The numerical computation of connecting orbits in dynamical systems: A rational spectral approach (Q1326692): Difference between revisions
From MaRDI portal
Set profile property. |
Set OpenAlex properties. |
||
Property / full work available at URL | |||
Property / full work available at URL: https://doi.org/10.1006/jcph.1994.1070 / rank | |||
Normal rank | |||
Property / OpenAlex ID | |||
Property / OpenAlex ID: W2169070947 / rank | |||
Normal rank |
Revision as of 18:13, 19 March 2024
scientific article
Language | Label | Description | Also known as |
---|---|---|---|
English | The numerical computation of connecting orbits in dynamical systems: A rational spectral approach |
scientific article |
Statements
The numerical computation of connecting orbits in dynamical systems: A rational spectral approach (English)
0 references
1994
0 references
This paper presents a numerical method to approximate connecting orbits (i.e. orbits which connect two stationary points) in dynamical systems governed by ordinary differential equations. Since the posed problem is a boundary value problem in an infinite interval, usual procedures replace the infinite interval by a finite interval and modify the end conditions appropriately. However in the approach proposed here the problem is solved numerically in the infinite interval by using a spectral type method with a suitable set of test functions. Thus the difficulties derived from the truncation of the interval are avoided. As basis functions the authors take \(R_ n(t) = \cos(n \cot^{-1}(t))\), \(n = 0,1,\dots\) which are orthogonal in the real line with respect to the weighting function \((1 + t^ 2)^{-1}\) and the collocation points are \(t_ j = \cot(j\pi / (M + 1))\), \(j = 1,\dots,M\). Moreover a new phase condition, which does not require information from the previous step, is proposed. Under these conditions and using a suitable scaling factor the nonlinear equations arising in the spectral approximation are solved numerically by a standard subroutine. To show the efficiency of the new method two numerical examples derived from Huxley- and Lorentz-equations are considered. It is seen that even with a small number of collocation points the new method provides more accurate results and requires a lower computational cost than other standard methods.
0 references
spectral method
0 references
connecting orbits
0 references
dynamical systems
0 references
collocation
0 references
scaling
0 references
numerical examples
0 references
Huxley- and Lorentz-equations
0 references