Characterizing matrices with \(\pmb{X}\)-simple image eigenspace in max-min semiring. (Q2829127)

This paper deals with a problem of max-min linear algebras with arithmetical operations \(a\oplus b:=\max(a,b)\) and \(a\otimes b:=\min(a,b)\) and it is defined over a linearly ordered set. A matrix \(A\) in max-min algebra (fuzzy matrix) is called weakly robust if \(A^k\otimes x\) is an eigenvector of \(A\) only if \(x\) is an eigenvector of \(A\). In the present paper, the authors consider an interval version of this condition. Namely, they describe matrices \(A\) such that for any eigenvector \(y\) belonging to an interval \(\pmb{X}=\left[\underline{x},\overline{x}\right]:=\left\{ x\in\mathbb{B}(n): \underline{x}\leq x\leq\overline{x}\right\}\) the system \(A\otimes x=y\) has a unique solution \(x=y\) in \(\pmb{X}\). This is what they mean by saying that \(A\) has an \(\pmb{X}\)-simple image eigenspace. It is shown that under a certain natural condition, \(A\) has an \(\pmb{X}\)-simple image eigenspace if and only if it satisfies a nontrivial combinatorial criterion, which makes use of threshold digraphs and to which they refer as \(\pmb{X}\)-conformism.