You are on page 1of 18

Ain Shams University

College of Engineering
Department of Electrical Engineering
2nd Year Electrical
System Dynamics and Control Components

RUNGE-KUTTA 4th ORDER METHOD


Under Supervision of Professor Dr. A. A. Wahdan

Mohammed Shadied Abbas


nd
2 Year Electrical Engineering
Group B
Section : 9
Bench No. : 6
Table of Contents :
I – Differential Equations .................................................... 3
i- Definition ....................................................................... 3
ii- Ordinary Differential Equations ................................... 4
iii- Numerical Ordinary Differential Equations ................ 5

II – Runge-Kutta Methods ................................................... 6


i- Introduction .................................................................... 6
ii- Fourth Order Runge-Kutta Method .............................. 6
iii- Computer-Aided 4th Order Runge-Kutta Method ........ 9
1- The Programming Language Python ...................... 9
2- The Program Code .................................................. 9
3- A Sample Runtime ................................................ 14

III – Utilising Runge-Kutta Method In Solving Higher Order


Differential Equations ............................................... 15

IV – References And Sources …........................................ 18

2
I – Differential Equations :
i- Definition :

A differential equation is a mathematical equation for an


unknown function of one or several variables that relates the
values of the function itself and its derivatives of various orders.
Differential equations play a prominent role in engineering,
physics, economics and other disciplines.

A simplified real world example of a differential equation is


modeling the acceleration of a ball falling through the air
(considering only gravity and air resistance). The ball's
acceleration towards the ground is the acceleration due to gravity
minus the deceleration due to air resistance. Gravity is a constant
but air resistance is proportional to the ball's velocity. This means
the ball's acceleration is dependent on its velocity. Because
acceleration is the derivative of velocity, solving this problem
requires a differential equation.

Differential equations are mathematically studied from several


different perspectives, mostly concerned with their solutions,
functions that make the equation hold true. Only the simplest
differential equations admit solutions given by explicit formulas.
Many properties of solutions of a given differential equation may
be determined without finding their exact form. If a self-contained
formula for the solution is not available, the solution may be
numerically approximated using computers. The theory of
dynamical systems puts emphasis on qualitative analysis of
systems described by differential equations, while many
numerical methods have been developed to determine solutions
with a given degree of accuracy.

3
ii- Ordinary Differential Equations :

In mathematics, an ordinary differential equation (or ODE) is a


relation that contains functions of only one independent variable,
and one or more of its derivatives with respect to that variable.

A simple example is Newton's second law of motion, which leads


to the differential equation:
..
m x = F t

For the motion of a particle of constant mass m. In general, the


force F depends upon the position of the particle x(t) at time t, and
thus the unknown function x(t) appears on both sides of the
differential equation, as is indicated in the notation F(x(t)).

Ordinary differential equations arise in many different contexts


including geometry, mechanics, astronomy and population
modeling. Many famous mathematicians have studied differential
equations and contributed to the field, including Newton, Leibniz,
the Bernoulli family, Riccati, Clairaut, d'Alembert and Euler.

Much study has been devoted to the solution of ordinary


differential equations. In the case where the equation is linear, it
can be solved by analytical methods. Unfortunately, most of the
interesting differential equations are non-linear and, with a few
exceptions, cannot be solved exactly. Approximate solutions are
arrived at using computer approximations (see numerical ordinary
differential equations).

4
iii- Numerical Ordinary Differential Equations :

Numerical ordinary differential equations is the part of numerical


analysis which studies the numerical solution of ordinary
differential equations (ODEs). This field is also known under the
name numerical integration, but some people reserve this term for
the computation of integrals.

Many differential equations cannot be solved analytically, in


which case we have to satisfy ourselves with an approximation to
the solution. The algorithms studied here can be used to compute
such an approximation. An alternative method is to use techniques
from calculus to obtain a series expansion of the solution.

Ordinary differential equations occur in many scientific


disciplines, for instance in mechanics, chemistry, biology, and
economics. In addition, some methods in numerical partial
differential equations convert the partial differential equation into
an ordinary differential equation, which must then be solved.

Some elementary methods were made to address this problem


numerically such as The Euler Method and The Backward Euler
Method. After that, other methods have emerged (which are
generally more accurate and efficient).

Fig.(1) : Illustration of numerical integration for


the differential equation y' = y,y(0) = 1. Blue: the
Euler method, Green: the midpoint method and
Red: the exact solution, y = et. The step size is h =
1.0.
Figure (1)

5
II – Runge-Kutta Methods :

i- Introduction :

A method of numerically integrating ordinary differential


equations by using a trial step at the midpoint of an interval to
cancel out lower-order error terms. The second-order formula is
k 1 =h f  x n , y n 
k 2=h f  x n1/ 2 h , y n 1/ 2 k 1 
3
y n1 = y n k 2 O h 

ii- Fourth Order Runge-Kutta Method :

The fourth order Runge-Kutta method has been derived from


Taylor formula where
2
y  x n1 = y n h f n h / 2 ! [ f x n  f n  f y n ]
so it's possible to develop formula equivalent to third, fourth,
fifth or even higher order Taylor formula, which do not involve
derivatives of f .

The derivation of a similar formula in the case of five term


Taylor series formula

' h 2 ' ' h3 ' ' ' h4 4


y n1 = y n h y n  y n  y n  y n
2! 3! 4!

where derivatives y'' , y''' and y(4) are obtained by successively


differentiating y'=f(x,y) is too lengthy and tedious but not
difficult.
6
The formula obtained after the derivation is
1
y n1 = y n  w 1 2w 2 2w 3 w 4 
6
where ;

w 1 =hf  x n , y n 
1 1
w 2 =hf  x n  h , y n  w 1 
2 2
1 1
w3=h f  x n  h , y n  w 2 
2 2
w 4=hf  x n h , y n w 3 

This is the classical Runge-Kutta formula. It's equivalent to five


term Taylor series above formula and hence it's fourth order
method.

Example :
Apply the Runge-Kutta method to the IVP
'
y =1− x4y ; y 0=1
to find y(0.2). Take h = 0.1 and use six digits.

Solution :

n x y w=h(1-x+4y) Δ y
0 0.00 1.00000 0.500000 0.50000
0.05 1.25000 0.595000 1.19000
0.05 1.29750 0.614000 0.22800
0.10 1.30700 0.612800 0.612800

1
y1 =1 2.5228=1.420466
6 2.5228
7
n x y w=h(1-x+4y) Δ y
1 0.10 1.42046 0.65818 0.65818
0.15 1.74955 0.78482 1.56964
0.15 1.81282 0.81014 1.42029
0.20 1.82553 0.81021 0.81021

1
y 2 =1.420466 4.45832=2.44469
6 4.45832

So, y(0.2) = 2.44469

8
iii- Computer-Aided 4th Order Runge-Kutta
Method :

1- The Programming Language Python :

Python is a general-purpose high-level programming language


that firstly appeared in 1991. Its design philosophy emphasizes
code readability. Python's core syntax and semantics are
minimalistic, while the standard library is large and
comprehensive. Its use of whitespace as block delimiters is
unusual among popular programming languages.

Python supports multiple programming paradigms (primarily


object oriented, imperative, and functional) and features a fully
dynamic type system and automatic memory management,
similar to Perl, Ruby, Scheme, and Tcl. Like other dynamic
languages, Python is often used as a scripting language.

The language has an open, community-based development model


managed by the non-profit Python Software Foundation, which
maintains the de facto standard definition of the language in
CPython, the reference implementation.

Python runs on Windows, Linux/Unix, Mac OS X, OS/2, Amiga,


Palm Handhelds, and Nokia mobile phones. Python has also been
ported to the Java and .NET virtual machines.

For more information about Python, you may refer to


"http://python.org/"

2- The Program Code :

9
H:\Users\Mohammed\Desktop\Runge\RK4.py

1 # In Th e N ame OF Al lah , M ost Gr aci ous , M ost Me rci ful


2 # Th e f oll owi ng cod e r epr ese nts a 1st or der OD E s olv er usi ng Run ge Kut ta 4
me tho d
3 # Th e e qua tio n s hou ld be wri tte n a s y ' = f( x,y ) a nd it sho uld be va lid
ac cor din g t o p yth on' s s ynt ax
4 # Wr itt en by Moh amm ed Abb as "ab bas @ie ee. org " u sin g p yth on scr ipt ing la ngu age
v2 .6. 1
5 # La st Edi t : 9: 52 PM , 1 6th of Ap ril 20 09
6
7
8
9
10 d ef RK4 ():
11
12
13 # u ser 's inp ut wit h p rob lem gi ven s :
14
15 pri nt "Nu mer ica l 1 st ord er ODE so lve r u sin g R ung e-K utt a 4 me tho d"
16 pri nt 'Al gor ith m w rit ten by Mo ham med Ab bas "a bba s@i eee .or g"\ nUs ing py tho n
sc rip tin g l ang uag e v 2. 6.1 '
17 pri nt "\n Imp ort ant no tes :\ n1- Wh en ent eri ng the eq uat ion , p lea se mak e
su re you r u se * f or mul tip lic ati on and ** fo r e xpo nen tia tio n
ex : x ^2 sho uld be wr itt en as x** 2. \n2 - U se par ent hes es () to ski p
or der of pr ece den ce. "
18
19 # e qua tio n f (x, y) is ent ere d a s a st rin g :
20
21 eqn = raw _in put ("\ nPl eas e e nte r t he equ ati on as y' = " )
22
23 # i nit ial co ndi tio ns, st ep siz e a nd the po int of so lut ion :
24
25 x0 = i npu t(" Ple ase en ter x0 = ")
26
27 y0 = i npu t(" Ple ase en ter y( x0) = ")
28
29 h = in put ("P lea se def ine th e s tep si ze h = ")
30
31
32 s = in put ("P lea se spe cif y the po int at wh ich yo u n eed th e s olu tio n y (x) >
> ")
33
34 # n re pre sen ts how ma ny tim es the it era tio n i s d one us ing th e f oll owi ng
f or loo p t ech niq ue :
35
36 n = (s *1. 0)/ h
37 n = in t(n )
38
39 # f our fl oat ing po int ar ray s f or eac h c olu mn of the it era tio n i nit ial ly
z ero ed :
40
41 xA = [ 0.0 ,0. 0,0 .0, 0.0 ]
42 yA = [ 0.0 ,0. 0,0 .0, 0.0 ]
43 wA = [ 0.0 ,0. 0,0 .0, 0.0 ]
44 dyA = [0. 0,0 .0, 0.0 ,0. 0]
45
46
47
48
H:\Users\Mohammed\Desktop\Runge\RK4.py

49
50
51
52 # t his is th e b egi nni ng of the it era tio n p roc ess , ' for ' l oop sh oul d g o a s
m any ti mes as n is :
53
54 for i in ran ge( n):
55
56 # her e w e s imp ly tak e t he val ues of in iti al con dit ion va ria ble s x 0,y 0
ent ere d b y u ser :
57
58 xA [0] =x0
59 yA [0] =y0
60
61 #c hec k
62 #p rin t x A[0 ],y A[0 ]
63
64 # the n w e a ssi gn the m t o v ari abl es cal led x and y to hel p e val ()
fun cti on eva lua te the en ter ed equ ati on as pyt hon 's val id exp res sio n,
it the n h app ens ea ch and ev ery ti me we' d l ike to us e e val () fun cti on
wit h o the r v alu es :
65
66
67 x= xA[ 0]
68 y= yA[ 0]
69
70 # her e w e u se eva l() fu nti on to eva lua te the en ter ed exp res sio n w ith
x0, y0 val ues to ge t w 1(0 ) a nd the n w e u se w1( 0) to get th e n ext
val ue of y :
71
72 wA [0] = h * e val (eq n)
73 yA [1] = y A[0 ] + (0 .5* wA[ 0])
74
75 #c hec k
76 #p rin t w A[0 ],y A[1 ]
77
78 # her e's ho w t he nex t v alu e o f x is ob tai ned an d h ow we use th e
pre vio us tec hni que ag ain :
79
80 xA [1] = x 0+( 0.5 *h)
81 x= xA[ 1]
82 y= yA[ 1]
83
84 #c hec k
85 #p rin t x A[1 ]
86
87 wA [1] = h * e val (eq n)
88 yA [2] = y A[0 ] + (0 .5* wA[ 1])
89
90 #c hec k
91 #p rin t w A[1 ],y A[2 ]
92
93
94
95
96
97
98
H:\Users\Mohammed\Desktop\Runge\RK4.py

99
100 xA [2] = x 0+( 0.5 *h)
101 x= xA[ 2]
102 y= yA[ 2]
103
104 #c hec k
105 #p rin t x A[2 ]
106
107 wA [2] = h * e val (eq n)
108 yA [3] = y A[0 ] + (0 .5* wA[ 2])
109
110 #c hec k
111 #p rin t w A[2 ],y A[3 ]
112
113 xA [3] = x 0+h
114 x= xA[ 3]
115 y= yA[ 3]
116
117 #c hec k
118 #p rin t x A[3 ]
119
120
121 wA [3] = h * e val (eq n)
122
123 #c hec k
124 #p rin t w A[3 ]
125
126 # fin all y w e c alc ula te the va lue of dy fo r t he ite rat ion :
127
128 dy A[0 ] = wA [0]
129 dy A[1 ] = 2* wA[ 1]
130 dy A[2 ] = 2* wA[ 2]
131 dy A[3 ] = wA [3]
132
133 #c hec k
134 #p rin t d yA[ 0], dyA [1] ,dy A[2 ],d yA[ 3]
135
136 su m = dy A[0 ]+d yA[ 1]+ dyA [2] +dy A[3 ]
137
138 # thi s i s h ow we get yn +1 :
139
140 re sul t = yA [0] + ( sum /6. 0)
141
142 # bef ore th e i ter ati on end s, we' d l ike to us e s ome va lue fr om it for
the ne xt one :
143
144 x0 =xA [3] # x n+1
145 y0 =re sul t # y n+1
146
147 #c hec k
148 #p rin t x 0,y 0
149
150 # f unc tio n r etu rns th e f ina l r esu lt of the lo op whi ch' s t he sol uti on of
t he DE at the po int of in ter est an d t he val ue of thi s p oin t :
151 # u nli ke C a nd som e o the r l ang uag es, py tho n i s c apa ble of re tur nin g
m ult ipl e s tuf f n ot jus t o ne var iab le wit hou t t he exp lic it use of
p oin ter s.
152
H:\Users\Mohammed\Desktop\Runge\RK4.py

153
154
155 ret urn re sul t,s
156
157 # e nd of RK4 () fun cti on
158
159
160
161
162
163
164 # he re we sim ply ca ll the fu nct ion to re cei ve the de sir ed val ues :
165
166 f ina l,s = RK4 ()
167
168 # th en we rou nd the so lut ion to 5 dig its :
169
170 f ina l = ro und (fi nal ,5)
171
172 # th e o utp ut :
173
174 p rin t " \ny (%s ) = %f " %(s ,fi nal )
175
176 # sc rip t t erm ina tio n :
177
178 e sc = r aw_ inp ut( "\n \nP res s a ny key to en d . .." )
179
3- A Sample Runtime :

A sample successful runtime of the code clearly represents the


previous example.
Numerical 1st order ODE solver using Runge-Kutta 4 method
Algorithm written by Mohammed Abbas "abbas@ieee.org"
Using python scripting language v 2.6.1

Important notes :
1- When entering the equation, please make sure your use * for multiplication
and ** for exponentiation ex: x^2 should be written as x**2.
2- Use parentheses () to skip order of precedence.

Please enter the equation as y' = 1-x+(4*y)


Please enter x0 = 0
Please enter y(x0) = 1
Please define the step size h = 0.1
Please specify the point at which you need the solution y(x) >> 0.2

y(0.2) = 2.444690

Press any key to end ...

14
III – Utilizing Runge-Kutta Method
In Solving Higher Order
Differential Equations :
Any differential equation of order n can be written as a system of
n first-order differential equations. Given an explicit ordinary
differential equation of order n and dimension 1,
' '' n−1 n
F  x , y , y , y ,..... , y = y
we define a new family of unknown functions
n−1
y n := y

We can then rewrite the original differential equation as a


system of differential equations with order 1 and dimension n.
'
y1 = y 2
'
y 2= y 3

.
.
.
'
y n−1 = y n
'
y n =F  y n ,.... , y 1 , x

which can be written concisely in vector notation as


'
y = F  x , y ; with y := y ,..... , y n 

15
Suppose we obtained a system from a higher order DE to be as
follows :
' '
x = f t , x , y , y = g t , x , y ; x 0= x 0 , y 0= y0

then the Runge-Kutta formulas for the system will be :

1
x n1= x n  v 12 v 22 v 3 v 4 
6
1
y n1 = y n  w 1 2w 2 2w 3w 4 
6

where :

v1 =hf t n , x n , y n 
w 1 =hg t n , x n , y n 

1 1 1
v 2 =hf t n h , x n v1 , y n  w 1 
2 2 2
1 1 1
w 2 =hg t n  h , x n  v 1 , y n w 1 
2 2 2

1 1 1
v 3=hf t n  h , x n  v 2 , y n w 2 
2 2 2
1 1 1
w 3 =hg t n  h , x n  v 2 , y n  w 2 
2 2 2

v 4 =hf t n h , x n v 3 , y n w 3 
w 3 =hg t n h , x n v 3 , y n w 3 

As the original formula error is O h5  , so the accumulated


formula error is O h 4  . This method is sufficiently accurate.

16
Example :
Use Runge-Kutta method with step size h = 0.25 to compute
x(0.5) for the IVP x ' ' −tx ' − x=0 ; x 0=0 , x ' 0=1
Use five digits.

Solution :

The equivalent system is :


'
x = y= f
'
y = xty=g
x 0=0 , y 0=1

The iteration goes as follows :

n t x y V=hf W=hg Δx Δy
0 0.000 0.00000 1.0000 0.25000 0.00000 0.25000 0.00000
0.125 0.12500 1.0000 0.25000 0.06250 0.50000 0.12500
0.125 0.12500 1.0313 0.25783 0.06347 0.51566 0.12696
0.250 0.25783 1.0635 0.26588 0.13093 0.26588 0.13093
x 1=00.25526=0.25526
y1 =10.063815=1.0638 0.28853 0.063815

n t x y V=hf W=hg Δx Δy
1 0.250 0.25526 1.0638 0.26595 0.13030 0.26595 0.13030
0.375 0.38824 1.1290 0.28225 0.20290 0.56450 0.40580
0.375 0.39639 1.1653 0.29133 0.20848 0.58266 0.41696
0.500 0.54659 1.2723 0.31808 0.29569 0.31808 0.29569
x 0.5≈ x 2 =0.255260.28853=0.54379
' 0.28853 0.20813
x 0.5= y 0.5≈ y 2 =1.06380.20813=1.2719
17
Although we took n = 2 only, the total error made in computing
x(0.5) is 0.00004 which is much better than any of the second
order methods with n = 5.

The code can be moderately modified to perform for this method.

IV – References And Sources :


1- Modeling and Analysis of Dynamic Systems
Publisher: Wiley; 3rd edition (August 6, 2001)
ISBN-13: 978-0471394426
By Charles M. Close, Dean K. Frederick and Jonathan
C. Newell

2- Engineering Mathematics (3) Part II, ASU 2nd Year Electrical


Publisher: Dar Teba Press
By Professor Dr. Reda El Barkouky

3- Learning Python 3rd Edition


Publisher: O'Reilly Media, Inc.; 3rd edition (July 16, 2008)
ISBN-13: 978-0596513986
By Mark Lutz

4- The Wikipedia, http://wikipedia.org/

5- Wolfram Mathworld, http://mathworld.wolfram.com/

18

You might also like