Treffer: Development and Implementation of a Large Scale Linear Programming Code
Weitere Informationen
The objective of this research is to develop a computationally efficient computer code for solving large linear programming (LP) problems up to size (100 x 400). This program can handle upper and lower bounds on variables without affecting the size of the constraint set. The value of nonbasic variables can be initiated at a prespecified value other than its lower or upper bound by using the concept of superbasic variables. Multiple problems can also be run in one compilation. This program uses additional features to improve computational efficiency and to allow the solution of large-scale problems. These features include: 1) Sparsity/Product Form of Inverse. 2) Partial Pricing/Multiple Pricing. 3) Periodic Reinversion. Before presenting the details of these features, some variables and vector definitions are needed, and a description of the computer code will be provided.