Polynomial preconditioning on vector computers (Q1316153)

The computation of the minimal eigenpair of the generalized eigenproblem \(Ax = \lambda Bx\), where \(A\) and \(B\) are \(n\times n\) symmetric positive definite matrices, by the preconditioned conjugate gradient (PCG) method is discussed. Choosing the preconditioner \(M\) is analyzed. Results are reported (tables are shown) on computer experiments to evaluate the effectiveness of the PCG method on the Cray Y-MP computer. The additive polynomial preconditioner and the SSOR polynomial preconditioner methods are compared.