Professional Documents
Culture Documents
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
3. Exercise 8.5.
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
4. Exercise 8.6.
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
5. Exercise 8.8.
(a) Show that D, U and L are each closed under addition and multiplication.
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.
Solution.
• Let M be the set of all matrices. By definition
D ∩ U = {A ∈ M | A ∈ D and A ∈ 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 }.
(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
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 ].
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.
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
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
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.
7
16. Exercise 9.11. Answers in S&B.
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.