Answer 4 questions only

a) Use bisection method and Newton-Raphson method to solve the equation: 𝑥 =50 𝑙𝑛𝑥.
Q1

Hint: the solution lies in the interval(1 , 1.25) .

b) {4 iterations are sufficient} Comment on the performance of the two methods.
{10 marks for each method and 5 marks on the comment}. [Q1 total: 25
marks]
Q2
Given the velocity vs. time of an object as follows:

t(s) 0 3 6 9 12 20
v(m/s) 0 0.16 0.55 6.08 20.17 81.25

a) Sketch the relation. [5

marks]
b) Use an exponential least square fit to estimate he velocity at t=8 s and t=15 s. [15
marks]
c) Comment on your results. [5 marks]
[Q2 total: 25 marks]
Q3
x=0.8 x
e cosx
Consider the integral I= ∫ 3
dx
x=0 1+ x
a) Approximate the integral by employing the trapezoidal rule and dividing the interval [0, 0.8]
into N = 4 equally sized intervals. [10 mark]
6
b) Use Romberg technique to reach an accuracyO(h ). [10 mark]
c) Calculate the error knowing that the “exact result” is: 0.9535106. [5 marks]
[Q3 total: 25 marks]
Q4
a) Consider the following system of linear equations:
x+30 y + 4z = 1
25x + 10y -5z = -10
−3x+ 6y + 12z = 3
Is the system diagonally dominant? Use four iterations of Gauss-Seidel method with
initial guess [0 , 0 ,0 ]to get an approximate solution. Calculate the relative error of your
results.
[13 marks]
b) Solve the following differential equation by using Runge- Kutta method over the interval
[ 0 , 0.5 ] with a step size h=0.1 :
dy
=−x y , and y ( 0 )=¿ 1.
dx
2
−x
Compare the solution with the analytical solution: y (x )=e 2 [12 marks]
[
Q4 total: 25 marks]
Q5

[Turn Over (only when instructed to do so)]

 Use two numerical techniques to solve the equation:
dy
=1+ xsiny , 0 ≤ x ≤ 0.3 , y ( 0 )=0.
dx
Sketch the solution. {Choose step sizeh=0.1
{10 marks for each method and 5 marks for the sketch}. [Q5 total: 25 marks]
The Formula Sheet
Newton- Raphson ( f ( x )=0 ¿
f (x n )
x n+1=x n −
f ' (x n )
Lagrange Interpolation

Newton Interpolation

( n+1 )
f (ξ)
Interpolation Error=f ( x ) −P n ( x )=( x−x 0 ) …(x−x n)
( n+1 ) !
Linear Least Square
y=a0 +a 1 x

xi [∑
n ∑ xi
∑ x 2i ][ ] [
a0
a1
= ∑ yi
∑ xi yi ]
Composite Trapezium & Simpson’s Rules for Integration

2 4
Composite Trapezium Error=C h , Simpson ’ s Rules Error=C h .
Simplest Formulae for Romberg Integration (h of LA is double of h of MA)
LA=Less Accurate, M A=More Accurate
4 (MA)−LA
From order of error h2 to order of error h 4 I=
3

16( MA )−LA
From order of error h 4 to order of error h6 I=
15
'
Initial Value Problem y =f x , y , y ( x 0 )= y 0
( )

[Turn Over (only when instructed to do so)]

 2
Euler Method y n +1= y n +h f ( x n , y n ) +O(h )
Runge-Kutta Method of order 2
k 1+ k 2
k 1=f ( x n , y n) ,k 2=f ( x n +h , y n +h k 1 ), y n +1= y n +h +O(h 3)
2

a) Use bisection method and Newton-Raphson method to solve the equation: 𝑥 =50 𝑙𝑛𝑥.
Q1

Hint: the solution lies in the interval(1 , 1.25) .

b) {4 iterations are sufficient} Comment on the performance of the two methods.
{10 marks for each method and 5 marks on the comment}.

f ( x )=x −lnx ¿

Bisection

Newton-Raphson f ( x )=x −lnx

1 x−1
f ' ( x )=1− =
x x
f
x new =x old − ' ( x old )
f

Newton-Raphson is faster but we need to derive a function and be sure that

it is not “numerically
Zero”

[Turn Over (only when instructed to do so)]

Q2
Given the velocity vs. time of an object as follows:

t(s) 0 3 6 9 12 20
v(m/s) 0 0.16 0.55 6.08 20.17 81.25

a) Sketch the relation

b) Use an exponential least square fit to estimate he velocity at t=8 s and t=15 s.
c) Comment on your results.

5
0.37293t
v=0.093104 e
At t=8 , v =1.839332089
At t=15 , v=25.02594548

[Turn Over (only when instructed to do so)]

 t v(Given) v(Calculated) Error Comment
3 0.16 0.285005182 0.125005182 SHAPE
4 4 SEEMS
6 0.55 0.872443259 0.322443259 ACCEPTABLE
4 4 BUT RESULTS
ARE FAR FROM
9 6.08 2.670678596
DATA
ESPECIALY AS
12 20.17 8.175344453 t increases

20 81.25 161.5094301 80.2594301 Very Large

Q3

[Turn Over (only when instructed to do so)]

 x=0.8 x
e cosx
Consider the integral I= ∫ 3
dx
x=0 1+ x
a) Approximate the integral by employing the trapezoidal rule and dividing the interval [0, 0.8]
into N = 4 equally sized intervals.
b) Use Romberg technique to reach an accuracyO(h 6).
c) Calculate the error knowing that the “exact result” is: 0.9535106.

x=0.8 x
e cosx
I= ∫ 3
dx =0.9533550
x=0 1+ x

APPROXIMATION METHOD: Romberg Integration Method

-----------------------ROMBERG INTEGRATION TABLE -----------------
O (h2 ¿ O (h 4 ¿ O (h6 ¿
0.8101982
0.9216636 0.9588187
0.9456882 0.9536964 0.9533550

Error= 0.9535106-0.9533550=.00001556

Q4
a) Consider the following system of linear equations:

[Turn Over (only when instructed to do so)]

 x+30 y + 4z = 1
25x + 10y -5z = -10
−3x+ 6y + 12z = 3
Is the system diagonally dominant? Use four iterations of Gauss-Seidel method with
initial guess [0 , 0 ,0 ]to get an approximate solution. Calculate the error of your results.
[13 marks]
Gauss Seidel

Direct - Method

[ ] [ ]
−0.383 −0.00001875
0.0275 −0.00005625
- 0.140 = .0000265

b) Solve the following differential equation by using Runge- Kutta method over the interval
[ 0 , 0.5 ] with a step size h=0.1 :
dy
=−x y , and y ( 0 )=¿ 1.
dx
2
−x
Compare the solution with the analytical solution: y (x )=e 2 [12 marks]

Exact Solution Runge-Kutta

0.0 1 1
0.1 0.9950124792 0.9950

[Turn Over (only when instructed to do so)]

 0.2 0.9801986733 0.9802
0.3 0.9559974818 0.9560
0.4 0.9231163464 0.9231
0.5 0.8824969026 0.8825

Q5
Use two numerical techniques to solve the equation:
dy
=1+ xsiny , 0 ≤ x ≤ 0.3 , y ( 0 )=0.
dx
Sketch the solution. {Choose step sizeh=0.1 ¿

x Runge-Kutta Euler Solution

0.0 0 0
0.1 0.1003 0.1
0.2 0. 2027 0. 2010
0.3 0.3091 0.3050

Euler is less accurate than R=K method

Error increases as x increases

x More Exact R-K-4 3rd Error Euler !st Error

Method order orderTaylor
Taylor
0 0 0 0 0 0
0.1 0.1003 0.1003 0.0 0.1 0.0
0,2 0.2027 0.2027 0.0 0.201 0.0017
0.3 0.3091 0.0391 0.0 0.0.305 0.0041
0.4 0.4216 0.4216 0.0 0.414 0.0076
0.5 0.5426 0.5425 0.0001 0.5301 0.0125

[Turn Over (only when instructed to do so)]

[Turn Over (only when instructed to do so)]
[Turn Over (only when instructed to do so)]