Time-Domain Decomposition for Mixed-Integer Optimal Control Problems

Abstract We consider mixed-integer optimal control problems, whose optimality conditions involve global combinatorial optimization aspects for the corresponding Hamiltonian pointwise in time. propose a time-domain decomposition, which makes this problem class accessible programming using parallel-in-time direct discretizations. The approach is based on decomposition of system and interpretation resulting subproblems as suitably chosen problems subintervals An iterative procedure then ensures continuity states at boundaries via co-state information encoded virtual controls. prove convergence scheme discrete-continuous linear-quadratic present numerical results both well nonlinear problems.