R ENDICONTI
del
S EMINARIO M ATEMATICO
della
U NIVERSITÀ DI P ADOVA
A LBERTO G. S ETTI
Eigenvalue estimates for the weighted laplacian on a riemannian manifold
Rendiconti del Seminario Matematico della Università di Padova, tome 100 (1998), p. 27-55
<http://www.numdam.org/item?id=RSMUP_1998__100__27_0>
© Rendiconti del Seminario Matematico della Università di Padova, 1998, tous droits réservés.
L’accès aux archives de la revue « Rendiconti del Seminario Matematico della Università di Padova » (http://rendiconti.math.unipd.it/) implique l’accord avec les conditions générales d’utilisation (http://www.numdam.org/conditions).
Toute utilisation commerciale ou impression systématique est constitutive d’une infraction pénale. Toute copie ou impression de ce fichier doit conte- nir la présente mention de copyright.
Article numérisé dans le cadre du programme Numérisation de documents anciens mathématiques
http://www.numdam.org/
on a
Riemannian Manifold.
ALBERTO G.
SETTI (*)
ABSTRACT - Given a
complete
Riemannian manifold M and a smoothpositive
func-tion w on M, let L = -L1 -
V(log w)
acting onL2(M,wdV). Generalizing techniques
used in the case of theLaplacian,
we obtain upper and lower bounds for the first non-zeroeigenvalue
of L, for M compact, and for the bot- tom of the spectrum, for M non-compact.1. - Introduction.
Let M n be an
n-dimensional, complete
Riemannian manifold. TheLaplace operator
onM, - LI,
can be defined as the differentialoperator
associated to the standard Dirichlet form
where I - I
is the norm inducedby
the Riemannian innerproduct
anddV is the volume element on M n . Now let w be a
given
smoothstrictly positive
function onM n,
that will be referred to as aweight
function. Ifwe
replace
the measure dV with theweighted
measure w dV in the defi-nition of
Q,
we obtain a newquadratic
formQw ,
and we denoteby
L theelliptic
differentialoperator
on inducedby Qw .
Inthis sense L arises as a natural
generalization
of theLaplacian.
It isclearly symmetric
andpositive
and extends to apositive self-adjoint
op- (*) Indirizzo dell’A.:Dipartimento
di Matematica, Universita di Milano, via Saldini 50, 20133 Milano,Italy.
1991 Mathematics
Subject Classification.
Primary 58G25;Secondary
35P15.erator on
L 2 ( w dV). By
Stokestheorem,
Thus
introducing
aweight
factor is the firststep
towardsdecoupling
theleading
term and the lower order terms of theoperator,
which in thecase of the
Laplace operator
arecompletely
determinedby
the metric of Mn .Weighted Laplacians
arisenaturally,
forexample right-invariant (sub) Laplacians
on a Lie group with left Haar measure can be viewed asweighted Laplacians
withweight given by
the modular function. Onemight
alsohope
that thestudy
ofweighted Laplacians
will increase theunderstanding
of differentialoperators acting
on L p spaces withrespect
to measures on M n not so
closely
related to the metric structure of themanifold,
e.g. the sum of squares of left invariant vector fields on a Lie group whichsatisfy
H6rmandercondition,
andgeneralisations. Finally, operators
of this form are the finite dimensional model ofoperators
oninfinite dimensional manifolds
(«the
infinite volumelimit»)
which arise in Statistical Mechanics.Operators
of the form(1.1)
havealready
been studied in[Bkl], [BkE], [Dal], [Da3], [Dl], [D1S],
whichmainly
deal withproperties
of theheat
semigroup generated by L,
and in[Bk2]
where the Riesz transform defined in terms of L isinvestigated.
In this paper westudy
estimatesfor the L
2-spectrum of L,
aproblem
also considered in[Dl]
and in[D1S].
Pointwise estimates or the heat kernel of L have been obtained in
[Se2].
Our
goal
is to find upper and lower bounds in terms of thegeometry
of M n and of the
properties
of theweight function,
for the first non-zeroeigenvalue
ofL,
when M n iscompact,
and for the bottom of thespectrum
if M n is not
compact.
Several
approaches
arepossible:
One could consider L as aperturba-
tion of - L1 and
apply perturbative
methods. Or one could observe that under the mapL is
unitarily equivalent
to theSchr6dinger operator
H = - d ++
(w -1~2 dw 1~2 )( ~ ) acting
on and deduce estimates for Lby Schr6dinger operator techniques.
Bothapproaches
have the disadvan-tage
ofrequiring hypotheses
thatdepend separately
upon thegeometry
of M n
(typically
via thecurvature)
and on the function w(via
uniformbounds for w and its
derivatives),
withoutkeeping
into account thepossi- bility
of mutual andperhaps competing
interactions.The
approach presented
here is more direct. Theinterplay
betweenthe
geometry
of M n and the behaviour of theweight
function w ismostly
taken into account
by
means of a modified Ricci curvature definedby
Then, generalising techniques successfully applied
in the case of - L1by
Lichnerowicz
([Lz]),
Li and Yau([LiY])
and Li([Li]),
we obtain upper and lower bounds for the first non-zeroeigenvalue
ofL,
if M n is com-pact,
and for the bottom of itsL ~-spectrum, non-compact.
Thesebounds will be
expressed
in terms of bounds forRw,
of the dimension n of M n and of its diameter.The paper is
organized
as follows. In § 2 westudy
lower bounds for the first non-zeroeigenvalue ,1,
1 of L on acompact
manifold. We prove first ageneralization
of Lichnerowicz theorem(cf. [Lz],
p.135)
thatgives
a lower bound for ~,1
in
termsRw .
Then we use thetechnique
ofgradient
estimates to obtain lower bounds for A
which generalize
the bounds ob- tainedby
Li and Yau([LiY])
and Li([Li])
for theLaplacian.
We conclude the sectionby briefly indicating
how Li and Yau’stechniques
can be usedto find lower bounds for the first Dirichlet
eigenvalue
of L on a domainQ c M with smooth
boundary.
In § 3 we consider the case of a
(non-compact)
manifold with apole
and we prove a lower bound for the bottom of the
spectrum A,
of L interms of a
negative
upper bound for the radial curvature and an upper bound for which is in the samespirit
as McKean’s lower bound for the bottom of thespectrum
of theLaplacian
on anegatively
curved manifold
(cf. [McK]). Assuming
further that the radial derivative aw/ar iseverywhere nonnegative
we also show thatA,
can be boundedbelow in terms of a
negative
upper bound for the radialcomponent
ofRw .
We conclude the section with two
examples
that show that the bounds obtained aresharp.
In the last section we prove a
comparison
theorem for w-volume(Theorem 4.1)
whichgeneralizes Bishop’s comparison
theorem(cf. [Cl],
p. 71
ff.).
This allows us to extend to L a version ofCheng’s comparison
theorem
([Cg])
thatgives
an upper bound for the firstgeneralized
Dirichlet
eigenvalue
of L on ageodesic
ball in M n in terms of the first Dirichleteigenvalue
of theLaplacian
on ageodesic
ball of the sameradius in a suitable space of constant curvature.
Using Cheng’s
argu-ment,
thisgives,
in thecompact
case, upper bounds for all theeigenval-
ues of L in terms of a lower bound for
Rw
and of the diameter ofM, and,
in the
non-compact
case, an upper bound for the bottom of thespectrum
in terms of a lower bound for
2. - The
compact
case: bounds from below for the first non-zeroeigenvalue.
Let M n = M be an n-dimensional Riemannian manifold with
weight
function w, and let L denote the differential
operator
defined as in(1.1)
above.Using elliptic regularity,
and anargument
as in Strichartz([St],
seealso
Bakry [Bk2]
where the details are carriedout),
one shows that for~, 0 the
equation (Lo )* u = ~,~c
hasonly
the zero solution inL 2 (M,
andconsequently ([RSi],
pp.122-137)
L isessentially
self-adjoint
on(M)
and extends to apositive self-adjoint operator
onL2(wdV).
Since L is a
positive self-adjoint operator
itsspectrum
is contained in thepositive
real axis.Moreover,
since theeigenfunctions
are smoothby elliptic regularity,
0 is aneigenvalue
iff the w-volume ofM,
defined asthe volume with
respect
to the measurew dV,
is finite. If u is in the do-main of
L,
then and([Bk2], Prop. 1.3).
It follows that for ~,0, (A -
is a continuous map Thus if M is acompact manifold,
thecompactness
of theembedding ([Au],
Theorem2.34), implies
that(~, - L ) -1
is acompact operator
onand,
as in the case of theLaplacian,
weconclude that the
spectrum
of L ispurely
discrete and theeigenvalues
can be
arranged
in adiverging
sequenceThe
corresponding eigenfunctions {uj}
are smooth on M and form a ba-sis of
Although
we aremainly
interested in the closedproblem
forL,
insome instances we shall also consider the Dirichlet
problem
for L on anopen
relatively compact
domain withnot-necessarily
smoothboundary
In this case theoperator
L with(generalized)
Dirichletboundary
conditions on aS2 is the Friedrichs extension of the differentialoperator Lo
= - d -V(log w)
that acts as apositive symmetric operator
on
(cf. [R-Si], p.177).
If theboundary
8Q ispiecewise smooth, elliptic regularity
and thecompactness
of theembedding
where is the closure of
( S~ )
in the norm =))w))f2 imply,
asabove,
that the spec- trum of L ispurely
discrete and theeigenvalues
can bearranged
in a di-verging
sequence 0~, o ( S~ ) ~,1 ( S~ ) ... ~
+ oo.At this
point
it isperhaps
worthnoticing that,
eventhough
’ can be written as the standard Dirichlet form relative to the metric
g = w 2in - 2 g,
this does notimply
that L isunitarily equivalent
to theLaplacian - 4
inducedby g.
Indeed L acts onwhile - 3 acts on where d V =
w2/n-2dV
is the volume element inducedby g.
Therefore we cannot reduce theproblem
ofestimating
theeigenvalues
of L to the sameproblem
for theLaplacian
relative to a con-.formally changed
metric. On the other hand Prof. C. Herz([Hz]) pointed
out that if N is a
compact
1-dimensional manifold(typically
N= ,S 1),
theaction of the
Laplacian
of thewarped product
M = Mx w N
on functionsconstant on N coincides with the action of L on functions defined on M.
Indeed,
since M is theproduct
manifold M x N endowed with the metricwhere .7r N are the
projections
and gm, 9N the metricsrespectively
onM and
N,
asimple computation
in local coordinates showsthat,
forIn
particular, if fo
is aneigenfunction
of Lbelonging
to theeigenvalue A,
then
f(x, 0) = fo ( x )
is aneigenfunction of -:1 belonging
to the sameeigenvalue.
This fact can be used to obtain lower bounds for the firstnon-zero
eigenvalue
of Lby appropriately adapting
some of the results which hold for theLaplacian.
Assume now that M is
compact.
As observedabove,
thespectrum
of Lis
purely
discrete. The firsteigenvalue being always
zero, we are inter- ested in lower bounds for the first non-zeroeigenvalue
of L. These lowerbounds
willnecessarily depend
on theinterplay
between thegeometry
of M and the behavior of the
weight
function w. In many instances thisinterplay
can beeffectively
taken into accountby
means of a modifiedRicci curvature defined
by
Following Bakry ([Bk2])
we alsoput
We start with a
generalization
of the classical Bochner-Lichnerowicz- Weitzenb6ch formula for functions([BGM],
p. 131ff.).
In a moregeneral
and
slightly
differentform,
the formula appears in[BkE]
and[Bk2].
The version which wepresent
is better suited to our purposes and admits asimpler proof.
PROPOSITION 2.1. Let M be a
(not necessarily compact) manifold
with
weight function
w.If
u EC2(M)
we havePROOF. It suffices to prove the first
equality,
which follows immedi-ately by combining
the BLWformula,
and the
following
formula for the drift termwhich,
inturn,
is a consequence ofThe next theorem extends to L a classical result
by
Lichnerowicz([Lz],
p.135]).
THEOREM 2.2. Let M be a
compact,
n-dimensional Riemannianmanifold
withweight function
w, and let ~,1be
thefirst
non-zeroeigen-
value
of L.
Assume thatThen
PROOF. The
proof
followsclosely
that of Lichnerowicz: Let u be aneigenfunction belonging
toA 1.
Since L1u = trace(Hess u),
we havewhere the last
inequality
follows from(x - y)2 ; ax 2 - a( 1 - a ) -1 y 2
Va E (0, 1),
witha = n/(n + 1 ). Substituting
this into(2.3),
andusing
Lu =
À 1
u and theassumption
we findIntegrating
withrespect
tow dV,
andusing
the identitiesthat hold for all
f
in the domain ofL,
one concludes as in[LZ],
p.135,
thatand
(2.4)
follows.REMARK. If w = 1 so that L = - L1 and =
Ric,
Theorem 2.2 fails toreproduce
Lichnerowicz estimate. We are indebted to J. D. Deuschel for thefollowing
variation of theargument
above. For every 6; >0,
letand assume that
Then, by using
with a = + in
(2.5)
one obtainsand, arguing
asabove,
thisyields
Notice that:
- For E = 1 we recover Theorem
2.2;
- If w =1 and Ric > x, then for all E we can take = K
and,
let-ting E
- oo in(2.6)
we obtain Lichnerowicz’s estimate for the first non- zeroeigenvalue
of theLaplacian, A 1
~nx/(n - 1);
Assuming
that andletting
ê ~ 0 in(2.6),
we recover the estimate ~,1 ~ x proven, with a differentproof,
in[Dl-S].
It is remarkable that the last result holds even for Mnoncompact.
Indeed thehypothesis ,Sw ; x implies
first that the w-volume of M(the
volume withrespect
tothe measure is finite
([Bkl]),
so that 0 is aneigenvalue
ofL,
andthen that the gap between 0 and the rest of the
spectrum
is at least x(cf.
[DIS]).
Adapting
Li and Yau’stechniques,
we shall derive next twogradient
estimates for the
eigenfunctions
of Lbelonging
to a non-zeroeigenvalue.
Using
thesegradient
estimates we will be able togive
bounds from be- low for the first non-zeroeigenvalue
of L.THEOREM 2.3. Let M be an n-dimensional
compact
Riemannianmanifold
withweight function w and
let u be aneigenfunction of L
witheigenvalue
A > 0. Assumethat Rw ;
K. Then 1 we havePROOF. The
proof
is a modification of theproof
of Theorem 1 in[LiY]. Assuming
without loss ofgenerality
that supu ~ I
= 1and 0
>1,
define a function F
by
and let x, be the
point
where F’ attains its maximum. Since u is smoothby elliptic regularity,
so isF, and, applying
the maximumprinciple,
wehave
Now,
so
that,
at xo ,A
straightforward computation
that uses(2.3), (2.8)
and Lu = Auyields
Now
be
a local orthonormal frame field in aneighbourhood
ofx,. An
argument
as in[LiY],
formula(1.9), yields
where the second
inequality
follows from(x - y)2 ~ ax 2 - a( 1 - a)-ly2
with a =
(n - 1)/2n,
and the lastinequality
is a consequence of L1u =- = £w - and of
( x
+y )2
~( 1 - 8)-1 X2
+s -1 y 2
with s ==
1/(n
+1).
Notice that our choice of a(Li
and Yau take a= 1/2)
is moti- vatedby
the desire ofgiving
a unified treatment of cases n = 2 and n ~ 2here and in the
sequel.
The definition of Hessian and
(2.8) imply that,
at x,,for
every i,
sothat, assuming
without loss ofgenerality that,
at Xo,81 w # 0,
and =0,
i >1,
we obtainPutting (2.11)
into(2.10), substituting
the result into(2.9)
andsimplify- ing,
weget
Since, by assumption,
supu I
=1,
we havelu/(,8 - u) ~ ~ 1 /(~3 - 1),
and1). Substituting
these into theinequality above,
andusing
thehypothesis
and the definition ofF,
wefinally
findwhich can be rewritten as
with
Notice that if K >
0,
then ~, ~(n
+ >(n - 1 ) x, by
Theorem2.2,
sothat C is
always positive.
Thequadratic
formulaapplied
to the last in-equality gives
and
(2.7)
follows.REMARK. A
similar,
butweaker,
result can be obtainedby using
therelationship
noted in the introduction betweeneigenfunctions
of L on Mand
eigenfunctions
of theLaplacian
of thewarped product
M = Mx w S 1.
Notice that M is a
compact, (n
+1 )-dimensional
manifold andthat, by [ON], Corollary 7.43,
we havefor
X,
Y E TM = TMX ~ 0 ~
cT(M),
O ET,S
and where a hat on asymbol
indicates that the
corresponding object
is defined on M. For u as inthe statement of the theorem define the function S on M
by 8 )
==
u(x), (x, e) e
Mso that - 2i = AS.Assuming
that and that- w -1 dw ~ x, [LiY],
Theorem1, yields
which
immediately gives
THEOREM 2.4. Let
M,
be as in Theorem2.3,
and assume thatSw ;
K. Then Va ~ 0 and~3 2 ~
sup(a + U)2,
REMARK. Notice
that,
since =Rw
+d(log w)
0d(log
(2.12) certainly
holds with the same constant if we assume instead thatRw > K.
PROOF. As in
[LiY],
Theorem4,
theproof
is carried outby applying
the
arguments
used above to the function G definedby
where a > 0 and > sup
(a
+u )2 .
In the next two theorems we use the
gradient
estimatesjust
derivedto obtain lower bounds for the first non-zero
eigenvalue
of L.THEOREM 2.5. Let M be a
compact,
n-dimensionalmanifoLd
withweight function
w and diameterd,
and assume that K.If
~ d - 2
then thefirst
non-zeroeigenvalue ~,1 of
Lsatisfies
PROOF. The
proof
mimics that of Theorem 7 in[LiY]
and weonly
sketch it here. Let u be an
eigenfunction belonging
to theeigenvalue X1.
We may assume that sup
u ~ I
= sup u. Since u isL 2 (w dV)-orthogonal
tothe
constants,
the nodal set N of u is notempty. Integrating u) along
the shortestgeodesic
from N to thepoint
xwhere u attains its
maximum,
andusing
Theorem2.3,
we findso that
squaring
andsimplifying
weget
for every 9
= (~3 -
E(0, 1).
Under theassumption
thatthe
right
hand side of(2.17)
is maximized forand, since,
(2.13)
follows.REMARK. If 2 nK >
d -2,
then theright
hand side of(2.14)
is maxi- mizedfor e
= 1 and thisgives
the estimate(r~
+1) - 1 nK
which isworse than the estimate
provided by
Theorem 2.2. ForRw non-negative,
the
following
theoremgives
a better bound for~,1.
THEOREM 2.6. Let M be a
compact
Riemannianmanifolds
withweight function
w and assurrze that K. ThenPROOF.
Again
let u be aneigenfunction belonging
to theeigenvalue By changing
thesign
of u if necessary, we may assume thatsup u ~
~
inf u 1. Therefore, letting a
= 0 and~3
= sup u in Theorem2.4,
we haveThe
proof
nowproceeds
as in[Li],
Theorem 3:Integrating (2.19) along
the
minimizing geodesic y joining
thepoint
where u achieves its supre-mum to the
point
where it attains itsinfimum,
andtaking
into accountthe fact that the
length
of y is at mostd,
weget
If the
multiplicity
ofX1
is >2,
Li’sargument
shows that there is aneigenfunction u
for which sup u =inf u I
and therefore(2.15)
holds with,~2 /( 2 d 2 ) replaced by Z2 Id 2.
In thegeneral
case one considers theprod-
uct manifold M = M x M with the
weight
functionw( x , y )
=w( x ) ~ w( y ).
It is easy to see that the
operator
L induced on M issimply Lx
+Ly ,
9where
Lx
= -L1 x - Vx (log w)(x),
so thatLf (x, y)
=Lx f(x, y)
++
Ly f(x, y ).
Thisimplies
that the first non-zeroeigenvalue ~,1
of L is~.1
.
with
multiplicity ->
2.Moreover,
forX,
Y Ewhere
Xi , X2 (resp. Yl , Y2 )
are theprojections
of X(resp. Y)
on TM xx ~ 0 ~ and {0}
x TM. Thus thehypothesis ,Sw ; x
holds and we can con-clude as above that
where d is the diameter of M = M x M. Since
Å1
andd 2 = 2 d 2, (2.15)
follows.REMARK. From the variational characterisation of
~,1
1 and theidentity
that holds for every finite measure 03BC and one concludes that
where XA1
is the first non-zeroeigenvalue
of the standardLaplacian
of M.min can be estimated
by integrating I V(Iog w) I along
a min-imising geodesic joining
thepoints
where w attainsrespectively
its mini-mum and maximum
value,
thusyielding
Together
with known estimates forÀ f
thisgives
bounds for ~,1in
terms of Riccicurvature,
diameter and of(sup I V (log W)
Thedependency
onM
the last
quantity
reflects theperturbative
nature of the method. The boundsprovided by
Theorems2.2, 2.5,
and2.6, by contrast,
are ex-pressed
in terms of the tensorRw
which takes into account themutual,
and
possibly competing,
interaction of the curvature of M and of the be- haviour of w.We conclude this section
showing
how thetechniques
introducedabove can be used to
give
lower bounds for the firsteigenvalue
of L withDirichlet
boundary
conditions. In what follows ,S~ will be an open rela-tively compact
domain in M with smoothboundary
We will denoteby
the outward unit normal to
and,
for x EH(x)
will be themean curvature of 8Q with
respect
toa/av,
defined as -THEOREM 2.7. Let M be a
complete
Riemannianmanifold
withweight
w and let Q be arelatively compact
open domain in M with smoothboundary
Assume that K inQ,
with x ~0,
and thatHo
is a lower bound
for
the mean curvature Hof If
u is apositive
sol-2Gt2on
of
then, 1,
eitheror
PROOF.
Assuming
without loss ofgenerality
that supu ~ I
= 1 andf3
>1,
we consideragain
the smooth function Qand let xo be the
point
where F attains its maximum.If then 0 and a
computation
as in theproof
ofTheorem 2 in
[LiY]
shows thatSince =
0, Vu(x,) =
313v so thatMoreover,
since u ~ 0 in S~ andu(xo ) = 0,
we have8u/8v(xo) =
= -
, and the
inequality
above becomesand
(2.18)
follows. Ifx, Ei Q, (2.17)
follows from Theorem 2.3. Notice that since now we cannotguarantee
that~,1; x,
thepositivity
of C in theproof
of Theorem 2.3 follows from theassumption K S
0.Now let u be the
eigenfunction belonging
to the first Dirichleteigen-
value ~,1
of
L on S~. Theproof
that onegives
in the case of theLaplacian (cf. [Bd],
ch.IV,
Lemma3)
can be used to show that u has constantsign
in S~.
Assuming u
>0,
we canapply
thegradient
estimate obtained above and deduce thefollowing
theorems.THEOREM 2.8. Let M and Q be as above.
Suppose
thatRw ;
K, with K S 0. LetHo
be the lower boundfor
the mean curvatureof
8Q and de-note
by
i the inscribed radiusof
Q. Thenwith
PROOF. As in
[LiY],
Theorem5,
one shows that iff3
in Theorem 2.7 ischosen so that
then
(2.17)
holds. One thenproceeds
as in theproof
of Theorem 2.5 above.3. - The
non-compact
case.In this section we
study
theoperator
L defined on anon-compact
manifold M. Our
goal
is to find lower bounds for the bottom of the spec- trumÀ 0
of L. Note thatby
thespectral
theorem and thedensity
ofC~°° (M)
in the domain ofL, ~, o
admits thefollowing
variational character- ization :where f varies
overCcOO(M),
orequivalently,
overH 1 (M, w dv),
definedas usual as the space of all
functions f such that f,
Recallthat if M is a Riemannian manifold with
weight
function w and D is ameasurable subset
of M,
the w-volumeof D,
denoted isby
defi-nition the measure of D with
respect
to w dV. Ifvolw (M)
oo,clearly
~, o = 0.
Theproposition below,
which extends to L a theorem provenby
Brooks
([Br])
for theLaplacian,
shows that much more is true.PROPOSITION 3.1.
Suppose
that the w-volumeof
thegeodesic
ballsof M
growssubexponentially,
i. e.for
some(and therefore all)
p E M andfor
all a >0,
Then
~,0=0.
After
replacing
the Riemannian volume with thew-volume,
theproof
is as in Davies
[Da2],
p.157,
and therefore will be omitted.Turning
topositive results,note
first that(3.1) immediately implies
that
where
~, o
is the bottom of thespectrum
of the standardLaplacian
of M.Of course this is
interesting only
when it is apriori
known that is bounded above and away from 0. In thegeneral
case a more direct ap-proach
is needed.In the
following
M will be a manifold with apole,
i.e. there exists p EE M such that is a
diffeomorphism.
Let(r, u)
E R + xx
,STp
M bespherical geodesic
coordinatesat p E M,
whereSTp
M =S n -
1is the unit
tangent
space at p. We denoteby ÝiJ( r, u)
the area element inthe coordinates
(r, u),
so that the Riemannian volume element of M isgiven by
dV =u) dr du,
dubeing
the standard measure onSTp M.
Therefore a standard
integration by parts argument (cf. [Cl],
p.47) yields:
THEOREM 3.2. Let M be a
manifold
withweight function
w and letp E M be a
pole. If
for
all(r, u)
eR+ xSTp M,
thenTo
apply
the result obtained above one needs to estimate Thefollowing proposition,
ofperturbative nature,
is a firststep
in this direction.PROPOSITION 3.3. Let M be a
manifold
with apole p
and withweight function
w.Suppose
that Sect 0for
all radial2-planes z
and that in
geodesic polar
coordinates at p we havefor
all(r, u).
ThenIn
particular
this holdsfor all p
E Mif
M iscomplete simply
connectedwith Sect
(M) ;
0 and Ric ~ -(~i
+I lw
-1Vwlloo )2.
PROOF. Observe first that the conclusion of Lemma 3 in
[Sel]
holds under the weakerhypothesis
that the radial sectional curvature and theradial
component
of the Ricci curvaturesatisfy
theinequalities
statedthere. Therefore
(3.4) implies
thatSince Lemma 4 in
[Sel]
holdsprovided ýg-1
0 and this isagain guaranteed by
ourassumption
on the radial sectionalcurvature,
we conclude that
and
(3.5)
follows fromREMARKS.
1) Example
1 below shows that eventhough
it is rathercrude, Proposition
3.3provides,
via Theorem2,
asharp
lower bound for2) Proposition
3.3 can beapplied
when the supremum of the sec-tional curvature is zero and the Ricci curvature is bounded above
by
anegative
constant(and
theperturbation
introducedby
the termis
suitably small).
This is the case, forinstance,
when M is theproduct
oftwo or more Riemannian manifold with
negative
curvature. Even in thissimple
case the bound forÀ 0
that one finds is far frombeing sharp.
Inthe
proof
ofProposition
3.3 above we have used theinequality
Therefore the bounds one can derive from it are more
interesting
whenis
negative.
We consider now the case 0.THEOREM 3.4. Let M be a
manifold
with apole
p andweight func-
tion w and denote
by
313r the radial unit vectorfield.
Assume thatfor
all q E M
i)
For all radical2-planes Tq M,
k ~0 ;
Then
PROOF. An
argument
as in theproof
of Lemma 3 in[Sel], which,
asremarked
above,
can be carriedthrough
under the weakerhypothesis
that M has a
pole
and the radial curvatures arecontrolled,
shows thati)
andii) imply
whence
The rest of the
proof
is now asimple
modification of theproof
of Lemma4 in
[Sel ].
Since~ -1 a ~~ar ~ (n - 1 ) r -1
asr 1
0 aw/ar issmooth and
bounded,
we havefor r small. Since
i)
andiii) imply
thatThe
argument
in[Se1]
shows that(3.6)
follows from(3.7).
We conclude the section with two
examples.
EXAMPLE 1. Let
M = Hn 4,
the n-dimensionalhyperbolic
space with constantcurvature -4,
which weidentify
with Rn endowed with the metricgiven
inpolar coordinates(r, g) by
+
11 /2 sinh ( 2 r) dg 12
and letw(r, ~) =
Since 1is C °° and even on
I~,
w is C °° on M.Since w -1
aw/ar = -(n - 1 ) tanr
we have(n - 1 )
so that we can writewith a = 1.
By Proposition
3.3 and Theorem 3.2 it follows that~, o ~
- l(n -1)2 /4.
On the other hand andconsequently w
dV =(sinh r)n -1
drd~.
Therefore we see thatif f is
a radi-al
function,
theRayleigh quotient
for Lcoincides with the
Rayleigh quotient
for the standardLaplacian - L1
onthe
hyperbolic
space with constant curvature -1. Since both the metric of M and theweight
function w areradially symmetric
one caneasily
see thatand since this is also true for the
Laplacian - A
on we concludethat the bottom of the
spectrum
of L on M coincides with the bottom of thespectrum
of - L1 on Thus~, o
=(n - 1 )2 /4 ([McK]), showing
thatthe bound obtained above is
sharp.
EXAMPLE 2. Let M
= R2
with the metricgiven
inpolar
coordinates(r, 8) by
and letw(r, 8)
= cosh r. Notice that in thiscase
Proposition
3.3 is notapplicable.
On the other hand we haveSect (M) = 0, a/ar) _ -1
and 3wl3r = sinh r >0,
soProposition
3.4
implies
thatA,, ->
1/4. We will show that in fact = 1 /4 whichimplies
that
Proposition
3.4gives
asharp
lower bound forA 0.
To see this weadapt
aproof by Pinsky [Pk]. Let f be
definedby f(r, 8) _ q5(r),
where~(r) = e -r~2 sin [ 2 ~t(r - d/2 ) ~ -1 ] if r E [~/2 , ~ ]
and~(r) = 0
otherwise.Then and on
[ d/2 , d ], ~
satisfies the differentialequation
Integrating by parts, using
the differentialequation
and the fact thatr -1
+ tanh r -1 isdecreasing
in(0, 00),
we find:Thus, integrating
overS1, dividing through by
andapplying
thequadratic
formulayields
whence, letting
ð~ + 00gives ~, o ~ 1/4,
and our claim follows. thus prov-ing
our claim.4. - E stimates from above.
In this
section, essentially extending
ideas of S. Y.Cheng presented
in
[Cg],
westudy
bounds from above for theeigenvalues
ofL,
in the com-pact
case, and for the bottom of thespectrum,
for Mnon-compact.
For
p e M
and let be the distancealong
thegeodesic
y u =
expp (tu)
from p to the cut locus of p. Notethat ~ (r, u)
E R + xx 0 r
c(u) ~
is the domain of thepolar geodesic
coordinates at p.Keeping
the notation introduced in theprevious section,
we havew(r, ~c)
> 0 dr E(0, c(u) ) ,
and therefore for everyweight
function w we have
~c)
> 0 in the same domain. Tosimplify
thenotation we let also
The
following
theoremgeneralizes Bishop’s comparison
theorem and will allow us to extend to L some ofCheng’s
results.THEOREM 4.1. Let M be a Riemannian
manifold
withweight function
w. Assume that inpolar geodesic
coordinatesat p
wehave
Then, VueSTpM
andc(u) ),
where
~3
= a/n andCf3
aredefined in (4.1 ). Moreover, if a>0,
PROOF.
Adopting
Chavel’s notation([CI],
p. 65ff.),
forU E TpM,
let
A(r, u)
be the selfadjoint isomorphism
of definedby
where is theparallel
translationalong
thegeodesic
= ThendetA(r, u) = w(r, u)
andif we define then U is self
adjoint,
tr U(r, u) = W-1 aw/ ar( r, u),
and thefollowing
differentialequation
holds:
Using
theinequalities
i)
trU2 > (tr U)2 /(n - 1),
which follows from the selfadjoint-
ness of
U,
iii)
andA 2 /( n - 1)
+B 2 ~ (A
+B )2 /n VAN, B,
withequality
iffA =
(n - 1 )B,
it follows thatThus the function
o(r)
definedby
for re
(0, c(u)),
satisfies the differentialinequality
where
fl
=a/n,
andSetting y(r)
= one verifies that 1jJ satis- fies the differentialequation
, (¡
and
therefore,
as in theproof
ofBishop’s comparison
theorem([CI],
p.
73),
one concludes thaty(r)
in(0,
n(0, c(u) ),
where+ 00 0.
0,
theproof
iscomplete.
If{3
~0,
takec( u ) )
n(0,
and notethat,
sincewe have
REMARK. Let M n and w be as in the statement of the
Theorem,
witha > 0. As in the
proof
of theBonnet-Myers Theorem,
since M iscomplete
and the
geodesic
y u is notminimizing
afterc(u),
it follows thatdist (q, p) ~
for allq E M,
and M n iscompact by
theHopf-Rinow
theorem. If we assume that a, with a >
0,
we can concludethat,
infact, diam(M) S
In connection withthis,
observe thatBakry ([Bkl])
has shown that a > 0implies
thatvoly(M)
oo.THEOREM 4.2. M and R >
0,
and consider thegeneralized
Dirichlet
problem for
L on thegeodesic
ball in M centeredat p
withIf
in
Bp (R),
thenwhere
Åo(Bp(R)
is the bottomof
thespectrum of
L on andI
~, -a (B; + 1»)
is the smallest Dirichleteigenvalue for -
L1 on the diskof
radius R in
Mn + 1,
the(n
+1 )-dimensional
space with constant curva-ture
{3
= a/n.REMARKS.
i)
Since we do not assume that is contained in the domain of the normal coordinates at p, 3B may fail to be smooth and it is necessary to consider thegeneralized
Dirichletproblem
on In any case thebottom of the
spectrum
of L isgiven by
where
f
ranges over or,equivalently,
overHo1 (Bp(R), wdV).
ii)
If a > 0 so that0,
thenMrp + 1
is the n + 1sphere
with curva-ture
/3
and Riemannian diameterjr/N F8. According
to Theorem 4.1 wealso have
d(p, dq E M. Consequently
the theorem has con-tent
only
for R Sn/ýp,
for if R > thenBp (R )
= M andand the theorem holds
trivially
withA 0 (M)
= 0 =~, -° (M~ + 1 ).
No-tice that if R then À -LI + 1
(R) )
= 0(cf. [Cl],
p.53),
and soagain ~, o (B~ (R ) ) = 0 = ~, - ° (B~ + 1 (R ) ) .
PROOF. Since the
proof
is very similar to that ofCheng’s
theorem([Cg],
Theorem1.1,
or[Cl],
pp.74-77)
we willonly
indicate how toadapt Cheng’s proof
to our case. Let T be the radialeigenfunction belonging
to~, -° (B,~ + 1 (R) )
and define Fby F(q)
=T(r(q»,
wherer(q)
=d(q, p).
Itis easy to see that
Letting
for convenienceb(u) =
= min
(c(u), R),
andusing
Theorem 4.1 instead ofBishop’s Comparison
Theorem in the
integration by parts argument
in[Cg],
pp.290-91, yields
so
that, integrating
we findand
(4.6)
follows from(4.7).
COROLLARY 4.3. Let M be a
compact manifold
withweight func-
tion w and assume that
be the sequence
of
theeigenvalues of L,
where eacheigenvalue
isrepeat-
ed
according
to itsmultiplicity.
Thenfor
every m we havewhere d is the diameter
of M and f3
and~, -° (B~ + 1 (R) )
aredefined
inTheorem 4.2.
PROOF. For every m,
let Øm
be the(normalized) eigenfunction
be-longing
toÀ m.
As in the case of theLaplacian
it is easy to see thatwith
equality Using
Theorem 4.2 above theproof
follows ex-actly
as in[Cg],
Theorem 2.1.Using
the estimate for~, -° (BK (R) )
obtainedby Cheng ([CG]),
wealso have:
COROLLARY 4.4. Let M be as in
Corollary 4.3,
and assume thatRw ~ -
a, with a ~ 0. Thenwith
C(n) depending onLy
upon n.We conclude
remarking
thatCheng’s
results relative tonon-compact
manifolds
([Cg], § 4)
also extendeffortlessly
to theoperator
L and allowus to state the
following corollary:
COROLLARY 4.5. Let M be an n-dimensional