Mathematical
Modelling
Part 1: Optimization Model
Content of part 1
Chapter 1: One-variable optimization
Chapter 2: Multivariable optimization
Chapter 3: Multi-objective optimization
Chapter 4: Computational methods for
optimization
3
Outline
• Five-step method
• Sensitivity analysis
• Sensitivity and Robustness
Optimization problem
• Optimization problems are the most common
applications.
• Objectives is to maximize profit or to minimize
cost.
• Business managers attempt to manipulate values
of control variables to achieve desired objective.
• Examples:
• Farming
• Manufacturing
• Finance
• Education
Five-step method
Five-step method is a general procedure that
can be used to solve problems using
mathematical modeling.
1. Ask the question
2. Select the modelling approach
3. Formulate the model
4. Solve the model
5. Answer the question
An example
• A pig weighing 200 pounds
gains 5 pounds per day
and costs 45 cents a day to
keep. The market price for
pigs is 65 cents per pound
but is falling 1 cent per day.
When should the pig be
sold?
Step 1: Ask a question
• Question must be phrased in the mathematical
terms:
• Start with a set of assumptions or suppositions
about the way things really are.
• Problem components:
• List of variables (appropriate units)
• List of relations between variables (equations/
inequalities)
• List of objectives (mathematical terms)
• Note that
• Be careful not to confuse variables and parameters
• Check units to make sure that your assumptions makes
sense
Step 1: Ask a question
9
Ask a question in illustrated
example
Ask a question in illustrated
example
Step 2: Select the modelling
approach
• We have already a problem stated in mathematical
language. We need to select a mathematical approach
to use to get an answer to the research question.
• How to select approaches:
• Characteristics of variables
• Characteristics of relations
• Complexity of problem
=> Experiences, Skill and Familiarity with the relevant literature
• Illustrated example: One-variable optimization
Step 3: Formulate model
• We need to take the question exhibited in step 1 and
reformulate it in the standard form selected in step 2,
so that we can apply the standard general solution
procedure.
Step 4: Solve model
• Apply the standard solution procedure
identified in step 2 to solve model.
• Check again your model
• Use appropriate tools/technologies
Step 5: Answer the question
• Explain results from mathematical models in
nontechnical terms such that anyone who can
understand the question be able to understand
your answer.
• Illustrated example: Sell the pig after 8 days to
get a net profit $133.2.
• Dealing with real-life problem, there exists
uncertainty in assumptions/suppositions, so we
should investigate several alternatives. This is
called by sensitivity analysis.
Summary (1)
16
Summary (2)
17
Summary (3)
18
Summary (4)
19
Summary (5)
20
Outline
• Five-step method
• Sensitivity analysis
• Sensitivity and Robustness
Sensitivity analysis
• Five-step method starts with making
some assumptions about the
problem. However, we are rarely
certain enough abouts things to be
able expect all these assumptions.
• Therefore, we need to consider how
sensitive our conclusions are to
each of the assumption we have
made.
• This plays an important role in
mathematical model.
Assumption of the illustrated
example
• Assumptions:
• Current weight of the pig
• Current price for pigs
• Cost per day of keeping the pig
• Rate of growth of the pig
• Rate of falling price
• Uncertain levels of assumptions are different to
each other due to nature of data
Illustrated example: Sensitivity of
decrease price rate
• Original assumption, rate at which the price is
falling, r = 0.01 dollars per day. We try to
change this value in different scenarios.
Illustrated example: Sensitivity of
growth rate
• Original assumption, growth rate g is 5 lbs per
day. We try to change this value in different
scenarios.
Sensitivity analysis
• We should interpret sensitivity data in terms of
relative change or percent change, rather than
in absolute terms.
• Illustrated example:
• 10% decrease in falling price rate r leads to a 39%
increase in optimal value
• 10% decrease in growth rate leads to 34%
decrease in optimal value
Sensitivity analysis
If 𝑥 changes by an amount Δ𝑥, the relative change
in 𝑥 is given by Δ𝑥/𝑥. If 𝑟 changes by Δ𝑟, resulting in
the change Δ𝑥 in 𝑥, then the ratio between the
relative changes is Δ𝑥/𝑥 divided by Δ𝑟/𝑟. Let Δ𝑟 →
0 and using the definition of the derivative, we
obtain
We call this limiting quantity the sensitivity of 𝑥 to 𝑟,
and we will denote it by 𝑆 𝑥, 𝑟 .
Sensitivity analysis
𝑑𝑥 7
In the example we have = − 2 =
𝑑𝑟 25𝑟
− 2800. With 𝑟 = 0.01, 𝑥 = 8, thus
It means that if 𝑟 goes up by 2%, then 𝑥 goes
down by 7%
Sensitivity analysis
We have
And
So that a 1% increase in the growth rate of the pig would cause us to wait
about 3% longer to sell the pig
Sensitivity analysis
• Question: What are
• 𝑆 = 𝑥, 𝑔
• 𝑆 = 𝑦, 𝑔
Outline
• Five-step method
• Sensitivity analysis
• Sensitivity and Robustness
Sensitivity and Robustness
• A mathematical model is robust if the conclusions it leads
to remain true even through the model is not completely
accurate.
• In real problems, we will never have perfect information,
and even if it were possible to construct a perfectly
accurate model, we might be better off with a simpler and
more tractable approximation.
• For this reason, a consideration of robustness is a
necessary ingredient in any mathematical modelling
project.
Illustrated example
• In assumption, both the weight of the pig and the
selling price per pound are linear functions of time.
These are not exactly valid in real life.
• According this assumptions, if we keep the pig a
year, so
• Assumptions should be as much realistic as
possible!
Definitions
• f reaches a global maximum in 𝑥 ∗ if 𝑓 𝑥 ≤
𝑓 𝑥 ∗ , ∀𝑥 ∈ 𝑋
• f reaches a global minimum in 𝑥 ∗ if 𝑓 𝑥 ≥
𝑓 𝑥 ∗ , ∀𝑥 ∈ 𝑋
• f reaches a local maximum in 𝑥 ∗ if 𝑓 𝑥 ≤
𝑓 𝑥 ∗ , ∀𝑥 ∈ 𝑋 and 𝑥 is a neighbor of 𝑥 ∗
• f reaches a local minimum in 𝑥 ∗ if 𝑓 𝑥 ≥
𝑓 𝑥 ∗ , ∀𝑥 ∈ 𝑋 and 𝑥 is a neighbor of 𝑥 ∗
• Note : Inequalities may be strict, in that case
the optima are unique
34
Illustration
35
Optimality conditions
• Definitions :
• Stationary point : 𝑓′ 𝑥 = 0
• Critical point : 𝑓′ 𝑥 does not exist
• 𝑓′(𝑥 + ) ≠ 𝑓′(𝑥 − )
• Or → ∞
36
Optimality conditions
Theorem 1 : First order necessary condition for
optimality
if 𝑓: 𝑋 → ℝ , 𝑥 ∗ is a local optimum and 𝑓′ 𝑥 ∗ exists,
then 𝑓 ′ (𝑥 ∗ ) = 0
Theorem 2 : Second order sufficient condition for
optimiality
Let 𝑥 ∗ be a stationary point and 𝑓 two times
derivable :
• If 𝑓 ′′ (𝑥 ∗ ) > 0, then 𝑥 ∗ is a local minimum
• If 𝑓 ′′ (𝑥 ∗ ) < 0, then 𝑥 ∗ is a local maximum
37
Optimality conditions
Theorem 3 :
Let 𝑥 ∗ be a stationary point
• If 𝑓 𝑥 is convex, then 𝑥 ∗ is a global minimum
• If 𝑓 𝑥 is concave, then 𝑥 ∗ is a global maximum
Theorem 4 :
Let 𝑓 be two times derivable :
• 𝑓 𝑥 is convex ⟺ 𝑓 ′′ (𝑥 ∗ ) ≥ 0, ∀𝑥 ∈ 𝑋
• 𝑓 𝑥 is concave ⟺ 𝑓 ′′ (𝑥 ∗ ) ≤ 0, ∀𝑥 ∈ 𝑋
38
Exercise
39
