J Braz. Soc. Mech. Sci. Eng.

DOI 10.1007/s40430-014-0214-3

TECHNICAL PAPER

New approaches to identification of the Lagrange multiplier

in the variational iteration method
S. S. Samaee • O. Yazdanpanah • D. D. Ganji

Received: 21 April 2013 / Accepted: 26 June 2014

 The Brazilian Society of Mechanical Sciences and Engineering 2014

Abstract In this paper, new methods to obtain the 1 Introduction

Lagrange multiplier are presented. Differential equation
consists of linear and nonlinear parts. We have infinite The variational iteration method (VIM) was developed by
equations that the linear parts of them are different together He [1–8]. The VIM is preferable over numerical methods
then any of the equations have different Lagrange multi- as it is free from rounding off errors as it does not involve
pliers. In this article, steps of the proposed methods are discretization and does not require large computer power
fully described manually. In this paper, we show that the or memory. This method has been proved by many
proposed methods attain precisely the Lagrange multipli- authors to be reliable and efficient for a wide variety of
ers. Exact identification of Lagrange multipliers in the VIM scientific and engineering applications. This method is
is very important for obtaining highly accurate solutions; more powerful than existing techniques such as the
on the other hand, it is complicate to determine the mul- Adomian decomposition method [9], perturbation method,
tipliers for strongly nonlinear equations. Lagrange coeffi- etc. [6]. The VIM gives rapidly convergent successive
cient is a parameter that is used in VIM. Therefore, this approximations of the exact solution if such a solution
study introduces a simple and efficient way to solve exists; otherwise, a few approximations can be used for
Lagrange coefficient. The results show that all three numerical purposes. The perturbation method suffers from
methods obtain the Lagrange coefficient without any errors the computational workload, especially when the degree
and manually are solved easily. of nonlinearity increases. Moreover, the Adomian
decomposition method involves tedious computations of
Keywords Solution differential equation  Lagrange Adomian polynomials. The VIM has no specific require-
multipliers (k)  Variational iteration method (VIM)  ments, such as linearization, small parameters, etc., for
Laplace method transform  New methods nonlinear operators. The VIM has been shown to solve
effectively, and accurately a large class of nonlinear
problems with approximations converging rapidly to
accurate solution.
To illustrate the basic concept of the He’s VIM, consider
the following general nonlinear equation:
LuðtÞ þ NuðtÞ ¼ gðtÞ ð1Þ
Technical Editor: Fernando Alves Rochinha. where L, N are linear and nonlinear operators, respec-
S. S. Samaee (&)  O. Yazdanpanah
tively, and g (t) is the source inhomogeneous term. He has
Department of Civil Engineering, Shomal University, Amol, Iran modified the general Lagrange multiplier method to an
e-mail: [email protected] iteration method called as correction functional. The basic
character of the method is to construct a correction
D. D. Ganji
Department of Mechanical Engineering, Babol university,
functional for the aforementioned equation, which reads
Babol, Iran as:

123
 J Braz. Soc. Mech. Sci. Eng.

Z x
unþ1 ðxÞ ¼ un ðxÞ þ kfL un ðsÞ þ N u~n ðsÞ  gðsÞgds ð2Þ where k(m-i), i = 2, …, m is (m - i)th derivative of k. The
0 last equation is calculated by getting extremum from the
where k is a general Lagrange multiplier. The subscript n following functional:
Z t Z t  
denotes the nth approximation, and u~n is a restricted 0

variation. kðLun Þds ¼ kf s; un ; un ; . . .; uðnmÞ ds ð6Þ

0 0
One of the problems with the VIM is the calculation of
k, Lagrange multiplier. This parameter plays the important Note that if functional I equals to
Z s2  
role in the convergence of VIM solution. For the first time,
I¼ F s; y; y0 ; . . .; yðnÞ ds; ð7Þ
the calculation of k was implemented by He [1–8], then s1
was developed future by Dehghan [10], Wazwaz [11–15]
and Ganji [16–21]. then extremum of Eq. (7) is calculated as follows:
    n  
In all of the implemented works, the process of oF d oF d2 oF nd oF
achieving k was not discussed clearly, even in some of the  þ     þ ð 1 Þ
oy ds oy0 ds2 oy00 dsn oyðnÞ
sources, k has been calculated wrongly, but, because of the ¼ 0:
high ability of the VIM, the results came to convergence
ð8Þ
with more iteration. In this paper, we studied about mod-
ifying the traditional method to achieve k with the fewer Equation (8) is called the Euler–Lagrange’s differential
steps which would be much easier and faster than the equations or Euler equation. In view (8), extremum of the
previous accomplishment. Because of needing the tradi- functional (6) is given as follows:
tional method to Leibnitz, we are not able to write a    
computer program to obtain k by means of traditional oðkf Þ d oðkf Þ d2 oðkf Þ
 0
þ 2  
method. In this paper, we present new methods that one of oy ds oy ds oy00
n 
its advantages over previous methods is resolving it d oðkf Þ
þ ð1Þn n
quickly and feels comfortable than manual methods. The ds oyðnÞ
purpose of this paper is introducing three new methods to ¼0 ð9Þ
solve the Lagrange coefficient. Section 2 introduces the old
Therefore, by solving the system of equations consisting
way of obtaining Lagrange coefficient. In Sect. 3, three
of (5) and (9), the Lagrange multiplier can be determined.
new methods are introduced this paper is fully explained
with examples. The first method is based on first and sec- Example 1 Consider the following nonlinear ordinary
ond order differential equations and higher. Second and differential equation
third approaches are based on Laplace’s method in second
method boundary conditions is considered zero, but the y0 ¼ 2y  y2 þ 1 ð10Þ
third method; we define the boundary conditions for each
Following equations are used to calculate k [22]:
equation.
k0 ð xÞ þ 2kð xÞ ¼ 0
1.1 Description of the pervious method ð11Þ
1 þ kð xÞx¼t ¼ 0
In this section, we present the method to find the Lagrangian So the multiplier can be identified as k = e2(t-x).
multiplier k. Let the derivative operator in Eq. (1) be of We purpose a different method to calculate k. As the
order m. Then, Eq. (2) assumes the following form: order of Eq. (11) is one m = 1, therefore, in view of (5) we
Z t   get,
0
unþ1 ¼ un þ kf s; un ; un ; . . .; uðnmÞ ds ð3Þ
0 1 þ kð xÞx¼t ¼ 0 ð12Þ
where f = Lun. For calculating k, we need m ? 1 equa-
further, we calculate extremum of following functional
tions. Integrating by part the formula
using Eq. (9):
Z t
unþ1 ¼ un þ kuðnmÞ ds ð4Þ Z t Z t
0
0 kf ðx; yð xÞ; y ð xÞÞdx ¼ kðy0 ð xÞ  2yð xÞ  1Þdx
0 0
We get the following m equations:
( ð13Þ
1 þ ð1Þm1 kðm1Þ
s¼t ¼0
ð5Þ The extremum of functional (13) is given by the following
kðmiÞ
s¼t ¼ 0; i ¼ 2; . . .; m; set of equations:

123
J Braz. Soc. Mech. Sci. Eng.

  c1 xðn1Þ
oðkf Þ d oðkf Þ where only the expression of the will remain
 ¼0 ðn1Þ!
oy dx oy0 through exercising the boundary conditions, and the
d ð14Þ rest answers will become zero.
2k  ðkÞ ¼ 0
dx 3. We define the boundary conditions of the problem as
k0 ð xÞ þ 2kð xÞ ¼ 0 following:
Using Eqs. (12) and (14), we get the same k as obtained by yð0Þ; y0 ð0Þ; y00 ð0Þ; . . .; yðn1Þ ð0Þ
Abbasbandy [22].
(n - 1) we suppose the initial boundary conditions
being equal to zero and place the last boundary con-
ditions being equal to A1 where A is the greater deriv-
2 The basic of the new modified method
ative coefficient of the linear section).
2.1 Description of the first method 4. We multiply the answer of the differential equation by
(-1)n (where n is the highest rank of the derivative of
Studying about k at the previous articles [1–23], we can the linear part).
observe that the k is obtained in three shapes which are 5. Wherever the result of solution differential equations
polynomial, exponential and trigonometry. All of these obtained parametrically, we change the variable, for
types are achieved from the solution of a homogeneous example, if the result of the solution differential
ordinary differential equation (ODE). Therefore, having equations obtained t, we replace it with (-1)n(s - t).
the coefficient of ODE and its boundary conditions, we
are able to calculate the k conveniently. First step to
obtain these methods are fully described then several 3 Application of the new method
examples are solved manually. To obtain the Lagrange
multiplier, we conduct the Following steps through dif- To give a clear overview of this method, we present the
ferential equation: following illustrative examples.
1. We solve the linear section of differential equation.
2. If the linear part of the solution is only the nth Example 2 Equation (1) from Table (1)
00 2
derivative of the y Function, we will take integral n y þ x y ¼ f ðtÞ: ð16Þ
time from it. (For example, y00 ? y2 = 0 where the
To solve this equation, we solve the linear Eq. (16). Since
linear part of this problem is y00 that we take integral
linear section derivative of the second order of y function
tow time from it).
must integrated two times and for simplicity solution we
For simplicity of the solution, we use the following
use formula (15).
formula:
Z Z
c1 xðn1Þ c2 xðn2Þ cm xðnnÞ c1 tð21Þ c2 tð22Þ
yn dx ¼ þ þ  þ ð15Þ y2 dt ¼ þ ¼ c1 t þ c2 n¼2 ð17Þ
ðn  1Þ! ðn  2Þ! ðn  nÞ! ð2  1Þ! ð2  2Þ!

Table 1 Comparisons between No. Equation Linear part k in article k obtained from new References
k obtained using the new equation methods
methods and k calculated in
different published articles 1 y00 þ x2 y ¼ f ðtÞ y00 þ x2 y k ¼ x1 sin xðs  tÞ k ¼ sinðxðstÞÞ [1]
x
00 2 00
2 y þ x y~ ¼ 0 y k=s-t k=s-t [1]
3 T 00 þ T þ x ¼ 0 T 00 þ T k = sin (s - x) k = sin (s - x) [2]
4 [1 ? eu]u0 ? u = 0 u0 ? u k = -es-t k = -es-t [16]
5 ut ? iuxx = 0 ut k = -1 k = -1 [11]
6 iut ? uxx ? 2|u|2u = 0 iut k=i k ¼  1i [11]
000
7 u000 þ 12 uu00 ¼ 0 u k¼  12 ðf  xÞ 2
k¼  12 ðf  xÞ 2 [12]
0 0
8 x  ex ða  byÞ ¼ 0 x k1 = -1 k1 = -1 [17]
y0 k2 = -1 k2 = -1
9 y0 = 2y - y2 ? 1 y0 - 2y k = -e2(t-f) k = -e2(t-f) [22]

10 u000 ðxÞ ¼ 1 þ 2u2 2x u000 (x) k¼  12 ðs  xÞ 2
k¼  12 ðs  xÞ 2 [23]

123
 J Braz. Soc. Mech. Sci. Eng.

Table 2 The results of solution No. Linear part of equation k obtained from new methods
differential equations for higher
pffiffi  pffiffi pffiffi 
order equations 1 u00 ðtÞ  2uðtÞ k ¼ 42 e 2ðftÞ  e 2ðftÞ
2 u000 ðtÞ þ u0 ðtÞ k = -1 ? cos (-f ? t)
000 00 4 4
3 5u ðtÞ  4u ðtÞ 5 5fþ5t
k ¼  14 f þ 14 t  16 e 5
þ 16
4 uð4Þ ðtÞ þ u00 ðtÞ k = f - t - sin (f - t)
(4) 000
þ 16 ðf  tÞ2  29 f þ 29 t  27
3 3
5 2u (t) ? 3u (t) 4 4 2fþ2t
k ¼ 27 e
pffiffiffi
6 u(5)(t) ? 3u000 (t) 1 1 2 1
k ¼ 9  6 ðf þ tÞ  9 cosð 3ðf þ tÞÞ
pffiffi pffiffi 
7 u(5)(t) ? u(4)(t) ? u000 (t) k ¼  12 ðf þ tÞ2  2 3 3 e2f2t sin 23 ðf þ tÞ  f þ t
1 1

8 u(7)(t) 1
k ¼  720 ðf þ tÞ6

(where n is the highest rank of the derivative of the linear n¼3

u ¼ ð1Þn  ð1  cos t Þ ! u ¼ ð1Þ3 ð1  cos t Þ
part). According to introduced point in section three the
first boundary condition will be zero and last boundary ¼  1 þ cos t ð28Þ
condition will be equal to one. With the following
t ¼ ð1Þn ðf  tÞ ¼ ð1Þ3 ðf  tÞ ¼ ðt  fÞ ð29Þ
boundary conditions:
Substituting Eq. (29) into Eq. (28)
yð0Þ ¼ 0; y0 ð0Þ ¼ 1
yð0Þ ¼ c1 ð0Þ þ c2 ) c2 ¼ 0 ð18Þ u ¼ 1 þ cosðt  fÞ ð30Þ
y0ð0Þ ¼ c1 ) c1 ¼ 1; y ¼ t
Example 4 Equation (3) from Table (2)
We multiply the answer of the differential equation by 000 00
5u ðtÞ  4u ðtÞ ð31Þ
(-1)n (where n is the highest rank of the derivative of the
linear part). We solve the linear section of Eq. (31):

y ¼ ð1Þ2  t ¼ t; n¼2 ð19Þ 4

5m3  4m2 ¼ 0 ) m2 ð5m  4Þ ¼ 0; m ¼ 0; m ¼
5
t ¼ ð1Þ2 ðs  tÞ ¼ ðs  tÞ ð20Þ ð32Þ
Substituting Eq. (20) into Eq. (19) General solution to the differential equation is as follows:
y ¼ ðs  tÞ: ð21Þ 4t
u ¼ c1 þ tc2 þ c3 e 5 ð33Þ
Example 3 Equation (2) from Table (2)
Because in this example derivative degree of linear differ-
u000 ðtÞ þ u0 ðtÞ ð22Þ ential equation is (3) we conclude that two initial boundary
condition of this equation is zero and the last boundary
We solve the linear section of Eq. (22)
  condition of it is (A1 ). A coefficient of the highest derivative
m3 þ m ¼ 0; m m2 þ 1 ¼ 0 ) m1 ¼ 0; m2 ¼ i is linear. With the following boundary conditions:
ð23Þ 1
uð0Þ ¼ 0; u0 ð0Þ ¼ 0 u00 ð0Þ ¼
General solution to the differential equation is as follows: 5
uð0Þ ¼ c1 þ c3 ) c1 ¼ c3
u ¼ c1 þ c2 cos t þ c3 sin t ð24Þ
4 4t 4 4
u0 ¼ c2 þ c3 e 5 ; u0 ð0Þ ¼ c2 þ c3 ) c2 ¼  c3
With the following boundary conditions: 5 5 5
uð0Þ ¼ 0; u0 ð0Þ ¼ 0; u00 ð0Þ ¼ 1 ð25Þ ð34Þ
16 4t 16 1 16 5
uð0Þ ¼ 0 ) c1 þ c2 cosð0Þ þ c3 sinð0Þ ¼ 0; c1 þ c2 ¼ 0 u00 ¼ c3 e 5 ; u00 ð0Þ ¼ c3 ; ¼ c3 ) c3 ¼
0
u ð0Þ ¼ 0 ) c2 sinð0Þ þ c3 cosð0Þ ¼ 0; c3 ¼ 0 25 25 5 25 16
u00 ð0Þ ¼ 1 ) c2 cosð0Þ  c3 sinð0Þ ¼ 1; c2 ¼ 1; c1 ¼ 1 ð35Þ
ð26Þ 5 1 5 4t
u ¼   t þ e5 ð36Þ
u ¼ 1  cos t ð27Þ 16 4 16
We multiply the answer of the differential equation by
We multiply the answer of the differential equation by
(-1)n
(-1)n

123
J Braz. Soc. Mech. Sci. Eng.

 
5 1 5 4t Substituting Eq. (49) into Eq. (48)
u ¼   t þ e5
16 4 16  
5 1 5 4t 5 T ¼ sinðs  xÞ ð50Þ
n¼3
 ð1Þn ! u ¼   t þ e 5  ð1Þ3 ¼
16 4 16 16 3.1 Description of the second method
1 5 4t
þ t  e5
4 16 For calculating Lagrange multiplier we are doing the fol-
ð37Þ lowing steps:
t ¼ ð1Þ3 ðf  tÞ ¼ ðt  fÞ ð38Þ 1. Linear part of differential equation is solved by
Laplace method and will be put equal (-1)n (where
Substituting Eq. (38) into Eq. (37)
n is the highest order derivative of the linear part.)
1 5 1 5 4f 4t 2. We assume zero the boundary conditions of the
) u ¼  f þ þ t  e 5 þ5 ð39Þ
4 16 4 16 problem.
Example 5 Equation (9) from Table 1 3. Wherever the result of Laplace is obtained paramet-
rically, we change the variable, for example, if the
y0 ¼ 2y  y2 þ 1 ð40Þ result of the Laplace obtained is t, we replace it with
With the following boundary conditions: (-1)n(s - t).

yð0Þ ¼ 1 ð41Þ
3.1.1 Application of the new method
We solve the linear section of Eq. (40):
R
Example 7 Equation (1) from Table 1
y ¼ ce 2dt ¼ ce2t ) yð0Þ ¼ ce0 ) c ¼ 1
ð42Þ
y ¼ e2t y00 þ x2 y ¼ f ðtÞ ð51Þ
We multiply the answer of the differential equation by (- We assume zero as the boundary conditions of the problem
1)n
) yð0Þ ¼ 0; y0 ð0Þ ¼ 0 ð52Þ
n 2t n¼1 1 2t 2t
y ¼ ð1Þ  e !y ¼ ð1Þ : e ¼  e ð43Þ
The Linear part of Eq. (51) is solved by Laplace method
n
t ¼ ð1Þ ðf  tÞ ) y ¼ e ¼  e 2t 2ðtfÞ
ð44Þ and will be put equal (-1)n

Example 6 Equation (3) from Table 1 n¼2

s2 L½ y  syð0Þ  y0 ð0Þ þ x2 L½ y ¼ ð1Þn !
T 00 þ T þ x ¼ 0 ð45Þ
s2 L½ y þ x2 L½ y ¼ 1; ðs2 þ x2 ÞL½ y ¼ 1; ð53Þ
With the following boundary conditions: h i sin xt
y ¼ L1 1 ðs2 þ x2 Þ ¼
Tð0Þ ¼ 0 T 0 ð0Þ ¼ 1 ð46Þ x
t ¼ ð1Þ2 ðs  tÞ ð54Þ
We solve the linear section of Eq. (45):
Substituting Eq. (54) into Eq. (53)
m2 þ 1 ¼ 0; m2 ¼ 1 ) m ¼ i
General solution to the differential equation is as follows: sin xðs  tÞ
)y¼ ð55Þ
x
T ¼ c1 sinð xÞ þ c2 cosð xÞ
Example 8 Equation (7) from Table 2
T ð0Þ ¼ c1 sinð0Þ þ c2 cosð0Þ ) 0 ¼ c2
ð47Þ
T 0 ð0Þ ¼ c1 cosð0Þ; 1 ¼ c1 ) c1 ¼ 1 uð5Þ ðtÞ þ uð4Þ ðtÞ þ u000 ðtÞ ð56Þ
T ¼ sinð xÞ
We assume zero as the boundary conditions of the problem
We multiply the answer of the differential equation by (-
) uð0Þ ¼ 0; u0 ð0Þ ¼ 0; u00 ð0Þ ¼ 0; u000 ð0Þ ¼ 0; uð4Þ ðtÞ ¼ 0
1)n
ð57Þ
T ¼ sinð xÞ  ð1Þ2 ¼ sinð xÞ ð48Þ
The linear part of Eq. (56) is solved by Laplace method and
x ¼ ð1Þ2  ðs  xÞ ¼ ðs  xÞ ð49Þ will be put equal (-1)n

123
 J Braz. Soc. Mech. Sci. Eng.

s5 L½u  s4 uð0Þ  s3 u0 ð0Þ  s2 u00 ð0Þ  su000 ð0Þ  uð4Þ ð0Þ 1. The linear part of differential equation is solved by
4 3 2 0 00
þ s L½u  s uð0Þ  s u ð0Þ  su ð0Þ  u ð0Þ 000 using the Laplace method transform.
2. The boundary conditions are defined as follows:
þ s3 L½u  s2 uð0Þ  su0 ð0Þ  u00 ð0Þ ¼ ð1Þ5 ; s5 L½u
þ s4 L½u þ s3 L½u ¼ 1 yð0Þ; y0 ð0Þ; y00 ð0Þ; . . .; yðn1Þ ð0Þ
 
) L½u s5 þ s4 þ s3 ¼ 1 3. (n - 1) Initial boundary condition will be zero and then
h i  
1 1 1 2 2 pffiffiffi 1t 1 pffiffiffi we put the last boundary condition, the presence of (1).
U¼L s þ s þ s ¼  2 t  3 3e sin 2 3t þ t
5 4 3 2 000
For example y ? y2 = 0 the linear part of equation is
00 0

ð58Þ y and the highest order derivative of the linear part is

(3). According to the point of above: (3 - 1 = 2) after
t ¼ ð1Þ5 ðf  tÞ ¼ ðt  fÞ ð59Þ that the initial boundary conditions equal to zero and
Substituting Eq. (59) into Eq. (58) the last boundary condition is equal to (1) (n is the
  highest rank of the derivative of the linear part).
1 2 pffiffiffi 1f1t 1 pffiffiffi
) U ¼  ðf þ tÞ2  3e2 2 sin 3ðf þ tÞ  f þ t 4. We multiply the result of Laplace to (-1)n (n is the
2 3 2
highest rank of the derivative of the linear part).
ð60Þ 5. Wherever there is a parametric equation, we change
Example 9 Equation (3) from Table 1 the variable. For example if we obtain t Variable then
T 00 þ T þ x ¼ 0 ð61Þ we will put (-1)n(s - t) instead of it.

We assume zero the boundary conditions of the problem

) Tð0Þ ¼ 0; T 0 ð0Þ ¼ 0 ð62Þ 4 Application of the new method
The linear part of Eq. (61) is solved by Laplace method and
will be put equal (-1)n Example 11 Equation (1) from Table 1
y00 þ x2 y ¼ f ðtÞ ð70Þ
s2 L½T   sT ð0Þ  T 0 ð0Þ þ L½T  ¼ ð1Þ2
 
2  1 1 ð63Þ With the following boundary conditions:
L½T  s þ 1 ¼ 1 ) T ¼ L ¼ sin x
s2 þ 1 yð0Þ ¼ 0; y0 ð0Þ ¼ 1 ð71Þ
2
x ¼ ð1Þ ðs  xÞ ¼ ðs  xÞ ð64Þ The linear part of Eq. (70) is solved by using the Laplace
Substituting Eq. (64) into Eq. (63) method transform.

T ¼ sinðs  xÞ ð65Þ s2 L½ y  syð0Þ  y0 ð0Þ þ x2 L½ y ¼ 0

Example 10 Equation (9) from Table 1 s2 L½ y  1 þ x2 L½ y ¼ 0; ðs2 þ x2 ÞL½ y ¼ 1 ) y

h i sin xt
y0 ¼ 2y  y2 þ 1 ð66Þ ¼ L1 1 ðs2 þ x2 Þ ¼ ð72Þ
x
We assume zero the boundary conditions of the problem We multiply the answer of the differential equation by
) yð0Þ ¼ 0 ð67Þ (-1)n
sin xt sin xt
The linear part of Eq. (66) is solved by Laplace method and y ¼ ð1Þ2  ¼ ð73Þ
will be put equal (-1)n x x
t ¼ ð1Þn ðs  tÞ ¼ ð1Þ2 ðs  tÞ ¼ ðs  tÞ ð74Þ
sL½ y  y0 ð0Þ  2L½ y ¼ ð1Þ1
  Substituting Eq. (74) into Eq. (73)
1 ð68Þ
L½ yðs  2Þ ¼ 1 ) y ¼ L1 ¼ e2t
s2 sin xðs  tÞ
)y¼ ð75Þ
t ¼ ð1Þ1 ðf  tÞ ¼ ðt  fÞ x
ð69Þ Example 12 Equation (7) from Table 2
y ¼ e2ðtfÞ
u ðtÞ þ u ðtÞ þ u000 ðtÞ
ð5Þ ð4Þ
ð76Þ
3.2 Description of the third method
With the following initial conditions:
For calculating Lagrange multiplier we are doing the fol- uð0Þ ¼ 0; u0 ð0Þ ¼ 0; u00 ð0Þ ¼ 0; u000 ð0Þ ¼ 0; uð4Þ ðtÞ ¼ 1
lowing steps.
ð77Þ

123
J Braz. Soc. Mech. Sci. Eng.

The linear part of Eq. (76) is solved by using the Laplace The linear part of Eq. (87) is solved by using the Laplace
method transform. method transform
s5 L½u  s4 uð0Þ  s3 u0 ð0Þ  s2 u00 ð0Þ  su000 ð0Þ  uð4Þ ð0Þ sL½ y  yð0Þ  2L½ y ¼ 0
 
4 3 2 0 00
þ s L½u  s uð0Þ  s u ð0Þ  su ð0Þ  u ð0Þ 000 1 ð89Þ
ðs  2ÞL½ y ¼ 1 ) y ¼ L1 ¼ e2t
þ s3 L½u  s2 uð0Þ  su0 ð0Þ  u00 ð0Þ ¼ 0; s5 L½u  1 s2
 
þ s4 L½u þ s3 L½u ¼ 0; L½u s5 þ s4 þ s3 ¼ 1 We multiply the answer of the differential equation by
h i 1   (-1)n
1 1 2 2 pffiffiffi 1t 1 pffiffiffi
U¼L s5 þ s4 þ s3 ¼ 2 t þ 3 3e sin 2 3t  t
2

y ¼ ð1Þ1 e2t ¼ e2t ð90Þ

ð78Þ
t ¼ ð1Þ1 ðf  tÞ ¼ ðt  fÞ ð91Þ
We multiply the answer of the differential equation by
Substituting Eq. (91) into Eq. (90)
(-1)n
  y ¼ e2ðtfÞ ð92Þ
1 2 pffiffiffi 1t 1 pffiffiffi
U ¼ ð1Þ5  t2 þ 3e 2 sin 3t  t
2 3  2
 5 Results
1 2 2 pffiffiffi 1t 1 pffiffiffi
¼ t  3e 2 sin 3t þ t ð79Þ
2 3 2 In order to check the new methods, we compared the results
of many articles with k that obtain from our solution dif-
t ¼ ð1Þ5 ðf  tÞ ð80Þ
ferential equations. Some of these comparisons have been
Substituting Eq. (80) into Eq. (79) collected into Table 1. Also the reference of each equation
  is indicated. These methods are able to obtain k from
1 2 pffiffiffi 1f1t 1 pffiffiffi
) k ¼  ðf þ tÞ2  3e2 2 sin 3ðf þ tÞ  f nonlinear differential equation, nonlinear partial differential
2 3 2
equation (PDE) and nonlinear system differential equation.
þt
At another approach, we tested the solution differential
equations with some higher order equations and respective
Example 13 Equation (3) from Table 1
k to each equation was calculated readily. The results of
00
T þT þx¼0 ð81Þ these tests have been listed in Table 2.
With the following initial conditions:
) Tð0Þ ¼ 0; T 0 ð0Þ ¼ 1 ð82Þ 6 Conclusions
The linear part of Eq. (81) is solved by using the Laplace
In this study, the Lagrange multiplier (k) was achieved by
method transform.
using new methods without any intricate manual calcula-
s2 L½T   sT ð0Þ  T 0 ð0Þ þ L½T  ¼ 0 tions, like traditional method. These methods are able to
  compute k by writing program on computer. Therefore, the
2  1 1 ð83Þ
L½T  s þ 1 ¼ 1 ) T ¼ L ¼ sin x solution differential equation to obtain k has advantages
s2 þ 1
that listed below:
We multiply the answer of the differential equation by
(-1)n 1. Achieving exact k without any error.
2. For cases encountering higher order nonlinear equa-
T ¼ ð1Þ2  sin x ¼ sin x ð84Þ tions. Calculation of k is too complicated manually. By
means of present solution differential equations,
x ¼ ð1Þ2 ðs  xÞ ¼ ðs  xÞ ð85Þ
Lagrange multiplier is acquired readily.
Substituting Eq. (85) into Eq. (84) 3. The three methods presented in this article are easy
than previous methods.
T ¼ sinðs  xÞ ¼ sinðs  xÞ ð86Þ
Example 14 Equation (9) from Table 1
0 2
y ¼ 2y  y þ 1 ð87Þ References
With the following initial conditions: 1. He JH (1999) Variational iteration method a kind of non-linear
analytical technique: some examples. Int J Nonlinear Mech
) yð0Þ ¼ 1 ð88Þ 34:699–708

123
 J Braz. Soc. Mech. Sci. Eng.

2. He JH (2000) Variational iteration method for autonomous 15. Wazwaz AM (2007) The variational iteration method for solving
ordinary differential systems. Comput Math Appl 114(2–3): a system of PDES. Comput Math Appl 54:895–902
115–123 16. Ganji DD, Tari H, Babazadeh H (2007) The application of He’s
3. He JH, Wu XH (2006) Construction of solitary solution and variational iteration method to nonlinear equations arising in heat
compaction-like solution by variational iteration method. Chaos transfer. Phys Lett A 363:213–217
Solitons Fractals 29(1):108–113 17. Rafei M, Daniali H, Ganji DD (2007) Variational iteration
4. He JH (2007) Variational iteration method—some recent results method for solving the epidemic model and the prey and predator
and new interpretations. J Comput Appl Math 207:3–17 problem. Appl Math Comput 186:1701–1709
5. He JH (1997) Variational iteration method for delay differential 18. Tari H, Ganji DD, Rostamian M (2007) Approximate solution of
equations. Commun Nonlinear Sci Numer Simul 2(4):235–236 K (2, 2), KdV and modified KdV equations by variational itera-
6. He JH (2006) Some asymptotic methods for strongly nonlinear tion method, homotopy perturbation method and homotopy ana-
equations. Int J Mod Phys B 20(10):1141–1199 lysis method. Int J Nonlinear Sci Numer Simul 8:203–210
7. He JH, Wu XH (2007) Variational iteration method: new 19. Ganji DD, Sadighi A (2006) Application of He’s methods to
development and applications. Comput Math Appl 54:881–894 nonlinear coupled systems of reactions. Int J Nonlinear Sci
8. He JH (2007) The variational iteration method for eighth-order Numer Simul 7(4):411–418
initial boundary value problems. Phys Scr 76(6):680–682 20. Ganji DD, Jannatabadi M, Mohseni E (2006) Application of He’s
9. Wazwaz AM (2002) Partial differential equations: methods and variational iteration method to nonlinear Jaulent–Miodek equa-
applications. Balkema, Rottesdam tions and comparing it with ADM. J Comput Appl Math 207:35–45
10. Tatari M, Dehghan M (2007) He’s variational iteration method 21. Ganji DD, Tari H, Jooybari MB (2007) Variational iteration
for computing a control parameter in a semi-linear inverse par- method and homotopy perturbation method for evolution equa-
abolic equation. Chaos Solitons Fractals 33(2):671–677 tions. Comput Math Appl 54:1018–1027
11. Wazwaz AM (2006) A study on linear and nonlinear Schrodinger 22. Abbasbandy S (2007) A new application of He’s variational
equations by the variational iteration method. Chaos Solitons iteration method for quadratic Riccati differential equation by
Fractals. doi:10.1016/j.chaos.10.009 using Adomian’s polynomials. J Comput App Math 207:59–63
12. Wazwaz AM (2007) The variational iteration method for solving 23. Khader MM (2012) Numerical and theoretical treatment for
tow forms of Blasius equation on a half-infinite domain. Appl solving linear and nonlinear delay differential equations using
Math Comput 188(1):485–491 variational iteration method. Arab J Math Sci
13. Wazwaz AM (2007) The variational iteration method for the
exact solution of Laplace equation. Phys Lett A 363:260–262
14. Wazwaz AM (2001) The numerical solution of fifth-order
boundary value problems by the decomposition method. J Com-
put Appl Math 136:259–270

123