Time Parallel Scalable Multiphysics/Multiscale Framework

Abstract

We propose a new computational framework that combines the recently developed time-parallel (TP) and the compound wavelet matrix (CWM) methods. The framework, termed tpCWM, offers significant computational acceleration by making multiscale/multiphysics simulations computationally scalable in time and space domains. We demonstrate the accuracy and the scalability of the method on a prototype problem with oscillatory trajectory. The method corrects the coarse solution by iterative use of the CWM, which compounds the fine and the coarse solutions for the processes. Computational savings, over the fine solution as well as the TP method, in terms of the real time required to perform the simulations, can reach several orders of magnitude. We believe that this method is general enough to be applicable to a wide-class of computational physics problems. Tendency towards large number of cores and processors in parallel computers is compatible with the computational scalability of the algorithm.