MATH-225 Midterm Exam — Solutions

Student’s surname: ................................................................

Student’s name: ................................................................

Student’s ID number: ..............................................

Course Section: ......

Department: ............................

Date: Saturday 20.11.2021

Time: 10:00–12:00

Classrooms: B-204 B-Z01 B-Z02 B-Z04 B-Z05 B-Z08 EE-03 EE-04 EE-05

1. This exam consists of 4 questions of 25 points each. Some questions are divided in sub-
questions, for which the corresponding number of points awarded is specified. All questions
are mandatory.
2. Each question is written on a separate sheet of paper. Please read each question carefully
and write your answers in the space below each question. BE NEAT.
3. SHOW ALL YOUR WORK. Correct answers without sufficient explanation might not
get full credit.
4. You may use the back side of the page, in which case you should specify the continuation.
No scratch paper is provided. You are not allowed to use any calculator and any other
electronic devices.

Do not write anything below this line.

1 2 3 4 TOTAL

25 25 25 25 100
 Question 1: (25 points) Consider the following system of linear equations

kx + y
 =a
3x − 2y + (k − 3)z = b

10kx + 3y − 2z = 15


where a, b and k are constants.

(a) (10 points) Find all the values of k for which the system has a unique solution.
(b) (10 points) For k = 2, find a in terms of b if the system is to be consistent.
(c) (5 points) Find the solution of the system if a = 1, b = 4 and k = 2.
Solutions (a) To have a unique solution, we need det(A) 6= 0 where A is the coefficient
matrix of the system.

k 1 0
−2 k − 3 3 k−3
det(A) = 3 −2 k − 3 = k −
3 −2 10k −2
10k 3 −2
= k(13 − 3k) − (30k − 10k 2 − 6) = (k − 2)(7k − 3)

Then, the system admits a unique solution for k ∈ R − {2, 73 }.

(b) The augmented coefficient matrix with k = 2 can be reduced to an echelon form as
follows
     
2 1 0 a 1 1/2 0 a/2 1 1/2 0 a/2
1/2R1 −3R1 +R2
 3 −2 −1 b  − −−→  3 −2 −1 b  −−−− −−→ 0 −7/2 −1 b − 3a/2 
−20R1 +R3
20 3 −2 15 20 3 −2 15 0 −7 −2 15 − 10a
 
1 1/2 0 a/2
−2R +R3
−−−2−−→ 0 −7/2 −1 b − 3a/2  = E
0 0 0 15 − 7a − 2b

Then, to have a consistent system, we have

1
15 − 7a − 2b = 0 ⇒ a = (15 − 2b).
7
(c) Substituting a = 1 and b = 4 in the above echelon matrix E, we have
 
1 1/2 0 1/2
E = 0 −7/2 −1 5/2
0 0 0 0

Hence, the system has infinitely many solutions and the solution takes the form
 
6 1 5 2
(x, y, z) = + t, − − t, t .
7 7 7 7

2
 Question 2: Let A be an invertible 3 × 3 matrix. We reduce A to the identity matrix
by applying the elementary row operations (1/2)R2 , −2R3 + R1 and (3/4)R3 + R2 , in this
order.
(a) (10 points) Find the elementary matrices E1 , E2 and E3 corresponding, respectively,
to these elementary row operations.
(b) (15 points) Find A and A−1 .
Solution. (a) We have the following elementary matrices:
 
1 0 0
1
R2 corresponds to E1 = 0 21 0
2
0 0 1
 
1 0 −2
−2R3 + R1 corresponds to E2 = 0 1
 0
0 0 1
 
1 0 0
3 3
R3 + R2 corresponds to E3 = 0 1

4
.
4
0 0 1

(b) We have
E3 E2 E1 A = I3 ,
hence  
1 0 −2
A−1 = E3 E2 E1 = 0 21 34 
0 0 1
and
   
1 0 0 1 0 2 1 0 0
A = E1−1 E2−1 E3−1 = 0 2 0 0 1 0 0 1 − 43 
0 0 1 0 0 1 0 0 1
 
1 0 2
= 0 2 − 23  .
0 0 1

3
 Question 3: Assume that the following conditions hold:
(i) A = [ a1 a2 a3 a4 ] is a 3 × 4-matrix in echelon form, where a1 , a2 , a3 , a4 are column
vectors.
(ii) B = [ b1 b2 b3 b4 ] is a 3 × 4-matrix in echelon form, where b1 , b2 , b3 , b4 are column
vectors.
(iii) c1 , c2 , c3 are real numbers.
(iv) c3 6= 0.
(v) c1 a1 + c2 a2 + c3 a3 = 0.
(vi) dim(Span{b1 , b2 , b3 , b4 }) ≥ 3.
(vii) The matrices A and B are row equivalent but not equal.
Then answer the following questions:
(1) (5 points) Is it possible to have c2 = 0?
(2) (5 points) Is it possible to have c2 6= 0?
(3) (5 points) Is it possible to have the row rank of [ a1 a2 a3 ] equal to 1?
(4) (10 points) Is it possible to have the row vector [ 0 0 1 1 ] in the row space of B?
(WARNING: If your answer is ”Yes!” then give an example for such matrices A, B and
numbers c1 , c2 , c3 . If your answer is ”No!” then explain why it is impossible.)
Solution. (1) Yes! Consider
   
1 0 1 0 1 0 1 0
A= 0 1
 0 0 , B = 0 1 0 1 , c1 = 1, c2 = 0, c3 = −1.
0 0 0 1 0 0 0 1
(2) Yes! Consider
   
1 0 1 0 1 0 1 0
A = 0 1 1 0 , B = 0 1 1 1 , c1 = 1, c2 = 1, c3 = −1.
0 0 0 1 0 0 0 1
(3) No! Assume that the row rank of [a1 a2 a3 ] is 1, hence its column rank is 1, hence
Span({a1 , a2 , a3 }) = Span({aj }) for some j ∈ {1, 2, 3}. But then Span({a1 , a2 , a3 , a4 }) =
Span({aj , a4 } has dimension ≤ 2 and dim(Span({a1 , a2 , a3 , a4 }) = dim(Span({b1 , b2 , b3 , b4 }) ≥
3, a contradiction.
(4) No! Assume that E is the result of the Gauss-Jordan algorithm applied to A. We
know that dim(Col(A)) = dim(Row(A)) = dim(Row(B)) = dim(Col(B)) ≥ 3. By (v) and
(iv) we know that
−c1 −c2
a3 = a1 + a2 ,
c3 c3
so E must have three leading entries and E has no leading entry in the 3rd column. So,
 
1 0 x 0
E = 0 1 y 0
0 0 0 1
for some x, y ∈ R. Consequently, the vector [0 0 1 1] is not in Row(E) = Row(A) =
Row(B).

4
 Question 4: Let v1 = (1, 0, 3, −2, 3), v2 = (0, 1, −2, 1, −3) and v3 = (2, −1, 8, −5, 9) be
vectors in R5 . Let W = Span({v1 , v2 , v3 }) be a subspace of R5 .
(a) (10 points) Find a bases of W .
(b) (10 points) Find a bases of the orthogonal complement W ⊥ of W .
(c) (5 points) Let v = (1, −1, 2, −2, 4) be a vector in R5 . Show that v is neither in W
nor in W ⊥ .
Solution: (a): Let W = Span({v1 , v2 , v3 }). Let A be a 3 × 5 matrix whose row vectors
are v1 , v2 and v3 . Hence W = Row(A). To find a basis of Row(A) we apply a sequence of
elementary row operations on A and obtain it’s Echalon matrix E.
 
1 0 3 −2 3
A= 0 1 −2 1 −3  (1)
2 −1 8 −5 9
Applying −2R1 + R3 we get  
1 0 3 −2 3
 0 1 −2 1 −3  (2)
0 −1 2 −1 3
Applying now R2 + R3 we the Echalon matrix
 
1 0 3 −2 3
E =  0 1 −2 1 −3  (3)
0 0 0 0 0
Hence v1 and v2 are independent and form a basis of Row(A) and hence a basis of W . The
dimension of W is 2.

(b): The orthogonal complement W ⊥ of W is the null space of A, Null(A) which is obtained
by solving the system of equations Ax = 0. Since Ex = 0 has the same solution space we
have
x1 + 3x3 − 2x4 + 3x5 = 0, (4)
x2 − 2x3 + x4 − 3x5 = 0 (5)

Leading variables: x1 and x2 .

Free variables: x3 = t, x4 = r, and x5 = s where t, r, s are any real numbers.
We find the leading variables in terms of t, r, s
x1 = −3t + 2r − 3s, (6)
x2 = 2t − r + 3s (7)
Hence the general solution x is
   
x1 −3t + 2r − 3s
 x2  
   2t − r + 3s 

x=  x 3
=
  t  = tu1 + ru2 + su3
 (8)
 x4   r 
x5 s

5
where      
−3 2 −3

 2 


 −1 


 3 

u1 = 
 1  , u2 = 
  0  , u3 = 
  0 
 (9)
 0   1   0 
0 0 1
Then a bases of Null(A) or W ⊥ is the set {u1 , u2 , u3 }. The dimension of W ⊥ is 3.

(c): Is v orthogonal to W ? No because v · v1 = 23, hence v is not in W ⊥ . Is v orthogonal

to W ⊥ ? No, because v · u1 = −3, hence it can not be in W either.