A global minimum search algorithm for estimating the distance to uncontrollability (Q1260792)

For the differential system \(\dot x=Ax+Bu\), the measure \(\mu(A,B)\) of the nearness of a controllable pair to an uncontrollable one is given by \((A,B)=\min\sigma_{\min}((A-\lambda I,B))\) with respect to \(\lambda\), where \(\sigma_{\min}\) holds for the smallest singular value of a matrix. By using simple properties due to \textit{R. Byers} [SIAM J. Sci. Stat. Comput. 9, No. 5, 875-881 (1988; Zbl 0658.65044)], the problem is converted into a minimization problem on a bounded region in the plane. The algorithm is described, and then a bisection algorithm is given for estimating the distance to uncontrollability within an error \(\varepsilon\). The error analysis of the approach is carefully expanded and a numerical example is provided. These algorithms are based upon partitions of regions.