An iterative solution method for solving \(f(A)x=b\), using Krylov subspace information obtained for the symmetric positive definite matrix A (Q1090063): Difference between revisions

For certain f an efficient iterative method for finding an approximate solution of \(f(A)x=b\) is developed, using intermediate information obtained by the conjugate gradient algorithm. The proposed algorithm requires the solution of an eigenvalue problem of low dimension as compared with the dimension of A.The particular case \(f(A)=A^ 2\) is also discussed. Numerical examples which provide evidence, supporting the proposed ideas, are presented.