Dr. Li Zhu
Rutgers University/Department of Physics
Friday, March 7, 2025
12:00 Noon
Room 120 – Meyerhoff Chemistry Building
Host: Dr. Joe Bennett
“Advancing Materials Discovery: From Computational Methods to Real-World Applications”
Advances in computational materials design have revolutionized the discovery and optimization of materials for diverse applications. This talk presents a comprehensive overview of my research, which integrates innovative computational methods with machine learning techniques to accelerate the exploration of novel materials. I will highlight key developments, including the libfp library for crystal fingerprint analysis, the ReformPy methodology for fingerprint-oriented structural optimization, and the PALLAS framework for predicting transition pathways using graph theory and swarm intelligence. Additionally, I will introduce EOSnet, a crystal graph neural network model that incorporates Gaussian Overlap Matrix-based fingerprints to predict material properties with unprecedented accuracy. These computational tools have been successfully applied to design and investigate various materials, including superconductors, ferroelectrics, and systems under extreme conditions. This presentation aims to showcase how these methodologies contribute to solving pressing challenges in materials science and to inspire new collaborations and directions in computational materials research.