Partial Padé prediction

When the first terms of a sequence (called sequence to predict) are known, a prediction method is a method which gives us an approximation of the following terms (the so-constructed sequence is called the predicted sequence). In this paper, we state two prediction methods, respectively called εp-pre...

Ausführliche Beschreibung

Gespeichert in:
Bibliographische Detailangaben
Veröffentlicht in:Numerical algorithms 1999-01, Vol.20 (1), p.23-50
Hauptverfasser: Prévost, M, Vekemans, D
Format: Artikel
Sprache:eng
Schlagworte:
Online-Zugang:Volltext
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
Beschreibung
Zusammenfassung:When the first terms of a sequence (called sequence to predict) are known, a prediction method is a method which gives us an approximation of the following terms (the so-constructed sequence is called the predicted sequence). In this paper, we state two prediction methods, respectively called εp-prediction and partial Padé prediction, which are generalizations of Aitken's Δ2-prediction of Brezinski and Redivo-Zaglia [6] and Padé prediction of Gilewicz [8] which are very simple to use. In order to choose among the different partial Padé predictions, we study some of their properties. The most important points of this paper are:• the use of an extrapolation algorithm (the ε-algorithm), to obtain a prediction algorithm for each partial Padé prediction (which avoids solving a system); • the results about consistency obtained for the partial Padé prediction (i.e., under certain conditions, each term of the predicted sequence converges to the analogous term of the sequence to predict).
ISSN:1017-1398
1572-9265
DOI:10.1023/A:1019187330365