NUMERICAL METHODS

Numerical methods are systems or algorithms for solving equations that cannot be solved by using
normal techniques or analytical methods.
Examples
1.1. APPROXIMATE INTEGRAL BY SIMPSON’S RULE:
An estimate of the area under the curve y  f  x  can be found by adding the area of a number of
rectangles whose tops closely match the shape of a curve. Simpson’s rule gives a fairly close result. A
good approximation can be obtained by dividing the intervals from a to b in to a number of smaller
intervals and applying the Simpson’s rule to the sub intervals as follows.
b
h
 f  x dx  3  f  x   f  x   4  x
a
0 n 1  x3  x5  ...  2  x2  x4  x6  ... 

h 
b n n

i.e.  f  x dx   f  x0   f  xn   4 x2 n 1   x2 n  ,
a
3 n 0 n 0 
ba
where h  , n  number of ordinates
n 1
Example1
3
1
Find, by Simpson’s rule, an approximate value of
x
0
5
1
dx , using 7 ordinates

Solution:
30
n7h  0.5
7 1
x 0 0.5 1.0 1.5 2.0 2.5 3.0
f  x  1.000 0.970 0.500 0.116 0.030 0.010 0.004
3
1 0.5
x dx  1.00  0.004   4  0.970  0.116  0.010   2  0.500  0.030    1.074915
0
5
1 3 
1.2. APPROXIMATE SOLUTION TO DIFFERENTIAL EQUATIONS
The Taylor’s series can be used to establish an approximate analytical solution to a differential
equation based on the initial conditions provided. Recall that the Taylor’s series is given as

 x  xo  f   xo    x  xo  f   xo   x  xo  f   xo 
2 3

f  x   f  xo     ...
1! 2! 3!
This can be conveniently written in terms of differential coefficients with respect to x as
 x  xo   dy   x  xo   x  xo   d 3 y   ... Where
2 3
 d2y 
f  x  yo      2 
  3 
1!  dx o 2!  dx o 3!  dx o
 dy   d y   d y   d n y  are the respective values of y, y  f  x  at x  xo
2 3
yo ,   , 2 
, 3 
,....  n 
 dx o  dx o  dx o  dx o
Example 2
dy
Solve the differential equation  x 2  y as a power series in ascending powers of x as far as the
dx
term in x . Given that when x  0, y  1 .
3

Solution:
 x  xo   dy   x  xo   x  xo   d 3 y   ...
2 3
 d2y 
f  x   yo      2 
  3 
1!  dx o 2!  dx o 3!  dx o
  dy 
xo  0, yo  1     x o  yo  1   0   1
2 2
But
 dx o
 d2y   d3y  x2 x3
 2 
 2 xo  1  2  0   1  1,  3 
 2  y  1  x  
 dx o  dx o 2 3
Example 3
Use the Taylor series expansion to obtain a solution in a series of ascending powers of x u the
d2y dy dy
 2 x  3 y 2  0 for which y  1, and  0 when x  0
4
term in x of the differential equation 2
dx dx dx
Solution
 x  xo   dy   x  xo   x  xo   d 3 y   ...
2 3
 d2y 
f  x   yo      2 
  3 
1!  dx o 2!  dx o 3!  dx o
 dy   d2y 
yo  1,    0 , xo  0   2   3
 dx o  dx o
d3y d2y dy  d3y 
 2 x  6 y  0   3   0;
dx3 dx 2 dx  dx o
 d4y 
2
d4y d3y d2y d2y  dy 
 2 x  2  6 y  6    0   4   21
dx 4 dx3 dx 2 dx 2  dx   dx o
x  dy  x2  d 2 y  x3  d 3 y  3 2 5 4
f  x   yo      2 
  3 
 ...  y  1  x  x  ...
1!  dx o 2!  dx o 3!  dx o 2 4
Example 4
dy
Given that  3 x  y and that y = 1 when x = 1, find the first three non – zero terms of the
dx
Taylor’s series expansion of y about x = 1
Solution
 x  xo   dy   x  xo   x  xo   d 3 y   ...
2 3
 d2y 
f  x   yo      2    3 
1!  dx o 2!  dx o 3!  dx o
 dy   d2 y   dy 
xo  1, yo  1     3(1)  1  2;  2 
 3   3 2 1
 dx o  dx o  dx o

 x  1
2
x2  2 x  1 1 x2
 y  1   x  1 2   1  1  2 x  2    x
2! 2 2 2
1.3. LINEAR APPROXIMATION FOR FIRST ORDER LINEAR DIFFERENTIAL EQUATIONS
Let the points P  xn , yn  and P  xn1 , yn1  lie on the curve y  f  x  , where xn1  xn  h as shown
below.
Q
P

As h  0, Q  P; PQ  straightline with gradient

 dy  yn 1  yn y  yn  dy   dy 
    n 1  yn 1  yn  h    yn 1  yn  h  
 dx n xn 1  xn h  dx  n  dx  n
 This formula can also be obtained from the Taylor series expansion of y about x  xo
 x  xn   dy   x  xn   x  xn   d 3 y   ...
2 3
 d2y 
i.e y  y      2 
  3 
n
1!  dx  n 2!  dx  n 3!  dx  n
h  dy  h2  d 2 y  h3  d 3 y 
If y  yn 1 , when x  xn  h , then yn 1  y n           ...
1!  dx n 2!  dx 2  n 3!  dx 3  n
2
If h is fairly small, then h and higher powers of h can be neglected therefore
 dy 
yn 1  y n  h  
 dx  n
To improve the accuracy of the approximation for a better approximation, consider three points
P  xn1 , yn1  , Q  xn , yn  and R  xn1 , yn1  on the curve y  f  x  , where
xn 1  xn  xn  xn 1  h as shown below
R

 dy  yn 1  yn 1 yn 1  yn 1 y  yn 1
At Q,    gradient of PR    n 1
 dx n xn 1  xn 1 xn 1  xn  xn  xn 1 2h
yn 1  yn 1  dy   dy 
     yn 1  yn 1  2h  
2h  dx n  dx n
1.4. APPLICATIONS OF THE APPROXIMATIONS FORMULAE
(i) STEP – BY STEP SOLUTION OF DIFFERENTIAL EQUATIONS
This method with the use of the above formulae will provide numerical values of y at regular
intervals along the x – axis and not the analytical solution to the differential equation.
If the given differential equation has the initial values y  y0 when x  x0 then a step–by–step

solution will give a sequence of values for y1 , y2 , y3 ,...., yn corresponding to x1 , x2 , x3 ,...., xn , where
x1  xo  h, x2  xo  2h, x3  xo  3h,.... as shown below
y

y1 y2 y3 y4 ……. ------ yn

o x1 x2 x3 x4 xn x
Example 5
 dy 
Using the approximation yn 1  yn  h   with h  0.1 , find an approximation to the value of y
 dx n
dy
when x  0.3 , given that  2 xy 2 and y  1 when x  0 .
dx
Solution
 dy  dy 
 2 xy 2     2 xn yn 2
dx  dx n
step1: n  0, x  xo  0, y  yo  1, h  1, y1  yo  h  2 x0 y0 2   1   0.1 2  0  1 2
  1
 
step 2 : n  1, x1  0.1, y1  1  y2  y1  h 2 x1 y12  1   0.1 2  0.1 1  0.99
2

step3 : n  2, x2  0.2, y2  0.99

 
 y3  y2  h 2 x2 y2 2  0.99  0.1 2  0.2  0.99   0.9704
2

 
x3  x2  h  0.3, y3  y2  h 2 x2 y2 2  0.99  0.1 2  0.2  0.99   0.9704
2

Example 6
 dy 
Using the approximation yn 1  yn  h   with a step length of 0.1, find an approximation to the
 dx n
dy
value of y when x  0.2 , given that  xy and y  1 when x  0
dx
Solution
 dy   dy 
   xn yn  yn 1  yn  h   yn 1  yn  hxn yn
 dx n  dx n
n  0, x0  0, y0  1, h  0.1  y1  y0  hx0 y0  1  0.1 0 1  1
n  1, x1  0.1, y1  1, h  0.1  y2  y1  hx1 y1  1  0.1 0.11  1.01
n  2, x2  0.2, y2  1.01, h  0.1  y3  y2  hx2 y2  1.01  0.1 0.2 1.01  1.0302

Example 7
4
Find, as a series of ascending powers of x up to and including the term in x an approximate solution
dy
to the differential equation  xy , given that y  1 when x  0 . Use the approximation
dx
 dy 
yn 1  yn 1  2h   with a step length of 0.1, to find an approximation to the value of y when
 dx n
x  0.3 , given that y  1 when x  0
Solution
 dy   dy 
   xn yn  0, y0  1     x0 y0   0 1  0
 dx n  dx 0
 d2 y   dy   d2y   dy 
 2 
 xn    y n ,   2 
 x0    y0   0  0   1  1
 dx  n  dx  n  dx 0  dx 0
 d3 y   d2 y   dy   d3 y   d2y   dy 
 3 
 xn  2 
 2     3 
 x0  2 
 2    0 1  2  0   0
 dx n  dx n  dx n  dx 0  dx 0  dx 0
 d4 y   d2 y   d3 y   d4y   d2 y   d3y 
 4  3 
 2  3   4  3  2    3    31   0   3
 dx n  dx n  dx  n  dx 0  dx 0  dx 0
 x  x0   dy   x  x0   x  x0   d 3 y   ...
2 3
 d2y 
y  yn      2 
  3 
1!  dx  n 2!  dx  n 3!  dx  n
1 2 1 4
 y  1 x  x  ....
2 8
 Using the approximation and differential equation
yn 1  yn 1  2hxn yn ; n  0, y1  y1  2  0.11 0   0  y1  y1
The value of y1 is not given but can be calculated using Taylor series.
1 2 1 4
i.e  y  1  x  x  .... one step is for each x – value. Then one step back ward
2 8
x  x0  h  0  0.1  0.1  y1 is one step backward or in the direction of the negative x – axis.
1 1
 0.1   0.1  1.00505
2 4
Hence when x  0.1, y1  1 
2 8
 y1  y1  2  0.1 0 1  1.00505  y1  y1  1.00505
n  1, x1  xo  h  0.1, yo  1, y1  1.00505  y2  y1  2  0.1 0.11.00505   1.0201
n  2, x2  x1  h  0.2, y2  1.0201, y1  1.00505  y3  1.00505  2  0.1 0.2 1.0201  1.045854
1.5. APPROXIMATION TO SECOND ORDER DIFFERENTIAL EQUATIONS
To estimate the solutions of second order formula, involve a second order differential equation.
Taylor series expansion of f  x  in ascending powers of x  x0 , where x  x0  h . We have
 x  xn   dy   x  xn   x  xn   d 3 y   ...
2 3
 d2y 
yn 1  y n      2    3
1!  dx n 2!  dx  n 3!  dx  n
Putting x  xn  h gives
 dy  h2  d 2 y  h3  d 3 y  h4  d 4 y 
yn 1  y n  h           4  ........................(1)
 dx  n 2!  dx  n 3!  dx  n 4!  dx  n
2 3

Putting x  xn  h , we have

Example 8
d2y dy dy
Given that 4 1  x 2
 2  y  0 and that y  0 and  1 when x  0 , find, as a series of
dx dx dx
4
ascending powers of x, up to and including the term in x , the solution of this differential equation.
Hence, using the approximations
 d2y   dy 
h 2  2   yn 1  2 yn  yn 1 and 2h    yn 1  yn 1 where h  0.2 to find, correct to 3 decimal
 dx n  dx n
places an approximate value of y when x  0.6 .
Solution
d2y dy
4 1  x  2
2 20
dx dx
 dy   d2y  1
y0  0,    1, x0  0   2   
 dx 0  dx 0 2
d3y 6 d2y 1 dy  d3y  1
     3 
dx 3
4(1  x) dx 2
4(1  x) dx  dx 0 2
 d4y 10 d 3 y 1 d2y  d4y  9 y  x
1 2 1 3 3 4
x  x  x  .....
     4  
dx 4 4(1  x) dx 3 4(1  x) dx 2  dx 0 8 4 12 64
yn 1  h  4  xn   8 yn 1  xn   yn h 2 3.8 y1
yn1  ; n  0, x0  0, y0  0, h  0.2  y1 
4  4 xn  h 4.2
 0.2  0.2 3  0.2 
2 3 4
3.8
y1   0.2      0.1907  y1   0.1907   0.1726
4 12 64 4.2
n  1, x1  0.2, y1  0.1726
y0  h  4  4 x1   8 y1 1  x1   y1 h 2 8  0.1726 1  0.2   0.1726  0.2 
2
1.6499
 y2     0.33
4  4 x1  h 4  4  0.2   0.2 5
n  2, x2  0.4, y2  0.33, h  0.2
y1  h  4  4 x2   8 y2 1  x2   y2 h2 0.1726  0.2  4  4  0.4    8  0.331  0.4   0.33  0.4 
2

 y3  
4  4 x2  h 4  4  0.4   0.2
4.5752
  0.7888 n  3, x3  0.6  y  y3  0.7888
5.8