Step-by-step averaging of linear differential inclusions of variable dimension on a finite interval (Q723519)

The paper considers linear differential inclusions with small perturbations of the form \[ x'(t)\in \varepsilon N(n(t))A(t)x(t)+\varepsilon N(n(t))F(t)\quad t\neq \tau _i, \] \[ x(0)=x_0,\quad x(\tau _i)=M(n(\tau _i))x(\tau _i-0), \] where \(\varepsilon>0\), \(I=[0,T]\), \(n:I\to {\mathbb N}\) is a piecewise constant right-continuous function bounded by a constant \(k>0\), \(\tau _i\in I\), \(\tau _i<\tau_{i+1}\), \(i=1,2,...,m\), \(n(\tau _i)\neq n(\tau _i-0)\), \(N(t),A(t)\) and \(M(t)\) are some particular \(k\times k\) matrix-valued functions and \(F(.):I\to {\mathcal P}({\mathbb R}^n)\) is a set-valued map with nonempty compact convex values. Under certain hypotheses it is proved that, for \(\varepsilon \) small enough, the Hausdorff distance between the set of solutions of the problem considered and the set of solutions of the associated averaged problem is less than \(\varepsilon \).