Journal of Lie Theory

?? (??) ????
c _ ?? Heldermann Verlag Berlin
Version of December 14, 2008
Equipartition of energy for the wave equation
associated to the Dunkl-Cherednik Laplacian
Fatma Ayadi

Abstract. This paper is concerned with energy properties of the wave equa-
tion associated to the Dunkl-Cherednik Laplacian. We establish the conservation
of the total energy, the strict equipartition of energy under suitable assumptions
and the asymptotic equipartition in the general case.
Mathematics Subject Index 2000: Primary 35L05 ; Secondary 22E30, 33C67,
35L65, 58J45
Keywords and phrases: Dunkl-Cherednik operator, wave equation, energy,
equipartition
1. Introduction
We use [10] as a reference for the Dunkl-Cherednik theory. Let a be a Euclidean
vector space of dimension d equipped with an inner product , ). Let 1 be a
crystallographic root system in a, 1
+
a positive subsystem and W the Weyl
group generated by the reections r

(x) = x2
,x

2
along the roots 1.
We let k : 1 [ 0, +) denote a multiplicity function on the root system 1,
and =
1
2

R
+
k

. We note that k is W-invariant. The Dunkl-Cherednik

operators are the following dierential-dierence operators, which are deformations
of partial derivatives and still commute pairwise :
T

f(x) =

f(x) , )f(x) +

R
+
k

, )
1 e
,x
f(x)f(r

x) .
Given an orthonormal basis
1
, . . . ,
d
of a, the Dunkl-Cherednik Lapla-
cian is dened by
Lf(x) =
d

j=1
T
2

j
f(x) .
More explicit formulas for L exist but they will not be used in this paper. The
Laplacian L is selfadjoint with respect to the measure (x)dx where
(x) =

R
+

2 sinh
, x)
2

2k
.

This work is part of my master thesis, done under the supervision of M. Si (Universite
de Tunis El Manar) and J.Ph. Anker (Universite dOrleans), within the program CMCU 07G
1501 Aspects analytique et probabiliste de la theorie de Dunkl
h
a
l
-
0
0
3
4
7
1
4
6
,

v
e
r
s
i
o
n

1

-

1
4

D
e
c

2
0
0
8
2 Ayadi
Consider the wave equation
_

2
t
u(t, x) = L
x
u(t, x),
u(0, x) = f(x),
t

t=0
u(t, x) = g(x),
(1)
with smooth and compactly supported initial data (f, g). Let us introduce:
the kinetic energy /[u](t) =
1
2
_
a

t
u(x, t)

2
(x) dx,
the potential energy T[u](t) =
1
2
_
a
Lu(x, t) u(x, t) (x) dx,
the total energy c[u](t) = /[u](t) +T[u](t).
In this paper we prove
the conservation of the total energy:
c[u](t) = constant, (2)
the strict equipartition of energy, under the assumptions that the dimension
d is odd and that all the multiplicities k

are integers:
/[u](t) = T[u](t) =
1
2
c[u] for [t[ large , (3)
the asymptotic equipartition of energy, for arbitrary d and R
+
-valued mul-
tiplicity function k:
/[u](t)
1
2
c[u] and T[u](t)
1
2
c[u] as [t[ goes to . (4)
The proofs follow [6] and use the Fourier transform in the Dunkl-Cherednik
setting, which we will recall in next section. We mention that during the past
twenty years, several works were devoted to Huygens principle and equipartition
of energy for wave equations on symmetric spaces and related settings. See for
instance [3], [8, ch. V, 5], [6], [7], [2], [4], [5], [1].
2. Generalized hypergeometric functions and Dunkl-Cherednik
transform
Opdam [9] introduced the following special functions, which are deformations of
exponential functions e
,x
, and the associated Fourier transform.
Theorem 2.1. There exist a neighborhood U of 0 in a and a unique holo-
morphic function (, x) G

(x) on a
C
(a+iU) such that
_
T

(x) = , ) G

(x) a ,
G

(0) = 1 .
Moreover, the following estimate holds on a
C
a :
[G

(x)[ [W[
1
2
e
Re x
.
h
a
l
-
0
0
3
4
7
1
4
6
,

v
e
r
s
i
o
n

1

-

1
4

D
e
c

2
0
0
8
Ayadi 3
Denition 2.2. Let f be a nice function on a, say f belongs to the space
(

c
(a) of smooth functions on a with compact support. Its Dunkl-Cherednik trans-
form is dened by
Tf() =
_
a
f(x) G
iw
0

(w
0
x) (x) dx.
Here w
0
denotes the longest element in the Weyl group W.
The involvement of w
0
in the denition of T is related to the below skew-
adjointness property of the Dunkl-Cherednik operators with respect to the inner
product
f, g) =
_
a
f(x) g(x) (x) dx, f, g (

c
(a).
Lemma 2.3. The adjoint of T

is w
0
T
w
0

w
0
:
T

f, g) = f, w
0
T
w
0

w
0
g) .
As an immediate consequence, we obtain:
Corollary 2.4. For every , a and f (

c
(a), we have
T(T

f)() = i, )Tf(),
and therefore
T(Lf)() = ||
2
Tf() .
Next we will recall from [9] three main results about the Dunkl-Cherednik
transform (see also [10]). For R > 0, let (

R
(a) be the space of smooth functions
on a vanishing outside the ball B
R
=xa [|x|R. We let H
R
(a
C
) denote the
space of holomorphic functions h on the complexication a
C
of a such that, for
every integer N>0,
sup
a
C
(1+||)
N
e
RIm
[h()[ < +.
Theorem 2.5. (Paley-Wiener) The transformation T is an isomorphism of
(

R
(a) onto H
R
(a
C
), for every R>0.
The Plancherel formula and the inversion formula of T involve the complex
measure () d with density
() =

R
+
0

_
i, +k
_

_
i,
_

_
i, +k
2
+k
2
_

_
i, +k
2
_

_
i, +k
_

_
i, +1
_

_
i, +k
2
+k
2
+1
_

_
i, +k
2
_
,
where 1
+
0
= 1
+
[

2
/ 1 is the set of positive indivisible roots, =2||
2

the coroot corresponding to , and k

2
=0 if 2 / 1. Notice that is an analytic
function on a, with a polynomial growth and which extends meromorphically
to a
C
. It is actually a polynomial if the multiplicity function k is integer-valued
and it has poles otherwise.
h
a
l
-
0
0
3
4
7
1
4
6
,

v
e
r
s
i
o
n

1

-

1
4

D
e
c

2
0
0
8
4 Ayadi
Theorem 2.6. (Inversion formula) There is a constant c
0
>0 such that, for
every f (

c
(a),
f(x) = c
0
_
a
Tf() G
i
(x) () d .
Theorem 2.7. (Plancherel formula) For every f, g (

c
(a),
_
a
f(x) g(x) (x) dx = c
0
_
a
Tf()

Tg() () d ,
where

Tg() := T(w
0
g)(w
0
) =
_
a
g(x) G
i
(x) (x) dx .
3. Conservation of energy
This section is devoted to the proof of (2). Via the Dunkl-Cherednik transform,
the wave equation (1) becomes
_

2
t
Tu(t, ) = ||
2
Tu(t, ),
Tu(0, ) = Tf(),
t

t=0
Tu(t, ) = Tg(),
and its solution satises
Tu(t, ) = (cos t||) Tf() +
sint

Tg() . (5)
By means of the Paley-Wiener Theorem 2.5, in [1, p. 52-53] the author proves the
following nite speed propagation property:
Assume that the initial data f and g belong to (

R
(a). Then the
solution u(t, x) belongs to (

R+|t|
(a) as a function of x.
Let us express the potential and kinetic energies dened in the introduction
via the Dunkl-Cherednik transform. Using the Plancherel formula and Corollary
2.4, we have
T[u](t) =
c
0
2
_
a
||
2
Tu(t, )

Tu(t, ) () d . (6)
Moreover, since the Dunkl-Cherednik Laplacian is W-invariant, it follows that
(w
0
u)(t, x) = u(t, w
0
x) is the solution to the wave equation (1) with the initial
data w
0
f and w
0
g. Thus

Tu(t, ) = (cos t||)

Tf() +
sint

Tg() . (7)
Now, by substituting (5) and (7) in (6), we get
T[u](t) =
c
0
2
_
a
||
2
(cos t||)
2
Tf()

Tf() () d
+
c
0
2
_
a
(sin t||)
2
Tg()

Tg() () d
+
c
0
4
_
a
|| (sin 2t||)
_
Tf()

Tg() +Tg()

Tf()
_
() d .
(8)
h
a
l
-
0
0
3
4
7
1
4
6
,

v
e
r
s
i
o
n

1

-

1
4

D
e
c

2
0
0
8
Ayadi 5
Similarly to T[u], we can rewrite the kinetic energy as
/[u](t) =
c
0
2
_
a

t
Tu(t, )
t

Tu(t, ) () d .
Using the following facts
_

t
Tu(t, ) = || (sin t||) Tf() + (cos t||) Tg() ,

t

Tu(t, ) = || (sin t||)

Tf() + (cos t||)

Tg() ,
we deduce that
/[u](t) =
c
0
2
_
a
||
2
(sin t||)
2
Tf()

Tf() () d
+
c
0
2
_
a
(cos t||)
2
Tg()

Tg() () d

c
0
4
_
a
|| (sin 2t||)
_
Tf()

Tg() +Tg()

Tf()
_
() d .
(9)
By suming up (8) and (9), we obtain the conservation of the total energy :
c[u](t) =
c
0
2
_
a
_
||
2
Tf()

Tf() +Tg()

Tg()
_
() d = c[u](0) .
That is c[u](t) is independent of t.
4. Equipartition of energy
This section is devoted to the proof of (3) and (4). Using the classical trigonometric
identities for double angles, we can rewrite the identities (8) and (9) respectively
as
T[u](t) =
c
0
4
_
a
_
||
2
Tf()

Tf() +Tg()

Tg()
_
() d
+
c
0
4
_
a
(cos 2t||)
_
||
2
Tf()

Tf() Tg()

Tg()
_
() d
+
c
0
4
_
a
|| (sin2t||)
_
Tf()

Tg() +Tg()

Tf()
_
() d
and
/[u](t) =
c
0
4
_
a
_
||
2
Tf()

Tf() +Tg()

Tg()
_
() d

c
0
4
_
a
(cos 2t||)
_
||
2
Tf()

Tf() Tg()

Tg()
_
() d

c
0
4
_
a
|| (sin 2t||)
_
Tf()

Tg() +Tg()

Tf()
_
() d .
Hence
T[u](t) /[u](t) =
=
c
0
2
_
a
(cos 2t||)
_
||
2
Tf()

Tf() Tg()

Tg()
_
() d
+
c
0
2
_
a
|| (sin2t||)
_
Tf()

Tg() +Tg()

Tf()
_
() d .
(10)
h
a
l
-
0
0
3
4
7
1
4
6
,

v
e
r
s
i
o
n

1

-

1
4

D
e
c

2
0
0
8
6 Ayadi
Introducing polar coordinates in a, (10) becomes
T[u](t) /[u](t) =
c
0
2
_
+
0
_
cos(2tr) (r) + sin(2tr) r (r)
_
r
d1
dr , (11)
where
(r) =
_
S(a)
_
r
2
Tf(r)

Tf(r) Tg(r)

Tg(r)
_
(r) d ,
(r) =
_
S(a)
_
Tf(r)

Tg(r) +Tg(r)

Tf(r)
_
(r) d ,
and d denotes the surface measure on the unit sphere S(a) in a. Let
0
(0, +]
be the width of the largest horizontal strip [Imz[ <
0
in which z (z) is
holomorphic for all directions S(a).
Lemma 4.1. (i) (z) and (z) extend to even holomorphic functions in the
strip [Imz[ <
0
.
(ii) If
0
<+, the following estimate holds in every substrip [Imz[
with <
0
: For every N > 0, there is a constant C > 0 (depending on
f, g (

R
(a), N and ) such that
[(z)[ +[(z)[ C [z[
|R
+
0
|
(1+[z[)
N
e
2R|Imz|
.
(iii) If
0
= +, the previous estimate holds uniformly in C.
Proof. (i) follows from the denition of and . Let us turn to the estimates
(ii) and (iii). On one hand, according to the Paley-Wiener Theorem (Theorem 2.5),
all transforms Tf(z),

Tf(z), Tg(z),

Tg(z) are O
_
1+[z[
N
e
R|Imz|
_
. On
the other hand, let us discuss the behavior of the Plancherel measure. Consider
rst the case where all multiplicities are integers. Without loss of generality, we
may assume that k

and k
2
N for every indivisible root . Then
() = const.

R
+
0
, )
_
, )+i(k

+2k
2
)
_

0<j<k
_
, )
2
+j
2
_
0
e
j<k
2
_
, )
2
+(k

+2

j)
2
_
is a polynomial of degree 2 [k[ = 2

R
+
k

. In general,
() = const. () () ,
where
() =

R
+
0
, )
is a homogeneous polynomial of degree [1
+
0
[ and
() =

R
+
0

_
i, +k
_

_
i, +1
_

_
i, +k
2
+k
2
_

_
i, +k
2
_

_
i, +k
_

_
i, +1
_

_
i, +k
2
+k
2
+1
_

_
i, +k
2
_
is an analytic function which never vanishes on a. Notice that z (z) or (z)
has poles for generic directions S(a) as soon as some multiplicities are not
integers. Using Stirlings formula
()

1
2
e

as [[ + with [ arg z[ < ,

h
a
l
-
0
0
3
4
7
1
4
6
,

v
e
r
s
i
o
n

1

-

1
4

D
e
c

2
0
0
8
Ayadi 7
we get the following estimate for the Plancherel density, in each strip [Imz[ <
with 0< <
0
:
[(z)[ C [z[
|R
+
0
|
(1+[z[)
2|k||R
+
0
|
.
The estimates (ii) and (iii) follow easily from these considerations.
Proposition 4.2. Assume that the dimension d is odd and that all multiplici-
ties are integers. Then there exists a constant C>0 (depending on the initial data
f, g (

R
(a)) such that, for every 0 and t R,
[T[u](t) /[u](t)[ C e
2(R|t|)
.
Proof. Evenness allows us to rewrite (11) as follows :
T[u](t) /[u](t) =
c
0
4
_
+

e
i2tr
_
(r)ir(r)
_
r
d1
dr .
Let us shift the contour of integration from R to Ri, according to the sign
of t, and estimate the resulting integral, using Lemma 4.1.iii. As a result, the
dierence of energy
T[u](t) /[u](t) =
c
0
4
e
2|t|
_
+

e
i2tr
_
(ri)i(r) (ri)
_
(ri)
d1
dr
is O
_
(1+)
N
e
2(R|t|)
_
.
As an immediate consequence of the above statement and in view of the
fact that
0
= when k is integer valued, we deduce the strict equipartition of
energy (3) for [t[ R, by letting .
Henceforth, we will drop the above assumption on k. By resuming the proof
of Proposition 4.2 and using Lemma 4.1.ii instead of Lemma 4.1.iii, we obtain the
following result.
Proposition 4.3. Assume that the dimension d is odd. Then, for every 0 <
<
0
, there is a constant C>0 (depending on the initial data f, g (

R
(a)) such
that
[T[u](t) /[u](t)[ C e
2|t|
t R.
As a corollary, we obtain the asymptotic equipartition of energy (4) in the
odd dimensional case, with an exponential rate of decay. In the even dimensional
case, the expression (11) cannot be handled by complex analysis and we proceed
dierently.
Proposition 4.4. Assume that the dimension d is even. Then there is a
constant C>0 (depending on the initial data f, g (

R
(a)) such that
[T[u](t) /[u](t)[ C (1+[t[)
d|R
+
0
|
t R.
h
a
l
-
0
0
3
4
7
1
4
6
,

v
e
r
s
i
o
n

1

-

1
4

D
e
c

2
0
0
8
8 Ayadi
Proof. The problem lies in the decay at innity. According to lemma 4.1, (r)
and (r) are divisible by r
D
, where D=[1
+
0
[ . Let us integrate (11) d+D times
by parts. This way
_
+
0
cos(2tr) r
d1
_
_
S(a)
Tg(r)

Tg(r) (r) d
_
. .
e
(r)
dr
becomes

1 or 0
(2t)
d+D
(d +D) !
(D+1) !
_

r
_
D+1

(r)

r=0

_
+
0
cos(2tr) or sin(2tr)
(2t)
d+D
_

r
_
d+D
_
r
d1

(r)
_
dr
which is O
_
[t[
dD
_
. Similarly
_
+
0
cos(2tr) r
d+1
_
_
S(a)
Tf(r)

Tf(r) (r) d
_
dr = O
_
[t[
dD2
_
and
_
+
0
cos(2tr) r
d
(r) dr = O
_
[t[
dD1
_
.
This concludes the proof of Proposition 4.4.
As a corollary, we obtain the asymptotic equipartition of energy (4) in the
even dimensional case, with a polynomial rate of decay.
Remark 4.5. Our result may not be optimal. In the W-invariant case, one
obtains indeed the rate of decay O
_
1+[t[
d2|R
+
0
|
_
as in [6].
References
[1] Ben Said, S., Huygens principle for the wave equation associated with
the trigonometric Dunkl-Cherednik operators, Math. Research Letters 13
(2006), no. 1, 4358.
[2] Ben Said, S. and rsted, B., The wave equation for Dunkl operators, Math.
(N.S.) 16 (2005), no. 34, 351391.
[3] Branson, T. and

Olafsson, G., Equipartition of energy for waves in sym-
metric space, J. Funct. Anal. 97 (1991), no. 2, 403416.
[4] Branson, T.

Olafsson, G. Pasquale, A., The PaleyWiener theorem and
the local Huygens principle for compact symmetric spaces, Math. (N.S.)
16 (2005), no. 34, 393428.
[5] Branson, T. and

Olafsson, G., The Paley-Wiener theorem for the Jacobi
transform and the local Huygens principle for root systems with even
multiplicities, Math. (N.S.) 16 (2005), no. 34, 429442.
[6] Branson, T.

Olafsson, G.Schlichtkrull, H., Huygens principle in Rieman-
nian symmetric spaces, Math. Ann. 301 (1995), no. 3, 445462.
[7] El Kamel, J. Yacoub, C., Huygens principle and equipartition of energy
for the modied wave equation associated to a generalized radial Laplacian,
Ann. Math Blaise Pascal 12 (2005), no. 1, 147160.
h
a
l
-
0
0
3
4
7
1
4
6
,

v
e
r
s
i
o
n

1

-

1
4

D
e
c

2
0
0
8
Ayadi 9
[8] Helgason, S., Geometric analysis on symmetric spaces, Math. Surveys
Monographs 39, Amer. Math. Soc. (1994).
[9] Opdam, E. M. , Harmonic analysis for certain representations of graded
Hecke algebras, Acta. Math. 175 (1995), 75121.
[10] Opdam, E. M. , Lecture notes on Dunkl operators for real and complex
reection groups, Math. Soc. Japan Mem. 8 (2000).
F. Ayadi
Universite de Tunis El Manar
Faculte des Sciences de Tunis
Departement de Mathematiques
2092 Tunis El Manar, Tunisie
ayfatm@yahoo.fr
&
Universite dOrleans & CNRS
Federation Denis Poisson (FR 2964)
Laboratoire MAPMO (UMR 6628)
B atiment de Mathematiques
B.P. 6759, 45067 Orleans cedex 2,
France
Fatma.Ayadi@etu.univ-orleans.fr
Received
and in nal form
h
a
l
-
0
0
3
4
7
1
4
6
,

v
e
r
s
i
o
n

1

-

1
4

D
e
c

2
0
0
8