State Estimation and Control for Stochastic Quantum Dynamics with Homodyne Measurement: Stabilizing Qubits under Uncertainty

Nahid Binandeh Dehaghani, A. Pedro Aguiar, Rafał WiśniewskiIEEE Access 2024

Abstract

This paper introduces a Lyapunov-based control approach with homodyne measurement. We study two filtering approaches: (i) the traditional quantum filtering and (ii) a modified version of the extended Kalman filtering. We examine both methods in order to directly estimate the evolution of the coherence vector elements, using sequential homodyne current measurements. The latter case explicitly addresses the dynamics of a stochastic master equation with correlated noise, transformed into a state-space representation, ensuring by construction the quantum properties of the estimated state variable. In addition, we consider the case where the quantum-mechanical Hamiltonian is unknown, and the system experiences uncertainties. In this case, we show as expected that both filters lose performance, exhibiting large expected estimation errors. To address this problem, we propose a simple multiple model estimation scheme that can be directly applied to any of the studied filters. We then reconstruct the estimated density operator ( \hatρ ), describing the full state of the system, and subject it to a control scheme. The proposed switching-based Lyapunov control scheme, which is fed with ( \hatρ ), guarantees noise-to-state practically stable in probability of the desired stationary target set with respect to the estimation error variance. We demonstrate our approach's efficacy in stabilizing a qubit coupled to a leaky cavity under homodyne detection in the presence of uncertainty in resonance frequency.

Introduction

In the development of reliable quantum computing, environmental noise and decoherence present significant hurdles. Quantum systems are inherently fragile, interacting with their surroundings and subject to the stochastic nature of quantum mechanics. This research addresses the challenge of stabilizing a qubit coupled to a "leaky cavity" under continuous homodyne measurement.

The primary difficulty lies in the extremely noisy nature of the measurement signal and the frequent lack of precise knowledge regarding physical parameters, such as the system's resonance frequency. Without accurate parameter estimation, standard control strategies often fail, resulting in substantial estimation errors and loss of stability. This paper proposes a framework to simultaneously estimate the qubit's state and its unknown parameters in real-time.

Methodology

The proposed framework utilizes a multi-layered approach for robust estimation and control:

  1. Quantum Kalman Filter (QKF): The authors adapted an Extended Kalman Filter (EKF) to accommodate the nonlinear dynamics of a qubit and the correlation between process and measurement noise. A key innovation is the inclusion of a projection operator, ensuring the estimated state remains within the "Bloch ball"—the space of physically valid quantum states.
  2. Multiple Model Adaptive Estimation (MMAE): To manage parameter uncertainty, a bank of filters operates in parallel, each assuming a different resonance frequency. By comparing predicted outputs against real-time measurement data, the system dynamically weights each model, converging on the most accurate physical representation.
  3. Switching Lyapunov Control: A Lyapunov-based controller switches the driving field ($\Omega$) based on the estimated state $\hat{\rho}$. This strategy provides a mathematical guarantee of "noise-to-state practical stability," keeping the system within a predictable neighborhood of the target state despite inherent noise.
Schematic of the quantum control system including the leaky cavity, homodyne detection setup, and the feedback controller.

Results

The efficacy of the proposed approach was evaluated through numerical simulations of a qubit in a leaky cavity across 100 stochastic realizations.

  • Enhanced Convergence: The modified Quantum Kalman Filter (KF) demonstrated significantly faster convergence to high fidelity compared to traditional quantum filters (QF). As shown in the fidelity comparison, the KF approach reaches a fidelity near 1.0 much more rapidly than the standard QF.
  • Parameter Identification: In scenarios with unknown resonance frequencies, the MMAE successfully identified the correct model. Weights assigned to the accurate model rose from an initial 20% to nearly 100% within a short timeframe (approximately 0.4s).
  • Robust Stabilization: The switching controller successfully stabilized the qubit at the target state with fidelity approaching 1.0 across all simulated realizations, proving its robustness against both quantum noise and initial parameter uncertainty.
Comparison of state estimation fidelity over time between the traditional Quantum Filter (QF) and the proposed Quantum Kalman Filter (KF).

Conclusion

This work presents a robust framework for qubit stabilization under uncertainty. By integrating structural quantum constraints into a Kalman filtering framework and employing multiple model estimation, the researchers achieved high-fidelity control in noisy environments. While currently demonstrated for a two-level system, future research will explore scaling these techniques to multi-qubit systems and optimizing for real-time experimental implementation where latency is a critical factor.

Summary updated

Abstract

This paper introduces a Lyapunov-based control approach with homodyne measurement. We study two filtering approaches: (i) the traditional quantum filtering and (ii) a modified version of the extended Kalman filtering. We examine both methods in order to directly estimate the evolution of the coherence vector elements, using sequential homodyne current measurements. The latter case explicitly addresses the dynamics of a stochastic master equation with correlated noise, transformed into a state-space representation, ensuring by construction the quantum properties of the estimated state variable. In addition, we consider the case where the quantum-mechanical Hamiltonian is unknown, and the system experiences uncertainties. In this case, we show as expected that both filters lose performance, exhibiting large expected estimation errors. To address this problem, we propose a simple multiple model estimation scheme that can be directly applied to any of the studied filters. We then reconstruct the estimated density operator \( \hatρ \), describing the full state of the system, and subject it to a control scheme. The proposed switching-based Lyapunov control scheme, which is fed with \( \hatρ \), guarantees noise-to-state practically stable in probability of the desired stationary target set with respect to the estimation error variance. We demonstrate our approach's efficacy in stabilizing a qubit coupled to a leaky cavity under homodyne detection in the presence of uncertainty in resonance frequency.
BibTeX
@article{dehaghani2024state,
  author = {Nahid Binandeh Dehaghani and A. Pedro Aguiar and Rafał Wiśniewski},
  title = {{State Estimation and Control for Stochastic Quantum Dynamics with Homodyne Measurement: Stabilizing Qubits under Uncertainty}},
  journal = {IEEE Access},
  year = {2024},
}
Figure