PHYSICS · MATH TOOLS
MATRIX ALGEBRA / SECTIONS 2.1–2.5

Matrix algebra,
step by step.

Your full syllabus in one place: clear notes, worked calculations, and 100 questions covering basic operations through simultaneous eigenvectors and diagonalization.

FROM OPERATIONS TO EIGENBASES
A ⊕ B   A ⊗ B   A†
Build, combine and classify matrices.
P⁻¹AP = D
Choose a basis that makes the map diagonal.
5syllabus sections25worked examples100practice questions100total marks

Your syllabus at a glance

SectionTopicsWorked examplesQuestions
2.1Basic matrix algebra
Operations, direct sum, Kronecker product, transpose, conjugates, trace and determinant
6Q1–Q20
2.2Inverse, rank & linear equations
Inverse methods, row reduction, rank and consistency of simultaneous equations
4Q21–Q40
2.3Special square matrices
Diagonal, triangular, symmetric, antisymmetric, orthogonal, Hermitian, anti-Hermitian, unitary and normal
5Q41–Q60
2.4Eigenvalues & common eigenvectors
Spectral properties of normal, Hermitian, anti-Hermitian, unitary and general matrices; simultaneous eigenvectors
3Q61–Q80
2.5Basis changes & diagonalization
Computing eigenpairs, similarity transformations, change of basis and diagonalization
7Q81–Q100
01 / LEARN

Notes for the full syllabus

Read in order, or use the syllabus links to jump to a section.

Notation: Aᵀ = transpose; A̅ = entrywise complex conjugate; A† = conjugate transpose (adjoint); ⊕ = direct sum; ⊗ = direct or Kronecker product. Unless stated otherwise, scalars are real or complex.
2.1

Basic matrix algebra

Operations, direct sum, Kronecker product, transpose, conjugates, trace and determinant

Dimensions & operations

Start with rows and columns

An m × n matrix has m rows and n columns. Two matrices can be added or subtracted only when their orders agree. Multiply a matrix by a scalar by multiplying every entry.

(A + B)ᵢⱼ = aᵢⱼ + bᵢⱼ   (cA)ᵢⱼ = caᵢⱼ

If A is m × n and B is n × p, their ordinary product AB is m × p. Use row × column: (AB)ᵢⱼ = Σₖ aᵢₖbₖⱼ. Usually AB ≠ BA. The identity I satisfies AI = IA = A whenever the products are defined.

Direct sum

Place the matrices in separate blocks

A ⊕ B =

For A of order m × n and B of order p × q, the direct sum has order (m + p) × (n + q). The off-diagonal blocks are zero matrices of the required sizes.

For square matrices, tr(A ⊕ B) = tr A + tr B and det(A ⊕ B) = (det A)(det B). The eigenvalues form the combined lists of eigenvalues of A and B.

Direct product = Kronecker product

Replace every entry of A by a block

A ⊗ B =

Here direct product means the Kronecker (tensor) product, not ordinary matrix multiplication. If A is m × n and B is p × q, then A ⊗ B is mp × nq.

In general A ⊗ B ≠ B ⊗ A. For compatible dimensions, (A ⊗ B)(C ⊗ D) = AC ⊗ BD. For square A and B, the eigenvalues are all products λᵢμⱼ.

Transpose & conjugates

Three different operations

(Aᵀ)ᵢⱼ = aⱼᵢ   (A̅)ᵢⱼ = a̅ᵢⱼ
A† = (A̅)ᵀ,   (A†)ᵢⱼ = a̅ⱼᵢ

Transpose Aᵀ: interchange rows and columns. Complex conjugate A̅: replace i by −i in every entry. Conjugate transpose A†: perform both operations; it is also called the Hermitian adjoint.

(AB)ᵀ = BᵀAᵀ and (AB)† = B†A†: both reverse the order. Also (A†)† = A. Some books use A* for the adjoint; this page uses † to avoid ambiguity.

Trace

Add the diagonal entries

tr A = a₁₁ + a₂₂ + ⋯ + aₙₙ

Trace is defined here for square matrices. It is linear: tr(A + B) = tr A + tr B and tr(cA) = c tr A. Also tr(Aᵀ) = tr A and tr(A†) = (tr A)̅.

For square matrices of the same order, tr(AB) = tr(BA), even when AB ≠ BA. More generally, trace is unchanged by cyclic rotations of a product: tr(ABC) = tr(BCA) = tr(CAB). It is not generally unchanged by an arbitrary reordering.

Determinant

Use minors and cofactors

det = ad − bc

A minor Mᵢⱼ is the determinant left after deleting row i and column j. The cofactor is Cᵢⱼ = (−1)ⁱ⁺ʲ Mᵢⱼ. Expand along any row i:

det A = Σⱼ aᵢⱼCᵢⱼ = Σⱼ (−1)ⁱ⁺ʲ aᵢⱼMᵢⱼ

This is a recursive rule: compute an n × n determinant from (n − 1) × (n − 1) determinants, stopping at 1 × 1. In a first-row expansion the signs alternate +, −, +, −, … .

Determinant rules worth remembering

Operation / propertyResult
Swap two rowsThe determinant changes sign.
Multiply one row by cThe determinant is multiplied by c.
Add c times one row to a different rowThe determinant is unchanged.
Two equal or proportional rowsThe determinant is zero.
Triangular matrixThe determinant is the product of the diagonal entries.
Square A and Bdet(AB) = det A · det B; det(Aᵀ) = det A.
n × n matrixdet(cA) = cⁿ det A, since all n rows are scaled.
2.2

Inverse, rank & linear equations

Inverse methods, row reduction, rank and consistency of simultaneous equations

Inverse

Undo the matrix multiplication

AA⁻¹ = A⁻¹A = I

A square matrix is invertible exactly when det A ≠ 0. A matrix with det A = 0 is singular and has no ordinary inverse.

⁻¹ = 1ad − bc

This 2 × 2 formula requires ad − bc ≠ 0. For a diagonal matrix with nonzero entries, invert each diagonal entry.

Methods & identities

Find an inverse systematically

Adjugate method: compute the cofactor matrix, transpose it to get adj A, then A⁻¹ = (adj A)/(det A).

Gauss–Jordan method: form the augmented matrix [A | I] and row-reduce the left block to I. The right block becomes A⁻¹. If the left block cannot become I, A is singular.

(AB)⁻¹ = B⁻¹A⁻¹   (A⁻¹)ᵀ = (Aᵀ)⁻¹

For invertible A: det(A⁻¹) = 1/det A and (A⁻¹)⁻¹ = A. Do not invert a product without reversing the order.

Rank & nullity

Count the independent rows or columns

The rank is the number of linearly independent rows, equal to the number of independent columns. Row-reduce to echelon form and count pivots, or nonzero rows. Row operations do not change rank.

0 ≤ rank A ≤ min(m, n)

For A with n columns, rank A + nullity A = n. Nullity is the dimension of the solution space of Ax = 0, equal to the number of free variables. A square n × n matrix is invertible exactly when its rank is n.

Equivalently, rank is the largest order of a nonzero minor. A zero matrix has rank 0.

Simultaneous linear equations

Compare the coefficient and augmented ranks

Ax = b   [A | b]

Write the coefficients in A, the unknowns in x and the right-hand sides in b. Use elimination on the augmented matrix. Let r = rank A, r′ = rank[A | b], and n be the number of unknowns.

r ≠ r′: no solution. r = r′ = n: one unique solution. r = r′ < n: infinitely many solutions over ℝ or ℂ, with n − r free variables.

For a square invertible A, x = A⁻¹b. Cramer’s rule gives xⱼ = det Aⱼ / det A, where Aⱼ replaces column j of A by b.

Homogeneous equations

Ax = 0 always has the zero solution. It also has a nonzero solution exactly when rank A < n, where n is the number of unknowns. For square A, this is equivalent to det A = 0. A row such as [0 0 0 | 1] in an augmented matrix means 0 = 1 and proves inconsistency.

2.3

Special square matrices

Diagonal, triangular, symmetric, antisymmetric, orthogonal, Hermitian, anti-Hermitian, unitary and normal

Definitions and a concrete example of every type

All matrices in this table are square. Orthogonal matrices are taken to be real. Complex conjugation matters for Hermitian, anti-Hermitian, unitary and normal matrices.

Matrix typeDefining testExample
Diagonalaᵢⱼ = 0 for i ≠ j
Upper triangularaᵢⱼ = 0 when i > j (below diagonal)
Lower triangularaᵢⱼ = 0 when i < j (above diagonal)
SymmetricAᵀ = A
Antisymmetric (skew-symmetric)Aᵀ = −A; every diagonal entry is zero
Orthogonal (real)QᵀQ = QQᵀ = I; Q⁻¹ = Qᵀ15
HermitianH† = H; diagonal entries are real
Anti-Hermitian (skew-Hermitian)K† = −K; diagonal entries are purely imaginary or zero
UnitaryU†U = UU† = I; U⁻¹ = U†
NormalNN† = N†N
Two decompositions

Separate symmetry from antisymmetry

A = S + K   S = (A + Aᵀ)/2   K = (A − Aᵀ)/2

Every square real matrix has a symmetric part S and an antisymmetric part K. Check Sᵀ = S and Kᵀ = −K.

A = H + L   H = (A + A†)/2   L = (A − A†)/2

Every square complex matrix has a Hermitian part H and an anti-Hermitian part L. Also L/i is Hermitian, so one can write A = H + iG with H and G Hermitian.

Relations & useful properties

Recognize overlapping classes

Every real symmetric matrix is Hermitian. Every real antisymmetric matrix is anti-Hermitian. A real unitary matrix is orthogonal. A diagonal complex matrix is always normal, but it is Hermitian only when its entries are real.

Hermitian, anti-Hermitian and unitary matrices are all normal. A normal matrix need not belong to any of these three narrower classes.

Orthogonal and unitary matrices preserve lengths and inner products. For real orthogonal Q, det Q = ±1. For unitary U, |det U| = 1. For an odd-order antisymmetric matrix over ℝ or ℂ, det A = 0.

2.4

Eigenvalues & common eigenvectors

Spectral properties of normal, Hermitian, anti-Hermitian, unitary and general matrices; simultaneous eigenvectors

Eigenpair definition

Find the special nonzero vectors

Av = λv,   v ≠ 0

An eigenvector is a nonzero vector that is only scaled by A; λ is the eigenvalue. The zero vector is never an eigenvector. Any nonzero scalar multiple of an eigenvector has the same eigenvalue.

An eigenspace contains all eigenvectors for one eigenvalue together with zero. Its dimension is called the geometric multiplicity. The number of repetitions of a root of the characteristic polynomial is its algebraic multiplicity. Always 1 ≤ geometric multiplicity ≤ algebraic multiplicity.

Inner product & normal matrices

Use conjugation for complex vectors

⟨u, v⟩ = u†v   ‖v‖² = v†v

For a complex vector, use conjugate transpose in the inner product. Orthogonal vectors satisfy u†v = 0; a normalized vector has v†v = 1.

Spectral theorem: a complex square matrix is normal if and only if it can be diagonalized by a unitary matrix. Eigenvectors of a normal matrix belonging to distinct eigenvalues are orthogonal. Within a repeated eigenspace, choose an orthonormal basis.

Hermitian & anti-Hermitian spectra

Reality and imaginary eigenvalues

For H† = H, every eigenvalue is real. If Hv = λv, then v†Hv = λv†v is real, so λ is real. Eigenvectors with distinct eigenvalues are orthogonal.

For K† = −K, eigenvalues are purely imaginary or zero: λ̅ = −λ. One can write λ = iα with α real. If H is Hermitian, iH is anti-Hermitian and its eigenvalues are i times those of H, with the same eigenvectors.

Unitary & general spectra

Know what is and is not guaranteed

For U†U = I and Uv = λv, preservation of length gives ‖Uv‖² = |λ|²‖v‖² = ‖v‖². Therefore |λ| = 1. Unitary eigenvalues lie on the unit circle; they need not be real.

An arbitrary n × n complex matrix has n characteristic roots counted with algebraic multiplicity, but may have fewer than n independent eigenvectors. Distinct eigenvalues give independent eigenvectors. A real matrix can have complex eigenvalues, occurring in conjugate pairs.

The sum and product of all eigenvalues equal tr A and det A, respectively, with algebraic multiplicity counted.

Simultaneous eigenvectors

Diagonalize compatible matrices together

A common eigenvector v obeys Av = λv and Bv = μv. A common eigenbasis means n independent vectors that are eigenvectors of both matrices.

Two commuting Hermitian matrices have a common orthonormal eigenbasis. More generally, two complex normal matrices are simultaneously unitarily diagonalizable if and only if AB = BA. If A and B are diagonalizable over the same field, commuting is equivalent to having a common eigenbasis over that field.

Commuting arbitrary matrices need not have a complete common eigenbasis: a defective matrix and I already give a counterexample. Sharing just one eigenvector does not imply that the matrices commute.

Degeneracy

Resolve the freedom inside an eigenspace

If Av = λv and AB = BA, then A(Bv) = B(Av) = λBv. Thus B maps the λ-eigenspace of A into itself.

When that eigenspace is one-dimensional, Bv must be proportional to v. When it is degenerate, an arbitrary eigenvector of A need not be an eigenvector of B. Diagonalize B within that eigenspace to choose common eigenvectors. Worked example 18 shows the steps.

Spectral properties at a glance

ClassEigenvaluesEigenvectors / diagonalization
HermitianRealAn orthonormal eigenbasis exists.
Anti-HermitianPurely imaginary, including zeroAn orthonormal eigenbasis exists over ℂ.
Unitary|λ| = 1An orthonormal eigenbasis exists over ℂ.
NormalMay be any complex numbersAn orthonormal eigenbasis exists over ℂ.
General squareMay be complex; repeated roots can occurA full eigenbasis may fail to exist.
2.5

Basis changes & diagonalization

Computing eigenpairs, similarity transformations, change of basis and diagonalization

Calculate eigenvalues

Solve the characteristic equation

det(A − λI) = 0

Subtract λ from every diagonal entry, calculate the determinant, factor the polynomial and solve for its roots. The equation det(λI − A) = 0 gives the same roots.

A = ⇒ λ² − (a + d)λ + (ad − bc) = 0

For a triangular matrix, the eigenvalues are its diagonal entries. For a 3 × 3 matrix, expand along a row or column with many zeros whenever possible.

Calculate eigenvectors

Solve a homogeneous system for each root

(A − λI)v = 0

Substitute one eigenvalue at a time, row-reduce, identify the free variables and choose nonzero solutions. For a repeated eigenvalue, find a basis for the entire eigenspace; do not choose two multiples of one vector and count them as independent.

Check every proposed vector directly using Av = λv. If desired, divide it by its length to normalize it.

Change of basis

Distinguish a vector from its coordinates

Let the columns of an invertible matrix P be the new basis vectors expressed in the old basis. Then:

[x]old = P[x]new   [x]new = P⁻¹[x]old
Anew = P⁻¹AoldP

The vector or linear transformation stays the same; its coordinates or matrix entries change. The direction of P in these formulas depends on this explicit column convention.

Similarity & diagonalization

Choose an eigenvector basis

P⁻¹AP = D   ⇔   AP = PD   ⇔   A = PDP⁻¹

Similar matrices have the same characteristic polynomial, eigenvalues, trace, determinant and rank. They need not have the same entries or eigenvectors in a fixed coordinate basis.

For diagonalization, put n independent eigenvectors in the columns of P. Put their matching eigenvalues on the diagonal of D, in the same order. Diagonalization is possible exactly when a full eigenbasis exists over the chosen field.

A real matrix with non-real eigenvalues cannot be diagonalized to a real diagonal matrix using a real P. It can still be diagonalizable over ℂ.

A reliable diagonalization checklist

  1. Find the roots of det(A − λI) = 0.
  2. Find an eigenspace basis for every distinct root.
  3. Check that there are n independent eigenvectors in total.
  4. Build P from these columns and D from the matching eigenvalues.
  5. Check det P ≠ 0 and verify AP = PD or P⁻¹AP = D.

For a real symmetric matrix, choose orthonormal eigenvectors and write QᵀAQ = D. For a complex normal matrix, use U†AU = D with U unitary. If A = PDP⁻¹, then Aᵏ = PDᵏP⁻¹ for positive integers k.

02 / WORK THROUGH

25 worked examples

Follow each calculation on paper, then use the answer to check your own work.

2.1 · Basic matrix algebra

01Ordinary addition and multiplicationRow × column, noncommutativity and a trace check
1

Add corresponding entries

A =   B =
A + B = =
2

Multiply rows of A by columns of B

AB = =
3

Reverse the order and compare

BA =

AB ≠ BA. However, tr(AB) = 4 + 3 = 7 and tr(BA) = 2 + 5 = 7. Trace can agree even when the matrices differ.

02Constructing a direct sumA 2 × 2 block and a 1 × 1 block
1

Write the two matrices

A =   B =
2

Place A and B on separate diagonal blocks

A ⊕ B =

The 2 × 2 and 1 × 1 blocks give a 3 × 3 matrix. Fill both off-diagonal blocks with zero.

3

Check its trace and determinant

tr(A ⊕ B) = (1 + 4) + 5 = 10
det(A ⊕ B) = (1·4 − 2·3)·5 = −10
03Constructing a direct (Kronecker) productReplace each scalar entry by a whole matrix block
1

Identify the blocks

A =   B =
A ⊗ B =
2

Expand each block into entries

A ⊗ B =

Each matrix is 2 × 2, so the Kronecker product is 4 × 4. It is different from the ordinary 2 × 2 product AB.

3

Compare with the reversed Kronecker product

B ⊗ A =

The two products have the same order but different entries. The order of the factors matters.

04Transpose, conjugate and conjugate transposeThree operations on one complex matrix
A =
1

Transpose: swap rows and columns

Aᵀ =

Do not change the signs of i when taking only the transpose.

2

Conjugate: change i to −i entry by entry

A̅ =

Do not swap rows and columns when taking only the complex conjugate.

3

Take the conjugate transpose

A† = (A̅)ᵀ =

Conjugate and transpose can be done in either order; they give the same A†.

05Calculating a trace with complex entriesOff-diagonal entries do not contribute
A =
1

Select only the main diagonal

The diagonal entries are 3 + i and 2 − i. The off-diagonal entries do not enter the trace.

2

Add and simplify

tr A = (3 + i) + (2 − i) = 5
3

Check transpose and adjoint traces

tr(Aᵀ) = 5   tr(A†) = (3 − i) + (2 + i) = 5

In general tr(A†) is the complex conjugate of tr A. They agree in this example because tr A is real.

06A 3 × 3 determinant from minors and cofactorsA full first-row expansion
A =
1

Expand along the first row with signs +, −, +

det A = 1 det − 2 det + 3 det
2

Evaluate the three 2 × 2 minors

M₁₁ = 1·0 − 4·6 = −24
M₁₂ = 0·0 − 4·5 = −20
M₁₃ = 0·6 − 1·5 = −5
3

Include the cofactor signs and add

det A = 1(−24) − 2(−20) + 3(−5)
= −24 + 40 − 15 = 1

The cofactor C₁₂ is −M₁₂ = 20. Keeping track of the middle minus sign is essential.

2.2 · Inverse, rank & linear equations

07An inverse by formula and by row reductionTwo methods, followed by a multiplication check
A =
1

Check that the inverse exists

det A = 2·1 − 1·1 = 1 ≠ 0
2

Apply the 2 × 2 formula

A⁻¹ = 11 =
3

Repeat using Gauss–Jordan elimination

[A | I] =

Swap R₁ and R₂, then set R₂ → R₂ − 2R₁:

Set R₂ → −R₂, then R₁ → R₁ − R₂:

The left block is I. The right block is A⁻¹.

4

Verify by multiplication

AA⁻¹ = =
08Rank and nullity by eliminationA 3 × 3 matrix with one dependent row
A =
1

Eliminate the first-column entries below the first pivot

Use R₂ → R₂ − 2R₁ and R₃ → R₃ − R₁.

2

Move the nonzero row up and simplify

Swap R₂ and R₃; then multiply R₂ by −1 and use R₁ → R₁ − 2R₂.

3

Count pivots and identify the nullity

There are two nonzero rows, hence rank A = 2. There are three columns, so nullity A = 3 − 2 = 1.

From Ax = 0: x − z = 0 and y + 2z = 0. Choose z = t, giving x = t and y = −2t.

x = t

The vector (1, −2, 1)ᵀ spans the nullspace. A nonzero 2 × 2 minor is det[[1, 2], [1, 1]] = −1, confirming rank is at least 2.

09Solving three simultaneous linear equationsElimination, back-substitution and consistency
1

Write the system and augmented matrix

x + y + z = 6   x − y + z = 2   x + y − z = 0
2

Subtract the first row from the other rows

R₂ → R₂ − R₁   R₃ → R₃ − R₁

The last two equations give y = 2 and z = 3.

3

Back-substitute and check

x = 6 − y − z = 6 − 2 − 3 = 1
(x, y, z) = (1, 2, 3)

Check: 1 + 2 + 3 = 6; 1 − 2 + 3 = 2; 1 + 2 − 3 = 0. All three pivots are nonzero, so both ranks equal 3 and the solution is unique.

10Infinitely many solutions versus no solutionWhy the augmented rank matters
1

A dependent, consistent pair

x + 2y = 3   2x + 4y = 6

Subtract twice the first equation from the second: 0 = 0. Both coefficient and augmented matrices have rank 1.

There are two unknowns and one free variable. Set y = t, then x = 3 − 2t.

(x, y) = (3 − 2t, t),   t ∈ ℝ
2

Change the second right-hand side

x + 2y = 3   2x + 4y = 7

Now the same elimination gives 0 = 1. The coefficient rank is 1 but the augmented rank is 2.

There is no solution. The left-hand sides are dependent, while the right-hand sides contradict one another.

2.3 · Special square matrices

11Symmetric and antisymmetric partsDecomposing a real matrix
A =
1

Find the symmetric part

S = (A + Aᵀ)/2 =

For example, the upper-right entry is (4 + 2)/2 = 3.

2

Find the antisymmetric part

K = (A − Aᵀ)/2 =

The upper-right entry is (4 − 2)/2 = 1, and the lower-left entry is −1.

3

Check both definitions and the sum

Sᵀ = S   Kᵀ = −K   S + K = A
12Hermitian and anti-Hermitian partsUse the adjoint rather than only the transpose
A =
1

Find the adjoint

A† =
2

Average A and A†

H = (A + A†)/2 =

The off-diagonal entries are complex conjugates and the diagonal entries are real. Thus H† = H.

3

Take half the difference

L = (A − A†)/2 =

Conjugating and transposing gives L† = −L. Adding H and L returns the original A.

13Testing an orthogonal matrixTranspose, inverse, determinant and length
Q = 15
1

Transpose Q

Qᵀ = 15
2

Multiply QᵀQ

QᵀQ = 125 = I

So Q is orthogonal and Q⁻¹ = Qᵀ. Also det Q = (9 + 16)/25 = 1.

3

Check length preservation on a vector

Q =
‖‖² = 1 + 4 = 5   ‖‖² = 1 + 4 = 5
14Testing a genuinely complex unitary matrixConjugation is essential
U = 1√2
1

Take the conjugate transpose

U† = 1√2
2

Multiply and use i² = −1

U†U = 12
= 12 = I
3

State the conclusion

U is unitary and U⁻¹ = U†. Its two columns are orthonormal under the complex inner product. U is not Hermitian because U† ≠ U. It is normal because every unitary matrix is normal.

15A normal matrix and a matrix that is not normalCompute both products rather than guessing
1

Test a diagonal complex matrix

N =   N† =
NN† = N†N =

N is normal. It is not Hermitian, anti-Hermitian or unitary. Normal is a broader class.

2

Test a nonsymmetric real matrix

J =   J† = Jᵀ =
JJ† =   J†J =

Since the two products differ, J is not normal. Having real entries does not make a matrix normal.

2.4 · Eigenvalues & common eigenvectors

16Eigenvalues of Hermitian and anti-Hermitian matricesA complex example with normalized eigenvectors
H =
1

Find the characteristic roots

det(H − λI) = (2 − λ)² − i(−i) = (2 − λ)² − 1
(λ − 3)(λ − 1) = 0   ⇒   λ = 3, 1
2

Find and normalize the eigenvectors

For λ = 3, −x + iy = 0 gives x = iy. Set y = 1. For λ = 1, x + iy = 0 gives x = −iy. Again set y = 1.

u₃ = 1√2   u₁ = 1√2

Both vectors have length 1. Their inner product is (i·i + 1)/2 = 0, so they are orthogonal.

3

Multiply the Hermitian matrix by i

K = iH =

K† = −K, so K is anti-Hermitian. The same vectors satisfy Ku₃ = 3iu₃ and Ku₁ = iu₁. Its eigenvalues are 3i and i, as expected.

17Eigenvalues of a unitary matrixComplex eigenvalues with unit modulus
U = 1√2
1

Try the vectors (1, 1)ᵀ and (1, −1)ᵀ

U = 1 + i√2
U = 1 − i√2
2

Read off the eigenvalues and check their moduli

λ₊ = (1 + i)/√2   λ₋ = (1 − i)/√2
|λ₊|² = |λ₋|² = (1² + 1²)/2 = 1
3

Normalize the common eigenvectors

u₊ = 1√2   u₋ = 1√2

With these vectors as the columns of Q, Q†UQ = diag((1 + i)/√2, (1 − i)/√2). The eigenvalues need not be real, but both lie on the unit circle.

18Simultaneous eigenvectors with degeneracyOne orthonormal basis diagonalizes two commuting matrices
A =   B =
1

Check commutation

AB = BA =

Both matrices are real symmetric, hence Hermitian. Their commutation guarantees a common orthonormal eigenbasis.

2

Resolve the degenerate eigenspace of A

The eigenvalue 1 of A has eigenspace {(x, y, 0)ᵀ}. In this plane, B acts as [[2, 1], [1, 2]]. Its eigenvectors are (1, 1)ᵀ and (1, −1)ᵀ, with eigenvalues 3 and 1.

u₁ = 1√2   u₂ = 1√2   u₃ =
3

Use the same columns for both diagonalizations

QᵀAQ =
QᵀBQ =

The vector e₁ = (1, 0, 0)ᵀ is an eigenvector of A but not of B: Be₁ = (2, 1, 0)ᵀ. Choosing a suitable basis within the degenerate plane is the key step.

2.5 · Basis changes & diagonalization

19Complete diagonalization of a symmetric 2 × 2 matrixEigenvalues, eigenvectors, P, inverse and final verification
A =
1

Compute the characteristic equation

det(A − λI) = (3 − λ)² − 1
λ² − 6λ + 8 = (λ − 4)(λ − 2) = 0
2

Find both eigenvectors

For λ = 4: −x + y = 0, so choose v₁ = (1, 1)ᵀ. For λ = 2: x + y = 0, so choose v₂ = (1, −1)ᵀ.

A = 4   A = 2
3

Build P, P⁻¹ and D in matching order

P =   det P = −2
P⁻¹ =   D =
4

Verify every entry of the final product

AP =
P⁻¹AP = =
20Diagonalization of a general nonsymmetric matrixA full eigenbasis need not be orthogonal
B =
1

Find the eigenvalues

det(B − λI) = (2 − λ)(3 − λ) = 0

The eigenvalues are 2 and 3.

2

Solve each homogeneous system

For λ = 2, y = 0; choose v₁ = (1, 0)ᵀ. For λ = 3, −x + y = 0; choose v₂ = (1, 1)ᵀ.

3

Construct the similarity transformation

P =   P⁻¹ =
BP = = P

Thus P⁻¹BP = diag(2, 3). The matrix is diagonalizable even though it is not symmetric or normal. Its eigenvectors need not be orthogonal.

21Complete diagonalization of a 3 × 3 matrixDeterminant expansion, eigenspaces and matrix verification
C =
1

Expand the determinant along the third row

det(C − λI) = (4 − λ)[(2 − λ)² − 1]
= (4 − λ)(1 − λ)(3 − λ) = 0

The eigenvalues are 1, 3 and 4.

2

Solve for three eigenvectors

For λ = 1: x + y = 0 and z = 0. Choose v₁ = (1, −1, 0)ᵀ.

For λ = 3: −x + y = 0 and z = 0. Choose v₂ = (1, 1, 0)ᵀ.

For λ = 4: −2x + y = 0 and x − 2y = 0 imply x = y = 0. Choose v₃ = (0, 0, 1)ᵀ.

3

Build the matrices

P =   det P = 2
P⁻¹ =   D =
4

Verify AP = PD and the diagonal form

CP = = PD
P⁻¹CP =
22Repeated eigenvalues: two different outcomesDiagonalizable and defective examples side by side
1

A repeated eigenvalue with enough eigenvectors

R =

For λ = 2, all vectors (x, y, 0)ᵀ belong to the eigenspace. Choose e₁ and e₂. For λ = 5, choose e₃. There are three independent eigenvectors, so R is diagonalizable. It is already diagonal.

2

A repeated eigenvalue without enough eigenvectors

J =   det(J − λI) = (2 − λ)²
(J − 2I) = =

The only condition is y = 0. Every eigenvector is t(1, 0)ᵀ with t ≠ 0. The eigenspace dimension is 1, although the eigenvalue has algebraic multiplicity 2.

3

Decide diagonalizability

J needs two independent eigenvectors but provides only one. Therefore J is not diagonalizable. Repeated eigenvalues alone do not settle the question; eigenspace dimensions do.

23A real rotation with complex eigenvectorsWhy the field of scalars matters
R =
1

Find the characteristic roots

det(R − λI) = λ² + 1 = 0   ⇒   λ = i, −i
2

Find complex eigenvectors

For λ = i, −ix − y = 0 gives y = −ix. Choose x = 1. For λ = −i, ix − y = 0 gives y = ix.

v₁ =   v₂ =
3

Diagonalize over ℂ

P =   det P = 2i ≠ 0
P⁻¹ = 12i
RP = = P

Thus P⁻¹RP = diag(i, −i). R is diagonalizable over ℂ but cannot be diagonalized to a real diagonal matrix over ℝ. Dividing both columns by √2 gives a unitary diagonalizing matrix.

24Changing basis for vectors and matricesFix the convention for P before using its inverse
1

Specify the old matrix and the new basis

Aold =   b₁ =   b₂ =
P =   P⁻¹ =

The columns of P are the new basis vectors written in the old coordinates.

2

Transform the coordinates of one vector

[x]old =   [x]new = P⁻¹ =

Indeed 2b₁ + b₂ = (3, 1)ᵀ.

3

Transform the matrix representing the same map

Anew = P⁻¹AoldP =
4

Check the result on the vector

Anew =
P⁻¹Aold = P⁻¹ =

Both coordinate descriptions agree. Similarity changes the matrix entries but preserves eigenvalues 2 and 4, trace 6 and determinant 8.

25Using diagonalization to calculate matrix powersAn application of the similarity transformation
A =   A = PDP⁻¹,   D =
1

Square the diagonal matrix

D² =
2

Transform back

A² = PD²P⁻¹ = 12
A² =
3

Generalize to positive integer powers

Aᵏ = 12

For example, k = 3 gives A³ = [[36, 28], [28, 36]]. Diagonalization turns repeated matrix multiplication into ordinary powers of eigenvalues.

03 / TEST YOURSELF

The 100-question quiz

20 questions from each section. Each correct answer earns 1 mark; there is no negative marking.

Student details

Optional for practice. Required to download or email a named answer sheet.

0 of 100 answered

The section filter changes the questions shown. Your total score always covers all 100 questions; unanswered questions earn 0.

2.1 · Basic matrix algebra Q1–Q20

Q1 A matrix with 2 rows and 3 columns has order:Basic
Q2 If A is 2 × 3 and B is 3 × 4, the ordinary product AB has order:Basic
Q3 For A = and B = , calculate AB.Application
Q4 For the same A = and B = , calculate BA.Application
Q5 If A = and B = , what is A ⊕ B?Application
Q6 If A is 2 × 3 and B is 4 × 1, the direct sum A ⊕ B has order:Basic
Q7 If A is 2 × 3 and B is 3 × 2, the Kronecker product A ⊗ B has order:Basic
Q8 For A = , which block is in block row 1, block column 2 of A ⊗ B?Basic
Q9 For A = and B = diag(1, −1), calculate A ⊗ B.Application
Q10 A has eigenvalues 2 and 3; B has eigenvalues 4 and 5. The eigenvalues of A ⊗ B, counting multiplicities, are:Challenge
Q11 What is the transpose of ?Basic
Q12 For compatible matrices A and B, which transpose rule is correct?Basic
Q13 The entrywise complex conjugate of is:Basic
Q14 If A = , then A† is:Application
Q15 The trace of is:Basic
Q16 For square matrices A and B of the same order, which statement is always true?Basic
Q17 Calculate det .Application
Q18 The determinant of the upper triangular matrix equals:Application
Q19 Calculate det .Application
Q20 A multiple of one row is added to a different row. What happens to the determinant?Basic

2.2 · Inverse, rank & linear equations Q21–Q40

Q21 A square matrix A has an ordinary inverse exactly when:Basic
Q22 Find the inverse of .Application
Q23 If A and B are invertible, (AB)⁻¹ equals:Basic
Q24 The inverse of diag(2, −4) is:Basic
Q25 If det A = 5, then det(A⁻¹) is:Basic
Q26 Which equation must hold for an inverse matrix?Basic
Q27 Does have an ordinary inverse?Basic
Q28 In Gauss–Jordan inversion, the correct augmented-matrix procedure is:Basic
Q29 The rank of a 3 × 4 zero matrix is:Basic
Q30 The rank of the identity matrix I₄ is:Basic
Q31 Find the rank of .Application
Q32 For an m × n matrix, the rank can never exceed:Basic
Q33 A matrix has 5 columns and rank 3. The nullity is:Application
Q34 An invertible 3 × 3 matrix must have rank:Basic
Q35 Solve x + y = 5 and x − y = 1.Application
Q36 The equations x + 2y = 3 and 2x + 4y = 6 have:Application
Q37 The equations x + 2y = 3 and 2x + 4y = 7 have:Application
Q38 A homogeneous system with 3 unknowns has coefficient rank 2. Its solution space has:Basic
Q39 If rank A = 2 but rank[A | b] = 3, then Ax = b is:Basic
Q40 For a real system with 3 unknowns, rank A = rank[A | b] = 2. The system has:Application

2.3 · Special square matrices Q41–Q60

Q41 A diagonal matrix has:Basic
Q42 In an upper triangular matrix, which entries must vanish?Basic
Q43 Which matrix is lower triangular?Basic
Q44 The defining condition for a symmetric matrix is:Basic
Q45 The diagonal entries of an antisymmetric matrix over ℝ or ℂ must be:Basic
Q46 For A = , the antisymmetric part (A − Aᵀ)/2 is:Application
Q47 For a real orthogonal matrix Q, the inverse is:Basic
Q48 The matrix Q = 15 is orthogonal. What does it do to a vector’s Euclidean length?Basic
Q49 The determinant of a real orthogonal matrix must be:Basic
Q50 Which matrix is Hermitian?Application
Q51 Which matrix is anti-Hermitian?Application
Q52 The conjugate transpose A† is obtained by:Basic
Q53 Every diagonal entry of a Hermitian matrix is:Basic
Q54 Every diagonal entry of an anti-Hermitian matrix is:Basic
Q55 Which condition defines a unitary matrix?Basic
Q56 If U is unitary, U⁻¹ equals:Basic
Q57 For U = 1√2, what is U†U?Application
Q58 A square matrix N is normal when:Basic
Q59 Which statement about normal matrices is correct?Basic
Q60 Which calculation proves J = is not normal?Application

2.4 · Eigenvalues & common eigenvectors Q61–Q80

Q61 Which equation defines an eigenvalue λ and a corresponding eigenvector v ≠ 0?Basic
Q62 Which vector is never counted as an eigenvector?Basic
Q63 If v is an eigenvector with eigenvalue λ and c ≠ 0, then cv is:Basic
Q64 All eigenvalues of a Hermitian matrix are:Basic
Q65 The eigenvalues of an anti-Hermitian matrix are:Basic
Q66 If λ is an eigenvalue of a unitary matrix, then:Basic
Q67 Every complex normal matrix can be diagonalized using:Basic
Q68 Which statement is correct for a general n × n complex matrix?Basic
Q69 Eigenvectors of a Hermitian matrix corresponding to distinct eigenvalues are:Basic
Q70 For a complex normal matrix, eigenvectors corresponding to distinct eigenvalues are:Basic
Q71 Find the eigenvalues of H = .Application
Q72 If H has eigenvalues 1 and 3, the eigenvalues of iH are:Application
Q73 The unitary matrix diag(1, i) has which eigenvalues?Basic
Q74 How many linearly independent eigenvectors does J = have?Application
Q75 The sum of all eigenvalues of a square matrix, with algebraic multiplicity counted, equals:Basic
Q76 The product of all eigenvalues of a square matrix, with algebraic multiplicity counted, equals:Basic
Q77 Two Hermitian matrices A and B of the same order admit a complete common orthonormal eigenbasis if and only if:Challenge
Q78 Let A = and B = I₂. They commute. Which conclusion is correct?Challenge
Q79 Let A = diag(1, 1, 2), B = and v = . The eigenvalues of A and B on this common eigenvector are, respectively:Application
Q80 A = diag(1, 2) and B = both have eigenvector (1, 0)ᵀ. What follows?Challenge

2.5 · Basis changes & diagonalization Q81–Q100

Q81 The eigenvalues of A are found from:Basic
Q82 The eigenvalues of A = are:Application
Q83 For A = , which vector has eigenvalue 4?Application
Q84 For A = , which vector has eigenvalue 2?Application
Q85 If the first eigenvector is (1, 1)ᵀ and the second is (1, −1)ᵀ, which P has them as columns in that order?Application
Q86 If columns 1 and 2 of P correspond to eigenvalues 4 and 2, respectively, then D is:Basic
Q87 The inverse of P = is:Application
Q88 Let the columns of P be the new basis vectors expressed in the old basis. The matrix of A in the new basis is:Basic
Q89 The columns of P = define the new basis in old coordinates. If [x]old = (3, 1)ᵀ, what is [x]new?Application
Q90 For Aold = diag(2, 4) and P = , calculate Anew = P⁻¹AoldP.Application
Q91 Which statement is true for similar matrices?Basic
Q92 Find the eigenvalues of C = .Application
Q93 For C = , an eigenvector with eigenvalue 1 is:Application
Q94 For C = and P = , what is P⁻¹CP?Application
Q95 Is diag(2, 2, 5) diagonalizable?Basic
Q96 Why is J = not diagonalizable?Basic
Q97 The real matrix R = is:Challenge
Q98 For R = , which vector is an eigenvector with eigenvalue i?Application
Q99 For A = , use its diagonalization to find A².Application
Q100 After proposing an invertible P and a diagonal D, which equation verifies diagonalization without calculating P⁻¹?Basic

Email submission sends your name, email, roll number, score and selected answers to josemathew@thecochincollege.edu.in when you press “Submit & email”. It needs an internet connection and the EmailJS service from the supplied sample. If email fails, download the CSV and attach it to an email yourself.