A combinatorial approach to convex quadratic programming

AbstractA convex quadratic program has Kuhn-Tucker conditions which are necessary and sufficient, and suggest a dual problem. These conditions may be formulated in a convenient schematic notation which leads to a finite class of equivalent problems displaying Cottle's duality. An inductive argument shows how the original problem may be reduced in a finite number of steps to an equivalent problem which is trivial. The fundamental duality theorem for convex quadratic programs follows directly