No label defined (Q4353365)

The authors deal with the optimal control problem \(\;J(x(t_1))\to \max\), \(\dot x(t)=A(t)x(t)+f(t,u(t))\), \(x(t_0)=x^0\); \(u\in \mathcal U=\{u\in L_\infty^r\mid u(t)\in U\) a.e. in \(T\}\); where \(T=[t_0,t_1]\), \(J:\mathbb R^n\to \mathbb R\) is convex and differentiable; \(x^0\), \(x(t)\in \mathbb R^n\) \(\forall t\in T\); \(U\) is compact in \(\mathbb R^n\), \(A(t)\) is an \((n\times n)\) matrix with components from \(L_\infty\), \(f(.,.)\) is continuous, \(f(.,u(.))\) is integrable for every \(u\in \mathcal U\). Let \(u_0\) be a control satisfying the Pontryagin's maximum principle. The control \(w\in \mathcal U\) is globally optimal iff \(\forall y\in \mathbb R^n\) with \(J(y)=J(x(t_1,w))\) the inequality \(\langle\nabla J(y),u_0(t_1,y)-y\rangle\leq 0\) holds. The global optimal control search algorithm is described. The authors construct the approximate resolving set \(\mathcal R(u^*,\varepsilon)\) enabling to escape from a stationary control \(u^*(.)\) satisfying the Pontryagin's maximum principle. An algorithm in 7 steps based on global optimality conditions is established. For the entire collection see [Zbl 0865.00047].