Quadratic programming with linear inequality constraints

The least-squares method of optimization of quadratic functions is the most common and widely practiced. The exact procedure in matrix form, is described by Boot, p.25 [2]. Some of the merits of the least-squares method are discussed in [1]. This thesis discusses this least-squares method of optimization in several restricted cases. The matrix format is used throughout, and the less than full rank case (the matrix in the quadratic part of the objective function is of less than full rank) is of particular interest. It is taken up in Chapter II along with the case of linear restrictions.