Quantitative Methods

Linear Programming (LP)

Kaconco James
+256772653191
Chapter Objectives
 By end of chapter students shall be in position to:
 Overview of Linear Programming (LP).
 Formulate medium, two decision variable linear
programming problems.
 Maximization Problem
 Minimization Problem
 Solve two decision variable linear programming problems
using the graphical method.
 Identify and explain special cases; for two decision
variable linear programming problems.
Linear Programming
(LP)
 Linear Programming (LP) is a decision making
technique for assigning scarce resources among
competing factors.
 The objective may be cost minimization or profit
maximization or optimize some measure of interest.
 Linear programming is a technique that helps to improve
a system’s (individual, company, ministry, government,
etc) efficiency and productivity.
 Scarce Resources include machinery, labor, money, time,
warehouse space, raw materials, operating restrictions, etc.
 The resources may be used to produce products (machinery,
furniture, food, clothing, etc) or services (schedules,
production, ads policies, investment decisions, etc).
 Linear programming improves the quality of decisions
by amplifying the analytic abilities of a decision maker.
 LP application follows a scientific approach to solving
management problems in order to help managers make better
decisions.
Linear Programming
(LP)
LP involves transformation of management problem
statement into a mathematical model.
 LP models can be solved
 Manually, Calculator, Computer software
 LP applications are numerous; (military, financial,
marketing, accounting, agricultural, transportation,
production, staffing, portfolio selection, facilities
selection, refinery operation, government, health
care, education, etc.).
LP Model Components
1. Decision variables are mathematical symbols that
represent levels of activity.
Dependent Decision Variable (z) [profit or cost].
Independent Decision Variable(s) (xi / xij / xijk) [competing
factors or operating restrictions].
 For example, an electrical manufacturing firm
desires to produce x1 radios, and x2 toasters,
where x1, and x2 are symbols representing
unknown variable quantities of each item
(competing factor). The final values of x1, and x2
as determined by the firm, constitute a
decision (e.g., the equation x1 = 100 radios is a
decision by the firm to produce 100 radios).
 100 is the decision variable value of x1
 LP problem decision variables are single (no
multiplication of variables and only to power 1 or
0).
LP Model Components
2. The objective function is a linear mathematical
function (relationship) that describes the objective of
the firm in terms of the decision variables.
 The objective function always consists of
either maximizing (Max) or minimizing (Min) some
value (Z).
 Max / Min Z = c1x1 + c2x2
 The parameters (c1 and c2) represent either profit or
cost against a decision variable.
 The parameters or numbers c1 and c2 are called
objective function coefficients and their values are
either known (determined from existing data) or an
estimated number; the parameters are expected not
to change during the time of study.
LP Model Components
3. The model constraints are linear mathematical relationships
(function) in terms of the decision variables.
 They represent the restrictions placed on the firm by the
operating environment (limited resources or restrictive
guidelines).
 Constraint examples include =, ≥, ≤.
 a1x1 + a2x2 = b1
 a1x1 + a2x2 ≥ b1
 a1x1 + a2x2 ≤ b1
 LP problems constraints could be of one or mixed type.
 Constraint right hand side (rhs) b1 represents the resource
limitation and is always nonnegative.
 The parameters or numbers a1 and a2 are known or is an
estimate and are called constraint technological
coefficients; and should not change during the time of study.
 Operating restriction describe the environment of
execution.
 example (for every one crate of soda consumed at least or at
most two crates of beer will have been consumed)
LP Model Components
4. Sign restriction (nonnegativity) is associated with
each decision variable.
 For any decision variable xi, the sign restriction specifies
that xi must be nonnegative (xi ≥ 0) or that xi may be
unrestricted in sign (urs).
 Only the positive quadrant is needed.
Summary
 Characteristics
(Assumptions) of Linear
Programming Problems
1. Numbers or parameters (objective function
coefficients, constraint right hand side, and constraint
technological coefficients) are known with certainty
and do not change during the period being studied.
2. Proportionality exists in the objective function and
constraints. The consumptions and contributions for
each activity are proportional to the actual activity
level.
3. Additivity exists in the objective function and
constraints (the total of all activities equals the sum of
the individual activities). The total consumption of each
resource, as well as the overall objective function value
are the aggregates of the resource consumptions and
the contributions to the problem objective function,
resulting by carrying out each activity independently
Characteristics
(Assumptions) of Linear
Programming Problems
4. Divisibility assumption requires that LP decision
variables are allowed to take on fractional values.
x1 = 2 or x1 = 2.3 or x1 = 0.25
This is not accepted in some situations.
5. All answers or decision variables (basic and non
basic) are nonnegative.
xi ≥ 0; or xij ≥ 0; or xijk ≥ 0
Only positive quadrant values are needed
Guidelines for Model
Formulation
 Understand the management problem
thoroughly well.
 Identify and define the decision variables.
 Identify and construct the objective function
in terms of decision variables.
 Identify and construct all the constraints in
terms of decision variables.
 Set the sign restriction (nonnegativity) on all
decision variables.
LP Formulation for Two Decision
Variable Problem
 Data can be summarized in a table format.
 Operating restriction constraints are some times
difficult to include in the table.
 They are developed outside the table and
thereafter added to the constraint list.
LP Formulation:
Maximization Problem
 A firm manufactures two products types A and B
and sells them at a profit of Ugx 200 on product
type A and Ugx 300 on product type B. Each
product type is processed on two machines G
and H. Product type A requires 1 minute of
processing time on machine G and 2 minutes of
processing on machine H; product type B
requires 1 minute of processing on machine G
and 1 minute of processing on machine H.
Machine G is available for not more than 400
minutes while machine H is available for not
more than 600 minutes during any working day.
The combined demand for product types A and B
is at least 100 units in a single working day.
Formulate the problem as a linear programming
model.
LP Formulation:
Minimization Problems
 A company owns 2 oil mills A and B which have
different production capacities for low and high grade
oil. On a single day oil mill A produces 6 and 4 barrels
of each grade respectively and oil mill B produces 2,
and 12 barrels of each grade respectively. It costs the
company USD 1,000 and USD 800 per day to run the
mills A and B respectively. The company enters into a
contract to supply 12 and 24 barrels of each grade
respectively every week. The combined days oil mills
A and B can operate should not exceed 6 in a week.
Formulate a Linear Programming Model to minimize
production costs (determine number of days per week
each oil mill will be operated in order to meet the
contract economically).
Solving LP Problems:
Graphical Method
 Formulate the LP problem.
 Graph each constraint.
 Assume constraint is an equation.
 Determine two points on the equation.
 Join the two points with a straight line.
 Shade off unwanted region.
 The points lying in the wanted region will satisfy all the constraints
and sign restrictions.
 Repeat process for all constraints.
 The common region is called feasible region; the area that
satisfies all of the LP’s constraints and the sign restrictions
simultaneously.
 Identify and name all corner points of the feasible region.
 Determine the coordinates of all the corner (extreme) points
with their corresponding Z value.
 Select the corner (extreme) point with the optimum Z value
(largest for maximization and smallest for minimization);
the coordinates become the decision variable values.
Constraint other Terms
 Binding constraint: This is the one that the left-
hand side and the right-hand side are equal when
the optimal values of the decision variables are
substituted into the constraint. Resource has been
completely utilized.
 Nonbinding constraint: This is the one that the
left-hand side and the right-hand side are unequal
when the optimal values of the decision variables
are substituted into the constraint. Resource has not
been completely utilized.
 Constraints of type ≤, will have slack or surplus value.
 Constraints of type ≥, will have excess value.
 Redundant Constraint: This is the one that does
not affect the feasible region in any way, and can
therefore be removed or ignored.
Special Cases
 Alternative Optimal Solutions
Two corner points have the same z-value, thus alternate
optimal solutions. The objective function line is parallel to a
binding or boundary constraint in the direction of
optimization. All points on this line segment being optimal.
 Infeasibility
No feasible region. The LP problem is over constrained or
poorly formulated (structured) so that no point satisfies all
the constraints and sign restrictions. The LP problem is said
to be infeasible.
 Redundancy
One or more constraints do not affect the feasible region
and can easily be dropped. This is common with large LP
problems
 Unboundedness
Feasible region is open. A decision variable(s) and objective
function value can be increased indefinitely. Such cases
only occur with maximization problems.
LP Limitations
1. Linear Relationship (non linear
programming).
2. Constant Value of objective and Constraint
Equations parameters (sensitivity analysis).
3. Accepts Fractional Value Solutions (integer
programming).
4. Degree Complexity; LP fails with very
complex problems. Dynamic programming
takes over.
5. Multiplicity of Goals; LP deals with only one
objective. Goal programming takes over in
case of multiple objectives.
End of Chapter 1