Abstract
This paper studies eigenvalues of the buckling problem of arbitrary order on bounded domains in Euclidean spaces and spheres. We prove universal bounds for the kth eigenvalue in terms of the lower ones independent of the domains. Our results strengthen the recent work in Jost et al. (Commun Partial Differ Equ 35:1563–1589, 2010) and generalize Cheng–Yang’s recent estimates (Cheng and Yang in Trans Am Math Soc 364:6139–6158, 2012) on the buckling eigenvalues of order two to arbitrary order.
Similar content being viewed by others
1 Introduction
Let \(\Omega \) be a bounded domain with smooth boundary in an \(n(\ge 2)\)-dimensional Riemannian manifold M and denote by \(\Delta \) the Laplace operator acting on functions on M. Let \(\nu \) be the outward unit normal vector field of \(\partial \Omega \), and let us consider the following eigenvalue problems :
They are called the fixed membrane problem and the bucking problem, respectively. It should be mentioned that the buckling problem (1.2) has interpretations in physics, that is, it describes the critical buckling load of a clamped plate subjected to a uniform compressive force around its boundary. Let
denote the successive eigenvalues for (1.1) and (1.2), respectively. Here each eigenvalue is repeated according to its multiplicity. An important theme of geometric analysis is to estimate these (and other) eigenvalues. When \(\Omega \) is a bounded domain in an n-dimensional Euclidean space \(\mathbf{R}^n\), Payne, Pólya and Weinberger (cf. [32, 33]) proved the bound
Inequality of this type is called a universal inequality since it does not depend on \(\Omega \).
On the other hand, Payne, Pólya and Weinberger also studied eigenvalues of the buckling problem (1.2) for bounded domains in \(\mathbf{R}^n\) and proved (cf. [32, 33])
For \(\Omega \subset \mathbf{R}^n\), this reads
Furthermore, Payne, Pólya and Weinberger proposed the following
Problem 1
(cf. [32, 33]). Can one obtain a universal inequality for the eigenvalues of the buckling problem (1.2) on a bounded domain in \(\mathbf{R}^n\) which is similar to the universal inequality (1.3) for the eigenvalues of the fixed membrane problem (1.1)?
With respect to the above problem, Hile and Yeh [27] obtained
Ashbaugh [2] proved :
This inequality has been improved to the following form in [30]:
By introducing a new method of constructing trial functions, Cheng and Yang [12] have obtained the following universal inequality:
Thus, the problem proposed by Payne, Pólya and Weinberger has been solved affirmatively. By making use of the asymptotic formula of Weyl for eigenvalues of the Dirichlet eigenvalue problem of the Laplacian and one of Agmon [1] and Pleijel [34] for eigenvalues of the clamped plate problem, we can have the asymptotic formula of eigenvalues for the buckling problem according to the variational characterization for eigenvalues of the buckling problem:
where \(\omega _n\) denotes the volume of the unit ball in \({\mathbf {R}}^n\). By the results of Li and Yau [31] and the variational characterization for eigenvalues, one can obtain a lower bound for eigenvalues of the buckling problem:
On the other hand, by making use of the recursion formula of Cheng and Yang [14], one can obtain an upper bound for eigenvalues of the buckling problem, which is sharp in the sense of the order of k, if one can get a sharp universal inequality for eigenvalues of the buckling problem as the following:
Conjecture. Eigenvalues of the buckling problem on a bounded domain in a Euclidean space \({\mathbf {R}}^n\) satisfy the following universal inequality:
which is proposed by Cheng and Yang [14]. Therefore, the next landmark goal for the study on eigenvalues of the buckling problem will be to prove the above sharp universal inequality. Recently, Cheng and Yang [17] have made an important breakthrough for it. They have obtained
In this paper, we will investigate eigenvalues of the buckling problem of arbitrary order:
where \(\Omega \) is a bounded domain in a Euclidean space and l is any integer no less than 2. Yang type inequalities for eigenvalues of the problem (1.9) have been obtained recently in [29]. We conjecture that the following sharp universal inequality:
holds for eigenvalues of the problem (1.9). The main purpose of this paper is to attack the above problem. We prove the following:
Theorem 1.1
Let \(\Lambda _i\) be the ith eigenvalue of the buckling problem (1.9), where \(\Omega \) is a bounded domain with smooth boundary in \(\mathbf{R}^n\). Then for any positive non-increasing monotone sequence \(\{\delta _i\}_{i=1}^k\), we have
Remark 1.1
Taking
in (1.11), we have
which improves the inequality (1.13) in [29]. From (1.12), we can obtain a quadratic inequality about \(\Lambda _1,\ldots ,\Lambda _{k+1}\).
Corollary 1.1
For any \(k\ge 1\), the first \(k+1\) eigenvalues of the buckling problem (1.9) with \(\Omega \subset \mathbf{R}^n\) satisfy the following inequality
Remark 1.2
When \(l=2\), (1.13) becomes Cheng–Yang’s inequality (1.8).
Furthermore, we prove the following universal inequality for eigenvalues of the buckling problem of arbitrary order on spherical domains.
Theorem 1.2
Let \(l\ge 2\) and let \(\Lambda _i\) be the ith eigenvalue of the buckling problem:
where \(\Omega \) is a domain with smooth boundary in \(S^n\). For each \(q= 1, \ldots , \) define the polynomials \(\Phi _q\) inductively by
Set
Then for any positive integer k and any positive non-increasing monotone sequence \(\{\delta _i\}_{i=1}^k\), we have
where
with \(a_j^{+}=\max \{a_j, 0\}\) and when \(l=2\) we use the convention that \(\sum _{j=1}^{l-2}a_j^{+}\Lambda _i^{j/(l-1)}=0\).
Remark 1.3
When \(l=2\), the inequality (1.18) is stronger than one of Cheng and Yang in [17].
Remark 1.4
Universal inequalities for eigenvalues of various elliptic operators have been studied extensively in recent years. For the developments in this direction, we refer to [1,2,3,4,5,6,7,8,9,10,11,12,13,14,15,16,17,18,19,20,21,22,23,24,25,26,27,28,29,30, 33,34,35,36,37,38,39,40,41,42] and the references therein.
2 Proofs of the results
First we recall a method of constructing trial functions developed by Cheng–Yang (cf. [12, 29, 37]). Let M be an n-dimensional complete submanifold in \(\mathbf{R}^{m}\). Denote by \(\langle , \rangle \) the canonical metric on \(\mathbf{R}^{m}\) as well as that induced on M. Denote by \(\Delta \) and \(\nabla \) the Laplacian and the gradient operator of M, respectively. Let \(\Omega \) be a bounded domain with smooth boundary in M and let \(\nu \) be the outward unit normal vector field of \(\partial \Omega \). For functions \(f, g\in W_0^{1,2}(\Omega )\) the Dirichlet inner product \((f, g)_D\) of f and g is given by
The Dirichlet norm of a function f is defined by
Consider the eigenvalue problem
Let
denote the successive eigenvalues, where each eigenvalue is repeated according to its multiplicity.
Let \(u_i\) be the ith orthonormal eigenfunction of the problem (2.1) corresponding to the eigenvalue \(\Lambda _i, i=1, 2, \ldots ,\) that is,
Consider the subspace \(W_{0,l}^{1,2}(\Omega )\) of \(W_{0}^{1,2}(\Omega )\) defined by
The eigenfunctions \(\{u_i\}_{i=1}^{\infty }\) defined in (2.2) form a complete orthonormal basis for the Hilbert space \(W_{0,l}^{1,2}(\Omega )\). If \(\phi \in W_{0,l}^2(\Omega )\) satisfies \((\phi , u_j)_D=0, \ \forall j=1, 2, \ldots , k\), then the Rayleigh–Ritz inequality tells us that
For vector-valued functions \(F=(f_1, f_2, \ldots , f_{m}), \ G=(g_1, g_2, \ldots , g_{m}): \Omega \rightarrow \mathbf{R}^{m}\), we define an inner product (F, G) by
The norm of F is given by
Let
Observe that a vector field on \(\Omega \) can be regarded as a vector-valued function from \(\Omega \) to \(\mathbf{R}^m\). Let \(\mathbf{L}_{0, 1}^2(\Omega )\subset \mathbf{L}^2(\Omega )\) be the subspace of \(\mathbf{L}^2(\Omega )\) spanned by the vector-valued functions \(\{ \nabla u_i\}_{i=1}^{\infty }\), which form a complete orthonormal basis for the Hilbert space \(\mathbf{L}_{0, 1}^2(\Omega )\). For any \(f\in W_{0,,l}^2(\Omega ), \) we have \(\nabla f\in \mathbf{L}_{0,1}^2(\Omega )\), and for any \(X\in \mathbf{L}_{0,1}^2(\Omega )\), there exists a function \(f\in W_{0,,l}^2(\Omega )\) such that \(X=\nabla f\).
Lemma 2.1
(cf. [29, 30]) Let \(u_i\) and \(\Lambda _i, i=1, 2, \ldots , \) be as in (2.2), then
We are now ready to prove the main results in this paper.
Proof of Theorem 1.1
With the notations as above, we consider now the special case that \(\Omega \) is a bounded domain in \(\mathbf{R}^{n}\). Denote by \(x_1,\ldots , x_n\) the coordinate functions of \(\mathbf{R}^{n}\) and let us decompose the vector-valued functions \(x_{\alpha }\nabla u_i\) as
where \(h_{\alpha i}\in W_{0,l}^{1,2}(\Omega ), \nabla h_{\alpha i}\) is the projection of \(x_{\alpha } \nabla u_i\) in \(\mathbf{L}_{0,1}^2(\Omega )\) and \(W_{\alpha i}\ \bot \ \mathbf{L}_{0,1}^2(\Omega )\). Thus we have
and from the discussions in [12] and [37] we know that
where for a vector field Z on \(\Omega , \mathrm{div}\ Z\) denotes the divergence of Z. \(\square \)
For each \(\alpha =1,\ldots , n, i=1,\ldots , k\), consider the functions \(\phi _{\alpha i}: \Omega \rightarrow \mathbf{R}\), given by
where
We have
It then follows from the Rayleigh–Ritz inequality for \(\Lambda _{k+1}\) that
After some calculations, we have (cf. (2.36) in [29])
It is easy to see that
where for a vector field Z on \(\Omega , ||Z||^2=\int _{\Omega } |Z|^2\). Combining (2.14)–(2.16), we infer
Observe that \(\nabla (x_{\alpha }u_i)=u_i\nabla x_{\alpha }+ x_{\alpha }\nabla u_i\in \mathbf{L}_{0,1}^2(\Omega )\). Set \(y_{\alpha i}=x_{\alpha } u_i-h_{\alpha i}\); then
and so
Substituting (2.18) into (2.17), we get
Summing on \(\alpha \) from 1 to n, we have
Using the divergence theorem, one can show that (cf. [12, 29])
Set
then \(d_{\alpha ij}=-d_{\alpha ji}\) and we have from (2.7), (2.8), (2.10) and (2.20) that
Thus, we have
where \(u_{i,\alpha }= \langle \nabla u_i, \nabla x_{\alpha }\rangle \). Summing on \(\alpha \) from 1 to n, we have by using (2.19) that
Summing on i from 1 to k and noticing the fact that \(a_{\alpha ij}=a_{\alpha ji},\ d_{\alpha ij}=-d_{\alpha ji}\), one gets
Since \(\{\delta _i\}_{i=1}^k\) is a non-increasing monotone sequence, we have
We conclude from (2.22) that
It follows from the divergence theorem and Lemma 2.1 that
where \(u_{i,\alpha \alpha }=\frac{\partial ^2 u_i}{\partial x_{\alpha }^2}\). Thus, we have
Before we can finish the proof of Theorem 1.2, we shall need two lemmas.
Lemma 2.2
For any i, we have
Proof
When \(l=2\), the above formula has been proved by Cheng and Yang [17]. We only consider the case that \(l>2\). In this case, we conclude from the boundary condition on \(u_i\) that \( y_{\alpha i}|_{\partial \Omega }=\nabla y_{\alpha i}|_{\partial \Omega }=\Delta y_{\alpha i}|_{\partial \Omega }=0\). Using the divergence theorem, we have
and
It follows from (2.26)–(2.29) that
Since
we get
On the other hand, we have
We obtain from (2.31) and (2.32) that
Combining (2.30) and (2.33), we infer
Summing on \(\alpha \), we get (2.25). \(\square \)
Lemma 2.3
For any i, we have
Proof
Using the definition of \(W_{\alpha i}\) and the divergence theorem and noticing (2.20), we have
On the other hand, for \(\epsilon >0\), we have
From (2.36), we have
Also, one can check that
Thus we have from (2.37)–(2.39) that
Taking
we get (2.35). This completes the proof of Lemma 2.3. \(\square \)
Let us continue the proof of Theorem 1.1. Since \(||u_i||^2=||W_{\alpha i}||^2+||\nabla y_{\alpha i}||^2\), we have from (2.35) that
which, combining with (2.25), implies that
Substituting (2.42) into (2.24) and using Lemma 2.1, we get
This completes the proof of Theorem 1.1.
Proof of Corollary 1.2
By induction, one can show that
which, combining with (1.12), gives (1.13). \(\square \)
Proof of Theorem 1.2
We use the same notations as in the beginning of this section and take M to be the unit n-sphere \(S^n\). Let \(x_1, x_2,\ldots , x_{n+1}\) be the standard coordinate functions of the Euclidean space \(\mathbf{R}^{n+1}\), then
It is well known that
As in the proof of Theorem 1.1, we decompose the vector-valued functions \(x_{\alpha }\nabla u_i\) as
where \(h_{\alpha i}\in W_{0,l}^{1,2}(\Omega ), \nabla h_{\alpha i}\) is the projection of \(x_{\alpha } \nabla u_i\) in \(\mathbf{L}_{0,1}^2(\Omega )\), \(W_{\alpha i}\ \bot \ \mathbf{L}_{0,1}^2(\Omega )\) and
We also consider the functions \(\phi _{\alpha i}: \Omega \rightarrow \mathbf{R}\), given by
Then
and we have the basic Rayleigh–Ritz inequality for \(\Lambda _{k+1}\) :
We have
For a function g on \(\Omega \), we have (cf. (2.31) in [37])
For each \(q=0, 1,\ldots \), thanks to (2.43) and (2.50), there are polynomials \(F_q\) and \(G_q\) of degree q such that
It is obvious that
It follows from (2.43) and (2.50) that
which gives
Also, when \(q\ge 2\), we have (cf. (2.65) and (2.66) in [29])
\(\square \)
For each \(q=1, 2, \ldots , \) let us set
We conclude from (2.52), (2.54)–(2.56) that the polynomials \(\Phi _q, \ q=1, 2, \ldots , \) are defined inductively by (1.15) and (1.16). Substituting
into (2.49), we get
Summing over \(\alpha \) and noticing
we infer
Set
then it is easy to check from Lemma 2.1 that
Substituting (2.62) into (2.60), we have
Observe from (2.44) and (2.46) that
Summing over \(\alpha \), one gets
Combining (2.47), (2.63) and (2.65), we get
Set
then \(c_{\alpha ij}=-c_{\alpha ji}\) (cf. Lemma in [37]). By using the same arguments as in the proof of (2.37) in [37], we have
Since
we have by summing over \(\alpha \) in (2.68) from 1 to \(n+1\) that
The following inequalities have been proved in [29]:
and
Thus, we have by combining (2.66), (2.70), (2.72) and (2.73) that
Since \(\{\delta _i\}_{i=1}^k\) is a positive non-increasing monotone sequence, we have
Hence, by summing over i from 1 to k in (2.74), we infer
That is
where \(S_i\) is given by (1.19). This completes the proof of Theorem 1.2.
References
Agmon, S.: On kernels, eigenvalues and eigenfunctions of operators related to elliptic problems. Commun. Pure Appl. Math. 18, 627–663 (1965)
Ashbaugh, M.S.: Isoperimetric and universal inequalities for eigenvalues. In: Davies, E.B., Safarov Y. (eds.) Spectral Theory and Geometry (Edingurgh, 1998), London Math. Soc. Lecture Notes, vol. 273, pp. 95–139. Cambridge University Press, Cambridge (1999)
Ashbaugh, M.S.: The universal eigenvalue bounds of Payne–Pólya–Weinberger, Hile-Protter and H C Yang. Proc. India Acad. Sci. Math. Sci. 112, 3–30 (2002)
Ashbaugh, M.S., Hermi, L.: A unified approach to universal inequalities for eigenvalues of elliptic operators. Pac. J. Math. 217, 201–219 (2004)
Ashbaugh, M.S., Hermi, L.: On Harrell–Stubbe Type Inequalities for the Discrete Spectrum of a Self-adjoint Operator. arXiv:0712.4396
Chen, D., Cheng, Q.-M.: Extrinsic estimates for eigenvalues of the Laplace operator. J. Math. Soc. Jpn. 60, 325–339 (2008)
Cheng, Q.-M., Ichikawa, T., Mametsuka, S.: Inequalities for eigenvalues of Laplacian with any order. Commun. Contemp. Math. 11, 639–655 (2009)
Cheng, Q.-M., Ichikawa, T., Mametsuka, S.: Estimates for eigenvalues of the poly-Laplacian with any order in a unit sphere. Calc. Var. Partial Differ. Equ. 36, 507–523 (2009)
Cheng, Q.-M., Ichikawa, T., Mametsuka, S.: Estimates for eigenvalues of a clamped plate problem on Riemannian manifolds. J. Math. Soc. Jpn. 62, 673–686 (2010)
Cheng, Q.-M., Yang, H.C.: Estimates on eigenvalues of Laplacian. Math. Ann. 331, 445–460 (2005)
Cheng, Q.-M., Yang, H.C.: Inequalities for eigenvalues of a clamped plate problem. Trans. Am. Math. Soc. 358, 2625–2635 (2006)
Cheng, Q.-M., Yang, H.C.: Universal bounds for eigenvalues of a buckling problem. Commun. Math. Phys. 262, 663–675 (2006)
Cheng, Q.-M., Yang, H.C.: Inequalities for eigenvalues of Laplacian on domains and compact complex hypersurfaces in complex projective spaces. J. Math. Soc. Jpn. 58, 545–561 (2006)
Cheng, Q.-M., Yang, H.C.: Bounds on eigenvalues of Dirichlet Laplacian. Math. Ann. 337, 159–175 (2007)
Cheng, Q.-M., Yang, H.C.: Estimates for eigenvalues on Riemannian manifolds. J. Differ. Equ. 247(8), 2270–2281 (2009)
Cheng, Q.-M., Yang, H.C.: Universal inequalities for eigenvalues of a system of elliptic equations. Proc. R. Soc. Edinb. Sect. A 139, 273–285 (2009)
Cheng, Q.-M., Yang, H.C.: Universal bounds for eigenvalues of a buckling problem II. Trans. Am. Math. Soc. 364, 6139–6158 (2012)
do Carmo, M.P., Wang, Q., Xia, C.: Inequalities for eigenvalues of elliptic operators in divergence form on Riemannian manifolds. Ann. Math. Pura Appl. 189, 643–660 (2010)
Harrell, E.M.: Some geometric bounds on eigenvalue gaps. Commun. Partial Differ. Equ. 18, 179–198 (1993)
Harrell, E.M., Hermi, L.: Differential inequalities for Riesz means and Weyl-type bounds for eigenvalues. J. Funct. Anal. 254, 3173–3191 (2008)
Harrell, E.M., Michel, P.L.: Commutator bounds for eigenvalues, with applications to spectral geometry. Commun. Partial Differ. Equ. 19, 2037–2055 (1994)
Harrell, E.M., Michel, P.L.: Commutator bounds for eigenvalues of some differential operators. In: Ferreyra, G., Goldstein, G.R., Neubrander, F. (eds.) Lecture Notes in Pure and Applied Mathematics, vol. 168, pp. 235–244. Marcel Dekker, New York (1995)
Harrell, E.M., Yildirim, S.: Yolcu, Eigenvalue inequalities for Klein–Gordon operators. J. Funct. Anal. 256, 3977–3995 (2009)
Harrell, E.M., Stubbe, J.: On trace identities and the universal eigenvalue estimates for some partial differential operators. Trans. Am. Math. Soc. 349, 1797–1809 (1997)
Hebey, E.: Nonlinear analysis on manifolds: Sobolev spaces and inequalities. Courant Lecture Notes in Mathematics, vol. 5. New York University, Courant Institute of Mathematical Sciences, New York. American Mathematical Society, Providence, RI (1999)
Hile, G.N., Protter, M.H.: Inequalities for eigenvalues of the Laplacian. Indiana Univ. Math. J. 29, 523–538 (1980)
Hile, G.N., Yeh, R.Z.: Inequalities for eigenvalues of the biharmonic operator. Pac. J. Math. 112, 115–133 (1984)
Ilias, S., Makhoul, O.: ‘Universal’ inequalities for the eigenvalues of the Hodge de Rham Laplacian. Ann. Glob. Anal. Geom. 36, 191–204 (2009)
Jost, J., Li-Jost, X., Wang, Q., Xia, C.: Universal inequalities for eigenvalues of the buckling problem of arbitrary order. Commun. Partial Differ. Equ. 35, 1563–1589 (2010)
Jost, J., Li-Jost, X., Wang, Q., Xia, C.: Universal bounds for eigenvalues of the polyharmonic operator. Trans. Am. Math. Soc. 363, 1821–1854 (2011)
Li, P., Yau, S.T.: On the Schrödinger equations and the eigenvalue problem. Commun. Math. Phys. 88, 309–318 (1983)
Payne, L.E., Pólya, G., Weinberger, H.F.: sur le quotient de deux fréquences propres cosécutives. Comptes Rendus Acad. Sci. Paris 241, 917–919 (1955)
Payne, L.E., Pólya, G., Weinberger, H.F.: On the ratio of consecutive eigenvalues. J. Math. Phys. 35, 289–298 (1956)
Pleijel, A.: On the eigenvalues and eigenfunctions of elastic plates. Commun. Pure Appl. Math. 3, 1–10 (1950)
Sun, H., Cheng, Q.-M., Yang, H.C.: Lower order eigenvalues of Dirichlet Laplacian. Manuscr. Math. 125, 139–156 (2008)
Wang, Q., Xia, C.: Universal bounds for eigenvalues of the biharmonic operator on Riemannian manifolds. J. Funct. Anal. 245, 334–352 (2007)
Wang, Q., Xia, C.: Universal bounds for eigenvalues of the buckling problem on spherical domains. Commun. Math. Phys. 270, 759–775 (2007)
Wang, Q., Xia, C.: Universal bounds for eigenvalues of Schrödinger operator on Riemannian manifolds. Ann. Acad. Sci. Fenn. Math. 33, 319–336 (2008)
Wang, Q., Xia, C.: Universal bounds for eigenvalues of the biharmonic operator. J. Math. Anal. Appl. 364, 1–17 (2010)
Wang, Q., Xia, C.: Inequalities for eigenvalues of the biharmonic operator with weight on Riemannian manifolds. J. Math. Soc. Jpn. 62, 597–622 (2010)
Wang, Q., Xia, C.: Inequalities for eigenvalues of a clamped plate problem. Calc. Var. Partial Differ. Equ. 40, 273–289 (2011)
Yang, H.C.: An estimate of the difference between consecutive eigenvalues, preprint IC/91/60 of ICTP, Trieste (1991)
Acknowledgements
The research of the first author is partially supported by a Grant-in-Aid for Scientific Research from JSPS, the research of the third author is partially supported by CNPq, and the main part of this paper has been done while the fourth author visited to Department of Mathematics, Saga University as a fellowship of JSPS. This author would like to express his gratitude to JSPS for finance support and to Professor Qing-Ming Cheng and Saga University for the worm hospitality. The authors are very grateful to the referee for the encouragements and valuable suggestions.
Author information
Authors and Affiliations
Corresponding author
Rights and permissions
About this article
Cite this article
Cheng, QM., Qi, X., Wang, Q. et al. Inequalities for eigenvalues of the buckling problem of arbitrary order. Annali di Matematica 197, 211–232 (2018). https://doi.org/10.1007/s10231-017-0676-x
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s10231-017-0676-x