Volume 2013, Issue 1 597193
Research Article
Open Access

Existence and Multiplicity of Nontrivial Solutions for a Class of Fourth-Order Elliptic Equations

Chun Li

Chun Li

School of Mathematics and Statistics, Southwest University, Chongqing 400715, China southwest.edu

Search for more papers by this author
Zeng-Qi Ou

Zeng-Qi Ou

School of Mathematics and Statistics, Southwest University, Chongqing 400715, China southwest.edu

Search for more papers by this author
Chun-Lei Tang

Corresponding Author

Chun-Lei Tang

School of Mathematics and Statistics, Southwest University, Chongqing 400715, China southwest.edu

Search for more papers by this author
First published: 05 December 2013
Academic Editor: Shaoyong Lai

Abstract

Using the Fountain theorem and a version of the Local Linking theorem, we obtain some existence and multiplicity results for a class of fourth-order elliptic equations.

1. Introduction and Main Results

Consider the fourth-order Navier boundary value problem
()
where ΩN  (N > 4) is a bounded smooth domain, aL(Ω),   c, and fC(Ω × , ). Δ is the Laplace operator and Δ2 is the biharmonic operator.
Let 0 < λ1 < λ2λ3 ≤ ⋯≤λk ≤ ⋯ be the eigenvalues of −Δ in . The eigenvalue problem
()
has infinitely many eigenvalues μi = λi(λic),  i = 1,2, ….
We will always assume c < λ1. Let E be the Hilbert space .  E is equipped with the inner product
()
and the norm
()
A weak solution of problem (1) is any uE such that
()
for any vE.
Let φ : E be the functional defined by
()
where . And, one has
()
for any u, vE, so that a critical point of the functional φ in E corresponds to a weak solution of problem (1).

In recent years, fourth-order problems have been studied by many authors. In [1], Lazer and McKenna have pointed out that problem (1) furnishes a model to study travelling waves in suspension bridges if f(x, u) = b((u + 1) + − 1), where u+ = max {u, 0} and b. Since then, more general nonlinear fourth-order elliptic boundary value problems have been studied.

In [2, 3], Micheletti and Pistoia proved that problem
()
admits two or three solutions by variational method. In [4], Zhang obtained the existence of weak solutions for problem (8) when f(x, u) is sublinear at . In [5], Zhang and Li showed that problem (8) has at least two nontrivial solutions by means of Morse theory and local linking. When f(x, u) is asymptotically linear at infinity, the existence of three nontrivial solutions has been obtained in [6] by using Morse theory. In [7], by using the mountain pass theorem, An and Liu gave the existence result for nontrivial solutions for a class of asymptotically linear fourth-order elliptic equations. In [8], Zhou and Wu got the existence of four sign-changing solutions or infinitely many sign-changing solutions for (8) by using the sign-changing critical point theorems. In [9], Yang and Zhang showed new results on invariant sets of the gradient flows of the corresponding variational functionals and proved the existence of positive, negative, and sign-changing solutions for some fourth-order semilinear elliptic boundary value problems. In [10], by using the variational method, Liu and Huang obtained an existence result of sign-changing solutions as well as positive and negative solutions for a fourth-order elliptic problem whose nonlinear term is asymptotically linear at both zero and infinity.

In this paper, we will study the existence of nontrivial solutions of problem (1). Our main results are the following theorems.

Theorem 1. Assume that F is even in u and the following conditions hold:

  • F1

    ()

  • uniformly in xΩ;

  • F2 there exist two constants 2 < θ < 2N/(N − 4) = 2** and k1 > 0 such that

    ()

  • uniformly for all xΩ;

  • F3 there exists a constant γ > N(θ − 2)/4 such that

    ()

  • uniformly for all xΩ.

Then problem (1) has infinitely many nontrivial solutions.

Theorem 2. Assume that F satisfies (F1) and

  • F4 there exist three positive constants L, m1, and m2 such that

    • j1 f(x, u)u − 2F(x, u) ≥ m1|u|2, if |u| ≥ L;

    • j2 |f(x,u)|σ/|u|σm2(f(x, u)u − 2F(x, u)), if |u| ≥ L, where σ > (N − 2)/2.

If F is even in u, problem (1) has infinitely many nontrivial solutions.

Remark 3. For Schrödinger equation, the corresponding condition (F4) is due to Ding and Luan [11]. The condition (F4) is weaker than the usual Ambrosetti-Rabinowitz-type condition (see [11, 12]).

Theorem 4. Assume that F satisfies (F1), (F2), (F3), and

  • F5

    ()

  • uniformly in xΩ.

If 0 is an eigenvalue of Δ2 + cΔ + a (with Navier boundary condition), assume also the condition that
  • F6 there exists δ > 0 such that

    • (i)

      F(x, u) ≥ 0,  for all |u| ≤ δ,  xΩ; or

    • (ii)

      F(x, u) ≤ 0,  for all |u| ≤ δ,  xΩ.

Then problem (1) has at least one nontrivial solution.

Theorem 5. Suppose that F satisfies (F1), (F4), and (F5). If 0 is an eigenvalue of Δ2 + cΔ + a (with Navier boundary condition), assuming also (F6), then problem (1) has at least one nontrivial solution.

Here, we have
()
where uE,   u+E+, and E(E+) is the space spanned by the eigenvectors corresponding to negative (positive) eigenvalues of Δ2 + cΔ + a. It is easy to know that φC1(E, ) under the conditions of our theorems.
It is well known that E is continuously embedded in Lλ(Ω) for every λ ∈ [1,2N/(N − 4)]. If 1 ≤ λ < 2N/(N − 4), the embedding is compact. It follows from (F2), (F3), and (F4) that
()
Hence, there is a positive constant K such that
()
for λ = 1, 2, θ, 2σ/(σ − 1), 2N/(N − 4) = 2**, where denotes the norm of Lλ(Ω).

2. Proof of Main Results

In this paper, we will use the Fountain Theorem of Bartsch ([13, Theorem 2.5], [14, Theorem 3.6]) to prove our Theorems 1 and 2. And, we will prove Theorems 4 and 5 by using a version of Local Linking theorem [12, Theorem 2.2] which extends theorems given by Li and Willem [15], Li and Szulkin [16].

In [13, 14], Bartsch established the Fountain Theorem under the (PS) c condition. Since the Deformation Theorem is still valid under the Cerami condition, the Fountain Theorem is true under the Cerami condition. So, we have the following Fountain Theorem.

Let X be a reflexive and separable Banach space. It is well known that there exist such that
  • (1)

    ψn, vm〉 = δn,m where δn,m = 1 for n = m and δn,m = 0 for nm.

  • (2)

    , .

Let Xj = vj; then . We define
()

Theorem A (Fountain theorem). Assume that φC1(X, ) satisfies the Cerami condition (C), φ(−u) = φ(u). If for almost every k, there exist ρk > rk > 0 such that

  • A1

    ()

  • A2

    ()

  • then φ has an unbounded sequence of critical values.

For the reader′s convenience, we state the following Local Linking theorem [12, Theorem 2.2]. Let X be a real Banach space with X = X1X2 and such that . For every multi-index α = (α1, α2) ∈ 2, let . We define that αβα1β1, α2β2. A sequence {αn} ⊂ 2 is admissible if for every α2 there is m such that nmαnα. We say that φC1(X, ) satisfies the (C*) condition if every sequence such that {αn} is admissible and satisfies
()
contains a subsequence which converges to a critical point of φ, where .

Theorem B (see [12], Theorem 2.2.)Suppose that φC1(X, ) satisfies the following assumptions:

  • i1 X ≠ {0} and φ has a local linking at 0; that is, for some r > 0,

    ()

  • i2 φ satisfies (C*) condition;

  • i3 φ maps bounded sets into bounded sets;

  • i4 for every m,   φ(u)→− as |u| → , on .

Then φ has at least one nonzero critical point.

Now, we can give the proof of our theorems.

Proof of Theorem 1. At first, we claim that φ satisfies the Cerami condition (C). Consider a sequence {un} such that φ(un) is bounded and ∥φ(un)∥(1 + ∥un∥) → 0 as n. Then there exists a constant M1 > 0 such that

()
By a standard argument, we only need to prove that {un} is a bounded sequence in E. Otherwise, going if necessary to a subsequence, we can assume that ∥un∥ → as n. From (F3), there exist two constants k2, k3 > 0 such that
()
So, by (21) and (22), we have
()
which implies that
()
for all n and some positive constant k4.

Since

()
On the one hand, we consider the case
()

Putting

()
one has 0 < α < 1. Let
()
where E0 = ker (Δ2 + cΔ + a). We can obtain from Hölder′s inequality, (15), and (24) that
()
for all n, where q = p/(p − 1) = 2**/(α + 1).

By (F2) and (29), one has

()
for all n. Since α < 1, we have
()

On the other hand, if γ satisfies,

()
then one sees 1 ≤ γ/(γθ + 1) ≤ 2N/(N − 4). So, we get
()
It follows from (24) that
()
By (F2) and (33), we obtain
()
for all n. Note that γ ≥ 2N(θ − 1)/(N + 4) and N > 4 imply that γ > θ − 1. So, it follows from (34) and the above expression that
()
Hence, we conclude from (31) and (36) that
()
Similarly for , we get
()

It follows from the equivalence of the norms on the finite dimensional space E0 that there exists K1 > 0 such that

()
Putting τ = 2N + (5 − N)γ + 1, one has
()
It follows from (24), (39), and Hölder′s inequality that
()
and consequently
()
Hence, by (37), (38), and (42), one sees
()
as n, which is a contradiction. So, we obtain that {un} is bounded in E. By a standard argument, we get that φ satisfies the condition (C).

Let with dim Xj < for any j ≥ 1. Set

()
Since dim (Yk) < , all the norms are equivalent. For uYk, there exists a constant K2 > 0 such that
()
From condition (F1), there exists k5 > 0 such that
()
For uYk, it follows from (45) and (46) that
()
which implies that
()
So, (A1) of Theorem A is satisfied for every ρk > 0 large enough.

Here, we obtain from (F2) that there exists a positive constant k6 such that

()
uniformly for all xΩ. Let us define
()
For k large enough, one has ZkE+. By (49), on Zk, we have
()
Choosing , we obtain, if uZk and ∥u∥ = rk,
()
Since, by Lemma 3.8 of [14], βk → 0 as k, (A2) is proved. Hence, the proof is completed by using Fountain theorem.

Proof of Theorem 2. Firstly, we claim that φ satisfies the Cerami condition (C). Consider a sequence {un} such that φ(un) is bounded from above and ∥φ(un)∥(1 + ∥un∥) → 0 as n. By a standard argument, we only need to prove that {un} is a bounded sequence in E. For otherwise, we can assume that ∥un∥ → as n.

From assumption (F4), there exist two positive constants m3 and m4, such that

()
So, one has
()
for all n and some positive constant m5.

Let vn = un/∥un∥; then ∥vn∥ = 1 and for all r ∈ [1,2N/(N − 4)). By (54), we have

()
as n. So, for r ∈ (2, (2N − 4)/(N − 4)), it follows from Hölder′s inequality and the above expression that
()
as n. It follows from (39) that
()
Hence, we get
()
From (F4), (53), and (56), there exists a positive constant m6 such that
()
as n; therefore, 1 = o(1), which is a contradiction. Hence, {un} is bounded.

In a way similar to the proof of Theorem 1, we can obtain that φ satisfies (A1) of Theorem A.

It follows from (F1) that there is L1 > 0 such that

()
Then, by (F4) and (60), for |u| ≥ L2 = max {L, L1},   xΩ, one has
()
which implies that
()
for |u| ≥ L2.

Therefore, there exist two positive constants m7 and m8 such that

()
where 2σ/(σ − 1) < 2N/(N − 4). As the proof of Theorem 1, we can get (A2). Therefore, Theorem 2 holds.

Proof of Theorem 4. The proof of this theorem is divided in several steps.

Step  1. We claim that φ has a local linking at zero with respect to (X1, X2).

By (F5), for any ε > 0, there exists δ1 > 0 such that

()
We obtain from the above expression and (49) that
()
where . Hence, we can get from (15) and (65) that
()
for all uE.

Here, we consider only the case where 0 is an eigenvalue of Δ2 + cΔ + a and case (ii) of (F6) holds. The case (i) is similar.

Let X = E,   X1 = E+E0, and X2 = E, where E0 = ker (Δ2 + cΔ + a). Choose a Hilbertian basis {en} n≥0 for X1 and define

()
Now, by (66), for each uX2 = E, one has
()
Letting ε = 1/(8K2) and by θ > 2, we have
()
for δ2 > 0 small enough.

Let u = u0 + u+E0E+ = X1 be such that ∥u∥ ≤ δ3 : = δ/(2K1). Put

()
Then, for all ∥u∥ ≤ δ3 and xΩ, by (39), one sees
()
On one hand, from above expression, for any xΩ1, we have
()
Hence, by condition (ii) of (F6), we get
()
On the other hand, for any xΩ2, one has
()
Hence, for all xΩ2 and uX1 with ∥u∥ ≤ δ3, we can obtain from (65) that
()
which implies that
()
Letting ε = 1/(16K2) in above expression, then for all xΩ2 and uX1 with ∥u∥ ≤ δ3, we have
()
which implies that
()
for δ4 > 0 small enough. Hence, φ has a local linking at zero with respect to (X1, X2) for δ5 = min {δ2, δ4} small enough.

Step  2. In a way similar to the proof of Theorem 1, we can get that φ satisfies the (C*) condition.

Step  3. Now, we claim that for each m, one has

()
Since dim (E0) < and , all the norms are equivalent. For , there exists a constant k7 > 0 such that
()
From condition (F1), there exists a constant k8 > 0 such that
()
For , it follows from (80) and (81) that
()
which implies that
()
Hence, all the assumptions of Theorem B are verified. Then, the proof of Theorem 4 is completed.

Proof of Theorem 5. In a way similar to the proof of Theorems 2 and 4, we can obtain that φ satisfies (i1), (i2), (i3), and (i4) of Theorem B. Therefore, Theorem 5 holds.

Conflict of Interests

The authors declare that there is no conflict of interests regarding the publication of this paper.

    Acknowledgments

    The authors would like to thank the referee for the valuable suggestions. This paper is supported by the National Natural Science Foundation of China (no. 11071198) and the Fundamental Research Funds for the Central Universities (no. XDJK2010C055).

        The full text of this article hosted at iucr.org is unavailable due to technical difficulties.