Structured interior point methods for optimal control

It is shown that primal-dual potential reduction algorithms for linear and quadratic programming can be put to use in solving problems in the optimal control of discrete-time systems, with general pointwise constraints on states and controls. The author describes an interior point algorithm for a discrete-time linear-quadratic regulator problem, and shows how it can be efficiently incorporated into a sequential quadratic programming algorithm for nonlinear problems. The key to the efficiency of the interior-point method is the banded structure of the coefficient matrix which is factorized at each iteration. This same feature makes it suitable for implementation on parallel computers.<<ETX>>