Introduction to Modeling and

Optimization
Instructor: Associate Professor Yasin Göçgün
e-mail: [email protected]
Week-3 – Lecture 1: Duality Theory
13 Duality theory

The theory of duality is a very elegant and important concept within the
field of OR.

The notion of duality within LP asserts that every linear program has
associated with it a related linear program called its dual.

The original problem in relation to its dual is termed the primal.

It is the relationship between the primal and its dual, both on a
mathematical and economic level, that is truly the essence of duality
theory.
14 Duality theory

The ‘Dual’ problem is an LP defined directly and systematically from the

original (or Primal) LP model.

The optimal solution of one problem yields the optimal solution to the other.

Duality ease the calculations for the problems, whose number of variables is
large.
15 Rules for converting Primal to Dual

If the Primal is to maximize, the dual is to minimize.

If the Primal is to minimize, the dual is to maximize.
For every constraint in the primal, there is a dual variable.
For every variable in the primal, there is a constraint in the dual.
16 Example
There is a small company in Melbourne which has recently become
engaged in the production of office furniture. The company manufactures
tables, desks and chairs :

 The production of a table requires 8 kg of wood and 5 kg of metal and is

sold for $80; a desk uses 6 kg of wood and 4 kg of metal and is sold for
$60; and a chair requires 4 kg of both metal and wood and is sold for $50.
 We would like to determine the revenue maximizing strategy for this
company, given that their resources are limited to 100 kg of wood and 60
kg of metal.
17 Example-Lp model
Let’s make an LP model quickly:

Maximize Z = 80 x1 + 60 x2 + 50 x3
subject to
8 x1 + 6 x2 + 4 x3 ≤ 100
5 x1 + 4 x2 + 4 x3 ≤ 60
x1, x2, x3 ≥ 0.

Now consider that there is a much bigger company in Melbourne which has been
the lone producer of this type of furniture for many years.

They don't appreciate the competition from this new company; so they have
decided to tender an offer to buy all of their competitor's resources and therefore
put them out of business.
18 Example-Lp model
The challenge for this large company then is to develop an LP model which will
determine the appropriate amount of money that should be offered for a unit of
each type of resource,
such that the offer will be acceptable to the smaller company while minimizing
the expenditures of the larger company.
Minimize W = 100 y1+60 y2
subject to
8 y1+ 5 y2 ≥ 80
6 y1+ 4 y2 ≥ 60
4 y1+ 4 y2 ≥ 50
y1, y2 ≥ 0.
19 Primal vs. Dual
 Now, let’s compare both Primal and Dual models

 Check the number of variables and constraints

Check the structure of objective functions and constraints

20 Primal and Dual conversion

Dual model can be found by using the following converting table:

For example:
21 Another Example
Consider the following Primal model:

Maximize Z = 3 x1 + 5 x2 Minimize W = 4 y1 + 12 y2 + 18 y3
subject to subject to
x1 ≤4 y1 + 3 y3 ≥ 3
2 x2 ≤ 12 2 y2 + 2 y3 ≥ 5
3 x1+2 x2 ≤ 18 y1, y2, y3 ≥ 0.
x1, x2 ≥ 0.

We define one dual variable for each primal constraint.

The Primal has: 2 variables and 3 constraints.

So the Dual has: 3 variables and 2 constraints

22 Advantage of dual problems

Consider the following Primal model:

Minimize Z = 2 x1 + 3 x2 + 4 x3 Maximize W = 8 y1 + 6 y2
subject to subject to
2 x1 + 4 x3 ≥ 8 2 y1 ≥2
3 x2 + 2 x3 ≥ 6 3 y2 ≤ 3
x1 ≤ 0, x2, x3 ≥ 0. 4 y1+2 y2 ≤ 4
y1, y2 ≥ 0.

How can we find the optimal solution using Graphical Method?

Is Dual model the fastest way to find the optimal solution?
23 Advantage of dual problems

Maximize W = 8 y1 + 6 y2
subject to
2 y1 ≥2
3 y2 ≤ 3
4 y1+2 y2 ≤ 4
y1, y2 ≥ 0.

The feasible region is just a point in the coordinate system; that is (y1, y2)=(1, 0).

Therefore, the optimal solution is located on this point:

(y*1, y*2)=(1, 0) and W*=8=Z*

24 Relations between Primal and Dual

The dual of the dual problem is again the primal problem.

Either of the two problems has an optimal solution if and only if the other
does;
if one problem is feasible but unbounded, then the other is infeasible;
if one is infeasible, then the other is either infeasible or
feasible/unbounded.
25 Weak Duality VS. Strong Duality

Weak Duality Theorem: The objective function value of the primal (dual)
to be maximized evaluated at any primal (dual) feasible solution cannot
exceed the dual (primal) objective function value evaluated at a dual (primal)
feasible solution.
cx ≤ yb (in the standard equality form)

Strong Duality Theorem: When there is an optimal solution, the optimal

objective value of the primal is the same as the optimal objective value of the
dual.
c x* = y* b
26 Four Possible Primal Dual Problems

1)Finite optimum vs. Finite optimum

2)Unbounded vs. Infeasible
3)Infeasible vs. Infeasible
4)Infeasible vs. Unbounded

Dual
Finite optimum Unbounded Infeasible
Primal

Finite optimum 

Unbounded 

Infeasible  
27 Examples
Let’s check some examples of different modes
Finite optimum VS. Finite optimum
Primal: Min Z = 2 x1 + 3 x2 Dual: Max W = 8 y1 + 6 y2
subject to subject to
x1 ≥8 y1 + 3 y2 ≤ 2
3 x1 + 2 x2 ≥ 6 2 y2 ≤ 3
x1, x2≥ 0. y1, y2 ≥ 0.

(x*1, x*2)=(8, 0) and Z*=16 (y*1, y*2)=(2, 0) and W*=16

28 Examples

Unbounded VS. Infeasible

Primal: Max Z = 2 x1 + 3 x2 Dual: Min W = 8 y1 + 6 y2

subject to subject to
x1 ≥8 y1 + 3 y2 ≥ 2
3 x1 + 2 x2 ≥ 6 2 y2 ≥ 3
x1, x2≥ 0. y1, y2 ≤ 0.

x*1 and/or x*2 → +∞ then Z* → +∞ Infeasible

29 Examples
Infeasible VS. Unbounded
Primal: Max Z = 2 x1 + 3 x2 Dual: Min W = 8 y1 + 6 y2
subject to subject to
x1 ≥8 y1 + 3 y2 ≥ 2
3 x1 + 2 x2 ≤ 6 2 y2 ≥ 3
x1, x2≥ 0. y1≤ 0, y2 ≥ 0.

Infeasible y*1 → -∞ then W* → -∞

30 Dual Simplex Method
Suppose a “Basic Solution” satisfies the optimality condition but not feasible,

then we apply dual Simplex method.

Minimize Z = 3x1 + 2x2

subject to
3x1 + x2 ≥ 3
4x1 + 3x2 ≥ 6
x1 + x2 ≤ 3
x1, x2 ≥ 0

In regular Simplex method, we start with a Basic Feasible Solution (which is not
optimal - usually the origin) and move towards optimality always retaining feasibility.
31 Dual Simplex Method
In the dual Simplex method, the exact opposite occurs.

We start with an “optimal” solution (which is not feasible) and move towards
feasibility always retaining optimality condition.

The algorithm ends once we obtain feasibility.

To start the dual Simplex method, the following two conditions are to be met:
The objective function must satisfy the optimality conditions of the regular
Simplex method.
All the constraints must be of the type ≤.
32 Dual Simplex Method
Now, consider the previous example:
Minimize Z = 3x1 + 2x2
subject to
3x1 + x2 ≥ 3
4x1 + 3x2 ≥ 6
x1 + x2 ≤ 3
x1, x2 ≥ 0

Step 1: The first two inequalities are multiplied by –1 to convert them to ≤

constraints and convert the objective function into maximization function (standard
form).
Maximize ZꞋ = - 3x1 - 2x2 where ZꞋ = - Z
33 Dual Simplex Method
Maximize ZꞋ = - 3x1 - 2x2
subject to
-3x1 - x2 ≤ -3
-4x1 - 3x2 ≤ -6
x1 + x2 ≤ 3
x1, x2 ≥ 0

Let s1, s2 and s3 be three slack variables

Model can rewritten as:

ZꞋ + 3x1+ 2x2 =0
-3x1 - x2 + s1 = -3
-4x1 - 3x2 + s2 = -6
x1 + x2 + s3 = 3
34 Dual Simplex Method
Initial Basic Solution is:
x1= 0, x2= 0, s1 = -3, s2 = -6, s3 = 3 and ZꞋ =0.

Let’s start the dual Simplex method:

Basic ZꞋ x1 x2 s1 s2 s3 RHS

ZꞋ 1 3 2 0 0 0 0

s1 0 -3 -1 1 0 0 -3

s2 0 -4 -3 0 1 0 -6
s3 0 1 1 0 0 1 3

Initial Basic Solution is Optimal (as the optimality condition is satisfied) but
infeasible.
35 Dual Simplex Method
Basic ZꞋ x1 x2 s1 s2 s3 RHS

ZꞋ 1 3 2 0 0 0 0

s1 0 -3 -1 1 0 0 -3

s2 0 -4 -3 0 1 0 -6

s3 0 1 1 0 0 1 3

Choose the most negative basic variable. Therefore, s2 is the leaving variable.
Minimum ratio test:
Calculate Ratio = |Z-row / s2-row| (values < 0 in s2-row)
Choose minimum ratio. Therefore, x2 is the entering variable
x1 x2
|3/-4|= 3/4 |2/-3|= 2/3
36 Dual Simplex Method
Basic ZꞋ x1 x2 s1 s2 s3 RHS

ZꞋ 1 1/3 0 0 2/3 0 4

s1 0 -5/3 0 1 -1/3 0 -1

x2 0 4/3 1 0 -1/3 0 2

s3 0 -1/3 0 0 1/3 1 1

Therefore, s1 is the leaving variable and x1 is the entering variable.

 Minimum ratio test:

x1 s2
1/5 2
37 Dual Simplex Method
Basic ZꞋ x1 x2 s1 s2 s3 RHS

ZꞋ 1 0 0 1/5 3/5 0 21/5

x1 0 1 0 -3/5 1/5 0 3/5

x2 0 0 1 4/5 -3/5 0 6/5

s3 0 0 0 -1/5 2/5 1 6/5

Optimal Solution is : x1*= 3/5, x2*= 6/5, Z* = 21/5

 Here, we do not need to convert ZꞋ to Z!

Lecture Resources
• Lecture Notes, Erfan Babaee Tirkolaee, Istinye
University