The PPS method-based constraint preconditioners for generalized saddle point problems

Abstract

For the large sparse generalized saddle point problems with non-Hermitian (1,1) blocks, we introduce a constraint preconditioner, which is based on the positive definite and semidefinite splitting (PPS) iteration method. Then we discuss one trait of the PPS-based constraint preconditioner, such as invertibility. We give the convergence of conditions of the preconditioning iteration method. Moreover, numerical experiments are given to illustrate that PPS-based constraint preconditioner has an obvious advantage of efficiency.