A sharp bound for the ratio of the first two Dirichlet eigenvalues of a domain in a hemisphere of 𝕊ⁿ (Q4523178)

Let \(\Omega\) be a bounded domain in \(\mathbb{R}^n\), and let \(\lambda_2/\lambda_1\) denote the ratio between the first two Dirichlet eigenvalues of \(\Omega\). The Payne-Polya-Weinberger conjecture [\textit{L. Payne, G. Pólya}, and \textit{H. Weinberger}, C. R. Acad. Sci., Paris 241, 917-919 (1955; Zbl 0065.08801)] is the following statement: \(\lambda_2/\lambda_1(\Omega)\leq \lambda_2/\lambda_1(\Omega^*),\) where \(\Omega^*\) is the \(n\)-ball having the same volume as \(\Omega\). This conjecture was proved by the present authors in [Bull. Am. Math. Soc. 25, 19-29 (1991; Zbl 0736.35075)] and [Ann. Math. (2) 135, 601-628 (1992; Zbl 0757.35052)] (for yet another proof, see [Commun. Math. Phys. 147, 181-190 (1992; Zbl 0758.34075)]). In the present paper, the authors consider the PWW conjecture on the sphere \(S^n\), and prove the following theorem. Theorem 1.4. Let \(\Omega\) be a domain contained in a hemisphere of \(S^n\). Then \[ \lambda_2/\lambda_1(\Omega)\leq \lambda_2/\lambda_1(\Omega^*), \] where \(\Omega^*\) is the geodesic ball in \(S^n\) having the same volume as \(\Omega\). Equality holds iff \(\Omega\) is itself a geodesic ball. The proof is an extension of their proof of the PWW conjecture in \(\mathbb{R}^n\). The first step is to show that, if \(B_{\lambda_1}\) is the geodesic ball which has the same value \(\lambda_1(\Omega)\) as its first eigenvalue, then: (2) \(\lambda_2(\Omega)\leq \lambda_2(B_{\lambda_1}).\) In the Euclidean case (2) immediately implies the PWW conjecture, because the ratio \(\lambda_2/\lambda_1\) is constant on Euclidean balls of arbitrary radius. In the spherical case, this property is no longer true, and the second main step of the proof is the following result. Theorem 1.3. For a geodesic ball in \(S^n\), contained in a hemisphere, the ratio \(\lambda_2/\lambda_1\) is an increasing function of the radius of the ball. \smallskip\textrm Taken together, Theorem 1.3 and (2) prove Theorem 1.4: \[ \lambda_2/\lambda_1(\Omega)\leq \lambda_2/\lambda_1(B_{\lambda_1})\leq \lambda_2/\lambda_1(\Omega^*), \] where the last inequality follows from Theorem 1.3 and the Faber-Krahn type result for domains in \(S^n\), previously proved by Sperner: \(\lambda_1(B_{\lambda_1})\geq \lambda_1(\Omega^*)\) which implies that, by monotonicity, radius \((B_{\lambda_1})\leq\) radius \((\Omega^*)\). Finally, in the introduction the authors discuss the following questions: a) Can one remove the restriction ``\(\Omega\) contained in a hemisphere''? b) Is it possible to prove similar results in other spaces, for example \(\mathbb{H}^n\)? About a), the authors say that they do not have any counterexamples to the ``full'' PWW conjecture on \(S^n\), although they point out that many geometric phenomena, which hold in a hemisphere, change drastically ``beyond the hemisphere''. For example, geodesic balls are no longer convex if they contain a hemisphere. About b), they observe that the PWW conjecture cannot hold in \(\mathbb{H}^n\), while inequality (2) could possibly hold, but has yet to be proved. The paper contains several other interesting estimates and facts about the eigenvalues of geodesic balls.