Prescribed Scalar Curvature problem on

Complete manifolds
David Holcman
Paris VI University
June 1998
Abstract

Conditions on the geometric structure of a complete Riemannian manifold are given to solve the prescribed scalar curvature problem in the
positive case. The conformal metric obtained is complete. A minimizing
sequence is constructed which converges strongly to a solution. In a second part, the prescribed scalar curvature problem of zero value is solved
which is equivalent to nd a solution to a linear PDE on a complete manifold. This solution helps to obtain a general result on the prescribed
scalar curvature problem, in the positive case [11].

This paper deals with the study of the prescribed curvature problem on noncompact complete manifolds. On compact manifolds it has been solved under
conditions on the function f . The main diculties in this case is the concentration phenomenon [19]. On a complete manifold having positive Yamabe
invariant, another phenomena can appear. Conditions must be prescribed on
the function f in order to avoid it.
Some results concerning the existence of conformal metrics of negative scalar
curvature manifolds were obtained by Aviles-MacOwen [5]. They proved that
it is possible to solve the Yamabe equation with a constant value equal to -1
for the prescribed curvature . In the positive case, Kim [14] solved the Yamabe
problem on a complete manifold assuming that the Yamabe quotient is less than
a certain number.
(Vn ; g) will denote a complete Riemannian manifold of dimension greater or
equal to 3 and R(g), the scalar curvature. We use the notation N = n2n2 , R = cRn
where cn = 4(nn 21) . We shall assume that the conformal Yamabe invariant is
positive :

2  2
VnR(ru) + Ru > 0
 = H inf
1 (Vn ) ( Vn uN )2=N

(1)

There are many instances of such manifolds : for example asymptotically

at manifolds with positive scalar curvature. We can construct other examples
1

starting with a compact manifold (Vn ; g). Let P be a point of Vn and GP be

the Green function of the conformal Laplacian with pole at P . Then (Vn
P; G4P=(n 2) g) is a complete manifold with the same Yamabe invariant than
(Vn ; g) [1].
The purpose of this paper is to nd a solution u of the following problem
 = f j u j nn+22 on Vn
(2)

g u + Ru
u
>0
where f is a strictly positive C 1 function such that the conformal metric g0 =
u4=(n 2)g is complete.
One important conclusion of this paper is that the prescribed scalar curvature problem can be solved on conformally at manifolds under conditions on
the geometry (see section 4 for the denition of bounded geometry) and local
conditions at a maximum point P of f plus some assumptions on the decay of f.
The mass of the manifold does not enter into the game as it does in [16]. Weyl
denotes in the following the Weyl conformal curvature tensor.
The main results of this paper are :
Theorem 1 Under the following assumptions
1. At a point P where f is maximal, Weyl(P ) 6= 0 ,
g f (P ) =
2g f (P ) = 0
and n > 6
2. Ricci > c, where c  0 is a constant and the injectivity radius is strictly
positive.
3. there exists a constant C such that
Z n=2 Z
Z 2=N
j rf j Cf and j
f j Cf
R  C f < +1 and
f < +1 (3)
Vn

Vn

Vn

and f is bounded.
4. The scalar curvature is bounded and nonnegative.
there exists a positive C 1 solution of equation (2).

Remarks

1. The rst assumption prevents the concentration phenomena locally. The

others, the concentration phenomenon "at innity".
2. More specically assumption 2 insures that the relevant Sobolev inequality
is valid : for any  > 0, there exists some constant A such that any
function u 2 H 1 (Vn ) satises

Vn

uN )2=N  (K 2 + )

Vn

j ru j2 + A

Vn

u2

where K is a constant. We choose the best constant possible [1].

2

3. We could use also the exact inequality

Vn

uN )2=N

 K2

Vn

j ru j2 + A

Z
Vn

u2

but this would require additional assumptions on the geometry of (Vn ; g).
4. Sobolev embbeding theorem holds in the following case
 The Ricci curvature is bounded from below and
 The injectivity radius is positive and the metric g is complete [17].
5. Examples of functions f satisfying assumption 3 are given below.
The second theorem is the following
Theorem 2 (Vn ; g) denotes a complete Riemannian manifold which is locally
conformally at. Then there exists a positive solution of (2) under the following
assumptions :
1. rf (P ) = ::: = rn 2 f (P ) = 0 at a maximum point of f and rf and
f
are bounded.
2. The Sobolev Inequality holds on (Vn ; g) and the scalar curvature R is nonnegative.
3. there exists aR nonconstant bounded solution h of the equation
h+R(g)h =
0 such that Vn fhN dVg is nite. Condition 3 on the decay at innity of
the function f of theorem 1 is valid.
Moreover when h converges at innity ( such is the case for asymptotically
at manifolds), the metric obtained in theorem 2 g0 = u4=(n 2) g is complete.
The conclusion holds also for theorem 1.
The basic ideas of the proof are the following. We solve Equation (2) by
a variational method. Using results from the compact case, we construct a
minimizing sequence [12]. To prove that this sequence converges we have to
show that the concentration phenomenon does not happen. This results from
the fact that f decays at innity. Indeed, since the solution has nite energy,
the number of possible concentration points is nite. A compact K is chosen,
avoiding these points. Then we prove that no concentration occurs, outside K
because of the decay condition on f, and inside K because of condition 1 in
theorem 1 or theorem 2 respectively.
This paper is organized as follows : in the rst part a sequence of minimizing solutions is constructed. In the second part the concentration at innity
phenomenon is dened and conditions are given to prevent it. In the third part,
this conditions are used to prove that concentration points and concentration
at innity do not occur. In the fourth part we use a variational method to solve
a linear equation (100). To nish estimates on the Green function are used to
obtain the behavior of the solution for the linear equation at innity.
3

1 Minimizing sequences approximation

In this part, a variational method produces a minimizing sequence. Let
k be
an exhaustion
of Vn by compact manifolds with smooth boundary such that
o

k 
k+1 . The variational problem is
Z
 2)
 k = Ainf I (u) = Ainf
(j ru j2 + Ru
(4)
k
k
k

where A k = fu 2 H1 (
k ) j f j u jN = k and u = 1 on @
k g. A minimizer of
I is called a minimizing Rsolution.
When the condition k fhNk < k is satised, the subset A k is not empty.
We suppose that this is the case in the following.

1.0.1 Construction of the minimizing sequence uk

Proposition 1 On each
k there exists a positive solution uk for (4), which
gives the Euler-Lagrange equation
 k = k f (uk + hk )N 1

g uk + Ru
uk = 0

on
k
on @
k

(5)
(6)

The Lagrange multiplier k is positive and hk is the positive solution of the

linear equation

 k = 0 on
k

g hk + Rh
hk = 1 on @
k

(7)
(8)

Proof

The existence and uniqueness of hk follow from classical linear theory, [9]. Since
the scalar curvature is bounded at innity, and the maximum of hk is reached
on the boundary of
k ,
g hk is uniformly bounded (sup
k hk = 1 ). Hence the
sequence hk is bounded in C 1; topology and it converges, uniformly on each
compact subset, to a solution of
 =0

g h + Rh
(9)

h is bounded by 1 on Vn by the maximum principle

It has been proved in [12] and [13], that there exists a minimizing solution
of (5) such that

f (uk + hk )N = k

k
where k is a number large enough, given below.
4

(10)

Moreover the sequence uk satises uk + hk > 0, since the variational problem

remains invariant substituting j u j for u in the expression :

 2)
(j ru j2 + Ru
(11)

k
R
with A k = fu 2 H1 (
k ) j f j u jN = k and u = 1 on @
k g. Using the
maximum principle for equation (5), uk > 0 since R > 0.
Recall the two conditions obtained in [12] or [13] : on each compact
k :

 k = Ainf I (u) = Ainf

k

K 2 ( k )(sup f )2=N k

k1 2=N
f juk +hk jN 2(uk +hk )hk

<1

(12)

and
k

K2
2=N ( k + C1 fhNk +
N 1 hk ( k )(sup f )

k
f
(
u
+
h
)
k
k

k
C2
k fuNk 1hk )1 2=N < 1

(13)

C1 and C2 are two constants and K is the best constant in the Sobolev
embedding.
When they are satised, the solution uk is minimizing. Later on we prove
that these conditions are satised. Moreover uk is a C 1 (
 k ) sequence by the
standard Trudinger argument.
The energy of the sequence uk is dened by :
R (j ruk j2 + Ru
 2)
E (uk ) = Vn (R fuN )2=N k
(14)
k
V
n

This energy is nite dues to the nonconcentration condititions (12) (13). When
the strong convergence will be proved, the energy of the limit u will be also
nite : it is well known that the number of concentration points in this case is
nite [19]. This will be important in the next section.

1.0.2 Convergence of uk

In this part, the convergence of the sequence uk is established. The sequence

k of energy is supposed to be equal to a large number.

Proposition 2 uk converges weakly on every compact K to a solution of

 = f (u + h)N 1

g u + Ru

on K
0

(15)
(16)

Proof
Suppose in this section that the Lagrange multiplier k is bounded in the equation (5) (this will be demonstrated in section 3.3)
 k = k f (uk + hk )N 1

g uk + Ru
(17)
then

 2k ) = k (
(j ruk j2 + Ru

f (uk + hk )N 1 hk ) < k
(18)

k
Hence on every compact subset K, as k goes to innity, there exists a k0 ,
such that for all k  k0 , K 
k and

 2k )  C
(j ruk j2 + Ru

(19)

where C is a constant. Finally uk is bounded in H 1 (K ). To prove that the

convergence is strong, the Lagrange multiplier sequence needs to be studied. We
will then prove the fact that uk is bounded in LN 0 (K ), where N 0 is greater than
N, the Sobolev critical exponent. Then by standard elliptic theory, the sequence
uk is bounded in C 2; (K ). So by Arzela-Ascoli theorem, the convergence is
uniform on each compact subset.

2 Strong convergence
In this section, conditions to obtain the strong convergence are established, and
also estimates on the Lagrange multiplier so that it converges. The following
notation is adopted k = uk + hk . This is a function of compact support in
k
Suppose
that K'0 is a given compact and N 0 > N , the critical Sobolev expoR
nent, if Vn K 0 fNk dVg is bounded, then
R for any  > 0, there exists a compact
K such that K 0  K and such that Vn K fNk dVg  , due to the fact that
f 2 L1 and Holder inequality. In a naive point of view everything happens as if
the phenomena at innity disappears.R
The study of the convergence for Vn fNk dVg is split into two parts :

Vn

fNk dVg =

fNk dVg +

Vn K

fNk dVg

(20)

K is a compact subset.
The convergence of the two integrals are proved separately :

Z
Vn K

fNk dVg !

K
fNk dVg

Vn K

fN dVg

(21)

fN dVg

(22)

2.0.3 Nonconcentration at innity condition

The convergence of the rst integral in (21) is a consequence of the nonconcentration condition locally. To study the second integral, rst we recall the
denition :

Denition 1 The sequence k (dened in the last section) is concentrated at

innity if there exists a real  > 0 such that for every R > 0
lim

k!+1 Vn BP (R)

fNk > 

where P is a given point.

The standard denition of the concentration is

Denition 2 Q is called a concentration point for the sequence p if

9 > O such that 8 > O , klim
!1

BQ ()

fNk > :

(23)

If the concentration phenomena at innity does not occur, the second integral
in (21) could be taken as small as possible by choosing a large compact set.
To prove that no concentration at innity occurs, equation (5) is multiplied
by a test function 2 f 2=N pk , where p > 1 is xed.  is a C 1 function equal to 1
in Vn 2K , vanishing identically on K, where K  2K . 2K denotes a compact
subset.
It is standard [4], that equation (5) is transformed
4p R
1=N (p+1)=2 ) j2 2(p 1)2 R
1=N
1=N p+1
(p+1)2 Vn K j r(f k
(p+1) Vn K f
(f )k +
R R 2f 2=N p+1 2 R j rf 1=N j2p+1
k
Vn K
Rk p2+1f 2=NVn+1KN +p 1
= p
(24)
k
Vn K

Using Holder's inequality the last term becomes :

Vn K

2 f 2=N +1 Nk +p 1  (

and

Vn K

fNk )1 2=N ( sup f )2=N (

Vn K

p = R p (N ) 1
Vn fp up

Z
Vn K

f(kp+1)N=2 )2=N (25)

(26)

We use the conditions at innity on f to bound the other terms ( the existence
of functions f which satisfy the decay conditions is given at the end of this paper).
7

Suppose the following decay at innity,

j rf j Cf
j
f j Cf

(27)
(28)

where C is a constant. A computation gives

f 1=N
f 1=N = NN 2 1 f 2=N 2 j rf j2 + f 2=N 1
f
Hence
j f 1=N
f 1=N j Cf 2=N

(29)
(30)

Moreover,
j f 1=N
(f 1=N ) j=j 2 f 1=N
f 1=N 2f 2=N 1 rrf + f 2=N 
 j Cf 2=N (31)
The computation of the second term in the left hand-side of (24) gives

Vn K

f 1=N
(f 1=N )p+1 j C

k

Vn K

f 2=N pk+1

(32)

By denition of the sequence k = uk + hk , the last expression becomes

Vn K

f 2=N pk+1  C (

Vn K

f 2=N hpk+1 +

Vn K

f 2=N upk+1 )

(33)

By Holder inequality, the rst term in the right hand-side is estimated as follow

Vn K

f 2=N hpk+1  (

Vn K

f 2=N )(p+1)=N (

Vn K

f 2=N hNk )1 (p+1)=N

(34)

R
Since Vn f 2=N exists, and hk is a bounded sequence, the two terms in the last
inequality can be arbitrary small.
The second term in (33) is evaluated rst by using Holder inequality:
Z

f 2=N upk+1  (

f 2=N uNk )(p+1)=N (

f 2=N )1 (p+1)=N (35)

Vn K
Since the Yamabe invariant  is positive, ( Vn uNk )2=N  1 ( Vn j ruk j2 +
 2k ). By construction of the sequence uk , which has a compact support in
k ,
Ru
Vn K

we have

Vn

Vn K

 2k = k
j ruk j2 + Ru

f (uk + hk )N 1 hk  k

(36)

Hence ,

Vn

uNk )2=N  1 k

(37)

but by assumptions the Lagrange multiplier satises k  K 2 (sup f )21=N 1 2=N

Finally,
Z
( uNk )2=N  1 2 1 2=N 2=N
(38)
K (sup f )
Vn
which is bounded.
Using these estimates, we obtain

Vn

f 2=N uNk  sup f 2=N

Vn

1
1
2=N
uNk  N=
2 K N (sup f ) sup f

(39)

R
This is term is bounded independently of the sequence uk . Finally, Vn K f 2=N upk+1
is bounded.
For the same reasons, using the inequality f 2=N 2 j rf j2  Cf 2=N , we deduce that
Z

j rf 1=N j2 pk+1  C

f 2=N p+1 + C 0

(f 1=N 1 ))2 j rf j2 pk+1 )  C (40)

2K K
Vn K
Using the conditions atR innity on the decay
R of f , we obtain that the last
term of (24) is bounded Vn K Rn=2  C Vn K f  . Indeed by Holder
inequality
Vn K

Vn K

R2 f 2=N pk+1  C (

Vn K

Rn=2 )2=n (

Vn K

f

(p+1)N 2=N
2 )

Finally (24) gives the following estimate :

j r(f 1=N pk+1=2 ) j  Cte +

Vn K
Z

k
1
2
=N
2
=N
(R

( sup f ) + )(
f(kp+1)N=2 )2=N (41)
N
1
Vn K
Vn K
Vn fk uk
According to Aubin [1] [2] [3] for any  > 0 there exists a constant A and
K such that for all functions in H 1 ,
(

Vn

uN )2=N  (K 2 + )

Vn

j ru j2 + A

Vn

u2

(42)

Hence if

k
1 2=N ( sup f )2=N < 1
N 1 uk
f
Vn K
Vn k

(43)

R
2
is satised, then Vn K j r(f 1=N pk+1=2 ) j is bounded and also
Z

for N 0 > N is bounded.

The nal condition is :

Vn K

fNp 0

( )
1 2=N ( sup f )2=N < 1
Vn f (u + h)N 1 h
Vn K
where u is the weak limit of (5) and ( ) denotes
( ) = inf
I (u) = inf
A
A

Vn

 2)
(j ru j2 + Ru

(44)

(45)

(46)

with A = fu 2 H1 (Vn ) Vn f j u jN = g.

2.0.4 Condition for nonconcentration phenomena on a compact set

In a compact set, conditions for nonconcentration phenomena are well known

[13]. The nonconcentration conditions for the solution u in a compact set is the
same as the conditions for each element of the converging sequence uk recalled
in the relation (12) and is adapted here :

( )
1 2=N (sup f )2=N < 1
N 1h
f
(
u
+
h
)
K
K

(47)

Where is a parameter to be selected. This identity is obtained in [12] by

using test functions.
K denotes an arbitrary compact. It is chosen, such that @K avoids possible
concentration points. This is possible since the sequence uk is of nite energy,
so only a nite number of concentration points can appear. The fact that the
energy is bounded is proved in the next section using geometric conditions.

3 Geometric conditions for nonconcentration at

innity

To construct the minimizing sequence uk and to obtain a minimum solution u

on Vn , an innite number of conditions are given associate to the sequence uk
of the last section. Another condition was found on the weak limit u.
10

For each k, (13) and (12) need to be satised plus the two following conditions
: If C denotes a compact subset of Vn

( )K 2
1 2=N ( sup f )2=N < 1
N 1
Vn C
Vn f (u + h) h

(48)

( )K 2
1 2=N (sup f )2=N < 1
N 1h
f
(
u
+
h
)
C
Vn

(49)

and

We prove that the conditions insuring the absence of concentration phenomena locally and at innity are satised by using test functions and the assumptions of the theorem 1 and 2. Moreover, the sequence k could be chosen equal
to a constant , large enough. Two cases appear :

3.1 The nonlocally conformally at case

In the case the manifold is not locally conformally at, we assume that there
exists a maximum point P of f where the Weyl tensor does not vanish. Denote
by r the geodesic distance to P.
Consider the following function

u(r) = (2 + r12 )n=2 1 (2 + 12 )n=2 1 in BP ()

= 0 in Vn BP (r)

(50)
(51)

 is xed.

We will see that the conditions, for the sequence (uk ) to be a minimizing on
each
k , is satised independently of
k when is a suciently large constant.
First let us choose

 =
and

(  ) =
by

Vn

Z
Vn

f (u + h)N

(52)

 2
j ru j2 + Ru

(53)

We introduce the following quantities : Q, P Qk and Pk . They are dened
2
f )2=N 1 2=N ( sup f )2=N
Q = ( R )Kf ((sup
u + h)N 1 h 


Vn K

Vn

11

(54)

2
f )2=N 1 2=N (sup f )2=N
P = ( R )Kf ((sup
u + h)N 1 h 

(55)

2
f )2=N 1 2=N (sup f )2=N
Qk = (R k )fK(u(sup

k
k
k
k + hk )N 1 hk k

(56)

Pk = k

Vn

K2
( k )(sup f )2=N ( k + C1
k fhNk +
f
(
u
k +hk )N 1 hk

k
C2
k fuNk 1hk )1 2=N

(57)

where C1 and C2 are two constants.

It remains to show that this quantities are strictly less than 1 independently
of k. Using the computations of [12]. Later on on this paper we should recall
and adapt the relevant parts of [12] for the convenience of the reader and we
shall point out where the boundedness assumptions on f are used.
The expansions are :

 =

Vn

f j u + h jN =

Z
0

rn 1 dr

S(r)

f j u + h(x) jN d
+

Vn BP

fhN (58)

where d
is the volume of the unit sphere S n 1 . Obviously, the second term
in this expression is independent of . The term k = , dened in the previous
part is also developed. By denition,

= k =

f j u + hk jN =

Z

rn 1 dr

f j u + hk

(x) jN d
+

fhNk (59)

k
0
S (r)

k BP
The second term in the last expression is always boundedR by assumption,
since the sequence hk converges to a bounded solution h and Vn fhN < 1. 
and k have the same expansion in the variable , . Let us compute only the
rst : The change of variable x = u gives

 =

Z = n n 1 Z
0

u

du

S (r )

p
f (u) g(u) 2n j v + n 2 h(u) jN d
+ O(1) (60)

with

v(r) = ( (r2 +1)1(n

v (P ) =

2)=2

n

(2 +2 )(n

developing the following expression

12

2)=2 )

on BP ()
on Vn BP

(61)
(62)

G(u) =

Z
S (r)

p
f (u) g(u) 2n j v + n 2hk (u) jN d

(63)

using Taylor expansion

j v + n 2 h(u) jN = j v jN +
R
n 2 01 N j v + tn 2 h(u) jN 2 (v + tn 2 h(u))h(u)dt
Then if R ( ) denotes the rest with

(64)

p R
R
R (u) =  n 2 S(r) f (u) g(u) 01 N j v + tn 2 h(u) jN 2
(v + tn 2 h(u))h(u)dtd

(65)

Z = n n 1

(66)

we obtain

R ( ) =

and deduce the following bound

Z = n 1 Z
R ( )  C 2
u du
0

S (u)

u

duR(u)

j f j j v + n 2 j sup h jjN 1 j sup h jd

+ O(1)

(67)
But h is bounded. C is a positive constant, which does not depend on  and k.
Finally,

 = ^ + R ( )

R
with ^ =

N
Vn fu .
Moreover ^ = 0( n) since :
=
^ =  n
rn 1 dr

S(r)

Z = n 1 N
f (x)vN  C n
r drv
0

Z =

but

(68)

rn 1 drvN  C
0
Similar types of inequalities gives an estimate of a bound for R :

Z = n 1
R( )   2 C
u duj v + n 2 j sup h j jN 1  C 2
0

(69)
(70)
(71)

Finally,

 = ^ (1 + O(n 2 ))
Q = S + o(4 )
13

(72)
(73)

and

S = (sup f )2=N K 2(u )^ (u) 2=N

(74)
R
+
1
p
To nish the computation S ,   is expanded ([12]) with Iq = 0 (1 +
t) p tq dt :
(u ) = (n 22)n w2n 1 Inn=2
2

n+4)
 (1 2 6Rn + 4 (T ((nn+2)(
6)(n 4)
and

4(n 1)
R R
R
4
cn ( 6n + 2nR (n 6)(n 4)(n 2) )) + o( ))

^ 2=N = (wn 1 f (P ) n (n2n 2) Inn=2 ) 2 nn (1 + 2 ( 2nf
(fP ) + 6Rn ) +

(n + 2)
n 1 R

f 2
4
4( (2 n n) (n An
4)(n 2)f (P ) + n 2 ( 6n + 2nf (P ) ) + o( )

with
and

2
A = f (P )T + (n +1 2)n (
4!f + R
12f
2
Rijkl

Ril fil )
4

2Rij Rij + R2 )
+
5
3
3
now using the assumptions on the geometry at the point P,
Q = 1 + Q4 4 + o(4 )
and

2 f
Q4 = 12n(n 14)(n 6) ( W 2 ) 4!n(ng0 4)
ij
T = (n +12)n4! ( 6
5R 2R15Rij

(75)

(76)
(77)
(78)
(79)
(80)

where g0 is a conformal metric built as in [2] to eliminate the tensors which

are not conformally invariant. Using assumption
2g0 f = 0, we are lead to
Q4 < 0 and the result
Q < 1:
(81)
To conclude this section we give the other bounded estimates : on Vn K ,
using supVn K f  supK f it follows that condition (49) is satised. Moreover
k could be chosen independently of k in the expression of (56),(57), and the
same conclusion holds : Pk < 1, Qk < 1.
To prove that P < 1 we use that the sequence uk is bounded in LN and the
main part of the denominator in this quantity is given by (). Indeed
Z
Z
f (uk + hk )N 1 hk  k1 1=N ( f (hk )N 1 )1=N
(82)

k
and the expansion in  gives
Z
k
f (uk + hk )N 1 hk = k (1 + O(n=N ))
(83)

k
14

3.2 The locally conformally at case

We assume the manifold to be complete and locally conformally at. For each
small open set, there exists a function  > 0 such that g = 4=(n 2) . In this
case, it has been proved that there exists a minimizing sequence uk such that
uk is a solution of (5) on each
k , k 2 N . Moreover we will see that if the
sequence of functions hk , dened in (7) are strictly positive then the minimal
solution uk exists, independently of the sign of the mass of
k .
To built the test function in this case, consider the Green function of the
conformal Laplacian, by denition
 k = P on
k

Gk + RG
(84)
Gk = 0 on @
k
(85)
and we have the following expansion in an Euclidean neighborhood of P:
Gk (r) = r2 n + Ak + k
(86)
where r is the geodesic distance relative to P and k is a C 1 function such
that k (P ) = 0. The sequence of the conformal Green function Gk converges
to G, uniformly on every compact.
To check the nonconcentration conditions for the limit u of the sequence uk ,
we shall use the following test function

 = (2 +r2)l(n 2)=2

0 (G + )
0 G

on BP ()
on BP (2) BP ()
on Vn BP (2)

G is the minimal Green function solution of

 = P

G + RG

(87)

(88)

To check the nonconcentration conditions depending on k, we use

p; (r) = (2 +r2)l(n 2)=2 on BP ()

0 (Gp + p ) on BP (2) BP ()
0 Gp
on
p BP (2)

(89)

where Gp is dened above (84).

In order that The test functions (87) and (89) are continuous the following
condition must be satised

l
=  (2 n + A)
2
2
( +  )(n 2)=2 0
15

(90)

for one case, l, 0 are constant numbers, such that 0 < l < (n 2)=n, is C 1
decreasing function in j r j, (x) = 1 on BP (), (x) = 0 for j x j 2 and
j r j 2 1 on  j x j 2. For the other case

l
=  (2 n + Ap )
2
2
( +  )(n 2)=2 0p

(91)

where 0p > 0

The quantities Q P Qk Pk will be evaluated as in the previous case, using
computations of [12].
We shall evaluate :

 =
and

k =

Vn

f (u + h)N and (  ) =

f (u + h)N and ( k ) =

k
the computations give [13]

Vn

 2
j ru j2 + Ru

(92)

 2
j ru j2 + Ru

(93)

Qk = Pk = 1 4hk (0)n 2 l Jn n=12 1 < 1

In

(94)

J depending on n only and

P = Q = 1 4h(0)n 2 l Jn n=12 1 < 1

In

(95)

rn(n1+2)=2 dr
where Jn = 01 (r2 +1)
In the last section, we prove that h > 0 hence it is not identically 0. Moreover
the sequence hk converges to h uniformly on every compact. To conclude this
section, remark that the mass of the manifold [15] does not appear as in other
cases. The function h plays the fundamental role. If h were identically zero,
then by [12],

Qk = Pk = 1

2 A n 2 < 1
k
nInn=2

(96)

Qk = Pk = 1

2 An 2 < 1
nInn=2

(97)

and

16

3.3 Estimation of the Lagrange multiplier

To nish this section, estimates of the Lagrange multiplier sequence k will be

obtain as a consequence of the previous study of nonconcentration. In [12] it is
proved that lim sup ( k )  ( ) when k converges to . The sequence f k g
satises the following inequality, for = k large enough, relation (18) gives
0 < k 

( k )
N 1 :

k f (uk + hk ) hk

(98)

Using the nonconcentration relation (12), one obtains the estimate 0 < k 
1
K 2 sup f 2=N 1 2=N
Finally the sequence k is bounded. Suppose that k converges to 0, then
the sequence k = uk + hk converges to 0 on each compact set since

 2k ) = k (
(j ruk j2 + Ru

f (uk + hk )N 1 hk )

(99)

But since uk converges strongly in LN andR since

k f (uk + hk )N = ,
N
N
Vn fh = . This contradict the fact that > Vn fh when is chosen large
enough.

3.4 Remarks and conclusion

R
R
If R(g) decays at innity as r1q , then the condition that Vn Rn=2  C f < 1
gives a upper bound for the decay rate q > 2.
The conditions on f at innity, stated in the two theorems, are satised, for
example, when f is decreasing like power functions
: f  r q , rf  r q 1 and
R
q
2
2
=N
rrf  r , with q > Nn=2 > 2. Indeed Vn f < +1, hence r1 n+(11 2=n)q
is in L1 (R ).
When h converges to 1 at innity (which is the case in asymptotically at
manifolds [6]) then the inequality h + u  h > 0, shows that the metric g0 =
(h + u)4=(n 2) g is complete.
In the next section, it is proved that the condition 2 of theorem 2 is ratied
under small additional assumptions on the geometry. This is done by studying
a linear PDE of the type introduced by E. Witten [20] in the case of Spinor
manifolds. Results on this linear PDE are used by Witten to prove in dimension
4 the positive mass theorem.

4 Conformal metric of zero scalar curvature on

complete manifolds
The purpose of this part is to prove the existence of a function h solution of the
problem
17

 = 0 on Vn

g h + Rh
h ! C > 0 when r ! +1

(100)

A point O 2 Vn is given, r is the Riemannian distance r = d(P; O).

h converges to C means that for every  > 0, there exists a compact K
containing O such that 8P 2 Vn K , j h(P ) C j .
The existence of such a function h is used to solve the prescribed scalar
curvature problem for a large class of complete manifolds [11], such as the
asymptotically at case or certain conformally at cases.
A certain number of assumptions are made on the geometry of the manifold.
Let us recall the denition of the Yamabe conformal invariant :
R (ru)2 + Ru
 2
VnR
(101)
 = H inf
1 (Vn ) ( Vn uN )2=N
the assumptions are :
1. The Yamabe invariant is positive :  > 0
R
2. An integral bound on the scalar curvature : Vn RdVg < +1
3. The Ricci curvature is bounded from below
4. The injectivity radius is strictly positive
When the last two conditions are satised, (Vn ; g) is said to have a bounded
geometry.
If (Vn ; g) has a bounded geometry, the Sobolev inequality is valid. The
conformal Laplacian has a minimal Green function which can be estimated.
This is a consequence of some works on the heat kernel [18] [7] [8].
The main theorem presented here is the following

Theorem 3 (Vn ; g) is a complete manifold, with Ricci curvature bounded from

below and injectivity radius strictly positive, bounded away from zero, and with
the Yamabe conformal invariant  > 0 .
R
If R denotes the scalar curvature, it is also assumed that R  0 and Vn R <
+
G the minimal Green function of the Laplacian exists and if moreover
R 1G. (Then
P;
Q
)
R(P )dVg goes to zero at innity, then there exists a strictly positive
Vn
solution of (100) which converges to a positive constant at innity.

Geometric conditions can be stated which insure that Vn G(P; Q)R(P )dVg
goes to zero at innity. For example, it will be proved in the last section, that
this is the case if the scalar curvature R converges to zero at innity or if R is
bounded in Lq with q > n=2.
The geometric interpretation of this result is the following : when the scalar
curvature is positive, then the solution h is strictly positive. This is summarized
in the proposition
18

Proposition 3 On manifolds with a bounded geometry and strictly positive

Yamabe invariant whose scalar curvature is positive and either belongs to L1 \ Lq
(q > n=2) or tend to zero at innity, there exists a conformal metric h4=(n 2) g
having zero scalar curvature and the manifold (h4=(n 2) g; Vn ) is complete.

As a consequence of the main theorem, one obtains :

Proposition 4 Let (Vn ; g) be a complete
R manifold. If in addition to the assumptions of the previous theorem, Vn j R jn=2 dVg is bounded, then the prescribed scalar curvature problem can be solved : there exists a positive solution
to the equation
n+2

 = fj u jn

g u + Ru

>0

on Vn

(102)

under the conditions on the function f described in the rst section. The metric

g0 = u4=n 2 g obtained is complete.

This proposition is proved exactly as the two theorems of the last section but
instead of choosing 1 as the boundary condition for the approximation sequence,
we use the restriction of the linear positive solution of (100). In this case, the
sequence hk converges exactly to the solution.
Previous results on the subject were obtained by Brill-Cantor [6]: they
proved the existence of a positive solution of (100) on asymptotically at manifolds with nite ends using weighted Sobolev spaces. The precise statement
is
Assume Vn has a nite number of ends, having neighborhoods N1 ; ::Nk diffeomorphic to the complement ofPa unit ball in <n . The weight Sobolev spaces
p = ff s.t.

+jj j p
are dened by : Hs;
L (Nl ) < +1 for each
p 2 jjs j D fi1 :::ij 
ends Nl g where  = 1 + r
The two following propositions are equivalent [6] :

 There exists a solution  of (100) on asymptotically at manifolds with

1 < p < 2n=n 2 , n=p <  < 2 + n(1 1=p) and s > n=p + 2. The
p . Moreover
metric g considered as a section of a vector bundle is in Hs;
p
the solution  satises  1 2 Hs; and  > 0.
 for all u =
6 0, u 2 C01 , the following inequality is satised :

Vn

j ru j2 + R (g)u2 > 0;

(103)

The dierentiability conditions on ; r; s stated above are very restrictive :

s > n=p + 2 > n=2 + 1. There
exists (Vn ; g) having bounded geometry such
p but such that
that g does not belong to Hs;
the scalar curvature R belongs to
L1 \ Lq with q > n=2.
19

5 A Variational problem
In this section, a variational problem is constructed to obtain a solution of
equation (100):
Consider
Z
Z 2
2 + Ru

 = inf
I
(
u
)
=
inf
j
r
u
j
(104)
A
A

R
where A = fu 2 H1 (Vn ) s.t.

Vn

Vn

Vn Ru = 1g:

Lemma 1 The minimum  is achieved by a positive function h which satises

:

=

Z
Vn

j rh j2 +

 =1
Rh

(105)

 2 < +1
Rh

(106)

Vn

Vn
 = R

g h + Rh
h 2 LN

(107)
(108)
The last fact implies that h is not a constant. Moreover  is strictly positive.
Proof :
R  2
R
Consider uRk a minimizing sequence satisfying :   Vn j ruk j2 + Vn Ru
k
 k =1
 + 1=k with Vn Ru
uk is bounded in LN (Vn ). On every compact set K, the sequence (uk ) is
bounded in H1 (K ) : this is a consequence of the fact that on every compact,
the rst eigenvalue of the operator
+ R is positive. This follow from the
inequality 0 <   CK 1 (
+ R ) where CK is a positive constant depending
only of K.
Choosing a subsequence if need be we canRassume that Ruk converges a.e on
 2 . Moreover
every
compact to a function u. Then  = Vn j ru j2 + Vn Ru
R Ru
 = 1. To see this all we have to do is to prove that for any  > 0 there
Vn
R  k  C where C = sup R Ru
 2
exists a compact set K such
that : Vn K Ru
Vn k .
R
In fact choose K so that Vn K R  2 then by the schwarz inequality

Vn K

 k
Ru

sZ

Vn K

 2k
Ru

Vn K

R  C

sZ

Vn K

R  C

(109)

R
the last inequality is a consequence of R 2 L1 and Vn Ru2k is bounded, since
uk is a minimizing sequence.
u is a solution of the variational problem. The Euler-Lagrange equation
implies that there exists a Lagrange multiplier 0 such that :
8 2 C 1 ;
0

 = 0
rru + Ru
20


R

(110)

 = 0 R ,  = 0 . Moreover u is not zero since RV Ru

 = 1 and u is
Hence
u + Ru
n
R
 2.
not constant since u belongs to LN .  is not zero because  = Vn j ru j2 + Ru
It remains to prove that u tends to zero at innity then it follows that the
function H =  u will be a solution of equation (100), which converges to a
positive constant  at innity.

6 Behavior of h at innity
In the previous section the existence of a solution of (100) was obtained by the
variational method. In this section estimates on the behavior at innity of the
solution are obtained by the use of the Green function.
The minimal Green function of the conformal Laplacian is the smallest of
the solutions of
 = P

g G + RG
(111)
Such a function exists [12]. P denotes the Dirac distribution localized at a given
point P. The Green function of the Laplacian is the minimal strictly positive
solution dened by

G0 = P

(112)

6.1 Existence of the Green function

There exists conditions for the existence of such a function : Varopoulos states
in [18] that if the Ricci curvature is positive then the
R existence of the Green
function is equivalent to the fact that the integral 1+1 V ol(Btt (x0 )) dt < +1
converges.
Here we will use the existence of the Green function under weaker conditions.
The Heat Kernel theory developed partially in [18] [8] [7], gives the existence
and estimates of such a function.
Suppose that the conditions of the bounded geometry are satised : the
Ricci Curvature bounded from below and the injectivity radius strictly positive.
If p(t,x,y) denotes the heat kernel solution of

p = @t p

(113)

Chavel [7] proves the estimations :

0 < p(t; x; x)  Ct n=2

(114)

 > 0; 8t > 0 and 8x; y 2 Vn

2
0 < p(t; x; y)  C t n=2 e d (x;y)=(4t(1+))

(115)
(116)

and

C and C are two constants.

21

R
Finally G(x; y) = 0+1 p(t; x; y)dt exists and
8x 6= y 2 Vn , 0 < G(x; y)  Cr n+2

(117)

where r is the Riemannian distance from x to y.

This is a consequence of a theorem of [7] which we will recall now. Using
the same notations :
is an open subset with compact closure and smooth
boundary, the isoperimetric quotient is :

A(@
)
I (
) = V (
)
1 1=n

(118)

where A denotes the n-1 dimensional Riemannian measure and V denotes the
n-dimensional Riemannian measure. The isoperimetric In constant is dened
to be the inmum of I (
) under all smooth

In = inf

I (
)

(119)

When the Sobolev inequalities are valid, which is the case in bounded geometry, Hp1  Lq with q1 = 1 p1 , the following equality holds by a theorem of
Federer-Fleming [7] :

In = Kn1 = f 2infC 1
0

R j rf j

R
>0
( j f j)1 1=n

(120)

nally to sum up the results :

Proposition 5 On bounded geometry, In > 0 and the following estimate holds:
8t > 0 and 8x; y 2 Vn 0 < p(t; x; y)  Ct n=2 e d2 (x;y)=(4t(1+))
(121)
Hence the solution G0 of
G0 = P exists and the following estimate on the
behavior of G0 is also valid : 0 < G0 (x; y)  Cr n+2

6.2 Application to the behavior of the solution

We can deduce from the results of the last paragraph and the fact that the
scalar curvature is positive, that 0 < G < G0 . Indeed since G is the minimal
Green function of the conformal Laplacian, the sequence Gi of Conformal Green
function with boundary value equal to 0 on
i (an exhaustion sequence) satises
G = limi!+1 Gi . Consider the bounded function Gi G0 , solution of
(G
 0  0 it satises the maximum principle Gi G0 < 0
G0 ) + R (G G0 ) = RG
on @
i so Gi G0 < 0. Hence 0 < G < G0 .
Using the Conformal Green function, the solution of equation (2) can be
written as:
22

u(P ) = 

Z
Vn

R(Q)G(P; Q)dVg

(122)

To prove that u goes to zero at innity under geometric conditions, take a

number  > 0. we shall prove that there exists a number R0 such that for all
P with d(P; 0)  R0 , 0  u(P ) R : To start x a constant R1 > 0 such that
K
is a compact set, K R(Q)G(P; Q)dVg   for d(P; 0)  R2 and
R = BRO (R1). Then
:
Vn K

u(P ) = 

Vn K

R(Q)G(P; Q)dVg + 

R(Q)G(P; Q)dVg

(123)

To prove that the rst integral can be bounded from above by any positive
constant arbitrary small, consider the decomposition

Vn K R(Q)G(P; Q)dVg = Vn K \fG1g R(Q)G(P; Q)dVg +

Vn K \fG1g R(Q)G(P; Q)dVg

Z
Vn K \fG1g

and

R(Q)G(P; Q)dVg 

Vn K \fG1g

R(Q)dVg

(125)

R(Q)dVg  

(126)

Vn K \fG1g

Z
Z

Z
Vn K

(124)

R(Q)G(P; Q)dVg 

R(Q)G(P; Q)dVg
Vn K \fG1g\fd(P;Q)C 1=(n 2)g
since the inequality d(P;QC)n 2  G(P; Q)  1 implies : d(P; Q)

(127)
(128)

 C 1=(n 2) At

this stage of the proof, two possibilities arise: rst if R tends to 0 at innity
then

Vn K \fG1g\fd(P;Q)C 1=(n 2) g

 sup R

R(Q)G(P; Q)dVg

(129)

G(P; Q)dVg
(130)
fd(P;Q)C 1=(n 2) g
0 2=(n 2) where C 0 is a constant. Hence
2) g G(P; Q)dVg  C C
Vn K

and fd(P;Q)C 1=(n

Vn K \fG1g

R(Q)G(P; Q)dVg  sup RC 0 C 2=(n 2)  

Vn K

23

(131)

If R tends to 0 at innity, the last expression is bounded by  for d(P; O)  R4 .

The second possibility is that
R
R

1=
Vn K \fG1g R(Q)G(P; Q)dVg  ( Vn K \fG1g R(Q) dVg ) 
R(

1=
(132)
Vn K \fG1g G(P; Q) dVg )
R
using Holder inequality.R Now Vn K \fG1g G(P; Q) dVg is bounded if  < nn 2
and since the integral Vn R(Q) dVg converges for  > n=2 by assumptions, we
can choose K being enough so that

Vn K

R(Q) dVg )1=  

(133)

Finally for P large enough (d(P; O)  Ri for i=1..3) the integral dening u(P )
goes to zero when P converges to innity under geometric conditions states
in the theorems. This shows that u(P ) tends to zero at innity. The main
theorem is solved by considering the function  u which is a solution of (100)
and converges to the positive constant .

References
[1] T. Aubin{ Some Nonlinear problems in Riemannian Geometry, 1998
Springer-Verlag.
[2] T. Aubin{ Equations dierentielles non lineaires et probleme de Yamabe
concernant la courbure scalaire, J. Math. pures et appl. 55, 1976 p269-296.
[3] T. Aubin{ Espace de Sobolev sur les varietes Riemanniennes, Bull. Sci.
math. 2, 100 1976 p 149-173
[4] T. Aubin{ Meilleures constantes dans le theoreme d'inclusion de Sobolev
et un theoreme de Fredholm non lineaire pour la transformation conforme
de la courbure scalaire J. Funct. Anal., 32, 1979 p148-174.
[5] P. Aviles{ R. McOwen Conformal deformation to constant negative
scalar curvature on noncompact Riemannian manifolds. J. Dierential Geometry, 27, 1988, 225-239
[6] M. Cantor D. Brill The Laplacian on asymptotically at manifolds,
Compositio Mathe. , 43, 1981 p 317-330.
[7] I. Chavel E. Feldman { Modied isoperimetric constants, and large
time heat diusion in Riemannian manifolds, Duke math. J. 64, 3, 1991 p
473-499.
[8] E.Davies. Gaussian Upper Bounds for Heat Kernels of Some SecondOrder Operators on Riemannian Manifolds Journ. Func. Anal. 80, 1988 p
16-32.
24

[9] D. Gilbarg { N. Trudinger { Elliptic Partial dierentiable equations

of second order; Springer-Verlag 1983, second edition.
[10] E. Hebey { M. Vaugon { The best constant problem in The Sobolev
embedding theorem for complete riemann Manifolds, Indiana, Vol.79, n1,
1995 . p235-273
[11] D.Holcman{ These de Doctorat de l'universite Paris VI, 1998.
[12] D. Holcman{ Solutions nodales sur les varietes riemanniennes compactes,
C.R. Acad. Sci. Paris t.326 SI, p 1205-1208, 1998
[13] D. Holcman{ Solutions nodales sur les varietes riemanniennes a bord,
C.R. Acad. Sci. Paris t.326 SI, p 1321-1324, 1998
[14] S.Kim The Yamabe problem and applications on noncompact complete
Riemannian manifolds and the specication of the scalar curvature Geometriae Dedicata 64: 1997 p373-381
[15] R. Schoen Variational theory for the total scalar curvature functional
for Riemannian metrics and related topics Montecatini notes, 1987 Lect.
Notes 1365, Springer Verlag, p120-155.
[16] R. Schoen Conformal deformation of a Riemannian metric to constant
scalar curvature, J. Di. Geom. 20, 1984 p 479-495.
[17] N.Varopoulos Hardy-Littlewood theory for semigroups, Journal of Functional Analysis, 1985, 63, p240-260
[18] N. Varopoulos The Poisson Kernel on Positively Curved Manifolds,
Journ. Func. Anal. 44, 3, 1981 p 359-380.
[19] M. Vaugon { Transformation de la courbure scalaire sur une variete Riemannienne compacte J. Funct. Anal., 71, No. 1, 1987 p 182-194.
[20] E. Witten { A new proof of the positive Energy theorem, Communication
in mathematical physics, 80, 1981 ,p 381-402.

25