The simplex method is a general mathematical solution technique for 5.

5. Assign values for the model variables in the top row and the basic
solving linear programming problems. In the simplex method, the feasible solution variables on the left side.
model is put into the form of a table, add then a number of
mathematical steps are performed on the table. These mathematical 6. Insert the model constraint coefficients into the body of the table
steps in effect replicate the process in graphical analysis of moving
from one extreme point on the solution boundary to another.
However, unlike the graphical method, in which we could simply  The row values are computed by multiplying the column
search through all the solution points to find the best one, the sim values by the variable column values and summing.
plex method moves from one better solution to another until the best  The simplex method works by moving from one solution
one is found, and then it stops. (extreme) point to an adjacent point until it locates the best
solution.
A basic feasible solution satisfies the model constraints and has the  The variable with the largest positive value is the entering
same number of variables with nonnegative values as there are variable.
constraints  The pivot column is the column corresponding to the
The basic feasible solution in the initial simplex tableau is the origin entering variable.
where all decision variables equal zero  The pivot row is the row corresponding to the leaving
variable.
Row operations are used to solve simultaneous equations where  The pivot number is the number at the intersection of the
equations are multiplied by constants and added or subtracted from pivot column and row.
each other  The various row values in the second tableau are
computed using several simplex formulas. First, the row,
The simplex method is a set of mathematical steps for solving a called the new tableau pivot row, is computed by dividing
linear programming problem carried out in a table called a simplex every value in the pivot row of the first (old) tableau by the
tableau. pivot number. The formula for these computations is
The initial simplex tableau always begins with the solution at the
origin, where and equal zero. Thus, the basic variables at the origin new tableau pivot row values =old tableau pivot row values
are the slack variables, and since the quantity values in the initial pivot number
solution always appear as the right-hand-side values of the constraint
equations, they can be read directly from the original constraint
equations. The top two rows and bottom two rows are standard for all
tableaus; however, the number middle rows is equivalent to the
number of constraints in the model. For example, this problem has
two constraints; therefore, it has two middle rows corresponding to
and (Recall that n variables minus m constraints equals the number
of variables in the problem with values of zero. This also means that
the number of basic variables with values other than zero will be
equal to m constraints.) Similarly, the three columns on the left side
of the tableau are standard, and the remaining columns are
equivalent to the number of variables. Since there are four variables
in this model, there are four columns on the right of the tableau,
corresponding to and The next step is to fill in the values, which are
the objective function coefficients, representing the contribution to
profit(or cost) for each variable or in the objective function.
This completes the process of filling in the initial simplex tableau. The
remaining values in the and rows, as well as subsequent tableau
values, are computed mathematically using simplex formulas.
The following list summarizes the steps of the simplex method (for a
maximization model) that have been presented so far:
1. First, transform all inequalities to equations by adding slack
variables.
2. Develop a simplex tableau with the number of columns equaling
the number of variables plus three, and the number of rows equaling A minimization problem requires a few changes in the normal simplex
the number of constraints plus four. process, which we will discuss in this section. In addition, several
3. Set up table headings that list the model decision variables and exceptions to the typical linear programming problem will be
slack variables. presented later in this module. These include problems with mixed
constraints and problems with more than one optimal solution, no
4. Insert the initial basic feasible solution, which are the slack feasible solution, or an unbounded solution; problems with a tie for
variables and their quantity values. the pivot column; problems with a tie for the pivot row; and problems
with constraints with negative quantity values. None of these kinds of Multiple optimal solutions are identified by cj- zj (or zj- cj) = 0 for a
problems require changes in the simplex method. They are basically nonbasic variable. To determine the alternate solution(s), enter the
unusual results in individual simplex tableaus that the reader should nonbasic variable(s) with a cj- zj value equal to zero. An infeasible
know how to interpret and work with problem is identified in the simplex procedure when an optimal
solution is achieved (i.e., when all cj- zj … 0 ) and one or more of the
Transforming a model into standard form by subtracting surplus basic variables are artificial. An unbounded problem is identified in
variables will not work in the simplex method. the simplex procedure when it is not possible to select a pivot row—
that is, when the values obtained by dividing the quantity values by
 An artificial variable allows for an initial basic feasible the corresponding pivot column values are negative or undefined
solution at the origin, but it has no real meaning. THE DUAL
 Artificial variables are assigned a large cost in the objective
function to eliminate them from the final solution. The original linear programming model is called the primal, and the
 The row is changed to in the simplex tableau for a alternative form is the dual. The dual solution variables provide the
minimization problem. zj-cj cj-zj Artificial variables are value of the resources, that is, shadow prices.
always included as part of the initial basic feasible solution
when they exist The specific relationships between the primal and the dual
demonstrated in this exam ple are as follows:
A mixed constraint problem includes a combination of and …, =, Ú
constraints. 1. The dual variables, y1, y2, and y3, correspond to the model
constraints in the primal. For every constraint in the primal there will
An artificial variable is added to an equality (=) constraint for standard be a variable in the dual. For example, in this case the primal has
form. three constraints; therefore, the dual has three decision variables.
An artificial variable in a maximization problem is given a large cost 2. The quantity values on the right-hand side of the primal inequality
contribution to drive it out of the problem. constraints are the objective function coefficients in the dual. The
constraint quantity values in the primal, 40, 216, and 240, form the
For a multiple optimal solution the (or zj- cj ) value for a nonbasic dual objective function: Z = 40y1 + 216y2 + 240y3.
variable in the final tableau equals zero.\
3. The model constraint coefficients in the primal are the decision
An alternate optimal solution is determined by selecting the non basic variable coefficients in the dual. For example, the labor constraint in
variable with as the entering variable. the primal has the coefficients 2 and 4. These values are the y1 2y1
An infeasible problem does not have a feasible solution space 4y1. variable coefficients in the model constraints of the dual: and

An infeasible problem has an artificial variable in the final simplex 4. The objective function coefficients in the primal, 160 and 200,
tableau. represent the model constraint requirements (quantity values on the
right-hand side of the constraint) in the dual.
In an unbounded problem the objective function can increase
indefinitely because the solution space is not closed. 5. Whereas the maximization primal model has < constraints, the
minimization dual model has > constraints.
A pivot row cannot be selected for an unbounded problem.
The values for slack variables are the marginal values of the
A tie for the pivot column is broken arbitrarily. constraint resources, i.e., shadow prices. If a resource is not
completely used, i.e., there is slack, its marginal value is zero.
Tie for the Pivot Row
The shadow price is the maximum amount that should be paid for
Degeneracy It is also possible to have a tie for the pivot row (i.e., two one additional unit of a resource. The total marginal value of the
rows may have identical lowest nonnegative values).Like a tie for a resources equals the optimal profit.
pivot column, a tie for a pivot row should be broken arbitrarily.
However, after the tie is broken, the basic variable that was the other The dual provides the decision maker with a basis for deciding how
choice for the leaving basic variable will have a quantity value of zero much to pay for more resources.
in the next tableau. This condition is commonly referred to as
degeneracy because theoretically it is possible for subsequent
simplex tableau solutions to degenerate so that the objective function CHAPTER 2 LINEAR PROGRAMMING
value never improves and optimality never results
One of the most frequent objectives of business firms is to gain the
Degeneracy occurs in a simplex problem when a problem continually most profit possible or, in other words, to maximize profit. The
loops back to the same solution or tableau. objective of individual organizational units within a firm (such as a
Standard form for simplex solution requires positive right hand-side production or packaging department) is often to minimize cost. When
values. A negative right-hand-side value is changed to a positive by a manager attempts to solve a general type of problem by seeking an
multiplying the constraint by , which changes the direction ofthe objective that is subject to restrictions, the management science
inequality. technique called linear programming is frequently used.

Summary of Simplex Irregularities  Decision variables are mathematical symbols that

represent levels of activity.
  The objective function is a linear relationship that reflects Alternate optimal solutions are at the endpoints of the constraint line
the objective of an operation. segment that the objective function parallels.
 A model constraint is a linear relationship that represents a
restriction on decision making. An infeasible problem has no feasible solution area; every possible
 Parameters are numerical values that are included in the solution point violates one or more constraints
objective functions and constraints. In an unbounded problem the objective function can increase
Nonnegativity constraints restrict the decision variables to zero or indefinitely without reaching a maximum value.
positive values. A linear programming problem requires a choice between alternative
A feasible solution does not violate any of the constraints. courses of action (i.e., a decision). The decision is represented in the
model by decision variables. A typical choice task for a business firm
The feasible solution area is an area on the graph that is bounded by is deciding how much of several different products to produce, as in
the constraint equations. the Beaver Creek Pottery Company example presented earlier in this
chapter. Identifying the choice task and defining the decision
An infeasible problem violates at least one of the constraints. variables is usually the first step in the formulation process because it
A common solution approach is to solve algebraically the set of is quite difficult to construct the objective function and constraints
mathematical relationships that form the model either manually or without first identifying the decision variables. The problem
using a computer program, thus determining the values for the encompasses an objective that the decision maker wants to achieve.
decision variables. However, because the relationships are linear, The two most frequently encountered objectives for a business are
some models and solutions can be illustrated graphically. The maximizing profit and minimizing cost. A third characteristic of a
graphical method is realistically limited to models with only two linear programming problem is that restrictions exist, making
decision variables, which can be represented on a graph of two unlimited achievement of the objective function impossible. In a
dimensions. Models with three decision variables can be graphed in business firm, these restrictions often take the form of limited
three dimensions, but the process is quite cumbersome, and models resources, such as labor or material; however, the sample models in
of four or more decision variables cannot be graphed at all this chapter exhibit a variety of problem restrictions. These
restrictions, as well as the objective, must be definable by
Although the graphical method is limited as a solution approach, it is mathematical functional relationships that are linear. Defining these
very useful at this point in our presentation of linear programming in relationships is typically the most difficult part of the formulation
that it gives a picture of how a solution is derived. Graphs can process.
provide a clearer understanding of how the computer and
mathematical solution approaches presented in subsequent chapters Properties of Linear Programming Models
work and, thus, a better understanding of the solutions In addition to encompassing only linear relationships, a linear
The optimal solution is the best feasible solution. programming model also has several other implicit properties, which
have been exhibited consistently throughout the examples in this
The optimal solution point is the last point the objective function chapter. The term linear not only means that the functions in the
touches as it leaves the feasible solution area. models are graphed as a straight line; it also means that the
relationships exhibit proportionality. In other words, the rate of
Extreme points are corner points on the boundary of the feasible change, or slope, of the function is constant; therefore, changes of a
solution area. given size in the value of a decision variable will result in exactly the
same relative changes in the functional value. Linear programming
Constraint equations are solved simultaneously at the optimal
also requires that the objective function terms and the constraint
extreme point to determine the variable solution values.
terms be additive. Also, the total resources used must equal the sum
The slope is computed as the “rise” over the “run.” of the resources used for each activity in a constraint (e.g., labor).

Sensitivity analysis is used to analyze changes in model parameters. Another property of linear programming models is that the solution
Multiple optimal solutions can occur when the objective function is values (of the decision variables) cannot be restricted to integer
parallel to a constraint line. values; the decision variables can take on any fractional value. Thus,
the variables are said to be continuous or divisible, as opposed to
SLACK VARIABLE integer or discrete. For example, although decision variables
representing bowls or mugs or airplanes or automobiles should
A slack variable is added to a < constraint to convert it to an equation
realistically have integer (whole number) solutions, the solution
(=). A slack variable represents unused resources.
methods for linear programming will not necessarily provide such
A slack variable contributes nothing to the objective function value. solutions. This is a property that will be discussed further as solution
methods are presented in subsequent chapters.
The final property of linear programming models is that the values of
In a minimization problem the boundary of the feasible solution area all the model parameters are assumed to be constant and known
closest to the origin contains the optimal extreme point with certainty. In real situations, however, model parameters are
frequently uncertain because they reflect the future as well as the
A surplus variable is subtracted from a > constraint to convert it to an present, and future conditions are rarely known with certainty.
equation (=). A surplus variable represents an excess above a
constraint requirement level.
To summarize, a linear programming model has the following general
properties: linearity, proportionality, additivity, divisibility, and
certainty. As various linear programming solution methods are
presented throughout this book, these properties will become more
obvious, and their impact on problem solution will be discussed in
greater detail