Introductory Mathematical Analysis

For Business, Economics, and The Life and Social Sciences

Fourteenth Canadian Edition

Chapter 7
Linear Programming

Copyright © 2019 Pearson Canada Inc. All rights reserved. 7-1

Chapter Objectives (1 of 2)
• To represent geometrically the solution of a linear
inequality in two variables and to extend this
representation to a system of linear inequalities.
• To state the nature of linear programming problems, to
introduce terminology associated with them, and to solve
them geometrically.
• To show how the simplex method is used to solve a
standard maximum linear programming problem. This
method enables the solution of problems that cannot be
solved geometrically.

Copyright © 2019 Pearson Canada Inc. All rights reserved. 7-2

Chapter Objectives (2 of 2)
• To use artificial variables to handle maximization problems
that are not of standard maximum form.
• To show how to solve a minimization problem by altering
the objective function so that a maximization problem
results.
• To first motivate and then formally define the dual of a
linear programming problem.

Copyright © 2019 Pearson Canada Inc. All rights reserved. 7-3

Chapter Outline
7.1) Linear Inequalities in Two Variables
7.2) Linear Programming
7.3) The Simplex Method
7.4) Artificial Variables
7.5) Minimization
7.6) The Dual

Copyright © 2019 Pearson Canada Inc. All rights reserved. 7-4

7.1 Linear Inequalities in 2 Variables (1 of 3)
A linear equality in the variables x and y is an inequality
that can be written in one of the forms
ax + by + c < 0, ax + by + c ≤ 0, ax + by + c > 0, ax + by + c ≥ 0
where a, b, and c are constants and not both
a and b are zero.

Geometrically, the solution (or graph) of a linear inequality

in x and y consists of all points ( x, y ) in the plane whose
coordinates satisfy the inequality. These regions are called
open half - planes, or closed half - planes.

Copyright © 2019 Pearson Canada Inc. All rights reserved. 7-5

7.1 Linear Inequalities in 2 Variables (2 of 3)
• Below are graphs of an open half-plane and a closed half-
plane. Note the dotted, versus filled, line:

Above and to the left is the graph of y < 5 − 2 x, forming the

open half-plane, while to the right is the graph of y ≤ 5 − 2 x,
forming the closed half-plane.
Copyright © 2019 Pearson Canada Inc. All rights reserved. 7-6
7.1 Linear Inequalities in 2 Variables (3 of 3)
Example 1 – Solving a Linear Inequality
Find the region defined by the inequality y ≤ 5.
Solution: Since x does not appear, the inequality
is assumed to be true for all values of x. The region
consists of the line y = 5, together with the half- plane
below it.
Example 3 – Solving a System of Linear
Inequalities
{
Solve the system y ≥ −2 x + 10
y ≥ x−2
Solution: The solution consists of all points that
are simultaneously on or above the line y = −2 x + 10
and on or above the line y = x − 2.
Copyright © 2019 Pearson Canada Inc. All rights reserved. 7-7
7.2 Linear Programming (1 of 3)
A linear function in x and y has the form P = P ( x, y ) = ax + by
where a and b are constants.
In a linear programming problem, the function to be maximized
or minimized is called the objective function. Its domain is
defined to be the set of all solutions to the system of linear
constraints that appear in the problem. The domain is also
called the set of feasible points, or the feasible region.
The feasible region contains corner points where the
constraint lines intersect.
A linear function defined on a nonempty, bounded feasible region
has both a maximum value and a minimum value.
Moreover, each of these values can be found at a corner point.

Copyright © 2019 Pearson Canada Inc. All rights reserved. 7-8

7.2 Linear Programming (2 of 3)
Example 1 – Solving a Linear Programming Problem
Maximize the objective function
P = 3 x + y subject to the constraints
2x + y ≤ 8
2 x + 3 y ≤ 12
x≥0
y≥0

Solution: The region is nonempty

and bounded, as seen in the figure above.
Thus, P attains a maximum at one of the four
corner points. The coordinates of A, B, and D are obvious.
To find the coordinates of C we solve the equations 2 x + y = 8 and
2 x + 3 y = 12, giving us x = 3, and y = 2.
Copyright © 2019 Pearson Canada Inc. All rights reserved. 7-9
7.2 Linear Programming (3 of 3)
Example 1 – Continued
Solution, continued

Thus A = (0, 0), B = (4, 0), C = (3, 2), and D = (0, 4).
Evaluating P at these points, we obtain
P ( A) = 3( 0 ) + 0 = 0
P ( B ) = 3 ( 4 ) + 0 = 12
P ( C ) = 3 ( 3) + 2 = 11
P ( D ) = 3( 0) + 4 = 4
Hence, the maximum value of P is 12, occurring when
x = 4 and y = 3.
Copyright © 2019 Pearson Canada Inc. All rights reserved. 7 - 10
7.3 The Simplex Method (1 of 5)
Standard Maximum Linear Programming Problem

Copyright © 2019 Pearson Canada Inc. All rights reserved. 7 - 11

7.3 The Simplex Method (1 of 5)
Standard Maximum Linear Programming Problem

Copyright © 2019 Pearson Canada Inc. All rights reserved. 7 - 12

7.3 The Simplex Method (1 of 5)
Standard Maximum Linear Programming Problem

Note that one feasible solution to a std.linear programming

problem is always, x1 = 0, x2 = 0, …, xn = 0 and at this feasible
solution the value of the objective function Z = 0

Copyright © 2019 Pearson Canada Inc. All rights reserved. 7 - 13

7.3 The Simplex Method (1 of 5)
Standard Maximum Linear Programming Problem

Copyright © 2019 Pearson Canada Inc. All rights reserved. 7 - 14

7.3 The Simplex Method (2 of 5)
Example 1 – The Simplex Method

 x1 + x2 ≤ 20
Maximize Z = 5 x1 + 4 x2 subject to  2 x1 + x2 ≤ 35
−3x1 + x2 ≤ 12
Solution: The initial simplex table is below.

Copyright © 2019 Pearson Canada Inc. All rights reserved. 7 - 15

7.3 The Simplex Method (3 of 5)
Example 1 – Continued
Solution continued: The most negative indicator, − 5, occurs in
the x1 column. Thus x1 is the entering variable. The smaller
quotient is 352 , so s2 is the departing variable. The pivot entry is
2. Using elementary row operations to get a 1 in the pivot
position and 0's elsewhere in its column, we have

Copyright © 2019 Pearson Canada Inc. All rights reserved. 7 - 16

7.3 The Simplex Method (4 of 5)
Example 1 – Continued
Solution continued: Note that in column B, which keeps track of
which variables are basic, x1 has replaced s2 . Since we still have
a negative indicator, − 32 , we must continue our process. s1 is the
departing variable and 12 is the pivot entry. Using elementary row
operations, we have

Since all indicators are nonnegative, the maximum value of

Z is 95 and occurs when x1 = 15, x2 = 5, and x3 = 0.
Copyright © 2019 Pearson Canada Inc. All rights reserved. 7 - 17
7.3 The Simplex Method (5 of 5)
Some Terminology
• If no quotient exists in a simplex table, then the standard
maximum linear programming problem has an unbounded
solution.
• An unbounded solution occurs when the objective function
does not attain a maximum value.
• In a table that gives an optimum solution, a zero indicator
for a nonbasic variable suggests the possibility of multiple
optimum solutions.

Copyright © 2019 Pearson Canada Inc. All rights reserved. 7 - 18

7.4 Artificial Variables (1 of 8)
• To start using the simplex method, a basic feasible
solution, BFS, is required.
• When an initial BFS is not obvious, there is an ingenious
method to arrive at one artificially.
• A new equation is formed by adding a nonnegative
variable t, called an artificial variable.
• We can eventually force t to be 0 if we alter the original
objective function. The altered objective function is called
the artificial objective function.

Copyright © 2019 Pearson Canada Inc. All rights reserved. 7 - 19

7.4 Artificial Variables (2 of 8)
Example 1 – Artificial Variables

Copyright © 2019 Pearson Canada Inc. All rights reserved. 7 - 20

7.4 Artificial Variables (3 of 8)
Example 1 – Continued

To get simplex table I, we replace the M in the objective row and

the artificial variable column by 0 by adding (− M ) times row 3 to
row 4 :

Copyright © 2019 Pearson Canada Inc. All rights reserved. 7 - 21

7.4 Artificial Variables (4 of 8)
Example 1 – Continued
Solution continued
Since M is a large positive number, (−1 − M ) is the most negative
indicator and, thus, the entering variable is x2 . The smallest
quotient is evidently 2 and, thus, the departing variable is t.
The pivot entry is shaded in the table above. Proceeding, we
get simplex table II:

Copyright © 2019 Pearson Canada Inc. All rights reserved. 7 - 22

7.4 Artificial Variables (5 of 8)
Example 1 – Continued
Solution continued
The BFS corresponding to table II has t = 0. Thus, we delete
the t -column and change W 's to Z 's in succeeding tables.
Continuing, we obtain table III:

All indicators are nonnegative. Hence, the maximum value of

Z is 17. It occurs when x1 = 5 and x2 = 7.

Copyright © 2019 Pearson Canada Inc. All rights reserved. 7 - 23

7.4 Artificial Variables (6 of 8)
Example 3 – An Empty Feasible Region
Use the simplex method to maximize Z = 2 x1 + x2 subject to
− x1 + x2 ≥ 2
 x1 + x2 ≤ 1
 x1 , x2 ≥ 0
Solution: An artificial variable will occur. The simplex tables are below:

Copyright © 2019 Pearson Canada Inc. All rights reserved. 7 - 24

7.4 Artificial Variables (7 of 8)
Example 3 – Continued

Copyright © 2019 Pearson Canada Inc. All rights reserved. 7 - 25

7.4 Artificial Variables (8 of 8)
Example 3 – Continued

Solution, continued

Since M is a large positive number, the

indicators in simplex table II are nonnegative,
so the simplex procedure terminates.
The value of the artificial variable, t1 , is 1.
Therefore, as previously stated, the feasible
region of the original problem is empty and,
hence, no solution exists. This result can be
obtained geometrically, as seen here.

Copyright © 2019 Pearson Canada Inc. All rights reserved. 7 - 26

7.5 Minimization (1 of 2)
• To minimize a function, it suffices to maximize the negative
of the function.
Example 1 – Minimization

Copyright © 2019 Pearson Canada Inc. All rights reserved. 7 - 27

7.5 Minimization (2 of 2)
Example 1 – Continued

Copyright © 2019 Pearson Canada Inc. All rights reserved. 7 - 28

7.6 The Dual (1 of 3)
• In general, with any given linear programming problem, we
can associate another linear programming problem called
its dual.
• The given problem is called primal.
• If the primal is a maximization problem, then its dual is a
minimization problem. Similarly, if the primal involves
minimalization, then the dual involves maximization.
• If the primal has an optimal solution, then so does the dual,
and the optimum value of the primal’s objective function is
the same as that of its dual.

Copyright © 2019 Pearson Canada Inc. All rights reserved. 7 - 29

7.6 The Dual (2 of 3)
Example 1 – Finding the Dual of a Maximization Problem

Copyright © 2019 Pearson Canada Inc. All rights reserved. 7 - 30

7.6 The Dual (3 of 3)
Example 3 – Applying the Simplex Method to the Dual
Use the dual and the simplex method to maximize Z = 4 x1 − x2 − x3
3x1 + x2 − x3 ≤ 4
subject to  x1 + x2 + x3 ≤ 2
 x1 , x2 , x3 ≥ 0
Solution: The dual is: Minimize W = 4 y1 + 2 y2 subject to
 3 y1 + y2 ≥ 4
 y1 + y2 ≥ 1
− y + y ≥ −1
 y ,1 y ≥2 0
 1 2
After computations, we find that the maximum value of − W is
− 112 , so the minimum value of W is 112 . Hence, the maximum value
of Z is also 112 , and it occurs when x1 = 32 , x2 = 0, and x3 = 12 .
Copyright © 2019 Pearson Canada Inc. All rights reserved. 7 - 31