Quantum Duffing Oscillator Dynamics

• Quantum Duffing oscillator is a driven nonlinear quantum system characterized by a quartic anharmonic potential or effective Kerr nonlinearity, modeling superconducting and Josephson circuits.
• It employs Floquet theory and perturbative approaches to reveal quasienergy spectra, multiphoton resonances, and metastable states under periodic driving and dissipation.
• Studies of this oscillator offer insights into quantum-classical transitions, experimental design for superconducting devices, and the limits of the Duffing approximation in realistic systems.

Searching arXiv for recent and foundational papers on the quantum Duffing oscillator to ground the article in published work. arxiv_search(query="quantum Duffing oscillator Josephson driven dissipative Floquet metastability", max_results=10, sort_by="relevance") The quantum Duffing oscillator is a driven nonlinear quantum oscillator whose dynamics are governed by a quartic anharmonicity or, in rotating-frame form, by an effective Kerr term. In the literature summarized here, it appears both as a direct quantization of a periodically driven quartic oscillator and as a reduced model for superconducting nonlinear resonators and Josephson circuits. Its defining problems are the structure of quasienergy spectra under periodic drive, the role of dissipation in producing unique steady states or long-lived metastable manifolds, and the extent to which classical Duffing notions such as bistability, hysteresis, and dynamical switching survive in the quantum regime (Vierheilig et al., 2010, Chen et al., 2022, Gosner et al., 2018).

1. Canonical definitions and model classes

A common laboratory-frame definition is the driven quartic oscillator

$$\hat{H}_{\rm DO}(t)=\frac{\hat{P}_y^2}{2M}+\frac{M\Omega^2}{2}\hat{y}^2+\frac{\alpha}{4}\hat{y}^4+\hat{y}F\cos(\omega_{ex} t),$$

with mass $M$, bare frequency $\Omega$, quartic nonlinearity $\alpha$, and monochromatic drive of amplitude $F$ and frequency $\omega_{ex}$. In the hard-nonlinearity case $\alpha>0$, the static potential is monostable, while the driven response acquires the characteristic nonlinear resonance structure of the Duffing problem (Vierheilig et al., 2010).

In superconducting implementations, the same driven nonlinear dynamics is often written directly in a rotating frame as a driven-dissipative Kerr oscillator,

$$H/\hbar = \Delta a^{\dagger}a + U a^{\dagger}a^{\dagger}aa + \xi \left(a+a^{\dagger}\right),$$

supplemented by single-photon loss $(\gamma/2)\mathcal D[a]$. Here $a$ and $M$0 act on the resonator mode, $M$1 is the detuning, $M$2 the Kerr nonlinearity, $M$3 the drive strength, and $M$4 the total energy dissipation rate. This formulation is the operative quantum Duffing model in superconducting nonlinear resonators (Chen et al., 2022).

A second, widely used dimensionless form emphasizes a double-well geometry and the semiclassical control parameter: $M$5 Its quantum Hamiltonian contains the quartic Duffing term, the inverted quadratic term generating the double well, the periodic drive, and an explicit damping-related symmetrized $M$6 contribution chosen so that expectation values recover the damped classical limit as $M$7. This scaling makes $M$8 the effective size or quantumness parameter: small $M$9 is the classical limit, while $\Omega$0 is more deeply quantum (Maris et al., 2020).

These model classes are not interchangeable in all regimes. The quartic or Kerr form is the quantum Duffing oscillator proper, but some physical systems only reduce to it after controlled approximations. That distinction is central in Josephson implementations.

2. Rotating-frame and Floquet structure

Because the Duffing oscillator is periodically driven, Floquet theory provides a natural description. The Schrödinger solutions are written as

$\Omega$1

with periodic Floquet modes $\Omega$2 and quasienergies $\Omega$3. In Sambe space, the quasienergies are defined only modulo $\Omega$4, so the relevant physics is organized by avoided crossings and multiphoton resonances rather than by the static energy spectrum alone (Vierheilig et al., 2010).

Two complementary perturbative Floquet schemes have been developed for the deep quantum regime. One treats the driven linear oscillator exactly and the quartic nonlinearity perturbatively; it is exact in drive amplitude and valid away from resonance. The other uses Van Vleck quasi-degenerate perturbation theory in Floquet space, perturbative in both drive and nonlinearity, and is designed for the vicinity of an $\Omega$5-photon resonance. Their overlap provides a controlled description across the driving-frequency axis (Vierheilig et al., 2010).

In the undriven nonlinear oscillator, the quartic term removes harmonic level equidistance: $\Omega$6 Consequently, the resonance condition becomes level dependent. For a one-photon resonance,

$\Omega$7

Near such resonances, the stationary dissipative response can exhibit antiresonant dips rather than resonant peaks; at weak damping the antiresonance is asymmetric, while stronger damping restores an ordinary broadened resonance centered at the shifted frequency (Vierheilig et al., 2010).

This Floquet viewpoint changes the interpretation of the driven quantum Duffing oscillator. In the deep quantum regime, the central objects are quasienergy doublets, multiphoton avoided crossings, and bath-controlled stationary Floquet populations, not the classical folded response curve.

3. Duffing as a controlled limit of Josephson dynamics

A current-driven Josephson junction provides a physically important case in which Duffing behavior is emergent rather than fundamental. The circuit degree of freedom is the superconducting phase difference $\Omega$8, conjugate to the charge $\Omega$9, with laboratory-frame Hamiltonian

$\alpha$0

Introducing $\alpha$1, moving to a frame rotating at $\alpha$2, and applying the rotating-wave approximation yields an exact-in-amplitude rotating-frame quasienergy operator whose nonlinearity is governed by a Bessel function,

$\alpha$3

The Duffing Hamiltonian appears only after a further small-amplitude expansion of this Bessel structure (Gosner et al., 2018).

The controlled Duffing regime requires four simultaneous conditions: small rotating-frame amplitude $\alpha$4, weak enough drive such that $\alpha$5 when $\alpha$6, weak quantum fluctuations $\alpha$7, and near-resonant driving with $\alpha$8 and $\alpha$9. When these conditions hold, the junction is quantitatively described by a driven Kerr oscillator. When they fail, especially for $F$0 and $F$1, the full cosine nonlinearity becomes essential (Gosner et al., 2018).

The qualitative failure of the Duffing approximation is sharp. In the standard driven softening regime, the Duffing oscillator supports at most bistability, whereas the full Josephson cosine can generate multiple stable dynamical states because its nonlinearity is periodic rather than merely quartic. In the cited study, parameters $F$2, $F$3, $F$4, and $F$5 produce multiple peaks in the steady-state Fock distribution around $F$6 in the full Josephson model, while the corresponding Duffing approximation gives only a single peak near $F$7. The same framework shows that quantum fluctuations renormalize the resonance frequency and Josephson energy, and that fluctuations around higher-amplitude fixed points depend nontrivially on Bessel functions $F$8 and $F$9, which are beyond-Duffing effects (Gosner et al., 2018).

A persistent misconception is therefore that a strongly driven Josephson junction is generically a Duffing oscillator. The cited analysis shows instead that it is a more general quantum nonlinear Josephson oscillator whose low-amplitude sector is Duffing-like.

4. Dissipation, metastability, and dissipative phase transitions

For finite-size driven-dissipative quantum Duffing oscillators, the asymptotic state is unique under general conditions. In Liouvillian language, if $\omega_{ex}$0 and the eigenvalues are ordered so that $\omega_{ex}$1, then $\omega_{ex}$2 and $\omega_{ex}$3. The smallest nonzero decay rate $\omega_{ex}$4 is the Liouvillian gap and sets the final relaxation time $\omega_{ex}$5. When $\omega_{ex}$6, the dynamics first relaxes rapidly onto a metastable manifold and only later converges to the unique steady state (Chen et al., 2022).

This spectral picture resolves the classical language of bistability and hysteresis. In a superconducting Duffing oscillator realized by a coplanar-waveguide resonator interrupted by a weakly asymmetric SQUID, the two classically identified “steady states” were measured to be long-lived metastable states rather than true stationary states. At $\omega_{ex}$7, delay-dependent measurements reproduced familiar hysteresis for $\omega_{ex}$8, yet increasing the waiting time made the branches collapse onto a single steady state. Around a critical drive $\omega_{ex}$9, the loop area in the complex reflection coefficient decayed as $\alpha>0$0 with two rates, $\alpha>0$1 and $\alpha>0$2, reflecting fast approach to the metastable manifold and slow relaxation to the true steady state. The Liouvillian gap dropped from approximately $\alpha>0$3 away from the transition to $\alpha>0$4 near criticality, which is the finite-size signature of critical slowing down (Chen et al., 2022).

Under the scaling

$\alpha>0$5

with $\alpha>0$6 at $\alpha>0$7 and $\alpha>0$8, the same device emulates the mean-field description of an $\alpha>0$9-site driven-dissipative Bose-Hubbard lattice. The critical rescaled drive $$H/\hbar = \Delta a^{\dagger}a + U a^{\dagger}a^{\dagger}aa + \xi \left(a+a^{\dagger}\right),$$0 is essentially independent of $$H/\hbar = \Delta a^{\dagger}a + U a^{\dagger}a^{\dagger}aa + \xi \left(a+a^{\dagger}\right),$$1, while the transition step sharpens as $$H/\hbar = \Delta a^{\dagger}a + U a^{\dagger}a^{\dagger}aa + \xi \left(a+a^{\dagger}\right),$$2 increases. Quantum state tomography identified the two sides of the transition as a coherent-state phase and a squeezed-state phase, with a two-component mixed Wigner function across $$H/\hbar = \Delta a^{\dagger}a + U a^{\dagger}a^{\dagger}aa + \xi \left(a+a^{\dagger}\right),$$3 (Chen et al., 2022).

A separate nonequilibrium result concerns global stability in the bistable Duffing regime. For the rotating-frame Lindblad description of the Duffing oscillator, the long-time stationary state can be reduced to an effective two-state model with transition rates $$H/\hbar = \Delta a^{\dagger}a + U a^{\dagger}a^{\dagger}aa + \xi \left(a+a^{\dagger}\right),$$4 and $$H/\hbar = \Delta a^{\dagger}a + U a^{\dagger}a^{\dagger}aa + \xi \left(a+a^{\dagger}\right),$$5 between low- and high-amplitude branches. The crossover point is where $$H/\hbar = \Delta a^{\dagger}a + U a^{\dagger}a^{\dagger}aa + \xi \left(a+a^{\dagger}\right),$$6, but that point is non-unique: the limits $$H/\hbar = \Delta a^{\dagger}a + U a^{\dagger}a^{\dagger}aa + \xi \left(a+a^{\dagger}\right),$$7, $$H/\hbar = \Delta a^{\dagger}a + U a^{\dagger}a^{\dagger}aa + \xi \left(a+a^{\dagger}\right),$$8, and $$H/\hbar = \Delta a^{\dagger}a + U a^{\dagger}a^{\dagger}aa + \xi \left(a+a^{\dagger}\right),$$9 do not commute. For $(\gamma/2)\mathcal D[a]$0, the cited study finds $(\gamma/2)\mathcal D[a]$1 on the classical thermal route and $(\gamma/2)\mathcal D[a]$2 on the quantum zero-temperature route (Yeh et al., 2012). This indicates a specifically nonequilibrium notion of tunneling and global stability, distinct from the static double-well criterion of equal potential minima.

5. Quantum dynamics near attractors

Near a stable classical state of the driven Duffing oscillator, attractor-centered descriptions are more accurate than bare linearization. Starting from

$(\gamma/2)\mathcal D[a]$3

one can displace the oscillator by $(\gamma/2)\mathcal D[a]$4 to the attractor center and then squeeze by $(\gamma/2)\mathcal D[a]$5 to eliminate local anomalous quadratic mixing. The resulting renormalized master equation contains an effective thermal occupation

$(\gamma/2)\mathcal D[a]$6

where $(\gamma/2)\mathcal D[a]$7 and $(\gamma/2)\mathcal D[a]$8 are the squeezing coefficients. Even at $(\gamma/2)\mathcal D[a]$9, one has $a$0, so the local oscillator is thermally populated in the squeezed frame by quantum fluctuations associated with the attractor (Feng et al., 24 Feb 2025).

This framework shows precisely what standard linearization misses. Around the high-amplitude stable state, the orbit amplitude scales as $a$1, so cubic and quartic fluctuation terms remain parametrically important even when the effective Planck constant $a$2 is small. A refined double perturbation theory for

$a$3

yields renormalized local level spacings, corrected eigenvectors, and level-dependent orbital displacements. The local quasienergy spectrum is not exactly harmonic near the attractor bottom: higher-order corrections make $a$4 level dependent and decrease the spacing toward the saddle point. Likewise, the average orbital center $a$5 becomes level dependent, whereas harmonic linearization would predict $a$6 for all local levels (Feng et al., 24 Feb 2025).

The same treatment gives a balance equation for local populations,

$a$7

valid in the weak-damping regime $a$8. In harmonic approximation the adjacent-level ratio is Bose-like,

$a$9

but higher-order corrections make the effective occupation number level dependent. Strong damping slightly renormalizes the effective level spacing extracted from the emission spectrum, and pure dephasing $M$00 raises the effective local temperature through

$M$01

Thus dephasing is not only coherence loss; after attractor-dependent displacement and squeezing it also acts as local heating (Feng et al., 24 Feb 2025).

6. Semiclassical correspondence, escape, and related realizations

In semiclassical open-system studies, quantum-classical differences in the Duffing oscillator are not controlled simply by how chaotic the classical motion is. Using Quantum State Diffusion, a Gaussian-moment semiclassical reduction, and noise-added classical surrogates, one study found that classically regular orbits can show the largest deviations from classical behavior, and that the growth of quantum-classical differences is not determined by the classical Lyapunov exponent. In an intermediate dissipation range, a parametrically invariant meta-attractor emerges at

$M$02

where attractors across different $M$03 values become nearly indistinguishable in temporal and spatial complexity. Below that scale, noise-added classical models deviate strongly from semiclassical dynamics (Maris et al., 2020).

A complementary semiclassical theory treats switching in the weak-dissipation regime as quasienergy diffusion over a dynamical barrier in the rotating frame rather than as coordinate-space escape. For the driven Duffing Hamiltonian

$M$04

the rotating-frame quasienergy landscape has two stable extrema and an unstable saddle for $M$05. In this regime, finite reflection at the phase-space barrier suppresses switching, but bath-induced quantum fluctuations generated by the rotating-frame $M$06- and $M$07-couplings enhance it, and the cited analysis finds that the environmental enhancement overcompensates the reflection-induced reduction. Near the bifurcation threshold, the barrier softens so strongly that the local dynamics becomes effectively overdamped, and the quantum energy scale associated with friction can exceed the thermal scale (Verso et al., 2010).

Experimental and structural generalizations show how the quantum Duffing problem connects to other nonlinear platforms. A single $M$08 ion in a linear RF Paul trap realizes a driven nonlinear mechanical oscillator whose steady-state motion is described by

$M$09

The experiment demonstrated hardening Duffing response, hysteresis, jump phenomena, phase discontinuities, and tunable linear and nonlinear damping through Doppler cooling, but it did not yet realize the deep quantum Duffing regime of nonclassical quasienergy states or quantum tunneling between attractors (Akerman et al., 2010). In a different direction, two weakly coupled classical Duffing oscillators in 1:1 resonance reduce after averaging to boson Josephson junction equations for an energy partition index $M$10 and relative phase $M$11, with effective Hamiltonian

$M$12

Within that reduced model, local modes correspond to macroscopic quantum self-trapping states, giving a semiclassical bridge between nonlinear vibration localization and Josephson dynamics (Pilipchuk, 2012).

Taken together, these results define the quantum Duffing oscillator as a family of closely related driven-dissipative quantum nonlinear problems rather than a single universal approximation. In low-amplitude weakly anharmonic regimes it is well described by a Kerr Hamiltonian; near resonances it is governed by Floquet quasienergy structure; in driven superconducting circuits it exhibits metastability and finite-size precursors of dissipative phase transitions; near attractors it supports squeezed, effectively thermal local modes with renormalized spectra; and in more general Josephson settings its Duffing description fails once the full periodic nonlinearity of the cosine becomes dynamically relevant.