Article ID: | iaor20051937 |
Country: | South Korea |
Volume: | 21 |
Issue: | 2 |
Start Page Number: | 43 |
End Page Number: | 60 |
Publication Date: | Nov 2004 |
Journal: | Korean Management Science Review |
Authors: | Seol Tongryeol |
Keywords: | interior point methods |
Every iteration of interior-point methods of large scale optimization requires computing at least one orthogonal projection. In the practice, symmetric variants of the Gaussian elimination such as Cholesky factorization are accepted as the most efficient and sufficiently stable method. In this paper several specific implementation issues of the symmetric factorization that can be applied for solving such equations are discussed. The code called McSML being the result of this work is shown to produce comparably sparse factors as other implementations in the MATLAB*** environment. It has been used for computing projections in an efficient implementation of self-regular based interior-point methods, McIPM. Although primary aim of developing McSML was to embed it into an interior-point methods optimizer, the code may equally well be used to solve general large sparse systems arising in different applications.