Two improvements of the deteriorated PSS preconditioner for generalized saddle point problems (Q526707)

The authors consider two new improved variants of the DPSS preconditioner for generalized saddle point problems. These preconditioners are better approximations to the generalized saddle point matrix \(A\) than the former DPSS preconditioner, as well as easier to implement than the DPSS one. Moreover the authors prove that the corresponding splitting iteration methods are unconditionally convergent and the new preconditioned matrices have clustered eigenvalue distribution.