a

r
X
i
v
:
q
u
a
n
t
-
p
h
/
0
3
0
1
0
4
5

v
1



1
1

J
a
n

2
0
0
3
Continuous-Variable Quantum Teleportation with a Conventional Laser
Mikio Fujii

Systems Integration Technology Center, Toshiba Corporation,

3-22 Katamachi, Fuchu, Tokyo, 183-8512, Japan
We give a description of balanced homodyne detection (BHD) using a conventional laser as a
local oscillator (LO), where the laser eld outside the cavity is a mixed state whose phase is com-
pletely unknown. Our description is based on the standard interpretation of the quantum theory
for measurement, and accords with the experimental result in the squeezed state generation scheme.
We apply our description of BHD to continuous-variable quantum teleportation (CVQT) with a
conventional laser to analyze the CVQT experiment [A. Furusawa et al., Science 282, 706 (1998)],
whose validity has been questioned on the ground of intrinsic phase indeterminacy of the laser eld
[T. Rudolph and B.C. Sanders, Phys. Rev. Lett. 87, 077903 (2001)]. We show that CVQT with a
laser is valid only if the unknown phase of the laser eld is shared among senders LOs, the EPR
state, and receivers LO. The CVQT experiment is considered valid with the aid of an optical path
other than the EPR channel and a classical channel, directly linking between a sender and a receiver.
We also propose a method to probabilistically generate a strongly phase-correlated quantum state
via continuous measurement of independent lasers, which is applicable to realizing CVQT without
the additional optical path.
PACS numbers: 03.67.Hk, 03.65.Ud, 42.50.-p
Quantum teleportation is a method to move quantum
states from a sender Alice to a receiver Bob by the
aid of entanglement. The original protocol [1] has later
been extended to continuous-variable quantum teleporta-
tion (CVQT) using a two-mode squeezed state and bal-
anced homodyne detection (BHD) [2]. Although experi-
mental demonstration of CVQT was reported as the rst
achievement of unconditional quantum teleportation [3],
there has been controversy over its validity on the ground
of intrinsic phase indeterminacy of the laser eld [4, 5].
The laser eld is often assumed to be a coherent state
having a xed phase, but the steady-state solution of
the master equation in the quantum theory of the laser
shows that the phase of the laser eld inside the cavity
is completely unknown in operation well above threshold
[4, 5, 6, 7].
In this Letter, we rst discuss a description of the laser
eld outside the cavity in line with the previous discus-
sions [4, 5]. We then analyze BHD using a laser as a local
oscillator (LO), and give a physically appropriate descrip-
tion of BHD based on the standard interpretation of the
quantum theory for measurement. We apply our descrip-
tion of BHD to CVQT with a laser to analyze the CVQT
experiment [3]. Finally, we propose a method to proba-
bilistically produce a strongly phase-correlated quantum
state via continuous measurement of independent lasers,
which is applicable to realizing CVQT without an optical
path between Alice and Bob for sharing the same laser
eld.
By employing the input-output theory [6], van Enk
and Fuchs nd that the laser eld outside the cavity is a
continuous-mode mixed state being quite similar to the
steady-state eld inside the cavity, whose phase is also
completely unknown. They then go on to express the
eld state in terms of noncontinuous operators given by
them, and discuss its coherency [5].
The denition of noncontinuous operators is based on
the narrow bandwidth approximation (NBA) B
0
,
where
0
is the central frequency of the bandwidth
B. The annihilation operator part of the continuous-
mode electric eld under NBA is written as

E
+
(z, t)
iE
0

n

n
(t z/c) c
n
, where E
0
[ h
0
/(2
0
cA)]
1/2
,
{
n
(t)} are basis functions giving a prole of the fun-
damental modes, and c
n
is a noncontinuous annihilation
operator [8].
In Ref. [5], basis functions are specically given as

n
(t) T
1/2
exp(i
0
t)(t/T n) where (t) is the
rectangle function, which are Fourier transformed into

n
() = [T/(2)]
1/2
exp[i(
0
)nT]sinc[(
0
)T/2].
But these basis functions do not dene correct non-
continuous operators, because T must be taken to the
limit T 0+ in order to satisfy the completeness
relation

n

n
(t)

n
(t

) = (t t

) for t = t

when
|t/T n| < 1/2 and |t

/T n| < 1/2, whereas T must

satisfy T 4/B 4/
0
(> 0) since (B/2)T/2
is necessary under NBA to reconstruct the given

n
(t) by the inverse Fourier transform, i.e.,
n
(t)
(2)
1/2
_
0+B/2
0B/2
d
n
() exp(it).
Hence, the laser eld outside the cavity cannot be ex-
pressed as Ref. [5], unless we impose another condition
|t t

| > T on t, t

whenever t = t

. It has not yet been

proved, however, whether we may justiably attribute
this added condition to some physical restriction, e.g.,
time resolution of photodetectors, or not. For simplicity,
we treat the laser eld outside the cavity as the single
traveling-wave mode throughout this Letter.
As long as the photon number operator well represents
an observable for an ecient photodetector lacking single
photon resolution [9, 10], we may regard a

l
a
s
+ a
l
a

s
as an
2
observable for BHD, where a
l
and a
s
are annihilation op-
erators for the LO eld and the signal eld, respectively
[11]. If the signal eld |
s
satises r
_
|

Xs()
2
|
s

_
| a

s
as|
s
, which holds when the intensity of the LO
eld is extremely larger than that of the signal eld, this
observable satises
( a

l
a
s
+ a
l
a

s
)|re
i

l
|
s
r

X
s
()|re
i

l
|
s
, (1)
where

X() ae
i
+ a

e
i
and |re
i
is the coherent
state in polar coordinates [6]. According to the standard
interpretation of the quantum theory [12], Eq. (1) implies
if we obtain the measurement outcome rx in one trial of
BHD with the prior knowledge of r, |
s
instantaneously
reduces to |x,
s
satisfying

X
s
()|x,
s
= x|x,
s
.
Since r of the laser eld is measurable beforehand, we
may dene the measurement operator [13] for BHD as

M(x, r, )

1
2
|re
i

l
|x,
s
x, |
s
re
i
|
l
, (2)
where |x, is the quadrature eigenstate written as
|x, = (2)

1
4
e

x
2
4
exp
_
xe
i
a

1
2
e
i2
a
2
_
|0. (3)
|x, satises the orthonormalization condition
x
1
, |x
2
, = (x
1
x
2
) and the completeness
relation
_
+

dx |x, x, | = 1 on x. Since the

coherent state also satises the completeness rela-
tion [6], Eq. (2) satises the completeness relation
_
+

dx
_

0
rdr
_
2
0
d

M

(x, r, )

M(x, r, ) = 1.
Considering that we cannot distinguish between |x,
1

and |x,
2
(
1
=
2
) by measurement results due to in-
trinsic phase indeterminacy of the laser eld [4, 7], the
probability of obtaining the measurement outcome x = x
with the prior knowledge of r is
P( x) =
_

0
rdr
_
2
0
d Tr
_

M( x, r, )
o

M

( x, r, )
_
, (4)
and the density operator after the measurement is
= P( x)
1
_

0
rdr
_
2
0
d

M( x, r, )
o

M

( x, r, ), (5)
where
o
is the density operator before the measurement.
We will denote the procedure described above the
observable-based projection method (OBPM) in the rest
of this Letter. Note that above discussion is not based on
the assumption that the laser eld is the coherent state
(partition ensemble fallacy [4, 14]). It is the property
of the observable for BHD that approximately projects
the strong laser eld of the LO mode onto the coherent
state after the measurement. On the contrary, the num-
ber states in the LO mode cannot be eigenstates of the
observable for BHD, because |n = |n 1 even in the
limit n + due to their rigid orthogonality.
As an example of BHD, we will calculate P( x) in the
squeezed light generation scheme [15] by OBPM. In the
scheme, the same laser source is used for supplying the
LO eld, and pumping the nonlinear medium to generate
the squeezed state. The density operator of the system
before the measurement is

o
=
_
2
0
d
2
|r
o
e
i(+)

l
|0, se
i2

s
0, se
i2
|
s
r
o
e
i(+)
|
l
, (6)
where is the unknown phase of the pump eld,
is the phase delay by a controllable phase shifter, and
|0,

S()|0 is the squeezed vacuum state [6]. The
unknown phase of the squeezed state is 2 instead of ,
because frequency of the pump eld is doubled by sec-
ond harmonic generation before the eld enters an optical
parametric oscillator. By using Eqs. (4), (6), orthogonal-
ity approximation of the coherent state |re
i
|r
o
e
io
|
2

(/r
o
)(r r
o
)(
o
) in the limit r
o
+ derived
from lim
0+
exp[t
2
/(4)]/(2

) = (t), and the rela-

tion
x, |0, se
i2()
=

n=0
x, |nn|0, se
i2()

n=0
(2)

1
4
(2
n
n!)

1
2
H
n
_
x

2
_
e

x
2
4
in
[2
n
n! cosh(s)]

1
2
[e
i2()
tanh(s)]
n
2
H
n
(0)
= x, |0, s
where H
n
(x) are Hermite polynomials, we nd P( x) =
| x, |0, s|
2
, which agrees with the experimental result
of Ref. [15].
Next, we will apply OBPM to CVQT with a laser
[2, 4]. In the measurement step by Alice, the probability
of obtaining x
1
in BHD1 and x
2
in BHD2 is P( x
1
, x
2
)
_

0
r
1
dr
1
_
2
0
d
1
_

0
r
2
dr
2
_
2
0
d
2
Tr{

M
2

M
1

I

M

2
} and
the density operator after the measurement is
II

P
1
( x
1
, x
2
)
_

0
r
1
dr
1
_
2
0
d
1
_

0
r
2
dr
2
_
2
0
d
2

M
2

M
1

I

M

2
,
where

M
j

1/2
|r
j
e
ij

lj
| x
j
,
j

sj
x
j
,
j
|
sj
r
j
e
ij
|
lj
(j =
1, 2).
I
is the density operator of the total system
before the measurement written as

I
=
_
2
0
d
2
|r
o
e
i

l1
|r
o
e
i(+

2
)

l2
|e
i2

1,2
|r
o
e
i

l3

in
r
o
e
i
|
l3
e
i2
|
1,2
r
o
e
i(+

2
)
|
l2
r
o
e
i
|
l1
, (7)
where the modes l1, l2 are for LOs of BHD1,2 in Al-
ice, l3 for LO in Bob, is the unknown phase of the
pump eld,
in
is an arbitrary density operator sup-
plied by a third party Victor to Alice, and |e
i2

1,2

_
1
2
exp(e
i2
a

1
a

2
)|0
1
|0
2
is a two-mode squeezed
state [6] as the EPR state. Again, the unknown phase in
the modes 1, 2 is 2 instead of . (See Fig. 1 in Ref. [4].)
By using Eq. (3) and a
s1
= ( a
in
a
1
)/

2, a
s2
=
( a
in
+ a
1
)/

2 where the modes s1, s2 are for the

signal eld of BHD1,2, the quadrature eigenstates of
the modes s1, s2 are written in the modes in, 1 as
3
| x
1
,
s1
| x
2
, +

2

s2
= [exp(
||
2
2
)/

2] exp[( a

in

1
)e
i
+ a

in
a

1
e
i2
]|0
in
|0
1
, where ( x
1
+
i x
2
)/

2. By using this and the relation for bosons

exp( a) exp( a

) = exp(

) exp( a

) exp( a) de-
rived from the Baker-Hausdor formula [16], we nd
x
1
, |
s1
x
2
, +

2
|
s2
|e
i2

1,2
= e

||
2
2
_
1
2
2
exp(e
i
a

2
)
_

n=0

n
|n
2
n|
in
_
exp(

e
i
a
in
). (8)
With orthogonality approximation of the coherent state,
we nd
II
includes Eq. (8). Ideal quantum teleporta-
tion is possible only when = 1, where a two-mode
squeezed state is maximally entangled [2]. Eq. (8)
shows that the unitary transform

U
2
applied by Bob
to the mode 2 in the reconstruction step must satisfy

U
2
|
=1
exp(
||
2
2
) exp(e
i
a

2
) exp(

e
i
a
2
) = const.,
because

n=0
|n
2
n|
in
transfers a state of the mode in
to the mode 2 with absolute precision.
The necessary condition for

U
2
is then found to be

U
2
|
=1
= exp(e
i
a

e
i
a
2
), which means Bob
needs not only the measurement results by Alice but
also the unknown phase of Alices LO elds to perform

U
2
. Hence, to share between Alice and Bob by a certain
means is essential to realizing CVQT with a laser.
In the experiment [3], Bob obtains from the LO at
hand directly connected to Alices LOs and the pump
eld of a two-mode squeezed state. If Bob performs
the unitary transform

U
2,l3
(, ) exp[(/r
o
)( a
l3
a

l3
a
2
)] after he obtains , the density operator of the
total system
III


U
2,l3
(, )
II

U

2,l3
(, ) becomes

III
P
1
( x
1
, x
2
)

_
2
0
d
2
|r
o
e
i

l1
| x
1
,
s1
|r
o
e
i(+

2
)

l2
| x
2
, +

2

s2
|r
o
e
i

l3

T
2,in
(, , )
in

T

2,in
(, , )r
o
e
i
|
l3
x
2
, +

2
|
s2
r
o
e
i(+

2
)
|
l2
x
1
, |
s1
r
o
e
i
|
l1
, (9)
where a

|r
o
e
i
r
o
e
i
|r
o
e
i
in the limit r
o
+
and

T
2,in
is dened as

T
2,in
(, , ) e

||
2
2
(1
2
)
_
1
2
2
exp(

e
i
a
2
)
_

n=0

n
|n
2
n|
in
_
exp(

e
i
a
in
), (10)
which corresponds to the transfer operator in Ref. [17]
from the mode in to the mode 2.
Eqs. (9) and (10) clearly show that in the special case
= 1

T
2,in
is independent of the unknown phase where
ideal quantum teleportation is realized, while in the usual
case 0 < 1

T
2,in
is dependent on the unknown phase
where the reconstructed density operator in the mode
2 is distorted from
in
.
We will subsequently discuss generation of a strongly
phase-correlated quantum state necessary in CVQT by
measuring two independent laser elds

o
=
_
2
0
d
a
2
_
2
0
d
b
2
|r
a
e
ia

a
|r
b
e
i
b

b
r
b
e
i
b
|
b
r
a
e
ia
|
a
. (11)
In the case of BHD, since the observ-
able satises ( a

b + a

)|r
a
e
i
|r
b
e
i()

2r
a
r
b
cos()|r
a
e
i
|r
b
e
i()
(0 ) in the
limit r
a
, r
b
+, the measurement operator for
OBPM may be dened as

M(cos(), r
1
, r
2
, )

1
|r
1
e
i
|r
2
e
i()
r
2
e
i()
|r
1
e
i
|. Then, the
density operator after the measurement becomes
=
1
2
_
2
0
d
2
|r
a
e
i

a
|r
b
e
i(+)

b
r
b
e
i(+)
|
b
r
a
e
i
|
a
+
1
2
_
2
0
d
2
|r
a
e
i

a
|r
b
e
i()

b
r
b
e
i()
|
b
r
a
e
i
|
a
, i.e.,
BHD does not determine a unique phase dierence
of two lasers except exactly when cos() = 1 with
negligible probability. Hence, the generated quantum
state by BHD is not applicable to CVQT to share the
unknown phase of the laser eld.
But if we perform continuous measurement [9, 16, 18]
presented in Fig. 1, a unique phase dierence of two lasers
is chosen with non-zero probability. In Fig. 1, two-level
atom beams are used as probes to ensure that photoab-
sorption occurs at most one time within the innitesimal
atom-eld interaction time , which is not feasible by a
present photodetector lacking single photon resolution in
the strong eld [9, 10].
Given that the total photoabsorption (quantum jump)
occurs either in the mode c or d at times t
1
, t
2
, . . . , t
s
in
the time interval [0, t] with no absorption between these
times, the conditional probability that photoabsorption
occurs p, q(=sp) times in the mode c, d, respectively, is
P(t; p, q|s)
_
s
p
_
Tr{ c
p
d
q
e
R( c

c+

d)t
(0)e
R( c

c+

d)t
d
q
c
p
}

p+q=s
_
s
p
_
Tr{ c
p
d
q
e
R( c

c+

d

d)t
(0)e
R( c

c+

d

d)t
d
q
c
p
}
=
1
_
s
p
_
B(p +
1
2
, q +
1
2
), (12)
and the density operator becomes
(t; p, q)
c
p

d
q
e
R( c

c+

d)t
(0)e
R( c

c+

d)t

d
q
c
p
Tr{ c
p
d
q
e
R( c

c+

d

d)t
(0)e
R( c

c+

d

d)t
d
q
c
p
}
=

B(p +
1
2
, q +
1
2
)
_
2
0
d
a
2
_
2
0
d
b
2
sin
2p
_
a
b
2
_
cos
2q
_
a
b
2
_
|r
t
e
ia

a
|r
t
e
i
b

b
r
t
e
i
b
|
b
r
t
e
ia
|
a
, (13)
where (0) is Eq. (11) with r
a
=r
b
(r
o
), e
R

d(p)

e
R c

c(q)
1, B(x, y) is the beta function, r
t
r
o
e
Rt
,
R g
2
/2, and g is the atom-eld coupling constant
[16].
The proposed continuous measurement is valid when
(

2s r
o
g)
1
. In Eq. (13), we nd that atoms simul-
taneously intersecting the output modes with no absorp-
tion (null measurement) damp both laser elds, leaving
4

































a
b

c
d

g
e
o

Laser
Laser
Two-Level Atoms
FIG. 1: Experimental setup for continuous measurement of
two independent laser elds. a,

b and c,

d are annihilation op-
erators for the input and output modes of a 50/50 beamsplit-
ter, respectively, satisfying c = ( a

b)/

2,

d = ( a +

b)/

2.
Two-level atoms resonant with the laser elds are all prepared
in the ground state beforehand, and go across the output elds
one by one at regular intervals.
phase correlation between the elds unchanged. The ab-
sorption rate is assumed to be quite high, where t is much
smaller than the dynamical time scale of an individual
laser.
For s 1, the distribution of the phase dierence of
states in the integrand of Eq. (13) has a peak at |
a

b
| = when p = s, or at |
a

b
| = 0 when p = 0.
Since Eq. (12) has peaks at p = 0, s, the probability of
obtaining Eq. (13) with p = 0, s is not negligible.
The photon number distribution of the mode c,
P
c
(m) m|Tr
d
{ (t; p, q)}|m
c
, is found to be
P
c
(m) = e
2rt
2
_
2r
t
2
_
m
m!
B(m + p +
1
2
, q +
1
2
)
B(p +
1
2
, q +
1
2
)

1
F
1
(q +
1
2
; m + p + q + 1; 2r
t
2
), (14)
where
1
F
1
(; ; z) is the conuent hypergeometric func-
tion of the rst kind. P
d
(n) is easily obtained by replac-
ing m with n and interchanging p q in Eq. (14). Fig.
2 is for P
c
(m), P
d
(n).
When p = 0, s with s 1, the generated quantum
state is applicable to CVQT as a means to share the
unknown phase of the laser eld between Alice and Bob,
though the phase correlation formed after the continuous
measurement will slowly be broken by the phase diusion
eect of lasers.
The famous experiment for interference of two inde-
pendent lasers by Peegor and Mandel [19], where weak
laser elds were mixed by beamsplitters and all the out-
put elds were continuously measured by photomulti-
pliers, should carefully be reviewed in terms of phase-
correlated quantum state generation by measurement.
In conclusion, we have pointed out that the eld state
outside the laser cavity is not equivalent to the expression
in terms of noncontinuous operators given in Ref. [5]. We
have presented OBPM for BHD to analyze CVQT with
a conventional laser whose phase is completely unknown.
500 1000 1500 2000
m
0.0025
0.005
0.0075
0.01
0.0125
Pc(m)
p=100,q=0
500 1000 1500 2000
n
0.05
0.1
0.15
0.2
Pd(n)
p=100,q=0
500 1000 1500 2000
m
0.0025
0.005
0.0075
0.01
0.0125
Pc(m)
p=0,q=0
500 1000 1500 2000
n
0.0025
0.005
0.0075
0.01
0.0125
Pd(n)
p=0,q=0
FIG. 2: Photon number distributions from Eq. (14) for p = s
with rt
2
= 10
3
. Given that s = 100, the probability Eq. (12)
for p = 0 or p = 100 is about 11.3%. If the Monte Carlo
wave-function procedure [18] is performed, gradual decay of
rt due to null measurement shall be seen besides the above
distribution change. Pc(m) approaches a Poisson distribution
as s becomes large.
CVQT is found to be possible only if the unknown phase
of the laser eld is shared among Alices LOs, the EPR
state, and Bobs LO by a certain means. The demon-
strated experiment for CVQT [3] is valid, but needs an
optical path other than the EPR channel and a classi-
cal channel allowed to use in the teleportation protocols
[1, 2] to share the unknown phase of the same laser eld
between Alice and Bob. We have proposed a method
to probabilistically generate a strongly phase-correlated
quantum state via continuous measurement of indepen-
dent lasers, which is applicable to realizing CVQT with-
out the additional optical path.
The author is greatly indebted to Kenji Ohkuma,
Kouichi Ichimura, and Noritsugu Shiokawa for intensive
and fruitful discussions on the subject of this Letter.
The author also acknowledges useful discussions with Mio
Murao, Hirofumi Muratani, and Toshiaki Iitaka.

Email address: mikio.fujii@toshiba.co.jp

[1] C.H. Bennett et al., Phys. Rev. Lett. 70, 1895 (1993).
[2] S.L. Braunstein and H.J. Kimble, Phys. Rev. Lett. 80,
869 (1998).
[3] A. Furusawa et al., Science 282, 706 (1998).
[4] T. Rudolph and B.C. Sanders, Phys. Rev. Lett. 87,
077903 (2001).
[5] S.J. van Enk and C.A. Fuchs, Phys. Rev. Lett. 88, 027902
(2002).
[6] D.F. Walls and G.J. Milburn, Quantum Optics (Springer,
Berlin, 1994).
[7] K. Mlmer, Phys. Rev. A 55, 3195 (1997).
[8] K.J. Blow et al., Phys. Rev. A 42, 4102 (1990).
[9] H.J. Carmichael, An Open Systems Approach to Quan-
tum Optics (Springer, Berlin, 1993).
[10] C.W. Gardiner and P. Zoller, Qauntum Noise, 2nd Ed.
5
(Springer, Berlin, 2000), Chap. 8.
[11] S.L. Braunstein, Phys. Rev. A 42, 474 (1990).
[12] J.J. Sakurai, Modern Quantum Mechanics, Rev. Ed.
(Addison-Wesley, MA, 1994), Chap. 1.
[13] M.A. Nielsen and I.L. Chuang, Quantum Computa-
tion and Quantum Information (Cambridge Univ. Press,
Cambridge, 2000), Chap. 2.
[14] P. Kok and S.L. Braunstein, Phys. Rev. A 61, 042304
(2000).
[15] S. Schiller et al., Phys. Rev. Lett. 77, 2933 (1996); G.
Breitenbach et al., Nature 387, 471 (1997).
[16] M.O. Scully and M.S. Zubairy, Quantum Optics (Cam-
bridge Univ. Press, Cambridge, 1997).
[17] H.F. Hofmann et al., Phys. Rev. A 62, 062304 (2000).
[18] J. Dalibard et al., Phys. Rev. Lett. 68, 580 (1992); K.
Mlmer et al., J. Opt. Soc. Am. B 10, 524 (1993).
[19] R.L. Peegor and L. Mandel, Phys. Rev. 159, 1084
(1967).