arxiv: 2605.10096 · v1 · submitted 2026-05-11 · 🧮 math.CO · math.MG

Recognition: 2 theorem links

· Lean Theorem

Randomly Shifted Steinhaus Longimeters and Buffon Discrepancy

Samuel Korsky

Pith reviewed 2026-05-12 03:13 UTC · model grok-4.3

classification 🧮 math.CO math.MG

keywords Buffon discrepancySteinhaus longimeterconvex domaindiscrepancy boundrandom shiftCrofton measureline intersectionsintegral geometry

0 comments

The pith

A randomly shifted Steinhaus longimeter improves the universal Buffon discrepancy bound from order L to the 1/3 to L to the 1/5 times (log L) to the 2/5 for every bounded convex domain.

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

The Buffon discrepancy problem asks for a one-dimensional set inside a convex domain whose line intersections closely track the average chord lengths given by Crofton measure, scaled by the total length L. Earlier constructions based on the Steinhaus longimeter achieved a universal error bound of order L to the one-third for any bounded convex shape. Introducing an independent random shift to the same construction yields a probabilistic improvement, lowering the worst-case bound to L to the one-fifth times (log L) to the two-fifths. A reader would care because the new guarantee applies uniformly to every convex domain rather than only to disks, where bounded discrepancy was already known.

Core claim

By taking a Steinhaus longimeter and translating it by a random vector chosen independently of the domain, the number of intersections with a random line deviates from the Crofton-normalized target by at most C L^{1/5} (log L)^{2/5} with positive probability; the same exponent holds for every bounded convex set in the plane.

What carries the argument

The randomly shifted Steinhaus longimeter, formed by taking the standard Steinhaus collection of segments and translating it by a uniform random vector so that intersection counts become random variables whose expectations and variances can be controlled uniformly.

If this is right

The Buffon discrepancy of the randomly shifted construction is at most order L^{1/5} (log L)^{2/5} with positive probability for any bounded convex domain.
The same construction works uniformly, without needing a different shift for each shape.
The exponent improvement is strictly better than the previous universal L^{1/3} result while still allowing the disk to retain its known bounded-discrepancy property.
Probabilistic averaging over shifts replaces deterministic worst-case arguments in the earlier Steinhaus analysis.

Where Pith is reading between the lines

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

The same randomization technique might reduce exponents in other integral-geometric discrepancy problems that count line intersections.
One could test whether repeated independent shifts or a more refined distribution on the shift vector produces still lower powers of L.
The result suggests that discrepancy bounds in convex geometry can often be improved by moving from deterministic to probabilistic constructions when uniform control over all domains is required.

Load-bearing premise

A single random shift, chosen without reference to the particular convex domain, suffices to make the probabilistic intersection estimates deliver the improved exponent for every bounded convex set at once.

What would settle it

Finding any bounded convex domain and length L large enough that every possible one-dimensional set of length L has Buffon discrepancy larger than C L^{1/5} (log L)^{2/5} for the implied constant C would falsify the claimed bound.

read the original abstract

Let $\Omega \subset \mathbb{R}^2$ be a bounded convex domain. Steinerberger (2026) introduced the Buffon discrepancy problem: given length $L$, construct a one-dimensional set $S\subset\Omega$ such that the number of intersections of $S$ with a line $\ell$ approximates the Crofton-normalized chord length $$ \frac{2L}{\pi|\Omega|}\cdot\mathcal{H}^1(\ell\cap\Omega).$$ Steinerberger proved a universal upper bound of order $L^{1/3}$ using a Steinhaus longimeter construction, and showed that the disk admits bounded discrepancy. We prove that a randomly shifted Steinhaus construction improves the order of the universal upper bound to $L^{1/5}(\log L)^{2/5}$.

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.

Desk Editor's Note private letter to a colleague

Random shift on the Steinhaus longimeter improves the universal Buffon discrepancy bound from L^{1/3} to L^{1/5}(log L)^{2/5}.

read the letter

The paper shows that a uniform random shift, chosen independently of the convex domain Ω, lets you average the intersection counts and extract a deterministic Steinhaus-type set whose discrepancy is O(L^{1/5}(log L)^{2/5}) uniformly in Ω. That is the concrete advance over Steinerberger's L^{1/3} construction. The argument balances the number of segments against the variance coming from the Crofton measure and uses a directional net plus union bound for the log factor. Because the shift period is fixed and the estimates rely only on convexity and boundedness, no hidden constants depending on the shape of Ω appear. The existence result follows directly from the probabilistic averaging, which is a clean way to upgrade the deterministic construction. The derivation looks reproducible from the stress-test description; the optimization step and the variance control are the parts that deliver the exponent gain. One minor soft spot is the log term itself. It comes from the net argument, and a more refined concentration or chaining might remove it, but that is an improvement rather than a flaw in what is proved. The paper does not claim optimality, so the current bound stands on its own. This is aimed at people working in geometric discrepancy or integral geometry who already know the Buffon problem and Steinerberger's earlier bound. A reader who cares about explicit rates in convex geometry will get a usable new exponent and a transparent probabilistic method. The work is grounded enough and the central claim is supported by the construction, so it deserves a serious referee rather than a desk rejection.

Referee Report

0 major / 3 minor

Summary. The paper claims to improve the universal upper bound on Buffon discrepancy from O(L^{1/3}) to O(L^{1/5}(log L)^{2/5}) for any bounded convex domain Ω ⊂ R². It does so by replacing Steinerberger's Steinhaus longimeter with a randomly shifted version whose shift is drawn uniformly from a fixed period independent of Ω; averaging the intersection statistics over the shift yields an existence result for a deterministic length-L set S whose discrepancy is controlled uniformly in Ω via convexity, boundedness, and an optimization of segment count against Crofton-measure variance (with the logarithmic factor arising from a union bound over a net of directions).

Significance. If the central estimates hold, the result meaningfully advances the Buffon discrepancy problem by exhibiting a strictly better universal exponent obtainable through a simple probabilistic averaging argument that introduces no hidden shape-dependent constants. The construction is parameter-free in the relevant sense and supplies an explicit existence proof rather than a fitted or self-referential bound.

minor comments (3)

The transition from the probabilistic averaging argument to the deterministic existence statement (via the probabilistic method) should be spelled out in a single dedicated paragraph, including the precise measure on the shift space.
Notation for the Crofton-normalized chord length and the discrepancy functional should be introduced once, early, and used consistently; the current abstract-to-introduction shift is slightly abrupt.
A brief remark on why the fixed-period shift can be chosen independently of Ω without losing uniformity would help readers who are not already familiar with the Steinhaus construction.

Simulated Author's Rebuttal

0 responses · 0 unresolved

We thank the referee for the positive summary, the recognition of the significance of the improved universal exponent, and the recommendation for minor revision. The referee's description accurately captures the probabilistic averaging argument over shifts and the resulting existence result for a deterministic set S.

Circularity Check

0 steps flagged

No significant circularity in the derivation

full rationale

The paper establishes the improved bound L^{1/5}(log L)^{2/5} via an explicit probabilistic construction: a Steinhaus longimeter with an independent uniform random shift (period fixed, independent of Ω), followed by averaging over the shift to obtain existence of a deterministic S. The exponent is derived by optimizing segment count against Crofton-measure variance and applying a union bound over a directional net; all estimates rest only on convexity and boundedness of Ω. No parameter is fitted to data and then renamed a prediction, no quantity is defined in terms of itself, and the cited Steinerberger (2026) result is used solely as a baseline comparison rather than a load-bearing premise. The argument is self-contained and does not reduce to its inputs by construction.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

The claim rests on the Crofton normalization (standard) and the existence of a suitable random shift whose intersection statistics improve the exponent; no free parameters or new entities are introduced in the abstract.

axioms (1)

standard math Crofton formula gives the normalized chord-length measure
Defines the target quantity that the intersection count must approximate.

pith-pipeline@v0.9.0 · 5425 in / 1114 out tokens · 81943 ms · 2026-05-12T03:13:12.409532+00:00 · methodology

discussion (0)

Lean theorems connected to this paper

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

IndisputableMonolith/Foundation/AlexanderDuality.lean alexander_duality_circle_linking unclear
We prove that a randomly shifted Steinhaus construction improves the order of the universal upper bound to L^{1/5}(log L)^{2/5}.

Reference graph

Works this paper leans on

6 extracted references · 6 canonical work pages

[1]

Steinerberger,Buffon Discrepancy and the Steinhaus Longimeter, arXiv:2603.27807, 2026

S. Steinerberger,Buffon Discrepancy and the Steinhaus Longimeter, arXiv:2603.27807, 2026

work page arXiv 2026
[2]

Hoeffding,Probability inequalities for sums of bounded random variables, Journal of the American Statistical Association58(1963), 13–30

W. Hoeffding,Probability inequalities for sums of bounded random variables, Journal of the American Statistical Association58(1963), 13–30

work page 1963
[3]

Chernoff,A measure of asymptotic efficiency for tests of a hypothesis based on the sum of observations, Annals of Mathematical Statistics23(1952), 493–507

H. Chernoff,A measure of asymptotic efficiency for tests of a hypothesis based on the sum of observations, Annals of Mathematical Statistics23(1952), 493–507

work page 1952
[4]

Alon and J

N. Alon and J. H. Spencer,The Probabilistic Method, 4th edition, Wiley, 2016

work page 2016
[5]

Matousek,Lectures on Discrete Geometry, Graduate Texts in Mathematics, vol

J. Matousek,Lectures on Discrete Geometry, Graduate Texts in Mathematics, vol. 212, Springer, 2002

work page 2002
[6]

Schneider,Convex Bodies: The Brunn–Minkowski Theory, 2nd expanded edition, Cam- bridge University Press, 2014

R. Schneider,Convex Bodies: The Brunn–Minkowski Theory, 2nd expanded edition, Cam- bridge University Press, 2014. 13

work page 2014