Enclosing the solution set of linear systems with inaccurate data by iterative methods based on incomplete LU-decompositions (Q5902744)

We present a class of iterative methods to enclose the solution set \(\{\) \(x|\) \(Ax=b\), \(A\in {\mathcal A}\), \(b\in b\}\) by an interval vector; A is varying in an \(n\times n\) interval H-matrix \({\mathcal A}\) and b is varying in an interval vector \(b\). The algorithm taken into consideration generalizes an iterative method of \textit{J. Meijerink} and \textit{H. van der Vorst} [Math. Comput. 31, 148-162 (1977; Zbl 0349.65020)] based on an incomplete LU-decomposition of an M-matrix A. Theorems concerning the feasibility of the algorithm, its rate of convergence and its quality of enclosure are given. Since the original method of Meijerink and van der Vorst is a special case of our algorithm we have thus shown its applicability to the larger class of H-matrices. Furthermore we relate the \(R_ 1\) factor (as defined by \textit{J. Ortega} and \textit{W. Rheinboldt} [Iterative solution of nonlinear equations in several variables (1970; Zbl 0241.65046)]) of the original method to the underlying set P of indices.