Multigrid Reduced in Time for Isogeometric Analysis
Roel Tielen (TU Delft - Numerical Analysis)
Matthias Moller (TU Delft - Numerical Analysis)
K. Vuik (TU Delft - Delft Institute of Applied Mathematics)
More Info
expand_more
Other than for strictly personal use, it is not permitted to download, forward or distribute the text or part of it, without the consent of the author(s) and/or copyright holder(s), unless the work is under an open content license such as Creative Commons.
Abstract
Isogeometric Analysis (IgA) can be seen as the natural extension of the Finite Element Method (FEM) to high-order B-spline basis functions. Combined with a time inte- gration scheme within the method of lines, IgA has become a viable alternative to FEM for time-dependent problems. However, as processors' clock speeds are no longer increasing but the number of cores are going up, traditional (i.e., sequential) time integration schemes become more and more the bottleneck within these large-scale computations. The Multigrid Reduced in Time (MGRIT) method is a parallel-in-time integration method that enables exploitation of parallelism not only in space but also in the temporal direction. In this paper, we apply MGRIT to discretizations arising from IgA for the _rst time in the literature. In particular, we investigate the (parallel) performance of MGRIT in this context for a variety of geometries, MGRIT hierarchies and time integration schemes. Numerical results show that the MGRIT method converges independent of the mesh width, spline degree of the B-spline basis functions and time step size _t and is highly parallelizable when applied in the context of IgA.