Approximation properties of a mapped Chebyshev method (Q1963918)

A mapped Chebyshev technique introduced by \textit{D. Kosloff} and \textit{H. Tal-Ezer} [J. Comput. Phys. 104, No. 2, 457-469 (1993; Zbl 0781.65082)] is analyzed. For suitable choices of parameters the mapped Chebyshev methods yield better approximation results for derivatives than the standard collocation method. Error bounds are presented which show that for the parameters suggested in the literature a low order of approximation occurs. However, this observation only appears for levels of accuracy much more stringent than those in usual computations.