You are on page 1of 14

Theoretical and Mathematical Physics, 188(1): 980–993 (2016)

SCHRÖDINGER POTENTIALS SOLVABLE IN TERMS OF THE


CONFLUENT HEUN FUNCTIONS
A. M. Ishkhanyan∗

We show that if the potential is proportional to an energy-independent continuous parameter, then there
exist 15 choices for the coordinate transformation that provide energy-independent potentials whose shape
is independent of that parameter and for which the one-dimensional stationary Schrödinger equation is
solvable in terms of the confluent Heun functions. All these potentials are also energy-independent and
are determined by seven parameters. Because the confluent Heun equation is symmetric under transpo-
sition of its regular singularities, only nine of these potentials are independent. Five of the independent
potentials are different generalizations of either a hypergeometric or a confluent hypergeometric classical
potential, one potential as special cases includes potentials of two hypergeometric types (the Morse conflu-
ent hypergeometric and the Eckart hypergeometric potentials), and the remaining three potentials include
five-parameter conditionally integrable confluent hypergeometric potentials. Not one of the confluent Heun
potentials, generally speaking, can be transformed into any other by a parameter choice.

Keywords: stationary Schrödinger equation, integrable potential, confluent Heun equation

DOI: 10.1134/S0040577916070023

1. Introduction
We consider the possibilities and strategies for reducing the one-dimensional stationary Schrödinger
equation to the confluent Heun equation [1], [2]. The latter is a linear second-order ordinary differential
equation that can be conventionally viewed as a natural generalization of the hypergeometric and conflu-
ent hypergeometric equations. Among the five Heun equations, precisely the confluent Heun equation is
particularly interesting because the structure of its singularities directly reflects the features of both hy-
pergeometric equations. This equation has two regular singularities at finite points in the complex-z plane
(conventionally, z = 0 and z = 1), which is characteristic for the hypergeometric equation, and one irregular
singularity of s-rank two [2] at z = ∞, as in the case of the confluent hypergeometric equation. We can
expect that because of such a singularity structure, the confluent Heun equation can help to generalize
all the hypergeometric potentials and also several nonhypergeometric potentials, for example, the Mathieu


Institute for Physical Research, National Academy of Sciences of Armenia, Ashtarak, Armenia; Armenian
State Pedagogical University, Yerevan, Armenia; Institute of Physics and Technology, National Research Tomsk
Polytechnic University, Tomsk, Russia, e-mail: aishkhanyan@gmail.com.
This research was performed within the scope of the International Associated Laboratory (CNRS-France & SCS-
Armenia) IRMAS and was supported by the Armenian State Committee of Science (SCS Grant Nos. 13RB-052 and
15T-1C323) and the project “Leading Research Universities of Russia” (Grant No. FTI 120 2014 Tomsk Polytechnic
University).

Prepared from an English manuscript submitted by the author; for the Russian version, see Teoreticheskaya i
Matematicheskaya Fizika, Vol. 188, No. 1, pp. 20–35, July, 2016. Original article submitted August 12, 2015; revised
October 23, 2015.

980 0040-5779/16/1881-0980 
c 2016 Pleiades Publishing, Ltd.
(in the algebraic form, the Mathieu equation is a particular specification of the confluent Heun equation),
spheroidal, and Coulomb spheroidal potentials [3].
Because the two hypergeometric and several other known equations (such as the mentioned Mathieu,
spheroidal, and Coulomb spheroidal equations) are obtained by simple choice of separate parameters, the
confluent Heun equation provides an understandable route to trace a particular extension when passing
from a known potential to a more general potential in the collection of confluent Heun potentials.
If the Schrödinger equation is reduced to the confluent Heun equation by an energy-independent coor-
dinate transformation, then the general form of the potential contains 11 parameters. But we show that if
the potential is proportional to an energy-independent parameter and if the potential shape is independent
of both this parameter and the energy, then the general family necessarily collapses to a restricted set of
particular potentials with a smaller number of independent continuous parameters. We first prove that
the coordinate transformation is necessarily independent of the abovementioned parameter. An immediate
consequence of this observation is that the Schwartzian derivative of the coordinate transformation cannot
have poles other than the singularities of the equation to which the Schrödinger equation reduces. This
leads to a discrete set of a few permissible forms for the coordinate transformation.
In the case of the confluent Heun equation, generally speaking, there exist 15 possible variants of
the coordinate transformation. Each of these variants leads to a separate seven-parameter potential. But
because the confluent Heun equation is symmetric under the transposition z ↔ 1 − z, the number of
independent potentials reduces to nine. Unlike the hypergeometric case, not one of the confluent Heun
potentials, generally speaking, can be transformed into any other by a parameter choice.
In the general case, the potentials are given parametrically as pairs of functions V (z) and x(z). But
in six cases, the inverse function z(x) and hence the potential V (z(x)) are written in terms of elementary
functions. Five of these potentials are different generalizations of either a hypergeometric or a confluent
hypergeometric classical potential [4]–[13], and the sixth potential as special cases contains classical hy-
pergeometric potentials of both types (the Morse confluent hypergeometric potential [5] and the Eckart
hypergeometric potential [6]). The remaining three potentials, for which the coordinate transformation
is not written in terms of known elementary functions, as special cases contain confluent hypergeometric
potentials that are only conditionally integrable. We note that one of these potentials can be represented
in terms of an implicit elementary function W (x), the Lambert function [14], [15], which is a solution of
the equation W eW = x.

2. Specialization of potentials

The Schrödinger equation for a particle with the mass m and energy E in the field of a potential V (x)

d2 ψ 2m  
+ 2 E − V (x) ψ = 0 (1)
dx 2 
via the transformation of the dependent and independent variables ψ = φ(z)u(z) and z = z(x) reduces to
the equation    
φz ρz φzz ρz φz 2m E − V (z)
uzz + 2 + uz + + + 2 u = 0, (2)
φ ρ φ ρ φ  ρ2
where ρ = z  (x). Here and hereafter, a lowercase Latin subscript denotes differentiation with respect to the
corresponding variable. This equation passes into the auxiliary equation

uzz + f (z)uz + g(z)u = 0 (3)

if
φz ρz φzz ρz φz 2m E − V (z)
2 + = f (z), + + 2 = g(z). (4)
φ ρ φ ρ φ  ρ2

981
We find φ(z) from the first equation,
  
1
φ(z) = ρ−1/2 exp f (z) dz , (5)
2

and rewrite the second equation in (4) as


   
fz f2 1 ρz 1 ρz 2 2m E − V (z)
g− − =− − + 2 . (6)
2 4 2 ρ z 4 ρ  ρ2

The expression in the left-hand side of this equation is the invariant of Eq. (3),

fz f2
I(z) = g − − ,
2 4

and the first two terms in the right-hand side determine the Schwartzian derivative {z, x} of z(x):
       
1 ρz 1 ρz 2 {z, x} z  (x) 1 z  (x) 2
− − =− , {z, x} ≡  − . (7)
2 ρ z 4 ρ 2ρ2 z (x) 2 z  (x)

Here and hereafter, the prime denotes the derivative with respect to x. Taking these relations into account,
we rewrite (6) as
1 2m  
(z  (x))2 I(z) + {z, x} = 2 E − V (x) . (8)
2 
This equation has been used by many authors starting from the work of Bose [16]. But for our derivations
to be more explicit, we use the equivalent Eq. (6).
Studying the potentials constructed using the energy-independent transformation z(x) under the con-
dition that the potential itself is assumed to be energy-independent, we can immediately see that the energy
and potential terms in (6) should be appropriately chosen such that each independently of the other corre-
sponds to one of the two terms in the left-hand side forming the invariant I(z). Applied to the confluent
Heun equation  
γ δ αz − q
uzz + + + ε uz + u = 0, (9)
z z−1 z(z − 1)
this invariant is a fourth-degree polynomial in z divided by z 2 (z − 1)2 . Hence, taking E/ρ2 ∼ I(z), we
directly obtain the transformation

±z(z − 1)
ρ(z) ≡ z  (x) = √ . (10)
r0 + r1 z + r2 z 2 + r3 z 3 + r4 z 4

Then taking V (z)/ρ2 ∼ I(z), we define the family of potentials

2mV (x) v0 + v1 z + v2 z 2 + v3 z 3 + v4 z 4 {z, x}


= − . (11)
 2 2 3
r0 + r1 z + r2 z + r3 z + r4 z 4 2

Because rk and vk , k = 0, 1, 2, 3, 4, and the integration constant in (10) are arbitrary, we conclude that this
family of potentials has 11 parameters.
In the case of the hypergeometric equations, the invariant is a second-degree polynomial r(z) divided
by z (z − 1)2 for hypergeometric equations or by z 2 for confluent hypergeometric equations. In both cases,
2

we obtain seven-parameter families of potentials (Natanzon potentials [17]). But it is known that the
classical hypergeometric potentials [4]–[13] are obtained under rather strong restrictions imposed on the

982
parameters of the polynomial r(z). For the hypergeometric potentials, it turns out that only a discrete set
of polynomials is admitted [16]

r(z) ∼ 1, r(z) ∼ z, r(z) ∼ 1 − z, r(z) ∼ z 2 ,


(12)
r(z) ∼ z(1 − z), r(z) ∼ (1 − z)2 .

For the confluent hypergeometric potentials, the admissible discrete set is [16]

r(z) ∼ 1, r(z) ∼ z, r(z) ∼ z 2 . (13)

Consequently, all classical potentials have five parameters. Three of these five parameters determine the
coordinate origin, the spatial scale, and the origin of the energy; only two parameters hence remain for
characterizing the physical potentials. In what follows, we trace this specification of general potential
family (10), (11) under rather general assumptions.
We consider the special case where the potential term in the Schrödinger equation is proportional to a
parameter μ that is independent of energy and assume that the potential shape is independent of both the
energy and that parameter:
2mV (x)
= μS(x), (14)
2
where μ is independent of E and S is independent of both μ and E. We can list several physical situations
where this assumption holds. Nevertheless, we must be careful here as follows. It might seem that we
could always set μ = m for mass-independent potentials, for example, in the case of purely electromagnetic
interactions. But this choice is singular because the mass is also involved in the energy term, and the limit
μ → 0, which we use below, cannot be directly applied in this case.
We note that in the hypergeometric cases, the roots of the polynomial r(z) always coincide with the
finite singularities of the equation to which the Schrödinger equation reduces, namely, z = 0 and z = 1 in
the case of the hypergeometric equation and z = 0 in the case of the confluent hypergeometric equation.
We prove that this is a general property following from assumption (14).
As a first step, we show that the transformation defined by Eq. (10) must be μ-independent. We can
prove this assertion by considering the behavior of the μ-independent functions S and s = log S  in the
vicinity of a point x = x∗ , where the variable z approaches a finite singularity of the equation. As such, we
can take any singularity that is an endpoint for the physical problem under consideration. For simplicity
of the derivations, we take z = 0 as such a singularity and assume that x∗ = −∞.
We consider the behavior of the potential in the vicinity of z = 0. If r0 = 0, then Eqs. (10) and (11)
yield the expansions

dx 1 r0 2r0 + r1
=  = + a0 + a1 z + a2 z 2 + . . . , a0 = √ , (15)
dz z (x) z 2 r0
2m
S(x) = (V0 + V1 z + V2 z 2 + . . . ) = S0 + S1 z + S2 z 2 + . . . . (16)
μ2

The first term in the right-hand side of (15) determines the asymptotic leading term
 
x − x0 √
z|x→−∞ ∼ exp √ = c0 ex/ r0 , (17)
r0

where x0 is the integration constant, which in general can depend on μ. Applying the identity

S  (x) V  (x) Vzz (z)


s (x) = 
= 
= ρ(z) + ρz (z) (18)
S (x) V (x) Vz (z)

983

and taking the conditions ρ(0) = 0 and ρz (0) = ±1/ r0 into account, we see that r0 is independent of μ.
Asymptotic equation (17) provides the useful formula
√ √
lim e−nx/ r0 n
z (x) = cn0 , c0 = e−x0 / r0
,
x→−∞

which holds for any n ≥ 1.



Now, successively subtracting the partial sums Pn (x) = nk=0 Sk z k (in which P−1 ≡ 0) from S(x) and
calculating the limits
√  
lim e−nx/ r0
S(x) − Pn−1 (x) = Sn cn0 , n = 0, 1, 2, . . . ,
x→−∞

we conclude that because S(x) is a μ-independent function, all Sn cn0 are independent of μ. Similarly,
considering the expansion
s(x) = s0 + s1 z + s2 z 2 + · · · (19)

of the function s(x) = log S  (x), successively subtracting the corresponding partial sums, and calculating
the limits as in the S(x) case, we see that all sn cn0 are also independent of μ.
Using identity (18), we can express the coefficients sn in terms of the coefficients Sn of expansion (16),
writing them in the form

n+1
sn = Ak (S0 , . . . , Sn+2 , r0 , . . . , r4 )ρ(k) (0), (20)
k=0

where the factors Ak involve the coefficients Si up to the order n + 2. We present the first three coefficients
in the case where the minus sign is chosen in Eq. (10):

2S2 ρ(0) 1
s0 = + ρz (0) = √ , (21)
S1 r0
 
6S3 4S 2 2S2 2S2 2r0 + r1
s1 = − 22 ρ(0) + ρz (0) + ρzz (0) = √ − 3/2
, (22)
S1 S1 S1 r0 S1 r0
   
12S4 18S2 S3 8S23 6S3 4S22 S2 1
s2 = − 2 + 3 ρ(0) + − 2 ρz (0) + ρzz (0) + ρzzz (0) =
S1 S1 S1 S1 S1 S1 2
3r2
= ···− 3/2
. (23)
2r0

Although slightly cumbersome, these equations are informative. For instance, precisely the equation for s0
helps establish that r0 is independent of μ. Because sn cn0 are independent of μ, Eqs. (20) multiplied by cn0
yield an infinite set of equations d(cn0 sn )/dμ = 0 for the six parameters rk , k = 0, 1, 2, 3, 4, and c0 = c0 (x0 ),
on which the coordinate transformation depends. It is easy to verify that the first six of these equations
already allow proving that all these parameters are μ-independent.
This conclusion can be obtained more easily and faster if we use information from the other singular
point z = 1 in parallel. For instance, considering the limit z → 1 in Eq. (18), we can immediately conclude

that the sum ri = r(1) is independent of μ. But we want to emphasize here that it suffices to consider
only one singularity. Therefore, if r0 = 0, then the coordinate transformation is independent of μ. Applying
this approach, we can easily prove that this property is also preserved in the alternative case r0 = 0.
Having established that the coordinate transformation z(x) and hence the function ρ(z) are independent
of μ and having passed to the limits E, μ → 0 in Eq. (6), we obtain
     
1 ρz 1 ρz 2 fz f2
+ =− g− − . (24)
2 ρ z 4 ρ 2 4 E,μ→0

984
Clearly, the logarithmic derivative ρz /ρ cannot have poles other than the singularities of the invariant
I(z) = g − (fz /2) − (f 2 /4). Consequently, the roots of the polynomial r(z) must coincide with the finite
singularities of Eq. (3), i.e., z = 0 and z = 1, if the confluent Heun equation is considered. As follows from
Eq. (10), the function ρ(z) is then written as a product z m1 (z − 1)m2 with integer or half-integer exponents
m1 and m2 .
We note that in the case of hypergeometric equations, r(z) is a second-degree polynomial. The admis-
sible polynomials for the hypergeometric equation and the confluent hypergeometric equation are given by
the respective formulas (12) and (13). Only two parameters are preserved under the coordinate transfor-
√ √
mation: the coordinate origin x0 and the spatial scale σ (we note that σ = r0 if r0 = 0, σ = r1 if r0 = 0
and r1 = 0, and so on). Therefore, the restrictions described above leads to five-parameter potentials in
all cases, six potentials in the case of the hypergeometric equation and three potentials in the case of the
confluent hypergeometric equation. For the hypergeometric equation, by virtue of the symmetry under the
transposition z ↔ 1 − z, the number of independent potentials decreases to four.
In what follows, we show that there are only two hypergeometric potentials that cannot be transformed
into each other by choosing the involved parameters. As such independent potentials, we can take the
Eckart [6] and Pöschl–Teller [11] potentials, which are obtained, for instance, by respectively choosing
r = r0 and r = r1 z. In the case of the confluent hypergeometric equation, all three admissible potentials
are independent. They are the Morse potential [5], the sum of the harmonic oscillator potential and the
inverse square potential [4], and the sum of the Coulomb potential and the inverse square potential (the
Kratzer potential [13]).

3. Confluent Heun potentials


According to what was said above, the admissible forms of the coordinate transformation independent
of E and μ reducing the Schrödinger equation to the confluent Heun equation are given by the formula

z m1 (z − 1)m2
z  (x) = ρ = , (25)
σ

where m1 and m2 are integers or half-integers and σ is an arbitrary scaling coefficient. For the energy term
E/ρ2 to agree with the left-hand side in Eq. (6), we must take the term z 2 (z − 1)2 /ρ2 as a polynomial in
z of at most the fourth degree. We can do this if m1 ≤ 1, m2 ≤ 1, and m1 + m2 ≥ 0. As a result, we
obtain 15 pairs (m1 , m2 ) (see Fig. 1). Taking these pairs and matching the potential term V (z)/ρ2 with
the invariant in Eq. (6), we obtain 15 seven-parameter potentials written as

V (z) = z 2m1 −2 (z − 1)2m2 −2 (v0 + v1 z + v2 z 2 + v3 z 3 + v4 z 4 ). (26)

Several of the presented potentials contain classical hypergeometric potentials as special cases. The
values (m1 , m2 ) corresponding to these cases are marked by squares in Fig. 1. For the values marked
by triangles, the potentials contain classical potentials integrable in confluent hypergeometric functions as
special cases. There are also cases where the confluent Heun potentials include classical hypergeometric
potentials of both types as special cases, corresponding to points marked by rhombuses. Finally, there are
potentials that in special cases are conditionally integrable in confluent hypergeometric functions (stars in
Fig.1).
We note that because of the symmetry under the transposition m1 ↔ m2 , the number of independent
potentials is nine. These nine cases are marked by black shapes in Fig. 1.
The nine independent confluent Heun potentials can be written in the forms presented in Table 1.
For six of these potentials, there are special cases of hypergeometric or confluent hypergeometric classical

985
Fig. 1. The 15 possible pairs (m1 , m2 ).

potentials or both types at once. These potentials were presented by Lamieux and Bose (see Table II in [18]).
In the table in [18], ten potentials are listed, but they are special cases of the mentioned six independent
confluent Heun potentials: the first three rows contain the potentials corresponding to (m1 , m2 ) = (0, 0),
(m1 , m2 ) = (1/2, 0), and (m1 , m2 ) = (1, 0); the fourth and fifth rows relate to the case (m1 , m2 ) = (1/2, 1/2);
the sixth and seventh rows give the potential with (m1 , m2 ) = (1/2, 1); and the last three rows describe the
potential with (m1 , m2 ) = (1, 1). Some nonhypergeometric cases from among these six potentials have been
considered by many authors and in different problems, for example, in studying the two-center Coulomb
problem [2], [3] or the Teukolsky equation [19].
The three remaining independent potentials, marked by black stars in Fig. 1, have five-parameter
potentials conditionally integrable in confluent hypergeometric functions [20], [21] as special cases. One of
these potentials with (m1 , m2 ) = (1, −1) is written in terms of the Lambert function, which is an implicit
elementary function [15]. Notably, this potential in turn has a particular four-parameter special case that
is exactly solvable in terms of confluent hypergeometric functions [21].
For all 15 presented potentials, the solution of stationary Schrödinger equation (1) is written explicitly
in terms of the confluent Heun function:

ψ = eα0 z z α1 (z − 1)α2 Hc (γ, δ, ε; α, q; z). (27)

Substituting this expression in Eq. (4) and collecting the coefficients of like powers of z, we obtain eight
equations linear in the five parameters γ, δ, ε, α, and q of the confluent Heun function Hc and quadratic
in the three parameters αk , k = 0, 1, 2, of the prefactors. Solving these equations, we obtain

γ = 2α1 + m1 , δ = 2α2 + m2 , ε = 2α0 , (28)


  2m
α = α0 2(−α0 + α1 + α2 ) + m1 + m2 + 2 (Er3 − v3 ), (29)

2m  
q = (2α1 + m1 )(α0 − α2 ) − m2 α1 + 2 E(2r0 + r1 ) − (2v0 + v1 ) (30)


986
Table 1
Coordinate transformation Particular
No. (m1 , m2 ) Potential V (z)
z(x) or x(z) potential
V1 V2
V0 + z + z2 + x−x0
1 (0, 0) V3 V4 z= σ 1F1 (z) [4]
+ z−1 + (z−1)2

1  V0 + V1 z + V2 z 2 +
2 1
2, 2 z = cosh2 x−x 0
2F1 (z) [11]
+ Vz3 + z−1
V4 2σ

1  V0 + V1 z + Vz2 + (x−x0 )2
3 2, 0 V3 V4 z= 4σ2 1F1 (z) [4]
+ z−1 + (z−1) 2

1  V0 + V1 V2
z + z−1 + x = x0 +
4 2, −2
1  √  —
+ V3 V4
(z−1)2 + (z−1)3 +σ z(z − 1) − sinh−1 z − 1

V0 + V1 z + V2 z 2 + 1
5 (1, 1) z= 2F1 (z) [6]
+ V3 z 3 + V4 z 4 e(x−x0 )/σ +1

 1 V0 + V1 z + V2 z 2 +
6 1, 2 z = sec2 x−x 0
2F1 (z) [11]
+ V3 z 3 + z−1
V4 2σ

V0 + V1 z + V2 z 2 + 1F1 (z) [5],


7 (1, 0) V3 V4 z = e(x−x0 )/σ
+ z−1 + (z−1) 2 F
2 1 (z) [6]

  V2
V0 + V1 z + z−1 + x = x0 +
8 1, − 21  √ √  —
V3 V4
+ (z−1)2 + (z−1) 3 + σ 2 z − 1 − 2 tan−1 z − 1
V1 V2
V0 + z−1 + (z−1)2 + x = x0 + σ(z − log z)
9 (1, −1) —
+ V3 V4
(z−1)3 + (z−1)4 z = −W (e−(x−x0 )/σ )
Nine independent seven-parameter confluent Heun potentials.

and
2m
α20 + (Er4 − v4 ) = 0, (31)
2
2m
α21 + α1 (m1 − 1) + (Er0 − v0 ) = 0, (32)
2

2m 
4
α22 + α2 (m2 − 1) + (Ern − vn ) = 0, (33)
2 n=0

where the auxiliary parameters vk , k = 0, 1, 2, 3, 4, are calculated for each case based on definition (26) and
the parameters rk , k = 0, 1, 2, 3, 4, are determined by the equation
z 2 (z − 1)2
= σ 2 z 2−2m1 (z − 1)2−2m2 = r0 + r1 z + r2 z 2 + r3 z 3 + r4 z 4 . (34)
ρ2
Hence, for any row in Table 1, we have a potential generally given parametrically as a pair of functions
V (z) and x(z). For z ∈ [0, 1], the coordinate transformation x(z) defined by Eq. (25) can generally be
written in terms of the incomplete beta function:

x = x0 + (−1)−m2 σBz (1 − m1 , 1 − m2 ). (35)

987
Fig. 2. Potential (37) for (V1 , V2 , V3 , V4 ) = (10, −10, 7, 1), σ = 100 (and x0 = −σ): the inset shows
coordinate transformation (38).

But because the parameters m1 and m2 are integers or half-integers, the beta functions always reduce to
elementary functions. These functions or the corresponding inverse transformations z(x) are presented in
the fourth column in Table 1 (we note that because the only requirement is that the resulting function
V (x) be real, the integration constant x0 and the scaling factor σ can be complex). We note that if we
have special cases of classical hypergeometric potentials, then the inverse transformation z(x) is written in
terms of elementary functions.
The case (m1 , m2 ) = (1, 0), for which z = e(x−x0 )/σ , is special because the confluent Heun poten-
tial for it is represented as the sum of the Morse confluent hypergeometric potential [5] and the Eckart
hypergeometric [6] potential:

V3 V4
V (x) = V0 + V1 e(x−x0 )/σ + V2 e2(x−x0 )/σ + + . (36)
e(x−x0 )/σ − 1 (e(x−x0 )/σ − 1)2

The Rosen–Morse [7], Woods–Saxon [8], Manning–Rosen [9], and Hulthön [10] potentials are special cases
of the Eckart potential (to obtain the Eckart, Rosen–Morse, and Woods–Saxon potentials, we must apply
the complex substitution x0 = x1 + iπσ).
There are no classical hypergeometric potentials for the three values (m1 , m2 ) in the lower-right quad-
rant in Fig. 1. We mention the potential with (m1 , m2 ) = (1, −1),

V1 V2 V3 V4
V (z) = V0 + + + + , (37)
z − 1 (z − 1)2 (z − 1)3 (z − 1)4

for which the transformation z(x) is written in terms of the Lambert W -function [15],

z(x) = −W (−e−(x−x0)/σ ). (38)

For x0 = −σ, σ > 0, this transformation yields a potential defined on the positive half-axis x > 0 with
only one singularity at the origin and tending to V∞ = V0 − V1 + V2 − V3 + V4 at infinity (see Fig. 2). The
behavior of the potential at infinity is given by

V |x→+∞ = V∞ − Ae−(x+σ)/σ + . . . , A = V1 − 2V2 + 3V3 − 4V4 , (39)

988
and in the vicinity of the origin, we have the expansion
V4 4V4 − 3V3 V2 − V3 + V4 d−1 √
V |x→0 = 2
+ √ + + √ + d0 + d1 x + . . . . (40)
4x 6 2x 3/2 2x x
Several special cases of potential (37) admit a solution in terms of simpler functions. For instance,
for the potential with V2 = V3 = V4 = 0 [21] and for the conditionally integrable potential with V3 =
−2 /2mσ 2 , V4 = 3V3 /4, the solution of the Schrödinger equation is written explicitly in terms of confluent
hypergeometric functions (see below).

4. Classical hypergeometric and confluent hypergeometric


potentials
Classical hypergeometric potentials. The confluent Heun equation reduces to the hypergeometric
equation if ε = α = 0. As follows from relations (28), (29), and (31), we have r3 = r4 = 0 and v3 = v4 = 0
in this case, and the fourth-degree polynomials r(z) and v(z) in the right-hand sides of Eqs. (34) and (26)
therefore become second-degree polynomials. It follows from Eq. (34) that m1 and m2 satisfy the inequality
m1 + m2 ≥ 1. Therefore, hypergeometric potentials exist for six points (m1 , m2 ) closing the upper-right
corner in Fig. 1. These are the classical five-parameter hypergeometric cases described by Bose [16]. The
potentials and corresponding transformations (with examples of real and complex parameters x0 and σ)
are presented in Table 2. In fact, only two of these potentials are independent; the others are obtained
from them by specifying appropriate parameter values. As such independent potentials, we can choose the
Eckart [6] and Pöschl–Teller [11] potentials.
Classical confluent hypergeometric potentials. The confluent Heun equation reduces to the
confluent hypergeometric equation if δ = 0 and q = α or if γ = 0 and q = 0. Because the equation is
symmetric under the transposition z ↔ z − 1, these two possibilities are equivalent, but the second is easier
to consider.
Indeed, let γ = q = 0. For m1 = 0, we easily obtain classical confluent hypergeometric potentials. First,
it is immediately seen from relations (28), (30), and (32) that r0 = r1 = 0 and v0 = v1 = 0, and the number
of the input parameters of the problem is decreased by two in this case. The resulting three independent
potentials, which can be also obtained by setting m2 = 0 and m1 = 0, 1/2, 1 (see Table 2), compose the set of
three classical five-parameter confluent hypergeometric potentials: the sum of the Coulomb and centrifugal
potentials [13], the sum of the harmonic oscillator and inverse square potentials [4], and also the Morse
potential [5].
Conditionally integrable confluent hypergeometric potentials. Considering the cases δ = 0
and q = α with m2 = 0 and γ = q = 0 with m1 = 0, we obtain six conditionally integrable potentials (these
cases correspond to values (m1 , m2 ) marked with stars in Fig. 1). Moreover, some of the parameters of
the potential are not independent or have a definite fixed value. Only three of the conditionally integrable
potentials are independent. They are presented in Table 3.
As an example of explicit solutions, we consider the conditionally integrable potential related to a
Heun-type potential with a Lambert W -function for which (m1 , m2 ) = (1, −1):
 
V1 V2 2 1 3/4
V = V0 + + − + ,
z − 1 (z − 1)2 2mσ 2 (z − 1)3 (z − 1)4 (41)
z = −W (−e−(x−x0)/σ ).
It can be verified directly that the general solution of the Schrödinger equation for this potential is written
explicitly using confluent hypergeometric functions as
 
ψ = eα0 z z α1 (z − 1)1/2 c1 · 1F1 (a; 1 + 2α1 ; −2α0 z) + c2 · U (a; 1 + 2α1 ; −2α0 z) , (42)

989
Table 2
Potential
No. (m1 , m2 ) z(x) Reference
V (z)

Classical hypergeometric potentials


z = 1 − e(x−x1 )/σ , Hulthön [10] (V2 = 0),
V1 V2
(0, 1) V0 + z + z2 x0 = x1 + iπσ; Manning–Rosen [9];
Woods–Saxon [8] (V2 = 0),
z = 1 + e(x−x0 )/σ
Rosen–Morse [7], Eckart [6]

1  Hyperbolic
V2 z = tanh2 x−x
2σ ;
0

2, 1 V0 + V1 z + z Pöschl–Teller [11];
Trigonometric
z = −tan2 x−x
2σ0 ,
0

Pöschl–Teller [11],
σ = iσ0
Scarf [12] (V1 = 0, V1 = V2 )
Hyperbolic
1 1
 V1 V2 z = cosh2 x−x
2σ ;
0

2 2, 2 V0 + z + z−1 Pöschl–Teller [11];


Trigonometric
z = cos2 x−x
2σ0 ,
0

Pöschl–Teller [11],
σ = iσ0
Scarf [12] (V1 = 0, V1 = V2 )
1 Eckart [6], Rosen–Morse [7],
2 z= 1+e(x−x0 )/σ
;
5 (1, 1) V0 + V1 z + V2 z Woods–Saxon [8] (V2 = 0);
1
z = 1−e(x−x 1 )/σ
, Manning–Rosen [9],
x0 = x1 + iπσ Hulthön [10] (V2 = 0)
Trigonometric
 1 V2 z = cos−2 x−x0
2σ ; Pöschl–Teller [11],
6 1, 2 V0 + V1 z + z−1
Scarf [12] (V1 = 0, V1 = V2 );
z = cosh−2 x−x
2σ0 ,
0
Hyperbolic
σ = iσ0 Pöschl-Teller [11]
Manning–Rosen [9],
V1 V2 z = e(x−x0 )/σ ;
7 (1, 0) V0 + z−1 + (z−1)2 Hulthön [10] (V2 = 0);
z = −e(x−x1 )/σ , Woods–Saxon [8] (V2 = 0),
x0 = x1 + iπσ Rosen–Morse [7], Eckart [6]
Classical confluent hypergeometric potentials
Coulomb [4] (V2 = 0),
x−x0
V1 V2 z= σ Centrifugal [4] (V1 = 0),
1 (0, 0) V0 + z + z2
Kratzer [13]
Harmonic oscillator
(x−x0 )2
1  V2 z= 4σ2 [4] (V2 = 0),
3 2, 0 V0 + V1 z + z
Centrifugal [4] (V1 = 0)
7 (1, 0) V0 + V1 z + V2 z 2 z = e(x−x0 )/σ Morse [5]
Classical hypergeometric potentials: the number in the first column indicates the number
of the independent Heun potential to which that classical hypergeometric potential relates.

990
Table 3
No. (m1 , m2 ) PotentialV (z) Coordinate transformation x(z) Restrictions
−32
1  V1 V2 x = x0 + V3 = 32mσ2 ,
V0 + z + z−1 + √ 
2, −2
1
4
+ V3 V4
(z−1)2 + (z−1)3
+ σ z(z − 1) − sinh−1 z − 1 V4 = 5V3
3
−32
  V2 x = x0 + √ V3 = 16mσ2 ,
8 1, − 21 V0 + V1 z + z−1 + √ 
V3 V4
+ (z−1)2 + (z−1) 3 + σ 2 z − 1 − 2 tan−1 z − 1 V4 = 5V3
6
−2
V0 + V1 V2
x = x0 + σ(z − log z), V3 = 2mσ2 ,
9 (1, −1) z−1 + (z−1)2 +
+ V3 V4
(z−1)3 + (z−1)4
z(x) = −W (−e−(x−x0)/σ ) V4 = 3V3
4

Independent conditionally integrable confluent hypergeometric potentials.

where

2m(V0 − E) 2m(V0 − V1 + V2 − E) 1
α0 = σ , α1 = σ + 2, (43)
2 2 4σ
and  
1 4mσ 4 V1
a= − α0 + α1 + V0 − − E . (44)
2 α0 2 4

The solution for the symmetric analogue of potential (41) with (m1 , m2 ) = (−1, 1) was previously presented
by Williams [20].
Concluding this section, we note that there are many other conditionally integrable special cases for
the confluent Heun potentials presented above, but these solutions have a different structure that is not
related to the ansatz ψ = φ(z)u(z), where u(z) is a Heun function, considered here. For instance, there are
potentials for which the solution is written in terms of combinations of hypergeometric functions [21], [22].

5. Discussion

There have been many studies of the problem of reducing the Schrödinger equation to Heun-class equa-
tions (see, e.g., [1], [2] [18]–[25] and the references therein). A large number of potentials, especially those
reducible to multiple-confluent Heun equations, are listed in [1], [2]. Lamieux and Bose [18] systematically
analyzed this question for all five Heun equations, but this work only embraces the cases where the coor-
dinate transformation can be explicitly written in terms of elementary functions. In a recent paper [25], a
more general consideration of the Heun potentials was presented, but the question of determining the form
of the potentials was not considered.
The breakdown of the general potential into a finite set of separate potentials involving fewer con-
tinuous parameters was previously noted by several authors. For example, it was shown in [26] that a
similar phenomenon inevitably occurs if the potential is proportional to an energy-independent parameter
μ. But the consideration there was restricted to the case of the two hypergeometric equations and, more
importantly, was based on the presupposition that the coordinate transformation is independent of μ.
Our basic result here is the statement that the independence of the coordinate transformation z(x)
from the parameter μ is necessary for any equation with a singularity at a finite point of the complex-z
plane under the condition that the range of the new independent variable z includes a neighborhood of this
singularity.
Considering the example of the confluent Heun equation, we showed that there are 15 choices for
the coordinate transformation, each of which leads to a seven-parameter potential. We note that this

991
situation is similar to the situation encountered in solving the quantum two-state problem, where a similar
picture is seen—the appearance of specific models related to the confluent Heun equation [27]. But the
mechanism is different in our case. The breakdown of the set of analytically integrable field configurations
into a discrete set of particular seven-parameter models is a consequence of the complex description of the
physical interactions, which is characteristic for laser physics.
Because the confluent Heun equation is symmetric under the transposition z ↔ 1 − z, only nine po-
tentials are independent. Three of these independent potentials are different generalizations of the Eckart
or Pöschl–Teller hypergeometric potentials, two other potentials generalize the oscillator potential and the
Coulomb confluent hypergeometric potential, and one potential is the sum of two potentials, the Morse con-
fluent hypergeometric potential and the Eckart hypergeometric potential. The remaining three independent
confluent Heun potentials have only special cases of conditionally integrable confluent hypergeometric po-
tentials. We presented an explicit solution for one of such potentials, which belongs to the class of potentials
of the confluent Heun equation with the Lambert W -function.
There exist other exactly solvable (see, e.g., [21], [22]) or conditionally integrable special cases of the
presented confluent Heun potentials whose solutions are written in terms of simpler special functions. But
these solutions have a different structure. For instance, there are potentials for which each of the two
fundamental solutions of the Schrödinger equation is written in terms of combinations of hypergeometric
functions (see examples in [21], [22], [28]–[34]). Several similar potentials can be obtained by applying
the deformed Heun equation [35]. We intend to present an analysis of such potentials in a subsequent
publication.

REFERENCES
1. A. Ronveaux, ed., Heun’s Differential Equations, Oxford Univ. Press, Oxford (1995).
2. S. Yu. Slavyanov and W. Lay, Special Functions: A Unified Theory Based on Singularities, Oxford Univ. Press,
New York (2000).
3. I. V. Komarov, L. I. Ponomarev, and S. Yu. Slavyanov, Spheroidal and Coulomb Spheroidal Functions [in
Russian], Nauka, Moscow (1976).
4. E. Schrödinger, Ann. Phys., 384, 361–376 (1926).
5. P. M. Morse, Phys. Rev., 34, 57–64 (1929).
6. C. Eckart, Phys. Rev., 35, 1303–1309 (1930).
7. N. Rosen and P. M. Morse, Phys. Rev., 42, 210–217 (1932).
8. R. D. Woods and D. S. Saxon, Phys. Rev., 95, 577–578 (1954).
9. M. F. Manning and N. Rosen, Phys. Rev., 44, 953 (1933).
10. L. Hulthén, Ark. Mat. Astr. Fys. A, 28, No. 5, 1–12 (1942); Ark. Mat. Astr. Fys. B, 29, No. 1, 1–11 (1942).
11. G. Pöschl and E. Teller, Z. Phys. A, 83, 143–151 (1933).
12. F. Scarf, Phys. Rev., 112, 1137–1140 (1958).
13. A. Kratzer, Z. Phys., 3, 289–307 (1920).
14. J. H. Lambert, Acta Helvetica, 3, 128–168 (1758).
15. L. Euler, Acta Acad. Scient. Petropol., 2, 29–51 (1783).
16. A. K. Bose, Phys. Lett., 7, 245–246 (1963).
17. G. A. Natanzon, Theor. Math. Phys., 38, 146–153 (1979).
18. A. Lemieux and A. K. Bose, Ann. Inst. H. Poincaré Sec. A, n.s., 10, 259–270 (1969).
19. E. W. Leaver, J. Math. Phys., 27, 1238–1265 (1986).
20. B. W. Williams, Phys. Lett. A, 334, 117–122 (2005).
21. A. M. Ishkhanyan, Phys. Lett. A, 380, 640–644 (2016); arXiv:1509.00846v2 [quant-ph] (2015).
22. A. M. Ishkhanyan, Eur. Phys. Lett., 112, 10006 (2015).
23. M. F. Manning, Phys. Rev., 48, 161–164 (1935).
24. L. J. El-Jaick and B. D. B. Figueiredo, J. Math. Phys., 49, 083508 (2008).

992
25. D. Batic, R. Williams, and M. Nowakowski, J. Phys. A: Math. Theor., 46, 245204 (2013).
26. A. M. Goncharenko, V. A. Karpenko, and V. N. Mogilevich, Theor. Math. Phys., 88, 715–720 (1991).
27. A. M. Ishkhanyan and A. E. Grigoryan, J. Phys. A: Math. Theor., 47, 465205 (2014).
28. C. Quesne, J. Phys. A: Math. Theor., 41, 392001 (2008).
29. S. Odake and R. Sasaki, Phys. Lett. B, 679, 414–417 (2009); arXiv:0906.0142v2 [math-ph] (2009).
30. A. M. Ishkhanyan, T. A. Shahverdyan, and T. A. Ishkhanyan, Eur. Phys. J. D, 69, 10 (2015); arXiv:1404.3922v2
[quant-ph] (2014).
31. T. A. Shahverdyan, T. A. Ishkhanyan, A. E. Grigoryan, and A. M. Ishkhanyan, J. Contemp. Phys. (Armenian
Acad. Sci.), 50, 211–226 (2015); arXiv:1412.1378v2 [quant-ph] (2014).
32. G. Junker and P. Roy, Phys. Lett. A, 232, 155–161 (1997).
33. G. Lévai and P. Roy, Phys. Lett. A, 264, 117–123 (1999); arXiv:quant-ph/9909004v2 (1999).
34. A. Sinha, G. Lévai, and P. Roy, Phys. Lett. A, 322, 78–83 (2004).
35. A. Ya. Kazakov and S. Yu. Slavyanov, Theor. Math. Phys., 179, 543–549 (2014).

993

You might also like