Ricerc@Sapienza

The paper studies the problem of determining the optimal control when singular arcs are present in the solution.
In the general classical approach the expressions obtained depend on the state and the costate variables at the
same time, so requiring a forward-backward integration for the computation of the control. In this paper,
sufficient conditions on the dynamics structure are provided and discussed in order to have both the control
and the switching function depending on the state only, so simplifying the computation avoiding the necessity

The paper studies the problem of determining the optimal control when singular arcs are present in the solution. In the general classical approach, the expressions obtained depend on the state and the costate variables at the same time, so requiring a forward-backward integration for the computation of the control. In this paper, firstly sufficient conditions on the dynamics structure are discussed, in order to have both the control and the switching function depending on the state only, computable by a simple forward integration.