This paper was converted on www.awesomepapers.org from LaTeX by an anonymous user.
Want to know more? Visit the Converter page.

11institutetext: Zhi-Xi Wang22institutetext: 22email: [email protected]
33institutetext: Shao-Ming Fei 44institutetext: 44email: [email protected]
55institutetext: 1 School of Mathematical Sciences, Capital Normal University, Beijing 100048, China
66institutetext: 2 Max Planck Institute for Mathematics in the Sciences, Leipzig 04103, Germany

Revealing hidden standard tripartite nonlocality by local filtering

Qiao-Qiao Lv1    Jin-Min Liang 1    Zhi-Xi Wang1    Shao-Ming Fei1,2
Abstract

Quantum nonlocality is a kind of significant quantum correlation that is stronger than quantum entanglement and EPR steering. The standard tripartite nonlocality can be detected by the violation of the Mermin inequality. By using local filtering operations, we give a tight upper bound on the maximal expected value of the Mermin operators. By detailed examples we show that the hidden standard nonlocality can be revealed by local filtering which can enhance the robustness of the noised entangled states.

Keywords:
Mermin inequality, Standard tripartite nonlocality, Local filtering

1 Introduction

Nonlocal quantum correlations are the essential nature of quantum physics EPRnonlocal . For bipartite systems the Bell nonlocality BellNonlocal ; Wiseman ; StrongNonlocal gives rise to stronger quantum correlations than quantum entanglement HHHH2009entangle ; LS2019 ; CG2005 and EPR steering steer2020 . As a fascinating counterintuitive phenomenon related to the foundation of quantum mechanics Bell1966 , nonlocality is understood as a resource in quantum information process ranging form quantum computation Liang2020 ; Liang2022 , quantum key distributed AK1991 and random numbers certification Nature2010 . It is significant to detect the nonlocality of given quantum systems.

A bipartite quantum state ρAB\rho_{AB} is said to be locally correlated if the joint probability distributions satisfies Wiseman ,

P(a,b|Ax,By)=λpλP(a|Ax,λ)P(b|By,λ),\displaystyle P(a,b|A_{x},B_{y})=\sum_{\lambda}p_{\lambda}P(a|A_{x},\lambda)P(b|B_{y},\lambda),

where pλp_{\lambda} is the probability distribution over the hidden variables λ\lambda, pλ0p_{\lambda}\geq 0, λpλ=1\sum_{\lambda}p_{\lambda}=1, P(a,b|Ax,By)P(a,b|A_{x},B_{y}) denotes the joint probability distribution that Alice performs measurement AxA_{x} on subsystem AA with outcome aa, and Bob performs measurement ByB_{y} on subsystem BB with outcome bb, P(a|Ax,λ)P(a|A_{x},\lambda) denotes the conditional probability of getting outcome aa when Alice performs the measurement AxA_{x} on her subsystem, P(b|By,λ)P(b|B_{y},\lambda) is similarly defined. The bipartite nonlocality can be detected by the violation of a kind of Bell inequality, the CHSH inequality CHSH1969 .

The maximal violation of a Bell inequality can be enhanced by local filtering operations CHSHviolate . In filterCHSH the authors presented a class of two-qubit entangled states admitting local hidden variable models, and shew that these states violate a Bell inequality after the local filtering. Namely, there exist entangled states whose so called hidden non-locality can be revealed by using a sequence of measurements. In fact, local filtering operations can not only reveal hidden quantum nonlocality, but also hidden quantum steerability FilterSteer .

For tripartite systems, the Mermin inequality Mermin1990 is a natural generalization of the CHSH inequality, which can be violated by not only genuine tripartite nonlocal states but also by standard tripartite nonlocal states. The upper bound on the maximal expectation value of the Mermin operator has been nicely derived in QIP2019 , though not always tight. Similar to the enhanced maximal violation of Bell inequalities, it is interesting to study the enhancement of the maximal violation of the Mermin inequality under local filtering.

In this work, we investigate the maximal violation of the Mermin inequality under local filtering. We first analyze and obtain the tight upper bounds on the maximal expected value of the Mermin operator under local filtering. Then applying our results to some special quantum states, the isotropic states and the noisy GHZ states, we show that local filtering can reveal the hidden standard nonlocality. Moreover, it is shown that the local filtering can transform the initial noisy state to a state with stronger tripartite nonlocality.

2 Tight upper bound on Mermin operator under local filtering

A tripartite state ρABC\rho_{ABC} is fully locally correlated if the joint probability distribution admits a local hidden variable (LHV) model Wiseman , namely,

P(a,b,c|Ax,By,Cz)=λpλP(a|Ax,λ)P(b|By,λ)P(c|Cz,λ)\displaystyle\begin{aligned} P(a,b,c|A_{x},B_{y},C_{z})=\sum_{\lambda}p_{\lambda}P(a|A_{x},\lambda)P(b|B_{y},\lambda)P(c|C_{z},\lambda)\end{aligned}

for all x,y,z,a,bx,y,z,a,b and cc, where P(a,b,c|Ax,By,Cz)P(a,b,c|A_{x},B_{y},C_{z}) is the joint probability when Alice, Bob and Charlie perform local measurements AxA_{x}, ByB_{y} and CzC_{z} with outcomes aa, bb and cc, respectively, pλ0p_{\lambda}\geq 0 is the probability distribution over the hidden variable λ\lambda, λpλ=1\sum_{\lambda}p_{\lambda}=1, P(a|Ax,λ)P(a|A_{x},\lambda) denotes the conditional probability of obtaining outcome aa when Alice performs the measurement AxA_{x} on her subsystem, P(b|By,λ)P(b|B_{y},\lambda) and P(c|Cz,λ)P(c|C_{z},\lambda) are similarly defined.

The tripartite non-locality of arbitrary 3-qubit states ρ\rho can be detected by the violation of the Mermin inequality |ρ||Tr[ρ]|2|\langle\mathcal{M}\rangle_{\rho}|\equiv|\textrm{Tr}[\mathcal{M}\rho]|\leq 2. The Mermin operator \mathcal{M} has the form Mermin1990 ,

=A0(B0C1+B1C0)+A1(B0C0B1C1),\mathcal{M}=A_{0}\otimes(B_{0}C_{1}+B_{1}C_{0})+A_{1}\otimes(B_{0}C_{0}-B_{1}C_{1}),

where A0A_{0}, A1A_{1}, B0B_{0}, B1B_{1}, C0C_{0} and C1C_{1} are quantum mechanical observables of the form K=kσ=i=13kiσiK=\vec{k}\cdot\vec{\sigma}=\sum\limits^{3}_{i=1}k_{i}\sigma_{i}, with a unit vector k{a,a,b,b,c,c}\vec{k}\in\{\vec{a},\vec{a}^{\prime},\vec{b},\vec{b}^{\prime},\vec{c},\vec{c}^{\prime}\}, the standard Pauli matrices σi\sigma_{i}, K{A0,A1,B0,B1,C0,C1}K\in\{A_{0},A_{1},B_{0},B_{1},C_{0},C_{1}\}.

In a recent work QIP2019 it has been shown that the maximal expectation value of the Mermin operator for arbitrary 3-qubit state ρ\rho is given by 𝒬:=max|ρ|22λmax\mathcal{Q}_{\mathcal{M}}:=\max|\langle\mathcal{M}\rangle_{\rho}|\leq 2\sqrt{2}\lambda_{\max}, where λmax\lambda_{\max} is the largest singular value of matrix CC, C=(Cj,ik)C=(C_{j,ik}) is the correlation matrix of ρ\rho with entries given by Cijk=Tr[ρ(σiσjσk)]C_{ijk}=\textrm{Tr}[\rho(\sigma_{i}\otimes\sigma_{j}\otimes\sigma_{k})], i,j,k=1,2,3i,j,k=1,2,3. This upper bound is tight if the degeneracy with respect to the largest singular value λmax\lambda_{\max} is more than 1, and the two degenerate nine-dimensional singular vectors corresponding to λmax\lambda_{\max} take the forms of acac\vec{a}\otimes\vec{c}-\vec{a}^{\prime}\otimes\vec{c}^{\prime} and ac+ac\vec{a}\otimes\vec{c}^{\prime}+\vec{a}^{\prime}\otimes\vec{c}.

Next, we investigate the violations of the Mermin inequality under local filtering, by computing the maximal expectation values of the Mermin operators with respect to the locally filtered 3-qubit states. For any 3-qubit state ρ\rho, after local filtering one gets filterCHSH ,

ρ=1F(FAFBFC)ρ(FAFBFC),\displaystyle\begin{aligned} \rho^{\prime}=\frac{1}{F}(F_{A}\otimes F_{B}\otimes F_{C})\rho(F_{A}\otimes F_{B}\otimes F_{C})^{{\dagger}},\end{aligned} (1)

where FAF_{A}, FBF_{B} and FCF_{C} are positive operators acted locally on the three subsystems respectively, F=Tr[(FAFBFC)ρ(FAFBFC)]F=\textrm{Tr}[(F_{A}\otimes F_{B}\otimes F_{C})\rho(F_{A}\otimes F_{B}\otimes F_{C})^{{\dagger}}] is the normalization constant. Suppose the filter operators FAF_{A}, FBF_{B} and FCF_{C} have the following spectral decompositions,

FA=UΣAU,FB=VΣBV,FC=WΣCW,\displaystyle F_{A}=U\Sigma_{A}U^{{\dagger}},~{}~{}F_{B}=V\Sigma_{B}V^{{\dagger}},~{}~{}F_{C}=W\Sigma_{C}W^{{\dagger}}, (2)

where UU, VV and WW are unitary operators. Set

αi=ΣAσiΣA,βj=ΣBσjΣB,γk=ΣCσkΣC,\displaystyle\begin{aligned} \alpha_{i}=\Sigma_{A}\sigma_{i}\Sigma_{A},~{}~{}\beta_{j}=\Sigma_{B}\sigma_{j}\Sigma_{B},~{}~{}\gamma_{k}=\Sigma_{C}\sigma_{k}\Sigma_{C},\end{aligned} (3)

for i,j,k=1,2,3i,j,k=1,2,3. Without loss of generality, we assume that the singular matrices have the forms,

ΣA=(l001),ΣB=(m001),ΣC=(n001),\displaystyle\begin{aligned} \Sigma_{A}=\left(\begin{array}[]{cc}l&0\\ 0&1\end{array}\right),~{}~{}\Sigma_{B}=\left(\begin{array}[]{cc}m&0\\ 0&1\end{array}\right),~{}~{}\Sigma_{C}=\left(\begin{array}[]{cc}n&0\\ 0&1\end{array}\right),\end{aligned} (4)

with l,m,n0l,m,n\geq 0.

We have the following theorem which provides a tight upper bound on the maximal violation value of the Mermin inequality.

Theorem 2.1

For an arbitrary 3-qubit quantum state ρ\rho, the maximal expectation value of the Mermin operator for the filtered state ρ\rho^{\prime} satifies

max|ρ|22λ,max\displaystyle\begin{aligned} \max|\langle\mathcal{M}\rangle_{\rho^{\prime}}|\leq 2\sqrt{2}\lambda{{}^{\prime}}_{\max},\end{aligned} (5)

where λmax\lambda{{}^{\prime}}_{\max} is the maximal singular value of D~F\frac{\tilde{D}}{F}, where D~=(D~j,ik)\tilde{D}=(\tilde{D}_{j,ik}) with D~ijk=Tr[ρ~(αiβjγk)]\tilde{D}_{ijk}=\textrm{Tr}[\tilde{\rho}(\alpha_{i}\otimes\beta_{j}\otimes\gamma_{k})], ρ~\tilde{\rho} is a state that is locally unitary equivalent to ρ\rho.

Proof

Based on the dual relation between SU(2)SU(2) and SO(3)SO(3) SO1995 , we have UσiU=iOiiσiU\sigma_{i}U^{{\dagger}}=\sum_{i{{}^{\prime}}}O_{ii{{}^{\prime}}}\sigma_{i{{}^{\prime}}}, where UU is a unitary operator and O=(Oij)O=(O_{ij}) belongs to SO(3)SO(3). Therefore, we have

Cijk=Tr[ρ(σiσjσk)]=Tr[(FAFBFC)ρ(FAFBFC)F(σiσjσk)]=1FTr[ρ(UΣAUVΣBVWΣCW)(σiσjσk)(UΣAUVΣBVWΣCW)]=1FTr[(UVW)ρ(UVW)(ΣAUσiUΣAΣBVσjVΣBΣCWσkWΣC)]=1FTr[ρ~(ΣAiOiiσiΣAΣBjOjjσjΣBΣCkOkkσkΣC)]=1FijkOiiOjjOkkTr[ρ~(ΣAσiΣAΣBσjΣBΣCσkΣC)]=1FijkOiiOjjOkkTr[ρ~(αiβjγk)]=1F[OAD~(OBTOCT)]ijk.\displaystyle\begin{aligned} C^{\prime}_{ijk}&=\textrm{Tr}[\rho^{\prime}(\sigma_{i}\otimes\sigma_{j}\otimes\sigma_{k})]\\ &=\textrm{Tr}[\frac{(F_{A}\otimes F_{B}\otimes F_{C})\rho(F_{A}\otimes F_{B}\otimes F_{C})^{{\dagger}}}{F}(\sigma_{i}\otimes\sigma_{j}\otimes\sigma_{k})]\\ &=\frac{1}{F}\textrm{Tr}[\rho(U\Sigma_{A}U^{{\dagger}}\otimes V\Sigma_{B}V^{{\dagger}}\otimes W\Sigma_{C}W^{{\dagger}})(\sigma_{i}\otimes\sigma_{j}\otimes\sigma_{k})(U\Sigma_{A}U^{{\dagger}}\otimes V\Sigma_{B}V^{{\dagger}}\otimes W\Sigma_{C}W^{{\dagger}})]\\ &=\frac{1}{F}\textrm{Tr}[(U^{{\dagger}}\otimes V^{{\dagger}}\otimes W^{{\dagger}})\rho(U\otimes V\otimes W)(\Sigma_{A}U^{{\dagger}}\sigma_{i}U\Sigma_{A}\otimes\Sigma_{B}V^{{\dagger}}\sigma_{j}V\Sigma_{B}\otimes\Sigma_{C}W^{{\dagger}}\sigma_{k}W\Sigma_{C})]\\ &=\frac{1}{F}\textrm{Tr}[\tilde{\rho}(\Sigma_{A}\sum\limits_{i^{\prime}}O_{ii^{\prime}}\sigma_{i^{\prime}}\Sigma_{A}\otimes\Sigma_{B}\sum\limits_{j^{\prime}}O_{jj^{\prime}}\sigma_{j^{\prime}}\Sigma_{B}\otimes\Sigma_{C}\sum\limits_{k^{\prime}}O_{kk^{\prime}}\sigma_{k^{\prime}}\Sigma_{C})]\\ &=\frac{1}{F}\sum\limits_{i^{\prime}j^{\prime}k^{\prime}}O_{ii^{\prime}}O_{jj^{\prime}}O_{kk^{\prime}}\textrm{Tr}[\tilde{\rho}(\Sigma_{A}\sigma_{i^{\prime}}\Sigma_{A}\otimes\Sigma_{B}\sigma_{j^{\prime}}\Sigma_{B}\otimes\Sigma_{C}\sigma_{k^{\prime}}\Sigma_{C})]\\ &=\frac{1}{F}\sum\limits_{i^{\prime}j^{\prime}k^{\prime}}O_{ii^{\prime}}O_{jj^{\prime}}O_{kk^{\prime}}\textrm{Tr}[\tilde{\rho}(\alpha_{i^{\prime}}\otimes\beta_{j^{\prime}}\otimes\gamma_{k^{\prime}})]\\ &=\frac{1}{F}[O_{A}\tilde{D}(O^{T}_{B}\otimes O^{T}_{C})]_{ijk}.\end{aligned} (6)

Hence, C=OAD~(OBTOCT)FC^{\prime}=\frac{O_{A}\tilde{D}(O^{T}_{B}\otimes O^{T}_{C})}{F}, and (C)C=1F2(OBOC)D~OAOAD~(OBOC)=1F2(OBOC)D~D~(OBOC)(C^{\prime})^{{\dagger}}C^{\prime}=\frac{1}{F^{2}}(O_{B}\otimes O_{C})\tilde{D}^{{\dagger}}O_{A}^{{\dagger}}O_{A}\tilde{D}(O_{B}\otimes O_{C})^{{\dagger}}=\frac{1}{F^{2}}(O_{B}\otimes O_{C})\tilde{D}^{{\dagger}}\tilde{D}(O_{B}\otimes O_{C})^{{\dagger}}. As OBO_{B} and OCO_{C} belong to SO(3)SO(3), (C)C(C^{\prime})^{{\dagger}}C^{\prime} has the same eigenvalues as D~D~F2\frac{\tilde{D}^{{\dagger}}\tilde{D}}{F^{2}}. That is to say, λmax\lambda{{}^{\prime}}_{\max} is also the maximal singular value of D~F\frac{\tilde{D}}{F}.

Remark The normalization factor FF has the following form

F=Tr[(FAFBFC)ρ(FAFBFC)]=Tr[(UΣAUVΣBVWΣCW)ρ(UΣAUVΣBVWΣCW)]=Tr[ρ(UΣA2UVΣB2VWΣC2W)]=Tr[(UVW)ρ(UVW)(ΣA2ΣB2ΣC2)]=Tr[ρ~(ΣA2ΣB2ΣC2)],\displaystyle\begin{aligned} F&=\textrm{Tr}[(F_{A}\otimes F_{B}\otimes F_{C})\rho(F_{A}\otimes F_{B}\otimes F_{C})^{{\dagger}}]\\ &=\textrm{Tr}[(U\Sigma_{A}U^{{\dagger}}\otimes V\Sigma_{B}V^{{\dagger}}\otimes W\Sigma_{C}W^{{\dagger}})\rho(U\Sigma_{A}U^{{\dagger}}\otimes V\Sigma_{B}V^{{\dagger}}\otimes W\Sigma_{C}W^{{\dagger}})]\\ &=\textrm{Tr}[\rho(U\Sigma^{2}_{A}U^{{\dagger}}\otimes V\Sigma^{2}_{B}V^{{\dagger}}\otimes W\Sigma^{2}_{C}W^{{\dagger}})]\\ &=\textrm{Tr}[(U^{{\dagger}}\otimes V^{{\dagger}}\otimes W^{{\dagger}})\rho(U\otimes V\otimes W)(\Sigma^{2}_{A}\otimes\Sigma^{2}_{B}\otimes\Sigma^{2}_{C})]\\ &=\textrm{Tr}[\tilde{\rho}(\Sigma^{2}_{A}\otimes\Sigma^{2}_{B}\otimes\Sigma^{2}_{C})],\end{aligned}

where ρ~\tilde{\rho} and ρ\rho are local unitary equivalent. They have the same maximal violation value of the Mermin inequality.

Note that the inequality (5) saturates if the degeneracy of λmax\lambda^{\prime}_{\max} is more than 1, and the two nine-dimensional singular vectors corresponding to λmax\lambda_{\max} take the forms of acac\vec{a}\otimes\vec{c}-\vec{a}^{\prime}\otimes\vec{c}^{\prime} and ac+ac\vec{a}\otimes\vec{c}{{}^{\prime}}+\vec{a}{{}^{\prime}}\otimes\vec{c}, respectively.

To illustrate the theorem let us consider the following examples.

Example 1. Consider the 3-qubit mixed Greenberger-Horne-Zeilinger (GHZ) state EX12015 ,

ρGHZ=p|GHZGHZ|+1p4I2I~,\rho_{GHZ}=p|GHZ\rangle\langle GHZ|+\frac{1-p}{4}I_{2}\otimes\tilde{I},

where 0p10\leq p\leq 1, |GHZ=|000+|1112|GHZ\rangle=\frac{|000\rangle+|111\rangle}{\sqrt{2}}, I2I_{2} is the 2×22\times 2 identity matrix and I~=\tilde{I}=diag(1,0,0,1)(1,0,0,1). The state ρGHZ\rho_{GHZ} is shown to be genuine multipartite entangled for 13<p1\frac{1}{3}<p\leq 1, and it admits bilocal hidden model for 0p0.416670\leq p\leq 0.41667 EX12015 . Later, Li et al. pointed out that ρGHZ\rho_{GHZ} is genuine multipartite nonlocal GN2013 ; LM2017 for 0.707107<p10.707107<p\leq 1, namely, it violates the Svetlichny inequality (SI) Svetlichny1987 when 0.707107<p10.707107<p\leq 1. The maximal violation of the Svetlichny inequality under local filtering has been also calculated FOP . Recently, the upper bound of the Mermin operator has been studied in QIP2019 , which shows that the state violates the Mermin inequality if 12<p1\frac{1}{2}<p\leq 1, i.e., the state is standard nonlocal for 12<p1\frac{1}{2}<p\leq 1.

By direct calculation, we have the correlation matrix of ρGHZ\rho_{GHZ},

C=(p000p00000p0p00000000000000)\displaystyle\begin{aligned} C=\left(\begin{array}[]{ccccccccc}p&0&0&0&-p&0&0&0&0\\ 0&-p&0&-p&0&0&0&0&0\\ 0&0&0&0&0&0&0&0&0\end{array}\right)\end{aligned} (7)

and

D=(plmn000plmn00000plmn0plmn0000000000000T),\displaystyle\begin{aligned} D=\left(\begin{array}[]{ccccccccc}plmn&0&0&0&-plmn&0&0&0&0\\ 0&-plmn&0&-plmn&0&0&0&0&0\\ 0&0&0&0&0&0&0&0&T\end{array}\right),\end{aligned} (8)

where D=(Dj,ik)D=(D_{j,ik}), Dijk=Tr[ρGHZ(αiβjγk)]D_{ijk}=\textrm{Tr}[\rho_{GHZ}(\alpha_{i}\otimes\beta_{j}\otimes\gamma_{k})]. The singular values of DD are 2plmn\sqrt{2}plmn, 2plmn\sqrt{2}plmn and T=(l21)(m2n2+1)4+(l2+1)(m2n21)4pT=\frac{(l^{2}-1)(m^{2}n^{2}+1)}{4}+\frac{(l^{2}+1)(m^{2}n^{2}-1)}{4}p. ρ~GHZ\tilde{\rho}_{GHZ} is locally unitary equivalent to ρGHZ\rho_{GHZ}. Then we have that 2plmnF\frac{\sqrt{2}plmn}{F}, 2plmnF\frac{\sqrt{2}plmn}{F} and TF\frac{T}{F} are the singular values of the matrix D~F\frac{\tilde{D}}{F}, where

F=Tr[ρGHZ(ΣA2ΣB2ΣC2)]=(l2+1)(m2n2+1)4+(l21)(m2n21)4p.\displaystyle F=\textrm{Tr}[\rho_{GHZ}(\Sigma^{2}_{A}\otimes\Sigma^{2}_{B}\otimes\Sigma^{2}_{C})]=\frac{(l^{2}+1)(m^{2}n^{2}+1)}{4}+\frac{(l^{2}-1)(m^{2}n^{2}-1)}{4}p.

The maximal singular value is λmax=2plmnF\lambda^{{}^{\prime}}_{\max}=\frac{\sqrt{2}plmn}{F} for given pp with 2plmnF>TF\frac{\sqrt{2}plmn}{F}>\frac{T}{F}. Then the upper bound of the maximal value of the Mermin operator is 22λ=max4plmnF2\sqrt{2}\lambda{{}^{\prime}}_{\max}=\frac{4plmn}{F}. Two singular vectors corresponding to the singular value λmax\lambda{{}^{\prime}}_{\max} with degeneracy 2 can be chosen as v1=(1,0,0,0,1,0,0,0,0)T\vec{v}_{1}=(-1,0,0,0,1,0,0,0,0)^{T} and v2=(0,1,0,1,0,0,0,0,0)T\vec{v}_{2}=(0,1,0,1,0,0,0,0,0)^{T}, which can be further written as

v1=(1,0,0)T(1,0,0)T(0,1,0)T(0,1,0)T,\displaystyle\vec{v}_{1}=(1,0,0)^{T}\otimes(-1,0,0)^{T}-(0,-1,0)^{T}\otimes(0,1,0)^{T},
v2=(1,0,0)T(0,1,0)T+(0,1,0)T(1,0,0)T.\displaystyle\vec{v}_{2}=(1,0,0)^{T}\otimes(0,1,0)^{T}+(0,-1,0)^{T}\otimes(-1,0,0)^{T}.

By taking a=(1,0,0)T\vec{a}=(1,0,0)^{T}, a=(0,1,0)T\vec{a}^{\prime}=(0,-1,0)^{T}, c=(1,0,0)T\vec{c}=(-1,0,0)^{T}, c=(0,1,0)T\vec{c}^{\prime}=(0,1,0)^{T}, b\vec{b} and b\vec{b}^{\prime} some suitable unit vectors, the upper bound 4plmnF\frac{4plmn}{F} of ρGHZ\rho_{GHZ} is attained. Therefore, the state violates the Mermin inequality if 4plmnF>2\frac{4plmn}{F}>2 under the restriction 2plmnF>TF\frac{\sqrt{2}plmn}{F}>\frac{T}{F}. As a result, the standard nonlocality of the state ρGHZ\rho^{\prime}_{GHZ} can be detected by the Mermin inequality for 0.471428<p10.471428<p\leq 1. However, the state violates the Mermin inequality if 12<p1\frac{1}{2}<p\leq 1 QIP2019 . Hence, the hidden standard nonlocality of ρGHZ\rho_{GHZ} is revealed by local filtering operation for 0.471428p0.50.471428\leq p\leq 0.5, see FIG. 1.

Refer to caption
Figure 1: The state ρGHZ\rho_{GHZ} violates Mermin inequality (MI) for 0.5<p10.5<p\leq 1. The locally filtered state shows standard nonlocality for 0.471428<p10.471428<p\leq 1. The hidden standard tripartite nonlocality is revealed for 0.471428p0.50.471428\leq p\leq 0.5.

Example 2. Consider the following state given in EX22007 ,

ρ=p|ΨΨ|+(1p)|0000|I22,\displaystyle\begin{aligned} \rho=p|\Psi\rangle\langle\Psi|+(1-p)|00\rangle\langle 00|\otimes\frac{I_{2}}{2},\end{aligned} (9)

where 0p10\leq p\leq 1, |Ψ=cosπ8|000+sinπ8|111|\Psi\rangle=\cos\frac{\pi}{8}|000\rangle+\sin\frac{\pi}{8}|111\rangle. Under local filtering we have

D=(plmn2000plmn200000plmn20plmn20000000000000T).\displaystyle\begin{aligned} D=\left(\begin{array}[]{ccccccccc}\frac{plmn}{\sqrt{2}}&0&0&0&-\frac{plmn}{\sqrt{2}}&0&0&0&0\\ 0&-\frac{plmn}{\sqrt{2}}&0&-\frac{plmn}{\sqrt{2}}&0&0&0&0&0\\ 0&0&0&0&0&0&0&0&T\end{array}\right).\end{aligned} (10)

The singular values of DD are plmnplmn, plmnplmn and T=l2m2(n21)2+2+2+2l2m2+2l2m2n24pT=\frac{l^{2}m^{2}(n^{2}-1)}{2}+\frac{-2+\sqrt{2}+2l^{2}m^{2}+\sqrt{2}l^{2}m^{2}n^{2}}{4}p. Since ρ~\tilde{\rho} is locally unitary equivalent to ρ\rho, the singular value of the matrix D~F\frac{\tilde{D}}{F} are plmnF\frac{plmn}{F}, plmnF\frac{plmn}{F} and TF\frac{T}{F}, where

F=Tr[ρ(ΣA2ΣB2ΣC2)]=l2m2(n2+1)2+222l2m2+2l2m2n24p.\displaystyle F=\textrm{Tr}[\rho(\Sigma^{2}_{A}\otimes\Sigma^{2}_{B}\otimes\Sigma^{2}_{C})]=\frac{l^{2}m^{2}(n^{2}+1)}{2}+\frac{2-\sqrt{2}-2l^{2}m^{2}+\sqrt{2}l^{2}m^{2}n^{2}}{4}p.

The maximal singular value is λmax=plmnF\lambda^{\prime}_{\max}=\frac{plmn}{F} for given pp with plmnF>TF\frac{plmn}{F}>\frac{T}{F}. Then the upper bound of the maximal value of the Mermin operator is 22λmax=22plmnF2\sqrt{2}\lambda^{\prime}_{\max}=\frac{2\sqrt{2}plmn}{F}. This bound can be attained by selecting the two singular vectors, corresponding to the singular value λmax\lambda^{\prime}_{\max} with degeneracy 2, to be v1=(1,0,0,0,1,0,0,0,0)T\vec{v}_{1}=(-1,0,0,0,1,0,0,0,0)^{T} and v2=(0,1,0,1,0,0,0,0,0)T\vec{v}_{2}=(0,1,0,1,0,0,0,0,0)^{T}, which can be decomposed to

v1=(1,0,0)T(1,0,0)T(0,1,0)T(0,1,0)T,\displaystyle\vec{v}_{1}=(1,0,0)^{T}\otimes(-1,0,0)^{T}-(0,-1,0)^{T}\otimes(0,1,0)^{T},
v2=(1,0,0)T(0,1,0)T+(0,1,0)T(1,0,0)T.\displaystyle\vec{v}_{2}=(1,0,0)^{T}\otimes(0,1,0)^{T}+(0,-1,0)^{T}\otimes(-1,0,0)^{T}.

Let a=(1,0,0)T\vec{a}=(1,0,0)^{T}, a=(0,1,0)T\vec{a}^{\prime}=(0,-1,0)^{T}, c=(1,0,0)T\vec{c}=(-1,0,0)^{T}, c=(0,1,0)T\vec{c}^{\prime}=(0,1,0)^{T}. Together with some suitable unit vectors b\vec{b} and b\vec{b}^{\prime}, the upper bound 22plmnF\frac{2\sqrt{2}plmn}{F} is attained. Therefore, the state violates the Mermin inequality if 22plmnF>2\frac{2\sqrt{2}plmn}{F}>2 under the restriction plmnF>TF\frac{plmn}{F}>\frac{T}{F}, namely, the state ρ\rho violates the Mermin inequality if 0.318675<p10.318675<p\leq 1, for which the standard nonlocality of the state ρ\rho^{\prime} is detected. The maximal violation of the Mermin inequality is shown in FIG. 2.

Refer to caption
Figure 2: f(p)f(p) is the maximal value of 𝒬(M)\mathcal{Q}(M). The red line represents the maximal violation of the locally filtered state ρ\rho^{\prime}. The blue line represents the maximal violation value of the initial state ρ\rho.

Based on the protocol introduced in QIP2019 , ρ\rho is standard tripartite nonlocal for 0.707107<p10.707107<p\leq 1. Therefore, the state ρ\rho shows hidden standard tripartite nonlocality for 0.318675p0.7071070.318675\leq p\leq 0.707107, see FIG. 3.

Refer to caption
Figure 3: The state ρ\rho does not violate SI for 0p10\leq p\leq 1. It violates the Mermin inequality for 0.707107<p10.707107<p\leq 1. After locally filtering ρ\rho^{\prime} is standard nonlocal for 0.318675<p10.318675<p\leq 1. The hidden standard nonlocality is revealed for 0.318675p0.7071070.318675\leq p\leq 0.707107.

Example 3. The interaction between a quantum system and its environment may reduce the entanglement and nonlocality of the system. The GHZ state is a genuine tripartite nonlocal state as violates the Svetlichny inequality. Let us consider that the GHZ state goes through the amplitude damping(AD) noise channel which maps a qubit state ρ\rho to AD(ρ)=E0ρE0+E1ρE1\mathcal{E}_{AD}(\rho)=E_{0}\rho E^{{\dagger}}_{0}+E_{1}\rho E^{{\dagger}}_{1}, where the Kraus operators are given by Nielsen ,

E0=(1001γ),E1=(0γ00),E_{0}=\left(\begin{array}[]{cc}1&0\\ 0&\sqrt{1-\gamma}\end{array}\right),~{}~{}~{}E_{1}=\left(\begin{array}[]{cc}0&\sqrt{\gamma}\\ 0&0\end{array}\right),

γ(0,1)\gamma\in(0,1) is the damping rate.

When each qubit of the GHZ state ρG\rho_{G} undergoes the amplitude damping noise channel, one gets

ρGADAD(ρG)=12((1+γ3)|000000|+(1γ)3|111111|+(1γ)3/2(|000111|+|111000|)+(1γ)γ2(|001001|+|010010|+|100100|)+(1γ)2γ(|011011|+|101101|+|110110|)).\displaystyle\begin{aligned} \rho_{G}^{AD}&\equiv\mathcal{E}_{AD}(\rho_{G})\\ &=\frac{1}{2}\Big{(}(1+\gamma^{3})|000\rangle\langle 000|+(1-\gamma)^{3}|111\rangle\langle 111|\\ &\quad+(1-\gamma)^{3/2}(|000\rangle\langle 111|+|111\rangle\langle 000|)\\ &\quad+(1-\gamma)\gamma^{2}(|001\rangle\langle 001|+|010\rangle\langle 010|+|100\rangle\langle 100|)\\ &\quad+(1-\gamma)^{2}\gamma(|011\rangle\langle 011|+|101\rangle\langle 101|+|110\rangle\langle 110|)\Big{)}.\end{aligned} (11)

The corresponding correlation matrix is

C=(s000s00000s0s0000000000000t),\displaystyle\begin{aligned} C=\left(\begin{array}[]{ccccccccc}s&0&0&0&-s&0&0&0&0\\ 0&-s&0&-s&0&0&0&0&0\\ 0&0&0&0&0&0&0&0&t\end{array}\right),\end{aligned} (12)

where s=(1γ)32s=(1-\gamma)^{\frac{3}{2}} and t=γ(36γ+4γ2)t=\gamma(3-6\gamma+4\gamma^{2}). The singular values of CC are 2(1γ)32\sqrt{2(1-\gamma)^{\frac{3}{2}}}, 2(1γ)32\sqrt{2(1-\gamma)^{\frac{3}{2}}} and γ(4γ26γ+3)\gamma(4\gamma^{2}-6\gamma+3). We choose the two nine-dimension vectors to be v1=(1,0,0,0,1,0,0,0,0)T\vec{v}_{1}=(-1,0,0,0,1,0,0,0,0)^{T} and v2=(0,1,0,1,0,0,0,0,0)T\vec{v}_{2}=(0,1,0,1,0,0,0,0,0)^{T}, which can be decomposed into

v1=(1,0,0)T(1,0,0)T(0,1,0)T(0,1,0)T,\displaystyle\vec{v}_{1}=(1,0,0)^{T}\otimes(-1,0,0)^{T}-(0,-1,0)^{T}\otimes(0,1,0)^{T},
v2=(1,0,0)T(0,1,0)T+(0,1,0)T(1,0,0)T.\displaystyle\vec{v}_{2}=(1,0,0)^{T}\otimes(0,1,0)^{T}+(0,-1,0)^{T}\otimes(-1,0,0)^{T}.

Let a=(1,0,0)T\vec{a}=(1,0,0)^{T}, a=(0,1,0)T\vec{a}^{\prime}=(0,-1,0)^{T}, c=(1,0,0)T\vec{c}=(-1,0,0)^{T}, c=(0,1,0)T\vec{c}^{\prime}=(0,1,0)^{T}. Together with suitable unit vectors b\vec{b} and b\vec{b}^{\prime}, the upper bound of the maximal expectation of the Mermin operator is 4(1γ)324\sqrt{(1-\gamma)^{\frac{3}{2}}} based on Mermin1990 . Hence, the state is standard tripartite nonlocal for γ(0,0.370039)\gamma\in(0,0.370039).

Now consider the filtering. The correlation matrix of the filtered state is

D=(S000S00000S0S0000000000000T),\displaystyle\begin{aligned} D=\left(\begin{array}[]{ccccccccc}S&0&0&0&-S&0&0&0&0\\ 0&-S&0&-S&0&0&0&0&0\\ 0&0&0&0&0&0&0&0&T\end{array}\right),\end{aligned} (13)

where S=lmn(1γ)32S=lmn(1-\gamma)^{\frac{3}{2}}. The singular values of DD are 2(1p)3l2m2n2\sqrt{2(1-p)^{3}l^{2}m^{2}n^{2}}, 2(1p)3l2m2n2\sqrt{2(1-p)^{3}l^{2}m^{2}n^{2}} and

T\displaystyle T =\displaystyle= 12(1+l2m2n2+γ3(l2+1)(m2+1)(n2+1)+γ(3+l2+m2+n2)\displaystyle\frac{1}{2}(-1+l^{2}m^{2}n^{2}+\gamma^{3}(l^{2}+1)(m^{2}+1)(n^{2}+1)+\gamma(3+l^{2}+m^{2}+n^{2})
γ2(3+2n2+m2(2+n2)+l2(2+m2+n2))).\displaystyle-\gamma^{2}(3+2n^{2}+m^{2}(2+n^{2})+l^{2}(2+m^{2}+n^{2}))).

As ρ~GAD\tilde{\rho}^{AD}_{G} is locally unitary equivalent to ρGAD\rho^{AD}_{G}, the singular values of the matrix D~F\frac{\tilde{D}}{F} are 2(1p)3l2m2n2F\frac{\sqrt{2(1-p)^{3}l^{2}m^{2}n^{2}}}{F}, 2(1p)3l2m2n2F\frac{\sqrt{2(1-p)^{3}l^{2}m^{2}n^{2}}}{F} and TF\frac{T}{F}, where

F\displaystyle F =\displaystyle= 12(1+l2m2n2+γ3(l21)(m21)(n21)\displaystyle\frac{1}{2}(1+l^{2}m^{2}n^{2}+\gamma^{3}(l^{2}-1)(m^{2}-1)(n^{2}-1)
+γ(3+l2+m2+n2)+γ2(32n2+m2(2+n2)+l2(2+m2+n2))).\displaystyle+\gamma(-3+l^{2}+m^{2}+n^{2})+\gamma^{2}(3-2n^{2}+m^{2}(-2+n^{2})+l^{2}(-2+m^{2}+n^{2}))).

The maximal violation value of the Mermin operator is 4(1γ)3l2m2n2F\frac{4\sqrt{(1-\gamma)^{3}l^{2}m^{2}n^{2}}}{F}, with the restriction 2(1p)3l2m2n2FTF\frac{\sqrt{2(1-p)^{3}l^{2}m^{2}n^{2}}}{F}\geq\frac{T}{F}. Therefore, the filtered state violates the Mermin inequality for γ(0,0.394752)\gamma\in(0,0.394752). That is to say, for the state with 0.370069γ0.3947520.370069\leq\gamma\leq 0.394752, the state ρGAD\rho^{AD}_{G} is not a standard nonlocal state. FIG. 4 shows that for γ(0,0.394752)\gamma\in(0,0.394752), the filtered state violates the Mermin inequality, i.e., it is a standard tripartite nonlocal state.

Refer to caption
Figure 4: f(γ)f(\gamma) is the maximal value of 𝒬(M)\mathcal{Q}(M). The red line represents the maximal violation value of the state after local filtering. The filtered state is standard nonlocal for 0γ<0.3947520\leq\gamma<0.394752.

3 Conclusion

In summary, we have investigated the maximal violation of the Mermin inequality under local filtering for any 3-qubit states. We have presented a tight upper bound for the maximal expectation value of the Mermin operator after local filtering. Furthermore, for the 3-qubit GHZ state, the standard tripartite nonlocal be revealed for 0.471428p0.50.471428\leq p\leq 0.5 by local filtering. Similarly, although the amplitude damping GHZ state is fully local for 0.370069γ0.3947520.370069\leq\gamma\leq 0.394752, the filtered one is standard nonlocal. The local filtering process may reveal certain hidden quantum correlations including nonlocality filterCHSH ; FOP and steerability FilterSteer . In order to improve the efficiency of quantum information process, a number of scheme have been put froward purify ; ErrorCorrection ; EntanglementConcentration ; QuantumRepeaters . The filter operations can also be used to improve the fidelity between quantum states and efficiency of information processing with noisy entangled state FilterNoise . Our approach presented in this article can also be used to deal with other Bell-type inequalities for tripartite or multipartite systems.

Acknowledgements.
This work is supported by NSFC (Grant Nos. 12075159, 12171044), Beijing Natural Science Foundation (Z190005), Academy for Multidisciplinary Studies, Capital Normal University, the Academician Innovation Platform of Hainan Province, and Shenzhen Institute for Quantum Science and Engineering, Southern University of Science and Technology (SIQSE202001).

Data Availability Statements All data generated or analysed during this study are available from the corresponding author on reasonable request.

References

  • (1) Einstein, A., Podolsky, B., and Rosen, N.: Can Quantum-Mechanical Description of Physical Reality Be Considered Complete? Phys. Rev. 47, 777 (1935).
  • (2) Brunner, N., Cavalcanti, D., Pironio, S., Scarani, V., and Wehner, S.: Bell nonlocality, Rev. Mod. Phys. 86, 419 (2014).
  • (3) Wiseman, H. M., Jones, S. J., and Doherty, A. C.: Steering, Entanglement, Nonlocality, and the Einstein-Podolsky-Rosen Paradox, Phys. Rev. Lett. 98, 140402 (2007).
  • (4) Halder, S., Banik, M., Agrawal, S., and Bandyopadhyay, S.: Strong Quantum Nonlocality without Entanglement, Phys. Rev. Lett. 120, 040403 (2019).
  • (5) Horodecki, R., Horodecki, P., Horodecki, M., and Horodecki, K.: Quantum entanglement, Rev. Mod. Phys. 81, 865 (2009).
  • (6) Levine, Y., Sharir, O., Cohen, N., and Shashua, A.: Quantum Entanglement in Deep Learning Architectures, Phys. Rev. Lett. 122, 065301 (2019).
  • (7) Cerf, V. J., Gisin, N., Massar, S., and Popescu, S.: Simulating Maximal Quantum Entanglement without Communication, Phys. Rev. Lett. 94, 220403 (2005).
  • (8) Uola, R., Costa, A. C. S., Nguyen, H. C., and Gühne, O.: Quantum steering, Rev. Mod. Phys. 92, 015001 (2020).
  • (9) BELL, J. S.: On the Problem of Hidden Variables in Quantum Mechanics, Rev. Mod. Phys. 38, 447 (1966).
  • (10) Liang, J. M., Shen, S. Q., Li, M., and Li, L.: Variational quantum algorithms for dimensionality reduction and classification, Phys. Rev. A 101, 032323 (2020).
  • (11) Liang, J. M., Wei, S. J., and Fei, S. M.: Quantum gradient descent algorithms for nonequilibrium steady states and linear algebraic systems, Sci. China Phys. Mech. Astron. 65, 250313 (2022).
  • (12) Ekert, A. K.: Quantum Cryptography based on Bell’s theorem, Phys. Rev. Lett. 67, 661 (1991).
  • (13) Pironio, S., Acín, A., Massar, S., Boyer de la Giroday, A., Matsukevich, D. N., Maunz, P., Olmschenk, S., Hayes, D., Luo, L., Manning, T. A., and Monroe, C.: Random numbers certified by Bell’s theorem, Nature 464, 1021 (2010).
  • (14) Clauser, J. F., Horne, M. A., Shimony, A., and Holt, R. A.: Proposed Experiment to Test Local Hidden-Variable Theories, Phys. Rev. Lett. 23, 880 (1969).
  • (15) Verstraete, F. and Wolf, M. M.: Entanglement versus Bell Violations and Their Behavior under Local Filtering Operations, Phys. Rev. Lett. 89, 170401 (2001).
  • (16) Hirsch, F., Quintino, M. T., Bowles, J., and Brunner, N.: Genuine Hidden Quantum Nonlocality, Phys. Rev. Lett. 111, 160402 (2013).
  • (17) Pramanik, T., Cho, Y. W., Han, S. W., Lee, S. Y., Kim, Y. S., and Moon, S.: Revealing hidden quantum steerability using local filtering operations, Phys. Rev. A 99, 030101(R) (2019).
  • (18) Mermin, N. D.: Extreme Quantum Entanglement in a Superposition of Macroscopically Distinct States, Phys. Rev. Lett. 65, 1838 (1990).
  • (19) Siddiqui, M. A. and Sazim, S.: Tight upper bound for the maximal expectation value of the Mermin operators, Quantum. Inf. Process. 18, 131 (2019).
  • (20) Sun, L. Y., Xu, L., Wang, J., Li, M., Shen, S. Q., Li, L., and Fei, S. M.: Tight upper bound on the quantum value of Svetlichny operators under local filtering and hidden genuine nonlocality, Front. Phys. 16, 31501 (2021).
  • (21) Schlienz, J. and Mahler, G.: Description of entanglement, Phys. Rev. A 52, 4396 (1995).
  • (22) Augusiak, R., Demianowicz, M., Tura, J., and Acín, A.: Entanglement and Nonlocality are Inequivalent for Any Number of Parties, Phys. Rev. Lett. 115, 030404 (2015).
  • (23) Bancal, J. D., Barrett, J., Gisin, N., and Pironio, S.: Definitions of multipartite nonlocality, Phys. Rev. A 88, 014102 (2013).
  • (24) Li, M., Shen, S. Q., Jing, N. H., Fei, S. M., and Li-Jost, X. Q.: Tight upper bound for the maximal quantum value of the Svetlichny operators, Phys. Rev. A 96, 042323 (2017).
  • (25) Svetlichny, G.: Distinguishing three-body from two-body nonseparability by a Bell-type inequality, Phys. Rev. D 35, 3066 (1987).
  • (26) Almeida, M. L., Pironio, S., Barrett, J., Tóth, G., and Acín, A.: Noise Robustness of the Nonlocality of Entangled Quantum States, Phys. Rev. Lett. 99, 040403 (2007).
  • (27) Nielsen, M. A. and Chuang, I. L.: Quantum Computation and Quantum Information, (Cambridge University Press, Cambridge, England, 2000).
  • (28) Bennett, C. H., Brassard, G., Popescu, S., Schumacher, B., Smolin, J. A., and Wootters, W. K.: Purification of Noisy Entanglement and Faithful Teleportation via Noisy Channels, Phys. Rev. Lett. 76, 722 (1997).
  • (29) Bennett, C. H., DiVincenzo, D. P., Smolin, J. A., and Wootters, W. K.: Mixed-state entanglement and quantum error correction, Phys. Rev. A 54, 3824 (1996).
  • (30) Zhao, Z., Pan, J. W., and Zhan, M. S.: Practical scheme for entanglement concentration, Phys. Rev. A 64, 014301 (2001).
  • (31) Briegel, H. J., Dür, W., Cirac, J. I., and Zoller, P.: Quantum Repeaters: The Role of Imperfect Local Operations in Quantum Communication, Phys. Rev. Lett. 81, 5932 (1998).
  • (32) Huang, Y. S., Xing, H. B., Yang, M., Yang, Q., Song, W., and Cao, Z. L.: Distillation of multipartite entanglement by local filtering operations, Phys. Rev. A 89, 062320 (2014).