Jump to ContentJump to Main Navigation
Show Summary Details
More options …

Advances in Nonlinear Analysis

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) 2017: 1.89

Open Access
See all formats and pricing
More options …

An elliptic system with logarithmic nonlinearity

Claudianor Alves
  • Unidade Acadêmica de Matemática, Universidade Federal de Campina Grande, Av. Aprigio Veloso, 882, CEP 58429-900, Campina Grande /PB, Brazil
  • Email
  • Other articles by this author:
  • De Gruyter OnlineGoogle Scholar
/ Abdelkrim Moussaoui / Leandro Tavares
  • Corresponding author
  • Universidade Federal do Cariri, Av. Ten. Raimundo Rocha, CEP 63048-080, Juazeiro do Norte /CE, Brazil
  • Email
  • Other articles by this author:
  • De Gruyter OnlineGoogle Scholar
Published Online: 2017-11-27 | DOI: https://doi.org/10.1515/anona-2017-0200


In the present paper, we study the existence of solutions for some classes of singular systems involving the Δp(x) and Δq(x) Laplacian operators. The approach is based on bifurcation theory and the sub-supersolution method for systems of quasilinear equations involving singular terms.

Keywords: Bifurcation; singular system; sub-supersolution

MSC 2010: 35J75; 35J48; 35J92

1 Introduction and statement of the main results

Let ΩN (N2) be a bounded domain with smooth boundary Ω. We are interested in the following quasilinear system:

{-Δp(x)u=-γlogv+θvα(x)in Ω,-Δq(x)v=-γlogu+θuβ(x)in Ω,u,v>0in Ω,u=v=0on Ω,(1.1)

which exhibits a singularity at zero through logarithm function. The variable exponents α(), β() are positive, the constants γ and θ are both greater than 0, and Δp(x) (resp. Δq(x)) stands for the p(x)-Laplacian (resp. q(x)-Laplacian) differential operator on W01,p(x)(Ω) (resp. W01,q(x)(Ω)) with p,qC1(Ω¯),


where p(x)=Np(x)N-p(x) and q(x)=Nq(x)N-q(x). In the sequel, we set

s-=infxΩs(x),s+=supxΩs(x)  for sC(Ω¯).

Throughout this paper, we denote by C1(Ω¯)×C1(Ω¯) the pair of functions (u,v)C1(Ω¯)×C1(Ω¯) such that there is a constant c>0, which depends on u and v, verifying

u(x),v(x)cd(x)in Ω,(1.3)

where d(x):=dist(x,Ω).

A weak solution of (1.1) is a pair (u,v)W01,p(x)(Ω)×W01,q(x)(Ω) with u, v being positive a.e. in Ω and satisfying


for all (ϕ,ψ)W01,p(x)(Ω)×W01,q(x)(Ω).

The study of problems involving variable exponents growth conditions is widely justified with many physical examples, and these problems arise from a variety of nonlinear phenomena. They are used in electrorheological fluids as well as in image restorations. For more inquiries on modeling physical phenomena involving the p(x)-growth condition, we refer to [1, 2, 4, 6, 7, 14, 19, 21, 20, 23, 22].

Elliptic problems involving the logarithmic nonlinearity appear in some physical models like in the dynamic of thin films of viscous fluids; see for instance [12]. An interesting point regarding these problems comes from the fact that -logx is sign changing and behaving at the origin like the power function tα for α<0 with a slow growth. In addition, the logarithmic function is not invariant by scaling, which does not hold for the power function. These facts motivated the recent studies in [16, 8, 12], where de Queiroz et al. considered the scalar semilinear case of (1.1) (that is, p(x)=q(x)=2) with constant exponents and by essentially using the linearity of the principal part. We also mention [15], focusing on problems with constant exponents involving nonlinear operators.

The essential point in this work is that the singularity in system (1.1) comes through logarithmic nonlinearities involving variable exponents growth conditions. According to our knowledge, this happens for the first time when such problems are studied. Our main results provide the existence and regularity of (positive) solutions for problem (1.1). They are stated as follows.

Theorem 1.1.

Assume (1.2) holds.

  • (i)



    then problem ( 1.1 ) has a solution (u,v) for all θ,γ>0.

  • (ii)



    then problem ( 1.1 ) has a solution (u,v) for γ small enough and for all θ>0.

  • (iii)



    then problem ( 1.1 ) admits a solution (u,v) for γ and θ small enough.

Theorem 1.2.

Assume (1.2) and that


holds for all xΩ, where e denotes the Euler number. Then problem (1.1) has a positive solution


satisfying (1.3).

The proof of Theorem 1.1 is done in Section 4. Our approach relies on the sub-supersolutions techniques. However, this method in its system version (see [5, p. 269]) does not work for problem (1.1) due to its noncooperative character, which means that the right-hand sides of the equations in (1.1) are not necessarily increasing whenever u (resp. v) is fixed in the first (resp. second) equation in (1.1). Another reason this approach cannot be directly implemented is the presence of singularities in (1.1). To overcome this difficulties, we disturb problem (1.1) by introducing a parameter ε>0. This gives rise to a regularized system for (1.1) depending on ε>0, whose study is relevant for our initial problem. We construct a sub-supersolution pair for the regularized system independent on ε, and we show the existence of a positive family of solutions (uε,vε)C1,γ(Ω¯)×C1,γ(Ω¯), for certain γ(0,1), through a new result regarding sub-supersolutions for quasilinear competitive (noncooperative) systems involving variable exponents growth conditions (see Section 3). Then a (positive) solution of (1.1) is obtained by passing to the limit as ε0 essentially relying on the independence on ε of the upper and lower bounds of the approximate solutions (uε,vε) and on Arzelà–Ascoli’s Theorem. An important part of our result lies in obtaining the sub and supersolution, which cannot be constructed easily. Precisely, this is due to the fact that the p(x)-Laplacian operator is inhomogeneous and, in general, it has no first eigenvalue, that is, the infimum of the eigenvalues of the p(x)-Laplacian equals 0 (see [9]). At this point, the choice of suitable functions with an adjustment of adequate constants is crucial.

The proof of Theorem 1.2 is done in Section 5. It is chiefly based on a theorem by Rabinowitz (see [18]) which establishes, for each ε>0, the existence of positive solutions (uε,vε) for the regularized problem of (1.1) in W01,p(x)(Ω)×W01,q(x)(Ω). The solution of (1.1) under assumption (1.7) is obtained by passing to the limit as ε0. This is based on a priori estimates, the Hardy–Sobolev inequality and Lebesgue’s dominated convergence theorem.

A significant feature of our existence results concerns the regularity part. In Theorem 1.1, the regularity of the obtained solution for problem (1.1) is derived through the weak comparison principle and the regularity result in [3].

2 Preliminaries

Let pC(Ω¯) with p(x)>1 in Ω. Consider the Lebesgue space

Lp(x)(Ω):={u:Ω:u is measurable and Ω|u(x)|p(x)𝑑x<+},

which is a Banach space with the Luxemburg norm


The Banach space W1,p(x)(Ω) is defined as


equipped with the norm


The space W01,p(x)(Ω) is defined as the closure of C0(Ω) in W1,p(x)(Ω) with respect to the norm. The space W01,p(x)(Ω) is a separable and reflexive Banach space when p->1. For a later use, we recall that the embedding


is compact with 1r(x)<p(x).

The following result gives important properties related to the logarithmic nonlinearity.

Lemma 2.1.

  • (i)

    For each α,θ>0 , there is a constant C that depends only on α and θ such that


    for all x>0.

  • (ii)

    For each θ,ε>0 , there is a constant C that depends only on ε and θ such that


    for all x0.

  • (iii)

    Let γ, θ and δ be real numbers. If γ,θ>0 and δ>γθe , then the function f(x)=γxδ-θlogx, x>0 , attains a positive global minimum.


With respect to the inequalities, we only prove (i) because (ii) can be justified similarly. A simple computation provides


Thus, there is a small m>0 such that

|log(x)|x-αfor x(0,m).

On the other hand, the limit


implies that there is M>0 such that

|log(x)|xθfor x(M,+).

Since the function |log(x)|/xθ, x>0, is continuous for all x>0, there is a constant which depends on α and θ such that |log(x)|Cxθ in [m,M]. Therefore, |log(x)|x-α+Cxθ for all x>0, where the constant C depends only on α and θ.

In order to show (iii), observe that f(x)=θδxδ-1-γx. Then f has a unique critical point at x0=(γθδ)1/δ. Thus, by solving the inequalities f(x)>0 and f(x)<0 for x>0, it follows that f is increasing on the interval [x0,+) and decreasing on (-,x0]. By noticing that


the condition δ>γθe implies that f(x0)>0, which proves the result. ∎

3 Sub-supersolution theorem

Let us introduce the quasilinear system

{-Δp(x)u=H(x,u,v)in Ω,-Δq(x)v=G(x,u,v)in Ω,u=v=0on Ω,(3.1)

where H,G:Ω×+×+ are Carathéodory functions satisfying the following assumption:

  • (I)

    Given T,S>0, there is a constant C>0 such that

    |H(x,s,t)|,|G(x,s,t)|Cfor all (x,s,t)Ω×[0,T]×[0,S].

The following result is a key point in the proof of Theorem 1.1.

Theorem 3.1.

Assume that H and G satisfy (I), and let u¯W01,p(x)(Ω)L(Ω) and v¯W01,q(x)(Ω)L(Ω), with u¯,v¯0 in Ω and u¯,v¯W1,(Ω) such that

u¯u¯𝑎𝑛𝑑v¯v¯  in Ω.

Suppose that




for all nonnegative functions (ϕ,ψ)W01,p(x)(Ω)×W01,q(x)(Ω). Then problem (3.1) has a (positive) solution



u¯(x)u(x)u¯(x)𝑎𝑛𝑑v¯(x)v(x)v¯(x)  for a.e. xΩ.


The proof is chiefly based on pseudomonotone operator theory. Define the functions

H1(x,s,t)={H(x,u¯(x),v¯(x)),su¯(x),H(x,s,v¯(x)),u¯(x)su¯(x) and tv¯(x),H(x,s,t),u¯(x)su¯(x) and v¯(x)tv¯(x),H(x,s,v¯(x)),u¯(x)su¯(x) and tv¯(x),H(x,u¯(x),v¯(x)),su¯(x) and tv¯(x),


G1(x,s,t)={G(x,u¯(x),v¯(x)),tv¯(x),G(x,s,v¯(x)),v¯(x)tv¯(x) and su¯(x),G(x,s,t),v¯(x)tv¯(x) and u¯(x)su¯(x),G(x,s,v¯(x)),v¯(x)tv¯(x) and su¯(x),G(x,u¯(x),v¯(x)),tv¯(x).

In what follows, we fix l(0,1) with min{p-,q-}>1+l and set


Using the above functions, we introduce the auxiliary problem

{-Δp(x)u=H2(x,u,v)in Ω,-Δq(x)v=G2(x,u,v)in Ω,u=v=0on Ω,(3.2)





By the Minty–Browder theorem (see, e.g., [17]), problem (3.2) has a solution (u,v) in W01,p(x)(Ω)×W01,q(x)(Ω). Indeed, let B:EE be a function defined by


where E is the Banach space W01,p(x)(Ω)×W01,q(x)(Ω) endowed with the norm


Let us show that the function B satisfies the hypotheses of the Minty–Browder theorem.

(i) B is continuous.

Let (un,vn)E be a sequence that converges to (u,v) in E. We need to prove that B(un,vn)-B(u,v)E0. To this end, let (ϕ,ψ)E with (ϕ,ψ)E1. By the Hölder inequality, one has


Up to a subsequence, we can assume that un(x)u(x) a.e in Ω and that there exists a function U(Lp(x)(Ω))N such that |un(x)|U(x) a.e in Ω. Therefore, Lebesgue’s dominated convergence theorem yields


Note that


Then the continuity and the boundedness of H, together with Lebesgue’s dominated convergence theorem and the Hölder inequality, gives


On the other hand, we can assume that un(x)u(x) a.e in Ω and that there exists wLp(x)(Ω) such that |un(x)|w(x) a.e in Ω. Arguing as before, we get


and so


Hence, the previous reasoning provides




which justify the continuity of B.

(ii) B is bounded.

Let us show that if UE is a bounded set, then B(U)E is bounded. To this end, consider a bounded set U and (ϕ,ψ)E such that (ϕ,ψ)1. Then for (u,v)U the Hölder inequality gives


Since H1(x,u,v) is bounded, we derive that


On the other hand, since


the Hölder inequality ensures


From the above arguments we obtain the boundedness of B.

(iii) B is coercive.

Next, we prove that

B(u,v),(u,v)(u,v)+as (u,v)+.

Note that


where C is a positive constant. The triangular inequality and the fact that (a+b)θaθ+bθ for nonnegative numbers a and b with θ(0,1) give


Gathering the last inequality with the embeddings


we derive


From (3.3)–(3.5) we have


where C is a positive constant. In the same manner, we can see that


  • If uLp(x)(Ω)1 and vLq(x)(Ω)<1, then


  • If uLp(x)(Ω)1 and vLq(x)(Ω)1, then


Consider in E a sequence {(un,vn)}n such that (un,vn)+. Thus,


Suppose that the first possibility happens and that unLp(x)(Ω)1 for all n. Then we consider two cases:

  • unLp(x)(Ω)1 and vnLq(x)(Ω)<1 for n. In this case, we have


  • unLp(x)(Ω)1 and vnLq(x)(Ω)1 for n. In this second case, we have


Consequently, in both cases studied above, one has

B(un,vn),(un,vn)(un,vn)Eas n+.

The other situations regarding unLp(x)(Ω) and vnLq(x)(Ω) can be handled in much the same way.

(iv) B is pseudomonotone.

We recall that B is a pseudomonotone operator if (un,vn)(u,v) in E and

lim supn+B(un,vn),(un,vn)-(u,v)0.(3.6)


lim infn+B(un,vn),(un,vn)-(ϕ,ψ)B(u,v),(u,v)-(ϕ,ψ)

for all (ϕ,ψ)E.

If (un,vn)(u,v), then unu and vnv in W1,p(x)(Ω) and W1,q(x)(Ω), respectively. Since H1 and G1 are bounded, we must have




Note that


The previous arguments can be repeated to show that


Gathering the above limits together with (3.6), we have


From the weak convergence we get








By using (3.8) and (3.9) in (3.7), the (S+) property of the operators -Δp(x) and -Δq(x) guarantees that unu in W01,p(x)(Ω) and vnv in W01,q(x)(Ω). Thus, by the continuity of B, it turns out that


for all (ϕ,ψ)E.

Finally, from properties (i)–(iv) we are in a position to apply [17, Theorem 3.3.6] which ensures that B is surjective. Thereby, there exists (u,v)E such that

B(u,v),(ϕ,ψ)=0for all (ϕ,ψ)E,

and, in particular, (u,v) is a solution of (3.2).

It remains to prove that

u¯uu¯andv¯vv¯  in Ω.(3.10)

We only prove the first inequalities in (3.10) because the second ones can be justified similarly. Set (ϕ,ψ):=((u-u¯)+,0). From the definition of H2 we obtain




wich implies that uu¯ in Ω. Using a quite similar argument for (ϕ,ψ):=((u¯-u)+,0), we get u¯u in Ω. This completes the proof. ∎

4 Proof of Theorem 1.1

For every ε>0, let us introduce the auxiliary problem

{-Δp(x)u=-γlog(|v|+ε)+θ|v|α(x)in Ω,-Δq(x)v=-γlog(|u|+ε)+θ|u|β(x)in Ω,u=v=0on Ω.(4.1)

Our goal is to show through Theorem 3.1 that (4.1) has a positive solution (uε,vε). Then, by passing to the limit as ε0+, we get a solution for the original problem (1.1).

Let Ω~ be a bounded domain in N with smooth boundary Ω~ such that Ω¯Ω~, and set d~(x)=dist(x,Ω~). In [24, Lemma 3.1], Yin and Zang have proved that, for δ>0 small enough and for constants η>0, the function


defined by

w(x)={ξd~(x)if d~(x)<δ,ξδ+δd~(x)ξ(2δ-tδ)2p--1if δd~(x)<2δ,ξδ+δ2δξ(2δ-tδ)2p--1if 2δd~(x),

is a subsolution of the problem

{-Δp(x)u=ηin Ω~,u=0on Ω~,

where δ>0 is a number that does not depend on η, and


with a fixed number τ(0,1) and c0>0 is a number depending only on δ, τ, Ω~ and p. Note that

w(x)=c0η1p+-1+τd~(x)for d~(x)<δ,c0η1p+-1+τδw(x)for d~(x)δ.

Given λ>1, let u¯ and v¯ in C1(Ω~¯) be the unique solutions of the problems

{-Δp(x)u¯=λσin Ω~,u¯=0on Ω~,and{-Δq(x)v¯=λσin Ω~,v¯=0on Ω~,(4.2)

where σ is a real constant.

If σ>0, considering the corresponding function w for η=λσ and applying the weak maximum principle, we get

{C0λσp+-1+τ1min{δ,d~(x)}u¯(x)C1λσp--1,C0λσq+-1+τ2min{δ,d~(x)}v¯(x)C1λσq--1,in Ω~,(4.3)

where C0,C0,C1,C1>0 and τ1,τ2(0,1) are constants that do not depend on λ. If -1<σ<0, from [11, Lemma 2.1] and for λ large one has

u¯(x)k2λσp--1c2λσp+-1andv¯(x)k2λσq--1c2λσq+-1  in Ω~,(4.4)

where k2, k2, c2 and c2 are positive constants independent of λ. Moreover, by the strong maximum principle there is a constant c0>0 (that can depend on λ) such that


Now, let u¯ and v¯ in C1(Ω¯) be the unique solutions of the homogeneous Dirichlet problems

{-Δp(x)u¯=λ-1in Ω,u¯=0on Ω,and{-Δq(x)v¯=λ-1in Ω,v¯=0on Ω.(4.6)

By [10, Lemma 2.1] and [11], there exist positive constants k0, K1 and K2 independent of λ such that

u¯(x)K1λ-1p--1andv¯(x)K2λ-1q--1  in Ω(4.7)


k0d(x)min{u¯(x),v¯(x)}in Ω.(4.8)

By the weak maximum principle, we have u¯u¯ and v¯v¯ in Ω¯ for λ>1 sufficiently large.

We state the following existence result for the regularized problem (4.1).

Theorem 4.1.

Under the assumptions of Theorem 1.1, there exists ε0>0 such that system (4.1) has a positive solution


for all ε(0,ε0). Moreover,

u¯(x)uε(x)u¯(x)𝑎𝑛𝑑v¯(x)vε(x)v¯(x)  for a.e. xΩ.(4.9)


First, let us show that (u¯,v¯) is a subsolution for problem (4.1) for all ε(0,ε0). To this end, pick ε012. Then from (4.6) and (4.7), for all ε(0,ε0), one has

-Δp(x)u¯=λ-1-γlog(K2λ-1q--1+ε0)-γlog(v¯(x)+ε)-γlog(v¯(x)+ε)+θv¯(x)α(x)in Ω


-Δq(x)v¯=λ-1-γlog(K1λ-1p--1+ε0)-γlog(u¯(x)+ε)-γlog(u¯(x)+ε)+θu¯(x)β(x)in Ω

for all γ,θ>0, provided that λ>1 is sufficiently large.

Next, we will show that (u¯,v¯) is a supersolution for problem (4.1) for all ε(0,ε0). Set δ:=dist(Ω~,Ω) and fix ε0(0,1). By Lemma 2.1, there are constants σ1,σ2(0,1) and Cσ1,α+,Cσ2,β+>0 such that, for all ε(0,ε0), one has




If (1.4) holds, it follows from (4.3), (4.10), (4.11) and for σ>0 in (4.2) that

-γlog(v¯+ε)+θv¯α(x)γ(1v¯σ1+Cσ1,α+(v¯+ε0)α+)+θ(v¯+1)α+γv¯σ1+2α+-1(γCσ1,α++θ)(v¯α++1)γ(v¯)σ1+2α+-1(γCσ1,α++θ)C1(λσα+q--1+1)γ(C0λσq+-1+τ2min{δ,δ})σ1+2α+-1(γCσ1,α++θ)(C1λσα+q--1+1)λσin Ω(4.12)


-γlog(u¯+ε)+θu¯β(x)γ(1u¯σ2+Cσ2,β+(u¯+ε)β+)+θ(u¯+1)β+γu¯σ2+2β+-1(γCσ2,β++θ)(u¯β++1)γ(C0λσp+-1+τ1min{δ,δ})σ2+2β+(γCσ2,α++θ)(C1λσβ+p--1+1)λσin Ω(4.13)

for all γ,θ>0, provided that λ is large enough.

If (1.5) is satisfied, combining Lemma 2.1 with (4.4) and (4.5), by (4.10), (4.11) and for σ(-1,0) in (4.2), we get

-γlog(v¯+ε)+θv¯α(x)γ(1(v¯+ε)σ1+Cσ,1α-(v¯+ε)α-)+θv¯α(x)γ(1v¯σ1+Cσ1,α-2α--1v¯α-+Cσ1,α-ε0α-2α--1)+θv¯α(x)γ(1(c0δ)σ1+Cσ1,α-2α--1λσα-q+-1+Cσ1,α-2α--1)+θc2λσα(x)q+-1γ(1(c0δ)σ1+Cσ1,α-2α--1λσα-q+-1+Cσ1,α-2α--1)+θc2λσα-q+-1λσin Ω(4.14)


-γlog(u¯+ε)+u¯β(x)γ(c0δ)σ1+γCσ1,β-2β--1λσα-q+-1+γCσ1,β-2β--1+c2λσβ-p+-1λσin Ω(4.15)

for γ>0 small enough, for all θ>0 and all ε(0,ε0), provided that λ is sufficiently large.

Finally, if (1.6) holds, using (4.3), (4.5), (4.10) and (4.11), for σ(-1,0) in (4.2), we obtain

-γlog(v¯+ε)+θv¯α(x)γ(1(v¯+ε)σ1+Cσ1,α+(v¯+1)α+)+θv¯α(x)γ(1v¯σ1+Cσ1,α+(v¯α++1))+θ(v¯+1)α+γ(1(c0δ)σ1+Cσ1,α+(v¯α++1))+2α+θ(v¯α++1)γ(1(c0δ)σ1+Cσ1,α+(λσα+q--1+1))+2α+θ(λσα+q--1+1)λσin Ω¯,(4.16)

and similarly

-γlog(u¯+ε)+θu¯β(x)γ(1u¯σ2+Cσ2,β+(u¯+1)β+)+θu¯β(x)λσin Ω¯(4.17)

for all γ,θ>0 small and all ε(0,ε0), provided that λ>0 is large enough.

Consequently, it turns out from (4.12), (4.13), (4.14), (4.15), (4.16) and (4.17) that




for all (ϕ,ψ)W01,p(x)(Ω)×W01,q(x)(Ω) with ϕ,ψ0. This shows that (u¯,v¯) is a supersolution for (4.1) for all ε(0,ε0).

Then, owing to Theorem 3.1, we conclude that the perturbed problem (4.1) has a solution


within [u¯,u¯]×[v¯,v¯] for all ε(0,ε0). Moreover, according to Lemma 2.1 combined with (4.8) and (4.9), we have that for σ1,σ2(0,1N) there are constants Cσ1,Cσ2>0 such that

-γlog(vε+ε)+θvεα(x)γ(vε-σ1+Cσ1vεσ1)+θvεα(x)=vε-σ1(γ+γCσ1vε2σ1+θvεα(x)+σ1)v¯-σ1(γ+γCσ1v¯2σ1+θv¯α(x)+σ1)(k0d(x))-σ1(γ+γCσ1v¯2σ1+θv¯α(x)+σ1)A1d(x)-σ1in Ω


-γloguε+uεβ(x)γ(uε-σ2+Cσ2uεσ2)+θuεβ(x)=u¯-σ2(γ+γCσ2u¯2σ2+θu¯β(x)+σ2)(k0d(x))-σ2(γ+γCσ2u¯2σ2+θu¯β(x)+σ2)A2d(x)-σ2in Ω

for some positive constants A1 and A2. Then, thanks to [3, Lemma 2], we deduce that


for certain ν(0,1). ∎

Proof of Theorem 1.1.

Set ε:=1n for n1/ε0. By Theorem 4.1, we know that there exists a positive solution (un,vn):=(u1/n,v1/n) bounded in C1,ν(Ω¯)×C1,ν(Ω¯), for certain ν(0,1), for the problem

{-Δp(x)un=-γlog(|vn|+1n)+θ|vn|α(x)in Ω,-Δq(x)vn=-γlog(|un|+1n)+θ|un|β(x)in Ω,un=vn=0on Ω.

Moreover, the property formulated in (4.9) holds true. Employing Arzelà–Ascoli’s theorem, we may pass to the limit in C1(Ω¯)×C1(Ω¯) and the limit functions (u,v)C1(Ω¯)×C1(Ω¯) satisfy (1.1) with (u,v)[u¯,u¯]×[v¯,v¯]. This completes the proof. ∎

5 Proof of Theorem 1.2

This section is devoted to the proof of Theorem 1.2. For ε>0, let us consider the regularized problem

{-Δp(x)u=-γlog(|v|+ε)+θ(|v|+ε)α(x)in Ω,-Δq(x)v=-γlog(|u|+ε)+θ(|u|+ε)β(x)in Ω,u=v=0on Ω.(5.1)

Our demonstration strategy will be to show, by applying the well-known result due to Rabinowitz [18], that for each λ>0 system (5.1) possesses a positive solution (uε,vε) in W01,p(x)(Ω)×W01,q(x)(Ω), and then derive a solution of (1.1) by taking the limit ε0.

5.1 Existence result for the regularized system

Fix ε>0, and for each pair (f,g)Lp(x)(Ω)×Lq(x)(Ω) let us consider the auxiliary problem

{-Δp(x)u=λ(-γlog(|g|+ε)+θ(|g|+ε)α(x))in Ω,-Δq(x)v=λ(-γlog(|f|+ε)+θ(|f|+ε)β(x))in Ω,u=v=0on Ω.(5.2)

Observe the following facts:

  • -log(|g|+ε)Lp(x)(Ω): Indeed, consider θ>0 such that 0<θ(p)+q(x) for all xΩ¯. By Lemma 2.1, one has


    From (1.2) the claim follows.

  • (|g(x)|+ε)α(x)Lp(x): By (1.2), notice that


    Since W1,q(x)(Ω)Lq(x)(Ω), the claim is proved.

In the same manner, we have |log(|f|+ε)|Lq(Ω) and (|f|+ε)β(x)Lq(x) for all fLp(x)(Ω). Then, on account of the above remarks, the unique solvability of (u,v)W01,p(x)(Ω)×W01,q(x)(Ω) in (5.2) is readily derived from the Minty–Browder theorem. Therefore, the solution operator


is well defined.

Lemma 5.1.

The operator T:R+×Lp(x)(Ω)×Lq(Ω)Lp(x)(Ω)×Lq(x)(Ω) is continuous and compact.


Consider a sequence (λn,fn,gn)(λ,f,g) in +×Lp(x)(Ω)×Lq(x)(Ω), and (un,vn):=𝒯(λn,fn,gn). By using un as a test function, one gets


Since {gn} is bounded in Lq(x)(Ω), by Lemma 2.1, {un} is bounded in W1,p(x)(Ω). Let (u,v):=𝒯(λ,f,g). Using un-u as a test function, we have

Ω|un|p(x)-2un-|u|p(x)-2u,(un-u)𝑑x=γΩ(λlog(|g|+ε)-λlog(|gn|+ε))(un-u)𝑑x   +θΩ(λ(|g|+ε)α(x)-λn(|gn|+ε)α(x))(un-u)𝑑x.(5.4)

Note that


where the constant C does not depend on n.

In the sequel, up to a subsequence, we can assume that gn(x)g(x) a.e in Ω and |gn(x)|h a.e in Ω for some hLq(x)(Ω). Then, by Lemma 2.1 and the Lebesgue theorem, we have


A similar reasoning leads to


Since {un} is bounded in W01,p(x)(Ω), from (5.4) we deduce that unu in W01,p(x)(Ω). This proves that 𝒯 is continuous.

In order to show that 𝒯 is compact, it suffices to prove that 𝒯(U)¯ is compact for all UE bounded. At this point, a quite similar argument as above produces the desired conclusion. This completes the proof. ∎

Theorem 5.2.

Under assumptions (1.2) and (1.7), problem (5.1) admits a solution (uε,vε) for all ε>0.


From Lemma 5.1 and by invoking [18], there is an unbounded continuum 𝒞 of solutions of the equation (u,v)=𝒯(λ,u,v), that is, (λ,u,v)𝒞 is a solution of (5.1).

On the other hand, by Lemma 2.1, the function f(x)=θxδ-γlogx, for x>0, attains a strictly positive minimum if δ>γθe. Since α-,α+>γθe, we obtain the following assertions:

  • If |u|+ε1, then


    where m-=min{-γlogx+θxα-:x>0}.

  • If |u|+ε<1, then


    where m+=min{-γlogx+θxα+:x>0}.

Therefore, -Δp(x)um1>0, where m1=min{m-,m+}, and, with a quite similar reasoning, we get -Δq(x)vm2>0 for some m2>0. Thus, by the maximum principle, 𝒞{(0,0,0)} must be constituted by strictly positive functions.

Next, we show that the component 𝒞 is unbounded with respect to λ0. By contradiction, suppose that there is λ>0 such that (λ,u,v)𝒞 implies that λλ. Fix 0<γ¯(q/p)-. Using u as a test function, we get




where C depends on λ, ε, γ and γ¯. Note that


where C depends on θ and ε. Now we will estimate the integral Ω|u||v|α(x)𝑑x. We have |v(x)|α(x)Lp(x)(Ω). In order to prove this, note that


The last function belongs to L1(Ω) because W1,q(x)(Ω)Lq(x)(Ω)Lq(x)(Ω). Thus, by the Hölder inequality we obtain


By the Hölder inequality and considering all the possibilities for the norms


we get


Using the embedding W1,q(x)(Ω)Lq(x)(Ω), considering all the possibilities for the norms vLq(x)(Ω) and uLp(x)(Ω) and estimates (5.9), (5.10), (5.11) and (5.12), and repeating the arguments for the integral Ω|u||v|γ𝑑x, we obtain




A similar reasoning leads to


Since α++1<p- and β++1<q-, it follows that the component 𝒞 is bounded, which is absurd. Consequently, 𝒞 crosses the set {1}×Lp(x)(Ω)×Lq(x)(Ω), and this implies that there is a solution (uε,vε) of (5.1). The proof is completed. ∎

5.2 Passage to the limit

Set ε=1n in (5.1) with any integer n1. By applying Theorem 5.2, we know that there exist u1/n:=un and v1/n:=vn that solve the problem (5.1) with ε=1n.


The sequences {un} and {vn} are bounded in W01,p(x)(Ω) and W01,q(x)(Ω), respectively, and the weak limits (that exist up to a subsequence) are strictly positive in Ω.

First of all, we know that -Δp(x)unm1>0, where m1=min{m-,m+}. If w1 denotes the unique positive solution of

{-Δp(x)w1=m1in Ω,w1=0on Ω,

the maximum principle gives

unw1>0in Ω.

By the strong maximum principle (see [11, Theorem 1.2]), we have w1η>0, where η is the inward normal vector of Ω. Let ϕq- be an eigenfunction associated to the first eigenvalue of the operator (-Δq-,W01,q-(Ω)). Note that


where P1 is a positive constant that does not depend on xΩ.

Denote by ϕp- an eigenfunction associated to the first eigenvalue of the operator (-Δp-,W01,p-(Ω)). Reasoning as above, we also have vnw2>0 and


where L1 is a positive constant that does not depend on xΩ, with w2 being the unique positive solution of

{-Δq(x)w2=m2in Ω,w2=0on Ω.

Let δ(0,1). By using un as a test function in its corresponding system of equations and arguing as in the set of inequalities (5.10) and (5.11), we get


where C is a constant that depends on γ and θ. By the Hardy–Sobolev inequality (see [13]), together with the embedding W01,p(x)(Ω)W01,p-(Ω) and the relation (5.15), it follows that


where the constant C does not depend on n.

By (5.16) and using the reasoning that leads to (5.13) and (5.14), we obtain that (un,vn) is bounded in E. Passing to a subsequence, we have

  • unu in W1,p(x)(Ω),

  • unu in Lp(x)(Ω),

  • unu a.e in Ω,

  • vnv in W1,p(x)(Ω),

  • vnv in Lp(x)(Ω),

  • vnv a.e in Ω

for some pair (u,v)E. From the previous pointwise convergence and the relations between w1, un and w2, vn, we conclude that u>0 and v>0, which proves the claim.

Taking un as a test function and repeating the arguments of relations (5.3)–(5.8), we get that unu in W01,p(x)(Ω). Notice that the same argument provides that vnv in W01,q(x)(Ω).

From the previous strong convergence of un and vn, combined with Lebesgue’s dominated convergence theorem, we obtain




for all (ϕ,ψ)W01,p(x)(Ω)×W01,q(x)(Ω), and the existence of a solution is proved.


The authors thank the anonymous referees and the editor, Prof. Vicentiu Rădulescu, for their valuable comments which helped to improve this work. The work was started while the second and the third authors were visiting the Federal University of Campina Grande. They thank Prof. Claudianor Alves and the other members of the department for hospitality.


  • [1]

    E. Acerbi and G. Mingione, Regularity results for stationary electrorheological fluids, Arch. Ration. Mech. Anal. 164 (2002), 213–259.  CrossrefGoogle Scholar

  • [2]

    E. Acerbi and G. Mingione, Regularity results for electrorheological fluids: Stationary case, C. R. Math. Acad. Sci. Paris 334 (2002), 817–822.  CrossrefGoogle Scholar

  • [3]

    C. O. Alves and A. Moussaoui, Existence and regularity of solutions for a class of singular (p(x),q(x))-Laplacian systems, Complex Var. Elliptic Equ. (2017), 10.1080/17476933.2017.1298589.  Web of ScienceGoogle Scholar

  • [4]

    S. N. Antontsev and J. F. Rodrigues, On stationary thermo-rheological viscous flows, Ann. Univ. Ferrara Sez. VII Sci. Mat. 52 (2006), 19–36.  CrossrefGoogle Scholar

  • [5]

    S. Carl, V. K. Le and D. Motreanu, Nonsmooth Variational Problems and Their Inequalities, Comparaison Principles and Applications, Springer, New York, 2007.  Google Scholar

  • [6]

    A. Chambolle and P. L. Lions, Image recovery via total variation minimization and related problems, Numer. Math. 76 (1997), 167–188.  CrossrefGoogle Scholar

  • [7]

    Y. Chen, S. Levine and M. Rao, Variable exponent, linear growth functionals in image restoration, SIAM J. Appl. Math. 66 (2006), 1383–1406.  CrossrefGoogle Scholar

  • [8]

    O. S. de Queiroz, A Neumann problem with logarithmic nonlinearity in a ball, Nonlinear Anal. 70 (2009), no. 4, 1656–1662.  Web of ScienceCrossrefGoogle Scholar

  • [9]

    X. Fan, Q. Zhang and D. Zhao, Eigenvalues of p(x)-Laplacian Dirichlet problem, J. Math. Anal. Appl. 302 (2005), 306–317.  Google Scholar

  • [10]

    X. L. Fan, On the sub-supersolution method for p(x)-Laplacian equations, J. Math. Anal. Appl. 330 (2007), 665–682.  Web of ScienceGoogle Scholar

  • [11]

    X. L. Fan, Y. Z. Zhao and Q. H. Zhang, A strong maximum principle for p(x)-Laplace equations, Chinese J. Contemp. Math. 24 (2003), no. 3, 1–7.  Google Scholar

  • [12]

    L. C. Ferreira and O. S. de Queiroz, A singular parabolic equation with logarithmic nonlinearity and Lp-initial data, J. Differential Equations 249 (2010), no. 2, 349–365.  Web of ScienceGoogle Scholar

  • [13]

    O. Kavian, Inegalité de Hardy–Sobolev et applications, Thése de Doctorate de 3eme cycle, Université de Paris VI, 1978.  Google Scholar

  • [14]

    M. Mihăilescu and V. Rădulescu, On a nonhomogeneous quasilinear eigenvalue problem in Sobolev spaces with variable exponent, Proc. Amer. Math. Soc. 135 (2007), no. 9, 2929–2937.  CrossrefWeb of ScienceGoogle Scholar

  • [15]

    J. Mo and Z. Yang, Existence of solutions to p-Laplace equations with logarithmic nonlinearity, Electron. J. Differential Equations 87 (2009), 1–10.  Google Scholar

  • [16]

    M. Montenegro and O. S. de Queiroz, Existence and regularity to an elliptic equation with logarithmic nonlinearity, J. Differential Equations 246 (2009), no. 2, 482–511.  Web of ScienceCrossrefGoogle Scholar

  • [17]

    J. Necas, Introduction to the Theory of Nonlinear Elliptic Equations, John Willey & Sons, Chichester, 1983.  Google Scholar

  • [18]

    P. H. Rabinowitz, Some global results for nonlinear eigenvalue problems, J. Funct. Anal. 7 (1971), 487–513.  CrossrefGoogle Scholar

  • [19]

    V. Rădulescu, Nonlinear elliptic equations with variable exponent: Old and new, Nonlinear Anal. 121 (2015), 336–369.  Web of ScienceCrossrefGoogle Scholar

  • [20]

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

  • [21]

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

  • [22]

    D. Repovš, Stationary waves of Schrödinger-type equations with variable exponent, Anal. Appl. (Singap.) 13 (2015), no. 6, 645–661.  CrossrefGoogle Scholar

  • [23]

    M. Ruzicka, Electrorheological Fluids: Modeling and Mathematical Theory, Lecture Notes in Math. 1748, Springer, Berlin, 2000.  Google Scholar

  • [24]

    H. Yin and Z. Yang, Existence and asymptotic behavior of positive solutions for a class of (p(x),q(x))-Laplacian systems, Differ. Equ. Appl. 6 (2014), 403–415.  Google Scholar

About the article

Received: 2017-09-11

Revised: 2017-10-11

Accepted: 2017-10-13

Published Online: 2017-11-27

Funding Source: Conselho Nacional de Desenvolvimento Científico e Tecnológico

Award identifier / Grant number: 304036/2013-7

Award identifier / Grant number: 402792/2015-7

The first author was partially supported by Conselho Nacional de Desenvolvimento Científico (304036/2013-7) and Instituto Nacional de Ciência de Tecnologia. The second author was supported by Conselho Nacional de Desenvolvimento Científico (402792/2015-7).

Citation Information: Advances in Nonlinear Analysis, Volume 8, Issue 1, Pages 928–945, ISSN (Online) 2191-950X, ISSN (Print) 2191-9496, DOI: https://doi.org/10.1515/anona-2017-0200.

Export Citation

© 2019 Walter de Gruyter GmbH, Berlin/Boston. This work is licensed under the Creative Commons Attribution 4.0 Public License. BY 4.0

Comments (0)

Please log in or register to comment.
Log in