Linear Programming:

The Simplex Method

Learning Objectives
1. Convert LP constraints to equalities with slack, surplus,
and artificial variables
2. Set up and solve LP problems with simplex tableaus
3. Interpret the meaning of every number in a simplex
tableau
4. Recognize special cases such as infeasibility,
unboundedness, and degeneracy
5. Use the simplex tables to conduct sensitivity analysis
6. Construct the dual problem from the primal problem
After completing this chapter, students will be able to:
A model consisting of linear relationships representing
a firms objective and resource constraints
Introduction to Linear
Programming (LP)
LP is a mathematical modeling technique used to
determine a level of operational activity in order to
achieve an objective, subject to restrictions called
constraints

LP Model Formulation
Decision variables
mathematical symbols representing levels of activity of an operation
Objective function
a linear relationship reflecting the objective of an operation
most frequent objective of business firms is to maximize profit
most frequent objective of individual operational units (such as a
production or packaging department) is to minimize cost
Constraint
a linear relationship representing a restriction on decision making
General LP Model Formulation
(cont.)
Max/min z = c
1
x
1
+ c
2
x
2
+ ... + c
n
x
n

subject to:
a
11
x
1
+ a
12
x
2
+ ... + a
1n
x
n
(, =, ) b
1
a
21
x
1
+ a
22
x
2
+ ... + a
2n
x
n
(, =, ) b
2
:
a
m1
x1 + a
m2
x
2
+ ... + a
mn
x
n
(, =, ) b
m

x
j
= decision variables
b
i
= constraint levels
c
j
= objective function coefficients (contribution coefficient for max. and
cost coefficient for min.)
a
ij
= constraint coefficients
With only two decision variables it is possible to use
graphical methods to solve LP problems
But most real life LP problems are too complex for simple
graphical procedures
We need a more powerful procedure called the simplex
method
The simplex method examines the corner points in a
systematic fashion using basic algebraic concepts
It does this in an iterative manner until an optimal solution
is found
Each iteration moves us closer to the optimal solution
Introduction
Why should we study the simplex method?
It is important to understand the ideas used to produce
solutions
It provides the optimal solution to the decision variables
and the maximum profit (or minimum cost)
It also provides important economic information
To be able to use computers successfully and to interpret
LP computer printouts, we need to know what the simplex
method is doing and why
The Flair Furniture Company makes two types of
furniture Table and Chair . The contribution for each
product as calculated by accounting department are $70
table and $ 50 chair. Both product are processed on two
departments. The time required by each product on each
department are as follows.
Department Table Chair Time avail
Painting 2hrs/table 1hr/chair 100 hours

Carpentry 4hrs/table 3hrs/chair 240 hours
i. Formulate this problem as LPP.
ii. How should the company utilize available
time in order to maximize contribution.
How To Set Up The Initial Simplex
Solution
Formulation of linear programming problem
Well use the simplex method to solve the problem
T = number of tables produced
C = number of chairs produced
Maximize profit = $70T + $50C (objective function)
subject to 2T + 1C 100 (painting hours constraint)
4T + 3C 240 (carpentry hours constraint)
T, C 0 (nonnegativity constraint)
and
Converting the Constraints to
Equations
The inequality constraints must be converted into
equations
Less-than-or-equal-to constraints () are converted to
equations by adding a slack variable to each
Slack variables represent unused resources
For the Flair Furniture problem, the slacks are
S
1
= slack variable representing unused hours
in the painting department
S
2
= slack variable representing unused hours
in the carpentry department
The constraints may now be written as
2T + 1C + S
1
= 100
4T + 3C + S
2
= 240
Converting the Constraints to
Equations
If the optimal solution uses less than the available
amount of a resource, the unused resource is slack
For example, if Flair produces T = 40 tables and C = 10
chairs, the painting constraint will be

2T + 1C + S
1
= 100
2(40) + 1(10) + S
1
= 100
S
1
= 10

There will be 10 hours of slack, or unused painting
capacity
Converting the Constraints to
Equations
Each slack variable must appear in every constraint
equation
Slack variables not actually needed for an equation have
a coefficient of 0
So
2T + 1C + 1S
1
+ 0S
2
= 100
4T + 3C + 0S
1
+ 1S
2
= 240
T, C, S
1
, S
2
0
The objective function becomes

Maximize profit = $70T + $50C + $0S
1
+ $0S
2
Finding an Initial Solution Algebraically
There are now two equations and four variables
When there are more unknowns than equations,
you have to set some of the variables equal to 0
and solve for the others
In this example, two variables must be set to 0 so
we can solve for the other two
A solution found in this manner is called a basic
feasible solution
Finding an Initial Solution Algebraically
The simplex method starts with an initial feasible solution
where all real variables are set to 0
While this is not an exciting solution, it is a corner point
solution
Starting from this point, the simplex method will move to
the corner point that yields the most improved profit
It repeats the process until it can further improve the
solution
On the following graph, the simplex method starts at point
A and then moves to B and finally to C, the optimal solution
Finding an Initial Solution Algebraically
Corner points for the
Flair Furniture Company
problem
100

80

60

40

20


C
| | | | |
0 20 40 60 80
T
N
u
m
b
e
r

o
f

C
h
a
i
r
s

Number of Tables Figure 9.1
B = (0, 80)
C = (30, 40)
2T + 1C 100
4T + 3C 240
D = (50, 0)
(0, 0) A
The First Simplex Tableau
Constraint equations
It simplifies handling the LP equations if we put them
in tabular form
These are the constraint equations for the Flair
Furniture problem
SOLUTION
MIX
QUANTITY
(RIGHT-
HAND SIDE) T C S
1
S
2

S
1
100 2 1 1 0
S
2
240 4 3 0 1
The First Simplex Tableau
The first tableau is is called a simplex tableau
C
j


SOLUTION
MIX
QUANTITY
$70
T
$50
C
$0
S
1

$0
S
2

$0 S
1
100 2 1 1 0
$0 S
2
240 4 3 0 1
Z
j
$0 $0 $0 $0 $0
Z
j
C
j
$0 -$70 -$50 $0 $0
Profit per
unit row
Constraint
equation rows
Gross profit
row
Net profit row
The First Simplex Tableau
The numbers in the first row represent the coefficients in the
first constraint and the numbers in the second the second
constraint
At the initial solution, T = 0 and C = 0, so S
1
= 100 and S
2
= 240
The two slack variables are the initial solution mix(Basic
variables)
The values are found in the QUANTITY column
The initial solution is a basic feasible solution(Zero Solution)
T
C
S
1

S
2
0
0
100
240

=
The First Simplex Tableau
Variables in the solution mix, called the basis in LP
terminology, are referred to as basic variables
Variables not in the solution mix or basis (value of 0) are
called nonbasic variables
The optimal solution was T = 30, C = 40, S
1
= 0, and S
2
= 0
The final basic variables would be
T
C
S
1

S
2
30
40
0
0

=
The First Simplex Tableau
Substitution rates
The numbers in the body of the tableau are the
coefficients of the constraint equations
These can also be thought of as substitution rates
Using the variable T as an example, if Flair were to
produce 1 table (T = 1), 2 units of S
1
and 4 units of S
2

would have to be removed from the solution
Similarly, the substitution rates for C are 1 unit of S
1

and 3 units of S
2

Also, for a variable to appear in the solution mix, it
must have a 1 someplace in its column and 0s in every
other place in that column

The First Simplex Tableau
Adding the objective function
We add a row to the tableau to reflect the objective
function values for each variable
These contribution rates are called C
j
and appear just
above each respective variable
In the leftmost column, C
j
indicates the unit profit for
each variable currently in the solution mix
C
j
$70 $50 $0 $0
SOLUTI
ON
MIX
QUANTI
TY

T

C

S
1


S
2

$0 S
1
100 2 1 1 0
$0 S
2
240 4 3 0 1
The First Simplex Tableau
The Z
j
and C
j
Z
j
rows
We can complete the initial tableau by adding two final
rows
These rows provide important economic information
including total profit and whether the current solution
is optimal
We compute the Z
j
value by multiplying the
contribution value of each number in a column by each
number in that row and the jth column, and summing
The First Simplex Tableau
The Z
j
value for the quantity column provides the total
contribution of the given solution
Z
j
(gross profit) = (Profit per unit of S
1
) (Number of units of S
1
)
+ (profit per unit of S
2
) (Number of units of S
2
)
= $0 100 units + $0 240 units
= $0 profit
The Z
j
values in the other columns represent the gross profit given up by
adding one unit of this variable into the current solution
Z
j
= (Profit per unit of S
1
) (Substitution rate in row 1)
+ (profit per unit of S
2
) (Substitution rate in row 2)
The First Simplex Tableau
Thus,
Z
j
(for column T) = ($0)(2) + ($0)(4) = $0
Z
j
(for column C) = ($0)(1) + ($0)(3) = $0
Z
j
(for column S
1
) = ($0)(1) + ($0)(0) = $0
Z
j
(for column S
2
) = ($0)(0) + ($0)(1) = $0
We can see that no profit is lost by adding one unit of
either T (tables), C (chairs), S
1
, or S
2
The First Simplex Tableau
The Z
j
C
j
number in each column represents the net
profit that will result from introducing 1 unit of each
product or variable into the solution
It is computed by subtracting the Z
j
total for each column
from the C
j
value at the very top of that variables column
COLUMN
T C S
1
S
2
C
j
for column $70 $50 $0 $0
Z
j
for column 0 0 0 0
Z
j

C
j
for column $70 $50 $0 $0
The First Simplex Tableau
Obviously with a profit of $0, the initial solution is not
optimal
By examining the numbers in the Z
j
C
j
row in Table, we
can see that the total profits can be increased by $70 for
each unit of T and $50 for each unit of C
A negative number in the number in the Z
j
C
j
row would
tell us that the profits would increase if the corresponding
variable were added to the solution mix
An optimal solution is reached when there are no negative
numbers in the Z
j
C
j
row
Simplex Solution Procedures
After an initial tableau has been completed, we
proceed through a series of five steps to compute
all the numbers needed in the next tableau
The calculations are not difficult, but they are
complex enough that even the smallest arithmetic
error can produce a wrong answer
Five Steps of the Simplex Method for
Maximization Problems
1. Determine the variable to enter the solution mix next. One
way of doing this is by identifying the column, and hence
the variable, with the largest negative number in the C
j
- Z
j

row of the preceding tableau. The column identified in this
step is called the pivot column(key column).
2. Determine which variable to replace. This is accomplished
by dividing the quantity column by the corresponding
number in the column selected in step 1. The row with the
smallest nonnegative number calculated in this fashion will
be replaced in the next tableau. This row is often referred
to as the pivot row(key row or replacing row). The number
at the intersection of the pivot row and pivot column is the
pivot number(key number).
Five Steps of the Simplex Method for
Maximization Problems
3. Compute new values for the pivot row. To do this, we simply
divide every number in the row by the pivot column.
4. Compute the new values for each remaining row. All remaining
rows are calculated as follows:

(New row numbers) = (Numbers in old row)
Number
above or
below
pivot number
Corresponding number in
the new row, that is, the
row replaced in step 3
x
Five Steps of the Simplex Method for
Maximization Problems
5. Compute the Z
j
and Z
j
C
j
rows, as demonstrated in the
initial tableau. If all the numbers in the Z
j
C
j
row are 0 or
negative, an optimal solution has been reached. If this is
not the case, return to step 1.
The Second Simplex Tableau
We can now apply these steps to the Flair Furniture
problem
Step 1. Select the variable with the largest positive Z
j
C
j

value to enter the solution next. In this case, variable T
with a contribution value of $70.
C
j
$70 $50 $0 $0
SOLUTION
MIX
QUANTITY
(RHS)
T C S
1
S
2

$0 S
1
100 2 1 1 0
$0 S
2
240 4 3 0 1
Z
j
$0 $0 $0 $0 $0
Z
j

C
j
-$70 -$50 $0 $0
Pivot column
total profit
The Second Simplex Tableau
Step 2. Select the variable to be replaced. Either S
1
or S
2
will
have to leave to make room for T in the basis. The following
ratios need to be calculated.
tables 50
table) per required 2(hours
available) time painting of 100(hours

For the S
1
row

tables 60
table) per required 4(hours
available) time carpentry of 240(hours

For the S
2
row

The Second Simplex Tableau
We choose the smaller ratio (50) and this determines the S
1

variable is to be replaced. This corresponds to point D on the
graph in Graph
C
j
$70 $50 $0 $0
SOLUTION
MIX
QUANTITY
(RHS)
T C S
1
S
2

Ratio
$0 S
1
100 2 1 1 0 50
$0 S
2
240 4 3 0 1 60
Z
j
$0 $0 $0 $0 $0
Z
j
- C
j
-$70 -$50 $0 $0
Pivot column
Pivot row
Pivot number
Mini. Ratio
The Second Simplex Tableau
Step 3. We can now begin to develop the second, improved
simplex tableau. We have to compute a replacement for the
pivot row. This is done by dividing every number in the pivot
row by the pivot number. The new version of the pivot row is
below.
1
2
2
5 0
2
1
.
5 0
2
1
.
*

0
2
0

50
2
100

C
j
SOLUTI
ON MIX
QUANTI
TY
T C S
1
S
2
$7
0
T 50 1 0.5 0.5 0
The Second Simplex Tableau
Step 4. Completing the rest of the tableau, the S
2
row, is
slightly more complicated. The right of the following
expression is used to find the left side.
Number in
New S
2
Row
=
Number in
Old S
2
Row

Intersecting
element in old
row

Corresponding Number
in the Replacement Row
0 = 4 (4) (1)
1 = 3 (4) (0.5)
2 = 0 (4) (0.5)
1 = 1 (4) (0)
40 = 240 (4) (50)
C
j
SOLUTION
MIX
QUANTI
TY
T C S
1
S
2
$7
0
T 50 1 0.5 0.5 0
$0 S
2
40 0 1 2 1
The Second Simplex Tableau
1
0
0
1
The T column contains and the S
2

column

Contains , necessary conditions for
variables to

be in the solution. The manipulations of steps 3
and 4 were designed to produce 0s and 1s in the
appropriate positions.
The Second Simplex Tableau
Step 5. The final step of the second iteration is to introduce
the effect of the objective function. This involves computing
the Z
j
C
j
rows. The Z
j
for the quantity row gives us the
gross profit and the other Z
j
represent the gross profit given
up by adding one unit of each variable into the solution.
Z
j
(for T column) = ($70)(1) + ($0)(0) = $70
Z
j
(for C column) = ($70)(0.5) + ($0)(1) = $35
Z
j
(for S
1
column) = ($70)(0.5) + ($0)(2) = $35
Z
j
(for S
2
column) = ($70)(0) + ($0)(1) = $0
Z
j
(for total profit) = ($70)(50) + ($0)(40) = $3,500
The Second Simplex Tableau
Completed second simplex tableau
C
j
$70 $50 $0 $0
SOLUTION
MIX
QUANTITY
(RHS) T C S
1
S
2

Rario
(RHS)
$70 T 50 1 0.5 0.5 0 50
$0 S
2
40 0 1 2 1 40
Z
j
$3,500 $70 $35 $35 $0 $3,500
Z
j
- C
j
$0 -$15 $35 $0
COLUMN
T C S
1
S
2
C
j
for column $70 $50 $0 $0
Z
j
for column $70 $35 $35 $0
Z
j
- C
j
for column $0 $15 $35 $0
Interpreting the Second Tableau
Current solution
The solution point of 50 tables and 0 chairs
(T = 50, C = 0) generates a profit of $3,500. T is a basic
variable and C is a nonbasic variable. This corresponds
to point D in Graph
Resource information
Slack variable S
2
is the unused time in the carpentry
department and is in the basis. Its value implies there
is 40 hours of unused carpentry time remaining. Slack
variable S
1
is nonbasic and has a value of 0 meaning
there is no slack time in the painting department.
Interpreting the Second Tableau
Substitution rates
Substitution rates are the coefficients in the heart of
the tableau. In column C, if 1 unit of C is added to the
current solution, 0.5 units of T and 1 unit of S
2
must be
given up. This is because the solution T = 50 uses up all
100 hours of painting time available.
Because these are marginal rates of substitution, so
only 1 more unit of S
2
is needed to produce 1 chair
In column S
1
, the substitution rates mean that if 1 hour
of slack painting time is added to producing a chair, 0.5
less of a table will be produced
Interpreting the Second Tableau
Net profit row
The Z
j
C
j
row is important for two reasons
First, it indicates whether the current solution is
optimal
When there are no positive values in the bottom row,
an optimal solution to a maximization LP has been
reached
The second reason is that we use this row to determine
which variable will enter the solution next
Developing the Third Tableau
Since the previous tableau is not optimal, we repeat the
five simplex steps
Step 1. Variable C will enter the solution as its Z
j
C
j
value of 15 is the largest
negative value. The C column is the new pivot column.
Step 2. Identify the pivot row by dividing the number in the quantity column by its
corresponding substitution rate in the C column.
chairs 100
5 0
50
row the For
.
: T
Ratio Minimum chairs 40
1
40
: row For the
2
S
Developing the Third Tableau
These ratios correspond to the values of C at points F and C in Graph. The S
2
row
has the smallest ratio so S
2
will leave the basis and will be replaced by C.
C
j
$70 $50 $0 $0
SOLUTION
MIX
QUANTITY T C S
1
S
2
Ratio
$70 T 50 1 0.5 0.5 0 50
$0 S
2
40 0 1 2 1 40Min
Z
j
$3,500 $70 $35 $35 $0 $3,500
Z
j
C
j
$0 $15 $35 $0
Pivot column
Pivot row
Pivot number
Developing the Third Tableau
Step 3. The pivot row is replaced by dividing every number in it by the pivot point
number
0
1
0
1
1
1
2
1
2

1
1
1
40
1
40

The new C row is

C
j
SOLUTION
MIX
QUANTITY T C S
1
S
2
$5 C 40 0 1 2 1
Developing the Third Tableau
Step 4. The new values for the T row may now be computed
Number in
new T row
=
Number in
old T row

Number above
pivot number

Corresponding number
in new C row
1 = 1 (0.5) (0)
0 = 0.5 (0.5) (1)
1.5 = 0.5 (0.5) (2)
0.5 = 0 (0.5) (1)
30 = 50 (0.5) (40)
C
j
SOLUTIO
N MIX
QUANTIT
Y
T C S
1
S
2
$7
0
T 30 1 0 1.5 0.5
$5
0
C 40 0 1 2 1
Developing the Third Tableau
Step 5. The Z
j
and Z
j
C
j
rows can now be calculated
Z
j
(for T column) = ($70)(1) + ($50)(0) = $70
Z
j
(for C column) = ($70)(0) + ($50)(1) = $50
Z
j
(for S
1
column) = ($70)(1.5) + ($50)(2) = $5
Z
j
(for S
2
column) = ($70)(0.5) + ($50)(1) = $15
Z
j
(for total profit) = ($70)(30) + ($50)(40) = $4,100
And the net profit per unit row is now
COLUMN
T C S
1
S
2
C
j
for column $70 $50 $0 $0
Z
j
for column $70 $50 $5 $15
Z
j
C
j
for column $0 $0 $5 $15
Developing the Third Tableau
Note that every number in the Z
j
C
j
row is 0 or positive indicating an optimal
solution has been reached
The optimal solution is
T = 30 tables
C = 40 chairs
S
1
= 0 slack hours in the painting department
S
2
= 0 slack hours in the carpentry department
profit = $4,100 for the optimal solution
Developing the Third Tableau
The final simplex tableau for the Flair Furniture problem corresponds to point
C in Graph.
C
j
$70 $50 $0 $0
SOLUTION
MIX QUANTITY T C S
1
S
2

$70 T 30 1 0 1.5 0.5
$50 C 40 0 1 2 1
Z
j
$4,100 $70 $50 $5 $15
Z
j
C
j
$0 $0 $5 $15
Arithmetic mistakes are easy to make
It is always a good idea to check your answer by going back to the
original constraints and objective function

Review of Procedures for Solving LP
Maximization Problems
I. Formulate the LP problems objective function and
constraints
II. Add slack variables to each less-than-or-equal-to constraint
and to the objective function
III. Develop and initial simplex tableau with slack variables in
the basis and decision variables set equal to 0. compute the
Z
j
and Z
j
C
j
values for this tableau.
IV. Follow the five steps until an optimal solution has been
reached
Review of Procedures for Solving LP
Maximization Problems
1. Choose the variable with the greatest negative Z
j
C
j

to enter the solution in the pivot column.
2. Determine the solution mix variable to be replaced and
the pivot row by selecting the row with the smallest
(nonnegative) ratio of the quantity-to-pivot column
substitution rate.
3. Calculate the new values for the pivot row
4. Calculate the new values for the other row(s)
5. Calculate the Z
j
and Z
j
C
j
values for this tableau. If
there are any Z
j
C
j
numbers lesser than 0, return to
step 1. If not, and optimal solution has been reached.
Surplus and Artificial Variables
Greater-than-or-equal-to () constraints are just as
common in real problems as less-than-or-equal-to ()
constraints and equalities
To use the simplex method with these constraints, they
must be converted to a special form similar to that made
for the less-than-or-equal-to () constraints
If they are not, the simplex technique is unable to set up an
initial solution in the first tableau
Consider the following two constraints
Constraint 1: 5X
1
+ 10X
2
+ 8X
3
210
Constraint 2: 25X
1
+ 30X
2
= 900

Surplus and Artificial Variables
Surplus variables
Greater-than-or-equal-to () constraints require a
different approach than the less-than-or-equal-to ()
constraints we have seen
They involve the subtraction of a surplus variable
rather than the addition of a slack variable
The surplus variable tells us how much the solution
exceeds the constraint amount
This is sometimes called negative slack
Surplus and Artificial Variables
To convert the first constraint we subtract a surplus variable, S
1
, to create an
equality
210 8 10 5 rewritten 1 Constraint
1 3 2 1
S X X X :
If we solved this for X
1
= 20, X
2
= 8, X
3
= 5, S
1
would be
210 8 10 5
1 3 2 1
S X X X
210 8(5) 10(8) 5(20)
1
S
210 40 80 100
1
S
220 210
1
S
units surplus 10
1
S
Surplus and Artificial Variables
Artificial variables
There is one more step in this process
If a surplus variable is added by itself, it would have a negative value in
the initial tableau where all real variables are set to zero
210 8(0) 10(0) 5(0)
1
S
210 0
1
S
210
1
S
But all variables in LP problems must be nonnegative at all times
Surplus and Artificial Variables
To resolve this we add in another variable called an artificial variable
210 8 10 5 completed 1 Constraint
1 1 3 2 1
A S X X X :
Now X
1
, X
2
, X
3
, and S
1
can all be 0 in the initial solution and A
1
will equal 210
The same situation applies in equality constraint equations as well
900 30 25 rewritten 2 Constraint
2 2 1
A X X :
Surplus and Artificial Variables
Artificial variables are inserted into equality constraints so
we can easily develop an initial feasible solution
When a problem has many constraint equations with many
variables, it is not possible to eyeball an initial solution
Using artificial variables allows us to use the automatic
initial solution of setting all the other variables to 0
Unlike slack or surplus variables, artificial variables have no
meaning in the problem formulation
They are strictly a computational tool, they will be gone in
the final solution
Surplus and Artificial Variables
Surplus and artificial variables in the objective
function
Both types of variables must be included in the
objective function
Surplus variables, like slack variables, carry a $0 cost
coefficient
Since artificial variables must be forced out of the
solution, we assign an arbitrarily high cost
By convention we use the coefficient M (or M in
maximization problems) which simply represents a
very large number
Surplus and Artificial Variables
A problem with this objective function
3 2 1
7 9 5 cost Minimize X X X $ $ $
And the constraint equations we saw before would appear as follows:
Minimize cost = $5X
1
+ $9X
2
+ $7X
3
+ $0S
1
+ $MA
1
+ $MA
2

subject to 5X
1
+ 10X
2
+ 8X
3
1S
1
+ 1A
1
+ 0A
2
= 210
25X
1
+ 30X
2
+ 0X
3
+ 0S
1
+ 0A
1
+ 1A
2
= 900
Solving Minimization Problems
Once the necessary equations are developed for a
minimization problem, we can use the simplex
method to solve for an optimal solution

The Muddy River Chemical Corporation Example
The Muddy River Chemical Corporation must produce exactly
1,000 pounds of a special mixture of phosphate and
potassium for a customer
Phosphate costs $5 per pound and potassium $6 per pound
No more than 300 pounds of phosphate can be used and at
least 150 pounds of potassium must be used
The company wants to find the least-cost blend of the two
ingredients
The Muddy River Chemical Corporation Example
The model formulation would be
Minimize cost = $5X
1
+ $6X
2
subject to X
1
+ X
2
= 1,000 lb
X
1
300 lb
X
2
150 lb
X
1
, X
2
0
where
X
1
= number of pounds of phosphate
X
2
= number of pounds of potassium
The Muddy River Chemical Corporation Example
Graphical analysis
Because there are only two decision variables, we can
plot the constraints and the feasible region as shown in
Figure 9.3
Because X
1
+ X
2
= 1,000 is an equality, the optimal
solution must lie on this line
It must also lie between points A and B because of the
X
1
300 constraint
It turns out the X
2
150 is redundant and nonbinding
The optimal corner point is point B (300, 700) for a
total cost of $5,700
The Muddy River Chemical Corporation Example

1,000

800

600

400

200
100
0
| | | | | |
200 400 600 800 1,000
X
2
X
1
X
2
150

X
1
+ X
2
= 1,000
X
1
300

C
A
B
D E
F G H
Figure 9.3
The Muddy River Chemical Corporation Example
Rarely will problems be this simple
The simplex method can be used to solve much
more complex problems
In this example, the simplex method will start at
coroner point E, move to point F, then G and
finally to point B which is the optimal solution
The Muddy River Chemical Corporation Example
Converting the constraints and objective function
The necessary artificial variables, slack variables, and
surplus variables need to be added to the equations
The revised model is
Minimize cost = $5X
1
+ $6X
2
+ $0S
1
+ $0S
2
+ $MA
1
+ $MA
2
subject to 1X
1
+ 1X
2
+ 0S
1
+ 0S
2
+ 1A
1
+ 0A
2
= 1,000
1X
1
+ 0X
2
+ 1S
1
+ 0S
2
+ 0A
1
+ 0A
2
= 300
0X
1
+ 1X
2
+ 0S
1
1S
2
+ 0A
1
+ 1A
2
= 150
X
1
, X
2
, S
1
, S
2
, A
1
, A
2
0
Rules of the Simplex Method for
Minimization Problems
Minimization problems are quite similar to the
maximization problems tackled earlier
The significant difference is the Z
j
C
j
row
We will now choose the variable with the positive Z
j
C
j

that gives the largest improvement
We select the variable that decreases costs the most
In minimization problems, an optimal solution is reached
when all the numbers in the Z
j
C
j
are 0 or negative
All other steps in the simplex method remain the same
Steps for Simplex Minimization Problems
1. Choose the variable with the greatest positive Z
j
C
j
to
enter the solution in the pivot column.
2. Determine the solution mix variable to be replaced and
the pivot row by selecting the row with the smallest
(nonnegative) ratio of the quantity-to-pivot column
substitution rate.
3. Calculate the new values for the pivot row
4. Calculate the new values for the other row(s)
5. Calculate the Z
j
and Z
j
C
j
values for this tableau. If
there are any Z
j
C
j
numbers greater than 0, return to
step 1. if not, and optimal solution has been reached.
First Simplex Tableau for the Muddy River
Chemical Corporation Example
The initial tableau is set up in the same manner as the in
the maximization problem
The first three rows are
Note the costs for the artificial variables are $M
We simply treat this as a very large number which forces
the artificial variables out of the solution quickly
C
j
SOLUTION
MIX
QUANTITY X
1
X
2
S
1
S
2
A
1
A
2
$M A
1
1,000 1 1 0 0 1 0
$0 S
1
300 1 0 1 0 0 0
$M A2 150 0 1 0 1 0 1
First Simplex Tableau for the Muddy River
Chemical Corporation Example
The numbers in the Z
j
are computed by multiplying the C
j

column on the far left of the table times the
corresponding numbers in each other column
Z
j
(for X
1
column) = $M(1) + $0(1) + $M(0) = $M
Z
j
(for X
2
column) = $M(1) + $0(0) + $M(1) = $2M

Z
j
(for S
1
column) = $M(0) + $0(1) + $M(0) = $0

Z
j
(for S
2
column) = $M(0) + $0(0) + $M(1) = $M

Z
j
(for A
1
column) = $M(1) + $0(0) + $M(0) = $M

Z
j
(for A
2
column) = $M(0) + $0(0) + $M(1) = $M

Z
j
(for total cost) = $M(1,000) + $0(300) + $M(150) = $1,150M

First Simplex Tableau for the Muddy River
Chemical Corporation Example
The Z
j
C
j
entires are determined as follows

COLUMN
X
1
X
2
S
1
S
2
A
1
A
2
C
j
for column

$5 $6 $0 $0 $M $M
Z
j
for column

$M $2M $0 $M $M $M
Z
j
- C
j
for column

$M $5 $2M $6 $0 $M $0 $0
First Simplex Tableau for the Muddy River
Chemical Corporation Example
The initial solution was obtained by letting each of the
variables X
1
, X
2
, and S
2
assume a value of 0
The current basic variables are A
1
= 1,000, S
1
= 150, and A
2

= 150
The complete solution could be expressed in vector form
as
=
X
1
X
2
S
1

S
2
A
1
A
2


0
0
300
0
1,000
150
First Simplex Tableau for the Muddy River
Chemical Corporation Example
The initial tableau
C
j
$5 $6 $0 $0 $M $M
SOLUTI
ON MIX
QUANTI
TY
X
1
X
2
S
1
S
2
A
1
A
2
Ratio

$M A
1
1,000 1 1 0 0 1 0 1000
$0 S
1
300 1 0 1 0 0 0
$M A2 150 0 1 0 1 0 1 150
Z
j
$1,150M $M $2M $0 $M $M $M
Z
j
- C
j
$M $5 2M $6 $0 $M $0 $0
Pivot column
Pivot number Pivot row

Developing the Second Tableau

In the Z
j
C
j
row there are two entries with positive
values, X
1
and X
2
This means an optimal solution does not yet exist
The positive entry for X
2
indicates it has the will result in
the largest improvement, which means it will enter the
solution next
To find the variable that will leave the solution, we divide
the elements in the quantity column by the respective
pivot column substitution rates
Developing the Second Tableau
1,000
1
000 1
row the For
1

,
A
0
300
row the For
1
S
150
1
150
row the For
2
A
(this is an undefined ratio, so
we ignore it)
(smallest quotient,
indicating pivot row)
Hence the pivot row is the A
2
row and the pivot number is
at the intersection of the X
2
column and the A
2
row
Infinite
Ratio
Developing the Second Tableau
The entering row for the next tableau is found by dividing each element in the
pivot row by the pivot number
(New row numbers) = (Numbers in old row)
Number above or
below pivot number
Corresponding number in
newly replaced row

A
1
Row S
1
Row
1 = 1 (1)(0) 1 = 1 (0)(0)
0 = 1 (1)(1) 0 = 0 (0)(1)
0 = 0 (1)(0) 1 = 1 (0)(0)
1 = 0 (1)(1) 0 = 0 (0)(1)
1 = 1 (1)(0) 0 = 0 (0)(0)
1 = 0 (1)(1) 0 = 0 (0)(1)
850 = 1,000 (1)(150) 300 = 300 (0)(150)
Developing the Second Tableau
The Z
j
and Z
j
C
j
rows are computed next
Z
j
(for X
1
) = $M(1) + $0(1) + $6(0) = $M
Z
j
(for X
2
) = $M(0) + $0(0) + $6(1) = $6
Z
j
(for S
1
) = $M(0) + $0(1) + $6(0) = $0

Z
j
(for S
2
) = $M(1) + $0(0) + $6(1) = $M 6

Z
j
(for A
1
) = $M(1) + $0(0) + $6(0) = $M

Z
j
(for A
2
) = $M(1) + $0(0) + $6(1) = $M + 6

Z
j
(for total cost) = $M(850) + $0(300) + $6(150) = $850M + 900
COLUMN
X
1
X
2
S
1
S
2
A
1
A
2
C
j
for column

$5 $6 $0 $0 $M $M
Z
j
for column

$M $6 $0 $M 6 $M $M + 6
Z
j
- C
j
for column

$M $5 $0 $0 $M 6 $0
$2M
+6
Developing the Second Tableau
Second simplex tableau
C
j
$5 $6
$
0
$0 $M $M
SOLUTIO
N MIX
QUANTI
TY
X
1
X
2
S
1
S
2
A
1
A
2
Ratio
$
M
A
1
850 1 0 0 1 1 1 850
$0 S
1
300 1 0 1 0 0 0 300
$6 X
2
150 0 1 0 1 0 1 150
Z
j
$850M +
$900
$M $6
$
0
$M 6 $M $M + 6
Z
j
- C
j
$M $5 $0
$
0
$M $6 $0
$2M +
6
Pivot column
Pivot number
Pivot row
Developing a Third Tableau
850
1
850
row the For
1
A
300
1
300
row the For
1
S
undefined
0
150
row the For
2
X
(smallest ratio)
Hence variable S
1
will be replaced by X
1

The new pivot column is the X
1
column and we check the quantity column-to-
pivot column ratio
Ratio
Developing a Third Tableau
To replace the pivot row we divide each number in the S
1
row by 1 leaving it
unchanged
The other calculations are shown below
A
1
Row S
1
Row
0 = 1 (1)(1) 0 = 0 (0)(1)
0 = 0 (1)(0) 1 = 1 (0)(0)
1 = 0 (1)(1) 0 = 0 (0)(1)
1 = 1 (1)(0) 1 = 1 (0)(0)
1 = 1 (1)(0) 0 = 0 (0)(0)
1 = 1 (1)(0) 1 = 1 (0)(0)
550 = 850 (1)(300) 150 = 150 (0)(300)
Developing a Third Tableau
The Z
j
and Z
j
C
j
rows are computed next
Z
j
(for X
1
) = $M(0) + $5(1) + $6(0) = $5
Z
j
(for X
2
) = $M(0) + $5(0) + $6(1) = $6
Z
j
(for S
1
) = $M(1) + $5(1) + $6(0) = $M + 5

Z
j
(for S
2
) = $M(1) + $5(0) + $6(1) = $M 6

Z
j
(for A
1
) = $M(1) + $5(0) + $6(0) = $M

Z
j
(for A
2
) = $M(1) + $5(0) + $6(1) = $M + 6

Z
j
(for total cost) = $M(550) + $5(300) + $6(150) = $550M + 2,400
COLUMN
X
1
X
2
S
1
S
2
A
1
A
2
C
j
for column

$5 $6 $0 $0 $M $M
Z
j
for column

$5 $6 $M + 5 $M 6 $M $M + 6
Z
j
- C
j
for column $0 $0 $M + 5 $M 6 $0 $2M + 6
Developing a Third Tableau
The third simplex tableau for the Muddy River Chemical problem
C
j
$5 $6 $0 $0
$
M
$M
SOLUTI
ON MIX
QUANTIT
Y
X
1
X
2
S
1
S
2
A
1
A
2
Ratio

$
M
A
1
550 0 0 1 1 1 1 550
$5 X
1
300 1 0 1 0 0 0
$6 X
2
150 0 1 0 1 0 1
Z
j
$550M +
2,400
$5 $6 $M + 5 $M 6
$
M
$M + 6
Z
j
- C
j
$0 $0 $M + 5 $M 6 $0 $2M + 6
Pivot column
Pivot number
Pivot row

_
Fourth Tableau for Muddy River
The new pivot column is the S
2
column
550
1
550
row the For
1
A
0
300
row the For
1
X
1
150
row the For
2

X
(row to be replaced)
(undefined)
(not considered because
it is negative)
Infinity

Fourth Tableau for Muddy River
Each number in the pivot row is again divided by 1
The other calculations are shown below
X
1
Row X
2
Row
1 = 1 (0)(0) 0 = 0 (1)(0)
0 = 0 (0)(0) 1 = 1 (1)(0)
1 = 1 (0)(1) 1 = 0 (1)(1)
0 = 0 (0)(1) 0 = 1 (1)(1)
0 = 0 (0)(1) 1 = 0 (1)(1)
0 = 0 (0)(1) 0 = 1 (1)(1)
300 = 300 (0)(550) 700 = 150 (1)(550)
Fourth Tableau for Muddy River
Finally the Z
j
and Z
j
C
j
rows are computed
Z
j
(for X
1
) = $0(0) + $5(1) + $6(0) = $5
Z
j
(for X
2
) = $(0) + $5(0) + $6(1) = $6
Z
j
(for S
1
) = $0(1) + $5(1) + $6(1) = $1

Z
j
(for S
2
) = $0(1) + $5(0) + $6(0) = $0

Z
j
(for A
1
) = $0(1) + $5(0) + $6(1) = $6

Z
j
(for A
2
) = $0(1) + $5(0) + $6(0) = $0

Z
j
(for total cost) = $0(550) + $5(300) + $6(700) = $5,700
COLUMN
X
1
X
2
S
1
S
2
A
1
A
2
C
j
for column

$5 $6 $0 $0 $M $M
Z
j
for column

$5 $6 $1 $0 $6 $0
Z
j
- C for column $0 $0 $1 $0 $M +6 $M
Fourth Tableau for Muddy River
Fourth and optimal tableau for the Muddy River Chemical Corporation
problem
C
j
$5 $6 $0 $0 $M $M
SOLUTIO
N MIX
QUANTI
TY
X
1
X
2
S
1
S
2
A
1
A
2
$0 S
2
550 0 0 1 1 1 1
$5 X
1
300 1 0 1 0 0 0
$6 X
2
700 0 1 1 0 1 0
Z
j
$5,700 $5 $6 $1 $0 $6 $0
Z
j
- C $0 $0 $1 $0
$M
+6
$M
Review of Procedures for Solving LP
Minimization Problems
I. Formulate the LP problems objective function and
constraints
II. Include slack variables to each less-than-or-equal-to
constraint and both surplus and artificial variables to
greater-than-or-equal-to constraints and add all
variables to the objective function
III. Develop and initial simplex tableau with artificial and
slack variables in the basis and the other variables set
equal to 0. compute the Z
j
and
Z
j
C
j
values for this tableau.
IV. Follow the five steps until an optimal solution has been
reached
Review of Procedures for Solving LP
Minimization Problems
1. Choose the variable with the positive Z
j
C
j
indicating
the greatest improvement to enter the solution in the
pivot column
2. Determine the row to be replaced and the pivot row by
selecting the row with the smallest (nonnegative)
quantity-to-pivot column substitution rate ratio
3. Calculate the new values for the pivot row
4. Calculate the new values for the other row(s)
5. Calculate the Z
j
and Z
j
C
j
values for the tableau. If
there are any Z
j
C
j
numbers greater than 0, return to
step 1. If not, and optimal solution has been reached.
Special Cases
We have seen how special cases arise when
solving LP problems graphically
They also apply to the simplex method
You remember the four cases are
Infeasibility
Unbounded Solutions
Degeneracy
Multiple Optimal Solutions
Infeasibility
Infeasibility comes about when there is no solution that
satisfies all of the problems constraints
In the simplex method, an infeasible solution is indicated
by looking at the final tableau
All Z
j
C
j
row entries will be of the proper sign to imply
optimality, but an artificial variable will still be in the
solution mix
A situation with no feasible solution may exist if the
problem was formulated improperly
Infeasibility
Illustration of infeasibility
C
j
$5 $8 $0 $0 $M $M
SOLUTIO
N MIX
QUANTI
TY
X
1
X
2
S
1
S
2
A
1
A
2
$5 X
1
200 1 0 2 3 1 0
$8 X
2
100 0 1 1 2 2 0
$
M
A
2
20 0 0 0 1 1 1
Z
j
$1,800 +
20M
$5 $8 $2 $31 M $21 M $M
Z
j
- C
j
$0 $0 $2 $31 M $21 2M $0
Unbounded Solutions
Unboundedness describes linear programs that do not
have finite solutions
It occurs in maximization problems when a solution
variable can be made infinitely large without violating a
constraint
In the simplex method this will be discovered prior to
reaching the final tableau
It will be manifested when trying to decide which variable
to remove from the solution mix
If all the ratios turn out to be negative or undefined, it
indicates that the problem is unbounded
Unbounded Solutions
Problem with an unbounded solution
C
j
$6 $9 $0 $0
SOLUTIO
N MIX
QUANTITY X
1
X
2
S
1
S
2
Ratio
$9 X
2
30 1 1 2 0 -
$0 S
2
10 2 0 1 1 -
Z
j
$270 $9 $9 $18 $0
Z
j
- C
j
$15 $0 $18 $0
Pivot column
Unbounded Solutions
The ratios from the pivot column
1
30
row the for Ratio
2

: X
2
10
row the for Ratio
2

: S
Negative ratios
unacceptable
Since both pivot column numbers are negative, an unbounded solution is
indicated
Degeneracy
Degeneracy develops when three constraints pass through
a single point
For example, suppose a problem has only these three
constraints X
1
10, X
2
10, and X
1
+ X
2
< 20
All three constraint lines will pass through the point (10,
10)
Degeneracy is first recognized when the ratio calculations
are made
If there is a tie for the smallest ratio, this is a signal that
degeneracy exists
As a result of this, when the next tableau is developed,
one of the variables in the solution mix will have a value of
zero
Degeneracy
Degeneracy could lead to a situation known as cycling in
which the simplex algorithm alternates back and forth
between the same nonoptimal solutions
One simple way of dealing with the issue is to select either
row in question arbitrarily
If unlucky and cycling does occur, simply go back and
select the other row
Degeneracy
Problem illustrating degeneracy
C
j
$5 $8 $2 $0 $0 $0
SOLUTIO
N MIX
QUANTIT
Y
X
1
X
2
X
3
S
1
S
2
S
3
Ratio
$8 X
2
10 0.25 1 1 2 0 0 40
$0 S
2
20 4 0 0.33 1 1 0 5Min
$0 S
3
10 2 0 2 0.4 0 1 5Min
Z
j
$80 $2 $8 $8 $16 $0 $0
Z
j
- C
j
$3 $0 $6 $16 $0 $0
Pivot column
Degeneracy
The ratios are computed as follows
40
25 0
10
row the For
2

.
: X
5
4
20
row the For
2
: S
5
2
10
row the For
3
: S
Tie for the smallest ratio
indicates degeneracy
Multiple Optimal Solutions
In the simplex method, multiple, or alternate, optimal
solutions can be spotted by looking at the final tableau
If the Z
j
C
j
value is equal to 0 for a variable that is not in
the solution mix, more than one optimal solution exists
Multiple Optimal Solutions
A problem with alternate optimal solutions
C
j
$3 $2 $0 $0
SOLUTION
MIX
QUANTITY X
1
X
2
S
1
S
2

$2 X
2
6 1.5 1 1 0
$0 S
2
3 1 0 0.5 1
Z
j
$12 $3 $2 $2 $0
Z
j
- C
j
$0 $0 $2 $0
Sensitivity Analysis with the Simplex
Tableau
Sensitivity analysis shows how the optimal solution and
the value of its objective function change given changes in
various inputs to the problem
Computer programs handling LP problems of all sizes
provide sensitivity analysis as an important output feature
Those programs use the information provided in the final
simplex tableau to compute ranges for the objective
function coefficients and ranges for the RHS values
They also provide shadow prices, a concept we will
introduce in this section
High Note Sound Company Revisited
You will recall the model formulation is
Maximize profit = $50X
1
+ $120X
2
subject to 2X
1
+ 4X
2
80 (hours of
electrician time)
3X
1
+ 1X
2
60 (hours of technician
time)
X
2
= 20 receivers
S
2
= 40 hours slack in technician time
X
1
= 0 CD players
S
1
= 0 hours slack in electrician time
Basic
variables
And the optimal solution is
Nonbasic
variables
High Note Sound Company Revisited
High Note Sound Company graphical solution
b = (16, 12)
Optimal Solution at Point a
X
1
= 0 CD Players
X
2
= 20 Receivers
Profits = $2,400
a = (0, 20)
Isoprofit Line: $2,400 = 50X
1
+ 120X
2

60

40

20
10
0
X
2

| | | | | |
10 20 30 40 50 60
X
1

(receivers)
(CD players) c = (20, 0)
Figure 9.4
Changes in the
Objective Function Coefficient
C
j
$50 $120 $0 $0
SOLUTION
MIX
QUANTITY X
1
X
2
S
1
S
2

$12
0
X
2
20 0.5 1 0.25 0
$0 S
2
40 2.5 0 0.25 1
Z
j
$2,400 $60 $120 $30 $0
Z
j
- C
j


$10 $0 $30 $0
Table 9.15
Optimal solution by the simplex method
Changes in the
Objective Function Coefficient
Nonbasic objective function coefficient
The goal is to find out how sensitive the problems optimal solution is to
changes in the contribution rates of variables not currently in the basis
How much would the objective function coefficients have to change
before X
1
or S
1
would enter the solution mix and replace one of the basic
variables?
The answer lies in the Z
j
C
j
row of the final simplex tableau
Changes in the
Objective Function Coefficient
This is a maximization problem so the basis will not
change unless the Z
j
C
j
value of one of the nonbasic
variables becomes greater than 0
The values in the basis will not change as long as C
j
Z
j

The solution will not change as long as X
1
does not exceed
$60 and the contribution rate of S
2
does not exceed $30
These values can also be made smaller without limit in
this situation
So the range of insignificance for the nonbasic variables is
60 for
1
$ ) ( X C
j
30 for
1
$ ) ( S C
j
Changes in the
Objective Function Coefficient
Basic objective function coefficient
Sensitivity analysis on objective function coefficients of
variables in the basis or solution mix is slightly more
complex
A change in the profit or cost of a basic variable can
affect the Z
j
C
j
values for all nonbasic variables
Thats because the C
j
value is in both the row and
column
This then impacts the Z
j
C
j
row
Changes in the
Objective Function Coefficient
Consider a change in the profit contribution of stereo
receivers
The current coefficient is $120
The changed coefficient will be represented as
The revised final tableau will then be
C
j
$50 $120 + $0 $0
SOLUTI
ON MIX
QUANTI
TY
X
1
X
2
S
1
S
2

$120 + X
2
20 0.5 1 0.25 0
$0 S
2
40 2.5 0 0.25 1
Z
j

$2,400 +
20
$60 + 0.5 $120 + $30 + 0.25 $0
Z
j
- C
j
$10 + 0.5 $0 $30 + 0.25 $0
Changes in the
Objective Function Coefficient
The new Z
j
C
j
values in the table were determined in
the same way as previous examples
How may the value of vary so that all Z
j
C
j
entries
remain negative?
To find out, solve for in each column
10 0.5 0
10 0.5
20 or 20
This inequality means the optimal solution will not change
unless X
2
s profit coefficient decreases by at least $20, =
20
Changes in the
Objective Function Coefficient
Variable X
1
will not enter the basis unless the profit per
receiver drops to $100 or less
For the S
1
column
30 0.25 0
30 0.25
120 or 120
Since the first inequality is more binding, we can say that
the range of optimality for X
2
s profit coefficient is
) ( $
2
for 100 X C
j
Changes in the
Objective Function Coefficient
In larger problems, we would use this procedure to test
for the range of optimality of every real decision variable
in the final solution mix
Using this procedure helps us avoid the time-consuming
process of reformulating and resolving the entire LP
problem each time a small change occurs
Within the bounds, changes in profit coefficients will not
force a change in the optimal solution
The value of the objective function will change, but this is
a comparatively simple calculation
Changes in Resources or RHS Values
Making changes in the RHS values of constraints result in
changes in the feasible region and often the optimal
solution
Shadow prices
How much should a firm be willing to pay for one
additional unit of a resource?
This is called the shadow price
Shadow pricing provides an important piece of economic
information
This information is available in the final tableau
Changes in Resources or RHS Values
Final tableau for High Note Sound
C
j
$50 $120 $0 $0
SOLUTION
MIX
QUANTITY X
1
X
2
S
1
S
2

$12
0
X
2
20 0.5 1 0.25 0
$0 S
2
40 2.5 0 0.25 1
Z
j
$2,400 $60 $120 $30 $0
Z
j
- C
j
$10 $0 $30 $0
Objective function increases by $30 if 1
additional hour of electricians time is
made available
Changes in Resources or RHS Values
An important property of the Z
j
C
j
row is that the negatives
of the numbers in its slack variable (S
i
) columns provide us
with shadow prices
A shadow price is the change in value of the objective
function from an increase of one unit of a scarce resource
High Note Sound is considering hiring an extra electrician at
$22 per hour
In the final tableau we see S
1
(electricians time) is fully
utilized and has a Z
j
C
j
value of $30
They should hire the electrician as the firm will net $8 (= $30
$22)
Changes in Resources or RHS Values
Should High Note Sound hire a part-time audio technician
at $14 per hour?
In the final tableau we see S
2
(audio technician time) has
slack capacity (40 hours) a Z
j
C
j
value of $0
Thus there would be no benefit to hiring an additional
audio technician
Changes in Resources or RHS Values
Right-hand side ranging
We cant add an unlimited amount of a resource
without eventually violating one of the other
constraints
Right-hand-side ranging tells us how much we can
change the RHS of a scarce resource without changing
the shadow price
Ranging is simple in that it resembles the simplex
process
Changes in Resources or RHS Values
This table repeats some of the information from the final
tableau for High Note Sound and includes the ratios
QUANTITY S
1
RATIO
20 0.25 20/0.25 = 80
40 0.25 40/0.25 = 160
The smallest positive ratio (80 in this example) tells us how many hours the
electricians time can be reduced without altering the current solution mix
Changes in Resources or RHS Values
The smallest negative ratio (160) tells us the number of
hours that can be added to the resource before the
solution mix changes
In this case, thats 160 hours
So the range over which the shadow price for electricians
time is valid is 0 to 240 hours
The audio technician resource is slightly different
There is slack in this resource (S
2
= 40) so we can reduce
the amount available by 40 before a shortage occurs
However, we can increase it indefinitely with no change in
the solution
Changes in Resources or RHS Values
The substitution rates in the slack variable column can
also be used to determine the actual values of the
solution mix variables if the right-hand-side of a constraint
is changed using the following relationship
New quantity Original
quantity
Substitution rate Change in the RHS
= +
Changes in Resources or RHS Values
For example, if 12 more electrician hours were made
available, the new values in the quantity column of the
simplex tableau are found as follows
ORIGINAL QUANTITY S
1
NEW QUANTITY
20 0.25 20 + 0.25(12) = 23
40 0.25 40 + (0.25)(12) = 37
If 12 hours were added, X
2
= 23 and S
2
= 37
Total profit would be 50(0) + 120(23) = $2,760, an increase of $360
This of course, is also equal to the shadow price of $30 times the 12
additional hours
Sensitivity Analysis by Computer
Solver in Excel has the capability of producing sensitivity
analysis that includes the shadow prices of resources
The following slides present the solution to the High Note
Sound problem and the sensitivity report showing shadow
prices and ranges
Sensitivity Analysis by Computer
Program 9.1a
Sensitivity Analysis by Computer
Program 9.1b
The Dual
Every LP problem has another LP problem associated with it
called the dual
The first way of stating a problem (what we have done so far)
is called the primal
The second way of stating it is called the dual
The solutions to the primal and dual are equivalent, but they
are derived through alternative procedures
The dual contains economic information useful to managers
and may be easier to formulate
The Dual
Generally, if the LP primal is a maximize profit problem with
less-than-or-equal-to resource constraints, the dual will
involve minimizing total opportunity cost subject to greater-
than-or-equal-to product profit constraints
Formulating a dual problem is not complex and once
formulated, it is solved using the same procedure as a regular
LP problem
The Dual
Illustrating the primal-dual relationship with the High Note
Sound Company data
The primal problem is to determine the best production mix
between CD players (X
1
) and receivers (X
2
) to maximize profit
Maximize profit = $50X
1
+ $120X
2
subject to 2X
1
+ 4X
2
80 (hours of available
electrician time)
3X
1
+ 1X
2
60 (hours of audio
technician time
available)
The Dual
The dual of this problem has the objective of minimizing
the opportunity cost of not using the resources in an
optimal manner
The variables in the dual are
U
1
= potential hourly contribution of
electrician time, or the dual value of 1
hour of electrician time
U
2
= the imputed worth of audio technician
time, or the dual of technician resource
Each constraint in the primal problem will have a
corresponding variable in the dual and each decision
variable in the primal will have a corresponding constraint
in the dual
The Dual
The RHS quantities of the primal constraints become the
duals objective function coefficients
The total opportunity cost will be represented by the
function
Minimize opportunity cost = 80U
1
+ 60U
2

The corresponding dual constraints are formed from the
transpose of the primal constraint coefficients
2 U
1
+ 3 U
2
50
4 U
1
+ 1 U
2
120
Primal profit coefficients
Coefficients from the second
primal constraint
Coefficients from the first primal
constraint
The Dual
The first constraint says that the total imputed value or
potential worth of the scarce resources needed to
produce a CD player must be at least equal to the profit
derived from the product
The second constraint makes an analogous statement for
the stereo receiver product
Steps to Form the Dual
If the primal is a maximization problem in the standard form,
the dual is a minimization, and vice versa
The RHS values of the primal constraints become the duals
objective coefficients
The primal objective function coefficients become the RHS
values of the dual constraints
The transpose of the primal constraint coefficients become
the dual constraint coefficients
Constraint inequality signs are reversed
Solving the Dual of the High Note Sound
Company Problem
The formulation can be restated as
= 80U
1
+ 60U
2
+ 0S
1
+ 0S
2
+ MA
1
+ MA
2


2U
1
+ 3U
2
0S
1
+ 1A
1
= 50
4U
1
+ 1U
2
0S
2
+ 1A
2
= 120
Minimize
opportunity
cost
subject to:
Solving the Dual of the High Note Sound
Company Problem
The first and second tableaus
C
j
80 60 0 0 M M
SOLUTIO
N MIX
QUANTIT
Y
U
1
U
2
S
1
S
2
A
1
A
2
First
tablea
u
$M A
1
50 2 3 1 0 1 0
$M A
2
120 4 1 0 1 0 1
Z
j
$170M $6M $4M $M $M $M $M
Z
j
- C
j

80
+6M
60 +
4M
M M 0 0
Secon
d
tablea
u
$8
0
U
1
25 1 1.5 0.5 0 0.5 0
$M A
2
20 0 5 2 1 2 1
Z
j
$2,000 +
20M
$80
$120
5M
$40 +
2M
$M $40 2M $M
Z
j
- C
j
0 5M + 60 2M40 M 3M 40 0
Solving the Dual of the High Note Sound
Company Problem
Comparison of the primal and dual optimal tableaus
Primals Optimal Solution
C
j
$50 $120 $0 $0
Solution Mix X
1
X
2
S
1
S
2
Quantity
$120 X
2
0.5 1 0.25 0 20
$0 S
2
2.5 0 0.25 1 40
Z
j
60 120 30 0 $2,400
C
j
Z
j
10 0 30 0
Duals Optimal Solution
C
j
80 60 0 0 M M
Solution Mix U
1
U
2
S
1
S
2
A
1
A
2
Quantity
80 U
1
1 0.25 0 0.25 0 0.5 30
0 S
1
0 2.5 1 0.5 1 0.25 10
Z
j
80 20 0 20 0 20 $2,400
C
j
Z
j
0 40 0 20 M M 20
Solving the Dual of the High Note Sound
Company Problem
In the final simplex tableau of a primal problem, the
absolute values of the numbers in the Z
j
C
j
row under
the slack variables represent the solutions to the dual
problem
They are shadow prices in the primal solution and
marginal profits in the dual
The absolute value of the numbers of the Z
j
C
j
values of
the slack variables represent the optimal values of the
primal X
1
and X
2
variables
The maximum opportunity cost derived in the dual must
always equal the maximum profit derived in the primal
Karmakars Algorithm
In 1984, Narendra Karmakar developed a new method of
solving linear programming problems called the Karmakar
algorithm
The simplex method follows a path of points on the
outside edge of feasible space
Karmakars algorithm works by following a path a points
inside the feasible space
It is much more efficient than the simplex method
requiring less computer time to solve problems
It can also handle extremely large problems allowing
organizations to solve previously unsolvable problems