Forces are not Enough: Benchmark and Critical Evaluation for Machine Learning Force Fields with Molecular Simulations

Abstract

Molecular dynamics (MD) simulation techniques are widely used for various natural science applications. Increasingly, machine learning (ML) force field (FF) models begin to replace ab-initio simulations by predicting forces directly from atomic structures. Despite significant progress in this area, such techniques are primarily benchmarked by their force/energy prediction errors, even though the practical use case would be to produce realistic MD trajectories. We aim to fill this gap by introducing a novel benchmark suite for learned MD simulation. We curate representative MD systems, including water, organic molecules, a peptide, and materials, and design evaluation metrics corresponding to the scientific objectives of respective systems. We benchmark a collection of state-of-the-art (SOTA) ML FF models and illustrate, in particular, how the commonly benchmarked force accuracy is not well aligned with relevant simulation metrics. We demonstrate when and how selected SOTA methods fail, along with offering directions for further improvement. Specifically, we identify stability as a key metric for ML models to improve. Our benchmark suite comes with a comprehensive open-source codebase for training and simulation with ML FFs to facilitate future work.

31 pages, 18 figures

Cited by in corpus (28)

• Evaluation of the MACE Force Field Architecture: from Medicinal Chemistry to Materials Science
• How to validate machine-learned interatomic potentials
• Machine Learning Coarse-Grained Potentials of Protein Thermodynamics
• Single-model uncertainty quantification in neural network potentials does not consistently outperform model ensembles
• Discrepancies and Error Evaluation Metrics for Machine Learning Interatomic Potentials
• Deep Ensembles vs. Committees for Uncertainty Estimation in Neural-Network Force Fields: Comparison and Application to Active Learning
• Machine-learning-accelerated simulations to enable automatic surface reconstruction
• Machine-learned molecular mechanics force field for the simulation of protein-ligand systems and beyond
• Force-Field-Enhanced Neural Network Interactions: from Local Equivariant Embedding to Atom-in-Molecule properties and long-range effects
• On the design space between molecular mechanics and machine learning force fields
• Denoise Pretraining on Nonequilibrium Molecules for Accurate and Transferable Neural Potentials
• Machine Learned Interatomic Potential for Dispersion Strengthened Plasma Facing Components
• Geometry-enhanced Pre-training on Interatomic Potentials
• Neural Network Potential with Multi-Resolution Approach Enables Accurate Prediction of Reaction Free Energies in Solution
• Data efficiency and extrapolation trends in neural network interatomic potentials
• MGNN: Moment Graph Neural Network for Universal Molecular Potentials
• Analyzing Atomic Interactions in Molecules as Learned by Neural Networks
• Implicit Solvent Approach Based on Generalised Born and Transferable Graph Neural Networks for Molecular Dynamics Simulations
• Probing the effects of broken symmetries in machine learning
• Towards Symbolic XAI -- Explanation Through Human Understandable Logical Relationships Between Features
• Does Hessian Data Improve the Performance of Machine Learning Potentials?
• Interpolating many-body wave functions for accelerated molecular dynamics on the near-exact electronic surface
• Enhanced sampling of robust molecular datasets with uncertainty-based collective variables
• Investigating Ionic Diffusivity in Amorphous Solid Electrolytes using Machine Learned Interatomic Potentials
• Scaling Graph Neural Networks to Large Proteins
• Bayesian Hierarchical Models for Quantitative Estimates for Performance metrics applied to Saddle Search Algorithms
• Learning atomic forces from uncertainty-calibrated adversarial attacks
• Knowledge Distillation of a Protein Language Model Yields a Foundational Implicit Solvent Model