Hindawi

Journal of Applied Mathematics

Volume 2018, Article ID 4029371, 19 pages
https://doi.org/10.1155/2018/4029371

Research Article
Exponentially Fitted and Trigonometrically Fitted Explicit
Modified Runge-Kutta Type Methods for Solving
𝑦󸀠󸀠󸀠(𝑥) = 𝑓(𝑥, 𝑦, 𝑦󸀠)

N. Ghawadri,1 N. Senu ,1,2 F. Ismail ,1,2 and Z. B. Ibrahim 1,2

1
Institute for Mathematical Research, Universiti Putra Malaysia (UPM), 43400 Serdang, Selangor, Malaysia
2
Department of Mathematics, Universiti Putra Malaysia (UPM), 43400 Serdang, Selangor, Malaysia

Correspondence should be addressed to N. Senu; [email protected]

Received 21 December 2017; Revised 1 May 2018; Accepted 28 May 2018; Published 4 July 2018

Academic Editor: Igor Andrianov

Copyright © 2018 N. Ghawadri et al. This is an open access article distributed under the Creative Commons Attribution License,
which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

Exponentially fitted and trigonometrically fitted explicit modified Runge-Kutta type (MRKT) methods for solving 𝑦󸀠󸀠󸀠 (𝑥) =
𝑓(𝑥, 𝑦, 𝑦󸀠 ) are derived in this paper. These methods are constructed which exactly integrate initial value problems whose solutions
are linear combinations of the set functions 𝑒𝜔𝑥 and 𝑒−𝜔𝑥 for exponentially fitted and sin(𝜔𝑥) and cos(𝜔𝑥) for trigonometrically
fitted with 𝜔 ∈ 𝑅 being the principal frequency of the problem and the frequency will be used to raise the accuracy of the methods.
The new four-stage fifth-order exponentially fitted and trigonometrically fitted explicit MRKT methods are called EFMRKT5 and
TFMRKT5, respectively, for solving initial value problems whose solutions involve exponential or trigonometric functions. The
numerical results indicate that the new exponentially fitted and trigonometrically fitted explicit modified Runge-Kutta type methods
are more efficient than existing methods in the literature.

1. Introduction Paternoster [1] developed Runge-Kutta-Nyström methods for

ODEs with periodic solutions based on trigonometric poly-
This work deals with exponentially fitted and trigonometri- nomials. Vanden Berghe et al. [2] developed exponentially fit-
cally fitted modified Runge-Kutta type methods for solving ted Runge-Kutta methods. Simos [3] extended exponentially
third-order ordinary differential equations (ODEs)
fitted Runge-Kutta methods for the numerical solution of the
𝑦󸀠󸀠󸀠 (𝑥) = 𝑓 (𝑥, 𝑦 (𝑥) , 𝑦󸀠 (𝑥)) , Schrodinger equation and related problems. Kalogiratou et
al. [[4, 5]] constructed trigonometrically and exponentially
𝑦 (𝑥0 ) = 𝑦0 , fitted Runge-Kutta-Nyström methods for the numerical solu-
tion of the Schrodinger equation and related problems which
𝑦󸀠 (𝑥0 ) = 𝑦0󸀠 , (1) is eighth algebraic order. Next Simos et al. [6] constructed
exponentially fitted Runge-Kutta-Nyström method for the
𝑦󸀠󸀠 (𝑥0 ) = 𝑦0󸀠󸀠 , numerical solution of initial value problems with oscillating
𝑥 ≥ 𝑥0 . solutions. Sakas et al. [7] developed a fifth algebraic order
trigonometrically fitted modified Runge-Kutta Zonneveld
This sort of problems is often found in numerous physical method for the numerical solution of orbital problems. Van
problems like thin film flow, gravity-driven flows, electro- de Vyver [8] in 2005 constructed Runge-Kutta-Nyström
magnetic waves, and so on. In the past and recent years many pair for the numerical integration of perturbed oscillators.
researchers constructed exponentially fitted and trigono- Then Yang et al. [9] constructed trigonometrically fitted
metrically fitted explicit Runge-Kutta methods for solving adapted Runge-Kutta-Nyström methods for perturbed oscil-
first-order and second-order ordinary differential equations. lators. Recently, Demba et al. [10] constructed an explicit
2 Journal of Applied Mathematics

trigonometrically fitted Runge-Kutta-Nyström method using Table 1: The Butcher tableau MRKT method.
Simos technique. 𝑐 ̂
𝛾 𝛾̂ 𝐴 𝐴
In this paper we construct explicit exponentially fit-
ted and trigonometrically fitted modified Runge-Kutta type 𝑏 𝑏󸀠 𝑏󸀠󸀠
methods with four-stage fifth-order, called EFMRKT5 and
TFMRKT5, respectively. Section 2 discussed the oscillatory
methods. In order to construct the exponentially fitted and
and nonoscillatory properties of the third-order linear dif-
trigonometrically fitted MRKT methods, the extra 𝛾𝑖 and 𝛾̂𝑖
ferential equation. In Section 3, the necessary conditions and
are absolutely necessary to insert at each stage and the MRKT
the derivation for exponentially fitted and trigonometrically
methods is given as follows:
fitted modified Runge-Kutta type methods for solving third-
order ODEs are given. The error analysis of the new EFM- 𝑠
ℎ2 󸀠󸀠
RKT5 and TFMRKT5 methods was discussed in Section 4, 𝑦𝑛+1 = 𝑦𝑛 + ℎ𝑦𝑛󸀠 + 𝑦𝑛 + ℎ3 ∑𝑏𝑖 𝑘𝑖 , (3)
respectively. The effectiveness of the new methods when 2 𝑖=1
compared with existing methods is given in Section 5. The 𝑠
󸀠
thin film flow problem is discussed in Section 6. 𝑦𝑛+1 = 𝑦𝑛󸀠 + ℎ𝑦𝑛󸀠󸀠 + ℎ2 ∑𝑏𝑖󸀠 𝑘𝑖 , (4)
𝑖=1

2. Third-Order Linear Differential 𝑠

󸀠󸀠
Equation with Oscillating 𝑦𝑛+1 = 𝑦𝑛󸀠󸀠 + ℎ∑𝑏𝑖󸀠󸀠 𝑘𝑖 , (5)
𝑖=1
and Nonoscillating Solutions
where
This section discusses the oscillatory and nonoscillatory
properties of the third-order linear differential equation 𝑘1 = 𝑓 (𝑥𝑛 , 𝑦𝑛 , 𝑦𝑛󸀠 ) , (6)
󸀠󸀠󸀠 󸀠
𝑦 (𝑥) + 𝑝 (𝑥) 𝑦 + 𝑞 (𝑥) 𝑦 = 0. (2)
ℎ2 2 󸀠󸀠
A solution of (2) will be said to be oscillatory if it changes 𝑘𝑖 = 𝑓 (𝑥𝑛 + 𝑐𝑖 ℎ, 𝛾𝑖 𝑦𝑛 + ℎ𝑐𝑖 𝑦𝑛󸀠 + 𝑐𝑦
2 𝑖 𝑛
signs for arbitrarily large values of 𝑥. The other solutions will
(7)
be said to be nonoscillatory. 𝑠 𝑠
If 𝑝(𝑥) < 0 and 𝑞(𝑥) < 0 are constants, then it is easy to + ℎ3 ∑ 𝑎𝑖𝑗 𝑘𝑗 , 𝑦𝑛󸀠 + 𝛾̂𝑖 ℎ𝑐𝑖 𝑦𝑛󸀠󸀠 + ℎ2 ∑𝑎̂𝑖𝑗 𝑘𝑗 )
show that if (2) has an oscillatory solution, then there are two 𝑗=1 𝑗=1
linearly independent oscillatory solutions of (2) whose zeroes
separate and such that any oscillatory solution of (2) is a linear for 𝑖 = 2, 3, . . . , 𝑠.
combination of them. Assuming that 𝑝(𝑥), 𝑝󸀠 (𝑥), and 𝑞(𝑥) are The parameters of the MRKT methods are 𝑐𝑖 , 𝑎𝑖𝑗 , 𝑎̂𝑖𝑗 , 𝑏𝑖 ,
continuous on [0, +∞) the following will be established (see 𝑏𝑖󸀠 , 𝑏𝑖󸀠󸀠 ,𝛾𝑖 and 𝛾̂𝑖 for 𝑖 = 1, 2, . . . , 𝑠 and 𝑗 = 1, 2, . . . , 𝑠 are
[11–14]). assumed to be real. If 𝑎𝑖𝑗 = 0 and 𝑎̂𝑖𝑗 = 0 for 𝑖 ⩽ 𝑗, it is an
explicit method and otherwise implicit method.
Definition 1. A solution of (2) will be called oscillatory iff it The MRKT method can be expressed in Butcher notation
has an infinity of zeroes in(0, +∞) and nonoscillatory iff it using the table of coefficients as follows (see Table 1).
has but a finite number of zeroes in this interval. Equation
(2) is said to be oscillatory iff it has at least one (nontrivial)
3.1. Exponentially Fitted MRKT Method. To construct the
oscillatory solution and nonoscillatory iff all of its (nontrivial)
exponentially fitted Runge-Kutta type four-stage fifth-order
solutions are nonoscillatory.
method the functions 𝑒𝜔𝑥 and 𝑒−𝜔𝑥 need to integrate exactly
Particularly, this paper deals with two cases based on (2) at each stage; therefore the following four equations are
when 𝑞(𝑥) = 0, as follows: obtained:
𝑠
(i) 𝑦󸀠󸀠󸀠 (𝑥) = 𝑝𝑦󸀠 , (𝑝 > 0); it is clear that the characteristic 1
𝑒±𝑐𝑖 V = 𝛾𝑖 ± 𝑐𝑖 V + 𝑐𝑖2 V2 ± V3 ∑ 𝑎𝑖𝑗 𝑒±𝑐𝑗 V , (8)
roots equations are real and one of them is zero; then 2 𝑗=1
solutions will consist of exponential functions.
𝑠
(ii) 𝑦󸀠󸀠󸀠 (𝑥) = −𝑝𝑦󸀠 , (𝑝 > 0); one of the characteristic roots
𝑒±𝑐𝑖 V = 1 ± 𝛾̂𝑖 𝑐𝑖 V ± V2 ∑𝑎̂𝑖𝑗 𝑒±𝑐𝑗 V , (9)
equations is zero and another two are conjugate roots 𝑗=1
and the solutions are in oscillatory form,
where 𝑝 is constant. and six more equations corresponding to 𝑦, 𝑦󸀠 , and 𝑦󸀠󸀠 :
𝑠
1
3. Exponentially Fitted and Trigonometrically 𝑒±V = 1 ± V + V2 ± V3 ∑𝑏𝑖 𝑒±𝑐𝑖 V , (10)
2 𝑖=1
Fitted MRKT Methods
𝑠
In this section, we will determine the conditions and develop 𝑒±V = 1 ± V + V2 ∑𝑏𝑖󸀠 𝑒±𝑐𝑖 V , (11)
exponentially fitted and trigonometrically fitted MRKT 𝑖=1
Journal of Applied Mathematics 3

𝑠
sinh (V𝑐𝑖 ) − V2 ∑𝑖−2 ̂𝑖,𝑗 sinh (V𝑐𝑗 )
𝑗=1 𝑎
𝑒±V = 1 ± V∑𝑏𝑖󸀠󸀠 𝑒±𝑐𝑖 V , (12) 𝛾̂𝑖 = ,
𝑖=1 V𝑐𝑖 (26)
where V = 𝜔ℎ, 𝜔 ∈ R. The relations cosh(V) = (𝑒V + 𝑒−V )/2 and 𝑖 = 2, . . . , 𝑠.
sinh(V) = (𝑒V − 𝑒−V )/2 will be used in the derivation process.
The following order conditions are obtained:
Referring to the following fifth-order four-stage method
𝑖−1 developed by Fawzi et al. [15]:
1
cosh (V𝑐𝑖 ) = 𝛾𝑖 + V2 𝑐𝑖2 + V3 ∑𝑎𝑖𝑗 sinh (V𝑐𝑗 ) , (13)
2 𝑗=1
𝑐1 = 0,
𝑖−1
sinh (V𝑐𝑖 ) = V𝑐𝑖 + V3 ∑𝑎𝑖𝑗 cosh (V𝑐𝑗 ) , 1
(14) 𝑐2 = ,
𝑗=1 5
𝑖−1 2
2 𝑐3 = ,
cosh (V𝑐𝑖 ) = 1 + V ∑ 𝑎̂𝑖𝑗 cosh (V𝑐𝑗 ) , (15) 3
𝑗=1
𝑐4 = 1,
𝑖−1
sinh (V𝑐𝑖 ) = 𝛾̂𝑖 𝑐𝑖 V + V2 ∑𝑎̂𝑖𝑗 sinh (V𝑐𝑗 ) , (16) 𝑎21 = 0,
𝑗=1
49
𝑎31 = − ,
and six equations corresponding to 𝑦, 𝑦 , and 𝑦 : 󸀠 󸀠󸀠 4860

𝑠 7
1 𝑎41 = ,
cosh (V) = 1 + V2 + V3 ∑𝑏𝑖 sinh (V𝑐𝑖 ) , (17) 50
2 𝑖=1
1
𝑠
𝑎42 = − ,
50
sinh (V) = V + V3 ∑𝑏𝑖 cosh (V𝑐𝑖 ) , (18)
𝑖=1 1
𝑎̂31 = − , (27)
𝑠 27
2
cosh (V) = 1 + V ∑𝑏𝑖󸀠 cosh (V𝑐𝑖 ) , (19) 3
𝑖=1 𝑎̂41 = ,
10
𝑠
sinh (V) = V + V2 ∑𝑏𝑖󸀠 sinh (V𝑐𝑖 ) , (20) 2
𝑎̂42 = − ,
𝑖=1 35
𝑠 3
cosh (V) = 1 + V∑𝑏𝑖󸀠󸀠 sinh (V𝑐𝑖 ) , (21) 𝑏3 = ,
112
𝑖=1

𝑠
𝑏4 = 0,
sinh (V) = V∑𝑏𝑖󸀠󸀠 cosh (V𝑐𝑖 ) . (22)
9
𝑖=1 𝑏3󸀠 = ,
56
Solving (13) to (16), we find 𝑎𝑖,𝑖−1 , 𝑎̂𝑖,𝑖−1 ,𝛾𝑖 , and 𝛾̂𝑖 .
𝑏4󸀠 = 0,
𝑖−1
1 7
𝛾𝑖 = cosh (V𝑐𝑖 ) − V2 𝑐𝑖 2 − V3 ∑ 𝑎𝑖,𝑗 sinh (V𝑐𝑗 ) , (23) 𝑏3󸀠󸀠 = ,
2 𝑗=1 56
5
sinh (V𝑐𝑖 ) − V𝑐𝑖 − V3 ∑𝑖−2
𝑗=1 𝑎𝑖,𝑗 cosh (V𝑐𝑗 ) 𝑏4󸀠 = ,
𝑎𝑖,𝑖−1 = , (24) 48
V3 cosh (V𝑐𝑖−1 )

cosh (V𝑐𝑖 ) − 1 − V2 ∑𝑖−2 ̂𝑖,𝑗 cosh (V𝑐𝑗 ) we solve (23) to (26) and let 𝑎̂21 , 𝑎̂32 , 𝑎̂43 , 𝑎32 , 𝑎43 , 𝛾2 , 𝛾3 , 𝛾4 ,
𝑗=1 𝑎
𝑎̂𝑖,𝑖−1 = , (25) 𝛾̂2 , 𝛾̂3 , and 𝛾̂4 be free parameters and yields.
V2 cosh (V𝑐𝑖−1 )

cosh (V/5) − 1
𝑎̂21 = ,
V2
cosh (2V/3) − 1 + (1/27) V2
𝑎̂32 = ,
V2 cosh (V/5)
4 Journal of Applied Mathematics

cosh (V) − 1 − V2 (3/10 − (2/35) cosh (V/5))

𝑎̂43 = ,
V2 cosh (2V/3)
sinh (2V/3) − 2V/3 + (49/4860) V3
𝑎32 = ,
V3 cosh (𝑐2 V)

sinh (V) − V − (7/50) V3 + (1/50) V3 cosh (V/5)

𝑎43 = ,
V3 cosh (2V/3)
V V2
𝛾2 = cosh ( ) − ,
5 50
2V 2V2 301V3 V
𝛾3 = cosh ( )− − sinh ( ) ,
3 9 4860 5
V2 1 V 1 2V
𝛾4 = cosh (V) − − V3 (− sinh ( ) + sinh ( )) ,
2 50 5 25 3
5 V
𝛾̂2 = sinh ( ) ,
V 5

2 2V 2 (cosh (2V/3) − 1 + V2 /27) sinh (V/5)

𝛾̂3 = sinh ( ) − ,
3V 3 3 cosh (V/5) V2

1 2V V (cosh (V) − 1 − (3V2 /10 − (2V2 /35) cosh (V/5))) sinh (2V/3)
𝛾̂4 = sinh (V) + sinh ( ) − .
V 35 5 V cosh (2V/3)
(28)

Next, we solve (17) to (22) and use the above coefficients to

find 𝑏1 , 𝑏2 , 𝑏1󸀠 , 𝑏2󸀠 , 𝑏1󸀠󸀠 , and 𝑏2󸀠󸀠 .

3 cosh (V/5) sinh (2V/3) − cosh (2V/3) sinh (V/5) −2 cosh (V/5) cosh (V) + 2 cosh (V/5) + cosh (V/5) V2 + 2 sinh (V) sinh (V/5) − 2V sinh (V/5)
𝑏1 = + ,
112 sinh (V/5) 2V3 sinh (V/5)
3 sinh (2V/3) −2 cosh (V) + 2 + V2
𝑏2 = − − ,
112 sinh (V/5) 2V3 sinh (V/5)
9 cosh (V/5) sinh (2V/3) − cosh (2V/3) sinh (V/5) cosh (V/5) sinh (V) − cosh (V/5) V − cosh (V) sinh (V/5) + sinh (V/5)
𝑏1󸀠 = − ,
56 sinh (V/5) V2 sinh (V/5)

𝑏2󸀠 = −
9 sinh (2V/3) sinh (V) − V
+ 2 ,
(29)
56 sinh (V/5) V sinh (V/5)
1 + (27/56) V sinh (2V/3) + (5/48) V sinh (V) − cosh (V)
𝑏2󸀠󸀠 = − ,
V sinh (V/5)
𝑏1󸀠󸀠
((27/56) cosh (V/5) V sinh (2V/3) + (5/48) cosh (V/5) V sinh (V) − cosh (V/5) cosh (V) + cosh (V/5) − (27/56) V cosh (2V/3) sinh (V/5) − (5/48) V cosh (V) sinh (V/5) + sinh (V) sinh (V/5))
= .
(V sinh (V/5))

57722879
These lead to our new exponentially fitted Runge-Kutta − V10 + ⋅ ⋅ ⋅ ,
type four-stage fifth-order explicit MRKT method denoted as 134057055132000000000
EFMRKT5. The corresponding Taylor series expansion of the 5 1 2 2921
solution is given by 𝑏2 = − V − V4
42 2160 136080000
1 1 2 101 1361
𝑏1 = + V + V4 − V6
48 2160 136080000 45927000000
5713 54293587
− V6 + V8
183708000000 40920957000000000
11330339 6964030429
− V8 + V10 + ⋅ ⋅ ⋅ ,
81841914000000000 670285275660000000000
Journal of Applied Mathematics 5

1 17 149 2216008103
𝑏1󸀠 = − V4 − V6 + V8
24 283500 218700000 1104865839000000000
1055069 15664491766661
− V8 − V10 + ⋅ ⋅ ⋅ ,
341007975000000 484759886861250000000000
23025689 7 221 2 114463 4
− V10 + ⋅ ⋅ ⋅ , 𝑎43 = − V + V
2538959377500000000 150 135000 637875000
25 241 22871 1010472889
𝑏2󸀠 = + V4 + V6 − V6
84 2268000 6123600000 34445250000000
112778137 222576819697
+ V8 + V8
2728063800000000 42625996875000000
13599351683 84318883418716333
+ V10 + ⋅ ⋅ ⋅ , − V10 + ⋅ ⋅ ⋅ ,
55857106305000000000 89770349418750000000000
1 1 167 1 1 1
𝑏1󸀠󸀠 = + V4 + V6 𝛾2 = 1 + V4 + V6 + V8
24 21600 58320000 15000 11250000 15750000000
528389 1
+ V8 + V10 + ⋅ ⋅ ⋅ ,
16533720000000 35437500000000
967343 59 2 9569 4 74831
+ V10 + ⋅ ⋅ ⋅ 𝛾3 = 1 − V − V − V6
5845851000000000 225 1215000 911250000
125 1 2867 26104429
𝑏2󸀠󸀠 = − V4 − V6 + V8
336 21600 408240000 11481750000000
3022109 495937229
− V8 − V10 + ⋅ ⋅ ⋅
16533720000000 25833937500000000
1457821 19 4 2267 6
− V10 + ⋅ ⋅ ⋅ , 𝛾4 = 1 − V2 + V + V
730731375000000 1000 4050000
1 1 1 4858267
𝑎̂21 = + V2 + V4 − V8
50 15000 11250000 255150000000
1 1 65060629
+ V6 + V8 + V10 + ⋅ ⋅ ⋅ ,
15750000000 35437500000000 344452500000000
1 1 2 1 1
+ V10 + ⋅ ⋅ ⋅ , 𝛾̂2 = 1 + V + V4 + V6
116943750000000000 150 75000 78750000
7 37 2 287 1
𝑎̂32 = + V + V4 + V8
27 12150 6561000 141750000000
68827 1
− V6 + V10 + ⋅ ⋅ ⋅ ,
516678750000 389812500000000
7097417 1 2 13 4 2032
+ V8 𝛾̂3 = 1 − V + V − V6
1674039150000000 270 60750 717609375
39720321233 136109
− V10 + ⋅ ⋅ ⋅ , + V8
621487034437500000000 2906317968750
9 43 2 3323 4 111136579 6 2726807
𝑎̂43 = − V + V − V − V10 + ⋅ ⋅ ⋅ ,
35 3000 1350000 255150000000 3596568486328125
26973882539 1 2 79 4 2161951 6
+ V8 𝛾̂4 = 1 + V + V − V
344452500000000 150 15000 2126250000
721361598388001 78384961
− V10 + ⋅ ⋅ ⋅ , + V8
51151196250000000000 425250000000
289 67 26141 5664933289891
𝑎32 = − V2 + V4 − V10 + ⋅ ⋅ ⋅
4860 729000 2755620000 170503987500000000
(30)
4671541
− V6
39858075000000 where 𝛾1 = 1, 𝛾̂1 = 1.
6 Journal of Applied Mathematics

This results in the new method called EFMRKT5. As the (31) to (34) and letting 𝑎̂21 , 𝑎̂32 , 𝑎̂43 , 𝑎32 , 𝑎43 , 𝛾2 , 𝛾3 , 𝛾4 , 𝛾̂2 ,
V 󳨀→ 0, the coefficients 𝑏1 , 𝑏2 , 𝑏1󸀠 , 𝑏2󸀠 , 𝑏2󸀠󸀠 , 𝑏2󸀠󸀠 , 𝑎32 , 𝑎42 , 𝑎̂21 , 𝛾̂3 , and 𝛾̂4 be free parameters will give
𝑎̂32 , 𝑎̂43 , 𝛾2 , 𝛾3 , 𝛾4 , 𝛾̂2 , 𝛾̂3 , and 𝛾̂4 of the new method EFMRKT5
reduce to the coefficients of the original method RKT5. That 1 − cos (V/5)
𝑎̂21 = ,
is to say, 𝑏1 (0), 𝑏2 (0), 𝑏1󸀠 (0), 𝑏2󸀠 (0), 𝑏2󸀠󸀠 (0), 𝑏2󸀠󸀠 (0), 𝑎32 (0), 𝑎42 (0), V2
𝑎̂21 (0), 𝑎̂32 (0), 𝑎̂43 (0), 𝛾2 (0), 𝛾3 (0), 𝛾4 (0), 𝛾̂2 (0), 𝛾̂3 (0), and 𝛾̂4 (0) 1 − cos (2V/3) + (1/27) V2
are identical to 𝑏1 , 𝑏2 , 𝑏1󸀠 , 𝑏2󸀠 , 𝑏2󸀠󸀠 , 𝑏2󸀠󸀠 , 𝑎32 , 𝑎42 , 𝑎̂21 , 𝑎̂32 , 𝑎̂43 , 𝛾2 , 𝛾3 , 𝑎̂32 = ,
V2 cos (V/5)
𝛾4 , 𝛾̂2 , 𝛾̂3 , and 𝛾̂4 of RKT5 method. Other than that, V 󳨀→ 0,
as EFMRKT5 method will have the same error constant as 1 − cos (V) − V2 (3/10 − (2/35) cos (V/5))
𝑎̂43 = ,
RKT5 method. V2 cos (2V/3)
−sin (2V/3) + 2V/3 + (49/4860) V3
𝑎32 = ,
V3 cos (V/5)
3.2. Trigonometrically Fitted MRKT Method. Exponentially
fitted method leads to trigonometrically fitted method when V − sin (V) − V3 (7/50 − (1/50) cos (V/5))
𝑎43 = ,
replacing V = 𝑤ℎ with 𝑖V and solving (8) to (9) to find 𝑎𝑖,𝑖−1 , V3 cos (2V/3)
𝑎̂𝑖,𝑖−1 ,𝛾𝑖 , and 𝛾̂𝑖 .
V V2
𝛾2 = cos ( ) − ,
5 50
(35)
𝑖−1 2V 2V2 301V3 V
1 𝛾3 = cos ( ) − − sin ( ) ,
𝛾𝑖 = cos (V𝑐𝑖 ) − V2 𝑐𝑖 2 − V3 ∑ 𝑎𝑖,𝑗 sin (V𝑐𝑗 ) , (31) 3 9 4860 5
2 𝑗=1
V2 1 V 1 2V
𝛾4 = cos (V) − − V3 (− sin ( ) + sin ( )) ,
2 50 5 25 3
1 − cos (V𝑐𝑖 ) − V2 ∑𝑖−2 ̂𝑖,𝑗 cos (V𝑐𝑗 )
𝑗=1 𝑎
𝑎̂𝑖,𝑖−1 = , (32) 5 V
V2 cos (V𝑐𝑖−1 ) 𝛾̂2 = sin ( ) ,
V 5
− sin (V𝑐𝑖 ) + V ⋅ 𝑐𝑖 − V3 ∑𝑖−2
𝑗=1 𝑎𝑖,𝑗 cos (V𝑐𝑗 ) 2 2V 2 (1 − cos (2V/3) + V2 /27) sin (V/5)
𝑎𝑖,𝑖−1 = , (33) 𝛾̂3 = sin ( ) + ,
V3 cos (V𝑐𝑖−1 ) 3V 3 3V2 cos (V/35)
sin (V) 2V V
sin (V𝑐𝑖 ) + V2 ∑𝑖−2 ̂𝑖,𝑗 sin (V𝑐𝑗 )
𝑗=1 𝑎
𝛾̂4 =
V
−
35
sin ( )
5
𝛾̂𝑖 = , 𝑖 = 2, . . . , 𝑠. (34)
V𝑐𝑖 (1 − cos (V) − (3V2 /10 − (2V2 /5) cos (V/5))) sin (2V/3)
+ .
V cos (2V/3)

Consider the same coefficients of fifth-order four-stage Next, solving (10) to (12), and using the above Fawzi
method developed by Fawzi et al.[15] as in Section 3.1. Solving coefficients to find 𝑏1 , 𝑏2 ,𝑏1󸀠 , 𝑏2󸀠 , 𝑏1󸀠󸀠 , and 𝑏2󸀠󸀠 ,

3 cos (V/5) sin (2V/3) − 3 cos (2V/3) sin (V/5) 2 cos (V/5) cos (V) − 2 cos (V/5) + cos (V/5) V2 + 2 sin (V) sin (V/5) − 2V sin (V/5)
𝑏1 = − ,
112 sin (V/5) 2V3 sin (V/5)
3 sin (2V/3) 2 cos (V) − 2 + V2
𝑏2 = − + ,
112 sin (V/5) 2V3 sin (V/5)
9 cos (V/5) sin (2V/3) − 9 cos (2V/3) sin (V/5) cos (V/5) sin (V) − V cos (V/5) − cos (V) sin (V/5) + sin (V/5)
𝑏1󸀠 = + ,
56 sin (V/5) V2 sin (V/5)
(25V/243) sin (V) + (80/81) cos (V) − 80/81 + (23V/48) sin (2V/3) (36)
𝑏1󸀠󸀠 = − ,
(80V/81) sin (V/5) − (V/125) sin (2V/3)
−1 + (27/56) V sin (2V/3) + (5/48) V sin (V) + cos (V)
𝑏2󸀠󸀠 = ,
V sin (V/5)
𝑏2󸀠
((27/56) cos (V/5) V sin (2V/3) + (5/48) cos (V/5) V sin (V) + cos (V/5) cos (V) − cos (V/5) − (27/56) V cos (2V/3) sin (V/5) − (5/48) V cos (V) sin (V/5) + sin (V) sin (V/5))
=−
V sin (V/5)

11330339
These lead to our new explicit trigonometrically fitted − V8
MRKT which is called TFMRKT5 method. The correspond- 81841914000000000
ing Taylor series expansion of the solution is given by 57722879
+ V10 + ⋅ ⋅ ⋅ ,
134057055132000000000
1 1 2 101
𝑏1 = − V + V4 5 1 2 2921
48 2160 136080000 𝑏2 = + V − V4
42 2160 136080000
5713 1361
+ V6 + V6
183708000000 45927000000
Journal of Applied Mathematics 7

54293587 721361598388001
+ V8 + V10 + ⋅ ⋅ ⋅ ,
40920957000000000 51151196250000000000
6964030429 289 67 26141
− V10 + ⋅ ⋅ ⋅ , 𝑎32 = + V2 + V4
670285275660000000000 4860 729000 2755620000
1 17 149 4671541
𝑏1󸀠 = − V4 + V6 + V6
24 283500 218700000 39858075000000
1055069 2216008103
− V8 + V8
341007975000000 1104865839000000000
23025689 15664491766661
+ V10 + ⋅ ⋅ ⋅ , + V10 + ⋅ ⋅ ⋅ ,
2538959377500000000 484759886861250000000000
25 241 22871 7 221 2 114463 4
𝑏2󸀠 = + V4 − V6 𝑎43 = + V + V
84 2268000 6123600000 150 135000 637875000
112778137 1010472889
+ V8 + V6
2728063800000000 34445250000000
13599351683 222576819697
− V10 + ⋅ ⋅ ⋅ , + V8
55857106305000000000 42625996875000000
1 1 167 84318883418716333
𝑏1󸀠󸀠 = + V4 − V6 + V10 + ⋅ ⋅ ⋅ ,
24 21600 58320000 89770349418750000000000
528389 1 2 1 1
+ V8 𝛾2 = 1 − V + V4 − V6
16533720000000 25 15000 11250000
967343 1 1
− V10 + ⋅ ⋅ ⋅ , + V8 − V10
5845851000000000 15750000000 35437500000000
125 1 2867 + ⋅⋅⋅ ,
𝑏2󸀠󸀠 = − V4 + V6
336 21600 408240000
59 2 9569 4 74831
3022109 𝛾3 = 1 − V − V − V6
− V8 225 1215000 911250000
16533720000000
26104429
1457821 + V8
+ V10 + ⋅ ⋅ ⋅ , 11481750000000
730731375000000
495937229
𝑎̂21 =
1
−
1
V2 +
1
V4 − V10 + ⋅ ⋅ ⋅ ,
50 15000 11250000 25833937500000000

1 1 19 4 2267 6
− V6 + V8 𝛾4 = 1 − V2 + V + V
15750000000 35437500000000 1000 4050000
1 4858267
− V10 + ⋅ ⋅ ⋅ , − V8
116943750000000000 255150000000
7 37 2 287 65060629
𝑎̂32 = − V + V4 + V10 + ⋅ ⋅ ⋅ ,
27 12150 6561000 344452500000000
68827 1 2 1 1
+ V6 𝛾̂2 = 1 − V + V4 − V6
516678750000 150 75000 78750000
7097417 1
+ V8 + V8
1674039150000000 141750000000
39720321233 1
+ V10 + ⋅ ⋅ ⋅ , − V10 + ⋅ ⋅ ⋅ ,
621487034437500000000 389812500000000
9 43 2 3323 4 111136579 6 1 2 13 4 2032
𝑎̂43 = + V + V + V 𝛾̂3 = 1 + V + V + V6
35 3000 1350000 255150000000 270 60750 717609375
26973882539 136109
+ V8 + V8
344452500000000 2906317968750
8 Journal of Applied Mathematics

2726807
+ V10 + ⋅ ⋅ ⋅ , equations and is solved using the existing Runge-Kutta of the
3596568486328125 same order.
1 2 79 4 2161951 6
𝛾̂4 = 1 − V + V + V (i) ℎ: step sizes.
150 15000 2126250000
78384961 (ii) TFMRKT5: the four-stage fifth-order trigonometri-
+ V8
425250000000 cally fitted RK type method derived in this paper.
5664933289891 (iii) EFMRKT5: the four-stage fifth-order exponentially
+ V10 + ⋅ ⋅ ⋅ ,
170503987500000000 fitted RK type method derived in this paper.
(37)
(iv) RKT5: the four-stage fifth-order RK type method
where 𝛾1 = 1, 𝛾̂1 = 1. given by Fawzi et al. [15].
This results in the new method called TFMRKT5. As
V 󳨀→ 0, the coefficients 𝑏1 , 𝑏2 , 𝑏1󸀠 , 𝑏2󸀠 , 𝑏2󸀠󸀠 , 𝑏2󸀠󸀠 , 𝑎32 , 𝑎42 , 𝑎̂21 , 𝑎̂32 , (v) RK5B: the six-stage fifth-order RK method given in
𝑎̂43 , 𝛾2 , 𝛾3 , 𝛾4 , 𝛾̂2 , 𝛾̂3 , and 𝛾̂4 of the new method TFMRKT5 Butcher [16].
reduce to the coefficients of the original method RKT5. That (vi) RKF5: the six-stage fifth-order RK method given in
is to say, 𝑏1 (0), 𝑏2 (0), 𝑏1󸀠 (0), 𝑏2󸀠 (0), 𝑏2󸀠󸀠 (0), 𝑏2󸀠󸀠 (0), 𝑎32 (0), 𝑎42 (0), Lambert [17].
𝑎̂21 (0), 𝑎̂32 (0), 𝑎̂43 (0), 𝛾2 (0), 𝛾3 (0), 𝛾4 (0), 𝛾̂2 (0), 𝛾̂3 (0), and 𝛾̂4 (0)
are identical to 𝑏1 , 𝑏2 , 𝑏1󸀠 , 𝑏2󸀠 , 𝑏2󸀠󸀠 , 𝑏2󸀠󸀠 , 𝑎32 , 𝑎42 , 𝑎̂21 , 𝑎̂32 , 𝑎̂43 , 𝛾2 , 𝛾3 , (vii) TFRK: the six-stage fifth-order trigonometrically fit-
𝛾4 , 𝛾̂2 , 𝛾̂3 , and 𝛾̂4 of RKT5 method. Other than that, V 󳨀→ 0, ted RK method given in Anastassi et al. [18].
as TFMRKT5 method will have the same error constant as
RKT5 method. Problem 2 (homogeneous linear problem).

4. Error Analysis 𝑦󸀠󸀠󸀠 (𝑥) = 2𝑦󸀠 (𝑥) ,

(39)
In this section, we will find the principal local truncation 𝑦 (0) = 0, 𝑦󸀠 (0) = 1, 𝑦󸀠󸀠 (0) = 0,
errors for 𝑦, 𝑦󸀠 , and 𝑦󸀠󸀠 (i.e., 𝜏𝑛+1 , 𝜏𝑛+1
󸀠 󸀠󸀠
, 𝜏𝑛+1 ) of the new expo-
nentially fitted and trigonometrically fitted explicit modified exact solution is
Runge-Kutta type methods, respectively. We first find the
Taylor series expansion of the actual solution 𝑦(𝑥𝑛 + ℎ), the
first derivative of the actual solution 𝑦󸀠 (𝑥𝑛 +ℎ), and the second √2𝑒√2𝑥 √2𝑒−√2𝑥
𝑦 (𝑥) = − . (40)
derivative of the actual solution 𝑦󸀠󸀠 (𝑥𝑛 + ℎ), the approximate 4 4
solution 𝑦𝑛+1 , the first derivative of the approximate solution
󸀠
𝑦𝑛+1 , and the second derivative of the approximate solution Estimated frequency 𝜔 = √2.
󸀠󸀠
𝑦𝑛+1 . The local truncation errors of 𝑦, 𝑦󸀠 , and 𝑦󸀠󸀠 are given as
Problem 3 ( homogeneous linear system).
𝜏𝑛+1 = 𝑦𝑛+1 − 𝑦 (𝑥𝑛 + ℎ) ,
󸀠 󸀠 𝑦1󸀠󸀠󸀠 (𝑥) = 8𝑦3󸀠 (𝑥) ,
𝜏𝑛+1 = 𝑦𝑛+1 − 𝑦󸀠 (𝑥𝑛 + ℎ) , (38)
󸀠󸀠 󸀠󸀠 𝑦1 (0) = 2, 𝑦1󸀠 (0) = 4, 𝑦1󸀠󸀠 (0) = 8,
𝜏𝑛+1 = 𝑦𝑛+1 − 𝑦󸀠󸀠 (𝑥𝑛 + ℎ)
𝑦2󸀠󸀠󸀠 (𝑥) = 8𝑦1󸀠 (𝑥) ,
󸀠 󸀠󸀠
The 𝜏𝑛+1 , 𝜏𝑛+1
and , 𝜏𝑛+1
of the methods are given in the (41)
Appendix. 𝑦2 (0) = 4, 𝑦2󸀠 (0) = 8, 𝑦2󸀠󸀠 (0) = 16,
󸀠 󸀠󸀠
Notes: from 𝜏𝑛+1 , 𝜏𝑛+1 , and 𝜏𝑛+1 , we can see that the order
of TFMRKT5 is order 5 because all of the coefficients up to ℎ5 𝑦3󸀠󸀠󸀠 (𝑥) = 𝑦2󸀠 (𝑥) ,
vanished.
𝑦3 (0) = 1, 𝑦3󸀠 (0) = 2, 𝑦3󸀠󸀠 (0) = 4,

5. Problems Tested and Numerical Results exact solutions are

In this section, we will apply the new explicit exponentially
fitted modified Runge-Kutta type method to some 𝑦󸀠󸀠󸀠 = 𝑦1 (𝑥) = 2𝑒2𝑥 ,
𝑓(𝑥, 𝑦, 𝑦󸀠 ) ODEs for problems (1)-(4) which consist of
exponential solutions and the new trigonometrically fitted 𝑦2 (𝑥) = 4𝑒2𝑥 , (42)
modified Runge-Kutta type method to some ODEs problems
(5)-(8) with trigonometric functions solutions. The numeri- 𝑦3 (𝑥) = 𝑒2𝑥 .
cal results are compared with the results obtained when the
same set of problems are reduced to a system of first-order Estimated frequency 𝜔 = 2.
Journal of Applied Mathematics 9

Problem 4 ( inhomogeneous linear system ). Problem 8 (inhomogeneous linear system).

𝑦1󸀠󸀠󸀠 (𝑥) = 𝑦3󸀠 (𝑥) + 1, 𝑦1󸀠󸀠󸀠 (𝑥) = −27𝑦1󸀠 (𝑥) ,

𝑦1 (0) = 2, 𝑦1󸀠 (0) = 3, 𝑦1󸀠󸀠 (0) = 5, 𝑦1 (0) = 0, 𝑦1󸀠 (0) = −1, 𝑦1󸀠 (0) = 0,

𝑦2󸀠󸀠󸀠 (𝑥) = 𝑦1󸀠 (𝑥) + 2, 𝑦2󸀠󸀠󸀠 (𝑥) = −27𝑦2󸀠 (𝑥) + cos (𝑥) ,
(43) (51)
𝑦2 (0) = 1, 𝑦2󸀠 (0) = 2, 𝑦2󸀠󸀠 (0) = 5, 𝑦2 (0) = 1, 𝑦2󸀠 (0) = −1, 𝑦2󸀠󸀠 (0) = 0,

𝑦3󸀠󸀠󸀠 (𝑥) = 𝑦2󸀠 (𝑥) + 3, 𝑦3󸀠󸀠󸀠 (𝑥) = −27𝑦3󸀠 (𝑥) ,

𝑦3 (0) = 0, 𝑦3󸀠 (0) = 4, 𝑦3󸀠󸀠 (0) = 5, 𝑦3 (0) = 1, 𝑦3󸀠 (0) = 0, 𝑦3󸀠󸀠 (0) = −1,
exact solutions are exact solutions are
𝑥
𝑦1 (𝑥) = 5𝑒 − 2𝑥 − 3, √3
𝑦1 (𝑥) = − sin (3√3𝑥) ,
𝑥
𝑦2 (𝑥) = 5𝑒 − 3𝑥 − 4, (44) 9

𝑦3 (𝑥) = 5𝑒𝑥 − 𝑥 − 5. 3√ 3 1
𝑦2 (𝑥) = 1 − sin (3√3𝑥) + sin (𝑥) , (52)
26 26
Estimated frequency 𝜔 = 1. 26 1
𝑦3 (𝑥) = + cos (3√3𝑥) .
Problem 5 (inhomogeneous linear problem). 27 27

𝑦󸀠󸀠󸀠 (𝑥) = 5𝑦󸀠 (𝑥) + sinh (𝑥) , Estimated frequency 𝜔 = 3√3.

1 1 (45) Problem 9 (inhomogeneous linear system).

𝑦 (0) = − , 𝑦󸀠 (0) = 0, 𝑦󸀠󸀠 (0) = − ,
4 4
𝑦1󸀠󸀠󸀠 (𝑥) = −7𝑦2󸀠 (𝑥) − cos (𝑥) ,
exact solution is
1
𝑒𝑥 𝑒−𝑥 𝑦1 (0) = , 𝑦1󸀠 (0) = 0, 𝑦1󸀠󸀠 (0) = −1,
𝑦 (𝑥) = − − . (46) 7
8 8
𝑦2󸀠󸀠󸀠 (𝑥) = −7𝑦1󸀠 (𝑥) − cos (𝑥) ,
Estimated frequency 𝜔 = 1. (53)
𝑦2 (0) = 1, 𝑦2󸀠 (0) = 0, 𝑦2󸀠󸀠 (0) = −1,
Problem 6 (homogeneous linear problem).
𝑦3󸀠󸀠󸀠 (𝑥) = −7𝑦3󸀠 (𝑥) − cos (𝑥) ,
𝑦󸀠󸀠󸀠 (𝑥) = −25𝑦󸀠 (𝑥) ,
(47) 𝑦3 (0) = 0, 𝑦3󸀠 (0) = 1, 𝑦3󸀠󸀠 (0) = 0,
𝑦 (0) = 0, 𝑦󸀠 (0) = 0, 𝑦󸀠󸀠 (0) = 1,
exact solutions are
exact solution is
1 1 1 √7 1
𝑦 (𝑥) = − cos (5𝑥) . (48) 𝑦1 (𝑥) = − sin (𝑥) + sin (√7𝑥) + cos (√7𝑥) ,
25 25 6 42 7
Estimated frequency 𝜔 = 5. 1 √7 6
𝑦2 (𝑥) = − sin (𝑥) + sin (𝑥) √7 +
6 42 7
Problem 7 (inhomogeneous linear problem). (54)
1
+ cos (√7𝑥) ,
𝑦󸀠󸀠󸀠 (𝑥) = −27𝑦󸀠 (𝑥) + sin (𝑥) , 7
(49)
√7 1
𝑦 (0) = 1, 𝑦󸀠 (0) = −1, 𝑦󸀠󸀠 (0) = 0, 𝑦3 (𝑥) = sin (√7𝑥) − sin (𝑥) .
6 6
exact solution is
Estimated frequency 𝜔 = 1.
√3 √3
𝑦 (𝑥) = cos (3√3𝑥) − sin (3√3𝑥)
702 9 6. An Application to a Problem in
(50)
1 28 Thin Film Flow
− cos (𝑥) + .
26 27
Here, we will use the suggested method to a famous problem
Estimated frequency 𝜔 = 3√3. in engineering and physics based on the thin film flow
10 Journal of Applied Mathematics

of a liquid. Many researchers in the literature explain this 2

problem more. Momoniat and Mahomed[19] constructed
0
symmetry reduction and numerical solution of a third-order
ODE from thin film flow. Tuck and Schwartz [20] discussed

ＦＩＡ10 (Max global error)

−2
the movement of a thin film of viscous fluid over a solid
surface and taken into account tension and gravity, as well as −4
viscosity. The problem was evaluated and solved using third-
order ODE as follows: −6
3
𝑑𝑦 −8
= 𝑓 (𝑦) . (55)
𝑑𝑥3
Many forms of the function were studied by [20] for the −10
drainage dry surface; it has the form of 𝑓(𝑦) which can be
−12
stated as 1.4 1.6 1.8 2 2.2 2.4 2.6 2.8 3
3 ＦＩＡ10 (Number of function evaluations)
𝑑𝑦 1
3
= −1 + 2 . (56)
𝑑𝑥 𝑦 EFMRKT5 RK5B
RKT5 RKF5
When the surface is prewetted by a thin film with thickness
𝛿 > 0 (where 𝛿 > 0 is very small), the function 𝑓 is given by Figure 1: The efficiency curve for EFMRKT5, RKT5, RK5B, and
RKF5 for Problem 2 with 𝑥𝑒𝑛𝑑 = 5 and ℎ = 0.1, 0.25, 0.5, 0.75.
1 + 𝛿 + 𝛿2 𝛿 + 𝛿2
𝑓 (𝑦) = −1 + − (57)
𝑦2 𝑦3
Problems concerning the flow of thin films of viscous fluid We have taken 𝑥0 = 0 and 𝑦0 = 𝑦0󸀠 = 𝑦0󸀠󸀠 = 1. Unfortunately,
with a free surface in which surface tension effects play a role for general 𝑘, (58) cannot be solved analytically. However, we
typically lead to third-order ODEs governing the shape of the can use these reductions to determine an efficient way to solve
free surface of the fluid,𝑦 = 𝑦(𝑥). As indicated by [20], one (1) numerically. Here, we are focusing on the cases 𝑘 = 2 and
such equation is 𝑘 = 3 (see Mechee et al.[22]).
𝑦󸀠󸀠󸀠 (𝑥) = 𝑦−𝑘 , 𝑥 ≥ 𝑥0 (58)
7. Discussion and Conclusion
with initial conditions
𝑦 (𝑥0 ) = 𝑦0 , In this research, we have derived exponentially fitted and
trigonometrically fitted explicit modified Runge-Kutta type
𝑦󸀠 (𝑥0 ) = 𝑦0󸀠 , (59) methods for solving 𝑦󸀠󸀠󸀠 (𝑥) = 𝑓(𝑥, 𝑦, 𝑦󸀠 ) with application
to thin film flow problem. Consequently, the new four-stage
𝑦󸀠󸀠 (𝑥0 ) = 𝑦0󸀠󸀠 , fifth-order exponentially-fitted and trigonometrically-fitted
methods which are denoted as EFMRKT5 and TFMRKT5,
where 𝑦0 , 𝑦0󸀠 , and 𝑦0󸀠󸀠 are constants, which is of specific respectively, were constructed and we used in numerical
significance since it portrays the dynamic balance amongst comparison the criteria based on computing the maximum
surface and gooey strengths in a thin fluid layer in disregard error in the solution (max(|𝑦(𝑡𝑛 ) − 𝑦𝑛 |)) which is equal to the
of gravity. For compare and contrast, we utilized Runge- maximum between absolute errors of the actual solutions and
Kutta methods which are fifth-order (RKT5, RK5B, RKF5, computed solutions. The numerical outcomes are plotted in
and TFRKT) strategies, individually. To utilize Runge-Kutta Figures 1–8. Figures 1–8 demonstrate that the new TFMRKT5
techniques we write (1) as a system of three first-order and EFMRKT5 methods require less capacity assessments
equations. Biazar et al. [21] we can write (58) as the following than the RKT5, RK5B, RKF5, and TFRK methods. The
system: figures showed the efficiency of the new methods where the
𝑑𝑦1 common logarithm of the maximum global error throughout
= 𝑦2 (𝑥) , the integration versus computational cost was measured by
𝑑𝑥
the number of function evaluations. The numerical results
𝑑𝑦2 obtained showed clearly that the global error for a short
= 𝑦3 (𝑥) , (60)
𝑑𝑥 period of integration for the new exponentially fitted method
and for a large period of integration for the new trigono-
𝑑𝑦2
= 𝑦1−𝑘 , metrically fitted explicit modified Runge-Kutta type method
𝑑𝑥 is smaller than that of the other existing methods. The new
where EFMRKT5 and TFMRKT5 methods are much more efficient
𝑦1 (0) = 1, than the other existing methods when solving third-order
ODEs of the form 𝑦󸀠󸀠󸀠 = 𝑓(𝑥, 𝑦, 𝑦󸀠 ) straightforwardly.
𝑦2 (0) = 1, (61) For Tables 2 and 3 we observed that the numerical results
using TFMRKT5 and EFMRKT5 methods are correct to five
𝑦3 (0) = 1. decimal places. Applying RK5B, RKF5, TFRK, and RKT5 to
Journal of Applied Mathematics 11

Table 2: Numerical results for problem in Thin Film Flow (58) taking ℎ = 0.1 and 𝑘 = 2.

𝑥 Exact Solution RK5B RKF5 RKT5 EMFRKT5 TFRK TFMRKT5

0.0 1.000000000 1.000000000 1.000000000 1.000000000 1.000000000 1.000000000 1.000000000
0.2 1.221211030 1.2212100068 1.2212100097 1.2212100039 1.2212100052 1.2212100218 1.2212100052
0.4 1.488834893 1.4888347851 1.4888347895 1.4888347797 1.4888347885 1.4888348090 1.4888347885
0.6 1.807361404 1.8073614063 1.8073614114 1.8073613988 1.8073614237 1.8073614357 1.8073614237
0.8 2.179819234 2.1798192463 2.1798192513 2.1798192371 2.1798192873 2.1798192788 2.1798192873
1.0 2.608275822 2.6082748841 2.6082748883 2.6082748735 2.6082749587 2.6082749176 2.6082749587

Table 3: Numerical results for problem in thin film flow (58) taking ℎ = 0.01 and 𝑘 = 2.

𝑥 Exact Solution RK5B RKF5 RKT5 EFMRKT5 TFRK TFMRKT5

0.0 1.000000000 1.000000000 1.000000000 1.000000000 1.000000000 1.000000000 1.000000000
0.2 1.221211030 1.2212100045 1.2212100045 1.2212100045 1.2212100045 1.2212100045 1.2212100045
0.4 1.488834893 1.4888347799 1.4888347799 1.4888347799 1.4888347799 1.4888347799 1.4888347799
0.6 1.807361404 1.8073613977 1.8073613977 1.8073613977 1.8073613977 1.8073613977 1.8073613977
0.8 2.179819234 2.1798192339 2.1798192339 2.1798192339 2.1798192340 2.1798192339 2.1798192340
1.0 2.608275822 2.6082748676 2.6082748676 2.6082748676 2.6082748677 2.6082748676 2.6082748677

(58) for 𝑘 = 2 also yields five-decimal place accuracy. Tables 2

4 and 5 show the numerical results for the case 𝑘 = 3 with 0
ℎ = 0.1 and ℎ = 0.01 since for 𝑘 = 3, Problem (58) cannot
be solved analytically. Table 4 shows that TFMRKT5 and −2
ＦＩＡ10 (Max global error)

EFMRKT5 manage to achieve the numerical results which

−4
agree to seven decimal places when compared to RK5B,
RKF5, TFRK, and RKT5 for ℎ = 0.1. In Table 5 the numerical −6
results for TFMRKT5 and EFMRKT5 agree to nine decimal
−8
places when compared to RK5B, RKF5, TFRK, and RKT5
for ℎ = 0.01. For Table 7 we observe that RK5B, RKF5, −10
RKT5, TFRK, TFMRKT5, and EFMRKT5 have similar order
of accuracy. In Table 6 values of the error are different. −12
Therefore it is consistent with results displayed in Tables 2 −14
and 3. Figures 9 and 10 show that the new EFMRKT5 and 1 1.2 1.4 1.6 1.8 2 2.2 2.4 2.6
TFMRKT5 methods require less function evaluations than ＦＩＡ10 (Number of function evaluations)
the RK5B, RKF5, TFRK, and RKT5 methods. This is because
EFRKTG5 RK5B
when problem (58) is solved using RK5B, RKF5, TFRK, and RKT5 RKF5
RKT5 methods, it needs to be reduced to a system of first-
Figure 2: The efficiency curve for EFMRKT5, RKT5, RK5B, and
order equations which is three times the dimension.
RKF5 for Problem 3 with 𝑥𝑒𝑛𝑑 = 2 and ℎ = 0.1, 0.25, 0.5, 0.75.

Appendix
1 1 1
The principal local truncation errors for 𝑦, 𝑦󸀠 , and 𝑦󸀠󸀠 (i.e., − 𝑦 𝐹𝑥𝑦 − 𝐹𝑦𝐹 − 𝑦 𝐹𝑦𝑥𝑥
󸀠 󸀠󸀠
3600 𝑥𝑥 144 3600 𝑥
𝜏𝑛+1 , 𝜏𝑛+1 , 𝜏𝑛+1 ) for EFMRKT5 are as follows:
1 527
− 𝐹𝑥𝑧𝑧𝑦𝑥𝑥 2 + 𝐹𝑥𝑧𝐹
1 1 3600 162000
𝜏𝑛+1 = (− 𝑦𝑥𝑥 2 𝐹𝑦𝑧 − 𝑦 3 𝐹𝑦𝑦𝑦
3600 10800 𝑥 1 971
− 𝑦𝑥𝑥 3 𝐹𝑧𝑧𝑧 + 𝐹𝑧𝐹𝑥
1 527 10800 907200
− 𝑦𝑥 𝐹𝑦𝑦𝑦𝑥𝑥 + 𝑦 𝐹𝑦𝑧𝐹
3600 162000 𝑥 971 1
+ 𝐹𝑧2 𝑦𝑥𝑥 − 𝑦 2 𝐹𝑥𝑦𝑦
527 1 907200 3600 𝑥
+ 𝐹𝑧𝑧𝑦𝑥x 𝐹 − 𝑦 𝐹𝑥𝑦𝑧𝑦𝑥𝑥
162000 1800 𝑥 1 1
+ 𝑤2 𝐹𝑧𝑦𝑥𝑥 − 𝐹𝑧𝑥𝑥𝑦𝑥𝑥
1 1 10800 3600
− 𝑦 2 𝐹𝑧𝑦𝑦𝑦𝑥𝑥 − 𝑦 𝑦 2 𝐹𝑧𝑦𝑧
3600 𝑥 3600 𝑥 𝑥𝑥 1
+ 𝑤2 𝐹𝑦𝑦𝑥 ) ℎ6 + 𝑂 (ℎ7 )
971 1 1 10800
+ 𝐹𝑧𝐹𝑦𝑦𝑥 − 𝐹𝑥𝑥𝑥 + 𝑤2 𝐹𝑥 (A.1)
907200 10800 10800
12 Journal of Applied Mathematics

Table 4: Numerical results for problem in thin film flow (58) taking ℎ = 0.1 and 𝑘 = 3.

𝑥 RK5B RKF5 RKT5 EMFRKT5 TFRK TFMRKT5

0.0 1.000000000 1.000000000 1.000000000 1.000000000 1.000000000 1.000000000
0.2 1.2211551491 1.2211551546 1.2211551394 1.2211551412 1.2211551831 1.2211551412
0.4 1.4881052974 1.4881053065 1.4881052807 1.4881052926 1.4881053519 1.4881052926
0.6 1.8042625677 1.8042625794 1.8042625459 1.8042625786 1.8042626364 1.8042625786
0.8 2.1715228242 2.1715228376 2.1715227987 2.1715228633 2.1715229031 2.1715228633
1.0 2.5909582923 2.5909583063 2.5909582638 2.5909583715 2.5909583783 2.5909583715

Table 5: Numerical results for problem in thin film flow (58) taking ℎ = 0.01 and 𝑘 = 3.

𝑥 RK5B RKF5 RKT5 EMFRKT5 TFRK TFMRKT5

0.0 1.000000000 1.000000000 1.000000000 1.000000000 1.000000000 1.000000000
0.2 1.2211551424 1.2211551424 1.2211551424 1.2211551424 1.2211551424 1.2211551424
0.4 1.4881052842 1.4881052842 1.4881052842 1.4881052842 1.4881052842 1.4881052842
0.6 1.8042625481 1.8042625481 1.8042625481 1.8042625482 1.8042625481 1.8042625482
0.8 2.1715227981 2.1715227981 2.1715227981 2.1715227982 2.1715227981 2.1715227982
1.0 2.5909582591 2.5909582591 2.5909582591 2.5909582592 2.5909582591 2.5909582592

2 2

0 0

−2 −2
ＦＩＡ10 (Max global error)

ＦＩＡ10 (Max global error)

−4 −4

−6 −6

−8 −8

−10 −10

−12 −12

−14 −14
1.4 1.6 1.8 2 2.2 2.4 2.6 2.8 3 3.2 1.2 1.4 1.6 1.8 2 2.2 2.4 2.6 2.8 3
ＦＩＡ10 (Number of function evaluations) ＦＩＡ10 (Number of function evaluations)

EFMRKT5 RK5B EFMRKT5 RK5B

RKT5 RKF5 RKT5 RKF5

Figure 3: The efficiency curve for EFMRKT5, RKT5, RK5B, and Figure 4: The efficiency curve for EFMRKT5, RKT5, RK5B, and
RKF5 for Problem 4 with 𝑥𝑒𝑛𝑑 = 6 and ℎ = 0.1, 0.25, 0.5, 0.75. RKF5 Problem 5 with 𝑥𝑒𝑛𝑑 = 4 and ℎ = 0.1, 0.25, 0.5, 0.75.

󸀠 1 1811 1 1
𝜏𝑛+1 =( 𝐹𝑥𝑥𝑥𝑥 + 𝐹𝑧𝑦𝑥 𝐹𝑥𝑦 + 𝑦 𝑦 2 𝐹𝑥𝑦𝑧𝑧 + 𝑦 2 𝑦 𝐹𝑥𝑦𝑦𝑧
21600 756000 1800 𝑥 𝑥𝑥 1800 𝑥 𝑥𝑥
1063 971 1 1
+ 𝑦𝑥𝑥 2 𝐹𝑧𝑧𝑧𝐹 + 𝐹𝑦𝑦𝑥 𝐹𝑥𝑧 + 𝑦𝑥 3 𝐹𝑧𝑦𝑦𝑦𝑦𝑥𝑥 + 𝑦 𝑦 3 𝐹𝑧𝑧𝑧𝑦
162000 226800 5400 5400 𝑥 𝑥𝑥
24853 15143 971 1063
+ 𝐹𝑧𝑦𝑥𝑥 2 𝐹𝑧𝑧 + 𝐹𝑧𝑦𝑥𝑥 𝐹𝑥𝑧 + 𝐹𝑦𝑦𝑥 2 𝐹𝑦𝑧 + 𝑦 𝐹𝑥𝑦𝑧𝐹
4536000 2268000 226800 81000 𝑥
1 1811 1063 1
+ 𝑦 2 𝑦 2 𝐹𝑧𝑧𝑦𝑦 + 𝐹𝑧𝑦𝑥 2 𝐹𝑦𝑦 + 𝑦 2 𝐹𝑧𝑦𝑦𝐹 + 𝑦 2 𝐹𝑦𝑦𝑦𝑦𝑥𝑥
3600 𝑥 𝑥𝑥 1512000 162000 𝑥 3600 𝑥
971 971 8689 1
+ 𝐹𝑧𝑧𝑦𝑥𝑥 𝐹𝑥 + 𝑦 𝐹𝑦𝑧𝐹𝑥 − 𝐹𝑧𝐹𝑦𝑦𝑥𝑥 + 𝑦 𝐹𝑥𝑦𝑦𝑦𝑥𝑥
226800 226800 𝑥 1512000 1800 𝑥
2087 1 1063 317
− 𝑦𝑥𝑥 𝐹𝑦𝑧𝐹 + 𝑦 𝑦 𝐹𝑥𝑥𝑦𝑧 + 𝐹𝑥𝑧𝑧𝑦𝑥𝑥 𝐹 + 𝐹𝑧𝐹𝑤2
162000 1800 𝑥 𝑥𝑥 81000 22680000
Journal of Applied Mathematics 13

Table 6: Comparison of error for problem in thin film flow (58) taking ℎ = 0.1 and 𝑘 = 2.

𝑥 RK5B RKF5 RKT5 EFMRKT5 TFRK TFMRKT5

0.0 0.0000(0) 0.0000(0) 0.0000(0) 0.0000(0) 0.0000(0) 0.0000(0)
0.2 1.0230 (-6) 1.0200(-6) 1.2600(-6) 1.0250(-6) 1.0080(-6) 1.0250(-6)
0.4 1.0800(-7) 1.0300(-7) 1.1300(-7) 1.0500(-7) 8.4100(-7) 1.0500(-7)
0.6 2.0000(-9) 7.0000(-9) 5.0000(-8) 2.0000(-8) 3.2000(-8) 2.0000(-8)
0.8 1.2000(-8) 1.7000(-8) 3.0000(-9) 5.300(-8) 4.5000(-8) 5.3000(-8)
1.0 9.3800(-7) 9.3400(-7) 9.4800 (-7) 8.6300 (-7) 9.0400(-7) 8.6300(-7)

Table 7: Comparison of error for problem in thin film flow (58) taking ℎ = 0.01 and 𝑘 = 2.

𝑥 RK5B RKF5 RKT5 EFMRKT5 TFRK TFMRKT5

0.0 0.0000(0) 0.0000(0) 0.0000(0) 0.0000(0) 0.0000(0) 0.0000(0)
0.2 1.0260(-6) 1.0260(-6) 1.0260(-6) 1.0260(-6) 1.0260(-6) 1.0260(-6)
0.4 6.0000(-7) 6.0000(-7) 6.0000(-7) 6.0000(-7) 6.0000(-7) 6.0000(-7)
0.6 9.0000(-9) 9.0000(-9) 9.0000(-9) 9.0000(-9) 9.0000(-9) 9.0000(-9)
0.8 0.0000(0) 0.0000(0) 0.0000(0) 0.0000(0) 0.0000(0) 0.0000(0)
1.0 9.5400(-7) 9.5400(-7) 9.5400(-7) 9.5400(-7) 9.5400(-7) 9.5400(-7)

0 1
−1 0
−2 −1
ＦＩＡ10 (Max global error)

ＦＩＡ10 (Max global error)

−3 −2
−4 −3
−5 −4
−6 −5
−7 −6
−8 −7
−9 −8
5.6 5.8 6 6.2 6.4 6.6 6.8 7 5.6 5.8 6 6.2 6.4 6.6 6.8 7
ＦＩＡ10 (Number of function evaluations) ＦＩＡ10 (Number of function evaluations)

TFMRKT5 RKF5 TFMRKT5 RKF5

RKT5 TFRK RKT5 TFRK
RK5B RK5B
Figure 5: The efficiency curve for TFMRKT5, RKT5, RK5B,
RKF5, and TFRK for Problem 6 with 𝑥𝑒𝑛𝑑 = 10000 and ℎ = Figure 6: The efficiency curve for TFMRKT5, RKT5, RK5B,
0.025, 0.05, 0.075, 0.1. RKF5, and TFRK for Problem 7 with 𝑥𝑒𝑛𝑑 = 10000 and ℎ =
0.025, 0.05, 0.075, 0.1.

1 7
+ 𝑦 𝐹𝑧𝑦𝑦𝑦𝑥𝑥 2 − 𝑦 𝐹𝑦𝑦𝐹
1800 𝑥 360 𝑥
1811 1
1 1 + 𝐹𝑧𝐹𝑥𝑥 + 𝑦 2 𝐹𝑥𝑥𝑧𝑧
+ 𝑦𝑥 𝐹𝑥𝑦𝑥𝑥 + 𝑦 4 𝐹𝑦𝑦𝑦𝑦 1512000 3600 𝑥𝑥
5400 21600 𝑥
1 1
1 31211 + 𝑦 2 𝐹𝑥𝑦𝑥𝑦 + 𝑦 2 𝐹𝑦𝑦
+ 𝑦𝑥𝑥 𝐹𝑦𝑥𝑥 + 𝐹𝑧2 𝐹 3600 𝑥 7200 𝑥𝑥
3600 22680000
1063 7 1 57889 1
+ 𝐹𝑧𝑥𝑥𝐹 − 𝐹𝐹𝑥𝑦 + 𝑦 3 𝐹𝑥𝑦𝑦𝑦 + 𝐹𝑧𝑧𝐹2 + 𝑦 2 𝐹𝑥𝑦𝑧
162000 360 5400 𝑥 14580000 1800 𝑥𝑥
1 1 1 1 1
− 𝐹𝑦𝐹𝑥 + 𝑦 3 𝐹𝑥𝑧𝑧𝑧 + 𝑦𝑥𝑥 3 𝐹𝑧𝑦𝑧 − 𝐹𝑦2 𝑦𝑥 + 𝐹𝑥𝑥𝑥𝑧𝑦𝑥𝑥
144 5400 𝑥𝑥 3600 144 5400
1 971 1 1063
+ 𝑦𝑥𝑥 4 𝐹𝑧𝑧𝑧𝑧 + 𝐹𝑥𝐹𝑥𝑧 − 𝑤4 𝐹 + 𝑦 𝑦 𝐹𝑧𝑦𝑧𝐹
21600 226800 21600 81000 𝑥 𝑥𝑥
14 Journal of Applied Mathematics

2 −1
1

ＦＩＡ10 (Number of function evaluations)

−1.5
0
ＦＩＡ10 (Max global error)

−1
−2
−2
−3 −2.5
−4
−5 −3

−6
−3.5
−7
−8 −4
5.6 5.8 6 6.2 6.4 6.6 6.8 7 1.5 2 2.5 3 3.5 4 4.5 5 5.5
ＦＩＡ10 (Number of function evaluations) ＦＩＡ10 (h)

TFMRKT5 RKF5 EFMRKT5 RKF5

RKT5 TFRK RKT5 RK5B
RK5B
Figure 9: Plot of graph for function evaluations against step size ℎ
Figure 7: The efficiency curve for TFMRKT5, RKT5, RK5B, for Problem (58) taking 𝑥𝑒𝑛𝑑 = 1,ℎ = 0.1, 0.01, 0.001, 0.0001, and
RKF5, and TFRK for Problem 8 with 𝑥𝑒𝑛𝑑 = 10000 and ℎ = 𝑘 = 2.
0.025, 0.05, 0.075, 0.1.

−1
−1
ＦＩＡ10 (Number of function evaluations)

−1.5
−2
−2
ＦＩＡ10 (Max global error)

−3

−4 −2.5

−5 −3

−6 −3.5

−7
−4
1.5 2 2.5 3 3.5 4 4.5 5 5.5
−8
5.6 5.8 6 6.2 6.4 6.6 6.8 7 ＦＩＡ10 (h)
ＦＩＡ10 (Number of function evaluations)
TFMRKT5 RK5B
TFMRKT5 RKF5 RKT5 TFRK
RKT5 TFRK RKF5
RK5B Figure 10: Plot of graph for function evaluations against step size
Figure 8: The efficiency curve for TFMRKT5, RKT5, RK5B, ℎ for Problem (58) taking 𝑥𝑒𝑛𝑑 = 1,ℎ = 0.1, 0.01, 0.001, 0.0001 and
RKF5, and TFRK for Problem 9 with 𝑥𝑒𝑛𝑑 = 10000 and ℎ = 𝑘 = 2.
0.025, 0.05, 0.075, 0.1.

32703767 3551
+ 𝐹𝑧𝐹𝐹𝑥𝑧 + 𝐹𝑧𝑦𝑥 3 𝐹𝑦𝑦𝑦
3061800000 3189375
971 15143
+ 𝐹𝑦𝑦𝑥 𝐹𝑧𝑧𝑦𝑥𝑥 + 𝐹𝑧𝑦𝑥 𝐹𝑦𝑧𝑦𝑥𝑥 ) 1 1
226800 2268000 − 𝑦𝑥 3 𝐹𝑥𝑦𝑦𝑦𝑧𝑦𝑥𝑥 − 𝑦 2 𝐹𝑥𝑥𝑥𝑧𝑧
5400 10800 𝑥𝑥
⋅ ℎ6 + 𝑂 (ℎ7 )
276941 10163 2
+ 𝑦 𝐹𝑧𝑧𝑧𝐹2 + 𝑦 𝐹𝑧𝑦𝑦𝐹𝑥
(A.2) 14580000 𝑥𝑥 907200 𝑥
󸀠󸀠 7589 13777 10163 1
𝜏𝑛+1 = (− 𝐹𝑦𝑦𝑥𝑥 𝐹𝑥𝑧 − 𝑦 𝐹𝑦𝑧𝐹𝑥 + 𝐹𝑥𝐹𝑥𝑧𝑧𝑦𝑥𝑥 − 𝑦 𝑦 2 𝐹𝑥𝑦𝑦𝑧
907200 907200 𝑥𝑥 453600 1800 𝑥 𝑥𝑥
5011 1193 1 1511
+ 𝑦𝑥 𝐹𝑧𝑧𝑦𝑥𝑥 2 𝐹𝑦𝑧 + 𝑦 𝐹𝑥𝑥𝑧𝑧𝐹 − 𝑦𝑥 𝑦𝑥𝑥 3 𝐹𝑧𝑧𝑦𝑦 + 𝐹𝑤2 𝐹𝑦𝑧𝑦𝑥
302400 40500 𝑥𝑥 3600 7560000
Journal of Applied Mathematics 15

19 392059 1 1
− 𝑦 𝐹𝑦𝑦𝐹𝑥 − 𝐹𝑧𝑦𝑥 𝐹𝑦𝑦𝑦𝑥𝑥 − 𝑦 2 𝐹𝑥𝑥𝑦𝑧 − 𝑦 3 𝐹𝑥𝑦𝑧𝑧
720 𝑥 17010000 3600 𝑥𝑥 3600 𝑥𝑥
5011 179 1
+ 𝑦 2 𝐹𝑦𝑦𝐹𝑧𝑧𝑦𝑥𝑥 − 𝐹𝑧𝐹𝑦𝐹 − 𝑦 4 𝐹𝑥𝑦𝑦𝑦𝑦
907200 𝑥 567000 21600 𝑥
989489 1 989489
+ 𝐹𝑧𝑦𝑥 2 𝐹𝑧𝑦𝑦𝑦𝑥𝑥 − 𝑦𝑥 𝐹𝑥𝑥𝑥𝑦𝑧𝑦𝑥𝑥 + 𝐹𝑧𝑥𝑥𝐹𝑧𝑦𝑥𝑥
68040000 5400 68040000
5011 10163 21449 46577
+ 𝑦 𝐹𝑥𝑧𝐹𝑦𝑧𝑦𝑥𝑥 + 𝐹𝑦𝑦𝑥 𝐹𝑥𝑧𝑧𝑦𝑥𝑥 + 𝐹𝑧𝑧𝐹𝐹𝑥 + 𝐹𝑧2 𝐹𝑦𝑦𝑥
226800 𝑥 453600 1632960 34992000
5011 1 3551
+ 𝑦 𝐹𝑥𝑦𝐹𝑧𝑧𝑦𝑥𝑥 − 𝑦 2 𝐹𝑥𝑦𝑦𝑦𝑦𝑥𝑥 + 𝐹𝑧𝑦𝑥 𝐹𝑦𝑥𝑥
453600 𝑥 3600 𝑥 1063125
989489 10163 276941
+ 𝐹𝑧𝑦𝑥 𝐹𝑥𝑦𝑧𝑦𝑥𝑥 + 𝐹𝑦𝑦𝑥 𝐹𝑧𝑥𝑥 + 𝑦 𝐹2 𝐹𝑧𝑦𝑧
34020000 907200 14580000 𝑥
10163 1289 1
+ 𝑦 𝐹𝑥𝐹𝑧𝑦𝑧𝑦𝑥𝑥 − 𝐹𝑦𝑦𝑥𝑥 2 𝐹𝑧𝑧 − 𝑦 3 𝐹𝑥𝑥𝑧𝑧𝑧
453600 𝑥 907200 10800 𝑥𝑥
10163 10163 1
+ 𝐹𝑦𝑦𝑥 2 𝐹𝑧𝑦𝑧𝑦𝑥𝑥 + 𝐹𝑦𝑦𝑥 2 𝐹𝑥𝑦𝑧 + 𝑤4 𝐹𝑧𝑦𝑥𝑥
453600 453600 108000
875857 5011 10163
+ 𝑦 𝑦 2 𝐹𝑧𝑦𝑧𝐹𝑧 + 𝑦 𝐹𝑥𝑧𝐹𝑥𝑦 + 𝑦 𝐹𝑥𝐹𝑥𝑦𝑧
34020000 𝑥 𝑥𝑥 453600 𝑥 453600 𝑥
1 1 1193 1193
− 𝑦 2 𝑦 3 𝐹𝑧𝑧𝑧𝑦𝑦 − 𝑦 𝑦 4 𝐹𝑧𝑧𝑧𝑧𝑦 + 𝑦 2 𝐹𝑥𝑧𝑧𝑧𝐹 + 𝑦 3 𝐹𝑧𝑧𝑧𝑧𝐹
10800 𝑥 𝑥𝑥 21600 𝑥 𝑥𝑥 40500 𝑥𝑥 121500 𝑥𝑥
1 3551 1187 10163
− 𝑦𝑥 2 𝐹𝑥𝑥𝑦𝑦𝑧𝑦𝑥𝑥 + 𝐹𝑧𝑦𝑥 2 𝐹𝑥𝑦𝑦 − 𝐹𝑦𝑦𝑥 𝐹𝑦𝑧𝑦𝑥𝑥 + 𝐹𝑥𝐹𝑧𝑥𝑥
3600 1063125 50400 907200
2513939 1 276941
+ 𝑦 3 𝐹𝑧𝑧𝑧𝐹𝑧 − 𝑦𝑥 𝐹𝑥𝑥𝑥𝑥𝑦 + 𝐹𝑥𝑧𝑧𝐹2
204120000 𝑥𝑥 21600 14580000
392059 5011 1 806011
− 𝐹𝑧𝑦𝑥𝑥 𝐹𝑥𝑦 + 𝐹𝑥𝑧𝑦𝑥𝑥 2 𝐹𝑧𝑧 − 𝑦 2 𝐹𝑥𝑦𝑦 − 𝐹𝑧𝑦𝑥𝑥 2 𝐹𝑦𝑧
17010000 302400 7200 𝑥𝑥 68040000
5011 875857 13 2 13
+ 𝑦 𝐹𝑦𝑧𝐹𝑥𝑥 + 𝐹𝑧𝑦𝑥𝑥 2 𝐹𝑥𝑧𝑧 − 𝑦 𝐹𝑦𝑦𝑦𝐹 − 𝑦 𝐹𝑥𝑦𝑦𝐹
907200 𝑥 34020000 360 𝑥 180 𝑥
10163 1193 1 1
+ 𝐹𝑦𝑦𝑥 𝑦𝑥𝑥 2 𝐹𝑧𝑧𝑧 + 𝑦 3 𝐹𝑧𝑦𝑦𝑦𝐹 − 𝑦𝑥 𝐹𝑦𝑦𝑦𝑦𝑥𝑥 2 − 𝑦 𝑦 3 𝐹𝑥𝑦𝑧𝑧𝑧
907200 121500 𝑥 7200 5400 𝑥 𝑥𝑥
1 1193 32703767
− 𝑦𝑥 2 𝑦𝑥𝑥 2 𝐹𝑥𝑦𝑦𝑧𝑧 + 𝑦 𝐹𝐹𝑥𝑥𝑦𝑧 + 𝐹𝑧𝑦𝑥 𝐹𝑦𝑧𝐹
3600 40500 𝑥 3061800000
1 1193 2 4169 21449
− 𝑦 3 𝐹𝑦𝑦𝑦𝑦𝑦𝑥𝑥 + 𝑦 𝐹𝐹𝑥𝑦𝑦𝑧 + 𝐹𝑧𝑤2 𝐹𝑦𝑦𝑥 + 𝐹𝑦𝑦𝑥 𝐹𝑧𝑧𝐹
10800 𝑥 40500 𝑥 34020000 1632960
4169 1 1 5011
+ 𝐹𝑧𝑤2 𝐹𝑥 − 𝑦 2 𝐹𝑥𝑥𝑥𝑦𝑦 + 𝑤4 𝐹𝑥 + 𝑦 3 𝐹𝑦𝑦𝐹𝑦𝑧
34020000 10800 𝑥 108000 907200 𝑥
5011 1 1 29
+ 𝐹𝑧𝑧2 𝑦𝑥𝑥 3 − 𝑦 4 𝐹𝑧𝑦𝑦𝑦𝑦𝑦𝑥𝑥 − 𝑦𝑥 3 𝑦𝑥𝑥 2 𝐹𝑧𝑧𝑦𝑦𝑦 − 𝐹𝑦𝑦𝑥 𝐹𝑥𝑦
907200 21600 𝑥 10800 720
5011 433 539 19
+ 𝑦𝑥 2 𝐹𝑦𝑧2 𝑦𝑥𝑥 − 𝑦 𝐹𝑥𝑦𝑧𝐹 − 𝑦𝑥𝑥 2 𝐹𝑧𝑦𝑧𝐹 − 𝐹𝑥𝐹𝑥𝑦
453600 10125 𝑥𝑥 81000 720
1 5011 1 433
− 𝑦 3 𝐹𝑥𝑥𝑦𝑦𝑦 + 𝐹𝑥𝑧𝐹𝑥𝑥 − 𝑦 4 𝐹𝑥𝑧𝑧𝑧𝑧 − 𝑦 𝐹𝑧𝑦𝑦𝑦𝑥𝑥 𝐹
10800 𝑥 907200 21600 𝑥𝑥 10125 𝑥
1 1511 1 1193
− 𝑦 5 𝐹𝑧𝑧𝑧𝑧𝑧 + 𝐹𝑤2 𝐹𝑥𝑧 − 𝐹𝑥𝑥𝑥𝑥𝑥 + 𝑦 𝑦 2 𝐹𝑧𝑧𝑧𝑦𝐹
108000 𝑥𝑥 7560000 108000 40500 𝑥 𝑥𝑥
13 1 1 1193 2
− 𝑦𝑥𝑥 𝐹𝑦𝑦𝐹 − 𝑦 𝐹𝑥𝑦𝑥𝑦𝑦𝑥𝑥 − 𝐹𝑦𝐹𝑥𝑥 + 𝑦 𝑦 𝐹𝑧𝑧𝑦𝑦𝐹
360 3600 𝑥 144 40500 𝑥 𝑥𝑥
16 Journal of Applied Mathematics

1193 1511 1 1 1
+ 𝑦 𝑦 𝐹𝑥𝑦𝑧𝑧𝐹 + 𝐹𝑤2 𝐹𝑧𝑧𝑦𝑥𝑥 − 𝑦 𝐹𝑥𝑦 − 𝑤2 𝐹𝑧𝑦𝑥𝑥 − 𝐹𝑥𝑥𝑥
20250 𝑥 𝑥𝑥 7560000 3600 𝑥𝑥 10800 10800
36460321 5011 1 1
+ 𝐹𝑧𝑦𝑥𝑥 𝐹𝑧𝑧𝐹 + 𝐹𝑧𝑧𝑦𝑥𝑥 𝐹𝑥𝑥 − 𝑦 𝐹𝑦𝑥𝑥 − 𝑤2 𝐹𝑦𝑦𝑥
1530900000 907200 3600 𝑥 10800
46577 46577 1
+ 𝐹𝑧2 𝐹𝑥 + 𝐹𝑧3 𝑦𝑥𝑥 − 𝑤2 𝐹𝑥) ℎ6 + 𝑂 (ℎ7 )
34992000 34992000 10800
1 1 (A.4)
− 𝐹2 𝐹𝑦𝑧 − 𝑦 𝑦 2 𝐹𝑥𝑥𝑦𝑧𝑧
10800 3600 𝑥 𝑥𝑥 󸀠 2087 1063
𝜏𝑛+1 = (− 𝑦𝑥𝑥 𝐹𝑦𝑧𝐹 + 𝑦 2 𝐹𝑧𝑦𝑦𝐹
5011 1 162000 162000 𝑥
+ 𝑦 2 𝐹𝑦𝑧𝐹𝑥𝑦 − 𝑦 𝐹𝑥𝑦𝑥𝑥
453600 𝑥 10800 𝑥𝑥 8689 971
− 𝐹𝑧𝐹𝑦𝑦𝑥𝑥 + 𝐹𝑦𝑦𝑥 𝐹𝑥𝑧
13 1 1512000 226800
− 𝐹𝐹𝑦𝑥𝑥 − 𝑦 5 𝐹𝑦𝑦𝑦𝑦𝑦
360 108000 𝑥 1811 1
+ 𝐹𝑧𝑦𝑥 𝐹𝑥𝑦 + 𝑦 2 𝑦 𝐹𝑥𝑦𝑦𝑧
1 1193 756000 1800 𝑥 𝑥𝑥
− 𝐹𝑦2 𝑦𝑥𝑥 + 𝐹𝑥𝑥𝑥𝑧𝐹
144 121500 1 1
+ 𝑦𝑥 2 𝑦𝑥𝑥 2 𝐹𝑧𝑧𝑦𝑦 + 𝑦 𝑦 𝐹𝑥𝑥𝑦𝑧
1 1 3600 1800 𝑥 𝑥𝑥
− 𝐹𝑥𝑥𝑥𝑥𝑧𝑦𝑥𝑥 − 𝑦 3 𝐹𝑧𝑦𝑦
21600 7200 𝑥𝑥 1 15143
+ 𝑦𝑥 𝑦𝑥𝑥 2 𝐹𝑥𝑦𝑧𝑧 + 𝐹𝑧𝑦𝑥𝑥 𝐹𝑥𝑧
1 10163 1800 2268000
+ 𝑤4 𝐹𝑦𝑦𝑥 + 𝐹𝑦𝑦𝑥 3 𝐹𝑧𝑦𝑦
108000 907200 24853 971
+ 𝐹𝑧𝑦𝑥𝑥 2 𝐹𝑧𝑧 + 𝑦 𝐹𝑦𝑧𝐹𝑥
1 1 4536000 226800 𝑥
− 𝑦𝑥 2 𝐹𝑦𝑦𝐹𝑦 − 𝑦 4 𝐹𝑧𝑧𝑧𝑦
30 10800 𝑥𝑥 971 1
+ 𝐹𝑧𝑧𝑦𝑥𝑥 𝐹𝑥 + 𝑦 𝑦 3 𝐹𝑧𝑧𝑧𝑦
4169 5011 226800 5400 𝑥 𝑥𝑥
+ 𝐹𝑧2 𝑤2 𝑦𝑥𝑥 + 𝐹𝑥𝑧2 𝑦𝑥𝑥
34020000 453600 1 1
+ 𝑦𝑥 3 𝐹𝑧𝑦𝑦𝑦𝑦𝑥𝑥 + 𝑦 𝐹𝑧𝑦𝑦𝑦𝑥𝑥 2
1 5011 5400 1800 𝑥
− 𝑦𝑥 2 𝐹𝑧𝑦𝑦𝑦𝑦𝑥𝑥 2 + 𝑦 2 𝐹𝑦𝑦𝐹𝑥𝑧
3600 907200 𝑥 971 1811
+ 𝐹𝑦𝑦𝑥 2 𝐹𝑦𝑧 + 𝐹𝑧𝑦𝑥 2 𝐹𝑦𝑦
10163 3551 226800 1512000
+ 𝑦 2 𝐹𝑧𝑧𝑧𝐹𝑥 + 𝐹𝑧𝐹𝑥𝑥𝑥) ℎ6
907200 𝑥𝑥 3189375 317 1
− 𝐹𝑧𝐹𝑤2 + 𝑦 2 𝐹𝑦𝑦𝑦𝑦𝑥𝑥
+ 𝑂 (ℎ )7 22680000 3600 𝑥
(A.3) 1063 1063
+ 𝐹𝑥𝑧𝑧𝑦𝑥𝑥 𝐹 + 𝑦 𝐹𝑥𝑦𝑧𝐹
81000 81000 𝑥
The principal local truncation errors for 𝑦, 𝑦󸀠 , and 𝑦󸀠󸀠 (i.e., 1 1063
󸀠
𝜏𝑛+1 , 𝜏𝑛+1 󸀠󸀠
, 𝜏𝑛+1 ) for TFMRKT5 are as follows: + 𝑦 𝐹𝑥𝑦𝑦𝑦𝑥𝑥 + 𝑦 2 𝐹𝑧𝑧𝑧𝐹
1800 𝑥 162000 𝑥𝑥
1 1 7 1063
𝜏𝑛+1 = (− 𝑦𝑥 3 𝐹𝑦𝑦𝑦 − 𝐹𝑧𝑥𝑥𝑦𝑥𝑥 − 𝑦𝑥 𝐹𝑦𝑦𝐹 + 𝑦 𝑦 𝐹𝑧𝑦𝑧𝐹
10800 3600 360 81000 𝑥 𝑥𝑥
1 1 971 15143
− 𝑦 𝐹𝑦𝑦𝑦𝑥𝑥 − 𝑦 2 𝐹𝑦𝑧 + 𝐹𝑦𝑦𝑥 𝐹𝑧𝑧𝑦𝑥𝑥 + 𝐹𝑧𝑦𝑥 𝐹𝑦𝑧𝑦𝑥𝑥
3600 𝑥 3600 𝑥𝑥 226800 2268000
527 1 1 1 1
+ 𝐹𝑧𝑧𝑦𝑥𝑥 𝐹 − 𝑦 𝐹𝑥𝑦𝑧𝑦𝑥𝑥 + 𝐹𝑥𝑥𝑥𝑥 − 𝑤4 𝐹 + 𝐹𝑥𝑥𝑥𝑧𝑦𝑥𝑥
162000 1800 𝑥 21600 21600 5400
1 527 1 1
− 𝑦𝑥 2 𝐹𝑧𝑦𝑦𝑦𝑥𝑥 + 𝑦 𝐹𝑦𝑧𝐹 + 𝑦 𝐹𝑦𝑥𝑥 + 𝑦 2 𝐹𝑥𝑦𝑧
3600 162000 𝑥 3600 𝑥𝑥 1800 𝑥𝑥
971 1 1 57889
+ 𝐹𝑧𝐹𝑦𝑦𝑥 − 𝐹𝑥𝑧𝑧𝑦𝑥𝑥 2 + 𝑦𝑥𝑥 2 𝐹𝑦𝑦 + 𝐹𝑧𝑧𝐹2
907200 3600 7200 14580000
1 1 1 1
− 𝑦 2 𝐹𝑥𝑦𝑦 − 𝑦 3 𝐹𝑧𝑧𝑧 + 𝑦 3 𝐹𝑧𝑦𝑧 + 𝑦 4 𝐹𝑦𝑦𝑦𝑦
3600 𝑥 10800 𝑥𝑥 3600 𝑥𝑥 21600 𝑥
971 971 1 1
+ 𝐹𝑧𝐹𝑥 + 𝐹𝑧2 𝑦𝑥𝑥 + 𝑦 3 𝐹𝑥𝑧𝑧𝑧 + 𝑦 4 𝐹𝑧𝑧𝑧𝑧
907200 907200 5400 𝑥𝑥 21600 𝑥𝑥
527 1 1 1 1 1
+ 𝐹𝑥𝑧𝐹 − 𝐹𝑦𝐹 − 𝑦 𝑦 2 𝐹𝑧𝑦𝑧 + 𝑦 3 𝐹𝑥𝑦𝑦𝑦 − 𝐹𝑦𝐹𝑥 + 𝑦 𝐹𝑥𝑦𝑥𝑥
162000 144 3600 𝑥 𝑥𝑥 5400 𝑥 144 5400 𝑥
Journal of Applied Mathematics 17

1 1 1717 1
+ 𝑦 2 𝐹𝑥𝑥𝑧𝑧 + 𝑦 2 𝐹𝑥𝑦𝑥𝑦 + 𝑦 2 𝐹𝑦𝑦𝐹𝑥𝑧 + 𝑤4 𝐹𝑧𝑦𝑥𝑥
3600 𝑥𝑥 3600 𝑥 302400 𝑥 108000
971 1811 1 1 9761
+ 𝐹𝑥𝑧𝐹𝑥 + 𝐹𝑧𝐹𝑥𝑥 − 𝐹𝑦2 𝑦𝑥 + 𝑤4 𝐹𝑦𝑦𝑥 + 𝐹𝑧𝑦𝑥 2 𝐹𝑥𝑦𝑦
226800 1512000 144 108000 2835000
31211 1063 29 10373
+ 𝐹𝑧2 𝐹 + 𝐹𝑧𝑥𝑥𝐹 − 𝐹𝑦𝑦𝑥 𝐹𝑥𝑦 + 𝑦 𝐹𝑥𝑦𝑧𝐹𝑥
22680000 162000 720 453600 𝑥
7 10373 298369
− 𝐹𝐹𝑥𝑦) ℎ6 + 𝑂 (ℎ7 ) + 𝐹𝑥𝑧𝑧𝑦𝑥𝑥 𝐹𝑥 + 𝐹𝑥𝑧𝑧𝑦𝑥𝑥 2 𝐹𝑧
360 453600 11340000
(A.5) 10373 856063
+ 𝑦 2 𝐹𝑧𝑧𝑧𝐹𝑥 + 𝑦 3 𝐹𝑧𝑧𝑧𝐹𝑧
󸀠󸀠 10373 907200 𝑥𝑥 68040000 𝑥𝑥
𝜏𝑛+1 =( 𝐹𝑦𝑦𝑥 𝑦𝑥𝑥 𝐹𝑥𝑧𝑧
453600 19 10373 2
− 𝑦 𝐹𝑦𝑦𝐹𝑥 + 𝑦 𝐹𝑧𝑦𝑦𝐹𝑥
10373 1717 720 𝑥 907200 𝑥
+ 𝐹𝑦𝑦𝑥 𝑦𝑥𝑥 2 𝐹𝑧𝑧𝑧 + 𝑦 𝐹𝑦𝑧𝑦𝑥𝑥 𝐹𝑥𝑧
907200 75600 𝑥 1717 10373
+ 𝑦𝑥 𝐹𝑥𝑧𝐹𝑥𝑦 + 𝐹𝑦𝑦𝑥 𝐹𝑧𝑥𝑥
10373 151200 907200
+ 𝑦 𝑦 𝐹𝑧𝑦𝑧𝐹𝑥
453600 𝑥 𝑥𝑥 9761 1
+ 𝐹𝑧𝑦𝑥 𝐹𝑦𝑥𝑥 − 𝑦 2 𝑦 𝐹𝑥𝑥𝑦𝑦𝑧
298369 2835000 3600 𝑥 𝑥𝑥
+ 𝑦 𝑦 2 𝐹𝑧𝑦𝑧𝐹𝑧
11340000 𝑥 𝑥𝑥 1 1
− 𝑦 2 𝑦 2 𝐹𝑥𝑦𝑦𝑧𝑧 − 𝑦 2 𝑦 3 𝐹𝑧𝑧𝑧𝑦𝑦
1717 3600 𝑥 𝑥𝑥 10800 𝑥 𝑥𝑥
+ 𝑦 𝐹𝑧𝑧𝑦𝑥𝑥 𝐹𝑥𝑦 1 1
151200 𝑥 − 𝑦 𝑦 𝐹𝑥𝑥𝑥𝑦𝑧 − 𝑦 𝑦 2 𝐹𝑥𝑥𝑦𝑧𝑧
5400 𝑥 𝑥𝑥 3600 𝑥 𝑥𝑥
1717 7811
+ 𝑦 𝐹𝑧𝑧𝑦𝑥𝑥 2 𝐹𝑦𝑧 + 𝐹𝑧2 𝐹𝑦𝑦𝑥 1 1717
100800 𝑥 5832000 − 𝑦𝑥 𝑦𝑥𝑥 3 𝐹𝑥𝑦𝑧𝑧𝑧 + 𝐹𝑥𝑧𝑦𝑥𝑥 2 𝐹𝑧𝑧
5400 100800
27487 67
− 𝑦 𝐹𝑥𝑦𝑧𝐹 − 𝐹𝑧𝑤2 𝐹𝑥 112471 1717
648000 𝑥𝑥 7560000 + 𝐹𝑧𝑦𝑥𝑥 𝐹𝑧𝑥𝑥 + 𝑦 𝐹𝑦𝑧𝐹𝑥𝑥
7560000 302400 𝑥
67 13 2
− 𝐹𝑧2 𝑤2 𝑦𝑥𝑥 − 𝑦 𝐹𝑦𝑦𝑦𝐹 1717 1
7560000 360 𝑥 + 𝐹𝑧𝑧𝑦𝑥𝑥 𝐹𝑥𝑥 − 𝑦 𝑦 4 𝐹𝑧𝑧𝑧𝑧𝑦
302400 21600 𝑥 𝑥𝑥
13 13 1 1
− 𝑦 𝐹𝑥𝑦𝑦𝐹 − 𝑦 𝐹𝑦𝑦𝐹 − 𝑦 3 𝑦 𝐹𝑥𝑦𝑦𝑦𝑧 − 𝑦 3 𝑦 2 𝐹𝑧𝑧𝑦𝑦𝑦
180 𝑥 360 𝑥𝑥 5400 𝑥 𝑥𝑥 10800 𝑥 𝑥𝑥
4087 61 1717 1717
− 𝑦 2 𝐹𝑧𝑦𝑧𝐹 − 𝐹𝑧𝐹𝑦𝐹 + 𝑦𝑥 2 𝐹𝑦𝑧𝐹𝑥𝑦 + 𝑦 2 𝐹𝑦𝑧2 𝑦𝑥𝑥
648000 𝑥𝑥 378000 151200 151200 𝑥
2483 130103 10373 1
− 𝐹𝑦𝑦𝑥𝑥 𝐹𝑥𝑧 − 𝐹𝑧𝑦𝑥𝑥 𝐹𝑥𝑦 + 𝐹𝑦𝑦𝑥 2 𝐹𝑥𝑦𝑧 − 𝑦 4 𝐹𝑧𝑦𝑦𝑦𝑦𝑦𝑥𝑥
302400 5670000 453600 21600 𝑥
19313 2 1717 1 1114379
+ 𝑦 𝐹𝐹𝑥𝑦𝑦𝑧 + 𝑦 3 𝐹𝑦𝑦𝐹𝑦𝑧 − 𝑦 𝐹𝑦𝑦𝑦𝑦𝑥𝑥 2 + 𝑦 𝐹𝑧𝑧𝑧𝐹2
648000 𝑥 302400 𝑥 7200 𝑥 58320000 𝑥𝑥
9761 1 1 1
+ 𝐹𝑧𝑦𝑥 3 𝐹𝑦𝑦𝑦 − 𝐹𝑦𝑦𝑥 2 𝐹𝑦𝑦 − 𝑦 𝑦 2 𝐹𝑥𝑦𝑦𝑧 − 𝑦 𝑦 3 𝐹𝑧𝑧𝑦𝑦
8505000 30 1800 𝑥 𝑥𝑥 3600 𝑥 𝑥𝑥
10373 3 1 1 1114379
+ 𝑦 𝐹𝑧𝑦𝑦𝐹𝑦 − 𝑦 2 𝐹𝑥𝑦𝑦𝑦𝑦𝑥𝑥 − 𝑦𝑥 2 𝐹𝑧𝑦𝑦𝑦𝑦𝑥𝑥 2 + 𝑦 𝐹2 𝐹𝑧𝑦𝑧
907200 𝑥 3600 𝑥 3600 58320000 𝑥
19313 10959589 383 87029
+ 𝑦𝑥 𝐹𝐹𝑥𝑥𝑦𝑧 + 𝐹𝑧𝐹𝐹𝑥𝑧 − 𝐹𝑦𝑦𝑥𝑥 2 𝐹𝑧𝑧 − 𝐹𝑧𝑦𝑥𝑥 2 𝐹𝑦𝑧
648000 1020600000 302400 7560000
13567 337511 1 1
− 𝑦 𝐹𝑦𝑧𝐹𝑥 + 𝐹𝑧𝑧𝐹𝐹𝑥 − 𝐹𝑥𝑥𝑥𝑥𝑥 + 𝑤4 𝐹𝑥
907200 𝑥𝑥 25515000 108000 108000
19313 1 1 1114379
+ 𝑦 3 𝐹𝑧𝑦𝑦𝑦𝐹 − 𝑦 3 𝐹𝑦𝑦𝑦𝑦𝑦𝑥𝑥 − 𝑦 5 𝐹𝑦𝑦𝑦𝑦𝑦 + 𝐹𝑥𝑧𝑧𝐹2
1944000 𝑥 10800 𝑥 108000 𝑥 58320000
19313 19313 1 11 2
+ 𝑦 2 𝐹𝑥𝑧𝑧𝑧𝐹 + 𝑦 3 𝐹𝑧𝑧𝑧𝑧𝐹 − 𝑦𝑥𝑥 2 𝐹𝑥𝑦𝑦 + 𝐹 𝐹𝑦𝑧
648000 𝑥𝑥 1944000 𝑥𝑥 7200 43200
19313 1 1 1
+ 𝑦𝑥𝑥 𝐹𝑥𝑥𝑧𝑧𝐹 − 𝑦 𝐹𝑥𝑦𝑥𝑦𝑦𝑥𝑥 − 𝑦𝑥𝑥 2 𝐹𝑥𝑥𝑦𝑧 − 𝑦 3 𝐹𝑥𝑦𝑧𝑧
648000 3600 𝑥 3600 3600 𝑥𝑥
18 Journal of Applied Mathematics

1 1
− 𝑦 4 𝐹𝑧𝑧𝑧𝑦 − 𝑦 3 𝐹𝑧𝑦𝑦 Conflicts of Interest
10800 𝑥𝑥 7200 𝑥𝑥
The authors declare that there are no conflicts of interest
13 1
− 𝐹𝐹𝑦𝑥𝑥 − 𝐹𝑥𝑥𝑥𝑥𝑧𝑦𝑥𝑥 regarding the publication of this paper.
360 21600
1 1 References
− 𝑦 𝐹𝑥𝑦𝑥𝑥 − 𝐹𝑦2 𝑦𝑥𝑥
10800 𝑥𝑥 144
[1] B. Paternoster, “Runge-Kutta(-Nyström) methods for ODEs
1717 1717 with periodic solutions based on trigonometric polynomials,”
+ 𝐹𝑥𝑧2 𝑦𝑥𝑥 + 𝐹𝑧𝑧2 𝑦𝑥𝑥 3
151200 302400 Applied Numerical Mathematics, vol. 28, no. 2–4, pp. 401–412,
1998.
1 1
− 𝑦 4 𝐹𝑥𝑦𝑦𝑦𝑦 − 𝑦 4 𝐹𝑥𝑧𝑧𝑧𝑧 [2] G. Vanden Berghe, H. De Meyer, M. Van Daele, and T. Van
21600 𝑥 21600 𝑥𝑥 Hecke, “Exponentially fitted Runge-Kutta methods,” Journal of
1 19 Computational and Applied Mathematics, vol. 125, no. 1-2, pp.
− 𝑦𝑥𝑥 5 𝐹𝑧𝑧𝑧𝑧𝑧 − 𝐹𝑥𝐹𝑥𝑦 107–115, 2000.
108000 720
[3] T. E. Simos, “Exponentially fitted Runge-Kutta methods for the
1 1717 numerical solution of the Schrödinger equation and related
− 𝐹𝑦𝐹𝑥𝑥 + 𝐹𝑥𝑥𝐹𝑥𝑧
144 302400 problems,” Computational Materials Science, vol. 18, no. 3-4, pp.
315–332, 2000.
9761 7811
+ 𝐹𝑧𝐹𝑥𝑥𝑥 + 𝐹𝑧2 𝐹𝑥 [4] Z. Kalogiratou and T. E. Simos, “Construction of trigonometri-
8505000 5832000 cally and exponentially fitted Runge-Kutta-Nyström methods
7811 1 for the numerical solution of the schrödinger equation and
+ 𝐹𝑧3 𝑦𝑥𝑥 − 𝑦 3 𝐹𝑥𝑥𝑧𝑧𝑧 related problems—a method of 8th algebraic order,” Journal of
5832000 10800 𝑥𝑥 Mathematical Chemistry, vol. 31, no. 2, pp. 211–232, 2002.
1 1 [5] Z. Kalogiratou, T. Monovasilis, and T. E. Simos, “Compu-
− 𝑦𝑥 3 𝐹𝑥𝑥𝑦𝑦𝑦 − 𝑦 𝐹𝑥𝑥𝑥𝑥𝑦
10800 21600 𝑥 tation of the eigenvalues of the Schrödinger equation by
exponentially-fitted Runge-Kutta-Nyström methods,” Com-
1 1
− 𝑦𝑥𝑥 2 𝐹𝑥𝑥𝑥𝑧𝑧 − 𝑦 2 𝐹𝑥𝑥𝑥𝑦𝑦 puter Physics Communications, vol. 180, no. 2, pp. 167–176, 2009.
10800 10800 𝑥 [6] T. E. Simos, “Exponentially-fitted Runge-Kutta-Nyström meth-
10373 19313 od for the numerical solution of initial-value problems with
+ 𝐹𝑥𝐹𝑧𝑥𝑥 + 𝐹𝑥𝑥𝑥𝑧𝐹 oscillating solutions,” Applied Mathematics Letters, vol. 15, no.
907200 1944000 2, pp. 217–225, 2002.
1331 112471 [7] D. P. Sakas and T. E. Simos, “A fifth algebraic order trigono-
− 𝐹𝑥𝑧𝐹𝑤2 + 𝐹𝑧𝑦𝑥 𝑦𝑥𝑥 𝐹𝑥𝑦𝑧
45360000 3780000 metrically-fitted modified Runge-Kutta Zonneveld method for
the numerical solution of orbital problems,” Mathematical and
1717
+ 𝑦 2 𝐹𝑦𝑦𝐹𝑧𝑧𝑦𝑥𝑥 Computer Modelling, vol. 42, no. 7-8, pp. 903–920, 2005.
302400 𝑥 [8] H. Van De Vyver, “A Runge-Kutta-Nyström pair for the numer-
112471 ical integration of perturbed oscillators,” Computer Physics
+ 𝐹𝑧𝑦𝑥 2 𝐹𝑧𝑦𝑦𝑦𝑥𝑥 Communications, vol. 167, no. 2, pp. 129–142, 2005.
7560000
[9] H. Yang and X. Wu, “Trigonometrically-fitted ARKN methods
10373 27487
+ 𝐹𝑦𝑦𝑥 2 𝐹𝑧𝑦𝑧𝑦𝑥𝑥 − 𝑦 𝐹𝑧𝑦𝑦𝐹𝑦𝑥𝑥 for perturbed oscillators,” Applied Numerical Mathematics, vol.
453600 648000 𝑥 58, no. 9, pp. 1375–1395, 2008.
[10] M. A. Demba, N. Senu, and F. Ismail, “Trigonometrically-
19313 2
+ 𝑦 𝑦 𝐹𝑧𝑧𝑦𝑦𝐹 fitted explicit four-stage fourth-order Runge–Kutta–Nyström
648000 𝑥 𝑥𝑥 method for the solution of initial value problems with oscilla-
19313 905927 tory behavior,” The Global Journal of Pure and Applied Mathe-
+ 𝑦𝑥 𝑦𝑥𝑥 𝐹𝑥𝑦𝑧𝑧𝐹 + 𝐹𝑧𝑦𝑥𝑥 𝐹𝑧𝑧𝐹 matics (GJPAM), vol. 12, no. 1, pp. 67–80, 2016.
324000 37800000
[11] M. Hanan, “Oscillation criteria for third-order linear differen-
19313 337511
+ 𝑦𝑥 𝑦𝑥𝑥 2 𝐹𝑧𝑧𝑧𝑦𝐹 + 𝐹𝑦𝑦𝑥 𝐹𝑧𝑧𝐹 tial equations,” Pacific Journal of Mathematics, vol. 11, pp. 919–
648000 25515000 944, 1961.
10959589 2627 [12] J. Rovder, “Oscillation criteria for third-order linear differential
+ 𝐹𝑧𝑦𝑥 𝐹𝑦𝑧𝐹 − 𝐹𝑦𝑦𝑥 𝐹𝑦𝑧𝑦𝑥𝑥 equations,” Matematický časopis, vol. 25, no. 3, pp. 231–244, 1975.
1020600000 113400
[13] A. C. Lazer, “The behavior of solutions of the differential
130103 1331 󸀠󸀠󸀠 󸀠

− 𝐹𝑧𝑦𝑥 𝐹𝑦𝑦𝑦𝑥𝑥 − 𝑦 𝐹𝑧𝑧𝐹𝑤2 equation y + p(x)y +q(x)y = 0,” Pacific Journal of Mathematics,
5670000 45360000 𝑥𝑥 vol. 17, pp. 435–466, 1966.
1331 67 [14] G. D. Jones, “Properties of solutions of a class of third-order
− 𝑦𝑥 𝐹𝑦𝑧𝐹𝑤2 − 𝐹𝑧𝑤2 𝐹𝑦𝑦𝑥 ) differential equations,” Journal of Mathematical Analysis and
45360000 7560000 Applications, vol. 48, pp. 165–169, 1974.
⋅ ℎ6 + 𝑂 (ℎ7 ) [15] F. A. Fawzi, N. Senu, F. Ismail, and Z. A. Majid, “An efficient
of direct integrator of Runge-Kutta Type Method for Solving
(A.6) y”’=f(x, y, y’) with Application to Thin Film Flow Problem,”
Journal of Applied Mathematics 19

International Journal of Pure and Applied Mathematics, vol. 117,

no. 4, accepted.
[16] J. C. Butcher, Numerical Methods for Ordinary Differential
Equations, John Wiley & Sons, New York, NY, USA, 2nd edition,
2008.
[17] J. D. Lambert, Numerical Methods for Ordinary Differential
Systems. The Initial Value Problem, John Wiley & Sons, New
York, NY, USA, 1993.
[18] Z. A. Anastassi and T. E. Simos, “Trigonometrically fit-
ted Runge-Kutta methods for the numerical solution of the
Schrödinger equation,” Journal of Mathematical Chemistry, vol.
37, no. 3, pp. 281–293, 2005.
[19] E. Momoniat and F. M. Mahomed, “Symmetry reduction and
numerical solution of a third-order ODE from thin film flow,”
Mathematical & Computational Applications, vol. 15, no. 4, pp.
709–719, 2010.
[20] E. O. Tuck and L. W. Schwartz, “A numerical and asymptotic
study of some third-order ordinary differential equations rele-
vant to draining and coating flows,” SIAM Review. A Publication
of the Society for Industrial and Applied Mathematics, vol. 32, no.
3, pp. 453–469, 1990.
[21] J. Biazar, E. Babolian, and R. Islam, “Solution of the system
of ordinary differential equations by Adomian decomposition
method,” Applied Mathematics and Computation, vol. 147, no. 3,
pp. 713–719, 2004.
[22] M. Mechee, N. Senu, F. Ismail, B. Nikouravan, and Z. Siri, “A
three-stage fifth-order Runge-Kutta method for directly solving
special third-order differential equation with application to thin
film flow problem,” Mathematical Problems in Engineering, vol.
2013, Article ID 795397, 7 pages, 2013.
Advances in Advances in Journal of The Scientific Journal of
Operations Research
Hindawi
Decision Sciences
Hindawi
Applied Mathematics
Hindawi
World Journal
Hindawi Publishing Corporation
Probability and Statistics
Hindawi
www.hindawi.com Volume 2018 www.hindawi.com Volume 2018 www.hindawi.com Volume 2018 http://www.hindawi.com
www.hindawi.com Volume 2018
2013 www.hindawi.com Volume 2018

International
Journal of
Mathematics and
Mathematical
Sciences

Journal of

Hindawi
Optimization
Hindawi
www.hindawi.com Volume 2018 www.hindawi.com Volume 2018

Submit your manuscripts at

www.hindawi.com

International Journal of
Engineering International Journal of
Mathematics
Hindawi
Analysis
Hindawi
www.hindawi.com Volume 2018 www.hindawi.com Volume 2018

Journal of Advances in Mathematical Problems International Journal of Discrete Dynamics in

Complex Analysis
Hindawi
Numerical Analysis
Hindawi
in Engineering
Hindawi
Differential Equations
Hindawi
Nature and Society
Hindawi
www.hindawi.com Volume 2018 www.hindawi.com Volume 2018 www.hindawi.com Volume 2018 www.hindawi.com Volume 2018 www.hindawi.com Volume 2018

International Journal of Journal of Journal of Abstract and Advances in

Stochastic Analysis
Hindawi
Mathematics
Hindawi
Function Spaces
Hindawi
Applied Analysis
Hindawi
Mathematical Physics
Hindawi
www.hindawi.com Volume 2018 www.hindawi.com Volume 2018 www.hindawi.com Volume 2018 www.hindawi.com Volume 2018 www.hindawi.com Volume 2018