Convergence Acceleration of Alternating Series

From MaRDI portal
Publication:4506286





This paper treats several algorithms to calculate the sum of an alternating series \(\sum_{k=0}^{\infty} (-1)^ka_k\) from the first \(n\) entries. The \(a_k\) have to be either the moments of a positive measure on \([0,1]\) or those of a weight \(w(x)\) on \([0,1]\) that extends analytically into the complex domain in a certain manner. In the first case the relative accuracy is of the order \(5.828^{-n}\); the algorithm uses \(O(1)\) storage and has a running time of \(O(1)\) for each \(a_k\) used. Chebyshev polynomials play a role in the construction of the algorithm. The second algorithm requires storage \(O(n)\) and running time \(O(n^2)\); the relative accuracy is of the order \(7.89^{-n}\) for a large class of series and \(17.93^{-n}\) for a smaller class. The conclusion is, that the choice of algorithm depends on the ease with which the terms of the series can be computed. A very interesting and nicely written paper.




Cited in
(35)








This page was built for publication: Convergence Acceleration of Alternating Series

Report a bug (only for logged in users!)Click here to report a bug for this page (MaRDI item Q4506286)