Hamiltonian Monte Carlo Without Detailed Balance

We present a method for performing Hamiltonian Monte Carlo that largely eliminates sample rejection for typical hyperparameters. In situations that would normally lead to rejection, instead a longer trajectory is computed until a new state is reached that can be accepted. This is achieved using Markov chain transitions that satisfy the fixed point equation, but do not satisfy detailed balance. The resulting algorithm significantly suppresses the random walk behavior and wasted function evaluations that are typically the consequence of update rejection. We demonstrate a greater than factor of two improvement in mixing time on three test problems. We release the source code as Python and MATLAB packages.

• Publication year

  2014

• Venue

  International Conference on Machine Learning

• Publication date

  2014-06-21

• Fields of study

  Mathematics, Computer Science

• Identifiers
  arXiv 1409.5191
• External record

  Open on Semantic Scholar

• Source metadata

  Semantic Scholar

• No claims are published for this paper.

• No concepts are published for this paper.

• Extra Chance Hybrid Monte Carlo

  2014cited by this paper
• Extra Chance Generalized Hybrid Monte Carlo

  2014cited by this paper
• Non-reversible Metropolis-Hastings

  2014cited by this paper
• An irreversible Markov-chain Monte Carlo method with skew detailed balance conditions

  2013cited by this paper
• Adaptive Hamiltonian and Riemann Manifold Monte Carlo

  2013cited by this paper
• Enhanced exchange algorithm without detailed balance condition for replica exchange method.

  2013cited by this paper
• Violation of detailed balance accelerates relaxation.

  2013cited by this paper
• Langevin dynamics neglecting detailed balance condition.

  2013cited by this paper
• Hamiltonian Annealed Importance Sampling for partition function estimation

  2012cited by this paper
• Hamiltonian Monte Carlo with Reduced Momentum Flips

  2012cited by this paper
• The No-U-turn sampler: adaptively setting path lengths in Hamiltonian Monte Carlo

  2011cited by this paper
• Riemann manifold Langevin and Hamiltonian Monte Carlo methods

  2011cited by this paper
• Probabilistic Inference Using Markov Chain Monte Carlo Methods

  2011cited by this paper
• Building a better probabilistic model of images by factorization

  2011influential reference
• Quasi-Newton Methods for Markov Chain Monte Carlo

  2011cited by this paper
• Hybrid Monte Carlo on Hilbert spaces

  2011cited by this paper
• MCMC Using Hamiltonian Dynamics

  2011cited by this paper
• Split Hamiltonian Monte Carlo

  2011cited by this paper
• Markov chain Monte Carlo method without detailed balance.

  2010cited by this paper
• Improving the Asymptotic Performance of Markov Chain Monte-Carlo by Inserting Vortices

  2010cited by this paper
• Phase Coupling Estimation from Multivariate Phase Statistics

  2009cited by this paper
• Shadow hybrid Monte Carlo: an efficient propagator in phase space of macromolecules

  2004cited by this paper
• Information Theory, Inference, and Learning Algorithms

  2004cited by this paper
• Geometric numerical integration illustrated by the Störmer–Verlet method

  2003cited by this paper
• Acta Numerica 2003: Geometric numerical integration illustrated by the Störmer–Verlet method

  2003cited by this paper
• ANALYSIS OF A NONREVERSIBLE MARKOV CHAIN SAMPLER

  2000cited by this paper
• A Direct Approach to Conformational Dynamics Based on Hybrid Monte Carlo

  1999cited by this paper
• Lifting Markov chains to speed up mixing

  1999cited by this paper
• An improved acceptance procedure for the hybrid Monte Carlo algorithm

  1992cited by this paper
• A generalized guided Monte Carlo algorithm

  1991cited by this paper
• Acceptances and autocorrelations in hybrid Monte Carlo

  1991cited by this paper
• Hybrid Monte Carlo

  1988cited by this paper
• Monte Carlo Sampling Methods Using Markov Chains and Their Applications

  1970influential reference

• Sampling From Multiscale Densities With Delayed Rejection Generalized Hamiltonian Monte Carlo

  2024cites this paper
• Graph-accelerated Markov Chain Monte Carlo using Approximate Samples

  2024cites this paper
• Natural gradient enables fast sampling in spiking neural networks

  2023cites this paper
• Principled Gradient-based Markov Chain Monte Carlo for Text Generation

  2023cites this paper
• Learning and Inference in Sparse Coding Models With Langevin Dynamics

  2022cites this paper
• Shadow Magnetic Hamiltonian Monte Carlo

  2022influential citation
• Metropolis Adjusted Langevin Trajectories: a robust alternative to Hamiltonian Monte Carlo

  2022cites this paper
• Geometric Methods for Sampling, Optimisation, Inference and Adaptive Agents

  2022cites this paper
• Toward Unlimited Self-Learning Monte Carlo with Annealing Process Using VAE's Implicit Isometricity

  2022cites this paper
• Toward Unlimited Self-Learning MCMC with Parallel Adaptive Annealing

  2022cites this paper
• Nonparametric Involutive Markov Chain Monte Carlo

  2022influential citation
• Peskun–Tierney ordering for Markovian Monte Carlo: Beyond the reversible scenario

  2021cites this paper
• Hamiltonian Dynamics with Non-Newtonian Momentum for Rapid Sampling

  2021cites this paper
• Entropy-based adaptive Hamiltonian Monte Carlo

  2021cites this paper
• Delayed rejection Hamiltonian Monte Carlo for sampling multiscale distributions

  2021influential citation
• A Unifying and Canonical Description of Measure-Preserving Diffusions

  2021cites this paper
• Shadow Manifold Hamiltonian Monte Carlo

  2021cites this paper
• Exact Langevin Dynamics with Stochastic Gradients

  2021cites this paper
• A general perspective on the Metropolis-Hastings kernel

  2020cites this paper
• Nonreversible MCMC from conditional invertible transforms: a complete recipe with convergence guarantees

  2020influential citation
• Exploring Probability Measures with Markov Processes

  2020cites this paper
• Involutive MCMC: a Unifying Framework

  2020influential citation
• A Neural Network MCMC Sampler That Maximizes Proposal Entropy

  2020cites this paper
• Neural Langevin Dynamical Sampling

  2020cites this paper
• Application of Exchanging Monte Carlo Method to Sample Deep Boltzmann Machines

  2020cites this paper
• Markov chain Monte Carlo algorithms with sequential proposals

  2019cites this paper
• Bayesian Neural Networks at Finite Temperature

  2019cites this paper
• Status and future perspectives for lattice gauge theory calculations to the exascale and beyond

  2019cites this paper
• Differentiable Visual Computing

  2019cites this paper
• Peskun-Tierney ordering for Markov chain and process Monte Carlo: beyond the reversible scenario

  2019cites this paper
• Cortical-like dynamics in recurrent circuits optimized for sampling-based probabilistic inference

  2019cites this paper
• Dynamical Sampling with Langevin Normalization Flows

  2019cites this paper
• Status and future perspectives for lattice gauge theory calculations to the exascale and beyond

  2019cites this paper
• SLIND: Identifying Stable Links in Online Social Networks

  2018cites this paper
• Stochastic approximation Hamiltonian Monte Carlo

  2018cites this paper
• Bayesian analysis of continuous time Markov chains with applications to phylogenetics

  2018cites this paper
• Binding Modes of Ligands Using Enhanced Sampling (BLUES): Rapid Decorrelation of Ligand Binding Modes via Nonequilibrium Candidate Monte Carlo.

  2018cites this paper
• On Link Stability Detection for Online Social Networks

  2018cites this paper
• Detecting Relational States in Online Social Networks

  2018cites this paper
• Forward Event-Chain Monte Carlo: Fast Sampling by Randomness Control in Irreversible Markov Chains

  2017cites this paper
• Forward Event-Chain Monte Carlo: a general rejection-free and irreversible Markov chain simulation method

  2017cites this paper
• Modified Hamiltonian Monte Carlo for Bayesian inference

  2017influential citation
• Binding Modes of Ligands Using Enhanced Sampling (BLUES): Rapid Decorrelation of Ligand Binding Modes Using Nonequilibrium Candidate Monte Carlo

  2017cites this paper
• Generalizing Hamiltonian Monte Carlo with Neural Networks

  2017influential citation
• Bayesian techniques for inverse uncertainty quantification

  2017cites this paper
• Learning and Inferring Representations of Data in Neural Networks

  2017cites this paper
• General and Efficient Bayesian Computation through Hamiltonian Monte Carlo Extensions

  2017cites this paper
• Geodesic Lagrangian Monte Carlo over the space of positive definite matrices: with application to Bayesian spectral density estimation

  2016cites this paper
• Enhancing sampling in computational statistics using modified hamiltonians

  2016cites this paper
• On the geometric ergodicity of Hamiltonian Monte Carlo

  2016cites this paper
• Geometrically Tempered Hamiltonian Monte Carlo

  2016cites this paper
• A Common Derivation for Markov Chain Monte Carlo Algorithms with Tractable and Intractable Targets

  2016cites this paper
• Identifying the Optimal Integration Time in Hamiltonian Monte Carlo

  2016cites this paper
• Magnetic Hamiltonian Monte Carlo

  2016cites this paper
• Laplacian Hamiltonian Monte Carlo

  2016cites this paper
• Bayesian Analysis of Continuous Time Markov Chains with Application to Phylogenetic Modelling

  2016cites this paper
• Gradient-based MCMC samplers for dynamic causal modelling

  2016cites this paper
• Hybrid Monte Carlo with Chaotic Mixing

  2016cites this paper
• A Markov Jump Process for More Efficient Hamiltonian Monte Carlo

  2015cites this paper
• Exponential Integration for Hamiltonian Monte Carlo

  2015cites this paper
• An investigation of gene-gene interactions in dose-response studies with Bayesian nonparametrics

  2015cites this paper
• Extra Chance Generalized Hybrid Monte Carlo

  2014influential citation
• The Geometric Foundations of Hamiltonian Monte Carlo

  2014cites this paper
• Extra Chance Hybrid Monte Carlo

  2014influential citation
• Compressible generalized hybrid Monte Carlo.

  2014cites this paper
• F OLLOW H AMILTONIAN L EADER : A N E FFICIENT E NERGY -G UIDED S AMPLING M ETHOD

  year unknowncites this paper