No label defined (Q4033308)

This paper deals with the local controllability of a dynamical system for three classes of admissible variations. The general form of a class of optimal control problem is \[ J(u)=\varphi(x(t^*))\to\min,\quad\dot x=f(x)+Bu,\quad x(0)=x_ 0,\quad x=x(t),\tag{1} \] where \(u=u(t)\) belongs to a certain class of permissible controls. The author defines the ``local controllability along \(u(t)\), \(t\in T\), with respect to the Hamiltonian \(H(x,\psi,u)=\psi'(f(x)+Bu)\) as the existence of such a number \(\varepsilon_ 0>0\) that for every aggregation \(\omega=(\omega^ j_ l,\;l\in k_ j,\;j\in L)\) there exists a function \(\Delta u(t,\omega,\varepsilon)\) so as for \(0<\varepsilon\leq\varepsilon_ 0\) the equality \(\rho_ l(t_ j)=\varepsilon\omega^ j_ l\) holds, where \(\rho\) means the cocontrol and \(k_ j\) a corresponding set of indices. By linearization the problem (1) becomes the controllability of the dynamical system \[ \dot\chi=A(t)\chi+B\mu,\quad\chi(0)=0,\quad A(t)=A(x(t)).\tag{2} \] The main result is the theorem: The dynamical system (1) is locally controllable along \(u(t)\), \(t\in T\), with respect to the Hamiltonian over \(\{T_ *,K_ *\}\), if the linearized model (2) is controllable, here \(T_ *\) means a set of points \(\{t_ j,\;j\in Z\}\) and \(K_ *=\bigcup_{j\in L} K_ j\). The admissible variations considered are those consisting of (1) Dirac functions, (2) needle-shaped functions and (3) a kind of basical functions cited from E. B. Lee and L. Markus.