External Ellipsoidal Harmonics for the Dunkl-Laplacian
The paper introduces external ellipsoidal and external sphero-conal h-harmonics for the Dunkl-Laplacian. These external h-harmonics admit integral representations, and they are connected by a formula of Niven's type.
Gespeichert in:
| Datum: | 2008 |
|---|---|
| 1. Verfasser: | |
| Format: | Artikel |
| Sprache: | English |
| Veröffentlicht: |
Інститут математики НАН України
2008
|
| Schriftenreihe: | Symmetry, Integrability and Geometry: Methods and Applications |
| Online Zugang: | http://dspace.nbuv.gov.ua/handle/123456789/147998 |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| Назва журналу: | Digital Library of Periodicals of National Academy of Sciences of Ukraine |
| Zitieren: | External Ellipsoidal Harmonics for the Dunkl-Laplacian / H. Volkmer // Symmetry, Integrability and Geometry: Methods and Applications. — 2008. — Т. 4. — Бібліогр.: 23 назв. — англ. |
Institution
Digital Library of Periodicals of National Academy of Sciences of Ukraine| id |
irk-123456789-147998 |
|---|---|
| record_format |
dspace |
| spelling |
irk-123456789-1479982019-02-17T01:26:51Z External Ellipsoidal Harmonics for the Dunkl-Laplacian Volkmer, H. The paper introduces external ellipsoidal and external sphero-conal h-harmonics for the Dunkl-Laplacian. These external h-harmonics admit integral representations, and they are connected by a formula of Niven's type. 2008 Article External Ellipsoidal Harmonics for the Dunkl-Laplacian / H. Volkmer // Symmetry, Integrability and Geometry: Methods and Applications. — 2008. — Т. 4. — Бібліогр.: 23 назв. — англ. 1815-0659 2000 Mathematics Subject Classification: 33C52; 35C10 http://dspace.nbuv.gov.ua/handle/123456789/147998 en Symmetry, Integrability and Geometry: Methods and Applications Інститут математики НАН України |
| institution |
Digital Library of Periodicals of National Academy of Sciences of Ukraine |
| collection |
DSpace DC |
| language |
English |
| description |
The paper introduces external ellipsoidal and external sphero-conal h-harmonics for the Dunkl-Laplacian. These external h-harmonics admit integral representations, and they are connected by a formula of Niven's type. |
| format |
Article |
| author |
Volkmer, H. |
| spellingShingle |
Volkmer, H. External Ellipsoidal Harmonics for the Dunkl-Laplacian Symmetry, Integrability and Geometry: Methods and Applications |
| author_facet |
Volkmer, H. |
| author_sort |
Volkmer, H. |
| title |
External Ellipsoidal Harmonics for the Dunkl-Laplacian |
| title_short |
External Ellipsoidal Harmonics for the Dunkl-Laplacian |
| title_full |
External Ellipsoidal Harmonics for the Dunkl-Laplacian |
| title_fullStr |
External Ellipsoidal Harmonics for the Dunkl-Laplacian |
| title_full_unstemmed |
External Ellipsoidal Harmonics for the Dunkl-Laplacian |
| title_sort |
external ellipsoidal harmonics for the dunkl-laplacian |
| publisher |
Інститут математики НАН України |
| publishDate |
2008 |
| url |
http://dspace.nbuv.gov.ua/handle/123456789/147998 |
| citation_txt |
External Ellipsoidal Harmonics for the Dunkl-Laplacian / H. Volkmer // Symmetry, Integrability and Geometry: Methods and Applications. — 2008. — Т. 4. — Бібліогр.: 23 назв. — англ. |
| series |
Symmetry, Integrability and Geometry: Methods and Applications |
| work_keys_str_mv |
AT volkmerh externalellipsoidalharmonicsforthedunkllaplacian |
| first_indexed |
2025-07-11T03:47:00Z |
| last_indexed |
2025-07-11T03:47:00Z |
| _version_ |
1837320761876938752 |
| fulltext |
Symmetry, Integrability and Geometry: Methods and Applications SIGMA 4 (2008), 091, 13 pages
External Ellipsoidal Harmonics
for the Dunkl–Laplacian?
Hans VOLKMER
Department of Mathematical Sciences, University of Wisconsin-Milwaukee,
P. O. Box 413, Milwaukee, WI 53201, USA
E-mail: volkmer@uwm.edu
URL: http://www.uwm.edu/∼volkmer/
Received September 22, 2008, in final form December 18, 2008; Published online December 23, 2008
Original article is available at http://www.emis.de/journals/SIGMA/2008/091/
Abstract. The paper introduces external ellipsoidal and external sphero-conal h-harmonics
for the Dunkl–Laplacian. These external h-harmonics admit integral representations, and
they are connected by a formula of Niven’s type. External h-harmonics in the plane are
expressed in terms of Jacobi polynomials Pα,β
n and Jacobi’s functions Qα,β
n of the second
kind.
Key words: external ellipsoidal harmonics; Stieltjes polynomials; Dunkl–Laplacian; funda-
mental solution; Niven’s formula; Jacobi’s function of the second kind
2000 Mathematics Subject Classification: 33C52; 35C10
1 Introduction
In a previous article [21] the author introduced (internal) ellipsoidal h-harmonics Fn,p and
sphero-conal h-harmonics Gn,p for the Dunkl–Laplacian. The functions Fn,p(x) and Gn,p(x)
are polynomials in the variables x = (x0, x1, . . . , xk) and they are h-harmonic, that is, they
satisfy the differential-difference equation
∆hu :=
(
D2
0 +D2
1 + · · ·+D2
k
)
u = 0, (1.1)
where Dj is defined by
Dju(x) :=
∂u(x)
∂xj
+ αj
u(x)− u(σj(x))
xj
and σj is the reflection at the jth coordinate plane. The parameters αj enter the weight function
h(x) := |x0|α0 |x1|α1 · · · |xk|αk . (1.2)
Following the book [9] by Dunkl and Xu and the paper [23] by Xu we will assume that αj ≥ 0
for each j = 0, 1, . . . , k although one would expect that the range of validity can be extended
analytically to the domain αj > −1
2 for each j. We will also exclude the case k = 1, α0 = α1 = 0
because we want the constant µ defined below in (1.5) to be positive.
The parity vector p = (p0, p1, . . . , pk) has components in {0, 1} and indicates that Fn,p
and Gn,p have parity p which means that they are sums of monomials xp0+2q0
0 xp1+2q1
1 · · ·xpk+2qk
k
with qj ∈ N0 = {0, 1, 2, . . . }. The vector n = (n1, n2, . . . , np) counts the zeros of the corres-
ponding Stieltjes quasi-polynomials inside k adjacent open intervals. If we set m = 2|n|+ |p| :=
?This paper is a contribution to the Special Issue on Dunkl Operators and Related Topics. The full collection
is available at http://www.emis.de/journals/SIGMA/Dunkl operators.html
mailto:volkmer@uwm.edu
http://www.uwm.edu/~volkmer/
http://www.emis.de/journals/SIGMA/2008/091/
http://www.emis.de/journals/SIGMA/Dunkl_operators.html
2 H. Volkmer
2(n1 + · · · + nk) + (p0 + · · · + pk) then Gn,p is a spherical h-harmonic of degree m, that is, it
is a h-harmonic homogeneous polynomial of degree m. Moreover, Fn,p is a polynomial of total
degree m which, however, is usually not homogeneous. The sphero-conal h-harmonics of fixed
degree m form an orthogonal basis in the linear space of spherical h-harmonics of degree m with
respect to the inner product
〈u, v〉 =
∫
Sk
h2(x)u(x)v(x) dS(x), (1.3)
where Sk denotes the unit sphere in Rk+1. Equation (1.3) uses the standard surface integral
over the sphere so that
∫
Sk 1dS is the surface area of Sk.
In the present paper the theory is extended to include external ellipsoidal h-harmonics Fn,p
and external sphero-conal h-harmonics Gn,p. Throughout, we will use calligraphic letters to
denote the “external” functions. The functions Gn,p and Fn,p are still solutions of (1.1) but
they are no longer polynomials. They have proper decay as ‖x‖ → ∞ which makes them
suitable for solving problems concerning the exterior of ellipsoids or spheres, respectively. The
external sphero-conal h-harmonics are easy to define. They are given by
Gn,p(x) := ‖x‖−2µ−2mGn,p(x), m := 2|n|+ |p|, (1.4)
where
µ := α0 + α1 + · · ·+ αk +
k − 1
2
. (1.5)
The definition of external ellipsoidal h-harmonics is postponed to Section 2.
After introducing the fundamental solution of (1.1) in Section 3 we obtain integral represen-
tations of Fn,p and Gn,p in terms of the fundamental solution in Section 4. Section 5 states some
integral formulas related to Section 5.2 in [9]. As a main result of this paper, in Section 6 we
prove a Niven type formula that expresses Fn,p in terms of Gn,p. In the classical situation k = 2,
α = 0, this formula was first given by Niven [16] in an amazing award winning paper. Modified
versions of some of Niven’s result are contained in the books by Whittaker and Watson [22] and
by Hobson [14]. It was the attempt to extend Niven’s results to external h-harmonics that led
to this paper. We also refer to paper [15] by Kalnins and Miller which contains a generalization
of Niven’s work from a different perspective.
In Section 7 we illustrate the results in the relatively simple but nontrivial planar case k = 1.
In this situation the internal ellipsoidal and sphero-conal h-harmonics are expressed in terms
of Jacobi polynomials P
(α,β)
n while the external ellipsoidal h-harmonic Fn,p involves the second
solution Q
(α,β)
n of the differential equation whose first solution is P
(α,β)
n .
We should mention that there is an increasing number of papers in Applied Mathematics
that use ellipsoidal harmonics; for example, see [3, 4, 5]. For applications of Dunkl operators
and the associated Laplacian to mathematical physics, special functions, probability theory and
geometry, we refer to [2, 6, 8, 13, 17, 18].
2 External ellipsoidal harmonics
We consider the Fuchsian differential equation
k∏
j=0
(t− aj)
v′′ +
k∑
j=0
αj + 1
2
t− aj
v′
+
−1
2
k∑
j=0
pjαjAj
t− aj
+
k−1∑
i=0
λit
i
v = 0 (2.1)
External Ellipsoidal Harmonics for the Dunkl–Laplacian 3
for the function v(t) where the prime denotes differentiation with respect to t, and
Aj :=
k∏
i=0
i6=j
(aj − ai).
The parameters αj are from (1.2), p = (p0, p1 . . . , pk) is a fixed parity vector,
a0 < a1 < · · · < ak (2.2)
are chosen numbers and λ0, λ1, . . . , λk−1 are real spectral parameters.
Let n = (n1, n2, . . . , nk) ∈ Nk
0. In [21, Section 2] we considered the quasi-polynomial En,p
originally introduced by Stieltjes [19] in the special case p = 0. The quasi-polynomial En,p
has nj zeros in (aj−1, aj) for each j = 1, 2, . . . , k, and it is a solution of (2.1) for special values of
the parameters λ0, λ1, . . . , λk−1. For this set of parameters, −m
2 and µ+ m
2 are the exponents of
equation (2.1) at infinity, where m and µ are according to (1.4), (1.5). Now En,p is the Frobenius
solution belonging to the exponent −m
2 . We introduce a second Frobenius solution En,p of
equation (2.1) belonging to the exponent µ + m
2 at infinity. This solution is defined for t > ak
and it is normalized by the condition that
lim
t→∞
tµ+m
2 En,p(t) = 1.
This normalization and the corresponding one of En,p leads to the Wronskian
En,p(t)E′
n,p(t)− En,p(t)E ′n,p(t) = (µ + m)
k∏
j=0
(t− aj)−αj− 1
2 (2.3)
which holds for t > ak. In the classical case k = 2, α = 0, the second solution En,p appears in
[1, § 9.7] and [11, § 100] .
We introduce ellipsoidal coordinates involving the parameters (2.2). For every (x0, . . . , xk)
in the positive cone of Rk+1
x0 > 0, . . . , xk > 0, (2.4)
its ellipsoidal coordinates t0, t1, . . . , tk lie in the intervals
ak < t0 < ∞, ai−1 < ti < ai, i = 1, . . . , k, (2.5)
and satisfy
k∑
j=0
x2
j
ti − aj
= 1 for i = 0, 1, . . . , k.
These coordinates provide a bijective mapping between the positive cone (2.4) and the cube (2.5).
As explained in [21, Section 3], equation (1.1) can be solved by the method of separation of
variables in ellipsoidal coordinates. In [21] we considered the internal ellipsoidal h-harmonic
Fn,p(x) =
k∏
j=0
En,p(tj),
where (t0, t1, . . . , tk) are ellipsoidal coordinates of x. We now introduce the external ellipsoidal
h-harmonic Fn,p by
Fn,p(x) := En,p(t0)
k∏
j=1
En,p(tj) =
En,p(t0)
En,p(t0)
Fn,p(x).
4 H. Volkmer
This function is originally defined in the positive cone (2.4) and can then be extended analytically
to the complement of the degenerate ellipsoid
k−1∑
j=0
x2
j
ak − aj
≤ 1, xk = 0. (2.6)
We see this as follows. The ellipsoidal coordinate t0 > ak is an analytic function of x for x
outside the degenerate ellipsoid (2.6). Further, En,p and En,p are analytic functions on the
interval (ak,∞), En,p has no zero there and Fn,p(x) is a polynomial in x.
We note that Fn,p has parity p, that is,
Fn,p(σj(x)) = (−1)pjFn,p(x).
Therefore, by construction, Fn,p satisfies equation (1.1). It may be appropriate here to remark
that a solution of (1.1) is always implicitly assumed to be defined on an open set which is
invariant under the reflections σj , j = 0, 1, . . . , k. Of course, our domain of definition of Fn,p
has this property.
3 The fundamental solution
We introduce the function
Φ(x, z) :=
Γ(µ)
4
∏k
j=0 Γ(αj + 1
2)
∫
[−1,1]k+1
(Ψ(x, z, τ ))−µ
k∏
j=0
cαj (1 + τj)(1− τ2
j )αj−1 dτ , (3.1)
where
Ψ(x, z, τ ) := ‖x‖2 − 2(τ0x0z0 + · · ·+ τkxkzk) + ‖z‖2 ≥ 0,
cν := (B(1
2 , ν))−1 with B denoting the Beta function, and µ is from (1.5). If αi = 0 then delete
the factor containing cαi , set τi = 1 in Ψ and omit the integration on τi. If αj = 0 for all j then
Φ reduces to the standard fundamental solution of Laplace’s equation [10, page 22].
If |xj | 6= |zj | for at least one j, then Ψ(x, z, ·) has a positive lower bound on [−1, 1]k+1.
Therefore, Φ(x, z) is analytic everywhere in Rk+1 × Rk+1 except at points where |xj | = |zj |
for all j. We recognize that Φ(x, z) = Φ(z,x), Φ(γx, γz) = |γ|−2µΦ(x, z) and Φ(x, z) > 0.
Moreover,
Φ(0, z) =
Γ(µ)
4
k∏
j=0
Γ(αj + 1
2)
‖z‖−2µ.
If z 6= 0 we can expand Φ(·, z) in a Taylor series at x = 0. Its coefficients are given by the
following lemma.
Lemma 1. If z 6= 0 then, for every (q0, q1, . . . , qk) ∈ Nk+1
0 ,
Dq0
x0
· · · Dqk
xk
(
xq0
0 · · ·xqk
k
)∂q0
x0 · · · ∂
qk
xkΦ(x, z)
q0! · · · qk!
∣∣∣∣
x=0
= Dq0
x0
· · · Dqk
xk
Φ(x, z) |x=0 = (−1)|q|Dq0
z0
· · · Dqk
zk
Φ(0, z), (3.2)
where |q| = q0 + · · ·+ qk.
External Ellipsoidal Harmonics for the Dunkl–Laplacian 5
Proof. We note that the first equality in (3.2) is trivial so it is enough to prove the second one.
The proof of the second equality involves some lengthy formulas, so we will use some common
abbreviations. If j = (j0, j1, . . . , jk) is a multi-index we set j! = j0! · · · jk!, ∂j
x = ∂j0
x0 · · · ∂
jk
xk and
xj = xj0
0 · · ·xjk
k . By Faa di Bruno’s formula (or by equation [21, (5.4)]), we write
∂q
xf(Ψ(x, z, τ )) =
∑
j
2|q|−2|j|f (|q|−|j|)(Ψ(x, z, τ ))
1
j!
∂2j
x (x0 − τ0z0)q0 · · · (xk − τkzk)qk ,
where f(u) := u−µ, and the summation is over all multi-indices j = (j0, j1, . . . , jk) with ji ≥ 0
and qi − 2ji ≥ 0, i = 0, 1, . . . , k. Computing the partial derivatives on the right-hand side and
setting x = 0 gives
∂q
xf(Ψ(x, z, τ ))|x=0
= (−1)|q|
∑
j
2|q|−2|j|f (|q|−|j|)(‖z‖2)
q!
j!(q− 2j)!
(τ0z0)q0−2j0 · · · (τkzk)qk−2jk .
We now carry out the integrations indicated in (3.1) taking into account that
cαj
∫ 1
−1
τm
j (1 + τj)(1− τ2
j )αj−1 dτj =
m!
Dm
zj
(zm
j )
.
Note that Dm
zj
(zm
j ) is a constant depending on j and m (the “Dunkl factorial”.) We obtain
∂q
xΦ̃(x, z)
∣∣∣
x=0
= (−1)|q|
∑
j
2|q|−2|j|f (|q|−|j|)(‖z‖2)
q!
j!Dq−2j
z zq−2j
zq−2j,
where Φ̃ is defined by the right-hand side of (3.1) but with the normalization factor in front of
the integral omitted. Now we use the first identity in (3.2) to obtain
Dq
xΦ̃(x, z)
∣∣∣
x=0
= (−1)|q|
∑
j
2|q|−2|j|f (|q|−|j|)(‖z‖2)
1
j!
Dq
z zq
Dq−2j
z zq−2j
zq−2j
= (−1)|q|
∑
j
2|q|−2|j|f (|q|−|j|)(‖z‖2)
1
j!
D2j
z zq.
Finally, we apply [21, Lemma 2] to the function A(v0, . . . , vk) = f(v0 + · · ·+ vk) and obtain the
second identity in (3.2). �
Since ‖z‖−2µ is h-harmonic, Lemma 1 shows that Φ(x, ·) is h-harmonic for fixed x and so,
by symmetry, Φ(·, z) is h-harmonic for fixed z.
Knowledge of the exact nature of the singularities of Φ will be immaterial in this paper. An
analysis of the singularities would be along the following lines. We see that the singularity of
Φ(x, z) at x = z is determined by integration over values τj close to 1 so it is convenient to
substitute σj = 1− τj . If all zj are nonzero we have the elementary inequality
Ψ(x, z, τ ) ≥ c(‖x− z‖2 + ‖σ‖)
if x is close enough to z and c is a positive constant; see the proof of [9, Theorem 5.5.7] for
similar estimates. It follows that Φ(·, z) and its normal derivative on spheres ‖x− z‖ = ε do not
grow faster as ε → 0 than in the classical case α = 0. Obvious modifications of the growth rate
are obtained when one or several zj vanish. If |xj | = |zj | for all j but xi 6= zi for at least one i
6 H. Volkmer
then the factor 1 + τi appearing in (3.1) comes into play and makes the singularity milder than
that at x = z.
We now express Φ in terms of the reproducing kernel Pm(x,y) of the linear space of spherical
h-harmonics of degree m with respect to the inner product (1.3); see [9, Section 5.3]. We
normalize the reproducing kernel so that
Y (x) =
∫
Sk
h2(x)Pm(x,y)Y (y) dS(y)
for every spherical h-harmonic Y of degree m and every x ∈ Sk. The generating function for
Gegenbauer polynomials [9, Definition 1.4.10] is given by
(1− 2us + s2)−µ =
∞∑
m=0
Cµ
m(u)sm. (3.3)
The convergence is uniform for −1 ≤ u ≤ 1, |s| ≤ q < 1. In (3.3) we set s = ‖x‖/‖z‖ and
u = ‖x‖−1‖z‖−1(τ0x0z0 + · · · + τkxkzk). Then (3.1) and the integral representation of the
reproducing kernel [9, Theorem 5.5.5] yields the following theorem.
Theorem 1. If ‖x‖ < ‖z‖ then
Φ(x, z) =
∞∑
m=0
1
2(µ + m)
‖x‖m‖z‖−2µ−mPm
(
x
‖x‖
,
z
‖z‖
)
. (3.4)
The convergence is uniform for ‖x‖
‖z‖ ≤ q < 1.
When k = 2, α = 0, formula (3.4) shows how to expand the reciprocal distance between z
and x into a series of Legendre polynomials. This expansion is due to Laplace and it marks the
very beginning of the history of spherical harmonics [11, page 3].
4 Integral formulas for external h-harmonics
Multiplying (3.4) by a spherical h-harmonic and integrating, we obtain an integral representation
of external spherical h-harmonics.
Theorem 2. Let Y be a spherical h-harmonic of degree m, and let Y denote the corresponding
external spherical h-harmonic Y(x) := ‖x‖−2µ−2mY (x). Then, for ‖z‖ > 1,
Y(z) = 2(µ + m)
∫
Sk
h2(x)Φ(x, z)Y (x) dS(x). (4.1)
In particular, Theorem 2 can be applied to a sphero-conal h-harmonic Y = Gn,p.
Let Y be any spherical h-harmonic of degree m. If ‖z‖ > 1 we use Green’s formula for
h-harmonic functions [9, Lemma 5.1.5] inside the unit ball to obtain
0 =
∫
Sk
h2(x)
(
Y (x)
∂Φ(x, z)
∂νx
− Φ(x, z)
∂Y (x)
∂ν
)
dS(x), (4.2)
where ν denotes the outward normal. On the unit sphere we have
∂Y
∂ν
= mY = mY,
∂Y
∂ν
= (−2µ−m)Y.
External Ellipsoidal Harmonics for the Dunkl–Laplacian 7
Therefore, adding (4.2) to (4.1) gives
Y(z) =
∫
Sk
h2(x)
(
Y(x)
∂Φ(x, z)
∂νx
− Φ(x, z)
∂Y(x)
∂ν
)
dS(x).
When α = 0 this formula corresponds exactly to the well-known representation formula of
harmonic functions in terms of a fundamental solution [10, page 34]. By using Green’s formula
in the region between two spheres we see that we also have
Y(z) =
∫
Sk
R
h2(x)
(
Y(x)
∂Φ(x, z)
∂νx
− Φ(x, z)
∂Y(x)
∂ν
)
dS(x) (4.3)
when ‖z‖ > R, where
Sk
R := {x : ‖x‖ = R}.
We introduce ellipsoidal coordinates by choosing numbers aj according to (2.2), and prove
an integral representation of external ellipsoidal h-harmonics.
Theorem 3. Let J denote the ellipsoid
J =
y :
k∑
j=0
y2
j
t− aj
= 1
(4.4)
for some fixed t > ak. Then, for all z exterior to J,
Fn,p(z) =
2(µ + m)
E2
n,p(t)
k∏
j=0
(t− aj)−αj− 1
2
∫
J
h2(y)w(y)Φ(y, z)Fn,p(y) dJ(y), (4.5)
where m = 2|n|+ |p| and w is defined by
1
w(y)
:=
k∑
j=0
y2
j
(t− aj)2
. (4.6)
Proof. We will suppress the subscripts n,p in this proof. Both sides of equation (4.5) are
analytic in the exterior of J so it will be sufficient to prove the equation for ‖z‖ > R, where R
is chosen so large that J fits inside the sphere Sk
R.
We claim that formula (4.3) is also true if we replace Y by F . To see this we expand F on
the sphere Sk
R in a series
F(x) =
∞∑
q=0
Yq(x),
where Yq is a spherical h-harmonic of degree q. Using an uniqueness theorem for h-harmonic
functions defined in the exterior of spheres with proper decay at infinity (the analogue of [7,
Theorem 2.9]) we obtain
F(x) =
∞∑
q=0
R2µ+2q‖x‖−2µ−2qYq(x).
This series and the series obtained by term-by-term differentiation converge uniformly to the
proper limits on and exterior to Sk
R. Therefore, by taking limits we obtain
F(z) =
∫
Sk
R
h2(x)
(
F(x)
∂Φ(x, z)
∂νx
− Φ(x, z)
∂F(x)
∂ν
)
dS(x).
8 H. Volkmer
Using Green’s formula in the region between J and Sk
R, we also have
F(z) =
∫
J
h2(y)
(
F(y)
∂Φ(y, z)
∂νy
− Φ(y, z)
∂F(y)
∂ν
)
dJ(y), (4.7)
where ν denotes the outward unit normal vector. We now apply Green’s formula to the internal
ellipsoidal h-harmonic F in the interior of J and obtain∫
J
h2(y)
(
F (y)
∂Φ(y, z)
∂νy
− Φ(y, z)
∂F (y)
∂ν
)
dJ(y) = 0. (4.8)
We multiply (4.7) by E(t), (4.8) by E(t) and subtract noting that E(t)F(y) = E(t)F (y) on J.
We obtain
E(t)F(z) =
∫
J
h2(y)Φ(y, z)
(
E(t)
∂F (y)
∂ν
− E(t)
∂F(y)
∂ν
)
dJ(y).
Since the ellipsoidal coordinates are orthogonal, the normal derivative is expressible in terms of
the derivative with respect to the ellipsoidal coordinate t0:
∂
∂ν
= 2w(y)
∂
∂t0
.
Therefore, we find
F(z) = 2
E(t)E′(t)− E(t)E ′(t)
E2(t)
∫
J
h2(y)w(y)Φ(y, z)F (y) dJ(y).
We replace the Wronskian by (2.3) and (4.5) is established. �
In Section 6 the integral in (4.5) will be evaluated using the formulas given in Section 5.
Theorem 3 leads to an expansion of Φ in ellipsoidal h-harmonics.
Theorem 4. Let the ellipsoidal t0-coordinate of z be larger than the corresponding one for y.
Then
Φ(y, z) =
∑
n,p
e2
n,p
2(µ + 2|n|+ |p|)
Fn,p(y)Fn,p(z), (4.9)
where the positive constants en,p are determined by
e2
n,p
∫
Sk
h2(x)G2
n,p(x) dS(x) = 1.
Proof. Let J be the ellipsoid (4.4) that contains y. The function Φ(·, z) is analytic and h-
harmonic on and inside J. Therefore, it can be expanded in ellipsoidal h-harmonics as in [21,
Section 7] and the expansion coefficients can be evaluated using (4.5). This gives (4.9) after
a simple calculation. �
When k = 2, α = 0, formula (4.9) can be found in [12, page 172].
External Ellipsoidal Harmonics for the Dunkl–Laplacian 9
5 Integral formulas for spherical h-harmonics
Theorem 5. Let f be a homogeneous polynomial of degree `, and let Y be a spherical h-harmonic
of degree m. If ` = m + 2r, r = 0, 1, 2, . . . , then
∫
Sk
h2(x)f(x)Y (x) dS(x) =
1
2`−1r!
k∏
i=0
Γ(αi + 1
2)
Γ(m + µ + r + 1)
∆r
hY (D)f(x), (5.1)
where Y (D) = Y (D0, . . . ,Dk). If ` < m or `−m is odd then the integral in (5.1) vanishes.
Proof. Theorem 5.1.15 in [9] gives the expansion of f in spherical h-harmonics on Sk. It follows
from this expansion and [9, Theorem 5.1.6] that the integral in (5.1) vanishes when ` < m or
when `−m is odd. If ` = m + 2r, r = 0, 1, 2, . . . , then [9, Theorem 5.1.15] shows that∫
Sk
h2(x)f(x)Y (x) dS(x) =
∫
Sk
h2(x)f̃(x)Y (x) dS(x), (5.2)
where f̃ is a homogeneous polynomial of degree m given by
f̃(x) =
1
4rr!(m + µ + 1)r
∆r
hf(x).
We evaluate the integral on the right-hand side of (5.2) by applying Theorem 5.2.4 in [9]. We
obtain∫
Sk
h2(x)f(x)Y (x) dS(x) =
1
c′2m4rr!(µ + 1)m(m + µ + 1)r
∆r
hY (D)f(x),
where
1
c′
=
∫
Sk
h2(x) dS(x) =
2
k∏
i=0
Γ(αi + 1
2)
Γ(µ + 1)
.
After simplification of the constant factors we arrive at the desired equation (5.1). �
In the special case k = 2, α = 0 formula (5.1) is due to Hobson [14, § 100].
We state a consequence of Theorem 1.
Corollary 1. Let f(x) =
∞∑̀
=0
f`(x), where f` is a homogeneous polynomial of degree `, and the
convergence is uniform on Sk. Let Y be a spherical h-harmonic of degree m. Then
∫
Sk
h2(x)f(x)Y (x) dS(x) =
∞∑
r=0
1
2m+2r−1r!
k∏
i=0
Γ(αi + 1
2)
Γ(m + µ + r + 1)
∆r
hY (D)fm+2r(x). (5.3)
6 Niven’s formula for external ellipsoidal h-harmonics
We consider the ellipsoid J from (4.4) with semi-axes dj :=
√
t− aj . By substituting yj = djxj ,
y ∈ J is transformed to x ∈ Sk. If f(x) is a continuous function on Sk, then
g(y) := f
(
y0
d0
, . . . ,
yk
dk
)
(6.1)
10 H. Volkmer
becomes a continuous function on J and∫
Sk
f(x) dS(x) =
k∏
j=0
d−1
j
∫
J
w(y)g(y) dJ(y) (6.2)
with w(y) from (4.6).
Let Y denote a spherical h-harmonic of degree m, and let f be as in Corollary 1. We use the
substitution yj = djxj and (6.2) to write (5.3) as∫
J
h2(y)w(y)g(y)Y
(
y0
d0
, . . . ,
yk
dk
)
dJ(y)
=
k∏
j=0
dj
2αj+1
∞∑
r=0
1
2m+2r−1r!
k∏
j=0
Γ(αj + 1
2)
Γ(m + µ + r + 1)
ΛrY (d0D0, . . . , dkDk)gm+2r(x), (6.3)
where Λ is the operator
Λ := d2
0D2
0 + · · ·+ d2
kD2
k
and g is defined by (6.1) with the same connection between f` and g`. We use (6.3) in the
special case Y = Gn,p, m = 2|n| + |p|, and g = Φ(·, z), where ‖z‖ > d0 (the largest semi-axis
of J). We also note that
En,p(t)Gn,p
(
y0
d0
, . . . ,
yk
dk
)
= Fn,p(y),
see [21, (6.1)]. Then the integral representation (4.5) for the external ellipsoidal h-harmonic Fn,p
and (6.3) yield
Fn,p(z) =
2(µ + m)
En,p(t)
∞∑
r=0
1
2m+2r−1r!
k∏
j=0
Γ(αj + 1
2)
Γ(m + µ + r + 1)
× ΛrGn,p(d0Dx0 , . . . , dkDxk
)Φm+2r(x, z).
We now use the identity
Gn,p(d0D0, . . . , dkDk)g(x) = En,p(t)Gn,p(D)g(x)
which holds for any h-harmonic function g; see the formula before (6.3) in [21]. Then we replace
the operator −Λ by a0D2
0 + · · ·+ akD2
k which is possible because Φ(·, z) is h-harmonic. Finally,
we use Lemma 1 and arrive at the following theorem.
Theorem 6. The external ellipsoidal h-harmonic Fn,p admits the differential-difference repre-
sentation
Fn,p(z) =
∞∑
r=0
(−1)r+m(µ + m)
2m+2rr!(µ)m+r+1
(
a0D2
0 + · · ·+ akD2
k
)r
Gn,p(D)‖z‖−2µ, (6.4)
where m = 2|n|+ |p| and µ is defined in (1.5). The expansion is valid for ‖z‖2 > ak − a0.
External Ellipsoidal Harmonics for the Dunkl–Laplacian 11
Formula (6.4) is true for all z with ‖z‖2 > ak − a0 because we can choose the number t
defining the ellipsoid (4.4) very close to ak.
We can rewrite (6.4) slightly by using the connection between internal and external spherical
h-harmonics [21, Corollary 2]:
Y(x) =
(−1)m
2m(µ)m
Y (D)‖x‖−2µ.
This leads to the following corollary.
Corollary 2. The external ellipsoidal h-harmonic Fn,p admits the differential-difference repre-
sentation
Fn,p(z) =
∞∑
r=0
(−1)r
22rr!(µ + m + 1)r
(
a0D2
0 + · · ·+ akD2
k
)rGn,p(z) (6.5)
for ‖z‖2 > ak − a0.
7 Examples
We illustrate some results of this paper in the planar case k = 1. We choose a0 = −1, a1 = 1
and set α := α1 − 1
2 , β := α0 − 1
2 . If p0 = p1 = 0 equation (2.1) becomes
(t2 − 1)
(
v′′ +
α + 1
t− 1
v′ +
β + 1
t + 1
v′
)
+ λ0v = 0. (7.1)
This equation is satisfied by the Jacobi polynomial v(t) = P
(α,β)
n (t) when λ0 = −n(n+α+β+1).
Therefore, the Stieltjes polynomial En,0 is a multiple of P
(α,β)
n . In [21] the Stieltjes polynomials
were normalized such that their leading coefficient equals 1. This gives
En,0 = anP (α,β)
n ,
where
an = a(α,β)
n :=
n!2n
(α + β + n + 1)n
.
There are similar formulas for the other three parity vectors p; compare [21, § 2].
A second solution of (7.1) with λ0 = −n(n+α+β +1) is Q
(α,β)
n ; see [20, § 4.61]. Comparing
the behavior at infinity, we see that
En,0 = bnQ(α,β)
n ,
where
bn = b(α,β)
n := 2−n−α−β Γ(2n + α + β + 2)
Γ(n + α + 1)Γ(n + β + 1)
.
Again, similar formulas hold for the other three parity vectors p.
The planar sphero-conal coordinates r, s1 as defined in [21, § 4] essentially agree with polar
coordinates x0 = r cos φ, x1 = r sinφ. The connection is given by cos(2φ) = s1. The internal
sphero-conal h-harmonic Gn,0 is equal to r2nEn,0(s1) in sphero-conal coordinates. Therefore, in
cartesian coordinates, we obtain
Gn,0(x0, x1) = an
(
x2
0 + x2
1
)n
P (α,β)
n
(
x2
0 − x2
1
x2
0 + x2
1
)
.
12 H. Volkmer
The external sphero-conal h-harmonic Gn,0 is given by
Gn,0(x0, x1) = an
(
x2
0 + x2
1
)−n−α−β−1
P (α,β)
n
(
x2
0 − x2
1
x2
0 + x2
1
)
. (7.2)
The internal ellipsoidal h-harmonic Fn,0 takes the form En,0(t0)En,0(t1) in ellipsoidal coordi-
nates. We can express t0, t1 explicitly as functions of x0, x1 as follows
t0 = 1
2
(
x2
0 + x2
1
)2 +
(
1
4
(
x2
0 + x2
1
)2 − x2
0 + x2
1 + 1
)1/2
, (7.3)
t1 = 1
2
(
x2
0 + x2
1
)2 −
(
1
4
(
x2
0 + x2
1
)2 − x2
0 + x2
1 + 1
)1/2
. (7.4)
It should be mentioned that analogous formulas do not exist when k ≥ 4 and would be very
complicated when k = 2 or k = 3. Now internal ellipsoidal h-harmonics can be written as
Fn,0(x0, x1) = a2
nP (α,β)
n (t0)P (α,β)
n (t1)
with t0, t1 given in (7.3), (7.4), respectively. Moreover, external ellipsoidal h-harmonics assume
the form
Fn,0(x0, x1) = anbnQ(α,β)
n (t0)P (α,β)
n (t1). (7.5)
Formula (7.5) is meaningful for all (x0, x1) which do not lie on the common focal line −
√
2 ≤
x0 ≤
√
2, x1 = 0 of the confocal ellipses determined by a constant value of the ellipsoidal
coordinate t0.
Formula (6.5) now expresses the function (7.5) in terms of the function (7.2) as follows.
Theorem 7. Let n ∈ N0. For every (x0, x1) ∈ R2 with x2
0 + x2
1 > 2 we have
bnQ(α,β)
n (t0)P (α,β)
n (t1)
=
∞∑
r=0
(−1)r
22rr!(2n + α + β + 2)r
(
D2
1 −D2
0
)r(
x2
0 + x2
1
)−n−α−β−1
P (α,β)
n
(
x2
0 − x2
1
x2
0 + x2
1
)
, (7.6)
where on the left-hand side t0, t1 are given by (7.3), (7.4), respectively.
Obviously, we could replace D2
1 − D2
0 by 2D2
1 on the right-hand side of (7.6). Formula (7.6)
expands the left-hand side in a series of homogeneous functions whose degree decreases with r.
The convergence will be especially good when x2
0 +x2
1 is large. The formula can be checked with
symbolic computer algebra by replacing xj by uxj on the left-hand side and then expanding in
an asymptotic series at u = ∞. Computer algebra confirmed the correctness of (7.6) for small
values of n. The natural question arises whether this formula or the more general formula (6.5)
can be proved more directly, that is, without using any integrals in the proof.
References
[1] Arscott F.M., Periodic differential equations. An introduction to Mathieu, Lamé, and allied functions,
International Series of Monographs in Pure and Applied Mathematics, Vol. 66, A Pergamon Press Book,
The Macmillan Co., New York, 1964.
[2] Ben Säıd S., Ørsted B., The wave equation for Dunkl operators, Indag. Math. (N.S.) 16 (2005), 351–391.
[3] Blimke J., Myklebust J., Volkmer H., Merrill S., Four-shell ellipsoidal model employing multipole expansion
in ellipsoidal coordinates, Med. Biol. Eng. Comput. 46 (2008), 859–869.
[4] Dassios G., The magnetic potential for the ellipsoidal MEG problem, J. Comput. Math. 25 (2007), 145–156.
External Ellipsoidal Harmonics for the Dunkl–Laplacian 13
[5] Dassios G., Kariotou F., Magnetoencephalography in ellipsoidal geometry, J. Math. Phys. 44 (2003), 220–
241.
[6] de Jeu M.F.E., The Dunkl transform, Invent. Math. 113 (1993), 147–162.
[7] Dunkl C.F., Reflection groups and orthogonal polynomials on the sphere, Math. Z. 197 (1988), 33–60.
[8] Dunkl C.F., A Laguerre polynomial orthogonality and the hydrogen atom, Anal. Appl. (Singap.) 1 (2003),
177–188, math-ph/0011021.
[9] Dunkl C.F., Xu Y., Orthogonal polynomials of several variables, Encyclopedia of Mathematics and Its
Applications, Vol. 81, Cambridge University Press, Cambridge, 2001.
[10] Evans L., Partial differential equations, Graduate Studies in Mathematics, Vol. 19, American Mathematical
Society, Providence, RI, 1998.
[11] Heine E., Handbuch der Kugelfunktionen, Vol. 1, G. Reimer Verlag, Berlin, 1878.
[12] Heine E., Handbuch der Kugelfunktionen, Vol. 2, G. Reimer Verlag, Berlin, 1881.
[13] Hikami K., Boundary K-matrix, elliptic Dunkl operator and quantum many-body systems, J. Phys. A:
Math. Gen. 29 (1996), 2135–2147.
[14] Hobson E.W., The theory of spherical and ellipsoidal harmonics, Cambridge University Press, Cambridge,
1931.
[15] Kalnins E.G., Miller W. Jr., Jacobi elliptic coordinates, functions of Heun and Lamé type and the Niven
transform, Regul. Chaotic Dyn. 10 (2005), 487–508.
[16] Niven W.D., On ellipsoidal harmonics, Phil. Trans. Royal Society London A 182 (1891), 231–278.
[17] Rösler M., Dunkl operators: theory and applications, in Orthogonal Polynomials and Special Functions
(Leuven, 2002), Lecture Notes in Math., Vol. 1817, Springer, Berlin, 2003, 93–135, math.CA/0210366.
[18] Rösler M., Voit M., Markov processes related with Dunkl operators, Adv. in Appl. Math. 21 (1998), 575–643.
[19] Stieltjes T.J., Sur certains polynômes qui vérifient une équation différentielle linéaire du second ordre et sur
la théorie des fonctions de Lamé, Acta Math. 5 (1885), 321–326.
[20] Szegö G., Orthogonal polynomials, 4th ed., American Mathematical Society, Providence, R.I., 1975.
[21] Volkmer H., Generalized ellipsoidal and sphero-conal harmonics, SIGMA 2 (2006), 071, 16 pages,
math.CA/0610718.
[22] Whittaker E.T., Watson G.N., A course in modern analysis, Cambridge University Press, Cambridge, 1927.
[23] Xu Y., Orthogonal polynomials for a family of product weight functions on the spheres, Canad. J. Math.
49 (1997), 175–192.
http://arxiv.org/abs/math-ph/0011021
http://arxiv.org/abs/math.CA/0210366
http://arxiv.org/abs/math.CA/0610718
1 Introduction
2 External ellipsoidal harmonics
3 The fundamental solution
4 Integral formulas for external h-harmonics
5 Integral formulas for spherical h-harmonics
6 Niven's formula for external ellipsoidal h-harmonics
7 Examples
References
|