www.getmyuni.

com

CHAPTER-6: SIMPLEX & DUAL SIMPLEX METHOD

1. © Simplex method

Introduction

Simplex method = Simplex technique = Simplex algorithm.

It is an iterative procedure for solving a linear programming problem in a

finite no. of steps. This method provides an algorithm which consist in moving
from one vertex of the region of feasible solution to another in such a way that
the value of the objective function at the succeeding vertex is less (or more)
than the preceding vertex so as to reach finally in the optimum solution.

The following important definitions are necessary to understand the

simplex method.

(a) Basic solution

Given a system 'm' simultaneous linear equations in 'n' unknowns (m<n).

Ax = b,

Where A is a m×n matrix of rank m. Let B be any m×m submatrix

formed by m linearly in dependent columns of A. Then a solution obtained by
setting n – m variable not associated with the columns of B equals to zero and
solving the resulting system is called a basic solution to the given system of
equations.

This can be explained with the following example.

Ex: Obtain basic solution to x1+2x2+x3 = 4 ---------------(1)

2x 1+x2+5x3 = 5 ---------------(2)

In this case m = 2 (i.e., Eq. (1) & Eq. (2))

n = 3 (i.e., x1, x2, x3) = m<n

the above system of equation can be written as:

www.getmyuni.com

1
1 2 1 4
2 =
2 1 5 5
3
A x b

(2×3) ×(3×1) = (2×1)

The different sub matrix from matrix A are:

1 2 1 1 2 1
. , &
2 1 2 5 1 5
The basic solutions are obtained from the following equations:

1 2 1 4
(I) = -------(1) setting x3 = 0
2 1 2 5
2 1 2 4
(II) = -----------(2) setting x1 = 0
1 5 3 5
1 1 1 4
(III) = -------------(3) setting x2 = 0
2 5 2 5
Considering Eq (1)

[x1+2x2 = 4] ×2

2x1+x2 = 5

4x 2-x2 = 8 - 5 = 3

= 3x2 = 3 = x2 = 1

2x 1+1 = 5 = x1 = 2

So basic x1 = 2, x2 = 1 & non basic x3 = 0

Similarly solving other 2 equations :

Basic x2 = 5/3, x3 = 2/3 non basic x1 = 0

Basic x1 = 5, x3 = -1 non basic x2 = 0

www.getmyuni.com

(b) Degenerate solution

A basic solution to the system is called Degenerate of one or more of the

basic variables vanish.

Ex: Show that the following system of linear equations has a degenerate
solution.

2x1+x2-x3 = 2

3x1+2x2+x3 = 3

Solution

The given system of linear equations can be written as:

1
2 1 −1 2
. 2 =
3 2 1 3
3
A x b

The basic solution can be obtained as:

2 1 1 2
. = -------------(1) by setting x3 = 0
3 2 2 3
1 −1 2 2
. = ------------(2) by setting x1 = 0
2 1 3 3
2 −1 1 2
. = ------------(3) by setting x2 = 0
3 1 3 3
Solving (1) Basic → x1 = 1, x2 = 0; non basic x3 = 0

(2) Basic → x2 = 5/3, x3 = -1/3; non basic x1 = 0

(3) Basic → x1 = 1, x3 = 0; non basic x2 = 0

In each is 2 basic solutions at least one of the basic variables is zero.

Hence two of the basic solutions are degenerate solutions.
www.getmyuni.com

(c) Basic feasible solutions

A feasible solution to a LPP which is also basic solution to the problem is

called as basic feasible solution to LPP. Let the LPP be to determine x so as to
maximize

z = cx

subjected to constraints:

Ax = b

& x≥0

Then xB is a basic feasible solution to this problem if B is an m×m non-

singular sub-matrix of A and BxB = b xB ≥0

Looking to the solutions of earlier problem.

(1)
= , =− ( ), =0←
−

(2) x1 = 1, x2 = 0 (basic); x3 = 0 (non basic)

Basic feasible solution

(3) x1 = 1, x3 = 0(basic) ; x2 = 0 (non basic)

(d) Associated cost vector

Let xB be a basic feasible solution to the LPP:

Maximize z = cx

Subject to constraint Ax = b and x ≥ 0.

The vector cB = (cB1, cB2…………cBm) where cBi are components of c

associated with the basic variables is called the cost vector associated with the
basic feasible solution xB.

It is obvious that the value of the objective function for the basic feasible
solution xB, is given by:
www.getmyuni.com

z0 = cBxB

(e) Improved Basic Feasible Solution

Let xB and be two basic feasible solution to the standard LPP, then
is said to be an improved basic feasible solution as compared to xB if

̂ ≥

Where ̂ is constituted of cost components corresponding to .

(f) Optimum Basic Feasible Solution

A basic feasible solution xB to LPP –

Maximize z = cx

Subjected to Ax = b and x ≥ 0.

Is called an optimum basic feasible solution if

z0 = cB xB ≥ z*

where z* is value of objective function for any feasible solution.

© Fundamental Properties of solution

The fundamental properties of solution will allow us to arrive at the

simplex algorithm. These properties are presented in the following in the form
of theorems:

Theorem- 1: (Reduction of Feasible solution to basic feasible solution) If a LPP

has a feasible solution, then it also has a basic feasible solution.

Theorem-2: (Replacement of a basic vector) Let a LPP have a basic feasible

solution. If we drop one of the basic vectors and introduce a non-basic vector
in the basic set, then the new solution obtained is also a basic feasible solution.

Theorem- 3: Improved basic feasible solution – stated earlier.

Theorem- 4 : Unbounded solution

www.getmyuni.com

Let there exist a basic feasible solution to a given LPP. If for at least one
j, for which yij ≤ 0 (i = 1,2,3,……..m), zj –cj is negative, then there does not exist
any optimum solution to this LPP.

Theorem-5: condition of optimality

A sufficient condition for a basic feasible solution to a LPP to be an

optimum (maximum) is that zj – cj ≥ 0 for all j for which the column vector is
not in the basis.

Theorem- 6: Any convex combination of k different optimum solutions to a LPP

is again on optimum solution to the problem.

Theorem- 7: Minimax theorem

Let f be a linear function of n variables such that f(x*) is its minimum

value for some point xT*, xT E Rn, Then – f(x) attains its maximum at the point
x*. Moreover for xT E Rn.

Minimum f(x) = -maximum{- f(x)}

© Computational procedure

The computational procedure for simplex algorithm can be explained

with the help of a typical example.

Use simplex algorithm to solve the following LPP.

Maximize z = 4x1+5x2+9x3

Subjected to constraints : x1+x2+2x3 ≤ 16

fx1+5x2+3x3 ≤ 25

x1≥0, x2≥0, x3≥0

Solution:

Step-1: Check whether the objective function is to maximize or minimize. if it is

a minimization case, then convert it into maximization by this relation
minimize z = maximize (-z).
www.getmyuni.com

Step-2: Check whether all bi are +ve. If any one bi is –ve, multiply the
corresponding in equation by (-1).

Step-3: Convert all the inequations of constraints into equation by introducing

slack/surplus variables

Now introducing slack variable, x4 & x5 the inequation of constraints because:

X1+x2+2x3+x4 = 16 ----------------(1)

fx1+5x2+3x3+x5 = 25 ---------------(2)

and hence the new objected function becomes:

z = 4x1+5x2+9x3+0.x 4+0.x5 ---------------------(3)

step-4: Obtain an initial basic feasible solution to the problem in the form of

xB = ( ) and put in the xB column of the simplex table.

The set of equation (1) & (2) can be written as:

1
⎡ 2⎤
1 1 2 1 0 ⎢ ⎥ 16
.
7 5 3 0 1 ⎢ 3⎥ = 25
⎢ 4⎥
⎣ 5⎦

To obtain an initial basic feasible solution more easily the following

matrix equation can be considered

i.e.

1 0 4 16
. =
0 1 5 25
:- x4 = 16

& x5 = 25

These values are substituted in the appropriate place in the initial

simplex table.
www.getmyuni.com

Step-5: Compute the net evaluation i.e., zj – cj (j = 1,2,……..n) by using the

relation zj –cj = cByj – cj.

Examine the sign of zj – cj

If all the zj – cj ≥ 0 ; than initial basic feasible solution xB is an optimal

basic feasible solution; otherwise if any one <0, we have to go to next step.

Now the initial simplex table becomes:

0 0
I
Cf, YB }CB Y1 y2- Y3 Y1 J/s
0 _y '( ~:=/{; I I @ f 0 i
2

0 JI5' 'X.~ ~ '2-_!,-

r 5 3 (J i 2
L 3
l.L~ 0
-4 -B -q () D
~
I

Step-6: If there are more than one –ve in zj – cj, choose the most –ve of them,
let Zr –Cr. If all yir ≤0 (i = 1,2,……….m) then there is an unbounded solution to
the given problem. If at least one yir > 0, then corresponding vector yr enters
the basis.

Step- 7: Compute the ratio , > 0, and choose the minimum of them. let
the minimum be , then the vector yk leaves the basis yB. The common
element is called pivotal element.

Step-8: Convert the pivotal element/leading element and all other elements in
the column to zeros.

Step-9: Repeat the computational procedure until either an optimum solution

is obtained or there is an indication of unbounded solution.

Following the procedure of simplex algorithm, the next simplex table

becomes:
www.getmyuni.com

q 0 0

1ltl Y, v: y3 _y Y_,.--
c~ 513 ~
1 s:I-- .L
q h e 2-
L
2-
0

0 .Y s: i ~
2- 03 0
-3-
2-
1.

µ::::. 'f2- 1 -L2- 0 !L 0

2- 2-
4
I

The next simplex table will be :-

0 D

c;?, .Ye, »« -:;, /j-2- y3 ,y'1 Yr

-t.- _/
er y~ 5~-
-=;::
0 L '1r r
"ff
t5 Yz.- 2-f1_ --r-11 t. D -3 2-
r- r-
'f2- L _L .L
s- ?--
0 0 -30r r
- ~

Since all zj – cj are ≥ 0 so no repeat procedure is required and the

maximum value of z = 72 ans

© Concept of duality in simplex methd

The original LPP is called → primal problem

This original LPP can also be expressed in different form – called → Dual
problem.

There is an unique dual problem associated with the primal problem and
vice versa.

The following example will clearly explain the duality of original.

Ex: The amount of vitamins (v1 & v2) present I 2 different food (f1 & f2),
cost and daily requirement are presented in the following table.
www.getmyuni.com

Food Minimum daily

Vitamin
F1 F2 requirement

V1 2 4 40

V2 3 2 50

Cost 3 2.5

The problem is to determine the minimum daily requirement of the 2

foods to have minimum cost.

Solution:

In order to formulate the problem mathematically, let x1 & x2 are the

units of foods from f1 & f2 respectively. So the objective function is to.

Minimize z = 3x1+2.5x2

Subject to constraints:

(i) 2x 1 +4x2 ≥ 40
(ii) 3x 1 +2x2 ≥ 5.

Formation of dual problem from this primal problem.

Suppose there is a wholesale dealer who sales 2 vitamins v1 & v2 along

with other commodities. The dealer knows very well that the foods f1 & f2 have
market values only because of their vitamin content.

Now the problem of the dealer is to fix maximum per unit selling prices
for the 2 vitamins in such a way that the resulting prices of foods f1 & f2 does
not exceed their existing market price.

Let the dealer decide to fix up 2 prices at w1 & w2 per unit respectively.
Then the dealer problem can be stated mathematically.
www.getmyuni.com

Maximize z = 40w1 + 50w2

Subject to constraints:

2w1 + 3w2 ≤ 3

4w1 + 2w2 ≤ 2.5

w1 w2 ≥ 0

A comparison between primal & dual problem is presented as follows:

&ij -F"

Ex: Obtain the dual problem of the following LPP.

Maximize f(x) 2x1+5x2+6x3

Subject to constraints

5x 1+6x2-x3 ≤ 3

-2x1+x2+4x3 ≤ 4

X1-5x2+3x3 ≤ 1

-3x1-3x2+7x3 ≤ 6

x1 x2 x3 ≥ 0.

Also verify that dual of the dual problem is the primal problem.
www.getmyuni.com

Solution: The primal can be restated as : Determine x1 x2 x3 so as to maximize :-

f(x) = 2x1+5x2+bx3.

Subjected to constraints:

5 6 −1 1 3
−2 1 4 4
.
1 −5 3 2 ≤ 1
7 3 6
−3 −3

4*3 3*1 4*1

Thus if w = (w1,w2,w3,w4) are the dual variable then the problem is to

determine w so, as to minimize.

d(w) = 3w1+4w2+w3+6w4

subject to constraints:

1 2
5 −2 1 −3
. 6 2 ≥
1 −5 −3 3 5
−1 4 3 7 4 6

3*4 4*1 3*1

w1 w2 w3 w4 ≥ 0

i.e. minimize

g(w) = 3w1+4w2+w3+6w4

subjected to constraints:

5w1-2w2+w3 – 3w4 ≥ 2

6w1+w2-5w3-3w4 ≥ 5

-w1+4w2+3w3+7w4 ≥ 6

w1,w2,w3,w4 ≥ 0.

Now, we can restate this dual problem as:

www.getmyuni.com

Maximize h(w) = -3w1 -4w2-w3-6w4

Subjected to constraints

1
5 −2 1 −3 −2
(-1) 6 2 ≤
1 −5 −3 3 −5
−1 4 3 7 4
−6

w1,w2,w3,w4 ≥ 0

We can write dual of this dual problem as

To minimize

h(x) = -2x1-5x2-6x 3

subjected to constraints:

5 6 −1 1 −3
(-1) −2 1 4 −4
2 ≥ −1
1 −5 3
7 3 −6
−3 −3

x1,x2,x3 ≥ 0

i.e, maximize f(x) = 2x1+5x2+6x3

subjected to constraints:

5 6 1 1 3
. −2 1 4 4
2 ≤ 1 ←original form proceed
1 −5 3
7 3 6
−3 −3

© Dual Simplex Algorithm

The various steps of dual simplex algorithm in presented in the

following.

Step-1: Convert the minimization L.P.P. into maximization if it is in the

minimization form.
www.getmyuni.com

Converts into ≥ type inequations representing the constraints into ≤ type

by multiplying with (-1).

Step-2: Introduce slack variables in the constraints and obtain an initial basic
solution. Put this solution in the starting dual simplex table.

Step-3: Test the nature of zj – cj in the starting simplex table.

(a) If all zj – cj and xBi are non-negative for all i & j then an optimum basic
feasible solution has been obtained.
(b) If all zj – cj are non- negative and at least one basic variable say xBi is
negative, then go to step 4.
(c) If at least one zj-cj is negative, the method is not applicable to the given
problem.

Step-4: Select the most negative xBis. The corresponding basic vector then
leaves the basis set yB. Let xBk be the most negative basic variable so that yk
leaves yB.

Step-5: Test the nature of yki, j = 1,2,……….n.

(a) If all yki are non-negative, there does not exist any easible solution to the
given problem.
(b) If at least one yki is negative, compute the replacement ratios { , yki <
0} j = 1,2,…….n. and choose the maximum of these. The corresponding
column vectors, say yr (corresponding to ), then enters the basic set
yB .

Step-6: Test the new iterated dual simplex table for optimality.

Repeat the procedure until either an optimum feasible solution has been
obtained (in finite number of steps) or there is an indication of the non-
existence of a feasible solution.

Sample problem

Use dual simplex method to solve the LPP

Minimize z = x1+2x2+3x3
www.getmyuni.com

Subject to constraints:

X1-x2+x3 ≥ 4

x1+x2+2x3 ≤ 8

x2 – x3 ≥ 2

x1, x2, x3 ≥ 0

solution:

the objective function should be maximize z = -x1-2x2-3x3

by introducing slack variables x4, x5, x6 ≥0 in the constraints (after converting

into ≤ condition) the equations for constraints would be

-x1+x2-x3+x4 = -4

X1+x2+2x3+x5 = 8

0×x1-x2+x3+x6 = -2

1
⎡ 2⎤
−1 +1 −1 +1 0 0 ⎢ −4
3⎥⎥
:- 1 1 2 0 1 0 ⎢ = 8
4⎥
0 −1 +1 0 0 +1 ⎢⎢ 5⎥ −2
⎣ 6⎦

Considering x1,x2&x3 = 0 the initial basic solution would be obtained from.

+1 0 0 4 −4
. 0 1 0 5 = 8
0 0 +1 6 −2

i.e.,

x4 = -4,

x5 = 8

x6 = -2
www.getmyuni.com

now the starting dual simplex table:-

_, 0 0 0

y, y2- y y't y y{;

4i .Y/'j ?le 3 s:
.v _,+ -( 0 0
0
y
If
-1.f
(
{

2-
' 0 ( 0
0 s: e 1

Y{;, - 2-
-I !.- c -
3 0 D 0
0 \ L 2
-
First iteration: y4 is dropped and y1 is introduced. Hence the resulting dual
simplex table becomes:

-r 0 0 0
~
C¬ ,, .YB ~ .YI y.2.- Y3 .s ; yr;- _yj.,
-
-r s, 4 i - 1 1 - f 0 0

0 Ys- 4 0 2. i L t. 0

0 .Yt, - 2. 0 -L..,.. 1 0 0 L '

() 0
-4 -3 2 i 0
-'

Second iteration: y6 is dropped and y2 is introduced and the resulting table:-

-( -3 0

~ _y r5 ~ Y1 y 2- Y3 ->!it »s: ~L_I

-I Y; ~ 4 0 v -e- / 0 ~1

o y s 0 0 0 3 I 1 2-
- 2-- vs: 2- 0 L -i 0 0 -/
~-Io 0 0 s I 0 3
I

Now since all xj-cj ≥ 0 and also all xBi ≥ 0 an optimum basic feasible
solution has been reached.
www.getmyuni.com

Hence the optimum basic feasible solution to the given LPP is x1=6, x2=2,
x3=0 and minimum z = -(-10) = 10 ans