Anda di halaman 1dari 153

LECTURE NOTES

SUBJECT NAME: NUMERICAL METHODS


SUBJECT CODE:MA1251

PREPARED BY
A.NALAYINI DEVI M.SC., M.PHIL.,
LECTURER, DEPARTMENT OF MATHEMATICS,
N.P.R.COLLEGE OF ENGINEERING AND
TECHNOLOGY,
NATHAM-624 401

UNIT I
SOLUTION OF EQUATIONS AND
EIGENVALUE PROBLEMS

Solution of equation
Fixed point iteration: x=g(x) method
Newtons method
Solution of linear system by Gaussian elimination
Gauss Jordon method
Gauss Jacobi method
Gauss Seidel method
Inverse of a matrix by Gauss Jordon method
Eigenvalue of a matrix by power method
Jacobi method for symmetric matrix.

Solution of Algebraic and Transcendental Equations


The equation of the form
f ( x)

a0 x n

a1 x n

a2 x n

.......

an 1 x

an

0 ------------(A)

is called rational integral

equation
Here, a 0 0 , n is a positive integer, a0, a1, a2, ....., an are constants.
The rational integral equation is classified into two parts
1. Algebraic Equation
2. Transcendental Equation

Algebraic Equation
In this equation, f(x) is a polynomial purely in x as in (A)
Example: x 3 3x 1 0, x 4 2 x 3 3x 2 2 x 1 0

Transcendental Equation
In this equation, f(x) contains some other functions such as trigonometric,
logarithmic or exponential etc.,
Example: 3x-cosx-1=0, xlog10x-1.2=0.

Properties
1. If f(a) and f(b) have opposite signs then one root of f(x)=0 lies between a
and b
2. To find an equation whose roots are with opposites signs to those of the
given equ. change x to (x)
3. To find an equation whose roots are reciprocals of the roots of of the given
equ. change x to 1
x

4. Every equation of an odd degree has atleast one real root whose sign is
opposite to that of its last term.
5. Every equation of an even degree with last term negative have atleast a pair
of real roots one positive and other negative.

Methods for solving Algebraic and Transcendental Equations


Fixed point iteration: x = g(x) method (or) Method of successive
approximation.
Newtons method (or) Newtons Raphson method

Fixed point iteration: x = g(x) method (or) Method of successive


approximation
Let f(x) =0 be the given equation whose roots are to be determined.

Steps for this method


1. Use the first property find a and b where the roots lies between.

Write the given equation in the form x = (x) with the condition ' ( x) 1
Let the initial approximation be x0 which is lies in the interval (a, b)
Continue the process using xn = (xn-1)
If the difference between the two consecutive values of xn is very small then
we stop the process and that value is the root of the equation.
Convergence of iteration method
The iteration process converges quickly if ' ( x) 1 where x= (x) is the
given equation. If
will become infinitely large and hence this
' ( x) 1 , xn
process will not converge. The convergence is linear.
Example: Consider the equation f(x) = x3+x-1 = 0
we can write x= (x) in three types
1. x=1-x3
2. x 1
2.
3.
4.
5.

1 x2

3. x 1 x 1 3
but we take the type which has the convergence property
f(x) = x3+x-1
f(0)=-ve and f(1)=+ve
hence the root lies between 0 and 1
Now consider The equation (1.) x=1-x3

' ( x) 1

Here (x)= 1-x3 , (x)=-3x2


at x=0.9 (x)=-3(0.9)2
(x)=-3(0.81)=-2.81

' ( x) 1

this equation x=1-x3 will not converge


so the iteration will not work if we consider this equation
Now consider
the equation x 1
1 x2

Here

' ( x)

2x
(1 x 2 ) 2

at x=0.9
' ( x)

2 0.9
(1 0.9 2 ) 2

1.8
3.2761

0.5494

' ( x) 1

the equation is converge


we use this equation.
no need to consider the third type.
Problems based on Fixed point iteration
1. Find a real root of the equation x3+x2-1=0 by iteration method.

Solution:
Let f(x)= x3+x2-1
f(0)= -ve and f(1)=+ve

Hence a real root lies between 0 and 1.


Now can be written as x

1
1 x

( x)

In this type only ' ( x) 1 in (0, 1)


Let the initial approximation be x0=0.5
x1

( x0 )

x2

( x1 )

x3

( x2 )

x4

( x3 )

x5

( x4 )

x6

( x5 )

x7

( x6 )

1
0.81649
0.5 1
1
0.74196
0.81649 1
1
0.75767
0.74196 1
1
0.75427
0.75767 1
1
0.75500
0.75427 1
1
0.75485
0.75500 1
1
0.75488
0.75485 1

Here the difference between x6 and x7 is very small.


therefore the root of the equation is 0.75488
-------------------------------------------------------------------------------------------2. Find the real root of the equation cosx = 3x-1, using iteration method.
Solution:
Let f(x)= cosx - 3x-1
f(0)= +ve and f(1)=-ve
A root lies between 0 and /2
The given equation can be written as
x

1
(1 cos x)
3
sin x
' ( x)
3

( x)

Clearly, ' ( x) 1 in (0, /2)


Let the initial approximation be x0=0

x1

( x0 )

x2

( x1 )

x3

( x2 )

x4

( x3 )

x5

( x4 )

x6

( x5 )

x7

( x6 )

x8

( x7 )

1
(1
3
1
(1
3
1
(1
3
1
(1
3
1
(1
3
1
(1
3
1
(1
3
1
(1
3

cos 0) 0.66667
cos 0.66667) 0.59529
cos 0.59529) 0.60933
cos 0.60933) 0.60668
cos 0.60668) 0.60718
cos 0.60718) 0.60709
cos 0.60709) 0.60710
cos 0.60710) 0.60710

Since the values of x7 and x8 are equal, the required root is 0.60710
--------------------------------------------------------------------------------------------3.
Find the negative root of the equation x3-2x+5=0
Solution:
The given equation is x3-2x+5=0 ------------------------------------------------(1)
we know that if , , are the roots of the equation (1), then the equation whose
roots are -, -, - is x3+(-1) 0 x2+(-1)2(-2x)+(-1)35=0------------------------(2)
The negative root of the equation (1) is same as the positive root of the equation
(2)
Let f(x)= x3-2x+5
Now f(2)=-ve and f(3) =+ve
Hence the root lies between 2 and 3. Equation (2) can be written as
1

( 2 x 5) 3

where

( x)

' ( x) 1 in (2, 3)

Let the initial approximation be x0=2


Since the values of x6 and x7 are equal, the root is 2.09455

Therefore the negative root of the given equation is -2.09455


--------------------------------------------------------------------------------------------------

Newtons method (or) Newtons Raphson method (Method of tangents)


Let f(x)=0 be the given equ. whose roots are to be determined.

FORMULA: x
n 1

f (x )
n
f '(x )
n

Steps for this method


i.
ii.
iii.
iv.

Use the first property find a and b where the roots lies between.
The initial approximation x0 is a if f (a) f (b) ; the initial approximation x0
is b if f (b) f (a) in the interval (a, b).
Use the formula and continue the process
If the difference between the two consecutive values of xn+1 is very small
then we stop the process and that value is the root of the equ.

Note
The process will evidently fail if f(x) = 0 in the neighbourhood of the root.
In such cases Regula-Falsi method should be used.
If we choose initial approximation x0 close to the root then we get the root of
the equ. very quickly.
The order of convergence is two

Condition for convergence of Newtons Raphson method

f ( x). f " ( x)

f ' ( x)

Problems based on Newtons Method


1. Compute the real root of xlog10x=1.2 correct to three decimal places using
Newtons Raphson Method

Solution:
Let f(x)= xlog10x-1.2

Now f(2)=-ve and f(3) =+ve


Hence the root lies between 2 and 3.
Let the initial approximation be x0=3
f(x)= xlog10x-1.2
1
f ' ( x) log10 x x. log10 e
x
log10 x 0.4343

d
1
(log a x)
log a e
dx
x
log10 e 0.4343

iteration

value of xn 1 xn

initial iteration x0
1

2
3

f ( xn )
f ' ( xn )

x0 log 10 x0 1.2
log 10 x0 0.4343

x1

2.746

x2

2.746

x1 log 10 x1 1.2
log 10 x1 0.4343

2.741

x3

2.741

x2 log 10 x2 1.2
log 10 x2 0.4343

2.741

Hence the real root of f(x)=0, correct to three decimal places is 2.741
-------------------------------------------------------------------------------------------2.
Evaluate 12 to four decimal places by Newtons Raphson Method
Solution:
2
Let x= 12 x =12
x2-12 =0

Let f(x)= x2-12 and f(x)=2x


Now f(3)=-ve and f(4) =+ve.Hence the root lies between 3 and 4. Here
f (3) f (4) the root is nearer to 3. Therefore the initial approximation is x0=3
f ( xn )
iteration
x
value of x
n 1

initial iteration x0
1
2
3
4

f ' ( xn )

3
x1
x2
x3
x4

f (3)
3.5
f ' (3)
f (3.5)
3.5
3.4642
f ' (3.5)
f (3.4642 )
3.4642
3.4641
f ' (3.4642 )
f (3.4641 )
3.4641
3.4641
f ' (3.4641 )

Hence the value of 12 is 3.4641


---------------------------------------------------------------------------------

Solving Simultaneous Equations with two variables using Newtons


Method
Let the simultaneous equations with two variables be f(x, y) =0 and g(x, y) =0
x1=x0+h and y1=y0+k
D1
D2
and k
D
D
fx 0 fy
where D
gx 0 gy

D1
D2

f0
g0
fx
gx

fy
gy
0
0

0
0

0
0

f0
g0

Problems
1. Find the solution of the equation 4x2+2xy+y2=30 and 2x2+3xy+y2=3
correct to 3 places of decimals, using Newtons Raphson method, given
that x0=-3 and y0=2.
Solution:
Let f(x, y) = 4x2+2xy+y2-30 and g(x, y) = 2x2+3xy+y2-3

fx=8x +2y, fy=2x+2y, gx=4x+3y, gy=3x+2y


x0

y0

fx

-3

-20

D
D1
D2
h
k

fx

fy

gx

gy

f0

fy

g0

gy

fx
gx
D1
D
D2
D

fy

-2

-6

- 20 - 2

-6

-5

-2 -2
1

-5

f0

- 20 - 2

g0

-6

12
0.1364
88
32
0.3636
88

x1=x0+h and y1=y0+k

gx

88

12
32

gy

-5

f0

g0

-2

x1= -3.1364 and y1=2.364

x1

y1

fx

-3.1364

2.364

-20.360

fx
gx

D1

f1
g1

D2

fx
gx

h
k

fy 1
gy 1

1
1

D1
D
D2
D

-1.544

gx

gy

-5.452

-4.680

gx

gy

f1

g1

0.0995

0.0169

0.4395

f1
g1

fy

86.8669

fy 1
gy 1
1

0.1984

0.0051
0.0023

x2= -3.131 and y2=2.362

x2

y2

fx

-3.131

2.362

-20.324

D
D1
D2
h
k

fx
gx

2
2

f2
g2
fx
gx
D1
D
D2
D

fy
gy
fy
gy

2
2

fy

-1.538

-5.438

-4.669

f2

g2

0.0008

-0.0009

86.5291

0.0051

f2
g2

0.0226

0.0001
0.0003

x3= -3.1309 and y3=2.3617.Since the two consecutive values of x2, x3 and y2, y3
are approximately equal, the correct solution can be taken as x= -3.1309 and
y=2.3617.
---------------------------------------------------------------------------------------------------------------------

Method of False Position (or) Regula Falsi Method (or) Method of


Chords
FORMULA: x
1

af (b) bf (a )
f (b) f (a )

This is the first approximation to the actual root.


Now if f(x1) and f(a) are of opposite signs, then the actual root lies between
x1 and a.
Replacing b by x1 and keeping a as it is we get the next approx. x2 to the
actual root.
Continuing this manner we get the real root.

Note:
The convergence of the root in this method is slower than Newtons
Raphson Method

Problems based on Regula Falsi Method


1. Sove the equation xtanx=-1 by Regula-Falsi method starting with x0=2.5 and
x1 = 3.0 correct to 3 decimal places.
Solution:
Let f(x) = xtanx+1
f(2.5)=-ve and f(3)=+ve
Let us take a=3 and b=2.5
af (b) bf (a)
iteration a
b
x
f (b)

f (a)

2.5 x 3 f (2.5) 2.5 f (3) 2.8012


1

2.8012 2.5 x
2

2.7984 2.5 x
3

f (2.5) f (3)
2.8012 f (2.5) 2.5 f (2.8012 )
f (2.5) f (2.8012 )
2.7984 f (2.5) 2.5 f (2.7984 )
f (2.5) f (2.7984 )

f(2.8012)=+ve
2.7984

f(2.7984)=+ve

2.7984

Since the two consecutive values of x2 and x3 are approximately equal, the
required root of the equation f(x)=0 is 2.7984
-------------------------------------------------------------------------------------------2. Find the root of xex = 3 by Regula-Falsi method correct to 3 decimal places.
Solution:
Let f(x)= xex 3
f(1)= +ve and f(1.5) = -ve
The root lies between 1 1nd 1.5
Take a = 1 and b = 1.5

iteration a
1

af (b) bf (a)
f (b) f (a)

1.5 x 1 f (1.5) 1.5 f (1) 1.035


1

1.035 1.5 x
2

1.045 1.5 x
3

1.048 1.5 x
4

f (1.5) f (1)
1.035 f (1.5) 1.5 f (1.035 )
1.045
f (1.5) f (1.035 )
1.045 f (1.5) 1.5 f (1.045 )
1.048
f (1.5) f (1.045 )
1.048 f (1.5) 1.5 f (1.048 )
1.048
f (1.5) f (1.048 )

f(1.035)=ve
f(1.045)=ve
f(1.045)=ve

Since the two consecutive values of x3 and x4 are equal, the required root of the
equation f(x)=0 is 1.048
--------------------------------------------------------------------------------------------

Solutions of linear algebraic equations


A system of m linear equations (or a set of m simultaneous linear equations)
in n unknowns x1 , x2 ,..., xn is a set of equations of the form,
a11 x1

a12 x2 ...... a1n xn

b1

a21 x1

a22 x2 ...... a2 n xn

b2

:::::::::: :::::::::: :::::::::: :::::::::: ::::


am1 x1

am 2 x2 ... amn xn

(1)

bm

Where the coefficients of x1 , x2 ,..., xn and b1 , b2 ,..., bm are constants.


The left hand side members of (1) may be specified by the square array of
the coefficients, known as the coefficient matrix.
a11 a12 .........a1n
A

a21 a22 .........a2 n


:::::::::: :::::::::: :
am1 am 2 .......amn

Whereas the complete set may be specified by the rectangular array


a11 a12 .........a1n b1
M

a21 a22 .........a2 n b2


:::::::::: :::::::::: :
am1 am 2 .......amn bm

is known as the augmented matrix.


There are two methods to solve such a system by numerical methods.
Direct methods
Iterative or indirect methods.

Gaussian elimination method, Gauss-Jordan method, belongs to Direct


methods,
Gauss-Seidel iterative method and relaxation method belongs to iterative
methods.

Back Substitution
Let A be a given square matrix of order n, b a given n-vector. We wish to
solve the linear system.
Ax b

For the unknown n-vector x. The solution vector x can be obtained without
difficulty in case A is upper-triangular with all diagonal entries are non-zero. In
that case the system has the form
a11 x1 a12 x2 ... a1,n 1 xn

a1n xn

a 22 x2 ... a 2,n 1 xn

b1

a2 n xn

b2

:::::::::: :::::::::: :::::::::: :::::::::: :::::


a n 2,n 2 xn 2 an 2,n 1 xn 1 bn
an

an

1, n 1 n 1

1, n

xn

an 2 xn

bn

ann xn

(1)
2

bn

In particular, the last equation involves only x n ; hence, since a nn 0 , we must have
bn
a nn

xn

Since we now know, the second last equation


an

an

1, n 1 n 1

1, n

xn

bn

Involves only one unknown, namely, x n 1 .


As a n 1,n 1 0 , it follows that
xn

bn

an

an

1,n

xn

1,n 1

With x n and x n 1 now determined, the third from last equation


an

2,n 2

xn

an

2,n 1 n 1

an

2 ,n

xn

bn

Contains only one true unknown, namely, x n 2 . Once again, wince a n


can solve for x n 2 .
xn

bn
2

an

2,n 1 n 1

an

bn

an

2,n

xn

2,n 2

and so on.

2,n 2

This process of determining the solution of (1) is called Back substitution.

0,

we

Gauss elimination method


Basically the most effective direct solution techniques, currently being used
are applications of Gauss elimination, method which Gauss proposed over a
century ago. In this method, the given system is transformed into an equivalent
system with upper-triangular coefficient matrix i.e., a matrix in which all elements
below the diagonal elements are zero which can be solved by back substitution.

Note
This method fails if the element in the top of the first column is zero.
Therefore in this case we can interchange the rows so as to get the pivot
element in the top of the first column.
If we are not interested in the elimination of x, y, z in a particular order, then
we can choose at each stage the numerically largest coefficient of the entire
coefficient matrix. This requires an interchange of equations and also an
interchange of the position of the variables.

Problems based on Gauss Elimination Method


1. Solve 2x+y+4z=12; 8x-3y+2z=20; 4x+11y-z=33 by gauss elimination method
Solution:
The given equations are,
2x+y+4z=12
8x-3y+2z=20
4x+11y-z=33

above equations can be written as


2

11
A

12

20

1 z

33

where, A-co efficient matrix; B-constants; X- unknown variables


The augmented matrix can be written as

2
A, B

8
3
4 11
2
~ 0

12

2 20
1 33

1
7

4 12
14 28

4 11

1 33

12

7 14
9 9

28
9

~ 0
0

2 1

12

~ 0 7

14

28

R2

4 R1 R2

R3

2 R1 R3

R3

9 R1 7 R3

0 0 189 189

solutions are obtained from above matrix by back substitution method as


2x+y +4z = 12
7y +14z = 28
189z = 189

(1)
(2)
(3)

from the above equations we get z=1, y=2, x=3


thus the solution of the equations are x=3 ; y= 2 ; z=1
------------------------------------------------------------------------------------------------------

2. Solve 3x+4y+5z=18; 2x-y+8z=13; 5x-2y+7z=20 by gauss elimination method


Solution:
The given equations are,
3x+4y+5z=18
2x-y+8z=13
5x-2y+7z=20
above equations can be written as
3

5 x

18

1 8 y

13

2 7 z

20

A X
B
where, A-co efficient matrix; B-constants; X- unknown variables
The augmented matrix can be written as

3
A, B

5 18

1 8 13

2 7 20

~ 0 11
0 26
3

18

14

30

R2

115

18

~ 0 11

14

408

408

R3

2 R1 3R2 , R3

5R1 3R3

26 R2 11R3

solutions are obtained from above matrix by back substitution method as


3x+4y +5z = 18
11y -14z = -3
-408z = -408

(1)
(2)
(3)

from the above equations we get z=1, y=1, x=3


thus the solution of the equations are x=3 ; y=1 ; z=1
------------------------------------------------------------------------------------------------------

Gauss Jordan Method


This method is a modified from Gaussian elimination method. In this
method, the coefficient matrix is reduced to a diagonal matrix (or even a unit
matrix) rather than a triangular matrix as in the Gaussian method. Here the
elimination of the unknowns is done not only in the equations below, but also in
the equations above the leading diagonal. Here we get the solution without using
the back substitution method since after completion of the Gauss Jordan method
the equations become
1
0
...
...
0

0
1
...
...
0

0
0
...
...
0

...
...
...
...
...

0
...
...
...
1

x1
x2
...
...
xn

a1
a2
...
...
an

Note:
This method involves more computation than in the Gaussian method.
In this method we can find the values of x1 , x2 ,..., xn immediately without
using back substitution.
Iteration method is self-correcting method, since the error made in any
computation is corrected in the subsequent iterations.

Problems based on Gauss Jordan Method


1. Solve 3x+4y+5z=18; 2x-y+8z=13; 5x-2y+7z=20 by gauss elimination method
Solution:
The given equations are,
3x+4y+5z=18
2x-y+8z=13
5x-2y+7z=20
above equations can be written as
3

5 x

18

1 8 y

13

2 7 z

20

A X
B
where, A-co efficient matrix; B-constants; X- unknown variables
The augmented matrix can be written as
3
A, B

2
5

5 18

1 8 13
2 7 20

18

~ 0 11
0 26

14 3
4
30

33 0
~ 0 11

111
14

210
3

408

408

13464
~
0

0
4488

0
0

R2

2 R1 3R2 , R3

R1

11R 1 4 R2 , R3

40392
4488

408

R1

5 R1 3R3

26 R2 11R3

408R 1 111R3 , R2

408R1 14 R

408

1 0 0 3
~ 0 1 0 1
0 0 1 1

R1

R1

13464

, R2

R2

4488

, R3

R3

408

without back substitution method we get z=1, y=1, x=3


Thus the solution of the equations are x=3 ; y=1 ; z=1
-------------------------------------------------------------------------------------------2. Solve 10x+y+z=12; 2x+10y+z=13; 2x+2y+10z=14 by gauss Jordan method
Solution:
The given equations are,

10x+y+z=12
2x+10y+z=13
2x+2y+10z=14
above equations can be written as
10

12

10

13

2 10 z
A
X

14
B

A-co efficient matrix


B-constants
X- unknown variables
The augmented matrix can be written as
A, B

10

12

10

13

10 14

10
~ 0
0
490
~ 0
0

1
49
9
0
49
0

1158850
~
0
0
1 0 0 1
~ 0 1 0 1
0 0 1 1

1
4
49
45
4

12
53
58

R2

535
53

R 1 - 5R 2 , R 3

R1

R 1 - 5R 3

49R 1 R 2 , R 3

9R 2 - 49R 3

2365 2365
0
0
1158850
115885
0
115885
0
2365
2365
R1

R1

,R
1158850 2

R1

R2

2365R 1 45R 3 , R 2

,R
- 115885 3

R3

2365R 2

2365

ut the back substitution x=1 ; y= 1 ; z=1


-------------------------------------------------------------------------------------------------------

Jacobis (or Gauss Jacobis) iteration method


Let the system of simultaneous equations be
a1 x b1 y c1 z

d1

a 2 x b2 y c 2 z

d2

a3 x b3 y c3 z

d3

(1)

4R 3

witho

This system of equations can also be written as


x
y
z

a1 d1

1
b1 y c1 z

b2 d 2

1
a2 x c2 z

c3 d 3

1
a3 x b3 y

(2)

Let the first approximation be x0 , y 0 and z 0 . Substituting x0 , y 0 and z 0 in (2) we get,


x1

1
d1 b1 y 0
a1

y1

1
d2
b2

a 2 x0

z1

1
d3
c3

a3 x0 b3 y 0

c1 z 0
c2 z 0

Substitution the values of x1 , y1 and z1 in (2) we get the second approximations


x 2 , y 2 and z 2 as given below
x2

1
d1 b1 y1 c1 z1
a1

y1

1
d2
b2

a 2 x1 c 2 z1

z1

1
d3
c3

a3 x1 b3 y1

Substituting the values of x2 , y 2 and z 2 in (2) we get the third approximations x3 , y 3


and z 3 .
This process may be repeated till the difference between two consecutive
approximations is negligible.

Problems based on Gauss Jacobi Method


1. Solve the following equations by Gauss Jacobis iteration method, 20x+y-2z=17,
3x+20y-z=-18, 2x-3y+20z=25.

Solution:
The given equations are
20x+y-2z=17,
3x+20y-z=-18,
2x-3y+20z=25.
The equations can be written as,

x
y
z

1
17 y 2 z
20
1
18 3 x z
20
1
25 2 x 3 y
20

Iterati
on

Initial

x0=0

1
17 y 2 z
20

x1

1
17
20

0.85

1
17 0.9 2.5
20
1.02
1
x3
17 0.965 2(1.3)
20
1.00125
1
17 1.0015 2(1.00325)
20
1.0004

18 3x z

y0=0

x2

1
20

x4

y1
y2

y3

y4

1
20

18

0.9

1
18 3(0.85) 1.25
20
0.965
1
18 3(1.02) 1.03
20
1.0015
1
18 3(1.00125) 1.00325
20
1.000025

1
25 2 x 3 y
20
z0=0
1
z1
25 1.25
20
z

1
25 1.7 2.7
20
1.03

z2

1
25 2(1.02) 3(0.965)
20
1.00325

z3

1
25 2(1.00125) 3( 1.0015
20
0.99965

z4

x3~x4; y3~y4 ; z3~z4

Therefore the solution is x=1; y= -1; z=1


----------------------------------------------------------------------------------------------------------

2. Solve the following equations by Gauss Jacobis method 9x+2y+4z=20;


x+10y+4z=6; 2x-4y+10z=-15
Solution:
The given system of equations is
9x+2y+4z=20;
x+10y+4z=6;
2x-4y+10z=-15
The equations can be written as
1
20 2 y 4 z
9
1
y
6 x 4z
10
1
z
15 2 x 4 y
10
x

Let the initial values be x0=y0=z0=0

Iterati
on

Initial

x0=0

x1

1
20 2 y 4 z
9

1
6 x 4z
10

y0=0
20
9

2.222

x2=2.7556
x3=2.762

1
10

15 2 x 4 y

z0=0

6
0.6
10
y2=0.9778

15
10

y1

z1

y3=1.0062

z2=-1.7044
z3=-1.66

1.5

x2~x3; y2~y3 ; z2~z3


Therefore the solution is x=2.8; y= 1 ; z=-1.7
----------------------------------------------------------------------------------------------------------

Gauss Seidal Iterative Method


Let the given system of equations be
a11 x1

a12 x2

a13 x3

...... a1n xn

C1

a 21 x1

a 22 x2

a 23 x3

...... a 2 n xn

C2

a31 x1

a32 x2

a33 x3

...... a3n xn

C3

..............................................................
..............................................................
a n1 x1 a n 2 x2 a n 3 x3 ...... a nn xn C n

Such system is often amenable to an iterative process in which the system is first
rewritten in the form
x1

x2
x3

1
C1 a12 x2 a13 x3 ...... a1n xn
a11
1
C 2 a 22 x2 a 23 x3 ...... a 2 n xn
a 22
1
C3 a32 x 2 a33 x3 ...... a3n x n
a33

(1)
(2)
(3)

.
xn

1
Cn
a nn

an 2 x2

a n 3 x3 ......

a nn xn

(4)

First let us assume that x2 x3 ..... xn 0 in (1) and find x1 . Let it be x1* . Putting x1*
for x1 and x3 x4 ..... xn 0 in (2) we get the value for x 2 and let it be x 2* . Putting
x1* for x1 and x 2* for x 2 and x3 x4 ..... xn 0 in (3) we get the value for x 3 and let
it be x 3* . In this way we can find the first approximate values for x1 , x2 ,..., xn .
Similarly we can find the better approximate value of x1 , x2 ,..., xn by using the
relation

1
a11
1
a 22
1
a33
1
a nn

x1*

x2*
x3*
xn*

C1 a12 x2

a13 x3 ......

a1n xn

C2

a 21 x1*

a 23 x3 ......

a2 n xn

C3

a31 x1*

a32 x3* ......

a3n x n

Cn

a n1 x1*

a n 2 x3* ......

a n ,n 1 xn*

Note:
This method is very useful with less work for the given systems of equation whose
augmented matrix have a large number of zero elements.
We say a matrix is diagonally dominant if the numerical value of the leading
diagonal element in each row is greater than or equal to the sum of the numerical
values of the other elements in that row.
For the Gauss Seidal method to coverage quickly, the coefficient matrix must be
diagonally dominant. If it is not so, we have to rearrange the equations in such a
way that the coefficient matrix is diagonally dominant and then only we can apply
Gauss Seidal method.

Problems based on Gauss Seidal Method


1. Solve x+y+54z=110, 27x+6y-z=85, 6x+15y+2z=72, by using Gauss Seidal

method.
Solution:
The system of equations is
x+y+54z=110,
27x+6y-z=85,
6x+15y+2z=72,
The co-efficient matrix is
1

54 1 1 54
27 6
1 6 27 1
6 15 2 2 6 15
27

1
6

1 27 6 1
1 54 1 1 54 Here R1
15 2 2 6 15

R2

27
6
1

1 27 6 1
15 2 15 6 2 Here R2
1 54 54 1 1

R3

Here the matrix is diagonally dominant


The diagonally dominant matrix is
27

15

54

Thus the matrix is diagonally dominant, now the system of equations is


27x+6y-z=85,
6x+15y+2z=72,
x+y+54z=110.
The equations can be written as
x= 1 (85-6y+z)
27

y= 1 (72-6x-2z)
z=

15
1
54

(110-x-y)

The initial values be x0=y0=z0=0.


Iteration x=
Initial
1

1 (85-6y+z)
27

x0=0
x1=85/27=3.148

y= 1 (72-6x-2z)
15

y0=0
y1= 1 (72-18.888)
15

=3.5408
2

x2= 1 (85-1.2448+1.9132) y2=


27

=2.432
3

=2.4257
4

x4= 1 (8527

21.4374+1.9259)
=2.4255
5

=3.572

x3= 1 (85-21.432+1.9258) y3=


27

1 (72-14.592-3.8264)
15

1 (72-14.5542-3.8516)
15

=3.5729
y4=

1 (72-14.553-3.8518)
15

=3.5730

1
x5= 1 (85-21.438+1.9259) y5= 15 (72-14.553-3.8518)

27

=2.4255

=3.5730

z= 1 (110-x-y)
54

z0=0
z1= 1 (110-3.14854

3.541)
=1.9132
z2= 1 (110-2.43254

3.572)
=1.9259
z3= 1 (110-2.425754

3.5729)
=1.9259
z4= 1 (110-2.425554

3.5730)
=1.9259
z5= 1 (110-2.425554

3.5730)
=1.9259

x4~x5 ; y4~y5 ; z4~z5, thus the solution is x=2.4255, y=3.5730, z=1.9259.


------------------------------------------------------------------------------------------------------------

2. Solve 8x-3y+2z=20; 6x+3y+12z=35; 4x+11y-z=33 by Gauss Seidal method


Solution:
The system of equations is
8x-3y+2z=20;
6x+3y+12z=35;
4x+11y-z=33
8

The co-efficient matrix is 6

11

2 8 3 2
12 3 6 12
1 1 4 11

Thus the matrix is not diagonally dominant


8
4
6

2 8 3 2
11
1 11 4 1
3 12 12 6 3

Now the matrix is diagonally dominant


The system of equations is
8x-3y+2z=20
4x+11y-z=33
6x+3y+12z=35
The equations can be written as
1
(20+3y-2z)
8
1
y= (33-4x+z)
11
1
z= (35-6x-3y)
12
Let the initial values be x0=y0=zo=0

x=

Iteration

x=

1
(20+3y-2z)
8

Initial
1

xo=0
x1=20/8=2.5

1
x2= (20+6-2.334)
8

y=

1
(33-4x+z)
11

yo=0
y1=

x3=

1
(3311

z1 =

1.8371)
=3.0260
4

x4=

x5=

12

6.2274)
=0.91875

1
(3311

z3 =

12.104+0.91857)
=1.9825

1
(20-5.94758

1.8154)
=3.0165
5

y3=

y4=

1.824)
=3.0166

y5=

1
(35-18.15612

5.9475)
=0.9077
z4= 1 (35-18.0996-

1
(3311

12

12.066+0.9077)
=1.9856

1
(20+5.95688

1
(35-1512

6)=11.439
z2= 1 (35-17.7498-

1
y2= (3311

11.833+1.607)
=2.0758

1
(20+6.22748

1
(35-6x-3y)
12

zo=0

10)=2.0909

=2.9583

z=

5.9568)
=0.9120
z5= 1 (35-18.0996-

1
(3311

12

12.0664+0.9120)
=1.9859

5.9577)
=0.9120

x4~x5 ; y4~y5 ; z4~z5,


Therefore the solution x=3.0165; y=1.9856; z=0.9120
--------------------------------------------------------------------------------------------------------

Inverse of a Matrix
Gauss Jordan Method
a1

b1

c1

Let A= a2 b2 c2 be the given matrix


a3

b3

c3

Step 1:
Write the augmented matrix A / I
Step 2:

a1

b1

c1

1 0 0

a2

b2

c2

0 1 0

a3

b3

c3

0 0 1

Use either row or column operations make the augmented matrix [A/I] as [I /A-1].
Here A-1 is the required inverse of the given matrix.

Problems based on Inverse of a Matrix


2 1 1

1. Point the inverse of a matrix 3 2 3 by using Gauss-Jordan Method.


1 4 9

Solution:
A/ I

2 1 1 1 0 0
3 2 3 0 1 0
1 4 9 0 0 1
2
~ 0
0
2

1
1
7

3 3
17 1

2
0

0 R2
2

~ 0
0

1
0

3 3
4 20

12

10

48

34

6 R1

4
0

1 0 0

3R1 2 R2 , R3

3
12
5

2.5
8.5

0.5
1.5

3.5

0.5

2.5
8.5
3.5

R1 2 R3

2 0 R1
14 2

20

~ 0 1 0 12
0 0 1
5
A

~ 0

14

R1

R2 , R3

2 R1

7 R2

R3 , R2

R3

R2

3R3

0.5
1.5 R1
0.5

R1
, R2
4

R2
, R3
4

R3
4

-------------------------------------------------------------------------------------------8

2. Point the inverse of a matrix

4
0

Solution:

4
8
4

0
4 by using Gauss-Jordan Method.
8

A/ I

8
4
0

4
8
4

0 1 0 0
4 0 1 0
8 0 0 1

8
4
~ 0 12
0
4

0 1 0 0
8 1 2 0 R2
8 0 0 1

24 0
~ 0 12

8 4 2 0
8 1 2 0 R1

16

1 2 3

48

9 6 3

~ 0
0

24 0 3 6 3 R1
0 16 1 2 3

3
16
1
~ 0 1 0
8
1
0 0 1
16
1 0 0

3
16
1
8
1
16

1
8
1
4
1
8

1 1
8 16
1 1
R1
4 8
1 3
8 16

R1 2 R2

3R1 R2 , R3

R2 3R3

2 R1 R3 , R2

2 R2

R1
, R2
48

R2
, R3
24

R3

R3
16

1
16
1
8
3
16

--------------------------------------------------------------------------------------------

Eigen values and Eigenvectors


Let A be any square matrix of order n. then for any scalar , we can form a
matrix A I where I is the nth order unit matrix. The determinant of this matrix
equated to zero is called the characteristic equation of A. i.e., the characteristic
equation of the matrix A is A 1 0 . Clearly this a polynomial of degree n in
having n roots for , say 1 , 2 ,..., n . These values are called eigenvalues of the
given matrix A.

For each of these eigenvalues, the system of equations A


x1
x2
. This solution X
...
xn

trivial solution for the vector X

I X

0 has a non-

x1
x2
is called a latent
...
xn

vector or eigen vector corresponding to the eigenvalue .


If A is of order n, then its characteristic equation is of n th degree. If n is large, it is
very difficult to find the exact roots of the characteristic equation and hence the
eigenvalues are difficult to find. But there are numerical methods available for
such cases. We list below two such methods called
Power method
Jacobis method
The second method can be applied only for symmetric matrices.
Power method
This method can be applied to find numerically the greatest eigenvalue of a square
matrix (also called the dominant eigenvalue). The method is explained below.
Let 1 , 2 ,..., n be the eigenvalues of A and let 1 be the dominant eigenvalue.
i.e., 1
2
3 .......
n
if the corresponding eigenvectors are x0 , x1 , x2 ,.... xn , then any arbitrary vector y can
be written as y a0 x0 a1 x1 ... an xn , since the eigenvectors are linearly
independent. Now
Ak y

a0

A k a0 x0
k
1

x0

... a n x n

a1 k2 x1 ... an

k
n n

x Ak X

k
k
1

a 0 x0

a1

x1 ......

But

1. i

2,...,n

. Hence A k y

k
1

a 0 x0

and A k 1 y

k 1
1
0

a x0 .

Hence, if k is large,

Ak 1 y
Ak y

where the division is carried out in the

corresponding components.
Here y is quite arbitrary. But generally we choose it as the vector having all its
components ones.

Note:
If

,...,

are the eigenvalues of A, then the eigenvalue

is dominant if

is called the

for i 2,3,...,n .

The eigenvector

x1
x2
...
xn

corresponding to the eigenvalue

dominant eigenvector.
If the eigenvalues of A are -3,1,2, then -3 is dominant.
If the eigen values of A are -4,1,4 then A has no dominant eigenvalue since
4.

The power method will work satisfactorily only if A has a dominant


eigenvalue.
1

Eigen vector may be 1 (or ) 0


1

(or )

0 for 3 x 3 matrix.
1

Problems based on Eigen value of a matrix


1. Using power method to find a dominant eigen value of a given matrix
2

1
0

2
1

0
1
2

Solution:
2

Let A=

1
0

1
2
1

1 and x0= 0
2
0

2
Ax0

1
0

1 0

1
2

12

1
Here

x1

12
0

2
Ax1

1
0

2.5

12
0

2.5

0. 5

0. 8
0. 2

1
Here

x2

0. 8
0. 2

2
Ax2

1
0

1
0.8
0.2

2. 8

2.8

2.8

1 .2

1
0.428

1
Here

x3

1
0.428

2
Ax3

1
0

1
1
0.428

3
3.428

0.875
3.428

1.856

1
0.541

0.875
Here

x4

1
0.541

Ax4

2
1

1
2

0
1

0.875
1

0.541

2.75
3.416
2.082

3.416

0.805
1
0.609

Here

0.805
1

x5

0.609
Ax5

0.805

2.61

1
0

2
1

1
2

1
0.609

3.414
2.3

0.764
3.414

1
0.65

0.764
Here

x6
2

Ax6

1
0

1
0.65
1

0.764
1

2.528
3.414

0.74
3.414

0.65

2. 3

0.674

0.726
1

0.74
Here

x7

1
0.674

Ax7

2
1

1
2

0
1

0.74
1

2.48
3.414

0.674

2.348

3.414

0.68

Here x7 =x8 approximately.


=3.414
Eigen value=3.414
0.72

Eigen vector=

1
0.68

--------------------------------------------------------------------------------------------

Steps for finding the smallest eigen value


First obtained the largest eigen value

of the given matrix.


Let B=A- 1I. Let be the largest eigen value of the matrix B then the
numerically smallest eigen value of A is + 1.
2. Find the largest eigen value and the corresponding eigen vector of the
1

matrix 0
0

4 2 and hence find the remaining eigen values.


0 7

Solution:
1

Let A= 0

4 2

Step 1

To find the largest Eigen value of A


0

Let x0= 0
1
1

A x0 = 0
0

0.43

0 = 2 =7 0.29
1
1
7

0.43

4 2

0.43

x1

0.29
1
1

A x1

0
0

4.01

0.29 = 0.84 =7 0.12


1
7
1

0.57

4 2

0.57

x2

0.12
1

A x2

0
0
0.54

x3

0.22
1

0.57

4 2
0

0.54

0.12 =7 0.22
1
1

A x3

0.54

0.22 =7 0.16
1
1

0.56

4 2

0.56

0.56
0.16

x4

1
1

A x4

0.16 =7 0.19
1
1

0.556

4 2

0.556

0.556

x5

0.19
1
1

A x5

0.19 =7 0.18
1
1

4 2

0.56

0.56

x6

0.18
1

x5

x6 approximately
Eigen value = =7
0.56

Eigen vector = 0.18


1

Step 2

To find the largest Eigen value of B


6

B=A- I= 0

11 2

Let y0= 0
1
6

B y0= 0
0

11 2
0

0 = 0 .
1
0

Here the we get eigen vector is zero . But the eigen vector should be non- zero. So
1

we consider th value of y0 be 0
0
6

B y0= 0

11 2

0 = 0
0
0

=-6 0
0

y1

0
0
1

Here y1

y0

0
0

The numerically largest eigen value of B = -6 =


Numerically smallest eigen value of A = + 1 = -6 + 7 =1= 2
1+ 2 + 3 = sum of the main diagonals of A = 1-4+7 =4
ie) 7 +1+ 3 = 4
3 =-4
The eigen values are 1, -4, 7.
--------------------------------------------------------------------------------------------

UNIT II
INTERPOLATION AND APPROXIMATION
Lagrangian Polynomials
Divided differences
Interpolating with a cubic spline
Newtons forward difference formula
Newtons backward difference formulA

LAGRANGIAN POLYNOMIALS
Formula
y ( x)

x x1 x x2 x x3 x x4 x x5
y0
x0 x1 x0 x2 x0 x3 x0 x4 x0 x5

x x0 x x2 x x3 x x4 x x5
y1
x1 x0 x1 x2 x1 x3 x1 x4 x1 x5

x x0 x x1 x x3 x x4 x x5
y2
x2 x0 x2 x1 x2 x3 x2 x4 x2 x5

x x0 x x1 x x2 x x4 x x5
y3
x3 x0 x3 x1 x3 x2 x3 x4 x3 x5

x x0 x x1 x x2 x x3 x x5
y4
x4 x0 x4 x1 x4 x2 x4 x3 x4 x5

x x0 x x1 x x2 x x3 x x4
y5
x5 x0 x5 x1 x5 x2 x5 x3 x5 x4

Problems based on Lagranges Method


1. Using Lagranges formula to calculate f(3) from the following table (A.U. N/D. 2007)
x
f(x)

0
1

1
14

2
15

4
5

5
6

6
19

x0=0

x1=1

x2=2

x3=4

x4=5

Solution:
x5=6

f(x)

y0=1

y1=14

y2=15

y3=5

y4=6

y5=19

We know that Lagranges formula is

y ( x)

y ( x)

x x1 x x2 x x3 x x4 x x5
y0
x0 x1 x0 x2 x0 x3 x0 x4 x0 x5

x x0 x x2 x x3 x x4 x x5
y1
x1 x0 x1 x2 x1 x3 x1 x4 x1 x5

x x0 x x1 x x3 x x4 x x5
y2
x2 x0 x2 x1 x2 x3 x2 x4 x2 x5

x x0 x x1 x x2 x x4 x x5
y3
x3 x0 x3 x1 x3 x2 x3 x4 x3 x5

x x0 x x1 x x2 x x3 x x5
y4
x4 x0 x4 x1 x4 x2 x4 x3 x4 x5

x x0 x x1 x x2 x x3 x x4
y5
x5 x0 x5 x1 x5 x2 x5 x3 x5 x4

x 1 x
0 1 0
x 0 x
2 0 2
x 0 x
5 0 5

2 x 4 x 5 x 6
x 0 x 2 x
(1)
2 0 4 0 5 0 6
1 0 1 2 1
1 x 4 x 5 x 6
x 0 x 1 x
(15)
1 2 4 2 5 2 6
4 0 4 1 4
1 x 2 x 4 x 6
x 0 x 1 x
(6)
1 5 2 5 4 5 6
6 0 6 1 6

4 x
4 1
2 x
2 4
2 x
2 6

5 x
5 1
5 x
5 4
4 x
4 6

6
6

(14)

6
(5)
6
5
(19)
5

x 1 x 2 x 4 x 5 x 6
x x 2 x 4 x 5 x 6
(14)
1 2 4 5 6
1 1 3 4 5
x x 1 x 4 x 5 x 6
x x 1 x 2 x 5 x 6
(15)
(5)
2 1 2 3 4
4 3 2 1 2
x x 1 x 2 x 4 x 6
x x 1 x 2 x 4 x 5
(6)
(19)
5 4 3 1 1
6 5 4 2 1
put x=3 we get
(3 1)(3 2)(3 4)(3 5)(3 6) 3(3 2)(3 4)(3 5)(3 6)
3(3 1)(3 4)(3 5)(3 6)
y (3)
(14)
(15)
240
60
48
3(3 1)(3 2)(3 5)(3 6)
3(3 1)(3 2)(3 4)(3 6)
3(3 1)(3 2)(3 4)(3 5)(3 6)
(5)
(6)
(19)
48
60
240
y ( x)

2(1)( 1)( 2)( 3) 3(1)( 1)( 2)( 3)


2(1)( 1)( 2)( 3)
(14)
(15)
240
60
48
3(2)(1)( 2)( 3)
3(2)(1)( 1)( 3)
3(2)(1)( 1)( 2)
(5)
(6)
(19)
48
60
240
12 252 540 180 108 228
240 60
48 48 60 240
12 1008 2700 900 432 228
240
2400
10
240
Answer: y(3) = 10.
-----------------------------------------------------------------------------------------------------------2. Using Lagranges formula fit a polynomial to the data (A.U. N/D. 2006)

Solution:
x
f(x)

y ( x)

x
f(x)

0
-12

1
0

3
6

4
12

x0=0

x1=1

x2=3

x3=4

y0= -12

y1=0

y2=6

y3=12

We know that Lagranges formula is


x x1 x x2 x x3 x x4 x x5
y0
x0 x1 x0 x2 x0 x3 x0 x4 x0 x5

x x0 x x2 x x3 x x4 x x5
y1
x1 x0 x1 x2 x1 x3 x1 x4 x1 x5

x x0 x x1 x x3 x x4 x x5
y2
x2 x0 x2 x1 x2 x3 x2 x4 x2 x5

x x0 x x1 x x2 x x4 x x5
y3
x3 x0 x3 x1 x3 x2 x3 x4 x3 x5

y ( x)

y ( x)
y ( x)

( x 1)(x 3)(x
(0 1)(0 3)(0
( x 0)(x 1)(x
(3 0)(3 1)(3
( x 1)(x 3)(x
( 1)( 3)( 4)
( x 1)(x 3)(x
( x 1)[x 2
( x 1)[x

y ( x)

4)
( x 0)(x 1)(x 3)
( 12)
(0)
4)
(1 0)(1 3)(1 4)
4)
( x 0)(x 1)(x 3)
(6)
(12)
4)
(4 0)(4 1)(4 3)
4)
( x)(x 1)(x 4)
( x)(x 1)(x 3)
( 12)
(6)
(12)
(3)(2)( 1)
(4)(3)(1)
4) x( x 1)(x 4) ( x)(x 1)(x 3)

3x 4 x 12 x 2

x2

4x

3 x]

6 x 12]

x3

6 x 2 12 x

x2

x3

7 x 2 18x 12

6 x 12
(1)

Substituting x=2 in (1), we get


y(2) = 23-7(22)+18(2)-12 = 8 28 +36 -12 = 44 -40 =4
Answer: y(2) = 4.
-----------------------------------------------------------------------------------------------------------3. Using Lagranges interpolation find the polynomial through (0, 0), (1, 1) and
(2, 2)
(A.U. M/J. 2007)
x
f(x)

0
0

1
1

2
2

Solution:

y ( x)
y ( x)
y ( x)

x0=0

x1=1

x2=2

f(x)

y0= 0

y1=1

y2=2

We know that Lagranges formula is


x x0 x x2
x x1 x x2
y0
y1
x0 x1 x0 x2
x1 x0 x1 x2

x 1 x 2
x 0 x 2
0
1
0 1 0 2
1 0 1 2
x x 2
x x 1
1
2
1 1
2 1
x( x 2) x( x 1)
x2

2x x2

x x0 x x1
y2
x2 x0 x2 x1

x 0 x 1
2
2 0 2 1

y ( x) x
The required polynomial is y x
------------------------------------------------------------------------------------------------------------

4. The following table gives certain corresponding values of x and log x. Compute the value of
10

log 323.5, by using Lagranges formula


10

x
f(x)

321.0
322.8
324.2
325.0
2.50651 2.50893 2.51081 2.51188

Solution:

y ( x)

x0=321.0

x1=322.8

x2=324.2

x3=325.0

f(x)

y0= 2.50651

y1= 2.50893

y2= 2.51081

y3= 2.51188

x x1 x x2 x x3 x x4 x x5
y0
x0 x1 x0 x2 x0 x3 x0 x4 x0 x5

x x0 x x2 x x3 x x4 x x5
y1
x1 x0 x1 x2 x1 x3 x1 x4 x1 x5

x x0 x x1 x x3 x x4 x x5
y2
x2 x0 x2 x1 x2 x3 x2 x4 x2 x5

x x0 x x1 x x2 x x4 x x5
y3
x3 x0 x3 x1 x3 x2 x3 x4 x3 x5

Substituting these values in Lagranges interpolation formula, we get,


(323.5 322.8)(323.5 324.2)(323.5 325)
f (323.5)
(2.5061)
(321 322.8)(321 324.2)(321 325)
(323.5 321)(323.5 324.2)(323.5 325)
(2.50893)
(322.8 321)(322.8 324.2)(322.8 325)
(323.5 321)(323.5 322.8)(323.5 325)
(2.51081)
(324.2 321)(324.2 322.8)(324.2 325)
(323.5 321)(323.5 322.8)(323.5 324.2)
(2.51188)
(325 321)(325 322.8)(325 324.2)
0.07996 1.18794 1.83897 0.43708
f (323.5) 2.50987
------------------------------------------------------------------------------------------------------------

Inverse Interpolation
The process of finding a value of x for the corresponding value of y is called inverse
interpolation.

y1 y y2 ... y yn
x0
y1 y0 y2 ... y0 yn

y0

y y0 y y2 ... y yn
x1
y1 y0 y1 y2 ... y1 yn
y y0 y
yn 0 yn

...

y1 ... y yn 1
xn
y1 ... yn yn 1

Problems based on Inverse Interpolation


1. Find the value of x when y=85, using Lagranges formula from the following table.
x
y

2
5
8
14
94.8 87.9 81.3 68.7
Solution:

x
y
x

x0=2
y0=94.8

x1=5
y1=87.9

y1 y y2 y y3
x0
y1 y0 y2 y0 y3

y0

x2=8
y2=81.3

x3=14
y3=68.7

y y0 y y2 y y3
x1
y1 y0 y1 y2 y1 y3

y0 y y1 y y3
y y0 y x1 y y2
x2
x3
y2 y0 x2 y1 y2 y3
y3 y0 y3 y1 y3 y2
Substituting the above values, we get,
(85 87.9)(85 81.3)(85 68.7)
(85 94.8)(85 81.3)(85 68.7)
(2)
(5)
(94.8 87.9)(94.8 81.3)(94.8 68.7)
(87.9 94.8)(87.9 81.3)(87.9 68.7)
(85 94.8)(85 87.9)(85 68.7)
(85 94.8)(85 87.9)(85 81.3)
(8)
(14)
(81.3 94.8)(81.3 87.9)(81.3 68.7)
(68.7 94.8)(68.7 87.9)(68.7 81.3)
x = 0.1438778+3.3798011+3.3010599-0.2331532 = 6.3038
Therefore the value of x when y = 6.3038
--------------------------------------------------------------------------------------------------x

2. The following table gives the value of the elliptic integral

y( )
0

certain values of . Find

y( )
Solution:

21
0.3706

if y( ) = 0.3887
23
0.4068

25
0.4433

1
sin 2
2

for

y1 y y 2 y y3
y1 y0 y 2 y0 y3

y0

y
y2

y0 y y1 y y3
y0 x2 y1 y 2 y3

y( )

y
0

y 0 y y 2 y y3
y0 y1 y 2 y1 y3

y1

y
2

y3

y0 y x1 y y 2
y0 y3 y1 y3 y 2

1=21

2=23

3=25

y1=0.3706

y2=0.4068

y3=0.4433

we have
(0.3887
(0.3706
(0.3887
(0.3706

0.4068)(0.3887
0.4068)(0.3706
0.3706)(0.3887
0.3706)(0.3706

0.4433)
(0.3887 0.3706)(0.3887 0.4433)
(21)
(23)
0.4433)
(0.4068 0.3706)(0.4068 0.4433)
0.4068)
(25)
0.4068)

21.999

Therefore the value of such that y( ) = 0.3887 is =21.999

DIVIDED DIFFERENCES
Problems based on Newtons Divided Difference Formula
2.2.1 Let the function y f x take the values f x0 , f x1 , f x2 ,... f xn corresponding to
the values x0 , x1, x2 ,... xn for the argument x where x1 x0 , x2 x1 ,..., xn xn 1 need not be
necessarily equal.
The first divided of f x for the argument x0 , x1 is defined as
f x1
f x0
x1 x0
f x2
f x1
Similarly f x1 , x2
x2 x1
f x3
f x2
f x2 , x3
and so on.
x3 x2

...(i)

f x0 , x1

Thus, for defining a first divided difference, we need the functional values
corresponding to two arguments.
The second divided difference of f(x) for three arguments
x0, x1, x2 is defined as
f x1 , x2
x2

f x0 , x1
x0
f x2 , x3
Similarly f x1 , x2 , x3
x3

... (ii)
f x1 , x2
x1

The third divided difference of f(x) for the four arguments


x0, x1, x2,x3 is defined as
f x0 , x1 , x2 , x3

f x1 , x2 , x3
x3

f x0 , x1 , x2
x0

... (iii)

The quantities in (i), (ii) and (iii) are called divided differences of orders 1, 2, 3
respectively.

1. Using Newtons Divided Difference formula, find the value of f(8) and f(5) given
the following data.
x
f(x)

4
48

5
100

7
294

10
900

11
1210

13
2028

Solution:
x
f(x)

x0=4
f(x0)=48

x1=5
f(x1)=100

x2=7
f(x2)=294

x3=10
f(x3)=900

x4=11
f(x4)=1210

x5=13
f(x5)=2028

The divided difference table for the given data is given below.
Newtons divided difference formula,
f(x) = f(x0) + (x-x0)f(x0,x1) + (x-x0) (x-x1) f(x0,x1,x2) + (x-x0) (x-x1) (x-x2) f(x0,x1,x2,x3)
+ ..
x

f(x)

48
100 48
5 4

2f(x)

f(x)

97 52
15
7 4

100

292 97
10 5

2028 1210
13 11

13

0
100 48
5 4

409 310
13 10

52

27

310

1210

0
100 48
5 4

310 202
11 7

52

21

202

900
1210 900
11 10

11

100 48
5 4

97

294
900 294
10 7

10

4f(x)

52

294 100
7 5

3f(x)

52

33

409

2028

Using divided differences and the given data in (1),


f(x) = 48 + 52(x-4) + 15(x-4)(x-5) +(x-4)(x-5)(x-7)
when x=8,
f(8) = 48 + 208 +180 +12 =448
Therefore f(8) =448
when x=15,
f(15) = 48 + 572 +1650 +880 =3150
Therefore f(15) =3150
-----------------------------------------------------------------------------------------------------------2. Use Newtons divided difference formula, to fit a polynomial to the data

and find y when x=1


x
y
Solution:
x
y

-1
8

0
3

x0=-1
y0=8

2
1

3
12

x1=0
y1=3

x2=2
y2=1

x3=3
y3=12

The divided difference table for the given data as follows.

-1

-8

2y

3f(x)

3 8
11
0 1

1 11
2 1

3
1 3
2 0

4
4 4
3 1

1
11 1
2 1

12 1
11
3 2

12

By Newtons divided difference formula,

f(x) = f(x0) + (x-x0)f(x0,x1) + (x-x0) (x-x1) f(x0,x1,x2) + (x-x0) (x-x1) (x-x2) f(x0,x1,x2,x3)
+ ..
Using these we get,
f(x) = -8+(x+1)11+(x+1)x(-4)+(x+1)x(x-2)2
= -8+11x+11-4x2-4x+2x3-2x2-4x
= 2x3-6x2+3x+3
y = 2x3-6x2+3x+3
y(1) = 2-6+3+3 =2
Answer: y(1)=2
------------------------------------------------------------------------------------------------------------

Interpolating with a Cubic Spline


Definition: Cubic Spline
A cubic polynomial approximating the curve in every subinterval is called is called
satisfying the following properties.
1. F(xi)=fi for i = 0, 1, 2, , n.
2. On each interval [xi-1, xi], 1 i n, F(x) is a third degree polynomial.
3. F(x), F(x) and F(x) are continuous on the interval [x0, xn].
The second derivatives at the end points of the given range are denoted as M0 and Mn
respectively.
Natural cubic spline
A cubic spline F(x) with end conditions M0=0 and Mn=0 where f(xi)=Mi in the interval
[x0, xn]
i.e., f(x0) = 0 and f(xn) = 0 is called a natural cubic spline.
Fitting a natural cubic spline for the given data:
In case of natural cubic spline the derivatives f(x0) = 0 and f(xn) = 0. i.e., M0=0 and
Mn=0
Suppose that the values of x are equally spaced with a spacing h.
The cubic spline approximation in the subinterval (xi-1, xi) is given by
( x xi 1 )3
xi x
x xi 1
1 ( xi x)3
h2
h2
Fi ( x)
Mi 1
Mi
yi 1
Mi 1
yi
Mi
h
6
6
h
6
h
6
6
[yi-1+yi+1-2yi] where i = 1, 2, .., n- 1 with M0=0 and Mn=0.
h2
Problems based on Cubic spline
1. Obtain the cubic spline approximation for the function y = f(x) from the following data,
given that y0=y3=0.

Mi-1+4Mi+Mi+1 =

x
y

-1
-1

0
1

1
3

2
35

Solution:
x
x0=-1 x1=0
x2=1
x3=2
y
y0=-1 y1=1
y2=3
y3=35
The values of x are equally spaced with h =1.
Therefore we have
Mi-1+4Mi+Mi+1 =6 [yi-1+yi+1-2yi] where i = 1, 2, .., n- 1
Further M0=0 and M3=0
M0+4M1+M2 =6 (y0+y2-2y1)
4M1+M2 =0 ----------------------------------------------------------------------------(1)
and

M1+4M2+M3 = 6 (y1+y3-2y2)
= 6(1-6+35) = 180
M1+4M2 = 180 -----------------------------------------------------------------------(2)
Solving (1) and (2), M1=-12 M2 = 48
The cubic spline in (xi-1, xi) is given by
y

x) 3

( xi
6

Mi

( x xi 1 ) 3
Mi
6

xi

x
1

yi

1
Mi
6

x xi
1

yi

1
Mi
6

---------------(3)

where i = 1, 2, .., n- 1.
In the interval -1 x 0, i.e., x0 x x1(i=1) the cubic spline is given by
x) 3

( x1

M0

( x x0 ) 3
M1
6

x1

x
1

y0

1
M0
6

x x0
1

y1

1
M1
6

1
[(x 1)3 ( 12)] ( x)( 1) ( x 1)(1 2)
6
= (-2)(x3+3x2+3x+1)+x+3x+3
y = -2x3-6x2-2x+1
In the interval 0 x1, i.e., x1 x x2(i=2) the cubic spline is given by
y

x) 3

( x2

M1

( x x1 ) 3
M2
6

x2

x
1

y1

1
M1
6

x x1
1

y2

1
M2
6

1
[(1 x)3 ( 12) x 3 (48)] (1 x)(1 2) ( x 0)(3 4)
6
= (-2)(1-x)3+8x3+3-3x-x
y = 2x3-6x2+6x-2+8x3+3-4x
y = 10x3-6x2+2x+1
In the interval 1 x2, i.e., x2 x x3(i=3) the cubic spline is given by
y

x) 3

( x3

M2

( x x2 ) 3
M3
6

x3

x
1

y2

1
M2
6

x x2
1

y3

1
M3
6

1
[(2 x)3 (48)] (2 x)(3 8) ( x 1)(35)
6
= 8(8-12x+6x2-x3)+5x-10+35x-35
y = -8x3+48x2-56x+19
y = -8x3+48x2-56x+19
Hence the required cubic spline approximation for the given function is
2x 3 6x 2 2x 1
for
1 x 0
y

10x3 6x 2 2x 1
8x
2.

48x

for

56x 19 for

0 x 1
1 x

Obtain the natural cubic spline which agrees with y(x) at the set of data points given
below:
x
y

Hence find y(2.5)

2
11

3
49

4
123

Solution:
x
y

2
11

3
49

4
123

The values of x are equally spaced with h =1.


Therefore we have
Mi-1+4Mi+Mi+1 =6 [yi-1+yi+1-2yi] where i = 1, 2, .., n- 1
Further M0=0 and M2=0
M0+4M1+M2 =6 (y0+y2-2y1)
4M1=6(11-98+123)
M1= 54
The cubic spline in (xi-1, xi) is given by
y

x) 3

( xi
6

Mi

( x xi 1 ) 3
Mi
6

xi

x
1

yi

1
Mi
6

x xi
1

yi

1
Mi
6

---------------(3)

where i = 1, 2, .., n- 1.
In the interval 2 x 3, i.e., x0 x x1(i=1) the cubic spline is given by
y

x) 3

( x1
6

M0

( x x0 ) 3
M1
6

x1

x
1

y0

1
M0
6

x x0
1

y1

1
M1
6

1
[(x 2)3 (54)] (3 x)(11) ( x 2)(49 9)
6
= 9(x3-6x2+12x-8)+33-11x+40x-80
y = 9x3-54x2+137x-119
In the interval 3 x4, i.e., x1 x x2(i=2) the cubic spline is given by
y

x) 3

( x2
6

M1

( x x1 ) 3
M2
6

x2

x
1

y1

1
M1
6

x x1
1

y2

1
M2
6

1
[(4 x)3 (54)] (4 x)(40) ( x 3)(123)
6
= 9(64-48x+12x2-x3)+160-40x+123x-369
y = -9x3+108x2+349x+367
Hence the required cubic spline approximation for the given function is
9x 3 54x 2 137x - 119
for 2 x 3
y
3
2
- 9x 108x 349x 367
for 3 x 4
-----------------------------------------------------------------------------------------------------------Newtons Forward and Backward Difference Formulas
y

Introduction:
If a function y=f(x) is not known explicitly the value of y can obtained when a set of
values of (xi, yi) i = 1, 2, 3, .n are known by using the methods based on the principles of finite
differences, provided the function y=f(x) is continuous. Here the values of x being equally
spaced, i.e., xn = x0 +nh, n = 0, 1, 2, ., n
Forward Differences

If y0, y1, y2,, yn denote the set of values of y, then the first forward differences of y =
f(x) are defined by
y0 = y1-y0; y1=y2-y1; .; yn-1=yn-yn-1
where is called the forward difference operator.
Forward Difference table

x0

y0

y0
x1

2y0

y1

3y0

y1
x2

2y1

y2

4y0

x3

2y2

y3

4y1
3y2

y3
x4

2y3

y4

y4
x5

y5

Formula
y( x0

nh)

y0

n y0

n(n 1)
2!

5y0

3y1

y2

y0

n(n 1)(n 2)
3!

y0 .......

Problems based on Newtons forward interpolation formula


1. Using Newtons Forward interpolation formula, find f(1.5) from the following data
x
0
1
2
3
4
f(x)
858.3
869.6
880.9
829.3
903.6
Solution:
x
f(x)

x0=0
858.3

x1=1
869.6

x2=2
880.9

x4=3
829.3

x5=4
903.6

Difference Table

To
find
y for
x =
1.5

2 y

858.3

3y

869.6-858.3=11.3
1
n

869.6

2 h

x0

11.3-11.3=0
880.9-869.6=11.3

1.5 1.0
n
880.9
1

0.1-0=0.1

0.5

11.4-11.3=0.1

892.3-880.9=11.4

nh)

y0

n y0

n( n

1)

2!

y (1.5)

869.6

(0.5)(11.3)

y (1.5)

869.6

(0.5)(11.3)

y (1.5)

869.6

5.65

y (1.5)

869.6

5.65

y (1.5)

875.7375

y0

(0.5)(0.5

n( n

1)(n

3!
1)(0.5 2)

2
(0.5)(0.5)(1.5)

(0.0375)
2
0.01875

(0.1)
2
(0.28125)

2)

(0)

-0.2-0.1=-0.3
-0.1-0.1=-0.2

By
3
892.3
Newt
903.6-892.3=11.3
ons
4
903.6
forw
ard interpolation formula,
y( x 0

4y

11.3-11.4=-0.1

y0

.......

(0.5)(0.5

1)(0.5

2
(0.5)(0.5)(1.5)(2.5)
6

2)

(0.1)

(0.5)(0.5

1)(0.5
6

2)(0.5

3)

( 0.3)

(0.3)

6
0.46875

-----------------------------------------------------------------------------------------------------------2.
Using Newtons forward interpolation, find the value of log 10 ,given
log3.141 =
0.4970679364
log3.142 = 0.4972061807
log3.143 = 0.4973443810
log3.144 = 0.49748253704
log3.145 = 0.4974825374
Solution

y=logx

2 y

3y

4y

0.4970679364
0.1382443x10-3

0.4972061807

11.3-11.3=0
0.1382003x10

-3

0.1381564x10

-3

0.1381124x10

-3

0.1-0=0.1

0.4973443810

11.4-11.3=0.1
-0.1-0.1=-0.2

0.49748253704

-0.2-0.1=-0.3

11.3-11.4=-0.1

0.4974825374

Here x0=3.141, h = 0.001, y0 = 0.4970679364


The Newtons forward interpolation formula is
x0+nh==3.1415926536
n=

y ( x0

nh)

3.1415926536 - 3.141
=0. 5926536
0.001
y0

n(n 1)
2!

n y0

y0

n(n 1)(n 2)
3!

y0

.......

y( )

0.5926536( 0.4073464)( 0.440 10 7 )


2
0.4970679364 0.0000819310 0.0000000053

y( )

0.4971498727

y( )

0.4970679364 (0.5926536)(0.000138244)

---------------------------------------------------------------------------------------------------------3. From the following data, estimate the no. of persons earning weekly wages between 60
and 70 rupees.
Wages(in Rs.)
Below 40 40-60 60-80 80-100 100-120
No.of person(in thousands)
250
120
100
70
50
Solution
x

x0=40

x1=60

x2=80

x4=100

x5=120

250

250+120=370

370+100=470

470+70=540

540+50=590

Here x0=40, h = 20, y0 = 250


The Newtons forward interpolation formula is
x0+nh=70
n=

70 - 40
=1.5
20

wages
Frequency(y)
x
Below

250

2y

3y

4y

40
n(n 1) 2
n(n 1)(n 120
2) 3
y0
y 0 .......
2!
3!
Below
(1.5)(1.5 1)
(1.5)(1.5 1)(1.5 2)
370
-20 (1.5)(1.5 1)(1.5 2)(1.5 3) (20)
y (70) 60
250 (1.5)(120)
( 20)
( 10)
2
6
24
100
-10
y (70) 250 180 7.5 0.625 0.46875

y ( x0

nh)

y0

n y0

y (70)Below
423.5937

470
-30
20
80
No
70
10
. of
per Below
540
-20
so
100
ns
50
wh Below
590
ose
120
we
ekly wages below 70 =423.5937
No. of persons whose weekly wages below 60 = 370
No. of persons whose weekly wages No. of persons whose weekly wages below 70between 60 and 70
= No. of persons whose weekly wages below 60
= 423.5937-370
---------------------------------------------------------------------------------------------------------

Newtons Backward Interpolation formula


The formula is used mainly to interpolate the values of y near the end of a set of values of
y a short distance ahead (to the right) of y.
Formula:
n(n 1) 2
n(n 1)(n 2) 3
y( xn nh) yn n yn
yn
yn .......
2!
3!
Problems based on Newtons backward difference formula
1. The following data are taken from the steam table:
Temp:C
140
150
160
170
180
2
Pressure kgf/cm 3.685
4.854
6.302
8.076
10.225

Find the pressure at temperature t = 142C and t = 175C


Solution: We form the difference table:

t t t0

u140
h

150
P4 142
160
170
180

142 p 140
3.685
10

1
5

0.2

4.8541
P4 140
10
5
6.302
uu 1
P0 u P0
2
8.076
10.225
3.685
0.2 1.169
0.2

2p

3p

4p

1.169
0.279
1.448

0.047
0.326

P01.774
...

0.002
0.049

0.375

2.149
0.2 0.8
0.279
2

0.8 1.8
0.2 0.8 1.8 2.8
0.047
6
24
=3.685+0.2338-0.02332+0.002256-0.0000672
=3.897668
=3.898
1
175 180
P4 t 175o P4 180
10 , where v
0.5
2
10
1 2
Pn
Pn
Pn ...
2
0.5 0.5
10.225
0.5 2.149
0.375
2
0.5 0.5 1.5
0.5 0.5 1.5 2.5
+
0.049
0.002
6
24
=10.225-1.0745-0.0046875-0.0030625-0.000078125
=9.10048438=9.100

0.002

UNIT III
NUMERICAL DIFFERENTIATION AND
INTEGRATION
Differentiation using interpolation formulae
Numerical integration by trapezoidal rule
Simpsons 1/3 and 3/8 rules
Rombergs method
Two and Three point Gaussian quadrature formulas
Double integrals using trapezoidal and simpsonss rules.

Numerical Differentiation
Differentiation using Forward Interpolation formula(for equal interval)
Differentiation using Backward Interpolation formula(for equal interval)
Differentiation using Stirlings(Central Difference) Formula(for equal interval)
Maximum and Minimum
Differentiation using divided difference(for unequal interval)
Forward difference formula to compute the derivative
Newtons forward interpolation formula is
rr 1 2
rr 1 r 2 3
rr 1 r 2 r 3 4
f x0 rh y0
y0
y0
y0
y0 ...
2!
3!
4!
(Here using of r for n is only for convenience)
Differentiating w.r.t. we get,
2r 1 2
3r 2 6r 2 3
2r 3 9r 2 11r 3 4
hf ' x0 rh
y0
y0
y0
y0 ...
2
6
12
(1)
6r 2 18 r 11 4
2
h 2 f " x0 rh
y0 r 1 3 y0
y0 ...
12
... (2)
2r 3 4
h3 f " ' x0 rh
3 y0
y0 ...
2
...(3)
Similarly we can find the remaining derivatives.
If we want to find the derivatives at a point x x0 , then x0 rh xo
i.e., r=0.
Hence on substituting this value of r=0 in the above formula (1), (2) and (3), we get
1
1 2
1 3
1 4
f ' x0
y0
y0
y0
y0 ...
h
2
3
4
1 2
11 4
3
f ' ' x0
y0
y0
y0 ...
2
h
12
1
3 4
f ' ' ' x0
3 y0
y0 ... and so on.
3
h
2
Note: If the x value is nearer to the starting of the given table we use Forward
Interpolation formula
Backward difference formula to compute the derivatives
Newtons backward difference formula is
rr 1 2
rr 1 r 2 3
rr 1 r 2 r 3 4
f xn rh yn
yn
yn
yn
yn ...
2!
3!
4!
1
2r 1 2
2r 2 6r 2 3
2r 3 6r 2 11r 3 4
'
f xn rh
yn
yn
yn
yn ...
h
2
6
12

---

h 2 f ' ' xn

rh

yn

r 1

yn

6r 2 18 r 11
12

yn

...

2r 3 4
yn ...
2
Similarly we can find the remaining derivatives.
x xn , i.e., xn rh xn , we have r=0.
At the point
h3 f ' ' ' xn

rh

yn

1
1 2
1 3
1 4
yn
yn
yn
yn ...
h
2
3
4
1
11 4
2
3
f ' ' xn
yn
yn
yn ...
2
h
12
1
3 4
3
f ' ' ' xn
yn
yn ... and so on.
3
h
2
Note: If the x value is nearer to the end of the given table we use backward Interpolation formula
Central difference formula for computing the derivatives
We know that Stirlings central difference formula is
y0
y 1
r2 2
r r2 1 3y 1 3y 2
f x0 rh y0 r
y 1
2
2!
3!
2
f ' xn

r2 r2 1 4
r r2 1 r2 4 5
y 2
y2
4!
5!
(Here using of r for n is only for convenience)
y0
y 1 r2 2
f x0 rh y0 r
y 1
2
2!

i.e.,

r3

y 1

3!

r4

y 2

r2

4!

...

y 2

r5

5r 3 4r 5
5
y2
y 3 ...
5!
Differentiating (1) w.r.t. r we get,
y0
y 1
3r 2 1
2
hy' x0 rh
y 1
2
12

2r 3 r
12
y ' x0

rh

1
h

y0

5r 4 15r
5!

y 1
2

2r 3 r
12

...(2)
Differentiating (2) w.r.t. r, we get,

y 1

y 1
5

y 2

3r 2 1
12

5r 4 15r 2
5!

y 2
...

y 3

y1
5
2

...

y ' ' x0

rh

1
h2

r
2

3
1

6r 2 1
12

-----------(3)
20r 3 30r 5
5
y2
y 3 ...
5!
Similarly we can find the remaining derivatives.
If we want to find the derivative at a point x x0 , then x0 rh x0
i.e., r=0
Substituting r=0 in (2) and (3) we get,
1
y0
y 1
1 3
1 5
3
5
y ' x0
y1
y2
y2
y 3 ...
h
2
12
30
1 2
1 4
y1
y 2 ...
h2
12
Similarly we can find the remaining derivatives.
Note:
If the x value is middle of the given table we use central Difference formula
Maximum and Minimum:
Steps
dy
Write the Newtons Forward difference formula y(x),
dx
Write the forward difference table
dy
Find
dx
dy
Put
=0 and find the value of x
dx
y ' ' x0

d2y
Find
dx 2

For every value of x find


If

d2y
dx 2

d2y
<0 , y is maximum at that x(maximum point)
dx 2

d2y
>0 , y is minimum at that x(minimum point)
dx 2
To find the maximum and minimum value substitute the maximum and minimum
points in y(x) formula respectively.

If

Problems based on Differentiation using Interpolation formula


1. Find the first, second and third derivatives of the function tabulated below at the point
x=1.5
x

1.5

2.0

2.5

3.0

3.5

4.0

f(x)

3.375

7.0

13.625

24.0

38.875

59.0

Solution
The difference table is as follows:
x

y=f(x)

1.5x0

3.375y0
3.625

2.0

3.0

7.0

0.75

13.625

y0

3.75

24.0

0
0.75

4.50
14.875

3.5

y0

10.375
3.0

y0

6.625
2.5

38.875

0
0.75

5.25
20.125

4.0

59.0

Here we have to find the derivative at the point x=1.5 which is the initial value of the table.
Therefore by Newtons forward difference formula for derivatives at x=x0, we have
1
1 2
1 3
f ' x0
y0
y0
y0 ...
h
2
3
Here x0=1.5, h=0.5
1
1
1
f ' 1.5
3.625
3.0
0.75 ...
0.5
2
3
f ' 1.5 4.75

At the point x

x0 ,

1
h2
Here x0=1.5, h=0.5

f ' ' x0

1
0.5

f ' ' 1.5

f ' ' 1.5 9.0


At the point x
1
h3

f ' ' ' x0

y0

11
12

y0

y0 ...

3.0 0.75

x0 ,
3

y0

3
2

y0

1
0.5

f ' ' ' 1.5

0.75

6.0

f ' ' ' 1.5 6.0


------------------------------------------------------------------------------------------------------------2.
Compute f(0) and f(4) from the data
x
0
1
2
3
4
y
1
2.718
7.381
20.086
54.598
Solution:
2
3
4
y
y
y
y
x y=f(x)
0x0

1y0
1.718

y0
2.945

2.718

y0
5.097

4.663
2

8.668
13.765

21.807

20.086
34.512

y0

8.042

7.381
12.705

yn

y0
4

yn

yn

yn

54.598

xn yn

Here we have to find f(0) .ie.x=0 which is the starting of the given table. So we use the
forward interpolation formula.

1 2
1 3
1 4
y0
y0
y0 ...
2
3
4
1
1
1
1
f' 0
1.718
2.945
5.097
8.668
0.2225
1
2
3
4
Here we have to find f(4) .ie.x=4 which is the end of the given table. So we use the
backward interpolation formula.
1
11 4
2
3
f ' ' xn
yn
yn
yn ...
2
h
12
1
11
f '' 4
21 .807 13 .765
8.668
43 .5177
2
1
12
-----------------------------------------------------------------------------------------------------------f ' x0

1
h

y0

dy
d2y
and 2 at x = 51 from the following data.
dx
dx
x
50
60
70
80
90
y
19.96
36.65
58.81
77.21
94.61
Solution:
Here h=10. To find the derivatives of y at x=51 we use Forward difference formula
taking the origin at x0=50.
x x0 51 50
0.1
We have r
h
10
at x=51, r=0.1

2. Find

dy
dx

x 51

dy
dx

1
2r 1 2 3r 2 6r 2 3 2r 3 9r 2 11r 3 4
y0
y0
y0
y0 ...
=
2
6
12
0 .1 h

The difference table is given by


x

y=f(x)

50x

19.96y

16.69
60

y0

36.65

5.47
2

y0

22.1
6

70

-9.23
3

y0

-3.76

58.81

11.09
4

18.4
0
80

77.21

2.76

-1.00
17.4
0

y0

90

dy
dx

94.61
dy
dx

x 51

=1

16.69

10

r 0 .1

=1
10

d2y
dx 2
d2y
dx 2

x 51

d2y
dx 2

x 51

d2y
dx 2

3(0.1) 2

0.2 1
(5.47)
2

16.69 2.188 2.1998 1.9863

1
h2

r 0.1

y0

1
5.47
100

r 0.1

r 1

y0

6(0.1) 2
6

2(0.1)3 9(0.1) 2 11(0.1) 3


11.99
12

9.23

1.0316

6r 2 18r 11
12

y0 ...

6(0.1)2 18(0.1) 11
(11.99)
12

0.1 1 ( 9.23)

1
5.47 8.307 9.2523 0.2303
100
-----------------------------------------------------------------------------------------------------------3.
Find the maximum and minimum value of y tabulated below.
x
-2
-1
0
1
2
3
4

0.25

0.25

15.75 56

Solution:
dy
dx

1
h

y0

2r 1
2

y=f(x)

-2

y0

3r 2 6r 2
6

y0

2r 3 9r 2 11r 3
12
3

y0 ...
5

2
-2.25

-1

-0.25

2.5
0.25

-3

-0.5
-0.25

-0.25

2.5
2.25

0
6

11.5
13.75

0
6

15

15.75

26.5
40.25

56
Choosing x0=0, r = x 1 0

dy
dx

1
1

0.25

2x 1
3x 2 6 x 2
(2.5)
(9)
2
6

2 x3 9 x 2 11x 3
(6)
12

1
1
dy
dx

x3

0.25 2.5 x 1.25 4.5 x 2 9 x 3 x3 4.5x 2 5.5 x 1.5


x

dy
0 x3 x 0
dx
x = 0, x = 1, x = -1.
2
d y
3x 2 1
dx 2
d2y
d2y
d2y
ve
ve
ve
at x=0
at
x=1
at
x=-1
dx 2
dx 2
dx 2
y is maximum at x=0, minimum at x=1 and -1
x( x 1) 2
y ( x)
y0 x y0
y0 ...
2!
Maximum value =y(0) =0, Minimum value =y(1)=-0.25.
-----------------------------------------------------------------------------------------------------------4. Consider the following table of data
x
0.2
0.4
0.6
0.8
1.0
Now

f(x)

0.9798652

0.9177710

0.8080348

0.6386093

0.3843735

Find f(0.25),f(0.6) and f(0.95).


Solution:
Here h=0.2
0.25 is nearer to the starting of the given table. So we use Newtons forward
interpolation formula to evaluate f(0.25)
0.95 is nearer to the ending of the given table. So we use Newtons backward
interpolation formula to evaluate f(0.95)
0.6 is middle point of the given table. So we use Central Difference formula to
evaluate f(0.6)
The difference table
2
3
4
y
x
y=f(x)
y
y
y
0.2x0 0.9798652y0
-0.0620942
y0
0.4

0.9177710

0.047642
-0.1097362

0.6

0.8080348

y0

-0.0120473
-0.0596893

y0
4

y0

0.01310985
4

-0.1694255

yn

0.02515715

0.8

0.6386093

0.08484645
0.25427195

1.0
xn

yn

yn

yn

0.3843735
yn

To find f(0.25)
Newtons forward interpolation formula for derivative
2r 1 2
3r 2 6r 2 3
2r 3 9r 2 11r 3
hf ' x0 rh
y0
y0
y0
2
6
12

y0

...

2(0.25) 1
( 0.047642)
2
6(0.25) 2
0.0120473
6
9(0.25) 2 11(0.25) 3
( 0.01310985)
12

- 0.0620942
f ' 0.25

1
0.2

3(0.25) 2
2(0.25) 3

=-0.2536(correct to four decimal places)


To find f(0.95)
x

y=f(x)

0.2x0

0.9798652(y-2)

0.4

0.9177710 (y-1)

-0.0620942

0.047642
-0.1097362

0.6

y1

0.8080348 (y0)

-0.0120473
-0.0596893

2
4

y1

-0.01310985

-0.1694255
0.8

0.6386093(y1)

-0.02515715
-0.08484645

-0.25427195

1.0

y0
2

y0

y1

0.3843735(y2)
Newtons backward interpolation formula for derivative

y1

f ' xn

1
h

rh

x xn
h

2r 1
2

yn

0.95 1
0.2

2r 2

yn

6r 2
6

yn

2r 3 6r 2 11r 3
12

yn ...

0.25
2( 0.25) 1
( 0.08484645)
2
6( 0.25) 2
0.02515715
6
9( 0.25) 2 11( 0.25) 3
( 0.01310985)
12

- 0.25427195
3( 0.25) 2

1
0.2

f ' 0.95

2( 0.25)3

f(0.95) =-1.71604
To find f(0.6)
Central Difference formula (Stirlings Formula)
f ' x0

rh

y0

1
h

f ' 0.6

1
0.2

3r 2 1
12

0.1694255 0.1097362
2
3(0.2) 2 1
12

3
1

2r 3 r
12

5r 4
2

15r 2
5!

5
2

...

(0.2)( 0.00596893)

0.02515715 0.0120473

2(0.2) 3 0.2
( 0.01310985)...
12

f(0.6)=-0.74295(correct to 5 decimal places)


-----------------------------------------------------------------------------------------------------------5. Given the following data, find y(6), y(5) and the maximum value of y
x
0
2
3
4
7
9
y
4
26
58
112
466
922
Solution:
Since the intervals are , we will use Newtons divided difference formula.
Divided Difference Table
x y=f(x)
f (x)
0

26

f ( x)

f ( x)

f ( x)

11=f(x0,x1)
7= f(x0,x1,x2)
32
3

58

112

1= f(x0,x1,x2,x3)
11

54

1
16

118
7

0= f(x0,x1,x2,x3, x4)

466

0
1

22
228

922

By Newtons Divided Difference formula,


y =f(x)=f(x0)+(x-x0)f(x0, x1)+ (x-x0) (x-x1)f(x0, x1,x2)+.
=4+(x-0)11+(x-0)(x-2)7+(x-0)(x-2)(x-3)1
3
2
=x +2x +3x+4

Therefore, f(x) = 3x2+4x+3


f'(6)=135
f(5)=98.
-----------------------------------------------------------------------------------------------------------------Numerical Integration
Single integral
Trapezoidal
Simpsons one-third rule
Simpsons three-eighth rule
Romberg method
Two and Three point Gaussian Quadrature Formulas
Double integral
Trapezoidal rule
Simpsons Rule
Single Integral
Trapezoidal Rule
xn
h
f ( x)dx
sum of the first and last ordinates
2
x0
xn

h
y0
2

ie, f ( x)dx
x0

yn

2 y1

2 sum of the remaining ordinates

y2 ... yn 1

Simpsons one third rule


xn

f ( x)dx
x0

h sum of the first and last ordinates 2 sum of the remaining odd ordinates
3 4 sum of the remaining even ordinates

xn

h
y0
2

ie, f ( x)dx
x0

yn

2 y1

y3 ... yn 1

4 y2

y4 ... yn

Simpsons three eight rule


xn

f ( x)dx
x0

xn

f ( x)dx
x0

3h
8

sum of the first and last ordinates


3 sum of the remaining ordinates which are not divisibleby 3
2 sum of the remaining ordinates which are divisibleby 3

h
y0
2

yn

3 y1

y2

y4

y5 ... yn 1

2 y3

y6 ... yn

Rule
Trapezoidal Rule

Simpsons one third rule

Simpsons three eight


rule

Degree of y(x)

No.of intervals

One

Any

Two

Error

h2

(b a )h 2 M
12

h4

(b a )h 4 M
180

h5

3h 5
8

Even

three

Order

Multiple of 3

Two and Three point Gaussian Quadrature Formulas


Gaussian Two point formula
1

If the Limit of the integral is -1 to 1 then we apply

f ( x)dx

1
3

1
. This
3

formula is exact for polynomials upto degree 3.


b

f ( x)dx then x=

If
a
b

b a
t
2

b a
b a
and dx =
using these conditions convert
2
2

f ( x)dx into
a

f (t )dt and then we apply the formula

f (t )dt
1

1
3

1
3

Gaussian Three point formula

If the Limit of the integral is -1 to 1 then we apply


1

5
f
9

f ( x)dx
1

3
5

3
5

8
f (0) . This formula is exact for polynomials upto
9

degree 5.
b

f ( x)dx then x=

Otherwise
a
b

a
1

f (t )dt
1

b a
b a
and dx =
using these conditions
2
2

f ( x)dx into

convert

b a
t
2

f (t )dt and then we apply the formula


1

5
f
9

3
5

3
5

8
f (0)
9

Problems based on single integrals


2

sin x dx

1. Using Trapezoidal rule evaluate


0

Solution
x 0

15

0 0.5087

Trapezoidal rule
xn
h
f ( x)dx
y0 yn
2
x0
2

sin x dx
0

2 y1

30

45

60

0.7071

0.8408

0.9306

75

90

0.9828

y2 ... yn 1

12 (0 1) 2(0.5087
2

0.7071

0.8408

0.9306

0.9828 )

= 1.17024
-----------------------------------------------------------------------------------------------------------2. By dividing the range into ten equal parts, evaluate sin xdx by Trapezoidal rule and
0

Simpsons Rule. Verify your answer with integration


Solution:
Range = -0=. Hence h=

10
we tabulate below the values of y at different xs
0
x
2
4
3
5
10
10
10
10
10

6
10

sinx 0

.9511 .809

.309

.5878 .809

Trapezoidal rule
xn
h
f ( x)dx
y0 yn
2
x0
2

sin xdx
0

2 y1

10 (0 0) 2(0.3090
2

.9511 1

7
10

8
10

9
10

.5878 .309

y2 ... yn 1

0.5878

0.8090

0.9306

0.9511 1 0.9511 0.9306

= 1.9843 nearly
Simpsons one third Rule:
we use Simpsons one third rule only when the no.of intervals is even
here the no of intervals =10(even)
xn
h
f ( x)dx
y0 yn 2 y1 y3 ... yn 1 4 y2 y4 ... yn 2
2
x0

0.8090

0.3090 )

10 (0 0) 2(0.9511
2

sin xdx
0

0.5878

0.9306

0.9511

0.9306 ) 4(0.3090

0.8090

1 0.8090

0.3090 )

= 2.00091 nearly
We use Simpsons three eight rule only when the no.of intervals divisible by 3
Here the no of intervals =10 which is not divisible by 3.
So we cannot use this method.
By actual integration,

sin xdx

cos x

Hence, Simpsons one third rule is more accurate than the Trapezoidal rule.
------------------------------------------------------------------------------------------------------------------1

3. Evaluate (i) (3x

5 x )dx and (ii) (3x 2

5 x 4 )dx by Gaussian two and three point

formulas
Solution:
1

(3x 2

(i)

5 x 4 )dx

Gaussian two point formula


Given interval is -1 and 1.
Hence we can apply
1
1
1
f ( x)dx f
f
3
3
1
1

(3x

1
3

5x )dx 3

1
3

1
3
3

1
5
3

=3.112

Gaussian two point formula


Given interval is -1 and 1.
Hence we can apply
1

f ( x)dx
1

5
f
9

3
5

(3x

5 x )dx

=
1

3
5

8
f (0)
9
2

(3x 2

5 x 4 )dx = 4

5
9

5
9

3
5

3
5

3
5

9
25

3
5

3
5

9
25

3
5

8
(0)
9

(ii) (3x 2

5 x 4 )dx

Gaussian two point formula


Here the interval is 0 to 1. So we use the formula
b a
b a b a
t
x=
and dx =
dt
2
2
2
t 1
dt
i.e, x
and dx=
2
2
1

t 1
3
2

t 1
5
2

dt 1.556

Gaussian two point formula


Here the interval is 0 to 1. So we use the formula
b a
b a b a
t 1
dt
t
x=
and dx =
dt i.e, x
and dx=
2
2
2
2
2
1

t 1
3
2

f (t )

t 1
2

3
5

3
1

t 1
2

dt
2

3
5

3
5

0.7745 1
2

t 1
3
2

t 1
5
2

dt

0.7745 1
2

8
f (0)
9

0 1
5
2

t 1
2

1
2

t 1
2

0.7745 1
2

0 1
3
2

f (0)

3
5

t 1
5
2

5
f
9

f (t )dt

0.7745 1
2

0.038138

4.28

1.0625
5
0.038138
9

dt

4.28

8
(1.0625 )
9

= 4(approximately)
1
2

3
1

t 1
2

t 1
2

dt

4
2

------------------------------------------------------------------------------------------------------------

Double Integral
Trapezoidal Rule
d b

f ( x, y )dxdy where a, b, c, d are constants.

Evaluate
c a

hk
[sum of values in]+2(sum of values in
4
Simpsons Rule
I=

+4[sum of remaining values]

sum of the values of f at four corners


2(sum of the values of f at the odd positions on the
boundaryexcept thecorners)

4(sum of the values of f at the even positions on the boundary)


{4(sum of the values of f at the odd positions)

hk
9 8(sum of the values of f at the even positions)
on the odd row f of the matrix except boundary rows}
{8(sum of the values of f at the odd positions)
16(sum of the values of f at the even positions)
on the even row f of the matrix }

Problems based on Double integrals


1.4
1
dxdy using Trapezoidal and Simpsons rule. Verify your result by actual
1. Evaluate
xy
1 22.4
integration.
Solution:
Divide the range of x and y into 4 equal parts
2. 4 2
h
0.1
4
1 .4 1
k
0.1
4
1
Get the values of f(x, y) = at nodal points
xy

y\x

2.1

2.2

2.3

2.4

0.5

0.4762

0.4545

0.4348

0.4167

1.1

0.4545

0.4329

0.4132

0.3953

0.3788

1.2

0.4167

0.3968

0.3788

0.3623

0.3472

1.3

0.3846

0.3663

0.3497

0.3344

0.3205

1.4

0.3571

0.3401

0.3247

0.3106

0.2976

(i) Trapezoidal Rule

(0.1)(0.1) 0.5 0.4167 0.3571 0.2976


4

0.4545 0.4167 0.3846 0.4762 0.4545


2 0.4348 0.3788 0.3472 0.3205 0.3106
0.3247 0.3401

4 0.4329 0.4132 0.3953 0.3968 0.3788 0.3623 0.3663 0.3497 0.3344


I=0.0614
(ii) Simpsons Rule:
0.5 0.4167 0.3571 0.2976
I

(0.1)(0.1)
9

2(0.4167 0.4545 0.3472 0.3247)

4(0.3846 0.4545 0.4762 0.4348 0.3788 0.3205 0.3106


0.3401 0.3788) 8(0.3968 0.3623 0.3497 0.4132)
16(0.3663 0.3344 0.4329 0.3953)

0.01(55.2116)
9
I=0.0613
-----------------------------------------------------------------------------------------------------------I=

UNIT IV
INITIAL VALUE PROBLEMS FOR
ORDINARY DIFFERENTIAL EQUATIONS
Taylor series method
Euler methods
Runge-Kutta method for solving first and second order equations
Milnespredictor and corrector method
Adams predictor and corrector method.

Introduction
An ordinary differential equation of order n is a relation of the form
dry
F x, y, y ' , y" ,..., y n
0 where y y x and y r
. The solution of this differential
dx r
equation involves n constants and these constants are determined with the help of n conditions
y, y ' , y" ,..., y n 1 are prescribed at x x0 , by
y x0

y0' ,... y

y0 , y ' x0

n 1

x0

y0n

These conditions are called the initial conditions because they depend only on x 0 .
The differential equation together with the initial conditions is called an initial value problem.
Taylors Series
Point wise solution
If y x is the solution of (1), then by Tailor series,
2

y x

y0

Put x1

x0

x x0 '
x x0
x x0
y0
y0"
y0"' ...
1!
2!
3!
h where h is the step-size, we have

h ' h 2 " h 3 '"


y0
y0
y0 ... (3)
1!
2!
3!
once y1 has been calculated form (1), y1' , y1" , y1'" can be calculated from
y x1

y1

y0

y' f x, y
Expanding y x in a Taylor series about x

x1 , we get

h 2 " h3 '"
y1
y1 ...
2!
3!
Where y2 y x2 and x2 x1 h
The Tailor algorithm is given as follows
h ' h 2 " h 3 '"
ym 1 ym
ym
ym
ym ...
1!
2!
3!
dry
Where ymr
at the point xm , ym where m 0,1,2,...
dx r
Problems based on Taylors Series
Solve y ' y 2 x; y 0 1 using Taylor series method and compute y 0.1 and y 0.2
Solution
Here x0 0, y0 1 .
y2

y1

h '
y1
1!

Given y ' y 2

To find

y0'

"
0

y0"'

y0iv

34

y" 2 yy' 1

;
;

y" ' 2 yy" 2 y '2


y iv

6 y" 2 yy" '

etc.,

Here h 0.1
By Taylor algorithm
h ' h 2 " h 3 '"
y1 y0
y0
y0
y0 ...
1!
2!
3!
0.01
0.001
y1 1 0.1 1
3
8
2
6

y 0.1 1.1164
To find y 0.2
x2 x1 h where x2

0.001
34
24

0.2

h ' h " h 3 '"


y1
y1
y1
1!
2!
3!
2
0.1 1.1164
1.3463

y2

y1

y1'

y1" 1 2 1.1164 1.3463


"'
1

2 1.1164 4.006

...

4.006
2

2 1.3463

= 12.5696

y 0.2

1.1164

0.01
4.006
2

0.1 1.3463

0.0001
12.5696
6

y 0.2 1.2732
---------------------------------------------------------------------------------------------------------2. Evaluate y 0.1 and y 0.2 , correct to four decimal places by Taylor series method, if y x
satisfies y ' xy 1, y 0 1
Solution
Here x0 0, y0 1
;
y0' 1
y' xy 1
;
y0" 1
y" xy' y
y" ' xy" 2 y '
y

iv

xy ' " 3 y"

;
;

y0"
y

iv
0

2
3

To find y 0.1
By Taylor algorithm
h ' h 2 " h 3 '"
y1 y0
y0
y0
y0 ...
1!
2!
3!
Let x1 x0 h where h 0.1

x1 0.1
y1 1

0.1

0.01
1
2

= 1.1057
To find y 0.2
Let y2 y x2 where x2

0.001
2
3

x1 h

0.0001
3
24

x2 0.2
By Taylor algorithm,
h ' h 2 " h 3 '"
y1
y1
y1
y1 ...
1!
2!
3!
1 0.1 1.1057 1.11057

y2

y1'
y1"

0.1 1.11057

"'
1

1.1057

1.216757

0.1 1.216757

y2 1.1057

2 1.11057
2.3428157
0.01
0.1 111057
1.216757
2

0.0001
2.342815
6

y2 1.2178
Hence y 0.1 1.1057 and y 0.2 1.2178 .
-------------------------------------------------------------------------------------------------------3.Solve by Taylor series method, y ' xy y 2 , y 0 1 at x 0.1 and 0.2, , correct to four decimal
places.
Solution
Given x0 0, y0 1
y ' xy
y" y

y2

;
;

xy' 2 yy'

y" ' 2 y' xy" 2 y '


y iv

2 yy" ;

3 y" xy" 6 y ' y" 2 yy" '

To find y1 y 0.1
Let x1 x0 h . Here h 0.1
x1 x0 h x1 0.1
By Taylor algorithm
h ' h 2 " h 3 '"
y1 y0
y0
y0
y0 ...
1!
2!
3!
0.01
0.001
y1 1 0.1 1
3
10
2
6
1.1 0.015 0.0017 0.0002

y 0.1 1.1169
To find y 0.2
y2 y x2 where x2 x1 h
Here x1 0.1, h 0.1, x2 0.02
By Taylor algorithm,
y2

y1

h '
y1
1!

h2 "
y1
2!

h 3 '"
y1
3!

x1 0.1, y1 1.1169
y1'

x1 y1

y12

y1' 1.35925

...

y0'

"
0

y0"

10

y0iv

47

0.0001
47
24

y1"

y1

x1 y1' 2 y1 y1'

y1' 1.35925

1.1169 0.1 1.35925


4.2891

2 1.1169 1.35925

y1"' 2 y1' x1 y1" 2 y1'


2 y1 y1"
2.7185 0.42891 3.6951 6.5810
16.4235
y1iv 3 y1" x1 y1"' 6 y1 y1" 2 y1 y1"'
212.8673 1.64235 34.9797 36.6868
86.17615
0.01
0.001
0.0001
y2 1.1169 0.1 1.35925
4.2891
16.4235
86.17615
2
6
24
1.1169 0.135925 0.02144 0.00274 0.00036
y 0.2 1.2774
Hence y 0.1 1.1169 and y 0.2 1.2774 .
-----------------------------------------------------------------------------------------------------------Taylor series method for simultaneous first order differential equation
dy
dy
f x, y , z ;
g x, y, z with initial
We can solve the equations of the form
dx
dx
y0 and z x0 z0 .
conditions y x0
The values of y and z at x1 x0 h are given by Taylor algorithm,
h ' h 2 " h 3 '"
y0
y0
y0 ...
1!
2!
3!
h ' h 2 " h 3 '"
z1 z0
z0
z0
z0 ...
1!
2!
3!
The derivatives on R.H.S of the above expressions are found at x x0 using the given
equations.
Similarly y 2 and z2 corresponding to x2 x1 h are calculated by Taylor series method.
Problems
dx
dy
1 ty;
tx given x 0, y 1 at t 0 by Taylor
1.Evaluate x 0.1 and y 0.1 given
dt
dt
series method.
Solution
Given t0 0, x0 0, y0 1
x' 1 ty
y' tx
y1

y0

x" y ty'
x'" 2 y' ty"

y"
y" '

x iv 3 y" ty" '


Then

y iv

x0'

y0'

x tx'
2 x' tx"
3x" tx" '
0

x 0"

"'
0

y 0"

"'
0

iv
0

iv
0

0
2
3

By Taylor algorithm, we have


h ' h 2 " h 3 '"
y1 y0
y0
y0
y0 ...
1!
2!
3!
0.01
0.001
y 0.1 1 0.1 0
0
2
6

y 0.1

0.0001
24

0.9997

x 0.1

x1

x0

0.1 1

h '
x0
1!

0.01
1
2

h2 "
x0
2!

h 3 '"
x0 ...
3!

.....

0.105
-----------------------------------------------------------------------------------------------------------dx
dy
xy 2t;
2ty x , with
2. Find x 0.2 and y 0.2 using Taylor series method given that
dt
dt
initial conditions x 1, y 1 at t 0 .
Solution
1
Given t 0 0, x0 1, y0
Taking h 0.2, t1 t0 h t1 0.2
x' xy 2t
y' 2ty x
x" xy' x' y 2
y" 2ty' 2 y x'
x'" xy" 2 x' y x" y
y"' 4 y' 2ty" x"
x0'

y 0'

"
0

"'
0

4
9

"
0

'
0

1
3

By Taylor algorithm, we have


h ' h 2 " h 3 '"
x1 x0
x0
x0
x0 ...
1!
2!
3!
h ' h 2 " h 3 '"
y1 y0
y0
y0
y0 ...
1!
2!
3!

y1

x 0.2

x 0.2

0.796

y 0.2

0.2

1 0.2

2 0.2

1.5 0.04

3
0.2
2

...

4
0.008
3

0.8493
0.8493
Hence x 0.2 0.796 and y 0.2
------------------------------------------------------------------------------------------------------------

Taylor series method for second order differential equations


d2y
dy
f x, y ,
Consider the differential equation
with the initial conditions y x0 y0
2
dx
dx
and y ' x0 y0' , where y0 , y0' are known values.
This equation can be reduced into a set of simultaneous equations, by putting y' p
we have y ' p, y x0 y0
(1)
'
And p ' f x, y, p , p x0
(2)
p0 y 0
Successively differentiating y" , the expression for y" ' , y iv etc., are known.
By Taylor series method, we find
h ' h 2 " h 3 '"
y1 y0
y0
y0
y0 ...
1!
2!
3!
h 2 " h 3 '"
y0 hp0
p0
p0 ...
2!
3!
Also by Taylor series method, we have
h ' h 2 " h 3 '"
p1 y1' p0
p0
p0
p0 ...
1!
2!
3!
Then the values y1" , y1"' , y1iv are found from y1" , y1"' etc

h ' h 2 " h 3 '"


y1
y1
y1 ...
1!
2!
3!
Thus we calculate y1 , y2 ,...
Problems
1.Find y 0.2 and 0.4 given y" xy if y 0
Solution
Given y" xy, x0 0, y0 1 and y' 0 1
Then we have
y" ' xy' y
y iv xy" 2 y '
y v 3 y" xy" ' etc
y2

y1

y0"

0, y0"'

1, y0iv

2, y0v

By Taylor Algorithm,
h ' h 2 " h 3 '"
y1 y0
y0
y0
y0 ...
1!
2!
3!
Taking, h 0.2, x1 x0 h x1 0.2
0.04
0.008
y1 y x1 1 0.2 1
0
1
2
6
y 0.2 1.2014
To find y1'

1, y ' 0

0.0016
2
24

1 by Taylor series method.

Set p y' . Then p' xy


p" xy' y; p" ' xy" 2 y ' ; p iv
p0

'
0

1, p

0, p

"
0

1, p

'"
0

2, p

3 y" xy" '


iv
0

By Taylor Algorithm,
h ' h 2 " h 3 '"
p1 p0
p0
p0
p0 ...
1!
2!
3!
0.04
0.08
p1 1 0.2 0
1
2
2
6
y1' 1 0.02 0.0027
1.0227
Let x2 x1 h . Since h 0.2, x1 0.2 x2 0.4
By Taylor series method,
h ' h 2 " h 3 '"
y x2
y 0.4 y1
y1
y1
y1 ...
1!
2!
3!
y" xy
y1" x1 y1 0.2 1.2014 0.24028
y" ' xy' y
y1'" 0.2 1.0227 1.2014 1.40594
y iv 2 y ' xy" y1iv 0.2 0.024028 2 1.0227 0.00048 2.0454 2.0459
0.04
0.008
0.0016
y2 1.2014 0.2 1.0227
0.024028
1.40594
2.0459
2
6
24

y 0.4 1.4084
Hence y 0.2 1.2014 and y 0.4 1.4084
-----------------------------------------------------------------------------------------------------------2.Find y at x 1.1,1.2 given y" x 3 y 2 y ' , y 1 and y' 1 1 , correct to four decimal places
using Taylor series method.
Solution
Given x0 1, y0 1; y0' 1
y" x 3

y 2 y'

y0"

y"' 3x 2 y y'
y 2 y" y0" 1
y iv 6 x 2 y 3 4 yy' y" 2 yy' y" y 2 y" '
y0iv
y

'
0

6 2 0 0 1 3
0, y0"

0, y0"'

1, y0iv

By Taylor series method,


h ' h 2 " h 3 '"
y1 y0
y0
y0
y0 ...
1!
2!
3!
Taking, h 0.1, x0 1, x1 x0 h 0.1
We have
0.01
0.001
0.0001
y 1.1 1 0.1 1
0
1
3
2
6
24

y 1.1

1.1002

To find y1'
Set p y' . Then p ' x 3
p

1
0

1 1 0, p

"
0

1, p

"'
0

y 2 y ' , p0

y0'

3 etc.,

h " h 3 '"
p0
p0 ...
2!
3!
0.01
0.01
p1 1 0.1 0
1
2
2
6
p1 1.0053
y1' 1.0053

p1

h '
p0
1!

p0

y" x 3

y 2 y'

y1"

1. 1

1.1002

1.0053

0.11415
y" ' 3 xy 2

2 y y'

y 2 y"

y1"'

3 1.1

1.268
By Taylor Algorithm,
h ' h 2 " h 3 '"
y2 y1
y1
y1
y1 ...
1!
2!
3!
Where y2 y x2 , x2 1.2
0.01
1.1002 0.1 1.0053
0.11415
2

2 1.1002 1.0053

1.1002

0.11415

0.01
1.268
6

y 1.2 1.2015
Hence y 1.1 1.1002 and y 1.2 1.2015
-----------------------------------------------------------------------------------------------------------Euler method
Let y1 y xi , where x1 x0 h
Then y1 y x0 h . Then by Taylors series,
h
h2
(1)
y ' x0
y" x0 .....
1!
2!
Neglecting the terms with h 2 and higher powers of h, we get from (1),
y1 y0 hf x0 , y0
(2)
y1

y x0

Expression (2) gives an approximate value of y at x1


Similarly, we get y2 y1 hf x1 , y1 for x2 x1 h .
for any m, ym 1 ym hf xm , ym , m 0,1,2,...

x0

h.

(3)

In Eulers method, we use (3) to compute successively y1 , y2 ,... etc., with an Error 0 h 2
Modified Euler method
The algorithm presented already in Modified Euler method in unit IV is sometimes
referred as Improved Euler Method.
Therefore a different algorithm for Modified Euler method to solve
dy
f x, y , y x0 y0 is explained with illustrations.
dx

Explanation: Modified Euler Method


h
h
yx h yx j f x
,y
f x, y
2
2

h
h
, yn
f x, y
2
2
Where yn 1 y xn h and h is the step size
Problems based on Eulers Method
(or) yn

yn h f xn

1,Using Eulers method, compute y in the range 0 x 0.5 , if y satisfies

dy
dx

3x y 2 , y 0

1.

Solution
Here f x, y 3 x y 2 , x0 0, y0 1
By Eulers, method
yn 1 yn hf xn , yn , n 0,1,2,...
Choosing h 0.1 , we compute the values of y using (1)
yn 1 yn hf xn , yn , n 0,1,2,...

y 0.1
1.1

y1

y0 hf x0 , y0

y 0.2

y2

y1

1.1

hf x1 , y1

0.1 3 0
0.1 0.3

1.1

1.251
2
y 0.3 y3 y2 hf x2 , y2 1.251 0.1 0.6 1.251
1.4675
2
y 0.4 y4 y3 hf x3 , y3 1.4675 0.1 0.9 1.4675
1.7728
2
y 0.5 y5 y4 hf x4 , y4 1.7728 0.1 1.2 1.7728
2.2071
-----------------------------------------------------------------------------------------------------------dy y x
, with y 1 for x 0 . Find y approximately for x
dx y x
five steps.
Solution
Given y0 1, x0 0 , choosing h 0.002 ,

2.Given

xi

x0 ih, i 1,2,3,4,5

To find y1 , y2 , y3 , y4 and y 5 where yi y xi


y x
Here f x, y
y x
Using yn 1 yn hf xn , yn , n 0,1,2,...
We get

0.1 by Eulers method in

y1

y0

hf x0 , y0

0.02

1 0
1 0

1.02
y2

y1 hf x1 , y1

1.02

0.02

1.02 0.02
1.02 0.02

1.0392
y3

y2

hf x2 , y2

1.0392

0.02

1.0392
1.0392

0.04
0.04

0.02

1.0577
1.0577

0.06
0.06

0.02

1.0756
1.0756

0.08
0.08

1.0577

y4

y3 hf x3 , y3

1.0577
1.0756

y5

y4

hf x4 , y4

1.0756

1.0928
Hence y 1.0928 and x 0.1
---------------------------------------------------------------------------------------------3.Compute y at x 0.25 by Modified Euler method given y' 2 xy, y 0 1
Solution

Here f x, y 2 xy, x0 0, y0 1
Choose h 0.25, x1 x0 h 0.25
By Modified Euler method
h
h
y1 y0 h f x0
, y0
f x0 , y0
2
2
f x0 , y0 2 0 1 0
f x0

h
, y0
2

h
f x0 , y0
2

f 0.125 ,1

0.25

y1 1 0.25 0.25 1.0625


Hence y 0.25 1.0625 .
-------------------------------------------------------------------------------------------------dy
x2 y2 , y 0 1 .
4.Using Modified Euler method, find y 0.1 and y 0.2 given
dx
Solution
Here x0 0, y0 1, h 0.1
h
0.1
f x, y x 2 y 2 , y0
f x0 , y0 1
0 1 1.05
2
2
y1

y0

hf x0

h
, y0
2

h
x0 , y0
2

1 0.1 f 0.05,1.05

1 0.1 0.05

y 0.1

f 0.1,1.1105

h
f x1 , y1
2
y2

1.1105

1.1105

f x1 , y1

y1

1.05

1.1105

y1 hf x1

h
, y1
2

0.1

1.1105

1.24321

0.05 1.24321 1.17266


h
f x1 , y1
2

1.1105

0.1 f 0.15,1.17266

1.1105

0.1 0.15

1.17266

y 0.2 y2 1.2503
y 0.2 1.2503
----------------------------------------------------------------------------------------------------Fourth-order Range-Kutta method
This method is commonly used for solving the initial value problem
dy
f x, y , y x0 y0
dx
Working Rule
The value of y1 y x1 where x1 x0 h where h is the step-size is obtained as follows.
We calculate successively.
k1

hf x0 , y0

h
k1
, y0
2
2
h
k2
k3 hf x0
, y0
2
2
k 4 hf x0 h, y0 k3
Finally compute the increment
k2

hf x0

1
k1 2k2 2k3 k4
6
The approximate value of y1 is given by
1
y1 y0
y y1 y0
k1 2k2 2k3 k4
6
Error in R-K fourth order method 0 h 5
In general, the algorithm can be written as
1
ym 1 ym
k1 2k2 2k3 k4 where
6
y

k1

hf xm , ym

k2

hf xm

h
, ym
2

k1
2

h
k2
, ym
2
2
k 4 hf xm h, ym k3
where m 0,1,2,...
Runge-Kutta method for second order differential equations
d2y
dy
f x, y ,
Consider the second order differential equation
with initial conditions
2
dx
dx
y x0
y0 and y ' x0
y0' . This can be reduced to a system of simultaneous linear first order
k3

hf xm

equations, by putting z

dy
dx
dz
dx

z with y x0

z x0

y0

f x, y, z , with z x0

dy
dx
dz
and
dx
i.e.,

dy
. Then we have,
dx

y ' x0

g x, y, z , where g x, y, z

y0'

f x, y, z with initial conditions y x0


z0 where z 0

y0 and

y 0'

Now, starting from x0 , y0 , z0 , the increments y and z in y and z are given by (h-step size)
l1 hf x0 , y0 , z0
k1 hg x0 , y0 , z0
k2

hg x0

k3

hg x0

k4

hg x0

h
k1
l
, y0
, z0 1
2
2
2
h
k2
l
, y0
, z0 2
2
2
2
h, y0 k3 , z0 l3

l2

hf x0

l3

hf x0

l4

hf x0

h
k1
l
, y0
, z0 1
2
2
2
h
k2
l
, y0
, z0 2
2
2
2
h, y0 k3 , z0 l3

1
1
k1 2k2 2k3 k4
z
l1 2l2 2l3 l4
6
6
z respectively.
y and z1 z0
Then for x1 x0 h , the values of y and z are y1 y0
By repeating the above algorithm the value of y at x2 x1 h can be found.
Problems based on RK Method
dy
x 2 y x, y 0 1 taking h 0.1 using Runge-Kutta
1.The value of y at x 0.2 if y satisfies
dx
method of fourth order.
Solution
Here f x, y x 2 y x, x0 0, y0 1.
Let x1 x0 h , choosing h 0.1, x1 0.1.
Then by R-K fourth order method,
1
y1 y0
k1 2k2 2k3 k4
6
y

k1

hf x0 , y0

k2

hf x0

k3

hf x0

k4

hf x0

y1 1

y 0.1
To find

0.1 0 0

h
k1
, y0
hf 0.05,1 0.00525
2
2
h
k2
, y0
hf 0.05,1.0026 0.00525
2
2
h, y0 k3 hf 0.1,1.00525 0.0110050

1
0 0.00525 0.00525 2 0.011005 1.0053
6
1.0053
y2 y x2 where x2 x1 h . Then x2 0.2
1
k1 2 k2 k3 k4
6
x1 , y1
0.1 0.1 0.01 1.0053
0.0110

y2

y1

k1

hf

k2

hf x1

k3

hf x2

h
k1
, y1
2
2
k2
h
, y2
2
2

0.1 0.15
0.1 0.15

0.15

0.15

1.0108
2

0.01727

1.013935

k4

0.01728

hf x1 h, y1 k3
0.1 0.2 0.2 1.02258 0.02409
1
y2 1.0053
0.0110 2 0.01727 0.01728 0.02409 1.0227
6
y 0.2 1.0227
Hence y 0.2 1.0227 .
--------------------------------------------------------------------------------------------------2. Apply Runge-Kutta method to find an approximate value of y for x = 0.2 in steps of 0.1 if
dy
x y 2 , y 0 1 , correct to four decimal places.
dx
Solution
Here f x, y x y 2 , x0 0, y0 1
We choose h 0.1
1
y1 yo
k1 2k2 2k3 k4
6
k1 hf x0 , y0
0.1 0 12 0.1
h
, y0
2
h
, y0
2

k1
2
k2
2

k2

hf x0

k3

hf x0

k4

hf x0 h, y0 k3

0.1 0.05

1.05

0.1 0.05

1.0576

0.1 0.1 1.1168

0.1152
2

0.1168

0.1347

1
0.1 2 0.1152 0.1168 0.1347
6
y1 1.1165
Hence y 0.1 1.1165 .
--------------------------------------------------------------------------------------------------dy
3.Use Runge-Kutta method to find y when x=1.2 in steps of 0.1, given that
dx
y 1 1.5 .
Solution
y1 1

2
2
Given f x, y x y , x0 1, y0 1.5
Let x1 x0 h , we choose h 0.1
1
y1 yo
k1 2k2 2k3 k4
6
2
k1 hf x0 , y0
0.1 1 1.5
0.325
h
k1
2
k 2 hf x0
, y0
0.1 1.05
1.6625
2
2
h
k2
2
k3 hf x0
, y0
0.1 1.05
2.8673
2
2

k4

x2

y 2 and

0.3866
0.397

hf x0 h, y0 k3
0.1 1.1
1.897
0.4809
1
y1 1.5
0.325 2 0.3866 0.397 0.4809 1.8955
6
y1 1.8955
To compute y 1.2 :
y2 y x2 where x2 x1 h 1.2 , since h 0.1
1
y2 y1
k1 2 k2 k3 k4
6
k1

hf x1 , y1

k2

hf x1

k3

hf x2

k4

hf x1

0.1 1.1

1.8955

0.4803

h
k1
2
2
, y1
0.1 1.15
2.1356
0.5883
2
2
h
k2
2
, y2
0.1 1.3225
2.1897
0.6117
2
2
h, y1 k3
0.1 1.44 6.286 0.7726

1
0.4803 2 0.5883 2 0.6117 0.7726 2.5043
6
Hence y 1.2 2.5043 .
--------------------------------------------------------------------------------------------------dy
dz
xz 1,
xy for x = 0.3 and 0.6. Given that y 0, z 1 when x 0
4.Solve the equation
dx
dx
y2 1.8955

Solution
Here f1 x, y, z

1 xz, x0

To find y 0.3 and z 0.3


k1 hf x0 , y0
0.3 0 1
k2

hf x0

k3

hf x0

k4

hf x0

h
k1
, y0
2
2
h
k2
, y0
2
2
h, y0 k3

l1

hf x0 , y0 , z0

l2

hf x0

l3

hf x0

l4

hf x0

y 0.3

0, y0

0, z0

0 and h

0.3
0.3 0.15 1

0.3450

0.3 0.15 0.9966 1

0.3 0.3 0.99224

0.3 0

0.3

0.3448

0.3893

h
k1
l
, y0
, z0 1
0.3 0.15 0.15
0.00675
2
2
2
h
k2
l
, y0
, z0 2
0.3 0.15 0.1725
0.00776
2
2
2
h, y0 k3 , z0 l3
0.3 0.3 0.3448
0.031036

y1

y0

1
k1 2k2 2k3 k4
6

1
0.3 2 0.3450 0.3448 0.3893
6
y 0.3 0.3448
1
z 0.3 z1 z0
l1 2l2 2l3 l4
6
1
z 0.3 1
0 2 0.00675 0.00776 0.031036
6
z 0.3 0.9899
To find y at x = 0.6, the starting values are x1 0.3, y1 0.3448 , z1 0.9899 and h 0.3
0

k1

hf x0 , y0

k2

hf x0

k3

hf x0

k4

hf x0

0.3 1

h
k1
, y0
2
2
h
k2
, y0
2
2
h, y0 k3

l1

hf x0 , y0 , z0

l2

hf x0

l3

hf x0

l4

hf x0

0.3

0.3 0.9899

0.3891

0.3 1

0.45 0.9744

0.4315

0.3 1

0.45 0.9535

0.4287

0.3 1

0.6 0.9142

0.3 0.3448

0.4645

0.0310

h
k1
l
, y0
, z0 1
0.3
0.45 0.53935
0.0728
2
2
2
h
k2
l
, y0
, z0 2
0.3 0.45 0.56055
0.0757
2
2
2
h, y0 k3 , z0 l3
0.3
0.6 0.7735
0.1392

1
k1 2k2 2k3 k4
6
1
0.3448
0.3891 2 0.4315 0.4287 0.4645
6
y 0.6 0.7738
1
z2 z1
l1 2l2 2l3 l4
6
1
0.9899
0.0310 2 0.0728 0.0757 0.1392
6
z 0.3 0.9210
--------------------------------------------------------------------------------------------------Using R-K method of fourth order solve y" x y ' 2 y 2 for x=0.2, given that y=1 and
y ' 0 when x = 0.
Solution
Let y' z then y" z '
Hence the given equation reduces to the form,
dy
dz
z and
xz 2 y 2
dx
dx
Given x0 0, y0 1, z0 0 and h 0.2
y2

y1

Take f1 x, y, z
k1 hf x0 , y0
k2

hf x0

k3

hf x0

k4

hf x0

h
, y0
2

k1
2

h
, y0
2
h, y0

0.2

k2
2
k3

0.1

0.02

0.999

0.3

0.1958

hf x0 , y0 , z0

0.2 0 1

l2

hf x0

l3

hf x0

l4

hf x0 h, y0 k3 , z0 l3

y1

y0

h
k1
l
, y0
, z0 1
2
2
2
h
k2
l
, y0
, z0 2
2
2
2

1
k1 2k2
6

y2

0.2

l1

xz 2

z , f 2 x, y , z
0.2 0 0

0.01998

0.03916

0.2

0.2 0.1
0.2 0.1

0.2 0.2

0.1

0.0999

0.99
2

0.1958

0.99
2

0.1998
2

0.98

0.1958
2

0.1905

2k 3 k 4

1
0.2 2 0.02 0.01998
6

y 0.2

0.9801

Also z 0.2

0.03916

1
0.2 2 0.1998 0.1958
6

0.1905

0.1969

------------------------------------------------------------------------------------------------------------------

Multi-Step Methods (Predictor-Corrector Methods)


Introduction
Predictor-corrector methods are methods which require function values at
xn , xn 1 , xn 2 , xn 3 for the computation of the function value at x n 1 . A predictor is used to find the
value of y at x n 1 and then the corrector formula is used to improve the value of y n 1 .
The following two methods are discussed in this chapter
(1) Milnes Method (2) Adams Method
Milnes Predictor-corrector method
Consider

the

initial

value

dy
f x, y , y x0 y0 . Assume that
dx
where xi 1 xi h, i 0,1,2,3 are known, these

problem

y0 y x0 , y1 y x1 , y2 y x2 , and y3
are the starting values.
Milnes predictor formula

y x3

4h '
2 y1 y2' 2 y3' and
3
Milnes corrector formula
h '
y4 ,c y2
y2 4 y3' y4' where y4' f x4 , y4, p
3
--------------------------------------------------------------------------------------------------Problems based on Predictor-Corrector method
dy
1 y 2 , y 0 1, y 0.2
1.Using Milnes method, compute y(0.8) given that
dx
y 0.4 0.4228 and y 0.6 0.6841
Solution
we have the following table of values
x
y
y' 1 y 2
0
0
1.0
0.2 0.2027 1.0411
0.4 0.4228 1.1787
0.6 0.6481 1.4681
y4 , p

x0

0, x1

y0

0.2, x2

y0

0, y1

0.2027 , y2

y0'

1, y1'

1.0411 , y2'

0.4, x3

0.6

0.4228 , y3
1.1787 , y3'

0.6841
1.4681

To find y (0.8):

x4 0.8 . Here h = 0.2


By Milnes predictor formula,
4h '
y 4 , p y0
2 y1 y2' 2 y3'
3
0.8
0
2 1.0411 1.1787 2 1.4681
3
y4, p

1.0239

0.2027 ,

y4'

f x4 , y 4

1.0239

= 2.0480
By Milnes corrector formula,
h '
y4 ,c y2
y2 4 y3' y4'
3
0.2
0.4228
1.1787 4 1.4681
3

2.0480

y 0.8 1.0294
--------------------------------------------------------------------------------------------------2.Given y ' x 2 y, y 0 1, y 0.1 0.9052 , y 0.2 0.8213 , find y(0.3) by Taylor series
method. Also fine y(0.4) by Milnes method
Solution
y0 1
Given x0 0 ,
x1 0.1 ,
y1 0.9052
x2 0.2 ,
y2 0.8213
x3 0.3 ,
y3 y x3
By Taylor algorithm
y3

hy2'

y2

y' x 2

h2 "
y2 ...
2!
y" 2 x y'

y" ' 2 y" , y iv

'
2

0.2

y" ' etc

0.8213

0.7813

y2"

2 0.2

y2'"

2 1.1813 0.8187

y2iv

y3

0.7813 1.1813

8187

0.8213

0.1 0.7813

0.01
1.1813
2

y 0.3 0.7492
For x3 0.3 y3 0.7492 and y3'
'
0

'
1

0.001
0.8187
6

0.09 0.7492

Also y
1, y
0.01 0.905
0.8952 and y
By Milnes method
4h
y 4 , p y0
2 y1' y2' 2 y3'
3
0.4
y4, p 1
2 0.8952 0.7813 2 0.6952
3
y4, p 0.6897
y4'
0.16
0.6897
0.5297
By Correctors formula,

0.6592
'
2

0.7813

0.0001
0.8187
24

h '
y2 4 y3' y4'
3
0.1
y4,c 0.8213
0.7813 4 0.6592 0.5297
3
y4,c 0.6897 y 0.4 0.6897
--------------------------------------------------------------------------------------------------Adam-Bash Forth Predictor-Corrector Method
Using Newtons backward difference interpolation formula, we derive a set of predictor
and corrector formulae. This method also requires past four values to estimate the fifth value.
Adams predictor formula
h
yn 1, p yn
55 yn' 59 yn' 1 37'n 2 9 yn' 3
24
Adams corrector formula
h
yn 1,c yn
59 yn' 1 19 yn' 5'n 1 yn' 2
24
The errors in these formulae are
251 5 iv
19 5 iv
h f
h f
and
respectively.
720
720
In particular,
h
y4, p y3
55 y3' 59 y2' 37 y1' 9 y0'
24
And
h
y4,c y3
9 y4' 19 y3' 5 y2' y1'
24
--------------------------------------------------------------------------------------------------Given y ' 1 y 2 , y 0 0, y 0.2 0.2027 , y 0.4 0.4228 , y 0.6 0.6841 , estimate
y4 ,c

y2

y 0.8 using Adams method.


Solution
Form the given data
y0 0 ,
x0 0 ,
x1 0.2
y1 0.2027
x2 0.4
y2 0.4228
x3

0.6

y3

y0'

0.6841

To fine y 4 for x4 0.8 . Here h =0.2


h
y4, p y3
55 y3' 59 y2' 37 y1' 9 y0'
24
0.2
y4, p 0.6841
55 1.4680 59 1.1786
24
y4, p 1.0235

y1' 1.0411
y2'

1.1786

'
3

1.4680

37 1.0411

y4 ,c
y4'

h
9 y4' 19 y3' 5 y2'
24
2
1 1.0235
2.0475

y4 ,c
y4 ,c

y3

0.6481

0.2
9 2.0475
24

y1'

19 1.4680

5 1.1786

1.0411

1.0297

y 0.8 1.0297
------------------------------------------------------------------------------------------------------------

UNIT V
BOUNDARY VALUE PROBLEMS IN
ORDINARY AND
PARTIALDIFFERENTIAL EQUATIONS
Finite difference solution of second order
ordinary differential equation
Finite difference solution of one dimensional
heat equation by explicit and implicit methods
One dimensional wave equation
Two dimensional Laplace equation
Poisson equation

Solution of Boundary Value Problems in ODE


Solution of One Dimensional Heat Equation
Solution of One Dimensional Wave Equation
Solution of Laplace Equation
Solution of Poisson Equation

There are number of methods for solving second order boundary value problems.
Finite Difference Method
Shooting Method
Finite difference solution of second order ordinary differential equation
Let us consider
y(x)+f(x)y(x)+g(x)y(x)=0
with boundary conditions y(x0)=a and y(xn)=b
Formula
yi
y" ( x)

2 yi
h2

yi

and y' ( x)

yi

yi 1

2h

Steps for solving the second order ODE

Put y(x), y(x), y(x) by yi " , yi ' , yi respectively


substitute the above formula
Form an equation
Substitute the boundary conditions
Solve the equations by putting i=1, 2, 3
Write the solution of the intermediate values of boundary conditions

Problems based on Finite difference method for ODE


1. Solve by finite difference method, the boundary value problem y(x) y(x)=2 where
1
y(0)=0 and y(1)=1, taking h=
4
Solution:
Given y(x) y(x)=2
Step1:
Put y(x), y(x) by yi " , yi respectively
i.e, yi " yi 2 -------------------(1)
Step2:

yi

substitute the formula y" ( x)


(1) becomes

yi

2 yi
h2

yi

yi

( 2 h 2 ) yi

yi

yi

( 2 h 2 ) yi

yi

yi

yi

33
yi yi
16
2.0625 yi

yi

2 yi
h2

yi

1
16
1
8

1
8

yi

0.125 ----------------------------(2)

Step3:
The boundary conditions are y0=0, y4=1
x0=0
y0=0

x1= 1

x2= 1

x3= 3

y1=?

y2=?

y3=?

x4=1

h=

1
gives
4

y4=1

Step4
Put i=1, 2, 3 we get the following equations
2.0625 y1 y21 0.125
----------------------------(3)
y1 2.0625 y2 y3 0.125
----------------------------(4)
y2 2.0625 y3
0.875
----------------------------(5)
Solving these three equations we get
y1=0.0451

y2=0.2183

y3=0.5301

--------------------------------------------------------------------------------------------------------1 2
2. Solve the equation y(x) x y(x)=0 where xi=0, , ,given that y(0)+y(0)=1 and y(1)=1.
3 3
Solution:
Step1:
Put y(x), y(x),x by yi " , yi , xi respectively
i.e, yi " xi yi 0 -------------------(1)
Step2:
yi 1 2 yi yi 1
substitute the formula y" ( x)
h2
y
2 yi yi 1
xi yi 0
(1) becomes i 1
h2

yi

1
xi yi
9

yi

0 ---------------------------(2)

Step3:
The conditions are y(0)+y(0)=1 and y(1)=1
x0=0
y0=?

x1= 1

x2= 2

x3=1

y1=?

y2=?

y3=1

Putting i=0, 1, 2 we get


y 1 2 y0 y1 0
-------------------------------(3)
55
y0
y1 y2 0
-------------------------------(4)
27
56
y1
y2 y3 0
-------------------------------(5)
27
Step4:
To solve the above equations we need the values of y-1 and y3
we have y3=1
So now we have to find the value of y-1 by using the condition y(0)+y(0)=1
y(0)+y(0)=1
y0 y0' 1
yi 1 yi 1
y1 y 1
y0
1 [put i=0 in yi'
]
2h
2h
1
put h= we get
3
2 y0 3 y1 2
y1
-----------------------------------------(6)
3
substitute (6) in (3),
2 y0 3 y1 1
-----------------------------------------(7)
55
y0
y1 y2 0 -----------------------------------------(8)
27
56
y1
y2 1 0 -------------------------------------------(9)
27
Solving the above three equations we get
y0=-0.9880

y1=-0.3253

y2=0.3253

----------------------------------------------------------------------------------------------------------Classification of second order PDE


The most general linear pde of second order can be written as
2
2
2
u
u
u
u
u
A 2 B
C 2 D
E
Fu 0 ------------------------(1)
x
x y
y
x
y
where A, B, C, D, E, F are in general functions of x and y.
Equation (1) is said to be
Elliptic
if B2-4AC<0

hyperbolic
Parabolic

if B2-4AC>0
if B2-4AC=0

Note:
The same differential equation may be elliptic in one region, parabolic in another region
and hyperbolic in some other region.
Examples
Elliptic

Parabolic

1. uxx+uyy=0 (Laplace
Equation)
2. uxx+uyy=f(x, y) (Poisson
Equation)

uxx=

1
2

Hyperbolic

ut (One dimensional

heat equation)

uxx=

1
2

utt (One dimensional

wave equation)

Problems based on the classification of 2nd order PDE


1. Classify uxx+4uxy+(x2+4y2)+uyy=sin(x+y)
Solution:
A = 1, B = 4, C = x2+4y2
B2-4AC=4[4-x2-4y2]
x2 y2
x2 y2
The equation is elliptic if 4-x2-4y2<0
1 (outside the ellipse
1)
4
1
4
1
x2 y2
x2 y2
Parabolic 4-x2-4y2=0
1 (on the ellipse
1)
4
1
4
1
x2 y2
x2 y2
Hyperbolic 4-x2-4y2>0
1 (inside the ellipse
1)
4
1
4
1
-----------------------------------------------------------------------------------------------------------2. Classify (x+1)uxx-2(x+2)uxy+(x+3)uyy=0
Solution:
A =(x+1), B=-2(x+2), C=(x+3)
B2-4AC=4>0
hyperbolic.
-----------------------------------------------------------------------------------------------------------One Dimensional Heat Equation
One dimensional heat equation is u
t

u
x2

Classification: Parabolic
There are two types for finding the Finite difference solution of one dimensional heat
flow equation
1
(i)
Bender-Schmidt Method(Explicit Method valid for 0
)
2
(ii)
Crank-Nicholoson Method(Implicit Method)

Bender-Schmidt Method(Explicit Method)


u
Coefficient of
=a, xi=x0+ih and tj=t0+jk
t
Case 1: (If h is given and k is not given)
Assume that =
k= ah2
Find k, using the formula
Using the given (i), (ii) and (iii) conditions fill the first row, first and last column of the table
Next using the Bender-Schmidt recurrence relation,

ui , j

1
ui
2

ui

1, j

1, j

i.e,
ui-1,j

ui+1,j

ui,j+1

a b
2

Case 2: (If h and k are given)


2

k= ah
Find , using the formula
Using the given (i), (ii) and (iii) conditions fill the first row, first and last column of the table
Next using the Bender-Schmidt recurrence relation,

ui , j

ui

1, j

ui

1, j

(1 2 )ui , j

Using the formula and tabulate the remaining rows.


Examples
1. Solve uxx=2ut when u(0, t) = u(4,t)=0 and with initial condition u(x, 0)=x(4-x) upto t=5
sec, assuming x=h=1.
Solution:

u
t
a=2 ; h=1 and k is not given.
Assume that = k= ah2 = 1 2 1 =1
Given: a=Coefficient of

By the given conditions x=0, 1, 2, 3, 4 and t=0, 1, 2, 3, 4, 5

Condition(i) u(0, t)=0 u(0, 0)=0, u(0, 1)=0, u(0, 2)=0, u(0, 3)=0, u(0, 4)=0, u(0, 5)=0
Condition(ii) u(4, t)=0 u(4, 0)=0, u(4, 1)=0, u(4, 2)=0, u(4, 3)=0, u(4, 4)=0, u(4, 5)=0
Condition(iii) u(x, 0)=x(4-x) u(1,0)=1(4-1)=3,
u(2,0)=2(4-2)=4,
u(3,0)=3(4-3)=3,
Bender-Schmidt recurrence relation,

ui , j

1
ui
2

ui

1, j

1, j

i.e,
ui-1,j

ui+1,j

ui,j+1

a b
2

xi=x0+ih and tj=t0+jk


i
0
1
2
3
4
j
x=0 x=1 x=2 x=3 x=4
0 t=0 0
3
4
3
0
1 t=1 0
2
3
2
0
2 t=2 0
1.5 2
1.5 0
3 t=3 0
1
1.5 1
0
4 t=4 0
0.75 1
0.75 0
5 t=5 0
0.5 0.75 0.5 0
-----------------------------------------------------------------------------------------------------------2. Find the values of the function satisfying the pde 4uxx=ut and the boundary conditions
u(0, t) = u(8,t)=0 and u(x, 0)=4x -

x2
2

for points x=0,1, 2, 3, 4, 5, 6, 7, 8, t= j , j=0,1, 2, 3,


8

4, 5
Solution:
u

Given: a=Coefficient of t
a= ; Here h and k is not given. Take h=1
Assume that = k= ah2 = 1 1 1 1
2 4

By the given conditions x=0,1, 2, 3, 4, 5, 6, 7, 8 and


1 2 3 4 5
0, , , , ,
8 8 8 8 8
Condition(ii) u(8, t)=0 t 0, 1 , 2 , 3 , 4 , 5
8 8 8 8 8
2
Condition(iii) u(x, 0)=4x - x
u(1,0)=
2

Condition(i) u(0, t)=0

1 2 3 4 5
0, , , , ,
8 8 8 8 8

4-

1
2

=3.5, u(2,0)= 8 -

4
2

=6, u(3,0)= 12 -

9
2

=7.5,

u(4,0)= 16 -

16 =8,
2

u(5,0)= 20 -

25 =7.5,
2

u(6,0)= 24 -

36 =6,
2

u(7,0)= 28 -

49 =3.5.
2

Bender-Schmidt recurrence relation,

ui , j

1
ui
2

ui

1, j

1, j

i.e,
ui-1,j

ui , j

ui

1, j

ui

1, j

(1 2 )u i , j

ui , j

ui

1, j

ui

1, j

(1 2 )u i , j

ui,j+1

xi=x0+ih and tj=t0+jk


i
0
1
2
j
x=0 x=1
x=2
0 t=0 0
3.5
6
1
1 t=
0
3
5.5
2
3
4
5

8
t= 2
8
3
t=
8
t= 4
8
t= 5
8

ui+1,j

a b
2

3
x=3
7.5
7

4
x=4
8
7.5

5
x=5
7.5
7

6
x=6
6
5.5

7
x=7
3.5
3

8
x=8
0
0

2.75

6.5

6.5

2.75

2.5

4.625

6.5

4.625

2.5

2.3125 4.25

2.125

5.5625 6

3.9875 5.125

5.5625 4.25

5.5625 5.125

2.3125 0

3.9875 2.125

-----------------------------------------------------------------------------------------------------------3. Solve uxx=ut when u(0, t) =0, u(4, t) =0 with initial boundary condition u(x, 0) = x(4-x),
1
h= 1, k= upto t=1 sec.
4
Solution:
Given: a=Coefficient of

u
t

a=1 ; Here h and k are given. Take h=1, k=


=

kh2
a

1
4

1
4

1
2

This is in the interval 0


So we use Bender-Schmidt method
Formula:

ui , j

ui

1, j

ui

1, j

(1 2 )ui , j

put

ui-1,j

ui , j

1
=
4

1
ui
4

ui,j

ui

1, j

ui+1,j

1
ui , j
2

1, j

ui,j+1
i
j
0 t=0
1
1
t=
4
2
1
t=
2
3
3
t=
4
4 t=1

0
1
x=0 x=1
0
3
0
2.5

4
3.5

3
2.5

4
x=4
0
0

2.125

2.125

1.8125 2.5625 1.8125 0

1.5468 2.1875 1.5468 0

boundary condition

2
x=2

1
b d
4

3
x=3

1
c
2

boundary condition

boundary condition

boundary condition

Crank-Nicholsons Method(Implicit Method)


Formula
Case(1) (h is given and k is not given)
k
Assume =1 where
find k using this formula
ah 2

ui , j

1
ui
4

1, j 1

ui

1, j 1

ui

ui

1, j

ui-1,j

1, j

ui+1,j
d

ui-1,j+1

ui,j+1

ui+1,j+1

1
a= (b c d e)
4
using the formula and tabulate the values
Case(2) (h and k are given)
Find the value of

k
ah 2

using the formula

Formula:
(2

2)ui , j

ui

1, j 1

ui

1, j 1

ui

1, j

ui

1, j

(2 2 )ui , j

Using the formula and tabulate the values.


Problems based on Crank- Nicholson Method
1. Solve uxx=2ut when u(0, t) = u(4,t)=0 and with initial condition u(x, 0)=x(4-x) assuming
x=h=1 and compute u for one time step.
Solution:
Here h is given and k is not given.
=1. k= ah2=2

Formula

1
ui 1, j 1 ui 1, j 1 ui 1, j ui 1, j
4
we have to find u11, u21, u31
put j=0,
1
ui ,1
ui 1,1 ui 1,1 ui 1,0 ui 1,0
4
put i=1, 2, 3 we get the following equations
4u11-4u21=4; -u11+4u21-u31=6; 0u11-u21+4u31=4 respectively.
We solve these equations we get the solution
u11=1.571, u21=2.2857, u31=1.571
ui , j

0
1
x=0 x=1
0
3
0
1.571

j
0 t=0
1 t=2

2
x=2

3
x=3

4
x=4
0
0

4
3
2.2857 1.571

One dimensional wave equation


One dimensional wave equation is uxx=

1
2

utt

There are two types


Type1:
Subject to the boundary conditions u(0, t) =0, u(l, t) =0 and the initial condition u(x,0)=f(x) and

ut(x, 0)=0.
Type2:
Subject to the boundary conditions u(0, t) =0, u(l, t) =0 and the initial condition u(x,0)=0 and ut(x, 0)=
f(x)
Formula

ui , j

21

a 2 ui , j

a 2 ui

1, j

ui

1, j

ui , j 1 -------------------------(1)

k
.
h

where

This formula is the explicit scheme for the wave equation.

h
the equation (1) takes the form
a

If k
ui, j

ui

1, j

ui

1, j

ui, j

ui,j-1
d

ui-1,j

ui,j

ui,j+1

ui+1,j

1
ui 1,0 ui
2
of the values of u at time t =0.
This is used to solve the type 1
and ut(x, 0)=0

ui ,1

a=b+c-d

1, 0

which gives the values of u at the first time step in terms

Suppose the problem is given in type2


We write ut(x, 0)=f(x) as ut(x, 0)=f(ih) and substitute in the formula
ui, j 1 ui, j 1
f (ih) for j =0
2k
And then find the value of u11, u21, u31.
The values of Remaining rows are same as type1
----------------------------------------------------------------------------------------------Problems based on one dimensional wave equation
1. Find the nodal values of the equation 16uxx=utt taking x=1 given that u(0, t) =0,
u(5,t)=0, ut(x, 0)=0 and u(x, 0)=x2(5-x) and upto one half of the period of vibration.
Solution:
1
One dimensional wave equation is uxx= 2 utt a=4.
u(5, t)=0

l=5

Period of vibration is =

2l
a

5
=2.5
2

Half period of vibration =1.25


x=1 h=1. Since k is not given we choose k

h
a

1
4

Formula:
ui, j 1
ui

1, j

ui

1, j

ui, j

----------------------------(1)

Conditions

1 1 3
5
1 1 3
5
, , ,1,
u(5,t)=0 for t=0, , , ,1,
4 2 4
4
4 2 4
4
2
u(x,0) = x (5-x) u(1,0)=4,
u(2,0)=12, u(3,0)=18,u(4, 0)=16
ui, j 1 ui, j 1
ut(x, 0) =0
0 for j=0
2k
u i ,1 u i , 1 we find the values of the second using this formula
ui-1,j b
c ui+1,j
u(0,t)=0 for t=0,

ui, j+1

b c
2

The remaining rows obtained by the following formula


ui,j-1
d

ui-1,j

ui,j

ui,j+1

ui+1,j

The values of u at the time steps


i
0
1
2
j
x=0 x=1 x=2
0 t=0 0
4
12
1
1 t=
0
6
11
2
3

4
1
t=
2
t= 3
4

4 t=1
5 t= 5
4

a=b+c-d

3
x=3
18
14

4
x=4
16
9

5
x=5
0
0

-2

-2

-8

-7

0
0

-9
-16

-14
-22

-11
-32

-10
-18

0
0

2. Solve uxx=utt,0<x<2, t>0 subject u(x,0)=0, ut(x, 0)=100(2x-x2), u(0, t) =0, u(2,t)=0.
Choosing h=, compute u for 4 time steps.
Solution
h
1
k
Here h= and a = 1. We choose k
a
2
The simplest explicit scheme is given by
ui, j

ui

ui

1, j

ui, j

1, j

--------------------------------(1)

The boundary conditions are


1
3
u(0,t)=0 for t=0, , 1, , 2
2
2
1
3
u(5,t)=0 for t=0, , 1, , 2
2
2
1
3
u(x,0) = 0 for x =0, , 1, , 2
2
2
ui, j

ut(x, 0)=100(2x-x2)

u i ,1 u i ,

Here h=k=

ui ,

2k

100 i

i2
4

u i ,1 100 i

ui, j

i2
4

100 (2ih i 2 h 2 ) for j =0

for i=1, 2, 3.

for i=1, 2, 3. ---------------------------(2)

From (1) we get ui ,1 ui 1, 0 ui 1,0 ui , 1


Sub (2) in (1) we get
ui 1,0 ui 1,0
i2
ui ,1
50 i
for i =1, 2, 3.
2
4
But u i , 0

0, i

i2
for i =1, 2, 3
4
So we have u11=37.5, u21=50, u31=37.5
ui ,1

50 i

0
x=0

1
x= 1

2
x=1

3
x= 3

0 t=0
1 t= 1

4
x=2

0
0

0
37.5

0
50

0
37.5

0
0

0
0

50
37.5

75
50

50
37.5

0
0

2 t=1
3 t= 3
2

4 t=2

Two dimensional Laplace Equation


Laplace equation is

u=uxx+uyy=0

Standard Five Point Formula(SFPF)


1
ui , j
ui 1, j ui 1, j ui , j 1 ui , j
4
ui,j-1
ui-1,j

ui,j

ui+1,j

ui,j+1

Diagonal Five Point Formula(DFPF)


1
ui , j
ui 1, j ui 1, j ui , j 1 ui , j
4

ui-1,j+1

ui+1,j+1

ui,j
ui-1,j-1
ui+1,j-1
Liebmanns Iteration Process
b13

b12

b11

b10

u1

u2

u3

u4

u5

u6

u7

u8

u9

b9
b8

b14
b15

b7
b6

b16
b1

b2

b3

b4

b5

We compute the initial values of u1, u2, ,u9 by using SFPF and DFPF. First we compute u5 by
SFPF.

1
b7 b15 b11 b3
4
We compute u1, u3, u7, u9 by using DFPF
u5

1
1
b1 u 5 b3 b15
u3
u5
4
4
1
1
u7
b13 u 5 b15 b11
u9
b7
4
4
Finally we compute u2,u4,u6,u8 using SFPF
u1

u2
u6

1
b3 u 5
4
1
u3 u9
4

u1

u3

u4

u5

b7

u8

b5

b3

b11

b11 u 5

b9

1
u1 u 7 b15
4
1
b11 u5 u 7
4

u5
u9

These initial values are called rough values.


After that we use only SFPF method
Once all the values are computed their accuracy can be improved by Gauss-Seidal method.
The Gauss-Seidal formula is given by
1 ( n 1)
u i(,nj 1)
u i 1, j u in 1, j u in, j 1 ui(,nj 11)
4