Location via proxy:   [ UP ]  
[Report a bug]   [Manage cookies]                
Scribd
Upload
46 views

Linear Programming I - Part 3

This document discusses duality in linear programming problems (LPPs). It explains that the primal and dual LPPs are related through transformations of the objective functions and constraints. The optimal solutions to the primal and dual problems are related through complementary slackness conditions. Examples are provided to demonstrate how to derive the dual problem from the primal and vice versa.

Uploaded by

Trisha elejorde
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
46 views

Linear Programming I - Part 3

This document discusses duality in linear programming problems (LPPs). It explains that the primal and dual LPPs are related through transformations of the objective functions and constraints. The optimal solutions to the primal and dual problems are related through complementary slackness conditions. Examples are provided to demonstrate how to derive the dual problem from the primal and vice versa.

Uploaded by

Trisha elejorde
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 15

DUALITY

LINEAR PROGRAMMING PROBLEM


Duality in Linear Programming

The pair of LPPs:

Minimize c1 x1+ c2 x2 +  + cn xn
subject to
a11 x1 + a12 x2 +  + a1n xn ≥ b1
a21 x1 + a22 x2 +  + a2 n xn ≥ b2

am1 x1 + am 2 x2 +  + amn xn ≥ bm
x1 ≥ 0,  , xn ≥ 0

are dual to each other.


The original problem, called the primal problem, is
transformed into a dual problem, by the following
procedures:
a) Change “maximize” to “minimize” and vice-versa.
b) Except for the nonnegativity restrictions, the inequality signs
in the primal constraints are reversed in the dual constraints.
c) The coefficient matrix of the dual constraints is the
transpose of the coefficient matrix of the primal constraints.
d) The row vector of coefficients in the primal objective
function is the column vector of constants in the dual
constraints.
Examples:

a) Max 45y1 + 55y2


subj. to
6y1 + 4y2 ≤ 120 6 4
3y1 + 10y2 ≤ 180 3 10
y1, y2 ≥ 0

The dual is given by:


Min 120 x1 + 180 x2
subj. to
6x1 + 3x2 ≥ 45 6 3
4 10
4x1 + 10x2 ≥ 55
x1, x2 ≥ 0
b) Min 2x1 + 10x2
subj. to
2x1 + x2 ≤ 6
5x1 + 4x2 ≥ 20
x1, x2 ≥ 0
Must be written in general form:
Min 2x1 + 10x2
subj. to
-2x1 - x2 ≥ -6
5x1 + 4x2 ≥ 20
x1, x2 ≥ 0
The dual is
Max -6y1 + 20y2
subj. to
-2y1 + 5y2 ≤ 2
-y1 + 4y2 ≤ 10
y1, y2 ≥ 0
Thm. If x is primal feasible and y is dual feasible then
VOF(x) ≥ VOF (y), i.e., the minimand can never fall below the
maximand.

Thm. If x* is primal feasible and y* is dual feasible, and


VOF (x*) = VOF(y*), then x* and y* are optimal solutions.

Thm. If the primal is unbounded, then the dual is


infeasible.
Thm. (Complementary Slackness) If x* and y* are
primal and dual optimal, respectively, then
(i) if yi* > 0, then ti* = 0 and if xj * > 0 then sj* = 0;
(ii) if ti * > 0, then yi * = 0 and if sj* > 0 then xj* = 0,
where t* and s* are vectors of surplus and slack
variables, respectively.

Minimize c1 x1+ c2 x2 +  + cn xn Maximize b1 y1+ b2 y2 +  + bm ym


subject to subject to
a11 x1 + a12 x2 +  + a1n xn − t1 = b1 a11 y1 + a21 y2 +  + an1 ym + s1 = c1
a21 x1 + a22 x2 +  + a2 n xn − t 2 = b2 a12 y1 + a22 y2 +  + an 2 ym + s 2 = c2
 
am1 x1 + am 2 x2 +  + amn xn − tm = bm a1n y1 + a2 m y2 +  + anm ym + sn ≤ cn
x1 ≥ 0, , xn ≥ 0 y1 ≥ 0,  , ym ≥ 0
Thm. (Complementary Slackness) If x* and y* are
primal and dual optimal, respectively, then
(i) if yi* > 0, then ti* = 0 and if xj * > 0 then sj* = 0;
(ii) if ti * > 0, then yi * = 0 and if sj* > 0 then xj* = 0,
where t* and s* are vectors of surplus and slack
variables, respectively.

There is no implication if:


(i) if yi* = 0 or xj* = 0.
(ii) if ti * = 0 or sj* = 0.
Recall JAMES
James produces two commodities, y1 and y2. These two
products are processed through 2 machines. Maximum
hours available are 120 and 180 for machines 1 and 2,
respectively. Profit is P45 for y1 and P55 for y2. Determine
the optimal y1 and y2 that should be manufactured in order
to maximize profit.
Determining the dual solution from the primal solutions:

Max. 45y1 + 55y2


subj. to y1* > 0 implies t1* = 0.
6y1 + 4y2 ≤ 120 y2* > 0 implies t2* = 0
3y1 + 10y2 ≤ 180
y1, y2 ≥ 0

y1* = 10; y2* = 15; VOF (y) = 1,275

Min 120 x1 + 180 x2


subj. to
6x1 + 3x2 ≥ 45 6x1 + 3x2 = 45
4x1 + 10 x2 ≥ 55 4x1 + 10 x2 = 55
x1, x2 ≥ 0

x1* = 95/16 or 5.9375 ; x2* = 3.125; VOF (x) = 1,275


Determining the dual solution from the primal solutions:

Max. y1 - y2
subj. to y1* > 0 implies t1* = 0.
-4y1 + 3y2 ≤ 3
y1 - y2 ≤ 3
y1, y2 ≥ 0

y1* = 3; y2* = 0; VOF (y) = 3

Min 3x1 + 3x2 3x1 + 3x2 = 3


subj. to
-4x1 + x2 ≥ 1 -4x1 + x2 = 1
3x1 - x2 ≥ -1
x1, x2 ≥ 0
x1* = 0; x2* = 1; VOF (x) = 3
Example 3 : BOURNE
Bourne manufactures tables (y1), chairs (y2) and
benches (y3) and derives a per unit profit of P6, P8 and
P7, respectively. Only two resources are consumed in
manufacturing: wood (300 board feet in inventory) and
labor (110 hours available). Bourne uses 30 board ft to
produce a table; 20 board ft for a chair and 25 board ft
for a bench. He also uses 5 hours of labor to produce
a table, 10 hours for a chair, and 7 hours for a bench.
Max π = 6y1 + 8y2 + 7y3
subject to
30y1 + 20y2 + 25y3 ≤ 300
5y1 + 10y2 + 7y3 ≤ 110
y1, y2 ≥ 0

Min C = 300x1 + 110x2


subject to
30x1 + 5x2 ≥ 6
20x1 + 10x2 ≥ 8
25x1 + 7x2 ≥ 7
x1, x2 ≥ 0
Max π = 6y1 + 8y2 + 7y3 6y1 + 8y2 + 7y3 = 98.1818
subject to
30y1 + 20y2 + 25y3 ≤ 300 30y1 + 20y2 + 25y3 = 300
5y1 + 10y2 + 7y3 ≤ 110 5y1 + 10y2 + 7y3 = 110
y1, y2 ≥ 0

y1* = 0; y2* = 5.9091; y3* = 7.2727

Min C = 300x1 + 110x2


subject to
30x1 + 5x2 ≥ 6 x1* > 0 implies s1* = 0.
20x1 + 10x2 ≥ 8 x2* > 0 implies s2* = 0
25x1 + 7x2 ≥ 7
x1, x2 ≥ 0

x1* = 7/55 or 0.1273; x2* = 6/11 or 0.5455 ; VOF (x) = 98.1818


Exercise

Solve the following linear programming problem by


solving the dual problem:

min G = 6y1 + 20y2 + 3y3 + 20y4


subject to
3y1 + 6y2 - y3 + 2y4 ≥ 4
-4y1 + 2y2 + y3 + 5y4 ≥ 2
yi ≥ 0 i = 1, ...., 4

Answer: x1=2.308; x2=3.077; VOF = 15.385


y1 = 0, y2 = 0.615, y3 = 0, y4 = 0.154

You might also like