A sequential method for a class of box constrained quadratic programming problems (Q2483014)

This paper studies the following class of box constrained problems: \[ \min f(x)={1\over 2} x^T Dx+ c^T x+{1\over 2} k(h^T x+ h_0)^2, \] \[ x\in B= \{x\in\mathbb{R}^n: l\leq x\leq u\} \] from both theoretical and algorithmic point of view. On one hand, to deepen on the properties of these problems, and on the other hand, to propose, in a unifying approach, an algorithm able to solve them for any fixed \(k\in\mathbb{R}\). Notice that for \(k< 0\) small enough, the quadratic objective function is not convex. The algorithm developed in the present paper is based on the optimal level solutions method and in general, stops after no more than \(2n-1\) iterations. A modified version of the algorithm is implemented and the results of a computationaly test are also presented.