Extension of the discretization bounds to general linear switching systems

Let A{\cal A} be a set of d×dd\times d matrices, let rr be the maximal spectral radius of the matrices, let ε\varepsilon be the prescribed accuracy, and let K(r,ε)K(r,\varepsilon) denote the critical discretization step size for the associated linear switching system. The preceding bounds were established under the assumption that all matrices in A{\cal A} have real spectra. General-spectrum conjecture. Theorem 10 and Corollary 10 remain true for general matrices, without the real-spectra assumption; in particular, the bound

K(r,ε)6ε(16d224d+11)r2K(r,\varepsilon)\geq \frac{6\varepsilon}{\bigl(16d^2-24d+11\bigr)r^2}

and the corresponding stability criterion for the discretization step remain valid. This would extend the paper's estimates for the discretization parameter from real-spectrum linear switching systems to arbitrary matrix spectra, a direction identified by the authors as a step toward the general problem.

Sources & referencesView supporting material

Primary source

Vladimir Yu. Protasov and Raphael M. Jungers, “Stability of linear switching systems and Markov-Bernstein inequalities for exponents”, arXiv:1407.3927 (2014).

Progress summary

Never refreshed

Nothing recorded yet. Refresh searches the literature and the public web for attempts on this problem, and writes the first summary here.

Solutions 0

No solutions have been posted yet.