Quantum Averaging Theory for Multi-Timescale Driven Quantum Systems

TL;DR

Quantum Averaging Theory (QAT) merges the unitarity-preserving Magnus expansion with multi-timescale averaging to model driven quantum systems across far-detuned and near-resonant regimes. By factorizing the unitary evolution into a fast propagator and a slow, effective Hamiltonian, QAT captures fast micromotion via a dynamical phase while governing long-time dynamics with a renormalized effective Hamiltonian $\hat{H}_{I,\mathrm{eff}}$. The framework provides a constructive algorithm to obtain higher-order corrections, establishes error bounds, and demonstrates rapid convergence to exact solutions in prototypical quantum-optical problems, including far-detuned Rabi and near-resonant Raman processes. It generalizes Floquet-based methods to almost-periodic and multi-timescale drives, enabling more accurate analytic control and design of quantum gates and simulations, with future extensions to open systems and many-body settings.

Abstract

We present a multi-timescale Quantum Averaging Theory (QAT), a unitarity-preserving generalized Floquet framework for analytically modeling periodically and almost-periodically driven quantum systems across multiple timescales. By integrating the Magnus expansion with the method of averaging on multiple scales, QAT captures the effects of both far-detuned and near-resonant interactions on system dynamics. The framework yields an effective Hamiltonian description while retaining fast oscillatory effects within a separate dynamical phase operator, ensuring accuracy across a wide range of driving regimes. We demonstrate the rapid convergence of QAT results toward exact numerical solutions in both detuning regimes for touchstone problems in quantum information science.

Figures (5)

• Figure 1: Far-Detuned Rabi Oscillations. The system is initialized in the ground state $\ket{g}$ and driven with a weak-field amplitude $\lambda = 0.5 \times 10^{-5}$. For a valid high-frequency QAT expansion, the detuning must satisfy $|\Lambda_\Delta| > \lambda$, with improved convergence as $\lambda/|\Lambda_\Delta| \ll 1$. To illustrate this, we choose a moderately large ratio $\lambda/|\Lambda_\Delta| = 0.4$. Second-order (blue) and fourth-order (red) QAT results are shown to match the envelope of the exact solution, with the fourth-order expansion exhibiting strong agreement over multiple cycles.
• Figure 2: Time-Evolution of the Particle Wavefunction in a Driven $1\mathrm{D}$ Double-Well Potential. A quantum particle of unit mass begins in a minimum-uncertainty coherent state centered at $x = -1$ (arb. units) and initially at rest. A weak, effective double-well potential arises due to the high-frequency modulation of the trap. While the classical trajectory follows the gradient toward the nearest potential minimum, the anharmonicity induces quantum spreading and rapid loss of coherence in the wavefunction. To illustrate these dynamics, we choose: natural oscillator frequency $\omega_0 = 0.2\omega$, DC confinement $\omega_{dc} = 1.5\omega_\mathrm{eff}$ with $\omega_\mathrm{eff} = \omega_0^2 / 2\omega$, and anharmonic coupling strength $g = 1.75\cdot10^{-2}$. At leading order, the QAT approximation moderately agrees with the full numerical solution; including the next non-zero correction significantly improves accuracy in both amplitude and phase estimation.
• Figure 3: Phase Space of Particle in a Driven $1\mathrm{D}$ Double-Well Potential. Using the same parameters as in Fig. 2, we plot the phase space trajectory of the particle. The high-frequency modulation gives rise to an effective time-averaged trapping potential, which generates a ponderomotive force. Under these conditions, the particle exhibits motion centered around the effective secular trajectory but exhibits pronounced excess micromotion due to the relatively large ratio $\omega_0/\omega_\mathrm{rf} = 0.2$.
• Figure 4: Near-Resonant Rabi Oscillations with Counter-Rotating Terms. The qubit is initialized in the superposition state $\ket{\psi_0} = \frac{1}{\sqrt{2}}(\ket{g} + \ket{e})$ and weakly driven with amplitude $\lambda = 0.05$. A near-resonant detuning $\Lambda_\Delta \simeq 1.14\lambda$ is chosen to eliminate the second-order light shift due to counter-rotating terms, thereby isolating higher-order contributions to the dynamics. Both the non-perturbative (red) and perturbative (blue) second-order QAT dynamics closely match the full numerical solution, even beyond the formal regime of guaranteed validity.
• Figure 5: Near-Resonant Raman Transition. The system is initialized in the ground state $\ket{\psi_0} = \ket{1}$. The pump and probe beams are driven with moderate coupling strengths $\Omega_1 = \Delta_1 / 4$ and $\Omega_2 = \Delta_1 / 5$, respectively, with detuning from the intermediate state $\ket{r}$ set by $\Delta_1 = 2\pi \times 10MHz$. The frequency difference $\Delta = \Delta_1 - \Delta_2$ is chosen to cancel second-order differential light shifts, ensuring $\bra{1}\hat{H}_\mathrm{eff}^{[2]}\ket{1} = \bra{2} \hat{H}_\mathrm{eff}^{[2]} \ket{2}$. Second-order QAT results (blue) agree with the numerical solution over half a Rabi cycle but deviate at longer times. In contrast, the next non-zero correction (red) shows rapid convergence and remains accurate over multiple cycles.