The quantum-to-classical transition and decoherence

Maximilian Schlosshauer Department of Physics, University of Portland, 5000 North Willamette Boulevard, Portland, Oregon 97203, USA

I give a pedagogical overview of decoherence and its role in providing a dynamical account of the quantum-to-classical transition. The formalism and concepts of decoherence theory are reviewed, followed by a survey of master equations and decoherence models. I also discuss methods for mitigating decoherence in quantum information processing and describe selected experimental investigations of decoherence processes. Note: Please see arXiv:1911.06282 [quant-ph] (published as Phys. Rep. 831, 1–57, 2019) for a much more extensive and up-to-date review of decoherence.

CONTENTS I. INTRODUCTION

I. Introduction1 Realistic quantum systems are never completely iso- lated from their environment. When a quantum system II. Basic formalism and concepts2 interacts with its environment, it will in general become A. Decoherence and interference damping2 entangled with a large number of environmental degrees B. Environmental monitoring and information of freedom. This entanglement influences what we can transfer3 locally observe upon measuring the system. In partic- C. Environment-induced superselection and ular, quantum interference effects with respect to cer- decoherence-free subspaces4 tain physical quantities (most notably, “classical” quan- 1. Pointer states and the commutativity tities such as position) become effectively suppressed, criterion5 making them prohibitively difficult to observe in most 2. Decoherence-free subspaces6 cases of practical interest. This is the process of deco- D. Proliferation of information and quantum herence, sometimes also called dynamical decoherence or Darwinism6 environment-induced decoherence [1–10]. Stated in gen- E. Decoherence versus dissipation and noise7 eral and interpretation-neutral terms, decoherence de- scribes how entangling interactions with the environment III. Master equations7 influence the statistics of results of future measurements A. Born–Markov master equations8 on the system. B. Lindblad master equations8 Formally, decoherence can be viewed as a dynamical C. Non-Markovian decoherence9 filter on the space of quantum states, singling out those states that, for a given system, can be stably prepared IV. Decoherence models 10 and maintained, while effectively excluding most other A. Collisional decoherence 10 states, in particular, nonclassical superposition states of B. Quantum Brownian motion 11 the kind popularized by Schr¨odinger’scat. In this way, C. Spin–boson models 13 decoherence lies at the heart of the quantum-to-classical D. Spin-environment models 13 transition. It ensures consistency between quantum and V. Qubit decoherence, quantum error correction, classical predictions for systems observed to behave clas- and error avoidance 14 sically. It provides a quantitative, dynamical account of A. Correction of decoherence-induced quantum the boundary between quantum and classical physics. In errors 14 any concrete experimental situation, decoherence theory arXiv:1404.2635v2 [quant-ph] 20 Nov 2019 B. Quantum computation on decoherence-free specifies the physical requirements, both qualitative and subspaces 15 quantitative, for pushing the quantum–classical bound- ary toward the quantum realm. Decoherence is a pure C. Environment engineering and dynamical quantum effect, to be distinguished from classical dissi- decoupling 16 pation and stochastic fluctuations (noise). VI. Experimental studies of decoherence 16 Decoherence processes are extremely efficient. Even A. Atoms in a cavity 17 when the environment does not, from a classical point B. Matter-wave interferometry 17 of view, impart significant classical perturbations on the C. Superconducting systems 17 system, quantum-mechanically the system will in most circumstances become rapidly and strongly entangled VII. Decoherence and the foundations of quantum with the environment. Furthermore, due to the many un- mechanics 19 controllable degrees of freedom of the environment, such entanglement is usually irreversible for all practical pur- References 19 poses. Increasingly realistic models of decoherence pro- 2 cesses have been developed, progressing from toy models freedom somewhere in the world that, if they were mea- to complex models tailored to specific experiments (see sured, would allow us to make, with a certain degree of Sec.IV). Advances in experimental techniques have made confidence, a statement about the path of the particle it possible to observe the gradual action of decoherence through the slits. While we cannot say that prior to in experiments such as matter-wave interferometry [11], their measurement, these degrees of freedom have en- cavity QED [12], and superconducting systems [13] (see coded information about a particular, definitive path of Sec.VI). the particle—instead, we have merely correlations involv- The superposition states necessary for quantum in- ing both possible paths—no actual measurement is re- formation processing are typically also those most sus- quired to bring about the decrease in interference visibil- ceptible to decoherence. Thus, decoherence is a major ity. It is enough that, in principle, we could make such barrier to implementing devices for quantum informa- a measurement to obtain which-path information. tion processing such as quantum computers (see Sec.V). This is somewhat loose talk, and conceptual caveats Qubit systems must be engineered to minimize environ- lurk. But it captures quite well the essence of what is mental interactions detrimental to the preparation and happening in decoherence, where those “degrees of free- longevity of the desired superposition states. At the dom somewhere in the world” are the degrees of freedom same time, they must remain sufficiently open to al- of the system’s environment interacting with the system, low for their control. Quantum error correction can leading to the creation of quantum correlations (entan- undo some of the decoherence-induced degradation of glement) between system and environment. Decoherence the superposition state and will be an integral part of can thus be thought of as a process arising from the con- quantum computers (see Sec.VA). Not only is deco- tinuous monitoring of the system by the environment; herence relevant to quantum information, but also vice effectively, the environment is performing nondemolition versa. An information-centric view of quantum mechan- measurements on the system (see Sec.IIB). We now give ics proves helpful in conveying the essence of the deco- a formal quantum-mechanical account of what we have herence process and is also used in recent explorations just tried to convey in words, and then flesh out the con- of the role of the environment as an information channel sequences and details. (see Sec.IIB). It is a curious “historical accident” (Joos’s term [14, p. 13]) that the role of the environment in quantum me- A. Decoherence and interference damping chanics was appreciated only relatively late. While one can find—for example, in Heisenberg’s writings [15]—a Consider again the double-slit experiment and denote few early anticipatory remarks about the role of environ- the quantum states of the particle (call it S, for “sys- mental interactions in the quantum-mechanical descrip- tem”) corresponding to passage through slit 1 and 2 by tion of systems, it wasn’t until the 1970s that the ubiquity |s1i and |s2i, respectively. Suppose that the particle in- and implications of environmental entanglement were re- teracts with another system E—for example, a detec- alized by Zeh [1, 16]. It took another decade for the for- tor or an environment—such that if the quantum state malism of decoherence to be developed, chiefly by Zurek of the particle before the interaction is |s1i, then the [2,3], and for concrete models and numerical estimates quantum state of E will become |E1i (and similarly for of decoherence rates to be worked out [17, 18]. |s2i), resulting in the final composite states |s1i |E1i and Review papers on decoherence include Refs. [4–6, 10, |s2i |E2i, respectively. For an initial superposition state 19]. There are two books on decoherence: a volume α |s1i+β |s2i, the final composite state will be entangled, by Joos et al. [8] (a collection of chapters written by |Ψi = α |s i |E i + β |s i |E i . (1) different authors) and a monograph by this author [9]. 1 1 2 2 Ref. [20] also contains material on decoherence. Foun- The statistics of all possible local measurements on S dational implications of decoherence are discussed in are exhaustively encoded in the reduced density matrix Refs. [6,7,9, 21]. ρS,

ρS = TrE(ρSE) = TrE|ΨihΨ| 2 2 II. BASIC FORMALISM AND CONCEPTS = |α| |s1ihs1| + |β| |s2ihs2| ∗ ∗ + αβ |s1ihs2|hE2|E1i + α β|s2ihs1|hE1|E2i. (2) In the double-slit experiment, we cannot observe an in- For example, suppose we measure particle’s position by terference pattern if we also measure which slit the parti- letting the particle impinge on a distant detection screen. cle went through (that is, if we obtain perfect which-path Statistically, the resulting particle probability density information). In fact, there is a continuous tradeoff be- p(x) will be given by tween interference (phase information) and which-path information: the better we can distinguish the two pos- p(x) = TrS(ρSx) sible paths, the less visible the interference pattern be- 2 2 2 2 = |α| |ψ1(x)| + |β| |ψ2(x)| comes [22]. What is more, for a decrease in interference ∗ ∗ visibility to occur it suffices that there are degrees of + 2 Re {αβ ψ1(x)ψ2 (x)hE2|E1i} , (3) 3 where ψi(x) ≡ hx|sii. The last term represents the in- interaction has form terference contribution. Thus, the visibility of the inter- X  X ference pattern is quantified by the overlap hE2|E1i, i.e., ci |sii |E0i −→ ci |sii |Ei(t)i . (4) by the distinguishability of |E1i and |E2i. In the lim- i i iting case of perfect distinguishability, hE2|E1i = 0, no interference pattern will be observable and we obtain the We have here introduced a time parameter t, where t = 0 classical prediction. Phase relations have become locally corresponds to the onset of the environmental interac- (i.e., with respect to S) inaccessible, and there is no mea- tion, with |Ei(t)i ≡ |E0i for all i; at t < 0 the system and environment are assumed to be uncorrelated (an as- surement on S that can reveal coherence between |s1i and sumption common to most decoherence models). |s2i. The coherence is now between the states |s1i |E1i A single environmental particle interacting with the and |s2i |E2i, requiring an appropriate global measure- ment (acting jointly on S and E) for it to be revealed. system will typically only insufficiently resolve the com- Conversely, if the interaction between S and E is such ponents |sii in the system’s superposition state. But be- that E is completely unable to resolve the path of the cause of the large number of such particles (and, hence, degrees of freedom), the overlap between their different particle, then |E1i and |E2i are indistinguishable and full coherence is retained at the level of S, as is also directly joint states |Ei(t)i will rapidly decrease as a result of obvious from Eq. (1). In the intermediary regime where the build-up of many interaction events. Specifically, in many decoherence models an exponential decay of over- 0 < |hE2|E1i| < 1, meaning that |E1i and |E2i can be distinguished in a one-shot measurement with nonzero lap is found [3,5,9, 17, 20, 28–31], 2 probability p = 1 − |hE2|E1i| < 1, an interference pat- h | i ∝ −t/τd 6 tern of reduced visibility is obtained. Equation (3) shows Ei(t) Ej(t) e for i = j. (5) that the reduction in visibility increases as |E1i and |E2i Here τd is the characteristic decoherence timescale, which become more distinguishable. can be evaluated for particular choices of the parameters Here is another way of putting the matter. Looking in each model (see Sec.IV). back at Eq. (1), we see that E encodes which-way infor- mation about S in the same “relative-state” sense [23] in which EPR correlations [24–26] may be said to encode B. Environmental monitoring and information “information.” That is, if hE2|E1i = 0 and we were to transfer measure E and found it to be in state |E1i, we could, in EPR’s words [24, p. 777], “predict with certainty” that 1 We will now motivate, in a different and more rigorous we will find S in |s1i. Whenever such a prediction is way, the picture of decoherence as a process of environ- possible were we to measure E, no interference effects be- mental monitoring. First, we express the influence of tween the components |s1i and |s2i can be measured at the environment in a completely general way. We as- S, even if E is never actually measured. If |hE2|E1i| > 0, sume that at t = 0 there are no correlations between then E encodes only partial which-way information about system S and environment E, ρSE(0) = ρS(0) ⊗ ρE(0). S, in the sense that a measurement of E could not reliably We write ρE(0) in its diagonal decomposition, ρE(0) = distinguish between |E1i and |E2i; instead, sometimes P P i pi|EiihEi|, where i pi = 1 and the states |Eii form the measurement will result in an outcome compatible an orthonormal basis of the Hilbert space of E. If with both |E1i and |E2i. Consequently, an interference H denotes the Hamiltonian (here assumed to be time- experiment carried out on S would find reduced visibil- independent) of SE and U(t) = e−iHt represents the uni- ity, representing diminished local coherence between the tary time evolution operator, then the density matrix of components |s1i and |s2i. S evolves according to As hinted above, the description developed so far de- scribes the essence of the decoherence process if we iden- ( " !# ) X † tify the particle S more generally with an arbitrary quan- ρS(t) = TrE U(t) ρS(0) ⊗ pi|EiihEi| U (t) tum system and the second system E with the environ- i ment of S. Then an idealized account of the decoherence X † = pi hEj| U(t) |Eii ρS(0) hEi| U (t) |Eji . (6) ij

Introducing the Kraus operators [32] defined by Eij ≡ 1 √ Of course, this must not be read as saying that S was already pi hEj| U(t) |Eii, we obtain in |s1i (i.e., “went through slit 1”) prior to the measurement of E. Nor does it mean that the result of a subsequent path X † measurement on S is necessarily determined, by virtue of the ρS(t) = EijρS(0)Eij. (7) measurement on E, prior to this S-measurement’s actually be- ij ing carried out. After all, as Peres has cautioned us [27], unper- formed measurements have no outcomes. So while the picture It is customary to combine the two indices i and j into a of E as “encoding which-path information” about S is certainly single index and write the Kraus operators as suggestive and helpful, it should be used with an understanding of its conceptual pitfalls. √ Wk ≡ pik hEjk | U(t) |Eik i , (8) 4 such that outcome α is

X † ρS(t) = WkρS(0)Wk . (9) (α) TrE {[I ⊗ Pα] ρSE(t)[I ⊗ Pα]} k ρS (t) = Prob (α | ρS(t))  † This Kraus-operator formalism (also called operator-sum TrE [I ⊗ Pα] U(t)[ρS(0) ⊗ ρE(0)] U (t)[I ⊗ Pα] = . formalism) represents the effect of the environment as Prob (α | ρS(t)) a sequence of (in general nonunitary) transformations of (12) ρS generated by the operators Wk. The Kraus operators exhaustively encode information about the initial state Inserting the diagonal decomposition ρE(0) = P of the environment and about the dynamics of the joint k pk|EkihEk| and carrying out the trace gives SE system. Because the evolution of SE is unitary, the [19] Kraus operators satisfy the completeness constraint † (α) X Mα,kρS(t)Mα,k X † ρ (t) = . (13) WkWk = IS, (10) S Prob (α | ρS(t)) k k where IS is the identity operator in the Hilbert space of Here we have introduced the measurement operators S. Equations (9) and (10) together imply that the Wk are √ the generators of a completely positive map Φ : ρS(0) 7→ Mα,k ≡ pk hα| U(t) |Eki , (14) ρS, also known as a quantum operation [32] or quantum channel.2 which obey the completeness constraint P † We will now use Eq. (9) to formally motivate the view α,k Mα,kMα,k = IS. Equation (12) describes the that decoherence corresponds to an indirect measurement effect of the indirect measurement on the state of the of the system by the environment, and that it thus re- system. If, however, we do not actually inquire about sults from a transfer of information from the system to the result of this measurement, we must assign to the the environment (see also Ref. [19]). In such an indi- system a density operator that is a sum over all the rect measurement, we let the system S interact with a (α) possible conditional states ρS (t) weighted by their probe—here the environment E—followed by a projec- probabilities Prob (α | ρS(t)), tive measurement on E. The probe is treated as a quan- tum system. This procedure aims to yield information X (α) ρS(t) = Prob (α | ρS(t)) ρS (t) about S without performing a projective (and thus de- α structive) direct measurement on S. To model such an X † indirect measurement, consider again an initial compos- = Mα,kρS(t)Mα,k. (15) α,k ite density operator ρSE(0) = ρS(0) ⊗ ρE(0) evolving under the action of U(t) = e−iHt, where H is the to- tal Hamiltonian. Consider a projective measurement M Note that this expression is formally analogous to the on E with eigenvalues α and corresponding projectors Kraus-operator expression of Eq. (9), which described 2 † the effect of a general environmental interaction on the Pα ≡ |αihα|, with Pα = Pα = Pα. The probability of obtaining outcome α in this measurement when S is de- state of the system. Recall, further, that the situation we encounter in decoherence is precisely one in which we do scribed by the density operator ρS(t) is not actually read out the environment—or, in the present picture, in which we do not inquire about the result of the Prob (α | ρ (t)) = Tr (P ρ (t)) S E α E indirect measurement. This suggests that decoherence   †  = TrE PαTrS U(t)(ρS(0) ⊗ ρE(0)) U (t) . (11) can indeed be understood as an indirect measurement— a monitoring—of the system by its environment. The density matrix of S conditioned on the particular

C. Environment-induced superselection and decoherence-free subspaces 2 The Kraus formalism is of limited use in calculating decoherence dynamics for concrete situations of physical interest. This is so because finding the Kraus operators corresponds to diagonaliz- Decoherence can occur in any basis; which observable ing the full Hamiltonian of SE, usually a prohibitively difficult is monitored by the environment depends on the spe- task. Moreover, the Kraus operators contain all contributions cific form of the system–environment interaction. The to the evolution of the reduced density matrix, while for con- preferred states (or preferred observables) of the system siderations of decoherence we are typically interested only in the nonunitary terms, and certain contributions—such as back- emerge dynamically as those states that are the most ro- action effects from the system on the environment—can often be bust to the interaction with the environment, in the sense neglected. (This is where master equations come into play; see that they become least entangled with the environment; Sec.III.) thus, they are the states most immune to decoherence. 5

This is the stability criterion for the selection of pre- Interaction Hamiltonians frequently describe the scat- ferred states, resulting in the dynamical selection of pre- tering of surrounding particles (photons, air molecules, ferred states (“environment-induced superselection”) [1– etc.), leading to collisional decoherence (see Sec.IVA). 3, 16]. These environment-superselected preferred states Since the force laws describing such processes typically (or observables) are sometimes also called pointer states depend on some power of distance, the interaction Hamil- (or pointer observables)[2], since they correspond to the tonian will then commute with the position operator. physical quantities that are most easily “read off” at the Thus, the pointer states will be approximate eigenstates level of the system, akin to the pointer on the dial of a of position (i.e., narrow position-space wave packets). measurement apparatus. This explains why superpositions of mesoscopically and macroscopically distinct positions are prohibitively diffi- cult to observe [2,3, 17, 31, 33–39]. Collisional decoher- 1. Pointer states and the commutativity criterion ence can also be dominant in microscopic systems when these systems occur in distinct spatial configurations that To find the preferred states, we decompose the couple strongly to the surrounding medium. For exam- total system–environment Hamiltonian into the self- ple, chiral molecules such as sugar are always observed to Hamiltonians of the system S and environment E rep- be in chirality eigenstates (left-handed or right-handed), resenting the intrinsic dynamics, and a part Hint repre- which are superpositions of different energy eigenstates. senting the interaction between system and environment, Any attempt to prepare such molecules in energy eigen- states leads to immediate decoherence into the environ- H = HS + HE + Hint. (16) mentally stable chirality eigenstates [40, 41]. The quantum limit of decoherence [42] arises when the In many cases of practical interest, Hint dominates modes of the environment are slow in comparison with the evolution of the system, H ≈ Hint (the quantum- the evolution of the system—that is, when the highest measurement limit of decoherence). We look for system frequencies (i.e., energies) available in the environment states |sii such that the composite system–environment are smaller than the separation between the energy eigen- state, when starting from a product state |sii |E0i at states of the system. Then the environment will be able t = 0, remains in the product form |sii |Ei(t)i for all to monitor only quantities that are constants of motion. t > 0 under the action of Hint (we shall assume here In the case of nondegeneracy, this quantity will be the en- that Hint is not explicitly time-dependent). That is, we ergy of the system, leading to the environment-induced demand that (setting ~ ≡ 1 from here on) superselection of energy eigenstates for the system [42].3 In many realistic situations, the commutativity crite- −iHintt −iHintt e |sii |E0i = λi |sii e |E0i ≡ |sii |Ei(t)i . rion, Eq. (18), can only be fulfilled approximately [43, 44]. (17) In addition, the self-Hamiltonian of the system and the Thus, the pointer states |sii are the eigenstates of the interaction Hamiltonian may contribute in roughly equal part of the interaction Hamiltonian Hint pertaining to the strengths (e.g., in models for quantum Brownian motion Hilbert space of the system, with eigenvalues λi. These [4, 45]; see Sec.IVB), rendering neither the quantum- states will be stationary under Hint [2]. It follows that the P measurement limit of negligible intrinsic dynamics nor pointer observable defined by OS = i oi|siihsi| com- the quantum limit of decoherence of a slow environ- mutes with Hint, ment appropriate. In such cases, more general methods   for determining the preferred states are required. The OS,Hint = 0. (18) predictability-sieve strategy [43, 44, 46] computes the time dependence of the amount of decoherence introduced into This commutativity criterion [2,3] is particularly easy to the system for a large set of initial states of the system apply when H takes the tensor-product form H = int int evolving under the total system–environment Hamilto- S ⊗ E, as is frequently the case. Then the environment- nian. Typically, this decoherence is measured using ei- superselected observables will be those observables that ther the purity Tr ρ2
or the von Neumann entropy commute with S. If S is Hermitian, it represents the S physical quantity monitored by the environment. In gen- eral, any Hint can be written as a diagonal decomposition of (unitary but not necessarily Hermitian) system and P 3 Textbooks on quantum mechanics usually attribute a special role environment operators Sα and Eα, Hint = α Sα ⊗ Eα. to such energy eigenstates (for closed systems) since they are If the Sα are Hermitian, such a Hamiltonian represents stationary under the action of the Hamiltonian. In this closed- the simultaneous environmental monitoring of different system picture, however, arbitrary superpositions of energy observables Sα of the system. A sufficient condition for eigenstates should nonetheless be perfectly legitimate. Thus, it {|s i} to form a set of pointer states of the system is then is important to realize that the environment-induced superselec- i tion of energy eigenstates is not equivalent to a situation in which given by the requirement that the |sii be simultaneous the presence of the environment could be neglected altogether; eigenstates of the operators Sα, instead, the environment plays the crucial role of continuously monitoring the energy of the system, leading to a local suppres- (α) sion of coherence between energy eigenstates. Sα |sii = λi |sii for all α and i. (19) 6

S(ρS) = −Tr (ρS log2 ρS) of the reduced density matrix D. Proliferation of information and quantum ρS. The states most immune to decoherence will be those Darwinism which lead to the smallest decrease in purity or the small- est increase in von Neumann entropy. Application of this method leads to a ranking of the possible preferred states with respect to their robustness to the interac- Quantum Darwinism [61–69] builds on the ideas of de- tion with the environment. For particular models it has coherence and environmental encoding of information, by been explicitly shown that the states picked out by the broadening the role of the environment to that of a com- predictability sieve are robust to the particular choice of munication and amplification channel. Interactions be- the measure of decoherence. For example, in the model tween the system and its environment lead to the redun- for quantum Brownian motion, different measures lead dant storage of selected information about the system in to the same minimum-uncertainty wave packets in phase many fragments of the environment. By measuring some space [5,8, 16, 44, 47, 48]. of these fragments, observers can indirectly obtain infor- mation about the system without appreciably disturbing the system itself. Indeed, this represents how we typi- cally observe objects. For example, we see an object not 2. Decoherence-free subspaces by directly interacting with it, but by intercepting scat- tered photons that encode information about the object’s The pointer-state condition of Eq. (19) can be spatial structure [67, 68]. strengthened to the concept of pointer subspaces [3] or decoherence-free subspaces (DFS) [49–58]. These are sub- spaces of the Hilbert space of the system in which every In this sense, quantum Darwinism provides a dynami- state in the subspace is immune to decoherence; this is cal explanation for the robustness of states of (especially) a nontrivial requirement, since in general superpositions macroscopic objects to observation. It was found that of pointer states will not be pointer states themselves. the observable of the system that can be imprinted most One important condition for this to happen is that the completely and redundantly in many distinct fragments preferred states |sii defined by Eq. (19) form an orthonor- of the environment coincides with the pointer observable (α) selected by the system–environment interaction [62–65]; mal basis of the subspace, and that the eigenvalues λ i conversely, most other states do not seem to be redun- in Eq. (19) are independent of i, i.e., that all |s i are i dantly storable. Indeed, it has been shown that the re- simultaneous degenerate eigenstates of each S , α dundant proliferation of information regarding pointer (α) states is as inevitable as decoherence itself [70]. Quantum Sα |sii = λ |sii for all α and i. (20) Darwinism has been studied in several concrete models, for example, in spin environments [64], quantum Brow- This condition states that the action of a given Sα must nian motion [71], and photon and photon-like environ- be the same for all basis states |sii of the DFS, and thus ments [67, 68, 70]. The efficiency of the amplification pro- the existence of a DFS corresponds to a symmetry in the cess described by quantum Darwinism can be expressed structure of the system–environment interaction, i.e., to in terms of the quantum Chernoff information [70]. a dynamical symmetry. A necessary condition for such a symmetry to obtain is the absence of terms in Hint that act jointly on system and environment in a nontrivial The structure and amount of information that the manner. environment encodes about the system can be quanti- An arbitrary state |ψi in the DFS can then be written fied using the measure of (classical [62, 63] or quantum P as |ψi = i ci |sii and will evolve according to [5, 64, 65]) mutual information. Classical mutual infor- mation is based on the choice of particular observables of −iH t −i P λ(α)E t the system S and the environment E and quantifies how int | i | i | i ( α α) | i e ψ E0 = ψ e E0 well one can predict the outcome of a measurement of a ≡ |ψi |Eψ(t)i . (21) given observable of S by measuring some observable on a fraction of E [62, 63]. Quantum mutual information is Thus, the state |ψi does not become entangled with the defined as S(ρS)+S(ρE)−S(ρ), where ρS, ρE, and ρ are environment and is therefore immune to decoherence. the density matrices of S, E, and the composite system When the self-Hamiltonian HS of the system cannot be SE, respectively, and S(ρ) = −Tr (ρ log2 ρ) is the von neglected, one needs to additionally ensure that none of Neumann entropy associated with ρ. Quantum mutual the basis states |sii of the DFS will drift out of the sub- information quantifies the degree of quantum correlations space under the evolution generated by HS. Otherwise between S and E. Classical and quantum mutual infor- an initially decoherence-free state would again become mation give similar results [5, 62–65] because the differ- prone to decoherence. The concept of DFS can be gener- ence between the two measures, known as the quantum alized to the formalism of noiseless subsystems (or noise- discord [72], disappears when decoherence is sufficiently less quantum codes)[58–60]. effective to select a well-defined pointer basis [72]. 7

E. Decoherence versus dissipation and noise always be undone using only local operations (witness, for example, the reversal of ensemble dephasing in NMR While the presence of dissipation implies the pres- experiments using the spin-echo technique). In any indi- ence of decoherence, the converse is not necessarily true. vidual realization of the noise process the dynamics of the When dissipation and decoherence are both present, they system are completely unitary, and thus no coherence is typically occur on vastly different timescales; the deco- lost from the system. By contrast, if the system becomes herence timescale is typically many orders of magnitude entangled with environmental degrees of freedom, at the shorter than the relaxation timescale. A rule-of-thumb very least we would need to perform a pair of measure- estimate for the ratio of the relaxation timescale τr to the ments on the environment before and after the interac- decoherence timescale τd for a massive object described tion with the system in order to gather enough informa- by a superposition of two different positions a distance tion to reverse the effect of decoherence by application ∆x apart is [18] of an appropriate countertransformation. Moreover, as also seen experimentally [77], these measurements would τ  ∆x 2 r ∼ , (22) not always constitute a sufficient procedure for “undo- τd λdB ing” decoherence (see also Sec. IV.C of Ref. [5]). −1/2 The loss of phase coherence due to environmental where λdB = (2mkT ) is the thermal de Broglie wave- entanglement is sometimes simulated (with the above length of the object. For an object of mass m = 1 g at caveats) by classical fluctuations perturbing the system, room temperature in a coherent superposition of two lo- i.e., by the addition of certain time-dependent terms to cations a distance ∆x = 1 cm apart, τr/τd is on the order 40 the self-Hamiltonian of the system. This strategy was of 10 . Thus, for macroscopic objects the dissipative in- implemented, for example, in theoretical [73, 78] and ex- fluence of the environment is usually completely negligi- perimental [77, 79] studies of the influence of fluctuating ble on the timescale relevant to the decoherence induced parameters in ion-trap quantum computers. by this environment. Decoherence is a consequence of environmental entan- glement. In the literature on quantum computing, how- III. MASTER EQUATIONS ever, the term “decoherence” is often used to refer to any process that affects the qubits, including perturba- In the usual approach to modeling decoherence, the tions due to classical fluctuations and imperfections. Ex- reduced density matrix ρ (t) is obtained from amples for sources of such classical noise in the context S of quantum computing are the fluctuations in the inten-  † ρS(t) = TrE ρSE(t) ≡ TrE U(t)ρSE(0)U (t) , (23) sity [73] and duration [74] of the laser beam incident on qubits in an ion trap, inhomogeneities in the magnetic where U(t) is the time-evolution operator for the compos- fields in NMR quantum computing [75], and bias fluctu- ite system SE. The task of calculating ρSE(t) is often ations in superconducting qubits [76]. The distinction be- computationally cumbersome or even intractable. It is tween classical noise and quantum decoherence has been also unnecessarily detailed, because we are usually only further blurred in quantum error correction, since the interested in the dynamics of the system. A master equa- error-correcting schemes are insensitive to the physical tion allows us to calculate ρS(t) directly from an expres- origin of the qubit errors (see Sec.VA). sion of the form Phenomenologically and formally the influence of clas- sical noise processes may be described in a manner simi- ρS(t) = V(t)ρS(0), (24) lar to the effect of environmental entanglement, namely, in terms of a decay of the off-diagonal elements (in- where the superoperator V(t) is the dynamical map gen- terference terms) in the local density matrix (in the erating the evolution of ρS(t). If the master equation environment-superselected basis). But in the case of is exact, then we merely have the identity V(t)ρS(0) ≡  † noise, the decay of the off-diagonal elements occurs be- TrE U(t)ρSE(0)U (t) and no computational advantage cause the system’s density matrix is identified with an is gained. Therefore, master equations are typically average over a physical ensemble of systems (or, put dif- based on simplifying approximations. ferently, over the different instances of particular noise In modeling decoherence, we focus on master equations processes), while in the case of decoherence the decay is that are first-order time-local differential equations of the due to an entanglement-induced delocalization of phase form coherence for individual systems. The fundamental dif- d 0 ference between these physical processes is masked by the ρS(t) = L [ρS(t)] ≡ −i [H , ρS(t)] + D[ρS(t)]. (25) dt S density-matrix description. Indeed, one can always find an experimental procedure that would, at least in princi- This equation is local in time in the sense that the change ple, distinguish between the different physical processes of ρS at time t depends only on ρS evaluated at t. The underlying formally similar density-matrix descriptions. superoperator L acting on ρS(t) typically depends on the In contrast with decoherence, noise does not create initial state of the environment and the different terms system–environment entanglement and can in principle in the Hamiltonian. We have decomposed L into two 8 parts to distinguish their physical interpretation. The omitting the superscript “I”; instead we use the conven- 0 first term, −i [HS, ρS(t)], is unitary and given by the tion that all operators bearing explicit time arguments Liouville–von Neumann commutator with the “renormal- are to be understood as interaction-picture operators. 0 ized” Hamiltonian HS of the system. (Because the en- (For density operators, however, we will maintain the vironment typically leads to a renormalization of the en- superscript notation in order to distinguish them from ergy levels of the system, this Hamiltonian does in general Schr¨odinger-picturedensity operators, which also carry not coincide with the unperturbed free Hamiltonian HS a time argument.) The quantities cαβ(τ) appearing in of S that would generate the evolution of S in absence of Eq. (27) are given by the environment.) The second, nonunitary term D[ρS(t)] represents decoherence (and often also dissipation) due to c (τ) ≡ hE (τ)E i . (28) αβ α β ρE the environment. These environment self-correlation functions quantify how much information the environment retains over time A. Born–Markov master equations about its interaction with the system. The Markov ap- proximation corresponds to the assumption of a rapid Born–Markov master equations allow for many deco- decay of the cαβ(τ) relative to the timescale set by the herence problems to be treated in a mathematically sim- evolution of the system. ple, and often closed, form. They are based on the fol- In many situations of interest, the general form of the lowing two approximations: Born–Markov master equation, Eq. (26), simplifies con- siderably. For example, typically only a single system 1. The Born approximation. The system– observable S is monitored by the environment, Hint = environment coupling is sufficiently weak and S ⊗ E. Also, the time dependence of the operators Sα(τ) the environment is reasonably large such that and Eα(τ) is often simple, facilitating the calculation of changes of the density operator ρE of the environ- the quantities Bα and Cα. Examples are discussed in ment are negligible and the system–environment Sec.IV. density operator remains remains approximately factorized at all times, ρSE(t) ≈ ρS(t) ⊗ ρE. 2. The Markov approximation. Memory effects of the B. Lindblad master equations environment are negligible, in the sense that any self-correlations within the environment created by Lindblad master equations constitute a particular, al- the coupling to the system decay rapidly compared beit quite general, class of time-local Markovian mas- to the characteristic relaxation timescale of the ter equations. They arise from the requirement that the open quantum system. evolution of the reduced density matrix generated by the master equation must ensure complete positivity [20, 80– Comparisons between the predictions of models based 85]. Complete positivity guarantees that the dynamical on Born–Markov master equations and experimental map ρS(0) 7→ ρS(t) = V(t)ρS(0) described by the master data indicate that the Born and Markov assumptions are equation generates physically consistent dynamics even reasonable in many physical situations (but see Sec.IIIC when S is initially entangled with another system. While below for exceptions and non-Markovian models). As- complete positivity is automatically fulfilled if the evo- suming these assumptions hold and writing the inter- P lution is exact, approximate master equations will not action Hamiltonian as Hint = α Sα ⊗ Eα, the Born– necessarily ensure complete positivity [20, 83–86]. The Markov master equation reads [9, 20] Lindblad master equation is a special case of the gen- eral Born–Markov master equation that ensures complete d ρS(t) = −i [HS, ρS(t)] positivity and takes the general form [81, 82] dt X − {[Sα,BαρS(t)] + [ρS(t)Cα,Sα]} , (26) d 0 ρS(t) = −i [H , ρS(t)] α dt S 1 X nh † i h † io where the system operators Bα and Cα are defined as + γαβ Sα, ρS(t)S + SαρS(t),S , (29) 2 β β Z ∞ αβ X (I) Bα ≡ dτ cαβ(τ)Sβ (−τ), (27a) 0 0 β where HS is the renormalized Hamiltonian of the sys- Z ∞ tem. The coefficients γαβ are time-independent and ex- X (I) Cα ≡ dτ cβα(−τ)Sβ (−τ). (27b) haustively encapsulate information about the physical 0 β parameters of the decoherence processes (and possibly dissipation processes). One can show that the matrix (I) Here Sα (−τ) denotes the operator Sα in the interaction Γ ≡ (γαβ) formed by the coefficients γαβ is positive, i.e., picture. In the following, we will simplify notation by all its eigenvalues κµ are ≥ 0. Therefore, Eq. (29) can be 9 simplified by diagonalizing Γ, which results in the diago- In this case, ρS(t) evolves unitarily. Since the Lµ are lin- nal form [82, 87] ear combinations of the Sα, Eq. (34) typically means that [Sα, ρS(t)] = 0 for all α, t. This implies that simultane- d 0 ous eigenstates of all Sα will be immune to decoherence, ρS(t) = −i [H , ρS(t)] dt S which is precisely the pointer-state criterion of Eq. (19). 1 X In quantum-jump and quantum-trajectory approaches, − κ L† L ρ (t) + ρ L† L − 2L ρ (t)L† . µ µ µ S S µ µ µ S µ the evolution of the reduced density matrix is conditioned 2 µ on an explicitly observed sequence of measurement re- (30) sults in the environment. This allows for the (formal) The Lindblad operators Lµ are linear combinations of the description of a single realization of the system evolv- original operators Sα, with coefficients determined by the ing stochastically, conditioned on a particular measure- diagonalization of Γ. The Lindblad structure of a mas- ment record. The dynamics are then described by a mas- ter equation can also be motivated from the requirement ter equation of the Lindblad type, Eq. (31), for the re- C that it gives rise to the most general form of generators duced density matrix ρS conditioned on the measurement of quantum dynamical semigroups [20, 81, 82, 84, 87–89]. records of the Lindblad operators Lµ, It is possible to bring any Born–Markov master equation into Lindblad form by imposing the rotating-wave ap- C C 1 X C dρ = −i H , ρ  dt − κ L , L , ρ  dt proximation. This assumption, ubiquituous in quantum S S S µ µ µ S 2 µ optics, is justified whenever the timescale set by the typ- 0 X √ C ical energy differences ~(ω − ω ) of the system Hamilto- + κµ W[Lµ]ρS dWµ. (35) nian is short in comparison with the relaxation timescale µ of the system. (See Sec. 3.3.1 of Ref. [20] for details.) †  † Because the Sα are not necessarily Hermitian, the Here, W[L]ρ ≡ Lρ+ρL −ρ Tr Lρ + ρL , and the dWµ Lindblad operators do not always correspond to physical denote so-called Wiener increments. Equation (35) corre- observables. But when they do, we can rewrite Eq. (30) sponds to a diffusive unraveling of the Lindblad equation in compact double-commutator form, into individual quantum trajectories, which can then be expressed by means of a stochastic Schr¨odingerequation d 0 1 X [90–102]. ρS(t) = −i [HS, ρS(t)] − κµ [Lµ, [Lµ, ρS(t)]] . dt 2 µ (31) As an example, consider a situation in which the envi- C. Non-Markovian decoherence ronment monitors the position of a system. With L = x 0 2 and the “free”-particle Hamiltonian HS = HS = p /2m, The derivation of the Born–Markov master equation Eq. (31) becomes assumes that the coupling between system and environ- ment is weak and memory effects of the environment can d i  2  1 be neglected. These conditions, however, are not met ρS(t) = − p , ρS(t) − κ [x, [x, ρS(t)]] . (32) dt 2m 2 in certain situations of physical interest. An example Expressing this master equation in the position represen- would be a superconducting qubit strongly coupled to a tation results in low-temperature environment of other two-level systems [103, 104]. Also, a recent experiment [105] has measured 0  2 2  ∂ρS(x, x , t) i ∂ ∂ 0 strongly non-Ohmic spectral densities for the environ- = − − ρS(x, x , t) ∂t 2m ∂x02 ∂x2 ment of a quantum nanomechanical system; such densi- ties lead to non-Markovian evolution. 1 0 2 0 − κ (x − x ) ρS(x, x , t). (33) In many cases, pronounced memory effects in the envi- 2 ronment will cause strong dependencies of the evolution This is the classic equation of motion for decoherence due of the reduced density operator on the past history of the to environmental scattering first derived in Ref. [17]. system–environment composite and therefore make it im- Lindblad master equations provide an intuitive and possible to describe the reduced dynamics by a differen- simple way of representing the environmental monitoring tial equation that is local in time. Surprisingly, however, of an open quantum system. Most of the real physics be- one can show that even non-Markovian dynamics some- hind this monitoring process is hidden in the coefficients times can still be described by a time-local differential κµ appearing in Eq. (30). If the Lindblad operators are equation of the form chosen to be dimensionless, the κµ can be directly in- terpreted as decoherence rates, since they have units of d ρS(t) = K(t)ρS(t), (36) inverse time. dt Equation (31) shows that the decoherence term van- ishes if where the superoperator K(t) depends only on t. For example, a non-Markovian master equation for quantum [Lµ, ρS(t)] = 0 for all µ, t. (34) Brownian motion (see Sec.IVB) can be obtained through 10 a formal modification of the Born–Markov master equa- If we assume that the central particle is much more tion [4,5]. In general, it is often possible to arrive at massive than the environmental particles such that its non-Markovian but time-local master equations via the center-of-mass state is not disturbed by the scattering so-called time-convolutionless projection operator tech- events (no recoil), the time evolution of the reduced den- nique [106–109]. sity matrix is given by [9, 17, 31, 34, 35] 0 ∂ρS(x, x , t) 0 0 = −F (x − x )ρS(x, x , t). (37) IV. DECOHERENCE MODELS ∂t This master equation describes pure spatial decoherence 0 Many physical systems can be represented either by without dissipation. The decoherence factor F (x − x ) a qubit if the state space of the system is discrete and plays the role of a localization rate. It represents the effectively two-dimensional, or by a particle described by characteristic decoherence rate at which spatial coher- 0 continuous phase-space coordinates. Needless to say, in ences between two positions x and x become locally sup- the case of quantum information processing the qubit pressed and is given by representation is of particular relevance. Z ∞ Similarly, a wide range of environments can be modeled F (x − x0) = dq %(q)v(q) as a collection of quantum harmonic oscillators or qubits. 0 Z 0 Harmonic-oscillator environments are of great generality. dˆn dˆn  0 0  2 × 1 − eiq(ˆn−ˆn )·(x−x ) |f(qˆn, qˆn0)| . (38) At low energies, many systems interacting with an en- 4π vironment can effectively be represented by one or two Here %(q) denotes the number density of incoming par- coordinates of the system linearly coupled to an environ- ticles with magnitude of momentum equal to q = |q|, ˆn ment of harmonic oscillators; indeed, sufficiently weak in- and ˆn0 are unit vectors (with dˆn and dˆn0 representing the teractions with an arbitrary environment can be mapped associated solid-angle differentials), and v(q) denotes the onto a system linearly coupled to a harmonic-oscillator speed of particles with momentum q. For the scattering environment [110, 111]. of massive environmental particles we have v(q) = q/m, Environments represented by qubits are often the ap- where m is each particle’s mass, while for the scatter- propriate model in the low-temperature regine, where de- ing of photons and other massless particles v(q) is equal coherence is typically dominated by interactions with lo- 0 2 calized modes, such as paramagnetic spins, paramagnetic to the speed of light. The quantity |f(qˆn, qˆn )| is the electronic impurities, tunneling charges, defects, and nu- differential cross section for the scattering of an environ- mental particle from initial momentum q = qˆn to final clear spins [103, 104, 112]. Each of the localized modes is 0 0 represented by a finite-dimensional Hilbert space with a momentum q = qˆn . finite energy cutoff. We can therefore model these modes Whenever the mass of the central particle becomes as a set of discrete states. Typically, only two such states comparable to the mass of the environmental particles (as are relevant, and thus the localized modes can be mapped in the case of air molecules scattered by small molecules onto an environment of qubits. Since qubits can be for- and free electrons [114]), the no-recoil assumption does 1 not hold and more general models for collisional deco- mally represented by spin- 2 particles, such models are known as spin-environment models. herence have to be considered [35, 36]. The resulting In the following, we will discuss four important stan- dynamics include dissipation, as well as decoherence in both position and momentum. dard models, namely, collisional decoherence (Sec.IVA), 0 quantum Brownian motion (Sec.IVB), the spin–boson To further evaluate the decoherence factor F (x − x ), model (Sec.IVC), and the spin–spin model (Sec.IVD). Eq. (38), we distinguish two important limiting cases. In For details on these and other decoherence models, in- the short-wavelength limit, the typical wavelength of the scattered environmental particles is much shorter than cluding derivations of the relevant master equations, see 0 Secs. 3 and 5 of Ref. [9]. the coherent separation ∆x = |x − x | between the well- localized wave packets in the spatial superposition state of the system. Then a single scattering event will be A. Collisional decoherence able to fully resolve this separation and thus carry away complete which-path information, leading to maximum spatial decoherence per scattering event. In this limit, Collisional decoherence arises from the scattering of en- F (x − x0) turns out to be simply equal to the total scat- vironmental particles by a massive free quantum particle. tering rate Γtot [9]. This implies the existence of an upper Models of collisional decoherence were first studied in the limit for the decoherence rate when increasing the sepa- classic paper by Joos and Zeh [17]. A more rigorous and ration ∆x, in contrast with decoherence rates obtained general treatment was later developed by Hornberger and from linear models [compare Eqs. (22) and (54)]. Equa- collaborators [31, 36–39] (see also [34, 35, 113]), which, tion (37) then shows that spatial interference terms will among other refinements, remedied a flaw in Joos and become exponentially suppressed at a rate set by Γ , Zeh’s original derivation that had resulted in decoher- tot 0 0 −Γtott ence rates that were too large by a factor of 2π [31]. ρS(x, x , t) = ρS(x, x , 0)e . (39) 11

is given by TABLE I. Estimates of decoherence timescales (in seconds) for the suppression of spatial interferences over a distance ∆x   3 X 1 2 1 2 2 equal to the size a of the object (∆x = a = 10− cm for a HE = pi + miωi qi , (41) 6 2m 2 dust grain and ∆x = a = 10− cm for a large molecule). See i i Ref. [9] for details. where mi and ωi denote the mass and natural frequency Environment Dust grain Large molecule of the ith oscillator, and qi and pi are the canonical posi- Cosmic background radiation 1 1024 tion and momentum operators. The interaction Hamilto- 18 6 nian Hint describes the bilinear coupling of the system’s Photons at room temperature 10− 10 14 2 position coordinate x to the positions qi of the environ- Best laboratory vacuum 10− 10− P 31 19 mental oscillators, Hint = x ⊗ i ciqi, where the ci de- Air at normal pressure 10− 10− note coupling strengths. This interaction represents the continuous environmental monitoring of the position co- ordinate of the system. In the opposite long-wavelength limit, the environmen- The Born–Markov master equation describing the evo- tal wavelengths are much larger than the coherent sep- 0 lution of the density matrix ρS(t) of the system is given aration ∆x = |x − x |, which implies that an individual by [9, 45] scattering event will reveal only incomplete which-path information. For this case, one can show that spatial co- d   ρS(t) = −i HS, ρS(t) herences become exponentially suppressed at a rate that dt depends on the square of the separation ∆x [9], Z ∞    2 − dτ ν(τ) x, x(−τ), ρS(t) ρ (x, x0, t) = ρ (x, x0, 0)e−Λ(∆x) t, (40) 0 S S   
− iη(τ) x, x(−τ), ρS(t) . (42) where Λ is a scattering constant that encapsulates the physical details of the interaction. Thus, the quantity Here, x(τ) denotes the system’s position operator in the 2 Λ(∆x) plays the role of a decoherence rate. The de- interaction picture, x(τ) = eiHS τ xe−iHS τ . The curly pendence of this rate on ∆x is reasonable: if the envi- brackets {· , ·} in the second line denote the anticom- ronmental wavelengths are much larger than ∆x, it will mutator {A, B} ≡ AB + BA. The functions require a large number of scattering events to encode an appreciable amount of which-path information in the Z ∞  ω  environment, and this amount will increase, for a given ν(τ) = dω J(ω) coth cos (ωτ) , (43) 0 2kT number of scattering events, as ∆x becomes larger. Note Z ∞ that if ∆x is increased beyond the typical wavelength of η(τ) = dω J(ω) sin (ωτ) , (44) the environment, the short-wavelength limit needs to be 0 considered instead, for which the decoherence rate is in- are known as the noise kernel and dissipation kernel, re- dependent of ∆x and attains its maximum possible value. spectively. The function J(ω), called the spectral density Numerical values of collisional decoherence rates ob- of the environment, is given by tained from Eqs. (39) and (40), with the physically rele- vant scattering parameters Γtot and Λ appropriately eval- 2 X ci uated, have shown the extreme efficiency of collisions in J(ω) ≡ δ(ω − ωi). (45) 2m ω suppressing spatial interferences; TableI shows a few i i i classic order-of-magnitude estimates [8,9, 17]. Excel- lent agreement between theory and experiment has been In general, spectral densities encapsulate the physi- demonstrated for the decoherence of fullerenes due to col- cal properties of the environment. One frequently re- lisions with background gas molecules in a Talbot–Lau places the collection of individual environmental oscilla- interferometer [31, 115–118] (see Sec.VIB and Fig.2), tors by an (often phenomenologically motivated) contin- and for the decoherence of sodium atoms in a Mach– uous function J(ω) of the environmental frequencies ω. Zehnder interferometer due to the scattering of photons If we specialize to the important case of the system rep- [119] and gas molecules [120]. resented by a harmonic oscillator with self-Hamiltonian

1 2 1 2 2 HS = p + MΩ x , (46) B. Quantum Brownian motion 2M 2 the resulting Born–Markov master equation is A classic and extensively studied model of decoherence and dissipation is the one-dimensional motion of a par- d  1 2 2     ticle weakly coupled to a thermal bath of noninteracting ρS(t) = −i HS + MΩe x , ρS(t) −iγ x, p, ρS(t) dt 2 harmonic oscillators, a model known as quantum Brown-       − D x, x, ρ (t) − f x, p, ρS(t) . (47) ian motion. The self-Hamiltonian HE of the environment S 12

The coefficients Ωe2, γ, D, and f are defined as 2 Z ∞ p p Ωe2 ≡ − dτ η(τ) cos (Ωτ) , (48a) M 0 Z ∞ 1 x x γ ≡ dτ η(τ) sin (Ωτ) , (48b) MΩ 0 Z ∞ D ≡ dτ ν(τ) cos (Ωτ) , (48c) 0 1 Z ∞ f ≡ − dτ ν(τ) sin (Ωτ) . (48d) MΩ 0 The first term on the right-hand side of Eq. (47) repre- sents the unitary dynamics of a harmonic oscillator whose natural frequency is shifted by Ω.e The second term de- scribes momentum damping (dissipation) at a rate pro- portional to γ, which depends only on the spectral den- sity but not the temperature of the environment. The third term is of the Lindblad double-commutator form [see Eq. (31)] and describes decoherence of spatial coher- ences over a distance ∆X at a rate D(∆X)2. Note that D depends on both the spectral density J(ω) and the temperature T of the environment. The fourth term also represents decoherence, but its influence on the dynam- ics of the system is usually negligible, especially at higher temperatures. In the long-time limit γt  1, the master equation (47) describes dispersion in position space given by

2 D ∆X (t) = t. (49) FIG. 1. Evolution of superpositions of Gaussian wave packets 2m2γ2 in quantum Brownian motion as studied in Ref. [28], visual- That is, the width ∆X(t) of the ensemble√ in position ized in the Wigner representation. Time increases from top space asymptotically scales as ∆X(t) ∝ t, just as in to bottom. In the left column, the initial wave packets are classical Brownian motion; hence the term “quantum separated in position; in the right column, the separation is Brownian motion.” in momentum. Figure1 shows the time evolution of position-space and momentum-space superpositions of two Gaussian wave equation [121], packets in the Wigner picture, as described by Eq. (47) [28]. Interference between the two wave packets is rep- d  0     ρS(t) = −i H , ρS(t) − iγ0 x, p, ρS(t) resented by oscillations between the direct peaks. The dt S    interaction with the environment damps these oscilla- − 2Mγ0kT x, x, ρS(t) , (51) tions. The damping occurs on different timescales for the two initial conditions. While the momentum coordi- where nate is not directly monitored by the environment, the 0 1 2 2 1 2 1  2  2 H = HS + MΩe x = p + M Ω − 2γ0Λ x intrinsic dynamics, through their creation of spatial su- S 2 2M 2 (52) perpositions from superpositions of momentum, result 0 in decoherence in momentum space. This interplay of is the frequency-shifted Hamiltonian HS of the system. environmental monitoring and intrinsic dynamics leads This equation has been widely and successfully used to to the emergence of pointer states that are minimum- model decoherence and dissipation processes, even in uncertainty Gaussians (coherent states) well-localized in cases where the assumptions were not strictly fulfilled both position and momentum, thus approximating clas- (for example, in quantum-optical settings, where often sical points in phase space [5,8, 16, 28, 44, 47, 48]. kT . Λ[122]). Let us consider the important case of an ohmic spectral In the position representation, the final term on the density J(ω) ∝ ω with a high-frequency cutoff Λ, right-hand side of Eq. (51) can be written as 2 2  − 0  2Mγ0 Λ x x 0 J(ω) = ω . (50) − γ0 ρS(x, x , t), (53) π Λ2 + ω2 λdB −1/2 In the limit of a high-temperature environment (kT  Ω where λdB = (2MkT ) is the thermal de Broglie wave- and kT  Λ), we arrive at the Caldeira–Leggett master length. This term describes spatial localization with a 13

−1 decoherence rate τ|x−x0| given by [18] The rich non-Markovian dynamics of this model have been analyzed in Refs. [29, 123]. The particular dynamics  0 2 −1 x − x strongly depend on the various parameters, such as the τ|x−x0| = γ0 . (54) λdB temperature of the environment, the form of the spec- tral density (subohmic, ohmic, or supraohmic), and the This is Eq. (22), and as discussed there, given that λdB is system–environment coupling strength. For each param- extremely small for macroscopic and even mesoscopic ob- eter regime, a characteristic dynamical behavior emerges: jects, we see that superpositions of macroscopically sepa- localization, exponential or incoherent relaxation, expo- rated center-of-mass positions will typically be decohered nential decay, and strongly or weakly damped coherent on timescales many orders of magnitude shorter than the oscillations [29]. −1 dissipation (relaxation) timescale γ0 . Over timescales In the weak-coupling limit, one can derive the Born– on the order of the decoherence time, we may therefore Markov master equation in much the same way as in often neglect the dissipative term in Eq. (51), leading to the case of quantum Brownian motion (note the similar the pure-decoherence master equation structure of the Hamiltonians). The result is (see Ref. [9] for details) d  0     ρS(t) = −i H , ρS(t) − 2Mγ0kT x, x, ρS(t) . (55) dt S d  0 0† ρS(t) = −i HSρS(t) − ρS(t)HS C. Spin–boson models dt ∗ − De [σz, [σz, ρS(t)]] + ζσzρS(t)σy + ζ σyρS(t)σz. (58) In the spin–boson model, a qubit interacts with an environment of harmonic oscillators. The seminal review The first term on the right-hand side of the master equa- tion (58) represents the evolution under the environment- paper by Leggett et al. [29] discusses the dynamics of the 0 spin–boson model in great detail. shifted self-Hamiltonian HS, the second term corre- Let us first consider a simplified spin–boson model sponds to decoherence in the σz eigenbasis of the system at a rate given by De, and the last two terms describe where the self-Hamiltonian of the system is taken to be 0 1 the decay of the two-level system. HS is the renormal- HS = 2 ω0σz, with eigenstates |0i and |1i. In contrast with the more general case discussed below, this Hamilto- ized (and in general non-Hermitian) Hamiltonian of the 1 system. The coefficients ζ∗, D, f, and γ are given by nian does not include a tunneling term − 2 ∆0σx, and thus e e e HS does not generate any nontrivial intrinsic dynamics. ∗ We employ the familiar self-Hamiltonian, Eq. (41), for ζ = fe− iγ,e (59a) an environment of harmonic oscillators, and choose the Z ∞ P bilinear interaction Hamiltonian Hint = σz ⊗ i ciqi. Us- De = dτ ν(τ) cos (∆0τ) , (59b) ing the raising and lowering operators a† and a, we can 0 Z ∞ recast the total Hamiltonian as fe = dτ ν(τ) sin (∆0τ) , (59c) 0 1 X † X  † ∗  H = ω0σz + ωia ai + σz ⊗ gia + g ai . (56) Z ∞ 2 i i i i i γe = dτ η(τ) sin (∆0τ) , (59d) 0   Note that since H, σz = 0, no transitions between |0i and |1i can be induced by H. There is no energy ex- with the noise and the dissipation kernels ν(τ) and η(τ) change between the system and the environment, and we taking the same form as in quantum Brownian motion therefore deal with a model of decoherence without dis- [see Eqs. (43) and (44)]. sipation. Such a model is a good representation of rapid decoherence processes during which the amount of dissi- pation is negligible, as is often the case in physical appli- D. Spin-environment models cations. The resulting evolution can be solved exactly [9]. For an ohmic spectral density with a high-frequency cut- off, it is found that superpositions of the form α |0i+β |1i A qubit linearly coupled to a collection of other are exponentially decohered on a timescale set by the qubits—known also as a spin–spin model—is often a good thermal correlation time (kT )−1 of the environment. model of a single two-level system, such as a supercon- 1 ducting qubit, strongly coupled to a low-temperature en- Inclusion of a tunneling term − 2 ∆0σx yields the gen- eral spin–boson model defined by the Hamiltonian vironment [103, 104]. The model of a harmonic oscil- lator interacting with a spin environment may be rele-   1 1 X 1 2 1 2 2 vant to the description of decoherence and dissipation in H = ω0σz − ∆0σx + pi + miωi qi 2 2 2mi 2 quantum-nanomechanical systems and micron-scale ion i traps [124]. For details on the theory of spin-environment X + σz ⊗ ciqi. (57) models, see Refs. [104, 125–127]. i A simple version of a spin–spin model is described by 14 the total Hamiltonian mitigation of the effects of decoherence through active quantum error correction [132–136]. N 1 1 X (i) We may distinguish two limiting cases for modeling H = HS + Hint = − ∆0σx + σz ⊗ giσ 2 2 z decoherence in qubits. The first limit is that of indepen- i=1 dent qubit decoherence. Here, each qubit couples indepen- 1 1 ≡ − ∆0σx + σz ⊗ E. (60) dently to its own environment, without any interactions 2 2 between these environments. For example, this may be the case if the qubits are spatially well-separated (rela- Here, HS represents the intrinsic dynamics given by a tive to the typical coherence length of the environment) tunneling term, while Hint describes the environmental and only couple to their immediate surroundings. Then monitoring of the observable σz. The model can be solved exactly [128, 129], and the error processes affecting the qubits will be completely the resulting dynamics illustrate the dependence of the uncorrelated. Thus, if the probability of a particular er- ror to affect one qubit is p, the probability of this error preferred basis on the relative strengths of the self- K Hamiltonian of the system and the interaction Hamil- to occur in K qubits will be p . Many error-correcting tonian. The preferred basis emerges as the local ba- schemes are only efficient in correcting such single-qubit sis that is most robust under the total Hamiltonian. errors, and thus the assumption of independent decoher- When the interaction Hamiltonian dominates over the ence frequently underlies these schemes. This assump- self-Hamiltonian, the pointer states are found to be eigen- tion, however, is unrealistic when the qubits are located states of the interaction Hamiltonian, in agreement with spatially close to each other. In this case, all qubits ap- the commutativity criterion, Eq. (18). Conversely, when proximately feel the same environment, and it is likely the modes of the environment are slow and the self- that errors will become correlated among multiple qubits. Hamiltonian dominates the evolution of the system (the The limiting case corresponding to this situation is that quantum limit of decoherence [42]), the pointer states are of collective qubit decoherence, in which all qubits couple the eigenstates of the Hamiltonian of the system. to exactly the same environment. In the weak-coupling limit, spin environments can be mapped onto oscillator environments [110, 130]. Specifi- A. Correction of decoherence-induced quantum cally, the reduced dynamics of a system weakly coupled errors to a spin environment can be described by the system coupled to an equivalent oscillator environment described by an explicitly temperature-dependent spectral density Consider a single qubit S, initially described by a pure of the form state |ψi and interacting with an environment E. One can show that an arbitrary evolution of the combined  ω  qubit–environment state can always be written in the Jeff(ω, T ) ≡ J(ω) tanh , (61) 2kT form X where J(ω) is the original spectral density of the spin |ψi |e0i −→ I |ψi |eI i + (σs |ψi) |esi , (62) environment. (See Sec. 5.4.2 of Ref. [9] for details and s=x,y,z examples.) where the Pauli operators σs act on the Hilbert space of S, and |eI i and {|esi} are environmental states that are V. QUBIT DECOHERENCE, QUANTUM not necessarily orthogonal or normalized. Thus, any in- ERROR CORRECTION, AND ERROR fluence of the environment on the qubit can be expressed AVOIDANCE simply in terms of a weighted sum of the Pauli operators and the identity operator acting on the original state of Quantum computation and quantum information pro- the qubit. The effects of σx and σz on the qubit state are cessing rely on coherent superpositions of mesoscopically often referred as a bit-flip error and phase-flip error, re- or macroscopically distinct states that are highly suscep- spectively. If we restrict our attention to environmental tible to decoherence. Avoiding, controlling, and mitigat- entanglement and the resulting decoherence effects, then ing decoherence is therefore of paramount importance. only phase-flip errors need to be taken into account. While the qubits need to be protected from detrimental For N qubits, Eq. (62) generalizes to environmental interactions, we also need to be able to X |ψi |e i −→ (E |ψi) |e i . (63) control and measure them via a macroscopic apparatus. 0 i i i The formidable challenge of designing a quantum com- puter consists of meeting both demands in a balanced Here |ψi is the initial N-qubit state, and the error op- way. Even so, decoherence induced by interactions with erators Ei are tensor products of N operators involv- the environment and the control apparatus, as well as ing identity and Pauli operators. Equation (63) repre- noise due to faulty gate operations, will likely be too sents a worst-case scenario. In many cases, simplified strong to allow for useful quantum computations to be versions can be used. One important case is that of par- carried out [74, 131]. What is also needed is an active tial decoherence. Here, only a small number K < N of 15 qubits become entangled with the environment between B. Quantum computation on decoherence-free two successive applications of an error-correcting mech- subspaces anism. Then it will be sufficient to restrict our attention to the 2K possible error operators made up of at most K We introduced the concept of decoherence-free sub- operators σz and N − K identity operators. In the case spaces (DFS) [49–58], or pointer subspaces [3], in of independent qubit decoherence, we only need to con- Sec. IIC2. DFS allow us to encode quantum informa- sider a collection of independent phase-flip errors acting tion in “quiet corners” of the Hilbert space to protect on single qubits, represented by error operators of the it from environmental effects. In contrast with quantum form E = I ⊗ · · · ⊗ I ⊗ σz ⊗ I ⊗ · · · ⊗ I. error correction, DFS prevent errors from happening in Given the entangled state on the right-hand side of the first place and thus represent a strategy for intrinsic Eq. (63), the goal of quantum error correction is to re- error avoidance. store the initial (unknown) state |ψi. We let an ancilla, The two limiting cases of independent qubit decoher- described by an initial state |a0i, interact with the qubit ence and collective qubit decoherence delineate the lim- system such that its on the size of a DFS. To illustrate this relation- " # ship, let us consider the case of collective decoherence X X |a0i (Ei |ψi) |eii −→ |aii (Ei |ψi) |eii . (64) of an N-qubit system interacting with an oscillator bath i i [49, 51, 53, 56, 137]. The interaction Hamiltonian for this generalized spin–boson model is taken to be [com- Let us assume that the ancilla states |a i are at least i pare Eq. (56)] approximately mutually orthogonal, such that they can be distinguished by measurement. We now measure the N N P X (i) X  † ∗  X (i) observable OA = i ai|aiihai| on the ancilla, with ai =6 Hint = σz ⊗ gijaj + gijaj ≡ σz ⊗ Ei. aj for i =6 j. The projective measurement will yield a i=1 j i=1 particular outcome, say, ak, and lead to the reduction of (67) the entangled state, The assumption of collective decoherence implies that X the couplings gij (and thus the environment operators | i | i | i −→ | i | i | i ai (Ei ψ ) ei ak (Ek ψ ) ek . (65) Ei) must be independent of the index i. Then Eq. (67) i becomes The outcome ak of the measurement tells us the counter- ! X (i) transformation needed to restore the initial qubit state. Hint = σz ⊗ E ≡ Sz ⊗ E. (68) −1 † Applying Ek = Ek to the system gives i

−1 Ek Recall that a DFS is spanned by a degenerate set of |aki (Ek |ψi) |eki −−−→|aki |ψi |eki . (66) eigenstates of the system operators Sα of the interaction Note that, as required in order to avoid introducing ad- Hamiltonian [see Eq. (20)]. Thus, in our case the DFS ditional decoherence in the computational basis of the will be spanned by degenerate eigenstates of the collec- qubit system, we have obtained no information whatso- tive spin operator Sz. Any N-qubit product state of the ever about the state of the system. computational basis states |0i and |1i (the eigenstates of This account of quantum error correction has been σz with eigenvalues +1 and −1, respectively) will be an highly idealized. Let us mention three complications. eigenstate of Sz. There are 2N +1 different possible inte- First, it is impossible to design an interaction between ger eigenvalues m, ranging from m = −N (corresponding the computational qubits and the ancilla that would al- to the basis state |1 ··· 1i) to m = +N (corresponding low us to distinguish, by measuring the ancilla, between to |0 ··· 0i). The largest number of mutually orthogonal all possible errors. Second, in realistic settings the error computational-basis states with the same eigenvalue m operators Ei may be very complex, and it remains to be of Sz is given by the set S0 of basis states with m = 0, seen whether and how the corresponding countertrans- i.e., those with N/2 qubits in the state |0i. There are N
formations can be applied without introducing signifi- n0 = N/2 such states in this set, spanning a DFS of di- cant additional decoherence. Third, the ancilla qubits mension n0. For large values of N, we can approximate are physically similar to the computational qubits and the binomial coefficient using Stirling’s formula, can therefore be expected to be equally prone to en-   vironmental interactions (and thus decoherence) as the N 1 N1 log2 ≈ N − log2(πN/2) −−−→ N. (69) computational qubits themselves. Since the inclusion of N/2 2 ancilla qubits increases the total number of qubits in the Therefore, in the limiting case of collective decoherence, quantum computer, and since decoherence rates typically the dimension of our DFS approaches the dimension of scale exponentially with the size of the system, it will re- the original Hilbert space, and the encoding efficiency quire sophisticated experimental designs to ensure not approaches unity. For example, for N = 4 qubits, the set only that quantum error correction works in practice, but also that it does not aggravate the problem of qubit S0 = { |0011i , |0101i , |0110i , |1001i , |1010i , |1100i } decoherence. (70) 16 spans a maximum-size DFS of dimension six, to be com- One way of overcoming this limitation is based on envi- pared with the dimension of the original Hilbert space, ronment engineering. Here, one tries to generate certain which is 24 = 16. Thus, given the model for collective de- symmetries in the structure of the system–environment coherence considered here, using four physical qubits we interactions. For example, an appropriately engineered can encode up to two logical qubits in a DFS (since en- symmetrization could make superposition states in Bose– coding three logical qubits would already require a DFS Einstein condensates correspond to (approximate) de- of dimension 23 = 8). generate eigenstates of the interaction Hamiltonian, in As mentioned in Sec. IIC2, the existence of a DFS which case such states would lie within a DFS, thereby corresponds to a dynamical symmetry. Our model rep- significantly enhancing their longevity [140]. In ion resents a case of perfect dynamical symmetry, since traps, changing the parameters in the effective interac- the system–environment interaction, Eq. (68), is com- tion Hamiltonian for the trapped ion allows one to se- pletely symmetric with respect to any permutations of lect different pointer subspaces and thereby control into the qubits, thereby leading to a DFS of maximum size. which DFS the trapped ion is driven [77, 79, 141, 142]. What happens if the symmetry is broken by additional Another approach to the active creation of DFS is small independent coupling terms? It has been shown known as dynamical decoupling [143–148]. Here time- [50, 138] that, to first order in the perturbation strength, dependent modifications are introduced into the Hamil- the storage of quantum information in DFS is stable to tonian of the system that counteract the influence of such perturbations to all orders in time, but that the pro- the environment. These modifications take the form of cessing of such quantum information encoded in DFS is sequences of rapid projective measurements or strong robust only to first order in time. control-field pulses acting on the system (“quantum In the case of purely independent qubit decoherence, bang-bang control” [143]). Even if the structure of the environment operators Ei appearing in Eq. (67) will the system–environment interaction Hamiltonian is not now differ from one another. To find a DFS, we follow known, decoherence can be suppressed arbitrarily well the usual strategy [see Eq. (20)] of determining a set of in the limit of an infinitely fast rate of the decoupling orthonormal basis states {|sii} such that control field, thus dynamically creating a DFS (which then represents a dynamically decoupled subspace). In h i the realistic case of a finite control rate, sufficient (albeit I(1) ⊗ · · · ⊗ I(j−1) ⊗ σ(j) ⊗ I(j+1) ⊗ · · · ⊗ I(N) |s i z i imperfect) protection from decoherence can be achieved (j) = λ |sii (71) via this decoupling technique, provided the control rate is larger than the fastest timescale set by the rate of for- for all i and 1 ≤ j ≤ N. The only state fulfilling this mation of environmental entanglement. eigenvalue problem is |0 ··· 0i. Since we need at least a two-dimensional subspace to encode a single logical qubit, the case of independent decoherence in the spin– VI. EXPERIMENTAL STUDIES OF boson model does not allow for the existence of a DFS for DECOHERENCE quantum computation. In the language of pointer sub- spaces, there is only a single exact pointer state, and this Decoherence, of course, happens all around us, and environment-superselected preferred state of the system in this sense its consequences are readily observed. But will be the ground state |0 ··· 0i. what we would like to do is to be able to experimen- In realistic settings, neither the assumption of purely tally study the gradual and controlled action of deco- independent decoherence nor the limit of entirely collec- herence. In this endeavor, several obstacles have to be tive decoherence will be entirely appropriate. We can, overcome. We need to prepare the system in a superpo- however, use a DFS to protect the qubits from collective sition of mesoscopically or even macroscopically distin- decoherence effects, and we can recover from single-qubit guishable states with a sufficiently long decoherence time errors due to independent decoherence using active error- such that the gradual action of decoherence can be re- correction methods. These two approaches can be con- solved. We must be able to monitor decoherence without catenated [54] to enable universal fault-tolerant quantum introducing a significant amount of additional, unwanted computation even when the restriction to single-qubit er- decoherence. We would also like to have sufficient con- rors is dropped [55, 139]. trol over the environment so we can tune the strength and form of its interaction with the system. Starting in the mid-1990s, several such experiments have been C. Environment engineering and dynamical performed, for example, using cavity QED [12], meso- decoupling scopic molecules [149], and superconducting systems such as SQUIDs and Cooper-pair boxes [13]. Bose–Einstein For reasonably large DFS to exist, the system– condensates [150] and quantum nanomechanical systems environment interaction must exhibit a sufficiently high [151, 152] are promising candidates for future experimen- degree of symmetry. Such symmetries are unlikely to tal tests of decoherence. arise naturally in typical experimental settings. These experiments are important for several reasons. 17

They are impressive demonstrations of the possibility ment correlation as a function of the wait time τ between of generating nonclassical quantum states in mesoscopic sending the first and second atom through the appara- and macroscopic systems. They show that the quantum– tus, the decoherence of the field state can be monitored. classical boundary is smooth and can be shifted by vary- Experimental results were in excellent agreement with ing the relevant experimental parameters. They allow theoretical predictions [158, 159]. It was found that de- us to test and improve decoherence models, and they coherence became faster as the phase shift φ and the help us design devices for quantum information process- mean numbern ¯ = |α|2 of photons in the cavity C was ing that are good at evading the detrimental influence increased. Both results are expected, since an increase of the environment. Finally, such experiments may be in φ andn ¯ means that the components in the superpo- used to test quantum mechanics itself [13]. Such tests re- sition become more distinguishable. Recent experiments quire sufficient shielding of the system from decoherence have realized superposition states involving several tens so that an observed (full or partial) collapse of the wave- of photons [160] and have monitored the gradual deco- function could be unambigously attributed to some novel herence of such states [161]. nonunitary mechanism in nature, such as those proposed in dynamical reduction models [153–155]. This shielding, however, is difficult to implement in practice, because B. Matter-wave interferometry the large number of particles required for the reduction mechanism to become effective will also lead to strong In these experiments (see Ref. [11] for a review), spatial decoherence [114, 156]. The superpositions realized in interference patterns are demonstrated for mesoscopic current experiments are still not sufficiently macroscopic molecules ranging from fullerenes [162] to molecular clus- to rule out collapse theories, although it has been demon- ters involving hundreds of atoms, with a total size of strated [118] that matter-wave interferometry with large up to 60 A˚ and masses of several thousand amu (see 6 molecular clusters (in the mass range between 10 and Fig.2)[163, 164]. Since the de Broglie wavelength of 8 10 amu) would be able to test the collapse theories pro- such molecules is on the order of picometers, standard posed in Refs. [154, 155]; such experiments may soon double-slit interferometry is out of reach. Instead, the become technologically feasible [11]. experiments make use of the Talbot effect, an interfer- ence phenomenon in which a plane wave incident on a diffraction grating creates an image of the grating at mul- A. Atoms in a cavity tiples of a distance L behind the grating. In the experi- ment, the molecular density (at a macroscopic distance L In 1996 Brune et al. generated a superposition of ra- from the grating) is scanned along the direction perpen- diation fields with classically distinguishable phases in- dicular to the molecular beam. An oscillatory density volving several photons [12, 150, 157]. This experiment pattern (corresponding to the image of the slits in the was the first to realize a mesoscopic Schr¨odinger-cat state grating) is observed, confirming the existence of coher- and allowed for the controlled observation and manipu- ence and interference between the different paths of each lation of its decoherence. A rubidium atom is prepared individual molecule passing through the grating. Recent in a superposition of energy eigenstates |gi and |ei cor- experiments have used an improved version of the origi- responding to two circular Rydberg states. The atom nal Talbot–Lau setup [165], as well as optical ionization enters a cavity C containing a radiation field contain- gratings [166]. ing a few photons. If the atom is in the state |gi, the Decoherence is measured as a decrease of the visibil- field remains unchanged, whereas if it is in the state ity of this pattern (Fig.2). The controlled decoherence |ei, the coherent state |αi of the field undergoes a phase due to collisions with background gas particles [115, 116] iφ shift φ, |αi −→ e α ; the experiment achieved φ ≈ π. and due to emission of thermal radiation from heated An initial superposition of the atom is therefore am- molecules [168] has been observed, showing a smooth de- plified into an entangled atom–field state of the form cay of visibility in agreement with theoretical predictions √1 | i | i | i |− i [31, 117, 167]. These successes have led to speculations 2 ( g α + e α ). The atom then passes through an additional cavity, further transforming the superposi- that one could perform similar experiments using even tion. Finally, the energy state of the atom is measured. larger particles such as proteins and viruses [115, 169] This disentangles the atom and the field and leaves the or carbonaceous aerosols [170]. Such experiments will be latter in a superposition of the mesoscopically distinct limited by collisional and thermal decoherence and by states |αi and |−αi. noise due to inertial forces and vibrations [115, 169, 170]. To monitor the decoherence of this superposition, a second rubidium atom is sent through the apparatus. Af- ter interacting with the field superposition state in the C. Superconducting systems cavity C, the atom will always be found in the same en- ergy state as the first atom if the superposition has not Superconducting quantum interference devices been decohered. This correlation rapidly decays with in- (SQUIDs) and Cooper-pair boxes have important creasing decoherence. Thus, by recording the measure- applications in quantum information processing. A ARTICLE NATURE COMMUNICATIONS | DOI: 10.1038/ncomms1263 18

(a) n many discussions on the foundations of physics, single-particle (a) (b) 1–4 5–12 di!raction at a double slit or grating is regarded as a para- ! 80% Idigmatic example for a highly non-classical feature of quantum supercurrent mechanics, which has never been observed for objects of our mac- 60% roscopic world. "e quantum superposition principle has become superconducting of paramount importance also for the growing #eld of quantum ring 40% information science13. Correspondingly, research in many labora-

tories around the world is focusing on our understanding of the Josephson junction probability for 0 5 10 15 20 25 30 35 role of decoherence for increasingly complex quantum systems and delay time τ (ns) possible practical or truly fundamental limits to the observation of quantum dynamics14,15. (b) Here we report on a new leap in quantum interference with large organic molecules. In contrast to earlier successful experi- (a) (b) 16 ments with internal molecular wave packets, our study focuses on ! 80% the wave evolution in the centre of mass motion of the molecule supercurrent as a whole, that is, pure de Broglie interference. We do this with 60% compounds that have been customized to provide useful molecu- superconducting lar beams at moderate temperatures17,18. Figure 1 compares the size ring 40% of two per$uoroalkylated nanospheres, PFNS8 and PFNS10, with 19 visibility (%) a single C60 fullerene and it relates a single tetraphenylporphyrin Figure 1 | Gallery of molecules used in our interference study. (a) The Josephson junction probability for 0 5 10 15 20 25 30 35 molecule (TPP) to its complex derivatives TPPF84 and TPPF152. fullerene C (m = 720 AMU, 60 atoms) serves as a size reference and 60 delay time τ (ns) We demonstrate the wave nature of all these molecules in a three- for calibration30 purposes; (b) The perfluoroalkylated nanosphere PFNS8 20,21 grating near-#eld interferometer of the Kapitza-Dirac-Talbot- (C60[C12F25]8, m = 5,672 AMU, 356 atoms) is a carbon cage with eight 22,23 Lau type , as shown in Figure 2. perfluoroalkyl20 chains. (c) PFNS10 (C60[C12F25]10, m = 6,910 AMU, 430 atoms) has ten side chains and is the most massive particle in the set. FIG. 3. (a) Schematic illustration of a SQUID. A supercon- Results (d) A single tetraphenylporphyrin TPP (C44H30N4, m = 614 AMU, 78 ducting ring is interrupted by Josephson junctions, leading Experimental setup. "e particles are evaporated in a thermal atoms) is the basis for the two derivatives (e) TPPF84 (C H F N S , 10 84 26 84 4 4 to a dissipationless supercurrent. (b) Decoherence of a su- source. "eir velocity is selected using the gravitational free-fall m = 2,814 AMU, 202 atoms) and (f) TPPF152 (C168H94F152O8N4S4, through a sequence of three slits. "e interferometer itself consists m = 5,310 AMU8 , 430 atoms). In its unfolded configuration, the latter is the perposition of clockwise and counterclockwise supercurrents in a superconducting qubit. The damping of the oscillation of three gratings G1, G2 and G3 in a vacuum chamber at a pressure largest molecule in the set. Measured by the number of atoms, TPPF152 − 8 6 of p < 10 mbar. "e #rst grating is a SiNx membrane with 90-nm and PFNS10 are equally complex. All molecules are displayed to scale. The amplitude corresponds to the gradual loss of coherence from wide slits arranged with a periodicity of d = 266 nm. Each slit of G1 scale bar corresponds to 10 Å. the system. Figure adapted from Ref. [173]. imposes a constraint onto the transverse molecular position that, 4 following Heisenberg’s uncertainty relation, leads to a momentum y 0 0.4 0.8Detector 1.2 1.6 6 uncertainty. "e latter turns into a growing delocalization and pressure (in 10− mbar) transverse coherence of the matter wave with increasing distance X that the two lowest-lying energy eigenstates |0i and |1i from G1. "e second grating, G2, is a standing laser light wave with a are equal-weight superpositions of the persistent-current wavelength of L = 532 nm. "e interaction between the electric laserFIG. 2. Left:Z Molecular clustersG3 used in recent interference light #eld and the molecular optical polarizability creates a sinusoidalexperiments, drawn to scale (the scale bar represents 10 A).˚ states |i and | i. G2 potential, which phase-modulates the incident matter waves. "Figuree from Ref. [163]. (a) Fullerene C60 (m = 720 amu, Such superposition states involving µA currents were distance between the #rst two gratings is chosen such that quantum 60 atoms). (b) PerfluoroalkylatedG1 nanosphereS3 PFNS8 (m = first experimentally observed in 2000 using spectroscopic interference leads to the formation of a periodic molecular density5672 amu, 356 atoms). (c) PFNS10 (m = 6910 amu, 430 measurements [171, 172]. Their decoherence was subse- pattern 105 mm behind G . "is molecular nanostructure is sampled 2 atoms). (d) Tetraphenylporphyrin TPP (m = 614 amu, 78 by scanning a second SiNx grating (G3, identical to G1) across the Lens quently measured using Ramsey interferometry [173], as molecular beam while counting the number of the transmittedatoms). (e) TPPF84 (m S=2 2814 amu, 202 atoms). (f) follows. Two consecutive microwave pulses are applied to particles in a quadrupole mass spectrometer (QMS). TPPF152 (m = 5310 amu, 430 atoms). Right: Visibility of the system. During the delay time τ between the pulses, In extension to earlier experiments, we have added various tech- S interference fringes1 of C70 fullerenes as a function of the pres- the system evolves freely. After application of the second nological re#nements: the oven was adapted to liquid samples,sure a of the background gas. Measured values (circles) agree Oven Laser pulse, the system is left in a superposition of | i and | i, liquid-nitrogen-cooled chamber became essential to maintain thewell with the theoretical prediction (solid line) [31, 117, 167]  source pressure low, a new mass analyser allowed us to increase with the relative amplitudes exhibiting an oscillatory de- the detected molecular $ux by a factor of four and many optimidescribing- an exponential decay of visibility with pressure. Figure 2 | Layout of the Kapitza-Dirac-Talbot-Lau (KDTL) interference pendence on τ. A series of measurements in the basis zation cycles in the interferometer alignment were needed to meetFigure adapted from Ref. [115]. experiment. The effusive source emits molecules that are velocity-selected all requirements for high-contrast experiments with very massive {| i , | i} over a range of delay times τ then allows one by the three delimiters S , S and S . The KDTL interferometer is composed  particles. 1 2 3 to trace out an oscillation of the occupation probabilities of two SiNx gratings G1 and G3, as well as the standing light wave G2. The optical dipole force grating imprints a phase modulation J(x)sA ·P/(v·w ) for | i and | i as a function of τ (Fig.3b). The envelope Observed interferograms. We recorded quantum interferograms opt y  onto the matter wave. Here A is the optical polarizability, P the laser for all molecules of Figure 1, as shown in Figure 3. In all cases theSQUID consists of a ringopt of superconducting material of the oscillation is damped as a consequence of decoher- power, v the molecular velocity and w the laser beam waist perpendicular measured fringe visibility V, that is, the amplitude of the sinusoidalinterrupted by thin insulatingy barriers, the Josephson ence acting on the system during the free evolution of to the molecular beam. The molecules are detected using electron impact modulation normalized to the mean of the signal, exceeds the maxi- junctionsionization (Fig. and quadrupole3a). Atmass sufficientlyspectrometry. low temperatures, duration τ. From the decay of the envelope we can infer mally expected classical moiré fringe contrast by a signi#cant multi- the decoherence timescale; the original experiment gave ple of the experimental uncertainty. "is is best shown for TPPF84electrons of opposite spin condense into bosonic Cooper and PFNS8, which reached the highest observed interference conpairs.- for TPPF152 Quantum-mechanical (see Figure 3), in which tunneling our classical of model Cooper predicts pairs 20 ns [173], while subsequent experiments have achieved trast in our high-mass experiments so far, with individual scansthrough Vclass = 1%. the " junctionsis supports our leads claim toof true the quantum flow of interference a resistance- for decoherence times of several µs [174]. up to Vobs = 33% for TPPF84 (m = 2,814 AMU) and Vobs = 49% forfree all supercurrent these complex molecules. around the loop (Josephson effect), Superpositions states and their decoherence have also PFNS8 at a mass of m = 5,672 AMU. In addition, we have observed "e most massive molecules are also the slowest and therefore which creates a magnetic flux threading the loop. The been observed in superconducting devices whose key vari- a maximum contrast of Vobs = 17 o 4% for PFNS10 and Vobs = 16 o 2% the most sensitive ones to external perturbations. In our particle collective center-of-mass motion of a macroscopic num- able is charge (or phase), instead of the flux variable used
 NATURE COMMUNberICATI (∼ ONS10 9| 2:263) of | D CooperOI: 10.1038/ncomms1263 pairs can | www.nature.com/naturecommunications then be represented by in SQUIDs. Such Cooper-pair boxes consist of a small © 2011 Macmillan Publishersa wave Limited. function All rights reserved. labeled by a single macroscopic variable, superconducting island onto which Cooper pairs can tun- namely, the total trapped flux Φ through the loop. nel from a reservoir through a Josephson junction. Two The two possible directions of the supercurrent define different charge states of the island, differing by at least a qubit with basis states {|i , | i}. By adjusting an one Cooper pair, define the basis states. Coherent os- external magnetic field, the SQUID can be biased such cillations between such charge states were first observed 19 in 1999 [175]. In 2002, Vion et al. [176] reported thou- rule. If we understand the “quantum measurement prob- sands of coherent oscillations with a decoherence time lem” as the question of how to reconcile the linear, de- of 0.5 µs. Similar results have been obtained for phase terministic evolution described by the Schr¨odingerequa- qubits [177, 178], demonstrating decoherence times of tion with the occurrence of random measurement out- several µs. comes, then decoherence has not solved this problem [6,9]. Decoherence does, however, address an aspect sometimes associated with the quantum measurement VII. DECOHERENCE AND THE problem, namely the preferred-basis problem (at least FOUNDATIONS OF QUANTUM MECHANICS in the sense described in Sec.IIC). Further explorations of the role of the environment, such as in quantum Dar- Can decoherence address foundational problems? If so, winism (see Sec.IID), can help illuminate fundamental which ones, and how? Addressing these subtle questions questions concerning information transfer and amplifica- is beyond the scope of this review; a few brief remarks tion in the quantum setting. must suffice here. (See Refs. [6,7,9, 21] for in-depth Decoherence has been used to identify internal con- discussions.) Decoherence, at its heart, is a technical re- cistency issues in interpretations of quantum mechanics, sult concerning the dynamics and measurement statistics and the picture associated with the decoherence process of open quantum systems. From this view, decoherence has sometimes been seen as suggestive of particular inter- merely addresses a consistency problem, by explaining pretations of quantum mechanics [6,7]. Indeed, histori- how and when the quantum probability distributions ap- cally decoherence theory arose in the context of Zeh’s [1] proach the classically expected distributions. Since deco- independent formulation of an Everett-style interpreta- herence follows directly from an application of the quan- tion (see Ref. [179] for a historical analysis). Ultimately, tum formalism to interacting quantum systems, it is not however, it seems that certain interpretations simply may tied to any particular interpretation of quantum mechan- be more in need of decoherence than others for defin- ics, nor does it supply such an interpretation, nor does it ing their structure; see Ref. [180] for the example of an amount to a theory that could make predictions beyond Everett-style interpretation [23]. At the end of the day, those of standard quantum mechanics. any interpretation that does not involve entities, claims, The predictively relevant part of decoherence theory or structures in contradiction with the predictions of de- relies on reduced density matrices, whose formalism and coherence theory (which is to say, with the predictions of interpretation presume the collapse postulate and Born’s quantum mechanics) will arguably remain viable.

[1] H.D. Zeh, Found. Phys. 1, 69 (1970) 73, 565 (2001) [2] W.H. Zurek, Phys. Rev. D 24, 1516 (1981) [13] A.J. Leggett, J. Phys.: Condens. Matter 14, R415 [3] W.H. Zurek, Phys. Rev. D 26, 1862 (1982). doi: (2002) 10.1103/PhysRevD.26.1862 [14] E. Joos, in Decoherence: Theoretical, Experimental, and [4] J.P. Paz, W.H. Zurek, in Coherent Atomic Matter Conceptual Problems, ed. by P. Blanchard, D. Giulini, Waves, Les Houches Session LXXII, Les Houches Sum- E. Joos, C. Kiefer, I.O. Stamatescu (Springer, Berlin, mer School Series, vol. 72, ed. by R. Kaiser, C. West- 2000), pp. 1–17 brook, F. David (Springer, Berlin, 2001), Les Houches [15] M. Schlosshauer, K. Camilleri, AIP Conf. Proc. 1327, Summer School Series, vol. 72, pp. 533–614 26 (2011) [5] W.H. Zurek, Rev. Mod. Phys. 75, 715 (2003). doi: [16] O. K¨ubler, H.D. Zeh, Ann. Phys. (N.Y.) 76, 405 (1973) 10.1103/RevModPhys.75.715 [17] E. Joos, H.D. Zeh, Z. Phys. B: Condens. Matter 59, 223 [6] M. Schlosshauer, Rev. Mod. Phys. 76, 1267 (2004) (1985) [7] G. Bacciagaluppi, in The Stanford Encyclopedia [18] W.H. Zurek, in Frontiers of Nonequilibrium Statistical of Philosophy, ed. by E.N. Zalta (2012). Online at Mechanics, ed. by G.T. Moore, M.O. Scully (Plenum http://plato.stanford.edu/archives/win2012/entries/qm- Press, New York, 1986), pp. 145–149. First published decoherence in 1984 as Los Alamos report LAUR 84-2750 [8] E. Joos, H.D. Zeh, C. Kiefer, D. Giulini, J. Kupsch, I.O. [19] K. Hornberger, in Entanglement and Decoherence: Stamatescu, Decoherence and the Appearance of a Clas- Foundations and Modern Trends, Lecture Notes in sical World in Quantum Theory, 2nd edn. (Springer, Physics, vol. 768, ed. by A. Buchleitner, C. Viviescas, New York, 2003) M. Tiersch (Springer, Berlin, 2009), pp. 221–276 [9] M. Schlosshauer, Decoherence and the Quantum- [20] H.P. Breuer, F. Petruccione, The Theory of Open Quan- to-Classical Transition (Springer, Berlin/Heidelberg, tum Systems (Oxford University Press, Oxford, 2002) 2007) [21] M. Schlosshauer, A. Fine, in Quantum Mechanics at the [10] M. Schlosshauer, Phys. Rep. 831, 1 (2019). doi: Crossroads: New Perspectives from History, Philosophy 10.1016/j.physrep.2019.10.001 and Physics, ed. by J. Evans, A. Thorndike (Springer, [11] K. Hornberger, S. Gerlich, S. Nimmrichter, P. Haslinger, Berlin, 2006), pp. 125–148 M. Arndt, Rev. Mod. Phys. 84, 157 (2012) [22] W.K. Wootters, W.H. Zurek, Phys. Rev. D 19, 473 [12] J.M. Raimond, M. Brune, S. Haroche, Rev. Mod. Phys. (1979) 20

[23] H. Everett, Rev. Mod. Phys. 29, 454 (1957) by F. Benatti, R. Floreanini (Springer, Berlin, 2003), [24] A. Einstein, B. Podolsky, N. Rosen, Phys. Rev. 47, 777 pp. 83–120. Also available as eprint quant-ph/0301032 (1935) [61] W.H. Zurek, in Science and Ultimate Reality, ed. by [25] J.S. Bell, Physics 1, 195 (1964) J.D. Barrow, P.C.W. Davies, C.H. Harper (Cambridge [26] J.S. Bell, Rev. Mod. Phys. 38, 447 (1966) University Press, Cambridge, England, 2004), pp. 121– [27] A. Peres, Am. J. Phys. 46 (1978) 137 [28] J.P. Paz, S. Habib, W.H. Zurek, Phys. Rev. D 47, 488 [62] H. Ollivier, D. Poulin, W.H. Zurek, Phys. Rev. Lett. 93, (1993) 220401 (2004) [29] A.J. Leggett, S. Chakravarty, A.T. Dorsey, M.P.A. [63] H. Ollivier, D. Poulin, W.H. Zurek, Phys. Rev. A 72, Fisher, A. Garg, Rev. Mod. Phys. 59, 1 (1987) 042113 (2005) [30] S.G. Mokarzel, A.N. Salgueiro, M.C. Nemes, Phys. Rev. [64] R. Blume-Kohout, W.H. Zurek, Found. Phys. 35, 1857 A 65, 044101 (2002) (2005) [31] K. Hornberger, J.E. Sipe, Phys. Rev. A 68, 012105 [65] R. Blume-Kohout, W.H. Zurek, Phys. Rev. A 73, (2003) 062310 (2006) [32] K. Kraus, States, Effects, and Operations (Springer, [66] W.H. Zurek, Nature Phys. 5, 181 (2009) Berlin, 1983) [67] C.J. Riedel, W.H. Zurek, Phys. Rev. Lett. 105, 020404 [33] W.H. Zurek, Phys. Today 44, 36 (1991). See also the (2010) updated version available as eprint quant-ph/0306072 [68] C.J. Riedel, W.H. Zurek, New J. Phys. 13, 073038 [34] M.R. Gallis, G.N. Fleming, Phys. Rev. A 42, 38 (1990) (2011) [35] L. Di´osi,Europhys. Lett. 30, 63 (1995) [69] C.J. Riedel, W.H. Zurek, M. Zwolak, New J. Phys. 14, [36] K. Hornberger, Phys. Rev. Lett. 97, 060601 (2006) 083010 (2012) [37] K. Hornberger, B. Vacchini, Phys. Rev. A 77, 022112 [70] M. Zwolak, C.J. Riedel, W.H. Zurek, Phys. Rev. Lett. (2008) 112, 140406 (2014) [38] M. Busse, K. Hornberger, J. Phys. A: Math. Theor. 42, [71] R. Blume-Kohout, W.H. Zurek, Phys. Rev. Lett. 101, 362001 (2009) 240405 (2008) [39] M. Busse, K. Hornberger, J. Phys. A: Math. Theor. 43, [72] H. Ollivier, W.H. Zurek, Phys. Rev. Lett. 88, 017901 015303 (2010) (2002) [40] R.A. Harris, L. Stodolsky, J. Chem. Phys. 74, 2145 [73] S. Schneider, G.J. Milburn, Phys. Rev. A 57, 3748 (1981) (1998) [41] H.D. Zeh, in Decoherence: Theoretical, Experimen- [74] C. Miquel, J.P. Paz, W.H. Zurek, Phys. Rev. Lett. 78, tal, and Conceptual Problems, ed. by P. Blanchard, 3971 (1997) D. Giulini, E. Joos, C. Kiefer, I. Stamatescu, Lecture [75] L.M.K. Vandersypen, I.L. Chuang, Rev. Mod. Phys. 76, Notes in Physics No. 538 (Springer, Berlin, 2000), pp. 1037 (2004) 19–42 [76] J.M. Martinis, S. Nam, J. Aumentado, K.M. Lang, [42] J.P. Paz, W.H. Zurek, Phys. Rev. Lett. 82, 5181 (1999) C. Urbina, Phys. Rev. B 67, 094510 (2003) [43] W.H. Zurek, S. Habib, J.P. Paz, Phys. Rev. Lett. 70, [77] C.J. Myatt, B.E. King, Q.A. Turchette, C.A. Sackett, 1187 (1993) D. Kielpinski, W.M. Itano, C. Monroe, D.J. Wineland, [44] W.H. Zurek, Prog. Theor. Phys. 89, 281 (1993) Nature 403, 269 (2000) [45] B.L. Hu, J.P. Paz, Y. Zhang, Phys. Rev. D 45, 2843 [78] S. Schneider, G.J. Milburn, Phys. Rev. A 59, 3766 (1992) (1999) [46] W.H. Zurek, Philos. Trans. R. Soc. London, Ser. A 356, [79] Q.A. Turchette, C.J. Myatt, B.E. King, C.A. Sackett, 1793 (1998) D. Kielpinski, W.M. Itano, C. Monroe, D.J. Wineland, [47] L. Di´osi,C. Kiefer, Phys. Rev. Lett. 85, 3552 (2000) Phys. Rev. A 62, 053807 (2000) [48] J. Eisert, Phys. Rev. Lett. 92, 210401 (2004) [80] K. Kraus, Ann. Phys. 64, 311 (1971) [49] G.M. Palma, K.A. Suominen, A.K. Ekert, Proc. R. Soc. [81] V. Gorini, A. Kossakowski, E.C.G. Sudarshan, J. Math. Lond. A 452, 567 (1996) Phys. 17, 821 (1976) [50] D.A. Lidar, I.L. Chuang, K.B. Whaley, Phys. Rev. Lett. [82] G. Lindblad, Commun. Math. Phys. 48, 119 (1976) 81, 2594 (1998) [83] R. Alicki, M. Fannes, Quantum Dynamical Systems [51] P. Zanardi, M. Rasetti, Phys. Rev. Lett. 79, 3306 (1997) (Oxford University Press, Oxford, 2001) [52] P. Zanardi, M. Rasetti, Mod. Phys. Lett. B 11, 1085 [84] R. Alicki, K. Lendi, Quantum Dynamical Semigroups (1997) and Applications, Lect. Notes Phys., vol. 717, 2nd edn. [53] P. Zanardi, Phys. Rev. A 57, 3276 (1998) (Springer, Berlin/Heidelberg, 2007) [54] D.A. Lidar, D. Bacon, K.B. Whaley, Phys. Rev. Lett. [85] F. Benatti, R. Floreanini, Int. J. Mod. Phys. B 19, 3063 82, 4556 (1999) (2005). doi:10.1142/S0217979205032097 [55] D. Bacon, J. Kempe, D.A. Lidar, K.B. Whaley, Phys. [86] R. D¨umcke, H. Spohn, Z. Phys. B 34, 419 (1979) Rev. Lett. 85, 1758 (2000) [87] V. Gorini, A. Frigerio, M. Verri, A. Kossakowski, E.C.G. [56] L.M. Duan, G.C. Guo, Phys. Rev. A 57, 737 (1998) Sudarshan, Rep. Math. Phys. 13, 149 (1978) [57] P. Zanardi, Phys. Rev. A 63, 012301 (2001) [88] E.B. Davies, Commun. Math. Phys. 39, 91 (1974) [58] E. Knill, R. Laflamme, L. Viola, Phys. Rev. Lett. 82, [89] A. Kossakowski, Rep. Math. Phys. 3, 247 (1972) 2525 (2000) [90] A. Barchielli, V.P. Belavkin, J. Phys. A: Math. Gen. 24, [59] J. Kempe, D. Bacon, D.A. Lidar, K.B. Whaley, Phys. 1495 (1991) Rev. A 63, 042307 (2001) [91] V.P. Belavkin, in Lecture Notes in Control and Infor- [60] D.A. Lidar, K.B. Whaley, in Irreversible Quantum Dy- mation Sciences, vol. 121 (Springer, Berlin, 1989), pp. namics, Springer Lecture Notes in Physics, vol. 622, ed. 245–265 21

[92] V.P. Belavkin, J. Phys. A: Math. Gen. 22, L1109 (1989) [129] F.M. Cucchietti, J.P. Paz, W.H. Zurek, Phys. Rev. A [93] V.P. Belavkin, Phys. Lett. A 140, 355 (1989) 72, 052113 (2005). doi:10.1103/PhysRevA.72.052113 [94] V.P. Belavkin, in Chaos: The Interplay Between [130] A.O. Caldeira, A.H. Castro Neto, T.O. de Carvalho, Stochastic and Deterministic Behaviour, ed. by P. Gar- Phys. Rev. B 48, 13974 (1993) baczewksi, M. Wolf, A. Veron, Lecture Notes in Physics [131] C. Miquel, J.P. Paz, R. Perazzo, Phys. Rev. A 54, 2605 (Springer, 1995), pp. 21–41 (1996) [95] L. Di´osi,Phys. Lett. A 129, 419 (1988) [132] A.M. Steane, Phys. Rev. Lett. 77, 793 (1996) [96] L. Di´osi,Phys. Lett. A 132, 233 (1988) [133] P.W. Shor, Phys. Rev. A 52, R2493 (1995) [97] L. Di´osi,J. Phys. A 21, 2885 (1988) [134] A.M. Steane, in Decoherence and Its Implications in [98] N. Gisin, Phys. Rev. Lett. 52, 1657 (1984) Quantum Computation and Information Transfer, ed. [99] N. Gisin, Helv. Phys. Acta 62, 363 (1989) by P. Turchi, A. Gonis (IOS Press, Amsterdam, 2001), [100] H.M. Wiseman, Phys. Rev. A 49, 2133 (1994) pp. 284–298. Also available as eprint quant-ph/0304016 [101] H.S. Goan, G.J. Milburn, H.M. Wiseman, H.B. Sun, [135] E. Knill, R. Laflamme, A. Ashikhmin, H. Barnum, Phys. Rev. B 63, 125326 (2001) L. Viola, W. Zurek, LA Science 27, 188 (2002) [102] M.B. Plenio, P.L. Knight, Rev. Mod. Phys. 70, 101 [136] M.A. Nielsen, I.L. Chuang, Quantum Computation and (1998) Quantum Information (Cambridge University Press, [103] N.V. Prokof’ev, P.C.E. Stamp, Rep. Prog. Phys. 63, Cambridge, 2000) 669 (2000) [137] J.H. Reina, L. Quiroga, N.F. Johnson, Phys. Rev. A 65 [104] M. Dub´e,P.C.E. Stamp, Chem. Phys. 268, 257 (2001) 65, 032326 (2002) [105] S. Gr¨oblacher, A. Trubarov, N. Prigge, M. Aspelmeyer, [138] D. Bacon, D.A. Lidar, K.B. Whaley, Phys. Rev. A 60, J. Eisert, Nature Comm. 6, 7606 (2015) 1944 (1999) [106] S. Chaturvedi, F. Shibata, Z. Phys. B 35, 297 (1979) [139] D.A. Lidar, D. Bacon, J. Kempe, K.B. Whaley, Phys. [107] F. Shibata, T. Arimitsu, J. Phys. Soc. Jpn. 49, 891 Rev. A 63, 022307 (2001) (1980) [140] D.A.R. Dalvit, J. Dziarmaga, W.H. Zurek, Phys. Rev. [108] A. Royer, Phys. Rev. A 6, 1741 (1972) A 62, 013607 (2000) [109] A. Royer, Phys. Lett. A 315, 335 (2003) [141] J.F. Poyatos, J.I. Cirac, P. Zoller, Phys. Rev. Lett. 77, [110] R. Feynman, F.L. Vernon, Ann. Phys. (N.Y.) 24, 118 4728 (1996) (1963) [142] A.R.R. Carvalho, P. Milman, R.L. de Matos Filho, [111] A. Caldeira, A. Leggett, Ann. Phys. (N.Y.) 149, 374 L. Davidovich, Phys. Rev. Lett. 86, 4988 (2001) (1983) [143] L. Viola, S. Lloyd, Phys. Rev. A 58, 2733 (1998) [112] O.V. Lounasmaa, Experimental Principles and Methods [144] L. Viola, E. Knill, S. Lloyd, Phys. Rev. Lett. 82, 2417 below 1 K (Academic Press, New York, 1974) (1999) [113] S.L. Adler, J. Phys. A: Math. Gen. 39, 14067 (2006) [145] P. Zanardi, Phys. Lett. A 258, 77 (1999) [114] M. Tegmark, Found. Phys. Lett. 6, 571 (1993) [146] L. Viola, E. Knill, S. Lloyd, Phys. Rev. Lett. 85, 3520 [115] L. Hackerm¨uller, K. Hornberger, B. Brezger, (2000) A. Zeilinger, M. Arndt, Appl. Phys. B 77, 781 [147] L.A. Wu, D.A. Lidar, Phys. Rev. Lett. 88, 207902 (2003) (2002) [116] K. Hornberger, S. Uttenthaler, B. Brezger, L. Hack- [148] L.A. Wu, M.S. Byrd, D.A. Lidar, Phys. Rev. Lett. 89, erm¨uller,M. Arndt, A. Zeilinger, Phys. Rev. Lett. 90, 127901 (2002) 160401 (2003) [149] M. Arndt, K. Hornberger, A. Zeilinger, Phys. World 18, [117] K. Hornberger, J.E. Sipe, M. Arndt, Phys. Rev. A 70, 35 (2005) 053608 (2004) [150] R. Kaiser, C. Westbrook, F. David (eds.). Coherent [118] S. Nimmrichter, K. Hornberger, P. Haslinger, M. Arndt, Atomic Matter Waves, Les Houches Session LXXII, Les Phys. Rev. A 83, 043621 (2011) Houches Summer School Series (Springer, Berlin, 2001) [119] D.A. Kokorowski, A.D. Cronin, T.D. Roberts, D.E. [151] M. Blencowe, Phys. Rep. 395, 159 (2004) Pritchard, Phys. Rev. Lett. 86, 2191 (2001) [152] M. Aspelmeyer, T.J. Kippenberg, F. Marquardt, Rev. [120] H. Uys, J.D. Perreault, A.D. Cronin, Phys. Rev. Lett. Mod. Phys. 86, 1391 (2014) 95, 150403 (2005) [153] A. Bassi, G.C. Ghirardi, Phys. Rep. 379, 257 (2003) [121] A.O. Caldeira, A.J. Leggett, Physica A 121, 587 (1983) [154] S.L. Adler, J. Phys. A 40, 2935 (2007) [122] D.F. Walls, M.J. Collett, G.J. Milburn, Phys. Rev. D [155] A. Bassi, D.A. Deckert, L. Ferialdi, EPL 92, 50006 32, 3208 (1985) (2010) [123] U. Weiss, Quantum Dissipative Systems (World Scien- [156] S. Nimmrichter, K. Hornberger, Phys. Rev. Lett. 110, tific, Singapore, 1999) 160403 (2013) [124] M. Schlosshauer, A.P. Hines, G.J. Milburn, Phys. Rev. [157] M. Brune, E. Hagley, J. Dreyer, X. Maˆıtre,A. Maali, A 77, 022111 (2008) C. Wunderlich, J.M. Raimond, S. Haroche, Phys. Rev. [125] P.C.E. Stamp, in Tunnelling in Complex Systems, ed. Lett. 77, 4887 (1996) by S. Tomsovic (World Scientific, Singapore, 1998), pp. [158] L. Davidovich, M. Brune, J.M. Raimond, S. Haroche, 101–197 Phys. Rev. A 53, 1295 (1996) [126] N.V. Prokof’ev, P.C.E. Stamp, (1995) [159] X. Maˆıtre, E. Hagley, J. Dreyer, A. Maali, C.W.M. [127] N.V. Prokof’ev, P.C.E. Stamp, J. Phys. Chem. Lett. 5, Brune, J.M. Raimond, S. Haroche, J. Mod. Opt. 44, L663 (1993) 2023 (1997) [128] V.V. Dobrovitski, H.A. De Raedt, M.I. Katsnelson, [160] A. Auffeves, P. Maioli, T. Meunier, S. Gleyzes, B.N. Harmon, Phys. Rev. Lett. 90, 210401 (2003). doi: G. Nogues, M. Brune, J.M. Raimond, S. Haroche, Phys. 10.1103/PhysRevLett.90.210401 Rev. Lett. 91, 230405 (2003) 22

[161] S. Del´eglise,I. Dotsenko, C. Sayrin, J. Bernu, M. Brune, Lukens, Nature 406, 43 (2000) J.M. Raimond, S. Haroche, Nature 455, 510 (2008) [172] C.H. van der Wal, A.C.J. ter Haar, F.K. Wilhelm, R.N. [162] M. Arndt, O. Nairz, J. Vos-Andreae, C. Keller, Schouten, C.J.P.M. Harmans, T.P. Orlando, S. Lloyd, G. van der Zouw, A. Zeilinger, Nature 401, 680 (1999) J.E. Mooij, Science 290, 773 (2000) [163] S. Gerlich, S. Eibenberger, M. Tomandl, S. Nimm- [173] I. Chiorescu, Y. Nakamura, C.J.P.M. Harmans, J.E. richter, K. Hornberger, P.J. Fagan, J. T¨uxen,M. Mayor, Mooij, Science 21, 1869 (2003) M. Arndt, Nature Comm. 2, 263 (2012) [174] P. Bertet, I. Chiorescu, G. Burkard, K. Semba, C.J.P.M. [164] S. Eibenberger, S. Gerlich, M. Arndt, M. Mayor, Harmans, D.P. DiVincenzo, J.E. Mooij, Phys. Rev. J. T¨uxen,Phys. Chem. Chem. Phys. 15, 14696 (2013) Lett. 95, 257002 (2005) [165] S. Gerlich, L. Hackerm¨uller,K. Hornberger, A. Stibor, [175] Y. Nakamura, Y.A. Pashkin, J.S. Tsai, Nature 398, 786 H. Ulbricht, F. Goldfarb, T. Savas, M. M¨uri,M. Mayor, (1999) M. Arndt, Nature Phys. 3, 711 (2007) [176] D. Vion, A. Aassime, A. Cottet, P. Joyez, H. Pothier, [166] P. Haslinger, N. D¨orre,P. Geyer, J. Rodewald, S. Nimm- C. Urbina, D. Esteve, M.H. Devoret, Science 296, 886 richter, M. Arndt, Nature Phys. 9, 144 (2013) (2002) [167] K. Hornberger, L. Hackerm¨uller,M. Arndt, Phys. Rev. [177] Y. Yu, S. Han, X. Chu, S.I. Chu, Z. Wang, Science 296, A 71, 023601 (2005) 889 (2002) [168] L. Hackerm¨uller, K. Hornberger, B. Brezger, [178] J.M. Martinis, S. Nam, J. Aumentado, C. Urbina, Phys. A. Zeilinger, M. Arndt, Nature 427, 711 (2004) Rev. Lett. 89, 117901 (2002) [169] M. Arndt, O. Nairz, A. Zeilinger, in Quantum [179] K. Camilleri, Stud. Hist. Phil. Mod. Phys. 40, 290 [Un]Speakables: From Bell to Quantum Information, (2009) ed. by R.A. Bertlmann, A. Zeilinger (Springer, Berlin, [180] D. Wallace, in Many Worlds? Everett, Quantum Theory 2002), pp. 333–351 and Reality, ed. by S. Saunders, J. Barrett, A. Kent, [170] K. Hornberger, Phys. Rev. A 73, 052102 (2006) D. Wallace (Oxford University Press, Oxford, 2010), pp. [171] J.R. Friedman, V. Patel, W. Chen, S.K. Yolpygo, J.E. 53–72