A closer look at local eigenvalue solvers for adaptive FETI-DP and BDDC

Abstract

Adaptive coarse spaces for domain decomposition methods are an active area of research to make iterative domain decomposition methods robust with respect to large discontinuities in the material parameters or almost incompressible elasticity. In order to make their use feasible for applications, the computational overhead of the adaptive methods has to be reduced while maintaining the robustness. Since the solution of the local eigenvalue problems is expensive, the iterative eigenvalue solvers are also considered in detail and potential improvements are presented.