On approximate solutions of infinite systems of linear inequalities (Q1121345)

Considering a theorem of \textit{A. J. Hoffman} [J. Res. Nat. Bur. Standards 49, 263-265 (1952; MR 14,155)], regarding the existence of a solution for the finite consistent system \(A_ x\leq b\), this paper extends the results on approximate solutions on finite systems of linear inequalities to infinite systems of linear inequalities. It is shown that, for a given infinite system of linear inequalities (satisfying certain conditions), the Euclidean distance from a vector x to the solution set of the system is equivalent to the ``biggest violation'' by x of the system. Any vector ``x'' that ``almost'' satisfies the system is then ``enough close'' to a solution of the system. There are many practical, as well as theoretical problems that can be formulated as semifinite linear programs. Generally, it is very difficult to find an optimal solution in a finite number of steps. The algorithms for solving semifinite linear programming can only provide an approximate solution at each step. The problem is to stop the algorithm as soon as a satisfactory approximate solution is discovered.