English

Heat kernel for reflected diffusion and extension property on uniform domains

Probability 2024-01-29 v3 Analysis of PDEs Functional Analysis

Abstract

We study reflected diffusion on uniform domains where the underlying space admits a symmetric diffusion that satisfies sub-Gaussian heat kernel estimates. A celebrated theorem of Jones (Acta Math. 1981) states that uniform domains in Euclidean space are extension domains for Sobolev spaces. In this work, we obtain a similar extension property for metric spaces equipped with a Dirichlet form whose heat kernel satisfies a sub-Gaussian estimate. We introduce a scale-invariant version of this extension property and apply it to show that the reflected diffusion process on such a uniform domain inherits various properties from the ambient space, such as Harnack inequalities, cutoff energy inequality, and sub-Gaussian heat kernel bounds. In particular, our work extends Neumann heat kernel estimates of Gyrya and Saloff-Coste (Ast\'erisque 2011) beyond the Gaussian space-time scaling. Furthermore, our estimates on the extension operator imply that the energy measure of the boundary of a uniform domain is always zero. This property of the energy measure is a broad generalization of Hino's result (PTRF 2013) that proves the vanishing of the energy measure on the outer square boundary of the standard Sierpi\'nski carpet equipped with the self-similar Dirichlet form.

Keywords

Cite

@article{arxiv.2304.03908,
  title  = {Heat kernel for reflected diffusion and extension property on uniform domains},
  author = {Mathav Murugan},
  journal= {arXiv preprint arXiv:2304.03908},
  year   = {2024}
}

Comments

62 pages, 1 Figures, to appear in Probability Theory and Related Fields

R2 v1 2026-06-28T09:55:10.679Z