Search

The stabilization of quantum states is a fundamental problem for realizing various quantum technologies. Measurement-based-feedback strategies have demonstrated powerful performance, and the construction of quantum control signals using measurement information has attracted great interest. However, the interaction between quantum systems and the environment is inevitable, especially when measurements are introduced, which leads to decoherence. To mitigate decoherence, it is desirable to stabilize quantum systems faster, thereby reducing the time of interaction with the environment. In this paper, we utilize information obtained from measurement and apply deep reinforcement learning (DRL) algorithms, without explicitly constructing specific complex measurement-control mappings, to rapidly drive random initial quantum state to the target state. The proposed DRL algorithm has the ability to speed up the convergence to a target state, which shortens the interaction between quantum systems and their environments to protect coherence. Simulations are performed on two-qubit and three-qubit systems, and the results show that our algorithm can successfully stabilize random initial quantum system to the target entangled state, with a convergence time faster than traditional methods such as Lyapunov feedback control. Moreover, it exhibits robustness against imperfect measurements and delays in system evolution.

This paper introduces an enhanced framework for managing Battery Energy Storage Systems (BESS) in residential communities. The non-convex BESS control problem is first addressed using a gradient-based optimizer, providing a benchmark solution. Subsequently, the problem is tackled using multiple Deep Reinforcement Learning (DRL) agents, with a specific emphasis on the off-policy Soft Actor-Critic (SAC) algorithm. This version of SAC incorporates reward refinement based on this non-convex problem, applying logarithmic scaling to enhance convergence rates. Additionally, a safety mechanism selects only feasible actions from the action space, aimed at improving the learning curve, accelerating convergence, and reducing computation times. Moreover, the state representation of this DRL approach now includes uncertainties quantified in the entropy term, enhancing the model's adaptability across various entropy types. This developed system adheres to strict limits on the battery's State of Charge (SOC), thus preventing breaches of SOC boundaries and extending the battery lifespan. The robustness of the model is validated across several Australian states' districts, each characterized by unique uncertainty distributions. By implementing the refined SAC, the SOC consistently surpasses 50 percent by the end of each day, enabling the BESS control to start smoothly for the next day with some reserve. Finally, this proposed DRL method achieves a mean reduction in optimization time by 50 percent and an average cost saving of 40 percent compared to the gradient-based optimization benchmark.

Efficiently controlling linear Gaussian quantum (LGQ) systems is a significant task in both the study of fundamental quantum theory and the development of modern quantum technology. Here, we propose a general quantum-learning-control method for optimally controlling LGQ systems based on the gradient-descent algorithm. Our approach flexibly designs the loss function for diverse tasks by utilizing first- and second-order moments that completely describe the quantum state of LGQ systems. We demonstrate both deep optomechanical cooling and large optomechanical entanglement using this approach. Our approach enables the fast and deep ground-state cooling of a mechanical resonator within a short time, surpassing the limitations of sideband cooling in the continuous-wave driven strong-coupling regime. Furthermore, optomechanical entanglement could be generated remarkably fast and surpass several times the corresponding steady-state entanglement, even when the thermal phonon occupation reaches one hundred. This work will not only broaden the application of quantum learning control, but also open an avenue for optimal control of LGQ systems., Comment: 14 pages, 7 figures

The 2019-20 Australia bushfire incurred numerous economic losses and significantly affected the operations of power systems. A power station or transmission line can be significantly affected due to bushfires, leading to an increase in operational costs. We study a fundamental but challenging problem of planning the optimal power flow (OPF) for power systems subject to bushfires. Considering the stochastic nature of bushfire spread, we develop a model to capture such dynamics based on Moore's neighborhood model. Under a periodic inspection scheme that reveals the in-situ bushfire status, we propose an online optimization modeling framework that sequentially plans the power flows in the electricity network. Our framework assumes that the spread of bushfires is non-stationary over time, and the spread and containment probabilities are unknown. To meet these challenges, we develop a contextual online learning algorithm that treats the in-situ geographical information of the bushfire as a 'spatial context'. The online learning algorithm learns the unknown probabilities sequentially based on the observed data and then makes the OPF decision accordingly. The sequential OPF decisions aim to minimize the regret function, which is defined as the cumulative loss against the clairvoyant strategy that knows the true model parameters. We provide a theoretical guarantee of our algorithm by deriving a bound on the regret function, which outperforms the regret bound achieved by other benchmark algorithms. Our model assumptions are verified by the real bushfire data from NSW, Australia, and we apply our model to two power systems to illustrate its applicability.

We present a switching control strategy based on Lyapunov control for arbitrary state transitions in open qubit systems. With coherent vector representation, we propose a switching control strategy, which can prevent the state of the qubit from entering invariant sets and singular value sets, effectively driving the system ultimately to a sufficiently small neighborhood of target states. In comparison to existing works, this control strategy relaxes the strict constraints on system models imposed by special target states. Furthermore, we identify conditions under which the open qubit system achieves finite-time stability (FTS) and finite-time contractive stability (FTCS), respectively. This represents a critical improvement in quantum state transitions, especially considering the asymptotic stability of arbitrary target states is unattainable in open quantum systems. The effectiveness of our proposed method is convincingly demonstrated through its application in a qubit system affected by various types of decoherence, including amplitude, dephasing and polarization decoherence., Comment: 12 pages, 7 figures

Quantum process tomography is a critical task for characterizing the dynamics of quantum systems and achieving precise quantum control. In this paper, we propose a two-stage solution for both trace-preserving and non-trace-preserving quantum process tomography. Utilizing a tensor structure, our algorithm exhibits a computational complexity of $O(MLd^2)$ where $d$ is the dimension of the quantum system and $ M $, $ L $ represent the numbers of different input states and measurement operators, respectively. We establish an analytical error upper bound and then design the optimal input states and the optimal measurement operators, which are both based on minimizing the error upper bound and maximizing the robustness characterized by the condition number. Numerical examples and testing on IBM quantum devices are presented to demonstrate the performance and efficiency of our algorithm., Comment: 41 pages, 7 figures

This article presents a robust control strategy using Time-Optimal Model Predictive Control (TOMPC) for a two-level quantum system subject to bounded uncertainties. In this method, the control field is optimized over a finite horizon using a nominal quantum system as the reference and then the optimal control for the first time interval is applied and a projective measurement is implemented on the uncertain system. The new control field for the next time interval will be iteratively optimized based on the measurement result. We present theoretical results to guarantee the stability of the TOMPC algorithm. We also characterize the robustness and the convergence rate of the TOMPC strategy for the control of two-level systems. Numerical simulations further demonstrate that, in the presence of uncertainties, our quantum TOMPC algorithm enhances robustness and steers the state to the desired state with high fidelity. This work contributes to the progress of Model Predictive Control in quantum control and explores its potential in practical applications of quantum technology., Comment: 22 pages, 3 figures

In the noisy intermediate-scale quantum (NISQ) era, the capabilities of variational quantum algorithms are greatly constrained due to a limited number of qubits and the shallow depth of quantum circuits. We may view these variational quantum algorithms as weak learners in supervised learning. Ensemble methods are general approaches to combining weak learners to construct a strong one in machine learning. In this paper, by focusing on classification, we theoretically establish and numerically verify a learning guarantee for quantum adaptive boosting (AdaBoost). The supervised-learning risk bound describes how the prediction error of quantum AdaBoost on binary classification decreases as the number of boosting rounds and sample size increase. We further empirically demonstrate the advantages of quantum AdaBoost by focusing on a 4-class classification. The quantum AdaBoost not only outperforms several other ensemble methods, but in the presence of noise it can also surpass the ideally noiseless but unboosted primitive classifier after only a few boosting rounds. Our work indicates that in the current NISQ era, introducing appropriate ensemble methods is particularly valuable in improving the performance of quantum machine learning algorithms., Comment: 9 figures; Add numerical simulations

Quantum kernel methods have been widely recognized as one of promising quantum machine learning algorithms that have potential to achieve quantum advantages. In this paper, we theoretically characterize the power of noisy quantum kernels and demonstrate that under global depolarization noise, for different input data the predictions of the optimal hypothesis inferred by the noisy quantum kernel approximately concentrate towards some fixed value. In particular, we depict the convergence rate in terms of the strength of quantum noise, the size of training samples, the number of qubits, the number of layers affected by quantum noises, as well as the number of measurement shots. Our results show that noises may make quantum kernel methods to only have poor prediction capability, even when the generalization error is small. Thus, we provide a crucial warning to employ noisy quantum kernel methods for quantum computation and the theoretical results can also serve as guidelines when developing practical quantum kernel algorithms for achieving quantum advantages., Comment: 3 figures

In this paper, we consider the real-time parameter estimation problem for a ZZ-coupled system composed of two qubits in the presence of spontaneous emission. To enhance the estimation precision of the coupling coefficient, we first propose two different control schemes, where the first one is feedback control based on quantum-jump detection, and the second one is hybrid control combining Markovian feedback and Hamiltonian control. The simulation results show that compared with free evolution, both control schemes can improve parameter precision and extend system coherence time. Next, on the basis of the two control schemes, we propose a practical single-parameter quantum recovery protocol based on Bayesian estimation theory. In this protocol, by employing batch-style adaptive measurement rules, parameter recovery is conducted to verify the effectiveness of both control schemes., Comment: 13 pages, 14 figures

Quantum annealers, coherent Ising machines and digital Ising machines for solving quantum-inspired optimization problems have been developing rapidly due to their near-term applications. The numerical solvers of the digital Ising machines are based on traditional computing devices. In this work, we propose a fast and efficient solver for the Ising optimization problems. The algorithm consists of a pruning method that exploits the graph information of the Ising model to reduce the computational complexity, and a domain selection method which introduces significant acceleration by relaxing the discrete feasible domain into a continuous one to incorporate the efficient gradient descent method. The experiment results show that our solver can be an order of magnitude faster than the classical solver, and at least two times faster than the quantum-inspired annealers including the simulated quantum annealing on the benchmark problems. With more relaxed requirements on hardware and lower cost than quantum annealing, the proposed solver has the potential for near-term application in solving challenging optimization problems as well as serving as a benchmark for evaluating the advantage of quantum devices.

Variational quantum eigensolvers (VQEs) are one of the most important and effective applications of quantum computing, especially in the current noisy intermediate-scale quantum (NISQ) era. There are mainly two ways for VQEs: problem-agnostic and problem-specific. For problem-agnostic methods, they often suffer from trainability issues. For problem-specific methods, their performance usually relies upon choices of initial reference states which are often hard to determine. In this paper, we propose an Entanglement-variational Hardware-efficient Ansatz (EHA), and numerically compare it with some widely used ansatzes by solving benchmark problems in quantum many-body systems and quantum chemistry. Our EHA is problem-agnostic and hardware-efficient, especially suitable for NISQ devices and having potential for wide applications. EHA can achieve a higher level of accuracy in finding ground states and their energies in most cases even compared with problem-specific methods. The performance of EHA is robust to choices of initial states and parameters initialization and it has the ability to quickly adjust the entanglement to the required amount, which is also the fundamental reason for its superiority., Comment: 18 pages, 23 figures

Quantum process tomography (QPT) is a fundamental task to characterize the dynamics of quantum systems. In contrast to standard QPT, ancilla-assisted process tomography (AAPT) framework introduces an extra ancilla system such that a single input state is needed. In this paper, we extend the two-stage solution, a method originally designed for standard QPT, to perform AAPT. Our algorithm has $O(Md_A^2d_B^2)$ computational complexity where $ M $ is the type number of the measurement operators, $ d_A $ is the dimension of the quantum system of interest, and $d_B$ is the dimension of the ancilla system. Then we establish an error upper bound and further discuss the optimal design on the input state in AAPT. A numerical example on a phase damping process demonstrates the effectiveness of the optimal design and illustrates the theoretical error analysis., Comment: 6 pages, 3 figures

With the rapid advancement of quantum computing, Quantum Approximate Optimization Algorithm (QAOA) is considered as a promising candidate to demonstrate quantum supremacy, which exponentially solves a class of Quadratic Unconstrained Binary Optimization (QUBO) problems. However, limited qubit availability and restricted coherence time challenge QAOA to solve large-scale pseudo-Boolean problems on currently available Near-term Intermediate Scale Quantum (NISQ) devices. In this paper, we propose a distributed QAOA which can solve a general pseudo-Boolean problem by converting it to a simplified Ising model. Different from existing distributed QAOAs' assuming that local solutions are part of a global one, which is not often the case, we introduce community detection using Louvian algorithm to partition the graph where subgraphs are further compressed by community representation and merged into a higher level subgraph. Recursively and backwards, local solutions of lower level subgraphs are updated by heuristics from solutions of higher level subgraphs. Compared with existing methods, our algorithm incorporates global heuristics into local solutions such that our algorithm is proven to achieve a higher approximation ratio and outperforms across different graph configurations. Also, ablation studies validate the effectiveness of each component in our method., Comment: 12 pages, 6 figures

Quantum state tomography (QST) is the process of reconstructing the complete state of a quantum system (mathematically described as a density matrix) through a series of different measurements. These measurements are performed on a number of identical copies of the quantum system, with outcomes gathered as frequencies. QST aims to recover the density matrix and the corresponding properties of the quantum state from the measured frequencies. Although an informationally complete set of measurements can specify quantum state accurately in an ideal scenario with a large number of identical copies, both measurements and identical copies are restricted and imperfect in practical scenarios, making QST highly ill-posed. The conventional QST methods usually assume adequate or accurate measured frequencies or rely on manually designed regularizers to handle the ill-posed reconstruction problem, suffering from limited applications in realistic scenarios. Recent advances in deep neural networks (DNNs) led to the emergence of deep learning (DL) in QST. However, existing DL-based QST approaches often employ generic DNN models that are not optimized for imperfect conditions of QST. In this paper, we propose a transformer-based autoencoder architecture tailored for QST with imperfect measurement data. Our method leverages a transformer-based encoder to extract an informative latent representation (ILR) from imperfect measurement data and employs a decoder to predict the quantum states based on the ILR. We anticipate that the high-dimensional ILR will capture more comprehensive information about quantum states. To achieve this, we conduct pre-training of the encoder using a pretext task that involves reconstructing high-quality frequencies from measured frequencies. Extensive simulations and experiments demonstrate the remarkable ability of the ILR in dealing with imperfect measurement data in QST.

One of the fundamental tasks in information theory is the compression of information. To achieve this in the quantum domain, quantum autoencoders that aim to compress quantum states to low-dimensional ones have been proposed. When taking a pure state as the reference state, there exists an upper bound for the encoding fidelity. This bound limits the compression rate for high-rank states that have high entropy. To overcome the entropy inconsistency between the initial states and the reconstructed states, we allow the reference state to be a mixed state. A new cost function that combines the encoding fidelity and the quantum mutual information is proposed for compressing general input states. In particular, we consider the reference states to be a mixture of maximally mixed states and pure states. To achieve efficient compression for different states, two strategies for setting the ratio of mixedness (in the mixture of maximally mixed states and pure states) are provided based on prior knowledge about quantum states or observations obtained from the training process. Numerical results on thermal states of the transverse-field Ising model, Werner states, and maximally mixed states blended with pure states illustrate the effectiveness of the proposed method. In addition, quantum autoencoders using mixed reference states are experimentally implemented on IBM Quantum devices to compress and reconstruct thermal states and Werner states.

Quantum entanglement plays a fundamental role in quantum computation and quantum communication. Feedback control has been widely used in stochastic quantum systems to generate given entangled states since it has good robustness, where the time required to compute filter states and conduct filter based control usually cannot be ignored in many practical applications. This paper designed two control strategies based on the Lyapunov method to prepare a class of entangled states for qubit systems with a constant delay time. The first one is bang bang like control strategy, which has a simple form with switching between a constant value and zero, the stability of which is proved. Another control strategy is switching Lyapunov control, where a constant delay time is introduced in the filter-based feedback control law to compensate for the computation time. Numerical results on a two qubit system illustrate the effectiveness of these two proposed control strategies.

Optical Parametric Oscillators (OPOs) have wide applications in quantum optics for generating squeezed states and developing advanced technologies. When the phase or/and the amplitude of the pumping field for an OPO have fluctuations due to fault signals, time-varying uncertainties will be introduced in the dynamic parameters of the system. In this paper, we investigate how to design a fault-tolerant $H^\infty$ controller for an OPO with a disturbance input and time-varying uncertainties, which can achieve the required $H^\infty$ performance of the quantum system. We apply robust $H^\infty$ control theory to a quantum system, and design a passive controller and an active controller based on the solutions to two Riccati equations. The passive controller has a simple structure and is easy to be implemented by using only passive optical components, while the active quantum controller may achieve improved performance. The control performance of the proposed two controllers and one controller that was designed without consideration of system uncertainties is compared by numerical simulations in a specific OPO, and the results show that the designed controllers work effectively for fluctuations in both the phase and amplitude of the pumping field.

Quantum state tomography (QST) is the process of reconstructing the state of a quantum system (mathematically described as a density matrix) through a series of different measurements, which can be solved by learning a parameterized function to translate experimentally measured statistics into physical density matrices. However, the specific structure of quantum measurements for characterizing a quantum state has been neglected in previous work. In this paper, we explore the similarity between highly structured sentences in natural language and intrinsically structured measurements in QST. To fully leverage the intrinsic quantum characteristics involved in QST, we design a quantum-aware transformer (QAT) model to capture the complex relationship between measured frequencies and density matrices. In particular, we query quantum operators in the architecture to facilitate informative representations of quantum data and integrate the Bures distance into the loss function to evaluate quantum state fidelity, thereby enabling the reconstruction of quantum states from measured data with high fidelity. Extensive simulations and experiments (on IBM quantum computers) demonstrate the superiority of the QAT in reconstructing quantum states with favorable robustness against experimental noise.

Unique properties of multimode non-Hermitian lattice dynamics can be utilized to construct exponentially sensitive sensors. However, the impact of noise remains unclear, which may severely degrade their sensitivity. We analytically characterize and highlight the impact of loss and gain on the sensitivity revival and stability of non-Hermitian sensors. Defying the general belief that the superiority of quantum sensing will vanish in the presence of loss, we find that by proactively tuning the loss, the exponential sensitivity can be surprisingly regained when the sensing dynamics is stable. Furthermore, we prove that gain is crucial to fully revive the ideally exponential sensitivity and to ensure the stability of non-Hermitian sensing by making a balanced loss and gain. Our paper opens a way to significantly enhance the sensitivity by proactively tuning the loss and gain, which may promote future quantum sensing and quantum engineering., Comment: 18 pages, 3 figures

Due to its property of not requiring prior knowledge of the environment, reinforcement learning has significant potential for quantum control problems. In this work, we investigate the effectiveness of continuous control policies based on deep deterministic policy gradient. To solve the sparse reward signal in quantum learning control problems, we propose an auxiliary task-based deep reinforcement learning (AT-DRL) for quantum control. In particular, we first design a guided reward function based on the fidelity of quantum states that enables incremental fidelity improvement. Then, we introduce the concept of an auxiliary task whose network shares parameters with the main network to predict the reward provided by the environment (called the main task). The auxiliary task learns synchronously with the main task, allowing one to select the most relevant features of the environment, thus aiding the agent in comprehending how to achieve the desired state. The numerical simulations demonstrate that the proposed AT-DRL can provide a solution to the sparse reward in quantum systems, and has great potential in designing control pulses that achieve efficient quantum state preparation., Comment: 13 pages, 11 figures

Parameterized quantum circuits (PQCs) have been widely used as a machine learning model to explore the potential of achieving quantum advantages for various tasks. However, training PQCs is notoriously challenging owing to the phenomenon of plateaus and/or the existence of (exponentially) many spurious local minima. To enhance trainability, in this work we propose an efficient parameter initialization strategy with theoretical guarantees. We prove that by reducing the initial domain of each parameter inversely proportional to the square root of circuit depth, the magnitude of the cost gradient decays at most polynomially with respect to qubit count and circuit depth. Our theoretical results are substantiated through numerical simulations of variational quantum eigensolver tasks. Moreover, we demonstrate that the reduced-domain initialization strategy can protect specific quantum neural networks from exponentially many spurious local minima. Our results highlight the significance of an appropriate parameter initialization strategy, offering insights to enhance the trainability and convergence of variational quantum algorithms., Comment: 7 figures; Numerical simulations added

High-fidelity quantum gate design is important for various quantum technologies, such as quantum computation and quantum communication. Numerous control policies for quantum gate design have been proposed given a dynamical model of the quantum system of interest. However, a quantum system is often highly sensitive to noise, and obtaining its accurate modeling can be difficult for many practical applications. Thus, the control policy based on a quantum system model may be unpractical for quantum gate design. Also, quantum measurements collapse quantum states, which makes it challenging to obtain information through measurements during the control process. In this paper, we propose a novel training framework using deep reinforcement learning for model-free quantum control. The proposed framework relies only on the measurement at the end of the control process and offers the ability to find the optimal control policy without access to quantum systems during the learning process. The effectiveness of the proposed technique is numerically demonstrated for model-free quantum gate design and quantum gate calibration using off-policy reinforcement learning algorithms., Comment: 12 pages, 17 figures, accepted for publication in the IEEE Transactions on Artificial Intelligence, in press

We present a combined analytical and numerical study for coherent terahertz control of a single molecular polariton, formed by strongly coupling two rotational states of a molecule with a single-mode cavity. Compared to the bare molecules driven by a single terahertz pulse, the presence of a cavity strongly modifies the post-pulse orientation of the polariton, making it difficult to obtain its maximal degree of orientation. To solve this challenging problem toward achieving complete quantum coherent control, we derive an analytical solution of a pulse-driven quantum Jaynes-Cummings model by expanding the wave function into entangled states and constructing an effective Hamiltonian. We utilize it to design a composite terahertz pulse and obtain the maximum degree of orientation of the polariton by exploiting photon blockade effects. This work offers a new strategy to study rotational dynamics in the strong-coupling regime and provides a method for complete quantum coherent control of a single molecular polariton. It, therefore, has direct applications in polariton chemistry and molecular polaritonics for exploring novel quantum optical phenomena., Comment: 16 pages, 5 figures , accepted by Physical Review Letters on 19 December, 2022

The rapid proliferation of learning systems in an arbitrarily changing environment mandates the need for managing tensions between exploration and exploitation. This work proposes a quantum-inspired bandit learning approach for the learning-and-adapting-based offloading problem where a client observes and learns the costs of each task offloaded to the candidate resource providers, e.g., fog nodes. In this approach, a new action update strategy and novel probabilistic action selection are adopted, provoked by the amplitude amplification and collapse postulate in quantum computation theory, respectively. We devise a locally linear mapping between a quantum-mechanical phase in a quantum domain, e.g., Grover-type search algorithm, and a distilled probability-magnitude in a value-based decision-making domain, e.g., adversarial multi-armed bandit algorithm. The proposed algorithm is generalized, via the devised mapping, for better learning weight adjustments on favourable/unfavourable actions and its effectiveness is verified via simulation., Comment: Accepted to appear in IEEE TKDE

Quantum detector tomography (QDT) is a fundamental technique for calibrating quantum devices and performing quantum engineering tasks. In this paper, we utilize regularization to improve the QDT accuracy whenever the probe states are informationally complete or informationally incomplete. In the informationally complete scenario, without regularization, we optimize the resource (probe state) distribution by converting it to a semidefinite programming problem. Then in both the informationally complete and informationally incomplete scenarios, we discuss different regularization forms and prove the mean squared error scales as $ O(\frac{1}{N}) $ or tends to a constant with $ N $ state copies under the static assumption. We also characterize the ideal best regularization for the identifiable parameters, accounting for both the informationally complete and informationally incomplete scenarios. Numerical examples demonstrate the effectiveness of different regularization forms and a quantum optical experiment test shows that a suitable regularization form can reach a reduced mean squared error., Comment: 19 pages, 10 figures

The current quantum reinforcement learning control models often assume that the quantum states are known a priori for control optimization. However, full observation of quantum state is experimentally infeasible due to the exponential scaling of the number of required quantum measurements on the number of qubits. In this paper, we investigate a robust reinforcement learning method using partial observations to overcome this difficulty. This control scheme is compatible with near-term quantum devices, where the noise is prevalent and predetermining the dynamics of quantum state is practically impossible. We show that this simplified control scheme can achieve similar or even better performance when compared to the conventional methods relying on full observation. We demonstrate the effectiveness of this scheme on examples of quantum state control and quantum approximate optimization algorithm. It has been shown that high-fidelity state control can be achieved even if the noise amplitude is at the same level as the control amplitude. Besides, an acceptable level of optimization accuracy can be achieved for QAOA with noisy control Hamiltonian. This robust control optimization model can be trained to compensate the uncertainties in practical quantum computing.

While reinforcement learning (RL) algorithms are achieving state-of-the-art performance in various challenging tasks, they can easily encounter catastrophic forgetting or interference when faced with lifelong streaming information. In the paper, we propose a scalable lifelong RL method that dynamically expands the network capacity to accommodate new knowledge while preventing past memories from being perturbed. We use a Dirichlet process mixture to model the non-stationary task distribution, which captures task relatedness by estimating the likelihood of task-to-cluster assignments and clusters the task models in a latent space. We formulate the prior distribution of the mixture as a Chinese restaurant process (CRP) that instantiates new mixture components as needed. The update and expansion of the mixture are governed by the Bayesian non-parametric framework with an expectation maximization (EM) procedure, which dynamically adapts the model complexity without explicit task boundaries or heuristics. Moreover, we use the domain randomization technique to train robust prior parameters for the initialization of each task model in the mixture, thus the resulting model can better generalize and adapt to unseen tasks. With extensive experiments conducted on robot navigation and locomotion domains, we show that our method successfully facilitates scalable lifelong RL and outperforms relevant existing methods., Comment: Manuscript accepted by IEEE Transactions on Cybernetics, 2022, DOI: DOI: 10.1109/TCYB.2022.3170485

Bayesian policy reuse (BPR) is a general policy transfer framework for selecting a source policy from an offline library by inferring the task belief based on some observation signals and a trained observation model. In this paper, we propose an improved BPR method to achieve more efficient policy transfer in deep reinforcement learning (DRL). First, most BPR algorithms use the episodic return as the observation signal that contains limited information and cannot be obtained until the end of an episode. Instead, we employ the state transition sample, which is informative and instantaneous, as the observation signal for faster and more accurate task inference. Second, BPR algorithms usually require numerous samples to estimate the probability distribution of the tabular-based observation model, which may be expensive and even infeasible to learn and maintain, especially when using the state transition sample as the signal. Hence, we propose a scalable observation model based on fitting state transition functions of source tasks from only a small number of samples, which can generalize to any signals observed in the target task. Moreover, we extend the offline-mode BPR to the continual learning setting by expanding the scalable observation model in a plug-and-play fashion, which can avoid negative transfer when faced with new unknown tasks. Experimental results show that our method can consistently facilitate faster and more efficient policy transfer., Comment: Published in IEEE Transactions on Neural Networks and Learning Systems, 2023, DOI: 10.1109/TNNLS.2023.3281604

Multi-agent settings remain a fundamental challenge in the reinforcement learning (RL) domain due to the partial observability and the lack of accurate real-time interactions across agents. In this paper, we propose a new method based on local communication learning to tackle the multi-agent RL (MARL) challenge within a large number of agents coexisting. First, we design a new communication protocol that exploits the ability of depthwise convolution to efficiently extract local relations and learn local communication between neighboring agents. To facilitate multi-agent coordination, we explicitly learn the effect of joint actions by taking the policies of neighboring agents as inputs. Second, we introduce the mean-field approximation into our method to reduce the scale of agent interactions. To more effectively coordinate behaviors of neighboring agents, we enhance the mean-field approximation by a supervised policy rectification network (PRN) for rectifying real-time agent interactions and by a learnable compensation term for correcting the approximation bias. The proposed method enables efficient coordination as well as outperforms several baseline approaches on the adaptive traffic signal control (ATSC) task and the StarCraft II multi-agent challenge (SMAC)., Comment: Accepted by IEEE Transactions on Neural Networks, 2022, DOI: 10.1109/TNNLS.2022.3230701

Maximally entangled states (MES) are highly valued in quantum information processing. In quantum control, the creation of MES is typically treated as a state transfer problem with a predefined MES as the target. However, this approach is limited by the requirement to predetermine the MES structure. This paper introduces an improved quantum Lyapunov control approach that relies on the quantum entanglement measure to construct the Lyapunov function, instead of using the distance between quantum states. This strategy enables the preparation of any MES, regardless of whether its structure is known beforehand, using a single control scheme. The proposed entanglement control technique is unaffected by the number of entangled subsystems since it targets the entanglement measure as a scalar. Initially applied to bipartite pure states, this method demonstrates its capability to generate Bell states and their equivalents. Subsequent applications to bipartite mixed states and multipartite systems illustrate that the technique can produce MES with unspecified structures., Comment: 16pages, 6 figures, 3 tables

The development of estimation and control theories for quantum systems is a fundamental task for practical quantum technology. This vision article presents a brief introduction to challenging problems and potential opportunities in the emerging areas of quantum estimation, control and learning. The topics cover quantum state estimation, quantum parameter identification, quantum filtering, quantum open-loop control, quantum feedback control, machine learning for estimation and control of quantum systems, and quantum machine learning., Comment: 11 pages, vision paper from the perspective of systems and control

Quantum state tomography aiming at reconstructing the density matrix of a quantum state plays an important role in various emerging quantum technologies. Inspired by the intuition that machine learning has favorable robustness and generalization, we propose a deep neural networks based quantum state tomography (DNN-QST) approach, which are applied to three measurement-constrained cases, including few measurement copies and incomplete measurements as well as noisy measurements. Numerical results demonstrate that DNN-QST exhibits a great potential to achieve high fidelity for quantum state tomography with limited measurement resources and can achieve improved estimation when tomographic measurements suffer from noise. In addition, the results for 2-qubit states from quantum optical devices demonstrate the generalization of DNN-QST and its robustness against possible error in the experimental devices.

In this paper, we consider the filtering problem of an optical parametric oscillator (OPO). The OPO pump power may fluctuate due to environmental disturbances, resulting in uncertainty in the system modeling. Thus, both the state and the unknown parameter may need to be estimated simultaneously. We formulate this problem using a state-space representation of the OPO dynamics. Under the assumption of Gaussianity and proper constraints, the dual Kalman filter method and the joint extended Kalman filter method are employed to simultaneously estimate the system state and the pump power. Numerical examples demonstrate the effectiveness of the proposed algorithms., Comment: 8 pages, 5 figures

This paper provides a stabilizing preparation method for quantum Gaussian states by utilizing continuous measurement. The stochastic evolution of the open quantum system is described in terms of the quantum stochastic master equation. We present necessary and sufficient conditions for the system to have a unique stabilizing steady Gaussian state. The conditions are much weaker than those existing results presented in the approach of preparing Gaussian states through environment engineering. Parametric conditions of how to prepare an arbitrary pure Gaussian state are provided. This approach provides more degrees of freedom to choose the system Hamiltonian and the system-environment coupling operators, as compared with the case where dissipation-induced approach is employed. The stabilizing conditions for the case of imperfect measurement efficiency are also presented. These results may benefit practical experimental implementation in preparing quantum Gaussian states., Comment: 16 pages

Distinct non-Hermitian dynamics has demonstrated its advantages in improving measurement precision over traditional sensing protocols. Multi-mode non-Hermitian lattice dynamics can provide exponentially-enhanced quantum sensing where the quantum Fisher information (QFI) per photon increases exponentially with the lattice size. However, somewhat surprisingly, it was also shown that the quintessential non-Hermitian skin effect does not provide any true advantage. In this paper, we demonstrate the importance of optimizing the phase of the coherent drive, and the position of the injection and detection in multi-mode non-Hermitian quantum sensing. The QFI per photon can be exponentially-enhanced or exponentially-reduced depending on parameters of the drive and detection. Specifically, it is demonstrated that for large amplification by choosing appropriate coherent drive parameters, the non-Hermitian skin effect can provide exponentially-enhanced quantum sensing. Moreover, in the regime beyond linear response, skin-effect can also provide a dramatic advantage as compared to the local perturbation, and the proposed protocol is robust in tuning the amplification factor., Comment: 16 pages, 4 figures

In cellular-connected unmanned aerial vehicle (UAV) network, a minimization problem on the weighted sum of time cost and expected outage duration is considered. Taking advantage of UAV's adjustable mobility, an intelligent UAV navigation approach is formulated to achieve the aforementioned optimization goal. Specifically, after mapping the navigation task into a Markov decision process (MDP), a deep reinforcement learning (DRL) solution with novel quantum-inspired experience replay (QiER) framework is proposed to help the UAV find the optimal flying direction within each time slot, and thus the designed trajectory towards the destination can be generated. Via relating experienced transition's importance to its associated quantum bit (qubit) and applying Grover iteration based amplitude amplification technique, the proposed DRL-QiER solution commits a better trade-off between sampling priority and diversity. Compared to several representative baselines, the effectiveness and supremacy of the proposed DRL-QiER solution are demonstrated and validated in numerical results., Comment: 30 pages in standard single-column format

States of quantum many-body systems are defined in a high-dimensional Hilbert space, where rich and complex interactions among subsystems can be modelled. In machine learning, complex multiple multilinear correlations may also exist within input features. In this paper, we present a quantum-inspired multilinear model, named Residual Tensor Train (ResTT), to capture the multiple multilinear correlations of features, from low to high orders, within a single model. ResTT is able to build a robust decision boundary in a high-dimensional space for solving fitting and classification tasks. In particular, we prove that the fully-connected layer and the Volterra series can be taken as special cases of ResTT. Furthermore, we derive the rule for weight initialization that stabilizes the training of ResTT based on a mean-field analysis. We prove that such a rule is much more relaxed than that of TT, which means ResTT can easily address the vanishing and exploding gradient problem that exists in the existing TT models. Numerical experiments demonstrate that ResTT outperforms the state-of-the-art tensor network and benchmark deep learning models on MNIST and Fashion-MNIST datasets. Moreover, ResTT achieves better performance than other statistical methods on two practical examples with limited data which are known to have complex feature interactions., Comment: 12 pages, 6 figures