Pith. sign in

REVIEW 2 major objections 5 minor 24 references

On one-dimensional Cluster cluster model

T0 review · 2 major / 5 minor · reviewed 2026-08-06 · deepseek-v4-flash

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

Pith's one-line read The paper proves that in one dimension the origin's cluster grows as $t^{1/(\alpha+2)}$ for $\alpha>-2$, with an exact limiting law at $\alpha=0$.

desk verdict First rigorous treatment of the 1D cluster-cluster model; growth exponents and blow-up likely survive, but the exact scaling limit has a real CLT normalization error and an unjustified reduction. read the letter →

arxiv 2507.03552 v1 pith:6QJH2JEB submitted 2025-07-04 math.PR math-phmath.MP

Noam Berger , Eviatar B. Procaccia , Daniel Sharon This is my paper · ORCID
classification math.PRmath-phmath.MP MSC 60K3560F05
keywords cluster-clusteraggregationdiffusion-limitedrandomwalkscalinglimitgrowthexponentgelationone-dimensionalinteractingparticlesystem
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 gives the first rigorous mathematical treatment of the one-dimensional cluster-cluster aggregation model, a 1984 physics model for flocculation in which clusters perform random walks and merge on contact, with move rate $|C|^{-\alpha}$. It proves that for $\alpha>-2$ the cluster containing the origin has size of order $t^{1/(\alpha+2)}$, while for $\alpha<-2$ an infinite cluster appears in finite time almost surely. For the equal-rate case $\alpha=0$, it derives an exact limiting distribution for $|C_0(t)|/\sqrt{t}$, with a closed-form CDF in terms of the standard normal CDF and density that depends only on the initial particle density $p$. A sympathetic reader should care because this settles the basic scaling question for a long-studied model and provides a benchmark for the physics literature.

What carries the argument

The load-bearing object is the difference process of two tagged particles. Before two clusters merge, each cluster center performs a rate-1 simple random walk, so the difference of the two original particles' positions is a rate-2 symmetric random walk whose initial value is a sum of geometric gaps with parameter $p$. Connection events are read off as this difference hitting a level, which reduces the $\alpha=0$ cluster-size CDF to running-minimum estimates for a single random walk. To go beyond $\alpha=0$, the paper uses the time-change identity $u(t)=\int_0^t |C(s)|^{-\alpha}ds$: the step times of a cluster with size-dependent rate are arrival times of a rate-1 Poisson process in this random clock, and Lemma 1 converts size-dependent-rate probabilities into comparisons with sums of exponential variables. The lower bound for negative $\alpha$ is carried by a ladder of target clusters $C_{n t^{\gamma_n}}$ with $\gamma_{n+1}=(1-\alpha\gamma_n)/2$, chosen so that the speed-up from small clusters balances the scale.

What would settle it

Simulate the $\alpha=0$ model on a long interval with $p=1/2$, record $|C_0(t)|/\sqrt{t}$ at large $t$, and compare the empirical distribution with the CDF in Theorem 1; a visibly different shape or a scaling exponent other than $1/2$ would falsify the difference-walk reduction. For $\alpha=1$, the same simulation should show median cluster size tracking $t^{1/3}$.

Watch

Extended reading notes

Core claim

The central claim is a phase transition in the growth exponent. Fix $d=1$ and $p\in(0,1)$. For $\alpha\ge 0$, Theorem 2 gives matching upper and lower power-law bounds: with probability at least $1-\varepsilon$, $c^{-1}t^{1/(\alpha+2)}\le |C_0(t)|\le c t^{1/(\alpha+2)}$ for large $t$. For $-2<\alpha<0$, Theorem 3 gives the same upper bound and a lower bound of order $t^{1/(\alpha+2)-\delta}$. For $\alpha=0$, Theorem 1 (proved as Theorem 4) gives the exact weak limit of $|C_0(t)|/\sqrt{t}$, namely the CDF $2\Phi(x(\eta-1)/2)-\frac{x(\eta-1)}{\sqrt{2\pi}}e^{-x^2(\eta-1)^2/8}-1$ with $\eta=1/p$. For $\alpha<-2$, equation (4) and Corollary 2.1 state that the origin's cluster is almost surely infinite in finite time, so the process is not well defined beyond that time. The paper also constructs the process for $-2<\alpha<0$ by coupling truncated versions.

Load-bearing premise

The exact $\alpha=0$ limit rests on treating a connection between the origin's cluster and the $(n+1)$-st particle as the difference of two original particles' positions hitting a level, even though the two clusters could merge through intermediate clusters; if that reduction is wrong, the explicit limiting law in Theorem 1 collapses.

Editorial extensions

If this is right

  • For every $\alpha>-2$ in $d=1$, the cluster-size exponent $1/(\alpha+2)$ is now rigorous, so simulations of the model can be tested against a proven power law.
  • At $\alpha=0$, the explicit limiting law gives asymptotic moments and quantiles of the cluster size; in particular, typical cluster size grows like $\sqrt{t}$ with a prefactor determined by $p$.
  • For $\alpha<-2$, the model almost surely develops an infinite cluster in finite time; Corollary 2.1 states the process is not well defined afterward.
  • The critical value $\alpha=-2$ is identified as the phase boundary, and the paper leaves its behavior as an open problem.

Reading between the lines

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

  • Editorial: the same difference-walk reduction may yield growth exponents in higher dimensions near the conjectured critical value $\alpha=-1$, but the paper only states this as a conjecture.
  • Editorial: the $\alpha=0$ closed-form law suggests a full hydrodynamic limit for the empirical cluster-size distribution in one dimension; proving such a limit would be a natural next step not taken in the paper.
  • Editorial: a direct check of the paper's mechanism would be to test whether, in the $\alpha=0$ process, the merger time between the origin's cluster and the $n$-th particle is well approximated by the hitting time of the difference random walk; the paper treats these as equal.
  • Editorial: the phase boundary at $\alpha=-2$ may be connected to a logarithmically corrected growth law, since the theorem gives divergence of the exponent as $\alpha\downarrow -2$; this is speculation, not a paper claim.
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 / 5 minor

Summary. The paper studies the one-dimensional cluster-cluster aggregation model introduced by Meakin et al. Each site of Z is occupied with probability p, and the set of occupied sites is partitioned into clusters; each cluster moves as a continuous-time simple random walk with rate |C|^{-alpha}, and two clusters that become adjacent merge. The paper's main results are: for alpha=0, a claimed exact limiting distribution for |C0(t)|/sqrt(t); for alpha>-2, high-probability upper and lower bounds on |C0(t)| of order t^{1/(alpha+2)} (with a delta-loss in the lower bound for -2<alpha<0); and for alpha<-2, finite-time blowup with probability tending to 1, from which the authors conclude the process is not well defined in that regime. The proofs are based on random-walk estimates, couplings between different alpha, and large-deviation bounds.

Significance. If correct, these results would be the first rigorous growth exponents and the first exact scaling limit for the cluster-cluster model in one dimension. The model is natural and the paper connects it to random-walk theory in a interesting way. A particular strength is that no parameters are fitted: the derivations start from the model and use standard tools such as the local central limit theorem, Hoeffding and Bernstein inequalities, and couplings. The growth-exponent theorems appear robust to the issues discussed below. However, the exact alpha=0 limit in Theorem 1 is derived from a specific CLT normalization that is incorrect, so the stated formula is not justified and must be corrected before the exact-result claim can be accepted.

major comments (2)
  1. [Section 4, eqs. (14)-(18)] The scaling limit for the rate-2 difference random walk is incorrectly normalized. A continuous-time rate-2 simple random walk S_t has variance 2t, so the correct CLT scaling is S_t/sqrt(2t) => N(0,1). The proof uses S_t/(2sqrt(t)) throughout, which has variance 1/2. This introduces a spurious factor of sqrt(2) in the arguments of Phi and in the exponential rate: the quantity n(eta-1)/(2sqrt(t)) should be n(eta-1)/sqrt(2t). The error propagates to eqs. (17)-(18) and to the final formula in Theorem 1/4. The exact distribution is therefore not justified as stated.
  2. [Remark 2.1] The density formula claimed in Remark 2.1 is not the derivative of the CDF in Theorem 1. Differentiating F(x)=2Phi(cx)- (2cx/sqrt(2pi)) e^{-c^2 x^2/2} - 1 with c=(eta-1)/2 gives f(x)= (2c^3/sqrt(2pi)) x^2 e^{-c^2 x^2/2}, not gamma x^2/sqrt(2pi) e^{-gamma x^2/2} with gamma=c^2. The coefficient is off by a factor 2c = eta-1. This internal inconsistency must be resolved together with the normalization issue.
minor comments (5)
  1. [Abstract and Theorem 3] The abstract states that for alpha>-2 the cluster size is of order t^{1/(alpha+2)}, but Theorem 3 only gives a lower bound of t^{1/(alpha+2)-delta} for -2<alpha<0. The abstract should be qualified to match the theorem.
  2. [Theorem 4] Theorem 4 says 'for all c>0' but the statement involves x; it should be 'for all x>0'.
  3. [Section 3] The definition of the '0th particle' as the particle closest to 0 is ambiguous when there are particles at both -1 and 1 and no particle at 0; a tie-breaking rule should be specified.
  4. [Section 5.1] The notation S_b^a in the proof of Theorem 5 is not defined. It would greatly help the reader to state explicitly whether this is a number of steps, a position, or a sum of step-time variables, and to relate it to the variables introduced in Section 3.
  5. [Throughout] There are several typographical errors, including 'and and' in the first sentence, 'preform', 'sulotions', and 'partally'; these should be corrected in a revision.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: the alpha=0 scaling limit and the t^(1/(alpha+2)) exponents are derived from the model definition with standard external tools; self-citations are context, technique provenance, and open-problem pointers only.

full rationale

The derivation chain is self-contained and no step reduces to its own input. Theorem 4/Theorem 1 derives the alpha=0 scaling limit from the model mechanics via the exact combinatorial identities (12)-(13), which express P(|C0(t)| = n+1) in terms of connection-probability differences Delta_n, followed by a first-principles estimate of P(0 t<-> n) as a hitting probability of the rate-2 difference walk, evaluated with Hoeffding, the reflection principle, and the external LCLT of [7]; no parameter is fitted and no limiting law is assumed in the course of computing it. Theorems 2-3 derive the exponent 1/(alpha+2) rather than importing it: Lemma 1's time-change bounds (Eqs. (5)-(10)) and the recursion gamma_1 = 1/2, gamma_{n+1} = (1 - alpha gamma_n)/2, whose fixed point is 1/(alpha+2), produce matching upper and lower bounds in Sections 5 and 6, and the critical alpha = -2 case is explicitly left open. Self-citations are peripheral, not load-bearing: [1] and [18] are introduction benchmarks on unrelated DLA-type growth rates; [2] is explicitly labeled 'Work in Progress' and cited only for the open problems; and [15], [17] are credited for the discrepancy-index coupling idea, but Theorem 8's proof (Eqs. (65)-(75)) reproduces the coupling in full detail rather than deferring to the citations. No uniqueness theorem from the authors' prior work is invoked, no ansatz is adopted by citation, and no known empirical pattern is renamed: the inserted phrase 'not a function of Cn' in Section 6.1 is a technical condition consistent with the later choice a_n = g_epsilon ln n, and the acknowledgments of open problems are limitations, not disguised inputs. The skeptic's CLT-normalization and Remark 2.1 density-mismatch concerns are mathematical-correctness issues (Eq. (16) normalizes the rate-2 walk by 2 sqrt(t) rather than sqrt(2t)), but they concern whether the printed formula is correct, not whether it was assumed; no fitted input is relabeled as a prediction anywhere in the paper.

Assumptions & free parameters 0 free parameters · 6 assumptions · 0 invented entities

The paper's central claims rest on standard probability theorems and the assumption that the model dynamics can be studied via random-walk hitting estimates. No free parameters are introduced and no new entities are postulated. The main hidden assumptions are the validity of the reduction in Theorem 1 and the coupling constructions.

assumptions (6)
  • standard math Local central limit theorem and Brownian approximations for rate-2 random walks
    Used in Section 4, equation (16), to evaluate hitting probabilities of the difference process.
  • standard math Hoeffding's inequality for sums of geometric random variables
    Used in Section 4, equation (15), and Section 5 for concentration of initial gaps.
  • standard math Time-change representation of non-homogeneous Poisson processes (Daley-Vere-Jones)
    Used in Section 3 to relate step times of clusters with size-dependent rates to unit-rate Poisson processes.
  • standard math Bernstein's inequality for sums of exponential variables
    Used in Section 6 in the lower and upper bound proofs.
  • standard math Coupling of processes with bounded rates to construct limits
    Used in Theorem 8 to construct the process for -2 < alpha < 0 via a limiting argument.
  • standard math Reflection principle for random walks
    Used in Section 4, equation (16), to estimate hitting probabilities.

how reviews work

0 comments
Cite this review

Pith. "Pith review of On one-dimensional Cluster cluster model." pith.science (2026). https://pith.science/paper/6QJH2JEB

@misc{pith2026250703552,
  author       = {Pith},
  title        = {Pith review of: On one-dimensional Cluster cluster model},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/6QJH2JEB}},
  note         = {Machine review of arXiv:2507.03552}
}
abstract

The Cluster-cluster model was introduced by Meakin et al in 1984. Each $x\in \mathbb{Z}^d$ starts with a cluster of size 1 with probability $p \in (0,1]$ independently. Each cluster $C$ performs a continuous-time SRW with rate $|C|^{-\alpha}$. If it attempts to move to a vertex occupied by another cluster, it does not move, and instead the two clusters connect via a new edge. Focusing on dimension $d=1$, we show that for $\alpha>-2$, at time $t$, the cluster size is of order $t^\frac{1}{\alpha + 2}$, and for $\alpha < -2$ we get an infinite cluster in finite time a.s. Additionally, for $\alpha = 0$ we show convergence in distribution of the scaling limit.

Figures

Figures reproduced from arXiv: 2507.03552 by the authors.

Figure 1
Figure 1. Cluster-cluster model on a torus with α = 1. (1) A0 is samples from a product measure with marginal distribution P(A0(v) = 1) = p. The edge set is empty at initiation, E0 = ∅. (2) Every cluster C has a Poisson clock at rate |C | −α. (3) • (d ≥ 2): Upon ringing a cluster attempts to preform a random walk move (all vertices and edges of the cluster move together). If destination is vacant, then cluster preforms the mo… view at source ↗
Figure 2
Figure 2. Example of move step and connection step Note that in dimension 1, the boundaries of clusters are always of size 2, thus when a cluster moves, it can encounter at most one other cluster. Thus, one does not need a mechanism to choose connections such as in d > 1. In this paper, we focus on the one-dimensional variant of the model, and provide sharp growth bounds for the size of a single cluster. 2. Results and Open P… view at source ↗
Figure 3
Figure 3. The clusters involved in the proof of the upper bound for α ≥ 0 The idea of the proof is to consider only slowing down the cluster after time intervals of t 2 and t 2+α time, as this gives us an upper bound on the size. By random walk properties, we get that |C0(t 2 )| ≈ t. After applying the slowdown to the clusters, we can see that it takes roughly t 2+α to perform the same amount of steps as in the initial t 2 ti… view at source ↗
Figures from the paper (3 more)
Figure 4
Figure 4. Figure 4: The clusters involved in the proof of the lower bound for α ≤ 0 We first show that under the above coupling, if for A 0 we have 0 t ↔ 2ct, then for A α we have |C0(t)| ≥ ct. We assume by contradiction that at time t we have 0 t ↔ 2ct and |C0(t)| < ct. This means that t…
Figure 5
Figure 5. Figure 5: The relevant clusters for the upper bound for −2 < α < 0 6.2. upper bound. Lemma 2. Let M > 0, −2 < α < 0 and consider the process A M t having rates upper bounded by M. ∀ε > 0, exists ∃c > 0, T > 0 such that ∀t > T (46) P  |C0(t)| ≥ ct 1 α+2  ≤ ε , where c = cα log …
Figure 6
Figure 6. Figure 6: The relevant cases for the bounds. The arrows show who needs to walk more steps, while the slash shows that a connection was not created [PITH_FULL_IMAGE:figures/full_fig_p015_6.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

24 extracted references · 19 canonical work pages

  1. [1]

    Berger, E

    N. Berger, E. B. Procaccia, and A. Turner. Growth of stationary hastings–levitov. The Annals of Applied Probability, 32(5):3331–3360, 2022

    work page 2022

  2. [2]

    Procaccia, Dominik Schmid, and Daniel Sharon

    Noam Berger, Eviatar B. Procaccia, Dominik Schmid, and Daniel Sharon. The cluster cluster model in dimension 2 and higher. Work in Progress

    work page

  3. [3]

    D. J. Daley and D. Vere-Jones. An introduction to the theory of point processes. Vol. I. Probability and its Applications (New York). Springer-Verlag, New York, second edition, 2003. Elementary theory and methods

    work page 2003

  4. [4]

    The critical one-dimensional multi-particle dla.arXiv preprint arXiv:2009.02761, 2020

    Dor Elboim, Danny Nam, and Allan Sly. The critical one-dimensional multi-particle dla.arXiv preprint arXiv:2009.02761, 2020

    arXiv 2009

  5. [5]

    Protein aggregation processes: In search of the mechanism.Protein Science, 16(11):2334– 2344, 2007

    Carl Frieden. Protein aggregation processes: In search of the mechanism.Protein Science, 16(11):2334– 2344, 2007

    2007

  6. [6]

    H. Kesten. How long are the arms in dla? Journal of Physics A: Mathematical and General, 20(1):L29, 1987

    work page 1987

  7. [7]

    Random walk: a modern introduction, volume 123

    Gregory F Lawler and Vlada Limic. Random walk: a modern introduction, volume 123. Cambridge University Press, 2010

    work page 2010

  8. [8]

    Scaling theory and exactly solved models in the kinetics of irreversible aggregation

    Francois Leyvraz. Scaling theory and exactly solved models in the kinetics of irreversible aggregation. Physics Reports, 383(2-3):95–212, 2003

    2003

Show all 24 references
  1. [9]

    How long are the arms in dbm? Communications in Mathematical Physics, 406(4):1–16, 2025

    Ilya Losev and Stanislav Smirnov. How long are the arms in dbm? Communications in Mathematical Physics, 406(4):1–16, 2025

    2025

  2. [10]

    Gelation in coagulating systems

    Alex A Lushnikov. Gelation in coagulating systems. Physica D: Nonlinear Phenomena, 222(1-2):37–53, 2006

    2006

  3. [11]

    Coagulation in finite systems

    Alexei A Lushnikov. Coagulation in finite systems. Journal of Colloid and interface science, 65(2):276– 285, 1978

    1978

  4. [12]

    Diffusion-limited aggregation in three dimensions: results from a new cluster-cluster aggregation model

    Paul Meakin. Diffusion-limited aggregation in three dimensions: results from a new cluster-cluster aggregation model. Journal of colloid and interface science, 102(2):491–504, 1984

    1984

  5. [13]

    Effects of cluster trajectories on cluster-cluster aggregation: A comparison of linear and brownian trajectories in two-and three-dimensional simulations

    Paul Meakin. Effects of cluster trajectories on cluster-cluster aggregation: A comparison of linear and brownian trajectories in two-and three-dimensional simulations. Physical Review A, 29(2):997, 1984

    1984

  6. [14]

    Dynamic cluster-size distribution in cluster-cluster aggregation: Effects of cluster diffusivity

    Paul Meakin, Tam´ as Vicsek, and Fereydoon Family. Dynamic cluster-size distribution in cluster-cluster aggregation: Effects of cluster diffusivity. Physical Review B, 31(1):564, 1985

    1985

  7. [15]

    Y. Mu, E. B. Procaccia, and Y. Zhang. Scaling limit of dla on a long line segment. Transactions of the American Mathematical Society, 2022

    2022

  8. [16]

    Lehrbuch der allgemeinen Chemie, volume 1

    Wilhelm Ostwald. Lehrbuch der allgemeinen Chemie, volume 1. W. Engelmann, 1903

    1903

  9. [17]

    E. B. Procaccia, J. Ye, and Y. Zhang. Stationary dla is well defined. Journal of Statistical Physics, 181:1089–1111, 2020

    2020

  10. [18]

    Dimension of diffusion-limited aggregates grown on a line

    Eviatar B Procaccia and Itamar Procaccia. Dimension of diffusion-limited aggregates grown on a line. Physical Review E, 103(2):L020101, 2021

    2021

  11. [19]

    Exact calculation of the probabilities of rare events in cluster-cluster aggregation

    R Rajesh, V Subashri, and Oleg Zaboronski. Exact calculation of the probabilities of rare events in cluster-cluster aggregation. Physical Review Letters, 133(9):097101, 2024

    2024

  12. [20]

    Erd˝ os-renyi random graphs forest fires self-organized criticality

    Balazs Rath and Balint Toth. Erd˝ os-renyi random graphs forest fires self-organized criticality. Elec- tronic Journal of Probability [electronic only], 14:1290–1327, 2009

    2009

  13. [21]

    One-dimensional multi-particle dla–a pde approach

    Vladas Sidoravicius and Balazs Rath. One-dimensional multi-particle dla–a pde approach. arXiv preprint arXiv:1709.00484, 2017

    2017 arXiv

  14. [22]

    On one-dimensional multi-particle diffusion limited aggregation

    Allan Sly. On one-dimensional multi-particle diffusion limited aggregation. In In and Out of Equilib- rium 3: Celebrating Vladas Sidoravicius, pages 755–774. Springer, 2020

    2020

  15. [23]

    Versuch einer mathematischen theorie der koagulationskinetik kolloider l¨ osungen

    M v Smoluchowski. Versuch einer mathematischen theorie der koagulationskinetik kolloider l¨ osungen. Zeitschrift f¨ ur physikalische Chemie, 92(1):129–168, 1918

    1918

  16. [24]

    The theory of ostwald ripening

    Peter W Voorhees. The theory of ostwald ripening. Journal of Statistical Physics, 38:231–252, 1985. 20 NOAM BERGER, EVIATAR B. PROCACCIA, AND DANIEL SHARON (Noam Berger) Department of mathematics, TUM URL: https://www.math.cit.tum.de/en/probability/people/berger/ Email address...

    1985

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.