Fourier, Gegenbauer and Jacobi Expansions for a Power-Law Fundamental Solution of the Polyharmonic Equation and Polyspherical Addition Theorems
We develop complex Jacobi, Gegenbauer and Chebyshev polynomial expansions for the kernels associated with power-law fundamental solutions of the polyharmonic equation on d-dimensional Euclidean space. From these series representations we derive Fourier expansions in certain rotationally-invariant co...
Gespeichert in:
| Datum: | 2013 |
|---|---|
| 1. Verfasser: | |
| Format: | Artikel |
| Sprache: | English |
| Veröffentlicht: |
Інститут математики НАН України
2013
|
| Schriftenreihe: | Symmetry, Integrability and Geometry: Methods and Applications |
| Online Zugang: | http://dspace.nbuv.gov.ua/handle/123456789/149201 |
| 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: | Fourier, Gegenbauer and Jacobi Expansions for a Power-Law Fundamental Solution of the Polyharmonic Equation and Polyspherical Addition Theorems / H.S. Cohl // Symmetry, Integrability and Geometry: Methods and Applications. — 2013. — Т. 9. — Бібліогр.: 35 назв. — англ. |
Institution
Digital Library of Periodicals of National Academy of Sciences of Ukraine| id |
irk-123456789-149201 |
|---|---|
| record_format |
dspace |
| spelling |
irk-123456789-1492012019-02-20T01:25:46Z Fourier, Gegenbauer and Jacobi Expansions for a Power-Law Fundamental Solution of the Polyharmonic Equation and Polyspherical Addition Theorems Cohl, H.S. We develop complex Jacobi, Gegenbauer and Chebyshev polynomial expansions for the kernels associated with power-law fundamental solutions of the polyharmonic equation on d-dimensional Euclidean space. From these series representations we derive Fourier expansions in certain rotationally-invariant coordinate systems and Gegenbauer polynomial expansions in Vilenkin's polyspherical coordinates. We compare both of these expansions to generate addition theorems for the azimuthal Fourier coefficients. 2013 Article Fourier, Gegenbauer and Jacobi Expansions for a Power-Law Fundamental Solution of the Polyharmonic Equation and Polyspherical Addition Theorems / H.S. Cohl // Symmetry, Integrability and Geometry: Methods and Applications. — 2013. — Т. 9. — Бібліогр.: 35 назв. — англ. 1815-0659 2010 Mathematics Subject Classification: 35A08; 31B30; 31C12; 33C05; 42A16 DOI: http://dx.doi.org/10.3842/SIGMA.2013.042 http://dspace.nbuv.gov.ua/handle/123456789/149201 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 |
We develop complex Jacobi, Gegenbauer and Chebyshev polynomial expansions for the kernels associated with power-law fundamental solutions of the polyharmonic equation on d-dimensional Euclidean space. From these series representations we derive Fourier expansions in certain rotationally-invariant coordinate systems and Gegenbauer polynomial expansions in Vilenkin's polyspherical coordinates. We compare both of these expansions to generate addition theorems for the azimuthal Fourier coefficients. |
| format |
Article |
| author |
Cohl, H.S. |
| spellingShingle |
Cohl, H.S. Fourier, Gegenbauer and Jacobi Expansions for a Power-Law Fundamental Solution of the Polyharmonic Equation and Polyspherical Addition Theorems Symmetry, Integrability and Geometry: Methods and Applications |
| author_facet |
Cohl, H.S. |
| author_sort |
Cohl, H.S. |
| title |
Fourier, Gegenbauer and Jacobi Expansions for a Power-Law Fundamental Solution of the Polyharmonic Equation and Polyspherical Addition Theorems |
| title_short |
Fourier, Gegenbauer and Jacobi Expansions for a Power-Law Fundamental Solution of the Polyharmonic Equation and Polyspherical Addition Theorems |
| title_full |
Fourier, Gegenbauer and Jacobi Expansions for a Power-Law Fundamental Solution of the Polyharmonic Equation and Polyspherical Addition Theorems |
| title_fullStr |
Fourier, Gegenbauer and Jacobi Expansions for a Power-Law Fundamental Solution of the Polyharmonic Equation and Polyspherical Addition Theorems |
| title_full_unstemmed |
Fourier, Gegenbauer and Jacobi Expansions for a Power-Law Fundamental Solution of the Polyharmonic Equation and Polyspherical Addition Theorems |
| title_sort |
fourier, gegenbauer and jacobi expansions for a power-law fundamental solution of the polyharmonic equation and polyspherical addition theorems |
| publisher |
Інститут математики НАН України |
| publishDate |
2013 |
| url |
http://dspace.nbuv.gov.ua/handle/123456789/149201 |
| citation_txt |
Fourier, Gegenbauer and Jacobi Expansions for a Power-Law Fundamental Solution of the Polyharmonic Equation and Polyspherical Addition Theorems / H.S. Cohl // Symmetry, Integrability and Geometry: Methods and Applications. — 2013. — Т. 9. — Бібліогр.: 35 назв. — англ. |
| series |
Symmetry, Integrability and Geometry: Methods and Applications |
| work_keys_str_mv |
AT cohlhs fouriergegenbauerandjacobiexpansionsforapowerlawfundamentalsolutionofthepolyharmonicequationandpolysphericaladditiontheorems |
| first_indexed |
2025-07-12T21:37:59Z |
| last_indexed |
2025-07-12T21:37:59Z |
| _version_ |
1837478746017234944 |
| fulltext |
Symmetry, Integrability and Geometry: Methods and Applications SIGMA 9 (2013), 042, 26 pages
Fourier, Gegenbauer and Jacobi Expansions
for a Power-Law Fundamental Solution
of the Polyharmonic Equation
and Polyspherical Addition Theorems
Howard S. COHL
Applied and Computational Mathematics Division, National Institute of Standards
and Technology, Gaithersburg, MD, 20899-8910, USA
E-mail: howard.cohl@nist.gov
URL: http://hcohl.sdf.org
Received November 29, 2012, in final form May 28, 2013; Published online June 05, 2013
http://dx.doi.org/10.3842/SIGMA.2013.042
Abstract. We develop complex Jacobi, Gegenbauer and Chebyshev polynomial expansions
for the kernels associated with power-law fundamental solutions of the polyharmonic equa-
tion on d-dimensional Euclidean space. From these series representations we derive Fourier
expansions in certain rotationally-invariant coordinate systems and Gegenbauer polynomial
expansions in Vilenkin’s polyspherical coordinates. We compare both of these expansions
to generate addition theorems for the azimuthal Fourier coefficients.
Key words: fundamental solutions; polyharmonic equation; Jacobi polynomials; Gegenbauer
polynomials; Chebyshev polynomials; eigenfunction expansions; separation of variables; ad-
dition theorems
2010 Mathematics Subject Classification: 35A08; 31B30; 31C12; 33C05; 42A16
1 Introduction
We have developed a technique for constructing addition theorems for the azimuthal Fourier
coefficients of fundamental solutions for linear homogeneous partial differential equations on
d-dimensional isotropic Riemannian manifolds. For a fundamental solution, we construct azi-
muthal Fourier expansions and compare them with eigenfunction expansions in rotationally-
invariant coordinate systems. The construction of eigenfunction expansions for fundamental
solutions in separable coordinate systems is in general non-trivial.
In connection with the Laplace operator, we have already constructed some of these addition
theorems on R3 [5, 10, 11], where we have treated the spherical, cylindrical, oblate spheroidal,
prolate spheroidal, parabolic, bispherical and toroidal coordinate systems. One may construct
addition theorems in this manner in any rotationally-invariant coordinate system which yields
solutions through separation of variables for the Laplace equation. In a similar setting, addition
theorems may be generated for other inhomogeneous linear partial differential equations, such as
for the Helmholtz, wave and heat equations in arbitrary dimensions. Furthermore an extension
of this concept is possible when working with linear partial differential operators on Riemannian
manifolds, such as for the Laplace–Beltrami operator (see [9, § 5.1]). Once a Fourier expansion
for a fundamental solution is obtained for a partial differential operator on a Riemannian mani-
fold, one must construct eigenfunction expansions for a fundamental solution corresponding to
that operator and identify those nested multi-summation and multi-integration eigenfunction
expansions which correspond to the Fourier coefficients for that operator.
mailto:howard.cohl@nist.gov
http://hcohl.sdf.org
http://dx.doi.org/10.3842/SIGMA.2013.042
2 H.S. Cohl
In this paper, we apply this technique to generate addition theorems from eigenfunction ex-
pansions for a fundamental solution of the polyharmonic operator on d-dimensional Euclidean
space Rd in Vilenkin–Kuznetsov–Smorodinskĭı polyspherical coordinate systems [32, § 9.5], [34],
[33, § 10.5] (hereafter Vilenkin). We have computed azimuthal Fourier expansions for a fun-
damental solution of this operator, as well as the corresponding eigenfunction expansions in
polyspherical coordinates. In each case, the comparison of these two expansions yields new
addition theorems for the azimuthal Fourier coefficients.
The main results of this paper are connected with closed-form expressions of hypergeometric
orthogonal polynomial expansions for a power-law fundamental solution of the polyharmonic
equation on d-dimensional Euclidean space. These expansions are the fundamental building
blocks for algorithms to compute the solution of inhomogeneous linear polyharmonic boundary
value problems through convolution with the source distribution. These problems are ubiqui-
tous in physics and engineering and include elasticity, electrostatics, magnetostatics, quantum
direct and exchange Coulomb interactions, Newtonian gravity, and potential fluid and heat flow,
just to name a few. The expansions presented in this paper are crucial for obtaining analytic
polyharmonic solutions and for numerical algorithms where expansions are needed for so-called
“fast algorithms”. Applications of generalized Hopf coordinates, one of the polyspherical co-
ordinate systems we study in this paper include particle physics [24], quantum field theory [2]
and cosmology [22]. Furthermore, expansions in hyperspherical coordinate systems have many
applications including general atomic multibody theory (see [14, 23] and references therein).
From a global analytic partial differential equation perspective, the most important results
of this paper are contained in Corollaries 2 and 4. These formulae represent multipole and
azimuthal Fourier decompositions for arbitrary powers of the Euclidean distance between two
points. They can be used to analytically and numerically solve, in a rapidly convergent fashion,
the inhomogeneous linear polyharmonic equation for isolated non-axisymmetric source distribu-
tions. Rapid convergence of the Fourier expansions is provided by the fact that for each azimuthal
mode there corresponds an infinite number of meridional modes, which are all summed over in
our expansions. Furthermore, if pure trigonometric azimuthal dependence exists for a particular
source distribution in the inhomogeneous partial differential equation, then Corollary 4 provides
a solution in a finite number of terms. In the case of an axisymmetric source distribution, the
inhomogeneous polyharmonic solution is obtained from a single m = 0 term in the expansion.
Corollary 2 is powerful in that it provides a mechanism for determining the multipole moments
associated with unrestricted powers of the distance between two points. This is in contrast with
the Gegenbauer generating function (which is generalized by Theorem 1 and Corollary 1) which
provides with the addition theorem for hyperspherical harmonics (B.13) a multipole expansion
for this kernel only for powers of the distance given by ν = 2− d, for d = 3, 4, 5, . . ..
From a special function theoretic perspective, the new results presented in this paper are
Theorem 1 and Corollary 1. These series expansions represent fundamental generalizations of
Heine’s formula [27, (14.28.2)], Gegenbauer’s generating function [27, (18.12.4)], and Heine’s
reciprocal square root identity [8, (3.11)]. Formula (A.14) which provides a connection between
the symmetric Jacobi function of the second kind and the associated Legendre function of
the second kind, is also interesting. As far as the author is aware, this has not previously
appeared in the literature. The addition theorems for associated Legendre functions given by
Theorems 2 and 3, and their Corollaries 5, 6, 7, also appear to be new. This paper only uses
eigenfunction expansions for a power-law fundamental solution of the polyharmonic equation in
Vilenkin’s polyspherical coordinates to obtain new addition theorems for the azimuthal Fourier
coefficients. We have only treated two different types of Vilenkin’s polyspherical coordinates.
In higher dimensions, many more types may be considered. The azimuthal Fourier expansion
presented in this paper, Corollary 4, can be used to provide new addition theorems for the
azimuthal Fourier coefficients in every rotationally-invariant coordinate system which separates
Fourier, Gegenbauer and Jacobi Expansions 3
the polyharmonic equation on Rd. These rotationally-invariant coordinate systems include those
of cylindrical, parabolic, and cyclidic type.
In this paper, we take advantage of previously derived closed-form expressions for the sepa-
rated eigenfunctions in Vilenkin’s polyspherical coordinates (found in [19, 33] and elsewhere) to
derive addition theorems from a power-law fundamental solution of the polyharmonic operator in
Euclidean space Rd. These addition theorems separate the complicated geometrically-relevant
quantity (the azimuthal Fourier coefficients) into functions of the individual variables in the
problem. We study all dimensions d ≥ 3 with an emphasis on the simplicity/explicitness of the
low-dimensional examples.
This paper is organized as follows. In Section 2, we describe the kernels associated with
a fundamental solution of the polyharmonic equation on Euclidean space Rd for d ≥ 2 and in-
troduce rotationally-invariant coordinate systems. In Section 3, we prove several new theorems
associated with Jacobi, Gegenbauer, and Chebyshev polynomial expansions for the kernels asso-
ciated with power-law fundamental solutions of the polyharmonic equation on Rd. In Section 4,
we derive and discuss new addition theorems in Vilenkin’s polyspherical coordinates for the
azimuthal Fourier coefficients of a fundamental solution for the polyharmonic equation on Rd.
In Appendix A, we summarize the definitions and properties of the special functions and ortho-
gonal polynomials that we use. In Appendix B, we review Vilenkin’s polyspherical coordinates
and the corresponding normalized hyperspherical harmonics.
Throughout this paper we rely on the following definitions. Let a1, a2, a3, . . . ∈ C, with C
being the set of complex numbers. If i, j ∈ Z and j < i, then
j∑
n=i
an = 0 and
j∏
n=i
an = 1. The
set of natural numbers is given by N := {1, 2, 3, . . .}, the set N0 := {0, 1, 2, . . .} = N ∪ {0}, and
Z := {0,±1,±2, . . .}. The sets Q and R represents the rational and real numbers respectively.
For d ∈ N, we denote by Rd, the finite-dimensional vector space, d-dimensional Euclidean space.
Furthermore, if x,x′ ∈ Rd then the Euclidean inner product (·, ·) : Rd × Rd → R defined by
(x,x′) := x1x
′
1 + · · ·+ xdx
′
d, (1.1)
induces a norm (the Euclidean norm) ‖ · ‖ : Rd → [0,∞), on Rd, given by ‖x‖ :=
√
(x,x).
Please see Appendix A for all notations used in this paper for special functions and orthogonal
polynomials.
2 Fundamental solution of the polyharmonic equation
in rotationally-invariant and polyspherical coordinate systems
In Euclidean space Rd, let the Laplacian operator ∆ : Cp(Rd)→ Cp−2(Rd) for p ≥ 2 be defined
by ∆ := ∂2
∂x21
+ · · ·+ ∂2
∂x2d
. If Φ : Rd → R satisfies the polyharmonic equation given by
(−∆)kΦ(x) = 0, (2.1)
where x ∈ Rd, k ∈ N and Φ ∈ C2k(Rd), then Φ is called polyharmonic. We use the nonnegative
Laplacian −∆ ≥ 0. The inhomogeneous polyharmonic equation is given by
(−∆)kΦ(x) = ρ(x), (2.2)
where we take ρ to be an integrable function so that a solution to (2.2) exists. A fundamental
solution for the polyharmonic equation on Rd is a function Gdk : (Rd×Rd)\{(x,x) : x ∈ Rd} → R
which satisfies the equation
(−∆)kGdk(x,x′) = δ(x− x′),
4 H.S. Cohl
where δ is the Dirac delta function (generalized function/distribution) and x′ ∈ Rd. Note that
this equation is satisfied in the sense of distributions.
A fundamental solution of the polyharmonic equation is given as follows (see for instance [4],
[15, p. 202], [28, p. 45]). Let d, k ∈ N. Define Gdk : (Rd × Rd) \ {(x,x) : x ∈ Rd} → R by
Gdk(x,x′) :=
(−1)k+d/2+1 ‖x− x′‖2k−d
(k − 1)! (k − d/2)!22k−1πd/2
(
log ‖x− x′‖ − βk−d/2,d
)
if d even, k ≥ d/2,
Γ(d/2− k)‖x− x′‖2k−d
(k − 1)!22kπd/2
otherwise,
(2.3)
where βp,d ∈ Q is defined as βp,d := 1
2
[
Hp +Hd/2+p−1 −Hd/2−1
]
, with Hj ∈ Q being the jth
harmonic number Hj :=
j∑
i=1
1
i . The gamma function Γ : C\−N0 → C, is a natural generalization
of the factorial function. Concerning the logarithmic contribution for d even, k ≥ d/2, the
polynomial ‖x − x′‖2k−d is polyharmonic, so any choice for the constant βp,d is valid. Our
choice for this constant is given such that −∆Gdk = Gdk−1 is satisfied for all k > d/2, and that
for k = d/2, the constant vanishes. Note that a solution of the inhomogeneous polyharmonic
equation (2.2) is obtained from Gdk via a convolution.
2.1 Fundamental solution decompositions
in rotationally-invariant coordinate systems
In this paper we restrict our attention to separable rotationally-invariant coordinate systems for
the polyharmonic equation on Rd which are given by
x1 = R(ξ1, . . . , ξd−1) cosφ,
x2 = R(ξ1, . . . , ξd−1) sinφ,
x3 = x3(ξ1, . . . , ξd−1),
· · · · · · · · · · · · · · · · · · · · · · · ·
xd = xd(ξ1, . . . , ξd−1).
(2.4)
These coordinate systems are described by d-coordinates: an angle φ ∈ [0, 2π) plus (d − 1)-
curvilinear coordinates (ξ1, . . . , ξd−1). Rotationally-invariant coordinate systems parametrize
points on the (d−1)-dimensional half-hyperplane given by φ = const and R ≥ 0 using the curvi-
linear coordinates (ξ1, . . . , ξd−1). A separable rotationally-invariant coordinate system trans-
forms the polyharmonic equation into a set of d-uncoupled ordinary differential equations with
separation constants m ∈ Z and kj ∈ R for 1 ≤ j ≤ d− 2. For a separable rotationally-invariant
coordinate system, this uncoupling is accomplished, in general, by assuming a product solution
to (2.1) of the form Φ(x) = eimφR(ξ1, . . . , ξd−1)
d−1∏
i=1
Ai(ξi,m, k1, . . . , kd−2), where the properties
of the functions R and Ai, for 1 ≤ i ≤ d− 1, and the constants kj for 1 ≤ j ≤ d− 2, depend on
the specific separable rotationally-invariant coordinate system in question. Separable coordinate
systems are divided into two different classes, those which are simply separable (R = const),
and those which are R-separable (see [26]).
The Euclidean distance between two points x,x′ ∈ Rd expressed in the rotationally-invariant
coordinate system described in (2.4) is
‖x− x′‖ =
√
2RR′
[
χ− cos(φ− φ′)
]1/2
, (2.5)
Fourier, Gegenbauer and Jacobi Expansions 5
where the toroidal parameter χ is
χ :=
R2 +R′2 +
d∑
i=3
(xi − x′i)2
2RR′
. (2.6)
The hypersurfaces χ = const are independent of coordinate system and represent hypertori of
revolution.
We now rewrite (2.3) in terms of the rotationally-invariant coordinate system (2.4).
From (2.3) we see that, apart from multiplicative constants, the expression ldk : (Rd×Rd)\{(x,x) :
x ∈ Rd} → R of a fundamental solution for the polyharmonic equation on Rd for d even, k ≥ d/2,
is given by
ldk(x,x
′) := ‖x− x′‖2k−d
(
log ‖x− x′‖ − βk−d/2,d
)
.
By expressing ldk in a rotationally-invariant coordinate system (2.4) we obtain
ldk(x,x
′) =
(
2RR′
)p [1
2
log
(
2RR′
)
− βp,d
] [
χ− cos(φ− φ′)
]p
+
1
2
(
2RR′
)p [
χ− cos(φ− φ′)
]p
log
[
χ− cos(φ− φ′)
]
, (2.7)
where p = k − d/2 ∈ N0. Similarly, when working on an even-dimensional Euclidean space Rd
with 1 ≤ k ≤ d/2 − 1, a fundamental solution of the polyharmonic equation hdk : (Rd × Rd) \
{(x,x) : x ∈ Rd} → (0,∞) is
hdk(x,x
′) := ‖x− x′‖2k−d.
By expressing hdk in a rotationally-invariant coordinate system we obtain
hdk(x,x
′) =
(
2RR′
)−q [
χ− cos(φ− φ′)
]−q
, (2.8)
where q = 2k − d ∈ N.
Examining (2.7) and (2.8), we see that for computation of Fourier expansions about the
azimuthal separation angle (φ − φ′) of ldk and hdk, all that is required is to compute the Fourier
cosine series for the following three functions fχ, hχ : R→ (0,∞) and gχ : R→ R defined as
fχ(ψ) := (χ− cosψ)p , gχ(ψ) := (χ− cosψ)p log (χ− cosψ) ,
hχ(ψ) := (χ− cosψ)−q ,
where p ∈ N0, q ∈ N and χ > 1 is a fixed parameter.
The Fourier series of fχ is given in [8] (cf. (4.4) therein)1, namely for p ∈ N0,
(z − x)p = (z2 − 1)p/2
p∑
n=0
εn(−p)n(p− n)!
(p+ n)!
Pnp
(
z√
z2 − 1
)
Tn(x), (2.9)
where εn ∈ {1, 2} is the Neumann factor defined by εn := 2− δn,0, δn,0 ∈ {0, 1} is the Kronecker
delta. The Fourier series of hχ is given in [8, (4.5)], namely for p ∈ N,
1
(z − x)p
=
(z2 − 1)−p/2
(p− 1)!
∞∑
n=0
εn(n+ p− 1)!P−np−1
(
z√
z2 − 1
)
Tn(x). (2.10)
In order to compute Fourier expansion of ldk (2.7) in separable rotationally-invariant coordinate
systems, all that remains is to determine the Fourier series of gχ (see [6]). A discussion of
Fourier cosine expansions for a logarithmic fundamental solution of the polyharmonic equation
on Rd (from gχ) can be found in [7]. The corresponding Gegenbauer polynomial expansions for
a logarithmic fundamental solution of the polyharmonic equation on Rd can be found in [6].
1We have used Whipple’s formula (A.10) in (2.9) and (2.10) to convert the associated Legendre functions of
the second kind Qµ
ν appearing in [8] to associated Legendre functions of the first kind Pµν .
6 H.S. Cohl
3 Jacobi polynomial and limiting expansions
for the Euler kernel
In this section we derive Jacobi, Gegenbauer and Chebyshev polynomial of the first kind series
expansions of the Euler kernel (z − x)−ν . These series expansions are used to obtain azimuthal
Fourier and hyperspherical harmonic expansions for a fundamental solution of the polyharmonic
equation on Rd.
Theorem 1. Let α, β > −1, such that if α, β ∈ (−1, 0) then α + β + 1 6= 0, x, z, ν ∈ C, with
z ∈ C \ (−∞, 1] on any ellipse with foci at ±1 and x in the interior of that ellipse. Then
1
(z − x)ν
=
(z − 1)α+1−ν(z + 1)β+1−ν
2α+β+1−ν
×
∞∑
n=0
(α+ β + 2n+ 1)Γ(α+ β + n+ 1)(ν)n
Γ(α+ 1 + n)Γ(β + 1 + n)
Q
(α+1−ν,β+1−ν)
n+ν−1 (z)P (α,β)
n (x). (3.1)
Note 1. It has been brought to the author’s attention by Tom Koornwinder that Theorem 1
for ν = −n, n ∈ N0, specializes to formula (21) in [21], namely
(z − x)n = (−2)nn!Γ(α+ β + 1)
n∑
k=0
(α+ β + 2k + 1)(α+ β + 1)k
Γ(α+ β + n+ k + 2)
× P (−α−n−1,−β−n−1)
n−k (z)P
(α,β)
k (x).
This equivalence is provided by the interesting identity
P
(−α−n−1,−β−n−1)
n−k (z) =
(−1)n+kΓ(α+ β + n+ k + 2)(z − 1)α+n+1(z + 1)β+n+1
2α+β+2n+1(n− k)! Γ(α+ k + 1)Γ(β + k + 1)
×Q(α+n+1,β+n+1)
k−n−1 (z),
which can be obtained by comparison of Gauss hypergeometric representations.
Proof. Consider the generating function for Gegenbauer polynomials (see, e.g., [27, (18.12.4)])
1
(1 + ρ2 − 2ρx)ν
=
∞∑
n=0
ρnCνn(x) =
∞∑
n=0
ρn
(2ν)n
(ν + 1
2)n
P (ν−1/2,ν−1/2)
n (x), (3.2)
where we have expressed the Gegenbauer polynomial as a symmetric Jacobi polynomial
using (A.7). Utilizing (A.6) in (3.2), reversing the order of the summations, and shifting the n
index yields
1
(1 + ρ2 − 2ρx)ν
=
∞∑
k=0
ρkΓ(α+ β + k + 1)
Γ(α+ β + 2k + 1)
P
(α,β)
k (x)
×
∞∑
n=0
ρn(2ν)n+k(ν + k + 1
2)n(2ν + n+ k)k
n!(ν + 1
2)n+k
× 3F2
(
−n, n+ 2ν + 2k, α+ k + 1
ν + k + 1
2 , α+ β + 2k + 2
; 1
)
.
Taking advantage of standard properties such as (A.2), (A.3) produces
1
(1 + ρ2 − 2ρx)ν
=
√
π
22ν−1Γ(ν)
∞∑
k=0
ρkΓ(2ν + 2k)Γ(α+ β + k + 1)
Γ(ν + k + 1
2)Γ(α+ β + 2k + 1)
P
(α,β)
k (x)
Fourier, Gegenbauer and Jacobi Expansions 7
×
∞∑
n=0
ρn(2ν + 2k)n
n!
3F2
(
−n, n+ 2ν + 2k, α+ k + 1
ν + k + 1
2 , α+ β + 2k + 2
; 1
)
.
We substitute the definition of the 3F2 generalized hypergeometric function (cf. (A.1)) in the
sum over n, and as previously, reverse the order of the two summations and shift the summation
index. It then follows using the duplication formula and (A.2)–(A.5), that one has
∞∑
n=0
ρn(2ν + 2k)n
n!
3F2
(
−n, n+ 2ν + 2k, α+ k + 1
ν + k + 1
2 , α+ β + 2k + 2
; 1
)
=
1
(1− ρ)2ν+2k 2F1
(
ν + k, α+ k + 1
α+ β + 2k + 2
;
−4ρ
(1− ρ)2
)
. (3.3)
If we apply the right-hand side of (3.3) to (A.13) noting Theorem 12.7.3 (expansion of an analytic
function in terms of orthogonal polynomials) in [31] to obtain the regions of convergence, we
obtain the desired result. �
Corollary 1. Let ν ∈ C \−N0, with µ ∈ (−1
2 ,∞) \ {0}, and z ∈ C \ (−∞, 1] on any ellipse with
foci at ±1 with x in the interior of that ellipse. Then
1
(z − x)ν
=
2µ+1/2Γ(µ)eiπ(µ−ν+1/2)
√
π Γ(ν)(z2 − 1)(ν−µ)/2−1/4
∞∑
n=0
(n+ µ)Q
ν−µ−1/2
n+µ−1/2(z)C
µ
n(x). (3.4)
Proof. Let α = β = µ− 1
2 in Theorem 1, and use (A.14) and the definition of the Gegenbauer
polynomial in terms of a symmetric Jacobi polynomial (A.7). The points ν ∈ −N0 which must
be removed are singularities originating from the associated Legendre function of the second
kind on the right-hand side of (3.4). This complete the proof. �
Note that these singularities are removable and correspond to non-negative integer powers of
the binomial z − x. See Note 1. These singularities can be removed by taking the limits as ν
approaches them.
Corollary 2. Let d ≥ 3, ν ∈ C \ {0, 2, 4, . . .}, x,x′ ∈ Rd with r = ‖x‖, r′ = ‖x′‖, and
cos γ = (x,x′)/(rr′). Then
‖x− x′‖ν =
eiπ(ν+d−1)/2Γ
(
d−2
2
)
2
√
π Γ
(
−ν
2
) (
r2> − r2<
)(ν+d−1)/2
(rr′)(d−1)/2
×
∞∑
n=0
(2n+ d− 2)Q
(1−ν−d)/2
n+(d−3)/2
(
r2 + r′2
2rr′
)
Cd/2−1n (cos γ), (3.5)
where r≶ = min
max{r, r
′}.
Proof. Map ν 7→ −ν/2 in (3.4) and substitute ‖x − x′‖ =
√
2rr′
√
z − x, x,x′ ∈ Rd for d ≥ 3
(B.3) with z = (r2 + r′2)/(2rr′), x = cos γ, i.e.,
1
(z − x)ν
7→ (
√
z − x)ν =
1
(2rr′)ν/2
‖x− x′‖ν .
This maps the singularities on the right-hand side of (3.4) at ν ∈ −N0 to singularities at
ν = 0, 2, 4, . . .. �
8 H.S. Cohl
Corollary 3. Let ν ∈ C \ −N0, and x, z ∈ C such that z ∈ C \ (−∞, 1] lie on any ellipse with
foci at ±1 with x in the interior of that ellipse. Then
1
(z − x)ν
=
√
2 e−iπ(ν−1/2)
√
π Γ(ν)(z2 − 1)ν/2−1/4
∞∑
n=0
εnTn(x)Q
ν−1/2
n−1/2(z). (3.6)
Proof. Take the limit as µ→ 0 on the right-hand side of (3.4) and use (A.8). �
Note that (3.6) is given in [8, (3.10)], so (3.1) and (3.4) represent generalizations of that
formula.
Corollary 4. Let d ≥ 2, ν ∈ C \ {0, 2, 4, . . .}, x,x′ ∈ Rd. Then
‖x− x′‖ν =
√
2eiπ(ν+1)/2(2RR′)ν/2√
2 Γ
(
−ν
2
)
(χ2 − 1)−(ν+1)/4
∞∑
m=0
εmTm(cos(φ− φ′))Q−(ν+1)/2
m−1/2 (χ),
where the toroidal parameter χ ∈ [1,∞) is defined in (2.6) such that χ = 1 when x = x′.
Proof. Combine (2.5) and Corollary 3. �
4 Addition theorems in Vilenkin’s polyspherical coordinates
In this section we construct power-law addition theorems in Vilenkin’s polyspherical coordinates.
4.1 Power-law addition theorem on Rd for d ≥ 3
in standard polyspherical coordinates
In standard polyspherical coordinates (B.11) we have the following multi-summation power-law
addition theorem.
Theorem 2. Let ν ∈ C \ {2m, 2m + 2, 2m + 4, . . .}, m ∈ Z, θi ∈ [0, π], for 1 ≤ i ≤ d − 2,
r, r′ ∈ [0,∞) with d ≥ 3. Then
Q
−(ν+1)/2
m−1/2 (χ) =
√
2eiπ(d−2)/2π(d−4)/2
(
2rr′
d−2∏
i=1
sin θisin θ
′
i
)−ν/2
×
(
χ2 − 1
)−(ν+1)/4
(
r2> − r2<
)(ν+d−1)/2
(rr′)(d−1)/2
×
∞∑
ld−2=m
(2ld−2 + 1)(ld−2 −m)!
(ld−2 +m)!
Pmld−2
(cos θd−2) Pmld−2
(
cos θ′d−2
)
×
∞∑
ld−3=ld−2
Θd
d−3 (ld−3, ld−2; θd−3) Θd
d−3
(
ld−3, ld−2; θ
′
d−3
)
· · ·
×
∞∑
l2=l3
Θd
2 (l2, l3; θ2) Θd
2
(
l2, l3; θ
′
2
)
×
∞∑
l1=l2
Θd
1 (l1, l2; θ1) Θd
1
(
l1, l2; θ
′
1
)
Q
(1−ν−d)/2
l1+(d−3)/2
(
r2 + r′2
2rr′
)
, (4.1)
Fourier, Gegenbauer and Jacobi Expansions 9
where
χ =
r2 + r′2 − 2rr′
d−2∑
i=1
cos θicos θ′i
i−1∏
j=1
sin θj sin θ′j
2rr′
d−2∏
i=1
sin θisin θ′i
.
Proof. If we adopt standard polyspherical coordinates (B.11) (see Fig. 4), then one can obtain
an eigenfunction expansion for a power-law fundamental solution of the polyharmonic equation
in standard polyspherical coordinates using the Gegenbauer expansion (3.5) with the addition
theorem for hyperspherical harmonics (B.13), and the normalized standard hyperspherical har-
monics (B.19), obtaining
‖x− x′‖ν =
4eiπ(ν+d−1)/2π(d−1)/2
Γ
(
−ν
2
) (
r2> − r2<
)(ν+d−1)/2
(rr′)(d−1)/2
×
∑
l1,K
Q
(1−ν−d)/2
l1+(d−3)/2
(
r2 + r′2
2rr′
)
Y K
l1 (x̂)Y K
l1
(x̂′). (4.2)
If we expand the product of polyspherical harmonics in (4.2) with (B.19) after reversing the
order of the summations, we obtain
‖x− x′‖ν =
∞∑
m=0
cos(m(φ− φ′))π
(d−3)/2eiπ(ν+d−1)/2εm
Γ
(
−ν
2
) (
r2> − r2<
)(ν+d−1)/2
(rr′)(d−1)/2
×
∞∑
ld−2=m
(2ld−2 + 1)(ld−2 −m)!
(ld−2 +m)!
Pmld−2
(cos θd−2) Pmld−2
(
cos θ′d−2
)
×
∞∑
ld−3=ld−2
Θd
d−3 (ld−3, ld−2; θd−3) Θd
d−3
(
ld−3, ld−2; θ
′
d−3
)
· · ·
×
∞∑
l2=l3
Θd
2 (l2, l3; θ2) Θd
2
(
l2, l3; θ
′
2
)
×
∞∑
l1=l2
Θd
1 (l1, l2; θ1) Θd
1
(
l1, l2; θ
′
1
)
Q
(1−ν−d)/2
l1+(d−3)/2
(
r2 + r′2
2rr′
)
, (4.3)
where Θd
j (lj , lj+1; θ), for 1 ≤ j ≤ d− 2 is defined in (B.20).
The Fourier expansion for a power-law fundamental solution of the polyharmonic equation in
standard polyspherical coordinates is obtained by substituting the expansion in terms of Cheby-
shev polynomials of the first kind (3.6) in the algebraic expression for a power-law fundamental
solution of the polyharmonic equation (cf. (2.5)). This results in
‖x− x′‖ν =
√
π
2
eiπ(ν+1)/2
Γ (−ν/2)
(
2rr′
d−2∏
i=1
sin θisin θ
′
i
)ν/2 (
χ2 − 1
)(ν+1)/4
×
∞∑
m=−∞
eim(φ−φ′)Q
−(ν+1)/2
m−1/2 (χ). (4.4)
By comparing the Fourier coefficients of (4.3) with (4.4), we complete the proof of this
theorem. �
10 H.S. Cohl
This is just one example of a derived multi-summation addition theorem for arbitrary di-
mensions. There are an unlimited number of such straightforward examples to generate. In the
next section we derive another example which is valid on Rd where d is given by a power of two,
generalized Hopf coordinates (B.12).
4.2 Power-law addition theorem on R2q
for q ≥ 2
in generalized Hopf coordinates
In generalized Hopf coordinates (B.12) we have the following multi-summation power-law addi-
tion theorem.
Theorem 3. Let ν ∈ C \ {2m, 2m + 2, 2m + 4, . . .}, m1 ∈ Z, r, r′ ∈ [0,∞), ϑi ∈
[
0, π2
]
with
1 ≤ i ≤ 2q−1 − 1, φi ∈ [0, 2π) such that 1 ≤ i ≤ 2q−1. Then
Q
−(ν+1)/2
m1−1/2 (χ) =
−
(
q−1∏
j=1
cosϑ2j−1 cosϑ′
2j−1
)−ν/2
2(ν+1)/2(χ2 − 1)(ν+1)/4
(
r2> − r2<
rr′
)(ν+2q−1)/2
×
∞∑
m2=0
εm2 cos(m2(φ2 − φ′2)) · · ·
∞∑
m2q−1=0
εm2q−1 cos(m2q−1(φ2q−1 − φ′2q−1))
×
∞∑
n2q−1−1=0
Υq
2q−1−1
(
n2q−1−1
|m2q−1−1|, |m2q−1 |;ϑ2q−1−1
)
×Υq
2q−1−1
(
n2q−1−1
|m2q−1−1|, |m2q−1 |;ϑ
′
2q−1−1
)
· · ·
×
∞∑
n2q−2=0
Υq
2q−2
(
n2q−2
|m1|, |m2|
;ϑ2q−2
)
Υq
2q−2
(
n2q−2
|m1|, |m2|
;ϑ′2q−2
)
×
∞∑
n2q−2−1=0
Υq
2q−2−1
(
n2q−2−1
l2q−1−2, l2q−1−1
;ϑ2q−2−1
)
×Υq
2q−2−1
(
n2q−2−1
l2q−1−2, l2q−1−1
;ϑ′2q−2−1
)
· · ·
×
∞∑
n1=0
Υq
1
(
n1
l2, l3
;ϑ1
)
Υq
1
(
n1
l2, l3
;ϑ′1
)
Q
(1−ν−2q)/2
2n1+l2+l3+(2q−3)/2
(
r2 + r′2
2rr′
)
, (4.5)
where
χ =
r2 + r′2 − 2rr′ cos γ + 2rr′ cos(φ1 − φ′1)
q−1∏
j=1
cosϑ2j−1 cosϑ′
2j−1
2rr′
q−1∏
j=1
cosϑ2j−1cosϑ′
2j−1
. (4.6)
Proof. If we adopt generalized Hopf coordinates (B.12) (see Fig. 5), then we can use the
corresponding harmonics (B.21) in combination with the addition theorem for hyperspherical
harmonics (B.13). We compare the Gegenbauer expansion for powers of the distance (3.5) with
the Fourier expansion
‖x− x′‖ν =
√
2ieiπν/2
(
2rr′
q−1∏
j=1
cosϑ2j−1 cosϑ′
2j−1
)ν/2
√
π Γ
(−ν
2
)
(χ2 − 1)−(ν+1)/4
Fourier, Gegenbauer and Jacobi Expansions 11
×
∞∑
m1=−∞
eim1(φ1−φ′1)Q
−(ν+1)/2
m1−1/2 (χ), (4.7)
where χ > 1 is given by (4.6). Notice that χ is independent of φ1−φ′1. By using the Gegenbauer
expansion (3.5) and inserting the appropriate Gegenbauer polynomial using the addition theorem
for hyperspherical harmonics (B.13), we obtain
‖x− x′‖ν =
−2iπ2
q−1
eiπν/2(r2> − r2<)(ν+2q−1)/2
√
π Γ
(−ν
2
)
(rr′)(2q−1)/2
×
∑
l1,K
Q
(1−ν−2q)/2
l1+(2q−3)/2
(
r2 + r′2
2rr′
)
Y K
l1 (x̂)Y K
l1
(x̂′). (4.8)
By expanding the product of polyspherical harmonics in (4.8) with (B.21) expressed in terms of
surrogate quantum numbers and reversing the order of the sums, we obtain a multi-summation
expression for the power of the Euclidean distance between two points in generalized Hopf coor-
dinates. Through comparison of the resulting equation with the m1 Fourier coefficients of (4.7),
we derive (4.5), a multi-summation addition theorem for the associated Legendre function of
the second kind with argument χ (cf. (4.6)). �
4.3 Power-law addition theorems on R3
In d = 3 there are two ways to construct polyspherical coordinates, with trees of type b′a (see
Fig. 2b) and ba (see Fig. 2c). We only treat the first tree since the addition theorem from the
second tree is trivially obtained from the first.
4.3.1 Type ba coordinates
Corollary 5. Let ν ∈ C \ {2m, 2m+ 2, 2m+ 4, . . .}, m ∈ N0, θ, θ′ ∈ [0, π], r, r′ ∈ [0,∞). Then
Q
−(ν+1)/2
m−1/2 (χ) = i
√
π2−(ν+3)/2(sin θ sin θ′)−ν/2(χ2 − 1)−(ν+1)/4
(
r2> − r2<
rr′
)(ν+2)/2
×
∞∑
l=m
(2l + 1)
(l −m)!
(l +m)!
Q
−(ν+2)/2
l
(
r2 + r′2
2rr′
)
Pml (cos θ)Pml (cos θ′), (4.9)
where
χ =
r2 + r′2 − 2rr′ cos θ cos θ′
2rr′ sin θ sin θ′
. (4.10)
Proof. Taking d = 3 in (4.1) for type ba coordinates converts χ to (4.10) and the rele-
vant Gegenbauer polynomials reduce to Ferrers functions through (A.12). This completes the
proof. �
Equation (4.9) is a generalization of one of the main results of [10]. This can be observed if
you substitute ν = −1 in (4.9) (this corresponds to a fundamental solution of Laplace’s equation
on R3), then the associated Legendre function of the second kind on the right-hand side reduces
to an elementary function through [1, (8.6.11)], producing
Qm−1/2(χ) = π
√
sin θ sin θ′
∞∑
l=|m|
(l −m)!
(l +m)!
(
r<
r>
)l+1/2
Pml (cos θ)Pml (cos θ′).
12 H.S. Cohl
4.4 Power-law addition theorems on R4
In d = 4 there are five ways to construct polyspherical coordinates, with trees of type b2a (see
Fig. 3a), bb′a (see Fig. 3b), b′ba (see Fig. 3c), b′2a (see Fig. 3d), ca2 (see Fig. 3e). We only
treat the first and the fifth trees since the addition theorems from the second, third and fourth
trees are trivially obtained from the first.
4.4.1 Type b2a coordinates
Corollary 6. Let ν ∈ C \ {2m, 2m+ 2, 2m+ 4, . . .}, m ∈ Z, r, r′ ∈ [0,∞), θ1, θ
′
1, θ2, θ
′
2 ∈ [0, π].
Then
Q
−(ν+1)/2
m−1/2 (χ) =
−1
2(ν+1)/2
(
r2> − r2<
rr′
)(ν+3)/2 (
χ2 − 1
)−(ν+1)/4
(sin θ1 sin θ′1 sin θ2 sin θ′2)
−ν/2
×
∞∑
l2=|m|
22l2(2l2 + 1)(l2!)
2(l2 −m)!
(l2 +m)!
(sin θ1 sin θ′1)
l2Pml2 (cos θ2)P
m
l2 (cos θ′2)
×
∞∑
l1=l2
(l1 + 1)(l1 − l2)!
(l1 + l2 + 1)!
Q
−(ν+3)/2
l1+1/2
(
r2 + r′2
2rr′
)
× C l2+1
l1−l2(cos θ1)C
l2+1
l1−l2(cos θ′1), (4.11)
where
χ =
r2 + r′2 − 2rr′ cos θ1 cos θ′1 − 2rr′ sin θ1 sin θ′1 cos θ2 cos θ′2
2rr′ sin θ1 sin θ′1 sin θ2 sin θ′2
.
Proof. Taking d = 4 in (4.1) completes the proof. �
If you substitute ν = −2 (a fundamental solution for the Laplacian on R4) in (4.11) then the
Legendre functions of the second kind reduce to elementary functions through [1, (8.6.10-11)],
and one obtains the following
(χ2 − 1)−1/2
(χ+
√
χ2 − 1)m
= 2 sin θ1 sin θ′1 sin θ2 sin θ′2
×
∞∑
l2=|m|
22l2(2l2 + 1)(l2!)
2(l2 −m)!
(l2 +m)!
(sin θ1 sin θ′1)
l2Pml2 (cos θ2)P
m
l2 (cos θ′2)
×
∞∑
l1=l2
(l1 − l2)!
(l1 + l2 + 1)!
(
r<
r>
)l1+1
C l2+1
l1−l2(cos θ1)C
l2+1
l1−l2(cos θ′1).
4.4.2 Type ca2 coordinates
Corollary 7. Let ν ∈ C \ {2m, 2m + 2, 2m + 4, . . .}, m1 ∈ Z, r, r′ ∈ [0,∞), ϑ, ϑ′ ∈ [0, π2 ],
φ2, φ
′
2 ∈ [0, 2π). Then
Q
−(ν+1)/2
m1−1/2 (χ) =
−2−(ν+1)/2
(χ2 − 1)(ν+1)/4
(
r2> − r2<
rr′
)(ν+3)/2 (
cosϑ cosϑ′
)|m1|−ν/2
×
∞∑
m2=0
εm2 cos(m2(φ2 − φ′2))(sinϑ sinϑ′)m2
Fourier, Gegenbauer and Jacobi Expansions 13
×
∞∑
n=0
(2n+ |m1|+m2 + 1)(|m1|+m2 + n)!n!
(|m1|+ n)!(m2 + n)!
×Q−(ν+3)/2
2n+|m1|+m2+1/2
(
r2 + r′2
2rr′
)
P (m2,|m1|)
n (cos 2ϑ)P (m2,|m1|)
n (cos 2ϑ′), (4.12)
where
χ =
r2 + r′2 − 2rr′ sinϑ sinϑ′ cos(φ2 − φ′2)
2rr′ cosϑ cosϑ′
.
Proof. Taking q = 2 in (4.5) completes the proof. �
If you substitute ν = −2 in (4.12), then the Legendre functions of the second kind reduce to
elementary functions through [1, (8.6.10-11)], and one obtains the following
(χ2 − 1)−1/2
(χ+
√
χ2 − 1)m1
= 2
(
cosϑ cosϑ′
)|m1|+1
∞∑
m2=0
εm2 cos
[
m2(φ2 − φ′2)
]
(sinϑ sinϑ′)|m2|
×
∞∑
n=0
(|m1|+ |m2|+ n)!n!
(|m1|+ n)!(|m2|+ n)!
(
r<
r>
)|m1|+|m2|+2n+1
× P (|m2|,|m1|)
n (cos 2ϑ)P (|m2|,|m1|)
n (cos 2ϑ′). (4.13)
Note that in the addition theorems (4.12) and (4.13), that if you make the map ϑ 7→ ϑ−π
2 , then
this transformation preserves the addition theorems such that m1 ↔ m2. This transformation
is equivalent to swapping the position of φ1 and φ2 for the tree in Fig. 3e.
A Special functions and orthogonal polynomials
The generalized hypergeometric function pFq : Cp × (C \ −N0)
q × {z ∈ C : |z| < 1} → C can be
defined as
pFq
(
a1, . . . , ap
b1, . . . , bq
; z
)
:=
∞∑
n=0
(a1)n · · · (ap)n
(b1)n · · · (bq)n
zn
n!
, (A.1)
where the Pochhammer symbol (rising factorial) (·)n : C → C for n ∈ N0 is defined by (z)n :=
n∏
i=1
(z + i− 1). Furthermore one has
(z)n =
Γ(z + n)
Γ(z)
(A.2)
for z ∈ C \ −N0, which implies
(z)n+k = (z)k(z + k)n, (A.3)
k ∈ N0, z ∈ C. One also has
(−n− k)k =
(−1)k(n+ k)!
n!
. (A.4)
In this paper, we will use two different generalized hypergeometric functions, namely 3F2 and
the Gauss hypergeometric function 2F1 (see for instance Chapter 15 in [27]). We also use the
binomial expansions for p = 1, q = 0, namely
1F0
(
α
−; z
)
= (1− z)−α, (A.5)
14 H.S. Cohl
where α, z ∈ C such that |z| < 1. The special functions used in this paper as well as and their
properties can be described in terms of these.
There are many important orthogonal polynomials which can be defined in terms of a termi-
nating generalized hypergeometric series. The Jacobi polynomials P
(α,β)
n : C → C, for n ∈ N0,
and α, β > −1 such that if α, β ∈ (−1, 0) then α+ β + 1 6= 0, are defined as [27, (18.5.7)]
P (α,β)
n (z) :=
(α+ 1)n
n!
2F1
(
−n, n+ α+ β + 1
α+ 1
;
1− z
2
)
.
These polynomials are orthogonal with respect to the positive weight w : (−1, 1) → [0,∞),
w(x) := (1− x)α(1 + x)β, with orthogonality relation∫ 1
−1
P (α,β)
m (x)P (α,β)
n (x)(1− x)α(1 + x)βdx =
δm,n(
Nα,β
n
)2 .
The normalization constant Nα,β
n ∈ (0,∞) is given by
Nα,β
n =
√
(2n+ α+ β + 1)Γ(n+ α+ β + 1)n!
2α+β+1Γ(n+ α+ 1)Γ(n+ β + 1)
.
The connection relation for Jacobi polynomials with two free parameters is given by (see for
instance [17, p. 256])
P (γ,δ)
n (x) =
n∑
k=0
cn,k(γ, δ;α, β)P
(α,β)
k (x), (A.6)
where γ, δ > −1, and such that if γ, δ ∈ (−1, 0) then γ + δ + 1 6= 0,
cn,k(γ, δ;α, β) =
(γ + k + 1)n−k(n+ γ + δ + 1)kΓ(α+ β + k + 1)
(n− k)!Γ(α+ β + 2k + 1)
× 3F2
(
−n+ k, n+ k + γ + δ + 1, α+ k + 1
γ + k + 1, α+ β + 2k + 2
; 1
)
.
Jacobi polynomials with parameters α = β are described as symmetric and are representable
in terms of Gegenbauer polynomials using (see (6.4.9) in [3])
Cνn(x) =
(2ν)n
(ν + 1
2)n
P (ν−1/2,ν−1/2)
n (x), (A.7)
where ν ∈ (−1
2 ,∞) \ {0}. The Chebyshev polynomial of the first kind Tn : C→ C is defined as
(see § 5.7.2 in [25])
Tn(z) := 2F1
(
−n, n
1
2
;
1− z
2
)
,
for n ∈ N0. These can be computed in terms of Gegenbauer polynomials using
lim
µ→0
n+ µ
µ
Cµn(x) = εnTn(x) (A.8)
(see for instance (6.4.13) in [3]).
Fourier, Gegenbauer and Jacobi Expansions 15
The associated Legendre function of the second kind Qµν : C \ (−∞, 1]→ C, ν +µ /∈ −N, can
be defined in terms of the Gauss hypergeometric function as follows [27, (14.3.7) and § 14.21]
Qµν (z) :=
√
πeiπµΓ(ν + µ+ 1)(z2 − 1)µ/2
2ν+1Γ(ν + 3
2)zν+µ+1 2F1
(ν+µ+1
2 , ν+µ+2
2
ν + 3
2
;
1
z2
)
,
for |z| > 1 and elsewhere in z by analytic continuation of the Gauss hypergeometric function.
One may also define the associated Legendre function of the second kind using [25, entry 24,
p. 161]
Qµν (z) :=
eiπµ2νΓ(ν + 1)Γ(ν + µ+ 1)(z + 1)µ/2
Γ(2ν + 2)(z − 1)b/2+a+1 2F1
(
ν + 1, ν + µ+ 1
2 + 2ν
;
2
1− z
)
, (A.9)
for |1−z| > 2. Similarly, the associated Legendre function of the first kind can be defined using
the Gauss hypergeometric function [27, (14.3.6) and § 14.21(i)]
Pµν (z) :=
1
Γ(1− µ)
(
z + 1
z − 1
)µ/2
2F1
(
−ν, ν + 1
1− µ ;
1− z
2
)
,
where |1−z| < 2, and elsewhere in z by analytic continuation. We can use Whipple’s formula to
relate the associated Legendre function of the first kind with the associated Legendre function
of the second kind. It is given by [1, (8.2.7)]
P
−ν−1/2
−µ−1/2
(
z√
z2 − 1
)
=
√
2
π
(z2 − 1)1/4e−iµπ
Γ(ν + µ+ 1)
Qµν (z), (A.10)
for Re z > 0. The Ferrers function of the first kind (associated Legendre function of the first
kind on-the-cut) Pµν : (−1, 1)→ C can be defined as [27, (14.3.1)]
Pµν (x) :=
1
Γ(1− µ)
(
1 + x
1− x
)µ/2
2F1
(
−ν, ν + 1
1− µ ;
1− x
2
)
. (A.11)
There is a relation between certain Gegenbauer polynomials on (−1, 1) and the Ferrers function
of the first kind (cf. (8.936.2) in [16]), namely
C
m+1/2
l−m (x) =
(−1)m(1− x2)−m/2
(2m− 1)!!
Pml (x), (A.12)
where the double factorial · : {−1, 0, 1, . . .} → N is defined such that
n!! :=
n · (n− 2) · · · 2 if n even ≥ 2,
n · (n− 2) · · · 1 if n odd ≥ 1,
1 if n = −1, 0.
The Jacobi function of the second kind Q
(α,β)
γ : C\(−∞, 1]→ C (cf. [13, (10.8.18)]) is defined by
Q(α,β)
γ (z) :=
2α+β+γΓ(α+ γ + 1)Γ(β + γ + 1)
Γ(α+ β + 2γ + 2)(z − 1)α+γ+1(z + 1)β
2F1
(
γ + 1, α+ γ + 1
α+ β + 2γ + 2
;
2
1− z
)
, (A.13)
where α + γ, β + γ /∈ −N. We can derive a relation between the symmetric Jacobi function of
the second kind and the associated Legendre function of the second kind
Q
(µ−ν+1/2,µ−ν+1/2)
n+ν−1 (z) =
2µ−ν+1/2Γ (µ+ n+ 1/2) eiπ(µ−ν+1/2)
Γ(ν + n)(z2 − 1)(µ−ν)/2+1/4
Q
ν−µ−1/2
n+µ−1/2(z), (A.14)
where n ∈ N0, µ ∈ C \
{
−1
2 ,−
3
2 ,−
5
2 , . . .
}
, ν ∈ C \ −N0. The relation (A.14) can be verified by
comparing (A.13) with (A.9).
16 H.S. Cohl
Figure 1. This figure shows the possibilities from left to right for branching nodes of type a, b, b′
and c. For type a, both branches end on a leaf node. The angle corresponding to this type of branching
node is φa ∈ [0, 2π). For type b, the left branch ends on a leaf node and the right branch ends on a
branching node. The angle corresponding to this type of branching node is θb ∈ [0, π]. For type b′, the
left-branch ends on a branching node and the right branch ends on a leaf node. The angle corresponding
to this type of branching node is θb′ ∈
[
−π2 ,
π
2
]
. For type c, both the left and right branches ends on
branching nodes (branching nodes of type c are only possible for d ≥ 4). The angle corresponding to this
type of branching node is ϑc ∈
[
0, π2
]
.
B Vilenkin’s polyspherical coordinates and the method of trees
Polyspherical coordinates are hyperspherical coordinates which are described by a radial co-
ordinate r ∈ [0,∞) plus (d − 1)-angles which together parametrize points on Sd−1r .2 We will
first discuss a general procedure for constructing polyspherical coordinate systems called the
“method of trees”3. We will give some examples of polyspherical coordinates which will be used
in the rest of the paper and then describe the harmonic separated product solutions.
In these rooted trees, there are two types of nodes, leaf nodes and branching nodes. For
a coordinate system on Rd, there are d-leaf nodes, each corresponding to the particular Cartesian
component of an arbitrary position vector x ∈ Rd. The branching nodes split into two separate
branches, one up to the left and one up to the right. Each branch emanating from a branching
node will end on either a leaf node or on another branching node. There are four possibilities
for branching nodes (see Fig. 1).
Separation of variables in polyspherical coordinates with (d − 1)-angles, for Laplace’s equa-
tion on Rd produces (d − 1)-separation constants, each of which are called quantum numbers.
The quantum numbers corresponding to these angles are all integers. Quantum numbers for
a particular tree label the basis of separable solutions for Laplace’s equation in that particular
coordinate system. With each branching node of the tree, we associate a quantum number as
well as an angle. The quantum number corresponding to a (2π)-periodic (azimuthal) angle is
called an azimuthal quantum number. Each azimuthal angle corresponds to a branching node
of type a and an azimuthal quantum number m ∈ Z. A natural consequence of the method of
trees is that there must exist at least one azimuthal angle for each tree, and therefore also for
each polyspherical coordinate system.
Branching nodes of type b, b′ and c (see Fig. 1), are associated with angles, which in turn are
associated with quantum numbers which we refer to as angular momentum quantum numbers
l ∈ N0 (see for instance Chapter 10 in [14]). There is always at least one branching node, the root
node, and all branching nodes correspond to a particular angle and quantum number. Let us
associate each branching node with an angle and its corresponding quantum number. These trees
2The Riemannian manifold Sdr is defined as the set of all points in Rd+1 such that x2
0 + · · ·+ x2
d = r2 (r > 0),
with the metric induced from that of the ambient Euclidean space. We denote the d-dimensional hypersphere of
unit radius as Sd := Sd1.
3Describing polyspherical coordinate systems in terms of rooted trees was originally developed in [34] (see also
[32, § 9.5], [33, § 10.5]) and has since been used extensively by others in a variety of contexts (see for instance
[18, 19, 20]).
Fourier, Gegenbauer and Jacobi Expansions 17
parametrize points in polyspherical coordinates as follows. Starting at the root node, traverse
the tree upward until you reach the leaf node corresponding to xi. The parametrization for xi
is given by the hyperspherical radius r multiplied by cosine or sine of each angle encountered as
you traverse the tree upward until you reach the leaf node corresponding to xi. If you branch
upwards to the left or upwards to the right at each branching node, multiply by the cosine or sine
of the corresponding angle respectively. This procedure produces the appropriate transformation
from polyspherical coordinates to Cartesian coordinates.
There are large numbers of equivalent trees and an even larger number of total trees, each
with their own specific polyspherical coordinate system. The enumeration of these trees are
characterized as follows. For d, bd ∈ N, let bd be the total number of trees. Then b1 = 1 is the
number of possible 1-branch trees. In our context, a 1-branch tree does not exist in isolation.
The following recurrence relation gives the total number of trees for arbitrary dimension
bd =
d−1∑
i=1
bibd−i. (B.1)
Using the recurrence relation (B.1), the first few elements of the sequence are given by
(bd : d ∈ {2, . . . , 13}) = (1, 2, 5, 14, 42, 132, 429, 1430, 4862, 16796, 58786, 208012).
The total number of trees are given in terms of the Catalan numbers Cn (see for instance Sloane
integer sequence A000108 [29] or p. 200 in [30]), i.e., bd = Cd−1 = 1
d
(
2d−2
d−1
)
, where
(
n
k
)
= n!
k!(n−k)!
is the binomial coefficient for k, n ∈ N0 with 0 ≤ k ≤ n. If ad ∈ N is the total number
of equivalence classes for equivalent trees for a given dimension (determined by a left-right
symmetry in the topology of the trees), then ad satisfies the following recurrence relation
ad =
bd/2c∑
i=1
aiad−i, if d odd,
d/2−1∑
i=1
aiad−i +
1
2
ad/2
(
ad/2 + 1
)
, if d even.
(B.2)
Using the recurrence relation (B.2), the first few elements of the sequence are given by
(ad : d ∈ {2, . . . , 13}) = (1, 1, 2, 3, 6, 11, 23, 46, 98, 207, 451, 983).
These are given in terms of the Wedderburn–Etherington numbers (see for instance Sloane
integer sequence A001190 [29]). We use a left-to-right recursive naming language for our trees
based on a depth-first search (see [12, pp. 540–549]). This naming language is given by listing
the types of branching nodes available in a particular tree.
In a polyspherical coordinate system the Euclidean distance between two points (2.5) can
also be given as
‖x− x′‖ =
√
2rr′
[
r2 + r′2
2rr′
− cos γ
]1/2
, (B.3)
where the separation angle γ ∈ [0, π] is defined through the relation
cos γ :=
(x,x′)
‖x‖‖x′‖
, (B.4)
using the Euclidean inner product and norm (cf. (1.1)). The method of trees constructs the
cosine of the separation angle (B.4) in a direct manner. The cosine of the separation angle will
18 H.S. Cohl
Figure 2. Tree diagrams for two and three dimensional polyspherical coordinate systems of type: (a) a,
(b) ba, (c) b′a.
be given by the sum of d-terms, each corresponding to a leaf node of the tree. There is a unique
path starting from the root node to each leaf node. It is
cos γ =
d∑
i=1
Ni∏
j=1
Ai,j(ψi,j)Ai,j(ψ
′
i,j),
where Ni is the number of branching nodes encountered from the root node to the leaf node,
Ai,j : R → [−1, 1] is either the trigonometric cosine or sine function depending respectively
on whether the left branch or right branch is chosen respectively, and ψi,j ∈ R is the angle
corresponding to the jth branching node for each ith leaf node. The formula for the cosine of
the separation angle is unique for each tree.
B.1 Examples of Vilenkin’s polyspherical coordinate systems
The simplest example of a polyspherical coordinate system on Rd occurs for d = 2 (polar
coordinates) where there is one branching node (the root node) and two leaf nodes (see Figs. 2a
and 5a). The left-branch ends on the leaf node corresponding to x1 and the right branch ends
on the leaf node corresponding to x2, i.e.,
x1 = r cosφ, x2 = r sinφ, (B.5)
where r ∈ [0,∞) and φ ∈ [0, 2π). We refer to this tree as type a. The cosine of the separation
angle is given by
cos γ = cos(φ− φ′), (B.6)
and corresponding to the angle φ is the azimuthal quantum number m ∈ Z (see Fig. 2a).
In d = 3 there are two possible topological trees, each corresponding to one of two different
trees. The first tree (see Fig. 2b) corresponds to the coordinate system
x1 = r cos θ, x2 = r sin θ cosφ, x3 = r sin θ sinφ, (B.7)
where θ ∈ [0, π], φ ∈ [0, 2π), i.e., standard spherical coordinates. This is a tree of type ba. The
cosine of the separation angle is given by
cos γ = cos θ cos θ′ + sin θ sin θ′ cos(φ− φ′),
and corresponding to the angles θ ∈ [0, π] and φ ∈ [0, 2π) are the quantum numbers l ∈ N0 and
m ∈ Z respectively. The second tree (see Fig. 2c) is of type b′a and is equivalent to the first.
Fourier, Gegenbauer and Jacobi Expansions 19
Figure 3. Tree diagrams for four dimensional polyspherical coordinate systems of type: (a) b2a,
(b) bb′a, (c) b′ba, (d) b′2a, (e) ca2.
In d = 4 there are five possible topological trees for polyspherical coordinates. The first tree
(see Figs. 3a and 4) corresponds to the coordinate system
x1 = r cos θ1, x2 = r sin θ1 cos θ2,
x3 = r sin θ1 sin θ2 cosφ, x4 = r sin θ1 sin θ2 sinφ, (B.8)
where θ1, θ2 ∈ [0, π]. This tree is of type b2a, and the cosine of the separation angle is given by
cos γ = cos θ1 cos θ′1 + sin θ1 sin θ′1
(
cos θ2 cos θ′2 + sin θ2 sin θ′2 cos(φ− φ′)
)
.
The second, third and fourth trees (see Figs. 3b, 3c, 3d) are equivalent to the first.
The fifth tree (see Figs. 3e and 5b) corresponds to Hopf coordinates,
x1 = r cosϑ cosφ1, x2 = r cosϑ sinφ1,
x3 = r sinϑ cosφ2, x4 = r sinϑ sinφ2, (B.9)
where ϑ ∈
[
0, π2
]
and φ1, φ2 ∈ [0, 2π). This tree is of type ca2. The cosine of the separation
angle is given by
cos γ = cosϑ cosϑ′ cos(φ1 − φ′1) + sinϑ sinϑ′ cos(φ2 − φ′2). (B.10)
There are many choices for polyspherical coordinates on Rd, suitably defined for any number
of dimensions d ≥ 2. The simplest example of a polyspherical coordinate system on Rd (which
generalizes type ba and b2a) are what we refer to as standard polyspherical coordinates (see
Fig. 4). These are
x1 = r cos θ1,
x2 = r sin θ1 cos θ2,
x3 = r sin θ1 sin θ2 cos θ3,
· · · · · · · · · · · · · · · · · · · · · · · ·
20 H.S. Cohl
Figure 4. Tree diagram for d-dimensional standard polyspherical coordinates of type bd−2a (standard
polyspherical coordinates).
xd−2 = r sin θ1 · · · sin θd−3 cos θd−2,
xd−1 = r sin θ1 · · · sin θd−3 sin θd−2 cosφ,
xd = r sin θ1 · · · sin θd−3 sin θd−2 sinφ, (B.11)
where θi ∈ [0, π] for 1 ≤ i ≤ d − 2 and φ ∈ [0, 2π). Using our naming procedure, this tree is of
type bd−2a. In these coordinates the cosine of the separation angle is given by
cos γ =
d−2∑
i=1
cos θicos θ′i
i−1∏
j=1
sin θjsin θ
′
j + cos(φ− φ′)
d−2∏
i=1
sin θisin θ
′
i.
Another example of a polyspherical coordinate system which is valid for large dimensions is
what we will refer to as generalized Hopf coordinates (see Fig. 5). These coordinates, valid on R2q
for q ≥ 1, generalize two-dimensional polar coordinates (B.5, type a) and four-dimensional Hopf
coordinates (B.9, type ca2). These coordinates are unique in that they correspond to the only
trees which contain only themselves in their equivalence class (see (B.2)). These coordinate sys-
tems have separated harmonic eigenfunctions which are given in terms of complex exponentials,
and for q ≥ 2, non-symmetric Jacobi polynomials (see Fig. 5).
This coordinate system is suitably defined for dimensions d = 2q for q ≥ 1. The transforma-
tion formulae to Cartesian coordinates are given by
x1 = r cosϑ1 cosϑ2 cosϑ4 cosϑ8 · · · cosϑ2q−2 cosφ1,
x2 = r cosϑ1 cosϑ2 cosϑ4 cosϑ8 · · · cosϑ2q−2 sinφ1,
· · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · ·
x2q−1−1 = r cosϑ1 sinϑ2 sinϑ5 sinϑ11 · · · sinϑ3·2q−3−1 cosφ2q−2 ,
x2q−1 = r cosϑ1 sinϑ2 sinϑ5 sinϑ11 · · · sinϑ3·2q−3−1 sinφ2q−2 ,
x2q−1+1 = r sinϑ1 cosϑ3 cosϑ6 cosϑ12 · · · cosϑ3·2q−3 cosφ2q−2+1,
x2q−1+2 = r sinϑ1 cosϑ3 cosϑ6 cosϑ12 · · · cosϑ3·2q−3 sinφ2q−2+1,
· · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · · ·
x2q−1 = r sinϑ1 sinϑ3 sinϑ7 sinϑ17 · · · sinϑ2q−1−1 cosφ2q−1 ,
x2q = r sinϑ1 sinϑ3 sinϑ7 sinϑ17 · · · sinϑ2q−1−1 sinφ2q−1 , (B.12)
Fourier, Gegenbauer and Jacobi Expansions 21
Figure 5. This figure is a tree diagram for polyspherical generalized Hopf coordinates of type V2q on R2q
with q = 1, 2, 3, 4 for (a), (b), (c), (d) respectively. The first (2q−1 − 1)-branching nodes are of type c
which correspond to the angles ϑi ∈
[
0, π2
]
and quantum numbers li ∈ N0. The following (2q)-branching
nodes are of type a which correspond to the angles φi ∈ [0, 2π) and quantum numbers mi ∈ Z. These
coordinates correspond to transformation (B.12).
where ϑi ∈
[
0, π2
]
for 1 ≤ i ≤ 2q−1 − 1 and φi ∈ [0, 2π) for 1 ≤ i ≤ 2q−1. Using our naming
procedure, these coordinates are of type
V2q = cV2q−1V2q−1 ,
where q ∈ N, with type V2 = a (polar) coordinates on R2 (see Figs. 2a and 5a).
The cosine of the separation angle in this coordinate system may be given as follows. Define
the symbol qG
r
s ∈ [−1, 1], where 0 ≤ s ≤ q and 1 ≤ r ≤ 2q − 1, by the recursive formula
qG
r
s = cosϑr−1+2q−s cosϑ′r−1+2q−s qG
2r−1
s−1 + sinϑr−1+2q−s sinϑ′r−1+2q−s qG
2r
s−1,
with qG
i
0 = 1. Then the cosine of the separation angle is given by
cos γ = qG
1
q
(cf. (B.6), (B.10)). Note that through the identification φi = ϑi−1+2q−1 , where 1 ≤ i ≤ 2q−1,
then qG
i
1 = cos(φi − φ′i). Thus, this shows one may stop this recursion at s = 1.
B.2 Hyperspherical harmonics in polyspherical coordinates
The eigenfunction expansions for a power-law fundamental solution of the polyharmonic equation
in polyspherical coordinates can be derived using a Gegenbauer polynomial expansion for the
22 H.S. Cohl
relevant kernel (see Corollary 2) in conjunction with the addition theorem for hyperspherical
harmonics. This addition theorem is given by
Cd/2−1n (cos γ) =
2(d− 2)πd/2
(2n+ d− 2)Γ(d/2)
∑
K
Y K
n (x̂)Y K
n (x̂′) (B.13)
(for a proof see [35]; see also § 10.2.1 in [14]), where K stands for a set of (d−2)-quantum numbers
identifying harmonics for a given value of n ∈ N0, and cos γ is the cosine of the separation angle
between two arbitrary vectors x,x′ ∈ Rd (see (B.4)). The functions Y K
n : Sd−1 → C are the
normalized hyperspherical harmonics. Normalization of the hyperspherical harmonics is achieved
through∫
Sd−1
Y K
n (x̂)Y K
n (x̂)dΩ = 1,
where dΩ is the Riemannian volume measure on Sd−1.
The general basis functions that one obtains by putting coordinates on the d-dimensional unit
hypersphere Sd−1, can be specified as solutions to the angular part of Laplace’s equation on Rd.
These correspond to separated solutions of Laplace’s equation, using the Laplace–Beltrami ope-
rator on the hypersphere Sd−1. The following numbers are associated with each branching node
m ∈ Z, l, lα, lβ ∈ N0 (see Fig. 1). The number of vertices above each branching node lα and lβ
are represented by Sα and Sβ respectively.
The following separated factors of eigenfunctions are generated at each branching node for
normalized hyperspherical harmonics in polyspherical coordinates using the method of trees (see
for instance [19, (2.3)–(2.6)], [33, § 10.5.3]):
• Type a:
Ψm(φa) =
1√
2π
eimφa , m ∈ Z, φa ∈ [0, 2π).
• Type b:
Ψα
n,lβ
(θb) = Nα,α
n (sin θb)lβP (α,α)
n (cos θb),
n = l − lβ, α = lβ +
Sβ
2
, n ∈ N0, θb ∈ [0, π]. (B.14)
• Type b′:
Ψβ
n,lα
(θb′) = Nβ,β
n (cos θb′)
lαP (β,β)
n (sin θb′),
n = l − lα, β = lα +
Sα
2
, n ∈ N0, θb′ ∈
[
−π
2
,
π
2
]
. (B.15)
• Type c:
Ψα,β
n,lα,lβ
(ϑc) = 2(α+β)/2+1Nα,β
n (sinϑc)lβ (cosϑc)lαP (β,α)
n (cos 2ϑc),
n =
1
2
(l − lα − lβ) , α = lα +
Sα
2
, β = lβ +
Sβ
2
,
n ∈ N0, ϑc ∈
[
0,
π
2
]
. (B.16)
Fourier, Gegenbauer and Jacobi Expansions 23
We refer to the quantum number n ∈ N0 in (B.14)–(B.16) as the surrogate quantum number
to l ∈ N0.
Notice that the eigenfunctions for branching nodes of type b and b′ can be expressed in terms
of Gegenbauer polynomials using (A.7). Therefore we can re-write (B.14) as
Ψα
n,lβ
(θb) =
(2α)!
Γ (α+ 1)
√
(2α+ 2n+ 1)n!
22α+1(2α+ n)!
(sin θb)lβCα+1/2
n (cos θb), (B.17)
and (B.15) as
Ψβ
n,lα
(θb′) =
(2β)!
Γ (β + 1)
√
(2β + 2n+ 1)n!
22β+1(2β + n)!
(sin θb′)
lαCβ+1/2
n (cos θb′).
Note that even though α, β are not necessarily integers, 2α, 2β ∈ N0.
With the simplest example, polar coordinates (see Fig. 2a, transformation (B.5)), the nor-
malized harmonics are given by
Ym(φ) =
eimφ√
2π
.
In d = 3 there are two equivalent polyspherical coordinate systems, that of type ba (B.7)
and b′a. Since they are equivalent coordinate systems, we treat only the first. This tree has
two branching nodes. Using (B.14) we see that α = m, n = l −m, lβ = m, and Sβ = 0 since
there are no vertices above the branching node m. Through reduction and multiplication by the
φ eigenfunction, the normalized spherical harmonics are
Yl,m(θ, φ) = (−1)m
√
2l + 1
4π
(l −m)!
(l +m)!
Pml (cos θ)eimφ. (B.18)
Notice we have used (A.12) with (B.17) to reduce the Gegenbauer polynomial to a Ferrers
function of the first kind. The functions Yl,m : [0, π]× [0, 2π)→ C are called standard spherical
harmonics.
In d = 4 we consider type b2a (see Fig. 3a) coordinates, whose normalized hyperspherical
harmonics are
Yl1,l2,m(θ1, θ2, φ) =
(−1)m(2l2)!!
π
√
(2l2 + 1)(l1 + 1)(l1 − l2)!(l2 −m)!
2(l1 + l2 + 1)!(l2 +m)!
× (sin θ1)
l2C l2+1
l1−l2(cos θ1)P
m
l2 (cos θ2)e
imφ.
In type ca2 (see Fig. 3e), the normalized hyperspherical harmonics are
Yl,m1,m2(ϑ, φ1, φ2) =
ei(m1φ1+m2φ2)
π
√
l + 1
2
[
1
2(l + |m1|+ |m2|
]
!
[
1
2(l − |m1| − |m2|)
]
![
1
2(l − |m1|+ |m2|)
]
!
[
1
2(l + |m1| − |m2|)
]
!
× (sinϑ)|m2|(cosϑ)|m1|P
(|m2|,|m1|)
(l−|m1|−|m2|)/2(cos 2ϑ),
with the restriction to the parameter space given by 1
2(l−|m1|−|m2|) ∈ N0. Using the surrogate
quantum number n ∈ N0, defined in terms of l ∈ N0 such that 2n = l−|m1|− |m2|, we can more
conveniently express the normalized hyperspherical harmonics in type ca2 coordinates, namely
Yn,m1,m2(ϑ, φ1, φ2) =
ei(m1φ1+m2φ2)
π
√
(2n+ |m1|+ |m2|+ 1)(n+ |m1|+ |m2|)!n!
2(n+ |m1|)!(n+ |m2|)!
24 H.S. Cohl
× (sinϑ)|m2|(cosϑ)|m1|P (|m2|,|m1|)
n (cos 2ϑ).
For arbitrary dimensions, we can use standard polyspherical coordinates (B.11) to construct
the normalized hyperspherical harmonics. The polyspherical harmonics corresponding to this
coordinate system are basis functions for the irreducible representations of O(d) (see [32]). In
terms of these coordinates, the normalized hyperspherical harmonics are
Y K
l (x̂) =
eimφ√
2π
d−2∏
j=1
Θd
j (lj , lj+1; θj), (B.19)
where x̂ ∈ Sd−1, K = {l2, l3, . . . , ld−1}, l = l1 ≥ l2 ≥ l3 ≥· · ·≥ ld−3 ≥ ld−2 = ` ≥ ld−1 = |m| ≥ 0,
and Θd
j : N2
0 × [0, π]→ R is defined by
Θd
j (lj , lj+1; θj) :=
Γ
(
lj+1 + d−j+1
2
)
2lj+1 + d− j − 1
√
22lj+1+d−j−1(2lj + d− j − 1)(lj − lj+1)!
π(lj + lj+1 + d− j − 2)!
× (sin θj)
lj+1C
lj+1+(d−j−1)/2
lj−lj+1
(cos θj). (B.20)
The computation of (B.20) is a straightforward consequence of (B.17), and doing the proper
node counting for Sβ, in the tree depicted in Fig. 4. The normalized spherical harmonics (B.18)
are the simplest example of these harmonics.
In generalized Hopf coordinates of type V2p coordinates with p ≥ 1 (see Fig. 5), the nor-
malized hyperspherical harmonics Y K
l1
: S2p−1 → C can be given more conveniently using the
surrogate quantum numbers nq ∈ N0 which are connected to lq ∈ N0 through the relation
lq = lα + lβ + 2nq such that 1 ≤ q ≤ 2p−1 − 1. The orthonormal polyspherical harmonics, which
can be generated using the method of trees, are
Y K
l1 (x̂) =
ei(m1φ1+···+m2p−1φ2p−1)
√
2π2p−2
p−1∏
j=1
2j−1∏
s=1
Υp
q
(
nq
lα, lβ
;ϑq
)
, (B.21)
where K = {l2, . . . , l2p−1−1,m1, . . . ,m2p−1} and Υp
q : N3
0 × [0, π2 ]→ R is defined by
Υp
q
(
nq
lα, lβ
;ϑq
)
:=
√
(2nq+ α+ β + 1)(nq+ α+ β)!nq!
(nq+ α)!(nq+ β)!
(cosϑq)
lα(sinϑq)
lβP (β,α)
nq (cos(2ϑq)),
with q = 2j−1 + s − 1, lα = l2q, lβ = l2q+1, α = l2q − 1 + 2p−2−blog2 qc, and β = l2q+1 − 1 +
2p−2−blog2 qc. Note that we use the identification
(l2p−1 , . . . , l2p−1) = (|m1| , . . . , |m2p−1 |).
Acknowledgements
I would like to thank A.F.M. Tom ter Elst and Heather Macbeth for valuable discussions.
I would like to express my gratitude to the anonymous referees and an editor at SIGMA whose
helpful comments improved this paper. Part of this work was conducted while H.S. Cohl was
a National Research Council Research Postdoctoral Associate in the Applied and Computational
Mathematics Division at the National Institute of Standards and Technology, Gaithersburg,
Maryland, USA.
Fourier, Gegenbauer and Jacobi Expansions 25
References
[1] Abramowitz M., Stegun I.A., Handbook of mathematical functions with formulas, graphs, and mathematical
tables, National Bureau of Standards Applied Mathematics Series, Vol. 55, U.S. Government Printing Office,
Washington, D.C., 1964.
[2] Alonso Izquierdo A., Fuertes W.G., de la Torre Mayado M., Guilarte J.M., One-loop corrections to the mass
of self-dual semi-local planar topological solitons, Nuclear Phys. B 797 (2008), 431–463, arXiv:0707.4592.
[3] Andrews G.E., Askey R., Roy R., Special functions, Encyclopedia of Mathematics and its Applications,
Vol. 71, Cambridge University Press, Cambridge, 1999.
[4] Boyling J.B., Green’s functions for polynomials in the Laplacian, Z. Angew. Math. Phys. 47 (1996), 485–492.
[5] Cohl H.S., Erratum: Developments in determining the gravitational potential using toroidal functions,
Astronom. Nachr. 333 (2012), 784–785.
[6] Cohl H.S., Fourier and Gegenbauer expansions for fundamental solutions of the Laplacian and powers in Rd
and Hd, Ph.D. thesis, The University of Auckland, Auckland, New Zealand, 2010.
[7] Cohl H.S., Fourier expansions for a logarithmic fundamental solution of the polyharmonic equation,
arXiv:1202.1811.
[8] Cohl H.S., Dominici D.E., Generalized Heine’s identity for complex Fourier series of binomials, Proc. R.
Soc. Lond. Ser. A Math. Phys. Eng. Sci. 467 (2011), 333–345, arXiv:0912.0126.
[9] Cohl H.S., Kalnins E.G., Fourier and Gegenbauer expansions for a fundamental solution of the Laplacian
in the hyperboloid model of hyperbolic geometry, J. Phys. A: Math. Theor. 45 (2012), 145206, 32 pages,
arXiv:1105.0386.
[10] Cohl H.S., Rau A.R.P., Tohline J.E., Browne D.A., Cazes J.E., Barnes E.I., Useful alternative to the
multipole expansion of 1/r potentials, Phys. Rev. A 64 (2001), 052509, 5 pages, arXiv:1104.1499.
[11] Cohl H.S., Tohline J.E., Rau A.R.P., Srivastava H.M., Developments in determining the gravitational po-
tential using toroidal functions, Astronom. Nachr. 321 (2000), 363–372.
[12] Cormen T.H., Leiserson C.E., Rivest R.L., Stein C., Introduction to algorithms, 2nd ed., MIT Press, Cam-
bridge, MA, 2001.
[13] Erdélyi A., Magnus W., Oberhettinger F., Tricomi F.G., Higher transcendental functions, Vol. II, McGraw-
Hill, New York – Toronto – London, 1953.
[14] Fano U., Rau A.R.P., Symmetries in quantum physics, Academic Press Inc., San Diego, CA, 1996.
[15] Gel’fand I.M., Shilov G.E., Generalized functions. Vol. I. Properties and operations, Academic Press, New
York, 1964.
[16] Gradshteyn I.S., Ryzhik I.M., Table of integrals, series, and products, 7th ed., Elsevier/Academic Press,
Amsterdam, 2007.
[17] Ismail M.E.H., Classical and quantum orthogonal polynomials in one variable, Encyclopedia of Mathematics
and its Applications, Vol. 98, Cambridge University Press, Cambridge, 2005.
[18] Izmest’ev A.A., Pogosyan G.S., Sissakian A.N., Winternitz P., Contractions of Lie algebras and separation
of variables. The n-dimensional sphere, J. Math. Phys. 40 (1999), 1549–1573.
[19] Izmest’ev A.A., Pogosyan G.S., Sissakian A.N., Winternitz P., Contractions of Lie algebras and the separa-
tion of variables: interbase expansions, J. Phys. A: Math. Gen. 34 (2001), 521–554.
[20] Kil’dyushov M.S., Hyperspherical functions of the “tree” type in the n-body problem, Soviet J. Nuclear
Phys. 15 (1972), 113–118.
[21] Koekoek J., Koekoek R., The Jacobi inversion formula, Complex Variables Theory Appl. 39 (1999), 1–18,
math.CA/9908148.
[22] Lake M., Thomas S., Ward J., Non-topological cycloops, J. Cosmol. Astropart. Phys. 2010 (2010), no. 1,
026, 27 pages, arXiv:0911.3118.
[23] Lin C.D., Hyperspherical coordinate approach to atomic and other Coulombic three-body systems, Phys.
Rep. 257 (1995), 1–83.
[24] Lin F., Yang Y., Energy splitting, substantial inequality, and minimization for the Faddeev and Skyrme
models, Comm. Math. Phys. 269 (2007), 137–152.
[25] Magnus W., Oberhettinger F., Soni R.P., Formulas and theorems for the special functions of mathematical
physics, 3rd ed., Die Grundlehren der mathematischen Wissenschaften, Bd. 52, Springer-Verlag, New York,
1966.
http://dx.doi.org/10.1016/j.nuclphysb.2007.11.023
http://arxiv.org/abs/0707.4592
http://dx.doi.org/10.1007/BF00916651
http://dx.doi.org/10.1002/asna.201211723
http://arxiv.org/abs/1202.1811
http://dx.doi.org/10.1098/rspa.2010.0222
http://dx.doi.org/10.1098/rspa.2010.0222
http://arxiv.org/abs/0912.0126
http://dx.doi.org/10.1088/1751-8113/45/14/145206
http://arxiv.org/abs/1105.0386
http://dx.doi.org/10.1103/PhysRevA.64.052509
http://arxiv.org/abs/1104.1499
http://dx.doi.org/10.1002/1521-3994(200012)321:5/6<363::AID-ASNA363>3.0.CO;2-X
http://dx.doi.org/10.1063/1.532820
http://dx.doi.org/10.1088/0305-4470/34/3/314
http://dx.doi.org/10.1080/17476939908815177
http://arxiv.org/abs/math.CA/9908148
http://dx.doi.org/10.1088/1475-7516/2010/01/026
http://arxiv.org/abs/0911.3118
http://dx.doi.org/10.1016/0370-1573(94)00094-J
http://dx.doi.org/10.1016/0370-1573(94)00094-J
http://dx.doi.org/10.1007/s00220-006-0123-0
26 H.S. Cohl
[26] Miller Jr. W., Symmetry and separation of variables, Encyclopedia of Mathematics and its Applications,
Vol. 4, Addison-Wesley Publishing Co., Reading, Mass. – London – Amsterdam, 1977.
[27] Olver F.W.J., Lozier D.W., Boisvert R.F., Clark C.W. (Editors), NIST handbook of mathematical functions,
U.S. Department of Commerce National Institute of Standards and Technology, Washington, DC, 2010.
[28] Schwartz L., Théorie des distributions. Tome I, Actualités Sci. Ind., no. 1091, Hermann & Cie., Paris, 1950.
[29] Sloane N.J.A., Plouffe S., The encyclopedia of integer sequences, Academic Press Inc., San Diego, CA, 1995.
[30] Stanley R.P., Enumerative combinatorics. Vol. 2, Cambridge Studies in Advanced Mathematics, Vol. 62,
Cambridge University Press, Cambridge, 1999.
[31] Szegő G., Orthogonal polynomials, American Mathematical Society Colloquium Publications, Vol. 23, Amer.
Math. Soc., Providence, R.I., 1959.
[32] Vilenkin N.Ja., Special functions and the theory of group representations, Translations of Mathematical
Monographs, Vol. 22, Amer. Math. Soc., Providence, R.I., 1968.
[33] Vilenkin N.Ja., Klimyk A.U., Representation of Lie groups and special functions. Vol. 2. Class I representa-
tions, special functions, and integral transforms, Mathematics and its Applications (Soviet Series), Vol. 74,
Kluwer Academic Publishers Group, Dordrecht, 1993.
[34] Vilenkin N.Ja., Kuznetsov G.I., Smorodinskĭı Ya.A., Eigenfunctions of the Laplace operator providing rep-
resentations of the U(2), SU(2), SO(3), U(3) and SU(3) groups and the symbolic method, Soviet J. Nuclear
Phys. 2 (1965), 645–652.
[35] Wen Z.Y., Avery J., Some properties of hyperspherical harmonics, J. Math. Phys. 26 (1985), 396–403.
http://dx.doi.org/10.1063/1.526621
1 Introduction
2 Fundamental solution of the polyharmonic equation in rotationally-invariant and polyspherical coordinate systems
2.1 Fundamental solution decompositions in rotationally-invariant coordinate systems
3 Jacobi polynomial and limiting expansions for the Euler kernel
4 Addition theorems in Vilenkin's polyspherical coordinates
4.1 Power-law addition theorem on Rd for d3 in standard polyspherical coordinates
4.2 Power-law addition theorem on R2q for q2 in generalized Hopf coordinates
4.3 Power-law addition theorems on R3
4.3.1 Type ba coordinates
4.4 Power-law addition theorems on R4
4.4.1 Type b2a coordinates
4.4.2 Type ca2 coordinates
A Special functions and orthogonal polynomials
B Vilenkin's polyspherical coordinates and the method of trees
B.1 Examples of Vilenkin's polyspherical coordinate systems
B.2 Hyperspherical harmonics in polyspherical coordinates
References
|