Module 1 - Unit 2: Models,

Algebra and Functions

Confidence Not Confident

Last Edited @October 31, 2024 12:28 PM

Introduction to Analytical Models in Supply

Chain Management
In this lesson, we'll dive into the basics of analysis, including models, algebra, and
mathematical functions. You might wonder why math is essential. The reason is
straightforward: decision-making is central to Supply Chain Management.
Throughout this course, you'll learn to address questions such as:

How many facilities should I open, and where?

What transportation mode should I use (truck, air, rail)?

How should I balance service versus cost?

Where should I source my raw materials?

How should I share risk with customers and suppliers?

How much inventory should I maintain?

How often should I replenish inventory?

What is the expected demand for next year?

How can I make my supply chain more resilient to disruptions?

These decisions are supported by analytical models. This course will focus on
heavy-duty analytical decision-making. While there is a people side of supply

Module 1 - Unit 2: Models, Algebra and Functions 1

 chain management, this course centers on mathematical models and their
applications.
Types of Models

Models can be classified based on their purpose and realism. On one side, we
have:

1. Operational Exercises: Practice scenarios that simulate real-world situations

but aren't the actual event (e.g., practice games, dress rehearsals).

2. Gaming: Scenarios where human responses are tested against a simulated

environment, replicating real-world interactions.

On the other side, we have:

1. Simulations: Mathematical models that simulate real-world scenarios without

human involvement, focusing on abstract representations.

2. Analytical Models: These are mathematical models used to capture essential

aspects of reality for decision-making. They are quicker and cheaper to solve
but more abstract.

We will primarily focus on analytical models for their efficiency and cost-
effectiveness. In later courses, we will explore operational exercises and gaming.

Module 1 - Unit 2: Models, Algebra and Functions 2

 Classification of Analytical Models

Analytical models can be categorized based on:

1. Evaluation vs. Generation:

Evaluation Models: Assess the performance of existing strategies.

Generation Models: Provide recommendations for new strategies.

2. Certainty vs. Uncertainty:

Certainty: Models dealing with known variables (e.g., linear programming,

network models).

Uncertainty: Models dealing with randomness and variability (e.g., Monte

Carlo simulation, queuing theory).

Model Types: Descriptive, Predictive, and Prescriptive

Module 1 - Unit 2: Models, Algebra and Functions 3

 1. Descriptive Models: Describe what has happened (e.g., queueing theory,
simulation).

2. Predictive Models: Forecast what could happen (e.g., regression, time series
analysis).

3. Prescriptive Models: Recommend actions or decisions (e.g., optimization

techniques).

Different models serve different roles. For instance, simulation can help evaluate
current solutions but does not provide recommendations. Understanding how
models align with your goals is crucial for effective decision-making.

Simulation will never reccomend or predict for example

Course Structure

1. Deterministic Models: Focus on known variables and prescriptive modeling.

Topics include:

Basic calculus and unconstrained optimization.

Module 1 - Unit 2: Models, Algebra and Functions 4

 Math programming and optimization techniques.

2. Stochastic and Uncertainty Models: Focus on predictive and descriptive

analysis. Topics include:

Basic probability and distributions.

Statistical analysis, regression modeling, and simulation.

Approach to Modeling

Our goal is not to make you an expert in mathematical proofs or the fine details of
each model. Instead, we aim to equip you with the knowledge to effectively use
and apply these models. Think of models as different types of cars: you don’t
need to be a mechanic, just a skilled driver. You should understand how to use the
models, what inputs are needed, and what outputs to expect.

Mathematical Functions

Module 1 - Unit 2: Models, Algebra and Functions 5

 Mathematical functions are fundamental elements in any type of analysis. A
function is essentially a relation between an input and an output, where each input
corresponds to exactly one output.

Function Notation
We write functions as: y = f(x) , where:
• y is the output
• f stands for function
• x is the input

We pronounce this as "f of x" or "y is a function of x".

Example: Quadratic Function

Let's consider a function f(x) = x²

Input (x) Output (y = f(x))

2 4

3.4 11.56

-2 4

Note: While each input gives exactly one output, the same output can come from
different inputs (e.g., both 2 and -2 give an output of 4).

Linear Functions
One of the most common types of functions is the linear function. In a linear
function, the dependent variable y changes linearly with the independent variable
x.

y changes linearly with x

The general form of a linear function is: y = ax + b , where:
• a is the slope (variable value)
• b is the y-intercept (constant or fixed value)

Notation Convention
• Constants are typically denoted by letters from the start of the alphabet (a, b, c,
etc.)

Module 1 - Unit 2: Models, Algebra and Functions 6

 • Variables are usually represented by letters from the end of the alphabet (x, y, z)

Graphical Representation
When plotted, a linear function forms a straight line:
• The y-intercept (b) is where the line crosses the y-axis (when x = 0)
• The slope (a) represents how much y increases for each unit increase in x

Applications in Supply Chain Management

1. Truckload Transportation Costs

Cost = $200 + $1.35 × distance

2. Warehousing Costs

Module 1 - Unit 2: Models, Algebra and Functions 7

 Cost = 2,500 + 2.5 × number_of_cases

3. Profit Function
Profit = (revenue - cost) × volume + fixed_cost

Linear functions are widely used in supply chain management due to their
simplicity and ability to represent many real-world scenarios accurately.

Quadratic Functions
Another widely used functional form is the quadratic function, also known as a
parabola or a polynomial function of degree 2. It has the general form:

y = ax² + bx + c

Where:

a, b, and c are constants

a ≠ 0 (if a = 0, it becomes a linear function)

x is the independent variable

y is the dependent variable

The shape of a quadratic function depends on the value of a:

Module 1 - Unit 2: Models, Algebra and Functions 8

 If a > 0, the parabola opens upward (convex)

If a < 0, the parabola opens downward (concave)

The constants affect the graph as follows:

a: determines the steepness of the curve

b: influences the position of the vertex

c: represents the y-intercept (where the curve crosses the y-axis)

Finding Roots of Quadratic Functions

Roots are x-values where y = 0. A quadratic function can have:

Two distinct roots

One root (tangent to x-axis)

No real roots (does not cross x-axis)

Module 1 - Unit 2: Models, Algebra and Functions 9

 To find roots, use the quadratic formula:

x = [-b ± √(b² - 4ac)] / (2a)

Example: For y = 2x² - 6x + 4

a = 2, b = -6, c = 4

Roots: x = 1 and x = 2

Quadratic functions are important in supply chain management for modeling

various non-linear relationships, such as diminishing returns or economies of
scale.

Practical Applications of Quadratic

Functions
Quadratic functions have widespread applications in various fields:

Physics and Engineering: Used to model trajectories of objects like balls,

missiles, or rockets.

Civil Engineering: Applied in designing road curves.

Manufacturing and Supply Chain Management: Used for pricing and cost
analysis.

Module 1 - Unit 2: Models, Algebra and Functions 10

 Example 1: Manufacturing Cost and Demand Analysis
Let's consider a manufacturing scenario for a product called "I-widgets":

Cost Function (Linear)

Cost = $500,000 + $75 × number of units produced

Demand Function (Linear)

Unit sales = 20,000 - 80 × price

This demand function provides important insights:

At a price of $0 (free), the maximum demand is 20,000 units, representing the

market size.

To find the price at which demand becomes zero, we can set the equation to
0:
20,000 - 80 × price = 0
Solving this gives us the price at which no one would buy the product.

By combining these functions, we can derive a quadratic profit function:

Profit = Revenue - Cost

= (20,000p - 80p²) - (2,000,000 - 6,000p)

Module 1 - Unit 2: Models, Algebra and Functions 11

 = -80p² + 26,000p - 2,000,000

This quadratic function helps determine the optimal price for maximum profit.

Example 2: Parcel Trucking Pricing

In parcel trucking, the cost per order often follows a quadratic function based on
weight:

Cost = 33 + 0.067w - 0.00005w²

Where w is the weight in kilograms.

This quadratic function captures the tapering effect of cost as weight increases,
reflecting the diminishing marginal cost for heavier shipments.

Module 1 - Unit 2: Models, Algebra and Functions 12

 Key Takeaways
Quadratic functions in supply chain management can model:

Non-linear pricing strategies

Profit optimization

Cost structures with diminishing marginal effects

Understanding these applications helps in making informed decisions in various

supply chain scenarios.

Power Function
There are many different functional forms you might encounter, but we'll focus on
two that you'll certainly see in supply chain management: the power function and
the exponential function. Let's briefly discuss both.

Module 1 - Unit 2: Models, Algebra and Functions 13

 The power function takes the form:
y = a ∗ xb 
where aand bare constants.

The shape of this curve is dictated by the value of b. Below are examples of how
different values of baffect the curve:

= 1, the function is linear, increasing at a constant rate. This means for
If b
every one unit increase in x, yincreases by a constant amount.

If b > 1, the function grows at an increasing rate (e.g., a * x² or a * x³). The
value of y increases exponentially.

If 0 < b < 1, the function grows slower than a linear function, as shown in
examples like parcel shipping cost vs. weight.

If b < 0, the function decreases asymptotically toward zero, meaning it gets
closer and closer to zero but never reaches it. For example, if b = -1, the
function becomes y = a / x. This is commonly seen when allocating fixed
costs.

Module 1 - Unit 2: Models, Algebra and Functions 14

 For instance, if a truck costs $1,000 to ship and carries up to 20 pallets, the cost
per pallet decreases as the number of pallets increases (e.g., $1,000 divided by 2
pallets = $500 per pallet).
So you'll see this a lot when we look at allocating [Link]'ll see it when we look
at inventory, when we're trying to figure out the effect of having a larger ordersize.
So we'll do this very frequently. So the power function will be widely used. And
we'll see it in inventory. We'll see it when we look at the propensity of disruptions.
You'll see it in segmentation.

Exponential Function
The exponential function is written as:
y = a ∗ bx ,
where a and b are constants, and x is the variable.

In an example where a = 1and b = 2, as x increases (e.g., 1, 2, 3, 4), yincreases

at an exponential rate. If b is larger (e.g., b = 3), the growth rate is even faster.
This is where the term "exponential growth" comes from.
Exponential functions are used to model fast-growing systems in fields like
physics, biology, economics, finance, and computer science. For example,
compound interest is calculated using an exponential function. The formula is:
Future Value = Principal * (1 + interest rate)^number of periods.

Module 1 - Unit 2: Models, Algebra and Functions 15

 For instance, if you invest $100 at a 20% interest rate for five years, the future
value will be:

$100 * (1 + 0.2)^5 = $248.80.

Euler’s Number (e)

Sometimes you'll see the letter e in exponential functions. e is a constant, similar
to pi, with a value of approximately 2.71828. It's used in natural logarithms and
appears in various growth-related scenarios.
For example, if an interest rate is 100% and you're doubling an amount every time
period, checking the growth more frequently (e.g., monthly, weekly, daily) will
eventually approach a limit, which is e. This is the basis for the natural logarithm,
and while the concept may seem complex, just remember that e = 2.718.
In supply chain management, both the power and exponential functions are widely
used. You'll see them applied in inventory management, cost allocation, disruption
analysis, and segmentation. Understanding these functions will help you model
and analyze various scenarios effectively.

Module 1 - Unit 2: Models, Algebra and Functions 16

 Logarithms
We've all encountered logarithms at some point in the past, even if it’s been a
while. They're important because they help solve specific types of equations.
Let’s break them down briefly.

What is a Logarithm?
A logarithm is the inverse function of an exponential function. For example, if we
have an exponential function where y = bx , the logarithm reverses this, showing
us the value of x that makes the equation true.
In simple terms, logarithms tell us what power we need to raise the base b to in
order to get a certain value y. It’s just another way of solving for x.

Inverse Functions
We use inverse functions regularly, even if we don't realize it. For example:

If y = a ∗ x, you divide yby a to find x = y/a

If y = a + x, you subtract a from y to find x = y − a
These are simple examples of inverse functions. For multiplication, the inverse is
division, and for addition, the inverse is subtraction. Similarly, for exponentials, the
inverse is a logarithm.

Example of Using Logarithms

Suppose we have an equation like 100 = 10^x. We can figure out that x = 2
because 10 * 10 = 100. However, if it weren’t obvious, we could use a logarithm to
solve for x:
log_10(100) = 2
Another example is if 10^x = 5. To find x, we would use the logarithm:
log_10(5) ≈ 0.7

This shows how logarithms help us solve for x in equations involving exponentials.
When you see a logarithm written without a base, like log(5), it usually means the
base is 10.

Module 1 - Unit 2: Models, Algebra and Functions 17

 Natural Logarithms (ln)
Another common type of logarithm is the natural logarithm, denoted as "ln," which
uses the base e (Euler’s number, approximately 2.718). For example, if we have an
equation where 1 = e^x, we use the natural logarithm to solve for x:
ln(1) = 0
This tells us that e^0 = 1.

Doubling an Investment
Let’s say you invest a sum of money with an annual interest rate of 7%. The
question is: How many years, t, will it take for the investment to double?
From previous discussions, we know that the future value F of an investment is
given by the formula:
F = P ∗ (1 + r)n 
where:

P is the initial principal (amount invested),

Module 1 - Unit 2: Models, Algebra and Functions 18

 r is the interest rate (0.07 for 7%),

n is the number of periods (years in this case).

Since we want the future value to be twice the principal F = 2P , we can set up
the equation:

2P = P (1 + r)n 
Simplifying the Equation
We cancel out P from both sides:
2 = (1 + r)n 
Substitute r = 0.07:
2 = (1.07)n 
Solving for n
To solve for n, we take the natural logarithm (ln) of both sides:

ln(2) = nln(1.07)
Now, solve for n:
n = ln(2)/ln(1.07)
Using a calculator:
n is approximately 0.6931 divided by 0.0677, which is about 10.24
This means it will take approximately 10.24 years for the investment to double at a
7% interest rate.

Module 1 - Unit 2: Models, Algebra and Functions 19

 Multivariate Functions
So far, we've been working with functions that have one input and one output.
However, there’s no reason why functions can't have more than one input. When
they do, these are called multivariate functions, which is just a fancy way of
saying a function with more than one variable. The functions we’ve seen before
are known as univariate functions, meaning they have a single variable.
In a multivariate function, you have more than one independent variable. The
general form would be:
y = f(x1 , x2 , ..., xn )
​ ​ ​

Here, you see subscripts to distinguish between different input variables (like x₁,
x₂, etc.). The ...means you can have as many x's as needed, up to xₙ, where n is
the number of variables. Despite having multiple inputs, the function still gives
only one output. This means that for any given set of input values, there will
always be one output. However, different inputs can give the same output.

Module 1 - Unit 2: Models, Algebra and Functions 20

 Example of a Multivariate Function
Consider this function with two variables:
f(x1 , x2 ) = x1 + 2 ∗ x2 + 5 ∗ x1 ∗ x2 
​ ​ ​ ​ ​ ​

Let’s say x1 ​
= 2and x2 = 4. Plugging these into the function, we get:
​

f(2, 4) = 2 + 2 ∗ 4 + 5 ∗ 2 ∗ 4 = 2 + 8 + 40 = 50
Now, if we change the inputs to x1 ​ = −1 and x2 = 0, we get:
​

f(−1, 0) = −1 + 2 ∗ 0 + 5 ∗ (−1) ∗ 0 = −1

Interestingly, if we choose another set of values like x1 ​
= 0and x2 = −1/2, we
​

again get:
f(0, −1/2) = 0 + 2 ∗ (−1/2) + 5 ∗ 0 ∗ (−1/2) = −1
So, even though the inputs were different, the output was the same.

Practical Application
Multivariate functions are widely used in practice. Let’s revisit a logistics example.
Suppose the cost of shipping depends not only on the weight (w) of the package
but also on the distance (d). In this case, the cost might be modeled as a function
of both variables:

Module 1 - Unit 2: Models, Algebra and Functions 21

 Cost(w, d) = c1 + c2 ∗ w + c3 ∗ w 2 + c4 ∗ d + c5 ∗ d2 + c6 ∗ w ∗ d
​ ​ ​ ​ ​ ​

Here, c1 might represent a fixed cost, and the other terms account for how the
​

weight, distance, and their interaction affect the total cost. Notice the squared
terms and the cross product term (w * d). In this function, coefficients like
c3 , c5 , c6 might be negative, as costs could taper off for very large shipments or
​ ​ ​

distances.
Another example of a multivariate function is in total logistics cost, which might
depend on annual demand (D), order cost (A), and order size (Q). A function for
this could look like:
T otalCost = c ∗ D + A ∗ (D/Q)
Here, the total cost depends on the number of units demanded (D), the cost of
placing each order (A), and the size of each order (Q). This equation has three
inputs (D, A, and Q), making it multivariate.

Key Takeaway
Multivariate functions are very common and can take many inputs, though they
still produce only one output. The same principles you’ve seen with single-
variable functions apply here, just with more variables involved.

Module 1 - Unit 2: Models, Algebra and Functions 22

 Properties of Functions: Convexity and
Continuity
As we begin discussing optimization, it's essential to understand two key
properties of functions: convexity and continuity.

Convexity
What is convexity?

A function is convex if it "holds water." Here's a simpler way to visualize it:

imagine a function that looks like a "U" shape. If you pour water into it, the water
stays in the curve. This indicates that the function is convex.
Formal definition: If you pick any two points on the curve of a convex function
and draw a line between them (like a tightrope), the value of the function at any
point on this line will be greater than or equal to the value of the function at that
same point on the curve. This is the technical definition of convexity.

A simpler way to think about convexity is just to remember: does it hold

water?

Now, let's contrast this with a concave function, which is the opposite of convex.
If you pour water on a concave function (imagine an upside-down "U"), the water
would run off the sides.
Sometimes, functions are neither purely convex nor purely concave. For
example, part of a function may be convex, while another part is concave. Imagine
drawing a line at x = 1. To the right of this line, the function is convex, but to the
left, it becomes concave. The ability to identify these sections is crucial for
optimization purposes.

Module 1 - Unit 2: Models, Algebra and Functions 23

 Continuity
The second important property is continuity.
A function is continuous if you can draw its graph without lifting your pen from
the paper (or in this case, from the screen).
For example, if we consider the function mentioned earlier (the one that's neither
fully convex nor concave), we can draw it continuously without lifting our pen, so
it's a continuous function.

Here are examples of discontinuous functions:

1. The function y = 1/x

As x approaches 0, y becomes infinitely large, so there's a break in the graph.
You would have to lift your pen to continue drawing it on the other side of the
graph.

2. Step function
This function looks like stairs. For certain values of xxx, the function suddenly
jumps from one value of yyy to another. You would have to lift your pen to
move from one step to the next, making it discontinuous.

Module 1 - Unit 2: Models, Algebra and Functions 24

 Key Points from Lesson
All right, we covered a lot of material in this lesson. It might feel like going back to
elementary or high school math, but this foundation is crucial. These are the
building blocks for our analytical models, so it's essential to get comfortable with
the terminology and manipulating equations. I promise it will get easier from here
—the first step is always the hardest.

1. Different Models and Their Uses:

Descriptive models: Help us understand what has happened.

Predictive models: Help us understand what could happen.

Prescriptive models: Make a recommendation on what we should do (we

will focus on optimization here). These models differ from the others
because they generate a recommendation, whereas descriptive and
predictive models evaluate or critique an existing situation.

2. Functions and Their Importance:

We spent a lot of time on mathematical functions, which are essential in

our analysis.

Types of functions:

Linear functions: y = ax + b. Variables are often denoted as xyz , and

constants as abc

Quadratic functions: y = ax2 + bx + c. These look like a “U” shape,

either facing up or down.

Power functions: y = axb . Depending on the value of b, the function

can increase or decrease exponentially. If b=1, it becomes linear; if b is
negative, it behaves like an inverse function.

Exponential functions: y = a ∗ bx . We explored these extensively as

well.

3. Logarithms:

Module 1 - Unit 2: Models, Algebra and Functions 25

 Logarithms are the inverse function of exponentials. If y = bx , then
solving for x requires a logarithm: logb(y) = x.
We introduced the natural logarithm (ln), which is simply the logarithm to
base e (Euler’s number, 2.718). It’s a tool used to solve for exponentials.

4. Multivariate Functions:

These functions have more than one input. However, it’s important to note
that even with multiple inputs, there is only one output.

yis a function of x1, x2, … , xn,where multiple inputs lead to a single

output value.

5. Convexity and Continuity:

Convexity: A function is convex if it “holds water” (easy to visualize in

simple cases, but more complex in larger problems).

Continuity: A function is continuous if it can be drawn without lifting your

pencil, meaning it exists at all points.

These properties are important when solving larger problems, and getting
familiar with the terminology now will help later.

Module 1 - Unit 2: Models, Algebra and Functions 26