Parallel computation of spectral portrait of large matrices by Davidson type methods (Q1385160)

The paper concerns spectral portrait computations and plotting for non-normal sparse and large matrices. The method chosen has been proposed by \textit{E. R. Davidson} [J. Comput. Phys. 17, 87-94 (1975; Zbl 0293.65022)]. The authors develop an efficient parallelization of this method which allows to compute the spectral portraits on distributed-memory parallel computer systems. For the general Davidson method, a selection of two preconditioners is shown which improves the performanc of parallel implementations. The computational experiments performed on a Paragon computer confirm the efficiency of the parallelization approach.