ENSEÑANZA REVISTA MEXICANA DE FÍSICA 55 (2) 196–215 DICIEMBRE 2009

The Cauchy problem for a forced harmonic oscillator

R.M. Lopez and S.K. Suslov
School of Mathematical and Statistical Sciences and Mathematical,
Computational, and Modeling Sciences Center,
Arizona State University, Tempe, AZ 85287–1804, U.S.A.
e-mail:rlopez14@asu.edu, sks@asu.edu
Recibido el 14 de agosto de 2009; aceptado el 20 de agosto de 2009
We construct an explicit solution of the Cauchy initial value problem for the one-dimensional Schrödinger equation with a time-dependent
Hamiltonian operator for the forced harmonic oscillator. The corresponding Green function (propagator) is derived with the help of the
generalized Fourier transform and a relation with representations of the Heisenberg–Weyl group N (3) in a certain special case first, and
then is extended to the general case. A three parameter extension of the classical Fourier integral is discussed as a by-product. Motion of
a particle with a spin in uniform perpendicular magnetic and electric fields is considered as an application; a transition amplitude between
Landau levels is evaluated in terms of Charlier polynomials. In addition, we also solve an initial value problem to a similar diffusion-type
equation.

Keywords: The Cauchy initial value problem; the Schrödinger equation; forced harmonic oscillator; Landau levels; the hypergeometric
functions; the Hermite polynomials; the Charlier polynomials; Green functions; Fourier transform and its generalizations; the Heisenberg–
Weyl group N (3).
En el presente trabajo construimos una solucion explı́cita unidimensional a la ecuación de Schrödinger con condiciones iniciales de Cauchy
y con un operador Hamiltoniano dependiente del tiempo para el oscilador armónico forzado. La correspondiente función de Green (propa-
gador) se deriva con aplicaciones de la transformada de Fourier generalizada y con una relación a las representaciones del grupo N (3) de
Heisenberg–Weyl, para un caso especial primero y después se extiende al caso general. Estudiamos por medio de un producto una extención
de tres parámetros a la integral clásica de Fourier. Consideramos, como una aplicacion, el movimiento de una partı́cula giratoria en un
campo eléctrico y en un campo magnético perpendicularmente uniforme; evaluamos en términos de polinomios de Charlier una transición
de amplitud entre los niveles de Landau. Además resolvemos una ecuación similar a la de difusión con valores iniciales.

Descriptores: Problema de valor inicial de Cauchy; ecuación de Schrödinger; osilador armónico forzado; niveles de Landau; funciones
hipergeometricas; polinomios de Hermite; polinomios de Charlier; funciones de Green; transformada de Fourier y sus generalizaciones; el
grupo Heisenberg–Weyl.
PACS: 45.20.-d; 02.30.-f; 02.30.Nw

1. Introduction where δ (t) is a complex valued function of time t and the

symbol ∗ denotes complex conjugation. This operator is Her-
The time-dependent Schrödinger equation for the one- mitian, namely, H † (t) = H (t) . It corresponds to the case
dimensional harmonic oscillator has the form of the forced harmonic oscillator which is of interest in many
∂ψ advanced problems. Examples include polyatomic molecules
i} = Hψ, (1)
∂t in varying external fields, crystals through which an electron
where the Hamiltonian is is passing and exciting the oscillator modes, and other in-
µ ¶
}ω ∂2 }ω ¡ † ¢ teractions of the modes with external fields. It has particular
H= − 2 + x2 = aa + a† a . (2) applications in quantum electrodynamics because the electro-
2 ∂x 2
magnetic field can be represented as a set of forced harmonic
Here a† and a are the creation and annihilation operators, re- oscillators [9, 20, 23, 34, 35, 45]. Extensively used propagator
spectively, given by techniques were originally introduced by Richard Feynman
µ ¶ µ ¶
† 1 ∂ 1 ∂ in Refs. 16 to 19.
a =√ x− , a= √ x+ ; (3)
2 ∂x 2 ∂x On this note we construct an exact solution of the time-
see [21] for another definition. They satisfy the familiar com- dependent Schrödinger equation
mutation relation ∂ψ
£ ¤ i} = H (t) ψ (6)
∂t
a, a† = aa† − a† a = 1. (4)
A natural modification of the Hamiltonian operator (2) is with the Hamiltonian of the form (5), subject to the initial
as follows: condition
}ω ¡ † ¢ ψ (x, t)|t=0 = ψ0 (x) , (7)
H → H (t) = aa + a† a
2
¡ ¢ where ψ0 (x) is an arbitrary square integrable complex-valued
+ } δ (t) a + δ ∗ (t) a† , (5) function from L2 (−∞, ∞) . We shall start with a particular
 THE CAUCHY PROBLEM FOR A FORCED HARMONIC OSCILLATOR 197

choice of the time-dependent function δ (t) given by (22) be- 2. The simple harmonic oscillator in one di-
low, which is later extended to the general case. The explicit mension
form of Eq. (6) is given by (25) and (77) below, and an ex-
tension to similar diffusion-type equations is also discussed. The time-dependent Hamiltonian operator (5) has the follow-
ing structure:
H (t) = H0 + H1 (t) , (8)
This paper is organized as follows. In Sec. 2 we re- where
}ω ¡ † ¢
mind the reader about the textbook solution of the station- H0 = aa + a† a (9)
2
ary Schrödinger equation for the one-dimensional simple har-
is the Hamiltonian of the harmonic oscillator and
monic oscillator. In Sec. 3 we consider the eigenfunction
¡ ¢
expansion for the time-dependent Schrödinger equation (6) H1 (t) = } δ (t) a + δ ∗ (t) a† (10)
and find its particular solutions in terms of the Charlier poly-
nomials for a certain forced harmonic oscillator. The series is the time-dependent “perturbation”, which corresponds to
solution to the corresponding initial value problem is ob- an external time-dependent force that does not depend on
tained in Sec. 4. It is further transformed into an integral the coordinate x (dipole interaction) and a similar velocity-
form in Sec. 7 after discussing two relevant technical tools, dependent term (see Refs. 20, 23, and 34 for more details).
namely, the representations of the Heisenberg–Weyl group The solution to the stationary Schrödinger equation for
N (3) and the generalized Fourier transform in Sec. 5 and 6, the one-dimensional harmonic oscillator
µ ¶
respectively. An important special case of the Cauchy ini- }ω ∂2 2
tial value problem for the simple harmonic oscillator is out- H0 Ψ = EΨ, H0 = − 2 +x (11)
2 ∂x
lined in Sec. 8 and a three-parameter generalization of the
Fourier transform is introduced in Sec. 9 as a by-product. is a standard textbook problem in quantum mechanics (see
In Secs. 10 and 11 we solve the initial value problem for Refs. 13, 21, 28, 34, 35, 39, 45, and 52 for example). The
the general forced harmonic oscillator in terms of the corre- orthonormal wave functions are given by
sponding Green function (or Feynman’s propagator) and the 1 2
−x /2
eigenfunction expansion, respectively, by a different method Ψ = Ψn (x) = p √ e Hn (x) (12)
2n n! π
that uses all technical tools developed before in the special
case. An extension to the case of time-dependent frequency with
is given in Sec. 12. Then in Sec. 13, we outline important Z∞ (
1, n = m,
special and limiting cases of the Feynman propagators. Fi- Ψ∗n (x) Ψm (x) dx = δnm = (13)
nally in Sec. 14, the motion of a charged particle with a spin 0, n 6= m,
−∞
in uniform magnetic and electric fields that are perpendicular
to each other is considered as an application; we evaluate a where Hn (x) are the Hermite polynomials, a family of the
transition amplitude between Landau levels under the influ- (very) classical orthogonal polynomials (see Refs. 1, 2, 4,
ence of the perpendicular electric field in terms of Charlier 12, 14, 38, 39, 41, and 49). The corresponding oscillator dis-
polynomials and find the corresponding propagator in three crete energy levels are
dimensions. Solutions to similar diffution-type equations are µ ¶
1
discussed in Sec. 15. E = En = }ω n + (n = 0, 1, 2, ... ) . (14)
2
The actions of the creation and annihilation operators (3)
on the oscillator wave functions (12) are given by
The Cauchy initial value problem for a forced harmonic
√ √
oscillator was originally considered by Feynman in his path a Ψn = n Ψn−1 , a† Ψn = n + 1 Ψn+1 . (15)
integrals approach to the nonrelativistic quantum mechanics
[16,17,20]. Since then this problem and its special and limit- These “ladder” equations follow from the differentiation for-
ing cases were discussed by many authors [7,23,25,29,34,50] mulas
the simple harmonic oscillator; [3, 10, 24, 37, 43] the particle d
in a constant external field; see also references therein. It is Hn (x) = 2nHn−1 (x) = 2xHn (x) − Hn+1 (x) , (16)
dx
worth noting that an exact solution to the n-dimensional time-
dependent Schrödinger equation for a certain modified oscil- which are valid for the Hermite polynomials.
lator is found in Ref. 30. These simple exactly solvable mod-
els may be of interest in a general treatment of the non-linear 3. Eigenfunction expansion for the time-
time-dependent Schrödinger equation (see Refs. 26, 27, 36, dependent Schrödinger equation
44, 46, 53, and references therein). They also provide explicit
solutions which can be useful for testing numerical methods In the spirit of Dirac’s time-dependent perturbation theory in
of solving the time-dependent Schrödinger equation. quantum mechanics (see Refs. 13, 21, 28, 35, and 45), we are

Rev. Mex. Fı́s. 55 (2) (2009) 196–215

198 R.M. LOPEZ AND S.K. SUSLOV

looking for a solution to the initial value problem in (6)–(7) In the next section we will obtain the function cnm (t) in
as an infinite series terms of the Charlier polynomials; see Eq. (33) below. In
∞
X Sec. 5 we establish a relation with the representations of the
ψ = ψ (x, t) = cn (t) Ψn (x) , (17) Heisenberg–Weyl group N (3); see Eq. (46). A generaliza-
n=0 tion to an arbitrary function δ (t) will be given later.
where Ψn (x) are the oscillator wave functions (12) which 4. Solution of the Cauchy problem
depend only on the space coordinate x and cn (t) are the
yet unknown time-dependent coefficients. Substituting this We can now construct the exact solution to the origi-
form of solution into the Schrödinger equation (6) with the nal Cauchy problem in (6)–(7) for the time-dependent
help of the orthogonality property (13) and the “ladder” rela- Schrödinger equation with the Hamiltonian of the
tions (15), we obtain the following linear infinite system: form (8)–(10) and (22). More explicitly, we will solve the
µ ¶ following partial differential equation:
dcn (t) 1 √
i =ω n+ cn (t) + δ (t) n + 1 cn+1 (t) µ 2 ¶
dt 2 ∂ψ ω ∂ ψ 2
i = − 2 +x ψ
∗
√ ∂t 2 ∂x
+ δ (t) n cn−1 (t) (n = 0, 1, 2, ... ) (18) µ ¶
p ∂ψ
of the first-order ordinary differential equations with + 2µ (cos (ω−1) t) xψ+i (sin (ω−1) t) (25)
∂x
c−1 (t) ≡ 0. The initial conditions are
subject to the initial condition
Z∞
ψ (x, t)|t=0 = ψ0 (x) (−∞ < x < ∞) . (26)
cn (0) = Ψ∗n (x)ψ0 (x) dx (19)
−∞
By (17), (19), and (23) our solution has the form
∞
X X∞
due to the initial data (7) and the orthogonality property (13). ψ (x, t) = Ψn (x) cnm (t)
Now we specify the exact form of the function δ (t) in or- n=0 m=0
der to find a particular solution to the system (18) in terms of Z∞
the so-called Charlier polynomials that belong to the classical
× Ψm (y) ψ0 (y) dy, (27)
orthogonal polynomials of a discrete variable (see Refs. 11,
−∞
12, 14, 38, and 39). One can easily verify that the following
Ansatz where
³ ´n−m r m!
µn/2
n cnm (t) = −µ 1/2
e−i(ω(n+1/2)−(n+µ))t
cn (t) = (−1) √ n!
n!
¡ ¢ ∞
µm X −ikt µ µk
× e−i(ω(n+1/2)−(n+µ))t e−iλt pn (λ) (20) × e cn (k) cµm (k) e−µ
m! k!
k=0
gives the three-term recurrence relation
n−m µ(n+m)/2 −i(ω(n+1/2)−(n+µ))t
λpn (λ)= − µpn+1 (λ)+(n + µ)pn (λ)−npn−1 (λ) (21) = (−1) e−µ √ e
n!m!
¡ −it ¢k
for the Charlier polynomials pn (λ) = cµn (λ) (see Refs. 38 ∞
X µe
µ µ
and 39 for example), when we choose × cn (k) cm (k) , (28)
k!
k=0
√ √
δ (t) = µ ei(ω−1)t , δ ∗ (t) = µ e−i(ω−1)t (22) in view of the superposition principle and the orthogonality
property
with the real parameter µ such that 0 < µ < 1. Thus with ∞
pn (λ) = cµn (λ) , Eq. (20) yields a particular solution of the X µk m!
cµn (k) cµm (k) e−µ = m δnm (0 < µ < 1) (29)
system (18) for any value of the spectral parameter λ. k! µ
k=0
By the superposition principle the solution to this linear
of the Charlier polynomials (see Ref. 38 for example).
system of ordinary differential equations, which satisfies the
The right-hand side of (28) can be transformed into a sin-
initial condition (19), can be constructed as a linear combina-
gle sum with the help of the following generating relation for
tion ∞
X the Charlier polynomials:
cn (t) = cnm (t) cm (0) , (23) ∞
X (µ1 µ2 s)
k
m=0 cµn1 (k) cµm2 (k) = eµ1 µ2 s (1 − µ1 s)m
where cnm (t) is a “Green” function, or a particular solution k!
k=0
that satisfies the simplest initial conditions µ ¶
s
× (1−µ2 s)n 2 F0 −n, −m; . (30)
cnm (0) = δnm . (24) (1−µ1 s) (1−µ2 s)

Rev. Mex. Fı́s. 55 (2) (2009) 196–215

 THE CAUCHY PROBLEM FOR A FORCED HARMONIC OSCILLATOR 199

(See Refs. 31 to 33, and 22 for more information and Ref. 6 where
for the definition of the generalized hypergeometric series.) Z∞
Choosing µ1 = µ2 = µ and s = e−it /µ we obtain
Tmn (α, β, γ) = Ψ∗m (x)ei(γ+βx) Ψn (x + α)dx
n+m
(−i) 2 −∞
cnm (t) = √ e−i((ω−1)n+(n+m)/2)t e−2µ sin (t/2)
2 n+m n!m! im−n i(γ−αβ/2) −ν/2
³ p ´n+m =√ e e
× e−i(µ sin t+(ω/2−µ)t) 2 2µ sin (t/2) m!n!
³ iα + β ´m ³ iα − β ´n
µ ¶
1 × √ √ cνm (n) (37)
×2 F0 −n, −m; − . (31) 2 2
4µ sin2 (t/2) ¡ ¢
with ν = α2 + β 2 /2 [38]. A similar integral
The hypergeometric series representation for the Charlier
polynomials is Z∞
µ ¶ Hm (x + y)Hn (x + z)e−x dx
2

1
cµn (x) = 2 F0 −n, −x; − (32)
µ −∞
√
(see [38] for example). Thus = π 2n m! z n−m Ln−m
m (−2yz) (38)

cnm (t) = e−i(µ sin t+(ω/2−µ)t) e−i((ω−1)n+(n+m)/2)t is evaluated in Ref. 15 in terms of the Laguerre polynomials
n+m Lαm (ξ) , whose relation with the Charlier polynomials is
(−i) 2 2
×√ e−β /4 β n+m cβn /2 (m) (33) −n
2 n+m n!m! cµn (x) = (−µ) n!Lx−n
n (µ) , (39)
√
with β = β (t) = 2 2µ sin (t/2) and, as a result, by sub- see Ref. 38. Its special case y = z in the form of
stitution of this expression into the series (27), we obtain
the eigenfunction expansion solution to the original Cauchy Z∞
2
problem (25)–(26). We shall be able to find an integral form Hm (x) Hn (x) e−(x−y) dx
of this solution in Sec. 7 after discussing representations of −∞
the Heisenberg–Weyl group N (3) and a generalization of the √ ¡ ¢
= π 2n m! y n−m Ln−m
m −2y 2 (40)
Fourier transform in the next two sections. This complete
solution to the particular initial value problem (25)–(26) will is of particular interest in this paper.
suggest a correct form of the Green function (propagator) for The unitary relation
the general forced harmonic oscillator in Secs. 10 and 11.
∞
X
∗
Tmn (α, β, γ) Tm0 n (α, β, γ) = δmm0 (41)
5. Relation with the Heisenberg–Weyl group n=0
N (3) holds due to the orthogonality property of the Charlier poly-
Let N (3) be the three-dimensional group of the upper trian- nomials (29).
gular real matrices of the form The relevant special case of these matrix elements is
 
1 α γ im+n
Tmn (0, β, 0) = tmn (β) = √
0 1 β  = (α, β, γ) . (34) 2m+n m!n!
0 0 1 2 2
× e−β /4
β m+n cβm /2 (n) , (42)
The map
which explicitly acts on the oscillator wave functions as fol-
T (α, β, γ) Ψ (x) = ei(γ+βx) Ψ (x + α) (35) lows:
X∞
defines a unitary representation of the Heisenberg–Weyl eiβx Ψn (x) = tmn (β) Ψm (x) . (43)
group N (3) in the space of square integrable functions m=0
Ψ ∈ L2 (−∞, ∞) (see Ref. 51, 38, and 47 for more details). Relations (30), (32) and (42) imply
The set {Ψn (x)}∞ n=0 of the wave functions of the har- ∞
X
monic oscillator (12) forms a complete orthonormal sys-
tmk (β1 ) tnk (β2 ) sk
tem in L2 (−∞, ∞) . The matrix elements of the representa-
k=0
tion (35) with respect to this basis are related to the Charlier
im+n
polynomials as follows: e−(β1 +β2 +2β1 β2 s)/4
2 2
=√
∞
X 2m+n m!n!
T (α, β, γ) Ψn (x) = Tmn (α, β, γ) Ψm (x) , (36) m
× (β1 + β2 s) (β2 + β1 s) cλm (n)
n
(44)
m=0

Rev. Mex. Fı́s. 55 (2) (2009) 196–215

200 R.M. LOPEZ AND S.K. SUSLOV

with where Ψn (z) are the oscillator wave functions defined

¡ ¡ ¢¢ by (12) and |r| ≤ 1, r 6= ±1 (see Refs. 14, 41, 49, and 52 for
λ = β12 + β22 + β1 β2 s + s−1 /2, example). Using the orthogonality property (13) one gets
which is an extension of the addition formula Z∞
n
∞
X r Ψn (x) = Kr (x, y) Ψn (y) dy, |r| < 1. (49)
tmk (β1 ) tnk (β2 ) = tmn (β1 + β2 ) (45) −∞
k=0
Thus the wave functions Ψn are also eigenfunctions of an
for the matrix elements. integral operator corresponding to the eigenvalues rn .
In order to obtain functions cnm (t) in terms of the matrix We denote
elements tmn (β) of the representations of the Heisenberg–
Weyl group, we compare (33) and (42). The result is ei(π/2−τ )/2
Kτ (x, y) = Keiτ (x, y) = √
n+m −i(µ sin t+(ω/2−µ)t) 2π sin τ
cnm (t) = (−1) e à ¡ 2 ¢ !
2xy − x + y 2 cos τ
× e−i((ω−1)n+(n+m)/2)t tmn (β) , (46) × exp i (50)
2 sin τ
√
where β = 2 2µ sin (t/2) . Our solution (27) takes the form
with 0 < τ < π and use the fact that the oscillator wave
∞
X ∞
X Z∞ functions are the eigenfunctions of the generalized Fourier
ψ (x, t) = Ψn (x) cnm (t) Ψm (y) ψ0 (y) dy transform
n=0 m=0 −∞ Z∞
inτ
∞
X e Ψn (x) = Kτ (x, y) Ψn (y) dy (51)
n
= e−i(µ sin t+(ω/2−µ)t) (−1) e−it(ω−1/2)n −∞
n=0

∞ Z∞ corresponding to the eigenvalues einτ . (See Refs. 5, 42, and

X
× Ψn (x) tmn (β)
m
(−1) e−imt/2 48 for more details on the generalized Fourier transform, its
m=0
inversion formula and their extensions. It is worth noting that
−∞
the classical Fourier transform corresponds to the particular
× Ψm (y) ψ0 (y) dy. (47) value τ = π/2 [52]. Its three-parameter extension will be
discussed in Sec. 9.)
In the next section, we will discuss a generalization of
the Fourier transformation, which will allow us to transform
this multiple series into a single integral form in Sec. 7; see 7. An integral form of solution
Eqs. (55) and (60)–(62) below.
Now let us transform the series (47) into a single integral
form. With the help of the inversion formula for the gen-
6. The generalized Fourier transform eralized Fourier transform (see (51) with τ → −τ ) and the
symmetry property
The Mehler generating function, or the Poisson kernel for
Hermite polynomials, is given by n
Hn (−x) = (−1) Hn (x)
∞
X 1 we get
Kr (x, y) = rn Ψn (x) Ψn (y) = p
n=0 π (1 − r2 )
Z∞
m −imt/2
à (−1) e Ψm (y) = K−t/2 (−y, z)Ψm (z)dz. (52)
¡ ¢¡ ¢!
4xyr − x2 + y 2 1 + r2 −∞
× exp , (48)
2 (1 − r2 )
Then by (43) and Fubuni’s theorem,

 
∞
X Z∞ ∞
X Z∞ Z∞
m 
tmn (β) (−1) e−imt/2 Ψm (y) ψ0 (y) dy= tmn (β) K−t/2 (−y, z)Ψm (z)dz  ψ0 (y)dy
m=0 −∞ m=0 −∞ −∞

Z Z∞ Ã ∞
! Z Z∞
X ¡ ¢
= K−t/2 (−y, z) tmn (β) Ψm (z) ψ0 (y)dydz = K−t/2 (−y, z) eiβz Ψn (z) ψ0 (y) dydz. (53)
−∞ m=0 −∞

Rev. Mex. Fı́s. 55 (2) (2009) 196–215

 THE CAUCHY PROBLEM FOR A FORCED HARMONIC OSCILLATOR 201

Now the series (47) takes the form

∞
X Z Z∞
−i(µ sin t+(ω/2−µ)t) n −it(ω−1/2)n ¡ ¢
ψ (x, t) = e (−1) e Ψn (x) K−t/2 (−y, z) eiβz Ψn (z) ψ0 (y) dydz
n=0 −∞

Z Z∞ Ã ∞
!
X
= e−i(µ sin t+(ω/2−µ)t) K−t/2 (−y, z)eiβz e−it(ω−1/2)n Ψn (−x) Ψn (z) ψ0 (y) dydz
−∞ n=0
 
Z∞ Z∞
= e−i(µ sin t+(ω/2−µ)t)  K−t/2 (−y, z) eiβz Kt(1/2−ω) (−x, z) dz  ψ0 (y) dy (54)
−∞ −∞

in view of the generating relations (48) and (50). Thus

Z∞
−i(µ sin t+(ω/2−µ)t)
ψ (x, t) = e Gt (x, y) ψ0 (y) dy, (55)
−∞

where we define the kernel as

Z∞
Gt (x, y) := Kt(1/2−ω) (−x, z) eiβz K−t/2 (−y, z) dz. (56)
−∞

This can be evaluated with the help of the familiar elementary integrals
Z∞ Z∞ r
−x2
√ i(az 2 +2bz ) πi −ib2 /a
e dx = π, e , dz = e (57)
a
−∞ −∞

(see Refs. 9, 15, and 40 also). Denoting τ1 = t (ω − 1/2) and τ2 = t/2, from (50) we get
ei(ωt−π)/2
ei(x cot τ1 +y cot τ2 )/2 ei(β+x/ sin τ1 +y/ sin τ2 )z ei(cot τ1 +cot τ2 )z /2 (58)
2 2 2
K−τ1 (−x, z) eiβz K−τ2 (−y, z) = √
2π sin τ1 sin τ2
and
Z∞
ei(ωt−π)/2
ei(x cot τ1 +y cot τ2 )/2
2 2
K−τ1 (−x, z) eiβz K−τ2 (−y, z) dz = √
2π sin τ1 sin τ2
−∞

Z∞
ei((β+x/ sin τ1 +y/ sin τ2 )z+(cot τ1 +cot τ2 )z /2)
2
× dz. (59)
−∞

As a result
à !
2
ei(ωt−π/2)/2 i(x2 cot τ1 +y2 cot τ2 )/2 sin τ1 sin τ2 (β + x/ sin τ1 + y/ sin τ2 )
Gt (x, y) = √ e exp (60)
2π sin ωt 2i sin ωt
√
with τ1 = t (ω − 1/2) , τ2 = t/2 and β = β(t) = 2 2µ sin(t/2). Thus the explicit form of this kernel is given by
á ¢ ¡ ¢ ! µ ¶
ei(ωt−π/2)/2 x2 + y 2 sin ωt − x2 − y 2 sin (ω − 1) t ikt2 (x, y)
Gt (x, y) = √ exp exp , (61)
2π sin ωt 2i (cos ωt − cos (ω − 1) t) sin ωt (cos ωt − cos (ω − 1) t)
where
kt (x, y) = (x + y) sin (ωt/2) cos ((ω − 1) t/2)
p
− (x − y) cos (ωt/2) sin ((ω − 1) t/2) − 2µ sin (t/2) (cos ωt − cos (ω − 1) t) . (62)
The last expression can be transformed into a somewhat more convenient form
µ ¶ µ ¶
∗ sin ((ω − 1/2) t) sin (t/2) β 2 (x sin (t/2) + y sin ((ω − 1/2) t)) β
Gt (x, y) = Kωt (x, y) exp exp (63)
2i sin ωt i sin ωt

Rev. Mex. Fı́s. 55 (2) (2009) 196–215

202 R.M. LOPEZ AND S.K. SUSLOV
√
with β = 2 2µ sin (t/2) in terms of the kernel of the has the following explicit solution:
generalized Fourier transform (50). Our formulas (55) and
(60)–(63) provide an integral form to the solution of the ∞
X
Cauchy initial value problem (25)–(26) in terms of a Green ψ (x, t) = e−iω(n+1/2)t Ψn (x)
function. n=0

By choosing ψ0 (x) = δ (x − x0 ) , where δ(x) is the Z∞

1
Dirac delta function, we formally obtain × Ψn (y) ψ0 (y) dy = p
2πi sin (ωt)
−∞
ψ (x, t) =G(x, x0 , t)=e−i(µ sin t+(ω/2−µ)t) Gt (x, x0 ), (64) Ã ¡ !
Z∞ ¢
x2 + y 2 cos (ωt) − 2xy
which is the fundamental solution to the time-dependent × exp i ψ0 (y)dy. (72)
2 sin (ωt)
Schrödinger equation (25). One can show that −∞

lim ψ (x, t) = ψ0 (x) (65) The last relation is valid when 0 < t < π/ω. Analytic con-
t→0+
tinuation in a larger domain is discussed in Ref. 29 and 50.
by methods of Refs. 5, 42, and 52. The details are left to the
Equation (72) gives the time evolution operator (66) for
reader.
the simple harmonic oscillator in terms of the generalized
The time evolution operator for the time-dependent
Fourier transform. This result and its extension to a general
Schrödinger equation (6) can formally be written as
forced harmonic oscillator without the velocity-dependent
   term in the Hamiltonian are well-known (see Refs. 7, 17,
Zt
i 20, 23, 25, 29, 34, 50, and references therein; further gener-
U (t, t0 ) = T exp − H (t0 ) dt0  , (66)
} alizations are given in Secs. 10–12; more special cases will
t0
be discussed in Sec. 13).
where T is the time ordering operator which orders operators
with larger times to the left [9], [21]. Namely, this unitary
operator takes a state at time t0 to a state at time t, so that
9. Three parameter generalization of the
ψ (x, t) = U (t, t0 ) ψ (x, t0 ) (67) Fourier transform

and The properties of the time evolution operator in (67)–(70)

U (t, t0 ) = U (t, t0 ) U (t0 , t0 ) , (68) suggest the following extension of the classical Fourier in-
tegral:
U −1 (t, t0 ) = U † (t, t0 ) = U (t0 , t) . (69)
Z∞
We have constructed this time evolution operator explic- f (x) = Lt (x, y) g (y) dy, (73)
itly as the following integral operator
−∞
−i(µ sin(t−t0 )+(ω/2−µ)(t−t0 ))
U (t, t0 ) ψ (x, t0 ) = e
where the kernel given by
Z∞
× Gt−t0 (x, y) ψ (y, t0 ) dy (70) Lt (x, y) = Kωt (x, y)
−∞ µ ¶
sin ((ω − 1/2) t) sin3 (t/2) 2
with the kernel given by (60)–(63), for the particular form of × exp i ε
2 sin ωt
the time-dependent Hamiltonian in (8)–(10) and (22). The ³ (x sin(t/2)+y sin((ω − 1/2)t)) sin(t/2) ´
Green function (propagator) for the general forced harmonic × exp i ε (74)
oscillator is constructed in Sec. 10; see Eqs. (79)–(84). sin ωt

depends on the three free parameters t, ω and ε. If ε = 0

8. The Cauchy problem for the simple har- and ωt = τ we arrive at the kernel of the generalized Fourier
monic oscillator transform (50). The formal inversion formula is given by

In an important special case µ = 0, the initial value problem Z∞

µ 2 ¶ g (y) = L∗t (x, y) f (x) dx. (75)
∂ψ ω ∂ ψ 2
i = − 2 +x ψ , −∞
∂t 2 ∂x
ψ (x, t)|t=0 = ψ0 (x) (−∞ < x < ∞) (71) The details are left to the reader. Note that, in terms of a

Rev. Mex. Fı́s. 55 (2) (2009) 196–215

 THE CAUCHY PROBLEM FOR A FORCED HARMONIC OSCILLATOR 203

distribution, provided a (0) = b (0) = c (0) = 0. The case g (t) ≡ 0 is

Z∞ discussed in Refs. 17, 20, and 34, but the answers for b (t)
L∗t (x, y) Lt (x, z) dx and c (t) are given in different forms; we shall elaborate on
this later.
−∞
Indeed, the previously found solution (63)–(64) in the
= ei cot(ωt)(y −z 2 )/2
2
eiε(z−y) sin((ω−1/2)t) sin(t/2)/ sin ωt special case of the forced oscillator (25) suggests to look for
a general Green function in the form (80), namely,
Z∞
1
× eix(z−y)/ sin ωt dx = δ(y − z), (76) ψ = u eiS , (85)
2π sin ωt
−∞
where u = G0 (x, y, t) is the fundamental solution of the
which gives the corresponding orthogonality property of the Schödinger equation for the simple harmonic oscillator (71)
L-kernel. These results admit further generalizations with the and S = a (t) x + b (t) y + c (t) . Its substitution into (77)
help of the time evolution operators found in Secs. 10 and 12. gives
µ ¶ ³
10. The general forced harmonic oscillator da db dc ω ´
x+ y+ u = ag + xf − a2 u
dt dt dt 2
Our solution to the initial value problem (25)–(26) obtained ∂u
in the previous sections admits a generalization. The Cauchy + i (aω − g) , (86)
∂x
problem for the general forced harmonic oscillator
µ 2 ¶ where by (81)
∂ψ ω ∂ ψ 2 ∂ψ
i = − 2 +x ψ −f (t) xψ+ig (t) , (77)
∂t 2 ∂x ∂x ∂u x cos ωt − y
=i u. (87)
∂x sin ωt
where f (t) and g (t) are two arbitrary real valued functions
of time only (such that the integrals in (82)–(84) below con- As a result
verge and a (0) = 0), with the initial data
da db dc ω
x+ y+ = ag + xf − a2
ψ (x, t)|t=0 = ψ0 (x) (−∞ < x < ∞) , (78) dt dt dt 2
x cos ωt − y
has the following explicit solution: − (aω − g) , (88)
sin ωt
Z∞
and equating the coefficients of x, y and 1, we obtain the fol-
ψ (x, t) = G (x, y, t) ψ0 (y) dy. (79)
lowing system of ordinary differential equations
−∞
d
Here the Green function (or Feynman’s propagator [17], [20], (sin ωt a (t)) = f (t) sin ωt + g (t) cos ωt, (89)
[34]) is given by dt
d ωa (t) − g (t)
G (x, y, t) = G0 (x, y, t) ei(a(t)x+b(t)y+c(t)) (80) b (t) = , (90)
dt sin ωt
with d ω
c (t) = g (t) a (t) − a2 (t) , (91)
1 dt 2
G0 (x, y, t)= √
2πi sin ωt whose solutions are (82)–(84), respectively, if the integrals
à ¡ ¢ ! converge. This method is equivalent to solving the quantum
x2 +y 2 cos ωt−2xy
× exp i (81) mechanical Hamilton–Jacobi equation for the general forced
2 sin ωt harmonic oscillator [34].
and Equation (83) can be rewritten as
Zt Zt Zt
1 g (s)
a (t) = (f (s) sin ωs + g (s) cos ωs) ds, (82) b (t) = − (sin ωs a (s)) d cot ωs − ds
sin ωt sin ωs
0 0 0

Zt and, integrating by parts,

ωa (s) − g (s)
b (t) = ds, (83)
sin ωs
0 b (t)=− cos ωt a (t)
Zt ³ ´ Zt
ω
c (t) = g (s) a (s) − a2 (s) ds (84) + (f (s) cos ωs−g (s) sin ωs) ds (92)
2
0 0

Rev. Mex. Fı́s. 55 (2) (2009) 196–215

204 R.M. LOPEZ AND S.K. SUSLOV

by (89). With the help of (82) and the addition formulas for
trigonometric functions we finally arrive at
Zt ×(−f (s) cos ωs + g(s) sin ωs)ds. (94)
1
b (t) = − (f (s) sin ω (s − t)
sin ωt This can be transformed into the form given in Refs. 17
0
and 20 when g (t) ≡ 0. The details are left to the reader.
+g (s) cos ω (s − t)) ds, (93) Evaluation of elementary integrals results in (63) again in
which is equivalent to the form obtain in Refs. 17, 20, and 34, the special case (25). The simple case f (t) = 2 cos ωt and
when g (t) ≡ 0. g (t) ≡ 0 gives
In a similar fashion,
sin ωt
Zt Zt a (t) = , b (t) = t,
1 2
ω
c (t) = g (s) a (s) ds + (sin ωs a (s)) d cot ωs 1 1
2 c (t) = sin 2ωt − t. (95)
0 0
8ω 2 4ω
and as a result
Zt The corresponding propagator in (80) does satisfy the
1 Schrödinger equation (77), which can be verified by a direct
c (t) = sin ωt cos ωt a2 (t) + sin ωs a (s)
2 differentiation with the help of a computer algebra system.
0 The details are left to the reader. A case of the forced modi-
fied oscillator is discussed in Ref. 30.

11. Eigenfunction expansion for the general forced harmonic oscillator

Separation of the x and y variables in Feynman’s propagator (80) –(84) with the help of the Mehler generating function (48)
written as ∞
X
G0 (x, y, t) = e−iω(k+1/2)t Ψk (x) Ψk (y) (96)
k=0
gives
∞
X ¡ ¢¡ ¢
G (x, y, t) = G0 (x, y, t) ei(ax+by+c) = ei(c−ωt/2) e−iωkt eiax Ψk (x) eiby Ψk (y)
k=0
∞
à ∞
!Ã ∞
!
X X X
= ei(c−ωt/2) e−iωkt tnk (a) Ψn (x) tmk (b) Ψm (y)
k=0 n=0 m=0
∞ X
∞
à ∞
!
X X
i(c−ωt/2) −iωkt
=e Ψn (x) Ψm (y) e tnk (a) tmk (b) (97)
n=0 m=0 k=0

by (43). The last series can be summed by using the addition formula (44) in the form
∞
X im+n 2 n m 2
e−iωkt tnk (a) tmk (b) = √ ei(ab sin ωt)/2 e−χ /4
(a + bz) (b + az) cχm /2 (n) , (98)
k=0
2m+n m!n!

with z = e−iωt and χ2 = a2 + b2 + 2ab cos ωt. As a result we arrive at the following eigenfunction expansion of the forced
harmonic oscillator propagator:
∞ X
X ∞
2 in+m n m 2
G (x, y, t) = ei(c−(ωt−ab sin ωt)/2) e−χ /4
Ψn (x) Ψm (y) √ (a + bz) (b + az) cχm /2 (n) (99)
n=0 m=0 2n+m n!m!

in terms of the Charlier polynomials. The special case g (t) ≡ 0 is discussed in Ref. 20 but the connection with the Charlier
polynomials is not emphasized.
The solution (79) takes the form
∞
X ∞
X Z∞
ψ (x, t) = Ψn (x) cnm (t) Ψm (y) ψ0 (y) dy, (100)
n=0 m=0 −∞

Rev. Mex. Fı́s. 55 (2) (2009) 196–215

 THE CAUCHY PROBLEM FOR A FORCED HARMONIC OSCILLATOR 205

where
2 in+m n m 2
cnm (t) = ei(c−(ωt−ab sin ωt)/2) e−χ /4
√ (a + bz) (b + az) cχm /2 (n) (101)
2n+m n!m!
with z = e−iωt and χ2 = a2 + b2 + 2ab cos ωt. Functions a = a (t) , b = b (t) and c = c (t) here are given by the integrals
(82)–(84), respectively, and limt→0+ cnm (t) = δnm .
If ψ0 (x) = ψ (x, t)|t=0 = Ψm (x) , Eq. (100) becomes
∞
X
ψ (x, t) = cnm (t) Ψn (x) . (102)
n=0

Thus function cnm (t) gives explicitly the quantum mechanical amplitude that the oscillator initially in state m is found at time
t in state n [20]. An application to the motion of a charged particle with a spin in uniform perpendicular magnetic and electric
fields is considered in Sec. 14.
As a by-product we found the fundamental solution cnm (t) √ of the system (18) in terms of the Charlier polynomials for
an arbitrary complex valued function δ (t) = (−f (t) + ig (t)) / 2. The explicit solution of the corresponding initial value
problem in given by (23).

12. Time-dependent frequency

and the system (89)–(91) becomes
An extension of the Schödinger equation to the case of
the forced harmonic oscillator with the time-dependent fre- d
(sin τ a (τ )) = f1 (τ ) sin τ + g1 (τ ) cos τ, (109)
quency is as follows: dτ
µ 2 ¶ d a (τ ) − g1 (τ )
∂ψ ω(t) ∂ ψ ∂ψ b (τ ) = , (110)
i = − 2 +x2 ψ −f (t)xψ+ig(t) , (103) dτ sin τ
∂t 2 ∂x ∂x
d 1
c (τ ) = g1 (τ ) a (τ ) − a2 (τ ) . (111)
where ω (t) > 0, f (t) and g (t) are arbitrary real valued dτ 2
functions of time only. It can be easily solved by the sub-
Thus
stitution
Zt Zt
dτ 1
τ = τ (t) = ω (s) ds, = ω (t) , (104) a (τ )= (f (s) sin τ (s) +g (s) cos τ (s)) ds, (112)
dt sin τ
0
0
Zt
which transforms this equation into a familiar form ω (s) a (τ (s)) −g (s)
b (τ )= ds, (113)
µ ¶ sin τ (s)
∂ψ 1 ∂2ψ ∂ψ 0
i = − +x2 ψ −f1 (τ ) xψ+ig1 (τ ) , (105) Zt µ ¶
∂τ 2 ∂x2 ∂x 1
c (τ )= g(s)a (τ (s)) − ω (s) a2 (τ (s)) ds, (114)
see the original Eq. (77) with respect to the new time variable 2
0
τ, where ω = 1 and
which is an extension of equations (82)–(83) to the case of
f (t) f (t) the forced harmonic oscillator with the time-dependent fre-
f1 (τ ) = , g1 (τ ) = . (106)
ω (t) ω (t) quency. The solution to the Cauchy initial-value problem is
given by
Therefore by (80)–(81) the propagator has the form
Z∞
G (x, y, τ ) = G0 (x, y, τ ) ei(a(τ )x+b(τ )y+c(τ )) (107) ψ (x, t) = G (x, y, τ ) ψ0 (y) dy (115)
−∞
with
1 with
G0 (x, y, τ ) = √
2πi sin τ Zt
à ¡ ¢ !
x2 + y 2 cos τ − 2xy τ= ω (s) ds.
× exp i (108) 0
2 sin τ
The details are left to the reader.

Rev. Mex. Fı́s. 55 (2) (2009) 196–215

206 R.M. LOPEZ AND S.K. SUSLOV

13. Some special cases Here

r
The time-dependent Schrödinger equation for the forced har- mω
G0 (x, y, t, t0 ) =
monic oscillator is usually written in the form 2πi} sin ω (t − t0 )
Ã
imω £ 2
∂Ψ × exp (x + y 2 )
i} = HΨ (116) 2} sin ω(t − t0 )
∂t
!
with the following Hamiltonian ¤
× cos ω(t − t0 ) − 2xy , (124)
p2 mω 2 2 } ∂
H= + x −F (t) x−G (t) p, p= , (117) Zt Ã
2m 2 i ∂x mω sin ω(s − t0 )
a (t, t0 ) = F (s)
} sin ω (t − t0 ) mω
where } is the Planck constant, m is the mass of the particle, t0
ω is the classical oscillation frequency, F (t) is a uniform in !
space external force depending on time, function G(t) rep- + G(s) cos ω(s − t0 ) ds, (125)
resents a similar velocity-dependent term, and p is the linear
momentum operator. The initial value problem is
b (t, t0 ) = −a (t0 , t) (126)
2 2 2
∂Ψ } ∂ Ψ mω 2
i} =− + x Ψ and
∂t 2m ∂x2 2
∂Ψ Zt µ ¶
−F (t) xΨ+i}G (t) (118) } 2
∂x c (t, t0 ) = G (s) a (s, t0 ) − a (s, t0 ) ds. (127)
2m
t0
with
The simple harmonic oscillator propagator, when
Ψ (x, t)|t=t0 = Ψ (x, t0 ) . (119) F = G = 0, is given by Eq. (124); see Refs. 7, 17, 20, 23,
25, 29, 34, 50, and references therein for more details. In the
Among important special cases are: the free particle, when limit ω → 0 we obtain
ω = F = G = 0; a particle in a constant external field, Ã !
where ω = G = 0 and F = constant; the simple harmonic r
m im(x − y)2
oscillator with F = G = 0. In this section for the benefits G0 (x, y, t, t0 )= exp (128)
2πi}(t−t0 ) 2}(t − t0 )
of the reader we provide explicit forms for the corresponding
propagators by taking certain limits in the general solution.
as the free particle propagator [20].
The usual change of the space variable
r
mω
Ψ (x, t) = ψ (ξ, t) , ξ= x (120)
}

reduces Eq. (118) to the form (77) with respect to ξ with

r
F (t) mω
f (t) = √ , g (t) = G (t) . (121)
}ωm }

The time evolution operator is

Z∞
Ψ (x, t) = G (x, y, t, t0 ) Ψ (y, t0 ) dy (122)
−∞

with the propagator of the form

G (x, y, t, t0 ) = G0 (x, y, t, t0 )
× ei(a(t,t0 )x+b(t,t0 )y+c(t,t0 )) . (123) F IGURE 1. Magnetic and electric fields in R3 .

Rev. Mex. Fı́s. 55 (2) (2009) 196–215

 THE CAUCHY PROBLEM FOR A FORCED HARMONIC OSCILLATOR 207

For a particle in a constant external field ω = G = 0 and wave function in the Schrödinger equation (131) can be taken
F = constant. The corresponding propagator is given by as an ordinary coordinate function Ψ = Ψ (r, t, σ) .
r à ! The Hamiltonian (133) does not contain the coordinates
2
m im (x − y) x and z explicitly. Therefore the operators pbx and pbz also
G (x, y, t, t0 ) = exp
2πi} (t − t0 ) 2} (t − t0 ) commute with the Hamiltonian and the x and z components
µ ¶ of the linear momentum are conserved. The corresponding
iF (x + y) iF 2 3
× exp (t − t0 ) − (t − t0 ) . (129) eigenvalues px and pz take all values from −∞ to ∞; see [28]
2} 24}m for more details. In this paper we consider the simplest case
This case was studied in detail in Refs. 3, 10, 20, 24, 37, when the magnetic field H is a constant and the electric force
and 43. We have corrected a typo in Ref. 20. F is a function of time t (see Fig. 1); a more general case
will be discussed elsewhere. Then the substitution

14. Motion in uniform perpendicular mag-

Ψ (r, t) = ei(xpx +zpz −S(t,t0 ))/} ψ (y, t) , (134)
netic and electric fields
14.1. Solution of a particular initial value problem where

A particle with a spin s has also an intrinsic magnetic mo- µ ¶ Zt

mentum µ with the operator p2z µσ cpx
S(t, t0 )= − H (t − t0 ) + F (τ ) dτ, (135)
2m s eH
t0
b = µb
µ s/s, (130)

where bs is the spin operator and µ is a constant character- results in the one-dimensional time-dependent Schrödinger
izing the particle, which is usually called the magnitude of equation of the harmonic oscillator driven by an external
the magnetic momentum. For the motion of a charged parti- force in the y-direction
cle in uniform magnetic H and electric E fields, which are
perpendicular to each other (Fig. 1), the corresponding three- ∂ψ }2 ∂ 2 ψ
dimensional time-dependent Schrödinger equation i} =−
∂t 2m ∂y 2
2
∂Ψ b mωH 2
i} = HΨ (131) + (y − y0 ) ψ − F (t) (y − y0 ) ψ (136)
∂t 2
has the Hamiltonian of the form [28]
with
µ ¶2
Hb = 1 pbx + eH y |e| H cpx
2m c ωH = , y0 = − . (137)
mc eH
1 2 1 2 µ
+ pby + pb − sbz H − yF, (132)
2m 2m z s The Cauchy initial-value problem subject to special data
where p b = −i}∇ is the linear momentum operator, functions
H and F/e are the magnitudes of the uniform magnetic and Ψ (r, t)|t=t0 = ei(xpx +zpz )/} ψ (y, t0 )
electric fields in z and y directions, respectively. The corre-
sponding vector potential A = −yH ex is defined up to a = ei(xpx +zpz )/} ϕ (y − y0 ) (138)
gauge transformation. Here we follow the original choice of
Ref. 28 (see a remark at the end of this section). has the following solution:
Since (132) does not contain the other components of the
spin, the operator sbz commutes with the Hamiltonian H b and
the z-component of the spin is conserved. Thus the operator Ψ (r, t) = Ψ (r, t, px , pz ) = ei(xpx +zpz −S(t,t0 ))/}
sbz can be replaced by its eigenvalue sz = σ in the Hamilto- Z∞
nian × G (y − y0 , η, t, t0 ) ϕ (η) dη, (139)
µ ¶2 −∞
b 1 eH 1 2
H= pbx + y + pb
2m c 2m y
where the propagator takes the form
1 2 µσ
+ pb − H − yF (133)
2m z s
G (y, η, t, t0 ) = G1 (y, η, t − t0 )
with σ = −s, −s + 1, ... , s − 1, s. Then the spin depen-
dence of the wave function becomes unimportant and the × ei(a(t,t0 )y+b(t,t0 )η+c(t,t0 )) (140)

Rev. Mex. Fı́s. 55 (2) (2009) 196–215

208 R.M. LOPEZ AND S.K. SUSLOV

with The first term here gives the discrete energy values corre-
r sponding to motion in a plane perpendicular to the field. They
mωH
G1 (y, η, t) = are called Landau levels. The expression (149) does not con-
2πi} sin ωH t
µ ¶ tain the quantity px , which takes all real values. Therefore
imωH ¡¡ 2 ¢ ¢ the total energy levels are continuously degenerate. For an
× exp y + η 2 cos ωH t − 2yη , (141)
2} sin ωH t electron, µ/s = − |e| }/mc, and formula (149) becomes
1 µ ¶
a (t, t0 ) = 1 p2
} sin ωH (t − t0 ) En = En (pz , σ) = }ωH n + + σ + z . (150)
2 2m
Zt
× F (τ ) sin ωH (τ − t0 ) dτ, (142) In this case, there is an additional degeneracy: the levels with
n, σ = 1/2 and n + 1, σ = −1/2 coincide:
t0

b (t, t0 ) = −a (t0 , t) (143) En (pz , 1/2) = En+1 (pz , −1/2) .

and The three-dimensional wave functions corresponding to the
Zt energy levels (149) are given by
}
c (t, t0 ) = − a2 (τ, t0 ) dτ. (144)
2m Ψn (r, t, σ) = Ψn (r, t, px , pz , σ)
t0

See Eqs. (122)–(127) with G ≡ 0. Function c (t, t0 ) can be = e−iEn (pz ,σ)(t−t0 )/} ei(xpx +zpz )/} χn (y) . (151)
written in several different forms.
They are the eigenfunctions of the following set of commut-
14.2. Landau levels b with F ≡ 0 :
ing operators pbx , pbz , sbz , and H

In an absence of the external force F ≡ 0, Eq. (136) is for- b n = En Ψn ,

HΨ sbz Ψn = σΨn ,
mally identical to the time-dependent Schrödinger equation
for a simple harmonic oscillator with the frequency ωH . The pbx Ψn = px Ψn , pbz Ψn = pz Ψn . (152)
standard substitution
The orthogonality relation in R3 is
−iε(t−t0 )/}
ψ (y, t) = e χ (y) (145) Z
gives the corresponding stationary Schrödinger equation as Ψ∗n (r, t, px , pz , σ) Ψm (r, t, p0x , p0z , σ 0 ) dxdydz
follows [28]: R3
µ ¶
2m 1 2
2
= (2π}) δnm δσσ0 δ (px − p0x ) δ (pz − p0z ) , (153)
χ00 + 2 ε − mωH 2
(y − y0 ) χ = 0, (146)
} 2
which has the square integrable solutions only when where
µ ¶ Z∞
1 1
ε = }ωH n + , n = 0, 1, 2, ... . (147) δ (α) = eiαξ dξ (154)
2 2π
−∞
The eigenfunctions are
1 is the Dirac delta function.
χn (y) = p √
2n n!aH π
à ! µ ¶ 14.3. Transition amplitudes
2
(y − y0 ) y − y0
× exp − Hn ,
2a2H aH In the presence of external force, the quantum mechanical
r amplitude of a transition between Landau’s levels under the
} influence of the perpendicular electric field can be explicitly
aH = , (148) found as a special case of our formulas (100)–(102). Indeed,
mωH
solution (139) takes the form
where Hn (η) are the Hermite polynomials.
∞
X
Thus the total energy levels of a particle in a uniform
magnetic field have the form Ψ (r, t, σ) = e−iS(t,t0 )/} Ψn (r, t0 , σ)
µ ¶ n=0
1 p2 µσ
En = En (pz , σ) = }ωH n + + z − H ∞
X Z∞
2 2m s
× cnm (t, t0 ) χm (η) ψ (η, t0 ) dη (155)
(n = 0, 1, 2, ... ) . (149) m=0 −∞

Rev. Mex. Fı́s. 55 (2) (2009) 196–215

 THE CAUCHY PROBLEM FOR A FORCED HARMONIC OSCILLATOR 209

in view of the bilinear generating relation (99). If where functions a (px , pz ) do not depend on time t and
ψ (y, t0 ) =χm (y) , this equation becomes S (t, t0 ) is given by (135). Now we replace the special initial
data (138) in R3 by the general one
Ψ (r, t, σ) = e−iS(t,t0 )/}
∞
X Ψ (r, t)|t=t0 = φ (x, y, z) , (160)
× cnm (t, t0 ) Ψn (r, t0 , σ) , (156)
n=0

where coefficients cnm (t, t0 ) are given by (101) in terms of which is independent of px (and y0 ). Letting t → t0 in (159)
Charlier polynomials as follows: and using the fundamental property of the Green function,

cnm (t, t0 ) = ei(c−(ωH (t−t0 )−ab sin ωH (t−t0 ))/2) e−γ

2
/4 Z∞
lim G (y − y0 , η, t, t0 ) ϕ (η) dη=ϕ (y − y0 ) , (161)
in+m n m 2 t→t+
0
×√ (a + bδ) (b + aδ) cγm /2 (n) (157) −∞
2n+m n!m!
one gets
with δ=e−iωH (t−t0 ) and γ 2 =a2 + b2 + 2ab cos ωH (t − t0 ) .
Functions a = a (t, t0 ) , b = b (t, t0 ) and c = c (t, t0 ) are Z Z∞
evaluated by the integrals (142)–(144), respectively. The last φ (x, y, z) = a (px , pz ) ϕ (y − y0 )
two formulas (156)–(157) and (135) give us the quantum me- −∞
chanical amplitude that the particle initially in Landau state
m is found at time t in state n. For the particle initially in the × ei(xpx +zpz )/} dpx dpz , (162)
ground state m = 0, the probability of occupying state n at
time t is given by the Poisson distribution where y0 is a function of px in view of (137). Thus

2 µn 1
|cn0 (t, t0 )| = e−µ , a (px , pz ) ϕ (y − y0 ) = 2
n! (2π})
1¡ 2 ¢ Z Z∞
µ= a + b2 + 2ab cos ωH (t − t0 ) < 1. (158)
2 × φ (ξ, y, ζ) e−i(ξpx +ζpz )/} dξdζ (163)
The details are left to the reader. −∞

14.4. Propagator in three dimensions by the inverse of the Fourier transform. Its substitution into
(159) gives
Our particular solutions (139) subject to special initial data
(138) have been constructed above as eigenfunctions of the Z Z∞
1
operators pbx and pbz , whose continuous eigenvalues px and Ψ (r, t) = 2 dpx dpz ei(xpx +zpz −S(t,t0 ))/}
pz vary from −∞ to ∞. By the superposition principle, one (2π})
−∞
can look for a general solution in R3 as a double Fourier in- Z∞
tegral of the particular solution
× dη G (y − y0 , η − y0 , t, t0 )
Z Z∞ −∞
Ψ (r, t) = a (px , pz ) Ψ (r, t, px , pz ) dpx dpz
Z Z∞
−∞
× φ (ξ, η, ζ) e−i(ξpx +ζpz )/} dξdζ (164)
Z Z∞ −∞
= dpx dpz a (px , pz )ei(xpx +zpz )/} e−iS(t,t0 )/}
−∞
as a solution of our initial value problem. A familiar integral
form of this solution is as follows:
Z∞ Z
× G (y − y0 , η, t, t0 ) ϕ (η) dη, (159) Ψ (r, t) = G (r, ρ, t, t0 ) φ (ξ, η, ζ) dξdηdζ, (165)
−∞
R3

where the Green function (propagator) is given as a double Fourier integral

Z Z∞
1
G (r, ρ, t, t0 ) = 2 ei((x−ξ)px +(z−ζ)pz )/} e−iS(t,t0 )/} G (y − y0 , η − y0 , t, t0 ) dpx dpz (166)
(2π})
−∞

with the help of the Fubini theorem.

Rev. Mex. Fı́s. 55 (2) (2009) 196–215

210 R.M. LOPEZ AND S.K. SUSLOV

This integral can be evaluated in terms of elementary functions as follows. Integration over pz gives the free particle
propagator of a motion in the direction of magnetic field
Z∞ µ µ ¶¶ r à !
2
1 i p2z m im (z − ζ)
G0 (z − ζ, t − t0 ) = exp (z − ζ) pz − (t − t0 ) dpz = exp (167)
2π} } 2m 2πi} (t − t0 ) 2} (t − t0 )
−∞

by the integral (57). Thus

 
µ ¶ Z∞ µ ¶ Zt
iµσH 1 i icpx
G (r, ρ, t, t0 ) = exp (t − t0 ) G0 (z − ζ, t − t0 ) exp (x − ξ) px exp − F (τ ) dτ 
}s 2π} } }eH
−∞ t0
µ ¶
iµσH
× G (y − y0 , η − y0 , t, t0 ) dpx = exp (t − t0 ) G0 (z − ζ, t − t0 ) G1 (y, η, t − t0 )
}s
  
Z∞ Zt
1 ip x c
× ei(a(t,t0 )y+b(t,t0 )η+c(t,t0 )) exp  x − ξ − F (τ ) dτ 
2π} } eH
−∞ t0
µ ¶ µ µ ¶ ¶
i (a (t, t0 ) + b (t, t0 )) cpx −i p2x |e|
× exp exp + (y+η) px tan (ωH (t−t0 ) /2) dpx . (168)
eH } mωH e
In view of (57), the last integral is given by
  
Z∞ Zt µ ¶
1 ipx  c i (a (t, t0 ) + b (t, t0 )) cpx
exp  x−ξ− F (τ ) dτ  exp
2π} } eH eH
−∞ t0
µ µ ¶ ¶ r
p2x
−i e mωH cot (ωH (t − t0 ) /2)
× exp + (y + η) px tan (ωH (t − t0 ) /2) dpx =
}
mωH |e| 4πi}
µ ¶
imωH cot (ωH (t − t0 ) /2) 2
× exp β , (169)
4}
where
e
β =x−ξ− (y + η) tan (ωH (t − t0 ) /2) + d (t, t0 ) (170)
|e|
with
Zt
c
d (t, t0 ) = F (τ ) (sin ωH (τ − t0 ) − sin ωH (τ − t) − sin ωH (t − t0 )) dτ. (171)
eH sin ωH (t − t0 )
t0

Here we have used (143)–(142). As a result, we arrive at the following factorization of our propagator:
G (r, ρ, t, t0 ) = G0 (z − ζ, t − t0 ) G1 (y, η, t − t0 ) ei(a(t,t0 )y+b(t,t0 )η+c(t,t0 )) G2 (x, ξ, y, η, t, t0 ) , (172)
where G0 (z − ζ, t − t0 ) is the free particle propagator in (167), G1 (y, η, t − t0 ) is the simple harmonic oscillator propagator
in (141), and
µ ¶r µ ¶
iµσH mωH cot (ωH (t − t0 ) /2) imωH cot (ωH (t − t0 ) /2) 2
G2 (x, ξ, y, η, t, t0 ) = exp (t − t0 ) exp β (173)
}s 4πi} 4}
with β = β (x, ξ, y, η, t, t0 ) given by (170)–(171).
Our propagator can be simplified to a somewhat more convenient form as follows:
G (r, ρ, t, t0 ) = G0 (z − ζ, t − t0 ) GH (x, ξ, y, η, t − t0 ) GF (x, ξ, y, η, t, t0 ) . (174)
Here G0 (z, t) is the free particle propagator in the direction of magnetic field. The function
µ ¶
iµσHt mωH
GH (x, ξ, y, η, t) = exp
}s 4πi} sin (ωH t/2)
µ µ ¶¶
imωH ³ 2 2
´ e
× exp (x − ξ) + (y − η) cot (ωH t/2) − 2 (x − ξ) (y + η) (175)
4} |e|

Rev. Mex. Fı́s. 55 (2) (2009) 196–215

 THE CAUCHY PROBLEM FOR A FORCED HARMONIC OSCILLATOR 211

is the propagator corresponding to a motion in a plane perpendicular to the magnetic field in the absence of an electric field
(compare our expression with one in Ref. 20, where F = µ = 0, and see a remark below in order to establish an identity of
two results). The third factor

GF (x, ξ, y, η, t, t0 ) = eiWF (t,t0 )/} (176)

with
1
WF (t, t0 ) = } (a (t, t0 ) y + b (t, t0 ) η + c (t, t0 )) + mωH d (t, t0 )
4
µ ¶
e
× (d (t, t0 ) + 2 (x − ξ)) cot (ωH (t − t0 ) /2) − 2 (y + η) (177)
|e|

is a contribution from the electric field. When F = 0,

WF = 0 and GF = 1. and in view of (179)
The solution to the Cauchy initial value-problem in R3
subject to the general initial data µ ¶
µ ¶
ief (x, y) ief (ξ, η)
GH → exp GH exp −
Ψ (r, t)|t=t0 = Ψ (r, t0 ) = φ (x, y, z) (178) }c }c
Ã
has the form
Z mωH imωH ³£ ¤
= exp (x − ξ)2 + (y − η)2
Ψ (r, t) = G (r, ρ, t, t0 ) Ψ (ρ, t0 ) dξdηdζ, (179) 4πi} sin(ωH t/2) 4}
!
R3
e ´
which gives explicitly the time evolution operator for a mo- × cot(ωH t/2) − 2 (xη − ξy) , (183)
|e|
tion of a charged particle in uniform perpendicular mag-
netic and electric fields with a given projection of the spin
sz = σ in the direction of the magnetic field. By choosing which is, essentially, Eq. (3-64) on page 64 of [20], where
Ψ (r, t0 ) = δ (r − r 0 ) , where δ (r) is the Dirac delta func- we have corrected a typo. The details are left to the reader.
tion in three dimensions, we formally obtain
Ψ (r, t) = G (r, r 0 , t, t0 ) (180)
as the wave function at time t of the particle initially located 15. Diffusion-type equation
at a point r = r 0 . Then Eq. (179) gives a general solution by
the superposition principle. 15.1. Special case
Remark. The vector potential of the uniform magnetic field
in the z-direction is defined up to a gauge transformation [28] A formal substitution√of t → −it and ψ → u into Eq. (25)
with ω = 2κ and 2µ = ε yields the following time-
A = −yH ex → A0 = A + ∇f
dependent diffusion-type equation:
1 1 1
= − yH ex + xH ey = H × r (181) µ 2 ¶
2 2 2 ∂u ∂ u 2
with f (x, y) = xyH/2. The corresponding transformation =κ − x u
∂t ∂x2
of the wave function is given by µ ¶
µ ¶ ∂u
0 ief (x, y) −ε (cosh (2κ−1) t) xu+ (sinh (2κ−1) t) , (184)
Ψ → Ψ = Ψ exp ∂x
}c
µ µ ¶¶
imωH e where the initial condition is
= Ψ exp 2 xy (182)
4} |e|
u (x, t)|t=0 = u0 (x) (−∞ < x < ∞) . (185)

As in the case of the time-dependent Schrödinger equation, in order to solve this initial value problem, we use the eigenfunction
expansion method. Hence the solution is given by

∞
X ∞
X Z∞
u (x, t) = Ψn (x) cnm (t) Ψm (y) u0 (y) dy, (186)
n=0 m=0 −∞

Rev. Mex. Fı́s. 55 (2) (2009) 196–215

212 R.M. LOPEZ AND S.K. SUSLOV

where
εn+m ¡ ¢m+n
e−(ε /2)(1−e−t ) −((2κ−1)n+κ−ε2 /2)t
n−m 2
cnm (t) = (−1) √ e 1 − e−t 2 F0
2n+m n!m!
à !
2e−t
× −n, −m; 2 (187)
ε2 (1 − e−t )

by analytic continuation t → −it with ω = 2κ and µ = ε2 /2 < 1 in (28) and (31). One can easily verify that

lim cnm (t) = δnm

t→0+
√
and that if 0 < ε < 2 and κ ≥ 1/2, κ > ε2 /2,

lim cnm (t) = 0 (m, n = 0, 1, 2, ... ) .

t→∞

Thus the limiting distribution is

lim u (x, t) ≡ 0 (−∞ < x < ∞) , (188)

t→∞

which is independent of the initial data (185). Relation (47) becomes

∞
X ∞
X Z∞
−(ε2 /2) sinh t−(κ−ε2 /2)t n −t(2κ−1/2)n m
u (x, t) =e (−1) e Ψn (x) tmn (β) (−1) e−mt/2 Ψm (y) u0 (y) dy (189)
n=0 m=0 −∞

with β = β (t) = −2iε sinh (t/2) . With the help of (49), (43) and the Fubuni theorem we transform
∞ Z∞ Z Z∞ Ã ∞ !
X X
−mt/2
tmn (β) e Ψm (−y) u0 (y) dy = Ke−t/2 (−y, z) tmn (β) Ψm (z) u0 (y) dydz
m=0 −∞ −∞ m=0

Z Z∞
= Ke−t/2 (−y, z) (eγz Ψn (z)) u0 (y) dydz, (190)
−∞

where γ = iβ = 2ε sinh (t/2) . The series (189) becomes

∞
X Z Z∞
−(ε2 /2) sinh t−(κ−ε2 /2)t −t(2κ−1/2)n
u (x, t) = e e Ψn (−x) Ke−t/2 (−y, z) (eγz Ψn (z)) u0 (y) dydz
n=0 −∞

Z Z∞ Ã ∞
!
X
−(ε2 /2) sinh t−(κ−ε2 /2)t γz −t(2κ−1/2)n
=e Ke−t/2 (−y, z) e e Ψn (−x) Ψn (z) u0 (y) dydz
−∞ n=0

Z∞ µZ ∞ ¶
−(ε2 /2) sinh t−(κ−ε2 /2)t γz
=e Ke−t/2 (−y, z) e Ke−t(2κ−1/2) (−x, z) dz u0 (y) dy (191)
−∞
−∞

in view of the generating relation (48). Therefore, the integral form of the solution (186)–(187) is
Z∞
−(ε2 /2) sinh t−(κ−ε2 /2)t
u (x, t) = e Ht (x, y) u0 (y) dy, (192)
−∞

where by the definition

Z∞
Ht (x, y) := Ke−t(2κ−1/2) (−x, z) eγz Ke−t/2 (−y, z) dz. (193)
−∞

Rev. Mex. Fı́s. 55 (2) (2009) 196–215

 THE CAUCHY PROBLEM FOR A FORCED HARMONIC OSCILLATOR 213

Denoting r1 = e−t(2κ−1/2) and r2 = e−t/2 we obtain by (48) that

Z∞ à ¡ ¢¡ ¢ ¡ ¢¡ ¢ !
γz 1 1 + r12 1 − r22 x2 + 1 − r12 1 + r22 y 2
Kr1 (−x, z) e Kr2 (−y, z) dz = p exp −
π (1 − r12 ) (1 − r22 ) 2 (1 − r12 ) (1 − r22 )
−∞

Z∞ á ¢¡ ¢ ¡ ¢ ¡ ¢ ! µ ¶
1 − r12 1 − r22 γ − 2r1 1 − r22 x − 2r2 1 − r12 y 1 − r12 r22 2
× exp z exp − z dz (194)
(1 − r12 ) (1 − r22 ) (1 − r12 ) (1 − r22 )
−∞

and the integral can be evaluated with the help of an elementary formula
Z∞ r
−az 2 +2bz π b2 /a
e dz = e , a > 0. (195)
a
−∞

As a result, an analog of the heat kernel in (192)–(193) is given by

à ¡ ¢¡ ¢ ¡ ¢¡ ¢!
1 1 − r12 r22 x2 + y 2 + r12 − r22 x2 − y 2
Ht (x, y) = p exp −
π (1 − r12 r22 ) 2 (1 − r12 ) (1 − r22 )
à £¡ ¢¡ ¢ ¤2 !
1 − r12 1 − r22 γ − (r1 + r2 ) (1 − r1 r2 ) (x + y) − (r1 − r2 ) (1 + r1 r2 ) (x − y)
× exp (196)
4 (1 − r12 ) (1 − r22 ) (1 − r12 r22 )

with r1 = e−t(2κ−1/2) , r2 = e−t/2 and γ = 2ε sinh (t/2) , t > 0. The last expression can be simplified to a somewhat more
convenient form
µ µ ¶ ¶ á ¢¡ ¢ !
r1 + r2 r1 − r2 γ 1 − r12 1 − r22 γ 2
Ht (x, y) = exp − (x + y) + (x − y) exp Kr1 r2 (x, y) (197)
1 + r1 r2 1 − r1 r2 2 1 − r12 r22 4

in terms of the Mehler kernel (48). One can show that tion

lim u (x, t) = u0 (x) (198)

t→0+

by methods of Ref. 52. The details are left to the reader.

A formal substitution of u0 (x) = δ (x − x0 ) into (192)
µ ¶
gives ∂u ∂2u 2 ∂u
=κ 2
− x u + f (t) xu − g (t) , (201)
∂t ∂x ∂x
u (x, t) = H (x, x0 , t)

= e−(ε /2) sinh t−(κ−ε2 /2)t

2
Ht (x, x0 ) (199)

as the fundamental solution of the diffusion Eq. (184).

In the limit ε → 0 we obtain
where f (t) and g (t) are real-valued functions of time, sub-
Z∞
ject to the initial condition
u (x, t) = e−κt Ke−2κt (x, y) u0 (y) dy (200)
−∞

as the exact solution of the corresponding initial value prob-

lem in terms of the Mehler kernel (48). This kernel gives also
a familiar expression in statistical mechanics for the density
matrix for a system consisting of a simple harmonic oscilla- u (x, t)|t=0 = u0 (x) (−∞ < x < ∞) . (202)
tor [20].

15.2. Generalization

A formal substitution of t → −it and ψ → u into (77) with

ω = 2κ and f → f, g → −ig yields a diffusion-type equa-

Rev. Mex. Fı́s. 55 (2) (2009) 196–215

214 R.M. LOPEZ AND S.K. SUSLOV

An analog of the expansion (99) is

The exact solution is ∞ X
∞
X
Z∞ H (x, y, t) = cnm (t) Ψn (x) Ψm (y) (212)
u (x, t) = H (x, y, t) u0 (y) dy (203) n=0 m=0

−∞ with
and the Green function can be found in the form 1 2
cnm (t) = √ ec−κt−(ab/2) sinh(2κt)+λ /4
2n+m n!m!
H (x, y, t) = H0 (x, y, t) ea(t)x+b(t)y+c(t) , (204) µ ¶
n m 2
where × (a + br) (b + ar)2 F0 −n, −m; 2 . (213)
λ
r
r
H0 (x, y, t) = Here r = e−2κt , λ2 = a2 +b2 +2ab cosh (2κt) and functions
π (1 − r2 ) a (t) , b (t) and c (t) are given by the integrals (209)–(211),
à ¡ ¢¡ ¢!
4xyr − x2 + y 2 1 + r2 respectively. This can be derived by expanding the kernel
× exp (205) (204) in the double series in the same fashion as in Sec. 11,
2 (1 − r2 )
or by the substitution t → −it, a → −ia, b → −ib, and
c → −ic in (99). The coefficients cnm (t) are positive when
with r = e−2κt . Indeed, substitution of (204) into (201) gives
t > 0.
the system of equations
The solution (203) takes the form
d
(sinh (2κt) a (t))=f (t) sinh (2κt) ∞
X ∞
X
dt
u (x, t) = Ψn (x) cnm (t)
+g (t) cosh (2κt) , (206) n=0 m=0

d 2κa (t) − g (t) Z∞

b (t) = , (207)
dt sinh (2κt) × Ψm (y) u0 (y) dy. (214)
−∞
d
c (t) = κa2 (t) − g (t) a (t) (208)
dt These results can be extended to the case when parameter κ
and the solutions are is a function of time in Eq. (201). The details are left to the
reader.
1
a (t) =
sinh (2κt)
Acknowledgments
Zt
× (f (s) sinh (2κs) + g (s) cosh (2κs)) ds, (209) This paper is written as a part of the summer 2007 program on
0 analysis of the Mathematical and Theoretical Biology Insti-
tute (MTBI) at Arizona State University. The MTBI/SUMS
Zt
2κa (s) − g (s) undergraduate research program is supported by The Na-
b (t) = ds, (210) tional Science Foundation (DMS–0502349), The National
sinh (2κs)
0 Security Agency (DOD–H982300710096), The Sloan Foun-
Zt dation, and Arizona State University. The authors are grateful
¡ ¢ to Professor Carlos Castillo-Chávez for support and Ref. 8.
c (t) = κa2 (s) − g (s) a (s) ds (211)
We thank Professors George Andrews, George Gasper, Slim
0
Ibrahim, Hunk Kuiper, Mizan Rahman, Svetlana Roudenko,
provided a (0) = b (0) = c (0) = 0. and Hal Smith for valuable comments.

1. G.E. Andrews and R.A. Askey, Classical orthogonal polynomi-
als, in: Polynômes orthogonaux et applications, Lecture Notes
in Math. 1171, (Springer-Verlag, 1985) p. 36.
2. G.E. Andrews, R.A. Askey, and R. Roy, Special Functions,
(Cambridge University Press, Cambridge, 1999).
3. G.P. Arrighini and N.L. Durante, Am. J. Phys. 64 (1996) 1036.
4. R.A. Askey Orthogonal Polynomials and Special Functions,
CBMS–NSF Regional Conferences Series in Applied Mathe-
matics, (SIAM, Philadelphia, Pennsylvania, 1975).
5. R.A. Askey, M. Rahman, and S.K. Suslov, J. Comp. Appl.
Math. 68 (1996) 25.
6. W.N. Bailey, Generalized Hypergeometric Series, (Cambridge
University Press, Cambridge, 1935).
7. L.A. Beauregard, Am. J. Phys. 34 (1966) 324.

Rev. Mex. Fı́s. 55 (2) (2009) 196–215

 THE CAUCHY PROBLEM FOR A FORCED HARMONIC OSCILLATOR
8. L.M.A. Bettencourt, A. Cintrón-Arias, D.I. Kaiser, and
C. Castillo-Chávez, Phisica A 364 (2006) 513.
9. N.N. Bogoliubov and D.V.Shirkov, Introduction to the Theory
of Quantized Fields, third edition, (John Wiley & Sons, New
York, Chichester, Brisbane, Toronto, 1980).
10. L.S. Brown and Y. Zhang, Am. J. Phys. 62 (1994) 806.
11. C.V.L. Charlier, Arkiv för Matematik, Astronomi och Fysik 2
(1905) 35.
12. T.S. Chihara, An Introduction to Orthogonal Polynomials,
(Gordon and Breach, New York, 1978).
13. A. S. Davydov, Quantum Mechanics, Pergamon Press, (Oxford
and New York, 1965).
14. A. Erdélyi, Higher Transcendental Functions, Vol. II,
A. Erdélyi, ed., (McGraw–Hill, 1953).
15. A. Erdélyi, Tables of Integral Transforms, Vols. I–II, A. Erdélyi,
ed., (McGraw–Hill, 1954).
16. R.P. Feynman, The Principle of Least Action in Quantum Me-
chanics, Ph. D. thesis, Princeton University, 1942; reprinted in:
textquotedblleft Feynman’s Thesis – A New Approach to Quan-
tum Theory, L.M. Brown, Editor, (World Scientific Publishers,
Singapore, 2005). pp. 1.
17. R.P. Feynman, Rev. Mod. Phys. 20 (1948) 367; reprinted in:
textquotedblleft Feynman’s Thesis – A New Approach to Quan-
tum Theory, L.M. Brown, Editor, (World Scientific Publishers,
Singapore, 2005) pp. 71.
18. R.P. Feynman, Phys. Rev. 76 (1949) 749.
19. R.P. Feynman, Phys. Rev. 76 (1949) 769.
20. R.P. Feynman and A.R. Hibbs, Quantum Mechanics and Path
Integrals, (McGraw–Hill, New York, 1965).
21. S. Flügge, Practical Quantum Mechanics, (Springer–Verlag,
Berlin, 1999).
22. G. Gasper, J. Math. Anal. Appl. 42 (1973) 438.
23. K. Gottfried and T.-M. Yan, Quantum Mechanics: Funda-
mentals, second edition, (Springer–Verlag, Berlin, New York,
2003).
24. B.R. Holstein, Am. J. Phys. 65 (1997) 414.
25. B.R. Holstein, Am. J. Phys. 67 (1998)583.
26. J. Howland, Indiana Univ. Math. J. 28 (1979) 471.
27. D.R. Jafaev, Teoret. Mat. Fiz. 45 (1980) 224.
28. L.D. Landau and E.M. Lifshitz, Quantum Mechanics: Nonrel-
ativistic Theory, (Pergamon Press, Oxford, 1977).
29. V.P. Maslov and M.V. Fedoriuk, Semiclassical Approximation
in Quantum Mechanics, (Reidel, Dordrecht, Boston, 1981).
30. M. Meiler, R. Cordero–Soto, and S.K. Suslov, J. Math. Phys.
49 (2008) 072102.
Rev. Mex. Fı́s. 55 (2) (2009) 196–215
215
31. J. Meixner, J. London Math. Soc. 9 (1934) 6.
32. J. Meixner, Mathematische Zeitscrift 44 (1939) 531.
33. J. Meixner, Deutsche Math. 6 (1942) 341.
34. E. Merzbacher, Quantum Mechanics, third edition, (John Wiley
& Sons, New York, 1998).
35. A. Messia, Quantum Mechanics, two volumes, (Dover Publica-
tions, New York, 1999).
36. V. Naibo and A. Stefanov, Math. Ann. 334 (2006) 325.
37. P. Nardone, Am. J. Phys. 61 (1993) 232.
38. A.F. Nikiforov, S.K. Suslov, and V.B. Uvarov, Classical Or-
thogonal Polynomials of a Discrete Variable, (Springer–Verlag,
Berlin, New York, 1991).
39. A.F. Nikiforov and V.B. Uvarov, Special Functions of Mathe-
matical Physics, (Birkhäuser, Basel, Boston, 1988).
40. J.D. Paliouras and D.S. Meadows, Complex Variables for Sci-
entists and Engineers, second edition, (Macmillan Publishing
Company, New York and London, 1990).
41. E.D. Rainville, Special Functions (Chelsea Publishing Com-
pany, New York, 1960).
42. M. Rahman and S.K. Suslov, Singular analogue of the Fourier
transformation for the Askey–Wilson polynomials, in: Sym-
metries and Integrability of Difference Equations; D. Levi, L.
Vinet, and P. Winternitz, Amer. Math. Soc. 9 (1996) 289.
43. R.W. Robinett, Am. J. Phys. 64 (1996) 803.
44. I. Rodnianski and W. Schlag, Invent. Math. 155 (2004) 451.
45. L.I. Schiff, Quantum Mechanics , third edition, (McGraw-Hill,
New York, 1968).
46. W. Schlag, Dispersive estimates for Schrödinger operators: a
survay, arXiv: math/0501037v3 [math.AP] 10 Jan 2005.
47. E.M. Stein, Harmonic Analysis: Real-Variable Methods, Or-
thogonality, and Oscillatory Integrals , (Princeton University
Press, Princeton, New Jersey, 1993).
48. S.K. Suslov, Electronic Transactions on Numerical Analysis
(ETNA) 27 (2007) 140.
49. G. Szegő, Amer. Math. Soc. (Colloq. Publ., Vol. 23, Rhode Is-
land, 1939).
50. N.S. Thomber, and E.F. Taylor, Am. J. Phys. 66 (1998) 1022.
51. N. Ya. Vilenkin, Special Functions and the Theory of Group
Representations, (American Mathematical Society, Providence,
1968).
52. N. Wiener, The Fourier Integral and Certain of Its Applica-
tions, (Cambridge University Press, Cambridge, 1933); (Dover
edition published in 1948).
53. K. Yajima, J. Math. Soc. Japan 29 (1977) 729.