Interval linear systems: the state of the art (Q880921)

The paper is devoted to the solution of a system of linear equations \(Ax=b\) in the case when \(A\) and \(b\) are a matrix and a vector with interval valued entries. In fact, \(A\) and \(b\) are considered as sets of values, e.g. \(A=\{a\in \mathbb R^{m\times n}:\;\underline A\leq a\leq \overline A\}\), where \(\underline A\) and \(\overline A\) are some matrices with real entries and the relation \(\leq\) is to be understood componentwise. The authors describe some characterizations of the set of solutions to \(Ax=b\) and consider computational algorithms for finding the interval hull for this set, i.e., the smallest interval to which all solutions to \(ax=\beta\) for all \(a\in A\), \(\beta\in b\) belong. Three methods are considered: the Krawczyk method, the interval Gauss-Seidel method and the interval Gauss elimination. Results of numerical simulations, and a list of available software tools are presented.