Contents

Study Guide for the topic: System of Linear Equations

Subject: Linear Algebra and Numerical Analysis
Module 2.2

Table of Contents:
1. Mind Map Overview
2. Introduction & Learning Objectives
• 2.1 Audio Overview
3. Revision Notes: Solving the Puzzles - Systems of Linear Equations
4. Practice Questions
5. Answer Keys & Explanations
6. Glossary of Key Terms

1. Mind Map Overview

For a visual summary of this module’s key concepts and their connections, please
refer to the mind map linked here:
Mind Map

2. Introduction & Learning Objectives

Hey Future AI & Data Mavericks! Let’s Solve Some Puzzles!
Welcome back! In Module 1, we learned the basic language of data – Scalars,
Vectors, and Matrices. Now, we’ll see how these tools help us tackle a funda-
mental problem that pops up everywhere, from ancient trade puzzles to modern
AI algorithms: solving systems of linear equations. Think of it as finding
the one specific answer (or answers!) that makes multiple conditions true simul-
taneously. Mastering this process is a core skill for many analytical tasks you’ll
encounter. (You can also get a visual summary from the Mind Map linked in
Section 1).
Learning Objectives for this Module:
• Understand what a system of linear equations represents, both alge-
braically and geometrically.
• Learn how to represent these systems efficiently using Augmented Ma-
trices.
• Master the steps of Gaussian Elimination (using Elementary Row Op-
erations) to systematically solve these systems.

1
 • Recognize what Row-Echelon Form (REF) looks like and why it’s
helpful.
• Understand how to use Back-Substitution to find the final answer.
• Identify the three possible outcomes for a system (Unique, Infinite, No
Solution) and how they appear during elimination.
• Appreciate why understanding this process is vital, even when computers
automate the solution, especially for AI/Data Science roles.
2.1 Audio Overview
Here is an audio overview on Systems of Linear Equations:
Gaussian_Elimination_Uncovered

3. Revision Notes: Solving the Puzzles - Systems of Linear Equations

3.1 What’s a System of Linear Equations? (The Puzzle)
Remember solving equations like 2x + 3 = 7? That’s one linear equation. A
system just means you have two or more linear equations that you need
to solve together.
• Linear Means Simple: The variables (like x, y, z) are only to the first
power (no x^2, y^3), not in denominators, and not multiplied together
(xy).
• Simultaneous Solution: You’re looking for a set of values for the vari-
ables (e.g., a specific value for x, y, and z) that makes all the equations
in the system true at the same time.
• Geometric Picture:
– For two variables (x, y), each equation is a line. The solution is the
single point where the lines intersect.
– For three variables (x, y, z), each equation is a plane. The solution
could be:
∗ A single point where all planes meet.
∗ A line where all planes meet.
∗ A whole plane (if all equations described the same plane).
∗ No common intersection point at all!
Example (The Shopping Puzzle Revisited):
Find x (bags) and y (boxes).
2x + 6y = 20 (Apple equation)
4x + 8y = 28 (Banana equation)
The solution (x=1, y=3) is the only pair of values that satisfies both equations.
3.2 Organizing the Puzzle: The Augmented Matrix

2
Writing out systems like the one above is okay for 2 equations, but imagine 10
equations with 10 variables! We need a neater way. That’s where matrices come
in.
• The Augmented Matrix [A | B]: We capture all the essential numbers
(coefficients and constants) in a grid.
1. Line up your variables (all x’s in column 1, y’s in column 2, etc.). If
a variable is missing, its coefficient is 0.
2. Write down the coefficients of the variables to form the Coefficient
Matrix (A).
3. Draw a vertical line.
4. Write down the constants from the right side of the equations to form
the Constant Vector (B).
• Example (Shopping Puzzle): System: 2x + 6y = 20 4x + 8y = 28
Augmented Matrix:
[ [ 2 6 | 20 ]
[ 4 8 | 28 ] ]
• Analogy: Think of it like a recipe ingredients list combined with the final
desired quantity. The matrix A lists the “ingredients” (coefficients), the
line separates it from the “desired outcome” B (constants).
3.3 Solving Systematically: Gaussian Elimination (The Master
Recipe!)
This is the classic, step-by-step algorithm to solve any system. It mimics the
“elimination” you did in school but uses matrix notation to keep things organized.
The goal is to simplify the matrix into a form where the solution is easy to see.
The Steps (The Recipe):
1. Write the Augmented Matrix [A | B]: (As done above).
2. Use Elementary Row Operations (The Legal Moves): You can ma-
nipulate the rows of the matrix using three specific operations. These are
“legal” because they change how the equations look but do not change
the actual solution of the system.
• Swap: Interchange any two rows (Ri <-> Rj). (Like reordering your
recipe steps if needed).
• Scale: Multiply all numbers in a row by the same non-zero constant
(c * Ri, where c != 0). (Like doubling an entire recipe).
• Add Multiple: Add a multiple of one row to another row (Ri ->
Ri + c * Rj). (This is the core elimination step – like adding ingre-
dients together in a specific ratio).
3. Reach Row-Echelon Form (REF) (The Simplified State): Keep
applying row operations with the goal of getting the matrix into this spe-
cific “staircase” pattern:
• Any rows containing only zeros are at the bottom.

3
 • The first non-zero number (called the leading entry or pivot) in
any non-zero row is strictly to the right of the leading entry of the
row above it.
• All entries in a column below a leading entry are zero.
• Visual: It looks like a triangle of zeros in the bottom-left corner of
the coefficient part A.
[ [ Pivot ... ... | ... ]
[ 0 Pivot ... | ... ]
[ 0 0 Pivot | ... ]
[ 0 0 0 | ... ] ]
(where Pivot is a non-zero number, Pivot != 0)
• Strategy: Usually, you work column by column from the left. Get a
leading entry (often a ‘1’ by scaling) in the top row, then use the “Add
Multiple” operation (Ri -> Ri + c * Rj) to create zeros below it.
Then move to the next row/column and repeat for the submatrix
below and to the right.
4. Use Back-Substitution (Getting the Answer):
• Convert the simplified REF matrix back into equations.
• The last non-zero equation will usually have only one variable, which
you can solve easily.
• Substitute this value back into the equation above it to solve for the
next variable.
• Continue substituting values upward until all variables are found.
(Optional: Reduced Row-Echelon Form - RREF): You can continue row
operations to make each leading entry a ‘1’ AND make all other entries in the
pivot columns zero (both above and below). This often makes the final solution
even easier to read directly from the matrix. Calculators often compute RREF.
3.4 What Can Happen? (Unique, Infinite, or No Solution)
When you perform Gaussian Elimination, the final REF matrix tells you exactly
what kind of solution your system has:
1. Unique Solution: The “normal” case. You end up with a nice stair-
case where the number of non-zero rows (the rank) equals the number of
variables. Back-substitution yields exactly one value for each variable.
• REF Look: Clean triangular form, no contradictions. Example:
[ [ 1 2 3 | 4 ]
[ 0 1 5 | 6 ]
[ 0 0 1 | 7 ] ]
(Ready for back-substitution)
• Geometric View: Lines/planes intersect at a single point.
2. Infinite Solutions: This happens when the system has dependent equa-
tions (one equation gives the same information as others).
• REF Look: You’ll get a row of all zeros ([0 0 ... 0 | 0]). This
0=0 row is true but provides no new information, meaning you have
fewer independent constraints (pivots) than variables. This leads to

4
 free variables – variables whose values can be chosen freely, deter-
mining the values of others.
• Geometric View: Lines are identical (coincident), or planes intersect
along a common line.
3. No Solution (Inconsistent): This happens when equations contradict
each other.
• REF Look: You’ll get a row that looks like [0 0 ... 0 | c] where
c is a non-zero number (c != 0). This corresponds to the equation
0 = c, which is impossible/contradictory.
• Geometric View: Lines are parallel but distinct, or planes have no
common intersection point.
Crucial Point: Recognizing these patterns in the final matrix form is essential
for correctly interpreting the result!
3.5 Why Learn This If Computers Do It? (Relevance & Automation)
It’s true! For any real-world sized problem, computers solve these systems
using highly optimized algorithms (often based on Gaussian Elimination or
related methods like LU Decomposition). Tools like Python with NumPy
([Link] or finding the inverse A^-1) or MATLAB handle this
instantly.
So why learn the manual process?
1. Understand the “Black Box”: Knowing Gaussian Elimination helps
you understand what the computer is fundamentally doing. This allows
you to:
• Interpret Results: Recognize if the output indicates infinite or no
solutions, which a simple numerical answer might obscure.
• Troubleshoot: Understand potential sources of error if the tool
gives strange results (e.g., issues related to near-zero pivots or ill-
conditioning).
• Choose Methods Wisely: Know when different methods (elimi-
nation vs. inverse) might be appropriate or have limitations.
2. Problem Formulation: Often, the hardest part in AI/Data Science isn’t
the calculation, but setting up the problem correctly. Many problems
reduce to solving a system of linear equations. You need to be able to
model a real-world scenario (like resource constraints or fitting a model to
data) as a system (AX=B) before you can even use the tool to solve it.
3. Foundation for Advanced Topics: Gaussian Elimination is the ba-
sis for understanding concepts like matrix rank, determinants, inverses,
and more advanced techniques like LU decomposition used in numerical
analysis and algorithm design.
3.6 Deep Dive: Systems in Your Future Job Roles
Let’s connect solving systems (AX=B) directly to tasks you might perform:
• Role: AI-Enabled Business Analyst

5
 – Scenario: A company wants to allocate its advertising budget (say,
�1 Crore) across three channels (Online Ads, TV Ads, Print Ads) to
achieve specific targets for reach (e.g., 5 Million people) and conver-
sions (e.g., 10,000 sales). Each channel has a different cost per reach
and conversion rate.
– LA Use: You can model this as a system of linear equations. Let x,
y, z be the amount spent on each channel.
∗ x + y + z = 1,00,00,000 (Budget Constraint)
∗ (ReachRate_X) * x + (ReachRate_Y) * y + (ReachRate_Z)
* z = 50,00,000 (Reach Target)
∗ (ConvRate_X) * x + (ConvRate_Y) * y + (ConvRate_Z) *
z = 10,000 (Conversion Target)
– Your Role: You’d define the coefficients (rates) in matrix A, the
unknowns x, y, z in vector X, and targets in vector B. You’d use a
tool (Excel Solver, Python script) to solve AX=B. Understanding the
setup and potential outcomes (maybe no combination meets targets
= No Solution, or maybe multiple combinations work = Infinite So-
lutions) is critical for providing actionable business insights, not just
a number.
• Role: Junior Data Scientist / AI-Enabled Data Analyst
– Scenario: Performing Linear Regression to predict house prices
based on size (sq. ft.) and number of bedrooms. You have data for
many houses.
– LA Use: Simple linear regression tries to find the best coefficients
(w0, w1, w2) for the equation: Price ~ w0 + w1 * Size + w2 *
Bedrooms. Finding the “best” coefficients often involves minimizing
prediction errors, which mathematically leads to solving a system of
linear equations called the Normal Equations: (X^T X) w = X^T
y. Here:
∗ X is a matrix derived from your input data (size, bedrooms).
∗ y is the vector of known house prices.
∗ w is the vector of unknown coefficients you want to find.
– Your Role: You’ll use libraries (like Scikit-learn) that solve this sys-
tem internally. Understanding that a linear system is being solved
helps you grasp how the model parameters are determined. It also
helps understand potential problems: if ‘size’ and ‘number of bed-
rooms’ are highly correlated (almost linearly dependent), the matrix
X^T X might be nearly singular (det(X^T X) ~ 0), making the solu-
tion numerically unstable or unreliable – something foundational LA
knowledge helps you anticipate.
• Role: Online Researcher / General Assistant
– Scenario: Validating financial data in a report where several line
items must sum up to reported totals across different categories.
– LA Use: Each summing rule is a linear equation (e.g., CategoryA +
CategoryB = Total1, Dept1_A + Dept2_A = CategoryA). You can
represent all these consistency rules as a system AX=B.

6
 – Your Role: Setting this up and using a tool to check for consistency
(does a solution exist? Does it match the reported numbers?) is much
faster than manual checks. If the system shows “No Solution” (a 0 =
non-zero contradiction during elimination), it flags an error in the
report automatically.
3.7 Module 2.2 Key Takeaways:
• Systems of linear equations represent problems with multiple conditions
and unknowns.
• Augmented matrices [A | B] provide a compact way to represent these
systems.
• Gaussian Elimination (using Row Operations: Swap, Scale, Add Multiple)
systematically transforms the matrix into Row-Echelon Form (REF).
• Back-substitution solves the system from the simplified REF.
• The final REF reveals if there’s a Unique Solution, Infinite Solutions (row
0=0), or No Solution (row 0=c, c != 0).
• Understanding this process is crucial for interpreting results from compu-
tational tools (like Python/NumPy) used heavily in AI/DS roles for tasks
like optimization, model fitting (regression), and data validation.

4. Practice Questions
4.1 Subjective Questions (Short & Long Answer)
(Using the 10 Subjective Questions created specifically for System of Linear
Equations)
(Short Answer Questions - Approx. 2-4 Marks Each)
Q1. Define what a “system of linear equations” is and explain what constitutes
a “solution” to such a system. Briefly mention the geometric interpretation of
a solution for a system of two linear equations in two variables.
Q2. Explain how to construct the augmented matrix [A | B] for a given
system of linear equations. Clearly state what the matrix A and the vector B
represent in this notation.
Q3. List the three types of Elementary Row Operations (EROs) that can
be performed on an augmented matrix during Gaussian Elimination. Why are
these specific operations considered “legal” or valid for solving the system?
Q4. Describe two key characteristics of a matrix that is in Row-Echelon Form
(REF). What is the main advantage of transforming an augmented matrix into
REF?
Q5. Explain the process of back-substitution. At what stage of solving a
linear system using Gaussian elimination is it applied, and what does it achieve?
(Medium Answer Questions - Approx. 5-7 Marks Each)

7
Q6. If, during Gaussian elimination, performing row operations on an aug-
mented matrix leads to a row like [0 0 0 | 7], what conclusion can you draw
about the original system of linear equations? Explain why this row leads to
that conclusion.
Q7. Consider a system of linear equations in three variables (x, y, z). If reducing
its augmented matrix to Row-Echelon Form results in only two non-zero rows
(e.g., the last row is [0 0 0 | 0]) and there are no contradictions, what does
this imply about the number of solutions? Briefly explain why, and state the
term used for variables not corresponding to leading entries (pivots).
Q8. Modern software (like Python’s NumPy library) can solve systems of linear
equations instantly using functions like [Link]. Given this, provide two
distinct reasons why it is still important for students in an Applied AI & Data
Science program to understand the process of Gaussian elimination.
(Longer Answer Questions - Approx. 8-10 Marks)
Q9. Imagine you are an AI-enabled Business Analyst trying to determine
the optimal mix of two products (Product X, Product Y) to manufacture, given
constraints on machine time (e.g., Machine M1, Machine M2) and raw materials
(e.g., Material R). Describe how you could potentially model this problem using
a system of linear equations. Explain what finding a “unique solution”, “no
solution”, or “infinite solutions” would mean in the context of this business
problem.
Q10. Summarize the overall strategy and key stages involved in solving a system
of linear equations using Gaussian elimination followed by back-substitution.
Start from the initial set of equations and describe the transformation process
and how the final answer (or solution type) is obtained.
4.2 Multiple Choice Questions (MCQs)
(Using the 10 MCQs created specifically for System of Linear Equations)
Q1. Consider the system of linear equations: 3x - y = 5 x + 2y = 4 Which
is the correct augmented matrix [A | B] representation for this system? (a)
[[31|5]
[12|4]]
(b)
[ [ 3 -1 | 5 ]
[12|4]]
©
[ [ 5 -1 | 3 ]
[42|1]]
(d)
[ [ 3 -1 ]
[ 1 2 ] ]‘
Q2. What is the primary purpose of applying Elementary Row Operations

8
(swapping rows, scaling a row, adding a multiple of one row to another) during
Gaussian Elimination? (a) To change the system into one with a different so-
lution. (b) To calculate the determinant of the matrix directly. © To simplify
the system into an equivalent system (with the same solution) that is easier to
solve. (d) To increase the number of variables in the system.
Q3. A matrix is said to be in Row-Echelon Form (REF) if it satisfies several
conditions. Which of these is a key characteristic of REF? (a) All entries in the
matrix must be either 0 or 1. (b) The first non-zero entry (leading entry) in
each row is 1. © All entries below a leading entry (pivot) in the same column
are zero. (d) The matrix must be square.
Q4. In the process of solving a linear system using Gaussian Elimination, once
the augmented matrix is transformed into Row-Echelon Form (REF), what is
the typical next step to find the values of the variables? (a) Calculate the
determinant of the REF matrix. (b) Perform column operations to get Reduced
Row-Echelon Form. © Use back-substitution, starting from the last non-zero
equation. (d) Find the inverse of the original coefficient matrix.
Q5. If, after performing Gaussian Elimination on an augmented matrix, one
of the rows becomes [0 0 0 | 5], what can you conclude about the system of
equations? (a) The system has a unique solution. (b) The system has infinitely
many solutions. © The system has no solution (it is inconsistent). (d) The
calculation must have contained an error.
Q6. Consider a system of 3 linear equations with 3 variables (x, y, z). After
reduction to Row-Echelon Form, the augmented matrix looks like:
[[123|4]
[015|6]
[000|0]]
What does this imply about the solution? (a) Unique solution. (b) Infinitely
many solutions (one free variable). © No solution. (d) Only the trivial solution
(x=y=z=0).
Q7. For a system AX=B where A is a 3 x 3 coefficient matrix, reaching a
Row-Echelon Form where the number of non-zero rows (rank) is 3 generally
indicates that the system has: (a) No solution. (b) Exactly one unique solution.
© Infinitely many solutions. (d) Exactly three solutions.
Q8. What is the primary advantage of using computational tools like Python
with NumPy (e.g., [Link]) to solve large systems of linear equa-
tions in data science compared to doing it manually? (a) Computers understand
the underlying mathematical theory better. (b) They can handle systems with
thousands or millions of variables efficiently and quickly. © They always find
a solution, even if one doesn’t exist mathematically. (d) They automatically
format the results into visually appealing graphs.
Q9. In Machine Learning, finding the optimal parameters for a Linear Regres-
sion model often mathematically reduces to: (a) Calculating the determinant of

9
the dataset. (b) Performing repeated matrix multiplication until the numbers
converge. © Solving a system of linear equations (often the ‘Normal Equations’).
(d) Finding the eigenvalues of the data matrix.
Q10. If you use software to solve AX=B and it reports that matrix A is “singular”
(or “not invertible”), what should you, as a data analyst, conclude? (a) The
system definitely has no solution. (b) The system definitely has infinitely many
solutions. © The software must be incorrect. (d) The system does not have a
unique solution, but could have either no solutions or infinitely many solutions.

5. Answer Keys & Explanations

5.1 Subjective Questions - Evaluator’s Guide & Model Answers
(Using the detailed Evaluator Key provided specifically for the System of Linear
Equations subjective questions)
Q1. Define System, Solution & Geometry (Short Answer - e.g., 4
Marks)
• Keywords: System, Linear equations, Simultaneous, Solution, Satisfies
all equations, Variables, Values, Geometric, Intersection point, Lines.
• (See detailed key provided previously for marking guidelines)
Q2. Constructing Augmented Matrix (Short Answer - e.g., 3 Marks)
• Keywords: Augmented matrix, Coefficient matrix (A), Constant
vector (B), [A | B], Standard form, Align variables, Rows=Equations,
Columns=Variables.
• (See detailed key provided previously for marking guidelines)
Q3. Elementary Row Operations & Validity (Short-Medium Answer
- e.g., 4 Marks)
• Keywords: Elementary Row Operations (EROs), Swap rows (Ri <->
Rj), Scale row (c * Ri, c != 0), Add multiple of row (Ri -> Ri + c *
Rj), Equivalent system, Same solution set, Valid algebraic manipulation.
• (See detailed key provided previously for marking guidelines)
Q4. Row-Echelon Form Characteristics & Goal (Medium Answer -
e.g., 5 Marks)
• Keywords: Row-Echelon Form (REF), Zero rows at bottom, Leading en-
try (pivot), Staircase pattern, Zeros below pivots, Simplify system, Easier
to solve, Back-substitution.
• (See detailed key provided previously for marking guidelines)
Q5. Back-Substitution Explanation (Short Answer - e.g., 3 Marks)
• Keywords: Back-substitution, After Row-Echelon Form (REF), Solve
bottom-up, Substitute values upwards, Find variable values.

10
 • (See detailed key provided previously for marking guidelines)
Q6. Interpretation of Contradictory Row (Short Answer - e.g., 3
Marks)
• Keywords: No solution, Inconsistent system, Contradiction, 0 =
non-zero.
• (See detailed key provided previously for marking guidelines)
Q7. Interpretation of All-Zero Row / Free Variables (Medium Answer
- e.g., 5 Marks)
• Keywords: Infinite solutions, Dependent system, Row of zeros (0=0),
Fewer pivots/equations than variables, Free variables.
• (See detailed key provided previously for marking guidelines)
Q8. Importance of Understanding Gaussian Elimination Despite Au-
tomation (Medium Answer - e.g., 6 Marks)
• Keywords: Interpret output, Solution types (unique, infinite, none),
Troubleshoot errors, Numerical issues, Problem formulation, Set up sys-
tem (AX=B), Foundational understanding, Algorithm basis.
• (See detailed key provided previously for marking guidelines)
Q9. Relevance for Business Analyst (Resource Allocation) (Medium-
Long Answer - e.g., 8 Marks)
• Keywords: Model, Linear equations/constraints, Variables (product
units), Coefficients (resource usage), Constants (resource limits), Unique
solution (specific plan), No solution (impossible constraints/targets),
Infinite solutions (flexibility/redundancy), Interpretation in context.
• (See detailed key provided previously for marking guidelines)
Q10. Summary of Gaussian Elimination Strategy (Medium Answer -
e.g., 6 Marks)
• Keywords: System of equations, Augmented matrix, Elementary Row
Operations, Row-Echelon Form (REF), Check consistency (solution type),
Back-substitution, Find variable values.
5.2 MCQ Answer Key with Explanations
• Q1. Correct Answer: (b)
– Explanation: The augmented matrix takes coefficients from the
variables (lined up: x then y) and the constants. The first equation
3x - 1y = 5 gives row [3 -1 | 5]. The second equation 1x + 2y
= 4 gives row [1 2 | 4].
• Q2. Correct Answer: ©
– Explanation: Elementary row operations (swapping, scaling,
adding a multiple of one row to another) are designed to transform
the system into a simpler, equivalent form (one with the same

11
 solution set) that is easier to solve, typically via back-substitution.
They don’t change the fundamental solution.
• Q3. Correct Answer: ©
– Explanation: A key defining property of Row-Echelon Form is that
all entries below a leading entry (the first non-zero number in a row)
must be zero. While leading entries are often made ‘1’ for RREF
(b), it’s not required for REF. Not all entries must be 0 or 1 (a), and
the matrix doesn’t have to be square (d) (though the coefficient part
often is).
• Q4. Correct Answer: ©
– Explanation: Once the matrix is in REF (“staircase” form), the sys-
tem is simplified. The last equation typically involves only one vari-
able, the second-to-last involves two, etc. Back-substitution solves
the last equation and substitutes the result upwards to find the other
variables efficiently.
• Q5. Correct Answer: ©
– Explanation: The row [0 0 0 | 5] translates back to the equation
0x + 0y + 0z = 5, which simplifies to 0 = 5. This is a contradic-
tion, meaning there is no set of (x, y, z) values that can satisfy this
impossible condition along with the others. Therefore, the system
has no solution (is inconsistent).
• Q6. Correct Answer: (b)
– Explanation: The matrix is in REF. The last row [0 0 0 | 0]
translates to 0 = 0, which is true but gives no information. There
are 3 variables (x, y, z) but only 2 leading entries (pivots) in the non-
zero rows (the ‘1’ in row 1 and the ‘1’ in row 2). Since there are fewer
pivots than variables, there will be at least one free variable (in this
case, z can be chosen freely), leading to infinitely many solutions.
• Q7. Correct Answer: (b)
– Explanation: For a square system (n x n, here 3 x 3), if the REF
has n non-zero rows (meaning rank = n, the number of variables),
and there are no contradiction rows (like 0 = c), then each variable
corresponds to a pivot, and back-substitution will yield exactly one
value for each variable, resulting in a unique solution.
• Q8. Correct Answer: (b)
– Explanation: Manual Gaussian elimination becomes incredibly
time-consuming and error-prone for systems larger than perhaps 3
x 3 or 4 x 4. Computers, using optimized linear algebra libraries
like NumPy, can solve systems with potentially millions of variables
extremely quickly and accurately, which is essential for real-world
data science problems.
• Q9. Correct Answer: ©
– Explanation: As mentioned in the notes, Linear Regression aims to
find the best-fitting line (or plane/hyperplane). The mathematical
process to determine the coefficients of this line/plane that minimize
the error often involves setting up and solving a system of linear

12
 equations known as the Normal Equations, typically represented as
(X^T X) w = X^T y.
• Q10. Correct Answer: (d)
– Explanation: A singular matrix (det(A)=0) means the matrix is
not invertible, which rules out finding a unique solution using the in-
verse method (X = A^-1 B). However, as seen in the study of solution
types, a system with a singular coefficient matrix can still have either
infinitely many solutions (if consistent, e.g., REF has 0=0 rows) or no
solution (if inconsistent, e.g., REF has 0=c row, c != 0). Therefore,
singularity only tells us there isn’t a unique solution.

6. Glossary of Key Terms

(Terms central to Module 2.2: System of Linear Equations)
• Augmented Matrix: A matrix representing a system of linear equations,
formed by combining the coefficient matrix (A) and the constant vector
(B), usually separated by a vertical line: [A | B].
• Back-substitution: The process of solving a system of linear equations
in row-echelon form by starting with the last equation and substituting
the found variable values back into the equations above it.
• Coefficient Matrix: The matrix (A in AX=B) containing the coefficients
of the variables in a system of linear equations.
• Consistent System: A system of linear equations that has at least one
solution (either unique or infinitely many).
• Constant Vector: The vector (B in AX=B) containing the constant terms
(right-hand side values) of a system of linear equations.
• Dependent System: A system of linear equations that has infinitely
many solutions (typically occurs when equations are redundant, leading
to fewer pivots than variables).
• Elementary Row Operations (EROs): Three basic operations used
in Gaussian elimination that do not change the solution set of the system:
(1) Swapping two rows (Ri <-> Rj), (2) Multiplying a row by a non-zero
scalar (c * Ri, c != 0), (3) Adding a multiple of one row to another row
(Ri -> Ri + c * Rj).
• Free Variable: In a system with infinitely many solutions, a variable
that does not correspond to a pivot (leading entry) in the row-echelon
form; its value can be chosen arbitrarily.
• Gaussian Elimination: A systematic algorithm for solving systems of
linear equations by using elementary row operations to transform the aug-
mented matrix into row-echelon form.
• Inconsistent System: A system of linear equations that has no solution
(indicated by a contradictory row like [0 0 ... 0 | c] where c != 0 in
REF).
• Leading Entry (Pivot): The first non-zero entry in a row of a matrix
in row-echelon form.

13
• Linear Equation: An equation where variables appear only to the first
power and are not multiplied together.
• Rank (of a matrix): The number of leading entries (pivots) or non-
zero rows in the row-echelon form of a matrix. Related to the number of
independent equations/variables.
• Row-Echelon Form (REF): A simplified form of a matrix achieved
through row operations, characterized by a staircase pattern of leading
entries with zeros below each leading entry.
• Solution (to a System): A set of values for all variables that simulta-
neously satisfies every equation in the system.
• System of Linear Equations: A collection of two or more linear equa-
tions involving the same set of variables.
• Unique Solution: A system has exactly one set of values for the variables
that satisfies all equations.

14