Efficient Optimization of Variational Autoregressive Networks with Natural Gradient

Abstract

Estimating free energy is a fundamental problem in statistical mechanics. Recently, machine-learning-based methods, particularly the variational autoregressive networks (VANs) have been proposed to minimize variational free energy and to approximate the Boltzmann distribution. VAN enjoys notable advantages, including the exact computation of the normalized joint distribution and fast sampling, which are critical features often missing in Markov chain Monte Carlo algorithms. However, VAN also faces significant computational challenges. These include difficulties in the optimization of variational free energy in a complicated parameter space and slow convergence of learning. In this work, we introduce an optimization technique based on natural gradients to the VAN framework, namely ng-VAN, to enhance the learning efficiency and accuracy of the conventional VAN. The method has computational complexity cubic in the batch size rather than in the number of model parameters, hence it can be efficiently implemented for a large VAN model. We carried out extensive numerical experiments on the Sherrington-Kirkpatrick model, spin glasses on random graphs, and the two-dimensional Ising model. Our results indicate that compared with the conventional VAN, ng-VAN significantly improves the accuracy in estimating free energy and converges much faster with shorter learning time. This allows extending the VAN framework's applicability to challenging statistical mechanics problems that were previously not accessible.

11 pages, 8 figures

References in corpus (22)

• Ising formulations of many NP problems
• Solving the Quantum Many-Body Problem with Artificial Neural Networks
• Unsupervised Generative Modeling Using Matrix Product States
• Deep autoregressive models for the efficient variational simulation of many-body quantum systems
• Solving Statistical Mechanics Using Variational Autoregressive Networks
• Hyper-optimized tensor network contraction
• Solving the sampling problem of the Sycamore quantum circuits
• Tree Tensor Networks for Generative Modeling
• Estimation of Thermodynamic Observables in Lattice Field Theories with Deep Generative Models
• Asymptotically unbiased estimation of physical observables with neural samplers
• Empowering deep neural quantum states through efficient optimization
• A simple linear algebra identity to optimize Large-Scale Neural Network Quantum States
• Boosting Monte Carlo simulations of spin glasses using autoregressive neural networks
• Tropical Tensor Network for Ground States of Spin Glasses
• Tensor networks for unsupervised machine learning
• Dynamics with autoregressive neural quantum states: application to critical quench dynamics
• Machine-learning-assisted Monte Carlo fails at sampling computationally hard problems
• jVMC: Versatile and performant variational Monte Carlo leveraging automated differentiation and GPU acceleration
• Solving Statistical Mechanics on Sparse Graphs with Feedback Set Variational Autoregressive Networks
• The autoregressive neural network architecture of the Boltzmann distribution of pairwise interacting spins systems
• Second-order optimisation strategies for neural network quantum states
• Sparse Autoregressive Neural Networks for Classical Spin Systems

Cited by in corpus (1)

• Large-scale portfolio optimization with variational neural annealing