A NNALES DE L ’I. H. P., SECTION B
M ARK A. P INSKY
On non-euclidean harmonic measure
Annales de l’I. H. P., section B, tome 21, n
o1 (1985), p. 39-46
<http://www.numdam.org/item?id=AIHPB_1985__21_1_39_0>
© Gauthier-Villars, 1985, tous droits réservés.
L’accès aux archives de la revue « Annales de l’I. H. P., section B » (http://www.elsevier.com/locate/anihpb) implique l’accord avec les condi- tions générales d’utilisation (http://www.numdam.org/conditions). Toute uti- lisation 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 Non-Euclidean Harmonic Measure
Mark A. PINSKY
Department of Mathematics, Northwestern University,
Lunt Hall Evanston, Ill 60201, U.S.A.
ABSTRACT. - Let
HE
be the harmonic measure of thes-sphere
in aRiemannian manifold. Riemannlan We compute
lim
in terms of thegradient
.
manifold. W e
compute lim ~~0
8 3 In terms 0 t e gra lenof the Gaussian curvature. ~~
°
RESUME. - Soit
HE
la mesureharmonique
de lasphere
du rayon 8 d’une variété riemannienne de dimension deux. Onexprime
lim(HE -
en fonction de la derivee de la courbure
gaussienne.
1. INTRODUCTION
Let
(M, g)
be an n-dimensional Riemannian manifold withBrownian
motionprocess {Xt,
t >0 }. According
to theOnsager-Machlup
for-mula
[3 ],
the « mostprobable path »
is that of a classicalparticle
in aconservative force field whose
potential
energy is one-twelfth the scalar curvature. This suggests that the « mostprobable path »
follows the nega- tivegradient
of the scalar curvature. The purpose of this note is to make thisprecise
in terms of the harmonic measure of a smallsphere.
For tech-nical reasons we restrict the discussion to surfaces. The
technique
followsthe method used to
study
the mean exit time[1 ].
Annales de l’Institut Henri Poincaré - Probabilités
40 M. A. PINSKY
2. NOTATIONS AND DEFINITIONS
Let
(M, g)
be an n-dimensional Riemannian manifold. We use the follow-ing
notations :M~
is the tangent space at m E M.Bm(e)
is the ball of radius E in M with center at m E M.is the ball of radius e in
M~
with center at6
EM~.
exp~ is the
exponential mapping (which
is defined on allof Mm
in case Mis
complete;
otherwise it is amapping)
from toBm(E)
forsufficiently
small e > 0. ,4Y~
is themapping
on functions definedby
D,
maps from to forsufficiently
small e > 0.A is the
Laplace-Beltrami
operator of the Riemannian manifold :The
following result,
which will be usedrepeatedly,
wasproved
in[1].
PROPOSITION 2. O. - There exist second order
differential
operators(Ll - 2, 03940, 03941, ...)
on such thatfor
each N >_ 0 and eachfe
l1j
mapspolynomials of degree
k topolynomials of degree
k +j.
In any nor- mal coordinate chart(x 1, - - - , xn)
we haveHere
Riajb
is the Riemann tensorand j
is the Ricci tensor atmEM.
Annales de l’Institut Henri Poincaré - Probabilités et Statistiques
Let
(Xt, Px)
be the Brownian motion process with infinitesimal gene-rator A. For each ~ ~ M let
Tg
be the exit time from thegeodesic
ballB~(e).
In case n = 2 we introduce
geodesic polar
coordinates(r, 8) by
Xl = r cos8,
x2 = r sin 8. The coordinate formulas for the first three operators become
~K where K is the Gaussian curvature and
Ko = K(m), Ki
=- (m),
~K .
~’
K2 = - (m).
Thiscomputation
is carried out in theAppendix,
Sec. 5.3. STATEMENT OF RESULTS
To
study
the harmonic measure we letf
E a smooth functionon the
circle. Extend f to R~B { 0 } by making f constant along
raysthrough
the
origin. Equivalently f
can bethought
of as a function onMmB ~
0}.
Then is a smooth function on
M B ~ m ~
which is constantalong
geo- desicseminating
from m. The harmonic measure operator isTHEOREM 3.1. - When
8 ~
0 we havewhere ~
is normalizedLebesgue
measureand ~ VK,
=Kl
cos o +K2
sino,
--71 ~ 0 71.
Proof
- We follow theperturbation
method introduced in[1 ].
LetUi E
(i
=0,2, 3)
be definedby
42 M. A. PINSKY
The
proof depends
on thefollowing
three lemmas : LEMMA 1. -LEMMA 2. -
LEMMA 3. -
Proof of Lemma
1. uo is’the
solution of the classical Dirichletproblem
in the unit disc. thus
In
particular
ProofofLemma
2.- Equivalently
we write uo as a Fourier seriesThus
.
00
This is a sum
of homogeneous polynomials hn. By
Lemma 6.3 of[7]
we haveu2(0)
==03A3 1 (n+2)2 ~ 03C0 -03C0 hn(03B8)03C9(d 0). But from the orthogonality
relations for
i
the functions
(cos
sin~0),
all of theseintegrals
are zero, henceM~(0)
= 0.D
(Note :
The abovedepends
on the fact that03940u0
is a harmonic functionwhich vanishes at the
origin.
This may also be seen from the relation0394-203940 - AoA-2
= -applied
touo.)
3
Annales de l’Institut Henri Poincaré - Probabilités et Statistiques
Proof of
Lemma 3. We haveReferring
to the formula(2.6),
we see that00 .
a sum
of homogeneous polynomials I hk. By
Lemma 6 . 3 of[1] we
haveu3(0) == I (k + 2) hk(03B8)03C9(d03B8). But from
the2
of Fourier series all of these
integrals
are zero except for k ==2,
I. e. n == I.This
gives
.Ct)
On the other hand
/(0)
=Ao + I 1 (An cos
n03B8 + and thus
1 We have
proved
thatu3(0)
.
== - - 1 fn cos 6 + K2
sin 03B8)03C9(d03B8)
as
required.
D 32 -n44 M. A. PINSKY
To
complete
theproof
of the theorem weappeal
toProposition
2.0with f =
uo +E2u2
+E~u~ ;
thuswhere we have used the
defining
relations(3.1)-(3.3).
Hence AU =0(E2)
where U = +
E2u2
+E3u3). Noting
that U = on we havethat for any x E
by Dynkin’s
formulaSetting
x = m we haveU(m)
=uo(O)
+e2u2(O)
+ from which the result follows. D4. COMPARISON
WITH THE NON-STOCHASTIC MEAN VALUE
The mean-value operator of a Riemannian manifold is defined
by
where 6E is the surface measure on the
e-sphere Se and fis
a smooth functionon the unit
sphere.
We have thefollowing
result.THEOREM 4 . 1. - When e
1
0Proof
From the method ofGray
and Willmore[2 ],
we haveAnnales de l’Institut Henri Poincaré - Probabilités et Statistiques
where
Using
theexpansion
we have
In
particular
Thus
which was to be
proved.
COROLLARY. - = every E >
then
(M, g)
has constant curvature.Proof
We haveIf this is zero for
aUf;
thenK2
=0,
i. e.VK(m)
=0,
and hence Kis constant.
, 5. APPENDIX. COMPUTATION
Ao, Ai
In
geodesic polar
coordinates we haveThe Jacobi
equation connecting
the metric with the curvature isVol. 21, n° 1-1985.
46 M. A. PINSKY
In a
neighborhood
of m, we can writeThis
yields
Performing
the indicatedoperations,
we haveSubstituting
in(5.1)
andequating
coefficients of ryields (2.4)-(2.6).
ACKNOWLEDGEMENT
I would like to thank Victor Mizel for some
helpful suggestions.
REFERENCES
[1] A. GRAY and M. PINSKY, The mean exit time from a small geodesic ball in a Riemannian manifold, Bulletin des Sciences Mathématiques, t. 107, 1983, p. 345-370.
[2] A. GRAY and T. J. WILLMORE, Mean-value theorems for Riemannian manifolds, Pro- ceedings of the Royal Society of Edinburgh, t. 92 A, 1982, p. 343-364.
[3] Y. TAKAHASKI and S. WATANABE, The Onsager-Machlup function of diffusion processes, Stochastic Integrals, Springer-Verlag Lecture Notes in Mathematics, t. 851, 1981,
p. 433-463.
(Manuscrit reçu le 10 Avril 1984) (Corrigé le 25 Septembre 1984)
Annales de l’Institut Henri Poincare - Probabilités et Statistiques