Dynamic Programming and Optimal Trajectories for Quadratic Variational Processes.
Dynamic programming provides a standard tool for determining optimal feedback control policies for linear systems with quadratic measures of cost. The situation has been less satisfactory, however, with regard to optimal trajectories. A one-sweep initial-value method is presented in this study for determining both optimal policies and optimal trajectories. It is shown also that the solution of the Cauchy problem satisfies the Euler equation and the boundary conditions. 18 pp. Ref