## Boundary Value Problems

Impact Factor 0.642

Open Access

# Multiple solutions for a fourth-order nonlinear elliptic problem which is superlinear at +∞ and linear at −∞

Boundary Value Problems20142014:12

DOI: 10.1186/1687-2770-2014-12

Accepted: 16 December 2013

Published: 10 January 2014

## Abstract

We consider a semilinear fourth-order elliptic equation with a right-hand side nonlinearity which exhibits an asymmetric growth at +∞ and at −∞. Namely, it is linear at −∞ and superlinear at +∞. Combining variational methods with Morse theory, we show that the problem has at least two nontrivial solutions, one of which is negative.

### Keywords

fourth-order elliptic boundary value problems multiple solutions critical groups Morse theory

## 1 Introduction

Consider the following Navier boundary value problem:
(1)

where ${\mathrm{△}}^{2}$ is the biharmonic operator, and Ω is a bounded smooth domain in ${\mathbb{R}}^{N}$ ($N>4$), and $c<{\lambda }_{1}^{\ast }$ the first eigenvalue of − in ${H}_{0}^{1}\left(\mathrm{\Omega }\right)$.

The conditions imposed on $f\left(x,t\right)$ are as follows:

(H1) $f\in {C}^{1}\left(\overline{\mathrm{\Omega }}×\mathbb{R},\mathbb{R}\right)$, $f\left(x,0\right)=0$ for all $x\in \mathrm{\Omega }$ and $f\left(x,t\right)<0$ for all $t<0$ and all $x\in \mathrm{\Omega }$;

(H2) there exist $r\in \left(2,{p}^{\ast }\right)$ and $A,B>0$ such that $|{f}_{t}^{\prime }\left(x,t\right)|\le A+B{|t|}^{r-2}$ for all $x\in \mathrm{\Omega }$, and $t\in \mathbb{R}$, where ${p}^{\ast }=\frac{2N}{N-4}$, if $N>4$;

(H3) ${lim}_{t\to -\mathrm{\infty }}\frac{f\left(x,t\right)}{t}=l$ uniformly for $x\in \mathrm{\Omega }$, where l is a nonnegative constant;

(H4) there exist $\beta ,\xi \in \mathbb{R}$ such that for $F\left(x,t\right)={\int }_{0}^{t}f\left(x,s\right)\phantom{\rule{0.2em}{0ex}}ds$, we have
(2)
(3)
and
(4)
(H5) there exist ${\vartheta }_{1},{\vartheta }_{2}\in {L}^{\mathrm{\infty }}{\left(\mathrm{\Omega }\right)}_{+}$ and an integer $k\ge 2$ such that
(5)
the first and the last inequality are strict on sets (not necessary the same) of positive measure, and
(6)
In view of the conditions (H3) and equation (3) in (H4), it is clear that for all $x\in \mathrm{\Omega }$, $f\left(x,t\right)$ is linear at −∞ and superlinear at +∞. Clearly, $u=0$ is a trivial solution of problem (1). It follows from (H1) and (H2) that the functional
$I\left(u\right)=\frac{1}{2}{\int }_{\mathrm{\Omega }}\left({|\mathrm{△}u|}^{2}-c{|\mathrm{\nabla }u|}^{2}\right)\phantom{\rule{0.2em}{0ex}}dx-{\int }_{\mathrm{\Omega }}F\left(x,u\right)\phantom{\rule{0.2em}{0ex}}dx$
(7)
is ${C}^{2}$ on the space ${H}_{0}^{1}\left(\mathrm{\Omega }\right)\cap {H}^{2}\left(\mathrm{\Omega }\right)$ with the norm
$\parallel u\parallel :={\left({\int }_{\mathrm{\Omega }}\left({|\mathrm{△}u|}^{2}-c{|\mathrm{\nabla }u|}^{2}\right)\phantom{\rule{0.2em}{0ex}}dx\right)}^{\frac{1}{2}},$
where $F\left(x,t\right)={\int }_{0}^{t}f\left(x,s\right)\phantom{\rule{0.2em}{0ex}}ds$. Under the condition (H2), the critical points of I are solutions of problem (1). Let $0<{\lambda }_{1}<{\lambda }_{2}<\cdots <{\lambda }_{k}<\cdots$ be the eigenvalues of $\left({\mathrm{△}}^{2}+c\mathrm{△},{H}^{2}\left(\mathrm{\Omega }\right)\cap {H}_{0}^{1}\left(\mathrm{\Omega }\right)\right)$ and ${\varphi }_{1}\left(x\right)>0$ be the eigenfunction corresponding to ${\lambda }_{1}$. In fact, ${\lambda }_{1}={\lambda }_{1}^{\ast }\left({\lambda }_{1}^{\ast }-c\right)$. Let ${E}_{{\lambda }_{k}}$ denote the eigenspace associated with ${\lambda }_{k}$. Throughout this article, we denoted by ${|\cdot |}_{p}$ the ${L}^{p}\left(\mathrm{\Omega }\right)$ norm and $E={H}^{2}\left(\mathrm{\Omega }\right)\cap {H}_{0}^{1}\left(\mathrm{\Omega }\right)$. The aim of this paper is to prove a multiplicity theorem for problem (1) when the nonlinearity term $f\left(x,t\right)$ exhibits an asymmetric behavior as $t\in \mathbb{R}$ approaches +∞ and −∞. In the past, some authors studied the following elliptic problem:
$-\mathrm{△}u=f\left(x,u\right),\phantom{\rule{1em}{0ex}}u\in {H}_{0}^{1}\left(\mathrm{\Omega }\right)$
(8)

with asymmetric nonlinearities by using the Fučík spectrum of the operator $\left(-\mathrm{△},{H}_{0}^{1}\left(\mathrm{\Omega }\right)\right)$. This approach requires that $f\left(x,t\right)$ exhibits linear growth at both +∞ and −∞ and that the limits ${lim}_{t\to ±\mathrm{\infty }}\frac{f\left(x,t\right)}{t}$ exist and belong to . See the works of Các [1], Dancer and Zhang [2], Magalhães [3], de Paiva [4], Schechter [5] and the references therein. Equations with nonlinearities which are superlinear in one direction and linear in the other were investigated by Arcoya and Villegas [6] and Perera [7]. They let the nonlinearity $f\left(x,t\right)$ be line at −∞ and satisfy the Ambrosetti-Rabinowitz condition at +∞. Particularly, it is worth noticing paper [8]. The authors relax several of the above restrictions on the nonlinearity $f\left(x,t\right)$. Their nonlinearity is only measurable in $x\in \mathrm{\Omega }$. The limit as $t\to -\mathrm{\infty }$ of $\frac{f\left(x,t\right)}{t}$ need not exist and the growth at −∞ can be linear or sublinear. Furthermore, their nonlinearity $f\left(x,t\right)$ does not satisfy the famous AR-condition. They use the truncated skill of first order weak derivative to verify the (PS) condition and obtain multiple solutions for problem (1) by combining variational methods and Morse theory.

To the authors’ knowledge, there seem to be few results on problem (1) when $f\left(x,t\right)$ is asymmetric nonlinearity at positive infinity and at negative infinity. However, the method in [8] cannot be applied directly to the biharmonic problems. For example, for the Laplacian problem, $u\in {H}_{0}^{1}\left(\mathrm{\Omega }\right)$ implies $|u|,{u}_{+},{u}_{-}\in {H}_{0}^{1}\left(\mathrm{\Omega }\right)$, where ${u}_{+}=max\left(u,0\right)$, ${u}_{-}=max\left(-u,0\right)$. We can use ${u}_{+}$ or ${u}_{-}$ as a test function, which is helpful in proving a solution nonnegative. While for the biharmonic problems, this trick fails completely since $u\in {H}_{0}^{2}\left(\mathrm{\Omega }\right)$ does not imply ${u}_{+},{u}_{-}\in {H}_{0}^{2}\left(\mathrm{\Omega }\right)$ (see [[9], Remark 2.1.10] and [10, 11]). As far as this point is concerned, we will make use of the new methods to overcome it.

This fourth-order semilinear elliptic problem can be considered as an analogue of a class of second-order problems which have been studied by many authors. In [12], there was a survey of results obtained in this direction. In [13], Micheletti and Pistoia showed that $\left({P}_{1}\right)$ admits at least two solutions by a variation of linking if $f\left(x,u\right)$ is sublinear. Chipot [14] proved that the problem $\left({P}_{1}\right)$ has at least three solutions by a variational reduction method and a degree argument. In [15], Zhang and Li showed that $\left({P}_{1}\right)$ admits at least two nontrivial solutions by Morse theory and local linking if $f\left(x,u\right)$ is superlinear and subcritical on u.

In this article, under the guidance of [8], we consider multiple solutions of problem (1) with the asymmetric nonlinearity by using variational methods and Morse theory.

## 2 Main result and auxiliary lemmas

Let us now state the main result.

Theorem 2.1 Assume conditions (H1)-(H5) hold. If $l<{\lambda }_{1}$, then problem (1) has at least two nontrivial solutions.

Lemma 2.2 Under the assumptions of Theorem  2.1, then I satisfies the (PS) condition.

Proof Let $\left\{{u}_{n}\right\}\subset E$ be a sequence such that for every $n\in N$,
$|\frac{1}{2}{\int }_{\mathrm{\Omega }}\left({|\mathrm{△}{u}_{n}|}^{2}-c{|\mathrm{\nabla }{u}_{n}|}^{2}\right)\phantom{\rule{0.2em}{0ex}}dx-{\int }_{\mathrm{\Omega }}F\left(x,{u}_{n}\right)\phantom{\rule{0.2em}{0ex}}dx|\le c,$
(9)
$|{\int }_{\mathrm{\Omega }}\left(\mathrm{△}{u}_{n}\mathrm{△}v-c\mathrm{\nabla }{u}_{n}\mathrm{\nabla }v\right)\phantom{\rule{0.2em}{0ex}}dx-{\int }_{\mathrm{\Omega }}f\left(x,{u}_{n}\right)v\phantom{\rule{0.2em}{0ex}}dx|\le {\epsilon }_{n}\parallel v\parallel ,\phantom{\rule{1em}{0ex}}v\in E,$
(10)
where $c>0$ is a positive constant and $\left\{{\epsilon }_{n}\right\}\subset {\mathbb{R}}^{+}$ is a sequence which converges to zero. By a standard argument, in order to prove that $\left\{{u}_{n}\right\}$ has a convergence subsequence, we have to show that it is a bounded sequence. To do this, we argue by contradiction assuming that for a subsequence, denoted by $\left\{{u}_{n}\right\}$, we have
Without loss of generality we can assume $\parallel {u}_{n}\parallel >1$ for all $n\in \mathbb{N}$ and define ${z}_{n}=\frac{{u}_{n}}{\parallel {u}_{n}\parallel }$. Obviously, $\parallel {z}_{n}\parallel =1$ $\mathrm{\forall }n\in \mathbb{N}$ and then it is possible to extract a subsequence (denoted also by $\left\{{z}_{n}\right\}$) such that
(11)
(12)
(13)
(14)
where ${z}_{0}\in E$ and $q\in {L}^{2}\left(\mathrm{\Omega }\right)$. Dividing both sides of inequality (10) by $\parallel {u}_{n}\parallel$, we obtain
Passing to the limit we deduce from equation (11) that
$\underset{n\to \mathrm{\infty }}{lim}{\int }_{\mathrm{\Omega }}\frac{f\left(x,{u}_{n}\right)}{\parallel {u}_{n}\parallel }v\phantom{\rule{0.2em}{0ex}}dx={\int }_{\mathrm{\Omega }}\left(\mathrm{△}{z}_{0}\mathrm{△}v-c\mathrm{\nabla }{z}_{0}\mathrm{\nabla }v\right)\phantom{\rule{0.2em}{0ex}}dx$
(15)

for all $v\in E$.

Now we claim that ${z}_{0}\left(x\right)\le 0$ a.e. $x\in \mathrm{\Omega }$. To verify this, let us observe that by choosing $v={z}_{0}$ in equation (15) we have
$\underset{n\to \mathrm{\infty }}{lim}{\int }_{{\mathrm{\Omega }}^{+}}\frac{f\left(x,{u}_{n}\right)}{\parallel {u}_{n}\parallel }{z}_{0}\phantom{\rule{0.2em}{0ex}}dx<+\mathrm{\infty },$
(16)
where ${\mathrm{\Omega }}^{+}=\left\{x\in \mathrm{\Omega }|{z}_{0}\left(x\right)>0\right\}$. But, on the other hand, from (H3) and equation (3) in (H4), we have
for some positive constant ${K}_{1}>0$. Moreover, using ${lim}_{n\to \mathrm{\infty }}{u}_{n}\left(x\right)=+\mathrm{\infty }$ a.e. $x\in {\mathrm{\Omega }}^{+}$, equation (13) and the superlinearity of f, we also deduce
Therefore, if $|{\mathrm{\Omega }}^{+}|>0$ we will obtain by Fatou’s lemma that
$\underset{n\to \mathrm{\infty }}{lim}{\int }_{{\mathrm{\Omega }}^{+}}\frac{f\left(x,{u}_{n}\left(x\right)\right)}{\parallel {u}_{n}\parallel }{z}_{0}\left(x\right)\phantom{\rule{0.2em}{0ex}}dx=+\mathrm{\infty }$

which contradicts inequality (16). Thus $|{\mathrm{\Omega }}^{+}|=0$ and the claim is proved.

Clearly, ${z}_{0}\left(x\right)\not\equiv 0$, by (H3), there exists $C>0$ such that $\frac{|f\left(x,{u}_{n}\right)|}{|{u}_{n}|}\le C$ for a.e. $x\in \mathrm{\Omega }$. By using Lebesgue dominated convergence theorem in equation (15), we have
${\int }_{\mathrm{\Omega }}\left(\mathrm{△}{z}_{0}\mathrm{△}v-c\mathrm{\nabla }{z}_{0}\mathrm{\nabla }v\right)\phantom{\rule{0.2em}{0ex}}dx-{\int }_{\mathrm{\Omega }}l{z}_{0}v\phantom{\rule{0.2em}{0ex}}dx=0$
(17)

for all $v\in E$. This contradicts $l<{\lambda }_{1}$. □

Lemma 2.3 Let $E=V\oplus W$, where $V={E}_{{\lambda }_{1}}\oplus {E}_{{\lambda }_{2}}\oplus \cdots \oplus {E}_{{\lambda }_{k}}$. If $k\ge 0$ is an integer, $\vartheta \in {L}^{\mathrm{\infty }}{\left(\mathrm{\Omega }\right)}_{+}$, $\vartheta \left(x\right)\le {\lambda }_{k+1}$ a.e. on Ω and the inequality is strict on a set of positive measure, then there exists $\gamma >0$ such that
${\parallel u\parallel }^{2}-{\int }_{\mathrm{\Omega }}\vartheta {u}^{2}\phantom{\rule{0.2em}{0ex}}dx\ge \gamma {\parallel u\parallel }^{2}$

for all $u\in W$.

Proof We claim that there exists a constant ${\vartheta }_{0}<1$ such that
${\int }_{\mathrm{\Omega }}\vartheta \left(x\right){u}^{2}\phantom{\rule{0.2em}{0ex}}dx\le {\vartheta }_{0}{\parallel u\parallel }^{2}$
(18)
for all $u\in W$. In fact, if not, there exists a sequence $\left\{{u}_{n}\right\}$ such that
${\int }_{\mathrm{\Omega }}\vartheta \left(x\right){|{u}_{n}|}^{2}\phantom{\rule{0.2em}{0ex}}dx>\left(1-\frac{1}{n}\right){\parallel {u}_{n}\parallel }^{2}$
for all $n\in N$, which implies ${u}_{n}\ne 0$ for all n. By the homogeneity of the above inequality, we may assume that $\parallel {u}_{n}\parallel =1$ and
${\int }_{\mathrm{\Omega }}\vartheta \left(x\right){|{u}_{n}|}^{2}\phantom{\rule{0.2em}{0ex}}dx>1-\frac{1}{n}$
(19)
for all n. It follows from the weak compactness of the unit ball of W that there exists a subsequence, say $\left\{{u}_{n}\right\}$, such that ${u}_{n}$ weakly converges to u in W. Now Sobolev’s embedding theorem suggests that $\left\{{u}_{n}\right\}$ converges to u in ${L}^{2}\left(\mathrm{\Omega }\right)$. From inequality (19) we obtain
${\int }_{\mathrm{\Omega }}\vartheta \left(x\right){|u|}^{2}\phantom{\rule{0.2em}{0ex}}dx\ge 1.$
Moreover one has
$1\ge {\parallel u\parallel }^{2}\ge {\lambda }_{k+1}{|u|}_{2}^{2}\ge {\int }_{\mathrm{\Omega }}\vartheta \left(x\right){|u|}^{2}\phantom{\rule{0.2em}{0ex}}dx\ge 1.$
Hence we have
${\parallel u\parallel }^{2}={\lambda }_{k+1}{|u|}_{2}^{2}$
and
${\int }_{\mathrm{\Omega }}\left({\lambda }_{k+1}-\vartheta \left(x\right)\right){u}^{2}\phantom{\rule{0.2em}{0ex}}dx=0$

which implies that $u\in {E}_{{\lambda }_{k+1}}\setminus \left\{0\right\}$ and $u=0$ on a positive measure subset. It contradicts the unique continuation property of the eigenfunction. □

## 3 Computation of the critical groups

It is well known that critical groups and Morse theory are the main tools in solving elliptic partial differential equation. Let us recall some results which will be used later. We refer the readers to the book [16] for more information on Morse theory.

Let H be a Hilbert space and $I\in {C}^{1}\left(H,\mathbb{R}\right)$ be a functional satisfying the (PS) condition or (C) condition, and ${H}_{q}\left(X,Y\right)$ be the q th singular relative homology group with integer coefficients. Let ${u}_{0}$ be an isolated critical point of I with $I\left({u}_{0}\right)=c$, $c\in \mathbb{R}$, and U be a neighborhood of ${u}_{0}$. The group
${C}_{q}\left(I,{u}_{0}\right):={H}_{q}\left({I}^{c}\cap U,{I}^{c}\cap U\setminus \left\{{u}_{0}\right\}\right),\phantom{\rule{1em}{0ex}}q\in Z$

is said to be the q th critical group of I at ${u}_{0}$, where ${I}^{c}=\left\{u\in H:I\left(u\right)\le c\right\}$.

Let $K:=\left\{u\in H:{I}^{\prime }\left(u\right)=0\right\}$ be the set of critical points of I and $a, the critical groups of I at infinity are formally defined by [17]
${C}_{q}\left(I,\mathrm{\infty }\right):={H}_{q}\left(H,{I}^{a}\right),\phantom{\rule{1em}{0ex}}q\in Z.$
From the deformation theorem, we see that the above definition is independent of the particular choice of $c. If $c then
${C}_{q}\left(I,\mathrm{\infty }\right):={H}_{q}\left(H,{\stackrel{˙}{I}}^{a}\right),\phantom{\rule{1em}{0ex}}q\in Z.$
(20)

For the convenience of our proof, we first recall two interesting results and prove two important propositions.

Proposition 3.1 [18]

Under (H2), if $u\in E:={H}^{2}\left(\mathrm{\Omega }\right)\cap {H}_{0}^{1}\left(\mathrm{\Omega }\right)$ is an isolated critical point of I, then ${C}_{\ast }\left(I,u\right)\cong {C}_{\ast }\left(I{|}_{{C}_{0}^{3}\left(\mathrm{\Omega }\right)},u\right)$.

Proposition 3.2 [19]

If ${D}_{1}\subset D\subset {E}_{0}\subset {E}_{1}\subset X$ and for some integer $k\ge 0$ we have ${H}_{k}\left(E,D\right)\ne 0$ and ${H}_{k}\left({E}_{1},{D}_{1}\right)=0$, then either ${H}_{k+1}\left({E}_{1},E\right)\ne 0$ or ${H}_{k-1}\left(D,{D}_{1}\right)\ne 0$.

Proposition 3.3 If the assumptions of Theorem  2.1 hold, then
Proof Under the guidance of [8] and [18], we begin to prove this result. Let ${I}_{1}=I{|}_{{C}_{0}^{3}\left(\overline{\mathrm{\Omega }}\right)}$. Indeed, it follows from above Proposition 3.1 that I and ${I}_{1}$ have same critical set. Since ${C}_{0}^{3}\left(\overline{\mathrm{\Omega }}\right)$ is dense in E, invoking Proposition 16 of Palais [20], we have
(21)
From equations (20) and (21), we see that in order to prove the proposition, it suffices to show that
(22)
In order to prove equation (22), we proceed as follows. We define the sets
$\partial {B}_{1}^{c}=\left\{u\in {C}_{0}^{3}\left(\overline{\mathrm{\Omega }}\right):{\parallel u\parallel }_{{C}_{0}^{3}\left(\overline{\mathrm{\Omega }}\right)}=1\right\}$
and
Consider the map ${h}_{+}:\left[0,1\right]×\partial {B}_{1,+}^{c}\to \partial {B}_{1,+}^{c}$ defined by

Clearly, ${h}_{+}$ is a continuous homotopy and $h\left(1,u\right)={\varphi }_{1}$ for all $x\in \partial {B}_{1,+}^{c}$. Therefore, $\partial {B}_{1,+}^{c}$ is contractible in itself.

By equation (3) in (H4), given any $\gamma >0$, we can find $C=C\left(\gamma \right)>0$ such that
(23)
Similarly, from condition (H3), and by choosing $C>0$ even bigger if necessary, we observe that there is a number ${\gamma }_{0}>0$ such that
(24)
Moreover, by condition (H2), we have
(25)

for some ${C}_{3}>0$.

Let $u\in \partial {B}_{1,+}^{c}$. By inequalities (23), (24), and (25), for all $t>0$ we have
$\begin{array}{rl}I\left(tu\right)& =\frac{{t}^{2}}{2}{\parallel u\parallel }^{2}-{\int }_{\mathrm{\Omega }}F\left(x,tu\right)\phantom{\rule{0.2em}{0ex}}dx\\ =\frac{{t}^{2}}{2}{\parallel u\parallel }^{2}-{\int }_{tu\ge C}F\left(x,tu\right)\phantom{\rule{0.2em}{0ex}}dx-{\int }_{tu\le -C}F\left(x,tu\right)\phantom{\rule{0.2em}{0ex}}dx-{\int }_{|tu|\le C}F\left(x,tu\right)\phantom{\rule{0.2em}{0ex}}dx\\ \le \frac{{t}^{2}}{2}\left[\left(1+\frac{{\gamma }_{0}}{{\lambda }_{1}}\right){\parallel u\parallel }^{2}-\gamma {\int }_{tu\ge C}{u}^{2}\phantom{\rule{0.2em}{0ex}}dx\right]+{C}_{3}|\mathrm{\Omega }|.\end{array}$
(26)
Recalling that $\gamma >0$ is arbitrary, from (26), we have
(27)
Using formula (4) in condition (H4), we see that there exist constants ${\xi }_{0}$ and $M>0$ such that
(28)
By (H2) and formula (2) in condition (H4), we have
(29)
for some $C>0$. By inequalities (28) and (29), for any $u\in E$ we have
${\int }_{\mathrm{\Omega }}\left(2F\left(x,t\right)-f\left(x,t\right)t\right)\phantom{\rule{0.2em}{0ex}}dx\le C,$
(30)
where C is a positive constant. Let $i:{C}_{0}^{3}\left(\overline{\mathrm{\Omega }}\right)\to E$ be the continuous embedding map. Let ${〈\cdot ,\cdot 〉}_{0}$ denote the duality brackets for the pair $\left({C}_{0}^{3}{\left(\overline{\mathrm{\Omega }}\right)}^{\ast },{C}_{0}^{3}\left(\overline{\mathrm{\Omega }}\right)\right)$. We let ${I}_{2}=I\circ i$, and so
(31)
Then, from equation (27), we obtain
(32)
From conditions (H2) and (H3), we see that given $ϵ>0$, we can find $M>0$ such that
(33)
Using inequality (33), we have
$\begin{array}{rl}I\left(u\right)& \ge \frac{1}{2}\left({\parallel u\parallel }^{2}-l{|u|}_{2}^{2}-ϵ{|u|}_{2}^{2}\right)-M\\ \ge C{\parallel u\parallel }^{2}-M\end{array}$
for $u\in -{C}_{+}$, where ${C}_{+}$ is defined as
and $C>0$ is a positive constant. So $I{|}_{-{C}^{+}}$ is coercive, thus we find ${C}^{\ast \ast }>0$ such that $I{|}_{-{C}^{+}}\ge -{C}^{\ast \ast }$. We pick
$a
Then inequality (32) implies that we can find $k\left(u\right)>1$ such that

Moreover, the implicit function theorem implies that $k\in C\left(\partial {B}_{1,+}^{c},\left[1,+\mathrm{\infty }\right)\right)$.

By the choice of a, we have
${I}_{2}^{a}=\left\{tu:u\in \partial {B}_{1,+}^{c},t\ge k\left(u\right)\right\}.$
(34)
We define the set ${E}_{+}=\left\{tu:u\in \partial {B}_{1,+}^{c},t\ge 1\right\}$. The map ${\stackrel{ˆ}{h}}_{+}:\left[0,1\right]×{E}_{+}\to {E}_{+}$ defined by
(35)
is a continuous deformation of ${E}_{+}$, ${\stackrel{ˆ}{h}}_{+}\left(1,{E}_{+}\right)\subset {I}_{2}^{a}$ and ${\stackrel{ˆ}{h}}_{+}\left(s,\cdot \right){|}_{{I}_{2}^{a}}=\mathit{id}{|}_{{I}_{2}^{a}}$ for all $s\in \left[0,1\right]$ (see equations (34) and (35)). Therefore, ${I}_{2}^{a}$ is a strong deformation retract of ${E}_{+}$. Hence we have
(36)
Recalling that in the first part of the proof, we established that $\partial {B}_{1,+}^{c}$ is contractible. This yields

Combining with equation (36) leads to equation (22), which completes the proof. □

Proposition 3.4 If the assumptions of Theorem  2.1 hold, then
${C}_{d}\left(I,0\right)\ne 0,$

where $d=dimV$ (V being defined in Lemma  2.3).

Proof By condition (H5), given $ϵ>0$, we can find ${\delta }_{\ast }>0$ such that
(37)
Since V is finite dimensional, all norms are equivalent. Thus we can find $\rho >0$ small such that
$\parallel u\parallel \le \rho \phantom{\rule{1em}{0ex}}⇔\phantom{\rule{1em}{0ex}}{\parallel u\parallel }_{\mathrm{\infty }}\le {\delta }_{\ast }$
(38)
for all $u\in V$. Taking inequalities (37) and (38) into account, for all $u\in V$ with $\parallel u\parallel \le \rho$ we have
$I\left(u\right)\le \frac{1}{2}{\int }_{\mathrm{\Omega }}\left({\lambda }_{k}-{\vartheta }_{1}\left(x\right)\right){u}^{2}\phantom{\rule{0.2em}{0ex}}dx+\frac{ϵ}{2}{|u|}_{2}^{2}.$
(39)
Similar to the proof of Lemma 2.3, there exists $C>0$ such that
$I\left(u\right)\le \left(-C+ϵ\right){\parallel u\parallel }^{2}\le 0$
(40)

for all $u\in V$ and $\parallel u\parallel \le \rho$.

On the other hand, for given $ϵ>0$, it follows from (H2) and (H5) that
$F\left(x,t\right)\le \frac{{\vartheta }_{2}\left(x\right)+ϵ}{2}{t}^{2}+{C}_{ϵ}{|t|}^{r}$
(41)
for all $x\in \mathrm{\Omega }$ and $t\in R$. By (41) and Lemma 2.3, we have
$I\left(u\right)\ge {C}_{4}{\parallel u\parallel }^{2}-{C}_{5}{\parallel u\parallel }^{r}$
(42)
for all $u\in W$. From inequality (42), we infer that for ρ small enough we have
(43)

From inequalities (40) and (43), we know that I has a local linking at 0. Then invoking Proposition 2.3 of Bartsch and Li [17], we obtain ${C}_{d}\left(I,0\right)\ne 0$. □

## 4 Proof of the main result

Proof of Theorem 2.1 We consider the following problem:
$\left\{\begin{array}{l}{\mathrm{△}}^{2}u+c\mathrm{△}u={f}_{-}\left(x,u\right),\phantom{\rule{1em}{0ex}}x\in \mathrm{\Omega },\\ u{|}_{\partial \mathrm{\Omega }}=\mathrm{△}u{|}_{\partial \mathrm{\Omega }}=0,\end{array}$
where
${f}_{-}\left(x,t\right)=\left\{\begin{array}{ll}f\left(x,t\right),& t<0,\\ 0,& t\ge 0.\end{array}$
Define a functional ${I}_{-}:E={H}^{2}\left(\mathrm{\Omega }\right)\cap {H}_{0}^{1}\left(\mathrm{\Omega }\right)\to \mathbb{R}$ by
${I}_{-}\left(u\right)=\frac{1}{2}{\int }_{\mathrm{\Omega }}\left({|\mathrm{△}u|}^{2}-c{|\mathrm{\nabla }u|}^{2}\right)\phantom{\rule{0.2em}{0ex}}dx-{\int }_{\mathrm{\Omega }}{F}_{-}\left(x,u\right)\phantom{\rule{0.2em}{0ex}}dx,$
where ${F}_{-}\left(x,t\right)={\int }_{0}^{t}{f}_{-}\left(x,s\right)\phantom{\rule{0.2em}{0ex}}ds$, then ${I}_{-}\in {C}^{2}\left(E,\mathbb{R}\right)$. Obviously, by conditions (H1) and (H3), we know that ${I}_{-}$ is coercive and boundedness from below. Thus we can find ${v}_{0}\in E$ such that
${I}_{-}\left({v}_{0}\right)=inf{I}_{-}=:{m}_{-}.$
(44)
Next, we claim that ${v}_{0}\ne 0$. By condition (H5), given $ϵ\in \left(0,{\lambda }_{k}-{\lambda }_{1}\right)$, there exists $\delta >0$ such that
(45)
For s small enough, it follows from inequality (45) that
$\begin{array}{rl}{I}_{-}\left(-s{\varphi }_{1}\right)& =\frac{{s}^{2}}{2}{\lambda }_{1}-{\int }_{\mathrm{\Omega }}{F}_{-}\left(x,-s{\varphi }_{1}\right)\phantom{\rule{0.2em}{0ex}}dx\\ \le \frac{{s}^{2}}{2}\left({\lambda }_{1}-{\lambda }_{k}+ϵ\right)<0,\end{array}$
and thus, by equation (44), ${I}_{-}\left({v}_{0}\right)\le {I}_{-}\left(-s{\varphi }_{1}\right)<0$, so ${v}_{0}\ne 0$. From condition (H1) and strong maximum principle, we have ${v}_{0}<0$ and
Since ${v}_{0}$ is an interior point of $-{C}_{+}$, from Proposition 3.1, we know
(46)
Let $\theta \in \mathbb{R}$, $ϵ>0$ be such that $\theta <{m}_{-}=I\left({v}_{0}\right)<-ϵ$. We consider the sublevel sets
${I}^{\theta }\subset {I}^{-ϵ}\subset {I}^{ϵ}\subset E.$
Suppose that 0 and ${v}_{0}$ are the only critical points of I. Otherwise, we have a second nontrivial smooth solution and so we are done. By Proposition 3.3, we have
(47)
We know that I satisfies the (PS) condition (see Lemma 2.2). Hence choosing $ϵ>0$ small enough, we have
${H}_{d}\left({I}^{ϵ},{I}^{-ϵ}\right)={C}_{d}\left(I,0\right)\ne 0$
(48)
(see Proposition 3.4). Because of equations (47) and (48), using Proposition 3.2, we obtain
${H}_{d+1}\left(E,{I}^{ϵ}\right)\ne 0\phantom{\rule{1em}{0ex}}\text{or}\phantom{\rule{1em}{0ex}}{H}_{d-1}\left({I}^{-ϵ},{I}^{\theta }\right)\ne 0.$
If ${H}_{d+1}\left(E,{I}^{ϵ}\right)\ne 0$, then there is a critical point ${v}^{\ast }\in E$ of I such that
$I\left({v}^{\ast }\right)>ϵ>0\phantom{\rule{1em}{0ex}}\text{and}\phantom{\rule{1em}{0ex}}{v}^{\ast }\ne 0,{v}_{0}.$
If ${H}_{d-1}\left({I}^{-ϵ},{I}^{\theta }\right)\ne 0$, then there is a critical point ${v}^{\ast }\in E$ of I such that
${C}_{d-1}\left(I,{v}^{\ast }\right)\ne 0.$
(49)

Since $d\ge 2$, from equations (46) and (49), we see that ${v}^{\ast }\ne {v}_{0}$. It is obvious that ${v}^{\ast }\ne 0$. Therefore ${v}_{0}$ and ${v}^{\ast }$ are two solutions of problem (1). □

## Declarations

### Acknowledgements

The authors would like to thank the referees for valuable comments and suggestions in improving this article. This study was supported by the National NSF (Grant No. 11101319) of China and Planned Projects for Postdoctoral Research Funds of Jiangsu Province (Grant No. 1301038C).

## Authors’ Affiliations

(1)
School of Mathematics and Statistics, Tianshui Normal University
(2)
School of Mathematics and Computer Sciences, Nanjing Normal University

## References

1. Các NP: On nontrivial solutions of a Dirichlet problem whose jumping nonlinearity crosses a multiple eigenvalue. J. Differ. Equ. 1989, 80: 379-404. 10.1016/0022-0396(89)90090-9
2. Dancer EN, Zhang Z: Fučík spectrum, sign-changing, and multiple solutions for semilinear elliptic boundary value problems with resonance at infinity. J. Math. Anal. Appl. 2000, 250: 449-464. 10.1006/jmaa.2000.6969
3. Magalhães CA: Multiplicity results for a semilinear elliptic problem with crossing of multiple eigenvalues. Differ. Integral Equ. 1991, 4: 129-136.
4. de Paiva FO: Multiple solutions for a class of quasilinear problems. Discrete Contin. Dyn. Syst. 2006, 15: 669-680.
5. Schechter M: The Fučík spectrum. Indiana Univ. Math. J. 1994, 43: 1139-1157. 10.1512/iumj.1994.43.43050
6. Arcoya D, Villegas S: Nontrivial solutions for a Neumann problem with a nonlinear term asymptotically linear at −∞ and superlinear at +∞. Math. Z. 1995, 219: 499-513. 10.1007/BF02572378
7. Perera K: Existence and multiplicity results for a Sturm-Liouville equation asymptotically linear at −∞ and superlinear at +∞. Nonlinear Anal. 2000, 39: 669-684. 10.1016/S0362-546X(98)00228-4
8. Motreanu D, Motreanu VV, Papageorgiou NS: Multiple solutions for Dirichlet problems which are superlinear at +∞ and (sub-) linear at −∞. Commun. Pure Appl. Anal. 2009, 13: 341-358.
9. Ziemer WP Grad. Texts in Math. 120. In Weakly Differentiable Functions. Springer, Berlin; 1989.
10. Liu Y, Wang ZP: Biharmonic equations with asymptotically linear nonlinearities. Acta Math. Sci. 2007, 27: 549-560. 10.1016/S0252-9602(07)60055-1
11. Pei RC: Multiple solutions for biharmonic equations with asymptotically linear nonlinearities. Bound. Value Probl. 2010., 2010: Article ID 241518Google Scholar
12. Lazer AC, Mckenna PJ: Large amplitude periodic oscillation in suspension bridges: some new connections with nonlinear analysis. SIAM Rev. 1990, 32: 537-578. 10.1137/1032120
13. Micheletti AM, Pistoia A: Multiplicity solutions for a fourth order semilinear elliptic problems. Nonlinear Anal. TMA 1998, 31: 895-908. 10.1016/S0362-546X(97)00446-X
14. Chipot M: Variational Inequalities and Flow in Porous Media. Springer, New York; 1984.
15. Zhang JH, Li SJ: Multiple nontrivial solutions for some fourth-order semilinear elliptic problems. Nonlinear Anal. TMA 2005, 60: 221-230.
16. Chang KC: Infinite Dimensional Morse Theory and Multiple Solutions Problems. Birkhäuser, Boston; 1993.
17. Bartsch T, Li SJ: Critical point theory for asymptotically quadratic functionals and applications to problems with resonance. Nonlinear Anal. TMA 1997, 28: 419-441. 10.1016/0362-546X(95)00167-T
18. Qian AX: Multiple solutions for a fourth-order asymptotically linear elliptic problem. Acta Math. Sin. 2006, 22: 1121-1126. 10.1007/s10114-005-0665-7
19. Perera K: Critical groups of critical points produced by local linking with applications. Abstr. Appl. Anal. 1998, 3: 437-446. 10.1155/S1085337598000657
20. Palais RS: Homotopy theory of infinite dimensional manifolds. Topology 1966, 5: 1-16. 10.1016/0040-9383(66)90002-4