MATH235: Linear Algebra 2

Practice Problems for Chapter 5: Orthogonal Diagonalization Fall 2022

Instructions
• These problems are meant to give you an opportunity to practice the material covered in class
and in the course notes. They are ungraded: you do not have to submit them. However, we
highly recommend that you attempt all of them and write up full solutions to each.

• You are welcome to discuss these problems with your instructor, classmates, and on Piazza.

Computational Problems
Q1. Determine if the following real matrices A are orthogonally diagonalizable. Either answer
“yes” or “no”, and provide a brief justification.
 
1 1
(a) .
1 −1
 
1 −1
(b) .
1 1
 
1 2 4
(c) 2 3 0 .
4 0 −1
 
5 0 0 0
0 4 0 0
(d) 
0 0 3 0.


0 0 0 2
Q2. Determine if the following matrices A are unitarily diagonalizable. Either answer “yes” or
“no”, and provide a brief justification.
 
1 0 0
(a) 0 1 0.
0 0 1
 
cos(θ) − sin(θ)
(b) where θ is any real number (this matrix rotates vectors by θ radians
sin(θ) cos(θ)
counterclockwise in R2 ).
 
1 2 + i −i
(c) 2 − i −1 −3.
i −3 0
 
1 0 1
(d) 1 1 0.

0 1 1
 
1 1 0
(e) 0 1 0.
0 0 1

1
MATH235: Linear Algebra 2
Practice Problems for Chapter 5: Orthogonal Diagonalization Fall 2022

Q3. Orthogonally diagonalize the following real matrices.

 
0 1
(a) .
1 0
 
3 −6 0
(b) −6 0 6 .
0 6 −3
 
5 −4 −2
(c) −4 5 2 .
−2 2 2
Q4. Unitarily diagonalize the following matrices.
 
0 4
(a) .
−4 0
 
0 3 4i
(b) −3 0 0 .
4i 0 0
 
0 1 i
(c)  1 0 −i.
−i i 0
Q5. Determine which of the following matrices are orthogonal.
 
1 2
(a) .
−2 1
(b) A matrix P obtained by permuting the columns of the n × n identity matrix. (Such
a matrix is called a permutation matrix. Examples of such matrices are given in Q9
below.)
 
−i 0 0
(c)  0 i 0 
0 0 −1
" 1+i 1+i #
√ √
(d) 6 3
2i
√ −i
√
6 3

Q6. Determine which of the following matrices are unitary.

 
i −i
(a) .
1 −1
 
2 3 + 2i
(b) .
3 − 2i −1
 1 1 1 
√ √ √
3 3 3
(c)  √13 1
− 2√3 − i 1
− 2√3 + i
2 .

2
√1 1
− 2√ + i 1
− 2√ − i
3 3 2 3 2

2
MATH235: Linear Algebra 2
Practice Problems for Chapter 5: Orthogonal Diagonalization Fall 2022

Q7. Determine which of the following matrices are Hermitian.

 
1 1 1
(a) 1 1 1.
1 1 1
 
i i i
(b) i i i.
i i i
 
1 1+i
(c) .
1 − i −1
 
i 0
(d) .
0 −i
(e) U where U is a unitary matrix.

Q8. Determine which of the following matrices are normal.

(a) A matrix A that satisfies AT = −A. (Such a matrix is called skew-symmetric.)

(b) A matrix A that satisfies A∗ = −A. (Such a matrix is called skew-Hermitian.)
(c) U where U is a unitary matrix.
 
0 i
(d) .
i 0
Q9. Show that the following matrices are unitarily diagonalizable, and unitarily diagonalize them
(that is, for each matrix A, find a unitary matrix U and a diagonal matrix D so that U ∗ AU =
D).  
  0 0 0 1
  0 0 1
0 1 1 0 0 0
, 1 0 0 , and  0 1 0 0 .

1 0
0 1 0
0 0 1 0
Do you have any guesses as to what happens in general? Try to prove anything you think
might be true.

Conceptual Problems
Q1. Give an example of a symmetric matrix A ∈ M3×3 (R) whose eigenvalues are 1, 2 and 3, with
corresponding eigenvectors      
1 0 1
 0  , 1 and 0 .
−1 0 1
 
1 1
Q2. (a) Unitarily diagonalize .
1 1

3
MATH235: Linear Algebra 2
Practice Problems for Chapter 5: Orthogonal Diagonalization Fall 2022

 
1 1 1
(b) Unitarily diagonalize 1 1 1.
1 1 1
(c) Unitarily diagonalize the n × n matrix where every entry is 1.

Q3. Rotations in R2 about the origin are linear maps.

(a) Let Rθ : R2 → R2 be the linear map defined by a rotation by θ about the origin
counterclockwise. With respect to the standard basis S, find the matrices [Rπ/2 ]S ,
[R2π/3 ]S , [Rπ/3 ]S , [Rπ/4 ]S , and [Rπ/6 ]S .
(b) Compute the eigenvalues for all of these matrices, and show none of them have any real
eigenvalues (why should you expect this?).
(c) Unitarily diagonalize these matrices (over C). What do you notice?
 
a b
Q4. Let Q = ∈ M2×2 (R) be an orthogonal matrix.
c d
(a) Show that the entries of A satisfy

a2 + c2 = 1, ab + cd = 0 and b2 + d2 = 1.

(b) Use part (a) to show that there is a θ ∈ R so that A is equal to either
   
cos θ − sin θ cos θ sin θ
or .
sin θ cos θ sin θ − cos θ

[Note: This shows that a 2 × 2 orthogonal matrix is either a rotation or a reflection.]

Q5. Prove/disprove: If A is invertible and unitarily diagonalizable, then A−1 is unitarily diago-
nalizable.

Q6. Prove/disprove: Let L : V → V be an operator on finite-dimensional vector space over R.

Suppose that the matrix [L]B of L with respect to some basis B for V is symmetric. Then
the matrix of L with respect to every basis for V is symmetric.

Q7. Let L : V → V be a linear operator on a finite-dimensional inner product space. Suppose that
there exists a linear operator M : V → V so that ⟨L( #» #» = ⟨ #»
v ), w⟩ #» for all #»
v , M ( w)⟩ v,w#» ∈ V .
Let B be an orthonormal basis for V . Prove that [M ]B = [L]∗B .

Q8. Let L : V → V be a linear operator on a 2-dimensional inner product space V . Prove that
there exists a linear operator M : V → V so that for all vectors #»
v,w#» ∈ V ,

⟨L( #» #» = ⟨ #»
v ), w⟩ #» .
v , M ( w)⟩

[Hint: Use the previous problem as a guide.]

[Note: This is true for any finite-dimensional inner product space V , not just 2-dimensional
ones.]

4
 MATH235: Linear Algebra 2
Practice Problems for Chapter 5: Orthogonal Diagonalization Fall 2022

Q9. Let V be an n-dimensional vector space and W an m-dimensional vector space. Let L : V →
W be a linear map with rank r. Prove that there exists bases B of V and C of W so that
C [L]B = [aij ]
where (
1 if i = j and i ≤ r
aij =
0 otherwise.

Q10. Let B = { #»
v 1 , . . . , #»
v n } and C = { #»
u 1 , . . . , #»
u n } be orthonormal basis for an inner product space
V . Prove that the change of basis matrix Q = C IB is a unitary matrix.
Q11. Let A be an n × n matrix.
(a) Let R(A) = 12 (A + A∗ ) and I(A) = 2i
1
(A − A∗ ). These are called the real and imaginary
parts of A. Prove that R(A) and I(A) are Hermitian matrices.
(b) If n = 1, that is A is a 1 × 1 matrix [a], show that R(A) and I(A) correspond to the
real and imaginary parts of a.
(c) Show that A = R(A) + iI(A).
(d) Prove that A is normal if and only if R(A)I(A) = I(A)R(A).
Q12. (a) Suppose A is a square matrix satisfying A2 = I. Prove that if λ is an eigenvalue of A,
then λ = 1 or λ = −1.
(b) Suppose A is a matrix that is both unitary and Hermitian. Prove all its eigenvalues are
either 1 or −1.
Q13. (a) Let A and B be Hermitian matrices with the property that AB = BA. Prove AB is
Hermitian.
(b) Find Hermitian matrices A and B such that AB is not Hermitian, therefore proving
that the condition AB = BA in part (a) is necessary.
Q14. We saw in the course notes that if A is normal and #»
v is a non-zero vector so that A #»
v = λ #»
v,
∗ #» #»
then A v = λ v .
(a) Show that the assumption that A is normal is necessary. That is, find an example of a
non-normal matrix A and an eigenvector #»v of A, so that #»
v is not an eigenvector of A∗ .
(b) Suppose A ∈ Mn×n (C) is normal, and suppose #» v ∈ Cn is an eigenvector of A. Let
#» #» #»
W = Span({ v }). Show that if w ∈ W ⊥ , then A( w) ∈ W ⊥ .
Q15. Let Q denote the set of all real quadratic forms in the variables x1 , . . . , xn .
(a) Show that the operations
(Q1 + Q2 )( #»
x ) = Q1 ( #»
x ) + Q2 ( #»
x ) for Q1 , Q2 ∈ Q
and
(cQ)( #»
x ) = c(Q( #»
x )) for Q ∈ Q and c ∈ R
turn Q into a vector space over R.

5
 MATH235: Linear Algebra 2
Practice Problems for Chapter 5: Orthogonal Diagonalization Fall 2022

(b) Find dim Q. [Hint: Construct an isomorphism between Q with the subspace of n × n
real symmetric matrices.]

Q16. Let A ∈ Mn×n (R) be a symmetric matrix, and let Q( #» x ) = #»

x T A #»
x be the associated quadratic
form.

(a) Suppose that A = diag(λ1 , . . . , λn ) is diagonal, so that Q( #»

x ) = ni=1 λi x2i . Show that
P
the maximum value of Q( #»x ) subject to the constraint ∥ #» x ∥2 = 1 is λmax = max{λi }, the
largest diagonal entry of A. [Hint: Begin by showing that Q( #» x ) ≤ λmax ∥ #»
x ∥.]
(b) Using the spectral theorem, prove that for general symmetric A, the maximum value of
Q( #»
x ) subject to the constraint ∥ #»
x ∥2 = 1 is the largest eigenvalue of A. What is the
minimum value?

[Note: If you have taken or are taking multivariable calculus, try to solve this problem using
Lagrange multipliers. You should spot an interesting notational coincidence.]