Applied Mathematics E-Notes, 16(2016), 11-20 c ISSN 1607-2510

Available free at mirror sites of http://www.math.nthu.edu.tw/ amen/

On The Spectrum Of (p(x); q(x))-Laplacian In RN

Abdelrachid El Amroussy, Anass Ourraouiz, Mostafa Allaouix

Received 27 October 2014

Abstract
In this work, we study the existence of a family of eigenvalues for a nonho-
mogeneous problem involving variable exponents in RN by using the variational
approach.

1 Introduction
Large number of papers was devoted to study elliptic equations and variational prob-
lems with variable exponent. They were of considerable importance in the theory of
partial di¤erential equations. Some of these problems come from di¤erent areas of
applied mathematics and physics such as Micro Electro-Mechanical systems, surface
di¤usion on solids or image processing and restoration. For more inquiries on modeling
physical phenomena involving p(x)-growth condition, we refer to [1–10].
Motivated by the works [11, 12], we consider the following problem
r(x) 2
p(x) u = q(x) u + m(x) juj u in RN ; (1)
where
p(x) u = div(jujp(x) 2
u)
N
is the p(x)-Laplacian operator and p; q : R ! R are Lipschitz continuous functions
with
1 < p := inf p(x) p(x) sup p(x) := p+ < N; N 3;
RN RN
N
where m is a positive weight for a.e x 2 R such that

m 2 L1 (RN ) \ L (x)
(RN )

with
q (x)
(x) = :
q (x) r(x)
Suppose that
1 < q(x) < r = inf r(x) r+ = sup r(x) < p(x) (2)
RN RN

Mathematics Subject Classi…cations: 35J60, 35D05, 35J20, 35J40.

y University Mohamed I, Faculty of sciences, Department of Mathematics, Oujda, Morocco
z National School of Applied Sciences of Al-Hoceima, Morocco
x University Mohamed I, Faculty of Sciences, Department of Mathematics, Oujda, Morocco

11
12 The Spectrum of Laplacian with Variable Expoment

and
N q(x)
r+ < q (x) with q (x) = : (3)
N q(x)
De…ne
R h i
1 p(x) 1 q(x)
RN p(x)jruj + q(x) jruj dx
1 = inf R m(x) r(x)
;
u2W 1;p(x) (RN )\W 1;q(x) (RN );u6=0 juj dx
RN r(x)

and R h i
p(x) q(x)
RN
jruj + jruj dx
= inf R r(x)
:
u2W 1;p(x) (RN )\W 1;q(x) (RN );u6=0 N m(x) juj dx
R

We state our main result.

THEOREM 1. Assume that conditions (2) and (3) hold. Then we have:

(i) 1 > 0 and 2[ 1 ; +1[ is an eigenvalue of the problem (1).

(ii) There exists 2]0; 1] such that 2]0; [ is not an eigenvalue of problem (1).

DEFINITION 1. We say that 2 R is an eigenvalue of the problem (1) if there

exists u 2 W 1;p(x) (RN ) \ W 1;q(x) (RN ) n f0g such that
Z Z Z
p(x) 2 q(x) 2 r(x) 2
jruj rurvdx + jruj rurvdx = m(x) juj uvdx;
RN RN RN

for all v 2 W 1;p(x) (RN ) \ W 1;q(x) (RN ) n f0g:

When p(x) = q(x) = 2 and m(x) = 1, the problem (1) is a normal Schrodinger
equation (see [13, 14]). The case m = 1 and p(x) = q(x) in a bounded domain has been
studied by Fan, Zhang and Zhao in [15, 16]. The case of an inde…nite weight m 6= 1 in
RN with p(x) = q(x) was considered by [11].
This article is organized as follows. In section 2, we give the necessary notations
and preliminaries. We include some useful results involving the variable exponents
Lebesgue and Sobolev spaces in order to facilitate the reading of the paper. Finally, in
section 3, we prove the main result.

2 Preliminary Notes
In order to deal with the problem (1), we need some theory of variable exponent Sobolev
Space. For convenience, we only recall some basic facts which will be used later.
De…ne the variable exponent Lebesgue space Lp(x) (RN ) by
Z
p(x)
Lp(x) (RN ) = u : RN ! R measurable : juj dx < 1 :
RN
El Amrouss et al. 13

Then Lp(x) ( ) endowed with the norm

Z p(x)
u
jujp(x) = inf >0: dx 1
RN

becomes a separable and re‡exive Banach space.

De…ne the variable exponent Sobolev space W 1;p(x) (RN ) by

W 1;p(x) (RN ) = fu 2 Lp(x) (RN ) : ru 2 Lp(x) (RN )g

equipped with the norm

( Z )
p(x)
ru
kukp(x) = inf >0: dx 1 ;
RN

which is also a separable and re‡exive Banach space.

R p(x)
PROPOSITION 1 (cf. [17]). Let (u) = RN
jruj dx and u 2 W 1;p(x) (RN ):
Then the following statements hold.
+
(1) If kukp(x) 1, then kukp(x) p (u) kukp(x) p :
+
(2) If kukp(x) 1, then kukp(x) p (u) kukp(x) p :

(3) limn!1 kun kp(x) = 0 (resp +1) if, and only if, limn!1 (un ) = 0 (resp +1).

REMARK 1. We have similar results (1) and (2) of Propositiion 1 for 1 (u) =
R p(x)
RN
juj dx.
0
PROPOSITION 2 (cf. [17]). For any u 2 Lp(x) (RN ) and v 2 Lp (x) (RN ); we have
Z
uvdx 2 jujp(x) jvjp0 (x)
RN

with
1 1
+ = 1:
p(x) p0 (x)

PROPOSITION 3 (cf. [17]). Suppose that p is Lipschitz continuous, q : RN ! R is

a measurable function and p(x) q(x) p (x); 8x 2 RN : Then there is a continuous
embedding
W 1;p(x) (RN ) ,! Lq(x) (RN ):
If is a bounded open subset of RN with cone property, then the embedding

W 1;p(x) ( ) ,!,! Lq(x) ( )

14 The Spectrum of Laplacian with Variable Expoment

is compact.

In what follows, for simplicity let

W = W 1;p(x) (RN ) \ W 1;q(x) (RN );

which will be endowed with the following norm

kuk = kukp(x) + kvkq(x) ;

R R
which makes W a Banach space separable and re‡exive and denote by = RN
: The
assertions of Proposition 1 remain valid with the norm k k :

PROPOSITION 4 (cf. [5]). Suppose that p; q are are Lipschitz continuous functions
and r is a measurable function. If q(x) r(x) p(x); then Lp(x) (RN ) \ Lq(x) (RN ) ,!
Lr(x) (RN ) with a continuous embedding.

Analogously, we have the following interesting result (see [5]).

PROPOSITION 5. Under the assumptions of Proposition 4, we have a continuous

embedding from W into Lr(x) (RN ):

3 Proof of the Main Result

First, by the assumption (2) we can see that
p(x) q(x) r+ p(x) q(x) r
jruj + jruj jruj , jruj + jruj jruj

and
r+ r r(x)
juj + juj juj :
i i
The continuous embedding from W 1;r (RN ) into Lr (RN ); i = ; implies that there
exist two positive constants C1 ; C2 such that
Z Z Z Z
r+ r+ r r
jruj dx C1 juj dx and jruj dx C1 juj dx:

Then there exists C > 0 such that

Z h i Z Z
p(x) q(x) r+ r C r(x)
2 jruj + jruj dx C juj + juj dx m(x) juj dx
jmj1

and thus
Z Z
1 p(x) 1 q(x) Cr m(x) r(x)
p+ jruj + jruj dx juj dx:
p(x) q(x) 2jmj1 r(x)

So 1 > 0: Similarly, we get > 0:

El Amrouss et al. 15

Before embarking in the proof of Theorem 1, we need some auxiliar lemmas. Let
Z Z
1 p(x) 1 q(x)
I(u) = jruj dx + jruj dx
p(x) q(x)

and Z
1 r(x)
J(u) = m(x) juj dx:
r(x)
The following lemma plays a crucial role in our di¤erent lines.

LEMMA 1.

(i) The functional I is weakly lower semi-continuous, that is, un * u implies that
I(u) lim inf n I(un ):

(ii) The functional J is weakly-strongly continuous, that is, un * u implies that

J(un ) ! J(u):

PROOF. (i) Using the convexity of the functional I, we see that the assertion is
immediate. (ii) Assume that un * u in W; which is re‡exive. Then fun g is a bounded
sequence. By Proposition 3, we have

W ,! Lq (x) (RN ):
n o
So we obtain a boundedness of jun jq (x) : So, there is a positive constant M > 0
such that
max jun jr(x) q (x)
; jujr(x) q (x)
M:
n
r(x) r(x)

Taking k = fx 2 RN : jxj < kg: Further, m 2 L (x)

(RN ) implies that

jmjL (x) (RN n

k)
! 0 as k ! +1:

Giving " > 0; we may …nd k1 > 0 large enough such that
"
jmjL (x) (RN n
k1 )
< :
8M

It follows from the compact embedding W 1;r(x) ( k1 ) ,!,! Lr(x) ( k1 ); that

Z Z
r(x)
m(x)jun j )dx ! m(x)jujr(x) dx;
k1 k1

because m 2 L1 (RN ): Hence, there exists n1 > 0 such that for n n1 ;

Z Z
"
m(x)jun jr(x) dx m(x)jujr(x) dx < :
k1 k1
2
16 The Spectrum of Laplacian with Variable Expoment

In view of Proposition 2, we get

Z h i
jJ(un ) J(u)j m(x)jun jr(x) m(x)jujr(x) dx
k1

Z
+ [m(x)jun jr(x) m(x)jujr(x) ]dx
RN n k1
Z h i
" r(x) r(x)
+ m(x) jun j + juj dx
2 RN n k1

"
+ 2jmjL (x) (RN n k1 )
jun jr(x) + jujr(x)
2 q (x)
r(x)
q (x)
r(x)
" "
+ = ":
2 2
Consequently, J(un ) ! J(u):

LEMMA 2. We have the following limits,

I(u)
lim = +1
kuk!+1 J(u)

and

I(u)
lim = +1: (4)
kuk!0 J(u)

PROOF. When kuk ! 0; we have kukq(x) ! 0: The preceding Proposition 1 to-

gether with Proposition 4 and 5 imply
R 1 p(x) R 1 q(x) 1 q+
I(u) p(x) jruj dx + q(x) jruj dx q+ kukq(x)
= R r(x) r
;
J(u) 1
m(x) r(x) juj dx c jmj1 kukq(x)

where c > 0: Since r > q + ; then the …rst relation hold. For the seconde assertion, let
Z
1 p(x)
K(u) = jruj dx:
p(x)

If kukp(x) 1; then we have

1 p 1 p+
kukp(x) K(u) kukp(x) :
p+ p

If kukp(x) 1; then we have

1 p+ 1 p
+
kukp(x) K(u) kukp(x) :
p p
El Amrouss et al. 17

In this case, we set

1 p 1 p+
C0 kukp(x) kukp(x) 0:
p+ p+
Then
1 p
K(u) kukp(x) C0 :
p+
We entail that
1 p
K(u) kuk C0 ; 8u 2 W 1;p(x) (RN ):
p+
Thereby, there exists C1 > 0 such that
Z Z
1 p(x) 1 q(x)
I(u) = jruj dx + jruj dx
p(x) q(x)
1 p 1 q
kukp(x) + + kukq(x) C1 ;
p+ q
whenever u 2 W:
On the other hand, according to Proposition 4 there exists C3 > 0 such that
Z
1 r(x) r+ r
J(u) = m(x) juj dx C3 (kuk + kuk ):
r(x)
Afterwards, we have
R 1 p(x) R 1 q(x)
I(u) p(x) jruj dx + q(x) jruj dx
= R r(x)
J(u) 1
r(x) m(x) juj dx
1 p 1 q
p+ kukp(x) + q+ kukq(x) C1
r+ r
;
C3 (kuk + kuk )
I(u)
since r+ < p ; we infer that J(u) ! +1 when kuk ! +1:

Proof of Theorem 1. We show that 1 is an eigenvalue of (1). By the de…nition

of 1 ; there exists fun g W n f0g such that

I(un )
1 = lim :
n J(un )

I(un )
Here fun g is bounded in W; because the coercivity of , and thus there exists
J(un )
u 2 W satisfying un * u; so

I(u) lim inf I(un ) and J(un ) ! J(u):

n

I(u)
Whether u 6= 0; then 1 = J(u) and it is done. Otherwise, assume that u = 0;
I(un )
so I(un ) ! 0 as n ! +1: We have I(un ) = J(un ) J(un ); passing to limit we get
18 The Spectrum of Laplacian with Variable Expoment

I(un )
I(un ) ! 0 as n ! +1; along with kun k ! 0; hence J(un ) ! +1 which is paradoxical.
We conclude that for any v 2 W we have

@ I(u + "v)
= 0:
@" J(u + "v) "=0

By a straightforward computation we have

Z Z
jrujp(x) 2
+ jrujq(x) 2
rurvdx J(u) = m(x)jujr(x) uvdx I(u); 8v 2 W:

Consequently, 1 is an eigenvalue of problem (1).

Now, we are to prove that any > 1 is an eigenvalue of (1). Set

L(u) = I(u) J(u):

From Lemma 1, we know that I is weakly lower semi-continuous and J is weakly-strongly

continuous, which yields L is weakly lower semicontinuous.
If we return to relation (4), it is clear that L is of class C 1 and coercive, therefore,
L admits a global minimum u in W; which is a critical point for L. We claim that u
is nontrivial. Indeed, by the characterization of 1 ; there exists v 2 W n f0g such that
I(v)
1 = J(v) ; since > 1 ; accordingly L(v) < 0, thence u 6= 0:
Hereinafter, we check that any 2]0; [ is not an eigenvalue of problem (1). First,
r+ +
we may observe that 1 ; because p 1 and r < p :
Next, let 2]0; [, by contradiction we assume that is an eigenvalue, then there
exists v 2 W n f0g which satis…es
Z Z
p(x) q(x) r(x)
jrvj + jrvj dx = m(x) juj dx:

On the other hand, according to de…nition of we obtain

R p(x) q(x)
jrvj + jrvj dx
R r(x)
= ;
m(x) jvj dx

which is contradictory and then the proof is achieved.

REMARK 2. When 0; for all u 2 W n f0g we have

Z Z Z
p(x) q(x) r(x)
I 0 (u)u = jruj dx + jruj dx m(x) juj dx > 0;

and then the problem (1) has no solution.

Whether m < 0; we may consider that the eigenvalue < 0; and the similar
argument works.
El Amrouss et al. 19

References
[1] E. Acerbi and G. Mingione, Regularity results for stationary electrorheological
‡uids, Arch. Ration. Mech. Anal., 164(2002), 213–259.

[2] M. Allaoui, A. El Amrouss and A. Ourraoui, Three Solutions For A Quasi-Linear

Elliptic Problem, Applied Mathematics E-Notes, 13(2013), 51–59

[3] Y. Chen, S. Levine and M. Rao, Variable exponent linear growth functionals in
image processing, SIAM J. Appl. Math., 66(2006), 1383–1406.

[4] F. Colasuonno and P. Pucci, Multiplicity of solutions for p(x)-polyharmonic elliptic

Kirchho¤ equations, Nonlinear Anal., 74(2011), no. 17, 5962–5974.

[5] L. Diening, P. Harjulehto, P. Hasto and M. Ruzicka, Lebesgue and Sobolev Spaces
with Variable Exponents.Springer Heidelberg Dordrecht London New York 2011.
Germany, 2002.

[6] S. G. Deng, A local mountain pass theorem and applications to a double perturbed
p(x)-Laplacian equations, Appl. Math. Comput, 211(2009), 234–241.

[7] A. El Amrouss and A. Ourraoui, Existence of solutions for a boundary problem

involving p(x)-biharmonic operator, Bol. Soc. Parana. Mat., 31(2013), 179–192.

[8] S. Esedoglu and S. Osher, Decomposition of images by the anisotropic Rudin-

Osher-Fatemi model, Comm. Pure Appl. Math., 57(2004), 1609–1626.

[9] A. Ourraoui, Multiplicity results for Steklov problem with variable exponent, Appl.
Math. Comput., 277(2016), 34–43.

[10] S. Samko, On a progress in the theory of Lebesgue spaces with variable exponent:
Maximal and singular operators, Integral Transforms Spec. Funct., 16(2005), 461–
482.

[11] N. Benouhiba, On the eigenvalues of weighted p(x)-Laplacian on RN , Nonlinear

Anal., 74(2011), 235–243 .

[12] B. Ge and Q. Zhou, Continuous spectrum of a fourth order nonhomogeneous

di¤erential operators with variable exponent, Electron. J. Qual. Theory Di¤er.
Equ., 2013, 18, 1–11.

[13] T. Bartsch, A. Pankov and Z. Q. Wang, Nonlinear Schrodinger equations with

steep potential well, Commun. Contemp. Math., 3(2001), 549–569.

[14] J. Byeon and Z. Q. Wang, Standing waves with a critical frequency for nonlinear
Schrodinger equation II, Calc. Var. Partial Di¤erential Equations, 18(2003), 207–
219.

[15] X. L. Fan, Eigenvalues of the p(x)-Laplacian Neumann problems, Nonlinear Anal.,

67(2007), 2982–2992 .
20 The Spectrum of Laplacian with Variable Expoment

[16] X. L. Fan, Q.H. Zhang and D. Zhao, Eigenvalues of p(x)-Laplacian Dirichlet prob-
lem, J. Math. Anal. Appl., 302(2005), 306–317 .
[17] X. L. Fan, J.S. Shen and D. Zhao, Sobolev, imbedding theorems for spaces
W k;p(x) ( ); J. Math. Anal. Appl., 262(2001), 749–760.