Linear Problem Simplex Adding a slack variable to each of the constraints

permits us to write the problem in standard form.

Method
Unit 6.3

An Algebraic Overview of the Simplex Method

HighTech Industries imports electronic components that

are used to assemble two different models of personal
computers.
The simplex method finds solutions for these equations
One model is called the Deskpro, and the other model is by assigning zero values to two of the variables and then
called the Portable. HighTech’s management is currently solving for the values of the remaining three variables.
interested in developing a weekly production schedule
for both products. For example, if we set x 1=0 and s1=0, the system of
constraint equations becomes
 The Deskpro generates a profit contribution of $50
per unit, and the Portable generates a profit
contribution of $40 per unit.
 A maximum of 150 hours of assembly time is
available.
 Each unit of the Deskpro requires 3 hours of
assembly time.
Each unit of the Portable requires 5 hours of Thus, we obtain the following solution to the three-
assembly time equation, five-variable set of linear equations:
 HighTech currently has only 20 Portable display
components in inventory; thus, no more than 20 x 1=50 , x2 =0 , s 1=0 , s 2=20 , s 3=−100
units of the Portable may be assembled.
The solution is referred to as a basic solution.
 Only 300 square feet of warehouse space can be
made available for new production. Assembly of Basic Solution
each Deskpro requires 8 square feet of warehouse
space; similarly, each Portable requires 5 square To determine a basic solution, set n – m of the variables
feet. equal to zero, and solve the m linear constraint
equations for the remaining m variables.
To develop a linear programming model for the
HighTech problem, we will use the following decision We shall refer to the n 2 m variables set equal to zero as
variables: the nonbasic variables and the remaining m variables
as the basic variables. Thus, in the preceding example,
X 1 = number of units of the Deskpro x 2 and s1 are the nonbasic variables, and x 1 , s2 , and S3
are the basic variables.
X 2 = number of units of the Portable
A basic solution can be either feasible or infeasible
The complete mathematical model for this problem is:
A basic feasible solution is a basic solution that also
satisfies the nonnegativity conditions. The basic solution
found by setting x 2 and s1 equal to zero and then solving
for x 1, s2, and s3 is not a basic feasible solution because
s3=−100

However, suppose that we had chosen to make x 1 and

x 2 nonbasic variables by setting x 1 = 0 and x 2 = 0.
Solving for the corresponding basic solution is easy
because with x 1=x 2=0, the three constraint equations
reduce to s1=150 , s2=20 , s3=300.

, The complete solution would be

 x 1=0 , x 2=0 , s 1=150 , s 2=20 , s 3=300 the basic variables are the values of the right-hand sides
of the constraint equations.

The second property that enables us to find a basic

The following graph shows all the constraint equations feasible solution requires the values of the right-hand
and basic solutions for the HighTech problem. Circle sides of the constraint equations be nonnegative. This
point 1 – 5 are basic feasible solutions; circled points 6 – nonnegativity ensures that the basic solution obtained by
9 are basic solutions that are feasible. setting the basic variables equal to the values of the
righthand sides will be feasible.

If a linear programming problem satisfies these two

properties, it is said to be in tableau form.

Thus, we see that the standard-form representation of

the HighTech problem is already in tableau form.

In fact, standard form and tableau form for linear

programs that have all less-than-or-equal-to constraints
and nonnegative right-hand-side values are the same.

To summarize, the following three steps are necessary

to prepare a linear programming problem for solution
using the simplex method:

Step 1. Formulate the problem

Tableau Form Step 2. Set up the standard form by adding slack and/or
subtracting surplus variables
A basic feasible solution to the system of m linear
constraint equations and n variables is required as a Step 3. Set up the tableau form
starting point for the simplex method. The purpose of
tableau form is to provide an initial basic feasible Setting Up the Initial Simplex Tableau
solution. Recall that for the HighTech problem, the After a linear programming problem has been converted
standard-form representation is to tableau form, we have an initial basic feasible solution
that can be used to begin the simplex method.

To provide a convenient means for performing the

calculations required by the simplex method, we will first
develop what is referred to as the initial simplex
tableau

Part of the initial simplex tableau is a table containing all

the coefficients shown in the tableau form of a linear
program. If we adopt the general notation

c j = objective function coefficient for variable j

We can identify two properties that make it possible to
find an initial basic feasible solution. The first property b j = right-hand-side value for constraint j
requires that the following conditions be satisfied:

a. For each constraint equation, the coefficient of one a ij = coefficient associated with variable j in constraint i
of the m basic variables in that equation must be 1,
Portion of the initial simplex tableau
and the coefficients for all the remaining basic
variables in that equation must be 0
b. The coefficient for each basic variable must be 1 in
only one constraint equation

When these conditions are satisfied, exactly one basic

variable with a coefficient of 1 is associated with each
constraint equation, and for each of the m constraint
equations, it is a different basic variable. Thus, if the n –
m nonbasic variables are set equal to zero, the values of
 Second, a row of the tableau is associated with each
basic variable. This row has a 1 in the unit column
corresponding to the basic variable. The value of each
basic variable is then given by the b i value in the row
Thus, for the HighTech problem, we obtain the following associated with the basic variable.
partial initial simplex tableau

In the example, row 1 is associated with basic variable

s1 because this row has a 1 in the unit column
corresponding to s1. Thus, the value of s1 is given by the
right-hand-side value b 1 : s1=b 1=150. In similar
Let: fashion, s2=b2=20 , and s3=b3=300.

c row = row of objective function coefficients

b column = column of right-hand-side values of the

constraint equations

A matric = m rows and n columns of coefficients of the

variables in the constraint equations

Using this notation, we can show these portions of the

initial simplex tableau:
To move from an initial basic feasible solution to a better
basic feasible solution, the simplex method must
generate a new basic feasible solution that yields as
better value for the objective function.

To do so requires changing the set of basic variables:

we select one of the current nonbasic variables to be
To help us recall that each of the columns contains the made basic and one of the current basic variables to be
coefficients for one of the variables, we write the variable made nonbasic.
associated with each column directly above the column.
By adding the variables, we obtain For computational convenience, we will add two new
columns to the complex tableau. One column is labeled
“Basis” and the other column is labeled “C B ”. In the
Basis column, we list the current variables, and in the C B
column, we list the corresponding objective function
coefficient for each of the basic variables. For the
HighTech problem:
This portion of the initial simplex tableau contains the
tableau-form representation of the problem; thus, it is
easy to identify the initial basic feasible solution. First,
we note that for each basic variable, a corresponding
column has a 1 in the only nonzero position. Such
columns are known as unit columns or unit vectors.
Note that in the column labeled Basis, s1 is listed as the Recall that the values in the c B column of the simplex
first basic variable because its value is given by the tableau are the objective function coefficients for the
right-hand-side value for the first equation. With s2 listed current basic variables.
second and s3 listed third, the Basis column and right- Hence, to compute the values in the z j row (the
hand-side values show the initial basic feasible solution
decrease in value of the objective function when x j is
has s1=150 , s2=20 , and s3=300.
increased by one), we form the sum of the products
obtained by multiplying the elements in the C B column
by the corresponding elements in the j th column of the A
matrix. Doing these calculations, we obtain

Can we improve the value of the objective function by

moving to a new basic feasible solution?

To find out whether it is possible, we add two rows to the

Because the objective function coefficient of x 1 is
bottom of the tableau. The first row, labeled z j , represent
c 1=50 , the value of c 1−z 1 is 50−0=50 . Then the net
the decrease in the value of the objective function that
will result if one unit of the variable corresponding to the result of bringing one unit of x 1 into the current basis will
j th column of the A matrix is brought into the basis. The be an increase in profit of $50. Hence, in the net
second row, labeled c j −z j , represents the net change evaluation row corresponding to x 1, we eneter 50. In the
in the value of the objective function if one unit of the same manner, we can calculate the c j −z j values for the
variable corresponding to the j th column of the A matrix remaining variables. The result is the following initial
simplex tableau:
is brought into the solution. We refer to the c j−z j row as
the net evaluation row.

Let us first see how the entries in the z j row are

computed. Suppose that we consider increasing the
value of the nonbasic variable x 1 by one unit – that is,
from x 1=0 to x 1=1

In order to make this change and at the same time

continue to satisfy the constraint equations, the values of
some of the other variables will have to be changed. The initial simplex tableau is now complete. The initial
As we will show, the simplex method requires that the basic feasible solution ( x 1=0 , x 2=0 , s 1=150 , s 2=20 ,
necessary changes be made to basic variables only. and s3=300) has an objective function value, or profit, of
$0. Also, the c j−z j or net evaluation row has values
For example, in the first constraint, we have
that will guide us in improving the solution by moving to
3 x 1+5 x 2 +1 s1 =150 a better basic feasible solution.

The current basic variable in this constraint equation is Improving the Solution
s1. Assuming that x 2 remains a nonbasic variable with a We must next determine which of the current basic
value of 0, if x 1 is increased in value by 1, then s1 must variables to make nonbasic.
be decreased by 3 for the constraint to be satisfied.
Notice that each of the coefficients in the x 1 column
Similarly, if we were to increase the value of x 1 by 1 (and indicated the amount of decrease in the corresponding
keep x 2=0 ¿, we can see from the second and third basic variable that would result from increasing x 1 by
equations that although s2 would not decrease, s3 would one unit. Considering the first row, we see that every unit
of the Deskpro produced will use 3 hours of assembly
decrease by 8.
time, reducing s1 by 3. In the current solution, s1 = 150
and x 1 = 0. Thus – considering this row only - the
maximum possible value of x 1 can be calculated by
solving

3 x 1=150

x 1=50

Finally, with 8 as the coefficient of x 1 in the third row,

every unit that we increase x 1 will cause a decrese of
eight units in s3. Because the value of s3 is currently
We see that c 1−z 1=50 is the largest positive value in
300, we can solve
the c j−z j row. Hence, x 1 is selected to become the new
8 x 1=300 bi
basic variable. Checking the ratios for values of a i1
To find the maximum possible increase in x 1 before s3 ai 1
will become nonbasic at a value of zero; thus, we see b3 300
greater than zero, we see that = =37.5 is the
300 a31 8
that x 1 cannot be any larger than =37.5
8 minimum of these ratios.

Considering the three rows (constraints) simultaneously,

we see that row 3 is the most restrictive. That is,
producing 37.5 units of the Deskpro will force the
corresponding slack variable to become nonbasic at a
value of s3=0

In making the decision to produce as many Deskpro

units as possible, we must change the set of variables in
the basic feasible solution, which means obtaining a new
basis. The simplex method moves from one basic
feasible solution to another by selecting a nonbasic
variable to replace one of the current basic variables.

This process of moving from one basic feasible solution

to another is called an iteration.

Criterion for Entering a New Variable into the Basis

Look at the net evaluation row (c j−z j ), and select the Thus, the current basic variable associated with row 3 (
variable to enter the basis that will cause the largest per- z 3) is the variable selected to leave the basis. In the
unit improvement in the value of the objective function.
tableau we have circled a 31=8 to indicate that the
In the case of a tie, follow the convention of selecting the
variable to enter the basis that corresponds to the left variable corresponding to the first column is to enter the
most of the columns. basis and that the basic variable corresponding to the
third row is to leave the basis.
To illustrate the computations involved, we add an extra
bi
column to the right of the tableau showing the ratios.
aij
Adapting the usual linear programming terminology, we Step 2. Add slack variables to each constraint to obtain
refer to this circled element as the pivot element. The standard form. This also provides the tableau form
column and the row containing the pivot element are necessary to identify an initial basic feasible solution for
called the pivot column and the pivot row, problems involving all less-than-or-equal-to constraints
respectively. with nonnegative right-hand-side values.

To improve the current solution of ( Step 3. Set up the initial simplex tableau
x 1=0 , x 2=0 , s 1=150 , s 2=20 , and s3=300), we should
Step 4. Choose the nonbasic variable with the largest
increase x 1 to 37.5. the production of 37.5 units of the
entry in the net evaluation row to bring into the basis.
Deskpro results in a profit of 50(37.5)=1875 . In This variable identifies the pivot column: the column
producing 37.5 units of the Deskpro, s3 will be reduced associated with the incoming variable.
to zero.
Step 5. Choose as the pivot row that row with the
Hence, x 1 will become the new basic variables, bi
smallest ratio of for a ij >0 where j is the pivot
replacing s3 in the previous basis. aij
column. This pivot row is the row of the variable leaving
Calculating the Next Tableau the basis when variable j enters.

We now want to update the simplex tableau in such a Step 6. Perform the necessary elementary row
fashion that the column associated with the new basic operations to convert the column for the incoming
variable is a unit column; in this way its value will be variable to a unit column with a 1 in the pivot row.
given by the righthand-side value of the corresponding
row. a. Divide each element of the pivot row by the pivot
element (the element in the pivot row and pivot
We would like the column in the new tableau column)
corresponding to x 1 to look just like the column b. Obtain zeroes in all other positions of the pivot
column by adding or subtracting an appropriate
corresponding to s3 in the original tableau, so our goal is
multiple of the new pivot row. Once the row
to make the column in A matrix corresponding to x 1 operations have been completed, the value of
appear as the new basic feasible solution can be read from
the b column of the tableau.
0
0
1 Step 7. Test for optimality. If c j −z j ≤0 for all columns,
the solution is optimal. If not, return to step 4.
The way in which we transform the simplex tableau so
that it still represents an equivalent system of constraint
equations is to use the following elementary row
operations.

Elementary Row Operations

1. Multiply any row (equation) by a nonzero number

2. Replace any row (equation) by the result of adding
or subtracting a multiple of another row (equation) to
it

Summary of the Simplex Method

Let us now summarize the steps followed to solve a

linear program using the simplex method. We assume
that the problem has all less-than-or-equal-to constraints
and involves maximization

Step 1. Formulate a linear programming model of the

problem