Abstract
The paper aims at answering the following question, in the scalar as well in the vector case: What do the famous Aitken’s \(\Delta ^{2}\) and Wynn’s \(\varepsilon \)-algorithm exactly do with the terms of the input sequence? Inspecting the rules of these algorithms from a geometric point of view leads to change the question into another one: By what kind of geometric object can the parallel (or harmonic) sum be represented? Thus, the paper begins with geometric considerations on the parallel addition and the parallel subtraction of vectors, including equivalent definition, and properties derived from the new point of view. It is shown how the parallel sum of vectors is related to the Bézier parabola controlled by these vectors and to their interpolating equiangular spiral. In the second part, observing that Aitken’s \(\Delta ^{2}\) and Wynn’s \(\varepsilon \)-algorithm may be defined through hybrid sums, mixing standard and parallel sums, the consequences are drawn regarding the way one can define and analyze these algorithms. New explanatory rules are derived for the \(\varepsilon \)-algorithm, providing a better understanding of its basic step and of its cross-rule.
Similar content being viewed by others
References
Aitken, A.C.: On Bernoulli’s numerical solution of algebraic equations. Proc. R. Soc. Edinb. 46, 289–305 (1926)
Anderson, W.N., Trapp, G.E.: The harmonic and geometric mean of vectors. Lin. Multilin. Alg. 22, 199–210 (1987)
Berlinet, A.F., Roland, C.: Geometric interpretation of some Cauchy related methods. Numer. Math. 119, 437–464 (2011)
Berlinet, A.F., Roland, C.: Acceleration of the EM algorithm: P-EM versus epsilon algorithm. Comp. Stat. Data An. 56, 4122–4137 (2012)
Brezinski, C., Redivo Zaglia, M.: New vector sequence transformations. Linear Algebra Appl. 389, 189–213 (2004)
Brezinski, C., Redivo Zaglia, M.: Generalizations of Aitken’s process for accelerating the convergence of sequences. Comput. Appl. Math. 26, 171–189 (2007)
Coxeter, H.S.M.: Introduction to geometry. Wiley, New York (1969)
Euler, L.: De Centro Similitudinis, Nova Acta Academiae Scientiarum Imperialis Petropolitaneae 9, 154–169 (1795)
Graves-Morris, P.R.: Vector valued rational interpolants. Numer. Math. 42, 331–348 (1983)
Graves-Morris, P.R.: Extrapolation methods for vector sequences. Numer. Math. 61, 475–487 (1992)
Henrici, P.: Elements of Numerical Analysis. Wiley, New York (1964)
Lee, E.T.Y.: The rational Bézier representation for conics. In: Farin, G.E. (ed.) Geometric Modeling: Algorithms and New Trends, pp. 3–19. SIAM, Philadelphia (1987)
Le Ferrand. H.: Une généralisation au cas vectoriel du procédé Δ2 d’Aitken et les suites à comportement linéaire. RAIRO, Modélisation mathématique et analyse numérique 29, 53–62 (1995)
MacLeod, A.J.: Acceleration of vector sequences by multi-dimensional Δ2 methods. Commun. Appl. Numer. Meth. 2, 385–392 (1986)
McLeod, J.B.: A note on the epsilon-algorithm. Computing 7, 17–24 (1971)
Sadok, H.: About Henrici’s transformation for accelerating vector sequences. J. Comput. Appl. Math. 29, 101–110 (1990)
Salam, A.: Non-commutative extrapolation algorithms. Numer. Alg. 7, 225–251 (1994)
Salam, A., Graves-Morris, P.R.: On the vector ε-algorithm for solving linear systems of equations. Numer. Alg. 29, 229–247 (2002)
Steele, J.A., Dolovich, A.T.: Toward the kernel of the vector ε-algorithm. Int. J. Numer. Meth. Eng. 48, 721–730 (2000)
White, H.S.: Leonard Euler on the Center of Similitude. http://www.math.dartmouth.edu/~euler/Estudies/E693t.pdf
Wynn, P.: Acceleration techniques for iterated vector and matrix problems. Math. Comp. 16, 301–322 (1962)
Author information
Authors and Affiliations
Corresponding author
Rights and permissions
About this article
Cite this article
Berlinet, A.F. Geometric approach to the parallel sum of vectors and application to the vector \(\varepsilon \)-algorithm. Numer Algor 65, 783–807 (2014). https://doi.org/10.1007/s11075-013-9714-y
Received:
Accepted:
Published:
Issue Date:
DOI: https://doi.org/10.1007/s11075-013-9714-y