Scientific applications (e.g., computational fluid dynamics, quantum chemistry, and power grid simulation) can be extremely time-consuming. We study how to use machine learning models to significantly reduce the execution time of those applications. In particular, we use machine learning models to approximate traditional numerical simulation in those applications, or generate more meaningful initial solutions to enable faster computation convergence in those applications. Our ultimate research goal is to build a system that can automatically accelerate scientific applications with machine learning models.

We are organizing a scientific machine learning workshop in SC'23! Welcome to make your submissions. :)

Research Outcome:

• • [HPDC'23] Wenqian Dong, Gokcen Kestor and Dong Li.  Auto-HPCnet: An Automatic Framework to Build Neural Network-based Surrogate for High-Performance Computing Applications. In 32nd International Symposium on High-Performance Parallel and Distributed Computing (acceptance rate:) 

  • [SC'20] Wenqian Dong, Zhen Xie, Gokcen Kestor and Dong Li. Smart-PGSim: Using Neural Network to Accelerate AC-OPF Power Grid Simulation. In 32nd ACM/IEEE International Conference for High Performance Computing, Performance Measurement, Modeling and Tools (acceptance rate: 22.3%) 

  • [SC'19] Wenqian Dong, Jie Liu, Zhen Xie and Dong Li. Adaptive Neural Network-Based Approximation to Accelerate Eulerian Fluid Simulation. In 31st ACM/IEEE International Conference for High Performance Computing, Performance Measurement, Modeling and Tools (acceptance rate: 22.6%) 

This research is under collaboration with: