Вы находитесь на странице: 1из 6

Connecting Quantum Discord with Coherence

Yu Guo1, and Sumit Goswami2

arXiv:1611.00413v3 [quant-ph] 17 Nov 2016

School of Mathematics and Computer Science, Shanxi Datong University, Datong, Shanxi 037009, China
2
Institute for Quantum Science and Technology, and Department of Physics & Astronomy,
University of Calgary, 2500 University Drive NW, Calgary, Alberta T2N 1N4, Canada
Quantum discord has been studied extensively as a measure of non-classical correlations which
includes entanglement as a subset. Although it is well known that non-zero discord can exist without
entanglement, the origin of quantum discord is not well understood as compared to entanglement
which manifests itself more simply as higher dimensional quantum superposition. In this Letter we
explore the relation between quantum discord and the correlated bipartite coherence. Consequently,
we establish a new measure of bipartite coherence called discord-like correlated coherence, minimum
of which coincides with the original quantum discord. This demonstrates quantum discord as the
irreducible correlated bipartite coherence. In addition, the discord-like correlated coherence is shown
to admit the postulates of the quantum resource theory (QRT) although the origin quantum discord
is not a good candidate under the QRT. We also find out that the relative entropy measure induced
from the discord-like coherence is a well-defined coherence measure for bipartite states.

Introduction.Correlated information always lies at


the heart of quantum information theory [13]. Quantum
discord was introduced [4] to quantify the total amount
of quantum correlation present in a bipartite system in
which entanglement,the most widely used quantum correlation, is included as a subset. Discord can exist without entanglement and hence is more robust against environment induced decoherence. Quantum discord has
been investigated extensively in the past decade since
it is proven to be an important quantum resource in a
plethora of quantum information processing tasks [5
19].
The relation between different quantum resources is
of great importance [3234]. It is shown that entanglement is a minimal quantum discord over state extensions
[32]. Another fundamental concept in quantum physics
which is closely connected to quantum superposition and
quantum correlations is coherence. An algorithmic characterization of quantum coherence as a resource and a set
of bona fide criteria for coherence monotones have been
identified[33, 3543]. Correlated coherence has been proposed to capture the mutual coherence between the two
subsystems of a bipartite system [34]. It is shown that coherence can be measured with entanglement [33] and the
correlated coherence is closely related to entanglement
[34]. The main purpose of this Letter is to investigate
the relation between quantum discord and coherence for
the bipartite systems. By replacing the von Neumann
entropy with the relative entropy measure of coherence,
replacing the mutual information by the correlated coherence and replacing the von Neumann measurements
by the local rank-one projective physically incoherent operations we establish a discord-like correlated coherence
for bipartite states and then compare it with the original
quantum discord: The original quantum discord turned
out to be the minimal discord-like correlated coherence
over all possible reference bases.
Later in the Letter we investigate the roles of quan-

tum discord and the discord-like coherence as quantum


resources in context of the recently developed quantum
resource theory (QRT) [28]. QRT was developed to create a unifying theoretical framework for different quantum resources. Considerable work on formulating QRT
has been done recently [2030]. A general structure of
QRT has three ingredients: (1) the free states, (2) the
resource states, and (3) the free operations. For example, in entanglement theory the resource states are the
entangled states, the free states are the separable ones
and the free operations are the local operations and classical communication (LOCC). In the theory of coherence,
the resource states are the coherent states, the free states
are the incoherent ones and the free operations are the
incoherent operations. However, not all the free operations can be implemented physically. Hence, the physically consistent conditions for QRT has been formulated
very recently[31], particularly in the context of coherence.
Physically incoherent operations turned out to be some
special incoherent operations [31]. Both the quantum discord and the discord-like coherence we established, have
been rigorously investigated as possible resources under
QRT. Consequently, we show that although the original
quantum discord does not obey the structure of QRT, the
discord-like coherence demonstrated itself as a reasonable
resource under QRT.
We recall the definitions of quantum discord and coherence at first. For a state ab of a bipartite system,
with finite dimensional subsystems A and B, described
by Hilbert space Ha Hb , the quantum discord of ab
(up to part A) is defined by [4]
D(ab ) := min
{I(ab ) I(ab |a )},
a

(1)

where, the minimum is taken over all local von Neumann


measurements a , I(ab ) := S(a ) + S(b ) S(ab ) is
interpreted as the quantum mutual information, S() :=
Tr( log ) is the von Neumann P
entropy, I(ab |a )} :=
S(b ) S(ab |a ), S(ab |a ) := k pk S(k ), and k =

2
1
a
pk (k

Ib )ab (ak Ib ) with pk = Tr[(ak Ib )ab (ak


Ib )], k = 1, 2, . . . , dim Hb . D(ab ) can be quantified as
D(ab ) = min
{I(ab ) I[(a 1b )ab ]}.
a

(2)

a combination of operations each with Kraus operators


{Kj }rj=1 of the form
Kj = Uj Pj =

1i

|j (i)ihi|Pj ,

(8)

Coherence is defined and quantified along the approach


of [35]. Let H be a finite dimensional Hilbert space with
dim H = d. Fixing a particular basis {|ii}d1
i=0 , we call all
quantum states represented by density operators that are
diagonal in this basis incoherent. This incoherent set of
quantum states will be labeled by I, all density operators
I are of the form
=

d1
X

i |iihi|.

(3)

i=0

Henceforth, we call the fixed basis reference basis. A


quantum operation is incoherent if its Kraus operators
fulfil Kn Kn /Tr(Kn Kn ) I for all I and for all n.
Discord-like correlation of bipartite coherence. The
relative entropy of coherence is defined by [35]
Cr () := min S(k),
I

(4)

where S(k) = Tr( log2 log2 ) is the relative entropy. Cr () can be calculated to be [35]
Cr () = S(d ) S(),

(5)

where d denotes the diagonalP


part of the in the reference basis, i.e., d = () = i |iihi||iihi|. Cr admits
the supper-additive property [46]
Cr (ab ) Cr (a ) + Cr (b ).

(6)

The equality holds whenever ab is a product state [46].


We remark here that S(ab ) S(a ) + S(b ) and the
equality holds if and only if ab is a product state. However, Cr (ab ) = Cr (a ) + Cr (b ) whenever ab is a product state but not vice versa. In fact, for any diagonal
bipartite state ab , we have Cr (ab ) = Cr (a ) + Cr (b ).
But ab is not necessarily a product state. If ab is a maximally coherent state, equality in Eq. (6) holds if and only
if it is a product state, provided that dim Ha = dim Hb .
But it is not valid when dim Ha 6= dim Hb [47].
Definition 1. We call the difference
Ic (ab ) := Cr (ab ) Cr (a ) Cr (b )

(7)

correlated coherence of ab with respect to the relative


entropy measure of coherence.
Ic (ab ) is the amounts of mutual coherence information contained in ab , which is similar to that of mutual
information I(ab ).
In [31], the physically incoherent operation (PIO) is
proposed according to the physically consistent QRT. A
CPTP map is proven to be a PIO if it can be written as

where {|ii} is the reference basis of part A, Pj form an


orthogonal and complete set of incoherent projectors on
part A and j are permutations. Particularly, we call it a
projective PIO (PPIO) if all its Kraus operators are in the
form of Eq.(8), namely, if it is not a convex combination.
Here, projective means that the Kraus operator Kj is
very close to the projective operator Pj . Moreover, it is
a rank-one PPIO if all its Kraus operators are rank-one.
a be any given local rank-one PPIO
Theorem 1. Let
a () :=
a 1b (), then
(on the subsystem A),
a (ab )) Ic (ab )
Ic (

(9)

is valid for any bipartite state ab .


a be a PPIO on part A. Then, for
Proof. Let
a on ab , the output state is =
any , we apply
ab
a

( 1b )ab . It follows that Ic (ab ) Ic (ab ) =


Cr (ab ) Cr (a ) Cr (b ) [Cr (ab ) Cr (a ) Cr (b )] =
Cr (ab )Cr (a )[Cr (ab )Cr (a )] = [S(( 1b )ab )
S(ab ) (S((a )) S(a ))] [S(( 1b )ab ) S(ab )
(S((a )) S(a ))] = S(ab ) S(ab ) + S(a ) S(a ).
a is
We show that S(ab )S(ab )+S(a )S(a ) 0. If
a rank-one PPIO with Ui = Uj for any i and j, it is clear
since S(ab ) S(ab ) + S(a ) S(a ) = I(ab ) I(ab )
in such a case. If Ui 6= Uj for some i and j, then Eq. (9)
is also true since S is concave.

Note. In general, for a PIO which is not a rank-one
PPIO, Eq. (9) is not true. Equation (9) is an analogue
of Eq. (14) for mutual information [4].
Definition 2. Let ab be a bipartite state. We define
a (ab )]},
Dc (ab ) := min{Ic (ab ) Ic [
a

(10)

where the minimum is taken over all local rank-one


a.
PPIOs
Hence, Dc (ab ) is defined as an analogue of quantum
discord: mutual information is replaced by correlated coherence and the von Neumann measurement is substituted P
by PPIO. It is easy to verify that Dc (ab ) = 0 iff
b
ab = P
i pi |iihi| i . By Theorem 3.1 in [44], Dc () = 0
for = i,j Aij |iihj| if and only if Aij s are mutually
commuting normal operators which are diagonal under
the reference basis. We denote the set of all states that
with zero Dc by Dc0 . Then Dc0 is a convex set. Let D0 be
the set of all zero-discordant states (up to part A), then
Dc0 is a proper subset of D0 .
Let {|ii} be the reference basis. An operation is
called an incoherently unitary operation (IUO) if
X
e 1i |(i)ihi|.
() = U U , U =
i

3
TABLE I. The comparison of the coherence-induced discord Dc with the original quantum discord D.
Correlation
D
Dc

Free states
D0
Dc0

Free Operation
Theorem 1
Proposition 1

Physically Free Operation


Theorem 2
Proposition 2

It is straightforward that Dc is invariant under local IUO.


In addition, (i) Dc cannot increase under local rank-one
PPIO and (ii) Dc can be generated under LOCC [48].
Comparison of D with Dc . From the proof in The a (ab )) is atorem 1, we know that the maximal Ic (
a

tained by the rank-one PPIO with Ui = Uj for


P

Ib ] =
anyPi and j and Ic [ j Uinc Pj Ib (ab )Pj Uinc
Ic [ j Pj Ib (ab )Pj Ib ]. We thus can conclude the
following.
Theorem 2. Let be set of all orthonormal basis of
Ha . Then
D(ab ) = min Dc (ab ),

(11)

where the minimum is taken over all possible reference


bases in .
Eq. (11) displays the relation between the quantum
discord and the correlated coherence: quantum discord
is the minimal correlated bipartite coherence. In addition, although the calculation of D is NP-complete [49],
Dc can be obtained directly since it is not dependent
on the choice of the rank-one PPIO. Furthermore, Dc ()
not only displays the quantum discord contained in but
also reflects the correlated coherence of . In other words,
Dc reveals quantum correlation and coherence simultaneously.
Another question is that whether the states in D0
0
Dc are quantum correlated. In fact, Dc is just the
quantum correlation up to the fixed reference basis. For
the fixed reference basis, the corresponding measurement
a is more easily performed than that of the arbitrarily

ones in laboratory: The direction of the corresponding


measurement is fixed for the given reference. That is,
replacing D by Dc is more useful both theoretically and
experimentally from this point of view.
The relative entropy of the discord-like coherence.
We follow the the axiomatic method for a reasonable measure of quantum coherence C() proposed in
Ref. [35]: (C1) C() 0, I [(C1 ) C() = 0
iff I]. (C2a) C() is nonincreasing under incoherent operations, i.e., C(()) C() for any IO .
(C2b) Monotonicity for average
coherence under selecP
p
C(
tive IO, i.e., C(())
i ), with probabilities
i i

pi = Tr(Ki Ki ), state i = Ki Ki /pi , and incoherent Kraus operators


Ki IKi I. (C3)
P Ki obeying P
Convexity,
j pj j ) for any set
j pj C(j ) C(
P i.e.,
{pj , j : j pj = 1}.
The relative entropy of a resource is an important
quantity in QRT [28]. We now discuss the relative en-

Invariant Operation
Local UO
Local IUO

Measurement
a
Rank-one PPIO

tropy of discord-like coherence. We define


C r (ab ) :=

inf

ab Dc0

S(ab kab ).

(12)

P
b
Dc0 and let dab =
Let
i be any state in P
P ab = i pi |iihi|
0
i |iihi| Bii Dc whenever ab =
i,j |iihj| Bij
under the reference basis {|ii} of Ha (that is, dab = (
1b )ab ). It follows that S(ab kab ) = S(dab ) S(ab ) +
S(dab kab ), and thus we get,
C r (ab ) = S(dab ) S(ab ).

(13)

It is clear that C r (ab ) fulfils (C1)(but it does not satisfies (C1 ), that is, C r (ab ) is not faithful). It satisfies
(C2a) and (C3) since the relative entropy is contractive
and jointly convex. It also fulfils (C2b) by Theorem 5
in Ref. [50] (also see the similar argument as that of
(C2b) for the original relative entropy of coherence in
[35]). That is, although Dc is not a well-defined coherence measure since they are not convex under the mixing
of states, the relative entropy measure C r is a measure
of bipartite coherence. The defect of this coherence measure is that it is not faithful, which is similar to that of
negativity as an entanglement measure [51]. So we can
also consider the symmetric discord-like measure and the
symmetric relative entropy of the discord-like quantity,
c and C r respectively. That is,
denoted by D
c (ab ) := min {Ic (ab ) Ic [(
a
b )ab ]}, (14)
D
a
b

a
b.
where the minimum is taken over all local PPIOs
c (ab ) = 0 if and only
For this symmetric measure, D
if ab is diagonal with respect to the reference basis
c0 be the set of all states with zero sym{|ii|ji}. Let D
metric discord-like measure. The corresponding relative
entropy measure is
C r (ab ) :=

inf

0
ab D
c

S(ab kab ).

(15)

That is, the symmetric measure of relative entropy coincides with the original relative entropy coherence measure.
In what follows, we discuss whether D and Dc satisfy the requirements of QRT. Let F be the set of
all free states (in all possible finite dimensions), and
Fm = F S(Hm ), where Hm = Hm1 Hm2 Hms ,
ao and Gour prodim Hmi = mi , i = 1, . . . , s. Brand
posed the following postulates [28]: (i) F is closed under tensor products; (ii) F is closed under the partial
trace of spatially separated subsystems; (iii) F is closed

4
under permutations of spatially separated subsystems;
(iv) Each Fm is a closed set; (v) Each Fm is a convex
set; (vi) The set of free operations cannot generate a
resource; they cannot convert free states into resource
states. We discuss below whether the quantum discord
and the discord-like correlation of coherence can be regarded as a quantum resource under QRT.
The free operation of quantum discord. For a Hilbert
space H we denote by B(H) and S(H) the space of all
bounded linear operators on H and the set of all quantum states on H, respectively. Recall that, a channel acting on an ndimensonal quantum system is
called an isotropic channel if it has the form () =
t() + (1 t)Tr() nI , where is either a unitary operation or unitarily equivalent to transpose [9]. If t is
nonzero, we call a nontrivial isotropic channel. A channel is called a completely decohering channel (CDC) if
(B(H)) is commutative. By Theorems 1-2 in Ref. [11]
and Theorem 1 in Ref. [8], if = a b , where a is
either a CDC or (i) a nontrivial isotropic channel (NIC)
on part A whenever dim Ha 3 and (ii) a unital channel
on part A whenever dim Ha = 2, and b is any quantum
channel on part B, then is a FOQD. For simplicity, we
call such a local channel a a local free operation of QD
(LFOQD). We show below that, the FOQDs contains the
operations above as special cases.
Theorem 3. A quantum operation : B(Ha Hb )
B(Ha Hb ) is a FOQD if and only if it has the Kraus
operators of the form
Kj = Aj Bj ,

(16)

P
P
where j Aj ()Aj is a LFOQD and j Bj ()Bj is any
quantum channel on part B.
Proof.The if part is clear.
We only need
to check the only if part. We write the Kraus
P
(j)
operators of as Kj =
k,l Akl |kihl|, where
{|ki} P
is an orthonormal basis ofPHb .
Taking
ab = i pi |ia ihia | bi , hia |ja i = ij , i pi = 1, pi > 0,
P P
P
(j)

then ab = (ab ) =
k,l Akl
j [(
j Kj Kj =
P
P
(j)
|tihs|)]
=
|kihl|)( i pi |ia ihia | bi )( s,t Ast
P
P P P
(j)
(j)
b
s,t hs|i |tiAks |ia ihia |Alt )]} |kihl|.
j [ i pi (
k,l {
P
P P
(j)
(j)
Let Akl (ab ) = j [ i pi ( s,t hs|bi |tiAks |ia ihia |Alt )].
Note that ab is a classical-quantum state, therefore,
by Theorem 3.1 in [44], Akl (ab ) are mutually commuting normal operators for any classical-quantum
state ab . That is, the action of on part A is in
fact a commutativity preserving map (or equivalently,
a normality preserving map by Corollary 1 in [45]).
P
(j)
(j)
by kl for
We denote the action
j,s,t Aks ()Alt
simplicity. It turns out from Theorem 3 in [45] that,
for any k and l, either kl (B(Ha )) is commutative, or
there exist a unitary operator Ua B(Ha ), a linear
function fkl on B(Ha ), and complex numbers tkl ,

P
(j)
(j)
= tkl Ua ()Ua + fkl ()Ia
such that
j,l,t Aks ()Alt

P
(j)
(j)
= tkl Ua ()T Ua + fkl ()Ia , where
or
j,l,t Aks ()Alt
X T denotes the transpose of X under the given basis.
P (j)
(j)
Together with the fact that j Aks ()Aks is a multiple
(j)

(j)

of a local channel, we thus can let Akl = kl Aj for


P
(j)

some complex number kl , where


j Aj ()Aj is a
commutativity preserving quantum channel on part A.
P
(j)
That is Kj = Aj Bj with Bj = k,l kl |kihl|. Then
we complete the proof by Theorems 1-2 in Ref. [11] and
Theorem 1 in Ref. [8].

The physically free operation of quantum discord.
When we process a quantum task, the interacting environment always consumes resources. To account for
this very recently Chitambar and Gour have investigated
the physically consistent QRT [31]. A QRT defined on
some quantum system S is physically consistent if any
free operation on S can be obtained by an auxiliary
state E , a joint unitary US,E , and a projective measurement {Pi } that are all free in an extended system S+E.
We call the free operation in the physically consistent
QRT physically free operation hereafter.
We begin with a bipartite system A+B and assume
that the environment is denoted by part E. Any physical
operation on this composite system can be decomposed
into three steps: (i) a joint unitary Uab,E is applied on
the input state ab and some fixed state E , i.e., the state

becomes Uab,E ab E Uab,E


, where Uab,E = Ua UbE
(note that the joint unitary operation is free, so it admits
this form), {|ii} is any orthonormal basis of Ha and UbE
is an arbitrary given unitary operator on Hb HE , (ii)
a von Neumann measurement act on the environment
encoding the measurement outcome as a classical index,
i.e., the state after this process is (1ab P )(Uab,E ab
Pt

E Uab,E
) = j=1 ab,j |jiE hj|, where

ab,j = TrE [(1ab Pj )(Uab,E ab E Uab,E


)],

and (iii) a classical processing channel is applied to


the measurement outcomes. That is, the final state is
Pt
Pt

k=1 ab,k |kiE hk|, where ab,k =


j=1 pk|j ab,j for
some classical channel pk|j . The CPTP map that can
be deduced in this way is said to be a physically free
operation of quantum discord (PFOQD).P
From the discussion above, let ab = ij |iihj| Bij
with respect to some basis {|ii} of Ha , then we can
conclude that ab,j = TrE [(1ab Pj )(Ua UbE ab
P

E Ua UbE
)] = TrE [ ij Ua |iihj|Ua 1b Pj UbE (Bij
P

E )UbE
] =
ij Ua |iihj|Ua TrE [1b Pj UbE (Bij

E )UbE
], which leads to the following theorem:
Theorem 4. A quantum operation : B(Ha Hb )
B(Ha Hb ) is a PFOQD if and only if it can be expressed
as a convex combination of maps each having Kraus operators {Kj }rj=1 of the form

K j = U a Bj ,

(17)

5
where Ua is an arbitrarily given unitary operator on
Ha and {Bj }rj=1 is any Kraus operators that satisfying
P
j Bj Bj = Ib .
It is clear that the set of PFOQDs is a proper subset
of that of FOQDs. In addition, quantum discord is based
on the local von Neumann measurements a , which is in
fact a special CDC, and thus a is a special FOQD. Note
that both entanglement and coherence, the free states are
closely related to the free operations. A state is entangled if and only if it can not be prepared by LOCC, and a
state is incoherent if and only if it can not be created via
incoherent operations [35]. However, quantum discord is
defined via the local von Neumann measurements, which
is only a proper set of its free operations. More remarkably, the free states of quantum discord is not convex.
From this point of view, discord as a quantum resource
is not a good candidate.
QRT of the discord-like coherence. For convenience,
(i) an isotropic channel is called incoherent if the associated unitary operation is incoherent, (ii) a CDC is incoherent if it is a CDC up to the reference basis, and (iii)
a unital channel is incoherent if it is unital with respect
to the reference basis. We call a LFOQD incoherent if
it is incoherent in the sense of items (i)-(iii) above. By
Theorems 3-4, the following are straightforward.
Proposition 1. A quantum operation is free for
discord-like correlated coherence
if and only if it has
P
the form as Eq. (16) with j Aj ()Aj is an incoherent
LFOQD.
Proposition 2. A quantum operation is a physically
free operation for the discord-like correlated coherence if
and only if it has the form as Eq. (17) with Uj are IUOs
on Ha .
When we regard Dc0 as the free states, and consider
the operations in Proposition 1 (or Proposition 2) as the
free operations (or physically free operations), we show
below that it is a new resource under QRT.
Since Dc is a measure of bipartite systems, we assume
that Hmi = Hai Hbi , i = 1, . . . , s, and S(Hm )
is free if any reduced state in Hms is free. We check the
postulates (i)-(v) for free states item by item. (i) With no
loss of generality, we consider the case of s =
P1. If ab
S(Ha Hb ) with Dc (ab ) = 0, then ab = i pi |iihi|
bi with respect to the reference basis {|ii} of Ha . It
follows that ab abP
S(Ha Hb HP
a Hb ) can be
expressed as abab = ( i pi |iihi| bi ) ( i pi |iihi| bi )
and thus it is a free state in S(Ha Ha Hb Hb ).
(ii) and (iii) are clear. (iv) is easy to check according to
Theorem 3.1 in [44]. (v) is clear and the properties for
the free operations are guaranteed by Proposition 1 and
Proposition 2. We thus conclude that the discord-like
correlation of bipartite coherence can also be regarded as
a quantum resource (the comparison between Dc and D
is listed in Table I).
It is worth mentioning that Dc can increase un-

der its free operation.


For example, we take =

1/4 1/8 1/8 1/8


1/8 1/4 1/8 1/8

1/8 1/8 1/4 1/8 under the reference basis, let


1/8 1/8 1/8 1/4
a = 1a and let the Kraus operators of b be
{|+ih0|, |ih1|}, where |i = 12 (|0i |1i). It is straightforward that Dc () 0.0202 Dc [(1a b ) ] 0.0228.
Conclusion. Quantum discord has two indispensable
shortcomings - it is difficult to compute and it does not
obey the structure of QRT. But the closely related correlation, the discord-like correlation of coherence we established here, overcomes these defects completely. It is
shown that the minimal correlated coherence over all possible reference basis turns out to be exactly the quantum
discord of the bipartite state. This shows the inherent
nature of quantum discord as a bipartite correlated coherence. Moreover, the discord-like correlation of coherence can be calculated in a straightforward manner as
well as obeying all requirements of QRT. Interestingly,
the relative entropy measure induced from this measure
can be presented as a coherence measure of the bipartite
system. It is far different from all the previous ways of
quantifying coherence in nature since Dc is obtained via
measurements. We believe that Dc would be more useful
than D as a quantum resource in quantum information
technology.
Discussion. If we replace Cr by the l1 norm measure
of coherence C1 in Definition 1 (i.e., the correlated coherence proposed in [34]), we can also define a corresponding
discord-like correlation. However, in such a case, it can
not reveal the close relation with the original quantum
discord. In addition, we conjecture that quantum discord can increase under its free operations since Dc is
not monotonic under it free operations and D is closely
related to Dc . However,it is difficult to prove due to the
fact that D() is hard to calculate in general.
This work was completed while Guo was visiting the
Institute of Quantum Science and Technology of the University of Calgary under the support of China Scholarship
Council. Guo thanks Professor C. Simon, Professor A.
Lvovsky and Professor G. Gour for their hospitality. The
authors thank Professor G. Gour and Professor S. Wu for
helpful discussions. Guo is supported by the Natural Science Foundation of China under Grant No. 11301312.

guoyu3@aliyun.com
[1] M. A. Nielsen and I. L. Chuang, Quantum Computation
and Quantum information (Cambridge University Press,
Cambridge, 2000).
[2] R. Horodecki, P. Horodecki, M. Horodecki, and K.
Horodecki, Rev. Mod. Phys. 81, 865 (2009).
[3] O. G
uhne and G. T
oth, Phys. Rep. 474, 1 (2009).
[4] H. Ollivier and W. H. Zurek, Phys. Rev. Lett. 88, 017901

6
(2001).
[5] K. Modi, A. Brodutch, H. Cable, T. Paterek, and V.
Vedral, Rev. Mod. Phys. 84, 1655 (2012).
[6] A. Streltsov and W. H. Zurek, Phys. Rev. Lett. 111,
040401 (2013).
[7] X. Yang, G. Huang, and M. Fang, Opt. Commun. 341,
91-96 (2015).
[8] A. Streltsov, H. Kampermann, and D. Bru, Phys. Rev.
lett. 107, 170502 (2011).
[9] X. Hu, H. Fan, D. L. Zhou, and W. M. Liu, Phys. Rev.
A 85, 032102 (2012).
[10] Y. Guo, Sci. Rep. 6, 25241 (2016).
[11] Y. Guo and J. Hou, J. Phys. A Math. Theor. 46, 155301
(2013).
[12] A. Datta, A. Shaji, and C. M. Caves, Phys. Rev. Lett.
100, 050502 (2008).
[13] A. Brodutch, Phys. Rev. A 88, 022307 (2013).
[14] X. Su, Chin. Sci. Bull. 59, 1083-1090 (2014).
[15] B. Dakic, Y. O. Lipp, X. Ma, M. Ringbauer, S.
Kropatschek, S. Barz, T. Paterek, V. Vedral, A.
Zeilinger, C. Brukner, and P. Walther, Nature Phys. 8,
666 (2012).
[16] G. L. Giorgi, Phys. Rev. A 88, 022315 (2013).
[17] S. Pirandola, Sci. Rep. 4, 6956 (2014).
[18] J.-S. Xu, K. Sun, C.-F. Li, X.-Y. Xu, G.-C. Guo, E. Andersson, R. L. Franco, and G. Compagno, Nat. Commun.
4, 2851 (2013).
[19] R. LoFranco, B. Bellomo, E. Andersson, and G. Compagno, Phys. Rev. A 85, 032318 (2012).
[20] F. G. S. L. Brand
ao, M. Horodecki, J. Oppenheim, J.
M. Renes, and R. W. Spekkens, Phys. Rev. Lett. 111,
250404 (2013).
[21] M. Horodecki and J. Oppenheim, Nat. Commun. 4, 2059
(2013).
[22] P. Faist, F. Dupuis, J. Oppenheim, and R. Renner, Nat.
Commun. 6, 7669 (2015).
[23] G. Gour, M. P. Muller, V. Narasimhachar, R. W.
Spekkens, and N. Y. Halpern, Phys. Rep. 583, 1 (2015).
[24] I. Marvian and R. W. Spekkens, Nat. Commun. 5, 3821
(2014).
[25] S. L. Braunstein and P. van Loock, Rev. Mod. Phys. 77,
513 (2005).
[26] A. Grudka, K. Horodecki, M. Horodecki, P. Horodecki,
R. Horodecki, P. Joshi, W. Klobus, and A. Wojcik, Phys.
Rev. Lett. 112, 120401 (2014).

[27] A. Rivas, S. F. Huelga, and M. B. Plenio, Rep. Prog.


Phys. 77, 094001 (2014).
[28] F. G. S. L Brand
ao and G. Gour, Phys. Rev. lett. 115,
070503 (2015).
[29] G. Gour, M. P. M
uller, V. Narasimhachar, R. W.
Spekkens, and N. Y. Halpern, Phys. Rep. 583, 1-58
(2015).
[30] A. Winter and D. Yang, Phys. Rev. Lett. 116, 120404
(2016).
[31] E. Chitambar and G. Gour, Phys. Rev. Lett., 117,
030401 (2016).
[32] S. Luo, Phys. Rev. A 94, 032129 (2016).
[33] A. Streltsov, U. Singh, H. S. Dhar, M. N. Bera, and G.
Adesso, Phys. Rev. Lett. 115, 020403 (2015).
[34] K. C. Tan, H. Kwon, C.-Y. Park, and H. Jeong, Phys.
Rev. A 94, 022329 (2016).
[35] T. Baumgratz, M. Cramer, and M. B. Plenio, Phys. Rev.
Lett. 113, 140401 (2014).
[36] J. Chen, S. Grogan, N. Johnston, C.-K. Li, and S.
Plosker, Phys. Rev. A 94, 042313 (2016).
[37] C. Napoli, T. R. Bromley, M. Cianciaruso, M. Piani, N.
Johnston, and G. Adesso, Phys. Rev. Lett. 116, 150502
(2016).
[38] A. E. Rastegin, Phys. Rev. A 93, 032136 (2016).
[39] S. Rana, P. Parashar, and M. Lewenstein, Phys. Rev. A
93, 012110 (2016).
[40] L.-H. Shao, Z. Xi, H. Fan, and Y. Li, Phys. Rev. A 91,
042120 (2015).
[41] D. Girolami, Phys. Rev. Lett. 113, 170401 (2014).
[42] S. P. Du and Z. F. Bai, Anna. Phys. 359, 136-140 (2015).
[43] S. P. Du, Z. F. Bai, and Y. Guo, Phys. Rev. A 91, 052120
(2015).
[44] Y. Guo and J. Hou, J. Phys. A-Math. Theor. 45, 505303
(2012).
[45] M. D. Choi, A. A. Jafarian, and H. Radjavi H, Lin. Alg.
Appl. 87 227-241 (1987).
[46] Z. Xi, Y. Li and H. Fan, Sci. Rep. 5,10922 (2015)
[47] Z. F. Bai and S. P. Du, Quant. Inf. Coumput. 15, 13551364 (2015).
[48] It can be easily checked by Theorems 1-2 and Propositions 1-2 in [11].
[49] Y. Huang, New J. Phys. 16, 033027 (2014).
[50] V. Vedral, Rev. Mod. Phys. 74, 197 (2002).
[51] G. Vidal and R. F. Werner, Phys. Rev. A 65, 032314
(2002).

Вам также может понравиться