Pointwise behaviour of M^1,1 Sobolev functions

From MaRDI portal
Publication:2384752





Let \((X,d)\) be a metric space equipped with a doubling measure \(\mu\). The space \(M^{1,1}(X)\), defined by \textit{P.\,Hajłasz} [Potential Anal. 5, No. 4, 403--415 (1996; Zbl 0859.46022)], is the Banach space of functions \(u\in L^{1}(X)\) for which there exists a generalized gradient \(g\in L^1(X)\) such that \[ | u(x)-u(y)| \leq d(x,y)(g(x)+g(y)) \] \(\mu\)-almost everywhere in \(X\), equipped with the norm \(\| u\|_{1,1}=\| u\|_{L^1(X)}+\inf_g\| g\|_{L^1(X)}\). It is known that \(M^{1,1}(\mathbb R^{n})\) does not coincide with the usual Sobolev space \(W^{1,1}(\mathbb R^n)\) and that the Hardy-Littlewood maximal operator is not bounded in \(L^1(X)\). The aim of the present paper is the extension to the case \(p=1\) of some properties of the spaces \(M^{1,p}(X)\) when \(p>1\). It is proved that \(u\in M^{1,1}(X)\) has a unique 1-quasicontinuous representative \(u^{*}(x)\) which is the limit, as \(r\to 0\), of the integral average over the ball with center \(x\) and radius \(r\) and therefore has Lebesgue points except on a set of 1-capacity zero. Moreover, \(u\) coincides on \(X\) with a Hölder continuous function, except on a set of small Hausdorff content. The main tools in the proofs are the discrete maximal operator \(\mathcal M^*\), defined in [\textit{J.\,Kinnunen} and \textit{V.\,Latvala}, Rev. Mat. Iberoam. 18, No. 3, 685--700 (2002; Zbl 1037.46031)], which is continuous in \(M^{1,1}(X)\), and an appropriate Sobolev-Poincaré type inequality.




Cited in
(26)








This page was built for publication: Pointwise behaviour of \(M^{1,1}\) Sobolev functions

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q2384752)