OPTIMIZATION TECHNIQUES

Ramalingam Sakthivel

Assistant Professor (Senior Grade)

School of Computer Science and Engineering (SCOPE)
Vellore Institute of Technology, Chennai.

1
 OPTIMIZATION TECHNIQUES

Module 1: Classical Optimization Techniques

• Introduction
• Methods
• Engineering applications of optimization
• Statement of an optimization problem
• Classification of optimization problems
• Single variable optimization
• Multivariable optimization with no constraints
• Multi variable optimization with equality and inequality
constraints: Lagrange multipliers methods, Kuhn-Tucker
conditions.

2
Ramalingam Sakthivel Assistant Professor (Senior Grade)
 INTRODUCTION

What is Optimization?
• Optimization is the process of finding the best possible solution or outcome under a given set of
circumstances.
• It involves maximizing or minimizing a particular objective function by adjusting input parameters
within specified constraints.
• Objective Function: A mathematical expression representing the goal (e.g., cost, profit, efficiency).
• Constraints: Limitations or restrictions on the variables (e.g., budget, resources, time).

3
Ramalingam Sakthivel Assistant Professor (Senior Grade)
 INTRODUCTION

Why is Optimization Needed?

[Link] Resource Utilization
• Helps in making the best use of limited resources like time, money, and materials.
[Link] Minimization
• Reduces operational costs in industries by improving processes.
[Link] Enhancement
• Optimizes performance in systems like transportation networks, supply chains, and manufacturing
units.
[Link] Decision-Making
• Provides quantitative solutions for complex decision-making problems.
[Link]-World Applications
• Used in designing energy-efficient buildings, developing machine learning algorithms, or optimizing
financial portfolios
4
Ramalingam Sakthivel Assistant Professor (Senior Grade)
 Applications of Optimization

Applications of
Optimization

Transportation and Machine Learning and

Business and Economics Healthcare Energy Systems
Logistics Artificial Intelligence

Supply chain management: Resource allocation in Power grid optimization: Route optimization: Finding Training models: Optimizing
Minimizing transportation hospitals: Assigning beds, Ensuring efficient power the shortest or fastest paths loss functions in algorithms
and inventory costs. staff, and equipment. distribution. for delivery. like deep learning.

Revenue management: Treatment planning: Renewable energy: Traffic management: Hyperparameter tuning:
Optimizing pricing strategies Optimizing radiation doses in Maximizing energy capture in Reducing congestion in urban Selecting the best model
for maximum profitability. cancer therapy. solar panels or wind turbines. areas. configurations.

Portfolio optimization:
Drug design: Finding the Energy storage: Optimizing Fleet management: Allocating Feature selection: Identifying
Allocating investments to
optimal combination of the use of battery systems in and scheduling vehicles for the most relevant features for
maximize returns while
molecules for new drugs. smart grids. logistics companies. prediction models.
minimizing risk.

5
Ramalingam Sakthivel Assistant Professor (Senior Grade)
 Engineering Applications of Optimization

Mechanical Structural Design: Thermal Systems: Vibration Control:

Minimizing material usage Optimizing heat exchangers Reducing unwanted
Engineering while maintaining strength. and refrigeration cycles. vibrations in machinery.

Construction Planning: Structural Optimization:

Traffic Flow Optimization:
Scheduling tasks to Designing bridges or
Civil Engineering minimize project costs and
Designing road networks
buildings for stability and
for smooth traffic flow.
time. cost-effectiveness.

Electrical Circuit Design: Minimizing Power Systems: Signal Processing:

power consumption in Optimizing load Improving clarity and
Engineering electronic circuits. distribution in power grids. reducing noise in signals.

Flight Path Optimization:

Aerospace Aircraft Design: Satellite Placement:
Minimizing fuel
Optimizing aerodynamics Determining optimal orbits
Engineering consumption and travel
for better performance. for coverage and efficiency.
time.

Chemical Process Optimization: Mixing and Blending: Catalyst Design: Improving

Enhancing efficiency in Optimizing proportions of reaction rates in chemical
Engineering chemical production plants. raw materials. processes.

Computer and Network Optimization: Cloud Computing:

Algorithm Design:
Ensuring efficient data flow Optimizing resource
Software in communication allocation in virtual
Creating algorithms for
Engineering optimal problem-solving
networks. environments.
6
Ramalingam Sakthivel Assistant Professor (Senior Grade)
 Statement of an Optimization Problem

Objective Function
• A Mathematical expression representing the goal of the optimization.
• Example: Maximize 𝑓(𝑥) or Minimize 𝑓 𝑥 , where 𝑓(𝑥) is the function to optimize.

Decision Variables
• Variables that can be adjusted to optimize the objective function.
• Example: 𝑥1 , 𝑥2 , 𝑥3 , … … … 𝑥𝑛 represent the variables to be determined.

Constraints
Restrictions or limitations on the decision variables. These can be:
• Equality Constraints: 𝑔𝑖 𝑥 = 0
• Inequality Constraints: ℎ𝑗 (𝑥) ≤ 𝑏𝑗

Feasible Region
• The set of all possible values of the decision variables that satisfy the constraints.

Optimization Type
• Specifies whether the problem is a maximization or minimization problem.

7
Ramalingam Sakthivel Assistant Professor (Senior Grade)
 Statement of an Optimization Problem

Mathematical Formulation

Optimize 𝑓(𝑥) (Objective Function)

𝑔𝑖 𝑥 = 0 (𝐸𝑞𝑢𝑎𝑙𝑖𝑡𝑦 𝐶𝑜𝑛𝑠𝑡𝑟𝑎𝑛𝑖𝑛𝑡𝑠)
Subject to: ൝
ℎ𝑗 𝑥 ≤ 𝑏𝑗 (𝐼𝑛𝑒𝑞𝑢𝑎𝑙𝑖𝑡𝑦 𝐶𝑜𝑛𝑠𝑡𝑟𝑎𝑖𝑛𝑡𝑠)

𝑥 ∈ 𝓍 (Feasible Region)

8
Ramalingam Sakthivel Assistant Professor (Senior Grade)
 Classification of Optimization Problems

Classification of
Optimization
Problems

Based on the Objective Based on the Nature of Based on Deterministic Based on Solution
Based on Constraints
Function Variables or Stochastic Nature Techniques

Linear Optimization Continuous Constrained Deterministic

Exact Methods
(Linear Programming) Optimization Optimization Optimization

Nonlinear Discrete Unconstrained Stochastic Heuristic

Optimization Optimization Optimization Optimization Methods

9
Ramalingam Sakthivel Assistant Professor (Senior Grade)
 SINGLE-VARIABLE OPTIMIZATION

• Single variable optimization involves finding the maximum or minimum value of a function that depends on a
single variable, typically within a specific domain.

• The function to be optimized is of the form 𝑓(𝑥), where 𝑥 is the decision variable.

• A function of one variable 𝑓(𝑥) is said to have a relative or local minimum at 𝑥 = 𝑥 ∗ if 𝑓(𝑥 ∗ ) ≤ 𝑓(𝑥 ∗ + ℎ) for
all sufficiently small positive and negative values of ℎ.

• Similarly a point 𝑥 ∗ is called a relative or local maximum if 𝑓(𝑥 ∗ ) ≥ 𝑓(𝑥 ∗ + ℎ) for all values of ℎ sufficiently
close to zero.

• A function 𝑓(𝑥) is said to have a global or absolute minimum at 𝑥 ∗ if 𝑓(𝑥 ∗ ) ≤ 𝑓(𝑥) for all 𝑥.

• Similarly, a point 𝑥 ∗ will be a global maximum of 𝑓(𝑥) if 𝑓(𝑥 ∗ ) ≥ 𝑓(𝑥) for all 𝑥.

10
Ramalingam Sakthivel Assistant Professor (Senior Grade)
 SINGLE-VARIABLE OPTIMIZATION

Why Single Variable Optimization is Needed?

Simplifies Problem Solving

• Single variable optimization deals with functions of only one variable, making it more straightforward than
multi-variable optimization.
• Useful for initial problem analysis or when the problem inherently involves only one variable.
Foundation for More Complex Methods
• Many real-world problems can be reduced to a single-variable case, either by fixing other variables or
through parameterization.
• It lays the groundwork for understanding and solving multi-variable optimization problems.
Efficiency in Decision-Making
• Helps in quick decision-making when optimizing processes or outcomes with a single varying factor.
• Examples include finding the optimal price, speed, or time for a specific goal.
Applicability in Constrained Settings
• Single variable optimization is often used in scenarios with natural constraints, such as time or distance,
making it an essential tool for such problems.
11
Ramalingam Sakthivel Assistant Professor (Senior Grade)
 SINGLE-VARIABLE OPTIMIZATION

Applications

Economics and
Engineering Physics Operations Research Computer Science Healthcare Transportation Environmental Studies
Business

Profit Design
Equilibrium Algorithm Treatment Route Pollution
Maximizati Optimizatio Scheduling
Points Design Planning Planning Control
on n

Cost Thermodyna Energy Inventory Machine Resource Speed Resource

Minimization mics Minimization Management Learning Allocation Optimization Conservation

12
Ramalingam Sakthivel Assistant Professor (Senior Grade)
 MULTI-VARIABLE OPTIMIZATION WITH NO CONSTRANTS

Definition
Multivariable optimization with no constraints involves finding the maximum or minimum of a function
𝑓(𝑥1 , 𝑥2 , … … , 𝑥𝑛 ) where there are no restrictions on the decision variables 𝑥1 , 𝑥2 , … … , 𝑥𝑛 .

Objective Function
The goal is to optimize (maximize or minimize) 𝑓(𝑥1 , 𝑥2 , … … , 𝑥𝑛 ) .

Critical Points
Critical points occur where all first partial derivatives are zero:
𝜕𝑓 𝜕𝑓 𝜕𝑓
= =⋯= =0
𝜕𝑥1 𝜕𝑥2 𝜕𝑥𝑛

Gradient Vector
𝜕𝑓 𝜕𝑓 𝜕𝑓
The gradient ∇𝑓 = (𝜕𝑥 , 𝜕𝑥 , … , 𝜕𝑥 ) represents the direction of steepest ascent.
1 2 𝑛

At a critical point, ∇𝑓 = 0
13
Ramalingam Sakthivel Assistant Professor (Senior Grade)
 MULTI-VARIABLE OPTIMIZATION WITH NO CONSTRANTS

Second Derivative Test (Hessian Matrix) Conditions for Positive Definite:

The Hessian matrix 𝐻 is used to determine the nature of 𝐻1 > 0, 𝐻2 > 0, 𝐻3 > 0, … … . … … .
the critical points.
⟹ All the principal minors are positive

𝜕2𝑓 𝜕2𝑓
... Conditions for Negative Definite:
𝜕𝑥12 𝜕𝑥1 𝜕𝑥2
𝐻1 < 0, 𝐻2 > 0, 𝐻3 < 0, … … . … … .
𝜕2𝑓 𝜕2𝑓
𝐻= ... ⟹ The principal minors are in opposite sings.
𝜕𝑥2 𝜕𝑥1 𝜕𝑥22
. . .
. . . Where,
. . .
𝐻1 is the determinant of 1 × 1 matrix of 𝐻.

Test for definiteness of 𝐻: 𝐻2 is the determinant of 2 × 2 matrix of 𝐻.

• Positive Definite: Local minimum. 𝐻3 is the determinant of 3 × 3 matrix of 𝐻.

• Negative Definite: Local maximum. .

• Indefinite: Saddle point. .

.

14
Ramalingam Sakthivel Assistant Professor (Senior Grade)
 MULTI-VARIABLE OPTIMIZATION WITH NO CONSTRANTS

Steps for Solving

• Find the first partial derivatives of 𝑓(𝑥1 , 𝑥2 , … … , 𝑥𝑛 ) .
• Solve ∇𝑓 = 0 to find critical points.
• Evaluate the Hessian at each critical point to classify it.

Global vs. Local Optima

Without constraints, the solution may not always be a global optimum; it could be a
• Local Maximum
• Local Minimum
• Saddle Point.

15
Ramalingam Sakthivel Assistant Professor (Senior Grade)
Figure 1.1 Relative and global minima

16
Ramalingam Sakthivel Assistant Professor (Senior Grade)
 APPLICATIONS OF MULTI-VARIABLE OPTIMIZATION WITH NO CONSTRANTS

Applications

Engineering Design Economics Physics Machine Learning Operations Research

Maximizing profits or
Optimizing the shape of a
minimizing costs as functions Minimizing energy functions Designing systems like
structure or a machine Optimizing model parameters
of multiple variables like in systems without boundary communication networks for
component for strength and (e.g., weights) during training.
labor, materials, and conditions. maximum efficiency.
efficiency.
production levels.

17
Ramalingam Sakthivel Assistant Professor (Senior Grade)
 MULTI-VARIABLE OPTIMIZATION WITH EQUALITY AND INEQUALITY CONSTRANTS

Definition
When solving multivariable optimization problems with constraints, the goal is to optimize (maximize or
minimize) a function 𝑓 𝑥1 , 𝑥2 , … … , 𝑥𝑛 , subject to equality and/or inequality constraints.

Objective Function
The function to be optimized, 𝑓 𝑥1 , 𝑥2 , … … , 𝑥𝑛 .

Constraints
• Equality Constraints: 𝑔𝑖 𝑥1 , 𝑥2 , … … , 𝑥𝑛 = 0, 𝑖 = 1,2, … , 𝑚.
• Inequality Constraints: ℎ𝑗 𝑥1 , 𝑥2 , … … , 𝑥𝑛 ≤ 0, 𝑗 = 1,2, . . , 𝑝.

Feasible Region
The set of all points that satisfy the constraints.

Optimization Goal
Find points in the feasible region where the objective function is maximized or minimized.

18
Ramalingam Sakthivel Assistant Professor (Senior Grade)
 MULTI-VARIABLE OPTIMIZATION WITH EQUALITY AND INEQUALITY CONSTRANTS

Methods for Solving Constrained Optimization Problems

1. Lagrange Multiplier Method (for Equality Constraints)
Used when only equality constraints are present.

• Objective Function:
Optimize (Minimize or Maximize): 𝑓 𝑥1 , 𝑥2 , … … , 𝑥𝑛 .
Subject to Equality Constraints:
𝑔𝑖 𝑥1 , 𝑥2 , … … , 𝑥𝑛 = 0, 𝑖 = 1,2, … , 𝑚.
Lagrangian Function Formulation
The Lagrangian function combines the objective function 𝑓 𝑥1 , 𝑥2 , … … , 𝑥𝑛 and the equality constraints
𝑔𝑖 𝑥1 , 𝑥2 , … … , 𝑥𝑛 using Lagrange multipliers λ𝑖 :
𝐿(𝑥1 , 𝑥2 , … … , 𝑥𝑛 , λ1 , λ2 , … … , λ𝑚 ) = 𝑓 𝑥1 , 𝑥2 , … … , 𝑥𝑛 − σ𝑚
𝑖=1 λ𝑖 𝑔𝑖 𝑥1 , 𝑥2 , … … , 𝑥𝑛
where
• λ𝑖 are Lagrange multipliers.
• 𝑔𝑖 𝑥1 , 𝑥2 , … … , 𝑥𝑛 : Equality constraints.
19
Ramalingam Sakthivel Assistant Professor (Senior Grade)
 MULTI-VARIABLE OPTIMIZATION WITH EQUALITY AND INEQUALITY CONSTRANTS

Lagrangian Optimization Conditions:

• Stationary Condition:
The gradient of the Lagrangian with respect to 𝑥𝑘 must be zero.

𝜕𝐿
= 0, 𝑘 = 1, … , 𝑛,
𝜕𝑥𝑘
• Constraint Satisfaction:
The equality constraints must hold
𝑔𝑖 𝑥1 , 𝑥2 , … … , 𝑥𝑛 = 0, 𝑖 = 1,2, … , 𝑚.

20
Ramalingam Sakthivel Assistant Professor (Senior Grade)
 MULTI-VARIABLE OPTIMIZATION WITH EQUALITY AND INEQUALITY CONSTRANTS

2. Karush-Kuhn-Tucker (KKT) Conditions (for Equality and Inequality

Constraints)

• The Karush-Kuhn-Tucker (KKT) conditions are used for solving optimization problems with both equality and
inequality constraints. These conditions generalize the Lagrangian method by incorporating inequality
constraints.

Optimization Problem Setup

Objective:
Optimize (Minimize or Maximize): 𝑓 𝑥1 , 𝑥2 , … … , 𝑥𝑛
Subject to Constraints:
[Link] Constraints: 𝑔𝑖 𝑥1 , 𝑥2 , … … , 𝑥𝑛 = 0, 𝑖 = 1,2, … , 𝑚.
[Link] Constraints: ℎ𝑗 𝑥1 , 𝑥2 , … … , 𝑥𝑛 ≤ 0, 𝑗 = 1,2, . . , 𝑝.

21
Ramalingam Sakthivel Assistant Professor (Senior Grade)
 MULTI-VARIABLE OPTIMIZATION WITH EQUALITY AND INEQUALITY CONSTRANTS

KKT Conditions
The KKT conditions consist of the following:
1. KKT Lagrangian Function
The Lagrangian function is extended to include both equality and inequality constraints
𝑝
𝐿(𝑥1 , … … , 𝑥𝑛 , λ1 , … … , λ𝑚 , 𝜇1 , … , 𝜇𝑝 ) = 𝑓 𝑥1 , 𝑥2 , … … , 𝑥𝑛 − σ𝑚
𝑖=1 λ𝑖 𝑔𝑖 𝑥1 , 𝑥2 , … … , 𝑥𝑛 − σ𝑗=1 𝜇𝑗 ℎ𝑗 𝑥1 , 𝑥2 , … … , 𝑥𝑛

Where,
• λ𝑖 : Lagrange multipliers for equality constraints.
• 𝜇𝑗 : Lagrange multipliers for inequality constraints (𝜇𝑗 ≥ 0).
• 𝑔𝑖 𝑥1 , 𝑥2 , … … , 𝑥𝑛 : Equality constraints.
• ℎ𝑗 𝑥1 , 𝑥2 , … … , 𝑥𝑛 : Inequality constraints.

2. Stationarity Condition
The gradient of the Lagrangian with respect to 𝑥 must be zero:
∇𝑥 𝐿(𝑥1 , … … , 𝑥𝑛 , λ1 , … … , λ𝑚 , 𝜇1 , … , 𝜇𝑝 ) = ∇𝑓 𝑥1 , 𝑥2 , … … , 𝑥𝑛 − σ𝑚
𝑖=1 λ𝑖 ∇𝑔𝑖 𝑥1 , 𝑥2 , … … , 𝑥𝑛 −
σ𝑝𝑗=1 𝜇𝑗 ∇ℎ𝑗 𝑥1 , 𝑥2 , … … , 𝑥𝑛

22
Ramalingam Sakthivel Assistant Professor (Senior Grade)
 MULTI-VARIABLE OPTIMIZATION WITH EQUALITY AND INEQUALITY CONSTRANTS

3. Primal Feasibility
The solution must satisfy the equality and inequality constraints
𝑔𝑖 𝑥1 , 𝑥2 , … … , 𝑥𝑛 = 0
ℎ𝑗 𝑥1 , 𝑥2 , … … , 𝑥𝑛 ≤ 0

4. Dual Feasibility
The multipliers for inequality constraints must be non-negative
𝜇𝑗 ≥ 0.

5. Complementary Slackness
For each inequality constraint, the product of the multiplier and the constraint must be zero
𝜇𝑗 ℎ𝑗 = 0. 𝑗 = 1,2, … , 𝑝.

23
Ramalingam Sakthivel Assistant Professor (Senior Grade)
 QUESTIONS AND PROBLEMS RELATED ON THIS MODULE

1. Explain briefly the statement of optimization problem.

Answer:
An optimization problem involves finding the best possible solution from a set of feasible solutions.
The goal is to maximize or minimize an objective function while satisfying certain constraints.

Components of an Optimization Problem:

1. Objective Function:
The mathematical expression to be optimized (minimized or maximized).
𝑓 𝑥 𝑜𝑟 𝑓 𝑥1 , 𝑥2 , … … , 𝑥𝑛
2. Decision Variables:
The variables that can be adjusted to achieve the optimal solution. Represented as 𝑥1 , 𝑥2 , … … , 𝑥𝑛 .

3. Constraints:
Conditions that the solution must satisfy, classified into
• Equality Constraints: 𝑔𝑖 𝑥1 , 𝑥2 , … … , 𝑥𝑛 = 0, 𝑖 = 1, . . , 𝑚

24
Ramalingam Sakthivel Assistant Professor (Senior Grade)
 QUESTIONS AND PROBLEMS RELATED ON THIS MODULE

• Inequality Constraints: ℎ𝑗 𝑥1 , 𝑥2 , … … , 𝑥𝑛 ≤ 0 , 𝑗 = 1, … . . , 𝑝.

4. Feasible Region:
The set of all points 𝑥 that satisfy the constraints.

Mathematical Formulation
Optimize (Minimize or Maximize): 𝑓 𝑥1 , 𝑥2 , … … , 𝑥𝑛
Subject to Constraints:
𝑔𝑖 𝑥1 , 𝑥2 , … … , 𝑥𝑛 = 0, 𝑖 = 1,2, … , 𝑚.
ℎ𝑗 𝑥1 , 𝑥2 , … … , 𝑥𝑛 ≤ 0, 𝑗 = 1,2, . . , 𝑝.

25
Ramalingam Sakthivel Assistant Professor (Senior Grade)
 QUESTIONS AND PROBLEMS RELATED ON THIS MODULE

Quadratic Expression
• An expression of degree 𝑛 in variable 𝑥 is 𝑎0 𝑥 𝑛 + 𝑎1 𝑥 𝑛−1 + 𝑎2 𝑥 𝑛−2 + ⋯ +𝑎𝑛−1 𝑥 + 𝑎𝑛 = 0 where 𝑎0 ≠ 0 and
𝑎1 , 𝑎2 , 𝑎3 , … , 𝑎𝑛 are real numbers. 𝑎0 , 𝑎1 , 𝑎2 , … , 𝑎𝑛 are called coefficients of the expression.
• In particular an expression of degree 2 is called a Quadratic Expression which is expressed as
𝒑 𝒙 = 𝒂𝒙𝟐 + 𝒃𝒙 + 𝒄, 𝑎 ≠ 0 and 𝑎, 𝑏, 𝑐 are real numbers.

Roots of Quadratic Equations

Let 𝑎𝑥 2 + 𝑏𝑥 + 𝑐 = 0, 𝑎 ≠ 0 be a quadratic equation. The value of 𝑥 such that the expression
𝑎𝑥 2 + 𝑏𝑥 + 𝑐 becomes zero are called roots of the quadratic equation 𝑎𝑥 2 + 𝑏𝑥 + 𝑐.

−𝒃 ± 𝒃𝟐 − 𝟒𝒂𝒄
𝒙=
𝟐𝒂

−𝑏+ 𝑏2 −4𝑎𝑐 −𝑏− 𝑏2 −4𝑎𝑐

∴ The roots are 𝑥 = and 𝑥 =
2𝑎 2𝑎

26
Ramalingam Sakthivel Assistant Professor (Senior Grade)
 QUESTIONS AND PROBLEMS RELATED ON THIS MODULE

Example:
Find the roots of the quadratic equation 𝒙𝟐 + 𝟖𝒙 + 𝟏𝟐 = 𝟎
Solution:
Compare given equation with the quadratic expression, we get
𝑎 = 1, 𝑏 = 8, 𝑐 = 12

−𝒃 ± 𝒃𝟐 − 𝟒𝒂𝒄
𝒙=
𝟐𝒂
−8 ± 82 − 4 × 1 × 12 −8 ± 64 − 48 −8 ± 16 −8 ± 4
𝑥= ⟹𝑥= ⟹𝑥= ⟹𝑥=
2×1 2 2 2

−8+4 −8−4
⟹𝑥= and 𝑥=
2 2
⟹ 𝑥 = −2 and 𝑥 = −6 are the roots of the given quadratic equation.

∴ The factors are 𝑥 + 2 and 𝑥 + 6.

27
Ramalingam Sakthivel Assistant Professor (Senior Grade)
 QUESTIONS AND PROBLEMS RELATED ON THIS MODULE

Synthetic Division
• We can use Synthetic Division method for solving Higher-Degree equations.

Example:
Find the roots of the equation 𝒙𝟒 + 𝒙𝟑 − 𝟏𝟏𝒙𝟐 − 𝟓𝒙 + 𝟑𝟎 = 𝟎
Solution:
Use synthetic division, because this equation is in degree 3.

1 1 − 11 −5 30
2 0 2 6 − 10 − 30
1 3 −5 − 15 0
−3 0 −3 0 15
1 0 −5 0

⟹ 𝑥 − 2 𝑥 + 3 𝑥2 − 5 = 0
∴ The roots are 𝑥 = 2, 𝑥 = −3, 𝑥 = ± 5.
28
Ramalingam Sakthivel Assistant Professor (Senior Grade)
 QUESTIONS AND PROBLEMS RELATED ON THIS MODULE

2. Determine the minimum value of the function

𝒇 𝒙 = 𝒙𝟐 − 𝟒𝒙 + 𝟓
Solution:
To find critical point:
Differentiate 𝑓(𝑥)
𝑓 ′ 𝑥 = 2𝑥 − 4,
Set 𝑓 ′ 𝑥 = 0
𝑥 = 2 is a critical point

To verify the nature of critical point:

Taking second derivative, and using critical point
𝑓″ 𝑥 = 2
At 𝑥 = 2, 𝑓 ″ 𝑥 = 2 > 0 and hence 𝑥 = 2 is a relative minimum. Therefore,
𝑓𝑚𝑖𝑛 = 𝑓 𝑥 = 2 = 22 − 4 2 + 5 = 4 − 8 + 5 = 1
∴ The minimum value of 𝑓 𝑥 = 1 at 𝑥 = 2.

29
Ramalingam Sakthivel Assistant Professor (Senior Grade)
 QUESTIONS AND PROBLEMS RELATED ON THIS MODULE

3. Determine the maximum and minimum values of the function

𝒇 𝒙 = 𝟏𝟐𝒙𝟓 − 𝟒𝟓𝒙𝟒 + 𝟒𝟎𝒙𝟑 + 𝟓
Solution:
To find critical point:
Differentiate 𝑓(𝑥)
𝑓 ′ 𝑥 = 60 𝑥 4 − 3𝑥 3 + 2𝑥 2 = 60𝑥 2 𝑥 2 − 3𝑥 + 2 = 60𝑥 2 𝑥 − 1 𝑥 − 2 ,
𝑓 ′ 𝑥 = 0 at 𝑥 = 0, 𝑥 = 1, 𝑥 = 2.

To verify the nature of critical point:

Taking second derivative, and using critical point
𝑓 ″ 𝑥 = 60(4𝑥 3 − 9𝑥 2 + 4𝑥)

At 𝑥 = 1, 𝑓 ″ 𝑥 = −60 < 0 and hence 𝑥 = 1 is a relative maximum. Therefore,

𝑓𝑚𝑎𝑥 = 𝑓 𝑥 = 1 = 12

30
Ramalingam Sakthivel Assistant Professor (Senior Grade)
 QUESTIONS AND PROBLEMS RELATED ON THIS MODULE

At 𝑥 = 2, 𝑓 ″ 𝑥 = 240 > 0 and hence 𝑥 = 2 is a relative minimum. Therefore,

𝑓𝑚𝑖𝑛 = 𝑓 𝑥 = 2 = −11

At 𝑥 = 0, 𝑓 ″ 𝑥 = 0 and hence we must investigate the next derivative:

𝑓 ‴ 𝑥 = 60 12𝑥 2 − 18𝑥 + 4 = 240 at 𝑥 = 0

Since 𝑓 ‴ 𝑥 ≠ 0 at 𝑥 = 0, 𝑥 = 0 is neither a maximum nor a minimum, and it is an inflection point.

31
Ramalingam Sakthivel Assistant Professor (Senior Grade)
 QUESTIONS AND PROBLEMS RELATED ON THIS MODULE

4. Find the maximum and minimum values of the function

𝒙𝟐
𝒇 𝒙 =
(𝒙 − 𝟏)(𝒙 − 𝟐)
Solution:
To find critical point:
𝑢(𝑥)
Let 𝑓 𝑥 = 𝑣(𝑥)

Where, 𝑢 𝑥 = 𝑥2 𝑣 𝑥 = 𝑥−1 𝑥−2

𝑢′ 𝑥 = 2𝑥 𝑣 ′ 𝑥 = 𝑥 − 1 + 𝑥 − 2 = 2𝑥 − 3

Differentiate 𝑓 𝑥
𝑣 𝑥 𝑢′ 𝑥 − 𝑢(𝑥)𝑣 ′ 𝑥
𝑓′ 𝑥 =
𝑣(𝑥) 2
𝑥 − 1 𝑥 − 2 2𝑥 − 𝑥 2 (2𝑥 − 3 )
𝑓′ 𝑥 = 2
𝑥 − 1 (𝑥 − 2)

32
Ramalingam Sakthivel Assistant Professor (Senior Grade)
 QUESTIONS AND PROBLEMS RELATED ON THIS MODULE

Set 𝑓 ′ 𝑥 = 0
′
𝑥 − 1 𝑥 − 2 2𝑥 − 𝑥 2 (2𝑥 − 3 )
𝑓 𝑥 = 2 =0
𝑥 − 1 (𝑥 − 2)
2𝑥 𝑥 2 − 3𝑥 + 2 − 𝑥 2 (2𝑥 − 3 ) = 0
2𝑥 3 − 6𝑥 2 + 4𝑥 − 2𝑥 3 + 3𝑥 2 = 0
−3𝑥 2 + 4𝑥 = 0
𝑥(4 − 3𝑥) = 0
∴ The critical points are
4
𝑥=0 𝑎𝑛𝑑 𝑥=3
To verify the nature of critical point:
Taking second derivative, and using critical point
2
(𝑥 − 1)(𝑥 − 2) 4 − 6𝑥 − 4𝑥 − 3𝑥 2 2(𝑥 − 1)(𝑥 − 2)[ 𝑥 − 1 + (𝑥 − 2)]
𝑓″ 𝑥 = 4
𝑥 − 1 (𝑥 − 2)
𝑥−1 𝑥−2 2 (4) 4×(−2)2 16
At 𝑥 = 0, 𝑓 ″ 𝑥 = 4 = = 16 = 1 > 0 and hence 𝑓(𝑥) is minimum at 𝑥 = 0. Therefore,
−1 (−2) (−2)4
0
𝑓𝑚𝑖𝑛 = 𝑓 𝑥 = 0 = (−1)(−2) =0
33
Ramalingam Sakthivel Assistant Professor (Senior Grade)
 QUESTIONS AND PROBLEMS RELATED ON THIS MODULE
2
(𝑥 − 1)(𝑥 − 2) 4 − 6𝑥 − 4𝑥 − 3𝑥 2 2(𝑥 − 1)(𝑥 − 2)[ 𝑥 − 1 + (𝑥 − 2)]
𝑓″ 𝑥 = 4
𝑥 − 1 (𝑥 − 2)
4
At 𝑥 = 3 ,
2 2
4 4 4 4 4 4 4 4 4
3−1 3−2 4−6 3 − 4 3 −3 3 2 3−1 3−2 − 1 +
3 3−2
𝑓″ 𝑥 = 4
4 4
3−1 3−2
2
1 −2 16 16 1 −2 1 −2
× 4−8 − 3 − 3 ×2×3× 3 × 3+ 3
3 3
= 4
1 −2
×
3 3
2
−2
× −4 − 0 −4 −4
= 9 = = = −81 < 0
−2 4
−2 2 4
9 9 81
4
and hence 𝑓(𝑥) is maximum at 𝑥 = 3 .

34
Ramalingam Sakthivel Assistant Professor (Senior Grade)
 QUESTIONS AND PROBLEMS RELATED ON THIS MODULE

Therefore,
4 2
4 3
𝑓𝑚𝑎𝑥 = 𝑓 𝑥 = = 4 4
3 −1 3−2
3
16 9
= ×
9 −2
= −8

35
Ramalingam Sakthivel Assistant Professor (Senior Grade)
 QUESTIONS AND PROBLEMS RELATED ON THIS MODULE

5. Find the minimum value of the function

𝒇 𝒙 = 𝒙𝟔 − 𝟔𝒙𝟒 + 𝟗𝒙𝟐
Solution:
To find critical point: Differentiate 𝑓(𝑥)
Since 𝑓 ′ 𝑥 = 6𝑥 5 − 24𝑥 3 + 18𝑥,
Set 𝑓 ′ 𝑥 = 0
𝑓 ′ 𝑥 = 6𝑥 5 − 24𝑥 3 + 18𝑥 = 0 = 6𝑥 𝑥 4 − 4𝑥 2 + 3 = 0
6𝑥 = 0 or 𝑥 4 − 4𝑥 2 + 3 = 0
𝑥=0
1 0 −4 0 3
1 0 1 1 −3 −3
1 1 −3 −3 0
−1 0 −1 0 3
1 0 −3 0

∴ The factors are (𝑥 − 1)(𝑥 + 1)(𝑥 2 − 3) = 0

36
Ramalingam Sakthivel Assistant Professor (Senior Grade)
 QUESTIONS AND PROBLEMS RELATED ON THIS MODULE

𝑥−1 =0 or 𝑥+1 =0 or (𝑥 2 − 3) = 0
𝑥=1 𝑥 = −1 𝑥=± 3

⟹ 𝑓 𝑥 = 0 at 𝑥 = 0, 𝑥 = 1, 𝑥 = −1, 𝑥 = ± 3

To verify the nature of critical point:

Taking second derivative, and using critical point
𝑓 ″ 𝑥 = 30𝑥 4 − 72𝑥 2 + 18
Evaluate 𝑓 ″ 𝑥 at the critical points:

At 𝑥 = 0, 𝑓 ″ 𝑥 = 18 > 0 and hence 𝑥 = 0 is a local minimum.

𝑓𝑚𝑖𝑛 = 𝑓 𝑥 = 0 = 06 − 6 0 4 +9 0 2 =0

At 𝑥 = ±1, 𝑓 ″ 𝑥 = −24 < 0 and hence 𝑥 = ±1 is local maximum.

𝑓𝑚𝑎𝑥 = 𝑓 𝑥 = ±1 = (±1)6 −6 ±1 4
+ 9 ±1 2
=4

37
Ramalingam Sakthivel Assistant Professor (Senior Grade)
 QUESTIONS AND PROBLEMS RELATED ON THIS MODULE

At 𝑥 = ± 3, 𝑓 ″ 𝑥 = 90 > 0 and hence 𝑥 = ± 3 is local minimum.

4 2
𝑓𝑚𝑖𝑛 = 𝑓 𝑥 = ± 3 = (± 3)6 −6 ± 3 +9 ± 3 = 27 − 54 + 27 = 0

The global minimum is 𝑓 𝑥 = 0 at 𝑥 = 0 and 𝑥 = ± 3.

38
Ramalingam Sakthivel Assistant Professor (Senior Grade)
 QUESTIONS AND PROBLEMS RELATED ON THIS MODULE

6. A rectangular garden is to be constructed using a rock wall as one side of the garden
and wire fencing for the other three sides (as in Figure 1). Given 100 ft of wire fencing,
determine the dimensions that would create a garden of maximum area. What is the
maximum area?

Figure 1
39
Ramalingam Sakthivel Assistant Professor (Senior Grade)
 QUESTIONS AND PROBLEMS RELATED ON THIS MODULE

Solution:
Let 𝑥 denote the length of the side of the garden perpendicular to the rock wall and 𝑦 denote the length of
the side parallel to the rock wall.

The are of the garden is

𝐴 = 𝑥 × 𝑦 −−−−−−−−−→ 1
we want to find the maximum possible area subject to the constraint that the total fencing is 100 ft.

As in the Figure 1, the total amount of fencing used will be

𝑥 + 𝑦 + 𝑥 = 100
2𝑥 + 𝑦 = 100
𝑦 = 100 − 2𝑥 −−−−−→ [2]
Substitute [2] in [1] we get,
𝐴 𝑥 = 𝑥 100 − 2𝑥
𝐴 𝑥 = 100𝑥 − 2𝑥 2

40
Ramalingam Sakthivel Assistant Professor (Senior Grade)
 QUESTIONS AND PROBLEMS RELATED ON THIS MODULE

To find critical point: Hence the dimensions are

Taking derivative 𝐴(𝑥) 𝑥 = 25 𝑓𝑡 and 𝑦 = 50 𝑓𝑡
′
𝐴 𝑥 = 100 − 4𝑥
Set 𝐴′ 𝑥 = 0 ⟹ 100 − 4𝑥 = 0 ∴ The maximum area of the garden is
4𝑥 = 100
𝑥 = 25 𝐴 = 25 × 50
∴ The critical point is 𝑥 = 25 𝐴 = 1250 𝑓𝑡 2

To verify the nature of critical point:

Taking second derivative, and using critical point
𝐴″ 𝑥 = −4
At 𝑥 = 25, 𝐴″ 𝑥 = −4 < 0 is local maximum

To find the value of 𝒚:

Substitute 𝑥 = 25 in [2], we get
𝑦 = 100 − 2 25 ⟹ 𝑦 = 50
41
Ramalingam Sakthivel Assistant Professor (Senior Grade)
 QUESTIONS AND PROBLEMS RELATED ON THIS MODULE

7. A rectangular box with a square base, an open top, and a volume of 𝟐𝟏𝟔 𝒊𝒏𝟑 is to be
constructed. What should the dimensions of the box be to minimize the surface area of
the box? What is the minimum surface area?
Solution:
Draw a rectangular box and introduce the variable 𝑥 to represent the length of each side of the square
base and 𝑦 to represent the height of the box.

Let 𝑆 denote the surface area of the open-top box. Since the box has an open top, we need only determine
the area of the four vertical sides and the base. The area of each of the four vertical sides is 𝑥 × 𝑦. The area of the
base is 𝑥 2 . Therefore, the surface area of the box is
𝑆 = 4𝑥𝑦 + 𝑥 2 −−−−−−→ [1]
42
Ramalingam Sakthivel Assistant Professor (Senior Grade)
 QUESTIONS AND PROBLEMS RELATED ON THIS MODULE

Since the volume of this box is 𝑥 2 𝑦 and the volume is given as 216 𝑖𝑛3 .
𝑥 2 𝑦 = 216
216
𝑦 = 2 −−−−−→ [2]
𝑥
Substitute [2] in [1], we get
216 2
864
𝑆 𝑥 = 4𝑥 2
+𝑥 = + 𝑥2
𝑥 𝑥
To find critical point:
Taking derivative S(𝑥)
864
𝑆′ 𝑥 = − + 2𝑥
𝑥2
864
Set 𝑆 ′ 𝑥 = 0 ⟹ − + 2𝑥 = 0
𝑥2
864
= 2𝑥
𝑥2
3 3 3
𝑥 3 = 432 ⟹ 𝑥 = 432 ⟹ 𝑥 = 6×6×6×2 ⟹𝑥 = 6 2
3
∴ The critical point is 𝑥 = 6 2

43
Ramalingam Sakthivel Assistant Professor (Senior Grade)
 QUESTIONS AND PROBLEMS RELATED ON THIS MODULE

To verify the nature of critical point:

Taking second derivative, and using critical point
2×864
𝑆″ 𝑥 = +2
𝑥3
3 2×864 2×864
At 𝑥 = 6 2 , 𝑆 ″ 𝑥 = 3 3 + 2 ⟹ 𝑆 ″ 𝑥 = 216×2 + 2 ⟹ 𝑆 ″ 𝑥 = 4 + 2 = 6 > 0 is minimum
6 2

To find the value of 𝒚:

3
Substitute 𝑥 = 6 2 in [2], we get
216
𝑦= 3 2
6 2
216
𝑦= 3
36( 2)2
6
𝑦= 3
4
3
6 2
𝑦= 3 ×3
4 2
3 3
6 2 6 2 3
𝑦= 3 ⟹𝑦= ⟹𝑦=3 2
8 2

44
Ramalingam Sakthivel Assistant Professor (Senior Grade)
 QUESTIONS AND PROBLEMS RELATED ON THIS MODULE

Hence the dimensions are

3 3
𝑥 = 6 2 𝑖𝑛 and 𝑦 = 3 2 𝑖𝑛

∴ The minimum surface area

3 3 3
𝑆=4 6 2 3 2 + (6 2)2
3 2 3 2
𝑆=4 6 3 2 + 36 2
3 2
𝑆 = (72 + 36) 2
3
𝑆 = 100 4 𝑖𝑛2

45
Ramalingam Sakthivel Assistant Professor (Senior Grade)
 QUESTIONS AND PROBLEMS RELATED ON THIS MODULE

8. If 𝑼 = 𝒇 𝒙, 𝒚, 𝒛 = −𝒙𝟑 + 𝟑𝒙𝒛 + 𝟐𝒚 − 𝒚𝟐 − 𝟑𝒛𝟐 . Find maximum or minimum values.

Solution:
To find the critical points:
Taking partial derivative of 𝑈 with respect to 𝑥, 𝑦, 𝑎𝑛𝑑 𝑧.
𝑓𝑥 = −3𝑥 2 + 3𝑧 = 0 −−−−−−−−−−−−−−−−−−−−−→ [1]
𝑓𝑦 = 2 − 2𝑦 = 0 −−−−−−−−−−−−−−−−−−−−−−−→ 2
𝑓𝑧 = 3𝑥 − 6𝑧 = 0 −−−−−−−−−−−−−−−−−−−−−−→ 3

From Eq. [2] we get,

2𝑦 = 2 = 𝑦 = 1
From Eq. [1] and Eq. [3] we get,
1 × 2 ==⇒ −6𝑥 2 + 6𝑧 = 0
3 ==⇒ 3𝑥 − 6𝑧 = 0
Solving the above two equations, we get
3𝑥 − 6𝑥 2 = 0
3𝑥 = 6𝑥 2
46
Ramalingam Sakthivel Assistant Professor (Senior Grade)
 QUESTIONS AND PROBLEMS RELATED ON THIS MODULE

1
𝑥=2
Substituting the value of 𝑥 on the Eq. [1], we get
1 2
1 ==⇒ −3 2
+ 3𝑧 =0
3
− 4 + 3𝑧 =0
3 1
3𝑧 = 4 = 𝑧 = 4
∴ The critical points are
1 1
𝑥 = 2, 𝑦 = 1, 𝑧 = 4.

To find the second order derivative:

Construct the Hessian matrix and using the critical points as
𝑓𝑥𝑥 𝑓𝑥𝑦 𝑓𝑥𝑧
𝐻 = 𝑓𝑦𝑥 𝑓𝑦𝑦 𝑓𝑦𝑧
𝑓𝑧𝑥 𝑓𝑧𝑦 𝑓𝑧𝑧
1
𝑓𝑥𝑥 = −6𝑥 = −6 = −3
2
47
Ramalingam Sakthivel Assistant Professor (Senior Grade)
 QUESTIONS AND PROBLEMS RELATED ON THIS MODULE

𝑓𝑦𝑦 = −2
𝐻1 = −3 < 0
𝑓𝑧𝑧 = −6 −3 0
𝐻2 = =6>0
𝑓𝑥𝑦 = 𝑓𝑦𝑥 = 0 0 −2
−3 0 3
𝑓𝑥𝑧 = 𝑓𝑧𝑥 = 3
𝐻3 = 𝐻 = 0 −2 0
𝑓𝑦𝑧 = 𝑓𝑧𝑦 = 0 3 0 −6
= −3 12 + 3 6 = −36 + 18 = −18 < 0
Condition for minimum: (+ve definite) ∴ This is the negative definite.
𝐻1 > 0, 𝐻2 > 0, 𝐻3 > 0 Maximum of 𝑈 = −𝑥 3 + 3𝑥𝑧 + 2𝑦 − 𝑦 2 − 3𝑧 2
Condition for maximum: (-ve definite) 1 3 1 1 1 2
=− +3 + 2 1 −1−3
2 2 4 4
𝐻1 < 0, 𝐻2 > 0, 𝐻3 < 0 1 3 3
= − 8 + 8 + 2 − 1 − 16
2 3
−3 0 3 = 8 + 1 − 16
𝐻 = 0 −2 0 4 3
3 0 −6 = +1−
16 16
1
= 16 + 1
−3 0 3
17
𝐻 = 0 −2 0 = 16
3 0 −6
48
Ramalingam Sakthivel Assistant Professor (Senior Grade)
 QUESTIONS AND PROBLEMS RELATED ON THIS MODULE

9. Find the local minima of the function

𝐕 = 𝒇 𝒙, 𝒚 = 𝒙𝟐 + 𝒚𝟐 − 𝟒𝒙 − 𝟔𝒚 + 𝟏𝟑
Solution:
To find the critical points:
Taking partial derivative of 𝑉 with respect to 𝑥, 𝑎𝑛𝑑 𝑦.
𝑓𝑥 = 2𝑥 − 4 = 0 −−−−−−−−−−−−−−−−−−−−−→ [1]
𝑓𝑦 = 2𝑦 − 6 = 0 −−−−−−−−−−−−−−−−−−−−−→ 2

From Eq. [1] we get,

2𝑥 = 4 ⟹ 𝑥 = 2
From Eq. [2] we get,
2𝑦 = 6 ⟹ 𝑦 = 3

∴ The critical points are

𝑥 = 2, 𝑦 = 3

49
Ramalingam Sakthivel Assistant Professor (Senior Grade)
 QUESTIONS AND PROBLEMS RELATED ON THIS MODULE

To find the second order derivative:

Construct the Hessian matrix and using the critical points as
𝑓𝑥𝑥 𝑓𝑥𝑦
𝐻=
𝑓𝑦𝑥 𝑓𝑦𝑦 Min 𝑉 = 𝑓 2,3
= 22 + 32 − 4 2 − 6 3 + 13
𝑓𝑥𝑥 = 2 = 4 + 9 − 8 − 18 + 13
𝑓𝑦𝑦 = 2 = 26 − 26
=0
𝑓𝑥𝑦 = 𝑓𝑦𝑥 =0
Condition for minimum: (+ve definite)
𝐻1 > 0, 𝐻2 > 0
2 0
𝐻=
0 2
2 0
𝐻 =
0 2
𝐻1 = 2 > 0
𝐻2 = 𝐻 = 4 > 0
Since both conditions are satisfied ( 𝐻1 > 0 𝑎𝑛𝑑 𝐻2 > 0),
the Hessian matrix is positive definite. Therefore, the critical point is
local minimum. 50
Ramalingam Sakthivel Assistant Professor (Senior Grade)
 QUESTIONS AND PROBLEMS RELATED ON THIS MODULE

10. Minimize 𝐙 = 𝟐𝒙𝟐𝟏 + 𝒙𝟐𝟐 + 𝟑𝒙𝟐𝟑 + 𝟏𝟎𝒙𝟏 + 𝟖𝒙𝟐 + 𝟔𝒙𝟑 − 𝟏𝟎𝟎

Subject to Constraints:
𝒙𝟏 + 𝒙𝟐 + 𝒙𝟑 = 𝟐𝟎
𝒙𝟏 , 𝒙𝟐 , 𝒙𝟑 ≥ 𝟎
Solution:
Let 𝑓 𝑥 = 2𝑥12 + 𝑥22 + 3𝑥32 + 10𝑥1 + 8𝑥2 + 6𝑥3 − 100
𝑔 𝑥 = 𝑥1 + 𝑥2 + 𝑥3 − 20
Taking 𝜆 as the Lagrangian multiplier.
Lagrangian function
𝐿 𝑥, 𝜆 = 𝑓 𝑥 − 𝜆𝑔(𝑥)
= 2𝑥12 + 𝑥22 + 3𝑥32 + 10𝑥1 + 8𝑥2 + 6𝑥3 − 100 − 𝜆(𝑥1 + 𝑥2 + 𝑥3 − 20)
The necessary condition for 𝑓(𝑥)
𝜕𝐿 −10 + 𝜆
= 4𝑥1 + 10 − 𝜆 = 0 =⇒ 𝑥1 = −−−−−−−−−−−−→ [1]
𝜕𝑥1 4

51
Ramalingam Sakthivel Assistant Professor (Senior Grade)
 QUESTIONS AND PROBLEMS RELATED ON THIS MODULE

𝜕𝐿 −8 + 𝜆
= 2𝑥2 + 8 − 𝜆 = 0 =⇒ 𝑥2 = −−−−−−−−−−−−→ [2]
𝜕𝑥2 2
𝜕𝐿 −6 + 𝜆
= 6𝑥3 + 6 − 𝜆 = 0 =⇒ 𝑥3 = −−−−−−−−−−−−→ [3]
𝜕𝑥3 6
𝜕𝐿
= − 𝑥1 + 𝑥2 + 𝑥3 − 20 = 0 −−−−−−−−−−−−−−−−−→ [4]
𝜕𝜆

Putting Eqs. [1], [2], [3] in [4], we get

−10 + 𝜆 −8 + 𝜆 −6 + 𝜆
4 ===⇒ + + − 20 = 0
4 2 6
−30 + 3𝜆 −48 + 6𝜆 −12 + 2𝜆
+ + − 20 = 0
12 12 12
−30 + 3𝜆 − 48 + 6𝜆 − 12 + 2𝜆
− 20 = 0
12
11𝜆 − 90
− 20 = 0
12
11𝜆 − 330 = 0
52
Ramalingam Sakthivel Assistant Professor (Senior Grade)
 QUESTIONS AND PROBLEMS RELATED ON THIS MODULE

11𝜆 = 330
330
𝜆=
11
𝜆 = 30
Substituting 𝜆 = 30 on Eqs. [1], [2], [3], we get
−10 + 30
1 ==⇒ 𝑥1 = =5
4
−8 + 30
2 ==⇒ 𝑥2 = = 11
2
−6 + 30
3 ==⇒ 𝑥3 = =4
6

The stationary points are

𝑥1 , 𝑥2 , 𝑥3 = (5,11,4)

53
Ramalingam Sakthivel Assistant Professor (Senior Grade)
 QUESTIONS AND PROBLEMS RELATED ON THIS MODULE

The sufficient condition for 𝑓 𝑥

Bordered Hessian matrix
𝜕2𝑓 𝜕2𝑓 𝜕2𝑓
𝜕𝑥12 𝜕𝑥1 𝜕𝑥2 𝜕𝑥1 𝜕𝑥3
𝐵
𝜕2𝑓 𝜕2𝑓 𝜕2𝑓
𝐻 =
𝜕𝑥2 𝜕𝑥1 𝜕𝑥22 𝜕𝑥2 𝜕𝑥3
𝜕2𝑓 𝜕2𝑓 𝜕2𝑓
𝜕𝑥3 𝜕𝑥1 𝜕𝑥3 𝜕𝑥2 𝜕𝑥32
𝜕2𝑓 𝜕 𝜕𝑓 𝜕
2 = = 4𝑥1 + 10 = 4
𝜕𝑥1 𝜕𝑥1 𝜕𝑥1 𝜕𝑥 1
Similarly,
𝜕2𝑓 𝜕2𝑓 𝜕2𝑓 𝜕2𝑓 𝜕2𝑓 𝜕2𝑓
= = = = = =0
𝜕𝑥1 𝜕𝑥2 𝜕𝑥1 𝜕𝑥3 𝜕𝑥2 𝜕𝑥1 𝜕𝑥2 𝜕𝑥3 𝜕𝑥3 𝜕𝑥1 𝜕𝑥3 𝜕𝑥2
𝜕2𝑓 𝜕2𝑓
= 2, =6
𝜕𝑥22 𝜕𝑥32

54
Ramalingam Sakthivel Assistant Professor (Senior Grade)
 QUESTIONS AND PROBLEMS RELATED ON THIS MODULE

4 0 0
𝐵
𝐻 = 0 2 0
0 0 6
𝐻1𝐵 = 4
𝐻2𝐵 = 8
𝐻3𝐵 = 48
∴ 𝑓(𝑥) is minimum at point (5, 11, 4) as all principal minor are positive.

Min 𝑓(𝑥) = 2(5)2 +112 + 3(4)2 +10 5 + 8 11 + 6 4 − 100

= 50 + 121 + 48 + 50 + 88 + 24 − 100
= 381 − 100
= 281

55
Ramalingam Sakthivel Assistant Professor (Senior Grade)
 QUESTIONS AND PROBLEMS RELATED ON THIS MODULE

11. Maximize Z= 𝟏𝟎𝒙𝟏 + 𝟒𝒙𝟐 − 𝟐𝒙𝟐𝟏 − 𝒙𝟐𝟐

Subject to Constraints:
𝟐𝒙𝟏 + 𝒙𝟐 ≤ 𝟓
𝒙𝟏 , 𝒙𝟐 ≥ 𝟎
Solution:
Let 𝑓 𝑥 = 10𝑥1 + 4𝑥2 − 2𝑥12 − 𝑥22
ℎ(𝑥) = 2𝑥1 + 𝑥2 − 5
Taking 𝜆 as the Lagrangian multiplier.
Lagrangian function
𝐿 𝑥, 𝜆 = 𝑓 𝑥 − 𝜆ℎ(𝑥)
= 10𝑥1 + 4𝑥2 − 2𝑥12 − 𝑥22 − 𝜆(2𝑥1 + 𝑥2 − 5)
The necessary condition for 𝑓(𝑥)
𝜕𝐿
= 0 ⟹ 10 − 4𝑥1 − 2𝜆 = 0 −−−−−−−−−−−−→ [1]
𝜕𝑥1

56
Ramalingam Sakthivel Assistant Professor (Senior Grade)
 QUESTIONS AND PROBLEMS RELATED ON THIS MODULE

𝜕𝐿
= 0 ⟹ 4 − 2𝑥2 − 𝜆 = 0 −−−−−−−−−→ [2]
𝜕𝑥2
Using KKT we’ve,
𝜆ℎ = 0 ===⇒ 𝜆 2𝑥1 + 𝑥2 − 5 = 0 −−−→ [3]

ℎ ≤ 0 ===⇒ 2𝑥1 + 𝑥2 ≤ 5 −−−−−−−−−→ 4

𝑥1 , 𝑥2 ≥ 0 −−−−−−−−−→ 5
𝜆 ≥ 0 −−−−−−−−−→ [6]
From Eq. [3]
===⇒ 𝜆=0 𝑜𝑟 2𝑥1 + 𝑥2 − 5 = 0

Case 1:
Let 𝜆 = 0, then
10 5
[1]===⇒ 10 − 4𝑥1 =0 ====> 𝑥1 = 4
=2
4
[2]===⇒ 4 − 2𝑥2 =0 ====> 𝑥2 = 2 =2
∴ 1 , 2 , 3 , 5 , 6 conditions are satisfied.
57
Ramalingam Sakthivel Assistant Professor (Senior Grade)
 QUESTIONS AND PROBLEMS RELATED ON THIS MODULE

But [4],
5
==⇒ 2 +2≤5
2
7≤5
Which is absurd.
∴ we reject this case.

Case 2:
Let 2𝑥1 +𝑥2 − 5 = 0, then
2𝑥1 +𝑥2 = 5 −−−−−−→ [7]
Now eliminating 𝜆 from [1] and [2],
[1] ===> 10 − 4𝑥1 − 2𝜆 = 0
2 × 2 =⇒ 8 − 4𝑥2 − 2𝜆 = 0

Solving we get, 2 − 4𝑥1 − 4𝑥2 = 0 ===⇒ 2𝑥1 − 2𝑥2 = 1 −−−−−−→ [8]

58
Ramalingam Sakthivel Assistant Professor (Senior Grade)
 QUESTIONS AND PROBLEMS RELATED ON THIS MODULE

Solve [7] and [8] we get

3𝑥2 = 4
4
𝑥2 = 3
4
Substitute 𝑥2 = 3 in [7], we get
4
2𝑥1 + 3 = 5
6𝑥1 + 4 = 15
11
𝑥1 = 6
From Eq. [1], we get
11
10 − 4 − 2𝜆 = 0
6
60 − 44 − 12𝜆 = 0
16
𝜆 = 12
4
𝜆=3
These values satisfying all the 6 conditions.

59
Ramalingam Sakthivel Assistant Professor (Senior Grade)
 QUESTIONS AND PROBLEMS RELATED ON THIS MODULE

Optimal Solution is
11 4
𝑥1 = , 𝑥2 = 3
6

11 4 121 16
∴ Maximize 𝑓 𝑥 = 10 +4 −2 −
6 3 36 9
660 + 192 − 242 − 64
=
36
546
=
36
91
=
6

60
Ramalingam Sakthivel Assistant Professor (Senior Grade)
 QUESTIONS AND PROBLEMS RELATED ON THIS MODULE

11. Maximize 𝒁 = 𝟏𝟎𝒙𝟏 + 𝟏𝟎𝒙𝟐 − 𝒙𝟐𝟏 − 𝒙𝟐𝟐

Subject to Constraints:
𝒙𝟏 +𝒙𝟐 ≤ 𝟖
−𝒙𝟏 + 𝒙𝟐 ≤ 𝟓
𝒙𝟏 , 𝒙𝟐 ≥ 𝟎
Solution:
Let 𝑓 𝑥 = 10𝑥1 + 10𝑥2 − 𝑥12 − 𝑥22
ℎ1 (𝑥) = 𝑥1 + 𝑥2 − 8
ℎ2 𝑥 = −𝑥1 + 𝑥2 − 5
Taking 𝜆1 𝑎𝑛𝑑 𝜆2 as the Lagrangian multiplier.
Lagrangian function
𝐿 𝑥, 𝜆 = 𝑓 𝑥 − 𝜆1 ℎ1 𝑥 − 𝜆2 ℎ2 (𝑥)
= 10𝑥1 + 10 − 𝑥12 − 𝑥22 − 𝜆1 𝑥1 + 𝑥2 − 8 − 𝜆2 (−𝑥1 + 𝑥2 − 5)
The necessary condition for 𝑓(𝑥)
𝜕𝐿
= 0 ⟹ 10 − 2𝑥1 − 𝜆1 + 𝜆2 = 0 −−−−−−−−−−−−→ [1]
𝜕𝑥1
61
Ramalingam Sakthivel Assistant Professor (Senior Grade)
 QUESTIONS AND PROBLEMS RELATED ON THIS MODULE

𝜕𝐿
= 0 ⟹ 10 − 2𝑥2 − 𝜆1 − 𝜆2 = 0 −−−−→ [2]
𝜕𝑥2
Using KKT we’ve,
𝜆1 ℎ1 = 0 ⟹ 𝜆1 𝑥1 + 𝑥2 − 8 = 0 −−−→ [3]
𝜆2 ℎ2 = 0 ⟹ 𝜆2 −𝑥1 + 𝑥2 − 5 = 0 −−−→ [4]
ℎ1 ≤ 0 ⟹ 𝑥1 + 𝑥2 − 8 ≤ 0 −−−−−−−−→ 5
ℎ2 ≤ 0 ⟹ −𝑥1 + 𝑥2 − 5 ≤ 0 −−−−−−−→ 6
𝑥1 , 𝑥2 , 𝜆1 , 𝜆2 ≥ 0 −−−−→ 7

Case 1:
Let 𝜆1 = 0, 𝜆2 = 0 then
10
[1] ⟹ 10 − 2𝑥1 = 0 ⟹ 𝑥1 = =5
2
10
[2]⟹ 10 − 2𝑥2 = 0 ⟹ 𝑥2 = =5
2
Since Eqn. [5] is not satisfied.
∴ We reject it.

62
Ramalingam Sakthivel Assistant Professor (Senior Grade)
 QUESTIONS AND PROBLEMS RELATED ON THIS MODULE

Case 2: Substitute, 𝑥2 = 7.5 in [10] we get

Let 𝜆1 = 0, 𝜆2 ≠ 0 then −𝑥1 + 7.5 − 5 = 0
[1] ⟹ 10 − 2𝑥1 + 𝜆2 = 0 −−−−−−−→ [8] 𝑥1 = 2.5
[2] ⟹ 10 − 2𝑥2 − 𝜆2 = 0 −−−−−−−→ [9] Substitute 𝑥1 in [8] we get
From Eqn. [4], since 𝜆2 ≠ 0 , 10 − 2 2.5 + 𝜆2 = 0
−𝑥1 + 𝑥2 − 5 = 0 −−−−−−→ [10] 𝜆2 = −10 + 5 = −5
Add Eqn. [8] and Eqn. [9] We get, 𝑥1 = 2.5, 𝑥2 = 7.5 and 𝜆2 = −5
10 − 2𝑥1 + 𝜆2 = 0
10 − 2𝑥2 − 𝜆2 = 0 Since Eqn. [5] is not satisfied.
20 − 2𝑥1 − 2𝑥2 = 0 −−−−−→ 11 ∴ We reject it.
Now, add
[10] × 2 −2𝑥1 + 2𝑥2 = 10
[11] 2𝑥1 + 2𝑥2 = 20
4𝑥2 = 30
30
𝑥2 = = 7.5
4

63
Ramalingam Sakthivel Assistant Professor (Senior Grade)
 QUESTIONS AND PROBLEMS RELATED ON THIS MODULE

Case 3: Substitute, 𝑥2 = 4 in [14] we get

Let 𝜆1 ≠ 0, 𝜆2 = 0 then 𝑥1 + 4 − 8 = 0
[1] ⟹ 10 − 2𝑥1 − 𝜆1 = 0 −−−−−−−→ [12] 𝑥1 = 4
[2] ⟹ 10 − 2𝑥2 − 𝜆1 = 0 −−−−−−−→ [13] Substitute 𝑥1 in [12] we get
From Eqn. [3], since 𝜆1 ≠ 0 , 10 − 2 4 − 𝜆1 = 0
𝑥1 + 𝑥2 − 8 = 0 −−−−−−→ [14] 𝜆1 = 10 − 8 = 2
Subtract Eqn. [12] and Eqn. [13] We get, 𝑥1 = 4, 𝑥2 = 4 and 𝜆1 = 2
10 − 2𝑥1 − 𝜆1 = 0
−10 + 2𝑥2 + 𝜆1 = 0 [5], [6] and [7] are satisfied
−2𝑥1 + 2𝑥2 = 0 −−−−−→ 15 ∴ All the conditions are satisfied.
Now, add
[14] × 2 2𝑥1 + 2𝑥2 = 16 Max 𝑍 = 10𝑥1 + 10𝑥2 − 𝑥12 − 𝑥22
= 10 4 + 10 4 − 4 2 − 4 2
[15] −2𝑥1 + 2𝑥2 = 0
4𝑥2 = 16 = 40 + 40 − 16 − 16
𝑥2 = 4 = 80 − 32
Max 𝑍 = 48
64
Ramalingam Sakthivel Assistant Professor (Senior Grade)
 QUESTIONS AND PROBLEMS RELATED ON THIS MODULE

Case 4: Put 𝑥1 and 𝑥2 in Eqn. [1] and Eqn. [2]

Let 𝜆1 ≠ 0, 𝜆2 ≠ 0 then −𝜆1 + 𝜆2 = −7
From Eqn. [3] and [4], −𝜆1 − 𝜆2 = 3
𝑥1 + 𝑥2 − 8 = 0 Solve the above two equations, we get
−𝑥1 + 𝑥2 − 5 = 0 𝜆1 = 2
2𝑥2 − 13 = 0 𝜆2 = −5
2𝑥2 = 13 Therefore, Eqn. [7] is not satisfied.
𝑥2 = 6.5 We reject it.

Substitute 𝑥2 = 6.5 in Eqn. [3], we get

𝑥1 + 6.5 − 8 = 0
𝑥1 = 1.5
[5] and [6] are satisfied.

65
Ramalingam Sakthivel Assistant Professor (Senior Grade)