On computing stabilizability radii of linear time-invariant continuous systems

In this paper we focus on a non-convex and non-smooth singular value optimization problem. Our framework encompasses the distance to stabilizability of a linear system (A, B) when both A and B or only one of them are perturbed. We propose a trisection algorithm for the numerical solution of the singular value optimization problem. This method requires O([n.sup.4]) operations on average, where n is the order of the system. Numerical experiments indicate that the method is reliable in practice. Key words. stabilizability radius, optimization, trisection algorithm, linear time-invariant continuous system AMS subject classifications. 65F15, 93D15, 65K10