On the existence of a common quadratic Lyapunov function for a rank one difference (Q865404)

A result of \textit{R. N. Shorten} and \textit{K. Narendra} [IEEE Trans. Automat. Control 48, No.~4, 618--621 (2003)] is discussed, which states that if \(A\) and \(B\) are real Hurwitz matrices and \(A-B\) is rank one, then the following are equivalent: (A) the pair \((A,B)\) has a common quadratic Lyapunov function (CQLF); (B) the matrix product \(AB\) has a real negative eigenvalue. A new proof is provided, which is based on results from convex analysis and the theory of moments and seems to be more natural from the viewpoint of matrix analysis. This proof includes the main theorem of the paper, which gives two new equivalent conditions for the existence of a CQLF, namely (C) there are not non-zero, positive semidefinite matrices \(X\) and \(Y\) such that \[ AX+XA^T+BY+YB^T=0; \] (D) there are not non-zero vectors \(v\) and \(w\) such that \[ Avv^T+vv^TA^T+Bww^T+ww^TB^T=0. \] This new approach can lead to further insights for the general CQLF problem.