Parallel implementation of nonlinear Schwarz domain decomposition methods
Please login to view abstract download link
In general, domain decomposition methods (DDMs) are iterative, robust, and highly parallel scalable solvers for discretized partial differential equations. The parallelization is achieved through a divide-and-conquer approach, where the computational domain is decomposed into smaller subdomains. Nonlinear DDMs are efficient alternatives to classical Newton-Krylov-DDMs. In contrast to the latter methods, in nonlinear DDMs, the nonlinear partial differential equation is decomposed into subdomains before linearization, yielding potential improvement in the nonlinear convergence behavior.
