Explore the application of machine learning in predicting energies from electron densities for reactive molecular dynamics in this 28-minute talk by NYU's Leslie Vogt-Maranto. Delve into topics such as generating free energy surfaces, calculating observables, and machine learning molecular energies. Examine the use of Density Functional Theory (DFT) in energy calculations via electron densities, and learn about machine learning approaches for DFT in molecules, hydrogen, and various compounds. Discover sampling strategies for training geometries and investigate the overlap of test and training data. Gain insights into future directions and ongoing developments in this cutting-edge field of computational chemistry and physics.
Predicting Energies from Electron Densities - Machine Learning for Reactive Molecular Dynamics
APS Physics via YouTube
Overview
Syllabus
Intro
Motivation Generating Free Energy Surfaces
Motivation Calculating Observables
Machine Leaming Molecular Energies
Energies via Electron Densities: DFT
Machine learning electron densities
Machine learning for DFT...for molecules!
Machine learning for DFT...for Hy
Machine learning for DFT: HO
Sampling strategy for training geometries
Machine learning for DFT. benzene
Machine learning for DFT. ethane
Machine learning for DFT malonaldehyde
Overlap of test and training data
Future directions (happening now!)
Taught by
APS Physics
