A Perspective On Data-Parallel Implicit Solvers For Mechanics

With the widespread acceptance of distributed memory multiprocessing as a cost-e ective means of solving very highly resolved problems in computational uid and structural mechanics, many engineers and scientists are encouraged with their initial ports of CFD or CSM codes for parallel execution, and are interested in learning whether applied mathematicians and computer scientists have anything to o er with the next step. This note is both a cautious a rmative and an attempt to de ne important roles that remain to be played by those familiar with the applications, themselves. We provide a high-level summary of the state of the art in domain decomposition methods, and a selective bibliography.