Beruflich Dokumente
Kultur Dokumente
Abstract. In this paper, we establish the existence of at least three weak solutions for a
parametric problem for doubly eigenvalue elliptic systems involving the (p1 (x), . . . , pn (x))-
Laplacian operator. Our technical approach is based on variational methods and recent
three critical points theorem obtained by Bonanno and Marano.
Key Words and Phrases: weak solutions, variable exponent spaces, (p1 (x), . . . , pn (x))-
Laplacian operator, three critical points theorem.
2010 Mathematics Subject Classifications: 35J60, 47J30, 58E05
1. Introduction
In this work, we deal with the multiplicity of weak solutions for nonlinear
elliptic system:
for 1 ≤ i ≤ n, where ∆pi (x) ui := div (|∇ui |pi (x)−2 ∇ui ) is the pi (x)-Laplacian
operator for all 1 ≤ i ≤ n, pi (x) are continuous real-valued functions such that
1 < p− +
i = inf x∈RN pi (x) ≤ pi (x) ≤ pi = supx∈RN pi (x) < N (N ≥ 2) for all x ∈
RN , λ is a positive parameter, ai ∈ L∞ (RN ) such that ai := ess inf x∈RN ai (x) > 0,
the real function F belongs to C 1 (RN × Rn ), Fui denotes the partial derivative
of F with respect to ui .
The p(x)-Laplacian operator possesses more complicated non-linearities than
p-Laplacian operator, mainly due to the fact that it is not homogeneous. The
study of various mathematical problems involving p(x) growth condition have
seen a strong rise of interest in recent years, we can, for example, refer to
∗
Corresponding author.
http://www.azjm.org 3
c 2010 AZJM All rights reserved.
4 N. Chems Eddine, A.A. Idrissi
[8, 9, 10, 11, 14, 17, 19, 22, 23, 28]. This great interest may be justified by
various physical applications of this operator. In fact, there are applications
concerning nonlinear elasticity theory [30], electro-rheological fluids [26, 27], sta-
tionary thermorheological viscous flows [2] and continuum mechanics [3], etc. It
also has wide applications in different research fields, such as image processing
model [12], and the mathematical description of the filtration process of an ideal
barotropic gas through a porous medium [4].
The goal of this work is to establish the existence of some interval which
includes λ, where the system (1) admits at least three weak solutions, by means
of a very recent abstract critical points result of G. Bonanno and S.A. Marano [7],
which is a more precise version of Theorem 3.2 of [6]. For other basic notations
and definitions we refer to [29].
Lemma 1 (see [7, Theorem 3.6]). Let X be a reflexive real Banach space; Φ :
X → R be a coercive, continuously Gâteaux differentiable and sequentially weakly
lower semicontinuous functional whose Gâteaux derivative admits a continuous
inverse on X ∗ ; Ψ : X → R be a continuously Gâteaux differentiable functional
whose Gâteaux derivative is compact such that
Φ(0) = Ψ(0) = 0.
Assume that there exist r > 0 and x ∈ X, with r < Φ(x), such that
supΦ(x)≤r Ψ(x) Ψ(x)
(a1 ) r < Φ(x) ;
Φ(x) r
(a2 ) for each λ ∈ Λr :=] Ψ(x) , sup Ψ(x) [ the functional Φ − λΨ is coercive.
Φ(x)≤r
Then, for each λ ∈ Λr the functional Φ − λΨ has at least three distinct critical
points in X.
The rest of the paper is organized as follows. Section 2 contains some basic
preliminary knowledge of the variable exponent spaces and some results which
will be needed later. In Section 3, we establish our main results and give an
example of potential function F satisfying the assumptions requested in our main
results.
In this section, we introduce some definitions and results which will be used in
the next section. Firstly, we introduce some theories of Lebesgue-Sobolev spaces
Multiple Solutions to a (p1 (x), . . . , pn (x))-Laplacian-type Systems 5
with variable exponent. The details can be found in [16, 20, 21]. Denote by
S(RN ) the set of all measurable real functions on RN . Set
Z
p(x) N N p(x)
L (R ) = u ∈ S(R ) : |u(x)| dx < ∞ ,
RN
Let a ∈ S(RN ), and a(x) > 0 for a.e x ∈ RN . Define the weighted variable
p(x)
exponent Lebesgue space La (RN ) by
Z
p(x) N N p(x)
La (R ) = u ∈ S(R ) : a(x)|u(x)| dx < ∞ ,
RN
From now on, we suppose that a ∈ L∞ (RN ) with a := ess inf x∈RN a(x) > 0.
p(x)
Then obviously La is a Banach space (see [13] for details).
On the other hand, the variable exponent Sobolev space W 1,p(x) (RN ) is de-
fined by
W 1,p(x) (RN ) = {u ∈ Lp(x) (RN ) : |∇u| ∈ Lp(x) (RN )},
and is endowed with the norm
1,p(x)
Then the norms kuka and kukp(x) are equivalent in Wa (RN ).
1,p(x)
If p− > 1, then the spaces Lp(x) (RN ), W 1,p(x) (RN ) and Wa (RN ) are
separable, reflexive and uniformly convex Banach spaces.
Here we display some facts which will be used later.
0
Proposition 1 (see [16, 20]). The conjugate space of Lp(x) (Ω) is Lp (x) (Ω), where
1 1
+ = 1.
p(x) p0 (x)
0
Moreover, for any (u, v) ∈ Lp(x) (Ω) × Lp (x) (Ω), we have
Z
1 1
| uvdx| ≤ ( −
+ 0 − )|u|p(x) |v|p0 (x) ≤ 2|u|p(x) |v|p0 (x) .
Ω p (p )
Proposition 2 (see [16, 20]). Denote ρ(u) = RN |u|p(x) dx, for all u ∈ Lp(x) (RN ).
R
We have
− + − +
min{|u|pp(x) , |u|pp(x) } ≤ ρ(u) ≤ max{|u|pp(x) , |u|pp(x) },
(i) |u|p(x) < 1 (resp. = 1, > 1) ⇔ ρ(u) < 1 (resp. = 1, > 1),
− +
(ii) |u|p(x) > 1 ⇒ |u|pp(x) ≤ ρ(u) ≤ |u|pp(x) ,
+ −
(iii) |u|p(x) < 1 ⇒ |u|pp(x) ≤ ρ(u) ≤ |u|pp(x) .
Z
− +
kukpa ≤ |∇u(x)|p(x) + a(x) |u(x)|p(x) dx ≤ kukpa if kuka ≥ 1. (2)
RN
Z
+ −
kukpa ≤ |∇u(x)|p(x) + a(x) |u(x)|p(x) dx ≤ kukpa if kuka ≤ 1. (3)
RN
Multiple Solutions to a (p1 (x), . . . , pn (x))-Laplacian-type Systems 7
Proposition 3 (see [18]). Let p(x) and q(x) be measurable functions such that
p ∈ L∞ (RN ) and 1 ≤ p(x), q(x) ≤ ∞ almost everywhere in RN . If u ∈ Lq(x) (RN ),
u 6= 0. Then, we have
− +
|u|p(x)q(x) ≤ 1 ⇒ |u|pp(x)q(x) ≤ |u|p(x) q(x) ≤ |u|pp(x)q(x) ,
+ −
|u|p(x)q(x) ≥ 1 ⇒ |u|pp(x)q(x) ≤ |u|p(x) q(x) ≤ |u|pp(x)q(x) .
Proposition 5 (see [16, 18]). Assume that p ∈ C(RN ) with p(x) > 1 for each
x ∈ RN . If q ∈ C(RN ) and 1 < q(x) < p∗ (x) for each x ∈ Ω, then there exists a
continuous and compact embedding W 1,p(x) (RN ) ,→ Lq(x) (RN ).
1,p (x)
where kukai is the norm of Wai i (RN ). The space X ? denotes the dual space
of X equipped with the usual dual norm.
8 N. Chems Eddine, A.A. Idrissi
n
X
−λ Fui (x, u1 , ...un )vi dx = 0
i=1
where 1 < µij < inf x∈RN pi (x), and pi (x) > N2 , for all x ∈ RN and for all
i ∈ {1, 2, ..., n}. The weight-functions bii (resp bij if i 6= j) belong to the
generalized Lebesgue spaces Lαi (RN ) (resp Lαij (RN )), with
(H3) Assume that there exist r > 0 and w = (w1 , ..., wn ) ∈ X such that the
following conditions are satisfied:
− +
Pn min kwi kpai ,kwi kpai
(c1) i=1 p+
>r
i
R R
RN sup(ξ1 ,...ξn )∈K(sr) F (x, ξ1 , ...ξn )dx N F (x, w1 , ..., wn )dx
(c2) <P R n − +
o
r n
max kw kpa , kw kpa
i=1 i i i i
(p? )− (p? )−
n n o o
where K(t) := (ξ1 , ...ξn ) ∈ RN : ni=1 min |ξi |p? (x) , |ξi |p? (x) ≤ t with
P
t>0 n ?−
(p?i )+
n oo
(pi )
and s = min p+ i min cpi (x) , cpi (x) , such that cpi (x) represent the con-
stants associated with Proposition (4).
We will use the three critical points theorem obtained by Bonanno and Marano
together with the following lemmas to get our main results.
Lemma 2. The functional Φ is continuously Gâteaux differentiable and sequen-
tially weakly lower semicontinuous, coercive whose Gâteaux derivative admits a
continuous inverse on X ? .
Proof. It is well-known that the functional Φ is well defined and is con-
tinuously Gâteaux differentiable functional whose derivative at the point u =
0
(u1 , ..., un ) ∈ X is the functional Φ (u) given by
Z Xn
0
Φ (u)(v) = |∇ui (x)|pi (x)−2 ∇ui (x)∇vi (x) + ai (x)|ui (x)|pi (x) ui (x)vi (x) dx,
Ω i=1
i=1 i
p+
10 N. Chems Eddine, A.A. Idrissi
Upi := x ∈ RN pi (x) ≥ 2 ,
n Z
X
= |∇u1i |pi (x)−2 ∇u1i − |∇u2i |pi (x)−2 ∇u2i (∇u1i − ∇u2i ) dx
i=1 RN
n Z
X
+ ai (x)|u1i |pi (x) u1i − ai (x)|u2i |pi (x) u2i (u1i − u2i ) dx,
i=1 RN
n Z
X 1
|∇(u1i − u2i )|pi (x) + ai (x)|u1i − u2i |pi (x) dx
i=1 Up i 2p(x)
n Z 2−pi (x)
|∇u1i − ∇u2i |
X
+ |∇(u1i − u2i )|pi (x) dx
i=1 Vpi |∇u1i | + |∇u2i |
n Z 2−pi (x)
|u1i − u2i |
X
+ ai (x)|u1i − u2i |pi (x) dx.
i=1 Vpi |u1i | + |u2i |
n
1 X 1 2 pi
−
1 2 pi
+
≥ N min kui − ui kai , kui − ui kai ≥ 0
2
i=1
for all u1 = (u11 , ..., u1n ), u2 = (u21 , ..., u2n ) ∈ X. Hence Φ0 is uniformly monotone
and therefore coercive (see (2)). Since Φ0 is semicontinuous in X, by applying
Minty-Browder theorem (Theorem 26.A of [29]), Φ0 admits a continuous inverse
on X ? . Moreover, the monotony of Φ0 on X ? assures us that Φ is sequentially
lower semicontinuous on X (see [29], Proposition 25. 20). J
Lemma 3. Under the assumptions (H1) and (H2), the functional Ψ is well
defined, and it is of class C 1 on X. Moreover, its derivative is
n Z
0
X ∂F
Ψ (u)h = (x, u1 (x), ..., un (x)hi (x) dx
RN ∂ui
i=1
Proof. For all u = (u1 , ..., un ) ∈ X, under the assumptions (H1) and (H2),
we can write
n Z ui
X ∂F
F (x, u1 , ..., un ) = (x, u1 , ..., s, ..., un ) ds + F (x, 0, ..., 0),
0 ∂s
i=1
Xn Xn
F (x, u1 , ..., un ) ≤ c1 bij (x)|uj (x)|µij −1 |ui (x)| . (5)
i=1 j=1
12 N. Chems Eddine, A.A. Idrissi
Then,
Z Xn Z n
X
F (x, u1 , ..., un ) dx ≤ c2 bij (x)|uj (x)|µij −1 |ui (x)| dx . (6)
RN i=1 RN j=1
If we consider the fact that W 1,p(x) (RN ) ,→ Lµ(x) (RN ), for µ(x) > 1, then there
exists c > 0 such that
and if we apply Propositions 1, 3 and 4 and take bii ∈ Lαi (x) , bij ∈ Lαij (x) if
i 6= j, then we have
Z n X
hX n i
F (x, u1 , ..., un ) ≤ c3 |bij |αij (x) ||uj |µij −1 |p?j (x) |ui |p?i (x) (7)
RN i=1 j=1
hXn Xn i
µ −1
≤ c3 |bij |αij (x) |uj |(µijij −1)p? (x) |ui |p?i (x) (8)
j
i=1 j=1
hXn Xn i
µ −1
≤ c3 |bij |αij (x) kuj kpjij(x) kui kpi (x) < ∞. (9)
i=1 j=1
Hence, Ψ is well defined. Moreover, one can easily see that Ψ0 is also well defined
on X. Indeed, using (F 2) for all h = (h1 , ..., hn ) ∈ X, we have
n Z
0
X ∂F
Ψ (u)h = (x, u1 , ..., un )hi dx
N ∂ui
i=1 R
n
XX n
≤ bij (x)|uj (x)|µij −1 |hi (x)|dx.
i=1 j=1
Now let us show that Ψ is differentiable in the sense of Frechet, that is, for
fixed u = (u1 , ..., un ) ∈ X and given ε > 0, there must be a δ = δε,u1 ,...,un > 0
such that
n
0
X
|Ψ(u1 +h1 , ..., un +hn )−Ψ(u1 , ..., un )−Ψ (u1 , ..., un )(h1 , ..., hn )| ≤ ε (khi kpi (x) )
i=1
Pn
for all h = (h1 , ..., hn ) ∈ X with i=1 (khi kpi (x) ) ≤ δ.
Let BR be the ball of radius R which is centered at the origin of RN and denote
0 1,p (x)
BR = RN − BR . Moreover, let us define the functional ΨR on ni=1 Wai i (BR )
Q
as follows: Z
ΨR (u) = F (x, u1 (x), ..., un (x)) dx.
BR
1,p (x)
If we consider (H1) and (H2), it is easy to see that ΨR ∈ C 1 ( ni=1 Wai i (BR )),
Q
1,p (x)
and in addition for all h = (h1 , ..., hn ) ∈ ni=1 Wai i (BR ), we have
Q
n Z
0
X ∂F
ΨR (u)h = (x, u1 (x), ..., un (x))hi (x) dx.
BR ∂ui
i=1
0
Also as we know, the operator ΨR : X → X ? is compact [20]. Then, for all
u = (u1 , ..., un ), h = (h1 , ..., hn ) ∈ X, we can write
0
|Ψ(u1 + h1 , ..., un + hn ) − Ψ(u1 , ..., un ) − Ψ (u1 , ..., un )(h1 , ..., hn )|
0
≤ |ΨR (u1 + h1 , ..., un + hn ) − ΨR (u1 , ..., un ) − ΨR (u1 , ..., un )(h1 , ..., hn )|
Z
+ 0 F (x, u1 + h1 , ..., un + hn ) − F (x, u1 , ..., un )−
B
R
n Z
X ∂F
− (x, u1 , ..., un )hi dx .
BR ∂ui
i=1
According to a classical theorem, there exist ξ1 , ..., ξn ∈]0, 1[ such that
Z
F (x, u + h , ..., u + h ) − F (x, u , ..., u n −
)
0 1 1 n n 1
B
R
n Z
X ∂F
− (x, u1 , ..., un )hi dx
BR ∂ui
i=1
Z n n
!
X ∂F X ∂F
= 0 (x, u1 , ..., ui + ξi hi , ..., un )hi − (x, u1 , ..., un ) dx .
B ∂ui ∂ui
R i=1 i=1
14 N. Chems Eddine, A.A. Idrissi
Xn X
n Z
≤ bij (x)(|uj + ξj hj |µij −1 − |uj |µij −1 )hi dx .
0
i=1 j=1 BR
|bij (x)|Lαij (B 0 ) −→ 0
R
n
n
X ∂F X
− (x, u1 , ..., un )hi dx ≤ ε (khi kpi (x) ).
∂ui
i=1 i=1
0
It remains only to show that Ψ is continuous on X. Let um = (um m
1 , ..., un ) be
such that um → u as m → ∞. Then, for h = (h1 , ..., hn ) ∈ X, we have
0 0 0 0
|Ψ (um )h − Ψ (u)h| ≤ |ΨR (um )h − ΨR (u)h|
n Z ∂F
X
m m ∂F
+ ( (x, u , ..., u )h − (x, u , ..., u )h ) dx .
1 n i 1 n i
BR ∂ui ∂ui
0
i=1
Multiple Solutions to a (p1 (x), . . . , pn (x))-Laplacian-type Systems 15
Qn 1,pi (x)
Since Ψ0R is continuous on i=1 Wai (BR )(see [20]), we have
0 0
|ΨR (um )h − ΨR (u)h| −→ 0,
as m → ∞. Now, using (H2) once again and taking into account that the other
terms on the right-hand side of the above inequality tend to zero, we conclude
that Ψ0 is continuous on X. J
0
Lemma 4. Under the assumptions (H1) and (H2), Ψ is compact from X to
X ?.
1,p (x)
Since the restriction operator is continuous, we have um * u in ni=1 Wai i (BR ).
Q
0
Because of the compactness of Ψ , the first expression on the right-hand side of
the inequality tends to 0, as m −→ ∞, and, as above, for sufficiently large R we
obtain
n Z ∂F
X
m m ∂F
(x, u , ..., u )h − (x, u , ..., u )h dx −→ 0.
1 n i 1 n i
∂ui ∂ui
0
i=1 BR
0
This implies Ψ is compact from X to X ? . J
Then,
Z
sup Ψ(u) = sup F (x, u1 , ..., un )dx,
u∈Φ−1 (]−∞,r[) u∈Φ−1 (]−∞,r[) RN
Z
≤ sup F (x, ξ1 , ..., ξn )dx.
RN (ξ1 ,...ξn )∈K(sr)
Ψ(w)
≤r .
Φ(w)
from which (a1 ) of Lemma 1 follows.
To show that the functional Φ − λΨ is coercive, we use the inequallity (9).
Then for all u ∈ X, we have by virtue of (H1) and (H2)
n Z
X 1
Φ(u) − λΨ(u) = |∇ui (x)|pi (x) dx + ai (x)|ui (x)|pi (x) dx
RN pi (x)
i=1
Multiple Solutions to a (p1 (x), . . . , pn (x))-Laplacian-type Systems 17
Z
−λ F (x, u1 (x), ..., un (x))dx,
RN
n Z
X 1 pi (x) pi (x)
≥ + |∇u i (x)| dx + a i (x)|ui (x)| dx
i=1
p i RN
hX n X n i
µ −1
− c3 |bij |αij (x) kuj kpjij(x) kui kpi (x) .
i=1 i=j
This shows that Φ−λΨ → +∞ as kukX → +∞, since 1 < µij < inf x∈RN pi (x);
that is, Φ − λΨ is coercive on X for every parameter λ, in particular, for every
Φ(w) r
λ ∈ Λr :=] Ψ(w) , sup [. Then, condition (a2 ) also holds. Now all the
Φ(w)≤r Ψ((w)
hypotheses of Lemma 1 are satisfied. Also note that the solutions of the equation
0 0
Φ (u) − λΨ (u) = 0 are exactly the weak solutions of(1). Thus for each
p−
Pn p+
i=1 max kwi kai , kwi kai
i i
r
λ∈ N F (x, w1 (x), ..., wn (x))dx , N sup
R R
(ξ1 ,...ξn )∈K(sr) F (x, ξ1 , ..., ξn )dx
R R
Example
Let
F (x, u, v) = a(x)|u|β(x) |v|γ(x) ,
β(x) γ(x)
where p(x) + q(x) < 1 and a is a positive function in Ls(x) (RN ) such that
p?(x) q ?(x)
s(x) = for each x ∈ RN .
p?(x) q ?(x) − β(x)q ?(x) − γ(x)p?(x)
18 N. Chems Eddine, A.A. Idrissi
We can easily verify that F (x, u, v) satisfies the conditions (H1 ) and (H3 ).
Moreover, by using Young inequality we easily check that the condition (H2 ) is
also satisfied, and then the conclusion of Theorem 1 holds true.
Acknowledgments
References
[4] S.N. Antontsev, S.I. Shmarev, A model porous medium equation with variable
exponent of nonlinearity: existence, uniqueness and localization properties of
solutions, Nonlinear Anal., 60, 2005, 515-545.
[7] G. Bonanno, S.A. Marano, On the structure of the critical set of non-
differentiable functions with a weak compactness condition, Appl. Anal., 89,
2010, 1-10.
[9] G. Bonanno, A. Chinni, Multiple solutions for elliptic problems involving the
p(x)-Laplacian, Matematiche (Catania), 66, 2011, 105-113.
Multiple Solutions to a (p1 (x), . . . , pn (x))-Laplacian-type Systems 19
[14] G. Dai, Three solutions for a nonlocal Dirichlet boundary value problem in-
volving the p(x)-Laplacien, Appl. Anal., 92, 2013, 191-210.
[18] D.E. Edmunds, J. Rakosnik, Sobolev embeddings with variable exponent, Stu-
dia Math., 143, 2000, 267-293.
[19] X.L Fan, Q.H Zangh, Existence of solutions for p(x)-Laplacien Dirichlet
problem, Nonlinear Anal, 52, 2003, 1843-1852.
[20] X. Fan, D. Zhao, On the spaces Lp (x)(Ω) and W m,p (Ω) , J. Math. Anal.
Appl., 263, 2001, 424-446.
[22] J. Liu, X. Shi, Existence of three solutions for a class of quasilinear elliptic
systems involving the (p(x), q(x))-Laplacian, Nonlinear Anal., 71, 2009, 550-
557.
[23] R.A. Mashiyev, Three solutions to a Neumann problem for elliptic equations
with variable exponent, Arab. J. Sci. Eng., 36, 2011, 1559-1567.
20 N. Chems Eddine, A.A. Idrissi
[24] F.I. Mamedov, F.M. Mammadova, A necessary and sufficient condition for
Hardy’s operator in Lp(.) (0, 1), Mathematische Nachrichten, 287(5-6), 2013,
666-676.
[28] X. Shi, X. Ding, Existence and multiplicity of solutions for a general p(x)-
Laplacian Neumann problem, Nonlinear Anal., 70, 2009, 3715-3720.
[29] E. Zeidler, Nonlinear Functional Analysis and its Applications, vol. II/B,
Berlin-Heidelberg-New York, 1990.