M2S1: Decision Theory o Decision Alternatives

 d1 = use building design A

Decision Analysis
 d2 = use building design B
- Problem Formulation
 d3 = use building design C
- Decision Making without Probabilities
- Decision Making with Probabilities
o States of Nature
- Risk Analysis and Sensitivity Analysis
 s1 = an average of 40 customers
- Decision Analysis with Sample Information
arriving per hour
- Computing Branch Probabilities
 s2 = an average of 60 customers o Influence Diagram
arriving per hour
 An influence diagram is a graphical
Problem Formulation  s3 = an average of 80 customers device showing the relationships
arriving per hour among the decisions, the chance
- The first step in the decision analysis process is
events, and the consequences.
problem formulation.
o Payoff Table  Squares or rectangles
- We begin with a verbal statement of the problem.
depict decision nodes.
- Then we identify:  The consequence resulting from a
 Circles or ovals depict
specific combination of a decision
o the decision alternatives chance nodes.
alternative and a state of nature is a
o the states of nature (uncertain future events) payoff.  Diamonds depict
 A table showing payoffs for all consequence nodes.
o the payoff (consequences) associated with
each specific combination of: combinations of decision  Lines or arcs connecting the
alternatives and states of nature is a nodes show the direction of
 decision alternative payoff table. influence.
 state of nature  Payoffs can be expressed in terms
of profit, cost, time, distance or any
other appropriate measure.
- Example: Burger Prince Restaurant
o Burger Prince Restaurant is considering
opening a new restaurant on Main Street. The
company has three different building designs
(A, B, and C), each with a different seating
capacity. Burger Prince estimates that the
average number of customers arriving per
hour will be 40, 60, or 80. o Decision Tree
o Payoff Table (Payoffs are Profit Per Week)
  A decision tree is a chronological o Three commonly used criteria for decision o Conservative (Maximin) Approach
representation of the decision making when probability information
problem.  The conservative approach would be
regarding the likelihood of the states of nature
used by a conservative decision
 A decision tree has two types of is unavailable are:
maker.
nodes:  the optimistic (maximax) approach
 For each decision the minimum
 round nodes correspond to  the conservative (maximin) payoff is listed.
chance events approach
 The decision corresponding to the
 square nodes correspond to  the minimax regret approach. maximum of these minimum
decisions payoffs is selected.
 Branches leaving a round node  If payoffs are in terms of costs, the
represent the different states of o Optimistic (Maximax) Approach
maximum costs will be determined
nature;  The optimistic approach would be for each decision and then the
 Branches leaving a square node used by an optimistic decision decision corresponding to the
represent the different decision maker. minimum of these maximum costs
alternatives. will be selected. (Hence, a minimax
 The decision with the overall largest
approach)
 At the end of a limb of the tree is payoff is chosen.
the payoff attained from the series  List the minimum payoff for each
 If the payoff table is in terms of
of branches making up the limb. decision. Choose the decision with
costs, the decision with the overall
the maximum of these minimum
lowest cost will be chosen (hence,
payoffs.
a mimin approach).
 The decision that has the largest
single value in the payoff table is
chosen

 Decision d1 – Maximin Decision

 Minimum d1 – Maximin Payoff

o Minimax Regret Approach

 d3 – Maximax Decision
Decision Making without Probabilities  d3 s3 – Maximax Payoff
- Criteria for Decision Making
 The minimax regret approach the minimum of these values. o We use the expected value approach to
requires the construction of a identify the best or recommended decision
regret table or an opportunity loss alternative.
table.
o The expected value of each decision
 This is done by calculating for each alternative is calculated
state of nature the difference
between each payoff and the largest o The decision alternative yielding the best
payoff for that state of nature. expected value is chosen.
 Then, using this regret table, the o The expected value of a decision alternative
maximum regret for each possible  Decision d3 – Minimax decision is the sum of the weighted payoffs for the
decision is listed. decision alternative.
 Maximum d3 – Maximax Regret
 The decision corresponding to the o The expected value (EV) of decision
minimum of the maximum regrets alternative di is defined as
is chosen. Decision Making with Probabilities
 First compute a regret table by - Assigning Probabilities
subtracting each payoff in a column
from the largest payoff in that column. o Once we have defined the decision
The resulting regret table is: alternatives and states of nature for the
chance events, we focus on determining
probabilities for the states of nature.
o The classical, relative frequency, or o where:
subjective method of assigning
probabilities may be used.  N = the number of states of nature
o Because only one of the N states of nature  P(sj ) = the probability of state of
can occur, the probabilities must satisfy nature sj
two conditions:  Vij = the payoff corresponding to
decision alternative di and state of
nature sj

 For each decision list the maximum - Expected Value Approach o Calculate the expected value (EV) for each
regret. Choose the decision with
decision.
 o Here d1, d2, d3 represent the decision - Expected value of perfect information is defined as o the expected value of a decision alternative,
alternatives of Designs A, B, and C. o EVPI = |EVwPI – EvwoPI| and
o And s1, s2, s3 represent the states of o the payoff that might actually occur
o Where
nature of 40, 60, and 80 customers per hour.
- The risk profile for a decision alternative shows the
 EVPI = expected value of perfect
o The decision alternative with the greatest EV possible payoffs for the decision alternative along with
information
is the optimal decision. their associated probabilities.
 EVwPI = expected value with perfect
information about the states of
- Risk Profile for Decision Alternative d3
o Decision Tree nature
 EVwoPI = expected value without
perfect information about the
states of nature

o EVPI Calculation
 Step 1: Determine the optimal
return corresponding to each state of
nature.
 Step 2: Compute the expected
value of these optimal returns.
 Step 3: Subtract the EV of the Sensitivity Analysis
optimal decision from the amount
determined in step (2) - Sensitivity analysis can be used to determine how
changes to the following inputs affect the
o Calculate the expected value for the optimum recommended decision alternative:
payoff for each state of nature and subtract
o probabilities for the states of nature
the EV of the optimal decision.
 EVPI= .4(10,000) +.2(18,000) o values of the payoffs
+.4(21,000) -14,000 = $2,000 - If a small change in the value of one of the inputs
causes a change in the recommended decision
alternative, extra effort and care should be taken in
estimating the input value.
o Choose the decision alternative with the - Resolving Using Different Values for the
largest EV: Design C Probabilities of the States of Nature
Risk Analysis
- Risk analysis helps the decision maker recognize the
Expected Value of Perfect Information difference between:
 - Decision Strategy
o A decision strategy is a sequence of
decisions and chance outcomes.
o The sequence of decisions chosen depends
on the yet to be determined outcomes of
chance events.
o The optimal decision strategy is based on the
Expected Value with Sample Information
(EVwSI).
Decision Analysis with Sample Information o The EVwSI is calculated by making a
- Knowledge of sample (survey) information can be backward pass through the decision tree.
used to revise the probability estimates for the o Influence Diagram
states of nature.
- Expected Value with Sample Information (EVwSI)
- Prior to obtaining this information, the probability
estimates for the states of nature are called prior o Step 1: Determine the optimal decision
probabilities. strategy and its expected returns for the
possible outcomes of the sample using the
- With knowledge of conditional probabilities for the
posterior probabilities for the states of nature.
outcomes or indicators of the sample or survey
information, these prior probabilities can be o Step 2: Compute the expected value of
revised by employing Bayes' Theorem. o Decision Tree (top half) these optimal returns.
- The outcomes of this analysis are called posterior
probabilities or branch probabilities for decision trees

- Example: Burger Prince

o Burger Prince must decide whether to
purchase a marketing survey from Stanton
Marketing for $1,000. The results of the
survey are "favorable” or "unfavorable". o Decision Tree (bottom half)
The branch probabilities corresponding to all
the chance nodes are listed on the next slide.

o Branch Probabilities
o EVwSI = .54($17,855) + .46($11,433) = o Where o Efficiency of Sample Information
$14,900.88
 EVSI = expected value of sample  The efficiency rating (E) of sample
information information is the ratio of EVSI to
EVPI expressed as a percent.
o Optimal Decision Strategy  EVwSI = expected value with sample
information about the states of
 If the outcome of the survey is nature
"favorable”, choose Design C.
 EVwoSI = expected value without
 If the outcome of the survey is sample information about the  The efficiency rating (E) of the
“unfavorable”, choose Design A. states of nature market survey for Burger Prince
Restaurant is:
o Risk Profile for Optimal Decision Strategy  EVSI Calculation
 Subtract the EVwoSI (the
value of the optimal decision
obtained without using the Computing Branch Probabilities
sample information) from the
EVwSI. - Bayes’ Theorem can be used to compute branch
probabilities for decision trees.
- For the computations we need to know:
 EVSI = $14,900.88 - $14,000
= $900.88 o the initial (prior) probabilities for the states
of nature,
 Conclusion o the conditional probabilities for the
 Because the EVSI ($900.88) outcomes or indicators of the sample
is less than the cost of the information, given each state of nature.
survey ($1000.00), the - A tabular approach is a convenient method for
survey should not be carrying out the computations.
purchased.

o The expected value of sample information

(EVSI) is the additional expected profit
possible through knowledge of the sample or
survey information. - Step 1
o EVSI = |EVwSI – EVwoSI| o For each state of nature, multiply the prior
 probability by its conditional probability
for the indicator. This gives the joint
probabilities for the states and indicator.
- Step 2
o Sum these joint probabilities over all states.
This gives the marginal probability for the
indicator.
- Step 3
o For each state, divide its joint probability
by the marginal probability for the
indicator. This gives the posterior probability
distribution.

- Example: Burger Prince Restaurant

o Recall that Burger Prince is considering
purchasing a marketing survey from Stanton
Marketing. The results of the survey are
"favorable“ or "unfavorable".

o The conditional probabilities are:

 P(favorable |40 customers per hour)
= .2
 P(favorable |60 customers per hour)
= .5
 P(favorable |80 customers per hour)
= .9

o Posterior Probability Distribution M2: Formative Questions

For a maximization problem, the conservative approach is often
referred to as the – Maximin Approach
 III. standard deviation I. a risk-averse decision maker will always prefer a to b but
Expected value is the sum of the weighted payoff possibilities at a may prefer c to a
- II
circular node in a decision tree – True
II. a risk neutral decision maker will always prefer c to a or b
The analysis of a computer decision situation by constructing a Which of the following methods of selecting a strategy is consistently III. A risk seeking decision-maker will always prefer c to a or
mathematical model of the situation and then performing a large with risk averting behavior? b
number of iterations in order to determine the probability distribution of I. if two strategies have the same expected profit select
outcomes is called - Simulation - I, II, & III
the one with the smaller standard deviation

The sequence of possible managerial decisions and their II. if two strategies have the same standard deviation Select A decision tree - Arranges decision alternatives and state of nature in
expected outcome under each set of circumstances can be the one with the smaller expected profit their natural chronological order
represented and analyzed by using - A decision tree III. Select the strategy with the larger coefficient variation

Circular nodes in a decision tree indicates that it would be incorrect - I A matrix that for each state of nature and strategy show the difference
to choose a path from the node - True between strategies payoff and the best strategy payoff is called - A
The coefficient of variation measures - The risk per unit of minimax regret matrix
A payoff - exist for each pair of decision alternative and state of expected payoff
nature
Which of the following is a way to deal with decision making under Situation in which a decision maker must choose between
Sample information with an efficiency rating of 100% is perfect uncertainty – simulation strategies that have more than one possible outcome when the
information - true probability of each outcome is unknown is referred to as - Uncertainty
For maximization problem, the conservative approach is often
The marginal utility of money diminishes for a decision maker who is referred to as - maximin approach Decision alternatives are structured so that several could occur
- A risk averter simultaneously - False
A decision maker has chosen .4 as a probability for which he cannot
States of nature___ choose between a certain loss of 10,000 and the lottery p (-25,000). Expected value is the sum of the weighted payoff possibilities at the
if the utility of -25,000 is zero and of 5000 is 1, then the utility of -10,000 circular node in a decision tree - True
I. can be selected by the decision maker is - 0.6
II. decision-maker can be described describe controllable
natural events such as floods or freezing temperatures Making a good decision
III. enumerate the decision maker I. implies that a desirable outcome will occur
- II II. requires a clear understanding of decision
alternatives, state of nature, and payoffs
A situation in which a decision-maker knows all the possible III. Requires probabilities for all states of nature
outcomes of a decision and also known as the probability M3S1: Linear Programming Graphical Method
associated with each outcome is referred to as - Risk - II
Linear Programming
The expected value of an alternative can never be negative - False
Which one of the following does measure risk? - A method of dealing with decision problems that can be
Strategy A has an expected value of 10 and a standard deviation of 3. expressed as constrained linear models.
I. coefficient of variation Strategy B has an expected value of 10 and a standard deviation of 5. - The primary objectives of all LP models are certainty of the
II. expected value strategy C has an expected value of 15 and a standard deviation of 10. parameters and linearity of the objective function and all
Which of the following is true? constraints.
 - A mathematical technique for finding the best uses of an o C = ai x + bi y (for minimization) lowest value (for minimization problem) of the
organization’s resources. objective function.
- It is initially referred as “programming in a linear structure.” - Optimal Solution
- Subject to a number of linear constraints of the form
- Was renamed as “linear programming” in 1948 as o is a combination of decision variable amounts that
suggested by Tjalling Koopmans. o ai x + bi y  c
yields the best possible value of the objective
- Note: LP = Linear Programming o ai x + bi y  c function and satisfies all the constraints.
o ai x + bi y = c - Feasible Region
George Bernard Dantzig
o is the set of combinations of values for the
- LP was developed by George Bernard Dantzig (1914- decision variables that satisfy the non-negativity
2005) in the 1940s. - Objective Function conditions and all the constraints simultaneously
- American mathematical scientist o is an expression, which shows the relationship that is the allowable decisions.
between the variables in the problem and the - Extreme point
- Professor Emeritus of Transportation Sciences and firm’s goal
Professor of Operations Research and of Computer o is the corner of the feasible region. It is the
Science at Stanford University location of the maximum and minimum point of
- Two Types of Constraints the feasible region.
Restriction of LP Graphical Method
o Structural Constraint (or explicit constraint) is a Fundamental Theorem of LP Problem
- LP graphical solution is limited in a two-dimensional set of
axes. limit on the availability of resources.

- Graphing software applications can be used in three o Non-negativity constraint (or implicit constraint) it - If a Linear Programming (LP) problem has optimal solution,
variables corresponds to planes in a coordinate space restricts all the variables to zero and positive there is always at least one extreme point (corner point)
(three-dimensional). solution. solution of the feasible region.
Linear Programing Model
- A Linear programming (LP) problem with bounded,
- Maximize: P = 1,200x + 1,600y  Objective Function nonempty feasible regions always contain optimal
- Subject to: solutions.

o Structural Constraints Steps in Solving LP Problem

 3x + 2y  18 - Graph the linear inequalities and determine the feasible
region.
 2x + 4y  20
- Determine the coordinates of the extreme points (corner
 x5 points)
Solving LP Problem - Substitute the coordinates of the extreme points to the
o Structural Constraints objective function and identify the highest (for maximization
- A LP problem in two unknowns x and y is one in which we
are to determine the maximum and minimum value of a  x  0, y  0 problem) or lowest (for minimization problem) result.
linear expression. - The highest or lowest result obtained in step 3 serves as
Optimal Solution of LP Model the optimal solution to the LP problem.
- Objective function:
- Optimal Value
o P = ai x + bi y (for maximization)
o is the highest (for maximization problem) or
Example 1: Maximization Problem
 - x  0, y  0
A local boutique produced two designs of gowns A and B and has the
following materials available: 3x + 2y  18
3x + 2y = 18
- 18 square meters of cotton, 20 square meters of silk, and 5
(0, 9) (6, 0)
square meters of wool.
Design A requires the following:
2x + 4y  20
- 3 square meters of cotton, 2 square meters of silk and 1 2x + 4y = 20
square meter of wool. (0, 5) (10, 0)
Design B requires the following:
x5
- 2 square meters of cotton, 4 square meters of silk. If x=5
Design A sells for ₧1,200 and
Design B for ₧1,600, how many of each garment should the boutique
produce to obtain the maximum amount of money?

Solution for Example 1

- Let x be the number of Design A gowns, and y be the
number of Design B gowns.
Plotting of Graphs

Materials Design A (x) Design B (y) Available

Cotton 3 2 18
Silk 2 4 20 Identifying the Extreme Points
Wool 1 0 5
Profit ₧1,200 ₧1,600

Solution for Example 1

Maximize:
- P = 1,200x + 1,600y  Objective Function

Subject to:
- Structural Constraints
o 3x + 2y  18  Cotton Determining the Feasible Region
o 2x + 4y  20  Silk
o x  5  Wool
Solution for 1st Equation and 2nd Equation
Non-negativity Constraints
 Constraint for Silk
2(3x + 2y = 18)  6x + 4y = 36 2x + 4y  20
3(2x + 4y = 20)  (–) 6x + 12y = 60 2(4) + 4(3)  20
0x – 8y = –24 8 + 12  20
– 8y = –24 20  20
y=3
Constraint for Wool
3x + 2(3) = 18 x5
3x + 6 = 18 45
3x = 18 – 6
- There is an excess of 1 square meter of wool
3x = 12
x=4

o The intersection of first equation and second

equation is (4, 3).

Solution for 1st Equation and 3rd Equation Optimal Solution for Example 1
3x + 2y = 18 Objective Function: P = 1,200x + 1,600y
x=5 Extreme points Values of the objective function
(0, 5) 1,200(0) + 1,600(5) = 0 + 8,000 = 8,000
Directly substitute x = 5 to the 3x + 2y = 18.
(5, 0) 1,200(5) + 1,600(0) = 6,000 + 0 = 6,000
3(5) + 2y = 18 (4, 3) 1,200(4) + 1,600(3) = 4,800 + 4,800 = 9,600
15 + 2y = 18 (5, 1.5) 1,200(5) + 1,600(1.5) = 6,000 + 2,400 = 8,400
2y = 18 – 15
2y = 3 Example 2: Minimization Problem
y = 1.5
Decision: A pharmacist produces a drug from two ingredients. Each ingredient
P = ₧9,600 contains the same three antibiotics in different proportions. Each
o The intersection of first equation and third x = 4 Design A gowns ingredient A produced results ₧80 in cost; each ingredient B results
equation is (5, 1.5). y = 3 Design B gowns ₧50 in cost. The production of the antibiotics is dependent on the
availability of limited resources. The resource requirements for the
Extreme Points Checking production are as follows.
- Solution: (x, y) = (4, 3) Antibiotic Resources Requirement Minimum
Ingredient A Ingredient B Requirement
Constraint for Cotton Antibiotic 1 3 units 1 unit 6
3x + 2y  18 Antibiotic 2 1 unit 1 unit 4
3(4) + 2(3)  18 Antibiotic 3 2 units 6 units 12
12 + 6  18
18  18 The company wants to determine the quantity of ingredient A and B that
must go in to drug in order to meet the antibiotics minimum
requirements at the minimum cost

Solution for Example 2

- Let x be the quantity of ingredient A, and y be the quantity
of ingredient B.

Materials Ingredient A (x) Ingredient B (y) Requirement

Antibiotic 1 3 1 6
Antibiotic 2 1 1 4
Antibiotic 3 2 6 12
Cost ₧80 ₧50

LP Model
Determining the Feasible Region
Minimize:
C = 80x + 50y

Subject to: Solution for 1st Equation and 2nd Equation

3x + y  6  Antibiotic 1
x + y  4  Antibiotic 2 1(3x + y = 6)  3x + y = 6
2x + 6y  12  Antibiotic 3 3( x + y = 4)  (–) 3x + 3y = 12
x  0, y  0 0x – 2y = –6
– 2y = –6
y=3
Solution for Example 2
3x + 3 = 6
3x + y  6 3x = 6 – 3
3x + y = 6 Identifying the Extreme Points 3x = 3
(0, 6) (2, 0) x=1
o The intersection of first equation and second
x+y4 equation is (1, 3).
x+y=4
(0, 4) (4, 0)

2x + 6y ≥ 12
2x + 6y = 12
(0, 2) (6,0)

Plotting of Graphs
Solution for 2nd Equation and 3rd Equation

2( x + y = 4)  2x + 2y = 8
1(2x + 6y = 12  (–) 2x + 6y = 12 (3, 1) 80(3) + 50(1) = 240 + 50 = 290
0x – 4y = –4 3x + 2y  18
– 4y = –4 Decision: 3x + 2y = 18
y=1 C = ₧230 (0, 9) (6, 0)
x = 1 unit of Ingredient A
x+1=4 y = 3 units of Ingredient B 2x + 4y = 20
x=4–1 (0, 5) (10, 0)
x=3
Checking 2y  8
Solution: (x, y) = (1, 3) 2y = 8
o The intersection of second equation and third
y=4
equation is (3, 1).
Constraint for Antibiotic 1
3x + y  6 Plotting of Graphs
3(1) + 3  6
Identifying the Extreme Points
3+36
66

Constraint for Antibiotic 2

x+y4
1+34
44

Constraint for Antibiotic 3

2x + 6y  12
2(1) + 6(3)  12
2 + 18  12
20  12

Restrictions on the LP Graphical Method

Maximization Involving Mixed Constraints

Maximize:
Optimal Solution for Example 2 P = 6x + 4y

Objective Function: Subject to:

C = 80x + 50y 3x + 2y  18
2x + 4y = 20
Extreme points Values of the objective function 2y  8
x  0, y  0
(0, 6) 80(0) + 50(6) = 0 + 300 = 300
(6, 0) 80(6) + 50(0) = 480 + 0 = 480
Determining the Coordinates
(1, 3) 80(1) + 50(3) = 80 + 150 = 230
 3x + 2y = 18
Plotting of Graphs (0, 9) (6, 0)
2(3x + 2y = 18)  6x + 4y = 36
3(2x + 4y = 20)  (–) 6x + 12y = 60 2x + 4y = 20
0x – 8y = –24 (0, 5) (10, 0)
– 8y = –24
y=3 2y  8
2y = 8
3x + 2(3) = 18 y=4
3x + 6 = 18
3x = 18 – 6 Plotting of Graphs
3x = 12
x=4
o The intersection of first equation and second
equation is (4, 3)

Optimal Solution
Feasible Solution Objective Function: P = 6x + 4y

Extreme points Values of the objective function

(4, 3) 6(4) + 4(3) = 24 + 12 = 36
(10, 0) 6(10) + 4(0) = 60 + 0 = 60

Decision:
P = 60
x = 10 units
y = 0 unit

Restrictions on the LP Graphical Method

Minimization Involving Mixed Constraints
Minimize:
C = 6x + 4y

Subject to:
3x + 2y  18
2x + 4y = 20
2y  8
x  0, y  0

Feasible Solution
Determining the Coordinates
3x + 2y  18
 (10, 0) 6(10) + 4(0) = 60 + 0 = 60

Decision:
C = 36
x = 4 units
y = 3 units

Special Cases in LP Graphical Method

- Multiple Optimal Solutions
- Infeasibility
- Redundancy
- Unbounded

Multiple Optimal Solutions

- An LP model that has a multiple optimal solution or more
than one optimal solution
Solution for 1st Equation and 2nd Equation
2(3x + 2y = 18)  6x + 4y = 36 - Example:
3(2x + 4y = 20)  (–) 6x + 12y = 60 o Maximize:
0x – 8y = –24
– 8y = –24 o P = 6x + 4y
y=3

3x + 2(3) = 18 o Subject to:

3x + 6 = 18
3x = 18 – 6 o 3x + 2y  18
3x = 12 o x + 2y  10
x=4
o x  0, y  0
o The intersection of first equation and second
equation is (4, 3).

Optimal Solution
Objective Function:
C = 6x + 4y

Extreme points Values of the objective function

(4, 3) 6(4) + 4(3) = 24 + 12 = 36
 Optimal Solution - (0, 9) (6, 0)
Objective Function:
P = 6x + 4y
- x≥7
(0, 9) 6(0) + 4(9) = 0 + 36 = 36 - x=7
(4, 3) 6(4) + 4(3) = 24 + 12 = 36
(0, 5) 6(0) + 4(5) = 0 + 20 = 20

Decision:
P = 36
x = 0 unit
y = 9 units

Alternative Solution:
P = 36
Solution for 1st Equation and 2nd Equation x = 4 units
1(3x + 2y = 18)  3x + 2y = 18 y = 3 units
3( x + 2y = 10)  (–) 3x + 6y = 30

0x – 4y = –12 Infeasibility
– 4y = –12
y=3 - A case where an LP model contains no feasible
solution even though all constraints are being
satisfied; that is, there no are points which
o The intersection of first equation and second
satisfy all constraints.
equation is (4, 3).
- Example:
Extreme Points Maximize:
P = 2x + y

Subject to:
3x + 2y  18
x7
x  0, y  0

- Determining the coordinates:

- 3x + 2y  18
- 3x + 2y = 18 Redundancy
 - A condition of an LP model wherein there is a constraint
which does not affect the feasible region.
- Example:
o Maximize:
o P = 2x + y

o Subject to:
o x+y9
o x+y6
o x  0, y  0

Determining the coordinates:

x+y9
x+y=9 Unbounded
(0, 9) (9, 0) - A condition of an LP model when the objective function of a
linear programming problem can be made infinitely large
x+y6 without violating any of the constraints
x+y=6(
- Example
0, 6) (6, 0)
- Maximize:
Plotting of Graphs o P = 2x + y
- Subject to:
o x3
o y6
o x  0, y  0
- Determining the coordinates:
o x≥3
o x=3
o y≥6
o y=6

Determining the Feasible Region

Plotting of Graphs
M3S2: Linear Programming Graphical Method
 - Proceed to step 4.
Simplex Method where b1, b2, b3 …, bn and c are numbers.
- is an iterative technique that begins with a feasible solution
that is not optimal, but serves as a starting point. Example:
Standard LP Maximization Model A local boutique produced two designs of gowns A and B and has the
Iteration following materials available:
Maximize:
- is a simplex method which consists of the sequence of P = a1 x1 + a2 x2 + a3 x3 +…+ an x - 18 square meters of cotton, 20 square meters of silk, and 5
steps (row operations) performed in moving one basic square meters of wool.
feasible solution to another. Subject to: Design A requires the following:
structural constraints - 3 square meters of cotton, 2 square meters of silk and 1
b1 x1 + b2 x2 + b3 x3 +…+ bn xn  c square meter of wool.
Design B requires the following:
Non-negativity constraints
x1  0, x2  0, x3  0, …, xn  0 - 2 square meters of cotton, 4 square meters of silk.
If Design A sells for ₧1,200 and Design B for ₧1,600, how many of
where c  0 and the structural constraints is strictly “” each garment should the boutique produce to obtain the maximum
amount of money?
Minimize:
C = a1 x1 + a2 x2 + a3 x3 +…+ an xn
Let x1 be the number of Design A gowns, and x2 be the number of
Subject to: Design B gowns.
structural constraints Materials Design A (x1) Design B (x2) Available
b1 x1 + b2 x2 + b3 x3 +…+ bn xn  c Cotton 3 2 18
Silk 2 4 20
non-negativity constraints Wool 1 0 5
x1  0, x2  0, x3  0, …, xn  0 Profit ₧1,200 ₧1,600

where c  0 and the structural constraints is strictly “”. Maximize:

P = 1,200x1 + 1,600x2  Objective Function

Subject to:
A LP problem in n unknowns x1 , x2 , x3 ,…,xn is one which we are to Steps of Solving LP Maximization Model Structural Constraints
determine the maximum or the minimum value of the objective function 3x1 + 2x2  18  Cotton
- Set up problem in an LP model. 2x1 + 4x2  20  Silk
a1 x1 + a2 x2 + a3 x3 +…+ an xn , - Introduce the necessary slack variables. x1  5  Wool x1  0, x2  0
- Establish the initial tableau. x1  0, x2  0  Non-negativity Constraints
where a1 , a2 , a3 ,…,an are constants, subject to a linear constraints of
the form - Examine the simplex tableau for optimal solutions. If the Slack variables
basic feasible solution is maximal, the problem is solved. - are variables added to constraints to convert them into
Otherwise, proceed to step 5. equations
b1 x1 + b2 x2 + b3 x3 +…+ bn xn  c - Compute a new simplex tableau: select the pivot column,
b1 x1 + b2 x2 + b3 x3 +…+ bn xn  c find pivot row, and pivot about the pivot entry. Maximize:
b1 x1 + b2 x2 + b3 x3 +…+ bn xn = c –1,200x1 – 1,600x2 + P = 0
 S1 2 0 1 -½ 0 0 8
Subject to: X2 ½ 1 0 ¼ 0 0 5
3x1 + 2x2 + S1 = 18 S3 1 0 0 0 1 0 5
2x1 + 4x2 + S2 = 20 P -400 0 0 400 0 1 8,000
x1 + S3 = 5
x1, x2, S1, S2, S3  0 BV X1 X2 S1 S2 S3 P RHS
x1 1 0 ½ -¼ 0 0 4
Right-Hand-Side (RHS) X2 0 1 -¼ 3/8 0 0 3
- is the column in a simplex tableau indicating the quantities S3 0 0 -½ ¼ 1 0 1
of the variables is in a solution. P 0 0 200 300 0 1 9,600
BV X1 X2 S1 S2 S3 P RHS Decision:
S1 3 2 1 0 0 0 18 x1 = 4 Design A gowns S1 = 0
Basic Variables (BV) x2 = 3 Design B gowns S2 = 0
S2 2 4 0 1 0 0 20
- are the variable included in a basic solution. S3 1 0 0 0 1 0 5 P = ₧9,600 profit S3 = 1
BV X1 X2 S1 S2 S3 P RHS P 1,200 1,600 0 0 0 1 0
S1 3 2 1 0 0 0 18 Tableau 1
S2 2 4 0 1 0 0 20 Replacing Row = Pivot Row  Pivot BV X1 X2 S1 S2 S3 P RHS
S3 1 0 0 0 1 0 5 S1 3 2 1 0 0 0 18
P 1,200 1,600 0 0 0 1 0 R2 2 = R2 1  P2 1 S2 2 4 0 1 0 0 20
Simplex Tableau = (2, 4, 0, 1, 0, 0, 20)  4 S3 1 0 0 0 1 0 5
= (½, 1, 0, ¼, 0, 0, 5) P -1,200 -1,600 0 0 0 1 0
- is a table use to keep track of the calculations made when BV X1 X2 S1 S2 S3 P RHS
the simplex method is employed. S1 Tableau 2
X2 BV X1 X2 S1 S2 S3 P RHS
Pivot column
S3 ½ 1 0 ¼ 0 0 5 S1 2 0 1 -½ 0 0 8
- is the column in any solution to a maximization problem P X2 ½ 1 0 ¼ 0 0 5
which has the lowest negative value in the last row. S3 1 0 0 0 1 0 5
Remaining Row = Previous Row – (Intersectional Element x Replacing P -400 0 0 400 0 1 8,000
Intersectional Elements Row)
- are elements common to both the pivot column and the Tableau 3
rows representing variables in the solution R1 2 = R1 1 – 2R2 2 = (3, 2, 1, 0, 0, 0, 18) – (2)(½, 1, 0, ¼, 0, 0, 5) BV X1 X2 S1 S2 S3 P RHS
Pivot = (3, 2, 1, 0, 0, 0, 18) – (1, 2, 0, ½, 0, 0, 10) x1 1 0 ½ -¼ 0 0 4
= (2, 0, 1, –½, 0, 0, 8) X2 0 1 -¼ 3/8 0 0 3
- is the element of the simplex tableau that is in both the S3 0 0 -½ ¼ 1 0 1
pivot row and the pivot column. R3 2 = R3 1 – 0R2 2 = (1, 0, 0, 0, 1, 0, 5) – (0)(½, 1, 0, ¼, 0, 0, 5) P 0 0 200 300 0 1 9,600
= (1, 0, 0, 0, 1, 0, 5) – (0, 0, 0, 0, 0, 0, 0)
= (1, 0, 0, 0, 1, 0, 5)

R4 2 = R4 1 + 1600R2 2
= (-1200, -1600, 0, 0, 0, 1, 0) + (1600)(½, 1, 0, ¼, 0, 0, 5)
= (-1200, -1600, 0, 0, 0, 1, 0) + (800, 1600, 0, 400, 0, 0, 8000) A pharmacist produces a drug from two ingredients. Each ingredient
= (-400, 0, 0, 400, 0, 1, 8000) contains the same three antibiotics in different proportions. Each
BV X1 X2 S1 S2 S3 P RHS ingredient A produced results ₧80 in cost; each ingredient B results
₧50 in cost. The production of the antibiotics is dependent on the 3x1 + x2  6
availability of limited resources. The resource requirements for the x1 + x2  4 Decision:
production are as follows. 2x1 + 6x2  12 x1 = 1 unit of ingredient 1
x1  0, x2  0 S1 = 0
Antibiotic Ingredient A Ingredient B Minimum x2 = 3 units of ingredient 2
Requirement Maximize:
Antibiotic 1 3 units 1 unit 6 P = –80x1 – 50x2 S2 = 0
Antibiotic 2 1 unit 1 unit 4 C = ₧230 cost
Antibiotic 3 2 units 6 units 12 Subject to: S3 = 8
3x1 + x2 – S1 = 6
The company wants to determine the quantity of ingredient A and B that x1 + x2 – S2 = 4 Maximize:
must go in to drug in order to meet the antibiotics minimum 2x1 + 6x2 – S3 = 12 P = 4x1 + 3x2
requirements at the minimum cost. –4x1 – 3x2 + P = 0
x1, x2, S1, S2, S3  0
Subject to:
- Let x1 be the quantity of ingredient A, and x2 be the 2x1 + x2  16  2x1 + x2 + S1 = 16
quantity of ingredient B. BV X1 X2 S1 S2 S3 P RHS Test Ratio 2x1 + 3x2  36  2x1 + 3x2 + S2 = 36
Materials Ingredient A Ingredient B Requirement S1 3 1 -1 0 0 0 6 6÷3=2 4x1 + 5x2  28  4x1 + 5x2 – S3 = 28
(X1) (X2) S2 2 1 0 -1 0 0 4 4÷1=4 x1  0, x2  0
Antibiotic 1 3 1 6 S3 1 6 0 0 -1 0 12 12 ÷ 2 = 6
Antibiotic 2 1 1 4 Maximize:
P 80 50 0 0 0 1 0
Antibiotic 3 2 6 12 P = 10x1 + 5x2
Cost Pst 80 Pst 50
BV X1 X2 S1 S2 S3 P RHS Test Ratio
x1 0 -1/3 -1 0 0 0 2 2÷=6 Subject to:
S2 0 1/3 0 -1 0 0 2 2÷=3 2x1 + x2  10
Minimize: C = 80x1 + 50x2 x2 = 4
x2 0 2/3 0 0 -1 0 8 8 ÷ 5 = 1.5
x1 + 4x2  20
Subject to: P 0 26 2/3 0 0 0 1 -160
x1  0, x2  0
3x1 + x2  6  Antibiotic 1
x1 + x2  4  Antibiotic 2 BV X1 X2 S1 S2 S3 P RHS Test Ratio Maximize:
2x1 + 6x2  12  Antibiotic 3 x1 1 -3/8 0 0 ⅟16 0 1 ½ 1 ½ ÷ ⅟16 = P = 10x1 + 5x2
x1  0, x2  0 24
S2 0 ¼ -1 -1 ⅛ 0 1 1÷⅛=8
x2 0 1/8 0 0 -³⁄16 0 1 ½ 1 ½ ÷ (-³⁄16) Subject to:
Minimizing C is the same as maximizing = -8 2x1 + x2  10
P 0 23 3/4 0 0 4⅜ 1 -195 x2  4
P = –C. x2  4
x1 + 4x2  20
C = 80x1 + 50x2 is now replaced by P = –80x1 – 50x2 . x1  0, x2  0
New LP model BV X1 X2 S1 S2 S3 P RHS
X1 1 0 -½ ½ 0 0 1 Maximize:
Maximize: S2 0 0 2 -8 1 0 8 Subject to:
P = –80x1 – 50x2 2x1 + x2 – S1 = 10
X2 0 1 ½ -1 ½ 0 0 3
x2 – S2 = 4
Subject to: P 0 0 15 35 0 1 -230
x2 + S3 = 4 S2 2 4 0 1 0 0 20 x1  7  x1 – S2 = 7
x1 + 4x2 + S4 = 20 S3 1 0 0 0 1 0 5 x1  0, x2  0  –2x1 – x2 + P = 0
–10x1 – 5x2 + P = 0 P -100 200 0 0 0 1 0
Tableau 1
Minimize: Tableau 2 BV X1 X2 S1 S2 P RHS
C = 9x1 + 3x2 BV X1 X2 S1 S2 S3 P RHS S1 3 2 1 0 0 18
9x1 + 3x2 + P = 0 S1 2 0 1 -½ 0 0 8 S2 1 0 0 -1 0 7
X2 ½ 1 0 ¼ 0 0 5 P -2 -1 0 0 1 0
Subject to: S3 1 0 0 0 1 0 5
x1 + 2x2  12  –x1 – 2x2 – S1 = –12 P 0 0 0 50 0 1 1,000 Tableau 2
x1 + x2  8  x1 + x2 – S2 = 8 BV X1 X2 S1 S2 P RHS
Decision:
2x1 + 3x2  6  2x1 + 3x2 – S3 = 6 S1 1 2/3 1/3 0 0 6
x1 = 0
x1  0, x2  0 S2 0 -2/3 -1/3 -1 0 1
x2=5
P = 1,000 P 0 1/3 2/3 0 1 12
S1=8
Special Cases in LP Model S2=0 Since there is an existence of starred variable in the final solution it
- Multiple Optimal Solution S3=5 makes the solution meaningless, this is not a real solution.
- Infeasibility Tableau 3
- Unbounded Solution BV X1 X2 S1 S2 S3 P RHS
X1 1 0 ½ -¼ 0 0 4
- Degeneracy (Tie for the Pivot Row) X2 0 1 -¼ 3/8 0 0 3
- Tie for the Optimum Column S3 0 0 -½ ¼ 1 0 1
P 0 0 0 50 0 1 1,000
If an LP model that has a multiple optimal solution or more than one Decision:
optimal solution. x1 = 4
x2 = 3
It arises if there is a zero entry in the final tableau where the variable P = 1,000 - This is a condition of an LP model when the objective
column is located, this indicates an alternative solution. S1=0 function of a linear programming problem can be made
S2=0 infinitely large without violating any of the constraints, it
Example: S3=1 occurs in maximization problems.
Maximize:
P = 100x1 + 200x2 Infeasibility Example:
It is a case where an LP model contains no feasible solution even Maximize:
Subject to: though all constraints are being satisfied; that is, there are no points P = 2x1 + x
3x1 + 2x2  18  3x1 + 2x2 + S1 = 18 which satisfy all constraints.
2x1 + 4x2  20  2x1 + 4x2 + S2 = 20 Subject to:
x1  5  x1 + S3 = 5 Example: x1  3  x1 – S1 = 3
x1  0, x2  0  –100x1 – 200x2 + P = 0 x2  6  x2 – S2 = 6
Maximize: x1  0, x2  0  –2x1 – x2 + P = 0
Tableau 1 P = 2x1 + x2
BV X1 X2 S1 S2 S3 P RHS Tableau 1
S1 3 2 1 0 0 0 18 Subject to: BV X1 X2 S1 S2 P RHS Test Ratio
3x1 + 2x2  18  3x1 + 2x2 + S1 = 18 S1 1 0 -1 0 0 3 3÷1=3
S2 0 1 0 -1 0 6 6÷0= BV X1 X2 S1 S2 S3 P RHS Test Ratio X1 1 0 0 -2 1 0 2 2 ÷ (–2) = –2
P -2 -1 0 0 1 0 S1 3 0 1 -2 0 0 6 6÷3=2 P 0 0 0 -1 2 1 13
X2 0 1 0 1 0 0 3 3÷0=
Tableau 2 S3 1 0 0 -2 1 0 2 2÷1=2 Tableau 4
BV X1 X2 S1 S2 P RHS Test Ratio P -2 0 0 3 0 1 9 BV X1 X2 S1 S2 S3 P RHS
x1 1 0 -1 0 0 3 3÷0= S2 0 0 ¼ 1 -3/4 0 0
S2 0 1 0 -1 0 6 6÷1=6 Tableau 3 X2 0 1 -¼ 0 3/4 0 3
P 0 -1 -2 0 1 6 BV X1 X2 S1 S2 S3 P RHS X1 1 0 ½ 0 -½ 0 2
X1 1 0 1/3 -2/3 0 0 2 P 0 0 ¼ 0 1¼ 1 13
Tableau 3 X2 0 1 0 1 0 0 3
BV X1 X2 S1 S2 P RHS Test Ratio S3 0 0 -1/3 -1 1/3 1 0 0
S1 1 0 -1 -1 0 3 3 ÷ (-1) =
-3 P 0 0 2/3 1 2/3 0 1 13
S2 0 1 0 -1 0 6 6÷0=
P 0 0 -2 -1 1 12
Decision: Decision:
x1 = 2 x1 = 2
x2 = 3 x2 = 3
P = 13 P = 13
S1 = 0
S2 = 0 S1 = 0
S3 = 0 S2 = 0
S3 = 0
Tableau 1
BV X1 X2 S1 S2 S3 P RHS Test Ratio This is a condition where the greatest positive or least negative entries
Degeneracy (Tie in Pivot Row) S1 3 2 -1 0 0 0 12 12 ÷ 2 = 6 in last row have the same values; thus, there is a tie in the pivot column.
S2 0 1 0 1 0 0 3 3÷1=3
- This is a condition resulting from a tie in the test ratios S3 1 2 0 0 1 0 8 8÷1=4 One of two tied columns should be selected arbitrarily.
determining the replaced row, which produces a basic P -2 -3 0 0 0 1 0
variable with a zero value. Although, one choice may require fewer iterations than the other but
- It may develop when a problem contains redundant Tableau 2 there is also no way of knowing this beforehand.
constraints, that is, one or more of the constraints in the BV X1 X2 S1 S2 S3 P RHS Test Ratio
formulation make it unnecessary. S1 3 0 1 -2 0 0 6 6÷3=2
X2 0 1 0 1 0 0 3 3÷0= M3: Formative Questions
S3 1 0 0 -2 1 0 2 2÷1=2 Department 3 has 2500 hours. Transfers are allowed to departments 2
Tableau 1
and 4, and from departments 1
BV X1 X2 S1 S2 S3 P RHS Test Ratio P -2 0 0 3 0 1 9
and 2. If Ai measures the labor hours allocated to department i and Tij
S1 3 2 -1 0 0 0 12 12 ÷ 2 = 6 the hours transferred from
S2 0 1 0 1 0 0 3 3÷1=3 Tableau 3 department i to department j, then - A3 − T13 − T23 + T32 + T34 =
S3 1 2 0 0 1 0 8 8÷1=4 BV X1 X2 S1 S2 S3 P RH Test Ratio 2500
P -2 -3 0 0 0 1 0 S
S1 0 0 1 4 -3 0 0 0÷4=0 The set of combinations of values for the decision variables that
Tableau 2 X2 0 1 0 1 0 0 3 3÷1=3 satisfy the non-negativity conditions and all the constraints
simultaneously that is the allowable decision is called the – Feasible If an LP problem is not correctly formulated, the computer software Department 3 has 2500 hours. Transfers are allowed to departments 2
region will indicate it is infeasible when and 4, and from departments 1 and 2. If Ai measures the labor hours
trying to solve it. - False allocated to department i and Tij the hours transferred from department i
There is no optimal solution if the feasible region is – unbounded to department j, then - A3 + T13 + T23 - T32 - T34 = 2500
Location of maximum and minimum points of the feasible region is
The objective of the product design and market share optimization called the – extreme points Let A, B, and C be the amounts invested in companies A, B, and C.
problem presented in the textbook is to choose the levels of each If no more than 50% of the total investment can be in company B then: -
product attribute that will maximize the number of sampled customers It is improper to combine manufacturing costs and overtime costs in -0.5A + 0.5B -0.5C <= 0
preferring the brand in question. – True the same objective function. – False
Let Pij = the production of product i in period j. To specify that
The constraint x1 − x2 = 0 implies that if project 1 is selected, project 2 Where is the location of the maximum and minimum points of the production of product 1 in period 3 and in period 4 differs by no more
cannot be. – False feasible region – corner than 100 units, - P13 −P14 ≤100; P14 −P13 ≤100

Linear Programming is a method of dealing with decision problems Slack and surplus variables are not useful in integer linear programs
that can be expressed as _____ linear models – Constrained – False https://quizlet.com/240846010/management-101-homework-flash-cards/
https://quizlet.com/82257829/ch-4-bus104-flash-cards/
What term refers to a Linear Programming model that contains no point Linear programming graphical method can be used in three https://quizlet.com/243296626/chapter-13-flash-cards/
which satisfy all constraints – Infeasibility variables that correspond to planes in a coordinate space and can be https://quizlet.com/85768952/bus104-ch-7-flash-cards/
graphical. What can you say about this – difficult but possible https://quizlet.com/205046643/chapter-13-qmb-flash-cards/
In general, rounding large values of decision variables to the
nearest integer value causes fewer problems than rounding small In a model involving fixed costs, the 0 - 1 variable guarantees that
values. – True the capacity is not available unless the cost has been incurred. – True

If Pij = the production of product i in period j, then to indicate that the The Theory of Linear Programming states that the optimal solution
limit on production of the will lie ____ of the feasible region – at the boundaries
company's three products in period 2 is 400, - P12 + P22 + P32 ≤ 400
The product design and market share optimization problem
Media selection problems can maximize exposure quality and use presented in the textbook is formulated as 0-1 integer linear
number of customers reached as a programming model. – True
constraint, or maximize the number of customers reached and use
exposure quality as a constraint. – True Let M be the number of units to make and B be the number of units to
buy. If it costs$2 to make a unit and $3 to buy a unit and 4000 units are
A linear Programming problem that has no optimal solution has needed, the objective function is - Min 2M+3B
always at least _______ solution of the feasible regions – two extreme
points Media selection problems can maximize exposure quality and use
number of customers reached as a constraint, or maximize the number
Which of the following shows the relationship between the variables of customers reached and use exposure quality as a constraint. – True
in the Linear Programming problem and the firm’s goal – Objective
function The solution to the LP Relaxation of a minimization problem will
always be less than equal to the value of the integer program
What is NOT TRUE about linear programming – none of the above minimization problem. – False

Revenue management methodology was originally developed for the

banking industry – False