Author ORCID Identifier

https://orcid.org/0000-0003-2867-1706

Semester

Spring

Date of Graduation

2025

Document Type

Dissertation

Degree Type

PhD

College

Eberly College of Arts and Sciences

Department

Physics and Astronomy

Committee Chair

Aldo H. Romero

Committee Co-Chair

Tudor Stanescu

Committee Member

Tudor Stanescu

Committee Member

Subhasish Mandal

Committee Member

Eduardo Hernandez

Committee Member

Prashnna Gyawali

Abstract

Modern materials science generates vast amounts of data from computational simulations and experiments, creating significant challenges for data processing and analysis. This thesis addresses these challenges through the development and application of computational tools within the framework of Material Data Science (MDS). Contributions span the four pillars of MDS: Material/Molecular Data, Algorithms, Databases, and High-Throughput Processes—with a primary focus on the Algorithm, Data, Database pillars.

For the Algorithm pillar, two Python libraries were developed to streamline common analysis tasks. PyProcar simplifies the post-processing and visualization of electronic structure data (band structures, density of states, Fermi surfaces) obtained from various Density Functional Theory (DFT) codes, providing a unified interface for researchers. MechElastic automates the calculation of mechanical properties (stability criteria, elastic moduli, anisotropy) from DFT-computed elastic tensors for both bulk and 2D materials, reducing manual effort and standardizing analysis.

Computational studies on the binary NiTi and TiAu systems served both as applications of these tools and contributions to the Data pillar, generating datasets on low-energy structures and their properties. These studies highlighted the need for efficient data management, motivating contributions to the Database pillar. ParquetDB was created as a lightweight database framework utilizing the columnar Parquet format for efficient storage and retrieval of large, complex datasets within Python workflows. MatGraphDB extends this foundation, providing specialized tools for managing graph-structured materials data, crucial for representing atomic connectivity and knowledge relationships.

Collectively, the tools and methodologies developed in this thesis enhance the ability to manage, process, analyze, and interpret complex materials data. By addressing key bottlenecks in the computational materials science workflow, this work supports the advancement of Material Data Science and contributes to accelerating the discovery and design of new materials.

Download

Share

COinS