Rounding error in evaluating continued fraction expansions
William B. Jones, W. J. Thron · 1974
It is well known that continued fraction expansions provide a useful means for representing and computing values of functions. Expansions for many functions of mathematical analysis and physics are contained in the literature [1, 7, 9, 10]. Other expansions can be developed from a Taylor series (convergent or asymptotic) by efficient non-linear sequence algorithms [4, 5]. In addition to questions of convergence and speed of convergence of an infinite continued fraction