Skip to Main Content

Title ImagePublic Abstract


DE-SC0019390: Ab initio Molecular Dynamics Beyond Density Functional Theory

Award Status: Active
  • Institution: California Institute of Technology, Pasadena, CA
  • DUNS: 009584210
  • Most Recent Award Date: 11/06/2023
  • Number of Support Periods: 4
  • PM: Holder, Aaron
  • Current Budget Period: 09/15/2021 - 09/14/2024
  • Current Project Period: 09/15/2018 - 09/14/2024
  • PI: Chan, Garnet
  • Supplement Budget Period: N/A

Public Abstract

Computational modeling of chemical and photochemical processes in complex systems faces extraordinary challenges in terms of the first-principle description of molecular motions and chemical reactions (i.e., ab initio molecular dynamics (MD)).   The current state of the art for ab initio MD is to describe the molecular interactions using density functional theory (DFT), due to DFT's reasonable compromise between accuracy and computational cost. However, the chemical sciences are permeated with systems for which the approximations of DFT fundamentally break down or for which the computational cost of DFT remains prohibitive for the MD simulation of necessary length- and timescales.  The planned research will use quantum embedding strategies to move ``beyond DFT" in both key dimensions of higher accuracy and lower computational cost, thereby bringing broad new chemical and materials application domains within the reach of ab initio MD simulation.  The research will pursue theoretical innovations that include coupling highly correlated system wavefunction methods with DFT environments, transferable machine-learning methods for electronic structure, and non-adiabatic dynamics methods based on the Feynman path-integral framework.  The work will address frontier chemical challenges ranging from non-adiabatic dynamics through conical intersections in metalloenzymes, to biological transport of energy and charge, to coupled electron-nuclear quantum dynamics in surface reactions. The new simulation capabilities will be implemented in open-source, high-performance and developer friendly packages.  The planned work is fully aligned with the Computational Chemistry Sciences (CCS) objectives by providing new, advanced capabilities in major open source chemical-simulation software to be used by the community to take advantage of gains in massively parallel computing platforms and systematically alleviate the need to employ semi-empirical corrections.  The work will combine theoretical, computational, and algorithmic advances to increase (1000-fold or more) the speed of accurate molecular simulation, including methods and applications that align with the Department of Energy mission in the chemical sciences, geosciences, and biosciences.  The planned work is particularly well aligned with targeted focus areas that include multi-scale (i.e., embedding) methods for describing natural and artificial photochemistry for solar driven energy conversion and storage, approaches to account for competing dissipative mechanisms (decoherence or other non-idealities) in molecular systems, and methods for modeling molecular complexes composed of interfaces.

Scroll to top