A note on the stability of Toeplitz matrix inversion formulas. (Q1767145)

It is shown that the algorithms of \textit{A. Ben-Artzi} and \textit{T. Shalom} [Linear Algebra Appl. 75, 173--192 (1986; Zbl 0586.15005)], \textit{G. Labahn} and \textit{T. Shalom} [ibid. 175, 143--158 (1992; Zbl 0760.15005)] and \textit{M. K. Ng, K. Rost} and \textit{Y.-W. Wen} [ibid. 348, No. 1--3, 145--151 (2002; Zbl 0998.15031)] for the inversion of a well conditioned Toeplitz matrix are forward stable. I.e., the difference between the computed inverse and the exact inverse has a finite \(2\)-norm.