Intrinsic linear programming (Q1077337)

The author describes a standard linear program P, maximize cx subject to Ax\(\leq b\), in terms of a diagram of vector spaces and linear transformations such as \[ P:\quad R\leftarrow^{c}V_ 1\to^{A}V_ 2\leftarrow^{b}R \] where R is the space of real numbers with the usual order structure, \(V_ 1\), \(V_ 2\) are preordered vector spaces, and A, b, c are linear transformations, and derives the duality theorem in a completely intrinsic fashion. The theory is applicable to the infinite dimensional case as well as the finite dimensional case, provided suitable topological hypotheses are made in the former case. The author develops two sets of conditions, under both of which the duality theorem holds. If the conditions in the first set are relaxed, the duality theorem can fail by the program or its dual not having a solution and the other not being unbounded. If the conditions in the second set are relaxed, then the duality theorem can fail also by both programs having solutions without the maximum of the primal being equal to the minimum of the dual.