JOURNAL OF PARTIAL DIFFERENTIAL EQUATIONS J. Part. Diff. Eq., Vol. 26, No. 1, pp.

25-38

doi: 10.4208/jpde.v26.n1.3 March 2013

Existence of Nontrivial Weak Solutions to Quasi-Linear Elliptic Equations with Exponential Growth
WANG Chong
Department of Mathematics, the George Washington University, Washington DC 20052, USA. Received 13 June 2012; Accepted 1 December 2012

Abstract. In this paper, we study the existence of nontrivial weak solutions to the following quasi-linear elliptic equations

n u + V ( x )|u|n2 u =

f ( x , u) , |x|
n

( n 2),

where n u = div(|u|n2 u), 0 < n, V : R n R is a continuous function, f ( x, u) is continuous in R n R and behaves like eu n1 as u +. AMS Subject Classications: 35J20, 35J60 Chinese Library Classications: O175 Key Words: Trudinger-Moser inequality; exponential growth.

1 Introduction
Consider nonlinear elliptic equations of the form

p u = f ( x, u),

in ,

(1.1)

where is a smooth bounded domain in R n , and p u = div (|u| p2 u). Br ezis [1], Br ezis-Nirenberg [2] and Bartsh-Willem [3] studied this problem under the assumptions that p = 2 and | f ( x, u)| c(|u| + |u|q1 ). Garcia-Alonso [4] studied this problem under the assumptions that p n and p2 n. When = R n and p = 2, Kryszewski-Szulkin [5], Alama-Li [6], Ding-Ni [7] and Jeanjean [8] studied the following equations in stead of (1.1): u + V ( x)u = f ( x, u), in R n .
Corresponding author.

Email address: chongwang@gwu.edu (C. Wang) 25

http://www.global-sci.org/jpde/

26

C. Wang / J. Partial Diff. Eq., 26 (2013), pp. 25-38

In this paper we consider quasi-linear elliptic equations in the whole Euclidean space

n u + V ( x)|u|n2 u =
where n u = div(|u|n2 u), 0

f ( x, u ) , | x|
n

(n

2) ,

(1.2)

< n, V : R n R is a continuous function, f ( x, u)

is continuous in R n R and behaves like eu n1 as u +. D. Cao [9] and Cao-Zhang [10] studied problem (1.2) in the case n = 2 and = 0. et al. [12, 13] and Alevs-Figueiredo [14] studied problem (1.2) in general Panda [11], do O et dimension and = 0. When = 0, (1.2) was studied by Adimurthi-Yang [15], do O 4 al. [16], Yang [17], Zhao [18], and others. Similar problems in R or complete noncompact Riemannian manifolds were also studied by Yang [19, 20]. We dene a function space E with the norm u
Rn

u W 1,n (R n ) :

Rn

V ( x )|u |n d x <

(|u| + V ( x)|u| )d x

1 n

(1.3)

We say that u E is a weak solution of problem (1.2) if for all E we have

Rn

(|u|n2 u + V ( x)|u|n2 u )d x =

Rn

f ( x, u ) d x. | x|

If a weak solution u satises u( x) 0 for almost every x R n , we say u is positive. Throughout this paper we assume the following two conditions on the potential V ( x): (V1 ) V ( x) V0 > 0; 1 (V2 ) The function V ( belongs to L1/(n1) (R n ). x) We also assume that the nonlinearity f ( x, s) satises the following: ( H1 ) There exist constants 0 , b1 , b2 > 0 such that for all ( x, s) R n R + ,

| f ( x, s) |

b1 s

n 1

+ b2 e

0 | s | n 1

kn n 2 k 0 | s | n 1

k=0

k!

( H2 ) There exists > n, such that for all x R n and s > 0,

0 < F ( x, s)
s 0

f ( x, t )dt

s f ( x, s );

( H3 ) There exist constants R0 , M0 > 0, such that for all x R n and s > R0 ,
F ( x, s) M0 f ( x , s ) ;

Existence of Nontrivial Weak Solutions to Quasi-Linear Elliptic Equations

27

( H4 )
s 0+

lim sup

n | F ( x, s) | < sn

uniformly with respect to x R n , where inf u

Rn n

u E,u =0

|u|n

dx |x|

( H5 ) There exist constants p > n and C p such that

f ( x, s) for all s 0 and all x R n , where pn ) Cp > ( p
1 pn n

C p s p 1 ,

n 0 ( n ) n

( n 1)( p n) n

Sp ,

n = nn1 n1 , n1 is the volume of the unit sphere S n1 , and Sp inf

Rn

(|u|n + V ( x)|u|n )d x) (
|u| p Rn |x| d x)
1 p

1 n

u E,u =0

= inf

u E,u =0

|u| p Rn |x| d x)

1 p

( H6 ) when s 0, f ( x, s) = 0 for all x R n .

Our main result is the following theorem: Theorem 1.1. Assume that V(x) is a continuous function satisfying (V1 ) and (V2 ). f : R n R R is a continuous function and the hypotheses (H1 )-(H6 ) hold. Then Eq. (1.2) has a nontrivial positive weak solution. Here the assumption ( H5 ) is different from that of [17]. ( H5 ) was also used in [16] and [18]. An example of f satisfying ( H1 )-( H6 ) reads f ( t) =
(t 2l l ! C p k= l
n 1 n 1 ( t ) t n 1 ) k

k!

t 0, t < 0,

0,

where l N is an integer, C p is as in ( H5 ), : [0, ) R is a smooth function such that 0 1, 0 on [0, A], 1 on [2 A, ), and | | 2/ A, where A is a large constant, say A > 4n1 . For details we refer the reader to in [20, Proposition 2.9]. Other examples were also given in [16] and [18] respectively.

28

C. Wang / J. Partial Diff. Eq., 26 (2013), pp. 25-38

2 Compactness analysis
We will give some preliminary results before proving Theorem 1.1. Dene a function : N R R by n 2 k s sk ( n, s) = es = . (2.1) k ! k= n 1 k ! k=0 Let s 0, p 1 be real numbers and n 2 be an integer, then there holds (see [17])
p

( n, s)

(n, ps).

(2.2)

Problem (1.2) is closely related to a singular Trudinger-Moser type inequality [15]. That is, for all > 0, 0 < n, and u W 1,n (R n ) (n 2), there holds e | u | n 1
Rn
n

n 2

k | u | n 1 k! k=0

kn

| x|

d x < .

(2.3)

Furthermore, we have for all

) n and > 0, (1 n

e | u | n 1 sup
( | u|n + |u|n )d x 1 Rn
Rn

n 2 k | u | n 1 k=0

kn

k!

| x|

d x < .

(2.4)

In this paper, we also need the following result which is taken from Lemma 2.4 in [17]. That is, if V : R n R is continuous and (V1 ), (V2 ) are satised, then for any q 1, there holds E Lq (R n ) compactly. (2.5) Dene a functional J : E R by J (u) 1 u n
n

F ( x, u ) d x, Rn | x|
s

where 0 < n, u is the norm of u E dened by (1.3), F ( x, s) = 0 f ( x, t)dt is the primitive of f ( x, s). Assume f ( x, u) satises the hypotheses ( H1 ), then there exist some positive constants 1 > 0 and b3 such that for all ( x, s) R n R, F ( x, s) b3 n, 1 |s| n1 ,
n

where (n, s) is dened by (2.1). Thus J is well dened thanks to (2.3).

Existence of Nontrivial Weak Solutions to Quasi-Linear Elliptic Equations

29

Lemma 2.1. Assume V ( x) V0 in R n , (H1 ), (H2 ) and (H3 ) hold. Then for any nonnegative, compactly supported function u W 1,n (R n )\{0}, there holds J (tu) as t + . Proof. We follow the line of [15]. ( H2 ) and ( H3 ) imply that there exists R0 > 0 such that for all s > R0 , s F ( x, s) = lnF ( x, s) , s F ( x, s) s therefore, s R0 It follows that F ( x, s)

F ( x, s) . F ( x, R0 )

F ( x, R0 ) R0 s .

Let c1 = F ( x, R0 ) R0 , then we have for all ( x, s) [0, +), F ( x, s) c1 s c2 , which is under the assumption that u is supported in a bounded domain and c2 is a positive constant. This implies that J (tu) tn u n
n

c1 t u d x = tn | x|

u n

c1 t n

u dx . | x|

Since > n, this implies J (tu) as t +. Lemma 2.2. Assume that V ( x) V0 in R n , (H1 ) and (H4 ) are satised. Then there exist > 0 and r > 0 such that J (u) for all u = r. Proof. According to ( H4 ), there exist , > 0 such that if |s| n | F ( x, s) | < . |s|n Therefore for all x R n , |s| , we have F ( x, s) n |s| . n , (2.7)

(2.6)

On the other hand, according to ( H1 ), we can obtain that for any |s| F ( x, s) where C = b1 n||
( 0 | | n 1 ) k k! k= n 1

C | s | n + 1 R ( 0 , s ) , b + 2n , | |

( 0 | s | n 1 ) k . R ( 0 , s ) = k! k= n 1

Combining (2.6) and (2.7), we have for all ( x, s) R n R n , F ( x, s) n | s | + C | s | n +1 R ( 0 , s ), n (2.8)

30 where C = C . Here we also use the inequality

C. Wang / J. Partial Diff. Eq., 26 (2013), pp. 25-38

| u | n +1 R ( 0 , u ) dx | x| Rn

C u

n +1

(2.9)

which is taken from Lemma 4.2 in [15]. According to the denition of , we get

|u |n dx Rn | x|
Thanks to (2.8), (2.9), and (2.10), we obtain J (u) 1 u n 1 u n

u n .

(2.10)

|u |n dx n | x| Rn u n n C u n
n

Rn

| u | n +1 R ( 0 , u ) dx | x|

n +1

= u

u n

n 1

C u

(2.11)

For sufciently small r > 0, we have n 1 Crn r n n 1 r , 2n (2.12)

which is due to > 0. Therefore, according to (2.11) and (2.12), for all u J (u) r r n 1 = rn . 2n 2n for all u = r.

= r,

Finally, let = 2n rn , we have J (u)

Lemma 2.3. Critical points of J are weak solutions of (1.2). Proof. Though the proof is standard, we write it for completeness. Dene a function g(t) = J (u + t ), namely g( t) = 1 n

Rn

(|(u + t )|n + V ( x)|u + t |n )d x

F ( x, u + t ) d x. | x| Rn

By a simple calculation, g ( t)
t =0

= J (u + t )
t =0

= J ( u ) .

Let f1 (t) = |(u + t )|n , f2 (t) = |u + t |n , and f 3 ( t) = F ( x, u + t ) d x. | x| Rn

Existence of Nontrivial Weak Solutions to Quasi-Linear Elliptic Equations

31

Clearly we have f 1 ( t)
t =0

= =
t =0

n 2 |u|n2 u = n|u|n2 u , 2 n n 2 | u | 2u = n | u | n 2 u , 2
Rn

f 2 ( t) f 3 ( t)
t =0

f ( x, u ) d x. | x|

Combining the above, we have for all E, J (u) =

Rn

(|u|n2 u + V ( x)|u|n2 u )d x

Rn

f ( x, u ) d x. | x|

(2.13)

Therefore, J (u) = 0 is equivalent to

Rn

(|u|n2 u + V ( x)|u|n2 u )d x

Rn

f ( x, u ) d x = 0. | x|

Hence we get the desired result. Lemma 2.4. Assume (H5 ) is satised, then there exists a function u p E which satises u p = S p , and for t [0, +), we dene J (tu p ) There holds max J (tu p ) <
t 0

tn up n

Rn

F ( x, tu p ) d x. | x|
n 1

1 n n n n 0

(2.14)

Proof. Similar to [18], assume {uk } is a bounded positive sequence of functions in E which satises |u k | p d x = 1 and uk S p . Rn | x| Meanwhile we assume that uk u p in E, uk u p in Lq (R n ) for all q 1, uk ( x) u p ( x) almost everywhere. Using the Holder inequality and the Mean Value Theorem, we can easily prove that for any > 0, there exists a constant K such that when k > K ,

Rn

|uk | p |u p | p d x < . | x| |u p | p d x = 1. | x|

Therefore,
Rn

|u k | p dx | x|

Rn

(2.15)

32 Next we will prove up

k

C. Wang / J. Partial Diff. Eq., 26 (2013), pp. 25-38

lim inf uk = S p .

(2.16)

Since uk u p weakly in E, we know uk u p weakly in Ln (R n ). According to the denition of weak convergence and the Holder inequality, we get
Rn

|u p |n d x

lim inf

Rn

|uk |n d x.

(2.17)

Similarly to the proof of (2.15), we know

Rn

V ( x )|u k |n d x

Rn

V ( x ) |u p |n d x.

(2.18)

Thanks to (2.17) and (2.18), (2.16) holds. Meanwhile, by the denition of S p , we know Sp u p . Therefore, we know u p = S p . According to ( H5 ), we have F ( x, tu p ) dx | x| Rn Due to the denition of J (tu p ) and (2.19), we have J (tu p ) Let f ( t) = tn n tp Sp Cp . n p tn n tp Sp Cp , n p S pn Cp tp p
1 pn

Cp

tp . p

(2.19)

and by calculation we know for any real number t, f ( t) This means tn n If we set pn Cp > ( ) p then we have p n Sp 1 < n np n C pp n In view of ( H5 ), we get (2.14) immediately.
np pn pn n

.
np

S pn C p

pn p n Sp n . np pn Cp

n 0 ( n ) n

( n 1)( p n) n

Sp ,

n n n 0

n 1

Existence of Nontrivial Weak Solutions to Quasi-Linear Elliptic Equations

33

Lemma 2.5. Assume that (V1 ), (V2 ), (H1 ), (H2 ) and (H3 ) hold and {uk } E be an arbitrary Palais-Smale sequence of J, i.e., J ( u k ) c, J ( u k ) 0 in E as k , where E denotes the dual space of E. Then there exists a subsequence of {uk } (still denoted by {uk } ) and u E such that uk u weakly in E, uk u strongly in Lq (R n ) for all q 1, and u k ( x ) u ( x ), a. e.in R n , f ( x, u ) f ( x, u k ) , stronglyin L1 (R n ), | x | | x | F ( x, u k ) F ( x, u ) , stronglyin L1 (R n ). | x| | x| Furthermore, u is a weak solution of (1.2). Proof. Assume {uk } is a Palais-Smale sequence of J . Since J (uk ) c, we obtain 1 u n k According to (2.13), we know
n

Rn

F ( x, u k ) d x c, as k . | x|

(2.20)

| J (uk ) | =

Rn

(|uk |n2 uk + V ( x)|uk |n2 uk )d x

Rn

f ( x, u k ) dx | x| (2.21)

k ,

for all E, where k = J (uk ) , and k 0 as k . Taking = uk in (2.21), we have uk By ( H2 ), we obtain

Rn n

Rn

f ( x, u k ) uk d x | x|

k uk .

(2.22)

F ( x, u k ) dx | x|
n,

Rn

u k f ( x, u k ) d x. | x|

(2.23)

Then considering ( n 1) uk 1 n uk
n

according to (2.23), we have 1 n uk F ( x, u k ) u k f ( x, u k ) dx | x| u k f ( x, u k ) d x uk n | x| Rn

n

Rn

= J (uk )+

2|c| + k uk .

(2.24)

34

C. Wang / J. Partial Diff. Eq., 26 (2013), pp. 25-38

According to (2.24), its easy to prove that uk is bounded. Due to (2.20) and (2.22), we get u k f ( x, u k ) F ( x, u k ) d x C, d x C, (2.25) | x| Rn Rn | x| where C is a constant which depends only on and n. According to (2.5) we obtain that for some u E and any q 1, up to a subsequence, uk u strongly in Lq (R n ). Then we know uk u almost everywhere in R n . Next we will prove that up to a subsequence
k

lim

Rn

| f ( x, u k ) f ( x, u ) | d x = 0. | x|

(2.26)

Due to f ( x, )

0, it is sufcient for us to prove that up to a subsequence lim f ( x, u k ) dx = | x| f ( x, u ) d x. | x| (2.27)

Rn

Rn

Due to

f ( x, u ) L1 (R n ), | x| lim f ( x, u ) d x = 0. | x|

we know
+ |u|

For any > 0, there exists M > C such that

|u| M

f ( x, u ) d x < , | x|

(2.28)

where C is the constant in (2.25). According to (2.25), we know f ( x, u k ) dx | x| 1 M f ( x, u k ) u k d x < . | x| (2.29)

|uk | M

|uk | M

For all x { x R n : |uk | < M }, by our assumption ( H1 ), we can deduce that

| f ( x, s) |
n

b1 + b2 e0 M

n n 1

| s | n 1 .

(2.30)

Let C1 = b1 + b2 e0 M n1 , according to (2.30), we know

| f ( x, uk ( x))|
Since

C1 |uk ( x)|n1 .

| u k | n 1 | u | n 1 strongly in L1 (R n ), and uk u almost everywhere in R n , (2.31) | x| | x|

Existence of Nontrivial Weak Solutions to Quasi-Linear Elliptic Equations

35

according to the generalized Lebesgues dominated convergence theorem, we know lim f ( x, u k ) dx = | x| f ( x, u ) d x. | x| (2.32)

k |u k |< M

|u|< M

According to (2.28), (2.29) and (2.32), we can prove that (2.27) holds. Therefore we get (2.26). By ( H3 ) and ( H1 ), we obtain that F ( x, u k )
n n 1

C1 |uk |n + C2 f ( x, uk ),

where C1 =(b1 /n)+ b2 e0 R0 and C2 = M0 . According to (2.5), (2.26), and the generalized Lebesgues Dominated Convergence Theorem, we know lim

| F ( x, u k ) F ( x, u ) | d x = 0. | x| Rn

Using the knowledge of (4.26) in [15], we know uk ( x) u( x) almost everywhere in n n R n and |uk |n2 uk |u|n2 u weakly in L n1 (R n ) . Let k in (2.21), and then we obtain that
Rn

(|u|n2 u + V ( x)|u|n2 u )d x

Rn

f ( x, u ) d x = 0, | x|

(R n ). This demonstrates that u is a weak solution of (1.2). for all C0

3 Proof of Theorem 1.1

Next we will prove Theorem 1.1. By Lemmas 2.1 and 2.2, we know J satises all the hypotheses of the Mountain-pass Theorem without the Palais-Smale condition: J C 1 ( E, R ); J (0) = 0; J (u) > 0, when u = r; J (e) < 0, for some e E with e > r. According to the Mountain-pass Theorem except for the Palais-Smale Condition [21], there exists a sequence {uk } E such that J (uk ) c > 0, in E , where c = min max J (u)
u

J (uk ) 0,

and

= C ([0,1], E) : (0) = 0, (1) = e .

36

C. Wang / J. Partial Diff. Eq., 26 (2013), pp. 25-38

According to Lemma 2.5, we know that up to a subsequence u u weakly in E, k q n uk u strongly in L (R ), for any q F ( x, u k ) F ( x, u ) lim dx = d x, k R n | x | Rn | x| u is a weak solution of (1.2).

1,

Next we will prove that the solution u which we get in the above is nontrivial. Suppose u 0. Due to F ( x, u) 0 for all x R n , we get lim F ( x, u k ) dx = Rn | x| F ( x, u ) d x = 0. Rn | x| (3.1)

According to (2.20), we have lim 1 uk n

n

Rn

F ( x, u k ) d x = c > 0. | x|

(3.2)

Combining (3.1) and (3.2), we can obtain that

k

lim

uk

= n c > 0.

(3.3)

By Lemma 2.4, we get 0 < nc < n n n 0

n 1

(3.4)

According to (3.3) and (3.4), we know there exists some 0 > 0 and K > 0, such that uk
n

n n 0 n 0

n 1

(3.5)

for all k > K . According to (3.5), we can choose q > 1 sufciently close to 1 such that q 0 u k
n n 1

0 0 n , n 2

(3.6)

for all k > K . By ( H1 ) and (2.1), we have

| f ( x, u k ) u k |
It follows that

b1 |uk |n + b2 |uk | (n, 0 |uk | n1 ).

| f ( x, u k ) u k | dx | x| Rn

b1

Rn

|u k |n d x + b2 | x|

| u k | ( n , 0 | u k | n 1 ) d x. | x| Rn

(3.7)

Existence of Nontrivial Weak Solutions to Quasi-Linear Elliptic Equations

37

Letting 1/q + 1/q = 1, and according to (3.6), (3.7), the Holder Inequality, (2.2), and (2.4), we have

| f ( x, u k ) u k | d x b1 n | x| R
b1

Rn

|u k |n d x + b2 | x| |u k |n dx + C | x|

|u k |q dx Rn | x| |u k |q dx Rn | x|

1 q

( n , q 0 | u k | n 1 ) dx n | x| R as k .

1 q

1 q

Rn

0,

(3.8)

According to (2.22) and (3.8), we have uk 0, as k .

This is in contradiction with (3.3). Thus the solution u of (1.2) is nontrivial. Testing Eq. (1.2) with u , the negative part of u, we conclude that u 0. Hence u 0.

Acknowledgments
The author is supported by the China Scholarship Council.
References [1] Br eis H., Elliptic equations with limiting Sobolev exponents. Comm. Pure Appl. Math., 39 (1986), 517-539. [2] Br eis H., Nirenberg L., Positive solutions of nonlinear elliptic equations involving critical Sobolev exponents. Comm. Pure Appl. Math., 36 (1983), 437-477. [3] Bartsh T., Willem M., On an elliptic equation with concave and convex nonlinearities. Proc. Amer. Math. Soc., 123 (1995), 3555-3561. [4] Garcia A., Alonso P., Existence and non-uniqueness for the p-Laplacian. Commun. Partial Differential Equations, 12 (1987), 1389-1430. [5] Kryszewski W., Szulkin A., Generalized linking theorem with an application to semilinear Schrodinger equation. Adv. Diff. Equ., 3 (1998), 441-472. [6] Alama S., Li Y., Existence of solutions for semilinear elliptic equations with indenite linear part. J. Diff. Equ., 96 (1992), 89-115. [7] Ding W., Ni W., On the existence of positive entire solutions of a semilinear elliptic equation. Arch. Rat. Math. Anal., 31 (1986), 283-308. [8] Jeanjean L., Solutions in spectral gaps for a nonlinear equation of Schrodinger type. J. Diff. Equ., 112 (1994), 53-80. [9] Cao D., Nontrivial solution of semilinear elliptic equations with critical exponent in R2 . Commun. Partial Differential Equations, 17 (1992), 407-435. [10] Cao D., Zhang Z., Eigenfunctions of the nonlinear elliptic equation with critical exponent in R2 . Acta Math. Sci., (English Ed.) 13 (1993), 74-88. [11] Panda R., Nontrivial solution of a quasilinear elliptic equation with critical growth in R n . Proc. Indian Acad. Sci. (Math. Sci.), 105 (1995), 425-444.

38

C. Wang / J. Partial Diff. Eq., 26 (2013), pp. 25-38

J. M., n-Laplacian equations in R n with critical growth. Abstr. Appl. Anal., 2 (1997), [12] do O 301-315. J. M., Medeiros E. and U. Severo, On a quasilinear nonhomogeneous elliptic equation [13] do O with critical growth in R n . J. Diff. Equ., 246 (2009), 1363-1386. [14] Alves C. O., Figueiredo G. M., On multiplicity and concentration of positive solutions for a class of quasilinear problems with critical exponential growth in R n . J. Diff. Equ., 246 (2009), 1288-1311. [15] Adimurthi, Yang Y., An interpolation of Hardy inequality and Trudinger-Moser inequaltiy in R n and its applicaitons. International Mathematics Research Notices, 13 (2010), 2394-2426. J. M., Souza M. de, On a class of singular Trudinger-Moser type inequalities and its [16] do O applications. Math. Narchr., 284 (2011), 1754-1776. [17] Yang Y., Existence of positive solutions to quasi-linear elliptic equations with exponential growth in the whole Euclidean space. Journal of Functional Analysis, 262 (2012), 1679-1704. [18] Zhao L., A multiplicity result for a singular and nonhomogeneous elliptic problem in R n . Journal of Partial Diff. Equ., 25 (2012), 90-102. [19] Yang Y., Adams type inequalities and related elliptic partial differential equations in dimension four. J. Diff. Equ., 252 (2012), 2266-2295. [20] Yang Y., Trudinger-Moser inequalities on complete noncompact Riemannian manifolds. Journal of Functional Analysis, 263 (2012), 1894-1938. [21] Rabinowitz P. H., On a class of nonlinear Schrodinger equations. Z. Angew. Math. Phys., 43 (1992), 270-291.