Solving Statistical Mechanics Using Variational Autoregressive Networks

Abstract

We propose a general framework for solving statistical mechanics of systems with finite size. The approach extends the celebrated variational mean-field approaches using autoregressive neural networks, which support direct sampling and exact calculation of normalized probability of configurations. It computes variational free energy, estimates physical quantities such as entropy, magnetizations and correlations, and generates uncorrelated samples all at once. Training of the network employs the policy gradient approach in reinforcement learning, which unbiasedly estimates the gradient of variational parameters. We apply our approach to several classic systems, including 2D Ising models, the Hopfield model, the Sherrington-Kirkpatrick model, and the inverse Ising model, for demonstrating its advantages over existing variational mean-field methods. Our approach sheds light on solving statistical physics problems using modern deep generative neural networks.

References in corpus (12)

• Geometric deep learning: going beyond Euclidean data
• Solving the Quantum Many-Body Problem with Artificial Neural Networks
• Tensor renormalization group approach to 2D classical lattice models
• Statistical physics of inference: Thresholds and algorithms
• Inverse statistical problems: from the inverse Ising problem to data science
• Second Renormalization of Tensor-Network States
• The Ising Model for Neural Data: Model Quality and Approximate Methods for Extracting Functional Connectivity
• Neural Network Renormalization Group
• Small-correlation expansions for the inverse Ising problem
• The Bethe approximation for solving the inverse Ising problem: a comparison with other inference methods
• Bethe-Peierls approximation and the inverse Ising model
• Fast Wavenet Generation Algorithm

Cited by in corpus (117)

• Machine learning and the physical sciences
• Towards Explainable Artificial Intelligence
• Recurrent Neural Network Wave Functions
• Deep autoregressive models for the efficient variational simulation of many-body quantum systems
• Equivariant flow-based sampling for lattice gauge theory
• Differentiable Quantum Architecture Search
• NetKet 3: Machine Learning Toolbox for Many-Body Quantum Systems
• Machine Learning for Quantum Matter
• Estimation of Thermodynamic Observables in Lattice Field Theories with Deep Generative Models
• Asymptotically unbiased estimation of physical observables with neural samplers
• Efficient generative modeling of protein sequences using simple autoregressive models
• Variational Neural Annealing
• Reinforcement Learning for Many-Body Ground-State Preparation Inspired by Counterdiabatic Driving
• Targeted free energy estimation via learned mappings
• Contracting Arbitrary Tensor Networks: General Approximate Algorithm and Applications in Graphical Models and Quantum Circuit Simulations
• Boosting Monte Carlo simulations of spin glasses using autoregressive neural networks
• A Perspective on Deep Learning for Molecular Modeling and Simulations
• Flow-based sampling for fermionic lattice field theories
• Gauge-equivariant flow models for sampling in lattice field theories with pseudofermions
• Enhanced Sampling in the Age of Machine Learning: Algorithms and Applications
• Drawing Phase Diagrams of Random Quantum Systems by Deep Learning the Wave Functions
• Riemannian geometry and automatic differentiation for optimization problems of quantum physics and quantum technologies
• Towards Novel Insights in Lattice Field Theory with Explainable Machine Learning
• Precision Machine Learning
• Hybrid convolutional neural network and PEPS wave functions for quantum many-particle states
• Unbiased Monte Carlo Cluster Updates with Autoregressive Neural Networks
• Roadmap on machine learning glassy dynamics
• U(1) symmetric recurrent neural networks for quantum state reconstruction
• Direct sampling of projected entangled-pair states
• Reducing Autocorrelation Times in Lattice Simulations with Generative Adversarial Networks
• Tensor networks for unsupervised machine learning
• Extending machine learning classification capabilities with histogram reweighting
• Scaling of neural-network quantum states for time evolution
• Machine Learning of consistent thermodynamic models using automatic differentiation
• Ab-initio study of interacting fermions at finite temperature with neural canonical transformation
• Deep Quantum Geometry of Matrices
• Variational autoencoder reconstruction of complex many-body physics
• Generative models for sampling and phase transition indication in spin systems
• Compressing deep neural networks by matrix product operators
• Machine-learning-assisted Monte Carlo fails at sampling computationally hard problems
• Neural Canonical Transformation with Symplectic Flows
• Reconstructing complex states of a 20-qubit quantum simulator
• Sampling with flows, diffusion and autoregressive neural networks: A spin-glass perspective
• Iterative Retraining of Quantum Spin Models Using Recurrent Neural Networks
• Deep Variational Free Energy Approach to Dense Hydrogen
• Calculating Renyi Entropies with Neural Autoregressive Quantum States
• A Bayesian generative neural network framework for epidemic inference problems
• Machine Learning of Thermodynamic Observables in the Presence of Mode Collapse
• Neural network approach to quasiparticle dispersions in doped antiferromagnets
• Continuous-mixture Autoregressive Networks for efficient variational calculation of many-body systems
• A learning by confusion approach to characterize phase transitions
• $m^{\ast }$ of two-dimensional electron gas: a neural canonical transformation study
• Analysis of autocorrelation times in Neural Markov Chain Monte Carlo simulations
• Replacing neural networks by optimal analytical predictors for the detection of phase transitions
• Mapping out phase diagrams with generative classifiers
• Solving Statistical Mechanics on Sparse Graphs with Feedback Set Variational Autoregressive Networks
• Thermal variational quantum simulation on a superconducting quantum processor
• Supplementing Recurrent Neural Networks with Annealing to Solve Combinatorial Optimization Problems
• Spin-1/2 kagome Heisenberg antiferromagnet: Machine learning discovery of the spinon pair density wave ground state
• Estimating the Euclidean quantum propagator with deep generative modeling of Feynman paths
• Neural canonical transformations for vibrational spectra of molecules
• Normalizing flows as an enhanced sampling method for atomistic supercooled liquids
• Neural networks in quantum many-body physics: a hands-on tutorial
• Autoregressive neural Slater-Jastrow ansatz for variational Monte Carlo simulation
• Reversible Upper Confidence Bound Algorithm to Generate Diverse Optimized Candidates
• Policy-guided Monte Carlo on general state spaces: Application to glass-forming mixtures
• Empirical Sample Complexity of Neural Network Mixed State Reconstruction
• The autoregressive neural network architecture of the Boltzmann distribution of pairwise interacting spins systems
• Accelerating equilibrium spin-glass simulations using quantum annealers via generative deep learning
• Numerically exact mimicking of quantum gas microscopy for interacting lattice fermions
• Creating equilibrium glassy states via random particle bonding
• Learning phase transitions from regression uncertainty: A new regression-based machine learning approach for automated detection of phases of matter
• Sampling U(1) gauge theory using a re-trainable conditional flow-based model
• Message Passing Variational Autoregressive Network for Solving Intractable Ising Models
• Variational Autoencoder Analysis of Ising Model Statistical Distributions and Phase Transitions
• Generative Neural Samplers for the Quantum Heisenberg Chain
• Computing Absolute Free Energy with Deep Generative Models
• Watch and learn -- a generalized approach for transferrable learning in deep neural networks via physical principles
• Network-Initialized Monte Carlo Based on Generative Neural Networks
• Solving Quantum Statistical Mechanics with Variational Autoregressive Networks and Quantum Circuits
• Training normalizing flows with computationally intensive target probability distributions
• Sparse Autoregressive Neural Networks for Classical Spin Systems
• Adaptive Monte Carlo augmented with normalizing flows
• Inference in conditioned dynamics through causality restoration
• Rényi entanglement entropy of spin chain with Generative Neural Networks
• Simulating first-order phase transition with hierarchical autoregressive networks
• Thermodynamics of the Ising model encoded in restricted Boltzmann machines
• Flow-Based Sampling for Entanglement Entropy and the Machine Learning of Defects
• Compressing Neural Networks Using Tensor Networks with Exponentially Fewer Variational Parameters
• Exploring explicit coarse-grained structure in artificial neural networks
• Efficient Bayesian Sampling Using Normalizing Flows to Assist Markov Chain Monte Carlo Methods
• Inverse Ising inference from high-temperature re-weighting of observations
• Performance of machine-learning-assisted Monte Carlo in sampling from simple statistical physics models
• Finding spin glass ground states through deep reinforcement learning
• Machine learning the Ising transition: A comparison between discriminative and generative approaches
• Efficient Optimization of Variational Autoregressive Networks with Natural Gradient
• Building imaginary-time thermal field theory with artificial neural networks
• Reinforcement Fine-Tuning for Materials Design
• Deep variational free energy prediction of dense hydrogen solid at 1200K
• Reconstruction of Pairwise Interactions using Energy-Based Models
• An Empirical Study of Quantum Dynamics as a Ground State Problem with Neural Quantum States
• Deep generative modelling of canonical ensemble with differentiable thermal properties
• Nearest-Neighbours Neural Network architecture for efficient sampling of statistical physics models
• Neural Canonical Transformations for Quantum Anharmonic Solids of Lithium
• Lattice Protein Folding with Variational Annealing
• The neural networks with tensor weights and emergent fermionic Wick rules in the large-width limit
• A method for quantifying the generalization capabilities of generative models for solving Ising models
• Nonequilibrium thermodynamics of self-supervised learning
• Omnisolver: an extensible interface to Ising spin glass solvers
• The statistical mechanics and machine learning of the $\alpha$-Rényi ensemble
• Integrating Neural Networks and Tensor Networks for Computing Free Energy
• Worldline algorithm by oracle-guided variational autoregressive network
• Global sampling of Feynman's diagrams through Normalizing Flow
• NeuMC -- a package for neural sampling for lattice field theories
• Hierarchical autoregressive neural networks in three-dimensional statistical system
• Ratio Divergence Learning Using Target Energy in Restricted Boltzmann Machines: Beyond Kullback--Leibler Divergence Learning
• Overcoming barriers to scalability in variational quantum Monte Carlo