On a minimax control problem (Q921564)

Let the motion of an n-dimensional vector x in \({\mathbb{R}}^ n\) be described by the equation \[ \dot x=f(t,x,u,v),\quad x(0)=x_ 0, \] where \(t\in I=[0,T]\), \(T>0\); \(u\in P\subset {\mathbb{R}}^ p\), \(v\in Q\subset {\mathbb{R}}^ q\), and P and Q are nonempty compact sets; the vector v is a model for the noises and perturbations, v(\(\cdot)\) belongs to the fixed nonempty class V of measurable functions such that v(t)\(\in Q\), \(t\in I\); the set U of admissible controls contains a measurable function u(\(\cdot)\) such that u(t)\(\in P\), \(t\in I.\) The function f(t,x,u,v) has the following properties: 1) f is continuous on \(I\times {\mathbb{R}}^ n\times P\times Q;\) 2) f is continuously differentiable with respect to \(x\in {\mathbb{R}}^ n\) for \(t\in I\), \(u\in P\), \(v\in Q;\) 3) the inequality \(<x,f(t,x,u,v)>\leq a(1+| x|^ 2)\) holds on \(I\times {\mathbb{R}}^ n\times P\times Q\) where a is a non-negative constant. The author investigates the problem: find an admissible control u(.)\(\in U\) minimizing the functional \[ J(u(\cdot))=\sup_{v(\cdot)\in V}\phi (x(T,u(\cdot),v(\cdot))) \] on U, where the scalar function \(\phi\) (x) is defined and continuously differentiable on \({\mathbb{R}}^ n\). The linear case of this problem is also considered.