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

Sending classical information through relativistic quantum channels

André G. S. Landulfo andre.landulfo@ufabc.edu.br Centro de Ciências Naturais e Humanas, Universidade Federal do ABC, Rua Santa Adélia 166, 09210-170, Santo André, São Paulo, Brazil    Adriano C. Torres actorres@ift.unesp.br Instituto de Física Teórica, Universidade Estadual Paulista, Rua Dr. Bento Teobaldo Ferraz, 271 - Bl. II, 01140-070, São Paulo, SP, Brazil
(August 12, 2025)
Abstract

We investigate how special relativity influences the transmission of classical information through quantum channels by evaluating the Holevo bound when the sender and the receiver are in (relativistic) relative motion. By using the spin degrees of freedom of spin-1/21/2 fermions to encode the classical information we show that, for some configurations, the accessible information in the receiver can be increased when the spin detector moves fast enough. This is possible by allowing the momentum wave packet of one of the particles to be wide enough while the momentum wave packets of other particles are kept relatively narrow. In this way, one can take advantage of the fact that boosts entangle the spin and momentum degrees of freedom of spin-1/21/2 fermions to increase the accessible information in the former. We close the paper with a discussion of how this relativistic quantum channel cannot in general be described by completely positive quantum maps.

pacs:
03.67.Hk, 03.65.Ta, 03.65.Ud, 03.30.+p

I Introduction

Remarkable effects are found in information processing when communication channels are allowed to be quantum mechanical. Examples of these are fast quantum algorithms fastquantalg , quantum teleportation tele1 , quantum cryptography cripto1 ; cripto2 ; cripto3 , dense coding densecoding1 and quantum error correction errorcorrection . Quantum information theory nielsen&chuang commonly deals only with nonrelativistic systems. However, a relativistic treatment is relevant not only to the logical completeness of the theory but also to the disclosure of new physical effects and bounds that arise in information transfer and processing when there is relative motion between the parts that trade information Peres&Terno2003 ; Peres&Terno2004 . Moreover, a better understanding of the relativistic extension of quantum information theory may shed light on several important conceptual issues, for instance, the black hole information “paradox” Hawking1976 ; Wald2001 . For the aforementioned reasons, a great deal of attention has been paid to quantum information theory in the context of special relativity PST02 ; GA02 , the Unruh effect Fetal ; FKMB11 ; LM09b ; CLMS10 , and black holes PJ08 ; HBK12 ; ML10 . Recently, an experimental setup in which a relativistic formulation of quantum information theory may be important was proposed. It consists in using free-space transmission of photons between ground stations and satellites in order to test quantum mechanics for large space distances and, eventually, to implement quantum information protocols in global scales satellite1 ; satellite2 ; satellite3 ; satellite4 . In previous works, we have studied how special relativity affects the correlations between an entangled pair of both fermions LM2009 and photons LMT2010 by examining the Clauser-Horne-Shimony-Holt Bell inequality CHSH1969 when the detectors are moving. Here, we adopt a rather information-theoretic approach and analyze the classical capacity of a relativistic quantum channel in comparison to that of a nonrelativistic one.

One of the keystones of quantum information is the indistinguishability of arbitrary quantum states: given two non-orthogonal quantum states, one cannot distinguish between them with full reliability by making any measurement, a result that is easily shown to be equivalent to the no-cloning theorem nocloning1 ; nocloning2 ; nielsen&chuang . More precisely, let X=1,,nX=1,...,n be an index that indicates each of the states from the set {ρ1,,ρn}\left\{\rho_{1},\ldots,\rho_{n}\right\}, and suppose that these states are prepared according to a probability distribution p1,,pnp_{1},\ldots,p_{n}. An experimentalist performs a measurement described by the positive operator-valued measure (POVM) {E1,,Em}\left\{E_{1},\ldots,E_{m}\right\} and is supposed to infer the state XX that was prepared from the measurement outcome Y=1,,mY=1,\ldots,m. A good measure of how much information the experimentalist can gain about the state through this procedure is given by the mutual information I(X:Y)I\left(X:Y\right), which can be defined as

I(X:Y)=H(X)+H(Y)H(X,Y).I\left(X:Y\right)=H(X)+H(Y)-H\left(X,Y\right). (1)

Here, H(X)H(X) and H(Y)H(Y) are the Shannon entropies associated with the probability distribution p1,,pnp_{1},\ldots,p_{n} of the preparation procedure and the probability distribution {tr(E1ρ),,tr(Enρ)}\{{\rm tr}(E_{1}\rho),...,{\rm tr}(E_{n}\rho)\} of the measurement outcomes, where ρx=1npxρx\rho\equiv\sum_{x=1}^{n}p_{x}\rho_{x}, respectively, and H(X,Y)H\left(X,Y\right) is their joint entropy. It is well known that I(X:Y)H(X)I\left(X:Y\right)\leq H(X) and that one can infer XX from YY if and only if I(X:Y)=H(X)I\left(X:Y\right)=H(X). The closer I(X:Y)I\left(X:Y\right) gets to H(X)H(X), the more accurately it is possible to infer XX from YY. Of course the mutual information depends on what measurement the experimentalist decides to perform, that is, on the POVM chosen. To avoid this indeterminacy, we define the accessible information to be the maximum of the mutual information over all possible measurement schemes. Although no general method for calculating the accessible information is known, it is possible to prove a very important upper bound known as the Holevo bound Holevo1973 . It states that, for any measurement the experimentalist may do, the inequality below holds:

I(X:Y)χ(ρ)S(x=1npxρx)x=1npxS(ρx),I\left(X:Y\right)\leq\chi(\rho)\equiv S\left(\sum_{x=1}^{n}p_{x}\rho_{x}\right)-\sum_{x=1}^{n}p_{x}S\left(\rho_{x}\right), (2)

where S(ω)=tr(ωlog2ω)S\left(\omega\right)=-\textup{tr}\left(\omega\,\log_{2}\omega\right) is the von Neumann entropy of the quantum state ω\omega. It is easy to show nielsen&chuang that χ(ρ)H(X)\chi(\rho)\leq H(X) and, therefore, inequality (2) implies that one qubit contains at most one bit of information. The Holevo bound is especially relevant due to the Holevo-Schumacher-Westmoreland (HSW) theorem, independently proved by Schumacher and Westmoreland Schumacher&Westmorland1997 and by Holevo Holevo1998 , according to which the rate χ(ρ)\chi(\rho) is asymptotically achievable and thus can be used to obtain an expression for the classical (product state) capacity of a quantum channel. Hence, in order to study the transmission of classical information through relativistic quantum channels, it is interesting to analyze the Holevo bound in a typical relativistic quantum communication setup. Here this is done by setting the detector in relative motion with respect to the preparation apparatus.

The indistinguishability of non-orthogonal quantum states mentioned above together with fact that any attempt to distinguish them ends up only imparting a disturbance to the states is what motivates their use in cryptographic protocols. This however is not the only setup in which non-orthogonal states play a relevant role. As it was first shown in fucs97 there are some noisy quantum channels in which the channel capacity is only achieved by the use of non-orthogonal states. In the present paper we will show that when the parts that trade information are in relative motion (and there is not any external noise afflicting the states), there is a certain class of non-orthogonal states for which the Holevo bound increases when compared to its value when the sender and the receiver are at rest relative to each other. This suggests that the relative motion may actually help to increase the capacity of some noisy quantum channels.

The paper is organized as follows. In Sec. II we analyze the transmission of two and four classical bits through nonrelativistic quantum communication channels. In Sec. III we study how relativity influences the previous quantum communication process. This is done by setting the receiver, Bob, in relative motion with respect to the sender, Alice. In this context we analyze how to optimize the accessible information on the receiver. In Sec. IV we define quantum maps that describe the quantum channels analyzed in the previous sections and show that there are cases in which they fail to be completely positive. Sec. V is dedicated to our final remarks. We adopt natural units c==1c=\hbar=1 unless stated otherwise.

II Nonrelativistic quantum communication setup

II.1 Two classical bits

Let us assume that Alice has a classical information source that produces symbols X=0,1X=0,1 according to the probability distribution p0=λ,p1=1λp_{0}=\lambda,p_{1}=1-\lambda, 0λ10\leq\lambda\leq 1. Depending on the value she obtains for XX, she prepares a pure quantum state ψX\psi_{X}, chosen from a fixed set {ψ0,ψ1}\left\{\psi_{0},\psi_{1}\right\}, of a spin-1/21/2 particle with mass mm and then sends it to Bob. He then makes a spin measurement of his choice on that state and has to identify XX based on the outcome YY. We assume that (see, e.g., Ref. Bogolubov1975 for the two-spinor notation used below)

ψ0(𝐩)\displaystyle\psi_{0}({\bf p}) =\displaystyle= (f𝐤0w0(𝐩)0),\displaystyle\left(\begin{array}[]{c}f_{\mathbf{k}_{0}}^{w_{0}}\left(\mathbf{p}\right)\\ 0\end{array}\right), (5)
ψ1(𝐩)\displaystyle\psi_{1}({\bf p}) =\displaystyle= cosθ(f𝐤1w1(𝐩)0)+sinθ(0f𝐤1w1(𝐩)),\displaystyle\cos\theta\left(\begin{array}[]{c}f_{\mathbf{k}_{1}}^{w_{1}}(\mathbf{p})\\ 0\\ \end{array}\right)+\sin\theta\left(\begin{array}[]{c}0\\ f_{\mathbf{k}_{1}}^{w_{1}}(\mathbf{p})\\ \end{array}\right), (10)

where

f𝐤iwi(𝐩)=π34wi32exp[(𝐩𝐤i)2/2wi2].f_{\mathbf{k}_{i}}^{w_{i}}\left(\mathbf{p}\right)=\pi^{-\frac{3}{4}}w_{i}^{-\frac{3}{2}}\,\exp\left[-\left(\mathbf{p}-\mathbf{k}_{i}\right)^{2}/2w_{i}^{2}\right]. (11)

Here, the parameters wi+w_{i}\in\mathbb{R}_{+} and 𝐤i=(ki,0,0)\mathbf{k}_{i}=\left(k_{i},0,0\right), i=0,1i=0,1, give the particle’s momentum dispersion and the average momentum, respectively. Note that ψ1\psi_{1} can be written as

ψ1(𝐩)=f𝐤1w1(𝐩)(cosθsinθ),\psi_{1}({\bf p})=f_{\mathbf{k}_{1}}^{w_{1}}(\mathbf{p})\left(\begin{array}[]{c}\cos\theta\\ \sin\theta\\ \end{array}\right),

making manifest its direct product structure.

First we summarize what happens when Bob is at rest relative to Alice, which is the typical quantum information scenario. All spin measurement results can be predicted through the reduced spin density operator, obtained by tracing out the momenta,

τd𝐩ρ(𝐩,𝐩),\tau\equiv\int\mbox{d}{\bf p}\;\rho\left(\mathbf{p},\mathbf{p}\right), (12)

where

ρ(𝐩,𝐩~)=λψ0(𝐩)ψ0(𝐩~)+(1λ)ψ1(𝐩)ψ1(𝐩~)\rho\left(\mathbf{p},\mathbf{\tilde{p}}\right)=\lambda\,\psi_{0}(\mathbf{p})\psi_{0}(\mathbf{\tilde{p}})^{\dagger}+(1-\lambda)\,\psi_{1}(\mathbf{p})\psi_{1}(\mathbf{\tilde{p}})^{\dagger} (13)

is the complete density operator of the system. We thus obtain

τ=λ(1000)+(1λ)(cos2θcosθsinθcosθsinθsin2θ).\tau=\lambda\left(\begin{array}[]{cc}1&0\\ 0&0\\ \end{array}\right)+(1-\lambda)\left(\begin{array}[]{cc}\cos^{2}\theta&\cos\theta\sin\theta\\ \cos\theta\sin\theta&\sin^{2}\theta\\ \end{array}\right). (14)

Note that, due to the linearity of the trace, we can write τλτ0+(1λ)τ1\tau\equiv\lambda\,\tau_{0}+(1-\lambda)\,\tau_{1}, where

τ0d𝐩ψ0(𝐩)ψ0(𝐩)=(1000),\tau_{0}\equiv\int\mbox{d}{\bf p}\;\psi_{0}(\mathbf{p})\psi_{0}(\mathbf{p})^{\dagger}=\left(\begin{array}[]{cc}1&0\\ 0&0\\ \end{array}\right), (15)

and

τ1d𝐩ψ1(𝐩)ψ1(𝐩)=(cos2θcosθsinθcosθsinθsin2θ).\tau_{1}\equiv\int\mbox{d}{\bf p}\;\psi_{1}(\mathbf{p})\psi_{1}(\mathbf{p})^{\dagger}=\left(\begin{array}[]{cc}\cos^{2}\theta&\cos\theta\sin\theta\\ \cos\theta\sin\theta&\sin^{2}\theta\\ \end{array}\right). (16)

As can be easily seen from Eq. (14), τ\tau is just a statistical mixture of the pure spin states ϕ\phi_{\uparrow} and

ϕθcosθϕ+sinθϕ,\phi_{\theta}\equiv\cos\theta\,\phi_{\uparrow}+\sin\theta\,\phi_{\downarrow}, (17)

where ϕ(10)\phi_{\uparrow}\equiv\left(\begin{array}[]{c}1\\ 0\\ \end{array}\right) and ϕ(01)\phi_{\downarrow}\equiv\left(\begin{array}[]{c}0\\ 1\\ \end{array}\right) are the eigenvectors of SzS_{\rm z} with eigenvalues 1/21/2 and 1/2-1/2, respectively. Note that ϕθ\phi_{\theta} is an eigenstate of 𝐒𝐧{\bf S}\cdot{\bf n} where 𝐧=(sin2θ,0,cos2θ){\bf n}=(\sin 2\theta,0,\cos 2\theta), 𝐒𝝈/2{\bf S}\equiv\boldsymbol{\sigma}/2, and 𝝈=(σx,σy,σz)\boldsymbol{\sigma}=\left(\sigma_{\rm x},\sigma_{\rm y},\sigma_{\rm z}\right) is the Pauli vector. Thus, we can say that the momentum degrees of freedom play absolutely no role in this case. As τ1\tau_{1} and τ2\tau_{2} correspond to pure states, we have S(τ1)=S(τ2)=0S\left(\tau_{1}\right)=S\left(\tau_{2}\right)=0, so that

χ(τ)=S(τ)=l=±βllog2βl,\chi\left(\tau\right)=S\left(\tau\right)=-\sum_{l=\pm}\beta_{l}\,\log_{2}\beta_{l}, (18)

where β±\beta_{\pm} are the eigenvalues of τ\tau, namely,

β±=12±121+4sin2θ(λ2λ).\beta_{\pm}=\frac{1}{2}\pm\frac{1}{2}\sqrt{1+4\sin^{2}\theta\left(\lambda^{2}-\lambda\right)}. (19)

Note that χ(τ)\chi(\tau) has periodicity π\pi with respect to θ\theta. It is straightforward to verify that the Holevo bound reaches its maximum value when θ=π/2\theta=\pi/2, corresponding to orthogonal spin states (see Fig. 1). At this point, and only at this point, it is possible for Bob to determine with certainty which state Alice has prepared, which he does simply by measuring SzS_{z}. This is actually a general property of orthogonal states, which can always be completely distinguished by appropriate measurements. By contrast, the minimum is attained at θ=0\theta=0 and θ=π\theta=\pi, which means that the states ψ0\psi_{0} and ψ1\psi_{1} are identical (up to a phase) when it comes solely to spin and, therefore, they cannot be distinguished at all.

Refer to caption
Figure 1: (Color online) The graph shows the behavior of the Holevo bound when Bob is at rest with respect to Alice as a function of the angle θ\theta, which characterize the spin part of the second state, for both the transmission of two (full line) and four (dashed line) classical bits.

II.2 Four classical bits

Now, let us see how the above results generalize to the case where more classical bits are sent through the quantum communication channel. For this purpose, let us suppose now that Alice has a classical information source that produces symbols X~=00,01,10,11\tilde{X}=00,01,10,11 according to the probability distribution p00=λ1,p01=λ2,p10=λ3p_{00}=\lambda_{1},p_{01}=\lambda_{2},p_{10}=\lambda_{3}, p11=λ4=1λ1λ2λ3p_{11}=\lambda_{4}=1-\lambda_{1}-\lambda_{2}-\lambda_{3}, 0λl~1,0\leq\lambda_{\tilde{l}}\leq 1, l~=1,2,3,4\tilde{l}=1,2,3,4. As in the previous case, Alice can choose pure quantum states from the set {ψ0,ψ1}\left\{\psi_{0},\psi_{1}\right\}, where ψ0\psi_{0} and ψ1\psi_{1} are given by Eqs. (5) and (10), respectively. Therefore, depending on the value of X~\tilde{X}, Alice prepares one of the four product states

ψ00(𝐩,𝐩~)\displaystyle\psi_{00}({\bf p},\tilde{{\bf p}})\!\! \displaystyle\equiv ψ0(𝐩)ψ0(𝐩~),ψ01(𝐩,𝐩~)ψ0(𝐩)ψ1(𝐩~),\displaystyle\!\!\psi_{0}({\bf p})\otimes\psi_{0}(\tilde{{\bf p}}),\;\psi_{01}({\bf p},\tilde{{\bf p}})\equiv\psi_{0}({\bf p})\otimes\psi_{1}(\tilde{{\bf p}}),
ψ10(𝐩,𝐩~)\displaystyle\psi_{10}({\bf p},\tilde{{\bf p}})\!\! \displaystyle\equiv ψ1(𝐩)ψ0(𝐩~),ψ11(𝐩,𝐩~)ψ1(𝐩)ψ1(𝐩~),\displaystyle\!\!\psi_{1}({\bf p})\otimes\psi_{0}(\tilde{{\bf p}}),\;\psi_{11}({\bf p},\tilde{{\bf p}})\equiv\psi_{1}({\bf p})\otimes\psi_{1}(\tilde{{\bf p}}),

and sends it to Bob. He then makes a spin measurement of his choice on that state and has to identify X~\tilde{X} based on the outcome Y~\tilde{Y}. The total density operator of the system is given by

ρ~(𝐩,𝐩~,𝐩,𝐩~)=λ1ψ00(𝐩,𝐩~)ψ00(𝐩,𝐩~)\displaystyle\tilde{\rho}\left(\mathbf{p},\mathbf{\tilde{p}},\mathbf{p}^{\prime},\mathbf{\tilde{p}}^{\prime}\right)=\lambda_{1}\psi_{00}({\bf p},\tilde{{\bf p}})\psi_{00}^{\dagger}({\bf p}^{\prime},\tilde{{\bf p}}^{\prime})
+λ2ψ01(𝐩,𝐩~)ψ01(𝐩,𝐩~)+λ3ψ10(𝐩,𝐩~)ψ10(𝐩,𝐩~)\displaystyle+\lambda_{2}\,\psi_{01}({\bf p},\tilde{{\bf p}})\psi_{01}^{\dagger}({\bf p}^{\prime},\tilde{{\bf p}}^{\prime})+\lambda_{3}\,\psi_{10}({\bf p},\tilde{{\bf p}})\psi_{10}^{\dagger}({\bf p}^{\prime},\tilde{{\bf p}}^{\prime})
+λ4ψ11(𝐩,𝐩~)ψ11(𝐩,𝐩~).\displaystyle+\lambda_{4}\,\psi_{11}({\bf p},\tilde{{\bf p}})\psi_{11}^{\dagger}({\bf p}^{\prime},\tilde{{\bf p}}^{\prime}). (21)

If the momentum degrees of freedom are traced out in Eq. (21), we obtain the reduced spin density operator

τ~d𝐩d𝐩~ρ~(𝐩,𝐩~,𝐩,𝐩~),\displaystyle\tilde{\tau}\equiv\int\mbox{d}{\bf p}\mbox{d}\tilde{{\bf p}}\;\tilde{\rho}\left(\mathbf{p},\tilde{\mathbf{p}},\mathbf{p},\tilde{\mathbf{p}}\right), (22)

which can be written as

τ~=λ1τ0τ0+λ2τ0τ1+λ3τ1τ0+λ4τ1τ1,\tilde{\tau}=\lambda_{1}\tau_{0}\otimes\tau_{0}+\lambda_{2}\tau_{0}\otimes\tau_{1}\\ +\lambda_{3}\tau_{1}\otimes\tau_{0}+\lambda_{4}\tau_{1}\otimes\tau_{1},

and from which all spin measurement results can be predicted. The density operators τ0\tau_{0} and τ1\tau_{1} are given in Eqs. (15) and (16), respectively. Now, by using that S(ω1ω2)=S(ω1)+S(ω2)S(\omega_{1}\otimes\omega_{2})=S(\omega_{1})+S(\omega_{2}) for any density matrices ω1\omega_{1} and ω2\omega_{2} and that S(τi)=0,S(\tau_{i})=0, we can write χ(τ~)\chi\left(\tilde{\tau}\right) as

χ(τ~)=S(τ~)=l~=14β~l~log2β~l~,\chi\left(\tilde{\tau}\right)=S\left(\tilde{\tau}\right)=-\sum_{\tilde{l}=1}^{4}\tilde{\beta}_{\tilde{l}}\,\log_{2}\tilde{\beta}_{\tilde{l}}, (23)

where β~l~\tilde{\beta}_{\tilde{l}} are the eigenvalues of τ~.\tilde{\tau}. In Fig 1, χ(τ~)\chi\left(\tilde{\tau}\right) is plotted when λ1=λ2=λ3=1/4\lambda_{1}=\lambda_{2}=\lambda_{3}=1/4. We can see that, as in the case of two bits, its maximum value is attained when θ=π/2.\theta=\pi/2.

III Relativistic Quantum Communication Optimization

Now we turn our attention to the relativistic case, in which Bob moves with three-velocity 𝐯=(v,0,0)\mathbf{v}=\left(v,0,0\right) relative to Alice.

III.1 Two classical bits

Let us suppose first that Alice wants to transmit two bits of classical information. Thus, as explained in Section II.1, she prepares the state ρ\rho given in Eq. (13) and sends it to Bob. Due to his motion, Bob sees the state ρ\rho prepared by Alice unitarily transformed in his proper frame as

ρ(𝐩,𝐩~)=λψ0(𝐩)ψ0(𝐩~)+(1λ)ψ1(𝐩)ψ1(𝐩~)\rho^{\prime}\left(\mathbf{p},\mathbf{\tilde{p}}\right)=\lambda\,\psi_{0}^{\prime}(\mathbf{p})\psi_{0}^{\prime}(\mathbf{\tilde{p}})^{\dagger}+(1-\lambda)\,\psi_{1}^{\prime}(\mathbf{p})\psi_{1}^{\prime}(\mathbf{\tilde{p}})^{\dagger} (24)

where Halpern1968 ; Weinberg1996

ψi(𝐩)(U(Λ)ψi)(𝐩)\psi_{i}^{\prime}\left(\mathbf{p}\right)\equiv\left(U(\Lambda)\psi_{i}\right)({\bf p}) (25)

with

(U(Λ)ψi)(𝐩)(Λ1p)0p0D(Λ,Λ1p)ψi(Λ1𝐩),\displaystyle\left(U(\Lambda)\psi_{i}\right)({\bf p})\equiv\sqrt{\frac{\left(\Lambda^{-1}p\right)^{0}}{p^{0}}}\,D\left(\Lambda,\Lambda^{-1}p\right)\psi_{i}\left(\Lambda^{-1}\mathbf{p}\right)\!\!, (26)

p=(𝐩2+m2,𝐩)p=\left(\sqrt{\mathbf{p}^{2}+m^{2}},\mathbf{p}\right), and Λ1𝐩\Lambda^{-1}\mathbf{p} denoting the spatial part of the four-vector Λ1p\Lambda^{-1}p. The Wigner rotation is given by

D(Λ,q)\displaystyle D\left(\Lambda,q\right) =\displaystyle= cosh(α/2)(q0+m)σ0[(p0+m)(q0+m)]1/2\displaystyle\frac{\cosh(\alpha/2)\,(q^{0}+m)\sigma^{0}}{[(p^{0}+m)(q^{0}+m)]^{1/2}} (27)
+\displaystyle+ sinh(α/2)[𝐪𝐞σ0+i(𝐞×𝐪)𝝈][(p0+m)(q0+m)]1/2,\displaystyle\frac{\sinh(\alpha/2)[{\bf q}\cdot{\bf e}\;\sigma^{0}+i({\bf e}\times{\bf q})\cdot\boldsymbol{\sigma}]}{[(p^{0}+m)(q^{0}+m)]^{1/2}},

where α=tanh1v\alpha=-\tanh^{-1}v, qΛ1p,q\equiv\Lambda^{-1}p, σ0=I\sigma^{0}=I is the identity matrix, and 𝐞\mathbf{e} gives the direction of the boost, which in our case is 𝐞x{\bf e}_{x}, so that

Λ=(coshαsinhα00sinhαcoshα0000100001).\Lambda=\left(\begin{array}[]{cccc}\cosh\alpha&\sinh\alpha&0&0\\ \sinh\alpha&\cosh\alpha&0&0\\ 0&0&1&0\\ 0&0&0&1\end{array}\right). (28)

By using Eq. (27) in Eq. (26) with ψi\psi_{i} given in Eqs. (5) and (10), we obtain

ψ0(𝐩)=(a𝐤0w0(𝐩)b𝐤0w0(𝐩))\psi_{0}^{\prime}(\mathbf{p})=\left(\begin{array}[]{c}a_{\mathbf{k}_{0}}^{w_{0}}(\mathbf{p})\\ b_{\mathbf{k}_{0}}^{w_{0}}(\mathbf{p})\end{array}\right) (29)

and

ψ1(𝐩)=cosθ(a𝐤1w1(𝐩)b𝐤1w1(𝐩))+sinθ(b𝐤1w1(𝐩)a𝐤1w1(𝐩)),\psi_{1}^{\prime}(\mathbf{p})=\cos\theta\left(\begin{array}[]{c}a_{\mathbf{k}_{1}}^{w_{1}}(\mathbf{p})\\ b_{\mathbf{k}_{1}}^{w_{1}}(\mathbf{p})\end{array}\right)+\sin\theta\left(\begin{array}[]{c}-b_{\mathbf{k}_{1}}^{w_{1}}(\mathbf{p})\\ a_{\mathbf{k}_{1}}^{w_{1}}(\mathbf{p})^{*}\end{array}\right), (30)

where the momentum wave packets are given by

a𝐤iwi(𝐩)\displaystyle a_{\mathbf{k}_{i}}^{w_{i}}(\mathbf{p}) =\displaystyle= Kf𝐤iwi(𝐪)[C(q0+m)+S(qx+iqy)],\displaystyle K\,f_{\mathbf{k}_{i}}^{w_{i}}(\mathbf{q})\left[C(q^{0}+m)+S(q_{x}+iq_{y})\right],
b𝐤iwi(𝐩)\displaystyle b_{\mathbf{k}_{i}}^{w_{i}}(\mathbf{p}) =\displaystyle= Kf𝐤iwi(𝐪)Sqz,\displaystyle K\,f_{\mathbf{k}_{i}}^{w_{i}}(\mathbf{q})\,Sq_{z},

with

K\displaystyle K \displaystyle\equiv (q0/p0)1/2/[(q0+m)(p0+m)]1/2,\displaystyle(q^{0}/p^{0})^{1/2}/[(q^{0}+m)(p^{0}+m)]^{1/2},
C\displaystyle C \displaystyle\equiv cosh(α/2),\displaystyle\cosh\left(\alpha/2\right),
S\displaystyle S \displaystyle\equiv sinh(α/2).\displaystyle\sinh\left(\alpha/2\right).

By tracing out the momentum degrees of freedom we obtain the reduced spin operator in Bob’s frame

τ=λτ0+(1λ)τ1,\tau^{\prime}=\lambda\,\tau_{0}^{\prime}+(1-\lambda)\tau_{1}^{\prime}, (31)

where

τd𝐩ρ(𝐩,𝐩),\tau^{\prime}\equiv\int\mbox{d}{\bf p}\;\rho^{\prime}(\mathbf{p},\mathbf{p}), (32)
τid𝐩ψi(𝐩)ψi(𝐩),\tau_{i}^{\prime}\equiv\int\mbox{d}{\bf p}\;\psi_{i}^{\prime}(\mathbf{p})\psi_{i}^{\prime}(\mathbf{p})^{\dagger}, (33)

and ρ(𝐩,𝐩)\rho^{\prime}(\mathbf{p},\mathbf{p}) is given in Eq (24). Explicitly, we have

τ0\displaystyle\tau_{0}^{\prime} =\displaystyle= (1V(α)00V(α)),\displaystyle\left(\begin{array}[]{cc}1-V(\alpha)&0\\ 0&V(\alpha)\end{array}\right), (36)
τ1\displaystyle\tau_{1}^{\prime} =\displaystyle= (A(α)B(α)B(α)1A(α)),\displaystyle\left(\begin{array}[]{cc}A(\alpha)&B(\alpha)\\ B(\alpha)&1-A(\alpha)\end{array}\right), (39)

where

A(α)\displaystyle A(\alpha) =\displaystyle= cos2θ[1U(α)]+sin2θU(α),\displaystyle\cos^{2}\theta\left[1-U(\alpha)\right]+\sin^{2}\theta\,U(\alpha), (40)
B(α)\displaystyle B(\alpha) =\displaystyle= cosθsinθ[14U(α)]\displaystyle\cos\theta\sin\theta\,\left[1-4\,U(\alpha)\right] (41)

with V(α)V(\alpha) and U(α)U(\alpha) being given by

V(α)sinh2(α2)d𝐪qz2|f𝐤0w0(𝐪)|2(q0+m)(p0+m),V(\alpha)\equiv\sinh^{2}\left(\frac{\alpha}{2}\right)\int\mbox{d}{\bf q}\,\frac{{q_{z}}^{2}\left|f_{\mathbf{k}_{0}}^{w_{0}}(\mathbf{q})\right|^{2}}{(q^{0}+m)(p^{0}+m)}, (42)

and

U(α)sinh2(α2)d𝐪qz2|f𝐤1w1(𝐪)|2(q0+m)(p0+m),U(\alpha)\equiv\sinh^{2}\left(\frac{\alpha}{2}\right)\int\mbox{d}{\bf q}\,\frac{{q_{z}}^{2}\left|f_{\mathbf{k}_{1}}^{w_{1}}(\mathbf{q})\right|^{2}}{(q^{0}+m)(p^{0}+m)}, (43)

respectively. In the above equations, we have used the fact that d𝐩/p0\mbox{d}{\bf p}/p^{0} is a relativistic invariant and performed the change of variables q=Λ1pq=\Lambda^{-1}p. Using Eqs. (36) and (39) in Eq (31), we can cast τ\tau^{\prime} as

τ=(λ(1V)+(1λ)A(1λ)B(1λ)BλV+(1λ)(1A)).\tau^{\prime}\!=\!\left(\!\!\!\begin{array}[]{cc}\lambda\left(1-V\right)+(1-\lambda)A&(1-\lambda)B\\ (1-\lambda)B&\lambda\,V+(1-\lambda)(1-A)\end{array}\!\!\!\right)\!\!. (44)

Note that, contrary to τ0\tau_{0} and τ1\tau_{1}, τ0\tau_{0}^{\prime} and τ1\tau_{1}^{\prime} are not pure states, so that S(τ0)S(\tau_{0}^{\prime}) and S(τ1)S(\tau_{1}^{\prime}) are both non-zero. The Holevo bound in Bob’s frame is given by

χ(τ)=S(τ)λS(τ0)(1λ)S(τ1).\chi(\tau^{\prime})=S(\tau^{\prime})-\lambda\,S(\tau_{0}^{\prime})-(1-\lambda)\,S(\tau_{1}^{\prime}). (45)

By using Eqs. (36), (39) and (44), the above equation can be rewritten as

χ(τ)\displaystyle\chi(\tau^{\prime}) =\displaystyle= l=±γllog2γl+λl=±δllog2δl\displaystyle-\sum_{l=\pm}\gamma_{l}\,\log_{2}\gamma_{l}+\lambda\sum_{l=\pm}\delta_{l}\,\log_{2}\delta_{l} (46)
+\displaystyle+ (1λ)l=±ϵllog2ϵl,\displaystyle(1-\lambda)\sum_{l=\pm}\epsilon_{l}\,\log_{2}\epsilon_{l},

where, for l=±l=\pm, γl\gamma_{l}, δl\delta_{l}, and ϵl\epsilon_{l} are the eigenvalues of τ\tau^{\prime}, τ0\tau_{0}^{\prime}, and τ1\tau_{1}^{\prime}, respectively. These can be easily calculated as functions of the integrals V(α)V(\alpha) and U(α)U(\alpha). In this paper, we proceed a numerical analysis of Eq. (46) to examine various aspects of the transmission of classical information through relativistic quantum channels. In order to do that, we rewrite Eqs. (42) and (43) as

V(α)=sinh2(α/2)πW03dQx0dQrG0(Qx,Qr),V(\alpha)=\frac{\sinh^{2}\left(\alpha/2\right)}{\sqrt{\pi}\;{W_{0}}^{3}}\int_{-\infty}^{\infty}\!\!\!\mbox{d}Q_{x}\int_{0}^{\infty}\!\!\!\mbox{d}Q_{r}\;G_{0}\left(Q_{x},Q_{r}\right), (47)

and

U(α)=sinh2(α/2)πW13dQx0dQrG1(Qx,Qr),U(\alpha)=\frac{\sinh^{2}\left(\alpha/2\right)}{\sqrt{\pi}\;{W_{1}}^{3}}\int_{-\infty}^{\infty}\!\!\!\mbox{d}Q_{x}\int_{0}^{\infty}\!\!\!\mbox{d}Q_{r}\;G_{1}\left(Q_{x},Q_{r}\right), (48)

respectively, where we have used Eq. (11) and cylindrical coordinates with qxq_{x} as the symmetry axis, and introduced

Gi(Qx,Qr)=Qr3exp{[(QxKi)2+Qr2]/Wi2}(Q0+1)(Q0coshαQxsinhα+1).G_{i}\left(Q_{x},Q_{r}\right)=\frac{{Q_{r}}^{3}\exp\left\{-\left[\left(Q_{x}-K_{i}\right)^{2}+{Q_{r}}^{2}\right]/{W_{i}}^{2}\right\}}{\left(Q^{0}+1\right)\left(Q^{0}\cosh\alpha-Q_{x}\sinh\alpha+1\right)}. (49)

In addition, we have defined the normalized non-dimensional variables Qrqr/mQ_{r}\equiv q_{r}/m, Qxqx/mQ_{x}\equiv q_{x}/m, Q0=Qx2+Qr2+1Q^{0}=\sqrt{{Q_{x}}^{2}+{Q_{r}}^{2}+1}, Wiwi/mW_{i}\equiv w_{i}/m and Kiki/mK_{i}\equiv k_{i}/m.

Refer to caption
Figure 2: (Color online) The Holevo bound χ(τ)\chi(\tau^{\prime}) is plotted as a function of α=tanh1v\alpha=-\tanh^{-1}v for different values of the normalized momentum K1=k1/mK_{1}=k_{1}/m. We have fixed W0=0.05,W_{0}=0.05, W1=6,W_{1}=6, λ=1/2,\lambda=1/2, K0=1K_{0}=1, and θ=π/8\theta=\pi/8. We can see that although χ(τ)\chi(\tau^{\prime}) decreases with α\alpha initially, it begins to increase for larger values of the rapidity, overcoming the value for α=0\alpha=0 when K1K_{1} is large enough.

Let us first analyze the behavior of Eq (45) [or, equivalently, Eq (46)] as a function of the rapidity α=tanh1v,\alpha=-\tanh^{-1}v, where we recall that vv is the relative velocity between Bob and Alice. First, it should be noted that if both W0W_{0} and W1W_{1} were much smaller than the unity, the effects of Bob’s velocity would be negligible and, for any value of α\alpha, χ(τ)\chi(\tau^{\prime}) would have, approximately, its value for α=0\alpha=0 (since in this case, ψ0\psi_{0} and ψ1\psi_{1} would be almost momentum eigenstates). However, if we allow W1W_{1} to be much larger than W0,W_{0}, the situation changes considerably, as can be seen in Fig. 2. There, χ(τ)\chi(\tau^{\prime}) is plotted as a function of the rapidity α\alpha, for different values of the normalized mean momentum K1K_{1} of ψ1\psi_{1}, when W0=0.05W_{0}=0.05, W1=6W_{1}=6, λ=1/2\lambda=1/2, K0=1K_{0}=1, and θ=π/8\theta=\pi/8. We can see that although χ(τ)\chi(\tau^{\prime}) initially decreases with α\alpha, it begins to increase for larger values of the rapidity, eventually overcoming the value for α=0\alpha=0 when K1K_{1} is large enough. Thus, for some codifications, i.e., for some values of θ\theta, the (relativistic) velocity of the receiver can actually increase his accessible information. This is possible because, when K1K_{1} is large, even though W1/W0102W_{1}/W_{0}\approx 10^{2}, the states ψ0\psi_{0} and ψ1\psi_{1} are almost orthogonal (despite of the fact that their spin parts, ϕ\phi_{\uparrow} and ϕθ\phi_{\theta}, are not), as can be easily checked. When Bob is moving, the momentum and spin degrees of freedom of the original state prepared by Alice are mixed up by the Wigner rotation. As a consequence, Bob can obtain some extra information in the spin degrees of freedom due to the fact that the states ψ0\psi_{0} and ψ1\psi_{1} prepared by Alice are more clearly distinguishable in momentum than in spin.

Refer to caption
Figure 3: (Color online) The graph shows the Holevo bound χ(τ)\chi(\tau^{\prime}) as a function of W1W_{1} for different values of the angle θ\theta when α\alpha\rightarrow\infty. We have fixed K0=1K_{0}=1, K1=10K_{1}=10, W0=0.05W_{0}=0.05, and λ=1/2.\lambda=1/2. When θ\theta is small and W1W_{1} is large enough, the accessible information always increases when Bob moves fast enough with respect to Alice. When θ\theta is larger (in particular, when it is closer to π/2\pi/2), the increase in W1W_{1} always leads to a decrease in χ(τ)\chi(\tau^{\prime}) and therefore in the accessible information.

In view of the above results, let us study now how the width of the wave packet of the second state influences the accessible information in different codification schemes. In Fig. 3, χ(τ)\chi(\tau^{\prime}) is plotted as a function of W1W_{1}, in the limit α\alpha\rightarrow\infty, for different values of θ\theta. We have fixed W0=0.05,W_{0}=0.05, K0=1,K_{0}=1, K1=10K_{1}=10, and λ=1/2\lambda=1/2. We can see that when θ\theta is small and W1W_{1} is large enough, the accessible information always increase when Bob moves with relativistic velocities with respect to Alice. When θ\theta is larger (in particular, when it is close to π/2\pi/2), the increase in W1W_{1} always leads to a decrease in χ(τ)\chi(\tau^{\prime}) and therefore in the accessible information. This is so because the spin parts, ϕ\phi_{\uparrow} and ϕθ,\phi_{\theta}, of the states ψ0\psi_{0} and ψ1\psi_{1} prepared by Alice are already quite distinguishable and thus the entanglement between the spin and momentum degrees of freedom caused by Bob’s movement ends up only making the spin degrees of freedom less distinguishable. It is interesting to note that Alice can codify the information in states ψ0\psi_{0} and ψ1\psi_{1} using θ=0\theta=0. This makes the states completely indistinguishable in spin in her frame. Yet, Bob can have a non zero value of χ(τ)\chi(\tau^{\prime}), as can be seen from Fig. 3. Thus, Alice can use the fact that spin and momentum are mixed up in Bob’s frame to “hide” the information in the momentum degrees of freedom in her frame, and let Bob movement make this information available to him in spin. This may be useful when considering noisy quantum channels in the spin degrees of freedom.

III.2 Four Classical Bits

We will analyze now if the above results still apply when Alice sends more classical bits through the quantum channel. For the sake of simplicity, we will describe here what happens when she tries to transmit four bits of classical information to Bob. Thus, as explained in Section II.2, Alice prepares the state ρ~\tilde{\rho} given in Eq. (21) and sends it to Bob. As Bob is moving with respect to her, he sees the state ρ~\tilde{\rho} prepared by Alice unitarily transformed in his proper frame as

ρ~(𝐩,𝐩~,𝐩,𝐩~)=λ1ψ00(𝐩,𝐩~)ψ00(𝐩,𝐩~)\displaystyle\tilde{\rho}^{\prime}\left(\mathbf{p},\mathbf{\tilde{p}},\mathbf{p}^{\prime},\mathbf{\tilde{p}}^{\prime}\right)=\lambda_{1}\,\psi_{00}^{\prime}({\bf p},\tilde{{\bf p}}){\psi^{\prime}}_{00}^{{\dagger}}({\bf p}^{\prime},\tilde{{\bf p}}^{\prime})
+λ2ψ01(𝐩,𝐩~)ψ01(𝐩,𝐩~)+λ3ψ10(𝐩,𝐩~)ψ10(𝐩,𝐩~)\displaystyle+\lambda_{2}\,\psi^{\prime}_{01}({\bf p},\tilde{{\bf p}}){\psi^{\prime}}_{01}^{{\dagger}}({\bf p}^{\prime},\tilde{{\bf p}}^{\prime})+\lambda_{3}\,\psi^{\prime}_{10}({\bf p},\tilde{{\bf p}}){\psi^{\prime}}_{10}^{{\dagger}}({\bf p}^{\prime},\tilde{{\bf p}}^{\prime})
+λ4ψ11(𝐩,𝐩~)ψ11(𝐩,𝐩~),\displaystyle+\lambda_{4}\,\psi^{\prime}_{11}({\bf p},\tilde{{\bf p}}){\psi^{\prime}}_{11}^{{\dagger}}({\bf p}^{\prime},\tilde{{\bf p}}^{\prime}), (50)

where ψij(𝐩,𝐩~)ψi(𝐩)ψj(𝐩~)\psi^{\prime}_{ij}({\bf p},\tilde{{\bf p}})\equiv\psi^{\prime}_{i}({\bf p})\otimes\psi^{\prime}_{j}(\tilde{{\bf p}}), i,j=0,1i,j=0,1, and ψ0\psi^{\prime}_{0} and ψ1\psi^{\prime}_{1} are given in Eqs. (29) and (30), respectively. If we trace out the momentum degrees of freedom in ρ~(𝐩,𝐩~,𝐩,𝐩~)\tilde{\rho}^{\prime}\left(\mathbf{p},\mathbf{\tilde{p}},\mathbf{p}^{\prime},\mathbf{\tilde{p}}^{\prime}\right) we obtain the density operator

τ~\displaystyle\tilde{\tau}^{\prime} =\displaystyle= d𝐩d𝐩~ρ~(𝐩,𝐩~,𝐩,𝐩~)\displaystyle\int\mbox{d}{\bf p}\mbox{d}\tilde{{\bf p}}\;\tilde{\rho}^{\prime}\left(\mathbf{p},\mathbf{\tilde{p}},\mathbf{p},\mathbf{\tilde{p}}\right) (51)

which can be written as

τ~\displaystyle\tilde{\tau}^{\prime} =\displaystyle= λ1τ0τ0+λ2τ0τ1\displaystyle\lambda_{1}\tau^{\prime}_{0}\otimes\tau^{\prime}_{0}+\lambda_{2}\tau^{\prime}_{0}\otimes\tau^{\prime}_{1} (52)
+\displaystyle+ λ3τ1τ0+λ4τ1τ1,\displaystyle\lambda_{3}\tau^{\prime}_{1}\otimes\tau^{\prime}_{0}+\lambda_{4}\tau^{\prime}_{1}\otimes\tau^{\prime}_{1},

and from which all spin measurement results can be predicted. We recall that τ0\tau^{\prime}_{0} and τ1\tau^{\prime}_{1} are given in Eqs. (36) and (39), respectively.

Refer to caption
Figure 4: (Color online) The graph exhibits the Holevo bound χ(τ~)\chi(\tilde{\tau}^{\prime}) as a function of α=tanh1v\alpha=-\tanh^{-1}v for different values of the angle θ\theta. We have chosen K0=1K_{0}=1, K1=50K_{1}=50, W0=0.05W_{0}=0.05, W1=6W_{1}=6, and λ1=λ2=λ3=λ4=1/4.\lambda_{1}=\lambda_{2}=\lambda_{3}=\lambda_{4}=1/4. For smaller angles (θ=0,\theta=0, θ=π/10\theta=\pi/10, and θ=π/8\theta=\pi/8), although χ(τ~)\chi(\tilde{\tau}^{\prime}) initially decreases with α\alpha it eventually increases as the rapidity gets larger, overcoming its value for α=0\alpha=0 when α\alpha\rightarrow\infty.

Now, using again the identity S(ω1ω2)=S(ω1)+S(ω2)S(\omega_{1}\otimes\omega_{2})=S(\omega_{1})+S(\omega_{2}), which is valid for any density matrices ω1\omega_{1} and ω2\omega_{2}, we can write the Holevo bound

χ(τ~)\displaystyle\chi(\tilde{\tau}^{\prime}) =\displaystyle= S(τ~)λ1S(τ0τ0)λ2S(τ0τ1)\displaystyle S(\tilde{\tau}^{\prime})-\lambda_{1}\,S(\tau_{0}^{\prime}\otimes\tau_{0}^{\prime})-\lambda_{2}\,S(\tau_{0}^{\prime}\otimes\tau_{1}^{\prime}) (53)
\displaystyle- λ3S(τ1τ0)λ4S(τ1τ1)\displaystyle\lambda_{3}\,S(\tau_{1}^{\prime}\otimes\tau_{0}^{\prime})-\lambda_{4}\,S(\tau_{1}^{\prime}\otimes\tau_{1}^{\prime})

as

χ(τ~)\displaystyle\chi(\tilde{\tau}^{\prime}) =\displaystyle= l~=14γ~l~log2γ~l~2λ1S(τ0)2λ4S(τ1)\displaystyle-\sum_{\tilde{l}=1}^{4}\tilde{\gamma}_{\tilde{l}}\,\log_{2}\tilde{\gamma}_{\tilde{l}}-2\lambda_{1}S(\tau_{0}^{\prime})-2\lambda_{4}S(\tau_{1}^{\prime}) (54)
\displaystyle- (λ2+λ3)[S(τ0)+S(τ1)],\displaystyle(\lambda_{2}+\lambda_{3})[S(\tau_{0}^{\prime})+S(\tau_{1}^{\prime})],

where γ~l~,\tilde{\gamma}_{\tilde{l}}, l~=1,2,3,4,\tilde{l}=1,2,3,4, are the eigenvalues of τ~\tilde{\tau}^{\prime}.

In Fig. 4 we plot χ(τ~)\chi(\tilde{\tau}^{\prime}) as a function of the rapidity α\alpha for different values of the angle θ\theta. We have verified that the behavior of the Holevo bound for τ~\tilde{\tau}^{\prime} as a function of α\alpha for different values of K1K_{1} is very similar to that of χ(τ)\chi(\tau^{\prime}), which is shown in Fig. 2. Similarly, the behavior χ(τ~)\chi(\tilde{\tau}^{\prime}), when α\alpha\rightarrow\infty, as a function of W1W_{1} looks very close to the behavior of χ(τ)\chi(\tau^{\prime}) described in Fig. 3. Therefore, in order to analyze the Holevo bound for different choices of the angle θ\theta of ψ1\psi_{1}, we have fixed K1K_{1} and W1W_{1} much larger than K0K_{0} and W0W_{0}, respectively. Explicitly, we have chosen K0=1K_{0}=1, K1=50K_{1}=50, W0=0.05W_{0}=0.05, W1=6W_{1}=6, and λ1=λ2=λ3=λ4=1/4.\lambda_{1}=\lambda_{2}=\lambda_{3}=\lambda_{4}=1/4. As can be seen from Fig. 4, χ(τ~)\chi(\tilde{\tau}^{\prime}) initially decreases as α\alpha increases. However, for small angles (for instance, θ=0,\theta=0, θ=π/10\theta=\pi/10, and θ=π/8\theta=\pi/8), the Holevo bound begins to increase as α\alpha gets larger and, eventually, χ(τ~)\chi(\tilde{\tau}^{\prime}) overcomes its value for α=0\alpha=0 (where Bob is at rest with respect to Alice). Thus, for some codifications of the classical bits, the (relativistic) velocity of the receiver can increase his accessible information. For angles closer to π/2,\pi/2, Bob’s movement only makes the Holevo bound smaller compared to the case where Alice and Bob do not have a relative motion. In particular, we can see from the graph that, when α\alpha is large enough, the use of orthogonal states, θ=π/2,\theta=\pi/2, is not even the best strategy anymore [i.e. orthogonal states do not maximize χ(τ~)\chi(\tilde{\tau}^{\prime})]. Thus, for larger angles, it is better to use both W0W_{0} and W1W_{1} much smaller than 11 so that ψ0\psi_{0} and ψ1\psi_{1} are “almost” momentum eigenstates and therefore, the effects of the motion of the receiver on the accessible information are negligible.

It is interesting to note that, as in the case where two bits are being sent through the quantum channel, Alice can “hide” the information on the momentum degrees of freedom by using states ψij,\psi_{ij}, i,j=0,1i,j=0,1, with θ=0\theta=0 (which are completely indistinguishable in spin), and let Bob’s movement make the information available to him in the spin degrees of freedom. As we have already pointed out in Section III.1, this might be useful to protect the communication against possible noises in the spin degrees of freedom.

IV Relativistic Quantum Map

Refer to caption
Figure 5: (Color online) The graph shows Δ2χ[(τ)]χ(τ)\Delta_{2}\equiv\chi\left[\mathcal{E}(\tau)\right]-\chi(\tau) as a function of θ\theta for different values of λ\lambda when α\alpha\rightarrow\infty. We have fixed W0=0.05,W_{0}=0.05, W1=6,W_{1}=6, K0=1K_{0}=1, and K1=50K_{1}=50. We can see that there are two range of angles in which Δ20\Delta_{2}\geq 0 and therefore the map \mathcal{E} is not CP for angles within these ranges.

Let us now define linear and trace-preserving maps that describe the relativistic quantum channels for both the transmission of two and four classical bits. For this purpose, we fix the values of θ(0<θ<π),\theta\;(0<\theta<\pi), KiK_{i}, and WiW_{i}, i=0,1,i=0,1, and define the convex sets

𝒰{λτ0+(1λ)τ1|0λ1},\mathcal{U}\equiv\{\lambda\tau_{0}+(1-\lambda)\tau_{1}|0\leq\lambda\leq 1\}, (55)

and

𝒲\displaystyle\mathcal{W}\equiv\!\! {λ1τ0τ0+λ2τ0τ1+λ3τ1τ0+λ4τ1τ1\displaystyle\!\!\!\bigl{\{}\lambda_{1}\tau_{0}\otimes\tau_{0}+\lambda_{2}\tau_{0}\otimes\tau_{1}+\lambda_{3}\tau_{1}\otimes\tau_{0}+\lambda_{4}\tau_{1}\otimes\tau_{1} (56)
|\displaystyle| 0λi1,i=14λi=1},\displaystyle 0\leq\lambda_{i}\leq 1,\sum_{i=1}^{4}\lambda_{i}=1\bigr{\}},

where τ0\tau_{0} and τ1\tau_{1} are given in Eqs. (15) and (16), respectively. If ()\mathcal{B}(\mathcal{H}) denotes the set of (bounded) operators over a Hilbert space \mathcal{H}, the maps :𝒰(2)\mathcal{E}:\mathcal{U}\rightarrow\mathcal{B}(\mathbb{C}^{2}) and 𝒩:𝒲(22)\mathcal{N}:\mathcal{W}\rightarrow\mathcal{B}(\mathbb{C}^{2}\otimes\mathbb{C}^{2}) are defined as

Refer to caption
Figure 6: (Color online) The graph shows Δ4χ[𝒩(τ~)]χ(τ~)\Delta_{4}\equiv\chi\left[\mathcal{N}(\tilde{\tau})\right]-\chi(\tilde{\tau}) as a function of θ\theta for different values of λ1,λ2,λ3\lambda_{1},\lambda_{2},\lambda_{3} when α\alpha\rightarrow\infty. We have fixed W0=0.05,W_{0}=0.05, W1=6,W_{1}=6, K0=1K_{0}=1, and K1=50K_{1}=50. We can see that there are two range of angles in which Δ40\Delta_{4}\geq 0 and therefore the map 𝒩\mathcal{N} is not CP for angles within these ranges.
(τ)d𝐩[U(Λ)ρ(𝐩,𝐩)U(Λ)]\mathcal{E}(\tau)\equiv\int\mbox{d}{\bf p}\left[U(\Lambda)\rho({\bf p},{\bf p})U^{\dagger}(\Lambda)\right] (57)

and

𝒩(τ~)d𝐩d𝐩[U(Λ)U(Λ)ρ~(𝐩,𝐩,𝐩,𝐩)U(Λ)U(Λ)]\!\mathcal{N}(\tilde{\tau})\!\equiv\!\!\!\int\!\!\!\mbox{d}{\bf p}\mbox{d}{\bf p}^{\prime}\!\left[U(\Lambda)\!\otimes\!U(\Lambda)\tilde{\rho}({\bf p},{\bf p}^{\prime},{\bf p},{\bf p}^{\prime})U^{\dagger}(\Lambda)\!\otimes\!U^{\dagger}(\Lambda)\right] (58)

where ρ(𝐩,𝐩~)\rho\left(\mathbf{p},\mathbf{\tilde{p}}\right), ρ~(𝐩,𝐩~,𝐩,𝐩~)\tilde{\rho}\left(\mathbf{p},\mathbf{\tilde{p}},\mathbf{p}^{\prime},\mathbf{\tilde{p}}^{\prime}\right), and U(Λ)U(\Lambda) are given in Eqs. (13), (21), and (26), respectively.

It is interesting to note that when W0W1W_{0}\ll W_{1} and K0K1K_{0}\ll K_{1}, the total state prepared by Alice, ρ(𝐩,𝐩~)\rho({\bf p},{\bf\tilde{p}}) for the two-bit case and ρ(𝐩,𝐩~,𝐩,𝐩~)\rho({\bf p},{\bf\tilde{p}},{\bf p}^{\prime},{\bf\tilde{p}}^{\prime}) for the four-bit case, not only presents correlations between the spin and momentum degrees of freedom but also have non-vanishing quantum discord 𝒟\mathcal{D} (with respect to measurements made on the spin degrees of freedom). For the sake of simplicity, let us prove this statement for the state ρ(𝐩,𝐩~)\rho({\bf p},{\bf\tilde{p}}). However, a complete analogous calculation also shows the non-vanishing of the quantum discord in the four-bits case. We first note that

[τI𝐩,ρ]=λ(1λ)cosθ(ϕϕθϕθϕ)(ρ1ρ0),\left[\tau\otimes I_{\bf p},\rho\right]=\lambda(1-\lambda)\cos\theta\left(\phi_{\uparrow}\phi_{\theta}^{\dagger}-\phi_{\theta}\phi_{\uparrow}^{\dagger}\right)\otimes(\rho_{1}-\rho_{0}), (59)

where I𝐩I_{\bf p} is the identity in the momentum space,

τd𝐩ρ(𝐩,𝐩)𝒰,\tau\equiv\int\mbox{d}{\bf p}\rho({\bf p},{\bf p})\in\mathcal{U},

and, in the momentum representation, ρi(𝐩,𝐩)f𝐤iwi(𝐩)f𝐤iwi(𝐩),\rho_{i}({\bf p},{\bf p}^{\prime})\equiv f_{{\bf k}_{i}}^{w_{i}}({\bf p})f_{{\bf k}_{i}}^{*w_{i}}({\bf p}^{\prime}), i.e., they are the reduced momentum density operator associated with ψi\psi_{i}. Now, we can see from Eq. (59) that whenever ρ0ρ1\rho_{0}\neq\rho_{1} (which is the case when W0W1W_{0}\ll W_{1} and K0K1K_{0}\ll K_{1}) the above commutator does not vanish and thus, as shown in facca10 , the state has non-vanishing quantum discord. Although it is not a sufficient condition, an initial state with non-vanishing discord indicates that the reduced spin dynamics (\mathcal{E} and 𝒩\mathcal{N}) may not be completely positive (CP) bdmrr12 ; mbcpv12 ; sl09 .

Before we proceed to investigate this possibility, it is important to remark that one must be careful in defining an effective dynamics when there are some prior correlations between the system of interest and its environment, which in our case are the spin and momentum degrees of freedom of spin 1/21/2 fermions, respectively. For example, even if the map describing the effective dynamics is linear and trace preserving, it may fail to be positive. Thus, to define a physically reasonable dynamics, one must restrict the domain of the map to a set in which it takes positive operators into positive operators mbcpv12 ; jss04 ; ctz08 ; mra12 ; rmg10 .

The maps \mathcal{E} and 𝒩\mathcal{N}, given in Eqs. (57) and (58), describe the relativistic quantum channels for transmitting two and four bits of classical information, respectively. It is easy to see that they are convex-linear, trace preserving (and can be extended, in a non-unique way, to linear and trace preserving maps acting on all linear operators), and positive. Thus, at least on 𝒰\mathcal{U} and 𝒲\mathcal{W}, the maps \mathcal{E} and 𝒩\mathcal{N} are well defined and describe the effective spin dynamics (which is enough for our purposes). It is well know that for any quantum map, i.e., any linear, trace-preserving, and CP operator 𝒦:()()\mathcal{K}:\mathcal{B}(\mathcal{H})\rightarrow\mathcal{B}(\mathcal{H}), where \mathcal{H} is a Hilbert space, the Holevo bound satisfies nielsen&chuang

χ[𝒦(ρ)]χ(ρ),\chi\left[\mathcal{K}(\rho)\right]\leq\chi(\rho), (60)

where ρ\rho is any density operator defined on .\mathcal{H}. Thus, quantum maps cannot increase the accessible information on the receiver. In Figs. 5 and 6 we plot

Δ2χ[(τ)]χ(τ)\Delta_{2}\equiv\chi\left[\mathcal{E}(\tau)\right]-\chi(\tau) (61)

and

Δ4χ[𝒩(τ~)]χ(τ~)\Delta_{4}\equiv\chi\left[\mathcal{N}(\tilde{\tau})\right]-\chi(\tilde{\tau}) (62)

as functions of θ\theta for different probability distributions {λ,1λ}\{\lambda,1-\lambda\} and {λ1,λ2,λ3,λ4}\{\lambda_{1},\lambda_{2},\lambda_{3},\lambda_{4}\}, respectively. In both figures we have fixed W0=0.05,W_{0}=0.05, W1=6,W_{1}=6, K0=1,K_{0}=1, and K1=50K_{1}=50. We can see from the plots that there are cases in which

χ[(τ)]χ(τ)\chi\left[\mathcal{E}(\tau)\right]\geq\chi(\tau) (63)

and

χ[𝒩(τ~)]χ(τ~).\chi\left[\mathcal{N}(\tilde{\tau})\right]\geq\chi(\tilde{\tau}). (64)

Therefore, in the relativistic case, if one wants to always maximize the accessible information on the receiver, there will be quantum channels that cannot be described by quantum maps, i.e. linear, trace-preserving, and CP operators (the impossibility to describe quantum channels by CP maps in relativistic setups using photons as information carriers was noticed in PT02 ). This will be the case when the classical information is encoded in states {ψ0,ψ1}\{\psi_{0},\psi_{1}\} with the angle θ\theta characterizing the state ψ1\psi_{1} being smaller (larger) than some angle ϑ\vartheta (πϑ\pi-\vartheta), with the value of ϑ\vartheta depending on the number of bits and on their probability distribution, as can be seen from Figs. 5 and 6.

For angles closer to π/2\pi/2 however, as we have already pointed out, the best strategy is to keep both W0W_{0} and W1W_{1} small to minimize the effects of the boost on the states. In the particular case where W0=W1W_{0}=W_{1}, K0=K1K_{0}=K_{1}, and θ=π/2\theta=\pi/2, it is easy to see that both \mathcal{E} and 𝒩\mathcal{N} are CPCP maps with Krauss decompositions

(τ)=μ=13ΓμτΓμ\mathcal{E}(\tau)=\sum_{\mu=1}^{3}\Gamma_{\mu}\tau\Gamma_{\mu} (65)
𝒩(τ~)=()(τ~)=μ,ν=13[ΓμΓν]τ~[ΓμΓν]\mathcal{N}(\tilde{\tau})=\left(\mathcal{E}\otimes\mathcal{E}\right)(\tilde{\tau})=\sum_{\mu,\nu=1}^{3}\left[\Gamma_{\mu}\otimes\Gamma_{\nu}\right]\tilde{\tau}\left[\Gamma_{\mu}\otimes\Gamma_{\nu}\right] (66)

respectively, where the Krauss operators Γμ\Gamma_{\mu} are given by

Γ11V(α)I,Γ2V(α)2σx,Γ3V(α)2σy,\displaystyle\Gamma_{1}\equiv\sqrt{1-V(\alpha)}\;I,\Gamma_{2}\equiv\sqrt{\frac{V(\alpha)}{2}}\;\sigma_{\rm x},\;\Gamma_{3}\equiv\sqrt{\frac{V(\alpha)}{2}}\;\sigma_{\rm y},

with II being the identity operator and V(α)V(\alpha) being given in Eq. (47).

Even in the CP regime, when one is dealing with quantum process tomography the initial correlations between the system of interest and its environment must be treated carefully. In such cases the preparation procedure plays a crucial role and the tomographically reconstructed quantum map may differ from the dynamical quantum map kmrs07 ; bgtw11 . It would be very interesting to investigate such issues in these relativistic scenarios.

V Final Remarks

In the present paper, we have used the Holevo bound to analyze how the relative motion between the sender and the receiver influences the capacity of a quantum communication channel to convey classical information. To this end, we have assumed that the sender, Alice, encodes the classical information in the spin degrees of freedom of spin-1/21/2 fermions of mass mm and sends the state prepared to the receiver, Bob, who is moving with velocity v=tanhαv=-\tanh\alpha with respect to her. Bob then makes a spin measurement on the state and has to identify the message sent by Alice based on its measurement outcome.

First it was analyzed the case where Alice has a classical information source that produces symbols X=0,1X=0,1 according to the probability distribution p0,p1p_{0},p_{1}. Depending on the value of XX, Alice prepares the spin-1/21/2 particle in a pure quantum state ψX\psi_{X} and sends it to Bob. The spin part of ψ0\psi_{0} and ψ1\psi_{1} were assumed to be eigenstates, with eigenvalue 1/21/2, of SzS_{\rm z} and 𝐒𝐧{\bf S}\cdot{\bf n}, with 𝐧=(sin2θ,0,cos2θ){\bf n}=(\sin 2\theta,0,\cos 2\theta), respectively. It was shown that when θ\theta is “close” to π/2\pi/2, Bob’s movement always reduces the Holevo bound and thus, the best strategy in this case is to use very narrow wave packets in the momentum degrees of freedom. This way, the states ψ0\psi_{0} and ψ1\psi_{1} are almost momentum eigenstates and therefore the effects of the boost on them are negligible. For small angles however, if Alice chooses W0W1W_{0}\ll W_{1} and K1K_{1} large enough, and Bob moves sufficiently fast, the Holevo bound χ(τ)\chi(\tau^{\prime}) overcomes χ(τ)\chi(\tau), the Holevo bound in the case where Bob is at rest relative to Alice.

We have also analyzed how the above results generalize to the case where more classical bits are sent through the quantum channel. We have shown explicitly in the case where Alice has an information source that produces four bits X~=00,01,10,11\tilde{X}=00,01,10,11 according to the probability distribution p00,p01,p10,p11p_{00},p_{01},p_{10},p_{11} and codifies each bit in one of the product states ψiψj,\psi_{i}\otimes\psi_{j}, i,j=0,1,i,j=0,1, that the conclusions reached in the two bit case can be extended to this one.

The use of non-orthogonal quantum states to convey classical information is important not only due to cryptographic purposes but also due to the fact that there are noisy quantum channels in which the optimal rate of information transmission is achieved only by using non-orthogonal states fucs97 . Our results seem to indicate that when the classical information is conveyed through such noisy quantum channels and the sender and receiver are in relativistic relative motion, one might actually increase the channel capacity by carefully preparing the momentum degrees of freedom of the states. As we have shown, this is possible because, relativistically, boosts entangle the spin and momentum degrees of freedom of a spin-1/21/2 particle. Therefore, owing to his motion, the receiver can obtain some extra information in the spin degrees of freedom due to the fact that the states prepared by the sender are more distinguishable in momentum than in spin.

Acknowledgements.
The authors are indebted to George Matsas for useful discussions and for reading the manuscript. We also thank Roberto Serra for his useful comments and for pointing out Ref. [40] to us. A.L. and A.T. acknowledge partial and full support from the Brazilian National Institute for Science and Technology of Quantum Information (INCT-IQ) and Coordenação de Aperfeiçoamento de Pessoal de Nível Superior (CAPES) respectivelly.

References

  • (1) P. W. Shor, SIAM J. Sci. Stat. Comput. 26, 1484 (1997).
  • (2) C. H. Bennett, G. Brassard, C. Crépeau, R. Jozsa, A. Peres, and W. K. Wootters, Phys. Rev. Lett. 70, 1895 (1993).
  • (3) S. Wiesner, SIGACT News 15, 78 (1983).
  • (4) C. H. Bennett and G. Brassard, in Proceedings of IEEE International Conference on Computers, Systems, and Signal Processing, Bangalore, 1984 (IEEE, New York, 1984), p. 175.
  • (5) A. K. Ekert, Phys. Rev. Lett. 67, 661 (1991).
  • (6) C. H. Bennett and S. J. Wiesner, Phys. Rev. Lett. 69, 2881 (1992).
  • (7) P. W. Shor, Phys. Rev. A 52, R2493 (1995).
  • (8) M. A. Nielsen and I. L. Chuang, Quantum Computation and Quantum Information (Cambridge University Press, Cambridge, 2000).
  • (9) A. Peres and D. R. Terno, Int. J. Quant. Info. 1, 225 (2003).
  • (10) A. Peres and D. R. Terno, Rev. Mod. Phys. 76, 93 (2004).
  • (11) S. W. Hawking, Phys. Rev. D 14, 2460 (1976).
  • (12) R. M. Wald, Living Rev. Rel. 4, 6 (2001).
  • (13) A. Peres, P. F. Scudo, and D. R. Terno, Phys. Rev. Lett. 88, 230402 (2002).
  • (14) R. M. Gingrich and C. Adami, Phys. Rev. Lett. 89, 270402 (2002).
  • (15) I. Fuentes-Schuller and R. B. Mann, Phys. Rev. Lett. 95, 120404 (2005) E. Martín-Martínez and I. Fuentes, Phys. Rev. A 83, 052306 (2011). G. Adesso, I. Fuentes-Schuller, and M. Ericsson, Phys. Rev. A 76, 062112 (2007).
  • (16) N. Friis, P. Kohler, ​ E. Martín-Martínez, and R. A. Bertlmann, Phys. Rev. A 84, 062111 (2011).
  • (17) A. G. S. Landulfo and G. E. A. Matsas, Phys. Rev. A 80, 032315 (2009).
  • (18) L. C. Céleri, A. G. S. Landulfo, R. M. Serra and G. E. A. Matsas, Phys. Rev. A 81, 062130 (2010).
  • (19) Q. Pan and J. Jing, Phys. Rev. D 78, 065015 (2008).
  • (20) E. Martín-Martínez and J. Leon, Phys. Rev. A 81, 032320 (2010).
  • (21) D. Hosler, C. van de Bruck, and P. Kok, Phys. Rev. A 85, 042312 (2012).
  • (22) M. Aspelmeyer, T. Jennewein, M. Pfennigbauer, W. Leeb, and A. Zeilinger, Selected Topics in Quantum Electronics, IEEE Journal of 9, 1541 (2003).
  • (23) P. Villoresi et al., New J. Phys. 10, 033038 (2008).
  • (24) R. Ursin et al., Space-QUEST: Experiments with quantum entanglement in space. IAC Proc. A2.1.3 (2008). arXiv: quant-ph/0806.0945v1.
  • (25) C. Bonato, A. Tomaello, V. Da Deppo, G. Naletto, and P. Villoresi, New J. Phys. 11, 045017 (2009).
  • (26) A. G. S. Landulfo and G. E. A. Matsas, Phys. Rev. A 79, 044103 (2009).
  • (27) A. G. S. Landulfo, G. E. A. Matsas, and A. C. Torres, Phys. Rev. A 81, 044103 (2010).
  • (28) J. F. Clauser, M. A. Horne, A. Shimony, and R. A. Holt, Phys. Rev. Lett. 23, 880 (1969).
  • (29) W. K. Wootters and W. H. Zurek, Nature 299, 802 (1982).
  • (30) V. Scarani, S. Iblisdir, and N. Gisin, Rev. Mod. Phys. 77, 1225 (2005).
  • (31) A. S. Holevo, in Proceedings of the Second Japan-USSR Symposium on Probability Theory, edited by G. Maruyama and J. V. Prokhorov (Springer-Verlag, Berlin, 1973), p. 104.
  • (32) B. Schumacher and M. D. Westmoreland, Phys. Rev. A 56, 131 (1997).
  • (33) A. S. Holevo, IEEE Trans. Inf. Theory 44, 269 (1998).
  • (34) C. A. Fuchs, Phys. Rev. Lett. 79, 1162 (1997).
  • (35) N. N. Bogolubov, A. A. Logunov and I. T. Todorov, Introduction to Axiomatic Quantum Field Theory (W. A. Benjamin, Massachusetts, 1975).
  • (36) F. R. Halpern, Special Relativity and Quantum Mechanics, (Prentice-Hall, Englewood Cliffs, NJ, 1968).
  • (37) S. Weinberg, The Quantum Theory of Fields (Cambridge University Press, Cambridge, 1996), Vol. I.
  • (38) E. P. Wigner, Ann. Math. 40, 149 (1939).
  • (39) A. Ferraro, L. Aolita, D. Cavalcanti, F. M. Cucchietti, and A. Acín, Phys. Rev. A 81, 052318 (2010).
  • (40) A. Brodutch, A. Datta, K. Modi, Ángel Rivas, and C. A. Rodríguez-Rosário, Phys. Rev. A 87, 042301 (2013).
  • (41) K. Modi, A. Brodutch, H. Cable, T. Paterek, and V. Vedral, Rev. Mod. Phys. 84, 1655 (2012).
  • (42) A. Shabani and D. A. Lidar, Phys. Rev. Lett 102, 100402 (2009).
  • (43) T. F. Jordan, A. Shaji, and E. C. G. Sudarshan, Phys. Rev. A 70, 052110 (2004).
  • (44) H. A. Carteret, D. R. Terno, and Karol Życzkowski, Phys. Rev. A 77, 042113 (2008).
  • (45) C. A. Rodríguez-Rosário, K. Modi, and A. Aspuru-Guzik, Phys. Rev. A 81, 012313 (2010).
  • (46) F. Masillo, G. Scolarici, and L. Solombrino, J. Math. Phys. 52, 012101 (2011).
  • (47) A. Peres and D. R. Terno, J. Mod. Optics 50, 1165 (2003).
  • (48) A. M. Kuah, K. Modi, C. A. Rodríguez-Rosário, and E. C. G. Sudarshan, Phys. Rev. A 76, 042113 (2007).
  • (49) A. Brodutch, A. Gilchrist, D. R. Terno, and C. J. Wood, J. Phys.: Conf. Ser. 306, 012030 (2011).