Asymptotically unbiased estimation of physical observables with neural samplers

Abstract

We propose a general framework for the estimation of observables with generative neural samplers focusing on modern deep generative neural networks that provide an exact sampling probability. In this framework, we present asymptotically unbiased estimators for generic observables, including those that explicitly depend on the partition function such as free energy or entropy, and derive corresponding variance estimators. We demonstrate their practical applicability by numerical experiments for the 2d Ising model which highlight the superiority over existing methods. Our approach greatly enhances the applicability of generative neural samplers to real-world physical systems.

arXiv comment5 figures

References in corpus (17)

• Solving the Quantum Many-Body Problem with Artificial Neural Networks
• Pixel Recurrent Neural Networks
• Monte Carlo errors with less errors
• Deep autoregressive models for the efficient variational simulation of many-body quantum systems
• Learning Thermodynamics with Boltzmann Machines
• Accelerate Monte Carlo Simulations with Restricted Boltzmann Machines
• Solving Statistical Mechanics Using Variational Autoregressive Networks
• Flow-based generative models for Markov chain Monte Carlo in lattice field theory
• Mutual Information, Neural Networks and the Renormalization Group
• Neural Network Renormalization Group
• CosmoGAN: creating high-fidelity weak lensing convergence maps using Generative Adversarial Networks
• Regressive and generative neural networks for scalar field theory
• Deep Learning the Ising Model Near Criticality
• Discrete Flows: Invertible Generative Models of Discrete Data
• Simulating the Ising Model with a Deep Convolutional Generative Adversarial Network
• Towards meaningful physics from generative models
• Comment on "Solving Statistical Mechanics Using VANs": Introducing saVANt - VANs Enhanced by Importance and MCMC Sampling

Cited by in corpus (63)

• Machine Learning for Quantum Matter
• Estimation of Thermodynamic Observables in Lattice Field Theories with Deep Generative Models
• Targeted free energy estimation via learned mappings
• Equivariant Flows: Exact Likelihood Generative Learning for Symmetric Densities
• Flow-matching -- efficient coarse-graining of molecular dynamics without forces
• Boosting Monte Carlo simulations of spin glasses using autoregressive neural networks
• Flow-based sampling for fermionic lattice field theories
• Gauge Invariant and Anyonic Symmetric Transformer and RNN Quantum States for Quantum Lattice Models
• Gauge-equivariant flow models for sampling in lattice field theories with pseudofermions
• Normalizing flows for atomic solids
• Towards Novel Insights in Lattice Field Theory with Explainable Machine Learning
• Mapping distinct phase transitions to a neural network
• Unbiased Monte Carlo Cluster Updates with Autoregressive Neural Networks
• Learning Lattice Quantum Field Theories with Equivariant Continuous Flows
• Reducing Autocorrelation Times in Lattice Simulations with Generative Adversarial Networks
• Tensor networks for unsupervised machine learning
• Extending machine learning classification capabilities with histogram reweighting
• Learning Trivializing Gradient Flows for Lattice Gauge Theories
• Normalizing Flows and the Real-Time Sign Problem
• Generative models for sampling and phase transition indication in spin systems
• Deep Variational Free Energy Approach to Dense Hydrogen
• Machine Learning of Thermodynamic Observables in the Presence of Mode Collapse
• Free energy calculation of crystalline solids using normalizing flow
• Flow-based density of states for complex actions
• Stochastic Normalizing Flows
• Neural Network flows of low q-state Potts and clock Models
• Analysis of autocorrelation times in Neural Markov Chain Monte Carlo simulations
• Sampling the lattice Nambu-Goto string using Continuous Normalizing Flows
• Neural Thermodynamic Integration: Free Energies from Energy-based Diffusion Models
• Hierarchical autoregressive neural networks for statistical systems
• Numerical determination of the width and shape of the effective string using Stochastic Normalizing Flows
• Accelerating equilibrium spin-glass simulations using quantum annealers via generative deep learning
• The autoregressive neural network architecture of the Boltzmann distribution of pairwise interacting spins systems
• Infinite Neural Network Quantum States: Entanglement and Training Dynamics
• Sampling U(1) gauge theory using a re-trainable conditional flow-based model
• On learning higher-order cumulants in diffusion models
• Generative Neural Samplers for the Quantum Heisenberg Chain
• Bounding free energy difference with flow matching
• Adaptive Monte Carlo augmented with normalizing flows
• Sparse Autoregressive Neural Networks for Classical Spin Systems
• Path Integral Sampler: a stochastic control approach for sampling
• Training normalizing flows with computationally intensive target probability distributions
• Watch and learn -- a generalized approach for transferrable learning in deep neural networks via physical principles
• Rényi entanglement entropy of spin chain with Generative Neural Networks
• Thermodynamics of the Ising model encoded in restricted Boltzmann machines
• Flow-Based Sampling for Entanglement Entropy and the Machine Learning of Defects
• Simulating first-order phase transition with hierarchical autoregressive networks
• Temperature-steerable flows
• Berezinskii-Kosterlitz-Thouless transition from Neural Network Flows
• Scaling Up Machine Learning For Quantum Field Theory with Equivariant Continuous Flows
• Application of the Variational Autoencoder to Detect the Critical Points of the Anisotropic Ising Model
• Annealed Flow Transport Monte Carlo
• Efficient Bayesian Sampling Using Normalizing Flows to Assist Markov Chain Monte Carlo Methods
• Machine learning the Ising transition: A comparison between discriminative and generative approaches
• Efficient Optimization of Variational Autoregressive Networks with Natural Gradient
• Scaling flow-based approaches for topology sampling in $\mathrm{SU}(3)$ gauge theory
• Reducing rejection exponentially improves Markov chain Monte Carlo sampling
• Neural-network-assisted Monte Carlo sampling trained by Quantum Approximate Optimization Algorithm
• Deep generative modelling of canonical ensemble with differentiable thermal properties
• NeuMC -- a package for neural sampling for lattice field theories
• Hierarchical autoregressive neural networks in three-dimensional statistical system
• Neural Mode Jump Monte Carlo
• Applications of flow models to the generation of correlated lattice QCD ensembles