REVIEW 2 major objections 6 minor 5 references
Buffon Needle Problem Over Convex Sets
T0 review · 2 major / 6 minor · reviewed 2026-08-12 · deepseek-v4-flash
Pith's one-line read For any non-disk convex set of perimeter $2\pi$, the disk's Buffon containment probability is strictly larger for all sufficiently short needles.
desk verdict A clean little isoperimetric theorem: the disk wins Buffon's containment game for small needles; the proof is sound in substance but needs a coarea formula made explicit. 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 argument is carried by the inner parallel set $X_l=X\ominus B_l(0)$ (all points whose distance from the boundary is at least $l$) together with a collar estimate. On $X_l$ every needle of length $l$ is contained in $X$, so $p_X(x,l)=1$. On the outer collar $X\setminus X_l$, the pointwise containment probability is bounded by $\frac{1}{2\pi}(\pi+2\arcsin(d(x,\partial X)/l))$, and integrating this bound over collar layers, with each layer's boundary length below $\ell(\partial X)=2\pi$, yields $\int_{X\setminus X_l}p_X(x,l)\,dx\le 2\pi l-2l$. Steiner's parallel-body formulas then control the area of $X_l$, and the explicit disk formula computed from kinematic density supplies the benchmark; the comparison is decided by the slope $h'(0)=2/A(X)-2/\pi>0$, which is exactly the isoperimetric inequality at fixed perimeter.
What would settle it
Compute or simulate $P_X(l)$ for a non-disk convex set of perimeter $2\pi$, such as an ellipse, at needle lengths $l=0.01,0.02,\dots,0.1$; the theorem requires $P_D(l)-P_X(l)>0$ near zero and, in particular, the slope of $P_X$ at $0$ to be strictly smaller than the disk's slope $-2/\pi$. A non-disk whose small-$l$ curve meets or exceeds the disk curve would disprove the claim.
Extended reading notes
Core claim
The central claim is stated in Theorem 3.1: among all compact convex sets of perimeter $2\pi$, the disk maximizes the Buffon containment probability for every sufficiently small needle length. The authors establish this by proving the upper bound $P_X(l)\le [A(X_l)+2\pi l-2l]/A(X)$, where $X_l$ is the inner parallel set of points at least $l$ from the boundary, and by computing exactly $P_D(l)=\frac{2}{\pi}(\arccos(l/2)-\frac{l}{2}\sqrt{1-l^2/4})$ for the unit disk. Subtracting the bound from this formula gives a gap whose derivative at $l=0$ is $2/A(X)-2/\pi$, which is strictly positive for every non-disk by the isoperimetric inequality; hence the gap remains positive on a small interval. The line-segment case is handled separately and is immediate.
Load-bearing premise
The argument rests on the assumption that the total length of every inner parallel curve is bounded above by the original perimeter $2\pi$, so the probability lost near the boundary is at most $2\pi l-2l$.
Editorial extensions
If this is right
- For every compact convex set of perimeter $2\pi$ that is not a disk, $P_X(l)<P_D(l)$ holds for all sufficiently small $l$, so the disk is the unique small-needle maximizer.
- The gap is quantitative: $P_D(l)-P_X(l)\ge h(l)-l h'(0)/2$ with $h'(0)=2/A(X)-2/\pi>0$, meaning the loss grows at least linearly in $l$ with a coefficient set by the isoperimetric deficit.
- The explicit formula for $P_D(l)$ gives a closed-form benchmark that any numerical or experimental study of short-needle containment can be checked against.
- The result covers degenerate convex sets as well: if $X$ is a line segment then $P_X(l)=0$, so the inequality is immediate.
Reading between the lines
- Formally, across equal-perimeter sets the first-order containment loss is governed by the ratio of perimeter to area, so $P_X(l)=1-\ell(\partial X)l/(\pi A(X))+O(l^2)$ for smooth shapes; the disk's maximal area at fixed perimeter is what makes it the best short-needle habitat.
- The same layer-integration scheme should adapt to higher dimensions using the higher-dimensional Steiner formula the paper cites, giving a small-ball containment version of the isoperimetric inequality in $\mathbb{R}^n$.
- The authors conjecture a uniform threshold $\delta$ independent of shape; the local argument suggests such a uniform gap might follow from a bound on curvature or on the isoperimetric deficit, replacing the worst-case tangent-line bound in Lemma 3.3.
Editorial analysis
A structured set of objections, weighed in public.
Referee Report
Summary. The paper studies the probability that a randomly oriented needle of length l, with its base point chosen uniformly in a bounded convex set X ⊂ R^2, is entirely contained in X. The main result (Theorem 3.1) states that among compact convex sets with fixed perimeter 2π, the unit disk maximizes this containment probability for all sufficiently small needle lengths. The proof computes the disk probability explicitly (Lemma 3.2), derives a boundary-layer estimate for general convex sets (Lemma 3.3), uses an inclusion/Steiner-formula bound on the area of inner parallel sets (Lemma 3.4), and compares the resulting upper bound with the disk value via the classical isoperimetric inequality. Numerical experiments in Section 6 support a stronger global conjecture.
Significance. If the proof is completed, the result is a clean and natural isoperimetric inequality for a probabilistic quantity, and the argument is quantitative with no fitted parameters. The explicit disk computation and the boundary-layer method are elegant and could be useful in related problems. The main theorem is modest in scope but seems new, and the conjecture in Section 6 gives a clear direction for future work. The paper is written accessibly and provides machine-checkable derivations in the sense that the main steps are concrete analytic estimates rather than abstract existence arguments.
major comments (2)
- [Section 5, proof of Lemma 3.3] The boundary-layer estimate that drives Theorem 3.1 is not fully proven. After the pointwise bound p_X(x,l) ≤ g_X(dist(x,∂X),l), the proof jumps to ∫_{X\X_l} p_X(x,l) dx ≤ ∫_0^l ℓ(∂X_t) g_X(t,l) dt without stating or proving the layer-cake/coarea formula for the distance-to-boundary function. The displayed equation is garbled: it mixes 'dxdt', introduces an undefined '∂A', and the surrounding sentence 'Since X_t ⊂ X, we have ℓ(∂X_l) ≤ ℓ(∂X)' addresses a different inequality. Because the quantitative bound 2πl − 2l, and hence the positivity of h'(0) in Theorem 3.1, depends on this step, the proof is incomplete as written. Please insert a lemma establishing ∫_{X\X_l} φ(dist(x,∂X)) dx = ∫_0^l ℓ(∂X_t) φ(t) dt for bounded convex X, for example via the coarea formula for Lipschitz functions or a direct layer-cake argument, and correct the display.
- [Section 5, proof of Lemma 3.3 (pointwise bound)] The geometric proof of the pointwise bound p_X(x,l) ≤ (1/2π)(π + 2 arcsin(|x−y_x|/l)) is only sketched. In particular, the contradiction argument for a point z in the arc with nonzero first coordinate is hard to follow: the sentence 'Convexity and the assumption that yx ∈ ∂X restricts the boundary of X to the second quadrant' is not justified, and the graph Z is not defined precisely. Since the pointwise bound is the starting point of the entire lemma, please rewrite the argument, for instance by using the supporting line at y_x and the half-plane containment, which gives the inequality directly and avoids the ambiguous arc-intersection discussion.
minor comments (6)
- [Section 5, proof of Lemma 3.3] In the line 'Since X_t ⊂ X, we have ℓ(∂X_l) ≤ ℓ(∂X)', the subscript/superscript notation is inconsistent: it should refer to the same inner parallel parameter, e.g. ℓ(∂X_t) ≤ ℓ(∂X), and the sentence should be separated from the coarea step.
- [Section 5, proof of Lemma 3.4] The statement of Lemma 3.4 quantifies over all r>0, but for r exceeding the inradius of X the inner parallel set X_r is empty and A(X_r), ℓ(∂X_r) are undefined; please qualify the statement to 0<r<inradius or add a convention for the empty set.
- [Section 5, proof of Lemma 3.4] In the limiting argument there are several typos: 'ℓ(∂(Xr)2)' should be 'ℓ(∂(X_r)^r)', 'A1/N' should be 'X_{1/N}', and 'choose c large enough' should be 'choose c close enough to 1'.
- [Section 5, proof of Lemma 3.4] The polygonal approximation should explicitly include the closing segment |cf(t_k)-cf(t_0)|, and the inequality ℓ(∂C) ≥ Σ|cf(t_i)-cf(t_{i-1})| needs a short justification that the scaled boundary points lie in cyclic order on c∂X, so their convex hull has these points as vertices in that order.
- [Definitions and Theorem 3.1] The definition of P_X(l) via division by A(X) requires X to have nonempty interior, yet Theorem 3.1 and its proof treat line segments separately; please either assume X has nonempty interior or define the uniform distribution on degenerate sets by a limiting or Hausdorff-measure prescription.
- [Throughout] There are several typographical errors: 'countained' in the proof of Lemma 3.3, 'formlae' in Theorem 5.1, and 'we havePD > PX (l)' in Conjecture 6.1; these should be corrected.
Circularity Check
No significant circularity: the disk comparison is derived from an exact computation, a convex-geometry boundary bound, and the external isoperimetric inequality.
full rationale
The derivation chain is self-contained and non-circular. P_D(l) is computed directly in Lemma 3.2 by integrating the characteristic function over the unit disk. The upper bound for P_X(l) is obtained from Lemma 3.3, which bounds the integral of the pointwise probability over the boundary layer X\X_l using a geometric tangent-line argument plus the perimeter bound ℓ(∂X_t) ≤ ℓ(∂X) = 2π, valid for inner parallel sets of a convex body. Lemma 3.4 is an application of the externally cited Steiner formulae, not an assumption of the conclusion. The final small-l comparison uses the classical isoperimetric inequality (Theorem 3.5) to get h'(0) > 0 when X is not a disk. The proof never fits a parameter to P_D or P_X, and it does not invoke any prior work by the authors as a load-bearing premise. The only weakness is an unstated coarea/layer-cake justification in Lemma 3.3 and a minor misstatement about ℓ(∂X_l), but these are completeness gaps, not circular reductions; the displayed inequality ∫_{X\X_l} p_X dx = ∫_0^l ℓ(∂X_t)g_X(t,l)dt is the standard coarea formula for the distance function and is not equivalent to the theorem being proved.
Assumptions & free parameters
assumptions (5)
- standard math Isoperimetric inequality in the plane: for a simple closed curve C enclosing area A, L^2 ≥ 4πA, with equality iff C is a circle.
- standard math Steiner's formulae for convex bodies: A(K+B_r)=A(K)+L(∂K)r+πr^2 and L(∂(K+B_r))=L(∂K)+2πr.
- standard math Properties of metric projection onto convex sets: the closest boundary point yields a supporting line perpendicular to the segment, and the convex set lies in the corresponding half-plane.
- standard math Coarea formula for distance function on convex sets: ∫_{X\X_l} f(dist(x,∂X)) dx = ∫_0^l f(t) ℓ(∂X_t) dt, where ℓ(∂X_t) is the length of the inner parallel curve.
- standard math Monotonicity of perimeter under inclusion of convex sets: if K1 ⊂ K2 are convex, then ℓ(∂K1) ≤ ℓ(∂K2).
Cite this review
Pith. "Pith review of Buffon Needle Problem Over Convex Sets." pith.science (2026). https://pith.science/paper/HWOBLPQ7
@misc{pith2026241116935,
author = {Pith},
title = {Pith review of: Buffon Needle Problem Over Convex Sets},
year = {2026},
howpublished = {\url{https://pith.science/paper/HWOBLPQ7}},
note = {Machine review of arXiv:2411.16935}
}
abstract
We solve a variant of the classical Buffon Needle problem. More specifically, we inspect the probability that a randomly oriented needle of length $l$ originating in a bounded convex set $X\subset\mathbb{R}^2$ lies entirely within $X$. Using techniques from convex geometry, we prove an isoperimetric type inequality, showing that among sets $X$ with equal perimeter, the disk maximizes this probability.
Figures
Reference graph
Works this paper leans on
-
[1]
Steiner's formula and a variational proof of the isoperimetric inequality
Joseph Ansel Hoisington. Steiner’s formula and a variational proof of the isoperimetric inequality.arXiv preprint, 1909.06347, 2019. 2, 6
work page Pith review arXiv 1909
-
[2]
Santal´ o.Integral Geometry and Geometric Probability
Luis A. Santal´ o.Integral Geometry and Geometric Probability. 2nd edition, Cambridge Mathematical Library, Cambridge University Press, 2004. 5
work page 2004
-
[3]
Introduction to the Calculus of Variations
Bernard Dacorogna. Introduction to the Calculus of Variations. 2009. 8
work page 2009
-
[4]
Einfacher Beweis der isoperimetrischen Haupts¨ atze
Jakob Steiner. Einfacher Beweis der isoperimetrischen Haupts¨ atze. Journal f¨ urdie reine und angewandte Mathematik, 18:281–296, 1838. 1
-
[5]
Robert Osserman. The Isoperimetric Inequality. Bulletin of the American Mathematical Society, 84(6):1182–1238, 1978. http://www.ams.org/journals/bull/1978-84-06/ S0002-9904-1978-14553-4/S0002-9904-1978-14553-4.pdf . 1, 3 Department of Mathematics, University of Rochester, Rochester, NY Email address: iosevich@math.rochester.edu
work page 1978
Reviewed August 12, 2026 · model on record in the stance chip above.
Discussion (0). Continue with ORCID to comment.