Hi! I'm Anthony Zhou, a PhD student at Carnegie Mellon University's
MAIL Lab
where I research machine learning methods for PDE solving and physics simulation.
I enjoy combining numerical methods with deep learning to create fast and accessible engineering tools.
In my free time I like to cook, go outside, or take pictures.
You can find my CV
and publications on Google Scholar.
Research
My research primarily focuses on applying prior knowledge from physics and numerical methods to improve neural PDE surrogates. As these surrogates become faster and more accurate, I'm interested in leveraging this to create new and interesting engineering tools. Here are some of my representative works:
-
Benchmarking Stochastic Interpolants for Modeling Preprint, 2026
Physical Systems -
Hamiltonian Neural PDE Solvers through Functional Neural Information Processing Systems (NeurIPS), 2025
Approximation -
Predicting Change, Not States: An Alternate Framework for Computer Methods in Applied Mechanics and Engineering, 2025
Neural PDE Surrogates -
Text2PDE: Latent Diffusion Models for Accessible Physics International Conference on Learning Representations (ICLR), 2025
Simulation
Other Works
-
Cross-Domain Pretraining for Steady-State Neural CFD Surrogates Preprint, 2026 -
Generative Neural Operators through Diffusion Last Layer International Conference on Machine Learning (ICML), 2026 -
Generative Latent Neural PDE Solver using Flow Matching Machine Learning: Science and Technology, 2026 -
Strategies for Pretraining Neural Operators Transactions on Machine Learning Research, 2024 -
Masked Autoencoders are PDE Learners Transactions on Machine Learning Research, 2024 -
CaFA: Global Weather Forecasting with Factorized Attention on Sphere Preprint, 2024 -
FaultFormer: Pretraining Transformers for Adaptable Bearing Fault Classification IEEE Access, 2024 -
The Variable Stiffness Treadmill 2: Development and Validation of a Unique Tool to Investigate Locomotion on Compliant Terrains Journal of Mechanisms and Robotics, 2025