Professional Documents
Culture Documents
Royle FinalReport
Royle FinalReport
Royle FinalReport
Approved:
_________________________________________
Ernesto Gutierrez-Miravete, Project Adviser
CONTENTS
Application of Perturbation Methods to Approximate the Solutions to Static and Nonlinear Oscillatory Problems .......................................................................................... i
LIST OF TABLES ............................................................................................................ iv
LIST OF FIGURES ........................................................................................................... v
ACKNOWLEDGMENT ................................................................................................. vii
ABSTRACT ................................................................................................................... viii
1. Introduction.................................................................................................................. 1
1.1
Background ........................................................................................................ 1
1.2
Project Scope...................................................................................................... 3
2. Methodology ................................................................................................................ 5
2.1
2.2
2.2.2
2.2.3
2.2.4
3. Results........................................................................................................................ 12
3.1
3.2
3.3
3.2.1
Perturbation Approximation................................................................. 14
3.2.2
Analytical Solution............................................................................... 17
3.2.3
Background .......................................................................................... 20
3.3.2
3.3.3
3.4
3.3.4
3.3.5
Background .......................................................................................... 34
3.4.2
3.4.3
3.4.4
3.4.5
3.4.6
4. Conclusion ................................................................................................................. 51
References........................................................................................................................ 53
A. Appendices ................................................................................................................ 54
A.1 Unforced Duffing Equation Numeric MAPLE Code ......................................... 55
A.2 Van Der Pol Equation Numeric MAPLE Code .................................................. 58
A.3 Numerical Value Tables for the Ordinary Differential Equation ....................... 61
A.4 Numerical Value Tables for the Duffing Equation ............................................. 62
A.5 Numerical Value Tables for the Van Der Pol Equation ..................................... 66
iii
LIST OF TABLES
Table 1: Perturbation and Exact Solutions to the Algebraic Equation .............................. 9
Table 2: Analytical Values Determined for the Ordinary Differential Equation ............ 19
Table 3: Perturbation and Exact Solutions to the Ordinary Differential Equation .......... 61
Table 4: Perturbation and Numerical Values Determined for the Unforced Duffing
Equation (=.01) .............................................................................................................. 62
Table 5: Perturbation and Numerical Values Determined for the Unforced Duffing
Equation (=.05) .............................................................................................................. 64
Table 6: Perturbation and Numerical Values Determined for the Van Der Pol Equation
(=.01) .............................................................................................................................. 66
Table 7: Perturbation and Numerical Values Determined for the Van Der Pol Equation
(=.05) .............................................................................................................................. 68
iv
LIST OF FIGURES
Figure 1: Comparative Solutions Plots for the Algebraic Equation .................................. 9
Figure 2: Perturbation Percent Error Plots for the Algebraic Equation ........................... 10
Figure 3: Comparative Solutions Plots for the Algebraic Equation as >1 .................... 11
Figure 4: Boundary Condition Visualization for Linear Ordinary Differential Equation 15
Figure 5: Comparative Solutions Plots for the Ordinary Differential Equation .............. 19
Figure 6: Regular Perturbation Percent Error Plot for the Ordinary Differential Equation
......................................................................................................................................... 20
Figure 7: Regular Perturbation versus Numeric Solution for Unforced Duffing Equation
(=.01) .............................................................................................................................. 27
Figure 8: Regular Perturbation versus Numeric Solution Percent Error Plot for Unforced
Duffing Equation (=.01)................................................................................................. 28
Figure 9: Poincare-Lindstedt versus Numeric Solution for Unforced Duffing Equation
(=.01) .............................................................................................................................. 28
Figure 10: Poincare-Lindstedt versus Numeric Solution Percent Error Plot for Unforced
Duffing Equation (=.01)................................................................................................. 29
Figure 11: Regular Perturbation versus Numeric Solution for Unforced Duffing Equation
(=.05) .............................................................................................................................. 30
Figure 12: Regular Perturbation versus Numeric Solution Percent Error Plot for
Unforced Duffing Equation (=.05)................................................................................. 31
Figure 13: Poincare-Lindstedt versus Numeric Solution for Unforced Duffing Equation
(=.05) .............................................................................................................................. 32
Figure 14: Poincare-Lindstedt versus Numeric Solution Percent Error Plot for Unforced
Duffing Equation (=.05)................................................................................................. 33
Figure 15: Regular Perturbation versus Numeric Solution for Van Der Pol Equation
(=.01) .............................................................................................................................. 43
Figure 16: Regular Perturbation versus Numeric Solution Absolute Error Plot for Van
Der Pol Equation (=.01) ................................................................................................. 44
Figure 17: Multiple Scales versus Numeric Solution for Van Der Pol Equation (=.01) 45
Figure 18: Multiple Scales versus Numeric Solution Absolute Error Plot for Van Der Pol
Equation (=.01) .............................................................................................................. 46
v
Figure 19: Perturbation versus Numeric Solution for Van Der Pol Equation (=.05) .... 47
Figure 20: Perturbation versus Numeric Solution Absolute Error Plot for Van Der Pol
Equation (=.05) .............................................................................................................. 48
Figure 21: Multiple Scales versus Numeric Solution for Van Der Pol Equation (=.05) 49
Figure 22: Figure 18: Multiple Scales versus Numeric Solution Absolute Error Plot for
Van Der Pol Equation (=.05) ......................................................................................... 50
vi
ACKNOWLEDGMENT
I would like to thank my family for their support over the course of my graduate study
especially during this final project. I would also like to thank the faculty and staff at
Rensselaer for their excellent education program. I would like to especially thank
Professor Gutierrez-Miravete for advising me throughout the duration of the project and
for making the cohort program a success. Additionally, I thank General Dynamics
Electric Boat Corporation and my work supervisor Thomas Lambert for supporting me
throughout my degree. I would like to thank one of my dearest friends and co-workers
Bernard Nasser Jr. for encouraging me to further my education by attending Rensselaer.
Finally my deepest thanks go to Jerold Lewandowski for spending countless time
mentoring me throughout my educational experience at Rensselaer.
vii
ABSTRACT
The purpose of this project is to learn and apply perturbation theory in order to
approximate solutions to engineering problems which would otherwise be intractable
through the use of traditional analytical methods. The report first outlines the technique
of perturbation theory with the aid of an algebraic equation. An introduction is provided
in the technique of non-dimensionalizing differential equations and how the term is
developed. Perturbation theory will then be applied to a linear ordinary differential
equation boundary layer problem. The boundary layer problem demonstrates the
technique required to match inner and outer solutions as well as the technique used to
develop a composite solution. Next, approximate solutions for several variations of a
non-linear mass spring dampener systems using various perturbation methods were
determined. The unforced Duffing and the Van Der Pol equations were investigated.
When regular perturbation approximations result with secular terms, a perturbation
approximation without the presence of secular terms will be developed through the use
of special perturbation methods; namely the Poincare-Lindstedt and Multiple Scales
methods. All problems investigated are also solved analytically or numerically as and
compared and contrasted to the approximations found through the use of perturbation
theory.
viii
1. Introduction
1.1 Background
Perturbation methods, also known as asymptotic, allow the simplification of
complex mathematical problems. Use of perturbation theory will allow approximate
solutions to be determined for problems which cannot be solved by traditional analytical
methods. Second order ordinary linear differential equations are solved by engineers and
scientists routinely. However in many cases, real life situations can require much more
difficult mathematical models, such as non-linear differential equations.
Numerical methods used on a computer of today are capable of solving extremely
complex mathematical problems; however, they are not perfect. The numerical methods
of today can still run into a multitude of problems ranging from diverging solutions to
tracking wrong solutions. Numerical methods on a computer do not provide much
insight to the engineers or scientists running them. Perturbation theory can offer an
alternative approach to solving certain types of problems. Solving problems analytically
often helps an engineer or scientist to understand a physical problem better, and may
help improve future procedures and designs used to solve their problems. Also, in a time
where there are tough economic circumstances, it is not unreasonable to consider that
future employers may prefer to rely on human ingenuity over the necessity of
continually purchasing expensive software package licenses to solve problems in which
analytical approximations can be made.
The first step required to start the implementation of perturbation theory nondimensionalizing of the governing equation. Once the equation is non-dimensionalized,
perturbation theory requires taking advantage of a small parameter that appears in an
equation. This parameter, usually denoted is on the order of 0 < << 1.
Next, through educated assumptions on the order of magnitude of terms, a rough
approximate solution is determined through the use of logical elimination of low
impacting terms. The perturbation method then solves this reduced outer problem.
Next an inner solution is constructed to satisfy the other constraints of the problem. A
composite solution is obtained through a matching process.
The Navier Stokes equations for a Newtonian fluid with constant density and viscosity in
Cartesian coordinates is as follows:
[1-1-2]
[1-1-3]
[1-1-4]
Assuming a steady, constant density and viscosity, and two dimensional flow, the
continuity and Navier stokes equations reduce to the following:
[1-1-5]
[ 1-1-6]
[1-1-7]
These equations are often used to model flow in boundary layer regions. Often
times, these equations are further simplified by engineers and scientist depending on the
physics of the problem being solved. This simplification can be performed by an order of
magnitude analysis. For example, the velocity in the vertical plane may be extremely
small compared to the velocity in the horizontal direction, therefore terms that carry the
vertical velocity term will be reduced to zero. While the vertical velocity may not be
exactly zero, this assumption will introduce some error into an eventual approximation.
The problem can be further simplified in this manor until an analytical solution is
obtainable. The mechanics of perturbation theory follows this same methodology
allowing analytical approximations to be found for equations which would otherwise be
impossible to solve without the use of a computer.
2. Methodology
2.1 Project Methodology
A polynomial algebraic equation will be solved using the traditional quadratic formula.
Next, solutions for the same equation will be approximated following the techniques of
perturbation theory. This will be done to develop the understanding of the methodology
required.
An analytical solution can be found for the ordinary linear differential equation by
using traditional methods for solving ordinary differential equations; however numerical
solutions will be required for the Duffing and Van Der Pol equations since they are nonlinear differential equations.
Microsoft Excel will be used to graph and compare analytical/numerical solutions to
the approximate solutions obtained through the use of perturbation theory. Maplesofts
MAPLE will be used to find numerical solutions as needed.
Sometimes during the development of a perturbation approximation, secular terms may
appear causing the perturbation approximation to diverge from the actual solution as time
increases. Secular terms are terms that grow as the approximation progresses without bound.
For these problems the Poincare-Lindstedt method will be used to develop perturbation
approximations without influence of secular terms. If the Poincare-Lindstedt method is
unable to eliminate all of the secular terms, the Multiple Scales method will be utilized.
2.2.1
It is important to note that the correction factor that is applied should always be smaller
than the leading term. Upon substitution of the leading order solution plus a correction,
(x), into the governing differential equation (x) is determined to be of the order
as
goes to zero.
Since this is a second degree polynomial equation, it is known that there are two
roots. Both roots are determined through perturbation the same way by substitution of
[2-2-2] into [1-2-1]. This calculation will further develop the positive root. Substitution
of the positive root seen in [2-2-2] into [1-2-1] yields:
6
[2-2-3]
Since both
and
and
extremely small terms are known as higher order terms (HOTs). In perturbation
nomenclature these HOTs are often abbreviated as since they carry little
significance to the solution resulting in often elimination. Solving the remainder for [22-3] for
yields:
[2-2-4]
Substitution of
and
These HOTs are eliminated from [2-2-7]. [2-2-7] is then used to solve for
yielding:
[2-2-8]
Substitution of
back into [2-2-5] for the yields the 3rd positive root
approximation:
[2-2-9]
It is important to note that each correction term is smaller than that of the preceding
term. Larger correction terms can be an indication that either an algebraic error has
occurred, or that a mistake could have occurred during the elimination of the HOTs.
2.2.2
used to determine the exact roots. The exact roots to [1-2-1] are:
[2-2-10]
2.2.3
The 1st term, 2nd term, and 3rd term perturbation approximations obtained in section 2.2.1
were compared to the exact solution determined in 2.2.2. Percent error was calculated
for each perturbation approximation. Percent Error was determined by the following
formula:
[2-2-11]
The actual value was taken to be the root solved by use of the quadratic formula.
Table 1 below was developed by utilizing equations [2-2-1], [2-2-4], [2-2-9], [2-2-10]
and [2-2-11].
0
0.000001
0.00001
0.0001
0.001
0.01
0.1
1
10
Exact
Positive
Root
1
1
0.999995
0.99995
0.9995
0.995012
0.951249
0.618034
0.09902
Perturbation 1st
Perturbation 2nd
Term
Term
Positive
%Error
Positive
%Error
Root
R1
Root
R1
1
0
1
0
1
5E-05
1 -1.3E-11
1
0.0005 0.999995 -1.3E-09
1
0.005 0.99995 -1.3E-07
1 0.050012
0.9995 -1.3E-05
1 0.50125
0.995 -0.00126
1 5.124922
0.95 -0.13132
1 61.8034
0.5 -19.0983
1 909.902
-4 -4139.61
Perturbation 3rd
Term
Positive
%Error
Root
R1
1
0
1
0
0.999995
0
0.99995 1.11E-14
0.9995 7.89E-13
0.995013 7.85E-09
0.95125
8.2E-05
0.625 1.127124
8.5 8484.167
1 can be
seen in Figure: 1.
Exact
Calculated Root
0.99
0.98
Perturbation
1st Term
0.97
Perturbation
2nd Term
0.96
0.95
Perturbation
3rd Term
0.94
0
0.02
0.04
0.06
0.08
0.1
Note that for any given value of the accuracy of the perturbation approximation
increases with the amount of corrections that were determined. The exact, 2nd term and
3rd term approximations are nearly indistinguishable at this magnification. Perturbations
approximations, unlike a typical series expansion, do not necessarily always become
9
more precise as additional terms are added to the approximation. Perturbation solutions
are developed in powers of (in the limit as goes to zero), whereas series solutions are
developed in powers of . This distinction leads to differences in solution convergence.
Engineers and scientists should be wary that distinct limitations exist with the accuracy
that can be achieved with perturbation approximations.
Plots of the percent error of the 2nd and 3rd term perturbation approximation for 0
0.01
-0.01 0
0.02
0.04
0.06
0.08
0.1
% Error
-0.03
Perturbation
2nd Term
-0.05
-0.07
-0.09
Perturbation
3rd Term
-0.11
-0.13
-0.15
2.2.4
10
9
Exact
7
5
Calculated Root
Perturbation
1st Term
3
1
Perturbation
2nd Term
-1 0
10
Perturbation
3rd Term
-3
-5
11
3. Results
3.1 Brief Introduction to Non-dimensionalizing Differential
Equations
Non-dimensionalizing the equation is the first step required in perturbation
methods. To introduce how this is to be accomplished, the typical linear ordinary
differential equation from a mass spring dash-pot dampener system is introduced below.
[3-1-1]
Here
dampener, and
And
[3-1-3]
Therefore:
[3-1-4]
Where
and
and
are dimensionalized
variables. It follows that utilizing the chain rule the first derivative of a function with
respect to t is:
[3-1-5]
12
[3-1-8]
Since the goal is to remove the dimensions for all the coefficients, let:
[3-1-9]
introducing the linear ordinary differential equation [1-2-2] as seen below with as
the dependent variable and as the independent variable:
[1-2-2]
And
[3-2-2]
3.2.1
Perturbation Approximation
Since
With this solution, only one of the boundary conditions from the initial problem can
be enforced. Using equation [3-2-1] and solving for
=0, which is firstly a trivial solution, but also would violate the initial problem as
shown in Figure: (4).
14
Y(X)
1
0.9
0.8
0.7
0.6
0.5
0.4
0.3
0.2
0.1
0
Initial
Conditions
0.2
0.4
0.6
0.8
is a positive value,
is also a
positive value (since the function is concave up). If this was true, then:
[3-2-8]
Equation [3-2-8] violates the initial problem in equation [1-2-2] and therefore
equation [3-2-1] is not the proper boundary condition for equation [3-2-7].
Using the boundary condition in equation [3-2-2] to solve for
in equation [3-2-7]
results in:
[3-2-9]
is required. Letting:
[3-2-11]
15
And:
[3-2-13]
Assuming that the first two terms balance and solving for
follows:
[3-2-15]
yields:
[3-2-16]
Since
=-2 and 0, the general solution of the equation takes the form:
[3-2-20]
Using the remaining boundary condition in equation [3-2-1] and solving for
yields:
[3-2-21]
Substitution of equation [3-2-21] into equation [3-2-20] yields the following inner
solution:
[3-2-22]
16
Since two separate solutions, equations [3-2-10] and [3-2-22], have been
obtained; matching is required to be performed in order to develop a composite solution
(
is
:
[3-2-23]
Equation [3-2-24] is not only used to determine the value of , but it also determines the
common solution of the limits of the inner and outer solution (
).
[3-2-25]
The composite solution is determined by combining the inner and outer solutions and by
shifting the solutions by removing the common solution:
[3-2-26]
Combining equations [3-2-10], [3-2-11], [3-2-22], [3-2-24], [3-2-25] and [3-2-26] and
simplifying yields the composite solution:
[3-2-27]
3.2.2
Analytical Solution
Since this is a second order linear ordinary differential equation, traditional
And substituting equation [3-2-28] into equation [1-2-2] and simplifying yields:
[3-2-29]
and
And
17
[3-2-31]
Since both roots are real numbers, the general solution takes the form:
[3-2-32]
Substitution of equations [3-2-30] and [3-2-31] into equation [3-2-32] yields the
following:
[3-2-33]
Enforcement of the initial condition seen in equation [3-2-1] to equation [3-2-33] yields:
[3-2-34]
Substituting equation [3-2-34] into [3-2-33] and enforcing the initial condition seen in
equation [3-2-2] into equation [3-2-33] yields:
[3-2-35]
yields:
[3-2-36]
[3-2-37]
Substitution of equations [3-2-36] and [3-2-37] into equation [3-2-33] yields the final
analytical solution:
[3-2-38]
18
3.2.3
0.01
Analytical Roots
Root 1
-1.00505
Analytical Constants
Root 2
-198.99494
C1
2.73204
C2
-2.73205
Data from Table 2 was used in conjunction with equations [2-2-11], [3-2-27],
and [3-2-38] to create the comparative data plots seen in Table 3. The formation of the
boundary layer become apparent upon the inspection of the roots in Table 2. Since Root
2 is large in magnitude compared to Root one, the influence of the solution dependent on
Root 2 on the total solution is quickly reduced as
layer can be seen for
Plots of the composite perturbation approximation and the analytical solutions for
the linear ordinary differential equation can be seen below in Figure: 5.
3
2.5
Analytical
Roots
2
1.5
Perturbation
1
0.5
0
0
0.2
0.4
0.6
0.8
1.2
19
0.2
0.1
% Error
0
-0.1 0
0.2
0.4
0.6
0.8
1.2
-0.2
-0.3
%Error
-0.4
-0.5
-0.6
Figure 6: Regular Perturbation Percent Error Plot for the Ordinary Differential
Equation
Background
The Duffing Oscillator is a differential equation that used to model non-linear
restoration force type problems. The Duffing Oscillator can be used to approximate the
physics of a pendulum problem [2].
Re-introducing the Duffing equation [1-2-3] as seen below with as the dependent
variable and as the independent variable:
[1-2-3]
And
20
[3-3-2]
3.3.2
Solving for:
[3-3-6]
is then:
[3-3-8]
And
[3-3-10]
21
Assuming (x) is some correction factor, the second solution approximation is as seen
below:
[3-3-12]
is then:
[3-3-13]
is then:
[3-3-14]
in equation [3-3-12],
in equation [3-3-13]
results in:
[3-3-15]
And
[3-3-16]
Substituting equation [3-3-14] and [3-3-12] into equation [1-2-3] and simplifying yields:
[3-3-17]
Expanding
yields:
=
[3-3-18]
Eliminating the HOTs from equation [3-3-18], the remaining terms are substituted back
into equation [3-3-17], which is re-written as:
[3-3-19]
Letting:
[3-3-20]
[3-3-21]
[3-3-22]
[3-3-23]
[3-3-24]
[3-3-25]
Solving for
Noting that general solution takes the same form as equation [3-2-7], yields:
[3-3-29]
are then:
[3-3-31]
Substitution of equations [3-3-30] and [3-3-31] into equation [3-3-27] and solving for
coefficients , , and
yeild:
23
[3-3-32]
[3-3-33]
[3-3-34]
Combining equations [3-3-28], [3-3-29], [3-3-30], [3-3-32], [3-3-33], and [3-3-34] and
simplifying with equation [3-3-20] yields:
[3-3-35]
yields:
[3-3-36]
Using the boundary conditions for equations [3-3-15] and [3-3-16] in equations [3-3-35]
and [3-3-36] and solving for
and
yields:
[3-3-37]
And
[3-3-38]
Combining equations [3-3-12], [3-3-35], [3-3-37], and [3-3-38] yields the 1st order
perturbation approximation is:
[3-3-39]
3.3.3
Poincare-Lindstedt Method
Upon a more detailed inspection of the 1st order perturbation approximation
24
progresses the
term (secular
is the correction
Using the chain rule, the first and second derivatives of [3-3-40] with respect to
are:
[3-3-41]
And
[3-3-42]
Allowing
[3-3-1] yields:
[3-3-43]
and
Following the same mathematical analysis as in section 3.3.2, the first PoincareLindstedt leading order solution is determined to be:
[3-3-45]
Assuming
below:
[3-3-46]
25
[3-3-47]
Substitution of equations [3-3-42], [3-3-46] and [3-3-47] into equation [1-2-3] yields:
[3-3-48]
Expanding and eliminating the HOTs in equation [3-3-58] in the same manner as
performed in section 3.3.2 and simplification yields:
[3-3-49]
Expansion of
Solving for
3.3.4
Numerical Solution
The numerical solution was obtained utilizing MAPLEs built in Fehlberg fourth-
fifth order Runge-Kutta method with degree four interpolant. The MAPLE file used to
perform the numerical analysis can be seen attached in Appendix A.1.
3.3.5
Case 1: = .01
26
Letting = .01 and =1, equations [3-3-39], [3-3-52] and [3-3-53] as well as the
numerical solution developed in section 3.2.4 was used in order to produce Table 4 in
Appendix A.4.
1.3
0.8
0.3
Y Numerical
-0.2 0
50
100
150
200
Y Perturbation
-0.7
-1.2
X
approximation tends to rapidly increase in error with respect to the the numerical
solution. The absolute error plot of the regular perturbation versus the numerical solution
is seen in Figure: 8.
27
0.25
Error
0.15
Absolute Error
(Perturbation
Numerical)
0.05
-0.05 0
50
100
150
200
-0.15
-0.25
Figure 8: Regular Perturbation versus Numeric Solution Percent Error Plot for
Unforced Duffing Equation (=.01)
1.3
0.8
Y Numerical
0.3
-0.2 0
50
100
150
200
Y Lindstedt
-0.7
-1.2
X
28
approximation track the numerical solution far better than the regular perturbation
approximation. The absolute error plot of the Poincare-Lindstedt perturbation versus the
numeric solution is seen in Figure: 10.
0.002
0.0015
0.001
Error
0.0005
0
-0.0005 0
50
100
150
-0.001
200
Absolute
Error
(Lindstedt
to
Numerical)
-0.0015
-0.002
Figure 10: Poincare-Lindstedt versus Numeric Solution Percent Error Plot for
Unforced Duffing Equation (=.01)
as the with the regular perturbation approximation however the magnitude of the
is continued to
Case 2: = .05
Letting = .05 and =1, equations [3-3-39], [3-3-52] and [3-3-53] as well as the
numerical solution developed in section 3.2.4 was used in order to produce Table 5 in
Appendix A.4.
29
Y Numerical
0
-1
50
100
150
200
-2
Y
Perturbation
-3
-4
-5
X
Figure 11: Regular Perturbation versus Numeric Solution for Unforced Duffing
Equation (=.05)
Since has increased in size, the secular term found in the regular perturbation
approximation now dominates the solution faster than seen in Case 1. The absolute error
plot of the regular perturbation versus the numeric solution is seen in Figure: 12.
30
5
4
3
2
Error
1
0
-1 0
50
100
150
200
-2
-3
Absolute
Error
(Perturbation
Numerical)
-4
-5
Figure 12: Regular Perturbation versus Numeric Solution Percent Error Plot for
Unforced Duffing Equation (=.05)
in both
31
1.5
1
Y
Numerical
0.5
0
0
50
100
150
-0.5
200
Y
Lindstedt
-1
-1.5
X
It is important to note that even with the increased value, the PoincareLindstedt perturbation approximation track the numerical solution far better than the
regular perturbation approximation. The absolute error plot of the Poincare-Lindstedt
perturbation versus the numeric solution is seen in Figure: 14.
32
0.05
0.04
0.03
Error
0.02
0.01
0
-0.01
50
100
150
200
Absolute
Error
(Lindstedt
to
Numerical)
-0.02
-0.03
-0.04
Figure 14: Poincare-Lindstedt versus Numeric Solution Percent Error Plot for
Unforced Duffing Equation (=.05)
33
Background
The Van Der Pol oscillator is a model of a non-conservative energy system. The
Van Der Pol equation can be used to model stick-oscillations, aero-elastic flutter and
biological oscillatory phenomena [Reference 2].
Re-introducing the Van Der Pol equation [1-2-4] as seen below with as the
dependent variable and as the independent variable:
[1-2-4]
And
[3-4-2]
3.4.2
This is the same equation as equation [3-3-3] in the unforced Duffing equation
section, and the boundary conditions in equations [3-4-1] and [3-4-2] are the same as [33-1] and [3-3-2]. Therefore the development of the leading order solution for [3-4-3] is
identical to that of [3-3-3]. Refer to section 3-3 for more information.
The leading order solution is determined to be:
[3-4-4]
34
[3-4-5]
are the
same as equations [3-3-13] and [3-3-14] in the Duffing equation section. The values of
and
Expanding equation [3-4-6] and eliminating the HOTs terms results in the following the
remaining terms rewritten as
[3-4-7]
Letting:
[3-4-8]
Performing the same type of trigonometric expansion as performed with equation [3-319] yields:
[3-4-10]
Solving for
Noting that general solution takes the same form as equation [3-4-3], yields:
[3-4-12]
35
are then:
[3-4-14]
Substitution of equations [3-4-13] and [3-4-14] into equation [3-4-10] and solving for
coefficients , , and
yield:
[3-4-15]
[3-4-16]
[3-4-17]
Combining equations [3-4-11], [3-4-12], [3-4-13], [3-4-15], [3-4-16], and [3-4-17] and
simplifying with equation [3-4-8] yields:
[3-4-18]
yields:
[3-4-19]
Using the boundary conditions for equations [3-3-15] and [3-3-16] in equations [3-4-18]
and [3-4-19] and solving for
and
yields:
[3-4-20]
And
[3-4-21]
Combining equations [3-4-5], [3-4-18], [3-4-20], and [3-4-21] yields the 1st order
perturbation approximation is:
[3-4-22]
36
3.4.3
Poincare-Lindstedt Method
As seen with the Duffing equation, the development of the regular perturbation
approximation for the Van Der Pol equation results the secular term, (
),
is identical as seen in
equation [3-3-47]:
[3-3-47]
Substitution of equations [3-3-42], [3-3-46], [3-3-47], and [3-4-23] into equation [1-2-4]
yields:
[3-4-24]
Expanding and eliminating the HOTs in equation [3-4-24] in the same manner as
performed in section 3.3.2 and simplification yields:
[3-4-25]
Expansion of
37
[3-4-26]
Unlike the Duffing equation from section 3-3-3, there is no value for
in which
prevention of the formation of the secular terms can be obtained. The Poincare-Lindstedt
method approximation is therefore unable to alleviate the unwanted effects of a secular
term.
3.4.4
Lindstedt method has failed to yield a perturbation approximation for the Van Der Pol
equation, the next attempt to eliminate the unwanted effects of secular terms by utilizing
the Multiple Scales method. The Multiple Scales method introduces a new variable ,
that forms the following relation to and :
[3-4-27]
Therefore, when
becomes
normal sized.
Leading Order Solution:
The first derivative of function
with respect to
is:
[3-4-28]
with respect to
27] is:
[3-4-30]
38
Simplification of equation [3-4-31] through the elimination of 2nd and higher order terms
in :
[3-4-32]
Setting
would results in the leading order problem reminisnt of the solution seen
in the ordinary differential equation in section 3-4-2, however the coefficients are now
unknown functions of
Using the boundary condition seen in equation [3-4-1] in conjunction with equation [34-34] and the boundary condition seen in equation [3-4-2] with conjunction with
equation [3-4-35] yields:
[3-4-36]
And
[3-4-37]
39
is:
[3-4-39]
once is:
[3-4-41]
Substitution of equations [3-4-38], [3-4-39], and [3-4-40] into equation [3-4-32] yields:
[3-4-42]
( , )2*
[3-4-43]
and
yields:
[3-4-44]
And
40
[3-4-45]
Noting that both equations [3-4-37] and [3-4-45] are equal to zero.
is determined
to be a constant 0.
Simplifying equation [3-4-44] yields:
[3-4-46]
Using practical fraction decomposition the left side of equation [3-4-47] and setting it
equal to the right side, then integrating once results in:
[3-4-48]
Where
[3-4-50]
And
[3-4-51]
yields:
[3-4-53]
Substituting equations and [3-4-27] and [3-4-53] into equation [3-4-35] and the
observation that
can be
[3-4-55]
3.4.5
Numerical Solution
As before, the numerical solution was obtained utilizing MAPLEs built in
Fehlberg fourth-fifth order Runge-Kutta method with degree four interpolant. The
MAPLE file used to perform the numerical analysis can be seen attached in Appendix
A.2.
3.4.6
Case 1: = .01
Letting = .01 equations [3-4-22], [3-4-56] and [3-3-53] as well as the numerical
solution developed in section 3.4.5 was used in order to produce Table 6 in Appendix
A.5
The regular perturbation approximation seen in Table 6 is plotted together with the
numerical solution that was obtained with MAPLE in Figure: 15 below.
42
2
1.5
Y
Numerical
0.5
0
0
50
100
-0.5
150
200
Y
Perturbati
on
-1
-1.5
-2
X
Figure 15: Regular Perturbation versus Numeric Solution for Van Der Pol
Equation (=.01)
The absolute error plot of the regular perturbation versus the numerical solution
is seen in Figure: 16.
43
0.05
0.04
0.03
0.02
Error
0.01
0
-0.01
50
100
-0.02
150
200
Absolute
Error
(Perturbation
Numerical)
-0.03
-0.04
-0.05
Figure 16: Regular Perturbation versus Numeric Solution Absolute Error Plot for
Van Der Pol Equation (=.01)
Once again as seen in the Duffing equation section, the regular perturbation
approximations absolute error increases as
44
2
1.5
1
Y
Numerical
0.5
0
0
50
100
-0.5
150
200
Y Multiple
Scales
-1
-1.5
-2
X
Figure 17: Multiple Scales versus Numeric Solution for Van Der Pol Equation
(=.01)
approximation tracks the numerical solution far better than the regular perturbation
approximation. The absolute error plot of the Multiple Scales perturbation versus the
numeric solution is seen in Figure: 18.
45
0.005
0.004
0.003
0.002
Error
0.001
0
-0.001
50
100
150
-0.002
Absolute Error
(Multiple
200
Scales to
Numerical)
-0.003
-0.004
-0.005
Figure 18: Multiple Scales versus Numeric Solution Absolute Error Plot for Van
Der Pol Equation (=.01)
The Multiple Scale method is able to provide a solution approximation that has
error two orders of magnitude small than the regular perturbation method.
Case 2: = .05
Letting = .05 equations [3-4-22], [3-3-56] and [3-3-53] as well as the numerical
solution developed in section 3.4.5 was used in order to produce Table 7 in Appendix
A.5.
46
5
4
3
Y Numerical
1
0
-1
50
100
150
200
Y
Perturbation
-2
-3
-4
-5
Figure 19: Perturbation versus Numeric Solution for Van Der Pol Equation (=.05)
Since has increased in size, the secular term found in the regular perturbation
approximation now dominates the solution faster than seen in Case 1. The absolute error
plot of the regular perturbation versus the numerical solution is seen in Figure: 20.
47
2.5
2
1.5
1
Error
0.5
0
-0.5
50
100
150
-1
200
Absolute
Error
(Perturbation
Numerical)
-1.5
-2
-2.5
Figure 20: Perturbation versus Numeric Solution Absolute Error Plot for Van Der
Pol Equation (=.05)
in both Cases
48
2.5
2
1.5
Y
Numerical
0.5
0
-0.5
50
100
150
200
Y Multiple
Scales
-1
-1.5
-2
-2.5
X
Figure 21: Multiple Scales versus Numeric Solution for Van Der Pol Equation
(=.05)
It is important to note that especially with the increased value, the Multiple
Scale perturbation approximation tracks the numerical solution far better than the regular
perturbation approximation. The absolute error plot of the Poincare-Lindstedt
perturbation versus the numeric solution is seen in Figure: 22.
49
0.03
0.02
Error
0.01
0
0
50
100
150
-0.01
200
Absolute
Error
(Multiple
Scales to
Numerical)
-0.02
-0.03
Figure 22: Figure 23: Multiple Scales versus Numeric Solution Absolute Error Plot
for Van Der Pol Equation (=.05)
The Multiple Scales method is able to provide a solution approximation that has
error two orders of magnitude small than the regular perturbation method.
50
4. Conclusion
The intent of the work reported in this paper was to demonstrate and convey the
idea of using perturbation methods to solve some selected engineering and mathematical
problems.
This paper first explained the theory of finding approximate solutions through the
use of perturbation methods through a simple algebraic example. Error of first, second,
and third order perturbation corrections were compared. The sensitivity of perturbation
approximations accuracy as increases was compared to the exact solution determined
through the use of the quadratic equation.
Next, a brief introduction into the process of non-dimensionalizing an ordinary
linear differential equation was discussed. The differential equation selected can be used
to model the physics of a typical mass spring dampener problem. This nondimensionalization allowed for the formation of , and was shown that nondimensionalization of the problem allowed the development of a single equation to
represent multiple physical parameter variations.
A similar linear second order ordinary differential equation was solved using
perturbation methods. Due to the location of in the differential equation, the equation
resulted in a specific subset known as a boundary layer problem. In order to enforce
both boundary conditions, the perturbation approximation developed an inner and outer
solution. Then, through the use of matching, a single composite solution was
determined. The perturbation approximation was compared to the exact analytical
solution obtained through normal application of differential equation theory.
A regular perturbation approximation was then developed for the unforced
Duffing equation. The regular perturbation approximation resulted in a secular term
being present. In order to develop a approximation without a secular term, the PoincareLindstedt method was used to shift the frequency of the perturbation approximation.
Both of these approximations were compared to a numerical solution which was
obtained through the use of MAPLE for two different values of . While both the regular
perturbation approximation and the Poincare-Lindstedt methods tracked the numerical
solution with low error at low values of
significantly lower error as values of
increased.
51
Finally, a regular perturbation approximation was then developed for the Van
Der Pol equation. The regular perturbation approximation resulted in a secular term
being present. In order to develop a approximation without a secular term, the PoincareLindstedt method was attempted. The Poincare-Lindstedt was unable to eliminate all the
terms that would result in secular term being present in a perturbation approximation.
The Multiple Scales method was then used to introduce a new variable which is
dependent on and . This new variable allowed the successful elimination of secular
terms from appearing in a perturbation approximation. Both the regular perturbation
approximation and the Multiple Scales method approximations were compared to a
numerical solution which was obtained through the use of MAPLE for two different
values of . While both the regular perturbation approximation and the Multiple Scales
methods tracked the numerical solution with low error at low values of , the Multiple
Scales method had significantly lower error as values of
52
increased.
References
[1] Introduction to Perturbation Methods. M.H Holmes; Springer; 1995
[2] Lecture Notes on Nonlinear Vibrations; Richard Rand; 2005
[3] Introduction to Singular Perturbation Methods Nonlinear Oscillations; A; Aceves,
N.Ercolani, C.Jones, J. Lega & J. Moloney; 1994
Additional Reading:
[4] Transport Phenomenan; Second Edition; Bird, Stewart and Lightfoot; John
Wiley& Sons; 2007
[5] Perturbation Methods; Ali Nayfeh; John Wiley& Sons; 1973
[6] Perturbation Theory & Stability Analysis University of Twente; T. Weinhart, A
Singh, A.R. Thornton; May 17, 2010
[7] Some Asymptotic Methods for Strongly Nonlinear Equations; Ji-Huan He; 2006
53
A. Appendices
54
55
56
57
58
59
60
Y Analytical
0
2.424557348
2.641622008
2.635230466
2.546917667
2.4591161
2.374339022
2.292484598
2.213452074
2.137144166
2.063466944
1.992329715
1.923644915
1.857327997
1.79329733
1.731474094
1.671782192
1.614148144
1.558501008
1.504772286
1.452895841
1.402807816
1.354446556
1.307752533
1.262668268
1.219138265
1.177108943
1.136528565
1.09734718
1.059516559
1.022990133
0.987722942
0.953671574
0.920794114
0.889050091
0.858400431
0.828807407
Y Composite
0
2.415660385
2.629258016
2.622507364
2.534980398
2.448021737
2.364043877
2.282946823
2.204631754
2.129003234
2.055969103
1.985440363
1.917331067
1.851558218
1.788041666
1.726704009
1.667470503
1.610268966
1.555029691
1.501685365
1.450170984
1.400423771
1.352383105
1.305990444
1.261189255
1.217924943
1.176144786
1.135797871
1.096835032
1.059208789
1.022873291
0.987784259
0.953898935
0.921176026
0.889575656
0.859059317
0.829589821
61
%Error
-0.36695
-0.46805
-0.48281
-0.46869
-0.45115
-0.4336
-0.41605
-0.39849
-0.38093
-0.36336
-0.34579
-0.32822
-0.31065
-0.29307
-0.27549
-0.25791
-0.24032
-0.22273
-0.20514
-0.18755
-0.16995
-0.15235
-0.13474
-0.11713
-0.09952
-0.08191
-0.06429
-0.04667
-0.02905
-0.01142
0.006208
0.023841
0.041476
0.059115
0.076757
0.094402
Y Perturbation
Y Lindstedt
Y Numerical
Absolute Error
(Lindstedt vs
Numerical)
0
3.926991
7.853982
11.78097
15.70796
19.63495
23.56194
27.48894
31.41593
35.34292
39.26991
43.1969
47.12389
51.05088
54.97787
58.90486
62.83185
66.75884
70.68583
74.61283
78.53982
82.46681
86.3938
90.32079
94.24778
98.17477
102.1018
106.0288
109.9557
113.8827
117.8097
121.7367
125.6637
1
-0.696251833
-0.029452431
0.73790386
-1
0.654599805
0.088357293
-0.779555888
1
-0.612947777
-0.147262156
0.821207915
-1
0.57129575
0.206167018
-0.862859943
1
-0.529643722
-0.26507188
0.90451197
-1
0.487991695
0.323976742
-0.946163998
1
-0.446339667
-0.382881605
0.987816025
-1
0.40468764
0.441786467
-1.029468053
1
1
-0.69661748
-0.029448173
0.737645704
-0.99826561
0.653172843
0.088242371
-0.776115199
0.993068457
-0.607462493
-0.146730474
0.81189252
-0.984426568
0.55964499
0.204709603
-0.844853565
0.97236992
-0.509886202
-0.261978638
0.874884
-0.956940336
0.458358731
0.318338928
-0.901879654
0.938191336
-0.405241314
-0.37359497
0.925746887
-0.916187957
0.350718205
0.427555093
-0.946402908
0.891006524
1
-0.696195777
-0.029347606
0.737164133
-0.998270494
0.652830896
0.087942462
-0.775592601
0.993084824
-0.607218292
-0.146236211
0.811346481
-0.984461257
0.559516144
0.204028918
-0.84430102
0.972430178
-0.509889556
-0.261122264
0.874340809
-0.957033969
0.458510253
0.317320245
-0.901360556
0.938326825
-0.405555365
-0.372429655
0.925265415
-0.916374533
0.351207602
0.42626083
-0.945971377
0.891254275
0
0.000421703
0.000100567
-0.000481572
-4.88334E-06
-0.000341947
-0.000299909
0.000522598
1.63671E-05
0.000244201
0.000494263
-0.000546039
-3.46884E-05
-0.000128846
-0.000680684
0.000552546
6.02572E-05
-3.35462E-06
0.000856375
-0.00054319
-9.36335E-05
0.000151522
-0.001018683
0.000519098
0.000135489
-0.000314051
0.001165315
-0.000481472
-0.000186576
0.000489397
-0.001294264
0.000431531
0.000247751
62
Absolute Error
(Perturbation
vs Numerical)
0
5.60552E-05
0.000104825
-0.000739727
0.001729506
-0.001768909
-0.000414832
0.003963287
-0.006915176
0.005729485
0.001025944
-0.009861434
0.015538743
-0.011779606
-0.002138099
0.018558923
-0.027569822
0.019754166
0.003949616
-0.030171161
0.042966031
-0.029481442
-0.006656497
0.044803441
-0.061673175
0.040784302
0.01045195
-0.06255061
0.083625467
-0.053480038
-0.015525637
0.083496676
-0.108745725
129.5907
133.5177
137.4447
141.3717
145.2987
149.2257
153.1526
157.0796
161.0066
164.9336
168.8606
172.7876
176.7146
180.6416
184.5686
188.4956
192.4226
196.3495
-0.363035612
-0.500691329
1.07112008
-1
0.321383585
0.559596191
-1.112772108
1
-0.279731557
-0.618501054
1.154424136
-1
0.23807953
0.677405916
-1.196076163
1
-0.196427502
-0.736310778
-0.294978531
-0.480032122
0.963776066
-0.862734386
0.238215642
0.530844026
-0.977806097
0.831469612
-0.180626435
-0.579814548
0.988444334
-0.797320654
0.122410675
0.626773822
-0.995653875
0.760405966
-0.0637703
-0.671558955
-0.29565392
-0.478628474
0.963405484
-0.863054266
0.239085462
0.529351867
-0.977506394
0.831873314
-0.181696578
-0.57825575
0.988224316
-0.797820567
0.123684165
0.625170868
-0.995521381
0.761015097
-0.065247244
-0.669934738
63
-0.000675389
0.001403648
-0.000370582
-0.00031988
0.000869821
-0.001492159
0.000299703
0.000403702
-0.001070143
0.001558798
-0.000220018
-0.000499914
0.00127349
-0.001602954
0.000132495
0.000609131
-0.001476945
0.001624217
0.067381692
0.022062855
-0.107714597
0.136945734
-0.082298122
-0.030244325
0.135265714
-0.168126686
0.098034979
0.040245303
-0.16619982
0.202179433
-0.114395364
-0.052235048
0.200554782
-0.238984903
0.131180258
0.06637604
Table 5: Perturbation and Numerical Values Determined for the Unforced Duffing
Equation (=.05)
Y Perturbation
Y Lindstedt
0
3.926991
7.853982
11.78097
15.70796
19.63495
23.56194
27.48894
31.41593
35.34292
39.26991
43.1969
47.12389
51.05088
54.97787
58.90486
62.83185
66.75884
70.68583
74.61283
78.53982
82.46681
86.3938
90.32079
94.24778
98.17477
102.1018
106.0288
109.9557
113.8827
117.8097
121.7367
125.6637
129.5907
1
-0.652832038
-0.147262156
0.861092176
-1
0.4445719
0.441786467
-1.069352314
1
-0.236311763
-0.736310778
1.277612451
-1
0.028051625
1.030835089
-1.485872589
1
0.180208513
-1.325359401
1.694132727
-1
-0.388468651
1.619883712
-1.902392864
1
0.596728788
-1.914408023
2.110653002
-1
-0.804988926
2.208932335
-2.31891314
1
1.013249064
1
-0.653172843
-0.146730474
0.844853565
-0.956940336
0.405241314
0.427555093
-0.963776066
0.831469612
-0.122410675
-0.671558955
0.999698819
-0.634393284
-0.170961889
0.85772861
-0.949528181
0.382683432
0.44961133
-0.970031253
0.817584813
-0.09801714
-0.689540545
0.998795456
-0.615231591
-0.195090322
0.870086991
-0.941544065
0.359895037
0.471396737
-0.97570213
0.803207531
-0.073564564
-0.707106781
0.997290457
Y Numerical
1
-0.651510889
-0.144319338
0.842120293
-0.957372181
0.406783672
0.421193484
-0.961865209
0.833423339
-0.128666376
-0.663478817
0.999876475
-0.639385186
-0.159940417
0.850611969
-0.952639425
0.392271396
0.435506458
-0.966117774
0.824514735
-0.112975562
-0.675287168
0.999497083
-0.627095178
-0.175529348
0.858892521
-0.947661839
0.377656432
0.449718518
-0.970125687
0.815394368
-0.097250226
-0.686932352
0.998861592
64
Absolute Error
(Lindstedt vs
Numerical)
Absolute Error
(Perturbation vs
Numerical)
0
0.001661954
0.002411136
-0.002733273
-0.000431845
0.001542358
-0.00636161
0.001910857
0.001953726
-0.006255701
0.008080138
0.000177657
-0.004991902
0.011021472
-0.007116641
-0.003111244
0.009587963
-0.014104872
0.003913479
0.006929922
-0.014958422
0.014253376
0.000701627
-0.011863587
0.019560974
-0.01119447
-0.006117774
0.017761395
-0.021678219
0.005576443
0.012186837
-0.023685663
0.020174429
0.001571135
0
0.001321149
0.002942818
-0.018971883
0.042627819
-0.037788228
-0.020592983
0.107487105
-0.166576661
0.107645387
0.072831961
-0.277735976
0.360614814
-0.187992042
-0.180223121
0.533233164
-0.607728604
0.255297945
0.359241626
-0.869617992
0.887024438
-0.286818518
-0.620386629
1.275297686
-1.175529348
0.262163732
0.966746184
-1.73299657
1.449718518
-0.165136761
-1.393537966
2.221662914
-1.686932352
-0.014387472
133.5177
137.4447
141.3717
145.2987
149.2257
153.1526
157.0796
161.0066
164.9336
168.8606
172.7876
176.7146
180.6416
184.5686
188.4956
192.4226
196.3495
-2.503456646
2.527173278
-1
-1.221509201
2.797980957
-2.735433415
1
1.429769339
-3.092505268
2.943693553
-1
-1.638029477
3.38702958
-3.151953691
1
1.846289615
-3.681553891
-0.595699304
-0.21910124
0.881921264
-0.932992799
0.336889853
0.492898192
-0.98078528
0.788346428
-0.049067674
-0.724247083
0.995184727
-0.575808191
-0.24298018
0.893224301
-0.923879533
0.31368174
0.514102744
-0.61464376
-0.191082216
0.866959421
-0.942440494
0.362942244
0.463825972
-0.973887501
0.806064396
-0.081494299
-0.698411033
0.997969843
-0.602033989
-0.206595202
0.874810338
-0.936976536
0.348132316
0.477825188
65
-0.018944456
0.028019024
-0.014961843
-0.009447695
0.026052391
-0.02907222
0.00689778
0.017717969
-0.032426624
0.02583605
0.002785117
-0.026225797
0.036384977
-0.018413963
-0.013097004
0.034450576
-0.036277556
1.888812886
-2.718255494
1.866959421
0.279068707
-2.435038713
3.199259387
-1.973887501
-0.623704943
3.01101097
-3.642104586
1.997969843
1.035995488
-3.593624782
4.026764029
-1.936976536
-1.498157298
4.159379079
A.5 Numerical Value Tables for the Van Der Pol Equation
Table 6: Perturbation and Numerical Values Determined for the Van Der Pol
Equation (=.01)
Y Perturbation
1
-0.715310079
-0.0025
0.740555511
-1.058904862
0.756962107
0.0025
-0.782207538
1.117809725
-0.798614134
-0.0025
0.823859566
-1.176714587
0.840266162
0.0025
-0.865511593
1.235619449
-0.88191819
-0.0025
0.907163621
-1.294524311
0.923570217
0.0025
-0.948815648
1.353429174
-0.965222245
-0.0025
0.990467676
-1.412334036
1.006874272
0.0025
-1.032119703
1.471238898
Y Multiple
Scales
Y Numerical
1
-0.717544628
8.54548E-15
0.738556114
-1.059416551
0.759718006
-2.61588E-14
-0.780991622
1.119573047
-0.802336314
4.87368E-14
0.823709742
-1.180010766
0.845068189
-7.35352E-14
-0.866366918
1.240236618
-0.887560558
9.14918E-14
0.908603521
-1.299739054
0.929450432
-1.10718E-13
-0.950056575
1.358006599
-0.970378338
1.50835E-13
0.990373647
-1.41454768
1.010002383
-1.73126E-13
-1.029226772
1.468910053
1
-0.715308572
-0.002534072
0.740870018
-1.059416984
0.757347418
0.002605122
-0.783449645
1.119573761
-0.799817226
-0.002680175
0.826326049
-1.180011592
0.842386545
0.002759142
-0.869155381
1.240237356
-0.884703033
-0.0028415
0.911576962
-1.299739495
0.926405093
0.002926298
-0.953225986
1.358006535
-0.967135468
-0.003012053
0.99374757
-1.414546908
1.006555088
0.0030967
-1.032810299
1.468908387
66
Absolute Error
(Multiple
Scales vs
Numerical)
0
0.002236056
-0.002534072
0.002313904
-4.32976E-07
-0.002370588
0.002605122
-0.002458023
7.14061E-07
0.002519089
-0.002680175
0.002616307
-8.26118E-07
-0.002681644
0.002759142
-0.002788463
7.37134E-07
0.002857525
-0.0028415
0.002973441
-4.41485E-07
-0.003045339
0.002926298
-0.003169411
-6.38997E-08
0.003242871
-0.003012053
0.003373923
7.72049E-07
-0.003447296
0.0030967
-0.003583527
-1.66624E-06
Absolute Error
(Perturbation vs
Numerical)
0
1.50724E-06
-3.40716E-05
0.000314507
-0.000512122
0.000385311
0.000105122
-0.001242107
0.001764037
-0.001203091
-0.000180175
0.002466483
-0.003297005
0.002120383
0.000259142
-0.003643788
0.004617907
-0.002784844
-0.0003415
0.004413341
-0.005215184
0.002834876
0.000426298
-0.004410338
0.004577362
-0.001913223
-0.000512053
0.003279894
-0.002212873
-0.000319184
0.0005967
-0.000690595
-0.002330511
-1.0485263
-0.0025
1.073771731
-1.53014376
1.090178327
0.0025
-1.115423758
1.589048623
-1.131830355
-0.0025
1.157075786
-1.647953485
1.173482382
0.0025
-1.198727814
1.706858347
-1.21513441
-0.0025
-1.048011733
2.17602E-13
1.066325186
-1.520698125
1.084138308
-2.64316E-13
-1.101425731
1.569586644
-1.118165687
3.13001E-13
1.134340085
-1.615329667
1.149934542
-3.16855E-13
-1.164938344
1.657764312
-1.179344367
3.67492E-13
-1.044356655
-0.003178281
1.070119773
-1.520695408
1.080275913
0.003254214
-1.105428917
1.569582752
-1.114100668
-0.003321966
1.138545217
-1.615324513
1.145675353
0.003378976
-1.169334777
1.657757855
-1.174903203
-0.003423445
67
0.003655078
-0.003178281
0.003794588
2.71691E-06
-0.003862395
0.003254214
-0.004003186
-3.89276E-06
0.004065018
-0.003321966
0.004205132
5.15366E-06
-0.004259188
0.003378976
-0.004396433
-6.45626E-06
0.004441164
-0.003423445
0.004169645
-0.000678281
-0.003651958
0.009448352
-0.009902415
0.000754214
0.009994841
-0.019465871
0.017729686
-0.000821966
-0.018530569
0.032628972
-0.027807029
0.000878976
0.029393037
-0.049100492
0.040231207
-0.000923445
Table 7: Perturbation and Numerical Values Determined for the Van Der Pol
Equation (=.05)
X
0
3.926991
7.853982
11.78097
15.70796
19.63495
23.56194
27.48894
31.41593
35.34292
39.26991
43.1969
47.12389
51.05088
54.97787
58.90486
62.83185
66.75884
70.68583
74.61283
78.53982
82.46681
86.3938
90.32079
94.24778
98.17477
102.1018
106.0288
109.9557
113.8827
117.8097
121.7367
125.6637
129.5907
Y Perturbation
1
-0.748123272
-0.0125
0.874350428
-1.294524311
0.95638341
0.0125
-1.082610566
1.589048623
-1.164643548
-0.0125
1.290870703
-1.883572934
1.372903685
0.0125
-1.499130841
2.178097245
-1.581163823
-0.0125
1.707390979
-2.472621556
1.789423961
0.0125
-1.915651117
2.767145868
-1.997684099
-0.0125
2.123911254
-3.061670179
2.205944236
0.0125
-2.332171392
3.35619449
-2.414204374
Y Multiple
Scales
1
-0.759718006
9.54311E-15
0.866366918
-1.299739054
0.970378338
-3.46252E-14
-1.066325186
1.569586644
-1.149934542
7.11143E-14
1.218955321
-1.764777172
1.273265719
-1.11311E-13
-1.314329515
1.881740046
-1.344415243
1.38149E-13
1.365939003
-1.943389019
1.381071207
-1.63382E-13
-1.391578068
1.973582897
-1.398809787
2.15621E-13
1.403757128
-1.987824615
1.407127574
-2.39125E-13
-1.409417186
1.994421102
-1.410969541
Y Numerical
1
-0.747966033
-0.013403983
0.880398134
-1.299528534
0.954246161
0.015498069
-1.085391266
1.56907201
-1.128744127
-0.017147211
1.242782828
-1.764004837
1.24846984
0.017093211
-1.340725129
1.880847285
-1.318537991
-0.015052885
1.39240169
-1.942499258
1.35620113
0.011594317
-1.416362673
1.972767831
-1.376194256
-0.007354759
1.425900721
-1.987115154
1.387399208
0.002739235
-1.428444397
1.993825342
-1.394422714
68
Absolute Error
(Multiple
Scales vs
Numerical)
0
0.011751972
-0.013403983
0.014031216
0.00021052
-0.016132177
0.015498069
-0.01906608
-0.000514635
0.021190415
-0.017147211
0.023827506
0.000772334
-0.024795879
0.017093211
-0.026395613
-0.00089276
0.025877251
-0.015052885
0.026462687
0.000889761
-0.024870076
0.011594317
-0.024784605
-0.000815066
0.022615531
-0.007354759
0.022143593
0.000709461
-0.019728366
0.002739235
-0.019027211
-0.00059576
0.016546827
Absolute Error
(Perturbation vs
Numerical)
0
0.000157239
-0.000903983
0.006047706
-0.005004223
-0.002137249
0.002998069
-0.0027807
-0.019976613
0.03589942
-0.004647211
-0.048087875
0.119568096
-0.124433846
0.004593211
0.158405712
-0.29724996
0.262625832
-0.002552885
-0.314989289
0.530122298
-0.43322283
-0.000905683
0.499288444
-0.794378037
0.621489843
0.005145241
-0.698010534
1.074555025
-0.818545028
-0.009760765
0.903726995
-1.362369148
1.01978166
133.5177
137.4447
141.3717
145.2987
149.2257
153.1526
157.0796
161.0066
164.9336
168.8606
172.7876
176.7146
180.6416
184.5686
188.4956
192.4226
196.3495
-0.0125
2.54043153
-3.650718801
2.622464512
0.0125
-2.748691667
3.945243113
-2.830724649
-0.0125
2.956951805
-4.239767424
3.038984787
0.0125
-3.165211943
4.534291735
-3.247244925
-0.0125
2.90529E-13
1.412020647
-1.997450568
1.412731717
-3.41747E-13
-1.413212462
1.998836407
-1.413537354
3.9285E-13
1.413756859
-1.999469221
1.413905133
-3.87055E-13
-1.414005279
1.999757945
-1.414072913
4.38051E-13
0.002044676
1.427706849
-1.996965526
1.399495979
-0.006901328
-1.425451914
1.998454034
-1.403666262
0.011788486
1.422497587
-1.999179232
1.407417697
-0.016688209
-1.419219222
1.999549071
-1.410973045
0.021592419
69
0.002044676
0.015686202
0.000485042
-0.013235738
-0.006901328
-0.012239452
-0.000382373
0.009871092
0.011788486
0.008740728
0.000289989
-0.006487436
-0.016688209
-0.005213943
-0.000208874
0.003099868
0.021592419
0.014544676
-1.112724681
1.653753275
-1.222968532
-0.019401328
1.323239753
-1.946789079
1.427058387
0.024288486
-1.534454218
2.240588192
-1.63156709
-0.029188209
1.745992721
-2.534742664
1.83627188
0.034092419