Skip to content
@AI4Science-WestlakeU

AI4Science-WestlakeU

AI for Scientific Simulation and Discovery Lab

Our research group at Westlake University (西湖大学) carries out long-term work on core and universal problems for AI + Science:

  • AI for scientific simulation, design, and control: Developing machine learning algorithms (based on Graph Neural Networks and Diffusion Models) for large-scale, multi-scale scientific simulation (applied to fluid dynamics, materials, plasmas), scientific design (protein design, materials design, mechanical design), and control (fluid control, PDE control);
  • AI for scientific discovery: Developing machine learning algorithms (based on foundation models and neuro-symbolic AI) to discover universal rules and internal structures in scientific systems (applied to life sciences and physics);

Group website: https://ai4s.lab.westlake.edu.cn/

Collaborators (a non-exhaustive list):

Popular repositories Loading

  1. flow_guidance flow_guidance Public

    [ICML 2025] The official implementation of the paper "On the Guidance of Flow Matching"

    Python 93 9

  2. BuildArena BuildArena Public

    BuildArena, where LLM agents design, build, and test rockets, cars, and bridges in a physics simulator given a goal-directed sentence.

    Python 87 3

  3. RealPDEBench RealPDEBench Public

    [ICLR26 Oral] RealPDEBench: A Benchmark for Complex Physical Systems with Paired Real-World and Simulated Data

    Python 65 9

  4. wdno wdno Public

    [ICLR 2025] Wavelet Diffusion Neural Operator (WDNO) uses diffusion models on wavelet space for generative PDE simulation and control.

    Python 59 2

  5. diffphycon diffphycon Public

    [NeurIPS2024] DiffPhyCon uses generative models to control complex physical systems

    Jupyter Notebook 47 4

  6. cindm cindm Public

    [ICLR24] CinDM uses compositional generative models to design boundaries and initial states significantly more complex than the ones seen in training for physical simulation

    Jupyter Notebook 40 3

Repositories

Showing 10 of 25 repositories
  • Frontiers-in-Computer-Science-and-Technology-2026 Public

    Frontiers in Computer Science and Technology 2026 Spring

    AI4Science-WestlakeU/Frontiers-in-Computer-Science-and-Technology-2026’s past year of commit activity
    Jupyter Notebook 32 4 0 0 Updated Apr 1, 2026
  • scDFM Public

    scDFM: Distributional Flow Matching for Robust Single-Cell Perturbation Prediction

    AI4Science-WestlakeU/scDFM’s past year of commit activity
    Python 9 MIT 0 3 0 Updated Mar 31, 2026
  • GenCP Public

    Code for reproducing the paper "GenCP: Towards Generative Modeling Paradigm of Coupled physics" (https://arxiv.org/abs/2601.19541).

    AI4Science-WestlakeU/GenCP’s past year of commit activity
    Python 11 MIT 2 0 0 Updated Mar 15, 2026
  • wdno Public

    [ICLR 2025] Wavelet Diffusion Neural Operator (WDNO) uses diffusion models on wavelet space for generative PDE simulation and control.

    AI4Science-WestlakeU/wdno’s past year of commit activity
    Python 59 MIT 2 0 0 Updated Mar 8, 2026
  • RealPDEBench Public

    [ICLR26 Oral] RealPDEBench: A Benchmark for Complex Physical Systems with Paired Real-World and Simulated Data

    AI4Science-WestlakeU/RealPDEBench’s past year of commit activity
    Python 65 9 0 0 Updated Mar 8, 2026
  • BuildArena Public

    BuildArena, where LLM agents design, build, and test rockets, cars, and bridges in a physics simulator given a goal-directed sentence.

    AI4Science-WestlakeU/BuildArena’s past year of commit activity
    Python 87 3 0 0 Updated Feb 25, 2026
  • cbi-cfm Public

    Repository for Bidirectional Conditional Flow Matching (Bi-CFM) and Conservation-constrained Bi-CFM (CBi-CFM).

    AI4Science-WestlakeU/cbi-cfm’s past year of commit activity
    Python 1 MIT 0 0 0 Updated Nov 4, 2025
  • diffphycon Public

    [NeurIPS2024] DiffPhyCon uses generative models to control complex physical systems

    AI4Science-WestlakeU/diffphycon’s past year of commit activity
    Jupyter Notebook 47 MIT 4 2 0 Updated Nov 4, 2025
  • t_scend Public

    This repo is the code for T-SCEND, a novel framework that significantly improves diffusion model’s reasoning capabilities with better energy-based training and scaling up test-time computation.

    AI4Science-WestlakeU/t_scend’s past year of commit activity
    Python 26 MIT 1 0 0 Updated Oct 19, 2025
  • FLDmamba Public
    AI4Science-WestlakeU/FLDmamba’s past year of commit activity
    Python 6 MIT 1 1 0 Updated Oct 19, 2025

Most used topics

Loading…