On the Solution of a Class od Quadratic Programs Using a Differentiable Exact Penalty Function

this paper we define Newton-type algorithms for the solution of box constrained quadratic programming problems. These algorithms are based on the unconstrained minimization of a continuously differentiable exact penalty function and make use of consistent approximations of the Newton's direction of the penalty function. Under suitable assumptions, finite termination at a solution of the constrained problem can be established.