Pith. sign in

REVIEW 3 major objections 3 minor 27 references

Decay of connection probability in high-dimensional continuum percolation

T0 review · 3 major / 3 minor · reviewed 2026-08-15 · deepseek-v4-flash

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

Pith's one-line read At the critical point in the high-dimensional random connection model, two points connect with probability decaying like $|x|^{-(d-2)}$.

desk verdict The L^p moment induction is a real advance and the diagrammatic estimates look coherent, but the main theorem leans on an unverified deconvolution preprint and the Z^d claim outruns what is proved. read the letter →

arxiv 2507.19288 v2 pith:PLQRT3ZX submitted 2025-07-25 math.PR math-phmath.MP

Matthew Dickson , Yucheng Liu This is my paper · ORCID
classification math.PRmath-phmath.MP MSC 60K3582B43
keywords randomconnectionmodelcontinuumpercolationcriticaltwo-pointfunctionmean-fieldexponentlaceexpansiondeconvolutionhigh-dimensionaleta=0
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 proves that in the random connection model on $\mathbb{R}^d$—vertices from a Poisson process, edges drawn independently with probability given by an integrable, even adjacency function—the critical connection probability between two points decays, for all sufficiently large $d$, like a constant times $|x|^{-(d-2)}$ with possible anisotropy. That is the mean-field critical exponent $\eta=0$ for a continuum percolation model. The proof combines the lace expansion with a deconvolution theorem that converts moment estimates on the lace kernel into an asymptotic Fourier integral, and it does so through an $L^p$ induction on the kernel's moments. The same argument is claimed to cover nearest-neighbour Bernoulli percolation on $\mathbb{Z}^d$ in $d\ge 11$ and to shorten the earlier 2008 proof for that model. If correct, the result supplies a standard input for studying the incipient infinite cluster and related high-dimensional questions in the continuum setting.

What carries the argument

The engine is the lace function $\Pi_\lambda$, which appears in the Ornstein-Zernike equation $\tau_\lambda=(\varphi+\Pi_\lambda)+\lambda(\varphi+\Pi_\lambda)*\tau_\lambda$, rearranged as $(\delta-J_\lambda)*\lambda\tau_\lambda=J_\lambda$. The proof upgrades the classical moment condition by working with $L^p$ moments: Definition 1.5 calls the $a$-th moment of $\Pi_\lambda$ good if $|x|^a\Pi_\lambda(x)$ is bounded in $L^{p_a}\cap L^2\cap L^\infty$ for a suitable $p_a$, and Proposition 1.6 inductively raises the good moment order from $\phi$ to $\phi+2$. The induction is driven by diagrammatic estimates (Proposition 2.1) that bound the weighted lace function by decorated bubble, triangle, and martini diagrams, whose finiteness is shown by Fourier-analytic derivative estimates. A deconvolution theorem then turns the final $(d-2)$-moment condition into the asymptotic formula for the inverse Fourier integral.

What would settle it

For a concrete adjacency function such as the Gaussian model $\varphi(x)=(2\pi)^{-d/2}e^{-|x|^2/2}$ in $d=9$, numerically evaluate the Fourier integral (1.23) at the critical intensity and check whether $|x|^{d-2}\tau_{\lambda_c}(x)$ converges to $a_d/(\lambda_c\sqrt{\det\Sigma})$ along each coordinate axis; a different power or a non-elliptic direction-dependent prefactor would refute Theorem 1.3. A complementary check is to verify hypothesis (iii) directly for the lace function, testing whether $|x|^{d-2}\Pi_{\lambda_c}(x)$ lies in $L^p\cap L^2\cap L^\infty$ for some $p<d/4$.

Watch

Extended reading notes

Core claim

The central claim is Theorem 1.3. Let the adjacency function $\varphi$ obey Assumption 1.1 and take $d>d_0$ with $d_0\ge 8$ sufficiently large. Then there exists a positive-definite diagonal matrix $\Sigma$, given explicitly as the second-moment matrix of the lace-expansion kernel $J_{\lambda_c}=\lambda_c(\varphi+\Pi_{\lambda_c})$, such that $\tau_{\lambda_c}(x)\sim \frac{a_d}{\lambda_c\sqrt{\det\Sigma}}\,(x\cdot\Sigma^{-1}x)^{-(d-2)/2}$ as $|x|\to\infty$, with $a_d=\Gamma((d-2)/2)/(2\pi^{d/2})$. In words, the critical two-point connection probability has exact power-law decay with the random-walk exponent $d-2$, i.e. $\eta=0$, including the correct prefactor. The paper further claims that the same proof applies to nearest-neighbour Bernoulli percolation on $\mathbb{Z}^d$ for $d\ge 11$. The route is to derive the Fourier integral $\lambda_c\tau_{\lambda_c}(x)=J_{\lambda_c}(x)+\int \frac{\hat J_{\lambda_c}(k)^2}{1-\hat J_{\lambda_c}(k)}e^{-ik\cdot x}\frac{dk}{(2\pi)^d}$, verify five moment and infrared conditions on $J_{\lambda_c}$, and invoke a deconvolution theorem that outputs the asymptotic form.

Load-bearing premise

The proof imports, without proving it here, a deconvolution theorem from a companion preprint (cited as [21, Theorem 1.3]) that turns verified moment bounds on the lace kernel into the exact asymptotic decay formula; if that theorem is false or cannot be applied at $\lambda_c$, the main conclusion does not follow from this paper.

Editorial extensions

If this is right

  • In high-dimensional random connection models at criticality, the probability that two distant points are connected decays as $|x|^{-(d-2)}$ along every direction, with an elliptically anisotropic prefactor governed by the second moments of the lace kernel.
  • The mean-field critical exponent $\eta$ takes the value $0$ for the continuum model, matching the exponents $\gamma=1$, $\beta=1$, and $\delta=2$ already known in this setting.
  • The method supplies the same $\eta=0$ result for nearest-neighbour Bernoulli percolation on $\mathbb{Z}^d$ in $d\ge 11$, with a proof the authors describe as considerably simpler than the 2008 argument.
  • The established two-point decay is the type of input used to construct the incipient infinite cluster and to study one-arm exponents, half-space percolation, and torus problems; the authors state the result is expected to be useful for those continuum analogues.
  • Because the proof is not tied to perturbative small parameters beyond convergence of the lace expansion, the same moment-induction scheme can be applied to any sufficiently spread-out random connection model in $d>8$ once the expansion converges.

Reading between the lines

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

  • If the deconvolution theorem cited as [21] holds, the same $L^p$-moment route could be transferred to other lace-expansion models in $\mathbb{R}^d$—for example self-avoiding walks or massive models—where moment bounds are cheaper than pointwise decay bounds; the paper does not make this extension.
  • The $d>8$ restriction is not intrinsic to the argument: the square diagram is used only for convenience, and the authors note that replacing it with a $b<1$ triangle should handle $d=7$ or $8$. A testable extension would be to run the same induction for spread-out models in those dimensions.
  • One could numerically test the prefactor, not just the exponent, by Monte Carlo simulation of the Palm connection probability in $d=9$ for the disk or Gaussian model and comparing $|x|^{d-2}\tau_{\lambda_c}(x)$ with $a_d/(\lambda_c\sqrt{\det\Sigma})$.
  • The paper's method requires $|x|^{d-2}\varphi(x)\in L^p$ with $p<d/4$; a natural open direction is to determine whether the same $\eta=0$ asymptotic survives for adjacency functions with heavier tails, where only an $L^1$ version of the $(d-2)$-moment is available.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 3 minor

Summary. The paper studies the random connection model on R^d in high dimensions (d > d_0 ≥ 8). Under Assumption 1.1 on the adjacency function φ, the authors prove that the critical two-point connection probability τ_{λ_c}(x) decays like (a_d/(λ_c √det Σ)) (x·Σ^{-1}x)^{-(d-2)/2} as |x|→∞, i.e., the critical exponent η = 0 with possibly anisotropic decay. The proof uses the lace expansion to write τ as the solution of a convolution equation with kernel J_{λ_c}, verifies moment and infrared conditions on J_{λ_c} (in particular, that |x|^{d-2}J_{λ_c}∈L^p∩L^2 for some p<d/4) via a new L^p induction for the lace function Π, and then invokes a deconvolution theorem from the preprint [21] to extract the sharp asymptotic from the Fourier integral. The abstract also claims the argument applies to nearest-neighbour Bernoulli percolation on Z^d for d≥11.

Significance. If the cited deconvolution theorem [21] is correct, the paper fills a gap by proving the mean-field value η=0 for the random connection model, complementing the known exponents γ=1, β=1, δ=2. The L^p moment induction is a genuine technical innovation that simplifies Hara's earlier approach and avoids the second diagrammatic estimate. The explicit verification of Assumption 1.1 for standard models (disk, Gaussian) is useful. However, the main theorem inherits its conclusion from an unproved external preprint, and the abstract's Z^d claim is not proved in the text; these dependencies must be resolved before the theorem can be considered self-contained.

major comments (3)
  1. [Section 1.2, Proof of Theorem 1.3 (Eq. (1.23)–(1.24))] The central asymptotic (1.24) is imported from [21, Theorem 1.3], which is not stated, proved, or even summarized in the present paper. The explicit constant and the matrix Σ in Theorem 1.3 are taken directly from (1.24). The authors list hypotheses (i)–(v) and say they verify them, but the reader cannot check that these are the complete and exact hypotheses of [21, Theorem 1.3]; if that theorem requires an additional condition (for instance a certain decay of Ĵ or a different unweighted moment), then the proof of Theorem 1.3 collapses. The authors should include a full statement of [21, Theorem 1.3], confirm that (i)–(v) are exactly its hypotheses, and ideally provide a proof of the deconvolution step in an appendix or cite a published version of [21].
  2. [Abstract and Remark 1.4] The assertion that the proof "also applies to nearest-neighbour Bernoulli percolation on Z^d in d ≥ 11" is not supported by any theorem, proof, or substantive sketch in the paper. The analysis is carried out only for the continuum random connection model; no Z^d analogue of the L^p induction (Proposition 1.6), the diagrammatic estimates, or the deconvolution step is presented. The paper cites [8] only for the existence of a convergent lace expansion with d0=10. The authors should either state and prove the Z^d result (or provide a detailed transfer argument), or soften the abstract to claim that the method is expected to extend to that setting.
  3. [Section 4, Lemma 4.9 and Eqs. (4.2)–(4.3)] The convergence of the diagrammatic expansion, which is essential for Proposition 4.1 and hence for the bounds on Π, relies on Proposition 7.1 and Lemma 5.7 of the preprint [5], and on the definitions of U_{λ_c} and V_{λ_c} taken from that paper. These results are not proved in the present manuscript, and [5] is not published. This is a second external dependency that is load-bearing for the proof. The authors should either reproduce the statements they need from [5], or provide the proofs in an appendix, or cite a published version if one becomes available.
minor comments (3)
  1. [Reference [6]] The word "dimensons" in the bibliographic entry for Duminil-Copin and Panis should be "dimensions".
  2. [Proof of Lemma 2.3, Eq. (2.16)] The limit in (2.16) is written as |k|→∞, but the pointwise limit in (2.15) is taken as |k|→0; the limit in (2.16) should also be |k|→0.
  3. [Section 3.3.4, last display] The bound ar{H}^{(a,b)}_1 ≤ (1/2 λ_c)^{-1} ar{E}^{(a)} ar{T}^{(b)} ar{S} would read more clearly with the factor λ^{-1} explained explicitly as coming from (3.36); this is a minor presentation point.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the target power-law decay is not assumed as an input; the deconvolution theorem and lace-expansion estimates supply independent hypotheses.

full rationale

The proof of Theorem 1.3 reduces the desired two-point asymptotics to verifying the five hypotheses (i)-(v) for J_{\lambda_c} and then applying the deconvolution theorem [21, Theorem 1.3] to the Fourier integral (1.23). None of the verified hypotheses is the target decay: (iii) is only an L^p integrability condition on |x|^{d-2}J_{\lambda_c}, much weaker than the pointwise asymptotic (1.13), and it is obtained by interpolation from the induction on moments of \Pi_\lambda rather than assumed. The base moment and infrared estimates come from the lace expansion of [15], whose convergence does not presuppose the two-point decay being proved. The use of [21] is a load-bearing external citation involving one of the present authors, but the cited theorem is a general deconvolution statement whose stated hypotheses are moment and infrared conditions, not the percolation conclusion; it therefore counts as independent support rather than a circular reduction. The abstract's assertion about nearest-neighbour Bernoulli percolation on Z^d in d \ge 11 is not proved in the paper, but that is an unverified extension or correctness risk, not a circularity. No fitted parameter is renamed as a prediction, no ansatz is smuggled in via citation, and no known result is merely relabelled. The derivation chain is therefore self-contained in the relevant sense: the output is not equivalent by construction to any of the inputs.

Assumptions & free parameters 0 free parameters · 4 assumptions · 0 invented entities

The central claim rests on external results rather than on fitted constants. Sigma is defined by integrals of J_{lambda_c} and is not fitted; lambda_c is the critical intensity of the model. No invented entities are introduced.

assumptions (4)
  • domain assumption The lace expansion for the random connection model yields (1.14) and the bounds (1.20), (1.21) for d > d_0, from [15, Theorem 1.2, Corollary 5.3, 6.1].
    The entire proof is conditional on this external convergence result, which is cited but not re-proved here.
  • domain assumption Deconvolution theorem [21, Theorem 1.3] converts moment conditions (i)-(v) into the asymptotic (1.24).
    This is the main black box; it is a preprint by one of the authors and not proved in this paper.
  • domain assumption Bounds U_{lambda_c} = O(1), V_{lambda_c} -> 0 from [5, Proposition 7.1, Lemma 5.7].
    Used to control the diagrammatic series for Pi_lambda in Proposition 4.1; [5] is a preprint by one of the authors.
  • standard math Hausdorff-Young inequality and L^p Fourier transform estimates.
    Used throughout Section 3 for weak differentiability and norm bounds; these are standard harmonic analysis tools.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Decay of connection probability in high-dimensional continuum percolation." pith.science (2026). https://pith.science/paper/PLQRT3ZX

@misc{pith2026250719288,
  author       = {Pith},
  title        = {Pith review of: Decay of connection probability in high-dimensional continuum percolation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/PLQRT3ZX}},
  note         = {Machine review of arXiv:2507.19288}
}
abstract

We study a percolation model on $\mathbb R^d$ called the random connection model. For $d$ large, we use the lace expansion to prove that the critical two-point connection probability decays like $|x|^{-(d-2)}$ as $|x| \to \infty$, with possible anisotropic decay. Our proof also applies to nearest-neighbour Bernoulli percolation on $\mathbb Z^d$ in $d \ge 11$ and simplifies considerably the proof given by Hara in 2008. The method is based on the recent deconvolution strategy of Liu and Slade and uses an $L^p$ version of Hara's induction argument.

Figures

Figures reproduced from arXiv: 2507.19288 by the authors.

Figure 1
Figure 1. Graphical notation for the ψ functions. Each • has a vertex factor λ. 1. Define τ ◦ λ (x) = 0 x = λ −1 δ(x) + τλ(x), where δ(x) is the Dirac delta. This means that the edge in the graphical notation can be contracted. 2. Define ψ0 = ψ (1) 0 + ψ (2) 0 + ψ (3) 0 , where ψ (1) 0 (w, u; s) = τλ(u − s)τλ(u − w)τλ(w − s), ψ (2) 0 (w, u; s) = λ −1 δ(w − s)τλ(u − s)λ(τλ ∗ τλ)(u − s), ψ (3) 0 (w, u; s) = λ −1 δ(w − s)φ(u − s… view at source ↗
Figure 2
Figure 2. Diagrams of the ψ¯ 0, and ψ¯ functions. We also define ψ¯ = ψ¯(1) + ψ¯(2) + ψ¯(3) + ψ¯(4), where ψ¯(1)(w, u; r, s) = τλ(w − u) Z R2d τλ(t − s)τλ(t − w)τλ(u − z)τλ(z − t)τ ◦ λ (z − r)λ 2dzdt, ψ¯(2)(w, u; r, s) = τλ(w − s) Z R2d τλ(t − z)τλ(z − u)τλ(u − t)τ ◦ λ (t − w)τ ◦ λ (z − r)λ 2dzdt, ψ¯(3)(w, u; r, s) = τλ(u − w)τλ(w − s)τλ(u − r), ψ¯(4)(w, u; r, s) = τλ(u − w)τλ(w − s)λ −1 δ(u − r). It is convenient to use diag… view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

27 extracted references · 18 canonical work pages

  1. [21]

    Y. Liu. Gaussian deconvolution onR d with application to self-repellent Brownian motion. Preprint,https://arxiv.org/pdf/2411.16058, (2024)

    work page Pith review arXiv 2024

  2. [5]

    Dickson and M

    M. Dickson and M. Heydenreich. The triangle condition for the marked random connection model. Preprint,https://arxiv.org/pdf/2210.07727, (2022)

    arXiv 2022

  3. [8]

    Fitzner and R

    R. Fitzner and R. van der Hofstad. Mean-field behavior for nearest-neighbor percolation in d>10.Electron. J. Probab.,22:1–65, (2017)

    work page 2017

  4. [1]

    Bolthausen, W

    E. Bolthausen, W. Koenig, and C. Mukherjee. Self-repellent Brownian bridges in an interacting Bose gas. Preprint,https://arxiv.org/pdf/2405.08753, (2024)

    arXiv 2024

  5. [2]

    Brydges and T

    D.C. Brydges and T. Spencer. Self-avoiding walk in 5 or more dimensions.Commun. Math. Phys.,97:125–148, (1985)

    1985

  6. [3]

    Caicedo and M

    A. Caicedo and M. Dickson. Critical exponents for marked random connection models.Electron. J. Probab.,29:article 151, 1–57, (2024)

    work page 2024

  7. [4]

    Chatterjee and J

    S. Chatterjee and J. Hanson. Restricted percolation critical exponents in high dimensions. Commun. Pure Appl. Math.,73:2370–2429, (2020)

    work page 2020

  8. [6]

    Duminil-Copin and R

    H. Duminil-Copin and R. Panis. An alternative approach for the mean-field behaviour of spread- out Bernoulli percolation in dimensonsd >6.Probab. Theory Related Fields, (2025).https: //doi.org/10.1007/s00440-025-01416-2

    doi:10.1007/s00440-025-01416-2 2025

Show all 27 references
  1. [7]

    Evans.Partial Differential Equations

    L.C. Evans.Partial Differential Equations. American Mathematical Society, Providence, 2nd edition, (2010)

    2010

  2. [9]

    E. Gilbert. Random plane networks.J. Soc. Ind. Appl. Math.,9:533–543, (1961). 30

    1961

  3. [10]

    T. Hara. Decay of correlations in nearest-neighbor self-avoiding walk, percolation, lattice trees and animals.Ann. Probab.,36:530–593, (2008)

    2008

  4. [11]

    T. Hara, R. van der Hofstad, and G. Slade. Critical two-point functions and the lace expansion for spread-out high-dimensional percolation and related models.Ann. Probab.,31:349–408, (2003)

    2003

  5. [12]

    Hara and G

    T. Hara and G. Slade. Mean-field critical behaviour for percolation in high dimensions.Commun. Math. Phys.,128:333–391, (1990)

    1990

  6. [13]

    Hara and G

    T. Hara and G. Slade. Self-avoiding walk in five or more dimensions. I. The critical behaviour. Commun. Math. Phys.,147:101–136, (1992)

    1992

  7. [14]

    Heydenreich and R

    M. Heydenreich and R. van der Hofstad.Progress in High-Dimensional Percolation and Random Graphs. Springer International Publishing Switzerland, (2017)

    2017

  8. [15]

    Heydenreich, R

    M. Heydenreich, R. van der Hofstad, G. Last, and K. Matzke. Lace expansion and mean-field behavior for the random connection model.Ann. I. Henri Poincar´ e Probab. Statist.,62(1):1–67, (2026)

    2026

  9. [16]

    van der Hofstad and A.A

    R. van der Hofstad and A.A. J´ arai. The incipient infinite cluster for high-dimensional unoriented percolation.J. Stat. Phys.,114:625–663, (2004)

    2004

  10. [17]

    Hutchcroft, E

    T. Hutchcroft, E. Michta, and G. Slade. High-dimensional near-critical percolation and the torus plateau.Ann. Probab.,51:580–625, (2023)

    2023

  11. [18]

    Kozma and A

    G. Kozma and A. Nachmias. The Alexander–Orbach conjecture holds in high dimensions.Invent. Math.,178:635–654, (2009)

    2009

  12. [19]

    Kozma and A

    G. Kozma and A. Nachmias. Arm exponents in high dimensional percolation.J. Am. Math. Soc.,24:375–409, (2011)

    2011

  13. [20]

    Lawler and V

    G.F. Lawler and V. Limic.Random Walk: A Modern Introduction. Cambridge University Press, Cambridge, (2010)

    2010

  14. [22]

    Y. Liu. A general approach to massive upper bound for two-point function with application to self-avoiding walk torus plateau.Electron. J. Probab.,30:article 110, 1–23, (2025)

    2025

  15. [23]

    Liu and G

    Y. Liu and G. Slade. Gaussian deconvolution and the lace expansion.Probab. Theory Related Fields, (2024). Published online.https://doi.org/10.1007/s00440-024-01350-9

    2024 doi

  16. [24]

    Liu and G

    Y. Liu and G. Slade. Gaussian deconvolution and the lace expansion for spread-out models. Ann. I. Henri Poincar´ e Probab. Statist.,62(1):68–85, (2026)

    2026

  17. [25]

    M. Penrose. On a continuum percolation model.Adv. Appl. Probab.,23(3):536–556, (1991)

    1991

  18. [26]

    A. Sakai. Lace expansion for the Ising model.Commun. Math. Phys.,272:283–344, (2007). Correction: A. Sakai. Correct bounds on the Ising lace-expansion coefficients.Commun. Math. Phys.,392:783–823, (2022)

    2007

  19. [27]

    Uchiyama

    K. Uchiyama. Green’s functions for random walks onZ N.Proc. London Math. Soc.,77:215–240, (1998). 31

    1998

Pith tools

Reviewed August 15, 2026 · model on record in the stance chip above.