A LINEAR PROGRAMMING APPLICATION OF A LEFT INVERSE OF A BASIS MATRIX

Abstract : Linear programming theorems are proved by using left inverse of a basis matrix in place of the ordinary inverse. It is shown that such left inverse always exists and reduces to the regular inverse in the event that the basis matrix is square. It is also proved that even though the left inverse is not unique it can still be used to give a unique expression for any Pj in terms of the basis. Thus it is possible to solve matrix equations of the form BXj=Pj where B is a basis, without considering whether or not B is square. (Author)