Modal and interior nodal masters in parallel condensation methods for generalized eigenvalue problems

For large-scale eigenvalue problems the authors introduced in [4] a coarse grained parallel algorithm for distributed memory computers based on substructuring and static condensation. The approach can be generalized to non-nodal masters if the support of each of the generalized masters is contained in the interior of one substructure. In this note we demonstrate that modal masters are superior to interior nodal masters