Extrapolation algorithms based on the asymptotic expansion of the inverse of the error; application to continued fractions (Q756352)

The author considers a convergent sequence \((S_ n)\) with a limit S and supposes that an asymptotic expansion of \((S_ n-S)^{-1}\) is known. On the basis of the algorithm published by \textit{C. Brezinski} [Numer. Math. 35, 175-187 (1980; Zbl 0444.65001)] an extrapolation scheme \((G_ n^{(j)})_ n\) \((j>1)\) is proposed such, that \(G_ n^{(j)}-S=o(G_ n^{(j-1)}-S)\). This is applied in the case, that \((S_ n-S)^{- 1}\approx a_ 1n^{\alpha_ 1}+a_ 2n^{\alpha_ 2}+...\) where \(\alpha_ 1>0\) and the first term in this expansion is dominant. The obtained extrapolation scheme is used for acceleration of continued fractions.