MA3251- Statistics and Numerical Methods

UNIT III
SOLUTION OF EQUATIONS AND EIGENVALUE PROBLEMS
Solution of algebraic and transcendental equations - Fixed point iteration method – Newton
Raphson method- Solution of linear system of equations - Gauss elimination method –
Pivoting - Gauss Jordan method – Iterative methods of Gauss Jacobi and Gauss Seidel –
Eigen values of a matrix by Power method and Jacobi’s method for symmetric matrices.
Solution of Algebraic and Transcendental Equations
Fixed Point Iteration Method
Fixed Point Theorem:
Let 𝑓 𝑥 = 0 and let 𝑥 = 𝜀 ∈ 𝐼 be a root of an equation 𝑓 𝑥 = 0 . Rewrite the equation
𝑓 𝑥 = 0 as 𝑥 = 𝑔 𝑥 , where 𝑔(𝑥) and 𝑔′ (𝑥) must be continuous in 𝐼 . If 𝑔′ (𝑥) < 1,
∀ 𝑥 ∈ 𝐼, then the sequence of approximations 𝑥0 , 𝑥1 , 𝑥2 , ........ 𝑥𝑛 given by 𝑥𝑛+1 = 𝑔(𝑥𝑛 )
converge to the root "𝜀", provided that the initial approximation 𝑥0 is chosen in 𝐼 .
Note:
1. A point "𝑥" is said to be fixed of the equation 𝑓 𝑥 = 0 if 𝑓 𝑥 = 𝑥.
2. The sufficient condition for the convergence of the sequence of approximations in the
fixed point iteration method is “ 𝒙 = 𝒈(𝒙) such that 𝒈′ (𝒙) < 1, ∀ 𝒙 ∈ 𝑰 = [𝒂, 𝒃]”.
Procedure to find the solution of the equation 𝒇 𝒙 = 𝟎:
Given that , 𝑓 𝑥 = 0..................(1)
We rewrite the equation (1) as 𝑥 = 𝑔(𝑥)...............(2) such that 𝑔′ (𝑥) < 1, ∀ 𝑥 ∈ 𝐼 = [𝑎, 𝑏].
Let 𝑥0 be an initial root of the Eqn(1).
Then, the first approximation is given by
𝑥1 = 𝑔(𝑥0 )
The successive approximations are given by
𝑥2 = 𝑔(𝑥1 )
𝑥3 = 𝑔(𝑥2 )
⋮
𝑥𝑛 = 𝑔(𝑥𝑛−1 )
∴ In general, we can give the iteration process as: 𝑥𝑛+1 = 𝑔 𝑥𝑛 ;𝑛 = 0,1,2, … …
Note:
1. The order of convergence of the iteration method is "1".
(i.e.) the order of convergence is linear.
2. Order of Convergence of an Iterative Scheme:

Dr.V.Nirmala, AP/Maths, UCET Page 1

Let the sequence of iterative values {𝑥𝑛 }∞
𝑛=0 converges to "𝑠".

Also, let 𝜀𝑛 = 𝑠 − 𝑥𝑛 and 𝜀𝑛+1 = 𝑠 − 𝑥𝑛+1 for 𝑛 ≥ 0 are the errors at 𝑛𝑡𝑕 and (𝑛 + 1)𝑡𝑕
iterations respectively.
If two positive constants 𝐴 ≠ 0 and 𝑅 > 0 exist, and
𝑠−𝑥 𝑛 +1 𝜀 𝑛 +1
lim𝑛→∞ = lim𝑛→∞ = 𝐴 , then the sequence is said to converge to "𝑠" with order
𝑠−𝑥 𝑛 𝑅 𝜀𝑛 𝑅

of convergence 𝑅. The number 𝐴 is called the asymptotic error constant.

3. Let 𝑋𝑛+1 = 𝑔(𝑋𝑛 ) define an iterative method, and let "𝑠" and 𝑥𝑛 respectively are the exact
and approximate solutions of 𝑥 = 𝑔 𝑥 . Then 𝑥𝑛 = 𝑠 + 𝜀𝑛 , where 𝜀𝑛 is the error in
𝑥𝑛 . Suppose that 𝑔 is differentiable any number of times, then from Taylor's formula
1
𝑥𝑛+1 = 𝑔 𝑥𝑛 = 𝑔 𝑠 + 𝜀𝑛 = 𝑔 𝑠 + 𝜀𝑛 𝑔′ 𝑠 + 2 𝜀𝑛 2 𝑔′′ 𝑠 + ⋯ … .. ............

The exponent of 𝜀𝑛 in the first non-vanishing term after 𝑔 𝑠 is called the order of the
iteration process defined by 𝑔.
4. Since the convergence of this scheme depends on the choice of 𝑔 𝑥 and the only
information available about 𝑔′ 𝑥 is 𝑔′ (𝑥) < 1 in some interval which brackets the root,
𝑔′ 𝑥 at 𝑥 = 𝑠 may or may not be zero. That is the order of fixed point iterative scheme is
only one.
Problems:
1. Find the root of 𝑒 𝑥 − 3𝑥 = 0 by the fixed point iteration method.
Solution:
Let 𝑓 𝑥 = 𝑒 𝑥 − 3𝑥 = 0 .............(1)
Now, 𝑓 0 = 𝑒 0 − 0 =
𝑓 1 = 𝑒1 − 3 =
∴ The root lies in the interval 𝐼 =
𝑒𝑥
Rewriting the Eqn(1), 𝑥 = = 𝑔(𝑥)...................(2)
3
𝑒𝑥
Now, 𝑔′ 𝑥 = 3

Dr.V.Nirmala, AP/Maths, UCET Page 2

Dr.V.Nirmala, AP/Maths, UCET Page 3
2. Find the root of 𝑥 3 − 2𝑥 2 − 5 = 0 by the fixed point iteration method.

Dr.V.Nirmala, AP/Maths, UCET Page 4

3. Find the root of 𝑥 3 + 𝑥 2 − 1 = 0 by the fixed point iteration method.

Dr.V.Nirmala, AP/Maths, UCET Page 5

4. Find the root of 3𝑥 = 1 + 𝑠𝑖𝑛𝑥 by the fixed point iteration method. {Change into radian
mode}

Dr.V.Nirmala, AP/Maths, UCET Page 6

5. Find the negative root of the equation 𝑥 3 − 2𝑥 + 5 = 0 .
Note: The negative root of 𝑓(𝑥) is = −{ + 𝑟𝑜𝑜𝑡 𝑜𝑓 𝑓 −𝑥 }.

Dr.V.Nirmala, AP/Maths, UCET Page 7

6. Find the root of 3𝑥 − log10 𝑥 = 6 by the fixed point iteration method.
Note:
log 𝑏 𝑎 = log 𝑐 𝑎 . log 𝑏 𝑐
⇒ log 𝑒 𝑥 = log10 𝑥 . log 𝑒 10
log 𝑒 𝑥
⇒ log10 𝑥 = log
𝑒 10

Solution:

Dr.V.Nirmala, AP/Maths, UCET Page 8

7. Find the cube root of “15”, correct to four decimal places by fixed point method.
Solution:
3
Let 𝑥 = 15

Dr.V.Nirmala, AP/Maths, UCET Page 9

8. Find the root of the equation 3𝑥 − 2 = 𝑐𝑜𝑠𝑥 by the fixed point iteration method.

Dr.V.Nirmala, AP/Maths, UCET Page 10

9. Find the root of the equation 2𝑥 − 3 = 𝑐𝑜𝑠𝑥 by the fixed point iteration method.

Dr.V.Nirmala, AP/Maths, UCET Page 11

10. Find the root of the equation 2𝑥 − log10 𝑥 = 7 by the fixed point iteration method.

Dr.V.Nirmala, AP/Maths, UCET Page 12

Newton’s Method (Newton Raphson Method)
Let 𝑥0 be an approximate root of the equation 𝑓 𝑥 = 0.The first approximation is given by
𝑓(𝑥 )
𝑥1 = 𝑥0 − 𝑓 ′ (𝑥0 ) , if 𝑓 ′ (𝑥0 ) ≠ 0.
0

Similarly,
𝑓(𝑥 )
𝑥2 = 𝑥1 − 𝑓 ′ (𝑥1 ) , if 𝑓 ′ (𝑥1 ) ≠ 0.
1

...........................
𝑓(𝑥 )
𝑥𝑛+1 = 𝑥𝑛 − 𝑓 ′ (𝑥𝑛 ) , if 𝑓 ′ (𝑥𝑛 ) ≠ 0, 𝑛 = 0,1,2, … ….
𝑛

Note:
2
1. The condition for the convergence of Newton Raphson method is 𝑓 𝑥 𝑓 ′′ (𝑥) < 𝑓 ′ (𝑥) .
2. The order of convergence of Newton’s method is two and the rate of convergence is
quadratic
Problems:
1. Find the smallest positive root of the equation 3𝑥 3 − 9𝑥 2 + 8 = 0, correct to 4 places of
decimals, using Newton-Raphson method.
Solution:
Given, 𝑓 𝑥 = 3𝑥 3 − 9𝑥 2 + 8 = 0

Dr.V.Nirmala, AP/Maths, UCET Page 13

Dr.V.Nirmala, AP/Maths, UCET Page 14
2. Find the root of the equation 𝑥 4 − 𝑥 − 10 = 0 by the Newton-Raphson method.

Dr.V.Nirmala, AP/Maths, UCET Page 15

Dr.V.Nirmala, AP/Maths, UCET Page 16
3. Find the root of the equation 𝑥 4 + 12𝑥 + 7 = 0, lying between " − 2" and " − 3",correct
to 4 places of decimals, using Newton Raphson method.

Dr.V.Nirmala, AP/Maths, UCET Page 17

4. Find the root of the equation 𝑥 2 + 4𝑠𝑖𝑛𝑥 = 0 that lies between -1 and -2 by the Newton-
Raphson method.

Dr.V.Nirmala, AP/Maths, UCET Page 18

 1 𝑁
5. Using Newton-Raphson method establish the formula 𝑥𝑛 +1 = 2 𝑥𝑛 + 𝑥 to calculate the
𝑛

square root of "𝑁" and hence find 5 correct to four places of decimals.
Solution:
Let 𝑥 = 𝑁 ⇒ 𝑥 2 = 𝑁
Let 𝑓 𝑥 = 𝑥 2 − 𝑁 = 0
Then, 𝑓 ′ 𝑥 = 2𝑥.
Now, the Newton’s iterative formula is:
𝑓(𝑥 )
𝑥𝑛+1 = 𝑥𝑛 − 𝑓 ′ (𝑥𝑛
𝑛 )

𝑥 𝑛 2 −𝑁
⇒𝑥𝑛+1 = 𝑥𝑛 − 2𝑥 𝑛

2𝑥 𝑛 2 −𝑥 𝑛 2 +𝑁
⇒𝑥𝑛+1 = 2𝑥 𝑛

𝑥𝑛 2 +𝑁
⇒𝑥𝑛+1 = 2𝑥 𝑛
1 𝑁
⇒𝑥𝑛+1 = 2 𝑥𝑛 + 𝑥 ...............(1)
𝑛

Dr.V.Nirmala, AP/Maths, UCET Page 19

 3
6. Find 17

Dr.V.Nirmala, AP/Maths, UCET Page 20

 1
7. Show that the iterative formula for finding the reciprocal of 𝑁 is 𝑥𝑛+1 = 𝑥𝑛 (2 − 𝑁𝑥𝑛 ) and
1
hence find the value of .
31

Solution:
1 1
Let 𝑥 = 𝑁 ⇒ 𝑥 = 𝑁
1
Let 𝑓 𝑥 = 𝑥 − 𝑁 = 0
1
Then, 𝑓 ′ 𝑥 = − 𝑥 2 .

Now, the Newton’s iterative formula is:

𝑓(𝑥 )
𝑥𝑛+1 = 𝑥𝑛 − 𝑓 ′ (𝑥𝑛
𝑛 )
1
−𝑁
𝑥𝑛
⇒𝑥𝑛+1 = 𝑥𝑛 − 1
−
𝑥𝑛 2

⇒𝑥𝑛+1 = 𝑥𝑛 + (𝑥𝑛 − 𝑥𝑛 2 𝑁)
⇒𝑥𝑛+1 = 2𝑥𝑛 − 𝑥𝑛 2 𝑁
⇒𝑥𝑛+1 = 𝑥𝑛 (2 − 𝑁𝑥𝑛 )...............(1)

Dr.V.Nirmala, AP/Maths, UCET Page 21

8. Find the root of the equation 𝑒 𝑥 − 𝑥 3 − 𝑐𝑜𝑠25𝑥 = 0 nearer to 𝑥 = 4.5 (correct to 3
decimal places).

Dr.V.Nirmala, AP/Maths, UCET Page 22

Solution of simultaneous linear system of equations
Direct methods:
In these methods the unknowns will be eliminated one by one and hence the given system of
equations will be transformed into an equivalent upper triangular system.
Gauss Elimination method:
Gauss elimination method is
1. A direct method
2. The co efficient matrix is transformed into upper triangular matrix.
3. The solution is obtained by back substitution method.
1. Solve the following system of equations by Gauss elimination method:
2𝑥 + 3𝑦 − 𝑧 = 5;4𝑥 + 4𝑦 − 3𝑧 = 3; 2𝑥 − 3𝑦 + 2𝑧 = 2.
Solution:
2 3 −1 : 5 2 3 −1 : 5 𝑅1 → 𝑅1
4 4 −3 : 3 ≅ 0 −2 −1 : −7 𝑅2 → 𝑅2 − 2 𝑅1
2 −3 2 : 2 0 −6 3 : −3 𝑅3 → 𝑅3 − 𝑅1
2 3 −1 : 5 𝑅1 → 𝑅1
≅ 0 −2 −1 : −7 𝑅2 → 𝑅2
0 0 6 : 18 𝑅3 → 𝑅3 − 3𝑅2
3rd row ⇒ 6𝑧 = 18 ⇒ 𝑧 = 3.
2nd row ⇒ −2𝑦 − 𝑧 = −7 ⇒ −2𝑦 − 3 = −7 ⇒ 𝑦 = 2
1st row ⇒ 2𝑥 + 3𝑦 − 𝑧 = 5 ⇒ 2𝑥 + 6 − 3 = 5 ⇒ 𝑥 = 1.
∴ The solution is {𝑥 = 1, 𝑦 = 2, 𝑧 = 3}.
2. Solve : 𝑥 + 2𝑦 + 𝑧 = 3; 2𝑥 + 3𝑦 + 3𝑧 = 10;3𝑥 − 𝑦 + 2𝑧 = 13.

Dr.V.Nirmala, AP/Maths, UCET Page 23

3. Solve:2𝑥 − 𝑦 + 3𝑧 = 9;𝑥 + 𝑦 + 𝑧 = 6; 𝑥 − 𝑦 + 𝑧 = 2.

4. Solve:3𝑥 + 4𝑦 + 5𝑧 = 18;2𝑥 − 𝑦 + 8𝑧 = 13; 5𝑥 − 2𝑦 + 7𝑧 = 20.

Dr.V.Nirmala, AP/Maths, UCET Page 24

5. Solve:𝑥 + 𝑦 + 𝑧 + 𝑤 = 2;𝑥 + 𝑦 + 3𝑧 − 2𝑤 = −6;
2𝑥 + 3𝑦 − 𝑧 + 2𝑤 = 7;𝑥 + 2𝑦 + 𝑧 − 𝑤 = −2.

Dr.V.Nirmala, AP/Maths, UCET Page 25

Gauss -Jordan method:
Gauss -Jordan method is
1. A direct method
2. The co efficient matrix is transformed into a diagonal matrix.
3. The solution is obtained by direct substitution method.
1. Solve the following system of equations by Gauss Jordan method:
2𝑥 + 3𝑦 − 𝑧 = 5;4𝑥 + 4𝑦 − 3𝑧 = 3; 2𝑥 − 3𝑦 + 2𝑧 = 2.
2 3 −1 : 5 2 3 −1 : 5 𝑅1 → 𝑅1
4 4 −3 : 3 ≅ 0 −2 −1 : −7 𝑅2 → 𝑅2 − 2 𝑅1
2 −3 2 : 2 0 −6 3 : −3 𝑅3 → 𝑅3 − 𝑅1
2 3 −1 : 5 𝑅1 → 𝑅1
≅ 0 −2 −1 : −7 𝑅2 → 𝑅2
0 0 6 : 18 𝑅3 → 𝑅3 − 3𝑅2
2 3 −1 : 5 𝑅1 → 𝑅1
≅ 0 −2 −1 : −7 𝑅2 → 𝑅2
0 0 1 : 3 𝑅3 → 𝑅3 /6
2 3 0 : 8 𝑅1 → 𝑅1 + 𝑅3
≅ 0 −2 0 : −4 𝑅2 → 𝑅2 + 𝑅3
0 0 1 : 3 𝑅3 → 𝑅3
4 0 0 : 4 𝑅1 → 2𝑅1 + 3𝑅2
≅ 0 −2 0 : −4 𝑅2 → 𝑅2
0 0 1 : 3 𝑅3 → 𝑅3
1st row ⇒ 𝑥 = 1.
2nd row ⇒ 𝑦 = 2
3rd row ⇒ 𝑧 = 3.
∴ The solution is {𝑥 = 1, 𝑦 = 2, 𝑧 = 3}.
2. Solve: 𝑥 + 2𝑦 + 𝑧 = 3; 2𝑥 + 3𝑦 + 3𝑧 = 10;3𝑥 − 𝑦 + 2𝑧 = 13.

Dr.V.Nirmala, AP/Maths, UCET Page 26

3. Solve:2𝑥 − 𝑦 + 3𝑧 = 9;𝑥 + 𝑦 + 𝑧 = 6; 𝑥 − 𝑦 + 𝑧 = 2.

Dr.V.Nirmala, AP/Maths, UCET Page 27

4. Solve:3𝑥 + 4𝑦 + 5𝑧 = 18;2𝑥 − 𝑦 + 8𝑧 = 13; 5𝑥 − 2𝑦 + 7𝑧 = 20.

Dr.V.Nirmala, AP/Maths, UCET Page 28

5. Solve: 𝑥 + 𝑦 + 𝑧 + 𝑤 = 2;𝑥 + 𝑦 + 3𝑧 − 2𝑤 = −6;
2𝑥 + 3𝑦 − 𝑧 + 2𝑤 = 7;𝑥 + 2𝑦 + 𝑧 − 𝑤 = −2

Dr.V.Nirmala, AP/Maths, UCET Page 29

6. Solve the following system of equations by Gauss elimination & Gauss Jordan method:
5𝑥 − 𝑦 + 𝑧 = 10 ; 2𝑥 + 4𝑦 = 12; 𝑥 + 𝑦 + 5𝑧 = −1.

Dr.V.Nirmala, AP/Maths, UCET Page 30

Dr.V.Nirmala, AP/Maths, UCET Page 31
7. Solve the following system of equations by Gauss elimination & Gauss Jordan method:
5𝑥 − 𝑦 = 9 ; −𝑥 + 5𝑦 − 𝑧 = 4; 𝑦 + 5𝑧 = −6.

Dr.V.Nirmala, AP/Maths, UCET Page 32

Dr.V.Nirmala, AP/Maths, UCET Page 33
Indirect Methods (Iterative methods)
In these methods the solution of the system of given equations is obtained by successive
approximations.
Gauss-Jacobi Method
Consider a system of three equations with three unknowns:
𝑎11 𝑥 + 𝑎12 𝑦 + 𝑎13 𝑧 = 𝑏1 ..............(1)
𝑎21 𝑥 + 𝑎22 𝑦 + 𝑎23 𝑧 = 𝑏2 ..............(2)
𝑎31 𝑥 + 𝑎32 𝑦 + 𝑎33 𝑧 = 𝑏3 ..............(3)
Suppose that the above system is diagonally dominant.
Then,
1
Eqn(1) ⇒ 𝑥 = 𝑎 𝑏1 − 𝑎12 𝑦 − 𝑎13 𝑧
11

1
Eqn(2) ⇒ 𝑦 = 𝑎 𝑏2 − 𝑎21 𝑥 − 𝑎23 𝑧
22

1
Eqn(3) ⇒ 𝑧 = 𝑏3 − 𝑎31 𝑥 − 𝑎32 𝑦
𝑎 33

Let 𝑥 (0) , 𝑦 (0) and 𝑧 (0) be the initial values.

Then the next approximation is given by
1
𝑥 (1) = 𝑎 𝑏1 − 𝑎12 𝑦 (0) − 𝑎13 𝑧 (0)
11

1
𝑦 (1) = 𝑎 𝑏2 − 𝑎21 𝑥 (0) − 𝑎23 𝑧 (0)
22

1
𝑧 (1) = 𝑎 𝑏3 − 𝑎31 𝑥 (0) − 𝑎32 𝑦 (0)
33

.............
In general
1
𝑥 (𝑛+1) = 𝑎 𝑏1 − 𝑎12 𝑦 (𝑛) − 𝑎13 𝑧 (𝑛 )
11

1
𝑦 (𝑛+1) = 𝑎 𝑏2 − 𝑎21 𝑥 (𝑛) − 𝑎23 𝑧 (𝑛)
22

1
𝑧 (𝑛+1) = 𝑎 𝑏3 − 𝑎31 𝑥 (𝑛) − 𝑎32 𝑦 (𝑛)
33

Note:
1. The iteration method can be applied only if the given system is diagonally dominant.
2. An 𝑛 ⨯ 𝑛 matrix 𝐴 is said to be diagonally dominant if the absolute value of each leading
diagonal element is greater than or equal to the sum of the absolute values of the remaining
elements in that row.
3. 𝑎11 𝑥 + 𝑎12 𝑦 + 𝑎13 𝑧 = 𝑏1 ;𝑎21 𝑥 + 𝑎22 𝑦 + 𝑎23 𝑧 = 𝑏2 ; 𝑎31 𝑥 + 𝑎32 𝑦 + 𝑎33 𝑧 = 𝑏3 .
This system is said to be diagonally dominant if

Dr.V.Nirmala, AP/Maths, UCET Page 34

𝑎11 ≥ 𝑎12 + 𝑎13 ;
𝑎22 ≥ 𝑎21 + 𝑎23 ;
𝑎33 ≥ 𝑎31 + 𝑎32 .
4. If the given matrix is not diagonally dominant, we re-arrange the equations in such a way
that is diagonally dominant.
5. The process of iteration converges quickly if the co- efficient matrix is diagonally
dominant.
1. Solve the following system of equations by Gauss-Jacobi Method
8𝑥 − 𝑦 + 𝑧 = 18;2𝑥 + 5𝑦 − 2𝑧 = 3;𝑥 + 𝑦 − 3𝑧 = 6.
Solution:

Dr.V.Nirmala, AP/Maths, UCET Page 35

Dr.V.Nirmala, AP/Maths, UCET Page 36
Gauss-Seidal method:
Consider a system of three equations with three unknowns:
𝑎11 𝑥 + 𝑎12 𝑦 + 𝑎13 𝑧 = 𝑏1 ..............(1)
𝑎21 𝑥 + 𝑎22 𝑦 + 𝑎23 𝑧 = 𝑏2 ..............(2)
𝑎31 𝑥 + 𝑎32 𝑦 + 𝑎33 𝑧 = 𝑏3 ..............(3)
Suppose that the above system is diagonally dominant.
Then,
1
Eqn(1) ⇒ 𝑥 = 𝑎 𝑏1 − 𝑎12 𝑦 − 𝑎13 𝑧
11

1
Eqn(2) ⇒ 𝑦 = 𝑎 𝑏2 − 𝑎21 𝑥 − 𝑎23 𝑧
22

1
Eqn(3) ⇒ 𝑧 = 𝑎 𝑏3 − 𝑎31 𝑥 − 𝑎32 𝑦
33

Let 𝑦 (0) and 𝑧 (0) be the initial values.

Then the next approximation is given by
1
𝑥 (1) = 𝑎 𝑏1 − 𝑎12 𝑦 (0) − 𝑎13 𝑧 (0)
11

1
𝑦 (1) = 𝑎 𝑏2 − 𝑎21 𝑥 (1) − 𝑎23 𝑧 (0)
22

1
𝑧 (1) = 𝑎 𝑏3 − 𝑎31 𝑥 (1) − 𝑎32 𝑦 (1)
33

The second approximation is given by

1
𝑥 (2) = 𝑎 𝑏1 − 𝑎12 𝑦 (1) − 𝑎13 𝑧 (1)
11

1
𝑦 (2) = 𝑎 𝑏2 − 𝑎21 𝑥 (2) − 𝑎23 𝑧 (1)
22

1
𝑧 (2) = 𝑎 𝑏3 − 𝑎31 𝑥 (2) − 𝑎32 𝑦 (2)
33

.
.
.
In general
1
𝑥 (𝑛+1) = 𝑎 𝑏1 − 𝑎12 𝑦 (𝑛) − 𝑎13 𝑧 (𝑛 )
11

1
𝑦 (𝑛+1) = 𝑎 𝑏2 − 𝑎21 𝑥 (𝑛+1) − 𝑎23 𝑧 (𝑛)
22

1
𝑧 (𝑛+1) = 𝑎 𝑏3 − 𝑎31 𝑥 (𝑛+1) − 𝑎32 𝑦 (𝑛+1)
33

Dr.V.Nirmala, AP/Maths, UCET Page 37

1. Solve the following system of equations by Gauss-Seidal Method:
8𝑥 − 𝑦 + 𝑧 = 18;2𝑥 + 5𝑦 − 2𝑧 = 3;𝑥 + 𝑦 − 3𝑧 = 6.

Dr.V.Nirmala, AP/Maths, UCET Page 38

Dr.V.Nirmala, AP/Maths, UCET Page 39
2. 27𝑥 + 6𝑦 − 𝑧 = 85;6𝑥 + 15𝑦 + 2𝑧 = 72; 𝑥 + 𝑦 + 54𝑧 = 110.

Dr.V.Nirmala, AP/Maths, UCET Page 40

Dr.V.Nirmala, AP/Maths, UCET Page 41
3. 4𝑥 − 10𝑦 + 3𝑧 = −3;𝑥 + 6𝑦 + 10𝑧 = −3; 10𝑥 − 5𝑦 − 2𝑧 = 3.

Dr.V.Nirmala, AP/Maths, UCET Page 42

Dr.V.Nirmala, AP/Maths, UCET Page 43
Eigen Value of a Matrix by Power Method
Power Method
This method is used to find the numerically largest Eigen value (called the dominant Eigen
value) and the corresponding Eigen vector of a square matrix A.
The method can be applied only if the Eigen values are real and the corresponding Eigen
vectors are linearly independent.
Working Rule:
Let 𝐴 be the given square matrix. Let 𝜆 be an Eigen value of 𝐴. ⇒ 𝐴𝜆 = 𝜆𝑋.
To find 𝜆:
Let 𝑋 (0) be the initial Eigen Vector.
Then the next approximation is given by
𝐴𝑋 (0) = 𝑌 (0) = 𝜆1 𝑋 (1)
𝐴𝑋 (1) = 𝑌 (1) = 𝜆2 𝑋 (2)
.......
𝐴𝑋 (𝑛+1) = 𝑌 (𝑛 +1) = 𝜆𝑛 𝑋 (𝑛) ,where 𝜆𝑖 denotes the numerically largest component of the
vector 𝑌 (𝑖+1) .
The process will be continued until 𝜆𝑖 s are repeated.
Generally the initial Eigen vector 𝑋 (0) is taken as a vector with all components equal to “1”.
1 1
Note: 1. The initial Eigen vector 𝑋 (0) may taken as follows: 𝑋 (0) = 0 or 𝑋 (0) = 1 or
0 0
1
𝑋 (0) = 1 etc.
1
2. To find the numerically smallest Eigen value of 𝐴, we obtain the dominant Eigen value of
𝐴 and also of the matrix 𝐵 = 𝐴 − 𝜆𝐼. Then,
the smallest value of 𝐴 = 𝑡𝑕𝑒 𝑑𝑜𝑚𝑖𝑛𝑎𝑛𝑡 𝐸𝑖𝑔𝑒𝑛 𝑣𝑎𝑙𝑢𝑒 𝑜𝑓 𝐵 + 𝜆.
3 −5
1. Find the numerically larger Eigen value of the matrix 𝐴 = by power method.
−2 4

Dr.V.Nirmala, AP/Maths, UCET Page 44

 1 −3 2
2. Find the numerically largest Eigen value of the matrix 𝐴 = 4 4 −1 by power
6 3 5
method.

Dr.V.Nirmala, AP/Maths, UCET Page 45

 1 6 1
3. Find the numerically largest Eigen value of the matrix 𝐴 = 1 2 0 by power method.
0 0 3

Dr.V.Nirmala, AP/Maths, UCET Page 46

 1 1 3
4. Find the numerically largest Eigen value of the matrix 𝐴 = 1 5 1 by power method.
3 1 1

Dr.V.Nirmala, AP/Maths, UCET Page 47

 4 2 2
5. Find the numerically larger Eigen value of the matrix 𝐴 = 2 5 1 by power method.
2 1 6

Dr.V.Nirmala, AP/Maths, UCET Page 48

 2 −1 0
6. Find the numerically larger Eigen value of the matrix 𝐴 = −1 2 −1 by power
0 −1 2
method. Take [1,0,0]𝑇 as the initial eigenvector.

Dr.V.Nirmala, AP/Maths, UCET Page 49

 5 0 1
7. Find the numerically larger Eigen value of the matrix 𝐴 = 0 −2 0 by power
1 0 5
method.

Dr.V.Nirmala, AP/Maths, UCET Page 50

 −15 4 3
6. Find the numerically smallest Eigen value of 𝐴 = 10 −12 6 if the numerically
20 −4 2
dominant Eigen value of 𝐴 is (−20).

Dr.V.Nirmala, AP/Maths, UCET Page 51

Jacobi’s method to find Eigen values
This method can be used to find all the Eigen values of only a symmetric matrix.
Procedure:
𝑎11 𝑎12 𝑎13
Consider a symmetric matrix 𝐴 = 𝑎21 𝑎22 𝑎23 .
𝑎31 𝑎32 𝑎33
Suppose that 𝑎𝑖𝑘 is the numerically largest off-diagonal element of the given matrix "𝐴".
1.Then chose a matrix 𝑆1 whose(𝑖, 𝑖), (𝑖, 𝑘), (𝑘, 𝑖)and (𝑘, 𝑘) positions elements are 𝑐𝑜𝑠𝜃 ,
−𝑠𝑖𝑛𝜃,𝑠𝑖𝑛𝜃 and 𝑐𝑜𝑠𝜃 respectively, where "𝜃" is found out by
1 2𝑎 𝑖𝑘
(a) When 𝑎𝑖𝑖 ≠ 𝑎𝑖𝑘 , 𝜃 = 2 𝑡𝑎𝑛−1 𝑎 𝑖𝑖 −𝑎 𝑘𝑘
𝜋
− 4 , 𝑖𝑓 𝑎𝑖𝑘 < 0
(b) When 𝑎𝑖𝑖 = 𝑎𝑖𝑘 , 𝜃 = 𝜋
, 𝑖𝑓 𝑎𝑖𝑘 > 0
4

2. All other elements of 𝑆1 are similar to those (either1 or 0) of a unit matrix.

3. Find a new matrix 𝐵1 = 𝑆1𝑇 𝐴𝑆1 .
4. Repeat step 1 to 3 for 𝐵1 and find 𝑆2 , so that we get 𝐵2 = 𝑆2𝑇 𝐵1 𝑆2 = 𝑆2𝑇 𝑆1𝑇 𝐴𝑆1 𝑆2 .
𝑇
5. Proceeding similarly, 𝐵𝑛 = 𝑆𝑛𝑇 𝑆𝑛−1 … . 𝑆2𝑇 𝑆1𝑇 𝐴𝑆1 𝑆2 … . 𝑆𝑛−1 𝑆𝑛 = 𝑆 𝑇 𝐴𝑆 , where 𝑆 =
𝑆1 𝑆2 … . 𝑆𝑛−1 𝑆𝑛 .
As "𝑛" is large, 𝐵𝑛 tends to a diagonal matrix whose diagonal elements the Eigen values are
of "𝐴" and the columns of "𝑆" are the corresponding Eigen vectors.
1 2 2
1. Find all the Eigen values and Eigen vectors of the matrix 𝐴 = 2 3 2 using
2 2 1
Jacobi’s method.
Solution:
Step 1: The numerically largest off-diagonal element is 𝑎13 = 2 .
1 2𝑎 13 1 4 1 1 𝜋 𝜋
Step 2: 𝜃 = 𝑡𝑎𝑛−1 = 𝑡𝑎𝑛−1 = 𝑡𝑎𝑛−1 ∞ = = .
2 𝑎 11 −𝑎 33 2 1−1 2 2 2 4
𝜋 𝜋 1 1
𝑐𝑜𝑠𝜃 0 −𝑠𝑖𝑛𝜃 cos 4 0 − sin 0 −
4 2 2
Step 3: ∴ 𝑆1 = 0 1 0 = 0 1 0 = 0 1 0 .
𝜋 𝜋 1 1
𝑠𝑖𝑛𝜃 0 𝑐𝑜𝑠𝜃 sin 4 0 cos 4 0
2 2
1 1
0
2 2
Step 4: 𝑆1𝑇 = 0 1 0
1 1
− 2 0 2

Dr.V.Nirmala, AP/Maths, UCET Page 52

 1 1 1 1
0 1 2 2 0 −
2 2 2 2
Step 5: 𝐵1 = 𝑆1𝑇 𝐴𝑆1 = 0 1 0 2 3 2 0 1 0
1 1 1 1
− 2 0 2 2 1 0
2 2 2
1 1 3 1
0 2
2 2 2 2
⇒𝐵1 = 0 1 0 2 3 0
1 1 3 −1
− 2 0 2
2 2 2

3 2 0
⇒𝐵1 = 2 3 0
0 0 −1
Since, 𝐵1 is not a diagonal matrix, we proceed to find 𝐵2 .
Step 6: The numerically largest off-diagonal element of 𝐵1 is 𝑎12 = 2 .
1 2𝑎 12 1 4 1 1 𝜋 𝜋
Step 7: 𝜃 = 2 𝑡𝑎𝑛−1 = 2 𝑡𝑎𝑛−1 = 2 𝑡𝑎𝑛−1 ∞ = 2 = 4.
𝑎 11 −𝑎 22 3−3 2
𝜋 𝜋 1 1
𝑐𝑜𝑠𝜃 −𝑠𝑖𝑛𝜃 0 cos 4 −sin 0 − 0
4 2 2
Step 8: ∴ 𝑆2 = 𝑠𝑖𝑛𝜃 𝑐𝑜𝑠𝜃 0 = sin 𝜋 cos
𝜋
0 =
1 1
0 .
0 0 1 4 4 2 2
0 0 1 0 0 1
1 1
0
2 2
Step 9: 𝑆2𝑇 = − 1 1
0
2 2
0 0 1
1 1 1 1
0 3 2 0 − 0
2 2 2 2
Step 10: 𝐵2 = 𝑆2𝑇 𝐵1 𝑆2 = − 1 1
0 2 3 0
1 1
0
2 2 0 0 −1 2 2
0 0 1 0 0 1
1 1 5 1
0 − 0
2 2 2 2
⇒𝐵2 = − 1 1
0
5 1
0
2 2 2 2
0 10 0 0 −1
5 0 0
⇒𝐵2 = 0 1 0
0 0 −1
Here, 𝐵2 is a diagonal matrix and hence we terminate the process.
The required Eigen values are 𝜆1 = 5; 𝜆2 = 1; 𝜆3 = −1.
To find the Eigen vectors:
1 1 1
1
0 −
1 1
−
1
0 −2 −
2 2
2 2 2 2
1 1
𝑆1 𝑆2 = 0 1 0 1 1
0 = 0
2 2
1 1 2 2
0 1
−2
1 1
2 2 0 0 1 2 2

Dr.V.Nirmala, AP/Maths, UCET Page 53

 1 1
−2 −
1
2
2
1 1
∴ The corresponding Eigen vectors are 𝑋1 = 2
; 𝑋2 = 2
; 𝑋3 = 0 ;
1
1 1
−2 2
2

Hence the problem.

1
1 1 2
1
2. Find all the Eigen values and Eigen vectors of the matrix 𝐴 = 1 1 4
using Jacobi’s
1 1
2
2 4

method.

Dr.V.Nirmala, AP/Maths, UCET Page 54

Dr.V.Nirmala, AP/Maths, UCET Page 55
 2 −1 0
3. Find all the Eigen values and Eigen vectors of the matrix 𝐴 = −1 2 −1 using
0 −1 2
Jacobi’s method.

Dr.V.Nirmala, AP/Maths, UCET Page 56

Dr.V.Nirmala, AP/Maths, UCET Page 57