Pseudospectral element method for computational fluid dynamics and its parallel implementation

The basic equations of fluid mechanics have been known since early in the nineteenth century, but their nonlinear character has made it very difficult both to understand their general features and to predict flow evolution in specific cases. The possibility of using computers to attack such problems numerically was one of the original motivations for the development of stored-program digital computers and has remained in the forefront of large-scale computing ever since. We review here the basic ideas and report on some recent work done at the Applied Physics Laboratory.