Machine-learning-inspired quantum optimal control of nonadiabatic geometric quantum computation via reverse engineering

Sep 17, 2026

Uploaded by Quinn

Quantum control is essential for practical quantum-computer operation, yet suitable and diverse control parameters remain difficult to obtain. The study introduces an average-fidelity-based, machine-learning-inspired optimization method that uses a neural network with periodic feature enhancement as a control ansatz. For single-qubit cat-state nonadiabatic geometric computation implemented through reverse engineering, phase gates such as the π/8 (T) gate achieve markedly higher fidelity (>99.99%) than simple trigonometric control. Numerical results show model-space expansion and enable high-quality cascaded multiqubit gates in a bosonic system. Single-qubit performance remains robust under systematic noise, additive white Gaussian noise, and decoherence, supporting feasibility in nonadiabatic NGQC.

LanguageEnglish

CategoryResearch & Report

FormatPDF

Pages11

Reading Time28 min

AccessFree to download

Introduction

Quantum control and multiqubit gates

Single-shot vs cascaded approaches

Geometric and nonadiabatic geometric phases

NGQC, NGQC+ and optimization strategies

Machine-learning-inspired quantum optimal control of nonadiabatic geometric quantum computation via reverse engineering Meng-Yun Mao , 1, 2 Zheng Cheng, 1, 2 Yan Xia ,3, 4 , * Andrzej M. Ole ,5, 6 ,† and Wen-Long You 1, 2 ,‡ 1 College of Physics, Nanjing University of Aeronautics and Astronautics, Nanjing 211106, China 2 Key Laboratory of Aerospace Information Materials and Physics (NUAA), MIIT, Nanjing 211106, China 3 Fujian Key Laboratory of Quantum Information and Quantum Optics, Fuzhou University, Fuzhou 350116, China 4 Department of Physics, Fuzhou University, Fuzhou 350116, China 5 Max Planck Institute for Solid State Research, Heisenbergstrasse 1, D-70569 Stuttgart, Germany 6 Institute of Theoretical Physics, Jagiellonian University, Prof. Stanisława Łojasiewicza 11, PL-30348 Kraków, Poland (Received 23 February 2023; accepted 11 September 2023; published 27 September 2023) Quantum control plays an irreplaceable role in practical use of quantum computers. However, some challenges have to be overcome to ﬁnd more suitable and diverse control parameters. We propose a promising and generalizable average-ﬁdelity-based machine-learning-inspired method to optimize the control parameters, in which a neural network with periodic feature enhancement is used as an ansatz. In the implementation of a single-qubit gate by cat-state nonadiabatic geometric quantum computation via reverse engineering, compared with the control parameters in the simple form of a trigonometric function, our approach can yield signiﬁcantly higher-ﬁdelity (>99.99%) phase gates, such as the π/8 gate (T gate) . Single-qubit gates are robust against systematic noise, additive white Gaussian noise, and decoherence. We numerically demonstrate that the neural network possesses the ability to expand the model space. With the help of our optimization, we provide a feasible way to implement cascaded multiqubit gates with high quality in a bosonic system. Therefore, the machine-learning-inspired method may be feasible in quantum optimal control of nonadiabatic geometric quantum computation. DOI: 10.1103/PhysRevA.108.032616 I. INTRODUCTION Multiqubit gates are widely used in quantum circuits [1–3], quantum error correction [4–6], and other ﬁelds [7–10] . Single-shot multiqubit gates [11–13] specify quantum circuits that evolute in a well-controlled, uninterrupted, and continuous-time way to implement the quantum computation [14] . Compared to the cascaded gates, single-shot multiqubit gates can greatly reduce the circuit depth and shorten the implementation time, thus suppressing decoherence [15, 16] . However, the single-shot method is difﬁcult to realize in experiments owing to the restricted conditions of simultaneously manipulating multiple physical systems and building complex couplings. One of the ways to mitigate the above difﬁculty is the application of single- and two-qubit gates to equivalently implement the function of multiqubit gates [17–19], and the *[xia-208@163.com](xia-208@163.com)[ ](xia-208@163.com)†[a.m.oles@fkf.mpi.de](a.m.oles@fkf.mpi.de)[ ](a.m.oles@fkf.mpi.de)‡[wlyou@nuaa.edu.cn](wlyou@nuaa.edu.cn) Published by the American Physical Society under the terms of the Creative Commons Attribution 4 .0 International license. Further distribution of this work must maintain attribution to the author(s) and the published article’s title, journal citation, and DOI. Open access publication funded by the Max Planck Society. decomposition is guaranteed by the Solovay-Kitaev theorem [20] . Although the decomposition method loses the upper hand in terms of the circuit depth and the implementation time, it has a wider scope of application due to direct execution on the quantum processing unit, namely, the brain of a quantum computer [21] . The realization of the synthetic gates depends on the universal single-qubit gates and a two-qubit entangling gate with high ﬁdelity [16] . However, statistical imprecision in the experimen

FAQs

What method does the paper propose for optimizing quantum control parameters?

It proposes an average-fidelity-based, machine-learning-inspired optimization using a neural network with periodic feature enhancement as an ansatz.

How does the proposed approach perform for the implementation of single-qubit phase gates?

Compared with trigonometric-form control parameters, it yields significantly higher-fidelity phase gates, achieving phase gates such as the π/8 (T) gate with fidelity greater than 99.99%.

Is the resulting single-qubit control robust to noise and imperfections?

Yes. The single-qubit gates are robust against systematic noise, additive white Gaussian noise, and decoherence, and numerical demonstrations show the neural network can expand the model space.

Sep 17, 2026

Uploaded by Quinn