Biography
Jiawei is a PhD candidate at UChicago, merging quantum chemistry with HPC and ML. During his undergrad, Jiawei collaborated with Prof. Jin Zhao and Prof. Noa Marom to research into excited molecular dynamics and surface generation. At Uchicago, he worked with Prof. Giulia Galli on utilizing deep learning for solving electronic structure problems as well as generating general Density Functional Approximation.
Interests
- Artificial Intelligence
- Applied Physics
- Quantum Chemistry
- High-Performance Computing
Education
BSc in Applied Physics, 2016-2020
University of Science and Technology of China
Ph.D. in Quantum Science and Engineering, 2020-2026
University of Chicago
Experience
Ph.D. Candidate
University of Chicago
Deign and implement largely-parallel quantum chemistry and deep learning methods for solveing electronic structure problems.
Research Scientist Intern
TikTok (ByteDance Research)
Project: Predicting Physical Models with Uncertainty-Aware Graph Neural Networks:
- Designed a graph representation for many-body systems.
- Implemented a Graph Neural Network for improved pattern extraction from graph data.
- Developed a multithread dataset with distributed-memory scheme to streamline data loading.
- Estimating the posterior Gaussian distribution of neural network’s weights for uncertainty estimation.
Visiting Scholar
Carnegie Mellon University
Project: Designing Symmetry-Preserving Neural Network for Multi-Objective Structural Optimization:
- Designed a specialized graph algorithm for efficient conversion between external and internal coordinates.
- Applied Gaussian Processes to determine the Pareto optimum in multi-objective structural relaxation.
Exchange Student
University of California, Los Angeles
Enhanced academic and cultural understanding by visiting UCLA.
Research Assistant
Institute of Software, Chinese Academy of Science
Developed a prototype for real-time taxi-hailing App.
Projects
*
Using multi-agent Proximal Policy Optimization to solve famous Overcooked game.
Recent Publications
Quickly discover relevant content by filtering publications.
(2024).
Screened-Exchange Range-Separated Hybrid Functionals for Surfaces and Interfaces.
Bulletin of the American Physical Society.
(2023).
Nonempirical Range-Separated Hybrid Functional with Spatially Dependent Screened Exchange.
Journal of Chemical Theory and Computation.
(2023).
Hybrid functionals for heterogenous materials.
APS March Meeting Abstracts.
(2020).
Ogre: A python package for molecular crystal surface generation with applications to surface energy and crystal habit prediction.
The Journal of Chemical Physics.