Chameli Devi Group of Institutions, Indore

Mathematics-III [BT-301]
0000000

Subject Notes

Module2: Numerical Methods–2

Contents
Numerical Differentiation, Numerical integration:
TrapezoidalruleandSimpson’s1/3rdand3/8rules.SolutionofSimultaneousLinearAlgebraic equations
byGauss’s Elimination,Gauss’s Jordan, Crout’smethods,Jacobi’s,Gauss-Seidal,and Relaxation
method.

NumericalDifferentiation

In numerical analysis, numerical differentiation describes algorithms for estimating

the derivative of a mathematical function or function subroutine using values of the function
and perhaps other knowledge about the function. i.e. Numerical different ion is the process of
calculating the derivative of a function at some particular value of the independent variable by
means of a set of given values of that function.

[A] Derivative at any point :- The general method of numerical differentiation of a function
consists in obtaining an explicit analytical relation 𝑦 = 𝑓 𝑥 with the help of an interpolation
formula ,and then differentiating 𝑦 with respect to 𝑥 as many times as required.

Let the function 𝑦 = 𝑓 𝑥 be obtained by the Newton Gregory’s forward interpolation formula

𝑢 𝑢−1 2 𝑢 𝑢−1 𝑢−2 3

𝑦 = 𝑦0 + 𝑢∆𝑦0 + ∆ 𝑦0 + ∆ 𝑦0 + … … … … … … … … (1)
2! 3!
𝑥−𝑥 0
Where𝑢 = h,

dy 2𝑢 − 1 2 3𝑢2 − 6𝑢 + 2 3 4𝑢3 − 18𝑢2 + 22𝑢 − 6 4

= ∆𝑦0 + ∆ 𝑦0 + ∆ 𝑦0 + ∆ 𝑦0 + ⋯ ….
du 2! 3! 4!
𝑑𝑢 1
and𝑑𝑥 = h

𝑑𝑦 𝑑𝑦 𝑑𝑢 1 2𝑢 −1 3𝑢 2 −6𝑢+2 4𝑢 3 −18𝑢 2 +22𝑢−6

Therefore,𝑑𝑥 = 𝑑𝑢 𝑑𝑥 =h [∆𝑦0 + ∆2 𝑦0 + ∆3 𝑦0 + ∆4 𝑦0 +
2! 3! 4!
⋯ … . ]….(2)
 Chameli Devi Group of Institutions, Indore

Mathematics-III [BT-301]
0000000

Subject Notes
𝑑2𝑦
The second derivative𝑑𝑥 2 may be determined by differentiating (2) with respect to 𝑥

𝑑2 𝑦 1 2 3
6𝑢2 − 18𝑢 + 11 4 2𝑢3 − 12𝑢2 + 21𝑢 − 10 5
= [∆ 𝑦0 + 𝑢 − 1 ∆ 𝑦0 + ∆ 𝑦0 + ∆ 𝑦0
𝑑𝑥 2 h2 12 12
+ ⋯….]

Similarly, derivatives of other higher orders may be obtained.

Example: A rod is rotating in a plane about one of it’s ends .If the following table gives the
angle 𝜃 radians throughh ,which the rod has turned for different values of time 𝑡 seconds,
find it’s angular velocity and angular acceleration ,when 𝑡 = 0.7 second.

𝑡 seconds 0.0 0.2 0.4 0.6 0.8 1.0

𝜽 radians 0.0 0.12 0.48 1.10 2.0 3.20

Solution: First, we will construct the difference table:

𝑡 seconds 𝜽 radians ∆𝜽 ∆𝟐 𝜽 ∆𝟑 𝜽 ∆𝟒 𝜽 ∆𝟓 𝜽
0.0 0.0 0.12
0.2 0.12 0.36 0.24 0.02 0.00 0.00
0.4 0.48 0.62 0.26 0.02 0.00
0.6 1.10 0.90 0.28 0.02
0.8 2.0 1.20 0.30
1.0 3.20
dθ
Then the angular velocity dt is

1 2𝑢 −1 3𝑢 2 −6𝑢+2 4𝑢 3 −18𝑢 2 +22𝑢−6

=h [∆𝜃0 + ∆2 𝜃0 + ∆3 𝜃0 + ∆4 𝜃0 + ⋯ … . ]……….(1)
2! 3! 4!

𝑡−𝑡 0 𝑡−0.0
Where 𝑢 = = = 5𝑡
h, 0.2

At = 0.7 ,then 𝑢 = 3.5

dθ
Therefore, the angular velocity dt = 4.496radian per second.

𝑑2𝜃
The angular acceleration is obtained with the help of equation (1),
𝑑𝑡 2
 Chameli Devi Group of Institutions, Indore

Mathematics-III [BT-301]
000000

Subject Notes

𝑑2 𝜃 1 2 2 6𝑢 − 6 3 du 1 2
2
= 0 + ∆ 𝜃0 + ∆ 𝜃0 = 2
[∆ 𝜃0 + 𝑢 − 1 ∆3 𝜃0 ]
𝑑𝑡 h 2! 3! dt h

At𝑡 = 0.7,then𝑢 = 3.5

Therefore,the angular acceleration

𝑑2𝜃 1
= 0.24 + 3.5 − 1 0.02 = 7.25 radianper𝑠𝑒𝑐𝑜𝑛𝑑 2 .
𝑑𝑡 2 (0.2)2

[B]Derivative at tabulated points: -In particular case, when the derivatives are required at one
of the tabulated points 𝑥0 , , 𝑦0 , , 𝑥1 , , 𝑦1 , … … … … … . 𝑥𝑛 , , 𝑦𝑛 , and not at the intermediate
points, the following formula may be used with advantage.
𝑑
If 𝐷 denotes the differential operator 𝑑𝑥 and ∆ is the difference operator defined by∆𝑓 𝑥 =
𝑓 𝑥 + 𝑕 − 𝑓 𝑥 ,then there holds the operational relation

𝑕𝐷 = 𝑙𝑜𝑔 1 + ∆
1 1 1
i.e.𝐷 ≡ 𝑕 [∆ − 2 ∆2 + 3 ∆3 − ⋯ … … . ]

Where,
𝑕 is the interval between any two successive values of 𝑥 ,at which the values of y are
1 11 5
prescribed.Similarly,𝐷2 ≡ h 2 [∆2 − ∆3 + 12 ∆4 − 6 ∆5 + ⋯ … … . ]

Example: A slider in a machine moves along a fixed straight rod. It’s distance 𝑥 cm along the
road are given in the following table for various values of the time 𝑡 seconds

𝑡 seconds 0.0 0.1 0.2 0.3 0.4 0.5

𝒙 cm 30.1 31.6 32.9 33.6 40.0 33.8
Find the velocity and acceleration,when 𝑡 = 0.3 second

Solution:First we will construct the difference table:

𝑡 seconds 𝒙 cm ∆𝒙 ∆𝟐 𝒙 ∆𝟑 𝒙 ∆𝟒 𝒙 ∆𝟓 𝒙
0.0 30.1 1.5
0.1 31.6 1.3 -0.2 -0.4 6.7 -31.3
0.2 32.9 0.7 -0.6 6.3 -24.6
0.3 33.6 6.4 5.7 -18.3
0.4 40.0 -6.2 -12.6
0.5 33.8
 Chameli Devi Group of Institutions, Indore

Mathematics-III [BT-301]
0000000

Subject Notes
dx
Then the velocity dt is

1 1 1
.𝐷 ≡ 𝑕 [∆ − 2 ∆2 + 3 ∆3 − ⋯ … … . ]

Where h = 0.1
dx
Therefore, the velocity dt = Dx at 𝑡 = 0.3 is given by

1 1 1 1
𝐷𝑥3 = 0.1 ∆𝑥3 − 2 ∆2 𝑥3 + 3 ∆3 𝑥3 − ⋯ … … . = 10 ∆𝑥3 − 2 ∆2 𝑥3 =127 cm per second

[Since other differences are not available in the table ]

Similarly, acceleration,when𝑡 = 0.3 second

1 11 5
𝐷2 𝑥3 = [∆2 𝑥3 − ∆3 𝑥3 + 12 ∆𝑥3 4 − 6 ∆5 𝑥3 + ⋯ … … . ]= 100[∆2 𝑥3 − 0]= 1260
h2
cmper𝑠𝑒𝑐𝑜𝑛𝑑 2 .

Numerical integration
𝑏
The process of computing the value of a definite integral 𝑎
𝑓 𝑥 𝑑𝑥 from a set of numerical
values of the integrand is called Numerical integration. When applied to the integration of a
function of one variable, the process is known as quadrature.

A General Quadrature Formula for Equidistant Ordinates:-

We shall, now establish a general formula for the numerical integration, from which other
special formulas will be deduced. For this, we assume that the values of 𝒙 are at equal intervals
.i.e 𝑎 = 𝑥0 , 𝑥1 = 𝑥0 + h, 𝑥2 = 𝑥0 + 2h … … … … … . 𝑥𝑛 = 𝑥0 + 𝑛h = b

Let the function 𝑦 = 𝑓 𝑥 be given by Newton-Gregory’s forward interpolation formula’s as

𝑢 𝑢−1 2 𝑢 𝑢−1 𝑢−2 3

𝑦 = 𝑓 𝑥 = 𝑦0 + 𝑢∆𝑦0 + ∆ 𝑦0 + ∆ 𝑦0 + … … … … … … … … (1)
2! 3!
𝑥−𝑥 0
Where𝑢 = => 𝑥 = 𝑥0 + 𝑢h
h,

Integrate (1) on the both sides w.r.t.𝑥Over the limits 𝑎 = 𝑥0 to 𝑥𝑛 = 𝑥0 + 𝑛h = b

𝑏 𝑥 0 +𝑛h
𝑓 𝑥 𝑑𝑥 = 𝑓 𝑥 𝑑𝑥
𝑎 𝑥0
 Chameli Devi Group of Institutions, Indore

Mathematics-III [BT-301]
000000

Subject Notes

But 𝒙 = 𝑥0 + 𝑢h , ∴ dx = h du , When 𝑥 = 𝑥0 , 𝑡h𝑒𝑛 𝑢 = 0 𝑎𝑛𝑑

𝑥 = 𝑥0 + 𝑛h, then u = n
𝑥 0 +𝑛h
𝑓 𝑥 𝑑𝑥
𝑥0
𝑛
𝑢 𝑢−1 2 𝑢 𝑢−1 𝑢−2 3
=h 𝑦0 + 𝑢∆𝑦0 + ∆ 𝑦0 + ∆ 𝑦0
0 2! 3!

+ … … … … … 𝑑𝑢

𝑥 0 +𝑛h 𝑛2 1 𝑛3 𝑛2 1 𝑛4
 𝑥0
𝑓 𝑥 𝑑𝑥 = h 𝑛𝑦0 + ∆𝑦0 + 2! − ∆2 𝑦0 + 3! − 𝑛3 + 𝑛2 ∆3 𝑦0 +
2 3 2 4

…………… …………………………………..(2)
𝑏−𝑎
 Where𝑕 = 𝑛

This is required general quadrature formula.

[I] Trapezoidal rule :- If ,we put 𝑛 = 1 in (2) ,Then there are only two ordinates 𝑦0 , , 𝑦1 , and
only one finite difference ∆𝑦0 = 𝑦1 , − 𝑦0 , exists, all other higher order finite differences
become zero.
𝑥1
1 1 h
𝑓 𝑥 𝑑𝑥 = h 𝑦0 + ∆𝑦0 = h 𝑦0 + (𝑦1 , − 𝑦0 , ) = [𝑦0 + 𝑦1 , ]
𝑥0 2 2 2

The geometrical meaning of this result is that the area between the curve𝑦 = 𝑓 𝑥 ,the𝑥-axis
and the ordinates 𝑦0 𝑎𝑛𝑑𝑦1 ,is approximated by the area of the trapezium who parallel sides
are𝑦0 𝑎𝑛𝑑𝑦1 , and whose breadth is h = 𝑥1 , − 𝑥0 ,
𝑥2 h
Similarly 𝑥1
𝑓 𝑥 𝑑𝑥 = [𝑦1 + 𝑦2 , ],
2

𝑥3
h
𝑓 𝑥 𝑑𝑥 = [𝑦 + 𝑦3 , ]
𝑥2 2 2

And so on
𝑥𝑛
h
𝑓 𝑥 𝑑𝑥 = [𝑦 + 𝑦𝑛 , ]
𝑛−1 2 𝑛−1
 Chameli Devi Group of Institutions, Indore

Mathematics-III [BT-301]
0000000

Subject Notes

Therefore, adding all the above results, we get

𝒃 𝐡 𝑏−𝑎
𝒂
𝒇 𝒙 𝒅𝒙 = 𝟐 𝒚𝟎 + 𝒚𝒏 , + 𝟐 𝒚𝟏 + 𝒚𝟐 , + 𝒚𝟑 , + ⋯ … … . . +𝒚𝒏−𝟏 Where𝑕 = 𝑛

This is required Trapezoidal rule.

[II] Simpson’s 1/3rd rule : If ,we put 𝑛 = 2 in (2) ,Then there are only three ordinates
𝑦0 , , 𝑦1 , 𝑦2 , and therefore only two finite differences ∆𝑦0 𝑎𝑛𝑑∆2 𝑦0 are exist and all other
differences become zero.
𝑥2
𝐡
𝑓 𝑥 𝑑𝑥 = [𝑦 + 4𝑦1 , + 𝑦2 , ]
𝑥0 𝟑 0,

The geometrical meaning of this result is that the curve 𝑦 = 𝑓 𝑥 between the lines
𝑥 = 𝑥0 𝑎𝑛𝑑𝑥 = 𝑥2 is approximated by the area of the parabola whose equation is
𝑥−𝑥 0 𝑥 −𝑥 0 𝑥−𝑥 1
𝑦 = 𝑦0 + ∆𝑦0 + ∆2 𝑦0 and which passes through the points
h 2!𝑕 2
𝑥0 , , 𝑦0 , , 𝑥1 , , 𝑦1 , , 𝑥2, , 𝑦2 , .

Proceeding in a similar way, we will get

𝑥4 𝐡
𝑥2
𝑓 𝑥 𝑑𝑥 = 𝟑 [𝑦2 , + 4𝑦3 , + 𝑦4 , ] ,

𝑥6 𝐡
𝑥4
𝑓 𝑥 𝑑𝑥 = 𝟑 [𝑦4 , + 4𝑦5 , + 𝑦6 , ] ,

And so on ……….
𝑥𝑛
𝐡
𝑓 𝑥 𝑑𝑥 = [𝑦 + 4𝑦𝑛−1 , + 𝑦𝑛 , ]
𝑥 𝑛 −2 𝟑 𝑛−2 ,

Therefore, adding all the above results, we get

𝒃
𝐡
𝒇 𝒙 𝒅𝒙 = [ 𝒚𝟎 + 𝒚𝒏 , + 𝟒 𝒚𝟏 , + 𝒚𝟑 , + ⋯ . +𝒚𝒏−𝟏 , + 𝟐 𝒚𝟐 , + 𝒚𝟒 , + ⋯ 𝒚𝒏−𝟐 , ]
𝒂 𝟑
𝑏−𝑎
This is required Simpson’s 1/3rdrule. Where𝑕 = 𝑛
 Chameli Devi Group of Institutions, Indore

0000000
Mathematics-III [BT-301]

Subject Notes

[III] Simpson’s 3/8rule :- If ,we put 𝑛 = 3 in (2) ,Then there are only four ordinates
𝑦0 , , 𝑦1 , 𝑦2 , 𝑦3 , and therefore only three finite differences ∆𝑦0 , ∆2 𝑦0 , 𝑎𝑛𝑑∆3 𝑦0 are exist and all
other differences become zero
𝑥3
𝟑𝐡
𝑓 𝑥 𝑑𝑥 = 𝑦 + 3(𝑦1 , + 𝑦2 , + 𝒚𝟑 , ]
𝑥0 𝟖 0,

The geometrical meaning of this result is that the curve 𝑦 = 𝑓 𝑥 between the lines
𝑥 = 𝑥0 𝑎𝑛𝑑𝑥 = 𝑥3 is approximated by the area of the cubical parabola whose equation is

𝑥−𝑥 0 𝑥 −𝑥 0 𝑥−𝑥 1 𝑥−𝑥 0 𝑥−𝑥 1 𝑥−𝑥 2

𝑦 = 𝑦0 + ∆𝑦0 + ∆2 𝑦0 + ∆3 𝑦0 And which passes through the
h 2!𝑕 2 3!𝑕 2
points 𝑥0 , , 𝑦0 , , 𝑥1 , , 𝑦1 , , 𝑥2, , 𝑦2 , 𝑎𝑛𝑑 𝑥3, , 𝑦3 , .

Proceeding in a similar way, we will get

𝑥6
𝟑𝐡
𝑓 𝑥 𝑑𝑥 = 𝑦 + 3(𝑦4 , + 𝑦5, + 𝒚𝟔 , ]
𝑥3 𝟖 3,

And so on ……….
𝑥𝑛
𝟑𝐡
𝑓 𝑥 𝑑𝑥 = 𝑦 + 3(𝑦𝑛−2 , + 𝑦𝑛−1, + 𝒚𝒏, ]
𝑥 𝑛 −3 𝟖 𝑛−3 ,

Therefore, adding all the above results, we get

𝒃
𝟑𝐡
𝒇 𝒙 𝒅𝒙 = [ 𝒚𝟎 + 𝒚𝒏 , + 𝟐 𝒚𝟑 , + 𝒚𝟔 , + ⋯ . . +𝒚𝒏−𝟑 ,
𝒂 𝟖
+ 𝟑 (𝒚𝟏 , + 𝒚𝟐 , + 𝒚𝟒 , + ⋯ + 𝒚𝒏−𝟏, ]
𝑏−𝑎
This is required Simpson’s 3/8th rule. Where𝑕 = 𝑛

1 1
Example: Evaluate the integral 0 1+𝑥 2
𝑑𝑥 by taking no. of subinterval 4 through, Trapezoidal
rd
rule, Simpson’s 1/3 rule and Simpson’s 3/8th rule.
𝑏−𝑎 1−0 1
Solution: We know that Where𝑕 = = =4
𝑛 4
 Chameli Devi Group of Institutions, Indore

000000
Mathematics-III [BT-301]

Subject Notes

𝒙 1
𝒚=
1 + 𝑥2

𝑥0 = 0 𝑦0 = 1
1 16
𝑥1 = 𝑥0 + 𝑕 = 𝑦1 =
4 17
1 4
𝑥2 = 𝑥0 + 2𝑕 = 𝑦2 =
2 5
3 16
𝑥3 = 𝑥0 + 3𝑕 = 𝑦3 =
4 25
𝑥4 = 𝑥0 + 4𝑕 = 1 1
𝑦4 =
2

We know that by Trapezoidal rule

𝒃
𝐡
𝒇 𝒙 𝒅𝒙 = [ 𝒚𝟎 + 𝒚𝒏 , + 𝟐(𝒚𝟏 + 𝒚𝟐 , + 𝒚𝟑 , + ⋯ … … . . +𝒚𝒏−𝟏 ]
𝒂 𝟐

Hence from above table the formula becomes

1 1 𝐡
0 1+𝑥 2
𝑑𝑥 = 𝟐 [ 𝒚𝟎 + 𝒚𝟒 , + 𝟐(𝒚𝟏 + 𝒚𝟐 , + 𝒚𝟑 , )] =0.782775

Now by Simpson’s 1/3rd rule

𝒃
𝐡
𝒇 𝒙 𝒅𝒙 = [ 𝒚𝟎 + 𝒚𝒏 , + 𝟒 𝒚𝟏 , + 𝒚𝟑 , + ⋯ . +𝒚𝒏−𝟏 , + 𝟐 𝒚𝟐 , + 𝒚𝟒 , + ⋯ 𝒚𝒏−𝟐 , ]
𝒂 𝟑

Hence from the above table the formula becomes

1
1 𝐡
𝑑𝑥 = 𝒚𝟎 + 𝒚𝟒 , + 𝟒 𝒚𝟏 , + 𝒚𝟑 , + 𝟐 𝒚𝟐 , = 𝟎. 𝟕𝟖𝟓𝟒.
0 1 + 𝑥2 𝟑

Now by Simpson’s 3/8th rule

𝒃
𝟑𝐡
𝒇 𝒙 𝒅𝒙 = [ 𝒚𝟎 + 𝒚𝒏 , + 𝟐 𝒚𝟑 , + 𝒚𝟔 , + ⋯ . . +𝒚𝒏−𝟑 ,
𝒂 𝟖
+ 𝟑 (𝒚𝟏 , + 𝒚𝟐 , + 𝒚𝟒 , + ⋯ + 𝒚𝒏−𝟏, ]

Hence from above table the formula becomes

 Chameli Devi Group of Institutions, Indore

000000
Mathematics-III [BT-301]

Subject Notes
𝒃
𝟑𝐡
𝒇 𝒙 𝒅𝒙 = 𝒚𝟎 + 𝒚𝟒 , + 𝟐 𝒚𝟑 , + 𝟑 (𝒚𝟏 , + 𝒚𝟐 , = 𝟎. 𝟕𝟓𝟎𝟑
𝒂 𝟖

Numerical Solution of Simultaneous Linear Algebraic Equations

Simultaneous Linear Algebraic Equations are very common in various fields of Engineering and
Science. We used matrix inversion method or Cramer’s rule to solve these equations in general.
But these methods prove to be tedious, when the system of equations contain a large number
of unknowns. To solve such equations there are other numerical methods, which are
particularly suited for computer operations. These are of two types:-

[I] Direct Method:

(1) Gauss’s Elimination Method.

(2) Gauss’s Jordan Method.

(3) Croute’smethods.

[II] Iterative Method

(1) Gauss’s Jacobi’s Method

(2) Gauss-Seidal Method
(3) Relaxation method
[I] Direct Method

(1) Gauss’s Elimination Method: In this method, the unknowns are eliminated successively
and the system is reduced to an upper triangular system from which, the unknowns are found
by back substitutions. The method is quite general and is well adapted for computer
operations.

Let us consider a system of𝑚 equations and in 𝑛 unknowns

a11x1  a12 x2    a1n xn  b1

a21x1  a22 x2    a2 n xn  b2

am1 x1  am 2 x2    amn xn  bm ------------------------------------- [1]
 Chameli Devi Group of Institutions, Indore

00000
Mathematics-III [BT-301]

Subject Notes

In this method for solving the above equations, we proceed stepwise asfollows:

Step-1:- Elimination of𝑥1 from the second, third, 𝑛𝑡𝑕Equation. We assume here that the order
of the equation and the order of unknowns in each equations are such that𝑎11 ≠ 0. The
𝑎
variable 𝑥1 can then be eliminated from the second equation by subtracting ( 𝑎 21 ) times the
11
𝑎 31
first equation from the second equation ( 𝑎 ) times the first equation from the third equation,
11
e.t.c. This gives new system say as follows:

𝑎11 𝑥1 + 𝑎12 𝑥2 + … … … + 𝑎1𝑛 𝑥𝑛 = 𝑏1

𝑎22
′
𝑥2 + … … … . +𝑎2𝑛
′
𝑥𝑛 = 𝑏2′

…………… ……………… -------------------------------- [2]

𝑎𝑛2
′
𝑥2 + … … … . +𝑎𝑛𝑛
′
𝑥𝑛 = 𝑏𝑛′

Here the first equation is called the pivotal equation and 𝑎11 is called the first pivot.

Step-2.Now, Elimination of 𝑥2 from thethird,𝑛𝑡𝑕Equation in (2).

If the coefficient 𝑎22

′
, … … . . , 𝑎𝑛𝑛
′
in (2) are not all zero,we may assume that the order of
equation and the unknowns is such that 𝑎22 ′
≠ 0 .Then ,we may eliminate 𝑥2 from the third,
….. 𝑛𝑡𝑕Equation of (2) by subtracting

𝑎′
( 𝑎 33
′ )times the second equation from the third equation,
22

𝑎′
( 𝑎 42
′ )times the second equation from the fourth equation e.t.c.
22

The further steps are now obvious. In the third step, we eliminate 𝑥3 and in the fourth step,
we eliminate 𝑥4 e.t.c.

By successive elimination, we arrive at a single equation in the unknown 𝑥𝑛 ,which can be

solved and substituting this in the preceding equation, we obtain the value of 𝑥𝑛−1 . In this,
manner, we find 𝑥𝑛 when the elimination is completed. Also when the elimination is complete
the system takes the form.

𝑐11 𝑥1 + 𝑐12 𝑥2 + … … … + 𝑐1𝑛 𝑥𝑛 = 𝑑1

𝑐22 𝑥2 + … … … . +𝑐2𝑛 𝑥𝑛 = 𝑑2
 Chameli Devi Group of Institutions, Indore

Mathematics-III [BT-301]
000000

Subject Notes

…………… …………… ………………

𝑐𝑛𝑛 𝑥𝑛 = 𝑑𝑛

In this case there exists a unique solution. The new coefficient matrix is an upper triangular
matrix; the diagonal element 𝑐𝑖𝑖 are usually equal to 1.

Example: - Apply Gauss’s Elimination Method to solve the equations

𝒙 + 𝟒𝒚 − 𝒛 = −𝟓

𝒙 + 𝒚 − 𝟔𝒛 = −𝟏𝟐

𝟑𝒙 − 𝒚 − 𝒛 = 𝟒

Solution: Given system of equations can be written in matrix form

1 4 −1 𝑥 −5
≈ 1 1 −6 𝑦 = −12
3 −1 −1 𝑧 4
→ 𝑅2 − 𝑅1 , → 𝑅3 − 3𝑅1

1 4 −1 𝑥 −5
≈ 0 −3 −5 𝑦 = −7
0 −13 2 𝑧 19
13
→ 𝑅3 − 𝑅
3 2
1 4 −1 𝑥 −5
≈ 0 −3 −5 𝑦 = −7
0 0 71/3 𝑧 148/3

This is an upper triangular matrix of coefficient matrix 𝐴

Therefore, the algebraic form of above Matrix form is

𝒙 + 𝟒𝒚 − 𝒛 = −𝟓 … … . . (𝟏)

−𝟑𝒚 − 𝟓𝒛 = −𝟕 … … . . (𝟐)

71
𝒛 = 148/3 … … . . (𝟐)
3
 Chameli Devi Group of Institutions, Indore

Mathematics-III [BT-301]
0000000

Subject Notes

Hence from above by back substitution, we get the desiredapproximate solution

𝑧 = 2.0845 , 𝑦 = −1.1408 , 𝑥 = 1.6479

(2) Gauss’s Jordan Method: - This is a modification ofGauss’s Elimination Method. In this
method, elimination of unknowns is performed not in the equations below but in the equations
above also. Ultimatelyreducingthesystem to a diagonal matrix. i.e. each equation involving only
one unknown. Thus in this method, thelabor of back substitution for finding the unknowns is
saved at the cost of additional calculations. This method is well explained by the following
example.

Example: - Apply Gauss’s Jordan Method to solve the equations

𝒙 + 𝒚 + 𝒛 = 𝟗; 𝟐𝒙 − 𝟑𝒚 + 𝟒𝒛 = 𝟏𝟑 ; 𝟑𝒙 + 𝟒𝒚 + 𝟓𝒛 = 𝟒𝟎

Solution: Given system of equations is

𝒙 + 𝒚 + 𝒛 = 𝟗 … … … . (𝟏)

𝟐𝒙 − 𝟑𝒚 + 𝟒𝒛 = 𝟏𝟑 … … … (𝟐)

𝟑𝒙 + 𝟒𝒚 + 𝟓𝒛 = 𝟒𝟎 … … … . (𝟑)

Step-1:- Operate (2)-2(1) and (3)-3(1) to eliminate 𝑥 from (2) and (3).

𝒙 + 𝒚 + 𝒛 = 𝟗 ………….(4)
−𝟓𝒚 + 𝟐𝒛 = −𝟓 … … … (𝟓)
𝒚 + 𝟐𝒛 = 𝟏𝟑 ……….(6)
Step-2:- operate (4) +1/5 (5) and (6) +1/5 (5) to eliminate 𝑦 from (4) and (6).

𝒙 + 7/5𝒛 = 𝟖 ……….(7)
−𝟓𝒚 + 𝟐𝒛 = −𝟓 … … … (𝟖)
𝟏𝟐
𝒛 = 𝟏𝟐 …………..(9)
𝟓
Step-3:- operate (7) -7/12 (9) and (8) -5/6 (9) to eliminate 𝑧 from (7) and (8).

𝒙=𝟏
𝒚=3
𝒛=𝟓

(4) Croute’s method:- Computation scheme : Let us consider

 Chameli Devi Group of Institutions, Indore

Mathematics-III [BT-301]
00000

Subject Notes

𝑎11 𝑥1 + 𝑎12 𝑥2 + 𝑎13 𝑥3 = 𝑏1

𝑎21 𝑥1 + 𝑎22 𝑥2 + 𝑎23 𝑥3 = 𝑏2

𝑎31 𝑥1 + 𝑎32 𝑥2 + 𝑎33 𝑥3 = 𝑏3

The Augmented Matrix of given system of equations is

𝑎11 𝑎12 𝑎13 𝑏1

𝐴∶𝐵 = 𝑎21 𝑎22 𝑎23 𝑏2
𝑎31 𝑎32 𝑎33 𝑏3

Now from the above matrix, the Matrix of 12 unknowns,so called derived matrix or Auxiliary
𝑙11 𝑢12 𝑢13 𝑦1
Matrix is 𝐷 ∶ 𝑀 = 𝑙21 𝑙22 𝑢23 𝑦2 and is to be calculated as follows:
𝑙31 𝑙32 𝑙33 𝑦3

Step-1:- The first column of the derived matrix is identical with the first column of 𝐴 ∶ 𝐵 .

i.e. 𝑙11 = 𝑎11 ; 𝑙21 = 𝑎21 ; 𝑙31 = 𝑎31 .

Step-2: The first row to the right of the first column of the 𝐷 ∶ 𝑀 is obtained by dividing the
corresponding element in 𝐴 ∶ 𝐵 by the leading diagonal element of that row.

𝑎 12 𝑎 13 𝑏1
i.e.𝑢12 = = ; 𝑢13 = = ; 𝑦1 = .
𝑎 11 𝑎 11 𝑎 11

Step-3: Remaining second column of 𝐷 ∶ 𝑀 is calculated as follows:

𝑙22 = 𝑎22 − 𝑙21 𝑢12 ; 𝑙32 = 𝑎32 − 𝑙31 𝑢12

Step-4: Remaining elements of second row of 𝐷 ∶ 𝑀 is calculated as follows;

𝑎23 − 𝑙21 𝑢13 𝑏2 − 𝑙21 𝑦1

𝑢23 = ; 𝑦2 =
𝑙22 𝑙22

Step-5: Remaining element of the third column of 𝐷 ∶ 𝑀 is calculated as follows:

𝑙33 = 𝑎33 − ( 𝑙31 𝑢13 + 𝑙32 𝑢23 )

Step-6: Remaining element of third row of 𝐷 ∶ 𝑀 is calculated as follows;

𝑏3 − ( 𝑙31 𝑦1 + 𝑙32 𝑦2 )
𝑦3 =
𝑙33
 Chameli Devi Group of Institutions, Indore

Mathematics-III [BT-301]
000000

Subject Notes

Step-7: Hence, from above the Derived Matrix or Auxiliary 𝑀𝑎𝑡𝑟𝑖𝑥 𝑖𝑠 ,

𝑙11 𝑢12 𝑢13 𝑦1

𝐷 :𝑀 = 𝑙21 𝑙22 𝑢23 𝑦2
𝑙31 𝑙32 𝑙33 𝑦3

Step-8: Therefore, from above 𝐷 : 𝑀 𝑀𝑎𝑡𝑟𝑖𝑥 , 𝑡𝑕𝑒 𝑟𝑒𝑞𝑢𝑖𝑟𝑒𝑑 𝑠𝑜𝑙𝑢𝑡𝑖𝑜𝑛 𝑖𝑠 obtained as follows:

𝑥3 = 𝑦3 ;

𝑥2 = 𝑦2 − 𝑢23 𝑥3 ;

𝑥1 = 𝑦1 − [ 𝑥2 𝑢12 + 𝑥3 𝑢13 ]

Example: Apply Croute’s Method to solve the equations

2 𝑥1 − 6 𝑥2 + 8 𝑥3 = 24 ; 5 𝑥1 + 4 𝑥2 − 3 𝑥3 = 2 ; 3 𝑥1 + 𝑥2 + 2 𝑥3 = 16

Solution: The Augmented Matrix of given system of equations is

𝑎11 𝑎12 𝑎13 𝑏1 2 −6 8 24

𝐴∶𝐵 = 𝑎21 𝑎22 𝑎23 𝑏2 = 5 4 −3 2
𝑎31 𝑎32 𝑎33 𝑏3 3 1 2 16

Now from the above matrix, the Matrix of 12 unknowns, so called derived matrix or Auxiliary
𝑙11 𝑢12 𝑢13 𝑦1
Matrix is 𝐷 ∶ 𝑀 = 𝑙21 𝑙22 𝑢23 𝑦2 and is to be calculated as follows:
𝑙31 𝑙32 𝑙33 𝑦3

Step-1:- The first column of the derived matrix 𝐷 ∶ 𝑀 is identical with the first column of
𝐴 ∶ 𝐵 .i.e 𝑙11 = 𝑎11 = 2; 𝑙21 = 𝑎21 = 5; 𝑙31 = 𝑎31 = 3

Step-2: The first row to the right of the first column of the 𝐷 ∶ 𝑀 is obtained by dividing the
corresponding element in 𝐴 ∶ 𝐵 by the leading diagonal element of that row.i.e.

𝑎12 −6 𝑎13 8 𝑏1 24
𝑢12 = = = −3 ; 𝑢13 = = = 4; 𝑦1 = = = 12
𝑎11 2 𝑎11 2 𝑎11 2

Step-3: Remaining second column of 𝐷 ∶ 𝑀 is calculated as follows:

𝑙22 = 𝑎22 − 𝑙21 𝑢12 = 4 − 5 −3 = 19 ; 𝑙32 = 𝑎32 − 𝑙31 𝑢12 = 1 − 3 −3 = 10

Step-4: Remaining elements of second row of 𝐷 ∶ 𝑀 is calculated as follows;

 Chameli Devi Group of Institutions, Indore

Mathematics-III [BT-301]
000000

Subject Notes

𝑎23 − 𝑙21 𝑢13 −3 − 5(4) 23 𝑏2 − 𝑙21 𝑦1 2 − 5(12) 58

𝑢23 = = =− ; 𝑦2 = = =−
𝑙22 19 19 𝑙22 19 19

Step-5: Remaining elements of the third column of 𝐷 ∶ 𝑀 is calculated as follows:

23 40
𝑙33 = 𝑎33 − 𝑙31 𝑢13 + 𝑙32 𝑢23 = 2 − [3 4 + 10 − =
19 19

Step-6: Remaining element of third row of 𝐷 ∶ 𝑀 is calculated as follows;

58
𝑏3 −( 𝑙 31 𝑦1 + 𝑙 32 𝑦2 ) 16−[3 12 +10(−19 )
𝑦3 = = 40 = 5.
𝑙 33
19

Step-7: Hence, from above the Derived Matrix or Auxiliary 𝑀𝑎𝑡𝑟𝑖𝑥 𝑖𝑠 ,

𝑙11 𝑢12 𝑢13 𝑦1 2 −3 4 12

𝐷 :𝑀 = 𝑙21 𝑙22 𝑢23 𝑦2 = 5 19 −23/19−58/19
𝑙31 𝑙32 𝑙33 𝑦3 3 10 40/19 5

Step-8: Therefore, from above 𝐷 : 𝑀 𝑀𝑎𝑡𝑟𝑖𝑥 , 𝑡𝑕𝑒 𝑟𝑒𝑞𝑢𝑖𝑟𝑒𝑑 𝑠𝑜𝑙𝑢𝑡𝑖𝑜𝑛 𝑖𝑠 obtained as follows:

𝑥3 = 𝑦3 = 5;

58 23
𝑥2 = 𝑦2 − 𝑢23 𝑥3 = − −5 − = 2;
19 19

𝑥1 = 𝑦1 − [ 𝑥2 𝑢12 + 𝑥3 𝑢13 ]=12- [3(-3)+5(4)] =1

[II] Iterative Methods:In these methods, we start from an initial approximation to the true
solution and obtain better and better approximations from a computation cycle repeated as
often as may be necessary for achieving a desired accuracy. For large systems, iterative
methods may be faster than the direct methods. Even the round -off errors in iterative methods
are smaller. In fact, iteration is a self-correcting process and any error made at any stage of
computation gets automatically corrected in the subsequent steps.

Simple iterative methods can be devised for systems in which the coefficient’s of the leading
diagonal are large as compared to others. We now describe three such methods:-

(1) Gauss’s Jacobi’s Method : Let us consider the system of equations

𝑎1 𝑥 + 𝑏1 𝑦 + 𝑐1 𝑧 = 𝑑1
 Chameli Devi Group of Institutions, Indore

Mathematics-III [BT-301]
000000

Subject Notes

𝑎2 𝑥 + 𝑏2 𝑦 + 𝑐2 𝑧 = 𝑑2 ……………..(1)

𝑎3 𝑥 + 𝑏3 𝑦 + 𝑐3 𝑧 = 𝑑3

If 𝑎1 , 𝑏2 , 𝑐3 are large as compared to other coefficients, solve the above system for
𝑥, 𝑦, 𝑧 respectively. Then system of equations can be written as

1
𝑥= ( 𝑑1 − 𝑏1 𝑦 − 𝑐1 𝑧)
𝑎1
1
𝑦= ( 𝑑2 − 𝑎2 𝑥 − 𝑐2 𝑧) ………………………… (2)
𝑏2

1
𝑧= ( 𝑑3 − 𝑎3 𝑥 − 𝑏3 𝑦)
𝑐3

Let us start with the initial approximation 𝑥0 , 𝑦0 , 𝑧0 for the values of

𝑥, 𝑦, 𝑧 respectively.Substituting these on the right sides of above, the first approximations are
given by
1
𝑥1 = ( 𝑑1 − 𝑏1 𝑦0 − 𝑐1 𝑧0 )
𝑎1

1
𝑦1 = ( 𝑑2 − 𝑎2 𝑥0 − 𝑐2 𝑧0 )
𝑏2

1
𝑧1 = ( 𝑑3 − 𝑎3 𝑥0 − 𝑏3 𝑦0 )
𝑐3

Now for second approximations, Substituting the𝑥1 ,𝑦1 , 𝑧1 on the right hand sides of (2).

1
𝑥2 = ( 𝑑1 − 𝑏1 𝑦1 − 𝑐1 𝑧1 )
𝑎1

1
𝑦2 = ( 𝑑2 − 𝑎2 𝑥1 − 𝑐2 𝑧1 )
𝑏2

1
𝑧2 = ( 𝑑3 − 𝑎3 𝑥1 − 𝑏3 𝑦1 )
𝑐3

This process is repeated till the diffence between two consecutive approximations is negligible.

Note: In the absence of any better estimates for 𝑥0 , 𝑦0 , 𝑧0 ,these may each be taken as zero.
 Chameli Devi Group of Institutions, Indore

000000
Mathematics-III [BT-301]

Subject Notes

Example: Solve by Gauss’s Jacobi’s Method

20𝑥 + 𝑦 − 2𝑧 = 17

3𝑥 + 20𝑦 − 𝑧 = −18……………..(1)

2𝑥 − 3𝑦 + 20𝑧 = 25

Solution: Then system of equations can be written as

1
𝑥= (17 − 𝑦 + 2𝑧)
20
1
𝑦 = 20 (−18 − 3𝑥 + 𝑧) ………………………… (2)

1
𝑧= (25 − 2𝑥 + 3𝑦)
20
For first approximations, substituting 𝑥0 = 0, 𝑦0 = 0, 𝑧0 = 0 for the values of
𝑥, 𝑦, 𝑧 respectively in above,

1
𝑥1 = 17 − 0 + 2.0 = 0.85;
20
1
𝑦1 = −18 − 3.0 + 0 = −0.9;
20
1
𝑧1 = 25 − 2. 0 + 3.0 = 1.25.
20
Now for second approximations, Substituting the𝑥1 ,𝑦1 , 𝑧1 on the right hand sides of (2).𝑥2 =
1
17 − 𝑦1 + 2 𝑧1 = 1.02;
20

1
𝑦2 = −18 − 3 𝑥1 + 𝑧1 = −0.965;
20
1
𝑧2 = 25 − 2𝑥1 + 3𝑦1 = 1.03.
20
Now for third approximations, Substituting the𝑥2 ,𝑦2 , 𝑧2 on the right hand sides of (2).𝑥3 =
1
17 − 𝑦2 + 2 𝑧2 = 1.00125;
20
 Chameli Devi Group of Institutions, Indore

Mathematics-III [BT-301]
000000

Subject Notes

1
𝑦3 = −18 − 3 𝑥2 + 𝑧2 = −1.0015;
20
1
𝑧3 = 25 − 2𝑥2 + 3𝑦2 = 1.00325.
20
Now for fourth approximations, substituting these values on the right hand sides of (2).𝑥4 =
1
17 − 𝑦3 + 2 𝑧3 = 1.0004;
20

1
𝑦4 = −18 − 3 𝑥3 + 𝑧3 = −1.000025;
20
1
𝑧4 = 25 − 2𝑥3 + 3𝑦3 = 0.9965.
20
Now for fifth approximations, substituting these values on the right hand sides of (2).𝑥5 =
1
20
17 − 𝑦4 + 2 𝑧4 = 0.999966;

1
𝑦5 = −18 − 3 𝑥4 + 𝑧4 = −1.000078;
20
1
𝑧5 = 25 − 2𝑥4 + 3𝑦4 = 0.999956.
20
Now for sixth approximations, substituting these values on the right hand sides of (2).𝑥6 =
1
17 − 𝑦5 + 2 𝑧5 = 1.0000;
20

1
𝑦6 = −18 − 3 𝑥5 + 𝑧5 = −0.999997;
20
1
𝑧6 = 25 − 2𝑥5 + 3𝑦5 = 0.999992.
20
As the values in the 5th and 6th approximations, i.e. Iterations being practically the same.We can
stop now, Hence the solution is

𝑥 = 1; 𝑦 = −1; 𝑧=1

(2) Gauss-Seidal Method: This is a modification of Jacobi’s method.

Let us consider the system of equations

𝑎1 𝑥 + 𝑏1 𝑦 + 𝑐1 𝑧 = 𝑑1
 Chameli Devi Group of Institutions, Indore

000000
Mathematics-III [BT-301]

Subject Notes

𝑎2 𝑥 + 𝑏2 𝑦 + 𝑐2 𝑧 = 𝑑2 ……………..(1)

𝑎3 𝑥 + 𝑏3 𝑦 + 𝑐3 𝑧 = 𝑑3

If 𝑎1 , 𝑏2 , 𝑐3 are large as compared to other coefficients, solve the above system for
𝑥, 𝑦, 𝑧 respectively. Then system of equations can be written as

1
𝑥= ( 𝑑1 − 𝑏1 𝑦 − 𝑐1 𝑧)
𝑎1
1
𝑦= ( 𝑑2 − 𝑎2 𝑥 − 𝑐2 𝑧) ………………………… (2)
𝑏2

1
𝑧= ( 𝑑3 − 𝑎3 𝑥 − 𝑏3 𝑦)
𝑐3

Let us start with the initial approximation 𝑥0 , 𝑦0 , 𝑧0 for the values of

𝑥, 𝑦, 𝑧 respectively. Which may each be taken as zero .Now,
Substituting 𝑦 = 𝑦0 , 𝑧 = 𝑧0 on the right sides of above first equation, the first
approximations are given by
1
𝑥1 = ( 𝑑1 − 𝑏1 𝑦0 − 𝑐1 𝑧0 )
𝑎1
Then putting 𝑥 = 𝑥1 , 𝑧 = 𝑧0 in the second of the equation of (2),we obtain

1
𝑦1 = ( 𝑑2 − 𝑎2 𝑥1 − 𝑐2 𝑧0 )
𝑏2

Next putting 𝑥 = 𝑥1 , 𝑦 = 𝑦1 in the third of the equation of (2),we obtain

1
𝑧1 = ( 𝑑3 − 𝑎3 𝑥1 − 𝑏3 𝑦1 )
𝑐3

And so on i.e. as soon as a new approximation for an unknown is found, it is

immediately used in the next step.
This process of iteration is repeated till the values of 𝑥, 𝑦, 𝑧 are obtained to desired
degree of accuracy.
Note :- (1) Since the most recent approximations of the unknowns are used,while
proceeding to the next step, the convergence in the Gauss-Seidal Method is twice
as fast as in Gauss’s Jacobi’s Method.
Gauss’s Jacobi’s Method and Gauss-Seidal Methods converge for any choice of the
initial approximations if in each equation of the system, the absolute value of the
 Chameli Devi Group of Institutions, Indore

Mathematics-III [BT-301]
00000

Subject Notes

largest coefficient is almost equal to or in at least one equation greater than the sum
of the absolute values of all the remaining coefficients.

Example: Apply Gauss-Seidal Method

20𝑥 + 𝑦 − 2𝑧 = 17

3𝑥 + 20𝑦 − 𝑧 = −18……………..(1)

2𝑥 − 3𝑦 + 20𝑧 = 25

Solution: Then system of equations can be written as

1
𝑥 = 20 (17 − 𝑦 + 2𝑧) ………….. (i)

1
𝑦 = 20 (−18 − 3𝑥 + 𝑧) …………(ii)

1
𝑧 = 20 (25 − 2𝑥 + 3𝑦)………….(iii)

For first approximations / Iteration,

Substituting 𝑦 = 𝑦0 = 0, 𝑧 = 𝑧0 = 0 for the values of 𝑦, 𝑧 respectively in (i),

1
𝑥1 = 17 − 0 + 2.0 = 0.8500;
20
Substituting 𝑥 = 𝑥1 , 𝑧 = 𝑧0 for the values of 𝑥 , 𝑧 respectively in (ii)

1
𝑦1 = −18 − 3𝑥1 + 𝑧0 = −1.0275;
20
Substituting 𝑥 = 𝑥1 , 𝑦 = 𝑦1 for the values of 𝑥 , 𝑦 respectively in (iii)

1
𝑧1 = 25 − 2𝑥1 + 3𝑦1 = 1.0109
20
For second approximations / Iteration

Substituting 𝑦 = 𝑦1 , 𝑧 = 𝑧1 for the values of 𝑦, 𝑧 respectively in (i),

1
𝑥2 = 17 − 𝑦1 + 2𝑧1 = 1.0025;
20
 Chameli Devi Group of Institutions, Indore

000000
Mathematics-III [BT-301]

Subject Notes

Substituting 𝑥 = 𝑥2 , 𝑧 = 𝑧1 for the values of 𝑥 , 𝑧 respectively in (ii)

1
𝑦2 = −18 − 3𝑥2 + 𝑧1 = −0.9998;
20
Substituting 𝑥 = 𝑥2 , 𝑦 = 𝑦2 for the values of 𝑥 , 𝑦 respectively in (iii)

1
𝑧2 = 25 − 2𝑥2 + 3𝑦2 = 0.9998
20
For third approximations / Iteration

Substituting 𝑦 = 𝑦2 , 𝑧 = 𝑧2 for the values of 𝑦, 𝑧 respectively in (i),

1
𝑥3 = 17 − 𝑦2 + 2𝑧2 = 1.0000;
20
Substituting 𝑥 = 𝑥3 , 𝑧 = 𝑧2 for the values of 𝑥 , 𝑧 respectively in (ii)

1
𝑦3 = −18 − 3𝑥3 + 𝑧2 = −1.0000;
20
Substituting 𝑥 = 𝑥3 , 𝑦 = 𝑦3 for the values of 𝑥 , 𝑦 respectively in (iii)

1
𝑧3 = 25 − 2𝑥3 + 3𝑦3 = 1.0000
20
Hence,the values in 2nd and 3rdApproximations being practically the same, now, we can stop,

The solution is 𝑥 = 1 ; 𝑦 = −1; 𝑧 = 1.

Therefore,we seen that the convergence is quite fast in Gauss-Seidal Method as compared to
Gauss’s Jacobi’s Method.

(3) Relaxation method: This method was original developed by R.V. South well in 1935,for
application to Structural engg. Problems.
Let us consider the system of equations

𝑎1 𝑥 + 𝑏1 𝑦 + 𝑐1 𝑧 = 𝑑1

𝑎2 𝑥 + 𝑏2 𝑦 + 𝑐2 𝑧 = 𝑑2 ……………..(1)

𝑎3 𝑥 + 𝑏3 𝑦 + 𝑐3 𝑧 = 𝑑3
 Chameli Devi Group of Institutions, Indore

Mathematics-III [BT-301]
000000

Subject Notes

We define the residuals 𝑅𝑥 , 𝑅𝑦 𝑅𝑧 , by the relations

𝑅𝑥 = 𝑑1 − 𝑎1 𝑥 − 𝑏1 𝑦 − 𝑐1 𝑧

𝑅𝑦 = 𝑑2 − 𝑎2 𝑥 − 𝑏2 𝑦 − 𝑐2 𝑧 …………….. (2)

𝑅𝑧 = 𝑑3 − 𝑎3 𝑥 − 𝑏3 𝑦 − 𝑐3 𝑧

To start with we assume 𝑥 = 0, 𝑦 = 0, 𝑧 = 0 and calculate the initial residuals. Then the
residuals are reduced step by step, by giving increments to the variables. For this purpose, we
construct the following operation table:

𝛿 𝑅𝑥 𝛿 𝑅𝑦 𝛿 𝑅𝑧
𝛿𝑥 = 1 − 𝑎1 −𝑎2 − 𝑎3
𝛿𝑦 = 1 − 𝑏1 − 𝑏2 − 𝑏3
𝛿𝑧 = 1 − 𝑐1 − 𝑐2 − 𝑐3
We note from the equations (2) that if 𝑥 is increased by 1 (keeping𝑦, 𝑧 constant), 𝑅𝑥 , 𝑅𝑦 and
𝑅𝑧 decrease by 𝑎1 , 𝑎2 , 𝑎3 respectively.This is shown in the above table along with the
effects on the residuals when 𝑦, 𝑎𝑛𝑑 𝑧 are given unit increments (The table is the transpose of
the coefficient matrix)

At each step, the numerically largest residual is reduced to almost zero,to reduce a particular
residual, the value of the corresponding variable is changed.

When all the residuals have been reduced to almost zero, the increments in 𝑥, 𝑦, 𝑎𝑛𝑑 𝑧 are
added separately to give the desired solution.

Obs.:The iterativeRelaxation method can be applied successfully only if the leading diagonal
elements of the coefficient matrix dominate the other coefficient’s in the corresponding row
i.e. if systems in which the coefficient’s of the leading diagonal are large as compared to others.
We now describe such method:-

The above method is well explained by the following example:-

Example: Apply Relaxation method

9𝑥 − 2𝑦 + 𝑧 = 50
 Chameli Devi Group of Institutions, Indore

Mathematics-III [BT-301]
000000

Subject Notes

𝑥 + 5𝑦 − 3𝑧 = 18……………..(1)

−2𝑥 + 2𝑦 + 7𝑧 = 19

Solution:Since the coefficients of the leading diagonal are large as compared to others. Hence,
we can apply this method

We define the residuals 𝑅𝑥 , 𝑅𝑦 𝑅𝑧 , by the relations

𝑅𝑥 = 50 − 9𝑥 + 2𝑦 − 𝑧

𝑅𝑦 = 18 − 𝑥 − 5𝑦 + 3𝑧 …………….. (2)

𝑅𝑧 = 19 + 2𝑥 − 2𝑦 − 7𝑧

The operation table is

𝛿 𝑅𝑥 𝛿 𝑅𝑦 𝛿 𝑅𝑧
𝛿𝑥 = 1 𝑎1 =-9 𝑎2 = −1 𝑎3 = 2
𝛿𝑦 = 1 𝑏1 = 2 𝑏2 = −5 𝑏3 =-2
𝛿𝑧 = 1 𝑐1 = −1 𝑐2 = 3 𝑐3 =-7
 Chameli Devi Group of Institutions, Indore

Mathematics-III [BT-301]
000000

Subject Notes

The Relaxation table is

Increments 𝐑𝐞𝐬𝐢𝐝𝐮𝐚𝐥 𝑹𝒙 𝐑𝐞𝐬𝐢𝐝𝐮𝐚𝐥 𝑹𝒚 𝐑𝐞𝐬𝐢𝐝𝐮𝐚𝐥 𝑹𝒛

= 𝟓𝟎 − 𝟗𝒙 + 𝟐𝒚 = 𝟏𝟖 − 𝒙 − 𝟓𝒚 + 𝟑𝒛 = 𝟏𝟗 + 𝟐𝒙 − 𝟐𝒚
−𝒛 − 𝟕𝒛
Initially ,𝑥 = 𝑦 = 𝑧 = 0 50 18 19
or
𝛿𝑥 = 𝛿𝑦 = 𝛿𝑧 = 0

Since 50 is numerically 5 13 29
largest residual ,
50
𝛿𝑥 = =5
9
Since 29 is numerically 1 25 1
largest residual ,
29
𝛿𝑧 = =4
7
Since 25 is numerically 11 0 5
largest residual ,
25
𝛿𝑦 = =5
5
Similarly , 𝛿𝑥 = 1 2 -1 -7
𝛿𝑧 = −1 3 -4 0
𝛿𝑦 = −0.8 1.4 0 1.6
𝛿𝑧 = 0.23 1.17 0.69 -0.1
𝛿𝑥 = 0.13 0 0.56 0.25
𝛿𝑦 = 0.112 0.224 0 0.026

Now, we stop, since the process of reducing residuals is carried as the all the residuals become
zero or negligible, Therefore, the desired solution is obtained by adding up the increments
𝛿𝑥, 𝛿𝑦, & 𝛿𝑧 𝑖𝑛 𝑥, 𝑦 & 𝑧 𝑠𝑒𝑝𝑎𝑟𝑎𝑡𝑒𝑙𝑦 𝑖. 𝑒.

𝑥= 𝛿𝑥 = 0 + 5 + 1 + 0.13 = 6.13 ;

𝑦= 𝛿𝑦 = 0 + 5 − 0.8 + 0.112 = 4.31 ;

𝑧= 𝛿𝑧 = 0 + 4 − 1 + 0.23 = 3.23 .