Molecular Electron Densities via Machine Learning - IPAM at UCLA

Explore a conference talk on using machine learning to predict molecular electron densities and energies for quantum mechanics applications. Delve into Leslie Vogt-Maranto's presentation at IPAM's Monte Carlo and Machine Learning Approaches in Quantum Mechanics Workshop, covering topics such as Kohn-Sham density functional theory, mapping nuclear potential to electron density, and leveraging delta-learning for ab initio energy calculations. Discover how these machine learning models can generate molecular dynamics trajectories, sample strained geometries, and capture conformer changes with sufficient accuracy. Gain insights into the potential of combining machine learning with quantum mechanics for more efficient and accurate molecular simulations.

Intro
Acknowledgements
Introduction: Generating Free Energy Surfaces
Motivation: Generating Free Energy Surfaces
Motivation: Calculating Observables
Machine Learning for Molecular Dynamics
Machine leaming electron densities
Machine learning for DFT.. for molecules!
Training using nuclear coordinates vs. densities
Sampling strategy for training geometries
Machine learning for DFT malonaldehyde
Overlap of test and training data
Machine leaming for DFT+
A-learning for coupled cluster (via DFT)
A-learning for coupled cluster optimizations
Molecular dynamics with coupled cluster energies?
MD using combined models
Machine leaming for molecular systems