The Penalty Method for Random Walks with Uncertain Energies
arXiv:physics/9812035 · doi:10.1063/1.478034
Abstract
We generalize the Metropolis et al. random walk algorithm to the situation where the energy is noisy and can only be estimated. Two possible applications are for long range potentials and for mixed quantum-classical simulations. If the noise is normally distributed we are able to modify the acceptance probability by applying a penalty to the energy difference and thereby achieve exact sampling even with very strong noise. When one has to estimate the variance we have an approximate formula, good in the limit of large number of independent estimates. We argue that the penalty method is nearly optimal. We also adapt an existing method by Kennedy and Kuti and compare to the penalty method on a one dimensional double well.
17 pages, 7 figures, accepted to Journal of Chemical Physics; Corrected swap of Figures 2 and 3. Added Figure 6
References in corpus (2)
Cited by in corpus (48)
- Correlated geminal wave function for molecules: an efficient resonating valence bond approach
- On the efficiency of pseudo-marginal random walk Metropolis algorithms
- Speeding Up MCMC by Efficient Data Subsampling
- Stable liquid Hydrogen at high pressure by a novel ab-initio molecular dynamics
- On Markov chain Monte Carlo methods for tall data
- Algorithmic differentiation and the calculation of forces by quantum Monte Carlo
- GPS-ABC: Gaussian Process Surrogate Approximate Bayesian Computation
- First principles simulations of dense hydrogen
- Coupled Electron Ion Monte Carlo Calculations of Dense Metallic Hydrogen
- Equation of state of metallic hydrogen from Coupled Electron-Ion Monte Carlo simulations
- Ab-initio molecular dynamics simulation of liquid water by Quantum Monte Carlo
- Quantum Monte Carlo Simulation of the High-Pressure Molecular-Atomic Crossover in Fluid Hydrogen
- The Coupled Electron-Ion Monte Carlo Method
- Near-optimal protocols in complex nonequilibrium transformations
- Characterisation of the probabilistic travelling salesman problem
- Speeding Up MCMC by Delayed Acceptance and Data Subsampling
- Finite temperature electronic simulations beyond the Born-Oppenheimer approximation
- Molecular-Atomic Transition in the Deuterium Hugoniot with Coupled Electron Ion Monte Carlo
- Linear scaling electronic structure calculations and accurate sampling with noisy forces
- Quantum computing for chemistry and physics applications from a Monte Carlo perspective
- On nonnegative unbiased estimators
- Coupled MCMC with a randomized acceptance probability
- Deep Variational Free Energy Approach to Dense Hydrogen
- Monte Carlo sampling for stochastic weight functions
- Modeling of many-body interactions between elastic spheres through symmetry functions
- Benchmarking vdW-DF first principle predictions against Coupled Electron-Ion Monte Carlo for high pressure liquid hydrogen
- Hamiltonian Monte Carlo with Energy Conserving Subsampling
- Stochastic Annealing
- A general perspective on the Metropolis-Hastings kernel
- Peskun-Tierney ordering for Markov chain and process Monte Carlo: beyond the reversible scenario
- Establishing some order amongst exact approximations of MCMCs
- Efficient Langevin dynamics for "noisy" forces
- When you can't count, sample! Computable entropies beyond equilibrium from basin volumes
- Scalable Monte Carlo inference for state-space models
- Coupled Electron Ion Monte Carlo Calculations of Atomic Hydrogen
- A Simple Algorithm for Scalable Monte Carlo Inference
- On the Fermion Sign Problem in Imaginary-Time Projection Continuum Quantum Monte Carlo with Local Interaction
- Implementing advanced trial wave functions in fermion quantum Monte Carlo via stochastic sampling
- Estimating Reciprocal Partition Functions to Enable Design Space Sampling
- Monte Carlo methods: Application to hydrogen gas and hard spheres
- ALAAMEE: Open-source software for fitting autologistic actor attribute models
- Thermodynamic Integration Methods, Infinite Swapping and the Calculation of Generalized Averages
- Subsampling MCMC - An introduction for the survey statistician
- Differentially Private Hamiltonian Monte Carlo
- Data Subsampling for Bayesian Neural Networks
- Probabilistic Programs with Stochastic Conditioning
- Uncertainty in AI-driven Monte Carlo simulations
- MEXIT: Maximal un-coupling times for stochastic processes