See

discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/230899020

Atom Optics Quantum Pendulum

Article in Journal of Russian Laser Research · May 2009

DOI: 10.1007/s10946-009-9078-x · Source: arXiv

CITATIONS READS

7 73

4 authors:

Muhammad Ayub Khalid Naseer

Pakistani Institute of Nuclear Science and Te… University of Sargodha
6 PUBLICATIONS 36 CITATIONS 11 PUBLICATIONS 42 CITATIONS

SEE PROFILE SEE PROFILE

Manzoor Ali Farhan Saif

Karakoram International University Quaid-i-Azam University
8 PUBLICATIONS 26 CITATIONS 106 PUBLICATIONS 731 CITATIONS

SEE PROFILE SEE PROFILE

All content following this page was uploaded by Muhammad Ayub on 01 February 2017.

The user has requested enhancement of the downloaded file. All in-text references underlined in blue are added to the original document
and are linked to publications on ResearchGate, letting you access and read them immediately.
Journal of Russian Laser Research, Volume 30, Number 3, 2009

ATOM OPTICS QUANTUM PENDULUM

Muhammad Ayub,1,2 Khalid Naseer,1,3 Manzoor Ali,1,4 and Farhan Saif1∗

1 Departmentof Electronics
Quaid-i-Azam University
Islamabad 45320, Pakistan
2 Pinstech, Nilor, Islamabad, Pakistan
3 Departmentof Physics
University of Sarghoda, 40100, Pakistan
4 Department
of Physics
Karakurum International University
Gilgit, Pakistan
*Corresponding author, e-mail: fsaif @ fulbrightweb.com
e-mail: ayubok @ yahoo.com

Abstract
We explain the dynamics of cold atoms, initially trapped and cooled in a magneto-optic trap, in a
monochromatic stationary standing electromagnetic wave field. In the large-detuning limit, the system
is modeled as a nonlinear quantum pendulum. We show that the wavepacket evolution of the quantum
particle probes parametric regimes in the quantum pendulum that support the classical period and
quantum-mechanical revival and superrevival phenomena. Interestingly, complete reconstruction, in
particular the parametric regime at quantum revival times, is independent of the potential height.

Keywords: quantum pendulum, optical lattices, cold atoms, Mathieu solutions, wavepacket revivals.

(Some figures in this article are in color only in the electronic version.)

1. Introduction
The pendulum in quantum mechanics [1–4] is a subject of great interest in explaining hindered
internal rotations in chemistry [5–8], quantum features of scattering atoms in quantum optics [9, 10],
the perturbation-theory methods to study weak field effects in quantum mechanics [11–13], dynamics
of Bose–Einstein condensates in optical lattices for small nonlinearities [14], and many other physical
systems. The comprehensive study of quantum pendulum has brought to light its various aspects, which
include the structure of energy spectrum, the time evolution focusing on the quantum revivals [15-31],
and asymptotic Mathieu solutions using algebraic methods [32].
Supercold atoms in optical standing fields is an area of both theoretical and experimental interest and
is modeled as a quantum pendulum [9]. The system is of great importance in both classical and quantum
domains since an explicit time dependence in phase or amplitude makes the classical counterpart chaotic.

Manuscript submitted by the authors in English first on March 19, 2009 and in final form on May 1, 2009.
1071-2836/09/3003-0205 c
2009 Springer Science+Business Media, Inc. 205
Journal of Russian Laser Research Volume 30, Number 3, 2009

The Bose–Einstein condensate in an optical lattice [33] is a milestone and an important paradigm to
study the atomic condensate evolution and its phase transition from the Mott insulator to the superfluid
state [34, 35–40]. Quantum revivals in the system present a profound manifestation of quantum interfer-
ence. The phenomenon occurs since the optical-lattice potential is not perfectly harmonic and the level
spacing varies with quantum number. The revivals provide useful information on the coherence time of
the atoms in the optical lattice.
In [29, 30], the energy spectrum and time scales encoded in the spectrum are discussed in detail. In
this paper, we extend the work following the same approach and explain the evolution of supercold weakly
condensed atoms in time-independent optical lattices. Furthermore, we explain the energy spectrum and
eigenstates of the system both analytically and numerically. The time dynamics of the wavepacket also
translates the relevance of the energy spectrum to the classical phase space. We probe the distribution of
energy eigenstates of the potential via the wavepacket evolution. For simplicity, we limit our discussion
to deep optical lattices where tunneling does not play a significant role. We show (i) an equally-spaced
local-energy spectrum deep in the well, which is reflected in the reconstruction of the wavepacket at
classical periods; (ii) beyond this domain, the behavior changes since the level spacing modifies itself,
and the nonlinear behavior dominates and controls the dynamics as we go further from the deep-in-
the-well conditions; (iii) numerical calculations with enhanced efficiency and accuracy in the presence
of analytical relationships explain the detailed dynamics of the quantum pendulum and its parametric
dependence on anharmonicity.
In Sec. 2, we discuss atomic interactions with the optical lattice and obtain the quantum pendulum.
In Sec. 3, we work out analytical solutions of the potential. In Sec. 4, a simplified quantum pendulum is
modeled as a series expansion of the cosine potential, and the effect of each term of the series that plays
an effective role in the atomic evolution in the quantum pendulum is explained. Results are discussed in
Sec. 5.

2. The Model
We consider supercold two-level atoms interacting with a classical monochromatic standing wave field

E(x, t) = êy εo cos(kL x)e−iωt + c.c. ,

 

where ω and kL are, respectively, the frequency and the wave number of the laser field and êy is the
polarization vector. We assume that the cavity end mirrors in the yz plane reflect the incoming light
wave along the x axis and, therefore, determine the position of the nodes of the standing light along
the axis. In the dipole and rotating-wave approximations, the atom–field interaction is controlled by the
Hamiltonian
p2x
 
*
iωt
H= + ~ωo |eihe| − d .êy εo cos(kL x) × e σ+ + h.c. , (1)
2M
*
where ~ω0 is the energy difference between the atomic excited state |ei and the ground state |gi and d
indicates the atomic dipole moment. Moreover, px is the center-of-mass momentum of the atom along the
cavity axis, M is the atomic mass, and σ± are the raising and lowering operators defined as σ+ = |eihg|
and σ− = |gihe|.
Furthermore, we have considered condensate densities so low that interparticle interactions are neg-
ligible and the single-particle approach remains applicable [41, 42]. Thus, we represent the wave function

206
Volume 30, Number 3, 2009 Journal of Russian Laser Research

of the system at any time of interaction t as

|ψ (x, t)i = ψg (x, t) |gi + ψe (x, t) e−iωt |ei.

In the presence of sufficiently large detuning between the atomic transition frequency and the field
frequency, i.e., δL = ω0 − ω, one can neglect the spontaneous emission. The same consideration allows
one to eliminate the excited state adiabatically and effectively describe the evolution of the atom within
the electromagnetic field in the ground state. Hence the dynamics of an atom is governed effectively by
the Hamiltonian
p2x ~Ωeff
H= − cos(2kL x),
2M 8
*
where Ωeff = Ω2 /δL is the effective Rabi frequency, with Ω = d .êy εo /~ being the Rabi frequency.
Therefore, the probability to find an atom in the excited state is negligible as a consequence of large
detuning, and the evolution properties of the atom in the field are completely determined by the ground-
state amplitude.
It is useful to define the following dimensionless quantities

2kL H(4kL2 )
t = ωt, x̄ = 2kL x, p= px , H̄ = .
Mω M ω2

We obtain the dimensionless effective Hamiltonian as follows:

p2
H̄ = − V0 cos x̄, (2)
2

where the parameters V0 = Ωeff /ω 2 is the effective potential depth and  = ~kL2 /2M is the recoil shift.
In the case δL > 0 and when the laser is tuned red to the atomic transition, we find V0 is positive.
Hence, a phase difference between the potential and the laser intensity shifts the location of the potential
minima such that they coincide with the locations of the laser intensity maxima. The atom is, therefore,
attracted towards the intensity maxima. In the other case, that is, δL < 0, the phase difference between
the potential and the laser intensity disappears, and the atom is attracted towards the intensity minima.
The quantized system has another controlling parameter, i.e., the scaled Planck’s constant k − = 8/ω,

which follows from the commutation relation i[p, x̄] = k − , where H̄ effectively defines the dynamics of a

quantum particle as a quantum pendulum.

The eigenstates of a quantum rotor have the spatial periodicity condition φn (x̄) = φn (x̄ + 2π), while
the eigenstates for an atom in the optical lattice obeys the general Bloch condition φn (x̄) = ei2πν φn (x̄),
where ν indicates the quasimomentum of the eigenstate. The general Bloch condition satisfies the spatial
boundary condition for a quantum rotor when ν has integer values. In contrast, the range of ν is
continuous for optical lattices. In both cases, the potential only couples the eigenstates on a momentum
ladder where the ladder spacing is 2~kL . However, for optical lattices, there are many momentum ladders
independent from each other, while for a quantum rotor there is only one centered at p = 0 [43].

207
Journal of Russian Laser Research Volume 30, Number 3, 2009

3. Schrödinger Equation for a Quantum Pendulum and the Mathieu

Equation
The dynamics of the atom interacting with a standing wave field in the large detuning limit is effec-
tively controlled by the time-independent Schrödinger equation
−2 ∂ 2 ψ(x̄)
k
− − V0 cos(x)ψ(x̄) = En ψ(x̄),
2 ∂x2
where, for simplicity, we write ψg = ψ and x̄ → x. We rewrite the position variable x as x ≡ x − π and
obtain the Mathieu equation
∂ 2 ψn (x)
+ [an − 2q cos(x)]ψn (x) = 0,
∂x2
which states that for any given an and q we have a series of solutions ψn (x) for the differential equation
labeled by index n. Hence, ψn (x) defines the eigenfunction of the system and
V0 ~Ωeff 2En
q≡ −2
= and an ≡ −2
(3)
k 64ER k
express, respectively, the effective potential depth and the Mathieu characteristic parameter leading to
the eigenenergies.
An atom in an optical lattice may observe deep or shallow potential depths corresponding to its
energy. Following Eq. (3), we scale the effective potential depth by recoil energy of an atom ER = ~.
The atom in the cosine potential observes a deep optical potential if the effective potential depth V0 is
of the order of a few hundred single photon recoil energies and the temperature of the atom is close
to the recoil temperature. In this case, the dynamics of the quantum particle in the individual well is
independent and one obtains multiple realization of anharmonic oscillators. On the other hand, when
the depth of the potential is just a few recoil energies and the atom is at about recoil temperatures, it
sees a shallow potential. In this case, the quantum-mechanical effects caused by the spatial periodicity
of optical lattices, such as formation of Bloch waves, become important. Furthermore, the levels are
broadened into bands due to resonant tunneling between adjacent wells [44]. Tunneling in the low-lying
bands is suppressed as the well depth increases and the particle motion is dominated by the single-well
dynamics, as discussed for large V0 .
Moderate values of V0 , with an effective Planck’s constant of order unity, indicates the deep quantum
regime. The semiclassical dynamics of the atom in the standing-wave field is observed, however, for large
values of V0 , which correspond to small values of the effective Planck’s constant k − , and here we find

several tightly bound energy bands.

Quantum-mechanical effects for small V0 become important once the atomic de Broglie wavelength
1 2π λL
2π~/P significantly exceeds the lattice constant d = = . This gives the condition
2 kL 2

P2 4~2 kL2
 ≡ 4ER .
2M 2M
For a fixed value of q, there are a countably infinite number of solutions labeled by n. However, only
for specific characteristic values of the parameter an are the solutions periodic, with periods π or 2π in the
variable x, which are denoted by an and bn , respectively, for the even and odd solutions. Because of the

208
Volume 30, Number 3, 2009 Journal of Russian Laser Research

intrinsic parity of the potential, the solutions can be characterized as being even, cen (x, q) or cosine-like
for integral values of n, with n ≥ 0, whereas they are odd, sen (x, q) or sine-like for integral values of n,
with n ≥ 1. Limiting cases, i.e., q = 0, q  1 and q  1, for a quantum pendulum are discussed in detail
in [29].
Approximate expressions for the characteristic values of an and bn both in q  1 and in q  1 limits
are provided by [45, 46]. For the limit q  1, we find that the an and bn are approximately degenerate
for n & 7, that is,
q2 (5n2 + 7)q 4
an ' bn = n2 + + + ··· (4)
2(n2 − 1) 32(n2 − 1)3 (n2 − 4)

The above expression is not limited to integral value of n and is a very good approximation when n is of
the form m + 1/2. In the case of integral value n = m, the series holds only up to the terms not involving
n2 − m2 in the denominator. The difference between the characteristic values for even and odd solutions
satisfies
 
bn
an − bn = O at n → ∞.
nn−1

In the other limiting case, when q  1 and the spectrum is the oscillator-like, we obtain

√ s2 + 1 s3 + 3s
an ≈ bn+1 ≈ −2q + 2s q − − 7√ − · · · , (5)
23 2 q

where s = 2n + 1. Thus, it has −

√ (n + 1/2) k ωh dependence in lower order, which resembles the harmonic-
oscillator energy for ωh = 2 V0 .
In the deep-optical-lattice limit, the band width is defined as [45]
p √
24n+5 2/πq (n/2)+(3/4) exp(−4 q)
bn+1 − an ' . (6)
n!

Equation (6) shows that in the deep optical lattice (q  1 limit), the energy bands are realized as
degenerate energy levels since the band width is negligible.
The band structure of the optical lattice is shown in Fig. 1 where for large q, near the bottom of the
lattice, thin bands are seen, the band width increases, and the band gap decreases as we move towards
the top of the lattice potential well.
As a consequence, we suppress atomic tunneling in the deep-optical-lattice limit. The hopping matrix
element J explains the tunneling between adjacent sites for deep optical lattices [36, 38, 40], viz.,

4 p
J = √ ER (Vo /ER )3/4 exp(−2 Vo /ER ). (7)
π

Equations (6) and (7) show that the width of the bands corresponds to tunneling of the atom from
one lattice site to the other. In the limit of deep lattice potentials, this probability is exponentially small,
and the band width, therefore, decreases exponentially as a function of the lattice-potential depth.

209
Journal of Russian Laser Research Volume 30, Number 3, 2009

Fig. 1. Characteristic values for the Mathieu equation a2m (for even solutions, blue (grey) curves) and b2m (for
odd solutions, red (dark) curves, mostly overridden by blue curves) versus q for the quantum pendulum. The
dotted lines correspond to the characteristic parameter an = ±2q, i.e., E = ±V0 . Note that for a given value of
q where q  an , the gap between the lowest energy state (lowest solid curve) is roughly one-half of the spacing
between solid and dashed curves corresponding to the zero-point energy in the oscillator limit.

After some algebra, we get asymptotic even Mathieu functions [32] as follows:
(n4 − 6n3 + 11n2 − 6n)Dn−4 + (4n − 4n2 )Dn−2 − 4Dn+2 − Dn+4
cen (x, q) = √
64 q

1 1 8
+ (n − 28n7 + 322n6 − 1960n5 + 6769n4 − 13132n3 + 13068n2 − 5040n)Dn−8
1024q 8
−(n6 − 15n5 + 85n4 − 225n3 + 274n2 − 120n)Dn−6 + 4(n5 − 7n4 + 17n3 − 17n2 + 6r)Dn−4

1
−(n4 + 26n3 − 37n2 + 10n)Dn−2 + (−36 − 25n + n2 )Dn+2 − 4(n + 2)Dn+4 + Dn+6 + Dn+8
8
 
1
+Dn + O 3/2 (8)
q
in terms of the Hermite polynomials
α2
   
1 α
Dn (α) = exp − Hn √
2n/2 4 2
and α is defined as α = 2q 1/4 cos(x). The normalization factor for cen (x, q) up to order O(1/q 2 ) is
√
2n + 1 n4 + 2n3 + 263n2 + 262n + 108

1 2n!
= √ 1 + √ +
Cn2 πq 1/4 8( q) 2048q
5 4 3 2
 
6n + 15n + 1280n + 1905n + 1778n + 572 1
+ 3/2
+O 2 , (9)
16384q q

210
Volume 30, Number 3, 2009 Journal of Russian Laser Research

where eigenstates are normalized to π.

4. Bound States of Optical Lattices

Around the minima of lattice sites, the harmonic evolution prevails, and in the presence of higher-order
terms it is gradually modified to the original potential. Microscopic investigations of the atom–optical
field system, using term by term the contribution of the cosine-potential expansion, reveal the dominant
role of the system’s particular parametric regime in the formation of eigenstates and eigenenergies. This
leads to simplified analytical solutions around the potential minima in the system, as discussed below.

4.1. Harmonic Oscillator-Like Limit

In the deep-lattice limit, the potential near the minima can be approximated as quadratic. Thus the
particle situated near the minima of the cosine potential experiences a harmonic potential. The energy
(0) √
− V − V , which can be identified in Eq. (5) by ignoring
in this regime is obtained as En = (2n + 1)k 0 0
the square and higher powers in s. The eigenstates of the quadratic potential are given as
s
−β 2 x2
 
β
φn (x) = √ Hn (βx) exp ,
2n n! π 2

where Hn (βx) are the Hermite polynomials and

 √ 1/2
2 V0 √ 1/4
β= −
= 2q .
k

The time evolution of the particle, which was initially in the state ψ(x, 0), is obtained by the time-
evolution operator Û such that
∞  
X En
ψ(x, t) = Û ψ(x, 0) = cn φn (x) exp −i − t ,
k
n=0

where En and φn (x) are the energy eigenvalues and eigenstates corresponding to quantum number n.
The probability amplitudes cn are defined as hφn (x)|ψ(x, 0)i. The quantum-particle’s wave packet in
the optical potential, narrowly peaked around a mean quantum number n̄, displays quantum recurrences
at different time scales defined as
2π
T(j) = ,
(j!k ) Enj |n=n̄
− −1

where Enj denotes the j th derivative of En with respect to n. The time scale T(1) is called the classical
period since it provides the time at which the particle completes its evolution following the classical
trajectory and reshapes itself, whereas at T(2) the particle reshapes itself as a consequence of quantum
interference in the nonlinear energy spectrum, which is purely a quantum phenomenon and thus named
the quantum revival time. In the parametric regime of a changing nonlinearity with respect to quantum
number n, we find the superrevival time T(3) for the quantum particle [48, 49].

211
Journal of Russian Laser Research Volume 30, Number 3, 2009

1
|A(t)|2

0.96

0.92

0 2 4 t 6 8 10
Fig. 2. Time evolution of the particle’s wavepacket placed at the bottom of the cosine potential. Dimensions
of the wavepacket are −k = 0.5 and ∆p = 0.5, with V0 = 10. We show autocorrelation function versus time (on
the right) and spatiotemporal behavior of the material’s wavepacket (on the left). The wavepacket shows equally
spaced energy levels and rebuilds after every classical period. Analytically calculated value of the classical period
and numerical results demonstrate an excellent agreement.

We study the time evolution of the material’s wavepacket, in view of the square of the autocorrelation
function [50]
∞ ∞  
2
X
4
X
2 2 t
|A(t)| = |cn | + 2 |cn | |cm | cos (En − Em ) − .
k
n=0 n6=m

In the parametric regime under discussion, the square of the autocorrelation function is written as
∞
X ∞
X h p i
|A(t)|2 = |cn |4 + 2 |cn |2 |cm |2 cos (n − m)2 V0 t , (10)
n=0 n6=m

∞
|cn |4 is independent of time and defines the interference-free, averaged value of |A(t)|2 .
P
where
n=0
The most dominant contribution to |A(t)|2 comes from n−m = 1 in the second part on the right-hand
side of Eq. (10). Other terms with m − n ≥ 1 play a √ negligible role because their oscillation frequency is
an integral multiple of the fundamental frequency 2 V0 and are averaged out to zero. For this reason,
the square of the autocorrelation function in the present case oscillates following a cosine law with the
√ (0) √
frequency 2 V0 , which leads to the classical time period Tcl = π/ V0 , as shown in Fig. 2. Here, zero in
(0)
the superscript of Tcl defines the system’s classical period in the absence of perturbation. At integral
multiples of the classical period |A(t)|2 is unity, whereas at time that is an odd integral multiple of half
of the classical period,
∞
X ∞
X
2 4
|A(t)| = |cn | + 2 |cn |2 |cm |2 cos[(n − m)π].
n=0 n6=m

Here, cos[(n − m)π] is alternately +1 and −1 when n − m is even or odd, respectively. Thus, after the
cancellation of positive terms with the negative ones, the second summation reduces to a minimum value

212
Volume 30, Number 3, 2009 Journal of Russian Laser Research

and |A(t)|2 attains its minima. Here, in the eigenstate expansion

  p ∞
 X
V0  p 
ψ(x, t) = exp −i 2 V0 − − t cn φn (x) exp −i2n V0 t , (11)
k
n=0

it is notable that the eigenstates φn (x) have the parity (−1)n .

In the case of even parity, only the even terms c2n are nonvanishing, and the n-dependent exponent
factor oscillates two times faster than in the general case. At half of the classical period, the wavepacket
reappears towards other turning points of the well.
In the case where the wave packet is initially placed at the center of the cosine well, it reappears
at half of the classical period at the same position but in the opposite direction. In this case, we see
classical revivals of the initial atomic wavepacket, and quantum revivals take place at an infinitely long
time. Hence, we find a revival of the atomic wavepacket after each classical period only. Experimentally
we may realize the situation by placing very few recoil energy atoms deep in the cosine potential well.
This reveals information on the level spacing around the bottom of the cosine potential. Interestingly,
we find an equal spacing between the energy levels in Fig. 1 for large q and small n. The spatiotemporal
behavior of the wavepacket in the quadratic potential, as shown in Fig. 2, confirms the above statements.

4.2. Quartic Oscillator Limit

Beyond the harmonic-oscillator limit,
we find the oscillator with nonlinearity
and the energy-level spacing different from
a constant value. The correction to the
energy of the harmonic oscillator comes
from the first-order perturbation (see Ap-
pendix 1 for energy corrections), that is,
−2
k
En(4) = − (2n2 + 2n + 1),
8
which again can be identified in Eq. (5) by
ignoring cubic and higher-order powers in
n.
The atoms with slightly higher energy,
which is equivalent to several recoil ener-
gies, see in another time scale where it re-
constructs itself beyond the classical pe-
riod, i.e., quantum revival time. The be-
havior of the autocorrelation function for Fig. 3. Autocorrelation function for a particle undergoing quan-
the wavepacket placed exactly in this re- tum revival evolution in time. The parameters are the same as
gion, where only first-order correction is in Fig. 2. The wavepacket was placed close to the bottom of the
sufficient, is shown in Fig. 3. We see that potential well in the regime where the first-order correction is suf-
ficient; it observes quantum revivals after many classical periods.
the wavepacket displays revivals at quan-
tum revival time. Thus, a little above the

213
Journal of Russian Laser Research Volume 30, Number 3, 2009

bottom of an optical lattice, the wavepacket sees variations in the energy-level spacing in a nonlinear way.
(1) (0) √
The classical time is modified as Tcl = α(1) Tcl , where α(1) = 1 + s̄/8 q, and the classical periodicity
for a particle in the present situation is related to the potential height and mean quantum number, here
√
s̄ = 2n̄ + 1. As n̄ increases, the classical revival time also increases, and the ratio s̄/8 q is always less
(1) −
than unity in the region where the first correction is sufficient. The quantum revival time Trev = 8π/k
is independent of the mean quantum number n̄, whereas the superrevival time, in this case, is infinite.
The eigenstates of a quartic oscillator due to the first-order perturbation are given as φqn (x) = φn (x)+
(1a) (1a)
φn (x), where φn (x) is the first-order correction to the harmonic-oscillator wave function and is defined
as
φ(1a)
n = D1 (η1 φn−4 + η2 φn−2 − η3 φn+2 − η4 φn+4 ) , (12)
where D1 , η1 , η2 , η3 , and η4 are constants and defined in Appendix 1.

Fig. 4. Comparison of the cosine potential with simplified potentials is made by calculating the projection S of the
eigenstates of the cosine potential on the eigenstates of the simplified potentials. For first few quantum numbers,
the cosine potential very much resembles to the harmonic potential; however, for higher quantum numbers, the
higher-order corrections to the harmonic oscillator are needed to make the resemblance.

In Fig. 4, the eigenstates of the quadratic, quartic, sixtic, and octic oscillators are mapped on nu-
merically obtained eigenstates of the cosine potential. From Fig. 4, we see that for V0 = 10 and k− = 0.5,

the first-order correction to the eigenstates of an unperturbed system matches the harmonic oscillator’s
eigenstates up to n = 3, and mapping of the quartic oscillator with the exact solution is much improved
compared with the harmonic oscillator. Similarly, mapping of the sixtic oscillator is better than that of
the quartic oscillator and is quite good for the octic oscillator for all bound bands. In this case, eight
bands exist inside the potential. From Fig. 1, we note that with increase in q the number of bound bands
can be increased.
The square of the autocorrelation function in this case takes the form
∞ ∞  −

2
X
4
X
2 2
p k
|A(t)| = |cn | + 2 |cn | |cm | cos (n − m)2 V0 t + (m − n)(n + m + 1) t , (13)
4
n=0 n6=m

where the nonlinear dependence of the energy eigenvalues on quantum number n makes the argument of
the cosine function nonlinear, as appears in the last term of the above expression (13). The nonlinear

214
Volume 30, Number 3, 2009 Journal of Russian Laser Research

term (m − n)(n + m + 1) removes the degeneracy present for the harmonic case between the cosine waves
corresponding to the nearest neighboring off diagonal terms and beyond. Hence, the overall evolution
displays a gradual decoherence leading to collapse, which later transforms in revival as the decoherence
− simplify as
in waves disappears. The values of |A(t)|2 in this regime at Trev = 8π/k
∞ ∞
2
X
4
X √
|A(t)| = |cn | + 2 |cn |2 |cm |2 cos [16π q(n − m)] . (14)
n=0 n6=m

√
Equation (14) shows that |A(t)|2 = 1 when q is an integral multiple of 1/8. Also, |A(t)|2 is unity at
√ √
half of the revival time if q is an integral multiple of 1/4. In the case where the q is not an integral
multiple of 1/8, the wavepacket revival occurs a little earlier than the revival time Trev = 8π/k − . Also,

|A(t)|2 approaches unity a little earlier than Trev /2. Here reproduction of the wavepacket at Trev /2 is out
of phase by π, i.e., at that time all the waves are moving exactly in opposite directions as initially. But
at Trev , they all are moving in the same direction, and each wave is in phase not only with their initial
states but also with each other.

Fig. 5. The time evolution of a localized wavepacket in a cosine potential is displayed for the same parameters as in
Fig. 2. For short times, the initial wavepacket shows classical revivals (left), but starts to display subwavepackets in
its long time dynamics (middle), which constructively interfere at quantum revival time Trev (right). Our analytical
and numerical results are in very good agreement.

Figure 5 shows the spatiotemporal behavior of the wavepacket in the cosine potential. We see that the
wavepacket spreads and oscillates in the cosine well, and after some time the original wavepacket is divided
into subwavepackets (Fig. 5, middle). At a quantum revival time, these subwavepackets constructively
interfere and the wavepacket assumes the original shape at the same initial position (Fig. 5, right). At
quantum revival time, the same classical pattern is seen as shown at the start of the time evolution
(Fig. 5, left).

4.3. Sixtic Oscillator Limit: Existence of Superrevivals and Beyond

Higher-order nonlinearities in the energy spectrum of the quantum pendulum show up beyond the
quartic limit. In the presence of the second correction, the energy is modified by the term
−3 (2n3 + 3n2 + 3n + 1)
k
En(6) = − √ .
V0 32

215
Journal of Russian Laser Research Volume 30, Number 3, 2009

(2)
The second correction to the energy modifies the time scales. The classical time period is now Tcl =
(0) (2) (1)
α(2) Tcl , whereas the quantum revival time is modified as Trev = |β (1) |Trev , where α(2) = α(1) +
2 8 (1)
3(s̄ + 1)/2 q and β = 3s̄/16q − 1 are constants. In addition, the system shows another time scale,
(2) √ −2 at which reconstruction of the original wavepacket takes place.
i.e., superrevival time Tspr = 64π V0 /k
The other quantum revival times occur at infinity.
The energy eigenstates in this regime are given as

φ(s) (1,a)
n = φn + φn + φ(1,b)
n + φ(2,a)
n ,
(2,a) (1,b)
where φn is perturbation in eigenstates due to the H (6) term, and φn is the second-order perturbation
(1,b) (2,a)
caused by the H (4) term [29]. The expressions φn and φn are given as

φ(1,b)
n = D2 [δ1 φn−8 + δ2 φn−6 + δ3 φn−4 + δ4 φn−2 + δ5 φn+2 + δ6 φn+4 + δ7 φn+6 + δ8 φn+8 ] ,
φ(2,a)
n = D6 [χ1 φn−6 + χ2 φn−4 + χ3 φn−2 + χ4 φn+2 + χ5 φn+4 + χ6 φn+6 ] .

The constants D2 , D6 , δj ’s, and χj ’s are calculated in Appendix 1 where j takes integral values.

Again, in this region, the classical time

increases with increase in n̄ and increases
faster than as it was in the quartic limit.
However, in the present regime, the quan-
tum revival time is not constant and de-
creases with increase in n̄. The n̄ depen-
dence of the quantum revival time is shown
in Fig. 6. The superrevival time is inde-
pendent of the mean quantum number. It
is directly proportional to the square root
of the potential height and inversely pro-
portional to the square of scaled Planck’s
constant. The temporal behavior of an
atom in the optical potential placed in this
regime shows three time scales — classi- Fig. 6. Quantum revival time vs mean quantum number is shown
cal periods enveloped in quantum revivals for simplified potentials. In the presence of only first-order cor-
and quantum revivals enveloped in super- rection (1), i.e., for the quadratic potential, the quantum revival
revivals (see Fig. 7). After each superre- time is constant. For higher-order corrections, i.e., the quartic (2),
sixtic (3), and octic (4) potentials, it decreases with increase in
vival time, the atomic-wavepacket evolu-
mean quantum number.
tion repeats itself.
Similarly, the third correction in energy modifies the energy by the factor
−4 (5n4 + 10n3 + 16n2 + 11n + 3)
k
En(8) = − .
V0 28
(3) (0) (3) (1)
The time scales in this case are Tcl = α(3) Tcl and Trev = |β (2) |Trev , where

(5s̄3 + 17s̄) 15s̄2 + 17

α(3) = α(2) + , β (2) = β (1) + ,
211 q 3/2 28 q

216
Volume 30, Number 3, 2009 Journal of Russian Laser Research

(3) (2) √
and the superrevival time is Tspr = |γ|Tspr , where γ = 5s̄/8 q −1. Furthermore, the super-quartic-revival
time T4 is independent of n̄. The higher-order corrections in energy show that other time scales do exist
in the system, but their times of recurrence are too large to consider them finite.
In the presence of the third-order correction to energy, the classical time increases with increase in
n̄ but increases a little faster than in the cases of quartic and sixtic corrections, whereas the quantum
revival time decreases faster with increase in n̄ compared with the case of the sixtic correction. The
superrevival time is not constant but decreases with increase in n̄.
Energy corrections increase anhar-
monicity in the system. We have shown
that the first-order correction to the
harmonic-potential energy led to the quan-
tum revivals, the second-order energy
correction led to superrevivals, and the
fourth-order correction led to the quar-
tic revival time. A comparison of revival
times for different energy corrections is
shown in Fig. 6. For the first-order en-
ergy correction, the quantum revival time
is constant, but for higher-order correc-
tions, the quantum revival time may de-
crease with increase in the mean quantum
number of the wavepacket.
In Fig. 7, we show the projection of
numerically calculated eigenstates of the
cosine potential on the eigenstates of the
quadratic, quartic (first correction to the Fig. 7. The wavepacket dynamics in the sixtic potential displays
three time scales — the classical periods (making the dense region),
quadratic potential), sixtic (second correc-
the quantum revival times (making the peaks in the dense region),
tion to the quadratic potential), and octic and the superrevival times (making the peaks in the envelop of the
(third correction to the quadratic poten- quantum revivals). The parameters are the same as in Fig. 2.
tial) potentials. We note that the eigen-
states of all the above-mentioned potentials match the eigenstates of the quadratic potential near the
bottom of the lattice potential. For slightly larger quantum numbers, we see that the projection of the
quadratic potential falls sharply and improves with higher-order potentials. This correction is quite good
for the octic potential when q = 40, justifying the q  1 condition. The eigenstates and eigenenergies of
the quartic, sixtic, and octic potentials are analytically calculated using the method given in Appendix 1.

5. Discussion
In this paper, we have extended the understanding of eigenenergy levels and eigenstates in deep
optical lattices, both analytically and numerically. We note that the solutions obtained through the
perturbation theory and the Mathieu solutions show similar results and are in very good agreement with
exact numerical solutions. The energy levels are equally spaced near the bottom and, by increasing the
lattice-potential depth, the number of equally spaced energy levels can be increased. A wavepacket placed
in this region revives after each classical period. From Fig. 4, it is also noted that all potentials discussed

217
Journal of Russian Laser Research Volume 30, Number 3, 2009

have equally spaced eigenlevels at the bottom of the potential well as their mapping with the harmonic
oscillator is unity. Beyond the linear regime, the energy dependence is quadratic and any wavepacket
evolved in this region shows complete quantum revivals enveloped by classical revivals. Interestingly,
(1) − ) is independent of the potential height but is
in this regime, the quantum revival time (Trev = 8π/k
inversely proportional to the effective Planck’s constant k − . We show that, for deep optical lattices, there

is a region where revivals are independent of the lattice depth, and this region expands with increase
in the lattice depth; however, beyond this region, the quantum revival time is no longer constant but
(2) √ −2 ,
decreases with increase in n̄. The higher-order time scale, the superrevival time, Tspr = 64π V0 /k
exists in this region and is independent of n̄. Again this region expands with increase in the potential
height, but the superrevival time in this region is directly proportional to the square root of the potential
height.
Above this region, other time scales also exist where the superrevival time is n̄ dependent and decreases
with increase in n̄, but these time scales are too long to consider them finite.

Acknowledgments
M.A. and K.N. thank the Higher Education Commission of Pakistan for financial support under
Grant No. 17-1-1 (Q.A.U) HEC/Sch/2004/5681. F.S. is supported by the Higher Education Commis-
sion of Pakistan under Research Grant 20-23 R & D/03143, the Abdus Salam International Center for
Theoretical Physics (Trieste, Italy), and the Pakistan Science Foundation. F.S. thanks S. Stanislav and
G. Ghirardi for fruitful discussions. The authors thank S. Iqbal, Rameez-ul-Islam, I. Rehman, and T.
Abbas for useful suggestions.

Appendix: Solution for Arbitrary Potential

An arbitrary potential U (r) in the vicinity of its minima can be solved by taking its Taylor’s expansion
[51], that is,
.
U (r) = U (r ) + G(1) (r − r ) + G(2) (r − r )2 + G(3) (r − r )3 + ..,
m m m m (15)
∂ j U (r
= rm )
where G(j) = (j!)−1 and j is an integer. The value G(j) for odd j is zero as the potential is
∂rj
cos(x) and it is calculated at the potential minima r = rm . Thus, in the presence of weak nonlinearity,
G(6)  G(4)  G(2) .
In our analysis, we consider expansion up to the second-order term to the unperturbed Hamiltonian
H0 . The eigenfunctions and eigenenergies of this Hamiltonian are those of the harmonic oscillator. The
effect of the higher-order terms in Taylor’s expansion is discussed as perturbation to the eigenenergies
and eigenfunctions of the harmonic oscillator. We express the effective Hamiltonian governing the atom
dynamics around the potential minima as
p̂2
H0 ∼= + U (rm ) + G(2) (r − rm )2 . (16)
2
The eigenfunctions and eigenenergies of the harmonic potential are
s
−β 2 x2
 
β −
p
φn (x) = n
√ Hn (βx) exp , En(0) = k V0 (2n + 1) + U (rm ),
2 n! π 2

218
Volume 30, Number 3, 2009 Journal of Russian Laser Research

where Hn (βx) are the Hermite polynomials.

The first- and second-order corrections to energy are quite well known.
The first-order correction to the energy of the quadratic potential is

En(p,a) = hφn |H (p) |φn i,

and the second-order correction is obtained from the relation

X |hφn |H (p) |φm i|2
En(p,b) = (0) (0)
.
m6=n En − Em

The third-order correction is determined by

X X hn|Ĥ (p) |jihj|Ĥ (p) |lihl|Ĥ (p) |ni X hn|Ĥ (p) |jihj|Ĥ (p) |ni
En(p,c) = (0) (0) (0) (0)
− hn|Ĥ (p) |ni (0) (0)
,
j6=n l6=n (En − Ej )(En − El ) j6=n ((En )2 − (Ej )2 )

where p = 3, 4, 5, 6, . . .
The presence of the perturbation term H (4) leads to the Hamiltonian

H = H0 + H (4) , (17)

where the first-order correction to the eigenfunctions, due to the H (4) term, is obtained as

|φ(1a)
n i = D1 (η1 φn−4 + η2 φn−2 − η3 φn+2 − η4 φn+4 ), (18)

where  2
1 1
D1 = G(4) √ −ω
4 q k h

and
p p
η1 = n(n − 1)(n − 2)(n − 3)/4, η2 = (2n − 1) n(n − 1),
p p
η3 = (2n + 3) (n + 1)(n + 2), η4 = (n + 1)(n + 2)(n + 3)(n + 4)/4.

Now the eigenfunctions in the presence of first-order perturbation, due to the correction H (4) , are
given as
φqn (x) = φn (x) + φ(1a)
n (x),

and the second-order correction due to the H (4) term is

φ(1,b)
n = D2 [δ1 φn−8 + δ2 φn−6 + δ3 φn−4 + δ4 φn−2 + δ5 φn+2 + δ6 φn+4 + δ7 φn+6 + δ8 φn+8 , (19)

where
  2  1 4  1 2
D2 = G(4) √ −ω
4 q k h

219
Journal of Russian Laser Research Volume 30, Number 3, 2009

and
p
n (n − 1) (n − 2) (n − 3) (n − 4) (n − 5) (n − 6) (n − 7)
δ1 = ,
32
(6n − 11) p
δ2 = n (n − 1) (n − 2) (n − 3) (n − 4) (n − 5),
12

p
δ3 = 2n2 − 9n + 7 n (n − 1) (n − 2) (n − 3) ,
3 2


56n − 228n + 214n − 146 p
δ4 = n (n − 1) ,
8
56n3 + 396n2 + 838n + 645 p


δ5 = (n + 1) (n + 2) ,
8
(31n2 + 197n + 258) p
δ6 = (n + 1) (n + 2) (n + 3) (n + 4) ,
16
(11n + 27) p
δ7 = (n + 1) (n + 2) (n + 3) (n + 4) (n + 5) (n + 6) ,
p 24
(n + 1) (n + 2) (n + 3) (n + 4) (n + 5) (n + 6) (n + 7) (n + 8)
δ8 = .
32

Hence, following the same procedure, the first-order correction due to the H (6) term changes the
Hamiltonian of the system as
H = H0 + H (4) + H (6) (20)

and the corrected eigenfunction in the presence of the correction due to the H (4) and H (6) terms appear
as
φ(s)
n = φn + φn
(1,a)
+ φ(1,b)
n + φ(2,a)
n ,

where
φ(2,a)
n = D6 [χ1 φn−6 + χ2 φn−4 + χ3 φn−2 + χ4 φn+2 + χ5 φn+4 + χ6 φn+6

and
 3
(6) 1 1
D6 = G √ −ω
,
4 q k h

with
p
χ1 = 6 n(n − 1)(n − 2)(n − 3)(n − 4)(n − 5),
3 p
χ2 = (2n − 3) n(n − 1)(n − 2)(n − 3),
4
15 2 p
χ3 = (n − n + 1) n(n − 1),
2
15 2 p
χ4 = (n + 3n + 3) (n + 1)(n + 2),
2
3 p
χ5 = (2n + 5) (n + 1) (n + 2) (n + 3) (n + 4),
4p
χ6 = 2 (n + 1) (n + 2) (n + 3) (n + 4) (n + 5) (n + 6).

220
Volume 30, Number 3, 2009 Journal of Russian Laser Research

Close to the potential minima, we can find the eigenenergies with considerable accuracy using the per-
turbation theory. The leading correction comes from H (4) using the first- and second-order perturbation
theory, respectively. The result is
−
En(4) = (α2 n2 + α1 n + α0 )k ωh ,

where
α0 = 3Da , α1 = 6Da , α2 = 6Da ,
and
  2  1 3 1 
1
2
(3) (4)
Cb= G √ −ω
, Da = G √ .
4 q k h 4 q
In the next step, the first-order perturbation of H (6) and the second-order perturbation of H (4)
contribute. The result can be written as
−
En(6) = (β3 n3 + β2 n2 + β1 n + β0 )k ωh ,

where
β0 = 3Ia − 21Jb , β1 = 8Ia − 59Jb , β2 = 6Ia − 51Jb , β3 = 4Ia − 34Jb ,
and  3  2  2
(6) 1 1 (4) 2 1 1
Ia = 5G √ −
, Jb = 2(G ) √ −
.
4 q k ωh 4 q k ωh
At the next higher order, we need to evaluate three contributions, namely, H (8) in the first order,
H (6) and H (4) in the second order, and H (4) in the third order. The energy expression is
−
En(8) = (γ4 n4 + γ3 n3 + γ2 n2 + γ1 n + γ0 )k ωh ,

where

γ0 = 3X − 12Y − 111Z, γ1 = 8X − 35Y − 347Z, γ2 = 10X − 46Y − 472Z,

γ3 = 4X − 22Y − 250Z, γ4 = 2X − 11Y − 125Z,

and
 4  5  2  6  2
(8) 1 1 (6) (4) 1 1 (4) 3 1 1
X = 35G √ −
, Y = 30G G √ −
, Z = 48(G ) √ −
.
4 q k ωh 4 q k ωh 4 q k ωh

Now the energy of the system is

En = En(0) + En(4) + En(6) + En(8) ,

or
−
E n = κ 4 n 4 + κ3 n 3 + κ2 n 2 + κ 1 n + κ 0 k ωh + U (rm ), (21)
where

κ0 = α0 + β0 + γ0 + 1/2, κ1 = α1 + β1 + γ1 + 1, κ2 = α2 + β2 + γ2 , κ3 = β3 + γ3 , κ4 = γ4 .

221
Journal of Russian Laser Research Volume 30, Number 3, 2009

References

1. E. U. Condon, Phys. Rev., 31, 891 (1928).

2. T. Pradhan and A. V. Khare, Am. J. Phys., 41, 59 (1973).
3. R. Aldrovandi and P. L. Ferreira, Am. J. Phys., 48, 660 (1980).
4. G. P. Cook and C. S. Zaidin, Am. J. Phys., 54, 259 (1986).
5. D. G. Lister, J. N. MacDonald, and N. L. Owen, Internal Rotation and Inversion, Academic Press,
London (1978).
6. W. J. Orville-Thomas (ed.), Internal Rotation in Molecules, Wiley, New York (1974).
7. G. Ercolani, J. Chem. Ed., 77, 1495 (2000).
8. G. L. Baker, J. A. Blackburn, and H. J. T. Smithm, Am. J. Phys., 70, 525 (2002).
9. S. Dyrting and J. G. Milburn, Phys. Rev. A, 47, R2484 (1993).
10. H. Müller, S. Chiow, and S. Chu, Phys. Rev. A, 77, 023609 (2008).
11. M. Schwartz and M. Martin, Am. J. Phys., 26, 639 (1958).
12. G. L. Johnston and G. Sposito, Am. J. Phys., 44, 723 (1976).
13. D. Kiang, Am. J. Phys., 46, 1188 (1978).
14. H. Pu, L. O. Baksmaty, W. Zang, et al., Phys. Rev. A, 67, 043605 (2003).
15. J. H. Eberly, N. B. Narozhny, and J. J. Sanchez-Mondragon, Phys. Rev. Lett., 44, 1323 (1980).
16. N. B. Narozhny, J. J. Sanchez-Mondragon, and J. H. Eberly, Phys. Rev. A, 23, 236 (1981).
17. G. Rempe, H. Walther, and N. Klein, Phys. Rev. Lett., 58, 353 (1987).
18. J. Parker and C. R. Stroud, Jr., Phys. Rev. Lett., 56, 716 (1986).
19. J. Yeazell, M. Mallalieu and C. R. Stroud, Jr., Phys. Rev. Lett., 64, 2007 (1990).
20. T. Baumert, V. Engel, C. Röttgermann, et al., Chem. Phys. Lett., 191, 639 (1992).
21. I. Sh. Averbukh and N. F. Perelman, Phys. Lett. A, 139, 449 (1989).
22. B. Yurke and D Stoler, Phys. Rev. Lett., 57, 13 (1986).
23. B. Yurke and D. Stoler, Phys. Rev. A, 35, 4846 (1987).
24. A. Mecozzi and P. Tombesi, Phys. Rev. Lett., 58, 1055 (1987).
25. D. L. Aronstein and C. R. Stroud, Jr., Phys. Rev. A, 55, 4526 (1997).
26. D. L. Aronstein and C. R. Stroud, Jr., Phys. Rev. A, 62, 22102 (2000).
27. R. Veilande and I. Bersons, J. Phys. B, 40, 2111 (2007).
28. I. D. Feranchuk and A. V. Leonov, Phys. Lett. A, 373, 517 (2009).
29. M. A. Doncheski and R. W. Robinnet, Ann. Phys., 308, 578 (2003).
30. For a review on the time evolution of wavepackets in bound systems and quantum revivals, see
R. W. Robinnet, Phys. Rep., 392, 1 (2004).
31. I. Marzoli, A. Kaplan, F. Saif, and W. P. Schleich, Fortschr. Phys., 56, 967 (2008).
32. D. Frenkel and R. Portugal, J. Phys. A, 34, 3541 (2001).
33. O. Morsch and M. Oberthaler, Rev. Mod. Phys., 78, 189 (2006).
34. M. A. P. Fisher, P. B. Weichman, G. Grinstein, and D. F. Fisher, Phys. Rev. B, 40, 546 (1989).
35. E. M. Wright, T. Wong, M. J. Collett, et al., Phys. Rev. A, 56, 591 (1997).
36. D Jaksch, C. Bruder, J. I. Cirac, et al., Phys. Rev. Lett., 81, 3108 (1998).
37. M. Greiner, O. Mandel, T. Esslinger, et al., Nature, 415, 51 (2002).
38. I. Bloch, J. Phys. B, 38, S269 (2005).
39. A. Eckardt, C. Wiss, and M. Halthous, Phys. Rev. Lett., 95, 260404 (2005).
40. I. Bloch, J. Dalibard, and W. Zwerger, Rev. Mod. Phys., 80, 885 (2008).

222
Volume 30, Number 3, 2009 Journal of Russian Laser Research

41. S. A. Gardiner, D. Jaksch, R. Dum, et al., Phys. Rev. A, 62, 023612 (2000).
42. A Eckardt, M. Holthaus, H. Lignier, et al., Phys. Rev. A, 79, 013611 (2009).
43. K. W. Madison, PhD Thesis, The University of Texas at Austin, USA (1996).
44. K. Drese and M Holthaus, Chem. Phys., 217, 201 (1997).
45. M. Abramowitz and I. A. Stegun (eds.), Handbook of Mathematical Functions, Dover, New York
(1970), Chapter 20.
46. N. W. McLachlan, Theory and Applications of Mathieu Functions, Oxford University Press, London
(1947).
47. C. Cohen-Tannoudji, J. Dupont-Roc, and Gilbert Grynberg, Atom–Photon Interactions, Wiley &
Sons, New York (1992).
48. C. Leichtle, I. Sh. Averbukh, and W. P. Schleich, Phys. Rev. Lett., 77, 3999 (1996).
49. C. Leichtle, I. Sh. Averbukh, and W. P. Schleich, Phys. Rev. A, 54, 5299 (1996).
50. M. Nauenberg, J. Phys. B., 23, L385 (1990).
51. R. L. Liboff, Introductory Quantum Mechanics, Addison-Wesley, New York (2002), 4th ed.

223

View publication stats