Fuzzy LR linear systems: quadratic and least squares models to characterize exact solutions and an algorithm to compute approximate solutions (Q344046)

The authors study a linear system of the form \(Ax=b\), where \(A\) is a rectangular matrix with real entries, \(b\) is an LR fuzzy vector and \(x\) is, in general, an unknown LR fuzzy vector. Necessary and sufficient conditions are derived for this system to have a solution. One of the characterizations is that a certain linear programming problem (defined by means of the matrix \(A\) and the vector \(b\)) has an optimal solution with the optimal value being zero. Another characterization involves the non-solvability of a system of linear inequalities, much like Farkas' lemma. A notion of an approximate solution using a least squares model is then proposed. The question of uniqueness of an approximate solution is considered. It is shown how such an approximate solution could be computed by solving a certain quadratic programming problem, leading to an algorithmic approach. Numerical examples are presented to illustrate the ideas.