphons

    Julien Petit, Renaud Lambiotte, and Timoteo Carletti. “Random walks on dense graphs and graphons”. In:SIAM Journal on Applied Mathematics81.6 (2021), pp. 2323–2345

    work page 2021

  31. [31]

    Emulating cellular automata in chemical reaction– diffusion networks

    Dominic Scalise and Rebecca Schulman. “Emulating cellular automata in chemical reaction– diffusion networks”. In:Natural Computing15 (2016), pp. 197–214

    work page 2016

  32. [32]

    Gamma-convergence of gradient flows on Hilbert and metric spaces and applications

    Sylvia Serfaty. “Gamma-convergence of gradient flows on Hilbert and metric spaces and applications”. In:Discrete and Continuous Dynamical Systems31.4 (2011), pp. 1427–1451

    work page 2011

  33. [33]

    A theory of pattern formation for reaction–diffusion systems on temporal networks

    Robert A Van Gorder. “A theory of pattern formation for reaction–diffusion systems on temporal networks”. In:Proceedings of the Royal Society A477.2247 (2021), pp. 1–28

    work page 2021

  34. [34]

    Cross-diffusion and pattern formation in reaction– diffusion systems

    Vladimir K Vanag and Irving R Epstein. “Cross-diffusion and pattern formation in reaction– diffusion systems”. In:Physical Chemistry Chemical Physics11.6 (2009), pp. 897–912

    work page 2009

  35. [35]

    Continuum limit of nonlocal diffusion on inhomogeneous networks

    Itsuki Watanabe. “Continuum limit of nonlocal diffusion on inhomogeneous networks”. In: Journal of Dynamics and Differential Equations(2022), pp. 1–20

    work page 2022

  36. [36]

    Internal observation systems and a theory of reaction-diffusion equation on a graph

    Hideo Yuasa and Masami Ito. “Internal observation systems and a theory of reaction-diffusion equation on a graph”. In:1998 IEEE International Conference on Systems, Man, and Cyber- netics. Vol. 4. IEEE. 1998, pp. 3669–3673

    work page 1998

  37. [37]

    Ginzburg–Landau Functionals in the Large-Graph Limit

    Edith Zhang, James Scott, Qiang Du, and Mason A Porter. “Ginzburg–Landau Functionals in the Large-Graph Limit”. In:Pure and Applied Functional Analysis1 (2026), pp. 169–209

    work page 2026