pith. sign in

arxiv: 2607.01946 · v1 · pith:GXJW2Y4Rnew · submitted 2026-07-02 · 🧮 math.PR

Second-order fluctuations for a phase transition in random partitions

Jaime Garza , Yizao Wang This is my paper

Pith reviewed 2026-07-03 07:08 UTC · model grok-4.3

classification 🧮 math.PR
keywords random partitionsChinese restaurant processphase transitionOrnstein-Uhlenbeck processM/M/∞ queuesecond-order fluctuationsKarlin infinite urn modelcomponent counts
0
0 comments X

The pith

Second-order fluctuations of component counts in random partitions exhibit a phase transition from an Ornstein-Uhlenbeck process to an M/M/∞ queue.

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

The paper establishes second-order limit theorems for the counting process of components of size j_n in partitions induced by the Chinese restaurant process. It shows that when j_n grows much slower than n to the power α/(1+α), the suitably normalized process converges to a stationary Ornstein-Uhlenbeck Gaussian process. When j_n scales exactly like r times that power for r>0, the limit becomes a stationary M/M/∞ queue, with an additional point-process convergence result in that regime. The proofs first derive the limits for the more general Karlin infinite urn model before adapting them to the partition setting, refining an earlier first-order phase transition result.

Core claim

We establish second-order limit theorems in both the subcritical (j_n ≪ n^{α/(1+α)}) and critical regimes for the counting process (C_{j_n}(n(1+t/j_n)_+))_{t∈R}. In the subcritical regime, after appropriate normalization, the limit is a stationary Ornstein-Uhlenbeck Gaussian process, whereas in the critical regime the limit is a stationary M/M/∞ queue. We also establish a more refined point-process convergence in the critical regime. In fact, we establish second-order limit theorems for the more general Karlin infinite urn model, and then adapt the analysis to the Chinese restaurant process.

What carries the argument

The scaled counting process (C_{j_n}(n(1+t/j_n)_+))_{t∈R} whose second-order fluctuations are tracked across the phase transition between the two limiting objects.

If this is right

  • The subcritical regime produces Gaussian fluctuations with a specific exponential covariance structure.
  • The critical regime produces fluctuations whose finite-dimensional distributions match those of the stationary M/M/∞ queue.
  • A point-process level convergence holds in the critical regime, giving more detail than the process-level limit.
  • The same second-order limits apply verbatim to the Karlin infinite urn model under analogous scalings.

Where Pith is reading between the lines

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

  • The M/M/∞ limit suggests that critical fluctuations admit an explicit Poisson-arrival interpretation that could be used for moment calculations.
  • Similar phase transitions between Gaussian and queueing limits may appear when the same scaling is applied to other exchangeable partition models.
  • The transfer step between urn and partition models could be tested directly by comparing finite-n distributions in both constructions.

Load-bearing premise

The second-order results transfer from the Karlin infinite urn model to the Chinese restaurant process without significant distortion under the given scaling regimes for j_n.

What would settle it

Numerical simulation of the normalized counting process in the subcritical regime whose empirical covariance deviates from the known covariance of the stationary Ornstein-Uhlenbeck process would falsify the claimed convergence.

read the original abstract

In a recent paper, Banderier et al. (2024) investigated the limiting behavior of component counts of random partitions induced by the Chinese restaurant process with parameter $\alpha\in(0,1)$ and $\theta>-\alpha$. Let $C_j(n)$ denote the number of components of size $j$ of a partition of $\{1,\ldots,n\}$ and consider $j=j_n\to\infty$ as $n\to\infty$. They revealed a phase transition in the first-order limit behavior of $C_{j_n}(n)$, where the critical regime corresponds to $j_n\sim rn^{\alpha/(1+\alpha)}$ for some $r>0$. A natural next question is to understand the corresponding second-order fluctuations. We establish second-order limit theorems in both the subcritical ($j_n\ll n^{\alpha/(1+\alpha)}$) and critical regimes for the counting process $(C_{j_n}(n(1+t/j_n)_+))_{t\in\mathbb R}$. In the subcritical regime, after appropriate normalization, the limit is a stationary Ornstein--Uhlenbeck Gaussian process, whereas in the critical regime the limit is a stationary $M/M/\infty$ queue. We also establish a more refined point-process convergence in the critical regime. In fact, we establish second-order limit theorems for the more general Karlin infinite urn model, and then adapt the analysis to the Chinese restaurant process.

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 / 0 minor

Summary. The paper establishes second-order limit theorems for the counting process (C_{j_n}(n(1+t/j_n)_+))_{t∈R} associated to component sizes in random partitions generated by the Chinese restaurant process (CRP) with α∈(0,1) and θ>-α. After first proving the results for the more general Karlin infinite-urn model, the authors adapt the analysis to obtain, in the subcritical regime j_n ≪ n^{α/(1+α)}, convergence after normalization to a stationary Ornstein-Uhlenbeck Gaussian process, and in the critical regime j_n ∼ r n^{α/(1+α)}, convergence to a stationary M/M/∞ queue together with a refined point-process limit.

Significance. If the claims hold, the work supplies a precise description of fluctuations around the phase transition identified in Banderier et al. (2024), connecting partition statistics to classical stochastic processes (OU and infinite-server queues) and extending the Karlin urn framework. The rigorous treatment of both regimes and the point-process refinement would be a substantive contribution to probabilistic combinatorics and the study of random discrete structures.

major comments (1)
  1. [Abstract (final paragraph) and the section describing the CRP adaptation] The central claim for the CRP rests on transferring the OU and M/M/∞ limits from the Karlin model. The abstract states only that the analysis is “adapted,” without indicating the location of the quantitative coupling or total-variation bound that shows the error vanishes after the second-order normalization uniformly in the critical window j_n ∼ r n^{α/(1+α)}. Because the CRP induces size-biased dependent sampling of the underlying Poisson-Dirichlet structure, this approximation step is load-bearing; the manuscript must exhibit an explicit error estimate (e.g., in the section containing the CRP transfer) that is o(1) after the scaling used for the limit theorems.

Simulated Author's Rebuttal

1 responses · 0 unresolved

We thank the referee for the careful reading and for identifying the need for greater explicitness in the CRP transfer step. We address the comment below and will revise the manuscript to incorporate the requested quantitative bounds.

read point-by-point responses

  1. Referee: [Abstract (final paragraph) and the section describing the CRP adaptation] The central claim for the CRP rests on transferring the OU and M/M/∞ limits from the Karlin model. The abstract states only that the analysis is “adapted,” without indicating the location of the quantitative coupling or total-variation bound that shows the error vanishes after the second-order normalization uniformly in the critical window j_n ∼ r n^{α/(1+α)}. Because the CRP induces size-biased dependent sampling of the underlying Poisson-Dirichlet structure, this approximation step is load-bearing; the manuscript must exhibit an explicit error estimate (e.g., in the section containing the CRP transfer) that is o(1) after the scaling used for the limit theorems.

    Authors: We agree that the transfer argument requires an explicit quantitative error bound, particularly uniformly over the critical window, and that the current description of the adaptation is insufficiently detailed on this point. The manuscript establishes the limits first for the Karlin model and then states that the CRP case follows by adaptation, but does not display the coupling or total-variation estimate that controls the difference after second-order scaling. In the revised version we will add, in the CRP adaptation section, a dedicated lemma that supplies a total-variation (or Wasserstein) bound between the finite-dimensional distributions of the suitably normalized counting processes for the two models; the bound will be shown to be o(1) uniformly for j_n ∼ r n^{α/(1+α)} by exploiting the explicit size-biasing relation between the CRP and the underlying Poisson-Dirichlet point process together with standard moment comparisons for the component counts. This addition will make the transfer fully rigorous and will be referenced from the abstract. revision: yes

Circularity Check

0 steps flagged

No circularity: second-order limits derived directly for Karlin model then adapted to CRP without reduction to self-citations or fitted inputs

full rationale

The paper states it first establishes the OU and M/M/∞ limits for the general Karlin infinite urn model under the stated scalings, then adapts the analysis to the CRP. The first-order phase transition is cited only for context from Banderier et al. (2024), not used as an input to the second-order derivations. No equations or steps reduce by construction to prior results or self-citations; the central claims rest on independent limit theorems for the urn model with the CRP transfer presented as a subsequent adaptation. This is self-contained against external benchmarks.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

The abstract indicates reliance on standard weak-convergence and point-process tools from probability theory together with an adaptation argument from the Karlin urn model; no free parameters, invented entities, or ad-hoc axioms are mentioned.

axioms (1)
  • standard math Standard results on weak convergence and stationary processes for Markovian systems apply to the normalized counting processes.
    Invoked implicitly when stating convergence to Ornstein-Uhlenbeck and M/M/∞ limits.

pith-pipeline@v0.9.1-grok · 5785 in / 1488 out tokens · 23245 ms · 2026-07-03T07:08:11.974281+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

32 extracted references · 2 canonical work pages · 1 internal anchor

  1. [1]

    D., and Tavaré, S

    Arratia, R., Barbour, A. D., and Tavaré, S. (2003).Logarithmic combinatorial structures: a probabilistic approach. EMSMonographsinMathematics.EuropeanMathematicalSociety (EMS), Zürich

    2003

  2. [2]

    and Najnudel, J

    Bahier, V. and Najnudel, J. (2022). On smooth mesoscopic linear statistics of the eigen- values of random permutation matrices.J. Theoret. Probab., 35(3):1640–1661

    2022

  3. [3]

    Banderier, C., Kuba, M., and Wallner, M. (2024). Phase transitions of composition schemes: Mittag–Leffler and mixed Poisson distributions.Ann. Appl. Probab., 34(5):4635– 4693

    2024

  4. [4]

    Barbour, A. D. and Gnedin, A. V. (2009). Small counts in the infinite occupancy scheme. Electron. J. Probab., 14:no. 13, 365–384

    2009

  5. [5]

    D., Holst, L., and Janson, S

    Barbour, A. D., Holst, L., and Janson, S. (1992).Poisson approximation, volume 2 of Oxford Studies in Probability. The Clarendon Press, Oxford University Press, New York. Oxford Science Publications

    1992

  6. [6]

    and Dang, K

    Ben Arous, G. and Dang, K. (2015). On fluctuations of eigenvalues of random permuta- tion matrices.Ann. Inst. Henri Poincaré Probab. Stat., 51(2):620–647. A PHASE TRANSITION IN RANDOM PARTITIONS 41

    2015

  7. [7]

    and Favaro, S

    Bercu, B. and Favaro, S. (2024). A martingale approach to Gaussian fluctuations and laws of iterated logarithm for Ewens-Pitman model.Stochastic Process. Appl., 178:Paper No. 104493, 19

    2024

  8. [8]

    (1999).Convergence of probability measures

    Billingsley, P. (1999).Convergence of probability measures. Wiley Series in Probability and Statistics: Probability and Statistics. John Wiley & Sons Inc., New York, second edition. A Wiley-Interscience Publication

    1999

  9. [9]

    and Kovalevskii, A

    Chebunin, M. and Kovalevskii, A. (2016). Functional central limit theorems for certain statistics in an infinite urn scheme.Statist. Probab. Lett., 119:344–348

    2016

  10. [10]

    Crane, H. (2016). The ubiquitous Ewens sampling formula.Statist. Sci., 31(1):1–19

    2016

  11. [11]

    and Révész, P

    Csörgő, M. and Révész, P. (1981).Strong approximations in probability and statistics. Probability and Mathematical Statistics. Academic Press, Inc. [Harcourt Brace Jovanovich, Publishers], New York-London

    1981

  12. [12]

    Dumitriu, I., Johnson, T., Pal, S., and Paquette, E. (2013). Functional limit theorems for random regular graphs.Probab. Theory Related Fields, 156(3-4):921–975

    2013

  13. [13]

    and Wang, Y

    Durieu, O. and Wang, Y. (2016). From infinite urn schemes to decompositions of self- similar Gaussian processes.Electron. J. Probab., 21:Paper No. 43, 23

    2016

  14. [14]

    Ewens, W. J. (1972). The sampling theory of selectively neutral alleles.Theoretical population biology, 3(1):87–112

    1972

  15. [15]

    (2010).The Poisson-Dirichlet distribution and related topics

    Feng, S. (2010).The Poisson-Dirichlet distribution and related topics. Probability and its Applications (New York). Springer, Heidelberg. Models and asymptotic behaviors

    2010

  16. [16]

    Ferguson, T. S. (1973). A Bayesian analysis of some nonparametric problems.Ann. Statist., 1:209–230

    1973

  17. [17]

    and Pal, S

    Ganguly, S. and Pal, S. (2020). The random transposition dynamics on random regular graphs and the Gaussian free field.Ann. Inst. Henri Poincaré Probab. Stat., 56(4):2935– 2970

    2020

  18. [18]

    and Wang, Y

    Garza, J. and Wang, Y. (2024). Limit theorems for random permutations induced by Chinese restaurant processes. Arxiv preprint,https://arxiv.org/abs/2412.02162

    work page arXiv 2024

  19. [19]

    and Wang, Y

    Garza, J. and Wang, Y. (2025). A functional central limit theorem for weighted occu- pancy processes of the Karlin model.Stochastic Process. Appl., 188:Paper No. 104665

    2025

  20. [20]

    and van der Vaart, A

    Ghosal, S. and van der Vaart, A. (2017).Fundamentals of nonparametric Bayesian in- ference, volume 44 ofCambridge Series in Statistical and Probabilistic Mathematics. Cam- bridge University Press, Cambridge

    2017

  21. [21]

    Gnedin, A., Hansen, B., and Pitman, J. (2007). Notes on the occupancy problem with infinitely many boxes: general asymptotics and power laws.Probab. Surv., 4:146–171

    2007

  22. [22]

    and Kabluchko, Z

    Grübel, R. and Kabluchko, Z. (2016). A functional central limit theorem for branching random walks, almost sure weak convergence and applications to random trees.Ann. Appl. Probab., 26(6):3659–3698

    2016

  23. [23]

    and Pal, S

    Johnson, T. and Pal, S. (2014). Cycles and eigenvalues of sequentially growing random regular graphs.Ann. Probab., 42(4):1396–1437

    2014

  24. [24]

    Karlin, S. (1967). Central limit theorems for certain infinite urn schemes.J. Math. Mech., 17:373–401

    1967

  25. [25]

    Kingman, J. F. C. (1982). The coalescent.Stochastic Process. Appl., 13(3):235–248. 42 JAIME GARZA AND YIZAO W ANG

    1982

  26. [26]

    (2006).Combinatorial stochastic processes, volume 1875 ofLecture Notes in Mathematics

    Pitman, J. (2006).Combinatorial stochastic processes, volume 1875 ofLecture Notes in Mathematics. Springer-Verlag, Berlin. Lectures from the 32nd Summer School on Proba- bility Theory held in Saint-Flour, July 7–24, 2002, With a foreword by Jean Picard

    2006

  27. [27]

    Resnick, S. I. (1987).Extreme values, regular variation, and point processes, volume 4 of Applied Probability. A Series of the Applied Probability Trust. Springer-Verlag, New York

    1987

  28. [28]

    and Yor, M

    Revuz, D. and Yor, M. (1999).Continuous martingales and Brownian motion, vol- ume 293 ofGrundlehren der Mathematischen Wissenschaften [Fundamental Principles of Mathematical Sciences]. Springer-Verlag, Berlin, third edition

    1999

  29. [29]

    Shao, Q. M. (1989). On a problem of Csörgő and Révész.Ann. Probab., 17(2):809–812

    1989

  30. [30]

    Wang, Y. (2026). On central limit theorems for Ewens–Pitman model. arXiv preprint arXiv:2603.16431

    work page internal anchor Pith review Pith/arXiv arXiv 2026

  31. [31]

    Whitt, W. (2007). Proofs of the martingale FCLT.Probab. Surv., 4:268–302

    2007

  32. [32]

    Wieand, K. (2000). Eigenvalue distributions of random permutation matrices.Ann. Probab., 28(4):1563–1587. Department of Mathematics and Statistics, University of Otta w a, 50 Louis Pasteur Pri- v ate, Otta w a, Ontario, K1N 6N5, Canada. Email address:jgarza@uottawa.ca Department of Mathematical Sciences, University of Cincinnati, 2815 Commons W ay, Cincin...

    2000