pith. sign in

arxiv: 2604.19220 · v1 · submitted 2026-04-21 · 🧮 math.PR

Split merge dynamics for expanding intervals and point processes on the real line

Serge Cohen (IMT) , Shambo Saha This is my paper

Pith reviewed 2026-05-10 02:18 UTC · model grok-4.3

classification 🧮 math.PR
keywords split-merge dynamicspoint processesregular variationweak convergenceinvariant distributionsexpanding intervalspartitions of the real line
0
0 comments X

The pith

Split-merge dynamics on regularly varying expanding intervals produce absolutely continuous limiting distributions for their break points.

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

The paper studies recursive partitions of a sequence of growing intervals I_n obtained by splitting subintervals in two and then merging at previous break points. For constant intervals the break points concentrate in a singular way at the ends, but when the lengths of I_n are regularly varying with positive index the empirical distribution of the n break points converges weakly to an absolutely continuous measure. The dynamics are then extended to partitions of the whole real line, where invariant distributions are characterized and certain expanding cases are shown to converge vaguely to those invariants. This matters for understanding how the growth of the underlying space influences the regularity of recursively generated point processes.

Core claim

When the lengths of the intervals I_n vary regularly with a positive index, the empirical distribution of the break points generated by the split-merge dynamics converges weakly to an absolutely continuous probability measure. For the real-line extension, the split-merge dynamics admit invariant distributions, and special cases of the expanding-interval dynamics converge vaguely in distribution to these invariants.

What carries the argument

The split-merge operation that splits each current subinterval into two according to a fixed rule and then merges pairs across the previous partition's break points, producing exactly n points after n steps on an interval of growing length.

If this is right

  • When interval lengths are regularly varying, the limiting point distribution is absolutely continuous rather than singular.
  • Invariant distributions for the real-line split-merge dynamics can be explicitly characterized.
  • Special expanding split-merge processes converge vaguely to the real-line invariants.
  • The growth rate of the intervals governs the transition from singular to continuous limits in these recursive partitions.

Where Pith is reading between the lines

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

  • The regular variation index may determine the specific form of the absolutely continuous density in the limit.
  • Similar split-merge rules could be applied to other growing domains to produce stationary point processes with continuous densities.
  • The construction might connect to fragmentation processes, allowing the invariants to be sampled via recursion on expanding intervals.

Load-bearing premise

The interval lengths must be regularly varying with a positive index and the splitting rule must be fixed in a way that supports the convergence proofs.

What would settle it

Simulate the split-merge process for large n on intervals with lengths n to the alpha for alpha greater than zero and check whether the histogram of the n break points fills the interval with a continuous density rather than concentrating at the endpoints.

Figures

Figures reproduced from arXiv: 2604.19220 by Serge Cohen (IMT), Shambo Saha.

Figure 1
Figure 1. Figure 1: Illustration of the first iteration of ( [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Illustration of the second iteration of ( [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
read the original abstract

We study sequences of partitions of a non decreasing sequence I n of intervals into subintervals, starting from the trivial partition, in which each partition is obtained from the one before by splitting its subintervals in two, according to a given rule, and then merging pairs of subintervals at the break points of the old partition. The nth partition then comprises n+1 subintervals with n break points. When I n = [0, 1] is constant, the empirical distribution of these points was shown to converge weakly to a singular probability supported in {0, 1} in a previous article. When the length of the intervals is regularly varying with a positive index, we show in this article that the limit can be absolutely continuous. In the last part we extend the split merge dynamics to partitions of R. In this case we characterize invariant distributions and show that special instances of split merge dynamics for expanding intervals converge to these invariant measures vaguely in distribution.

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

2 major / 2 minor

Summary. The paper studies split-merge dynamics on a non-decreasing sequence of intervals I_n, beginning from the trivial partition and iteratively splitting each subinterval according to a fixed rule then merging pairs at prior break points, yielding n+1 subintervals after n steps. For constant length I_n = [0,1] the empirical distribution of break points converges weakly to a singular measure supported at the endpoints (prior work). The central claims are that when |I_n| is regularly varying with positive index the limiting empirical measure is absolutely continuous, and that the dynamics extend to partitions of R where invariant distributions are characterized via a fixed-point equation on the intensity measure, with specific expanding-interval instances converging vaguely in distribution to these invariants.

Significance. If the stated convergence and invariance results hold, the work supplies a concrete mechanism by which interval expansion produces absolutely continuous limiting point distributions (contrasting the singular constant-length case) and gives an explicit Markov-kernel characterization of stationary measures for infinite-line partitions. These are technically non-trivial extensions that could serve as building blocks for models of growing random partitions and vague limits of point processes on R.

major comments (2)
  1. [§3] §3 (proof of absolute continuity): the argument relies on tightness plus explicit density evolution under regular variation of |I_n| with index α>0, but the manuscript does not state the precise range of α for which the density remains bounded away from zero and infinity; without this the absolute-continuity claim is not fully delimited.
  2. [§5] §5 (vague convergence on R): the coupling argument between the expanding-interval process and the stationary Markov process on partitions of R is only sketched; the error bound controlling the vague distance after n steps is not displayed, making it impossible to verify that the regular-variation tail indeed yields convergence.
minor comments (2)
  1. [§2] Notation for the splitting rule is introduced only in §2; an explicit formula or pseudocode at the first appearance would improve readability.
  2. [Abstract] The abstract claims 'the limit can be absolutely continuous' but the body shows it is absolutely continuous under the regular-variation hypothesis; the abstract should be aligned.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the careful reading of our manuscript and for highlighting the need for greater precision in Sections 3 and 5. We agree that both points require clarification and will revise the paper accordingly. Our point-by-point responses follow.

read point-by-point responses

  1. Referee: [§3] §3 (proof of absolute continuity): the argument relies on tightness plus explicit density evolution under regular variation of |I_n| with index α>0, but the manuscript does not state the precise range of α for which the density remains bounded away from zero and infinity; without this the absolute-continuity claim is not fully delimited.

    Authors: We agree that an explicit statement of the range is needed. The proof in §3 proceeds via the explicit recursive evolution of the density under the regular-variation assumption and establishes absolute continuity together with bounds away from zero and infinity (on compact subsets of (0,1)) for every α>0. In the revision we will add a precise statement at the beginning of §3: the limiting measure is absolutely continuous with density bounded away from 0 and ∞ on any [δ,1-δ] whenever |I_n| is regularly varying with index α>0. This delimits the claim without altering the argument. revision: yes

  2. Referee: [§5] §5 (vague convergence on R): the coupling argument between the expanding-interval process and the stationary Markov process on partitions of R is only sketched; the error bound controlling the vague distance after n steps is not displayed, making it impossible to verify that the regular-variation tail indeed yields convergence.

    Authors: We accept that the error bound should be displayed explicitly rather than left implicit. The coupling construction in §5 already yields a bound on the vague distance of order o(1) times the tail integral of the regularly varying function; the regular-variation assumption then forces this quantity to zero. In the revised manuscript we will insert the explicit inequality (displayed as a new displayed equation in §5) together with the short verification that the tail vanishes for any positive index α. This makes the convergence statement fully verifiable while leaving the overall argument unchanged. revision: yes

Circularity Check

0 steps flagged

Minor self-citation to prior constant-case result; new claims on expanding intervals and R-partitions are independent

full rationale

The paper cites a previous article solely for the constant-length I_n=[0,1] case (singular limit supported at {0,1}). All load-bearing steps for the regularly-varying expanding case and the R-extension are self-contained: they begin from an explicit splitting rule (uniform or deterministic break-point), apply the regular-variation tail assumption directly to obtain tightness and density control, and characterize invariants via a fixed-point equation on the intensity measure together with a coupling argument. No equation reduces to a fitted parameter, no ansatz is smuggled via self-citation, and the central claims do not rely on the prior result for their validity.

Axiom & Free-Parameter Ledger

0 free parameters · 2 axioms · 0 invented entities

The paper relies on standard results from probability theory such as weak convergence of measures and properties of regularly varying functions; no free parameters, invented entities, or ad-hoc axioms are evident from the abstract.

axioms (2)
  • standard math Weak convergence of empirical measures to a limiting distribution
    Invoked for the convergence statements in both the fixed and expanding interval cases.
  • domain assumption Regular variation of interval lengths with positive index
    Central assumption enabling the absolutely continuous limit; stated in the abstract as the condition under which the result holds.

pith-pipeline@v0.9.0 · 5460 in / 1324 out tokens · 35139 ms · 2026-05-10T02:18:46.789371+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

7 extracted references · 7 canonical work pages

  1. [1]

    Billingsley

    P. Billingsley. Convergence of probability measures . John Wiley & Sons Inc., New York, 1999. 2nd Edition

    work page 1999

  2. [2]

    N. H. Bingham, C. M. Goldie, and J. L. Teugels. Regular variation, vol- ume 27 of Encyclopedia of Mathematics and its Applications . Cambridge University Press, Cambridge, 1989

    work page 1989

  3. [3]

    Cohen, J

    S. Cohen, J. Norris, M. Pain, and G. Samorodnitsky. Inter lacing sequnces resulting from an interval split-merge dynamics and the ind uced proba- bility. https://hal.science/hal-04947651 , to appear in A LEA, February 2025

    work page 2025

  4. [4]

    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 a nd methods

    work page 2003

  5. [5]

    An Introduction to Probability Theory and its Applications , volume 2

    W Feller. An Introduction to Probability Theory and its Applications , volume 2. Wiley, 1950

    work page 1950

  6. [6]

    Kallenberg

    O. Kallenberg. Random measures, theory and applications , volume 77 of Probability Theory and Stochastic Modelling . Springer, Cham, 2017

    work page 2017

  7. [7]

    Meyn and R

    S. Meyn and R. L. Tweedie. Markov chains and stochastic stability . Cambridge University Press, Cambridge, second edition, 20 09. With a prologue by Peter W. Glynn. 30

    work page