Graduation Year

2024

Date of Submission

4-2024

Document Type

Campus Only Senior Thesis

Degree Name

Bachelor of Arts

Department

Physics

Reader 1

Marivi Fernández-Serra

Reader 2

Scot Gould

Abstract

The time-evolution of many-body systems can be described using molecular dynamics (MD) simulations, which are split into two principal approaches. Classical force fields use analytic expressions for a system's potential energy, predicting molecular behavior by solving Newton's equations of motion. On the other hand, ab initio MD simulations incorporate electronic structure by numerically solving the many-body Schrödinger equation. In collaboration with Professor Fernández-Serra's research group at Stony Brook University, the purpose of this thesis is to present an intermediate approach called an electron force field (EFF) that aims to achieve the accuracy of AIMD simulations, while maintaining the efficiency of classical methods. The EFF models electrons in the system as spherical Gaussians, parameterized only by their position, width, and spin. The analytic functional form of the EFF is related to the energy decomposition of density functional theory (DFT), featuring contribution terms for the electrostatic interactions between constituents of the system, as well as kinetic, exchange, and correlation energy components. Additive corrections are made using spline interpolations for same-spin, opposite-spin, and core-valence interactions. While this successfully improves the performance of the force field, it also increases the number of parameters, creating a more complex energy landscape to optimize. Despite these challenges, the EFF achieves good accuracy in predicting molecular energies and equilibrium geometries on a range of molecules.

Download

This thesis is restricted to the Claremont Colleges current faculty, students, and staff.

Share

COinS