Arxiv:2105.14032V1 [Quant-Ph] 28 May 2021 the 2Nd Eigenstate of Ρ Until the ﬁrst Eigenstate Has Time of Coherent State Classical Wavepackets

Copycat process in the early stages of einselection

Rose Baunach∗ and Andreas Albrecht† Center for Quantum Mathematics and Physics and Department of Physics and Astronomy UC Davis, One Shields Ave, Davis CA.

Andrew Arrasmith‡ Theoretical Division, Los Alamos National Laboratory, Los Alamos, NM USA. (Dated: June 1, 2021) We identify and describe unique early time behavior of a quantum system initially in a superposition, interacting with its environment. This behavior—the copycat process—occurs after the system begins to decohere, but before complete einselection. To illustrate this behavior analytic solutions for the system density matrix, its eigenvalues, and eigenstates a short time after system-environment interactions begin are provided. Features of the solutions and their connection to observables are discussed, including predictions for the continued evolution of the eigenstates towards einselection, time dependence of spin expectation values, and an estimate of the system’s decoherence time. In particular we explore which aspects of the early stages of decoherence exhibit quadratic evolution to leading order, and which aspects exhibit more rapid linear behavior. Many features of our early time perturbative solutions are agnostic of the spectrum of the environment. We also extend our work beyond short time perturbation theory to compare with numerical work from a companion paper.

I. INTRODUCTION The evolution shown in Fig. 1 starts with the system and environment in a product state. At that mo- There are many reasons why one would want to study ment the SHO density matrix, ρ, has only one nonzero the effects of decoherence and einselection on a quantum eigenvalue. An instant later a second nonzero (but in- system interacting with its environment—from interest finitesimal) eigenvalue emerges. As soon as this second in theoretical interpretations of quantum mechanics to eigenvalue becomes resolved in our calculations, the cor- applications in quantum computing [1–8]. In [9] we in- responding eigenstate takes on the intriguing “mirror im- troduced the “adapted Caldeira Leggett” (ACL) model, age” (copycat) form shown in Fig. 1. The first and sec- a tool designed to explore these phenomena using fully ond eigenstates keep these forms as the second eigenvalue unitary calculations in the combined system-environment evolves over many decades in magnitude. Eventually on space. This tool enables us to examine behaviors outside a timescale given by the decoherence time, einselection 1 of the standard approximation schemes common in the takes place. field. While initially the copycat phenomenon seemed strik- Our original aim was a study of the relationship being and unusual to us, and we were a bit concerned about tween einselection and the arrow of time. We present the the possible role of numerical artifacts in our results, we outcome of that work in [10]. On the path of exploring have come to understand this process in relatively sim- decoherence and einselection with the ACL model numer- ple physical terms. The “transient stability” we observe ically, we witnessed a curious phenomenon—the copycat relates to the slow quadratic start to the early time evo- process—which we investigate in this paper. Figure 1 lution of the eigenstates. Furthermore, the mirror im- (Fig. 12 in [9]) gives a general picture of this process. age copycat form seems quite natural if one thinks of the two wavepackets as spanning a two dimensional re- The ACL model describes a simple harmonic oscilla- duced Hilbert space (effectively a single qubit). From tor (SHO) coupled to an environment. For Fig. 1 the that point of view, the form shown is the only option for SHO was set up in a “Schrödingercat” superposition arXiv:2105.14032v1 [quant-ph] 28 May 2021 the 2nd eigenstate of ρ until the first eigenstate has time of coherent state classical wavepackets. Over time, de- to evolve appreciably. In fact, the copycat picture we coherence with the environment brings the SHO into a describe here is implicit in standard results from Nuclear classical mixture of wavepackets described by a density Magnetic Resonance (NMR) physics, although they have matrix (ρ) with the wavepackets as eigenstates. In this not been previously presented from that point of view. manner the classical wavepackets are specially selected by This work originally came about through our need to the specifics of the decoherence physics, a process called fully understand the behaviors of the ACL model before “einselection.” applying it to the new problems explored in [10]. Having

∗ [email protected] 1 The SHO has a period of 2π. Thus, choosing t = 4π for the ﬁnal † [email protected] time shown in Fig. 1 simpliﬁes our presentation. More details ‡ [email protected] about the behavior of the ACL model can be found in [9]. 2

behave as essentially a two-state system in our numerical work [9]. Section IV is a discussion of some speciﬁc Properties of eigenstates features and applications of our derived solutions for the 100 probability of 2nd two state system, including predictions for the continued eigenstate evolution of the system eigenstates, early time behav- -10 ior of spin observables, and an estimate of the system’s 10 decoherence time. In Sect. V we extend our solutions be- 10-5 t 100 yond short time perturbation theory to compare with our 1st eigenstate 2nd eigenstate 0.2 0.2 numerical work in [9]. Our further discussion and con- -6 t=10-6 t=10 clusions are provided in Sect. VI. Appendix A extends 0.1 0.1 our technical results from a qubit model to the qutrit case, providing additional perspective on the possibility 0 0 of linear behaviors (which we summarize is Sect. VI). -10 0 10 -10 0 10 0.2 0.2 t=0.1 t=0.1 0.1 0.1 II. FORMALISM

0 0 A. Basics -10 0 10 -10 0 10 0.3 t=4 t=4 We consider a “world” Hilbert space, comprised of a system and an environment: w = s ⊗ e. We take |ψwi initially to be a product state and study the entanglement 0 0 caused by system-environment interactions: -10 0 10 -10 0 10 q q −−−−−−−−−−→ X |ψwi = |ψsi |ψei entanglement bij |iis |jie . (1) i,j FIG. 1. Copycats in the early stages of entanglement: The SHO is initially taken to be in a pure Schrödingercat state, The onset of system-environment entanglement is called which becomes entangled with the environment as it evolves. decoherence. Once entanglement has taken place, the The 2nd eigenvalue and |ψ(q)|2 for the first two eigenstates system is described by the density matrix of ρ are shown from very early times. The 2nd eigenstate takes the mirror image “copycat” form over several decades ρs = T re(|ψwi hψw|). (2) of evolution. The bottom row shows these two states after einselection is complete, and the initial superposition has be- Einselection is the special case of decoherence where en- come a mixture of classical wave packets. vironmental interactions induce the system density matrix to become diagonal in a preferred basis. These preferred basis states are called pointer states in the literature [2, 11]. satisfied ourselves that the copycat process was indeed Under certain conditions (which apply in Fig. 1) co- physical, and having extended it in a number of direc- herent states will be einselected as the pointer states tions, we’ve decided those explorations are worth report- [1, 9, 12]. Under those conditions, a system that starts ing in this paper. We feel they may be of interest to in a superposition of coherent states those studying the early stages of decoherence, perhaps in the context of quantum technologies. In particular we |ψsi = a1 |α1i + a2 |α2i (3) offer explorations of the extent to which the early stages of decoherence are quadratic to leading order, and which would evolve into the density matrix aspects evolve with a more rapid linear behavior. 2 2 ρs = |a1| |α1i hα1| + |a2| |α2i hα2| . (4) The rest of this paper is devoted to describing and analytically deriving the copycat process. In Sect. II Here we have labeled coherent states with the parame- we briefly review the ACL and reduced Caldeira-Leggett ter α, a standard convention articulated in detail for the (RCL) models as presented in [9]. Then in Sect. III ACL model in [9]. Indeed Eqn. 4 roughly describes what we present perturbative analysis of the copycat pro- we see in Fig. 1. (Inspection of the full analysis presented cess by calculating the system density matrix, associated in [9] reveals that in this particular case, the finite form eigenvalues, and eigenstates of a two-state system en- of the ACL model leads to a number of small deviations tangled with its environment a short time after system- from the idealized picture described by Eqn. 4.) It is in environment interactions begin. In our derivations we the early stages of the evolution toward the Eqn. 4 form chose to model the superposition of coherent states in that we notice the copycat behavior. Eqn. 3 as a two state system, both for simplicity and As illustrated in Fig. 1, at early times there is an because such coherent superpositions were observed to eigenstate of the system density matrix that resembles 3 the initial state and one that has the “copycat” form. As with the full SHO case, we start with a pure product The initial evolution of these eigenstates are quadratic in state at t = 0 with no initial entanglement. At t = 0 the time—as we will prove subsequently—or “slow,” hence system (now just a qubit) is a two-state superposition the appearance of “transient stability.” We call the ap- and the environment is in some pure state which we call pearance of a copycat state and its subsequent behavior |φei: the “copycat process”. To the best of our knowledge the |ψw(0)i = |ψs(0)i |φei copycat process and its implications have not been directly explored in the literature, although we will discuss = a |↑i + b |↓i |φei . (6) how earlier work has come very close to this topic in an Here a and b can be complex, and |a|2 + |b|2 = 1. We indirect way. consider a Hamiltonian given by: B. The ACL Model ↑ ↓ Hw ≡ H = λ |↑i h↑| He + |↓i h↓| He (7) As discussed in [9], the original Caldeira Leggett (CL) where λ is a real parameter to adjust the strength of the model is a toy model describing a system interacting with ↑ ↓ interaction, and He and He only operate in the subspace its environment with a Hamiltonian of the form [1, 7, 13]: of the environment. We will refer to Hw as H for brevity in what follows. s e e e s ↑ ↓ Hw = HSHO ⊗ 1 + qSHO ⊗ HI + H ⊗ 1 . (5) We take He , He , and H to all be time independent, and we generally allow H↑H↓ 6= H↓H↑. For most of what In the CL model the system is a simple harmonic oscil- e e e e follows, no additional assumptions are made about the lator (SHO) moving in the standard SHO potential, and eigenvalue spectra of the H ’s or the dimensionality of the the environment is an infinite set of SHOs. The system e environment. We note that H in Eqn. 7 is very similar and environment together describe a closed system un- to what we call the “reduced Caldeira-Leggett” (RCL) dergoing unitary evolution, but generally the system and model Hamiltonian in [9], although there we considered environment individually do not have to evolve unitarily. the special case where H↑ = −H↓. We call the model we The adapted Caldeira Leggett (ACL) model was intro- e e use here an RCL model as well, and note that as discussed duced as an adaptation of the CL model which operates in [9] this model will einselect the pointer states |↑i and in a finite dimensional Hilbert space—so its evolution can |↓i. be investigated numerically in its full unitary form [9]. In Working in the Schrodinger picture with: the ACL model, the Hamiltonian is also given by Eqn. 5, but the components are modified since the Hilbert space ∂ −ıHt ı |ψi = H |ψi U(t) = e ~ (8) is finite—for example, the system is given by a truncated ~∂t SHO. Full technical details are given in [9]. we can perturbatively compute the state of the system and environment at a short time t = ∆, using the series III. MODELLING THE COPYCAT PROCESS expansion of the time evolution operator

|ψw(∆)i = U(∆) |ψw(0)i A. Setting up and the RCL model 2 ıH∆ 1 (H∆) 3 = 1 − − 2 + O(∆ ) |ψw(0)i . We start our technical explorations of the copycat pro- ~ 2 ~ (9) cess with the following observation: The ACL model used to produce Fig. 1 had parameters adjusted to make The result is: the coherent states especially stable, making them the |ψ (∆)i = a |↑i + b |↓i |φ i pointer states. We also note that while the coherent w e state wavefunctions are nowhere truly zero, the overlap ı∆λ ↑ ↓ − a |↑i He |φei + b |↓i He |φei between the two coherent states shown in Fig. 1 is expo- ~ 2 2 nentially suppressed making the two coherent states es- ∆ λ ↑ ↑ ↓ ↓ − a |↑i He He |φei + b |↓i He He |φei . sentially orthogonal. The SHO dynamics will ultimately 2~2 move the two packets into positions of greater overlap, (10) but the copycat process takes place on time scales short We have found the important leading order behavior oc- compared to the SHO evolution. We use both the sta- curs at second order, so we keep terms up to O(∆2) in bility of the coherent states and their lack of overlap to what follows. argue heuristically that they span a two dimensional subspace, which is eﬀectively decoupled from the rest of the SHO Hilbert space at early times. Based on these consid- B. System Reduced Density Matrix erations, we model the superposition of coherent states in Eqn. 3 (used to generate Fig. 1) with a single qubit cou- We compute the reduced density matrix of the system pled to an environment and undertake analytical calcu- after a short time t = ∆: lations of early time behavior using perturbation theory in the small time parameter. ρs(t = ∆) = Tre |ψw(∆)i hψw(∆)| . (11) 4

↑ ↓ ↑ ↓ ↓ ↑ The result expressed in the |↑i , |↓i basis is: hφe| He He |φei can be complex (since He He 6= He He ). This requires β to be purely real, but allows η to be com-   aa∗ ab∗ 1 + ıβ∆ − η∆2 plex. It follows that the system density matrix in Eqn. 12 † ρs(∆) = is Hermitian and properly normalized since ρ = ρ, and  ∗ ∗ 2 ∗  ba 1 − ıβ∆ − η ∆ bb T r[ρ] = |a|2 + |b|2 = 1. (12) where the coeﬃcients β and η are given by:

" ! ! # b∗ 2|a|2|b|2 + |a|2 2|a|2|b|2 − |a|2 |ψ i = a 1 + ıβ∆ + ∆2 − η |↑i + b 1 + ∆2 |↓i (15) 1 |b| 2 2 " ! ! # −a 2|a|2|b|2 + |b|2 2|a|2|b|2 − |b|2 |ψ i = b∗ 1 + ıβ∆ + ∆2 − η |↑i − a∗ 1 + ∆2 |↓i (16) 2 |a| 2 2

with associated eigenvalues: the two states have the “mirror image” feature which led us to call the second eigenstate a“copycat” state in Fig. 1. 2 2 2 p1 = 1 − |a| |b| ∆ (17) Thus, we see that at very early times the copycat proﬁle 2 2 2 p2 = |a| |b| ∆ (18) is achieved automatically in this simple illustration. We noted in the introduction that the initial time where the parameter is deﬁned by: evolution of the copycat state appeared to be “slow”

∗ 2 in our numerical simulations. This is also apparent in ≡ η + η − β . (19) our analytic solutions—they show the time dependence of the eigenstates and their associated probabilities to We note that can also be written as be quadratic to leading order modulo a linear complex 2 λ h ↓ ↑ 2 ↓ ↑ 2i phase (which we have by convention placed in the co- = hφe| (He − He ) |φei − hφe| (He − He ) |φei ~2 efficients of |↑i). The quadratic early time dependence (20) of the eigenvalues has been anticipated before by calcu- (using Eqns. 13 and 14). Equation 20 shows that is lations in [3, 14]—where their “rate of deseparation” is just the variance of λ (H↓ − H↑), so ≥ 0 by definition. analogous to the quantity |a|2|b|2—but the eigenstate ~ e e Furthermore, = 0 is a degenerate case where einse- solutions and their copycat nature is a new feature of lection does not occur—you can see, for example, that our analysis. when = 0 the second eigenvalue in Eqn. 18 exactly disappears and the only state remaining with any probability is the initial state. Thus, for cases of interest here 3 IV. FURTHER PERTURBATIVE ANALYSIS > 0. Also note that hψ1|ψ2i = hψ2|ψ1i = 0 + O(∆ ) 3 and hψ1|ψ1i = hψ2|ψ2i = 1+O(∆ ), as you would expect from an O(∆2) calculation. A. Continued evolution of eigenstates Inspecting Eqns. 15 and 16, the zeroth order terms identify |ψ1i with the original state of the system (|ψs(0)i To investigate what happens to the system density ma- from Eqn. 6 apart from an irrelevant overall phase) and trix eigenstates after the copycat state appears, let us |ψ2i as the orthogonal state. In a two dimensional Hilbert again consider Eqns. 15 and 16. At the onset of the space, there is (up to an overall phase) only one orthog- copycat process, for small t = ∆, the system has already onal state to |ψ1i. From the way a and b alternate loca- begun to decohere, but einselection has hardly started. tions in the expressions for |ψ1i vs |ψ2i, one can see that While the system and environment are clearly entangled, 5 the system density matrix eigenstates are not yet de- order O(∆2). One could interpret this as evidence for a scribed by the pointer states |↑i and |↓i of the Hamil- static system—that after the onset of the copycat process tonian in Eqn. 7. no further evolution of the eigenstates occurs. However, We now explore the perturbative behavior as ∆ in- in the exactly degenerate limit |a|2 = |b|2 all states are creases. Here we focus on |ψ|2 of each eigenstate, given equally “good” eigenstates of the system density matrix, by so einselection into a specific basis of pointer states has no meaning in this limit.2 2h 2 2 2 2 i hψ1|ψ1i = |a| 1 + ∆ (2|a| |b| − |b| ) h↑ | ↑i h i + |b|2 1 + ∆2(2|a|2|b|2 − |a|2) h↓ | ↓i B. Decoherence Time (21) Full einselection will take place on the timescale set by 2h 2 2 2 2 i hψ2|ψ2i = |b| 1 + ∆ (2|a| |b| − |a| ) h↑ | ↑i the decoherence processes. A system that has fully einselected will have the off-diagonal elements of its density 2h 2 2 2 2 i + |a| 1 + ∆ (2|a| |b| − |b| ) h↓ | ↓i matrix close to zero when ρs is expressed in the pointer (22) state basis [2]. Although this stage is only reached outside of the range of our perturbative calculations, we can where we have made use of Eqn. 19. still estimate the decoherence time by solving for the For > 0—which is true in all cases where our analysis value of ∆ where the off-diagonal elements are zero for holds (see the discussion below Eqns. 19 and 20)—we can our perturbative calculations. Applying this to Eqn. 12 re-write Eqns. 21 and 22 as: gives: h ↑ i h ↓ i ∗ 2 2 2 ρ|↑ih↓| = ab 1 + ıβ∆ − η∆ = 0 hψ1|ψ1i = |a| 1 + C1 (∆) h↑ | ↑i + |b| 1 + C1 (∆) h↓ | ↓i (23) ∗ ∗ 2 ρ|↓ih↑| = ba 1 − ıβ∆ − η ∆ = 0 (25) h i h i hψ |ψ i = |b|2 1 + C↑(∆) h↑ | ↑i + |a|2 1 + C↓(∆) h↓ | ↓i 2 2 2 2 which can be rewritten as (24) ∗ ∗ ∗ ∗ 2 ∗ ∗ ∗ and construct the following chart for the sign of the time ab + ba + ıβ∆(ab − ba ) − ∆ (ab η + ba η ) = 0 dependent coefficients as time increases: (26) ∗ ∗ ∗ ∗ 2 ∗ ∗ ∗ ↑ ↓ ↑ ↓ ab − ba + ıβ∆(ab + ba ) − ∆ (ab η − ba η ) = 0. Original State C1 (∆) C1 (∆) C2 (∆) C2 (∆) (27) |a|2 > |b|2 + - - + 2 2 Solving Eqns. 26 and 27 together and simplifying yields |a| < |b| - + + - the following perturbative estimate for the decoherence 2 2 |a| = |b| 0 0 0 0 time: r Comparing Eqns. 21 and 22 with Eqns. 23, 24, and 2 ∆d = (28) the chart illustrates that the subsequent evolution of η + η∗ the system eigenstates towards einselection is determined where η is given by Eqn. 14. Note that this result is inde- by the hierarchy of |a|2 and |b|2, the initial system pendent of a and b (the initial state of the system). We probabilities—and that interactions with the environ- have compared this expression with our numerical work ment control how fast this evolution occurs through the and found it gives reasonable estimates of the decoher- parameter . ence time. For example, suppose the initial state of the system is such that |a|2 > |b|2 at t = 0. As t = ∆ grows we see that the probability to observe |ψ1i in the |↑i C. Spin Observables ↑ state increases, since C1 (∆) is increasing over time, and that the probability of observing |ψ1i in the |↓i state Here we consider the behavior of the Pauli spin opera- is decreasing by the same token. The exact opposite tors in our RCL solutions. This will allow contact to be trends occur in |ψ2i, the orthogonal state. So long as 3 hψ1|ψ1i = hψ2|ψ2i = 1 + O(∆ ) is preserved (namely that the perturbation expansion remains valid), the full system will exhibit these trends. The system approaches 2 In cases with very small deviations away from complete degen- complete einselection once 1 + C↓(∆) ≈ 1 + C↑(∆) ≈ 0. eracy, small irregularities (due, for example, to the finite size of 1 2 the environment) can disrupt any tendency toward einselection. An exactly analogous explanation occurs for the case of We have seen this phenomenon in our numerical work, where for 2 2 |a| < |b| . sufficiently degenerate cases finite size effects introduced large For the case of |a|2 = |b|2, all time dependent coeffi- random fluctuations which dominated over the einselection pro- cients vanish for a properly normalized state to leading cess. 6 made with various experimental contexts such as NMR A. An Alternate Derivation and quantum computing [1, 6, 15]. Our basis states for our system density matrix, To derive a non-perturbative version of Eqn. 12, we {|↑i , |↓i}, can be identified with the Sz eigenbasis for begin by specifying the following: 1 spin- 2 , so we can compute the expectation values for the N spin operators Sx, Sy and Sz by ↑ X He = ~ωi |ωii hωi| (34) i ~ hSii = T r(ρsSi) = T r(ρsσi) (29) N 2 ↓ X ↑↓ He = ~f ωi |ωii hωi| (35) i where the σi are the usual Pauli matrices so that 0 1 0 −ı 1 0 σx = σy = σz = . (30) ↓ ↑↓ ↑ 1 0 ı 0 0 −1 ωi = f ωi (36)

↑ ↓ This gives i.e. He and He are almost identical, except for a tune- able real dimensionless constant f ↑↓. (The RCL model discussed in [9] has this form with f ↑↓ = −1.) This sim- ~ 2 2 ↑ hSzi = (|a| − |b| ) (31) plifying assumption about the relationship between He 2 ↓ and He enables the analysis which follows. ~ ∗ ∗ 2 ∗ hSxi = 2Re[ab ] − 2β∆Im[ab ] − 2∆ Re[ab η] We can express the state of the environment in the 2 energy eigenbasis of H↑ and H↓, so that (32) e e N ~ ∗ ∗ 2 ∗ hSyi = − 2Im[ab ] − 2β∆Re[ab ] + 2∆ Im[ab η] . X 2 |φei = αi |ωii (37) (33) i with the normalization condition: Note that the system will have fully deco- N hered/completed einselection when hSxi = hSyi = 0. For X ↑ 2 our perturbative expressions, this condition is the same |αi | = 1. (38) as that imposed by Eqns. 26 and 27. i Both the ACL and RCL models operate within a finite dimensional Hilbert space, so we have made that explicit ↑ ↓ V. BEYOND PERTURBATION THEORY in our forms for He and He . The exact frequency spectrum of the ωi’s is still arbitrary. However, when com- parisons with our numerical work are made in the next An intriguing part of the copycat process is that its section, we will take their distribution to be random and general features are agnostic about the spectrum of the centered around zero—to coincide with the random na- environment. The Hamiltonian in Eqn. 7 used to de- ↑ ↓ ture of He and He in the ACL model [9]. rive our results thus far makes no assumptions about the Given the definitions in Eqns. 34 - 38, we can re-express ↑ pieces that operate on the state of the environment, He the original state of the system and environment as: ↓ and He , except that they are time independent. This gives our results a flavor of generality often missing from N X canonical toy models in the literature—reviewed in [1, 7] |ψw(0)i = (a |↑i + b |↓i) ⊗ αi |ωii (39) and others—which typically make specific assumptions i of “ohmic” environments and the like in order to arrive at concrete mathematical expressions. and the RCL Hamiltonian originally given in Eqn. 7 be- ↑ ↓ comes: However, if we do further specify He and He we can compute a non-perturbative version of the density ma- N trix in Eqn. 12, closely following an approach by Zurek X Hw =(λ |↑i h↑| ⊗ ~ωi |ωii hωi|) for a similar model [11]. This has two benefits. First, it i is possible to re-derive a form of the copycat results as N leading order terms in the time series expansion of the X ↑↓ + (λ |↓i h↓| ⊗ ~ωif |ωii hωi|) (40) non-perturbative solutions—as should be the case. Sec- i ond, a non-perturbative approach allows us to interpret the full time range of our numerical results discussed in Since we have made the eigenvalues of our Hamiltonian [9] from an analytic perspective. explicit from the start, we may write down the full time 7

↑ evolved state as: and 14, and then substituting in the specific forms of He , H↓, and |φ i given in this section—the result will be the N e e X same as Eqn. 45. An analogous expression for z∗(∆) |ψ (t)i =(a α e−ıλωit |↑i ⊗ |ω i) w i i holds, which enables us to re-express Eqn. 12 as i N 2 ∗ X ↑↓ ρs(∆) =|a| |↑i h↑| + ab z(∆) |↑i h↓| + (b α e−ıλf ωit |↓i ⊗ |ω i) (41) i i ∗ ∗ 2 i + ba z (∆) |↓i h↑| + |b| |↓i h↓| (46) Tracing over the environment then gives the following From Eqns. 45 and 46, one can then go on to determine system density matrix: the eigenvalues, eigenvectors and decoherence time of the system density matrix. The results will match the more 2 ∗ ρs(t) =|a| |↑i h↑| + ab z(t) |↑i h↓| general calculations in Sections III and IV, for the specific + ba∗z∗(t) |↓i h↑| + |b|2 |↓i h↓| (42) versions of η and β given in Eqn. 45. To summarize, in this section we have derived a specific where the quantity z(t) has been called the correlation realization of the copycat results as leading order terms in amplitude or decoherence factor [2, 6, 11, 15] for similar the time series expansion of Eqns. 42 and 43. Note that if ↑ ↓ toy models and is given in our notation by: the relationship between He and He in Eqns. 34 and 35 was more complicated—if they did not share the same N ↑↓ energy eigenbasis or if the relationship between eigen- X 2 −ıtλωi(1−f ) z(t) = |αi| e (43) values was non-linear, for example—then a derivation of i the early time density matrix from a non-perturbative Note that z(t) is a sum of complex exponentials that approach might not proceed as smoothly as we just de- directly depends on the difference in eigenvalues of our scribed. However, the early time results of Sect. III will two environmental Hamiltonians. As Zurek originally have a more general range of validity. 2 discussed in [11], the quantity |αi| describes the probability of finding the environment in the different eigenstates of the interaction Hamiltonian, and it is possible B. Comparison with numerical results to show that the average value of z(t) will approach zero for sufficiently long times, effectively damping out the In this section we consider two quantities that de- off-diagonal system density matrix elements. pend strongly on the off-diagonal elements of the system Together Eqns. 42 and 43 are the non-perturbative ver- density matrix—the linear entropy and hSxi—and com- sion of the system density matrix in Eqn. 12, for the pare numerical non-perturbative results to semi-analytic specific realization of environment parameters given by early-time expressions. Eqns. 34 - 38. To show how these results connect with The linear entropy for a density matrix is defined as: the copycat process solutions, first write z(t) in terms of 2 trigonometric functions. Sl(ρ) = 1 − T r[ρ ] (47)

N which is bounded according to 0 ≤ Sl ≤ 1 [7]. For the X 2 ↑↓ ↑↓ z(t) = |αi| cos(λωi(1 − f )t) − ı sin(λωi(1 − f )t) system density matrix given in our perturbative analysis i (Eqn. 12), this yields: (44) 2 2 2 3 Then we take the limit t → ∆ by keeping only the first Sl,P = 2|a| |b| ∆ + O(∆ ) (48) non-trivial term in each trigonometic function’s series expansion. We still keep the sum over the states of the en- with given by: vironment, all we are doing is an early time expansion. " N N !2# This yields: 2 ↑↓ 2 X 2 2 X 2 = λ (f − 1) ωi |αi| − ωi|αi| (49) N i i X z(∆) =1 + ı |α |2λω (f ↑↓ − 1)∆ i i assuming Eqns. 34 - 38 from the previous section. i For the non-perturbative case, given Eqns. 42 and 43, N X λ2(f ↑↓ − 1)2 one obtains: − |α |2ω2 ∆2 i i 2 i 4 4 2 2 2 Sl,NP = 1 − |a| − |b| − 2|a| |b| |z(t)| (50) =1 + ıβ∆ − η∆2 (45) with which is exactly the time dependent off-diagonal element N in Eqn. 12, given the definitions in Eqns. 34 - 38. One can ↑↓ 2 X 2 2 −ıλ(1−f )(ωi−ωj )t verify the equivalence between the two lines of Eqn. 45 by |z(t)| = |αi| |αj| e (51) starting with the definitions of β and η given by Eqns. 13 i,j 8

1 Numerical 0.4 Perturbative 0.5

0.3 >

x S 0.2

FIG. 3. hS i, giving the real part of the off-diagonal element FIG. 2. Linear entropy curves. Solid: Non-perturbative (cor- x of ρ . The solid curve is non-perturbative, corresponding to responding to numerically evaluating Eqns. 50 and 51). Dot- s numerically evaluating Eqn 53. The dotted curve corresponds ted: Perturbative expression from Eqn. 48, using (given by to the early time analytic expression in Eqn. 52, with the Eqn. 49) drawn from the same numerical calculation shown. the quantities defined in Eqn 45 drawn from the numerical calculation. We take ~ = 1. Figure 2 shows the perturbative and non-perturbative linear entropies as functions of time. To numerically gen- in Fig. 3. We further discuss the interpretation of these erate the non-perturbative solid curve, our simulations oscillations in [9] and link them to phenomena seen in effectively evaluate the summation in Eqn. 51 followed NMR experiments. We also identify a modification to by Eqn. 50 at each time-step and plot the result. For the the RCL model which reduces these oscillations, thereby dotted curve corresponding to Eqn. 48, the summation further illuminating their physical origins. in Eqn. 49 is evaluated once numerically and then the Both Figs. 2 and 3 demonstrate that the perturbative expression in Eqn. 48 is plotted for the same time-steps √ regime can be a sizable portion of the full time evolution as those√ used for Eqn. 50. For both curves: a = 1/ 5, ↑↓ of the system’s linear entropy and hSxi. The duration b = 2/ 5, f = −1, and the distribution of environmen- may vary somewhat for different system-environment tal frequencies, ωi, is taken to be random and centered ↑ coupling strengths or environmental frequency spec- around zero—to coincide with the random nature of He ↓ tra, but the overall presence of a significant period of and He in the ACL model [9]. quadratic time behavior is clear. Next, consider hSxi. For the early-time regime we simply have Eqn. 32, reprinted here: VI. DISCUSSION AND CONCLUSIONS hS i = ~ 2Re[ab∗] − 2β∆Im[ab∗] − 2∆2Re[ab∗η] x P 2 (52) In this paper we have identified and described unique with the quantities η and β as defined in Eqn. 45. For early time behavior of a quantum system interacting with the non-perturbative case we obtain: its environment—the copycat process. The copycat process is a new and potentially important addition to the ~ ∗ hS i = (2Re[ab z(t)]) (53) narrative of decoherence and einselection. By considering x NP 2 the evolution of the system density matrix from an eigen- with z(t) given by Eqn. 43. state perspective, we were able to recognize the early- Figure 3 shows the perturbative and non-perturbative time emergence of a distinct transiently stable “copycat” results for hSxi as a function of time. As with the lin- state, as illustrated in Fig. 1. We have derived the same ear entropy, the non-perturbative curve was generated effect analytically in Sect. III, and then utilize the solu- numerically in our simulations essentially by evaluating tions and their implications to obtain new insights into Eqn. 53 for each time step, while the early-time result is how small quantum systems einselect in Sects. IV and V. the analytical function in Eqn. 52 with the summations Furthermore, the comparison with our numerical work in for η and β in Eqn. 45 evaluated numerically. The values section V.B demonstrates that one might expect key fea- ↑↓ for a, b, f , and ωi are the same as in Fig. 2, and we tures of the copycat process to dominate for a significant take ~ = 1. portion of the full time to full einselection. We note here that the RCL model is not highly efficient The generality of our results in Sects. III and IV is at completing the process of einselection, as evidenced by also noteworthy. As we briefly commented in Sect. V, the small oscillation around zero of the numerical curve an intriguing part of the copycat process is that it is 9 agnostic about the spectrum of the environment. The Hamiltonian in Eqn. 7 used to derive the copycat results makes no assumption of any of the standard en- 1st eigenstate 2nd eigenstate vironmental spectra—such as “ohmic” environments— 1 0.2 t=0 2nd eigenstate not typically employed in the literature to make analytical determined at t=0 0.5 progress [1, 7], and it also does not assume the random 0.1 environment that we utilize for our numerical work in 0 0 section V.B and [9]. This suggests that the onset of ein- -10 0 10 0.2 0.4 0.6 0.8 1 selection could begin with the copycat process in a wide 0.4 -6 variety of cases. 0.2 t=10-6 t=10 0.2 While we acknowledge our analytical modelling of 0.1 system-environment interactions is fairly simplified in the 0 0 RCL model, our numerical work with the ACL model in -10 0 10 -10 0 10 [9] demonstrates that the copycat process persists even 0.4 -1 in the presence of strong self Hamiltonians of the system 0.2 t=10-1 t=10 and environment. Furthermore, we expect the copycat 0.2 0.1 process to be present in some form for larger and more complicated systems. The orthogonal nature of the copy- 0 0 -10 0 10 -10 0 10 cat eigenstates might be unsurprising for the two-state 1 system results—the small Hilbert space greatly limits 1 t=10 t=10 the possibilities—but we have also seen that the same 0.5 copycat behavior holds for superpositions of two coher- 0.5 ent states of an SHO. 0 0 When the evolution of the global (w) system is unitary, -10 0 10 -10 0 10 the evolution of a subsystem density matrix is always deterministic. Thus (except for the case of degenerate eigenvalues) the evolution the density matrix eigenstates is also deterministic. The copycat process is an example of a form of this deterministic evolution which generi- FIG. 4. The evolution of an initial eight-cat state, showing cally appears in two state systems, as well as some larger the top two eigenstates of ρs. The term “copycat” might not systems that are started in “two cat” states. be a great description of the second eigenstate, but we have found that the quadratic transient stability is still present. To explore further, in Appendix A we extend our ana- The process of einselection is essentially complete at the final lytical perturbative analysis to the case where the system time shown. We use the RCL model, with a d = 30 qudit is a qutrit. There we see that many of the same copycat system. The markers show the amplitude squared for each features appear, although the pattern of early quadratic basis vector, and the lines are added for illustrative purposes. behavior is partially broken by the possibility of linear evolution in the (2d) system subspace orthogonal to the original system state. tems appear in discussions of the “quantum Zeno para- Exploring further still, Fig. 4 shows the evolution of dox” [7, 19, 20], where the behavior of the system density a particular example with a higher dimensional system, matrix eigenstates is also typically not considered. set up in a form that might be thought of as “eight Generally, we have found that features of the evolution Schrödingercats.” In that case we did observe over- that might be associated with “decay,” or the onset of en- all quadratic behavior to leading order, resulting in the tanglement, are quadratic to leading order. Such features same transient stability. One would be hard pressed to are controlled by the eigenvalues of ρs, which are always describe the states that appear in the right column as quadratic (to leading order) in our results and in the lit- “copycat states,” but these are the states which emerge erature we cite here. We have also quite generally found from the deterministic Schrödinger evolution in the w aspects of the evolution which are linear at lowest order space for the particular chosen initial state. These results in time. These aspects describe evolution of the system give some sense how the equivalent process can look in a in ways not associated with the onset of entanglement. more complex situation. For example, the linear piece ∝ β in Eqns. 15 and 16 There are several connections between the results we describes the evolution of the relative phase between the present here and the existing literature. Early-time coefficients of the pointer states, and in Appendix A we quadratic decay of the off-diagonal elements of the sys- saw a real linear part to the evolution of the second and tem density matrix for decohering systems has been third eigenstates of ρs. mentioned to varying degrees in several places [1, 3, 5– We also note that many approaches to studying de- 7, 11, 14–18]. However, these references typically do not coherence and einselection—see reviews in [1, 7], for look explicitly at the eigenstates of ρs. Other explo- example—utilize a master equation approach to ana- rations of the early-time behavior of open quantum sys- lyze the time evolution of the system density matrix. 10

As discussed in [9], this master equation approach typi- Section III, in that we begin with an initial state cally carries with it assumptions of Markovian evolution |ψ (0)i = (a |1i + b |0i + c |−1i) |φ i (A1) and the resulting exponential decay of off-diagonal sys- w e tem density matrix elements. Even within the master with |a|2 + |b|2 + |c|2 = 1 and Hamiltonian of the form3 equation approach it is known that exponential decay H = λ |1i h1| H1 + |0i h0| H0 + |−1i h−1| H−1. (A2) is not always valid [1, 7, 21], however the exponential w e e e case remains the focus of much of the literature. There Following the same methods as Section III, we obtained are some exceptions to this focus. Zurek and collabo- the reduced density matrix rators [6, 11, 15] explicitly note the generally dominant  ρ ρ ρ  early quadratic behaviors and point out that, in the con- |1ih1| |1ih0| |1ih−1| ρs =  ρ|0ih1| ρ|0ih0| ρ|0ih−1|  (A3) text of the formalisms they develop, the exponential be- ρ ρ ρ havior is a very special case. And Peres [22] offers a |−1ih1| |−1ih0| |−1ih−1| general analysis of the diverse range of possible behav- with entries defined as: iors. Our approach in this paper and in our numerical ρ = aa∗ work [9] is agnostic of Markovian assumptions by simply |1ih1| ∗ 2 3 4 solving the Schrödingerequation directly. This has led ρ|1ih0| = ab 1 + ı∆β10 − ∆ η10 + ı∆ ν10 + ∆ κ10 ∗ 2 3 4 to us observing more complicated non-Markovian time ρ|1ih−1| = ac 1 + ı∆β1−1 − ∆ η1−1 + ı∆ ν1−1 + ∆ κ1−1 dependence in our system density matrix, including the ∗ 2 ∗ 3 ∗ 4 ∗ copycat regime. ρ|0ih1| = ba 1 − ı∆β10 − ∆ η10 − ı∆ ν10 + ∆ κ10 ∗ Looking forward, we are curious whether our calcula- ρ|0ih0| = bb tions of the copycat process could provide a useful tool for ∗ 2 3 4 ρ|0ih−1| = bc 1 + ı∆β0−1 − ∆ η0−1 + ı∆ ν0−1 + ∆ κ0−1 studying decoherence and einselection in open quantum ∗ 2 ∗ 3 ∗ 4 ∗ systems. Examining the time dependence of the system ρ|−1ih1| = ca 1 − ı∆β1−1 − ∆ η1−1 − ı∆ ν1−1 + ∆ κ1−1 ∗ 2 ∗ 3 ∗ 4 ∗ density matrix eigenstates and eigenvalues allows one to ρ|−1ih0| = cb 1 − ı∆β0−1 − ∆ η0−1 − ı∆ ν0−1 + ∆ κ0−1 see the system smoothly transition from an initial quan- ρ = cc∗. (A4) tum superposition to a classical mixture of pointer states, |−1ih−1| with the copycat process describing the first stages of this It ends up being necessary to calculate the reduced den- transition. sity matrix to O(∆4), in order to not lose information when calculating the eigenvalues and eigenstates to O(∆2). The β and η parameters are defined analogously VII. ACKNOWLEDGEMENTS to Eqns. 13 and 14, i.e.

λ j i We are grateful to F. Anza, N. Curro, P. Coles, B. βij = hφe| He |φei − hφe| He |φei (A5) ~ Nachtergaele, R. Singh, A. Sornborger, and Z. Wang for 2 i i j j λ hφe| HeHe |φei + hφe| He He |φei valuable conversations. This work was supported in part ηij = by the U.S. Department of Energy, Office of Science, Of- ~2 2 j i fice of High Energy Physics QuantISED program under − hφe| He He |φei (A6) Contract No. KA2401032. with ∗ 2 ij = ηij + ηij − βij (A7) Appendix A: The qutrit RCL and the additional third and forth order parameters, νij and κij, are defined according to: Here we extend our perturbative treatment of the RCL 3 i 3 j 3 model to the case where the single qubit system is re- λ hφe| (He) |φei − hφe| (He ) |φei νij = placed with a qutrit. This enables us, among other ~3 6 things, to study the evolution of three cat Schrödinger hφ | (Hj)2Hi |φ i − hφ | Hj(Hi)2 |φ i + e e e e e e e e (A8) cat initial states. 2 Our results show that the early time behavior of the 4 i 4 j 4 λ hφe| (He) |φei + hφe| (He ) |φei eigenvalues is quadratic to leading order, as we have al- κij = 4 24 ready shown analytically in the qubit case and have also ~ hφ | (Hj)3Hi |φ i + hφ | Hj(Hi)3 |φ i observed numerically in much larger systems. In many − e e e e e e e e respects the behavior of the eigenstates also reflects what 6 hφ | (Hj)2(Hi)2 |φ i we saw for the qubit case. However, we have identified + e e e e . (A9) circumstances where the leading behavior of the second 4 and third eigenstates has a real linear contribution, in contrast to the qubit case where the linear piece just showed up in a relative phase. 3 Note that the “−1” superscript here is an index, not an inverse The derivation is nearly identical to the treatment in operation. 11

Because this 3 x 3 system reduced density matrix is Her- given mitian, general analytical solutions for the eigenvalues p 2 2 and eigenstates exist [23, 24]. Using these exact solutions N0 = 1 + |x0| + |y0| (A20) as a starting point, we then performed another sequence 2 2 ∗ ∗ N2 = |x1| + |y1| + 2Re[x0x2] + 2Re[y0y2 ] (A21) of series expansions for the small parameter t = ∆ to obtain analytic solutions. For the sequence of series ex- with 4 pansions we kept terms up to O(∆ ), only truncating the b 2 results to O(∆ ) at the end. As mentioned earlier, this y0 = (A22) c is essential to not lose information when calculating the 2 b eigenvalues and eigenstates to O(∆ )—for example, one y1 = β1−1 − β10 (A23) needs to keep up to O(∆4) to navigate the series expan- c h sion of the ratio involving a square root in Eqn. 9 of [23] b ∗ ∗ y2 = (β10 − β1−1)β1−1 + η1−1 − η10 correctly. The eigenvalues to lowest order in time are c 2 2 i given by: + |c| (δ1 − 0−1) + |b| δ1 (A24) 2 2 2 2 2 2 2 p1 = 1 − ∆ |a| |b| 10 + |a| |c| 1−1 + |b| |c| 0−1 and 2 = 1 − ∆ λ1 (A10) a x0 = (A25) 2 √ c ∆ a p2 = λ1 + Λ x = β (A26) 2 1 c 1−1 2 = ∆ λ2 (A11) a 1 h x = (1 − |c|2)( − η ) − λ − cb∗y ∆2 √ 2 c |a|2 1−1 1−1 1 2 p3 = λ1 − Λ 2 2 2 ∗ i + |b| β0−1 − 2β0−1β1−1 − 1−1 + η1−1 + η0−1 = ∆2λ (A12) 3 (A27) with where λ is deﬁned in Eqn. A10, y in Eqn. A24, and δ 2 2 2 2 2 2 2 1 2 1 Λ = λ1 + 4|a| |b| |c| |η0−1| + |η10| + |η1−1| is shorthand for:

+ β10β1−1Re[η0−1] + β0−1β1−1Re[η10] ∗ ∗ ∗ δ1 = β0−1β10 + η0−1 + η10 − η1−1. (A28) − β0−1β10Re[η1−1] − Re[η0−1η10] As with the two-state solutions, it is purely a matter of ∗ ∗ − Re[η1−1η0−1] − Re[η1−1η10] . (A13) our chosen convention (chosen for convenience) that the linear complex phase is present in |1i and |0i but not To obtain the above it is necessary to recognize that |−1i in Eqns. A17 - A19. (β0−1 + β10 − β1−1) = 0 (A14) Additional complexity is present for |ψ2i and |ψ3i. For these eigenstates, χ , γ , and ζ are defined as: generically, simply following from the definition in 2,3 2,3 2,3 Eqn. A5. This sets the leading order time dependence u0 h u1 M1 of the eigenvalues to be quadratic, as with the two-state χ2,3 = 1 + ı ∆ + 2 ∆ √ M0 u0 (M0) results earlier in this paper. Note the ± Λ part of 2 u2 u1M1 3(M1) M2 2i Eqns. A11 and A12 is what saves p2 and p3 from be- + + 2 − 4 − 2 ∆ ing degenerate at O(∆2). In the limit that any of the u0 2u0(M0) 8(M0) 2(M0) initial state coefficients a, b, or c are sent to zero, we ex- (A29) actly recover the two state eigenvalues given by Eqns. 17 v0 h v1 M1 γ2,3 = 1 + ı ∆ + 2 ∆ and 18 from Eqns. A10 - A13. M0 v0 (M0) The normalized eigenstate results have the general v v M 3(M )2 M i form: 2 1 1 1 2 2 + + 2 − 4 − 2 ∆ v0 2v0(M0) 8(M0) 2(M0) |ψ1i = χ1 |1i + γ1 |0i + ζ1 |−1i (A15) (A30) |ψ2,3i = χ2,3 |1i + γ2,3 |0i + ζ2,3 |−1i (A16) 2 1 h M1 3(M1) M2 2i ζ2,3 = 1 + 2 ∆ − 4 + 2 ∆ where for the top eigenstate χ, γ, and ζ are defined by: M0 (M0) 8(M0) 2(M0) (A31) x0 h x1 x2 1 N2 2i χ1 = 1 + ı ∆ + − 2 ∆ (A17) N0 x0 x0 2 (N0) given: y0 h y1 y2 1 N2 2i p 2 2 γ1 = 1 + ı ∆ + − 2 ∆ (A18) M0 = 1 + |u0| + |v0| (A32) N0 y0 y0 2 (N0) ∗ ∗ M1 = Im[u1u0] + Im[v1v0 ] (A33) 1 h 1 N2 2i ζ1 = 1 − 2 ∆ (A19) 2 2 ∗ ∗ N0 2 (N0) M2 = |u1| + |v1| + 2Re[u0u2] + 2Re[v0v2 ] (A34) 12 with These solutions for the eigenstates exhibit the in- creased complexity present in the three dimensional case. " 2 # b λ2,3 + |c| (δ1 − 0−1) The numerical factors in Eqns. A22 - A27 and Eqns. A35 v0 = 2 (A35) c λ2,3 − |b| δ1 - A40 showcase a complicated interplay between the en- " # vironmental factors βij, ηij, ij, νij, and κij for the three b 1 different states. The time dependence of the top eigen- v1 = 2 2 c (λ2,3 − |b| δ1) state, |ψ1i, is reminiscent of the two state solutions in " that the leading order real time dependence is quadratic × λ + |c|2(δ − )λ β − |b|2δ with a linear complex phase. However, note that for gen- 2,3 1 0−1 2,3 1−1 3 1 0 −1 eral He , He , and He the leading order real time depen- # dence for |ψ2i and |ψ3i is actually linear if one considers 2 2 Eqns. A29 - A31. However, this linear time dependence + λ2,3β10 − |c| δ2 |b| δ1 − λ2,3 (A36) will disappear if all the environmental factors βij, ηij, ij, " # and νij are purely real, due to the vanishing of Eqn. A33 b 1 in that limit. v2 = 2 3 c (λ2,3 − |b| δ1) In the case where the linear time dependence is present, " one can think of it this way: All aspects of the way in 2 2 2 × (λ2,3 − |b| δ1)(λ2,3β1−1 − |b| δ3)(λ2,3β10 − |c| δ2) which the initial state is being diminished are occurring at a quadratic rate (to leading order). But in the qutrit − (λ − |b|2δ )2(λ η∗ − |c|2δ ) case, the probability is flowing from the initial state into a 2,3 1 2,3 10 4 two dimensional subspace orthogonal to the initial state. h 2 2 2 − (λ2,3 − |c| (δ1 − 0−1)) (λ2,3β1−1 − |b| δ3) Under certain conditions it is possible for the description # of the system in this orthogonal subspace to move around i with a linear time dependence, even as the profile of the − (λ − |b|2δ )(|b|2δ + λ η∗ ) (A37) 2,3 1 5 2,3 1−1 evolution of the initial state remains quadratic.

and −1 h i u = |c|2 + v cb∗ (A38) 0 ca∗ 0 −1 h i u = |c|2β + cb∗v + v (β − β ) 1 ca∗ 1−1 1 0 1−1 0−1 (A39) " 1 u = λ + |c|2(η − ) 2 ca∗ 2,3 1−1 1−1 ∗ − cb v2 + v1(β0−1 − β1−1) # ∗ + v0(β0−1β1−1 + 1−1 − η1−1 − η0−1) (A40)

~~with the additional mixing parameters:~~

∗ ∗ δ2 = β1−1η0−1 − β0−1η1−1 + ν0−1 + ν10 − ν1−1 (A41) ∗ ∗ ∗ ∗ ∗ δ3 = β10η0−1 − β0−1η10 − ν0−1 − ν10 + ν1−1 (A42) ∗ ∗ δ4 = β1−1ν0−1 − β0−1ν1−1 + η0−1η1−1 ∗ ∗ + κ0−1 − κ10 + κ1−1 (A43) ∗ ∗ ∗ ∗ δ5 = β10ν0−1 + β0−1ν10 − η0−1η10 ∗ ∗ ∗ − κ0−1 − κ10 + κ1−1 (A44) where for |ψ2i one chooses λ2 deﬁned in Eqn. A11, with a similar convention for |ψ3i. 13

[1] M. Schlosshauer, Decoherence and the Quantum-To- [11] W. H. Zurek, Phys. Rev. D 26, 1862 (1982). Classical Transition (Springer, 2007). [12] W. H. Zurek, S. Habib, and J. P. Paz, Phys. Rev. Lett. [2] W. H. Zurek, Rev. Mod. Phys. 75, 715 (2003), 70, 1187 (1993). arXiv:quant-ph/0105127 [quant-ph]. [13] A. Caldeira and A. Leggett, Physica A 121, 587 (1983). [3] E. Joos and H. Zeh, Z. Physik B - Condensed Matter 59, [14] O. K¨ublerand H. Zeh, Annals of Physics 76, 405 (1973). 223 (1985). [15] W. H. Zurek, F. M. Cucchietti, and J. P. Paz, “Gaussian [4] J. Preskill, Quantum Information: Chapter 3, Lecture decoherence and gaussian echo from spin environments,” Notes for Ph219/CS219 (California Institute of Technol- (2006), arXiv:quant-ph/0611200 [quant-ph]. ogy, 2018). [16] G. M. Palma, K.-A. Suominen, and A. K. Ekert, [5] W. G. Unruh, Phys. Rev. A51, 992 (1995), arXiv:hep- Proc. Roy. Soc. Lond. A452, 567 (1996), arXiv:quant- th/9406058 [hep-th]. ph/9702001 [quant-ph]. [6] F. M. Cucchietti, J. P. Paz, and W. H. Zurek, Phys. [17] V. V. Dobrovitski, H. A. De Raedt, M. I. Katsnelson, Rev. A 72, 052113 (2005). and B. N. Harmon, Phys. Rev. Lett. 90, 210401 (2003). [7] H. P. Breuer and F. Petruccione, The Theory of Open [18] D. Braun, F. Haake, and W. T. Strunz, Phys. Rev. Lett. Quantum Systems (Oxford University Press, 2002). 86, 2913 (2001). [8] M. A. Nielsen and I. L. Chuang, Quantum Computa- [19] E. Joos, Phys. Rev. D 29, 1626 (1984). tion and Quantum Information: 10th Anniversary Edi- [20] A. Peres, Am. J. Phys. 48, 931 (1980). tion (Cambridge University Press, 2011). [21] J. R. Anglin, J. P. Paz, and W. H. Zurek, Physical Re- [9] A. Albrecht, R. Baunach, and A. Arrasmith, “Adapted view A 55, 4041–4053 (1997). Caldeira-Leggett model,” (2021), submitted simultane- [22] A. Peres, Annals of Physics 129, 33 (1980). ously with this paper to the arXiv. [23] C.-A. Deledalle, L. Denis, S. Tabti, and F. Tupin, [10] A. Albrecht, R. Baunach, and A. Arrasmith, “Einse- “Closed-form expressions of the eigen decomposition of 2 lection, equilibrium and cosmology,” (2021), submitted x 2 and 3 x 3 hermitian matrices,” (2017), hal-01501221. simultaneously with this paper to the arXiv. [24] J. Kopp, Int. J. Mod. Phys. C19, 523 (2008), arXiv:physics/0610206 [physics].