TY - CHAP T1 - An Online Stabilization Method for Parametrized Viscous Flows T2 - Reduction, Approximation, Machine Learning, Surrogates, Emulators and Simulators. Y1 - 2024 A1 - Shafqat Ali A1 - Francesco Ballarin A1 - Gianluigi Rozza AB -
The purpose of this work is to investigate the inf-sup stability of reduced basis (RB) method applied to parametric Stokes problem. While performing the Galerkin projection on the reduced space, the inf-sup approximation stability has always been a challenge for the RB community, even if the construction of reduced basis is done using a stable high-fidelity method. In this work we propose a new online stabilization strategy for RB approximation of parametrized Stokes problem. In this strategy, a stable high-fidelity method is used to construct the RB spaces, and then, online solution is improved by a post processing based on rectification method [8, 13, 16]. This approach involves the computation of less expensive (but less consistent) FE approximation during the online stage and hence the improvement of online solutions using a RB-based rectification method. The consistency of the RB solution is also improved. We compare this approach with existing offline-online stabilization approach presented in our earlier work [2]. All the numerical simulations are carried out using RBniCS [4, 14], an open-source reduced order modelling library, built on top of FEniCS [15].
JF - Reduction, Approximation, Machine Learning, Surrogates, Emulators and Simulators. PB - Springer, Cham SN - 978-3-031-55060-7 UR - https://link.springer.com/chapter/10.1007/978-3-031-55060-7_1 ER -