Some ergodic results on stochastic iterative discrete events systems (Q1360356)

The paper deals with the asymptotic behaviour of a class of discrete event dynamic systems. The class is characterized by a recurrence relation in \(\mathbb{R}^K\) of the form \(T(n+1)= \varphi_{n+1} (T(n))\), where \(\varphi_n\) is a random operator which is monotone and 1-linear. (An operator \(\varphi: \mathbb{R}^K \to \mathbb{R}^K\) is linear if \(\varphi(x) < \varphi(y)\) when \(x_i<y_i\) for all \(i\), and 1-linear if \(\varphi (x+\lambda 1) =\varphi (x)+ \lambda 1\) for all scalars \(\lambda\).) By an application of Kingman's subadditive ergodic theorem, it is shown that the largest and smallest components of the stochastic vector \(T(n)\) behave asymptotically linearly in \(n\). Some bounds for the constants involved in the asymptotic relations are given. The author also derives some stability properties of the systems at hand and discusses several examples.