Yulong Liu
Ph.D. student in Earth and Atmospheric Sciences at Cornell University.
My work is about getting AI to learn physics instead of only fitting data. I think about this through implicit neural representations, operator learning, LLM-based reasoning, and physics-informed neural networks, with applications to computational mechanics, bio-inspired mechanics, and coupled Earth science problems.
Computational mechanics
Implicit neural representation
Operator learning
LLMs for physics
Physics-informed neural networks
Bio-inspired mechanics
- Program
- Ph.D. in Earth Science
- Advisor
- Chloé Arson
- Focus
- AI for physics · INR · operator learning · LLMs · PINNs · bio-inspired mechanics
- Base
- Ithaca, New York
- yl3825@cornell.edu
Overview
Current directions in my work.
The question behind most of my work is straightforward: how can we make learning-based models respect physics instead of only matching data? I am especially interested in settings where geometry, constitutive behavior, and coupled processes matter, and where physical structure can make AI more useful for real mechanics and Earth science problems.