Using incomplete indefinite $LDL^T$ preconditioning for inexact interior point methods for linear programming
Abstract
Most linear algebra kernels in interior point methods for linear programming require the solution of linear systems of equation with the matrix (or ), where denotes the constraint matrix of the linear program. This matrix arises from the reduced KKT system by block elimination. If the number of non-zeros in or in its Cholesky factorization is very large, the computational cost and memory requirement to solve the linear systems of equations with may be prohibitively large. In this work we implement an interior point method described by R. Freund and F. Jarre. Forming the normal equation matrix is avoided altogether and we work with the reduced KKT system instead. We solve the linear systems for the Newton directions iteratively only to low accuracy using SQMR and an indefinite multilevel preconditioner. Preliminary numerical results are encouraging.
Cite
@article{arxiv.1708.04298,
title = {Using incomplete indefinite $LDL^T$ preconditioning for inexact interior point methods for linear programming},
author = {Robert Luce},
journal= {arXiv preprint arXiv:1708.04298},
year = {2017}
}
Comments
APMOD 2012 extended abstract; posted for archivational purpose