REVIEW 3 major objections 4 minor 21 references
Concentration inequalities for functionals of Poisson cylinder processes
T0 review · 3 major / 4 minor · reviewed 2026-08-14 · deepseek-v4-flash
Pith's one-line read For a stationary Poisson $k$-cylinder process in a compact window, the union volume has upper tail $\exp(-\Theta(r\log r))$, and under isotropy the same order holds for every intrinsic volume.
desk verdict Volume concentration for Poisson cylinders is solid and new; the intrinsic-volume claim breaks at j=0 in the Boolean model and needs revision. read the letter →
The pith
A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.
The reading
What carries the argument
The engine is a general exponential concentration inequality for Poisson functionals (Lemmas 2.2 and 2.3), which bounds $P(F-\mathbb{E}F\ge r)$ by $\exp(\inf_{s\ge0}(\int_0^s v(u)\,du-rs))$ in terms of an integrated add-one-cost functional $V_F(s)$. The paper feeds this engine with the deterministic geometric estimate $\lambda_d(Z(x,\theta,K)\cap W)\le \lambda_{d-k}(K)\,\mathrm{diam}(W)^k$ (inequality (3.3)), which converts the abstract bound into the explicit $\Psi$-function expressions; the number of cylinders that can touch $W$ is then controlled by the projection formula $\lambda_{d-k}(P_{d-k}(\Theta^T W)+\Xi^*)$. For intrinsic volumes, the same scheme is powered by a new mean-value formula (Proposition 5.1), obtained from the principal kinematic formula, together with isoperimetric inequalities that express higher intrinsic volumes of a cylinder cut by $W$ as powers of $V_j$.
What would settle it
Test Corollary 4.3 numerically: take $W=[0,1]^3$, $k=1$, and a unit-square base $M$, compute $\alpha$ and $\beta$ from (4.1), simulate the isotropic Poisson cylinder process for a fixed $\gamma$, and estimate $P(F-\mathbb{E}F\ge r)$ at several $r$ with enough repetitions; if the empirical upper tail significantly exceeds the deterministic value $\exp(r/\alpha-(\beta+r/\alpha)\log(1+r/(\alpha\beta)))$ supplied by the corollary, the central claim would be refuted.
Extended reading notes
Core claim
The central claim is Theorem 3.2. For a stationary Poisson $k$-cylinder process with intensity $\gamma$ and typical base volume $m_{d-k}$, the volume $F=\lambda_d(Z\cap W)$ satisfies upper and lower tail bounds whose exponents are infimums over $s\ge 0$ of an expectation involving $\Psi(s\,\lambda_{d-k}(\Xi)\,\mathrm{diam}(W)^k)$, multiplied by the projection integral $\lambda_{d-k}(P_{d-k}(\Theta^T W)+\Xi^*)$, where $\Psi(x)=e^x-x-1$. When the base is a random rotation of a fixed convex body $M$ and the direction is uniform, Corollary 4.3 reduces this to $P(F-\mathbb{E}F\ge r)\le \exp(r/\alpha-(\beta+r/\alpha)\log(1+r/(\alpha\beta)))$, with $\alpha=\lambda_{d-k}(M)\,\mathrm{diam}(W)^k$ and $\beta$ given by (4.1); hence $\exp(-\Theta(r\log r))$ for a fixed window and $\exp(-\Theta(r^2))$ for the lower tail. For isotropic processes with convex bases, Theorem 5.4 and Corollary 5.6 extend the same structure to every intrinsic volume $V_j(Z\cap W)$ with $j\ge k$, using new mean-value formulas for the intrinsic volumes. The case $k=0$ recovers the known Boolean-model inequalities and adds intrinsic-volume concentration for the Boolean model.
Load-bearing premise
The whole chain rests on inequality (3.3), stated without proof, that a window cut by one cylinder has volume at most (base volume) times (window diameter)$^k$; if this geometric bound ever fails, the explicit constants in Corollaries 4.3 and 5.6 lose their justification.
Editorial extensions
If this is right
- For a fixed window $W$, the volume of the union has upper tail of order $\exp(-\Theta(r\log r))$ and lower tail of order $\exp(-\Theta(r^2))$, matching the order for a Poisson random variable and for the Boolean model despite the long-range correlations of cylinder processes.
- For a window growing as $r^{1/d}W$, the upper-tail bound becomes $\exp(-\Theta(r^{1-k/d}))$, explicitly degrading as the cylinder dimension $k$ grows and reducing to the Boolean order at $k=0$.
- For isotropic processes with convex bases, every intrinsic volume $V_j(Z\cap W)$ with $j\ge k$ satisfies the same explicit exponential concentration with constants built from the window and base body (Corollary 5.6).
- The fixed-window bound improves on what follows from the cumulant-based large-deviation estimates in the existing cylinder-process literature, which give only $\exp(-\Theta(r))$ for the upper tail.
- The $k=0$ case recovers, and for intrinsic volumes extends, the known concentration inequalities for the classical Boolean model.
Reading between the lines
- The method localizes all anisotropy in the projection integral $\mathbb{E}[\lambda_{d-k}(P_{d-k}(\Theta^T W)+\Xi^*)]$; a natural extension would be to non-isotropic cylinder processes once that integral is controlled by geometric estimates.
- Because the constants $\alpha$ and $\beta$ in Corollary 4.3 are explicit, one could compare base shapes of equal volume and ask which body $M$ makes the upper tail sharpest; this is an optimization problem the paper does not address.
- The growing-window exponent $r^{1-k/d}$ suggests that, on the paper's scaling, concentration weakens dramatically as $k$ approaches $d$; deciding whether this bound is tight would require matching lower bounds on the tail.
Signed reviews
Editorial analysis
A structured set of objections, weighed in public.
Referee Report
Summary. The paper develops concentration inequalities for the volume and the intrinsic volumes of the union set generated by a stationary Poisson process of k-cylinders in R^d, observed in a compact window W. For the volume functional F = λ_d(Z ∩ W), the authors apply a general concentration inequality for Poisson functionals from Gieringer and Last, control the add-one cost through the estimate λ_d(Z(x,θ,K) ∩ W) ≤ λ_{d−k}(K) diam(W)^k, and obtain upper and lower tail bounds. In the isotropic case with randomly rotated convex base, this gives explicit exp(−Θ(r log r)) upper tails and Gaussian-type lower tails; expanding windows are also treated. In the second part, assuming isotropy and convex bases, mean value formulas for V_j(Z ∩ W) are derived and used to state concentration inequalities for intrinsic volumes of arbitrary order, with the Boolean model k = 0 as the special case.
Significance. The volume part is a natural and nontrivial generalization of the Boolean-model inequalities of [3], and the resulting tail orders, Θ(r log r) for fixed windows and r^{1−k/d} for growing windows, are informative and likely close to optimal. The mean-value formulas for intrinsic volumes of isotropic Poisson cylinder processes are also useful new material. If the intrinsic-volume concentration theorems were correct as stated, they would be a significant contribution. However, the advertised full range k = 0, j = 0 is not well-defined, and the proof of the basic integrability lemma fails precisely for the Euler characteristic; these are load-bearing gaps in Section 5. The volume results, by contrast, appear coherent, up to small presentational issues.
major comments (3)
- [Theorem 5.4 and Corollary 5.6] The statements are not well-defined for j = 0. Since Theorem 5.4 assumes j ≥ k and Corollary 5.6 states k ≤ j ≤ d, the Boolean case k = 0 includes j = 0. The displayed exponents contain (∑ diam(W)^{j−i} binomial(k, j−i) V_i(Ξ))^{m/j}, and the lower-tail bound and the definition of β in Corollary 5.6 contain the same power m/j with j = 0. The paper explicitly advertises k = 0, so this is not a harmless convention; as written, Theorem 5.4 and Corollary 5.6 have no meaning for V_0.
- [Lemma 5.3] The proof of Lemma 5.3 states that the intrinsic volumes are non-negative and monotone under set inclusion on the family of convex bodies and concludes that D_{(x,θ,K)}F_j ≤ V_j(W). But Z ∩ W and Z ∩ Z(x,θ,K) ∩ W are not convex in general. For j = 0 the assertion is false: V_0 is the Euler characteristic, and a union of thin convex rectangles arranged as an m×m grid inside a convex cell has Euler characteristic 1 − m^2, so the add-one cost D_{(x,θ,K)}F_0 = 1 − χ(Z ∩ Z(x,θ,K)) can be of order m^2 and is not bounded by V_0(W). Consequently the proof that s_{F_0} = s^{(lt)}_{F_0} = ∞ collapses, and the optimization over all s ≥ 0 in Theorem 5.4 is not justified for the Boolean-model case k = 0, j = 0.
- [Theorem 5.4, proof; Lemma 5.3] Even for 1 ≤ j < d the integrability step needs more than the hypotheses stated. The lemma's monotonicity argument applies, if at all, to convex bodies, not to the non-convex set Z ∩ Z(x,θ,K) ∩ W; and the condition m_i < ∞ in Theorem 5.4 does not by itself guarantee the exponential integrability needed for s_{F_j} = ∞. At minimum the proof must either establish the relevant nonnegativity and exponential moment bounds on the convex ring, or add explicit moment assumptions and restrict s to a finite interval. This is load-bearing because the final inf over s ≥ 0 in Lemma 2.2 is taken over [0, s_F).
minor comments (4)
- [Section 3, Eq. (3.3)] The geometric estimate λ_d(Z(x,θ,K) ∩ W) ≤ λ_{d−k}(K) diam(W)^k is the key step converting the abstract Poisson bound into explicit constants, but it is stated without proof; it follows from the isodiametric inequality applied to the projection of W onto the k-dimensional subspace, and should be stated as a lemma with proof or reference.
- [Theorem 5.4] The formula for β_m is typeset in an illegible way, for example the expression 'm−2p−1 d−k ...' is ambiguous in the provided text; please reformat all coefficients and verify that they agree with the quantities α_m defined in the proof.
- [Abstract] The abstract contains the typo 'payed' for 'paid'.
- [Corollary 4.3, proof] The inequality αβ ≥ EF is derived using a monotonicity result from [14, Theorem 1] for the gamma function; since this inequality is needed to justify the range 0 ≤ r ≤ EF in the lower tail, a short statement of the cited result would help readability.
Circularity Check
No significant circularity: the tail bounds are derived from an independent concentration inequality and explicit geometric formulas, with no fitted parameters or load-bearing self-citation.
full rationale
I find no circularity in the paper's derivation chain. The central concentration inequalities for the volume and intrinsic volumes are obtained by specializing the independent concentration inequality of Gieringer and Last [3], restated as Lemmas 2.2 and 2.3, to the cylinder-process functionals F and F_j. No parameter is fitted to a target tail: the constants alpha and beta in Corollaries 4.3 and 5.6 are closed-form expressions in the model parameters gamma, Q, W, M, and diam(W). The volume mean EF = lambda_d(W)(1 - exp(-gamma m_{d-k})) is quoted from the existing literature [5,17], and the intrinsic-volume means in Proposition 5.1 are derived, not assumed, from the principal kinematic formula in [16]. The lower-tail condition alpha*beta >= EF is proved in Corollary 4.3 using the isoperimetric inequality and monotonicity of a Gamma function, rather than imposed as an input. The geometric inequality (3.3) is indeed stated without proof and is load-bearing for the explicit constants, but it is a deterministic geometric bound independent of the probabilistic conclusion; if it failed the proof would collapse, but this is a correctness or completeness concern, not a circular reduction. Likewise, the possible ill-definedness of Theorem 5.4 for j=0 is a well-definedness issue, not a circularity. There are no relevant self-citations by the present authors, and no uniqueness theorem from the authors' prior work is invoked to force the choice of bound. The paper is self-contained against external benchmarks in the sense that the probabilistic input is an external theorem and the output is not normalized to reproduce a known result. Therefore the appropriate circularity score is 0.
Assumptions & free parameters
assumptions (7)
- standard math General concentration inequality for Poisson functionals from Gieringer and Last [3, Corollary 2.3].
- standard math Principal kinematic formula for cylinders from Schneider and Weil [16, Corollary 6.3.1].
- standard math Rotational integral formula and mean projection formula from [16, Theorems 6.1.1 and 6.2.2].
- standard math Isoperimetric inequalities for intrinsic volumes of convex bodies [16, Eq. (14.31)].
- standard math Steiner formula for convex bodies [16, Eq. (14.5)].
- standard math Monotonicity of Gamma(1+x/2)^(1/x) from [14, Theorem 1].
- domain assumption Model assumptions: stationarity of the marked Poisson process, condition (2.1) or m_{d-k}<infinity, convexity of Xi and rotational invariance of Q for intrinsic volumes, and convexity of W for intrinsic volumes.
Cite this review
Pith. "Pith review of Concentration inequalities for functionals of Poisson cylinder processes." pith.science (2026). https://pith.science/paper/YOBJSNRT
@misc{pith2026190802112,
author = {Pith},
title = {Pith review of: Concentration inequalities for functionals of Poisson cylinder processes},
year = {2026},
howpublished = {\url{https://pith.science/paper/YOBJSNRT}},
note = {Machine review of arXiv:1908.02112}
}
abstract
Random union sets $Z$ associated with stationary Poisson processes of $k$-cylinders in $\mathbb{R}^d$ are considered. Under general conditions on the typical cylinder base a concentration inequality for the volume of $Z$ restricted to a compact window is derived. Assuming convexity of the typical cylinder base and isotropy of $Z$ a concentration inequality for intrinsic volumes of arbitrary order is established. A number of special cases are discussed, for example the case when the cylinder bases arise from a random rotation of a fixed convex body. Also the situation of expanding windows is studied. Special attention is payed to the case $k=0$, which corresponds to the classical Boolean model.
Figures
Reference graph
Works this paper leans on
-
[3]
and Last, G.: Concentration inequalities for measures of a Boolean model
Gieringer, F. and Last, G.: Concentration inequalities for measures of a Boolean model. ALEA, Lat. Am. J. Probab. Math. Stat. 15, 151–166 (2018)
work page 2018
-
[1]
and Tykesson, J.: Connectedness of Poisson cylinders in Euclidean space
Borman, E.I. and Tykesson, J.: Connectedness of Poisson cylinders in Euclidean space. Ann. Inst. H. Poincaré Probab. Statist.52, 102–126 (2016)
work page 2016
-
[2]
Dissertation KIT, Karlsruhe (2016)
Gieringer, F.: Konzentrationsungleichungen für Poisson- und Binomialfunktionale in der Stochastischen Geometrie. Dissertation KIT, Karlsruhe (2016)
work page 2016
-
[4]
Heinrich, L.: Large deviations of the empirical volume fraction for stationary Poisson grain models. Ann. Appl. Probab. 15, 392–420 (2005)
work page 2005
-
[5]
Heinrich, L. and Spiess, M.: Berry-Esseen bounds and Cramér-type large deviations for the volume distri- bution of Poisson cylinder processes. Lithuanian Math. J.49, 381–398 (2009)
work page 2009
-
[6]
Heinrich, L. and Spiess, M.: Central limit theorems for volume and surface content of stationary Poisson cylinder processes in expanding domains. Adv. Appl. Probab.45, 312–331 (2013)
work page 2013
-
[7]
and Teixeira, A.: Cylinders percolation in three dimensions
Hilario, M.R., Sidoravicius, V. and Teixeira, A.: Cylinders percolation in three dimensions. Probab. Theory Relat. Fields 163, 613–642 (2015)
work page 2015
- [8]
Show all 21 references
-
[9]
Houdré, C.: Remarks on the deviation inequalities for functions of infinitely divisible random vectors. Ann. Probab. 30, 1223–1237 (2002)
2002
-
[10]
and Schulte, M.: Second-order properties and central limit theorems for geometric functionals of Boolean models
Hug, D., Last, G. and Schulte, M.: Second-order properties and central limit theorems for geometric functionals of Boolean models. Ann. Appl. Probab.26, 73–135 (2016)
2016
-
[11]
and Penrose, M.:Lectures on the Poisson Process
Last, G. and Penrose, M.:Lectures on the Poisson Process . Cambridge University Press (2018)
2018
-
[12]
Wiley (1975)
Matheron, G.: Random Sets and Integral Geometry . Wiley (1975)
1975
-
[13]
In:Stochastic Geometry
Miles, R.E.: A synopsis of Poisson flats in Euclidean spaces. In:Stochastic Geometry. A Tribute to the Memory of Rollo Davidson , edited by Harding, E.F. and Kendall D., Wiley (1974)
1974
-
[14]
and Guo, B.-N.: Complete monotonicities of functions involving the gamma and digamma functions
Qi, F. and Guo, B.-N.: Complete monotonicities of functions involving the gamma and digamma functions. RGMIA Research Report Collection7, article 8 (2004)
2004
-
[15]
and Statulevičius, V.A.:Limit Theorems for Large Deviations
Saulis, L. and Statulevičius, V.A.:Limit Theorems for Large Deviations . Kluwer Academic Publishers (1991)
1991
-
[16]
and Weil, W.:Stochastic and Integral Geometry
Schneider, R. and Weil, W.:Stochastic and Integral Geometry . Springer (2008)
2008
-
[17]
and Spodarev, E.: Anisotropic Poisson processes of cylinders
Spiess, M. and Spodarev, E.: Anisotropic Poisson processes of cylinders. Methodol. Comput. Appl. Probab. 13, 801–819 (2011)
2011
-
[18]
and Windisch D.: Percolation in the vacant set of Poisson cylinders
Tykesson, J. and Windisch D.: Percolation in the vacant set of Poisson cylinders. Probab. Theory Related Fields 154, 165–191 (2012)
2012
-
[19]
Acta Appl
Weil, W.: Point processes of cylinders, particles and flats. Acta Appl. Math.9, 103–136 (1987)
1987
-
[20]
Weil, W.: Iterations of translative integral formulae and non-isotropic Poisson processes of particles. Math. Z. 205, 531–549 (1990)
1990
-
[21]
Wu, L.: Anew modified logarithmicSobolev inequality forPoisson point processes andseveral applications. Probab. Theory Relat. Fields118, 427–438 (2000). 24
2000
Reviewed August 14, 2026 · model on record in the stance chip above.
Discussion (0). Continue with ORCID to comment.