Trigonometrically fitted multi-step RKN methods for second-order oscillatory initial value problems☆
Introduction
In this work we consider the effective numerical integration of the initial value problem of second-order differential equations in the form whose solution has an oscillatory character, where y ∈ Rd, f: [t0, T] × Rd → Rd is sufficiently differentiable and the first derivative y′ does not appear explicitly. Such a problem often arises in different fields of applied sciences such as celestial mechanics, molecular dynamics, quantum mechanics, spatial semi-discretizations of wave equations, electronics, and so on [1], [2], [3].
Considering the oscillatory feature of the problem (1), some researchers have proposed to develop numerical integrators with frequency-dependent coefficients by the techniques like trigonometrical/exponential fitting (see [4], [5]). Early presentations of these techniques come from Gautschi [6] and Lyche [7]. Since then, a lot of exponentially (or trigonometrically) fitted linear multi-step methods have been proposed and applied to other scientific fields. Recently, in the context of Runge–Kutta–Nyström (RKN) methods, exponentially-fitted and trigonometrically-fitted methods have been considered in [8], [9], [10], [11], [12]. Meanwhile mixed-collocation versions have been introduced by Coleman and Duxbury in [13]. All of these methods integrate exactly second-order systems (1) whose solution can be expressed as linear combination of functions from the set of functions or equivalently the set {cos (wt), sin (wt)}, (see [14], [15]). At the same time, taking the basic frame of two-step hybrid method given by Coleman [16], exponentially (or trigonometrically) fitted two-step hybrid method attracted a wide spread attention [17], [18], [19], [20], [21].
More recently, Li and Wang [22] proposed a family of multi-step Runge–Kutta–Nyström (MSRKN) methods and derived the corresponding order conditions by using the theory of SN-series. Compared with classical one-step RKN methods, MSRKN methods can reach higher order with the same function evaluations per step. In fact, MSRKN methods form a subclass of general linear methods studied by R. D’Ambrosio et al. [23] recently.
Inspired by the previous work, in this paper, we will extend the idea of trigonometrical fitting to MSRKN methods. The rest of this paper is organized as follows: In Section 2, we restate the general formulation of MSRKN methods for the problems (1). Trigonometrical fitting conditions are presented and algebraic order conditions for trigonometrically fitted MSRKN (TFMSRKN) methods are given. In Section 3, the stability properties are analyzed. With the order conditions, two new explicit trigonometrically fitted MSRKN methods of order five, are constructed in Section 4. In Section 5, numerical experiments are carried out and the numerical results show the robustness of the new methods. Section 6 is concerned with conclusions and discussions.
Section snippets
Trigonometrically fitted multi-step Runge–Kutta–Nyström methods
Multi-step Runge–Kutta–Nyström (MSRKN) methods for the problem (1) are defined by the formulas (see [22]) where ci, pik, qik, aij, uk, vk, wk and bi with are all real coefficients. The method (2) can be presented by the Butcher-type tableau
Stability
Applying a TFMSRKN method (2) to the following second-order homogeneous linear test mode [25], [26] the resulting recursion relation can also be expressed as where and with and given by Because the stability matrix M(H, v) depends on the variables H and v, geometrically, the characterization of stability
Construction of explicit methods
In this section, we focus our attentions on the construction of the explicit TFMSRKN methods. We consider the following explicit TFMSRKN methods with FSAL property defined by the table of coefficients For the coefficients of this method, the conditions (3) give the following fundamental relations Under the first two simplifying conditions of (13),
Numerical experiments
In this section, in order to show the competence and superiority of the new methods compared with the well-known methods in the scientific literature, we use four model problems. The criterion used in the numerical comparisons is the decimal logarithm of the global error (GE) versus the computational effort measured in the decimal logarithm of the number of function evaluations required by each method. The integrators we select for comparison are
- •
L1TFMSRKN2L4S5P: the four-stage
Conclusions and discussions
This paper proposes and studies the trigonometrically fitted multi-step RKN (TFMSRKN) methods for solving oscillatory initial value problem (1). We derived the necessary and sufficient conditions for an TFMSRKN method to be of algebraic order p via the theory of SN-series. Based on the order conditions, two new explicit two-step four-stage methods of order five were constructed. The numerical experiments carried out with several oscillatory problems show that for the problem with oscillatory
Acknowledgements
The authors are grateful to the two anonymous reviewers for their valuable suggestions, which help improve this paper significantly.
References (36)
- et al.
An optimized Runge–Kutta method for the solution of orbital problems
J. Comput. Appl. Math.
(2005) Exponentially fitted explicit Runge–Kutta–NystrÖm methods
J. Comput. Appl. Math.
(2004)Runge–Kutta–Nystrom methods for ODEs with periodic solutions based on trigonometric polynomials
Appl. Numer. Math.
(1998)- et al.
Extended RKN-type methods for numerical integration of perturbed oscillators
Comput. Phys. Commun.
(2009) - et al.
Mixed collocation methods for
J. Comput. Appl. Math.
(2000) - et al.
Exponentially fitted explicit Runge–Kutta methods
J. Comput. Appl. Math.
(2000) - et al.
Two-step extended RKN methods for oscillatory systems
Comput. Phys. Commun.
(2011) - et al.
Exponentially fitted two-step hybrid methods for
J. Comput. Appl. Math.
(2011) - et al.
Trigonometrically-fitted scheifele two-step methods for perturbed oscillators
Comput. Phys. Commun.
(2011) - et al.
Multi-step Runge–Kutta–Nyström methods for special second-order initial value problems
Appl. Numer. Math.
(2017)
A new approach on the construction of trigonometrically fitted two step hybrid methods
J. Comput. Appl. Math.
Dissipative Chebyshev exponential-fitted methods for numerical solution of second-order differential equations
J. Comput. Appl. Math.
Variable-stepsize Chebyshev-type methods for the integration of second-order I.V.P.’s
J. Comput. Appl. Math.
On the frequency choice in trigonometrically fitted methods
Appl. Math. Lett.
On the choice of the frequency in trigonometrically-fitted methods for periodic problems
J. Comput. Appl. Math.
Numerical solution of highly oscillatory ordinary differential equations
Acta Numer.
Diagonally implicit Rungez–Kuttaz–Nyström methods for oscillatory problems
SIAM J. Numer. Anal.
Solving Ordinary Differential Equations I: Nonstiff Problems
Cited by (2)
Multi-step hybrid methods adapted to the numerical integration of oscillatory second-order systems
2019, Journal of Applied Mathematics and Computing
- ☆
The research was supported in part by the National Natural Science Foundation of China under Grant No: 11401164, by Hebei Natural Science Foundation of China under Grant No: A2014205136, by the Natural Science Foundation of China under Grant No: 11201113 and and by the Specialized Research Foundation for the Doctoral Program of Higher Education under Grant No: 20121303120001