Abstract
Algorithms for the inversion of a sparse square matrix have generally been designed with the aim of minimizing the number of calculations required to factorize the matrix and to solve for one or more right-hand sides. This usually involves attempting to minimize the number of non-zeros in the factors of the inverse. An inversion routine for use in the product-form simplex method must satisfy a rather different requirement. The inverse is updated during iterations to reflect changes in the basis, and additional non-zeros created by updating become significant. We consider the use of various forms of the Hellerman-Rarick P4 inversion algorithm in conjunction with the Forrest-Tomlin update of the factors, and suggest a heuristic for selection between these forms during the inversion process.
Similar content being viewed by others
Author information
Authors and Affiliations
Rights and permissions
About this article
Cite this article
Hattersley, B., Mackley, L. Construction of LU Factors of the Basis to Reduce Build-Up during Simplex Iterations. J Oper Res Soc 43, 507–518 (1992). https://doi.org/10.1057/jors.1992.76
Published:
Issue Date:
DOI: https://doi.org/10.1057/jors.1992.76