Pith. sign in

REVIEW 2 major objections 4 minor 15 references

Infinite collisions of simple random walks on random recursive trees generated by Bernoulli sequences

T0 review · 2 major / 4 minor · reviewed 2026-07-12 · grok-4.5

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

Pith's one-line read A Bernoulli recursive tree has one end and two random walks collide infinitely often.

desk verdict Clean one-end theorem and explicit trunk/branch geometry for a new Bernoulli recursive tree; the infinite-collision claim rests on an incomplete transfer from a comb embedding. read the letter →

arxiv 2607.02916 v1 pith:2EAPOZV2 submitted 2026-07-03 math.PR

Jia-shan Tang , Feng Wang , Xian-Yuan Wu This is my paper · ORCID
classification math.PR MSC 60J1005C81
keywords randomrecursivetreetopologicalendinfinitecollisionpropertyBernoullisequencesimplewalkcombgraph
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 constructs an infinite random tree by a simple Bernoulli rule: each new vertex attaches either to the last vertex or the second-last one. It proves that this tree almost surely has exactly one infinite simple path (one topological end) and that two independent simple random walks started on it meet at the same vertex at the same time infinitely often. The infinite-collision property is a stronger, non-monotonic refinement of recurrence: it asks whether particles can interact forever. Because the construction is elementary and the tree can be mapped onto a one-sided comb whose tooth lengths are controlled by the same Bernoulli sequence, the result supplies a concrete new family of graphs where meetings never stop.

What carries the argument

The projection map Φ that sends the unique trunk of T onto the non-negative integers and each finite branch onto a vertical tooth, producing a one-sided comb Comb*(Z,f) whose tooth lengths f(i) are the branch lengths of T; the infinite-collision criterion for such combs then transfers back to T.

What would settle it

Exhibit a positive-probability set of Bernoulli sequences for which the projected comb has summable reciprocal tooth lengths, or construct an explicit pair of walks on T that meet only finitely often with positive probability.

Watch

Extended reading notes

Core claim

The random recursive tree T generated by a Bernoulli sequence almost surely has exactly one topological end, and two independent simple random walks on T collide infinitely often almost surely.

Load-bearing premise

The claim that the natural projection from the tree onto a one-sided comb preserves the infinite-collision property of simple random walks, so that a known comb criterion can be applied directly.

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

2 major / 4 minor

Summary. The paper constructs an infinite random recursive tree T by attaching each new vertex either to the most recent vertex (probability p) or the second-most recent (probability q=1-p). Theorem 1.1 asserts that T has almost surely exactly one topological end; the short proof uses a parity argument on the underlying Bernoulli sequence and shows that two ends would force an infinite terminal run of zeros, an event of probability zero. Theorem 1.2 asserts that T has the infinite-collision property for two independent simple random walks. The argument maps T onto a one-sided comb Comb*(Z,f) via a projection Φ that sends the unique trunk to the non-negative integers and each finite branch to a vertical tooth of length L_i, verifies that the reciprocal-branch-length sum diverges almost surely (two elementary proofs M1 and M2), and invokes the Chen–Chen criterion for infinite collisions on such combs.

Significance. The model is a simple, explicitly constructible random tree whose geometry is completely determined by a Bernoulli sequence; the almost-sure uniqueness of the topological end is clean and of independent interest. Establishing the infinite-collision property for this family would enlarge the short list of non-transitive recurrent graphs known to possess the property and would illustrate how the Chen–Chen comb criterion can be applied beyond deterministic combs. The two divergence proofs (record lengths and large-deviation control of longest zero runs) are elementary and self-contained. The reduction step itself, however, is incomplete, so the main claim remains conditional on a missing transfer argument.

major comments (2)
  1. Section 3.2–3.3 and Lemma 3.3: the projection Φ is many-to-one and does not preserve degrees or transition probabilities of simple random walk (a trunk vertex of degree 2+1_{L_i>0} is sent to a comb vertex of degree 2 or 3). The paper only sketches that horizontal projections behave like reflecting walks and that first-meeting times of the projected walks dominate those of the original walks (display (3.4)). No coupling, no Green-function comparison, and no argument that infinite collisions on the image lift to infinite collisions on T are supplied. Without such a transfer the reduction to the Chen–Chen criterion is incomplete, and Theorem 1.2 is not proved.
  2. The same gap appears when the authors claim that Lemma 3.2 (originally stated for two-sided combs) extends to one-sided Comb*(Z,f). The short comparison of stopping times τ_N and heta_dN is written for the absolute-value map of a two-sided walk; it is not verified that the same domination holds for the reflecting walk that actually lives on Comb*(Z,f). A rigorous justification of this extension is needed before the criterion can be applied to Φ(T).
minor comments (4)
  1. Page 4, line after Definition 2.2: the Bernoulli sequence is introduced with values {1,0} but later written with values {1,-1}; the two conventions should be reconciled.
  2. Equation (3.1): the transition probabilities for the trunk indices σ_i are derived correctly, yet the subsequent claim that they are independent of i is used without explicit statement; a one-line remark would help the reader.
  3. References [5] and [8] are listed as arXiv preprints without final publication data; if they have appeared, the journal citations should be updated.
  4. The abstract and the last paragraph of the introduction both assert that T has the infinite-collision property; once the transfer gap is closed these statements will be accurate, but until then they overstate the proved content.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: theorems proved from first principles on Bernoulli sequences; external comb criterion is independently verified, not self-defined.

full rationale

The derivation chain is self-contained and non-circular. Theorem 1.1 follows directly from the recursive construction: each vertex has at most two chances to produce descendants, so at most two ends; two ends would force an infinite terminal run of b_n=0, which has probability zero by the Bernoulli product measure. Theorem 1.2 maps T to a one-sided comb Comb*(Z,f) via the explicit projection Φ of Lemma 3.3 (trunk to base line, branches to vertical teeth of lengths L_i whose law is computed from the same Bernoulli sequence in (3.2)), then verifies the Chen–Chen divergence hypothesis ∑ 1/f̃(n)=∞ by two independent arguments (M1 via successive record lengths of branches, M2 via large-deviation control on longest runs of zeros). The only external input is the already-published sufficient condition of Chen & Chen [4] (Lemma 3.2), whose hypotheses are checked on the image graph without fitting parameters or redefining the target. No quantity is defined in terms of the infinite-collision conclusion, no uniqueness theorem is imported from the authors’ prior work, and no ansatz is smuggled via self-citation. The reader’s and skeptic’s concerns about whether Φ preserves SRW transitions are correctness/transfer issues, not circularity; they do not make any step reduce to its own input by construction. Score 0 is therefore the honest finding.

Assumptions & free parameters 1 free parameters · 3 assumptions · 0 invented entities

The paper works entirely inside classical discrete probability. The only free parameter is the attachment probability p ∈ (0,1), which is part of the model definition rather than a fitted constant. All other ingredients are either standard mathematical facts or previously published lemmas that are cited and whose hypotheses are checked.

free parameters (1)
  • attachment probability p
    Fixed but arbitrary parameter of the recursive construction; the statements hold for every p ∈ (0,1). Not fitted to data.
assumptions (3)
  • standard math Definition of topological ends of a tree as the number of infinite simple paths from a fixed root (Pemantle).
    Used in Definition 2.2 and Theorem 1.1; classical.
  • domain assumption Chen & Chen sufficient condition: if ∑ 1/f̃(n) = ∞ then Comb(Z,f) (and its one-sided version) has the infinite-collision property.
    Invoked as Lemma 3.2; the paper verifies the hypothesis for the random f arising from the Bernoulli tree.
  • standard math Large-deviation asymptotics for the longest run of successes in i.i.d. Bernoulli trials (Mao–Wang–Wu).
    Cited as Theorem 1.1 of [15] and used in the second proof of divergence.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Infinite collisions of simple random walks on random recursive trees generated by Bernoulli sequences." pith.science (2026). https://pith.science/paper/2EAPOZV2

@misc{pith2026260702916,
  author       = {Pith},
  title        = {Pith review of: Infinite collisions of simple random walks on random recursive trees generated by Bernoulli sequences},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/2EAPOZV2}},
  note         = {Machine review of arXiv:2607.02916}
}
abstract

In this paper, we study random recursive trees generated by Bernoulli sequences. Starting from a graph with two vertices and one edge, each new vertex is connected to the last vertex with probability $ p $, or to the second-last vertex with probability $ q = 1-p $, this recursive construction yields a random infinite recursive tree $T$. We prove that $T$ almost surely has exactly one topological end. Furthermore, we establish that $T$ has the infinite collision property: two independent simple random walks on $T$ collide infinitely often almost surely.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

15 extracted references · 7 linked inside Pith

  1. [1]

    Krishnapur and Y

    M. Krishnapur and Y. Peres, Recurrent graphs where two independent ran- dom walks collide finitely often,Electronic Communications in Probability 9(2004), 72–81

    2004

  2. [2]

    M. T. Barlow, Y. Peres and P. Sousi, Collisions of random walks,Probability Surveys8(2011), 1–59. Available atarXiv:1003.3255

    arXiv 2011

  3. [3]

    Chen and D

    X. Chen and D. Chen, Two random walks on the open cluster ofZ 2 meet infinitely often,Science China Mathematics53(2010), 1971–1978

    2010

  4. [4]

    Chen and D

    X. Chen and D. Chen, Some sufficient conditions for infinite collisions of simple random walks on a wedge comb,Electronic Journal of Probability 16(2011), 1341–1355

    2011

  5. [5]

    Hutchcroft and Y

    T. Hutchcroft and Y. Peres, Collisions of random walks in reversible ran- dom graphs, Available atarXiv:1507.02974

    arXiv

  6. [6]

    J. F. Richey, Collisions of random walks and related diffusions, Available athttps://jfrichey.github.io/pagedocs/rw collisions.pdf(2018)

    2018

  7. [7]

    Halberstam and T

    N. Halberstam and T. Hutchcroft, Collisions of Random Walks in Dynamic Random Environments,Electronic Journal of Probability27(2022), no. 10, 1–30. Available atarXiv:2009.13951

    arXiv 2022

  8. [8]

    Watanabe, Infinite collision property for the three-dimensional uniform spanning tree, Available atarXiv:2301.08547

    S. Watanabe, Infinite collision property for the three-dimensional uniform spanning tree, Available atarXiv:2301.08547

    arXiv

Show all 15 references
  1. [9]

    Astoquillca, On the stationary measures of two variants of the voter model, Available atarXiv:2409.16064

    J. Astoquillca, On the stationary measures of two variants of the voter model, Available atarXiv:2409.16064

    arXiv

  2. [10]

    De Ambroggio, M

    U. De Ambroggio, M. Nitzschner and C. Scali, Collisions of random walks on comb graphs with a planar base, Available atarXiv:2508.19814

    arXiv

  3. [11]

    D. A. Croydon, D. Shiraishi and S. Watanabe, Collision properties of the four-dimensional random walk trace, Available atarXiv:2605.30755. 12

    arXiv

  4. [12]

    Freudenthal, Neuaufbau Der Endentheorie,Annals of Mathematics 43(1942), no

    H. Freudenthal, Neuaufbau Der Endentheorie,Annals of Mathematics 43(1942), no. 2, 261–279

    1942

  5. [13]

    Pemantle, Choosing a spanning tree for the integer lattice uniformly, Annals of Probability29(2001), no

    R. Pemantle, Choosing a spanning tree for the integer lattice uniformly, Annals of Probability29(2001), no. 3, 1559–1574

    2001

  6. [14]

    J. A. Arredondo, Sa´ ul Quispe, and Camilo Ram´ ırez Maluendas, On the fiber product of noncompact Riemann surfaces,Science51(2026), no. 1, 127–144

    2026

  7. [15]

    Y.-H. Mao, F. Wang and X.-Y. Wu, Large deviation behavior for the longest head run in an i.i.d. Bernoulli sequence,Journal of Theoretical Probability28(2015), no. 1, 259–268. School of Mathematical Sciences, Capital Normal University, Beijing, 100048, People R China. Email:152...

    2015

Pith tools

Reviewed July 12, 2026 · model on record in the stance chip above.