Applied Mathematics and Computation 191 (2007) 334–346 www.elsevier.com/locate/amc
Application of homotopy perturbation method for solving eighth-order boundary value problems A. Golbabai, M. Javidi
*
Department of Mathematics, Iran University of Science and Technology, Narmak, Tehran 16844, Iran
Abstract Homotopy perturbation method is applied to the numerical solution for solving eighth-order boundary value problems. Comparison of the result obtained by the present method with that obtained by modified decomposition method [M. Mesrovic, The modified decomposition method for eighth-order boundary value problems, Appl. Math. Comput. (2006), doi:10.1016/j.amc.2006.11.015] reveals that the present method is very effective and convenient. Ó 2007 Elsevier Inc. All rights reserved. Keywords: Eighth-order boundary value problems; Homotopy perturbation method
1. Introduction A new perturbation method called homotopy perturbation method (HPM) was proposed by He in 1997 and systematical description in 2000 which is, in fact, a coupling of the traditional perturbation method and homotopy in topology [1]. This new method was further developed and improved by He and applied to nonlinear oscillators with discontinuities [2], nonlinear wave equations [3], asymptotology [4], boundary value problem [5], Limit cycle and bifurcation of nonlinear problems [6] and many other subjects. Thus He’s method is a universal one which can solve various kinds of nonlinear equations. For example, it was applied to the quadratic Ricatti differential equation by Abbasbandy [7]; to the axisymmetric flow over a stretching sheet by Ariel et al. [8]; to the nonlinear systems of reaction–diffusion equations by Ganji and Sadighi [9]; to the Helmholtz equation and fifth-order KdV equation by Rafei and Ganji [10]; for the thin film flow of a fourth grade fluid down a vertical cylinder by Siddiqui et al. [11]; to the nonlinear Voltra–Fredholm integral equations by Ghasemi et al. [12]. Recently various powerful mathematical methods such as variational iteration method [13–21], Exp-function method [22,23], F-expansion method [24], Adomian decomposition method [25] and others [26,27] have been proposed to obtain exact and approximate analytic solutions for linear and nonlinear problems.
*
Corresponding author. E-mail address:
[email protected] (M. Javidi).
0096-3003/$ - see front matter Ó 2007 Elsevier Inc. All rights reserved. doi:10.1016/j.amc.2007.02.091
A. Golbabai, M. Javidi / Applied Mathematics and Computation 191 (2007) 334–346
335
In this paper, we consider the general eighth-order boundary value problems of the type: y ðviiiÞ ðxÞ ¼ f ðx; y; y 0 ; y 00 ; y 000 ; y ðivÞ ; y ðvÞ ; y ðviÞ ; y ðviiÞ Þ;
a < x < b;
ð1Þ
with boundary conditions yðaÞ ¼ a0 ; y ðivÞ ðaÞ ¼ a4 ;
y 0 ðaÞ ¼ a1 ;
y 00 ðaÞ ¼ a2 ;
y ðvÞ ðaÞ ¼ a5 ;
y 000 ðaÞ ¼ a3 ;
yðbÞ ¼ b0 ;
y 0 ðbÞ ¼ b1 ;
ð2Þ
where f is continuous function on [a,b] and the parameters ai ; i ¼ 0; 1; . . . ; 6 and bi ; i ¼ 0; 1 are real constants. Such type of boundary value problems arise in the mathematical modeling of the viscoelastic flows and other branches of mathematical, physical and engineering sciences, see [28–31] and references therein. Several numerical methods including spectral Galerkin and collocation [29,30], sixth B-spline method [28], decomposition method [32], spline collocation approximation [33], Chow–Yorke algorithm [34] and others [35,36] have been developed for solving the problem of type (1). This paper, applies the homotopy perturbation method [1– 12] to the discussed problem. 2. Homotopy perturbation method Using the transformation dy d2 y d3 y ¼ y2; ¼ y ; ¼ y4; 3 dx dx2 dx3 ð3Þ d4 y d5 y d6 y d7 y ¼ y5; ¼ y6; ¼ y7; ¼ y8; dx4 dx5 dx6 dx7 we can rewrite the eighth-order boundary value problem (1), (2) as the system of ordinary differential equations: 8 dy 1 ¼ y2; > > dx > > > dy 2 > ¼ y3; > dx > > > dy 3 > > ¼ y4; > dx > > > < dy 4 ¼ y ; 5 dx ð4Þ dy 5 > ¼ y > 6; dx > > > dy 6 > > ¼ y7; > dx > > > dy 7 > > ¼ y8; > dx > > : dy 8 ¼ f ðx; y 1 ; y 2 ; y 3 ; y 4 ; y 5 ; y 6 ; y 7 Þ; dx y 1 ¼ y;
with the boundary conditions y 1 ðaÞ ¼ a0 ;
y 2 ðaÞ ¼ a1 ;
y 3 ðaÞ ¼ a2 ;
y 4 ðaÞ ¼ a3 ;
y 5 ðaÞ ¼ a4 ;
y 6 ðaÞ ¼ a5 ;
y 1 ðbÞ ¼ b0 ;
y 2 ðbÞ ¼ b1 ;
which can be written as a system of integral equations: 8 Rx y 1 ¼ a0 þ 0 y 2 ðtÞ dt; > > > Rx > > > > y 2 ¼ a1 þ R0 y 3 ðtÞ dt; > > > > y 3 ¼ a2 þ 0x y 4 ðtÞ dt; > > > < y ¼ a þ R x y ðtÞ dt; 3 4 R0x 5 > y ¼ a þ y ðtÞ dt; 4 > 5 > R0x 6 > > > y ¼ a5 þ 0 y 7 ðtÞ dt; > > Rx > 6 > > > > y 7 ¼ A þ R0 y 8 ðtÞ dt; > x : y 8 ¼ B þ 0 f ðt; y 1 ðtÞ; y 2 ðtÞ; y 3 ðtÞ; y 4 ðtÞ; y 5 ðtÞ; y 6 ðtÞ; y 7 ðtÞÞ dt;
ð5Þ
ð6Þ
336
A. Golbabai, M. Javidi / Applied Mathematics and Computation 191 (2007) 334–346
where y 7 ðaÞ ¼ A;
y 8 ðaÞ ¼ B:
To explain HPM, we consider (6) as 2 3 L1 ðy 1 ; y 2 ; . . . ; y 8 Þ 6 7 .. 7 ¼ 0; Lðy 1 ; y 2 ; . . . ; y 8 Þ ¼ 6 . 4 5 L8 ðy 1 ; y 2 ; . . . ; y 8 Þ
ð7Þ
with solution ðf1 ; f2 ; . . . ; f8 Þ where Rx 8 L1 ðy 1 ; y 2 ; . . . ; y 8 Þ ¼ y 1 a0 0 y 2 ðtÞ dt; > > < .. . > > Rx : L8 ðy 1 ; y 2 ; . . . ; y 8 Þ ¼ y 8 B 0 f ðt; y 1 ðtÞ; y 2 ðtÞ; y 3 ðtÞ; y 4 ðtÞ; y 5 ðtÞ; y 6 ðtÞ; y 7 ðtÞÞ dt:
ð8Þ
We can define homotopy Hðy 1 ; y 2 ; . . . ; y 8 ; pÞ by Hðy 1 ; y 2 ; . . . ; y 8 ; 0Þ ¼ Fðy 1 ; y 2 ; . . . ; y 8 Þ;
Hðy 1 ; y 2 ; . . . ; y 8 ; 1Þ ¼ Lðy 1 ; y 2 ; . . . ; y 8 Þ;
where T
Fðy 1 ; y 2 ; . . . ; y 8 Þ ¼ ½F 1 ðy 1 ; y 2 ; . . . ; y 8 Þ; . . . ; F 8 ðy 1 ; y 2 ; . . . ; y 8 Þ T
ð9Þ
¼ ½y 1 a0 ; . . . ; y 7 A; y 8 B ; T
Hðy 1 ; y 2 ; . . . ; y 8 ; pÞ ¼ ½H 1 ðy 1 ; y 2 ; . . . ; y 8 ; pÞ; . . . ; H 8 ðy 1 ; y 2 ; . . . ; y 8 ; pÞ : Typically we may choose a convex homotopy by Hðy 1 ; y 2 ; . . . ; y 8 ; pÞ ¼ ð1 pÞFðy 1 ; y 2 ; . . . ; y 8 Þ þ pLðy 1 ; y 2 ; . . . ; y 8 Þ ¼ 0:
ð10Þ
The convex homotopy (10) continuously trace an implicitly defined curve from a starting point Hðy 1 a0 ; . . . ; y 7 A; y 8 B; 0Þ ¼ 0 to a solution function Hðy 1 ; . . . ; y 7 ; y 8 ; 1Þ ¼ 0. The embedding parameter p monotonically increases from zero to unit as trivial problem Fðy 1 ; y 2 ; . . . ; y 8 Þ ¼ 0 is continuously deformed to original problem Lðy 1 ; y 2 ; . . . ; y 8 Þ ¼ 0. The HPM uses the homotopy parameter p as an expanding parameter to obtain [16]: y 1 ¼ y 10 þ py 11 þ p2 y 12 þ ; .. .
ð11Þ
y 8 ¼ y 80 þ py 81 þ p2 y 82 þ ; when p ! 1, (8) corresponds to f1 ¼ lim y 1 ¼ y 10 þ y 11 þ y 12 þ ; p!1
.. .
ð12Þ
f8 ¼ lim y 8 ¼ y 80 þ y 81 þ y 82 þ : p!1
For the application of HPM to (6) we can write (10) as follows: 8 H 1 ðy 1 ; y 2 ; . . . ; y 8 ; pÞ ¼ ð1 pÞF 1 ðy 1 ; y 2 ; . . . ; y 8 Þ þ pL1 ðy 1 ; y 2 ; . . . ; y 8 Þ ¼ 0; > > < .. . > > : H 8 ðy 1 ; y 2 ; . . . ; y 8 ; pÞ ¼ ð1 pÞF 8 ðy 1 ; y 2 ; . . . ; y 8 Þ þ pL8 ðy 1 ; y 2 ; . . . ; y 8 Þ ¼ 0:
ð13Þ
A. Golbabai, M. Javidi / Applied Mathematics and Computation 191 (2007) 334–346
Substitution of (8), (9) and (11) into (13) yields Rx 8 ð1 pÞðy 1 a0 Þ þ p y 1 a0 0 y 2 ðtÞ dt ¼ 0; > > < .. . > > Rx : ð1 pÞðy 8 BÞ þ p y 8 B 0 f ðt; y 1 ðtÞ; y 2 ðtÞ; y 3 ðtÞ; y 4 ðtÞ; y 5 ðtÞ; y 6 ðtÞ; y 7 ðtÞÞ dt ¼ 0:
337
ð14Þ
By equating the terms with identical powers of p, we have 8 y 10 > > > > > . > > y 70 > > : y 70 8 y 11 > > > > > > < .. . p1 : > > > > y 71 > > : y 81 8 > y > > 12 > > > . > < .. p2 : > > y 72 > > > > > :y 81
¼ a0 ;
ð15Þ ¼ A; ¼ B ! y 80 ¼ B Rx ¼ 0 y 20 ðtÞ dt;
¼ ¼
ð16Þ
Rx
y ðtÞ dt; 0 80 Rx d f ðt; y 1 ðtÞ; y 2 ðtÞ; y 3 ðtÞ; y 4 ðtÞ; y 5 ðtÞ; y 6 ðtÞ; y 7 ðtÞÞjp¼0 dt; 0 dp
¼
Rx
¼
Rx
0
0
¼ 2!1
y 21 ðtÞ dt;
ð17Þ
y 81 ðtÞ dt; R x d2 Rx f ðt; y 1 ðtÞ; y 2 ðtÞ; y 3 ðtÞ; y 4 ðtÞ; y 5 ðtÞ; y 6 ðtÞ; y 7 ðtÞÞjp¼0 dt ! y 82 ¼ 2!1 0 0 dp2
d2 dp2
f ðt; y 1 ðtÞ; y 2 ðtÞ; y 3 ðtÞ; y 4 ðtÞ; y 5 ðtÞ; y 6 ðtÞ; y 7 ðtÞÞjp¼0 dt;
.. . Rx 8 y 1;n ¼ 0 y 2;n1 ðtÞ dt; > > > > > > < .. . n p : Rx > > y 7;n ¼ 0 y 8;n1 ðtÞ dt; > > > > R x dn : y 8;n ¼ n!1 0 dp n f ðt; y 1 ðtÞ; y 2 ðtÞ; y 3 ðtÞ; y 4 ðtÞ; y 5 ðtÞ; y 6 ðtÞ; y 7 ðtÞÞjp¼0 dt:
ð18Þ
Combining all the terms: (15)–(18) gives the solution of the problem. Using the boundary conditions y 1 ðbÞ ¼ b0 and y 2 ðbÞ ¼ b1 we can obtained A and B. 3. Applications In this section, in order to verify numerically whether the proposed methodology leads to higher accuracy, we evaluate the numerical solution of the problem (1). To show the efficiency of the present method for our problem in comparison with the exact solution we report absolute error which is defined by Ey N ðxÞ ¼ jy Exact ðxÞ y N ðxÞj; P where y N ðxÞ ¼ Nm¼0 y 1m for N = 0, 1, 2, . . .. Example 1. Consider the following linear eighth-order problem [32] y ðviiiÞ ðxÞ ¼ 8ex þ yðxÞ;
ð19Þ
0 < x < 1;
with the following boundary conditions: yð0Þ ¼ 1;
y 0 ð0Þ ¼ 0;
y ðivÞ ð0Þ ¼ 3;
y 00 ð0Þ ¼ 1;
y ðvÞ ð0Þ ¼ 4;
y 000 ð0Þ ¼ 2;
y ðviÞ ð0Þ ¼ 5;
y ðviiÞ ð0Þ ¼ 6:
ð20Þ
338
A. Golbabai, M. Javidi / Applied Mathematics and Computation 191 (2007) 334–346
The analytic solution is yðxÞ ¼ ð1 xÞex : Using the transformation (3) we can rewrite the eighth-order boundary value problem (19) as the system integral equations Rx 8 y 1 ¼ 1 þ 0 y 2 ðtÞ dt; > > > Rx > > > y 2 ¼ 0 y 3 ðtÞ dt; > > > Rx > > > y 3 ¼ 1 þ 0 y 4 ðtÞ dt; > > > > R > < y 4 ¼ 2 þ x y 5 ðtÞ dt; 0 ð21Þ Rx > y ¼ 3 þ y 6 ðtÞ dt; > 5 > 0 > > Rx > > > y 6 ¼ 4 þ 0 y 7 ðtÞ dt; > > > Rx > > > y 7 ¼ 5 þ 0 y 8 ðtÞ dt; > > > Rx : y 8 ¼ 6 þ 0 ð8et þ y 1 ðtÞÞ dt: Using (14) for (21) we have Rx 8 y 10 þ py 11 þ p2 y 12 þ ¼ 1 þ p 0 ðy 20 þ py 21 þ p2 y 22 þ Þ dt; > > > Rx > > > < y 20 þ py 21 þ p2 y 22 þ ¼ p 0 ðy 30 þ py 31 þ p2 y 32 þ Þ dt; .. > > > . > > > Rx : y 80 þ py 81 þ p2 y 82 þ ¼ 6 þ p 0 ð8et þ y 10 þ py 11 þ p2 y 12 þ Þ dt: Comparing the coefficient of like powers of p, we have: 8 y 10 ¼ 1; > > > > > > > > y 20 ¼ 0; > > > > y 30 ¼ 1; > > > > > < y 40 ¼ 2; p0 : > y 50 ¼ 3; > > > > > > > y 60 ¼ 4; > > > > > > y 70 ¼ 5; > > > : y 80 ¼ 6; 8 y 11 ¼ 0; > > > > > > y 21 ¼ x; > > > > > > y 31 ¼ 2x; > > > > > < y 41 ¼ 3x; p1 : > > > y 51 ¼ 4x; > > > > > y 61 ¼ 5x; > > > > > > y 71 ¼ 6x; > > > : y 81 ¼ 8ex þ x þ 8;
ð22Þ
ð23Þ
ð24Þ
A. Golbabai, M. Javidi / Applied Mathematics and Computation 191 (2007) 334–346
8 > y > > 12 > > > > y 22 > > > > > > y 32 > > > > > y 52 > > > > > y 62 > > > > > > y 72 > > > > : y 82
339
¼ 12 x2 ; ¼ x2 ; ¼ 32 x2 ; ¼ 2x2 ;
ð25Þ
¼ 52 x2 ; ¼ 3x2 ; ¼ 8ex þ 12 x2 þ 8x þ 8; ¼ 0;
.. .
p3 :
8 y 13 > > > > > > y 23 > > > > > > y 33 > > > > > < y 43 > y 53 > > > > > > y 63 > > > > > > y 73 > > > > : y 83
¼ 13 x3 ; ¼ 12 x3 ; ¼ 23 x3 ; ¼ 56 x3 ;
ð26Þ
¼ x3 ; ¼ 8ex þ 16 x3 þ 4x2 þ 8x þ 8; ¼ 0; ¼ 16 x3 :
Combining all the terms: (23)–(26) gives 7 5 1 1 6 1 7 1 y 8 ðxÞ ¼ 9 þ x2 þ x3 þ x4 þ x5 þ x þ x 8ex þ x8 þ 8x: 2 24 30 240 2520 40320 The numerical results obtained in Table 1. In Table 1, we list the results obtained by homotopy perturbation method and compared with modified decomposition method (MDM) results given in [32] at x = 0.25(0.25)1. As we see from this Table, it is clear that the result obtained by the present method is very superior to that obtained by MDM method. As can be seen from Table 1, the error decreased when the integer N is increased until N = 8. Example 2. Consider the following nonlinear eighth-order problem [32]: y ðviiiÞ ðxÞ ¼ ex y 2 ðxÞ;
ð27Þ
0 < x < 1;
with the following boundary conditions: y ðiÞ ð0Þ ¼ 1;
i ¼ 0; 1; . . . ; 7:
ð28Þ
Table 1 Numerical results for Example 1 x
0.25 0.50 0.75 1.00
Error HPM, N = 2
HPM, N = 4
HPM, N = 8
MDM
0.0057 0.0506 0.1895 0.5000
3.4323e5 0.0012 0.0093 0.0417
2.7611e13 2.9519e10 1.7849e8 3.3265e7
1.0e7 3.1e6 5.5e5 4.2e4
340
A. Golbabai, M. Javidi / Applied Mathematics and Computation 191 (2007) 334–346
The analytic solution is yðxÞ ¼ ex : Using the transformation (3) we can rewrite the eighth-order boundary value problem (27) as the system integral equations 8 Rx y 1 ¼ 1 þ 0 y 2 ðtÞ dt; > > > Rx > > > y 2 ¼ 1 þ 0 y 3 ðtÞ dt; > > Rx > > > y 3 ¼ 1 þ 0 y 4 ðtÞ dt; > > Rx > < y 4 ¼ 1 þ 0 y 5 ðtÞ dt; Rx ð29Þ > y 5 ¼ 1 þ 0 y 6 ðtÞ dt; > > Rx > > > y 6 ¼ 1 þ 0 y 7 ðtÞ dt; > > Rx > > > y 7 ¼ 1 þ 0 y 8 ðtÞ dt; > > > Rx : y 8 ¼ 1 þ 0 ðet y 21 ðtÞÞ dt: Using (14) for (29) we have 8 Rx y 10 þ py 11 þ p2 y 12 þ ¼ 1 þ p 0 ðy 20 þ py 21 þ p2 y 22 þ Þ dt; > > > Rx > > < y 20 þ py 21 þ p2 y 22 þ ¼ 1 þ p 0 ðy 30 þ py 31 þ p2 y 32 þ Þ dt; .. > > . > > > Rx : y 80 þ py 81 þ p2 y 82 þ ¼ 1 þ p 0 et ðy 10 þ py 11 þ p2 y 12 þ Þ2 dt:
ð30Þ
Comparing the coefficient of like powers of p, we have: p0 : y i;0 ¼ 1; i ¼ 1; 2; . . . ; 8; y i;1 ¼ x; i ¼ 1; 2; . . . ; 7; 1 p : y ¼ ex þ 1; 8 81 1 2 > < y i;2 ¼ 2 x ; i ¼ 1; 2; . . . ; 6; p2 : y 72 ¼ x þ ex 1; > : y 82 ¼ 2ex x 2ex þ 2; .. . 8 y i;3 ¼ 13 x3 ; i ¼ 1; 2; . . . ; 5; > > > < y ¼ 1 x2 ex x þ 1; 63 2 p3 : > y ¼ 2ex x þ 4ex þ 2x 4; > 73 > : y 83 ¼ 2ex x2 4ex x 4ex þ 4:
ð31Þ ð32Þ
ð33Þ
ð34Þ
Combining all the terms: (31)–(34) gives 1 1 1 1 5 1 6 1 7 y 7 ðxÞ ¼ 1 þ x þ x2 þ x3 þ x4 þ x þ x þ x: 2 6 24 120 720 5040 The numerical results obtained in Table 2. In Table 2, we list the results obtained by homotopy perturbation method and compared with modified decomposition method (MDM) results given in [32] at Table 2 Numerical results for Example 2 x
Error HPM, N = 3
HPM, N = 5
HPM, N = 7
MDM
0.25 0.50 0.75
0.0314 0.1279 0.2967
1.6311e4 0.0026 0.0135
3.8922e10 1.0255e7 2.7065e6
4.91e5 7.04e5 4.98e5
A. Golbabai, M. Javidi / Applied Mathematics and Computation 191 (2007) 334–346
341
x = 0.25(0.25)0.75. As we see from this Table, it is clear that the result obtained by the present method is very superior to that obtained by MDM method. As can be seen from Table 2, the error decreased when the integer N is increased until N = 7. Example 3. Consider the following linear eighth-order problem [32]: y ðviiiÞ ðxÞ ¼ 8ex þ yðxÞ;
ð35Þ
0 < x < 1;
with the following boundary conditions: yð0Þ ¼ 1;
y 0 ð0Þ ¼ 0;
y ðivÞ ð0Þ ¼ 3;
y 00 ð0Þ ¼ 1;
y ðvÞ ð0Þ ¼ 4;
y 000 ð0Þ ¼ 2;
y 0 ð1Þ ¼ e;
y 00 ð1Þ ¼ 2e:
ð36Þ
The analytic solution is yðxÞ ¼ ð1 xÞex : Using the transformation (3) we can rewrite the eighth-order boundary value problem (19) as the system integral equations Rx 8 y 1 ¼ 1 þ 0 y 2 ðtÞ dt; > > > Rx > > > y 2 ¼ 0 y 3 ðtÞ dt; > > > Rx > > > y 3 ¼ 1 þ 0 y 4 ðtÞ dt; > > > R > < y ¼ 2 þ x y ðtÞ dt; 4 0 5 ð37Þ Rx > y 5 ¼ 3 þ 0 y 6 ðtÞ dt; > > > Rx > > > y 6 ¼ 4 þ 0 y 7 ðtÞ dt; > > > Rx > > > ¼ A þ y ðtÞ dt; y > 7 0 8 > > Rx : y 8 ¼ B þ 0 ð8et þ y 1 ðtÞÞ dt: Using (14) for (37) we have Rx 8 y 10 þ py 11 þ p2 y 12 þ ¼ 1 þ p 0 ðy 20 þ py 21 þ p2 y 22 þ Þ dt; > > R > x > > y þ py 21 þ p2 y 22 þ ¼ p 0 ðy 30 þ py 31 þ p2 y 32 þ Þ dt; > > < 20 .. . > > Rx > > > y 70 þ py 71 þ p2 y 72 þ ¼ A þ p 0 ðy 80 þ py 81 þ p2 y 82 þ Þ dt; > > Rx : y 80 þ py 81 þ p2 y 82 þ ¼ B þ p 0 ð8et þ y 10 þ py 11 þ p2 y 12 þ Þ dt: Comparing the coefficient of like powers of p, we have: 8 y 10 ¼ 1; > > > > > y 20 ¼ 0; > > > > > > > > y 30 ¼ 1; > > < y ¼ 2; 40 0 p : > y 50 ¼ 3; > > > > > > y 60 ¼ 4; > > > > > y 70 ¼ A; > > > : y 80 ¼ B;
ð38Þ
ð39Þ
342
A. Golbabai, M. Javidi / Applied Mathematics and Computation 191 (2007) 334–346
8 y 11 > > > > > y 21 > > > > > y 31 > > > > y 51 > > > > > > y 61 > > > > > y 71 > > : y 81 8 y 12 > > > > > y 22 > > > > > y > > > 32 > y > 52 > > > > > y 62 > > > > > y 72 > > : y 82 .. . 8 y 13 > > > > > y 23 > > > > > > > > y 33 > > y 53 > > > > > > y 63 > > > >y > > 73 > > : y 83
¼ 0; ¼ x; ¼ 2x; ¼ 3x;
ð40Þ
¼ 4x; ¼ Ax; ¼ Bx; ¼ 8ex þ x þ 8; ¼ 12 x2 ; ¼ x2 ; ¼ 32 x2 ; ¼ 2x2 ;
ð41Þ
¼ 12 Ax2 ; ¼ 12 Bx2 ; ¼ 8ex þ 12 x2 þ 8x þ 8; ¼ 0;
¼ 13 x3 ; ¼ 12 x3 ; ¼ 23 x3 ; ¼ 16 Ax3 ;
ð42Þ
¼ 16 Bx3 ; ¼ 8ex þ 16 x3 þ 4x2 þ 8x þ 8; ¼ 0; ¼ 16 x3 :
Combining all the terms: (39)–(42) gives 7 5 1 1 1 1 1 7 1 y 8 ðxÞ ¼ 9 þ x2 þ x3 þ x4 þ x5 þ Ax6 þ Bx7 8ex þ x8 þ x þ x6 þ 8x: 2 24 30 720 5040 40320 630 90
ð43Þ
Using the boundary conditions y 0 ð1Þ ¼ e and y 00 ð1Þ ¼ 2e leads to the system for unknown coefficients DX ¼ b; where
" D¼
and
" b¼
1 120 1 24
1 720 1 120
# ð44Þ
# þ 8e 2:7183 32051 1680 ; 5:4366 11929 þ 8e 720
X ¼ ½A; BT :
ð45Þ
The solution of this system is A ¼ 5:0074;
B ¼ 5:9710:
ð46Þ
A. Golbabai, M. Javidi / Applied Mathematics and Computation 191 (2007) 334–346
343
Table 3 Numerical results for Example 3 x
Error
0.25 0.50 0.75 1.00
HPM, N = 6
HPM, N = 7
HPM, N = 8
7.2883e8 9.8635e6 1.7626e4 0.0014
5.7238e9 6.0777e7 1.8121e5 1.9386e4
2.1630e9 1.1571e7 1.0479e6 4.2188e6
The numerical results were obtained in Table 3. In Table 3, we list the results obtained by homotopy perturbation method at x = 0.25(0.25)1. As can be seen from Table 3, the error decreased when the integer N is increased until N = 8. Example 4. One of the problems considered by Scott and Watts [37] is the linear fourth-order problem 1 y ð4Þ ðxÞ ðc þ 1Þy 00 ðxÞ þ cyðxÞ ¼ cx2 1 ¼ gðxÞ; 2
0 < x < 1;
ð47Þ
subject to the boundary conditions yð0Þ ¼ 1; y 0 ð0Þ ¼ 1;
3 þ sinhð1Þ; 2 y 0 ð1Þ ¼ 1 þ coshð1Þ:
yð1Þ ¼
ð48Þ
The analytic solution is 1 yðxÞ ¼ 1 þ x2 þ sinhðxÞ: 2 This is an interesting problem and has some unusual behavior. If we regroup (47) in terms of c we get 1 2 ð4Þ 00 00 ½y ðxÞ y ðxÞ þ 1 c y ðxÞ yðxÞ þ x ¼ 0: ð49Þ 2 We can rewrite (49) as follows: 00 1 1 y 00 ðxÞ yðxÞ þ x2 c y 00 ðxÞ yðxÞ þ x2 ¼ 0: 2 2
ð50Þ
The solution of fourth-order problem is also a solution of 1 y 00 ðxÞ yðxÞ þ x2 ¼ 0: 2
ð51Þ
This results in the solution of the original problem being independent of constant c. However, for procedures which solve initial value problems, the solutions of the intermediate problems are not independent of c. Hence for large values of c, any procedure solving initial value problems will have difficulty. Using the transformation (3) we can rewrite the fourth-order boundary value problem (47) and (48) as the system integral equations Rx 8 y 1 ¼ 1 þ 0 y 2 ðtÞ dt; > > > < y ¼ 1 þ R x y ðtÞ dt; 2 R0x 3 ð52Þ > y ¼ A þ y ðtÞ dt; > 3 0 4 > Rx : y 4 ¼ B þ 0 ððc þ 1Þy 3 ðtÞ cy 1 ðtÞ þ gðtÞÞ dt:
344
A. Golbabai, M. Javidi / Applied Mathematics and Computation 191 (2007) 334–346
Using (14) for (49) we have 8 Rx y 10 þ py 11 þ p2 y 12 þ ¼ 1 þ p 0 ðy 20 þ py 21 þ p2 y 22 þ Þ dt; > > > Rx > 2 2 > > < y 20 þ py 21 þ p y 22 þ ¼ 1 þ p R0 ðy 30 þ py 31 þ p y 32 þ Þ dt; x y 30 þ py 31 þ p2 y 32 þ ¼ A þ p 0 ðy 40 þ py 41 þ p2 y 42 þ Þ dt; > Rx > > > y 40 þ py 41 þ p2 y 42 þ ¼ B þ p 0 ððc þ 1Þðy 30 þ py 31 þ p2 y 32 þ Þ > > : cðy 10 þ py 11 þ p2 y 12 þ Þ þ gðtÞÞ dt: Comparing the coefficient of like powers of p, we have: 8 y 10 ¼ 1; > > > < y ¼ 1; 20 p0 : > y > 30 ¼ A; > : y 40 ¼ B; 8 y 11 ¼ x; > > > < y ¼ Ax; 21 p1 : > y > 31 ¼ Bx; > : y 41 ¼ Acx þ Ax cx x þ 16 cx3 ; 8 y 12 ¼ 12 12 Ax2 ; > > > < y 22 ¼ 12 12 Bx2 ; p2 : > y 32 ¼ 121 241 þ 12 ðAc þ A c 1Þx2 ; > > : y 42 ¼ 12 ððc þ 1Þb cÞx2 ; .. . 8 y 13 ¼ 16 Bx3 ; > > > < 1 y 23 ¼ 120 cx5 þ 13 12 Ac þ 12 A 12 c 12 x3 ; 3 p : > y 33 ¼ 13 12 ðc þ 1ÞB 12 c x3 ; > > : 1 y 43 ¼ 120 ðc þ 1Þcx5 þ 13 ðc þ 1Þ 12 Ac þ 12 A 12 c 12 12 Ac x3 :
ð53Þ
ð54Þ
ð55Þ
ð56Þ
ð57Þ
Combining all the terms: (51)–(54) gives
1 1 1 1 1 1 1 1 1 1 1 y 6 ðxÞ ¼ 1 þ x þ Ax2 þ Bx3 þ cx6 þ Ac þ A c x4 þ ðc þ 1ÞB c x5 2 6 720 4 6 6 6 6 5 24 24 1 1 1 1 1 1 1 1 ðc þ 1Þcx8 þ ðc þ 1Þ Ac þ A c Ac x6 : þ 40320 6 60 2 2 2 2 120
ð58Þ
Using the boundary conditions yð1Þ ¼ 2 þ sinhð1Þ and y 0 ð1Þ ¼ 2 þ coshð1Þ leads to the system for unknown coefficients DX ¼ b; where
" D¼
13 24
29 1 þ 720 c þ 720 ðc þ 1Þ2
1 6
1 þ 120 ðc þ 1Þ
7 6
19 1 þ 120 c þ 120 ðc þ 1Þ2
1 2
þ 241 ðc þ 1Þ
# ð59Þ
and 2
3 62
b¼4
þ sinhð1Þ
n
23 24
n
2
o3
2
7 o5
7 1 1 144 c þ 40320 ðc þ 1Þc 720 ðc þ 1Þ
1 1 2 þ coshð1Þ 16 15 c þ 5040 ðc þ 1Þc 120 ðc þ 1Þ
ð60Þ
A. Golbabai, M. Javidi / Applied Mathematics and Computation 191 (2007) 334–346
345
Table 4 Numerical results for Example 4 x
Error HPM, N = 5
HPM, N = 10
HPM, N = 14
c=1 0.25 0.5 0.75
N=5 3.2696e7 2.0146e5 2.2050e4
N = 10 2.4034e13 1.0804e11 7.9793e10
N = 14 2.3484e13 5.9036e13 5.5106e13
c = 100 0.25 0.50 0.75
1.4046e4 0.0034 0.0397
2.0768e4 5.0700e4 0.0017
2.0763e4 4.4177e4 6.3589e5
and X ¼ ½A; BT :
ð61Þ
We can solve this system to obtain A and B. The numerical results obtained in Table 4. In Table 4, we list the results obtained by homotopy perturbation method at x = 0.25(0.25)0.75 with c = 1 and c = 100. This results in the solution of the original problem being independent of constant c. However, As can be seen from Table 4, for procedures which solve initial value problems, the solutions of the intermediate problems are not independent of c. Hence for large values of c, any procedure solving initial value problems will have difficulty. 4. Conclusions Homotopy perturbation method is applied to the numerical solution for solving eighth-order boundary value problems. Comparison of the result obtained by the present method with that obtained by modified decomposition method [32] reveals that the present method is very effective and convenient. References [1] J.H. He, A coupling method of a homotopy technique and a perturbation technique for non-linear problems, Int. J. Non-Linear Mech. 35 (1) (2000) 37–43. [2] J.H. He, The homotopy perturbation method for non-linear oscillators with discontinuities, Appl. Math. Comput. 151 (1) (2004) 287– 292. [3] J.H. He, Application of homotopy perturbation method to nonlinear wave equations, Chaos, Solitons, Fractals 26 (3) (2005) 695–700. [4] J.H. He, Asymptotology by homotopy perturbation method, Appl. Math. Comput. 156 (3) (2004) 591–596. [5] J.H. He, Homotopy perturbation method for solving boundary problems, Phys. Lett. A 350 (1–2) (2006) 87–88. [6] J.H. He, Limit cycle and bifurcation of nonlinear problems, Chaos, Solitons, Fractals 26 (3) (2005) 827–833. [7] S. Abbasbandy, Iterated He’s homotopy perturbation method for quadratic Riccati differential equation, Appl. Math. Comput. 175 (1) (2006) 581–589. [8] P.D. Ariel, T. Hayat, S. Asghar, Homotopy perturbation method and axisymmetric flow over a stretching sheet, Int. J. Nonlinear Sci. Numer. Simul. 7 (4) (2006) 399–406. [9] D.D. Ganji, A. Sadighi, Application of He’s homotopy-perturbation method to nonlinear coupled systems of reaction–diffusion equations, Int. J. Nonlinear Sci. Numer. Simul. 7 (4) (2006) 411–418. [10] M. Rafei, D.D. Ganji, Explicit solutions of Helmholtz equation and fifth-order KdV equation using homotopy perturbation method, Int. J. Nonlinear Sci. Numer. Simul. 7 (3) (2006) 321–328. [11] A.M. Siddiqui, R. Mahmood, Q.K. Ghori, Homotopy perturbation method for thin film flow of a fourth grade fluid down a vertical cylinder, Phys. Lett. A 352 (4–5) (2006) 404–410. [12] M. Ghasemi et al., Numerical solution of the nonlinear Voltra–Fredholm integral equations by using homotopy perturbation method, Appl. Math. Comput. (2006), doi:10.1016/j.amc.2006.10.015. [13] J.H. He, Approximate solution of nonlinear differential equations with convolution product nonlinearities, Comput. Meth. Appl. Mech. Eng. 167 (12) (1998) 6973.
346
A. Golbabai, M. Javidi / Applied Mathematics and Computation 191 (2007) 334–346
[14] J.H. He, Approximate analytical solution for seepage flow with fractional derivatives in porous media, Comput. Meth. Appl. Mech. Eng. 167 (12) (1998) 5768. [15] J.H. He, Variational iteration method kind of non-linear analytical technique: some examples, Int. J. Non-Linear Mech. 34 (4) (1999) 699–708. [16] J.H. He, Variational iteration method for autonomous ordinary differential systems, Appl. Math. Comput. 114 (2-3) (2000) 11523. [17] S. Momani, S. Abuasad, Application of He’s variational iteration method to Helmholtz equation, Chaos, Solitons, Fractals 27 (5) (2006) 111923. [18] A.A. Soliman, A numerical simulation and explicit solutions of KdV Burgers’ and Lax’s seventh-order KdV equations, Chaos, Solitons, Fractals, in press, doi:10.1016/j.chaos.2005.08.054. [19] E.M. Abulwafa, M.A. Abdou, A.A. Mahmoud, The solution of nonlinear coagulation problem with mass loss, Chaos, Solitons, Fractals, in press, doi:10.1016/j.chaos.2005.08.044. [20] Z.M. Odibat, S. Momani, Application of variational iteration method to nonlinear differential equations of fractional order, Int. J. Nonlinear Sci. Numer. Simul. 7 (1) (2006) 2736. [21] M. Javidi, A. Golbabai, Exact and numerical solitary wave solutions of generalized Zakharov equation by the variational iteration method, Chaos, Solitons Fractals, in press, doi:10.1016/j.chaos.2006.06.088. [22] J.H. He, Exp-function method for nonlinear wave equations, Chaos, Solitons, Fractals 30 (3) (2006) 700–708. [23] J.H. He, M.A. Abdou, New periodic solutions for nonlinear evolution equations using Exp-function method, Chaos, Solitons, Fractals, in press, doi:10.1016/j.chaos.2006.05.072. [24] M. Wang, X. Li, Applications of F-expansion to periodic wave solutions for a new Hamiltonian amplitude equation, Chaos, Solitons, Fractals 24 (5) (2005) 1257–1268. [25] D. Kaya, S.M. El-Sayed, A numerical simulation and explicit solutions of the generalized Burger–Fisher equation, Appl. Math. Comput. 152 (2004) 403–413. [26] J.H. He, Some asymptotic methods for strongly nonlinear equations, Int. J. Mod. Phys. B 20 (10) (2006) 1141–1199. [27] J.H. He, Non-perturbative methods for strongly nonlinear problems, Berlin: dissertation.de-Verlag im Internet GmbH, 2006. [28] H.N. Caglar, S.H. Caglar, E.E. Twizel, The numerical solution of fifth-order boundary value problems with sixth degree B-spline functions, Appl. Math. Lett. 12 (1999) 25–30. [29] A.R. Davies, A. Karageoghis, T.N. Phllips, Spectral Galerkin methods for the primary two-point boundary-value problems in modelling viscoelastic flows, Int. J. Numer. Methods Eng. 26 (1988) 647–662. [30] A. Karageoghis, T.N. Phllips, A.R. Davies, Spectral collocation methods for the primary two-point boundary-value problems in modelling viscoelastic flows, Int. J. Numer. Methods Eng. 26 (1998) 805–813. [31] G.L. Liu, New research directions in singular perturbation theory: artificial parameter approach and inverse-perturbation technique, in: Proceeding of the Conference of the 7th Modern Mathematics and Mechanics, Shanghai, 1997. [32] M. Mestrovic, The modified decomposition method for eighth-order boundary value problems, Appl. Math. Comput. (2006), doi:10.1016/j.amc.2006.11.015. [33] Layne T. Watson, Melvin R. Scott, Solving spline-collocation approximations to nonlinear two-point boundary-value problems by a homotopy method, Appl. Math. Comput. 24 (1987) 333–357. [34] Layne Watson, Engineering applications of the Chow–Yorke algorithm, Appl. Math. Comput. 9 (1981) 111–133. [35] M.R. Scott, H.A. Watts, A systematized collection of codes for solving twopoint boundary-value problems, in: L. Lapidus, W.E. Schiesser (Eds.), Numerical Methods for Differential Systems, Academic, New York, 1976. [36] M.R. Scott, H.A. Watts, Computational solution of linear two-point boundary value problems via orthonormalization, SIAM J. Numer. Anal. 14 (1977) 40–70. [37] M.R. Scott, H.A. Watts, SUPORT—a computer code for two-point boundary-value problems via orthonormalization, SAND750198, Sandia Laboratories, Albuquerque, NM, 1975.