Professional Documents
Culture Documents
On
Partial Differential Equation
Differential equation (DE)
An equation involves derivatives of one or more dependent variables with respect to one or more
independent variables is said to be a differential equation.
𝑑2 𝑦 𝑑𝑦
Example: 𝑑𝑥 2 + 2𝑦 = 4𝑦
𝑑𝑥
2
Linear PDE
A partial differential equation is called linear if
i. The power of the dependent variable and each partial derivatives contained in the
equation is one
ii. The co-efficient of the dependent variable and each partial derivatives are constants or
independent variables
Example: y11+py1+qy=0
Non-Linear PDE: A PDE which is not linear is called a non-linear PDE.
Example:𝑦𝑦" + 𝑦𝑦′ = 0
Homogenous PDE: If all the terms of a PDE contains the dependent variable or its partial
derivatives, then such a PDE is called homogenous PDE.
Example:𝑢𝑥 + 𝑦𝑢𝑦 = 3𝑢
Inhomogeneous PDE: A PDE which is not homogeneous is called inhomogeneous PDE.
Example:𝑢𝑥 + 𝑢𝑦 = 𝑥 2 + 4𝑥𝑦 + 𝑦 2
Order and Degree: The largest derivative present in the differential equation is called the order
of the PDE.
The degree of the highest order derivative occurs in the DE.
Example:𝑢𝑥 + 𝑢𝑦 = 0, first order.
𝑢𝑥𝑥 + 𝑢𝑦 = 0, second order.
𝑑2 𝑥 𝑑𝑥
And 𝑑𝑡 2 + 𝑑𝑡 = 0, first order.
𝑑2 𝑥 𝑑𝑥
( 𝑑𝑡 2 )2 + 𝑑𝑡 = 0, second
Initial value problem: An initial value problem is a DE together with a point in the domain of f.
Boundary value problem: A boundary value problem is a DE together with a set of additional
constraints, called boundary conditions.
General Solution: The general solution to a differential equation is the most general form that
the solution can take doesn’t take any initial condition into account.
Actual Solution: The specific solution that not only satisfies the DE but also satisfies the given
initial conditions.
Particular Solution: The particular solution is obtained by assigning particular value to the
arbitrary constants in the general solution.
Series Solution: Series solution to a DE is such a solution that assumes a power series with
unknown co-efficient.
3
Closed form Solution: An equation is said to be a closed form solution if it solves a given
problem in terms of functions and mathematical operations from a given generally accepted sets.
Example: The solution of any quadratic equation with complex coefficient can be expressed in
closed form in terms of addition, substraction, multiplication, division and square root extraction,
each of which is an elementary function. The quadratic equation,
𝑎𝑥 2 + 𝑏𝑥 + 𝑐 = 0,
−𝑏 ± √𝑏 2 − 4𝑎𝑐
𝑥=
2𝑎
Travelling wave solution: The KdV equation posses travelling wave solutions. One particular
travelling wave solutions and it was discovered experimentally by John Scott Russel I 1834.
Decomposition method: The adomian decomposition method(ADM) is a powerful method
which considers the approximate solution of a non linear equation as an infinite series which
usually converges to the exact solution.It was developed and introduced by George Adomian and
this method is applied to a wide class of linear and non-linear ODE and PDE and integral
equations.It consists of decomposing the unknown function u(x,y) of any equation into a sum of
an infinite number of components define by the decomposition series,
∞
𝑢(𝑥, 𝑦) = ∑ 𝑢𝑛 (𝑥, 𝑦)
𝑛=0
Where the function f represents the terms arising from integrating the source term g and from
using the given conditions that are assumed to be prescribed. Adomian method defines the
solution u by an infinite series of components given by
∝
𝑢 = ∑ 𝑢𝑛 … … … … … … … . . (𝑖𝑖 )
𝑛=0
4
∝ ∝
−1
∑ 𝑢𝑛 = 𝑓 − 𝐿 (𝑅 (∑ 𝑢𝑛 )) … … … … … (𝑖𝑖𝑖)
𝑛=0 𝑛=0
Applying 𝐿−1
𝑥 to both sides,
𝐿−1 −1 −1
𝑥 𝐿𝑥 𝑢 = 𝐿𝑥 (𝑥 + 𝑦) − 𝐿𝑥 (𝐿𝑦 𝑢)
1
Or equivalently, 𝑢(𝑥, 𝑦) = 𝑢(0, 𝑦) + 2 𝑥 2 + 𝑥𝑦 − 𝐿−1
𝑥 (𝐿𝑦 𝑢)
1 2
= 𝑥 + 𝑥𝑦 − 𝐿−1
𝑥 (𝐿𝑦 𝑢) … … … … … … … (𝑖)[𝑢(0, 𝑦)]
2
The decomposition method sets the solution u(x,y) in a series form by,
∝
5
∝ ∝
1
∑ 𝑢𝑛 (𝑥, 𝑦) = 𝑥 2 + 𝑥𝑦 − 𝐿−1
𝑥 (𝐿𝑦 (∑ 𝑢𝑛 (𝑥, 𝑦)))
2
𝑛=0 𝑛=0
1
𝑢0 + 𝑢1 + 𝑢2 + ⋯ … … … … . = 𝑥 2 + 𝑥𝑦 − 𝐿−1
𝑥 (𝐿𝑦 (𝑢0 + 𝑢1 + 𝑢2 + ⋯ … ))
2
The decomposition method identifies the zeroth component 𝑢0 by all terms arising from the given
condition and from
Integrating 𝑓(𝑥, 𝑦) = 𝑥 + 𝑦 , therefore we set ,
1 2
𝑢0 (𝑥, 𝑦) = 𝑥 + 𝑥𝑦
2
𝑢𝑘+1 (𝑥, 𝑦) = −𝐿−1
𝑥 (𝐿𝑦 (𝑢𝑘 )),k≥0
𝑢(𝑥, 𝑦) = 𝑦 + 𝐿−1
𝑥 (𝐿𝑦 𝑢)(2)[𝑢(0, 𝑦) = 0]
6
∞
∑ 𝑢 (𝑥, 𝑦) = 𝑦 + 𝐿−1
𝑥 (𝐿𝑦 𝑢)
𝑛=0 𝑛
⇒ 𝑢0 +𝑢1 + 𝑢2 + ⋯ = 𝑦 + 𝐿−1
𝑥 (𝐿𝑦 (𝑢0 + 𝑢1 + 𝑢2 + ⋯ ))
𝑢1 (𝑥, 𝑦) = 𝐿−1 −1
𝑥 (𝐿𝑦 𝑢0 ) = 𝐿𝑥 (𝐿𝑦 (𝑦)) = 𝑥
𝑢1 (𝑥, 𝑦) = 𝐿−1 −1
𝑥 (𝐿𝑦 𝑢1 ) = 𝐿𝑥 (𝐿𝑦 (𝑥)) = 0
1. Use the decomposition method and the noise terms phenomenon to solve the
following inhomogeneous PDE
ux+ uy=(1+x)ey , u(0,y) = 0, u(x,0) = x
𝑥2
∑∞
𝑛=0 𝑢𝑛 (𝑥, 𝑦)= (x + ) ey - Lx-1 Ly(∑∞
𝑛=0 𝑢𝑛 (𝑥, 𝑦))
2!
7
𝑥2
u0+ u1+ u2 + ……………= (x + )ey - Lx-1 Ly (u0+ u1+ u2 + ……)
2!
𝑥2 𝑥3
u1(x,y) = Lx-1 Ly (u0) = -( 2! + 3! ) ey
𝑥3 𝑥4
u2(x,y) = Lx-1 Ly (u1) = ( 3! + 4! ) ey
𝑥2 y
Considering the first two components uo and u1, it is easily observed that the noise terms e and
2!
𝑥2 y
- e appear in u0 and u1 respectively. By canceling the noise terms, we find that the exact
2!
solution is given by,
u (x,y) = xey
2. Use the decomposition method and the noise terms phenomenon to solve the following
PDE,
𝑥3
∑∞
𝑛=0 𝑢𝑛 (𝑥, 𝑦) = + 2yx2 + y2x - Lx-1 Ly(∑∞
𝑛=0 𝑢𝑛 (𝑥, 𝑦))
3
𝑥3
equivlently, u0+ u1+ u2 + ……………= 3 + 2yx2 + y2x - Lx-1 Ly(u0+ u1+ u2…….)
𝑥3
𝑢0 = + 2𝑦𝑥 2 + 𝑦 2 𝑥
3
𝑥3
𝑢1 = −𝐿−1
𝑥 𝐿𝑦 ( + 2𝑦𝑥 2 + 𝑦 2 𝑥)
3
8
=−𝐿−1 2
𝑥 (2𝑥 + 2𝑦𝑥)
𝑥3 𝑥2
= − (2 + 2 𝑦)
3 2
2
= − ( 𝑥 3 + 𝑥 2 𝑦)
3
𝑥3
𝑢2 = 𝐿−1
𝑥 𝐿𝑦 (2 + 𝑥 2 𝑦)
3
= 𝐿−1 2
𝑥 (𝑥 )
𝑥3
=
3
𝑥3
𝑢3 = 𝐿−1
𝑥 𝑦𝐿 ( )=0
3
𝑢𝑘 = 0, 𝑘 ≥ 0.
We can easily observe that the two components 𝑢0 and 𝑢1 do not contain noise terms.
Hense, 𝑢(𝑥, 𝑦) = 𝑢0 + 𝑢1 + 𝑢2 + ⋯ … … … … ..
1 3 2 1
= 𝑥 + 2𝑥 2 𝑦 + 𝑥𝑦 2 − 𝑥 2 𝑦 − 𝑥 3 + 𝑥 3 = 𝑥𝑦 2 + 𝑥 2 𝑦
3 3 3
The Modified Decomposition method: The modified decomposition method was developed by
Wazwaz which will accelerate the convergence of the series solution. It can be applied to all
PDE’s of any order.
Description of the method: Consider, the partial differential equation in an operator form
𝐿𝑢 + 𝑅𝑢 = 𝑔
Where, L is the highest order derivative, R is a linear differential operator of less order or equal
order to L, any g is the source term. Operating with the inverse operator L-1,
u=f-L-1(Ru)
Where, f represents the terms arising from the given initial condition and from integrating the
source term g. Defining the decomposition series,
∝
𝑢 = ∑ 𝑢𝑛
𝑛=0
9
Which admits the use of the recursive relation,
U0=f,
Uk+1=-L-1(Ruk), k≥0
The modified decomposition method introduces a slight variation to the recursive relation that
will lead to the determination of the components of u in a faster and easier way. We can set,
𝑓 = 𝑓1 + 𝑓2
We identify the zeroth component 𝑢0 by one part of f, namely 𝑓1 𝑜𝑟 𝑓2 . The other part of f can
be added to component 𝑢1 among other terms. The modified recursive relation can be identified
by,
𝑢0 = 𝑓1
𝑢1 = 𝑓2 − 𝐿−1 (𝑅𝑢0 )
𝑢𝑘+1 = −𝐿−1 (𝑅𝑢𝑘 ), 𝑘 ≥ 1
1. Use the modified decomposition method to solve the first order partial differential
equation:
𝒖𝒙 +𝒖𝒚 = 𝟑𝒙𝟐 𝒚𝟑 + 𝟑𝒙𝟑 𝒚𝟐 , 𝒖(𝟎, 𝒚) = 𝟎.
𝒖𝟎 (x,y)=𝒙𝟑 𝒚𝟑
3
𝑢1 (x,y)=4 𝑥 4 𝑦 2-𝐿𝑥 −1 (𝑢0)𝑦
3
=4 𝑥 4 𝑦 2 -𝐿𝑥 −1 (3𝑥 3 𝑦 2 )
3 3
=4 𝑥 4 𝑦 2 -4 𝑥 4 𝑦 2
=0
10
𝑢𝑘+1 (x,y)=0, k≥ 1
Then the exact solution is,
u(x,y)=𝑥 3 𝑦 3
2.Use the modified decomposition method to solve the first order partial differential
𝟏
equation: 𝒖𝒙 -𝒖𝒚 =𝒙𝟑 - 𝒚𝟑 , u(0,y)=𝟒 𝒚𝟒
Solution: In an operator form ,
𝐿𝑥 (u)=𝑥 3 -𝑦 3 +𝐿𝑦 (u)
Applying the inverse operator 𝐿𝑥 −1 on both sides,
𝑥4 𝑦4
u(x,y)= 4 + 4 -x𝑦 3 +L𝑥 −1 𝐿𝑦 (u)
𝑥4 𝑦4
we get, 𝑢0 +𝑢1 +.........= 4 + 4 -x𝑦 3 +L𝑥 −1 𝐿𝑦 (𝑢0 +𝑢1 +...........)
1
=4(𝑥 4 +𝑦 4 )-x𝑦 3 +𝐿𝑥 −1 𝐿𝑦 (𝑢0 +𝑢1 +..........)
𝑓2 (x,y)=-x𝑦 3
Consequently, We set the modified recursive relation,
1
𝑢0 (x,y)=4(𝑥 4 +𝑦 4 )
1
𝑢1 (x,y)=-x𝑦 3 +𝐿𝑥 −1 (4(𝑥 4 +𝑦 4 ))
=-x𝑦 3 +𝐿𝑥 −1 (𝑦 3 )
=-x𝑦 3 +x𝑦 3
=0
𝑢𝑘+1 (x,y)=0, k≥1
1 1
The exact solution, u(x,y)=4 𝑥 4 +4 𝑦 4
11
3.Use the modified decomposition method to solve the first order partial differential
equation:
𝒖𝒙 +𝒖𝒚 =coshx+coshy, u(x,0)=sinhx
Solution: We can rewrite in an operator form,
𝐿𝑦 u=coshx+coshy-𝑢𝑥
Applying the universe operator 𝐿𝑦 −1 ,
u(x,y)=sinhx+sinhy+ycoshx-𝐿𝑦 −1 𝐿𝑥 (u)
The function f(x,y) can be wtitten as,
𝑓1 (x,y)=sinhx+sinhy
𝑓2 (x,y)=ycoshx
The decomposition series,
𝛼
𝑢(𝑥, 𝑦) = ∑ 𝑢𝑛 (𝑥, 𝑦)
𝑛=0
Then
𝑢0 + 𝑢1 + 𝑢2 … … … … … … … = 𝑠𝑖𝑛ℎ𝑥 + 𝑠𝑖𝑛ℎ𝑦 + 𝑦𝑐𝑜𝑠ℎ𝑥 − 𝐿−1
𝑦 (𝑢0 + 𝑢1 + 𝑢2 … … )𝑥
𝑢𝑘+1 = 0, 𝑘≥1
12
The variational iteration method
Introduction: The variational iteration method established by Ji-Huan He, is thoroughly used by
mathematicians to handle a wide variety of scientific and engineering applications: linear and non-
linear, and homogeneous and inhomogeneous as well. It was shown that this method is effective
and reliable for analytic and numerical purpose. This method gives rapidly convergent successive
approximations of the exact solution if such a solution exists. The VIM does not require specific
treatments for nonlinear problems as in Adomian method and others. In what follows, we present
the main steps of the method.
Where L and N are linear and nonlinear operators respectively, and g(t) is the source
inhomogeneous term.
The VIM method presents a correction functional for Eq. (1) in the form
t
un 1 t un t Lun Nun g d (2)
0
Where is a general Lagrange multiplier, which can be identified optimally via the variational
theory, and un is a restricted variation which means un 0.
It is obvious now that the main steps of the He’s variational iteration method require first.
The determination of the Lagrange multiplier that will be identified optimally. Integration
by parts is usually used for the determination of the Lagrange multiplier . In other words, we
can use
u d u u d
n n n
u d u u u d
n n n n (3)
and soon. The last two identities can be obtained by integrating by parts.
13
Having determined the lagrange multiplier , the successive approximations un 1 , n 0 of the
solution u will be readily obtained upon using any selective function u0 . Consequently, the
solution
limn un (4)
In other words, the correction function (2) will give several approximations, and therefore the exact
solution is obtained as the limit of the resulting successive approximations.
Example 1: Use variational iteration method to solve the following inhomogeneous PDE
𝑼 𝒙 + 𝑽𝒚 = 𝒙 + 𝒚 U(0,y) = 0 U(x,0) = 0
𝑈𝑥 + 𝑉𝑦 = 𝑥 + 𝑦 ...........(I)
The correction functional for eq. (I) is given by
𝑥 𝛿𝑈𝑛 (𝜉,𝑦) 𝛿 2 𝑈𝑛 (𝜉,𝑦)
𝑈𝑛+1 (x,y) = 𝑈𝑛 (x,y) + ∫0 𝜆(𝜉) ( + -𝜉-y) d 𝜉.........(II)
𝛿𝜉 𝛿𝑦
As stated before we can select 𝑈0 (x,y) = U (0,y) = 0 from the given conditions. using this selection
into (III) we obtain the successive approximations
𝑈0 (x,y) = 0
𝑥 𝛿𝑈 𝛿𝑈
𝑈1 (x,y) = 0-∫0 (( 𝛿𝜉0 + 𝛿𝑦0 - 𝜉-y) d 𝜉)
𝑥
= ∫0 (0 + 0 − 𝜉-y) d 𝜉
𝑥2
= + 𝑥𝑦
2
14
𝑥2 𝑥 𝛿𝑈 𝛿𝑈1
𝑈2 (x,y) = ( + xy)- ∫0 ( 𝛿𝜉1 + − 𝜉-y) d 𝜉 = xy
2 𝛿𝑦
𝑥 𝛿𝑈 𝛿𝑈2
𝑈3 (x,y) = xy - ∫0 ( 𝛿𝜉2 + − 𝜉-y) dξ = xy
𝛿𝑦
.
.
.
𝑈𝑛 (x,y) = xy
𝑼𝒙 − 𝑼𝒚 = 0, U(0,y) = y , U(x,0) = x
15
Approximation u0(x,y)=y
𝑥 𝜕𝑢 𝜕𝑢0
u1(x,y) = y- ∫0 ( 𝜕𝜁0 − ) 𝑑𝜁
𝜕𝑦
𝑥
=y- ∫0 (0 − 1)𝑑𝜁
=y+x
=x+y.
𝑥 𝜕𝑢 𝜕𝑢1
u2(x,y)=x+y- ∫0 ( 𝜕𝜁1 − ) 𝑑𝜁
𝜕𝑦
=x+y
.
.
.
un(x,y)=x+y.
The VIM admits the use of u(x,y)= lim 𝑢𝑛 (𝑥, 𝑦) that gives the exact solution by u(x,y)=x+y.
𝑛→∞
Method of characteristics
Introduction: The first mathematician (who used the methods of characteristics for differential
equations) seems to be Paul de villecourt (June 30,1784).
will be investigated by using the traditional method of characteristics. It is important to note that
a,b and f depend on x,y and u but not on the derivative of u. In addition, we also assume that a,b
and f are continuously differentiable of their arguments.
Assuming that u(x,y) is a solution of (1), then by using the chain rule we obtain
𝑑𝑥 𝑑𝑦 𝑑𝑢
𝑎
= 𝑏
= 𝑓(𝑥,𝑦)+𝑘𝑢 …………(3)
16
𝑑𝑥 𝑑𝑦
The pair =
𝑎 𝑏
⇒ bx-ay=c …………...(4)
Where, c is a constant.
we next consider the pair
𝑑𝑥 𝑑𝑢
= 𝑓(𝑥,𝑦)+𝑘𝑢
𝑎
𝑑𝑥 𝑑𝑢
⇒ = 𝑏𝑥−𝑐
𝑎 𝑓(𝑥, )+𝑘𝑢
𝑎
𝑑𝑦 𝑘 1 𝑏𝑥−𝑐
⇒𝑑𝑥 − 𝑎 𝑢 = 𝑎 𝑓 (𝑥, )……. (5)
𝑎
Which is a first order linear ODE. The integrating factor of (5) is given by
𝑘 𝑘
I.F = 𝑒 ∫ −𝑎𝑑𝑥 =𝑒 −𝑎𝑥
Example 1: Use the method of characteristics to solve the first order Partial differential
equation𝑼𝒙 +𝑼𝒚 =x+y,U(x,0)=0
𝒅𝒙 𝒅𝒚 𝒅𝒖
𝟏
= 𝟏 =𝒙+𝒚
17
Then dx=dy
𝒅𝒖
and dy=𝒙+𝒚
=>du=(x+y)dy
=>du=(2y+c)dy
Where g(x-y) is an arbitrary function. To determine g(x,y).We substitute the given condition into
(3), then U(x,0)=0+0+g(x-0)
Then g(x)=0
Then U(x,y)=xy.
Example 2: Use the method of characteristics to solve the first order partial differential
equation x𝒖𝒙 +𝒖𝒚 =xsinhy +u,u(0,y)=0
𝒅𝒙 𝒅𝒚 𝒅𝒖
From (1) and (2), = = 𝒙𝒔𝒊𝒏𝒉𝒚+𝒖
𝒙 𝟏
𝑑𝑢 𝑑𝑦
Then =
𝑥 1
=> ln 𝑥=y+ln 𝑐
18
𝑥
=> 𝑐 =𝑒 𝑦
Then x=c𝑒 𝑦
𝑑𝑢
And 𝑥𝑠𝑖𝑛ℎ𝑦+𝑢=dy
𝑑𝑢
=>dy=𝑥𝑠𝑖𝑛ℎ𝑦+𝑢
𝑑𝑢
=> 𝑑𝑦=xsinhy+u
𝑑𝑢
=> 𝑑𝑦-u=xsinhy…………(3)
U(0,y)=0+𝑒 0 g(c)
Then g(c) =0
Introduction: Systems of PDEs, linear or nonlinear have attracted much concern in studying
evolution equations that describe wave propagation in investigating shallow water wave and in
examining the endemial relation-diffusion model for Brusselator. The general ideas and the
essential features of this systems are of wide applicability. To solve the systems of linear PDEs.
Adomian decomposition method is more easy and simple. For simplicity reasons, we will use
Adomian method to solve this section.
19
Methodology: Consider the systems of linear PDEs written in an operator form as
Lt u Lx v g1
(1)
Lt v Lx u g2
With initial data
u x, 0 f1 x
(2)
v x, 0 f 2 x
Where Lt and Lx are considered, without loss of generality first order partial differential operator
and g1 and g 2 are in homogeneous term. Now applying inverse operator into (1) and using initial
condition,
u x, t f1 x Lt 1 g1 Lt 1 Lx v
(3)
v x, t f 2 x Lt 1 g 2 Lt 1 Lx u
The adomian method suggested that the linear terms u x, t and v x, t be decomposed by an
u x, t f x L
n 0
n 1 t
1
g1 Lt 1 ( Lx vn x, t )
n 0
(5)
v x, t f x L
n 0
2 t
1
g 2 Lt ( Lx un x, t )
1
n 0
20
Equation (6) and (7) will give the solution of the system u x, t and v x, t respectively.
𝐿 + 𝑣 + 𝐿𝑥 𝑢 = 0
𝑜𝑟 𝐿 + 𝑢 = −𝐿𝑥 𝑣
𝐿 + 𝑣 = −𝐿𝑥 𝑢
Taking 𝐿−1
𝑡 to both sides of equation 1,
𝐿−1 −1
𝑡 (𝐿 + 𝑢) = −𝐿𝑡 𝐿𝑥 𝑣
𝐿−1 −1
𝑡 (𝐿 + 𝑣) = −𝐿𝑡 𝐿𝑥 𝑢
𝑜𝑟 𝑢(𝑥, 𝑡) = 𝑒 𝑥 − 𝐿−1
𝑡 𝐿𝑥 𝑣
𝑣(𝑥, 𝑡) = 𝑒 −𝑥 − 𝐿−1
𝑡 𝐿𝑥 𝑢
𝑢(𝑥, 𝑡) = ∑ 𝑢𝑛 (𝑥, 𝑡)
𝑛=0
∞
𝑣(𝑥, 𝑡) = ∑ 𝑣𝑛 (𝑥, 𝑡)
𝑛=0
∞ ∞
∑ 𝑢𝑛 (𝑥, 𝑡) = 𝑒 𝑥 − 𝐿−1
𝑡 𝐿𝑥 ∑ 𝑣𝑛 (𝑥, 𝑡)
𝑛=0 𝑛=0
∞ ∞
−𝑥
∑ 𝑣𝑛 (𝑥, 𝑡) = 𝑒 − 𝐿−1
𝑡 𝐿𝑥 ∑ 𝑢𝑛 (𝑥, 𝑡)
𝑛=0 𝑛=0
𝑜𝑟 (𝑢0 + 𝑢1 + ⋯ … … … … … … ) = 𝑒 𝑥 − 𝐿−1
𝑡 𝐿𝑥 (𝑣0 + 𝑣1 + ⋯ … … … … … … . . )
21
𝑣0 + 𝑣1 + ⋯ … … … … … … . ) = 𝑒 −𝑥 − 𝐿−1
𝑡 𝐿𝑥 (𝑢0 + 𝑢1 + ⋯ … … … … … … … )
By decomposition method, the zeroth component of the linear systems are 𝑢0 (𝑥, 𝑡) = 𝑒 𝑥
𝑣0 (𝑥, 𝑡) = 𝑒 −𝑥
𝑎𝑛𝑑 𝑢𝑘+1 (𝑥, 𝑡) = −𝐿−1
𝑡 𝐿𝑥 𝑣𝑘 , 𝑘 ≥ 0
𝑢1 (𝑥, 𝑡) = −𝐿−1 −𝑥
𝑡 𝐿𝑥 𝑣0 = 𝑒 𝑡
𝑡2
𝑢2 (𝑥, 𝑡) = −𝐿−1
𝑡 𝐿𝑥 𝑣1 =𝑒𝑥
2!
𝑡3
𝑈3 (𝑥, 𝑡) = −𝐿𝑡 −1 𝐿𝑥 𝑉2 = 𝑒 −𝑥
3!
𝑡4
𝑈4 (𝑥, 𝑡) = −𝐿𝑡 −1 𝐿𝑥 𝑉3 = 𝑒 −𝑥
4!
⋮
𝑉1 (𝑥, 𝑡) = −𝐿𝑡 −1 𝐿𝑥 𝑈0 = −𝑡𝑒 𝑥
𝑡2
𝑉2 (𝑥, 𝑡) = −𝐿𝑡 −1 𝐿𝑥 𝑈1 = 𝑒 −𝑥
2!
𝑡4
𝑉3 (𝑥, 𝑡) = −𝐿𝑡 −1 𝐿𝑥 𝑈3 = 𝑒 −𝑥
4!
⋮
Now 𝑈(𝑥, 𝑡) = 𝑈0 + 𝑈1 + 𝑈2 + 𝑈3+⋯⋯
𝑡2 𝑡3
𝑈(𝑥, 𝑡) = 𝑒 𝑥 + 𝑒 −𝑥 𝑡 + 𝑒 𝑥 + 𝑒 −𝑥 + ⋯
2! 3!
𝑡2 𝑡4 𝑡3
𝑈(𝑥, 𝑡) = 𝑒 𝑥 (1 + + ) + 𝑒 −𝑥 (𝑡 + + ⋯ )
2! 4! 3!
and 𝑉(𝑥, 𝑡) = 𝑉0 + 𝑉1 + 𝑉2 + 𝑉3 + ⋯
𝑡2 𝑡3
𝑈(𝑥, 𝑡) = 𝑒 −𝑥 − 𝑡𝑒 𝑥 + 𝑒 −𝑥 − 𝑒𝑥 + ⋯
2! 3!
𝑡2 𝑡4 𝑡3
𝑈(𝑥, 𝑡) = 𝑒 −𝑥 (1 + + + ⋯ ) − 𝑒 𝑥 (𝑡 + + ⋯ )
2! 4! 3
Hence the exact solution is given by
22
𝑈(𝑥, 𝑡) = 𝑒 𝑥 𝐶𝑜𝑠ℎ𝑡 + 𝑒 −𝑥 𝑆𝑖𝑛ℎ𝑡
𝑉(𝑥, 𝑡) = 𝑒 −𝑥 𝐶𝑜𝑠ℎ𝑡 − 𝑒 𝑥 𝑆𝑖𝑛ℎ𝑡
(𝑈, 𝑉) = (𝑒 𝑥 𝐶𝑜𝑠ℎ𝑡 + 𝑒 −𝑥 𝑆𝑖𝑛ℎ𝑡, 𝑒 −𝑥 𝐶𝑜𝑠ℎ𝑡 − 𝑒 𝑥 𝑆𝑖𝑛ℎ𝑡)
𝑈𝐿 (𝑥, 𝑡) = −2𝐿−1
𝑡 𝐿𝑥 𝑈0 = −𝑡𝑆𝑖𝑛𝑥
𝑡2
𝑢2 (𝑥, 𝑡) = −2𝑙𝑡−1 𝑣1 − 2𝑙𝑡−1 𝑙𝑥 𝑢1 = cos(𝑥)
2!
𝑡3
𝑢3 (𝑥, 𝑡) = −2𝑙𝑡−1 𝑣2 − 2𝑙𝑡−1 𝑙𝑥 𝑢2 = cos(𝑥)
3!
23
.
.
.
𝑣1 (𝑥, 𝑡) = 2𝑙𝑡−1 𝑢0 − 2𝑙𝑡−1 𝑙𝑥 𝑣0 = 𝑡 cos(𝑥)
𝑡2
𝑣2 (𝑥, 𝑡) = 2𝑙𝑡−1 𝑢1 − 2𝑙𝑡−1 𝑙𝑥 𝑣1 = − s 𝑖𝑛(𝑥)
2!
𝑡3
𝑣3 (𝑥, 𝑡) = 2𝑙𝑡−1 𝑢2 − 2𝑙𝑡−1 𝑙𝑥 𝑣2 = − cos(𝑥)
3!
.
.
.
Now 𝑢(𝑥, 𝑡) = 𝑢0 + 𝑢1 + 𝑢3 + ⋯
𝑡2 𝑡3
= cos(𝑥) (1 − + ⋯ ) − sin(𝑥) (𝑡 − + ⋯ )
2! 3!
And
𝑡2 𝑡3
𝑢(𝑥, 𝑡) = sin(x) (1 − + ⋯ ) + 𝑐𝑜𝑠(𝑥) (𝑡 − + ⋯ )
2! 3!
The pair (u,v) is known in a closed form by
(𝑢, 𝑣) = (cos(𝑥 + 𝑡) , sin(𝑥 + 𝑡))
By obtained upon using Taylor and trigonometric identities
Methodology: In this section we will apply the variational iteration method for solving systems
of linear partial differential equations. We write a system in an operator form by
𝑙 + 𝑢 + 𝑅1 (𝑢. 𝑣) = 𝑔1
𝑙 + 𝑢 + 𝑅2 (𝑢. 𝑣) = 𝑔2
where u = u(x,t), with initial data
𝑢(𝑥, 0) = 𝑓1 (𝑥)
𝑢(𝑥, 0) = 𝑓2 (𝑥)
24
Where Lt is considered a first order partial differential operator and Rj,1 ≤ 𝑗 ≤ 3 are linear
operators and g1,and g2 are source terms.
the following correction functionals for the system (i)can be set in the form
𝑡
𝑢𝑛+1 (𝑥, 𝑡) = 𝑢𝑛 (𝑥, 𝑡) + ∫ 𝜆1 (𝐿𝑢𝑛 (𝜉) + 𝑅1 (𝑢𝑛 , 𝑣𝑛 ) − 𝑔1 (𝜉)𝑑𝜉
0
𝑡
𝑣𝑛+1 (𝑥, 𝑡) = 𝑣𝑛 (𝑥, 𝑡) + ∫ 𝜆2 (𝐿𝑢𝑛 (𝜉) + 𝑅2 (𝑢𝑛 , 𝑣𝑛 ) − 𝑔2 (𝜉)𝑑𝜉
0
Where λ j, j = 1,2 are general Lagrange multipliers which can be identified optimally via the
variational theory and un, and vn as restricted variations which means δun = 0 and δvn = 0. The
Lagrange multipliers λj, j = 1,2 will be identified optimally via integration by parts as introduced
before as,
And
To successive approximation 𝑢𝑛+1 (𝑥, 𝑡), 𝑣𝑛+1 (𝑥, 𝑡) 𝑛 ≥ 0 of the solution u(u.t) and v(u,t) are
obtained by using Lagrange multipliers and selective functions u0 and v0.
Consequently, the solutions are given by,
𝑢(𝑥, 𝑡) = lim 𝑢𝑛 (𝑥, 𝑡)
𝑛→∞
25
u x, 0 e x
v x, 0 e x
Where u u x, t , v v x, t .
We can select u0 x, t e x and v0 x, t eu from given initial condition. By using this selection,
we get
u0 x, t e x
v0 x, t e x
u1 x, t e x te x
v1 x, t e x te x
1 2 x
u2 x, t e x te x t e
2!
1 2 x
v2 x, t e x te x t e
2!
26
1 2 x 1 3 x 1 4 x
un x, t e x te x t e t e t e ........
2! 3! 4!
1 2 x 1 3 x 1 4 x
vn x, t e x te x t e t e t e ........
2! 3! 4!
u x, t limn un x, t e x cosh t e x sinh t
As a result, we find
λ1 =λ2 =λ3 = −1.
We can select
27
u0 = y+t, v0= y−t, w0= −y+t by using the given initial values. Accordingly, we obtain the
following successive approximations
𝑢0 (x, y, t) = y + t 𝑣0 (x, y, t) = y − t 𝑤0 (x, y, t) = −y + t
𝑢1 = x + y + t 𝑣1 = x + y − t 𝑤1 = x − y + t
. . .
. . .
. . .
𝑢𝑛 = x + y + t 𝑣𝑛 = x + y − t 𝑤𝑛 = x − y + t
Where 𝑛 ≥ 2,
This gives the following equations,
𝑢 = x+y+t 𝑣 = x+y−t 𝑤 =x−y+t
Chapter 3
Methodology: An important point can be made here in that the method attacks the problem,
homogeneous or inhomogeneous, in a straightforward manner without any need for transformation
formulas. No need to change the inhomogeneous boundary conditions to homogeneous conditions
as required by the method of separation of variables.
Without loss of generality, we study the initial-boundary value problem
PDE 𝑢𝑡 = 𝑢𝑥𝑥 , 0 < 𝑥 < 𝜋, 𝑡 > 0
BC 𝑢(0, 𝑡) = 0, 𝑡 ≥ 0
And 𝑢(𝐿, 𝑡) = 0, 𝑡 ≥ 0 … … … … … … … … … … … … … … … … … … . . (1)
IC 𝑢(𝑥, 0) = 𝑓(𝑥), 0 ≤ 𝑥 ≤ 𝜋
to achieve our goal.
To achieve our goal, equation (1) written in an operator form by
𝐿𝑡 𝑢(𝑥, 𝑡) = 𝐿𝑥 𝑢(𝑥, 𝑡) … … … … … … … … … … … … … … … … … … … … … . (2)
28
𝑑 𝑑2
𝐿𝑡 = 𝑑𝑡 , 𝐿𝑥 = 𝑑𝑥 2
It is obvious that Lt-1 and Lx-1 exist and defined by
𝑡 𝑥
Lt-1 = ∫0 𝑑𝑡 , Lx-1 = ∫0 𝑑𝑥𝑑𝑥.
29
Example -1: Use the Adomian decomposition method to solve the initial boundary value problem
PDE ut = uxx , 0<x<π , t>0
BC u(01t) = 01t , 𝑡 ≥ 0
u(π1t) = 01t , 𝑡 ≥ 0
IC u(x1t) = sinx
𝐿 + 𝑢 = 𝐿𝑥 𝑢 ……………………….(i)
𝜕 𝑡
Where, 𝐿𝑡 = , 𝐿−1
𝑡 = ∫0 𝑑𝑡
𝜕𝑡
𝜕2 𝑥 𝑥
𝐿𝑥 = 𝜕𝑥 2 , 𝐿−1
𝑥 = ∫0 ∫0 𝑑𝑥𝑑𝑥
Applying 𝐿−1
𝑡 to both sides of (i)
𝐿−1 −1
𝑡 𝐿𝑡 𝑢 = 𝐿𝑡 𝐿𝑥 𝑢
𝑢(𝑥, 𝑡) = ∑ 𝑢𝑛 (𝑥, 𝑡)
𝑛=0
∞ ∞
𝑢0 (𝑥, 𝑡) = 𝑠𝑖𝑛𝑥
𝑢1 (𝑥, 𝑡) = 𝐿−1
𝑡 𝐿𝑥 𝑢0 = −𝑡𝑠𝑖𝑛𝑥
30
𝑡2
𝑢2 (𝑥, 𝑡) = 𝐿−1
𝑡 𝐿𝑥 𝑢1 = 𝑠𝑖𝑛𝑥
2!
𝑢(𝑥, 𝑡) = 𝑢0 + 𝑢1 + 𝑢2 + ⋯
𝑡2
= 𝑠𝑖𝑛𝑥 (1 − 𝑡 + −⋯)
2!
= 𝑠𝑖𝑛𝑥𝑒 −𝑡
= 𝑒 −𝑡 𝑠𝑖𝑛𝑥
Which satisfies the PDE, the initial condition and boundary condition.
Example-1: Use the adomian decomposition method to solve the inhomogeneous PDE
BC 𝑢(0, 𝑡) = 𝑒 −𝑡 , 𝑡 ≥ 0
𝑢(𝜋, 𝑡) = −𝑒 −𝑡 , 𝑡 ≥ 0
IC 𝑢(𝑥, 0) = 𝑐𝑜𝑠𝑥
𝑢𝑡 = 𝑢𝑥𝑥 + 𝑠𝑖𝑛𝑥
𝐿 + 𝑢 = 𝐿𝑥 𝑢 + 𝑠𝑖𝑛𝑥 …………………(i)
𝜕 𝑡
Where, 𝐿𝑡 = 𝜕𝑡 , 𝐿−1
𝑡 = ∫0 𝑑𝑡
𝜕2 𝑥 𝑥
𝐿𝑥 = 𝜕𝑥 2 , 𝐿−1
𝑥 = ∫0 ∫0 𝑑𝑥𝑑𝑥
31
Applying 𝐿−1
𝑡 to both sides of (i), we get
𝐿−1 −1 −1
𝑡 𝐿𝑡 𝑢 = 𝐿𝑡 𝐿𝑥 𝑢 + 𝐿𝑡 𝑠𝑖𝑛𝑥
We know,
∑ 𝑢𝑛 (𝑥, 𝑡) = 𝑢(𝑥, 𝑡)
𝑛=0
∞ ∞
Now
𝑡2
𝑢1 (𝑥, 𝑡) = 𝐿𝑡 −1 𝐿𝑥 𝑢0 = − sin 𝑥 − t cos 𝑥.
2!
𝑡3 𝑡2
𝑢2 (𝑥, 𝑡) = 𝐿𝑡 −1 𝐿𝑥 𝑢1 = sin 𝑥 + cos 𝑥.
3! 2!
Hence the solution u(x, t) is given by
𝑢(𝑥, 𝑡) = 𝑢0 + 𝑢1 + 𝑢2 + ⋯ ⋯ ⋯
𝑡2 𝑡3 𝑡2
= sin 𝑥 (𝑡 − 2! + 3! − ⋯ ⋯ ) + cos 𝑥 (1 − 𝑡 + 2! − ⋯ ⋯ ⋯ )
𝑡2 𝑡3 𝑡2
= sin 𝑥 (1 − 1 + 𝑡 − 2! + 3! − ⋯ ⋯ ) + cos 𝑥 (1 − 𝑡 + 2! − ⋯ ⋯ ⋯ )
= sin 𝑥 (1 − 𝑒 −𝑡 ) + 𝑒 −𝑡 cos 𝑥.
32
The Variational Iteration Method.
Homogeneous Heat Equation.
Example -1: Use the variational iteration method to solve the initial-boundary value problem
𝐼𝐶 𝑢(𝑥, 0) = sin 𝑥.
1 + 𝜆 ∣𝜉=𝑥 = 0
𝜆′ ∣𝜉=𝑥 = 0
We now select 𝑢0 (𝑥, 𝑡) = 𝑢(𝑥, 0) = sin 𝑥 from the given conditions. Using this selection in (3), we
obtain the following successive approximations
𝑢0 (𝑥, 𝑡) = sin 𝑥
𝑡
𝑢1 (𝑥, 𝑡) = sin 𝑥 − ∫0 (0 − (− sin 𝑥)) 𝑑ξ = sin 𝑥 − t sin x.
𝑡 𝑡2
𝑢2 (𝑥, 𝑡) = sin 𝑥 − t sin 𝑥 − ∫0 {− sin 𝑥 − ( − sin 𝑥 + t sin 𝑥)} 𝑑ξ = sin 𝑥 − t sin x + 2! sin 𝑥.
𝑡2 𝑡3
𝑢3 (𝑥, 𝑡) = sin 𝑥 − t sin x + sin 𝑥 − sin 𝑥 + 0 + 0.
2! 3!
⋮
𝑡2 𝑡3
𝑢𝑛 (𝑥, 𝑡) = sin 𝑥 − t sin x + sin 𝑥 − sin 𝑥 + ⋯ ⋯ ⋯ ⋯ ⋯.
2! 3!
The variational iteration method admits the use of
∴ 𝑢(𝑥, 𝑡) = 𝑒 −𝑡 sin 𝑥.
33
Inhomogeneous Heat Equations.
The variational iteration method handles the inhomogeneous heat problem in a similar way to that used
in the homogeneous type of equation.
Example 1: Use the VIM method to solve the inhomogeneous partial differential equation
BC 𝑢(0, 𝑡) = 𝑒 −𝑡 , 𝑡 ≥ 0
𝑢(𝜋, 𝑡) = −𝑒 −𝑡 , 𝑡 ≥ 0
IC 𝑢(𝑥, 0) = 𝑐𝑜𝑠𝑥.
𝑡2 𝑡2
𝑢2 (𝑥, 𝑡) = 𝑐𝑜𝑠𝑥 − 𝑡𝑐𝑜𝑠𝑥 + 𝑡𝑠𝑖𝑛𝑥 − 𝑠𝑖𝑛𝑥 + 𝑐𝑜𝑠𝑥.
2! 2!
𝑡2 𝑡3 𝑡2 𝑡3
𝑢3 (𝑥, 𝑡) = 𝑐𝑜𝑠𝑥 (1 − 𝑡 + − ) + 𝑠𝑖𝑛𝑥 (𝑡 − + )
2! 3! 2! 3!
𝑡2 𝑡3 𝑡2 𝑡3
𝑢𝑛 (𝑥, 𝑡) = 𝑐𝑜𝑠𝑥 (1 − 𝑡 + − + ⋯ ) + 𝑠𝑖𝑛𝑥 (𝑡 − + + ⋯ )
2! 3! 2! 3!
34
Method of separation of variables.
Introduction: The most important feature of the method of separation of variables is that it successively
replace the PDE by a system of ODEs that are usually easy to handle. Unlike the decomposition method,
the method of separation of variables employs specific assumptions and transformation formulas in
handling PDE. In particular the method of separation of variables requires that the boundary conditions be
homogeneous. For inhomogeneous boundary conditions a transformation formula should be employed to
transform inhomogeneous boundary condition to homogeneous boundary conditions.
Analysis of the method: We begin our analysis by writing the homogeneous PDE with homogeneous
boundary conditions, that describes the heat flow by the partial DE
IC 𝑢(𝑥, 0) = 𝑓(𝑥)
The method of separation of variables consists of assuming that the temperature U(x,t) is identified as the
product of two distinct functions F(x) and T(x), where F(x) depends on the space variable x and T(x) on t.
This assumption allows us to set
Differentiating both sides of (1) with respect to t and twice with respect to 𝑥 , we obtain
𝑈𝑡 (𝑥, 𝑡) = 𝐹(𝑥)𝑇ˈ(𝑥)
(3)
𝑈𝑥𝑥 (𝑥, 𝑡) = 𝐹ˈˈ(𝑥)𝑇(𝑥)
𝐹(𝑥)𝑇ˈ(𝑥) = 𝑘 𝐹ˈˈ(𝑥)𝑇(𝑥)
𝑇ˈ(𝑡) 𝐹ˈˈ(𝑥)
∴ = = - 𝜆2 (say) (4)
𝑘(𝑡) 𝐹(𝑥)
The selection of - 𝜆2 and not 𝜆2 in (4) is the only selection for which non- trivial solution exists. However
we can easily show that selecting the constant to be zero or positive value will lead to the trivial solution
U(x,t) = 0. It is clear that (4) gives two distinct ordinary DE given by
𝑇ˈ(𝑡) + 𝑘 𝜆2 𝑇(𝑡) = 0
(5)
𝐹ˈˈ(𝑥) + 𝜆2 𝐹(𝑥) = 0
This means that The PDE (1) is redueed to the more familiar ODE (5), where each equation relics only on
one variable .
𝑇ˈ(𝑡) + 𝑘 𝜆2 𝑇(𝑡) = 0
35
To find that
2𝑡
T(t)= c 𝑒 −𝑘𝜆 (6)
Where c is a constant.
On the other hand. The function F(x) can be easily determined by solving the second order LODE
𝐹ˈˈ(𝑥) + 𝜆2 𝐹(𝑥) = 0
To find that
Where 𝐴 and B are constant .to determine 𝐴, 𝐵 and 𝜆 , we use the homogeneous boundary conditions
𝑈(0, 𝑡) = 0
(8)
𝑈(𝐿, 𝑡) = 0
Substituting (8) into the assumption (2) that gives
𝐹(0) 𝑇(𝑡) = 0
𝐹(𝐿) 𝑇(𝑡) = 0
Which gives
𝐹(0) = 0
(9)
𝐹(𝐿) = 0
36
𝑛𝜋
𝐹𝑛 (𝑥) = 𝑠𝑖𝑛 ( 𝑥)
𝐿
𝑛𝜋 2
−𝑘𝑡( )
𝑇𝑛 (𝑡) = 𝑒 𝐿 ,𝑛 = 1,2,3 … ….
Ignoring the constants 𝐵 and 𝐶, we conclude that the functions, called the fundamental solutions
𝑢𝑛 (𝑥, 𝑡) = 𝐹𝑛 (𝑥)𝑇𝑛 (𝑡)
𝑛𝜋 𝑛𝜋 2
𝑢𝑛 (𝑥, 𝑡) = sin( 𝑥) 𝑒 −𝑘𝑡( 𝐿 ) , 𝑛 = 1,2,3 … ….
𝐿
That satisfy equation (1) and given boundary conditions using the superposition principle,that
gives the general solution by
∞
𝑛𝜋 𝑛𝜋 2
−𝑘𝑡( )
𝑢(𝑥, 𝑡) = ∑ 𝐵𝑛 sin ( 𝑥) 𝑒 𝐿 … … … … . . (11)
𝐿
𝑛=1
The constant 𝐵𝑛 can be determined in this case by using Fouriar co-efficients given by the
formula
2 𝐿 𝑛𝜋
𝐵𝑛 = ∫ 𝑓(𝑥)𝑠𝑖𝑛 ( 𝑥) 𝑑𝑥.
𝐿 0 𝐿
Having determined the constants 𝐵𝑛 ,the particular solution 𝑢(𝑥, 𝑡) follows immediately.
𝑛𝜋
On the other hand, if the initial condition 𝑓(𝑥) is given in terms of 𝑠𝑖𝑛 ( 𝐿 𝑥) , 𝑛 ≥ 1. The
constant 𝐵𝑛 can be completely determined by expanding (11), using the initial condition and by
equating the co-efficient of like terms on both sides.
37
Example 1. Use the method of separation of variables to solve the following initial boundary value problem
BC 𝑢(0, 𝑡) = 0, 𝑢(𝜋, 𝑡) = 0, 𝑡 ≥ 0
Solution:
We first set
Now differentiating (1) with respect to t and twice with respect to 𝑥 ,we get
𝑢(𝑥, 𝑡) = 𝐹(𝑥)𝑇’(𝑡)
𝑢𝑥𝑥(𝑥,𝑡)=𝐹′′ (𝑋)𝑇(𝑡)
𝑇 ′ (𝑡) 𝐹 ′′ (𝑥)
From (1) 𝑇(𝑡)
= 𝐹(𝑥) =-𝜆2 (say)
𝑇 ′ (𝑡) + 𝜆2 𝑇(𝑡) = 0
and
𝐴 = 0 𝑎𝑛𝑑 𝑠𝑖𝑛𝜆𝜋 = 0
Here 𝑛 = 0 excluded 𝜆𝑛 because it gives trival solution 𝑢(𝑥, 𝑡) = 0.In accordance with the infinite number
of values of 𝜆𝑛 .we therefore write
𝐹𝑛(𝑥) = sin(𝑛𝑥)
2
𝐹𝑛 (𝑡) = 𝑒 −𝑥 𝑡 , 𝑛 = 1,2
38
2
= 𝑠𝑖𝑛𝑛𝑥 𝑒 −𝑥 𝑡 , 𝑛 = 1, 2,3. ..
Which satisfies the given PDE and boundary value condition .Using superposition, we obtain
2
𝑢𝑡 (x, t) = ∑∞
𝑛=1 Bn e −
x t
𝑠𝑖𝑛𝑛𝑥
Where 𝐵𝑛 ,𝑛 ≥ 1 are arbitrary constant .to determine 𝐵𝑛 , 𝑛 ≥ 1.we use initial condition and substitute 𝑡 =
0 in (5) to find
𝐵1 = 1, 𝐵2 = 3, 𝐵𝑘 = 0 𝑓𝑜𝑟 𝑘 ≥ 3
Example-1:
Using the method of separation of variables to solve the following initial-boundary value problem
𝐵𝐶 𝑢𝑥 (0, 𝑡) = 0, 𝑡≥0
𝑢𝑥 (𝜋, 𝑡) = 0, 𝑡≥0
𝐼𝐶 𝑢(𝑥, 0) = 𝑥,
Solution:
We first set
and
𝐵 = 0,
39
and
𝜆𝑛 = 𝑛, 𝑛 = 0,1,2,3,···
𝐹𝑛 (𝑥) = 𝑐𝑜𝑠(𝑛𝑥),
2
𝑇𝑛 (𝑡) = 𝑒 −𝑛 𝑡 , 𝑛 = 0,1,2,···
∑∞
𝑛=0 𝐴𝑛 cos(𝑛𝑥) = 𝑥 .
The arbitrary constants 𝐴𝑛 are determined by using the Fourier method, therefore we find
1 𝜋 𝜋
𝐴0 = 𝜋 ∫0 𝑥𝑑𝑥 = 2
2 𝜋 2 1 1
𝐴𝑛 = 𝜋 ∫0 𝑥𝑐𝑜𝑠(𝑛𝑥)𝑑𝑥 = 𝜋 (𝑛2 cos(𝑛𝜋) − 𝑛2 ) , 𝑛 = 1, 2, 3 …
so that
0 𝑖𝑓 𝑛 𝑖𝑠 𝑒𝑣𝑒𝑛, 𝑛 ≠ 0
𝐴𝑛 = { 4 }
− 𝑖𝑓 𝑛 𝑖𝑠 𝑜𝑑𝑑
𝜋𝑛2
Based on these results for the constants An, the particular solution is given by
𝜋 4 1 2
𝑢(𝑥, 𝑡) = 4 − 𝜋 ∑∞
𝑚=0 (2𝑚+1)2 𝑒
−(2𝑚+1)
cos(2𝑚 + 1) 𝑥.
BC 𝑢(0, 𝑡) = 𝛼, 𝑡 ≥ 0,
𝑢(𝐿, 𝑡) = 𝛽, 𝑡 ≥ 0, (1)
IC 𝑢(𝑥, 0) = 𝑓 (𝑥).
To convert the boundary conditions from inhomogeneous to homogeneous we simply use the following
transformation formula:
40
𝑥
𝑈(𝑥, 𝑡) = (𝛼 + (𝛽 − 𝛼)) + 𝑉(𝑥, 𝑡)
𝑡
This means that 𝑈(𝑥, 𝑡) consists of a steady-state solution, that doesn’t depend on time, defined by
𝑥
𝑊(𝑥) = 𝛼 + (𝛽 − 𝛼)
𝑡
That satisfies the boundary conditions, and a transient solution given by 𝑉(𝑥, 𝑡). We can easily show that
𝑉(𝑥, 𝑡) will be generated by the initial boundary value problem
Be 𝑉(0, 𝑡) = 0 , 𝑉(𝐿, 𝑡) = 0 , 𝑡 ≥ 0
𝑥
IC 𝑉(𝑥, 0) = 𝑓(𝑥) − (𝛼 + 𝑡 (𝛽 − 𝛼))
Be 𝑈(0, 𝑡) = 1 , 𝑈(1, 𝑡) = 2 , 𝑡 ≥ 0
IC 𝑈(𝑥, 0) = 1 + 𝑥 + 2sin(𝜋𝑥)
WE obtain that
2𝑡
𝑇(𝑡) = 𝐶𝑒 −𝜆
𝐵1 = 2 , 𝐵𝑘 = 0 , 𝑘 ≥ 2
41
2
This gives the solution for 𝑉(𝑥, 𝑡) = 2𝑒 −𝜆 𝑡 sin(𝜋𝑥)
Follows immediately.
BC 𝑢(0, 𝑡) = 0, (4)
𝑢(𝐿, 𝑡) = 0,
IC 𝑢(𝑥, 0) = 𝑓(𝑥),
It is easily observed that this equation is not the standard heat equation we discussed so far. Instead, It
includes the term −𝑐𝑢(𝑥, 𝑡) due to the lateral heat loss.
We will focus our attention on converting Eq. (4) to a standard heat equation. Thereafter, we can implement
the separation of variables method in a straightforward way. The goal can be achieved by using the
transformation formula
BC 𝑤(0, 𝑡) = 0, (6)
𝑤(𝐿, 𝑡) = 0,
IC 𝑤(𝑥, 0) = 𝑓(𝑥).
Where 𝑤(𝑥, 𝑡) can be easily obtained in a similar manner to the discussion started above. The following
example illustrates the use of the transformation formula (7)
BC 𝑢(0, 𝑡) = 0,
𝑢(1, 𝑡) = 0,
42
Solution.
Carries into
BC 𝑤(0, 𝑡) = 0,
𝑤(1, 𝑡) = 0,
Setting
𝑤(𝑥, 𝑡) = 𝐹(𝑥)𝑇(𝑡),
Obtained upon using the boundary conditions. To determine the arbitrary constants 𝐵𝑛 , 𝑛 ≥ 1 we substitute
𝑡 = 0 in (3.229) and use the initial condition to find
Which give
𝐵1 = 1, 𝐵2 = 2, 𝐵𝑘 = 0, 𝑘 ≠ 1,3 (2)
43
CHAPTER-08
Where Lx is the highest order differential in x, Ly is the highest order differential in y, R contains
the remaining linear terms of lower derivatives, F(u(x,y)) is an analytic non linear term, and
g(x,y) is an in homogeneous or forcing term. The decision as to which operator Lx or Ly should
be used to solve the problem depends mainly on two bases:
(i) The operator of lowest order should be selected to minimize the size of computational
work.
(ii) The selected operator of lowest order should be of best known conditions to accelerate
the evaluation of the components of the solution.
Assuming that the operator Lx meets the two bases of selection, therefore we set
We first consider the nonlinear partial differential equation given in an operator form
44
ApplyingL−1 x to both sides of (ii)gives
𝑢(𝑥, 𝑦) = 𝜙0 − 𝐿−1 −1 −1 −1
𝑋 𝑔(𝑥, 𝑦) − 𝐿𝑋 𝐿𝑦 𝑢(𝑥, 𝑦) − 𝐿𝑋 𝑅(𝑢(𝑥, 𝑦)) − 𝐿𝑋 𝐹(𝑢(𝑥, 𝑦))….(iii)
Where,
𝜕
𝑢(0, 𝑦) 𝑓𝑜𝑟 𝐿 = 𝜕𝑥
𝜕2
𝑢(0, 𝑦) + 𝑥𝑢𝑥 (0, 𝑦) 𝑓𝑜𝑟 𝐿 = 𝜕𝑥 2
Φ0 = 1 𝜕3
𝑢(0, 𝑦) + 𝑥𝑢𝑥 (0, 𝑦) + 2! 𝑥 2 𝑢𝑥𝑥 (0, 𝑦) 𝑓𝑜𝑟 𝐿 = 𝜕𝑥 3 ……….(iv)
1 1 𝜕4
𝑢(0, 𝑦) + 𝑥𝑢𝑥 (0, 𝑦) + 2! 𝑥 2 𝑢𝑥𝑥 (0, 𝑦) + 3! 𝑥 3 𝑢𝑥𝑥𝑥 𝑓𝑜𝑟 𝐿 = 𝜕𝑥 4
{ }
We proceed in exactly the same manner by calculating the solution u(x, y) in a series form
∞
𝐹(𝑢(𝑥, 𝑦)) = ∑ 𝐴𝑛
𝑛=0
Where 𝐴𝑛 are adomian polynomials that can be generated for all forms of nonlinearity ,
Equation (iii) becomes
∞ ∞ ∞
−1
∑ 𝑢𝑛 (𝑥, 𝑦) = ∅° − 𝐿−1
𝑥 𝑔(𝑥, 𝑦) − 𝐿𝑥 𝐿𝑦 (∑ 𝑢𝑛 (𝑥, 𝑦)) − 𝐿−1
𝑥 𝑅 (∑ 𝑢𝑛 (𝑥, 𝑦))
𝑛=0 𝑛=0 𝑛=0
∞
− 𝐿−1
𝑥 (∑ 𝐴𝑛 )
𝑛=0
The components 𝑢𝑛 (𝑥, 𝑦), 𝑛 ≥ 0 of the solution 𝑢(𝑥, 𝑦) can be recursively determined by using
the relation
𝑢° (𝑥, 𝑦) = ∅° − 𝐿−1
𝑥 𝑔(𝑥, 𝑦)
Using the algorithms described before for calculating 𝐴𝑛 for the nonlinear form F(u), the first few
components can be identified by
𝑈° (𝑥, 𝑦) = ∅° − 𝐿−1
𝑥 𝑔(𝑥, 𝑦)
45
𝑈1 (𝑥, 𝑦) = −𝐿−1 −1 −1
𝑥 𝐿𝑦 𝑈° (𝑥, 𝑦) − 𝐿𝑥 R(𝑈° (𝑥, 𝑦))-𝐿𝑥 𝐴°
𝑈2 (𝑥, 𝑦) = −𝐿−1 −1 −1
𝑥 𝐿𝑦 𝑈1 (𝑥, 𝑦) − 𝐿𝑥 𝑅(𝑈1 (𝑥, 𝑦)) − 𝐿𝑥 𝐴1
𝑈3 (𝑥, 𝑦) = −𝐿−1 −1 −1
𝑥 𝐿𝑦 𝑈2 (𝑥, 𝑦) − 𝐿𝑥 𝑅(𝑈2 (𝑥, 𝑦)) − 𝐿𝑥 𝐴2
𝑈4 (𝑥, 𝑦) = −𝐿−1 −1 −1
𝑥 𝐿𝑦 𝑈3 (𝑥, 𝑦) − 𝐿𝑥 𝑅(𝑈3 (𝑥, 𝑦)) − 𝐿𝑥 𝐴3
Where each component can be determined by using the preceding component. Having calculated
the components 𝑈𝑛 (𝑥, 𝑦), 𝑛 ≥ 0, the solution in a series form is readily obtained.
t
Lt 1 * *dt
0
Applying 𝐿−1
𝑡 to both sides of (ii) and using the initial condition are obtain
Substituting,
U x, t U n x, t
n 0
46
UU x An
n 0
1
U n x, t x Lt An
n 0 n 0
This gives the recursive relation
U 0 ( x, t ) x
U k 1 ( x, t ) Lt 1 ( Ak ), k 0
U 0 ( x, t ) x
U1 ( x, t ) Lt 1 A0 Lt 1 ( x) xt
Where additional terms can be easily computed combining the results obtained above, the
solution in a series form is given by
U(x,t)=x(1-t+t2-t3+…..)
And in a closed form by
U x, t
x
, t 1
1 t
Example 2: Use the modified decompositions method to solve the nonlinear partial
differential equation U t UU x x xt 2 ,U x,0 0, t 0
.where U=U(x,t)
Solution: Given solution,
Ut UU x x xt 2 ,U x,0 0, t 0 (i)
47
Applying Lt 1 to both sides of (ii) and using the initial condition we find
1 3
U ( x, t ) xt xt Lt 1UU x (iii)
3
Using the decomposition assumptions for the linear terms U(x,t) and for the nonlinear term UUx
defined by
U x, t U n x, t
n 0
UU x An
n 0
U x, t xt 3 xt
1
Form (iii) n
3
Lt 1 An (iv)
n 0 n 0
The use of a modified recursive relation given by
U 0 ( x, t ) xt
1 3
U 1 ( x, t ) xt Lt 1 ( A0 ) V
3
U k 2 ( x, t ) Lt 1 ( Ak 1 ), K 0
Consequently, we obtain,
U 0 ( x, t ) xt
1 1 1
U 1 ( x, t ) xt 3 Lt 1 ( xt 2 ) xt 3 xt 3 0
3 3 3 V
U k 2 ( x, t ) 0, K 0 i
48
Nonlinear PDEs by VIM
Where, L is a linear time derivative operator, R is a linear operator which has partial derivative
Where, 𝜆 is a Lagrange multiplier which can be identified optimally via variation iteration method.
0. The successive approximation 𝑢𝑛+1 , 𝑛 ≥ 0 of the solution, 𝑢 will be readily obtained upon
using the determined Lagrange multiplier and any selective function 𝑢0 , consequently, the solution
is given by:
𝑢 = lim 𝑢𝑛
𝑛→∞
49
1 + 𝜆|𝜉=𝑡 = 0|
𝜆′ |𝜉=𝑡 = 0|
Gives, 𝜆 = −1
Substituting this value of the Lagrange multiplier 𝜆 = −1 into the functional equation (2) gives
Selecting 𝑢0 (𝑥, 𝑡) = 𝑥 from the given initial condition yields the successive approximations,
𝑢0 (𝑥, 𝑡) = 𝑥
𝑢1 (𝑥, 𝑡) = 𝑥 − 𝑥𝑡
1
𝑢2 (𝑥, 𝑡) = 𝑥 − 𝑥𝑡 + 𝑥𝑡 2 − 𝑥𝑡 3 + ⋯
3
2
𝑢3 (𝑥, 𝑡) = 𝑥 − 𝑥𝑡 + 𝑥𝑡 2 − 𝑥𝑡 3 + 𝑥𝑡 4 + ⋯
3
𝑢𝑛 (𝑥, 𝑡) = 𝑥(1 − 𝑡 + 𝑡 2 − 𝑡 3 + 𝑡 4 + ⋯ )
50
ut + 𝑢𝑢𝑥 = 𝑥 + 𝑥𝑡 2 , 𝑢(𝑥, 0) = 0, 𝑡 > 0 (1)
1 + 𝜆|𝜉=𝑡 = 0
𝜆′ |𝜉=𝑡 = 0
Gives, 𝜆 = −1
Substituting this value of the Lagrange multiplier 𝜆 = −1 into the functional equation (2) gives
Selecting 𝑢0 (𝑥, 𝑡) = 0 from the given initial condition yields the successive approximations,
𝑢0 (𝑥, 𝑡) = 0
1
𝑢1 (𝑥, 𝑡) = 𝑥𝑡 + 𝑥𝑡 3
3
1 1 2
𝑢2 (𝑥, 𝑡) = 𝑥𝑡 + ( 𝑥𝑡 3 − 𝑥𝑡 3 ) − 𝑥𝑡 3 + ⋯
3 3 5
1 1 2 2
𝑢3 (𝑥, 𝑡) = 𝑥𝑡 + (3 𝑥𝑡 3 − 3 𝑥𝑡 3 ) − (5 𝑥𝑡 3 − 5 𝑥𝑡 3 ) + ⋯
𝑢𝑛 (𝑥, 𝑡) = 𝑥𝑡
Which is the exact solution obtained upon cancelling the noise terms.
51
Example 3: Solve the nonlinear partial differential equation by the VIM,
𝟏
𝐮𝐭 = 𝒙𝟐 + 𝟒 𝒖𝒙 , 𝒖(𝒙, 𝟎) = 𝟎 ,where, u= u (x, t)
1 + 𝜆|𝜉=𝑡 = 0
𝜆′ |𝜉=𝑡 = 0
Gives, 𝜆 = −1
Substituting this value of the Lagrange multiplier 𝜆 = −1 into the functional equation (2) gives
Selecting 𝑢0 (𝑥, 𝑡) = 0 from the given initial condition yields the successive approximations,
𝑢0 (𝑥, 𝑡) = 0
𝑢1 (𝑥, 𝑡) = 𝑥 2 𝑡
1
𝑢2 (𝑥, 𝑡) = 𝑥 2 𝑡 + 𝑥 2 𝑡 3 + ⋯
3
1 2 1
𝑢3 (𝑥, 𝑡) = 𝑥 2 𝑡 + 3 𝑥 2 𝑡 3 + 15 𝑥 2 𝑡 5 + 63 𝑥 2 𝑡 7 …
52
1 2 17 7
𝑢𝑛 (𝑥, 𝑡) = 𝑥(𝑡 + 𝑡 3 + 𝑡 5 + 𝑡 …
3 15 315
𝑢(𝑥, 𝑡) = 𝑥 2 𝑡𝑎𝑛𝑡
53
Un(x,t)= t1(x) + Lt -1 g1 - Lt -1 Lx Vn - Lt -1 An
Vn(x,t)= t2(x) + Lt -1 g2 - Lt -1 Lx Un - Lt -1 Bn .....(vi)
Two recursive relations can be constructed from (vi) given by,
U0(x,t)= f1(x)+Lt -1 g1
Uk+1 (x,t) =-Lt -1 (Lx Vt)- Lt -1 (Ak),k>=0 ...(vii)
and
V0(x,t)= f2(x)+Lt -1 g2
Vk+1 (x,t) =-Lt -1 (Lx Ux)- Lt -1 (Bk),k>=0 ...(viii)
It is an essential feature of the decomposition method that the zeroth components U0(x,t) and
V0(x,t) are defined always by all terms that arise from initial data and from integrating the
source terms. Having defined the zeroth pair (U0,V0) , the remaining pair (Uk,Vk), k>=1 can be
obtained in a recurrent manner by using (vii) and (viii) additional pairs for the decomposition
series. Solutions normally account for higher accuracy. Having determined the components of
U(x, t) and V(x, t) the solution (U,V) of the system follows immediately in the form of a power
series expansion upon using(iv).
54
The linear terms U(x, t) and V(x, t) can be represented by the decomposition series
U(x, t)=∑∞
𝑛=0 Un(x, t)
V(x, t)=∑∞
𝑛=0 Vn(x, t) iv
Where An and Bn are the Adomian Polynomials that can be generated for any form of non-
linearity. Substituting (iv) and (v) into (iii) gives
∑∞ 𝑥 -1 ∞
𝑛=0 Un(x, t)=𝑒 +t- L t (∑𝑛=0 𝐴𝑛 + Un)
∑∞ −𝑥 -1 ∞
𝑛=0 Vn(x, t)=𝑒 +t- L t (∑𝑛=0 𝐵𝑛 + 𝑉 n) v
The modified decomposition method defines the recursive relation in the form
U0(x, t)=ex
U1(x, t)=t-L-1t(A0+U0)
vi
-1
Uk+1(x, t)= -L t(Ak +Uk), k≥ 1
55
Using the derived adomian Polynomials into (vii) and (viii) we obtain the following pairs of
components
U 0 ,V0 e x , e x
U 1 ,V1 te x , te x
2
t 2 x
U 2 ,V2 t ex , e
2! 2!
3
t 3 x
U 3 ,V3 t ex , e
3! 3!
Accordingly, the solution of the system in a series form is given by
2
t3 t2 t3
U ,V e x 1 t t ........ , e x 1 t ........
2! 3! 2! 3!
U ,V e xt , e xt
Vt WxU y 5
Wt U xV y 5
Vt WxU y 5
i
Wt U xV y 5
56
With initial conditions
U(x,y,0)=x+2y,
V(x,y,0)=x-2Y,
ii
W(x,y,0)=-x+2y
Following the analysis presented above we obtain
Substituting the decomposition representations for linear and non linear terms into (iii) yields
1
U n x, y , t x 2 y t Lt An
n 0 n 0
1
Vn x, y , t x 2 y 5t Lt Bn
n 0 n 0
1
Wn x, y , t x 2 y 5t Lt Cn
n 0 n 0
Where An,Bn, Cn are adomian Polynomials for the nonlinear terms VxWy , WxU y , U xV y
respectively the first three adomian polynomials for An,Bn, Cn As follows:
For VxWy we find
A0 V0 xW0 y
A1 V1xW0 y V0 xW1 y
B0 W0 xU 0 y
B1 W1xU 0 y W0 xU1 y
B2 W2 xU 0 y W1xU1 y W0 xU 2 y
57
And for UxVy We find
C0 U 0 xV0 y
C1 U1xV0 y U 0 xV1 y
U 0 ,V0 ,W0 x 2 y t , x 2 y 5t , x 2 y 5t
U1 ,V1 ,W1 2t,2t,2t
U k ,Vk ,Wk 0,0,0 , K>=2
Consequently, the exact solution of the system of nonlinear partial differential equations is given
by
(U,V,W)= (x+2y+3t, x-2y+3t, -x+2y+3t)
58
U n1 x, t U n x, t 1 LU n x, t s R1 U n ,Vn ,Wn N1 U n ,Vn ,Wn g1 t s dts
t
0
Vn1 x, t Vn x, t 2 LVn x, t s R2 U n ,Vn ,Wn N 2 U n ,Vn ,Wn g 2 t s dts
0
t
(iii
)
Wn1 x, t Wn x, t 3 LWn x, t s R3 U n ,Vn ,Wn N 3 U n ,Vn ,Wn g 3 t s dts
t
0
where 𝛾𝑗, 1 ≤ 𝑗 ≤ 3 are generally lagrange’s multipliers ,which can be identified optimally via
variational theory and 𝑈̃𝑛 , 𝑉̃𝑛 𝑎𝑛𝑑 𝑊
̃𝑛 asrestricted variations which means 𝑆𝑈
̃𝑛 = 0, 𝑆𝑉̃𝑛 =
0 𝑎𝑛𝑑 𝑆𝑊 ̃𝑛 = 0.The successive approximations 𝑈𝑛+1 (𝑥, 𝑡), 𝑉𝑛+1 (𝑥, 𝑡) 𝑎𝑛𝑑 𝑊𝑛+1 (𝑥, 𝑡) , 𝑛 ≥ 0 of
the solutions 𝑈(𝑥, 𝑡), 𝑉(𝑥, 𝑡) 𝑎𝑛𝑑 𝑊(𝑥, 𝑡) will follow immediately upon using the obtained
lagrange multipliers and by using selrctive functions 𝑈0, 𝑉0 𝑎𝑛𝑑 𝑊0 .
Consequently, the solutions are given by
𝑈(𝑥, 𝑡) = lim 𝑈𝑛 (𝑥, 𝑡)
𝑛→∞
59
𝑡
𝜕𝑈𝑛 (𝑥, 𝜏) 𝜕𝑈𝑛 (𝑥, 𝜏)
𝑈𝑛+1 (𝑥, 𝑡) = 𝑈𝑛 (𝑥, 𝑡) + ∫ 𝛾1 (𝜏) ( + 𝑉̃𝑛 (𝑥, 𝜏) ̃𝑛 (𝑥, 𝜏) − 1) 𝑑𝜏
+𝑈
𝜕𝜏 𝜕𝑥
0
𝑡 𝜕𝑉𝑛 (𝑥,𝜏) 𝜕𝑉𝑛 (𝑥,𝜏)
𝑉𝑛+1 (𝑥, 𝑡) = 𝑉(𝑥, 𝑡) + ∫0 𝛾2 (𝜏) ( ̃𝑛 (𝑥, 𝜏)
−𝑈 − 𝑉̃𝑛 (𝑥, 𝜏) −
𝜕𝜏 𝜕𝑥
1) 𝑑𝜏..................................................................................................................(ii)
− 1) 𝑑𝜏 … … … … … … … … … … … … … … … … … … … … … … … … … … … … … . (𝑖𝑣)
The zeroth approximation 𝑈0 (𝑥, 𝑡) = 𝑒 𝑥 𝑎𝑛𝑑 𝑉0 (𝑥, 𝑡) = 𝑒 −𝑥 are selected by using y the given
initial conditions. Therefore, we obtain the following successive approximations
𝑈0 (𝑥, 𝑡) = 𝑒 𝑥 , 𝑉0 (𝑥, 𝑡) = 𝑒 −𝑥
𝑈1 (𝑥, 𝑡) = 𝑒 𝑥 − 𝑡𝑒 𝑥 , 𝑉1 (𝑥, 𝑡) = 𝑒 −𝑥 + 𝑡𝑒 −𝑥 ............(v)
𝑡2
𝑈2 (𝑥, 𝑡) = 𝑒 𝑥 − 𝑡𝑒 𝑥 + 𝑒 𝑥 + 𝑛𝑜𝑖𝑠𝑒 𝑡𝑒𝑟𝑚𝑠
2!
𝑡2
𝑉2 (𝑥, 𝑡) = 𝑒 −𝑥 + 𝑡𝑒 −𝑥 + 𝑒 −𝑥 + 𝑛𝑜𝑖𝑠𝑒 𝑡𝑒𝑟𝑚𝑠
2!
60
𝑡2 𝑡3
𝑈𝑛 (𝑥, 𝑡) = 𝑒 𝑥 (1 − 𝑡 + 2! − 3! + ⋯ … . )
𝑡2 𝑡3
𝑉𝑛 (𝑥, 𝑡) = 𝑒 −𝑥 (1 + 𝑡 + 2! + 3! + ⋯ … . ) .............(vi)
𝑈(𝑥, 𝑡) = 𝑒 𝑥−𝑡
𝑉(𝑥, 𝑡) = 𝑒 −𝑥+𝑡
𝑼𝒕 − 𝑽𝒙 𝑾𝒚 = 𝟏
𝑽 − 𝑾𝒙 𝑼𝒚 = 𝟓
𝑾𝒕 − 𝑼𝒙 𝑽𝒚 = 𝟓
𝑈𝑡 − 𝑉𝑥 𝑊𝑦 = 1
𝑉 − 𝑊𝑥 𝑈𝑦 = 5
𝑊𝑡 − 𝑈𝑥 𝑉𝑦 = 5................................................................(i)
so that
𝛾1 = 𝛾2 = 𝛾3 = −1................................(iv)
61
𝑡 𝜕𝑈𝑛 (𝑥,𝑦,𝜏) 𝜕𝑉𝑛 (𝑥,𝑦,𝜏) 𝜕𝑊𝑛 (𝑥,𝑦,𝜏)
𝑈𝑛+1 (𝑥, 𝑦, 𝑡) = 𝑈𝑛 (𝑥, 𝑦, 𝑡) − ∫0 ( − × − 1) 𝑑𝜏
𝜕𝜏 𝜕𝑥 𝜕𝑦
are selected by using the given initial conditions consequently, the following successive
approximations.
u0 (x,y,t) = x+2y
v0 (x,y,t) = x-2y
w0(x,y,t ) = - x+2y
u1 (x,y,t) = x+2y+3t
v1 (x,y,t) = x-2y+3t
w1 (x,y,t) = -x+2y+3t
.
.
un (x,y,t) = x+2y+3t
vn (x,y,t) = x-2y+3t
wn (x,y,t) = -x+2y+3t
are readily obtained.
The exact solution are given by
u (x,y,t) = x+2y+3t
v (x,y,t) = x-2y+3t
w (x,y,t) = -x+2y+3t
62
Chapter-13
The family of the kdv Equations
Introduction:
The Ubiquities kdv equation in dimentional variables and reals
Ut+aUUx+Uxxx=0-----------(1)
Where subscribes denoted partials derivatives the parameter can be scale to any real number when
the commonly used values are a=+-1 or a=+-6
The kdv equation models a variety of nonlinear phenomena including acoustic waves in plasmas
and shallow water waves the derivative UT characterized the time evolution of the wire
propagating in one direction the nonlinear terms UUx describes the steeping of the world and the
linear term you access account for the spreading or dispersion of the web.
The function you U(x,t ) represents the waters free surface in one dimension variables then only
in kdv equation gives large variety of solution, the solution propagate at speed see while rerating,
its identity we usually introduced the new waves variable S=x-ct so that
U(x,t)=U(ţ)………(ii)
63
-cu´+6uu´+u´´´ = 0 …………(iii)
That gives
𝑐 √𝑐
u(x,t)= 𝑠𝑒𝑐ℎ2 ţ (x- ct)……………..(vi)
2 2
It is obvious that u(u,t) in (vi) along with its derivatives approaches zero as ţ → ∞ . The amsplitude
of the wave is directly proportional to its speed c , and this is turn mean that the taller the above
the faster it moves . It moves to the right for (-c) and to the left if we replace (-c) by (+c). It is also
clear that the wave has no dispersion because of the balance between the dispersion effect and the
nonlinear effect. Consequently, the wave retains its identity and shape. Based on (vi), the following
𝑐 √𝑐
U(x,t) = − 2 𝑠𝑒𝑐ℎ2 ( (𝑥 − 𝑐𝑡)) …………….. (vii)
2
64
Where u(x,t) is a function of space and time variable
αu
αu2
f(u) = αun
αux
2αu-3βu2
αun-βu2n
Ut -6u2ux+uxxx = 0
𝑇1 = 𝑈
1 2
𝑇2 = 𝑈
2
1 4 1 2
𝑇3 = 𝑈 − 𝑈𝑥
4 4
𝑋1 = 2𝑈 3 + 𝑈𝑥𝑥
3 4 1
𝑋2 = 𝑈 + 𝑈𝑈𝑥𝑥 − 𝑈𝑥2
2 2
1 1 2
𝑋3 = 𝑈 6 + 𝑈 3 𝑈𝑥𝑥 − 3𝑈 2 𝑈𝑥2 − 𝑈𝑥 𝑈𝑥𝑥𝑥 + 𝑈𝑥𝑥
2 4
4. Generalized KdV equation
𝑈𝑡 + 𝑎𝑈 𝑛 𝑈𝑥 + 𝑈𝑥𝑥𝑥 = 0, 𝑛 > 2
65
5. The potential KdV equation
𝑈𝑡 + 𝑎𝑈𝑥2 + 𝑈 ′′′ = 0
using the transformation 𝑈 = 𝑉𝑥 and integrating once
𝑈𝑡 − 6𝑈 2 𝑈𝑥 + 𝑈𝑥𝑥𝑥 = 0 … … …(i)
The mKdV equation is completely integrable and is known to exhibit N-soliton solutions and
infinite number of conserved densities. A conservation law is given by the relation
𝜕𝑇 𝜕𝑋
+ = 0 … … … (ii)
𝜕𝑡 𝜕𝑥
where T and X are the density and flux respectively. This in turn gives the first three conservation
laws
𝑇1 = 𝑈
1 2
𝑇2 = 𝑈
2
1 4 1 2
𝑇3 = 𝑈 − 𝑈𝑥
4 4
𝑋1 = 2𝑈 3 + 𝑈𝑥𝑥
3 4 1
𝑋2 = 𝑈 + 𝑈𝑈𝑥𝑥 − 𝑈𝑥2
2 2
66
1 1 2
𝑋3 = 𝑈 6 + 𝑈 3 𝑈𝑥𝑥 − 3𝑈 2 𝑈𝑥2 − 𝑈𝑥 𝑈𝑥𝑥𝑥 + 𝑈𝑥𝑥
2 4
The modified KdV equation differs from the original KdV equation in the nonlinear term only,
where it includes 𝑈 2 𝑈𝑥 in stead of 𝑈𝑈𝑥 but include the dispersion term 𝑈𝑥𝑥𝑥 .
The kdv and the mkdv equations are linked at a depper level by the so called Miura transformation,
given by
𝑢 = 𝑣 2 + 𝑣𝑥 (3)
that gives
𝑢𝑡 = 2𝑣𝑣𝑡 + 𝑣𝑥𝑡
𝑢𝑥 = 2𝑣𝑣𝑥 + 𝑣𝑥𝑥
𝑢𝑡 − 6𝑢𝑢𝑥 + 𝑢𝑥𝑥𝑥 = 0
leads to
𝜕
(2𝑣 + 𝜕𝑥)(𝑣𝑡 − 6𝑣 2 𝑣𝑥 + 𝑣𝑥𝑥𝑥𝑥 = 0
𝑣 2 − 6𝑣𝑣𝑥 + 𝑣𝑥𝑥𝑥 = 0
Substituting the wave variable 𝑠 = 𝑥 − 𝑐𝑡 into the mkdv equation
𝑢𝑡 + 𝑎𝑢2 𝑢𝑥 + 𝑢𝑥𝑥𝑥 = 0
𝑎
And integrating once we find −𝑐𝑢 + 3 𝑢3 + 𝑢𝑢 = 0
To an ODE
67
Equation (2) is then integral as long as all terms contain derivatives where integration constants
are considered zeros.
Since, all derivatives of a tanh are represented by tanh itself, If we set,𝑇 = 𝑡𝑎𝑛ℎ(𝑠),then we have
T = tanh(s)
𝑇′ = 1 − 𝑇2
𝑇′′ = −2𝑇 + 2𝑇 3
𝑇 ′′′ = −2 + 8𝑇 2 − 6𝑇 4
𝑑 2 2 2)
𝑑 2 2 2
𝑑 2
= −2µ 𝑦(1 − 𝑦 + µ (1 − 𝑦 )
𝑑𝑠 2 𝑑𝑦 𝑑𝑦 2
𝑑3 3 (1 2 )(3𝑌 2
𝑑 3 2 )2
𝑑2 3 2 3
𝑑3
= 2𝜇 −𝑌 − 1) − 6𝜇 𝑌(1 − 𝑌 + 𝜇 (1 − 𝑌 )
𝑑𝜉 3 𝑑𝑌 𝑑𝑌 2 𝑑𝑌 3
𝑑4 4 2 )(3𝑌 2
𝑑 4 (1 2 )2 (9𝑌 2
𝑑2
= −8𝜇 𝑌(1 − 𝑌 − 2) + 4𝜇 − 𝑌 − 2)
𝑑𝜉 4 𝑑𝑌 𝑑𝑌 2
𝑑3 𝑑4
−12𝜇 4 𝑌(1 − 𝑌 2 )3 + 𝜇 4 (1
− 𝑌 2 )4
… … … … . … … … … … … (𝑣)
𝑑𝑌 3 𝑑𝑌 4
The tanh-coth method admits the use of the finite expansion
𝑀 𝑀
𝑘=0 𝑘=1
Where M is a positive integer, in most cases, that will be determined. For non-integer M, a
transformation formula is used to overcome this difficulty. Expansion (vi) reduce to the standard
tanh method for 𝑏𝑘 = 0 , 1 ≤ 𝑘 ≤ 𝑀. Substituting (vi) into the reduced ODE results in an algebraic
equation in power of Y.
To carry out the balance method, we notice from (v) and (vi) that the highest exponents for the
function u and it’s derivatives are as follows
68
𝑢 → 𝑀,
𝑢𝑛 → 𝑛𝑀, … … … … … … … … … . . (𝑣𝑖𝑖)
𝑢′ → 𝑀 + 1,
𝑢′′ → 𝑀 + 2,
𝑢𝑟 → 𝑀 + 𝑟, … … … … … … … … … (𝑣𝑖𝑖𝑖)
To determine the parameter M, we usually balance the linear terms of highest order in the resulting
equation with the highest order nonlinear terms by using the scheme given above. We then collect
all coefficients of powers of Y in the resulting equation where these coefficients have to vanish.
This will give a system of algebraic equations involving the parameters 𝑎𝑘 , 𝑏𝑘 , 𝜇 𝑎𝑛𝑑 𝑐. Having
determined these parameters we obtain an analytic solution 𝑢(𝑥, t) in a closed form. The solutions
we obtain may be solitons in terms of 𝑠𝑒𝑐ℎ2 , or may be kinks in terms of tanh. However, this
method may give periodic solutions as well.
𝜕𝑈 𝜕Ӻ
𝑈𝑥 = 𝜕Ӻ . 𝜕𝑥 = U׳
𝑈𝑥𝑥𝑥 = U׳
Then,
-cU׳+aUU ׳+U=׳׳׳0
Intregrating this
𝑎
-cu+2 𝑈 2 + 𝑈 = ׳׳0 … … . . (𝑖𝑖)
M=2
From tanh-coth method we get,
U=∑2𝑗=0 𝑎𝑗 𝑌𝑗 +∑2𝑖=1 𝑏𝑖 𝑌 −𝑖
=𝑎0 +𝑎1 Y+𝑎2 𝑌 2 +𝑏1 𝑌 −1 +𝑏2 𝑌 −2
69
U=𝑎0 + 𝑎1 tanh(µӺ) + 𝑎2 𝑡𝑎𝑛ℎ2 (µӺ) + 𝑏1 𝑡𝑎𝑛ℎ−1 (µӺ) + 𝑏2 𝑡𝑎𝑛ℎ−2 (µӺ)
𝑢𝑎 = ׳1 𝑠𝑒𝑐ℎ2 (µӺ). µ+𝑎2 2 tanh(µӺ) 𝑠𝑒𝑐ℎ2 (µӺ). µ + 𝑏1 (−1)𝑡𝑎𝑛ℎ−2 (µӺ). 𝑠𝑒𝑐ℎ2 (µӺ)µ
+ 𝑏2 (−2)𝑡𝑎𝑛ℎ−3 (µӺ) sec ℎ2 (µӺ) µ
𝑎1 µ𝑠𝑒𝑐ℎ2 (µӺ) + 2𝑎2 µ tanh(µӺ) 𝑠𝑒𝑐ℎ2 (µӺ) − 𝑏1 µ𝑡𝑎𝑛ℎ−2 (µӺ)(1 − 𝑡𝑎𝑛ℎ2 (µӺ))
− 2𝑏2 µ𝑡𝑎𝑛ℎ−3 (µӺ)(1 − 𝑡𝑎𝑛ℎ2 (µӺ))
𝑢𝑎 = ׳׳1 µ(0 − 2 tanh(µӺ) 𝑠𝑒𝑐ℎ2 (µӺ)µ)2𝑎2 µ(𝑠𝑒𝑐ℎ2 (µӺ)µ − 3𝑡𝑎𝑛ℎ2 (µӺ)𝑠𝑒𝑐ℎ2 (µӺ)µ)
− 𝑏1 µ(−2𝑡𝑎𝑛ℎ−3 (µӺ)𝑠𝑒𝑐ℎ2 (µӺ)µ − 0)
− 2𝑏2 µ(−3𝑡𝑎𝑛ℎ−4 (µӺ) sec ℎ2 (µӺ)µ + 𝑡𝑎𝑛ℎ−2 (µӺ)𝑠𝑒𝑐ℎ2 (µӺ)µ)
=−𝑐(𝑎0 + 𝑎1 𝑌 + 𝑎2 𝑌 2 + 𝑏1 𝑦 −1 +
𝑎
𝑏2 𝑌 −2 )+2(𝑎02 +𝑎12 𝑌 2 +𝑎22 𝑌 4 +2𝑎0 𝑎1Y+2𝑎1 𝑎2 𝑌 3 +2𝑎0 𝑎2 𝑌 2 +2𝑎0 𝑏1 𝑌 −1 +2𝑎0 𝑏2 𝑌 −2 +
2𝑎1 𝑏1 +𝑎1 𝑏2 𝑌 −1 +𝑎2 𝑏1 𝑌+𝑎2 𝑏2 +𝑏12 𝑌 −2 +𝑏22 𝑌 −4+2𝑎1 𝑏1 𝑌 −3 )−2𝑎µ2 (𝑦 − 𝑦 3 ) + 2𝑎2 µ2 (1 − 𝑦 2 ) −
6𝑎2 µ2 (𝑦 2 − 𝑦 4 ) + 2𝑏1 µ2 (𝑦 −3 − 𝑦 −1 ) + 6𝑏2 µ2 (𝑦 −4 − 𝑦 −2 ) − 2𝑏2 µ2 (𝑦 −2 − 1)
Collecting the co-efficients of each occur of year. 0 ≤ 𝑟 ≤ 8, setting each coefficient to zero.
𝑎𝑎22 + 12𝑎2 µ2 = 0
70
2𝑎𝑎1 𝑎2 + 4𝑎1 µ2 = 0
2𝑎𝑏1 𝑏2 + 4𝑏1 µ2 = 0
𝑎𝑏22 + 12𝑏2 µ2 = 0
Solving the above equation by any computational software we can find the values if the unknowns
𝑎0 , 𝑎1 , 𝑎3 , 𝑏1 , 𝑏2 𝑎𝑛𝑑 µ.Using this value of the unknowns in solution equation (3) we obtain the
solution of the given equation (2).
Example 2: Solve the modified 𝒌𝒅𝒗 (𝒎𝒌𝒅𝒗) equation by using the 𝒕𝒂𝒏𝒉 − 𝒄𝒐𝒕𝒉 method.
Solution:
The 𝑚𝑘𝑑𝑣 equation reads
𝑈𝑡 − 6𝑈 2 𝑈𝑥 + 𝑈𝑥𝑥𝑥 = 0 … … … … (𝑖)
which is PDE. Convert it into ODE and then integrating we get,
𝑎
−𝐶𝑈 + 𝑈 3 + 𝑈 ′′ = 0 … … … … (𝑖𝑖)
3
Solving this by using 𝑡𝑎𝑛ℎ − 𝑐𝑜𝑡ℎ method.
Balancing the non-linear term 𝑈 3 , that has the exponent 3𝑀 with the highest order derivative 𝑈 ′′ ,
that has the exponent 𝑀 + 2.
∴ 3𝑀 = 𝑀 + 2
⇒𝑀=1
= 𝑎0 + 𝑎1 𝑌 + 𝑏1 𝑌 −1
71
= 𝑎1 𝜇(1 − 𝑡𝑎𝑛ℎ2 (µξ)) − 𝑏1 𝜇[𝑡𝑎𝑛ℎ(µξ) − 𝑡𝑎𝑛ℎ3 (µξ)]
= −2𝑎1 𝜇 2 𝑌(1 − 𝑌 2 ) − 𝑏1 𝜇 2 [1 − 𝑌 2 − 3𝑌 2 + 3𝑌 4 ]
= −2𝑎1 𝜇 2 𝑌(1 − 𝑌 2 ) − 𝑏1 𝜇 2 [1 − 4𝑌 2 + 3𝑌 4 ]
From (ii)
𝑎
−𝐶(𝑎0 + 𝑎1 𝑌 + 𝑏1 𝑌 −1 ) + (𝑎0 + 𝑎1 𝑌 + 𝑏1 𝑌 −1 )3 − 2𝑎1 𝜇 2 𝑌(1 − 𝑌 2 ) − 𝑏2 𝜇 2 [1 − 4𝑌 2 + 3𝑌 4 ]
3
1
= −𝐶𝑎0 − 𝐶𝑎1 𝑌 − 𝑏1 𝐶𝑌 −1 + [𝑎𝑎03 + 3𝑎𝑎02 𝑎1 𝑌 + 3𝑎𝑎0 𝑎12 𝑌 2 + 𝑎𝑎13 𝑌 3 + 3𝑎𝑎02 𝑏1 𝑌 −1 +
3
6𝑎𝑎0 𝑎1 𝑏1 + 3𝑎𝑎12 𝑏1 𝑌 + 3𝑎𝑎0 𝑏12 𝑌 −2 + 3𝑎𝑎1 𝑏12 𝑌 −1 + 𝑎𝑏13 𝑌 −3 ] − 2𝑎1 𝜇 2 𝑌 + 2𝑎1 𝜇 2 𝑌 3 −
𝑏1 𝜇 2 + 4𝑏1 𝜇 2 𝑌 2 − 3𝑏1 𝜇 2 𝑌 4
𝑎𝑎13 + 6𝑎1 𝜇 2 = 0
3𝑎𝑎0 𝑏12 = 0
𝑎𝑏13
Solving the above equation by any computational software. We can find the values of the
unknowns in solution equation (iii) we obtain the solution of the given equation (ii).
72
3𝑎𝑎0 𝑎12 + 12𝑏1 𝜇 2 = 0
−3𝐶𝑎1 + 3𝑎𝑎02 𝑎1 + 3𝑎𝑎12 𝑏1 − 6𝑎1 𝜇 2 = 0
−3𝐶𝑎0 + 𝑎𝑎03 + 6𝑎𝑎0 𝑎1 𝑏1 − 3𝑏1 𝜇 2
−3𝑏1 𝐶 + 3𝑎𝑎02 𝑏1 + 3𝑎𝑎1 𝑏12
3𝑎𝑎0 𝑏12 = 0
𝑎𝑏13
Solving the above equation by any computational software. We can find the values of the
unknowns in solution equation (iii) we obtain the solution of the given equation (ii).
The polynomial can be determined by considering the homogeneous balance between the highest
order derivatives and nonlinear terms appearing in a given nonlinear evolution equation ,and the
coefficients of the polynomial can be obtained by solving a set of algebraic equations resulted from
the process of the using proposed method. It will be seen that more travelling wave solutions of
many nonlinear evolution equation can be obtained by using the (Gʹ/G)-expansion method.
73
Where U=U(x,t) is an unknown function, p is a polynomial in U=U(x,t) and its various partial
derivatives, in which the highest order derivatives and nonlinear terms are involved.
Step-1: Combining the independent variables x and t into one variable ζ=x-vt
U(x,t)=U(ζ), ζ=x-vt ………….(ii)
From (i)
P(u,-vuʹ,uʹ,vuʹʹ,-vuʹʹ,uʹʹ,………)=0 …………...(iii)
Step-2: Suppose that the solution of ODE (iii) can be expressed by a polynomial in (Gʹ/G) as
follows:
𝐺ʹ
U(ζ)=𝛼𝑚 ( 𝐺 )𝑚 +……… …………(iv)
Step-3: By substituting (iv) into equation (iii) and using second order LODE (v), collecting all
terms with the same order of (Gʹ/G) together. Equation (iii) is converted into another polynomial
in (Gʹ/G). Equating each coefficient of this polynomial to zero.
Step-4: Assuming that the constants 𝛼𝑚 ,…….𝑉,λ and µ can be obtained by solving the algebric
equation. The general solution of 𝑒𝑞 𝑛 . (v) into 𝑒𝑞 𝑛 . (iv) we have more travelling wave solutions
of evolution equation(i).
Example-1: Solve the modified kdV equation by using the original the (Gʹ/G)-expansion
method.
74
Balancing the terms U3 and Uʹʹ
3M=M+2
⇒M=1
Using (Gʹ/G)-expansion method
𝐺ʹ
U(x,t)=a0+a1( 𝐺 ) ……………..(iii)
Now
𝐺𝐺ʹʹ−𝐺ʹ2
( 𝐺ʹ⁄𝐺 )ʹ = 𝐺2
𝐺(−𝜆𝐺ʹ−µ𝐺)−𝐺ʹ2
= [as Gʹʹ+λGʹ+µG=0]
𝐺2
2
𝐺′ 𝐺′ 𝐺′
∴ = −𝜆 ( ) − 𝜇 − (( )
𝐺 𝐺 𝐺
From (𝑖𝑖𝑖)
2
𝐺′ 𝐺′
U′ = 𝑎1 {𝜆 ( 𝐺 ) − 𝜇 ( 𝐺 ) }
2 2
″
𝐺′ 𝐺′ 𝐺′ 𝐺′ 𝐺′
𝑈 = 𝑎1 [−𝜆 {−𝜆 ( ) − 𝜇 − ( ) } − 2 ( ) {−𝜆 ( ) − 𝜇 − ( ) }]
𝐺 𝐺 𝐺 𝐺 𝐺
2 2 3
𝐺′ 𝐺′ 𝐺′ 𝐺′ 𝐺′
= 𝑎1 [𝜆2 ( 𝐺 ) + 𝜆𝜇 + 𝜆 ( 𝐺 ) + 2𝜆 ( 𝐺 ) + 2𝜇 ( 𝐺 ) + 2 ( 𝐺 ) ]
2 3
𝐺′ 𝐺′ 𝐺′
= 𝑎1 [(𝜆2 + 2𝜇) ( 𝐺 ) + 3𝜆 ( 𝐺 ) + 2 ( 𝐺 ) + 𝜆𝜇]
From (𝑖𝑖)
3 2 3
𝐺′ 𝑎 𝐺′ 2 𝐺′ 𝐺′ 𝐺′
−𝑐 (𝑎0 + 𝑎1 ( )) + (𝑎0 + 𝑎1 ( )) + 𝑎1 [(𝜆 + 2𝜇) ( ) + 3𝜆 ( ) + 2 (𝐺 ) + 𝜆𝜇] = 0
𝐺 3 𝐺 𝐺 𝐺
2 3
𝐺′ 𝑎𝑎0 3 2
𝐺′ 2
𝐺′ 𝑎 3 𝐺′
=> −𝑐𝑎0 − 𝑐𝑎1 ( ) + + 𝑎𝑎0 𝑎1 ( ) + 𝑎𝑎0 𝑎1 ( ) + 𝑎1 ( )
𝐺 3 𝐺 𝐺 3 𝐺
2 3
𝐺′ 𝐺′ 𝐺′
+𝑎1 (𝜆2 + 2𝜇) ( 𝐺 ) + 3𝜆𝑎1 ( 𝐺 ) + 2𝑎1 ( 𝐺 ) + 𝑎1 𝜆𝜇 = 0
2 3
𝐺′ 𝐺′ 𝐺′ 𝐺′
-3 𝑐𝑎0 − 3𝑐𝑎1 ( 𝐺 ) + 𝑎𝑎0 3 + 3𝑎𝑎0 2 𝑎1 ( 𝐺 ) + 3𝑎𝑎0 𝑎1 2 ( 𝐺 ) + 𝑎𝑎1 3 ( 𝐺 ) +
75
2 3
2
𝐺′ 𝐺′ 𝐺′
3𝑎1 (𝜆 + 2𝜇) ( ) + 9𝜆𝑎1 ( ) + 6𝑎1 ( ) + 3𝑎1 𝜆𝜇 = 0
𝐺 𝐺 𝐺
𝐺′
Equating the coefficients of ( )
𝐺
0
𝐺′
( ) : − 3𝑐 𝑎0 + 𝑎𝑎0 3 + 3𝑎1 𝜆𝜇
𝐺
1
𝐺′
( ) : −3𝑐 𝑎1 + 3𝑎𝑎0 2 𝑎1 + 3𝑎1 (𝜆2 + 2𝜇)
𝐺
2
𝐺′
( ) : 3𝑎𝑎0 𝑎1 2 + 9𝜆𝑎1
𝐺
3
𝐺′
( ) : 𝑎1 3 + 6𝑎1
𝐺
By setting each coefficients equal to zero and applying any computational software to solve and getting
the unknowns we can have the solution of (𝑖).
Example 2: Solve the (2+1) Dimensional Cubic Nonlinear Klein – Gordon Equation by using
MSE method.
Solution: The (2+1) Dimensional Cubic Nonlinear Klein – Gordon Equation has the form
𝑈𝑥𝑥 + 𝑈𝑦𝑦 − 𝑈𝑡𝑡 + 𝛼𝑈 − 𝛽𝑈 3 = 0 … … … … … … … … … … (1)
76
𝜓′
𝑈(𝜉) = 𝐴0 + 𝐴1 ( )
𝜓
𝜓′
= 𝐴0 + 𝐴1 𝑀 [𝑀 = ]
𝜓
𝑈 ′ = 𝐴1 𝑀′
𝑈 ′′ = 𝐴1 𝑀′′
From (3)
(2 − 𝑐 2 )𝐴1 𝑀′′ + 𝛼(𝐴0 + 𝐴1 𝑀) − 𝛽(𝐴0 + 𝐴1 𝑀)3 = 0
=> 2𝐴1 𝑀′′ − 𝑐 2 𝐴1 𝑀′′ + 𝛼𝐴0 + 𝛼𝐴1 𝑀 − 𝛽𝐴0 3 − 3𝛽𝐴0 2 𝐴1 𝑀 − 3𝛽𝐴0 𝐴1 2 𝑀2 − 𝛽𝐴1 3 𝑀3 = 0
=> 𝑀(𝛼𝐴1 − 3𝛽𝐴0 2 𝐴1 ) + 𝑀2 (−3𝛽𝐴0 𝐴1 2 ) + 𝑀3 (−𝛽𝐴1 3 ) + 𝑀′′ (2𝐴1 − 𝑐 2 𝐴1 ) + 𝐴0 (𝛼 − 𝛽𝐴0 2 ) = 0
2 3 3
𝜓′ 𝜓′ 𝜓′ 𝜓′′′ 3𝜓′𝜓′′ 2𝜓′
=> (𝛼𝐴1 − 3𝛽𝐴0 2 𝐴1 ) + ( ) (−3𝛽𝐴0 𝐴1 2 ) + ( ) (−𝛽𝐴1 3 ) + ( − 2
+ 3
) (2𝐴1 − 𝑐 2 𝐴1 )
𝜓 𝜓 𝜓 𝜓 𝜓 𝜓
+ 𝐴0 (𝛼 − 𝛽𝐴0 2 ) = 0
−1 ′ −2 ′2 −3 ′3 −1 ′′′
=> 𝜓 (𝜓 (𝛼𝐴1 − 3𝛽𝐴0 2 𝐴1 )) + 𝜓 (𝜓 (−3𝛽𝐴0 𝐴1 2 )) + 𝜓 (𝜓 (−𝛽𝐴1 3 )) + 𝜓 (𝜓 (2𝐴1 − 𝑐 2 𝐴1 ))
−2 ′ ′′ −3 3
−𝜓 (3𝜓 𝜓 (2𝐴1 − 𝑐 2 𝐴1 )) + 𝜓 (2𝜓′ (2𝐴1 − 𝑐 2 𝐴1 )) + 𝐴0 (𝛼 − 𝛽𝐴0 2 ) = 0
−3 ′3 3 0
+𝜓 (−𝜓 𝛽𝐴1 3 + 2𝜓′ 𝐴1 (2 − 𝑐 2 )) + 𝜓 (𝐴0 (𝛼 − 𝛽𝐴0 2 )) = 0
77
Thanks to All
78