A splitting preconditioner for saddle point problems.

From MaRDI portal





For the large sparse saddle point problem, the authors studied a splitting iteration method and its preconditioning property. This method is a special case of the generalized parameterized inexact Uzawa (GPIU) method, a generalization of the parameterized inexact Uzawa (PIU) method introduced and studied by \textit{Z. Z. Bai}, \textit{B. N. Parlett} and \textit{Z.-Q. Wang} [Numer. Math. 102, No. 1, 1--38 (2005; Zbl 1083.65034)] and by \textit{Z.-Z. Bai} and \textit{Z.-Q. Wang} [Linear Algebra Appl. 428, No. 11--12, 2900--2932 (2008; Zbl 1144.65020)]. The PIU method converges fast under certain restrictions on the preconditioning matrix, which motivates the authors of this paper to present the splitting iteration method. Theoretical analyses have shown that this splitting iteration method is convergent for all \(t>0\), and its convergence rate is controlled by \(t\), where \(t\) is an iteration parameter. Compared with the PIU and the GPIU methods, this splitting iteration method requires less computer memory. In addition, the eigenvalues of the corresponding preconditioned matrix is discussed and an upper bound for the degree of the minimal polynomials of the preconditioned matrix is derived. Numerical results about a model Stokes problem and a linear-constraint least-squares problem are given to illustrate the effectiveness of the splitting iteration method.



Cites work


Cited in
(68)








This page was built for publication: A splitting preconditioner for saddle point problems.

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q2897409)