On an Algorithm for Optimal Control Using an Augmented Hamiltonian

Abstract

An algorithm on optimal control due to Y. Sakowa and Y. Shindo is considered. it is based on Pontryagins minimum principle and the Hamiltonian is extended by a penalty term. Using another penalty term for this algorithm, under suitable assumptions we can show that each sequence of control vectors generated by the algorithm converges with respect to the $L_1$-norm to an admissible control fulfilling the first order optimality condition.