Degree

Doctor of Philosophy (PhD)

Department

Division of Computer Science and Engineering (CSE)

Document Type

Dissertation

Abstract

Understanding the segregation of bulk Earth melt systems into metallic (core) and silicate (mantle) phases under high-pressure and high-temperature conditions is central to modeling Earth’s interior, yet relevant experimental and computational studies remain limited. This work develops a machine learning–based simulation pipeline that iteratively couples first-principles (quantum mechanical) calculations with neural network training to generate high-fidelity force fields. Using major-element Fe–Mg–Si–O melt systems, with and without H and N, as testbeds, we demonstrate that this framework enables large-scale molecular dynamics simulations at near first-principles accuracy. We further introduce a sequence of phase identification methods, progressing from statistical binning of elemental distributions to density-aware clustering that resolves diffuse nonlinear boundaries, and ultimately to weakly supervised graph neural networks that generalize phase segmentation to substantially larger systems. Together, these approaches enhance our ability to model core–mantle differentiation and predict the interior distribution of key life-essential elements.

Date

4-10-2026

Committee Chair

Karki, Bijaya B.

LSU Acknowledgement

1

LSU Accessibility Acknowledgment

1

Share

COinS