作者: Charbel Farhat , Marion Chandesris
DOI: 10.1002/NME.860
关键词:
摘要: A methodology for squeezing the most out of massively parallel processors when solving partial differential evolution equations by implicit schemes is presented. Its key components include a preferred time-integrator, decomposition time-domain into time-slices, independent time-integrations in each time-slice semi-discrete equations, and Newton-type iterations on coarse time-grid. Hence, this parallelizes time-loop time-dependent equation solver without interfering with its sequential or space-computations. It particularly interesting problems few degrees freedom such as those arising robotics protein folding applications, where opportunities parallelization over are limited. Error stability analyses proposed performed first- second-order hyperbolic problems. feasibility impact reducing solution time below what attainable methods which address only parallelism space-domain highlighted fluid, structure, coupled fluid–structure model Copyright © 2003 John Wiley & Sons, Ltd.