Quantitative differentiability on uniformly rectifiable sets
Jonas Azzam, Mihalis Mourgoglou, Michele Villa
TL;DR
The paper develops a quantitative differentiability theory for Sobolev functions on uniformly rectifiable sets by proving a Dorronsoro-type estimate in the L^p setting. The authors introduce a novel square function G^q built from affine-deviation and flatness coefficients γ_f^q and Ω_f^q, and show that ||∇f||_{L^p(E)} is comparable to ||G^q f||_{L^p(E)} under UR and Poincaré hypotheses, extending classical Euclidean results to non-smooth sets. The core method combines good-λ inequalities, square-function estimates, stopping-time arguments, and a Lipschitz-graph corona decomposition to control multi-scale geometry and affine approximations via Tolsa’s α-numbers, Jones’ β-numbers, and the introduced γ-coefficients. A key result is the two-part achievement: (1) an L^p gradient estimate for Hajłasz-Sobolev functions on UR sets, and (2) a corresponding tangential-gradient bound, with corollaries on extensions and traces in domains with UR boundaries. The framework provides a robust bridge between geometric measure theory and quantitative differentiation, enabling extensions and trace results in domains with UR boundaries and offering tools for further analysis of elliptic problems on non-smooth sets.
Abstract
We prove $L^p$ quantitative differentiability estimates for functions defined on uniformly rectifiable subsets of the Euclidean space. More precisely, we show that a Dorronsoro-type theorem holds in this context: the $L^p$ norm of the gradient of a Sobolev function $f: E \to \mathbb{R}$ is comparable to the $L^p$ norm of a new square function measuring both the affine deviation of $f$ and how flat the subset $E$ is. A corollary dealing with extensions and traces of Sobolev functions may be found in a companion article.
