Comments, with reply, on "Convergence properties of Gram-Schmidt and SMI adaptive algorithms" by K. Gerlach and F.F. Kretschmer, Jr

The commenter summarizes Q-R decomposition techniques for solving the least squares (LS) problem and comments on associated aspects of the work presented by K. Gerlach and F.F. Kretschmer, Jr. (ibid., vol.267, no.1, Jan.90). In response to the commenter's statement that the statistical properties of the LS that determine the convergence performance are well known. Gerlach and Kretschmer assert that this is true only under the assumptions that have been used in the past to analyze the convergence performance of the canceler and for only a limited number of convergence performance measures. Gerlach and Kretschmer also address the commenter's points on overmatching degrees of freedom.<
>