This paper was converted on www.awesomepapers.org from LaTeX by an anonymous user.
Want to know more? Visit the Converter page.

Eigenvalues of elliptic operators with density

Bruno Colbois, Luigi Provenzano Bruno Colbois, Laboratoire de Mathématiques, Université de Neuchâtel, 13 Rue E. Argand, 2007 Neuchâtel, Switzerland. E-mail: bruno.colbois@unine.ch Luigi Provenzano, EPFL, SB Institute of Mathematics, Station 8, CH-1015 Lausanne, Switzerland. E-mail: luigi.provenzano@epfl.ch
Abstract.

We consider eigenvalue problems for elliptic operators of arbitrary order 2m2m subject to Neumann boundary conditions on bounded domains of the Euclidean NN-dimensional space. We study the dependence of the eigenvalues upon variations of mass density and in particular we discuss the existence and characterization of upper and lower bounds under both the condition that the total mass is fixed and the condition that the LN2mL^{\frac{N}{2m}}-norm of the density is fixed. We highlight that the interplay between the order of the operator and the space dimension plays a crucial role in the existence of eigenvalue bounds.

Key words and phrases:
High order elliptic operators, eigenvalues, mass densities, eigenvalue bounds, Weyl eigenvalue asymptotics
2010 Mathematics Subject Classification:
Primary: 35P15. Secondary: 35J40, 35P20.
The second author is member of the Gruppo Nazionale per l’Analisi Matematica, la Probabilità e le loro Applicazioni (GNAMPA) of the Istituto Nazionale di Alta Matematica (INdAM)

1. Introduction

We consider the eigenvalue problem

(1.1) (Δ)mu=μρu(-\Delta)^{m}u=\mu\rho u

on a connected bounded open subset Ω\Omega of N\mathbb{R}^{N}, where ρ\rho is a positive function bounded away from zero and infinity and where we impose Neumann boundary conditions on uu. Under suitable regularity assumptions on the boundary of Ω\Omega (e.g., if Ω\Omega has a Lipschitz boundary) it is standard to prove that problem (1.1) admits an increasing sequence of non-negative eigenvalues of finite multiplicity

0=μ1[ρ]==μdN,m[ρ]<μdN,m+1[ρ]μj[ρ]+,0=\mu_{1}[\rho]=\cdots=\mu_{d_{N,m}}[\rho]<\mu_{d_{N,m}+1}[\rho]\leq\cdots\leq\mu_{j}[\rho]\cdots\nearrow+\infty,

where dN,md_{N,m} denotes the dimension of the space of polynomials of degree at most m1m-1 in N\mathbb{R}^{N}.

In this paper we will prove a few results on the dependence of the eigenvalues μj[ρ]\mu_{j}[\rho] upon variation of ρ\rho. In particular we will consider the problem of finding upper bounds for μj[ρ]\mu_{j}[\rho] among all positive and bounded densities ρ\rho satisfying suitable constraints. We shall also consider the issue of lower bounds, which also presents some interesting features.

Keeping in mind important problems for the Laplace and the biharmonic operators in linear elasticity (see e.g., [16]) we shall think of the weight ρ\rho as a mass density of the body Ω\Omega and we shall refer to the quantity M=Ωρ𝑑xM=\int_{\Omega}\rho dx as to the total mass of Ω\Omega. In fact, when N=2N=2 the eigenvalues μj[ρ]\mu_{j}[\rho] describe the vibrations of a non-homogeneous membrane with free edge when m=1m=1 (see e.g., [28, § 9]) and of a non-homogeneous plate with free edge when m=2m=2 (see [9, 12]).

Relevant questions on the dependence of the eigenvalues μj[ρ]\mu_{j}[\rho] upon ρ\rho are whether it is possible to minimize or maximize the eigenvalues under the assumption that the total mass is fixed, or whether it is possible to have uniform upper or lower bounds for the eigenvalues (i.e., bounds which depend only on the total mass, the dimension and the eigenvalue index) under the same constraint, and which have the correct behavior in jj\in\mathbb{N} as described by the Weyl’s asymptotic law.

Most of the existing literature treats the case of the Laplace operator with Dirichlet boundary conditions. In particular, we recall the famous result of Krein [31] on the eigenvalues of the Dirichlet Laplacian in one dimension (fixed string) which completely answers the questions raised above. In fact he finds sharp upper and lower bounds which depend only on j,M,H,lj,M,H,l for all the eigenvalues of the Laplacian on the string ]0,l[]0,l[ upon densities 0ρH0\leq\rho\leq H for which M=0lρ𝑑xM=\int_{0}^{l}\rho dx is fixed (see Remark 3.24). We refer also to the extensive work of Banks and collaborators for generalizations and extensions of Krein’s results (see [2, 3, 5, 6, 26] and the references therein). We mention also [21, § 5] which contains a detailed analysis of the eigenvalues of Sturm-Liouville problems with Dirichlet conditions with density (and also other types of weight). In particular, in [21, § 5], the authors provide estimates (upper and lower bounds) under various type of linear and non-linear constraints on the weights. Existence of minimizers and maximizers under mass constraint in higher dimensions for the Dirichlet Laplacian has been investigated in [11, 17, 18, 19, 24], where the authors impose the additional constraint that admissible densities are uniformly bounded from below and above by some fixed constants. We refer to [28, § 9] and to the references therein for further discussions on eigenvalue problems for inhomogeneous strings and membranes with fixed edges.

As for Neumann boundary conditions, much less is known. Very recently the problem of finding uniform upper bounds for the Neumann eigenvalues of the Laplacian with weights has been solved (for N2N\geq 2) by Colbois and El Soufi [14] in the more general context of Riemannian manifolds, by exploiting a general result of decomposition of a metric measure space by annuli (see [27], see also [29]). The authors have not considered the case N=1N=1 and, in fact, as we shall see in the present paper, upper bounds with mass constraint do not exists in dimension one.

There are very few results for what concerns higher order operators. We recall [4, 35], where the authors consider the case of the biharmonic operator in one dimension with intermediate boundary conditions (hinged rod) and [7], where the author considers the case of the biharmonic operator with Dirichlet conditions in dimension one (clamped rod) and two (clamped plate). We also refer to [21, § 7.9] where it is possible to find some estimates for the eigenvalues of elliptic operators of order 2m2m with density subject to Dirichlet boundary conditions. We refer again to [28, § 11] for a more detailed discussion on eigenvalue problems for inhomogeneous rods and plates with hinged and clamped edges. Up to our knowledge, there are no results in the literature on the existence and characterization of upper and lower bounds with respect to mass densities for higher order operators subject to Neumann boundary conditions (already for the biharmonic operator or the Laplacian in dimension one).

Finally, we refer to [32] where the authors prove continuity and differentiability results for the dependence of the eigenvalues of a quite wide class of higher order elliptic operators and homogeneous boundary conditions upon variation of the mass density and in most of the cases (except, again, that of Neumann boundary conditions), they establish a maximum principle for extremum problems related to mass density perturbations which preserve the total mass. We remark that in [32] partial results are obtained in the case of Neumann boundary conditions only for the Laplace operator.

In this paper we shall primarily address the issue of finding upper bounds for the eigenvalues μj[ρ]\mu_{j}[\rho] of the polyharmonic operators with Neumann boundary conditions which are consistent with the power of jj in the Weyl’s asymptotic formula (see (2.9)), among all densities which satisfy a suitable constraint. In particular, we consider two very natural constraints: Ωρ𝑑x=const.\int_{\Omega}\rho dx={\rm const.} and ΩρN2m𝑑x=const.\int_{\Omega}\rho^{\frac{N}{2m}}dx={\rm const.}. This second constraint arises naturally since it is well-known (see e.g., [23]) that if we set N(μ):={#μj:μj<μ}N(\mu):=\left\{\#\mu_{j}:\mu_{j}<\mu\right\}, then N(μ)ωN(2π)1NμN2mΩρN2m𝑑xN(\mu)\sim\omega_{N}(2\pi)^{\frac{1}{N}}\mu^{\frac{N}{2m}}\int_{\Omega}\rho^{\frac{N}{2m}}dx. This means that ρLN2m(Ω)\|\rho\|_{L^{\frac{N}{2m}}(\Omega)} describes the asymptotic distribution of the eigenvalues of problem (1.1) (and in particular implies the Weyl’s law (2.9)). Most of the literature mentioned above considers only the fixed mass constraint.

In view of the physical interpretation of problem (1.1) when m=1m=1 and N=1N=1 or N=2N=2, it is very natural to ask whether it is possible to redistribute a fixed amount of mass on a string (of fixed length) or on a membrane (of fixed shape) such that all the eigenvalues become arbitrarily large when the body is left free to move, or, on the contrary, if there exists uniform upper bounds for all the eigenvalues. As highlithed in [14], uniform upper bounds with mass constraint exist if N2N\geq 2. In this paper, by using the techniques of [14] we prove that if N2mN\geq 2m, uniform upper bounds exist (see Theorem 3.4), namely we prove that if N2mN\geq 2m

(1.2) μj[ρ]CN,m|Ω|Ωρ𝑑x(j|Ω|)2mN,\mu_{j}[\rho]\leq C_{N,m}\frac{|\Omega|}{\int_{\Omega}\rho dx}\left(\frac{j}{|\Omega|}\right)^{\frac{2m}{N}},

where CN,mC_{N,m} depends only on mm and NN. Surprisingly, in lower dimensions, uniform upper bounds do not hold. In fact we find explicit examples of densities with fixed mass and arbitrarily large eigenvalues (see Theorem 3.29). In this case, however, we are able to find upper bounds which depend also on ρL(Ω)\|\rho\|_{L^{\infty}(\Omega)} (see Theorem 3.22), namely we prove that if N<2mN<2m

(1.3) μj[ρ]CN,mρL(Ω)2mN1(Ωρ𝑑x)2mNj2mN,\mu_{j}[\rho]\leq C_{N,m}\frac{\|\rho\|_{L^{\infty}(\Omega)}^{\frac{2m}{N}-1}}{\left(\int_{\Omega}\rho dx\right)^{\frac{2m}{N}}}j^{\frac{2m}{N}},

where again CN,mC_{N,m} depends only on mm and NN and the exponent of ρL(Ω)\|\rho\|_{L^{\infty}(\Omega)} is sharp. We remark that this inequality holds when m=1m=1 for N=1N=1, and it is the analogue of the upper bounds (3.25) proved by Krein [31] for the Dirichlet Laplacian on an interval (up to a universal constant). We note that in order to prove that certain eigenvalue bounds under some natural constraints do not hold, one has to provide counterexamples. It is then natural to ask whether it is possible to find ‘weaker’ bounds which include the correct quantities that explain the counterexamples. This is the case of the bounds (1.3).

We note that the interplay between the dimension of the space and the order of the operator plays a crucial role in the existence of uniform upper bounds for the eigenvalues of problem (1.1) under mass constraint. We can summarize our first result in this way:

“If N2mN\geq 2m there exist uniform upper bounds with mass constraint for all the eigenvalues of (1.1), while if N<2mN<2m we can always redistribute a fixed amount of mass such that all the eigenvalues of (1.1) become arbitrarily large”.

As for the the non-linear constraint ΩρN2m𝑑x=const.\int_{\Omega}\rho^{\frac{N}{2m}}dx={\rm const.}, in view of the fact that N(μ)ωN(2π)1NμN2mΩρN2m𝑑xN(\mu)\sim\omega_{N}(2\pi)^{\frac{1}{N}}\mu^{\frac{N}{2m}}\int_{\Omega}\rho^{\frac{N}{2m}}dx, it is natural to ask whether upper bounds of the form

(1.4) μj[ρ]CN,m(jΩρN2m𝑑x)2mN\mu_{j}[\rho]\leq C_{N,m}\left(\frac{j}{\int_{\Omega}\rho^{\frac{N}{2m}}dx}\right)^{\frac{2m}{N}}

hold. We will call bounds of the form (1.4) “Weyl-type bounds”. Clearly, for N=2mN=2m inequality (1.4) is equivalent to (1.2). For N>2mN>2m we are able to find densities with fixed LN2mL^{\frac{N}{2m}}-norm and which produce arbitrarily large eigenvalues (see Theorem 4.16). However, we are able to prove upper bounds for all the eigenvalues which involve both ρLN2m(Ω)\|\rho\|_{L^{\frac{N}{2m}}(\Omega)} and ρL(Ω)\|\rho\|_{L^{\infty}(\Omega)} (see Theorem 4.11), namely we prove that if N>2mN>2m then

(1.5) μj[ρ]CN,m(|Ω|ρL(Ω)N2mΩρN2m𝑑x)12mN(jΩρN2m𝑑x)2mN,\mu_{j}[\rho]\leq C_{N,m}\left(\frac{|\Omega|\|\rho\|_{L^{\infty}(\Omega)}^{\frac{N}{2m}}}{\int_{\Omega}\rho^{\frac{N}{2m}}dx}\right)^{1-\frac{2m}{N}}\left(\frac{j}{\int_{\Omega}\rho^{\frac{N}{2m}}dx}\right)^{\frac{2m}{N}},

where CN,mC_{N,m} depends only on mm and NN and the exponent of ρL(Ω)\|\rho\|_{L^{\infty}(\Omega)} is sharp. Since (1.4) holds for N=2mN=2m we are led to conjecture that it must hold for any N<2mN<2m. We are still not able to prove (1.4) for N<2mN<2m, and actually it seems to be a quite difficult issue. However we can prove the weaker inequality

(1.6) μj[ρ]CN,m(|Ω|ρL(Ω)N2mΩρN2m𝑑x)2mN1(jΩρN2m𝑑x)2mN.\mu_{j}[\rho]\leq C_{N,m}\left(\frac{|\Omega|\|\rho\|_{L^{\infty}(\Omega)}^{\frac{N}{2m}}}{\int_{\Omega}\rho^{\frac{N}{2m}}dx}\right)^{\frac{2m}{N}-1}\left(\frac{j}{\int_{\Omega}\rho^{\frac{N}{2m}}dx}\right)^{\frac{2m}{N}}.

We leave the validity of (1.4) for N<2mN<2m as an open question. We refer to Remark 4.10 where we discuss relevant examples in support of the validity of our conjecture. In particular, we note that if (1.4) holds true for N<2mN<2m, when m=1m=1 and N=1N=1 we would find uniform upper bounds for the eigenvalues of the Neumann Laplacian under the constraint that Ωρ𝑑x=const.\int_{\Omega}\sqrt{\rho}dx={\rm const.}. We can summarize our second result as follows:

“If N>2mN>2m, we can always find a density ρ\rho with fixed ρLN2m(Ω)\|\rho\|_{L^{\frac{N}{2m}}(\Omega)} such that all the eigenvalues of (1.1) are arbitrarily large, while we have uniform Weyl-type upper bounds when N=2mN=2m. We conjecture the existence of uniform Weyl-type upper bounds when N<2mN<2m.

We also mention [28, § 9.2.3] where it is considered a spectral optimization problem for the Dirichlet Laplacian with the non-linear constraint Ωρp𝑑x=const.\int_{\Omega}\rho^{p}dx={\rm const.}, where p>N/2p>N/2 and N2N\geq 2 (see also [21, § 5]).

We have also considered the issue of lower bounds and we have found that ‘surprisingly’ the interplay between the space dimension NN and the order mm of the operator plays a fundamental role also in the existence of lower bounds. In fact we are able to prove the following facts (see Theorems 5.1,5.4,5.13 and 5.15):

“If N<2mN<2m there exists a positive constant CC which depends only on m,Nm,N and Ω\Omega such that the first positive eigenvalue of problem (1.1) is bounded from below by C(Ωρ𝑑x)1C\left(\int_{\Omega}\rho dx\right)^{-1}, while if N2mN\geq 2m, for all jj\in\mathbb{N} we can always redistribute a fixed amount of mass such that the first jj eigenvalues of (1.1) are arbitrarily close to zero”

and

“If N>2mN>2m there exists a positive constant CC which depends only on m,Nm,N and Ω\Omega such that the first positive eigenvalue of problem (1.1) is bounded from below by CρLN2m(Ω)1C\|\rho\|_{L^{\frac{N}{2m}}(\Omega)}^{-1}, while if N2mN\leq 2m for all jj\in\mathbb{N} we can always find densities with fixed LN2mL^{\frac{N}{2m}}-norm such that the first jj eigenvalues of (1.1) are arbitrarily close to zero”.

We note that lower bounds for the first eigenvalue under one of the two constraints exist in the case that upper bounds with the same constraint do not exist. We remark that the situation is very different if we consider for example the issue of the minimization of the eigenvalues of (1.1) with ρ1\rho\equiv 1 among all bounded domains with fixed measure: it is standard to prove that there exist domains with fixed volume and such that the first jj eigenvalues can be made arbitrarily close to zero, in any dimension N2N\geq 2.

Finally we remark that all the results of this paper can be adapted to the more general eigenvalue problem

u=μρu,\mathcal{L}u=\mu\rho u,

with Neumann boundary conditions, where \mathcal{L} is defined by

u:=0|α|,|β|m(1)|α|α(Aαββu)\mathcal{L}u:=\sum_{0\leq|\alpha|,|\beta|\leq m}(-1)^{|\alpha|}\partial^{\alpha}(A_{\alpha\beta}\partial^{\beta}u)

and is an elliptic operator of order 2m2m, under suitable assumptions on the domain Ω\Omega and the coefficients of AαβA_{\alpha\beta}. We refer to [32] for a detailed description of eigenvalue problems for higher order elliptic operators with density (see also [21, § 7]).

The present paper is organized as follows: Section 2 is dedicated to some preliminaries. In Section 3 we consider the problem of finding uniform upper bounds with mass constraint. In particular in Subsection 3.1 we prove uniform upper bounds (1.2) for N2mN\geq 2m, in Subsection 3.2 we prove upper bounds (1.3) for N<2mN<2m and in Subsection 3.3 we provide counterexamples to uniform upper bounds in dimension N<2mN<2m. In Section 4 we investigate the existence of upper bounds with the non-linear constraint ΩρN2m𝑑x=const.\int_{\Omega}\rho^{\frac{N}{2m}}dx={\rm const.}. In particular in Subsections 4.1 and 4.2 we prove upper bounds (1.6) and (1.5), respectively, while in Subsection 4.3 we provide counterexamples to uniform upper bounds (1.4) for N>2mN>2m. In Subsection 4.1 we state the open question whether bounds of the form (1.4) hold if N<2mN<2m. In Section 5 we consider lower bounds and in particular we discuss how the constraint, the space dimension and the order of the operator influence their existence. At the end of the paper we have two appendices, Appendix A and Appendix B. In Appendix A we discuss Neumann boundary conditions for higher order operators and develop some basic spectral theory for such operators. In Appendix B we prove some useful functional inequalities which are crucial in the proof of Theorem 3.29 in Subsection 3.3.

2. Preliminaries and notation

Let Ω\Omega be a bounded domain (i.e., an open connected bounded set) of N\mathbb{R}^{N}. By Hm(Ω)H^{m}(\Omega) we shall denote the standard Sobolev space of functions in L2(Ω)L^{2}(\Omega) with weak derivatives up to order mm in L2(Ω)L^{2}(\Omega), endowed with its standard norm defined by

uHm(Ω):=(Ω|Dmu|2+u2dx)12\|u\|_{H^{m}(\Omega)}:=\left(\int_{\Omega}|D^{m}u|^{2}+u^{2}dx\right)^{\frac{1}{2}}

for all uHm(Ω)u\in H^{m}(\Omega), where

|Dmu|2:=αN,|α|=m|αu|2.|D^{m}u|^{2}:=\sum_{\begin{subarray}{c}\alpha\in\mathbb{N}^{N},\\ |\alpha|=m\end{subarray}}|\partial^{\alpha}u|^{2}.

In what follow we will use the standard multi-index notation. Hence, for αN\alpha\in\mathbb{N}^{N}, α=(α1,,αN)\alpha=(\alpha_{1},...,\alpha_{N}), we shall denote by |α||\alpha| the quantity |α|=α1++αN|\alpha|=\alpha_{1}+\cdots+\alpha_{N}. Moreover, for α,βN\alpha,\beta\in\mathbb{N}^{N}, α+β=(α1+β1,,αN+βN)\alpha+\beta=(\alpha_{1}+\beta_{1},...,\alpha_{N}+\beta_{N}) and α!=α1!αN!\alpha!=\alpha_{1}!\cdots\alpha_{N}!. For xNx\in\mathbb{R}^{N}, x=(x1,,xN)x=(x_{1},...,x_{N}), we will write xα=x1α1xNαNx^{\alpha}=x_{1}^{\alpha_{1}}\cdots x_{N}^{\alpha_{N}}. For a function uu of class C|α|C^{|\alpha|}, we write αu=|α|ux1α1xNαN\partial^{\alpha}u=\frac{\partial^{|\alpha|}u}{\partial_{x_{1}}^{\alpha_{1}}\cdots\partial_{x_{N}}^{\alpha_{N}}}. Finally, for a function U:U:\mathbb{R}\rightarrow\mathbb{R} and ll\in\mathbb{N}, we shall write U(l)(x)U^{(l)}(x) to denote the ll-th derivative of UU with respect to xx.

In the sequel we shall assume that the domain Ω\Omega is such that the embedding of Hm(Ω)H^{m}(\Omega) into L2(Ω)L^{2}(\Omega) is compact (which is ensured, for example, if Ω\Omega is a bounded domain with Lipschitz boundary). By \mathcal{R} we shall denote the subset of L(Ω)L^{\infty}(\Omega) of those functions ρL(Ω)\rho\in L^{\infty}(\Omega) such that essinfΩρ>0{\rm ess}\inf_{\Omega}\rho>0.

We shall consider the following eigenvalue problem:

(2.1) ΩDmu:Dmφdx=μΩρuφ𝑑x,φHm(Ω)\int_{\Omega}D^{m}u:D^{m}\varphi dx=\mu\int_{\Omega}\rho u\varphi dx\,,\ \ \ \forall\varphi\in H^{m}(\Omega)

in the unknowns uHm(Ω)u\in H^{m}(\Omega) (the eigenfunction), μ\mu\in\mathbb{R} (the eigenvalue), where

Dmu:Dmφ:=|α|=mαuαφ.D^{m}u:D^{m}\varphi:=\sum_{|\alpha|=m}{\partial^{\alpha}u}{\partial^{\alpha}\varphi}.

We note that problem (2.1) is the weak formulation of the following eigenvalue problem:

(2.2) {(Δ)mu=μρu,inΩ,𝒩0u==𝒩m1u=0,onΩ,\begin{cases}(-\Delta)^{m}u=\mu\rho u,&{\rm in\ }\Omega,\\ \mathcal{N}_{0}u=\cdots=\mathcal{N}_{m-1}u=0,&{\rm on\ }\partial\Omega,\end{cases}

in the unknowns uC2m(Ω)C2m1(Ω¯)u\in C^{2m}(\Omega)\cap C^{2m-1}(\overline{\Omega}) and μ\mu\in\mathbb{R}. Here 𝒩j\mathcal{N}_{j} are uniquely defined ‘complementing’ boundary operators of degree at most 2m12m-1 (see [25] for details), which we will call Neumann boundary conditions (see Appendix A.1).

Example 2.3.

If m=1m=1, 𝒩0u=uν\mathcal{N}_{0}u=\frac{\partial u}{\partial\nu} and (2.2) is the classical formulation of the Neumann eigenvalue problem for the Laplace operator, namely

(2.4) {Δu=μρu,inΩ,uν=0,onΩ,\begin{cases}-\Delta u=\mu\rho u,&{\rm in\ }\Omega,\\ \frac{\partial u}{\partial\nu}=0,&{\rm on\ }\partial\Omega,\end{cases}

in the unknowns uC2(Ω)C1(Ω¯)u\in C^{2}(\Omega)\cap C^{1}(\overline{\Omega}) and μ\mu\in\mathbb{R}, while if m=2m=2 we have the Neumann eigenvalue problem for the biharmonic operator, namely

(2.5) {Δ2u=μρu,inΩ,2uν2=0,onΩ,divΩ(D2uν)+Δuν=0,onΩ,\begin{cases}\Delta^{2}u=\mu\rho u,&{\rm in\ }\Omega,\\ \frac{\partial^{2}u}{\partial\nu^{2}}=0,&{\rm on\ }\partial\Omega,\\ {\rm div}_{\partial\Omega}(D^{2}u\cdot\nu)+\frac{\partial\Delta u}{\partial\nu}=0,&{\rm on\ }\partial\Omega,\end{cases}

in the unknowns uC4(Ω)C3(Ω¯)u\in C^{4}(\Omega)\cap C^{3}(\overline{\Omega}) and μ\mu\in\mathbb{R}. Here divΩ{\rm div}_{\partial\Omega} denotes the tangential divergence operator on Ω\partial\Omega (we refer to [20, § 7] for more details on tangential operators).

In Appendix A.1 we discuss in more detail boundary conditions for problems (2.2) and (2.5) and, more in general, Neumann boundary conditions for the polyharmonic operators.

It is standard to prove (see Theorem A.5) that the eigenvalues of (2.1) are non-negative, have finite multiplicity and consist of a sequence diverging to ++\infty of the form

0=μ1[ρ]==μdN,m[ρ]<μdN,m+1[ρ]μj[ρ]+,0=\mu_{1}[\rho]=\cdots=\mu_{d_{N,m}}[\rho]<\mu_{d_{N,m}+1}[\rho]\leq\cdots\leq\mu_{j}[\rho]\leq\cdots\nearrow+\infty,

where

dN,m:=(N+m1N).d_{N,m}:=\binom{N+m-1}{N}.

The eigenfunctions associated with the eigenvalue μ=0\mu=0 are the polynomials of degree at most m1m-1 in N\mathbb{R}^{N} (the dimension of the space spanned by the polynomials of degree at most m1m-1 in N\mathbb{R}^{N} is exactly dN,md_{N,m}). We note that we have highlithed the dependence of the eigenvalues upon the density ρ\rho, which is the main object of study of the present paper.

By standard spectral theory, we deduce the validity of the following variational representation of the eigenvalues (see [16, § IV] for more details):

Theorem 2.6.

Let Ω\Omega be a bounded domain in N\mathbb{R}^{N} such that the embedding Hm(Ω)L2(Ω)H^{m}(\Omega)\subset L^{2}(\Omega) is compact. Let ρ\rho\in\mathcal{R}. Then for all jj\in\mathbb{N} we have

(2.7) μj[ρ]=infVHm(Ω)dimV=jsup0uVΩ|Dmu|2𝑑xΩρu2𝑑x.\mu_{j}[\rho]=\inf_{\begin{subarray}{c}V\leq H^{m}(\Omega)\\ {\rm dim}V=j\end{subarray}}\sup_{\begin{subarray}{c}0\neq u\in V\end{subarray}}\frac{\int_{\Omega}|D^{m}u|^{2}dx}{\int_{\Omega}\rho u^{2}dx}.

We conclude this section by recalling the asymptotic behavior of the eigenvalues as j+j\rightarrow+\infty, which is described by the Weyl’s law.

Theorem 2.8.

Let Ω\Omega be a bounded domain in N\mathbb{R}^{N} with Lipschitz boundary. Let ρ\rho\in\mathcal{R}. Then

(2.9) μj[ρ](2π)2mωN2mN(jΩρN2m)2mN\mu_{j}[\rho]\sim\frac{(2\pi)^{2m}}{\omega_{N}^{\frac{2m}{N}}}\left(\frac{j}{\int_{\Omega}\rho^{\frac{N}{2m}}}\right)^{\frac{2m}{N}}

as j+j\rightarrow+\infty.

We refer to [23] for a proof of Theorem 2.8.

3. Upper bounds with mass constraint

In this section we consider the problem of finding uniform upper bounds for the jj-th eigenvalue μj[ρ]\mu_{j}[\rho] among all mass densities ρ\rho\in\mathcal{R} which preserve the mass (that is, among all ρ\rho\in\mathcal{R} such that Ωρ𝑑x=const.\int_{\Omega}\rho dx={\rm const.}), and which show the correct growth in the power of jj with respect to the Weyl’s law (2.9). In particular, in Subsection 3.1 we prove that such bounds exist if N2mN\geq 2m (see Theorem 3.4), while in Subsection 3.3 we will give counter-examples in dimension N<2mN<2m (see Theorem 3.29). Moreover, in Subsection 3.2 we establish upper bounds in the case N<2mN<2m which involve also a suitable power of ρL(Ω)\|\rho\|_{L^{\infty}(\Omega)} (see Theorem 3.22) which turns out to be sharp.

3.1. Uniform upper bounds with mass constraint for N2mN\geq 2m

In this subsection we will prove the existence of uniform upper bounds for N2mN\geq 2m with respect to mass preserving densities.

The main tool which we will use is a result of decomposition of a metric measure space by annuli (see [27, Theorem 1.1]). We recall it here for the reader’s convenience.

Let (X,d)(X,d) be a metric space. By an annulus in XX we mean any set AXA\subset X of the form

A=A(a,r,R)={xX:r<d(x,a)<R},A=A(a,r,R)=\left\{x\in X:r<d(x,a)<R\right\},

where aXa\in X and 0r<R<+0\leq r<R<+\infty. By 2A2A we denote

2A=2A(a,r,R)={xX:r2<d(x,a)<2R}.2A=2A(a,r,R)=\left\{x\in X:\frac{r}{2}<d(x,a)<2R\right\}.

We are ready to state the following theorem (see [27, Theorem 11]):

Theorem 3.1.

Let (X,d)(X,d) be a metric space and ν\nu be a Radon measure on it. Assume that the following properties are satisfied:

  1. i)

    there exists a constant Γ\Gamma such that any metric ball of radius rr can be covered by at most Γ\Gamma balls of radius r/2r/2;

  2. ii)

    all metric balls in XX are precompact sets;

  3. iii)

    the measure ν\nu is non-atomic.

Then for any integer jj there exist a sequence {Ai}i=1j\left\{A_{i}\right\}_{i=1}^{j} of jj annuli in XX such that, for any i=1,,ji=1,...,j

ν(Ai)cν(X)j,\nu(A_{i})\geq c\frac{\nu(X)}{j},

and the annuli 2Ai2A_{i} are pairwise disjoint. The constant cc depends only on Γ\Gamma.

In the sequel we will need also the following corollary of Theorem 3.1 (see [27, Remark 3.13]):

Corollary 3.2.

Let the assumptions of Theorem 3.1 hold. If in addition 0<ν(X)<0<\nu(X)<\infty, each annulus AiA_{i} has either internal radius rir_{i} such that

(3.3) ri12inf{r:V(r)vj},r_{i}\geq\frac{1}{2}\inf\left\{r\in\mathbb{R}:V(r)\geq v_{j}\right\},

where V(r):=supxXν(B(x,r))V(r):=\sup_{x\in X}\nu(B(x,r)) and vj=cν(X)jv_{j}=c\frac{\nu(X)}{j} , or is a ball of radius rir_{i} satisfying (3.3).

We are now ready to state the main result of this section.

Theorem 3.4.

Let Ω\Omega be a bounded domain in N\mathbb{R}^{N}, N2mN\geq 2m, such that the embedding Hm(Ω)L2(Ω)H^{m}(\Omega)\subset L^{2}(\Omega) is compact. Let ρ\rho\in\mathcal{R}. Then for every jj\in\mathbb{N} we have

(3.5) μj[ρ]CN,m1Ωρ𝑑x|Ω|1(j|Ω|)2mN,\mu_{j}[\rho]\leq C_{N,m}\frac{1}{\int_{\Omega}\rho dx|\Omega|^{-1}}\left(\frac{j}{|\Omega|}\right)^{\frac{2m}{N}},

where CN,mC_{N,m} is a constant which depends only on NN and mm.

Remark 3.6.

Inequality (3.5) says that there exists a uniform upper bound for all the eigenvalues μj[ρ]\mu_{j}[\rho] with respect to those densities ρ\rho\in\mathcal{R} which give the same mass M=Ωρ𝑑xM=\int_{\Omega}\rho dx. We note that the quantity Ωρ𝑑x|Ω|1=M/|Ω|\int_{\Omega}\rho dx|\Omega|^{-1}=M/|\Omega| is an average density, i.e., the total mass over the total volume of Ω\Omega. Moreover, from (3.5) it follows that

μj[ρ]CN,m(j|Ω|)2mN,\mu_{j}[\rho]\leq C_{N,m}\left(\frac{j}{|\Omega|}\right)^{\frac{2m}{N}},

for all densities ρ\rho\in\mathcal{R} and bounded domains Ω\Omega (with Hm(Ω)L2(Ω)H^{m}(\Omega)\subset L^{2}(\Omega) compact) such that Ωρ𝑑x=|Ω|\int_{\Omega}\rho dx=|\Omega|.

Proof of Theorem 3.4.

The proof is based on the general method described by Grigor’yan, Netrusov and Yau in [27] (see Theorem 3.1; see also [14, 15] for the case of the Laplace operator). In particular, we will build a suitable family of disjointly supported test functions with controlled Rayleigh quotient.

Let 0r<R<+0\leq r<R<+\infty. Let Ur,R:[0,+[[0,1]U_{r,R}:[0,+\infty[\rightarrow[0,1] be defined by

(3.7) Ur,R(t):={0ift[0,r/2[[2R,+[,i=02m1airitiift[r/2,r[,1ift[r,R[,i=02m1biRitiift[R,2R[.U_{r,R}(t):=\begin{cases}0&{\rm if\ }t\in[0,r/2[\cup[2R,+\infty[,\\ \sum_{i=0}^{2m-1}\frac{a_{i}}{r^{i}}t^{i}&{\rm if\ }t\in[r/2,r[,\\ 1&{\rm if\ }t\in[r,R[,\\ \sum_{i=0}^{2m-1}\frac{b_{i}}{R^{i}}t^{i}&{\rm if\ }t\in[R,2R[.\end{cases}

The coefficients {ai}i=02m1\left\{a_{i}\right\}_{i=0}^{2m-1}, {bi}i=02m1\left\{b_{i}\right\}_{i=0}^{2m-1} are uniquely determined by the equations

(3.8) {Ur,R(r/2)=0,Ur,R(r)=1,Ur,R(R)=1,Ur,R(2R)=0,Ur,R(l)(r)=Ur,R(l)(r/2)=Ur,R(l)(R)=Ur,R(l)(2R)=0,l=1,,m1.\begin{cases}U_{r,R}(r/2)=0,&\\ U_{r,R}(r)=1,&\\ U_{r,R}(R)=1,&\\ U_{r,R}(2R)=0,&\\ U_{r,R}^{(l)}(r)=U_{r,R}^{(l)}(r/2)=U_{r,R}^{(l)}(R)=U_{r,R}^{(l)}(2R)=0,&\forall l=1,...,m-1.\end{cases}

We note that (3.8) can be written as

(3.9) {i=02m1ai2i=0,i=02m1ai=1,i=02m1bi=1,i=02m12ibi=0,i=l2m1i(i1)(il+1)2liai=0,l=1,,m1,i=l2m1i(i1)(il+1)ai=0,l=1,,m1,i=l2m1i(i1)(il+1)2ilbi=0,l=1,,m1,i=l2m1i(i1)(il+1)bi=0,l=1,,m1,\begin{cases}\sum_{i=0}^{2m-1}\frac{a_{i}}{2^{i}}=0,&\\ \sum_{i=0}^{2m-1}{a_{i}}=1,&\\ \sum_{i=0}^{2m-1}{b_{i}}=1,&\\ \sum_{i=0}^{2m-1}{2^{i}b_{i}}=0,&\\ \sum_{i=l}^{2m-1}i(i-1)\cdots(i-l+1)2^{l-i}a_{i}=0,&\forall l=1,...,m-1,\\ \sum_{i=l}^{2m-1}i(i-1)\cdots(i-l+1)a_{i}=0,&\forall l=1,...,m-1,\\ \sum_{i=l}^{2m-1}i(i-1)\cdots(i-l+1)2^{i-l}b_{i}=0,&\forall l=1,...,m-1,\\ \sum_{i=l}^{2m-1}i(i-1)\cdots(i-l+1)b_{i}=0,&\forall l=1,...,m-1,\end{cases}

which is a system of 4m4m equations in 4m4m unknowns a0,,a2m1a_{0},...,a_{2m-1}, b0,,b2m1b_{0},...,b_{2m-1}. It is standard to see that (3.9) admits an unique non-zero solution. Moreover, the coefficients a0,,a2m1a_{0},...,a_{2m-1}, b0,,b2m1b_{0},...,b_{2m-1} depend only on mm.

We note that by construction Ur,RCm1([0,+[)Cm1,1([0,2R])U_{r,R}\in C^{m-1}([0,+\infty[)\cap C^{m-1,1}([0,2R]). Let now A=A(a,r,R)A=A(a,r,R) be an annulus in N\mathbb{R}^{N}. We define a function ua,r,Ru_{a,r,R} supported on 2A2A and such that ua,r,R1u_{a,r,R}\leq 1 on 2A2A and ua,r,R1u_{a,r,R}\equiv 1 on AA by setting

(3.10) ua,r,R(x):=Ur,R(|xa|).u_{a,r,R}(x):=U_{r,R}(|x-a|).

By construction, the restriction of this function to Ω\Omega belongs to the Sobolev space Hm(Ω)H^{m}(\Omega). Now we exploit Theorem 3.1 with X=ΩX=\Omega endowed with the Euclidean distance, and the measure ν\nu given by ν(E):=Eρ𝑑x\nu(E):=\int_{E}\rho dx for all measurable EΩE\subset\Omega. The hypothesis of Theorem 3.1 are clearly satisfied. Hence, for each index jj\in\mathbb{N} we find 2j2j annuli {Ai}i=12j\left\{A_{i}\right\}_{i=1}^{2j} such that 2Ai2A_{i} are disjoint and

AiΩρ𝑑xcNΩρ𝑑x2j,\int_{A_{i}\cap\Omega}\rho dx\geq c_{N}\frac{\int_{\Omega}\rho dx}{2j},

where cN>0c_{N}>0 depends only on NN. Since we have 2j2j annuli 2Ai=2Ai(ai,ri,Ri)2A_{i}=2A_{i}(a_{i},r_{i},R_{i}), i=1,,2ji=1,...,2j, we can choose jj of such annuli, say {2Ai1,,2Aij}\left\{2A_{i_{1}},...,2A_{i_{j}}\right\} such that

(3.11) |2Aik||Ω|j|2A_{i_{k}}|\leq\frac{|\Omega|}{j}

for all k=1,,jk=1,...,j. To each of such annuli, we associate a function uiku_{i_{k}} defined by

(3.12) uik(x):=uaik,rik,Rik(x).u_{i_{k}}(x):=u_{a_{i_{k}},r_{i_{k}},R_{i_{k}}}(x).

We have then built a family of jj disjointly supported functions, which we relabel as u1,,uju_{1},...,u_{j}, and whose restriction to Ω\Omega belong to the space Hm(Ω)H^{m}(\Omega).

From the min-max principle (2.7) it follows that

(3.13) μj[ρ]max0uVjΩ|Dmu|2𝑑xΩρu2𝑑x,\mu_{j}[\rho]\leq\max_{\begin{subarray}{c}0\neq u\in V_{j}\end{subarray}}\frac{\int_{\Omega}|D^{m}u|^{2}dx}{\int_{\Omega}\rho u^{2}dx},

where VjV_{j} is the subspace of Hm(Ω)H^{m}(\Omega) generated by u1,,uju_{1},...,u_{j} (and which has dimension jj). Since the space is generated by jj disjointly supported functions, it is standard to prove that (3.13) is equivalent to the following:

(3.14) μj[ρ]maxi=1,,jΩ|Dmui|2𝑑xΩρui2𝑑x.\mu_{j}[\rho]\leq\max_{i=1,...,j}\frac{\int_{\Omega}|D^{m}u_{i}|^{2}dx}{\int_{\Omega}\rho u_{i}^{2}dx}.

This means that it is sufficient to have a control on the Rayleigh quotient of each of the generating functions uiu_{i} in order to bound μj[ρ]\mu_{j}[\rho]. It is standard to see that, if uCm(Ω)u\in C^{m}(\Omega) is given by u(x)=U(|xa|)u(x)=U(|x-a|), for some function UU of one real variable, then for all αN\alpha\in\mathbb{N}^{N} with |α|=m|\alpha|=m

(3.15) αu(x)=k=1mcN,k,α|xa|mkU(k)(|xa|),{\partial^{\alpha}u(x)}=\sum_{k=1}^{m}\frac{c_{N,k,\alpha}}{|x-a|^{m-k}}U^{(k)}(|x-a|),

where cN,k,αc_{N,k,\alpha} depends only on NN, kk and α\alpha. From (3.15) it follows then that there exists a constant CN,m>0C_{N,m}>0 which depends only on mm and NN such that

(3.16) |Dmu(x)|2CN,mk=1m(U(k)(xa))2|xa|2(mk).|D^{m}u(x)|^{2}\leq C_{N,m}\sum_{k=1}^{m}\frac{(U^{(k)}(x-a))^{2}}{|x-a|^{2(m-k)}}.

Through the rest of the proof we will denote by CN,mC_{N,m} a positive constant which depends only on mm and NN and which can be eventually re-defined line by line.

By standard approximation of functions in Hm(Ω)H^{m}(\Omega) by smooth functions (see [22, § 5.3]), from (3.16) we deduce that if uHm(Ω)u\in H^{m}(\Omega) is given by u(x)=U(|xa|)u(x)=U(|x-a|), for some function UU of one real variable, then

(3.17) Ω|Dmu|2p𝑑xCN,mΩk=1m(U(k)(|xa|))2p|xa|2p(mk)dx,\int_{\Omega}|D^{m}u|^{2p}dx\leq C_{N,m}\int_{\Omega}\sum_{k=1}^{m}\frac{(U^{(k)}(|x-a|))^{2p}}{|x-a|^{2p(m-k)}}dx,

for all p>0p>0. We are now ready to estimate the right-hand side of (3.14). For the denominator we have

(3.18) Ωρui2𝑑x=2AiΩρui2𝑑xAiΩρui2𝑑x=AiΩρ𝑑xcNΩρ𝑑x2j.\int_{\Omega}\rho u_{i}^{2}dx=\int_{2A_{i}\cap\Omega}\rho u_{i}^{2}dx\geq\int_{A_{i}\cap\Omega}\rho u_{i}^{2}dx=\int_{A_{i}\cap\Omega}\rho dx\geq c_{N}\frac{\int_{\Omega}\rho dx}{2j}.

This follows from the fact that ui1u_{i}\leq 1 and ui1u_{i}\equiv 1 on AiA_{i} and from Theorem 3.1. For the numerator, since N2mN\geq 2m, we have

(3.19) Ω|Dmui|2𝑑x=Ω2Ai|Dmui|2𝑑x2Ai|Dmui|2𝑑x(2Ai|Dmui|N/m𝑑x)2m/N|2Ai|12m/N.\int_{\Omega}|D^{m}u_{i}|^{2}dx=\int_{\Omega\cap 2A_{i}}|D^{m}u_{i}|^{2}dx\leq\int_{2A_{i}}|D^{m}u_{i}|^{2}dx\\ \leq\left(\int_{2A_{i}}|D^{m}u_{i}|^{N/m}dx\right)^{2m/N}|2A_{i}|^{1-2m/N}.

From (3.11), we have that |2Ai|12m/N(|Ω|/j)12m/N|2A_{i}|^{1-2m/N}\leq\left({|\Omega|}/{j}\right)^{1-2m/N}. Moreover, from (3.7), (3.17) and standard calculus we have that

(3.20) (2Ai|Dmui|N/2𝑑x)2m/NCN,m.\left(\int_{2A_{i}}|D^{m}u_{i}|^{N/2}dx\right)^{2m/N}\leq C_{N,m}.

From (3.18), (3.19) and (3.20) we have that

(3.21) Ω|Dmui|2𝑑xΩρui2𝑑xCN,m1Ωρ𝑑x|Ω|1(j|Ω|)2mN\frac{\int_{\Omega}|D^{m}u_{i}|^{2}dx}{\int_{\Omega}\rho u_{i}^{2}dx}\leq C_{N,m}\frac{1}{\int_{\Omega}\rho dx|\Omega|^{-1}}\left(\frac{j}{|\Omega|}\right)^{\frac{2m}{N}}

for all i=1,,ji=1,...,j. From (3.14) and (3.21) we deduce the validity of (3.5). This concludes the proof. ∎

3.2. Upper bounds with mass constraint for N<2mN<2m

We note that the proof of Theorem 3.4 does not work in the case N<2mN<2m. Indeed, as we will see in Subsection 3.3 (see Theorem 3.29), bounds of the form (3.5) do not hold if N<2mN<2m. In this subsection we prove upper bounds for the eigenvalues μj[ρ]\mu_{j}[\rho] which involve also a suitable power of ρL(Ω)\|\rho\|_{L^{\infty}(\Omega)}, namely, we prove the following theorem:

Theorem 3.22.

Let Ω\Omega be a bounded domain in N\mathbb{R}^{N}, N<2mN<2m, such that the embedding Hm(Ω)L2(Ω)H^{m}(\Omega)\subset L^{2}(\Omega) is compact. Let ρ\rho\in\mathcal{R}. Then for every jj\in\mathbb{N} we have

(3.23) μj[ρ]CN,mρL2mN1(Ωρ𝑑x)2mNj2mN,\mu_{j}[\rho]\leq C_{N,m}\frac{\|\rho\|_{L^{\infty}}^{\frac{2m}{N}-1}}{\left(\int_{\Omega}\rho dx\right)^{\frac{2m}{N}}}j^{\frac{2m}{N}},

where CN,m>0C_{N,m}>0 depends only on mm and NN.

Remark 3.24.

We recall the following well-known result by Krein [31] which states that in the case of the equation u′′(x)=μρ(x)u(x)-u^{\prime\prime}(x)=\mu\rho(x)u(x) on ]0,l[]0,l[ with Dirichlet boundary conditions, we have

(3.25) μj[ρ]π2ρL(]0,l[)(0lρ𝑑x)2j2,\mu_{j}[\rho]\leq\frac{\pi^{2}\|\rho\|_{L^{\infty}(]0,l[)}}{\left(\int_{0}^{l}\rho dx\right)^{2}}j^{2},

which is the analogous of (3.23) for the Laplace operator (m=1m=1) in dimension N=1N=1. Actually, inequality (3.25) is sharp, i.e., for all jj\in\mathbb{N} there exists ρj\rho_{j}\in\mathcal{R} such that the equality holds in (3.25) when ρ=ρj\rho=\rho_{j}.

Proof of Theorem 3.22.

In order to prove (3.23) we exploit in more detail Theorem 3.1 and Corollary 3.2. As in the proof of Theorem 3.4, for any measurable EΩE\subset\Omega, let ν(E):=Eρ𝑑x\nu(E):=\int_{E}\rho dx. By following the same lines of the proof of Theorem 3.4, we find for each jj\in\mathbb{N}, jj annuli A1,,AjA_{1},...,A_{j} such that the annuli 2Ai2A_{i} are disjoint, AiΩρ𝑑xcNΩρ𝑑x2j\int_{A_{i}\cap\Omega}\rho dx\geq c_{N}\frac{\int_{\Omega}\rho dx}{2j} for all i=1,,ji=1,...,j, where cN>0c_{N}>0 depends only on NN, and moreover |2Ai||Ω|/j|2A_{i}|\leq{|\Omega|}/{j} for all i=1,,ji=1,...,j.

Let rir_{i} and RiR_{i} denote the inner and outer radius of AiA_{i}, respectively (rir_{i} denotes the radius of AiA_{i} if AiA_{i} is a ball). Associated to each annulus AiA_{i} we construct a test function uiu_{i} supported on 2Ai2A_{i} and such that ui1u_{i}\equiv 1 on AiA_{i}, and which satisfies

2AiΩρui2𝑑xAiΩρ𝑑xcNΩρ𝑑x2j.\int_{2A_{i}\cap\Omega}\rho u_{i}^{2}dx\geq\int_{A_{i}\cap\Omega}\rho dx\geq c_{N}\frac{\int_{\Omega}\rho dx}{2j}.

and

2Ai|Dmui|2𝑑xCN,m(RiN2m+riN2m)2CN,mriN2m,\int_{2A_{i}}|D^{m}u_{i}|^{2}dx\leq C_{N,m}(R_{i}^{N-2m}+r_{i}^{N-2m})\leq 2C_{N,m}r_{i}^{N-2m},

where CN,m>0C_{N,m}>0 depends only on mm and NN (see (3.7), (3.10), (3.12) and (3.17)). In what follows we shall denote by CN,mC_{N,m} a positive constant which depends only on mm and NN and which can be eventually re-defined line by line. Then, if AiA_{i} is an annulus of inner radius rir_{i} or a ball of radius rir_{i}, we have

(3.26) Ω|Dmui|2𝑑xΩρui2𝑑xCN,mriN2mjΩρ𝑑x,\frac{\int_{\Omega}|D^{m}u_{i}|^{2}dx}{\int_{\Omega}\rho u_{i}^{2}dx}\leq C_{N,m}r_{i}^{N-2m}\frac{j}{\int_{\Omega}\rho dx},

for all i=1,,ji=1,...,j.

We note that Corollary 3.2 provides an estimate for the inner radius of the annuli given by the decomposition of Theorem 3.1 (and, respectively, an estimate of the radius of the ball in the case that the decomposition of the space produces a ball). In particular, if rir_{i} is the inner radius of AiA_{i}, we have for all i=1,,ji=1,...,j

ri12inf{r:V(r)vj},r_{i}\geq\frac{1}{2}\inf\left\{r\in\mathbb{R}:V(r)\geq v_{j}\right\},

where V(r):=supxΩB(x,r)ρ𝑑xV(r):=\sup_{x\in\Omega}\int_{B(x,r)}\rho dx and vj=cNΩρ𝑑x2jv_{j}=c_{N}\frac{\int_{\Omega}\rho dx}{2j}. Let Bj:={r:V(r)>vj}B_{j}:=\left\{r\in\mathbb{R}:V(r)>v_{j}\right\}. If rBjr\in B_{j}, then

cNΩρ𝑑x2j=vj<supxΩB(x,r)ρ𝑑xωNrNρL(Ω),c_{N}\frac{\int_{\Omega}\rho dx}{2j}=v_{j}<\sup_{x\in\Omega}\int_{B(x,r)}\rho dx\leq\omega_{N}r^{N}\|\rho\|_{L^{\infty}(\Omega)},

where ωN\omega_{N} denotes the volume of the unit ball in N\mathbb{R}^{N}. This means that

rN>cNΩρ𝑑x2jωNρL(Ω)r^{N}>c_{N}\frac{\int_{\Omega}\rho dx}{2j\omega_{N}\|\rho\|_{L^{\infty}(\Omega)}}

for all rBjr\in B_{j}. Hence, if r0:=infBjr_{0}:=\inf B_{j}, then r0NcNΩρ𝑑x2jωNρL(Ω)r_{0}^{N}\geq c_{N}\frac{\int_{\Omega}\rho dx}{2j\omega_{N}\|\rho\|_{L^{\infty}(\Omega)}} and therefore

(3.27) riNcNΩρ𝑑x2N+1jωNρL(Ω)r_{i}^{N}\geq c_{N}\frac{\int_{\Omega}\rho dx}{2^{N+1}j\omega_{N}\|\rho\|_{L^{\infty}(\Omega)}}

for all i=1,,ji=1,...,j. From (3.26) and (3.27) it follows that

Ω|Dmui|2𝑑xΩρui2𝑑xCN,mρL(Ω)2mN1(Ωρ𝑑x)2mNj2mN,\frac{\int_{\Omega}|D^{m}u_{i}|^{2}dx}{\int_{\Omega}\rho u_{i}^{2}dx}\leq C_{N,m}\frac{\|\rho\|_{L^{\infty}(\Omega)}^{\frac{2m}{N}-1}}{\left(\int_{\Omega}\rho dx\right)^{\frac{2m}{N}}}j^{\frac{2m}{N}},

for all i=1,,ji=1,...,j, which implies (3.23) by (2.7) and by the fact that uiu_{i} are disjointly supported (see also (3.13) and (3.14)). This concludes the proof.

3.3. Non-existence of uniform upper bounds for N<2mN<2m and sharpness of the exponent of ρL(Ω)\|\rho\|_{L^{\infty}(\Omega)} in (3.23)

In this subsection we will prove that if N<2mN<2m, there exist families {ρε}ε]0,ε0[\left\{\rho_{\varepsilon}\right\}_{\varepsilon\in]0,\varepsilon_{0}[}\subset\mathcal{R} such that Ωρε𝑑xωN\int_{\Omega}\rho_{\varepsilon}dx\rightarrow\omega_{N} as ε0+\varepsilon\rightarrow 0^{+} and μj[ρε]+\mu_{j}[\rho_{\varepsilon}]\rightarrow+\infty for all jdN,m+1j\geq d_{N,m}+1 as ε0+\varepsilon\rightarrow 0^{+}, and moreover we will provide the rate of divergence to ++\infty of the eigenvalues with respect to the parameter ε\varepsilon. This means that in dimension N<2mN<2m we can redistribute a bounded amount of mass in Ω\Omega in such a way that all the positive eigenvalues become arbitrarily large. This is achieved, for example, by concentrating all the mass at one point of Ω\Omega. Moreover, the families {ρε}ε]0,ε0[\left\{\rho_{\varepsilon}\right\}_{\varepsilon\in]0,\varepsilon_{0}[} considered in this subsection will provide the sharpness of the power of ρL(Ω)\|\rho\|_{L^{\infty}(\Omega)} in (3.23).

Through all this subsection, Ω\Omega will be a bounded domain in N\mathbb{R}^{N} with Lipschitz boundary. Assume without loss of generality that 0Ω0\in\Omega and let ε0]0,1[\varepsilon_{0}\in]0,1[ be such that B(0,ε)ΩB(0,\varepsilon)\subset\subset\Omega for all ε]0,ε0[\varepsilon\in]0,\varepsilon_{0}[ (all the result of this section hold true if we substitute 0 with any other x0Ωx_{0}\in\Omega). For all ε]0,ε0[\varepsilon\in]0,\varepsilon_{0}[ let ρε\rho_{\varepsilon}\in\mathcal{R} be defined by

(3.28) ρε:=ε2mNδ+εNχB(0,ε),\rho_{\varepsilon}:=\varepsilon^{2m-N-\delta}+\varepsilon^{-N}\chi_{B(0,\varepsilon)},

for some δ]0,1/2[\delta\in]0,1/2[, which we fix once for all and which can be chosen arbitrarily close to zero. We have the following theorem:

Theorem 3.29.

Let Ω\Omega be a bounded domain in N\mathbb{R}^{N}, N<2mN<2m, with Lipschitz boundary. Let ρε\rho_{\varepsilon}\in\mathcal{R} be defined by (3.28) for all ε]0,ε0[\varepsilon\in]0,\varepsilon_{0}[. Then

  1. i)

    limε0+Ωρε𝑑x=ωN\lim_{\varepsilon\rightarrow 0^{+}}\int_{\Omega}\rho_{\varepsilon}dx=\omega_{N};

  2. ii)

    for all jj\in\mathbb{N}, jdN,m+1j\geq d_{N,m}+1, there exists cj>0c_{j}>0 which depends only on mm, NN, Ω\Omega and jj such that (up to subsequences) limε0+μj[ρε]ε2mNδ=cj\lim_{\varepsilon\rightarrow 0^{+}}\mu_{j}[\rho_{\varepsilon}]\varepsilon^{2m-N-\delta}=c_{j} (for all δ]0,1/2[\delta\in]0,1/2[).

From Theorem 3.29 it follows that for N<2mN<2m and jdN,m+1j\geq d_{N,m}+1

limε0+μj[ρε]=+.\lim_{\varepsilon\rightarrow 0^{+}}\mu_{j}[\rho_{\varepsilon}]=+\infty.

Moreover, since ρεL(Ω)=εN\|\rho_{\varepsilon}\|_{L^{\infty}(\Omega)}=\varepsilon^{-N}, it follows that the power of ρL(Ω)\|\rho\|_{L^{\infty}(\Omega)} in (3.23) is sharp.

We remark that the proof of Theorem 3.29 requires some precise inequalities for function in Hm(Ω)H^{m}(\Omega) and N<2mN<2m which we prove in Lemmas B.5 and B.13 of the Appendix B.

Proof of Theorem 3.29.

The proof of point i)i) is a standard computation. In fact

limε0+Ωρε𝑑x=limε0+ε2mNδ|Ω|+limε0+εN|B(0,ε)|=ωN.\lim_{\varepsilon\rightarrow 0^{+}}\int_{\Omega}\rho_{\varepsilon}dx=\lim_{\varepsilon\rightarrow 0^{+}}\varepsilon^{2m-N-\delta}|\Omega|+\lim_{\varepsilon\rightarrow 0^{+}}\varepsilon^{-N}|B(0,\varepsilon)|=\omega_{N}.

We prove now ii)ii). In order to simplify the notation, from now on we will denote an eigenvalue μj[ρε]\mu_{j}[\rho_{\varepsilon}] simply as μj[ε]\mu_{j}[\varepsilon]. The proof of ii)ii) is divided into two steps. In the first step we will prove that there exists a positive constant cj(1)>0c_{j}^{(1)}>0 which depends only on mm, NN, jj and Ω\Omega such that μj[ε]cj(1)εN2m+δ\mu_{j}[\varepsilon]\leq c_{j}^{(1)}\varepsilon^{N-2m+\delta}. In the second step we will prove that there exists a positive constant cj(2)>0c_{j}^{(2)}>0 which depends only on mm, NN, jj and Ω\Omega such that μj[ε]cj(2)εN2m+δ\mu_{j}[\varepsilon]\geq c_{j}^{(2)}\varepsilon^{N-2m+\delta}. This yields the result (up to choosing a suitable subsequence of {ρε}ε]0,ε0[\left\{\rho_{\varepsilon}\right\}_{\varepsilon\in]0,\varepsilon_{0}[}).

Step 1. We note that ρεε2mNδ\rho_{\varepsilon}\geq\varepsilon^{2m-N-\delta}, hence for all uHm(Ω)u\in H^{m}(\Omega),

(3.30) Ω|Dmu|2𝑑xΩρεu2𝑑xεN2m+δΩ|Dmu|2𝑑xΩu2𝑑x.\frac{\int_{\Omega}|D^{m}u|^{2}dx}{\int_{\Omega}\rho_{\varepsilon}u^{2}dx}\leq\varepsilon^{N-2m+\delta}\frac{\int_{\Omega}|D^{m}u|^{2}dx}{\int_{\Omega}u^{2}dx}.

By taking the minimum and the maximum into (3.30), by (2.7) we have that μj[ε]εN2m+δμj[1]\mu_{j}[\varepsilon]\leq\varepsilon^{N-2m+\delta}\mu_{j}[1] for all jj\in\mathbb{N}. Hence cj(1)=μj[1]c_{j}^{(1)}=\mu_{j}[1].

Step 2. We introduce the function ρ~ε\tilde{\rho}_{\varepsilon} defined by

(3.31) ρ~ε:=1+ε2m+δχB(0,ε).\tilde{\rho}_{\varepsilon}:=1+\varepsilon^{-2m+\delta}\chi_{B(0,\varepsilon)}.

We note that μj[ε]\mu_{j}[\varepsilon] is an eigenvalue of (2.1) with ρ=ρε\rho=\rho_{\varepsilon} if and only if μ~j[ε]:=ε2mNδμj[ε]\tilde{\mu}_{j}[\varepsilon]:=\varepsilon^{2m-N-\delta}\mu_{j}[\varepsilon] is an eigenvalue of problem (2.1) with ρ=ρ~ε\rho=\tilde{\rho}_{\varepsilon}, where ρ~ε\tilde{\rho}_{\varepsilon} is defined by (3.31).

We prove now that for all jdN,m+1j\geq d_{N,m}+1 there exist c>0c>0 which depends only on m,nm,n and Ω\Omega such that μ~j[ε]c\tilde{\mu}_{j}[\varepsilon]\geq c for all ε]0,ε0[\varepsilon\in]0,\varepsilon_{0}[. This implies the existence of constants cj(2)>0c_{j}^{(2)}>0 such that μj[ε]cj(2)εN2m+δ\mu_{j}[\varepsilon]\geq c_{j}^{(2)}\varepsilon^{N-2m+\delta}.

We recall that the first positive eigenvalue is μ~dN,m+1[ε]\tilde{\mu}_{d_{N,m}+1}[\varepsilon]. From the min-max principle (2.7) we have

μ~dN,m+1[ε]=infuVεΩ|Dmu|2𝑑xΩu2𝑑x+ε2m+δB(0,ε)u2𝑑x,\tilde{\mu}_{d_{N,m}+1}[\varepsilon]=\inf_{u\in V_{\varepsilon}}\frac{\int_{\Omega}|D^{m}u|^{2}dx}{\int_{\Omega}u^{2}dx+\varepsilon^{-2m+\delta}\int_{B(0,\varepsilon)}u^{2}dx},

where

Vε:={uHm(Ω):Ωρ~εu(x)xα𝑑x=0, 0|α|m1}.V_{\varepsilon}:=\left\{u\in H^{m}(\Omega):\int_{\Omega}\tilde{\rho}_{\varepsilon}u(x)x^{\alpha}dx=0,\ \forall\,0\leq|\alpha|\leq m-1\right\}.

We argue by contradiction. Assume that μ~dN,m+1[ε]0\tilde{\mu}_{d_{N,m}+1}[\varepsilon]\rightarrow 0 as ε0+\varepsilon\rightarrow 0^{+}. Let uεHm(Ω)u_{\varepsilon}\in H^{m}(\Omega) be an eigenfunction associated with μ~dN,m+1[ε]\tilde{\mu}_{d_{N,m}+1}[\varepsilon] normalized by Ωρ~εuε2𝑑x=1\int_{\Omega}\tilde{\rho}_{\varepsilon}u_{\varepsilon}^{2}dx=1. Clearly uεL2(Ω)2Ωρ~εuε2𝑑x=1\|u_{\varepsilon}\|_{L^{2}(\Omega)}^{2}\leq\int_{\Omega}\tilde{\rho}_{\varepsilon}u_{\varepsilon}^{2}dx=1 (see formula (3.31)) and since Ω|Dmuε|2𝑑x0\int_{\Omega}|D^{m}u_{\varepsilon}|^{2}dx\rightarrow 0 as ε0+\varepsilon\rightarrow 0^{+}, we have that the sequence {uε}ε]0,ε0[\left\{u_{\varepsilon}\right\}_{\varepsilon\in]0,\varepsilon_{0}[} is bounded in Hm(Ω)H^{m}(\Omega). Then there exists u0Hm(Ω)u_{0}\in H^{m}(\Omega) such that, up to subsequences, uεu0u_{\varepsilon}\rightharpoonup u_{0} in Hm(Ω)H^{m}(\Omega) and uεu0u_{\varepsilon}\rightarrow u_{0} in Hm1(Ω)H^{m-1}(\Omega) as ε0+\varepsilon\rightarrow 0^{+} by the compactness of the embedding Hm(Ω)Hm1(Ω)H^{m}(\Omega)\subset H^{m-1}(\Omega).

Since limε0+Ω|Dmuε|2𝑑x=0\lim_{\varepsilon\rightarrow 0^{+}}\int_{\Omega}|D^{m}u_{\varepsilon}|^{2}dx=0, it is standard (see e.g., [22, § 5.8]) to prove that αu0=0{\partial^{\alpha}u_{0}}=0 in L2(Ω)L^{2}(\Omega) for all |α|=m|\alpha|=m, and hence u0=|α|m1aαxαu_{0}=\sum_{|\alpha|\leq m-1}a_{\alpha}x^{\alpha} for some constants {aα}|α|m1\left\{a_{\alpha}\right\}_{|\alpha|\leq m-1}\subset\mathbb{R}. This means that u0u_{0} is a polymonial of degree at most m1m-1. Moreover, from Lemma B.13, point i)i) and ii)ii), it follows that αu0(0)=0\partial^{\alpha}u_{0}(0)=0 for all |α|k|\alpha|\leq k, where k=mN212k=m-\frac{N}{2}-\frac{1}{2} if NN is odd, and k=mN21k=m-\frac{N}{2}-1 if NN is even. Hence aα=0a_{\alpha}=0 for all |α|k|\alpha|\leq k. Then

(3.32) u0=k+1|α|m1aαxα.u_{0}=\sum_{k+1\leq|\alpha|\leq m-1}a_{\alpha}x^{\alpha}.

Moreover, from Lemma B.13, points v)v) and vi)vi) it follows that Ωu0xα𝑑x=0\int_{\Omega}u_{0}x^{\alpha}dx=0 for all k+1|α|m1k+1\leq|\alpha|\leq m-1 which implies along with (3.32) that u00u_{0}\equiv 0.

Again, from Lemma B.13, iv)iv) and v)v), it follows that 1=Ωρ~εuε2𝑑xΩu02𝑑x1=\int_{\Omega}\tilde{\rho}_{\varepsilon}u_{\varepsilon}^{2}dx\rightarrow\int_{\Omega}u_{0}^{2}dx as ε0+\varepsilon\rightarrow 0^{+}, which yields the contradiction. This concludes the proof.

4. Weyl-type upper bounds

In this section we investigate the existence of uniform upper bounds for μj[ρ]\mu_{j}[\rho] which are compatible with the Weyl’s law (2.9), namely we look for uniform upper bounds of the form

(4.1) μj[ρ]CN,m(jΩρN2m𝑑x)2mN,\mu_{j}[\rho]\leq C_{N,m}\left(\frac{j}{\int_{\Omega}\rho^{\frac{N}{2m}}dx}\right)^{\frac{2m}{N}},

where the constant CN,mC_{N,m} depends only on mm and NN. Actually we will not prove (4.1), but a weaker form involving also ρL(Ω)\|\rho\|_{L^{\infty}(\Omega)} in the case N<2mN<2m (see Theorem 4.2). We remark that in the case N=2mN=2m, the bounds (4.1) hold, in fact this is already contained in Theorem 3.4. Moreover, we shall prove upper bounds involving ρL(Ω)\|\rho\|_{L^{\infty}(\Omega)} in the case N>2mN>2m (see Theorem 4.11) in which the power of ρL(Ω)\|\rho\|_{L^{\infty}(\Omega)} turns out to be sharp (see Theorem 4.16). In particular this implies that bounds of the form (4.1) do not hold if N>2mN>2m. We will be left with the open question (see Remark 4.10) of the existence of bounds of the form (4.1) in the case N<2mN<2m.

4.1. Upper bounds for N2mN\leq 2m

In this subsection we prove upper bounds of the form (4.1) in the case N2mN\leq 2m involving a certain power of ρL(Ω)\|\rho\|_{L^{\infty}(\Omega)}, namely, we prove the following theorem:

Theorem 4.2.

Let Ω\Omega be a bounded domain in N\mathbb{R}^{N}, N2mN\leq 2m, such that the embedding Hm(Ω)L2(Ω)H^{m}(\Omega)\subset L^{2}(\Omega) is compact. Let ρ\rho\in\mathcal{R}. Then there exists a constant CN,m>0C_{N,m}>0 which depends only on mm and NN such that for all jj\in\mathbb{N} it holds

(4.3) μj[ρ]CN,m(|Ω|ρL(Ω)N2mΩρN2m𝑑x)2mN1(jΩρN2m𝑑x)2mN.\mu_{j}[\rho]\leq C_{N,m}\left(\frac{|\Omega|\|\rho\|_{L^{\infty}(\Omega)}^{\frac{N}{2m}}}{\int_{\Omega}\rho^{\frac{N}{2m}}dx}\right)^{\frac{2m}{N}-1}\left(\frac{j}{\int_{\Omega}\rho^{\frac{N}{2m}}dx}\right)^{\frac{2m}{N}}.
Proof.

First, we remark that (4.3) with N=2mN=2m has already been proved in Theorem 3.4. Hence, from now on we let N<2mN<2m.

The proof is very similar to that of Theorem 3.4. It differs from the choice of the measure ν\nu in Theorem 3.1. In fact, we exploit Theorem 3.1 with X=ΩX=\Omega endowed with the Euclidean distance and ν\nu defined by ν(E):=EΩρN2m𝑑x\nu(E):=\int_{E\cap\Omega}\rho^{\frac{N}{2m}}dx for all measurable EΩE\subset\Omega.

The hypothesis of Theorem 3.1 are clearly satisfied. Then for each index jj\in\mathbb{N} we find jj metric annuli {Ai}i=1j\left\{A_{i}\right\}_{i=1}^{j} such that 2Ai2A_{i} are disjoint,

AiΩρN2m𝑑xcNΩρN2m𝑑x2j,\int_{A_{i}\cap\Omega}\rho^{\frac{N}{2m}}dx\geq c_{N}\frac{\int_{\Omega}\rho^{\frac{N}{2m}}dx}{2j},

where cN>0c_{N}>0 depends only on NN, and such that

(4.4) |2Ai||Ω|j|2A_{i}|\leq\frac{|\Omega|}{j}

for all i=1,,ji=1,...,j.

As in the proof of Theorem 3.4, for all i=1,,ji=1,...,j, we define a function uai,ri,Riu_{a_{i},r_{i},R_{i}} supported on 2Ai2A_{i} and such that uai,ri,Ri1u_{a_{i},r_{i},R_{i}}\leq 1 on 2Ai2A_{i} and uai,ri,Ri1u_{a_{i},r_{i},R_{i}}\equiv 1 on AiA_{i} by setting

uai,ri,Ri(x):=Uri,Ri(|xai|),u_{a_{i},r_{i},R_{i}}(x):=U_{r_{i},R_{i}}(|x-a_{i}|),

where Ur,RU_{r,R} is given by (3.7). By construction, the restriction of this function to Ω\Omega belongs to the Sobolev space Hm(Ω)H^{m}(\Omega). In order to simplify the notation, we will set ui(x):=uai,ri,Ri(x)u_{i}(x):=u_{a_{i},r_{i},R_{i}}(x).

We have then built a family of jj disjointly supported functions belonging to the space Hm(Ω)H^{m}(\Omega). We estimate now the Rayleigh quotient Ω|Dmui|2𝑑xΩρui2𝑑x\frac{\int_{\Omega}|D^{m}u_{i}|^{2}dx}{\int_{\Omega}\rho u_{i}^{2}dx} of uiu_{i} for all i=1,,ji=1,...,j. For the denominator we have

(4.5) Ωρui2𝑑x=2AiΩρui2𝑑xAiΩρui2𝑑x=AiΩρ𝑑x(AiΩρN2m𝑑x)2mN|Ai|12mNcN(ΩρN2m𝑑x2j)2mN|Ai|12mN.\int_{\Omega}\rho u_{i}^{2}dx=\int_{2A_{i}\cap\Omega}\rho u_{i}^{2}dx\geq\int_{A_{i}\cap\Omega}\rho u_{i}^{2}dx=\int_{A_{i}\cap\Omega}\rho dx\\ \geq\left(\int_{A_{i}\cap\Omega}\rho^{\frac{N}{2m}}dx\right)^{\frac{2m}{N}}|A_{i}|^{1-\frac{2m}{N}}\geq c_{N}\left(\frac{\int_{\Omega}\rho^{\frac{N}{2m}}dx}{2j}\right)^{\frac{2m}{N}}|A_{i}|^{1-\frac{2m}{N}}.

This follows from the fact that ui1u_{i}\leq 1 and ui1u_{i}\equiv 1 on AiA_{i}, from Hölder’s inequality and from Theorem 3.1. For the numerator we have

(4.6) Ω|Dmui|2𝑑x=Ω2Ai|Dmui|2𝑑x2Ai|Dmui|2𝑑xCN,m(riN2m+RiN2m)2CN,mriN2m,\int_{\Omega}|D^{m}u_{i}|^{2}dx=\int_{\Omega\cap 2A_{i}}|D^{m}u_{i}|^{2}dx\leq\int_{2A_{i}}|D^{m}u_{i}|^{2}dx\\ \leq C_{N,m}(r_{i}^{N-2m}+R_{i}^{N-2m})\leq 2C_{N,m}r_{i}^{N-2m},

where CN,mC_{N,m} is a constant which depends only on mm and NN. From now on we shall denote by CN,mC_{N,m} a positive constant which depends only on mm and NN and which can be eventually re-defined line by line. Assume now that Ai=B(ai,ri)A_{i}=B(a_{i},r_{i}) is a ball of center aia_{i} and radius rir_{i}. From (4.5) and (4.6) we have that

(4.7) Ω|Dmui|2𝑑xΩρui2𝑑xCN,m(jΩρN2m𝑑x)2mN,\frac{\int_{\Omega}|D^{m}u_{i}|^{2}dx}{\int_{\Omega}\rho u_{i}^{2}dx}\leq C_{N,m}\left(\frac{j}{\int_{\Omega}\rho^{\frac{N}{2m}}dx}\right)^{\frac{2m}{N}},

Assume now that AiA_{i} is a proper annulus (i.e., 0<ri<Ri0<r_{i}<R_{i}). From Corollary 3.2 it follows that for all i=1,,ji=1,...,j

ri12inf{r:V(r)vj},r_{i}\geq\frac{1}{2}\inf\left\{r\in\mathbb{R}:V(r)\geq v_{j}\right\},

where V(r):=supxΩB(x,r)ρN2m𝑑xV(r):=\sup_{x\in\Omega}\int_{B(x,r)}\rho^{\frac{N}{2m}}dx and vj=cNΩρN2m2jv_{j}=c_{N}\frac{\int_{\Omega}\rho^{\frac{N}{2m}}}{2j}. As in the proof of Theorem 3.22, we see that

(4.8) riNcNΩρN2m𝑑x2N+1jωNρL(Ω)N2mcNΩρN2m𝑑x|Ai|2N+1ωNρL(Ω)N2m|Ω|,r_{i}^{N}\geq\frac{c_{N}\int_{\Omega}\rho^{\frac{N}{2m}}dx}{2^{N+1}j\omega_{N}\|\rho\|_{L^{\infty}(\Omega)}^{\frac{N}{2m}}}\geq\frac{c_{N}\int_{\Omega}\rho^{\frac{N}{2m}}dx|A_{i}|}{2^{N+1}\omega_{N}\|\rho\|_{L^{\infty}(\Omega)}^{\frac{N}{2m}}|\Omega|},

where the second inequality follows from (4.4). By combining (4.5), (4.6) and (4.8) we obtain

(4.9) Ω|Dmui|2𝑑xΩρui2𝑑xCN,m(|Ω|ρL(Ω)N2mΩρN2m𝑑x)2mN1(jΩρN2m𝑑x)2mN,\frac{\int_{\Omega}|D^{m}u_{i}|^{2}dx}{\int_{\Omega}\rho u_{i}^{2}dx}\leq C_{N,m}\left(\frac{|\Omega|\|\rho\|_{L^{\infty}(\Omega)}^{\frac{N}{2m}}}{\int_{\Omega}\rho^{\frac{N}{2m}}dx}\right)^{\frac{2m}{N}-1}\left(\frac{j}{\int_{\Omega}\rho^{\frac{N}{2m}}dx}\right)^{\frac{2m}{N}},

By combining (4.7) and (4.9) and by the fact that |Ω|ρL(Ω)N2mΩρN2m𝑑x1\frac{|\Omega|\|\rho\|_{L^{\infty}(\Omega)}^{\frac{N}{2m}}}{\int_{\Omega}\rho^{\frac{N}{2m}}dx}\geq 1 for all ρ\rho\in\mathcal{R}, we obtain (4.3) thanks to (2.7) (see also (3.13) and (3.14)). This concludes the proof.

Remark 4.10.

From Theorem 4.2 it naturally arises the question whether bounds of the form (4.1) hold in the case N<2mN<2m. We conjecture an affirmative answer. In fact, in order to produce a family of densities {ρε}ε]0,ε0[\left\{\rho_{\varepsilon}\right\}_{\varepsilon\in]0,\varepsilon_{0}[} such that μj[ρε]+\mu_{j}[\rho_{\varepsilon}]\rightarrow+\infty as ε0+\varepsilon\rightarrow 0^{+}, a necessary condition is that ρε(x)0\rho_{\varepsilon}(x)\rightarrow 0 for almost every xΩx\in\Omega (otherwise we will find a subset EΩE\subset\Omega of positive measure where ρεc>0\rho_{\varepsilon}\geq c>0 for all ε]0,ε0[\varepsilon\in]0,\varepsilon_{0}[ and construct suitable test functions supported in EE which can be used to prove upper bounds for all the eigenvalues independent of ε\varepsilon as is done in Theorem 3.4). Hence, concentration phenomena are the right candidates in order to produce the blow-up of the eigenvalues. We may think to very simple toy models, like concentration around a point or in a neighborhood of the boundary (or in general, in a neighborhood of submanifolds contained in Ω\Omega). It is possible, for example, to show that if we concentrate all the mass in a single point, then the eigenvalues remain bounded (one can adapt the same arguments used in the proof of Theorem 3.29 or explicitly construct test functions for the Rayleigh quotient). If we concentrate all the mass in a neighborhood of the boundary, it is possible to prove that μj[ρε]0\mu_{j}[\rho_{\varepsilon}]\rightarrow 0 as ε0+\varepsilon\rightarrow 0^{+} (see Theorem 5.15 here below). These two types of concentration are somehow the extremal cases of mass concentration around submanifolds contained in Ω\Omega.

Moreover, we note that if for a fixed ρ\rho\in\mathcal{R} and a fixed jj\in\mathbb{N} all the 2j2j annuli given by the decomposition of Theorem 3.1 are actually balls, then from (4.7) we immediately deduce the validity of (4.1).

4.2. Upper bounds for N>2mN>2m

In this subsection we prove upper bounds for the eigenvalues μj[ρε]\mu_{j}[\rho_{\varepsilon}] which involve a suitable power of ρL(Ω)\|\rho\|_{L^{\infty}(\Omega)}. We have the following theorem:

Theorem 4.11.

Let Ω\Omega be a bounded domain in N\mathbb{R}^{N}, N>2mN>2m, such that the embedding Hm(Ω)L2(Ω)H^{m}(\Omega)\subset L^{2}(\Omega) is compact. Let ρ\rho\in\mathcal{R}. Then there exists a constant CN,m>0C_{N,m}>0 such that for all jj\in\mathbb{N}

(4.12) μj[ρ]CN,m(|Ω|ρL(Ω)N2mΩρN2m𝑑x)12mN(jΩρN2m𝑑x)2mN.\mu_{j}[\rho]\leq C_{N,m}\left(\frac{|\Omega|\|\rho\|_{L^{\infty}(\Omega)}^{\frac{N}{2m}}}{\int_{\Omega}\rho^{\frac{N}{2m}}dx}\right)^{1-\frac{2m}{N}}\left(\frac{j}{\int_{\Omega}\rho^{\frac{N}{2m}}dx}\right)^{\frac{2m}{N}}.
Proof.

Formula (4.12) follows directly from formula (3.5) by observing that for N>2mN>2m

ΩρN2m𝑑x=ΩρρN2m1𝑑xρL(Ω)N2m1Ωρ𝑑x,\int_{\Omega}\rho^{\frac{N}{2m}}dx=\int_{\Omega}\rho\rho^{\frac{N}{2m}-1}dx\leq\|\rho\|_{L^{\infty}(\Omega)}^{\frac{N}{2m}-1}\int_{\Omega}\rho dx,

and hence

(4.13) (Ωρ𝑑x)1ρL(Ω)N2m1ΩρN2m𝑑x.\left(\int_{\Omega}\rho dx\right)^{-1}\leq\frac{\|\rho\|_{L^{\infty}(\Omega)}^{\frac{N}{2m}-1}}{\int_{\Omega}\rho^{\frac{N}{2m}}dx}.

By plugging (4.13) into (3.5) and by standard calculus, (4.12) immediately follows. ∎

4.3. Non-existence of Weyl-type upper bounds for N>2mN>2m and sharpness of the exponent of ρL(Ω)\|\rho\|_{L^{\infty}(\Omega)} in (4.12)

In this subsection we prove that for N>2mN>2m there exist sequences {ρε}ε]0,ε0[\left\{\rho_{\varepsilon}\right\}_{\varepsilon\in]0,\varepsilon_{0}[}\subset\mathcal{R} such that ΩρεN2m𝑑x|Ω|\int_{\Omega}\rho_{\varepsilon}^{\frac{N}{2m}}dx\rightarrow|\partial\Omega| as ε0+\varepsilon\rightarrow 0^{+}, and μj[ρε]+\mu_{j}[\rho_{\varepsilon}]\rightarrow+\infty for all jdN,m+1j\geq d_{N,m}+1 as ε0+\varepsilon\rightarrow 0^{+}, and we also provide the rate of divergence to ++\infty of the eigenvalues with respect to ε\varepsilon. This means that if N>2mN>2m bounds of the form (4.1) do not hold. This result can be achieved, for example, by concentrating all the mass in a neighborhood of the boundary. Thus, mass densities with fixed LN2mL^{\frac{N}{2m}}-norm and which concentrate on particular submanifolds may produce blow-up of the eigenvalues if N>2mN>2m. Moreover the families of densities considered in this subsection will provide the sharpness of the power of ρL(Ω)\|\rho\|_{L^{\infty}(\Omega)} in (4.12).

Through all this subsection Ω\Omega will be a bounded domain in N\mathbb{R}^{N} of class C2C^{2}. Let

(4.14) ωε:={xΩ:dist(x,Ω)<ε}\omega_{\varepsilon}:=\left\{x\in\Omega:{\rm dist}(x,\partial\Omega)<\varepsilon\right\}

be the ε\varepsilon-tubular neighborhood of Ω\partial\Omega. Since Ω\Omega is of class C2C^{2} it follows that there exist ε0]0,1[\varepsilon_{0}\in]0,1[ such that for all ε]0,ε0[\varepsilon\in]0,\varepsilon_{0}[, each point in ωε\omega_{\varepsilon} has a unique nearest point on Ω\partial\Omega (see e.g., [30]). For all ε]0,ε0[\varepsilon\in]0,\varepsilon_{0}[ let ρε\rho_{\varepsilon}\in\mathcal{R} be defined by

(4.15) ρε:={ε2mN,inωε,ε22mN,inΩω¯ε.\rho_{\varepsilon}:=\begin{cases}\varepsilon^{-\frac{2m}{N}},&{\rm in\ }\omega_{\varepsilon},\\ \varepsilon^{2-\frac{2m}{N}},&{\rm in\ }\Omega\setminus\overline{\omega}_{\varepsilon}.\end{cases}

We have the following theorem:

Theorem 4.16.

Let Ω\Omega be a bounded domain in N\mathbb{R}^{N}, with N>2mN>2m, of class C2C^{2}. Let ρε\rho_{\varepsilon}\in\mathcal{R} be defined by (4.15) for all ε]0,ε0[\varepsilon\in]0,\varepsilon_{0}[. Then

  1. i)

    limε0+ΩρεN2m𝑑x=|Ω|\lim_{\varepsilon\rightarrow 0^{+}}\int_{\Omega}\rho_{\varepsilon}^{\frac{N}{2m}}dx=|\partial\Omega|;

  2. ii)

    for all jj\in\mathbb{N}, jdN,m+1j\geq d_{N,m}+1, there exists cj>0c_{j}>0 which depends only on mm, NN, Ω\Omega and jj such that limε0+μj[ρε]ε12mN=cj\lim_{\varepsilon\rightarrow 0^{+}}\mu_{j}[\rho_{\varepsilon}]\varepsilon^{1-\frac{2m}{N}}=c_{j}.

From Theorem 4.16 it follows that limε0+μj[ρε]=+\lim_{\varepsilon\rightarrow 0^{+}}\mu_{j}[\rho_{\varepsilon}]=+\infty. Moreover, since ρεL(Ω)=ε2mN\|\rho_{\varepsilon}\|_{L^{\infty}(\Omega)}=\varepsilon^{-\frac{2m}{N}}, the power of ρL(Ω)\|\rho\|_{L^{\infty}(\Omega)} in (4.12) is sharp.

In order to prove Theorem 4.16 we will exploit a result on the convergence of the Neumann eigenvalues of the polyharmonic operator to the corresponding Steklov eigenvalues.

The weak formulation of the polyharmonic Steklov eigenvalue problem reads:

(4.17) ΩDmu:Dmφdx=σΩuφ𝑑x,φHm(Ω)\int_{\Omega}D^{m}u:D^{m}\varphi dx=\sigma\int_{\partial\Omega}u\varphi dx\,,\ \ \ \forall\varphi\in H^{m}(\Omega)

in the unknowns uHm(Ω)u\in H^{m}(\Omega) (the eigenfunction), σ\sigma\in\mathbb{R} (the eigenvalue). It is standard to prove that the eigenvalues of (4.17) are non-negative and of finite multiplicity and are given by

0=σ1==σdN,m<σdN,m+1σj+.0=\sigma_{1}=\cdots=\sigma_{d_{N,m}}<\sigma_{d_{N,m}+1}\leq\cdots\leq\sigma_{j}\leq\cdots\nearrow+\infty.

We refer e.g., to [9] for a more detailed discussion on the Steklov eigenvalue problem for the biharmonic operator. We have the following theorem:

Theorem 4.18.

Let Ω\Omega be a bounded domain in N\mathbb{R}^{N} of class C2C^{2}. Let ξε:=ε+ε1χωε\xi_{\varepsilon}:=\varepsilon+\varepsilon^{-1}\chi_{\omega_{\varepsilon}}, where ωε\omega_{\varepsilon} is defined by (4.14). Let μj[ξε]\mu_{j}[\xi_{\varepsilon}] denote the eigenvalues of problem (2.1) with ρ=ξε\rho=\xi_{\varepsilon}. Then for all jj\in\mathbb{N}

limε0+μj[ξε]=|Ω|σj,\lim_{\varepsilon\rightarrow 0^{+}}\mu_{j}[\xi_{\varepsilon}]=|\partial\Omega|\sigma_{j},

where {σj}j\left\{\sigma_{j}\right\}_{j\in\mathbb{N}} are the eigenvalues of problem (4.17).

We refer to [33, 34] for the proof of Theorem 4.18 in the case of the Laplace operator and to [9] for the proof of Theorem 4.18 in the case of the biharmonic operator, and for more information on the convergence of Neumann eigenvalues to Steklov eigenvalues via mass concentration to the boundary. We remark that the proof of Theorem 4.18 for all values of mm\in\mathbb{N} follows exactly the same lines as the proof of the case m=1m=1 and m=2m=2.

Proof of Theorem 4.16.

We start from point i)i). It is standard to see that ΩρεN2m𝑑x=ε2N2m1|Ωω¯ε|+ε1|ωε|\int_{\Omega}\rho_{\varepsilon}^{\frac{N}{2m}}dx=\varepsilon^{\frac{2N}{2m}-1}|\Omega\setminus\overline{\omega}_{\varepsilon}|+\varepsilon^{-1}|\omega_{\varepsilon}|. The first summand goes to zero as ε0\varepsilon\rightarrow 0. For the second summand we note that since Ω\Omega is of class C2C^{2}, it is standard to prove that |ωε|=ε|Ω|+o(ε)|\omega_{\varepsilon}|=\varepsilon|\partial\Omega|+o(\varepsilon) as ε0+\varepsilon\rightarrow 0^{+}. This concludes the proof of point i)i).

We consider now point ii)ii). We note that for all ε]0,ε0[\varepsilon\in]0,\varepsilon_{0}[, ρε=ε12mNρ~ε\rho_{\varepsilon}=\varepsilon^{1-\frac{2m}{N}}\tilde{\rho}_{\varepsilon}, where

ρ~ε:={ε1,inωε,ε,inΩω¯ε.\tilde{\rho}_{\varepsilon}:=\begin{cases}\varepsilon^{-1},&{\rm in\ }\omega_{\varepsilon},\\ \varepsilon,&{\rm in\ }\Omega\setminus\overline{\omega}_{\varepsilon}.\end{cases}

We note that μ\mu\in\mathbb{R} is an eigenvalue of (2.1) with ρ=ρε\rho=\rho_{\varepsilon} if and only if μ~:=ε12mNμ\tilde{\mu}:=\varepsilon^{1-\frac{2m}{N}}\mu is an eigenvalue of problem (2.1) with ρ=ρ~ε\rho=\tilde{\rho}_{\varepsilon}. Problem (2.1) with ρ=ρ~ε\rho=\tilde{\rho}_{\varepsilon} admits an increasing sequence of non-negative eigenvalues of finite multiplicity given by

0=μ1[ρ~ε]==μdN,m[ρ~ε]<μdN,m+1[ρ~ε]μj[ρ~ε]+.0=\mu_{1}[\tilde{\rho}_{\varepsilon}]=\cdots=\mu_{d_{N,m}}[\tilde{\rho}_{\varepsilon}]<\mu_{d_{N,m}+1}[\tilde{\rho}_{\varepsilon}]\leq\cdots\leq\mu_{j}[\tilde{\rho}_{\varepsilon}]\leq\cdots\nearrow+\infty.

Now from Theorem 4.18, it follows that for all jj\in\mathbb{N},

limε0ε12mNμj[ρε]=limε0μj[ρ~ε]=|Ω|σj,\lim_{\varepsilon\rightarrow 0}\varepsilon^{1-\frac{2m}{N}}\mu_{j}[\rho_{\varepsilon}]=\lim_{\varepsilon\rightarrow 0}\mu_{j}[\tilde{\rho}_{\varepsilon}]=|\partial\Omega|\sigma_{j},

hence, for jdN,m+1j\geq d_{N,m}+1 limε0+ε12mNμj[ρε]=|Ω|σj>0\lim_{\varepsilon\rightarrow 0^{+}}\varepsilon^{1-\frac{2m}{N}}\mu_{j}[\rho_{\varepsilon}]=|\partial\Omega|\sigma_{j}>0. The proof is now complete.

5. Lower bounds

In this last section we shall discuss the issue of the lower bounds. In many situations (e.g., shape optimization problems) the problem of minimization of the eigenvalues leads to trivial solutions in the case of Neumann boundary conditions. Nevertheless, the eigenvalue problems with density which we have considered in this paper show an interesting behavior with respect to lower bounds, both if we fix the total mass or the LN2mL^{\frac{N}{2m}}-norm of the density. In the first case, we are able to show that there exist densities which preserve the total mass for which the jj-th eigenvalue can be made arbitrarily close to zero if N2mN\geq 2m (which is the case when upper bounds with mass constraint exist). This is stated in Theorem 5.4. On the contrary, if N<2mN<2m, the first positive eigenvalue is uniformly bounded from below by a positive constant which depends only on mm, NN and Ω\Omega divided by the total mass (in this case we recall that upper bounds with mass constraint do not exist). This is stated in Theorem 5.1.

When we choose as a constraint the LN2mL^{\frac{N}{2m}}-norm of the density, we see that exactly the opposite happens: if N2mN\leq 2m we find densities with prescribed LN2mL^{\frac{N}{2m}}-norm such that the jj-th eigenvalue can be made arbitrarily close to zero (in this case we have conjectured the existence of upper bounds of the form (4.1)), see Theorem 5.15; if N>2mN>2m, then the first positive eigenvalue is uniformly bounded from below by a positive constant which depends only on mm, NN and Ω\Omega divided by the LN2mL^{\frac{N}{2m}}-norm of the density, and in this case we recall that upper bounds with LN2mL^{\frac{N}{2m}} constraint do not exist. This is stated in Theorem 5.13.

We present now the precise statements and the corresponding proofs of such phenomena.

We start with the following theorem concerning lower bounds with mass constraint:

Theorem 5.1.

Let Ω\Omega be a bounded domain in N\mathbb{R}^{N}, N<2mN<2m, with Lipschitz boundary. Then there exists a positive constant Cm,N,ΩC_{m,N,\Omega} which depends only on mm, NN and Ω\Omega such that for every ρ\rho\in\mathcal{R}

(5.2) μdN,m+1[ρ]Cm,N,ΩΩρ𝑑x.\mu_{d_{N,m}+1}[\rho]\geq\frac{C_{m,N,\Omega}}{\int_{\Omega}\rho dx}.
Proof.

We recall from (2.7) that

μdN,m+1[ρ]=infVHm(Ω)dimV=dN,m+1sup0uVΩ|Dmu|2𝑑xΩρu2𝑑x.\mu_{d_{N,m}+1}[\rho]=\inf_{\begin{subarray}{c}V\leq H^{m}(\Omega)\\ {\rm dim}V=d_{N,m}+1\end{subarray}}\sup_{\begin{subarray}{c}0\neq u\in V\end{subarray}}\frac{\int_{\Omega}|D^{m}u|^{2}dx}{\int_{\Omega}\rho u^{2}dx}.

Since uHm(Ω)u\in H^{m}(\Omega) and N<2mN<2m, the standard Sobolev inequality (B.2) implies that there exists a constant CC which depends only on mm, NN and Ω\Omega such that

uC0(Ω)2CuHm(Ω)2.\|u\|_{C^{0}(\Omega)}^{2}\leq C\|u\|_{H^{m}(\Omega)}^{2}.

Then for all uH2(Ω)u\in H^{2}(\Omega)

Ω|Dmu|2𝑑xΩρu2𝑑xΩ|Dmu|2𝑑xuC0(Ω)2Ωρ𝑑x1CΩρ𝑑xΩ|Dmu|2𝑑x(Ω|Dmu|2𝑑x+Ωu2𝑑x).\frac{\int_{\Omega}|D^{m}u|^{2}dx}{\int_{\Omega}\rho u^{2}dx}\geq\frac{\int_{\Omega}|D^{m}u|^{2}dx}{\|u\|_{C^{0}(\Omega)}^{2}\int_{\Omega}\rho dx}\geq\frac{1}{C\int_{\Omega}\rho dx}\frac{\int_{\Omega}|D^{m}u|^{2}dx}{(\int_{\Omega}|D^{m}u|^{2}dx+\int_{\Omega}u^{2}dx)}.

Hence

μdN,m+1[ρ]=infVHm(Ω)dimV=dN,m+1sup0uVΩ|Dmu|2𝑑xΩρu2𝑑x1CΩρ𝑑xinfVHm(Ω)dimV=dN,m+1sup0uVΩ|Dmu|2𝑑xΩ|Dmu|2𝑑x+Ωu2𝑑x1CΩρ𝑑xμdN,m+1[1]1+μdN,m+1[1].\mu_{d_{N,m}+1}[\rho]=\inf_{\begin{subarray}{c}V\leq H^{m}(\Omega)\\ {\rm dim}V=d_{N,m}+1\end{subarray}}\sup_{\begin{subarray}{c}0\neq u\in V\end{subarray}}\frac{\int_{\Omega}|D^{m}u|^{2}dx}{\int_{\Omega}\rho u^{2}dx}\\ \geq\frac{1}{C\int_{\Omega}\rho dx}\inf_{\begin{subarray}{c}V\leq H^{m}(\Omega)\\ {\rm dim}V=d_{N,m}+1\end{subarray}}\sup_{\begin{subarray}{c}0\neq u\in V\end{subarray}}\frac{\int_{\Omega}|D^{m}u|^{2}dx}{\int_{\Omega}|D^{m}u|^{2}dx+\int_{\Omega}u^{2}dx}\\ \geq\frac{1}{C\int_{\Omega}\rho dx}\frac{\mu_{d_{N,m}+1}[1]}{1+\mu_{d_{N,m}+1}[1]}.

Then (5.2) holds with Cm,N,Ω=μdN,m+1[1]C(1+μdN,m+1[1])C_{m,N,\Omega}=\frac{\mu_{d_{N,m}+1}[1]}{C(1+\mu_{d_{N,m}+1}[1])}. This concludes the proof. ∎

Densities which preserve the total mass and produce jj arbitrarily small eigenvalues can be given, for example, by concentrating all the mass around jj distinct points of Ω\Omega. For all jj\in\mathbb{N} let us fix once for all jj points a1,,ajΩa_{1},...,a_{j}\in\Omega and a number ε0]0,1[\varepsilon_{0}\in]0,1[ such that Bi0:=B(ai,ε0)ΩB_{i}^{0}:=B(a_{i},\varepsilon_{0})\subset\subset\Omega and Bi0B_{i}^{0} are disjoint. For ε]0,ε0[\varepsilon\in]0,\varepsilon_{0}[, we will write Biε:=B(ai,ε)B_{i}^{\varepsilon}:=B(a_{i},\varepsilon). Let ρε,j\rho_{\varepsilon,j}\in\mathcal{R} defined by

(5.3) ρε,j:=ε+i=1jεNχBiε.\rho_{\varepsilon,j}:=\varepsilon+\sum_{i=1}^{j}\varepsilon^{-N}\chi_{B_{i}^{\varepsilon}}.

We have the following theorem:

Theorem 5.4.

Let Ω\Omega be a bounded domain in N\mathbb{R}^{N}, with N2mN\geq 2m, such that the embedding Hm(Ω)L2(Ω)H^{m}(\Omega)\subset L^{2}(\Omega) is compact. Let ρε,j\rho_{\varepsilon,j}\in\mathcal{R} be defined by (5.3) for all ε]0,ε0[\varepsilon\in]0,\varepsilon_{0}[ and jj\in\mathbb{N}. Then

  1. i)

    limε0+Ωρε,j𝑑x=jωN\lim_{\varepsilon\rightarrow 0^{+}}\int_{\Omega}\rho_{\varepsilon,j}dx=j\omega_{N};

  2. ii)

    μj[ρε]CN,mεN2m\mu_{j}[\rho_{\varepsilon}]\leq C_{N,m}\varepsilon^{N-2m} if N>2mN>2m;

  3. iii)

    μj[ρε]Cm,Ω,j|log(ε)|\mu_{j}[\rho_{\varepsilon}]\leq\frac{C_{m,\Omega,j}}{|\log(\varepsilon)|} if N=2mN=2m,

where CN,m,Cm,Ω,jC_{N,m},C_{m,\Omega,j} are positive constants which depend only on m,Nm,N and m,Ω,jm,\Omega,j respectively.

Proof.

We start with point i)i). We have

Ωρε,j𝑑x=ε|Ω|+i=1jεN|Biε|=ε|Ω|+jωN,\int_{\Omega}\rho_{\varepsilon,j}dx=\varepsilon|\Omega|+\sum_{i=1}^{j}\varepsilon^{-N}|B_{i}^{\varepsilon}|=\varepsilon|\Omega|+j\omega_{N},

which yields the result.

We prove now ii)ii). Let N>2mN>2m and let us fix jj\in\mathbb{N}. Let aiΩa_{i}\in\Omega, i=1,,ji=1,...,j and ε0]0,1[\varepsilon_{0}\in]0,1[ be as in the definition of ρε,j\rho_{\varepsilon,j} in (5.3). Let 21Biε:=B(ai,ε/2)2^{-1}B_{i}^{\varepsilon}:=B(a_{i},\varepsilon/2). Associated to each BiεB_{i}^{\varepsilon} we construct a function uiHm(Ω)u_{i}\in H^{m}(\Omega) which is supported in BiεB_{i}^{\varepsilon} and such that ui1u_{i}\equiv 1 on 21Biε2^{-1}B_{i}^{\varepsilon} in the following way:

ui(x):={U(|xai|),ifε2|xai|ε,1,if|xai|ε2,0,if|xai|ε,u_{i}(x):=\begin{cases}U(|x-a_{i}|),&{\rm if\ }\frac{\varepsilon}{2}\leq|x-a_{i}|\leq\varepsilon,\\ 1,&{\rm if\ }|x-a_{i}|\leq\frac{\varepsilon}{2},\\ 0,&{\rm if\ }|x-a_{i}|\geq\varepsilon,\end{cases}

where

U(t):=i=02m1αiεiti.U(t):=\sum_{i=0}^{2m-1}\frac{\alpha_{i}}{\varepsilon^{i}}t^{i}.

The coefficients αi\alpha_{i}, i=0,,2m1i=0,...,2m-1 are uniquely determined by imposing U(ε/2)=1U(\varepsilon/2)=1, U(ε)=0U(\varepsilon)=0, U(l)(ε/2)=U(l)(ε)=0U^{(l)}(\varepsilon/2)=U^{(l)}(\varepsilon)=0 for all l=1,,m1l=1,...,m-1 and depend only on mm (see also (3.7), (3.8) and (3.9) in the proof of Theorem 3.4).

Now we estimate the Rayleigh quotient Ω|Dmui|2𝑑xΩρε,jui2𝑑x\frac{\int_{\Omega}|D^{m}u_{i}|^{2}dx}{\int_{\Omega}\rho_{\varepsilon,j}u_{i}^{2}dx} of uiu_{i}, for all i=1,..,ji=1,..,j. We start from the numerator. As in the proof of Theorem 3.4 (see (3.19) and (3.20)) we have that

(5.5) Ω|Dmui2|𝑑x=Biε21Biε|Dmui|2𝑑xCN,m|Biε21Biε|12mN(12N)CN,mεN2m,\int_{\Omega}|D^{m}u_{i}^{2}|dx=\int_{B_{i}^{\varepsilon}\setminus 2^{-1}B_{i}^{\varepsilon}}|D^{m}u_{i}|^{2}dx\leq C_{N,m}|B_{i}^{\varepsilon}\setminus 2^{-1}B_{i}^{\varepsilon}|^{1-\frac{2m}{N}}\\ \leq(1-2^{-N})C_{N,m}\varepsilon^{N-2m},

where CN,m>0C_{N,m}>0 depends only on mm and NN and can be eventually re-defined through the rest of the proof. For the denominator we have

(5.6) Ωρε,jui2𝑑x21Biερε,jui2𝑑x=εN|21Biε|=2NωN.\int_{\Omega}\rho_{\varepsilon,j}u_{i}^{2}dx\geq\int_{2^{-1}B_{i}^{\varepsilon}}\rho_{\varepsilon,j}u_{i}^{2}dx=\varepsilon^{-N}|2^{-1}B_{i}^{\varepsilon}|=2^{-N}\omega_{N}.

From (5.5), (5.6) and the min-max principle (2.7) and from the fact that {ui}i=1j\left\{u_{i}\right\}_{i=1}^{j} is a set of jj disjointly supported functions, it follows that

μj[ρε,j]maxu1,,ujΩ|Dmui|2𝑑xΩρε,jui2𝑑xCN,mεN2m\mu_{j}[\rho_{\varepsilon,j}]\leq\max_{u_{1},...,u_{j}}\frac{\int_{\Omega}|D^{m}u_{i}|^{2}dx}{\int_{\Omega}\rho_{\varepsilon,j}u_{i}^{2}dx}\leq C_{N,m}\varepsilon^{N-2m}

(see also (3.13), (3.14) in the proof of Theorem 3.4). This concludes the proof of ii)ii).

Consier now iii)iii). Let N=2mN=2m. Again, let us fix jj\in\mathbb{N}. Let aiΩa_{i}\in\Omega, i=1,,ji=1,...,j and ε0]0,1[\varepsilon_{0}\in]0,1[ be as in the definition of (5.3) (we note that admissible values for ε0\varepsilon_{0} depend on Ω\Omega and jj). Associated to each Bi0B_{i}^{0} we construct for all ε]0,ε0[\varepsilon\in]0,\varepsilon_{0}[ a function uε,iHm(Ω)u_{\varepsilon,i}\in H^{m}(\Omega) which is supported in Bi0B_{i}^{0} in the following way:

ui,ε(x):={U1(|xai|),ifε|xai|ε0,U2(|xai|),if|xai|ε,0,if|xai|ε0,u_{i,\varepsilon}(x):=\begin{cases}U_{1}(|x-a_{i}|),&{\rm if\ }\varepsilon\leq|x-a_{i}|\leq\varepsilon_{0},\\ U_{2}(|x-a_{i}|),&{\rm if\ }|x-a_{i}|\leq\varepsilon,\\ 0,&{\rm if\ }|x-a_{i}|\geq\varepsilon_{0},\end{cases}

where

U1(t):=log(t)+log(ε0)k=1m1(ε0t)kkε0kU_{1}(t):=-\log(t)+\log(\varepsilon_{0})-\sum_{k=1}^{m-1}\frac{(\varepsilon_{0}-t)^{k}}{k\varepsilon_{0}^{k}}

and

U2(t):=α(ε)+k=0m2αk(ε)tm+k,U_{2}(t):=\alpha(\varepsilon)+\sum_{k=0}^{m-2}\alpha_{k}(\varepsilon)t^{m+k},

where the coefficients α(ε)\alpha(\varepsilon) and αk(ε)\alpha_{k}(\varepsilon) are uniquely determined by imposing U1(l)(ε)=U2(l)(ε)U_{1}^{(l)}(\varepsilon)=U_{2}^{(l)}(\varepsilon) for all 0lm10\leq l\leq m-1. Moreover (possibly choosing a smaller value for ε0\varepsilon_{0}), it is standard to prove that there exist positive constants c1c_{1} and c2c_{2} which depend only on mm and ε0\varepsilon_{0} (and hence on mm, Ω\Omega and jj) such that c1|log(t)||U1(t)|c2|log(t)|c_{1}|\log(t)|\leq|U_{1}(t)|\leq c_{2}|\log(t)|, c1tl|U1(l)(t)|c2tlc_{1}t^{-l}\leq|U_{1}^{(l)}(t)|\leq c_{2}t^{-l} for all t]ε,ε0[t\in]\varepsilon,\varepsilon_{0}[ and 0lm10\leq l\leq m-1, and that c1|log(ε)||α(ε)|c2|log(ε)|c_{1}|\log(\varepsilon)|\leq|\alpha(\varepsilon)|\leq c_{2}|\log(\varepsilon)| and c1εmk|αk(ε)|c2εmkc_{1}\varepsilon^{-m-k}\leq|\alpha_{k}(\varepsilon)|\leq c_{2}\varepsilon^{-m-k}. In particular (possibly re-defining the constants c1c_{1} and c2c_{2} and choosing a smaller value for ε0\varepsilon_{0}), we have that c1|log(ε)||U2(t)|c2|log(ε)|c_{1}|\log(\varepsilon)|\leq|U_{2}(t)|\leq c_{2}|\log(\varepsilon)| for all t[0,ε]t\in[0,\varepsilon].

From the min-max principle (2.7) and from the fact that {ui}i=1j\left\{u_{i}\right\}_{i=1}^{j} is a set of jj disjointly supported functions, it follows that

(5.7) μj[ρε,j]maxu1,,ujΩ|Dmui|2𝑑xΩρε,jui2𝑑x\mu_{j}[\rho_{\varepsilon,j}]\leq\max_{u_{1},...,u_{j}}\frac{\int_{\Omega}|D^{m}u_{i}|^{2}dx}{\int_{\Omega}\rho_{\varepsilon,j}u_{i}^{2}dx}

(see also (3.13), (3.14) in the proof of Theorem 3.4). It remains then to estimate the Rayleigh quotient of all the function uiu_{i}. We have for the denominator

(5.8) Ωρε,jui2𝑑xε2mBiεui2𝑑xε2mc1|log(ε)|2|Biε|Cm,Ω,j|log(ε)|2.\int_{\Omega}\rho_{\varepsilon,j}u_{i}^{2}dx\geq\varepsilon^{-2m}\int_{B_{i}^{\varepsilon}}u_{i}^{2}dx\geq\varepsilon^{-2m}c_{1}|\log(\varepsilon)|^{2}|B_{i}^{\varepsilon}|\geq C_{m,\Omega,j}|\log(\varepsilon)|^{2}.

From now on Cm,Ω,jC_{m,\Omega,j} will denote a positive constat which depends only on m,Ωm,\Omega and jj. For the numerator, we have, since the functions uiu_{i} are radial with respect to aia_{i} (see also (3.17))

(5.9) Ω|Dmui|2𝑑x=Bi0|Dmui|2𝑑x=Bi0Biε|Dmui|2𝑑x+Biε|Dmui|2𝑑xCm,Ω,jk=1mBi0Biε(U1(k)(|xai|))2|xai|2(mk)𝑑x+Cm,Ω,jk=1mBiε(U2(k)(|xai|))2|xai|2(mk)𝑑x.\int_{\Omega}|D^{m}u_{i}|^{2}dx=\int_{B_{i}^{0}}|D^{m}u_{i}|^{2}dx=\int_{B_{i}^{0}\setminus B_{i}^{\varepsilon}}|D^{m}u_{i}|^{2}dx+\int_{B_{i}^{\varepsilon}}|D^{m}u_{i}|^{2}dx\\ \leq C_{m,\Omega,j}\sum_{k=1}^{m}\int_{B_{i}^{0}\setminus B_{i}^{\varepsilon}}\frac{(U_{1}^{(k)}(|x-a_{i}|))^{2}}{|x-a_{i}|^{2(m-k)}}dx+C_{m,\Omega,j}\sum_{k=1}^{m}\int_{B_{i}^{\varepsilon}}\frac{(U_{2}^{(k)}(|x-a_{i}|))^{2}}{|x-a_{i}|^{2(m-k)}}dx.

Since |U1(k)(t)|2c22t2k|U_{1}^{(k)}(t)|^{2}\leq c_{2}^{2}t^{-2k} and N=2mN=2m, we have

(5.10) Bi0Biε(U1(k)(|xai|))2|xai|2(mk)𝑑xNCm,Ω,jεε0t2m+N1𝑑tCm,Ω,j|log(ε)|.\int_{B_{i}^{0}\setminus B_{i}^{\varepsilon}}\frac{(U_{1}^{(k)}(|x-a_{i}|))^{2}}{|x-a_{i}|^{2(m-k)}}dx\leq NC_{m,\Omega,j}\int_{\varepsilon}^{\varepsilon_{0}}t^{-2m+N-1}dt\leq C_{m,\Omega,j}|\log(\varepsilon)|.

Moreover, |U2(k)(t)|2(m1)c22i=0m2ε2m2it2m+2i2k|U_{2}^{(k)}(t)|^{2}\leq(m-1)c_{2}^{2}\sum_{i=0}^{m-2}\varepsilon^{-2m-2i}t^{2m+2i-2k}, hence

(5.11) k=1mBiε(U2(k)(|xai|))2|xai|2(mk)𝑑xCm,Ω,ji=0m2ε2m2i0εt2i+N1Cm,Ω,j.\sum_{k=1}^{m}\int_{B_{i}^{\varepsilon}}\frac{(U_{2}^{(k)}(|x-a_{i}|))^{2}}{|x-a_{i}|^{2(m-k)}}dx\leq C_{m,\Omega,j}\sum_{i=0}^{m-2}\varepsilon^{-2m-2i}\int_{0}^{\varepsilon}t^{2i+N-1}\leq C_{m,\Omega,j}.

From (5.9), (5.10) and (5.11) we have that

(5.12) Ω|Dmui|2𝑑xCm,Ω,j|log(ε)|.\int_{\Omega}|D^{m}u_{i}|^{2}dx\leq C_{m,\Omega,j}|\log(\varepsilon)|.

By combining (5.8) and (5.12), from (5.7) we deduce that μj[ρε,j]Cm,Ω,j|log(ε)|\mu_{j}[\rho_{\varepsilon,j}]\leq\frac{C_{m,\Omega,j}}{|\log(\varepsilon)|}. This concludes the proof for the case N=2mN=2m and of the theorem. ∎

We consider now lower bounds with the non-linear constraint ΩρN2m𝑑x=const.\int_{\Omega}\rho^{\frac{N}{2m}}dx={\rm const.} and N>2mN>2m. We have the following theorem:

Theorem 5.13.

Let Ω\Omega be a bounded domain in N\mathbb{R}^{N}, N>2mN>2m, with Lipschitz boundary. Then there exists a positive constant Cm,N,ΩC_{m,N,\Omega} which depends only on mm, NN and Ω\Omega such that for every ρ\rho\in\mathcal{R}

(5.14) μdN,m+1[ρ]Cm,N,Ω(ΩρN2m𝑑x)2mN.\mu_{d_{N,m}+1}[\rho]\geq\frac{C_{m,N,\Omega}}{\left(\int_{\Omega}\rho^{\frac{N}{2m}}dx\right)^{\frac{2m}{N}}}.
Proof.

We recall from (2.7) that

μdN,m+1[ρ]=infVHm(Ω)dimV=dN,m+1sup0uVΩ|Dmu|2𝑑xΩρu2𝑑x.\mu_{d_{N,m}+1}[\rho]=\inf_{\begin{subarray}{c}V\leq H^{m}(\Omega)\\ {\rm dim}V=d_{N,m}+1\end{subarray}}\sup_{\begin{subarray}{c}0\neq u\in V\end{subarray}}\frac{\int_{\Omega}|D^{m}u|^{2}dx}{\int_{\Omega}\rho u^{2}dx}.

Since uHm(Ω)u\in H^{m}(\Omega) and N>2mN>2m, the standard Sobolev inequality (B.3) implies that there exists a constant CC which depends only on mm, NN and Ω\Omega such that

uL2NN2m(Ω)2CuHm(Ω)2.\|u\|_{L^{\frac{2N}{N-2m}}(\Omega)}^{2}\leq C\|u\|_{H^{m}(\Omega)}^{2}.

Then for all uH2(Ω)u\in H^{2}(\Omega)

Ω|Dmu|2𝑑xΩρu2𝑑xΩ|Dmu|2𝑑x(ΩρN2m𝑑x)2mN(Ωu2NN2m𝑑x)N2mN1CρLN2m(Ω)Ω|Dmu|2𝑑xΩ|Dmu|2𝑑x+Ωu2𝑑x,\frac{\int_{\Omega}|D^{m}u|^{2}dx}{\int_{\Omega}\rho u^{2}dx}\geq\frac{\int_{\Omega}|D^{m}u|^{2}dx}{\left(\int_{\Omega}\rho^{\frac{N}{2m}}dx\right)^{\frac{2m}{N}}\left(\int_{\Omega}u^{\frac{2N}{N-2m}}dx\right)^{\frac{N-2m}{N}}}\\ \geq\frac{1}{C\|\rho\|_{L^{\frac{N}{2m}}(\Omega)}}\frac{\int_{\Omega}|D^{m}u|^{2}dx}{\int_{\Omega}|D^{m}u|^{2}dx+\int_{\Omega}u^{2}dx},

where we have used a Hölder inequality in the first line. Hence

μdN,m+1[ρ]=infVHm(Ω)dimV=dN,m+1sup0uVΩ|Dmu|2𝑑xΩρu2𝑑x1CρLN2m(Ω)infVHm(Ω)dimV=dN,m+1sup0uVΩ|Dmu|2𝑑xΩ|Dmu|2𝑑x+Ωu2𝑑x1CρLN2m(Ω)μdN,m+1[1]1+μdN,m+1[1].\mu_{d_{N,m}+1}[\rho]=\inf_{\begin{subarray}{c}V\leq H^{m}(\Omega)\\ {\rm dim}V=d_{N,m}+1\end{subarray}}\sup_{\begin{subarray}{c}0\neq u\in V\end{subarray}}\frac{\int_{\Omega}|D^{m}u|^{2}dx}{\int_{\Omega}\rho u^{2}dx}\\ \geq\frac{1}{C\|\rho\|_{L^{\frac{N}{2m}}(\Omega)}}\inf_{\begin{subarray}{c}V\leq H^{m}(\Omega)\\ {\rm dim}V=d_{N,m}+1\end{subarray}}\sup_{\begin{subarray}{c}0\neq u\in V\end{subarray}}\frac{\int_{\Omega}|D^{m}u|^{2}dx}{\int_{\Omega}|D^{m}u|^{2}dx+\int_{\Omega}u^{2}dx}\\ \geq\frac{1}{C\|\rho\|_{L^{\frac{N}{2m}}(\Omega)}}\frac{\mu_{d_{N,m}+1}[1]}{1+\mu_{d_{N,m}+1}[1]}.

Hence formula (5.14) holds with Cm,N,Ω=μdN,m+1[1]C(1+μdN,m+1[1])C_{m,N,\Omega}=\frac{\mu_{d_{N,m}+1}[1]}{C(1+\mu_{d_{N,m}+1}[1])}. This ends the proof. ∎

Densities with prescribed LN2mL^{\frac{N}{2m}}-norm and which made the jj-th eigenvalue arbitrarily small in dimension N<2mN<2m are, for example, densities which explode in a ε\varepsilon-tubular neighborhood of the boundary of Ω\Omega. This is contained in the following theorem.

Theorem 5.15.

Let Ω\Omega be a bounded domain in N\mathbb{R}^{N} with N<2mN<2m, of class C2C^{2}. Let ρε\rho_{\varepsilon}\in\mathcal{R} be defined by (4.15) for all ε]0,ε0[\varepsilon\in]0,\varepsilon_{0}[ and jj\in\mathbb{N}. Then

  1. i)

    limε0+ΩρεN2m𝑑x=|Ω|\lim_{\varepsilon\rightarrow 0^{+}}\int_{\Omega}\rho_{\varepsilon}^{\frac{N}{2m}}dx=|\partial\Omega|;

  2. ii)

    for all jj\in\mathbb{N}, jdN,m+1j\geq d_{N,m}+1, there exists cj>0c_{j}>0 which depends only on mm, NN, Ω\Omega and jj such that limε0+μj[ρε]ε12mN=cj\lim_{\varepsilon\rightarrow 0^{+}}\mu_{j}[\rho_{\varepsilon}]\varepsilon^{1-\frac{2m}{N}}=c_{j}.

Proof.

The proof is the same as that of Theorem 4.16 and is accordingly omitted. ∎

From Theorem 5.15 it follows that limε0+μj[ρε]=0\lim_{\varepsilon\rightarrow 0^{+}}\mu_{j}[\rho_{\varepsilon}]=0 for all jj\in\mathbb{N} if N<2mN<2m. We remark that also in the case N=2mN=2m we find densities which make the jj-th eigenvalue arbitrarily small, in fact this is stated by point iii)iii) of Theorem 5.4.

Appendices

Appendix A Eigenvalues of polyharmonic operators

In this section we shall present some basics of spectral theory for the polyharmonic operators. In particular, we will discuss Neumann boundary conditions, mainly for the Laplace and the biharmonic operator. Then we will characterize the spectrum of the polyharmonic operators subjet to Neumann boundary conditions by exploiting classical tools of spectral theory for compact self-adjoint operators. We refer to [21, 32] and to the references therein for a discussion on eigenvalue problems for general elliptic operators of order 2m2m with density subject to homogeneous boundary conditions.

A.1. Neumann boundary conditions

Neumann boundary conditions are usually called ‘natural’ boundary conditions. This is well understood for the Laplace operator. In fact, assume that uu is a classical solution of (2.4). If we multiply the equation Δu=μρu-\Delta u=\mu\rho u by a test function φC(Ω)\varphi\in C^{\infty}(\Omega) and integrate both sides of the resulting identity over Ω\Omega, thanks to Green’s formula we obtain:

Ωuφdx=μΩρuφ𝑑x+Ωuνφ𝑑σ=μΩρuφ𝑑x.\int_{\Omega}\nabla u\cdot\nabla\varphi dx=\mu\int_{\Omega}\rho u\varphi dx+\int_{\partial\Omega}\frac{\partial u}{\partial\nu}\varphi d\sigma=\mu\int_{\Omega}\rho u\varphi dx.

Hence (2.1) with m=1m=1 holds for all φC(Ω)\varphi\in C^{\infty}(\Omega) when uu is a solution of (2.2). We can relax our hypothesis on uu and just require that uH1(Ω)u\in H^{1}(\Omega) and that (2.1) holds for all φH1(Ω)\varphi\in H^{1}(\Omega). Hence (2.1) is the weak formulation of the Neumann eigenvalue problem for the Laplace operator. We note that the boundary condition in (2.2) arises naturally and is not imposed a priori with the choice of a subspace of H1(Ω)H^{1}(\Omega) in the weak formulation (as in the case of H01(Ω)H^{1}_{0}(\Omega) for Dirichlet conditions): if a weak solution of (2.1) for m=1m=1 exists and is sufficiently smooth, then it solves Δu=λρu-\Delta u=\lambda\rho u in the classical sense and satisfies the Neumann boundary condition uν=0\frac{\partial u}{\partial\nu}=0.

Let us consider now more in detail the case of the biharmonic operator. Assume that uu is a classical solution of problem (2.5). We multiply the equation Δ2u=μρu\Delta^{2}u=\mu\rho u by a test function φC(Ω)\varphi\in C^{\infty}(\Omega) and apply the biharmonic Green’s formula (see [1, Lemma 8.56]). We obtain:

ΩΔ2uφ𝑑x=ΩD2u:D2φdx+Ω(divΩ(D2u.ν)Ω+Δuν)φdσΩ2uν2φνdσ=ΩD2u:D2φdx=μΩρuφ𝑑x,\int_{\Omega}\Delta^{2}u\varphi dx=\int_{\Omega}D^{2}u:D^{2}\varphi dx\\ +\int_{\partial\Omega}\left({\rm div}_{\partial\Omega}\left(D^{2}u.\nu\right)_{\partial\Omega}+\frac{\partial\Delta u}{\partial\nu}\right)\varphi d\sigma-\int_{\partial\Omega}\frac{\partial^{2}u}{\partial\nu^{2}}\frac{\partial\varphi}{\partial\nu}d\sigma\\ =\int_{\Omega}D^{2}u:D^{2}\varphi dx=\mu\int_{\Omega}\rho u\varphi dx,

where (D2u.ν)Ω(D^{2}u.\nu)_{\partial\Omega} denotes the tangential component of D2u.νD^{2}u.\nu. Hence (2.1) with m=2m=2 holds for all φC(Ω)\varphi\in C^{\infty}(\Omega) when uu is a solution of problem (2.5) (we remark that if 2uν2=0\frac{\partial^{2}u}{\partial\nu^{2}}=0 then (D2u.ν)Ω=(D2u.ν)(D^{2}u.\nu)_{\partial\Omega}=(D^{2}u.\nu)). We can relax our hypothesis on uu and just require that uH2(Ω)u\in H^{2}(\Omega) and that (2.1) holds for all φH2(Ω)\varphi\in H^{2}(\Omega). This is exactly the weak formulation of the Neumann eigenvalue problem for the biharmonic operator. We note again that the two boundary conditions in (2.5) arise naturally and are not imposed a priori: if a weak solution of (2.1) exists and is sufficiently smooth, then it satisfies the two Neumann boundary conditions. We also remark that if Ω\Omega is sufficiently regular, e.g., if it is of class CkC^{k} with k>4+N2k>4+\frac{N}{2} and ρ\rho is continuous, then a weak solution of (2.1) with m=2m=2 is actually a classical solution of (2.5) (see [25, § 2]). The choice of the whole space H2(Ω)H^{2}(\Omega) in the weak formulation (2.1) contains the information on the boundary conditions in (2.2).

It is natural then to consider problem (2.1) for any mm\in\mathbb{N} as the weak formulation of an eigenvalue problem for the polyharmonic operator with Neumann boundary conditions. In the case of a generic value of mm it is much more difficult to write explicitly the boundary operators 𝒩0,,𝒩m1\mathcal{N}_{0},...,\mathcal{N}_{m-1} (this is already extremely involved for m=3m=3). If moreover Ω\Omega is sufficiently regular and ρ\rho is continuous, then weak solutions of (2.1) are actually classical solution of (2.2), and the mm bounday conditions are uniquely determined and arise naturally from the choice of the whole space Hm(Ω)H^{m}(\Omega) (see [25] for further discussions on higher order elliptic operators and eigenvalue problems).

A.2. Characterization of the spectrum

The aim of this subsection is to prove that problem (2.1) admits an increasing sequence of non-negative eigenvalues of finite multiplicity diverging to ++\infty, and to provide some additional information on the spectrum. To do so, we will reduce problem (2.1) to an eigenvalue problem for a compact self-adjoint operator on a Hilbert space.

We define first the following (equivalent) problem: find uHm(Ω)u\in H^{m}(\Omega) and Λ\Lambda\in\mathbb{R} such that

(A.1) ΩDmu:Dmφ+ρuφdx=ΛΩρuφ𝑑x,φHm(Ω).\int_{\Omega}D^{m}u:D^{m}\varphi+\rho u\varphi dx=\Lambda\int_{\Omega}\rho u\varphi dx\,,\ \ \ \forall\varphi\in H^{m}(\Omega).

Clearly the eigenfuctions of (A.1) coincide with the eigenfunctions of (2.1), while all the eigenvalues μ\mu of (2.1) are given by μ=Λ1\mu=\Lambda-1, where Λ\Lambda is an eigenvalue of (A.1).

We consider the operator (Δ)m+ρId(-\Delta)^{m}+\rho I_{d} as a map from Hm(Ω)H^{m}(\Omega) to its dual Hm(Ω)H^{m}(\Omega)^{\prime} defined by

((Δ)m+ρId)[u][φ]:=ΩDmu:Dmφ+ρuφdx, 8φHm(Ω).((-\Delta)^{m}+\rho I_{d})[u][\varphi]:=\int_{\Omega}D^{m}u:D^{m}\varphi+\rho u\varphi dx,\ \ \ \mathcal{8}\varphi\in H^{m}(\Omega).

The operator (Δ)m+ρId(-\Delta)^{m}+\rho I_{d} is a continuous isomorphism between Hm(Ω)H^{m}(\Omega) and Hm(Ω)H^{m}(\Omega)^{\prime}. In fact it follows immediately that there exist C1,C2>0C_{1},C_{2}>0 such that

(A.2) C1uHm(Ω)2((Δ)m+ρId)[u][u]C2uHm(Ω)2,uHm(Ω).C_{1}\|u\|^{2}_{H^{m}(\Omega)}\leq((-\Delta)^{m}+\rho I_{d})[u][u]\leq C_{2}\|u\|^{2}_{H^{m}(\Omega)},\ \forall\,u\in H^{m}(\Omega).

Next we denote by ii the canonical embedding of Hm(Ω)H^{m}(\Omega) into L2(Ω)L^{2}(\Omega) and by JρJ_{\rho} the embedding of L2(Ω)L^{2}(\Omega) into Hm(Ω)H^{m}(\Omega)^{\prime}, defined by

Jρ[u][φ]:=Ωρuφ𝑑x 8uL2(Ω),φHm(Ω).J_{\rho}[u][\varphi]:=\int_{\Omega}\rho u\varphi dx\ \ \ \mathcal{8}u\in L^{2}(\Omega),\varphi\in H^{m}(\Omega).

Let TρT_{\rho} be the operator from Hm(Ω)H^{m}(\Omega) to itself defined by Tρ:=((Δ)m+ρId)(1)JρiT_{\rho}:=((-\Delta)^{m}+\rho I_{d})^{(-1)}\circ J_{\rho}\circ i. Problem (A.1) is then equivalent to

Tρu=Λ1u,T_{\rho}u=\Lambda^{-1}u,

in the unknows uHm(Ω)u\in H^{m}(\Omega), Λ\Lambda\in\mathbb{R}. We now consider the space Hm(Ω)H^{m}(\Omega) endowed with the bilinear form

(A.3) u,vρ=ΩDmu:Dmv+ρuvdx, 8u,vHm(Ω).\langle u,v\rangle_{\rho}=\int_{\Omega}D^{m}u:D^{m}v+\rho uvdx,\ \ \ \mathcal{8}u,v\in H^{m}(\Omega).

From (A.2) it follows that (A.3) is a scalar product on Hm(Ω)H^{m}(\Omega) whose induced norm is equivalent to the standard one. We denote by Hρm(Ω){H^{m}_{\rho}}(\Omega) the space Hm(Ω)H^{m}(\Omega) endowed with the scalar product defined by (A.3). Then we can state the following theorem:

Theorem A.4.

Let Ω\Omega be a bounded domain in N\mathbb{R}^{N} such that the embedding Hm(Ω)L2(Ω)H^{m}(\Omega)\subset L^{2}(\Omega) is compact. Let ρ\rho\in\mathcal{R}. The operator Tρ:=((Δ)m+ρId)(1)JρiT_{\rho}:=((-\Delta)^{m}+\rho I_{d})^{(-1)}\circ J_{\rho}\circ i is a compact self-adjoint operator in Hρm(Ω){H^{m}_{\rho}}(\Omega), whose eigenvalues coincide with the reciprocals of the eigenvalues of problem (A.1) for all jj\in\mathbb{N}.

The proof of Theorem A.4 is standard, hence we omit it (see e.g., [8, § IX]). As a consequence of Theorem A.4 we have the following:

Theorem A.5.

Let Ω\Omega be a bounded domain in N\mathbb{R}^{N} such that the embedding Hm(Ω)L2(Ω)H^{m}(\Omega)\subset L^{2}(\Omega) is compact. Let ρ\rho\in\mathcal{R}. Then the set of the eigenvalues of (2.1) is contained in [0,+[[0,+\infty[ and consists of the image of a sequence increasing to ++\infty. The eigenvalue μ=0\mu=0 has multiplicity dN,m=(N+m1N)d_{N,m}=\binom{N+m-1}{N} and the eigenfunctions corresponding to the eigenvalue μ=0\mu=0 are the polynomials of degree at most m1m-1 in N\mathbb{R}^{N}. Each eigenvalue has finite multiplicity. Moreover the space Hρm(Ω)H^{m}_{\rho}(\Omega) has a Hilbert basis of eigenfunctions of problem (2.1).

Proof.

We note that ker(Tρ)={0}ker(T_{\rho})=\left\{0\right\}, hence by standard spectral theory it follows that the eigenvalues of TρT_{\rho} are positive and bounded and form an infinite sequence {λj}j\left\{\lambda_{j}\right\}_{j\in\mathbb{N}} converging to zero. Moreover to each eigenvalue λj\lambda_{j} is possible to associate an eigenfunction uju_{j} such that {uj}j\left\{u_{j}\right\}_{j\in\mathbb{N}} is a orthonormal basis of Hρm(Ω)H^{m}_{\rho}(\Omega).

From Theorem A.4 it follows that the eigenvalues of (A.1) form a sequence of real numbers increasing to ++\infty which is given by {Λj=λj1}j\left\{\Lambda_{j}=\lambda_{j}^{-1}\right\}_{j\in\mathbb{N}} and that the space Hρm(Ω)H^{m}_{\rho}(\Omega) has a Hilbert basis of eigenfunctions of (A.1). The eigenvalues μj\mu_{j} of (2.1) are given by μj=Λj1\mu_{j}=\Lambda_{j}-1 for all jj\in\mathbb{N}, where {Λj}j\left\{\Lambda_{j}\right\}_{j\in\mathbb{N}} are the eigenvalues of (A.1) and the eigenfunctions associated with Λj\Lambda_{j} coincide with the eigenfunctions associated with μj=Λj1\mu_{j}=\Lambda_{j}-1. Moreover, given an eigenvalue μ\mu of (2.1) and a corresponding eigenfunction uu, we have that

Ω|Dmu|2𝑑x=μΩρu2𝑑x,\int_{\Omega}|D^{m}u|^{2}dx=\mu\int_{\Omega}\rho u^{2}dx,

thus μ[0,+[\mu\in[0,+\infty[. Finally, if μ=0\mu=0, then Ω|Dmu|2𝑑x=0\int_{\Omega}|D^{m}u|^{2}dx=0, thus uu is a polynomial of degree at most m1m-1 in N\mathbb{R}^{N}. The eigenspace associated with the eigenvalue μ=0\mu=0 has dimension (N+m1N)\binom{N+m-1}{N} and coincides with the space of the polynomials of degree at most m1m-1 in N\mathbb{R}^{N}. This concludes the proof.

Appendix B A few useful functional inequalities

In this section we will prove some useful functional inequalities which are crucial in the proof of the results of Subsection 3.3, in particular of Theorem 3.29. Since we think that they are interesting on their own, we shall provide here all the details of the proofs. Through this section Ω\Omega will be a bounded domain in N\mathbb{R}^{N} with Lipschitz boundary. We start this section by recalling the standard Sobolev embeddings.

Theorem B.1.

Let Ω\Omega be a bounded domain with Lipschitz boundary. Let mm\in\mathbb{N} and assume that uHm(Ω)u\in H^{m}(\Omega).

  1. i)

    If N<2mN<2m then uCm[N2]1,γ(Ω)u\in C^{m-\left[\frac{N}{2}\right]-1,\gamma}(\Omega), where

    γ={[N2]+1N2,ifN2anynumberin]0,1[,ifN2.\gamma=\begin{cases}\left[\frac{N}{2}\right]+1-\frac{N}{2},&{\rm if\ }\frac{N}{2}\notin\mathbb{N}\\ {\rm any\ number\ in\ }]0,1[,&{\rm if\ }\frac{N}{2}\in\mathbb{N}.\end{cases}

    Moreover there exists a positive constant CC which depends only on m,Nm,N and Ω\Omega such that

    (B.2) uCm[N2]1,γ(Ω)CuHm(Ω).\|u\|_{C^{m-\left[\frac{N}{2}\right]-1,\gamma}(\Omega)}\leq C\|u\|_{H^{m}(\Omega)}.
  2. ii)

    If N>2mN>2m then uL2NN2m(Ω)u\in L^{\frac{2N}{N-2m}}(\Omega) and

    (B.3) uL2NN2m(Ω)CuHm(Ω)\|u\|_{L^{\frac{2N}{N-2m}}(\Omega)}\leq C\|u\|_{H^{m}(\Omega)}

    the constant CC depending only on m,Nm,N and Ω\Omega.

  3. iii)

    If N=2mN=2m there exist constants C1,C2>0C_{1},C_{2}>0 which depend only on mm and Ω\Omega such that

    (B.4) ΩeC1(u(x)uHm(Ω))2𝑑xC2.\int_{\Omega}e^{C_{1}\left(\frac{u(x)}{\|u\|_{H^{m}(\Omega)}}\right)^{2}}dx\leq C_{2}.

We refer to [10, § 4.6-4.7] and to [22, § 5.6.3] for the proof ot points i)i) and ii)ii) of Theorem B.1. We refer to [13, Theorem 1.1] for the proof of (B.4) (see also [10, § 4.7]).

From Theorem B.1 it follows that if N<2mN<2m then a function uHm(Ω)u\in H^{m}(\Omega) is (equivalent to) a function of class Cm[N2]1(Ω)C^{m-\left[\frac{N}{2}\right]-1}(\Omega). In particular if NN is odd, we can write N=2m2k1N=2m-2k-1 for some k{0,,m1}k\in\left\{0,...,m-1\right\} and a function uHm(Ω)u\in H^{m}(\Omega) is (equivalent to) a function of class Ck,12(Ω)C^{k,\frac{1}{2}}(\Omega). If NN is even, we can write N=2m2k2N=2m-2k-2 for some k{0,,m2}k\in\left\{0,...,m-2\right\} and a function uHm(Ω)u\in H^{m}(\Omega) is (equivalent to) a function of class Ck,γ(Ω)C^{k,\gamma}(\Omega) for any γ]0,1[\gamma\in]0,1[.

Assume now that a function uHm(Ω)u\in H^{m}(\Omega) has all its partial derivatives up to the kk-th order vanishing at a point x0Ωx_{0}\in\Omega. Then the integral of u2u^{2} over B(x0,ε)B(x_{0},\varepsilon) (where ε>0\varepsilon>0 is such that B(x0,ε)ΩB(x_{0},\varepsilon)\subset\subset\Omega) can be controlled by ε2muHm(Ω)2\varepsilon^{2m}\|u\|^{2}_{H^{m}(\Omega)} if N<2mN<2m is odd, and by ε2m(1+|log(ε)|)uHm(Ω)2\varepsilon^{2m}(1+|\log(\varepsilon)|)\|u\|^{2}_{H^{m}(\Omega)} if N<2mN<2m is even. This is proved in the following lemma, where without loss of generality we set x0=0x_{0}=0.

Lemma B.5.

Let Ω\Omega be a bounded domain in N\mathbb{R}^{N}, N<2mN<2m, with Lipschitz boundary. Assume that 0Ω0\in\Omega and let ε>0\varepsilon>0 be such that B(0,ε)ΩB(0,\varepsilon)\subset\subset\Omega. Let uHm(Ω)u\in H^{m}(\Omega). Then there exists a positive constant CC which depends only on m,km,k and Ω\Omega such that

  1. i)

    B(0,ε)|u(x)|α|kαu(0)α!xα|2𝑑xCε2muHm(Ω)2\int_{B(0,\varepsilon)}{\left|u(x)-\sum_{|\alpha|\leq k}\frac{\partial^{\alpha}u(0)}{\alpha!}x^{\alpha}\right|^{2}dx}\leq C\varepsilon^{2m}\|u\|_{H^{m}(\Omega)}^{2} if N=2m2k1N=2m-2k-1, k=0,,m1\forall\,k=0,...,m-1;

  2. ii)

    B(0,ε)|u(x)|α|kαu(0)α!xα|2𝑑xCε2m(1+|log(ε)|)uHm(Ω)2\int_{B(0,\varepsilon)}{\left|u(x)-\sum_{|\alpha|\leq k}\frac{\partial^{\alpha}u(0)}{\alpha!}x^{\alpha}\right|^{2}dx}\leq C\varepsilon^{2m}(1+|\log(\varepsilon)|)\|u\|_{H^{m}(\Omega)}^{2} if N=2m2k2N=2m-2k-2, k=0,,m2\forall\,k=0,...,m-2.

Proof.

We start by proving i)i). Let N=2m2k1N=2m-2k-1 for some k{0,,m1}k\in\left\{0,...,m-1\right\}. Actually, we will prove i)i) for a function uCk+1(Ω)Hm(Ω)u\in C^{k+1}(\Omega)\cap H^{m}(\Omega). The result for a function uHm(Ω)u\in H^{m}(\Omega) will follow from standard approximation of functions in the space Hm(Ω)H^{m}(\Omega) by smooth functions (see [10, § 2.3] and [22, § 5.3]). Let then uCk+1(Ω)Hm(Ω)u\in C^{k+1}(\Omega)\cap H^{m}(\Omega). Through the rest of the proof we shall denote by CC a positive constant which depends only on m,km,k and Ω\Omega and which we can eventually re-define line by line. From the standard Sobolev embedding theorem, it follows that uCk+1(Ω)Ck,12(Ω)u\in C^{k+1}(\Omega)\cap C^{k,\frac{1}{2}}(\Omega). From Taylor’s Theorem it follows that

(B.6) u(x)|α|kαu(0)α!xα=|β|=k+1|β|β!01(1t)kβu(tx)dtxβ.u(x)-\sum_{|\alpha|\leq k}\frac{\partial^{\alpha}u(0)}{\alpha!}x^{\alpha}=\sum_{|\beta|=k+1}\frac{|\beta|}{\beta!}\int_{0}^{1}(1-t)^{k}\partial^{\beta}u(tx)dtx^{\beta}.

We consider now the absolute value of the expression in the right-hand side of (B.6) and integrate over B(0,ε)B(0,\varepsilon) each integral which appears in the sum. We have

(B.7) B(0,ε)|01(1t)kβu(tx)xβdt|𝑑x01(1t)kB(0,ε)|βu(tx)||x|k+1𝑑x𝑑t=01t2m+k(1t)kB(0,tε)|βu(y)||y|k+1𝑑y𝑑t01t2m+k(1t)k(B(0,tε)|βu(y)|2(2m2k1)𝑑y)12(2m2k1)(B(0,tε)|y|2(2m2k1)(k+1)2(2m2k1)1𝑑y)2(2m2k1)12(2m2k1)dtC01t2m+k(1t)kβuL2(2m2k1)(Ω)ε2mk12t2mk12𝑑t=Cε2mk12βuL2(2m2k1)(Ω)Cε2mk12βuHmk1(Ω)Cε2mk12uHm(Ω),\int_{B(0,\varepsilon)}\left|\int_{0}^{1}(1-t)^{k}\partial^{\beta}u(tx)x^{\beta}dt\right|dx\leq\int_{0}^{1}(1-t)^{k}\int_{B(0,\varepsilon)}|\partial^{\beta}u(tx)||x|^{k+1}dxdt\\ =\int_{0}^{1}t^{-2m+k}(1-t)^{k}\int_{B(0,t\varepsilon)}|\partial^{\beta}u(y)||y|^{k+1}dydt\\ \leq\int_{0}^{1}t^{-2m+k}(1-t)^{k}\left(\int_{B(0,t\varepsilon)}|\partial^{\beta}u(y)|^{2(2m-2k-1)}dy\right)^{\frac{1}{2(2m-2k-1)}}\\ \cdot\left(\int_{B(0,t\varepsilon)}|y|^{\frac{2(2m-2k-1)(k+1)}{2(2m-2k-1)-1}}dy\right)^{\frac{2(2m-2k-1)-1}{2(2m-2k-1)}}dt\\ \leq C\int_{0}^{1}t^{-2m+k}(1-t)^{k}\|\partial^{\beta}u\|_{L^{2(2m-2k-1)}(\Omega)}\varepsilon^{2m-k-\frac{1}{2}}t^{2m-k-\frac{1}{2}}dt\\ =C\varepsilon^{2m-k-\frac{1}{2}}\|\partial^{\beta}u\|_{L^{2(2m-2k-1)}(\Omega)}\leq C\varepsilon^{2m-k-\frac{1}{2}}\|\partial^{\beta}u\|_{H^{m-k-1}(\Omega)}\\ \leq C\varepsilon^{2m-k-\frac{1}{2}}\|u\|_{H^{m}(\Omega)},

where in the last line we have used the Sobolev inequality (B.3) for functions in Hmk1(Ω)H^{m-k-1}(\Omega) with Ω2m2k1\Omega\in\mathbb{R}^{2m-2k-1}.

Next, we estimate the quantity

u(x)|α|kαu(0)α!xαC0(B(0,ε)):=maxxB(0,ε)|u(x)|α|kαu(0)α!xα|\left\|u(x)-\sum_{|\alpha|\leq k}\frac{\partial^{\alpha}u(0)}{\alpha!}x^{\alpha}\right\|_{C^{0}(B(0,\varepsilon))}:=\max_{x\in B(0,\varepsilon)}\left|u(x)-\sum_{|\alpha|\leq k}\frac{\partial^{\alpha}u(0)}{\alpha!}x^{\alpha}\right|

for a function uCk,12(Ω)u\in C^{k,\frac{1}{2}}(\Omega). First we note that there exits t]0,1[t\in]0,1[ such that

u(x)|α|k1αu(0)α!xα=|α|=kαu(tx)xαα!,u(x)-\sum_{|\alpha|\leq{k-1}}\frac{\partial^{\alpha}u(0)}{\alpha!}x^{\alpha}=\sum_{|\alpha|=k}\partial^{\alpha}u(tx)\frac{x^{\alpha}}{\alpha!},

then

|u(x)|α|kαu(0)α!xα|=||α|=kxαα!(αu(0)αu(tx)||x|k|α|=k1α!|αu(0)αu(tx)|,\left|u(x)-\sum_{|\alpha|\leq k}\frac{\partial^{\alpha}u(0)}{\alpha!}x^{\alpha}\right|=\left|\sum_{|\alpha|=k}\frac{x^{\alpha}}{\alpha!}(\partial^{\alpha}u(0)-\partial^{\alpha}u(tx)\right|\\ \leq|x|^{k}\sum_{|\alpha|=k}\frac{1}{\alpha!}|\partial^{\alpha}u(0)-\partial^{\alpha}u(tx)|,

which implies

(B.8) u(x)|α|kαu(0)α!xαC0(B(0,ε))Cεk+12uHm(Ω),\left\|u(x)-\sum_{|\alpha|\leq k}\frac{\partial^{\alpha}u(0)}{\alpha!}x^{\alpha}\right\|_{C^{0}(B(0,\varepsilon))}\leq C\varepsilon^{k+\frac{1}{2}}\|u\|_{H^{m}(\Omega)},

where the constant CC depends only on m,km,k and Ω\Omega (see also (B.2)).

Consider now (B.6). We take the squares of both sides and integrate over B(0,ε)B(0,\varepsilon). We have

(B.9) B(0,ε)|u(x)|α|kαu(0)α!xα|2𝑑xu(x)|α|kαu(0)α!xαC0(B(0,ε))B(0,ε)|u(x)|α|kαu(0)α!xα|𝑑xCε2muHm(Ω)2,\int_{B(0,\varepsilon)}\left|u(x)-\sum_{|\alpha|\leq k}\frac{\partial^{\alpha}u(0)}{\alpha!}x^{\alpha}\right|^{2}dx\\ \leq\left\|u(x)-\sum_{|\alpha|\leq k}\frac{\partial^{\alpha}u(0)}{\alpha!}x^{\alpha}\right\|_{C^{0}(B(0,\varepsilon))}\int_{B(0,\varepsilon)}\left|u(x)-\sum_{|\alpha|\leq k}\frac{\partial^{\alpha}u(0)}{\alpha!}x^{\alpha}\right|dx\\ \leq C\varepsilon^{2m}\|u\|_{H^{m}(\Omega)}^{2},

where the last inequality follows from (B.7) and (B.8) and the constant CC depends only on m,km,k and Ω\Omega. Since inequality (B.9) holds for all uCk+1(Ω)Hm(Ω)u\in C^{k+1}(\Omega)\cap H^{m}(\Omega), by standard approximation of Hm(Ω)H^{m}(\Omega) functions by smooth functions, we conclude that it holds for all uHm(Ω)u\in H^{m}(\Omega). This proves i)i).

Consider now ii)ii). Let N=2m2k2N=2m-2k-2 for some k{0,,m2}k\in\left\{0,...,m-2\right\}. Again, we shall prove ii)ii) for a function uCk+1(Ω)Hm(Ω)u\in C^{k+1}(\Omega)\cap H^{m}(\Omega). The result for a function uHm(Ω)u\in H^{m}(\Omega) will follows from standard approximation of functions in the space Hm(Ω)H^{m}(\Omega) by smooth functions.

We prove first the following inequality:

(B.10) fL2(B(0,ε))2Cε2m2k2(1+|log(ε)|)fHmk1(Ω)2,\|f\|_{L^{2}(B(0,\varepsilon))}^{2}\leq C\varepsilon^{2m-2k-2}(1+|\log(\varepsilon)|)\|f\|_{H^{m-k-1}(\Omega)}^{2},

for all fHmk1(Ω)f\in H^{m-k-1}(\Omega) (the constant C>0C>0 depending only on m,km,k and Ω\Omega). In order to prove (B.10) we will use the exponential inequality (B.4) which describes the limiting behavior of the Sobolev inequality (B.3) when N=2m2k2N=2m-2k-2 for functions in Hmk1(Ω)H^{m-k-1}(\Omega). Let then fHmk1(Ω)f\in H^{m-k-1}(\Omega) and let C1,C2C_{1},C_{2} be the constants appearing in (B.4). We have

B(0,ε)f(x)2𝑑x=|B(0,ε)|C1fHmk1(Ω)2B(0,ε)C1(f(x)fHmk1(Ω))2dx|B(0,ε)|=|B(0,ε)|C1fHmk1(Ω)2B(0,ε)log(eC1(f(x)fHmk1(Ω))2)dx|B(0,ε)||B(0,ε)|C1fHmk1(Ω)2log(B(0,ε)eC1(f(x)fHmk1(Ω))2dx|B(0,ε)|)|B(0,ε)|C1fHmk1(Ω)2log(1|B(0,ε)|ΩeC1(f(x)fHmk1(Ω))2𝑑x)|B(0,ε)|C1fHmk1(Ω)2log(C2|B(0,ε)|)=|B(0,ε)|C1fHmk1(Ω)2(log(C2)log(|B(0,ε)|))Cε2m2k2(1+|log(ε)|)fHmk1(Ω)2,\int_{B(0,\varepsilon)}f(x)^{2}dx=\frac{|B(0,\varepsilon)|}{C_{1}}\|f\|_{H^{m-k-1}(\Omega)}^{2}\int_{B(0,\varepsilon)}C_{1}\left(\frac{f(x)}{\|f\|_{H^{m-k-1}(\Omega)}}\right)^{2}\frac{dx}{|B(0,\varepsilon)|}\\ =\frac{|B(0,\varepsilon)|}{C_{1}}\|f\|_{H^{m-k-1}(\Omega)}^{2}\int_{B(0,\varepsilon)}\log\left(e^{C_{1}\left(\frac{f(x)}{\|f\|_{H^{m-k-1}(\Omega)}}\right)^{2}}\right)\frac{dx}{|B(0,\varepsilon)|}\\ \leq\frac{|B(0,\varepsilon)|}{C_{1}}\|f\|_{H^{m-k-1}(\Omega)}^{2}\log\left(\int_{B(0,\varepsilon)}e^{C_{1}\left(\frac{f(x)}{\|f\|_{H^{m-k-1}(\Omega)}}\right)^{2}}\frac{dx}{|B(0,\varepsilon)|}\right)\\ \leq\frac{|B(0,\varepsilon)|}{C_{1}}\|f\|_{H^{m-k-1}(\Omega)}^{2}\log\left(\frac{1}{|B(0,\varepsilon)|}\int_{\Omega}e^{C_{1}\left(\frac{f(x)}{\|f\|_{H^{m-k-1}(\Omega)}}\right)^{2}}dx\right)\\ \leq\frac{|B(0,\varepsilon)|}{C_{1}}\|f\|_{H^{m-k-1}(\Omega)}^{2}\log\left(\frac{C_{2}}{|B(0,\varepsilon)|}\right)\\ =\frac{|B(0,\varepsilon)|}{C_{1}}\|f\|_{H^{m-k-1}(\Omega)}^{2}\left(\log(C_{2})-\log(|B(0,\varepsilon)|)\right)\\ \leq C\varepsilon^{2m-2k-2}(1+|\log(\varepsilon)|)\|f\|_{H^{m-k-1}(\Omega)}^{2},

where in the first inequality we have used the concavity of the logarithm and Jensen’s inequality and in the third inequality we have applied (B.4). Inequality (B.10) is now proved.

Let now uCk+1(Ω)Hm(Ω)u\in C^{k+1}(\Omega)\cap H^{m}(\Omega). From the Sobolev inequality (B.2), it follows that uCk+1(Ω)Ck,γ(Ω)u\in C^{k+1}(\Omega)\cap C^{k,\gamma}(\Omega) for all γ]0,1[\gamma\in]0,1[. From Taylor’s Theorem (see also (B.6)) it follows it follows then

(B.11) B(0,ε)|u(x)|α|kαu(0)α!xα|2𝑑x|β|=k+1(k+1)!(β!)2B(0,ε)01(1t)2k|βu(tx)|2|x|2(k+1)𝑑t𝑑x.\int_{B(0,\varepsilon)}\left|u(x)-\sum_{|\alpha|\leq k}\frac{\partial^{\alpha}u(0)}{\alpha!}x^{\alpha}\right|^{2}dx\\ \leq\sum_{|\beta|=k+1}\frac{(k+1)!}{(\beta!)^{2}}\int_{B(0,\varepsilon)}\int_{0}^{1}(1-t)^{2k}|\partial^{\beta}u(tx)|^{2}|x|^{2(k+1)}dtdx.

We estimate the integrals appearing in the right-hand side of (B.11)

(B.12) B(0,ε)01(1t)2k|βu(tx)|2|x|2(k+1)𝑑t𝑑x=01(1t)2kB(0,ε)|βu(tx)|2|x|2(k+1)𝑑x𝑑t=01t2m(1t)2kB(0,tε)|βu(y)|2|y|2(k+1)𝑑x𝑑t01t2m+2k+2ε2k+2(1t)2kβuL2(B(0,tε))2𝑑tCε2m(1+|log(ε)|)uHm(Ω)2,\int_{B(0,\varepsilon)}\int_{0}^{1}(1-t)^{2k}|\partial^{\beta}u(tx)|^{2}|x|^{2(k+1)}dtdx\\ =\int_{0}^{1}(1-t)^{2k}\int_{B(0,\varepsilon)}|\partial^{\beta}u(tx)|^{2}|x|^{2(k+1)}dxdt\\ =\int_{0}^{1}t^{-2m}(1-t)^{2k}\int_{B(0,t\varepsilon)}|\partial^{\beta}u(y)|^{2}|y|^{2(k+1)}dxdt\\ \leq\int_{0}^{1}t^{-2m+2k+2}\varepsilon^{2k+2}(1-t)^{2k}\|\partial^{\beta}u\|_{L^{2}(B(0,t\varepsilon))}^{2}dt\\ \leq C\varepsilon^{2m}(1+|\log(\varepsilon)|)\|u\|_{H^{m}(\Omega)}^{2},

where the last inequality follows from (B.10) applied with f=βuf=\partial^{\beta}u. From (B.11) and (B.12), ii)ii) immediately follows. This ends the proof of the lemma. ∎

Assume now that uHm(Ω)u\in H^{m}(\Omega) is such that Ωρ~εu(x)xα𝑑x=0\int_{\Omega}\tilde{\rho}_{\varepsilon}u(x)x^{\alpha}dx=0 for all |α|m1|\alpha|\leq m-1, that is, for a fixed δ]0,1/2[\delta\in]0,1/2[

Ωu(x)xα𝑑x+ε2m+δB(0,ε)u(x)xα𝑑x=0\int_{\Omega}u(x)x^{\alpha}dx+\varepsilon^{-2m+\delta}\int_{B(0,\varepsilon)}u(x)x^{\alpha}dx=0

for all |α|m1|\alpha|\leq m-1. In view of Lemma B.5 we expect that the quantities |αu(0)||\partial^{\alpha}u(0)|, (B(0,ε)u2𝑑x)12\left(\int_{B(0,\varepsilon)}u^{2}dx\right)^{\frac{1}{2}} and Ωu(x)xα𝑑x\int_{\Omega}u(x)x^{\alpha}dx can be bounded by the uHm(Ω)\|u\|_{H^{m}(\Omega)} and a suitable power of ε\varepsilon. In particular we expect that all these quantities vanish as ε0+\varepsilon\rightarrow 0^{+}. The aim of the next lemma is to prove that this is exactly what happens. We shall also provide the correct powers of ε\varepsilon in the estimates which are crucial in the proof of Theorem 3.29.

Lemma B.13.

Let Ω\Omega be a bounded domain in N\mathbb{R}^{N}, N<2mN<2m, with Lipschitz boundary. Assume that 0Ω0\in\Omega and let ε>0\varepsilon>0 be such that B(0,ε)ΩB(0,\varepsilon)\subset\subset\Omega. Let δ]0,1/2[\delta\in]0,1/2[ be fixed. Let uHm(Ω)u\in H^{m}(\Omega) be such that Ωρ~εu(x)xβ𝑑x=0\int_{\Omega}\tilde{\rho}_{\varepsilon}u(x)x^{\beta}dx=0 for all βN\beta\in\mathbb{N}^{N} with |β|m1|\beta|\leq m-1. Then there exists a positive constant CC which depends only on m,km,k and Ω\Omega such that

  1. i)

    |αu(0)|Cεk+12|α|uHm(Ω)|\partial^{\alpha}u(0)|\leq C\varepsilon^{k+\frac{1}{2}-|\alpha|}\|u\|_{H^{m}(\Omega)} for all αN\alpha\in\mathbb{N}^{N} with |α|k|\alpha|\leq k, if N=2m2k1N=2m-2k-1, k=0,,m1\forall\,k=0,...,m-1;

  2. ii)

    |αu(0)|Cεk+1|α|(1+|log(ε)|)12uHm(Ω)|\partial^{\alpha}u(0)|\leq C\varepsilon^{k+1-|\alpha|}(1+|\log(\varepsilon)|)^{\frac{1}{2}}\|u\|_{H^{m}(\Omega)} for all αN\alpha\in\mathbb{N}^{N} with |α|k|\alpha|\leq k, if N=2m2k2N=2m-2k-2, k=0,,m2\forall\,k=0,...,m-2;

  3. iii)

    B(0,ε)u2𝑑xCε2muHm(Ω)2\int_{B(0,\varepsilon)}u^{2}dx\leq C\varepsilon^{2m}\|u\|_{H^{m}(\Omega)}^{2} if N=2m2k1N=2m-2k-1, k=1,,m1\forall\,k=1,...,m-1;

  4. iv)

    B(0,ε)u2𝑑xCε2m(1+|log(ε)|)uHm(Ω)2\int_{B(0,\varepsilon)}u^{2}dx\leq C\varepsilon^{2m}(1+|\log(\varepsilon)|)\|u\|_{H^{m}(\Omega)}^{2} if N=2m2k2N=2m-2k-2, k=1,,m2\forall\,k=1,...,m-2;

  5. v)

    |Ωuxα𝑑x|Cε|α|+δk12uHm(Ω)\left|\int_{\Omega}ux^{\alpha}dx\right|\leq C\varepsilon^{|\alpha|+\delta-k-\frac{1}{2}}\|u\|_{H^{m}(\Omega)} for all αN\alpha\in\mathbb{N}^{N} with k+1|α|m1k+1\leq|\alpha|\leq m-1, if N=2m2k1N=2m-2k-1, k=0,,m2\forall\,k=0,...,m-2;

  6. vi)

    |Ωuxα𝑑x|Cε|α|+δk1(1+|log(ε)|)12uHm(Ω)\left|\int_{\Omega}ux^{\alpha}dx\right|\leq C\varepsilon^{|\alpha|+\delta-k-1}(1+|\log(\varepsilon)|)^{\frac{1}{2}}\|u\|_{H^{m}(\Omega)} for all αN\alpha\in\mathbb{N}^{N} with k+1|α|m1k+1\leq|\alpha|\leq m-1, if N=2m2k2N=2m-2k-2, k=0,,m2\forall\,k=0,...,m-2.

Proof.

Let uHm(Ω)u\in H^{m}(\Omega) be such that Ωρ~εu(x)xβ𝑑x=0\int_{\Omega}\tilde{\rho}_{\varepsilon}u(x)x^{\beta}dx=0 for all βN\beta\in\mathbb{N}^{N} with |β|m1|\beta|\leq m-1. We start by proving i)i) and ii)ii). We recall that by Sobolev inequality (B.2), uCk,12(Ω)u\in C^{k,\frac{1}{2}}(\Omega) if N=2m2k1N=2m-2k-1, while uCk,γ(Ω)u\in C^{k,\gamma}(\Omega) for all γ]0,1[\gamma\in]0,1[ if N=2m2k2N=2m-2k-2. We have, for |β|m1|\beta|\leq m-1, that

(B.14) B(0,ε)ρ~εu(x)xβ=B(0,ε)ρ~ε(u(x)|α|kαu(0)α!xα)xβ𝑑x+|α|kαu(0)α!B(0,ε)ρ~εxαxβ𝑑x.\int_{B(0,\varepsilon)}\tilde{\rho}_{\varepsilon}u(x)x^{\beta}\\ =\int_{B(0,\varepsilon)}\tilde{\rho}_{\varepsilon}\left(u(x)-\sum_{|\alpha|\leq k}\frac{\partial^{\alpha}u(0)}{\alpha!}x^{\alpha}\right)x^{\beta}dx+\sum_{|\alpha|\leq k}\frac{\partial^{\alpha}u(0)}{\alpha!}\int_{B(0,\varepsilon)}\tilde{\rho}_{\varepsilon}x^{\alpha}x^{\beta}dx.

From (B.14) and from the fact that Ωρ~εu(x)xβ𝑑x=0\int_{\Omega}\tilde{\rho}_{\varepsilon}u(x)x^{\beta}dx=0 it follows that

(B.15) |α|kαu(0)α!ε2m+δB(0,ε)xαxβ𝑑x=Ωu(x)xβ𝑑xε2m+δB(0,ε)(u(x)|α|kαu(0)α!xα)xβ𝑑x.\sum_{|\alpha|\leq k}\frac{\partial^{\alpha}u(0)}{\alpha!}\varepsilon^{-2m+\delta}\int_{B(0,\varepsilon)}x^{\alpha}x^{\beta}dx\\ =-\int_{\Omega}u(x)x^{\beta}dx-\varepsilon^{-2m+\delta}\int_{B(0,\varepsilon)}\left(u(x)-\sum_{|\alpha|\leq k}\frac{\partial^{\alpha}u(0)}{\alpha!}x^{\alpha}\right)x^{\beta}dx.

We compute now B(0,ε)ε2m+δxαxβ𝑑x\int_{B(0,\varepsilon)}\varepsilon^{-2m+\delta}x^{\alpha}x^{\beta}dx. It is convenient to pass to the spherical coordinates (r,θ)=(r,θ1,θN1)[0,+[×B(r,\theta)=(r,\theta_{1},...\theta_{N-1})\in[0,+\infty[\times\partial B, where B\partial B denotes the unit sphere in N\mathbb{R}^{N} endowed with the N1N-1 dimensional volume element dσ(θ)d\sigma(\theta). With respect to these new variables, we write the coordinate functions xix_{i} as xi=rHi(θ)x_{i}=rH_{i}(\theta), where Hi(θ)H_{i}(\theta) are the standard spherical harmonics of degree 11 in N\mathbb{R}^{N}. We have then

B(0,ε)ε2m+δxαxβ𝑑x=B0εH1α1+β1HNαN+βNr|α|+|β|rN1𝑑r𝑑σ(θ)=Kα,βεN+|α|+|β|2m+δ,\int_{B(0,\varepsilon)}\varepsilon^{-2m+\delta}x^{\alpha}x^{\beta}dx\\ =\int_{\partial B}\int_{0}^{\varepsilon}H_{1}^{\alpha_{1}+\beta_{1}}\cdots H_{N}^{\alpha_{N}+\beta_{N}}r^{|\alpha|+|\beta|}r^{N-1}drd\sigma(\theta)=K_{\alpha,\beta}\varepsilon^{N+|\alpha|+|\beta|-2m+\delta},

where

(B.16) Kα,β=1N+|α|+|β|BH1α1+β1HNαN+βN𝑑σ(θ).K_{\alpha,\beta}=\frac{1}{N+|\alpha|+|\beta|}\int_{\partial B}H_{1}^{\alpha_{1}+\beta_{1}}\cdots H_{N}^{\alpha_{N}+\beta_{N}}d\sigma(\theta).

From (B.15) it follows that for all |β|k|\beta|\leq k

(B.17) |α|kKα,βα!ε|α|αu(0)=ε2mN|β|δΩu(x)xβ𝑑xεN|β|B(0,ε)(u(x)|α|kαu(0)α!xα)xβ𝑑x.\sum_{|\alpha|\leq k}\frac{K_{\alpha,\beta}}{\alpha!}\varepsilon^{|\alpha|}\partial^{\alpha}u(0)\\ =-\varepsilon^{2m-N-|\beta|-\delta}\int_{\Omega}u(x)x^{\beta}dx-\varepsilon^{-N-|\beta|}\int_{B(0,\varepsilon)}\left(u(x)-\sum_{|\alpha|\leq k}\frac{\partial^{\alpha}u(0)}{\alpha!}x^{\alpha}\right)x^{\beta}dx.

Clearly

(B.18) |Ωu(x)xβ𝑑x|CΩ,βuL2(Ω),\left|\int_{\Omega}u(x)x^{\beta}dx\right|\leq C_{\Omega,\beta}\|u\|_{L^{2}(\Omega)},

where CΩ,βC_{\Omega,\beta} depends only on Ω\Omega and |β||\beta|. Moreover, from Lemma B.5 and Hölder’s inequality it follows that

(B.19) |B(0,ε)(u(x)|α|kαu(0)α!xα)xβ𝑑x|Cε|β|+2mk12uHm(Ω),\left|\int_{B(0,\varepsilon)}\left(u(x)-\sum_{|\alpha|\leq k}\frac{\partial^{\alpha}u(0)}{\alpha!}x^{\alpha}\right)x^{\beta}dx\right|\leq C\varepsilon^{|\beta|+2m-k-\frac{1}{2}}\|u\|_{H^{m}(\Omega)},

if N=2m2k1N=2m-2k-1, while

(B.20) |B(0,ε)(u(x)|α|kαu(0)α!xα)xβ𝑑x|Cε|β|+2mk1(1+|log(ε)|)12uHm(Ω),\left|\int_{B(0,\varepsilon)}\left(u(x)-\sum_{|\alpha|\leq k}\frac{\partial^{\alpha}u(0)}{\alpha!}x^{\alpha}\right)x^{\beta}dx\right|\leq C\varepsilon^{|\beta|+2m-k-1}(1+|\log(\varepsilon)|)^{\frac{1}{2}}\|u\|_{H^{m}(\Omega)},

if N=2m2k2N=2m-2k-2. From (B.17), (B.18), (B.19), (B.20) and since δ]0,1/2[\delta\in]0,1/2[, we deduce that

|α|kKα,βα!ε|α|αu(0)C(ε2k+1|β|δ+εk+12)uHm(Ω)2Cεk+12uHm(Ω),\sum_{|\alpha|\leq k}\frac{K_{\alpha,\beta}}{\alpha!}\varepsilon^{|\alpha|}\partial^{\alpha}u(0)\leq C(\varepsilon^{2k+1-|\beta|-\delta}+\varepsilon^{k+\frac{1}{2}})\|u\|_{H^{m}(\Omega)}\\ \leq 2C\varepsilon^{k+\frac{1}{2}}\|u\|_{H^{m}(\Omega)},

if N=2m2k1N=2m-2k-1, while

|α|kKα,βα!ε|α|αu(0)C(ε2k+2|β|δ+εk+1(1+|log(ε)|12))uHm(Ω)2Cεk+1(1+|log(ε)|12)uHm(Ω),\sum_{|\alpha|\leq k}\frac{K_{\alpha,\beta}}{\alpha!}\varepsilon^{|\alpha|}\partial^{\alpha}u(0)\leq C(\varepsilon^{2k+2-|\beta|-\delta}+\varepsilon^{k+1}(1+|\log(\varepsilon)|^{\frac{1}{2}}))\|u\|_{H^{m}(\Omega)}\\ \leq 2C\varepsilon^{k+1}(1+|\log(\varepsilon)|^{\frac{1}{2}})\|u\|_{H^{m}(\Omega)},

if N=2m2k2N=2m-2k-2. Since for all |α|k|\alpha|\leq k, Kα,α>0K_{\alpha,\alpha}>0 by definition (see (B.16)), necessairily inequalities i)i) and ii)ii) must hold with a constant C>0C>0 which depends only on m,km,k and Ω\Omega.

Consider now iii)iii). We have

B(0,ε)u2𝑑x=B(0,ε)(u(x)|α|kαu(0)α!xα+|α|kαu(0)α!xα)2𝑑x2B(0,ε)(u(x)|α|kαu(0)α!xα)2𝑑x+2(N+kk)|α|k|αu(0)|2(α!)2B(0,ε)|x|2α𝑑xCε2muHm(Ω)2+C|α|kε2k+12αB(0,ε)|x|2α𝑑xCε2muHm(Ω)2,\int_{B(0,\varepsilon)}u^{2}dx=\int_{B(0,\varepsilon)}\left(u(x)-\sum_{|\alpha|\leq k}\frac{\partial^{\alpha}u(0)}{\alpha!}x^{\alpha}+\sum_{|\alpha|\leq k}\frac{\partial^{\alpha}u(0)}{\alpha!}x^{\alpha}\right)^{2}dx\\ \leq 2\int_{B(0,\varepsilon)}\left(u(x)-\sum_{|\alpha|\leq k}\frac{\partial^{\alpha}u(0)}{\alpha!}x^{\alpha}\right)^{2}dx\\ +2\binom{N+k}{k}\sum_{|\alpha|\leq k}\frac{|\partial^{\alpha}u(0)|^{2}}{(\alpha!)^{2}}\int_{B(0,\varepsilon)}|x|^{2\alpha}dx\\ \leq C\varepsilon^{2m}\|u\|_{H^{m}(\Omega)}^{2}+C\sum_{|\alpha|\leq k}\varepsilon^{2k+1-2\alpha}\int_{B(0,\varepsilon)}|x|^{2\alpha}dx\\ \leq C\varepsilon^{2m}\|u\|_{H^{m}(\Omega)}^{2},

where in the second inequality we have used point i)i) of Lemma B.5 and in the last inequality we have used point i)i) of the present lemma and the fact that

|B(0,ε)x2α𝑑x|ε2|α||B(0,ε)|=ω2m2k1ε2|α|+2m2k1.\left|\int_{B(0,\varepsilon)}x^{2\alpha}dx\right|\leq\varepsilon^{2|\alpha|}|B(0,\varepsilon)|=\omega_{2m-2k-1}\varepsilon^{2|\alpha|+2m-2k-1}.

This concludes the proof of iii)iii). Point iv)iv) is proved exactly as point iii)iii), by using point ii)ii) of Lemma B.5 and point ii)ii) of the present lemma.

We consider now point v)v). Let N=2m2k1N=2m-2k-1, with 0km20\leq k\leq m-2 and let βN\beta\in\mathbb{N}^{N} such that k+1|β|m1k+1\leq|\beta|\leq m-1. We have

(B.21) Ωu(x)xβ𝑑x=ε2m+δB(0,ε)u(x)xβ𝑑x=ε2m+δB(0,ε)(u(x)|α|kαu(0)α!xα)xβ𝑑xε2m+δ|α|kαu(0)α!B(0,ε)xαxβ𝑑x.\int_{\Omega}u(x)x^{\beta}dx=-\varepsilon^{-2m+\delta}\int_{B(0,\varepsilon)}u(x)x^{\beta}dx\\ =-\varepsilon^{-2m+\delta}\int_{B(0,\varepsilon)}\left(u(x)-\sum_{|\alpha|\leq k}\frac{\partial^{\alpha}u(0)}{\alpha!}x^{\alpha}\right)x^{\beta}dx\\ -\varepsilon^{-2m+\delta}\sum_{|\alpha|\leq k}\frac{\partial^{\alpha}u(0)}{\alpha!}\int_{B(0,\varepsilon)}x^{\alpha}x^{\beta}dx.

From Lemma B.5 point i)i) and from Hölder’s inequality, we have that

(B.22) ε2m+δB(0,ε)|(u(x)|α|kαu(0)α!xα)xβ|𝑑xε2m+δ(B(0,ε)(u(x)|α|kαu(0)α!xα)2𝑑x)12(B(0,ε)|x|2|β|𝑑x)12Cε|β|+δk12uHm(Ω).\varepsilon^{-2m+\delta}\int_{B(0,\varepsilon)}\left|\left(u(x)-\sum_{|\alpha|\leq k}\frac{\partial^{\alpha}u(0)}{\alpha!}x^{\alpha}\right)x^{\beta}\right|dx\\ \leq\varepsilon^{-2m+\delta}\left(\int_{B(0,\varepsilon)}(u(x)-\sum_{|\alpha|\leq k}\frac{\partial^{\alpha}u(0)}{\alpha!}x^{\alpha})^{2}dx\right)^{\frac{1}{2}}\left(\int_{B(0,\varepsilon)}|x|^{2|\beta|}dx\right)^{\frac{1}{2}}\\ \leq C\varepsilon^{|\beta|+\delta-k-\frac{1}{2}}\|u\|_{H^{m}(\Omega)}.

Moreover, from point i)i) of the present lemma we have that for all |α|k|\alpha|\leq k

(B.23) ε2m+δ|αu(0)|α!B(0,ε)|x||α|+|β|𝑑xCε|β|+δk12.\varepsilon^{-2m+\delta}\frac{|\partial^{\alpha}u(0)|}{\alpha!}\int_{B(0,\varepsilon)}|x|^{|\alpha|+|\beta|}dx\leq C\varepsilon^{|\beta|+\delta-k-\frac{1}{2}}.

The proof of v)v) follows by combining (B.21) with (B.22) and (B.23). The proof of vi)vi) is identical to that of v)v) and follows form point ii)ii) of Lemma B.5 and point ii)ii) of the present lemma. This concludes the proof.

References

  • [1] M. J. Arrieta and P. D. Lamberti. Higher order elliptic operators on variable domains. stability results and boundary oscillations for intermediate problems. Online preprint ArXiv:1502.04373v2, 2015.
  • [2] D. Banks. Bounds for the eigenvalues of some vibrating systems. Pacific J. Math., 10:439–474, 1960.
  • [3] D. Banks. Upper bounds for the eigenvalues of some vibrating systems. Pacific J. Math., 11:1183–1203, 1961.
  • [4] D. O. Banks. Bounds for the eigenvalues of nonhomogeneous hinged vibrating rods. J. Math. Mech., 16:949–966, 1967.
  • [5] D. O. Banks. Lower bounds for the eigenvalues of a vibrating string whose density satisfies a Lipschitz condition. Pacific J. Math., 20:393–410, 1967.
  • [6] D. C. Barnes and D. O. Banks. Asymptotic bounds for the eigenvalues of vibrating systems. J. Differential Equations, 7:497–508, 1970.
  • [7] P. R. Beesack. Isoperimetric inequalities for the nonhomogeneous clamped rod and plate. J. Math. Mech., 8:471–482, 1959.
  • [8] H. Brezis. Analyse fonctionnelle. Collection Mathématiques Appliquées pour la Maîtrise. [Collection of Applied Mathematics for the Master’s Degree]. Masson, Paris, 1983. Théorie et applications. [Theory and applications].
  • [9] D. Buoso and L. Provenzano. A few shape optimization results for a biharmonic Steklov problem. J. Differential Equations, 259(5):1778–1818, 2015.
  • [10] V. I. Burenkov. Sobolev spaces on domains, volume 137 of Teubner-Texte zur Mathematik [Teubner Texts in Mathematics]. B. G. Teubner Verlagsgesellschaft mbH, Stuttgart, 1998.
  • [11] S. Chanillo, D. Grieser, M. Imai, K. Kurata, and I. Ohnishi. Symmetry breaking and other phenomena in the optimization of eigenvalues for composite membranes. Comm. Math. Phys., 214(2):315–337, 2000.
  • [12] L. M. Chasman. An isoperimetric inequality for fundamental tones of free plates. Comm. Math. Phys., 303(2):421–449, 2011.
  • [13] A. Cianchi. Moser-Trudinger inequalities without boundary conditions and isoperimetric problems. Indiana Univ. Math. J., 54(3):669–705, 2005.
  • [14] B. Colbois and A. El Soufi. Spectrum of the laplacian with weights. Online preprint ArXiv:1606.04095 [math.DG], 2016.
  • [15] B. Colbois, A. El Soufi, and A. Savo. Eigenvalues of the Laplacian on a compact manifold with density. Comm. Anal. Geom., 23(3):639–670, 2015.
  • [16] R. Courant and D. Hilbert. Methods of mathematical physics. Vol. I. Interscience Publishers, Inc., New York, N.Y., 1953.
  • [17] S. J. Cox. Extremal eigenvalue problems for the Laplacian. In Recent advances in partial differential equations (El Escorial, 1992), volume 30 of RAM Res. Appl. Math., pages 37–53. Masson, Paris, 1994.
  • [18] S. J. Cox and J. R. McLaughlin. Extremal eigenvalue problems for composite membranes. I,. Appl. Math. Optim., 22(2):153–167, 1990.
  • [19] S. J. Cox and J. R. McLaughlin. Extremal eigenvalue problems for composite membranes, II. Appl. Math. Optim., 22(1):169–187, 1990.
  • [20] M. C. Delfour and J.-P. Zolésio. Shapes and geometries, volume 22 of Advances in Design and Control. Society for Industrial and Applied Mathematics (SIAM), Philadelphia, PA, second edition, 2011. Metrics, analysis, differential calculus, and optimization.
  • [21] Y. Egorov and V. Kondratiev. On spectral theory of elliptic operators, volume 89 of Operator Theory: Advances and Applications. Birkhäuser Verlag, Basel, 1996.
  • [22] L. C. Evans. Partial differential equations, volume 19 of Graduate Studies in Mathematics. American Mathematical Society, Providence, RI, second edition, 2010.
  • [23] J. Fleckinger and M. L. Lapidus. Eigenvalues of elliptic boundary value problems with an indefinite weight function. Trans. Amer. Math. Soc., 295(1):305–324, 1986.
  • [24] S. Friedland. Extremal eigenvalue problems defined for certain classes of functions. Arch. Rational Mech. Anal., 67(1):73–81, 1977.
  • [25] F. Gazzola, H.-C. Grunau, and G. Sweers. Polyharmonic boundary value problems, volume 1991 of Lecture Notes in Mathematics. Springer-Verlag, Berlin, 2010. Positivity preserving and nonlinear higher order elliptic equations in bounded domains.
  • [26] R. D. Gentry and D. O. Banks. Bounds for functions of eigenvalues of vibrating systems. J. Math. Anal. Appl., 51:100–128, 1975.
  • [27] A. Grigor’yan, Y. Netrusov, and S.-T. Yau. Eigenvalues of elliptic operators and geometric applications. In Surveys in differential geometry. Vol. IX, volume 9 of Surv. Differ. Geom., pages 147–217. Int. Press, Somerville, MA, 2004.
  • [28] A. Henrot. Extremum problems for eigenvalues of elliptic operators. Frontiers in Mathematics. Birkhäuser Verlag, Basel, 2006.
  • [29] N. Korevaar. Upper bounds for eigenvalues of conformal metrics. J. Differential Geom., 37(1):73–93, 1993.
  • [30] S. G. Krantz and H. R. Parks. Distance to CkC^{k} hypersurfaces. J. Differential Equations, 40(1):116–120, 1981.
  • [31] M. G. Krein. On certain problems on the maximum and minimum of characteristic values and on the Lyapunov zones of stability. Amer. Math. Soc. Transl. (2), 1:163–187, 1955.
  • [32] P. Lamberti and L. Provenzano. A maximum principle in spectral optimization problems for elliptic operators subject to mass density perturbations. Eurasian Math. J., 4(3):70–83, 2013.
  • [33] P. Lamberti and L. Provenzano. Viewing the Steklov eigenvalues of the Laplace operator as critical Neumann eigenvalues. In V. V. Mityushev and M. V. Ruzhansky, editors, Current Trends in Analysis and Its Applications, Trends in Mathematics, pages 171–178. Springer International Publishing, 2015.
  • [34] P. D. Lamberti and L. Provenzano. Neumann to Steklov eigenvalues: asymptotic and monotonicity results. Proc. Roy. Soc. Edinburgh Sect. A, 147(2):429–447, 2017.
  • [35] B. Schwarz. Some results on the frequencies of nonhomogeneous rods. J. Math. Anal. Appl., 5:169–175, 1962.