Pivoting for Size and Sparsity in Linear Programming Inversion Routines

This note reports some experimental results on the inversion of real linear programming bases, with particular emphasis on the compromise between minimum density and maximum numerical stability. The general features of a linear programming inversion routine are outlined and the special structure of linear programs considered. The main result is a suitable and apparently safe "pivot tolerance " level, together with more general data on the nature and behaviour of the problems. 1