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

Scaling laws of the out-of-time-order correlators at the transition to the spontaneous 𝒫𝒯\cal{PT}-symmetry breaking in a Floquet system

Wen-Lei Zhao [email protected] School of Science, Jiangxi University of Science and Technology, Ganzhou 341000, China    Ru-Ru Wang School of Science, Jiangxi University of Science and Technology, Ganzhou 341000, China    Han Ke School of Science, Jiangxi University of Science and Technology, Ganzhou 341000, China    Jie Liu [email protected] Graduate School, China Academy of Engineering Physics, Beijing 100193, China HEDPS, Center for Applied Physics and Technology, and College of Engineering, Peking University, Beijing 100871, China
Abstract

We investigate both numerically and analytically the dynamics of out-of-time-order correlators (OTOCs) in a non-Hermitian kicked rotor model, addressing the scaling laws of the time dependence of OTOCs at the transition to the spontaneous 𝒫𝒯\mathcal{PT} symmetry breaking. In the unbroken phase of 𝒫𝒯\mathcal{PT} symmetry, the OTOCs increase monotonically and eventually saturate with time, demonstrating the freezing of information scrambling. Just beyond the phase transition points, the OTOCs increase in the power-laws of time, with the exponent larger than two. Interestingly, the quadratic growth of OTOCs with time emerges when the system is far beyond the phase transition points. Above numerical findings have been validated by our theoretical analysis, which provides a general framework with important implications for Floquet engineering and the information scrambling in chaotic systems.

pacs:
03.65.-w, 03.65. YZ, 05.45.-a, 05.45.Mt

I Introduction

Non-Hermiticity has been regarded as a fundamental modification to the conventional quantum mechanics Markum99 ; Stephanov99 ; Berry2004 ; Graefe2008 ; Rotter2009 ; Ott2013 ; Harsh2014 , a subclass of which with 𝒫𝒯\cal{PT} symmetry even displays the transition from the real energy spectrum to complex one. Such intrinsic spontaneous 𝒫𝒯\cal{PT}-symmetry breaking occurs at the exceptional points (EP), at which both the eigenstates and eigenvalues coalesce Bender1998 ; Bender2002 ; Mosta2002 ; Bender2007 ; Mosta2010 ; Harsh2010 ; Ganainy2018 ; Klaiman2008 ; Musslimani2008 . The existence of EP leads to rich physics, such as the enhancement of precision in quantum sensors Cai20 , the topological phase transition Bergholtz21 ; ZhongWang18 ; Mzhao1 ; FYu ; Mzhao2 , the nonadiabatic transition FengYu21 ; WYWang22 , and the unidirectional propagation of light YinHuang , just to name a few. Theoretical advances have enabled exponential realizations of 𝒫𝒯\cal{PT} symmetric systems in various fields, such as optical settings Sergi2014 ; Makris2008 ; Ganainy2007 ; Guo2009 ; Regensburger2012 ; Hodaei2014 ; Feng2014 ; Jiahua2016 ; Longhi2009 ; Longhi2010 ; Ruter2010 ; YongmeiXue , electronic circuits Stegmaier2021 , and optomechanical systems Luxy15 . Moreover, the extension of Floquet-driven systems to the 𝒫𝒯\mathcal{PT} symmetric regime has opened up unique opportunities for understanding fundamental concepts such as quantum chaos Bender09 and quantum-classical transition Bender10 ; wlzhao23a . Interestingly, chaos is found to facilitate the scaling law of the spontaneous 𝒫𝒯\mathcal{PT} symmetry breaking in a 𝒫𝒯\mathcal{PT} symmetric kicked rotor (PTKR) model West2010 . This system even displays ballistic energy diffusion Longhi2017 and the quantized acceleration of momentum current Zhao19 , which enriches our understanding on the unique transport phenomena in the presence of chaos.

The dynamics of OTOCs, originally introduced by Lakin et al., in the study of quasiclassical theory of superconductivity Larkin1969 , has received extensive studies in the fields of high energy physics Roberts2016 ; Maldacena2016 ; Polchinski2016 , condensed matter physics Bohrdt2016 ; Garttner2017 ; Banerjee2017 ; Shen2017 ; Fan2017 ; Huang2017 and quantum information Li2017 ; Weinstein22 ; Hu23 . It has been found that OTOCs can effectively detect quantum chaos Pappalardi22 ; GMata18 ; JiaoziWang21 ; Kidd2021 , quantum thermalization Balachandran21 , and information scrambling Harris2022 ; Roberts2022 ; Zhang2019 ; Yan2020 ; Patel2017 ; WLZhao23 . In the semiclassical limit, the exponential growth of OTOCs is governed by the Lyapunov exponent of classical chaos, which demonstrates a route of quantum-classical correspondence Hashimoto2017 . In Floquet-driven systems, OTOCs has been used to diagnose dynamical quantum phase transition Zamani2022 and entanglement Martin2018 ; Lewis-Swan2019 . Intrinsically, we previously found a quantized response of OTOCs when varying the kicking potential of the PTKR model WlZhao22 . State-of-art experimental advances have observed different kinds of OTOCs in the setting of nuclear magnetic resonance Wei2018 ; Nie2019 , trapping ions Landsman2019 and qubit under Floquet engineering Zhao2021 .

In this context, we both numerically and analytically investigate the dynamics of OTOCs when the PTKR model is in different phases of 𝒫𝒯\cal{PT}-symmetry. We use a machine learning method, namely a long short-term memory network (LSTM), to classify the phase diagram of 𝒫𝒯\cal{PT}-symmetry breaking and extract the phase boundary in a wide range of system parameters. We find that in the unbroken phase of 𝒫𝒯\mathcal{PT} symmetry, OTOCs increase monotonically with time evolution and eventually saturate, demonstrating the freezing of operator growth. We analytically prove that the saturation of OTOCs is a power-law function of the real part of the kicking potential. In the broken phase of the 𝒫𝒯\mathcal{PT} symmetry, we find a power-law increase of OTOCs with time, for which the characteristic exponent is larger than two when the system is just beyond the phase transition point, and is equal to two for the system far beyond the phase transition point. Through the detailed analysis of the wavepacket’s dynamics in the time reversal process, we uncover the mechanisms of both the dynamical localization and the power-law increase of OTOCs. Our investigations reveal that the dynamics of OTOCs can be utilized to diagnose spontaneous 𝒫𝒯\cal{PT}-symmetry breaking.

The paper is organized as follows. In Sec. II, we describe the PTKR model and show the scaling-law of spontaneous 𝒫𝒯\cal{PT}-symmetry breaking. In Sec. III, we show the scaling-laws of the dynamics of OTOCs at the transition to the 𝒫𝒯\cal{PT}-symmetry breaking. Sec. IV contains the theoretical analysis of the scaling-laws of OTOCs. The conclusion and discussion are presented in Sec. V.

II Transition to spontaneous 𝒫𝒯\cal{PT}-symmetry breaking in Floquet systems

II.1 Model

The Hamiltonian of the PTKR model in dimensionless unites reads

H=p22+V(θ)nδ(ttn),\textrm{H}=\frac{p^{2}}{2}+V(\theta)\sum_{n}\delta(t-t_{n})\;, (1)

where the kicking potential V(θ)=K[cos(θ)+iλsin(θ)]V(\theta)=K\left[\cos(\theta)+i\lambda\sin(\theta)\right] satisfies the 𝒫𝒯\cal{PT}-symmetric condition V(θ)=V(θ)V(\theta)=V^{*}(-\theta) Longhi2017 ; West2010 ; Zhao19 . The parameters KK and λ\lambda indicate the strength of the real and imaginary parts of the kick potential, respectively. The p=ieff/θp=-i\hbar_{\text{eff}}{\partial}/{\partial\theta} is angular momentum operator, θ\theta is the angle coordinate, and eff\hbar_{\text{eff}} denotes the effective planck constant. The time tn(=0,1,2)t_{n}(=0,1,2\ldots) is integer, indicating kicking numbers. The eigenequation of angular momentum operator is p|φn=neff|φnp|\varphi_{n}\rangle=n\hbar_{\text{eff}}|\varphi_{n}\rangle with eigenstate θ|φn=einθ/2π\langle\theta|\varphi_{n}\rangle=e^{in\theta}/\sqrt{2\pi} and eigenvalue pn=neffp_{n}=n\hbar_{\text{eff}}. On the basis of |φn|\varphi_{n}\rangle, an arbitrary quantum state can be expanded as |ψ=nψn|φn|\psi\rangle=\sum_{n}\psi_{n}|\varphi_{n}\rangle.

For time-periodic systems, i.e., H(t+T)=H(t)\textrm{H}(t+T)=\textrm{H}(t), the Floquet theory predicts the eigenequation of the evolution operator U|ψε=eiε|ψεU|\psi_{\varepsilon}\rangle=e^{-i\varepsilon}|\psi_{\varepsilon}\rangle, where the eigenphase ε\varepsilon is referred to as quasienergy. One-period time evolution of a quantum state of the PTKR system is given by |ψ(tj+1=U|ψ(tj|\psi(t_{j+1}\rangle=U|\psi(t_{j}\rangle with the Floquet operator

U=UfUK=exp(ieffp22)exp[ieffV(θ)].U=U_{f}U_{K}=\exp\left(-\frac{i}{\hbar_{\text{eff}}}\frac{p^{2}}{2}\right)\exp\left[-\frac{i}{\hbar_{\text{eff}}}V(\theta)\right]\;. (2)

This demonstrates that in numerical simulations, one period evolution is split into two steps, namely the kicking evolution UKU_{K} and the free evolution UfU_{f}. The kick evolution is realized in angle coordinate space, i.e., ψ(θ)=UK(θ)ψ(θ,tj)\psi^{\prime}(\theta)=U_{K}(\theta)\psi(\theta,t_{j}). Then, one can utilize the fast Fourier transform to change the state ψ(θ)\psi^{\prime}(\theta) to angular momentum space, thereby obtaining its component ψn\psi^{\prime}_{n} on the eigenstate |n|n\rangle. Finally, the free evolution is conducted in angular momentum space, i.e., ψn(tj+1)=Uf(pn)ψn\psi_{n}(t_{j+1})=U_{f}(p_{n})\psi^{\prime}_{n}. By repeating the same procedure, one can get the quantum state at arbitrary time Casati79 .

II.2 Spontaneous 𝒫𝒯\cal{PT}-symmetry breaking

It is straightforward to prove that the Floquet operator of the PTKR satisfies the 𝒫𝒯\cal{PT} symmetry U=(𝒫𝒯)U𝒫𝒯U=(\mathcal{P}\mathcal{T})^{\dagger}U\mathcal{P}\mathcal{T}, where 𝒫\cal{P} and 𝒯\cal{T} are the parity and time reversal operators, respectively. Based on conventional understanding of quantum mechanics, one asserts that the two operators, i.e., UU and 𝒫𝒯\cal{PT} have simultaneous eigenstates, that is to say, the quasieigenstate |ψε|\psi_{\varepsilon}\rangle is also the eigenstate of the 𝒫𝒯\cal{PT} operator, i.e., 𝒫𝒯|ψε=±|ψε\cal{PT}|\psi_{\varepsilon}\rangle=\pm|\psi_{\varepsilon}\rangle. This conclusion is indeed valid for positive quasienergies ε>0\varepsilon>0. However, a notable feature of the PTKR system is that complex quasienergies ε=εr±εi\varepsilon=\varepsilon_{r}\pm\varepsilon_{i} emerge when the strength of the imaginary part of the complex potential exceeds a threshold value, i.e., λ>λc\lambda>\lambda_{c} Longhi2017 ; West2010 ; Zhao19 . The threshold value λc\lambda_{c} is just the exceptional points of the system. It can be proven that the quasieigenstate |ψε|\psi_{\varepsilon}\rangle is no longer an eigenstate of the 𝒫𝒯\cal{PT} operator due to the complex quasienergies, thus demonstrating the spontaneous 𝒫𝒯\mathcal{PT}-symmetry breaking. An intrinsic quality of the PTKR system is that 𝒫𝒯\cal{PT} symmetry is helpful in protecting the real spectrum of the Floquet operator.

We assume that the initial state is expanded as |ψ(t0)=ερε|ψε|\psi(t_{0})\rangle=\sum_{\varepsilon}\rho_{\varepsilon}|\psi_{\varepsilon}\rangle. Then, after the nnth kick, the quantum state has the expression |ψ(tn)=Utn|ψ(t0)=ερεeiεrtneεitn|ψε|\psi(t_{n})\rangle=U^{t_{n}}|\psi(t_{0})\rangle=\sum_{\varepsilon}\rho_{\varepsilon}e^{-i\varepsilon_{r}t_{n}}e^{\varepsilon_{i}t_{n}}|\psi_{\varepsilon}\rangle, whose norm 𝒩=ψ(tn)|ψ(tn)\mathcal{N}=\langle\psi(t_{n})|\psi(t_{n})\rangle exponentially increases with time due to positive εi\varepsilon_{i}. We numerically investigate the time evolution of 𝒩\cal{N} for different λ\lambda. Without loss of generality, we choose a Gaussian wavepacket, i.e., ψ(θ,t0)=(σ/π)1/4exp(σθ2/2)\psi{(\theta,t_{0})}=(\sigma/\pi)^{1/4}\exp(-\sigma\theta^{2}/2) with σ=10\sigma=10 as the initial state in numerical simulations. Figure 1(a) shows that for very small λ\lambda (e.g., λ=0.01\lambda=0.01 and 0.05), the value of 𝒩\mathcal{N} equals almost to unity with time evolution, which implies that quasienergies are all real. Interestingly, for sufficiently large λ\lambda (e.g., λ=0.15\lambda=0.15), 𝒩\mathcal{N} increases exponentially with time, i.e., 𝒩=eμt\mathcal{N}=e^{\mu t}, and the growth rate μ\mu increases with the increase of λ\lambda. The non-unitary feature of the Floquet operator, UK=exp[Kλsin(θ)/]U_{K}=\exp[K\lambda\sin(\theta)/\hbar], leads to the growth of the norm. A rough estimation of the norm yields a time dependence of the form 𝒩exp(Kλt/)\mathcal{N}\propto\exp(K\lambda t/\hbar), indicating the relation of the growth rate μλ\mu\propto\lambda, which is confirmed by our numerical results [see inset in Figure 1(a)]. We further investigate the long-time average value of the norm, 𝒩¯=n=1N𝒩(tn)/N\bar{\mathcal{N}}=\sum_{n=1}^{N}\mathcal{N}(t_{n})/N, for a wide range of λ\lambda. Figure 1(b) shows that, for a specific eff\hbar_{\text{eff}} (e.g., eff=0.1\hbar_{\text{eff}}=0.1), 𝒩¯\bar{\mathcal{N}} remains at unity for λ\lambda smaller than a threshold value λc\lambda_{c}, beyond which it monotonically increases with λ\lambda. It is reasonable to believe that the threshold value λc\lambda_{c} corresponds to the emergence of spontaneous 𝒫𝒯\cal{PT}-symmetry breaking.

Refer to caption
Figure 1: (a) Norm 𝒩\cal{N} versus time for λ=0.01\lambda=0.01 (squares), 0.05 (circles), 0.15 (triangles), 0.2 (diamonds), and 0.3 (pentagrams). The parameters are K=5K=5 and eff=1\hbar_{\text{eff}}=1. Solid lines indicate the exponential increase, i.e., 𝒩(t)=eμt\mathcal{N}(t)=e^{\mu t}, while dash-dotted line denotes 𝒩=1\mathcal{N}=1. Inset: the μ\mu versus λ\lambda. Solid line indicates the linear increase μλ\mu\propto\lambda. (b) Average value 𝒩¯\bar{\cal{N}} versus λ\lambda for K=5K=5 with eff=0.1\hbar_{\text{eff}}=0.1 (squares), 0.5 (circles), and 1 (triangles). Solid line denotes 𝒩=1\mathcal{N}=1. (c) Phase diagram of the spontaneous 𝒫𝒯\cal{PT}-symmetry breaking for eff=1\hbar_{\text{eff}}=1. One can see clearly a phase boundary λc\lambda_{c}. (d) The value of λc\lambda_{c} in the parameter space (K,eff)(K,\hbar_{\text{eff}}).

Recently, the long short-term memory (LSTM) network has been exploited to extract the character of time series and thus to predict the phase diagram of quantum diffusion Mano21 . Based on the character of the time evolution of 𝒩\mathcal{N}, we conducted supervised training on the LSTM network and used it to evaluate the feature of 𝒩(t)\mathcal{N}(t), namely, whether 𝒩(t)=eμt\mathcal{N}(t)=e^{\mu t} or not, for different system parameters. This highly effective machine learning method outputs the probability ρ\rho of the time series 𝒩(t)\mathcal{N}(t) to be exponentially increasing or not, which can predict the phase diagram of spontaneous 𝒫𝒯\mathcal{PT}-symmetry breaking. Interestingly, our results show that the ρ\rho increases with the increase of both KK and λ\lambda [see Fig. 1(c)]. We identify two phases in the parameters space (K,λ)(K,\lambda), the boundary of which is clearly visible in Fig. 1(c). We further investigate the λc\lambda_{c} for different KK and eff\hbar_{\text{eff}}. Our results demonstrate that the critical parameter λc\lambda_{c} increases with the increase of eff\hbar_{\text{eff}} and decreases with the increase of KK [see Fig. 1(d)]. This behavior is rooted in the fact that the mean spacing level Δ\Delta of the quasienergies of the quantum kicked rotor (QKR) model is proportional to /K\hbar/K West2010 . The smaller the Δ\Delta is, the easier it is for the non-Hermitian parameter λ\lambda to cause the coalescence of two quasienergies, implying the relation λc/K\lambda_{c}\propto\hbar/K.

III Scaling laws of the OTOCs at the transition to the spontaneous 𝒫𝒯\cal{PT}-symmetry breaking

The OTOCs are defined by C(tn)=[A^(tn),B^]2C(t_{n})=-\langle[\hat{A}(t_{n}),\hat{B}]^{2}\rangle, with the operators A^(tn)=U(tn)AU(tn)\hat{A}(t_{n})=U^{\dagger}(t_{n})AU(t_{n}) and BB being evaluated in the Heisenberg picture Chen2017 ; Hashimoto2017 ; Dora2017 ; Mata2018 ; Alavirad2018 ; Swingle2016 ; Hafezi2016 ; Li2017 ; Garttner2017 . The average, i.e., =ψ(t0)||ψ(t0)\langle\cdots\rangle=\langle\psi(t_{0})|\cdots|\psi(t_{0})\rangle, is taken over an initial state |ψ(t0)|\psi(t_{0})\rangle Heyl2018 . In this work, we consider the case where both A^\hat{A} and B^\hat{B} are angular momentum operators, i.e., C(t)=[p(t),p]2C(t)=-\langle[{p}(t),{p}]^{2}\rangle. We use a Gaussian wavepacket, i.e., ψ(θ,t0)=(σ/π)1/4exp(σθ2/2)\psi{(\theta,t_{0})}=(\sigma/\pi)^{1/4}\exp(-\sigma\theta^{2}/2) with σ=10\sigma=10 as the initial state. It is worth noting that, as opposed to static-lattice systems, periodically-driven systems have no thermal states, as the temperature grows to infinity with time evolution D'Alessio14 . Thus, there is no need to average over the initially thermal states in the definition of C(tn)C(t_{n}) in our system WlZhao22 ; WLZhao21 .

Straightforward derivation yields the equivalence

C(tn)=\displaystyle C(t_{n})= C1(tn)+C2(tn)2Re[C3(tn)],\displaystyle C_{1}(t_{n})+C_{2}(t_{n})-2\text{Re}\left[C_{3}(t_{n})\right]\;, (3)

where the two-points correlators, namely, the first two terms in right side are defined by

C1(tn)=ψR(t0)|p2|ψR(t0),C_{1}(t_{n})=\langle\psi_{R}(t_{0})|{p}^{2}|\psi_{R}(t_{0})\rangle\;, (4)
C2(tn)=φR(t0)|φR(t0),C_{2}(t_{n})=\langle\varphi_{R}(t_{0})|\varphi_{R}(t_{0})\rangle\;, (5)

and the four-points correlator is

C3(tn)=ψR(t0)|p|φR(t0),C_{3}(t_{n})=\langle\psi_{R}(t_{0})|p|\varphi_{R}(t_{0})\rangle\;, (6)

with |ψR(t0)=U(t0,tn)pU(t0,tn)|ψ(t0)|\psi_{R}(t_{0})\rangle={U}^{\dagger}(t_{0},t_{n})pU(t_{0},t_{n})|\psi(t_{0})\rangle and |φR(t0)=U(t0,tn)pU(t0,tn)p|ψ(t0)|\varphi_{R}(t_{0})\rangle={U}^{\dagger}(t_{0},t_{n})pU(t_{0},t_{n})p|\psi(t_{0})\rangle Ueda2018 . The symbol Re[]\text{Re}[\cdots] denotes the real part of a complex variable.

To obtain the state |ψR(t0)|\psi_{R}(t_{0})\rangle, three steps must be carried out: i) the forward evolution from t0t_{0} to tnt_{n}, i.e. |ψ(tn)=U(t0,tn)|ψ(t0)|\psi(t_{n})\rangle=U(t_{0},t_{n})|\psi(t_{0}), ii) the action of the operator pp on the state |ψ(tn)|\psi(t_{n})\rangle, i.e. |ψ~(tn)=p|ψ(tn)|\tilde{\psi}(t_{n})\rangle=p|\psi(t_{n})\rangle, and iii) the backward evolution from tnt_{n} to t0t_{0}, i.e. |ψR(t0)=U(t0,tn)|ψ~(tn)|\psi_{R}(t_{0})\rangle={U}^{\dagger}(t_{0},t_{n})|\tilde{\psi}(t_{n})\rangle. The expectation value of the square of the momentum can then be calculated using |ψR(t0)|\psi_{R}(t_{0})\rangle to obtain C1(tn)C_{1}(t_{n}) [see Eq. (4)]. To numerically simulate C2(t)C_{2}(t), the operator pp should first be applied to the initial state |ψ(t0)|\psi(t_{0})\rangle, yielding the new state |φ(t0)=p|ψ(t0)|\varphi(t_{0})\rangle=p|\psi(t_{0})\rangle. Then, the forward evolution is conducted, i.e., |φ(tn)=U(t0,tn)|φ(t0)|\varphi(t_{n})\rangle=U(t_{0},t_{n})|\varphi(t_{0})\rangle. Subsequently, the action of pp is performed on |φ(tn)|\varphi(t_{n})\rangle, obtaining |φ~(tn)=p|φ(tn)|\tilde{\varphi}(t_{n})\rangle=p|\varphi(t_{n})\rangle, afterwards, the time-reversal is applied to |φ~(tn)|\tilde{\varphi}(t_{n})\rangle, resulting in |φR(t0)=U(t0,tn)|φ~(tn)|\varphi_{R}(t_{0})\rangle={U}^{\dagger}(t_{0},t_{n})|\tilde{\varphi}(t_{n})\rangle. Using Eq. (5), C2(tn)C_{2}(t_{n}) can then be calculated by evaluating the norm of |φR(t0)|\varphi_{R}(t_{0})\rangle. Lastly, the term C3(tn)C_{3}(t_{n}) [seen in Eq. (6)] can be determined using the two states |ψR(t0)|\psi_{R}(t_{0})\rangle and |φR(t0)|\varphi_{R}(t_{0})\rangle, which is usually complex since they are not identical.

In the 𝒫𝒯\cal{PT}-symmetry breaking phase, the norm of the quantum state 𝒩ψ(tn)=ψ(tn)|ψ(tn)\mathcal{N}_{\psi}(t_{n})=\langle\psi(t_{n})|\psi(t_{n})\rangle increases exponentially with time regardless of the forward or backward evolution. To address this issue and eliminate its contribution to the OTOCs, we normalize the time-evolved state. For the forward evolution t0tnt_{0}\rightarrow t_{n} of |ψ(t0)|\psi(t_{0})\rangle, we set the norm of the quantum state to be the same as that of the initial state, i.e., 𝒩ψ(tj)=ψ(t0)|ψ(t0)\mathcal{N}_{\psi}(t_{j})=\langle\psi(t_{0})|\psi(t_{0})\rangle with 0jn0\leq j\leq n. The backward evolution starts from the state |ψ~(tn)|\tilde{\psi}(t_{n})\rangle, whose norm 𝒩ψ~(tn)=ψ(tn)|p2|ψ(tn)\mathcal{N}_{\tilde{\psi}}(t_{n})=\langle\psi(t_{n})|p^{2}|\psi(t_{n})\rangle is the mean energy of the state |ψ(tn)|\psi(t_{n})\rangle. Thus, it is reasonable to take the norm of the quantum state during the backward evolution tnt0t_{n}\rightarrow t_{0} to be 𝒩ψ~(tn)\mathcal{N}_{\tilde{\psi}}(t_{n}), i.e., 𝒩ψR(tj)=𝒩ψ~(tn)\mathcal{N}_{\psi_{R}}(t_{j})=\mathcal{N}_{\tilde{\psi}}(t_{n}). In short, the norm of the time-evolved state for both the forward and backward evolution is equal to that of the state it starts from. If the same normalization procedure is applied to the evolution of |φ(tn)|\varphi(t_{n})\rangle, then we will have 𝒩φ(tj)=φ(t0)|φ(t0)\mathcal{N}_{\varphi}(t_{j})=\langle\varphi(t_{0})|\varphi(t_{0})\rangle and 𝒩φR(tj)=φ~(tn)|φ~(tn)\mathcal{N}_{\varphi_{R}}(t_{j})=\langle\tilde{\varphi}(t_{n})|\tilde{\varphi}(t_{n})\rangle (0jn0\leq j\leq n) for the forward and time reversal evolutions, respectively.

Refer to caption
Figure 2: (a) Time dependence of CC for K=6K=6 with λ=105\lambda=10^{-5} (squares), 0.0220.022 (circles), 0.20.2 (diamonds), 0.50.5 (triangles), and 0.90.9 (pentagrams). Solid lines in red denote theoretical prediction in Eqs. (12) and (13), i.e., C(t)tηC(t)\propto t^{\eta}. (b) The η\eta versus λ\lambda for eff=0.3\hbar_{\text{eff}}=0.3 (squares) and 0.5 (circles). Arrow marks the phase transition point λc0.001\lambda_{c}\approx 0.001 for eff=0.3\hbar_{\text{eff}}=0.3.

In order to understand the effects of 𝒫𝒯\cal{PT}-symmetry breaking on the dynamics of C(tn)C(t_{n}), we numerically investigated the time evolution of C(tn)C(t_{n}) for different λ\lambda. Figure 2(a) shows that, for values of λ\lambda smaller than the phase transition point (e.g., λ=105λc\lambda=10^{-5}\ll\lambda_{c}), the C(tn)C(t_{n}) increases gradually up to saturation. Interestingly, for λ\lambda slightly larger than λc\lambda_{c}, the C(tn)C(t_{n}) increases in a power-law of time, i.e., C(tn)tηC(t_{n})\propto t^{\eta} with η>2\eta>2 [see λ=0.022\lambda=0.022 with η=3.4\eta=3.4 in Fig. 2(a)]. We dub this phenomenon as a super-quadratic growth (SQG) of C(tn)C(t_{n}). When the value of λ\lambda is much larger than the phase transition point, i.e., λλc\lambda\gg\lambda_{c} [e.g., λ=0.5\lambda=0.5 and 0.9 in Fig.2(a)], the quadratic growth (QG) of OTOCs C(tn)t2C(t_{n})\propto t^{2} emerges. We further investigate the exponent η\eta for different λ\lambda. Our results show that η\eta is zero for λ<λc\lambda<\lambda_{c}, increases abruptly to a maximum value greater than two for λ\lambda slightly larger than λc\lambda_{c}, and finally saturates to two for sufficiently large λ\lambda [e.g., see eff=0.3\hbar_{\text{eff}}=0.3 in Fig. 2(b)]. It is evident that the scaling-law of OTOCs reveals the emergence of the spontaneous 𝒫𝒯\mathcal{PT}-symmetry breaking and unveils the correlation between information scrambling and the 𝒫𝒯\mathcal{PT}-symmetry phase transition.

IV Theoretical analysis of the dynamics of OTOCs

IV.1 Mechanism of the saturation of C(t)C(t) for λ<λc\lambda<\lambda_{c}

We numerically investigate the time evolution of the three parts of the OTOCs, i.e., C1C_{1}, C2C_{2}, and C3C_{3} for λ<λc\lambda<\lambda_{c}. Figure 3(a) shows that the time dependence of C1C_{1} and CC almost overlap, displaying rapid growth up to saturation. Since the real part of C3C_{3}, i.e., Re(C3)\textrm{Re}(C_{3}) fluctuates between positive and negative values, we plot the absolute value |Re(C3)||\textrm{Re}(C_{3})| in Fig. 3(b). Both C2C_{2} and |Re[C3]||\textrm{Re}[C_{3}]| saturate after a very short time evolution. Importantly, C1C_{1} is at least 4 orders of magnitude larger than both C2C_{2} and |Re[C3]||\textrm{Re}[C_{3}]|, leading to a perfect consistency between C1C_{1} and CC. Consequently, based on Eq. (3), we can safely use the approximation

C(tn)ψR(t0)|p2|ψR(t0)=p2(t0)R𝒩ψR(t0),\displaystyle C(t_{n})\approx\langle\psi_{R}(t_{0})|{p}^{2}|\psi_{R}(t_{0})\rangle=\langle p^{2}(t_{0})\rangle_{R}\mathcal{N}_{\psi_{R}}(t_{0})\;, (7)

where p2(t0)R=ψR(t0)|p2|ψR(t0)/𝒩ψR(t0)\langle p^{2}(t_{0})\rangle_{R}=\langle\psi_{R}(t_{0})|{p}^{2}|\psi_{R}(t_{0})\rangle/\mathcal{N}_{\psi_{R}}(t_{0}) denotes the exceptional value of energy of the state |ψR(t0)|\psi_{R}(t_{0})\rangle divided by its norm 𝒩ψR(t0)=ψR(t0)|ψR(t0)\mathcal{N}_{\psi_{R}}(t_{0})=\langle\psi_{R}(t_{0})|\psi_{R}(t_{0})\rangle.

Refer to caption
Figure 3: (a) Time evolution of CC (circles) and C1C_{1} (squares). Note that CC almost fully overlaps with C1C_{1}. (b) Dependence of C2C_{2} (squares) and |Re(C3)||\textrm{Re}(C_{3})| (circles) on time. The parameters are K=6K=6, λ=105\lambda=10^{-5}, and eff=0.3\hbar_{\text{eff}}=0.3. (c) The C¯\bar{C} versus KK with λ=105\lambda=10^{-5} for eff=1\hbar_{\text{eff}}=1 (squares), 5 (circles), 10 (triangles), and 15 (diamonds). Solid lines indicates our theoretical prediction in Eq. (8).

The normalization procedure for time reversal yields the equivalence 𝒩ψR(t0)=𝒩ψ~(tn)=ψ(tn)|p2|ψ(tn)\mathcal{N}_{\psi_{R}}(t_{0})=\mathcal{N}_{\tilde{\psi}}(t_{n})=\langle\psi(t_{n})|p^{2}|\psi(t_{n})\rangle, which shows that the value of 𝒩ψR(t0)\mathcal{N}_{\psi_{R}}(t_{0}) is just the mean energy of the state |ψ(tn)|\psi(t_{n})\rangle at the time t=tnt=t_{n}. For λ<λc\lambda<\lambda_{c}, the quasienergies are all real, thus the dynamics of the PTKR is the same as that of the Hermitian QKR. A noteworthy characteristic of the QKR’s energy diffusion is the phenomenon of DL, i.e. the mean energy p2\langle p^{2}\rangle gradually approaches to saturation level with increasing time due to quantum coherence. It is reasonable to believe that the mechanism of DL suppresses the growth of both 𝒩ψR(t0)\mathcal{N}_{\psi_{R}}(t_{0}) and p2(t0)R\langle p^{2}(t_{0})\rangle_{R}, and therefore leads to the saturation of C(tn)C(t_{n}).

To confirm this conjecture, we consider a specific time, i.e., t=tnt=t_{n}, and numerically trace the evolution of p2\langle p^{2}\rangle for both the forward (t<tnt<t_{n}) and backward (t>tnt>t_{n}) evolution. Figure 4(a) shows that for tn=2500t_{n}=2500, p2\langle p^{2}\rangle increases rapidly to saturation during forward time evolution from t0t_{0} to t2500t_{2500}, then jumps to a specific value at the start of the time reversal (i.e., at t=t2500t=t_{2500}) before finally saturating for the backward evolution from t2500t_{2500} to t0t_{0}. This clearly demonstrates the emergence of the DL, which is also reflected by the probability density distribution in momentum space. We compare the momentum distributions at the end of the forward evolution (i.e., t=t2500t=t_{2500}) and the end of time reversal (i.e., t=t0t=t_{0}) in Fig. 4(b). One can see that the two quantum states almost overlap with each other, both of which are exponentially localized in momentum space, i.e., |ψ(tn)|2exp(|p|/L)|\psi(t_{n})|^{2}\sim\exp(-|p|/L) [see Fig. 4(b)]. A rough estimation yields 𝒩ψR(t0)=ψ(tn)|p2|ψ(tn)L2\mathcal{N}_{\psi_{R}}(t_{0})=\langle\psi(t_{n})|p^{2}|\psi(t_{n})\rangle\sim L^{2} and p2(t0)Rp2exp(|p|/L)𝑑pL2\langle p^{2}(t_{0})\rangle_{R}\sim\int_{-\infty}^{\infty}p^{2}\exp(-|p|/L)dp\sim L^{2}. Plugging the two relations into Eq. (7), we can immediately get the estimation of the OTOCs, i.e., C(tn)L4C(t_{n})\sim L^{4}. It is known that the localization length is in a quadratic function of KK, i.e., LK2L\propto K^{2} Izrailev90 , which results in the relation

CK8.C\propto K^{8}\;. (8)

This clearly demonstrates that the CC is time-independent after the long term evolution, verifying our numerical results in Fig. 2(a) and Fig. 3(a).

To provide evidence of our analytical prediction, we investigate the time-averaged value of OTOCs, i.e., C¯=j=1NC(tj)/N\bar{C}=\sum_{j=1}^{N}C(t_{j})/N, numerically for different KK. In the numerical simulations, we ensure that NN is large enough for the long-term saturation of C(t)C(t) to be well quantified by C¯\bar{C}. Our numerical results show that for a specific \hbar, C¯\bar{C} increases in a power-law of KK [see Fig. 3(c)], which is well described by our theoretical prediction in Eq. (8). This is a strong indication of the validity of our analytical analysis. Our findings of the dependence of the OTOCs on the kick strength provide an opportunity to control the operator growth with an external driven potential.

Refer to caption
Figure 4: (a) Time trace of p2\langle p^{2}\rangle (squares) and 𝒩\mathcal{N} (circles) during the forward evolution t0t2500t_{0}\rightarrow t_{2500}, the action of pp at the time t=t2500t=t_{2500}, and the time reversal t2500t5000t_{2500}\rightarrow t_{5000}. Green dashed line marks tn=2500t_{n}=2500. (b) Momentum distributions for the state |ψ(t)|\psi(t)\rangle (squares) at the time tn=2500t_{n}=2500 and the state |ψR(t0)|\psi_{R}(t_{0})\rangle (circles) at the end of time reversal. Solid line indicates the exponentially-localized shape |ψ(p)|2e|p|/L|\psi(p)|^{2}\propto e^{-|p|/L} with L46L\approx 46. The parameters are K=6K=6, λ=105\lambda=10^{-5}, and eff=0.3\hbar_{\text{eff}}=0.3.

The discontinuous jump in the mean square momentum, p2\langle p^{2}\rangle, at t=t2500t=t_{2500}, the beginning of time reversal, is due to the action of the operator pp on the quantum state |ψ(t2500)|\psi(t_{2500})\rangle. This action generates the quantum state |ψ~(t2500)=p|ψ(t2500)|\tilde{\psi}(t_{2500})\rangle=p|\psi(t_{2500})\rangle, for which the mean value is given by ψ~(t2500)|p2|ψ~(t2500)=ψ(t2500)|p4|ψ(t2500)\langle\tilde{\psi}(t_{2500})|p^{2}|\tilde{\psi}(t_{2500})\rangle=\langle\psi(t_{2500})|p^{4}|\psi(t_{2500})\rangle. The exponentially-localized shape of the quantum state, |ψ(t2500)|2exp(|p|/L)|\psi(t_{2500})|^{2}\sim\exp(-|p|/L) [see Fig. 4(b)], allows us to obtain the expectation values ψ(t2500)|p2|ψ(t2500)L2eff2\langle\psi(t_{2500})|p^{2}|\psi(t_{2500})\rangle\sim L^{2}\hbar_{\text{eff}}^{2} and ψ~(t2500)|p2|ψ~(t2500)=ψ(t2500)|p4|ψ(t2500)L4eff4\langle\tilde{\psi}(t_{2500})|p^{2}|\tilde{\psi}(t_{2500})\rangle=\langle\psi(t_{2500})|p^{4}|\psi(t_{2500})\rangle\sim L^{4}\hbar_{\text{eff}}^{4}, which quantitatively explains the discontinuous increase in the mean energy from L2eff2L^{2}\hbar_{\text{eff}}^{2} to L4eff4L^{4}\hbar_{\text{eff}}^{4}.

IV.2 Mechanism of the SQG of C(t)C(t) for λλc\lambda\gtrsim\lambda_{c}

Figure 5(a) shows the time evolution of C1C_{1}, C2C_{2}, |Re(C3)||\textrm{Re}(C_{3})|, and CC for λ\lambda just larger than the 𝒫𝒯\cal{PT}-symmetry phase transition point λλc\lambda\gtrsim\lambda_{c}. It is clear that the time dependence of C1C_{1} corresponds perfectly to that of CC, both of which increase following the SQG Ct3.4C\propto t^{3.4}. The time evolution of C2C_{2} displays the QG, i.e., C2(t)t2C_{2}(t)\propto t^{2}, while |Re(C3)||\textrm{Re}(C_{3})| follows the SQG |Re(C3)|t2.9|\textrm{Re}(C_{3})|\sim t^{2.9}. In addition, one can see that C1C_{1} is larger than both C2C_{2} and |Re(C3)||\textrm{Re}(C_{3})| by approximately four orders of magnitude. Therefore, it is sufficient to analyze the time evolution of the term C1C_{1} to uncover the mechanism of the SQG of CC.

Refer to caption
Figure 5: Time evolution of C1C_{1} (squares), C2C_{2} (diamonds), |Re(C3)||\textrm{Re}(C_{3})| (triangles), and CC (circles) for λ=0.022\lambda=0.022 (a) and 0.9 (b). In (a): Red lines indicates the power-law fitting. In (b): Red lines indicate the quadratic function C144t2C_{1}\approx 44t^{2}, C217t2C_{2}\approx 17t^{2}, and |Re(C3)|2t2|\textrm{Re}(C_{3})|\approx 2t^{2}. Inset: The Δ12\Delta_{12} (squares) and Δ13\Delta_{13} (circles) versus time. Solid (dashed-dotted) line denotes Δ122.5\Delta_{12}\approx 2.5 (Δ1322\Delta_{13}\approx 22). Other parameters are same as Fig. 4.

Since the value of C(t)C(t) at a specific time t=tnt=t_{n} is dependent on both the mean energy p2(t0)R\langle p^{2}(t_{0})\rangle_{R} and the norm 𝒩ψR(t0)\mathcal{N}_{\psi_{R}}(t_{0}) of the state |ψR(t0)|\psi_{R}(t_{0})\rangle at the end of time reversal [see Eq. (7)], we numerically calculate the forward and backward time evolution of p2\langle p^{2}\rangle, p\langle p\rangle, and 𝒩\mathcal{N} with a fixed tnt_{n} (e.g., tn=2500t_{n}=2500 in Fig. 6). Figure 6(a) demonstrates that the mean energy diffuses ballistically with time p2γ2t2\langle p^{2}\rangle\approx\gamma^{2}t^{2} during the forward evolution t0t2500t_{0}\rightarrow t_{2500}, and it displays the intrinsic time reversal during t2500t0t_{2500}\rightarrow t_{0}. Meanwhile, the mean momentum p\langle p\rangle linearly increases for t0<t<t2500t_{0}<t<t_{2500}, and linearly decays for t2500t0t_{2500}\rightarrow t_{0} [see Fig. 6(b)]. Moreover, Fig. 6(a) reveals that the norm remains unity, i.e., 𝒩(t)=1\mathcal{N}(t)=1, during the forward evolution and equals the mean energy at the time t=t2500t=t_{2500}, i.e., 𝒩ψR(tj)=p2(t2500)\mathcal{N}_{\psi_{R}}(t_{j})=\langle p^{2}(t_{2500})\rangle during the time reversal. Taking the ballistic diffusion of energy into account, the following equivalence can be derived

𝒩ψR(t0)=γ2tn2.\mathcal{N}_{\psi_{R}}(t_{0})=\gamma^{2}t_{n}^{2}\;. (9)

In order to measure the degree of time reversal for a fixed tnt_{n}, we define the ratio of mean energy between forward (t<tnt<t_{n}) and backward (t>tnt>t_{n}) time evolution as

(tj)=p2(2tntj)Rp2(tj),\mathcal{R}(t_{j})=\frac{\langle p^{2}(2t_{n}-t_{j})\rangle_{R}}{\langle p^{2}(t_{j})\rangle}\;, (10)

where p2(2tntj)R\langle p^{2}(2t_{n}-t_{j})\rangle_{R} and p2(tj)\langle p^{2}(t_{j})\rangle (0jn0\leq j\leq n) denote the mean square of momentum for the forward evolution and time reversal, respectively. The inset in Fig. 6(b) shows that \mathcal{R} is very large (i.e., 104\mathcal{R}\gtrsim 10^{4} for t=t0t=t_{0}) and approaches almost one with time evolution. This reveals that the mean energy at the end of time reversal is much greater than that at the initial time, i.e., p2(t0)Rp2(t0)\langle p^{2}(t_{0})\rangle_{R}\gg\langle p^{2}(t_{0})\rangle. We further investigate the time evolution of p2(t0)R\langle p^{2}(t_{0})\rangle_{R}, and find [see Fig. 6(c)] that the p2(t0)R\langle p^{2}(t_{0})\rangle_{R} increases in the power-law of time

p2(t0)Rt1.4.\langle p^{2}(t_{0})\rangle_{R}\propto t^{1.4}\;. (11)

Substituting Eqs. (9) and (11) into Eq. (7) yields the SQG of OTOCs

C(t)tηwithη3.4.C(t)\propto t^{\eta}\quad\text{with}\quad\eta\approx 3.4\;. (12)
Refer to caption
Figure 6: Top two panels: Time trace of p2\langle p^{2}\rangle (squares) in (a), 𝒩\mathcal{N} (circles) in (a), and p\langle p\rangle in (b) during the forward evolution t0t2500t_{0}\rightarrow t_{2500}, the action of pp at the time t=t2500t=t_{2500}, and the time reversal t2500t0t_{2500}\rightarrow t_{0}. Green dashed lines makes t=t2500t=t_{2500}. In (a): Red line indicates the quadratic function p2=γ2t2\langle p^{2}\rangle=\gamma^{2}t^{2} with γ3.2\gamma\approx 3.2. In (b): Red line indicates the linear growth p=γt\langle p\rangle=\gamma t. Inset: \cal{R} versus time. Red line marks =1\mathcal{R}=1. In (c): Time dependence of p2(t0)R\langle p^{2}(t_{0})\rangle_{R}. Red line indicates the power-law fitting p2(t0)Rt1.4\langle p^{2}(t_{0})\rangle_{R}\propto t^{1.4}. The parameter is λ=0.022\lambda=0.022. Other parameters are same as Fig. 4.

Figure 7 shows the probability density distribution of the state at forward |ψ(tj)|\psi(t_{j})\rangle and backward |ψR(tj)|\psi_{R}(t_{j})\rangle evolution in both the real and momentum space. The initial state is a Gaussian wavepacket ψ(θ,t0)=(σ/π)1/4exp(σθ2/2)\psi{(\theta,t_{0})}=(\sigma/\pi)^{1/4}\exp(-\sigma\theta^{2}/2) centered at θ=0\theta=0 and p=0p=0 [see Figs. 7(a) and (b)]. Interestingly, one can observe that the quantum state is mainly distributed in the region 0<θ<π0<\theta<\pi for both the forward and backward evolution. This is due to the fact that the action of the Floquet operator of the kicking term UK(θ)=exp[Kλsin(θ)/eff]exp[iKcos(θ)/eff]U_{K}(\theta)=\exp[K\lambda\sin(\theta)/\hbar_{\text{eff}}]\exp[-iK\cos(\theta)/\hbar_{\text{eff}}] on a quantum state, i.e., UK(θ)ψ(θ)U_{K}(\theta)\psi(\theta) helps to amplify the state within the region 0<θ<π0<\theta<\pi as Kλsin(θ)>0K\lambda\sin(\theta)>0. Assuming that the real part of the kicking potential provides the driven force F=Ksin(θ)F=K\sin(\theta), the PTKR experiences a positive magnitude force F>0F>0 during the forward evolution, thus the momentum grows with time, as shown in Fig. 6(b). For the time reversal, the sign of kick strength KK flips, i.e., KKK\rightarrow-K, so the mean momentum decreases with time evolution. Our conjecture is supported by the numerical results of momentum distributions. Figures 7(b), (d) and (f) show that the wavepacket, like a soliton, moves to the positive direction in momentum space for t0t2500t_{0}\rightarrow t_{2500}, resulting in p=γt\langle p\rangle=\gamma t [in Fig. 6(b)], and moves back to the opposite direction for t2500t0t_{2500}\rightarrow t_{0}. In addition, the width of the wavepacket in momentum space is so narrow that one can safely use the approximation p2(p)2\langle p^{2}\rangle\sim(\langle p\rangle)^{2}, which is verified by our numerical results in Figs. 6(a).

Refer to caption
Figure 7: Probability density distributions in real (left panels) and momentum (right) space for λ=0.022\lambda=0.022. (a)-(d) Black and blue lines separately correspond to the states at forward |ψ(t)|\psi(t)\rangle and backward |ψR(t)|\psi_{R}(t)\rangle evolution with t=t0t=t_{0} (top panels) and t=t1250t=t_{1250} (middle panels). Inset in (b) shows a magnified view of the momentum distribution for the initial Gaussian wavepacket around p=0p=0. Bottom panels: Probability density distributions in real (e) and momentum (f) space at the time t=t2500t=t_{2500}. Black and blue lines indicate the state |ψ(t2500)|\psi(t_{2500})\rangle and |ψ~(t2500)=p|ψ(t2500)|\tilde{\psi}(t_{2500})\rangle=p|\psi(t_{2500})\rangle, respectively. Other parameters are same as Fig. 4.

IV.3 Mechanism of the QG of C(t)C(t) for λλc\lambda\gg\lambda_{c}

We numerically investigate the time evolution of C1C_{1}, C2C_{2}, C3C_{3}, and CC for λλc\lambda\gg\lambda_{c}. As shown in Fig. 5(b), all of them increase in the way of QG (i.e., t2\propto t^{2}). We use the ratios Δ12=C1/C2\Delta_{12}=C_{1}/C_{2} and Δ13=C1/|Re[C3]|\Delta_{13}=C_{1}/|\textrm{Re}[C_{3}]| to quantify the differences among C1C_{1}, C2C_{2}, and |Re[C3]||\textrm{Re}[C_{3}]|. Our investigation show that both of them are larger than one, specifically Δ122.5\Delta_{12}\approx 2.5 and Δ1322\Delta_{13}\approx 22 [see the inset in Fig. 5]. This suggests that C1C_{1} contributes mainly to the CC, which is verified by the good agreement between C1C_{1} and CC [see Fig. 5(b)]. To reveal the mechanism of the QG of CC, we proceed to analyze the time evolution of C1C_{1} by thoroughly investigating both the forward and backward evolution of the mean values p2\langle p^{2}\rangle, p\langle p\rangle, and the norm 𝒩\cal{N} for a given tnt_{n}.

Figure 8(a) shows that the mean energy exhibits ballistic diffusion p2γ2t2\langle p^{2}\rangle\approx\gamma^{2}t^{2}, during the forward evolution from t0t_{0} to tnt_{n}, and decays as the inverse of a quadratic function, with p2t2\langle p^{2}\rangle\propto t^{-2}, during the backward evolution from tnt_{n} to t0t_{0}. This decay is symmetric with respect to the p2\langle p^{2}\rangle of t<tnt<t_{n}. The dynamics of the mean momentum also exhibits perfect time reversal, namely it linearly increases as p=γt\langle p\rangle=\gamma t during t0t2500t_{0}\rightarrow t_{2500} and decreases linearly during t2500t0t_{2500}\rightarrow t_{0}. The ratio \mathcal{R} remains close to one throughout the time evolution, except at the end, i.e., (t=2500)2.5\mathcal{R}(t=2500)\approx 2.5 [see the inset in Fig. 8(b)], providing a clear evidence of time reversal. For the forward evolution from t0t_{0} to t2500t_{2500}, the norm 𝒩(tj)\mathcal{N}(t_{j}) is equal to unity, while for the interval t2500t0t_{2500}\rightarrow t_{0}, it is equal to the value of p2(t2500)\langle p^{2}(t_{2500})\rangle, i.e., 𝒩ψR(tj)=p2(t2500)\mathcal{N}_{\psi_{R}}(t_{j})=\langle p^{2}(t_{2500})\rangle [see Fig. 8(a)]. By utilizing the ballistic diffusion of mean energy, we establish the relationship 𝒩ψR(t0)𝒩ψR(tn)γ2tn2\mathcal{N}_{\psi_{R}}(t_{0})\approx\mathcal{N}_{\psi_{R}}(t_{n})\approx\gamma^{2}t_{n}^{2}, where tnt_{n} is an arbitrary time. We further evaluate the behavior of p2(t0)R\langle p^{2}(t_{0})\rangle_{R} for different tnt_{n}. As shown in Fig. 8(c), the p2(t0)R\langle p^{2}(t_{0})\rangle_{R} remains almost constant at a value of one, indicating that it is independent of time. Plugging in the values of 𝒩ψR(t0)\mathcal{N}_{\psi_{R}}(t_{0}) and p2(t0)R\langle p^{2}(t_{0})\rangle_{R} into Eq. (7), we obtain the QG of OTOCs

C(t)γ2tηwithη=2.C(t)\approx\gamma^{2}t^{\eta}\quad\text{with}\quad\eta=2. (13)
Refer to caption
Figure 8: Same as in Fig. 6 but for λ=0.9\lambda=0.9. In (a): Red line indicates the quadratic function p2=γ2t2\langle p^{2}\rangle=\gamma^{2}t^{2} with γ6.3\gamma\approx 6.3. In (b): Red line denotes the linear growth p=γt\langle p\rangle=\gamma t. In (c): Red line indicates p(t0)R1\langle p(t_{0})\rangle_{R}\approx 1.

The time reversal of a wavepacket’s dynamics is clearly seen in the evolution of its momentum distributions. For the forward time evolution, the quantum state is localized at the point θc=π/2\theta_{c}=\pi/2 [seen in Figs. 9(a), (c) and (e)], which is the result of the localization-effect of the imaginary part of the Floquet operator UK(θ)U_{K}(\theta). With the wavepacket mimicking a classical particle, it experiences a kicking force of magnitude F=Ksin(θc)=KF=K\sin(\theta_{c})=K, resulting in a constant acceleration of momentum Δp=K\Delta p=K, which is reflected in the linear growth of momentum. This phenomenon of the directed current is also seen in the propagation of momentum distributions in Figs. 9(b), (d) and (f), where a soliton can be observed moving unidirectionally towards the positive direction in momentum space.

During the backward evolution, the wavepacket of the real space remains centered at θc=π/2\theta_{c}=\pi/2, with a width much smaller than the corresponding state at the time of forward evolution [see Figs. 9(a), (c) and (e)]. As the particle is exposed to the kicking force with F=KF=-K during time reversal, its momentum decreases linearly in time, which is also reflected in the propagation of the wavepackets in momentum space [Figs. 9(b), (d) and (f)]. It is evident that the |ψ(tn)|2|\psi(t_{n})|^{2} is in perfect overlap with the |ψR(tn)|2|\psi_{R}(t_{n})|^{2}, apart from the initial state [see Figs. 9(b)]. The width of |ψR(t0)|2|\psi_{R}(t_{0})|^{2} is considerably larger than that of |ψ(t0)|2|\psi(t_{0})|^{2}, leading to the ratio of energy being larger than one, i.e., =p2(t5000)R/p2(t0)2.5\mathcal{R}=\langle p^{2}(t_{5000})\rangle_{R}/\langle p^{2}(t_{0})\rangle\approx 2.5 [see the inset in Fig. 8(b)].

Refer to caption
Figure 9: Same as in Fig. 7 but for λ=0.9\lambda=0.9.

V Conclusion and discussion

In this work, we investigate the dynamics of OTOCs in a PTKR model and achieve its scaling laws in different phases of 𝒫𝒯\cal{PT} symmetry. We use the time series of the norm to train a LSTM, which enables us to extract a clear phase diagram of 𝒫𝒯\cal{PT}-symmetry breaking, with a phase boundary at λc\lambda_{c}. For λ<λc\lambda<\lambda_{c}, we find that the DL of energy diffusion suppresses the growth of OTOCs, and prove analytically the dependence of OTOCs on the kicking strength, i.e., CK8C\propto K^{8}. At the vicinity of the phase transition points, i.e., λλc\lambda\gtrsim\lambda_{c}, we observe a SQG of OTOCs, i.e., C(t)tηC(t)\propto t^{\eta} with an exponent η>2\eta>2. Interestingly, a QG of OTOCs, i.e., C(t)t2C(t)\propto t^{2} emerges for λλc\lambda\gg\lambda_{c}. We elucidate the mechanisms of both the SQG and QG by analyzing the time-reversed wavepacket’s dynamics. Our results demonstrate that the spontaneous 𝒫𝒯\cal{PT}-symmetry breaking profoundly affects the dynamics of OTOCs, providing an unprecedented opportunity for diagnosing the spontaneous 𝒫𝒯\cal{PT}-symmetry breaking with OTOCs.

In recent years, the OTOCs have been widely used to investigate the operator growth in quantum mapping systems Moudgalya19 , the information scrambling in spin chains TianciZhou20 , and the quantum thermalization in many-body chaotic systems KenXuanWei19 . Theoretical studies have demonstrated that the QKR model is mathematically equivalent to the kicked Heisenberg spin XXZ chain Boness10 , indicating a connection between the magnon dynamics and quantum diffusion of chaotic systems. Our findings therefore bridge the gap between the information scrambling in condensed matter physics and the operator growth in quantum chaotic systems. This also paves the way for the experimental observation of OTOCs dynamics in chaotic systems using spin chain platforms.

ACKNOWLEDGMENTS

Wen-Lei Zhao is supported by the National Natural Science Foundation of China (Grant Nos. 12065009), the Natural Science Foundation of Jiangxi province (Grant Nos. 20224ACB201006 and 20224BAB201023) and the Science and Technology Planning Project of Ganzhou City (Grant No. 202101095077). Jie Liu is supported by the NSAF (Contract No. U1930403).

References

  • (1) H. Markum, R. Pullirsch, and T. Wettig, Non-Hermitian Random Matrix Theory and Lattice QCD with Chemical Potential, Phys. Rev. Lett. 83, 484 (1999).
  • (2) N. Hatano and David R. Nelson, Localization Transitions in Non-Hermitian Quantum Mechanics, Phys. Rev. Lett. 77, 570 (1996).
  • (3) M. Berry, Physics of Nonhermitian Degeneracies, Czech. J. Phys. 54, 1039 (2004).
  • (4) E. M. Graefe, H. J. Korsch, and A. E. Niederle, Mean-field dynamics of a non-Hermitian Bose-Hubbard dimer, Phys. Rev. Lett. 101, 150408 (2008).
  • (5) I. Rotter, A non-Hermitian Hamilton operator and the physics of open quantum systems, J. Phys. A 42, 153001 (2009).
  • (6) G. Barontini, R. Labouvie, F. Stubenrauch, A. Vogler, V. Guarrera, and H. Ott, Controlling the Dynamics of an Open Many-Body Quantum System with Localized Dissipation, Phys. Rev. Lett. 110, 035302 (2013).
  • (7) K. Jones-Smith and H. Mathur, Relativistic Non-Hermitian Quantum Mechanics, Phys. Rev. D89, 125014 (2014).
  • (8) C. M. Bender and S. Boettcher, Real Spectra in Non-Hermitian Hamiltonians Having PT Symmetry, Phys. Rev. Lett. 80, 5243 (1998).
  • (9) C. M. Bender, D. C. Brody, and H. F. Jones, Complex Extension of Quantum Mechanics, Phys. Rev. Lett. 89, 270401 (2002).
  • (10) A. Mostafazadeh, Pseudo-Hermiticity versus PT-symmetry. II. A complete characterization of non-Hermitian Hamiltonians with a real spectrum, J. Math. Phys. (N.Y.), 43, 2814 (2002).
  • (11) C. M. Bender, Making sense of non-Hermitian Hamiltonians, Rep. Prog. Phys. 70, 947 (2007).
  • (12) A. Mostafazadeh, Pseudo-hermitian representation of quantum mechanics, Int. J. Geom. Meth. Mod. Phys. 07, 1191 (2010).
  • (13) K. Jones-Smith and H. Mathur, Non-Hermitian quantum Hamiltonians with PT symmetry, Phys. Rev. A82, 042101 (2010).
  • (14) R. El-Ganainy, K. G. Makris, M. Khajavikhan, Z. H. Musslimani, S. Rotter, and D. N. Christodoulides, Non-Hermitian physics and PT symmetry, Nat. Phys. 14, 11 (2018).
  • (15) S. Klaiman, U. Günther, and N. Moiseyev, Visualization of Branch Points in PT-Symmetric Waveguides, Phys. Rev. Lett. 101, 080402 (2008).
  • (16) Z. H. Musslimani, K. G. Makris, R. El-Ganainy, and D. N. Christodoulides, Optical Solitons in PT Periodic Potentials, Phys. Rev. Lett. 100, 030402 (2008).
  • (17) Y. Chu, Y. Liu, H. Liu, and J. Cai, Quantum Sensing with a Single-Qubit Pseudo-Hermitian System, Phys. Rev. Lett. 124, 020501 (2020).
  • (18) E. J. Bergholtz, J. C. Budich, and F. K. Kunst, Exceptional topology of non-Hermitian systems, Rev. Mod. Phys. 93, 015005 (2021).
  • (19) S. Yao and Z. Wang, Edge States and Topological Invariants of Non-Hermitian Systems, Phys. Rev. Lett. 121, 086803 (2018).
  • (20) X. M. Zhao, C. X. Guo, M. L. Yang, H.Wang,W. M. Liu, and S. P. Kou, Anomalous non-Abelian statistics for non-Hermitian generalization of Majorana zero modes, Phys. Rev. B104, 214502 (2021).
  • (21) Z. F. Yu, J. K. Xue, L. Zhuang, J. Zhao, and W. M. Liu, Non-Hermitian spectrum and multistability in exciton-polariton condensates, Phys. Rev. B104, 235408 (2021).
  • (22) X. M. Zhao, C. X. Guo, S. P. Kou, L. Zhuang, and W. M. Liu, Defective Majorana zero modes in a non-Hermitian Kitaev chain, Phys. Rev. B104, 205131 (2021).
  • (23) F. Yu, X. L. Zhang, Z. N. Tian, Q. D. Chen, and H. B. Sun, General Rules Governing the Dynamical Encircling of an Arbitrary Number of Exceptional Points, Phys. Rev. Lett. 127, 253901 (2021).
  • (24) W. Y. Wang, B. Sun, and J. Liu, Adiabaticity in nonreciprocal Landau-Zener tunneling, Phys. Rev. A106, 063708 (2022).
  • (25) Y. Huang, Y. Shen, C. Min, S. Fan and G. Veronis, Unidirectional reflectionless light propagation at exceptional points, Nanophotonics 6, 977 (2017).
  • (26) K. G. Zloshchastiev, and A. Sergi, Comparison and unification of non-Hermitian and Lindblad approaches with applications to open quantum optical systems, J. Mod. Opt. 61, 1298 (2014).
  • (27) K. G. Makris, R. El-Ganainy, D. N. Christodoulides, and Z. H. Musslimani, Beam Dynamics in PT Symmetric Optical Lattices, Phys. Rev. Lett. 100, 103904 (2008).
  • (28) R. El-Ganainy, K. G. Makris, D. N. Christodoulides, and Z. H. Musslimani, Theory of coupled optical PT-symmetric structures, Opt. Lett. 32, 2632 (2007).
  • (29) A. Guo, G. J. Salamo, D. Duchesne, R. Morandotti, M. Volatier-Ravat, V. Aimez, G. A. Siviloglou, and D. N. Christodoulides, Observation of PT-Symmetry Breaking in Complex Optical Potentials, Phys. Rev. Lett. 103, 093902 (2009).
  • (30) A. Regensburger, C. Bersch, M.-A. Miri, G. Onishchukov, D. N. Christodoulides, and U. Peschel, Parity-time synthetic photonic lattices, Nature 488, 167 (2012).
  • (31) H. Hodaei, M. A. Miri, M. Heinrich, and M. Khajavikhan, Parity-time-symmetric microring lasers, Science 346, 975 (2014).
  • (32) L. Feng, Z. J. Wong, R. M. Ma, Y . Wang, and X. Zhang, Single-mode laser by parity-time symmetry breaking, Science 346, 972 (2014).
  • (33) J. Li, R. Yu, C. Ding, and Y. Wu, PT-symmetry-induced evolution of sharp asymmetric line shapes and high-sensitivity refractive index sensors in a three-cavity array, Phys. Rev. A93, 023814 (2016).
  • (34) S. Longhi, Bloch Oscillations in Complex Crystals with PT Symmetry, Phys. Rev. Lett. 103, 123601 (2009).
  • (35) S. Longhi, Optical Realization of Relativistic Non-Hermitian Quantum Mechanics, Phys. Rev. Lett. 105, 013903 (2010).
  • (36) C. E. Rüter, K. G. Makris, R. El-Ganainy, D. N. Christodoulides, M. Segev, and D. Kip, Observation of parity-time symmetry in optics, Nat. Phys. 6, 192 (2010).
  • (37) Y. Xue, C. Hang, Y. He, Z. Bai, Y. Jiao, G. Huang, J. Zhao, and S. Jia, Experimental observation of partial parity-time symmetry and its phase transition with a laser-driven cesium atomic gas, Phys. Rev. A105, 053516 (2022).
  • (38) A. Stegmaier, S. Imhof, and T. Helbig, Topological Defect Engineering and PT Symmetry in Non-Hermitian Electrical Circuits, Phys. Rev. Lett. 126, 215302 (2021).
  • (39) X. Y. Lü, H. Jing, J. Y. Ma, and Y. Wu, PT-Symmetry-Breaking Chaos in Optomechanics Phys. Rev. Lett. 114, 253601 (2015).
  • (40) C. M. Bender, J. Feinberg, D. W. Hook, et al, Chaotic systems in complex phase space, Pramana - J Phys 73, 453 (2009).
  • (41) C. M. Bender, D. W. Hook, P. N. Meisinger, and Q. H. Wang, Complex Correspondence Principle, Phys. Rev. Lett. 104, 061601 (2010).
  • (42) W. Zhao and H. Zhang, Dynamical stability in a non-Hermitian kicked rotor model, Symmetry 15, 113 (2013).
  • (43) C. T. West, T. Kottos, and T. Prosen, PT-Symmetric Wave Chaos, Phys. Rev. Lett. 104, 054102 (2010).
  • (44) S. Longhi, Localization, quantum resonances, and ratchet acceleration in a periodically kicked PT -symmetric quantum rotator, Phys. Rev. A95 012125 (2017).
  • (45) W. L. Zhao, J. Wang, X. Wang, and P. Tong, Directed momentum current induced by the PT-symmetric driving, Phys. Rev. E99, 042201 (2019).
  • (46) A. Larkin and Y. N. Ovchinnikov, Quasiclassical method in the theory of superconductivity, Sov. Phys. JETP 28, 1200 (1969).
  • (47) D. A. Roberts, and B. Swingle, Lieb-Robinson bound and the butterfly effect in quantum field theories, Phys. Rev. Lett. 117, 091602 (2016).
  • (48) J. Maldacena and D. Stanford, Remarks on the Sachdev-Ye-Kitaev model. Phys. Rev. D94, 106002 (2016).
  • (49) J. Polchinski and V. Rosenhaus, The spectrum in the Sachdev-Ye-Kitaev model, J. High Energ. Phys 2016, 1 (2016).
  • (50) A. Bohrdt, C. Mendl, M. Endres, and M. Knap, Scrambling and thermalization in a diffusive quantum many-body system, New J. Phys. 19, 063001 (2017).
  • (51) M. Gärttner, J. G. Bohnet, A. Safavi-Naini, M. L. Wall, J. J. Bollinger, and A. M. Rey, Measuring out-of-time-order correlations and multiple quantum spectra in a trapped-ion quantum magnet, Nat. Phys. 13, 781 (2017).
  • (52) S. Banerjee and E. Altman, Solvable model for a dynamical quantum phase transition from fast to slow scrambling, Phys. Rev. B95, 134302 (2017).
  • (53) H. Shen, P. Zhang, R. Fan, and H. Zhai, Out-of-time-order correlation at a quantum phase transition, Phys. Rev. B96, 054503 (2017).
  • (54) R. Fan, P. Zhang, H. Shen, and H. Zhai, Out-of-time-order correlation for many-body localization, Science Bulletin 62, 707 (2017).
  • (55) Y. Huang, Y.-L. Zhang, and X. Chen, Out-of-Time- Ordered Correlators in Many-Body Localized Systems, Ann. Phys. (Berlin) 529, 1600318 (2017).
  • (56) J. Li, R. Fan, H. Wang, B. Ye, B. Zeng, H. Zhai, X. Peng, and J. Du, Measuring out-of-time-order correlators on a nuclear magnetic resonance quantum simulator, Phys. Rev. X 7, 031011 (2017).
  • (57) Z. Weinstein, S. P. Kelly, J. Marino, E. Altman, Scrambling Transition in a Radiative Random Unitary Circuit, arXiv:2210.14242 quant-ph.
  • (58) Xi. Hu, T. Luo, and D. Zhang, Quantum algorithm for evaluating operator size with Bell measurements, Phys. Rev. A107, 022407 (2023).
  • (59) S. Pappalardi, J. Kurchan, Low temperature quantum bounds on simple models, SciPost Phys. 13, 006 (2022).
  • (60) I. García-Mata, M. Saraceno, R. A. Jalabert, A. J. Roncaglia, and D. A. Wisniacki, Chaos Signatures in the Short and Long Time Behavior of the Out-of-Time Ordered Correlator, Phys. Rev. Lett. 121, 210601 (2018).
  • (61) J. Wang, G. Benenti, G. Casati, and W. G. Wang, Quantum chaos and the correspondence principle, Phys. Rev. E103, L030201 (2021).
  • (62) R. A. Kidd, A. Safavi-Naini, and J. F. Corney, Saddle-point scrambling without thermalization, Phys. Rev. A103, 033304 (2021).
  • (63) V. Balachandran, G. Benenti, G. Casati, and D. Poletti, From the eigenstate thermalization hypothesis to algebraic relaxation of OTOCs in systems with conserved quantities, Phys. Rev. B104, 104306 (2021).
  • (64) J. Harris, Bin Yan, and N. A. Sinitsyn, Benchmarking Information Scrambling, Phys. Rev. Lett. 129, 050602 (2022).
  • (65) J. Braumüller, et al, Probing quantum information propagation with out-of-time-ordered correlators, Nat. Phys. 18, 172 (2022).
  • (66) Y. L. Zhang, Y. Huang, and X. Chen, Information scrambling in chaotic systems with dissipation, Phys. Rev. B99, 014303 (2019).
  • (67) B. Yan, and N. A. Sinitsyn, Recovery of damaged information and the out-of-time-ordered correlators, Phys. Rev. Lett. 125, 040605 (2020).
  • (68) A. A. Patel, D. Chowdhury, S. Sachdev, and B. Swingle, Quantum butterfly effect in weakly interacting diffusive metals, Phys. Rev. X 7, 031047 (2017).
  • (69) W. Zhao and R. Wang, Scaling laws of out-of-time-order correlators in a non-Hermitian kicked rotor model, Front. Phys 11, 1130225 (2023).
  • (70) K. Hashimoto, K. Murata, and R. Yoshii, Out-of-time-order correlators in quantum mechanics, J. High Energ. Phys. 2017, 138 (2017).
  • (71) S. Zamani, R. Jafari, and A. Langari, Out-of-time-order correlations and floquet dynamical quantum phase transition, Phys. Rev. B105, 094304 (2022).
  • (72) Martin Grttner, P. Hauke, and A. M. Rey, Relating Out-of-Time-Order Correlations to Entanglement via Multiple-Quantum Coherences, Phys. Rev. Lett. 120, 040402 (2018).
  • (73) R. J. Lewis-Swan, A. Safavi-Naini, A. M. Kaufman, and A. M. Rey, Dynamics of Quantum Information, Nat. Rev. Phys. 1, 627 (2019).
  • (74) W. L. Zhao, Quantization of out-of-time-ordered correlators in non-Hermitian chaotic systems, Phys. Rev. Research 4, 023004 (2022)
  • (75) K. X. Wei, C. Ramanathan, and P. Cappellaro, Exploring Localization in Nuclear Spin Chains, Phys. Rev. Lett. 120, 070501 (2018).
  • (76) X. Nie, Z. Zhang, X. Zhao, T. Xin, D. Lu, and J. Li, Detecting scrambling via statistical correlations between randomized measurements on an NMR quantum simulator arXiv:1903.12237.
  • (77) K. A. Landsman, C. Figgatt, T. Schuster, N. M. Linke, B. Yoshida, N. Y. Yao, and C. Monroe, Verified quantum information scrambling, Nature, 567, 61 (2019).
  • (78) S. K. Zhao, Z. Y. Ge, Z. Xiang, G. M. Xue, and S. P, Zhao, Probing Operator Spreading via Floquet Engineering in a Superconducting Circuit arXiv.2108.01276.
  • (79) G. Casati, B. V. Chirikov, F. M. Izrailev, and J. Ford, Stochastic Behavior in Classical and Quantum Hamiltonian Systems, edited by G. Casati and J. Ford, Lecture Notes in Physics, Vol. 93 (Springer, Berlin, 1979).
  • (80) T. Mano and T. Ohtsuki, Machine learning the dynamics of quantum kicked rotor, Annals of Physics, 435, 168500 (2021).
  • (81) X. Chen, T. Zhou, D. A. Huse, and E. Fradkin, Out-of-time-order correlations in many-body localized and thermal phases, Ann. Phys. 529, 1600332.
  • (82) B. Dóra and R. Moessner, Out-of-time-ordered density correlators in Luttinger liquids, Phys. Rev. Lett. 119, 026802 (2017).
  • (83) I. García-Mata, M. Saraceno, R. A. Jalabert, A. J. Roncaglia, and D. A. Wisniacki, Chaos signatures in the short and long time behavior of the out-of-time ordered correlator, Phys. Rev. Lett. 121, 210601 (2018).
  • (84) Y. Alavirad and A. Lavasani, Scrambling in the Dicke model, Phys. Rev. A99, 043602. (2018).
  • (85) B. Swingle, G. Bentsen, M. Schleier-Smith, and P. Hayden, Measuring the scrambling of quantum information, Phys. Rev. A94, 040302 (2016).
  • (86) G. Zhu, M. Hafezi, and T. Grover, Measurement of many-body chaos using a quantum clock, Phys. Rev. A94, 062329 (2016).
  • (87) M. Heyl, F. Pollmann, and B. Dóra, Detecting equilibrium and dynamical quantum phase transitions in Ising chains via out-of-time-ordered correlators, Phys. Rev. Lett. 121, 016801 (2018).
  • (88) L. D’Alessio and M. Rigol, Long-time Behavior of Isolated Periodically Driven Interacting Lattice Systems, Phys. Rev. X 4, 041048 (2014).
  • (89) W. L. Zhao, Y. Hu, Z. Li, and Q. Wang, Super-exponential growth of Out-of-time-ordered correlators, Phys. Rev. B103, 184311 (2021).
  • (90) R. Hamazaki, K. Fujimoto, and M. Ueda, Operator Noncommutativity and Irreversibility in Quantum Chaos arXiv:1807.02360.
  • (91) F.M. Izrailev, Simple modelds of quantum chaos: spectrum and eigenfunctions, Phys. Rep. 196, 299 (1990), and references therein.
  • (92) S. Moudgalya, T. Devakul, C. W. von Keyserlingk, and S. L. Sondhi, Operator spreading in quantum maps, Phys. Rev. B, 99 094312 (2019).
  • (93) T. Zhou, S. Xu, X. Chen, A. Guo, and B. Swingle, Operator Lévy Flight: Light Cones in Chaotic Long-Range Interacting Systems, Phys. Rev. Lett. , 124, 180601 (2020).
  • (94) K. X. Wei, P. Peng, O. Shtanko, I. Marvian, S. Lloyd, C. Ramanathan, and P. Cappellaro, Emergent Prethermalization Signatures in Out-of-Time Ordered Correlations, Phys. Rev. Lett. 123, 090605 (2019).
  • (95) T. Boness, K. Kudo, and T. S. Monteiro, Doubly excited ferromagnetic spin-chain as a pair of coupled kicked rotors, Phys. Rev. E81, 046201 (2010).