Thesis Defense: Andrew Ma, Learning Simple Chemical Heuristics to Model and Discover Materials
Room 4-331
Doctoral Thesis Title: Learning Simple Chemical Heuristics to Model and Discover Materials
Presenter: Andrew Ma
Presenter’s Affiliation: RLE
Thesis Supervisor: Marin Soljačić
Date: May 9th, 2025
Time: 1:00 PM
Location: Room 4-331 (the Duboc Room, inside the Center for Theoretical Physics) and Zoom.
Abstract: Computational materials science has traditionally relied on first-principles methods, which involve direct calculation of the quantum mechanical wave function. Driven in part by the emergence of large-scale materials datasets, machine learning approaches have become increasingly prevalent in the 21st century. While deep learning models can predict material properties with high accuracy, they often function as black boxes that are difficult to understand. In contrast, chemists have long done well with simple and intuitive heuristics, such as identifying bond types using elements’ electronegativities. In this thesis, we present remarkably simple machine learning models for predicting important properties of materials, including topology, metallicity, and band gap. These models take the form of highly interpretable chemical heuristics. We also integrate the topology prediction model into a workflow for discovering new topological materials. Altogether, this work revisits the classic idea of chemical heuristics through a modern data-driven lens, shedding new light on fundamental problems in materials science.
Details
- Date: Friday, May 9
- Time: 1:00 pm - 2:00 pm
- Category: Thesis Defense
- Location: Room 4-331