Efficient algorithm for a class of least squares estimation problems

Abstract

A computationally efficient algorithm for solving least squares estimation problems is proposed. It is well suited for problems with the normal equation matrix factorizable in terms of Kronecker’s products. Three classes of identification problems factorizable in this sense are pointed out. Computational complexity of the algorithm and its robustness against round off errors is also discussed.

Publication
IEEE Transactions on Automatic Control
Wojciech Myszka
Wojciech Myszka
Site administrator

I’m just looking around…