A lower bound of concurrence for multipartite quantum systems

Xue-Na Zhu1 , Ming Li2 , and Shao-Ming Fei3,4

1
School of Mathematics and Statistics Science, Ludong University, Yantai 264025, China
2
College of the Science, China University of Petroleum, Qingdao 266580, China
3
School of Mathematical Sciences, Capital Normal University, Beijing 100048, China
4
Max-Planck-Institute for Mathematics in the Sciences, 04103 Leipzig, Germany

We present a lower bound of concurrence for four-partite systems in terms of the concurrence for
M (2 ≤ M ≤ 3) part quantum systems and give an analytical lower bound for 2 ⊗ 2 ⊗ 2 ⊗ 2 mixed
quantum sates. It is shown that these lower bounds are able to improve the existing bounds and
detect entanglement better. Furthermore, our approach can be generalized to multipartite quantum
arXiv:1801.00078v1 [quant-ph] 30 Dec 2017

systems.

PACS numbers: 03.65.Ud, 03.67.Mn

I. INTRODUCTION

Quantifying entanglement is a basic and long standing problem in quantum information theory [1]. Concurrence
[2–6] is one of the well-accepted entanglement measures [7–12]. Different from the entanglement of formation which
is defined for bipartite systems, concurrence can be generalized to arbitrary multipartite systems. Nevertheless,
calculation of concurrence is a formidable task for higher-dimensional cases. For arbitrary S-dimensional bipartite
quantum states, Ref. [13] provided an analytical lower bound of concurrence by decomposing the joint Hilbert space
into many s (2 ≤ s ≤ S − 1)-dimensional subspaces, which may be used to improve all the known lower bounds of
concurrence. For arbitrary qubit systems, Ref. [14] provided analytical lower bounds of concurrence in terms of the
monogamy inequality of concurrence for qubit systems. For arbitrary N-partite S-dimensional quantum states, Ref.
[15] provided an analytical lower bound of concurrence in terms of the concurrence for N-partite s (2 ≤ s ≤ S − 1)-
dimensional quantum systems. More generally, for arbitrary N-partite arbitrary dimensional quantum states, Ref.
[17] provided an analytical lower bound of concurrence in terms of the concurrence for two part quantum systems.
A natural problem is whether the arbitrary dimensional N-partite quantum states can be dealt with M -partite
(2 ≤ M ≤ N − 1) quantum systems.
In this paper we provide the lower bound of concurrence for 4-partite quantum states in terms of tripartite and
bipartite quantum systems. The generalized lower bound of concurrence can be generalized to the multipartite case.

II. LOWER BOUND OF CONCURRENCE FOR FOUR-PARTITE QUANTUM SYSTEMS

To investigate multi-entanglement we first introduce the M-partite concurrence in N-partite systems. For a pure
N-partite quantum state |ψi ∈ H1 ⊗ H2 ⊗ · · · ⊗ HN , dimHi = di , i = 1, ..., N , the concurrence of |ψi is defined by
[6, 16]
s
1− N
X
CN (|ψi) = 2 2 (2N − 2) − T r(ρ2α ), (1)
α

where ρα = T rᾱ (|ψihψ|), α ⊆ {1, 2, ..., N }, ᾱ is the compliment of α, ρα labels all the different re-
duced density matrices of |ψihψ|. We list all the 2N − 2 reduced matrices in the following way:
, ..., ρ23...N }, by noticing that T r(ρ2α ) = T r(ρ2ᾱ ) for any pure states.
{ρ1 , ρ2 , ..., ρN , ρ12 , ρ13 , ..., ρ1N , ρ23 , ..., ρ12...N −1P
For a mixed multipartite quantum state, ρ = i pi |ψi ihψi | ∈ H1 ⊗ H2 ⊗ · · · ⊗ HN , the corresponding concurrence is
given by the convex roof extension,
X
CN (ρ) = min pi CN (|ψi i), (2)
{pi ,|ψi }i
i

where the minimum is taken over all possible pure state decompositions {pi , |ψi i} of ρ. This multipartite concurrence
can be used to detect and classify various genuine multipartite entanglements. It has been shown in [17] that a
multipartite quantum state is genuinely multipartite entangled if the multipartite concurrence is larger than certain
quantities given by the number and the dimension of the subsystems.
For the N-partite quantum state |ψi ∈ H1 ⊗ H2 ⊗ · · · ⊗ HN , we denote the general M -partite decomposition of |ψi,
M M
{i }, ..., {iM1 }, {k11 , k21 }, ..., {k1M2 , k2M2 }, ..., {q1 j , ..., qj j }, where all the subspaces in one bracket {} are taken to be
1
 2

M M Pj Pj
one part. Hence, {i1 , ..., iM1 , k11 , k21 , ..., k1M2 , k2M2 , ..., q1 j , ..., qj j } = {1, 2, ..., N } and k=1 Mk = M , k=1 kMk = N .
The concurrence of the state |ψi under such M -partite partition is given by
s
M
X
CM (|ψi) = 21− 2 (2M − 2) − T r(ρ2β ), (3)
β

M M
where β ∈ {{i1}, ..., {iM1 }, ..., {q1 j , ..., qj j }}. Take N = 4 and M = 3 as an example, one has M1 = 2 and M2 = 1.
There are six different tripartite decompositions: 1|2|34, 1|23|4, 1|24|3, 12|3|4, 13|2|4 and 14|2|3. For convenience, we
denote Ci|jk|l (ρ) = C3 (ρi|jk|l ) and Cij|kl (ρ) = C2 (ρij|kl ).
Theorem 1: For any mixed quantum state, ρ ∈ H1 ⊗ H2 ⊗ H3 ⊗ H4 , the concurrence is bounded by
1
C42 (ρ) ≥ 2
2C1|2|34 2
(ρ) + 2C1|3|24 (ρ)
12
2 2
+ 2C1|4|23 (ρ) + 2C12|3|4 (ρ)
2 2
+ 2C13|2|4 (ρ) + 2C14|2|3 (ρ)
2 2 2


+ C12|34 (ρ) + C13|24 + C14|23 (ρ) . (4)
[Proof:] We start the proof with a pure state |ψi ∈ H1 ⊗ H2 ⊗ H3 ⊗ H4 . According to the definition (1), one has
that
s
1 X
C4 (|ψi) = 14 − T r(ρ2α )
2 α
s
1 X
= (1 − T r(ρ2α )), (5)
2 α

where ρα labels all the different reduced density matrices of |ψihψ|. On the other hand, we have
2 1
Ci|jk|l (|ψi) = (1 − T r(ρ2i )) + (1 − T r(ρ2jk ))
2
+ (1 − T r(ρ2l )) + (1 − T r(ρ2ijk ))
+ (1 − T r(ρ2il )) + (1 − T r(ρ2jkl )) ,

(6)
and
2
Cij|kl (|ψi) = (1 − T r(ρ2ij )) + (1 − T r(ρ2kl )). (7)
From (5), (6) and (7) , we have
1
C42 (|ψi) = 2
2C1|2|34 2
(|ψi) + 2C1|3|24 (|ψi)
12
2 2
+ 2C1|4|23 (|ψi) + 2C12|3|4 (|ψi)
2 2
+ 2C13|2|4 (|ψi) + 2C14|2|3 (|ψi)
2 2 2


+ C12|34 (|ψi) + C13|24 (|ψi) + C14|23 (|ψi) . (8)
P
Let ρ = i pi |ψii hψ| be the optimal pure state decomposition of (2). We have
X
C4 (ρ) = pi C4 (|ψii )
i
r
X 1 2 1 2 1 2 1 2
= pi C (|ψii ) + ... + C14|2|3 (|ψii ) + C12|34 (|ψii ) + ... + C14|24 (|ψii )
i
6 1|2|34 6 12 12
s
X 1 X 1
≥ ( pi √ C1|2|34 (|ψii ))2 + ... + ( pi √ C14|23 (|ψii ))2
i
6 i
12
1 2 2 2 2 2 2
≥ √ 2C1|2|34 (ρ) + 2C1|3|24 (ρ) + 2C1|4|23 (ρ) + 2C12|3|4 (ρ) + 2C13|2|4 (ρ) + 2C14|2|3 (ρ)
12
2 2 2

1
+C12|34 (ρ) + C13|24 (ρ) + C14|23 (ρ) 2 , (9)
 3
1 P P 2 1
where the Cauchy-Schwarz inequality ( j ( i yij )2 ) 2 ≤ i ( j yij
P P
) 2 has been used in the second inequality.
For a mixed quantum state ρ ∈ H1 ⊗ H2 ⊗ H3 ⊗ H4 , a lower bound of C42 (ρ) has been derived based on bipartite
partitions in Ref. [17]. By using the following relation [17],

2 1 2 2 2

Ci|j|kl (ρ) ≥ Ci|jkl (ρ) + Cij|kl (ρ) + Cikl|j (ρ) , (10)
2
from (4) we have

1
C42 (ρ) ≥ 2
2C1|2|34 2
(ρ) + 2C1|3|24 2
(ρ) + 2C1|4|23 2
(ρ) + 2C12|3|4 2
(ρ) + 2C13|2|4 2
(ρ) + 2C14|2|3 (ρ)
12
2 2 2


+ C12|34 (ρ) + C13|24 + C14|23 (ρ)
1 2 2 2 2
≥ C (ρ) + C2|134 (ρ) + C3|124 (ρ) + C4|123 (ρ)
4 1|234
2 2 2


+ C12|34 (ρ) + C13|24 (ρ) + C14|23 (ρ) = ∆, (11)

where ∆ is the lower bound obtained in [17]. Hence, our bound (4) is better than the lower bound in [17] for
four-partite quantum mixed states.

III. ANALYTICAL LOWER BOUND FOR 2 ⊗ 2 ⊗ 2 ⊗ 2 MIXED STATES

Let HA , HB and HC be 2, 2 and 4-dimensional Hilbert spaces associated with the systems A, B and C, respectively.
A pure state |ϕi ∈ HA ⊗ HB ⊗ HC has the form
1 X
X 1 X
3
|ϕi = aijk |ijki, (12)
i=0 j=0 k=0

|aijk |2 = 1, {|ijki} is the basis of HA ⊗ HB ⊗ HC . The concurrence of |ϕi can be equivalently

P
where aijk ∈ C, ijk
written as [6],

r
1X
C3 (|ϕi) = (|aijk apqm − aijm apqk |2 + |aijk apqm − aiqk apjm |2 + |aijk apqm − apjk aiqm |2 ). (13)
2
To evaluate C3 (ρ), we project 2 ⊗ 2 ⊗ 4 dimensional states to 2 ⊗ 2 ⊗ 2 sub-states. For a given 2 ⊗ 2 ⊗ 4
P1 P1 P
pure state, we define its “2 ⊗ 2 ⊗ 2” pure state |ϕi2⊗2⊗2 = i=0 j=0 k∈{k1 ,k2 } aijk |ijki, where {k1 , k2 } ∈
{{0, 1}, {0, 2}, {0, 3}, {1, 2}, {1, 3}, {2, 3}}. In fact, for any 2 ⊗ 2 ⊗ 4 pure state |ϕi , there are 6 different 2 ⊗ 2 ⊗ 2
substates with respect to |ϕi. Without causing confusion, in the following we simply use |ϕi2⊗2⊗2 to denote one of
such states, as these substates will always be considered together. The concurrence C(|ϕi2⊗2⊗2 ) is similarly given by
Eq. (13), with the subindices i and j, associated with the systems A and B respectively, running from 0 to 1, and
with the subindex k associated with the system C taking values k1 and k2
Correspondingly, for a mixed state ρ, wePdefine its 2⊗2⊗2 mixed (unnormalized) substates ρ2⊗2⊗2 . The concurrence
of ρ2⊗2⊗2 is defined
P by C(ρ2⊗2⊗2 P) = min i pi C(|φi i), minimized over all possible 2 ⊗ 2 ⊗ 2 pure-state decompositions
of ρ2⊗2⊗2 = i pi |φi ihφi |, with i pi = T r(ρ2⊗2⊗2 ). The 2 ⊗ 2 ⊗ 2 submatrices ρ2⊗2⊗2 have the following form,

ρ00k1 ,00k1 ρ00k1 ,00k2 ρ00k1 ,01k1 ρ00k1 ,01k2 ρ00k1 ,10k1 ρ00k1 ,10k2 ρ00k1 ,11k1 ρ00k1 ,11k2
 
ρ00k2 ,00k1 ρ00k2 ,00k2 ρ00k2 ,01k1 ρ00k2 ,01k2 ρ00k2 ,10k1 ρ00k2 ,10k2 ρ00k2 ,11k1 ρ00k2 ,11k2 
ρ01k1 ,01k1 ρ01k1 ,00k2 ρ01k1 ,01k1 ρ01k1 ,01k2 ρ01k1 ,10k1 ρ01k1 ,10k2 ρ01k1 ,11k1 ρ01k1 ,11k2 
 
ρ ρ01k2 ,00k2 ρ01k2 ,01k1 ρ01k2 ,01k2 ρ01k2 ,10k1 ρ01k2 ,10k2 ρ01k2 ,11k1 ρ01k2 ,11k2 
 
ρ2⊗2⊗2 =  01k2 ,00k1 , (14)
ρ10k1 ,00k1 ρ10k1 ,00k2 ρ10k1 ,01k1 ρ10k1 ,01k2 ρ10k1 ,10k1 ρ10k1 ,10k2 ρ10k1 ,11k1 ρ10k1 ,11k2 
ρ ρ10k2 ,00k2 ρ10k2 ,01k1 ρ10k2 ,01k2 ρ10k2 ,10k1 ρ10k2 ,10k2 ρ10k2 ,11k1 ρ10k2 ,11k2 
 10k2 ,00k1 
ρ ρ11k1 ,00k2 ρ11k1 ,01k1 ρ11k1 ,01k2 ρ11k1 ,10k1 ρ11k1 ,10k2 ρ11k1 ,11k1 ρ11k1 ,11k2 
11k1 ,00k1
ρ11k2 ,00k1 ρ11k2 ,00k2 ρ11k2 ,01k1 ρ11k2 ,01k2 ρ11k2 ,10k1 ρ11k2 ,10k2 ρ11k2 ,11k1 ρ11k2 ,11k2
 4

where 0 ≤ k1 < k2 ≤ 3 associated to the space HC .

Theorem 2: For any 2 ⊗ 2 ⊗ 4 tripartite mixed quantum state ρ, the concurrence C(ρ) satisfies
1X 2
C32 (ρ) ≥ C3 (ρ2⊗2⊗2 ), (15)
3
P
where stands for summing over all possible 2 ⊗ 2 ⊗ 2 mixed sub-states ρ2⊗2⊗2 .
[Proof]. From the expression of concurrence (13), it is straightforward to prove that the concurrence of pure state
|ϕi and the concurrence of |ϕi2⊗2⊗2 with respect to |ϕi have the following relation,
X1
C32 (|ϕi) ≥ C32 (|ϕi2⊗2⊗2 ). (16)
3
P
Therefore for mixed state ρ = pi |ϕi ihϕi |, we have
X
C3 (ρ) = min pi C3 (|ϕi i)
ri
1 X X 2  12
≥ min pi C3 (|ϕi i2⊗2⊗2 )
3 i
r  !2  12
1 X X
≥ min pi C3 (|ϕi i2⊗2⊗2 ) 
3 i
r  !2  21
1 X X
≥ min pi C3 (|ϕi i2⊗2⊗2 ) 
3 i
r h i 21
1 X 2
= C3 (ρ2⊗2⊗2 ) ,
3
1 1
where the relation ( j ( i xij )2 ) 2 ≤ i ( j x2ij ) 2 has been used in the second inequality, the first three minimizations
P P P P
run over all possible pure state decompositions of the mixed state ρ, while the last minimization runs over all 2 ⊗ 2 ⊗ 2
pure state decompositions of ρ2⊗2⊗2 associated with ρ.
According to Theorem 1 and Theorem 2, we have the following Corollary 1:
Corollary 1: For any 2 ⊗ 2 ⊗ 2 ⊗ 2 mixed quantum state ρ, the concurrence C4 (ρ) satisfies
 
1 X2 X
C42 (ρ) ≥  C 2 (ρ2⊗2⊗2 ) + 2
C1j|{1,2,3,4}\{1,j} (ρ) , (17)
12 3 3
2≤j≤4
P
where stands for summing over all possible 2 ⊗ 2 ⊗ 2 mixed sub-states ρ2⊗2⊗2 of ρi|j|kl , 1 ≤ i < j ≤ 4, {k, l} =
{1, 2, 3, 4} \ {i, j}).
For any four-qubit mixed quantum state ρ, Ref. [14] provided analytical lower bounds of concurrence in terms of
the monogamy inequality of concurrence:
3 4
1 XX
C42 (ρ) ≥ 2
(Ti + Tj )Cij (ρ), (18)
2 i=1 j>i

where Ti (i = 1, 2, 3, 4) are given in Ref. [14] and the difference of a constant factor 12 defining the concurrence for
four qubit pure states has already been taken into account. The bounds given in Corollary 1 can be used to improve
the bounds of concurrence presented in [14]. Let us consider the following example.
Example: We consider the quantum state ρ = 1−t 16 I16 + t|ψihψ|, where |ψi = (|0000i + |0011i + |1100i + |1111i)/2,
I16 is the 16 × 16 identity matrix.
From our Theorem 1, we need to compute the lower bounds of Ci|j|kl (ρ) and Cij|kl (ρ). For convenience, we denote

(5t − 1)2
Z1 = ,
128

1
Z2 = (1 − 9t)2 (1 + t)2 ,
128
 5

1 h √ √ i
Z3 = − (1 + t)2 5(−51 + 4 17)t2 + (26 + 4 17)t − 3 ,
256
and
s s !2
3(1 + t)4 1 + 7t 1−t
Z4 = −3 .
128 4(t + 1) 4(t + 1)

For C(ρi|j|kl ), we use Theorem 2 and the lower bound of [15] of 2 ⊗ 2 ⊗ 2 mixed states. We obtain the lower bound
2
P
Z of 1≤i<j≤4,{k,l}={1,2,3,4}/{i,j} Ci|j|kl (ρ):

1

 2Z2 , t ∈ ( , 0.2],

9

Z = 32Z + 2Z , t ∈ (0.2, 0.308051],

 1 2
32Z1 + Z2 + Z3 , t ∈ (0.308051, 1].


For C(ρij|kl ), we use the lower bound of [13]. We have

X
2
C1j|kl (ρ) ≥ Z4
1<j≤4,{k,l}={1,2,3,4}\{1,j}

with t ∈ (0.5, 1].

Then the lower bound of C42 (ρ) can be obtained, see Fig. 1.

C2

0.4

0.3

0.2

0.1

t
0.0 0.2 0.4 0.6 0.8 1.0

Fig. 1: Lower bound of C42 (ρ) for 0 ≤ t ≤ 1 .

From Fig. 1, we see that the lower bound can detect entanglement of ρ for t > 91 . From Fig. 2, we see that the our
result is better than the lower bound from [17] and [14] for t ∈ ( 19 , 0.4), where the difference of a constant factor 21 in
defining the concurrence for four qubit pure states has already been taken into account.
By generalizing our results to arbitrary dimensional n-partite systems, we have
Corollary 2: For any N -partite arbitrary dimensional mixed state ρ ∈ H1 ⊗ H2 ⊗ · · · ⊗ HN ,
M
X X
2
CN (ρ) ≥ qm Ci21 i1 ...i1 2 2 2 m m m (ρ), (19)
1 2 k1 |i1 i2 ...ik2 |...|i1 i2 ...ikm
m=2 i
Pm
where 2 ≤ M ≤ N − 1, i=1 km = N , {i11 , ..., i1k1 , ..., im m m
1 i2 ...ikm } = {1, 2, ..., n} and qm is a fixed number depending
on k1 , k2 , · · · , km .
Corollary 2 says that the lower bound of the concurrence of an N-partite quantum state can be expressed by the
concurrences of its 2, 3, ..., N − 1-partite substates.
 6

C2
0.025

0.020

0.015

0.010

0.005

t
0.15 0.20 0.25 0.30 0.35 0.40

Fig. 2: Dashed line for the bound (4), dotted line for that from Ref. [17], solid line for that from Ref. [14].

IV. CONCLUSIONS AND REMARKS

In summary, we have proposed a new approach in constructing hierarchy of lower bounds of concurrence for mixed
multipartite quantum states in terms of the less part decomposed quantum systems. Besides, our approach can be
generalized to N part systems to obtain the lower bound of the concurrence for M (2 ≤ M ≤ N − 1) part systems.

Acknowledgments This work is supported by NSFC under numbers 11675113 and 11605083.

[1] R. Horodecki, P. Horodecki, M. Horodecki, and K. Horodecki, Quantum entanglement, Rev. Mod. Phys. 81, 865(2009):867-
942.
[2] W. K. Wootters, Entanglement of Formation of an Arbitrary State of Two Qubits, Phys. Rev. Lett. 80, 2245 (1998):2245-
2248.
[3] X. F. Qi, T. Gao, and F. L. Yan, Lower bounds of concurrence for N-qubit systems and the detection of k-nonseparability
of multipartite quantum systems, Quantum Inf Process 16, 23 (2017):1-8.
[4] A. Uhlmann, Fidelity and concurrence of conjugated states, Phys. Rev. A 62, 032307 (2000):1-9.
[5] P. Rungta, V. Bužek, C. M. Caves, M. Hillery and G. J. Milburn, Universal state inversion and concurrence in arbitrary
dimensions, Phys. Rev. A 64, 042315 (2001):1-13.
[6] S. Albeverio, and S. M. Fei, A note on invariants and entanglements, J. Opt. B: Quantum Semiclass. Opt. 3, 223 (2001):1-11.
[7] F. Mintert, M. Kuś, and A. Buchleitner, Concurrence of Mixed Bipartite Quantum States in Arbitrary Dimensions, Phys.
Rev. Lett. 92, 167902 (2004):1-4.
[8] K. Chen, S. Albeverio, and S. M. Fei, Concurrence of Arbitrary Dimensional Bipartite Quantum States, Phys. Rev. Lett.
95, 040504 (2005):1-4.
[9] H. P. Breuer, Separability criteria and bounds for entanglement measures, J. Phys. A: Math. Gen. 39, 11847 (2006):1-8.
[10] H. P. Breuer, Optimal Entanglement Criterion for Mixed Quantum States, Phys. Rev. Lett. 97, 080501 (2006):1-4.
[11] J. I. de Vicente, Lower bounds on concurrence and separability conditions, Phys. Rev. A 75, 052320 (2007):1-5.
[12] C. J. Zhang, Y. S. Zhang, S. Zhang, and G. C. Guo, Optimal entanglement witnesses based on local orthogonal observables,
Phys. Rev. A 76, 012334 (2007):1-6.
[13] M. J. Zhao, X. N. Zhu, and S. M. Fei, Lower bound on concurrence and distillation for arbitrary-dimensional bipartite
quantum states, Phys. Rev. A 84, 062322 (2011):1-5.
[14] X. N. Zhu, and S.M. Fei, Lower bound of concurrence for qubit systems, Quantum Inf Process (2014) 13:815-823.
[15] X. N. Zhu, M. J. Zhao, and S. M. Fei, Lower bound of multipartite concurrence based on subquantum state decomposition,
Phys. Rev. A 86 , 022307 (2012):1-4.
[16] A.R.R. Carvalho, F. Mintert, and A. Buchleitner, Decoherence and Multipartite Entanglement, Phys. Rev. Lett. 93,
230501 (2004):1-4.
[17] M. Li, S. M. Fei, X. Q. Li-Jost, and H. Fan, Genuine multipartite entanglement detection and lower bound of multipartite
concurrence, Phys. Rev. A 92, 062338 (2015):1-6.