Machine Learning-Aided Optimal Control of a Qubit Subjected to External Noise

TL;DR

The paper tackles robust quantum control of an open qubit under non-Markovian dephasing by introducing a greybox framework that couples a whitebox physics-based model with a transformer-based blackbox trained on synthetic data. The dynamics are modeled with $H(t) = H_ctrl(t) + g \beta(t) \sigma_z$ under RTN and OU noise, enabling the neural network to learn environmental effects that influence control performance. Gradient-based optimization of ten-pulse control sequences yields gate fidelities above $99\%$ at low noise and above $90\%$ even at strong coupling across both noise types, validating the approach's robustness. The work demonstrates a viable path toward scalable, high-fidelity quantum control in noisy environments and outlines open challenges related to scalability to two-qubit gates and broader noise spectra.

Abstract

We apply a machine-learning-enhanced greybox framework to a quantum optimal control protocol for open quantum systems. Combining a whitebox physical model with a neural-network blackbox trained on synthetic data, the method captures non-Markovian noise effects and achieves gate fidelities above 90% under Random Telegraph and Ornstein-Uhlenbeck noise. Critical issues of the approach are discussed.

Figures (1)

• Figure 1: The transformer-based graybox architecture we used to model Markovian and non-Markovian open-system qubit dynamics.