Review of domain-decomposition methods for the implementation of FEM on massively parallel computers

The finite-element solution of electromagnetic phenomena, involving electrically large and complex structures, is a major challenge, since both accuracy and efficiency are hard to obtain. The authors introduce the different methods for performing the domain decomposition for distributed memory, multiple-instruction multiple-data (MIMD) multiprocessors. In discussing the advantages and disadvantages of the different methods, we assume that an iterative-matrix method, such as the conjugate gradient method, is used to solve the matrix equation.>