On a Generalized Oscillator System: Interbasis Expansions

arXiv:quant-ph/9608027v1 16 Aug 1996

MAURICE KIBLER
Institut de Physique Nucléaire de Lyon, IN2P3-CNRS et Université Claude Bernard,
43 Bd du 11 Novembre 1918, F-69622 Villeurbanne Cedex, France

LEVON G. MARDOYAN
Laboratory of Nuclear Problems,
Joint Institute for Nuclear Research, 141980 Dubna, Moscow region, Russia

GEORGE S. POGOSYAN
Bogoliubov Laboratory of Theoretical Physics,
Joint Institute for Nuclear Research, 141980 Dubna, Moscow region, Russia

ABSTRACT
This article deals with a nonrelativistic quantum mechanical study of a dynamical sys-
tem which generalizes the isotropic harmonic oscillator system in three dimensions. The
Schrödinger equation for this generalized oscillator system is separable in spherical, cylin-
drical, and spheroidal (prolate and oblate) coordinates. The quantum mechanical spec-
trum of this system is worked out in some details. The problem of interbasis expansions of
the wavefunctions is completely solved. The coefficients for the expansion of the cylindrical
basis in terms of the spherical basis, and vice-versa, are found to be analytic continuations
(to real values of their arguments) of Clebsch-Gordan coefficients for the group SU(2).
The interbasis expansion coefficients for the prolate and oblate spheroidal bases in terms
of the spherical or the cylindrical bases are shown to satisfy three-term recursion relations.
Finally, a connection between the generalized oscillator system (projected on the z-line)
and the Morse system (in one dimension) is discussed.

0
 Introduction
The purpose of this paper is to study the quantum mechanical motion of a particle in
the three-dimensional axially symmetric potential
Ω2 2 P 1 Q 1
V = (x + y 2 + z 2 ) + + , (1)
2 2 z2 2 x2 + y 2
where Ω, P , and Q are constants with Ω > 0, P > − 41 , and Q ≥ 0. In the last decade,
this potential (including the case P = 0) has been the object of numerous studies [1-12].
The Schrödinger and Hamilton-Jacobi equations for this generalized oscillator potential
are separable in spherical, cylindrical, and spheroidal (prolate and oblate) coordinates.
In the case when P = 0 we get the well-known ring-shape oscillator potential which was
investigated in many articles [1, 3, 4, 7, 8] in recent years as a companion of the Hartmann
potential [4, 6, 8, 13-17]. If P = Q = 0 we have the ordinary isotropic harmonic oscillator
in three dimensions.
The plan of this article is as follows. We solve the Schrödinger equation
HΨ = EΨ, (2)
with
1
H = − (∂xx + ∂yy + ∂zz ) + V, (3)
2
in spherical coordinates (in the second section) and in cylindrical coordinates (in the
third section). (The constant h̄ and the reduced mass are taken to be equal to 1. In
the whole paper, we use Ψ to denote the total wavefunction whatsoever the coordinate
system is; the wavefunctions Ψ in spherical, cylindrical, and spheroidal coordinates are
then distinguished by the corresponding quantum numbers. Note also that we use s to
denote the fraction 21 in the following.) In the fourth section, we determine the interbasis
expansion coefficients between the cylindrical and the spherical bases. The fifth and sixth
sections deal with the generalized oscillator system in spheroidal coordinates. In particular,
the prolate and oblate spheroidal bases are expanded in terms of both the spherical basis
and the cylindrical basis. Two appendices close this article. The first one is devoted to the
bi-orthogonality of the radial wavefunctions (in spherical coordinates) for the generalized
oscillator system. The second appendix concerns a connection between the Smorodinsky-
Winternitz system (that is a basic component for the generalized oscillator system) and
the Morse system.
The generalized oscillator system constitutes a pending part to the generalized Kepler-
Coulomb system studied in Ref. [11, 12, 18]. The latter two nonrelativistic systems gener-
alize two important paradigms in quantum mechanics, namely, the oscillator system and
the Kepler-Coulomb system. The main results of this work and the one in Ref. [18] concern
the separability in spheroidal coodinates as well as the SU(2) approach and the three-term
recursion relations for the interbasis expansion coefficients.
The authors are very pleased to contribute to this memorial volume in honour of
Jean-Louis Calais. Professor Jean-Louis Calais achieved, among other important works,
an original job [19] on the derivation of the SU(2) Clebsch-Gordan coefficients by the
(Löwdin) projection operator method. We are glad to present here a work where an
analytic continuation of SU(2) Clebsch-Gordan coefficients plays an important rôle in the
analysis of interbasis expansions.

1
 The use of spheroidal coordinates is now well established in quantum chemistry [20].
There exist now powerful techniques [21] for evaluating (angular and radial) prolate
spheroidal wavefunctions from differential equations. It is hoped that this paper will
shed some new light on expansions of spheroidal wavefunctions.
Spherical Basis
The Schrödinger equation (2) in spherical coordinates (r, θ, ϕ) for the potential (1),
i.e.,

Ω2 2 P 1 Q 1
V = r + + ,
2 2 2
2 r cos θ 2 r sin2 θ
2

may be solved by seeking a wavefunction Ψ of the form

eimϕ
Ψ(r, θ, ϕ) = R(r)Θ(θ) √ , (4)
2π
with m ∈ Z. This amounts to find the eigenfunctions of the set {H, Lz , M} of commuting
operators, where the constant of motion M reads
P Q
M = L2 + + (5)
cos θ sin2 θ
2

(L2 is the square of the angular momentum and Lz its z-component). We are thus left
with the system of coupled differential equations:

(M − A)Θ = 0, (6)
1 A
 
2 2 2
dr (r dr ) + 2E − Ω r − 2 R = 0, (7)
r2 r
where A is a (spherical) separation constant. √
Let us consider the angular equation (6). By putting Θ(θ) = f (θ)/ sin θ, we can
rewrite Eq. (6) in the Pöschl-Teller form:

1 b2 − 41 c2 − 41
!
dθθ + A + − − f = 0,
4 cos2 θ sin2 θ
(8)
s
1 q
b= P+ , c= Q + m2 .
4
In the case where b > s, the angular potential is repulsive for θ = π2 . In this case, the
θ domain is separated in two regions (θ ∈ ]0, π2 [ and θ ∈ ] π2 , π[) and the “motion” takes
place in one or another region. Furthermore, in this case Eq. (8) corresponds to a genuine
Pöschl-Teller potential. In the case where 0 < b < s, we can call the angular potential
an attractive Pöschl-Teller potential. When b = s, i.e., P = 0, we get the well-known
ring-shape oscillator potential [1, 3, 4, 7, 8]. The solution Θ(θ) ≡ Θq (θ; c, ±b) of Eq. (6)
(for both 0 < b < s and b > s), with the conditions Θ(0) = Θ( π2 ) = 0, is easily found to
be (cf., [22, 23])
Θ(θ) = Nq (c, ±b)(sin θ)c (cos θ)s±bPq(c,±b) (cos 2θ), (9)

2
with q ∈ N, where Pn(α,β) denotes a Jacobi polynomial. Then, the constant A is quantized
as
Aq (c, ±b) = (2q + c ± b + s)(2q + c ± b + 3s). (10)
The normalization constant Nq (c, ±b) in (9) is given (up to a phase factor) by
π
Z
2 1
Θq′ Θq sin θdθ = δq′ q . (11)
0 2
This leads to
v
u (2q + c ± b + 1)q!Γ(q + c ± b + 1)
u
Nq (c, ±b) = t . (12)
Γ(q + c + 1)Γ(q ± b + 1)

Note that only the positive sign in front of b has to be taken when b > s while both the
positive and negative signs have to be considered for 0 < b < s.
Let us go to the radial equation (7). The introduction of (10) into (7) yields an
equation that is very reminiscent of the radial equation for the three-dimensional isotropic
oscillator except that the orbital quantum number l is replaced by 2q + c ± b + s. The
solution R(r) ≡ Rnr q (r; c, ±b) of the obtained equation, in terms of Laguerre polynomials
Lαn , is √ 2
R(r) = Nnr q (c, ±b)( Ωr)2q+c±b+s e−sΩr L2q+c±b+1
nr (Ωr 2 ), (13)
with nr ∈ N. In Eq. (13), the radial wavefunctions Rnr q satisfy the orthogonality relation
Z ∞
Rn′r q Rnr q r 2 dr = δn′r nr (14)
0

[cf., Eq. (71)] so that the normalization factor Nnr q (c, ±b) is
v
2Ω3s nr !
u
u
Nnr q (c, ±b) = t . (15)
Γ(nr + 2q + c ± b + 2)

The normalized total wavefunction Ψ(r, θ, ϕ) ≡ Ψnr qm (r, θ, ϕ; c, ±b) is then given by
Eqs. (4), (9), (12), (13), and (15). The energies E corresponding to nr + q fixed are

En (c, ±b) = Ω (2n + c ± b + 2), (16)

with n = nr + q. Equation (16) shows that, for each quantum number n, we have two
levels (for +b and −b) in the 0 < b < s region and one level (for +b) in the b > s region.
Note that this spectrum was obtained through a path integral approach in [2, 6] for the
b > s case and in [12] for the general case (see also Refs. [10, 11]).
In the 0 < b < s region, for the limiting situation where b = s− , i.e., P = 0− , we have
for the separation constant A:

Aq (c, +s) = (2q + c + 1)(2q + c + 2),

(17)
Aq (c, −s) = (2q + c)(2q + c + 1).

3
Then, by using the connecting formulas [24]

(λ)n+1
λ
C2n+1 (x) = x Pn(λ−s,+s)(2x2 − 1),
(s)n+1
λ (λ)n (λ−s,−s) 2
C2n (x) = P (2x − 1),
(s)n n

between the Jacobi polynomial Pn(α,β) and the Gegenbauer polynomial Cnλ , we have the
following odd and even angular solutions (with respect to cos θ 7→ − cos θ)
v
u (4q + 2c + 3)(2q + 1)!
u
Θq (θ; c, +s) = t
2πΓ(2q + 2c + 2)
×2c Γ(c + s)(sin θ)c C2q+1
c+s
(cos θ), (18)
v
u (4q + 2c + 1)(2q)!
u
Θq (θ; c, −s) = t
2πΓ(2q + 2c + 1)
×2c Γ(c + s)(sin θ)c C2q
c+s
(cos θ). (19)

Let us introduce (a new orbital quantum number) l and (a new principial quantum number)
N through
2q + 1 for the + sign
 
l − |m| = ,
2q for the − sign
(20)
2n + 1 for the + sign
 
N − |m| = .
2n for the − sign

Note that N = 2nr +l both for the + and − signs. Then, the separation constant [Eq. (17)]
and the energy [Eq. (16)] can be expressed as

Aq (c, ±s) ≡ Al (δ) = (l + δ)(l + δ + 1),

(21)
En (c, ±s) ≡ EN (δ) = Ω (N + δ + 3s),

respectively, where
q
δ= Q + m2 − |m|.

Thus, the two parts of the energy spectrum for the signs ± correspond now to odd
(for +) and even (for −) values of N − |m|. In terms of N, l, and δ, the functions
Rnr q (r; c, ±s) ≡ RN l (r; δ) [cf., Eq. (13)] and Θq (θ; c, ±s) ≡ Θlm (θ; δ) [cf., Eqs. (18) and
(19)] can be rewritten as
v
2Ω3s ( N2−l )!
u
u
RN l (r; δ) = t
Γ( N2+l + δ + 3s)
√ 2
×( Ωr)l+δ e−sΩr Ll+δ+s 2
N−l (Ωr ), (22)
2

4
 Θlm (θ; δ) = 2|m|+δ Γ(|m| + δ + s)
v
u (2l + 2δ + 1)(l − |m|)!
u
×t
2πΓ(l + |m| + 2δ + 1)
|m|+δ+s
×(sin θ)|m|+δ Cl−|m| (cos θ). (23)
Equations (22) and (23 compare with the corresponding results for the ring-shape oscillator
in [1, 3]. Note that (23) was given in terms of Legendre functions in Refs. [1] and [3] and
was studied in details in Ref. [9].
In the b > s region, for the limiting situation where b = s+ , i.e., P = 0+ , we have only
odd solutions. In other words when P → 0+ , the eigenvalues and eigenfunctions of the
generalized oscillator do not restrict to the eigenvalues and eigenfunctions, respectively, of
the ring-shape oscillator. This fact may be explained in the following manner. To make
P = 0 in the wavefunction Ψnr qm (r, θ, ϕ; c, +b) amounts to changing the Hamiltonian
into a Hamiltonian corresponding to P = 0 and to introducing an unpenetrable barrier.
(Another way to describe this phenomenon is to say that for very small P , the potential
V is infinite in the θ = π2 plan and equal to the ring-shape potential only for P = 0.) This
phenomenon is known as the Klauder phenomenon [25].
A further limit can be obtained in the case when δ = 0, i.e., Q = 0. It is enough to
use the connecting formula [24]

|m| (−2)|m| |m|+s

Pl (x) = √ Γ(|m| + s)(1 − x2 )s|m| Cl−|m| (x)
π
|m|
between the Gegenbauer polynomial Cnλ and the Legendre polynomial Pl . In fact for
Q = 0, Eq. (23) can be reduced to
v
|m| t 2l + 1 (l − |m|)! |m|
u
u
Θlm (θ; 0) = (−1) P (cos θ),
2 (l + |m|)! l
√
so that Θlm (θ; 0)eimϕ / 2π coincides with the usual spherical harmonic
√ Ylm (θ, ϕ) (up to a
phase factor, e.g., see [26]). The wavefunctions Θq (θ; c, ±b)eimϕ / 2π may thus be consid-
ered as a generalisation of the spherical harmonics.
Cylindrical Basis
In the cylindrical coordinates (ρ, ϕ, z), the potential V reads
Ω2 2 P 1 Q1
V = (ρ + z 2 ) + 2
+ .
2 2z 2 ρ2
Equation (2), with this potential, admits a solution Ψ of the form
eimϕ
Ψ(ρ, ϕ, z) = R(ρ)Z(z) √ , (24)
2π
where m ∈ Z. In other words, we look for the eigenfunctions of the set {H, Lz , N} of
commuting operators, where the constant of motion N is
P
N = Dzz + , (25)
z2

5
Dzz being the zz component of

Dzi zj = −∂zi zj + Ω2 zi zj ,

the so-called Demkov tensor [27] for the isotropic harmonic oscillator in R3 . It is sufficient
to solve the two coupled equations

(N − 2Ez )Z = 0, (26)
Q + m2
" #
1 2 2
dρ (ρdρ ) + 2Eρ − Ω ρ − R = 0, (27)
ρ ρ2

where the two cylindrical separation constants Eρ and Ez obey Eρ +Ez = E. The solutions
Ψ(ρ, ϕ, z) ≡ Ψnρ pm (ρ, ϕ, z; c, ±b) of (26-27) lead to the normalized wavefunction

eimϕ
Ψ(ρ, ϕ, z) = Rnρ (ρ; c)Zp (z; ±b) √ , (28)
2π
where
v
2Ωnρ ! 2 √
u
e−sΩρ ( Ωρ)c Lcnρ (Ωρ2 )
u
Rnρ (ρ; c) = t
Γ(nρ + c + 1)

and

2 √
s
Ωs p!
Zp (z; ±b) = (−1)p e−sΩz ( Ωz)s±b L±b 2
p (Ωz ), (29)
Γ(p ± b + 1)

with nρ ∈ N and p ∈ N. The normalization of the wavefunction (28) is ensured by

Z ∞
Rn′ρ Rnρ ρdρ = δn′ρ nρ ,
0
∞ 1
Z
Zp′ Zp dz = δp′ p .
0 2
Furthermore, the constants Eρ and Ez in (26-27) become

Eρ (nρ , c) = Ω (2nρ + c + 1),

(30)
Ez (p, ±b) = Ω (2p ± b + 1).

Therefore, the quantized values of the energy E are given by (16) where now the quantum
number n is n = nρ + p. As in the second section, the sign in front of b in Eqs. (28)-(30)
may be only positive when b > s. When 0 < b < s, both the signs + and − are admissible.
In the 0 < b < s region, in the limiting case where b = s− , due to the connecting
formulas [24]

H2n+1 (x) = (−1)n 22n+1 n!xL+s 2

n (x ),
H2n (x) = (−1)n 22n n!L−s 2
n (x ),

6
between the odd H2n+1 and even H2n Hermite polynomials and the Laguerre polynomials
L±s
n , we immediately have

 14 2
√
Ω e−sΩz

Zp (z; +s) = q H2p+1 ( Ωz),
π 22p+1 (2p + 1)!
 14 2
√
Ω e−sΩz

Zp (z; −s) = q H2p ( Ωz).
π 22p (2p)!

Introducing (a new quantum number) n3 such that n3 = 2p + 1 for the + sign and n3 = 2p
for the − sign, we obtain
 41 2
√
Ω e−sΩz

Zp (z; ±s) = √ n Hn3 ( Ωz).
π 2 3 n3 !

The energy is then given by (21) where N = 2nρ + n3 + |m|. Note that the spectrum in
the case b = s− , which corresponds to the ring-shape oscillator system, was obtained in
Refs. [1-3].
In the b > s region, in the limiting situation where b = s+ , we get only the odd solution
of the ring-shape oscillator system.
Connecting the Cylindrical and Spherical Bases
According to first principles, any cylindrical wavefunction (24) corresponding to a given
value of E can be developed in terms of the spherical wavefunctions (4) associated to the
eigenvalue E (see also Ref. [11]). Thus, we have
n
q
X
Ψnρ pm = Wnp (c, ±b)Ψnr qm , (31)
q=0

where nρ + p = nr + q = n. In Eq. (31), it is understood that the wavefunctions in the

left- and right-hand sides are written in spherical coordinates (r, θ, ϕ) owing to ρ = r sin θ
and z = r cos θ. The dependence on eimϕ can be eliminated in both sides of Eq. (31).
Furthermore, by using the formula Lαn (x) ∼ (−1)n xn /n!, valid for x arbitrarily large, (31)
yields an equation that depends only on the variable θ. Thus, by using the orthonormality
relation (11), for the quantum numbers q, we can derive the following expression for the
interbasis expansion coefficients
q
Wnp (c, ±b) = (−1)q−p Bnp
q q
(c, ±b) Enp (c, ±b), (32)

where
v
(2q + c ± b + 1)(n − q)!q!Γ(q + c ± b + 1)Γ(n + q + c ± b + 2)
u
q
u
Bnp (c, ±b) = t
(n − p)!p!Γ(q + c + 1)Γ(q ± b + 1)Γ(n − p + c + 1)Γ(p ± b + 1)
(33)
Z π
2
q
Enp (c, ±b) = 2 (sin θ)2n−2p+2c (cos θ)2p+1±2b Pq(c,±b) (cos 2θ) sin θdθ.
0

7
By making the change of variable x = cos 2θ and by using the Rodrigues formula for the
Jacobi polynomial [24]

(−1)n −β d
n
Pn(α,β) (x) = (1 − x) −α
(1 + x) [(1 − x)α+n (1 + x)β+n ],
2n n! dxn
Eqs. (32)-(33) lead to the integral expression

q (−1)p Z 1 dq
Wnp (c, ±b) = (1 − x)n−p (1 + x)p [(1 − x)q+c (1 + x)q±b ]dx
2n+q+c±b+1 −1 dx q
v
(2q + c ± b + 1)(n − q)!Γ(q + c ± b + 1)Γ(n + q + c ± b + 2)
u
u
×t (34)
p!q!(n − p)!Γ(q + c + 1)Γ(q ± b + 1)Γ(n − p + c + 1)Γ(p ± b + 1)
q
for the coefficient Wnp (c, ±b). Equation (34) can be compared with the integral represen-
tation [26]
v
(2c + 1)(J + 1)!(J − 2c)!(c + γ)!
u
u
(abαβ|cγ) = δα+β,γ t
(J − 2a)!(J − 2b)!(a − α)!(a + α)!(b − β)!(b + β)!(c − γ)!
(−1)a−c+β 1 dc−γ
Z
× (1 − x)a−α (1 + x)b−β [(1 − x)J−2a (1 + x)J−2b ]dx
2J+1 −1 dxc−γ
cγ
(with J = a + b + c) for the Clebsch-Gordan coefficients Caα;bβ ≡ (abαβ|cγ) of the compact
Lie group SU(2). This yields
q
Wnp (c, ±b) = (−1)n−q (a0 b0 αβ|c0 , α + β),
n±b n+c c±b
a0 = , b0 = , c0 = q + , (35)
2 2 2
n∓b n+c
α=p− , β= − p.
2 2
Since the quantum numbers in Eq. (35) are not necessarily integers or half of odd integers,
the coefficients for the expansion of the cylindrical basis in terms of the spherical basis
may be considered as analytical continuation, for real values of their arguments, of the
SU(2) Clebsch-Gordan coefficients. The inverse of Eq. (31), namely
n
p
X
Ψnr qm = W̃nq (c, ±b) Ψnρ pm (36)
p=0

follows from the orthonormality property of the SU(2) Clebsch-Gordan coefficients. The
expansion coefficients in (36) are thus
p q
W̃nq (c, ±b) = Wnp (c, ±b).
q
Note that in order to compute the coefficients Wnp (c, ±b) through (35), we can use the
F
3 2 (a, b, c; d, e; 1) representation [26] of the SU(2) Clebsch-Gordan coefficients.
We close this section with some considerations concerning the limiting cases (P = 0,
Q 6= 0) and (P = 0, Q = 0). It is to be observed that the passage from (P 6= 0, Q 6= 0)

8
to (P = 0, Q 6= 0) needs some caution. Indeed for b = s− , Eq. (35) can be rewritten in
terms of the quantum numbers N, l, and n3 as
N−l
q
Wnp (c, ±s) = (−1) 2(a0 b0 αβ|c0 , α + β),
N − |m| − s ± s
a0 = ,
4
N + |m| − s ∓ s δ
b0 = + ,
4 2
(37)
2l − 1 δ
c0 =
+ ,
4 2
2n3 − N + |m| − s ± s
α= ,
4
−2n3 + N + |m| + s ± s δ
β= + .
4 2
By using the ordinary symmetry property [26]

(abαβ|cγ) = (−1)a+b−c (ab, −α, −β|c, −γ)

and the Regge symmetry [26]

!
a+b+γ a+b−γ a−b+α−β a−b−α+β
(abαβ|cγ) = , , , |c, a − b
2 2 2 2

in Eq. (37) with the sign + and by using the ordinary symmetry property [26]

(abαβ|cγ) = (−1)a+b−c (baβα|cγ)

in Eq. (37) with the sign −, we get

q
Wnp (c, ±s) ≡ WNl mn3 (δ) = (a0 b0 αβ|c0, α + β),
N + |m| δ N − |m| − 1 2l − 1 δ
a0 = + , b0 = , c0 = + , (38)
4 2 4 4 2
N + |m| − 2n3 δ 2n3 − N + |m| − 1
α= + , β= .
4 2 4
As a conclusion, when b = s− we have an expansion of the type [3]

WNl mn3 (δ) ΨN lm (r, θ, ϕ; δ),

X
ΨN mn3 (ρ, ϕ, z; δ) = (39)
l

where the summation on l goes, by steps of 2, from |m| or |m| + 1 to N according to

whether as N − |m| is even or odd (because N − l is always even). Equations (38)-(39)
were obtained in Ref. [3] for the ring-shape oscillator system. Finally, the case P = Q = 0
can be easily deduced from (38)-(39) by taking δ = 0: we thus recover the result obtained
in Refs. [28, 29] for the isotropic harmonic oscillator in three dimensions. Note that in the
case P = Q = 0, the expansion coefficients in Eq. (39) become Clebsch-Gordan coefficients
for the noncompact Lie group SU(1,1) (cf., Ref. [30]).

9
 Prolate and Oblate Spheroidal Bases
SEPARATION IN PROLATE SPHEROIDAL COORDINATES
The prolate spheroidal coordinates (ξ, η, ϕ) are defined via

Rq 2
x = (ξ − 1)(1 − η 2 ) cos ϕ,
2
Rq 2
y = (ξ − 1)(1 − η 2 ) sin ϕ,
2
R
z = ξη,
2
(with 1 ≤ ξ < ∞, −1 ≤ η ≤ 1, and 0 ≤ ϕ < 2π), where R is the interfocus distance. As is
well-known [31], in the limits where R → 0 and R → ∞, the prolate spheroidal coordinates
reduce to the spherical coordinates and the cylindrical coordinates, respectively. In prolate
spheroidal coordinates, the potential V reads

Ω2 R2 2
" #
2 P Q
V = (ξ + η 2 − 1) + 2 2 2 + 2 . (40)
8 R ξ η (ξ − 1)(1 − η 2 )

By looking for a solution Ψ of Eq. (2), with the potential (40), in the form

eimϕ
Ψ(ξ, η, ϕ) = ψ1 (ξ)ψ2 (η) √ , (41)
2π
with m ∈ Z, we obtain the two ordinary differential equations

Q + m2 ER2 2 Ω2 R4 2 2
" #
2 P
dξ (ξ − 1)dξ − 2 + ξ − ξ (ξ − 1) + 2 ψ1 = +λ(R)ψ1 , (42)
ξ −1 2 16 ξ

Q + m2 ER2 2 Ω2 R4 2
" #
2 P
dη (1 − η )dη − 2
− η − η (1 − η 2 ) − 2 ψ2 = −λ(R)ψ2 , (43)
1−η 2 16 η
where λ(R) is a separation constant in prolate spheroidal coordinates. The combination
of Eqs. (42) and (43), leads to the operator
1 h
2 2 2 2
i
Λ= − η ∂ξ (ξ − 1)∂ξ + ξ ∂η (1 − η )∂η
ξ 2 − η2
ξ 2 + η2 − 1
+ (Q − ∂ϕϕ )
(ξ 2 − 1)(1 − η 2 )
Ω2 R4 2 2 ξ 2 + η2
+ ξ η +P 2 2
16 ξ η

after eliminating the energy E. The eigenvalues of the operator Λ are λ(R) while its
eigenfunctions are given by (41). The significance of the (self-adjoint) operator Λ is to be
found in the connecting formula

R2
Λ=M+ N, (44)
4

10
where M and N are the constants of motion (5) and (25). The operator Λ is of pivotal
importance for the derivation of the interbasis expansion coefficients from the spherical
basis or the cylindrical basis to the prolate spheroidal basis. In particular, it allows us
to derive the latter coefficients without knowing the wavefunctions in prolate spheroidal
basis. (In this respect, credit should be put on the work by Coulson and Joseph [32]
who considered an operator similar to Λ for the hydrogen atom.) Therefore, we shall
not derive the prolate spheroidal wavefunctions ψ1 and ψ2 which could be obtained by
solving Eqs. (42) and (43). It is more economical to proceed in the following way that
presents the advantage of giving, at the same time, the global wavefunction Ψ(ξ, η, ϕ) ≡
Ψ(ξ, η, ϕ; R, c, ±b) and the interbasis expansion coefficients.
INTERBASIS EXPANSIONS FOR
THE PROLATE SPHEROIDAL WAVEFUNCTIONS
The three constants of motion M, N, and Λ, which occur in Eq. (44), can be seen to
satisfy the following eigenequations

MΨnr qm = Aq (c, ±b)Ψnr qm , (45)

NΨnρ pm = 2Ez (p, ±b)Ψnρ pm , (46)

and
ΛΨnkm = λk (R)Ψnkm (47)
for the spherical, cylindrical, and prolate spheroidal bases, respectively. [In Eq. (47), the
index k labels the eigenvalues of the operator Λ and varies in the range 0 ≤ k ≤ n.] The
spherical, cylindrical, and prolate spheroidal bases are indeed eigenbases for the three sets
of commuting operators {H, Lz , M}, {H, Lz , N}, and {H, Lz , Λ}, respectively. We are now
in a position to deal with the interbasis expansions
n
p
X
Ψnkm = Unk (R; c, ±b)Ψnρ pm (48)
p=0
n
q
X
Ψnkm = Tnk (R; c, ±b)Ψnr qm (49)
q=0

for the prolate spheroidal basis in terms of the cylindrical and spherical bases.
First, we consider Eq. (48). Let the operator Λ act on both sides of (48). Then, by
using Eqs. (44), (46), and (47) along with the orthonormality property of the cylindrical
basis, we find that
1 R2
 
p
λk (R) − Ez (p, ±b) Unk (R; c, ±b)
2 2
n
X p ′ (±)
= Unk (R; c, ±b)Mpp′ , (50)
p′ =0

where Z ∞ Z 2π Z ∞
(±)
Mpp′ = Ψ∗nρ pm MΨnρ p′m ρdρdϕdz. (51)
0 0 0
(±)
The calculation of the matrix element Mpp′ can be done by expanding the cylindrical
wavefunctions in (51) in terms of spherical wavefunctions [see Eq. (31)] and by making

11
use of the eigenvalue equation for M [see Eq. (45)]. This leads to
n
(±) 1X q q
Mpp′ = Aq (c, ±b)Wnp (c, ±b)Wnp ′ (c, ±b). (52)
2 q=0
To calculate the sum in Eq. (52), we need some recursion relation for the coefficient
q
Wnp (c, ±b) involving p − 1, p, and p + 1. Owing to Eq. (35), this amounts to use the
following recursion relations [33]:
[−a(a + 1) − b(b + 1) + c(c + 1) − 2αβ](abαβ|cγ)
q
= (a + α)(a − α + 1)(b − β)(b + β + 1)
×(a, b, α − 1, β + 1|cγ)
q
+ (a − α)(a + α + 1)(b + β)(b − β + 1)
×(a, b, α + 1, β − 1|cγ). (53)
Then, by introducing Eq. (53) into Eq. (52) and by using the orthonormality condition
(abαβ|cγ)(abα′ β ′|cγ) = δαα′ δββ ′ ,
X
(54)
c,γ

(±)
we find that Mpp′ is given by
(±)
Mpp′ = 2[p(p ± b)(n − p + 1)(n + c − p + 1)]s δp′ ,p−1
+[s (c ∓ b + s) (c ∓ b + 3s) + 2(p + 1)(n − p) + 2(p ± b)(n + c − p + 1)]δp′ p
+2[(p + 1)(p + 1 ± b)(n − p)(n + c − p)]s δp′ ,p+1. (55)
Now by introducing (55) into (50), we get the following three-term recursion relation
[(p + 1)(n − p) + (p ± b)(n + c − p + 1)
1 R2 1 p
+ (c ∓ b + s)(c ∓ b + 3s) + Ez (p, ±b) − λk (R)]Unk
4 8 4
+[(p + 1)(p + 1 ± b)(n − p)(n + c − p)]s Unk
p+1

+[p(p ± b)(n − p + 1)(n + c − p + 1)]s Unkp−1

=0 (56)
q q
for the expansion coefficients Unk ≡ Unk (R; c, ±b). The recursion relation (56) provides us
with a system of n + 1 linear homogeneous equations which can be solved by taking into
account the normalization condition
n
p
(R; c, ±b)|2 = 1.
X
|Unk
p=0

The eigenvalues λk (R) of the operator Λ then follow from the vanishing of the determinant
for the latter system.
Second, let us concentrate on the expansion (49) of the prolate spheroidal basis in
terms of the spherical basis. By employing a technique similar to the one used for deriving
Eq. (50), we get
q
[λk (R) − Aq (c, ±b)] Tnk (R; c, ±b)
2 Xn
R q′ (±)
= Tnk (R; c, ±b) Nqq′ , (57)
2 q′ =0

12
where
π
Z ∞Z Z 2π
(±) 2
Nqq′ = Ψ∗nr qm NΨnr q′ m r 2 sin θdrdθdϕ.
0 0 0

(±) (±)
The matrix elements Nqq′ can be calculated in the same way as Mpp′ except that we must
use the relation [26]
#s
c2 (2c + 1)(2c − 1)
"
(abαβ|cγ) = − 2
(c − γ 2 )(−a + b + c)(a − b + c)(a + b − c + 1)(a + b + c + 1)
(" #s
(c − γ − 1)(c + γ − 1)(−a + b + c − 1)(a − b + c − 1)(a + b − c + 2)(a + b + c)
×
(c − 1)2 (2c − 3)(2c − 1)
)
(α − β)c(c − 1) − γa(a + 1) + γb(b + 1)
×(abαβ|c − 2, γ) − (abαβ|c − 1, γ)
c(c − 1)

and the orthonormality condition

(abαβ|cγ)(abαβ|c′ γ ′ ) = δc′ c δγ ′ γ ,
X

α,β

instead of Eqs. (53) and (54). This produces the matrix element

(±) 2q(q + 1) + (c ± b)(2q ± b + 1)

Nqq′ = En (c, ±b) δq′ q
(2q + c ± b)(2q + c ± b + 2)
−2Ω[Aq+1 q
n (c, ±b)δq ′ ,q+1 + An (c, ±b)δq ′ ,q−1 ], (58)

where
" #s
q(n − q + 1)(q + c ± b)(q ± b)(q + c)(n + q + c ± b + 1)
Aqn (c, ±b) = .
(2q + c ± b)2 (2q + c ± b − 1)(2q + c ± b + 1)

Finally, the introduction of (58) into (57) leads to the three-term recursion relation

R2
"
λk (R) − Aq (c, ±b) −
En (c, ±b)
2
#
2q(q + 1) + (c ± b)(2q ± b + 1) q
× Tnk
(2q + c ± b)(2q + c ± b + 2)
h i
q+1 q−1
+ΩR2 Aq+1 q
n (c, ±b)Tnk + An (c, ±b)Tnk =0 (59)
p p
for the expansion coefficients Tnk ≡ Tnk (R; c, ±b). This relation can be iterated by taking
into account the normalization condition
n
q
(R; c, ±b)|2 = 1.
X
|Tnk
q=0

Here again, the eigenvalues λk (R) may be obtained from the vanishing of the determinant
of a system of n + 1 linear homogeneous equations.

13
 LIMITING CASES
Putting b = s− , i.e., P = 0− , in the matrix element (58) with Q 6= 0 and by using (20),
we have

(±) 2Al (δ) − 2(|m| + δ)2 − 1

Nqq′ = EN (δ) δl′ l
(2l + 2δ − 1)(2l + 2δ + 3)
−2Ω[Al+2 l
N (δ)δl′ ,l+2 + AN (δ)δl′ ,l−2 ],

where
" #s
l− (l− − 1)(l+ + 2δ)(l+ + 2δ − 1)(N − l + 2)(N + l + 2δ + 1)
AlN (δ) =
4(2l + 2δ − 1)2 (2l + 2δ − 3)(2l + 2δ + 1)

(with l± = l ± |m|) and finally we get the following three-term recursion relation

R2 2Al (δ) − 2(|m| + δ)2 − 1

" #
λk (R) − Al (δ) − EN (δ) T l (R; δ)
2 (2l + 2δ − 1)(2l + 2δ + 3) N k
h i
+ΩR2 Al+2 l+2 l l−2
N (δ)TN k (R; δ) + AN (δ)TN k (R; δ) = 0

for TNl k (R; δ) ≡ TNl k (R; c, ±s). By analogy it is easy to obtain a three-term recursion
relation for the interbasis expansion coefficients UNn3k (R; δ) ≡ UNn3k (R; c, ±s). We get

ΩR2
[(2n3 + 1)(N − n3 + δ + 1) + (|m| + δ)2 − 1 + (2n3 + 1) − λk (R)]UNn3k (R; δ)
4
+[(n3 + 1)(n3 + 2)(N − |m| − n3 )(N + |m| − n3 + 2δ)]s UNn3k+2 (R; δ)
+[n3 (n3 − 1)(N − |m| − n3 + 2)(N + |m| − n3 + 2δ + 2)]s Unk
n3 −2
(R; δ) = 0.

Consequently, when b = s− we have the expansions [cf., Eqs. (48) and (49)]
N
UNn3k (R; δ)ΨN mn3 ,
X
ΨN km =
n3
N
TNl k (R; δ)ΨN lm ,
X
ΨN km =
l

for the ring-shape oscillator. The summations on l and n3 go, by steps of 2, from |m| or
|m| + 1 to N and from 0 or 1 to N − |m| according to whether as N − |m| is even or odd
(because N − l and N − |m| − n3 are always even).
The next limiting case δ = 0, i.e., Q = 0, is trivial and the corresponding results for
the isotropic harmonic oscillator agree with the ones obtained in Ref. [34].
Finally, it should be noted that the following two limits
p p
lim Unk (R; c, ±b) = W̃nk (c, ±b),
R→0
q q
lim Tnk (R; c, ±b) = Wnk (c, ±b)
R→∞

furnish a useful means for checking the calculations presented in the fourth and fifth
sections.

14
 SEPARATION AND INTERBASIS EXPANSIONS FOR
THE OBLATE SPHEROIDAL WAVEFUNCTIONS
The oblate spheroidal coordinates (ξ, η, ϕ) are defined by

R
q
2
x = (ξ + 1)(1 − η 2 ) cos ϕ,
2
R
q
2
y = (ξ + 1)(1 − η 2 ) sin ϕ,
2
R
z = ξ η,
2
(with 0 ≤ ξ < ∞, −1 ≤ η ≤ 1, and 0 ≤ ϕ < 2π), where R is the interfocus distance in the
oblate spheroidal coordinate system. As in the prolate system, in the limits R → 0 and
R → ∞, the oblate spheroidal coordinates give the spherical and cylindrical coordinates,
respectively [31, 34].
The potential V , the Schrödinger equation, the oblate spheroidal constant of motion
Λ, and the interbasis expansion coefficients for the oblate spheroidal coordinates can be
obtained from the corresponding expressions for the prolate spheroidal coordinates by
means of the trick: ξ → −iξ and R → iR.
Spheroidal Corrections for the Spherical and Cylindrical Bases
As we have already mentioned, the spheroidal system of coordinates is one of the
most general one-parameter systems of coordinates which contains spherical and cylindri-
cal coordinates as some limiting cases. Accordingly, the prolate spheroidal basis of the
generalized oscillator as R → 0 and R → ∞ degenerates into the spherical and cylindrical
bases that can be treated as zeroth order approximations in some perturbation series. The
three-term recursion relations for the expansion coefficients of the prolate spheroidal basis
in the cylindrical and spherical bases, which have been obtained in the fifth section, may
serve as a basis for constructing an algebraic perturbation theory, respectively, at large
(R ≫ 1) and small (R ≪ 1) values of the interfocus distance R. Thus it is possible to
derive prolate spheroidal corrections for the spherical and cylindrical bases.
THE CASE R ≪ 1
Let us rewrite the three-term recursion relation (59) in the following form
q
[λk (R) − Aq (c, ±b) − ΩR2 Bnq (c, ±b)]Tnk
h i
q+1 q−1
+ΩR2 Aq+1 q
n (c, ±b)Tnk + An (c, ±b)Tnk = 0, (60)

where
1
Bnq (c, ±b) = (2n + c ± b + 2)
2
2q(q + 1) + (c ± b)(2q ± b + 1)
× .
(2q + c ± b)(2q + c ± b + 2)

The zeroth order approximation for the separation constant λk (R) and the coefficients
p
Tnk (R; c, ±b) can immediately be derived from the recursion relation (60). Indeed, from

15
Eq. (60), we obtain

lim λk (R) = Ak (c, ±b),

R→0
q
lim Tnk (R; c, ±b) = δkq ,
R→0

so that, for the wavefunction, we have

lim Ψnkm (ξ, η, ϕ; R, c, ±b) = Ψnr km (r, θ, ϕ, z; c, ±b).

R→0

As is seen from these limiting relations, the quantum number k, labeling the spheroidal
separation constant and being (according to the oscillation theorem [31]) the number of
zeros of the prolate angular spheroidal function ψ2 (η) in the interval −1 ≤ η ≤ 1, turns
into a spherical quantum number determining the number of zeros of the angular function
(9). It is clear that this fact is a consequence of the independence of the number of zeros
of the wavefunction on R.
In order to calculate higher order corrections, we represent the interbasis coefficients
q
Tnk (R; c, ±b) and the spheroidal separation constant λk (R) as expansions in powers of
ΩR2 : ∞
q (j)
Tkq (ΩR2 )j ,
X
Tnk (R; c, ±b) = δkq + (61)
j=1
∞
(j)
λk (ΩR2 )j .
X
λk (R) = Ak (c, ±b) + (62)
j=1

Substituting Eqs. (61) and (62) into the three-term recursion relation (60) and equating
the coefficients with the same power of R, we arrive at the equation for the coefficients
(j) (j)
Tkq and λk
(j)
4(k − q)(k + q + c ± b + 1)Tkq =
(j−1) (j−1)
−Aq+1 q
n (c, ±b)Tk,q+1 + Bn (c, ±b)Tkq
j−1
(j−1) (j−t) (t)
−Aqn (c, ±b)Tk,q−1
X
− λk Tkq . (63)
t=0

(0) (j)
Equation (63) with the initial condition Tkq = δkq and the condition Tqq = δj0 arising
(j)
in the standard perturbation theory [35] allow us to derive a formula expressing λk for
(j−1) (j−1)
j ≥ 1 through the coefficients Tkk and Tk,k±1 :
(j) (j−1) (j−1) (j−1)
λk = −Ak+1 k
n (c, ±b)Tk,k+1 + Bn (c, ±b)Tk,k − Akn (c, ±b)Tk,k−1 . (64)
(j) (j)
This gives a possibility to determine, step by step, the coefficients λk and Tkq in Eqs. (61)
and (62). As an example, let us write down the first and second order corrections in (62)
q
for λk (R) and the first order correction in (61) for Tnk (R; c, ±b). It follows from Eq. (64)
that
(1)
λk = Bnk (c, ±b),
(2) (1) (1)
λk = −Ak+1 k
n (c, ±b)Tk,k+1 − An (c, ±b)Tk,k−1

16
and Eq. (63) for j = 1 results in

(1) Akn (c, ±b) Ak+1

n (c, ±b)
Tkq = − δq,k−1 + δq,k+1 . (65)
4(2k + c ± b) 4(2k + c ± b + 2)
Thus, for the spheroidal separation constant, with an accuracy up to the term (ΩR2 )2 , we
get

λk (R) = Ak (c, ±b) + ΩR2 Bnk (c, ±b)

2 2
Ω2 R4 Akn (c, ±b) Ak+1
" #
(c, ±b)
+ − n .
4 2k + c ± b 2k + c ± b + 2
Introducing (65) into (61) and then using (49) for the expansion of the prolate spheroidal
basis over the spherical one, we get the following approximate formula

Ψnkm (ξ, η, ϕ; R, c, ±b) = Ψnkm(r, θ, ϕ; c, ±b)

Ω2 R4 Akn (c, ±b)
"
− Ψn,k−1,m(r, θ, ϕ; c, ±b)
4 2k + c ± b
Ak+1
#
n (c, ±b)
− Ψn,k+1,m(r, θ, ϕ; c, ±b) .
2k + c ± b + 2

THE CASE R ≫ 1
Now let us consider the case R ≫ 1. The three-term recursion relation (56) can be
written as
R2 λk (R) p
[Dnp (c, ±b) + Ez (p, ±b) − ]Unk
8 4
p+1 p−1
+[Cnp+1 (c, ±b)Unk + Cnp (c, ±b)Unk ] = 0, (66)

where

Cnp (c, ±b) = [p(p ± b)(n − p + 1)(n + c − p + 1)]s ,

Dnp (c, ±b) = (p + 1)(n − p)
+(p ± b)(n + c − p + 1)
1
+ (c ∓ b + s)(c ∓ b + 3s).
4
It follows from Eq. (66) that
λk (R) 1
lim = Ez (k, ±b),
R→∞ R2 2
p
lim Unk (R; c, ±b) = δkp .
R→∞

For R ≫ 1, the interbasis expansion coefficients and the spheroidal separation constant
are developed in negative powers of ΩR2 :
∞
p (j)
Ukp (ΩR2 )−j ,
X
Unk (R; c, ±b) = δkp + (67)
j=1

17
 ∞
λk (R) 1 (j)
λk (ΩR2 )−j .
X
= Ez (k, ±b) + (68)
ΩR2 2Ω j=1

Substituting Eqs. (67) and (68) into Eq. (66), we get

1 (j) (j−1) (j−1)
(p − k)Ukp + Cnp+1 (c, ±b)Uk,p+1 + Dnp (c, ±b)Ukp
4
(j−1) 1 j−1
X (j−t) (t)
+Cnp (c, ±b)Uk,p−1 − λ Ukp = 0. (69)
4 t=1 k
(0) (j)
Using the conditions Ukp = δkp and Upp = δj0 , one easily obtain
1 (j) (j−1) (j−1) (j−1)
λk = Cnp+1 (c, ±b)Uk,p+1 + Dnp (c, ±b)Ukp + Cnp (c, ±b)Uk,p−1 . (70)
4
Equations (69) and (70) completely solve the problem of determining the expansion coef-
(j) (j)
ficients λk and Ukp . For instance, we have the approximate formulae
λk (R) 1 4
= Ez (k, ±b) + D k (c, ±b)
ΩR 2 2Ω ΩR2 n
16 h k 2 k+1 2
i
+ C (c, ±b) − C (c, ±b) ,
(ΩR2 )2 n n

Ψnkm (ξ, η, ϕ; R, c, ±b) = Ψnkm (ρ, ϕ, z; c, ±b)

"
4
+ C k (c, ±b)Ψn,k−1,m (ρ, ϕ, z; c, ±b)
ΩR2 n
#
−Cnk+1 (c, ±b)Ψn,k+1,m(ρ, ϕ, z; c, ±b) .

ACKNOWLEDGMENTS
One of the authors (G.S.P.) is grateful to the Institut de Physique Nucléaire de Lyon
for the kind hospitality extended to him during his stay in Lyon-Villeurbanne. A prelim-
inary version of this work was discussed at the International Workshop “Finite Dimen-
sional Integrable Systems” (Dubna, Russia, July 1994). In this respect, the authors thank
A.N. Sissakian, V.M. Ter-Antonyan, and P. Winternitz for interesting discussions.
Appendix: Bi-Orthogonality of the Radial Wavefunctions
Besides the orthonormality relation (14) in the quantum numbers nr for the function
Rnr q , we also have an orthogonality relation in the quantum numbers q, viz.,
(±)
Z ∞ Ω
Jqq′ = Rn′r q′ Rnr q dr = δq′ q , (71)
0 2q + c ± b + 1
for a given value n′r + q ′ = nr + q of the principal quantum number n. The proof of (71)
is as follows. In the integral in Eq. (71), we replace the two radial wavefunctions by their
expressions (13). Then, with the help of the formula [36]
Z ∞
e−cx xα−1 Lγm (cx)Lλn (cx)dx =
0
(γ + 1)m (λ − α + 1)n Γ(α)
×
m!n!cα
×3 F2 (−m, α, α − λ; γ + 1, α − λ − n; 1),

18
we arrive at
(±) Γ(q ′ + q + c ± b + 1)
Jqq′ = Ω
Γ(2q ′ + c ± b + 2)Γ(q − q ′ + 1)
v
u (n − q ′ )!Γ(n + q ′ + c ± b + 2)
u
×t
(n − q)!Γ(n + q + c ± b + 2)
×2 F1 (−q + q ′ , q + q ′ + c ± b + 1; 2q ′ + c ± b + 2; 1). (72)

By using the Gauss summation formula [24]

Γ(c)Γ(c − a − b)
2 F1 (a, b; c; 1) =
Γ(c − a)Γ(c − b)
we can rewrite (72) as

(±) Ω
Jqq′ =
q+ q′ +c±b+1
v
u (n − q ′ )!Γ(n + q ′
+ c ± b + 2)
u
×t
(n − q)!Γ(n + q + c ± b + 2)
×[Γ(q − q ′ + 1)Γ(q ′ − q + 1)]−1 .

This completes the proof of Eq. (71) since [Γ(q − q ′ + 1)Γ(q ′ − q + 1)]−1 = δqq′ .
Appendix: The Smorodinsky-Winternitz and Morse Systems
The Morse system with the potential

VM = V0 (e−2ax − 2e−ax )

can be connected to the dynamical system with the potential

Ω2 2 P 1
VSW = z + .
2 2 z2
(The latter potential may be considered as a one-dimensional component of the so-called
Smorodinsky-Winternitz [37, 38, 39] potential. The potential VSW was investigated by
Calogero [40].)
The Schrödinger equation for the Morse potential VM , i.e.,
h i
dxx + 2E − 2V0 (e−2ax − 2e−ax ) ψ = 0 (73)

admits a discrete spectrum (with E < 0) and a continuous spectrum. For the discrete
spectrum, by making the change of variable

y = ax, y ∈ R, z = e−sy , z ∈ R+

and the change of function

1
ψ(x) = √ f (z)
z

19
in Eq. (73), we get
8E 1 1
   
2 2
dzz + 4λ (2 − z ) + + f = 0, (74)
a2 4 z2
where
√
2V0
λ= .
a
Equation (74) has the same form as Eq. (26) for z > 0 with
8E 1
Ez = 4λ2 , Ω = 2λ, P =− − .
a2 4
Therefore, we must consider two admissible regions for the energy E: (i) −32E > a2 and
(ii) 0 < −32E < a2 .
In the case (i), by employing the energy formula (30) for Ez , we obtain that E is
determined by the relation
√
−2E
= λ − (p + s) , p = 0, 1, · · · , [λ − s] . (75)
a
In Eq. (75), [x] stands for the integral value of x. As a result, we have
2
1

E = −V0 1 − (p + s) , p = 0, 1, · · · , [λ − s] . (76)
λ
Equation (76) is in agreement with the well-known result according to which the discrete
spectrum of the Morse system has a finite number (here [λ − s] + 1) of energy levels with
the condition λ > s.
In the case (ii), we have
√
−2E
= ±[λ − (p + s)],
a
which has no solution for p ∈ N.
The connection just described between the Morse and Smorodinsky-Winternitz systems
can be used also to deduce the wavefunctions of one system from the wavefunctions of the
other. For instance, from Eq. (29), we immediately get the normalized solution ψ(x) ≡
ψp (z; λ) of (73):
s
p λ−p ap!
ψp (z; λ) = (−1) (2λ)
Γ(2λ − p)
2
×e−λz z 2λ−2p−1 L2λ−2p−1
p (2λz 2 ), (77)

with

z = e−sax , p = 0, 1, · · · , [λ − s].

Our result (77) differs from the one of Nieto and Simmons [41] (by the fact that the factor
p! in (77) is 2p − λ in Ref. [41]).

20
References
[1] C. Quesne, J. Phys. A: Math. Gen. 21, 3093 (1988).

[2] M. V. Carpio-Bernido and C. C. Bernido, Phys. Lett. A 134, 395 (1989); 137, 1
(1989).

[3] I. V. Lutsenko, L. G. Mardoyan, G. S. Pogosyan, and A. N. Sissakian, Communication

P2-89-814, JINR, Dubna (1989).

[4] M. Kibler and P. Winternitz, Phys. Lett. A 147, 338 (1990).

[5] N. W. Evans, Phys. Rev. A 41, 5666 (1990).

[6] M. V. Carpio-Bernido, J. Math. Phys. 32, 1799 (1991).

[7] O. F. Gal’bert, Ya. I. Granovskii, and A. S. Zhedanov, Phys. Lett. A 153, 177 (1991).

[8] M. Kibler, G.-H. Lamot, and P. Winternitz, Int. J. Quantum Chem. 43, 625 (1992).

[9] L. G. Mardoyan, A. N. Sissakian, V. M. Ter-Antonyan, and T. A. Chatrchian, Preprint

P2-92-511, JINR, Dubna (1992).

[10] M. Kibler and C. Campigotto, Phys. Lett. A 181, 1 (1993).

[11] A. S. Zhedanov, J. Phys. A: Math. Phys. 26, 4633 (1993).

[12] C. Grosche, G. S. Pogosyan, and A. N. Sissakian, Fortschr. Phys. 43, 453 (1995).

[13] H. Hartmann, Theor. Chim. Acta 24, 201 (1972).

[14] H. Hartmann, R. Schuck, and J. Radtke, Theor. Chim. Acta 42, 1 (1976).

[15] H. Hartmann and D. Schuch, Int. J. Quantum Chem. 18, 125 (1980).

[16] M. Kibler and P. Winternitz, J. Phys. A: Math. Gen. 20, 4097 (1987).

[17] I. V. Lutsenko, G. S. Pogosyan, A. N. Sissakyan, and V. M. Ter-Antonyan, Teor.

Mat. Fiz. 83, 419 (1990).

[18] M. Kibler, L. G. Mardoyan, and G. S. Pogosyan, Int. J. Quantum Chem. 52, 1301
(1994).

[19] J.-L. Calais, Int. J. Quantum Chem. 2, 715 (1968).

[20] E. Ley-Koo and S. A. Cruz, J. Chem. Phys. 74, 4603 (1981). M. A. Núñez and G.
Izquierdo B., Int. J. Quantum Chem.: Quantum Chem. Symp. 28, 241 (1994). S. A.
Cruz, E. Ley-Koo, J. L. Marı́n, and A. Taylor-Armitage, Int. J. Quantum Chem. 54,
3 (1995).

[21] A. A. Abramov, A. L. Dyshko, N. B. Konyukhova, T. V. Pak, and B. S. Pariiskii, J.

Comp. Math. and Math. Phys. 24, 3 (1984).

21
[22] N. W. Evans, J. Math. Phys. 32, 3369 (1991).

[23] S. Flügge, Practical Quantum Mechanics (Springer-Verlag, Berlin, 1971).

[24] A. Erdélyi, W. Magnus, F. Oberhettinger, and F. Tricomi, Higher Transcendental

Functions (McGraw-Hill, New York, 1953), Vols. I and II.

[25] J. Klauder, Acta Phys. Austriaca, Suppl. 11, 341 (1973).

[26] D. A. Varshalovich, A. N. Moskalev, and V. K. Khersonskii, Quantum Theory of

Angular Momentum (World Scientific, Singapore, 1988).

[27] Yu. N. Demkov, JETP 26, 757 (1954); 36, 89 (1959).

[28] G. S. Pogosyan and V. M. Ter-Antonyan, Communication P2-11962, JINR, Dubna

(1978).

[29] G. S. Pogosyan, Ya. A. Smorodinsky, and V. M. Ter-Antonyan, Communication P2-

82-118, JINR, Dubna (1982).

[30] V. A. Knyr, P. P. Pipirayte, and Yu. F. Smirnov, Yad. Fiz. 22, 1063 (1975).

[31] I. V. Komarov, L. I. Ponomarev, and S. Yu. Slovyanov, Spheroidal and Coulomb

Spheroidal Functions (Nauka, Moscow, 1976).

[32] C. A. Coulson and A. Joseph, Proc. Phys. Soc. 90, 887 (1967).

[33] M. Kibler and G. Grenet, J. Math. Phys. 21, 422 (1980).

[34] L. G. Mardoyan, G. S. Pogosyan, A. N. Sissakian, and V. M. Ter-Antonyan, Com-

munications P2-85-139, P2-85-140, and P2-85-141, JINR, Dubna (1985).

[35] L. D. Landau and E. M. Lifshitz, Quantum Mechanics (Pergamon Press, Oxford,

1977).

[36] A. P. Prudnikov, Yu. A. Brichkov, and O. I. Marichev, Integrals and Series, Special
Functions (Nauka, Moscow, 1983).

[37] J. Friš, V. Mandrosov, Ya. A. Smorodinsky, M. Uhlı́ř, and P. Winternitz, Phys. Lett.
16, 354 (1965).

[38] P. Winternitz, Ya. A. Smorodinskiı̆, M. Uhlir, and J. Fris, Yad. Fiz. 4, 625 (1966)
[English translation: Sov. J. Nucl. Phys. 4, 444 (1967)].

[39] A. A. Makarov, J. A. Smorodinsky, Kh. Valiev, and P. Winternitz, Nuovo Cimento

A 52, 1061 (1967).

[40] F. Calogero, J. Math. Phys. 10, 2191 (1969).

[41] M. M. Nieto and L. M. Simmons, Jr., Phys. Rev. A 19, 438 (1979).

22