\(X^{A E}\) and \(X^{E A}\) robustness of max-min matrices (Q2322878)

The paper deals with eigenvectors and eigenspaces of square matrices over a lattice (see also [\textit{K. Cechlárová}, Fuzzy Sets Syst. 138, No. 3, 627--641 (2003; Zbl 1075.05537); \textit{Y.-J. Tan}, Linear Algebra Appl. 283, No. 1--3, 257--272 (1998; Zbl 0932.15005); \textit{J. S. Golan}, Semirings and their applications. Dordrecht: Kluwer Academic Publishers (1999; Zbl 0947.16034)]). The most popular domain is a lattice \((B,0,1,\max,\min)\), where \(B\) is a bounded chain with bounds denoted by \(0\) and \(1\) (see [\textit{M. Gavalec}, Linear Algebra Appl. 345, No. 1--3, 149--167 (2002; Zbl 0994.15010)]) and some results can be generalized to the lattice of closed intervals from \(B\) (see [\textit{M. Gavalec} and \textit{J. Plavka}, Kybernetika 46, No. 3, 387--396 (2010; Zbl 1202.15013)]). In this paper, the authors consider a mixed problem, where interval eigenvectors for matrices over \(B\) are studied . This is done by creating \(n\) additional free variables, which can be bound by quantifiers. If at first, some variables are bound by a universal quantifier and next, other variables are bound by an existential quantifier, then we have a so-called AE-problem. In the opposite order, we have a so-called EA-problem. The authors apply known results concerning robustness of matrices over \(B\) for the above new problems with estimation of computational complexity of the presented algorithms.