Nonconvex Duality in Optimal Control

From MaRDI portal





The authors consider the optimal control problem of minimizing \(\ell(T,x(T))\) over all solutions \(x(t)\) of a differential inclusion and all end times \(T\geq 0\). The main result in the paper states that the optimal value function of the problem is the supremum \(\sup_\psi \psi(0,x(0))\) over all locally Lipschitz functions \(\psi\) satisfying the proximal Hamilton-Jacobi inequality \(\bar h(x,\partial_L\psi(t,x))\geq 0\) for some suitably defined extended Hamiltonian \(\bar h\) and the boundary condition \(\psi(t,x)\leq \ell(t,x)\). The paper gives a self-contained proof of this result under mild regularity hypotheses on the underlying differential inclusion and discusses a number of consequences and special cases.











This page was built for publication: Nonconvex Duality in Optimal Control

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q5317107)