Arxiv:1508.06383V1 [Quant-Ph] 26 Aug 2015 Bles at Room Temperature [19–21], Bose-Einstein Conden- Is So: There Are Two Possibilities

Decoherence of Einstein-Podolsky-Rosen steering

L. Rosales-Zárate, R. Y. Teh, S. Kiesewetter, A. Brolis, K. Ng and M. D. Reid1 1Centre for Quantum and Optical Science, Swinburne University of Technology, Melbourne, 3122 Australia We consider two systems A and B that share Einstein-Podolsky-Rosen (EPR) steering correlations and study how these correlations will decay, when each of the systems are independently coupled to a reservoir. EPR steering is a directional form of entanglement, and the measure of steering can change depending on whether the system A is steered by B, or vice versa. First, we examine the decay of the steering correlations of the two-mode squeezed state. We find that if the system B is coupled to a reservoir, then the decoherence of the steering of A by B is particularly marked, to the extent that there is a sudden death of steering after a finite time. We find a different directional effect, if the reservoirs are thermally excited. Second, we study the decoherence of the steering of a Schrodinger cat state, modelled as the entangled state of a spin and harmonic oscillator, when the macroscopic system (the cat) is coupled to a reservoir. OCIS: 270.0270; 270.6570; 270.5568

I. INTRODUCTION

The sort of nonlocality we call “Einstein-Podolsky- Rosen-steering” [1–6] originated in 1935 with the Einstein-Podolsky-Rosen (EPR) paradox [7]. The EPR paradox is the argument that was put forward by Einstein, Podolsky and Rosen for the incompleteness of quantum mechanics. The argument was based on premises (sometimes called Local Realism or in Einstein’s language, no “spooky action-at-a-distance”) that were not restricted to classical mechanics, but were thought essen- tial to any physical theory [8]. The EPR argument re- veals the inconsistency between Local Realism and the completeness of quantum mechanics. It does not in itself Figure 1. Our question is this: We consider two well-separated rule out all completions of quantum mechanics, that are systems, A and B, that possess the type of quantum corre- compatible with local realism. Nowadays, after the work lation we call EPR-steering. Each system is independently of Bell, we realise this ruling out of all local realistic the- coupled to a thermal reservoir, at a time t = 0. This coupling ories can be done in some special experimental situations induces a decay of the EPR-steering. We ask how does the [9–12]. Any realisation of the EPR paradox [13–15], as a decay of the steering depend on: the coupling γa, γb to each special case of EPR steering, is nonetheless important in reservoir; the thermal excitation na, nb of each reservoir; and the measure that gives the strength of the EPR-steering. We giving a concrete intermediate result: the fact that quan- will consider two types of EPR systems: The first is a two- tum mechanics without completion cannot be consistent mode squeezed state. The second is a mesoscopic/ macro- with local realism. A great advantage of EPR-steering scopic entangled “Schrodinger cat” state. over Bell-type tests is that they are more accessible to mesoscopic or macroscopic systems. EPR steering tests have been quantitatively studied or proposed for opti- experiments can justify the claim of loophole-free EPR cal down conversion [14–17], optical systems in nonlinear correlations [36], or steering without detection efficiency regimes near or at critical points [18], atomic gas ensem- loopholes [14, 15, 37, 38]. This paper examines why this arXiv:1508.06383v1 [quant-ph] 26 Aug 2015 bles at room temperature [19–21], Bose-Einstein conden- is so: There are two possibilities. sates (BEC) [22–31], and opto-mechanics [32–35]. This (1) It could be that quantum mechanics needs modi- may lead to experimental tests of Schrodinger cat-type fication to the extent that EPR correlations cannot be states. predicted. If we accept irrefutable evidence now exists The EPR paradox manifests as a strong correlation for EPR correlations, then this probability would appear between the positions and momenta of two spatially sep- falsified. However, we comment that the experimental re- arated particles (or some equivalent correlations). These alisations for EPR steering have been for optical systems, correlations, once one assumes Local Realism, become not yet for atoms or mesoscopic devices. inconsistent with the uncertainty principle. Given the (2) Or else: a popular belief is that quantum mechan- strangeness of the EPR correlations, a likely hypothe- ics is correct as it is, and predicts exactly why, to a great sis would be that they do not exist in real physical sys- accuracy, the EPR correlations are extraordinarily diffi- tems. Indeed, the sort of correlations needed for EPR are cult to measure. In that case, these predictions need to not easily realisable in experiment. To date, only a few be evaluated and verified by experiment. 2

This last point is what we examine in this paper: We One of the nicest examples of such EPR steering has study in detail why, according to quantum predictions, been realised experimentally in optics [15] as the two- the EPR correlations are not easily realisable. As with mode squeezed state [69]. Here, each mode of the light the well known “Schrodinger cat” [39–48], we expect this field is modeled as a quantum harmonic oscillator. An ex- is so, because of the effect that occurs when a quantum ample of a coupling between the oscillators that generates system is coupled to its environment, modelled as a large this type of state is called parametric down conversion, system - a reservoir [49]. This effect is called “decoher- and can be described by the interaction Hamiltonian in ence”. an appropriate rotating frame [16] Here, we examine some different sorts of decoherence H = iκE(ab − a†b†) (1) that occur for EPR steering. As with the decay of entanglement, there are many cases one can study. It is Similar two-mode squeezed states can be generated by worth noting that EPR steering was realised using high impinging two single-mode squeezed states into the two efficiency detection for optical amplitudes [14, 15], and different input ports of an optical beam splitter [70]. has also been unambiguously detected in photonic sce- The two-mode squeezed state system is fundamentally narios with exceptional losses (∼ 87%) [38]. Tests have interesting, as it can be quite accurately realised in op- also begun for its genuine multipartite form, distributed tics, but also it gives a reasonable approximation to the among different locations [50–53]. effects we expect to see in many other physical realisa- In this paper, we restrict our investigation to the fol- tions of steering, such as in opto-mechanics [32, 34, 35], lowing cases: We study position-momentum measure- atomic ensembles [19–21], and BEC [22–30]. The study of ments or their optical equivalent, quadrature phase am- the effect of the thermal environment on the EPR steer- plitude measurements. First, we analyze the steering of a ing for a two-mode squeezed state will therefore give us two-mode squeezed state, to understand the decoherence insight into a broad set of physical scenarios. The two- effects of coupling to a thermal reservoir. We examine mode quantum correlation effects were noticed originally two properties of the decoherence: the effect of simple in the context of photon number correlations between the losses (damping); and the effect of thermal noise. Sec- two beams, which gives rise to noise reduction [71, 72]. ond, we consider a special case of EPR-steering, that relates to a macroscopic/mesoscopic superposition state - a “Schrodinger cat” state. In fact, the EPR-steering A. The two-mode squeezing Hamiltonian paradox can be used to signify the Schrodinger cat superposition [54]. We examine the effect of decoherence We begin by reviewing the solution for the EPR cor- as caused by interaction with the environment, on this relations given by the Hamiltonian of Eq. (1). The signature. EPR solutions were originally derived in [16]. We de- An interesting new feature evident for EPR-steering fine the quadrature phase amplitudes XA, PA, XB and † † is that the nonlocality can manifest asymmetrically with PB for each mode: XA = a + a ,PA = (a − a )/i and † † respect to the observers [55–60]. In our results, we focus XB = b + b ,PB = (b − b )/i. This choice of scaling for on the asymmetry of the decoherence effects on the EPR- the amplitudes will imply the Heisenberg uncertainty re- steering. This is valuable to understand, not only from lations ∆XA∆PA ≥ 1 and ∆XB∆PB ≥ 1. We note that the fundamental perspective of testing quantum mechan- depending on the physical system modelled by the Hamil- ics, but from the point of view of the potential applica- tonian, the amplitudes can also correspond to scaled po- tions of EPR-steering, which include cryptography [61– sition and momentum observables. We now ask what 66] and no-cloning teleportation [65]. One-sided device- correlations are generated between the quadrature ampli- independent cryptography has been proposed using EPR tudes, after an interaction time τ between the two modes, steering inequalities [64]. In recent papers, it was shown at sites denoted by A and B. how the asymmetric effects of loss on EPR-steering could On examining (1), the resulting coupled equations for affect the optimal location of teleportation stations [67], a and b can be readily solved, to give or which entanglement criterion should be used to faith- p 2 † fully verify entanglement in cases of untrusted devices a(τ) = ηa(0) − (η − 1)b (0) [64–68]. b(τ) = ηb(0) − p(η2 − 1)a†(0) (2) where η = cosh r. We define the squeezing parameter as r = Ek τ. The quadrature phase amplitudes are given by II. DECOHERENCE OF STEERING FOR THE ~ TWO-MODE SQUEEZED STATE XA (τ) = cosh rXA (0) − sinh rXB (0)

PA (τ) = cosh rPA (0) + sinh rPB (0) Consider two quantum harmonic oscillators, with bo- XB (τ) = cosh rXB (0) − sinh rXA (0) son operators a and b, and depicted in the Figure 1 as P (τ) = cosh rP (0) + sinh rP (0) (3) systems A and B. Such oscillators can be coupled, so B B A 2 that EPR-steering correlations are induced between the This enables us to calculate correlations such as hXAi, two oscillators. hXAXBi after a time τ, assuming initial uncorrelated 3 vacuum states for all modes. We can then evaluate These LHS models are similar to the local hidden vari- the variance of XA − gxXB, which will be denoted able models introduced by Bell [9, 10]. Here λ represents 2 D 2E hidden variable parameters and P (λ) is the probability ∆ (XA − gxXB) and which equals (XA − gxXB) distribution for these parameters. The P (λ) is indepen- when we assume vacuum inputs at τ = 0. We have intro- dent of the choice of measurement (θ and φ), which is duced constants g , which can be chosen to minimise the x made after the generation and separation of the subsys- variance relative to the Heisenberg quantum noise level θ tems A and B. In this model, the hXAiλ is the average of the amplitudes XA and PA. We find this value of gx θ value for the result XA, given the hidden variable state by standard procedures [15, 16]: φ specified by λ. The hXBiλ is defined similarly. For the LHS model however, there is the additional asymmetrical ∂ 2 hXAXBi ∆ (XA − gxXB) = 0, ⇔ gx = 2 (4) constraint that the local hidden variable moments (such ∂gx hXBi θ as hXAiλ) for system A be consistent with measurements For the system described by the Hamiltonian of Eq. (1), of some local observables (for example position and mo- we find the optimal value is g = − tanh 2r. The optimal mentum) at A, and is thus able to be described as arising x λ variance is given by: from a local quantum density operator ρA. We can define the minimum value of EPR after opti- 2 2 2 hXAXBi mising gx and gp as: EPRA|B = min{EPRA|B(gx, gp)}. ∆ (XA − gxXB) = XA − 2 (5) Thus, for the Hamiltonian of Eq. (1): hXBi which for the parametric system (1) becomes 1 2 1 EPRA|B = (10) ∆ (XA − gxXB) = cosh 2r . Similarly we evaluate cosh 2r the variance of PA + gpPB , which we will denote as 2 as derived originally in [16]. Ideal EPR correlations are ∆ (PA + gpPB): The variance minimizes when obtained as r → ∞, in which case EPRA|B → 0, and we say the “steering is perfect”. ∂ 2 hPAPBi ∆ (PA + gpPB) = 0, ⇔ gp = − 2 ∂gp hPBi

This optimal value of gp for system (1) is gp = − tanh 2r. C. Reservoir coupling The optimal variance is given by

2 We now examine the eﬀect on EPR steering if the sys- 2 2 hPAPBi tems A and B are independently coupled to heat bath ∆ (PA + gpPB) = PA − 2 (6) hPBi reservoirs (Figure 1). We assume that the parametric

2 1 interaction H given by (1) that generates a two-mode which becomes ∆ (PA + gpPB) = cosh 2r for Eq. (1). squeezed state is turned off at the time t = 0, and the system left to decay. The two-mode squeezed state is a so-called Gaussian state, meaning that its characteristic B. EPR steering correlations function is Gaussian [73]. Within the constraint of two- mode Gaussian states and measurements, the condition The criterion for the EPR paradox introduced in [16] is (8) will provide a necessary and sufficient test of steering also a criterion for EPR steering [3–5]. This EPR steering [3, 4]. This makes the criterion valuable, for understand- criterion is defined as the square root of the variance ing the effects of decoherence in Gaussian systems. product: Thus, we consider a system prepared in a two-mode squeezed state at time t = 0. In principle, the modes a EPRA|B(gx, gp) = ∆ (XA − gxXB) ∆ (PA + gpPB) (7) and b can be spatially separated. We consider the coupling of each mode a and b to independent heat baths We observe EPR steering (of system A by B) whenever (reservoirs) with thermal occupation numbers na and nb respectively. The solutions after coupling to the reser- EPR (g , g ) < 1 (8) A|B x p voir are straightforward to evaluate using the operator The ideal correlations created by the parametric down Langevin equations that describe the evolution of the conversion Hamiltonian interaction (1) are obtained in mode operators [74, 75]: the limit of r → ∞ and correspond to the EPR variances a˙ = −γ a+p2γ Γ (t) becoming zero, and hence EPR → 0. a a a ˙ p The EPR steering condition (8) by its very definition b = −γbb+ 2γbΓb(t) (11) will negate all local hidden state (LHS) models of the type [3, 4] Here γa and γb describe the decay rates (losses) that are induced by the reservoirs. The quantum reser- θ φ ˆ θ φ θ φ voir operators Γ(t) have nonzero correlations given by hXAXBi ≡ hXAXBi = P (λ)hXAiλhXBiλdλ (9) † 0 0 † 0 ˆ hΓa(t)Γa(t )i = naδ(t − t ) and hΓa(t)Γa(t )i = (na + 4

0 1)δ(t − t ) where the numbers na and nb give the level of II.A. The ﬁnal moments after reservoir coupling are thermal occupation of the reservoirs. Solutions are † −2tγa −2tγa 2 a (t)a(t) = na 1 − e + e sinh (r) ha(t)b(t)i = −e−(γa+γb)t cosh r sinh r (14) t 0 −γat p −γa(t−t ) 0 0 a(t) = e a(0)+ 2γa e Γa (t ) dt (12) The covariance matrix V is deﬁned as: ˆ0  2  XA hXAPAi hXAXBi hXAPBi 2  hPAXAi PA hPAXBi hPAPBi  From this, we can calculate the moments at a later time, V =  2  (15)  hXBXAi hXBPAi XB hXBPBi  in terms of the initial moments: 2 hPBXAi hPBPAi hPBXBi PB

Hence we ﬁnd V14 = V23 = 0, V12 = V34 = i , V11 = V22, † −2γat † −2γat ha (t)a(t)i = e ha (0)a(0)i + na(1 − e ) V33 = V44, V24 = −V13 where −(γa+γb)t ha(t)b(t)i = e ha(0)b(0)i (13) −2γat −2γat V11 = e cosh 2r + 1 − e (1 + 2na)

−2γbt −2γbt V33 = e cosh 2r + 1 − e (1 + 2nb) † The solution for hb (t)b(t)i is obtained from that for −(γa+γb)t V24 = e sinh 2r (16) ha†(t)a(t)i, but exchanging the letters a with b. The initial moments are given by the solutions found in Section Applying the results of the Section II.A, we calculate that

−2γ t −2γ t −2γ t −2γ t −2(γ +γ )t −2γ t −2γ t cosh 2r e a 1 − e b (1 + 2nb) + e b 1 − e a (1 + 2na) + e a b + (1 + 2na) (1 + 2nb) 1 − e b 1 − e a EPR = A|B −2γ t −2γ t e b cosh 2r + (1 − e b ) (1 + 2nb) (17)

of loss to be asymmetrical with respect to the two sys- 1.5 1.5 tems A and B.

1 1 1. Damping the steering system: Steering sudden-death 0.5 0.5 Let us suppose that the reservoirs are coupled asym- 0 0 0 0.2 0.4 0.6 0.8 1 0 0.2 0.4 0.6 0.8 1 metrically, so that γa → 0 i.e. only the mode b is coupled to a reservoir. In this case, we find for the steering of A (mode a) Figure 2. The decoherence of two-mode EPR steering with e−2γbt + cosh 2r 1 − e−2γbt no thermal excitation (na = nb = 0). A sudden-death type of EPRA|B = −2γ t (18) decoherence is noted when it is the system “doing the steering” 1 + e b (cosh 2r − 1) that is is coupled to the reservoir. The steering signatures 0 The plots of Figure 2 (left graph) indicate that where EPRA|B and EPRB|A are plotted versus t = γbt, for various the losses are entirely on the steering mode B, the de- values of squeeze parameter r. EPR steering is obtained for coherence of EPR steering is substantial. After a time the Gaussian system iff EP R < 1 and the strongest steering is ln 2 when EPR → 0. Left graph: Reservoir coupling to system given by γbt = 2 , steering (as measured by EPRA|B) in that direction is lost. The loss is inevitable, regardless B only (γa = 0). Solid lines correspond to EPRA|B while dashed lines correspond to EPRB|A. Curves are (bottom to of how much steering is present in the initial two-mode top on the far left) r = 2 (black), r = 1 (red), r = 0.5 (blue). quantum system, as shown by the results for the differ- Right graph: Plots of EPRA|B for symmetric coupling to ent values of squeeze parameter r: We also note that the the reservoirs: γa = γb. In this case, EPRA|B = EPRB|A. cut-off time for steering is independent of the amount of steering r in the initial two-mode system. The behavior observed here is analogous to the “entanglement sudden- D. Decoherence of EPR steering with no thermal death”, that has been observed for the decoherence of noise entanglement [76–78].

We first assume negligible thermal noises: we put 2. Damping the steered system na = nb = 0 in the solution given by Eq. (17). The decoherence effect on the system by the reservoirs manifests as losses, parametrised by γa and γb. We will study However, the sudden-death effect is not apparent when the effect on the steering parameters, showing the effect the “steered” system is lossy. We again assume γa → 0 5 and evaluate the steering of the damped system, by the undamped system: 1.5

e−2γbt + cosh 2r 1 − e−2γbt EPR = (19) B|A cosh 2r 1 The plots of Figure 2 (left graph) reveal the nature of the 0.5 decoherence when only the steered system is damped. Here, we see that there is much less sensitivity of the steering to the losses. In fact, while the EPR steering is 0 certainly diminished by the dissipation, there is no cut- 0 0.1 0.2 off, or sudden death, but rather steering is still possible for arbitrary times, t → ∞. We remark that the in- creased sensitivity to losses affecting the steering system, Figure 3. The decoherence of two-mode EPR steering with as compared to the steered system, has been noted in a thermal reservoir coupled to system B only. Here γa = 0, 0 earlier papers [15, 55, 56, 59]. r = 1. We plot the steering signature versus t = γbt. The thermal noise creates a sudden-death type decoherence effect when it is the system that is “being steered” that is coupled to 3. Symmetric decoherence the reservoir. Solid lines correspond to EPRA|B while dashed lines correspond to EPRB|A. EPR-steering is obtained when We next consider a reservoir with symmetric damping EP R < 1 and the strongest steering is for EPR → 0. We plot different values of nb: from bottom to top (for each line γa = γb. type), nb = 1 (blue), nb = 5 (red), nb = 10 (black). 2 e−4γbt + 1 − e−2γbt EPRA|B = 1 + e−2γbt (cosh 2r − 1) 1.5 1.5 2 cosh 2r e−2γbt − e−4γbt + (20) 1 + e−2γbt (cosh 2r − 1) 1 1

The results of Figure 2 (right graph) show, as might be 0.5 0.5 expected, the eﬀect is dominated by the fact that the steering system is damped. Here steering (as measured 0 0 ln 2 0 0.1 0.2 0 0.1 0.2 by EPRA|B) is also lost at γbt = 2 .

E. Decoherence of EPR steering with thermal noise Figure 4. The decoherence of two-mode EPR steering with thermal reservoirs. EPR-steering is obtained when EP R < 1 and the strongest steering is for EPR → 0. Left graph: Now we consider the more complex interaction where Reservoirs coupled to both systems A and B symmetrically. there is additional thermal noise for each reservoir. Here 0 We plot EPRA|B vs t = γbt. Here γa = γb, na = nb, r = we use the full expressions given by Eq. (17). Our re- 1. We plot different values of nb, given from bottom to top sults in Figures 3 and 4 reveal several features. We note (for each line type) , nb = 1 (blue), nb = 5 (red), nb = 10 that the loss of steering is rapid and complete (“sudden- (black). Solid lines correspond to r = 1 while dashed-dotted death”) if the thermal noise is placed on the system that lines correspond to r = 2. Right graph: Thermal reservoir is being steered. This effect has been noticed in previous coupling to system A and cold reservoir coupling to B (γa = studies of thermal steering and is especially important for γb, na = 1, 5, 10). Solid lines correspond to EPRA|B while asymmetrical systems such as found in opto-mechanics dashed lines correspond to EPRB|A. [21, 31, 34, 35]. The effect is more significant as the thermal noise increases, but is reduced for higher r (Figure 4). the exchange A ←→ B. The relationships between entanglement and EPR steering for Gaussian states have been recently studied [79, 80]. For Gaussian states, it is F. Comparing with the decoherence of possible to give a quantification of steering [3, 4, 79, 80], entanglement which is an asymmetrical quantum correlation, similar to quantum discord [81–83]. The effects of the reservoir on the steering are asym- We may evaluate the entanglement via the parameter 2 metrical. This will not be true for entanglement: En- Ent = ∆(XA − gXB)∆(PA + gPB)/{1 + g } [84, 85]. tanglement is symmetrically defined, with respect to the The inequality Ent < 1 is a necessary and sufficient con- two different systems involved, and the numerical value dition for entanglement of two-mode Gaussian systems we assign to the entanglement will be unchanged under for the case we examine here where ∆ (XA − gXB) = 6

1 1 2 2 We note that if XA = XB , then the value of g that 0.75 0.75 minimises the expression is g = 1, but where we have asymmetric reservoir effects, like different reservoir cou- 0.5 0.5 plings or thermal noises, the optimal value will be differ- 0.25 ent to 1. Here we will use from Section II.A that: 0.25 0 X2 = e−2γat cosh 2r + 1 − e−2γat (1 + 2n ) 0 0 0.1 0.2 A a 0 0.1 0.2 −(γa+γb)t hXAXBi = −e sinh 2r 2 −2γbt −2γbt XB = e cosh 2r + 1 − e (1 + 2nb) (24) Figure 5. The decoherence of the entanglement of the two- mode squeezed state under the action of a reservoir. The Figure 5 plots the entanglement Ent for various reser- entanglement signature for Gaussian systems Ent is plotted voir couplings. We see from Figure 5 (left graph) that versus scaled time t0, under different sorts of reservoir coupling. Entanglement is obtained iff Ent < 1 and stronger where there is no thermal noise, the entanglement de- entanglement corresponds to Ent → 0. Left graph: Here cays steadily, but is never destroyed completely. This we take all thermal noise zero (na = nb = 0). Solid curves effect was noted in [86, 87]. The larger values of r corre- show the symmetric case γa = γb. Dashed-dotted curves are spond to larger amounts of entanglement for the initial for the asymmetric case, for a reservoir coupling to system B state before decoherence, and we note that while the de- 0 only (γa = 0) when t = γbt; or, for a reservoir coupling to cay is sharper for higher r, a larger initial amount of EPR 0 system A only (γb = 0) when t = γat. We plot different val- entanglement will ensure larger EPR entanglement for all ues of r, given from bottom to top (for each line type): r = 2 later times. The Figure 5 (right graph) shows that when (black), r = 1 (red), r = 0.5 (blue). Right graph: Here we the reservoirs are thermally excited, the entanglement is study thermal reservoirs. Solid curves show the symmetric totally destroyed (in sudden-death fashion) after a finite case n = n and γ = γ . Dashed-dotted curves are for the a b a b time. This time is shortened, by increasing the amount asymmetric case, for a reservoir coupling to system B only 0 of thermal excitation of the reservoir (on either system). (γa = 0) when t = γbt. We plot different values of thermal noise, given from bottom to top (for each line type) nb = 1 (blue), nb = 5 (red), nb = 10 (black). III. DECOHERENCE OF THE STEERABILITY OF A S-CAT STATE

∆ (PA + gPB). The condition for optimally chosen g has been shown equivalent to the Peres-Simon condition in Traditionally, most studies of decoherence have cen- that case [79]. Thus, in this section, we consider the tred around the Schrodinger cat [45–49]. We noticed entanglement parameter in the study of the decoherence of steering for the two- mode squeezed state that for stronger EPR effects, the ∆(X − gX )∆(P + gP ) decay was brought about more sharply, to give overall Ent = A B A B (21) 1 + g2 decoherence times of a similar order (Figure 2). This is consistent with the overall intuition about decoherence First we find the value of g that minimizes the param- in quantum mechanics, that it acts to destroy the more ∂Ent eter Ent. We evaluate ∂g , in this case we know that extreme (larger) effects more quickly so that they are ∆ (XA − gXB) = ∆ (PA + gPB) hence: not observed in nature. This has been studied for the Schrodinger cat example, where one considers a system ∂Ent ∂ ∆2 (X − gX ) initially prepared in superposition of two states macro- = A B ∂g ∂g 1 + g2 scopically distinguishable e.g. in phase space [43–49]. 2 2 2 The coupling to a reservoir induces a decay of the super- −2 hXAXBi + 2g XB − XA + 2g hXAXBi = 2 position. Where the separation in phase space increases, (1 + g2) calculation and experiments show that the decay rate (22) will increase. Thus, there is an explanation of the transition from microscopic to macroscopic quantum mechan- ∂Ent Next ∂g = 0, ics. Here, motivated by this, we examine the decoherence of the EPR steering of a Schrodinger cat state. 2 2 2 ⇔ − hXAXBi + g XB − XA + g hXAXBi = 0

Hence the value of g that minimizes Ent is [79]: A. Steering signature for a Schrodinger cat q 2 2 2 2 2 2 − XB − XA + (hXBi − hXAi) + 4 hXAXBi Consider the state: g = 2 hX X i A B 1 iθ |ψi = √ {| − αiA| ↑iB + e |αiA| ↓iB} (25) (23) 2 7

0.4 0.8

0.3 0.6

0.2 0.4

0.1 0.2

0 0 −12 −8 −4 0 4 8 12 −4−3 −2 −10 1 2 3 4

Figure 6. The conditional distributions associated with an EPR steering signature for the Schrodinger cat. Left Graph: Conditional probability for XA given that the result of a mea- Figure 7. The steering of a Schrodinger cat. We have pro- surement σz is +1 (red solid curve is for α = 2, green dash- posed a signature for the cat, based on EPR steering param- dotted curve is for α = 3) or −1 (blue long dashed curve is eter EPRA|B . Here, measurements made on the microscopic for α = 2, black short dashed curve is for α = 3). Right spin system B “steer” the macroscopic system (the cat) A. Graph: Conditional probability for PA given that the result The “cat” in this case is modelled as the superposition in- of the measurement σx is +1 (red solid curve) or −1 (blue volving two coherent states, |αi and | − αi. The value of α long dashed curve). Here α = 2. determines the size of the Schrodinger cat. We consider that the “cat” being macroscopic is coupled to a thermal reservoir. We examine the decay of the steering signature, with respect which is the entangled Schrodinger cat state. We select to the parameters of the reservoir coupling. θ = π/2 and α to be real. Here, |αi is the coherent state for mode a of system A and | ↑i, | ↓i are the eigenstates spin σx. of the Pauli spin σz of system B. This state is similar to that described in Schrodinger’s original gedanken exper- 1 iment [39], where a microscopic system becomes entan- |ψi = (| − αiA + i|αiA)| ↑ix,B 2 gled with a macroscopic one, the “cat”. The spin-mode 1 entangled state given by Eq. (25) has been the subject + (i|αi − | − αi )| ↓i (27) 2 A A x,B of several experiments [43]. We first explain how one can signify the Schrodinger cat, using EPR steering. At this point, we note the for other choices of θ, normal- We will derive the probability distributions P (XA) (or isation factors are more complicated. We find that the P (PA)) for the system A, given that a measurement of probability for the momentum given the results for spin the Pauli spin σz (or σx) at B has been carried out to σx is give a result +1 or −1. We define the scaled amplitudes 1 1 2 2 for position and momentum by a = (ˆx + ipˆ) where a 2 1 −2 p pα c c2 P p σx = ±1 = e 1 ± sin 4 is the boson operator,√ and note that for real position π c c and momentum, c = 2. In terms of the quadrature (28) phase amplitudes XA/B and PA/B defined in Section II, however, we chose the scaling c = 2. We will distinguish This allows us to calculate the conditional variances the two cases by using lower and upper case, respectively, that give us a signature for EPR steering. An EPR steer- and note we have dropped the use of the operator hats ing of the “cat” is observed when where the meaning is clear, or to denote the outcomes of the measurements. As expected from direct examination EPR = V ar(XA σz)V ar(PA σx) < 1 (29) of the state (25), the distributions P (x +1) and P (x −1) for the result x at A given the outcomes +1 or −1 for σz where here at B, are the two Gaussian hills (Figure 6, left graph):

Speciﬁcally, V ar(XA σz) = Pz(+1)V ar(XA +1)+Pz(−1)V ar(XA −1) 1 2 2 1 x2 xα P x σ = ±1 = exp −2 − 2α2 ∓ 4 denotes the conditional variance for XA averaged over the z 2 π c c c outcomes of σz. We have used the notation V ar(XA ±1) (26)

to mean the variance of P (XA ±1), which is the probability distribution for XA given the outcome ±1 for the Next, we derive the conditional distributions for P (PA) measurement σz, respectively. The V ar (PA|σx) is de- given that a measurement of Pauli spin σx at B yields ﬁned similarly, but for outcomes of σx. The right-side the result +1 or −1. To evaluate this, we rewrite the bound is the quantum noise limit, as determined by the state in terms of the basis states | ↑ix and | ↓ix for the Heisenberg uncertainty relation which in this case for the 8 choice c = 2 is ∆XB∆PB ≥ 1. We note that the deriva- tion of the inequality as a signature of EPR steering has 1.5 been given in the Refs. [5, 15, 16]. The EPR quantity (29) can be evaluated: We see that 1.25 for the two Gaussian hills, the variance is at the quan-

1 tum noise level (Figure 6, left graph): V ar(XA σz) = 1. On the other hand, the variance associated with the mo- 0.75 mentum distributions is reduced below 1 (Figure 6, right 0.5 graph), implying that EPRA|B < 1. We can calculate the variance from the conditional distributions. Alterna- 00.050.1 0.15 0.2 tively, noting that the collapsed state for system B given 1 a measurement σx is the superposition 2 (|−αiA +i|αiA) (for result +1) or 1 (i|αi − | − αi ) (for result −1), it is 2 A A Figure 8. The decoherence of the Schrodinger cat EPR- easy to use the methods and results of the next section, steering signature EPR for the system depicted in Fig. 7. to find that Here, a thermal reservoir is coupled to the spin system B. We 0 2 plot the steering signature versus t = γt. Here we consider 2 −4|α| V ar(PA σx) = 1 − 4α e (30) different sizes of the “cat” (solid line lower to top, α = 0.5, 1 and 1.5). For each α, we plot different values of n: n = 0 The steering inequality (29) has been suggested in (solid), n = 1 (dashed-dotted). Ref. [54], as a way to realise an EPR paradox with the Schrodinger cat state. The steering cannot be obtained if the system is in the mixture of states, | − αiA| ↑iB and We see that the variance for P can be reduced below the |αiA| ↓iB, that allows a classical dead or alive descrip- quantum limit (given by 1). The reduction of the vari- tion. This case is especially interesting, because it focuses ance of P below the quantum limit is in itself a signature on the steering of a mesoscopic system (the Schrodinger for the “cat” state. We see this as follows. If we take cat) [34]. If we assume Local Realism is valid, then it is α large, and assume no thermal noise and γ = 0, then the local state of the mesoscopic system that is shown the probability distribution P (x) associated with each of to be inconsistent with the quantum mechanics (refer the “dead” and “alive” states is a Gaussian hill, with vari- to the LHS model of the Section II.B). This contrasts ance 1. If the system were to actually be in any kind of with EPR’s original argument, which showed the incon- mixture of these quantum states, then the variance for P sistency for a microscopic system [7]. could not drop below 1 because of the uncertainty relation (and the fact that mixing states cannot decrease the variance) [54, 88]. The observation of a reduced variance B. Decoherence of the Schrodinger cat with a heat for P can thus occur for a superposition but not for a bath mixture of the two quantum states that possess a distribution P (x) given by the Gaussian hills. The details are It will prove useful to next study the interaction of the not studied further here however, since our objective is single mode “Schrodinger cat” state [45–48] to study the decoherence of steering.

1 |ψi = √ {| − αi + i|αi} (31) 2 C. Decoherence of the steering of the Schrodinger cat with a reservoir. This was analysed by many authors, including Yurke and Stoler [45–49]. We consider that Motivated by this, we now examine the decoherence of the single mode, prepared in the Schrodinger cat state, the EPR steering of a Schrodinger cat state, as given by is then coupled to a thermal heat bath with dissipation. the signature Eq. (29). We consider the two-mode case Using the solutions from Section II.C, we can calculate where A is a harmonic oscillator coupled to a heat bath the moments at a later time t, in terms of the initial from time t = 0 and the second system is the spin B (no moments: heat bath coupling), as depicted in Figure 7. The system is prepared in the entangled Schrodinger cat state a†(t)a(t) = e−2γt a†(0)a(0) + n 1 − e−2γat 1 |ψi = √ {| − αiA| ↑iB + i|αiA| ↓iB} (33) ha(t)a(t)i = e−2γt ha(0)a(0)i 2

Denoting the variances using shorthand notation, by We can rewrite this state in terms of the basis states | ↑ix 2 2 2 2 ∆ P = (∆P ) and ∆ X = (∆X) , we ﬁnd and | ↓ix for the spin σx as shown in Section III.A. It is straightforward to calculate the moments haσ i, ha2σ i, 2 z z ∆2P = 1 + 2n 1 − e−2γt − 4α2e−2γte−4α (32) etc. at the time t = 0. The moments at a later time t 9

(after the interaction with the heat bath has been turned We find that with n = 0 (no thermal noise), the value on) can be evaluated using the solution (12) of Section of EPR is identical to that of V ar(PA σx), which is also II.C. We note that since the reservoir interaction does not identical to the variance ∆2P given by Eq. (32). This involve the spins, we have σz(t) = σz(0), σx(t) = σx(0). Therefore, we can calculate the moments at the time t, value is plotted in Figure 8 (solid lines). By the above in terms of the moments at the initial time t = 0. argument, the signature of the S-cat is the drop below 1 We next consider the moments of the distributions of EPR. We note that as α increases, the EPR value tends to 1, the signature therefore becoming more sen- at the time t for XA and PA, conditional on getting the result either +1 or −1 for the Pauli spin measure- sitive to decoherence. This reduction in variance of P ment on B. These moments can be readily calculated. as α increases is directly associated with the interference We outline the method. We use the notation, for ex- fringes in the distribution P (PA) [45–48]. These fringes become finer as α increases, as evident from the function ample, hXA 1iz to denote the moment hXAi of system given by Eq. (28). With thermal noise, we see that the A, conditioned on the result +1 for spin σz at B. Let value of EPR increases: a much greater sensitivity to us suppose then that we obtain the outcomes XA at decoherence is apparent, as illustrated in Figure 8 by the A, and σz at B. We define the measurable probabil- dashed-dotted curves. ity P (XA, σz) for the joint outcomes. We see that:

P (XA,σz ) P (XA σz) = . It follows that P (XA, ±1) = Pz (σz )

P (X ±1)P (±1) = 1 P (X ±1). Hence, hX σ i = IV. CONCLUSION A z 2 A A z

P (XA, 1)XA − P (XA, −1)XA = P (XA 1)Pz(1)XA − The directional properties of EPR steering are signifi- 1 cant when it comes to understanding the decoherence of P (XA −1)Pz(−1)XA = {hXA 1i − hXA −1i}. Follow- 2 EPR-steering. We have shown how when two systems ing this procedure, we see that: are coupled independently to a reservoir, the steering is 1 asymmetrically affected. If we consider the steering of hX σ i = {hX 1i − hX −1i } A z 2 A z A z a system A by measurements made on the system B, 1 then the steering is sensitive to the reservoir coupling to hX i = {hX 1i + hX −1i } (34) A 2 A z A z system B. This is intuitively not surprising, given the Local Hidden State (LHS) definition of EPR steering as 2 and similarly for the moments involving XA. This a nonlocality. In terms of the LHS description that is allows us to solve for the conditional moments using to be negated in order to confirm such steering by B [3, 4], it is the hidden states of system B that are like the relations such as hXA 1iz = hXAσzi + hXAi and those considered by Bell. The sensitivity of the steering hXA −1iz = −hXAσzi + hXAi. We also define the to the losses on the system B has been studied experi- same relations, but replacing XA with PA and σz with mentally, in the context of the EPR paradox [54, 86, 87]. σ . The final solutions for the conditional moments are: There, it was known that the EPR criterion could not be x −γt 2 −2γt achieved, with 50% or more losses on the steering sys- hXA ±1iz = ∓2αe , hXA ±1iz = 1 + 2n 1 − e + 2 tem A. Recent work considers this result in terms of 2 −2γt −γt −2α 2 4α e , hPA ±1ix = ∓2αe e and hPA ±1ix = the monogamy properties of EPR steering [59]. The sen- 1 + 2n 1 − e−2γt. Hence we solve for the average vari- sitivity of the steering signatures to losses (which may represent an eavesdropper on the channel) has poten- ance V ar(XA σz) of X conditioned on the measurement tially important implications for quantum cryptography of spin σz at B: We obtain [64, 68]. The behaviour with respect to the thermal noise ap- −2γt V ar(XA σz) = 1 + 2n 1 − e (35) pears almost reversed. This has implications for detecting steering of Schrodinger cats which are coupled

Similarly, we can solve for variance V ar(P σ ) of P to hot reservoirs. For the Schrodinger cat example, we A x A explained how the steering acts as a signature of the conditioned on spin σx at B, to obtain Schrodinger cat. Adding thermal noise to the system B

2 significantly affects the amount of EPR steering of sys- −2γt 2 −2γt −4α V ar(PA σx) = 1 + 2n 1 − e − 4α e e (36) tem A. There have been recent theoretical studies of the steering of a mechanical oscillator, which we liken to a To show how the EPR steering signature decoheres with “cat”, by measurements made on an optical pulse [34, 35]. time, we evaluate In those studies, the thermal noise of the oscillator was shown to have a quite dramatic effect on the amount

EPR = V ar(XA σz)V ar(PA σx) (37) of steering possible. By comparison, the entanglement between the thermal oscillator and the pulse was quite Here, we use the notation deﬁned in Section III.A. robust. Similar results have been obtained for Bose Ein- 10 stein condensates [31]. These results are consistent with Dalton for many helpful suggestions. the simple descriptions of decoherence given in this paper.

ACKNOWLEDGMENTS

This work was supported by an Australian Research Council Discovery Project Grant. We are grateful to B.

[1] E. Schroedinger, “Discussion of probability relations be- alcanti, P. K. Lam, H. A. Bachor, U. L. Andersen, and tween separated systems,” Proc. Cambridge Philos. Soc. G. Leuchs, “Colloquium: The Einstein-Podolsky-Rosen 31, 555-563 (1935). paradox: From concepts to applications,” Rev. Mod. [2] E. Schroedinger, “Probability relations between sepa- Phys. 81, 1727-1751 (2009). rated systems ,” Proc. Cambridge Philos. Soc. 32, 446- [16] M. D. Reid, “Demonstration of the Einstein-Podolsky- 452 (1936). Rosen paradox using nondegenerate parametric amplifi- [3] H. M. Wiseman, S. J. Jones, and A. C. Doherty, cation,” Phys. Rev. A 40, 913-923 (1989). “Steering, entanglement, nonlocality, and the Einstein- [17] M. D. Reid, and P. D. Drummond, “Quantum correla- Podolsky-Rosen paradox,” Phys. Rev. Lett. 98, 140402 tions of phase in nondegenerate parametric oscillation,” (2007). Phys. Rev. Lett. 60, 2731-2733 (1988). [4] S. J. Jones, H. M. Wiseman, and A. C. Doherty, “Entan- [18] K. Dechoum, S. Chaturvedi, P. Drummond, and M. glement, Einstein-Podolsky-Rosen correlations, Bell non- D. Reid, “Critical fluctuations and entanglement in the locality, and steering,” Phys. Rev. A 76, 052116 (2007). nondegenerate parametric oscillator,” Phys. Rev. A70, [5] E. G. Cavalcanti, S. J. Jones, H. M. Wiseman, and 053807 (2004). M. D. Reid, “Experimental criteria for steering and the [19] B. Julsgaard, A. Kozhekin, and E. S. Polzik, “Experimen- Einstein-Podolsky-Rosen paradox,” Phys. Rev. A 80, tal long-lived entanglement of two macroscopic objects,” 032112 (2009). Nature 413, 400-403 (2001). [6] D. J. Saunders, S. J. Jones, H. M. Wiseman, and G. [20] H. Krauter, C. A. Muschik, K. Jensen, W. Wasilewski, J. J. Pryde, “Experimental EPR-steering using Bell-local M. Petersen, J. I. Cirac, and E. S. Polzik, “Entanglement states,” Nature Physics 6, 845-849 (2010). generated by dissipation and steady state entanglement [7] A. Einstein, B. Podolsky and N. Rosen, “Can quantum- of two macroscopic objects,” Phys. Rev. Lett. 107 080503 mechanical description of physical reality be considered (2011). complete?,” Phys. Rev. 47, 777-780 (1935). [21] Q. He, and M. Reid, “Towards an Ein- [8] J. S. Bell, “Speakable and unspeakable in quantum stein–Podolsky–Rosen paradox between two macroscopic mechanics” (Cambridge University Press, Cambridge, atomic ensembles at room temperature,” New J. Phys., 1987). 15, 063027 (2013). [9] J. S. Bell, “On the Einstein-Podolsky-Rosen paradox,” [22] C. Gross, H. Strobel, E. Nicklas, T. Zibold, N. Bar- Physics 1, 195-200 (1964). Gill, G. Kurizki, and M. K. Oberthaler, “Atomic ho- [10] J. F. Clauser, M. A. Horne, A. Shimony, and R. A. Holt, modyne detection of continuous-variable entangled twin- “Proposed experiment to test local hidden-variable theo- atom states,” Nature 480, 219-233 (2011). ries,” Phys. Rev. Lett. 23, 880-884 (1969). [23] N. Bar-Gill, C. Gross, I. Mazets, M. Oberthaler, and G. [11] M. Giustina, A. Mech, S. Ramelow, B. Wittmann, J. Kurizki, “Einstein-Podolsky-Rosen correlations of ultra- Kofler, J. Beyer, A. Lita, B. Calkins, T. Gerrits, S. W. cold atomic gases,” Phy. Rev. Lett, 106, 120404 (2011). Nam, R. Ursin, and A. Zeilinger, “Bell violation using en- [24] K. Kheruntsyan, M. Olsen, and P. D. Drummond, tangled photons without the fair-sampling assumption,” “Einstein-Podolsky-Rosen correlations via dissociation of Nature 497, 227-230 (2013). a molecular Bose-Einstein condensate,” Phys. Rev. Lett. [12] B. G. Christensen, K. T. McCusker, J. B. Altepeter, B. 95, 150405 (2005). Calkins, T. Gerrits, A. E. Lita, A. Miller, L. K. Shalm, [25] M Ögren, and K. V. Kheruntsyan, “Role of spatial inho- Y. Zhang, S. W. Nam, N. Brunner, C. C. W. Lim, N. mogeneity in dissociation of trapped molecular conden- Gisin, and P. G. Kwiat, “Detection-loophole-free test of sates,” Phys. Rev. A 82, 013641 (2010). quantum nonlocality, and applications,” Phys. Rev. Lett. [26] J. Jaskula, M. Bonneau, G. B. Partridge, V. Krachmalni- 111, 130406 (2013). coff, P. Deuar, K. V. Kheruntsyan, A. Aspect, D. Boiron, [13] C. S. Wu, and I. Shaknov, “The angular correlation of and C. I. Westbrook, “Sub-Poissonian number differences scattered annihilation radiation,” Physical Review A 77, in four-wave mixing of matter waves,” Phys Rev. Lett. 136 (1950). 105, 190402 (2010). [14] Z. Y. Ou, S. F. Pereira, H. J. Kimble, and K. C. Peng, [27] A. Ferris, M. Olsen, E. Cavalcanti, and M. Davis, “Detec- “Realization of the Einstein-Podolsky-Rosen paradox for tion of continuous variable entanglement without coher- continuous variables,” Phys. Rev. Lett. 68, 3663-3666 ent local oscillators,” Phys. Rev. A 78, 060104 (2008). (1992). [28] Q. He, M. Reid, T. Vaughan, C. Gross, M. Oberthaler, [15] M. D. Reid, P. D. Drummond, W. P. Bowen, E. G. Cav- and P. Drummond, “Einstein-Podolsky-Rosen entangle- 11

ment strategies in two-well Bose-Einstein condensates,” Wineland, “A “Schrödinger Cat” superposition state of Phys. Rev. Lett. 106, 120405 (2011). an atom,” Science 272, 1131-1136 (1996). [29] Q. He, P. Drummond, M. Olsen, and M. D. Reid, [45] B. Yurke, and D. Stoler, “Generating quantum me- “Einstein-Podolsky-Rosen entanglement and steering in chanical superpositions of macroscopically distinguish- two-well Bose-Einstein-condensate ground states,” Phys. able states via amplitude dispersion,” Phys. Rev.Lett. Rev. A 86, 023626 (2012) 57, 13-16 (1986). [30] B. Opanchuk, Q. He, M. Reid, and P. Drummond, “Dy- [46] A. Gilchrist and W. J. Munro, “Signatures of the pair- namical preparation of Einstein-Podolsky-Rosen entan- coherent state,”J. Opt. B 2 47-52 (2000). glement in two-well Bose-Einstein condensates,” Phys. [47] G. Milburn, and C. Holmes, “Dissipative quantum and Rev. A 86, 023625 (2012). classical Liouville mechanics of the anharmonic oscilla- [31] R. J. Lewis-Swan, and K. V. Kheruntsyan, “Sensitivity tor,” Phys. Rev. Lett. 56, 2237-2240 (1986). to thermal noise of atomic Einstein-Podolsky-Rosen en- [48] A. Dragan, and K. Banaszek, “Homodyne Bell’s inequali- tanglement,” Phys. Rev. A 87, 063635 (2013). ties for entangled mesoscopic superpositions,” Phys. Rev. [32] S. G. Hofer, W. Wieczorek, M. Aspelmeyer, and K. A 63, 062102 (2001). Hammerer, “Quantum entanglement and teleportation in [49] A. Caldeira, and A. J. Leggett, “Inﬂuence of dissipation pulsed cavity optomechanics,” Phys. Rev. A 84, 052327 on quantum tunneling in macroscopic systems,” Phys. (2011). Rev. Lett., 46, 211-214, (1981). [33] V. Giovannetti, S. Mancini, and P. Tombesi, “Radiation [50] S. Armstrong, M. Wang, R. Y. Teh, Q. Gong, Q. He, pressure induced Einstein-Podolsky-Rosen paradox,” Eu- J. Janousek, Hans-Albert Bachor, M. D. Reid, and P. rophys. Lett. 54, 559-565 (2001). K. Lam, “Multipartite Einstein–Podolsky–Rosen steer- [34] Q. He, and M. Reid, “Einstein-Podolsky-Rosen paradox ing and genuine tripartite entanglement with optical net- and quantum steering in pulsed optomechanics,” Phys. works,” Nat. Phys. 11, 167-172 (2015). Rev.A 88, 052121 (2013). [51] Q. Y. He, and M. D. Reid, “Genuine multipartite [35] S. Kiesewetter, Q. Y. He, P. D. Drummond, and M. D. Einstein-Podolsky-Rosen steering,” Phys Rev Lett. 111, Reid, “Scalable quantum simulation of pulsed entangle- 250403 (2013). ment and Einstein-Podolsky-Rosen steering in optome- [52] D. Cavalcanti, P. Skrzypczyk, G. H. Aguilar, R. V. Nery, chanics,” Phys. Rev. A 90, 043805 (2014). P. H. Souto Ribeiro, and S. P. Walborn, “Detecting [36] B. Wittmann, S. Ramelow, F. Steinlechner, N. K. Lang- multipartite entanglement with untrusted measurements ford, N. Brunner, H. M. Wiseman, R. Ursin, and A. in asymmetric quantum networks,” arXiv:1412.7730v2 Zeilinger, “Loophole-free Einstein–Podolsky–Rosen ex- (2014). periment via quantum steering,” New J. Phys. 14, 053030 [53] C. M. Li, K. Chen, Y. Chen, Q. Zhang, Y. Chen, J. W. (2012). Pan, “Genuine high-order Einstein-Podolsky-Rosen steer- [37] D. Smith, G. Gillett, M. P. de Almeida, C. Bran- ing,” arXiv:1501.01452 (2015). ciard, A. Fedrizzi, T. J. Weinhold, A. Lita, B. Calkins, [54] E. G. Cavalcanti, and M. D. Reid, “Criteria for gener- T. Gerrits, H. M. Wiseman, S. W. Nam, and A. G. alized macroscopic and mesoscopic quantum coherence,” White, “Conclusive quantum steering with superconduct- Phys. Rev. A 77, 062108 (2008). ing transition-edge sensors,” Nature Communications 3, [55] S. L. W. Midgley, A. J. Ferris, and M. K. Olsen, “Asym- 625-630 (2012). metric Gaussian steering: When Alice and Bob disagree,” [38] A. J. Bennett, D. A. Evans, D. J. Saunders, C. Bran- Phys. Rev. A 81, 022101 (2010). ciard, E. G. Cavalcanti, H. M. Wiseman, and G. J. Pryde, [56] V. Händchen, T. Eberle , S. Steinlechner, A. Samblowski “Arbitrarily loss-tolerant Einstein-Podolsky-Rosen steer- , T. Franz , R. F. Werner, and R. Schnabel, “Observa- ing allowing a demonstration over 1 km of optical ﬁber tion of one-way Einstein-Podolsky-Rosen steering,” Na- with no detection loophole,” Physical Review X 2, 031003 ture Photonics 6, 598-601 (2012). (2012). [57] S. P. Walborn, A. Salles, R. M. Gomes, F. Toscano, [39] E. Schroedinger, “Die gegenwärtige situation in der quan- and P. H. Souto Ribeiro, “Revealing hidden Einstein- tenmechanik,” Naturwiss. 23, 807-812 (1935). Podolsky-Rosen nonlocality,” Phys. Rev. Lett. 106, [40] D. Leibfried, E. Knill, S. Seidelin, J. Britton, R. B. 130402 (2011). Blakestad, J. Chiaverini, D. B. Hume, W. M. Itano, [58] J. Schneeloch, C. J. Broadbent, S. P. Walborn, E. G. J. D. Jost, C. Langer, R. Ozeri, R. Reichle, and D. Cavalcanti, and J. C. Howell, “Einstein-Podolsky-Rosen J. Wineland, “Creation of a six-atom ’Schrödinger cat’ steering inequalities from entropic uncertainty relations,” state,” Nature, 438, 639-642 (2005). Phys. Rev. A 87, 062103 (2013). [41] T. Monz, P. Schindler, J. T. Barreiro, M. Chwalla, D. [59] M. D. Reid, “Monogamy inequalities for the Einstein- Nigg, W. A. Coish, M. Harlander, W. Hänsel, M. Hen- Podolsky-Rosen paradox and quantum steering,” Phys. nrich, and R. Blatt, “14-qubit entanglement: Creation Rev. A 88, 062108 (2013). and coherence,” Phys Rev Lett 106, 130506 (2011). [60] J. Bowles, T. Vértesi, M. T. Quintino, and N. Brunner, [42] A. Ourjoumtsev, R. Tualle-Brouri, J. Laurat, and P. “One-way Einstein-Podolsky-Rosen steering,” Phys. Rev. Grangier, “Generating optical Schrödinger kittens for Lett. 112, 200402 (2014). quantum information processing,” Science 312, 83-86, [61] T. C. Ralph, “Continuous variable quantum cryptogra- (2006). phy,” Phys. Rev. A 61, 010303(R) (1999). [43] M. Brune, E. Hagley, J. Dreyer, X. Maître, A. Maali, C. [62] M. D. Reid, “Quantum cryptography with a predeter- Wunderlich, J. M. Raimond, and S. Haroche, “Observing mined key, using continuous-variable Einstein-Podolsky- the progressive decoherence of the “meter” in a quantum Rosen correlations,” Phys. Rev. A 62, 062308 (2000). measurement,” Phys. Rev. Lett. 77, 4887-4890 (1996). [63] M. D. Reid, in “Quanutm Squeezing” editors P. D. Drum- [44] C. Monroe, D. M. Meekhof, B. E. King, and D. J. mond, and Z. Ficek (Springer, 2004), p. 337. 12

[64] C. Branciard, E. G. Cavalcanti, S. P. Walborn, V. amplification,” Phys. Rev. A 30, 1386-1391 (1984). Scarani, and H. M. Wiseman, “One-sided device- [76] T. Yu, and J. H. Eberly, “Finite-time disentanglement independent quantum key distribution: Security, feasi- via spontaneous emission,” Phys. Rev. Lett. 93, 140404 bility, and the connection with steering,” Phys. Rev. A (2004) 85, 010301(R) (2012). [77] M. P. Almeida, F. de Melo, M. Hor-Meyll, A. Salles, S. [65] F. Grosshans, and P. Grangier, “Quantum cloning and P. Walborn, P. H. Souto Ribeiro, and L. Davidovich, teleportation criteria for continuous quantum variables,” “Environment-induced sudden death of entanglement,” Phys. Rev. A 64, 010301(R) (2001). Science 316, 579-582 (2007). [66] X. Ma, and N. Lutkenhaus, “Improved data post- [78] S. Chan, M. D. Reid, and Z. Ficek, “Entanglement evolu- processing in quantum key distribution and application tion of two remote and non-identical Jaynes–Cummings to loss thresholds in device independent QKD,” Quantum atoms,” J. Phys. B: At. Mol. Opt. Phys. 42, 065507 Inf. Comput. 12, 203-214 (2012). (2009). [67] M. D. Reid, “Signifying quantum benchmarks for qubit [79] Q. Y. He, Q. H. Gong, and M. D. Reid, “Classifying direc- teleportation and secure quantum communication us- tional Gaussian entanglement, Einstein-Podolsky-Rosen ing Einstein-Podolsky-Rosen steering inequalities,” Phys. steering, and discord,” Phys. Rev. Lett. 114, 060402 Rev. A 88, 062338 (2013). (2015). [68] B. Opanchuk, L. Arnaud, and M. D. Reid, “Detecting [80] I. Kogias, A. R. Lee, S. Ragy, and G. Adesso, “Quantifi- faked continuous-variable entanglement using one-sided cation of Gaussian quantum steering,” Phys. Rev. Lett. device-independent entanglement witnesses,” Phys. Rev. 114, 060403 (2015). A 89, 062101 (2014). [81] P. Giorda and M. G. A. Paris, “Gaussian Quantum Dis- [69] B. L. Schumaker, and C. M. Caves, “New formalism for cord”, Phys. Rev. Lett. 105, 020503 (2010). two-photon quantum optics. II. Mathematical founda- [82] G. Adesso and A. Datta, “Quantum versus Classical tion and compact notation,” Phys. Rev. A 31, 3093-3111 Correlations in Gaussian states”, Phys. Rev. Lett. 105, (1985). 030501 (2010). [70] A. Furusawa, J. L. Sørensen, S. L. Braunstein, C. A. [83] S. Pirandola, G. Spedalieri, S. L. Braunstein, N. J. Cerf, Fuchs, H. J. Kimble, and E. S. Polzik, “Unconditional and S. Lloyd, "Optimality of Gaussian Discord", Phys. quantum teleportation,” Science 282, 706-709 (1998). Rev. Lett. 113, 140405 (2014). [71] A. Heidmann, R. J. Horowicz, S. Reynaud, E. Giacobino, [84] L. M. Duan, G. Giedke, J. I. Cirac, and P. Zoller, “Insep- C. Fabre, and G. Camy, “Observation of quantum noise arability criterion for continuous variable systems,” Phys. reduction on twin laser beams,” Phys Rev. Lett. 59, 2555- Rev. Lett. 84, 2722-2725 (2000). 2557 (1987). [85] V. Giovannetti, S. Mancini, D. Vitali, and P. Tombesi, [72] A. S. Lane, M. D. Reid, and D. F. Walls, “Absorp- “Characterizing the entanglement of bipartite quantum tion spectroscopy beyond the shot-noise limit,” Phys Rev systems,” Phys. Rev. A 67, 022320 (2003). Lett. 60, 1940-1942 (1988). [86] D. Buono, G. Nocerino, A. Porzio, and S. Solimeno, “Ex- [73] C. Weedbrook, S. Pirandola, R. García-Patrón, N. J. perimental analysis of decoherence in continuous-variable Cerf, T. C. Ralph, J. H. Shapiro, and S. Lloyd, “Gaus- bipartite systems,” Phys. Rev. A 86, 042308 (2012). sian quantum information,” Rev. Mod. Phys. 84, 621-669 [87] W. P. Bowen, R. Schnabel, P. K. Lam, and T. C. Ralph, (2012). “Experimental Investigation of Criteria for Continuous [74] C. W. Gardiner, and M. Collett, “Input and output in Variable Entanglement,” Phys. Rev. Lett. 90, 043601 damped quantum systems: Quantum stochastic differen- (2003). tial equations and the master equation,” Phys. Rev. A [88] E. G. Cavalcanti, and M.D. Reid, “Signatures for gener- 31, 3761-3774 (1985). alized macroscopic superpositions,” Phys. Rev. Lett. 97, [75] M. J. Collett, and C. Gardiner, “Squeezing of intracavity 170405 (2006). and traveling-wave light fields produced in parametric