Rectifiability of the straight visible boundary and the limits of segment-based Hardy criteria

Technical Summary: Rectifiability of the Straight Visible Boundary and the Limits of Segment-Based Hardy Criteria

Problem Statement
The paper investigates the geometric conditions under which a domain $\Omega \subset \mathbb{R}^n$ admits the weighted $(p, \beta)$-Hardy inequality:
$$\int_{\Omega} |u(x)|^p d_{\Omega}(x)^{\beta-p} dx \leq C \int_{\Omega} |\nabla u(x)|^p d_{\Omega}(x)^{\beta} dx$$
where $d_{\Omega}(x) = \text{dist}(x, \partial \Omega)$. Modern sufficient conditions for this inequality rely on lower bounds for the Hausdorff content of the boundary. Two primary approaches exist:

1. Density Criteria: Lower bounds on the Hausdorff content of the full boundary $\partial \Omega$ (Lehrbäck [8]). These are effective for exponents $\beta < p - 1$ but fail to guarantee the inequality for $\beta \geq p - 1$.
2. Visibility Criteria: Lower bounds on the "visual boundary" $v_x(c)\text{-}\partial \Omega$, consisting of points reachable from $x$ by curves satisfying a quantitative John-type condition (Koskela and Lehrbäck [5]). This approach is necessary for the regime $\beta \geq p - 1$.

Since verifying curve-based visibility is geometrically complex, the authors examine whether "straight" visibility—replacing curves with line segments—can serve as a simpler, equivalent criterion. Specifically, they analyze the quantitatively straight visible boundary $\partial^{str}_{x, \alpha}\Omega$, defined as boundary points $\xi$ such that the segment $[x, \xi)$ lies in $\Omega$ and satisfies a uniform thickness condition: $d_{\Omega}((1-t)x + t\xi) \geq \alpha \min\{t, 1-t\}|x-\xi|$.

Methodology
The authors employ geometric analysis and measure theory to characterize the structure of $\partial^{str}_{x, \alpha}\Omega$.

1. Cone Property: They prove that the uniform thickness condition along a segment implies the existence of an interior truncated cone at the boundary point $\xi$ (Theorem 3.1).
2. Rectifiability: Using the cone property, they demonstrate that a set of boundary points, each carrying an interior cone with axes confined to a small angular sector, forms a Lipschitz graph. By covering the sphere of directions with finitely many sectors, they show that the straight visible boundary is a finite union of Lipschitz graphs (Theorem 3.3).
3. Measure Bounds: They establish that such sets are $(n-1)$-rectifiable and carry zero $\lambda$-Hausdorff content for any $\lambda > n-1$.
4. Counterexamples and Sharpness: The authors construct specific domains (convex domains, Lipschitz graph domains, and a planar domain with "teeth") to test the sharpness of their results and the strictness of the inclusion between straight and curve-based visibility.

Key Results

1. Rectifiability Theorem (Theorem 1.1 / 3.3):
   For any domain $\Omega$, point $x \in \Omega$, and parameter $\alpha \in (0, 1]$, the set $\partial^{str}_{x, \alpha}\Omega \cap B(x, C_0 d_{\Omega}(x))$ is contained in a finite union of Lipschitz graphs. Consequently, it is $(n-1)$-rectifiable, and its $\lambda$-Hausdorff content vanishes for all $\lambda > n-1$.

2. The Dichotomy of Segment-Based Criteria (Theorem 4.2):
   The paper establishes a fundamental limitation for criteria based on $\partial^{str}_{x, \alpha}\Omega$:

   • Case $\lambda > n-1$: No domain satisfies a lower Hausdorff content bound on $\partial^{str}_{x, \alpha}\Omega$ with exponent $\lambda > n-1$. The condition is unsatisfiable because the set is too "thin" (dimension $n-1$).
   • Case $\lambda \leq n-1$: If a domain satisfies the condition, it implies the standard density condition on the full boundary $\partial \Omega$. Thus, the segment-based criterion yields no new information beyond what is already known from density criteria alone, which do not require visibility assumptions.
   • Conclusion: Segment-based criteria are either unsatisfiable in the regime where visibility is genuinely needed ($\beta \geq p-1$, corresponding to $\lambda > n-1$) or redundant in the regime where they can be satisfied ($\beta < p-1$).

3. Sharpness of the Exponent (Theorem 4.4, 4.6):
   The bound $\lambda = n-1$ is attained. Convex domains and Lipschitz graph domains realize this exponent, confirming that the rectifiability result is sharp and the dichotomy is not an artifact of loose estimates.

4. Strictness of Inclusion (Theorem 5.1):
   The inclusion $\partial^{str}_{x, \alpha}\Omega \subset v_x(\alpha^{-1})\text{-}\partial \Omega$ is strict. The authors construct a planar domain where a boundary point $\xi$ is directly visible from $x$ and reachable by a John-type curve, yet fails to belong to $\partial^{str}_{x, \alpha}\Omega$ for any $\alpha$. This point lacks an interior cone, illustrating the geometric gap between straight segments and general curves.

5. Raw Straight Visibility (Section 6):
   If the thickness condition is removed (considering $\partial^{dir}_{x}\Omega$), the cone obstruction disappears. The authors note that for $\lambda \leq n-1$, the criterion still holds trivially. However, for $\lambda > n-1$, the question remains open: can a "raw" straight visible boundary (without thickness) carry $\lambda$-content for $\lambda > n-1$? This is linked to whether radial graphs over fractal boundaries can be uniformly visible.

Significance and Claims
The paper claims to demonstrate that the apparent gain of using straight segments to simplify visibility conditions is illusory.

• Intrinsic Obstruction: The requirement of uniform thickness along a segment intrinsically forces the boundary to be $(n-1)$-rectifiable. This geometric constraint prevents the straight visible boundary from supporting the higher-dimensional content ($\lambda > n-1$) required to prove Hardy inequalities in the critical regime $\beta \geq p-1$.
• Limit of Segments: The authors conclude that curve-based visual boundaries cannot be replaced by segments in the regime where visibility is essential. The "thickness" condition, often viewed as a technical convenience, is the very mechanism that creates the obstruction.
• Open Problem: The paper identifies that the only potential avenue for a segment-based criterion to reach $\beta \geq p-1$ is to drop the thickness condition entirely, leaving the validity of such a criterion for $\lambda > n-1$ as an open problem.

The work serves to delimit the reach of straight-line visibility in the theory of weighted Hardy inequalities, clarifying that the complexity of curve-based conditions is not merely a technical hurdle but a geometric necessity for handling fractal or high-dimensional boundary structures.