Explore reduced order modeling techniques for finite element simulations in this conference talk from the MFEM Workshop 2022. Discover how MFEM and libROM partnership can lower computational complexity while maintaining accuracy in large-scale scientific simulations. Learn about intrusive projection-based model reduction and non-intrusive approaches, and see their applications in various MFEM examples, including Poisson problems, linear elasticity, nonlinear heat conduction, and Euler equations. Gain insights into advanced techniques such as hyper reduction, dynamic mode decomposition, localization, time windowing, and parametric interpolation for optimizing finite element calculations.
Reduced Order Modeling for Finite Element Simulations with MFEM and libROM
Inside Livermore Lab via YouTube
Overview
Syllabus
Introduction
Projectionbased MFEM
Hyper Reduction
Dynamic Mode Decomposition
Localization
Time Windowing
Parametric Interpolation
Training Parameters
Taught by
Inside Livermore Lab