2006
DOI: 10.1016/j.cam.2005.07.001
|View full text |Cite
|
Sign up to set email alerts
|

RCMS: Right Correction Magnus Series approach for oscillatory ODEs

Abstract: We consider RCMS, a method for integrating differential equations of the form y = [ A + A 1 (t)]y with highly oscillatory solution. It is shown analytically and numerically that RCMS can accurately integrate problems using stepsizes determined only by the characteristic scales of A 1 (t), typically much larger than the solution "wavelength". In fact, for a given t grid the error decays with, or is independent of, increasing solution oscillation. RCMS consists of two basic steps, a transformation which we call … Show more

Help me understand this report

Search citation statements

Order By: Relevance

Paper Sections

Select...
3
1
1

Citation Types

0
49
0

Year Published

2008
2008
2022
2022

Publication Types

Select...
5
1

Relationship

0
6

Authors

Journals

citations
Cited by 14 publications
(49 citation statements)
references
References 18 publications
(57 reference statements)
0
49
0
Order By: Relevance
“…Thus, to compute y i = e hĀ e σ(h) y i−1 with h = h i , we need to approximate σ(h) by truncating the expansion (27) and replacing integrals by quadrature (see 4.3). As shown in [26], truncating all but the first integral leads to a fourth order method, while including also σ 2 gives us a scheme of order eight. Having approximated σ(h), its 2 × 2 matrix exponential must be computed.…”
Section: Modified Neumann and Magnus Schemes For The Schrödinger Equamentioning
confidence: 99%
See 1 more Smart Citation
“…Thus, to compute y i = e hĀ e σ(h) y i−1 with h = h i , we need to approximate σ(h) by truncating the expansion (27) and replacing integrals by quadrature (see 4.3). As shown in [26], truncating all but the first integral leads to a fourth order method, while including also σ 2 gives us a scheme of order eight. Having approximated σ(h), its 2 × 2 matrix exponential must be computed.…”
Section: Modified Neumann and Magnus Schemes For The Schrödinger Equamentioning
confidence: 99%
“…When scanning over λ we have used only three values, those such that Z(h i ) = (q − λ)h 2 i = −m 2 π 2 , m = 0, 1, 2. The selection of only these was mainly intended to speed up the evaluation but this is enough as the error decreases like O(1/ √ λ) for large Z-values [26,28], and as confirmed by experimental tests showing that the error is indeed larger for smaller values of Z.…”
Section: Some Experimentsmentioning
confidence: 99%
“…1-2 possess a unique countable family of solutions {λ j , y λ j } j ∈Z + 0 and that its eigenvalues are simple, bounded from below and accumulate only at infinity as well as that its eigenfunctions oscillate as functions of the eigenvalues [13], namely λ j < λ j +1 , lim j →+∞ (λ j /j 2 ) = (π/(b − a)) 2 , y λ j has exactly j zeros in (a, b).…”
Section: Uniform Versus Asymptotic Error Estimatesmentioning
confidence: 99%
“…e.g., in the piecewise perturbation methods [7], in the right-correction Magnus series [2] and in the modified Magnus methods [9]. In addition, one can consider another regime.…”
Section: Uniform Versus Asymptotic Error Estimatesmentioning
confidence: 99%
See 1 more Smart Citation