Quantum to classical transition induced by a classically small influence
Zhao Wen-Lei1, †, Jie Quanlin2
School of Science, Jiangxi University of Science and Technology, Ganzhou 341000, China
Department of Physics, Wuhan University, Wuhan 430072, China

 

† Corresponding author. E-mail: wlzhao@jxust.edu.cn

Project supported by the National Natural Science Foundation of China (Grant Nos. 11864014 and 11804130).

Abstract

We investigate the quantum to classical transition induced by two-particle interaction via a system of periodically kicked particles. The classical dynamics of particle 1 is almost unaffected in condition that its mass is much larger than that of particle 2. Interestingly, such classically weak influence leads to the quantum to classical transition of the dynamical behavior of particle 1. Namely, the quantum diffusion of this particle undergoes the transition from dynamical localization to the classically chaotic diffusion with the decrease of the effective Planck constant eff. The behind physics is due to the growth of entanglement in the system. The classically very weak interaction leads to the exponential decay of purity in condition that the classical dynamics of external degrees freedom is strongly chaotic.

1. Introduction

Rich dynamical behavior in periodically-driven systems has attracted extensive attention in recent years.[13] It is found that periodical driving induces quantized transport,[4,5] topological phase transition,[68] and thermalization of quantum dynamics,[9,10] just to name a few. Among these, the quantum transport in momentum space is particularly important, as it provides an insight on the mapping of transport phenomenon from position to momentum space. As an example, the dynamical localization (DL) in momentum space[1113] is an analog of the Anderson localization in position space,[14] both of which result from the disordered feature of the system.[15] Nowadays, the transport phenomenon in momentum-space lattice has been a fruitful subject in physics, where the intriguing phenomena, such as topologically-protected quantum walk[16] and exponentially-fast diffusion[1721] have been reported.

Theoretical investigations have stimulated experimental studies on the transport phenomenon in periodically-driven quantum systems. Recently, the DL was observed in the atom-optics experiments on the periodically-driven ultracold atoms[22] and molecules.[23] Even more remarkable, experimental advances in atom-optics have made it possible to actively tune the driven laser to be a quasiperiodic[2426] or random sequence,[27] and thus to investigate the diffusion behavior in a controlled manner, e.g., the Anderson transition for the quasiperiodical kicking systems, and the decoherence-induced subdiffusion for the random kicking systems. These investigations greatly broaden our understanding on the exotic transport phenomenon induced by quantum coherence.[2830]

It is known that the unavoidable coupling between system and environment destroys quantum coherence,[31] and consequently leads to the appearance of the classically chaotic diffusion from the underlying quantum dynamics.[3241] Interestingly, the influence from a single chaotic particle can effectively produce the decoherence effects.[4246] The quantum–classical correspondence (QCC) which is induced by the interaction with a few degrees of freedom[4750] has been recently observed in a ultracold atoms experiment.[51] In the past decades, the influence of interatomic interactions on the quantum diffusion has received considerable investigations.[5254] Rich diffusion behaviors, such as subdiffusion[5561] and exponentially-fast diffusion,[18,19,21] have been observed. This topic regains attention as it is closely associated with the issue of dynamical many-body localization,[6265] since the system of coupled periodically-kicking particles is an ideal system to explore the diffusion behavior in discrete momentum-space lattice.[6669] At present, the fate of dynamical many-body localization under the interaction of particles is still an open question.[70]

Motivated by these studies, we investigate the quantum diffusion in a system involving two-coupled particles. The system is periodically driven by impulsive fields. We concentrate on both the classical and quantum dynamics of one particle (say particle 1) under the interaction with the other one (say particle 2). Interestingly, the effects of particle 2 with mass m on the classical diffusion of particle 1 with unity mass decrease as m decreases, and it is negligibly small when m ≪ 1. The reason is that particle 2 of very small mass possesses little energy to affect the classical motion of particle 1. More importantly, the classically vanishing small influence is able to destroy the DL of particle 1. It even leads to the appearance of classically chaotic diffusion in quantum dynamics when the effective Planck constant eff is small enough. To characterize the quantum entanglement, we numerically investigate the purity of the quantum state. The issue we address here is the sensitivity of entanglement to the classical chaoticity of the external degrees of freedom. We find that the time dependence of purity exhibits the power law decay for the regular motion of the classical dynamics of particle 2, and the exponential decay for the chaotic motion.

The paper is organized as follows. In Section 2, we describe the system and show the QCC of particle 1 subjected to the influence from a very small particle. In Section 3, we give a purity description of quantum entanglement. Summary is present in Section 4.

2. The quantum and classical diffusions

We consider two interacting particles which are trapped in an infinitely square well and periodically kicked by optical lattices. The Hamiltonian reads

where Hi (i = 1,2) indicates the Hamiltonian of individual particles

with V(xi) being an infinite square well

and the interaction Hamiltonian HI takes the form

The variable pi is the momentum, xi is the coordinate, mi is the mass of individual particles, kl denotes the wave number, Vi is the amplitude of the kicking optical lattices with period T, and L is the width of the infinite square well. We consider the repulsive interaction for which the parameter σ is positive and the coupling strength is controlled by ε.

It is useful to introduce a set of scaled dimensionless units. Time is scaled by T, i.e., t′ = t/T, which means that t′ denotes the number of kicks. The canonical coordinate and momentum variables are redefined as and , which satisfy the commutation relation , with the effective Planck constant . The masses of the two particles are rescalled by m1, thereby a dimensionless mass is m = m2/m1. The kicking strength in dimensionless units takes the form . For the interaction term, the scaled coupling strength is , and . The dimensionless width of the infinite square well has the expression L′ = 2kl L, For clarity, from here on, we omit the superscript of prime in these variables. After rescalling, the dimensionless Hamiltonian is described as

where the Hamiltonian for the first particle is

the Hamiltonian for the particle 2 is

and that for the interaction term takes the form

In this system, the mass of particle 1 is unity and that of particle 2 is a dimensionless quantity m = m2/m1 which can be adjusted by modifying m1 and m2. This opens the opportunity for investigating the dynamics of particle 1 under the influence of particle 2 with tunable mass.

We concentrate on both the classical and quantum dynamics of particle 1. Numerically, we use the split-operator method to simulate the time evolution of the quantum dynamics. The initial state is a product state of the ground state of each particle, i.e., ψ(0) = ϕ1(0)ϕ2(0) with . The classical evolution of a trajectory is addressed by the fourth order Runge–Kutta Method. The initial condition of a trajectory (p1,x1,p2,x2) is such that the momentum of each particle is zero, p1 = p2 = 0, the coordinates x1 and x2 are uniformly distributed in the region [0,L].

The typical character of classically chaotic diffusion is the linear growth of the mean energy with time, i.e., 〈 p2(t) 〉c = Dc t, where 〈 ⋅ 〉c stands for the ensemble average over many trajectories and Dc is usually termed as the classical diffusion coefficient (CDC).[7174] This phenomenon is traditionally referred as normal diffusion. Our numerical investigations show that the mean energy of particle 1 linearly increases with time if the kick strength is strong enough (see the inset in Fig. 1). In our numerical simulations, the number of the classical trajectories is 104 and the evolution over one period is separated into 105 numbers of steps. In order to investigate the effects of particle 2 on the classical diffusion of particle 1, we numerically calculate the CDC of this particle, i.e., for different m. Here, tf is the total time during which one may numerically track the time evolution. In our numerical experiments, tf is on the scale of several hundreds of kicking periods, which can ensure the high precision of the numerical results due to the linear growth of mean energy with time. Interestingly, with the decrease of m, Dc decreases to the saturation level which almost equals with the CDC of the unperturbed case (see Fig. 1 for ε = 2 and 4). This demonstrates that the influence from particle 2 decreases as m decreases and it is even negligible if m ≪ 1. For weak coupling (e.g., ε = 0.1), the Dc of particle 1 has slight difference from that of the unperturbed case as m varies, since the two particles have little effects on each other for very weak coupling.

Fig. 1. Classical diffusion coefficient Dc of particle 1 versus m for ε = 0.1 (triangles), 2 (squares), and 4 (circles). The dashed line (in red) denotes the classical diffusion coefficient of unperturbed case, i.e., ε = 0. Inset: mean square of classical momentum of particle 1 versus time for m = 0.001. Dashed line (in red) denotes the fitting function of the form with Dc = 0.087. The parameters are K1 = 1.8, K2 = 0, λ = 10, and L = π.

We numerically investigate the quantum diffusion of particle 1 when its classical counterpart is almost unaffected which is ensured by the condition m ≪ 1. The quantum mean energy of this subsystem exhibits DL for large eff [e.g., eff = 0.05 in Fig. 2(a)]. Interestingly, the quantum diffusion of particle 1 undergoes the transition from DL to classically-chaotic diffusion as eff decreases. For small enough eff (e.g., for eff = 0.01), the quantum diffusion is in good agreement with its classical counterpart, which is a clear evidence of QCC. At first glance, such QCC happening for very small eff is trivial. However, by comparing with the quantum mean energy with ε = 0, one can find that the quantum diffusion of the unperturbed case is eventually suppressed by quantum coherence after long time evolution [see Fig. 2(a) for eff = 0.01]. Therefore, the interaction plays a key role for the appearance of QCC. To reveal rich features of quantum diffusion, we numerically investigate the probability density distribution in momentum space. Our numerical results show that for large eff [e.g., eff = 0.05 in Fig. 2(b)], the wave packet is exponentially localized in momentum space, which is a feature of DL. For small enough eff [e.g., eff = 0.01 in Fig. 2(b)], the momentum distribution is in a good agreement with the Gaussian function, which is a signature of the appearance classically chaotic diffusion. The change from the exponentially-localized profile of wavepackets to the Gaussian form reveals the quantum to classical transition.

Fig. 2. (a) Time dependence of the classical (red line) mean energy and quantum mean energy (black lines) of particle 1. From top to bottom, black solid lines correspond to eff = 0.01, 0.02, 0.03, 0.04, and 0.05, respectively. The parameters are K1 = 1.8, K2 = 0, m = 0.001, ε = 2, λ = 10, and L = π. For comparison, the dashed line (in blue) denotes the quantum mean energy of the unperturbed case, i.e., ε = 0 with K = 1.8 and eff = 0.01. (b) Momentum distribution at the time t = 500 with eff = 0.01 (blue curve) and 0.05 (black curve). Dash-dotted line (in red) indicates the fitting function of the Gaussian form |ψ1(p)|2∝ e−p2/ζ. Dashed line (in red) denotes the exponential fitting |ψ1(p)|2∝ e−|p|/ξ.

In chaotic situation, the motion of particle 2 of very small mass behaves like random noise. It is known that external noises can destroy quantum interference and thus leads to the classically chaotic diffusion.[7577] As a further step, we consider the case that the kick strength contains random noises, for which the Hamiltonian reads

where δK is random numbers uniformly distributed in the interval , and V(x) denotes an infinite square well. We numerically investigate the quantum diffusion for different . In numerical simulations, we take the average of the expectation value 〈 p2〉 over 50 realizations of random δK to reduce the fluctuations of the time dependence of the physical observable. Our numerical results show that for a specific eff the quantum diffusion exhibits the transition from DL to the classically chaotic diffusion with the increase of , as shown in Fig. 3(a). Note that the strength of the random noise is much smaller than that of the kick strength, i.e., , thus the chaotic diffusion emerging from quantum dynamics with is in good consistence with the classically normal diffusion of the unperturbed case. In the presence of random noise, the quantum diffusion also undergoes the quantum to classical transition with the decrease of eff [see Fig. 3(b) for ]. The comparison of the quantum states between the random kicking system with and the two-particle system with m = 0.001 shows the good agreement both for DL and chaotic diffusion, as shown in Fig. 3(c). It is therefore reasonable to believe that the influence from a small particle resembles random noises on the system.

Fig. 3. (a), (b) Time dependence of the mean energy. In (a), eff = 0.01, from top to bottom solid lines correspond to , 0.008, 0.005, and 0. In (b), , from top to bottom solid lines correspond to eff = 0.01, 0.02, 0.03, and 0.05. For comparison, red-dashed lines denote the classical mean energy with . (c) Momentum distribution at the time t = 500 with for eff = 0.01 and 0.05, respectively. The parameters are K = 1.8 and L = π. Black curves denote the momentum distributions for the two-particle systems with m = 0.001, K1 = 1.8, ε = 2, λ = 10, and L = π (same as in Fig. 1(b)).

We quantify the difference between quantum and diffusion coefficients of particle 1 by using the ratio = Dq/Dc, where the quantum diffusion coefficient is defined as . In the ideal case, the DL (Dq = 0) and QCC (Dq = Dc) correspond to = 0 and 1, respectively. In numerical simulations, we evolve the system for 500 periods to calculate both Dc and Dq, and thus obtain the value . Figure 4(a) shows the dependence of on eff for various m. It is seen that with increasing eff, the exponentially decays as ∝ eαℏeff from almost unity to the saturation level which is around = 0.1. This clearly demonstrates the exponentially-fast departure of the quantum dynamics from its classical limit. Note that the saturation region of ≈ 0.1 indicates the appearance of DL. Due to the finite time evolution in numerical simulations, the value of Dq for DL is not exactly zero, which results in the saturation of ≈ 0.1. Detailed observations show that the decay rate α of increases with the decrease of m, which reveals that the quantum dynamics of particle 1 is much easier to deviate from its classical counterpart if the mass of particle 2 gets smaller. This is due to the fact that the influence from particle 2 on particle 1 decreases as m decreases.

Fig. 4. (a) The ratio versus eff with m = 0.001 (triangles), 0.01 (circles), and 1 (squares). Dashed lines (in red) denote the exponential fitting, i.e., ∝ eαℏeff. (b) Phase diagram of quantum diffusion as a function of eff and m, where circles and triangles indicate the threshold values of and which correspond to the appearance of classically chaotic diffusion and dynamical localization, respectively. The coupling strength is ε = 2. Other parameters are the same as those in Fig. 2.

Furthermore, we define the criteria of the appearance of the classically chaotic diffusion and DL as = 0.9 and 0.1, respectively. We numerically obtain the threshold values of eff corresponding to = 0.9 and 0.1, which are separately termed as and . The dependence of and on m is depicted in Fig. 4(b). One can find three different regions in the parameter space (, m). Blow the curve of , there is the zone of QCC for which the quantum system exhibits chaotic diffusion. Above the curve , the quantum system exhibits the DL since eff is so large that the the quantum coherence is not destroyed by the interaction. There is a transition region between the classically chaotic diffusion zone and the DL zone. Our numerical result is useful for guiding the investigation on the diffusion behavior in cold-atom experiments.

3. A purity description of entanglement

A commonly used measure of quantum entanglement is purity , where ρ1 is the reduced density matrix of particle 1 by tracing out the degree of particle 2 from the density matrix of the two-particle system.[78,79] The sensitivity of the quantum entanglement to external perturbations has been extensively studied in the context of quantum chaos.[8082] In the present work, we numerically investigate the effects of the classically chaotic dynamics of particle 2 on the generation of quantum entanglement. We consider the case that the coupling strength is in the form , for which the coupling is almost negligible in the semiclassical limit eff ≪ 1. Our numerical results show that for small K2, the purity decays in the power law of time, i.e., (see Fig. 5). For strong kick strength (e.g. K2 = 0.09 and 0.12), the purity decays exponentially fast with time, i.e., . Moreover, the time-dependence of purity is almost unchanged as long as the K2 is sufficiently strong. Our numerical result demonstrates that there is a regime where the exponentially-fast growth of entanglement is stable provided that the chaoticity of the external degree of freedom is strong enough.

Fig. 5. Purity versus time with K2 = 0 (circles), 0.03 (triangles), 0.05 (diamonds), 0.09 (pentagrams), and 0.12 (squares). Dashed lines (in blue) denote the fitting function of the form . Dashed-dotted (in red) line indicates the exponential fitting . The parameters are K1 = 5, m = 0.01, = 10−4, ε = 20, λ = 10, and L = π.
4. Summary

We numerically investigate the entanglement involving a few degrees of freedom via two coupled particles. We show that, the effect of particle 2 on the classical behavior of particle 1 decreases as its mass (m) decreases. Under the classically weak perturbation (m ≪ 1), the quantum diffusion behavior of particle 1 undergoes a transition from DL to chaotic diffusion with the decrease of eff. We numerically investigate the difference between quantum and classical diffusions for a wide regime of eff and m, and find the exponential decay of with eff. By using this quantity, we define the boundary for the appearance of DL and classically-chaotic diffusion. Numerically, we obtain a “phase” diagram of the quantum diffusion in the parameters space (eff, m). The quantum to classical transition is accompanied by the growth of entanglement. For the vanishingly small interaction ε with eff ≪ 1, the time decay of purity is exponentially fast in condition that the classical dynamics of particle 2 is strongly chaotic. Such exponentially-fast entanglement is stable in the sense that the time dependence of purity is almost unchanged as K2 varies. Our investigation has important implication for the fundamental problem of the quantum to classical transition.

Reference
[1] Dana I 2014 Phys. Rev. 89 205111
[2] Ben-Harush M Dana I 2016 Phys. Rev. 93 052207
[3] Wang J Gong J B 2009 Phys. Rev. Lett. 102 244102
[4] Ho D Y H Gong J B 2012 Phys. Rev. Lett. 109 010601
[5] Zhao W L Wang J Z Wang X H Tong P Q 2019 Phys. Rev. 99 042201
[6] Zhou L W Chen C Gong J B 2016 Phys. Rev. 94 075443
[7] Li Z Wang H Q Zhang D W Zhu S L Xing D Y 2016 Phys. Rev. 94 043617
[8] Zhou L W Gong J B 2018 Phys. Rev. 97 063603
[9] Borgonovi F Izrailev F M Santos L F Zelevinsky V G 2016 Phys. Rep. 626 1
[10] Neill C Roushan P Fang M et al. 2016 Nat. Phys. 12 1037
[11] Casati G Chirikov B V Izraelev F M Ford J 1979 Stochastic Behavior in Classical and Quantum Hamiltonian Systems Lecture Notes in Physics Berlin Springer 334 10.1007/BFb0021732
[12] Izrailev F M 1990 Phys. Rep. 196 299
[13] Dana I 2014 Can. J. Chem. 92 77
[14] Anderson P W 1958 Phys. Rev. 109 1492
[15] Fishman S Grempel D R Prange R E 1982 Phys. Rev. Lett. 49 509
[16] Dadras S Gresch A Groiseau C Wimberger S Summy G S 2018 Phys. Rev. Lett. 121 070402
[17] Wang J Guarneri I Casati G Gong J B 2011 Phys. Rev. Lett. 107 234104
[18] Mieck B Graham G 2005 J. Phys. A: Math. Gen. 38 L139
[19] Zhao W L Gong J B Wang W G Casati G Liu J Fu L B 2016 Phys. Rev. 94 053631
[20] Zhao W L Wang J Z Wang W G 2019 J. Phys. A: Math. Theor. 52 305101
[21] Guarneri I 2017 Phys. Rev. 95 032206
[22] Lopez M Clement J F Szriftgiser P Garreau J C Delande D 2012 Phys. Rev. Lett. 108 095701
[23] Bitter M Milner V 2017 Phys. Rev. Lett. 118 034101
[24] Chabé J Lemarie G Gremaud B Delande D Szriftgiser P Garreau J C 2008 Phys. Rev. Lett. 101 255702
[25] Lemarié G Chabé J Szriftgiser P Garreau J C Grémaud B Delande D 2009 Phys. Rev. 80 043626
[26] Garreau J C 2017 C. R. Phys. 18 31
[27] Sarkar S Paul S Vishwakarma C Kumar S Verma G Sainath S Rapol U D Santhanam M S 2017 Phys. Rev. Lett. 118 174101
[28] Hainaut C Manai I Chicireanu R Clément J Zemmouri S Garreau J C 2017 Phys. Rev. Lett. 118 184101
[29] Tan J T Luo Y R Zhou Z Hai W H 2016 Chin. Phys. Lett. bf 33 070302
[30] Li Z Wang J Z Fu L B 2013 Chin. Phys. Lett. 30 010301
[31] Yang Y B Wang W G 2015 Chin. Phys. Lett. 32 030301
[32] Joos E Zeh H D Kiefer C Giulini D Kupsch J Stamatescu I O 2003 Decoherence and the Appearance of a Classical World in Quantum Theory Berlin Springer 41 10.1007/978-3-662-03263-3
[33] Zurek W H 2003 Rev. Mod. Phys. 75 715
[34] Schlosshauer M 2004 Rev. Mod. Phys. 76 1267
[35] Schlosshauer M 2008 Found. Phys. 38 796
[36] Chaudhry A Z Gong J B 2014 Phys. Rev. 89 014104
[37] Wang W G He L W Gong J B 2012 Phys. Rev. Lett. 108 070403
[38] Feynman R P Vernon F L 1963 Ann. Phys. (Leipzig) 24 118
[39] Caldeira A O Leggett A J 1983 Physica (Amsterdam) 121A 587
[40] Wisniacki D Toscano F 2009 Phys. Rev. 79 025203(R)
[41] Petitjean C Jacquod P 2006 Phys. Rev. Lett. 97 194103
[42] Rossini D Benenti G Casati G 2006 Phys. Rev. 74 036209
[43] Bandyopadhyay J N 2009 Europhys. Lett. 85 50006
[44] Adachi S Toda M Ikeda K 1988 Phys. Rev. Lett. 61 659
[45] Graham R Kolovsky A R 1996 Phys. Rev. A 222 47
[46] Park H K Kim S W 2003 Phys. Rev. 67 060102(R)
[47] Takahashi K 1989 Prog. Theor. Phys. Suppl. 98 109
[48] Zhao W L Jie Q L 2009 Commun. Theor. Phys. 51 465
[49] Zhao W L Jie Q L Zhou B 2010 Commun. Theor. Phys. 54 247
[50] Zhao W L Wang J Z Dou F Q 2012 Acta Phys. Sin. 61 240302 in China
[51] Gadway B Reeves J Krinner L Schneble D 2013 Phys. Rev. Lett. 110 190401
[52] Shepelyansky D L 1994 Phys. Rev. Lett. 73 2607
[53] Borgonovi F Shepelyansky D L 1995 Nonlinearity 8 877
[54] Qin P Q Andreanov A Park H C Flach S 2017 Sci. Rep. 7 41139
[55] Shepelyansky D L 1993 Phys. Rev. Lett. 70 1787
[56] García-Mata I Shepelyansky D L 2009 Phys. Rev. 79 026205
[57] Flach S Krimer D O Skokos C 2009 Phys. Rev. 79 026205
[58] Veksler H Krivolapov Y Fishman S 2009 Phys. Rev. 80 037201
[59] Pikovsky A Fishman S 2011 Phys. Rev. 83 025201
[60] Michaely E Fishman S 2012 Phys. Rev. 85 046218
[61] Fishman S Krivolapov Y Soffer A 2012 Nonlinearity 25 53
[62] Rozenbaum E B Galitski V 2017 Phys. Rev. 95 064303
[63] Notarnicola S Iemini F Rossini D Fazio R Silva A Russomanno A 2018 Phys. Rev. 97 022202
[64] Keser A C Ganeshan S Refael G Galitski V 2016 Phys. Rev. 94 085120
[65] Čadež T Mondaini R Sacramento P D 2017 Phys. Rev. 96 144301
[66] Gadway B 2015 Phys. Rev. 92 043606
[67] Alex An F Meier E J Ang’ong’a J Gadway B 2018 Phys. Rev. Lett. 120 040407
[68] Alex An F Meier E J Gadway B 2017 Nat. Comm 8 325
[69] Meier E J Alex An F Gadway B 2016 Phys. Rev. 93 051602(R)
[70] Abanin D A Roeck W D Huveneers F 2015 Phys. Rev. Lett. 115 256803
[71] Hu B Li B W Liu J Gu Y 1999 Phys. Rev. Lett. 82 4224
[72] Liu J Cheng W T Cheng C G 2000 Commun. Theor. Phys. 33 15
[73] Chen H S Wang J Gu Y 2000 Chin. Phys. Lett. 17 85
[74] Paul S Pal H Santhanam M S 2016 Phys. Rev. A 93 060203(R)
[75] Milner V Steck D A Oskay W H Raizen M G 2000 Phys. Rev. E. 61 7223
[76] Karney C F F Rechester A B White R B 1982 Physica D 3 425
[77] Ott E Antonsen T M Jr Hanson J D 1984 Phys. Rev. Lett. 53 2187
[78] Lakshminarayan A 2001 Phys. Rev. 64 036207
[79] Bandyopadhyay J N Lakshminarayan A 2002 Phys. Rev. A 89 060402
[80] Cohen D Heller E J 2000 Phys. Rev. Lett. 84 2841
[81] Wisniacki D A Cohen D 2002 Phys. Rev. 66 046209
[82] Wisniacki D A Ares N Vergini E G 2010 Phys. Rev. Lett. 104 254101