On the critical parameter value for the higher sigma-exponent of linear differential systems (Q2459702)

Consider the linear ODE \[ \frac{dx}{dt} = A(t) x \in \mathbb{R}^n,\;t \geq 0, \] with piecewise continuous, uniformly bounded matrix \(A \in \mathbb{R}^{n \times n}\) and the perturbed system \[ \frac{dy}{dt} = A(t) y + Q(t) y, y \in \mathbb{R}^n,\;t \geq 0 \] with piecewise continuous perturbation matrix \(Q\) satisfying the \(\sigma\)-condition \[ \exists N(Q)=\text{constant} \; \forall t \geq 0 : | | Q(t)| | \leq N(Q) \exp (-\sigma t) \] with constant \(\sigma >0\). The largest characteristic exponent of the perturbed system is denoted by \(\lambda_n (A+Q)\). The quantity \[ I_\sigma (A) := \sup_{Q_\sigma \in \mathbb{R}^{n\times n}} \lambda_n (A+Q_\sigma) \] taken over all exponentially decaying matrices \(Q_\sigma\) with \(\sigma\)-condition is called the Izobov exponent or higher \(\sigma\)-exponent according to the Russian-English translation of this paper (for fixed \(\sigma\)). The author obtains explicit formulas expressing the critical values \(\sigma_0 (A)\) of \(I_\sigma (A)\) in terms of the Cauchy matrix of the unperturbed system. For this purpose, he uses an algorithm of \textit{N. A. Izobov} [Differ. Uravn. 5, No. 7, 1186--1192 (1969; Zbl 0175.09902)] for the computation of the Izobov exponent.