07-02-2024

Session-2

Report the
Results Understand The Real World Problem

Mathematical Modelling
Interpretations
of the Math Mathematical Formulation
Solutions Mathematical Analysis
Algorithms

Solution of the Math Problem

A Big Picture: Key Components of Optimization Problem

1
 07-02-2024

Mathematical Optimization
Model

Five Components, namely:

• Sets and Indices
• Parameters
• Decision Variables
• Objective Function(s)
• Constraints

Introduction to Linear Programming Problems

• The most important and the most popular

• Try to optimize a linear objective function subject to linear

constraints and with non-negativity restrictions on the
variables.

2
 07-02-2024

Advantages and Disadvantages:

• Advantages:
• Relative Simplicity of LP models.
• Scalability of state-to-art algorithms (software can handle LP
models with millions of decision variables and constraints).

• Disadvantages:
• Most real-life problems are not linear.
• The implicit assumptions of LP models are rather restrictive.

Assumption of LP
• Proportionality Assumption:
The contribution of a variable is proportional to its value.
• Additivity Assumption:
Contributions of variables are independent.
• Divisibility Assumption:
Decision variables can take fractional values.
• Certainty Assumption:
Each parameter is known with certainty.

3
 07-02-2024

Abstract and Generalize Form of LP/MILP

A Production Planning Problem

• A manufacturing facility makes n products say P1, P2,
P3….Pn using m different resources (Such as raw material,
labor, machine, etc.), R1, R2…..RM.
• It is known that each unit of product Pj will sell for cj dollars.
• We have bi units of resources Ri available.
• Producing one unit of product Pj requires aij units of resource
Ri.
Decide the production quantities of each product to maximize
the total revenue.

4
 07-02-2024

• Indices:
• i will be used for i-th resource, i =1,……,m;
• j will be used for the j-th product, j = 1,……..,n.

• The decision variables:

• Xj = the number of units of product j to manufacture, j
= 1,……..,n.

Objective function coefficients

𝑛
Objective Decision Variables
𝑀𝑎𝑥 ෍ c𝑗 𝑥𝑗
Function
𝑗=1

RHS: Right Hand Side

𝑛
Less than or
෍ ai , j 𝑥𝑗 ≤ b i (𝑖 = 1 .. . . 𝑚) Set of indices for
equal
Constraints
Constraints j=1
Technology
coefficients
Non-Negativity Set of indices for
𝑥𝑗 ≥ 0 (𝑗 = 1… . 𝑛) Variables
Constraints

5
 07-02-2024

LP in Matrix Form
• Objective Function: Maximize 𝑐1 𝑥1 + 𝑐2 𝑥2 + 𝑐3 𝑥3 +……….+ 𝑐𝑛 𝑥𝑛
• Subject to:
𝑎1,1 𝑥1,1 + 𝑎1,2 𝑥1,2 + 𝑐1,3 𝑥1,3 +……….+ 𝑐1,𝑛 𝑥1,𝑛 ≤ 𝑏1
𝑎2,1 𝑥2,1 + 𝑎2,2 𝑥2,2 + 𝑐2,3 𝑥2,3 +……….+ 𝑐2,𝑛 𝑥2,𝑛 ≤ 𝑏2

𝑎𝑚,1 𝑥𝑚,1 + 𝑎𝑚,2 𝑥𝑚,2 + 𝑐𝑚,3 𝑥𝑚,3 +……….+ 𝑐𝑚,𝑛 𝑥𝑚,𝑛 ≤ 𝑏𝑚

𝑥1 , 𝑥2 , 𝑥3 ,……….., 𝑥𝑛 ≥ 0
Matrix Form:
Maximize 𝑐 𝑇 *x, Subject to Ax ≤ b, x ≥ 0

Where c= [𝑐𝑗 ]𝑛𝑗=1 , b= [𝑏𝑖 ]𝑚 𝑛

𝑖=1 , A = [𝑎𝑖𝑗 ]𝑚∗𝑛 , x = [𝑥𝑗 ]𝑗=1

Two Variables and Two Constraints Problem

n = (2) number of decision variables (products)
m = (2) number of constraints (resources)
Objective Function:
𝑀𝑎𝑥 c1𝑥1 + c2𝑥2
Subject to:
𝑎1,1𝑥+ 𝑎 1,2 𝑥2 ≤ b 1
1
𝑎2,1𝑥 + 𝑎2,2 𝑥2 ≤ b2
1
𝒙𝟏, 𝒙𝟐 ≥ 𝟎

6
 07-02-2024

The Furniture Factory Problem

A data scientist is in charge of developing the Weekly

Production Plan for two key products that the furniture factory
makes: chairs and tables. The data scientist, using machine
learning techniques, predicts that the selling price of a chair is
$45 and the selling price of a table is $80 dollars.

• There are two critical resources in the production of chairs and

tables:
• Mahogany (measured in board square-feet) and labor (measured
in work hours).
• There are 400 units of mahogany available at the beginning of
each week.
• There are 450 units of labor available during each week.

7
 07-02-2024

• The data scientist estimates that

•One chair requires 5 units of mahogany and 10 units of
labor.
•One table requires 20 units of mahogany and 15 units of
labor.

• The marketing department has told the data scientist

that ALL the production chairs and tables can be sold.

• Identify all five components of the optimization model.

• Formulate an LP to help the data scientists make a Production

Plan that maximizes total revenue.

• Translate the LP model to Abstract and Generalize Form

8
 07-02-2024

Linear Programming model components

Actual Problem: LP Formulation Summary

1.0 . Max revenue = 45𝑥1 + 80𝑥2

2.0 5𝑥1 + 20𝑥2 ≤ 400 Units of mahogany capacity
3.0 . 10, 𝑥1 + 15𝑥2 ≤ 450 Labor hours capacity

𝒙𝟏, 𝒙𝟐 ≥ 𝟎 , Non − negativity

1.0 . Max (b1 = 45)𝑥1 + (𝑏2 = 80)𝑥2

2.0 𝑎1 1 = 5 𝑥1 + (𝑎1 2 = 20)𝑥2 ≤ 𝐾1 = 400
3.0 . (𝑎2 2 = 10)𝑥1 + (𝑎2 2 = 15)𝑥2≤ 𝐾2 = 450

𝒙𝟏, 𝒙𝟐 ≥ 𝟎

Parametrization of input data

Parametrization of input data of an LP problem allows
one to separate the data from the model. That is, one
can change the values of the data without changing the
model.

Abstraction and Generalization of Furniture Problem

Parametrized LP problem formulation

Prices

1.0 𝑀𝑎𝑥 𝑏1𝑥1 + 𝑏2𝑥2 ➔ Max σ2𝑗=1 𝑏𝑗 ∗ 𝑥 𝑗

2.0 𝑎1,1𝑥1 + 𝑎1,2𝑥2 ≤ 𝐾1
3.0 𝑎2,1 𝑥1 + 𝑎2,2 𝑥2 ≤ 𝐾2 ➔ σ2𝑗=1 𝑎𝑖,𝑗 ≤ 𝐾𝑖 (𝑖 = 1,2)
Resources
Technology capacity
coefficients
𝒙𝟏, 𝒙𝟐 ≥ 𝟎 ➔ 𝑥𝑗 ≥ 0, (𝑗 = 1,2)

9
 07-02-2024

Retail Store Problem

A Retail store is planning an advertising campaign aiming to
increase the number of customers visiting its physical location as
well as its online store.
• Each 1000 dollars invested in magazine ads will attract 100
new customers to store, as well as 500 new website visitors.
• Each 1000 dollars invested in online ads will attract 50 new
customers to store, as well as 1000 new website visitors.
• The target is to bring at least 500 new guests to the physical
store and at least 5000 new visitors to the online store.
Formulate an LP to help the store to minimize the cost of the
advertising campaign.

10