Optimal Solution Properties of an Overdetermined System of Linear Equations

Abstract

The paper considers the solution properties of an overdetermined system of linear equations in a given norm. The problem is observed as a minimization of the corresponding functional of the errors. Presenting the main results of $p$ norm it is shown that the functional is convex. Following the convex properties we examine minimization properties showing that the problem possesses regression, scale, and affine equivariant properties. As an example we illustrated the problem of finding weighted mean and weighted median of the data.