Australian Catholic University

Semester 2 2015

Strathfield

Geometry and Topology

MATH312

Limits and Continuity

35.

In each case decide whether f : R R is continuous at the given point.

f (x) = 3x + 5 at x = 1

(b)

f (x) =

x1
2

(c)

f (x) =

(d)

f (x) = x2 at x = 0

(e)

f (x) = x2 at x = 2

at x = 0

x,
x + 1,

if x > 0
if x 0

at x = 0.

Solution.
First we remind ourselves that we only need to consider open balls. Suppose that we have shown
that the inverse image of every open ball contains an open ball centred at the point of interest.
Now let B be an open set and let A = f 1 (B). For any point x A we know that f (x) B
the
and so we can find an open ball about f (x) that lies completely in B. But we know that
inverse image of that open ball is an openset that must contain x. Thus we can find an open
ball about x that lies completely within f 1 (B). Thus, given anypoint of f 1 (B) we can find
an open ball centred at that point the lies entirely within f 1 (B) which must therefore be open.
Hence, we only need to show that the inverse image of every open ball contains an open ball.
In this particular question, we only need consider open balls centred at the given point.
(a)

Let > 0 and consider the open ball B f (1) = B (8). This is the set of numbers
between 8 and 8 + .
As we will soon see, polynomials of degree 1 are too easy.
It is clear that f 1 B (8) is the set of numbers between a and b where f (a) = 8
and f (b) = 8 + . You can check that

f 1

f 1+


=8
3

=8+
3

Thus f 1 B (8) is the interval from 1 3 to 1 + 3 and hence actually is an open
ball (with one third the radius). Therefore, f is continuous at 1.
312t0515

35

(b)

A very similar argument will show that, in this case, that the in inverse image of
an open ball about 21 is an open ball with twice the radius. Hence this function
is continuous at x = 0.

(c)

This time we are suspicious. Looked at from the high school perspective, the limit
at 0 from the left is 0, while the limit from the right is 1. While not a proof (for
us) this is very good evidence that this function is not continuous at x = 0. To
show this, we need to find a counter-example. That is, we must find an open ball
about f (0) = 1 whose inverse image does not contain an open ball about 0. Here
the function jumps exactly 1, so we will choose a positive
value for that is less
than 1, say we choose = 21 > 0. Consider B 21 (1) = 12 , 23 . If x > 0 then
f (x) = x

and so f 1 1, 23 = 1, 32 . We observe that f (0) = 1 and if x < 0 then f (x) = x+1
and so

 

1
1
1
f
,1 = ,0
2
2
Hence
f


 1  1 3
B 12 (1) =
,0
,
2
2 2

It is clear that there is no open ball about 0 that is contained entirely within this
set. Hence, we have found an open ball about 1 whose inverse image is not open,
and so f is not continuous at x = 0.
(d)

We now meet a function that is not a polynomial of degree 1 and start to see
the problems we encounter. We note that f (0) = 0, and so we consider an
open ball B (0) for some > 0. We seek the set of real numbers x that satisfy
f (x) B (0), that is the numbers such that
< x2 <
We can safely ignore the negative half of this interval and consider those numbers
that satisfy
0x<
It is easy to see that this set is the interval

( , )
given that we may take the square root because > 0. This interval is an open
ball about 0 and so we are finished.

(e)

Now the fun really starts. Consider the open ball B f (2) = B (4) = (4 , 4+ ).
There are two possibilities: 4 > 0 or 4 0. This consideration tells us whether
the inverse image falls into two pieces or not.

36

In the first case, with two regions, the inverse image will be

( 4 + , 4 ) ( 4 , 4 + )
For the second case (where we have changed the scale to fit it in) there is one region
of the form

( 4 + , 4 + )
In both cases it can be seen that there is an open ball centred at 2 that is contained
in the inverse image.
As an exercise, you should try to write down the radius of the largest open ball
about 2 that lies within the inverse image of B (4).

36.

Consider R2 with the normal distance as the metric.

Define f : R2 R2 by

f (x, y) = (y, x)

Show that f is continuous.

Only for those who have done, or are doing MATH307, show that all isometries are
continuous.

Solution.
It is easier to do these all at once. If we have an isometry then distance is preserved and all
points of R2 are in the range of f . That is, for all points x and y we have

d(x, y ) = d f (x), f (y )

We can now show that the inverse image of every open ball is not only open but is in fact an
open ball.
Given (a, b) there is a point (x, y) such that
f (x, y) = (a, b), because all points are in the
1
range of f . Let >
B (a, b) . Suppose that (c, d) B (x, y), then we must

and consider f
have d (x, y), (c, d) < and hence




d f (x, y), f (c, d) = d (x, y), (c, d) <

Thus f (c, d) B (x, y) and so




B (x, y) f 1 B f (x, y)

A similar proof will show that, in fact




B (x, y) = f 1 B f (x, y)




Hence, f 1 B f (x, y) is an open set.

As always, this implies that the inverse image of every open set is open.
37

37.

Given the metric space A, show that the sets and A are open.

Solution.
Suppose our metric is d and let x A. Picking any positive number, say 1, we consider the
open ball B1 (x). By definition this is a subset of A. Thus, for any element x A we can find
an open ball about x that is contained entirely within A. Hence A is open.
To show that the empty set is open is actually easier than this but most people dont like the
proof. We need to show that for every point in we can find an open ball about that point that
lies entirely within the empty set. This is vacuously true. That is, there can be no points that
violate this condition and so the requirement is satisfied! Hence, is open.
If you dont like either of these proofs, dont worry you are not alone. The good news is that
very soon we will avoid this by moving to a situation were it will be an axiom that A and are
open.

38. Let A and B be metric spaces, and let b B be a chosen element of B. We define a (constant)
function from A to B as follows. Let f : A B be the function such that f (x) = b for all x A.
Show that f is continuous.
Solution.
As always we need to show that the inverse image of any open set is open. Here there are
only two cases. Let S B be an open set. We wish to evaluate f 1 (S). There are two cases,
either b S or b
/ S.
If b S then for all x A we have f (x) = b S and so f 1 (S) = A, which we have just seen is
an open set.
On the other hand, if b
/ S then no element of A has its image in S and so f 1 (S) = , which
is also open.
Thus in any case the inverse image of an open set is open. (In fact we have shown that the
inverse image of every set is open!) Hence f is continuous.

39. Let A be metric space, we define the identity function from A to A as follows. Let f : A A
be the function such that f (x) = x for all x A.
Show that f is continuous.
Solution.
After you have completed a few continuity questions the identity function comes as a bit of
relief. As f (x) = x for all x we also know that f 1 (S) = S for all subsets S A. So if S is an
open set then so is f 1 (S) = S! Hence the identity function is always continuous.
38

40.

Given functions f : A B and g : B C and a subset D C, show that

Solution.


If x (g f )1 (D), then g f (x) D. Thus f (x) g 1 (D) and hence x f 1 g1 (D) . This
tells us that every element of the set on the right is also an element of the set on the left, that is


(g f )1 (D) f 1 g1 (D)


1 1
It only remains to show the reverse inclusion.
Suppose
that
y

f
g
(D)
, then we know




1
that f (y) g (D) and hence g f (y) D. But g f (y) = g f (y) and so y (g f )1 (D)
and we are finished.

41. Suppose that f : A B and g : B C are continuous functions, with A, B and C metric
spaces. Show that g f : A C is continuous.
Solution.
It is this proof that demonstrates the benefit of talking in terms of open sets. We need to show
that the inverse image of every open set is open. Let S be an open set in C. As g is continuous


we know that g 1 (S) is open. We also know that f is continuous and so f 1 g1 (S) is open.
By the last exercise we now know that


(g f )1 (S) = f 1 g1 (S)

is open. Thus g f is continuous.

42.

(a)

Consider the sequence

2,

s r
r q
q

2 2, 2 2 2,
2 2 2 2, . . .

(i)

Evaluate the first few terms and try to decide what the limit of this sequence might
be. Assuming it has a limit, of course.

(ii)

Again assuming that this sequence has a limit evaluate that limit. That is, show
that your guess from the first part is correct.

(iii)

Show that all terms of this sequence are less than or equal to 2.

(iv) Show that this is an increasing sequence. That is, each term is at least as big as
the term before.
39

(b)

Consider the sequence

2,

2+

2,

2+

2+

2,

2+

2+

2+

2, . . .

(i)

Evaluate the first few terms and try to decide what the limit of this sequence might
be. Assuming it has a limit, of course.

(ii)

Again assuming that this sequence has a limit evaluate that limit. That is, show
that your guess from the first part is correct.

(iii)

Show that all terms of this sequence are less than 2.

(iv) Show that this is an increasing sequence. That is, each term is at least as big as
the term before.
(c) What can you say if 2 is replaced in both of these by the positive integer m?
Solution.
To complete this question properly we need one fact. I wont prove it here but it
is known that if a sequence is increasing and all elements are less than some fixed
number, then that sequence must converge. (And similarly if it is decreasing and
greater than a fixed number.)
(a)

(i)

Hopefully you used your index laws for this one, if you did then you get a
big hint about the limit.

1
2 = 22
q

1
3
1
2 2 = 22 24 = 24
r q

1
1
1
7
2 2 2 = 22 24 28 = 28
s r
q

1
1
1
1
15
2 2 2 2 = 2 2 2 4 2 8 2 16 = 2 16

(ii)

If you can be bothered you might use induction to prove that the nth member
of this sequence is
2
And clearly the limit of

(iii)

2n 1
2n

2n 1
2n

is 1 and so the limit of this sequence is 21 = 2.

th
We do the last
the n term of the sequence and
two parts together. Let an be
so an+1 = 2an . We know that 0 < a1 = 2 2. Suppose that 0 < an 2,
then we have

0 < an 2

a2n 2an
a2n 2an 0

(an 2an )(an 2an ) 0

as an > 0


But we know
that
a
and
2an are positive and hence their sum is positive.
n

Thus an 2an 0, that is

an 2an = an+1
It only remains to show that an+1 2. Now 0 < an 2 and 0 < 2 2 and
so
2an 2 2
a2n+1 22

an+1 2

as everything is positive here. Thus 0 < an an+1 2. Hence, by induction,

Our arguments for this part will be close to but (obviously) different to the last
part.
(i)

This time we have no nice simplification, all we can do is calculate the terms.
Let bn be the nth term of the sequence.

2 = 1.4142 . . .
q

2 + 2 = 1.8477 . . .
r
q

2 + 2 + 2 = 1.9615 . . .
s
r
q

2 + 2 + 2 + 2 = 1.9903 . . .

(ii)

Here we need to use a trick. Suppose that lim bn = l. The following works
n
once we know that there is a limit but you
should be suspicious as it is a
very sneaky trick. We know that bn+1 = 2 + bn so, if we take the limit of
both sides we get.
p
2 + bn
lim bn+1 = lim
n
n

l = 2+l
2
l =2+l
2
l l2=0
(l 2)(l + 1) = 0

Thus l = 2 or l = 1. As all terms are clearly positive we must have a

Another induction proof to do this properly. However, we can convince you

an + 2 < 2 + 2 = 4

an + 2 < 4 = 2
an+1 < 2
41

Filling in the rest will show, by induction, that the claim is true.

(iv)

We know that 0 < an < 2 and so an + 1 > 0 and an 2 < 0, hence

a2n an 2 < 0
a2n < 2 + an

an < 2 + an = an+1

Hopefully you realise that to find the start of this little proof we worked
backwards. If an < an+1 , then . . ., and then checked to see that we could
reverse the argument.

(c)

You should have no problem in convincing yourselves that the limit of the sequence

m,

m m,

s r
r q
q

m m m, m m m m, . . .

is m. As for the case m = 2 the terms of this sequence are equal to m

2n 1
2n

and the

For the sequence

m,

m+

m,

m+

m+

m,

m+

m+

m+

m, . . .

things are a little more complicated. If we assume that this sequence has a limit l,
then the same argument we used before shows that
l=

m+l

l lm=0

1 1 + 4m
l=
2
1 + 1 + 4m
=
2
As all the terms of the sequence are positive we may ignore the possible negative
value for l.

All that remains is for you to show that this sequence in increasing and bounded
above. That is the posh way of saying that an < an+1 for all n and that we can
find a number that is bigger than all the terms of the sequence.
42

43$. Recall the situation. We know the angle of the ecliptic to the celestial equation and the angle
of the sun from the first point of Aries. That is, we know and in the diagram below and we wish
to find and .
N

C
M

Recall that N is the north pole and so the arc AN and hence the length AN is
and AC are both 2 as that is how we chose AC.

.
2

Similarly AB

(a)

Given that AB and AC are both

(b)

Using the result of question 24, show that tan = tan cos .

(c)

Using variants of Menelaus theorem (yes, you will need to find or discover them) show
that sin = sin sin .

show that AC is .

Solution.
(a)

As with the calculations for the area of a lune, we observe that the angle between
the planes forming the great circles is . We now look at the plane forming the
great circle joining A and C. The lines joining A and C to the centre of the sphere
must form an angle . As the radius is 1, we find AC = .
A

(b)

Now things get tricky! (As if they werent already.) From before we know the
following.
sin(arcA N ) sin(arcC M )
sin(arcA L )
=

sin(arcL B)
sin(arcN C)
sin(arcM B)
43

First some easy ones: arcA N = 2 and so sin(arcA N ) = 1. Now observe

sin
cos
sin
sin
sin
= cos
cos
cos
tan = tan cos

(c)

Keep trying!

44

taking reciprocals