Detail publikace
Predictor–corrector Obreshkov pairs
SEHNALOVÁ, P. BUTCHER, J.
Anglický název
Predictor–corrector Obreshkov pairs
Typ
Článek recenzovaný mimo WoS a Scopus
Jazyk
en
Originální abstrakt
The combination of predictor–corrector (PEC) pairs of Adams methods can be generalized to high derivative methods using Obreshkov quadrature formulae. It is convenient to construct predictor–corrector pairs using a combination of explicit (Adams–Bashforth for traditional PEC methods) and implicit (Adams–Moulton for traditional PEC methods) forms of the methods. This paper will focus on one special case of a fourth order method consisting of a two-step predictor followed by a one-step corrector, each using second derivative formulae. There is always a choice in predictor–corrector pairs of the so-called mode of the method and we will consider both PEC and PECE modes. The Nordsieck representation of Adams methods, as developed by C. W. Gear and others, adapts well to the multiderivative situation and will be used to make variable stepsize convenient. In the first part of the paper we explain the basic approximations used in the predictor–corrector formula. Those can be written in terms of Obreshkov quadrature. Next section we discuss the equations in terms of Nordsieck vectors. This provides an opportunity to extend the Gear Nordsieck factorization to achieve a variable stepsize formulation. Numerical tests with the new method are also discussed. The paper will present Prothero–Robinson and Kepler problem to illustrate the power of the approach.
Klíčová slova anglicky
PEC methods, Adams methods, Nordsieck representation, Ordinary differential equations, Numerical methods
Vydáno
2013-01-10
Nakladatel
Springer-Verlag Wien
Místo
AT
ISSN
0010-485X
Časopis
COMPUTING
Ročník
95
Číslo
5
Strany od–do
355–371
Počet stran
17
BIBTEX
@article{BUT97130,
author="Pavla {Sehnalová} and John {Butcher}",
title="Predictor–corrector Obreshkov pairs",
journal="COMPUTING",
year="2013",
volume="95",
number="5",
pages="355--371",
doi="10.1007/s00607-012-0258-0",
issn="0010-485X",
url="http://link.springer.com/article/10.1007%2Fs00607-012-0258-0"
}