A new universal inequality for Neumann eigenvalues of the Laplacian on a planar convex domain

Kei Funano · 2025

Read A new universal inequality for Neumann eigenvalues of the Laplacian on a planar convex domain by Kei Funano (2025) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2511-18043.txt.

Introduction

Kei Funano establishes two new universal inequalities for Neumann eigenvalues of the Laplacian on a planar convex domain. Eigenvalues in this setting are the natural vibration frequencies of a membrane with a free edge, and the Laplacian is the core operator behind them. Universal inequalities bound such values in ways that hold for every domain in the class, regardless of its exact shape. The note states the two results for flat, convex regions, adding to the family of shape-free bounds known for this problem.

Catalog record

CREATORKei Funano
YEAR2025
COLLECTIONPAPERS
FORMATTXT
FILE SIZE10 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2511-18043

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

A new universal inequality for Neumann eigenvalues of the Laplacian on a planar convex domain Kei Funano Abstract . In this paper we establish a new universal inequality for Neumann eigenvalues of the Laplacian on a planar convex domain. 2010 Mathematics Subject Classification 35P15, 53C23, 58J50 1. Introduction Let \Omega be a bounded domain in \mathbb{R}^{n} with Lipschitz boundary. We consider the following Neumann eigenvalue problem of the Laplacian: (1.1) \begin{cases}-\Delta u=\mu u\quad\text{in }\Omega,\\ \partial_{n}u=0\quad\text{on }\partial\Omega,\end{cases} where \partial_{n} denotes the directional derivative with respect to n , the outward unit normal vector to \partial\Omega . In this situation it is known the compactness of the embedding H^{1}(\Omega) to L^{2}(\Omega) and it gives that the eigenvalues of ( 1.1 ) can be listed with multiplicity as follows: \displaystyle 0=\mu_{0}(\Omega)<\mu_{1}(\Omega)\leq\mu_{2}(\Omega)\leq\dots\leq\mu_{k}(\Omega)\leq\mu_{k+1}(\Omega)\leq\dots\to\infty. Our main concern in this paper is the relation between those eigenvalues. The main theorem is the following. We denote by \alpha\lesssim\beta if \alpha\leq c\beta for some universal concrete constant c>0 (which means c does not depend on any parameters such as dimension and k etc). Theorem 1.1 . Let \Omega be a planar bounded convex domain. Then for any k\geq l we have \displaystyle\mu_{k}(\Omega)\lesssim\Big(\frac{k}{l}\Big)^{2}\mu_{l}(\Omega). Let us mention the related known inequalities between the eigenvalues. In [ 10 , Theorem 1.1] , Liu showed the sharp dimension-free universal inequality \mu_{k}(\Omega)\lesssim k^{2}\mu_{1}(\Omega) for any bounded convex domain \Omega in \mathbb{R}^{n} . In [ 4 , Theorem 1.2] the author proved that (1.2) \displaystyle\mu_{k+1}(\Omega)\lesssim n^{4}\mu_{k}(\Omega) under the same situation of that of Liu’s. The above theorem unifies the Liu inequality and the inequality ( 1.2 ) in the planar case. In the proof we use a variant of domain monotonicity for eigenvalues to reduce the problem in the case of rectangular domains. In this aspect we give a new brief proof for an weaker inequality of the inequality ( 1.2 ) in the last section (see Theorem 4.1 ). It is weaker but its proof is new and rather shorter than the proof of the inequality ( 1.2 ) so the author include it in this paper. 2. Preliminaries We recall several results which will be used in the proof of main theorems. In the following proposition Hatcher showed that bounded convex domains can be approximated by rectangular domains in some sense. He used the John theorem which asserts that bounded convex domains can be approximated by ellipsoids ( [ 6 , Theorem III] ). Proposition 2.1 ( [ 5 , Proposition 2.3] ) . Let \Omega be a bounded convex domain in \mathbb{R}^{n} . Tlanslating \Omega if necessary we can find a sequence a_{1},a_{2},\cdots,a_{n} of positive real numbers such that \displaystyle\Big[-\frac{a_{1}}{\sqrt{n}},\frac{a_{1}}{\sqrt{n}}\Big]\times\cdots\times\Big[-\frac{a_{n}}{\sqrt{n}},\frac{a_{n}}{\sqrt{n}}\Big]\subseteq\Omega\subseteq[-a_{1}n,a_{1}n]\times\cdots[-a_{n}n,a_{n}n]. Combining the above proposition with the next theorem we see that the proof of the main theorem reduce to the case where the domain is rectanglular. It is a variant of domain monotonicity, it states that the smaller domain has a larger eigenvalue up to a multiplicative constant factor under convexity assumption. Theorem 2.2 ( [ 3 , Theorem 1.1] ) . For any two bounded convex domains \Omega\subseteq\Omega^{\prime} in \mathbb{R}^{n} their Neumann eigenvalues of the Laplacian satisfy \displaystyle\mu_{k}(\Omega^{\prime})\lesssim n^{2}\mu_{k}(\Omega). For a convex domain Neumann eigenvalues of the Laplacian are related with diameter as follows: Theorem 2.3 (Payne–Weinberger, [ 11 , (1.9)] ) . Suppose that \Omega is a bounded convex domain in \mathbb{R}^{n} . Then we have \mu_{1}(\Omega)\geq\frac{\pi^{2}}{(\operatorname{Diam}\Omega)^{2}} . In the following theorem j_{\nu,k} denotes the k -th positive zero of the Bessel function J_{\nu} of order \nu . Theorem 2.4 (Kröger, [ 7 , Theorem 1] ) . Let \Omega be a bounded convex domain in \mathbb{R}^{n} . Assume that n>2 . Then we have (1) (\operatorname{Diam}\Omega)^{2}\mu_{k}(\Omega)\leq 4j_{\frac{n-2}{2},\frac{k+1}{2}}^{2} if k is odd, and (2) (\operatorname{Diam}\Omega)^{2}\mu_{k}(\Omega)\leq(j_{\frac{n-2}{2},\frac{k}{2}}+j_{\frac{n-2}{2},\frac{k+2}{2}})^{2} if k is even. Now assume that n=2 . Then we have \displaystyle(\operatorname{Diam}\Omega)^{2}\mu_{k}(\Omega)\leq(2j_{0,1}+(k-1)\pi)^{2}\lesssim k^{2}. Kröger proved the following upper bound for Neumann eigenvalues in terms of volume using harmonic analysis. Theorem 2.5 (Kröger, [ 8 , Corollary 2] ) . Let \Omega be a (not necessarily convex) bounded domain in \mathbb{R}^{n} with Lipschitz boundary. Then for any natural number k we have \displaystyle\mu_{k}(\Omega)\leq(2\pi)^{2}\Big(\frac{n+2}{2}\Big)^{\frac{2}{n}}\Big(\frac{k}{\omega_{n}\operatorname{vol}\Omega}\Big)^{\frac{2}{n}}\lesssim\Big(\frac{k}{\omega_{n}\operatorname{vol}\Omega}\Big)^{\frac{2}{n}}, where \omega_{n} is the volume of a unit ball in \mathbb{R}^{n} . Let \Omega be a bounded domain in a Euclidean space and \{\Omega_{i}\}_{i=0}^{l} be a finite partition of \Omega by subdomains; \Omega=\bigcup_{i}\Omega_{i} and \operatorname{vol}(\Omega_{i}\cap\Omega_{j})=0 for different i\neq j . We need the following Buser theorem to give a lower bound of eigenvalues of the Laplacian. Theorem 2.6 (Buser, [ 1 , 8.2.1 Theorem] ) . Under the above situation we have \displaystyle\mu_{k}(\Omega)\geq\min_{i}\mu_{1}(\Omega_{i}). 3. Proof of the Main Theorem In this section we prove the main theorem. Proof of Theorem 1.1 . By virtue of Proposition 2.1 and Theorem 2.2 we may assume that our \Omega is a rectangular region, say \Omega=[-a,a]\times[-b,b] . Without loss of generality we may also assume that a\leq b . Put R:=\frac{Ck}{l\sqrt{\mu_{k}(\Omega)}} , where C>0 is a sufficiently large constant which will be specified later. We want to show that \Omega can be decomposed into l^{\prime} convex subdomains \{\Omega_{i}\}_{i=1}^{l^{\prime}} such that \operatorname{Diam}\Omega_{i}^{\prime}\lesssim R and l^{\prime}\leq l . If this holds, then using Buser’s theorem (Theorem 2.6 ) and the Payne-Weinberger inequality (Theorem 2.3 ) we obtain \displaystyle\mu_{l}(\Omega)\geq\mu_{l^{\prime}}(\Omega)\geq\min_{i}\mu_{1}(\Omega_{i})\gtrsim\frac{1}{R^{2}}=\frac{l^{2}\mu_{k}(\Omega)}{C^{2}k^{2}}, which completes the proof of the theorem. We first consider the case where a\leq 2R . In this case since Kröger’s inequality (Theorem 2.4 ) gives \displaystyle R\gtrsim\frac{Ck}{l\frac{k}{\operatorname{Diam}\Omega}}=\frac{C\operatorname{Diam}\Omega}{l}\gtrsim\frac{C(2b)}{l}, if C>0 is large enough, then we can have a partition \displaystyle[-b,b]=[b_{1},b_{2}]\cup[b_{2},b_{3}]\cup\cdots\cup[b_{l^{\prime}},b_{l^{\prime}+1}], such that l^{\prime}\leq l and b_{i+1}-b_{i}\leq R . Thus setting \Omega_{i}:=[-a,a]\times[b_{i},b_{i+1}] we get the desired convex partition of \Omega ; \Omega=\bigcup_{i=1}^{l^{\prime}}\Omega_{i} , where \operatorname{Diam}\Omega_{i}\lesssim R . Let us consider the case where a>2R . We put \displaystyle\Omega^{\prime}:=\{x\in\Omega\mid d(x,\partial\Omega)\geq R\}, where \displaystyle d(x,\partial\Omega):=\inf\{|x-y|\mid y\in\partial\Omega\}. Note that \Omega^{\prime} is a nonempty (noncollapsed) rectangular region. Let \{x_{i}\}_{i=1}^{l^{\prime}} be a maximal (2R) -separated points in \Omega^{\prime} . Since \{B(x_{i},R)\}_{i=1}^{l^{\prime}} are disjoint balls in \Omega , we have \displaystyle\frac{l^{\prime}\omega_{2}(Ck)^{2}}{l^{2}\mu_{k}(\Omega)}=l^{\prime}\omega_{2}R^{2}=\sum_{i=1}^{l^{\prime}}\operatorname{vol}B(x_{i},R)\leq\operatorname{vol}\Omega. Applying Theorem 2.5 to the above inequality we have \displaystyle l^{\prime}\lesssim\frac{l}{k}\cdot\frac{l}{C^{2}}\leq\frac{l}{C^{2}}. If C is large enough we then obtain l^{\prime}\leq l . The maximality of the points \{x_{i}\}_{i=1}^{l^{\prime}} gives \Omega^{\prime}\subseteq\bigcup_{i=1}^{l^{\prime}}B(x_{i},2R) . Also observe that the \sqrt{2}R -neighbourhood of \Omega^{\prime} covers \Omega (note that the R -neighbourhood of \Omega^{\prime} does not cover \Omega ). We thereby have \Omega\subseteq\bigcup_{i=1}^{l^{\prime}}B(x_{i},(2+\sqrt{2})R) . Therefore if we introduce the Voronoi cells \displaystyle\Omega_{i}:=\{x\in\Omega\mid|x-x_{i}|\leq|x-x_{j}|\text{ for all }j\neq i\}, then \{\Omega_{i}\}_{i=1}^{l^{\prime}} is a convex partition of \Omega . Since \Omega_{i}\subseteq B(x_{i},(2+\sqrt{2})R) we obtain the desired convex partition \{\Omega_{i}\}_{i=1}^{l^{\prime}} of \Omega . This completes the proof of the theorem. \Box 4. An weaker inequality of ( 1.2 ) In this section we give a short proof of the following: Theorem 4.1 . Let \Omega be a bounded convex domain in \mathbb{R}^{n} . Then we have \displaystyle\mu_{k+1}(\Omega)\lesssim n^{7}\mu_{k}(\Omega) for any natural number k . Proof. Proposition 2.1 gives the existence of a rectangular domain R such that \displaystyle\frac{1}{\sqrt{n}}R\subseteq\Omega\subseteq nR. Let \widetilde{R} be the flat torus whose fundamental domain is R . Observe that the k -the eigenvalue \lambda_{k}(\widetilde{R}) of the Laplacian on \widetilde{R} is equal to \mu_{k}(R) ( [ 2 , Section II.2] ). Also since \widetilde{R} is a homogeneous manifold, we use P. Li’s theorem ( [ 9 , Theorem 11] ) to get \lambda_{k+1}(\widetilde{R})\lesssim\lambda_{k}(\widetilde{R}) , yielding that \displaystyle\mu_{k+1}(R)\lesssim\mu_{k}(R). Applying Theorem 2.2 twice we thus obtain

Read the complete text