No label defined (Q4237484)

The paper investigates necessary and sufficient conditions of optimality for the control problem: \(\dot x \in F(x,t)\), \(t\in I=[t_1,t_2]\); \(p\in P\), \(p=(t_1,t_2,x_1,x_2)\), \(x_1=x(t_1)\), \(x_2=x(t_2)\); \(x(t)\in G\) \(\forall t\in I\); \(J(p)\to\min\); where \(x\in R^n\), \(t\in R^1\), the multivalued mapping \(F\) takes convex compact values from \(R^n\) and is locally Lipschitzian; \(P\), \(G\) are nonempty closed subsets of \(R^{2n+2}\) and \(R^n\), respectively. The necessary optimality conditions generalize Pontryagin's maximum principle. They contain the Hamiltonian \(H(F(x,t),\psi)= \max\{\langle f,\psi\rangle: f\in F(x,t)\}\) and the Borel vector measure \(\nu\) with the norm \(\| \nu\|=\sup_{\| x\|_{C(I)=1}}\int_{t_1}^{t_2} x(t)d\nu\). The questions of nondegeneracy of the maximum principle have been solved. The sufficient conditions of the absence of a singular component of the measure are derived. The special autonomous time-optimal problem is studied, too. For the entire collection see [Zbl 0910.00033].