A note on the existence of positive realizations (Q1886543)

This paper contains the following result: Let \(G(s)\) be a strictly proper rational transfer function of degree \(n\). Then (A) and (B) are equivalent: (A) \(G(s)\) admits an asymptotically stable minimal positive diagonal realization. (B) \(G(s)\) is asymptotically stable, minimum phase and the poles and zeros of \(G(s)\) are real and interlaced. The authors also prove that (A) and (B) are equivalent to a statement about two submatrices of the infinite Hankel matrix associated to \(G(s)\). A similar result is given for a discrete-time transfer function. The above result is also extended to possibly unstable transfer functions in continuous time. The above results are applied to a number of other cases as well, leading for instance to the following corollary: Let \(d(s)\) be a monic polynomial with all real roots. Then \(G(s):=d(s)^{-1}\) admits a minimal positive realization. The well written paper ends with a number of examples and an adequate list of references.