Zihan Zhou
I am a Ph.D. student in Data Science at The Chinese University of Hong Kong, Shenzhen, working under the supervision of Prof. Tianshu Yu. My research focuses on AI4Science, particularly on developing machine learning methods for solving physical dynamics and generating physically plausible structures.
Research Interests
My work spans several interconnected areas at the intersection of machine learning and scientific computing:
-
Physics-Informed Machine Learning: Developing diffusion models and neural operators that incorporate physical priors (PDEs, conservation laws) and geometric invariances (SE(3)-equivariance) for learning and generating physical dynamics.
-
Differential Equation Solving: Designing neural differential equation solvers and projection operators for forward and inverse problems in partial differential equations.
-
Manifold-Aware Generation: Developing projection-free SDE/ODE solvers on SE(3)-invariant manifolds for efficient 3D structure generation (molecules, protein conformations).
Applications: Molecular conformation generation, protein dynamics simulation, climate modeling, and irregular time-series analysis.
Recent Highlights
For a complete list of publications, please see the Publications page.