Research Theme
We use AI to understand, model, and discover complex scientific laws, moving scientific practice from purely data-driven workflows toward mechanism-aware discovery.
Core Issues
- Integrating data-driven methods with physical and chemical mechanisms
- Low-data scientific problems
- Interpretability and scientific discovery capacity
- Model generalization and transfer across systems
Research Content
PDEs and Scientific Computing
Keywords: continuous-system modeling / learning physical laws
- Physics-informed neural networks (PINNs)
- Neural operators such as FNO and DeepONet
- Accelerated scientific simulation and surrogate models
AI Drug Discovery and Molecular Design
Keywords: discrete structures plus biochemical systems
- Molecular generative models including Diffusion and Graph Models
- Protein structure and interaction modeling
- Drug screening and property prediction (ADMET)
- Structure-property learning
Data-driven Scientific Discovery
Keywords: moving from calculation to discovery
- Scientific law discovery with symbolic regression and discovery tools
- AI-driven experimentation
- Multi-scale modeling
Application Scenarios
- Materials design
- AI drug discovery
- Energy and chemical systems
- Industrial process modeling