Scribd
Upload
14 views

Opt Lecture 7

Uploaded by

pptforwork3011
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
14 views

Opt Lecture 7

Uploaded by

pptforwork3011
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 11

Lecture 7.

Duality of LP problem

Situations lead to dual problems


Definition of dual problem of standard problem
Definition of dual problem of general problem
Relationship between primal and dual problems

1/1
Situations lead to dual problems

Enterprise A has a plan producing n items for the market.


m: number of types of resources.
bi : amount of i th resource, i = 1, . . . , m
n: number of products
aij : units of i th resource to produce 1 unit of j th product (i = 1, . . . , m,
j = 1, . . . , n)
pj : price of 1 unit of j th product (j = 1, . . . , n)
a. Construct a problem finding a production plan so that revenue of
enterprise A reaches its maximum.
b. Enterprise A might not produce anything and sell all inputs for enterprise
B. When will enterprise A decide to sell all inputs instead of producing?
2/1
Definition of dual problem of a standard problem
Consider standard linear programming problem
n
X
f (xx ) = cj xj → max
j =1
 n
X
 aij xj = bi i = 1, m


j =1 (1)

x ≥ 0

j = 1, n
j

The dual problem of (6) has the form:


m
X
fe(yy ) = bi yi → min
i =1
m
½X
aij yi ≥ cj j = 1, n (2)
i =1
Comment on the relationship between two problems?
3/1
Definition of dual problem of a standard problem

Example 1.
Write the dual problem of the following problem:

f (xx ) = x1 + 2x2 + x3 → max





 2x1 +x2 −x3 = −4
x1 −2x2 +x3 = 5


x1 x2 x3 ≥ 0

4/1
Definition of dual problem of a standard problem

Example 2.
Write the dual problem of the following problem:

f (xx ) = −2x1 − x2 + x3 → max





 x1 −x2 +2x3 ≤ 1


 2x1 +x2 −x3 ≥ 5



x2 +4x3 = 3





 x1 ≥ 0

x2 0

 ≤

5/1
Writing dual problem for a general LP

Primal Problem Dual Problem


n
X m
X
f (xx ) = cj xj −→ max fe(yy ) = bi yi −→ min
j =1 i =1
n
X
aij xj = bi yi has no sign constraint
j =1
X n
aij xj ≥ bi yi ≤ 0
j =1
X n
aij xj ≤ bi yi ≥ 0
j =1
m
X
xj ≥ 0 aij yi ≥ cj
i =1
X m
xj ≤ 0 aij yi ≤ cj
i =1
X m
xj has no sign constraint aij yi = cj
i =1
6/1
Relationship between primal and dual problems

1 If both problems (1) and (2) have feasible solutions then for all
feasible solutions x of (1) and y of (2) we have f (xx ) ≤ fe(yy ).
2 Let (xb, yb) be feasible solutions of primal and dual problem (1) and
(2). If f (xb) = fe(yb) then xb is the optimal solution of (1), yb is the
optimal solution of (2).
3 If either primal or dual problem is solvable, then the other is solvable
also, and the optimal value of two problems are equal.
4 Necessary and sufficient conditions for xb to be the optimal solution of
(1) and yb to be the optimal solution of dual problem (2) are :
m
X
a. If xbj > 0 then aij ybi = cj
i =1
m
X
b. If aij ybi > cj then xbj = 0
i =1
7/1
Relationship between primal and dual problems

Remarks
1 Necessary and sufficient condition for primal-dual problems to be
solvable is each problem having at least one feasible solution.
2 Necessary and sufficient condition for primal-dual problems to have
one problem having feasible solution and the other not is that the
objective function of the one having feasible solution must be
unbounded on the feasible region.
3 If a constraint of a problem is non-binding at its optimal solution then
the corresponding dual constraint must be binding at the optimal
solution of the dual problem.

8/1
Applications of properties of dual problem

Example 3
Given a problem: f (xx ) = 2x1 + c2 x2 + 2x3 − c4 x4 → max

 2x1 −x2 +3x3 −x4 = 6


2x1 −3x2 −x3 ≤ 6


x1 , x2 , x3 , x4 ≥ 0

a. Solve this problem using simplex algorithm given that c2 = −1, c4 = 0.


b. Write the dual problem of the above problem.
c. From the simplex table in a., comment on the solvability of the dual
problem when c4 changes.
d. Prove that if c4 = 0 then for all c2 , both the primal and dual problem
are insolvable.

9/1
Applications of properties of dual problem

Example 4
Given a problem:

f (xx ) = 5x1 − 4x2 + 2x3 + 3x4 → min



 3x1 −2x2 +x3 +2x4 ≥ 20


x1 +x2 −3x3 −x4 ≤ −15



2x1 −2x2 +x3 +x4 ≥ 30

a. Does this problem has an optimal solution? an optimal solution which is


an extreme point?
b. Comment on the solvability of this problem when b changes.
c. In the above problem, if we replace f (xx ) → min by f (xx ) → max then
does the solvability of the problem change?
10 / 1
Applications of properties of dual problem

Example 5
Given a problem

f (xx ) = −8x1 + 6x2 + 4x3 + 5x4 → min





 x1 −2x3 +x4 ≤ 7

 −2x

+x2 −x3 +3x4 = −4
1


 3x1 −x2 +2x3 −6x4 ≥ 5


 x ,
1 x2 , x4 ≥ 0

and a vector x 0 = (3; 0; −2; 0).


a. What are the properties of x 0 in this problem (a feasible solution, an
optimal solution, an optimal solution which is an extreme point)?
b. Find the optimal set of the dual problem and find the optimal solution
of the dual problem having y1 = −3. 11 / 1

You might also like