A recursive quadratic programming algorithm that uses a new nondifferentiable penalty functions

Abstract

In this paper, a recursive quadratic programming algorithm is proposed and studied. The line search functions used are Han’s nondifferentiable penalty functions with a second order penalty term. In order to avoid maratos effect, Fukushima’s mixed direction is used as the direction of line search. Finally, we prove the global convergence and the local second order convergence of the algorithm.