State Predictive Information Bottleneck

Abstract

The ability to make sense of the massive amounts of high-dimensional data generated from molecular dynamics (MD) simulations is heavily dependent on the knowledge of a low dimensional manifold (parameterized by a reaction coordinate or RC) that typically distinguishes between relevant metastable states and which captures the relevant slow dynamics of interest. Methods based on machine learning and artificial intelligence have been proposed over the years to deal with learning such low-dimensional manifolds, but they are often criticized for a disconnect from more traditional and physically interpretable approaches. To deal with such concerns, in this work, we propose a deep learning based State Predictive Information Bottleneck (SPIB) approach to learn the RC from high dimensional molecular simulation trajectories. We demonstrate analytically and numerically how the RC learnt in this approach is deeply connected to the committor in chemical physics, and can be used to accurately identify transition states. A crucial hyperparameter in this approach is the time-delay, or how far into the future the algorithm should make predictions about. Through careful comparisons for benchmark systems, we demonstrate that this hyperparameter choice gives useful control over how coarse-grained we want the metastable state classification of the system to be. We thus believe that this work represents a step forward in systematic application of deep learning based ideas to molecular simulations in a way that bridges the gap between artificial intelligence and traditional chemical physics.

11 pages, 13 figures

References in corpus (10)

• Adam: A Method for Stochastic Optimization
• Canonical sampling through velocity-rescaling
• PLUMED 2: New feathers for an old bird
• Accurate sampling using Langevin dynamics
• Caliber based spectral gap optimization of order parameters (SGOOP) for sampling complex molecular systems
• Variational Encoding of Complex Dynamics
• Galerkin Approximation of Dynamical Quantities using Trajectory Data
• Learning reaction coordinates via cross-entropy minimization: Application to alanine dipeptide
• VAE with a VampPrior
• TherML: Thermodynamics of Machine Learning

Cited by in corpus (17)

• JARVIS-Leaderboard: A Large Scale Benchmark of Materials Design Methods
• Explaining reaction coordinates of alanine dipeptide isomerization obtained from deep neural networks using Explainable Artificial Intelligence (XAI)
• Enhanced Sampling in the Age of Machine Learning: Algorithms and Applications
• Driving and characterizing nucleation of urea and glycine polymorphs in water
• From Free-Energy Profiles to Activation Free Energies
• Recent advances in describing and driving crystal nucleation using machine learning and artificial intelligence
• Multidimensional minimum-work control of a 2D Ising model
• Is the Local Ion Density Sufficient to Drive NaCl Nucleation from the Melt and Aqueous Solution?
• Atomic scale insights into NaCl nucleation in nanoconfined environments
• Differentiable Physics: A Position Piece
• Flow Matching for Optimal Reaction Coordinates of Biomolecular System
• Spectral Map for Slow Collective Variables, Markovian Dynamics, and Transition State Ensembles
• Information Thermodynamics of the Transition-Path Ensemble
• Thermodynamically Optimized Machine-learned Reaction Coordinates for Hydrophobic Ligand Dissociation
• Unveiling interatomic distances influencing the reaction coordinates in alanine dipeptide isomerization: An explainable deep learning approach
• Recovering Hidden Degrees of Freedom Using Gaussian Processes
• Deep learning of committor and explainable artificial intelligence analysis for identifying reaction coordinates