Matrix Algebra Exercise Answers

All the exercises are from Simon & Blume.

1. Exercise 8.1. Some answers in S&B. Should be straightforward.

2. Show that if AB is defined, then B T AT is defined but AT B T need not be defined.

Solution. If AB is defined for two matrices A and B, it must be that the number
of columns in A is equal to the number of rows in B. So let A be an n × k matrix
and let B be a k × m matrix. Then AT is a k × n matrix and B T is a m × k matrix.
Since B T has the same number of columns as AT has rows (i.e. k) it follows that the
matrix product B T AT is defined. On the other hand, AT B T need not be defined,
because we might not have m = n. As a counterexample take
 
1 4  
7 8
A= 2 5 ,B = .
9 0
3 6

Then
 
43 8
AB =  59 16  .
75 24

Now
   
T 1 2 3 T 7 9
A = ,B = .
4 5 6 8 0

Hence, since AT is 2 × 3 and B T is 2 × 2, AT B T is undefined. 

3. Exercise 8.5.

(a) Answer in S&B.

(b) Show that if B is a scalar multiple of the 2 × 2 identity matrix, then AB = BA
for all 2 × 2 matrices A.

Solution. Let B be a scalar multiple of the 2 × 2 identity matrix. Then B is of

the form
   
1 0 λ 0
B=λ = ,
0 1 0 λ

for some λ ∈ R. Let A be any 2 × 2 matrix. Then

 
a b
A= ,
c d

1
 for some a, b, c, d ∈ R. We want to show that AB = BA (∗). First we compute
the LHS of (*).
    
a b λ 0 λa λb
AB = = .
c d 0 λ λc λd

Now we compute the RHS of (∗).

    
λ 0 a b λa λb
BA = = .
0 λ c d λc λd

So AB = BA. (Note that AB = Aλ = λA = BA. Multiplication by a diagonal

matrix with the same entries on the diagonal gives the same result as scalar
multiplication.) 

4. Exercise 8.6.

Square Matrix Diagonal Matrix

     

8 3 −6
 8 0 0 0 0 0 1 0 0
 0 1 0 , 0 0 0 , 0 1 0 .
 4 1 −5  .
0 −2 7 0 0 7 0 0 0 0 0 1

Upper-Triangular Matrix
Column Matrix
     
  8 3 −6 0 0 0 1 0 0
8  0 1 −5  ,  0 0 0  ,  0 1 0  .
 4 .
0 0 7 0 0 0 0 0 1
0
Lower-Triangular Matrix
Row Matrix      
8 0 0 0 0 0 1 0 0

 4 1 0 , 0 0 0 , 0 1 0 .
8 3 −6 . 0 −2 7 0 0 0 0 0 1

Symmetric Matrix
     
8 3 −6 0 0 0 1 0 0
 3 1 −5  ,  0 0 0  ,  0 1 0  .
−6 −5 7 0 0 0 0 0 1

Idempotent Matrix
For an idempotent matrix, we need a square matrix B for which BB = B.
If B 6= I, then B must be singular. Suppose B were not singular. Then B has
an inverse B −1 . If B is idempotent we have BB = B. Multiplying both sides
on the left by the inverse we get B = B −1 B = I, contradicting our assumption
that B 6= I. This shows that an idempotent matrix B 6= I must have a less
than maximal rank and that |B| = 0.

2
 In fact this means that the rows are linearly dependent, so that they are linear
combinations of each other. We can use this fact to help construct examples of
idempotent matrices.
     
5 −5 5 0 0 0 6 2 12
 4 −4 4  ,  0 0 0  ,  3 1 6 .
0 0 0 0 0 0 −3 −1 −6

Permutation Matrix Nonsingular Matrix

   
1 0 0 1 0 0    
8 3 −6 1 0 0
 0 0 1 , 0 1 0 .  4 1 −5  ,  0 1 0  .
0 1 0 0 0 1 0 −2 7 0 0 1

5. Exercise 8.8.

(a) Show that D, U and L are each closed under addition and multiplication.

Solution. Consider any two diagonal matrices, A, B ∈ D. By definition A and

B are square (say n×n) and aij = bij = 0 for all i 6= j where i, j ∈ {1, 2, . . . , n}.
Now, by the definition of matrix addition, entry (i, j) of the matrix A + B is
aij + bij . When i 6= j, the (i, j)th entry of A + B is simply 0 + 0 = 0, and so
A + B ∈ D. Hence D is closed under matrix addition. To show D is closed
under matrix multiplication, consider the (i, j)th entry of AB given by
 
b1j
 b2j 
(ai1 ai2 · · · ain )  ..  = ai1 b1j + ai2 b2j + · · · + ain bnj
 
 . 
bnj
n
X
= aih bhj .
h=1

When i 6= j, it must be the case that one of aih or bhj is 0, and so aih bhj = 0
for all h ∈ {1, . . . , m}. Therefore the (i, j)th entry of AB is 0 for all i 6= j. So
we have shown that AB ∈ D and D is closed under matrix multiplication. The
argument for sets U and L is similar. 

(b) Show that D ∩ U = D, S ∩ U = D, and D ⊂ S.

Solution.
• Let M be the set of all matrices. By definition

D ∩ U = {A ∈ M | A ∈ D and A ∈ U }.

We know that D ∩ U ⊆ D from a general result we saw in set theory. (We

could prove this in this special case in the following way. Let A ∈ D ∩ U .

3
 Then matrix A is a diagonal matrix (and upper-triangular but that is
irrelevant) so A ∈ D and it follows that D ∩ U ⊆ D.) We are left to show
that D ⊆ D ∩ U . Let A ∈ D. Then aij = 0 for i 6= j and therefore aij = 0
for i > j. Hence A ∈ U and we have shown that A ∈ D ∩ U as required.
Putting the two results together we find that D ∩ U = D.
• By definition

S ∩ U = {A ∈ M | A ∈ S and A ∈ U }.

We first show that S ∩ U ⊆ D. Let A ∈ S ∩ U . Then matrix A is a

symmetric matrix and aij = aji for all i, j. Also, since A is an upper-
triangular matrix, aij = 0 for i > j. Putting these two properties together
we see that aji = 0 for i > j and it follows that aij = 0 for all i 6= j. Thus
A ∈ D and S ∩ U ⊆ D. Now we show that D ⊆ S ∩ U . Let A ∈ D. Then
aij = 0 for all i 6= j. It follows (from simply swapping the role of i and j in
our labelling of indices) that aji = 0 for all i 6= j. But then aij = aji for all
i 6= j. Since in any matrix it is always the case that aij = aji for i = j, we
have shown that A is symmetric. We showed above that a diagonal matrix
is upper-triangular i.e. A ∈ U . It follows that A ∈ S ∩ U and D ⊆ S ∩ U .
Hence S ∩ U = D.
• We showed above that a diagonal matrix is necessarily symmetric. That
is, we showed that A ∈ D implies A ∈ S, so by definition D ⊆ S. To show
that D ⊂ S we must also show that S 6⊆ D. Consider the counterexample
 
1 2
,
2 1

which is symmetric but not diagonal. 

(c) Show that all matrices in D commute with each other. Is this true for matrices
in U or S too?

Solution.
• Let A, B ∈ D both be n × n. Matrix addition is commutative for any two
matrices for which it is defined. Since D ⊂ M it follows that A and B
commute under addition. Now consider matrix multiplication. By defini-
tion, the (i, j)th entry of AB is nh=1 aih bhj , while the (i, j)th entry of BA
P

is nh=1 bih ahj . When i 6= j the (i, j)th entry is 0 in both AB and BA by
P

an argument we used in (a). When i = j, the entry (i, j) or equivalently

(i, i) (on the diagonal) in AB is given by nh=1 aih bhi . The (i, i)th entry
P

of BA is given by nh=1 aih bhi and so the (i, i)th entries of AB and BA
P

are the same for all i. Since all entries of both matrices are equal, we have

4
 shown that AB = BA and matrix multiplication of diagonal matrices is
commutative.
• Matrix multiplication of matrices in S is not commutative. For example
    
−1 1 1 1 0 0
= ,
1 0 1 1 1 1

but     
1 1 −1 1 0 1
= .
1 1 1 0 0 1
• Matrix multiplication of matrices in U is not commutative. For example
    
3 1 1 2 3 10
= ,
0 2 0 4 0 8

but     
1 2 3 1 3 5
= .
0 4 0 2 0 8


(d) Show that S is closed under addition but not under multiplication.

Solution. The proof that S is closed under addition is very similar to the other
proofs in (a). To show that S is not closed under multiplication, consider the
following counterexample (or the counterexample we used to show symmetric
matrices were not commutative).
    
1 2 3 2 7 4
= ∈
/ S.
2 4 2 1 14 8

6. Exercise 8.17. Follow example 8.4, but first interchange the rows, then add −a/c
times the first row to the second row to get the augmented matrix [A|I] in row
echelon form. Then reduce it further to get the matrix [I|A−1 ].

7. Exercise 8.19. Answers in S&B.

8. Exercise 8.20. Answers in S&B.

9. Exercise 8.22. Answer in S&B.

10. Exercise 8.24. Consider a general 2 × 2 upper-triangular matrix U and a general

lower-triangular matrix L:
   
a b a 0
U= ,L = .
0 d c d

5
 (a) Suppose U is invertible. By theorem 8.8 it follows that ad − bc = ad − b(0) =
ad 6= 0, which implies that a 6= 0 and d 6= 0 as required. Now, suppose that
each diagonal entry is nonzero i.e. a 6= 0 and d 6= 0. Then, for our matrix U ,
we have ad − bc = ad − b(0) = ad 6= 0. Hence by theorem 8.8 U is invertible.
A similar argument shows that the result is true for L.
(b) By example 8.4, the inverse of L is
 
−1 1 d 0
L = ,
ad −c a

which is lower-triangular.
(c) By example 8.4, the inverse of U is
 
−1 1 d −b
U = ,
ad 0 a

which is upper-triangular.

11. Exercise 8.28. Answer in S&B.

12. Exercise 9.4.

Solution. Consider a general 2 × 2 matrix A given by

 
a11 a12
.
a21 a22

for some a11 , a12 , a21 , a22 ∈ R.

Along the First Row

|A| = a11 (−1)1+1 det (a22 ) + a12 (−1)1+2 det (a21 )

= a11 a22 − a12 a21 .

Along the Second Row

|A| = a21 (−1)2+1 det (a12 ) + a22 (−1)2+2 det (a11 )

= −a12 a21 + a22 a11
= a11 a22 − a12 a21 .

This shows that we ge the same result no matter which row we use. Similar
working shows we can also expand along any column to find the determinant.

13. Exercise 9.7. You need to calculate the determinant of the original matrix and of the
row echelon form and compare the two numbers. Let A denote the original matrix.

6
 (a) |A| = 1(1) − 1(2) = −1. A row echelon form is obtained by subtracting 2 time
row 1 from row 2 to get
 
1 1
R= .
0 −1

We have |R| = 1(−1) − 1(0) = −1 and |A| = |R|.

(b) Expand along column 3 since it has two zeros.
 
 2 4 
|A| = −3  
−6 −10 
= −3[2(−10) − 4(−6)]
= −12.

To obtain a row echelon form, first add −2 times row 1 to row 2, and adding
3 times row 1 to row 2. Then add row 2 to row 3 to get
 
2 4 0
R =  0 −2 3  .
0 0 3

By theorem 9.1 we have |R| = 2(−2)(3) = −12 and so |A| = |R|.

(c) Expand along column 1 since it has two zeros.
 
 1 2 
|A| = −3  
7 8 
= −3[1(8) − 2(7)]
= 18

To obtain a row echelon form, first interchange rows 1 and 2. Then add −7
times row 2 to row 3 to get
 
3 4 5
R= 0 1 2 .
0 0 −6

By theorem 9.1 we have |R| = 3(1)(−6) = −18. We have |A| = −|R| because
one row interchange was used. If we had used two row interchanges, we would
have obtained a determinant equal to |A|.

14. Exercise 9.8. Reduce the matrix A to a row echelon form R, taking note of the
number of row interchanges. Then find the determinant of R by multiplying the
diagonal entries. Finally, calculate |A| = (−1)i |R| where i is the number of row
interchanges.

15. Exercise 9.9. Answer in S&B.

7
16. Exercise 9.11. Answers in S&B.

17. Exercise 9.13. Answers in S&B.

18. Exercise 9.17.

Solution. Write the system in matrix form with the endogenous variables Y and r
in a column matrix on the left and the exogenous variables or parameters on the
right and in the coefficient matrix:
    
(1 − t)s a Y I0 + G
= .
m −h r MS − M 0
Now use Cramer’s rule to solve for Y and r.

 
 I0 + G a 

 MS − M 0 −h 
Y =  
 (1 − t)s a 
 m −h 
(I 0 + G)(−h) − a(MS − M 0 )
=
(1 − t)s(−h) − am
(I + G)h + a(MS − M 0 )
0
= ,
(1 − t)sh + am
 
 (1 − t)s I 0 + G 
 
 m MS − M 0 
r= 
 (1 − t)s

 a 

 m −h 
(1 − t)s(MS − M 0 ) − (I 0 + G)m
=
(1 − t)s(−h) − am
(I + G)m − (1 − t)s(MS − M 0 )
0
= .
(1 − t)sh + am

To find the effect of the tax rate on Y and r take the partial derivative (the derivative
holding all other exogenous variables constant) with respect to t.
∂Y sh[(I 0 + G)h + a(MS − M 0 )]
=− .
∂t [(1 − t)sh + am]2
We have to use the quotient rule to find the effect of a change in t on r:
∂r s(M s − M 0 )[(1 − t)sh + am] − [(I 0 + G)m − (1 − t)s(M S − M 0 )](−sh)
=
∂t [(1 − t)sh + am]2
sm[(I + G)h + a(MS − M 0 )]
0
=
[(1 − t)sh + am]2
If MS > M 0 then net national product falls with the tax rate and the interest rate
rises. 

8
19. Use the fact that |AB| = |A||B| to show that if A is invertible then
1
|A−1 | = .
|A|

Solution. Let B = A−1 . Then by the fact we have |AA−1 | = |A||A−1 |. Now |AA−1 | =
|I| = 1 since the determinant of a diagonal matrix is just the product of the diagonal
entries which are all equal to one. So we have 1 = |A||A−1 |, and rearranging gives
the result.