Adams inequality on metric measure spaces (Q734614): Difference between revisions

The Adams inequality [\textit{D.\,R.\thinspace Adams}, Stud.\ Math.\ 48, 99--105 (1973; Zbl 0237.46037)] says in \(\mathbb{R}^n\) that, for \(1 \leq p < q\) and \(p<n\), \(\| u \|_q \leq CM^{1/q} \| \nabla u \|_p\) for every \(u \in C^{\infty}_0(\mathbb{R}^n)\) whenever \(\nu\) is a Borel measure satisfying \[ \nu(B(x,r)) \leq Mr^{\alpha}, \, \alpha = q(n-p)/p.\tag{\(*\)} \] Here, \(\| u \|_q\) refers to the \(q\)-norm of \(u\) with respect to the measure \(\nu\). The author proves the corresponding inequality for Lipschitz functions with compact support in a complete metric measure space \((X, \mu)\) admitting a weak \((1, 1)\)-Poincaré inequality. The measure \(\mu\) is assumed to be a doubling Radon measure; for the assumptions on \(X\) and \(\mu\), see [\textit{P.\,Hajłasz} and \textit{P.\,Koskela}, ``Sobolev met Poincaré'' (Mem.\ Am.\ Math.\ Soc.\ 688) (2000; Zbl 0954.46022)]. The proofs for \(p = 1\) and \(p >1\) require a different treatment. Inequality (\(*\)) is expressed in terms of the measure \(\mu\). The proof makes use of the isoperimetric inequality and the co-area formula in \(X\) [see \textit{L.\,Ambrosio}, Set-Valued Anal.\ 10, No.\,2--3, 111--128 (2002; Zbl 1037.28002) and \textit{M.\,Miranda}, J.~Math.\ Pures Appl.\ (9) 82, No.\,8, 975--1004 (2003; Zbl 1109.46030)], as well as the Marcinkiewicz interpolation theorem.