ebook img

Entangled Nonorthogonal States and Their Decoherence Properties PDF

13 Pages·0.12 MB·English
Save to my drive
Quick download
Download
Most books are stored in the elastic cloud where traffic is expensive. For this reason, we have a limit on daily download.

Preview Entangled Nonorthogonal States and Their Decoherence Properties

Entangled Nonorthogonal States and Their Decoherence Properties Osamu Hirota1,2, Steven J. van Enk+, Kazuo Nakamura∗1,2, Masaki Sohma†, and Kentaro Kato1,2 1.Tamagawa University, Machida, Tokyo, Japan. 1 0 2. CREST, JST:Japan Science and Technology. 0 2 +Bell Labs, Lucent Technologies, Murray Hill, NJ, USA n a J ∗1 Fundamental Research Laboratory, NEC, Tsukuba, Japan. 9 1 ∗2 CREST, JST:Japan Science and Technology. 1 † Matsushita Research Institute Inc. Tokyo, Japan. v 6 9 0 1 0 Abstract 1 0 / h This paper presents properties of the so-called quasi-Bell states: entangled p - t states written as superpositions of nonorthogonal states. It is shown that a n a u specialclassofthosestates,namelyentangledcoherentstates,aremorerobust q : v against decoherence due to photon absorption than the standard bi-photon i X Bell states. r a PACS numbers: 03.65.Bz Typeset using REVTEX 1 I. INTRODUCTION Entanglement and its information-theoretic aspects have been studied by many authors [1–5]. Here we give a short survey of the theory of entanglement that we will later apply to quasi-Bell states. For a pure entangled state of a bipartite system Ψ , the measure of AB | i entanglement defined as [1,6] E( Ψ ) = Tr ρ logρ , ρ = Tr Ψ Ψ , (1) AB A A A A B AB | i − | i h | is called the “entropy of entanglement”. This quantity enjoys two kinds of information- theoretic interpretations. One is that E gives the entanglement of formation, which is defined as the asymptotic number k of standard singlet states required to faithfully locally prepare n identical copies of a system in the bipartite state Ψ for very large k and n. AB | i The other is that E gives the amount of distillable entanglement, which is the asymptotic number of singlets k that can be distilled from n identical copies of Ψ . With either of AB | i these definitions of k and n, E satisfies k lim = E( Ψ ). (2) AB n,k→∞ n | i For pure states we can rewrite 1 E( Ψ ) = H (1+ 1 C( Ψ )2) (3) AB AB | i (cid:18)2 q − | i (cid:19) where H(x) = xlogx (1 x)log(1 x) is the binary entropy function and C( Ψ ) is AB − − − − | i the “concurrence” defined by C( Ψ ) = Ψ Ψ˜ with Ψ˜ = σ Ψ ∗ . The above | iAB |ABh | iAB| | iAB | iAB expression is valid for mixed states of two qubit systems as well [4]. For mixed states of qubits one may also define an expression for the entanglement of formation [1,4,5]. It is defined as the average entanglement of the pure states of a decom- position ρ = p ψ ψ of the density operator ρ, minimized over all decompositions [1] i i| iih i| P E(ρ) = min p E( ψ ). (4) i i | i X 2 In general, it is difficult to find the exact amount of entanglement of formation except for special cases. However, there is a lower bound which is expressed in terms of a quantity called the “fully entangled fraction”, which we denote by f(ρ) and is defined as f(ρ) = max e ρ e , (5) h | | i where the maximum is over all completely entangled states e . A lower bound on the | i entanglement of formation is [1] E(ρ) h[f(ρ)], (6) ≥ where H[1 + f(1 f)] (f 1) h[f(ρ)] =  2 q − ≥ 2 (7)  0 (f < 1) 2  Usually in order to construct entangled states one writes superpositions of orthogonalstates. Forinstance thestandardBell basisuses states like and ,andofcourseitsproperties | li | ↔i arewell known. Our concern hereis what kindof properties appearif we have superpositions of nonorthogonal states. In this paper, we will clarify properties of entangled states of nonorthogonal states such as coherent states based on the above basic theory. II. QUASI-BELL STATES A. General definition Let us consider entangled states based on two nonorthogonal states ψ and ψ such 1 2 | i | i that ψ ψ = κ where κ is real. We can define a set of 4 entangled states as follows: 1 2 h | i Ψ = h ( ψ ψ + ψ ψ ) | 1iAB 1 | 1iA| 2iB | 2iA| 1iB |Ψ2iAB = h2(|ψ1iA|ψ2iB −|ψ2iA|ψ1iB) (8) Ψ = h ( ψ ψ + ψ ψ ) 3 AB 3 1 A 1 B 2 A 2 B | i | i | i | i | i |Ψ4iAB = h4(|ψ1iA|ψ1iB −|ψ2iA|ψ2iB) 3 where h are normalization constants: h = h = 1/ 2(1+κ2), h = h = 1/ 2(1 κ2). i 1 3 2 4 { } q q − We call these states “quasi-Bell states”. They are not orthogonal to each other. In fact for κ real their Gram matrix G = Ψ Ψ becomes ij AB i j AB | h | i | 1 0 D 0    0 1 0 0   G =   (9)    D 0 1 0        0 0 0 1     where D = 2κ/(1+κ2). If the basic states are orthogonal (κ = 0), then these states reduce to standard Bell states. Let us discuss the entropy of entanglement for the above states. We first calculate the reduced density operators of the quasi-Bell states. They are ρ (1) = ρ (3) A A and ρ (2) = ρ (4) with A A 1 ρ (1) = ψ ψ +κ ψ ψ +κ ψ ψ + ψ ψ , (10) A 2(1+κ2){| 1ih 1| | 1ih 2| | 2ih 1| | 2ih 2} 1 ρ (2) = ψ ψ κ ψ ψ κ ψ ψ + ψ ψ . (11) A 2(1 κ2){| 1iAh 1|− | 1iAh 2|− | 2iAh 1| | 2iAh 2|} − The eigenvalues of the above density operators ρ (1)(or ρ (3)) are given in terms of the Gram A A matrix elements G as follows, ij 1+G (1+κ)2 1 G (1 κ)2 13 13 λ = = , λ = − = − , (12) 1/1 2 2(1+κ2) 2/1 2 2(1+κ2) and for ρ (2)(or ρ (4)) we have A A 1+G 1 1 G 1 24 24 λ = = , λ = − = . (13) 1/2 2/2 2 2 2 2 Hence, the entropy of entanglement is 1+D 1+D 1 D 1 D E( Ψ ) = E( Ψ ) = log − log − , (14) 1 AB 3 AB | i | i − 2 2 − 2 2 and E( Ψ ) = E( Ψ ) = 1, (15) 2 AB 4 AB | i | i 4 because G = D, and G = 0. Thus Ψ and Ψ are maximally entangled, even 13 24 2 AB 4 AB | i | i thoughtheentangled states consist ofnonorthogonalstates ineach subsystem. These results aretrueforarbitrarynonorthogonalstates with ψ ψ = ψ ψ =κanddo notdepend on 1 2 2 1 h | i h | i the physical dimension of the systems. This property may be unexpected but can be under- stood easily by noting that the states Ψ and Ψ are equivalent to ( + + )/√2 2 4 | i | i | i|−i±|−i| i in terms of the orthogonal basis = ( ψ ψ )/ N , (16) 1 2 ± |±i | i±| i q with N = 2 2κ. ± ± B. Mixtures of quasi Bell states We can construct a quasi-Werner mixed state based on quasi-Bell states by 1 F W = F Ψ Ψ + − Ψ Ψ + Ψ Ψ + Ψ Ψ , (17) 2 AB 2 1 AB 1 3 AB 3 4 AB 4 | i h | 3 {| i h | | i h | | i h |} where 0 F 1. If Ψ , Ψ , Ψ , Ψ are Bell states, then the above equation 1 AB 2 AB 3 AB 4 AB ≤ ≤ | i | i | i | i gives a standard Werner state [7]. It is known that the fully entangled fractionof the Werner state is F, and the entanglement of formation of the Werner state is given by 1 E(W) = H + F[1 F] . (18) (cid:18)2 q − (cid:19) The fully entangled fraction of the quasi Werner state is analogously given by f(W) = Ψ W Ψ = F, (19) AB 2 2 AB h | | i becausethequasiBellstatesareorthogonaltoeachother,except forthepairofstates Ψ 1 AB | i and Ψ , as one can see from the Gram matrix G. However, the quasi Werner state and 3 AB | i Werner state are completely different states. In particular, the eigenvalues of quasi Werner states are different from those of Werner states. The eigenvalues of the density operator are given by those of the modified Gram matrix [10]. For the quasi Werner state, the Gram matrix is 5 1(1 F) 0 1(1 F)D 0  3 − 3 −   0 F 0 0    G =   (20) W    1(1 F)D 0 1(1 F) 0   3 − 3 −       0 0 0 1(1 F)   3 −    As a result, we have F, 1(1 F), 1(1+D)(1 F), 1(1 D)(1 F) as the eigenvalues of 3 − 3 − 3 − − the quasi Werner state. Thus the lower bound of the entropy of formation of quasi Werner state is the same as that of Werner state. For more general mixtures of quasi Bell states we will need a more advanced analysis which is reported on in a subsequent paper. III. QUASI-BELL STATES BASED ON BOSONIC COHERENT STATES Let us consider two coherent states of a bosonic mode α , α , e.g., let α be {| i | − i} ± the coherent amplitude of a light field. Using previous notation, we have κ = α α = h | − i exp 2 α 2 . Then one can construct the quasi Bell states as follows: {− | | } Ψ = h ( α α + α α ) | 1iAB 1 | iA|− iB |− iA| iB |Ψ2iAB = h2(|αiA|−αiB −|−αiA|αiB) (21) Ψ = h ( α α + α α ) 3 AB 3 A B A B | i | i | i |− i |− i |Ψ4iAB = h4(|αiA|αiB −|−αiA|−αiB) Since the coherent states are nonorthogonal, we can apply the results of Section 2 to these states. States of similar form were discussed by Sanders [8], and Wielinga [9], who called these states entangled coherent states. From the results in the Section 2, we know that Ψ and Ψ have one ebit of entanglement independent of α. This is an interesting 2 AB 4 AB | i | i and potentially useful property (see next Section). On the other hand, Ψ and Ψ 1 AB 3 AB | i | i are maximally entangled states only in the limit α . In order to avoid confusion with → ∞ continuous variable states [14], we should mention here that, of course, the dimension of the space spanned by the quasi Bell states is 4 even though they are embedded in a vector space 6 of infinite dimension. This implies that the maximum value of the von Neumann entropy for thequasiBellstatesisunity. Differentamplitudes αjust givethedegreeofnonorthogonality. It also means that one cannot use coherent entangled states for teleportation of continuous quantum variables [14,15]. In a separate paper [16], we will show how to use entangled coherent states for teleportation of Schr¨odinger cat states. The average photon numbers of the reduced states of the quasi Bell states read (1 κ2) (1+κ2) n(1) = − α 2, n(2) = α 2. (22) h A i (1+κ2)| | h A i (1 κ2)| | − Thus the quasi Bell states can have arbitrary photon numbers. As said above, however, the quasi Bell states are not truly continuous variable states and in particular do not belong to the class of Gaussian states [11], [12] in contrast to, e.g., the two mode squeezed state [13]. This is shown in the following way. The characteristic functions of the quasi Bell states are given by C(ξ,η) = Tr[ Ψ Ψ exp(ξa†)exp( ξ∗a )exp(ηa† )exp( η∗a )] | iABh | A − A B − B exp ( ξ 2+ η 2)/2 (23) × {− | | | | } where a and a† are the annihilation and creation operators, respectively. They can be calculated, with the result C(ξ,η i = 1,2) = h 2exp ( ξ 2 + η 2)/2 exp(A B )α i 1 1 | {− | | | | }{ − + exp( A +B )α exp(A B )α 1 1 2 2 − ± − exp( A +B )α (24) 2 2 ± − } C(ξ,η i = 3,4) = h 2exp ( ξ 2 + η 2)/2 exp(A +B )α i 1 1 | {− | | | | }{ + exp( A B )α exp(A +B )α 1 1 2 2 − − ± exp( A B )α (25) 2 2 ± − − } where A = (ξ ξ∗),A = (ξ + ξ∗),B = (η η∗),B = (η + η∗). The characteristic 1 2 1 2 − − functions are indeed not Gaussian. Finally let us explore one more property of a similar set 7 of entangled states. If the amplitudes of the modes A and mode B in Ψ (or Ψ ) 2 AB 4 AB | i | i are chosen to be different, say, α and β, respectively, the eigenvalues of the reduced density operator are (1+κ )(1 κ ) (1 κ )(1+κ ) A B A B λ = − , λ = − , (26) 1 2 2(1 κ κ ) 2(1 κ κ ) A B A B − − where κ = exp( 2 α 2) and κ = exp( 2 β 2). We can then easily see that the entropy of A B − | | − | | entanglement attains its maximum value of 1 only when the amplitudes of both modes are the same. IV. DECOHERENCE PROPERTIES In this section, we will discuss decoherence properties of the state Ψ . We are 2 AB | i concerned with the decoherence due to energy loss or photon absorption. In particular, we would like to demonstrate that entangled coherent states possess a certain degree of robustness against decoherence when compared to a standard bi-photon Bell state. We assume that Alice produces a coherent entangled state Ψ , keeps one part (A) and 2 AB | i transmits the other part B to Bob through a lossy channel. Bob will receive an attenuated optical state. Thus, Alice prepares Ψ = h ( α α α α ) Ψ (α) (27) 2 AB 2 A B A B 2 AB | i | i |− i −|− i | i ≡ | i where h = 1/ 2(1 κ 2), κ = α α . When we employ a half mirror model for the 2 A A q − h |− i noisy channel, the effect of energy losses is described by a linear coupling with an external vacuum field as follows: U α 0 = √ηα (1 η)α . (28) BE B E B E | i | i | i |q − i where the mode E is an external mode responsible for the energy loss, η is the noise parameter, and α is taken as real. If we use Ψ (α) as the initial state, the final state 2 AB | i entangled with the environment is 8 Iˆ U Ψ (α) 0 A BE 2 AB E ⊗ | i ⊗| i = h ( α √ηα (1 η)α α √ηα (1 η)α ). (29) 2 A B E A B E | i |− i |−q − i −|− i | i |q − i The normalized density operator shared by Alice and Bob is given by a super operator calculation [17], ρ = h2 α α √ηα √ηα + α α √ηα √ηα AB 2{| ih |⊗|− ih− | |− ih− |⊗| ih | α α L √ηα √ηα α α L √ηα √ηα (30) − |− ih |⊗ | ih− |−| ih− |⊗ |− ih |} where L = exp 2(1 η) α 2 . Let us discuss the entanglement of the above density {− − | | } operator. As discussed in Section 2, we can use the entangled fraction to measure the entanglement of this mixed state. The fully entangled fraction of ρ is given by AB f(ρ ) = max Ψ (β) ρ Ψ (β) (31) AB AB 2 AB 2 AB β h | | i because Ψ (β) is indeed maximally entangled. The above is given by 2 AB | i (1+L)(κ 2(β)κ 2(β)+κ 2(β)κ 2(β) 2κ (β)κ (β)κ (β)κ (β)) 1 2 3 4 1 2 3 4 f(ρ ) = max − (32) AB β 2(1 κA2)(1 κ02(β)) − − where κ (β) = exp 2 β 2 , κ (β) = exp α β 2/2 , κ (β) = exp β √ηα 2/2 , 0 1 2 {− | | } {−| − | } {−| − | } κ (β) = exp α+ β 2/2 , and κ (β) = exp β + √ηα 2/2 . The maximum is attained 3 4 {−| | } {−| | } for the value α+√ηα β = (33) 2 that is, exactly halfway between the original coherent amplitude and the attenuated ampli- tude. InFigure1 we plotf(ρ )asa functionof α forvariousvalues ofthenoise parameter AB | | η. 9 FIGURES 1 0.9 0.8 0.7 )B A ρ f(0.6 0.5 0.4 0 1 2 3 4 5 |α| FIG. 1. Overlap of ρ with appropriately chosen fully entangled state (see text) as a function AB of α for noise parameters η = 0.9,0.7,0.5,0.3,0.1 for top to bottom curves. | | For comparison, we consider the biphoton Bell state: 1 Ψ = ( ), (34) | 2(p)iAB √2 | liA| ↔iB −| ↔iA| liB where , denotes single photon polarization directions. After passing through the {| li | ↔i} same lossy channel, Alice and Bob share the state 1 ρpol = η Ψ Ψ +(1 η) I 0 0 (35) AB | 2(p)iABh 2(p)| − 2 A ⊗| iBh | with I the identity on mode A. As a result, the fully entangled fraction is A f(ρpol) = η (36) A in that case, which is clearly less than for entangled coherent states with sufficiently small amplitudes (see Figure 1). Finally, we note that in [16] we give the analogous decoherence properties for a symmetric noise channel, describing the case where both Alice’s and Bob’s mode suffer losses. V. CONCLUDING REMARKS In this paper, our effort was devoted to clarify several properties of entangled nonorthog- onal states. We constructed 4 entangled states that generalize the standard Bell states. 10

See more

The list of books you might like

Most books are stored in the elastic cloud where traffic is expensive. For this reason, we have a limit on daily download.