A filter diagonalization for generalized eigenvalue problems based on the Sakurai-Sugiura projection method (Q2654201)

The authors deal with the generalized eigenvalue problem by the Sakurai-Sugiura (SS) projection method [cf. the second author and \textit{H. Sugiura}, J. Comput. Appl. Math. 159, No.\,1, 119--128 (2003; Zbl 1037.65040)], which aims at finding certain eigenvalues in a given domain. Here the SS method is formulated within the frame of the resolvent theory. A block version is also proposed, which allows an improvment in numerical accuracy. The method is also interpreted in the context of filter diagonalization. This method may exhibit a larger computational cost than the usual methods. However the advantage of the block SS method is the fact that it can be easily parallelized. Some significant numerical examples are discussed in details.