Global Journal of Pure and Applied Mathematics.

ISSN 0973-1768 Volume 12, Number 1 (2016), pp. 67–80

© Research India Publications
http://www.ripublication.com/gjpam.htm

Trigonometrically-fitted explicit four-stage

fourth-order Runge–Kutta–Nyström method
for the solution of initial value problems with
oscillatory behavior

M. A. Demba1
Department of Mathematics,
Universiti Putra Malaysia,
43400 UPM Serdang, Selangor, Malaysia.

N. Senu and F. Ismail

Department of Mathematics and
Institute for Mathematical Research,
Universiti Putra Malaysia,
43400 UPM Serdang, Selangor, Malaysia.

Abstract

An explicit trigonometrically-fitted Runge–Kutta–Nyström (ETFRKN) method is

constructed in this paper based on Simos technique, which exactly integrates initial-
value problems whose solutions are linear combinations of functions of the form
eiwx and e−iwx or equivalently sin(wx) and cos(wx) with w > 0 the principal fre-
quency of the problem. The numerical results show the efficacy of the new method
in comparison with other existing methods.

AMS subject classification: 65L05, 65L06.

Keywords: Trigonometric-fitting, Runge-Kutta-Nystrom (RKN) Methods, Oscil-
latory Initial Value Problems.

1
Corresponding author.
68 M. A. Demba, N. Senu and F. Ismail

1. Introduction
In the last decades, methods for the numerical solution of the initial value problems

y

= f (x, y), y(x0 ) = y0 , y
(x0 ) = y
0 (1.1)

whose solution shows a pronounced oscillatory behavior has attracted the interest of
many researchers. Such problems occur in several fields of applied sciences such as:
molecular dynamics, celestial mechanics, theoretical physics, physical chemistry and
electronics. Bettis in [1] proposed the first Runge–Kutta (RK) methods with 3 and 4
stages for the solution of Ordinary Differential Equations (ODEs) with periodic solu-
tions. More recently, monovasilis in [2], Franco in [3], Ambrosio in [4], Ramos in
[5] proposed Runge–Kutta–Nyström (RKN) methods for the solutions of second order
ordinary differential equations. Similarly, Simos in [6], Coleman in [7] constructed an
explicit RK method which solves certain first order initial value problems with oscillatory
solutions. In the same way, Berghe in [8] proposed some exponentially-fitted RK (EFRK)
methods that integrates exactly first order systems whose solutions can be represented
as the linear combination of some functions. Motivated by the work of Simos in [9], we
construct a new accurate fourth order four stage explicit trigonometrically-fitted RKN



method. The constructed method can exactly integrate the test equation y = −w2 y
and the numerical results show the efficiency of the new method. The remaining part
of this paper is designed as follows: Section 2 deals with the derivation of the proposed
method. In section 3 we analyze the local truncation error of the new method. In section
4 we present the numerical results and the last section deals with the conclusion.

2. Derivation of the proposed method

In this section, we will derive four-stage fourth order explicit trigonometrically-fitted
RKN method using Simos technique.
Consider the four-stage explicit RKN method given by:

4
yn+1 = yn + hyn
+ h2 bi f (xn + ci h, Yi ), (2.1)
i=1


4


yn+1 = yn
+ h di f (xn + cih, Yi ), (2.2)
i=1


3
Yi = yn + ci hyn
+h 2
aij f (xn + cih, Yj ). (2.3)
j =1

or in Butcher Tableau:
In this study, the four-stage fourth order dispersive of order eight RKN method will
be considered as given in [10]. The coefficients of the method are given in Table 1 below:
Four-stage fourth-order Runge–Kutta–Nyström method 69

c2 a21

c3 a31 a32

c4 a41 a42 a43

b1 b2 b3 b4

d1 d2 d3 d4

Table 1: The RKN4(4,8,5)M Method [10]

0

1 1
4 32
7 19 16
10 600 75
32 58 3
1
315 315 14

1 28 50 1
21 81 567 54
1 32 250 5
14 81 567 54

Applying an explicit Runge-Kutta-Nyström method (2.1)–(2.3) to y = −w2 y, we
obtained:

4
yn+1 = yn + hy
n + h2 bi (−w2 Yi ), (2.4)
i=1

and

4
yn+1 = yn
+h di (−w2 Yi ). (2.5)
i=1
70 M. A. Demba, N. Senu and F. Ismail

where

Y1 = yn + c1 hyn , (2.6)


Y2 = yn + c2 hyn − h a21 w Y1 ,
2 2
(2.7)


Y3 = yn + c3 hyn + h (−a31 w Y1 − a32 w Y2 ),
2 2 2
(2.8)


Y4 = yn + c4 hyn + h (−a41 w Y1 − a42 w Y2 − a43 w Y3 ).
2 2 2 2
(2.9)

Now, let yn = eI wx , computing the value of yn+1 , yn
and yn+1

and substituting in
the equations (2.4)–(2.9) and by using e = cos(v) + I sin(v) and comparing the real
Iv

and imaginary part, we get the following system of equations:

4

3
cos(v) = 1 − v 2
bi (1 − v 2
aij Yj e−I wx ), (2.10)
i=1 j =1

4
sin(v) = v − v 2 bi ci v, (2.11)
i=1


4

3
sin(v) = v di (1 − v 2
aij Yj e−I wx ), (2.12)
i=1 j =1

4
cos(v) = 1 − v 2
di ci . (2.13)
i=1

where v = wh.
Solving (2.10)–(2.13) using the coefficients of the method in Table 1 for a31 , b2 , c2 , c3 ,
we obtain the solution as given in (2.14):

272160 sin(v) − 2097v 5 + 36v 7 + 41560v 3 − 272160v

a31 = ,
600v 3 (9v 2 − 200)

16(2v 4 − 37v 2 + 162v sin(v) − 810 + 810 cos(v))

b2 = ,
405v 2 (v 2 − 32)

3(cos(v) + 4)(15v 3 + v 3 sin(v)2 + 5 cos(v) sin(v)v 2 − 480v − 32v sin(v)2 )

c2 =
16(sin(v)2 + 15)v(v 2 + 6v sin(v) + 30 cos(v) − 30)

3(160 cos(v) sin(v) − 20 sin(v)v 2 + 640 sin(v))

− ,
16(sin(v)2 + 15)v(v 2 + 6v sin(v) + 30 cos(v) − 30)
Four-stage fourth-order Runge–Kutta–Nyström method 71

3(90720v sin(v)3 + 21280v 2 sin(v)2 − 699v 6 sin(v)2 + 453600 sin(v)2 cos(v))

c3 =
100(sin(v)2 + 15)(v 2 + 6v sin(v) + 30 cos(v) − 30)v 2 (9v 2 − 200)

3(12v 8 sin(v)2 − 2721600 sin(v)2 + 11200v 4 sin(v)2 + 13980v 5 sin(v) − 190400 sin(v)v 3 )
+
100(sin(v)2 + 15)(v 2 + 6v sin(v) + 30 cos(v) − 30)v 2 (9v 2 − 200)

3(263200v sin(v) − 3495 cos(v)v 5 sin(v) + 47600 cos(v) sin(v)v 3 − 240v 7 sin(v))
+
100(sin(v)2 + 15)(v 2 + 6v sin(v) + 30 cos(v) − 30)v 2 (9v 2 − 200)

3(274400 cos(v) sin(v)v + 60 cos(v)v 7 sin(v) − 1400v 4 cos(v) + 173600v 4 + 4536000)

+
100(sin(v)2 + 15)(v 2 + 6v sin(v) + 30 cos(v) − 30)v 2 (9v 2 − 200)

3(10485v 6 + 180v 8 − 823200v 2 − 4536000 cos(v) + 285600v 2 cos(v))(cos(v) + 4)

− . (2.14)
100(sin(v)2 + 15)(v 2 + 6v sin(v) + 30 cos(v) − 30)v 2 (9v 2 − 200)

Which lead to our new method ETFRKN4(4,8,5)M.

The corresponding Taylor series expansion of the solution is given in (2.15).
19 3 1 349
a31 = + v4 + v6 + v8
600 20000 2000000 4400000000

109999 1618447
+ v 10 + v 12
34320000000000 11088000000000000
292360553
+ v 14 + · · · ,
44553600000000000000

28 1 4 1 19 449
b2 = − v + v6 − v8 + v 10
81 3600 161280 232243200 1226244096000

32869 312811
− v 12 − v 14 + · · · ,
3570822807552000 1713994947624960000

1 1 19 1019 481093
c2 = − v4 + v6 − v8 + v 10
4 26880 7741440 13624934400 119027426918400

8198117 647998097
− v 12 + v 14 + · · · ,
57133164920832000 93241325150797824000

7 47 4 1759 335651 2163024121

c3 = + v − v6 + v8 − v 10
10 60000 54000000 1663200000000 51891840000000000

1913285557
+ v 12 + · · · (2.15)
124540416000000000000
72 M. A. Demba, N. Senu and F. Ismail

As v → 0, the coefficients a31 , b2 , c2 and c3 of the new method (ETFRKN4(4,8,5)M)

reduces to the coefficients of the original method (RKN4(4,8,5)M). That is to say
a31 (0), b2 (0), c2 (0) and c3 (0) are identical to a31 , b2 , c2 and c3 of RKN4(4,8,5)M method.

3. Algebraic Order and Error Analysis

In this section, we are going to compute the local truncation error(LTE) of the new method
and verify its algebraic order. We start by computing the Taylor series expansions of the


theoretical solution y(xn + h), of its derivative y (xn + h), the approximate solution yn+1


and of its derivative yn+1 . Then, we compute the LTE of y and that of its derivative as
given below:

LT E = yn+1 − y(xn + h),

LT Eder = yn+1 − y (xn + h). (3.1)

The LT E and LT Eder of the method ETFRKN4(4,8,5)M are given below:

h5







LT E = − (3y y fyy + 3y fyxx + (y )3 fyyy + 3y fxy
1440


+ 3(y )2 fxyy + fxxx ) + O(h6 ),
h5







LT Eder = (6(y )2 y fyyy + 5(y )2 fyy fy + 12y y fxyy
120




+ 6y fxy fy + 4y fyy fx + 4(y )3 fxyyy






+ 6y fyxx + (fy )2 y + 4y fxxxy + fy fxx + 4fxy fx



+ 6(y )2 fxxyy + (y )4 fyyyy



+ 3(y )2 fyy + fxxxx ) + O(h6 ). (3.2)

From equation (3.2), we can see that the order of ETFRKN4(4,8,5)M is 4 because all of
the coefficients up to h4 vanished.

4. Problems Tested and Numerical Results

In this section, we will apply the new method to some second-order ordinary differential
equation problems. The following RKN and RK method are used for the numerical
comparisons.

• ETFRKN4(4,8,5)M: The new method derived in this paper,

• RKN4G: The fourth-order RKN method obtained by Garcia et al in [16],

• RKN4-4: An explicit RKN method given by Xinyuan Wu et al in [12],

Four-stage fourth-order Runge–Kutta–Nyström method 73

Table 2: Numerical results for problem 1.

h Methods T = 100 T = 1000 T = 10000
0.025 RKN4(4,8,5)M 6.972504(−6) 6.991171(−5) 6.999528(−4)
ETFRKN4(4,8,5)M 4.589344(−11) 5.240339(−9) 6.214849(−7)
RKN4G 2.060889(−3) 2.060542(−3) 2.041889(−1)
RKN4(4,8,5)S 1.268534(−6) 8.952827(−6) 8.771342(−5)
RKN4-4 4.126979(−3) 4.117858(−2) 4.011603(−1)
RK4 8.723902(−2) 8.701857(−1) 7.692638(+0)
0.05 RKN4(4,8,5)M 2.244837(−4) 2.255854(−3) 2.281038(−2)
ETFRKN4(4,8,5)M 2.903877(−11) 2.974295(−9) 1.733919(−7)
RKN4G 3.281656(−2) 3.202012(−1) 1.631807(+0)
RKN 4(4,8,5)S 4.113511(−5) 3.805704(−4) 3.781592(−3)
RKN4-4 6.576666(−2) 6.081756(−1) 1.556586(+0)
RK4 1.349974(+0) 9.652030(+0) 1.125925(+0)
0.075 RKN4(4,8,5)M 1.722928(−3) 1.746987(−2) 1.888961(−1)
ETFRKN4(4,8,5)M 1.719080(−11) 1.955921(−9) 1.158436(−7)
RKN4G 1.628263(−1) 1.232240(+0) 1.491109(+0)
RKN 4(4,8,5)S 4.189894(−4) 4.147285(−3) 4.076388(−2)
RKN4-4 3.194580(−1) 1.405130(+0) 1.405130(+0)
RK4 5.671354(+0) 1.001198(+1) 1.001198(+1)
0.1 RKN4(4,8,5)M 7.486879(−3) 7.699014(−2) 1.088929(+0)
ETFRKN 4(4,8,5)M 5.150516(−10) 4.419555(−9) 1.313724(−7)
RKN4G 4.835378(−1) 1.381059(+0) 1.381059(+0)
RKN4(4,8,5)S 2.468325(−3) 2.503959(−2) 2.264669(−1)
RKN4-4 8.242491(+0) 9.207004(+0) 9.207001(+0)
RK4 9.159293(−1) 1.292106(+0) 1.292106(+0)

• RKN4(4,8,5)M: The RKN method with dispersion order eight and dissipation
order five obtained by Senu in [10],

• RKN4(4,8,5)S: The RKN method with dispersion order eight and dissipation order
five obtained by Senu in [10] and

• RK4: The classical RK method given by Butcher in [11].

Problem 1 Ahmad et al in [13]

y = −64y, y(0) = 1, y (0) = −2

The exact solution is

1
y(x) = − sin(8x) + cos(8x).
4
74 M. A. Demba, N. Senu and F. Ismail

Table 3: Numerical results for problem 2

h Methods T = 100 T = 1000 T = 10000
0.025 RKN4(4,8,5)M 3.639342(-5) 3.660083(-4) 3.671962(-3)
ETFRKN4(4,8,5)M 2.922861(-9) 1.118849(-8) 1.068178(-6)
RKN4G 8.600175(-3) 8.609385(-2) 8.230410(-1)
RKN4(4,8,5)S 1.201895(-5) 5.484566(-5) 4.939859(-4)
RKN4-4 1.719217(-2) 1.712476(-1) 1.495620(+0)
RK4 4.582085(-1) 4.424328(+0) 2.167420(+1)
0.05 RKN4(4,8,5)M 1.172498(-3) 1.191337(-2) 1.237884(-1)
ETFRKN4(4,8,5)M 4.41177(-8) 4.814676(-8) 2.687882(-7)
RKN4G 1.373487(-1) 2.136231(+0) 2.136231(+0)
RKN4(5,8,5)S 3.525533(-4) 2.479343(-3) 2.366650(-2)
RKN4-4 2.694122(-1) 1.846476(+0) 2.023791(+0)
RK4 6.561813(+0) 1.812164(+1) 1.812164(+1)
0.075 RKN4(4,8,5)M 9.143561(-3) 9.486361(-2) 1.293305(+0)
ETFRKN4(4,8,5)M 2.451447(-7) 2.472938(-7) 4.375056(-7)
RKN4G 6.408312(-1) 1.929449(+0) 1.929449(+0)
RKN 4(4,8,5)S 3.498339(-3) 2.909897(-2) 2.619785(-1)
RKN4-4 1.123727(+0) 1.806809(+0) 1.806809(+0)
RK4 1.594005(+1) 1.607897(+1) 1.607897(+1)
0.1 RKN4(4,8,5)M 3.925749(-2) 4.576288(-1) 2.206548(+1)
ETFRKN 4(4,8,5)M 8.652444(-7) 1.081217(-6) 3.116065(-6)
RKN4G 1.496559(+0) 1.773207(+0) 1.773207(+0)
RKN4(4,8,5)S 2.114117(-2) 1.808480(-1) 1.0775279(+0)
RKN4-4 1.654061(+0) 1.654061(+0) 1.654061(+0)
RK4 1.492280(+1) 1.492280(+1) 1.492280(+1)

Problem 2 Anastassi and Kosti in [14]

y = −100y + 99 sin(x), y(0) = 1, y (0) = 11

The exact solution is

y(x) = sin(10x) + cos(10x) + sin(x).

Problem 3 Moo et al in [17]

y1 = −y1 + 0.001 cos(x), y1 (0) = 1, y1 = 0,




y2 = −y2 + 0.001 sin(x), y2 (0) = 0, y2 = 0.9995
Four-stage fourth-order Runge–Kutta–Nyström method 75

The exact solution is

y1 (x) = cos(x) + 0.0005 x cos(x)

y2 (x) = sin(x) − 0.0005 x sin(x).

Table 4: Numerical results for problem 3

h Methods T = 100 T = 1000 T = 10000
0.025 RKN4(4,8,5)M 3.417711(-11) 9.723635(-10) 1.067510(-7)
ETFRKN4(4,8,5)M 7.858603(-12) 8.986945(-10) 1.062072(-7)
RKN4G 6.034852(-8) 3.464664(-6) 1.613212(-5)
RKN4(4,8,5)S 7.84666(-10) 1.575359(-9) 1.069329(-7)
RKN4-4 1.206630(-7) 1.254851(-6) 3.255984(-5)
RK4 4.520694(-7) 4.601692(-6) 4.591613(-5)
0.05 RKN4(4,8,5)M 1.081503(-9) 1.163141(-8) 1.171338(-7)
ETFRKN4(4,8,5)M 5.258016(-12) 5.02382(-10) 2.977504(-8)
RKN4G 9.654375(-7) 5.517615(-5) 2.627705(-4)
RKN4(4,8,5)S 1.261104(-8) 1.381034(-8) 2.867273(-8)
RKN4-4 1.930491(-6) 2.009134(-5) 5.256414(-4)
RK4 7.232536(-6) 7.364302(-5) 7.361172(-4)
0.075 RKN4(4,8,5)M 6.317846(-9) 6.792010(-8) 1.781351(-6)
ETFRKN4(4,8,5)M 1.089411(-9) 1.120594(-8) 1.211640(-7)
RKN4G 4.886862(-6) 2.794100(-4) 1.330306(-3)
RKN 4(4,8,5)S 6.369031(-8) 7.033042(-8) 2.461517(-7)
RKN4-4 9.771535(-6) 1.017168(-4) 2.661288(-3)
RK4 2.608892(-5) 2.711084(-4) 7.097847(-3)
0.1 RKN4(4,8,5)M 3.460976(-8) 3.721067(-7) 3.739522(-6)
ETFRKN4(4,8,5)M 1.875833(-12) 2.308980(-10) 2.656270(-8)
RKN4G 1.544134(-5) 8.831508(-4) 4.205253(-3)
RKN4(4,8,5)S 2.043120(-7) 2.333710(-7) 5.823003(-7)
RKN4-4 3.087523(-5) 3.214332(-4) 8.413109(-3)
RK4 1.157343(-4) 1.1778225(-3) 1.177426(-2)

Problem 4 Senu et al in [15]

y = −y + x, y(0) = 1, y (0) = 2

The exact solution is

y(x) = cos(x) + sin(x) + x.
The accuracy strategy used is finding log10 of the maximum global error,

Accuracy = log10 max y(xn ) − yn  , (4.1)

76 M. A. Demba, N. Senu and F. Ismail

where xn = x0 + nh, n = 1, 2, 3, . . . (T − x0 )/ h. In this paper, we denote T as the

interval used for the integration.

The numerical results are shown in Tables 2-5.

Table 5: Numerical results for problem 4

h Methods T = 100 T = 1000 T = 10000
0.025 RKN4(4,8,5)M 3.201817(-11) 6.383933(-10) 2.989178(-7)
ETFRKN4(4,8,5)M 1.880414(-11) 5.408549(-10) 2.979411(-7)
RKN4G 8.653461(-8) 3.466190(-6) 8.524657(-6)
RKN4(4,8,5)S 7.805048(-10) 8.739462(-10) 2.91846(-7)
RKN4-4 1.708399(-7) 1.721004(-6) 1.715370(-5)
RK4 3.219511(-7) 3.347823(-6) 8.732349(-5)
0.05 RKN4(4,8,5)M 8.483979(-10) 9.229263(-9) 2.664778(-7)
ETFRKN4(4,8,5)M 2.112705(-10) 1.812355(-9) 6.845243(-8)
RKN4G 1.384249(-6) 5.523357(-5) 1.380617(-4)
RKN4(4,8,5)S 1.253343(-8) 1.324117(-8) 7.672772(-8)
RKN4-4 2.734130(-6) 2.754721(-5) 2.761332(-4)
RK4 5.151875(-6) 5.357128(-5) 1.401884(-3)
0.075 RKN4(4,8,5)M 8.212211(-9) 8.827681(-8) 8.875049(-7)
ETFRKN4(4,8,5)M 3.424816(-12) 3.370815(-10) 1.977423(-8)
RKN4G 7.006304(-6) 2.795714(-4) 6.991754(-4)
RKN 4(4,8,5)S 6.421809(-8) 7.074550(-8) 1.506032(-7)
RKN4-4 1.384465(-5) 1.394256(-4) 1.398006(-3)
RK4 3.663125(-5) 3.727062(-4) 3.725955(-3)
0.1 RKN4(4,8,5)M 2.641282(-8) 2.819148(-7) 7.435934(-6)
ETFRKN4(4,8,5)M 3.433840(-9) 3.478845(-8) 3.733085(-7)
RKN4G 2.215882(-5) 8.838602(-4) 2.209373(-3)
RKN4(4,8,5)S 2.022193(-7) 2.302874(-7) 9.318692(-7)
RKN4-4 4.377299(-5) 4.409937(-4) 4.417562(-3)
RK4 8.246080(-5) 8.565301(-4) 2.243462(-2)
Four-stage fourth-order Runge–Kutta–Nyström method 77

−1
log10 (MAXERR)

−2

−3

−4

−5 ETFRKN4(4,8,5)M
RKN4-4
RKN4(4,8,5)S
−6 RKN4(4,8,5)M
RKN4G
−7 RK4

5.5 5.6 5.7 5.8 5.9 6 6.1 6.2

log10 (Function Evaluations)

Figure 1: The efficiency curve for Problem 1 with tend = 10000 and h = i(0.025),
i = 1, 2, 3, 4.

−1
log10 (MAXERR)

−2

−3

−4
ETFRKN4(4,8,5)M
−5 RKN4-4
RKN4(4,8,5)S
−6 RKN4(4,8,5)M
RKN4G
RK4
−7
5.5 5.6 5.7 5.8 5.9 6 6.1 6.2
log10 (Function Evaluations)

Figure 2: The efficiency curve for Problem 2 with tend = 10000 and h = i(0.025),
i = 1, 2, 3, 4.

We also display the accuracy of these methods graphically in figures 1 to 4.

5. Conclusion
In this study, we have presented the fourth order four-stage explicit trigonometrically-
fitted RKN method for the solutions of oscillatory problems. The numerical results
obtained show clearly that the global error of the new method is smaller than that of the
other existing methods; The new method is much more efficient than the other existing
methods.
78 M. A. Demba, N. Senu and F. Ismail

−1
ETFRKN4(4,8,5)M
RKN4-4
−2 RKN4(4,8,5)S
RKN4(4,8,5)M
RKN4G
−3 RK4
log10 (MAXERR)

−4

−5

−6

−7

−8
5.5 5.6 5.7 5.8 5.9 6 6.1 6.2
log10 (Function Evaluations)

Figure 3: The efficiency curve for Problem 3 with tend = 10000 and h = i(0.025),
i = 1, 2, 3, 4.

−1
ETFRKN4(4,8,5)M
RKN4-4
−2 RKN4(4,8,5)S
RKN4(4,8,5)M
RKN4G
−3 RK4
log10 (MAXERR)

−4

−5

−6

−7

−8
5.5 5.6 5.7 5.8 5.9 6 6.1 6.2
log10 (Function Evaluations)

Figure 4: The efficiency curve for Problem 4 with tend = 10000 and h = i(0.025),
i = 1, 2, 3, 4.

Acknowledgment

We are thankful to the reviewers of this manuscript for his/her fruitful comments and
remarks, Institute of Mathematical Research (INSPEM) and the Department of Math-
ematics, Universiti Putra Malaysia for the support and assistance during the research
work.
Four-stage fourth-order Runge–Kutta–Nyström method 79

References

[1] D. Bettis, “Runge–Kutta algorithms for oscillatory problems”, Zeitschrift für ange-
wandte Mathematik und Physik ZAMP, vol. 30, pp. 699–704, 1979.
[2] T. Monovasilis, Z. Kalogiratou and T.E. Simos, “Exponentially-fitted symplectic
Runge-Kutta Nyström methods”, Appl. Math. Inf. Sci, vol. 7, pp. 81–85, 2013.
[3] J. Franco, I. Gomez, “Symplectic Explicit methods of Runge-Kutta Nyström" type
for solving pertubed oscillators, Journal of Computational and Applied Mathemat-
ics, vol. 260, pp. 482–493, 2014.
[4] D’Ambrosio, B. Paternoster, and G. Santomauro, “Revised exponentially-fitted
Runge-Kutta Nyström methods”, Applied Mathematics Letters, vol. 30, pp. 56–60,
2014.
[5] H. Ramos, and J. Vigo-Aguiar, “A trigonometrically-fitted method with two fre-
quencies, one for the solution and another one for the derivative”, Computer Physics
Communications, vol 185, pp. 1230–1236, 2014.
[6] T. Simos, “An exponentially-fitted Runge-Kutta method for the numerical inte-
gration of initial-value problems with periodic or oscillating solutions, Computer
Physics Communications, vol. 115, pp. 1–8, 1998.
[7] J. P. Coleman and S. C. Duxbury, “Mixed collocation methods for y

=
f (x, y)”, Journal of Computational and Applied Mathematics, vol. 126, pp. 47–75,
2000.
[8] G. V. Berghe, H. De Meyer, M. Vn Daele, and T. Van Hecke, “Exponentially-fitted
explicit Runge-Kutta methods”, Computer Physics Communications, vol. 123, pp.
7–15, 1999.
[9] T. E. Simos, “Exponentially-fitted Runge-Kutta Nyström method for the numerical
integration of initial-value problems with oscillating solutions, Applied Mathemat-
ics Letters, vol. 15, pp. 217–225, 2002.
[10] N. Senu,“Runge-Kutta Nyström methods for solving oscillatory problems”, Ph.D
diss., Department of mathematics, faculty of science, 43400 Serdang, Malaysia,
2010.
[11] J. C. Butcher, “Numerical methods for ordinary differential equations”, Wiley &
Sons LTD., England, 2008.
[12] Xinyuan Wu, Xiong You and Bin Wang, “Structure preserving algorithms for os-
cillatory differential equations”, Science Press Beijing & Springer-Verlag Berlin
Heidelberg, Beijing, 2013.
[13] S. Ahmed, F. Ismail, N. Senu, and M. Suleiman, “Zero-dissipative phase-
fitted hybrid methods for solving oscillatory second order ordinary differential
equations”, Applied Mathematics and Computation, vol. 19, pp. 10096–10104,
2013.
[14] Z. Anastassi and A. Kosti, “A 6(4) optimized embedded Runge-Kutta Nyström
80 M. A. Demba, N. Senu and F. Ismail

pair for the numerical solution of periodic problems”, Journal of Computational

Applied Mathematics, vol. 275, pp. 311–320, 2013.
[15] N. Senu, M. Suleiman, F. Ismail and M. Othman, “A singly diago-
nally implicit Runge-Kutta-Nyström method for solving oscillatory problems”,
IAENG International Journal of Applied Mathematics, vol. 41, pp. 155–161,
2011.
[16] A. Garcia, P. Marti, andA. B. Gonzalez et al, “New methods for oscillatory problems
based on classical codes”, Applied Numerical Mathematics, vol. 42, pp. 141–157,
2002.
[17] K. Moo, N. Senu, F. Ismail, and M. Suleiman, “New phase-fitted and amplification-
fitted fourth-order and fifth-order Runge-Kutta-Nyström methods for oscillatory
problems”, Abstract and Applied Analysis, vol. 2013, 2013.
[18] Fang, Yonglei, You, Xiong, Ming and Qinghe, “Trigonometrically fitted two-
derivative Runge-Kutta methods for solving oscillatory differential equations”, Nu-
merical Algorithms, vol. 65, pp. 651–667, 2014.