Show Summary Details
More options …

Editor-in-Chief: Radulescu, Vicentiu / Squassina, Marco

IMPACT FACTOR 2018: 6.636

CiteScore 2018: 5.03

SCImago Journal Rank (SJR) 2018: 3.215
Source Normalized Impact per Paper (SNIP) 2018: 3.225

Mathematical Citation Quotient (MCQ) 2018: 3.18

Open Access
Online
ISSN
2191-950X
See all formats and pricing
More options …
Volume 8, Issue 1

# On the existence of a weak solution for some singular $p\left(x\right)$-biharmonic equation with Navier boundary conditions

Khaled Kefi
• Corresponding author
• Community College of Rafha, Northern Border University, 73222 Arar, Kingdom of Saudi Arabia; and Mathematics Department, Faculty of Sciences Tunis El Manar, 1060 Tunis, Tunisia
• Email
• Other articles by this author:
/ Kamel Saoudi
Published Online: 2018-03-13 | DOI: https://doi.org/10.1515/anona-2016-0260

## Abstract

In the present paper, we investigate the existence of solutions for the following inhomogeneous singular equation involving the $p\left(x\right)$-biharmonic operator:

where $\mathrm{\Omega }\subset {ℝ}^{N}$ ($N\ge 3$) is a bounded domain with ${C}^{2}$ boundary, λ is a positive parameter, $\gamma :\overline{\mathrm{\Omega }}\to \left(0,1\right)$ is a continuous function, $p\in C\left(\overline{\mathrm{\Omega }}\right)$ with $1<{p}^{-}:={inf}_{x\in \mathrm{\Omega }}p\left(x\right)\le {p}^{+}:={sup}_{x\in \mathrm{\Omega }}p\left(x\right)<\frac{N}{2}$, as usual, ${p}^{*}\left(x\right)=\frac{Np\left(x\right)}{N-2p\left(x\right)}$,

$g\in {L}^{\frac{{p}^{*}\left(x\right)}{{p}^{*}\left(x\right)+\gamma \left(x\right)-1}}\left(\mathrm{\Omega }\right),$

and $f\left(x,u\right)$ is assumed to satisfy assumptions (f1)–(f6) in Section 3. In the proofs of our results, we use variational techniques and monotonicity arguments combined with the theory of the generalized Lebesgue Sobolev spaces. In addition, an example to illustrate our result is given.

MSC 2010: 35J20; 35J60; 35G30; 35J35

## 1 Introduction

Let Ω be a smooth bounded domain in ${ℝ}^{N}$ ($N\ge 3$) with ${C}^{2}$ boundary condition. In this paper, we are dealing with the following problem:

(P-+λ)

where λ is a positive parameter, $\gamma \left(x\right)\in C\left(\overline{\mathrm{\Omega }}\right)$ satisfying $0<{\gamma }^{-}={inf}_{x\in \mathrm{\Omega }}\gamma \left(x\right)\le {\gamma }^{+}={sup}_{x\in \mathrm{\Omega }}\gamma \left(x\right)<1$, $p\in C\left(\overline{\mathrm{\Omega }}\right)$ with $1<{p}^{-}:={inf}_{x\in \mathrm{\Omega }}p\left(x\right)\le {p}^{+}:={sup}_{x\in \mathrm{\Omega }}p\left(x\right)<\frac{N}{2}$, as usual, ${p}^{*}\left(x\right)=\frac{Np\left(x\right)}{N-2p\left(x\right)}$, and

$g\in {L}^{\frac{{p}^{*}\left(x\right)}{{p}^{*}\left(x\right)+\gamma \left(x\right)-1}}\left(\mathrm{\Omega }\right)$

and is almost everywhere positive in Ω. In the sequel, X will denote the Sobolev space ${W}^{2,p\left(x\right)}\left(\mathrm{\Omega }\right)\cap {W}_{0}^{1,p\left(x\right)}\left(\mathrm{\Omega }\right)$. Associated to problem (P-+λ), we have the singular functional ${I}_{\mp \lambda }:X\to ℝ$ given by

${I}_{\mp \lambda }\left(u\right)=J\left(u\right)-{\mathrm{\Phi }}_{\mp \lambda }\left(u\right),$

where

$J\left(u\right)={\int }_{\mathrm{\Omega }}\frac{1}{p\left(x\right)}{|\mathrm{\Delta }u|}^{p\left(x\right)}dx$

and

${\mathrm{\Phi }}_{\mp \lambda }\left(u\right)={\int }_{\mathrm{\Omega }}\frac{g\left(x\right)}{1-\gamma \left(x\right)}{|u|}^{1-\gamma \left(x\right)}dx\mp \lambda {\int }_{\mathrm{\Omega }}F\left(x,u\left(x\right)\right)dx,$

where $F\left(x,t\right)={\int }_{0}^{t}f\left(x,s\right)ds$.

#### Definition 1.1.

If for all $v\in X$,

${\int }_{\mathrm{\Omega }}{|\mathrm{\Delta }u|}^{p\left(x\right)-2}\mathrm{\Delta }u\mathrm{\Delta }vdx={\int }_{\mathrm{\Omega }}g\left(x\right){|u|}^{-\gamma \left(x\right)}vdx\mp \lambda {\int }_{\mathrm{\Omega }}f\left(x,u\right)vdx,$

then $u\in X$ is called a weak solution of (P-+λ).

The operator ${\mathrm{\Delta }}_{p\left(x\right)}^{2}u:=\mathrm{\Delta }\left({|\mathrm{\Delta }u|}^{p\left(x\right)-2}\mathrm{\Delta }u\right)$ is called the $p\left(x\right)$-biharmonic operator of fourth order, where p is a continuous non-constant function. This differential operator is a natural generalization of the p-biharmonic operator ${\mathrm{\Delta }}_{p}^{2}u:=\mathrm{\Delta }\left({|\mathrm{\Delta }u|}^{p-2}\mathrm{\Delta }u\right)$, where $p>1$ is a real constant. However, the $p\left(x\right)$-biharmonic operator possesses more complicated non-linearity than the p-biharmonic operator, due to the fact that ${\mathrm{\Delta }}_{p\left(x\right)}^{2}$ is not homogeneous. This fact implies some difficulties; for example, we can not use the Lagrange multiplier theorem in many problems involving this operator.

The study of this kind of operators occurs in interesting areas such as electrorheological fluids (see [19]), elastic mechanics (see [25]), stationary thermo-rheological viscous flows of non-Newtonian fluids, image processing (see [6]) and the mathematical description of the processes filtration of an ideal barotropic gas through a porous medium (see [1]).

Problem (P-+λ) is a new variant of $p\left(x\right)$-biharmonic equations due to the singular term and the indefinite one. Note that results for $p\left(x\right)$-Laplace equations with singular non-linearity are rare. Meanwhile, elliptic and singular elliptic equations involving the $p\left(x\right)$-Laplace and the $p\left(x\right)$-biharmonic operators can be found in [1, 2, 5, 8, 9, 10, 13, 14, 17, 18, 21, 20, 22, 23].

Recently, Ayoujil and Amrouss [4] studied the following problem:

(P)

In the case when ${\mathrm{max}}_{x\in \mathrm{\Omega }}q\left(x\right)<{\mathrm{min}}_{x\in \mathrm{\Omega }}p\left(x\right)$, they proved that the energy functional associated to problem (P) has a nontrivial minimum for any positive λ; see [4, Theorem 3.1]. In the case when ${\mathrm{min}}_{x\in \mathrm{\Omega }}q\left(x\right)<{\mathrm{min}}_{x\in \mathrm{\Omega }}p\left(x\right)$ and $q\left(x\right)$ has a subcritical growth, they used Ekeland’s variational principle in order to prove the existence of a continuous family of eigenvalues which lies in a neighborhood of the origin. Finally, when

$\underset{x\in \mathrm{\Omega }}{\mathrm{max}}p\left(x\right)<\underset{x\in \mathrm{\Omega }}{\mathrm{min}}q\left(x\right)\le \underset{x\in \mathrm{\Omega }}{\mathrm{max}}q\left(x\right)<\frac{Np\left(x\right)}{N-2p\left(x\right)},$

they showed (see [4, Theorem 3.8]) that for every $\mathrm{\Lambda }>0$ the energy functional ${\mathrm{\Phi }}_{\lambda }$ corresponding to (P) has a Mountain Pass-type critical point which is nontrivial and nonnegative, and hence $\mathrm{\Lambda }=\left(0,+\mathrm{\infty }\right)$, where Λ is the set of the eigenvalues. The same problem, for $p\left(x\right)=q\left(x\right)$, is studied by Ayoujil and Amrouss in [3]. They established the existence of infinitely many eigenvalues for problem (P) by using an argument based on the Ljusternik–Schnirelmann critical point theory. They showed that $sup\mathrm{\Lambda }=+\mathrm{\infty }$, and they pointed out that only under special conditions, which are somehow connected with a kind of monotony of the function $p\left(x\right)$, one has $inf\mathrm{\Lambda }>0$ (this is in contrast with the case when $p\left(x\right)$ is a constant, where one always has $inf\mathrm{\Lambda }>0$).

Later, Ge, Zhou and Wu [10] considered the following problem:

(P1)

where V is an indefinite weight and λ is a positive real number. They considered different situations concerning the growth rates, and they proved, using the mountain pass lemma and Ekeland’s principle, the existence of a continuous family of eigenvalues. A recent paper concerning this type of problems is [12].

Inspired by the above-mentioned papers, we study problem (P-+λ), which contains a singular term and indefinite many more general terms than the one studied in [10]. In this new situation, we will show the existence of a weak solution for problem (P-+λ). The paper is organized as follows: In Section 2, we recall some definitions concerning variable exponent Lebesgue spaces, ${L}^{p\left(x\right)}\left(\mathrm{\Omega }\right)$, as well as Sobolev spaces, ${W}^{k,p\left(x\right)}\left(\mathrm{\Omega }\right)$. Moreover, some properties of these spaces will also be exhibited to be used later. Our main results are stated in Section 3. The proofs of our results will be presented in Section 4 and Section 5.

## 2 Notations and preliminaries

To study $p\left(x\right)$-biharmonic problems, we need some results on the spaces ${L}^{p\left(x\right)}\left(\mathrm{\Omega }\right)$, ${W}^{1,p\left(x\right)}\left(\mathrm{\Omega }\right)$ and ${W}^{k,p\left(x\right)}\left(\mathrm{\Omega }\right)$ (for details, see [18, 9]) and some properties of the $p\left(x\right)$-biharmonic operator, which will be needed later. Set

Let p be a Lipschitz continuous function on $\overline{\mathrm{\Omega }}$. We set $1<{p}^{-}:={\mathrm{min}}_{x\in \overline{\mathrm{\Omega }}}p\left(x\right)\le {p}^{+}={\mathrm{max}}_{x\in \overline{\mathrm{\Omega }}}p\left(x\right)<\mathrm{\infty }$ and

We recall the following so-called Luxemburg norm on this space defined by the formula

${|u|}_{p\left(x\right)}=inf\left\{\mu >0:{\int }_{\mathrm{\Omega }}{|\frac{u\left(x\right)}{\mu }|}^{p\left(x\right)}dx\le 1\right\}.$

Clearly, when $p\left(x\right)=p$, a positive constant, the space ${L}^{p\left(x\right)}\left(\mathrm{\Omega }\right)$ reduces to the classical Lebesgue space ${L}^{p}\left(\mathrm{\Omega }\right)$, and the norm ${|u|}_{p\left(x\right)}$ reduces to the standard norm

For any positive integer k, as in the constant exponent case, let

${W}^{k,p\left(x\right)}\left(\mathrm{\Omega }\right)=\left\{u\in {L}^{p\left(x\right)}\left(\mathrm{\Omega }\right):{D}^{\alpha }u\in {L}^{p\left(x\right)}\left(\mathrm{\Omega }\right),|\alpha |\le k\right\},$

where $\alpha =\left({\alpha }_{1},\mathrm{\dots },{\alpha }_{N}\right)$ is a multi-index, $|\alpha |={\sum }_{i=1}^{N}{\alpha }_{i}$ and

${D}^{\alpha }u=\frac{{\partial }^{|\alpha |}u}{{\partial }^{{\alpha }_{1}}{x}_{1}\mathrm{\cdots }{\partial }^{{\alpha }_{N}}{x}_{N}}.$

Then ${W}^{k,p\left(x\right)}\left(\mathrm{\Omega }\right)$ is a separable and reflexive Banach space equipped with the norm

${\parallel u\parallel }_{k,p\left(x\right)}=\sum _{|\alpha |\le k}{|{D}^{\alpha }u|}_{p\left(x\right)}.$

Furthermore, ${W}_{0}^{k,p\left(x\right)}\left(\mathrm{\Omega }\right)$ is the closure of ${C}_{0}^{\mathrm{\infty }}\left(\mathrm{\Omega }\right)$ in ${W}^{k,p\left(x\right)}\left(\mathrm{\Omega }\right)$. Let ${L}^{{p}^{\prime }\left(x\right)}\left(\mathrm{\Omega }\right)$ be the conjugate space of ${L}^{p\left(x\right)}\left(\mathrm{\Omega }\right)$ with $\frac{1}{p}+\frac{1}{{p}^{\prime }}=1$. Then the Hölder-type inequality

$|{\int }_{\mathrm{\Omega }}uvdx|\le \left(\frac{1}{{p}^{-}}+\frac{1}{{\left({p}^{\prime }\right)}^{-}}\right){|u|}_{p\left(x\right)}{|v|}_{{p}^{\prime }\left(x\right)},u\in {L}^{p\left(x\right)}\left(\mathrm{\Omega }\right),v\in {L}^{{p}^{\prime }\left(x\right)}\left(\mathrm{\Omega }\right)$(2.1)

holds.

The modular on the space ${L}^{p\left(x\right)}\left(\mathrm{\Omega }\right)$ is the map ${\rho }_{p\left(x\right)}:{L}^{p\left(x\right)}\left(\mathrm{\Omega }\right)\to ℝ$ defined by

${\rho }_{p\left(x\right)}\left(u\right):={\int }_{\mathrm{\Omega }}{|u|}^{p\left(x\right)}dx,$

and it satisfies the following propositions.

#### Proposition 2.1 (see [16]).

For all $u\mathrm{\in }{L}^{p\mathit{}\mathrm{\left(}x\mathrm{\right)}}\mathit{}\mathrm{\left(}\mathrm{\Omega }\mathrm{\right)}$, we have the following assertions:

• (i)

${|u|}_{p\left(x\right)}<1$ (resp. $=1$, $>1$ ) if and only if ${\rho }_{p\left(x\right)}\left(u\right)<1$ (resp. $=1$, $>1$ ).

• (ii)

$\mathrm{min}\left({|u|}_{p\left(x\right)}^{{p}^{-}},{|u|}_{p\left(x\right)}^{{p}^{+}}\right)\le {\rho }_{p\left(x\right)}\left(u\right)\le \mathrm{max}\left({|u|}_{p\left(x\right)}^{{p}^{-}},{|u|}_{p\left(x\right)}^{{p}^{+}}\right)$.

• (iii)

${\rho }_{p\left(x\right)}\left({u}_{n}-u\right)\to 0$ if and only if ${|{u}_{n}-u|}_{p\left(x\right)}\to 0$.

#### Proposition 2.2 (see [7]).

Let p and q be two measurable functions such that $p\mathrm{\in }{L}^{\mathrm{\infty }}\mathit{}\mathrm{\left(}\mathrm{\Omega }\mathrm{\right)}$ and $\mathrm{1}\mathrm{\le }p\mathit{}\mathrm{\left(}x\mathrm{\right)}\mathit{}q\mathit{}\mathrm{\left(}x\mathrm{\right)}\mathrm{\le }\mathrm{\infty }$ for a.e. $x\mathrm{\in }\mathrm{\Omega }$. Let $u\mathrm{\in }{L}^{q\mathit{}\mathrm{\left(}x\mathrm{\right)}}\mathit{}\mathrm{\left(}\mathrm{\Omega }\mathrm{\right)}$, $u\mathrm{\ne }\mathrm{0}$. Then

$\mathrm{min}\left({|u|}_{p\left(x\right)q\left(x\right)}^{{p}^{+}},{|u|}_{p\left(x\right)q\left(x\right)}^{{p}^{-}}\right)\le {|{|u|}^{p\left(x\right)}|}_{q\left(x\right)}\le \mathrm{max}\left({|u|}_{p\left(x\right)q\left(x\right)}^{{p}^{-}},{|u|}_{p\left(x\right)q\left(x\right)}^{{p}^{+}}\right).$

For more details concerning the modular, see [9, 16].

#### Definition 2.3.

Assuming that E and F are Banach spaces, we define the norm on the space $X:=E\cap F$ as ${\parallel u\parallel }_{X}={\parallel u\parallel }_{E}+{\parallel u\parallel }_{F}$ .

In order to discuss problems (P-+λ), we need some theories on the space $X:={W}_{0}^{1,p\left(x\right)}\left(\mathrm{\Omega }\right)\cap {W}^{2,p\left(x\right)}\left(\mathrm{\Omega }\right)$. From Definition 2.3 we know that for any $u\in X$,

$\parallel u\parallel ={\parallel u\parallel }_{1,p\left(x\right)}+{\parallel u\parallel }_{2,p\left(x\right)},$

and thus

$\parallel u\parallel ={|u|}_{p\left(x\right)}+{|\nabla u|}_{p\left(x\right)}+\sum _{|\alpha |=2}{|{D}^{\alpha }u|}_{p\left(x\right)}.$

Zang and Fu [24], proved the equivalence of the norms, and they even proved that the norm ${|\mathrm{\Delta }u|}_{p\left(x\right)}$ is equivalent to the norm $\parallel u\parallel$ (see [24, Theorem 4.4]). Let us choose on X the norm defined by $\parallel u\parallel ={|\mathrm{\Delta }u|}_{p\left(x\right)}$. Note that $\left(X,\parallel \cdot \parallel \right)$ is also a separable and reflexive Banach space and that the modular is defined as ${\rho }_{p\left(x\right)}:X\to ℝ$ by ${\rho }_{p\left(x\right)}\left(\mathrm{\Delta }u\right)={\int }_{\mathrm{\Omega }}|\mathrm{\Delta }u|dx$ and satisfies the same properties as in Proposition 2.1. Hereafter, let

${p}^{*}\left(x\right)=\left\{\begin{array}{cc}\frac{Np\left(x\right)}{N-2p\left(x\right)},\hfill & p\left(x\right)<\frac{N}{2},\hfill \\ +\mathrm{\infty },\hfill & p\left(x\right)\ge \frac{N}{2}.\hfill \end{array}$

#### Remark 2.4.

If $q\in {C}^{+}\left(\overline{\mathrm{\Omega }}\right)$ and $q\left(x\right)<{p}^{*}\left(x\right)$ for any $x\in \mathrm{\Omega }$, by [4, Theorem 3.2] we deduce that X is continuously and compactly embedded in ${L}^{q\left(x\right)}\left(\mathrm{\Omega }\right)$.

Throughout this paper, the letters $k,c,C,{C}_{i},i=1,2,\mathrm{\dots }$ , denote positive constants which may change from line to line.

## 3 Hypotheses and main results

Let us impose the following hypotheses on the non-linearity $f:\overline{\mathrm{\Omega }}×ℝ\to ℝ$:

• (f1)

f is a ${C}^{1}$ function such that $f\left(x,0\right)=0$.

• (f2)

There exists ${\mathrm{\Omega }}_{1}⋐\mathrm{\Omega }$ with $|{\mathrm{\Omega }}_{1}|>0$, and a nonnegative function ${h}_{1}$ on ${\mathrm{\Omega }}_{1}$ such that ${h}_{1}\in {L}^{{s}_{1}\left(x\right)}\left(\mathrm{\Omega }\right)$ with

• (f3)

There exists a positive function h on Ω such that $h\in {L}^{s\left(x\right)}\left(\mathrm{\Omega }\right)$ and

where $s,{s}_{1},r$ and ${r}_{1}\in C\left(\overline{\mathrm{\Omega }}\right)$ are such that $1<\mathrm{max}\left\{r\left(x\right),{r}_{1}\left(x\right)\right\} for all $x\in \mathrm{\Omega }$.

• (f4)

There exists $A>0$ such that

• (f5)

$f\left(x,t\right)\le Ch\left(x\right){|t|}^{r\left(x\right)-2}t$ for all $t\in ℝ$ and all $x\in \overline{\mathrm{\Omega }}$, where C is a positive constant, $h\in {L}^{s\left(x\right)}\left(\mathrm{\Omega }\right)$ and $s,r\in C\left(\overline{\mathrm{\Omega }}\right)$ are such that for all $x\in \overline{\mathrm{\Omega }}$ we have $1.

• (f6)

There exists ${\mathrm{\Omega }}_{1}⋐\mathrm{\Omega }$ with $|{\mathrm{\Omega }}_{1}|>0$ such that $f\left(x,t\right),h\left(x\right)>0$ in ${\mathrm{\Omega }}_{1}$.

Some remarks regarding the hypotheses are in order.

#### Remark 3.1.

Under assumptions (f3) and (f4), we have the following assertions.

• (i)

$\left(1-\gamma \left(x\right)\right)p\left(x\right) for any $x\in \overline{\mathrm{\Omega }}$, so the injection of $X↪{L}^{\left(1-\gamma \left(x\right)\right)p\left(x\right)}\left(\mathrm{\Omega }\right)$ is compact and continuous.

• (ii)

${s}^{\prime }\left(x\right)r\left(x\right)<{p}^{*}\left(x\right)$ for any $x\in \overline{\mathrm{\Omega }}$, where $\frac{1}{s\left(x\right)}+\frac{1}{{s}^{\prime }\left(x\right)}=1$, so $X↪{L}^{{s}^{\prime }\left(x\right)r\left(x\right)}\left(\mathrm{\Omega }\right)$ is compact and continuous.

• (iii)

${s}_{1}^{\prime }\left(x\right){r}_{1}\left(x\right)<{p}^{*}\left(x\right)$ for any $x\in \overline{\mathrm{\Omega }}$, where $\frac{1}{{s}_{1}\left(x\right)}+\frac{1}{{s}_{1}^{\prime }\left(x\right)}=1$, so $X↪{L}^{{s}_{1}^{\prime }\left(x\right){r}_{1}\left(x\right)}\left(\mathrm{\Omega }\right)$ is compact and continuous.

• (iv)

$X↪{L}^{{p}^{*}\left(x\right)}\left(\mathrm{\Omega }\right)$ is continuous.

Moreover, under conditions (f5) and (f6), we remark the following.

#### Remark 3.2.

• (i)

There exists $K>0$ such that $\eta f\left(x,\eta \right)\le Kr\left(x\right)F\left(x,\eta \right)$ for all $x\in {\mathrm{\Omega }}_{1},\eta \in {ℝ}_{+}$.

• (ii)

Due to condition (f5), there exists ${C}^{\prime }>0$ such that

• (iii)

Conditions (f5) and (f6) assure that $F\left(x,\eta \right)>0$ for all $x\in {\mathrm{\Omega }}_{1},\eta \in ℝ$.

• (iv)

Put $f\left(x,t\right)=Ch\left(x\right){|t|}^{r\left(x\right)-2}t,x\in \mathrm{\Omega },t\in ℝ$. Then the first condition in the remark is satisfied.

Here we state our main results asserted in the following two theorems.

#### Theorem 3.3.

Assume that hypotheses (f1), (f2), (f3) and (f4) are fulfilled. Then for all $\lambda \mathrm{>}\mathrm{0}$ problem (P${}_{-\lambda }$) has at least one nontrivial weak solution with negative energy.

#### Theorem 3.4.

Assume that hypotheses (f5) and (f6) are fulfilled. Then for all $\lambda \mathrm{>}\mathrm{0}$ problem (P${}_{+\lambda }$) has at least one nontrivial weak solution with negative energy.

## 4 Proof of Theorem 3.3

The study of the existence of solutions to problem (P${}_{-\lambda }$) is done by looking for critical points to the functional ${I}_{-\lambda }:X\to ℝ$ defined by

${I}_{-\lambda }\left({u}_{\lambda }\right)={\int }_{\mathrm{\Omega }}\frac{{|\mathrm{\Delta }{u}_{\lambda }|}^{p\left(x\right)}}{p\left(x\right)}dx-{\int }_{\mathrm{\Omega }}\frac{g\left(x\right){u}_{\lambda }^{1-\gamma \left(x\right)}}{1-\gamma \left(x\right)}dx+\lambda {\int }_{\mathrm{\Omega }}F\left(x,{u}_{\lambda }\right)dx$

in the Sobolev space X. The proof of the Theorem 3.3 is organized in several lemmas. Firstly, under Remark 3.1 one has

(4.1)

and

Now, we are in a position to show that ${I}_{-\lambda }$ possesses a nontrivial global minimum point in X.

#### Lemma 4.1.

Under assumptions (f2), (f3) and (f4), the functional ${I}_{\mathrm{-}\lambda }$ is coercive on X.

#### Proof.

First, we recall that in view of assumptions (f3), (f4), inequality (4.1), Remark 3.1 and Proposition 2.1 one has for any ${u}_{\lambda }\in X$ with $\parallel {u}_{\lambda }\parallel >\mathrm{max}\left(1,A\right)$,

${I}_{-\lambda }\left({u}_{\lambda }\right)\ge {\int }_{\mathrm{\Omega }}\frac{1}{p\left(x\right)}{|\mathrm{\Delta }{u}_{\lambda }|}^{p\left(x\right)}dx-{C}_{4}{|g|}_{\frac{{p}^{*}\left(x\right)}{{p}^{*}\left(x\right)+\gamma \left(x\right)-1}}{|{|{u}_{\lambda }|}^{1-\gamma \left(x\right)}|}_{{p}^{*}\left(x\right)}+\lambda {\int }_{\mathrm{\Omega }}F\left(x,{u}_{\lambda }\right)dx$$\ge {\int }_{\mathrm{\Omega }}\frac{1}{p\left(x\right)}{|\mathrm{\Delta }{u}_{\lambda }|}^{p\left(x\right)}dx-{C}_{4}{|g|}_{\frac{{p}^{*}\left(x\right)}{{p}^{*}\left(x\right)+\gamma \left(x\right)-1}}{|{|{u}_{\lambda }|}^{1-\gamma \left(x\right)}|}_{{p}^{\left(}x\right)}$$\ge {\int }_{\mathrm{\Omega }}\frac{1}{p\left(x\right)}{|\mathrm{\Delta }{u}_{\lambda }|}^{p\left(x\right)}dx-{C}_{4}{|g|}_{\frac{{p}^{*}\left(x\right)}{{p}^{*}\left(x\right)+\gamma \left(x\right)-1}}{\parallel {u}_{\lambda }\parallel }^{1-{\gamma }^{-}}$$\ge \frac{1}{{p}^{+}}{\rho }_{{p}_{\left(x\right)}}\left(\mathrm{\Delta }{u}_{\lambda }\right)-{C}_{4}{|g|}_{\frac{{p}^{*}\left(x\right)}{{p}^{*}\left(x\right)+\gamma \left(x\right)-1}}{\parallel {u}_{\lambda }\parallel }^{1-{\gamma }^{-}}$$\ge \frac{1}{{p}^{+}}{\parallel {u}_{\lambda }\parallel }^{{p}^{-}}-{C}_{4}{|g|}_{\frac{{p}^{*}\left(x\right)}{{p}^{*}\left(x\right)+\gamma \left(x\right)-1}}{\parallel {u}_{\lambda }\parallel }^{1-{\gamma }^{-}}.$

Since $1-{\gamma }^{-}<{p}^{-}$, we infer that ${I}_{-\lambda }\left({u}_{\lambda }\right)\to \mathrm{\infty }$ as $\parallel {u}_{\lambda }\parallel \to \mathrm{\infty }$; in other words, ${I}_{-\lambda }$ is coercive on X. The proof of Lemma 4.1 is now completed. ∎

#### Lemma 4.2.

Suppose assumptions (f2) and (f3) are fulfilled. Then there exists $\phi \mathrm{\in }X$ such that $\phi \mathrm{\ge }\mathrm{0}$, $\phi \mathrm{\ne }\mathrm{0}$ and ${I}_{\mathrm{-}\lambda }\mathit{}\mathrm{\left(}t\mathit{}\phi \mathrm{\right)}\mathrm{<}\mathrm{0}$ for $t\mathrm{>}\mathrm{0}$ small enough.

#### Proof.

Let $\phi \in {C}_{0}^{\mathrm{\infty }}\left(\mathrm{\Omega }\right)$ such that $\mathrm{supp}\left(\phi \right)\subset {\mathrm{\Omega }}_{1}⋐\mathrm{\Omega },\phi =1$ in a subset ${\mathrm{\Omega }}^{\prime }\subset \mathrm{supp}\left(\phi \right)$ and $0\le \phi \le 1$ in ${\mathrm{\Omega }}_{1}$. Using assertions on the functions g and F and assumption (f2), we have

${I}_{-\lambda }\left(t\phi \right)={\int }_{\mathrm{\Omega }}\frac{1}{p\left(x\right)}{|\mathrm{\Delta }t\phi |}^{p\left(x\right)}dx-{\int }_{\mathrm{\Omega }}\frac{1}{1-\gamma \left(x\right)}g\left(x\right){|t\phi |}^{1-\gamma \left(x\right)}dx+\lambda {\int }_{\mathrm{\Omega }}F\left(x,t\phi \right)dx$$\le \frac{{t}^{{p}^{-}}}{{p}^{-}}{\rho }_{p\left(x\right)}\left(\mathrm{\Delta }\phi \right)-\frac{{t}^{1-{\gamma }^{-}}}{1-{\gamma }^{-}}{\int }_{\mathrm{\Omega }}g\left(x\right){|\phi |}^{1-\gamma \left(x\right)}dx+\lambda {C}_{1}{t}^{{r}_{1}^{-}}{\int }_{{\mathrm{\Omega }}_{1}}{h}_{1}\left(x\right){|\phi |}^{{r}_{1}\left(x\right)}dx$$\le {t}^{{r}_{1}^{-}}\left[\frac{1}{{p}^{-}}{\rho }_{p\left(x\right)}\left(\mathrm{\Delta }\phi \right)+\lambda {C}_{1}{\int }_{{\mathrm{\Omega }}_{1}}{h}_{1}\left(x\right){|\phi |}^{{r}_{1}\left(x\right)}dx\right]-\frac{{t}^{1-{\gamma }^{-}}}{1-{\gamma }^{-}}{\int }_{\mathrm{\Omega }}g\left(x\right){|\phi |}^{1-\gamma \left(x\right)}dx,$

so

with

$0<\psi <\mathrm{min}\left\{1,\frac{\frac{1}{1-{\gamma }^{-}}{\int }_{\mathrm{\Omega }}g\left(x\right){|\phi |}^{1-\gamma \left(x\right)}dx}{\frac{1}{{p}^{-}}{\rho }_{p\left(x\right)}\left(\mathrm{\Delta }\phi \right)+\lambda {C}_{1}{\int }_{{\mathrm{\Omega }}_{1}}{h}_{1}\left(x\right){|\phi |}^{{r}_{1}\left(x\right)}dx}\right\}.$

Finally, we point out, using the hypothesis on φ and the definition of the modular on X, that

$\frac{1}{{p}^{-}}{\rho }_{p\left(x\right)}\left(\mathrm{\Delta }\phi \right)+\lambda C{\int }_{{\mathrm{\Omega }}_{1}}{h}_{1}\left(x\right){|\phi |}^{{r}_{1}\left(x\right)}dx>0.$

In fact, if

$\frac{1}{{p}^{-}}{\rho }_{p\left(x\right)}\left(\mathrm{\Delta }\phi \right)+\lambda {C}_{1}{\int }_{{\mathrm{\Omega }}_{1}}{h}_{1}\left(x\right){|\phi |}^{{r}_{1}\left(x\right)}dx=0,$

then ${\rho }_{p\left(x\right)}\left(\mathrm{\Delta }\phi \right)=0$, and consequently $\parallel \phi \parallel =0$, which contradicts the choice of φ and gives the proof of Lemma 4.2. ∎

In the sequel, we put ${m}_{\lambda }={inf}_{{u}_{\lambda }\in X}{I}_{-\lambda }\left({u}_{\lambda }\right)$. Then we have the following lemma.

#### Lemma 4.3.

Let $\lambda \mathrm{\ge }\mathrm{0}$, $\gamma \mathrm{\in }C\mathit{}\mathrm{\left(}\overline{\mathrm{\Omega }}\mathrm{,}\mathrm{\left(}\mathrm{0}\mathrm{,}\mathrm{1}\mathrm{\right)}\mathrm{\right)}$,

$g\in {L}^{\frac{{p}^{*}\left(x\right)}{{p}^{*}\left(x\right)+\gamma \left(x\right)-1}}\left(\mathrm{\Omega }\right)$

with $g\mathit{}\mathrm{\left(}x\mathrm{\right)}\mathrm{>}\mathrm{0}$ for almost every $x\mathrm{\in }\mathrm{\Omega }$, and assume that hypothesis (f1), (f2), (f3) and (f4) are fulfilled. Then ${I}_{\mathrm{-}\lambda }$ reaches its global minimizer in X, that is, there exists ${u}_{\lambda }\mathrm{\in }X$ such that ${I}_{\mathrm{-}\lambda }\mathit{}\mathrm{\left(}{u}_{\lambda }\mathrm{\right)}\mathrm{=}{m}_{\lambda }\mathrm{<}\mathrm{0}$.

#### Proof.

Let $\left\{{u}_{n}\right\}$ be a minimizing sequence, that is to say ${I}_{-\lambda }\left({u}_{n}\right)\to {m}_{\lambda }$. Suppose $\left\{{u}_{n}\right\}$ is not bounded, so $\parallel {u}_{n}\parallel \to +\mathrm{\infty }$ as $n\to +\mathrm{\infty }$. Since ${I}_{\lambda }$ is coercive, we have

This contradicts the fact that $\left\{{u}_{n}\right\}$ is a minimizing sequence, so $\left\{{u}_{n}\right\}$ is bounded in X, and therefore, up to a subsequence, there exists ${u}_{\lambda }\in X$ such that

${u}_{n}⇀{u}_{\lambda }$${u}_{n}\to {u}_{\lambda }$${u}_{n}\left(x\right)\to {u}_{\lambda }\left(x\right)$

Since $J:X\to ℝ$ is sequentially weakly lower semi-continuous (see [12]), we have

${\int }_{\mathrm{\Omega }}\frac{1}{p\left(x\right)}{|\mathrm{\Delta }{u}_{\lambda }|}^{p\left(x\right)}\le \underset{n\to +\mathrm{\infty }}{lim inf}{\int }_{\mathrm{\Omega }}\frac{1}{p\left(x\right)}{|\mathrm{\Delta }{u}_{n}|}^{p\left(x\right)}.$(4.2)

On the other hand, by Vital’s theorem (see [16, p. 113]), we can claim that

$\underset{n\to \mathrm{\infty }}{lim}{\int }_{\mathrm{\Omega }}g\left(x\right){|{u}_{n}|}^{1-\gamma \left(x\right)}dx={\int }_{\mathrm{\Omega }}g\left(x\right){|{u}_{\lambda }|}^{1-\gamma \left(x\right)}dx.$(4.3)

Indeed, we only need to prove that

$\left\{{\int }_{\mathrm{\Omega }}g\left(x\right){|{u}_{n}|}^{1-\gamma \left(x\right)}dx:n\in ℕ\right\}$

is equi-absolutely-continuous. Note that $\left\{{u}_{n}\right\}$ is bounded in X, so Remark 3.1 implies that $\left\{{u}_{n}\right\}$ is bounded in ${L}^{{p}^{*}\left(x\right)}\left(\mathrm{\Omega }\right)$. For every $\epsilon >0$, using Proposition 2.1 and the absolutely-continuity of

${\int }_{\mathrm{\Omega }}{|g\left(x\right)|}^{\frac{{p}^{*}\left(x\right)}{{p}^{*}\left(x\right)+\gamma \left(x\right)-1}}dx,$

there exist $\zeta ,\xi >0$ such that

Consequently, by the Hölder inequality and Proposition 2.1 one has

${\int }_{\mathrm{\Omega }}|g\left(x\right)|{|{u}_{n}|}^{1-\gamma \left(x\right)}dx\le {|g|}_{\frac{{p}^{*}\left(x\right)}{{p}^{*}\left(x\right)+\gamma \left(x\right)-1}}{|{|{u}_{n}|}^{1-\gamma \left(x\right)}|}_{{p}^{*}\left(x\right)}\le {|g|}_{\frac{{p}^{*}\left(x\right)}{{p}^{*}\left(x\right)+\gamma \left(x\right)-1}}{|{u}_{n}|}_{\left(1-\gamma \left(x\right)\right){p}^{*}\left(x\right)}^{k}.$

Since $\left(1-\gamma \left(x\right)\right){p}^{*}\left(x\right)<{p}^{*}\left(x\right)$, we have

${|{u}_{n}|}_{\left(1-\gamma \left(x\right)\right){p}^{*}\left(x\right)}\le {C}_{7}{|{u}_{n}|}_{{p}^{*}\left(x\right)},$

so

${\int }_{\mathrm{\Omega }}|g\left(x\right)|{|{u}_{n}|}^{1-\gamma \left(x\right)}dx\le {|g|}_{\frac{{p}^{*}\left(x\right)}{{p}^{*}\left(x\right)+\gamma \left(x\right)-1}}{|{|{u}_{n}|}^{1-\gamma \left(x\right)}|}_{{p}^{*}\left(x\right)}\le {|g|}_{\frac{{p}^{*}\left(x\right)}{{p}^{*}\left(x\right)+\gamma \left(x\right)-1}}{|{u}_{n}|}_{\left(1-\gamma \left(x\right)\right){p}^{*}\left(x\right)}^{k}<\epsilon {C}_{7}^{k}{|{u}_{n}|}_{{p}^{*}\left(x\right)}^{k}.$

Since ${|{u}_{n}|}_{{p}^{*}\left(x\right)}$ is bounded, claim (4.3) is valid.

In what follows, we remark, using assumptions (f2) and (f3), that for all $\epsilon >0$ there exists ${C}_{\epsilon }$ such that

$|F\left(x,{u}_{\lambda }\left(x\right)\right)|\le \epsilon \frac{{C}_{1}}{{r}_{1}^{-}}|{h}_{1}\left(x\right)|{|{u}_{\lambda }|}^{{r}_{1}\left(x\right)}+{C}_{\epsilon }\frac{C}{{r}^{-}}|h\left(x\right)|{|{u}_{\lambda }|}^{r\left(x\right)}.$

Then by the Hölder inequality one has

${\int }_{\mathrm{\Omega }}|F\left(x,{u}_{\lambda }\left(x\right)|\le \epsilon \frac{{C}_{1}}{{r}_{1}^{-}}{|{h}_{1}|}_{{s}_{1}\left(x\right)}{|{|{u}_{\lambda }|}^{{r}_{1}\left(x\right)}|}_{{s}_{1}^{\prime }\left(x\right)}+{C}_{\epsilon }\frac{C}{{r}^{-}}{|h|}_{s\left(x\right)}{|{|{u}_{\lambda }|}^{r\left(x\right)}|}_{{s}^{\prime }\left(x\right)}.$

Besides, if ${u}_{n}⇀{u}_{\lambda }$ in X, then we have strong convergence in ${L}^{{s}^{\prime }\left(x\right)r\left(x\right)}\left(\mathrm{\Omega }\right)$ and ${L}^{s_{1}{}^{\prime }\left(x\right){r}_{1}\left(x\right)}\left(\mathrm{\Omega }\right)$. So the Lebesgue dominated convergence theorem and Proposition 2.2 enable us to state the following assertion: If

$u↦\lambda {\int }_{\mathrm{\Omega }}F\left(x,{u}_{\lambda }\left(x\right)\right)dx$

is weakly continuous, then

$\underset{n\to +\mathrm{\infty }}{lim}{\mathrm{\Phi }}_{-\lambda }\left({u}_{n}\right)={\mathrm{\Phi }}_{-\lambda }\left({u}_{\lambda }\right).$(4.4)

Using (4.2), (4.3) and (4.4), we deduce that ${I}_{-\lambda }$ is weakly lower semi-continuous, and consequently

${m}_{\lambda }\le {I}_{-\lambda }\left({u}_{\lambda }\right)\le \underset{n\to +\mathrm{\infty }}{lim inf}{I}_{-\lambda }\left({u}_{n}\right)={m}_{\lambda }.$

The proof of Lemma 4.3 is now completed. ∎

#### Proof of Theorem 3.3.

Now, let us show that the weak limit ${u}_{\lambda }$ is a weak solution of problem (P${}_{-\lambda }$) if $\lambda >0$ is sufficiently large. Firstly, observe that ${I}_{-\lambda }\left(0\right)=0$. So, in order to prove that the solution is nontrivial, it suffices to prove that there exists ${\lambda }_{\ast }>0$ such that

For this purpose, we consider the variational problem with constraints

(4.5)

and define

${\mathrm{\Lambda }}_{*}:=inf\left\{\lambda >0:\text{(P-}\lambda \text{) admits a nontrivial weak solution}\right\}.$

From above we have

Therefore, the above remarks show that ${\lambda }_{\ast }\ge {\mathrm{\Lambda }}_{*}$ and that problem (P${}_{-\lambda }$) has a solution for all $\lambda >{\lambda }_{\ast }$.

We now argue that problem (P${}_{-\lambda }$) has a solution for all $\lambda >{\mathrm{\Lambda }}_{*}$. Fixing $\lambda >{\mathrm{\Lambda }}_{*}$, by the definition of ${\mathrm{\Lambda }}_{*}$ we can take $\mu \in \left({\mathrm{\Lambda }}_{*},\lambda \right)$ such that ${I}_{-\mu }$ has a nontrivial critical point ${u}_{\mu }\in X$. Since $\mu <\lambda$, we obtain that ${u}_{\mu }$ is a sub-solution of problem (P${}_{-\lambda }$). We now want to construct a super-solution of problem (P${}_{-\lambda }$) which dominates ${u}_{\mu }$. For this purpose, we introduce the constrained minimization problem

By using the previous arguments to treat (4.5), follows that the above minimization problem has a solution ${u}_{\lambda }>{u}_{\mu }$. Moreover, ${u}_{\lambda }$ is also a weak solution of problem (P${}_{-\lambda }$) for all $\lambda >{\mathrm{\Lambda }}_{*}$. With the arguments developed in [15], we deduce that problem (P${}_{-\lambda }$) has a solution if $\lambda ={\mathrm{\Lambda }}_{*}$.

Now, it remains to show that $\mathrm{\Delta }{u}_{\lambda }=0$ on $\partial \mathrm{\Omega }$. Due to the above arguments, one has

(4.6)

where

$m\left(x\right)=g\left(x\right){u}_{\lambda }^{-\gamma \left(x\right)}-\lambda f\left(x,{u}_{\lambda }\right).$

Relation (4.6) implies that

(4.7)

Let ζ be the unique solution of the problem

Relation (4.7) yields

Using the Green formula, we have

${\int }_{\mathrm{\Omega }}\left(\mathrm{\Delta }\zeta \right)vdx={\int }_{\mathrm{\Omega }}\zeta \mathrm{\Delta }vdx.$

Therefore,

(4.8)

On the other hand, for all $\stackrel{~}{{u}_{\lambda }}\in {C}_{0}^{\mathrm{\infty }}\left(\mathrm{\Omega }\right)$ there exists a unique $v\in {C}_{0}^{\mathrm{\infty }}\left(\mathrm{\Omega }\right)$ such that $\mathrm{\Delta }v=\stackrel{~}{{u}_{\lambda }}$ in Ω. Thus, relation (4.8) can be rewritten as

Applying the fundamental lemma of the calculus of variations, we deduce that

Since $\zeta =0$ on $\partial \mathrm{\Omega }$, we conclude that $\mathrm{\Delta }{u}_{\lambda }=0$ on $\partial \mathrm{\Omega }$. Thus, ${u}_{\lambda }$ is a nontrivial weak solution of problem (P${}_{-\lambda }$) such that $\mathrm{\Delta }{u}_{\lambda }=0$. This completes the proof of Theorem 3.3. ∎

## 5 Proof of Theorem 3.4

The proof of Theorem 3.4 is organized in several lemmas. Firstly, we show the existence of a local minimum for ${I}_{+\lambda }$ in a small neighborhood of the origin in X.

#### Lemma 5.1.

Under assumption (f5), the functional ${I}_{\mathrm{+}\lambda }$ is coercive on X.

#### Proof.

Using Remark 3.1, inequality (2.1) and Proposition 2.1, we obtain that for any ${v}_{\lambda }\in X$ with $\parallel {v}_{\lambda }\parallel >1$,

${I}_{+\lambda }\left({v}_{\lambda }\right)={\int }_{\mathrm{\Omega }}\frac{1}{p\left(x\right)}{|\mathrm{\Delta }{v}_{\lambda }|}^{p\left(x\right)}dx-\frac{1}{1-\gamma \left(x\right)}{\int }_{\mathrm{\Omega }}g\left(x\right){|{v}_{\lambda }|}^{1-\gamma \left(x\right)}dx-\lambda {\int }_{\mathrm{\Omega }}F\left(x,{v}_{\lambda }\left(x\right)\right)dx$$\ge \frac{1}{{p}^{+}}{\rho }_{p\left(x\right)}\left(\mathrm{\Delta }{v}_{\lambda }\right)-\frac{1}{1-{\gamma }^{+}}{\int }_{\mathrm{\Omega }}g\left(x\right){|{v}_{\lambda }|}^{1-\gamma \left(x\right)}dx-\lambda {\int }_{\mathrm{\Omega }}F\left(x,{v}_{\lambda }\left(x\right)\right)dx$$\ge \frac{1}{{p}^{+}}{\parallel {v}_{\lambda }\parallel }^{{p}^{-}}-\frac{1}{1-{\gamma }^{+}}{|g|}_{\frac{{p}^{*}\left(x\right)}{{p}^{*}\left(x\right)+\gamma \left(x\right)-1}}{|{|{v}_{\lambda }|}^{1-\gamma \left(x\right)}|}_{{p}^{*}\left(x\right)}-{C}^{\prime }\lambda {\int }_{\mathrm{\Omega }}h\left(x\right){|{v}_{\lambda }|}^{r\left(x\right)}dx$$\ge \frac{1}{{p}^{+}}{\parallel {v}_{\lambda }\parallel }^{{p}^{-}}-\frac{1}{1-{\gamma }^{+}}{|g|}_{\frac{{p}^{*}\left(x\right)}{{p}^{*}\left(x\right)+\gamma \left(x\right)-1}}{|{|{v}_{\lambda }|}^{1-\gamma \left(x\right)}|}_{p\left(x\right)}-{C}^{\prime }\lambda {|h|}_{s\left(x\right)}{|{|{v}_{\lambda }|}^{r\left(x\right)}|}_{{s}^{\prime }\left(x\right)}\mathrm{d}x$$\ge \frac{1}{{p}^{+}}{\parallel {v}_{\lambda }\parallel }^{{p}^{-}}-\frac{1}{1-{\gamma }^{+}}{|g|}_{\frac{{p}^{*}\left(x\right)}{{p}^{*}\left(x\right)+\gamma \left(x\right)-1}}\mathrm{min}\left({|{v}_{\lambda }|}_{\left(1-\gamma \left(x\right)\right)p\left(x\right)}^{1-{\gamma }^{+}},{|{v}_{\lambda }|}_{\left(1-\gamma \left(x\right)\right)p\left(x\right)}^{1-{\gamma }^{-}}\right)$$-{C}^{\prime }\lambda {|h|}_{s\left(x\right)}\mathrm{min}\left({|{v}_{\lambda }|}_{{s}^{\prime }\left(x\right)r\left(x\right)}^{{r}^{+}},{|{v}_{\lambda }|}_{{s}^{\prime }\left(x\right)r\left(x\right)}^{{r}^{-}}\right)$$\ge \frac{1}{{p}^{+}}{\parallel {v}_{\lambda }\parallel }^{{p}^{-}}-\frac{1}{1-{\gamma }^{+}}{|g|}_{\frac{{p}^{*}\left(x\right)}{{p}^{*}\left(x\right)+\gamma -1}}\mathrm{min}\left({C}_{1}^{1-{\gamma }^{+}}{\parallel {v}_{\lambda }\parallel }^{1-{\gamma }^{+}},{C}_{1}^{1-{\gamma }^{-}}{\parallel {v}_{\lambda }\parallel }^{1-{\gamma }^{-}}\right)$$-{C}^{\prime }\lambda {|h|}_{s\left(x\right)}\mathrm{min}\left({C}_{2}^{{r}^{+}}{\parallel {v}_{\lambda }\parallel }^{{r}^{+}},{C}_{2}^{{r}^{-}}{\parallel {v}_{\lambda }\parallel }^{{r}^{-}}\right).$

Since $1-{\gamma }^{-}<{r}^{+}<{p}^{-}$, we infer that ${I}_{\lambda }\left({v}_{\lambda }\right)\to \mathrm{\infty }$ as $\parallel {v}_{\lambda }\parallel \to \mathrm{\infty }$ and ${I}_{+\lambda }$ is coercive on X. This ends the proof of Lemma 5.1. ∎

#### Lemma 5.2.

Under assumptions (f5) and (f6), there exists $\phi \mathrm{\in }X$ such that $\phi \mathrm{\ge }\mathrm{0}\mathrm{,}\phi \mathrm{\ne }\mathrm{0}$ and ${I}_{\mathrm{+}\lambda }\mathit{}\mathrm{\left(}t\mathit{}\phi \mathrm{\right)}\mathrm{<}\mathrm{0}$ for $t\mathrm{>}\mathrm{0}$ small enough.

#### Proof.

Let $\phi \in {C}_{0}^{\mathrm{\infty }}\left(\mathrm{\Omega }\right)$ such that $\mathrm{supp}\left(\phi \right)\subset {\mathrm{\Omega }}_{1}⋐\mathrm{\Omega },\phi =1$ in a subset ${\mathrm{\Omega }}^{\prime }\subset \mathrm{supp}\left(\phi \right)$ and $0\le \phi \le 1$ in ${\mathrm{\Omega }}_{1}$. Using assertions on the functions g and F, we have

${I}_{+\lambda }\left(t\phi \right)={\int }_{\mathrm{\Omega }}\frac{1}{p\left(x\right)}{|\mathrm{\Delta }t\phi |}^{p\left(x\right)}dx-{\int }_{\mathrm{\Omega }}\frac{1}{1-\gamma \left(x\right)}g\left(x\right){|t\phi |}^{1-\gamma \left(x\right)}dx-\lambda {\int }_{\mathrm{\Omega }}F\left(x,t\phi \right)dx$$\le {\int }_{\mathrm{\Omega }}\frac{{t}^{p\left(x\right)}}{{p}^{-}}{|\mathrm{\Delta }\phi |}^{p\left(x\right)}dx-{\int }_{\mathrm{\Omega }}\frac{g\left(x\right)}{1-\gamma \left(x\right)}{|t\phi |}^{1-\gamma \left(x\right)}dx-\lambda {\int }_{\mathrm{\Omega }}F\left(x,t\phi \right)dx$$\le \frac{{t}^{{p}^{-}}}{{p}^{-}}{\int }_{\mathrm{\Omega }}{|\mathrm{\Delta }\phi |}^{p\left(x\right)}dx-{t}^{1-{\gamma }^{-}}{\int }_{\mathrm{\Omega }}\frac{g\left(x\right){|\phi |}^{1-\gamma \left(x\right)}}{1-\gamma \left(x\right)}dx-\lambda {\int }_{\mathrm{\Omega }1}F\left(x,t\phi \right)dx.$

Then

${I}_{+\lambda }\left(t\phi \right)\le \frac{{t}^{{p}^{-}}}{{p}^{-}}{\int }_{\mathrm{\Omega }}{|\mathrm{\Delta }\phi |}^{p\left(x\right)}dx-\frac{{t}^{1-{\gamma }^{-}}}{1-{\gamma }^{-}}{\int }_{\mathrm{\Omega }}g\left(x\right){|\phi |}^{1-\gamma \left(x\right)}dx.$

Since ${p}^{-}>1-{\gamma }^{-}$, we have ${I}_{+\lambda }\left(t\phi \right)<0$ for $t<{\psi }^{1/\left({p}^{-}-\left(1-{\gamma }^{-}\right)\right)}$ with

$0<\psi <\mathrm{min}\left\{1,\frac{\frac{{p}^{-}}{1-{\gamma }^{-}}{\int }_{\mathrm{\Omega }}g\left(x\right){|\phi |}^{1-\gamma \left(x\right)}dx}{{\rho }_{p\left(x\right)}\left(\mathrm{\Delta }\phi \right)}\right\}.$

Finally, we point out that ${\rho }_{p\left(x\right)}\left(\mathrm{\Delta }\phi \right)>0$. In fact, if ${\rho }_{p\left(x\right)}\left(\mathrm{\Delta }\phi \right)=0$, then $\parallel \phi \parallel =0$, and consequently $\phi =0$ in Ω, which is a contradiction. ∎

In the sequel, put ${m}_{\lambda }^{1}={inf}_{{v}_{\lambda }\in X}{I}_{+\lambda }\left({v}_{\lambda }\right)$. As a last proposition, we have the following.

#### Lemma 5.3.

Let $\lambda \mathrm{\ge }\mathrm{0}$, $\gamma \mathrm{\in }C\mathit{}\mathrm{\left(}\overline{\mathrm{\Omega }}\mathrm{,}\mathrm{\left(}\mathrm{0}\mathrm{,}\mathrm{1}\mathrm{\right)}\mathrm{\right)}$,

$g\in {L}^{\frac{{p}^{*}\left(x\right)}{{p}^{*}\left(x\right)+\gamma \left(x\right)-1}}\left(\mathrm{\Omega }\right)$

with $g\mathit{}\mathrm{\left(}x\mathrm{\right)}\mathrm{>}\mathrm{0}$ for almost every $x\mathrm{\in }\mathrm{\Omega }$, and assume that assertions (f5) and (f6) hold. Then ${I}_{\mathrm{+}\lambda }$ reaches its global minimizer in X, that is, there exists ${v}_{\lambda }\mathrm{\in }X$ such that ${I}_{\mathrm{+}\lambda }\mathit{}\mathrm{\left(}{v}_{\lambda }\mathrm{\right)}\mathrm{=}{m}_{\lambda }^{\mathrm{1}}\mathrm{<}\mathrm{0}$.

#### Proof.

The proof of Lemma 5.3 is word for word as the one of Lemma 4.3. ∎

#### Proof of Theorem 3.4.

From Lemma 5.3, ${v}_{\lambda }$ is a local minimizer for ${I}_{+\lambda }$, with ${I}_{+\lambda }\left({v}_{\lambda }\right)={m}_{\lambda }<0$, which implies that ${v}_{\lambda }$ is nontrivial. Now, we prove that ${v}_{\lambda }$ is a positive solution of problem (P${}_{+\lambda }$). Our proof is inspired by Saoudi and Ghanmi in [11].

Let $\varphi \in X$ and $0<ϵ<1$. We define $\mathrm{\Psi }\in X$ by $\mathrm{\Psi }:={\left({v}_{\lambda }+ϵ\varphi \right)}^{+}$, where ${\left({v}_{\lambda }+ϵ\varphi \right)}^{+}=\mathrm{max}\left\{{v}_{\lambda }+ϵ\varphi ,0\right\}$. Since ${v}_{\lambda }$ is a local minimizer for ${I}_{+\lambda }$, one has

$0\le {\int }_{\mathrm{\Omega }}{|\mathrm{\Delta }{v}_{\lambda }|}^{p\left(x\right)-2}\mathrm{\Delta }{v}_{\lambda }\mathrm{\Delta }\mathrm{\Psi }dx-{\int }_{\mathrm{\Omega }}g\left(x\right){v}_{\lambda }^{-\gamma \left(x\right)}\mathrm{\Psi }dx-\lambda {\int }_{\mathrm{\Omega }}f\left(x,{v}_{\lambda }\right)\mathrm{\Psi }dx$$={\int }_{\left\{x:{v}_{\lambda }+ϵ\varphi >0\right\}}{|\mathrm{\Delta }{v}_{\lambda }|}^{p\left(x\right)-2}\mathrm{\Delta }{v}_{\lambda }\mathrm{\Delta }\left({v}_{\lambda }+ϵ\varphi \right)dx-{\int }_{\left\{x:{v}_{\lambda }+ϵ\varphi >0\right\}}g\left(x\right){u}^{-\gamma \left(x\right)}\left({v}_{\lambda }+ϵ\varphi \right)dx$$-\lambda {\int }_{\left\{x:{v}_{\lambda }+ϵ\varphi >0\right\}}f\left(x,{v}_{\lambda }\right)\left({v}_{\lambda }+ϵ\varphi \right)dx$$={\int }_{\mathrm{\Omega }}{|\mathrm{\Delta }{v}_{\lambda }|}^{p\left(x\right)-2}\mathrm{\Delta }{v}_{\lambda }\mathrm{\Delta }\left({v}_{\lambda }+ϵ\varphi \right)dx-{\int }_{\mathrm{\Omega }}g\left(x\right){u}^{-\gamma \left(x\right)}\left({v}_{\lambda }+ϵ\varphi \right)dx-\lambda {\int }_{\mathrm{\Omega }}f\left(x,{v}_{\lambda }\right)\left({v}_{\lambda }+ϵ\varphi \right)dx$$-{\int }_{\left\{x:{v}_{\lambda }+ϵ\varphi \le 0\right\}}{|\mathrm{\Delta }{v}_{\lambda }|}^{p\left(x\right)-2}\mathrm{\Delta }{v}_{\lambda }\mathrm{\Delta }\left({v}_{\lambda }+ϵ\varphi \right)dx-{\int }_{\left\{x:{v}_{\lambda }+ϵ\varphi \le 0\right\}}g\left(x\right){u}^{-\gamma \left(x\right)}\left({v}_{\lambda }+ϵ\varphi \right)dx$$-\lambda {\int }_{\left\{x:{v}_{\lambda }+ϵ\varphi \le 0\right\}}f\left(x,{v}_{\lambda }\right)\left({v}_{\lambda }+ϵ\varphi \right)dx$$={\int }_{\mathrm{\Omega }}{|\mathrm{\Delta }{v}_{\lambda }|}^{p\left(x\right)}dx-{\int }_{\mathrm{\Omega }}g\left(x\right){v}_{\lambda }^{1-\gamma \left(x\right)}dx-\lambda {\int }_{\mathrm{\Omega }}f\left(x,{v}_{\lambda }\right){v}_{\lambda }dx$$+ϵ\left({\int }_{\mathrm{\Omega }}{|\mathrm{\Delta }{v}_{\lambda }|}^{p\left(x\right)-2}\mathrm{\Delta }{v}_{\lambda }\nabla {v}_{\lambda }\nabla \varphi \mathrm{d}x-{\int }_{\mathrm{\Omega }}g\left(x\right){v}_{\lambda }^{-\gamma \left(x\right)}\varphi dx-\lambda {\int }_{\mathrm{\Omega }}f\left(x,{v}_{\lambda }\right)\varphi dx\right)$$+{ϵ}^{2}{\int }_{\mathrm{\Omega }}{|\mathrm{\Delta }{v}_{\lambda }|}^{p\left(x\right)-2}\mathrm{\Delta }{v}_{\lambda }\mathrm{\Delta }\varphi dx-{\int }_{\left\{x:{v}_{\lambda }+ϵ\varphi \le 0\right\}}{|\mathrm{\Delta }{v}_{\lambda }|}^{p\left(x\right)-2}\mathrm{\Delta }{v}_{\lambda }\mathrm{\Delta }\left({v}_{\lambda }+ϵ\varphi \right)dx$$-{\int }_{\left\{x:{v}_{\lambda }+ϵ\varphi \le 0\right\}}g\left(x\right){v}_{\lambda }^{-\gamma \left(x\right)}\left({v}_{\lambda }+ϵ\varphi \right)dx-\lambda {\int }_{\left\{x:{v}_{\lambda }+ϵ\varphi \le 0\right\}}f\left(x,{v}_{\lambda }\right)\left({v}_{\lambda }+ϵ\varphi \right)dx$$=ϵ\left({\int }_{\mathrm{\Omega }}{|\mathrm{\Delta }{v}_{\lambda }|}^{p\left(x\right)-2}\mathrm{\Delta }{v}_{\lambda }\nabla u\nabla \varphi \mathrm{d}x-{\int }_{\mathrm{\Omega }}g\left(x\right){v}_{\lambda }^{-\gamma \left(x\right)}\varphi dx-\lambda {\int }_{\mathrm{\Omega }}f\left(x,{v}_{\lambda }\right)\varphi dx\right)$$+{ϵ}^{2}{\int }_{\mathrm{\Omega }}{|\mathrm{\Delta }{v}_{\lambda }|}^{p\left(x\right)-2}\mathrm{\Delta }{v}_{\lambda }\mathrm{\Delta }\varphi dx-{\int }_{\left\{x:{v}_{\lambda }+ϵ\varphi \le 0\right\}}{|\mathrm{\Delta }{v}_{\lambda }|}^{p\left(x\right)-2}\mathrm{\Delta }{v}_{\lambda }\mathrm{\Delta }\left({v}_{\lambda }+ϵ\varphi \right)dx$$-{\int }_{\left\{x:{v}_{\lambda }+ϵ\varphi \le 0\right\}}g\left(x\right){v}_{\lambda }^{-\gamma \left(x\right)}\left({v}_{\lambda }+ϵ\varphi \right)dx-\lambda {\int }_{\left\{x:{v}_{\lambda }+ϵ\varphi \le 0\right\}}f\left(x,{v}_{\lambda }\right)\left({v}_{\lambda }+ϵ\varphi \right)dx$$\le ϵ\left(ϵ{\int }_{\mathrm{\Omega }}{|\mathrm{\Delta }{v}_{\lambda }|}^{p\left(x\right)-2}\mathrm{\Delta }{v}_{\lambda }\mathrm{\Delta }\varphi dx-{\int }_{\mathrm{\Omega }}g\left(x\right){v}_{\lambda }^{-\gamma \left(x\right)}\varphi dx-\lambda {\int }_{\mathrm{\Omega }}f\left(x,{v}_{\lambda }\right)\varphi dx\right)$$+ϵ{\int }_{\mathrm{\Omega }}{|\mathrm{\Delta }{v}_{\lambda }|}^{p\left(x\right)-2}\mathrm{\Delta }{v}_{\lambda }\nabla {v}_{\lambda }\nabla \varphi \mathrm{d}x-{ϵ}^{2}{\int }_{\left\{x:{v}_{\lambda }+ϵ\varphi \le 0\right\}}{|\mathrm{\Delta }{v}_{\lambda }|}^{p\left(x\right)-2}\mathrm{\Delta }{v}_{\lambda }\mathrm{\Delta }\varphi dx$$\le ϵ\left({\int }_{\mathrm{\Omega }}{|\mathrm{\Delta }{v}_{\lambda }|}^{p\left(x\right)-2}\mathrm{\Delta }{v}_{\lambda }\mathrm{\Delta }\varphi dx-{\int }_{\mathrm{\Omega }}g\left(x\right){v}_{\lambda }^{-\gamma \left(x\right)}\varphi dx-\lambda {\int }_{\mathrm{\Omega }}f\left(x,{v}_{\lambda }\right)\varphi dx\right)$$-ϵ{\int }_{\left\{x:{v}_{\lambda }+ϵ\varphi \le 0\right\}}{|\mathrm{\Delta }{v}_{\lambda }|}^{p\left(x\right)-2}\mathrm{\Delta }{v}_{\lambda }\mathrm{\Delta }\varphi dx.$

Since the measure of the domain of integration $\left\{x:{v}_{\lambda }+ϵ\varphi \le 0\right\}$ tends to zero as $ϵ\to {0}^{+}$, it follows as $ϵ\to {0}^{+}$ that

${\int }_{\left\{x:{v}_{\lambda }+ϵ\varphi \le 0\right\}}{|\mathrm{\Delta }{v}_{\lambda }|}^{p\left(x\right)-2}\mathrm{\Delta }{v}_{\lambda }\mathrm{\Delta }\varphi dx\to 0.$

Dividing by ϵ and letting $ϵ\to {0}^{+}$, we get

${\int }_{\mathrm{\Omega }}{|\mathrm{\Delta }{v}_{\lambda }|}^{p\left(x\right)-2}\mathrm{\Delta }{v}_{\lambda }\mathrm{\Delta }\varphi dx-{\int }_{\mathrm{\Omega }}g\left(x\right){v}_{\lambda }^{-\gamma \left(x\right)}\varphi dx-\lambda {\int }_{\mathrm{\Omega }}f\left(x,{v}_{\lambda }\right)\varphi dx\ge 0.$

Since the equality holds if we replace ϕ by $-\varphi$, which implies that ${v}_{\lambda }$ is a positive solution of problem (P${}_{+\lambda }$), this completes the proof of Theorem 3.3. ∎

## 6 An example

In this section, we give an example to illustrate our results.

#### Example 6.1.

Let Ω be a smooth bounded domain in ${ℝ}^{N}$ ($N\ge 3$), let p be a Lipschitz continuous function on $\overline{\mathrm{\Omega }}$ with $1<{p}^{-}\le {p}^{+}<\frac{N}{2}$ and ${p}^{*}\left(x\right)=\frac{Np\left(x\right)}{N-2p\left(x\right)}$, let $s,{s}_{1},r$ and ${r}_{1}$ be continuous functions on $\overline{\mathrm{\Omega }}$ such that $1<\mathrm{max}\left(r\left(x\right),{r}_{1}\left(x\right)\right) for all $x\in \mathrm{\Omega }$, let $\gamma :\overline{\mathrm{\Omega }}\to \left(0,1\right)$ be a continuous function, let

$g\in {L}^{\frac{{p}^{*}\left(x\right)}{{p}^{*}\left(x\right)+\gamma \left(x\right)-1}}\left(\mathrm{\Omega }\right),$

and let h and ${h}_{1}$ be two positive functions such that $h\in {L}^{s\left(x\right)}\left(\mathrm{\Omega }\right)$ and ${h}_{1}\in {L}^{{s}_{1}\left(x\right)}\left(\mathrm{\Omega }\right)$. Put

$f\left(x,t\right)=\left\{\begin{array}{cc}h\left(x\right){|t|}^{\beta \left(x\right)-1},\hfill & |t|\le 1,\hfill \\ {h}_{1}\left(x\right){|t|}^{\alpha \left(x\right)-1},\hfill & |t|>1,\hfill \end{array}$

with $r\left(x\right)<\beta \left(x\right)$ and $\alpha \left(x\right)<{r}_{1}\left(x\right)$ for all $x\in \mathrm{\Omega }$. Then conditions (I1), (I2) and (I3) are satisfied, so for any $\lambda \ge 0$ problem (P${}_{-\lambda }$) has a weak solution.

Moreover, if we suppose that $f\left(x,t\right)=Ch\left(x\right){|t|}^{r\left(x\right)-2}t$ for all $x\in \mathrm{\Omega }$, then assumptions (f5) and (f6) hold, and consequently, for any $\lambda \ge 0$, problem (P${}_{+\lambda }$) has at least one nontrivial weak solution in ${W}^{2,p\left(x\right)}\left(\mathrm{\Omega }\right)\cap {W}_{0}^{1,p\left(x\right)}\left(\mathrm{\Omega }\right)$.

## Acknowledgements

The authors would like to thank the anonymous referees for their suggestions and helpful comments which improved the presentation of the original manuscript.

## References

• [1]

S. N. Antontsev and S. I. Shmarev, A model porous medium equation with variable exponent of nonlinearity: Existence, uniqueness and localization properties of solutions, Nonlinear Anal. 60 (2005), no. 3, 515–545.

• [2]

M. Avci, Existence of weak solutions for a nonlocal problem involving the $p\left(x\right)$-Laplace operator, Univer. J. Appl. Math. 3 (2013), 192–197.  Google Scholar

• [3]

A. Ayoujil and A. R. El Amrouss, On the spectrum of a fourth order elliptic equation with variable exponent, Nonlinear Anal. 71 (2009), no. 10, 4916–4926.

• [4]

A. Ayoujil and A. R. El Amrouss, Continuous spectrum of a fourth order nonhomogeneous elliptic equation with variable exponent, Electron. J. Differential Equations 2011 (2011), Paper No. 24.  Google Scholar

• [5]

K. Ben Ali, A. Ghanmi and K. Kefi, On the Steklov problem involving the $p\left(x\right)$-Laplacian with indefinite weight, Opuscula Math. 37 (2017), no. 6, 779–794.

• [6]

Y. Chen, S. Levine and M. Rao, Variable exponent, linear growth functionals in image restoration, SIAM J. Appl. Math. 66 (2006), no. 4, 1383–1406.

• [7]

D. E. Edmunds and J. Rákosník, Sobolev embeddings with variable exponent, Studia Math. 143 (2000), no. 3, 267–293.

• [8]

X. Fan, On nonlocal $p\left(x\right)$-Laplacian Dirichlet problems, Nonlinear Anal. 72 (2010), no. 7–8, 3314–3323.

• [9]

X. Fan, J. Shen and D. Zhao, Sobolev embedding theorems for spaces ${W}^{k,p\left(x\right)}\left(\mathrm{\Omega }\right)$, J. Math. Anal. Appl. 262 (2001), no. 2, 749–760.  Google Scholar

• [10]

B. Ge, Q.-M. Zhou and Y.-H. Wu, Eigenvalues of the $p\left(x\right)$-biharmonic operator with indefinite weight, Z. Angew. Math. Phys. 66 (2015), no. 3, 1007–1021.

• [11]

A. Ghanmi and K. Saoudi, A multiplicity results for a singular problem involving the fractional p-Laplacian operator, Complex Var. Elliptic Equ. 61 (2016), no. 9, 1199–1216.

• [12]

S. Heidarkhani, G. A. Afrouzi, S. Moradi, G. Caristi and B. Ge, Existence of one weak solution for $p\left(x\right)$-biharmonic equations with Navier boundary conditions, Z. Angew. Math. Phys. 67 (2016), no. 3, Article ID 73.

• [13]

K. Kefi, $p\left(x\right)$-Laplacian with indefinite weight, Proc. Amer. Math. Soc. 139 (2011), no. 12, 4351–4360.

• [14]

K. Kefi and V. D. Rădulescu, On a $p\left(x\right)$-biharmonic problem with singular weights, Z. Angew. Math. Phys. 68 (2017), no. 4, Article ID 80.

• [15]

L. Kong, Multiple solutions for fourth order elliptic problems with $p\left(x\right)$-biharmonic operators, Opuscula Math. 36 (2016), no. 2, 253–264.

• [16]

R. A. Mashiyev, B. Cekic, M. Avci and Z. Yucedag, Existence and multiplicity of weak solutions for nonuniformly elliptic equations with nonstandard growth condition, Complex Var. Elliptic Equ. 57 (2012), no. 5, 579–595.

• [17]

V. D. Rădulescu and D. D. Repovš, Combined effects in nonlinear problems arising in the study of anisotropic continuous media, Nonlinear Anal. 75 (2012), no. 3, 1524–1530.

• [18]

V. D. Rădulescu and D. D. Repovš, Partial Differential Equations with Variable Exponents. Variational Methods and Qualitative Analysis, Monogr. Res. Notes Math., CRC Press, Boca Raton, 2015.  Google Scholar

• [19]

M. Růžička, Electrortheological Fluids: Modeling and Mathematical Theory, Springer, Berlin, 2000.  Google Scholar

• [20]

K. Saoudi, Existence and non-existence of solution for a singular nonlinear Dirichlet problem involving the $\mathrm{p}\left(x\right)$-Laplace operator, J. Adv. Math. Stud. 9 (2016), no. 2, 291–302.  Google Scholar

• [21]

K. Saoudi, Existence and multiplicity of solutions for a quasilinear equation involving the $p\left(x\right)$-Laplace operator, Complex Var. Elliptic Equ. 62 (2017), no. 3, 318–332.

• [22]

K. Saoudi and A. Ghanmi, A multiplicity results for a singular equation involving the $p\left(x\right)$-Laplace operator, Complex Var. Elliptic Equ. 62 (2017), no. 5, 695–725.

• [23]

K. Saoudi, M. Kratou and S. Alsadhan, Multiplicity results for the $p\left(x\right)$-Laplacian equation with singular nonlinearities and nonlinear Neumann boundary condition, Int. J. Differ. Equ. 2016 (2016), Article ID 3149482.  Google Scholar

• [24]

A. Zang and Y. Fu, Interpolation inequalities for derivatives in variable exponent Lebesgue–Sobolev spaces, Nonlinear Anal. 69 (2008), no. 10, 3629–3636.

• [25]

V. V. Zhikov, Averaging of functionals of the calculus of variations and elasticity theory (in Russian), Izv. Akad. Nauk SSSR Ser. Mat. 50 (1986), no. 4, 675–710, 877; translation in Math. USSR-Izv. 29 (1987), no. 1, 33-66.  Google Scholar

Revised: 2017-11-11

Accepted: 2018-02-28

Published Online: 2018-03-13

Citation Information: Advances in Nonlinear Analysis, Volume 8, Issue 1, Pages 1171–1183, ISSN (Online) 2191-950X, ISSN (Print) 2191-9496,

Export Citation