Faber-Krahn inequalities in sharp quantitative form

Eigenvalue problems for linear operators (47A75) Optimization of shapes other than minimal surfaces (49Q10) Variational problems in a geometric measure-theoretic setting (49Q20) Variational methods for eigenvalues of operators (49R05)

Abstract: The classical Faber-Krahn inequality asserts that balls (uniquely) minimize the first eigenvalue of the Dirichlet-Laplacian among sets with given volume. In this paper we prove a sharp quantitative enhancement of this result, thus confirming a conjecture by Nadirashvili and Bhattacharya-Weitsman. More generally, the result applies to every optimal Poincar'e-Sobolev constant for the embeddings

W_{0}^{1, 2} (O m e g a) h o o k r i g h t a r r o w L^{q} (O m e g a)

.

Review text

The aim of this paper is to prove the following quantitative version of the classical Faber-Krahn inequality: given an open set with finite measure \(\Omega\subset{\mathbb R}^N\) and first eigenvalue of the Dirichlet Laplacian \(\lambda(\Omega)\), there exists a dimensional constant \(\sigma>0\) such that \[ |\Omega|^{2/N}\lambda(\Omega)-|B|^{2/N}\lambda(B)\geq \sigma \mathcal{A}(\Omega)^2, \] where \(B\subset {\mathbb R}^N\) is an open ball and \(\mathcal{A}(\Omega)\) is the Frankel asymmetry of \(\Omega\), defined as the infimum of the quotients \(|\Omega\Delta B'|/|B'|\), where \(B'\) is a ball of volume \(|\Omega|\). As indicated, the power \(2\) on \(\mathcal{A}(\Omega)\) is sharp in the sense it that cannot be replaced by a smaller constant. The inequality was conjectured by \textit{T. Bhattacharya} and \textit{A. Weitsman} [Contemp. Math. 221, 31--58 (1999; Zbl 0914.31001)] and \textit{N. Nadirashvili} [in: Proceedings of the Ashkelon workshop on complex function theory, Ashkelon, Israel, 1996. Ramat-Gan: Bar-Ilan University, Isr. Math. Conf. Proc. 11, 197--201 (1997; Zbl 0890.35097)]. This quantitative version of the Faber-Krahn's inequality follows from a more general result stated as the main theorem of the paper: if \(1\leq q<2^*\), where \(2^*\) is defined as \(2N/(N-2)\) when \(N\geq 3\) and as \(+\infty\) if \(N=2\), then there is a constant \(\sigma_q>0\) depending only on \(q\) and \(N\) such that \[ |\Omega|^{{2\over N}+{2\over q}-1}\lambda_{2,q}(\Omega)-|B|^{{2\over N}+{2\over q}-1}\lambda_{2,q}(B)\geq\sigma_q\mathcal{A}(\Omega)^2, \] where \(\lambda_{2,q}:=\min_{u\in W_{0}^{1,2}(\Omega)}\{\int_\Omega |\nabla u|^2dx: ||u||_{L^q(\Omega)}=1\}\). The proof is not based on the qualitative extensions of the isoperimetric inequality (see [\textit{N. Fusco} et al., Ann. Math. (2) 168, No. 3, 941--980 (2008; Zbl 1187.52009)] and [\textit{A. Figalli} et al., Invent. Math. 182, No. 1, 167--211 (2010; Zbl 1196.49033)]), but on the selection principle by \textit{M. Cicalese} and \textit{G. P. Leonardi} [Arch. Ration. Mech. Anal. 206, No. 2, 617--643 (2012; Zbl 1257.49045)].

Recommendations

Cites work

Cited in

(83)

This page was built for publication: Faber-Krahn inequalities in sharp quantitative form

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q2354908)