Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
4.1 Introduction
Every linear programming problem (LPP) is associated with another linear
programming problem involving the same data and optimal solutions. Such
two problems are said to be duals of each other. One problem is called the
primal, while the other problem is called the dual.
The dual formulation is derived from the same data and solved in a manner
similar to the original 'primal' formulation. In other words, you can say that
dual is the 'inverse' of the primal formulation because of the following
reasons.
If the primal objective function is 'maximisation' function, then the dual
objective function is 'minimisation' function and vice-versa.
The column co-efficient in the primal constraint is the row co-efficient in
the dual constraint.
The co-efficients in the primal objective function are the RHS constraint
in the dual constraint.
Sikkim Manipal University Page No. 75
Operations Research Unit 4
The RHS column of constants of the primal constraints becomes the row
of co-efficient of the dual objective function.
The concept of duality is useful to obtain additional information about the
variation in the optimal solution. These changes could be effected in the
constraint co-efficient, in resource availabilities and/or objective function co-
efficient. This effect is termed as post optimality or sensitivity analysis.
Learning objectives
By the end of this unit, you should be able to:
Describe the duality concept
Solve a dual problem
Describe the economical interpretation
Apply sensitivity analysis
Formation of dual LPP is easier when the standard form of LPP for
maximisation problem must contain “≤” type of constraints, while for
minimisation problem, it must contain “≥” type of constraints.
Solved Problem 1
Write dual of
Max Z = 4x1 + 5x2
Subject to 3x1 + x2 ≤ 15
x1 +2 x2 ≤ 10
5x1 + 2x2 ≤ 20
x1, x2, ≥ 0
Solution: The given problem is in its standard form. Therefore, its dual is
Mini W = 15y1 + 10 y2 + 20 y3
Subject to 3y1 + y2 + 5 y3 ≥ 4
y1 + 2y2 + 2y3 ≥ 5
y1, y2, y3, ≥ 0
Solved Problem 2
Write the dual of
Min Z = 10x1 + 12x2
Subject to 2x1 +3 x2 ≥ 10
5x1 +6 x2 ≥ 20
x1 + 2x2 ≥ 15
2x1 + 3x2 ≥ 12
x1, x2, ≥ 0
Solution: The given problem is in its standard form. Therefore, its dual
problem is
Max W = 10y1 + 20 y2 + 15 y3 + 12y4
Subject to 2y1 + 5y2 + y3 = 2y4 ≤ 12
6y1 + 2y2 + 3 y3 ≤ 10
y1, y2, y3, ≥ 0
Solved Problem 3
Write the dual of
Max Z = 100x1 + 200x2
Subject to 3x1 - 10 x2 ≥ 15
+4x1 +15 x2 ≤ 20
x1, x2, ≥ 0
Solution: First we convert
3x1 - 10x2 ≥ 15
to “≤” type as shown here.
-3x1 + 10x2 ≤ -15
Therefore, its dual is
Mini Z = -15y1 + 20 y2
Subject to -3y1 + 4y2 ≥ 100
10y1 + 15y2 ≥ 200
y1, y2, ≥ 0
Solved Problem 4
When the constraints contain “=” sign, write the dual of
Max Z = 40x1 + 30x2
Subject to 10x1 +6 x2 ≤ 15
5x1 +6 x2 ≥ -10
x1 + x2 = 9
x1 + x2 ≥ 10
x1, x2, ≥ 0
Solution: Firstly,
5x1 – 7x2 ≥ -10
is rewritten as
-5x1 +7x2 ≤ 10
Secondly,
x1 + x2 = 9
is written as
x1 + x2 ≤ 9
Also
x1 + x2 ≥ 9 this is same as -x1 - x2 ≤ -9
Solved Problem 5
Write the dual of
Max Z = 12x1 + 15x2
Subject to 5x1 +36 x2 ≥ 10
x1 + x2 ≤ 5
x1 ≥ 0
x2 is unrestricted in sign
Therefore, dual is
5y1 – y2 ≤ 12 ………….…(1)
3y1 – y2 ≤ 15 ………….…(2)
-3y1 + y2 ≤ - 15 ……………(3)
No.3 constraint is 3y1 – y2 ≥ 15 …………(4)
Constraints 2 and 4 give
3y1 – y2 = 15
The dual problem is
Max W = 10y1 –15 y2
Subject to 5y1 – y2 ≤ 12
3y1 – y2 =15
y1 ≥ 0 y2 unrestricted in sign
Solved Problem 6
Write the dual of the following LPP
Minimise Z = 3x1 – 2x2 + 4x3
Subject to 3x1 + 5x2 + 4x3 ≥ 7
6x1 + x2 + 3x3 ≥ 4
7x1 - 2x2 - x3 ≤ 10
x1 - 2x2 + 5x3 ≥ 3
4x1 + 7x2 - 2x3 ≥ 2,
x1, x2, x3 ≥ 0
Solved Problem 7
Find the dual of
Maximise 12x1 + 10x2
Subject to 2x1 + 3x2 ≤ 18
2x1 + x2 ≤ 14
x1, x2 ≥ 0
Solved Problem 8
Obtain the dual problem of the following primal formulation.
Maximise Z = 2x1 + 5x2 + 6x3
Subject to 5x1 + 6x2 -x3 ≤3
-2x1 + x2 + 4x ≤ 4
x1 - 5x2 + 3x3 ≤ 1
-3x1 - 3x2 + 7x3≤ 6
x1, x2, x3≥ 0
Solution: Step 1: Write the objective function of the dual. As there are
four constraints in the primal, the objective function of the dual will have 4
variables.
Minimise r* = 3 w 1 + 4 w 2 + w 3 + 6 w 4
Step 2: Write the constraints of the dual. As all the constraints in the
primal are '<', the constraints in the dual will be '>'. The column COM+
efficient of the primal becomes the row co-efficient of the dual.
Constraints 5w1 - 2w2 + w3 - 3w4 ≥2
6w1 + w2 - 5w3 -w4 ≥5
-w1 + 4w2 + 3w3 + 7w4≥ 6
Step 3 : Therefore the dual of the primal is :
Minimise Z = 3w1 + 4w2 + w3 + 6w4
Subject to: 5w1 - 2w2 + w3 - 3w4 ≥2
6w1+ w2 - 5w3 - 3w4 ≥ 5
- w1 + 4w2 + 3w3 + 7w4≥ 6
w1, w2, w3, w4 >= 0
Solved Problem 9
Obtain the dual of the following linear programming problem.
Minimise Z = 5x1 -6x2 + 4x3
Subject to the constraints: 3x1 + 4x2 + 6x3 ≥ 9
x1 + 3x2 + 2x3 ≥ 5
7x1 - 2x2 - x3 ≤ 10
x1 - 2x2 + 4x3 ≥ 4
2x1 + 5x2 - 3x3≥ 3
x1, x2, x3 ≥0
Solution: As you can see, one of the primal constraints is a "≤ " constraint
while the others are all "≥” constraints. The dual cannot be worked out
unless all the constraints are in the same direction. To convert this into "≥”
constraint, multiply both the sides of the equation by "-" sign. After
multiplying the constraint by "-" sign, it will become 7x1 + 2x2 + x3 ≥ -10.
Now that all the constraints are in the same direction and the dual can be
worked out, the dual formulation is:
Maximise Z = 9w1 + 5w2 - 10w3 + 4w4 + 3w5
Subject to: 3w1 + w2 - 7w3 + w4 + 2w5 ≤ 5
4w1 + 3w2 + 2w3 - 2w4 + 5w5 ≤ -6
6w1 + 2w2 + w3 + 4w4 + -3w5 ≤ 4
w1, w2, w3, w4, w5 ≥0
Primal Dual
Maximise Minimise
n m
z c j. x j w bi . y i
j1 i1
Subject to Subject to
n m
xj ≥ 0, j = 1,2,…., n yj ≥ 0, i = 1,2,…., m
In the above resource allocation model, the primal problem has „n‟ economic
activities and „m‟ resources. The coefficient cj in the primal represents the
profit per unit of activity j. Resource i, whose maximum availability is bi, is
consumed at the rate aij units per unit of activity j.
4.4.1 Economic interpretation of dual variables:
For any pair of feasible primal and dual solutions,
(Objective value in the maximisation problem) ≤ (Objective value in the
minimisation problem).
At the optimum, the relationship holds as a strict equation.
Note: Here, the sense of optimisation is very important.
Sikkim Manipal University Page No. 84
Operations Research Unit 4
Hence, for any two primal and dual feasible solutions, the values of the
objective functions, when finite, must satisfy the following inequality.
n m
z cj xj bi y i w
j 1 i 1
The strict equality, z = w, holds true when both the primal and dual solutions
are optimal. Consider the optimal condition z = w. Given that the primal
problem represents a resource allocation model, you can think of z as
representing the profit in rupees. bi represents the number of units available
of the resource i. Therefore, we can express the equation z = w as profit
(Rs) = (units of resource i) x (profit per unit of resource i)
This means that the dual variables yi, represent the worth per unit of
resource i.
Note: Variables yi are also called as dual prices, shadow prices and simplex
multipliers.
With the same logic, the inequality z < w associated with any two feasible
primal and dual solutions is interpreted as (profit) < (worth of resources).
This relationship implies that as long as the total return from all the activities
is less than the worth of the resources, the corresponding primal and dual
solutions are not optimal. Optimality is achieved only when the resources
have been exploited to their fullest extent, which can happen only when the
input equals the output (profit). Economically, the system is said to be
unstable (non optimal) when the input (worth of the resources) exceeds the
output (return). Stability occurs only when the two quantities are equal.
Caselet
Luminous lamps produce three types of lamps A, B and C. These lamps
are processed on three machines X, Y and Z.
To find out the optimal solution in order to maximise profit and minimise
cost, first you find out whether a previously determined optimal solution
remains optimal if the contribution rate is changed. An increase in Cj of a
variable would mean that resources from other products should be
diverted to this more profitable product.
The reverse is true for a minimisation problem.
Solved Problem 10
For the above scenario, the full technology and input restrictions are given
in the following table.
Table 4.1: Lamp production and profit per unit
Machine
Product Profit Per Unit
M1 M2 M3
A 10 7 2 12
B 2 3 4 3
C 1 2 1 1
Available time 100 77 80
Cj 12 3 1 0 0 0
Basic Solution values
CB variables x1 x2 x3 X4 x5 x6 b (=xB )
B
1 3 1 73
x1 – –
12 1 0 16 16 8 0 8
3 x2 0 1
13 –
7 5 0
35
16 16 8 8
17 11 9 177
0 x6 0 0 1
8 8 4 4
27 15 3
zj 12 3 0
16 16 8
11 15 3
zj – cj 0 0 0
16 16 8
From the above final simplex table, you can see that profit contribution for
product C is Re 1, which is not greater than its zj. Therefore, to bring x3 into
27
the basis, its profit contribution rate cj must exceed Rs. to make zj – cj
16
value negative or zero that is zj – cj 0. Specifically,
If cj* – cj > zj – cj, then a new optimal solution must be derived
If cj* – cj = zj – cj, then alternative optimal solution exist.
If cj* – cj < zj – cj, then current optimal solution remains unchanged.
11
In case of c3 = 1 and z3 – c3 = ,
16
11
C3* – 1
16
11 27
C3* +1 =
16 16
x3 can be introduced in the basis, if its contribution rate c3 increases at least
29
Rs . If it increases beyond that, then the current solution will no longer
16
be optimal.
4.5.2 Change in Cj of a basic variable
Consider the case of product A (x1 column) and divide each zj - cj entry in the
index row by the corresponding co-efficient in the x1 row as shown below:
2j Cj 2j Cj
- Minimum ; y ij 0 Minimum ; y ij 0
y ij 1 y ij
Referring to the final simplex table, you observe that corresponding to the
non basic variables x3 and x5, y13, y15 <0.
Therefore,
11 3
Minimum 16 , 8 = minimum (11, 3) = 3
1 1
16 8
Corresponding to the non- basic variable x4 ,y14 >0, therefore,
15
Minimum 16 =5
3
16
Hence,
–5 c1* -12 3 i.e. 7 c1* 15
Thus, you can say that the optimal solution is insensitive as long as the
changed profit coefficient c1* varies between Rs 7 and Rs. 15
4.5.3 Change in available resources
If the available resources change, you need to investigate whether a
previous optimal solution remains feasible or not. For long term planning, it
is important to know the bounds within which each available resource (for
instance machine hours) can vary without causing violent changes in the
current optimal solution. To illustrate this, divide each quantity in the XB
column by the corresponding coefficient in the X4 column of the table.
Sikkim Manipal University Page No. 89
Operations Research Unit 4
XB
XB X4
X4
73 3 146
8 16 3
35 7
– 10
8 16
177 11 354
4 8 11
354
The least positive ratio indicates the number of hours of machine M1
11
that can be decreased. The least negative ratio (–10) indicates the increase
in the number of hours of machine M1.
4.5.4 Calculating the range
354 746
Lower limit = 100 – =
11 11
Upper limit = 100 – (–10) = 110
746
Therefore, the range of hours for M1 is to 110.
11
You can calculate the range of hours for machine M2 and M3 in the same
manner. The management of a company rarely restricts its interest to the
numerical values of an optimal solution. It is interested in knowing the
impact of changes in the input parameter values on the optimal solution.
This process is called sensitivity analysis.
4.6 Summary
For every linear programming problem there exists a dual linear
programming problem. The dual LPP helps us to reduce the amount of
calculation involved in original the LPP. They also help us to interpret the
economic variables more effectively.
7. False
8. True
9. Final
10. Resource, levels
11. Upper, lower, bounce
4.9 References
The following sources have been referred to for additional content for the
unit.
http://www.psnacet.edu.in/courses/MBA/Linearprogramming/lecture-
11.pdf
http://agecon2.tamu.edu/people/faculty/mccarl-bruce/mccspr/new04.pdf
http://www.sce.carleton.ca/faculty/chinneck/po/Chapter6.pdf.
http://www-ee.eng.hawaii.edu/~jyee/491.06/lectures/chapter06.pdf