We analyze the effect which a fold and simple cusp singularity in the flow of a parametrized family of extremal trajectories of an optimal control problem has on the corresponding parametrized cost or value function. A fold singularity in the flow of extremals generates an edge of regression of the value implying the well-known results that trajectories stay strongly locally optimal until the fold-locus is reached, but lose optimality beyond. Thus fold points correspond to conjugate points. A simple cusp point in the parametrized flow of extremals generates a swallowtail in the parametrized value. More specifically, there exists a region in the state space which is covered 3:1 with both locally minimizing and maximizing branches. The changes from the locally minimizing to the maximizing branch occur at the fold-loci and there trajectories lose strong local optimality. However, the branches intersect and generate a cut-locus which limits the optimality of close-by trajectories and eliminates these trajectories from optimality near the cusp point prior to the conjugate point. In the language of partial differential equations, a simple cusp point generates a shock in the solutions to the Hamilton-Jacobi-Bellman equation while fold points will not be part of the synthesis of optimal controls near the simple cusp point.
Introduction.We study singularities in solutions to the Hamilton-JacobiBellman equation for the value-function of an optimal control problem for ordinary differential equations. It is well known (see, for instance, [3]) that the necessary conditions for optimality given in the Pontryagin maximum principle [28] also give the characteristic equations for the Hamilton-Jacobi-Bellman equation. Thus, if the flow of extremals (trajectories which satisfy the necessary conditions of the Pontryagin maximum principle) covers an open region R of the state-space diffeomorphically, then a smooth solution to the Hamilton-Jacobi-Bellman equation can be constructed on R by the method of characteristics. In general, however, except for special problems like the linear-quadratic regulator, the value function is typically not smooth. The difficulties in finding solutions to optimal control problems or, equivalently, in finding solutions to the Hamilton-Jacobi-Bellman equation, precisely lie in analyzing the singularities.There is a large body of literature on singularities of solutions to the HamiltonJacobi-Bellman equation. Most of these papers deal with general topological properties of the singular set or try to establish smoothness of solutions. In [17] Fleming proves that the singular set is closed and of Hausdorff dimension at most equal to the dimension of the state space. Dafermos [16] analyzes singular sets for more general hyperbolic conservation laws, but only in dimension one. Cannarsa and Soner [11] analyze the local structure of the singular set for viscosity solutions which satisfy cer-