How to Recognise Extension domains

Abstract: Let $Ω\subset \mathbb{R}^n$ be a bounded domain and $1 < p < \infty$. We characterize $(1,p)$-extension domains in terms of inequalities of Bourgain--Brezis--Mironescu type. More precisely, we show that $Ω$ is a $(1,p)$-extension domain if and only if it is Ahlfors regular and satisfies, for all $f \in \dot{W}^{1,p}(Ω)$, [(1-s)[f]{W^{s,p}(Ω)}^p \leq C [f]{W^{1,p}(Ω)}^p,] for all $s$ sufficiently close to $1$, where $C > 0$ is a constant independent of $s$ and $f$. As a key ingredient, we establish a fractional Poincaré-type inequality under the assumption of Ahlfors regularity alone, improving a result of Ponce (2004). As a further application, we prove that, under a mild Hausdorff measure condition on the boundary $\partial Ω$, fractional extension (from $\dot{W}^{1,p}(Ω)$ to $\dot{W}^{s,p}(\mathbb{R}^n)$) at a single exponent $s > 1/p$ self-improves to full first-order Sobolev extension (from $\dot{W}^{1,p}(Ω)$ to $\dot{W}^{1,p}(\mathbb{R}^n)$). These results clarify the role of nonlocal estimates in the geometry of Sobolev extension domains.

• The paper establishes that a BBM-type fractional Sobolev inequality, combined with Ahlfors regularity, is necessary and sufficient for a domain to be a (1,p)-extension domain.
• It develops a fractional Poincaré inequality under minimal geometric assumptions to control constants in constructing extension operators.
• The self-improvement result demonstrates that a single fractional extension property can bootstrap to full first-order extension under precise boundary conditions.

Characterizing Sobolev Extension Domains via Nonlocal Inequalities

Introduction and Context

The paper "How to Recognise Extension domains" (2604.23598) addresses the characterization of $(1,p)$-extension domains in $\mathbb{R}^n$—that is, domains $\Omega$ for which every $f \in \dot W^{1,p}(\Omega)$ admits an extension $Ef$ to $\mathbb{R}^n$ such that $Ef \in \dot W^{1,p}(\mathbb{R}^n)$ and $Ef|_\Omega = f$, with a norm bound controlled by $f$. While classical results provide geometric conditions such as Lipschitz or $(\varepsilon,\delta)$ domains that guarantee the extension property, this work seeks a functional analytic characterization in terms of fractional Sobolev inequalities of Bourgain–Brezis–Mironescu (BBM) type.

Specifically, the authors prove that on Ahlfors regular domains, the existence of a BBM-type inequality with sharp dependence on the order $\mathbb{R}^n$0 serves as a necessary and sufficient condition for the domain to admit a Sobolev extension operator. This result brings nonlocal, scale-dependent estimates to the fore as powerful analytic probes for the geometry of extension domains, and offers new perspectives on the interplay between function space theory and geometric measure theory.

Main Results

Sharp Characterization of Extension Domains

Let $\mathbb{R}^n$1 be a bounded open set and $\mathbb{R}^n$2. The principal result establishes the following equivalence:

1. $\mathbb{R}^n$3 is a $\mathbb{R}^n$4-extension domain.
2. $\mathbb{R}^n$5 is Ahlfors regular and satisfies, for all $\mathbb{R}^n$6 and all $\mathbb{R}^n$7 close to $\mathbb{R}^n$8,

$\mathbb{R}^n$9

with $\Omega$0 independent of $\Omega$1 and $\Omega$2.

1. The same as (2) but for $\Omega$3 in the non-homogeneous $\Omega$4 with norms on both sides.

Thus, the BBM inequality with the correct scaling—as $\Omega$5—plus Ahlfors regularity, is a complete functional characterization of extension domains within this geometric class.

The argument is robust: neither hypothesis can be dropped. Examples are given where each fails independently and the extension property breaks down, highlighting the necessity and sharpness of the result.

Fractional Poincaré Inequality under Weak Assumptions

A notable technical contribution is the derivation of a fractional Poincaré-type inequality that requires only Ahlfors regularity, not the full extension property. This strengthens prior results (notably Ponce 2004) and is essential for controlling the constants in extension operator constructions.

For $\Omega$6 and $\Omega$7 near $\Omega$8, the inequality

$\Omega$9

holds for any ball $f \in \dot W^{1,p}(\Omega)$0, with constants depending only on Ahlfors regularity. Here, $f \in \dot W^{1,p}(\Omega)$1 and $f \in \dot W^{1,p}(\Omega)$2 denotes the mean of $f \in \dot W^{1,p}(\Omega)$3 on $f \in \dot W^{1,p}(\Omega)$4.

Self-Improvement: From Fractional to First-Order Extension

If for some $f \in \dot W^{1,p}(\Omega)$5 the domain $f \in \dot W^{1,p}(\Omega)$6 admits an extension from $f \in \dot W^{1,p}(\Omega)$7 to $f \in \dot W^{1,p}(\Omega)$8 at a single fixed exponent $f \in \dot W^{1,p}(\Omega)$9, then—provided the boundary has sufficiently small Hausdorff measure ($Ef$0)—this property self-improves: the domain is a true $Ef$1-extension domain. The proof relies on slicing techniques, fine measure estimates on the boundary, and refined extension operator constructions without excessive geometric regularity.

Counterexamples and Necessity

The analysis includes explicit constructions (e.g., cusp and slit-disk domains) showing that:

• The BBM inequality can hold in the absence of extension properties if Ahlfors regularity fails.
• Ahlfors regularity alone does not suffice: some domains satisfy the measure condition but violate BBM, and hence the extension property fails.

Implications and Discussion

Functional vs. Geometric Characterizations

By tying the extension property to nonlocal analytic inequalities, the paper closes the gap between classical geometric criteria (such as Lipschitz or $Ef$2 domain conditions) and the more modern approach via function space embeddings. This exposes how fine-scale geometric structure (captured by Ahlfors regularity) and scale-invariant nonlocal estimates together suffice to diagnose the extendability of Sobolev functions.

Practical Significance

The unified characterization gives practitioners a powerful analytic test to recognize extension domains through verification of the BBM inequality and measure regularity, rather than detailed geometric analysis of boundaries. This approach may be particularly advantageous in the analysis of irregular or fractal domains in applications to PDE, calculus of variations, and geometric analysis.

Theoretical Developments

• The results clarify the prominence of nonlocal estimates (fractional seminorms) as robust probes of domain geometry, with direct implications for the study of nonlocal operators and equations.
• The "self-improvement" property suggests that fractional regularity conditions on extension can, under controllable boundary measure assumptions, bootstrap to full first-order extension—a phenomenon reminiscent of self-improving properties in harmonic analysis and quasiconformal theory.
• The refined Poincaré estimates without the extension property assumption broaden the toolkit for analysis in metric measure spaces with minimal geometry.

Future Directions

The techniques and characterizations developed could be extended in several directions:

• Metric Measure Spaces: Adaptation of these results to more general settings, such as doubling metric measure spaces or Carnot groups, where analogues of the Sobolev and fractional Sobolev spaces have been studied.
• Sharpness of Boundary Conditions: Further analysis of the boundary measure threshold in the self-improvement theorem to determine necessity and optimality.
• Algorithmic Detection: Leveraging the BBM-type inequalities for algorithmic criteria in numerical analysis, where automated recognition of extension domains is beneficial for finite element and variational methods.
• Nonlocal PDE: Application to nonlocal boundary value problems, where domain geometry and fractional order Sobolev spaces play a pivotal role.

Conclusion

The paper provides a sharp, analytic characterization of $Ef$3-extension domains via BBM-type inequalities and Ahlfors regularity, bridging geometry and analysis in the theory of Sobolev spaces. It enhances the understanding of the geometry-function interplay underlying extension problems, offers optimal fractional Poincaré inequalities under weak hypotheses, and demonstrates a self-improvement effect from fractional to full first-order extension under mild boundary measure assumptions. These contributions deepen connections across analysis, geometric measure theory, and the burgeoning field of nonlocal functional analysis.