Challenges
Datasets
Workspaces
Papers
Discussions
Leaderboard
Log inSign up
Challenges
Datasets
Workspaces
Papers
Discussions
Leaderboard
Blog
Job Board
Q3AS

© 2026 Aqora Quantum S.A.S.

TermsPrivacyLegal Notice

arXiv:2307.14715 · Physical Review A

Enhanced quantum state preparation via stochastic predictions of neural networks

Chaochao Li, Run-Hong He, Zhao-Ming Wang

Published Nov 17, 2023• 9 citations
arXivHTMLPDF

Abstract

In pursuit of enhancing the prediction capabilities of neural networks, it has been a longstanding objective to create datasets encompassing a diverse array of samples. The purpose is to broaden the horizons of the neural network and continually strive for improved prediction accuracy during the training process, which serves as the ultimate evaluation metric. However, in this paper, we explore an intriguing avenue for enhancing algorithm effectiveness through exploiting the knowledge blindness of the neural network. Our approach centers around a machine learning algorithm utilized for preparing arbitrary quantum states in a semiconductor double quantum dot system, a system characterized by highly constrained control degrees of freedom. We artificially define the characteristics of local optima in the state preparation task. When the optimization falls into a local optimum, disturbances are introduced through stochastic prediction, which gives a chance to escape from the local optimum. Notably, unlike previous methodologies that employ reinforcement learning to identify pulse patterns, we adopt a training approach akin to supervised learning, using it to dynamically design the pulse sequence. This approach has broad applicability and improves the efficiency of the algorithm.

DOI
10.1103/physreva.108.052418

Related papers

Machine-Learning-Assisted Pulse Design for State Preparation in a Noisy Environment

Zhaowei Wang, Hongyang Ma, Yun‐An Yan +2 more

High-precision quantum control is essential for quantum computing and quantum information processing. However, its practical implementation is challenged by environmental noise, which affects the stability and accuracy of quantum systems. In this paper, using machine learning techniques we propose a quantum control approach that incorporates environmental factors into the design of control schemes, improving the control fidelity in noisy environments. Specifically, we investigate arbitrary quantum state preparation in a two-level system coupled to a bosonic bath. We use both Deep Reinforcement Learning (DRL) and Supervised Learning (SL) algorithms to design specific control pulses that mitigate the noise. These two neural network (NN) based algorithm both have the advantage that the well trained NN can output the optimal pulse sequence for any environmental parameters. Comparing the performance of these two algorithms, our results show that DRL is more effective in low-noise environments due to its strong optimization capabilities, while SL provides greater stability and performs better in high-noise conditions. These findings highlight the potential of machine learning techniques to enhance the quantum control fidelity in practical applications.

arXiv:2508.20377•Aug 28, 2025•0 citations

Deep reinforcement learning for universal quantum state preparation via dynamic pulse control

Run-Hong He, Rui Wang, Shen-Shuang Nie +3 more

Abstract Accurate and efficient preparation of quantum state is a core issue in building a quantum computer. In this paper, we investigate how to prepare a certain single- or two-qubit target state from arbitrary initial states in semiconductor double quantum dots with only a few discrete control pulses by leveraging the deep reinforcement learning. Our method is based on the training of the network over numerous preparing tasks. The results show that once the network is well trained, it works for any initial states in the continuous Hilbert space. Thus repeated training for new preparation tasks is avoided. Our scheme outperforms the traditional optimization approaches based on gradient with both the higher efficiency and the preparation quality in discrete control space. Moreover, we find that the control trajectories designed by our scheme are robust against stochastic fluctuations within certain thresholds, such as the charge and nuclear noises.

arXiv:2012.00326•EPJ Quantum Technology•Dec 1, 2021•26 citations

Arbitrary quantum states preparation aided by deep reinforcement learning

Zhaowei Wang, Zhao-Ming Wang

Abstract The preparation of quantum states is essential in the realm of quantum information processing, and the development of efficient methodologies can significantly alleviate the strain on quantum resources. Under the framework of deep reinforcement learning (DRL), we integrate the initial and the target state information within the state preparation task together, so as to realize the control trajectory design between two arbitrary quantum states. Taking a semiconductor double quantum dots (DQDs) model as an example, our results demonstrate that the resulting control trajectories can effectively achieve arbitrary quantum state preparation (AQSP) for both single-qubit and two-qubit systems, with average fidelities of 0.9868 and 0.9556 for the test sets, respectively. For the DQDs, charge and nuclear noises will exist, thus decreases the average fidelity of AQSP. We have also incorporated noise amplitude as an input feature in the training process, leading to the development of noise-aware models. The control trajectories designed by these models can effectively suppress the influence of noise. Our research validates the effectiveness of DRL in quantum state preparation and offers solutions for multi-initial and multi-objective quantum control tasks, with potential for broader application in quantum control problems.

arXiv:2407.16368•Physica Scripta•Feb 26, 2025•3 citations

Quantum-Enhanced Optimization of Neural Network Architecture

Jincheng Zhang

Quantum computing holds the potential to revolutionize various fields, and the field of neural networks is no exception. This research investigates the application of quantum optimization algorithms to enhance the design of neural network architectures, aiming to improve performance across specific tasks. We explore the feasibility of utilizing a novel quantum optimization technique to systematically explore network architectures and identify configurations that yield superior results. This work presents a theoretical framework and preliminary results demonstrating the potential for quantum acceleration in neural network architecture design. The core claim centers around the design of a quantum algorithm capable of efficiently navigating a complex space of possible network architectures, ultimately leading to optimized neural network configurations. The research will focus on identifying key architectural parameters and evaluating their impact on performance using benchmark datasets. This paper provides a foundational analysis of this approach and outlines potential avenues for future exploration.

•Sep 6, 2026•0 citations

Universal quantum state preparation via revised greedy algorithm

Run-Hong He, Hai-Da Liu, Shengbin Wang +3 more

Abstract Preparation of quantum state lies at the heart of quantum information processing. The greedy algorithm provides a potential method to effectively prepare quantum state. However, the standard greedy (SG) algorithm, in general, cannot take the global maxima and instead becomes stuck on a local maxima. Based on the SG algorithm, in this paper we propose a revised version to design dynamic pulses to realize universal quantum state preparation, i.e. preparing an arbitrary state from another arbitrary one. As applications, we implement this scheme to the universal preparation of single- and two-qubit state in the context of semiconductor quantum dots and superconducting circuits. Evaluation results show that our scheme outperforms the alternative numerical optimizations with higher preparation quality while possesses the comparable high efficiency. Compared with the emerging machine learning, it shows better accessibility and does not require any training. Moreover, the numerical results show that the pulse sequences generated by our scheme are robust against various errors and noises. Our scheme opens a new avenue of optimization in few-level system and limited action space quantum control problems.

arXiv:2108.03351•Quantum Science and Technology•Aug 16, 2021•9 citations

Machine learning techniques for state recognition and auto-tuning in quantum dots

Sandesh S. Kalantre, Justyna P. Zwolak, Stephen Ragole +4 more

Abstract Recent progress in building large-scale quantum devices for exploring quantum computing and simulation has relied upon effective tools for achieving and maintaining good experimental parameters, i.e., tuning up devices. In many cases, including quantum dot-based architectures, the parameter space grows substantially with the number of qubits, and may become a limit to scalability. Fortunately, machine learning techniques for pattern recognition and image classification, using so-called deep neural networks, have shown surprising successes for computer-aided understanding of complex systems. We propose a new paradigm for fully automated experimental initialization through a closed-loop system relying on machine learning and optimization techniques. We use deep convolutional neural networks to characterize states and charge configurations of semiconductor quantum dot arrays when only measurements of a current−voltage characteristic of transport are available. For simplicity, we model a semiconductor nanowire connected to leads and capacitively coupled to depletion gates using the Thomas−Fermi approximation and Coulomb blockade physics. We then generate labeled training data for the neural networks, and find at least 90 % accuracy for charge and state identification for single and double dots. Using these characterization networks, we can then optimize the parameter space to achieve a desired configuration of the array, a technique we call “auto-tuning”. Finally, we show how such techniques can be implemented in an experimental setting by applying our approach to an experimental dataset, and outline further problems in this domain, from using charge sensing data to extensions to full one- and two-dimensional arrays, that can be tackled with machine learning.

arXiv:1712.04914•npj Quantum Information•Jan 14, 2019•108 citations
In pursuit of enhancing the prediction capabilities of neural networks, it has been a longstanding objective to create datasets encompassing a diverse array of samples. The purpose is to broaden the horizons of the neural network and continually strive for improved prediction accuracy during the training process, which serves as the ultimate evaluation metric. However, in this paper, we explore an intriguing avenue for enhancing algorithm effectiveness through exploiting the knowledge blindness of the neural network. Our approach centers around a machine learning algorithm utilized for preparing arbitrary quantum states in a semiconductor double quantum dot system, a system characterized by highly constrained control degrees of freedom. We artificially define the characteristics of local optima in the state preparation task. When the optimization falls into a local optimum, disturbances are introduced through stochastic prediction, which gives a chance to escape from the local optimum. Notably, unlike previous methodologies that employ reinforcement learning to identify pulse patterns, we adopt a training approach akin to supervised learning, using it to dynamically design the pulse sequence. This approach has broad applicability and improves the efficiency of the algorithm.
High-precision quantum control is essential for quantum computing and quantum information processing. However, its practical implementation is challenged by environmental noise, which affects the stability and accuracy of quantum systems. In this paper, using machine learning techniques we propose a quantum control approach that incorporates environmental factors into the design of control schemes, improving the control fidelity in noisy environments. Specifically, we investigate arbitrary quantum state preparation in a two-level system coupled to a bosonic bath. We use both Deep Reinforcement Learning (DRL) and Supervised Learning (SL) algorithms to design specific control pulses that mitigate the noise. These two neural network (NN) based algorithm both have the advantage that the well trained NN can output the optimal pulse sequence for any environmental parameters. Comparing the performance of these two algorithms, our results show that DRL is more effective in low-noise environments due to its strong optimization capabilities, while SL provides greater stability and performs better in high-noise conditions. These findings highlight the potential of machine learning techniques to enhance the quantum control fidelity in practical applications.
Abstract Accurate and efficient preparation of quantum state is a core issue in building a quantum computer. In this paper, we investigate how to prepare a certain single- or two-qubit target state from arbitrary initial states in semiconductor double quantum dots with only a few discrete control pulses by leveraging the deep reinforcement learning. Our method is based on the training of the network over numerous preparing tasks. The results show that once the network is well trained, it works for any initial states in the continuous Hilbert space. Thus repeated training for new preparation tasks is avoided. Our scheme outperforms the traditional optimization approaches based on gradient with both the higher efficiency and the preparation quality in discrete control space. Moreover, we find that the control trajectories designed by our scheme are robust against stochastic fluctuations within certain thresholds, such as the charge and nuclear noises.
Abstract The preparation of quantum states is essential in the realm of quantum information processing, and the development of efficient methodologies can significantly alleviate the strain on quantum resources. Under the framework of deep reinforcement learning (DRL), we integrate the initial and the target state information within the state preparation task together, so as to realize the control trajectory design between two arbitrary quantum states. Taking a semiconductor double quantum dots (DQDs) model as an example, our results demonstrate that the resulting control trajectories can effectively achieve arbitrary quantum state preparation (AQSP) for both single-qubit and two-qubit systems, with average fidelities of 0.9868 and 0.9556 for the test sets, respectively. For the DQDs, charge and nuclear noises will exist, thus decreases the average fidelity of AQSP. We have also incorporated noise amplitude as an input feature in the training process, leading to the development of noise-aware models. The control trajectories designed by these models can effectively suppress the influence of noise. Our research validates the effectiveness of DRL in quantum state preparation and offers solutions for multi-initial and multi-objective quantum control tasks, with potential for broader application in quantum control problems.
Quantum computing holds the potential to revolutionize various fields, and the field of neural networks is no exception. This research investigates the application of quantum optimization algorithms to enhance the design of neural network architectures, aiming to improve performance across specific tasks. We explore the feasibility of utilizing a novel quantum optimization technique to systematically explore network architectures and identify configurations that yield superior results. This work presents a theoretical framework and preliminary results demonstrating the potential for quantum acceleration in neural network architecture design. The core claim centers around the design of a quantum algorithm capable of efficiently navigating a complex space of possible network architectures, ultimately leading to optimized neural network configurations. The research will focus on identifying key architectural parameters and evaluating their impact on performance using benchmark datasets. This paper provides a foundational analysis of this approach and outlines potential avenues for future exploration.
Abstract Preparation of quantum state lies at the heart of quantum information processing. The greedy algorithm provides a potential method to effectively prepare quantum state. However, the standard greedy (SG) algorithm, in general, cannot take the global maxima and instead becomes stuck on a local maxima. Based on the SG algorithm, in this paper we propose a revised version to design dynamic pulses to realize universal quantum state preparation, i.e. preparing an arbitrary state from another arbitrary one. As applications, we implement this scheme to the universal preparation of single- and two-qubit state in the context of semiconductor quantum dots and superconducting circuits. Evaluation results show that our scheme outperforms the alternative numerical optimizations with higher preparation quality while possesses the comparable high efficiency. Compared with the emerging machine learning, it shows better accessibility and does not require any training. Moreover, the numerical results show that the pulse sequences generated by our scheme are robust against various errors and noises. Our scheme opens a new avenue of optimization in few-level system and limited action space quantum control problems.
Abstract Recent progress in building large-scale quantum devices for exploring quantum computing and simulation has relied upon effective tools for achieving and maintaining good experimental parameters, i.e., tuning up devices. In many cases, including quantum dot-based architectures, the parameter space grows substantially with the number of qubits, and may become a limit to scalability. Fortunately, machine learning techniques for pattern recognition and image classification, using so-called deep neural networks, have shown surprising successes for computer-aided understanding of complex systems. We propose a new paradigm for fully automated experimental initialization through a closed-loop system relying on machine learning and optimization techniques. We use deep convolutional neural networks to characterize states and charge configurations of semiconductor quantum dot arrays when only measurements of a current−voltage characteristic of transport are available. For simplicity, we model a semiconductor nanowire connected to leads and capacitively coupled to depletion gates using the Thomas−Fermi approximation and Coulomb blockade physics. We then generate labeled training data for the neural networks, and find at least 90 % accuracy for charge and state identification for single and double dots. Using these characterization networks, we can then optimize the parameter space to achieve a desired configuration of the array, a technique we call “auto-tuning”. Finally, we show how such techniques can be implemented in an experimental setting by applying our approach to an experimental dataset, and outline further problems in this domain, from using charge sensing data to extensions to full one- and two-dimensional arrays, that can be tackled with machine learning.