A Parallel Algorithm for Large Scale Convex Optimal Control Problems

A parallel algorithm is developed for long horizon optimal control problems. This is done by first decomposing along the time axis the original problem into subproblems with shorter time horizon, and then using incentive coordination scheme to take the interaction of subproblems into account. For strictly convex problems, it is proved that the decomposed problem with linear incentive coordination is equivalent to the original problem. Thus the high level is a parameter optimization problem. Based upon the gradient method, a parallel algorithm is developed. A numerical example is used to illustrate its feasibility.