Suplemento de la Revista Mexicana de Fı́sica 1 (3) 31–35 JULY–AUGUST 2020

Numerical implementation of a Mach-Zehnder

interferometer for Bose-Einstein condensates
J. Gil-Londoño, G. Marı́n-Alvarado, and K. Rodrı́guez-Ramı́rez
Departamento de Fı́sica, Universidad del Valle, A.A. 25360, Cali, Colombia.
e-mai: [email protected]
Received 7 September 2019; accepted 29 January 2020

We numerically implement a Mach-Zehnder interferometer, where the coherence and oscillatory properties of Bose-Einstein condensates are
explored and the system is modeled by the Gross-Pitaevskii equation. Several time-dependent external trapping potentials were engineered
seeking the adiabatic regime which is quantified using fidelity measurements with respect to the actual ground-state of the trap. The dynamics
of both conjugate variables, namely density and phase of the matter-wave function, are shown. Moreover, the density and fidelity profiles
of the system are presented when the phase-shifter is switching-on and -off, being found in the presented profiles that the system exhibits
three different regimes during the recombination stage among them even an orthogonal BEC to the original one is obtained. We achieve
the numerical solution through an adequate implementation of the finite-difference method for the spatial discretization and a Runge-Kutta
method for the time evolution.

Keywords: Bose-Einstein condensates; atom interferometry; atomic and molecular physics; coherent optical effects.

Se implementa numéricamente un interferómetro Mach-Zehnder, donde se exploran la coherencia y las propiedades oscilatorias de los con-
densados de Bose-Einstein, el sistema se modela mediante la ecuación de Gross-Pitaevskii. Se diseñaron varios potenciales de atrapamiento
externo dependientes del tiempo con el objetivo de encontrar el régimen adiabático que se cuantifica utilizando mediciones de fidelidad con
respecto al estado base actual del potencial. Se muestra la dinámica de ambas variables conjugadas, esto es, la densidad y la fase de la
función de onda de materia. Además, se muestran los perfiles de densidad y fidelidad del sistema cuando el modificador de fase se activa,
encontrándose en los perfiles presentados que el sistema exhibe tres regı́menes diferentes durante la etapa de recombinación, entre ellos
se obtiene incluso un BEC ortogonal al original. La solución numérica se logra a través de una implementación adecuada del método de
diferencias finitas para la discretización espacial y el método de Runge-Kutta para la evolución temporal.

Descriptores: Condensados de Bose-Einstein; interferometrı́a atómica; fı́sica atómica y molecular; efectos ópticos coherentes.

PACS: 03.75.-b; 03.75.Dg; 67.85.-d; 02.70.-c

1. Introduction

Ultracold atoms in external optical traps constitute an ex-

traordinary tool for the analysis of coherent matter under
extremely well-controlled conditions as highlighted by the
achievement of the Bose-Einstein condensation [1] and the
“superatom”-properties it exhibits [2]. The field is growing
very fast due to the novel possibilities for studies of funda-
mental quantum-mechanical processes. High research activ-
ity has been reported in the last decade including experimen-
tal, theoretical and numerical developments [3, 4]. F IGURE 1. Scheme of the implemented Mach-Zehnder interferom-
eter by means of a time-dependent trapping potential described by
One of the particular interest focuses on the precise Eq. (2), see the text.
manipulation of the quantum-mechanical macroscopic wave
function of the Bose-Einstein condensates (BEC) [5]: It has In this work, we present a numerical implementation of
millions of identical atoms in the same quantum state, and a Mach-Zehnder interferometer (MZI) [13], where the coher-
it is large enough to be optically imaged [6]. On the other ence and oscillatory properties of ultracold bosonic matter-
hand, quantum interference constitutes a major challenge for waves [14] are explored. The system is modeled by the
testing quantum-mechanical foundations [7] besides the real- Gross-Pitaevskii equation (GPE) [15, 16], which is a non-
ization of ultra-precise measurements [8]. linear Schödinger equation and is introduced in Sec. 2.
The particle-wave duality allows us to analyze the appli- The proposed scheme is shown in Fig. 1, and consists of a
cation of matter-wave interferometers from the experimental, time-dependent trapping potential which is discussed and de-
theoretical and numerical points of view [9,10]. Furthermore, scribed in Sec. 3. The incoming wave is divided into two
the development of atom-optics counterparts to beam split- beams by a 50 : 50 beam splitter, one beam travels through
ters, phase-shifters and recombiners makes this field a com- a controlled phase-shifter acquiring a phase. Afterwards, the
plement to the optical interferometers [11, 12]. two beams are recombined registering characteristic density
32 J. GIL-LONDOÑO, G. MARÍN-ALVARADO, AND K. RODRÍGUEZ-RAMÍREZ

and phase patterns. As presented in Sec. 4, we develop sev-

eral time-dependent external trapping potentials, and for each
case, the fidelity is monitored to analyze the adiabaticity of
the dynamical process. Section 5 is devoted to the conclu-
sions and final remarks.
F IGURE 2. As discussed in the text, three different ways to ramp up
2. Model the optical lattice on time to build the double-well potential out of
the initial harmonic confinement are engineered: linear (a), square-
We are interested in studying a bosonic cloud in which tem- root (b) and Gaussian (c).
perature is very close to the absolute zero, i.e. of the order
of few micro-Kelvin (µK). Under this circumstance, the sys- where b = 15 is the strength of the ramping, t2 = (t −
tem undergoes Bose-Einstein condensation and is described tf /2)/b in β2 , t3 = (t − tf /2)/(b/3) in β3 , and tf = 40 is
by a macroscopic wave function, ψ = ψ(~x, t), where the time the final simulation time. The schemes of the potentials gen-
evolution is determined by the GP equation. After a normal- erated from the β-functions are presented in Fig. 2, where
ization, introducing a change of variables and providing the panel (a) corresponds to the potential generated by β1 (t) and
effective confinement to be only in one-dimension, the di- panels (b) and (c) to the potentials obtained by β2 (t) and
mensionless Gross-Pitaevskii equation gets as follows [17], β3 (t), respectively.
The phase shift in the MZI is carried out by entering the
∂ψ(x, t) h ² 2 i
parameter δ(x, t) in the trapping potential. This parame-
i = − ∇ +V (x, t)+g|ψ(x, t)|2 ψ(x, t). (1)
∂t 2 ter is defined in Eq. (4) and allows us to generate a depth
Here the kinetic energyp is measured by ² = ~/(ωx mx2s ) = difference between the minima of the double-well potential,
2 which translates into a phase difference among the two matter
(a0 /xs ) where a0 = ~/ωx m is the harmonic-oscillator
ground-state length in the x−direction, xs is the characteris- beams,
tic BEC length, and m the atomic mass. We consider the con- · ¸
(x − x0 )2 (t − t0 )2
finement frequencies as ωy,z ¿ ωx . The interaction strength δ(x, t) = −δ0 exp − − , (4)
σx σt
among the particles is g = U0 ²3/2 which is governed by a s-
wave contact potential determined by U0 = (4πaS N )/(a0 ), where x0 = 2.530, t0 = tf /2 = 20, σx = 1, σt = 25 and δ0
being aS the s-wave scattering length, and N the number of scales the depth of the phase-shifter. The reason for choosing
particles. these parameters is because once we set our harmonic con-
The numerical solutions are achieved through an ade- finement, the numerical box is settled as well, therefore, all
quate implementation of the finite-differences method for the the potential changes must be performed in such a way that
spatial discretization [18] and the Runge-Kutta method for the dimensions of this numerical box are preserved. In par-
the time evolution [19]. Hence, the dynamics of both con- ticular, the δ-function should be written such that all changes
jugate variables, namely density and phase as well as the fi- in the trapping potential are clearly separated in the spatial
delity, of the matter-wave function are simulated in this way. and temporal axes.
The dynamical processes are intended to be in the most
adiabatic possible manner. Hence, the particles are prepared
3. BEC time-dependent trapping potential
in the ground-state of the initial harmonic trap, and then we
The BEC is initially trapped in a harmonic potential and an let the BEC evolves following the potential. To analyze the
optical lattice is ramped up on time. The resulting time- adiabaticity, we monitored the fidelity (F(t)) by projecting
dependent potential is obtained by a suitable superposition the time-evolved wave-function to the ground state of the ex-
of the form, ternal potential at the corresponding time [21],

V (x, t) = αx2 + β(t) cos2 (ωx) + δ(x, t), (2) F(t) = |hψGS (t)|ψ(t)i|2 . (5)

being α = 0.85 half of the constant force. Once we have defined the construction of the confinement
The ramping up of the optical lattice is a crucial point potential, we expose in Sec. 4 the dynamics of the imple-
under the experimental point of view since this process can mented Mach-Zehnder interferometer.
lead non-adiabiticities that could be end up in an undesired
heating of the system [20]. In order to select an adequate 4. Interferometer dynamics
beam-splitter, three different ways to rise the lattice are ana-
lyzed which are defined as: We present the density and phase of the matter-wave as a
q function of time, as well as the fidelity using the designed
β1 (t) = bt, β2 (t) = b 1 − t22 , trapping potentials. In Sec. 4.1, we present the dynamics for
¡ ¢ the three different activation functions of the optical lattice in
and β3 (t) = b exp −t23 , (3) order to rise the double-well potential. Furthermore, using

Rev. Mex. Fı́s. 1 (3) 31–35

 NUMERICAL IMPLEMENTATION OF A MACH-ZEHNDER INTERFEROMETER FOR BOSE-EINSTEIN CONDENSATES 33

F IGURE 3. Matter-wave interferometer in absence of the phase- F IGURE 5. Matter-wave interferometer in absence of the phase-
shifter, using β1 (t), the evolution of the density (a), fidelity (b) and shifter, using β3 (t), the evolution of the density (a), fidelity (b) and
phase (c). phase (c).

β-functions (see Eq. (3)). Since our purpose is to determine

which of the splitters works more properly for our interest
the dynamics are initially presented without introducing any
phase shifting in the system.
At first, let us start with the dynamics display by using
the β1 (t) function shown in Fig. 3. In panel 3(a) we observe
that the splitted-beams stage lasts longer than the recombi-
nation process, being not a good sign for our purposes. The
fidelity behavior is presented in panel 3(b), it clearly fits our
expectations of having adiabatic evolutions. Indeed, we can
see tiny deviations from the unity when the beam-splitter is
switching-on and -off. On the other hand, in the phase profile
presented in panel 3(c), symmetrical behavior is observed as
a function of the position, as expected. Although manifesting
edge effects the phase propagates properly.
In the second place, we analyze the matter-wave evolu-
tion when the β2 (t) function is used as the splitter, see Fig. 4.
The density profile presented in panel 4(a) exhibits strong
fluctuations, which is evidenced by the strong fall from the
unity of the fidelity profile (panel 4(b)), during the splitting
F IGURE 4. Matter-wave interferometer in absence of the phase- and even stronger during the recombination process. Hence,
shifter, using β2 (t), the evolution of the density (a), fidelity (b) and
since the evolution of the system moves significantly away
phase (c).
from the ground state, even when a phase is not printed in
the ramping-time function β3 (t) (see Fig. 2(c)), we present the system, we discard this kind of beam-splitters. The phase
in Sec. 4.2, three different regimes resulting in the evolution profile (panel 4(c)) shows symmetric behavior as a function
as a function of the phase-shifter depth. of the position, however, it presents even stronger undesirable
edge effects.
4.1. Splitting matter-waves On the other hand, Fig. 5 manifests adequate behavior by
using β3 (t), that is, the recombination process in the density
We activate the splitting of matter-waves through the term profile lasts longer as it is observed in panel 5(a); besides,
β(t) introduced in Eq. (2). The Figs. 3, 4 and 5 present the the dynamics of the system is adiabatic enough to fulfill our
obtained dynamics by using the three introduced ramped-up requirements, which is evidenced in the fidelity profile (see

Rev. Mex. Fı́s. 1 (3) 31–35

34 J. GIL-LONDOÑO, G. MARÍN-ALVARADO, AND K. RODRÍGUEZ-RAMÍREZ

panel 5(a)). Here, the evolved state presents a small devia-

tion from the ground state only at the times when a change
in the confinement potential is introduced but the fidelity re-
turns to one rapidly in the recombination stage. Contrary to
the phase profiles illustrated in Figs. 3 and 4, no edge effects
are observed at all, as it is shown in panel 5(c). Therefore, in
order to study the dynamics developed by the inclusion of the
phase-shifter, Eq. (4), in the interferometer, we conclude that
the best analyzed beam-splitter is generated by β3 (t), hence,
all the subsequent results are obtained using it.

4.2. Phase-shifter

Once the initial beam is splitted in two, a phase difference is

introduced by the phase-shifter described in Eq. (4).
The term δ0 in the parameter δ(x, t), take values between F IGURE 7. Matter-wave interferometer in presence of the phase-
δ0 = 0 and δ0 = 12.0, establishing a maximum relative depth shifter, using β3 (t), the evolution of the density ((a), (c), (e)) and
between the minima of the double-well potential of h ≈ 5.51, fidelity ((b), (d), (f)). Regime 2.
while the horizontal distance between the wells is x ≈ 3.31.
Figures 6, 7 and 8 show the density and fidelity profiles for
different relative depths, and three distinct behaviors can be
seen.
Panels (a), (c) and (e) of Fig. 6 present the density pro-
files for the relative depths generated by δ0 = 2.0, δ0 = 6.0
and δ0 = 9.4 respectively. In these panels, a minimum pop-
ulation in the central position during the recombination stage
is evidenced. On the other hand, panels (b), (d) and (f) show
the obtained fidelity profiles. Here, a sudden decay in two
steps is observed, one in the beam splitting stage and then in
the recombination. Clearly showing that the recombined state
is orthogonal to the ground-state of the single-well potential,
and an identical behavior for these three relative depths is ob-
served.
Figure 7 displays the density profiles (panels (a), (c) and
(e)) and the fidelity curves (panels (b), (d) and (f)) introduc- F IGURE 8. Matter-wave interferometer in presence of the phase-
ing potential depths of δ0 = 3.0, δ0 = 6.8 and δ0 = 10.2, shifter, using β3 (t), the evolution of the density ((a), (c), (e)) and
respectively. The density profiles show that during the recom- fidelity ((b), (d), (f)). Regime 3.

bination process the maximum distribution of the particles

tends to stay in the classical returning points in an alternative
manner. Panels (b), (d) and (f) show that the fidelity moves
away notoriously from the unit once the potential is modi-
fied, although reaching a stable value higher than the one dis-
cussed in the previous case. It suggests that the recombined
state has at least a tiny probability to be in the ground-state
among its superposition.
Figure 8 exhibits the density profiles for depths of δ0 =
4.3, δ0 = 8.0 and δ0 = 11.2 (panels (a), (c) and (e), re-
spectively), it is observed that the density in the recombina-
tion stage oscillates around the central position. On the other
hand, panels (b), (d) and (f) show the corresponding fidelity
profiles and it can be seen deviations from the unit during
the beam splitting process but it recovers, although not com-
F IGURE 6. Matter-wave interferometer in presence of the phase- pletely but still good enough in the recombination stage.
shifter, using β3 (t), the evolution of the density ((a), (c), (e)) and In summary, three specific regimes have been manifested
fidelity ((b), (d), (f)). Regime 1. in the dynamics of the implemented interferometer.

Rev. Mex. Fı́s. 1 (3) 31–35

 NUMERICAL IMPLEMENTATION OF A MACH-ZEHNDER INTERFEROMETER FOR BOSE-EINSTEIN CONDENSATES 35

5. Conclusions Once the beam-splitter is established, the phase-shifter

δ(x, t) is introduced in the confinement potential, through
We build numerically a time-dependent external trapping po- which a relative depth was generated between the minima of
tential to confine Bose-Einstein condensates and simulate a the double-well potential. δ(x, t) had a range of variation me-
Mach-Zehnder interferometer. In order to do so, three ac- diated by parameter δ0 , which generates relative depth values
tivation functions of the double-well potential were consid- between h = 0 and h ≈ 5.51, while the distance between the
ered and, based on the set of defined observables, the β3 (t) wells is kept at x ≈ 3.31.
function has been chosen to lead to the most adiabatic possi- It was found that the dynamics of the system exhibits
ble dynamics. This function takes the important role of the three distinct regimes in the recombination stage, which must
matter-wave beam-splitter for our interferometer. be analyzed in detail in future works.

1. M. H. Anderson, J. R. Ensher, M. R. Matthews, C. E. Wieman, 12. T. Berrada, S. van Frank, R. Bücker, T. Schumm, J.-F. Schaff,
and E. A. Cornell, Sci. 269 (1995) 198. and J. Schmiedmayer, Nat. Commun. 4 (2013) 2077.
2. K. B. Davis et al., Phys. Rev. Lett. 75 (1995) 3969. 13. G. D. McDonald et al., Phys. Rev. A 87 (2013) 013632.
3. I. Bloch, J. Dalibard, and W. Zwerger, Rev. Mod. Phys. 80 14. B. Lücke et al., Sci. 334 (2011) 773.
(2008) 885.
15. E. P. Gross, II Nuovo Cimento (1955-1965) 20 (1961) 454.
4. S. Giorgini, L. P. Pitaevskii, and S. Stringari, Rev. Mod. Phys.
16. L. Pitaevskii, Sov. Phys. JETP 13 (1961) 451.
80 (2008) 1215.
17. W. Bao, D. Jaksch, and P. A. Markowich, J. Comput. Phys. 187
5. R. Corgier et al., New J. Phys. 20 (2018) 055002.
(2003) 318.
6. F. V. Pepe, Bose-einstein condensation: static and dynamical
aspects, Ph.D. thesis, Universit Degli Studi Di Bari Aldo Moro 18. W. C. David Kincaid, Numerical Analysis: Mathematics of Sci-
(2013). entific Computing, 3rd ed. (American Mathematical Society,
Providence Rhode Island, 2009), pages 589-593.
7. T. van Zoest et al., Sci. 328 (2010) 1540.
19. A. D. S. R. Kent Nagle, Edward B. Saff, Ecuaciones Diferen-
8. A. O. Jamison, Precision interferometry with bose-einstein con- ciales y Problemas con Valores en la Frontera, 4 ed. (Pearson
densates, Ph.D. thesis, University of Washington (2014). Educación, México, 2009), págs. 135-137.
9. T. Schumm et al., Nat. Phys. 1 (2005) 57. 20. M. Dolfi, A. Kantian, B. Bauer, and M. Troyer, Phys. Rev. A 91
10. S. van Frank et al., Nat. Commun. 5 (2014) 4009. (2015) 033407.
11. A. D. Cronin, J. Schmiedmayer, and D. E. Pritchard, Rev. Mod. 21. O. Lychkovskiy, O. Gamayun, and V. Cheianov, Phys. Rev. B
Phys. 81 (2009) 1051. 98 (2018) 024307.

Rev. Mex. Fı́s. 1 (3) 31–35