Adomian decomposition method for solving higher order boundary value problems

Mathematical Theory and Modeling www.iiste.org
ISSN 2224-5804 (Paper) ISSN 2225-0522 (Online)
Vol.2, No.1, 2011

Adomian Decomposition Method for Solving Higher Order
Boundary Value Problems
Neelima Singh* Manoj Kumar

Department of Mathematics,Motilal Nehru National Institute of Technology, Allahabad 211004 (U.P.) India
*E-mail: sneelima_2000in@yahoo.com

Abstract

In this paper, we present efficient numerical algorithms for the approximate solution of linear
and non-linear higher order boundary value problems. Algorithms are, based on Adomian
decomposition. Also, the Laplace Transformation with Adomian decomposition technique is
proposed to solve the problems when Adomian series diverges. Three examples are given to
illustrate the performance of each technique.

Keyword: Higher order Singular boundary value problems, Adomian decomposition techniques,
Laplace transformations.

1. Introduction

The Adomian decomposition method, proposed initially with the aims to solve frontier physical
problems, has been applied to a wide class of deterministic and stochastic problems, linear and
nonlinear problems, in physics, biology and chemical reactions etc (Adomian 1992). Inspired
and motivated by the ongoing research in this area, the method assumes a series solution for the
unknown quantity. It has been shown ( Abbaooui & Cherruault 1999, Hosseini & Nasabzadeh
2006, that the series converges fast, and with a few terms this series approximate the exact
solution with a fairly reasonable error, normally very less. Each term of this series is a
generalized polynomial called the Adomian polynomial. Besides, it also has certain advantages
over standard numerical methods as it is free from rounding-off errors and computationally
inexpensive since it does not involve discretization. In this paper, we adopt the algorithm to the
solution of boundary value problems arising in the modelling of real life problems. The fourth
order two point boundary value problems have received a lot of attention in the literature due to
their many applications in elasticity, was investigated by Kosmatov (2004). The objective of this
paper is to implement a symbolic code (discussed in Kumar at el. 2010 and Kumar at el. 2011)
for fully reflecting a simple and reliable technique for third order singular boundary value
problems and fourth order beam bending problem using Mathematica 6.0.

The balance of this paper is as follows. In the next Section, we briefly introduce the Adomian
decomposition method. In Section 3, we will explain Adomian decomposition method for higher
order singular boundary value problems. In Section 4, Laplace transformation with Adomian
decomposition method for oscillatory solutions for which Adomian method diverges is
described. In Section 5, we present the numerical experiment which shows effectiveness of the
proposed method. In the last Section, concluding remark with summary of the paper is given.

2. Adomian decomposition method:

In this section, we describe the Adomian decomposition method as it applies to a general
nonlinear equation of the form

11

Vol.2, No.1, 2011

y  Ny  f (2.1)
where N is a nonlinear operator from a Hilbert space H into H , f is a given function in H . By
the decomposition method y is a series solution given by


y   yn (2.2)
n 0
Nonlinear operator N is decomposed by

Ny   An (2.3)
n 0
From (2.1), (2.2) and (2.3) we have
 

 yn   An  f
n 0 n 0
(2.4)

An are Adomian polynomials that can be constructed for various classes of nonlinearity
according to specific algorithms set by Adomian 1992.
A0  f  y0 
A1  y1 f '  y0 
y12
A2  y 2 f  y 0   f  y 0 ,
2!


An 
1 dn
n! dn

f i yi  0 , n  0,1, 2...... (2.5)

If the series (2.3) is convergent, then (2.4) hold as
y0  f
y1  A0 ( y0 )
y2  A1 ( y0 , y1 )


yn  An1 ( y0 , y1 ,......., yn1 )

Thus, we can recursively determines every term of the series  y n .
n 0
3. Adomian decomposition method for higher order singular boundary value problems

For the higher order singular boundary value problems of type
m n
y n1  y  q( x) F ( y )  g ( x) , (3.1)
x
y(0)  a0 , y(0)  a1 , , y n1 (0)  an1 , y(b)  c (3.2)

12

Vol.2, No.1, 2011

where F is a nonlinear operator of order less than n . Therefore equation (3.1) and (3.2) can be
rewritten in operator form as
Ly  g ( x)  q( x) F ( y) (3.3)
where n fold operator L is defined by Hasan et. al., 2009 as
d n 1 n m d m  n
1
L1  x x x (.) (3.4)
dx n dx
n
1 d d d
L2  x x n m x m  n1 (.) (3.5)
dx n dx dx
Hasan et al., 2009 defined the inverse operator for (3.4) and (3.5) respectively as:
x x x x

x     x (.) dx dx
1 nm m  n 1
L x
1 (3.6)
b 0 0 0
x x x x
L  x
1
2
n  m 1
 x    x (.) dx dx
mn
(3.7)
0 b 0 0

By applying L1 1 and L1 to the equation (3.1) we have

2
 
y 1 ( x)   yn  1 ( x)  L11 g ( x)  L11[q( x) An ]
 
(3.8)
n 0 n 0
 
y  2  ( x)   yn  2 ( x)  L1 g ( x)  L1[q( x) An ]
2 2 (3.9)
n 0 n 0

where Ny   An , specific formula for finding Adomian polynomial is explained in (2.5) for
n 0

nonlinear term F (y) . y1 ( x) and y 2 ( x) are approximate solution using inverse operator (3.6) and
(3.7) respectively. The components y n can be obtained by applying modified Adomian
decomposition method from the recurrence relation:
 y0   ( x)
 1 1
 y1  L g ( x)  L A1 (3.10)
 1
 yn 1   L An , n  1
The n term approximation can be obtained by,
n
 n   yi (3.11)
i 0
The Adomian decomposition method (ADM) outlined above (detailed in Hassan 2009) is easy to
implement and does not need discretization. However, it has some drawbacks. Its efficiency and
accuracy rely on the convergence and the rate of convergence of the series solution. We found,
that the ADM gives a series solution which may have a slow rate of convergence over wider
regions. Furthermore, if the solution of the problem is oscillatory, then the ADM series solution
may be divergent. To overcome these drawbacks, ADM needs to be modified in order to work
for problems where the solutions are of oscillatory in nature. Due to this difficulty the idea of
Laplace transform is introduced with Adomian decomposition method to solve such problems.

13

Vol.2, No.1, 2011

4. Laplace Transformation with Adomian decomposition method for higher order
boundary value problems

In this section, the Laplace Transform with Adomian decomposition Method discussed (Hajji et
al., 2008) for solving nonlinear higher order boundary value problems in the interval [0, b] of the
form
y ( n)  q( x) F ( y)  g ( x), 0 xb (4.1)
subject to the condition
y(0)  a0 , y(0)  a1 , , y n1 (0)  an1 , y(b)  c (4.2)
Appling Laplace Transform integral operator (denoted by L ) both side of equation (4.1)
L[ y ( n) ]  L[q( x) F ( y)]  L[ g ( x)], 0 xb (4.3)
It gives
s n L[ y]  s n1a0  s n2 a1    san2  an1  L[q( x) F ( y)]  L[ g ( x)], (4.4)
Substituting boundary conditions from (4.2) we have
s n L[ y]  s n 1 y(0)  s n  2 y(0)    syn  2 (0)  y n 1 (0)  L[q( x) F ( y)]  L[ g ( x)], (4.5)
where the constants which are unknown are determined by imposing the boundary conditions
at x  b .
1
L[ y]  n ( s n1 a0  s n2 a1    s a n2  a n1  L[q( x) F ( y)]  L[ g ( x)] ) (4.6)
s
 
Putting y ( x)   yn ( x) and F ( y )   An in (4.6) and comparing the terms we will get
n 0 n 0
1
L[ y0 ]  n ( s n1a0  s n2 a1    s a n2  an 1  L[ g ( x)] ) (4.7)
s
1
L[ y1 ]   n ( L[q( x) A0 ] ) (4.8)
s
1
L[ y 2 ]   n ( L[q( x) A1 ] ) (4.9)
s
1
L[ y3 ]   n ( L[q( x) A2 ] ) (4.10)
s

Applying the inverse Laplace transform to equation (4.7), gives the zeroth component of
Adomian solution. After substituting the value of A0 , A1 , A2 , , An in (4.8-4.10) respectively and
applying the inverse Laplace transform, we obtain the solution
components y0 , y1 , y 2 ,, y n successively. The n term approximation is given by
n 1
 n ( x, c)   yi ( x, c), (4.11)
i 1
To determine the constant c , we require that  n ( x, c) satisfies the boundary conditions at x  b .
Solving the algebraic equation, we obtain the required constant, which complete the numerical
solution of our nonlinear higher-order boundary value problems.

14

Vol.2, No.1, 2011

5. Computer Simulation

In this section, we apply the proposed algorithm on two third order non-linear singular boundary
value problems and one fourth order oscillatory boundary value problem.

Example 1 :( Third order singular problem)
2
y  y  y  y 2  7 x 2 e x  6 x e x  6 e x  x 6 e 2 x , (5.1)
x
y(0)  y(0)  0, y(1)  e, (5.2)
Operator form of the equation is
Ly  q( x)  F ( y, x), (5.3)
where
q( x)  7 x2e x  6 xe x  6e x  x6e2 x and F ( y, x)  y 2
So the inverse operators for the equation (5.1), (5.2) from (3.6), (3.7) are

x x x
L  x x
1
1
4 5
  x(.) dx dx dx (5.4)
1 0 0
x x x
L  x
1
2
4
x 
5
x(.) dxdxdx (5.5)
0 1 0
Applying L1 on the both side of equation (5.2) we get

y(i ) ( x)  2.71828 x 4  L1q( x)  L1  An ,
i i (5.7)
k 1
2
Nonlinear term y , is calculated by the Adomian polynomial as,
A0  y0 ,
2

A1  2 y0 y1 ,
A2  2 y0 y2  y1 ,
2


After computing the values of series components y1 , y 2 , y3 , we get the third order
approximation of series solution
3
3   y i , (5.8)
i 1

The exact solution of the (5.1), subject to the (5.2) is y  x e , the comparison of exact
3 x

solution and approximate solution is given in table 1. Figure 1 shows the graphical representation
of the approximate solutions which is very close to the exact solution.

15

Vol.2, No.1, 2011

Table 1

x y(ADM ) y(Exact) Error
0 0 0 0
0.1 0.00110664 0.00110666 1.88E-8
0.2 0.00986585 0.00986616 3.01E-7
0.3 0.03751990 0.03752150 1.52E-6
0.4 0.10147900 0.10148400 4.83E-6
0.5 0.22886500 0.22887700 1.19E-5
0.6 0.46120300 0.46122900 2.56E-5
0.7 0.86027700 0.86032800 5.07E-5
0.8 1.51512000 1.51522000 9.82E-5
0.9 2.55015000 2.55035000 1.92E-4
1.0 4.13454000 4.13492000 1.92E-4
y

2.5

2.0

approx
1.5
exact

1.0

0.5

x
0.2 0.4 0.6 0.8 1.0

Figure 1: Exact and Approximate solution of example 1

Example 2 :( Third order singular problem)
2
y  y  y 3  g ( x), (5.9)
x
y(0)  y(0)  0, y(1)  10.873, (5.10)
Operator form of the equation is
Ly  g ( x)  F ( y, x), (5.11)
where g ( x)  7 x e  6 xe  6e  x e  x e and F ( y, x)  y
2 x x x 9 3x 3 x 3

So the inverse operator for the equation (5.10) and (5.9) is
x x x
L1 ()  x 4  x 5   x() dx dx dx (5.12)
1 0 0

Applying L1 on the both side of equation (5.11) we get

16

Vol.2, No.1, 2011

 
y ( x)  L1 g ( x)  L1  y  L1  An  y (1) x 4 , (5.13)
k 1 k 1
2
Nonlinear term y is calculated by the Adomian polynomial as,
A0  y0
3

A1  3 y0 y1 ,
2

A2  3 y0 y2  3 y1 y0 ,
3 2


After computing the values of series components y1 , y 2 , y3 , third approximation of series
3
solution is 3   y i , (5.14)
i 1

The exact solution of the (5.9) with (5.10) is y  x e , the comparison of exact solution and
3 x

approximate solution is given in table 2. Figure 2 shows the graphical representation of the
approximate solutions which is very close to the exact solution.

Table 2

x y(ADM) y(Exact) Error
0 0 0 0
0.1 0.00110664 0.00110666 1.88E-8
0.2 0.00986585 0.00986616 3.01E-7
0.3 0.03751990 0.03752150 1.52E-6
0.4 0.10147900 0.10148400 4.83E-6
0.5 0.22886500 0.22887700 1.19E-5
0.6 0.46120300 0.46122900 2.56E-5
0.7 0.86027700 0.86032800 5.07E-5
0.8 1.51512000 1.51522000 9.82E-5
0.9 2.55015000 2.55035000 1.92E-4
1.0 4.13454000 4.13492000 1.92E-4

17

Vol.2, No.1, 2011

2.0

1.8

1.6 approx

1.4 exact

1.2

0.2 0.4 0.6 0.8 1.0
Figure 2: Exact and Approximate solution of example 2

Example 3: (Elastic beam equation rigidly fixed at both ends)

Fourth-order nonlinear differential equations have many applications such as balancing condition
of an elastic beam whose two ends are simply supported, may be described by nonlinear fourth-
order ordinary differential equations. Beams can be used in many different settings, as long as
their capabilities are understood. Many different types of beams are available; each with their
own identity which depends on the types of material, length, width, depth, and external forces
being placed on the beam. One major concern that needs to be considered, when deciding what
type of beam to use in a certain structure, is deflection.

Deflection is the displacement of any point along the beam from its original position, measured
in the y direction. It is generally shown in a graph of the deflection curve, representing the
deflection versus incremental load values. The specific values of deflection can be found
through differential equations. The moment-curvature equation is a second-order, ordinary
differential equation; whereas, the load-deflection equation is a fourth-order, ordinary differential
equation. Both are very useful equations and can be easily programmed for solution by
computer. Beams are used everywhere (stadiums, airports, bridges, etc.), but only when their
capabilities are known can they be used safely.

We consider the problem of bending of a long uniformly rectangular plate supported over the
entire surface by an elastic foundation and rigidly supported along the edges. The vertical
deflection y at every point satisfies the system
D y (4) ( x)  f ( x, y)  g y, 0  x  b, (5.15)
y(0)  y (0)  y(1)  y (1)  0 (5.16)
where D, is the flexural rigidity of the plate, f is the intensity of the load acting on the plate,
and g is the reaction of the foundation. The more details we refer Agarwal et al., (1982).

Another mathematical model:

18

Vol.2, No.1, 2011

Suppose y, represents an elastic beam of length b, which is clamped at its left side x  0 , and
resting on a kind of elastic bearing at its right side x  b , along its length, a load f is added to
cause deformations. Then the differential equation that models this phenomenon is given by
y (4) ( x)  f ( x, y), 0  x  b, (5.17)
y(0)  y(0)  0, y(b)  0, y (b)  g ( y(b)) (5.18)
where f  C ([0,1]  R) and g  C ( R) .Owing to its importance in physics, the existence of
solutions to this nonsingular problem has been studied by many authors (Pang et al., (2006),
Wazwaz, (2002), Cabada, (1994)), However, in practice only its positive solutions are
significant. Corresponding problems modeling vibrating beams on elastic bearings were
considered in Ma, 2001. The detail of the mechanical interpretation of the above two models
belongs to a general class of boundary value problems of the form
y 4 ( x)  g ( x) F ( x, y)  f ( x), 0  x  1, (5.19)
y(0)  0 , y(0)  1 , y(b)  0 , y(b)  1 (5.20)
We consider the problem of bending a rectangular clamped beam of length b.  resting on an
elastic foundation. The vertical deflection y (x) of the beam satisfies the system
y (4)  64 y  Sin 2 x, 0 x  (5.21)
y(0)  0 , y (0)  0, y( )  0, y ( )  0 (5.22)

Adomian decomposition method:
y 0  a0
3
1 i
y1   ai x  64 L1 [ y 0 ]  L1 [ Sin 2 x]
i 0 i!
y 2  64 L1[ y1 ]
y3  64 L1[ y 2 ]


y ( x)   y n ( x) (5.23)
n 0

   3
y ( x)   yk ( x)    ai x

i  64 x
4 k


 64k Sin 2 x 

k 0 k 0  i 0
 i  4k ! 2 4k 4 

(5.24)

The series solution (5.24) shows the divergent geometric series. Thus we can observe that the
ADM approach diverges, therefore it is not applicable to that type of problem in which solution
is oscillatory. To overcome this difficulty we will use Laplace Transformation (discussed in Hajji
et al., (2008))

Laplace Adomian decomposition technique:

Now applying Laplace transform integral operator to the both sides of the equation (5.19), we get
L[ y (4) ]   L[64 y]  L[Sin 2 x]
Using the formula of the Laplace transform of the derivatives, we obtain

19

Vol.2, No.1, 2011

  2
L[ y 0 ]    (5.25)
s 3
s 4
s ( s 2  4)
4

(64) n
L[ y n ]   ( L[ y n 1 ] ), n  1 (5.26)
s4
Following the ADM procedure outlined in the previous section, we find that the general form of
the series solution by Laplace transformation is given by
  2 (64) n
L[ y ]  3  4  4 2  4 ( L[ y n 1 ] )
s s s ( s  4) s
 s 2
y ( x)    4 (5.27)
64  s 64  s ( s  64)( s 2  4)
4 4

For this example, it can be easily verified that the exact solution can be recovered by taking the
inverse Laplace transform. By the above analysis, we see that the coupling of the Laplace
transform with the Adomian decomposition made it possible to obtain a convergent series
expansion in the Laplace domain. A n-term approximate solution is obtained by calculating n
solution components, as described above n ( x,  ,  ) . The constants  and  are determined by
imposing the boundary conditions at x   . Using Mathematica 6.0,  and  were found to be
  0.099672 and    0.3 . Then we obtain the approximate solution is given by

7 ( x)  1638.38 x  0.0498136 x 2 - 1092.3 x 3  218.467 x 5 - 0.00885575 x 6 - 20.8025 x 7
 1.15556 x 9  0.000112454 x10 - 0.042051 x11  0.00107876 x 13- 2.99578  10 - 7 x 14
- 0.0000204876 x 15  3.00699  10 - 7 x17  2.6107  10 -10 x 18-3.55832  10 .-9 x19 
3.41515  10 -11 x 21- 9.51724  10 -14 x 22-2.57512  10 -13 x 23  1.66137  10 -15 x 25 
1.69761  10 -17 x 26 -1.13598  10 -17 x 27  6.21772  10 - 20 x 29- 1.65188  10 - 21 x 30
-1.06971  10 - 22 x 31 -819.19 sin ( 2 x).

The above series is the convergent series in Laplace domain, shows that the Adomian
decomposition method coupled with Laplace transformation gives convergent series solution for
oscillatory problems.

6. Conclusion

In this paper, we explained the Adomian decomposition method for solving higher order singular
boundary value problems and Laplace Adomian decomposition method for solving fourth-order
boundary value problems for which the Adomian decomposition method diverges. It can be a
potential tool to solve the oscillatory nonlinear higher order boundary value problems. Although
there are some other methods which can be used to solve such systems or more complex ones,
but Adomian decomposition method shows its advantages is that calculations are relatively easy
to follow and understand, besides, it can be fulfilled by mathematical software like Mathematica
6.0 though the solution is of the form of an infinite series in many cases, it can be written in a
closed form in some cases, otherwise, it can also be satisfactorily represented by proper
truncations for it shares a relatively rapid convergence.

References

20

Vol.2, No.1, 2011

Abbaoui K., Cherruault Y. (1995), New ideas for proving convergence of decomposition
methods, Computers & Mathematics with Applications. 29(1995), 103-108.

Adomain G (1992), Differential equations with singular coefficients. Applied Mathematics and
Computation. 47(1992),179–184.

Agarwal R. P. and Akrivis, C., (1982) Boundary value problems occurring in plate deflection
theory. Journal of Computational and Applied Mathematics, 8(1982), 145–154.

Cabada A. (1994). The method of lower and upper solutions for second, third, fourth and higher
order boundary value problems, Journal of Mathematical Analysis and Applications, 185, 302-
320.

Hajji M. A. and Al Khaled K. (2008), Numerical methods for nonlinear fourth order boundary
value problems with applications. International Journal of Computer Mathematics. 85(2008),
83-104.
Hasan Y. Q. and Zhu L. M. (2009). A note on the use of modified Adomian decomposition
method for solving singular boundary value problems of higher-order ordinary differential
equations. Communications in Nonlinear Science and Numerical Simulation,
volume 14, Issue 8, August 2009, Pages 3261-3265

Hasan Y. Q. and Zhu L. M. (2009). Solving singular boundary value problems of higher-order
ordinary differential equations by modified Adomian decomposition method, Communications in
Nonlinear Science and Numerical Simulation, 14(2009), 2592–2596

Hosseini M. M. and Nasabzadeh H. (2006), On the convergence of Adomian decomposition
method. Applied Mathematics and Computation, 182(2006), 536–543.

Kosmatov N. (2004), Countably many solutions of a fourth order boundary value problem,
Electronic Journal of Qualitative Theory of Differential Equations, 12(2004), 1–15.

Kumar M. and Singh N. (2009), A collection of computational techniques for solving singular
boundary-value problems. Advances in Engineering Software, Volume 40, Issue 4, April 2009,
Pages 288-297.

Kumar M. and Singh N. (2010), Modified Adomian decomposition method and computer
implementation for solving singular boundary value problems arising in various physical
problems. Computers & Chemical Engineering, Volume 34, Issue 11, 8 November 2010, Pages
1750-1760.

Kumar M. and Singh N. (2011), Computer simulation of third-order nonlinear singular problems
arising in draining and coating flows. Article first published online: International Journal for
Numerical Methods in Fluids, 11 Jul 2011,doi: 10.1002/fld.2635.

Ma T.F. (2001), Boundary stabilization for a non-linear beam on elastic bearings. Mathematical
Methods in Applied Science, 24(2001), 583–594.

Pang C., Dong W. and Wei Z. (2006), Multiple solutions for fourth-order boundary value
problem. Journal of Mathematical Analysis and Applications, 314(2006), 464–476.

21

Vol.2, No.1, 2011

Rashidinia J., Golbabaee A., (2005), Convergence of numerical solution of a fourth-order
boundary value problem. Applied Mathematics and Computation, 171(2005) 1296-1305.

Wazwaz A.M. (2002), The numerical solution of special fourth-order boundary value problems
by the modified decomposition method. International Journal of Computer Mathematics,
79(2002), 345–356.

Yang B. (2005), Positive solutions for the beam equation under certain boundary conditions,
Electronic Journal of Differential Equations, 78(2005), 1–8.

22

Adomian decomposition method for solving higher order boundary value problems

More Related Content

What's hot (20)

Similar to Adomian decomposition method for solving higher order boundary value problems (20)

More from Alexander Decker (20)

Recently uploaded (20)

Adomian decomposition method for solving higher order boundary value problems