Optimality of central and projection algorithms for bounded uncertainty (Q1100150)

The authors study the optimality in the sense of Traub and Wozniakowski [see \textit{J. F. Traub} and \textit{H. Wozniakowski}, A general theory of optimal algorithms (1980; Zbl 0441.68046)] of central and projection algorithms for identification of linear systems subject to a norm-bounded noise. It is shown that the least squares algorithms are strongly optimal, while other projection algorithms such as least-absolute-value and minimax algorithms do not possess this property.