You are on page 1of 5

a

r
X
i
v
:
1
3
1
1
.
3
8
5
5
v
1


[
q
u
a
n
t
-
p
h
]


1
5

N
o
v

2
0
1
3
Cauchy-Schwarz inequality and particle entanglement
T. Wasak
1
, P. Sza nkowski
1
, P. Zi n
2
, M. Trippenbach
1
and J. Chwede nczuk
1
1
Faculty of Physics, University of Warsaw, ul. Ho za 69, PL00681 Warszawa, Poland
2
National Centre for Nuclear Research, ul. Ho za 69, PL-00-681 Warsaw, Poland
The Glauber-Sudarshan P-representation is used in quantum optics to distinguish between
semi-classical and genuinely quantum electromagnetic elds. We employ the analog of the P-
representation to show that the violation of the Cauchy-Schwarz inequality for the second-order
correlation function is a proof of entanglement between identical massive bosons. The presented
derivation is valid both in systems with xed and uctuating number of particles. Thanks to the
recent advances in techniques of detecting positions of separate particles, the violation of the Cauchy-
Schwarz inequality can be used as a simple entanglement criterion in various many-body quantum
systems.
Although the foundations of quantum and classical
physics are much dierent, it is often dicult to tell
whether a particular object is quantum. An example of
non-classicality is, according to the Schrodinger equation,
the possibility for particles to exist in superpositions of
quantum states. The most prominent manifestation of
such superposition is the Young double-slit experiment
for massive particles, which conrms the wave aspect of
the quantum world. For optical waves it is the opposite
the non-classical is the quantized electromagnetic eld,
consisting of individual photons. The challenging ques-
tion whether the pulse of light is quantized or not, was
one of the key points which triggered the development
of quantum optics. One way to address this problem is
to express the density matrix of the system in the basis
of coherent states [ using the Glauber-Sudarshan P-
representation [1, 2], which for a single-mode case reads
=
__
dd

P(,

)[[. (1)
Whether the state of light described by is quantum or
not depends on the properties of the P-representation.
When P is either a
non negative function of and

or (2a)
convex sum of Dirac delta distributions, (2b)
then is a statistical mixture of coherent states, which
resemble classical waves described by Maxwell equations.
On the other hand, when P does not posses the aforemen-
tioned features, description of exceeds the laws of classi-
cal electromagnetism and the eld is said to be quantum.
Once the electromagnetic eld is quantized, photons
and other particles can be treated on a more equal foot-
ing. It is then reasonable to ask what are the correla-
tions between individual particles and in this context the
concept of entanglement emerges [3, 4]. The possibility
for particles to be entangled, which is a purely quantum
phenomenon, has rather dramatic consequences. The
quantumness of entanglement is underlined by the word
paradox, which is often used to describe some highly
counter-intuitive phenomena involving entangled parti-
cles. Among these is are the the Einstein-Podolsky-
Rosen (EPR) paradox [5] and the related Schrodingers
cat problem. Apart from fundamental aspects, systems
of entangled particles have applications in quantum infor-
mation [6], teleportation [7, 8] or ultra-precise metrology
[9, 10]. Entanglement is also believed to be the crucial in-
gredient for the extreme eciency of the energy transfer
in the process of photosynthesis [11].
Much as a fascinating consequence of quantum me-
chanics, entanglement is also elusive. It is not simple
to entangle particles on demand, because this requires
complicated experimental strategies. Entanglement is
dicult to protect from the destructive inuence of the
environment, which inevitably leads to decoherence [12
16]. Finally, even if a non-classical state reaches detectors
rather intact, it is often not clear which quantity should
be measured to witness entanglement.
This last diculty is related to a fundamental problem,
which is encountered already at the level of denition.
An entangled state is such that is not separable, mean-
ing that it cannot be written as a statistical mixture of
product states of N particles [1719]

sep
=

i
p
i

(i)
1
. . .
(i)
N
, (3)
where p
i
s are non-negative weights that add up to unity.
The consequence of this indirect denition is that to char-
acterize the entanglement we must always refer to some
bounds achievable by separable states. A good example
is the two-mode quantum interferometry, where a col-
lection of N qubits in state is used to determine an
unknown phase . If the precision of the parameter es-
timation is better than shot-noise = N
1/2
(the
smallest error reachable with separable states), then
is entangled [10]. However, in most cases the argument
cannot be reversed, just because we do not know what is
the entangled state.
In this work we show that, apart from its applications
in quantum optics, the analog of the P-representation
allows to formulate a simple criterion for entanglement
of identical massive bosons using the Cauchy-Schwarz in-
equality (CSI) for their second order correlation function,
2
which reads
(
(2)
(x, x

) =
_

(x)

(x

(x

(x)
_
. (4)
The average value of the above operator is the trace of
its product with the density matrix of the system, while

(x) is the bosonic eld operator.


We introduce two regions a and b having volumes V
a
and V
b
and a following set of three integrals of (4)
(
(2)
ij
=
_
Vi
dx
_
Vj
dx

(
(2)
(x, x

), (5)
where i and j are either a or b. The CSI relates the
integrated cross- and local correlations as follows
(
(
(2)
ab
_
(
(2)
aa
(
(2)
bb
1. (6)
Using the analog of the P-representation (1) we will now
demonstrate that that the violation of the CSI for iden-
tical massive bosons is a proof of particle-entanglement,
following the general scheme outlined above by rst
showing that the CSI cannot be violated for separable
states.
The discussion begins by stating that any separable
pure state of N identical bosons is a product of N iden-
tical single-particle orbitals [, i.e.
[
N
[
N
. (7)
The state (7) satises

(x) [
N
=

N(x) [
N1
, (8)
which is a xed-N counterpart of the property

c
(+)
(x) [ = (x) [ of a coherent state of the positive-
frequency part of the electromagnetic eld

c(x). In
Eq. (8), (x) is a single-particle function, which deter-
mines the spatial properties of the system.
The general separable state of identical bosons is a
statistical mixture of dierent states (7),
=
_
T [
N

N
[ T(), (9)
where T() is either a
non negative function of or (10a)
convex sum of Dirac delta distributions. (10b)
The symbol T denotes the measure of the integration
over the set of states [
N
. There is a direct analogy be-
tween the P-representation from Eq. (1) and T() from
Eq. (9). If the former does not satisfy the conditions
(2), then the electromagnetic eld is genuinely quantum.
Analogically for N indistinguishable bosons, if conditions
(10) are not fullled, then the density matrix cannot be
written as a statistical mixture (9) meaning that particles
are entangled.
Using the expression (9) and the property (8), we ob-
tain that the integrated second-order correlation between
the regions a and b reads
(
(2)
ab
= N(N 1)
_
TT()I
a
()I
b
(), (11)
where I
a/b
() =
_
V
a/b
dx[(x)[
2
. Applying the CSI for
integrals

_
Tf
a
() f
b
()

_
T[f
a
()[
2
_
T[f
b
()[
2
(12)
with f
a/b
() =
_
T() I
a/b
gives
(
(2)
ab
_

(
(2)
aa

(
(2)
bb
1, (13)
where

(
(2)
aa
= N(N 1)
_
T [T()[ I
2
a
() (14)
for a and analogously for b. Note that the condition (13)
is equivalent to (6) only when [T()[ = T() which is
true if (10) is satised, i.e. for all separable states. In
general, by combining (6) and (13) we obtain that
(

(
(2)
aa

(
(2)
bb
(
(2)
aa
(
(2)
bb
1, (15)
which shows that ( can reach values higher than unity.
This can happen only when either conditions (10) are not
satisies, i.e. [T()[ ,= T(), or when the state cannot
be represented in the form (9). Therefore, we conclude
that the violation of the inequality (6) is possible only
when is entangled.
The result (13) can be generalized to systems where
the number of massive particles is not xed. In such case,
assuming that the super-selection rule applies, which ex-
cludes coherences between states with dierent numbers
of particles, the density matrix of a separable state reads
=

N=0
p
N
_
T [
N

N
[ T
N
(), (16)
where p
N
is the probability for having N particles in the
system, while T
N
(), which depends on N, satises (10).
The CSI now involves the integrated second-order cor-
relation functions averaged with p
N
. The only modi-
cation to Eq. (13) is that in Eq. (11) N(N 1)T() is
replaced by
T
N
()

N
p
N
N(N 1)T
N
(). (17)
3
Note that T
N
() might be positive, even when for some
particular N, T
N
() is partially negative, meaning that
the system is entangled. From the point of view of in-
equality (6), the separable part of T
N
can overshadow
the entangled component.
To nd out which kind of entangled states the CSI is
sensitive to, we introduce the number-squeezing parame-
ter dened as a variance of the population imbalance op-
erator between the two regions,
2
=

n
2
_
n
2
. Here
n = n
a
n
b
, where n
a/b
=
_
V
a/b
dx

(x)

(x). This
parameter can be expressed in terms of the local- and
cross-correlations as follows

2
= n
a
+ n
b
(n
a
n
b
)
2
+(
(2)
aa
+(
(2)
bb
2(
(2)
ab
, (18)
where n
a/b
= n
a/b
. First, consider a balanced state
having n
a
= n
b
and (
(2)
aa
= (
(2)
bb
. The system is number
squeezed,
2
< n
a
+ n
b
, only when the correlation part
is negative, which according to Eq. (6) requires ( > 1.
Therefore, in this case, the number-squeezing is equiv-
alent to the violation of the CSI and signies particle
entanglement. If the state does not posses this symme-
try, then one cannot link the number squeezing with the
CSI, due to presence of the non-vanishing term (n
a
n
b
)
2
.
Thus the CSI criterion is more universal then the num-
ber squeezing, because it does not make any assumptions
about the regions a and b. Nevertheless, the relation
between the integrated correlation functions (5) and
2
from Eq. (18) is a strong suggestion that the violation of
the CSI is more likely to manifest in systems, where the
uctuations between the two regions are reduced rather
then enhanced. Indeed, a maximally entangled two-mode
NOON state
1

2
([N, 0+[0, N) gives
2
= N
2
and ( = 0,
so it seems not to be entangled according to the criterion
of the CSI.
It is instructive to compare the CSI criterion with an-
other method of detecting entanglement in many-body
systems, which is known from quantum metrology. The
Quantum Fisher information (QFI), denoted here by F
Q
[20], provides a lower bound for the precision of the
estimation of an unknown parameter in a series of m
experiments,
1

m
1

FQ
. The value of F
Q
is deter-
mined by the properties of the state and the trans-
formation, which introduced the dependence on in the
system. A particularly important is the case, when is
the relative phase between two modes of , imprinted by
an interferometer, which can be represented by a unitary
transformation e
in

J
. Here n

J is a product of a unit
vector and the vector of angular momentum operators
[21]. This interferometric transformation acts on each
particle independently, so it does not entangle them, and
F
Q
N for all separable [10]. In consequence, all two-
mode states which give F
Q
> N are entangled [10]. Typ-
ically in the laboratory it is not necessary to nd to
estimate the value of F
Q
. Usually, some quantity F
Q
is measured, such as the inverse of the spin-squeezing pa-
rameter [2231]. If the experiment outcome gives > N,
then also F
Q
> N and the system is particle-entangled.
In contrary to the interferometric criterion, the vio-
lation of the CSI (6) makes no assumptions about the
modal structure of . Moreover, the QFI is inevitably
related to some interferometric transformation, so to ver-
ify if the state is entangled one usually must implement
this interferometer [32]. The CSI criterion is not linked
with any transformation, so aside from the measurement
of the integrated correlation function (5) it does not re-
quire any further manipulation of the system.
However, a simple illustration shows that the metro-
logical approach is more powerful than that based on
the violation of the CSI. Take a pure Fock state of N
particles equally occupying two modes described by the
bosonic annihilation operators a and

b, i.e.

N
2
,
N
2
_
. If
this evolves in the Mach-Zehnder interferometer, which
is represented by the interferometric transformation with
n = (0, 1, 0)
T
, then the QFI is equal to F
Q
= 4

J
2
y
=
N +
N
2
2
[20]. Therefore, according to this interferometric
criterion, the state is strongly entangled and the correc-
tion to the no-entanglement limit F
Q
= N is quadratic
in N. On the other hand, using Eq. (6) for this state,
we obtain ( = 1 +
2
N2
, so in this case the correction
to the classical value ( = 1 becomes negligible for large
N. This example underlines the main dierence between
the QFI approach and the CSI criterion. The former,
although usually dicult to implement, utilizes infor-
mation about whole density matrix. The latter, which
should be much easier to check experimentally, is based
solely on the properties of the second-order correlation
function. This way, when N is large, much knowledge
about the non-classical relations between the particles,
contained in higher order correlations, is lost.
Another important dierence can be illustrated by con-
sidering the phase-imprinting operation, represented by
the transformation with n = (0, 0, 1)
T
. This interferom-
eter fed by the NOON state gives F
Q
= N
2
, while the
CSI criterion does not detect any entanglement, as ar-
gued above. This shows another advantage of the metro-
logical approach thanks to the freedom of choice of the
interferometric apparatus, it is sensitive to a wider spec-
trum of entangled states.
Finally, we demonstrate that the indistinguishability
of particles is crucial to establish relation between the
violation of the CSI and entanglement. To this end, we
calculate the second-order correlation function for a sep-
arable state (3) without imposing the indistinguishability
of particles
(
(2)
(x, x

) =
N

n=m=1
Tr
_

sep

n
(x)

m
(x

)
_
. (19)
Here,

n
(x) projects the n-th particle onto the position
state [x, while the sum ensures that all possible combi-
4
nations of one particle being at position x and the other
at x

contribute to the correlation function. Using Eq. (3)


we obtain that
(
(2)
(x, x

) =

i
p
i
N

n=m=1
P
(n)
i
(x)P
(m)
i
(x

), (20)
where P
(n)
i
(x) = Tr
_

(n)
i

n
(x)
_
is the one-body proba-
bility for nding the n-th particle in state
(n)
i
at position
x. If particles are identical, then these probabilities do
not depend on indices n and m. In this case, the sum
over n ,= m gives the coecient N(N 1) and after in-
tegrating x over volume V
a
and x

over V
b
, we obtain the
discrete version of Eq. (11). However, if particles are not
identical, then the sum over n ,= m gives
(
(2)
(x, x

) =

i
p
i
f
i
(x, x

), (21)
where f
i
(x, x

) does not factorize into a product of func-


tions of x and x

. In consequence, no such relation as in


Eq. (13) can be established.
Indeed, a following example shows that for distinguish-
able particles, the CSI can be violated even for separable
states. Consider two particles occupying two modes
(a)
and
(b)
in a Werner state [19]

w
=
1 p
4

1 + p [
1

1
[ , (22)
where [
1
=
1

2
([
(a)
1
,
(b)
2
+[
(a)
2
,
(b)
1
) and 0 p 1.
Since the identity operator is spanned by the triplet
of bosonic vectors [
1
, [
2
= [
(a)
1
,
(a)
2
, [
3
=
[
(b)
1
,
(b)
2
and a fermionic singlet [
4
=
1

2
([
(a)
1
,
(b)
2

[
(a)
2
,
(b)
1
), then
w
is not a state of indistinguishable
bosons apart from p = 1. For this state, the second
order correlation function can be easily calculated. For
instance, (
(2)
aa
= Tr [
w
[
2

2
[] and similarly for (
(2)
bb
and (
(2)
ab
, giving ( = 2
1+p
1p
> 1 for all 0 p 1. On the
other hand, the Peres-Horodecki positive partial trans-
pose criterion [18, 33] tells, that
w
is entangled when
p >
1
3
. This simple example shows, that violation of the
CSI does not necessarily imply entanglement of distin-
guishable particles.
In conclusion, we have demonstrated that the analog
of the Glauber-Sudarshan P-representation known from
quantum optics, can be used to show that the viola-
tion of the Cauchy-Schwarz inequality for the second-
order correlation function proofs entanglement in a sys-
tem of identical bosons. The violation of the CSI was
already observed in [34], where correlated atomic pairs
were emited due to collisions of two counter-propagating
Bose-Einstein condensates (BECs). Since the particles
forming the BECs are indistinguishable, we conclude that
this experiment demonstrated entanglement in a many-
body system. The CSI condition is not as powerful
as those inferred from quantum metrology, but usually
is much simpler to implement. It could be used as a
rst test of entanglement in systems, where the number-
squeezing is likely to be present, like in the twin-beam se-
tups [35, 36], scattering of particles in the BEC collisions
[3741] and many other correlated systems, to indicate
its potential applicability for ultra-precise metrology, vi-
olation of Bell inequalities [4, 42, 43] or to demonstrate
the EPR phenomenon.
T.W. and P. Sz. acknowledge the Foundation for
Polish Science International Ph.D. Projects Programme
co-nanced by the EU European Regional Development
Fund. M. T. acknowledges the support of the National
Science Center grant N202 167840. J. Ch. acknowledges
the Foundation for Polish Science International TEAM
Programme co-nanced by the EU European Regional
Development Fund. T. W., P. Sz., P. Z. and J.Ch.
were supported by the National Science Center grant no.
DEC-2011/03/D/ST2/00200.
[1] E. C. G. Sudarshan, Phys. Rev. Lett. 10, 277 (1963)
[2] Roy J. Glauber, Phys. Rev. 131, 2766 (1963)
[3] I. Bengtsson and K.

Zyczkowski, Geometry of Quan-
tum States. An Introduction to Quantum Entangle-
ment,Cambridge University Press (2006)
[4] R. Horodecki, P. Horodecki, M. Horodecki and K.
Horodecki, Rev. Mod. Phys. 81, 865 (2009)
[5] A. Einstein, B. Podolsky, and N. Rosen, Phys. Rev. 47,
777 (1935)
[6] M. A. Nielsen and I. L. Chuang, Quantum computation
and quantum information, Cambridge University Press
(2000).
[7] C. H. Bennett, G. Brassard, C. Crepeau, R. Jozsa, A.
Peres, and W. K. Wootters, Phys. Rev. Lett. 70, 1895
(1993)
[8] C. H. Bennett, G. Brassard, S. Popescu, B. Schumacher,
J. A. Smolin, and W. K. Wootters, Phys. Rev. Lett. 76,
722 (1996)
[9] V. Giovanetti, S. Lloyd and L. Maccone, Science 306,
1330 (2004)
[10] L. Pezze and A. Smerzi, Phys. Rev. Lett. 102, 100401
(2009)
[11] M. Sarovar, A. Ishizaki, G. R. Fleming and K. B. Whaley,
Nat. Phys. 6, 462 (2010)
[12] M. Schlosshauer, Decoherence and the Quantum-to-
Classical Transition, Springer (2007)
[13] E. Joos, H. D. Zeh, C. Kiefer, D.J.W. Giulini, J. Kupsch
and I.-O. Stamatescu, Decoherence and the Appearance
of a Classical World in Quantum Theory, Springer (2003)
[14] R. Omnes, Understanding Quantum Mechanics, Prince-
ton University Press (1999)
[15] W. H. Zurek, arXiv:quant-ph/0306072
[16] M. Schlosshauer, Rev. Mod. Phys. 76, 1267 (2004)
[17] A. Sorensen, L.-M. Duan, J.I. Cirac and P. Zoller, Nature
69, 63 (2001)
[18] A. Peres, Phys. Rev. Lett. 77, 1413 (1996)
[19] R. F. Werner, Phys. Rev. A 40, 4277 (1989)
[20] S. L. Braunstein and C. M. Caves, Phys. Rev. Lett. 72,
5
3439 (1994).
[21] These angular momentum operators are dened as

Jx
( a

b +

a )/2,

Jy ( a

a )/2i and

Jz ( a

b )/2.
[22] J. Appel, P. J. Windpassinger, D. Oblak, U. B. Ho, N.
Kjrgaard, and E. S. Polzik, PNAS 106, 10960 (2009)
[23] C. Gross, T. Zibold, E. Nicklas, J. Esteve and M. K.
Oberthaler, Nature 464, 1165 (2010)
[24] Max F. Riedel, Pascal Bohi, Yun Li, Theodor W. Hansch,
Alice Sinatra and Philipp Treutlein, Nature 464, 1170
(2010)
[25] I. D. Leroux, M. H. Schleier-Smith and V. Vuletic, Phys.
Rev. Lett. 104, 1170 (2010)
[26] Z. Chen, J. G. Bohnet, S. R. Sankar, J. Dai, and J. K.
Thompson, Phys. Rev. Lett. 106, 133601 (2011)
[27] D. J. Wineland, J. J. Bollinger, W. M. Itano and D. J.
Heinzen, Phys. Rev. A 50, 67 (1994)
[28] M. Kitagawa and M. Ueda, Phys. Rev. A 47, 5138 (1993)
[29] D. J. Wineland, J. J. Bollinger, W. M. Itano, F. L. Moore
and D. J. Heinzen, Phys. Rev. A 46, 6797 (1992)
[30] J. Esteve, C. Gross, A. Weller, S. Giovanazzi and M. K.
Oberthaler, Nature 455, 1216 (2008)
[31] T. Berrada, S. van Frank, R. B ucker, T. Schumm, J.-F.
Scha and J. Schmiedmayer, Nat. Comm. 4, 2077 (2013)
[32] B. L ucke, M. Scherer, J. Kruse, L. Pezz e, F. Deuret-
zbacher, P. Hyllus, O. Topic, J. Peise, W. Ertmer, J.
Arlt, L. Santos, A. Smerzi and C. Klempt, Science 11,
773 (2011)
[33] M. Horodecki, P. Horodecki and R. Horodecki, Phys.
Lett. A 223, 1 (1996)
[34] K. V. Kheruntsyan, J.-C. Jaskula, P. Deuar, M. Bonneau,
G. B. Partridge, J. Ruaudel, R. Lopes, D. Boiron, and
C. I. Westbrook, Phys. Rev. Lett. 108, 260401 (2012)
[35] R. B ucker, J. Grond, S. Manz, T. Berrada, T. Betz,
C. Koller, U. Hohenester, T. Schumm, A. Perrin and J.
Schmiedmayer, Nat. Phys. 7, 608 (2011)
[36] M. Bonneau, J. Ruaudel, R. Lopes, J.-C. Jaskula, A.
Aspect, D. Boiron, and C. I. Westbrook, Phys. Rev. A
87, 061603 (2013)
[37] A. Perrin, H. Chang, V. Krachmalnico, M. Schellekens,
D. Boiron, A. Aspect, C.I. Westbrook, Phys. Rev.
Lett. 99, 150405 (2007).
[38] J.-C. Jaskula, M. Bonneau, G. B. Partridge, V. Krach-
malnico, P. Deuar, K. V. Kheruntsyan, A. Aspect, D.
Boiron, and C. I. Westbrook, Phys. Rev. Lett. 105,
190402 (2010)
[39] R. G. Dall, L. J. Byron, A. G. Truscott, G. R. Dennis, M.
T. Johnsson, and J. J. Hope, Phys. Rev. A 79, 011601
(2009)
[40] D. Pertot, B. Gadway, and D. Schneble, Phys. Rev. Lett.
104, 200402 (2010)
[41] Wu RuGway, S. S. Hodgman, R. G. Dall, M. T. Johnsson,
and A. G. Truscott, Phys. Rev. Lett. 107, 075301 (2011)
[42] P. G. Kwiat, K. Mattle, H. Weinfurter, A. Zeilinger, A. V.
Sergienko and Y. Shih, Phys. Rev. Lett. 75, 4337 (1995)
[43] M. D. Reid, P. D. Drummond, W.P. Bowen, E. G. Cav-
alcanti, P. H. Lam, H. A. Bachor, U. L. Andersen, G.
Leuchs, Rev. Mod. Phys. 81, 1727 (2009).

You might also like