The Dual Problem: Minimization with problem constraints of the form

Linear programming problems exist in pairs. That is in linear programming problem, every maximization problem is associated with a minimization problem. Conversely, associated with every minimization problem is a maximization problem. Once we have a problem with its objective function as maximization, we can write by using duality relationship of linear programming problems, its minimization version. The original linear programming problem is known as primal problem, and the derived problem is known as dual problem.

Thus, the dual problem uses exactly the same parameters as the primal problem, but in different locations. To highlight the comparison, now look at these same two problems in matrix notation.
Primal Problem Minimize Z=cx Dual Problem Maximize W=yb

Subject to
and

Axb
x0

Subject to
And

yAc
y0

Primal problem 11 11 11 12 12 12 13 13 13

Dual problem 11 12 13 11 12 13 11 12 13

A=

A=

As an example, Primal Problem in algebraic form Minimize C=3x1+5x2 Subject to x1 4 2x2 12 3x1+2x2 18 x10, x20 A= Primal problem 1 0 3 3 0 2 2 5 4 12 18 1 A=
T

Dual problem 1 0 4 0 2 12 3 2 18 3 5 1

Dual Problem in algebraic form Maximize Z=4y1+12y2+18y3 Subject to y1+3y3 3 2y2+2y3 5

and

y10 , y20 ,y30

and

Consequently, (1) the parameters for a constraint in either problem are the coefficients of a variable in the other problem and (2) the coefficients for the objective function of either problem are the right sides for the other problem.

Summary
Primal (a) Maximize. (b) Objective Function. (c) Right hand side. (d) i th row of input-output coefficients. (e) j th column of input-output coefficients. Dual Minimize Right hand side. Objective function. i th column of input output coefficients. j the row of input-output coefficients.

EXAMPLE 1
The procedure for forming the dual problem is summarized in the box below: Formation of the Dual Problem Given a minimization problem with problem constraints, Step 1. Use the coefficients and constants in the problem constraints and the objective function to form a matrix A with the coefficients of the objective function in the last row. Step 2. Interchange the rows and columns of matrix A to form the matrix AT, the transpose of A.

Step 3. Use the rows of AT to form a maximization problem with problem constraints.
Forming the Dual Problem Minimize C = subject to 40x1 + 12x2 + 40x3 2x1 + x2 + 5x3 20 4x1 + x2 + x3 30 x1, x2, X3 0

EXAMPLE 2

Form the dual problem: Minimize C = 16 x1 + 9x2 + 21x3 subject to x1 + x2 + 3x3 12 2x1 + x2 +x3 16 x1, x2, x3 0

EXAMPLE 3 Solution of Minimization Problems

ORIGINAL PROBLEM (1) Minimize C = 16x1 + 45x2 DUAL PROBLEM (2) Maximize P = 50y1 + 27y2 subject to 2y1 + y2 16 5y1 + 3y2 45 y1,y2 0

subject to 2x1 + 5x2 50 x1 + 3x2 27 x1, x2 0

Corner Point (x1, x2) C= 16x1+45 x2 (0,10) 450 (15,4) 420 (27,0) 432 Min C=420 at (15,4)

Corner Point (y1, y2) P=50y1+27 y2 (0,0) 0 (0,15) 405 (3,10) 420 (8,0) 400 Max P=420 at (3,10)

For reasons that will become clear later, we will use the variables x1 and x2 from the original problem as the slack variables in the dual problem: 2y1 + 5y1 + -50y1y2 3y2 27y2 + x1 + x2 +P =16 =45 =0 (initial system for the dual problem)

y1 x1 x2
2 5 -50

y2
1 3 -27 0.5 3 -27 0.5 0.5 -2 0 1 0

x1
1 0 0 0.5 0 0 0.5 -2.5 25 3 -5 15

x2
0 1 0 0 1 0 0 1 0 -1 2 4

P
0 0 1 0 0 1 0 0 1 0 0 1 16 45 0 8 45 0 8 5 400 3 10 420

x1 x2

1 5 -50

y1 x2
P

1 0 0 1 0 0

y1 y2
P

Since all indicators in the bottom row are nonnegative, the solution to the dual problem is y1 = 3, y2 = 10, x1 = 0, x2 = 0, P = 420 which agrees with our earlier geometric solution. Furthermore, examining the bottom row of the final simplex tableau, we see the same optimal solution to the minimization problem that we obtained directly by the geometric method:

Min C = 420
This is not achieved with mistake.

at

x1 = 15, x2 = 4

An optimal solution to a minimization problem always can be obtained from the bottom row of the final simplex tableau for the dual problem. Now we can see that using x1 and x2 as slack variables in the dual problem makes it easy to identify the solution of the original problem.

EXAMPLE 4

Solve the following minimization problem by maximizing the dual: Minimize C = 40x1 + 12x2 + 40x3 subject to 2x1 +x2 + 5x3 20 4x1 + x2 + x3 30 Nonnegativity x1, x2, x3 0

Maximization and minimization with mixed problem constraints

In this section we present a generalized version of the simplex method that will solve both maximization and minimization problems with any combination of , , and = problem constraints.
When the constraint is in the form of we introduce slack variable (unused capacity), and when the constraint is in the form of we subtract the surplus (excess amount).

 In order to use the simplex method on problems with mixed constraints, we turn to an ingenious device called an artificial variable. This variable has no physical meaning in the original problem (which explains the use of the word "artificial") and is introduced solely for the purpose of obtaining a basic feasible solution so that we can apply the simplex method. An artificial variable is a variable introduced into each equation that has a surplus variable. As before, to ensure that we consider only basic feasible solutions, an artificial variable is required to satisfy the nonnegative constraint.

To prevent an artificial variable from becoming part of an optimal solution to the original problem, a very large "penalty" is introduced into the objective function. This penalty is created by choosing a positive constant M so large that the artificial variable is forced to be 0 in any final optimal solution of the original problem.

Big M Method: Introducing Slack, Surplus, and Artificial Variables to Form the Modified Problem
Step 1. If any problem constraints have negative constants on the right side, multiply both sides by -1 to obtain a constraint with a nonnegative constant. (If the constraint is an inequality, this will reverse the direction of the inequality.) Step 2. Introduce a slack variable (S) in each constraint. Step 3. Introduce a surplus variable (E) and an artificial variable (A) in each constraint. Step 4. Introduce an artificial variable (A) in each = constraint. Step 5. For each artificial variable Ai subtract- MAi from the objective function.

EXAMPLE 5
Find the modified problem for the following linear programming problem. (Do not attempt to solve the problem.) Maximize P = 2x1 + 5x2 + 3x3 subject to x1 + 2x2 - x3 7 -x1 + x2 - 2x3 -5 x1 + 4x2 + 3x3 1 2x1- x2 + 4x3 = 6 x1, x2, x3 0

SOLUTION
First, we multiply the second constraint by -1 to change -5 to 5: (-1)(-x1 + x2 -2x3 ) (-1)(-5) x1 - x2 + 2x3 5

Next, we introduce the slack, surplus, and artificial variables according to the procedure stated in the box:

x1 + 2x2 - x3 +S1 x1 - x2 + 2x3 -E1+A1 x1 + 4x2 + 3x3 2x1- x2 + 4x3

-E2+A2
+A3

=7 =5 =1 =6

Finally, we subtract MA1, MA2, and MA3 from the objective function to penalize the artificial variables: P = 2x1 + 5x2 + 3x3 - MA1 - MA2 - MA3 The modified problem is

Maximize P = 2x1 + 5x2 + 3x3 - MA1 - MA2 - MA3

subject to x1 + 2x2 - x3 +S1 x1 - x2 + 2x3 -E1+A1 x1 + 4x2 + 3x3 E2+A2 2x1- x2 + 4x3 +A3 x1, x2, x3, S1, E1, E2, A1, A2, A3 0 =7 =5 =1 =6

After introducing the slack, surplus and artificial variables we continue to solve the problem with following steps; Step 1. Form the preliminary simplex tableau for the modified problem.

Step 2. Use row operations to eliminate the Ms in the bottom row of the preliminary simplex tableau in the columns corresponding to the artificial variables. The resulting tableau is the initial simplex tableau.
Step 3. Solve the modified problem by applying the simplex method to the initial simplex tableau found in step 2.

EXAMPLE 6
Solve the following linear programming problem using the big M method:
Maximize P = x1 - x2 + 3x3 subject to x1 + x2 20 x1 + x3 = 5 x2 + x3 10 x1 , x2, x3 0

Minimization by the big M method

In addition to solving any maximization problem, the big M method can be used to solve minimization problems. To minimize an objective function, we have only to maximize its negative. Furthermore, if M is the minimum value of f, then M is the maximum value of -f, and conversely. Thus, we can find the minimum value of a function f by finding the maximum value of -f and then changing the sign of the maximum value.

EXAMPLE 7
Production Scheduling: Minimization Problem A small jewelry manufacturing company employs a person who is a highly skilled gem cutter, and it wishes to use this person at least 6 hours per day for this purpose. On the other hand, the polishing facilities can be used in any amounts up to 10 hours per day. The company specializes in three kinds of semiprecious gemstones, J, K, and L. Relevant cutting, polishing, and cost requirements are listed in the table. How many gemstones of each type should be processed each day to minimize the cost of the finished stones? What is the minimum cost?

J Cutting Polishing Cost per stone 1hr 2hr $30

K 1hr 1 hr $30

L 1hr 2hr $10