Recently developed tools such as neural-network based generative models and incremental tensor networks are promising methods for simplifying the learning of functions in high-dimensional space. We consider the application of these tools in two particular areas: modeling quantum states and imitation learning from high-dimensional data. In the first application, we consider the use of generative models such as normalizing flows as neural quantum states. Because quantum states are functions ψ such that |ψ|2 is a probability density, using normalizing flows as neural quantum states is a natural choice. Neural quantum states are commonly used in Variational Monte Carlo (VMC), where the parameters θ of a parameterized quantum state ψθ are varied in order to minimize some objective function. Due to the ability to exactly sample from and estimate probability densities from normalizing flows, the use of normalizing flows as neural quantum states simplifies the calculation of integrals used in Variational Monte Carlo. This is applied in two settings: first we consider using VMC for minimizing the ground state energy of high-dimensional continuous quantum states. Second, we explore using VMC for numerically solving time-dependent partial differential equations. For the second application, we explore using generative models and incremental tensor networks for extracting a low-dimensional latent space for use in downstream tasks like imitation learning. We first explore the use of neural-network architectures like Variational Autoencoders for learning a latent space that can be sampled from. Doing this allows for synthetic states to be generated, as well as predicted actions for those states. However, the use of these neural network-based feature extractors can be expensive, due to hyperparameter tuning and the gradient-based optimization methods used to find a good latent space representation. We explore the use of incremental tensor network approaches, which can extract a latent space in one pass through the training data and requires minimal hyperparameter tuning. We find that these incremental tensor network approaches learn a latent space representation that leads to similar imitation learning accuracy, but provides significant computational speedups over the neural network approaches.
Recently developed tools such as neural-network based generative models and incremental tensor networks are promising methods for simplifying the learning of functions in high-dimensional space. We consider the application of these tools in two particular areas: modeling quantum states and imitation learning from high-dimensional data. In the first application, we consider the use of generative models such as normalizing flows as neural quantum states. Because quantum states are functions ψ such that |ψ|2 is a probability density, using normalizing flows as neural quantum states is a natural choice. Neural quantum states are commonly used in Variational Monte Carlo (VMC), where the parameters θ of a parameterized quantum state ψθ are varied in order to minimize some objective function. Due to the ability to exactly sample from and estimate probability densities from normalizing flows, the use of normalizing flows as neural quantum states simplifies the calculation of integrals used in Variational Monte Carlo. This is applied in two settings: first we consider using VMC for minimizing the ground state energy of high-dimensional continuous quantum states. Second, we explore using VMC for numerically solving time-dependent partial differential equations. For the second application, we explore using generative models and incremental tensor networks for extracting a low-dimensional latent space for use in downstream tasks like imitation learning. We first explore the use of neural-network architectures like Variational Autoencoders for learning a latent space that can be sampled from. Doing this allows for synthetic states to be generated, as well as predicted actions for those states. However, the use of these neural network-based feature extractors can be expensive, due to hyperparameter tuning and the gradient-based optimization methods used to find a good latent space representation. We explore the use of incremental tensor network approaches, which can extract a latent space in one pass through the training data and requires minimal hyperparameter tuning. We find that these incremental tensor network approaches learn a latent space representation that leads to similar imitation learning accuracy, but provides significant computational speedups over the neural network approaches.
Optimizing hyperparameters in deep learning models remains a significant bottleneck in the development and deployment of effective neural networks. Traditional methods, such as grid search and random search, often struggle with the exponentially growing search space, leading to substantial computational costs and limited exploration of potentially optimal configurations. This paper proposes a novel approach to deep learning hyperparameter optimization that leverages the principles of quantum-inspired optimization, specifically employing tensor network methods, notably Matrix Product States (MPS), to efficiently approximate the complex optimization landscape. Our method utilizes quantum-inspired sampling techniques to guide the exploration, effectively mimicking the probabilistic nature of quantum measurements. We demonstrate the feasibility and potential advantages of this approach through theoretical analysis and a conceptual framework. The core of our method lies in representing the hyperparameter space as a tensor network, allowing for efficient calculations of the objective function (e.g., validation loss) without exhaustively evaluating every possible combination. This significantly reduces the computational burden compared to conventional methods. The paper outlines a detailed algorithm and discusses key considerations for implementation, including tensor network construction, sampling strategies, and convergence criteria. We propose a framework for iteratively refining the tensor network representation based on observed validation performance, ultimately converging towards a high-performing hyperparameter configuration. The theoretical underpinning draws upon concepts from quantum mechanics, particularly the idea of measurement and probability distribution, to drive the optimization process. We believe this quantum-inspired approach offers a promising avenue for accelerating deep learning hyperparameter tuning and achieving state-of-the-art performance.
Studying the dynamics of open quantum systems can enable breakthroughs both in fundamental physics and applications to quantum engineering and quantum computation. Since the density matrix ρ, which is the fundamental description for the dynamics of such systems, is high-dimensional, customized deep generative neural networks have been instrumental in modeling ρ. However, the complex-valued nature and normalization constraints of ρ, as well as its complicated dynamics, prohibit a seamless connection between open quantum systems and the recent advances in deep generative modeling. Here we lift that limitation by utilizing a reformulation of open quantum system dynamics to a partial differential equation (PDE) for a corresponding probability distribution Q, the Husimi Q function. Thus, we model the Q function seamlessly with off-the-shelf deep generative models such as normalizing flows. Additionally, we develop novel methods for learning normalizing flow evolution governed by high-dimensional PDEs based on the Euler method and the application of the time-dependent variational principle. We name the resulting approach Q-Flow and demonstrate the scalability and efficiency of Q-Flow on open quantum system simulations, including the dissipative harmonic oscillator and the dissipative bosonic model. Q-Flow is superior to conventional PDE solvers and state-of-the-art physics-informed neural network solvers, especially in high-dimensional systems.
The representation of a quantum wave function as a neural network quantum state (NQS) provides a powerful variational ansatz for finding the ground states of many-body quantum systems. Nevertheless, due to the complex variational landscape, traditional methods often employ the computation of quantum geometric tensor, consequently complicating optimization techniques. Contributing to efforts aiming to formulate alternative methods, we introduce an approach that bypasses the computation of the metric tensor and instead relies exclusively on first-order gradient descent with Euclidean metric. This allows for the application of larger neural networks and the use of more standard optimization methods from other machine learning domains. Our approach leverages the principle of imaginary time evolution by constructing a target wave function derived from the Schrödinger equation, and then training the neural network to approximate this target. We make this method adaptive and stable by determining the optimal time step and keeping the target fixed until the energy of the NQS decreases. We demonstrate the benefits of our scheme via numerical experiments with 2D J_1-J_2 J 1 − J 2 , Heisenberg model, which showcase enhanced stability and energy accuracy in comparison to direct energy loss minimization. Importantly, our approach displays competitiveness with the well-established density matrix renormalization group method and NQS optimization with stochastic reconfiguration.
Modeling the joint distribution of high-dimensional data is a central task in unsupervised machine learning. In recent years, many interests have been attracted to developing learning models based on tensor networks, which have the advantages of a principle understanding of the expressive power using entanglement properties, and as a bridge connecting classical computation and quantum computation. Despite the great potential, however, existing tensor network models for unsupervised machine learning only work as a proof of principle, as their performance is much worse than the standard models such as restricted Boltzmann machines and neural networks. In this Letter, we present autoregressive matrix product states (AMPS), a tensor network model combining matrix product states from quantum many-body physics and autoregressive modeling from machine learning. Our model enjoys the exact calculation of normalized probability and unbiased sampling. We demonstrate the performance of our model using two applications, generative modeling on synthetic and real-world data, and reinforcement learning in statistical physics. Using extensive numerical experiments, we show that the proposed model significantly outperforms the existing tensor network models and the restricted Boltzmann machines, and is competitive with state-of-the-art neural network models.
Quantum many-body computation is a central challenge across multiple domains of scientific research, and Monte Carlo method has emerged as one of the most effective numerical techniques for investigating large-scale many-body systems. In recent years, the integration of Monte Carlo methods with machine learning has led to the development of neural quantum state approach. Quantum neural networks (QNNs) offer superior expressiveness compared to classical neural networks. In this paper, we propose a novel Quantum Neural Network Quantum State (QNNQS) method, based on the variational Monte Carlo framework. This approach leverages a quantum neural network to represent the many-body quantum state, and optimizes it iteratively through gradient descent to determine the ground-state energy. To assess the performance of this method, we conduct simulations of the one-dimensional transverse-field Ising model, investigating the influence of quantum circuit architecture on circuit expressibility, entangling capability, and computational accuracy. The results demonstrate that the QNNQS method achieves comparable simulation accuracy while utilizing significantly fewer parameters and requiring less training resources than classical network quantum state methods.
Neural network quantum states (NQS) have emerged as a powerful and flexible framework for addressing quantum many-body problems. While successful for model Hamiltonians, their application to molecular systems remains challenging for several reasons. In this work, we introduce three innovations to overcome some of the key limitations. (1) We develop a bounded-degree graph recurrent neural network (BDG–RNN) ansatz, which hybridizes the tensor network and neural network states and is more suitable to molecular electronic structure problems. As matrix product states (MPS) can be embedded into this ansatz, good initialization is possible for complex systems. (2) We introduce neural network correlators (NNCs) to further enhance expressivity and improve accuracy, without dramatically modifying the underlying variational Monte Carlo (VMC) optimization framework. Specifically, we introduce two types of restricted Boltzmann machine (RBM)-inspired correlators, namely, cos-RBM and Ising-RBM, which unlike previous correlators, such as Jastrow and real RBM, can adjust the sign structure of the wave function. (3) We introduce a semistochastic algorithm for local energy evaluation, which significantly reduces computational cost while maintaining high accuracy. Combining these advances, we demonstrate that our approaches can achieve chemical accuracy in challenging systems, including the one-dimensional hydrogen chain H 50, the iron–sulfur cluster [Fe 2 S 2 (SCH 3 ) 4 ] 2–, and a three-dimensional 3 × 3 × 2 hydrogen cluster H 18 . These methods are implemented in an open-source package, P y NQS ( https://github.com/Quantum-Chemistry-Group-BNU/PyNQS ), to advance NQS methodologies for quantum chemistry.