PHYSICAL REVIEW A 67, 062313 共2003兲

Qutrit quantum computer with trapped ions

A. B. Klimov,1 R. Guzmán,2 J. C. Retamal,2 and C. Saavedra3,*
1
Departamento de Fı́sica, Universidad de Guadalajara, Revolución 1500, 44410, Guadalajara, Jalisco, Mexico
2
Departamento de Fı́sica, Universidad de Santiago de Chile, Casilla 307, Correo 2, Santiago, Chile
3
Departamento de Fı́sica, Universidad de Concepción, Casilla 160-C, Concepción, Chile
共Received 25 June 2002; revised manuscript received 14 January 2003; published 26 June 2003兲
We study the physical implementation of a qutrit quantum computer in the context of trapped ions. Qutrits
are defined in terms of electronic levels of trapped ions. We concentrate our attention on a universal two-qutrit
gate, which corresponds to a controlled-NOT gate between qutrits. Using this gate and a general gate of an
individual qutrit, any gate can be decomposed into a sequence of these gates. In particular, we show how this
works for performing the quantum Fourier transform for n qutrits.

DOI: 10.1103/PhysRevA.67.062313 PACS number共s兲: 03.67.Lx

I. INTRODUCTION photon states from a parametric down-conversion process

The building blocks of a quantum computer are qubits 关1兴,
which are distinguishable two-level physical systems, which
are manipulated individually as well as collectively. These
operations are performed using the so-called quantum gates
in analogy to their classical counterparts. In the simplest
case, collective manipulation reduces to bipartite quantum
gates, for instance, the controlled-NOT gate 关2兴, also called
XOR gate. In this gate one qubit acts as a control and the
other as the target, so that if the state of the control qubit is
in state 兩 1 典 , the state of the target qubit is flipped, and in
other cases the target qubit remains unchanged. The experi-
mental implementation of a two-qubit XOR has been studied
in different physical contexts, as for instance, in trapped ions
关3,4兴, nuclear magnetic resonance 关5–7兴, cavity QED sys-
tems and quantum heterostructures 关8 –10兴.
In recent works, the use of quantum entanglement in
higher-dimensional quantum systems has been studied. The
notion of entanglement generation and characterization in the
case of three-level quantum systems, qutrits, has been con-
sidered by authors in Ref. 关11兴. The use of qutrits instead of
qubits has been proven to be more secure against a symmet-
ric attack on a quantum key distribution protocol 关12兴. These
studies require a generalized version of an XOR, which has
been given in Ref. 关13兴, which is called the GXOR gate. This
gate operates on a tensor product of two qudits, states lying
in a d-dimensional Hilbert space. Further studies on qudit
systems have been considered, for example, bounds on en-
tanglement between qudits 关14兴, discrimination among the
Bell states of qudits 关15兴, entanglement among qudits 关16兴,
Greenberger-Horne-Zeilinger paradox for many qudits 关17兴,
quantum computing with qudits 关18兴, quantum tomography
for qudits states 关19兴, entanglement swapping between mul-
tiqudit systems 关20兴. Recently, a quantum communication
complexity protocol with two entangled qutrits has been pro-
posed 关21兴. Finally, an important step towards the use of
higher-dimensional quantum systems has been given with
the experimental generation of entangled qutrits using two-
关22兴.
In the present work, we study the accomplishment of
quantum gates for a qutrit based quantum computer, by ex-
ploiting the possibility of coherent manipulation of ions in a
linear trap 关3兴. The main goal in studying quantum compu-
tation with qutrits instead of qubits is the exponential in-
crease of the available Hilbert space with the same amount of
physical resources. In particular, we focus our attention both
to implementing a quantum Fourier transform for one qutrit
and to the conditional quantum gate between two qutrits,
XOR (3) . We also give a protocol for a quantum Fourier trans-
form among many qutrits. The extension of the physical
implementation from qubits to qutrits is, of course, nontrivial
because of the coherent operations required in a three-
dimensional Hilbert space. As far as we know, there is a
proposal for an implementation of a conditional gate for qu-
dits, called GXOR based on two-mode field interactions 关13兴,
which assumes the existence of a discrete Fourier transform
for an arbitrary field state lying in a D-dimensional subspace
of one of the field modes.
Here, conditional gates between two qutrits are conve-
niently expressed as
XOR12
(3)
兩 i 典 1 兩 j 典 2 ⫽ 兩 i 典 1 兩 j⫾i 典 2 , 共1兲
where j⫾i denotes the addition 共difference兲 i⫾ j, modulo 3.
(3)
As will be shown, this gate is decomposed as XOR mn
⫺1
⫽F n P mn F n , where F n is the quantum Fourier transform
for one qutrit, and P mn is a conditional phase-shift gate for
qutrits, where m and n are the control and target qutrits,
respectively. As indicated above, the XOR (3) gate is given in
two main steps. One is the generation of the discrete quan-
tum Fourier transform, which requires the coherent manipu-
lation of populations in a three-dimensional Hilbert space.
The second is the conditional phase-shift gate, which re-
quires the intervention of an ancillary quantum channel be-
tween the qutrits.
II. ARBITRARY ONE-QUTRIT GATE

Here we consider the achievement of a general gate for a

*Electronic address: carlos.saavedra@udec.cl single qutrit. An arbitrary unitary operation on a qutrit state,

1050-2947/2003/67共6兲/062313共7兲/$20.00 67 062313-1 ©2003 The American Physical Society

KLIMOV et al. PHYSICAL REVIEW A 67, 062313 共2003兲

FIG. 2. Electronic level structure of trapped ion. Quantum in-

formation of qutrits is stored in levels 兩 0 典 , 兩 1 典 , and 兩 2 典 . The tran-
sitions involving effective interactions between levels 兩 0 典 → 兩 1 典 and
兩 0 典 → 兩 2 典 are driven by classical fields with different polarizations.

FIG. 1. Raman configurations for defining the logical states of a
qutrit.
U共3兲 operation, is split up into a sequence of SU共2兲 opera-
tions. A physical setup for this is obtained in a linear array of
trapped ions, where qutrit states are defined using Zeeman’s
level structure, in a 138Ba⫹ ion in a Paul trap. A representa-
tion of this level structure is depicted in Fig. 1. Two Raman
configurations are independently implemented between the
levels 6S 1/2(m⫽⫺1/2), 6 P 1/2(m⫽⫺1/2), 5D 3/2(m
⫽⫺1/2) and levels 6S 1/2(m⫽⫺1/2), 6 P 3/2(m⫽⫺3/2),
5D 3/2(m⫽⫺3/2). The first and second Raman configura-
tions require a ␴ ⫹ polarized light and a ␴ ⫺ polarized light,
respectively. Diode lasers 关23兴 are used in order to achieve
the interactions. In this case, typical values of magnetic fields
for the Zeeman splitting are of the order of 10 G, and the
conditional two-qutrit gate. A key ingredient is the existence
of electric dipole forbidden transitions 兩 0 典 → 兩 1 典 and 兩 0 典
→ 兩 2 典 . These transitions are addressed via Raman transitions
through the independent channels associated with orthogonal
polarizations, driven by classical fields ⍀ 03 , ⍀ 13 , ⍀ 04 , and
⍀ 24 .
In this system, the ion level populations are manipulated
by selecting the desired coherent operation. For instance, we
can independently operate with transitions 兩 0 典 → 兩 1 典 , 兩 0 典
→ 兩 2 典 , 兩 1 典 → 兩 2 典 by adjusting the parameters. The Hamil-
tonian describing this effective system, under the standard
dipole and rotating wave approximations, is given by
H⫽ 兺j ប ␻ j 兩 j 典具 j 兩 ⫹ប 兵 e ⫺i ␯ t共 ⍀ 04兩 4 典具 0 兩 ⫹⍀ 03兩 3 典具 0 兩 兲
2

corresponding energy differences between two consecutive

ជ兩 共MHz/G兲 for level S 1/2 , ⫹e ⫺i ␯ 1 t 共 ⍀ 13兩 3 典具 1 兩 ⫹⍀ 24兩 4 典具 2 兩 兲 ⫹H.c.其 , 共2兲
Zeeman sublevels are u⫽0.14兩 B
u⫽ 3 0.14兩 B
2 ជ兩 共MHz/G兲 for level P 1/2 , and u⫽ 54 0.14兩 B ជ兩
where j⫽0,1,2,3,4. In the case of single-qutrit gates, only
共MHz/G兲 for D 3/2 . From these values, one finds that a typi- the carrier transition in the ion is considered, so that no ex-
cal energy difference is of few MHz. The transition fre- plicit effects on the center-of-mass motion of the ion are
quency between levels S⇔ P and D⇔ P is typically hun- included. Thus, the spatial dependences of Raman fields have
dreds of MHz, and the trap oscillation frequency reported is been included as phase factors. Assuming the following con-
of the order of tens of MHz 关1,24兴. ditions: ⌬⫽( ␻ 4 ⫺ ␻ 0 )⫺ ␯ 2 ⫽( ␻ 3 ⫺ ␻ 0 )⫺ ␯ 2 ⫽( ␻ 4 ⫺ ␻ 2 )
An extra single-ion operation is needed for implementing ⫺ ␯ 1 ⫽( ␻ 3 ⫺ ␻ 1 )⫺ ␯ 1 and ⌬Ⰷ⍀ 04 , ⍀ 03 , ⍀ 31 , ⍀ 42 , rapidly
the controlled quantum gate operation, due to the presence of decaying upper levels 兩 3 典 and 兩 4 典 are adiabatically elimi-
a phase if only two Raman configurations are used. This nated, leading to an effective Hamiltonian:
phase is removed with an additional Raman configuration, in
this case 6S 1/2(m⫽⫺1/2), 6 P 1/2(m⫽⫺1/2), 5D 3/2(m H 兩 ⍀ 31兩 2 兩 ⍀ 42兩 2 兩 ⍀ 30兩 2 ⫹ 兩 ⍀ 40兩 2
⫽⫺1/2). This transition S⇔ P is achieved with a ␲ polar- ⫽⫺ 兩 1 典具 1 兩 ⫺ 兩 2 典具 2 兩 ⫺ 兩 0 典具 0 兩
ization, and the S⇔D transition using a ␴ ⫺ polarization for ប ⌬ ⌬ ⌬
the electromagnetic field. A similar physical implementation
is performed using Ca⫹ . However, using Ca⫹ extra opera-
tions are done via a quadrupole direct transition between the
⫺ 再 ⍀ 31⍀ 30
⌬
*
兩 0 典具 1 兩 ⫹
⍀ 42⍀ 40
⌬
*
冎
兩 0 典具 2 兩 ⫹H.c. . 共3兲

levels 6S 1/2 and 5D 5/2 , which are forbidden in dipole cou-

plings. All of these physical parameters make possible the Assuming the additional condition
implementation of a qutrit quantum computer with barium
trapped ions using the fine structure 关25兴. 兩 ⍀ 31兩 2 兩 ⍀ 42兩 2 兩 ⍀ 30兩 2 ⫹ 兩 ⍀ 40兩 2
⫽ ⫽ , 共4兲
For the sake of simplicity, here we consider a simplified ⌬ ⌬ ⌬
level structure of ions, only depicting the relevant electronic
transitions of ions, as is shown in Fig. 2. Levels we can find the evolution for the system. After some calcu-
兵 兩 0 典 , 兩 1 典 , 兩 2 典 其 are the logical states of a single qutrit. The lations, the evolution operator in the restricted three-
level 兩 0 ⬘ 典 is an auxiliary level necessary when defining the dimensional space 兵 兩 2 典 , 兩 1 典 , 兩 0 典 其 is given by

062313-2
QUTRIT QUANTUM COMPUTER WITH TRAPPED IONS PHYSICAL REVIEW A 67, 062313 共2003兲

冉 1⫹ 兩 g 兩 2 C 共 ␸ 兲 gg ⬘ * C 共 ␸ 兲 ⫺ig sin ␸

冊
2
1
U共 ␸ 兲⫽ g ⬘g *C共 ␸ 兲 1⫹ 兩 g ⬘ 兩 2 C 共 ␸ 兲 ⫺ig ⬘ sin ␸ ,
F兩 j典⫽
冑3 兺 e 2i ␲ l j/3兩 l 典 .
l⫽0
共11兲

⫺ig * sin ␸ ⫺ig ⬘ * sin ␸ cos ␸

共5兲 Explicitly, the transformed states 兩 j̄ 典 ⫽F 兩 j 典 read as

where ␸ ⫽⍀t is an adimensional interaction time, C( ␸ ) 1

兩 0̄ 典 ⫽ 共 兩 0 典 ⫹ 兩 1 典 ⫹ 兩 2 典 ), 共12兲
⫽cos ␸⫺1, and ⍀ 2 ⫽ 兩 ␬ ⬘ 兩 2 ⫹ 兩 ␬ 兩 2 . We have introduced the 冑3
notations g⫽ ␬ /⍀ and g ⬘ ⫽ ␬ ⬘ /⍀, where ␬ ⫽⍀ 40⍀ 42 * /⌬ and
␬ ⬘ ⫽⍀ 30⍀ 31* /⌬. This evolution operator allows implement- 1
ing all the required coherent operations between any two 兩 1̄ 典 ⫽ 共 兩 0 典 ⫹e 2i ␲ /3兩 1 典 ⫹e ⫺2i ␲ /3兩 2 典 ),
logical states. For instance, to activate the transition 兩 1 典 冑3
→ 兩 2 典 , we assume ␸ ⫽ ␲ in Eq. 共5兲, so that

冉 冊
1
兩 2̄ 典 ⫽ 共 兩 0 典 ⫹e ⫺2i ␲ /3兩 1 典 ⫹e 2i ␲ /3兩 2 典 ).
cos ␣ ⫺e i ␤ 1 sin ␣ 0 冑3
⫺i ␤
U 1 ⫽ ⫺e 1 sin ␣ ⫺cos ␣ 0 , 共6兲
This transformation is obtained by using the general operator
0 0 ⫺1 we have calculated before, for the carrier transition, and
combining with adiabatic transitions in both polarization
where we have defined channels. After some calculations, it is found that the Fourier
transform is decomposed into the form
兩 ␬ ⬘兩 2⫺ 兩 ␬ 兩 2
cos ␣ ⫽ and e i ␤ 1 ⫽ ␬␬ ⬘ * / 兩 ␬␬ ⬘ * 兩 . 共7兲 F⫽iU D U 2 U 3 U 1 , 共13兲
兩 ␬ ⬘兩 2⫹ 兩 ␬ 兩 2
where each one of these operations is obtained from the
Other transitions are addressed by manipulating the ␬ cou- above process: In U D , if %⫽ ␲ /3, ␧⫽ ␲ /6, we get

冉 冊
plings in the transition 兩 0 典 → 兩 2 典 , or ␬ ⬘ in the transition 兩 0 典
→ 兩 1 典 . For example, the transition 兩 0 典 → 兩 2 典 is addressed by e i ␲ /3 0 0
assuming ␬ ⬘ ⫽0 so that U D⫽ 0 e i ␲ /6
0 . 共14兲

冉 冊
⫺i ␲ /2
0 0 e
cos ␸ 2 0 ⫺ie ⫺i ␤ 2 sin ␸ 2
U 2⫽ 0 1 0 , 共8兲 In U 1 with ␣ ⫽ ␲ /4,␤ 1 ⫽⫺2 ␲ /3,

冉 冊
i␤2
⫺ie sin ␸ 2 0 cos ␸ 2 1 ⫺e ⫺2i ␲ /3 0
1 2i ␲ /3
where ␬ ⫽e i ␤ 2 兩 ␬ 兩 and ␸ 3 ⫽ 兩 ␬ 兩 t. Finally, for the transition U 1⫽ ⫺e ⫺1 0 . 共15兲
冑2
兩 0 典 → 兩 1 典 , we assume ␬ ⫽0: 0 0 ⫺ 冑2

U 3⫽ 0
0
冉 1 0
cos ␸ 3
⫺ie ⫺i ␤ 3 sin ␸ 3
0
⫺ie i ␤ 3 sin ␸ 3 ,
cos ␸ 3
冊 共9兲
If ␸ 2 ⫽ ␲ /4,␤ 2 ⫽⫺ ␲ /3 in U 2 ,

U 2⫽
1
冑2 冉 1
0 冑2
0 ie 2i ␲ /3
0
冊 . 共16兲
⫺2i ␲ /3
where ␬ ⬘ ⫽e i ␤ 3 兩 ␬ ⬘ 兩 and ␸ 3 ⫽ 兩 ␬ ⬘ 兩 t. In order to generate any ie 0 1
SU共3兲 operator 关26兴, we need to have a decomposition of
Finally, assuming ␸ 3 ⫽⫺ ␲ /3,␤ 3 ⫽7 ␲ /6 in U 3 ,

冉 冊
height-independent parameters. In the previous cases, we
have three operations involving six independent parameters. 冑3 0 0
By connecting interactions of classical fields with 兩 0 典 → 兩 1 典 1
and the 兩 0 典 → 兩 2 典 transitions in the far off-resonance limit, U 3⫽ 0 ⫺ 冑2 ie ⫺i ␲ /6
. 共17兲
冑3
we obtain the following dispersive evolution: 0 ie i ␲ /6
⫺ 冑2

U D⫽ 冉 e i%
0
0
e
0

0
i␧

e
0
0
⫺i(%⫹␧)
冊 , 共10兲
In the same way, any arbitrary one-qutrit gate is decomposed
as in the case of the Fourier transform using the evolution
operator given by Eq. 共5兲.

III. CONDITIONAL TWO-QUTRIT GATE

which provides two additional parameters.
In particular, we can now focus on the calculation of the The Fourier transform that we have found plays the same
quantum Fourier transform for one qutrit, which is a unitary role that the Hadamard gate plays for qubits. For example,
operation defined by the entangled states of two qubits are generated by the appli-

062313-3
KLIMOV et al. PHYSICAL REVIEW A 67, 062313 共2003兲

cation of a Hadamard gate followed by a conditional phase- where we have defined ⍀ q ␩ t/2⫽l ␲ /2. As we shall see, these
shift gate. In what follows we will describe the accomplish- coherent operations allow selecting the quantum channel for
ment of a conditional phase shift between two qutrits, which transferring the information to the center-of-mass. After this
allows the achievement of the XOR (3) gate we have intro- it is necessary to introduce a phase change in the qutrit state,
duced in Sec. I. A basic requirement for performing a condi- which depends on the energy of the center-of-mass state.
tional gate between two qutrits is providing a mechanism to This phase change is accomplished through the dispersive
distinguish independent quantum paths, in order to satisfy regime of the first red sideband in Eq. 共18兲, that is,
the conditional change in the target qutrit depending on the
† †
state of the control. In our case, we have defined the XOR (3) q
Dm 共 ␸ 兲 ⫽e i ␸ aa 兩 q 典 m 具 q 兩 ⫹e ⫺i ␸ a a 兩 0 典 m 具 0 兩 , 共19兲
in such a way that the target changes only when the control
qutrit is in state 兩 1 典 or 兩 2 典 , provided that the state of the where ␸ ⫽ (⍀ q ␩ ) 2 /4␦ , which allows for an intensity-
target will be 兩 j 丣 0 典 ⫽ 兩 j 典 when the control qubit is in state dependent phase shift of the electronic levels.
兩 0 典 . Thus, we only need to implement a protocol considering From Eq. 共12兲 it is not difficult to figure out that the
independent quantum channels through states 兩 1 典 and 兩 2 典 of conditional phase shift needed to implement the
the control qutrit. In a set of ions in a linear trap, in the XOR (3) 兩 i 典 m 兩 j 典 n ⫽ 兩 i 典 m 兩 j両i 典 n is given by
electronic configuration given in Fig. 2, such quantum chan-
nels are established with the compromise of the collective 兩 0 典 m兩 0 典 n 兩 0 典 m兩 0 典 n
center-of-mass 共CM兲 motion of ions inside the trap. 兩 0 典 m兩 1 典 n 兩 0 典 m兩 1 典 n
In the ion-trap quantum computer with qubits 关3兴, the
兩 0 典 m兩 2 典 n 兩 0 典 m兩 2 典 n
quantum channel between ions is established through the
center-of-mass motion, which is addressed by adjusting a 兩 1 典 m兩 0 典 n 兩 1 典 m兩 0 典 n
Raman transition to a given red sideband. As we shall see in 2兲 共1兲
P 共mn P mn
what follows, we can proceed along the same line of reason- 兩 1 典 m兩 1 典 n e 4i ␲ /3 兩 1 典 m 兩 1 典 n
——→
ing, considering the ion model described in Fig. 2. Let us
assume that we adjust the field amplitudes such that ⍀ 04 兩 1 典 m兩 2 典 n e 2i ␲ /3 兩 1 典 m 兩 2 典 n
⫽⍀ 24⫽0, and ⍀ 31 , ⍀ 03⫽0; or ⍀ 04⫽⍀ 24⫽0, and ⍀ 31 , 兩 2 典 m兩 0 典 n 兩 2 典 m兩 0 典 n
⍀ 03⫽0. In both cases, after eliminating the upper excited 兩 2 典 m兩 1 典 n e 2i ␲ /3
兩 2 典 m兩 1 典 n
level and adjusting to the first red sideband transition, we
4i ␲ /3
obtain the Hamiltonian describing the ion center of mass 兩 2 典 m兩 2 典 n e 兩 2 典 m兩 2 典 n, 共20兲
coupled to the electronic transition 兩 0 典 → 兩 q 典 : (1) (2)
where P mn and P mn are defined as follows:
⍀ q␩
关 兩 q 典 n 具 0 兩 ae ⫺i ␦ t⫺i ␾ ⫹a † 兩 0 典 n 具 q 兩 e i ␦ t⫹i ␾ 兴 . 共 ␾ 1 , ␾ 2 兲 ⫽R 00⬘ 共 ␲ 兲 U m 共 3 ␲ /2兲 D 2n 共 ␰ 2 兲 D 2n 共 ␾ 2 兲 D 1n 共 ␰ 1 兲
(1) 1,1
H n,q ⫽ P mn
2
共18兲 ⫻D 1n 共 ␾ 1 兲 U m
1,1
共 ␲ /2兲 R 00⬘ 共 ␲ 兲 ,
Here a and a † are the annihilation and creation operators of (2)
P mn 共 ␾ 2 , ␾ 1 兲 ⫽R 00⬘ 共 ␲ 兲 U m
1,2
共 3 ␲ /2兲 D 2n 共 ␰ 1 兲 D 2n 共 ␾ 1 兲 D 1n 共 ␰ 2 兲
the CM phonons, respectively, ⍀ q is the effective Rabi fre-
quency after adiabatic elimination of upper excited levels, ␾ ⫻D 1n 共 ␾ 2 兲 U m
1,2
共 ␲ /2兲 R 00⬘ 共 ␲ 兲 , 共21兲
is the laser phase, ␦ ⫽ ␻ 2 ⫺ ␻ 0 ⫺ ␯ 2 ⫹ ␯ 1 ⫹ ␯ x ⫽ ␻ 1 ⫺ ␻ 0 ⫺ ␯ 2
⫹ ␯ 1 ⫹ ␯ x , and ␩ ⫽ 冑បk 2␪ /(2M ␯ x ) is the Lamb-Dicke param- with ␰ i ⫽2 ␲ ⫺ ␾ i . The operation R 00⬘ ( ␲ ) is a rotation which
eter (k ␪ ⫽lcos␪, with kជ the laser wave vector and ␪ the angle only impinges on the ion when it is in the 兩 0 典 level, sending
between the X axis and the direction of laser propagation兲. it to the 兩 0 ⬘ 典 level, avoiding producing any phase shift in this
The subscript q⫽1,2 refers to the transition excited by the state. The dispersive operations affecting transitions 兩 0 典
→ 兩 1 典 and 兩 0 典 → 兩 2 典 are given by

冉 冊
laser in the nth ion. The center-of-mass motion coupled to
electronic transitions is coherently manipulated by the selec-
1 0 0
tion of the effective interaction time and laser polarizations.
aa †
This Hamiltonian allows implementing the coherent interac- D 1n 共 ␾ 1 兲 ⫽ 0 e i␾1 0 ,
tion between qutrits and the collective center-of-mass mo- a†a
tion. In particular, in order to implement a conditional phase 0 0 e ⫺i ␾ 1

冉 冊
shift, the following operations are needed:
1 0 0
l,q
Um 共 ␾ 兲兩 0 典 m兩 0 典 ⫽ 兩 0 典 m兩 0 典 , D 1n 共 ␸ 1 兲 ⫽ 0 e i␸1
0 ,

冉 冊 冉 冊
⫺i ␸
0 0 e 1
l␲ l␲

冉 冊
l,q
Um 共 ␾ 兲 兩 0 典 m 兩 1 典 ⫽cos 兩 0 典 m 兩 1 典 ⫺ie ⫺i ␾ sin 兩 q 典 m兩 0 典 ,
2 2 e i ␾ 2 aa
†
0 0

l,q
Um 共 ␾ 兲 兩 q 典 m 兩 0 典 ⫽cos 冉 冊
l␲
2
兩 q 典 m 兩 0 典 ⫺ie i ␾ sin
l␲
2 冉 冊
兩 0 典 m兩 1 典 ,
D 2n 共 ␾ 2 兲 ⫽ 0
0
1
0 e
0
⫺i ␾ 2 a † a
,

062313-4
QUTRIT QUANTUM COMPUTER WITH TRAPPED IONS PHYSICAL REVIEW A 67, 062313 共2003兲

D 2n 共 ␸ 2 兲 ⫽ 冉 e i␸2
0
0
0
1
0 e
0
0
⫺i ␸
2
冊 .
from the mathematical guidelines developed to define the
Fourier transform for a system of n qubits 关1兴. In basis 3, the
general expression for the Fourier transform is given by
3 n ⫺1
1
兺
n
Thus, the particular phase shift in Eq. 共20兲 is achieved for 兩 j̄ 典 ⫽ e 2 ␲ i j(k/3 ) 兩 k 典 . 共24兲
(1)
P mn (4 ␲ /3,2 ␲ /3) and for P mn
(2)
(2 ␲ /3,4 ␲ /3). Finally, the ef- 3 n/2 k⫽0
fective conditional change of the state of the target qutrit
In this context, one important element is the decomposition
gives rise to the XOR (3) gate. In brief,
of an integer number 0⭐ j⭐3 n ⫺1 in basis 3, which is con-
⫺1 veniently given as
XORmn ⫽F n 共22兲
(3) (2) (1)
P mn P mn F n .
(3)
In this way, the XOR mn produces the following evolution of j⫽ j 1 3 n⫺1 ⫹ j 2 3 n⫺2 ⫹•••⫹ j n 3 0 . 共25兲
state of two qutrits: The fraction k/3n can be written in the 3-basis as
兩0典兩0典 兩 0 典 兩 0̄ 典 兩 0 典 兩 0̄ 典 兩0典兩0典 n
k k 1 3 n⫺1 k 2 3 n⫺2 k n3 0
兩0典兩1典 兩 0 典 兩 1̄ 典 兩 0 典 兩 1̄ 典 兩0典兩1典 3n
⫽
3n
⫹
3n
⫹•••⫹
3n
⫽ 兺 k l 3 ⫺l .
l⫽1
共26兲

兩0典兩2典 兩 0 典 兩 2̄ 典 兩 0 典 兩 2̄ 典 兩0典兩2典
Following an analysis similar to that carried out for a system
兩1典兩0典 兩 1 典 兩 0̄ 典 兩 1 典 兩 2̄ 典 兩1典兩2典 of n qubit, so that the Fourier transform now is written in the
form
兩1典兩1典 Fn 2兲 共1兲
P 共mn P mn F ⫺1 兩1典兩0典

冋兺 册
→
兩 1 典 兩 1̄ 典 ——→
兩 1 典 兩 0̄ 典 n
→ 2
1 ⫺l
兩1典兩2典 兩 1 典 兩 2̄ 典 兩 1 典 兩 1̄ 典 兩1典兩1典 兩 j̄ 典 ⫽ n/2
丢 l⫽1
n
e 2 ␲ i jk l 3 兩 k l 典 . 共27兲
3 k l ⫽0

兩2典兩0典 兩 2 典 兩 0̄ 典 兩 2 典 兩 1̄ 典 兩2典兩1典
If the condition l⬍n is satisfied, we get
兩2典兩1典 兩 2 典 兩 1̄ 典 兩 2 典 兩 2̄ 典 兩2典兩2典
j j n⫹1⫺l j n⫹2⫺l jn
兩2典兩2典 兩 2 典 兩 2̄ 典 兩 2 典 兩 0̄ 典 兩2典兩0典. ⫽int⫹ ⫹ ⫹•••⫹ l
3 l 3 3 2
3
共23兲
⫽int⫹0 j n⫹1⫺l j n⫹2⫺l ••• j n , 共28兲
Universal quantum computation requires, in addition, a
measurement scheme in the computational basis. In our case, where ‘‘int’’ denotes an integer number. Then
von Neumann measurements distinguishing among three di-
rections 兩 0 典 , 兩 1 典 , 兩 2 典 are accomplished by connecting reso-
nant interactions from 兩 1 典 , 兩 2 典 to states 兩 3 典 , 兩 4 典 , respec-
兩 j̄ 典 ⫽
1
3 n/2
n
丢 l⫽1
2

k l ⫽0
冋兺 e 2 ␲ ik l [0 j n⫹1⫺l j n⫹2⫺l ••• j n ] 兩 k l 典 . 册
tively. Fast decay of excited optical levels through separated 共29兲
polarization channels allows us to discriminate between oc- This product is written in an equivalent form as

冉兺 冊冉 兺 冊
cupation of levels 兩 1 典 , 兩 2 典 , when fluorescence is observed; 2 2
or level 兩 0 典 when nothing is observed. 1 2 ␲ ik 1 0 j n
兩 j̄ 典 ⫽ e 兩 k 1典 e 2 ␲ ik 2 0 j n⫺l j n 兩 k 2 典
As a final remark, it should be stated that a conditional 3 n/2 k 1 ⫽0 k 2 ⫽0
modular addition 兩 j 丣 i 典 , as the conditional operation be-
tween qutrits, is defined instead of a modular difference op-
eration, just by adjusting the conditional phase shift in Eq.
共20兲 to
••• 冉兺 2

k n ⫽0
冊
e 2 ␲ ik n 0 j 1 j 2 ••• j n 兩 k n 典 . 共30兲

If we expand the summation for each factor, for example, in

兩 1 典 m 兩 1 典 n →e 2i ␲ /3兩 1 典 m 兩 1 典 n ,
the last term we have
兩 1 典 m 兩 2 典 n →e 4i ␲ /3兩 1 典 m 兩 2 典 n ,
3⫺1

兺
k n ⫽0
e 2 ␲ ik n (0 j 1 j 2 ••• j n ) 兩 k n 典
兩 2 典 m 兩 1 典 n →e 4i ␲ /3兩 2 典 m 兩 1 典 n ,
1
兩 2 典 m 兩 2 典 n →e 2i ␲ /3兩 2 典 m 兩 2 典 n . ⫽ 共 兩 0 典 ⫹e 2 ␲ i(0 j 1 j 2 ••• j n ) 兩 1 典 ⫹e 4 ␲ i(0 j 1 j 2 ••• j n ) 兩 2 典 )
冑3
IV. THE FOURIER TRANSFORM FOR n QUTRITS 1 2 ••• j /3n )
⫽ 共 兩 0 典 ⫹e 2 ␲ i( j 1 /3⫹ j 2 /3 n 兩1典
A natural extension of the previous sections is to consider 冑3
the analysis of the general protocol for the quantum Fourier 2 ••• j /3n )
transform for a system of n qutrits. We can benefit, of course, ⫹e 4 ␲ i( j 1 /3⫹ j 2 /3 n 兩 2 典 ). 共31兲

062313-5
KLIMOV et al. PHYSICAL REVIEW A 67, 062313 共2003兲

This state is generated starting with the application of a Fou- 2

rier transform on the first qutrit, 兺
k n ⫽0
e 2 ␲ ik n (0 j 2 j 3 ••• j n ) 兩 k n 典

F „3…
1 兩 j 1 典 兩 j 2 典 , 兩 j 3 典 ••• 兩 j n 典 1
⫽ 共 兩 0 典 ⫹e 2 ␲ i(0 j 2 j 3 ••• j n ) 兩 1 典 ⫹e 4 ␲ i(0 j 2 j 3 ••• j n ) 兩 2 典 )
1 冑3
⫽ 共 兩 0 典 ⫹e 2 ␲ i j 1 /3兩 1 典 ⫹e 4 ␲ i j 1 /3兩 2 典 ) 兩 j 2 典 , 兩 j 3 典 ••• 兩 j n 典 ,
冑3 1 2 ⫹•••⫹ j /3n⫺1 )
⫽ 共 兩 0 典 ⫹e 2 ␲ i( j 2 /3⫹ j 3 /3 n 兩1典
and after applying conditional phase transformations on this 冑3
qutrit state, conditioned to the initial state of remaining 2 ⫹•••⫹ j /3 n⫺1 )
兩 j 2 典 , 兩 j 3 典 , . . . , 兩 j n 典 qutrit states, ⫹e 4 ␲ i( j 2 /3⫹ j 3 /3 n 兩2典) 共34兲
is generated by applying a Fourier transform on the qutrit
1 兩 j 1 典 兩 j 2 典 兩 j 3 典 ••• 兩 j n 典
R n1 •••R 31R 21F (3) 兩 j 2 典 and after applying conditional phase operations on this
qutrit state, conditioned to the state of the remaining
1 2 ••• j /3n ) 兩 j 3 典 , . . . , 兩 j n 典 qutrit states.
⫽ 共 兩 0 典 ⫹e 2 ␲ i( j 1 /3⫹ j 2 /3 n 兩1典
冑3
V. SUMMARY
2 ••• j /3n )
⫹e 4 ␲ i( j 1 /3⫹ j 2 /3 n 兩 2 典 ) 兩 j 2 典 兩 j 3 典 ••• 兩 j n 典 . We have described a physical implementation of a univer-
sal qutrit quantum computer based on trapped ions. The logi-
The conditional phase is given by cal states of qutrits are codified into the electronic levels of

冉 冊 冉 冊
trapped ions. We have shown how to implement an arbitrary
4 ␲ 8 ␲ (1) 2 ␲ 4 ␲ one-qutrit gate. Besides, we have built a two-qutrit gate,
R 21⫽ P 21
(2)
, P 21 , , 共32兲 which is decomposed into a quantum Fourier transform and a
32 32 32 32 phase-shift gate. The main result of this work is that the same

冉 冊 冉 冊
kind of physical setup used for a qubit quantum computer is
4 ␲ 8 ␲ (1) 2 ␲ 4 ␲ suitable for a qutrit quantum computer, which gives an ex-
R 31⫽ P 31
(2)
, P 31 , , ponential increase of the available Hilbert space for the same
33 33 33 33
amount of physical resources. In principle, this scheme also
allows for entanglement distributions between qutrits allo-
.... cated in widely distant nodes of a quantum network.

ACKNOWLEDGMENTS
R k1 ⫽ P k1
(2) (1)
P k1 ,
We acknowledge A. Delgado for stimulating comments.
(j )
P k1k ⫽ 冉 2 j k␲ 4 j k␲
3k
, k .
3
冊 共33兲
This work was partially supported by grants from CONA-
CYT 共Mexico兲 and CONICYT 共Chile兲 ‘‘Información Cuán-
tica e Interacciones Colectivas,’’ Milenio Grant No. ICM
P99-135F and FONDECYT Grant No. 1010010. R.G. thanks
In the same way, the state to MECESUP Grant No. USA0108.

关1兴 M.A. Nielsen and I.L. Chuang, Quantum Computation and
Quantum Information 共Cambridge University Press, Cam-
bridge, 2000兲.
关2兴 C. Monroe, D.M. Meekhof, B.E. King, W.M. Itano, and D.J.
Wineland, Phys. Rev. Lett. 75, 4714 共1995兲.
关3兴 J.I. Cirac and P. Zoller, Phys. Rev. Lett. 74, 4091 共1995兲.
关4兴 Q.A. Turchette et al., Phys. Rev. Lett. 81, 3631 共1998兲.
关5兴 I.L. Chuang, N. Gershenfeld, and M. Kubinec, Phys. Rev. Lett.
80, 3408 共1998兲.
关6兴 R. Laflamme, E. Knill, W.H. Zurek, P. Catasti, and S.V.S.
Mariappan, Philos. Trans. R. Soc. London, Ser. A 356, 1941
共1998兲.
关7兴 D.G. Cory, M.D. Price, W. Maas, E. Knill, R. Laflamme, W.H.
Zurek, T.F. Havel, and S.S. Somaroo, Phys. Rev. Lett. 81, 2152
共1998兲.
关8兴 D. Loss and D.P. DiVincenzo, Phys. Rev. A 57, 120 共1998兲.
关9兴 B.E. Kane, Nature 共London兲 393, 133 共1998兲.
关10兴 G. Burkard, D. Loss, and D.P. DiVincenzo, Phys. Rev. B 59,
2070 共1999兲.
关11兴 Carlton M. Caves and Gerard J. Milburn, Opt. Commun. 179,
439 共2000兲.
关12兴 D. Bruss and C. Macchiavello, Phys. Rev. Lett. 88, 127901
共2002兲; N.J. Cerf, M. Bourennane, A. Karlsson, and N. Gisin,
ibid. 88, 127902 共2002兲.
关13兴 G. Alber, A. Delgado, N. Gisin, and I. Jex, J. Phys. A 34, 42
共2001兲.
关14兴 Vivien M. Kendon, Karol Zyczkowski, and William J. Munro,
Phys. Rev. A 66, 062310 共2002兲.
关15兴 Miloslav Dusek, Opt. Commun. 199, 161 共2001兲.
关16兴 Kenneth A. Dennison and William K. Wootters, Phys. Rev. A
65, 010301 共2002兲; P. Rungta, V. Buzek, C.M. Caves, M. Hil-
lery, and G.J. Milburn, ibid. 64, 042315 共2001兲.
关17兴 Nicolas J. Cerf, Serge Massar, and Stefano Pironio, Phys. Rev.
Lett. 89, 080402 共2002兲.

062313-6
QUTRIT QUANTUM COMPUTER WITH TRAPPED IONS
关18兴 Stephen D. Bartlett, Hubert de Guise, and Barry C. Sanders,
Phys. Rev. A 65, 052316 共2002兲.
关19兴 R.T. Thew, K. Nemoto, A.G. White, and W.J. Munro, Phys.
Rev. A 66, 012303 共2002兲.
关20兴 Jan Bouda and Vladimir Buzek, J. Phys. A 34, 4301 共2001兲.
关21兴 Caslav Brukner, Marek Zukowski, and Anton Zeilinger, Phys.
Rev. Lett. 89, 197901 共2002兲.
关22兴 Alipasha Vaziri, George Weihs, and Anton Zeilinger, Phys.
Rev. Lett. 89, 240401 共2002兲.
062313-7
PHYSICAL REVIEW A 67, 062313 共2003兲
关23兴 C. Raab, J. Bolle, H. Obetrst, J. Eschner, F. Schmidt-Kaler, and
R. Blatt, Appl. Phys. B: Lasers Opt. 67, 683 共1998兲.
关24兴 D. Reiss, A. Lindner, and R. Blatt, Phys. Rev. A 54, 5133
共1996兲.
关25兴 Hilmart Oberst, diplomarbeit, Innsbruck University, 1999.
关26兴 N. Vilenkin and A. Klimyk, Representation of Lie Group and
Special Functions 共Kluwer Academic, Dordrecht, 1991兲, Vols.
1–3.