"Quantum Monte Carlo and Machine Learning Simulations of Dense Hydrogen"

David Ceperley
University of Illinois at Urbana-Champaign

We have developed Coupled-Electron QMC methods to simulate dense hydrogen, incorporating advanced wavefunctions and using reptation QMC for electronic energies and Path Integral MC for the proton distribution. Recent advances of the calculation of electronic energy gaps allow direct comparison with experimental measurements. Using QMC we have constructed a data-base of forces on the protons of dense hydrogen configurations and trained machine learnt force fields. Simulations of those force field predict novel solid hydrogen structures.

Back to Workshop IV: Monte Carlo and Machine Learning Approaches in Quantum Mechanics