A new solution for the least squares problem

Abstract

This paper presents a new iterative method to obtain the generalized inverse of a continuous linear mapping. This method gives the solution of the least squares problem and, from numerical point of view, it is a very efficient one.

Our method generalizes the results from [3] and [6] and it is not based on the QR or SVD decompositions ([1, 5]).

Keywords:

• Least squares problem
• Hilbert space
• generalized inverse

AMS (MOS):

• 47A58
• 46N10
• 65F20

C.R. Categories:

• G.1.6
• G.1.3