Solutions for Mid-Semester Exam

Question 1.
(a) (5 marks) Find the values of x for which the following is an augmented
matrix corresponding to a consistent system:
 
1 −2 1 x
0 5 −2 x2 
4 −23 10 x3

(b) (5 marks ) Determine the RREF of the matrix formed by substituting

x with π in the matrix in part (a).

Solution.
(a) We reduce the given augmented matrix to echelon form:
   
1 −2 1 x 1 −2 1 x
R →R −4R1
0 5 −2 x2  −−3−−−3−−−→ 0 5 −2 x2 
4 −23 10 x 3 0 −15 6 x − 4x 3

 
1 −2 1 x
R →R +3R2
−−3−−−3−−−→ 0 5 −2 x2 
0 0 3
0 x − 4x + 3x 2

The corresponding linear system is consistent if and only if the aug-

mented column is not a pivot column. This condition holds if is x is a
root of the polynomial x3 + 3x2 − 4x. Now

x3 + 3x2 − 4x = x(x2 + 3x − 4)
= x(x + 4)(x − 1)

Therefore the given matrix is an augmented matrix corresponding to a

consistent linear system when x = 0, x = −4 or x = 1.
(b) We find the RREF of the matrix
 
1 −2 1 π
0 5 −2 π 2 
4 −23 10 π 3

Using part (a), we know that

   
1 −2 1 π 1 −2 1 π
0 2
5 −2 π ∼ 0   5 −2 π2 
4 −23 10 π 3 0 0 3 2
0 π + 3π − 4π

We continue with the row reduction process to reduce the matrix to

RREF:
1 −2
 
  1 π
1 −2 1 π π2
R → 51 R2 0
 −−2−−− 1 − 25 
0 5 −2 π2 −→ 

5 

0 0 0 π 3 + 3π 2 − 4π 3 2
0 0 0 π + 3π − 4π
 
1 2π 2
1 0 5 π + 5
 
 
R1 →R1 +2R2  2
−−−−−−−−→ 0 1 − 25 π 
5

 
 
0 0 3
0 π + 3π − 4π 2
 2

1
1 0 5 π + 2π5
1
R3 →
 
R3 
π + 3π 2 − 4π
3 
2 π 2
−−−−−−−−−−−−−−−−→ 0 1 −5
 
5

 
 
0 0 0 1
1 2
π + 2π5

2 1 0 5
π
R2 →R2 − R3  
5
−−−−−−−−−−→ 0

2

 1 −5 0  
 
0 0 0 1
 1 
2π 2 1 0 0
 
5
R1 →R1 −π+ R 3 
5

− 25
 
−−−−−−−−−−−−−−−→  0 1 0
 
0 0 0 1
Question 2.
(a) (5 marks) Let  
1 2 x
A = 0 3 4 
0 0 5
Express the inverse of A as a function of x (i.e. a matrix whose entries are
functions of x), without calculating the determinant or using Cramer’s
rule.
(b) Is the span of the columns of A−1 all of R3 (YES/NO) ? Justify your
answer briefly.
(Note for Section A students: You may assume that x is a fixed scalar in
Part (b).)

Solution.
(a) We reduce [A I] to RREF:
 
  1 2 x 1 0 0
1 2 x 1 0 0 R →1R   
0 3 4 0 1 0 −−2−−− 3 2 
−→ 0 1 34 1

0 3 0 
0 0 5 0 0 1  
0 0 5 0 0 1

1 0 x − 83 1 − 23
 
0
 
R →R −2R2  4
0 13

−−1−−−1−−−→ 0 1
 3 0
 
0 0 5 0 0 1
1 0 x − 38 1 − 23
 
0
 
R3 → 51 R3 
4
0 31

−−−−−−→  0 1 3 0
 
1
0 0 1 0 0 5

1 0 x − 38 1 − 23
 
0
 
R2 →R2 − 4 R3 
0 31 4 

−−−−−−−3−→  0 1 0 − 15 
 
1
0 0 1 0 0 5
 1 0 0 1 − 23 8−3x 

15
 
R1 →R1 −(x− 38 )R3 
1 4

−−−−−−−−−−−−→  0 1 0 0 3 − 15 

 
1
0 0 1 0 0 5

Thus
1 − 23 8−3x 

15
 
−1 1 4
 
A 0 3
= − 15 

 
1
0 0 5

(b) We first note that A−1 exists for every choice of x ∈ R. Having fixed
x ∈ R, the span of the columns of A−1 is all of R3 . This may be justified
in any of the following ways:

Method 1. For any vector b ∈ R3 ,

A−1 (Ab) = b.

Therefore b can be expressed as a linear combination of the

columns of A−1 using the entries of Ab as coefficients. As
the choice of b was arbitrary, the span of the columns of A−1
is R3 .
Method 2. Since A−1 is invertible, it follows by the Invertible Matrix
Theorem that for any vector b ∈ R3 , the equation

A−1 x = b

has a solution. Therefore every b ∈ R3 can be expressed as

a linear combination of the columns of A−1 . Hence the span
of the columns of A−1 is R3 .
Method 3. For any vector b ∈ R3 , the equation

A−1 x = b

has a solution because the last columns of the augmented

matrix [A−1 b] is not a pivot column. Therefore every
b ∈ R3 can be expressed as a linear combination of the
columns of A−1 . Hence the span of the columns of A−1 is
R3 .
 Method 4. Since A−1 has a pivot position in every row, it follows by
Theorem 4 on p. 43, Section 1.4 of the course textbook (3rd
edition), that the columns of A−1 span R3 .

Question 3. Let
V = {x ∈ R : x > 0},
and define addition for V by

x ⊕ y := xy

and scalar multiplication by any α ∈ R by

α ∗ x = xα .

(a) (7 marks) Verify the closure axioms, the commutative, zero and inverse
properties for addition, and the property 1 ∗ x = x for all x ∈ V .

(Remark: V is in fact a vector space over the field R. You need not
verify the other properties of a vector space.)

(b) (3 marks) Is V a subspace of R regarded as a vector space over itself

(YES/NO) ? Justify your answer clearly.

Solution.

(a) • First closure axiom:

Let x, y ∈ V . Then xy ∈ R and x > 0, y > 0 =⇒ xy > 0.
Therefore xy ∈ V =⇒ x ⊕ y ∈ V .
• Second closure axiom:
Let x ∈ V and α ∈ R. As x > 0 and x ∈ R, it follows that xα ∈ R
and xα > 0 (A proof of this fact is outlined in the first chapter of
the text “Principles of Mathematical Analysis” by W. Rudin - p.
22, 3rd edition). Hence xα ∈ V =⇒ α ∗ x ∈ V .
• Commutativity of addition:
Let x, y ∈ V . Then

x ⊕ y = xy = yx = y ⊕ x.
• Existence of additive identity:
For every x ∈ V ,
x⊕1=x×1=x
Hence 1 is a zero vector in V .
• Existence of additive inverse:
Let x ∈ V . Since x1 > 0 it follows that 1
x ∈ V . Further

1
x⊕ = 1.
x
1
Hence x is an additive inverse of x in V .
• Lastly, let x ∈ V . Then 1 ∗ x = x1 = x.
(b) When we consider the question whether V is a subspace of R (with
the usual vector space structure on R), we strip V of the vector space
structure defined in part (a) and view it simply as a subset of R. In
order for V to be a subspace of R it must be a vector space under the
usual operations of vector addition and scalar multiplication
which are defined on R.

Thus the answer to our question is the following.

No, V is not a subspace of R, when R is regarded as a vector space over

itself.

One reason is that V is not closed under scalar multiplication. Any of

the following counterexamples can be cited.

1. Let x ∈ V . Then 0 × x = 0, but 0 ∈

/ V.
2. Let x ∈ V . Then −1 × x = −x, but −x ∈
/ V.
3. Let x ∈ V . Let c ∈ R be chosen such that c ≤ 0. Then

x > 0 =⇒ cx ≤ 0.

Hence cx ∈
/ V.
4. 13 ∈ V, −13 ∈ R, but −169 ∈
/ V.

Alternatively, the fact that 0 ∈

/ V can be used as a standalone justifica-
tion. In this case, V is not a subspace of R, by the definition listed on
p. 220 of the course textbook (3rd edition).

Question 4 (10 marks). Choose any four of the five sets below. For each
set you choose, state whether or not it is a subspace of M3×3 (the space of
all 3 × 3 matrices having real entries). Justify each answer. All choices carry
equal marks.
(a) The set of all invertible 3 × 3 matrices

(b) The set of all 3 × 3 matrices whose trace is 0 (The trace of a square
matrix A is the sum of its diagonal entries.)

(c) The set of all 3 × 3 echelon matrices

(d) The set of all symmetric 3 × 3 matrices (A square matrix A is said to

be symmetric if AT = A)
(e) The set of all skew-symmetric 3 × 3 matrices (A square matrix A is said
to be skew-symmetric if AT = −A)
(Note for Section B students: M3×3 is the same as R3×3 ).

Solution.
(a) Let V be the set of all invertible 3 × 3 matrices. V is not a subspace of
M3×3 . This can be justified in any of the following ways:
1. V is not closed under vector addition. For example, I ∈ V, −I ∈ V ,
but I − I = 0 and 0 ∈/ V.
2. V is not closed under scalar multiplication. If c = 0, then I ∈ V but
cI = 0 and cI ∈/ V.
3. 0 ∈
/ V . Thus V is not a subspace of M3×3 , by the definition listed on
p. 220 of the course textbook (3rd edition).
(b) Let V be the set of all 3 × 3 matrices whose trace is 0. V is a subspace
of M3×3 . This is verified as follows:
Let A, B ∈ V . Let A = (aij ) and B = (bij ). Then
3
X
trace(A + B) = aii + bii
i=1
3
X 3
X
= aii + bii
i=1 i=1
= trace(A) + trace(B)
=0
Hence A + B ∈ V . Thus V is closed under vector addition.
Let A ∈ V and c ∈ R. Then
3
X
trace(cA) = caii
i=1
X3
=c aii
i=1
= c trace(A)
=0
Hence cA ∈ V . Thus V is closed under scalar multiplication.
(c) Let V be the set of all 3 × 3 echelon matrices. V is not a subspace of
M3×3 . This is because V is not closed under vector addition. Coun-
terexamples abound. One is given here for your reference: Let
 
1 0 0
A = 0 −1 0
0 0 1

Then A ∈ V and I ∈ V . However

 
1 0 0
A + I = 0 0 0 ∈ /V
0 0 1

because A + I has a row of zeros in the middle, which violates the

conditions for being an echelon matrix.

(d) Let V be the set of all symmetric 3 × 3 matrices. V is a subspace of

M3×3 . This is verified as follows:

Let A, B ∈ V . Then

(A + B)T = AT + B T
=A+B

Hence A + B ∈ V . Thus V is closed under vector addition.

Let A ∈ V and c ∈ R. Then

(cA)T = cAT
= cA

Hence cA ∈ V . Thus V is closed under scalar multiplication.

(e) Let V be the set of all skew-symmetric 3 × 3 matrices. V is a subspace

of M3×3 . This is verified as follows:

Let A, B ∈ V . Then

(A + B)T = AT + B T
= −A − B
= −(A + B)
 Hence A + B ∈ V . Thus V is closed under vector addition.

Let A ∈ V and c ∈ R. Then

(cA)T = cAT
= c(−A)
= −cA

Hence cA ∈ V . Thus V is closed under scalar multiplication.

Question 5. (10 marks) Let V be a vector space over a field F . Suppose

v1 , v2 , . . . , vn are linearly independent in V and w ∈ V .
Show that if v1 + w, v2 + w, . . . , vn + w are linearly dependent in V , then
w ∈ Span{v1 , v2 , . . . , vn }.
(Note for Section A students: You may assume that F = R.)

Solution.

Method 1. As v1 + w, v2 + w, . . . , vn + w are linearly dependent in V , there

exist scalars c1 , . . . , cn ∈ R, not all zero, such that
n
X
0= ci (vi + w)
i=1
n n
! (1)
X X
= ci vi + ci w
i=1 i=1
Pn
Put c = i=1 ci . Suppose if possible that c = 0. Then (1) gives
us
n
X
ci vi = 0.
i=1

As v1 , v2 , . . . , vn are linearly independent in V , it follows that

ci = 0 for i = 1, . . . , n. Contradiction.
1
Therefore c 6= 0. Now multiply both sides of (1) by c to obtain
n
1X
w+ ci vi = 0
c
i=1
 Hence
n
X ci
w= vi ∈ Span{v1 , v2 , . . . , vn }.
c
i=1

Method 2. The following result was proved in class for both Section A as
well as Section B:

Let {v1 , . . . , vn } be a linearly independent set in a vector space

V . If w ∈ / Span({v1 , . . . , vn }) then the set {v1 , . . . , vn , w} is
linearly independent.

Using the above result, it suffices to show that the set {v1 , . . . , vn , w}
is not linearly independent in V .

As v1 + w, v2 + w, . . . , vn + w are linearly dependent in V , there

exist scalars c1 , . . . , cn ∈ R, not all zero, such that
n
X
ci (vi + w) = 0
i=1

Hence
n n
!
X X
ci vi + ci w = 0. (2)
i=1 i=1

As c1 , . . . , cn are not all zero, (2) is a linear dependence relation

between v1 , . . . , vn , w.

Method 3. We prove the contrapositive of the given statement, i.e.

w∈
/ Span{v1 , v2 , . . . , vn } =⇒ {v1 + w, v2 + w, . . . , vn + w} is l.i.

So, assume that w ∈

/ Span{v1 , v2 , . . . , vn }. Suppose
n
X
ci (vi + w) = 0,
i=1
Pn
where c1 , . . . , cn ∈ R. Put c = i=1 ci . Then
n
X
cw = − ci vi (3)
i=1
Were c 6= 0, we would obtain w ∈ Span{v1 , v2 , . . . , vn } by mul-
tiplying both sides of (3) by 1c , which would then contradict our
assumption. Therefore c = 0. Therefore
n
X
ci vi = 0
i=1

Now, since v1 , v2 , . . . , vn are linearly independent in V , it follows

that ci = 0 for i = 1, . . . , n. Thus the vectors v1 + w, . . . , vn + w
are linearly independent in V .