A Convergence Acceleration Method based on a good Estimation of the Absolute value of the Error
Ana C. Matos · IMA Journal of Numerical Analysis · 1990
Suppose a sequence (Sn)n, converges to a limit S. In this paper we construct, from a good estimate (Dn)n of (∣Sn−S∣), an acceleration method for (Sn)n based on: (a) an automatic selection between two transformations if (Dn)n converges logarithmically; (b) composite sequence transformations if (Dn)n converges linearly.