Date: Friday, December 03, 2021
Location: ZOOM East Hall (3:00 PM to 4:00 PM)
Title: Physicsconstrained datadriven methods for accurately accelerating simulations
Abstract: A datadriven model can be built to accurately accelerate computationally expensive physical simulations, which is essential in multiquery problems, such as inverse problem, uncertainty quantification, design optimization, and optimal control. In this talk, two types of datadriven model order reduction techniques will be discussed, i.e., the blackbox approach that incorporates only data and the physicsconstrained approach that incorporates the first principle as well as data. The advantages and disadvantages of each method will be discussed. Several recent developments of generalizable and robust datadriven physicsconstrained reduced order models will be demonstrated for various physical simulations as well. For example, a hyperreduced timewindowing reduced order model overcomes the difficulty of advectiondominated shock propagation phenomenon, achieving a speedup of O(20~100) with a relative error much less than 1% for Lagrangian hydrodynamics problems, such as 3D Sedov blast problem, 3D triple point problem, 3D TaylorGreen vortex problem, 2D Gresho vortex problem, and 2D RayleighTaylor instability problem. The nonlinear manifold reduced order model also overcomes the challenges posed by the problems with Kolmogorov's width decaying slowly by representing the solution field with a compact neural network decoder, i.e., nonlinear manifold. The spacetime reduced order model accelerates a largescale particle Boltzmann transport simulation by a factor of 2,700 with a relative error less than 1%. Furthermore, successful application of these reduced order models for matematerial latticestructure design optimization problems will be presented. Finally, the library for reduced order models, i.e., libROM (https://www.librom.net), and its webpage and several YouTube tutorial videos will be introduced, which is useful for education as well as research purpose.
Files:
Speaker: Youngsoo Choi
Institution: LLNL
Event Organizer: MICDE mcteja@umich.edu
