See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/351410078

Existence results for anisotropic fractional (p1(x,.), p2(x,.))-Kirchhoff type

problems

Article in Journal of Applied Analysis & Computation · May 2021

DOI: 10.11948/20200394

CITATIONS READS
3 531

4 authors:

El Houssine Azroul A. Benkirane

Sidi Mohamed Ben Abdellah University Sidi Mohamed Ben Abdellah University
156 PUBLICATIONS 1,186 CITATIONS 171 PUBLICATIONS 1,738 CITATIONS

SEE PROFILE SEE PROFILE

Nguyen Thanh Chung Mohammed Shimi

Quang Binh University Sidi Mohamed Ben Abdellah University
104 PUBLICATIONS 850 CITATIONS 18 PUBLICATIONS 242 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Mohammed Shimi on 30 September 2021.

The user has requested enhancement of the downloaded file.

Journal of Applied Analysis and Computation Website:http://www.jaac-online.com
Volume 11, Number 5, October 2021, 2363–2386 DOI:10.11948/20200394

EXISTENCE RESULTS FOR ANISOTROPIC

FRACTIONAL (p1 (x, .), p2 (x, .))-KIRCHHOFF
TYPE PROBLEMS

E. Azroul1 , A. Benkirane1 , N. T. Chung2 , and M. Shimi1,†

Abstract In this paper, we investigate the existence and multiplicity of so-

lutions for a class of fractional (p1 (x, .), p2 (x, .))-Kirchhoff type problems with
Dirichlet boundary data of the following form
 2 (∫ )

 ∑ 1 |u(x) − u(y)|pi (x,y)

 M dxdy (−∆)spi (x,.) u(x)


i
p (x, y) |x − y| N +spi (x,y)
 i=1 i
( s )  ∑
Q
2
PMi

 + |u|p̄i (x)−2 u = f (x, u) in Ω,



 i=1

u = 0 in RN \ Ω.

More precisely, by means of mountain pass theorem with Cerami condition, we

show that the above problem has at least
( s one) nontrivial solution. Moreover,
using Fountain theorem, we prove that PM i
possesses infinitely many (pairs)
of solutions with unbounded energy.
Keywords Kirchhoff type problems, fractional p(x, .)-Laplacian operator,
mountain pass theorem, fountain theorem, cerami condition.
MSC(2010) 35R11, 47G20, 35S15, 35A15.

1. Introduction and statement of the main results

The study of differential equations involving p(x)-Laplacian operators have been a
very interesting and exciting topic in recent years (see in particular the fascinating
monograph [24] and the references therein for further details). This type of problems
are extremely attractive because they can be used to model dynamical phenomena
which arise from the study of electrorheological fluids or elastic mechanics. Problems
with variable exponent growth conditions also appear in the modelling of stationary
thermorheological viscous flows of non-Newtonian fluids and in the mathematical
description of the processes of filtration of an ideal barotropic gas through a porous
medium. The detailed application backgrounds of the p(x)-Laplacian can be found
in [4, 19, 27, 36, 40] and the references therein.
For problems involving different growth rates depending on the underlying do-
mains, they also involve equations with (p(x), q(x))-growth conditions where several
†
The corresponding author. Email: [email protected] (M. Shimi)
1
Department of Mathematics, Sidi Mohamed Ben Abdellah University, Faculty of
Sciences Dhar el Mahraz, Laboratory of Mathematical Analysis and Applications,
B.P. 1796 Fez-Atlas, 30003, Fez, Morocco
2
Department of Mathematics, Quang Binh University, 312 Ly Thuong Kiet, Dong
Hoi, Quang Binh, Vietnam
2364 E. Azroul, A. Benkirane, N. T. Chung, and M. Shimi

p(x)-Laplacian operators involved, interacting with one another. This (p(x), q(x))-
growth condition is a natural generalization of the anisotropic (p, q)-growth condi-
tion. In that context, the systems involving the (p(x), q(x))-Laplacian (or (p1 (x),
p2 (x), · · · , pn (x))-Laplacian) can be good candidates for modeling phenomena which
ask for distinct behavior of partial differential derivatives in various directions, for
related problems we just mention [3, 21, 33, 37].
On the other hand, in the recent years increasing attention has been paid to
the study of pseudo-differential and nonlocal fractional operators (as (−∆)s , (−∆)sp
and their generalizations) and related fractional differential equations. This type
of operators arises in a quite natural way in many different applications, such as,
continuum mechanics, phase transition phenomena, population dynamics, minimal
surfaces and game theory, as they are the typical outcome of stochastic stabilization
of Lévy processes, see for instance [5, 18, 31] and the references therein. For a self-
contained overview of the basic properties of fractional Sobolev spaces and fractional
Laplacian (or fractional p-Laplacian operator), we refer the reader to [16, 35] and
to the references included.
It is therefore, a natural question is to see which results “survive” when the
p(x)-Laplacian is replaced by the fractional p(x, .)-Laplacian. In a few last years, to
our best knowledge, there have been some mathematicians extending the study of
classical exponent variable case to include fractional case (see for instance [2, 6–11,
13, 14, 17, 23, 29, 32]), the authors established some definitions and basic properties
about new fractional Sobolev spaces with variable exponents and obtained some
existence results for nonlocal fractional problems.
Motivated by the papers mentioned above and the results introduced in [1, 22]
and the references therein, we aim to discuss the existence of a nontrivial solu-
tions for a fractional (p1 (x, .), p1 (x, .))-Kirchhoff type problem with homogeneous
Dirichlet boundary data of the following form
 2 Z 

 X 1 |u(x) − u(y)|pi (x,y)

 Mi
s
dxdy (−∆)pi (x,.) u(x)

 p (x, y) |x − y|N +spi (x,y)
 Q i

 i=1X 2
PMi s

 + |u|p̄i (x)−2 u = f (x, u) in Ω,



 i=1

u = 0 in RN \ Ω,
where:
• Ω ⊂ RN is a Lipschitz bounded open domain and Q := R2N \ (Ωc × Ωc ) with
Ωc = RN \ Ω, N ⩾ 3.
• pi : Q −→ (1, +∞) is a bounded continuous function, i = 1, 2, p̄(x) = p(x, x) for
any x ∈ Ω, and s ∈ (0, 1).
• Here, for i = 1, 2, the operator (−∆)spi (x,.) is the fractional pi (x, .)-Laplacian
defined as follows
Z
|u(x) − u(y)|pi (x,y)−2 (u(x) − u(y))
(−∆)spi (x,.) u(x) = p.v. dy for all x ∈ RN
R N |x − y| N +spi (x,y)

with p.v. is a commonly used abbreviation in the principal value sense.

• Mi : R+ −→ R+ , i = 1, 2, is a Kirchhoff function with the following assumptions
(K0 ) : Mi : R+ −→ R+ is a differentiable function and there exists mi > 0 such
that
Mi (t) ⩾ mi for all t ⩾ 0 i = 1, 2.
Fractional (p1 (x, .), p2 (x, .))-Kirchhoff type problems 2365

(K1 ) : For i = 1, 2, there exists αi ∈ (0, 1) such that

αi Mi (t) ⩾ (1 − αi )tMi′ (t) for all t ⩾ 0.

Note that (K1 ) implies that

Z t
c1 ) :
(K ci (t) ⩾ (1 − αi )Mi (t)t,
M c(t) =
for all t ⩾ 0, where M M (τ )dτ.
0
• The nonlinearity f : Ω × R −→ R is a Carathéodory function such that

(f0 ) : |f (x, t)| ⩽ c1 (1 + |t|q(x)−1 ) for all (x, t) ∈ Ω × R,

∗
where c1 is a positive constant and q ∈ C+ (Ω) such that 1 < q − ⩽ q(x) < (p̄max )s (x)
for any x ∈ Ω (see Notation 2.1 and Section 2).
F (x, t)
(f1 ) : lim + = 0, uniformly for a.e. x ∈ Ω, where α = min{α1 , α2 } and
|t|→∞ pmax
|t| 1−α
Z t
F (x, t) = f (x, τ )dτ.
0
(f2 ) : There exists θ ⩾ 1 such that

θH(x, t) ⩾ H(x, βt) for any (x, t) ∈ Ω × R and β ∈ [0, 1],

p+
where H(x, t) = f (x, t)t − max
F (x, t) (see Notation 2.1).
1−α
F (x, t)
(f3 ) : lim + = 0, uniformly for a.e. x ∈ Ω.
|t|pmax
t→0
Condition (f2 ) originates in the study of L. Jeanjean [28] in the case p(x, y) ≡ 2
for the Laplacian equation. This condition is crucial to obtain the compactness
condition of the Palais-Smale or Cerami type for an elliptic equation in the whole
space RN . In that context, these results for superlinear problems in bounded do-
mains have been initially investigated by Miyagaki and Souto [34], Motivated by
this work, many authors studied the existence of nontrivial solutions for nonlinear
elliptic problems under the following condition:
(f ∗ ): There is constant C∗ > 0 such that

tf (x, t) − pF (x, t) ≤ sf (x, s) − pF (x, s) + C∗

for any x ∈ Ω, 0 < t < s or s < t < 0.

In our study, we suppose that the nonlinearity f satisfies the condition (f2 )
instead of the well-known Ambrosetti-Rabinowitz (AR) type condition:
(AR): ∃γ > p+ max , L > 0 such that

0 ≤ γF (x, t) ≤ f (x, t)t for all x ∈ Ω, |t| ≥ L.

If (x, t) 7−→ f (x, t) is increasing in t, then (AR) implies (f2 ) when t is large enough.
1
In fact, we can take θ = > 1, then
p+
1 − γ(1−α)
max

θH(x, t) − H(x, βt) ≥ f (x, t)t − f (x, βt)βt ≥ 0.

But, in general, (AR) does not imply (f2 ), (see [38, Example 3.4]).
2366 E. Azroul, A. Benkirane, N. T. Chung, and M. Shimi

Now, it is worth mentioning that (−∆)sp(x,.) is a nonlocal pseudo-differential

operator of elliptic type which can be seen as a generalization of the fractional p-
Laplacian operator (−∆)sp in the constant exponent case (i.e., when p(x, y) = p =
constant). On the other hand, we remark that the above expression (
is the fractional)
version of the well-known p(x)-Laplacian operator ∆p(x) u(x) = div |∇u(x)|p(x)−2 u(x)
(where p(x) = p(x, x)) which is associated with the variable exponent Sobolev space.
For the Kirchhoff function M , a typical prototype is due to Kirchhoff in 1883,
and it is given by

M (t) = a + btα−1 , a, b ⩾ 0, a + b > 0, t ⩾ 0, (1.1)

and

 α ∈ (1, +∞) if b > 0,
α = 1 if b = 0,

when M (t) > 0 for all t ⩾ 0, Kirchhoff problems are said to be nondegenerate and
this happens for example if a > 0 and b ⩾ 0 in the model case (1.1). Otherwise, if
M (0) = 0 and M (t) > 0 for all t > 0, the Kirchhoff problems are called degenerate
and this occurs in the model case (1.1) when a = 0 and b > 0.
s
One typical feature of problem (PM i
) is the nonlocality, in the sense that the
value of (−∆)p(x,.) u(x) at any point x ∈ Ω depends not only on the values of u on
s

Ω, but actually on the entire space RN . Moreover, the presence of the functions
s
Mi , i = 1, 2, which implies that the first equation in (PM i
) is no longer a pointwise
equation, it is no longer a pointwise identity, Therefore, the Dirichlet datum is
given in RN \ Ω (which is different from the classical case of the p(x)-Laplacian)
and not simply on ∂Ω. Hence, it is often called nonlocal problem. This causes
some mathematical difficulties which make the study of such a problem particularly
interesting.
As far as we know, there is no work that deals with a nonlocal problem involv-
ing fractional (p1 (x, .), p2 (x, .))-Laplacian operator except [23] in which the authors
considered problem (PM s
i
) for the case M1 = M2 ≡ 1 and they established some
existence results for the problem with indefinite weights in an appropriate space
of functions by means of variational techniques and Ekeland’s variational princi-
ple. Moreover, in [6], using mountain pass theorem, the authors studied the ex-
istence of weak solutions for a quasilinear elliptic system involving the fractional
(p(x, .), q(x, .))-Laplacian operators. Very recently, the authors in [12] studied the
equation (−∆)sp(x,.) u(x) = f (x, u(x)) without assuming the (AR) type condition.
Therefore, without this condition it becomes a very difficult task to get the compact-
ness condition. That is why, to our best knowledge, the present studied anisotropic
Kirchhoff type problem is the first contribution in this direction. The purpose of
this work is to improve the results of the above-mentioned papers. So, using the
weaker assumption (f2 ) instead of (AR)-condition and some variant min-max the-
orem, we overcome these difficulties and we prove the existence and multiplicity of
s
weak solutions for problem (PM i
). Hence, our main results can be stated as follows.
Theorem 1.1. Assume that the assumptions (K0 ), (K1 ) and (f0 )-(f3 ) hold. If
−
p+ s
max < q , then problem (PMi ) has at least one nontrivial solution.

Theorem 1.2. Assume that (K0 ), (K1 ), and (f0 )-(f3 ) are satisfied. Moreover, we
Fractional (p1 (x, .), p2 (x, .))-Kirchhoff type problems 2367

suppose that
(f4 ) : f (x, −t) = −f (x, t) for all (x, t) ∈ Ω × R.
−
If q > p+
max ,
s
then problem (PM i
) has a sequence of weak solutions {±uk }∞
k=1 such
that
J (±uk ) −→ +∞ as k → +∞.
The rest of this paper is organized as follows: In section 2, we give some def-
initions and fundamental properties of generalized Lebesgue spaces and fractional
Sobolev spaces with variable exponent. In section 3, we discuss the existence of
s
nontrivial weak solutions of problem (PM i
) by means of mountain pass theorem
with Cerami condition. Furthermore, using Fountain theorem, we show that prob-
s
lem (PM i
) has infinitely many (pairs) of solutions with unbounded energy. As a
conclusion, we extend all our results directly to the fractional multi p(x, .)-Laplacian
case. Moreover, in order to illustrate our results, we consider a particular example
of the Kirchhoff functions Mi and the nonlinearity f .

2. Variatoinal setting and preliminary results

For the reader’s convenience, we briefly review the definitions and list some useful
properties of the generalized Lebesgue spaces. Furthermore, we recall some qualita-
tive properties of the fractional Sobolev spaces with variable exponent and several
important properties of fractional p(x, .)-Laplacian operator.

2.1. Variable exponent Lebesgue spaces

In this subsection, we give some basic results of variable exponent Lebesgue spaces
Lq(·) (Ω). For more details, we refer the reader to [25,30] and the references therein.
Consider the set

C+ (Ω) = q ∈ C(Ω) : q(x) > 1 for all x ∈ Ω .

For all q ∈ C+ (Ω), we define

q + = sup q(x) and q − = inf q(x),
x∈Ω x∈Ω

such that
1 < q − ⩽ q(x) ⩽ q + < +∞. (2.1)
For any q ∈ C+ (Ω), we define the variable exponent Lebesgue space as
 Z 
L q(x)
(Ω) = u : Ω −→ R measurable : |u(x)|q(x)
dx < +∞ .
Ω

This vector space endowed with the Luxemburg norm, which is defined by
 Z q(x) 
u(x)
kukLq(x) (Ω) = inf λ > 0 : dx ⩽ 1
Ω λ
is a separable reflexive Banach space.
1 1
Let q̂ ∈ C+ (Ω) be the conjugate exponent of q, that is, + = 1. Then
q(x) q̂(x)
we have the following Hölder-type inequality.
2368 E. Azroul, A. Benkirane, N. T. Chung, and M. Shimi

Lemma 2.1. If u ∈ Lq(·) (Ω) and v ∈ Lq̂(·) (Ω), then

Z  
1 1
uvdx ⩽ + kukLq(·) (Ω) kvkLq̂(·) (Ω) ⩽ 2kukLq(·) (Ω) kvkLq̂(·) (Ω) .
Ω q− q̂ −

A very important role in manipulating the generalized Lebesgue spaces with

variable exponent is played by the modular of the Lq(·) (Ω) space, which is defined
by

ρq(.) : Lq(·) (Ω) −→ R

Z
u 7−→ ρq(.) (u) = |u(x)|q(x) dx.
Ω

Proposition 2.1. Let u ∈ Lq(·) (Ω), {uk } ⊂ Lq(·) (Ω), k ∈ N, then we have
(i) kukLq(·) (Ω) < 1 (resp. = 1, > 1) ⇔ ρq(.) (u) < 1 (resp. = 1, > 1),
(ii) kukLq(·) (Ω) < 1 ⇒ kukq+
Lq(·) (Ω)
⩽ ρq(.) (u) ⩽ kukq−
Lq(·) (Ω)
,
(iii) kukLq(·) (Ω) > 1 ⇒ kukq−
Lq(·) (Ω)
⩽ ρq(.) (u) ⩽ kukq+
Lq(·) (Ω)
,
(iv) lim kuk − ukLq(·) (Ω) = 0 ⇐⇒ lim ρq(.) (uk − u) = 0.
k→+∞ k→+∞

2.2. Fractional Sobolev spaces with variable exponent

In this subsection, we present some preliminary results and basic properties of
fractional Sobolev spaces with variable exponent that were introduced in [10]. For
a deeper treatment on these spaces, we refer the reader to [9, 14, 29].
Let Ω be a Lipschitz bounded open set in RN . We denote by Q the set

Q := RN × RN \ (Ωc × Ωc ), where Ωc = RN \ Ω.

Let p : Q −→ (1, +∞) be a continuous bounded function, we assume that

1 < p− = min p(x, y) ⩽ p(x, y) ⩽ p+ = max p(x, y) < +∞ (2.2)

(x,y)∈Q (x,y)∈Q

and
p is symmetric, that is, p(x, y) = p(y, x) for all (x, y) ∈ Q. (2.3)
We set
p̄(x) = p(x, x) for any x ∈ Ω.
Throughout this paper, s is a fixed real number such that 0 < s < 1.
Due to the non-locality of the operator (−∆)sp(x,.) , we introduce the general
fractional Sobolev space with variable exponent as in [10] as follows
 

 u : RN −→ R measurable such that u|Ω ∈ Lp̄(x) (Ω) with 

X =W s,p(x,y)
(Q) = Z |u(x) − u(y)|p(x,y) ,

 dxdy < +∞, for some λ > 0 

Q λ
p(x,y) |x − y|N +sp(x,y)

with the norm

kukX = kukLp̄(x) (Ω) + [u]X ,
Fractional (p1 (x, .), p2 (x, .))-Kirchhoff type problems 2369

where [.]X is a Gagliardo seminorm with variable exponent, defined by

 Z 
|u(x) − u(y)|p(x,y)
[u]X = [u]s,p(x,y) (Q) = inf λ > 0 : dxdy ⩽ 1 .
Q λp(x,y) |x − y|N +sp(x,y)

The space (X, k.kX ) is a separable reflexive Banach space.

Definition 2.1. Let p : Q −→ (1, +∞) be a continuous variable exponent and let
s ∈ (0, 1), we define the modular ρp(x,y) : X −→ R, by
Z Z
|u(x) − u(y)|p(x,y)
ρp(x,y) (u) = dxdy + |u(x)|p̄(x) dx.
Q |x − y|N +sp(x,y) Ω
   
u
Consequently, kukρp(x,y) = inf λ > 0 : ρp(x,y) ⩽1 .
λ
Now, let us denote by X0 the following linear subspace of X

X0 = u ∈ X : u = 0 a.e. in RN \ Ω ,

with the norm

 Z 
|u(x) − u(y)|p(x,y)
kukX0 = [u]X = inf λ > 0 : dxdy ⩽ 1 .
Q λp(x,y) |x − y|N +sp(x,y)

The space (X0 , k.kX0 ) is a separable reflexive Banach space (see [10, Lemma 2.3]).
We define the modular ρ0p(x,y) : X0 −→ R, by
Z
|u(x) − u(y)|p(x,y)
ρ0p(x,y) (u) = dxdy.
Q |x − y|N +sp(x,y)
   
u
Consequently, kukρ0p(x,y) = inf λ > 0 : ⩽ 1 = [u]X . Similar to Propo-
ρ0p(x,y)
λ
0
sition 2.1, ρp(x,y) satisfies the following assertions.
Lemma 2.2. For any u ∈ X0 and {uk } ⊂ X0 , we have

(i) kukX0 < 1 (resp. = 1, > 1) ⇔ ρ0p(x,y) (u) < 1 (resp. = 1, > 1),
u
(ii) for u ∈ X0 \ {0}, kukX0 = λ ⇔ ρ0p(x,y) = 1,
λ
− +
(iii) 1 ⩽ kukX0 ⇒ kukpX0 ⩽ ρ0p(x,y) (u) ⩽ kukpX0 ,
+ −
(iv) kukX0 ⩽ 1 ⇒ kukpX0 ⩽ ρ0p(x,y) (u) ⩽ kukpX0 .
(v) kuk kX0 → 0 (resp → ∞) ⇔ ρp(x,y)
0
(uk ) → 0 (resp → ∞).

In [29], the authors introduced the variable exponent Sobolev fractional space
as follows

F = W s,q(x),p(x,y) (Ω)
 Z 
|u(x) − u(y)|p(x,y)
= u ∈ Lq(x) (Ω) : dxdy < +∞, for some λ > 0 ,
Ω×Ω λp(x,y) |x − y|N +sp(x,y)
2370 E. Azroul, A. Benkirane, N. T. Chung, and M. Shimi

where q : Ω −→ (1, +∞) is a continuous function satisfies (2.1).

We would like to mention that a continuous and compact embedding theorem is
proved in [29] under the assumption q(x) > p̄(x) = p(x, x). The authors in [9] give a
slightly different version of continuous compact embedding theorem assuming that
q(x) = p̄(x) = p(x, x), in this case the space E becomes

W = W s,p(x,y) (Ω)
 Z 
|u(x) − u(y)|p(x,y)
= u∈L p̄(x)
(Ω) : dxdy < +∞, for some λ > 0 .
Ω×Ω λp(x,y) |x − y|N +sp(x,y)

Theorem 2.1 ( [9]). Let Ω be a Lipschitz bounded domain in RN and let s ∈ (0, 1).
Let p : Ω × Ω −→ (1, +∞) be a continuous function satisfying (2.2) and (2.3) with
sp+ < N . Let r : Ω −→ (1, +∞) be a continuous variable exponent such that

N p̄(x)
1 < r− = min r(x) ⩽ r(x) < p∗s (x) = for all x ∈ Ω.
x∈Ω N − sp̄(x)

Then, there exists a constant C = C(N, s, p, r, Ω) > 0 such that, for any u ∈ W ,

kukLr(x) (Ω) ⩽ CkukW .

That is, the space W is continuously embedded in Lr(x) (Ω). Moreover, this embed-
ding is compact.
In [10], we compared the spaces W and X, and we established the compact and
continuous embedding of X into Lebesgue spaces with variable exponent.
Theorem 2.2. Let Ω be a Lipschitz bounded domain in RN and let s ∈ (0, 1). Let
p : Q −→ (1, +∞) be a continuous function satisfying (2.2) and (2.3) on Q with
sp+ < N . Then the following assertions hold:

(i) If u ∈ X, then u ∈ W . Moreover,

kukW ⩽ kukX ;

(ii) if u ∈ X0 , then u ∈ W s,p(x,y) (RN ). Moreover,

kukW ⩽ kukW s,p(x,y) (RN ) = kukX ;

(iii) if r : Ω −→ (1, +∞) be a continuous variable exponent such that

N p̄(x)
1 < r− ⩽ r(x) < p∗s (x) = for all x ∈ Ω.
N − sp̄(x)

Then, there exists a constant C = C(N, s, p, r, Ω) > 0 such that, for any
u ∈ X,
kukLr(x) (Ω) ⩽ CkukX .

That is, the space X is continuously embedded in Lr(x) (Ω). Moreover, this
embedding is compact.

Remark 2.1.
Fractional (p1 (x, .), p2 (x, .))-Kirchhoff type problems 2371

(i) The assertion (iii) in Theorem 2.2 remains true if we replace X by X0 .

(ii) Since 1 < p− ⩽ p̄(x) < p∗s (x) for all x ∈ Ω, then by Theorem 2.2-(iii) we have
that, k.kX0 = [.]X and k.kX are equivalent on X0 .
Let denote by L the operator associated to the (−∆p(x,.) )s defined as

L : X0 −→ X0∗
u 7−→ L(u) : X0 −→ R
ϕ 7−→ hL(u), ϕi
such that
Z
|u(x) − u(y)|p(x,y)−2 (u(x) − u(y))(ϕ(x) − ϕ(y))
hL(u), ϕi = dxdy,
Q |x − y|N +sp(x,y)
where h., .i denotes the usual duality between X0 and its dual space X0∗ .
Lemma 2.3 ( [14]). Assume that assumptions (2.2) and (2.3) are satisfied. Then,
the following assertions hold:
(i) L is a bounded and strictly monotone operator.
(ii) L is a mapping of type (S+ ), that is,
if uk * u in X0 and lim sup ⟨L(uk ) − L(u), uk − u⟩ ⩽ 0, then uk −→ u in X0 .
k−→+∞

(iii) L is a homeomorphism.
s
In order to facilitate the investigation of problem (PM i
), the following notations
are required.
Notation 2.1. For all (x, y) ∈ Q, let us denote
• pmax (x, y) = max {p1 (x, y), p2 (x, y)} = max pi (x, y),
i=1,2

• pmin (x, y) = min pi (x, y),

i=1,2

• p̄max (x) = max p̄i (x) = max pi (x, x), p̄min (x) = min p̄i (x),
i=1,2 i=1,2 i=1,2

• p+
max = sup pmax (x, y) and p−
min = inf pmin (x, y),
(x,y)∈Q (x,y)∈Q

∗ N p̄max (x)
• (p̄max )s (x) = for any x ∈ Ω.
N − sp̄max (x)
It is easy to see that p̄max , p̄min ∈ C+ (Ω) and pmax , pmin ∈ C+ (Q). For simplicity,
we set E = W s,pmax (x,y) (Ω) and

E0 = u ∈ E : u = 0 a.e. in RN \ Ω .
It is clear that E and E0 are separable and reflexive Banach spaces under the norms
kukE = kukLp̄(x) (Ω) + [u]s,pmax (x,y) and kukE0 = [u]s,pmax (x,y) .
s,pi (x,y)
For i = 1, 2, we denote by ρ0pi (x,y) the modular on W0 (Q) defined by
Z
|u(x) − u(y)|pi (x,y)
ρ0pi (x,y) (u) = dxdy.
Q |x − y|N +spi (x,y)
2372 E. Azroul, A. Benkirane, N. T. Chung, and M. Shimi

Remark 2.2. Obviously, from Theorem 2.2, for any q ∈ C + (Ω) such that q(x) <
∗
(p̄max )s (x) for all x ∈ Ω, we have E ,→ Lq(x) (Ω), and this embedding is continuous
and compact. Moreover, by Remark 2.1-(i), this result remains true if we replace
E by E0 .
Now, we give the definition of the Cerami condition (C) which is introduced by
Cerami in [20].
Definition 2.2. Let X be a Banach space and J ∈ C 1 (X, R). Given c ∈ R, we say
that Φ satisfies the Cerami c condition (we denote condition (Cc )), if

(C1 ): any bounded sequence {un } ⊂ X such that Φ (un ) → c and Φ′ (un ) → 0 has
a convergent subsequence,
(C2 ): there exist constants δ, R, β > 0 such that

kΦ′ (u)k kuk ≥ β ∀u ∈ Φ−1 ([c − δ, c + δ]) with kuk ≥ R.

If Φ ∈ C 1 (X, R) satisfies condition (Cc ) for every c ∈ R, we say that Φ satisfies

condition (C).

Note that condition (C) is weaker than the Palais-Smale condition. However,
it was shown in [15] that from condition (C) it is possible to obtain a deformation
lemma, which is fundamental in order to get some min-max theorems. More pre-
cisely, let us recall the following version of the mountain pass lemma with Cerami
condition which will be used in the sequel.
Proposition 2.2. Let X a Banach space, Φ ∈ C 1 (X, R), u0 ∈ X and r > 0, be
such that ku0 kX > r and

b := inf Φ(u) > Φ(0) ≥ Φ(u0 ).

∥u∥X =r

If Φ satisfies the condition (Cc ) with

c := inf max Φ(γ(t)),

γ∈Γ t∈[0,1]

Γ := {γ ∈ C ([0, 1], X) |γ(0) = 0, γ(1) = u0 }.

Then c is a critical value of Φ.

Now, since X is a separable and reflexive Banach space, from [39, Section 17,
Theorems 2-3], there exist {en }∞ ∗ ∞ ∗
n=1 ⊂ X and {en }n=1 ⊂ X such that

 1 if n = m,
e∗n (em ) = δn,m =
 0 if n =
6 m.

Hence, X = span{en , n = 1, 2, ...} and X ∗ = span{e∗n , n = 1, 2, ...}. For k =

1, 2, ..., denote

Xk = span{ek }, Yk = ⊕ki=0 Xi , Z k = ⊕∞
i=k Xi .

Next, we introduce the Fountain theorem with the condition (C) as in [41].
Fractional (p1 (x, .), p2 (x, .))-Kirchhoff type problems 2373

Proposition 2.3. Assume that X is a separable Banach space, Φ ∈ C 1 (X, R)

is an even functional satisfying the Cerami condition (C). Moreover, for each
k = 1, 2, . . . , there exist Rk > rk > 0 such that

• (Φ1 ) inf Φ(u) → +∞ as k → ∞,

{u∈Zk :∥u∥=rk }

• (Φ2 ) max Φ(u) ≤ 0.

|u∈Yk :∥u∥=Rk }

Then, Φ has a sequence of critical values which tends to +∞.

3. Proof of Existence and multiplicity results

s
By a weak solution for (PM i
), we mean a function u ∈ E0 such that
Z


X
2

|u(x) − u(y)|pi (x,y)−2 u(x) − u(y) ϕ(x) − ϕ(y)
Mi σpi (x,y) (u) dxdy
i=1 Q |x − y|N +spi (x,y)
2 Z
X Z
+ |u(x)| p̄i (x)−2
u(x)ϕ(x)dx − f (x, u)ϕdx = 0,
i=1 Ω Ω
(3.1)
Z
1 |u(x) − u(y)|pi (x,y)
for all ϕ ∈ E0 , where σpi (x,y) (u) = dxdy. In this case,
Q pi (x, y) |x − y|
N +spi (x,y)

the weak formulation (3.1) is the Euler-Lagrange equation of the energy functional
J : E0 −→ R defined by

X Z Z
2

X2
1
J (u) = ci σp (x,y) (u) +
M |u|p̄i (x) dx − F (x, u)dx.
i
i=1 i=1 Ω p̄i (x) Ω

Standard arguments (see, for instance [9, Lemma 3.1]) and the continuity of Mi ,
i = 1, 2, imply that J is well defined and J ∈ C 1 (E0 , R). Moreover, for all u, ϕ ∈
E0 , its Gateaux derivative is given by
Z


X
2

|u(x) − u(y)|pi (x,y)−2 u(x) − u(y) ϕ(x) − ϕ(y)
′
hJ (u), ϕi = Mi σpi (x,y) (u) dxdy
i=1 Q |x − y|N +spi (x,y)
2 Z
X Z
+ |u|p̄i (x)−2
uϕdx − f (x, u)ϕdx.
i=1 Ω Ω

s
Thus, the weak solutions of (PM i
) coincide with the critical points of J .

3.1. Compactness Cerami condition for the functional J

In this subsection, we establish the following compactness result which plays the
most important role in this chapter.
Lemma 3.1. Suppose that the conditions (K0 ), (K1 ), and (f0 )-(f2 ) hold. Then,
J satisfies the Cerami condition (Cc ).
2374 E. Azroul, A. Benkirane, N. T. Chung, and M. Shimi

Proof. We first show that J satisfies the assertion (C1 ) of Cerami condition (Cc )
(see Definition 2.2). Indeed, for all c ∈ R, let {un } ⊂ E0 be a bounded sequence
such that
J (un ) −→ c and J ′ (un ) −→ 0. (3.2)
n→+∞ n→+∞

Since E0 is a reflexive space, then without loss of generality, we can assume that
un * u in E0 , which implies that

hJ ′ (un ), un − ui −→ 0.
n→+∞

Thus, we have

hJ ′ (un ), un − ui =
X
2


Mi σpi (x,y) (un ) ×
i=1
Z
|un (x) − un (y)|pi (x,y)−2 (un (x) − un (y)) ((un (x) − un (y)) − (u(x) − u(y)))
dxdy
Q |x − y|N +spi (x,y)
2 Z
X Z
+ |un |p̄i (x)−2 un (un − u)dx − f (x, un )(un − u)dx −→ 0. (3.3)
Ω Ω n→+∞
i=1

On the other hand, by (f0 ) and Hölder’s inequality in Lemma 2.1, we have
Z
f (x, un )(un−u)dx ⩽ 2c1 k1kLq̂(x) (Ω) kun−ukLq(x) (Ω)+2c1 kun kLq(x) (Ω) kun−ukLq(x) (Ω) ,
Ω

1 1 ∗
where + = 1. Hence, as 1 < q − ⩽ q(x) < (p̄max )s (x) for all x ∈ Ω, we
q(x) q̂(x)
have that E0 is compactly embedded in Lq(x) (Ω). It follows that
Z
f (x, un )(un − u)dx −→ 0. (3.4)
Ω n→+∞

∗
Besides this, since 1 < p−
i ⩽ p̄i (x) < (p̄max )s (x), i = 1, 2, for all x ∈ Ω, then E0 is
compactly embedded in Lp̄i (x) (Ω), for i = 1, 2. So, again by Hölder’s inequality in
Lemma 2.1, we get
2 Z
X
|un |p̄i (x)−2 un (un − u)dx −→ 0. (3.5)
Ω n→+∞
i=1

Combining (3.3)-(3.5), we obtain

X
2


lim Mi σpi (x,y) (un ) ×
n→+∞
i=1
Z
|un (x) − un (y)|pi (x,y)−2 (un (x) − un (y)) ((un (x) − un (y)) − (u(x) − u(y)))
dxdy = 0.
Q |x − y|N +spi (x,y)

c1 ), for i = 1, 2, we can easily obtain that

Using (K

ci (t) ⩽ c̃i t 1−αi .

1
M
Fractional (p1 (x, .), p2 (x, .))-Kirchhoff type problems 2375

Hence, from (K0 ), it follows that

ci (t) ⩽ c̃i αi
m0 ⩽ Mi (t) ⩽ M t 1−αi .
(1 − αi )


Since {un } ⊂ E0 and
u ∈ E0 , by Lemma 2.2, we deduce that Mi σpi (x,y) (un )
and Mi σpi (x,y) (u) are bounded. Thence, by assumption (K0 ), we get

2 Z
X |un (x) − un (y)|pi (x,y)−2 (un (x) − un (y)) ((un (x) − un (y)) − (u(x) − u(y)))
dxdy
Q |x − y|N +spi (x,y)
i=1
D E
e n ), un − u
:= L(u −→ 0 (3.6)
n→+∞

Now, since un * u in E0 , using (3.2), we get

hJ ′ (u), un − ui −→ 0.
n→+∞

Hence, by the same argument as before, we deduce that

X2 Z
|u(x) − u(y)|pi (x,y)−2 (u(x) − u(y)) ((un (x) − un (y)) − (u(x) − u(y)))
dxdy
i=1 Q
|x − y|N +spi (x,y)
D E
e
:= L(u), un − u −→ 0 (3.7)
n→+∞

Combining (3.6) and (3.7), we conclude that

D E
e n ) − L(u),
lim sup L(u e un − u ⩽ 0.
n→+∞

From Lemma 2.3-(iii), L is a mapping of type (S+ ), and since L e is a sum of two
e
operators of type (S+ ). Then, by [26, Lemma 6.8-(b)], L is also of type (S+ ). Hence
 D E
 e n ) − L(u),
 lim sup L(u e un − u ⩽ 0,


 n−→+∞
 un * u in E0 ,



L e is a mapping of type (S+ ).

Consequently, un −→ u (strongly) in E0 .

Next, we show that J satisfies the assertion (C2 ) of Cerami condition (Cc ) (see
Definition 2.2), we argue by contradiction. Indeed, we assume that there exists
c ∈ R and {un } ⊂ E0 such that

J (un ) −→ c, kun kE0 −→ ∞, and kJ ′ (un )kE0∗ kun kE0 −→ 0. (3.8)

n→+∞ n→+∞ n→+∞

From (3.8), it is easy to see that

1−α
J (un ) − hJ ′ (un ), un i −→ c. (3.9)
p+
max n→+∞
2376 E. Azroul, A. Benkirane, N. T. Chung, and M. Shimi

un
Denote ϕn = , then kϕn kE0 = 1, which implies that {ϕn } is bounded in E0 .
kun kE0
Hence, for a subsequence of {ϕn }, still denoted by {ϕn }, and ϕ ∈ E0 , we get

ϕn * ϕ in E0 , (3.10)
ϕn −→ ϕ in L q(x)
(Ω), (3.11)
ϕn (x) −→ ϕ(x) a.e. in Ω, (3.12)

where q is given in assumption (f0 ).

• If ϕ = 0, as in [28], we can define a sequence {tn } ⊂ R such that

J (tn un ) = max J (tun ).

t∈[0,1]

If there is a number of tn satisfying the above equality, one choose one of them. Fix
1
−1
any A > + , let vn = 2Ap+ max
pmax ϕ . By (3.11), we get
n
2pmax

vn −→ 0 in Lq(x) (Ω).

From (f0 ), we have  

|F (x, t)| ⩽ c1 1 + |t|q(x) .
Hence, by the continuity of t 7−→ F (., t), we have

F (., vn ) −→ 0 in L1 (Ω).
n→+∞

Therefore, Z
lim F (x, vn )dx = 0. (3.13)
n→+∞ Ω

−1
2Ap+
pmax
pmax

Then, for n large enough, ∈ (0, 1), using (K0 ), Lemma 2.2, Propo-
kun kE0
sition 2.1, and Remark 2.2, we obtain

J (tn un ) ⩾J (vn )
X Z Z
2

X2
1
= ci σp (x,y) (vn ) +
M |vn |p̄i (x) dx − F (x, vn )dx
i
i=1 i=1 Ω i
p̄ (x) Ω

X2 Z X 2 Z
mi |vn (x) − vn (y)|pi (x,y) 1
⩾ dxdy + |vn |p̄i (x) dx
i=1
p+
i Q |x − y|N +spi (x,y)
i=1
p +
i Ω
Z
− F (x, vn )dx
Ω
X Z
2
mi
|ϕn (x) − ϕn (y)|pi (x,y)
⩾ 2Ap+ dxdy
i=1
p+
max Q
max
|x − y|N +spi (x,y)
X Z Z
2
1

+ 2Apmax |ϕn |
+ p̄i (x)
dx − F (x, vn )dx
p+
max Ω
i=1 Ω

X2 Z X2 Z
|ϕn (x) − ϕn (y)|pi (x,y)
⩾2A mi dxdy + 2A |ϕn |p̄i (x) dx
i=1 Q |x − y| N +spi (x,y)
i=1 Ω
Fractional (p1 (x, .), p2 (x, .))-Kirchhoff type problems 2377
Z
− F (x, vn )dx
Ω
X
2 X
2 Z
p− p−
⩾2A mi kϕn kEi0 + 2A kϕn kLip̄i (x) (Ω) − F (x, vn )dx
i=1 i=1 Ω

X
2 X
2 Z
p− p− p−
⩾2A mi kϕn kEi0 + 2A c̄i i kϕn kEi0 − F (x, vn )dx
i=1 i=1 Ω
Z
p− p−
⩾2A min{m1 , m2 } + 2A min{c̄11 , c̄22 } − F (x, vn )dx,
Ω

that is,
J (tn un ) −→ +∞, as n → +∞. (3.14)
Since J (0) = 0 and J (un ) −→ c, then
n→+∞

d
tn ∈ (0, 1) and hJ ′ (tn un ), tn un i = tn J (tn un ) = 0. (3.15)
dt t=tn

From (3.9) and (f2 ), we have

 
1−α
c = lim J (un ) − + hJ ′ (un ), un i
n→+∞ pmax
( 2 2 Z Z
X
X 1
= lim ci σp (x,y) (un ) +
M |un |p̄i (x)
dx − F (x, un )dx
i
n→+∞
i=1 i=1 Ω i
p̄ (x) Ω
2 Z
1−α X
X
2
− + Mi σpi (x,y) (un ) ρ0pi (x,y) (un ) + |un |p̄i (x) dx
pmax i=1 i=1 Ω
Z 
− f (x, un )un dx ,
Ω
( 2 2 Z
X
X 1
c = lim c
Mi σpi (x,y) (un ) + |un |p̄i (x) dx
n→+∞
i=1 i=1 Ω i
p̄ (x)
2 Z
1−α X
X
2
− + Mi σpi (x,y) (un ) ρ0pi (x,y) (un ) + |un |p̄i (x) dx (3.16)
pmax i=1 i=1 Ω
Z 
− H(x, un )dx .
Ω

Now, we consider the following function

   
ci σp (x,y) (tu) − 1 − α Mi σp (x,y) (tu) ρ0
t ∈ (0, 1) 7→ θi (t) = M pi (x,y) (tu).
i
p+max
i

For i = 1, 2, using (K1 ), we obtain

  dMi  
αMi σpi (x,y) (tu) ⩾ (1 − α) σpi (x,y) (tu) σpi (x,y) (tu).
dt
dθi (t)
This fact implies that ⩾ 0 for all t ⩾ 0, i = 1, 2. Hence t 7→ θi (t) is increasing.
dt
2378 E. Azroul, A. Benkirane, N. T. Chung, and M. Shimi

Thus, from (f2 ) and (3.16) and since tn ∈ (0, 1), we obtain
( 2 2 Z
X
X 1
c ⩾ lim Mci σp (x,y) (tn un ) + |tn un |p̄i (x) dx
i
n→+∞
i=1 i=1 Ω p̄i (x)
X X2 Z
1−α
2

0
− + Mi σpi (x,y) (tn un ) ρpi (x,y) (tn un ) + |tn un |p̄i (x) dx
pmax i=1 i=1 Ω
Z 
H(x, tn un )
− dx
Ω θ
 
1 1−α
⩾ lim J (tn un ) − + hJ ′ (tn un ), tn un i .
n→+∞ θ pmax

Then, by (3.14) and (3.15), we get a contradiction.

• If ϕ 6= 0, then the set Ω0 = {x ∈ Ω : ϕ(x) 6= 0} has a positive Lebesgue measure.
For x ∈ Ω0 , we have |un (x)| −→ +∞. Thus, by (f1 ), we get
n→+∞

F (x, un (x)) p+
max
+ |ϕn (x)| 1−α −→ +∞. (3.17)
pmax n→+∞
|un (x)| 1−α

c1 ), for i = 1, 2, we can easily deduce that

Using (K

ci (t) ⩽ c̃i t 1−αi ,

1
M (3.18)

where c̃i is a positive constant. Since J (un ) −→ c, using (3.18), Lemma 2.2,
n→+∞
Proposition 2.1, and the continuous embedding of E0 into Lp̄i (x) (Ω), i = 1, 2, we
deduce via the Fatou lemma that
Z
c̃1 + c̃2 c̄1 + c̄2 c + o(1) F (x, un )
− 1 + − − + ⩾ + dx
(pmin ) 1−α pmin pmax
Ω
pmax

kun kE1−α ku k
n E0
1−α
0
Z
F (x, un ) p+
max
= +
pmax
|ϕ(x)| 1−α dx

φn ̸=0
|un (x)| 1−α
Z
F (x, un ) p+
max
+ +
pmax
|ϕ(x)| 1−α dx.

φn =0
|un (x)| 1−α

By (3.17), we obtain a contradiction.

3.2. Existence of weak solution via mountain pass theorem

By means of mountain pas theorem with Cerami condition given in Proposition 2.2,
we establish the first main result of this paper which is an existence theorem for
s
problem (PM i
) as stated in Theorem 1.1.
Proof of Theorem 1.1. By Lemma 3.1, J satisfies the Cerami condition (Cc )
in E0 . To apply Proposition 2.2, we will show that J possesses the mountain pass
geometry.
Fractional (p1 (x, .), p2 (x, .))-Kirchhoff type problems 2379

• Firstly, we claim that there exist R, a > 0 such that

J (u) ⩾ a for any u ∈ E0 with kukE0 = R. (3.19)

Indeed, Since p̄i (x) < (p̄max )∗s (x) for any x ∈ Ω, from Remark 2.2, we have that E0
embedded in Lp̄i (x) (Ω), that is, there exist c̄i > 0, i = 1, 2, such that

kukLp̄1 (x) (Ω) ⩽ c̄1 kukE0 and kukLp̄2 (x) (Ω) ⩽ c̄2 kukE0 . (3.20)

∗
max , q(x) < (p̄max )s (x) for any x ∈ Ω, then there exist c2 , c3 > 0 such
Moreover, as p+
that
kukLp+ max (Ω)
⩽ c2 kukE0 and kukLq(x) (Ω) ⩽ c3 kukE0 . (3.21)

p+ p+
p+ m1 + m2 + c̄11 + c̄22
Next, let ε > 0 be such that εc2max < . Combining (f0 ) and
2p+
max
(f3 ), we obtain
+
|F (x, t)| ⩽ ε|t|pmax + cε |t|q(x) for all (x, t) ∈ Ω × R. (3.22)

Using (3.20)-(3.22), (K0 ), Lemma 2.2, Proposition 2.1, and Remark 2.2, for all
kukE0 sufficiently small, we get

X Z Z
2

X2
1
J (u) = ci σp (x,y) (u) +
M |u|p̄i (x) dx − F (x, u)dx
i
i=1 i=1 Ω i
p̄ (x) Ω

X 2 Z Z Z
1 X
2
+
⩾ mi σpi (x,y) (u) + + |u|p̄i (x) dx − ε |u|pmax dx − cε |u|q(x) dx
i=1
p max i=1 Ω Ω Ω

1 X 1 X
2 2
p+ p+ −
⩾ mi ρ0pi (x,y) (u)+ kukLip̄i (x) (Ω) −εkuk max
+ −cε kukqLq(x) (Ω)
p+
max i=1 p+
max i=1 Lpmax (Ω)

1 X
2
p+ 1 X
2
p+ p+ p+ p+ − −
⩾ mi kukEi0 + c̄i i kukEi0 − εc2max kukEmax − cε cq3 kukqE0
p+
max i=1
+
pmax i=1 0

p+ p+
!
p+ m1 + m2 + c̄11 + c̄22 p+ − q − −p+
= kukEmax
0 + − εc2max − cε cq3 kukE0 max .
pmax

−
max < q , then there exist R ∈ (0, 1) and a > 0 such that (3.19) hold true.
As p+

• Secondly, we affirm that there exists u0 ∈ E0 \ BR6 (0) such that

J (u0 ) < 0. (3.23)

P2   1−α
1
1 0
p−
c̃
i=1 i ρ pi (x,y) (ϕ0 )
In fact, from (f1 ), we choose a constant B > min
R 1 , and
Ω
|ϕ0 | 1−α dx
a constant cB > 0 depending on B such that
p+
max
F (x, t) > B|t| 1−α , for all |t| > cB and uniformly in x ∈ Ω.
2380 E. Azroul, A. Benkirane, N. T. Chung, and M. Shimi

Let l > 1 be large enough, by the above inequality and (3.18), we have

X Z Z
2

X2
1
J (lϕ0 ) = ci σp (x,y) (lϕ0 ) +
M |lϕ0 |p̄i (x) dx − F (x, lϕ0 )dx
i
i=1 i=1 Ω i
p̄ (x) Ω
+ Z
X2 p
+
i
1−α
1 X 2
l pi
⩽ l 1−α c̃i σpi (x,y) (ϕ0 ) + − |ϕ0 |p̄i (x) dx
i=1
p i Ω
Z Z i=1
− F (x, lϕ0 )dx − F (x, lϕ0 )dx
{|lφ0 |>cB } {|lφ0 |⩽cB }
  1−α 2 Z
1 X lpmax X
2 p
+ 1 +
i
⩽ − l 0
c̃i ρpi (x,y) (ϕ0 )
1−α + − |ϕ0 |p̄i (x) dx
pmin i=1 pmin i=1 Ω
Z Z Z
p+
max p+
max p+max
− F (x, lϕ0 )dx−Bl 1−α |ϕ0 | 1−α dx + B |lϕ0 | 1−α dx
{|lφ0 |⩽cB } Ω {|lφ0 |⩽cB }
p+
X   1−α 2 Z
lpmax X
max 2 1 +
l1−α
⩽ c̃ ρ 0
(ϕ ) + |ϕ0 |p̄i (x) dx
p− −
i pi (x,y) 0
min i=1 p min i=1 Ω
Z
p+
max p+max
− Bl 1−α |ϕ0 | 1−α dx + c4 .
Ω

p+
max
Hence, as > p+
max , it follows
1−α

J (lϕ0 ) −→ −∞.
l→+∞

Consequently, there exist l0 > 1 and u0 = l0 ϕ0 ∈ E0 \ BR (0) such that (3.23)

hold true. Hence, in the light of mountain pass theorem with Cerami condition
(Proposition 2.2), we deduce that J has at least one nontrivial critical value, that
s
is, problem (PM i
) has at least one nontrivial solution. This completes the proof. □

s
3.3. Infinitely many solutions for problem (PM i
)
s
In this subsection, we provide a multiplicity result for problem (PM i
). The main
tools used here is the Fountain theorem with Cerami condition (see, Proposition
2.3).
Since E0 is a separable and reflexive Banach space, from [39, Section 17, Theo-
rems 2-3], there exist {en }∞ ∗ ∞ ∗
n=1 ⊂ E0 and {en }n=1 ⊂ E0 such that

 1 if n = m,
e∗n (em ) = δn,m =
 0 if n =
6 m.

Hence, E0 = span{en , n = 1, 2, · · · } and E0∗ = span{e∗n , n = 1, 2, · · · }. For

k = 1, 2, · · · , denote

E0k = span{ek }, Yk = ⊕ki=0 E0i , Z k = ⊕∞

i=k E0i .

To establish the proof of the above result, we need the following auxiliary lemma.
Fractional (p1 (x, .), p2 (x, .))-Kirchhoff type problems 2381

Lemma 3.2. Let r ∈ C+ (Ω) such that r(x) < (p̄max )∗s (x) for any x ∈ Ω, define

θk = sup kukLr(x) (Ω) : kukE0 = 1, u ∈ Zk .

Then lim θk = 0.
k→+∞

Proof. Clearly, 0 < θk+1 ⩽ θk , so θk −→ θ̄ ⩾ 0. Let uk ∈ Zk satisfy

1
kuk kE0 = 1 and 0 ⩽ θk − kuk kLr(x) (Ω) < .
k
Since E0 is a reflexive space, so {uk } has a weakly convergent subsequence, which
we still denoted by {uk }, we suppose that uk * u. We claim that u = 0. In fact,
for any e∗n , n = 1, 2, · · · , we have he∗n , uk i = 0 when k > n, hence he∗n , uk i −→ 0
as k → +∞, which implies that for any e∗n , n = 1, 2, · · · , he∗n , ui = 0. Therefore,
u = 0. That is, uk * 0 into E0 as k → +∞. By the compact embedding of E0 in
Lr(x) (Ω), we have that uk −→ 0 in Lr(x) (Ω). Thus, θk −→ 0 as k → +∞.
Now, we are ready to prove Theorem 1.2.
Proof of Theorem 1.2. We show that J verifies the assumptions of Fountain
theorem given in Proposition 2.3. Indeed, from condition (f4 ), J is an even function.
By Lemma 3.1, J satisfies conditon (Cc ) . Next, we verify that (Φ1 ) and (Φ2 ) in
Theorem 2.3 are satisfied.
(Φ1 ) : For u ∈ Zk such that kukE0 = rk > 1 (rk will be given below), using (K0 ),
(f0 ), Lemma 2.2, Proposition 2.1, and Remark 2.2, we get

X Z Z
2

X2
1
J (u) = ci σp (x,y) (u) +
M |u|p̄i (x) dx − F (x, u)dx
i
i=1 i=1 Ω i
p̄ (x) Ω

X2 X 2 Z
mi p− 1 p−
⩾ + kuk E0
i
+ + kuk i
Lp̄i (x) (Ω)
− c1 (|u| + |u|q(x) )dx
i=1
p max i=1
p max Ω
Z
1 X p−
2
m̄ p− p−
⩾ + kukE0 + + min
c̄ kukE0 − c1
i i
|u|q(x) dx − c1 kukE0
pmax pmax i=1 i Ω
Z
m̄ p− c̄ p−
⩾ + kukEmin + kuk E0
min
− c1 |u|q(x) dx − c1 kukE0
pmax 0
p+
max Ω
 m̄ + c̄
 p− q−
 + kukE0 − c1 kukLq(x) (Ω) − c1 kukE0 if kukLq(x) (Ω) ⩽ 1
min

⩾ pmax
 m̄ + c̄ −
 + kukpEmin − c1 kukq q(x)
+
− c1 kukE0 if kukLq(x) (Ω) > 1
0 L (Ω)
pmax
 m̄ + c̄
 p−
 + kukEmin 0
− c1 − c1 kukE0 if kukLq(x) (Ω) ⩽ 1
⩾ m̄pmax
 + c̄ −
 + kukpEmin − c1 (θk kukE0 )q − c1 kukE0 if kukLq(x) (Ω) > 1
+

0
pmax
m̄ + c̄ p− + +
⩾ + kukEmin 0
− c1 θkq kukqE0 − c1 kukE0 − c1 .
pmax
Thus, we obtain
 
p− m̄ + c̄ + q + −p−
J (u) ⩾ rkmin + − c1 θkq rk min
− c1 rk − c1 ,
pmax
2382 E. Azroul, A. Benkirane, N. T. Chung, and M. Shimi

p− p−
when m̄ = min{m1 , m2 } and c̄ = {c̄11 , c̄22 }. We fix rk as follows
+ ! −
1

c1 q + θkq p
min
−q +

rk = .
m̄ + c̄

It follows that
 
p− 1 1
J (u) ⩾ (m̄ + c̄)rkmin − − c1 rk − c1
p+
max q+
   
p− −1 1 1
= rk (m̄ + c̄)rkmin − − c 1 − c1 .
p+
max q+
Combining Lemma 3.2 with the fact that 1 < p−
min ⩽ pmax < q , we deduce that
+ +

rk −→ +∞ as k → +∞. Consequently,
J (u) −→ +∞ as kukE0 → +∞.
The assertion (Φ1 ) is verified.
(Φ2 ) : Since Yk = ⊕ki=0 E0i , then dim Yk < +∞ and as all norms are equivalents in
the finite-dimensional space, there exists bk > 0, for all u ∈ Yk with kukE0 ⩾ 1, by
(3.18) and Lemma 2.2, we have
X
2

X2

1
ci σp (x,y) (u) ⩽
M c̃i σpi (x,y) (u) 1−α
i
i=1 i=1

1 X
2 +
p
i
⩽ c̃i kukE1−α (3.24)
p−
min i=1
0

p+
max
⩽ bk kuk 1−α
+
pmax
.
L 1−α (Ω)

Using Proposition 2.1 and since p̄i (x) ⩽ p+

max , for all x ∈ Ω, we obtain
2 Z
X 2 Z
1 1 X
|u|p̄i (x)
dx ⩽ − |u|p̄i (x) dx
i=1 Ω p̄i (x) pmin i=1 Ω

1 X
2
p+
⩽ kukLip̄i (x)
p−
min i=1 (3.25)
1 X
2
p+ p+
⩽ − ci i kuk max
+
pmin i=1 Lpmax (Ω)

p+
⩽ c5 kuk max
+ .
Lpmax (Ω)

Combining assumptions (f1 ) and (f3 ), we deduce that

p+
max +
F (x, t) ⩾ 2bk |t| 1−α − ε|t|pmax for all (x, t) ∈ Ω × R. (3.26)
For (3.24)-(3.26), for any u ∈ Yk such that kukE0 = Rk > rk , we deduce that
X Z Z
2

X2
1
J (u) = ci σp (x,y) (u) +
M |u|p̄i (x) dx − F (x, u)dx
i
i=1 i=1 Ω p̄i (x) Ω
Fractional (p1 (x, .), p2 (x, .))-Kirchhoff type problems 2383

p+
max
p+
⩽ −bk kuk 1−α
+
pmax
+ (ε + c5 )kuk max
+
Lpmax (Ω)
L 1−α (Ω)
p+
max
p+
⩽ −c6 kukE1−α
0
+ c7 (ε + c5 )kukEmax
0
.

Hence, for Rk large enough (Rk > rk ), from the above fact, we conclude that

max J (u) ≤ 0,
{u∈Yk :∥u∥E0 =Rk }

which implies that the assertion (Φ2 ) is verified. Consequently, by the Fountain
theorem, we achieve the proof of Theorem 1.2. □

3.4. A multi fractional p(x, .)-Laplacian Kirchhoff type prob-

lem
We could extend all our results directly to the fractional multi p(x, .)-Laplacian
case by considering the fractional (p1 (x, .), p2 (x, .), · · · , pn (x, .))-Laplacian problem
of the following form
 n
 ∑ ( ) ∑
n

 Mi σpi (x,y) (u) (−∆)spi (x,.) u(x) + |u|p̄i (x)−2 u = f (x, u) in Ω,
(Psn ) i=1 i=1


 u=0 in RN \ Ω.

✠ As a particular case of Kirchhoff functions Mi , we consider

Mi (t) = ai + bi t, ai > 0, bi ⩾ 0 for all t ⩾ 0, i = 1, · · · , n.

✠ For the nonlinearity f , we can take the function

p+ p+
|t| 1−α −2 t ln(|t|),
max max
f (x, t) =
1−µ
where α = min{α1 , α2 }. Note that the above function does not satisfy (AR). But
it is easy to see that it is satisfies (f0 )-(f4 ). In this case problem (Psn ) becomes
 n
 X


 s


ai + bi σpi (x,y) (u) (−∆)pi (x,.) u(x)
 i=1


s,n
(PK ) Xn
p+ p+
 1−α −2 t ln(|t|)
max

 + |u| p̄i (x)−2
u = max
|t| in Ω,

 i=1 1−µ



u = 0 in RN \ Ω.

It is clear that
• Mi (t) ⩾ ai for all t ⩾ 0, i = 1, · · · , n.
• Mi′ (t) = bi for all t ⩾ 0, i = 1, · · · , n.
1
If we take in (K1 ), αi = , for all i = 1, . . . , n. It follows that Mi satisfies the
2
assumptions (K0 ) and (K1 ). Therefore, the results obtained in Theorems 1.1 and
s,n
1.2 stay true for problem (PK ). The problem and results are all new.
2384 E. Azroul, A. Benkirane, N. T. Chung, and M. Shimi

Acknowledgments
The authors would like to thank the anonymous referee who provided useful and
detailed comments on this manuscript.

References
[1] G. A. Afrouzi, M. Mirzapour and N. T. Chung, Existence and multiplicity of
solutions for a p(x)-Kirchhoff type equation, Rend. Semin. Mat. Univ. Padova,
2016, 136 , 95–109, https://doi.org/10.4171/RSMUP/136-8.
[2] K. B. Ali, M. Hsini, K. Kefi and N. T. Chung, On a Nonlocal Fractional
p(., .)-Laplacian Problem with Competing Nonlinearities, Complex Analysis
and Operator Theory, 2019, 13(3), 1377–1399, https://doi.org/10.1007/
s11785-018-00885-9.
[3] M. Allaoui and O. Darhouche, Existence results for a class of nonlocal problems
involving the(p1 (x), p2 (x))-Laplace operator, Complex Var. Elliptic Equ., 2018,
63(1), 76–89, https://doi.org/10.1080/17476933.2017.1282950.
[4] S. N. Antontsev and J. F. Rodrigues, On stationary thermorheological viscous
flows, Annali Dell’ Università Di Ferrara, 2006, 52(1), 19–36, https://doi.
org/10.1007/s11565-006-0002-9.
[5] D. Applebaum, Lévy processes-from probability to fnance quantum groups, Not.
Am. Math. Soc., 2004, 51, 1336–1347 http://www.ams.org/notices/200411/
fea-applebaum.pdf.
[6] E. Azroul and A. Boumazourh, On a class of fractional systems with non-
standard growth conditions, J. Pseudo–Differ. Oper. Appl., 2019, https:
//doi.org/10.1007/s11868-019-00310-5.
[7] E. Azroul, A. Benkirane and M. Shimi, Existence and Multiplicity of solutions
for fractional p(x, .)-Kirchhoff type problems in RN , Applicable Analysis, 2019,
https://doi.org/10.1080/00036811.2019.1673373.
[8] E. Azroul, A. Benkirane and M. Shimi, General fractional Sobolev space
with variable exponent and applications to nonlocal problems, Advances
in Operator Theory, 2020, 5(4), 1512–1540, https://doi.org/10.1007/
s43036-020-00062-w.
[9] E. Azroul, A. Benkirane and M. Shimi, Eigenvalue problems involving the
fractional p(x)-Laplacian operator. Adv. Oper. Theory, 2019, 4(2), 539–555,
https://doi.org/10.15352/aot.1809-1420.
[10] E. Azroul, A. Benkirane, M. Shimi and M. Srati, On a class of fractional
p(x)-Kirchhoff type problems, J. Applicable Analysis, 2021, 100(2), 383–402,
https://doi.org/10.1080/00036811.2019.1603372.
[11] E. Azroul, A. Benkirane, A. Boumazourh and M. Shimi, Existence results
for fractional p(x, .)-Laplacian problem via the Nehari manifold approach,
Applied Mathematics and Optimization, 2020, https://doi.org/10.1007/
s00245-020-09686-z.
[12] E. Azroul, A. Benkirane and M. Shimi, On a nonlocal problem involving the
fractional p(x, .)-Laplacian satisfying Cerami condition, J. Discrete and Con-
tinuous Dynamical Systems Series S, 2020, doi: http://www.aimsciences.
org/article/doi/10.3934/dcdss.2020425.
Fractional (p1 (x, .), p2 (x, .))-Kirchhoff type problems 2385

[13] E. Azroul and M. Shimi, Nonlocal eigenvalue problems with variable exponent,
Moroccan J. of Pure and Appl. Anal., 2018, 4(1), 46–61, https://doi.org/
10.1515/mjpaa-2018-0006.
[14] A. Bahrouni and V. Rădulescu, On a new fractional Sobolev space and applica-
tions to nonlocal variational problems with variable exponent, Discrete Contin.
Dyn. Syst., 2018, 11, 379–389, https://doi.org/10.3934/dcdss.2018021.
[15] P. Bartolo, V. Benci and D. Fortunato, Abstract critical point theorems
and applications to some nonlinear problems with ”strong” resonance at in-
finity, Nonlinear Anal., 1983, 7(9), 981–1012, https://doi.org/10.1016/
0362-546X(83)90115-3.
[16] G. M. Bisci, V. Rădulescu and R. Servadi , Variational Methods for Nonlocal
Fractional Problems, Encyclopedia of mathematics and its applications, Cam-
bridge University Press, Cambridge, 2016, http://hdl.handle.net/11576/
2664250.
[17] L. M. M. Bonaldo, E. J. Hurtado and O. H. Miyagaki, A class of elliptic equa-
tions involving nonlocal integrodifferential operators with sign-changing weight
functions, J. Math. Phys., 2020, 61, 051503, https://doi.org/10.1063/1.
514515.
[18] L. Caffarelli, Nonlocal diffusions, drifts and games, Nonlinear Par-
tial Differential Equations, 2012, 7, 37–52, https://doi.org/10.1007/
978-3-642-25361-4_3.
[19] Y. Chen, S. Levine and M. Rao, Variable exponent, linear growth functionals
in image processing, SIAM J. Appl. Math., 2006, 66 , 1383–1406, https://
doi.org/10.1137/050624522.
[20] G. Cerami, An existence criterion for the critical points on unbounded mani-
folds, Istit. Lombardo Accad. Sci. Lett. Rend. A, 1978, 112, 332–336, https:
//ci.nii.ac.jp/naid/10006654437/#cit.
[21] N. T. Chung, Multiple solutions for a class of (p1 (x), p2 (x))-Laplacian problems
with Neumann boundary conditions, Acta Universitatis Apulensis, 2018, 56,
125–141, https://doi.org/10.17114/j.aua.2018.56.11.
[22] N. T. Chung, Multiplicity results for a class of p(x)-Kirchhoff type equa-
tions with combined nonlinearities, Electronic Journal of Qualitative Theory of
Differential Equations, 2012, 42, 1–13 https://doi.org/10.14232/ejqtde.
2012.1.42.
[23] N. T. Chung and H. Q. Toan, On a class of fractional Laplacian problems with
variable exponents and indefinite weights, Collectanea Mathematica, 2019, 1–
15, https://doi.org/10.1007/s13348-019-00254-5.
[24] L. Diening, P. Harjulehto, P. Hästö and M. Råužička, Lebesgue and Sobolev
Spaces with Variable Exponents, Exponents, Lecture notes in Mathematics, 1st
Edition, 2017, https://doi.org/10.1007/978-3-642-18363-8.
[25] X. Fan and D. Zhao, On the Spaces Lp(x) (Ω) and W m,p(x) (Ω), J. Math. Anal.
Appl., 2001 263, 424–446 https://doi.org/10.1006/jmaa.2000.7617.
[26] J. Francåu, Monotone operators. A survey directed to applications to differen-
tial equations, Aplikace matematiky, 1990, 35(4), 257–301, http://dml.cz/
dmlcz/104411.
 2386 E. Azroul, A. Benkirane, N. T. Chung, and M. Shimi

[27] T. C. Halsey, Electrorheological fluids, Science, 1992, 258(5083), 761–766,

https://doi.org/10.1126/science.258.5083.761.
[28] L. Jeanjean, On the existence of bounded Palais-Smale sequences and
application to Landesman-lazer type problem set on RN , Proc. Roy.
Soc. Edinburgh Sect. A., 1999, 129, 787–809, https://doi.org/10.1017/
S0308210500013147.
[29] U. Kaufmann, J. D. Rossi and R. Vidal, Fractional Sobolev spaces with vari-
able exponents and fractional p(x)-Laplacians, Elec. Jour. of Qual. Th, of Diff.
Equa., 2017, 76, 1–10, https://doi.org/10.14232/ejqtde.2017.1.76.
[30] O. Kováčik and J. Rákosník, On Spaces Lp(x) (Ω) and W m,p(x) (Ω), Czechoslovak
Math. Jour., 1991, 41(4), 592–618, https://dml.cz/handle/10338.dmlcz/
102493.
[31] N. Laskin, Fractional quantum mechanics and Lévy path integrals, Phys.
Lett. A., 2000, 268, 298–305, https://doi.org/10.1016/S0375-9601(00)
00201-2.
[32] J. I. Lee, J. M. Kim , Y. H. Kim and J. Lee, Multiplicity of weak solutions
to non-local elliptic equations involving the fractional p(x)-Laplacian, J. Math.
Phys., 2020, 61, 011505, https://doi.org/10.1063/1.5111786.
[33] D. Liu, X. Wang and J. Yao, On (p1 (x), p2 (x))-Laplace equations, 2012, http:
//arxiv.org/pdf/1205.1854.pdf.
[34] O. H. Miyagaki and M. A. S. Souto, Superlinear problems without Ambrosetti
and Rabinowitz growth condition, J. Differential Equations, 2008, 245, 3628–
3638, https://doi.org/10.1016/j.jde.2008.02.035.
[35] E. Di Nezza, G. Palatucci and E. Valdinoci, Hitchhiker’s guide to the fractional
Sobolev spaces, Bull. Sci. Math, 2012 136(5), 521–573, https://doi.org/10.
1016/j.bulsci.2011.12.004.
[36] M. Ruzicka, Electrorheological Fluids: Modeling and Mathematical Teory, Lec-
ture Notes in Mathematics, Springer, Berlin,Germany, 2000, https://doi.
org/10.1007/BFb0104029.
[37] Z. Yucedag, Infinitely many nontrivial solutions for nonlinear problem involv-
ing (p1 (x), p2 (x))-Laplace operator, Acta Univer. Apulensis, 2014, 40, 315–331,
https://doi.org/10.17114/j.aua.2014.40.26.
[38] A. Zang, p(x)-Laplacian equations satisfying Cerami condition, J. Math. Anal.
Appl., 2008, 337, 547–555, https://doi.org/10.1016/j.jmaa.2007.04.007.
[39] J. Zhao, Structure theory of Banach spaces, Wuhan Univ. Press, Wuhan, 1991.
[40] V. V. Zhikov, Averaging of functionals of the calculus of variations and elas-
ticity theory, Math. USSR. Izv., 1987, 9, 33–66, https://doi.org/10.1070/
IM1987v029n01ABEH000958.
[41] W. Zou, Variant Fountain theorems and their applications, Manuscripta Math.,
2001, 104, 343–358, https://doi.org/10.1007/s002290170032.

View publication stats