On sign-real spectral radii and sign-real expansive matrices (Q6087879)

Let \({\mathbb M}_n\) be the space of \(n\times n\) real matrices. The sign-real spectral radius of \(A\in {\mathbb M}_n\) is \[ \xi(A):=\max\{\lambda\in {\mathbb R}_+ : |Ax|\ge \lambda |x| \mbox{ for some } x\in {\mathbb R}^n\setminus\{0\}\}, \] where \(|x|:=(|x_1|, \dots, |x_n|)^\top\) if \(x = (x_1, \dots, x_n)^\top\in {\mathbb R}^n\) and the vector inequality \(x\le y\) is defined component-wise. It has been studied by researchers since its introduction by \textit{S. M. Rump} [Linear Algebra Appl. 266, 1--42 (1997; Zbl 0901.15002)] in 1997. The notion is related to estimating the component-wise distance to singularity. The function \(\xi\) bears some resemblance of or is related to the unitarily invariant matrix norms in some ways. A matrix \(A\in {\mathbb M}_n\) is called sign-real expansive if \(\xi (A) \ge 1\). Let \[ \Omega_n =\{ A\in {\mathbb M}_n: \xi(A) \ge 1\} \] be the set of sign-real expansive matrices. The authors obtain new properties of \(\xi\) and study the structure of \(\Omega_n\), such as checking membership based on computing principal minors. Another method presented is new volumetric technique for testing sign-real expansiveness. Some topological and invariance properties are given, for example, they prove that \(\Omega_n\) is a regular closed, unbounded, path-connected, and semi-algebraic set. They come up with a conjecture which is stronger than the conjecture in [\textit{S. M. Rump}, ``100 Euro Problem'', \url{https://www.tuhh.de/ti3/rump/100EuroProblem.pdf}] and further provide evidence supporting their conjecture.