You are on page 1of 30

This article was downloaded by: [Michigan State University]

On: 11 January 2015, At: 19:14


Publisher: Taylor & Francis
Informa Ltd Registered in England and Wales Registered Number: 1072954 Registered office: Mortimer
House, 37-41 Mortimer Street, London W1T 3JH, UK

Communications in Partial Differential Equations


Publication details, including instructions for authors and subscription information:
http://www.tandfonline.com/loi/lpde20

Nontrivial Solution of Semilinear Elliptic Equations


with Critical Exponent in R
a
D. M. Cao
a
Wuhan Institute of Mathematical Sciences, Academia Sinica , P.O. Box 71007,
Wuhan, 430071, P.R. China
Published online: 18 May 2010.

To cite this article: D. M. Cao (1992) Nontrivial Solution of Semilinear Elliptic Equations with Critical Exponent in R ,
Communications in Partial Differential Equations, 17:3-4, 407-435, DOI: 10.1080/03605309208820848

To link to this article: http://dx.doi.org/10.1080/03605309208820848

PLEASE SCROLL DOWN FOR ARTICLE

Taylor & Francis makes every effort to ensure the accuracy of all the information (the “Content”) contained
in the publications on our platform. However, Taylor & Francis, our agents, and our licensors make no
representations or warranties whatsoever as to the accuracy, completeness, or suitability for any purpose
of the Content. Any opinions and views expressed in this publication are the opinions and views of the
authors, and are not the views of or endorsed by Taylor & Francis. The accuracy of the Content should
not be relied upon and should be independently verified with primary sources of information. Taylor
and Francis shall not be liable for any losses, actions, claims, proceedings, demands, costs, expenses,
damages, and other liabilities whatsoever or howsoever caused arising directly or indirectly in connection
with, in relation to or arising out of the use of the Content.
This article may be used for research, teaching, and private study purposes. Any substantial or systematic
reproduction, redistribution, reselling, loan, sub-licensing, systematic supply, or distribution in any
form to anyone is expressly forbidden. Terms & Conditions of access and use can be found at http://
www.tandfonline.com/page/terms-and-conditions
COMMUN. IN PARTIAL DIFFERENTIAL EQUATIONS, 17(3&4), 407-435 (1992)

NONTRIVIAL SOLUTION
OF SEMILINEAR ELLIPTIC EQUATION
WITH CRITICAL EXPONENT IN R2

D.M. CAO

Wuhan Institute of Mathematical Sciences,


Academia Sinica
P.O. Box 71007
Wuhan 430071,
P.R. China
Downloaded by [Michigan State University] at 19:14 11 January 2015

~
I
I. Introduction.

I We consider the problem

1 where f ( x , t ) = O(t) as t -+ 0 and I f ( x , t ) ( 5 ~


constant C and for all x E W2. t E R.
e for~ some
" positive
~ ~

When If(x,t)( < CeltlY (0 < v < 2), G.B. Li and X.P. Zhu [4],
P.L. Lions [7] have studied (1.1). M.C. Shaw (101 has obtained the
existence of nontrivial solution for the following eigenvalue problem

where R is a bounded domain in W2,g(x, t) is a continuous function on


S1 x 64 satisfying

' 3
Copyright @ 1992 by Marcel Dekker, Inc.
g(x:O) = 0
Idx,t)I I A + BItI
m at2
e uniformly in x
B, rn and cu
for some nonnegative constants -4, > 0.
It is well-known that the embedding of H 1 ( R 2 )into Lp(W2) (2 5 p <
+co) and of the set V = { u E H,'(n) , Jn IVuI2 dx <
1 ) into Orlicz
space Lv (a)
where d ( t ) = e4nt2- 1 is not compact and consequently
we can not hope functional

to satisfy the (PS) (Palais-Smale) condition in W if we apply variational


method. In this paper, with the help of concentration-compactness prin-
ciple put forward by P.L.Lions [6],[7], we show that I(u) satisfies (PS),
condition for all c E ( 0 ,J ) (for the definition of J , see section 3). Then,
we obtain the existence of nontrivial solution for (1.1) by using Moun-
tain Pass lemma in H. Brezis and L. Nirenberg (31 to construct a critical
Downloaded by [Michigan State University] at 19:14 11 January 2015

point 1 ( u ) with a critical value in ( 0 ,J).

The main result in this paper are as follows

THEOREM 1.1. Suppose f (x,t) E C(R2 x W) satisfies the following


conditions :

(fl) 1f (+,t)l < ~e~~~~ for x E R2 , t E R where C > 0 is some


constant.

f ( 5 ,t ) = 0
lim - unifonnly with respect to x E W 2 ;
t-0 t
lim- f (x,f > = $03 unifonnlywithrespecttox~~~.
t-co t
NONTRIVIAL SOLUTION OF SEMILINEAR ELLIPTIC EQUATION
-
(f4) limlZl,, f (x, t ) = f (t) uniformly for t bounded,

where 6 > 0 is some constant, E ( R )--+ 0 as R -r +a.


(f5) f ( ~ , t ) 1 T ( t ) > ;s;(I - zql-5 ~ t l p - ~for
t x E R ~ t, E W,
where 2 < p < +m,

f (z, t) f T(t) for fixed t E R with respect to x E WZ. Then (1.1) has a
nontrivial s o h tion.

If we are interested in positive solution, we have a similar result.


Downloaded by [Michigan State University] at 19:14 11 January 2015

THEOREM
1.2. Suppose f(x, t) satisfies (f1)-(f4) and
(f5 :,
f(x,t) >0 for x E W 2 , t > 0 ,
f (x, t) > 3(t) > ;s;(l - 26)'-f ltlp-'t for x E R2 , t 2 0 .

where 2 < p < +a,


Sp as defined in Theorem 1.1.

f(x,t) (x,t)Efix(0,6)
where R is some domain in RZ with positive measure and 6 is some
positive number.
Then (1.1) has a positive solution.

It should be remarked that assumptions like (f3), (f5) and (f5') seem
just technical and leave room for improvement.

2. Preliminary.

We begin this section by giving some notations and definitions. Define


CAO

(L2(l~u12
1

(2.1) \lu// = + u2)dx)' . ~)


for u E H ~ ( [ W

for u E L Q ( R ~. )

The variational functional of (1.1) is

where
F(x,u ) = lu f(x, -
Let

where
lu
Downloaded by [Michigan State University] at 19:14 11 January 2015

-
F(u) = f ( s )ds .

cO" = inf { I m ( u ) / u E M m ) , if Mw # 0;
(2.6)
Cw = +m , ifMm=8.

If f ( t ) satisfies ( f l )and ( f 2 )then I? > 0.


REMARK:
NONTRIVIAL SOLUTION OF SEMILINEAR ELLIPTIC EQUATION 411

PROOF:Suppose, on the contrary, IF = 0, then there exists a sequence


{u,) such that
=
l2 dx

i2 iVu,12 dz 4 0.

By Lemma 2.2 (to be proved), ( f l ) and (f2)

Ii2 ~ ( u . 1dxl
1
z k 2/u,I2 dx + Cl k2 ~ u ~ J ( e -~ 1' -~ 4r(un12)dx
~ ~ l ~
1

:A2 /un12dx + c2(i2 1vun12 dx)' l2


Iunl2 dx
where C1, C2 > 0 are some constants.
Thus
Downloaded by [Michigan State University] at 19:14 11 January 2015

therefore

a contradiction which concludes our remark.

Similar to J..4.Moser [8], we have the following embedding lemma :

where C ( m ,M) > 0 is some constant independent of u.

PROOF:As in J.M. Moser [8], we use the method of symmetrization.


Let u* be the symmetrization of u, then u* depends on 1x1 only and is
nonnegative nonincreasing function of 1x1. It is well-known that

where r > 1 is a number to be determined.


Downloaded by [Michigan State University] at 19:14 11 January 2015

Setting IxI2 = w(t) = 2&u*(x), we deduce

By Holder's inequality,

where s o is some unit vector in R2


NONTRIVIAL SOLUTION OF SEMILINEAR ELLIPTIC EQUATION 413

= nr2 1 05

ew2(')-' dt

<
- Tr2e4~+lu'('.o)12
1°5 e(l+c)lvu*l:'-t dt (for any E > 0)

On the other hand, by radial lemma A.IV in H. Berestycki and


P.L. Lions [2] we have
(2.16) Iu*(x)I 5 &Ju*lzJzl-' for l x l # O
Thus
Downloaded by [Michigan State University] at 19:14 11 January 2015

Choosing r > &1un12, we have

(2.17) (e4nlu'1z - 1)dz 5 C(M) .

(2.15) and (2.17) deduces (2.8).

LEMMA 2.2. Suppose u E H'(W2), JR2 (VuI2dx < &. Then

where C > 0 is some constant independent of u.


i

I
PROOF:By the result of G. Talenti [ll]or Th. Aubin [I],we know that
IJ 5 5
if s,t > 1, = - $ and lVhl E Li(W2)then
I

I (2.19) lhls 5 K ( t ) IVhlt

I~
414

where

Let h = Iulv, v = 2n-I ' n 2 2. Then (Vh[ = vlulU-'IVuI. Taking


s = 2n - 1, t = 2 - &, using Hijlder's inequality, we get

By direct computation we obtain

where > 0 is some constant.


Thus
Downloaded by [Michigan State University] at 19:14 11 January 2015

In the above inequalities, we have used the same C to denote various


constants. In what follows we shall adopt such a notation.

Because limn,+, += e
(n!) i
i
: IVu12 < 1
, xL=27IVu1;
ca (4-n)"

converges. Hence

Denote BR = {t E W 2 , 1x1 < R).We have the following lemma which


can be regarded as an extension of lemma 1.1 in [7]due to
P.L. Lions.

LEMMA2.3. Suppose {un) C H1(R2)satisfies


NONTRIVIAL SOLUTION OF SEMILINEAR ELLIPTIC EQUATION 415

and

lim sup ( J V U , ~ ~ + ~ U ~ ) =~ O) ~ X forsome R > O .


n--.co yes2 I y + B n

f (x,t ) satisfies (fl). Then

PROOF:Let E Cr(WZ)such that t r 1 for 1x1 < 5 R and r 0 for


1x1 > R ;lVJl < 6 . Let Jy = E ( . + y ) , then
Downloaded by [Michigan State University] at 19:14 11 January 2015

For n large enough, we have IV(tyun)li < &. Hence, it follows from
Lemma 2.2 that
Covering W 2 by balls of radius $ in such a way that any point of R2
is contained in at most m balls of radius R. For large n we have

Taking n -+ co,then E -, O+ in (2.23) we deduce

Similarly we have
Downloaded by [Michigan State University] at 19:14 11 January 2015

LEMMA2.4. 1f lff(t)satisfies (fl),(f2)and there exists ii E H'(W2) such


that ii $ 0, f R 2 F(E)dx 2 sR2
ii2 dx, SR2IVE(*dx < 1. Then Ir can
be achieved. Moreover IF < JR2 IViiI2 dx.

PROOF:It is easy to see that there exists f E (O,l] such that

Thus, I? < Jm2 IVii12 dx. Letting { u n ) be a minimizing sequence of

I F , without loss of generality we can assume


I

1 NONTRIVIAL SOLUTION OF SEMILINEAR ELLIPTIC EQUATION

Denoting the symmetrization of un by u:, we know that u: depends


on 1x1 only and is nonincreasing as a function of 1x1. Furthermore

Thus, {u:) is still a minimizing sequence of I?. We denote it by (u, )


simply in what follows. Suppose
un + u
n
,
weakly in H'(FP~)

-
Downloaded by [Michigan State University] at 19:14 11 January 2015

Atz
+
,
Let Q(t) = eltm - 1 t2, P(t) = F(t). It follows from ( f l ) and (f2)
that P ( t ) + 0 as t 4 0, $#
4 0 as it( 03.

Due to invariance of the Dirichlet integral by the transformation u -+


U, = u(c.) we can assume Sw2
IunI2dx = 1 (otherwise, replace u,(x) by
un(anx), properly choose a,). By Lemma 2.1,

L (e*iun12 - 1) dx 5 C for same constant

Thus, by Strauss lemma and un + 0 uniformly as n --+


C>0.

cx, (by radial


n
lemma A.IV of H. Berestycki and P.L. Lions [2))we have

Therefore u $ 0.

= Ja2~(x)dx.

If ~ ~ , u 2 =d JR2F(u)dx,
x then u achieves IT. If Jm,u2dx <
CAO

JR, F ( u )dz? then there exists t E (O? 1) such that

Hence

lim
n-+a,
- J12
I V U , ~ dz
~ = 17,
a contradiction which concludes our proof.

3. Proof of Main Results.

First, we prove a local compactness result. That is, I ( u ) satisfies


( P S ) ccondition for c E (0, J ) where J = min (CM, !j - 0).

L E M M A 3.1. Suppose f(x,t) satisfies (f1)-(f4) and there exists a


Downloaded by [Michigan State University] at 19:14 11 January 2015

sequence {u,) C H1(R2) such that


I(un) + c E (0, J)

I1(un) -+ 0 in H - ' ( w ~ ) .
n

Then there exists a subsequence of { u , ) which converges strongly in


H1(R2).

PROOF:It follows from the assumption that

for any E H1(W2),where o(1) denote the quantities that tend to 0 as


n -+ CO, Cn -4
n
0 in H-'(R2).

Taking y = u , in (3.2) we obtain


NONTRIVIAL SOLUTION OF SEMILINEAR ELLIPTIC EQUATION 41 9

we shall prove { I , ) has a consequence converging strongly in H ' ( R ~ )


by applying concentration-compactness principle of P.L. Lions [?I. By
( 3 . 1 ) , ( 3 . 3 ) and ( f 3 ) :we have

11unlI
2
5 -21 + - C
1-20
+
o(1).

Therefore for n large enough

where m < 1 is some positive number.


Suppose for some u E H 1 ( R 2 )

(3.5) u, + u weakly in H'(W2)


(3.6) un -+ u a.e. in R2 .
n

Without loss of generality, we may assume


Downloaded by [Michigan State University] at 19:14 11 January 2015

b 2 1 ~ u n 1 2 d x 5 m 1< forall n 2 1 .

If I = 0, by (fl),(f2) and Lemma 2.2, we have


which contradicts c = lim,,, I ( u n ) # 0. Thus, 1 > 0.

Let pn = IVunI2 +IunI2. By lemmaI.l of P.L. Lions in [6], there are


three possibilities : (1) vanishing, (2) dichotomy and ( 3 ) compactness.
We shall show that only ( 3 ) compactness occurs. We argue by way of
contradiction.

Suppose (1) vanishing occurs, that is,

~ X0
( I V U , / ~ + / U , ~ ~ )+ forall R>O
Y"p2
ER 4 + B n

Lemma 2.3 deduces immediately


Downloaded by [Michigan State University] at 19:14 11 January 2015

/ 2 dx
The above two limits combined with (3.3) imply ~ w 2 ( / ~ u n1unl2) +
+ 0 which contradicts 1 > 0. Thus, vanishing does not occurs.
n

Suppose (2) dichotomy occurs. Let Q, be the concentration function


of p, defined by
P

Q , is a sequence of nondecreasjng nonnegative uniformly bounded func-


tion on R+. By extracting subsequence, we can assume that there
exists Q ( t ) such that Q,(t) 4 Q ( t ) and Q ( t ) = cu E (0'1)
(dichotomy occurs). For any E > 0, E < &, we can choose t o > 0
such that Q ( t ) 2 cr - $ if t 2 t o . Then for n large enough (n 2 no),
+
cr - f 5 Q n ( t ) 5 cu 6. Furthermore there exists y, E R2 such that

and n large enough. We can find t , -t +oo such that


n
NONTRIVIAL SOLUTION OF SEMILINEAR E L L I P T I C EQUATION 421

Let (,y E C,00(R2)be cut-off functions such that 0 5 (, y _< 1. < 0


if 1x1 2 3 ; ( ZE 1 if 1x1 -< 2 ; p = 1 if 1x1 2 2, y =
0 if 1x1 5 1. Denote
[, = [(y), tl > 0 is a positive constant to be determined later on
and satisfies

for n large enough.

Denote p, = (F),
U; = E n t i n , U: = v n u n . By computation we
deduce

Choosing tl large enough we have

> 3tl
Downloaded by [Michigan State University] at 19:14 11 January 2015

For n large enough so that t , we have

where 7, E C?(W2) such that 0 5 7, 5 1, 7, = - 0 if )x - y,J < t 1 or


+
12 - yn( 2 2 t n IX
2 ; q n ZE 1 if 2t1 < - ynl < 2 t n , IVg,] 5 1 for a: E R2.
Since
2 J
t,< ( ~ - - ~ ~ l < 2 t ~ + 2
+
(IVU,I~ /un12)dl
1
5 2 & < -
4xe '
using Lemma 2.2 we obtain

lun12 dx 5 CE .
C / t1<lz-yn l<2tn+2

Similarly we have
Downloaded by [Michigan State University] at 19:14 11 January 2015

(3.7),(3.8) combined with (3.11) deduce

and

It follows from (fl) and (f4) that

Applying the same method as in the proof of (3.10) we obtain


NONTRIVIAL SOLUTION OF SEMILINEAR ELLIPTIC EQUATION 423

From (3.13),(3.14) and (3.15) we have

(3.16) I(u,) L I(uA) + I ( u i ) - CE .


If {y,) is bounded, then
Downloaded by [Michigan State University] at 19:14 11 January 2015

When 1x1 large enough, for (uZ,I < we have

where E(R) --, 0 as R -, +m.


Thus

1
( F ( x ,ti:) - F(u:)) dx < ~(t.) mes ( 6 < lull < -)
6
wu:ll+
CAO

By (fl) and (f4),

h"
F(+:t)
= O , lim
w1 = 0.
,*t2
'-+Soo e'+m -1 t-+m 8*t2
e'+m - 1

Hence

F ( Z ,u i ) da:
lX:l>1

a=lun12
(e l+m -1)dx (byLemma2.1 and (3.4))
Downloaded by [Michigan State University] at 19:14 11 January 2015

Thus

therefore

where U(E)denotes the quantities that tend to 0 as E -, 0+, o(1) denotes


the quantities which tend to 0 as n -+ +co.

Next, we want to show that I ( u 2 ) > Cm - C L ( -


~ )o(1). To this end,
some careful estimates are needed.
NONTRIVIAL SOLUTION OF SEMILINEAR ELLIPTIC EQUATION 42 5

Arguing as the proof of (3.21) we can show that

Let v n ( x ) = u : ( a x ) , then
Downloaded by [Michigan State University] at 19:14 11 January 2015

In order to obtain on close to 1 such that

We first prove that IP2


JVuil2 dx has a positive lower bound A. which
is independent of E small enough and is independent of n.

Indeed, if there were 6, + 0 such that


m

n--tw
Js2 IVu:(6rn)12 dx = g 6 m ) , -
~(6,)+
m
0 as 6, +0,
m

where u:(6,) is a subsequence selected by above process for each 6,.


We denote ui(6,) simply by u:. By dichotomy we have
CAO

(for 6, small enough).


Thus
Downloaded by [Michigan State University] at 19:14 11 January 2015

+
Hence, choosing an = 1 cn(JR2IVu;I2 dz)-' we have v, E M a ,
which implies that M a is non empty if dichotomy occurs. It is easy to
see that.
+
on = 1 P ( E ) -I- o(1) .

Since (3.13), (3.14), we can assum IVu: l2 < if E is small


enough, and therefore Im(u;) is bounded. Hence
rm(u2.) 2 I="(v,) - ~ ( € 1 -o(1) 2 "&I a J ( u )- p(&)- o(1)
= C a - ,u(e) - o(1) .
Thus
(3.25) ) o(1) .
I " ( u ~ , ) 2 Coo- p ( ~ -
NONTRIVIAL SOLUTION OF SEMILINEAR ELLIPTIC EQUATION 42 7

Setting p = un<2in (3.2) we have


(3.26)
r

It follows from (3.9),(3.10) and (3.26) that


Downloaded by [Michigan State University] at 19:14 11 January 2015

Letting n -+ m, then E -, 0 we have c 2 Cm + (4 - qo,a contra-


diction.

If {y.} is unbounded, we change the role of {u:) and {u:). Then


we can still get a contradiction as above. Thus, we have ruled out
dichotomy. Therefore only compactness occurs. So, there exists {y.) c
R* such that for any E > 0,there is t = t ( s ) > 0 such that
We claim that {y,) is bounded. Otherwise, without loss of generality,
suppose (y, 1 -+ cm,then
n

Let 77, be a cut-off function such that 0 5 7, I 1 for x E R2, qn 0


for I x - y , l L t , q , r l f o r I x - y n l > t + l . IVq,lL2. F o r € < G,1

Thus
Downloaded by [Michigan State University] at 19:14 11 January 2015

lunl2 dx I P(E).

As in the proof of (3.21) we have


-
(3.31)

Thus
X ~-~,l<t+l
( F ( x , un) - F(un)) dx I o(l) .

(3.32) %
(I), 2 Irn(&) - p(&)- 4 1 ) .
Similar argument of ruling out dichotomy enable us t o have a, such
+
that a, = 1 - P(E) 0(1), vn(x) = u,(anx) E MOD and
+
I ( u n ) L IOD(v,) - P(E) o(1) >
C" - p(e) O(1) . +
In this case, we also get M" # 0. Taking n -+ m, then E + 0 we obtain
>
c C M , a contradiction which implies {y,) is bounded. Therefore, for
NONTRIVIAL SOLUTION OF SEMILINEAR ELLIPTIC EQUATION 429

any E > 0 , there exists t = t ( ~>) 0 such that

P(s> = 0
lim - , lim -
P(s> = 0 .
s-O Q(s) IsI-- Q(s)
r

1. Q ( u n )dx _< C
By Strauss lemma in H. Berestycki and
for some constant C >0 .
P.L.Lions [Z],for any bounded
Bore1 set R, we have

As in the proof of (3.30) we have

(3.35)
Downloaded by [Michigan State University] at 19:14 11 January 2015

Thus

- --
On the other hand, taking y = u in (3.2) and letting n m, we
obtain

~ --
Therefore from (3.3), (3.36) and (3.37) l l ~ ~ 2l l IIuIl, and thus un + u
n
strongly in H 1( R ~ )Thus
. we have completed the proof of Lemma 3.1.

THEOREM 3.2. Suppose f(x, t ) satisfies (fl)-(f4) and there exists uo E


H ' ( R ~ such
) that

Then (1.1) has a nontrivial solution.


PROOF:It is easy to see from ( f l ) , ( f 2 and
) Lemma 2.2 that there exist
p > 0 , cr > 0 such that

I ( u ) > cr for all u satisfying ((ull= p but 1luoll > p .

As in W.Y. Ding and W.M. Ni [4], we can construct a path vo


connecting 0 and tOuOsuch that max,~,, I ( u ) = I ( u o ) , where t o > 0
is so large that I ( t o u o )< 0 , lltouoll > p. Let

(3.38) c = inf s u p I ( u )
u E u~E v

where r is the set of all paths which connect 0 to tOuOcontinuously.


By Mountain Pass lemma in H. Br6zis and L. Nirenberg [3], there is a
sequence {u,} c H1(W2)such that
Downloaded by [Michigan State University] at 19:14 11 January 2015

Applying Lemma 3.1, we can prove our conclusion.

If f ( x , t ) E T ( t ) does not depend on x , then we have

THEOREM 3.2: Suppose f (z, t ) r f ( t ) does not depend on x and (fl)-


(£3) hold. Moreover there exists uo E H ' ( R 2 ) such that

I(u0) < -1 - 9 .
2
Then (1.1) has a nontrivial solution 'ii such that I @ ) 5 I(u0).

PROOF:The same as in the proof of Theorem 3.2, we can get a sequence


{ u n ) C H1(W2)such that

where c < - 6.
NONTRIVIAL SOLUTION OF SEMILINEAR ELLIPTIC EQUATION 431

In the proof of Lemma 3.1, we have used c < Ca to rule out lyn1 --, co
n
in the case that compactness occurs only. In the present case, we may
+ +
replace un by un(x y,) and prove that un(x y,) converges strongly
to some u E H 1(W2). The same method in the proof of Theorem 3.2 can
deduce our conclusion.

THEOREM 3.3. Suppose f ( x , t) = J(t) does not depend on x and satis-


fies (f1)-(B), and

where
Sp = inf ll4l ,
-
U E H ~ ( R ~lulp
)

Then (1.1) has a nontrivial s o h tion.


Downloaded by [Michigan State University] at 19:14 11 January 2015

for any E > 0, there exists u, E H1(R2)


PROOF: By the definition of Sp>
such that

Let v, = - w u . , l/u,/12= 1 - 28,

Choosing E small enough so that


CAO

then we have

Thus, from Lemma 2.4, I. can be achieved and I. < 1 - 28. Assume
u o achieves I. then
Downloaded by [Michigan State University] at 19:14 11 January 2015

for some Lagrange multiplier X E W. By (f3) and the fact that I. > 0
we get X > 0. Let u = u O ( f i x ) , then u satisfies (1.1). By Pohozaev
identity (see P.L. Lions [7])

So, we have proved Theorem 3.3.

We now prove Theorem 1.1.

PROOFOF THEOREM1.1:
By the assumptions of Theorem 1.1 we know that f(t) satisfies the
conditions of Theorem 3.3. Thus there exists E E H ' ( R 2 ) satisfying
NONTRIVIAL SOLUTION OF SEMILINEAR ELLIPTIC EQUATION

moreover I m ( E ) < 4 - 0. Let

By (f2) and Jw2(lVz12+z2)dx = Soan


? ( ~ ) z d xwe have

h l ( t ) >_ 0 for 0 5 t < 1 , h l ( l )= 0


hl(t) 5 0 for t > 1 .

Hence
) max Im(E) .
I w ( ~=
t>o

It is easy to see that there exists to E ( 0 ,+m) such that

I(to.E) = rnax I ( t E ) .
t20
By (f5) we have

(3.41) I ( t o n ) < Ico(ton) < max Ioo(tn)= Im(iz)


Downloaded by [Michigan State University] at 19:14 11 January 2015

t2o

Choosing t > t o sufficiently large that I(t?l) < 0 for t tl . Let > r be
the set of all continuous paths connecting 0 and t l Z . Define

(3.42) c = inf sup I ( u )


vEr UEV

Then, as in the proof of Theorem 3.2, we have

(3.43) c > 0.

On the other hand,

c max I ( t E ) 5 ma I ( t E )
IO<t_<tl 120

By Mountain Pass lemma in H. BrCzis and L. Nirenberg [3],


there exists
a sequence {u,) c H'(W2) such that
Since for c defined in (3.42), I(u) satisfies (PS), condition, we can ex-
tract a subsequence of {un) converging to some u E H1(BB2)strongly.
Thus I ( u ) = c, I1(u)= 0 which implies u $ 0 and u is a nontrivial
solution of (1.1).

By slightly improved arguments (which are standard), we can prove


Theorem 1.2.

ACKNOWLEDGMENTS
The author would like to thank the referee for some helpful comments.
This paper was revised while the author was visiting CEREMADE,
University of Paris-Dauphine. He would like to express his gratitude
for the hospitality he received during the stay.
Downloaded by [Michigan State University] at 19:14 11 January 2015

[I] Th. Aubin, Nonlinear analysis on manifolds : Monge-AmpZre


equations, Springer-Verlag, 1982.

[2] H. Berestycki and P.L. Lions, Nonlinear scalar field equations, I.


Existence ofground state? Arch. Rat. Mech. Anal. 82 (1983), 313-346.
[3] H. Brezis and L. Nirenberg, Positive solution of nonlinear
elliptic equations involving critical SoboIev exponent, Comm. Pure
Appl. Math. 36 (1983), 437-477.

141 W.Y. Ding and W.M. Ni, On the existence of positive entire
solution of a semihnear elliptic equations, Arch. Rat. Mech. Anal.,
-
91 (1986), 283-308.

(51 G .B. Li and X.P. Zhu, Nontrivial solution of nonlinear scalar field
equation with strong nonlinearity, Acta Math. Scientia 3 (1988), 431-
448.

[6] P.L. Lions, The concentration-compactness principle in the calcu-


lus of variations. The locally case, Part I, Ann. I.H.P.Anal. Nonlin. 1
(1984), 109-145.
NONTRIVIAL SOLUTION OF SEMILINEAR ELLIPTIC EQUATION 435

[7]P.L. Lions, The concentration-compactness principle in the calcu-


lus of variations. The locally case, Part 11, Ann. I.H.P. And. Nonlin. 1
(1984), 223-283.

[8] J.A. Moser, A sharp form of an inequality by N. Ti-udinger,


Indiana Univ. Math. J. 20 (1971), 1077-1092.

[9]S.J.Pohozaev, Eigenfunctions of the equation Au + X f ( u ) = 0,


Dokl. Akad. Nauk. SSSR. 165 (1965), 1408-1411.

[lo] M.C. Shaw, Eigenfunctions of the nonlinear equation Au+


v f ( z , u )= 0 inWZ,Pacific J. Math., 129 (1987), 349-356.
[ll]G . Talenti, Best constants in Sobolev inequality, Ann. di Math.,
-
110 (1976), 333-372.

[12]N.S. Trudinger, On imbedding into Orlicz spaces and some


applications, J. Math. Mech. 11(19671, 473-484.
Downloaded by [Michigan State University] at 19:14 11 January 2015

Received December 1990


Revised December 1991

You might also like