MAS 109 Introduction to Linear Algebra

Final Exam June 15 (Wednesday), 2022 16:00 PM ∼ 18:45 PM

Spring 2022 MAS 109
Full Name :

Student Number:

Class Section: A B C D E F G H

Do not write in this box.

Instruction

• No items other than scratch papers (given by TA), pen, pencil, eraser and your id are al-
lowed. You must leave all of your belongings that are not allowed at the designated area
in the front of the classroom.

• Before you start, fill out the identification section on the title page with an inerasable pen.

• Show your work for full credit, unless noted otherwise. Write the final answer to each
problem in the box provided.

• The exam is for two hours and 45 minutes. Ask permission by raising your hand if you
have any question or need to go to toilet. You are not allowed to go to toilet for the first 30
minutes and the last 15 minutes.

• Any attempt to cheat or a failure to follow this instruction lead to serious disciplinary
actions not limited to failing the exam or the course.

Do not write in this table.

Problem Score Problem Score Problem Score Problem Score
1 / 50 5 / 20 9 / 20 13 / 20

2 / 20 6 / 30 10 / 20 14 / 20

3 / 20 7 / 30 11 / 20 15 / 20

4 / 20 8 / 20 12 / 20 Total / 350

–1–
MAS 109 2022 spring

1. Let A be a square matrix of size n × n. Determine whether each claim below is true or false.
Fully justify your answer. (10pts each)

(a) Elementary row operations do not change row(A).

Ans:

(b) Elementary row operations do not change col(A).

Ans:

–2–
MAS 109 2022 spring

(c) Elementary row operations do not change nullity(AT ).

Ans:

(d) The non-zero row vectors of any row echelon form R of A form a basis of row(A), even
when R is not the reduced row echelon form.

Ans:

–3–
MAS 109 2022 spring

(e) The non-zero column vectors of any row echelon form R of A form a basis of col(A),
even when R is not the reduced row echelon form.

Ans:

–4–
MAS 109 2022 spring

2. Given an m × n matrix A, let B be a basis for the row space of A and let B ′ be a basis for
the kernel of A. Prove that B ∪ B ′ is a basis of Rn . (20pts)

–5–
MAS 109 2022 spring

 
−1 0
3. Let A =  0 1 and bT = [0 0 1]. Solve the followings. (10pts each)
0 1

(a) What is the normal equation associated to the linear system Ax = b?

Ans:

–6–
MAS 109 2022 spring

(b) Find all least squares solutions to this linear system. If there is a unique least squares
solution, indicate that it is unique.

Ans:

–7–
MAS 109 2022 spring

4. Prove or disprove that if P is the standard matrix for an orthogonal projection of Rn onto a
subspace of Rn , then tr(P ) =rank(P ). (20pts)

–8–
MAS 109 2022 spring

5. Let A be the following matrix:  

1 1 1

 1 2 22 

A=
 1 3 32 .

 1 4 42 
1 5 52
Does A have full column rank? If not, find a basis for col(A) consisting of column vectors
of A. Make sure to show your work.

Ans:

–9–
MAS 109 2022 spring

6. Let A be a 4 × 4 matrix given as follows:

 
2 0 0 0
 0 −2 0 2 
A=  0 0 −3 0
.

0 2 0 1

Solve the followings. (10pts each)

(a) Find all eigenvalues of A along with their algebraic multiplicities.

Ans:

– 10 –
MAS 109 2022 spring

(b) Is A orthogonally diagonalizable? If so, find such an orthonormal basis of R4 consisting

of eigenvectors of A. If not, explain why.

Ans:

– 11 –
MAS 109 2022 spring

(c) Find the solution set of Ax = b where b = [200, 150, 300, −75]T .

Ans:

– 12 –
MAS 109 2022 spring

7. Show that a square matrix is nilpotent if and only if 0 is the only eigenvalue. (30pts)

– 13 –
MAS 109 2022 spring

8. Find a column-row factorization (a.k.a. CR decomposition) for each of the following ma-
trices. (10pts each)
 
1 2 3 4
 2 4 6 8 
(a) A =   3 6 9 12 


4 8 12 16

Ans:

– 14 –
MAS 109 2022 spring

 
1 2 3
(b) B =  0 2 4 
0 0 −1

Ans:

– 15 –
MAS 109 2022 spring

9. Let W be a plane through the origin in R3 . Solve the followings. (10pts each)

(a) Show that the standard matrix of the orthogonal projection of R3 onto W is symmetric.

– 16 –
MAS 109 2022 spring

(b) Find the standard matrix P for the orthogonal projection of R3 onto W which is given
by the equation x − y + z = 0.

Ans:

– 17 –
MAS 109 2022 spring

10. Let  
1 0 1
1 0 −1
A=
1
.
1 0
−1 1 1
Solve the following questions. (10pts each)

(a) Find a QR-decomposition of A.

Ans:

– 18 –
MAS 109 2022 spring

 T
(b) Find the least squares solution x̂ of Ax = 1 0 0 1

Ans:

– 19 –
MAS 109 2022 spring

11. Prove the followings. (10pts each)

(a) Let S be a subset of nonzero vectors in Rn consisting of orthogonal vectors. Show that
S is linearly independent.

– 20 –
MAS 109 2022 spring

(b) If u1 , u2 , . . . , um are eigenvectors of a matrix A that correspond to distinct eigenvalues

λ1 , λ1 , . . . , λm , then the set {u1 , u2 , . . . , um } is linearly independent.

– 21 –
MAS 109 2022 spring

12. Solve the followings. (10pts each)

(a) Let W be a subspace of Rn . Show that the subset U of all vectors in Rn , that are
orthogonal to all vectors in W forms a subspace of Rn .

– 22 –
MAS 109 2022 spring

(b) Let W be a subspace of R5 spanned by the vectors u1 = (2, 2, −1, 0, 1),

u2 = (1, 1, −2, 0, −1) and u3 = (0, 0, 1, 1, 1). Find a basis for a subspace U of R5 .

Ans:

– 23 –
MAS 109 2022 spring

13. Suppose a SVD of an invertible square matrix A is A = U ΣV T . Give a SVD of AT and A−1
in terms of U, Σ, V . (20pts)

Ans:

– 24 –
MAS 109 2022 spring

 
−1 1
14. Find a singular value decomposition of A = . (20pts)
−1 −1

Ans:

– 25 –
MAS 109 2022 spring

15. We will see how much the column space of a (singular) symmetric matrix A is tilted by
a perturbation, leading to a symmetric perturbed matrix B. An eigenvector correspond-
ing to the largest eigenvalue is called a dominating eigenvector. The following MATLAB
code computes the angle between the dominating eigenvectors of A and B, and the angle
between the column space of A and the dominating eigenvector of B. Fill in the blanks
(1)-(9). (20points in total, 3+1+3+2+2+1+3+2+3 pts)

% Input: a (singular) symmetric n by n matrix A.

n = size(A,1);
E = ___(1)___; % Generate an n by n matrix with random numbers
% from a standard normal distribution.
epsilon = 0.001;
B = A + epsilon*1/2*(E+E’);

% Compute vA and vB that are eigenvectors

% corresponding to the largest eigenvalues of A and B, respectively.
vA = final_MAS109(A, 1); vB = final_MAS109(B, 1);

% Compute cos(theta), where theta is an angle between vA and vB.

cos_theta = ___(2)___;

% Compute CA whose columns are orthonormal eigenvectors of A

% that span the column space of A
CA = ___(3)___;

% Express the vector vB as a sum of two orthogonal terms vB1 and vB2
vB1 = ___(4)___; % orthogonal projection of vB onto the column space of A
vB2 = ___(5)___; % orthogonal projection of vB onto the null space of A

% Compute cos(phi), where phi is an angle between vB

% and the column space of A
cos_phi = ___(6)___;

See the next page!

– 26 –
MAS 109 2022 spring

This page’s codes are the continuation of the previous page’s codes.

% Define a function with the name final_MAS109 that outputs a matrix CA,
% whose columns are orthonormal eigenvectors corresponding to k largest
% eigenvalues of A
function ___(7)___
[V, D] = eig(A); % If A is symmetric, it outputs eigenvalues
% and corresponding orthonormal eigenvectors of A
lambdas = ___(8)___; % eigenvalues in a vector form

% Sort eigenvalues in a descending order

[sorted_lambdas index] = sort(lambdas, ’descend’);
% Ex) if lambdas = [10 -5 4],
% then sorted_lambdas = [10 4 -5] and index = [1 3 2]

CA = ___(9)___;
end

(1) (2) (3) (4) (5)

(6) (7) (8) (9)

– 27 –
MAS 109 2022 spring

– 28 –