《《2016 Inexact constraint preconditioners for linear systems arising in interior point methods》.pdf

Comput Optim Appl (2007) 36: 137–147 DOI 10.1007/s10589-006-9001-0 Inexact constraint preconditioners for linear systems arising in interior point methods Luca Bergamaschi · Jacek Gondzio · Manolo Venturin · Giovanni Zilli Published online: 15 February 2007 © Springer Science+Business Media, LLC 2007 Abstract Issues of indefinite preconditioning of reduced Newton systems arising in optimization with interior point methods are addressed in this paper. Constraint pre- conditioners have shown much promise in this context. However, there are situations in which an unfavorable sparsity pattern of Jacobian matrix may adversely affect the preconditioner and make its inverse representation unacceptably dense hence too ex- pensive to be used in practice. A remedy to such situations is proposed in this paper. An approximate constraint preconditioner is considered in which sparse approxima- tion of the Jacobian is used instead of the complete matrix. Spectral analysis of the preconditioned matrix is performed and bounds on its non-unit eigenvalues are pro- vided. Preliminary computational results are encouraging. Keywords Interior-point methods · Iterative solvers · Preconditioners · Approximate Jacobian L. Bergamaschi ( ) · G. Zilli Department of Mathematical Methods and Models for Scientific Applications, University of Padua, Padua, Italy e-mail: berga@dmsa.unipd.it G. Zilli e-mail: zilli@dmsa.unipd.it J. Gondzio School of Mathematics, University of Edinburgh, Edinburgh, Scotland, UK e-mail: J.Gondzio@ed.ac.uk M. Venturin Department of Pure and Applied Mathematics, University of Padua, Padua, Italy e-mail: mventuri@math.unipd.it 138 L. Bergamaschi et al. 1 Introduction Interior point methods for linear, quadratic or nonlinear programming are the key op- timization methodology. Their theory [ 12] and implementation [1] is well understood. Wh