BIMSA >
BIMSA Computational Math Seminar
Machine learning-based optimization of numerical methods for solving PDEs
Machine learning-based optimization of numerical methods for solving PDEs
Organizers
Tahereh Eftekhari
,
Pipi Hu
,
Xin Liang
,
Zhiting Ma
,
Hamid Mofidi
, Chunmei Su
,
Axel G.R. Turnquist
,
Li Wang
,
Fansheng Xiong
,
Shuo Yang
,
Wuyue Yang
Speaker
Qian Wang
Time
Wednesday, March 11, 2026 2:00 PM - 3:00 PM
Venue
A3-4-312
Online
Zoom 518 868 7656
(BIMSA)
Abstract
High-fidelity simulations are extensively used to address complex scientific and engineering problems. The accuracy and efficiency of these simulations are primarily determined by the numerical methods employed in space and time discretization. The development of numerical methods can be viewed in two phases: construction and optimization. Effective optimization can substantially enhance the accuracy and efficiency of the numerical method, unlocking its full potential. Optimization involves selecting optimal values for free parameters; however, the relationship between the optimal parameter values and their dependencies is often complex and nonlinear. This lack of theoretical guidance for parameter optimization typically results in reliance on empirical experience for parameter selection. To address this challenge, we propose using machine learning techniques to overcome the complexities of parameter dependencies and develop a robust optimization framework for numerical methods. This presentation examines optimization strategies and methodologies through selected case studies.