University of Toronto Scarborough

Department of Computer & Mathematical Sciences

MAT A23H 2014/2015

Solutions #3

1. (Fraleigh & Beauregard, pages 46 – 48)

 
  2 −1  
4 1 −2  14 −2
(16) B C = 0 6 = .
5 −1 3 1 −5
  −3 2  
2 −1   3 3 −7
4 1 −2
CB = 0 6  =  30 −6 18 .
5 −1 3
−3 2 −2 −5 12
   2   
2 0 0 2 0 0 4 0 0
(17) A =  0 −1 0 . (a) A2 =  0 (−1)2 0  =  0 1 0 .
0 0 1 0 0 12 0 0 1
 7   
2 0 0 128 0 0
7 7
(b) A =  0 (−1) 0 =  0 −1 0 .
7
0 0 1 0 0 1
      
0 0 −1 0 0 −1 0 0 −1 −2 0 0
(18) A =  0 2 0 . (a) A2 =  0 2 0  0 2 0 = 0 4 0 
2 0 0 2 0 0 2 0 0 0 0 −2
   
(−2)3 0 0 0 0 −1
7 2 3 3
(b) A = (A ) A =  0 4 0   0 2 0 =
3
  0 0 (−2)  2 0 0
−8 0 0 0 0 −1 0 0 8
 0 64 0   0 2 0 =
  0 128 0 .
0 0 −8 2 0 0 −16 0 0
 
  4  

(19) x y = −2 3 −1  −1  = −14 = −14
  3  
4   −8 12 −4
y x =  −1  −2 3 −1 =  2 −3 1 .
3 −6 9 −3
(21) (a) True; (b) False; (c) False; (d) True; (e) False; (f) True; (g) True; (h) True;
(i) False; (j) True.
(23) We know that if v, w are column vectors with n components we can write the dot
product v·w as v T w in matrix notation. (Since v·w = w·v we could even use
w·v.) For the A, b and c of this question we have (A b)·(A c) = (A b)T (A c) =
(bT AT ) (A c).
MATA23H Solutions # 3 page 2

(24) Using the hint, c A = ((c A)T )T = (AT cT )T . We know that AT cT is a linear
combination of the column vectors of AT where the coefficient of the j th column
vector is cj . Since the transpose of a linear combination of column vectors is a
linear combination of row vectors with the same coefficients, we see that c A =
(AT cT )T is the linear combination of the row vectors of A, where the coefficient
of the ith row vector is ci .
     
(25) Let A = aij , B = bij ∈ Mn,k (R).  Now A + B  = a ij + b ij ∈ Mn,k (R). Since
addition in R commutes, aij + bij = bij + aij = B + A. Since B + A is also
n × k, we have A + B = B + A.
 
(27) Let A = aij ∈ Mn,k (R) and consider the the ij th entry of (r + s) A, (r + s) aij .
Now (r + s) aij = r aij + s aij , which is the ij th entry in r A + s A. Since both
(r + s) A and r A + s A ∈ Mn,k (R) we have (r + s) A = r A + s A.
(29) Let A ∈ Mn,k (R) and let B, C ∈ Mk,p (R). Now the ij th entry of A(B + C)
is (ith row of A) · (j th column of B + C) =(ith row of A) · (j th column of B) +
(ith row of A) · (j th column of C) (using the 2nd property of the dot product). This
is just the ij th entry in A B + A C. Since A(B + C) and A B + A C ∈ Mn,p (R),
they are equal.
  T
  T T
 T
(30) If
 A = a ij ∈ M n,k (R), then A = a ji ∈ M k,n (R), so (A ) = aji =
T T
aij ∈ Mn,k (R). Hence (A ) = A.
     
(31) Let A = aij , B = bij  ∈ M n,k
   (R). Now A + B = a ij + b ij ∈ Mn,k (R), so
(A + B)T = aji + bji = aji + bji = AT + B T ∈ Mk,n (R)
     
(32) Let A = aiℓ ∈ Mn,k (R), B = bℓj ∈ Mk,p (R) so A B = cij ∈ Mn,p (R). Mow
Xk X k
T
 
(A B) = cji , where cji = aiℓ bℓj = bℓj aiℓ , but this is just the (j th column
ℓ=1 ℓ=1
of B)·(ith row of A) = (j th row of B T )·(ith column of AT ), which is the jith entry
of B T AT . Since both (A B)T and B T AT ∈ Mp,n (R) we have (A B)T = B T AT .
(35) If B ∈ Mm,n (R) and B = AT then
(a) A ∈ Mn,m (R).
(b) A AT ∈ Mn,n (R).
(c) AT A ∈ Mm,m (R).
(36) Since v, w are column vectors in Rn we can regard v as an n × 1 matrix and
wT as an 1 × n matrix. Hence v wT is n × n. Now (v wT )T = (wT )T v T = w v T
and (w v T )T = (v T )T wT = v wT , we have that v wT and w v T are transposes of
each other.
   
1 1
= hji = HnT . Since Hn = HnT , Hn
   
(37) Now Hn = hij = =
i+j−1 j+i−1
is a symmetric matrix.
(38) Since A is a square matrix, AT is also square and of the same size so A + AT is
defined. Now (A + AT )T = AT + (AT )T = AT + A = A + AT ; i.e., A + AT is
symmetric.
MATA23H Solutions # 3 page 3

(39) Let A ∈ Mn,k (R), then AT ∈ Mk,n (R) and A AT is defined. Now (A AT )T =
(AT )T AT = A AT ; i.e., A AT is symmetric.
(40) (a) (A2 )T = (A A)T = AT AT = (AT )2 and (A3 )T = (A A2 )T = (A2 )T AT =
(AT )2 AT = (AT )3 (using #32).
(b) The generalization is An )T = (AT )n , n ≥ 1. To prove, we will use two steps.
1. base step (n = 1): where we have AT = AT .
2. inductive step (assume true for n = k and show true for n = k + 1): Now
inductive
(Ak+1 )T = (A Ak )T = (Ak )T AT = (AT )k AT = (AT )k+1 .
hypothesis
Hence it is true for all n
 
(41) (a) Let A = aij ∈ Mm,n (R). Now the j th component of A ej can be regarded as
the (ith row of A)·(ej ) = ai1 , ai2 , · · · , aij , · · · , ain ·ej = aij , j 
= 1, 2, ·· · , n,
a1j
 a2j 
since only the j th component of ej is non-zero. Hence A ej =  .. , the
 
 . 
am j
th
j column vector of A.
(b) (i) If A x = 0 for all x we have A ej = 0 for j = 1, 2, · · · , n. Now part (a)
=⇒ that the j th column of A is 0 for all j = 1, 2, · · · , n. Hence A = O,
the zero matrix.
(ii) (A x = B x, for each x) if and only if ((A − B) x = 0, for each x) if and
only if (A − B = O, by part (b)(i)) if and only if (A = B).
(42) (A + B)2 = (A + B)(A + B) = A2 + B A + A B + B 2 = A2 + 2 A B + B 2 if and
only if A B = B A. We already know that A B is generally not equal to B A, so
we seek acounterexample:
      
1 0 0 1 0 1 1 0
Let A = , B= so A B = and B A = . Now
1 0  0 0 0 1    00 
2 2 1 2 2 1 0 0 2 0 0
(A + B) = and A + 2 A B + B = + + =
   1 1  1 0 0 2 0 0
1 2 2 1
6= .
1 2 1 1
1 1
(44) We will try writing A as A = (A + AT ) + (A − AT ). From #38 we know that
2 2
 T
1 1 1 1
(A+AT ) is symmetric. Now (A − AT ) = (AT −(AT )T ) = (AT −A) =
2 2 2 2
1 T
− (A − A ), so this matrix is skew-symmetric. Hence A is written as the sum
2
of a symmetric matrix and a skew-symmetric matrix.
If A = B + C is a 2nd sum, with B symmetric and C skew-symmetric, we have
AT = (B + C)T = B T + C T = B − C. Adding the two expressions, we have
1
A + AT = 2B =⇒ B = (A + AT ) and subtracting we have A − AT = 2 C =⇒
2
MATA23H Solutions # 3 page 4

1
C = (A − AT ). Hence the decomposition is unique.
2
       
2 0 0 1 0 1 2 0 2 1 0 1 2 0 0
(46)  0 r 0   0 1 0  =  0 r 0  and  0 1 0   0 r 0  =
 0 0 2  1 0 1 2 0 2 1 0 1 0 0 2
2 0 2
 0 r 0 . Since both products are equal for all r, we say they commute
2 0 2
for all r ∈ R.
   
2. Let A = aij , B = bij ∈ Mn (R).
 
(a) Since A and B are square matrices, so is A + B = aij + bij . Now tr(A + B) =
(a11 + b11 ) + (a22 + b22 ) + · · · + (ann + bnn ) = (a11 + a22 + · · · + ann ) + (b11 +
b22 + · · · + bnn = tr(A) + tr(B).
 
(b) If λ ∈ R, λ A = λ aij and tr(λ A) = λ a11 + λ a22 + · · · + λ ann = λ (a11 +
a22 + · · · + ann ) = λ tr(A).
 T  
(c) Since AT = aij = aji . tr(AT ) = a11 + a22 + · · · + ann = tr(A).
 Xn 
(d) Since A and B are n × n matrices, A B = cij = aiℓ bℓj and B A =
ℓ=1
 n
X 
dij = bik akj are also n × n matrices. Hence tr(A B) = c11 + c22 + · · · + cnn =
k=1
(a11 b11 + a12 b21 + · · · + a1n bn1 ) + (a21 b12 + a22 b22 + · · · + a2n bn2 ) + · · · + (an1 b1n +
an2 b2n + · · · + ann bnn . Now tr(B A) = d11 + d22 + · · · + dnn = (b11 a11 + b12 a21 + · · · +
b1n an1 ) + (b21 a12 + b22 a22 + · · · + b2n an2 ) + · · · + (bn1 a1n + bn2 a2n + · · · + bnn ann ) =
(a11 b11 + a21 b12 + · · · an1 b1n ) + (a12 b21 + a22 b22 + · · · + an2 b2n ) + · · · (a1n bn1 + a2n bn2 +
collect all f irst terms
· · · ann bnn ) = (a11 b11 + a12 b21 + · · · + a1n bn1 ) + (a21 b12 + a22 b22 +
second terms etc
· · · + a2n bn2 ) + · · · + (an1 b1n + an2 b2n + · · · + ann bnn = tr(A B).

3. (Fraleigh & Beauregard, pages 68 – 71)

     
2 4 −2 R1 → 21 R1 1 2 −1 R2 →R2 −4 R1 1 2 −1
(2) (a)  4 8 3   4 8 3  R3 →R3 +R1  0 0 7 
−1 −3 0 ∼ −1 −3 0 ∼ 0 −1 −1
 
1 2 −1
R2 ↔R3
 0 −1 −1 .
∼
0 0 7
(b) Since
 thereis a pivot in each column we conclude that this matrix reduces to
1 0 0
 0 1 0 .
0 0 1
(It is also correct to continue to row reduce.)
MATA23H Solutions # 3 page 5

     
0 2 −1 3 1 1 −3 3 1 1 −3 3
R2 →R2 +R1
 −1 1 2 0  R1 ↔R3  −1 1
 2 0  R4 →R4 −R1  0 2 −1 3 
 
(3) (a) 
 1 1 −3 3  ∼  0 2 −1 3   0 2 −1 3 
∼
1 5 5 9 1 5 5 9 0 4 8 6
   
1 1 −3 3 1 1 −3 3
R3 →R3 −R2
 0 2 −1 3  R3 ↔R4  0 2 −1 3 
R4 →R4 −2 R2    .
 0 0 0 0  ∼  0 0 10 0 
∼
0 0 10 0 0 0 0 0
   
1
1 1 −3 3 1
1 1 0 3
R2 → 2 R2
 0 1 − 1 3  R2 →R2 + 2 R3  0 1 0 3  R1 →R1 −R2
(b) R3 → 10 R3 
1
 0 0
2 2  R →R +3 R
1 1 3
 2 
1 0   0 0 1 0  ∼
∼ ∼
0 0 0 0 0 0 0 0
3 
1 0 0 2

 0 1 0 3 
 0 0 1 0 .
 2 

0 0 0 0
   
0 0 1 2 −1 −4 2 4 −1 3 2 −1
R1 ↔R3 R2 ↔R3
(6) (a)  0 0 0 1 −1 3   0 0 0 1 −1 3 
∼ ∼
 2 4 −1 3 2 −1  0 0 1 2 −1 −4
2 4 −1 3 2 −1
 0 0 1 2 −1 −4 .
0 0 0 1 −1 3
1
1 2 − 2 23 1 − 12 1 2 − 12 0 5
   
1
R2 →R2 −2 R3
2
−5
R1 → 2 R1
(b)  0 0 1 2 −1 −4  R1 →R1 − 32 R3  0 0 1 0 1 −2 
∼
0 0 0 1 −1 3  ∼ 0 0 0 1 −1 3
1 2 0 0 3 −6
R1 →R1 + 21 R2
 0 0 1 0 1 −2 .
∼
0 0 0 1 −1 3
 
1 0 2 0 1
x1 + 2 x3 = 1
(9)  0 1 1 3 −2  gives equations . Since
x2 + x3 + 3 x4 = −2
0 0 0 0 0
the third and fourth columns are pivot free, we put x4 = t and  x3 = s. Then 
1 − 2s
 −2 − s − 3t 
x2 = −2 − s − 3t and x1 = 1 − 2s giving the general solution x =  
 s 
t
with s, t ∈ R. Putting x3 = 3 and x4 = −2, we have the particular solution,
−5
 1 
x=  3 .


−2
MATA23H Solutions # 3 page 6

 
1 0
0 0 0
x1 = 0
 0 1
0 0 2 
  x2 = 2
(11)  0 0
1 0 −5   gives equations . The unique solution is
 x 3 = −5
 0 0
0 1 2 
x4 = 2
0 0
0 0 0
 
0
 2 
x=  −5 .


2
 
1 −1 2 0 3 1
 0 0 0 1 4 2 
(12) 
 0
. Since the third row has all zeros on the left of the
0 0 0 0 −1 
0 0 0 0 0 0
partition and −1 6= 0 on the right of the partition, the system is inconsistent and
there is no solution.
4 x1 − 3 x2 = 10
(14) Writing the given equations, , as an augmented ma-
  8 x1 −  x2 = 10 
4 −3 10 R2 →R2 −2 R1 4 −3 10
trix we have . The new equations
8 −1 10 ∼ 0 5 −10
4 x1 − 3 x2 = 10
are . The second gives x2 = −2 so the first becomes
5 x2 = −10  
1
4 x1 −3(−2) = 10 =⇒ 4 x1 = 4 =⇒ x1 = 1. The unique solution is x = .
−2
2x + y − 3z = 0
(16) The linear system, 6x + 3y − 8z = 0 , has augmented matrix
  2 x − y
 + 5 z = −4   
2 1 −3 0 R2 →R2 −3 R1 2 1 −3 0 2 1 −3 0
R2 ↔R3
 6 3 −8 0  R3 →R3 −R1  0 0 1 0   0 −2 8 −4 .
∼
2 −1 5 −4 ∼ 0 −2 8 −4 0 0 1 0
2x + y − 3z = 0
We now have equations − 2 y + 8 z = −4 . The third gives z = 0.
z = 0
Then the second becomes −2y + 8(0) = −4 =⇒ y = 2 and  the first becomes
−1
2x + 2 − 3(0) = 0 =⇒ x = −1. The unique solution is x =  2 .
0
x1 − 3 x2 + x3 = 2
(18) The linear system, , has augmented matrix
  3 x 1 − 8 x2 + 2 x3 = 5
 
1 −3 1 2 R2 →R2 −3 R1 1 −3 1 2
. We now have equations
3 −8 2 5 ∼ 0 1 −1 −1
x1 − 3 x2 + x3 = 2
. Since the third column is pivot free, we put
x2 − x3 = −1
x3 = t. Then x2 = −1 + t and the first equation becomes x1 − 3(−1 + t) = 2 =⇒
MATA23H Solutions # 3 page 7

 
−1 + 2 t
x1 = −1 + 2 t. The general solution is x =  −1 + t  , t ∈ R.
t
x1 + 4 x2 − 2 x3 = 4
(19) The linear system, 2 x1 + 7 x2 − x3 = −2 , has augmented matrix
  2 x1 + 9x2 − 7 x3 = 1

1 4 −2 4 R2 →R2 −2 R1 1 4 −2 4
R3 →R3 +R2
 2 7 −1 −2  R3 →R3 −2 R1  0 −1 3 −10 
∼
 2 9 −7 1  ∼ 0 1 −3 −7
1 4 −2 4
 0 −1 3 −10  which indicates an inconsistent system. There is no so-
0 0 0 −17
lution.  
2 x1 + 8 x2 = 16 2 8 16
(22) The linear system, , has augmented matrix
 5 x1 − 4 x2 =  −8   5 −4 −8
R1 → 12 R1 1 4 8 R2 →R2 −5 R1 1 4 8 R2 → −1
24
R 2 1 4 8
∼ 5 −4 −8  ∼ 0 −24 −48   ∼ 0 1 2
R1 →R1 −4 R2 1 0 0 0
. The unique solution is x = .
∼ 0 1 2 2
x1 − 2 x3 + x4 = 6
(23) The linear system, 2 x1 − x2 + x3 − 3 x4 = 0 , has augmented
 9 x1 − 3 x2 − x3  7 x4 = 4 
1 0 −2 1 6 R2 →R2 −2 R1 1 0 −2 1 6
R2 →−R2
matrix  2 −1 1 −3 0  R3 →R3 −9 R1  0 −1 5 −5 −12 
∼
 9 −3 −1 −7 4 ∼  0 −3 17 −16 −50
1 0 −2 1 6 1 0 −2 1 6
R3 →R3 +3 R2 R3 → 21 R3
 0 1 −5 5 12   0 1 −5 5 12 
∼ ∼
0 −3 17 −16 −50 0 0 2 −1 −14
1 0 0 0 −8
   
1 0 −2 1 6 R2 →R2 +5 R3
5
 0 1 −5 5 12  R1 →R1 +2 R3  0 1 0 2
−23 . Since the fourth
1
0 0 1 − 2 −7 ∼ 1
0 0 1 − 2 −7
column is pivot free, we put x 4 = t giving the general solution
−8
 
 −23 − 52 t 
x=  −7 + 1 t  , t ∈ R.

2
t
(26) b is in the span of v 1 , v 2 , v 3 , if there are scalars  x1, x2 , x3 such that v =
  x 1
x1 v 1 + x2 v 2 + x3 v 3 ; i.e., if the system v 1 v 2 v 3  x2  = b has a solution. We
 x3 
1 −4 1 8 R2 →R2 −3 R1
write this system as an augmented matrix:  3 −12 5 26  R3 →R3 −2 R1
2 −9 −1 14 ∼
MATA23H Solutions # 3 page 8

   
1 −4 1 8 1 −4 1 8
R2 ↔R3
 0 0 2 2   0 −1 −3 −2 . Since this system is consis-
∼
0 −1 −3 −2 0 0 2 2
tent, b is in the span of v 1 , v 2 , v 3 .
(29) (a) False; (b) False; (c) True; (d) True; (e) False; (f) True; (g) True; (h) True;
(i) False; (j) True.