Toward Accelerated Molecular Dynamics for Photofragmentation of Ions

Poster Number

7C

Lead Author Affiliation

Pre-Dental

Lead Author Status

Undergraduate - Sophomore

Second Author Affiliation

Chemistry

Second Author Status

Doctoral Student

Third Author Affiliation

Chemistry Department

Third Author Status

Faculty

Faculty Mentor Name

Anthony Dutoi

Format

Poster Presentation

Research or Creativity Area

Natural Sciences

Abstract

Molecular dynamics (MD) simulations model how atoms move over time by repeatedly computing the forces acting on each atom. This work implements the Velocity Verlet algorithm, which is a standard numerical integration scheme that conserves total energy (kinetic plus potential) over time, developed with an in-house Python framework. At each time step, we use the atomic simulation environment (ASE) interface to obtain forces from an underlying electronic structure method, converted to atomic units, and divided by atomic masses to yield accelerations. Our first attempt was to use the GxTB semiempirical method because of its promised efficiency. Although the GxTB calculator wasn’t sufficiently accurate, this opens up the door to testing other ASE interfaces to be substituted into the algorithm, as the fundamental integration remains the same. ASE interfaces will be tested against Q-Chem outputs to validate accuracy.

Location

University of the Pacific, DeRosa University Center

Start Date

24-4-2026 11:00 AM

End Date

24-4-2026 2:00 PM

This document is currently not available here.

Share

COinS
 
Apr 24th, 11:00 AM Apr 24th, 2:00 PM

Toward Accelerated Molecular Dynamics for Photofragmentation of Ions

University of the Pacific, DeRosa University Center

Molecular dynamics (MD) simulations model how atoms move over time by repeatedly computing the forces acting on each atom. This work implements the Velocity Verlet algorithm, which is a standard numerical integration scheme that conserves total energy (kinetic plus potential) over time, developed with an in-house Python framework. At each time step, we use the atomic simulation environment (ASE) interface to obtain forces from an underlying electronic structure method, converted to atomic units, and divided by atomic masses to yield accelerations. Our first attempt was to use the GxTB semiempirical method because of its promised efficiency. Although the GxTB calculator wasn’t sufficiently accurate, this opens up the door to testing other ASE interfaces to be substituted into the algorithm, as the fundamental integration remains the same. ASE interfaces will be tested against Q-Chem outputs to validate accuracy.