Author ORCID Identifier

https://orcid.org/0000-0002-3080-5868

Semester

Summer

Date of Graduation

2025

Document Type

Dissertation

Degree Type

PhD

College

Eberly College of Arts and Sciences

Department

Physics and Astronomy

Committee Chair

Aldo Romero

Committee Co-Chair

Juan Felipe Carrasquilla Alvarez

Committee Member

Subhasish Mandal

Committee Member

Tudor Stanescu

Abstract

Understanding and predicting the magnetic behavior of materials from first principles is one of the central challenges in condensed matter physics. This dissertation presents a systematic framework that bridges ab initio calculations, many-body physics, and effective spin models to analyze magnetic materials in particular, but also more general quantum systems. Starting from the electron many-body Hamiltonian and the second quantization formalism, we derive Density Functional Theory (DFT) and explain how magnetic properties emerge from exchange interactions and can be interpreted as perturbations to the magnetization density. To capture these effects efficiently, we construct Heisenberg models from first principles using the Magnetic Force Theorem (MFT), which relates small variations in Kohn–Sham eigenvalues to magnetic exchange parameters. The combination of MFT and the Heisenberg model offers a computationally tractable approach to the many-body problem without compromising accuracy. However, it poses challenges due to assumptions of fixed localized spins, homogeneous rotations, and prior knowledge of the magnetic ground state. To address these limitations, we develop a self-consistent workflow based on linear spin-wave theory (LSWT) to identify stable spin configurations and predict the magnetic ground state. We also introduce a ligand downfolding procedure that systematically corrects for the limitations of MFT in systems with significant hybridization, addressing the ambiguities of magnetic region definitions and rotational homogeneity. We demonstrate the accuracy of these methods across a diverse set of transition metal compounds, showing agreement with experimental magnetic structures and spin-wave spectra measured by neutron scattering. Finally, we explore a deep learning–aided quantum computing approach to the many-body problem. We design a neural network architecture that learns to predict variational ansätze for quantum algorithms, enabling fast and generalizable solutions to electronic structure calculations. Together, these contributions advance a comprehensive strategy for magnetic materials design and quantum system modeling grounded in quantum theory and numerical methods.

Download

Share

COinS