Adaptive Monte Carlo augmented with normalizing flows
arXiv:2105.12603 · doi:10.1073/pnas.2109420119
Abstract
Many problems in the physical sciences, machine learning, and statistical inference necessitate sampling from a high-dimensional, multi-modal probability distribution. Markov Chain Monte Carlo (MCMC) algorithms, the ubiquitous tool for this task, typically rely on random local updates to propagate configurations of a given system in a way that ensures that generated configurations will be distributed according to a target probability distribution asymptotically. In high-dimensional settings with multiple relevant metastable basins, local approaches require either immense computational effort or intricately designed importance sampling strategies to capture information about, for example, the relative populations of such basins. Here we analyze an adaptive MCMC which augments MCMC sampling with nonlocal transition kernels parameterized with generative models known as normalizing flows. We focus on a setting where there is no preexisting data, as is commonly the case for problems in which MCMC is used. Our method uses: (i) a MCMC strategy that blends local moves obtained from any standard transition kernel with those from a generative model to accelerate the sampling and (ii) the data generated this way to adapt the generative model and improve its efficacy in the MCMC algorithm. We provide a theoretical analysis of the convergence properties of this algorithm, and investigate numerically its efficiency, in particular in terms of its propensity to equilibrate fast between metastable modes whose rough location is known \textit{a~priori} but respective probability weight is not. We show that our algorithm can sample effectively across large free energy barriers, providing dramatic accelerations relative to traditional MCMC algorithms.
References in corpus (33)
- Auto-Encoding Variational Bayes
- NICE: Non-linear Independent Components Estimation
- Variational Inference with Normalizing Flows
- Score-Based Generative Modeling through Stochastic Differential Equations
- Normalizing Flows for Probabilistic Modeling and Inference
- MADE: Masked Autoencoder for Distribution Estimation
- On the ergodicity properties of some adaptive MCMC algorithms
- 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
- An Introduction to Stochastic PDEs
- Estimation of Thermodynamic Observables in Lattice Field Theories with Deep Generative Models
- The dissipation-time uncertainty relation
- Asymptotically unbiased estimation of physical observables with neural samplers
- Transport map accelerated Markov chain Monte Carlo
- Reaction Pathways Based on the Gradient of the Mean First-Passage Time
- Normalizing Flows on Tori and Spheres
- A-NICE-MC: Adversarial Training for MCMC
- Auto-Encoding Sequential Monte Carlo
- Equivariant Flows: Exact Likelihood Generative Learning for Symmetric Densities
- Boosting Monte Carlo simulations of spin glasses using autoregressive neural networks
- Efficient Modelling of Trivializing Maps for Lattice Theory Using Normalizing Flows: A First Look at Scalability
- Equivariant Flows: sampling configurations for multi-body systems with symmetric energies
- Dissipation bounds the amplification of transition rates far from equilibrium
- On nonlinear Markov chain Monte Carlo
- Kernel Adaptive Metropolis-Hastings
- An Eyring-Kramers law for the stochastic Allen-Cahn equation in dimension two
- Generalizing Hamiltonian Monte Carlo with Neural Networks
- Accelerating Langevin Sampling with Birth-death
- Learning Model Reparametrizations: Implicit Variational Inference by Fitting MCMC distributions
- Markovian Score Climbing: Variational Inference with KL(p||q)
- Self-Supervised Learning of Generative Spin-Glasses with Normalizing Flows
- Temperature-steerable flows