pith. sign in

arxiv: 2510.18780 · v4 · submitted 2025-10-21 · 🧮 math.PR

Ranges of Extremal Processes and Heavy-Tailed Random Walks in Spaces of Growing Dimension

Bochen Jin , Ilya Molchanov This is my paper

Pith reviewed 2026-05-18 04:52 UTC · model grok-4.3

classification 🧮 math.PR
keywords heavy-tailed random walksextremal processesPoisson cluster processGromov-Hausdorff convergencel_p metricstriangular array limitsgrowing dimensionmetric spaces
0
0 comments X

The pith

Heavy-tailed random walks in growing dimensions converge to a Poisson cluster process after isometries of the l_p metric.

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

The paper develops limit theorems for extremal processes and random walks produced by heavy-tailed random vectors in high-dimensional l_p spaces. In the regime where both the number of steps n and the dimension d tend to infinity, it shows that suitable isometries can be applied to the paths so that they converge in distribution to a Poisson cluster process. This convergence also implies that the paths, when viewed as finite metric spaces, converge in the Gromov-Hausdorff metric, and that the transformed paths converge in the space of counting measures equipped with an induced Hausdorff metric.

Core claim

It is possible to transform the paths by suitable isometries of l_p such that the transformed paths converge in distribution to a Poisson cluster process; these results also imply the convergence in distribution of the paths viewed as finite metric spaces in the Gromov-Hausdorff metric.

What carries the argument

Suitable isometries of the l_p metric that normalize the heavy-tailed paths for convergence to a Poisson cluster process in the triangular array setting.

If this is right

  • The paths of the extremal processes and random walks converge in distribution to a Poisson cluster process after transformation.
  • The paths viewed as finite metric spaces converge in the Gromov-Hausdorff metric.
  • The transformed paths converge in distribution in the space of counting measures on the line with a suitable l_p-type Hausdorff metric.

Where Pith is reading between the lines

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

  • This suggests the limit behavior may hold for other symmetric norms when dimension grows with sample size.
  • The Poisson cluster structure could inform simulation methods for high-dimensional extremes without explicit dimension dependence.
  • Extensions might examine how the cluster process changes under different regular variation indices.

Load-bearing premise

The random vectors are heavy-tailed in a way that allows regular variation or equivalent tail conditions, with both the number of steps and dimension tending to infinity.

What would settle it

A specific sequence of heavy-tailed distributions in increasing dimensions where no choice of isometries leads to convergence of the paths to a Poisson cluster process, or where the Gromov-Hausdorff distance does not converge to zero.

read the original abstract

We consider extremal processes and random walks generated by heavy-tailed random vectors taking values in $\mathbb{R}^d$ endowed with the $\ell_p$ metric. We establish limit theorems for the associated paths in the triangular array setting when both the number of steps $n$ and the dimension $d$ grow to infinity. It is shown that it is possible to transform the paths by suitable isometries of $\ell_p$ such that the transformed paths converge in distribution and to identify the limit in terms of a Poisson cluster process. These results also imply the convergence in distribution of the paths viewed as finite metric spaces in the space of metric spaces equipped with the Gromov-Hausdorff metric. Furthermore, we prove convergence in distribution of the transformed paths in the space of counting measures on the line equipped with a Hausdorff metric induced by a suitable $\ell_p$-type distance between counting measures.

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 establishes limit theorems for extremal processes and heavy-tailed random walks in R^d equipped with the ell_p metric, in the triangular array regime where both the number of steps n and dimension d tend to infinity. It shows that paths can be transformed by suitable isometries of ell_p so that the transformed paths converge in distribution to a Poisson cluster process; this implies Gromov-Hausdorff convergence of the paths viewed as finite metric spaces, and also convergence of the paths as counting measures under a Hausdorff metric induced by an ell_p-type distance.

Significance. If the results hold, they provide a mechanism for reducing high-dimensional heavy-tailed path behavior to a universal Poisson cluster limit via isometries, which is of interest for extreme-value theory in growing dimensions and for linking functional convergence to convergence of intrinsic metric spaces in the Gromov-Hausdorff topology. The explicit identification of the limit and the counting-measure extension add technical value.

major comments (2)
  1. [Theorem 3.2 and §4.1] The central construction in the proof of path convergence (around the statement of Theorem 3.2 and the subsequent isometry selection) defines the isometry by ordering coordinates according to the size of the largest increments. For p ≠ 2 the isometry group is essentially signed permutations, so the selection map from the Skorokhod space of cadlag paths into this group must be shown to be measurable (and preferably continuous at continuity points of the limit) for the composition to preserve weak convergence. The manuscript does not contain an explicit verification of this measurability or of the required continuity property with respect to the topology used for the convergence statement.
  2. [§5 and the proof of Theorem 5.1] The joint limit n, d → ∞ is handled by a triangular-array argument that relies on regular variation of the tails. It is not clear from the proof sketch whether the constants in the Poisson cluster intensity are uniform in d or whether additional truncation arguments are needed when d grows faster than any function of n; this uniformity is load-bearing for the claimed convergence in the Gromov-Hausdorff metric.
minor comments (2)
  1. [§2] The notation for the ell_p-type distance on counting measures (Definition 2.4) could be introduced earlier and compared more explicitly with the standard Hausdorff metric on compact sets.
  2. [Eq. (3.5)] A few typographical inconsistencies appear in the indexing of the Poisson points (e.g., the use of both k and j for the same summation index in the cluster representation).

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the thorough review and the positive evaluation of our work. We appreciate the suggestions for improving the rigor of the proofs, particularly regarding measurability and uniformity issues. We address the major comments point by point below and plan to revise the manuscript accordingly.

read point-by-point responses

  1. Referee: [Theorem 3.2 and §4.1] The central construction in the proof of path convergence (around the statement of Theorem 3.2 and the subsequent isometry selection) defines the isometry by ordering coordinates according to the size of the largest increments. For p ≠ 2 the isometry group is essentially signed permutations, so the selection map from the Skorokhod space of cadlag paths into this group must be shown to be measurable (and preferably continuous at continuity points of the limit) for the composition to preserve weak convergence. The manuscript does not contain an explicit verification of this measurability or of the required continuity property with respect to the topology used for the convergence statement.

    Authors: We agree with the referee that an explicit verification of the measurability of the isometry selection map is necessary, especially for p ≠ 2 where the isometry group consists of signed permutations. In the original manuscript, this was implicitly assumed through the construction, but we acknowledge that a detailed proof is warranted. In the revised version, we will add a new lemma in Section 4.1 that establishes the measurability of the map which orders the coordinates by the magnitude of the largest increments. This map will be shown to be Borel measurable with respect to the Skorokhod topology on the space of cadlag functions. Additionally, we will prove that it is continuous at points where the limiting process has distinct jump sizes almost surely, ensuring the preservation of weak convergence under the composition. We believe this addresses the concern fully. revision: yes

  2. Referee: [§5 and the proof of Theorem 5.1] The joint limit n, d → ∞ is handled by a triangular-array argument that relies on regular variation of the tails. It is not clear from the proof sketch whether the constants in the Poisson cluster intensity are uniform in d or whether additional truncation arguments are needed when d grows faster than any function of n; this uniformity is load-bearing for the claimed convergence in the Gromov-Hausdorff metric.

    Authors: We thank the referee for highlighting this point on uniformity. The regular variation assumption on the tails holds uniformly in the dimension d, since the marginal distributions are identical and regularly varying with index α independently of d. To make this explicit and address potential issues when d grows rapidly with n, we will revise the proof of Theorem 5.1 to include detailed truncation arguments. These will demonstrate that the error from truncating high-dimensional coordinates is negligible uniformly in d, using bounds derived from the regular variation property. Consequently, the intensity constants of the limiting Poisson cluster process remain uniform, supporting the Gromov-Hausdorff convergence without requiring specific growth rates between d and n. We will incorporate these clarifications and additional estimates in the revised manuscript. revision: yes

Circularity Check

0 steps flagged

No significant circularity; derivations rely on standard extreme-value and point-process tools

full rationale

The paper proves convergence in distribution of suitably transformed paths to a Poisson cluster process and Gromov-Hausdorff convergence of the associated metric spaces, under regular variation assumptions on heavy-tailed vectors in the triangular array regime with n and d to infinity. These limits are obtained via established weak convergence arguments for extremal processes and point processes on metric spaces; the isometry transformations are constructed explicitly from the path data to align large jumps but do not redefine the target limit or fit parameters to the output. No load-bearing step reduces by construction to a self-citation, fitted input renamed as prediction, or ansatz smuggled from prior work by the same authors. The central claims remain independent of the paper's own fitted quantities and rest on external benchmarks in probability theory.

Axiom & Free-Parameter Ledger

0 free parameters · 2 axioms · 0 invented entities

The central claims rest on standard mathematical background plus domain assumptions about heavy tails; no free parameters or invented entities are introduced in the abstract.

axioms (2)
  • domain assumption Heavy-tailed random vectors admit regular variation or equivalent tail conditions that permit Poisson point process limits
    Invoked to obtain the Poisson cluster process limit when both n and d grow.
  • domain assumption The l_p spaces admit sufficient isometries to align the paths for convergence
    Used to transform paths before taking the limit.

pith-pipeline@v0.9.0 · 5682 in / 1449 out tokens · 48182 ms · 2026-05-18T04:52:30.142102+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

What do these tags mean?
matches
The paper's claim is directly supported by a theorem in the formal canon.
supports
The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends
The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses
The paper appears to rely on the theorem as machinery.
contradicts
The paper's claim conflicts with a theorem or certificate in the canon.
unclear
Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score.

  1. Set-indexed and multiple sums in high dimensions

    math.PR 2026-05 unverdicted novelty 6.0

    Sets of values of multiple and set-indexed sums of random vectors converge in probability to a generalization of the Wiener spiral as dimension grows.

Reference graph

Works this paper leans on

25 extracted references · 25 canonical work pages · cited by 1 Pith paper

  1. [1]

    A. D. Barbour and T. C. Brown. Stein’s method and point process approximation. Stochastic Process. Appl., 43(1):9–31, 1992

    work page 1992

  2. [2]

    Basrak, N

    B. Basrak, N. Milin ˇcevi´c, and I. Molchanov. Foundations of regular variation on topological spaces. Technical report, arxiv math:2503.00921, 2025

    work page arXiv 2025

  3. [3]

    Basrak and H

    B. Basrak and H. Planini ´c. A note on vague convergence of measures.Statist. Probab. Lett., 153:180–186, 2019

    work page 2019

  4. [4]

    Basrak, H

    B. Basrak, H. Planini ´c, and P. Soulier. An invariance principle for sums and record times of regularly varying stationary sequences.Probab. Theory Related Fields, 172(3-4):869–914, 2018

    work page 2018

  5. [5]

    Billingsley.Convergence of Probability Measures

    P. Billingsley.Convergence of Probability Measures. John Wiley & Sons, Inc., New York, second edition, 1999. 34

    work page 1999

  6. [6]

    Buraczewski, E

    D. Buraczewski, E. Damek, and T. Mikosch.Stochastic Models with Power-Law Tails. The equation X=AX+B. Springer, Cham, 2016

    work page 2016

  7. [7]

    Burago, Y

    D. Burago, Y . Burago, and S. Ivanov.A course in metric geometry, volume 33 of Graduate Studies in Mathematics. American Mathematical Society, Providence, RI, 2001

    work page 2001

  8. [8]

    Cheng, J

    R. Cheng, J. Mashreghi, and W. T. Ross.Function Theory andℓ p Spaces. Ameri- can Mathematical Society, Providence, RI, 2020

    work page 2020

  9. [9]

    Davydov and V

    Y . Davydov and V . Egorov. On convergence of empirical point processes.Statist. Probab. Lett., 76(17):1836–1844, 2006

    work page 2006

  10. [10]

    Davydov, I

    Y . Davydov, I. Molchanov, and S. Zuyev. Stability for random measures, point processes and discrete semigroups.Bernoulli, 17(3):1015–1043, 2011

    work page 2011

  11. [11]

    Dombry, C

    C. Dombry, C. Tillier, and O. Wintenberger. Hidden regular variation for point processes and the single/multiple large point heuristic.Ann. Appl. Probab., 32(1):191–234, 2022

    work page 2022

  12. [12]

    R. J. Fleming and J. E. Jamison.Isometries on Banach Spaces: Function Spaces, volume 129. 2003

    work page 2003

  13. [13]

    Fristedt and L

    B. Fristedt and L. Gray.A Modern Approach to Probability Theory. Birkh ¨auser, Boston, MA, 1997

    work page 1997

  14. [14]

    C. C. Heyde. On large deviation probabilities in the case of attraction to a non- normal stable law.Sankhy ¯a Ser. A, 30:253–258, 1968

    work page 1968

  15. [15]

    Hu and Z

    J. Hu and Z. Bai. Strong representation of weak convergence.Sci. China Math., 57(11):2399–2406, 2014

    work page 2014

  16. [16]

    Kabluchko and A

    Z. Kabluchko and A. Marynych. Random walks in the high-dimensional limit I: The Wiener spiral.Ann. Inst. Henri Poincar ´e Probab. Stat., 60(4):2945–2974, 2024

    work page 2024

  17. [17]

    Kabluchko, A

    Z. Kabluchko, A. Marynych, and K. Raschel. Random walks in the high- dimensional limit II: The crinkled subordinator.Stochastic Process. Appl., 176:Pa- per No. 104428, 13, 2024

    work page 2024

  18. [18]

    Lamperti

    J. Lamperti. On the isometries of certain function-spaces.Pacific J. Math., 8:459– 466, 1958

    work page 1958

  19. [19]

    Lederer and M

    J. Lederer and M. Oesting. Extremes in high dimensions: methods and scalable algorithms. Technical report, ArXiv math.PR/2303.04258, 2023. 35

    work page arXiv 2023

  20. [20]

    J. P. Nolan.Univariate Stable Distributions: Models for Heavy Tailed Data. Springer, Cham, 2020

    work page 2020

  21. [21]

    S. I. Resnick.Extreme Values, Regular Variation, and Point Processes. Springer- Verlag, New York, 1987

    work page 1987

  22. [22]

    Schuhmacher and A

    D. Schuhmacher and A. Xia. A new metric between distributions of point pro- cesses.Adv. in Appl. Probab., 40(3):651–672, 2008

    work page 2008

  23. [23]

    V . V . Uchaikin and V . M. Zolotarev.Chance and Stability. VSP, Utrecht, 1999

    work page 1999

  24. [24]

    A. Vince. A rearrangement inequality and the permutahedron.Amer. Math. Monthly, 97(4):319–323, 1990

    work page 1990

  25. [25]

    Weissman

    I. Weissman. On some dependence measures for multivariate extreme value distri- butions. InAdvances in mathematical and statistical modeling, Stat. Ind. Technol., pages 171–180. Birkh¨auser Boston, Boston, MA, 2008. 36

    work page 2008