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

Super-exponential scrambling of Out-of-time-ordered correlators

Wen-Lei Zhao wlzhao@jxust.edu.cn School of Science, Jiangxi University of Science and Technology, Ganzhou 341000, China    Yue Hu Guangdong Provincial Key Laboratory of Quantum Engineering and Quantum Materials, GPETR Center for Quantum Precision Measurement, Frontier Research Institute for Physics and SPTE, South China Normal University, Guangzhou 510006, China    Zhi Li lizhi˙phys@m.scnu.edu.cn Guangdong Provincial Key Laboratory of Quantum Engineering and Quantum Materials, GPETR Center for Quantum Precision Measurement, Frontier Research Institute for Physics and SPTE, South China Normal University, Guangzhou 510006, China    Qian Wang qwang@zjnu.edu.cn Department of Physics, Zhejiang Normal University, Jinhua 321004, China
Abstract

Out-of-time-ordered correlators (OTOCs) are an effective tool in characterizing black hole chaos, many-body thermalization and quantum dynamics instability. Previous research findings have shown that the OTOCs’ exponential growth (EG) marks the limit for quantum systems. However, we report in this letter a periodically-modulated nonlinear Schrödinger system, in which we interestingly find a novel way of information scrambling: super-EG. We show that the quantum OTOCs’ growth, which stems from the quantum chaotic dynamics, will increase in a super-exponential way. We also find that in the classical limit, the hyper-chaos revealed by a linearly-increasing Lyapunov exponent actually triggers the super-EG of classical OTOCs. The results in this paper break the restraints of EG as the limit for quantum systems, which give us new insight into the nature of information scrambling in various fields of physics from black hole to many-body system.

Introduction.— Akin to quantum butterfly effects, quantum scrambling is the process of encoded information spreading from local degrees of freedom to multiple degrees of freedom, hence comes its unattainability by measuring the local operators Swingle19 ; Lewis19 . However, through the time evolution of out-of-time-ordered correlators (OTOCs)  Xu20 ; Prakash20 ; Sekino08 ; Landsman19 ; Rozenbaum20 , this elusive process can be well quantified, which explains the enthusiastic and extensive study of OTOCs in many frontiers of physics such as quantum holography, quantum chaos and black hole physics. Recent research has proved that OTOCs act as an effective indicator of quantum phase transition Dag19 ; Sahu19 ; Wangqian19 ; Heyl2018 , many-body localization Rammensee18 ; Ray18 ; Huang16 and quantum entanglement Garttner18 . Besides, experimental progress has made it possible to observe the OTOCs in atomic-optics setups Rammensee18 ; Ray18 and nuclear spins Li2017 . Up to now, a wide range of OTOCs with logarithmic, power-law or exponential growth (EG) has been found in various systems including many-body systems and quantum chaotic systems.

As proposed in landmark studies of quantum chaos, the OTOCs can be used to describe the exponential instability in quantum dynamics Hashimoto2017 ; Mata2018B ; Fortes19 ; Dora2017 ; Rozenbaum17 ; Rozenbaum19 ; Yan19 . As for systems with the well-defined classical limit, the EG will occur within Ehrenfest time with a rate determined by the classical Lyapuonv exponent Mata2018B ; Fortes19 ; Rozenbaum17 ; Lakshmi19 . Interestingly, the mathematically verified correlation between OTOCs and Loschmidt echo provides theoretical basis for the irreversibility of scrambling dynamics Zurek20 . Note that, the boundary set by chaos dynamics for exponential scrambling of OTOCs, which means that an exponent actually marks the greatest rate of increase for OTOCs in a quantum system, is obtained by the conjecture of thermalization in quantum systems with a large number of degrees of freedom Murthy19 ; Maldacena16 . At present, massive research efforts have been focused on how many-body chaos affects the dynamics of OTOCs. For instance, a recent study has shown that the OTOCs’ EG is equal to the classical Lyapunov exponent in the presence of interatomic interaction Mata2018 . However, we noticed that the growth rate of OTOCs in previous systems fails to break the exponential limit. Therefore, is it safe to say that the EG actually tops all OTOCs’ rates of growth in quantum systems?

Model and results.— We consider a Schrödinger system with the temporally modulated nonlinear interaction, the corresponding Hamiltonian reads Zhao16 ; Zhao19

H=p22+g|ψ(θ,t)|2jδ(tj),{\rm H}=\frac{p^{2}}{2}+g|\psi(\theta,t)|^{2}\sum_{j}\delta(t-j)\;, (1)

where the angular momentum operator p=ieff/θp=-i\hbar_{\mathrm{eff}}\partial/\partial\theta with eff\hbar_{\mathrm{eff}} being the effective Planck constant, θ\theta is angle coordinate, gg denotes the nonlinear interaction strength. All variables are properly scaled and thus in dimensionless units. An arbitrary quantum state is expanded in terms of the complete basis of angular momentum operator (p|φn=neff|φnp|\varphi_{n}\rangle=n\hbar_{\mathrm{eff}}|\varphi_{n}\rangle), i.e., |ψ=nψn|φn|\psi\rangle=\sum_{n}\psi_{n}|\varphi_{n}\rangle, hence, it is periodical in θ\theta, i.e., ψ(θ)=ψ(θ+2π)\psi(\theta)=\psi(\theta+2\pi). The one-period evolution operator from time tt to t+1t+1 is given by U(t,t+1)=exp(ip2/2eff)exp[ig|ψ(θ,t)|2/eff]U(t,t+1)=\exp\left(-i{p^{2}}/{2\hbar_{\mathrm{eff}}}\right)\exp\left[-i{g|\psi(\theta,t)|^{2}}/{\hbar_{\mathrm{eff}}}\right]. The quantum OTOCs are defined as the average of the squared commutator, i.e., C(t)=[A^(t),B^(0)]2C(t)=-\langle[\hat{A}(t),\hat{B}(0)]^{2}\rangle, where A^(t)=eiHtA(0)eiHt\hat{A}(t)=e^{iHt}A(0)e^{-iHt} is the time-dependent operator in Heisenberg picture, with \langle\cdot\rangle being the average of the initial state. In the many-body systems, the average of C(t)C(t) comes from thermal states. For the quantum mapping systems, however, there is no definition of thermal states, as the temperature tends to be infinitely large after long time evolution Rigol14 . Indeed, our previous investigations have proved that the system in Eq. (1) exhibits the unbounded heating, which is quantified by the exponentially-fast growth of mean energy p2\langle p^{2}\rangle Zhao16 ; Zhao19 .

Here, we consider the case of a pure state, i.e., a Gaussian wavepacket ψ(θ,0)=(σ/π)1/4exp(σθ2/2)\psi(\theta,0)=(\sigma/\pi)^{1/4}\exp(-\sigma\theta^{2}/2). As in ref. Rozenbaum17 , A^=B^=p\hat{A}=\hat{B}=p, namely

C(t)=[p(t),p(0)]2.C(t)=-\left\langle[{p}(t),{p}(0)]^{2}\right\rangle\;. (2)

Our main result is the analytical prediction of the Super-EG scrambling of the quantum OTOCs

C(t)exp[αγt+βln(gπeff)2t+ηγt2],C(t)\propto\exp\left[\alpha\gamma t+\beta\ln\left(\frac{g}{\pi\hbar_{\mathrm{eff}}}\right)^{2}t+\eta\gamma t^{2}\right]\;, (3)

where

γ=ln[1+(g𝒩0πeff)2],\gamma=\ln\left[1+\left(\frac{g\mathcal{N}_{0}}{\pi\hbar_{\mathrm{eff}}}\right)^{2}\right]\;, (4)

α\alpha, β\beta and η\eta are prefactors and 𝒩0\mathcal{N}_{0} is the normalization constant of the initial state (usually 𝒩0=1\mathcal{N}_{0}=1). Numerical results of OTOCs are in good agreement with the analytical expression (see Fig. 1). It is worth noting that EG is usually believed to be the boundary of the scrambling of OTOCs in chaotic systems Maldacena16 , therefore, our finding of the super-EG scrambling sheds new light in the field of quantum information Dressel18 ; Alonso19 . We have also investigated, both numerically and analytically, the OTOCs defined as C(t)=[θ(t),p]2C(t)=-\langle[\theta(t),p]^{2}\rangle. Interestingly, we found the scaling

C(t)=μ(t)Nexp(νγt),C(t)=\mu(t)N\exp(\nu\gamma t)\;, (5)

where μ\mu is the time-dependent coefficient, NN is number of basis, namely the dimension of the system, ν\nu is a constant, and the growth rate reads

γ=ln{1+[g𝒩~(t)πeff]2}\gamma=\ln\left\{1+\left[\frac{g\tilde{\mathcal{N}}(t)}{\pi\hbar_{\mathrm{eff}}}\right]^{2}\right\}\; (6)

with 𝒩~(t)=ψ(t)|θ2|ψ(t)\tilde{\mathcal{N}}(t)=\langle\psi(t)|\theta^{2}|\psi(t)\rangle Supple . It is obvious that the C(t)C(t) approaches to infinite with the increase of the dimension of the system NN, which demonstrates that there is no bound on the OTOCs in our system.

The bound of the exponential growth of OTOCs applies to many-body systems with finite temperature where the thermal states are well defined. For such systems, the two-point correlator in OTOCs saturates rapidly. The exponential growth of OTOCs mainly results from the four-point correlator Maldacena16 . It has proved that the periodically-driven systems are equivalent to the systems with infinite temperature Rigol14 , for which the thermal states cannot be well defined without an effective Hamiltonian. These essential differences lead to the inconsistency between the super-EG of OTOCs in our system and the bound of chaos in ref. Maldacena16 . Remarkably, for periodically-driven systems, the two-point correlator contributes mainly to the OTOCs, and the four-point correlator saturates rapidly Rozenbaum17 ; Ueda2018 . In the following, we will show that the two-point correlator part exhibits the super-exponential growth with time. Here, the super-EG of OTOCs is rooted in the exponentially-fast diffusion of mean energy. It is a clear evidence that the scrambling dynamics is closely related to quantum thermalization, which is highlighted in such interdisciplinary topics as quantum information scrambling and quantum chaos. On the other hand, as a variant of kicked rotor model, our system is an ideal platform for investigating the wavepacet dynamics, such as quantum walk Dadras18 and topologically-protected transport Ho12 in momentum-space lattice. Therefore, our finding opens a new prospective in the field of the scrambling dynamics in momentum-space lattice.

Refer to caption
Figure 1: (color online). (on log-linear scale) Quantum OTOC C(t)C(t) versus time with eff=0.6\hbar_{\mathrm{eff}}=0.6 for g=1.3g=1.3 (squares), 1.51.5 (circles), 2.02.0 (triangles) and 3.03.0 (diamonds). Red solid lines indicate our theoretical prediction in Eq. (S6). The width of Gaussian wavepacket is σ=1.0\sigma=1.0.

We proceed to evaluate the scrambling in the semiclassical limit. While approaching to the semiclassical limit, the quantum commutator reduces to Possion bracket [p(t),p(0)]=eff{p(t),p(0)}=effp(t)/x(0)[p(t),p(0)]=\hbar_{\mathrm{eff}}\{p(t),p(0)\}=\hbar_{\mathrm{eff}}{\partial p(t)}/{\partial x(0)}. Then, a natural definition of the classical OTOCs are

Ccl(t)=(p(t)x(0))2,C_{cl}(t)=\left\langle\left(\frac{\partial p(t)}{\partial x(0)}\right)^{2}\right\rangle\;, (7)

where \langle\cdot\rangle denotes the average of the ensemble of classical trajectories Rozenbaum17 . In numerical calculations, the classical OTOCs are approximated as Ccl(t)(δp(t)/δx(0))2C_{cl}(t)\approx\langle(\delta p(t)/\delta x(0))^{2}\rangle, where δp\delta p and δx\delta x denote the difference of the two nearest neighboring trajectories Rozenbaum17 .

We have proved that the system in Eq. (1) is mathematically equivalent to a generalized kicked rotor (GKR) model Zhao16 ; Zhao19 ; Supple , whose Hamiltonian takes the form

H=p22+n=1+Kn(t)cos(nθ)j(tj).\displaystyle\textrm{H}=\frac{p^{2}}{2}+\sum_{n=1}^{+\infty}K_{n}(t)\cos(n\theta)\sum_{j}(t-j)\;. (8)

The kicking strength dependent on the Fourier components of the quantum state reads

Kn(t)=gπm=+ψm(t)ψm+n(t).K_{n}(t)=\frac{g}{\pi}\sum_{m=-\infty}^{+\infty}\psi_{m}^{*}(t)\psi_{m+n}(t). (9)

The classical mapping equation of this GKR model can be formulated as

{p(t+1)p(t)=n=1+nKn(t)sin[nθ(t)]θ(t+1)θ(t)=p(t+1),\begin{cases}p(t+1)-p(t)=\sum_{n=1}^{+\infty}nK_{n}(t)\sin\left[n\theta(t)\right]\\ \theta(t+1)-\theta(t)=p(t+1),\end{cases} (10)

where p(t)p(t) and θ(t)\theta(t) indicate the classical momentum and angle variables after the tt-th kick.

Based on the classical mapping equations, we numerically investigate the classical OTOCs for different interaction gg. In numerical simulations, we set the initial values of pp and θ\theta as random variables with the probability distribution given by Gaussian function in phase space. The difference of initial trajectories is δθ(0)=105\delta\theta(0)=10^{-5}. Interestingly, the classical OTOCs increase in the super-EG way, i.e.,

Ccl(t)exp(γt2),C_{cl}(t)\propto\exp\left(\gamma t^{2}\right)\;, (11)

which is in good agreement with our theoretical prediction (see Fig. 2(a)). In addition, we numerically investigate the maximal Lyapunov exponent λ=limtlimδθ0log[δp(t)/δθ(0)]/t\lambda=\lim_{t\rightarrow\infty}\lim_{\delta\theta\rightarrow 0}\langle\log[\delta p(t)/\delta\theta(0)]\rangle/t. Remarkably, the maximal Lyapunov exponent λ\lambda linearly increases with time λ(t)γt\lambda(t)\propto\gamma t (see Fig. 2(a)), which is in consistence with the theoretical prediction in Eq. (18). This clearly demonstrates the existence of the hyper-chaotic dynamics Andreev19 . Different from previous research, in which the classical OTOCs depend on the λ\lambda in an exponential way of Ccl(t)eλtC_{cl}(t)\propto e^{\lambda t} Mata2018B ; Fortes19 ; Rozenbaum17 ; Lakshmi19 , here we get the super-EG with the form Ccl(t)eγt2C_{cl}(t)\propto e^{\gamma t^{2}}. Compared with traditional quantum systems Guarneri17 ; Gisin95 , richer physics will be exhibited in the periodically-modulated nonlinear Schrödinger system from the perspective of either quantum dynamics or semi-classical dynamics.

Refer to caption
Figure 2: (color online). Time dependence of the classical OTOC Ccl(t)C_{cl}(t) (a) (on log-linear scale) and the maximal Lyapunov exponent (b) with g=1.3g=1.3 (squares), 1.51.5 (circles), 2.02.0 (triangles) and 3.03.0 (diamonds). Red dashed lines in (a) and (b) indicate our theoretical predictions in Eqs. (11) and (18), respectively. Other parameters are the same as in Fig. 1.

Theoretical analysis.— It is straightforward to decompose the OTOCs in Eq. (2) as

C(t)=C1(t)+C2(t)2Re[C3(t)],C(t)=C_{1}(t)+C_{2}(t)-2\text{Re}[C_{3}(t)]\;, (12)

where we define the terms in the right of the above equation as

C1(t)=ψ(0)|U(t)pU(t)p2U(t)pU^(t)|ψ(0),C2(t)=ψ(0)|pU(t)pU(t)U(t)pU(t)p|ψ(0),C3(t)=ψ(0)|U(t)pU(t)pU(t)pU^(t)p|ψ(0).\begin{split}C_{1}(t)&=\left\langle\psi(0)\left|{U}^{\dagger}(t)p{U}(t)p^{2}{U}^{\dagger}(t)p\hat{U}(t)\right|\psi(0)\right\rangle\;,\\ C_{2}(t)&=\left\langle\psi(0)\left|p{U}^{\dagger}(t)p{U}(t){U}^{\dagger}(t)p{U}(t)p\right|\psi(0)\right\rangle\;,\\ C_{3}(t)&=\left\langle\psi(0)\left|{U}^{\dagger}(t)p{U}(t)p{U}^{\dagger}(t)p\hat{U}(t)p\right|\psi(0)\right\rangle\;.\end{split} (13)

Our numerical investigation shows that the contribution of the first term C1(t)C_{1}(t) is larger than the other two about several orders of magnitude, which means C(t)C1(t)C(t)\approx C_{1}(t) Ueda2018 . The two-point correlation function quantifies the quantum irreversibility measured by the expectation value of p2p^{2} under the perturbation of pp at the time tt. Hereinafter, we will show how the exponentially-fast diffusion of mean energy (p2eγt\langle p^{2}\rangle\propto e^{\gamma t}) induces the super-EG of C1(t)C_{1}(t).

The definition of C1(t)C_{1}(t) in Eq. (13) means that there are four steps in the calculation of this part. The first step is the forward time evolution of an initial state from t0t_{0}(=0=0) to t+=t0+tt_{+}=t_{0}+t, which yields the quantum state |ψ(t+)=U(t)|ψ(0)|\psi(t_{+})\rangle={U}(t)|\psi(0)\rangle. In the second step, the operator pp is exerted on the state |ψ(t+)|\psi(t_{+})\rangle, i.e., |ψ~(t+)p|ψ(t+)=pU(t)|ψ(0)|\tilde{\psi}(t_{+})\rangle\equiv p|\psi(t_{+})\rangle={p}{U}(t)|\psi(0)\rangle. The norm of this state has the expression 𝒩t+=ψ~(t+)|ψ~(t+)=ψ(0)|U(t)p2U(t)|ψ(0)\mathcal{N}_{t_{+}}=\langle\tilde{\psi}(t_{+})|\tilde{\psi}(t_{+})\rangle=\langle\psi(0)|{U}^{\dagger}(t){p}^{2}{U}(t)|\psi(0)\rangle, which is just the mean energy at the time t+t_{+}. Our previous investigation has shown that the mean energy exponentially increases Zhao16 ; Zhao19 , i.e.,

𝒩t+𝒩0exp(γt)=exp(γt),\mathcal{N}_{t_{+}}\propto\mathcal{N}_{0}\exp(\gamma t)=\exp(\gamma t)\;, (14)

where we adopted the condition 𝒩0=ψ(0)|ψ(0)=1\mathcal{N}_{0}=\langle\psi(0)|\psi(0)\rangle=1. The third step is the time reversal from t+t_{+} to t0t_{0} (tt steps), which results in a state |φ(t0)U(t)|ψ~(t+)=U(t)p^U^(t)|ψ(0)|\varphi(t_{0})\rangle\equiv{U}^{\dagger}(t)|\tilde{\psi}(t_{+})\rangle={U}^{\dagger}(t)\hat{p}\hat{U}(t)|\psi(0)\rangle. Finally, in the fourth step, one can obtain the expectation value of p2p^{2} with the state |φ(t0)|\varphi(t_{0})\rangle, i.e., C1(t)φ(t0)|p2|φ(t0)C_{1}(t)\approx\langle\varphi(t_{0})|p^{2}|\varphi(t_{0})\rangle =ψ~(t+)|U(t)p2U(t)|ψ~(t+)=\langle\tilde{\psi}(t_{+})|{U}(t)p^{2}{U}^{\dagger}(t)|\tilde{\psi}(t_{+})\rangle.

Note that, the process of time reversal (the third step) can be viewed as a process starting from a new initial state |ψ~(t+)|\tilde{\psi}(t_{+})\rangle normalized to 𝒩t+\mathcal{N}_{t_{+}}, and then evolving into a quantum state |φ(t0)|\varphi(t_{0})\rangle in a time interval tt. The two-point correlator C1(t)C_{1}(t) is just the “mean energy” of the state |φ(t0)|\varphi(t_{0})\rangle, which follows the exponentially-fast way

C1(t)𝒩t+exp(γt+t)exp(γt+γt+t)C_{1}(t)\propto\mathcal{N}_{t_{+}}\exp(\gamma_{t_{+}}t)\propto\exp(\gamma t+\gamma_{t_{+}}t)\; (15)

with the rate γt+=ln[1+(g𝒩t+/πeff)2]\gamma_{t_{+}}=\ln\left[1+\left({g{{\cal{N}}_{t_{+}}}}/{\pi\hbar_{\mathrm{eff}}}\right)^{2}\right] Zhao16 ; Zhao19 . Taking Eq. (14) into account, the γt+{\gamma}_{t_{+}} has the expression

γt+\displaystyle{\gamma}_{t_{+}} ln[1+(gπeff)2e2γt]\displaystyle\propto\ln\left[1+\left(\frac{g}{\pi\hbar_{\mathrm{eff}}}\right)^{2}e^{2\gamma t}\right] (16)
ln(gπeff)2+γt,\displaystyle\propto\ln\left(\frac{g}{\pi\hbar_{\mathrm{eff}}}\right)^{2}+\gamma t\;,

where we use the condition e2γt1e^{2\gamma t}\gg 1. As a consequence, the OTOCs take the form

C(t)C1(t)exp[αγt+βln(gπeff)2t+ηγt2],C(t)\approx C_{1}(t)\propto\exp\left[\alpha\gamma t+\beta\ln\left(\frac{g}{\pi\hbar_{\mathrm{eff}}}\right)^{2}t+\eta\gamma t^{2}\right]\;, (17)

where the prefactors α\alpha, β\beta and η\eta can not be exactly obtained, since in the process of derivation we have used approximations in Eqs. (14), (15) and (16). According to the above analysis, the super-EG of OTOCs is mainly caused by the positive feedback mechanism of the temporally-modulated interaction, which is the key point in this letter. The exponential increase of the kicking strength, which is absent in the traditional kicked rotor model, is responsible for the appearance of super-EG in the classical limits of the system.

Next, we analyze the time evolution of the classical OTOCs. For the kicked rotor model, the maximum Lyapunov depends on the kicking strength by the way of λln(K)\lambda\propto\ln(K) Chirikov79 ; Rozenbaum17 . In the GKR model, the definition of one of the components of the kicking strength KnK_{n} in Eq. (9) indicates that the KnK_{n} is the quantum correlation in momentum space. A significant feature of this system is the exponential localization of quantum states, i.e., |ψ(p)|2exp(|p|/ξ)|\psi(p)|^{2}\sim\exp(-|p|/\xi) with ξ\xi being the localization length. Through simple calculation, one can obtain that the quantum correlation also has the exponential decay, i.e., Knexp(|p|/ξ)K_{n}\propto\exp(-|p|/\xi) Zhao16 ; Zhao19 . Then, a rough estimation of the kicking strength in the GKR model (see Eq. (10)) is K(t)=nnKn(t)nne|n|eff/ξξK(t)=\sum_{n}nK_{n}(t)\propto\sum_{n}ne^{-|n|\hbar_{\mathrm{eff}}/\xi}\propto\xi. Previously, we have predicted the exponentially-fast increase of the localization length ξeγt\xi\propto e^{\gamma t} by means of the hybrid quantum-classical theory Zhao16 ; Zhao19 . As a consequence, the kick strength exponentially grows with time, i.e., KeγtK\propto e^{\gamma t}. Accordingly, the Lyapunov exponent linearly increases with time

λln(K)γt.\lambda\propto\ln(K)\propto\gamma t\;. (18)

For strong chaotic case, the classical OTOCs exponentially increase with the growth rate proportional to Lyapunov exponent Mata2018B ; Fortes19 ; Rozenbaum17 ; Lakshmi19 , i.e., Ccl(t)eλtC_{cl}(t)\propto e^{\lambda t}. Taking Eq. (18) into account, one can get Ccl(t)exp(γt2)C_{cl}(t)\propto\exp(\gamma t^{2}), which is confirmed by our numerical results (see Fig. 2).

Conclusion and prospects—We have proved the existence of OTOCs’ super-EG. The results not only give evidence to verify the association between quantum hyper-chaos and quantum scrambling, but also confirm the super-exponential sensitivity of quantum dynamics to initial conditions. Since the conventional theory has it that a EG boundary is imposed on the chaotic dynamics, our finding, by breaking traditional restraints, bears great significance in the research of information scrambling and quantum chaos Ray18 . More importantly, the super-EG is the fastest growth rate of OTOC today, for which, the underlying mechanism is the positive feedback of the periodic modulated interaction. Our finding helps review the issue of the chaotic systems’ boundary.

Experimentally, due to the generality of the nonlinear Schrödiner system, our findings will serve as a universal theory in broad fields including the cold atomic gases Dalfovo99 , nonlinear optics Szameit10 ; Longhi11 ; Blomer06 and complex Ginzburg-Landau equation in condensed-matter physics Anderson2002 . More remarkably, the interaction in the above-mentioned systems features high controllability Inouye98 ; Kevrekidis03 ; Theis04 ; Anderson2002 ; Chin10 ; Szameit10 ; Longhi11 ; Blomer06 , which will facilitate the experimental realization and observation of the predictions.

Acknowledgements.
We are grateful to Jie Liu and Jiaozi Wang for valuable suggestions and discussions. This work was supported by the Natural Science Foundation of China under Grant Nos. 12065009 and 11704132, the PCSIRT (Grant No. IRT1243), the Natural Science Foundation of Guangdong Province (No. 2018A030313322, and No. 2018A0303130066), the KPST of Guangzhou (Grant No. 201804020055) and science and technology program of Guangzhou (No. 2019050001)

References

  • (1) B. Swingle, Unscrambling the physics of out-of-time-order correlators, Nat. Phys. 14, 988 (2018).
  • (2) R. J. Lewis-Swan, A. Safavi-Naini, J. J. Bollinger, and A. M. Rey, Unifying scrambling, thermalization and entanglement through measurement of fidelity out-of-time-order correlators in the Dicke model, Nat. Commun. 10, 1581 (2019).
  • (3) T. Xu, T. Scaffidi, and X. Cao, Does Scrambling Equal Chaos? Phys. Rev. Lett. 124, 140602 (2020).
  • (4) R. Prakash and A. Lakshminarayan, Scrambling in strongly chaotic weakly coupled bipartite systems: Universality beyond the Ehrenfest timescale, Phys. Rev. B101 121108(R) (2020).
  • (5) Y. Sekino and L. Susskind, Fast scramblers, Fast scramblers, J. High Energy Phys. 08, 065 (2008).
  • (6) K. A. Landsman, C. Figgatt, T. Schuster, N.M. Linke, B. Yoshida, N. Y. Yao, and C. Monroe, Verified quantum information scrambling, Nature (London) 567, 61 (2019).
  • (7) E. B. Rozenbaum, S. Ganeshan, and V. Galitski, Universal level statistics of the out-of-time-ordered operator, Phys. Rev. B100 035112 (2019).
  • (8) C. B. Daǧ , Kai Sun, and L.-M. Duan, Detection of Quantum Phases via Out-of-Time-Order Correlators, Phys. Rev. Lett. 123 140602 (2019).
  • (9) S. Sahu, S. Xu, and B. Swingle, Scrambling Dynamics across a Thermalization-Localization Quantum Phase Transition, Phys. Rev. Lett. 123, 165902 (2019).
  • (10) Q. Wang and F. Pérez-Bernal, Probing an excited-state quantum phase transition in a quantum many-body system via an out-of-time-order correlator, Phys. Rev. A100, 062113 (2019).
  • (11) 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).
  • (12) S. Ray, S. Sinha, and K. Sengupta, Signature of chaos and delocalization in a periodically driven many-body system: An out-of-time-order-correlation study, Phys. Rev. A98, 053631 (2018).
  • (13) J. Rammensee, J. D. Urbina, and K. Richter,  it Many-Body Quantum Interference and the Saturation of Out-of-Time-Order Correlators, Phys. Rev. Lett. 121, 124101 (2018).
  • (14) Y. Huang, Y. L. Zhang, and X. Chen, Out-of-time-ordered correlators in many-body localized systems, Ann. Phys. 529, 1600332 (2017).
  • (15) M. Gärttner, P. Hauke, and A. M. Rey, Relating Out-of-Time-Order Correlations to Entanglement via Multiple-Quantum Coherences, Phys. Rev. Lett. 120 040402 (2018).
  • (16) 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).
  • (17) K. Hashimoto, K. Murata, and R. Yoshii, Out-of-time-order correlators in quantum mechanics, J. High Energ. Phys. 10, 138 (2017).
  • (18) R. A. Jalabert, I. García-Mata, and D. A. Wisniacki, Semiclassical theory of out-of-time-order correlators for low-dimensional classically chaotic systems, Phys. Rev. E98, 062218 (2018).
  • (19) E. M. Fortes, I. García-Mata, R. A. Jalabert, and D. A. Wisniacki, Gauging classical and quantum integrability through out-of-time-ordered correlators, Phys. Rev. E100 042201 (2019).
  • (20) B. Dóra and R. Moessner, Out-of-Time-Ordered Density Correlators in Luttinger Liquids, Phys. Rev. Lett. 119, 026802 (2017).
  • (21) E. B. Rozenbaum, S. Ganeshan, and V. Galitski, Lyapunov Exponent and Out-of-Time-Ordered Correlator¡¯s Growth Rate in a Chaotic System, Phys. Rev. Lett. 118 086801 (2017).
  • (22) E. B. Rozenbaum, L. A. Bunimovich, and V. Galitski, Early-Time Exponential Instabilities in Nonchaotic Quantum Systems, Phys. Rev. Lett. 125 014101 (2020).
  • (23) H. Yan, J. Z. Wang, and W. G. Wang, Similar early growth of out-of-time-ordered correlators in quantum chaotic and integrable ising chains, Commun. Theor. Phys. (Beijing, China) 71, 1359 (2019).
  • (24) A. Lakshminarayan, Out-of-time-ordered correlator in the quantum bakers map and truncated unitary matrices, Phys. Rev. E99, 012201 (2019).
  • (25) B. Yan , L. Cincio, and W. H. Zurek, Information Scrambling and Loschmidt Echo, Phys. Rev. Lett. 124 160603 (2020).
  • (26) C. Murthy and M. Srednicki, Bounds on Chaos from the Eigenstate Thermalization Hypothesis, Phys. Rev. Lett. 123, 230606 (2019).
  • (27) J. Maldacena, S. H. Shenker, and D. Stanford, A bound on chaos, J. High Energy Phys. 08 106 (2016).
  • (28) 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).
  • (29) W. Zhao, J. Gong, W. Wang, G. Casati, J. Liu, and L. Fu, Exponential wave-packet spreading via self-interaction time modulation, Phys. Rev. A94, 053631 (2016).
  • (30) W. Zhao, J. Wang, and W. Wang, Quantum-classical correspondence in a nonlinear Gross-Pitaevski system, J. Phys. A: Math. Theor. 52, 305101 (2019).
  • (31) Luca D’Alessio and Marcos Rigol, Long-time Behavior of Isolated Periodically Driven Interacting Lattice Systems, Phys. Rev. X 4, 041048 (2014).
  • (32) J. Dressel, J. R. González Alonso, M. Waegell, and N. Y. Halpern, Strengthening weak measurements of qubit out-of-time-order correlators, Phys. Rev. A98, 012132 (2018).
  • (33) J. R. González Alonso, N. Yunger Halpern, and J. Dressel, Out-of-Time-Ordered-Correlator Quasiprobabilities Robustly Witness Scrambling Phys. Rev. Lett. 122, 040404 (2019).
  • (34) See Supplementary materials for detailed derivation of the scaling of OTOCs in Eq. (S8) and the derivation of the Hamiltonian in Eq. (S4). There is numerical procedure for calculating the OTOCs and numerical results for the comparision of quantum and classical OTOCs during short time evolution.
  • (35) R. Hamazaki, K. Fujimoto, and M. Ueda, Operator Noncommutativity and Irreversibility in Quantum Chaos arXiv:1807.02360 [condmat. stat-mech].
  • (36) S. Dadras, A. Gresch, C. Groiseau, S. Wimberger, and G. S. Summy Quantum Walk in Momentum Space with a Bose-Einstein Condensate, Phys. Rev. Lett. , 121, 070402 (2018).
  • (37) D. Y. H. Ho and J. Gong, Quantized Adiabatic Transport in Momentum Space, Phys. Rev. Lett. 109, 010601 (2012).
  • (38) A. V. Andreev, A. G. Balanov, T. M. Fromhold, M. T. Greenaway, A. E. Hramov, W. Li, V. V. Makarov, A. M. Zagoskin, Emergence and Control of Complex Behaviours in Driven Systems of Interacting Qubits with Dissipation, arXiv:1907.03602 [quant-ph]
  • (39) I. Guarneri, Gross-Pitaevski map as a chaotic dynamical system, Phys. Rev. E95, 032206 (2017).
  • (40) N. Gisin and M. Rigo, Relevant and irrelevant nonlinear Schrodinger equations, J. Phys. A (Math. Gen.) 28, 7375 (1995).
  • (41) B. V. Chirikov, A universal instability of many-dimensional oscillator systems, Phys. Rep. 52, 263 (1979).
  • (42) F. Dalfovo and S. Giorgini, Theory of Bose-Einstein condensation in trapped gases, Rev. Mod. Phys. 71, 463 (1999).
  • (43) A. Szameit and S. Nolte, Discrete optics in femtosecond-laser-written photonic structures, J. Phys. B 43, 163001 (2010).
  • (44) S. Longhi, Many-body coherent destruction of tunneling in photonic lattices, Phys. Rev. A83, 034102 (2011).
  • (45) D. Blömer, A. Szameit, F. Dreisow, T. Schreiber, S. Nolte, and A. Tünnermann, Nonlinear refractive index of fs-laser-written waveguides in fused silica, Opt. Express 14, 2151 (2006).
  • (46) I. S. Aranson and L. Kramer, The world of the complex Ginzburg-Landau equation, Rev. Mod. Phys. 74, 99 (2002).
  • (47) S. Inouye, M. R. Andrews, J. Stenger, H. J. Miesner, D. M. Stamper-Kurn, and W. Ketterle, Observation of Feshbach resonances in a Bose-Einstein condensate, Nature(London) 392, 151 (1998).
  • (48) P. G. Kevrekidis, G. Theocharis, D. J. Frantzeskakis, and B. A. Malomed, Feshbach Resonance Management for Bose-Einstein Condensates, Phys. Rev. Lett. 90, 230401 (2003).
  • (49) M. Theis, G. Thalhammer, K. Winkler, M. Hellwig, G. Ruff, R. Grimm, and J. H. Denschlag, Tuning the Scattering Length with an Optically Induced Feshbach Resonance, Phys. Rev. Lett. 93, 123001 (2004).
  • (50) C. Chin, R. Grimm, P. Julienne, and E. Tiesinga, Feshbach resonances in ultracold gases, Rev. Mod. Phys. 82, 1225 (2010).

Supplemental Material:
Super-exponential scrambling of Out-of-time-ordered correlators

I S1. Details about the mathematical equivalence between the nonlinear Schrödinger system and the GKR model

The periodically modulated nonlinear Schrödinger system reads

H=p22+g|ψ(θ,t)|2jδ(tj).{\rm H}=\frac{p^{2}}{2}+g|\psi(\theta,t)|^{2}\sum_{j}\delta(t-j)\;. (S1)

For a symmetric initial state (ψ(θ,0)=ψ(θ,0)\psi(\theta,0)=\psi(-\theta,0)), since the kicking evolution operator UK=exp[ig|ψ(θ,t)|2/eff]U_{K}=\exp\left[-i{g|\psi(\theta,t)|^{2}}/{\hbar_{\mathrm{eff}}}\right] is wavefunction-dependent, the quantum state will preserve the symmetry in the duration of evolution, i.e., ψ(θ,t)=ψ(θ,t)\psi(\theta,t)=\psi(-\theta,t). The wavefunction can be expanded as ψ(θ,t)=n=+ψn(t)einθ/2π\psi(\theta,t)=\sum_{n=-\infty}^{+\infty}\psi_{n}(t)e^{in\theta}/\sqrt{2\pi}, therefore one can obain

|ψ(θ,t)|2=2n=+Yn(t)einθ=2Y0+4n=1+Yn(t)cos(nθ),\displaystyle\begin{split}|\psi(\theta,t)|^{2}&=2\sum_{n=-\infty}^{+\infty}Y_{n}(t)e^{in\theta}\\ &=2Y_{0}+4\sum_{n=1}^{+\infty}Y_{n}(t)\cos(n\theta),\end{split} (S2)

where

Yn(t)=14πm=+ψm(t)ψm+n(t).Y_{n}(t)=\frac{1}{4\pi}\sum_{m=-\infty}^{+\infty}\psi_{m}^{*}(t)\psi_{m+n}(t). (S3)

The expression Eq. (S3) describes the correlation of the quantum state in the momentum space. Besides, one can get Yn(t)=Yn(t)Y_{n}(t)=Y_{-n}(t) Zhao16 ; Zhao19 . By plugging Eq. (S2) into Eq. (S1), we have

H=p22+n=1+Kn(t)cos(nθ)j(tj).\displaystyle\textrm{H}=\frac{p^{2}}{2}+\sum_{n=1}^{+\infty}K_{n}(t)\cos(n\theta)\sum_{j}(t-j)\;. (S4)

where the kicking strength Kn(t)=4gYn(t)K_{n}(t)=4gY_{n}(t). Since the term with Y0=1/4πY_{0}=1/4\pi only contributes a global phase in the evolution, which has no physical effects, we can drop it safely.

We numerically investigate the phase space of classical trajectories. Our results show that, for a specific value of gg, the classical phase space exhibits regular diffusion of trajectories for short time evolution (e.g., t=3t=3 in Fig. S1(a)), the coexistence of both the regular diffusion and the chaotic diffusion for intermediate time interval (e.g., t=5t=5 in Fig. S1(b)), and the full chaotic diffusion after long enough time evolution (e.g., t=15t=15 in Fig. S1(c)). This clearly demonstrates the regular-to-chaotic transition of the classical dynamics, which stems from the time-dependent increase of the GKR model’s kick strength. The regular behavior leads to the early-time deviation of the Lyapunov exponent from the theoretical prediction in Eq.(18) of the main text, whose validity is guaranteed under the strong chaotic condition (see Fig.1(a) in main text).

Refer to caption
Figure S1: Phase space portrait of the classical GKR model for an ensemble of N=104N=10^{4} trajectories with t=3t=3 (a), 55 (b) and 1515 (c). The parameters are g=1.5g=1.5 and eff=0.6\hbar_{\mathrm{eff}}=0.6.

II S2. Quantum and classical OTOCs for short time evolution

For systems described by Schrödinger equation, the quantum OTOCs are consistent with its classical counterpart, i.e., C(t)=eff2Ccl(t)C(t)=\hbar_{\mathrm{eff}}^{2}C_{cl}(t) in the semiclassical limit (eff1\hbar_{\mathrm{eff}}\ll 1Rozenbaum17 ; Mata2018 . The comparison between C(t)C(t) and eff2Ccl(t)\hbar_{\mathrm{eff}}^{2}C_{cl}(t) for our system is shown in Fig. S2. One can see that, during short time interval, the quantum OTOC is larger than its classical counterpart, which is in sharp contrast to that of the Schrödinger system Zhao16 ; Zhao19 . Unfortunately, since the wavepackets spread in the super-exponential way, the long-time evolution faces the severer computation limit on system size. Our theoretical prediction of quantum OTOCs, i.e., C(t)exp[αγt+βln(gπeff)2t+ηγt2]C(t)\propto\exp\left[\alpha\gamma t+\beta\ln\left(\frac{g}{\pi\hbar_{\mathrm{eff}}}\right)^{2}t+\eta\gamma t^{2}\right] (see Eq.(3) in the main text), yields C(t)exp(ηγt2)C(t)\propto\exp\left(\eta\gamma t^{2}\right) after long time evolution, which is consistent with the time dependence of classical OTOCs Ccl(t)exp(γt2)C_{cl}(t)\propto\exp\left(\gamma t^{2}\right) (see Eq.(9) in the main text) regardless of the factor η\eta.

Refer to caption
Figure S2: (color online). (on log-linear scale) Time dependence of the C(t)C(t) (squares) and eff2Ccl(t)\hbar_{\mathrm{eff}}^{2}C_{cl}(t) (circles) with eff=0.001\hbar_{\mathrm{eff}}=0.001 and g=0.01g=0.01 . Dashed lines (in red) indicate our theoretical predictions. The width of Gaussian wavepacket is σ=10\sigma=10.

III S3. Details about the quantum evolution

Since the exact form of the temporal modulation function is the periodical-delta kicks, to get the Floquet operator, one can use the standard method of the time integral. The evolution of a quantum state from t=nt=n to t=n+1t=n+1 can be separated into two steps: i). the free evolution from t=n+t=n^{+} to t=(n+1)t=(n+1)^{-}, where the superscripts ‘+’ (‘-’) indicate the time immediately after (before) the nn-th kick, i.e., |ψ[(n+1)]=Uf|ψ(n+)|\psi[(n+1)^{-}]\rangle=U_{f}|\psi(n^{+})\rangle; and ii). the kick evolution during the infinitely small time interval from t=(n+1)t=(n+1)^{-} to t=(n+1)+t=(n+1)^{+}, i.e., |ψ[(n+1)+]=UK|ψ[(n+1)]|\psi[(n+1)^{+}]\rangle=U_{K}|\psi[(n+1)^{-}]\rangle. The integral for the free evolution yields the Floquet operator Uf=exp(ip2/2eff)U_{f}=\exp\left({-ip^{2}}/{2\hbar_{\mathrm{eff}}}\right). For the integral of delta kick, it is straightforward to get the Floquet operator UK=exp{ig|ψ[θ,(n+1)]|2/eff}U_{K}=\exp\left\{{-ig|\psi[\theta,(n+1)^{-}]|^{2}}/{\hbar_{\mathrm{eff}}}\right\}. Then, one can get the one-period evolution operator

U=UfUK=exp(ip22eff)exp[ig|ψ(θ,t)|2eff].U=U_{f}U_{K}=\exp\left(\frac{-ip^{2}}{2\hbar_{\mathrm{eff}}}\right)\exp\left[\frac{-ig|\psi(\theta,t)|^{2}}{\hbar_{\mathrm{eff}}}\right]\;. (S5)

Our main result is the analytical prediction of the Super-EG scrambling of the quantum OTOCs

C(t)exp[αγt+βln(gπeff)2t+ηγt2],C(t)\propto\exp\left[\alpha\gamma t+\beta\ln\left(\frac{g}{\pi\hbar_{\mathrm{eff}}}\right)^{2}t+\eta\gamma t^{2}\right]\;, (S6)

where

γ=ln[1+(g𝒩0πeff)2],\gamma=\ln\left[1+\left(\frac{g\mathcal{N}_{0}}{\pi\hbar_{\mathrm{eff}}}\right)^{2}\right]\;, (S7)

α\alpha, β\beta and η\eta are prefactors and 𝒩0\mathcal{N}_{0} is the normalization constant of the initial state (usually 𝒩0=1\mathcal{N}_{0}=1). We cannot analytically get the values of α\alpha, β\beta, η\eta and the proportionality constant. To fit the numerical results in Fig. S3(a), we select a set of (α,β,η)(\alpha,\beta,\eta) and a set of proportionality constants, with the aim of making the curve of the analytical formula in Eq. (S6) and the numerical results a good match. Same method is taken for the red curves in Fig.2 in the main text. Therefore, the values of α\alpha, β\beta, η\eta and the proportionality constant are dependent on the value of gg.

Refer to caption
Figure S3: (color online). Time dependence of the C1C_{1} (a), C2C_{2} (b) and Re[C3]\text{Re}[C_{3}] (c) with eff=0.6\hbar_{\mathrm{eff}}=0.6 for g=1.3g=1.3 (squares), 1.51.5 (circles), 2.02.0 (triangles) and 3.03.0 (diamonds). Red solid lines indicate our theoretical prediction in Eq. (S6). The width of Gaussian wavepacket is σ=1.0\sigma=1.0.

IV S4. Scaling of the OTOCs C(t)=[θ(t),p]2C(t)=-\langle[\theta(t),p]^{2}\rangle

In this section, we show the scaling of the OTOCs C(t)=[θ(t),p]2C(t)=-\langle[\theta(t),p]^{2}\rangle with the dimension of the system. Theoretically, the law of time-dependent takes the form

C(t)=μ(t)Nexp(νγt),C(t)=\mu(t)N\exp(\nu\gamma t)\;, (S8)

where μ\mu is a time-dependent coefficient, NN is number of basis, ν\nu is a constant, and the growth rate γ\gamma has the expression

γ=ln{1+[g𝒩~(t)πeff]2}\gamma=\ln\left\{1+\left[\frac{g\tilde{\mathcal{N}}(t)}{\pi\hbar_{\mathrm{eff}}}\right]^{2}\right\}\; (S9)

with 𝒩~(t)=ψ(t)|θ2|ψ(t)\tilde{\mathcal{N}}(t)=\langle\psi(t)|\theta^{2}|\psi(t)\rangle. Equation (S8) means that, at a specific time tt, the C(t)C(t) scales linearly with the dimension of the system, hence it approaches to infinity with the increase of NN. The divergence of C(t)C(t) demonstrates that there is indeed no bound on the growth of OTOCs in our system. To verify our theoretical prediction, we numerically investigate the C(t)C(t) for different gg and NN. In numerical simulations, the initial state is a Gaussian wavepacket ψ(θ,0)=(σ/π)1/4exp(σθ2/2)\psi(\theta,0)=(\sigma/\pi)^{1/4}\exp(-\sigma\theta^{2}/2). Figure (S4)(a) shows that, at a specific time tt, the C(t)C(t) linearly increase with NN, which is in good agreement with our theoretical prediction in Eq. (S8).

We proceed to show the details in the numerical calculation of the OTOCs with the general definition C(t)=[A(t),B]2C(t)=-\langle[A(t),B]^{2}\rangle. In our present work, we have considered two cases with (A=p,B=p)(A=p,B=p) and (A=θ,B=p)(A=\theta,B=p). Numerical procedures of calculating these two different OTOCs are the same. As shown in the main text, the OTOCs can be decomposed as

C(t)=C1(t)+C2(t)2Re[C3(t)],C(t)=C_{1}(t)+C_{2}(t)-2\text{Re}[C_{3}(t)]\;, (S10)

where we define the terms in the right of the above equation as

C1(t)\displaystyle C_{1}(t) =ψR(0)|B2|ψR(0),\displaystyle=\left\langle\psi_{R}(0)\left|B^{2}\right|\psi_{R}(0)\right\rangle\;, (S11)
C2(t)\displaystyle C_{2}(t) =φR(0)|φR(0),\displaystyle=\left\langle\varphi_{R}(0)|\varphi_{R}(0)\right\rangle\;, (S12)
C3(t)\displaystyle C_{3}(t) =ψR(0)|B|φR(0),\displaystyle=\left\langle\psi_{R}(0)|B|\varphi_{R}(0)\right\rangle\;, (S13)

with |ψR(0)=U(t)AU^(t)|ψ(0)|\psi_{R}(0)\rangle={U}^{\dagger}(t)A\hat{U}(t)|\psi(0)\rangle and |φR(0)=U(t)AU(t)B|ψ(0)|\varphi_{R}(0)\rangle={U}^{\dagger}(t)A{U}(t)B|\psi(0)\rangle. There are five steps to calculate both C1(t)C_{1}(t) and C2(t)C_{2}(t) at a specific time t=tt=t^{*}:

  • (i)

    select an initial state |ψ(0)|\psi(0)\rangle,

  • (ii)

    forward evolution 0t0\rightarrow t^{*} yields |ψ(t)=U(t,0)|ψ(0)|\psi(t^{*})\rangle=U(t^{*},0)|\psi(0)\rangle,

  • (iii)

    at the time tt^{*}, exert the operator AA on the state |ψ(t)|\psi(t^{*})\rangle and get a new state |ψ~(t)=A|ψ(t)|\tilde{\psi}(t^{*})\rangle=A|\psi(t^{*})\rangle,

  • (iv)

    backward evolution t0t^{*}\rightarrow 0 yields |ψR(0)=U(t,0)|ψ~(t)|\psi_{R}(0)\rangle=U^{\dagger}(t^{*},0)|\tilde{\psi}(t^{*})\rangle,

  • (v)

    calculate the expectation value C1(t)=ψR(0)|B2|ψR(0)C_{1}(t^{*})=\left\langle\psi_{R}(0)\left|B^{2}\right|\psi_{R}(0)\right\rangle.

To get C2(t)C_{2}(t^{*}), at the step (i), one should exert the operator BB on the state |ψ(0)|\psi(0)\rangle, i.e., |φ(0)=B|ψ(0)|\varphi(0)\rangle=B|\psi(0)\rangle. The steps (ii)-(iv) are similar to that of calculating C1(t)C_{1}(t^{*}). At the step (v), one can obtain the term C2(t)C_{2}(t^{*}) (see Eq. (S12)) by calculating the inner product of the state |φR(0)|\varphi_{R}(0)\rangle.

At the end of the time reversal, one can use the two state |ψR(0)|\psi_{R}(0)\rangle and |φR(0)|\varphi_{R}(0)\rangle to get the third term of the OTOCs, i.e., C3(t)=ψR(0)|B|φR(0)C_{3}(t^{*})=\left\langle\psi_{R}(0)|B|\varphi_{R}(0)\right\rangle (see Eq. (S13)). The schematic diagram for the evolution progress of the quantum state to calculate the C1(t)C_{1}(t^{*}) with (A=θ,B=p)(A=\theta,B=p) is shown in Table 1.

Forward: tt^{*} steps θ\theta action Backward: tt^{*} steps
|ψ(0)|\psi(0)\rangle \rightarrow |ψ(1)|\psi(1)\rangle \rightarrow \cdots |ψ(t)|\psi(t^{*})\rangle |ψ~(t)=θ|ψ(t)|\tilde{\psi}(t^{*})\rangle=\theta|\psi(t^{*})\rangle |ψ~(t)|\tilde{\psi}(t^{*})\rangle \cdots \rightarrow |ψR(1)|\psi_{R}(1)\rangle \rightarrow |ψR(0)|\psi_{R}(0)\rangle
E(t)=ψ(t)|p2|ψ(t)E(t)=\langle\psi(t)|p^{2}|\psi(t)\rangle E(0)E(0)   \rightarrow  E(1)E(1) \rightarrow \cdots  E(t)E(t^{*}) E~(t)=ψ~(t)|p2|ψ~(t)\tilde{E}(t^{*})=\langle\tilde{\psi}(t^{*})|p^{2}|\tilde{\psi}(t^{*})\rangle E~(t)\tilde{E}(t^{*})  \cdots \rightarrowER(1)E_{R}(1)    \rightarrowER(0)E_{R}(0)
𝒩(t)=ψ(t)|ψ(t)\mathcal{N}(t)=\langle\psi(t)|\psi(t)\rangle 𝒩(0)\mathcal{N}(0)   ==  𝒩(1)\mathcal{N}(1) ==   \cdots𝒩(t)=1\mathcal{N}(t^{*})=1 𝒩~(t)=ψ(t)|θ2|ψ(t)\tilde{\mathcal{N}}(t^{*})=\langle\psi(t^{*})|\theta^{2}|\psi(t^{*})\rangle 𝒩~(t)\tilde{\mathcal{N}}(t^{*})\cdots==   𝒩R(1){\cal{N}}_{R}(1)    ==  𝒩R(0){\cal{N}}_{R}(0)
Table 1: Schematic diagram of time evolution to calculate the term C1(t)=ψR(0)|p2|ψR(0)C_{1}(t^{*})=\langle\psi_{R}(0)|p^{2}|\psi_{R}(0)\rangle.

Numerically, we find that both C2(t)C_{2}(t) and Re[C3(t)]\textrm{Re}[C_{3}(t)] are negligibly small compared with C(t)C(t) (see Fig. S4), which demonstrates that C(t)C1(t)C(t)\approx C_{1}(t). Then, we proceed to theoretically evaluate the time evolution of C1(t)C_{1}(t). Equation. (S11) demonstrates that the C1(t)C_{1}(t) is just the expectation value of the square of momentum for the state at the end of time reversal, i.e., C1(t)=ψR(0)|p2|ψR(0)C_{1}(t)=\left\langle\psi_{R}(0)\left|p^{2}\right|\psi_{R}(0)\right\rangle. We numerically investigate the evolution of the mean square of momentum for a specific time interval t=tt=t^{*}. From Fig. S5(a), one can see that the values of p2\langle p^{2}\rangle remain almost the same during the forward evolution, i.e., ttt\leq t^{*}. Moreover, the time evolution of the mean energy is independent on the number of basis, which demonstrates the convergence of numerical results. After the action of the θ\theta operator, i.e., |ψ~(t)=θ|ψ(t)|\tilde{\psi}(t^{*})\rangle=\theta|\psi(t^{*})\rangle, however, the value of p2\langle p^{2}\rangle has a clear jump, namely p~2(t)p2(t)\langle\tilde{p}^{2}(t^{*})\rangle\gg\langle p^{2}(t^{*})\rangle. Interestingly, during the time reversal t>tt>t^{*}, the mean energy exponentially increases and has shown clear distinction for different NN (see Fig. S5(a)). Our theoretical prediction of the exponential increase of mean energy is

p2(t)=p~2exp(νγt),\langle p^{2}(t^{\prime})\rangle=\langle\tilde{p}^{2}\rangle\exp(\nu\gamma t^{\prime})\;, (S14)

where p~2\langle\tilde{p}^{2}\rangle is the mean energy at the time t=tt=t^{*}, ν\nu is a constant, and the expression of the growth rate is in Eq. (S9). The detailed derivations of Eq. (S14) are provided in the following. Before that, we will first show the details for the derivation of C(t)C(t)’s scaling in Eq. (S8).

Note that, the time reversal starts from the time t=tt=t^{*} (see Table. 1 for calculating C1(t)C_{1}(t^{*})), therefore t=0t^{\prime}=0 means the time t=tt=t^{*}. Since the time reversal lasts tt^{*} steps (see Table. 1), the maximum value of tt^{\prime} equals to tt^{*}. Accordingly, the mean energy at the end of time reversal is

p2(t=0)R=p~2(t)exp(νγt),\langle p^{2}(t=0)\rangle_{R}=\langle\tilde{p}^{2}(t^{*})\rangle\exp(\nu\gamma t^{*})\;, (S15)

which is just the C1(t)C_{1}(t) at the time t=tt=t^{*} (see Eq. (S11)). Then, we get the time-dependence of the OTOCs,

C(t)p~2(t)exp(νγt).C(t)\approx\langle\tilde{p}^{2}(t)\rangle\exp(\nu\gamma t)\;. (S16)

The scaling of C(t)C(t) with NN results from the dependence of the mean energy p~2\langle\tilde{p}^{2}\rangle on NN. Interestingly, we find that, at a specific time, e.g., t=tt=t^{*}, the p~2\langle\tilde{p}^{2}\rangle increases with NN by the way of

p~2(t)=μ(t)N,\langle\tilde{p}^{2}(t^{*})\rangle=\mu(t^{*})N, (S17)

where μ\mu is a coefficient (see Fig. S5(b) for t=7t^{*}=7). To reveal the mechanism of such linear scaling, we numerically investigate the distribution of wavepacket at the time t=tt=t^{*} in the momentum space. Our results show that, after the action of θ\theta operator on a quantum state, i.e., |ψ~=θ|ψ|\tilde{\psi}\rangle=\theta|\psi\rangle, the momentum distribution exhibits the power-law decay (see Figs. S5(c) and (d))

|ψ~(p)|2|p|2,|\tilde{\psi}(p)|^{2}\propto|p|^{-2}\;, (S18)

for which the mean energy is p~2=0pNp2|ψ~(p)|2𝑑ppN\langle\tilde{p}^{2}\rangle=\int_{0}^{p_{N}}p^{2}|\tilde{\psi}(p)|^{-2}dp\propto p_{N}. Then, a rigorous expression of the mean energy is p~2=μN\langle\tilde{p}^{2}\rangle=\mu N since pN=Neffp_{N}=N\hbar_{\mathrm{eff}}, which is confirmed by our numerical results in Fig. S5(b).

Refer to caption
Figure S4: (color online). (a) Dependence of C1(t)C_{1}(t) (a), C2(t)C_{2}(t) (b), and Re[C3(t)]\textrm{Re}[C_{3}(t)] (b) on NN with t=7t=7 for g=0.4g=0.4 (squares), 0.50.5 (circles) and 0.60.6 (triangles). In (a): red lines denote our theoretical prediction in Eq. (S8). In (b): solid (empty) symbols denote C2(t)C_{2}(t) (Re[C3(t)]\textrm{Re}[C_{3}(t)]). The parameter is eff=0.6\hbar_{\mathrm{eff}}=0.6. The width of Gaussian wavepacket is σ=10\sigma=10.

The following can explain why the power-law decayed wavefuction appears. In the momentum space, the quantum state |ψ~|\tilde{\psi}\rangle is expressed as

n|ψ~=mn|θ|mm|ψ,\langle n|\tilde{\psi}\rangle=\sum_{m}\langle n|\theta|m\rangle\langle m|\psi\rangle\;, (S19)

where the matrix element of n|θ|m\langle n|\theta|m\rangle takes the form

n|θ|m={πform=n,1i(mn)formn.\langle n|\theta|m\rangle=\begin{cases}\pi&\text{for}\quad m=n\;,\\ \frac{1}{i(m-n)}&\text{for}\quad m\neq n.\end{cases} (S20)

The power-law decay of n|θ|m\langle n|\theta|m\rangle is a kind of long-range interaction, which effectively leads to the transition among momentum sites. Even if the quantum state |ψ|\psi\rangle is exponentially-localized in the momentum space |ψ(p)|2exp(|p|/ξ)|\psi(p)|^{2}\propto\exp(-|p|/\xi) (see Fig. S5(c)), the action of θ\theta operator can induce the power-law decay of |ψ~(p)|2|\tilde{\psi}(p)|^{2}.

We further show the details for the derivation of the exponential growth of the mean energy in Eq. (S14) during the time reversal. Our previous works have it that, for strong enough nonlinear interaction, the mean energy of our system obeys the iterative equation

p2(t+1)p2(t)[1+g2(t)],\langle p^{2}(t^{\prime}+1)\rangle\approx\langle p^{2}(t^{\prime})\rangle\left[1+g^{2}(t^{\prime})\right]\;, (S21)

where g(t)=g|ψ(t)|2g(t^{\prime})=g|\psi(t^{\prime})|^{2}. Note that, during the time reversal, the norm of the quantum state is a constant, i.e., ψR(t)|ψR(t)=ψ~(t)|ψ~(t)=𝒩~(t)\langle\psi_{R}(t^{\prime})|\psi_{R}(t^{\prime})\rangle=\langle\tilde{\psi}(t^{*})|\tilde{\psi}(t^{*})\rangle=\tilde{\mathcal{N}}(t^{*}). Thus, a rough estimation of the nonlinear interaction strength is g(t)g𝒩~(t)g(t^{\prime})\simeq g\tilde{\mathcal{N}}(t^{*}) with 𝒩~(t)=ψ(t)|θ2|ψ(t)\tilde{\mathcal{N}}(t^{*})=\langle\psi(t^{*})|\theta^{2}|\psi(t^{*})\rangle (see Table. 1 for 𝒩~(t)\tilde{\mathcal{N}}(t^{*})). From Eq. (S21), it is straightforward to get the law of the time-dependence of the mean energy

p2(t)=p2(t=0)exp(νγt)=p~2exp(νγt)\langle p^{2}(t^{\prime})\rangle=\langle p^{2}(t^{\prime}=0)\rangle\exp(\nu\gamma t^{\prime})=\langle\tilde{p}^{2}\rangle\exp(\nu\gamma t^{\prime})\; (S22)

with

γ=ln{1+[g𝒩~(t)πeff]2},\gamma=\ln\left\{1+\left[\frac{g\tilde{\mathcal{N}}(t^{*})}{\pi\hbar_{\mathrm{eff}}}\right]^{2}\right\}\;, (S23)

which is confirmed by our numerical results in Fig. S5(a).

Refer to caption
Figure S5: (color online). (a) Time dependence of p2\langle p^{2}\rangle with t=7t=7 and g=0.6g=0.6 for N=215N=2^{15} (squares) and 2182^{18} (circles). Red solid lines are the theoretical prediction in Eq. (S14). Green line is the auxiliary line. (b) The p~2(t=7)\langle\tilde{p}^{2}(t=7)\rangle versus NN with g=0.4g=0.4 (squares), 0.50.5 (circles) and 0.60.6 (triangles). Red solid lines are the theoretical prediction in Eq. (S17). (c) Momentum distribution at the time t=7t=7. Squares (in black) and circles (in blue) indicate the state |ψ|\psi\rangle and |ψ~=θ|ψ|\tilde{\psi}\rangle=\theta|\psi\rangle, respectively. Red and green lines denote the power-law decay |ψ~(p)|2|p|2|\tilde{\psi}(p)|^{2}\propto|p|^{-2} and exponential decay |ψ(p)|2exp(|p|/ξ)|\psi(p)|^{2}\propto\exp(-|p|/\xi) with ξ2.0\xi\approx 2.0. (d) Momentum distribution for the state |ψ~|\tilde{\psi}\rangle with N=215N=2^{15} (circles) and 2182^{18} (squares). We only plot the |ψ~(p)|2|\tilde{\psi}(p)|^{2} of positive pp, since it is an even function. Red solid line indicates the power-law decay ψ~(p)|p|2\tilde{\psi}(p)\propto|p|^{-2}. Other parameters are the same as in Fig. S4.

References

  • (1) W. Zhao, J. Gong, W. Wang, G. Casati, J. Liu, and L. Fu, Exponential wave-packet spreading via self-interaction time modulation, Phys. Rev. A94, 053631 (2016).
  • (2) W. Zhao, J. Wang, and W. Wang, J. Phys. A: Math. Theor. Quantum-classical correspondence in a nonlinear Gross-Pitaevski system, 52 305101 (2019).
  • (3) E. B. Rozenbaum, S. Ganeshan, and V. Galitski, Lyapunov Exponent and Out-of-Time-Ordered Correlator¡¯s Growth Rate in a Chaotic System, Phys. Rev. Lett. 118 086801 (2017).
  • (4) 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).