Two-dimensional analysis of a gradient method in function space optimal control algorithm

Non-linear optimal control problems often require solution using iterative procedures and, hence, they fall naturally in the realm of 2D systems where the two dimensions are response time horizon and iteration index, respectively. The paper employs 2D systems theory, in the form of unit memory repetitive process techniques, to analyse local stability behaviour of a gradient method in function space algorithm, for solving continuous non-linear dynamic optimal control problems. It is shown that 2D systems theory can be usefully applied to analyse the properties of the algorithm producing a novel local stability condition.