Abstract: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 $\pi / 8$ 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 multi-qubit 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.

What problem does this paper attempt to address?

### What problem does this paper attempt to solve? This paper aims to solve a key problem in quantum computing: how to optimize the control parameters of Nonadiabatic Geometric Quantum Computation (NGQC) to achieve high - fidelity single - qubit and multi - qubit quantum - gate operations. Specifically, the researchers proposed a machine - learning - based average - fidelity - optimization method, optimizing the control parameters through a neural network with enhanced periodic features. #### Main problems and challenges: 1. **Fidelity of quantum - gate operations**: The traditional forms of control parameters (such as simple trigonometric functions) have the problem of insufficient fidelity when implementing single - qubit gates. For example, it is difficult for the T - gate (π/8 - gate) to achieve a fidelity of more than 99.99%. 2. **Influence of noise and decoherence**: In actual experiments, the system will be affected by system noise, additive white Gaussian noise and decoherence effects, resulting in a decrease in fidelity. 3. **Implementation of multi - qubit gates**: In order to achieve efficient multi - qubit - gate operations, a method that can optimize multiple physical systems simultaneously needs to be found, while existing methods are often limited by complex coupling conditions. #### Proposed solutions: 1. **Machine - learning - based optimization method**: The researchers introduced an optimization method based on a neural network, which uses a neural network with enhanced periodic features as a model. Through this method, the fidelity of single - qubit gates can be significantly improved (exceeding 99.99%), and these gates are more robust against noise and decoherence. 2. **Expansion of model space**: Through the nonlinear activation function of the neural network, any smooth periodic and aperiodic functions can be fitted, thus expanding the model space and making the optimized control parameters easier to be physically implemented. 3. **Application of reverse engineering**: By combining reverse - engineering technology, a Hamiltonian based on invariants is constructed, so that more free parameters can be used to control the evolution path of the system. #### Experimental verification: - The researchers verified the effectiveness of this method through numerical simulation, especially when implementing the T - gate (π/8 - gate), the average fidelity of the final state reached 0.9999. - In addition, they also evaluated the performance of this protocol under system noise, random noise and decoherence effects, and the results showed that this method has good robustness. In conclusion, this paper proposes an innovative machine - learning - inspired method for optimizing the control parameters in non - adiabatic geometric quantum computing, thereby achieving high - fidelity single - qubit and multi - qubit quantum - gate operations and performing well in practical applications.

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

Using optimal control to guide neural-network interpolation of continuously-parameterized gates

Machine-learning-inspired quantum control in many-body dynamics

Quantum optimal control of superconducting qubits based on machine-learning characterization

Quantum Geometric Machine Learning for Quantum Circuits and Control

Nonadiabatic Holonomic Quantum Computation and Its Optimal Control

Universal Quantum Control through Deep Reinforcement Learning

Flexible Scheme for the Implementation of Nonadiabatic Geometric Quantum Computation

Optimal control of linear Gaussian quantum systems via quantum learning control

Learning in Quantum Control: High-Dimensional Global Optimization for Noisy Quantum Dynamics

Deep Learning of Robust and High-Precision Quantum Controls

Path-optimized nonadiabatic geometric quantum computation on superconducting qubits

Searching nonadiabatic holonomic quantum gates via an optimization algorithm

Learning Robust and High-Precision Quantum Controls

Nonadiabatic geometric quantum gates with on-demand trajectories

Optimal Control Theory for Continuous-Variable Quantum Gates

Experimental Realization of Universal Time-optimal non-Abelian Geometric Gates

Large-scale quantum approximate optimization on non-planar graphs with machine learning noise mitigation

Nonadiabatic geometric quantum gates that are robust against systematic errors

Hybrid Quantum-Classical Approach to Quantum Optimal Control.