61

DUALITY

61.1 INTRODUCTION
In the previous lesson we have seen that in the case of a
maximum problem the simplex method can be applied easily in
order to find the optimal solution. Whereas when the problem is
of minimization we have to take recourse to the two phase
method, which is a bit complicated.
Is there any other method to solve these problems of
minimization?
Among a number of techniques for solving such problems is
one developed by John Von Neumann and others, in which the
solution (if it exists) of a minimum problem is found by solving
a related maximum problem called the dual problem.

61.2 OBJECTIVES :
After the study of this lesson you will be able to:
l

define the primal problem

define the dual problem

find the dual of a given problem

solve the minium problem

61.3 BASIC CONCEPTS OF DUALITY

The concept of duality plays a very important role in the
development of linear programming theory.

72 :: Mathematics

Associated with
another intimately
the dual problem
The original linear
problem.

every linear programming problem there is

related linear programming problem, called
of the original linear programming problem.
programming problem is called the primal

It is interesting that in general either problem can be taken

as the primal and the remaining problem as its dual. The optimal
solution of either problem gives us information concerning the
optimal solution of the other.
If the optimal solution of either problem is known, the optimal
solution of the other can be found.
The concept of a dual problem is sometimes very fruitful when
the dual is easier to solve than that of the primal. A particular
case when the number of variables is less than the constraints
in a linear programming problem, its solution can be found more
easily by sovling its dual instead the primal problem.

61.4 STANDARD FORM OF A MINIMUM PROBLEM

A minimum problem is said to be in standard form provided
the following conditions are met:
1. All the variables are nonnegative.
2. All the other constraints must be written with sign.
3. The objective function to be minimized must be written
with non-negative coefficients.
Example A :
Determine which of the following problems are in standard
form :
(a)

Minimize z = 2x 1+3x2
subject to the constraints
x1 +3x 2 24,
2x 1+x 2 18,
x1 0, x 2 0

(b)

Minimize z = 2x1 +x 2 +x3

subject to the constraints

Duality :: 73

x1 3x2 +x 3 12,
x1 +x 2+x3 1,
x1 0, x 2 0, x3 0
Solution:
(a)

Since all the three conditions are met, this minimum

problem is in standard form.

(b)

Conditions (1) and (3) are met but condition (2) is not.
Since the first constraint x 13x2 +x 3 12 is not written with
a sign. Thus, the minimum problem as stated is not in
the standard form.
Notice however, that by multiplying by 1, we can write
this constraint as x 1+3x 2 x 3 12. Written in this way the
problem is in the standard form.

Check point 1:
Is the following minimum problem in the standard form?
Minimize z = 2x1 +x2 +3x 3
subject to the constraints
x 1 +2x 2+x 3 2,
x1 +x 2+x3 6,
x1 0, x 2 0, x 3 0

61.5 THE DUAL OF A MINIMUM PROBLEM

We will illustrate by the following example that how to obtain
the dual of a minimum problem.
Example B :
Obtain the dual of the minimum problem
Minimize z = 300x 1+480x2
subject to the constraints
x1 +3x 2 0.25,
2x1 +2x2 0.45,
x1 0, x2 0

74 :: Mathematics

Solution :
Here the minimum problem is in the standard form. We begin
by constructing a special matrix for the coefficients of the
constraints of this problem without introducing slack variables.
As in a simplex table, we place the objective function in the last
row. The special matrix is
x1

x2

0.25

0.45

300

480

Taking transpose of this matrix, i.e., interchanging the rows to

columns, we get
1

300

480

0.25

0.45

Translating this matrix into a maximum problem in standard

form, we get.
Maximize z = 0.25y 1+0.45y 2
subject to the constraints
y 1+2y 2 300,
3y 1 +2y 2 480,
y 1 0, y 2 0
This is the dual of the given minimum problem. This duality
relationship is significant because of the following duality
principle:
The optimal solution of a minimum linear programming
problem, if the solution exists, has the same value as the
optimal solution of the maximum problem, that is, its dual.
In obtaining the dual of a minimum problem following steps
are employed :
Step 1: Write the minimum problem in standard form.
Step 2: Construct the special matrix from the constraints and
the objective function.

Duality :: 75

Step 3: Interchange the rows and columns to form the special

matrix of the dual problem.
Step 4: Translate this matrix into a maximum problem in
standard form.
Example C :
Find the dual of the minimum problem
Minimize z = 2x 1+3x2
subject to the constraints
2x 1+x 2 6,
x1 +2x 2 4,
x1 +x 2 5,
x1 0, x2 0
Solution :
Observe that the minimum problem is in the standard form. The
special matrix will be
2

Interchanging the rows and columns, we get

2

Translate it into the following maximum problem

Maximize z = 6y 1 +4y 2+5y 3
subject to the constraints
2y 1 +y 2+y 3 2,
y 1+2y 2+y 3 3,
y 1 0, y 2 0, y 3 0
This is a maximum problem in standard form and the dual of
the original problem.

76 :: Mathematics

Some observations about this example:

1. The variables (x1, x2 ) of the minimum problem have different
names from the variables of its dual problem (y1, y2, y3).
2. The minimum problem has three constraints and two
variables, while the dual problem has two constraints and
three variables (In general, if a problem has m
constraints and n variables, its dual will have n
constraints and m variables).
3. The inequalities defining the constraints are for the
minimum problem and for the maximum problem.
4. Since the coefficients in the minimal objective function are
positive, the dual problem has nonnegative numbers to
the right of the signs.
Example D :
Obtain the dual of
Minimize z = 16x1 +9x 2+21x3
subject to the constraints
x1 +x 2+3x3 16,
2x1 +x 2 +x3 12,
x1 0, x2 0, x3 0,
Solution :
We observe that the minimum problem is in the standard form.
The special matrix will be
1

16

12

16

21

Interchanging the rows and columns, we get

1

16

21

16

12

Duality :: 77

Translating it into the maximum problem, i.e., the dual of the

minimum problem is
Maximize z 1 = 16y 1+12y 2
subject to the conditions
y 1+2y 2 16,
y 1+y 2

9,

3y 1 +y 2 21,
y 1 0, y 2 0
Check point 2:
Write the dual of the following linear programming problem
Minimize z = 2x1 +x 2
subject to the conditions
3,

3x 1+x 2

4x 1+3x2 6,
2,

x1 +2x 2

x1 0, x2 0,
Example E :
Find the dual of the minimum problem
Minimize z = 2x 1+3x2
subject to the constraints
2x1 +x 2 6,
x1 +2x 2 4,
x1 +x 2

5,

x1 0, x2 0,
and find its optimal solution.
Solution :
In the previous example we have found the dual of this minimum
problem, which is
Maximize z 1 = 6y 1 +4y 2 +5y 3
Subject to the conditions

78 :: Mathematics

2y 1 +y 2 +y 3 2,
y 1+2y 2 +y 3 3,
y 1 0, y 2 0, y 3 0,
This is a maximum problem in the standard form. So, we apply
the simplex method to solve this problem (as in the previous lesson).
We introduce slack variables s1 and s2 to get
2y 1 +y 2 +y 3+s1 = 2
y 1+2y 2+y 3+s2 = 3
The initial simplex table is

y1

y2

y3

s1

s2

s1

s2

Here we have drawn an extra vertical line in the simplex table

(as compared to the previous lesson) to get off the columns of
s1 and s2 . (As you will see, the slack variable columns play a
special role in the solution of a dual problem).
As discussed in the previous lesson, we find that the pivot column
is column 1 and the pivot row is row 1. The pivot element is
circled in the above table. Dividing row 1 by 2, we get
y1

y2

y3

s1

s2

1
2

1
2

1
2

s1

s2

Applying R2 R1 ,

R 3 +6R 1 , we get

y1

y2

y3

s1

s2

1
2

1
2

1
2

y1

3
2

1
2

1
2

s2

Duality :: 79

The pivot column is column 3 and the pivot row is row 1. The
1
pivot element is circled one. Dividing R 1 by
, we get
2
y1
y2
y3
s1
s2

0
0

y1

s2

1
3
1
22

Applying R 2

R1 ,

R 3 +2R 1 , we get

y1

y2

y3

s1

s2

y3

s2

10

Here we see that there is no negative entry in the objective row.

Thus, we have found the optimal solution. The solution to the
maximum problem is
z1 = 10, y 1 = 0, y 2 = 0, y 3 = 2
The duality principal states that the minimum value of the
objective function in the original problem is the same as the
maximum value in the dual, that is,
Min. z = 10
But which value of x 1

and x2 will yield this minimum value?

The entries in the objective row will give the values of x1 and x2
corresponding to the entries in the columns headed by slack
variables. The entry of the objective row corresponding to s1 will
give x 1 and similarly the entry corresponding to s2 will give x 2.
Hence, in this problem
x1 = 5, x 2 =0, z 1 =10
Moreover, the entire minimal solution can be read from the right
end of the objective row of the final simplex table.
The following steps be employed in solving a minimal problem:
Step 1: Write the dual (maximum) problem.
Step 2: Solve this maximum problem by simplex method.

80 :: Mathematics

Step 3: Read the optimal solution for the original problem from
the objective row of the final simplex table. The variables
will appear as the last entries in the columns
corresponding to the slack variables.
Step 4: The minimum value of the objective function z will
appear in the lower right corner of the final table and
is equal to the maximum value of the dual objective
function z 1 .
Example F :
Minimize z = 6x1 +8x 2+x 3
subject to the conditions
3x1 +5x 2+3x3 20,
x1 +3x 2+2x3

9,

6x1 +2x 2 +5x3 30,

x1 0, x 2 0, x 3 0
Find the optimal solution of the above problem.
Solution :
This minimum problem is in standard form. The special matrix
of this problem is
3

20

30

Interchanging rows and columns, we get

3

20

30

Duality :: 81

The dual problem is

Maximize z1 = 20y 1+9y 2+30y 3
subject to the conditions
3y 1 +y 2+6y 3

6,

5y 1+3y 2+2y 3

8,

3y 1+2y 2+5y 3

1,

y 1 0, y 2 0, y 3 0
This maximum problem is in the standard form. We introduce
slack variables s1 , s2 , s 3 to get
3y 1+y 2+6y 3+s 1

= 6,

5y 1+3y 2+2y 3+s 2 = 8,

3y 1+2y 2+5y 3+s 3 = 1,
The initial simplex table is

1
3
2
5

y1

y2

y3

s1

s2

s3

s1

s2

s3

20

30

The pivot column is column 3 and pivot row is row 3.

The pivot element is circled in the intial simplex table.
Dividing row 3 by 5, we get
y1

y2

y3

s1

s2

s3

s1

s2

30

20

s3
0

82 :: Mathematics

Applying R1 6R3 , R 2 2R 3 , R 4+30R 3 , we get

y1

y2

y3

s1

s2

s3

s1

s2

y3

The pivot column is column 1 and the pivot row is row 3.

The pivot element is circled in the above simplex table.
Dividing row 3 by
y1

y2

y3

s1

s2

s1

s2

y3

1
2

, we get

s3

Applying R 1+

3
5

R3 , R2

19
5

R 3 , R 4 +2R 3, we get

y1

y2

y3

s1

s2

s3

s2

y1

s1

Duality :: 83

Now all the entries in the objective row are greater than or
equal to zero. So this is the final simplex table and the optimal
solution has been reached.
The solution to the maximum problem is
Max. z 1 =

y1 =

, y 2 = 0,

y3 = 0

For the minimum problem the values of x 1, x 2, and x3 are read

as the last entries in the column under s1 , s 2 and s 3 respectively.
Hence, the optimal solution to the minimal problem is
x 1 = 0,

x2 = 0,

x3 =

and the minimum value is z =

Check point 3:
Minimize z = 6x 1+3x2
subject to the constraints
x1 +x 2 2,
2x 1+6x2 6,

1
20
33

x1 0, x2 0
Find the optimal solution of the above problem.

INTEXT QUESTIONS 61.1

1.

Find the dual of the following problems:

(a) Minimize z = 2x1 +x 2

subject to the conditions
3x 1+x 2

3,

4x 1+3x2 6,
x1 +2x 2

2,

x1 0, x2 0
(b) Maximize z = 3x1+4x 2
subject to the conditions
x1 x2

1,

84 :: Mathematics

x1 +x 2 0,
x1 0, x2 0
(c) Maximize z = 4x1 +3x 2
subject to the conditions
x1 6,
x2 8,
x1 +x 2 7,
3x 1+x2 15
x2 1,
x1 0, x2 0
(d) Maximize z = 3x 12x 2
subject to the conditiosn
x1 4,
x2 6,
x1 +x 2 5,
x2 1,
x1 0, x 2 0

61.6 DUAL FOR THE MAXIMUM PROBLEMS AND

THE LINEAR PROGRAMMING PROBLEMS
WITH MIXED CONSTRAINTS
In the preceeding section we have obtained the dual for minimal
problem and find the solution of the minimal problem by solving
the corresponding dual problem. In this section, we will find the
dual for the maximal problems and for the problems which have
the mixed constraints.
Example G :
Give the dual of the following problem:
Max. z = x1 +2x 2
subject to the constraints
2x1 3x2 3,
4x 1+x 2

4,

x1 0, x2 0

Duality :: 85

Solution :
Here the maximum problem is in standard form. We begin by
constructing a special matrix for the coefficients of the constraints
of this problem without introducing slack variables. As in the
simplex table, we place the objective function in the last row.
The special matrix is
2

Interchaning rows and columns, we get

2

Translating this matrix into a minimum problem in the standard

form, we get the dual of the maximum problem as:
Minimum z 1 = 3y1 +4y 2
subject to the constraints
2y 1+4y 2 1,
3y 1+y 2 2,
y 1 0, y 2 0
Check point 4:
Find the dual of the problem
Maximum

z = x1 +3x 2

subject to the constraints

3x1 +2x2 6,
3x 1+x 2

4,

x1 0, x2 0

86 :: Mathematics

Example H :
Find the dual of the following problem :
Min. z = 2x 1+2x 2+4x3
subject to the constraints
2x1 +3x 2 +5x3 2,
3x1 +x 2+7x3

3,

x1 +4x 2+6x3

5,

x1 0, x2 0, x 3 0
Solution :
This problem is not in the standard form because the second
and third constraints have the sign . In order to convert
into sign, multiply the second and third constraints by 1.
We get the original problem in the standard form as :
Min. z = 2x 1+2x 2+4x3
subject to the constraints
2x1 +3x 2 +5x3 2,
3x1 x 2 7x 3

3,

x 1 4x 2 6x 3

5,

x1 0, x2 0, x 3 0
The special matrix for this problem will be
2

Interchanging rows and columns, we get

2

Duality :: 87

Translating this matrix into a maximum problem in the standard

form, we get the dual of this problem which is
Max. z1 = 2y 1 3y 25y 3
subject to constraints
2y 13y 2 y 3

2,

3y 1y 2 4y 3

2,

5y 1 7y 2 6y 3 4,
y 1 0, y 2 0, y 3 0
Example I :
Find the dual of the following problem
Minimum z = x 1+x2 +x 3
subject to the constraints
x1 3x2 +4x 3 = 5,
2x1 2x2 3,
2x2 x3 5,
x1 0, x2 0, x 3 0
Solution :
This problem is not in the standard form. First we should convert
the given problem into the standard primal form. This problem
is of minimization so all the constraints must involve the
sign.
First constraint can be written as
x1 3x2 +4x 3 5,
x1 3x2 +4x 3 5,
First one is not in the standard form yet. So multiplying it by
1, we get
x1 +3x 24x3 5,
Again multiplying 2x1 2x2 3 by 1, we get
2x1 +2x 2 3,

88 :: Mathematics

Thus, the given problem in standard form takes the form

Minimum z = x 1+x2 +x 3
subject to the constraints
x1 3x2 +4x 3 5,
x1 +3x 24x3 5,
2x1 +2x 2 3,
2x 2x3 5,
x1 0, x2 0, x 3 0
The special matrix of this problem is
1

Interchanging rows and columns, we get

1

Translating this matrix into a maximum problem in standard

form, we get the dual problem as
Maximum z 1 = 5y 1 5y 23y 3+5y 4
subject to the constraints
y 1y 2 2y3 1,
3y 1 +3y 2 +2y 3 +2y 4 1,
4y 1 4y 2 y 4 1,
y 1 0, y 2 0, y 3 0, y 4 0

Duality :: 89

Check point 5 :
Find the dual of the following problem
Maximum z = 2x1 +3x 2+x3
subject to the constraints
4x1 +3x 2+x3 = 6,
x1 +2x 2+5x3 = 4,
x1 0, x2 0, x 3 0

INTEXT QUESTIONS 61.2

1.

Find the dual of the following problems :

(a) Minimum z = 2x 1+3x2 +4x 3

subject to the conditions
2x1 +3x 2+5x3 2,
3x 1+x 2+7x3 3,
x1 +4x 2+6x3 5,
x 1 , x 2, x 3 0
(b) Minimum z = 3x1 2x2 +4x 3
subject to the conditions
3x1 +5x 2+4x3 7,
6x1 +x 2+3x3

4,

7x1 2x2 x3

10,

x1 2x2 +5x 3

3,

4x1 +7x 22x 3 2,

x 1 , x 2, x 3 0
(c) Maximum z = 3x1 +2x 2
subject to the conditions
x1 +x 2

1,

x1 +x 2

7,

x1 +2x 2 10,
x2 3,
x1, x 2 0

90 :: Mathematics

(d) Maximum z = x1 +3x 2

subject to the conditions
3x1 +2x2 6,
3x1 +x 2 = 4,
x1, x 2 0

WHAT YOU HAVE LEARNT

l

Associated with every linear programming problem there is

another intimately related linear programming problem, called
the dual problem of the original linear programming problem.
The original linear programming problem is called primal
problem.

Any minimum linear programming problem has the following

matrix representation:
Min Z = AX
subject to the following constraints
BX C
X 0
Then its dual will be
Max. Z = CY
subject to the following constraints
B Y A
Y 0
where A, B, C are the transpose of the matrices A,B and C
respectively.

If a problem has m constraints and n variables, its dual will

have n constraints and m variables.

The optimal solution of a minimum linear programming

problem, if the solution exists, has the same value as the
optimal solution of the maximum problem, i.e., its dual.

Duality :: 91

TERMINAL QUESTIONS
1. Find the dual of the following problems:
(a) Min. z = x1 +x 2
subject to the condition
2x 1+x 2

4,

x1 +7x 2

7,

x1, x 2

(b) Max. z = x1+2x 2

subject to the conditions
2x1 3x2

3,

4x 1+x 2

4,

x1, x 2

2. Find the dual of the following problems with mixed constraints:

(a) Min. z = 3x1 2x2 +x 3
subject to the conditions
2x1 3x2 +x 3 = 1,
2x1 +3x2 8,
x1, x 2, x 3 0
(b) Max. z = 2x1 +3x 2+x 3
subject to the conditions
4x1 +3x2+x3 = 6,
x1 +2x 2+5x3 = 4,
x1, x 2, x 3 0
(c) Min. z = x2 3x3 +2x 5
subject to the conditions
x1 +3x 2x 3+2x 5 = 7,
2x2 +4x 3+x 4 = 12
4x2 +3x 3 +8x 5+x6 = 10,
x 1 , x 2, x 3, x 4 , x 5 , x 6 0

92 :: Mathematics

3. Write the dual of the following problems and solve by the

simplex mehtod.
(a) Max. z = x 1+x2 +x 3
subject to the conditions
2x1 +x 2+2x3

2,

4x1 +2x2 +x3

2,
0

x1, x 2, x3

(b) Min. z = 2x1 3x2

subject to the conditions
2x1 x2 x3 3,
x1 x2 +x 3

2,

x 1 , x 2, x 3 0

ANSWERS TO CHECK POINTS

Check point 1:
Standard Form
Check point 2:
Max. z1 = 3y 1 + 6y 2 + 2y 3
subject to the constraints
3y 1+ 4y 2 + y 3 2,
y 1+ 3y 2 + 2y 3 1,
y 1 0, y 2 0, y 3 0
Check point 3:
Min z = 6, x 1 = 0, x2 = 2
Check point 4:
Min. z1 = 6y 1+ 4y 2
subject to the constraints
3y 1 +3y 2 1,

Duality :: 93

2y 1 +y 2 3,
y 1 0, y 2 0.
Check point 5 :
Minimum z1 = 6y 16y 2 +4y 3 4y 4
subject to the constraints
4y 14y 2 +y 3 y 4 2,
3y 13y 2 +2y 3 2y 4 3,
y 1y 2 +5y 3 5y 4 1,
y 1 0, y 2 0, y 3 0, y 4 0

ANSWERS TO INTEXT QUESTIONS

61.1
1

(a) Max. z = 3y 1 +6y 2+2y 3

subject to the constraints
3y 1 +4y 2+y 3 2,
y 1+3y 2+2y 3 1,
y 1, y 2 , y3 0
(b) Min. z = y 1
subject to the constraints
y 1y 2

3,

y 1+y 2 4,
y 1, y 2 0
(c) Min. z = 6y 1+8y 2+7y 3+15y 4+y 5
subject to the constraints
y 1+y 3+3y 4 4,
y 2+ y 3 + y 4y 4 3,
y 1, y 2 , y 3, y 4 , y 5 0
(d) Min z = 4y 1+6y 2+5y 3y 4
subject to the constraints

94 :: Mathematics

y 1+y 3 3
y 2+y 3y 4 2
y 1, y 2 , y 3, y 4 0
61.2
1

(a) Maximum z = 2y 13y 25y 3

subject to the constraints
2y 13y 2 y 3

2,

3y 1y 2 4y 3

2,

5y 1 7y 2 6y 3 4,
y 1, y 2 , y 3

(b) Maximum z = 7y 1+4y 2 10y 3+3y 4+2y 5

subject to the conditions
3y 1+6y 2 7y 3 +y 4 +4y 5

3,

5y 1+y 2+2y 32y 4 +7y 5

2,

4y 1+3y 2 +y 3+5y 42y 5

4,

y 1, y 2 , y 3, y 4 , y 5

(c) Min. z = y 1+7y 2 10y 3+3y 4

subject to the conditions
y 1+y 2y 3 3,
y 1+y 22y 3+y 4 2,
y 1, y 2 , y 3, y 4

(d) Min. z = 6y 1+4y 2 4y 3

subject to the conditions
3y 1 +3y 2 3y 3 1,
2y 1 +y 2 y 3

3,

y 1, y 2 , y 3,

Duality :: 95

ANSWERS TO TERMINAL QUESTIONS

1. (a) Min. z = 6y 1+4y 2 4y 3
subject to the conditions
3y 1 +3y 23y 3 1,
2y 1 +y 2y 3

3,

y 1, y 2 , y 3, 0
(b) Min. z = 3y1 +4y 2
subject to the conditions
2y 1 +4y 2 1,
3y 1+y 2 2,
y 1, y 2 0
2. (a) Max. z = y 1 y2 +8y 3
subject to the conditions
2y 1 2y 2 +2y 3 3,
3y 1+3y 2+3y 3 2,
y 1 y 2 1,
y 1, y 2 , y 3 0
(b) Min. z = 6y 16y 2+4y 34y 4
subject to the conditions
4y 1 4y 2 +y 3 y4 2,
3y 13y 2 +2y 3 2y 4 3,
y 1y 2 +5y 3 5y 4 1,
y i 0, i = 1,2,3,4
(c) Max. z = 7y 17y 2 +12y 3 12y 4+10y 510y 6
subject to the conditions
y1 y2

3y 13y 2 2y 3+2y4 4y 5+4y 6 1,

y 1+y 2 +4y 3 4y 4 +3y 5 3y 6
y 3 y 4 0,

3,

96 :: Mathematics

2y 12y 2 +8y 5 8y 6
y 5 y 6 0,

y 1, y 2 , y 3, y 4 , y 5 , y 6 0
3. (a) Dual problem is
Min. z = 2y 1+2y 2
subject to the conditions
2y 1 +4y 2 1,
y 1+2y 2

1,

2y 1 +y 2

1,

y i 0, i = 1,2
Solution to the dual problem is
z =

, y1 =

, y2 =

(b) Dual problem is

Max. z = 3y 1+2y 2
subject to the conditions
2y 1 +y 2 2,
y 1y 2 3,
y 1+y 2

0,

Solution of the primal problem is unbounded.

The dual has no feasible solution.