ISSN 0081-5438, Proceedings of the Steklov Institute of Mathematics, 2008, Vol. 261, pp. 287294. c Pleiades Publishing, Ltd.

s Publishing, Ltd., 2008.

Some Issues on the p-Laplace Equation
in Cylindrical Domains
M. Chipot
a
and Y. Xie
b
Received January 2007
AbstractWe investigate the asymptotic behavior of the solution to equations of the p-Lapla-
cian type in cylindrical domains becoming unbounded and address some issues regarding the
solution in unbounded domains.
DOI: 10.1134/S0081543808020235
1. INTRODUCTION
Many problems of mathematical physics are set in cylindrical domains. For instance, these are
porous media ows in channels, plate theory, elasticity theory, etc. In this note we will address
the problem of the p-Laplace equation. To be more precise, suppose that

is a two-dimensional
domain (for simplicity) pictured in the gure below, and let u

be the unique solution to the

nonlinear problem
_

p
u

= f(x
2
) in

,
u

= 0 on

,
(1.1)
where
p
is the usual p-Laplace operator dened as
p
u := div(|u|
p2
u), p > 1, and

denotes the boundary of

. Note that the p-Laplace operator reduces to the usual Laplacian when
p = 2. In this note, we mainly consider the case p = 2. One refers to [27] for results in the
linear case.
We notice that the data f of (1.1) depends only on the x
2
-variable. Of course, due to the
boundary conditions at the ends of the cylinder, u

is not a function independent of x

1
. However,
one expects that when +, u

is close to a function depending on x

2
only. To be more precise,
let u

be the solution to
_

p
u

= f(x
2
) in (1, 1),
u

(1) = u

(1) = 0.
(1.2)
Then a natural question is whether
u

as + in any bounded subdomain of R (1, 1).

We will investigate such convergence in Section 2. In the following Section 3, some property of
the solution to the p-Laplace equation in an unbounded domain will be discussed.
Before we go into details, let us quote some useful inequalities:
a
Institut f ur Mathematik, Abteilung f ur Angewandte Mathematik, Universitat Z urich, Winterthurerstrasse 190,
CH-8057 Z urich, Switzerland.
b
Department of Mathematics, Ecole Polytechnique Federale de Lausanne, 1015 Lausanne, Switzerland.
E-mail addresses: m.m.chipot@math.unizh.ch (M. Chipot), yitian.xie@ep.ch (Y. Xie).
287
288 M. CHIPOT, Y. XIE
0
1
1

x
1
x
2
Figure.
Lemma 1.1 (see [1, 8] for a proof). For all p > 1, 0, and , R
n
, it holds that for some
constant c depending on p

||
p2
||
p2

c| |
1
(|| +||)
p2+
, (1.3)
_
||
p2
||
p2
,
_
c| |
2+
(|| +||)
p2
. (1.4)
2. A CONVERGENCE RESULT
Let us consider the problem mentioned above in a more general setting. Denote by X =
(X
1
, X
2
) = (x
1
, . . . , x
q
, x
q+1
, . . . , x
n
) any point in R
n
, and let

= (, )
q
, where is an open
bounded domain in R
nq
.
Assume that u

and u

are solutions to
_

p
u

= f(X
2
) in

,
u

= 0 on

(2.1)
and
_

p
u

= f(X
2
) in ,
u

= 0 on ,
(2.2)
respectively. We consider here the weak solutions to (2.1) and (2.2) and assume u

W
1,p
0
(

)
and u

W
1,p
0
(), where W
1,p
0
(

) and W
1,p
0
() stand for the usual Sobolev spaces. We refer, for
instance, to [2] for more information regarding these spaces.
Theorem 2.1. Suppose that f(X
2
) L
p

() is nonnegative (or nonpositive for an analo-

gous statement ) and is any bounded subset in R
q
. Then u

is a nondecreasing sequence of
nonnegative functions bounded above by u

, and it holds that

u

in W
1,p
().
Proof. We notice that u

, the solution to (2.1), satises

_

|u

|
p2
u

v dx =
_

fv dx v W
1,p
0
(

). (2.3)
Take v = u

W
1,p
0
(

), the negative part of u

, in (2.3). We obtain
_

|u

|
p2
u

dx =
_

fu

dx 0,
PROCEEDINGS OF THE STEKLOV INSTITUTE OF MATHEMATICS Vol. 261 2008
SOME ISSUES ON THE p-LAPLACE EQUATION 289
i.e.,
_

|u

|
p
dx 0.
Hence we derive that u

is nonnegative. Following the same arguments, one can show that u

, the
solution to (2.2), is nonnegative.
Letting <

, one has
_

|u

|
p2
u

v dx =
_

fv dx v W
1,p
0
(

), (2.4)
_

|u

|
p2
u

v dx =
_

fv dx v W
1,p
0
(

). (2.5)
We remark that u

is nonnegative in

. So
(u

)
+
W
1,p
0
(

) ((u

)
+
is the positive part of u

).
Therefore, one derives from (2.4) and (2.5) that
_

|u

|
p2
u

(u

)
+
dx =
_

|u

|
p2
u

(u

)
+
dx.
The above equation (see (1.4)) leads to
(u

)
+
= 0,
which shows that {u

} is a nondecreasing sequence in . On the other hand, we have

p
u

+
p
u

= 0 in

,
u

=
_
u

0 on (, )
q
,
0 on (, )
q
.
(2.6)
By the weak maximum principle, this implies
0 u

in

. (2.7)
Consider now a smooth nonnegative function (x) such that
0 1, 1 in

0
, 0 outside

0
+1
, || is bounded. (2.8)
Taking u

p
in (2.1), we obtain
_

|u

|
p2
u

{u

p
} dx =
_

fu

p
dx.
Therefore, for some constant c we have
_

0
+1
|u

|
p

p
dx = p
_

0
+1
|u

|
p2
u

p1
dx +
_

0
+1
fu

p
dx
c
_

0
+1
|u

|
p1
u

p1
dx +
_

0
+1
fu

p
dx.
PROCEEDINGS OF THE STEKLOV INSTITUTE OF MATHEMATICS Vol. 261 2008
290 M. CHIPOT, Y. XIE
Applying the Young inequality on the right-hand side, we come to
_

0
+1
|u

|
p

p
dx
1
2
_

0
+1
|u

|
p

p
dx + c
_

0
+1
u
p

dx +
_

0
+1
fu

p
dx (2.9)
for some constant c. Since (2.7) holds, we derive from (2.9) that
_

0
|u

|
p
dx 2c
_

0
+1
u
p

dx + 2
_

0
+1
fu

dx 2c
_

0
+1
u
p

dx + 2
_

0
+1
fu

dx C;
i.e., {u

} is uniformly bounded in

0
. Up to a subsequence of u

, labeled still by , there exists a

u
0
W
1,p
(

0
) vanishing on (
0
,
0
) such that
u

u
0
in L
p
(

0
), u

u
0
in W
1,p
(

0
).
We remark here that due to the monotonicity of {u

}, the whole sequence converges to u

0
in L
p
.
Fixing a positive constant h, for any integer 1 i q we consider u

(X
1
+ he
i
, X
2
) and
u
+h
(X
1
, X
2
) in the domain (, )
i1
( h, h) (, )
qi
. We have

p
u

(X
1
+ he
i
, X
2
) +
p
u
+h
(X
1
, X
2
) = f(X
2
) f(X
2
) = 0
in (, )
i1
( h, h) (, )
qi
,
together with the boundary condition
u

(X
1
+ he
i
, X
2
) u
+h
(X
1
, X
2
)
=

0 on (, )
i1
( h, h) (, )
qi
,
0 on (, )
i1
{ h} (, )
qi
,
u
+h
(X
1
, X
2
) 0 on (, )
i1
{ h} (, )
qi
.
By the maximum principle we have
u

(X
1
+ he
i
, X
2
) u
+h
(X
1
, X
2
) in (, )
i1
( h, h) (, )
qi
.
Passing to the limit in

0
leads to
u
0
(X
1
+ he
i
, X
2
) u
0
(X
1
, X
2
).
By changing h to h and noticing that this holds for all i, one has
u
0
(X
1
, X
2
) = u
0
(X
2
).
Now we denote by a smooth nonnegative function such that (this is slightly dierent from (2.8))
0 1, 1 in

0
1
, 0 outside

0
, || is bounded.
Taking v = (u

u
0
) W
1,p
0
(

0
) as a test function in (2.1), we obtain
_

0
|u

|
p2
u

{(u

u
0
)} dx =
_

0
f(u

u
0
) dx,
PROCEEDINGS OF THE STEKLOV INSTITUTE OF MATHEMATICS Vol. 261 2008
SOME ISSUES ON THE p-LAPLACE EQUATION 291
i.e.,
_

0
|u

|
p2
u

(u

u
0
) dx =
_

0
|u

|
p2
u

(u

u
0
) dx +
_

0
f(u

u
0
) dx. (2.10)
Since u

u
0
in L
p
(

0
), it follows that
_

0
f(u

u
0
) dx 0
and

0
|u

|
p2
u

(u

u
0
) dx

c
_

0
|u

|
p2
|u

| |u

u
0
| dx
c

0
|u

|
p
dx

1/p

0
|u

u
0
|
p
dx

1/p
C

0
|u

u
0
|
p
dx

1/p
0.
Then we can derive from (2.10)
_

0
|u

|
p2
u

(u

u
0
) dx 0.
Therefore, one has
_

0
{|u

|
p2
u

|u
0
|
p2
u
0
}(u

u
0
) dx
=
_

0
|u

|
p2
u

(u

u
0
) dx
_

0
|u
0
|
p2
u
0
(u

u
0
) dx 0.
(Recall that u

u
0
in W
1,p
(

0
).) We nd that when p 2 (see (1.4)),
_

0
|(u

u
0
)|
p
dx c
_

0
_
|u

|
p2
u

|u
0
|
p2
u
0
_
(u

u
0
) dx 0. (2.11)
When 1 < p < 2, it holds that
_

0
|(u

u
0
)|
p
dx

0
_
|u

| +|u
0
|
_
p2
|(u

u
0
)|
2
dx

p/2

0
_
|u

| +|u
0
|
_
p
dx

(2p)/2
c

0
_
|u

|
p2
u

|u
0
|
p2
u
0
_
(u

u
0
) dx

p/2

0
_
|u

| +|u
0
|
_
p
dx

(2p)/2
0. (2.12)
PROCEEDINGS OF THE STEKLOV INSTITUTE OF MATHEMATICS Vol. 261 2008
292 M. CHIPOT, Y. XIE
From above, we derive that for 1 < p < +
_

0
1
|(u

u
0
)|
p
dx
_

0
|(u

u
0
)|
p
dx 0,
which shows that
u

u
0
in W
1,p
(

0
1
).
Therefore, u
0
satises

p
u
0
= f.
We nally conclude that u
0
= u

since u
0
is independent of X
1
. This completes the proof of the
theorem.
Remark 2.2. Some other convergence results including some rate of convergence are available.
We refer the reader to [9] for details.
3. A LIOUVILLE TYPE THEOREM
As we have seen at the end of the proof of Theorem 2.1, a key point in the convergence issue is
the uniqueness of the solution to
_

p
u
0
= f(X
2
) in R
q
,
u
0
= 0 on R
q
,
(3.1)
when, eventually, some other assumptions on u
0
, like boundedness, are imposed. This has a avor
of the Liouville theorem. In this direction let us prove
Theorem 3.1. Let u L

(R
q
; L
p
()) be a weak solution to
_

p
u = 0 in R
q
,
u = 0 on R
q
.
(3.2)
Then u 0; i.e., problem (3.2) does not admit any nontrivial solution.
Before turning to the proof of the theorem, let us make our assumptions precise. L

(R
q
; L
p
())
denotes the space of functions from R
q
with values in L
p
() that are essentially bounded. By a
weak solution to (3.2) we mean a function u such that, for any domain (, )
q
,
u W
1,p
((, )
q
) L

(R
q
; L
p
()), u = 0 on (, )
q
,
_
(,)
q

|u|
p2
uv dx = 0 v W
1,p
0
((, )
q
).
(3.3)
Let us now prove the above theorem.
Proof of Theorem 3.1. Let be a nonnegative smooth function such that
0 1, 1 in
_

1
2
,
1
2
_
, 0 outside (1, 1), |

| is bounded,
and set

(x) =
_
x

_
, :=

q
i=1

(x
i
). We have, if
p
denotes the pth power of ,
u
p
W
1,p
0
((, )
q
).
PROCEEDINGS OF THE STEKLOV INSTITUTE OF MATHEMATICS Vol. 261 2008
SOME ISSUES ON THE p-LAPLACE EQUATION 293
Plugging the above function into (3.3) leads to
_
(,)
q

|u|
p2
u(u
p
) dx = 0.
Therefore, we derive that
_
(,)
q

|u|
p2
uu
p
dx =
_
(,)
q

|u|
p2
u
p
udx
= p
_
(,)
q

|u|
p2
uu
p1
dx

_
(,)
q

|u|
p1
|u|
p1
dx.
Applying the Holder inequality to the right-hand side of the above inequality, we obtain
_
(,)
q

|u|
p

p
dx
c

_
(,)
q

|u|
p

p
dx

(p1)/p

_
(,)
q

|u|
p
dx

1/p
,
which is equivalent to
_
(,)
q

|u|
p

p
dx
c

p
_
(,)
q

|u|
p
dx. (3.4)
We remark that u vanishes on R
q
, i.e., for a.e. X
1
, u(X
1
, ) W
1,p
0
(). Exploiting Poincares
inequality in the X
2
-direction (see [3]) yields
_
(,)
q

|u|
p

p
dx
c

p
_
(,)
q

|
X
2
u|
p
dx
(
X
2
= (x
q+1
, . . . , x
n
)). Due to the choice that we made for , we obtain
_
(

2
,

2
)
q

|u|
p
dx
c

p
_
(,)
q

|u|
p
dx.
After iterating k 1 times this inequality, it comes that
_
(

2
,

2
)
q

|u|
p
dx
c

p(k1)
_
(2
k1
, 2
k1
)
q

|u|
p
dx.
Hence we derive from (3.4) that
_
(

2
,

2
)
q

|u|
p
dx
c

pk
_
(2
k
, 2
k
)
q

|u|
p
dx.
PROCEEDINGS OF THE STEKLOV INSTITUTE OF MATHEMATICS Vol. 261 2008
294 M. CHIPOT, Y. XIE
Since u L

(R
q
; L
p
()), we obtain
_
(

2
,

2
)
q

|u|
p
dx
c

pk
(2
k+1
)
q
ess sup |u(X
1
, )|
p
L
p
()
(ess sup denotes the essential supremum in X
1
). Choosing k so large that pk > q and letting
lead to
u = 0.
This completes the proof of the theorem.
Remark 3.2. One should notice that one cannot remove entirely the assumption
u L

(R
q
; L
p
()). (3.5)
Indeed, suppose that
p = 2, q = 1, = (, ). (3.6)
Then the function
u = e
x
1
sin x
2
(3.7)
is a nontrivial weak solution to
_
u = 0 in R ,
u = 0 on R .
(3.8)
ACKNOWLEDGMENTS
The authors have been supported by the Swiss National Science Foundation under the contracts
#20-103300/1 and #20-111543/1.
REFERENCES
1. J. W. Barrett and W. B. Liu, Finite Element Approximation of the Parabolic p-Laplacian, SIAM J. Numer.
Anal. 31 (2), 413428 (1994).
2. M. Chipot, Elements of Nonlinear Analysis (Birkhauser, Basel, 2000).
3. M. Chipot, Goes to Plus Innity (Birkhauser, Basel, 2002).
4. M. Chipot and A. Rougirel, On the Asymptotic Behaviour of the Solution of Elliptic Problems in Cylindrical
Domains Becoming Unbounded, Commun. Contemp. Math. 4 (1), 1544 (2002).
5. M. Chipot and A. Rougirel, On the Asymptotic Behaviour of the Solution of Parabolic Problems in Cylindrical
Domains of Large Size in Some Directions, Discrete Contin. Dyn. Syst. B 1 (3), 319338 (2001).
6. M. Chipot and A. Rougirel, Remarks on the Asymptotic Behaviour of the Solution to Parabolic Problems in
Domains Becoming Unbounded, Nonlinear Anal., Theory Methods Appl. 47 (1), 311 (2001).
7. M. Chipot and Y. Xie, On the Asymptotic Behaviour of the p-Laplace Equation in Cylinders Becoming Un-
bounded, in Nonlinear Partial Dierential Equations and Their Applications: Proc. Int. Conf., Shanghai, China,
2003, Ed. by N. Kenmochi, M.

Otani, and S. Zheng (Gakkotosho, Tokyo, 2004), pp. 1627.
8. W. Liu and N. Yan, Quasi-norm Local Error Estimators for p-Laplacian, SIAM J. Numer. Anal. 39 (1), 100127
(2001).
9. Y. Xie, On Asymptotic Problems in Cylinders and Other Mathematical Issues, Thes. (Univ. Z urich, May 2006).
This article was submitted by the authors in English
PROCEEDINGS OF THE STEKLOV INSTITUTE OF MATHEMATICS Vol. 261 2008