A Stability Version of the Jones Opaque Set Inequality
Stefan Steinerberger
TL;DR
This paper addresses the opaque-set problem for convex planar domains, focusing on the Jones inequality $L \geq \frac{|\,\partial\Omega\,|}{2}$ and its (long-standing) difficulty of improvement. It introduces an angular-measure framework by assigning to any opaque set $\mathcal{O}$ a measure $\mu_{\mathcal{O}}$ on $[0,2\pi)$ and to the boundary $\mu_{\partial\Omega}$, with total masses $L$ and $|\partial\Omega|/2$, respectively. The authors prove a quantitative stability bound in the Sobolev space $\dot H^{-2}(\mathbb{T})$: $\|\mu_{\mathcal{O}} - \mu_{\partial\Omega}\|_{\dot H^{-2}} \leq \frac{L^{1/4}}{\sqrt{2}}\left(L - \frac{|\partial\Omega|}{2}\right)^{3/4}$, linking near-optimal length to a near-equivalence of angular distributions via Fourier analysis. The results yield structural information about near-optimal opaque sets (e.g., for $\Omega=[0,1]^2$, the orientation distribution must be nearly axis-aligned) and suggest a broader geometric-analytic program to understand opaque sets through $\mathbb{R}^2 \times \mathbb{S}^1$ viewpoints and energy methods, with potential implications for disks and other shapes.
Abstract
Let $Ω\subset \mathbb{R}^2$ be a bounded, convex set. A set $O \subset \mathbb{R}^2$ is an opaque set (for $Ω$) if every line that intersects $Ω$ also intersects $O$. What is the minimal possible length $L$ of an opaque set? The best lower bound $L \geq |\partial Ω|/2$ is due to Jones (1962). It has been remarkably difficult to improve this bound, even in special cases where it is presumably very far from optimal. We prove a stability version: if $L - |\partial Ω|/2$ is small, then any corresponding opaque set $O$ has to be made up of curves whose tangents behave very much like the tangents of the boundary $\partial Ω$ in a precise sense.
