A stationary iterative pseudoinverse algorithm (Q1267019): Difference between revisions

If the matrix of a linear least squares problem is rank-deficient, then many methods for solving this problem fail to produce the unique solution. In this paper an iterative scheme is described that overcomes this difficulty for the case of stationary methods. The algorithm combines two stationary methods: the first one produces any least squares solution, the second one produces the minimum norm solution of a consistent system.