Scribd
Upload
59 views25 pages

Duality Theorems Finding The Dual Optimal Solution From The Primal Optimal Tableau

This document discusses duality theorems and finding the dual optimal solution from the primal optimal tableau. It presents the primal and dual linear programming problems in matrix form. It proves the weak duality theorem and corollaries. It provides examples to show an infeasible primal problem with an unbounded dual, and vice versa. The document also presents a theorem stating that the dual optimal solution can be found by taking the negative inverse of the primal optimal basis matrix multiplied by the primal objective coefficients vector.

Uploaded by

Potnuru Vinay
Copyright
© Attribution Non-Commercial (BY-NC)
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPT, PDF, TXT or read online on Scribd
59 views25 pages

Duality Theorems Finding The Dual Optimal Solution From The Primal Optimal Tableau

This document discusses duality theorems and finding the dual optimal solution from the primal optimal tableau. It presents the primal and dual linear programming problems in matrix form. It proves the weak duality theorem and corollaries. It provides examples to show an infeasible primal problem with an unbounded dual, and vice versa. The document also presents a theorem stating that the dual optimal solution can be found by taking the negative inverse of the primal optimal basis matrix multiplied by the primal objective coefficients vector.

Uploaded by

Potnuru Vinay
Copyright
© Attribution Non-Commercial (BY-NC)
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PPT, PDF, TXT or read online on Scribd
You are on page 1/ 25

Duality theorems

Finding the dual optimal


solution from the primal
optimal tableau
In this lecture we shall present the primal
and dual problems in matrix form and
prove certain results on the feasible and
optimal solutions of the primal and dual
problems. The material of this lecture is
treated in Chapter 7.5 of your Textbook.
Dual problem in Matrix form
Suppose the primal in (Matrix and ) standard
form is
0 b 0 X
b, X A
> >
=
,
subject to
X c = z
Dual problem in Matrix form
Maximize
where
| |
n
c c c . .
2 1
= C
(
(
(
(
(
(

=
n
x
x
x
.
.
2
1
X
,
,
(
(
(
(
(
(

=
mn m m
n
n
a a a
a a a
a a a
. .
.
.
. .
. .
2 1
2 22 21
1 12 11
A
1
2
.
.
m
b
b
b
(
(
(
(
=
(
(
(

b
Letting
| |
m
y y y . .
2 1
= Y
the dual LPP in matrix form becomes
Minimize b Y = w
subject to
> YA c,
Y unrestricted in sign
If the primal LPP is a minimization problem
the dual LPP in matrix form becomes
Maximize
b Y = w
subject to
s YA c,
Y unrestricted in sign
Minimize
X c = z
0 b 0 X
b, X A
> >
=
,
subject to
Weak Duality Theorem: For any pair of
feasible primal and dual solutions (X, Y), the
value of the objective function in the
minimization problem is an upper bound for
the value of objective function in the
maximization problem. For the optimal pair
(X*, Y*), the values of the objective functions
of the two problems are equal.
Proof: We first look at the case where the
primal is a maximization problem (so that the
dual is a minimization problem).
Since X is a feasible solution of the primal,
X c = z
(1)
=
> >
AX b,
X 0 b 0 ,
(2)
Since Y is a feasible solution of the dual,
b Y = w
(3)
and
(4)
> YA c,
Y unrestricted in sign
and
Premultiplying the equation (2) by Y we get
w = = YAX Yb
Postmultiplying the inequality (4) by the non-
negative vector X we get
Thus we get z w>
The proof when the primal is a minimization
problem is similar.
z = > X c X YA
Corollary(1): If X* is a feasible solution of the
primal and Y* is a feasible solution of the dual
such that
= = = b * Y * X c * z
then X* is an optimal solution of the primal
and Y* is an optimal solution of the dual.
* w
Remark: The second part of the theorem says
the converse to the corollary is also true.
Corollary(2): If one of the problems has an
unbounded solution then the other has an
infeasible solution.
For if it is not, then both problems have
feasible solutions and the relationship,
w z s
must hold, impossible as by assumption either
or z w = + =
We can only say the following:
If one of the problems has an infeasible
solution, then the other problem either has an
unbounded solution or has an infeasible
solution.
The converse of this corollary is NOT true.
Example (1) Problem 2 Problem Set 7.5B
Page 325
Consider the LPP
3 2 1
10 30 50 x x x z + + =
Maximize
subject to
Show that the given primal problem is infeasible
and its dual is unbounded.
The primal is infeasible as 2
nd
constraint
violates
0
2
> x
0 , ,
6 4
5 2
1 2
3 2 1
3 1
2
2 1
>
= +
=
= +
x x x
x x
x
x x
The dual is the LPP
3 2 1
6 5 y y y w + =
Minimize
subject to
1 3 2
5, 10, any number 15 y y y = = >
is a feasible solution of the dual. Hence the
dual is unbounded.
1 3
1 2
3
1 2 3
2 4 50
2 30
10
, , unrestricted in sign
y y
y y
y
y y y
+ >
+ >
>
Example 2. Consider the Primal LPP
Maximize
2 1
x x z + =
subject to
Its dual is the LPP
Minimize
subject to
2 1
y y w =
Both the problems
have infeasible
solutions as we can
easily show.
0 ,
1
1
2 1
2 1
2 1
>
s +
s
x x
x x
x x
0 ,
1
1
2 1
2 1
2 1
>
> +
>
y y
y y
y y
x
1
x
2
1
2 1
s x x
y
1
y
2
1
2 1
> y y
1
2 1
> + y y
1
2 1
s + x x
Primal
Dual
The following theorem tells us how we can find
the optimal solution of the dual from the optimal
tableau of the primal.
Theorem: Given the optimal primal basis B and
its associated objective coefficient vector c
B
, the
optimal solution of the dual problem is
is often called the Simplex Multiplier.
=
-1
B
Y c B
-1
B
c B
Proof Again we assume the primal is a
maximization problem. Since the table is optimal,
0 for all j
j j
z c >
That is
or
So Y is feasible.
The optimal solution of the primal is
1
* z w

= = =
B
c B b Yb
And hence by the corollary, Y is the optimal
solution of the dual. Q.E.D.
1
0

>
B
c B A c
c A Y >
In words,
Row Vector of Optimal dual solution =
( Row Vector of Original Objective
coefficients of optimal primal basic
variables) ( optimal primal inverse )
We can also say
Primal objective Row coefficient of
variable x
j
in the primal optimal tableau
= (LHS of corresponding dual constraint)
- (RHS of corresponding dual constraint)
Problem 4 Problem Set 7.5B Page 326
Consider the following LPP
4 3 2 1
3 4 4 2 x x x x z + + =
subject to
(a) Write the dual.
(b) Verify that B=(A
2
,A
3
) is optimal.
(c) Find the associated optimal dual solution.
0 , , ,
8 4
4
4 3 2 1
4 2 1
3 2 1
>
= + +
= + +
x x x x
x x x
x x x
Maximize
Solution:
(b)
| | 4 4 =
B
C
(

4 / 1 1
4 / 1 0
1
B
| |
(

= =
0 4
1 1
3 2
A A B
For non-basic variables x
1
, x
4
| |
| | 0 2 2 4 2
1
1
0 4
2
1
1
4 / 1 1
4 / 1 0
4 4
1 1
1
1 1
> = =
(

= =

c A B C c z
B
| |
| | 0 3 3
1
0
0 4
) 3 (
1
0
4 / 1 1
4 / 1 0
4 4
4 4
1
4 4
> = +
(

=

(

= =

c A B C c z
B
Also the primal solution is
Thus we get the optimal solution as
0 , 2 , 2 , 0
4 3 2 1
= = = = x x x x
0
2
2
8
4
4 / 1 1
4 / 1 0
1
>
(

=
(

b B
And Max z = 16
(a) The dual problem is:
Minimize
2 1
8 4 y y w + =
subject to 1 2
1 2
1
2
1 2
8 2
4 4
4
3
, unrestricted in sign
y y
y y
y
y
y y
+ >
+ >
>
>
(c)
Thus the dual optimal solution is:
=
-1
B
C B
| |
4 4
0 1/ 4
1 1/ 4
(
(


| |
4 0 =
We also note that Optimum w =
w* = 16 = z*
y
1
= 4, y
2
= 0
***

You might also like