Simultaneous Iteration for Partial Eigensolution of Real Matrices
Alan Jennings, William J. Stewart · IMA Journal of Applied Mathematics · 1975
Simultaneous iteration methods are presented for obtaining dominant eigenvalues and corresponding eigenvectors of real unsymmetric matrices. These methods involve the accurate eigensolution of the smaller interaction matrix at each iteration, the dimensions of which depend on the number of vectors processed simultaneously. A bi-iteration procedure is proposed when both left and right eigenvectors are required, and a lopsided iteration procedure when only one set of eigenvectors is required. Numerical tests are discussed.