MAS201 PROBLEM SHEET 2

Exercise 1. Consider the vectors

v
1
=
_
1 2 1 0

T
v
2
=
_
3 1 4 2

T
v
3
=
_
1 5 6 2

T
v =
_
5 4 9 4

T
w =
_
4 2 3 1

T
and the matrix
A =
_
_
v
1
v
2
v
3
v w
_
_
.
(a) Row-reduce A.
(b) Is v a linear combination of v
1
, v
2
and v
3
?
(c) Is w a linear combination of v
1
, v
2
and v
3
?
Solution: We have
A =
_
_
v
1
v
2
v
3
v w
_
_
=
_

_
1 3 1 5 4
2 1 5 4 2
1 4 6 9 3
0 2 2 4 1
_

_
1 3 1 5 4
0 7 7 14 10
0 7 7 14 7
0 2 2 4 1
_

_

_

_
1 3 1 5 4
0 1 1 2 1
0 7 7 14 10
0 2 2 4 1
_

_
1 3 1 5 4
0 1 1 2 1
0 0 0 0 3
0 0 0 0 3
_

_

_

_
1 0 2 1 1
0 1 1 2 1
0 0 0 0 1
0 0 0 0 0
_

_
1 0 2 1 0
0 1 1 2 0
0 0 0 0 1
0 0 0 0 0
_

_
=:
_
_
D
1
D
2
D
3
E F
_
_
,
in reduced row echelon form.
(a) Delete all the 5th columns from the above matrices: the same sequence of EROs shows that
_
_
v
1
v
2
v
3
v
_
_

_
_
D
1
D
2
D
3
E
_
_
=
_

_
1 0 2 1
0 1 1 2
0 0 0 0
0 0 0 0
_

_
.
Since E = D
1
+ 2D
2
, we see that D
1
2D
2
+ E = 0 and
_

_
1 0 2 1
0 1 1 2
0 0 0 0
0 0 0 0
_

_
_

_
1
2
0
1
_

_
= 0, so that
_
_
v
1
v
2
v
3
v
_
_
_

_
1
2
0
1
_

_
= 0.
Therefore v
1
2v
2
+ v = 0 and v = v
1
+ 2v
2
= v
1
+ 2v
2
+ 0v
3
is a linear combination of v
1
, v
2
, v
3
.
(b) Now delete all the 4th columns from the matrices at the beginning of this solution: the same sequence
of EROs shows that
_
A

|w
_
=
_
_
v
1
v
2
v
3
w
_
_

_
_
D
1
D
2
D
3
F
_
_
=
_

_
1 0 2 0
0 1 1 0
0 0 0 1
0 0 0 0
_

_
.
1
Thus the system of linear equations A

1

2

3

T
= w has no solution, so that w is not a linear
combination of v
1
, v
2
, v
3
.
Exercise 2. Consider the vectors
v
1
=
_
1 2 3 4

T
and v
2
=
_
3 2 1 0

T
,
and let {e
1
, e
2
, e
3
, e
4
} be the standard basis.
(a) Row-reduce the matrix
A =
_
_
v
1
v
2
e
1
e
2
e
3
e
4
_
_
.
(b) Find a basis B for R
4
that contains v
1
and v
2
.
(c) Express each of e
1
, e
2
, e
3
and e
4
as a linear combination of the vectors in B.
Solution: We have
A =
_
_
v
1
v
2
e
1
e
2
e
3
e
4
_
_
=
_

_
1 3 1 0 0 0
2 2 0 1 0 0
3 1 0 0 1 0
4 0 0 0 0 1
_

_
1 3 1 0 0 0
0 4 2 1 0 0
0 8 3 0 1 0
0 12 4 0 0 1
_

_

_

_
1 3 1 0 0 0
0 4 2 1 0 0
0 0 1 2 1 0
0 0 2 3 0 1
_

_
1 0
1
2
3
4
0 0
0 1
1
2

1
4
0 0
0 0 1 2 1 0
0 0 0 1 2 1
_

_

_

_
1 0 0
1
4
1
2
0
0 1 0
3
4

1
2
0
0 0 1 2 1 0
0 0 0 1 2 1
_

_
1 0 0 0 0
1
4
0 1 0 0 1
3
4
0 0 1 0 3 2
0 0 0 1 2 1
_

_
=: E,
in reduced row echelon form. Write
E =
_
_
G
1
G
2
G
3
G
4
G
5
G
6
_
_
.
Delete all the 5th and 6th columns from the above matrices, and add a 4 1 column of 0s to each
one, as a new 5th column: the same sequence of EROs shows that
_
_
v
1
v
2
e
1
e
2
0
_
_

_

_
1 0 0 0 0
0 1 0 0 0
0 0 1 0 0
0 0 0 1 0
_

_
=
_
I
4
|0
_
.
Therefore the system of linear equations
_
_
v
1
v
2
e
1
e
2
_
_
_

1

2

3

4

T
= 0
has
1
=
2
=
3
=
4
= 0 as its only solution. Therefore the vectors v
1
, v
2
, e
1
, e
2
are linearly
independent. We then note that 4 linearly independent vectors in R
4
automatically span R
4
, and so
form a basis for R
4
.
Of course, e
1
= 0v
1
+ 0v
2
+ 1e
1
+ 0e
2
and e
2
= 0v
1
+ 0v
2
+ 0e
1
+ 1e
2
.
2
Note that

A =
_
_
v
T
1
v
T
2
e
T
1
e
T
2
e
T
3
_
_

_
_
G
1
G
2
G
3
G
4
G
5
_
_
=
_

_
1 0 0 0 0
0 1 0 0 1
0 0 1 0 3
0 0 0 1 2
_

_
.
Since G
5
= 0G
1
+ 1G
2
3G
3
2G
4
, so that
_
_
G
1
G
2
G
3
G
4
G
5
_
_
_

_
0
1
3
2
1
_

_
= 0
it follows that

A
_
0 1 3 2 1

T
= 0, so that e
3
= 0v
1
+ 1v
2
3e
1
2e
2
.
Similarly, since
_
_
v
1
v
2
e
1
e
2
e
4
_
_

_
_
G
1
G
2
G
3
G
4
G
6
_
_
=
_

_
1 0 0 0
1
4
0 1 0 0
3
4
0 0 1 0 2
0 0 0 1 1
_

_
and G
6
=
1
4
G
1

3
4
G
2
+ 2G
3
+ 1G
4
, we obtain e
4
=
1
4
v
1

3
4
v
2
+ 2e
1
+ 1e
2
.
Exercise 3. Put a =
_
1 3 5 7

R
4
.
(a) Suppose we have vectors u
1
, . . . , u
4
R
4
with a.u
1
= a.u
2
= a.u
3
= a.u
4
= 0. Prove that the
list u
1
, . . . , u
4
does not span R
4
.
(b) Give an example of a list v
1
, . . . , v
4
that satises a.v
1
= a.v
2
= a.v
3
= a.v
4
= 1 and also spans
R
4
.
(c) Give an example of a list w
1
, . . . , w
4
that satises a.w
1
= a.w
2
= a.w
3
= a.w
4
= 1 and does not
span R
4
.
Solution:
(a) If x is a linear combination of the vectors u
i
, we have x =
1
u
1
+ +
4
u
4
for some scalars

1
, . . . ,
4
, so
a.x = a.(
1
u
1
+ +
4
u
4
) =
1
(a.u
1
) + +
4
(a.u
4
),
but a.u
1
= a.u
2
= a.u
3
= a.u
4
= 0 so a.x = 0. On the other hand, we have a.e
1
= 1 = 0, so e
1
cannot be a linear combination of the vectors u
i
. This means that the u
i
do not span R
4
.
(b) The obvious example is
v
1
=
_

_
1
0
0
0
_

_
v
2
=
_

_
0
1/3
0
0
_

_
v
3
=
_

_
0
0
1/5
0
_

_
v
4
=
_

_
0
0
0
1/7
_

_
.
To see that this spans, note that an arbitrary vector x =
_
a b c d

T
in R
4
can be expressed
as
x = av
1
+ 3bv
2
+ 5cv
3
+ 7dv
4
,
which is a linear combination of the list v
1
, . . . , v
4
.
3
(c) The most obvious solution is to take w
1
= w
2
= w
3
= w
4
= e
1
. If we prefer to avoid repetitions,
we can instead use
w
1
=
_

_
1
0
0
0
_

_
w
2
=
_

_
4
1
0
0
_

_
w
3
=
_

_
7
2
0
0
_

_
w
4
=
_

_
10
3
0
0
_

_
.
It is clear that any linear combination of w
1
, . . . , w
4
has zeros in the third and fourth places. In
particular, the standard vector e
4
is not a linear combination of the list w
1
, . . . , w
4
, so the list
does not span R
4
.
Exercise 4. The vectors
u
1
=
_

_
1
0
0
0
_

_
u
2
=
_

_
1
1
0
0
_

_
u
3
=
_

_
1
1
1
0
_

_
u
4
=
_

_
1
1
1
1
_

_
u
5
=
_

_
0
1
1
1
_

_
u
6
=
_

_
0
0
1
1
_

_
u
7
=
_

_
0
0
0
1
_

_
span R
4
, because an arbitrary vector x =
_
a b c d

T
can be expressed as a linear combination of u
i
by the formula
x = (a b)u
1
+ bu
2
+ cu
6
+ (d c)u
7
,
or alternatively by the formula
x = bu
1
+ bu
2
du
3
+ (a + d)u
4
au
5
+ cu
6
cu
7
.
(a) Check the above formulae.
(b) Give a similar explicit formula to prove that the following vectors span R
4
:
v
1
=
_

_
1
1
1
0
_

_
v
2
=
_

_
1
1
0
1
_

_
v
3
=
_

_
1
1
1
1
_

_
v
4
=
_

_
1
0
1
1
_

_
v
5
=
_

_
0
1
1
1
_

_
(c) Use the row-reduction method to show again that the vectors v
i
span R
4
.
Solution:
(a) For the rst formula we have
(a b)u
1
+ bu
2
+ cu
6
+ (d c)u
7
=
_

_
a b
0
0
0
_

_
+
_

_
b
b
0
0
_

_
+
_

_
0
0
c
c
_

_
+
_

_
0
0
0
d c
_

_
=
_

_
a
b
c
d
_

_
.
For the second, we have
bu
1
+ bu
2
du
3
+ (a + d)u
4
au
5
+ cu
6
cu
7
=
_

_
b
0
0
0
_

_
+
_

_
b
b
0
0
_

_
+
_

_
d
d
d
0
_

_
+
_

_
a + d
a + d
a + d
a + d
_

_
+
_

_
0
a
a
a
_

_
+
_

_
0
0
c
c
_

_
+
_

_
0
0
0
c
_

_
=
_

_
a
b
c
d
_

_
.
(b) One possible formula is as follows: if x =
_
a b c d

T
, then
x = dv
1
cv
2
+ (a + b + c + d)v
3
bv
4
av
5
.
This can be found as follows: we note that
e
1
= v
3
v
5
e
2
= v
3
v
4
e
3
= v
3
v
2
e
4
= v
3
v
1
,
and it follows that
x = ae
1
+ be
2
+ ce
3
+ de
4
= a(v
3
v
5
) + b(v
3
v
4
) + c(v
3
v
2
) + d(v
3
v
1
)
= dv
1
cv
2
+ (a + b + c + d)v
3
bv
4
av
5
.
4
(c) The general method for these kinds of questions is to construct a matrix A whose rows are the
vectors v
T
i
, and then row-reduce it:
A =
_

_
1 1 1 0
1 1 0 1
1 1 1 1
1 0 1 1
0 1 1 1
_

_
1 1 1 0
0 0 1 1
0 0 0 1
0 1 0 1
0 1 1 1
_

_
1 1 1 0
0 0 1 1
0 0 0 1
0 0 1 2
0 1 1 1
_

_
1 1 1 0
0 0 1 1
0 0 0 1
0 0 0 3
0 1 1 1
_

_
1 1 1 0
0 1 1 1
0 0 1 1
0 0 0 1
0 0 0 3
_

_
1 0 0 0
0 1 0 0
0 0 1 0
0 0 0 1
0 0 0 0
_

_
The nal matrix has a pivot in every column, so the vectors v
i
span R
4
.
Exercise 5. Over a period of 5 minutes, in a typical MAS201 lecture, 90% of students who are awake at
the beginning of the 5-minute period will still be so at the end of it (but the other 10% will fall asleep)
and 90% of students who are asleep at the beginning of the 5-minute period will still be so at the end of
it (and the other 10% will wake up). If all the students are awake at the beginning of the lecture, what
percentage will be awake at the end of the lecture, 50 minutes later?
Solution: For each k = 0, . . . , 10, let a
k
and b
k
be the proportions of students who are awake and who
are asleep after 5k minutes of the lecture, respectively, and set v
k
= ( a
k
b
k
)
T
. We have
_
a
0
b
0
_
=
_
1
0
_
and
_
a
k+1
b
k+1
_
=
_
9
10
1
10
1
10
9
10
_
_
a
k
b
k
_
for k = 0, . . . , 9. Set
A =
_
9
10
1
10
1
10
9
10
_
.
We are thus considering the dierence equation v
k+1
= Av
k
, so that v
k
= A
k
v
0
for k = 0, . . . , 10, and
we wish to nd v
10
= A
10
v
0
.
The matrix A is stochastic, and so has 1 as an eigenvalue. The characteristic polynomial of A is

A
(t) =

9
10
t
1
10
1
10
9
10
t

=
_
t
9
10
_
2

_
1
10
_
2
= (t 1)
_
t
8
10
_
.
Thus the eigenvalues of A are 1 and
8
10
. Since A has 2 distinct eigenvalues, we can nd a basis for R
2
consisting of eigenvectors of A.
To nd an eigenvector of A corresponding to the eigenvalue
1
:= 1, we need to solve the system of
linear equations (AI
2
)( x y )
T
= 0. This has augmented matrix
_
AI
2
|0
_
=
_

1
10
1
10
0
1
10

1
10
0
_

_

1
10
1
10
0
0 0 0
_
,
and so w
1
:=
_
1 1

T
is an eigenvector of A corresponding to the eigenvalue 1.
To nd an eigenvector of A corresponding to the eigenvalue
2
:=
8
10
, we need to solve the system of
linear equations
_
A
8
10
I
2
_
x y

T
= 0. This has augmented matrix
_
A
8
10
I
2
|0
_
=
_
1
10
1
10
0
1
10
1
10
0
_

_
1
10
1
10
0
0 0 0
_
,
and so w
2
:=
_
1 1

T
is an eigenvector of A corresponding to the eigenvalue
8
10
.
Now, w
1
=
_
1
1
_
and w
2
=
_
1
1
_
, being eigenvectors corresponding to distinct eigenvalues of A,
form a basis for R
2
. We express v
0
as a linear combination of these two eigenvectors:
v
0
=
_
1
0
_
=
1
2
_
1
1
_
+
1
2
_
1
1
_
=
1
2
w
1
+
1
2
w
2
.
5
We have
v
k
= A
k
v
0
= A
k
_
1
2
w
1
+
1
2
w
2
_
=
1
2
A
k
w
1
+
1
2
A
k
w
2
=
1
2

k
1
w
1
+
1
2

k
2
w
2
=
1
2
1
k
w
1
+
1
2
(0.8)
k
w
2
.
In particular,
v
10
=
_
a
10
b
10
_
=
1
2
w
1
+
1
2
(0.8)
10
w
2
=
1
2
_
1
1
_
+
1
2
(0.8)
10
_
1
1
_
.
Since (0.8)
10
0.107374, we conclude that approximately 55.37% of students are awake at the end of
the lecture.
Exercise 6. Let A be the 5 5 matrix in which every entry is one.
(a) Show that A
2
= 5A.
(b) Suppose that is an eigenvalue of A, so there exists a nonzero vector u with Au = u. By
considering A
2
u, show that
2
= 5, so = 0 or = 5. (You should not write out any matrices
here, or attempt to calculate the characteristic polynomial; just use part (a).)
(c) Find an eigenvector v of eigenvalue 5, and a linearly independent list w
1
, . . . , w
4
of eigenvectors
of eigenvalue 0.
(d) Now put B =
1
2
I
5
+
1
10
A. Show that B is stochastic.
(e) Prove by induction on k that B
k
= 2
k
I
5
+ (1 2
k
)A/5 for all k 0. (You should not write
out any matrices here; just use part (a).) What happens when k is large?
Solution:
(a) One way to say this is to introduce the vector v =
_
1 1 1 1 1

T
, so v.v = 5. We also have
A =
_

_
1 1 1 1 1
1 1 1 1 1
1 1 1 1 1
1 1 1 1 1
1 1 1 1 1
_

_
=
_

_
v
T
v
T
v
T
v
T
v
T
_

_
=
_

_
v v v v v
_

_
so
A
2
=
_

_
v.v v.v v.v v.v v.v
v.v v.v v.v v.v v.v
v.v v.v v.v v.v v.v
v.v v.v v.v v.v v.v
v.v v.v v.v v.v v.v
_

_
=
_

_
5 5 5 5 5
5 5 5 5 5
5 5 5 5 5
5 5 5 5 5
5 5 5 5 5
_

_
= 5A.
(b) Suppose we have an eigenvalue , and an associated eigenvector u (so u = 0 and Au = u). We
then have
A
2
u = A(Au) = A(u) = Au =
2
u.
On the other hand, we have A
2
= 5A, so
A
2
u = 5Au = 5u.
By comparing these two equations, we see that
2
u = 5u, so (
2
5)u = 0 or ( 5)u = 0.
As u = 0 it follows that = 0 or = 5.
(c) Put
v =
_

_
1
1
1
1
1
_

_
w
1
=
_

_
1
1
0
0
0
_

_
w
2
=
_

_
0
1
1
0
0
_

_
w
3
=
_

_
0
0
1
1
0
_

_
w
4
=
_

_
0
0
0
1
1
_

_
.
It is easy to see that Av = 5v and Aw
i
= 0 for all i, so v is an eigenvector of eigenvalue 5, and
w
1
, . . . , w
4
are eigenvectors of eigenvalue 0. It is also clear that the list w
1
, . . . , w
4
is linearly
independent. This is not the only possible answer. For example, the list
w

1
=
_

_
1
0
0
0
1
_

_
w

2
=
_

_
0
1
0
0
1
_

_
w

3
=
_

_
0
0
1
0
1
_

_
w

4
=
_

_
0
0
0
1
1
_

_
would do equally well.
6
(d) In the metrix B, every entry away from the diagonal is
1
10
, and every entry on the diagonal is
1
10
+
1
2
=
6
10
. In particular, all entries are positive. Moreover, each column contains four entries
equal to
1
10
and one entry equal to
6
10
, adding up to (4 1 + 6)/10 = 1. Thus, the matrix is
stochastic.
(e) We claim that for all k 0 we have B
k
= 2
k
I
5
+(1 2
k
)A/5. When k = 0 the left hand side
is B
0
= I
5
, whereas the right hand side is 2
0
I
5
+ (1 2
0
)A = I
5
, as required. When k = 1 the
left hand side is B
1
= B =
1
2
I
5
+
1
10
A. We also have 2
1
= 1 2
1
=
1
2
so on the right hand
side we have
1
2
I
5
+
1
10
A again, as required.
Now suppose that the claim is true for a particular value of k. We can the equation B =
1
2
I
5
+
1
10
A by the equation B
k
= 2
k
I
5
+ (1 2
k
)A/5 and expand out to get
B
k+1
= (
1
2
I
5
+
1
10
A)(2
k
I
5
+ (1 2
k
)A/5)
=
1
2
2
k
I
5
+
1
2
1
5
(1 2
k
)A +
1
10
2
k
A +
1
10
1
5
(1 2
k
)A
2
.
Using A
2
= 5A this becomes
B
k+1
=
1
2
2
k
I
5
+
1
2
1
5
(1 2
k
)A +
1
10
2
k
A +
1
10
(1 2
k
)A
= 2
k1
I
5
+
_
1
2
(1 2
k
) +
1
2
2
k
+
1
2
(1 2
k
)
_
A/5
= 2
k1
I
5
+
_
1
2
2
k1
+ 2
k1
+
1
2
2
k1
)
_
A/5
= 2
k1
I
5
+ (1 2
k1
)A/5.
This is the case k + 1 of our claim. It follows by induction that the claim holds for all k.
Exercise 7. Consider the matrix A =
_
1 1
2 1
_
.
(a) Find the eigenvalues of A.
(b) For each eigenvalue, nd a corresponding eigenvector of A.
(c) Dene recursively a sequence of vectors
_
u
n
v
n
_
as follows: we have u
0
= 1 and v
0
= 0, and for all
n > 0 we have
u
n
= u
n1
+ v
n1
v
n
= 2u
n1
+ v
n1
.
Use your eigenvectors of A to nd expressions for u
n
and v
n
(for a general positive integer n).
Solution:
We have

A
(t) =

1 t 1
2 1 t

= (t 1)
2
2 = t
2
2t 1 =
_
t 1

2
_ _
t 1 +

2
_
.
We thus see that the eigenvalues of A are
1
= 1 +

2 and
2
= 1

2. There are two distinct

eigenvalues, and so it is possible to nd a basis for R
2
consisting of eigenvectors of A. Notice that
(1 +

2)(1

2) = 1 and (1 +

2) + (1

2) = 2.
To nd an eigenvector corresponding to
1
, we consider (A
1
I
2
)
_
x y

T
= 0:
_
A
1
I
2
|0
_
=
_
1 (1 +

2) 1 0
2 1 (1 +

2) 0
_
=
_

2 1 0
2

2 0
_

2 1 0
0 0 0
_
,
so that w
1
:=
_
1

2
_
is an eigenvector of A corresponding to
1
.
To nd an eigenvector corresponding to
2
, we consider (A
2
I
2
)
_
x y

T
= 0:
_
A
2
I
2
|0
_
=
_
1 (1

2) 1 0
2 1 (1

2) 0
_
=
_
2 1 0
2

2 0
_

_
2 1 0
0 0 0
_
,
7
so that w
2
:=
_
1

2
_
is an eigenvector of A corresponding to
2
.
We have
_
u
0
v
0
_
=
_
1
0
_
and
_
u
n
v
n
_
=
_
1 1
2 1
_ _
u
n1
v
n1
_
= A
_
u
n1
v
n1
_
for n > 0.
Therefore
_
u
n
v
n
_
= A
n
_
1
0
_
for n > 0.
We calculate this by using the above eigenvectors w
1
and w
2
of A. Since w
1
and w
2
are eigenvectors of
A corresponding to dierent eigenvalues, they are linearly independent, and so form a basis for R
2
. We
express
_
1 0

T
as a linear combination of w
1
and w
2
:
_
1
0
_
=
1
2
_
1

2
_
+
1
2
_
1

2
_
=
1
2
w
1
+
1
2
w
2
.
Therefore, for all n > 0,
_
u
n
v
n
_
= A
n
_
1
0
_
= A
n
_
1
2
w
1
+
1
2
w
2
_
=
1
2
A
n
w
1
+
1
2
A
n
w
2
=
1
2

n
1
w
1
+
1
2

n
2
w
2
=
1
2
_
1 +

2
_
n
_
1

2
_
+
1
2
_
1

2
_
n
_
1

2
_
=
_
_
1
2
_
_
1 +

2
_
n
+
_
1

2
_
n
_

2
2
_
_
1 +

2
_
n

_
1

2
_
n
_
_
_
.
Thus
u
n
=
1
2
_
(1 +

2)
n
+ (1

2)
n
_
and v
n
=
1

2
_
(1 +

2)
n
(1

2)
n
_
for all n > 0.
Exercise 8. The sequence (a
n
)

n=0
is given by a
0
= 1001001, a
1
= 1010100, a
2
= 1110000 and
a
n+3
= 111a
n+2
1110a
n+1
+ 1000a
n
( for n > 2)
(a) Write down a matrix equation relating the vector u
n
to u
n+1
, where u
n
=
_
_
a
n+2
a
n+1
a
n
_
_
.
(b) Find the eigenvalues and eigenvectors of the matrix occuring in (a). (If you have done this
correctly, the answers will be integers with a nice pattern.)
(c) Express u
0
as a linear combination of the eigenvectors in (b).
(d) Give a general formula for a
n
.
(e) Check directly that your formula satises a
n+3
= 111a
n+2
1110a
n+1
+1000a
n
and that a
0
, a
1
and a
2
are as they should be.
Solution:
(a) We have
u
n+1
=
_
_
a
n+3
a
n+2
a
n+1
_
_
=
_
_
111a
n+2
1110a
n+1
+ 1000a
n
a
n+2
a
n+1
_
_
=
_
_
111 1110 1000
1 0 0
0 1 0
_
_
_
_
a
n+2
a
n+1
a
n
_
_
.
In other words, if we put
A =
_
_
111 1110 1000
1 0 0
0 1 0
_
_
then u
n+1
= Au
n
. It follows that for all n 0 we have
u
n
= A
n
u
0
= A
n
_
_
1110000
1010100
1001001
_
_
.
8
(b) The characteristic polynomial is

A
(t) = det
_
_
111 t 1110 1000
1 t 0
0 1 t
_
_
= (111 t) det
_
t 0
1 t
_
+ 1110 det
_
1 0
0 t
_
+ 1000 det
_
1 t
0 1
_
= (111 t)t
2
1110t + 1000 = 1000 1110t + 111t
2
t
3
= (1 t)(10 t)(100 t).
Thus, the eigenvalues are 1, 10 and 100. To nd the corresponding eigenvectors, we perform the
following row-reductions:
AI =
_
_
110 1110 1000
1 1 0
0 1 1
_
_

_
_
1 0 1
0 1 1
0 0 0
_
_
=: B
1
A10I =
_
_
101 1110 1000
1 10 0
0 1 10
_
_

_
_
1 0 100
0 1 10
0 0 0
_
_
=: B
2
A100I =
_
_
10 1110 1000
1 100 0
0 1 100
_
_

_
_
1 0 10000
0 1 100
0 0 0
_
_
=: B
3
.
To nd an eigenvector w
2
=
_
x y z

T
of eigenvalue 10, we need to solve (A10I)w
2
= 0, or
equivalently B
2
w
2
= 0, which just reduces to x = 100z and y = 10z with z arbitrary. Taking
z = 1, we see that
_
100 10 1

T
is an eigenvector of eigenvalue 10. Treating the other two
eigenvalues in the same way, we nd that the vectors
w
1
=
_
_
1
1
1
_
_
w
2
=
_
_
100
10
1
_
_
w
3
=
_
_
10000
100
1
_
_
are eigenvectors of eigenvalues 1, 10 and 100 respectively.
(c) By inspection we have
u
0
=
_
_
1110000
1010100
1001001
_
_
=
_
_
1000000
1000000
1000000
_
_
+
_
_
100000
10000
1000
_
_
+
_
_
10000
100
1
_
_
= 1000000
_
_
1
1
1
_
_
+ 1000
_
_
100
10
1
_
_
+
_
_
10000
100
1
_
_
= 10
6
w
1
+ 10
3
w
2
+ w
3
.
(d) Recall that Aw
1
= w
1
and Aw
2
= 10w
2
and Aw
3
= 100w
3
. It follows that for all n 0 we have
A
n
w
1
= w
1
and A
n
w
2
= 10
n
w
2
and A
n
w
3
= 100
n
w
3
= 10
2n
w
3
. This gives
u
n
= A
n
u
0
= A
n
(10
6
w
1
+ 10
3
w
2
+ w
3
) = 10
6
A
n
w
1
+ 10
3
A
n
w
2
+ A
n
w
3
= 10
6
w
1
+ 10
n+3
w
2
+ 10
2n
w
3
= 10
6
_
_
1
1
1
_
_
+ 10
n+3
_
_
10
2
10
1
_
_
+ 10
2n
_
_
10
4
10
2
1
_
_
=
_
_
10
6
+ 10
n+5
+ 10
2n+4
10
6
+ 10
n+4
+ 10
2n+2
10
6
+ 10
n+3
+ 10
2n
_
_
.
In particular, a
n
is the bottom entry in u
n
, which is
a
n
= 10
6
+ 10
n+3
+ 10
2n
.
(e) Our formula gives
a
0
= 10
6
+ 10
3
+ 10
0
= 1001001
a
1
= 10
6
+ 10
4
+ 10
2
= 1010100
a
0
= 10
6
+ 10
5
+ 10
4
= 1110000
9
as it should. We also have
111a
n+2
1110a
n+1
+ 1000a
n
=111(10
6
+ 10
n+5
+ 10
2n+4
) 1110(10
6
+ 10
n+4
+ 10
2n+2
) + 1000(10
6
+ 10
n+3
+ 10
2n
)
=10
6
(111 1110 + 1000) + 10
n+3
(11100 11100 + 1000) + 10
2n
(1110000 111000 + 1000)
=10
6
+ 1000 10
n+3
+ 1000000 10
2n
= 10
6
+ 10
n+6
+ 10
2n+6
= a
n+3
.
Exercise 9. A be an nn matrix, and let
1
, . . . ,
h
be h distinct eigenvalues of A. For each i = 1, . . . , h,
let the vectors v
i,1
, . . . , v
i,t
i
be linearly independent eigenvectors of A all corresponding to the eigenvalue

i
. We collect these lists together into a single list
v
1,1
, . . . , v
1,t
1
, v
2,1
, . . . , v
2,t
2
, . . . , v
h,1
, . . . , v
h,t
h
.
Prove (as was stated in lectures) that this list is linearly independent.
Solution: For each i = 1, . . . , h, the vectors v
i1
, . . . , v
it
i
are linearly independent eigenvectors of A all
corresponding to the eigenvalue
i
. We show that
v
11
, . . . , v
1t
1
, v
21
, . . . , v
2t
2
, . . . , v
h1
, . . . , v
ht
h
(taken all together) are linearly independent by induction on h.
When h = 1, there is nothing to prove, because we are given that v
11
, . . . , v
1t
1
are linearly independent.
Assume now that h > 1 and that the claim is true for h 1 distinct eigenvalues of A.
Let
a
11
, . . . , a
1t
1
, a
21
, . . . , a
2t
2
, . . . , a
h1
, . . . , a
ht
h
be scalars such that
(1)
h

i=1
t
i

j=1
a
ij
v
ij
= 0.
Then
0 = A0 = A
_
_
h

i=1
t
i

j=1
a
ij
v
ij
_
_
=
h

i=1
t
i

j=1
a
ij
Av
ij
and so
(2)
h

i=1
t
i

j=1
a
ij

i
v
ij
= 0
because Av
ij
=
i
v
ij
for all j = 1, . . . , t
i
and i = 1, . . . , h. If we now subtract
h
times (1) from (2) we
get
h

i=1
t
i

j=1
a
ij
(
i

h
)v
ij
= 0,
that is
h1

i=1
t
i

j=1
a
ij
(
i

h
)v
ij
= 0.
By the induction hypothesis,
v
11
, . . . , v
1t
1
, v
21
, . . . , v
2t
2
, . . . , v
h1,1
, . . . , v
h1,t
h1
(taken all together) are linearly independent. Therefore
a
ij
(
i

h
) = 0 for all j = 1, . . . , t
i
and i = 1, . . . , h 1.
Since
i

h
= 0 for all i = 1, . . . , h 1, it follows that
a
ij
= 0 for all j = 1, . . . , t
i
and i = 1, . . . , h 1.
With this information, equation (1) now simplies to
t
h

j=1
a
hj
v
hj
= 0,
and so it follows from the fact that v
h1
, . . . , v
ht
h
are linearly independent that a
h1
= = a
ht
h
= 0.
10
We have now shown that
v
11
, . . . , v
1t
1
, v
21
, . . . , v
2t
2
, . . . , v
h1
, . . . , v
ht
h
are linearly independent. This completes the inductive step. By the Principle of Mathematical Induction,
the claim is proved.
11