Pith. sign in

REVIEW 4 minor 32 references

Collisions of random walks in unimodular random graphs: applications to the random geometric graph and long-range percolation

T0 review · 0 major / 4 minor · reviewed 2026-07-11 · grok-4.5

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

Pith's one-line read On unimodular random graphs, a Green-function moment decides whether two walks collide only finitely often, settling the voter model on geometric and long-range graphs.

desk verdict Clean finite-collision criterion for unimodular graphs plus complete voter-model classification on standard unbounded-degree geometric and long-range models; the stronger Green integrability is flagged by the authors and does not break the claims. read the letter →

arxiv 2607.04002 v1 pith:AWZKNDVX submitted 2026-07-04 math.PR

Jhon Astoquillca , Carlos Martinez-Arevalo This is my paper · ORCID
classification math.PR MSC 60K3560G5082C2205C81
keywords unimodularrandomgraphsrandom-walkcollisionsGreenfunctionvotermodelGilbertgraphDelaunaytriangulationlong-rangepercolationmasstransport
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 continues the study of when two independent simple random walks on a random graph meet only finitely many times or infinitely often. Earlier work already showed that recurrence plus finite mean degree at the root forces infinitely many collisions almost surely. The new result is the matching finite-collision theorem: if the expectation of the degree times the Green function at the root is finite, then the walks collide only finitely often almost surely (in both discrete and continuous time). Because many natural random geometric graphs have unbounded degrees, classical bounded-degree arguments do not apply; the unimodular mass-transport principle supplies the missing control. The authors verify the moment condition for the supercritical Gilbert graph, the Delaunay triangulation, the Gabriel graph, and selected regimes of long-range percolation. As a direct corollary they obtain a complete description of the extremal stationary measures of the voter model on each of these graphs: only the two consensus measures when collisions are infinite, and the whole one-parameter family of Bernoulli product measures when collisions are finite and the walk tail is trivial.

What carries the argument

Mass-transport principle applied to the expected number of collisions: the transport function f(G,u,v)=p_n(u,v)p_n(v,u)deg(v) converts the double sum of return probabilities into an expectation of deg(ρ)·G(ρ,ρ), which is finite by hypothesis and therefore forces the number of collisions to be finite a.s.

What would settle it

Construct (or disprove the existence of) a unimodular random rooted graph that is almost surely transient, has finite mean degree at the root, yet infinite expected degree-times-Green-function, and check whether two independent walks still collide only finitely often.

Watch

Extended reading notes

Core claim

If (G,ρ) is a unimodular random rooted graph satisfying E[deg(ρ)·∑_n p_n(ρ,ρ)] < ∞, then two independent simple random walks on G collide only finitely often almost surely, both in discrete and in continuous time. Together with the known infinite-collision statement under recurrence, this dichotomy completely determines the set of extremal stationary measures of the voter model on the graphs under study.

Load-bearing premise

The argument needs a finite expectation of degree times Green function at the root; mere almost-sure transience plus finite mean degree may not be enough, and the paper leaves open whether that weaker pair already implies finite collisions.

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

0 major / 4 minor

Summary. The paper studies discrete- and continuous-time collision properties of simple random walks on unimodular random rooted graphs. It recalls that recurrence plus E[deg( ho)]<\infty implies the infinite-collision property (Theorem 1, adapting Hutchcroft–Peres), and proves that the stronger integrability E[deg( ho)·G( ho, ho)]<\infty implies the finite-collision property almost surely (Theorem 2, via mass transport). These criteria are applied to the infinite components of the supercritical Gilbert graph, the Delaunay triangulation and the Gabriel graph in R^d (Theorem 3), and to long-range percolation clusters on Z^d in the regimes where an infinite cluster exists (Theorem 4). The collision properties, together with triviality of the random-walk tail ho-algebra, completely characterize the extremal stationary measures of the voter model on these graphs (Corollary 1).

Significance. If correct, the work supplies a clean, usable criterion for the finite-collision property on unbounded-degree unimodular graphs and thereby settles the structure of I_e for several natural geometric and long-range models that lie outside the classical bounded-degree or vertex-transitive setting. The mass-transport argument of Theorem 2 is short and transparent; the geometric applications rest on carefully transferred heat-kernel and resistance estimates (Barlow, Crawford–Sly, Rousselle) via an explicit comparison lemma (Lemma 9). The resulting complete description of stationary measures for the voter model is a concrete payoff that will be of interest both to interacting-particle-systems and to random-geometry communities.

minor comments (4)
  1. [Introduction] p. 2, after Theorem 2: the open question whether E[deg( ho)]<\infty plus a.s. transience already implies finite collisions is correctly flagged; a one-sentence remark that the present applications all verify the stronger moment would help the reader.
  2. [§3.3.2] Lemma 9 and the subsequent geometric constructions (Sections 3.3.3–3.3.4) are dense; a short schematic diagram of the good-box renormalization and the path heta oΘ would improve readability.
  3. [§2.3.2] Equation (11) equating continuous and discrete Green functions is standard but could be given a one-line reference or derivation for non-specialists.
  4. [§4.2] In the long-range section the restriction s otin[d+2,2d] is explained by the lack of quantitative tails on the random time T_x; a pointer to the available heat-kernel bounds of Can–Croydon–Kumagai would be useful even if they are not yet strong enough for the argument.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: Theorem 2 follows directly from mass transport under the stated Green-function integrability; applications verify that hypothesis via external heat-kernel/resistance estimates with no author overlap.

full rationale

The central finite-collision criterion (Theorem 2) is obtained in §2.3.2 by a short mass-transport argument: the expected number of collisions of two independent walks started at the root is bounded above by E[deg(ρ)·G(ρ,ρ)], which is finite by hypothesis, hence the number of collisions is a.s. finite. The same bound (via the continuous Green function identity (11)) yields the continuous finite-collision property. No parameter is fitted, no uniqueness theorem is imported from the authors’ prior work, and the argument does not reduce to a definitional identity. The infinite-collision companion (Theorem 1) is an adaptation of the external result of Hutchcroft–Peres (2015) under the weaker recurrence + E[deg(ρ)]<\infty hypothesis; the adaptation is spelled out and does not rely on self-citation. All applications (Gilbert/Delaunay/Gabriel via Lemmas 6–9 and the comparison scheme of Lemma 9; long-range percolation via Theorem 7 of Crawford–Sly) consist of verifying precisely the integrability hypothesis of Theorem 2 by means of heat-kernel upper bounds and effective-resistance tail estimates whose authors do not overlap with the present paper. The voter-model characterization (Theorem 5 / Corollary 1) is the classical duality statement of Liggett, merely restated in the form of the first author’s earlier note; it is applied after the collision properties have already been established and is not used to prove those properties. The only self-citations are therefore non-load-bearing. The derivation chain is consequently free of the six circularity patterns.

Assumptions & free parameters 0 free parameters · 5 assumptions · 0 invented entities

The paper rests on the mass-transport principle for unimodular graphs, classical electrical-network identities, and a collection of external heat-kernel/resistance estimates. No free parameters are fitted; the only ad-hoc elements are the concrete renormalization scales chosen large enough for domination, which are existential rather than numerical fits.

assumptions (5)
  • domain assumption Mass-Transport Principle for unimodular random rooted graphs (Aldous–Lyons)
    Used as the main tool in the proofs of Theorems 1 and 2 (Sections 2.3.1–2.3.2).
  • standard math Green function equals degree times effective resistance to infinity
    Equation (3); classical electrical-network identity used throughout Section 3.
  • domain assumption Heat-kernel upper bounds and tail controls on the random time after which they hold, for supercritical bond percolation (Barlow) and long-range percolation (Crawford–Sly)
    Proposition 3 and Theorem 7; imported to verify the Green-function moment in high dimensions.
  • domain assumption Recurrence of Delaunay and Gabriel graphs in d=2 (Rousselle)
    Cited for the infinite-collision regime; the authors supply their own proof only for the Gilbert graph.
  • domain assumption Existence and uniqueness of the infinite cluster for the listed long-range percolation regimes (Aizenman–Kesten–Newman, Berger)
    Background for Theorem 4; the paper conditions on the origin belonging to that cluster.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Collisions of random walks in unimodular random graphs: applications to the random geometric graph and long-range percolation." pith.science (2026). https://pith.science/paper/AWZKNDVX

@misc{pith2026260704002,
  author       = {Pith},
  title        = {Pith review of: Collisions of random walks in unimodular random graphs: applications to the random geometric graph and long-range percolation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/AWZKNDVX}},
  note         = {Machine review of arXiv:2607.04002}
}
read the original abstract

We study collision properties of simple random walks in unimodular random rooted graphs. This work continues the study initiated in~\cite{HutchcroftPeres2015}: under recurrence and an integrability condition on the root, two independent random walks collide infinitely often a.s. We prove that, under transience and an integrability assumption involving the Green function on the root, two independent random walks collide only finitely often a.s. We apply these results to several random graphs with unbounded degree: the Gilbert graph, the Delaunay graph, the Gabriel graph; and the long-range percolation model. We use these collision properties to characterize stationary measures of the voter model on these graphs.

Figures

Figures reproduced from arXiv: 2607.04002 by the authors.

Figure 1
Figure 1. Illustration of a corner of Cn+1 \Cn−1 partitioned into r-squares on the inner and outer boundaries of Cn. The number of edges in Πn with an endpoint in the red region is bounded from above by (number of vertices in the red region)×(number of vertices in the blue region). 3.3 Dimension 3 and higher In this section, we prove part 2 of Theorem 3, namely the finite collision property. The section is organized as follow… view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

32 extracted references · 1 linked inside Pith

  1. [1]

    Aizenman, H

    M. Aizenman, H. Kesten, and C. M. Newman. Uniqueness of the infinite cluster and con- tinuity of connectivity functions for short and long range percolation.Communications in Mathematical Physics, 111(4):505 – 531, 1987. 4

    1987

  2. [2]

    Aldous and R

    D. Aldous and R. Lyons. Processes on unimodular random networks.Electronic Journal of Probability, 12, 2007. 2.3, 2.3.1, 3.1, 3.1, 4

    2007

  3. [3]

    U. D. Ambroggio, M. Nitzschner, and C. Scali. Collisions of random walks on comb graphs with a planar base, 2025. arXiv preprint arXiv:2508.19814. 1

    arXiv 2025

  4. [4]

    Astoquillca

    J. Astoquillca. On the Stationary Measures of Two Variants of the Voter Model.Journal of Theoretical Probability, 39(34), 2026. 1

    2026

  5. [5]

    Barlow, Y

    M. Barlow, Y. Peres, and P. Sousi. Collisions of random walks.Ann. Inst. Henri Poincar´ e, Probab. et Stat., 48(no. 4):922–946, 2012. 1, 2.2

    2012

  6. [6]

    M. T. Barlow. Random walks on supercritical percolation clusters.The Annals of Probability,

  7. [7]

    Benjamini and N

    I. Benjamini and N. Curien. Ergodic theory on stationary random graphs.Electronic Journal of Probability, 17(93):1–20, 2012. 3, 2.3, 1, 2.3.3, 2, 2.3.3, 4

    2012

  8. [8]

    N. Berger. Transience, Recurrence and Critical Behavior for Long-Range Percolation.Com- munications in Mathematical Physics, 226:531–558, 2002. 4, 4.1

    2002

Show all 32 references
  1. [9]

    Billingsley.Convergence of Probability Measures

    P. Billingsley.Convergence of Probability Measures. Wiley-Interscience, 1968. 3.2.2

    1968

  2. [10]

    M. Biskup. Graph diameter in long-range percolation.Random Structures & Algorithms, 39(2):210–227, 2011. 4

    2011

  3. [11]

    Bonnet and N

    G. Bonnet and N. Chenavier. The maximal degree in a poisson–delaunay graph.Bernoulli, 26(2):948–979, 2020. 3.1

    2020

  4. [12]

    V. H. Can, D. A. Croydon, and T. Kumagai. Spectral dimension of simple random walk on a long-range percolation cluster.Electronic Journal of Probability, 27:1–37, 2022. 1, 4.2

    2022

  5. [13]

    D. Chen, B. Wei, and F. Zhang. A note on the finite collision property of random walks. Statistics&Probability Letters, 78(13):1742–1747, 2008. 1, 2.2

    2008

  6. [14]

    Chen and D

    X. Chen and D. Chen. Two random walks on the open cluster ofZ 2 meet infinitely often. Science China Mathematics, 53:1971–1978, 2010. 1

    1971

  7. [15]

    Crawford and A

    N. Crawford and A. Sly. Simple random walk on long range percolation clusters I: heat kernel bounds.Probab. Theory Relat. Fields, 154:753–786, 2012. 1, 1.1, 4.2

    2012

  8. [16]

    Friedrich, T

    T. Friedrich, T. Sauerwald, and A. Stauffer. Diameter and broadcast time of random geo- metric graphs in arbitrary dimensions.Algorithmica, 67(67):65–88, 2013. 3.3.3, 3

    2013

  9. [17]

    Grimmett.Percolation, volume 321 ofGrundlehren der mathematischen Wissenschaften

    G. Grimmett.Percolation, volume 321 ofGrundlehren der mathematischen Wissenschaften. Springer, Berlin, 2 edition, 1999. 3.3.2, 3.3.2

    1999

  10. [18]

    G. R. Grimmett, A. E. Holroyd, and G. Kozma. Percolation of finite clusters and infinite surfaces. InMathematical Proceedings of the Cambridge Philosophical Society, volume 156, pages 263–279. Cambridge University Press, 2014. 3.3.2

    2014

  11. [19]

    G. R. Grimmett, H. Kesten, and Y. Zhang. Random walk on the infinite cluster of the percolation model.Probability Theory and Related Fields, 96:33–44, 1993. 1 26

    1993

  12. [20]

    H¨ aggstr¨ om

    O. H¨ aggstr¨ om. Infinite clusters in dependent automorphism invariant percolation on trees. The Annals of Probability, 25(3):1423 – 1436, 1997. 2.3

    1997

  13. [21]

    Hutchcroft and Y

    T. Hutchcroft and Y. Peres. Collisions of random walks in reversible random graphs.Elec- tronic Communications in Probability, 20:1–6, 2015. (document), 1, 1, 2.2, 2.3.1, 2.3.1

    2015

  14. [22]

    Krishnapur and Y

    M. Krishnapur and Y. Peres. Recurrent Graphs where Two Independent Random Walks Collide Finitely Often.Electronic Communications in Probability, 9:72 – 81, 2004. 1

    2004

  15. [23]

    Last and M

    G. Last and M. Penrose.Lectures on the Poisson Process, volume 7 ofInstitute of Mathe- matical Statistics Textbooks. Cambridge University Press, Cambridge, 2017. 3, 3.1

    2017

  16. [24]

    T. M. Liggett.Interacting Particle Systems. Grundlehren der Mathematischen Wis- senschaften [Fundamental Principles of Mathematical Sciences]. Springer, New York, 1985. 1

    1985

  17. [25]

    T. M. Liggett, R. H. Schonmann, and A. M. Stacey. Domination by product measures.The Annals of Probability, 25(1):71–95, 1997. 3

    1997

  18. [26]

    Lyons and Y

    R. Lyons and Y. Peres.Probability on Trees and Networks. Cambridge Series in Statistical and Probabilistic Mathematics. Cambridge University Press, 2017. 2.1

    2017

  19. [27]

    Montgomery.Topics in Random Walks

    A. Montgomery.Topics in Random Walks. PhD thesis, University of Oregon, 2013. 2.2

    2013

  20. [28]

    Nachmias.Planar Maps, Random Walks and Circle Packing

    A. Nachmias.Planar Maps, Random Walks and Circle Packing. ´Ecole d’ ´Et´ e de Probabilit´ es de Saint-Flour XLVIII - 2018. Lecture Notes in Mathematics. Springer Cham, 2020. 2.1

    2018

  21. [29]

    Penrose.Random Geometric Graphs

    M. Penrose.Random Geometric Graphs. Oxford studies in probability. Oxford University Press, 2003. 3

    2003

  22. [30]

    L. P. R. Pimentel. On some fundamental aspects of polyominoes on random Voronoi tilings. Brazilian Journal of Probability and Statistics, 27(1):54 – 69, 2013. 3.4

    2013

  23. [31]

    Rousselle

    A. Rousselle. Recurrence or transience of random walks on random graphs generated by point processes inR d.Stochastic Processes and their Applications, 125(12):4351–4374, 2015. 3.2, 3.2.1, 3.3.2, 3.3.4

    2015

  24. [32]

    J. M. Swart.A Course in Interacting Particle Systems. Cambridge Series in Statistical and Probabilistic Mathematics. Cambridge University Press, 2026. 1 27

    2026

Pith tools

Reviewed July 11, 2026 · model on record in the stance chip above.