Asymptotic Methods in Non linear dynamics

∗
R Dutta
Department of Mathematics, The Ohio State University

I. ABSTRACT

This paper features and elaborates recent developments and modifications in asymptotic techniques in solving
differential equation in non linear dynamics. These methods are proved to be powerful to solve weakly as well as
arXiv:1607.07835v2 [math-ph] 16 Aug 2016

strongly non linear cases. Obtained approximate analytical solutions are valid for the whole solution domain. In this
paper, limitations of traditional perturbation methods are illustrated with various modified techniques. Mathemat-
ical tools such as variational approach, homotopy and iteration technique are discussed to solve various problems
efficiently. Asymptotic methods such as Variational Method, modified Lindstedt-Poincare method, Linearized
perturbation method, Parameter Expansion method, Homotopy Perturbation method and Perturbation-Iteration
methods(singular and non singular cases) have been discussed in various situations. Main emphasis is given on
Singular perturbation method and WKB method in various numerical problems.

keywords: analytical solution, nonlinear equation, perturbation method, variational theory, homotopy perturbation,
variational iteration method, correctional functional.

II. INTRODUCTION

With rapid development of research in non linear science, there appears ever increasing interest of scientists and
engineers in asymptotic techniques to solve non linear problems, specially non linear differential equations. First order
non linear partial equations model non linear waves, gas dynamics, water wave, elastodynamics, chemical reaction,
transport of pollutants, flood wave in river, chromatography, traffic flow and wide range of biological and ecological
systems. One of the most important non linear phenomena without non linear counterpart is the breakdown of
solution in finite time resulting in formation of discontinuous shock wave. A striking example is the supersonic boom
produced by an airplane. In linear wave equation, the signal propagates along with the characteristics. But in non
linear case, the characteristics can cross each other precipitating an onset of shock. The characterization of shock
dynamics requires additional physical information. Parabolic second order differential equations govern non linear
diffusion processes including thermodynamics, chemical reaction, dispersion of pollutants and population dynamics.
The simplest and most well understood is Burger equation which can surprisingly be linearized by transforming it to
heat equation. In the limit of diffusion or viscosity tending to zero, solution of Burger equation tends to shock wave
solution to the limiting first order dispersion less equation and thus provides alternative mechanism for unravelling
shock dynamics. Second order non linear equations arise in fluid dynamics, water wave, Blasius layer theory i.e
Falkner Skan equation, heat wave and heat equation that include wide variety of equations such as autonomous
equation, Emden-Fowler equation , Navier Stokes equation. Third order partial differential equations arise in study
of dispersive wave motion including water wave, plasma wave, wave in elastic medium( very suitable in arterial blood
flow studies). It is very easy to find solution in linear system numerically but becomes very difficult to find solution
for non linear equation analytically or numerically. Because almost all discretized methods or numerical simulation
use iteration method technique in solving equation which is very sensitive to initial solutions [10, 19, 29, 51]. It implies

∗ [email protected]
 2

that it is very difficult to find solution in convergence in presence of non linearity, often impossible in case of strong
one. The problem that we encounter quite often is that amplitude of non linear oscillation which depends on initial
parameter becomes lost during numerical simulation. Analytical asymptotic approaches to solve non linear equations
include Non Perturbative method [24], Adomian Decomposition method [2, 4, 23, 77] , Modified Lindstedt-Poincare
method [20, 39], Variational Iteration method [31, 37, 63] , Energy Balance method [38], F-expansion method
[69, 75] and Perturbative methods [8, 12, 25, 56, 60] are most common approximate methods in studying non linear
mathematical models arising in physics and engineering that are constantly being developed or applied to more
complex non linear systems. A major limitation of these methods is the restriction of small parameter which makes
the solution valid for weakly non linear systems mostly, very difficult to find solution in strongly non linear case.
Alternative perturbative techniques such as Linearized Perturbation method [31], Modified Linstedt-Poincare method
[20], Homotopy Perturbation method [32] and Multiple Scale Lindstedt-Poincare method [64] are being developed
and analyzed in various systems. Alternative attempt to validate solutions for strongly non linear systems is the
use of Iteration - Perturbation technique [53, 80] . On the other hand, Stochastic methods[non asymptotic method]
[26]are being applied numerically to solve qualitative and quantitative properties of the solution for various design of
engineering and Industrial processes. Non linear equations [30] assumes the system to be iterative and these methods
prove to be powerful tool to solve non linear dynamics undergoing random fluctuation in physical parameters
through perturbative technique. Many of these methods or numerical simulations apply iteration technique to find
their numerical solutions of non linear problems and nearly all iterative solutions are sensitive to initial solutions.
Convergence of these solutions is the main difficulty. In addition, natural frequency of non linear oscillation depends
on initial condition that means amplitude of the oscillation will be lost during the procedure of numerical simulation.
Variation Iteration method (VIM) based on the use of restricted variations and correction functionals has found wide
applications for solution of non linear ordinary and partial differential equations. Almost all perturbative methods
are based on assumption with the existence of a small parameter in equation. This parameter assumption greatly
restricts applications of perturbation techniques, a well known problem in solving an overwhelming majority of
non linear problems. It is even more difficult to determine this small parameter which requires special technique
and often arises from understanding of the physical system. An inappropriate choice of small parameter results
in bad effect (sometimes serious effect) on the solution. In presence of suitable parameter value, approximate
solution obtained using perturbative methods are valid only for small values of the parameter in most cases. For
example, approximations solved by multiple scaling are uniformly valid as long as specific system parameter is
small. However, we can not rely on approximation fully because there is no criterion on how small the parameter
should be. Thus numerical result or experimental value of the approximations is necessary. We discuss various
analytical and numerical methods to solve various non linear situations . First , we will discuss variational approaches
such as Ritz method, Soliton solution, Bifurcation , Variational Iteration method. Under modified expanding
parameter method, Lindstedt - Poincare method(modified form) , Parametrized Perturbation method and Homotopy
perturbation method with various equations under non linear dynamics. Finally, we will discuss about Singular per-
turbation method and WKB methods implemented in various non linear situations. If we consider following equations

u′ + u − ǫ = 0 · · · with u(0) = 1
(1)
u′ − u − ǫ = 0 · · · withu(0) = 1

The solution of the first of equation (1) is

u(t) = ǫ + (1 − ǫ)e−t

while the unperturbed solution is

uo (t) = e−t

. In case ǫ ≪ 1, the perturbation solution leads to high accuracy for

| u(t) − u0 (t) |≤ ǫ
 3

By similar analysis of the second of equation (1) we have ,

| u(t) − u0 (t) |= ǫ | (1 − e−t ) |

It is obvious that u0 can never be considered as approximate solution of equation (1) for t ≫ 1 . So, normal
perturbative method is not valid even ǫ ≪ 1.
This indicates that we should introduce new developed method when normal perturbative method fails. There exist
analytical asymptotic approaches such as non perturbative method [24] , Weighted linearized method [5, 67] , Ado-
mian decomposition method [2, 23, 77] , Modified Lindstedt-Poincare method [20, 39] , Variational Iteration method
[31, 37] , Energy Balance Method [38] , tanh method [9] , F-expansion method [69] and so on. Some new perturbation
methods such as Artificial Parameter Method [39] and δ method [8? ], Perturbation Incremental method [20], Homo-
topy perturbation method [35] are being proposed to solve these systems which does not depend on small parameter.
Recent study reveals that numerical technique can also be powerfully applied to perturbative methods. A wide litera-
ture dealing with problem of approximate solutions in non linear equations with various methods have been reviewed.
But comprehensive review of asymptotic methods for non linear equations still requires more research. There exist
some alternative analytical asymptotic approaches such as non perturbative method to probe non linear equations
such as Modified linearization method, Adomian decomposition method [1, 4], Lindstedt-Poincare method [20], Vari-
ational Iteration method [15], Parameter Expanding method [42] and so on. In parameter expanding perturbative
method, parameter can be expanded as power series of ǫ which can be a small parameter, an artificial parameter or
book keeping parameter. Perturbative method proves to be useful tool to analyze stochastic structure of an chaotic
system very well. In this work, we limit to non linear partial differential equation (NPDE) equations because most
PDE can be converted to ODE by linear transformation . We start with KdV equation,

∂t φ + bφ∂x φ + ǫ∂xxx φ = 0 (2)

KdV is a non linear, dispersive PDE for a function φ of two variable , space and time. If we assume that φ(x, t) =
φ(ξ) = φ(x − ct). where c is wave speed. Substituting above into wave equation, we get

− cuξ = +uuξ + ǫuξξξ (3)

Integrating Eq (), we have an Ordinary differential equation

1
− cu + u2 + ǫuξξ = A (4)
2
with A as constant of integration. Interpreting ξ as virtual time variable, this signifies that φ satisfies Newton motion
in cubic potential. If ǫ is adjusted so that potential V (φ) has local minimum for φ = 0 function , there exists solution
in which φ(ξ) starts at virtual time and eventually reaches to local minima. The characteristic shape of the function
is called Solitary wave used in solution of soliton equation. More precisely, the solution is
√
1 c

2
φ(x, t) = csech (x − ct − a) (5)
2 2

This describes right moving soliton. KdV equation has infinitely integrals of motion. In next section, variational
approach to soliton equation , bifurcation, limit cycle and periodic solution of non linear equations are illustrated
including Ritz method, energy method , Iterative variational method. A recent study [36] reveals that the numerical
technique can also be applied powerfully to perturbation method.
 4

III. VARIATIONAL APPROACH

A. Ritz method

Variational methods [28, 52, 81] such as Raleigh-Ritz and Galerkin methods have been popular tools for non linear
analysis. Compared to other methods, variational method provides physical insight into the nature of the solution.
Ritz method is a direct method to find an approximate solution for boundary value problem. It is exactly the finite
element method used to compute Eigen vector and Eigen values of a hamiltonian system. Rayley - Ritz method
allows computation of ritz pairs ( λ̃i , x̃i ) which approximates solution to Eigen value problem. In this method, the
problem must be stated in a variational form as a minimization problem and solution is approximated by a finite
linear combination of all possible trial functions. As an illustration, we apply Ritz method to Lambert equation
2
k2 ẏ(x)
ÿ(x) + y(x) − (1 − n) =0 (6)
n y(x)
and arrived at variational formulation with energy integral in Hamiltonian system as
1
Z
J(y) = (−n2 y 2n−2 ẏ 2 + k 2 y 2n )dt (7)
2
By transformation z = y n , we obtain
1
Z
J(z) = (−ż 2 + k 2 z 2 )dt (8)
2
which leads to linear Euler Lagrange equation
z̈ + k 2 z = 0 (9)
whose solution is of the form
1/n
y = (C coskx + D sinkx) (10)
Considering non linear wave equation i.e kdV equation
∂u ∂u ∂ 3 u
− 6u + 3 = 0 (11)
∂t ∂x ∂x
We seek traveling wave solution of following form
u(x, t) = u(ξ) (12)
with ξ = x − ct. Substituting the solution into differential equation we obtain
− cu̇ − 6uu̇ + ü = 0 (13)
Here prime denotes differential with respect to ξ. This is soliton equation. And after integrating the above equation,
one obtains
− cu − 3u2 + ü = 0 (14)
By semi inverse method [73] , following variational formulations is obtained
2
Z ∞ !
1 2 3 1 du
J= cu + u + ( ) dξ (15)
0 2 2 dξ
 5

Implementing Ritz method, we try to find solution of the form

u = psech2 (qξ) (16)

Upon substituting in the differential equation, we finally obtain solution as

s
c c
u = − sech2 (x − ct − ξ0 ) (17)
2 4

which is exact solitary wave solution of KdV equation. We now focus on Ritz method in bifurcation problem.
Bifurcation arises in various non linear problems [82]. One can apply Variational method to search for dual solution of
a nonlinear equation if it exists. Ritz method method helps to identify bifurcation point. Considering Bratu equation,
we apply Ritz method to nonlinear Bratu equation of the form

ü + λeu = 0 (18)

with boundary condition u(0) = u(1) = 0 Bratu type equation arises in many physical & Engineering problems such
as chemical reaction, radiative heat transfer , nanotechnology and universe expansion model. As an example, the
above equation is typically obtained from solid fuel ignition model in thermal combustion which has exact solution.
The above problem can be solved with high accuracy using variational method. From above equation, one can obtain
variational formulation as
Z 1
1 2
J(u) = u̇ − λeu dx (19)
0 2

We can choose trial function as

u = Ax(1 − x) (20)

Here A is unknown constant to be determined. Substituting u into variational formulation(integral form), one can
obtain
Z 1
1 2 2
J(A) = A (1 − 2x) − λeAx(1−x) dx (21)
0 2
Using minimization principle with respect to A , this yields
Z 1
∂J 2
= A(1 − 2x) − λx(1 − x)eAx(1−x) dx = 0 (22)
∂A 0

One can find out value of critical value for unknown constant A numerically showing that Bratu equation has two
solutions which is the indication bifurcation phenomena for certain A.

B. Limit Cycle Method

In mathematics, in the study of dynamical system with two dimensional phase space, a limit cycle is a closed
trajectory that spirals into it either as time approaches negative infinity or positive infinity. Such behavior is exhibited
in some non linear systems. Limit cycle can be used to model the behavior of many real world oscillatory systems.
To illustrate the basic idea of the method, one can consider non linear oscillation as

ẍ + x + ǫf (x, ẋ) = 0 (23)

 6

of this equation Limit cycle can be determined in general form as

X
x = a + b(t) cosωt + cn cos(nωt) + dn sin(nωt) (24)

in terms of Fourier Series [41]. Assuming amplitude as varying function of time that is

a(t) ≈ Aeαt

, one can calculate α for Van der Pols equation based on limit cycle method as
ǫ2 (A2 − 4)ωπ
α= (25)
4(2πω − ǫA2
In case of electro spin problem, any nth order solution can be written as
Z z  2
∂ un (s)

un+1 (z) = un (z) + (s − z) + f (u n (s) ds (26)
0 ∂s2
Limit cycle method helps to study if the system exhibits any oscillation or not. He et al [41] implemented limit cycle
method for non linear oscillator system which represents various engineering nano mechanical problem.

IV. VARIATIONAL ITERATION METHOD

The variational iteration method [31, 41, 63] has been proved to solve a large class of non linear problems very
effectively and accurately with approximations converging rapidly to accurate solutions. Many authors found that
shortcoming of Adomian method can be improved using Variational Iteration method [59, 63, 72]. Illustrating basic
idea of variational Iteration method, we consider equation of general non linear system as

Lu + N u = g(x) (27)

where L is the linear operator and N is the non linear one. According to variational iteration method, we can construct
following iteration formulation
Z x
un+1 (x) = un (x) + λ(Lun (s) + N un˜(s) − g(s))ds (28)
0

where λ is the Lagrangian multiplier which can be found implementing using variational theory i,e δun = 0 .

A. Variational method to Duffing Equation

Duffing equation is a classic example in non linear dynamics to illustrate numerical procedure. Duffing equation
is an example of a dynamical system that exhibits chaotic behavior. This equation is a non linear second order
differential equation used to model certain damped and force driven oscillator given by

ü + δ u̇ + αu + βu3 = γcos(ωt) (29)

˙ 0) = 0 The equation describes the motion of a damped oscillator with more
with initial condition u( t = 0) = 0 , u(t =
complicated potential than in simple harmonic motion. Suppose that the angular frequency of the system is ω, then
we have the following homogeneous linearized equation

ü + ω 2 u = 0 (30)
 7

so that we can write above equation () in following form

ü + ω 2 u + g(u) = 0 (31)

where g(u) = δ u̇ + βu3 − γ cos(ωt) Implementing variational iteration method, we can construct functional as
Z t
un+1 (t) = un (t) + ˜n )dτ
λu¨n (τ ) + ω 2 un (τ ) + g(u (32)
0

where g(u ˜n ) is considered as restricted variation in variational theory i.e δg̃ = 0. Calculating variation with respect
to un , and using restricted variation theory , we have following stationary conditions

τ̈ + ω 2 λ(τ ) = 0
(33)
λ(τ )τ =t = 0

The multiplier can then be identified as

1
λ= sinω (34)
ω
Substituting lambda value into iteration equation , we obtain

1
un+1 (t) = un (t) + sin(τ − t) (35)
ω
Substituting this into iteration formula, we get

1 t
Z
un+1 (t) = un (t) + sin(ω(τ − t)u¨n (τ ) + un (τ ) + βun 3 (τ ) + f (τ )dτ (36)
ω 0

Assuming initial approximate solution of the form

u0 (t) = A cos(ωt) (37)

leads to calculation of residual function as

3 1 1
 
R0 (t) = 1 − ω 2 + A2 A cosωt + cosωt + cos3ωt (38)
4 4 4

And using iteration formulation, one can calculate higher order also

ǫA3
u1 (t) = A cos ωt + ( cos3ωt − cos ωt) (39)
32ω 2

B. Variational Iteration approach to Bratu equation

Due to various engineering and physical importance, Bratu problem has been studied extensively. It arises in a wide
variety of physical applications ranging from chemical reaction theory to radiative heat transfer and nanotechnology.
Because of simplicity , the equation is widely used as benchmark tool for numerical method. The problem can be
easily solved with high accuracy by the variational iteration method. Furthermore, two branches of the solutions can
 8

be simultaneously determined and bifurcation point can be identified with ease, if there is any. The classical Bratu
problem is

∆u + λeu = 0 (40)

with planer version as

¨ + λeu(x) = 0
u(x) (41)

with 0 ≤ x ≤ 1 and with Dirichlet boundary condition u(0) = 0 and u(1) = 0 This equation is used in combustion
modeling. It is easy to establish a variational formulation for Eq () which reads
Z 1
1 ′2
J(u) = u − λeu dx (42)
0 2
Variational iteration method can be successfully applied to bratu type equation to construct correction functional
Z x
dun 2 (s)
un+1 (x) = un (x) + [ + αeu˜n (s) ]ds (43)
0 ds2
n ≥ 0 with restricted variation δ u˜n = 0 to find multiplier value. Using Iteration Variation method gives,
Z x
dun 2
un+1 (x) = un (x) + (s − x)[ 2 + αeun (s) ]ds (44)
0 ds
Batiha et al [11] showed using numerical results that variation iteration method is more effective than Adomian
decomposition method.

V. MODIFIED LINDSTEDT POINCARE METHOD

In perturbation theory, Lindstedt-Poincare method is a technique of uniformly approximating periodic solutions to

ordinary differential equations when regular perturbation method fails. This method removes secular terms without
bound- arising straightforward application of perturbation theory to weakly non linear problems with finite oscillatory
solutions. Lindstedt-Poincare method [60, 61] gives uniformly valid asymptotic expansion for the periodic solutions of
weakly nonlinear oscillations. This method does not work for strongly non linear term. The basic idea in this method
is to introduce a new variable.

τ = ω(ǫ)t (45)

where ω is the frequency of the system. Then, any general nonlinear oscillator equation can be written as

ω 2 mü + ω0 2 u + ǫf (u) = 0 (46)

We can expand frequency in power series in ǫ. Cheung et al [20] introduced a new parameter τ . All these modifications
are valid for conservative system. if we write the transformation as

τ = t + f (ǫ, t)

with f (0, t) = 0 , the differential equation can be written as

2
∂f ∂ 2 u ∂2u

ü = 1+ = G(ǫ, τ ) (47)
∂t ∂τ 2 ∂τ 2
 9

The best example is unforced, undamped Duffing equation. A number of variants of the classical perturbation
methods have been proposed to analyze conservative oscillators. Burton [16] defined a parameter α = α(ǫ) in such
a way that asymptotic solutions in power series in α converges more quickly than asymptotic solution. Lindstedt -
Poincare method gives uniformly valid asymptotic expansion for periodic solutions of weakly non linear oscillations.
However, this technique does not work in case of strongly non linear terms.

The basic idea of Lindstedt - Poincare method is to introduce a new variable in terms of oscillation parameter
τ = ω(ǫ)t (48)
where ω is the frequency of the system. Equation () then becomes
ω 2 mu′ + ω0 2 u + ǫf (u) = 0 (49)
Here prime denotes derivative with respect to τ . The frequency is also expanded in powers of ǫ. i.e
ω = ω0 + ǫω1 + ǫ2 ω2 + · · · (50)
It is of interest to note that a much more accurate relation for frequency ω can be found by expanding ω 2 in power
series of ǫ. All these modifications are limited to conservation system, instead of linear transformation.
We can rewrite the transformation of standard Lindstedt Poincare method in the form
τ = t + f (ǫ, t)
(51)
f (0, t) = 0

where f (ǫ, t) is an unknown function(not functional). We can also see from the following examples that the identifi-
cation of the unknown function f is much easier than that of functional in Dai method [? ]. From the relation, we
obtain
2
∂2u ∂f ∂ 2 u ∂2u
= (1 + ) = G (52)
∂t2 ∂t ∂τ 2 ∂τ 2
where
2
∂f

G(ǫ, t) = 1+ (53)
∂t
Applying taylor series, we have,
G(ǫ, t) = G0 + ǫG1 + ǫ2 G2 + · · · (54)
where Gi i=1,2,3 · · · can be solved from equation () . If we simplify new parameter as
τ = f (t, ǫ) (55)
with f(0,t) as t, then we have
2
∂2u ∂f ∂ 2 u
2
=( ) (56)
∂t ∂t ∂τ 2
2
∂f
And expanding in power series
∂t
2
∂f

= 1 + ǫf1 + ǫ2 f2 + · · · (57)
∂t
 10

If we consider Van der Pol equation, i.e

ü + u − ǫ(1 − u2 )u̇ = 0 (58)

With above transformation and introducing transformation variable, we can calculate period of oscillation for ǫ ≪ 1
very accurately as that obtained by regular perturbation method. For large ǫ , approximate period has same feature
as exact one with relative error < 37.5%. For even larger value of ǫ , modified Lindstedt-Poincare method gives very
good result when classical method fails to work. Parameter expansion method [42] developed by He alternatively can
deal this type of situation very successfully. This method does not require time transformation like Lindstedt-Poincare
method . Main idea is to expand solution in terms of perturbation parameter. To search for periodic solution, , we
can assume

u = u0 + ǫu1 + ǫ2 u2 + · · · (59)

Implementing this method in pendulum equation i.e

ü + sinu = 0 (60)

One solves for time period using Bessel function. We used parameter expansion method in Classical Van der Pol -
Duffing equation which arises in electrical damped oscillator circuit given by

ẍ − µ(1 − x2 )ẋ + x + αx3 = 0 (61)

˙ = x(0)
with initial boundary condition x(0) = x0 and x(0) ˙ Here dot denotes time derivative with µ and α as positive
unknown coefficient. Applying parameter expansion method to solve this equation

x = x0 + ǫx1 + ǫ2 x2 + · · · (62)

ǫ is the expansion parameter . We expand the coefficients of the system in terms of ǫ we obtain

µ = ǫµ1 + ǫ2 µ2 + · · ·
α = ǫα1 + ǫ2 α2 + · · · (63)
2 2
1 = ω + ǫω1 + ǫ ω2 + · · ·

ω is the frequency of non linear oscillation. Substituting these into differential equation, we obtain

x¨0 + ω 2 x0 = 0
s s
3 3 (64)
x = Asin 1 + α(A2 + B 2 ) + Bcos 1 + α(A2 + B 2 )t
4 4

with frequency value obtained same as perturbation method. Nayfeh [60] studied free oscillation with non linearity
of quadratic and cubic order.

VI. PERTURBATION METHOD

Perturbation method has proved to be powerful tool in solving various boundary layer problems , interface flow or
nano fluid mixture [? ]. In regular perturbation problem, as ǫ → 0 (perturbation parameter) , the solution tends to
the solution for ǫ = 0 . Consider the equation

ÿ + 2ǫẏ − y = 0 (65)
 11

with boundary condition y(0) = 0, y(1) = 1 and 0 ≪ ǫ ≪ 1. The general solution is

e m1 x − e m2 x
y(x, ǫ) = (66)
e m1 − e m2
√ √
where m1 = −ǫ + 1 + ǫ2 and m2 = −ǫ + 1 + ǫ2 As epsilon → 0, the solution becomes
sinh(x)
y→ (67)
sinh(1)
We can expand solution as perturbative solution. This is regular perturbation problem. In boundary layer theory ,
perturbative methods prove to be very useful tool in fluid dynamics. Part of the problem can be linearized as part
of perturbation procedure. The ideas stem from rigorous background of Liapounov - Schmidt theory of bifurcation.
Falkner - Skan equation is the generalized equation for Blasius solution for wedge flow in fluid mechanics i.e flows are
not parallel to the plate. In physics and fluid mechanics, Blasius boundary layer describes steady two dimensional
laminar boundary layer. Considering Falkner - Skan equation,
...
f + f f¨ + β(1 − f˙2 ) = 0 (68)

with Mixed boundary condition f (0) = f˙(0) = 0 and f˙(∞) = 1 to have solution f (η) for 0 < η < 1. If we expand
solution in terms of perturbation parameter as
f (η) = f0 (η) + (β − β0 )f1 (η) + (β − β0 )2 f2 (η) + · · · (69)
On equating coefficients of powers of small parameter β − β0 in Falkner Skan equation, this gives us a succession of
linear inhomogeneous problems to solve
...
Lf1 = f 1 + (β − β0 )f1 (η) + (β − β0 )2 f2 (η) + · · ·
(70)
Lf2 = β0 f˙2 − f1 f¨1 + 2f˙0 f˙1
1

We can implement perturbation method to find solution. And we obtain solution of following form
1/2 ′
f (η) = f0 (η) ± (β − β0 )(I1 /I2 ) f0 (η) + O(β − β⋆ ) (71)
with existence of bifurcation near β = β0 . The Blasius boundary layer solution is only one member of a larger family
of boundary layer solution known as Falkner Skan flow. Considering Blasius boundary layer equation applied to two
dimensional flow over a wedge with slip flow boundary, one can implement perturbative method very successfully in
numerical analysis [54].

VII. ITERATIVE PERTURBATION METHOD

In this section, we will illustrate a new perturbation technique coupling with the iterative method. Consider the
following nonlinear oscillation:
u′ + u + ǫf (u, u′ ) = 0 (72)
We write the equation in the following form
u′ + u + ǫug(u, u′) = 0 (73)
We construct an iteration formula for the above equation
un + 1′ + un+1 + ǫun+1 g(un , un ′ ) = 0 (74)
 12

For nonlinear oscillation, Eq () is of Mathieu type. We will use the perturbation method to find approximately un+1 .
Consider Eq () and assume that initial approximate solution can be expressed as u0 = Acosωt where ω is the angular
frequency of oscillation. We rewrite Eq () approximately as follows
d2 u
+ ǫA2 ucosωt2 = 0 (75)
dt2
or
d2 u 1 2
+ A uǫA2 ucos2ωt = 0 (76)
dt2 2
which is Mathieu type. Suppose that
u = u0 + ǫu1 + ǫ2 u2 (77)
To obtain a second order approximate solution, we substitute u0 and u1 we find
c1 A2
c2 = (78)
32ω 2
To obtain approximate solution with higher accuracy, we

VIII. PARAMETRIZED PERTURBATION METHOD

An expanding parameter is introduced by a linear transformation

u = ǫv + b (79)
where ǫ is the introduced perturbation parameter, b is a constant.

IX. MULTIPLE SCALE LINDSTEDT POINCARE METHOD(MMS)

The method of Multiple Scale (MMS) is routinely applied to many weakly nonlinear systems of both ordinary
and partial differential equations. For delay differential equations or retarded functional differential equations in
the neighborhood of Hopf bifurcation , this method had been applied very successfully without applying manifold
reduction scheme. Manifold reduction as suggested by many authors [17] eliminates a priori many exponentially
decaying part of the solution leaving behind slowly evolving dynamics on the center manifold. MMS method can
efficiently eliminate rapidly decaying part in the center manifold. This method is very efficient in system with time
delay [66]. If we consider autonomous equation with limit cycle
π
ẏ(t) = −αy(t − ) − y 3 (t) (80)
2
with α > 0 . The characteristic equation for the linearized form is

λ̃eλ̃ + α̃ = 0 (81)
π π
For 0 < α̃ < , this equation undergoes Hopf bifurcation at α̃ = or α = 1. To study equation near bifurcation, we
2 2
take α = 1 + ǫ and equation () becomes
π
ẏ(t) = −(1 + ǫ)y(t − ) − y 3 (t) (82)
2
 13

with 0 < ǫ ≪ 1. If we allow transformation solution y = ǫx , equation () becomes

π π
ẋ(t) = −x(t − ) − ǫ[x(t − ) + x3 (t)] (83)
2 2
Hopf bifurcation occurs at ǫ = 0 and a stable periodic solution exists for small ǫ > 0. To find solution near bifurcation
point, we can define multiple time scale to implement MMS of the form

x(t) = X0 (t, T0 , T1 , T2 ) + ǫX1 (t, T0 , T1 , T2 ) + ǫ2 X2 (t, T0 , T1 , T2 ) + · · · (84)

Substituting the above solution in the differential equation, we obtain

∂X0 ∂X1 ∂X0 π ∂X0 π π π π

 
+ǫ + − (t − ) = −X0 (t − ) − ǫ(t − ) + X0 (t − ) + X0 3 (t)] + O(ǫ2 ) (85)
∂t ∂t ∂t 2 ∂t 2 2 2 2

This equation has of the form

X0 = Asint + Bcost (86)

And if one solves for solution , one observes MMS result is very close to Direct Numerical Simulation result for
oscillation. MMS method can be implemented to strongly nonlinear case where the method needs to be carefully
implemented.

X. HOMOTOPY PERTURBATION METHOD

Homotopy Analysis method (HAM) is a semi analytical technique to solve non linear/partial differential equations.
The homotopy analysis method employs the concept of homotopy from topology to generate a convergent series
solution for non linear systems. This is enabled by utilizing a homotopy - Maclaurin series to deal with the non
linearity of the system. Homotopy perturbation method have been applied to various kinds of non linear problems.
The main property of the method is its flexibility and ability to solve non linear equations accurately and conveniently
used. Since perturbation method depends on small parameter which is very difficult to be found in non linear problems,
Variational Perturbation Method(VPM) and Homotopy Perturbation Method(HPM) have been introduced to solve
non linear problems. HPM method is being as most effective method in non linear analysis of engineering problems.
In this method, according to homotopy technique, a homotopy with embedded parameter p ∈ [0, 1] is constructed ,
so the parameter embedded is small , so the method is called homotopy perturbation method. This method can take
the full advantage of homotopy techniques and perturbation method. This method is widely used in Duffing equation
with high order of non linearity. Consider a non linear equation in the form

A(u) + f (r) = 0 (87)

where r ∈ Ω with boundary conditions

∂u
B(u, )=0 (88)
∂n
with r ∈ Γ. where A is general operator and B is boundary operator. f(r) is a known analytic function and Γ is t he
boundary of the domain Ω. The operator , in general be divided into two parts L and N, where L is the linear and
M is the non linear part. Equation() can be written as

L(u) + N (u) − f (r) = 0 (89)

 14

In case of non linear equation, it has no small parameter. We can embed an artificial parameter in the above equation
as

L(u) + pN (u) − f (r) = 0 (90)

In this method, according to homotopy technique, a homotopy with embedded parameter p ∈ [0, 1] is constructed , so
the parameter embedded is small . The above equation is called perturbation equation with artificial small parameter
p and it assumes that u can be expressed as a series of power of p.

u = u0 + pu1 + p2 u − 2 + p3 u3 + · · · (91)

In order to use homotopy perturbation, a suitable construction of homotopy equation is of vital importance. For
example ,

y′ + y2 = 1 (92)

with t > 0 with initial condition with initial condition y(0) = 0. We embed an artificial parameter p such that

y ′ + py 2 = 1 (93)

This equation does not have approximate solution in power of p. So, Homotopy technique proposed by Liu et al [? ]
is being applied here which satisfies

(1 − p)[L(v) − L(u0 )] + p[A(v) − f (r)] = 0 (94)

which satisfies the boundary condition. Transformation of p from zero to unity is equivalent to deformation in topology
and L(v) − L(u0 )andA(v) − f (r) are homotopic. v(r,p) changes from u0 tou(r) homotopically. It is very natural to
express v as

v = v0 + pv1 + p2 v2 + c . . . (95)

In the limit p → 1,

u = lim p → 1v = v − 0 + v1 + v2 + · · · (96)

Convergence of this series has been proved in [33], [34]. Using method of weighted residual methods, one can obtain
p value from the above equations. We cite some examples. Consider very simple system

y′ + y2 = 0 (97)

x ≥ 0 , x ∈ Ω with boundary condition y(0)=1. with exact solution .

We construct solution following homotopy Ω × [0, 1] which satisfies

(1 − p)(y ′ − y0′ ) + p(y ′ + y 2 ) = 0 (98)

with initial approximation y0 = 1. Suppose the solution has the form

y = y0 + py1 + p2 y2 + · · · (99)

Now we consider a typical non linear equation, i.e Duffing equation with non linearity with higher order non linearity
. Duffing equation represents simple harmonic motion of damped pendulum in may cases.

d2 u
+ u + ǫu5 = 0 (100)
dt2
 15

t ∈ Ω with initial condition u(0) = u′ (0) = 0. we construct a homotopy Ω × [0, 1] which satisfies
L(v) − L(u0 ) + pL(u0 ) + pǫv 5 = 0 (101)
d2 u
where Lu = + u. We assume initial guess of equation () as
dt2
u0 (t) = A cosαt (102)
where α(ǫ) is a unknown constant with α(0) = 1.
with similar approximate solution for v we have,
L(v0 ) − L(u0 ) = 0
(103)
L(v1 ) + L(u0 ) + ǫv0 5 = 0
with v1 ′ (0) = v1 (0) = 0. From equation () and () we obtain,
v0 = A cosαt
2 5 (104)
d v1 5 ǫA
 
+ v1 + A −α2 + 1 + ǫA4 cos αt + (cos5αt + 5 cos3αt)
dt2 8 16
The solution of the above equation can be obtained using variational Iteration method. i.e
Z t
5
v( t) = sin(t − τ )[A(−α62 = 1 + ǫA4 ) cosατ (105)
0 8
The constant α can be identified as weighted residual (least square method, collocation method, Galerkin method).
So, we found v1 (t) as
5 A ǫA5 5ǫA5
 
v1 (t) = −α2 + 1 + ǫA4 2
cosαt + 2
cos5αt + (106)
8 α −1 16(25α − 1) 16(9α2 − 1)
Finally we calculate u1 (t)
u1 (t) = v0 (t) + v1 (t)
5 5 (107)
5ǫA ǫA 5ǫA5
u1 (t) = cosαt + cos 5αt + c cos3αt
8(α2 − 1) 16(25α2 − 1) 16(9α2 − 1)
Nofel et al [62] implemented homotopy perturbation method in obtaining approximate periodic solutions for non
linear heat conduction equation and Van der Pol equation. Biazar et al [14] implemented homotopy perturbation
method in detail for solving hyperbolic equations. Hyperbolic partial differential equations play a dominant role in
many branches of science in engineering. Wave equation is typical example , which models problems such as vibrating
string fixed at both ends , longitudinal vibration in bar, propagation of sound waves, electromagnetic waves. They
also occur in fluid mechanics. Euler equations in gas dynamics are typical examples of hyperbolic equations. They
considered following homotopy equations for the quasi linear hyperbolic partial differential equation
L(u) + N (u) − f (r) = 0 (108)
Using homotopy technique, one can construct homotopy equation as
H(u, p) = (1 − p) (109)
p ∈ [0, 1] is the homotopy parameter with u0 as initial guess satisfying boundary conditions. Considering the problem
of combined convective -radiative cooling of lumped system represented by equation.
 16

XI. SINGULAR PERTURBATION METHOD

Singular perturbation problems often arise in fluid mechanics with thin boundary layer , the properties of slightly
viscous fluid dramatically changes on both sides of boundary layer so that fluid exhibits multiple spatial scales.
Reaction diffusion systems are classic example of this system. A wide variety of techniques have been applied to solve
singular perturbation problem. In one form or other, these techniques consist of dividing the problem into an inner
solution(or boundary layer) and an outer layer problem. Express the inner and outer as asymptotic series , equating
various terms on inner and outer solutions in the series. Finally, one obtains solution combining the inner and outer
solutions in some fashion to obtain uniformly valid solution. Commonly the outer solution is obtained as a function of
small(perturbation) parameter with ǫ → 0. Kaplun [44] first developed ideas about singular perturbation in studying
asymptotic solution of Navier Stokes equation for low Reynold number based on limit expansion idea. Under outer
limit, finite body shrinks to a point which can not disturb the flow or presence of small body causes perturbation which
is small everywhere except surface. The inner limit is the one where viscosity tends to infinity. Mathematical technique
for constructing asymptotic expansion consists in first finding the leading term of outer solution and matching inner
solution to outer approximation. Kaplun considered partially ordered set of equivalence classes of functions f(Re). To
each such f(Re), corresponding limit process is defined and implemented in N-S equations. Boundary value technique is
being introduced by Roberts [68] to obtain asymptotic matched coefficients in various situations. Boundary technique
partitions original singular perturbation equation into two problems , an inner solution differential equation and an
outer solution differential equation. The solution of the outer equation provides provides terminal conditions for inner
equation. And in turn, solution for inner equation provides initial condition for outer equation. The problem is
solved iteratively for various values of terminal point and terminal boundary conditions of inner solution until profile
stabilizes and original boundary condition is satisfied. Consider following singular perturbation equation

ǫy ” (x) + y ′ (x) + y(x) = 0 (110)

with boundary condition y(0) = α , y(1) = β and 0 ≤ x ≤ 1. Converting the original singular perturbation equation
to the outer solution with right boundary condition, outer equation reduces to

y ′ (x) + y(x) = 0 (111)

x
with y(1) = β. Parameter scaling is very important to develop inner differential equation. Here we set t = and
ǫ
equation becomes

y(x) = Y (t)
Y ′ (t) (112)
y ” (x) =
ǫ2
to obtain differential equation of inner solution

Y ” (t) + y ′ (t) + ǫY (t) = 0 (113)

with boundary condition . choosing tf as terminal point for inner solution, one obtains inner equation as initial value
problem with

y(tf ǫ) = Y (tf ) = β̃ (114)

The original boundary value perturbation equation becomes initial value equation and can be solved numerically
iteratively. Singular perturbation problem is one where perturbed problem is qualitatively different from unperturbed
one. Solution of singular perturbation problems involving differential equations often depend on several widely different
length or time scale. Such problems can be divided into two broad classes: layer problem being treated using method
of matched asymptotic expansions (MMAE) and multiple scale problem treated by method of multiple scale (MMS).
 17

Prandtl boundary layer theory for high Reynold number flow of viscous fluid over a solid body is classic example of
MME method and semi classical limit of quantum mechanics is an example of MMS problem. An asymptotic expansion
as z → ∞ of a complex function with essential singularity is valid only in wedge shaped region. MMS method is
needed for problems in which the solutions depend simultaneously on widely different scales. A typical example is
the modulation of oscillatory solution over time scale that is much greater than period of oscillation. Considering
boundary layer problem in fluid mechanics , Boundary layer theory and asymptotic matching are collection of singular
perturbation methods for constructing a uniformly and globally valid solution by calculating the separated outer and
inner solutions and matching them at boundary. Considering non linear boundary layer problem .
For example , Navier Stokes equation can be expanded ( a highly non linear equation) in asymptotic power series in
small parameter ǫ which is determined as a function of Reynold number by asymptotic technique. Flow over constant
pressure flat plate turbulent boundary layer had been studied by asymptotic technique [79], [55]. Starting with two
dimensional Navier Stokes equation in asymptotic power series of ǫ in both region i.e inner and outer region and
then matching both equations in the overlap region in asymptotic manner. We assume, asymptotic parameter ǫ as
function of Reynold number and goes to zero as Reynold number → ∞. And we can write velocity field in Navier
Stokes equation as
U = ĩ + ǫU1 (X − T, Y, Z) + · · · (115)
Matching outer and inner equations at boundary , Meller et al [55] developed asymptotic expansion identity for
velocity field


U = ǫ κ−1 ln(ǫRe ) + Bi (116)
Meller showed in their asymptotic analysis that overlapping zone exists in solution and it indicates coupling between
inner and outer layer at the turbulent boundary layer. Method of matched asymptotic expansion is particularly used
for singular perturbed differential equation.
Fundamental idea of asymptotic ( or perturbative) analysis is to relate unknown solution of original complex problem
to the known simple solution using simple transition. Consider a general equation,
L(u) + ǫN (u) = 0 (117)
where L(u) = 0 is an equation that can be readily solved with the solution u0 . We assume N is a bounded operator
defined on a space in real valued function. We need to find u(ǫ) to above equation with condition u(ǫ) → 0 as ǫ → 0.
And since L(u) is a linear operator
L(c − 1u + c2 v) = c1 L(u) + c2 L(v) (118)
holds and the solution can be found as
u(ǫ) = u − 0 + ǫu1 + ǫ2 u2 + · · · (119)
This is regular perturbation method. For singular perturbed boundary problem, For boundary problem, we employ
asymptotic expansion(outer and inner ). We will use MMS method to construct leading asymptotic terms of the
outer and inner expansion of the solution to the singularly perturbed solution. Passing to the limit, as ǫ → 0, we
observe that Dirichlet boundary condition disappear and we arrive to following limit. All non trivial solutions have
singularities and behavior of functions can be analyzed through matching of the outer with inner at the overlap region.
To construct inner asymptotic expansion of the solution, generally we pass through the stretched coordinates ξ j and
construct as
uǫ (x) = µ0 (ǫ)w0 j (ξ j + µ1 (ξ)w1 j (ξ j ) + · · · (120)
To calculate asymptotic series outer expansion , we construct leading asymptotic term first
w0 j (ξ) = ξ (121)
 18

A. WKB Method

WKB theory is powerful tool to solve linear system to find global uniform solution and can be applied to boundary
layer problem. This method is generally applied in solving linear differential equation and can be implemented
effectively in case of non linear equations also. WKB method is a method of approximating the solution of a differential
equation whose highest derivative is multiplied by a small parameter ǫ. Miura & Kruskal [58] first implemented WKB
method effectively to solve kdV equation in presence of small dispersion parameter. Let us consider general differential
equation

d2 y
ǫ2 + F (x)y = 0 (122)
dx2
using WKB method. We can assume the solution in the form

iu(x)
 
y(x, ǫ) = A(x, ǫ)exp (123)
ǫ

so that differential equation becomes

2 iu iu
(−(u′ )A + ǫi(Au” + 2A′ u′ ) + ǫ2 A” ) exp( + F (x)exp( = 0 (124)
ǫ ǫ
Now let

A(x, epsilon) ≈ A0 (x) + ǫA1 (x) + · · · (125)

And the solution just comes as

Z xp
u(x) = ± F (s)ds (126)
x0

WKB method can be implemented in nth order differential equation [13? ]. One of the most common application
of WKB method occurs in solving Schroedinger equation in quantum mechanics. WKB method is also used in
geometrical optics, Ray theory, various wave patterns specially with Kelvin ship wave. WKB method can be used in
solving Schroedinger- Newton equation which form a class of non linear differential equations. These equations are
obtained by coupling together the ordinary Schroedinger equation and Newton field of equations for the gravitational
potential. The equations describe a particle moving in its own gravitational field , where the field is generated via
classical Newtons field equation. This equation is originally obtained by Penrose [65] . Considering time independent
equations of the form

2m
∆ψ = − (e − φ)ψ
~2 (127)
∆φ = −4πGm2 | ψ |2

where ψ is the wave function and φ denotes grav. potential. E is the energy eigen value. If we define variables as
2m ~ψ
U= (E − φ) and S = p , one may evolve equations
~ 8piGm3

∆S = −SU
(128)
∆U = −S 2
 19

These equations are known as Schroedinger Newton equation. Implementing WKB method to these equations, one
obtains
Z r
x
S(r) = S0 − x(1 − )S(x)U (x)dx
0 r
Z x (129)
x 2
U (r) = U0 − x(1 − )S(x) dx
0 r

In nanoscale semi conductor structure studies, quantum effects occur and have to be taken into account by means
of Schroedinger equation. Computational difficulties develop in numerical calculation of Schroedinger equation. On
the other hand behaviour of the solution requires a refined spatial grid. Wang et al [76] developed WKB-LDG method
in simulating resonant tunneling diode which provides computational and memory cost significantly.

[1] Abassy T.A, El-Tawil M.A & Saleh H.K : The solution of kdV and mkdV equations using Adomian Pade
approximation, Int. J. Nonl. Sci. Num Simulation 5(4), 327-339 (2004).

[2] Abbasbandy S. : A numerical solution of Blasius equation by Adomian decomposition method and compar-
ison to homotopy perturbation method, Chaos Solitons Fractals 5(1) , 257-260 (2007).

[3] Acton J.R. & Squire P.T : Solving Equation with Physical Understanding , Adam Hilger Ltd., Boston (1985).

[4] Adomian G. : A Review of the Decomposition method in Applied Mathematics, J. Math. Anal. & Appl.
135, 501-544 (1988).

[5] Agrawal V.P. & Denman H.: Weighted Linearization Technique for Period Approximation in large amplitude
nonlinear oscillations, 463- (1985).

[6] Aksoy Y. & Pakdermirli M. : New Perturbation -iteration solutions for Bratu type equations , Computers &
Mathematics with Applications 59, 2802-2808 (2010).

[7] Argatov I.I, & Nazarov S.A. : Introduction to Asymptotic Modeling in Mechanics, Polytechnics , St. Petersberg.

[8] Awrejcewicz J., Andrianov I.V. & Manevitch L.I.: Asymptotic approaches in Nonlinear Dynamics: New Trends
and Applications (Springer Verlag, Heidelberg) (1988).

[9] Bai C.L. & Zhao H.: Generalized Extended tanh-function method and its application, Chaos Solitons Fractals
27(4), 1026-1035 (2006).

[10] Bao F.B. & Lin J.H.: Linear Stability analysis for various forms of one dimensional burnett equations, Int.
J. Nonl. Sci. Num. Simulations 6(3), 295-303 (2005).

[11] Batiha B.: Numerical Solution of Bratu Type equations by Variational Iteration method, Jouurnal of
Mathematics & Statistics 39(1), 23-29 (2010).

[12] Bellman R. : Perturbation Techniques in Mathematics, Physics & Engineering , (Holt, Rinehart & Winston
Inc., (1964).

[13] Bender C.M.: Advanced Mathematical Methods for Scientists & Engineers, Asymptotic Methods and
perturbation Theory (1999).
 20

[14] Biazar J. & Ghavini H.: Homotopy Perturbation method for solving Hyperbolic partial differential equations,
Computers & Mathematics with Applications 56, 453-458 (2008).

[15] Bildik N. & Konuralp A.: The use of variational iteration method, differential transform method and
Adomian decomposition method for solving different types of nonlinear partial differential equations, Int.
J. Nonl. Sci. Num Simulation 7(1), 65-70 (2006).

[16] Burton T.D.: A perturbation method for certain non linear oscillators, Int. J. Nonlinear Mechanics 19(5),
397-407 (1984).

[17] Cambell S.A. , Belair J., Ohira T. & Milton J.: Complex Dynamics and multistability in a damped harmonic
oscillator with delayed negative feedback, Chaos, 5(4), 640-645 (1995).

[18] Chung K.W, Chan C.L & Mahmoud G.M: A perturbation incremental method for strongly Nonlinear au-
tonomous oscillators with many degrees of freedom, Nonlinear Dynamis 28, 243-259 (2002).

[19] Chen X.P, Chung K.W & Xu Z.: Symmetric boundary knot method for membrane vibrations under mixed-
type boundary conditions, Int. J. Nonl. Sci. Num. Simulation 6(4), 421-424 (2005).

[20] Cheung Y.K.Chen S.H & Lau S.L.: A modified Lindstedt-Poincare method certain strongly nonlinear oscilla-
tors, Int. J. Nonl. Mech. 26(3/4), 367-378 (1990).

[21] Cole J.D Perturbation methods in applied mathematics, Blaisdell Pub. Co., Waltham Mass.

[22] Cveticanin L. : Homotopy perturbation method for pure nonlinear differential equation, Chaos Solitons in
Fractals 30(5), 1221-1230 (2006).

[23] Danaf T.S El, Ramadan M.A & Alaal F.E.I.A .: The use of Adomian decomposition method for solving
regularized long wave equation, Chaos, Soliton , Fractals 26(3), 747-757 (2005).

[24] Delamotte B.: Non perturbative(but approximate) method for solving differential equations and finding
limit cycles, Phys. Rev. Lett 70(22), 3361-3364 (1993).

[25] Dutta. R.: Asymptotic Method in Fluid Fluid Interface Stability Analysis , Applied Mathematics & Computa-
tion(communicated) (2016).

[26] Dutta R.: Numerical Studies(stochastic method) of Particle Laden Flow in Dispersed Phase, Int. J.
Numerical Method in Fluids, 64(8), 973-978 (2010).

[27] Dyke V.: Perturbation Methods in Fluid Mechanics, Academic Press, New York.

[28] Finlayson B.A. : The Method of Weighted Residual and Variational Principles, Academic Press, New York, (1972).

[29] Gu B.H. & Zhang H.P. : Ballistic impact of laminated composite penetrated by cylindrical rigid projectile: a
finite element simulation, Int. J. Nonl. Sci. Num. Simulation 6(3), 215-222 (2005).

[30] Higham J. D. & Kloden P ; Numerical Methods for Non linear Stochastic Differential Equations with Jumps
, Numeriche Mathematik .

[31] He J.H. : Variational Theory method - a kind of non linear analytical technique: some examples, Int. J.
Nonl. Mechanics 34(4), 699-708 (1999).

[32] He J.H.: Homotopy Perturbation Method for solving boundary value problems, Phys. Lett. A 350(4-6),
87-88 (2006).
 21

[33] He J.H. : Homotopy perturbation method for bifurcation of nonlinear problems, Int. J. Nonl. Sci. Num.
Simulations 6(2), 207-208 (2005).

[34] He J.H.: A simple perturbation approach to Blasius equation, Appl. Math. Comput. 140(2-3), 217-222 (2003).

[35] He J.H. : Homotopy perturbation method : a new nonlinear analytical technique , Appl. Math. Comput.
135(1), 73-79 (2003).

[36] He J.H.: Iteration perturbation method for strongly nonlinear oscillations, J. Vib. Cont. 7(5) 631-642 (2001).

[37] He J.H, Wan Q & Guo Q.: An Iteration Formulation for Normalized diode Characteristics, Int. J. Circ.
Theor. Appl. 32(6), 629-632 (2004).

[38] He J.H.: Preliminary Report on the energy balance for non linear oscillations Mech. Res. Comm. 29(2-3),
107-111 (2012).

[39] He J.H.: Modified Lindstedt-Poincare methods for some strongly non linear oscillations Part II, a new
transformation, Int. J. Nonl. Mech. 37(2), 309-314 (2002).

[40] He J.H.: Variational approach to the Lane-Emden equation, Appl. Math. Comput. 143(2-3) 539-541 (2003).

[41] He J.H.: Limit Cycle and bifurcation of nonlinear problems, Chaos Solitons & Fractals, 26(3), 827-833 (2005).

[42] He J.H.: Bookkeeping Parameter in Perturbation Methods, Int. J. Nonl. Sci. , 2(3), 257-264 (2001).

[43] Matinfar M. & Ghasemi M.: Construction of Solitary Solution and Compacton-like convolution product
nonlinearities, J. Computational Methods Applied Research in Mechanical Engineering 2(2), 59-68 (1988).

[44] Kaplun S. : The Role of Coordinate Systems in Boundary Layer Theory, PhD Thesis, 1954.

[45] Kawai T & takei Y (2005), Algebraic Analysis of Singular Perturbation Theory, Trans Goro Kato, AMS,
Providence (2005).

[46] Kim Y.H, Kim K. & Kim H.S.: Effects of moment arms on spinal stability subjected to follower roads: A
finite element analysis Int. J. Nonl. Sci. Num. Simulation, 6(3), 311-316 (2005).

[47] Kiymaz O. & Cetinkaya A.: Variational Iteration Method for a class of non linear differential Equations , Int.
J. Contemp. Math. Sciences 5(37), 1819-1826 (2010).

[48] Li P.S & Wu B.S.: An Iteration approach to non linear oscillations of conservative single -degree -of -freedom
systems, Acta Mechanica 170, 69-75 (2004).

[49] Liao S.J.: Boundary element method for general nonlinear differential operators, Eng. Anal. Boundary Elem.
20(2), 91-99 (1997).

[50] Liu H.M.: Approximate period of nonlinear oscillators with discontinuities by modified Lindstedt-Poincare
method , Chaos, Solitons Fractals 23(2), 577-579 (2005).

[51] Liu G.L.: New Research Directions in Singular Perturbation Theory: Artificial Parameter Approach and
Inverse Perturbation Technique, National Conf. of 7th Modern Mathematics & Mechanics Sanghai, 47-53 (1997).

[52] Liu H.M.: Variational approach to nonlinear electrochemical system , Int. J. Nonl. Sci. Num. Simul 5(1),
95 (2004).

[53] Marinca V. & Herisanu N.: A Modified Iteration perturbation method for some nonlinear oscillation problems,
 22

Acta Mechanica 184(1), 231-242 (2006).

[54] Martin M. & Boyd I.D.: Falkner Skan flow over a wedge with slip boundary conditions, 47th AIAA Aerospace
Sciences Meeting , Orlando, Florida, 5-8 January, Florida.

[55] Meller G.L.: The large Reynold number , asymptotic theory of turbulent boundary layers, Int. J. Eng.
Science 10, 851-873 (1972).

[56] Mickens R.E.: An Introduction to Non linear Oscillations , Cambridge University Press (1981).

[57] Millikan C.B.: A critical discussion of turbulent flows in channels and circular tubes , Proc. of 5th Int. Congress
of Appl. Math ed J.P Hartog & H. peters , Wiley, New York, 386 (1936).

[58] Miura R. M. & Kruskal M. : Application of a Non Linear WKB method to kdV Equation, Siam J. Appl.
Math 26(2), 376-395 (1974).

[59] Momani S. & Odibat Z.: Numerical Comparison of methods for solving linear differential equations of
fractional order, Chaos Solitons Fractals , 31(5), 1248-1255 (2007).

[60] Nayfeh A.H (1981), Introduction to Perturbation Techniques, ohn Wiley & Sons, New York (1981).

[61] Nayfeh A.H & Mook D.T.: Nonlinear Oscillations, John Wiley , New York, 1979.

[62] Nofel T.A.: Application of the Homotopy Perturbation Method to Nonlinear Heat-Conduction and
Fractional Van der Pol Damped Non linear Oscillator, Applied Mathematics letters 5, 852-861 (2014).

[63] Odibat Z.M. & Momani S.: Application to Variational Iteration Method to Non Linear Differential Equation
of Fractional Order, Int. J. Nonl. Sci. & Num. Simulation 7(1), 27-34 (2006).

[64] Pakdemirli M., Karahan M.M.F & Boyaci H.: A new perturbation algorithm with better convergence prop-
erties: Multiple Scales Lindstedt Poincare method, Mathematical and Computational Applications 14, 31-44 (2009).

[65] Penrose R. : Quantum Computation, Entanglement and State Reduction , Phil Trans. R. Soc. A 356, 1927
(1998).

[66] Plaut R.H & hseih J.C.: Non-linear structural vibrations involving a time delay in damping, Journal of Sound
and Vibration, 117(3), 497-510 (1987).

[67] Ramos J.I. : Piecewise-Linearized methods for oscillators with limit cycles , Chaos, Solitons & Fractals 27(5),
1229-1238 (2006).

[68] Roberts S.M.: A Boundary value technique for Singular Perturbation Problems, J. Math. Analysis Appl. 87,
489-508 (1982).

[69] Ren Y.J. & Zhang H.Q.: A generalized F-expansion method to find abundant families of Jacobi Elliptic
Function solutions of the (2+1) dimensional Nizhnik -Novikov-Veselov equation, Chaos Solitons Fractals
27(4), 959-979 (2006).

[70] Shohat J.: On Van der Pols and Related nonlinear differential equations, J. Appl. Phys. 15, 568-574 (1944).

[71] Siddiqui A.M., Mahmood R. & Ghori Q.K. : Coutte and Poiseuille flows for Non Newtonian fluids, Int. J.
Nonl. Sci. Num Simulation 7(1), 15-26 (2006).

[72] Sweilam N.H. & Khader M.M.: Variational iteration method for one dimensional nonlinear thermo elasticity,
Chaos Solitons Fractals, 32(1), 145-149 (2007).
 23

[73] Taniuti T. & Wei C.C.: Reductive Perturbation Method in Non Linear Wave propagation, J. Phys. Soc.
Japan 24, 941-946 (1968).

[74] Turgut O. & Yildirm A.: Application of He’s semi inverse method to the non linear Scroedinger Equation,
Computers & Mathematics with Applications 54(7-8), 1039-1042 (2007).

[75] Wang M.L & Zhang H.Q.: Further improved F-expansion method and new exact solutions of Konopelchenko-
Dubrovsky equation, Chaos Soliton and Fractals 25(3), 601-610 (2005).

[76] Wang W. & Wang Shu C. : The WKB Local Discontinuous Galerkin Method for the SImulation of
Schroedinger Equation in a Resonant Tunneling Diode, J. Sci. Computing 40(1), 360-374 (2009).

[77] Wazwaz A.M. : A comparison between Variational Iteration Method and Adomian Decomposition Method
, J. Comput. Appl. Math. 207(1), 129-136 (2007).

[78] Xu Y., Li Y., Liu D. & Wantao J.: Responses of Duffing oscillator with Fractional Damping and Random
Phase, Nonlinear Dynamics 74, 745-753 (2013).

[79] Yajnik K.S.: Asymptotic theory of turbulent shear flows, J. Fluid Mech 42, 411-427 (1970).

[80] Yu P. , Hu S.Y., Chen L.Q. & Du Q.: An Iterative perturbation scheme for treating inhomogeneous elasticity
in phase field models, J. Comput. Phys. 208, 34-50 (2005).

[81] Zhang J. , Yu J.Y. & Pan N. : Variational Principles for Nonlinear Fiber Optics , Chaos Solitons Fractals 24(1)
, 309-311 (2005).
[82] Zhang C.R., Zu Y.G & Zheng B.D.: Stability and Bifurcation of a discrete red blood cell survival model, Chaos
Solitons & Fractals 28(2), 386-394 (2006).