Академический Документы
Профессиональный Документы
Культура Документы
CHAPTER1
INTRODUCTION
CHAPTER1
LITERATURE REVIEW
CHAPTER3
OBJECTIVE
CHAPTER3
RESEARCH METHODOLOGY
CHAPTER4
DATA ANALYSIS
REFERENCES
1
CHAPTER1
INTRODUCTION
2
One of the most important tasks in a study of dynamical systems is the numerical
calculation of the trajectories. Thus far we have considered the integration method to be a
black box into which we pop the system, initial conditions, method and time range and
out pops a plot of the trajectories. Although this approach is common in courses on
dynamical systems it obscures many of the pitfalls of numerical integration.
It is not possible at the present state of the art to choose a ‘best’ algorithm for the
calculation of trajectories. There are several types of numerical algorithm, each with
their own advantages and disadvantages. We shall consider a class of methods known as
discrete variable methods. These methods approximate the continuous time dynamical
system by a discrete time system. This means that we are not really simulating the
continuous system but a discrete system which may have different dynamical
properties. This is an extremely important point.
The discrete variable methods which we consider fall into two main types, Runge-Kutta
methods and Linear Multistep methods. Maple has implementations of both types of
method as well as a number of more sophisticated techniques designed to overcome
some of the pitfalls of numerical solution. The more sophisticated methods still fall into
the discrete variable category.
3
5.1 TYPES OF METHOD
Although the dynamical systems which we are simulating are usually in more than one
dimension we can, without loss, restrict our numercal anlaysis of the methods to the
single non-autonomous differential equation
bg
x f t , x
bg
x t 0 x0
subject to the initial condition. We shall usually refer to the differential equation
together with the initial condition as an initial value problem (IVP). Discrete variable
methods yield a series of approximations
X n x (tn )
t n 1 tn h, n 0,1 N
4
Taylor Series Method
h2 h3
x(tn 1 ) x (tn ) hx(tn ) x tn ) (
( x tn )
2! 3!
h2 h3 h p ( p)
X n1 X n hX n X n X n Xn
2! 3! p!
dpX
where X ( p ) .
dt p
Runge-Kutta Methods
These methods are based on the notion of finding a formula which agrees with the
Taylor series as closely as possible without involving derivatives. For example consider
the possibility of matching the second order Taylor series method
5
2
h
X n 1 X n hX n Xn
2!
X n 1 X n h (t n , X n , h)
where
(t , x , h) f (t , x ) f (t ah, x bhf (t , x ))
h
(t , x , h) x(t ) (
x t)
2
h
f (t , x ) f t (t , x ) f x (t , x ) f (t , x )
2
(t , x , h) ( ) f (t , x ) h af t (t , x ) bf x (t , x ) f (t , x ) O(h 2 )
1, a 21 , b 1
2
6
1
1 , , a b
2
X n 1 X n
h
2
b
f (tn , X n ) f tn h, X n hf (tn , X n ) g
b
X n1 X n hf tn h2 , X n h2 f (tn , X n ) g
Unfortunately the terminology for naming second order Runge-Kutta methods is not
standardised and Maple calls the improved Euler method the Heun formula and the
modified Euler method the improved polygon method.
The procedure above can be extended to give higher order methods such as the
7
h
X n 1 X n ( k1 2 k 2 2 k 3 k 4 )
6
k1 f (t n , X n )
k 2 f (t n 21 h, X n 21 hk1 )
k 3 f (t n 21 h, X n 21 hk 2 )
k 4 f (t n h, X n hk 3 )
8
CHAPTER2
LITERATURE REVIEW
9
Classical Methods
Maple contains a number of one-step methods for the numerical solution of initial value
problems. These are referred to as classical methods and are invoked by including the
option method=classical[type] in the call to dsolve. Here type can be one of
heunform the Heun formula (also known as the trapezoidal rule, or the improved
Euler method);
10
Worked Example 2 - The Forward Euler Method
y 2 xy 2 , bg
y 0 1
> ivp:={diff(y(x),x)=-2*x*y(x)^2,y(0)=1}:
In this case Maple can find the exact solution using dsolve
> Exactsoln:=rhs(dsolve(ivp,y(x)));
1
Exactsoln :=
x 21
Now we use Euler's method to obtain the numerical solution. Note that this method like
all the other methods of type classical uses a fixed stepsize which we provide.
> es0:=dsolve(ivp,y(x),type=numeric,
method=classical[foreuler],stepsize=0.001):
> es0(0.4);
[ x.4, y( x ).8623085097414066 ]
11
and plot the solution for a range of values of x
> odeplot(es0,[x,y(x)],0..6,labels=[x,y]);
12
CHAPTER3
OBJECTIVE
\
13
CHAPTER4
RESEARCH
METHODOLOGY
14
Research Design:
Descriptive Research: Descriptive research includes survey and fact-findings enquire of
different kinds. The major purpose of descriptive research is description of the state
affairs, as it exists at present.
Data Collection:
The study is based on the data collected through primary and secondary sources.
Primary Data:
An interview schedule was designed to collect primary data from various broadband
users.
Secondary Data:
Secondary data was collected from journals, magazines, web sites and from other
relevant publications.
Sampling Design:
The sampling design mainly consists of the sample taken for the study along with the
sample size, sample frame and sampling method.
Sample Universe:
All customers using broadband connection was taken as the sample universe.
Sample Size:
From the universe, sample sizes of 200 problems were selected for the purpose of the
study.
Sampling Method:
15
Convenience sampling was used, based on the willingness and availability of the
respondents. The study was conducted on consumers with different type of business.
Sample Size —200 respondents Sample Unit- Students of Graduation and the Post
Graduation have been taken as sample unit. Sampling Area – Bhubaneswar. Sampling
Technique - Random Sampling technique
DATA COLLECTION:
• Primary data has been used by me in the form of Questionnaire & Observation, which
are the two basic methods of collecting primary data, which suffices all research
objectives.
• Secondary data sources like catalogue of the company, product range book of the
company & various internet sites such as motorola.com & google.com have been used.
16
CHAPTER5
DATA ANALYSIS
17
Construct an array whose elements compare the exact solution to the numerical
solution for three different stepsizes.
> mm:=array(1..8,1..5):
mm[1,1]:=`x(k)`:mm[1,2]:=`Exactsoln`:mm[1,3]:=`h=0.1`:
mm[1,4]:=`h=0.01`:mm[1,5]:=`h=0.001`:
for i from 2 to 8 do
mm[i,1]:=0.1*(i-2):
mm[i,2]:=evalf(ExactSoln(x(i-2)),5):
for j from 3 to 5 do
mm[i,j]:=evalf(EulerSoln(x(i-2),10^(-j+2)),5)
od:
od:
> eval(mm);
18
x(k) Exact soln h=0.1 h=0.01 h=0.001
0 1. 1. 1. 1.
.1 .99010 1. .99107 .99020
.2 .96154 .98000 .96330 .96171
.3 .91743 .94158 .91969 .91766
.4 .86207 .88839 .86448 .86231
.5 .80000 .82525 .80229 .80023
.6 .73529 .75715 .73727 .73549
Another possibility is to compare the errors at each stepsize. Firstly define a function
giving the error
> err:=(x,h)->ExactSoln(x)-EulerSoln(x,h):
> tt:=array(1..8,1..4):
tt[1,1]:=`x(k)`:tt[1,2]:=`h=0.1`:tt[1,3]:=`h=0.01`:tt[1,4]:=`h=0.001`:
for i from 2 to 8 do
tt[i,1]:=0.1*(i-2);
for j from 2 to 4 do
tt[i,j]:=evalf(err(x(i-2),10^(-j+1)),5);
od:
od:
> eval(tt);
19
x(k) h=0.1 h=0.01 h=0.001
0 0 0 0
.1 -.00990 -.00097 -.00010
.2 -.01846 -.00176 -.00017
.3 -.02415 -.00226 -.00023
.4 -.02632 -.00241 -.00024
.5 -.02525 -.00229 -.00023
.6 -.02186 -.00198 -.00020
y 4 y 4 x, bg
y 0 1
> IVP:={diff(y(x),x)=-4*y(x)+4*x,y(0)=1}:
> dsolve(IVP,y(x));
1 5 ( 4 x )
y( x )x e
4 4
20
Use the 2nd order classical Runge-Kutta method
> rk2:=h->dsolve(IVP,y(x),type=numeric,method=classical[rk2],stepsize=h):
> x:=k->k*0.5:
> RK2Soln:=(x,h)->rhs(rk2(h)(x)[2]):
> ExactSoln:=x->x-1/4+5/4*exp(-4*x);
1 5 ( 4 x )
ExactSoln := xx e
4 4
> mm:=array(1..10,1..5):
mm[1,1]:=`x(k)`:mm[1,2]:=`Exactsoln`:mm[1,3]:=`h=0.25`:
mm[1,4]:=`h=0.5`:mm[1,5]:=`h=0.75`:
for i from 2 to 10 do
mm[i,1]:=0.5*(i-2):
mm[i,2]:=evalf(ExactSoln(x(i-2)),5):
od:
> eval(nm);
21
x(k) Exact soln h=0.25 h=0.5 h=0.75
0 1. 1. 1. 1.
.5 .41918 .56250 1.5000 1.5000
1.0 .77290 .82813 2. 2.3125
1.5 1.2531 1.2695 2.5000 9.0625
2.0 1.7504 1.7549 3. 9.5625
2.5 2.2501 2.2512 3.5000 12.016
3.0 2.7500 2.7503 4. 51.578
3.5 3.2500 3.2501 4.5000 52.078
4.0 3.7500 3.7500 5. 64.785
Note that as the stepsize is increased the numerical solution fails to represent the exact
solution accurately. Indeed for a stepsize of 0.75 the numerical solution 'blows up'. This
is due to non-convergence as a result of the numerical method becoming unstable. We
shall consider this phenomenon next.
22
Exercises 1
1. Use the classical numerical methods foreuler, heunform, rk3, rk4 and adambash to
attempt to obtain a numerical solution of the IVPs
(a)
dx
dt
bg
2tx 2 , x 0 1
(b)
dx
dt
b g bg
5x 1 x , x 0 0.5
Use a range of stepsizes in the interval [0,1]. At what approximate value of the stepsize
do the methods become unstable.
2. Use each of the methods above to solve the systems of differential equations
x x xy , x (0) 0.5
(a)
y y xy , y (0) 0.5
x y, x(0) 0
(b)
y x (1 x 2 ) y, y(0) 0.5
23
x ( y z), x (0) 1
(c) y x 0.2 y , y (0) 1
z 0.2 8z xz, z(0) 1
In each case use odeplot to obtain time series and phase plots.
24
5.3 LOCAL AND GLOBAL ERRORS
l q
The output of a discrete variable method is a set of points t n , X n and the output of the
bg
dynamical system is a continuous trajectory x t . For the numerical results to provide a
bg
X N x tN
where is some defined error tolerance, at each solution point. This difference is called
the global error and is the accumulated error over all solution steps. Unfortunately it is
extremely difficult to accomplish this and we have to confine ourselves to controlling
the local error
~
bg
X n x tn
~
at each step where X n is the numerical solution obtained on the assumption that the
numerical solution at the previous solution point is exact.
There are two sources of local error,the roundoff error and the truncation error.
Roundoff Error
The roundoff error is the error which arises from the fact that numerical methods are
implemented on digital computers which only calculate results to a fixed precision
25
which is dependent on the computer system used. Note that since roundoff errors
depend only on the number and type of arithmetic operations per step and is thus
independent of the integration stepsize h.
Truncation Error
26
REFERENCE
1.
27
7. Rombaldi, Jean-Étienne (2004). Éléments d'analyse réelle : CAPES et
9. Rudin, Walter (1987). Real and Complex Analysis (3rd ed.). New
58807-2.
28