Targeted free energy estimation via learned mappings
arXiv:2002.04913 · doi:10.1063/5.0018903
Abstract
Free energy perturbation (FEP) was proposed by Zwanzig more than six decades ago as a method to estimate free energy differences, and has since inspired a huge body of related methods that use it as an integral building block. Being an importance sampling based estimator, however, FEP suffers from a severe limitation: the requirement of sufficient overlap between distributions. One strategy to mitigate this problem, called Targeted Free Energy Perturbation, uses a high-dimensional mapping in configuration space to increase overlap of the underlying distributions. Despite its potential, this method has attracted only limited attention due to the formidable challenge of formulating a tractable mapping. Here, we cast Targeted FEP as a machine learning problem in which the mapping is parameterized as a neural network that is optimized so as to increase overlap. We develop a new model architecture that respects permutational and periodic symmetries often encountered in atomistic simulations and test our method on a fully-periodic solvation system. We demonstrate that our method leads to a substantial variance reduction in free energy estimates when compared against baselines, without requiring any additional data.
arXiv commentAdded figure 3, added data augmentation for octahedral symmetries, updated experimental results and revised text (11 pages, 6 figures)
References in corpus (14)
- Statistically optimal analysis of samples from multiple equilibrium states
- How van der Waals interactions determine the unique properties of water
- Rare events and the convergence of exponentially averaged work values
- Solving Statistical Mechanics Using Variational Autoregressive Networks
- Flow-based generative models for Markov chain Monte Carlo in lattice field theory
- Adaptive Importance Sampling in General Mixture Classes
- Neural Network Renormalization Group
- Asymptotically unbiased estimation of physical observables with neural samplers
- Targeted free energy perturbation
- Optimal estimates of free energies from multi-state nonequilibrium work data
- Estimating Ratios of Normalizing Constants Using Linked Importance Sampling
- Using bijective maps to improve free energy estimates
- Controlling cargo trafficking in multicomponent membranes
- Residual and Destroyed Accessible Information after Measurements
Cited by in corpus (26)
- Normalizing Flows: An Introduction and Review of Current Methods
- Flow-matching -- efficient coarse-graining of molecular dynamics without forces
- Perspectives on adaptive dynamical systems
- Enhanced Sampling in the Age of Machine Learning: Algorithms and Applications
- Normalizing flows for atomic solids
- Targeted free energy perturbation revisited: Accurate free energies from mapped reference potentials
- Contrastive Learning of Coarse-Grained Force Fields
- Skipping the Replica Exchange Ladder with Normalizing Flows
- Gravitational wave populations and cosmology with neural posterior estimation
- Ab-initio study of interacting fermions at finite temperature with neural canonical transformation
- Deep Variational Free Energy Approach to Dense Hydrogen
- Multimap targeted free energy estimation
- of two-dimensional electron gas: a neural canonical transformation study
- Neural Thermodynamic Integration: Free Energies from Energy-based Diffusion Models
- Efficient mapping of phase diagrams with conditional Boltzmann Generators
- Space Group Informed Transformer for Crystalline Materials Generation
- Bounding free energy difference with flow matching
- Normalizing flows for microscopic many-body calculations: an application to the nuclear equation of state
- Solvation Free Energies from Neural Thermodynamic Integration
- Scaling of Stochastic Normalizing Flows in lattice gauge theory
- Learned Mappings for Targeted Free Energy Perturbation between Peptide Conformations
- Energy-Based Coarse-Graining in Molecular Dynamics: A Flow-Based Framework without Data
- Fokker-Planck Score Learning: Efficient Free-Energy Estimation under Periodic Boundary Conditions
- Sampling the Liquid-Gas Critical Point with Boltzmann Generators
- Nonadiabatic force matching for alchemical free-energy estimation
- Neural Canonical Transformation for the Spectra of Fluxional Molecule CH5+