Sparsity Preserving Computation for Spectral Projectors

Abstract

Several areas of applications such as model order reduction preconditioner design and eigenvalue problems for spectral projectors can be found in the literature. In this paper a fast and sparsity preserving approach for computing the spectral projectors is proposed. The suggested approach can be used in both Newton iteration and integral representation based methods. A comparison of the original and the suggested approaches in terms of computation time sparsity preservation and accuracy is presented in this paper.