Linear Programming (Chapter 7)  Formulating a linear program involves

developing a mathematical model to represent

 Linear programming is a widely used the managerial problem
mathematical modeling technique designed to  Steps in formulating a linear program:
help managers in planning and decision making o Understand the managerial problem
relative to resource allocation. being faced
 Mathematical programming has little to do with o Identify the objective and constraints
computer programming
o Define the decision variables
 Programming refers to modeling and solving a
o Use the decision variables to write
problem mathematically
mathematical expressions for the
 All LP problems have several properties and
objective function and the constraints
assumptions in common.
 Product mix problems use LP to decide how
 Objective function is the property of
much of each product to make, given a series of
maximizing or minimizing some quantity, such
resource restrictions
as profit or cost
 Technically, in LP, we maximize total
 Objectives must be stated clearly and defined
contribution margin which is the difference
mathematically.
between unit selling price and variable costs
 It does not matter whether profits and costs are excluding depreciation, fixed general expense,
measured in cents, dollars, or millions of dollars and advertising.
 Constraints the restrictions that limit the degree
to which we can pursue our objective Graphical solution method
 Objectives and constraints in LP problems must
be expressed in terms of linear equations or  The easiest way to solve a small LP problem is
inequalities. with the graphical solution approach
 Linear mathematical relationships mean that all  The graphical procedure is useful only when
terms used in the objective function and there are two decision variables. When there are
constraints are of the first degree more than two variables, it is not possible to plot
 The term linear implies both proportionality and the solution on a 2D graph
additivity (all activities equals the sum of the  However, the graphical method is invaluable in
individual activities) providing us with insights into how other
 We assume that conditions of certainty exist: approaches work
that is, numbers in the objective and constraints  We are always working in the first quadrant of
are known with certainty and do not change the graph
during the period being studied  The shaded region represents the area of
 We also assume divisibility where solutions solutions that does not exceed either of the set
need not be in whole numbers (integers), constraints; feasible region
instead they are divisible and may take any  The feasible region must satisfy all conditions
fractional value (work in process) specified in the problem’s constraints
 We assume that all answers or variables are  An infeasible solution represents any point
nonnegative since negative values of physical outside the shaded area
quantities are impossible, as you cannot
produce negative number of items. Isoprofit Line Solution Method

To summarize the properties of linear programs  The optimal solution is the point lying in the
feasible region that produces the highest profit
1. One objective function  The speediest one to apply in solving for the
2. One or more constraints optimal solution is this method
3. Alternative courses of action  Started by letting profit equal an arbitrary but
4. Objective function and constraints are linear— small dollar amount and to find the intercepts of
proportionality and divisibility such solution
5. Certainty  Isoprofit involves graphing parallel profit lines
6. Divisibility  Slope can be obtained by arranging the
7. Nonnegative variables objective function from b=mx+y to y=mx+b
  Gradually increase the profit (DECREASE THE Four special cases in LP
COST) and find x1 and x2 by using the same
constants so that the lines are parallel to each 1) No feasible solution—when there is no solution
other to an LP problem that satisfies all of the
 The farther we move from the origin, the higher constraints given
the profit will be 2) Unboundedness—when the profit in a
maximization problem can be infinitely large and
Corner Point Solution Method is missing one or more constraints, then the
problem is unbounded
 This technique is simpler conceptually that the 3) Redundancy—a constraint that does not affect
isoprofit line approach, but it involves looking at the feasible solution region
the profit at every corner point of the feasible 4) Alternate optimal solutions—there are multiple
region optimal solutions wherein the objective function’s
 The mathematical theory behind LP states that iso-objective line runs perfectly parallel to one of
an optimal solution to any problem will lie at a the problem’s constraints
corner point, or extreme point, of the feasible
region Sensitivity analysis
 Elimination method is used to solve equations
 Optimal solutions have been found under
simultaneously
deterministic assumptions where we assume
Slack and surplus complete certainty in the data and relationships
of a problem
 Slack is the amount of a resource that is not  When an optimal solution is found, we recognize
used in a less-than-or-equal-to constraint the importance of seeing just how sensitive that
 Slack = right hand side – left hand side solution is to model assumptions and data
 Surplus is used with greater-than-or-equal-to  Changes in the following areas are tested to see
constraints to indicate the amount by which the its effect on the feasible solution found
right hand side of a constraint is exceeded o Contribution rates for each variable
 Surplus = actual amount – required o Technological coefficients (the numbers
 A constraint that has zero slack or surplus for in the constraint equations)
the optimal solution is called a binding o Available resources
constraint  Sensitivity analysis is the examination of
 A constraint with positive slack or surplus for the changes after the optimal solution has been
optimal solution is called a nonbinding reached
constraint  Two approaches to determining just how
sensitive an optimal solution is to changes:
Solving minimization problems o Trial and error
 Commonly for minimizing an objective such as o Analytic postoptimality method
cost  If a constraint is binding, obtaining additional
 For minimization problems, it is easier to units of that resource will usually result in higher
simultaneously solve for the equations by profits.
method of elimination between lines that  For a nonbinding constraint, obtaining additional
intersect units of that resource will not result in higher
 Minimization problems are often unbounded profits and will only increase slack
outward
Changes in the objective function coefficient
 Since the constraints are now greater-than-or-
equal-to, the feasible solution area will be above  Change in coefficient usual mean no change
the constraint lines in this example in the feasible region but the slope of the iso-
 The optimal solution is usually the extreme in the objective line will change
lower left  An extreme change in coefficient may also
 The lowest isocost line (the one closes to the cause a new corner point to become optimal due
origin) provides us with the optimal solution  We can increase or decrease the objective
corner function coefficient of any variable and the
 current corner point may remain optimal if the  The dual price for a constraint is the
change is not too large improvement in the objective function value that
 However, if we increase or decrease the results from a one unit increase in the right hand
coefficient by too much, the optimal solution side value of the constraint.
could change to a different corner point  The upper and lower bounds of a constraint are
 The current solution remains optimal unless an the amounts at which the dual price will be
objective function coefficient is increased to a nulled (the profit will stop increasing) when the
value above the upper bound, or decreased to a RHS of the constraint exceed such bounds
value below the lower bound  Dual prices will change if the amount of the RHS
 If the change equals to the upper or lower of the constraint goes above the upper bound or
bound, there would be two optimal solutions as below the lower bound
the objective function would be parallel to the  If the constraint has slack, increase in RHS
first constraint would not increase profit but simply increase the
 The upper and lower bounds relate to changing amount of slack.
only one coefficient at a time  A dual price of 0 will remain relevant as long as
the RHS does not go below the lower bound or
Changes in the technological coefficients’ above upper bound.
 A shadow price is the change in the objective
 Changes in the technological coefficients
function value that results from a one-unit
(coefficient constraints) affect the shape of
increase in the RHS of a constraint (simply, dual
the feasible solution region, and therefore,
price)
the optimal solution.
 Maximization
 These changes will have no effect on the
o Improvement in objective function value
objective function of an LP problem, but they can
is the same as a positive change
produce a significant change in the shape of the
o Dual and shadow price are exactly the
feasible solution region, and hence in the
optimal profit or cost same
 Minimization
Changes in the resources or right-hand side values o Improvement in the objective function is
the same as a negative change (a
 If the RHS of a constraint is changed, the decrease in cost)
feasible region will change and often the o The shadow price will be the negative of
optimal solution will change. the dual price
 If the constraint is nonbinding, increase in RHS
will increase profit

The Simplex Method (Module 7) How to Set Up the Initial Simplex Solution

 Most real life problems have more than two  The theory of LP states the optimal solution will
variables and are thus too large for the simple lie at a corner point of the feasible region. In
graphical solution procedure large LP problems, the feasible region cannot be
 The simplex method examines the corner points graphed because it has many dimensions, but
in an iterative manner, that is, repeating the the concept is the same.
same set of procedures time after time until an  The simplex method systematically examines
optimal solution is reached corner points, using algebraic steps, until an
 Each iteration brings a higher value for the optimal solution is found
objective function so that we are always moving  The first step of the simplex method requires
closer to the optimal solution that we convert each inequality constraint
 The simplex approach yields not only the optimal (except non-negativity constraints) in an LP
solution to the decision variables and maximum formulation into an equation
profit (minimum cost), but also valuable  ≤ Add slack variable (+s1)
economic information.  The slack variable represent unused resources
 Looking at a graphical solution first will help us in the department which concerns the certain
understand the steps in the simplex method constraint
  If nothing is produced, we have a slack of those  We subtract a surplus variable and add an
resources equal to the RHS of the constraint artificial variable to form an equality when
 Slack variables yield no profit dealing with a ≥ constraint
 ≥ Minus surplus variable (-s2) and add artificial  Artificial variables have no physical meaning and
variable (+A1) drop out of the solution mix before the final
 = Add artificial variable (+A1) tableau
 If an artificial variable has a positive (nonzero)
Finding an Initial Solution Algebraically value, then the original constraint where this
artificial variable was added has not been
 Set the real variables in the constraints to 0 to satisfied.
find the value of the slack/surplus variables
 Before the final simplex solution has been
 A basic feasible solution to a system is reached, all artificial variables must be gone
established wherein n equations is found by from the solution mix
setting all but n variables equal to 0 and solving
 Since artificial variables must be forced out of
for the other variables
the solution, we can assign a very high Cj cost to
 This first solution produces a profit of 0 as well each
as slack variables equal to the constant (RHS)
terms in the constraint equations
 The simplex method considers only feasible Maximization
solutions and hence will touch no possible
combinations other than the corner points of the  Pivot column (entering variable)—the most
shaded region. positive net profit
 The initial solution begins where the x1 and x2 are  Pivot row (leaving variable)—the row with the
0, or the (0,0) corner point smallest nonnegative calculated ratio
 Basic variables are the variables in the solution (quantity:pivot column equivalent) disregard 0
mix while those not in the solution mix are called and negative ratios
the nonbasic variables  When there are no more positive numbers in the
 For any corner point, exactly two of the four net profit row (only negative and zero), an
variables will equal zero optimum solution has been reached
 Substitution rates are numbers in the body of  -M
the table
Minimization
 For every unit of a variable introduced into the
current solution n1 units of s1 and n2 units of s2  Pivot column (entering variable)—the most
must be removed from the solution. n=constraint negative cost (Cj – zj)
coefficient for a specific variable
 Pivot row (leaving variable)—the row with the
 Or any variable ever to appear in the solution smallest nonnegative calculated ratio
mix column, it must have the number 1 (quantity:pivot column equivalent) disregard 0
someplace in its column and 0s in every other and negative ratios
place in that column.
 When there are no more negative numbers in
 The Z-row entry in the quantity column provides the cost row (only positive and zero), an
the gross profit optimum solution has been reached
 The Cj – zj gives the net profit from introducing  +M
one unit of each variable into the solution
 To identify which corner point the tableau is, look More than one optimal solution
at the basic variables. The quantity of the real
variable in the solution mix is the point for that  Alternate optimal solutions may exist if the Cj – zj
variable and if the variable is non-basic, the is equal to 0 for a variable not in the solution mix
point is 0.
 The net profit row indicates (1) whether the
current solution is optimal and if it is not, (2)
which variable should enter and leave the next
solution mix.

Surplus and Artificial Variables