“…This kind of system of linear equations arises in a variety of scientific and engineering applications, such as computational fluid dynamics, constrained optimization, optimal control, weighted least-squares problems, electronic networks, computer graphic, the constrained least squares problems and generalized least squares problems etc; see [2,9,18,23,28,29] and the references therein. In addition, we can also obtain saddle point linear systems from the meshfree discretization of some partial differential equations [12,20] or the mixed hybrid finite element discretization of second order elliptic problems [11]. A comprehensive summary about various applications leading to saddle point matrices and a general framework of preconditioning methods and their theoretical analyses were given in [8].…”