The fractional Laplacian in power-weighted $L^p$ spaces: integration-by-parts formulas and self-adjointness

TL;DR

This paper investigates the validity of integration-by-parts formulas for the fractional Laplacian in weighted $L^2$ and $L^p$ spaces, revealing conditions under which these formulas hold or fail, impacting the understanding of self-adjointness.

Contribution

It provides new results on when the classical integration-by-parts formula holds for fractional Laplacians in weighted spaces, and explores implications for operator self-adjointness.

Findings

Integration-by-parts formula validity depends on weight behavior at infinity.

Results extend to weighted $L^p$ spaces.

Implications for self-adjointness of fractional Laplacian operators.

Abstract

We consider the fractional Laplacian operator $(-\Delta)^s$ (let $ s \in (0,1) $) on Euclidean space and investigate the validity of the classical integration-by-parts formula that connects the $ L^2(\mathbb{R}^d) $ scalar product between a function and its fractional Laplacian to the nonlocal norm of the fractional Sobolev space $ \dot{H}^s(\mathbb{R}^d) $. More precisely, we focus on functions belonging to some weighted $ L^2 $ space whose fractional Laplacian belongs to another weighted $ L^2 $ space: we prove and disprove the validity of the integration-by-parts formula depending on the behaviour of the weight $ \rho(x) $ at infinity. The latter is assumed to be like a power both near the origin and at infinity (the two powers being possibly different). Our results have direct consequences for the self-adjointness of the linear operator formally given by $ \rho^{-1}(-\Delta)^s $. The generality of the techniques developed allows us to deal with weighted $ L^p $ spaces as well.