Show Summary Details
More options …

# Zeitschrift für Naturforschung A

### A Journal of Physical Sciences

Editor-in-Chief: Holthaus, Martin

Editorial Board: Fetecau, Corina / Kiefer, Claus / Röpke, Gerd / Steeb, Willi-Hans

12 Issues per year

IMPACT FACTOR 2016: 1.432

Cite Score 2016: 1.35

SCImago Journal Rank (SJR) 2016: 0.368
Source Normalized Impact per Paper (SNIP) 2016: 0.594

Online
ISSN
1865-7109
See all formats and pricing
More options …

# Magnus Expansion Approach to Parametric Oscillator Systems in a Thermal Bath

Beilei Zhu
• Corresponding author
• Zentrum für Optische Quantentechnologien and Institut für Laserphysik, Universität Hamburg, 22761 Hamburg, Germany
• Email
• Other articles by this author:
• De Gruyter OnlineGoogle Scholar
/ Tobias Rexin
• Zentrum für Optische Quantentechnologien and Institut für Laserphysik, Universität Hamburg, 22761 Hamburg, Germany
• Other articles by this author:
• De Gruyter OnlineGoogle Scholar
/ Ludwig Mathey
• Zentrum für Optische Quantentechnologien and Institut für Laserphysik, Universität Hamburg, 22761 Hamburg, Germany
• The Hamburg Centre for Ultrafast Imaging, Luruper Chaussee 149, Hamburg 22761, Germany
• Other articles by this author:
• De Gruyter OnlineGoogle Scholar
Published Online: 2016-09-21 | DOI: https://doi.org/10.1515/zna-2016-0135

## Abstract

We develop a Magnus formalism for periodically driven systems which provides an expansion both in the driving term and in the inverse driving frequency, applicable to isolated and dissipative systems. We derive explicit formulas for a driving term with a cosine dependence on time, up to fourth order. We apply these to the steady state of a classical parametric oscillator coupled to a thermal bath, which we solve numerically for comparison. Beyond dynamical stabilisation at second order, we find that the higher orders further renormalise the oscillator frequency, and additionally create a weakly renormalised effective temperature. The renormalised oscillator frequency is quantitatively accurate almost up to the parametric instability, as we confirm numerically. Additionally, a cut-off dependent term is generated, which indicates the break down of the hierarchy of time scales of the system, as a precursor to the instability. Finally, we apply this formalism to a parametrically driven chain, as an example for the control of the dispersion of a many-body system.

## 1 Introduction

The study of periodically driven systems has experienced renewed interest in recent times. Both in solid state and ultra cold atom systems, strong periodic driving has been used to control nonequilibrium states. In ultra-cold atom systems, periodic lattice driving has been used to realise an effective, synthetic gauge field, see [1]. In solid-state systems, pump–probe experiments [2], on high-Tc superconductors and on graphene have been performed, see [3], [4], [5]. Theoretical studies on light-induced superconductivity were reported in [6], [7], [8], [9], [10], [11].

Remarkably, in both cases, external high-frequency driving is used to control the low-frequency behaviour of each system. The quintessential example for this phenomenon is the Kapitza effect [12]. In the case of the effective synthetic field in an ultra-cold atom system, this process is explicitly described by an approximate, effective low-energy Hamiltonian, which, in contrast to the original, nondriven Hamiltonian, has a synthetic field. In the case of a driven high-Tc superconductor, the near-resonant driving of an optical phonon mode results in a modified response in the low-frequency optical conductivity. Both of these observations exemplify the development of a new field of emergence in driven many-body systems.

In this article, we give a systematic expansion of the emergent low-energy description of a driven system. This discussion applies and extends the Magnus formalism, as discussed in [13], [14], [15], [16]. Our formalism provides a systematic expansion both in the driving amplitude and in the inverse driving frequency and is applicable to closed and open classical systems, to closed quantum systems. We derive explicit, general expressions for the leading terms beyond second order. As a key example, we apply this formalism to a parametrically driven oscillator, coupled to a thermal bath [17], [18], and determine the properties of its steady state. An insightful discussion of parametric oscillators was given in [19], [20], as well as in [21]. We then apply our results to a chain of parametrically driven oscillators. This provides insight in how the dispersion of a system can be controlled via parametric driving.

This article is organised as follows: In Section 2, we describe the dissipatively coupled, parametrically driven oscillator, and give a discussion of its properties using elementary ansatz functions. In Section 3, we develop the Magnus expansion in full generality first and then apply it to the parametric oscillator in Section 4. In Section 5, we discuss the control of the dispersion of a parametrically driven chain of oscillators, and in Section 6, we conclude.

## 2 Parametric Oscillator

As the key example to which we apply the Magnus expansion, we consider a parametrically driven oscillator, described by the Hamiltonian

$H=H0+Hdr$(1)

with

$H0=p22m+mω022x2$(2)

$Hdr=mω022Acos(ωmt)x2.$(3)

p and x are the momentum and spatial coordinate of the oscillator, m the mass, and ω0 the bare oscillator frequency. A is the amplitude of the parametric driving term and ωm the driving frequency.

We assume that this oscillator is coupled to a thermal bath of temperature T, via a dissipative term. The resulting equations of motion are of the Langevin form:

$dxdt=pm$(4)

$dpdt=−mω02(1+Acos(ωmt))x−γp+ξ.$(5)

γ is the damping rate, and ξ describes white noise, with the correlation function 〈ξ(t1)ξ(t2)〉=2γkBTmδ(t1t2), where kB is the Boltzmann constant. In thermal equilibrium, in the absence of driving, the system is described by the canonical distribution ρ0(x, p)=exp(−βH0(x, p))/Z, with β=1/(kBT). Z is the partition function, which normalises this probability distribution. For this distribution, the variances of x and p are $〈{x}^{2}〉={x}_{T}^{2}$ and $〈{p}^{2}〉={p}_{T}^{2},$ with ${p}_{T}=\sqrt{m{k}_{B}T}$ and ${x}_{T}=\sqrt{\frac{{k}_{B}T}{m{\omega }_{0}^{2}}}.$ Furthermore, we have 〈x〉=〈p〉=〈xp〉=0. We note that for a classical oscillator, xT and pT can be used to rescale x and p. With this choice, the temperature does not appear in any of the remaining quantities and simply provides an energy scale for the system. For a quantum mechanical oscillator, this rescaling cannot be performed. Here, an additional regime appears in which quantum fluctuations dominate, for kBT=ћω0. A full discussion of the driven, dissipative quantum mechanical oscillator will be given elsewhere. The analysis presented here addresses isolated quantum systems, in addition to dissipative classical systems.

In Figure 1, we depict the time averaged variance 〈x(t)2〉, of the steady state of the driven system, as a function of A and ωm/ω0. Here, and in the examples throughout this article, we choose γ/ω0=0.1. The most striking feature of this plot is the parametric resonance that appears near ωm≈2ω0, for small A. This feature is then power broadened for increasing A. In this regime, the magnitude of 〈x(t)2〉 is increased by orders of magnitude, compared to the equilibrium value. In addition to this strong heating effect, there is a regime for large ωm/ω0, and large amplitude, for which a reduction of 〈x(t)2〉 is observed. Here, the parametric driving leads to a dynamic stabilisation of the fluctuations of x. It is this counterintuitive and quintessential example of reducing fluctuations via high-frequency driving that we study systematically in this article.

Figure 1:

We depict the time averaged magnitude of $〈{x}^{2}\left(t\right)〉/{x}_{T}^{2}$ in the steady state as a function of driving frequency and driving amplitude. Panels (A) and (B) depict the same data on different scales. The system displays a power broadened instability emerging from ~2ω0, and a dynamical stabilisation for large ωm and A. In panel (A), we show the comparison to (8), in panel (B) to (7).

In Figure 2, we show the same quantity in the steady state as a function of A, for a fixed value of the driving frequency, ωm/ω0=20, to give a clearer insight into the quantitative behaviour. The magnitude of these fluctuations is visibly reduced with increasing driving amplitude. However, eventually this trend of decreasing fluctuations is rapidly reverted, resulting in a steep increase of the fluctuations. As visible from Figure 1, this steep increase is due to the power-broadened parametric instability. The onset of this instability determines the location of the minimal amount of fluctuations that can be achieved with this type of driving. It is therefore imperative to understand the origin of this steep increase of the fluctuations and provide a systematic approach to determine its behaviour.

Figure 2:

We depict the time averaged expectation value of 〈x2〉, in units of ${x}_{T}^{2},$ as a function of the driving amplitude A, for the driving frequency ωm/ω0=20, and for γ/ω0=0.1. We compare the numerically obtained result to the prediction in (10). The dashed, vertical line corresponds to (8).

In Figure 3, we show a histogram of the distribution ρdr in the steady state, in comparison to the equilibrium distribution ρ0; we show ρdrρ0. The distribution ρdr is generated from trajectories of the Langevin equation, which have been low-frequency filtered via ${x}_{c}\left(t\right)=\int \text{d}s{G}_{\sigma }\left(s-t\right)x\left(t\right),$ and similarly for pc(t), derived from p(t). Gσ(s) is a normalised Gaussian, with a time scale σ, for which we choose σ=1/ω0. In Figure 3, furthermore, we choose A=10 and ωm/ω0=20. We observe that the width of the distribution along x-direction is reduced, due to the dynamical stabilisation that is described in the following sections. Along the p-direction, the distribution is only weakly affected. We emphasise that for a quantitative comparison of the driven state to the effective, low-frequency predictions, the exclusion of the high-frequency contributions in the numerics is essential. We elaborate on this point in Appendix A.

Figure 3:

Distribution in phase space of the driven system in the steady state. The trajectories of the time evolution have been smoothed out on a time scale of σ=1/ω0. We use ωm/ω0=20, A=10, and γ/ω0=0.1. The binning size is Δx/xTp/pT=0.01. The reduction of the width of the distribution in the x-direction is clearly visible.

## 2.1 Elementary Approach

Before we develop the renormalisation of the oscillator due to the periodic driving systematically in the following section, we give estimates of its behaviour using various ansatz functions.

We start out by giving an estimate for the instability regime, and note that a more detailed discussion is given in Appendix B. We consider the equation of motion of the isolated system, $\stackrel{¨}{x}+{\omega }_{0}^{2}\left(1+A\mathrm{cos}\left({\omega }_{m}t\right)\right)x=0.$ We consider the ansatz x(t)=a0 cos(ωefft)+a1 cos((ωmωeff)t), where a0 and a1 are constant coefficients. We solve for the effective frequency ωeff, which gives

$ωeff=ωm−ωm2+4ω02−2ω0A2ω02+4ωm22$(6)

The instability regime is reached when the expression under the outer square root becomes negative. This occurs at

$ωm,prω0≈2A+4,$(7)

which simplifies to

$ωm,prω0≈2A,$(8)

for large A. This provides an estimate for the instability regime for large driving amplitudes and frequencies, which we show in Figure 1, and which gives good agreement.

To give an estimate for the renormalisation of ωeff, we extend this ansatz to include not only the frequencies ωeff and ωmωeff, but also the next three contributing terms, corresponding to the frequencies ωm+ωeff, 2ωmωeff and 2ωm+ωeff. This ansatz is explicitly written in (B5). This ansatz results in (B6) for the effective frequency. We solve this equation iteratively in the amplitude A, which gives the expansion

$ωeff2≈ω02+A2ω042(ωm2−4ω02)+25A4ω0832ωm6$(9)

At second order in A and at second order in the inverse driving frequency, this is

$ωeff2ω02≈1+A2ω022ωm2$(10)

This approximation for the effective frequency is shown in Figure 2. We note that the fourth-order term in (9) is positive. This is indeed confirmed further down by the systematic Magnus expansion. However, the Magnus expansion determines the correct prefactor, which differs from the one found here.

## 3 Magnus Expansion

We now turn to the Magnus expansion of the system. This expansion provides a time-independent approximation of the low-frequency sector of the system, derived from the original, time-dependent Hamiltonian that describes all frequencies. After deriving general expressions for the Magnus terms beyond second order, we ask the question if and how the key features of the parametric oscillator, the dynamical stabilisation and the instability regime, can be captured within this approach. We note that these features, as they were described in the previous section, might suggest that such an approach might not be possible in a consistent fashion for the fourth-order correction. This is due to the following two observations. We observed, as shown in (9), that the fourth-order correction has a positive prefactor, which results in an addition stabilisation of the oscillator. This term would be derived from a term in an effective Hamiltonian that is of the form $~{A}^{4}/{\omega }_{m}^{6},$ with a positive prefactor. On the other hand, if the instability of (8) is derived from an effective Hamiltonian, it also needs to be derived from a term of the form $~{A}^{4}/{\omega }_{m}^{6},$ but now with a negative prefactor.

Interestingly, as we discuss below, the Magnus expansion provides two types of terms at fourth order. One of them is cut-off independent and features a positive prefactor. The resulting renormalisation due to this term is in agreement with the numerically obtained result. The other term is cut-off dependent. It indicates that the hierarchy of time scales that is required for the Magnus expansion breaks down. We interpret this as a precursor of the instability regime, and indeed find that the scaling for this regime, as shown in (8), is predicted correctly.

## 3.1 Kramers Equation

To apply the Magnus expansion, we formulate the time evolution of the system (4 and 5), as a time evolution of the phase space distribution ρ(x, p, t). This is given by the Kramers equation

$∂tρ=L(t)ρ.$(11)

Here, L(t)=L0+Ldr(t) with

$L0ρ=−v∂xρ+ω02x∂vρ+γ(∂v(vρ)+kBTm∂vvρ)$(12)

and

$Ldr=Ldr,0cos(ωmt)$(13)

with

$Ldr,0=Aω02x∂v.$(14)

We refer to (11) as the Kramers equation to distinguish it from the Fokker–Planck equation, which we reserve for the over-damped limit, in accordance with the terminology of [22].

## 3.2 General Expansion

We now derive the expansion of the low-energy description in full generality. We consider a general, dynamical system that is described by the same equation of motion

$∂tρ=L(t)ρ$(15)

as before, without the assumption of the specific form of the equation of motion, as in the previous section. The parametrically driven oscillator will serve as the example to which we apply our results further down. The system under consideration can be either a closed or an open classical system or a closed quantum system. For a closed quantum system, we interpret the operator L(t) as a Hamiltonian, divided by , i.e. L(t)=H(t)/(). For an open system, we also include dissipative terms, as in (11). We again assume that L(t) has the form

$L(t)=L0+Ldr(t),$(16)

where L0 describes the time-independent part of the system and Ldr(t) is the driving term, again of the form

$Ldr(t)=Ldr,0cos(ωmt).$(17)

We perform the Magnus expansion in the interaction picture. In this picture, the order of the Magnus expansion coincides with the order of the driving term. In the case of the parametric oscillator, this is the order of the driving amplitude A. For the interaction picture, we define

$Ldr,i(t, s)=exp(−L0s)Ldr(t)exp(L0s)$(18)

where the standard interaction picture term is Ldr,i(t)=Ldr,i (t, t). Then the equation of motion is

$∂tρi=Ldr,i(t)ρi.$(19)

Its solution is

$ρi(t)=Ttexp(∫t0tdsLdr,i(s))ρi(t0)$(20)

where Tt is the time ordering operator and ρi(t0) the initial state at t0. The Magnus expansion consists of re-expressing this solution in the form $\mathrm{exp}\left({\sum }_{i}{M}_{i}\right),$ where Mi is the Magnus term of i-th order, see [15].

We time average each of these terms over a time interval [t0, t]. The time interval is long compared to the driving period but short compared to the dynamics that is created by H0. For the parametric oscillator, we demand 1/ω0tt0≫1/ωm. The time interval Δtc=tt0 is also the inverse of a frequency cut-off ωc=2πtc, for which we equivalently demand ω0ωcωm. For a general system, the frequency ω0 has to be replaced by a typical frequency that is characteristic for the dynamics of H0.

The second-order Magnus term in the interaction picture is given by

$M2,i=−12∫t0tdt2∫t0t2dt1[Ldr,i(t1), Ldr,i(t2)]$(21)

We transfer this expression back to the Schrödinger picture and project this term on the frequency range below ωc. The resulting effective ${L}_{\text{eff}}^{\left(2\right)}$ is time independent because all the oscillatory contributions oscillate with a frequency above the cut-off frequency. The time evolution that results from this term is of the form $\mathrm{exp}\left({L}_{\text{eff}}^{\left(2\right)}\Delta {t}_{c}\right).$ Therefore, we can simplify (21) by taking the time derivative with respect to t, which reduces the number of integrations. The resulting second-order term is therefore

$Leff(2)=[−12∫t0tdt1[Ldr,i(t1, t˜1), Ldr(t)]]ω < ωc$(22)

with ${\stackrel{˜}{t}}_{1}={t}_{1}-t.$ We expand the expression in (18) to first order:

$Ldr,i(t, s)≈Ldr(t)−s[L0, Ldr(t)]$(23)

We use this first-order expansion, with $s\to {\stackrel{˜}{t}}_{1}$ and tt1, and the time dependence of the driving term (17),

$Leff(2)≈12[[L0, Ldr,0], Ldr,0]×[∫t0tdt1t˜1cos(ωmt)cos(ωmt1)]ω < ωc$(24)

The low-frequency part of the time integral, which refers to frequencies below ωc, is

$[∫t0tdt1t˜1cos(ωmt)cos(ωmt1)]ω < ωc=12ωm2$(25)

Therefore, we obtain

$Leff(2,2)=14ωm2[[L0, Ldr,0], Ldr,0]$(26)

Here, and throughout the article, we use the notation ${L}_{\text{eff}}^{\left(n,m\right)}$ refer to the n-th order of the Magnus expansion and to the m-th order in the inverse driving frequency.

## 3.3 Fourth Order in ${\omega }_{m}^{-1}$

We now derive the next order term in the inverse frequency. We consider the expansion in (18) to third order

$Ldr,i(t, s)≈Ldr(t)−s[L0, Ldr(t)]+s22adL02Ldr(t)−s33!adL03Ldr(t)$(27)

where we introduced the notation of the adjoint derivative ${\text{ad}}_{{L}_{0}}^{n}{L}_{\text{dr}}\left(t\right).$ It is defined via ${\text{ad}}_{{L}_{0}}^{n}{L}_{\text{dr}}\left(t\right)=\left[{L}_{0},{\text{\hspace{0.17em}ad}}_{{L}_{0}}^{n\text{\hspace{0.17em}}-\text{\hspace{0.17em}}1}{L}_{\text{dr}}\left(t\right)\right],$ and ${\text{ad}}_{{L}_{0}}^{0}{L}_{\text{dr}}\left(t\right)={L}_{\text{dr}}\left(t\right).$ The term that is quadratic in s gives no low-frequency contribution, therefore ${L}_{\text{eff}}^{\left(2,3\right)}=0.$ The fourth-order term is

$Leff(2,4)=12[adL03Ldr, Ldr,0]×[∫t0tdt1t˜133!cos(ωmt)cos(ωmt1)]ω < ωc$(28)

We use the integral property

${\left[{\int }_{{t}_{0}}^{t}\text{d}{t}_{1}{\stackrel{˜}{t}}_{1}^{3}\mathrm{cos}\left({\omega }_{m}t\right)\mathrm{cos}\left({\omega }_{m}{t}_{1}\right)\right]}_{\omega \text{\hspace{0.17em}}<\text{\hspace{0.17em}}{\omega }_{c}}=-\frac{3}{{\omega }_{m}^{4}}$

which results in

$Leff(2,4)=−14ωm4[adL03Ldr, Ldr,0]$(29)

Higher order terms of the form ${L}_{\text{eff}}^{\left(2,m\right)}$ can be derived in a similar manner.

## 3.4 Fourth-Order Magnus Expansion

For the fourth-order term in the driving term, we proceed along the same lines as for the quadratic term in the previous sections. The fourth-order term in the interaction picture has the form

$M4,i=−112∫t0tdt4∫t0t4dt3∫t0t3dt2∫t0t2dt1([Ldr,i(t1), [[Ldr,i(t2), Ldr,i(t3)], Ldr,i(t4)]]+[[Ldr,i(t1), [Ldr,i(t2), Ldr,i(t3)]], Ldr,i(t4)]+[[Ldr,i(t1), Ldr,i(t2)], [Ldr,i(t3), Ldr,i(t4)]]+[[Ldr,i(t1), Ldr,i(t3)], [Ldr,i(t2), Ldr,i(t4)]])$(30)

Again, we transform this expression to the Schrödinger picture. We project this term on the low-frequency regime. Interestingly, we find two contributions, as we show below. The first is proportional to Δtc. Therefore, it lends itself to an interpretation as an effective low-energy description. The second term is cubic in Δtc, which means that we can write

$[M4]ω < ωc=Leff(4)Δtc+L˜eff,c(4)Δtc3,$(31)

and we also introduce the definition ${L}_{\text{eff,}c}^{\left(4\right)}={\stackrel{˜}{L}}_{\text{eff,}c}^{\left(4\right)}\Delta {t}_{c}^{2}.$ We again obtain the operators ${L}_{\text{eff}}^{\left(4\right)}$ and ${L}_{\text{eff,}c}^{\left(4\right)}$ by considering the low-frequency sector of the time derivative of M4, i.e.

$Leff(4)+3Leff,c(4)=[−112∫t0tdt3∫t0t3dt2∫t0t2dt1([Ldr,i(t1, t˜1), [[Ldr,i(t2, t˜2), Ldr,i(t3, t˜3)], Ldr(t)]]+[[Ldr,i(t1, t˜1), [Ldr,i(t2, t˜2), Ldr,i(t3, t˜3)]], Ldr(t)]+[[Ldr,i(t1, t˜1), Ldr,i(t2, t˜2)], [Ldr,i(t3, t˜3), Ldr(t)]] +[[Ldr,i(t1, t˜1), Ldr,i(t3, t˜3)], [Ldr,i(t2, t˜2), Ldr(t)]])]ω < ωc$(32)

with ${\stackrel{˜}{t}}_{i}={t}_{i}-t.$ The factor of 3 in front of ${L}_{\text{eff},c}^{\left(4\right)}$ is due to the derivative of (31). We use the expansion of Ldr,i, given in (27). We order the resulting terms according to the combined order of the times ${\stackrel{˜}{t}}_{i},$ i.e. ${\stackrel{˜}{t}}_{1}^{{k}_{1}}{\stackrel{˜}{t}}_{2}^{{k}_{2}}{\stackrel{˜}{t}}_{3}^{{k}_{3}},$ and k=k1+k2+k3. The first- and second-order terms with k=1 and k=2 give no contribution. For the k=3 term, a number of contributions are generated in this expansion. These contain time integrals of the form

$ck1,k2,k3=[cos(ωmt)∫t0tdt3∫t0t3dt2∫t0t2dt1cos(ωmt1)cos(ωmt2)cos(ωmt3)t˜1k1t˜2k2t˜3k3]ω < ωc$(33)

The integrals that are necessary to derive ${L}_{\text{eff}}^{\left(4\right)}+3{L}_{\text{eff},c}^{\left(4\right)}$ are given in Table 1. All the terms that scale as $1/{\omega }_{m}^{6}$ contribute to ${L}_{\text{eff}}^{\left(4\right)}.$ These are written out and simplified in Appendix C. We obtain ${L}_{\text{eff}}^{\left(4\right)}$ to be

Table 1:

The value of the integrals of the form given in (33), which are necessary to evaluate (32) at order k=3.

$Leff(4,6)=112ωm6(3964[Ldr,0, [[adL03Ldr,0, Ldr,0], Ldr,0]]+6164[Ldr,0, [[adL02Ldr,0, adL0Ldr,0], Ldr,0]]+8732[[adL02Ldr,0, [adL0Ldr,0, Ldr,0]], Ldr,0]−332[[adL02Ldr,0, Ldr,0], [adL0Ldr,0, Ldr,0]])$(34)

The term that scales as $\Delta {t}_{c}^{2}/{\omega }_{m}^{4},$ which is due to the c0,2,1 integral, gives $3{L}_{\text{eff,}c}^{\left(4\right)}.$ Therefore, the cut-off-dependent contribution is

$Leff,c(4,6)=Δtc2144ωm4[Ldr,0, [[adL02Ldr,0, adL0Ldr,0], Ldr,0]]$(35)

We emphasise again that for any system that can be written in the form of (15–17), the results given in (26, 29, 34, and 35) apply. They constitute the main conceptual result of this paper.

## 4 Magnus Expansion of the Parametric Oscillator

We now apply our results to the case of the parametric oscillator, introduced above. For the ${L}_{\text{eff}}^{\left(2,2\right)}$ correction, we use (26) and find

$Leff(2,2)=A2ω042ωm2x∂v$(36)

This implies a renormalisation of the oscillator frequency of the form

$ωeff2ω02=1+A2ω022ωm2$(37)

This coincides with the second-order term that was obtained in (10). At the fourth in the inverse driving frequency, we have

$Leff2,4=A2ω044ωm4(2(4ω02−γ2)x∂v+8γ(T/m)∂vv)$(38)

where we applied (29). Interestingly, in addition to a further renormalisation of the oscillator frequency, a renormalisation of the temperature is created:

$ωeff2ω02=1+A2ω022ωm2+A2ω02(4ω02−γ2)2ωm4$(39)

$TeffT=1+2A2ω04ωm4.$(40)

It is generated because the white-noise dissipative term contains fluctuations at all frequencies, in particular at the driving frequency ωm. This results in an additional renormalisation of the low-frequency regime, via time averaging, of the system at this higher order. For the example presented here, the magnitude of the renormalisation is small. However, nonlinear systems will in general create nonlinear effective dissipative terms at this order. Finally, we determine the two terms at order A4. The cut-off independent term is

$Leff(4,6)=10796A4ω08ωm6x∂v$(41)

This term generates an additional renormalisation of the oscillator frequency, resulting in

$ωeff2ω02=1+A2ω022ωm2+A2ω02(4ω02−γ2)2ωm4+107A4ω0696ωm6$(42)

We note that this renormalisation at fourth order in A has a positive prefactor, as in the estimate in (9). However, the systematic Magnus expansion gives the correct magnitude of the prefactor.

In Figure 4, we depict the power spectrum Sp(ω) of the momentum p in steady state, as a function of the driving amplitude A, and for the fixed driving frequency ωm/ω0=20. The power spectrum is defined via

Figure 4:

Power spectrum Sp(ω) as a function of the driving amplitude A, depicted on a logarithmic scale. For the driving frequency, we use ωm/ω0=20. We show the second-order estimate of the effective frequency, ωeff,2, which refers to (37). In addition, we show the fourth-order estimate ωeff,4, based on (42).

$Sp(ω)=〈p(−ω)p(ω)〉$(43)

with $p\left(\omega \right)=\left(1/{\sqrt{T}}_{s}\right)\int \text{d}{t}^{\prime }\mathrm{exp}\left(-i\omega {t}^{\prime }\right)p\left({t}^{\prime }\right),$ where Ts is the sampling interval. At A=0, the power spectrum reduces to that of a harmonic oscillator, with a single peak at ω0. As the driving is turned on, additional peaks appear at m±ω0, where n is an integer describing the Floquet band. We note that these frequencies are approximately the ones that were used in the ansatz functions in Section 2.1 and Appendix B. With increasing driving amplitude, the effective oscillator frequency increases. We compare this increase to the second-order prediction, given in (37), and the fourth-order prediction (42). The fourth-order estimate describes the oscillator frequency well almost up to the instability, which is reached around A≈180, in this example. We emphasise that the orange bar at A≈180 is numerical data. Here, the magnitude of power spectrum increases rapidly by many orders of magnitude.

The cut-off dependent term is

$Leff,c(4,6)=−A4ω08Δtc218ωm4x∂v$(44)

This term competes with the previously discussed terms that stabilise the oscillator. For simplicity, we only consider the dominant term of the effective frequency (36). We relate the time scale Δtc to a frequency cut-off via Δtc=2π/ωc. We assume to be in the strongly renormalised regime, ${\omega }_{\text{eff}}^{2}/{\omega }_{0}^{2}\approx \frac{{A}^{2}{\omega }_{0}^{2}}{2{\omega }_{m}^{2}}.$ Therefore, ${L}_{\text{eff,}c}^{\left(4,6\right)}$ competes with this renormalisation if

$A2ω022ωm2≈2π2A4ω069ωm4ωc2$(45)

This results in the criterium

$ωmωcω0≈A$(46)

If we consider a cut-off frequency chosen as fraction of the driving frequency, and therefore ωc~ωm, we recover

$ωmω0≈A$(47)

which displays the same scaling as in (8). The scaling displayed in (46) can also be motivated by comparing the cut-off frequency ωc to ωeff/ω0~0/ωm. Again, this condition indicates that the originally assumed hierarchy of energy scales is no longer valid. This property of the system derives from the cut-off dependent term ${L}_{\text{eff,}c}^{\left(4,6\right)}.$ While this term in itself cannot be interpreted as a contribution to the effective equation of motion, it can give an insight into the breakdown of the necessary hierarchy of time scales of the system.

## 5 Parametrically Driven Chain

We apply this formalism to the stabilisation of a chain of oscillators via parametric driving. This, and related mechanisms have been considered in the context of light enhanced superconductivity, with the following motivation. If we imagine a complex order parameter field describing fluctuating superconducting order, a key feature of this system is its phase stiffness. In equilibrium, it controls the superconducting stability and the critical current. The phase stiffness in turn is related to how steeply the dispersion of the system increases with increasing momentum. Therefore, one possible explanation of light enhanced superconductivity might entail stabilising and steepening the dispersion of the system.

We here give the simplest, yet generic, case of a one-dimensional chain of oscillators. The system is described by H=H0+Hdr(t), with

$H0=∑i(pi22m+mω022(xi−xi+1)2)$(48)

with i=1, …,N. The driving term is

$Hdr=∑imω022Acos(ωmt)(xi−xi+1)2$(49)

We therefore have a parametrically driven lattice of oscillators. We Fourier transform the system via ${x}_{i}=\frac{1}{\sqrt{N}}{\sum }_{k}\mathrm{exp}\left(ik{r}_{i}\right){x}_{k}$ and ${p}_{i}=\frac{1}{\sqrt{N}}{\sum }_{k}\mathrm{exp}\left(ik{r}_{i}\right){p}_{k},$ and note that $\left[{x}_{{k}_{1}},\text{\hspace{0.17em}}{p}_{{k}_{2}}\right]=i\hslash {\delta }_{{k}_{1},-{k}_{2}}.$ The Langevin equations for the system are

$dxkdt=pkm$(50)

$dpkdt=−mωk,02(1+Acos(ωmt))xk−γpk+ξk$(51)

with $〈{\xi }_{{k}_{1}}\left({t}_{1}\right){\xi }_{{k}_{2}}\left({t}_{2}\right)〉=2\gamma {k}_{B}Tm{\delta }_{{k}_{1},-{k}_{2}}\delta \left({t}_{1}-{t}_{2}\right).$ In real space, this corresponds to 〈ξi(t1)ξ(t2)〉=2γkBTmδijδ(t1t2). The dispersion ωk,0 is

$ωk,0=ω02−2cosk=2ω0|sin(k/2)|$(52)

We observe that (50 and 51) are equivalent to (4 and 5), with the replacement ω0ωk,0. We can therefore apply the results for the single oscillator (42), to each momentum mode and obtain the effective dispersion

$ωk,eff2=ωk,02(1+A2ωk,022ωm2+A2ωk,02(4ωk,02−γ2)2ωm4+107A4ωk,0696ωm6)$(53)

The second-order correction, derived from ${L}_{\text{eff}}^{\left(2,2\right)},$ contains contributions of the form ~cos(2k). This can be seen by substituting ωk,0=2ω0|sin(k/2)|, see (52). This describes coupling to the next-nearest neighbor, induced by the periodic driving, because a next-nearest coupling term of the form ${\sum }_{i}{x}_{i}{x}_{i\text{\hspace{0.17em}}+\text{\hspace{0.17em}}2}$ gives rise to cos(2k) terms in momentum space when Fourier transformed. When substituting (52), in the term that is quadratic in A and quartic in ${\omega }_{m}^{-1},$ we obtain terms up to ~cos(3k), which corresponds to coupling to the third neighbor. Finally, the term quartic in A contains coupling to the fourth nearest neighbor.

In Figure 5, we show the two point correlation Γ(k, ω) in momentum and coordinate space in the steady state for a one-dimensional chain of parametrically driven oscillators. The two point correlation function is defined by

Figure 5:

Two-point correlation function 𝒢(k, ω), depicted on a logarithmic scale. We use ωm/ω0=80 for the driving frequency with driving amplitude A=400. We compare the numerics with the analytical estimates (white lines), respectively, thermal situation (dotted line), second-order estimate (dashed line), and fourth-order estimate (solid line) of the effective frequency, ωk,eff, which refers to (53).

$G(k, ω)=$(54)

with

$X\left(k,\text{\hspace{0.17em}}\omega \right)=\frac{1}{\sqrt{{K}_{s}{T}_{s}}}\int \text{d}{t}^{\prime }\int \text{d}{\text{r}}^{\prime }x\left({r}^{\prime },{t}^{\prime }\right)\mathrm{exp}\left(-ik{r}^{\prime }\right)\mathrm{exp}\left(-i\omega {t}^{\prime }\right)$

where Ts and Ks are sampling time interval and space interval, respectively. Compared with the nondriven situation, the driven dispersion line has a steeper slope, which means that the driving term stiffens the system significantly. We compare the numerics with effective dispersion ωk,eff, (53). It is clearly seen that at higher k modes, the second-order correction deviates from the numerics while the fourth-order correction describes the numerics precisely.

We also observe that the strongest renormalisation of the dispersion occurs at its upper edge. This includes the onset of the parametric instability. We now have the condition

$ωmωk,0≈A$(55)

which is first reached for the maximum of the band. This sets the upper limit for the driving amplitude A that can be used to stabilise the dispersion. However, we note that, depending on the physical system, the range of A might be much more limited. For example, the value of the spring constant between neighboring oscillators might not allow for negative values, meaning that A<1. With this constraint, the magnitude of the renormalisation is small, of the order of ${\omega }_{k,0}^{2}/{\omega }_{m}^{2}.$

## 6 Conclusions

We have developed a systematic Magnus expansion in the driving term and the inverse driving frequency. In this formalism, we have derived explicit expressions for a system with a driving term with cosine time dependence. This system can be either a quantum mechanical system or a classical system including dissipative terms. The main, conceptual formulas are given in (29, 34, and 35), which are the terms beyond the widely discussed lowest order term in (26). At fourth order in the driving term, we find two contributions, one cut-off independent and one cut-off dependent. The cut-off independent term contributes to the effective Kramers or Hamilton operator, whereas the increasing magnitude of the cut-off dependent term indicates the breakdown of the hierarchy of time scales that was originally assumed. We apply this formalism to a parametrically driven oscillator, coupled to a thermal bath, and to a parametric oscillator chain. We obtain the magnitude of stabilisation that can be achieved for these systems and the onset of the instability. We emphasise that our formalism can be applied to a wide range of driven systems, including nonlinear systems and many-body systems. It will be of particular interest to the emerging field of controlling many-body systems via external driving.

## Acknowledgements

We gratefully acknowledge discussions with Andrea Cavalleri, Robert Höppner, and Junichi Okamoto. We acknowledge support from the Deutsche Forschungsgemeinschaft through the SFB 925 and through Project No. MA 5900/1-1, the Hamburg Centre for Ultrafast Imaging, and from the Landesexzellenzinitiative Hamburg, supported by the Joachim Herz Stiftung. B.Z. acknowledges support from the China Scholarship Council, under scholarship No. 2012 0614 0012.

## Appendix A: Frequency Cut-Off

In this section, we discuss the comparison of the predictions of the effective description to the observables extracted from the full system. Because the effective description is a low-frequency description, it is, in general, imperative to apply a frequency cut-off on the observables, for a quantitative comparison. While for some observables depend only weakly on the introduction of this cut-off, in general, the low-pass-filtered observable will differ from the observable that includes all frequencies.

As discussed in Section 2, we have depicted the phase space distribution that is derived from the low-frequency-filtered trajectories (xc(t), pc(t)) in Figure 3. For comparison, we depict the phase space distribution that is derived from original trajectories (x(t), p(t)) that include all frequencies, in Figure 6. As is clearly visible, for this distribution a broadening of the distribution in the p-direction occurs, in contrast to Figure 3.

Figure 6:

Distribution in phase space of the driven system in the steady state, for the same parameters as in Figure 3, but without the low-frequency filtering. For this distribution, an increase in the width in the p-direction is observed, which is due to high-frequency contributions.

To elaborate on this further, we depict the time average of 〈pc(t)2〉 and 〈p(t)2〉 in the steady state, as a function of A, in Figure 7. 〈p(t)2〉 has a strong dependence on A, which is approximately quadratic. 〈pc(t)2〉, however, has only a very weak A dependence, only given the weak temperature renormalisation that was given in (40).

Figure 7:

Time average value of p2(t), shown as the blue line, and ${p}_{\text{c}}^{2}\left(t\right),$ shown as the red line. The additional increase in p2(t) is due to high-frequency contributions. For a quantitative comparison to effective, low-frequency descriptions, the low-frequency filtered observable has to be used.

## Appendix B: Elementary Ansatz

We consider the equation of motion for the isolated system

$x¨+ω02(1+Acos(ωmt))x=0.$(B1)

To estimate the regime in which the instability of the system occurs, we consider the ansatz

$x(t)=a0cos(ωefft)+a1cos((ωm−ωeff)t)$(B2)

where a0 and a1 are constant coefficients. ωeff is the effective oscillation frequency, which we solve for. Substituting this ansatz in the equation of motion, and ignoring further frequencies, this results in the equations $\left({\omega }_{\text{eff}}^{2}-{\omega }_{0}^{2}\right){a}_{0}=A{\omega }_{0}^{2}{a}_{1}/2$ and $\left(\left({\omega }_{m}-{\omega }_{\text{eff}}{\right)}^{2}-{\omega }_{0}^{2}\right){a}_{1}=A{\omega }_{0}^{2}{a}_{0}/2.$ We eliminate a0 and a1 and obtain the equation

$(ωeff2−ω02)((ωm−ωeff)2−ω02)=A2ω044$(B3)

The resulting ωeff is

$ωeff=ωm−ωm2+4ω02−2ω0A2ω02+4ωm22$(B4)

The parametric resonance is reached when the expression under the square root becomes negative. We note that the effective frequency ωeff increases monotonously, with increasing A. The instability occurs when the two frequencies ωeff and ωmωeff equal each other. We confirm this behaviour by calculating the power spectrum of the driven state, which is shown in Figure 4.

To give a more accurate estimate of the renormalisation of ωeff, we consider the following ansatz

$x(t)=a0cos(ωefft)+a1cos((ωm−ωeff)t)+a2cos((ωm+ωeff)t)+a3cos(((ωm−ωeff)t)+a4cos(((ωm+ωeff)t).$(B5)

When we substitute this in the equation of motion, we obtain the following equation for ωeff:

$ωeff2−ω02=−A2ω044(1ω02−(ωm−ωeff)2+A2ω04/4(2ωm−ωeff)2−ω02+1ω02−(ωm+ωeff)2+A2ω04/4(2ωm+ωeff)2−ω02)$(B6)

We solve this equation iteratively in the driving amplitude A, which gives

$ωeff2≈ω02+A2ω042(ωm2−4ω02)+25A4ω0832ωm6.$(B7)

Here, we kept the leading order in the inverse frequency 1/ωm for the fourth-order term, which scales as $1/{\omega }_{m}^{6}.$ We kept all orders in 1/ωm for the term that is second ordering A.

## Appendix C: Fourth-Order Term of the Magnus Expansion

After expanding (32) to the order k=3, evaluating the integrals of the form of (33), and collecting the terms that scale as $1/\text{​}{\omega }_{m}^{6},$ we obtain for ${L}_{\text{eff}}^{\left(4,6\right)}:$

$Leff(4,6)=112ωm6(164564[Ldr,0, [[adL03Ldr,0, Ldr,0], Ldr,0]]+12132[Ldr,0, [[adL02Ldr,0, adL0Ldr,0], Ldr,0]]+12716[Ldr,0, [[adL0Ldr,0, adL02Ldr,0], Ldr,0]]−1698[Ldr,0, [[Ldr,0, adL03Ldr,0], Ldr,0]]+122764[adL0Ldr,0, [[adL02Ldr,0, Ldr,0], Ldr,0]]−121516[adL0Ldr,0, [[Ldr,0, adL02Ldr,0], Ldr,0]]+122164[adL02Ldr,0, [[adL0Ldr,0, Ldr,0], Ldr,0]]−123332[adL02Ldr,0, [[Ldr,0, adL0Ldr,0], Ldr,0]]+164564[[Ldr,0, [adL03Ldr,0, Ldr,0]], Ldr,0]+12132[[Ldr,0, [adL02Ldr,0, adL0Ldr,0]], Ldr,0]+12716[[Ldr,0, [adL0Ldr,0, adL02Ldr,0]], Ldr,0]−1698[[Ldr,0, [Ldr,0, adL03Ldr,0]], Ldr,0]+122764[[adL0Ldr,0, [adL02Ldr,0, Ldr,0]], Ldr,0]−121516[[adL0Ldr,0, [Ldr,0, adL02Ldr,0]], Ldr,0]+122164[[adL02Ldr,0, [adL0Ldr,0, Ldr,0]], Ldr,0]−123332[[adL02Ldr,0, [Ldr,0, adL0Ldr,0]], Ldr,0]−123332[[adL02Ldr,0, Ldr,0], [adL0Ldr,0, Ldr,0]]+12132[[Ldr,0, adL02Ldr,0], [adL0Ldr,0, Ldr,0]]−121516[[adL0Ldr,0, Ldr,0], [adL02Ldr,0, Ldr,0]]+12716[[Ldr,0, adL0Ldr,0, [adL02Ldr,0, Ldr,0]]+122164[[adL02Ldr,0, Ldr,0], [adL0Ldr,0, Ldr,0]]+12716[[Ldr,0, adL02Ldr,0], [adL0Ldr,0, Ldr,0]]+122764[[adL0Ldr,0, Ldr,0], [adL02Ldr,0, Ldr,0]]+12132[[Ldr,0, adL0Ldr,0], [adL02Ldr,0, Ldr,0]])$(C1)

To simplify this expression, we first combine the terms that are related by commutation. In addition, we use that

$[adL0Ldr,0, [[adL02Ldr,0, Ldr,0], Ldr,0]]$(C2)

$+[adL02Ldr,0, [[adL0Ldr,0, Ldr,0], Ldr,0]]$(C3)

$=[[adL0Ldr,0, [adL02Ldr,0, Ldr,0]], Ldr,0]$(C4)

$+[[adL02Ldr,0, [adL0Ldr,0, Ldr,0]], Ldr,0]$(C5)

and

$[adL0Ldr,0, [adL02Ldr,0, Ldr,0]]$(C6)

$=[adL02Ldr,0, [adL0Ldr,0, Ldr,0]]$(C7)

$−[Ldr,0, [adL0Ldr,0, adL02Ldr,0]]$(C8)

With these identities, we simplify the expression to the form given in (34).

## References

• [1]

J. Struck, M. Weinberg, C. Ölschläger, P. Windpassinger, J. Simonet, et al., Nat. Phys. 9, 738 (2013). Google Scholar

• [2]

C. Giannetti, M. Capone, D. Fausti, M. Fabrizio, F. Parmigiani, et al., Adv. Phys. 65, 58 (2016). Google Scholar

• [3]

W. Hu, S. Kaiser, D. Nicoletti, C. R. Hunt, I. Gierz, et al., Nat. Mater. 13, 705 (2014). Google Scholar

• [4]

R. Mankowsky, A. Subedi, M. Först, S. O. Mariager, M. Chollet, et al., Nature 516, 71 (2014). Google Scholar

• [5]

I. Gierz, J. C. Petersen, M. Mitrano, C. Cacho, E. Turcu, et al., Nat. Mater. 12, 1119 (2013). Google Scholar

• [6]

R. Höppner, B. Zhu, T. Rexin, A. Cavalleri, and L. Mathey, Phys. Rev. B 91, 104507 (2015). Google Scholar

• [7]

J.-i. Okamoto, A. Cavalleri, and L. Mathey, arXiv:1606.09276. Web of Science

• [8]

M. A. Sentef, A. F. Kemper, A. Georges, and C. Kollath, Phys. Rev. B 93, 144506 (2016). Google Scholar

• [9]

Z. M. Raines, V. Stanev, and V. M. Galitski, Phys. Rev. B 91, 184506 (2015). Google Scholar

• [10]

S. J. Denny, S. R. Clark, Y. Laplace, A. Cavalleri, and D. Jaksch, Phys. Rev. Lett. 114, 137001 (2015). Google Scholar

• [11]

A. A. Patel and A. Eberlein, Phys. Rev. B 93, 195139 (2016). Google Scholar

• [12]

L. D. Landau and E. M. Lifshitz, Mechanics Vol. 1 (1st ed.), Pergamon Press, Headington Hill Hall, Oxford 1960. Google Scholar

• [13]

W. Magnus, Commun. Pure Appl. Math. 7, 649 (1954). Google Scholar

• [14]

S. Blanes, F. Casas, J. A. Oteo, and J. Ros, Phys. Rep. 470, 151 (2009). Google Scholar

• [15]

W. R. Salzman, J. Chem. Phys. 82, 822 (1985). Google Scholar

• [16]

L. D’Alessio and A. Polkovnikov, Ann. Phys. 333, 19 (2013). Google Scholar

• [17]

C. Zerbe, P. Jung, and P. Hänggi, Phys. Rev. E 49, 3626 (1994). Google Scholar

• [18]

C. Zerbe and P. Hänggi, Phys. Rev. E 52, 1533 (1995). Google Scholar

• [19]

E. I. Butikov, J. Phys. A 35, 6209 (2002). Google Scholar

• [20]

E. I. Butikov, Regular and Chaotic Motions of the Parametrically Forced Pendulum: Theory and Simulations, Springer Berlin, Heidelberg 2002. Google Scholar

• [21]

M. G. Clerc, C. Falcón, C. Fernández-Oto, and E. Tirapegui, Europephys. Lett. 98, 30006 (2012). Google Scholar

• [22]

N. G. van Kampen, Stochastic Processes in Physics and Chemistry, Elsevier, Amsterdam, Netherlands 1992. Google Scholar

## About the article

Accepted: 2016-08-23

Published Online: 2016-09-21

Published in Print: 2016-10-01

Citation Information: Zeitschrift für Naturforschung A, Volume 71, Issue 10, Pages 921–932, ISSN (Online) 1865-7109, ISSN (Print) 0932-0784,

Export Citation

©2016 Walter de Gruyter GmbH, Berlin/Boston.

## Citing Articles

[1]
Jun-ichi Okamoto, Wanzheng Hu, Andrea Cavalleri, and Ludwig Mathey
Physical Review B, 2017, Volume 96, Number 14
[2]
Jun-ichi Okamoto, Andrea Cavalleri, and Ludwig Mathey
Physical Review Letters, 2016, Volume 117, Number 22