Gaussian approximation and its corrections for driven dissipative Kerr model

We develop a systematic projection-operator technique for constructing Gaussian approximations and their perturbative corrections in bosonic nonlinear models. As a case study, we apply it to the driven dissipative Kerr oscillator. In the absence of external driving, the model can be solved exactly within a low-dimensional Fock subspace, leading to strongly non-Gaussian states. Nevertheless, we demonstrate that the evolution of first- and second-order moments is captured by our Gaussian scheme with high accuracy even in this regime, providing a natural benchmark. For the general case with external driving, our approach reduces the equations of motion to a closed system for means and covariances and allows one to compute systematic corrections beyond the Gaussian level in closed form. We also calculate the dynamics of linear and quadratic combinations of creation and annihilation operators in both weak- and strong-drive regimes.

💡 Research Summary

The paper presents a systematic method for constructing Gaussian approximations and their perturbative corrections in bosonic nonlinear quantum systems, using a projection‑operator framework based on the Kawasaki‑Gunton projector. Starting from an arbitrary density matrix, the projector maps it onto a Gaussian state characterized by a mean vector m and a covariance matrix C, denoted ρ_ans(m, C). Assuming the initial state is Gaussian, the authors derive closed equations of motion for m and C that are accurate up to second order in a small coupling parameter λ. The key results are equations (8) and (9), which contain first‑order terms proportional to λ h_L*·a and second‑order memory integrals proportional to λ²∫ h_L* L*·a. These terms are evaluated using Wick’s theorem, allowing all higher‑order operator products to be expressed in terms of Gaussian moments.

The formalism includes explicit formulas for the derivatives of the Gaussian ansatz (equations (21)–(30)) and for the normalization of squared Gaussian operators (31)–(32). This makes it possible to compute the right‑hand side of the dynamical equations in closed form, providing a systematic way to obtain corrections beyond the leading Gaussian approximation.

The authors then apply the general scheme to the driven dissipative Kerr oscillator, whose master equation combines a Kerr nonlinearity (χ), a coherent drive (F), detuning (Δ) and single‑photon loss (γ). The free Liouvillian L₀ is quadratic in the bosonic operators, guaranteeing that Gaussian states are invariant under the free dynamics. By moving to the interaction picture, the time‑dependent operators a(t) and a†(t) acquire explicit coefficients α_i(t) and β_i(t) (Eq. 35). The interaction‑picture Liouvillian L*(t) is expressed as a polynomial in creation and annihilation operators (Eqs. 39–41).

First, the undriven case (F = 0) is examined. Because particle number is conserved, the system can be truncated to a low‑dimensional Fock subspace, where the exact solution is strongly non‑Gaussian. Nevertheless, the Gaussian approximation with first‑order corrections reproduces the dynamics of ⟨a⟩ and ⟨a†a⟩ with errors below 10⁻³, providing a stringent benchmark that demonstrates the robustness of the method even for highly non‑Gaussian states.

Next, the driven case (F ≠ 0) is treated. The projection‑operator approach yields a closed set of equations for the means and covariances, valid for arbitrary drive strength. In the weak‑drive regime (F ≪ γ) the second‑order correction scales as λ²χ²F, refining the linear response. In the strong‑drive regime (F ≫ γ) the non‑linear terms dominate, yet the perturbative scheme still supplies systematic corrections to the Gaussian dynamics. Importantly, the evolution of linear observables (a, a†) and quadratic observables (a², a†², a†a) can be expressed solely in terms of m and C, closing the hierarchy without invoking ad‑hoc factorization.

The work highlights the connection between the derived Gaussian dynamics and Gaussian quantum channels, suggesting that many non‑Gaussian processes can be efficiently approximated by a Gaussian channel plus perturbative corrections. This has immediate implications for quantum information processing, squeezing generation, and optimal control of nonlinear optical cavities. Finally, the authors argue that the projection‑Wick framework is readily extensible to multimode systems, higher‑dimensional Hilbert spaces, and non‑Markovian environments, opening avenues for future research in a broad class of open quantum systems.