An algorithm for generating transfer functions uniformly distributed over H∞ balls

Probabilistic methods have recently been the subject of considerable attention in the context of robust performance assessment. However, in spite of their potential, these methods have been limited to the case of parametric uncertainty; the problem of sampling causal bounded operators is largely open. In this paper, we take steps towards removing this limitation by providing a computationally efficient algorithm aimed at uniform sampling over balls contained in suitably chosen proper subspaces of H_∞. As shown in the paper, samples generated from these balls can be used, for instance, by Monte Carlo methods to assess robust performance for uncertainty models involving the H_∞ norm