On the static effective Lindbladian of the squeezed Kerr oscillator

Jayameenakshi Venkatraman, Xu Xiao, Rodrigo G. Cortiñas, Michel DevoretarXiv 2022 · DOI · arXiv

This paper, authored by researchers at Yale University, provides a critical theoretical missing link in our understanding of decoherence in superconducting quantum circuits—specifically for the "Kerr-cat" qubit.

TL;DR

The authors derive a new mathematical model (an effective Lindbladian) for driven nonlinear oscillators that goes beyond the standard Rotating Wave Approximation (RWA). This model accounts for "hidden" dissipative processes that explain why experimental Kerr-cat qubits decohere orders of magnitude faster than previously predicted, providing a roadmap for better quantum error correction.

What Problem Does This Paper Address?

In the race to build a fault-tolerant quantum computer, "Kerr-cat" qubits are a frontrunner. They use a squeezed Kerr oscillator to create quantum states that are naturally protected against bit-flip errors. However, researchers noticed a frustrating discrepancy: in experiments, these qubits lose their quantum information (decohere) much faster than the standard theoretical models suggest.

The standard model relies on the Rotating Wave Approximation (RWA), which simplifies equations by ignoring very fast-oscillating terms. While usually reliable, the RWA fails to capture how a strong microwave drive (used to create the qubit) interacts with the surrounding environment (the "bath"). This paper addresses this gap by calculating the "beyond-RWA" effects that turn the drive's energy into unwanted noise.

Methods

The authors employ a systematic analytical approach:

  1. Starting Hamiltonian: They begin with a driven nonlinear oscillator (the system) coupled to a bath of harmonic oscillators (the environment).
  2. Schrieffer-Wolff Transformation: To find a "static" description of this rapidly oscillating system, they use a series of unitary transformations. This is akin to moving into a sophisticated "rotating frame" where the physics becomes time-independent.
  3. Perturbative Expansion: They expand the system-bath coupling in powers of $\phi_{zps}$ (the zero-point spread of the phase). While standard RWA only looks at the 0th order, this paper derives terms up to the 2nd order.
  4. Effective Lindbladian: They derive the master equation that describes how the qubit's density matrix evolves over time, identifying new "dissipators" (mathematical terms representing specific types of noise).

Datasets

As a theoretical physics paper, it does not use traditional machine learning datasets. Instead, it validates its model against experimental data from recent superconducting circuit studies, specifically the Kerr-cat qubit implementation by Frattini et al. (2022) [Ref 31 in the paper].

Key Insights and Ideas

  • The "Unruh-like" Heating: The authors show that the microwave drive effectively "heats up" the environment from the qubit's perspective. Even at zero temperature, the drive can convert a drive photon into a qubit excitation and an environment excitation simultaneously.
  • Two-Photon Heating is the Culprit: The most significant discovery is a "two-photon heating" term ($D[\hat{a}^{\dagger 2}]$). This process preserves the parity of the quantum state but destroys the coherence between the "cat" states.
  • Non-Linear Scaling: Crucially, this new two-photon noise scales with the photon number as $|\alpha|^4$, whereas standard single-photon loss scales as $|\alpha|^2$. This explains why the decoherence becomes catastrophic as the "size" of the cat state increases.
  • Kerr-Dependence: At the second order of the expansion, the decoherence rate becomes highly sensitive to the Kerr nonlinearity ($K$). This explains why different experimental setups with varying $K$ values saw vastly different lifetimes.

Key Results

  • Bridging the Gap: The standard RWA model (Order 0) predicted a coherent state lifetime ($T_X$) of roughly 1 second. Experimental results showed closer to 1 millisecond. The authors' new "Order 1" and "Order 2" models match the experimental millisecond-scale data almost perfectly (see Figure 1 and 3).
  • Order of Magnitude Accuracy: The model accounts for a 1,000x difference in decoherence rates that was previously unexplained.
  • Mitigation Strategy: In Appendix A, the authors demonstrate that adding "engineered" two-photon cooling can largely compensate for these beyond-RWA heating effects, offering a practical path to improving qubit lifetimes (Figure 4).

Limitations and Future Work

  • White Noise Assumption: The model assumes the environment is "white" (the noise is the same across all frequencies). While this works for high frequencies, it doesn't account for low-frequency "1/f" noise.
  • Low-Photon Discrepancy: The model still shows some disagreement with experimental data at very low photon numbers ($|\alpha|^2 < 2$), likely because non-Markovian (memory-based) noise effects become dominant there.
  • Future Refinement: The authors suggest that future work should involve more detailed modeling of specific noise sources (like quasi-particles or inductive loss) to refine the low-frequency accuracy of the Lindbladian.

Summary updated

Abstract

We derive the static effective Lindbladian beyond the rotating wave approximation (RWA) for a driven nonlinear oscillator coupled to a bath of harmonic oscillators. The associated dissipative effects may explain orders of magnitude differences between the predictions of the ordinary RWA model and results from recent superconducting circuits experiments on the Kerr-cat qubit. The higher-order dissipators found in our calculations have important consequences for quantum error-correction protocols and parametric processses.
BibTeX
@misc{venkatraman2022static,
  author = {Jayameenakshi Venkatraman and Xu Xiao and Rodrigo G. Cortiñas and Michel Devoret},
  title = {{On the static effective Lindbladian of the squeezed Kerr oscillator}},
  year = {2022},
  eprint = {2209.11193},
  archivePrefix = {arXiv},
  url = {https://arxiv.org/abs/2209.11193},
}

Loading related content...

Figure