On the Subspace Projected Approximate Matrix method

摘要: We provide a comparative study of the Subspace Projected Approximate Matrix method, abbreviated SPAM, which is fairly recent iterative method computing few eigenvalues Hermitian matrix A. It falls in category inner-outer iteration methods and aims to reduce costs matrix-vector products with A within its inner iteration. This done by choosing an approximation A0 A, then, based on both A0, define sequence (Ak)k=0n matrices that increasingly better approximate as process progresses. Then Ak used kth instead