Comprehensive Study Hand Note: STT163 Linear Algebra

(2018–2024 Exam Prep)

This document serves as a rigorous pedagogical guide for students preparing for the STT163 Linear Algebra
examinations. It synthesizes core theoretical principles with the specific computational techniques emphasized in
recent exam cycles (2018–2024).

1. Fundamental Matrix Definitions and Properties

Core Definitions
• Matrix: A rectangular array of elements (scalars) arranged in m rows and n columns.
• Non-singular Matrix: A square matrix A where the determinant is non-zero (|𝐴| ≠ 0). Theoretical Insight: Non-
singularity is the necessary and sufficient condition for a matrix to possess an inverse.
• Orthogonal Matrix: A square matrix A such that 𝐴𝐴𝑇 = 𝐼. This implies 𝐴𝑇 = 𝐴−1 .
• Symmetric Matrix: A matrix A such that 𝐴 = 𝐴𝑇 .
2 1
◦ Example: 𝐴 = [ ] is symmetric because 𝑎12 = 𝑎21 = 1.
1 3

Quick Reference: Transpose and Symmetry Properties

Property Mathematical Expression / Proof
Product Transpose (𝐴𝐵)𝑇 = 𝐵𝑇 𝐴𝑇
Inverse Transpose (𝐴𝑇 )−1 = (𝐴−1 )𝑇
For any square matrix A, 𝐶 = 𝐴 + 𝐴𝑇 is symmetric.
Sum Symmetry 𝑇
Proof: 𝐶 𝑇 = (𝐴 + 𝐴𝑇 )𝑇 = 𝐴𝑇 + 𝐴𝑇 = 𝐴𝑇 + 𝐴 = 𝐶
Power Symmetry If A is symmetric, 𝐴𝑛 is symmetric for n > 0: (𝐴𝑛 )𝑇 = 𝐴𝑛

Professor’s Proof: Symmetry of Aⁿ

To prove (Aⁿ)ᵀ = Aⁿ for a symmetric matrix A:
1. Base Case: For n = 1, 𝐴𝑇 = 𝐴 (given).
𝑇
2. Inductive Step: Assume (𝐴𝑘 ) = 𝐴𝑘
𝑇 𝑇 𝑇
3. Expansion: (𝐴𝑘+1 ) = (𝐴𝑘 · 𝐴) = 𝐴𝑇 · (𝐴𝑘 ) (by product property).
𝑇
4. Substitution: Since 𝐴𝑇 = 𝐴 and (𝐴𝑘 ) = 𝐴𝑘 , we have 𝐴 · 𝐴𝑘 = 𝐴𝑘+1. Thus, 𝐴𝑘+1 is symmetric.

2. Determinants, Minors, and Adjoint Identities

Minors and Cofactors: Symbolic Illustration
To find the determinant of a matrix of order 3 or higher, we utilize Minors (Mij) and Cofactors (Aij). Given:
𝑎11 𝑎12 𝑎13
𝐴 = [𝑎21 𝑎22 𝑎23 ]
𝑎31 𝑎32 𝑎33
𝑎22 𝑎23
• Minor M11: Delete row 1 and column 1 ⇒ 𝑀11 = |𝑎 𝑎33 |
32
• Cofactor Aij: Defined as 𝐴𝑖𝑗 = (−1)𝑖+𝑗 𝑀𝑖𝑗
Critical Exam Identities
5. |𝑎𝑑𝑗 𝐴| = |𝐴|𝑛−1 for a non-singular matrix of order n.
1
6. |𝐴−1 | = |𝐴|
. Proof Note: Since 𝐴𝐴−1 = 𝐼, |𝐴𝐴−1 | = |𝐴||𝐴−1 | = |𝐼| = 1
7. Similar Matrices: If 𝐵 = 𝑃−1 𝐴𝑃, then |𝐵| = |𝐴|

Worked Example: Matrix Inverse by Adjoint Method

Find A-1 for:
1 1 1
𝐴 = [1 2 3]
1 4 9
8. Determinant: |𝐴| = 1(18 − 12) − 1(9 − 3) + 1(4 − 2) = 6 − 6 + 2 = 2 (Consistent, since 2 ≠ 0).
2 3
9. Cofactors (Sample calculation for A11): 𝐴11 = (−1)1+1 | | = (18 − 12) = 6
4 9
10. Full Adjoint:
6 −5 1
𝑎𝑑𝑗 𝐴 = [−6 8 −2]
2 −3 1
11. Inverse:
1 6 −5 1
𝐴−1 = [−6 8 −2]
2
2 −3 1

3. Systems of Linear Equations: Solving Techniques

System Classification
• Homogeneous (𝐴𝑋 = 0): Always has the Trivial Solution (𝑋 = 0). Non-trivial solutions exist if and only if
Rank(𝐴) < 𝑛.
• Non-homogeneous (𝐴𝑋 = 𝐵): Subject to consistency checks via Rouché–Capelli.

Algorithm: Cramer’s Rule (2024 Exam Favorite)

For a 4-variable system where 𝐷 = |𝐴| ≠ 0:
𝐷𝑥 𝐷𝑦 𝐷𝑧 𝐷𝑤
𝑥 = , 𝑦 = , 𝑧 = , 𝑤 =
𝐷 𝐷 𝐷 𝐷
Numerical Case: Given 𝐷 = 10, 𝐷𝑥 = 20, 𝐷𝑦 = 10, 𝐷𝑧 = 0, 𝐷𝑤 = −10.
Solution: 𝑥 = 2, 𝑦 = 1, 𝑧 = 0, 𝑤 = −1.

Consistency Analysis (Rouché–Capelli Flowchart)

Let Rank(A) be 𝑅𝑎 and Rank(A|B) be 𝑅𝑐 , where n is the number of unknowns:
12. Is 𝑅𝑎 = 𝑅𝑐 ?
◦ NO ⇒ Inconsistent (No Solution).
◦ YES ⇒ Consistent.
◦ Is 𝑅𝑎 = 𝑛?
▪ YES ⇒ Unique Solution.
▪ NO (𝑅𝑎 < 𝑛) ⇒ Infinite Solutions.
Solving with Parameters (λ and μ)
Consider:
𝑥 + 𝑦 + 𝑧 = 6
𝑥 + 2𝑦 + 5𝑧 = 10
2𝑥 + 3𝑦 + 𝜆𝑧 = 𝜇
Reduction of the augmented matrix (A|B) yields the final row: [0, 0, 𝜆 − 6 ∶ 𝜇 − 16].
• No Solution: Rank(𝐴) = 2, Rank(𝐴|𝐵) = 3. Occurs if 𝜆 = 6 𝑎𝑛𝑑 𝜇 ≠ 16.
• Infinite Solutions: Rank(𝐴) = Rank(𝐴|𝐵) = 2. Occurs if 𝜆 = 6 𝑎𝑛𝑑 𝜇 = 16.

4. Matrix Reduction: REF, RREF, and Normal Form

REF vs. RREF Comparison
Feature Row Echelon Form (REF) Reduced Row Echelon Form (RREF)
Leading Entries Must be 1. Must be 1.
Below Pivots All zeros. All zeros.
Above Pivots May be non-zero. Must be all zeros.
Utility Used for Rank and Back-substitution. Provides direct solutions to systems.

Algorithm: Normal Form (PAQ)

To find non-singular matrices P and Q such that:
𝐼𝑟 0
𝑃𝐴𝑄 = [ ]
0 0
13. Setup: Write 𝐴𝑚×𝑛 = 𝐼𝑚 𝐴𝐼𝑛 .
14. Operations: Apply Row operations to A and the pre-factor I_m. Apply Column operations to A and the post-
factor I_n.
15. Example: For:
1 1 1
𝐴 = [1 −1 −1]
3 1 1
Setup I₃ A I₃.
1 0 0
• After reduction, 𝑃𝐴𝑄 = [0 1 0], identifying Rank(𝐴) = 2.
0 0 0
1 0 0 1 −1 0
1 1
• Final matrices: 𝑃 = [ 2 − 2 0], 𝑄 = [0 1 −1].
−2 −1 1 0 0 1

Algorithm: Matrix Inverse by Row Operations

Transform [𝐴 | 𝐼] → [𝐼 | 𝐴−1 ]. Example: For:
2 5 3
𝐴 = [1 2 3]
1 0 8
row reduction yields:
 16 −40 9
𝐴−1 = [−5 13 −3]
−2 5 −1

5. Vector Spaces: Linear Independence and Basis

Core Concepts
• Vector Space: A set of vectors closed under addition and scalar multiplication.
• Orthonormal Vectors: A set of orthogonal vectors (𝑢 · 𝑣 = 0) where each vector has unit length (‖𝑢‖ = 1).

Finding Basis and Dimension of span(S)

Given:
𝑆 = [(1, 2, 3, −1, 0), (3, 6, 8, −2, 0), (−1, −1, −3, 1, 1), (−2, −3, −5, 1, 1)]
16. Matrix Construction: Arrange vectors as rows of matrix A.
17. Reduction to RREF:
1 0 0 2 −2
0 1 0 0 1
[ ]
0 0 1 −1 0
0 0 0 0 0
18. Basis: The non-zero rows form the basis:
𝐵 = [(1, 0, 0, 2, −2), (0, 1, 0, 0, 1), (0, 0, 1, −1, 0)]
19. dim(span(𝑆)) = 3 (Number of pivots).

6. Eigenvalues, Eigenvectors, and Diagonalization

The Characteristic Equation
To find eigenvalues 𝜆, solve:
|𝐴 − 𝜆𝐼| = 0
Example (2×2):
1 1
𝐴 = [ ]
3 −1
|𝐴 − 𝜆𝐼| = 𝜆2 − 4 = 0 ⇒ 𝜆1 = 2, 𝜆2 = −2
1
• For 𝜆 = 2, solving (𝐴 − 2𝐼)𝑋 = 0 gives eigenvector 𝑋1 = [ ].
1
1
• For 𝜆 = −2, solving (𝐴 + 2𝐼)𝑋 = 0 gives eigenvector 𝑋2 = [ ].
−3

Cayley-Hamilton (CH) Theorem

Statement: Every square matrix satisfies its own characteristic equation. Professor’s Tip: Use CH to simplify high-
order polynomials.
Example (Exam 8c): For:
1 0
𝐴 = [ ]
2 4
the characteristic equation is 𝜆2 − 5𝜆 + 4 = 0. By CH: 𝐴2 − 5𝐴 + 4𝐼 = 0.
To find the eigenvalues of 4𝐴−1 + 3𝐴 + 2𝐼, substitute the eigenvalues of A (which are 1 and 4) into the
expression.
7. Quadratic Forms (QF) and Advanced Operations
Matrix Representation
A QF 𝑓(𝑥1 , 𝑥2 , 𝑥3 ) is represented by 𝑋 𝑇 𝐴𝑋 where A is symmetric.
Conversion Example:
4𝑥1 2 + 2𝑥2 2 − 3𝑥3 2 + 2𝑥1 𝑥2 + 4𝑥1 𝑥3
• Diagonal elements: 𝑎11 = 4, 𝑎22 = 2, 𝑎33 = −3
2 4
• Off-diagonal: 𝑎12 = 𝑎21 = 2
= 1; 𝑎13 = 𝑎31 = 2
= 2
4 1 2
𝐴 = [1 2 0]
2 0 −3

Canonical Form Metrics

After reduction to ∑ 𝜆𝑖 𝑦𝑖 2 :
• Rank (r): Number of non-zero terms.
• Index (p): Number of positive terms.
• Signature (s): 𝑠 = 2𝑝 − 𝑟 (Difference between positive and negative terms).

Advanced Identities
• Vectorization (vec): Stacking columns of a matrix into a vector.
• Kronecker Product (⊗): 𝐴 ⊗ 𝐵 results in a block matrix [𝑎𝑖𝑗 𝐵].
• Verification: vec(𝐴𝐵𝐶) = (𝐶 𝑇 ⊗ 𝐴)vec(𝐵)