Understanding Analysis

Solution of exercise problems.

Quasar Chunawala
Version 1.0.1, last revised on 2022-10-15.
Abstract
This is a solution manual for Understanding Analysis, 2nd edition, by Stephen Abbott.

Example 1.3.7. Let A ⊆ R be nonempty and bounded above and let c ∈ R. Define the set
c + A by

c + A = {c + a : a ∈ A}
Then, sup(c + A) = c + sup A.
Proof.
Since A is non-empty, bounded subset of R, by the Axiom Of Completeness, there exists a
least upper bound of A; we denote it by sup A.
By the definition of least upper bound, we are required to prove that:
(1) c + sup A is an upper bound of c + A.
(2) If u is an upper bound of c + A, then c + sup A ≤ u.
(1) We know that, a ≤ sup A, for all a ∈ A.
So, c + a ≤ c + sup A for all c + a ∈ c + A. So, c + sup A is an upper bound of c + A.
(2) Let u be an arbitrary upper bound of c + A. Then, c + a ≤ u for all a ∈ A. Thus,
a ≤ u - c for all a ∈ A. Thus, u - c is an upper bound for A. Consequently, sup A ≤ u - c.
Thus, c + sup A ≤ u. As u was arbitrary to begin with, this is true for all such upper
bounds.
Consequently, sup(c + A) = c + sup A.
Lemma 1.3.8. Assume s ∈ R is an upper bound for a set A ⊆ R. Then, s = sup A if and
only if, for all 𝜖 > 0, there exists an element a ∈ A satisfying s - 𝜖 < a.
Proof.
⟹ direction.
We are given that, s be an upper bound for the set A ⊆ R. We proceed by contradiction.
Assume that s = sup A, such that there exists 𝜖0 > 0, such that for all a ∈ A, s - 𝜖0 > a.
Clearly, s - 𝜖0 is an upper bound for A. But, s - 𝜖0 < s = sup A. This is a contradiction.
Hence our initial assumption is false.
It follows, that if s = sup A, then for all 𝜖 > 0, there exists a ∈ A, such that s - 𝜖 < a < s.
⟸ direction.
s is an upper bound for A. We are given that, for all 𝜖 > 0, ∃a ∈ A, such that s - 𝜖 < a.
We proceed by contradiction. Assume that s is not the supremum of A. Then, there exists
an upper bound t for A, such that t < s.
Put 𝜖 = s - t. Then, ∃a ∈ A, such that s - (s - t) = t < a. But, this contradicts the fact
that t is an upper bound for A. Hence, our initial assumption is false. s = sup A.
□
[Abbott 1.3.1. (a)] Write a formal definition in the style of definition 1.3.2 for the
infimum or the greatest lower bound of a set.
Definition (Greatest Lower Bound). A real number l is the greatest lower bound for a
set A ⊆ R, if it meets the following two criteria:
(i) l is a lower bound for A.
(ii) If m is an arbitrary lower bound for A, m ≤ l.
(b) Now, state and prove a version Lemma 1.3.8. for the greatest lower bound.
Let l be a lower bound for A ⊆ R.Then, l = inf A, if and only for all 𝜖 > 0, there exists
a ∈ A, such that a < l + 𝜖.

⟹ direction. l is a lower bound for A. We are given that l = inf A. We proceed by

contradiction. Assume that ∃𝜖0 > 0, such that ∀a ∈ A, a ≥ l + 𝜖0 . Thus, l + 𝜖0 is a lower
bound for A. But, l + 𝜖0 > l = inf A. This is a contradiction. Hence, our initial assumption
is false.
∀𝜖 > 0, ∃a ∈ A, such that a < l + 𝜖.

⟸ direction.
lis a lower bound of A. We are given that, ∀𝜖 > 0, ∃a ∈ A, such that a < l + 𝜖. We
proceed by contradiction. Assume that l is not the infimum of A. Then, there exists a lower
bound m for A, such that m > l.
Put 𝜖 = m - l. It follows that, there exists a ∈ A, such a < l + (m - l) = m. But, this
contradicts the fact that m is a lower bound for A. Hence, our initial assumption must be
false.
l = inf A.

Exercise 1.3.2. Give an example of each of the following, or state the request is impossible.
(a) A set B with inf B ≥ sup B.
It is impossible for a set to satisfy inf B > sup B. For a singleton set, inf B = sup B.
(b) A finite set that contains its infimum but not its supremum.
This request is impossible. A finite set always contains its infimum and supremum.
(c) A bounded subset of Q, that contains its supremum but not it's infimum.
Consider A = { r : 0 < r ≤ 1, r ∈ Q }. This is a bounded subset of Q.
We have, inf A = 0. 0 ∉ A. sup A = 1. 1 ∈ A.
Exercise 1.3.4. Let A1 , A2 , A3 , … be a collection of non-empty sets, each of which is
bounded above.
N
(a) Find a formula for sup(A1 ∪ A2 ). Extend this to sup ⋃ Ak .
k=1

Proof.
Our claim is that sup(A1 ∪ A2 ) = max sup A1 , sup A2 .
Let s1 = sup A1 and s2 = sup A2 . Let m = max{ s1 , s2 }.
(a)Claim: m is an upper bound for A1 ∪ A2 .
Let a be an arbitrary element of A1 ∪ A2 . Then, either a ∈ A1 or a ∈ A2 or both are true.
Consequently, either a ≤ s1 or a ≤ s2 or both are true. Since, s1 ≤ and s2 ≤ m, in both
cases, we must have that, a ≤ m.
(b)Claim: If u is an upper bound for A1 ∪ A2 , then m ≤ u.
Let u be an upper bound for A1 ∪ A2 . Then, a ≤ u for all a ∈ A1 ∪ A2 .
So, a' ≤ u for all a' ∈ A1 and a'' ≤ u for all a'' ∈ A2 . So, u is an upper bound for A1 and
A 2 . So, s 1 ≤ u and s 2 ≤ u are both true. Consequently, m ≤ u.

From (a) and (b) it follows that, m = sup(A1 ∪ A2 ).

We can extend this to finite N.
Claim: sup ∪ kn=1 Ak = max1≤k≤n sup Ak .
The claim is true for k = 1 and k = 2. We proceed by the principle of mathematical
induction.
Assume that P(n) is true. That is, we assume that

sup ∪ kn=1 Ak = max1≤k≤n sup Ak

Our inductive hypotheses is:
+1
sup ∪ kn=1 A k = max1≤k≤n+1 sup A k

We have:

+1
sup ∪ kn=1 ∪ kn=1 Ak ∪ An+1
A k = sup

= max sup ∪ kn=1 Ak , sup An+1

= max max1≤k≤n sup Ak , sup An+1

= max1≤k≤n+1 sup Ak
 ∞
(b) Consider sup ⋃ Ak . Does the formula in (a) extend to the infinite case?
k=1

Exercise 1.4.1. Recall that I stands for the set of irrational numbers.
(a) Show that if a, b ∈ Q, then ab and a + b are elements of Q as well.
Proof.
m p
Let a = , b = with n and q matural numbers, m and p as integers in Z.
n q

The addition of rational numbers a + b is defined as :

m p mq + np
+ :=
n q nq
Since, both n and q are non-zero, nq ≠ 0. The product of two natural numbers is a natural
number so, nq ∈ N. Moreover, mq + np ∈ Z. Consequently, m / n + p / q ∈ Q. Hence, Q
is closed under addition.
The product of rational numbers ab is defined as:
mp
a×b =
nq

Since, Z is closed under multiplication, mp ∈ Z. Similarly, nq ∈ N and nq ≠ 0.

Consequently, the product a × b belongs to Q.
(b) Show that if a ∈ Q and t ∈ I, then a + t ∈ I and at ∈ I as long as a ≠ 0.
Exercise 1.4.2. Let A ⊆ R be non-empty and bounded above, and let s ∈ R have the
1 1
property that for all n ∈ N, s + is an upper bound for A and s - is not an upper
n n
bound for A. Show that s = sup A.
Proof.
Claim. s is an upper bound for A.
We proceed by contradiction.
Assume that s is not an upper bound for A. Then there exists a0 ∈ A, such that s < a0 . By
the Archimedean property, there exists a natural number n0 ∈ N, such that

1
< a0 - s
n0
Thus, there exists n0 ∈ N such that:
1
s+ < s + (a 0 - s)
n0
1
s+ < a0
n0
1
That is, there exists n0 ∈ N, such that s + is not an upper bound for A.
n0
1
But, this is a contradiction. ∀n ∈ N, s + is an upper bound for A.
n

Claim. If t is any other upper bound for A, s ≤ t.

We proceed by contradiction.
Assume that there exists an upper bound t0 , for A, such that t0 < s. By the Archimedean
property there exists m0 ∈ N such that

1
< s - t0
m0
So, there exists m0 such that:

1
s- > s - (s - t 0 )
m0
= t0
1
That is, there exists m0 ∈ N, such that s - is an upper bound for A.
m0
 1
But this is a contradiction. ∀m ∈ N, s - is not an upper bound for A.
m

This closes the proof. s = sup A.

□
1
Exercise 1.4.3. Prove that ∩ n∞=1 0, = ∅. Notice that this demonstrates that the
n
intervals in the Nested Interval Property must be closed for the conclusion of the theorem
to hold.
Proof.
1
Let In = 0, . Let x be an arbitrary real number in I1 = (0, 1). By the Archimedean
n
property, there exists a natural number n0 ∈ N, such that

1
<x
n0
Thus, x ∉ In and therefore x ∉ ∩ n∞=1 In .
0

Since x was arbitrary, this holds for all 0 < x < 1.

∩ n∞=1 In = ∅

Cantor's Diagonalization Method

Cantor published his discovery that R is uncountable in 1874. Althoug it has some modern
polish on it, the argument presented in theorem 1.5.6 (ii) is actually quite similar to the one
Cantor originally found. In 1891, Cantor offered another proof of this same fact that is
startling in its simplicity. It relies on decimal representations for real numbers, which we will
accept and use without any formal definitions.
Theorem 1.6.1. The open interval (0, 1) = { x ∈ R : 0 < x < 1 } is uncountable.
Exercise 1.6.1. Show that (0, 1) is countable if and only if R is uncountable.
Proof. As with Theorem 1.5.6., we proceed by contradiction and assume that there does
exist a function f : N → (0, 1) that is 1 - 1 and onto. For each m ∈ N, f(m) is a real
number between 0 and 1 and we represent it using the decimal notation

f( m) = . a m 1 a m 2 a m 3 a m 4 …

What is meant here is that for each m, n ∈ N, amn is the digit from the set { 0, 1, 2, … , 9 }
that represents the nth digit in the decimal expansion of f(m). The 1 - 1 correspondence
between N and (0, 1) can be summarized in the doubly indexed array:

N (0, 1)
1 ⟷ f(1) = . a11  a12  a13  a14  a15  a16   …
2 ⟷ f(2) = . a21  a22  a23  a24  a25  a26   …
3 ⟷ f(3) = . a31  a32  a33  a34  a35  a36   …
4 ⟷ f(4) = . a41  a42  a43  a44  a45  a46   …
5 ⟷ f(5) = . a51  a52  a53  a54  a55  a56   …
6 ⟷ f(6) = . a61  a62  a63  a64  a65  a66   …
⋮ ⋮ ⋮ ⋮ ⋮ ⋮ ⋮ ⋮ ⋱

The key assumption abou this correspondence is that every real number in (0, 1) is assumed
to appear somewhere on the list.
Now for the pearl of the argument. Define a real number x ∈ (0, 1) with the decimal
expansion x = . b1 b2 b3 b4 … using the rule:

bn =
2 if ann ≠ 2
3 if ann = 2

Exercise 1.6.2. (a) Explain why the real number x = . b1 b2 b3 b4 … cannot be f(1).
Since the first digit b1 in the decimal expansion of x differs from the first digit a11 in the
decimal expansion of f(1), x ≠ f(1).
(b) Now, explain why x ≠ f(2) and in general why x ≠ f(n) for any n ∈ N.
Since the second digit b2 in the decimal expansion of x differs from the second digit a22 in
the decimal expansion of f(2), x ≠ f(2).
In general, since the nth digit bn in the decimal expansion of x differs from the nth digit
ann in the decimal expansion of f(n), x ≠ f(n), for all n ∈ N.

(c) Point out the contradiction that arises from these observations and conclude that (0, 1)
is uncountable.
This shows that x does not belong to the set { f(1), f(2), f(3), … }. This contradicts the
fact that x ∈ (0, 1). Hence our initial assumption is false. The set (0, 1) is uncountable.
Exercise 2.2.1. What happens if we reverse the order of the quantifiers in definition 2.2.3?
Definition. A sequence (xn ) verconges to x if there exists 𝜖 > 0 such that for all N ∈ N, it
is true that n ≥ N implies |xn - x| < 𝜖.
Give an example of a vercongent sequence. Is there an example of a vercongent sequence
that is divergent? Can a sequence verconge to two different values? What exactly is being
described in this strange definition?
Proof.
1
The sequence (xn ) = is vercongent.
n

Consider the sequence (xn )n∞=1 = (-1) n . Pick 𝜖 = 2. For all N ∈ N, if n ≥ N, it follows that

|xn - 0| = 1 < 2 = 𝜖
Thus, it is a vercongent sequence.
A sequence can verconge to two different values. Let (xn ) = (-1) n .
Pick 𝜖 = 2. For all N ∈ N, if n ≥ N, it follows that
 1
|x n - | < 2 = 𝜖
2
1
Consequently, (xn ) also verconges to .
2

Exercise 2.2.2. Verify using the definition of the convergence of a sequence, that the
following sequences converge to the proposed limit.
2n + 1 2
(a) lim = .
5n + 4 5

Solution.
2n + 1 2
We are interested to make the distance | - | as small as we please.
5n + 4 5
Let's explore the inequality:

2n + 1 2
| - |<𝜖
5n + 4 5
|5(2n + 1) - 2(5n + 4)|
∴ <𝜖
|5(5n + 4)|
3
<𝜖
5(5n + 4)
3 3
We know that < . We can strengthen the inequality we wish to prove, by
5(5n + 4) 5(5n)
choosing an upper bound for the left hand side of the inequality. So, we are interested to
have:

3
<𝜖
25n
3
Let's choose N > . To show that this choice of N indeed works, pick an arbitrary
25𝜖
𝜖 > 0. Then, n ≥ N implies that:
 3
n >
25𝜖
1 25𝜖
⟹ <
n 3
3
⟹ <𝜖
25n
2n + 1 2
⟹ | - | <𝜖
5n + 4 5

2n 2
(b) lim = 0.
n3 + 3

Solution.
2n 2
We are interested to make the distance | - 0| as small as we please.
n3 + 3

Let's explore the inequality:

2n 2
| |<𝜖
n3 + 3
Since all quantities are positive, we may write:

2n 2
<𝜖
n3 + 3
2n 2 2n 2
Now, 3 < 3 . We can strengthen the condition we wish to prove by replacing the
n +3 n
left hand side of the above inequality by it's upper bound. Therefore, we are interested in
making

2
<𝜖
n
2
We choose N > . To show that this choice of N indeed works, we pick an arbitrary 𝜖 > 0.
𝜖
Then, for all n ≥ N, we have
 2
n >
𝜖
2
⟹ <𝜖
n
2n 2
⟹ | | <𝜖
n3 + 3
2n 2
Hence, lim = 0.
n3 + 3

sin n 2
(c) lim = 0.
∛n

Proof.
sin n 2
We are interested to make the distance | | as small as we please.
∛n

Let's explore the inequality :

sin n 2
| |<𝜖 (1)
∛n
We know that,

|sin n 2 | < 1

We can strengthen the condition we are interested to prove by replacing left-hand side in
the inequality (1) by it's upper bound. We have:

1
<𝜖
∛n
1
Let's pick N > 3 . Then, for all n ≥ N, it follows that:
𝜖
 1
n >
𝜖3
1
⟹ <𝜖
∛n
sin n 2
⟹ | | <𝜖
∛n

n2
Consequently, lim sin = 0.
∛n

Example 2.2.4. Give an example of each or state that the request is impossible. For any
that are impossible, give a compelling argument for why that is the case.
(a) A sequence with an infinite number of ones that does not converge to one.
Proof.
Consider the sequence

(xn ) = (1, 0, 1, 0, 1, 0, … )
This sequence does not converge to 1, but has an infinite number of 1s.
(b) A sequence with an infinite number of ones that converges to a limit not equal to one.
This request is impossible. We proceed by contradiction. Assume that there exists a
sequence (xn ) with an infinite number of 1s and converges to a limit l ≠ 1.
The sequence (xn ) has infinite number of 1s. The distance between 1 and l is |l - 1|.
|l - 1|
Pick 𝜖0 = .For all N ∈ N, it follows that there are atleast some terms of the
2
sequence (xn ) beginning with or after the Nth term, such that the distance |xn - l| ≥ 𝜖0 .
Consequently, the sequence does not converge to l. This is a contradiction.
Thus, there is no sequence (xn ).
(c) A divergent sequence such that for every n ∈ N, it is possible to find conseceutive ones
somewhere in the sequence.
Solution.
Consider the sequence (xn ) = (1, 1, -1, 1, 1, -1, 1, 1, -1, … ). This is a divergent sequence,
where for all n ∈ N, it is possible to find consecutive ones in the sequence.
Exercise 2.2.6. Prove theorem 2.2.7. To get started, assume (an ) → a and also that
(an ) → b. Now argue that a = b.

Assume that (an ) → a and (an ) → b. Pick an arbitrary 𝜖 > 0.

There exists N1 (𝜖) ∈ N such that for all n ≥ N1 :

| a n - a| < 𝜖 / 2 (1)
There exists N2 (𝜖) ∈ N such that for all n ≥ N2 :

| a n - b| < 𝜖 / 2 (2)
Pick N = max{ N1 , N2 }. For n ≥ N, both the inequalities (1) and (2) are satisfied.
We have:

| ( a n - b ) - ( a n - a) | ≤ | a n - b | + | a n - a| Triangle Inequality
𝜖 𝜖
| a - b| < + =𝜖
2 2
Since 𝜖 was arbitrary to begin with, this is true for all 𝜖 > 0. So, for all 𝜖 > 0, the distance
|a - b| can be made smaller than 𝜖. Consequently, a = b.

Exercise 2.3.1. Let xn ≥ 0 for all n ∈ N.

(a) If (xn ) → 0, show that x n → 0.

Proof.
We are interested to make the distance | xn | as small as we please.
Let's explore the inequality:

| xn | < 𝜖
2
| xn | 2 = | x n | = |x n | < 𝜖 2

Pick an arbitrary 𝜖 > 0. As (xn ) → 0, we know that there exists N𝜖 ∈ N such that, for all
n ≥ N𝜖 , we have:

|x n | < 𝜖 2

Since 𝜖 was arbitrary, it follows that for all 𝜖 > 0, there exists N ∈ N, such that, for all
n ≥ N,

| xn | < 𝜖

(b) If (xn ) → x, show that xn → x.

If x = 0, then (a) holds and we are done. Assume that x ≠ 0.

We are interested to prove that the distance | xn - x| can be made as small as well
please. Pick an arbitrary 𝜖 > 0. Let's explore the inequality :

| xn - x| < 𝜖
This can be rewritten as:

| x n - x|
<𝜖
xn + x
1
We can strengthen the condition we wish to prove, by replacing by its upper
xn + x
bound. Since, xn ≥ 0, x n ≥ 0. So,
 1 1
<
xn + x x
So, our claim is:

| x n - x|
<𝜖
x
that is, |xn - x| <𝜖 x

Since (xn ) → x, there exists N ∈ N, such that for all n ≥ N, we have:

| x n - x| < 𝜖 x
Thus, for all n ≥ N,

| xn - x| < 𝜖
Consequently, x n → x.

Exercise 2.3.2. Using only definition 2.2.3, prove that if (xn ) → 2, then
2xn - 1
(a) → 1.
3

Proof.
2xn - 1
We are interested to make the distance -1 as small as we please.
3

Pick an arbitrary 𝜖 > 0. We would like to have:

 2xn - 1
-1 <𝜖
3
2xn - 1 - 3
or , <𝜖
3
2
| xn - 2 | < 𝜖
3
3𝜖
| xn - 2 | <
2
Since (xn ) → 2, there exists N ∈ N, such that for all n ≥ N, |xn - 2| < 3𝜖 / 2.
Consequently, for all n ≥ N the above inequality would be satisfied, and therefore
2xn - 1
|(2xn - 1) / 3 - 1| < 𝜖. Hence, → 1.
3

1 1
(b) We are interested to make the distance - as small as we please. Pick an
xn 2
arbitrary 𝜖 > 0.
We would like to show that:
1 1
- <𝜖
xn 2
|xn - 2|
<𝜖
2|xn |
We can strengthen the condition we wish to prove by finding an upper bound for the left
hand side of the above inequality.
Let's pick 𝜖 = 1. Since (xn ) → 2, there exists N1 ∈ N, such that for all n ≥ N1 , the
distance |xn - 2| < 1. In other words,

1 < xn < 3
Thus, |xn | > 1. Consequently,

|xn - 2| |xn - 2|
<
2|xn | 2
So, our claim is:
 |xn - 2|
<𝜖
2
Since (xn ) → 2, there exists N2 ∈ N, such that for all n ≥ N2 , the distance

|xn - 2| < 2𝜖

We pick N = max{ N1 , N2 }. To show that this choice of N works, let n ≥ N. We have:

1 1 |xn - 2| 1 1
- = < |xn - 2| < ⋅ 2𝜖 = 𝜖
xn 2 2|xn | 2 2

Exercise 2.3.3. (Squeeze Theorem.) Show that if xn ≤ yn ≤ zn for all n ∈ N, and if

lim xn = lim zn = l, then lim yn = l as well.

Proof.
Pick an arbitrary 𝜖 > 0.
Since (xn ) → l, there exists N1 ∈ N, such that for all n ≥ N1 :

l - 𝜖 < xn < l + 𝜖

Since (yn ) → l, there exists N2 ∈ N, such that for all n ≥ N2 :

l - 𝜖 < yn < l + 𝜖

Pick N = max{ N1 , N2 }. Since, xn ≤ yn ≤ zn , for all n ∈ N, we have that for all n ≥ N,

l - 𝜖 < xn ≤ yn ≤ zn < l + 𝜖
That is:

l - 𝜖 < yn < l + 𝜖
As 𝜖 was arbitrary to begin with, this holds true for all 𝜖. Consequently, (yn ) is a
convergent sequence and (yn ) → l.
Exercise 2.3.4. Let (an ) → 0 and use the Algebraic Limit Theorem to compute each of the
following limits (assuming the fractions are always defined):
1 + 2an
(a) lim
1 + 3an - 4an2

Proof.

1 + 2an lim 1 + 2an

lim =
1 + 3an - 4an2 lim 1 + 3an - 4an2
1+2⋅0
=
1+3⋅0-4⋅0
=1
( a n + 2) 2 - 4
(b) lim
an

Proof.

( a n + 2) 2 - 4 an2 + 4an + 4 - 4
lim = lim
an an
an2 + 4an
= lim
an
a n ( a n + 4)
= lim Since an ≠ 0
an
= lim (an + 4)
=4
 2
an
+3
(c) lim 1
.
an
+5

Proof.

2
an
+3 2 + 3an
lim = lim
1
+5 1 + 5an
an

lim 2 + 3an
=
lim 1 + 5an
2+3⋅0
=
1+5⋅0
2

Exercise 2.3.5. Let (xn ) and (yn ) be given and define (zn ) to be the shuffled sequence
(x1 , y1 , x2 , y2 , x3 , y3 , … , xn , yn , … ). Prove that (zn ) is convergent if and only if (xn ) and
(yn ) are both convergent with lim xn = lim yn .

Proof.
⟹ direction.

We are given that (zn ) is convergent. We are interested to prove that both (xn ) and (yn )
are convergent with lim xn = lim yn .
Assume that (zn ) → l. Pick an arbitrary 𝜖 > 0. There exists N ∈ N, such that for all
n ≥ N, we have:

l - 𝜖 < zn < l + 𝜖
If N = 2M (even), then for all m > M, both (xm ) and (ym ) fall in the interval (l - 𝜖, l + 𝜖).
If N = 2K + 1 (odd), then for all k ≥ K, both (xk ) and (yk ) fall in the interval (l - 𝜖, l + 𝜖).
In both cases, we are able to find a response to the given 𝜖 - challenge and the sequences
(xn ) and (yn ) eventually settle in (l - 𝜖, l + 𝜖). Consequently, lim xn = lim yn = l.

⟸ direction.

We are given that both (xn ) and (yn ) are convergent sequences with lim xn = lim yn and
we are interested to prove that the shuffled sequence (zn ) is also convergent.
Pick an arbitrary 𝜖 > 0.
Since (xn ) → l, there exists N1 ∈ N, such that for all n ≥ N1 , we have:

|x n - l| < 𝜖

Since (yn ) → l, there exists N2 ∈ N, such that for all n ≥ N2 , we have:

|y n - l| < 𝜖

Note that xn = z2n-1 and yn = z2n . Let N = max{ 2N1 - 1, 2N2 }. Then for all n ≥ N, it
follows that

l - 𝜖 < zn < l + 𝜖
Consequently, (zn ) → l.

Exercise 2.3.6. Consider the sequence given by bn = n - n 2 + 2n. Taking (1 / n) → 0 as

given and using both the Algebraic Limit Theorem and the result in exercise 2.3.1., show
that lim bn exists and find the value of the limit.
Proof.
We have:
 2 2 n+ n 2 + 2n
lim n - n + 2n = lim n - n + 2n ⋅
n+ n 2 + 2n
n 2 - n 2 + 2n
= lim
n+ n 2 + 2n
2n
= -lim
n+ n 2 + 2n
2
= -lim
2
1+ 1+
n
2
=- = -1
1+1

Exercise 2.3.7. Give an example of each of the following, or state that such a request is
impossible by referencing the proper theorem(s):
(a) Sequences (xn ) and (yn ), which both diverge, but whose sum (xn + yn ) converges.
Proof.
Consider (xn ) = n and (yn ) = - n. Both (xn ) and (yn ) are divergent sequences but
(xn + yn ) is constant zero sequence.

(b) sequences (xn ) and (yn ) where (xn ) converges, (yn ) diverges and (xn + yn ) converges.
Proof.
This request is impossible. We have:

yn = (x n + y n ) - x n
lim yn = lim [(xn + yn ) - xn ]
= lim (xn + yn ) - lim xn
Since both lim(xn + yn ) and lim xn are well-defined, the sequence (yn ) must be convergent.
(c) a convergent sequence (bn ) with bn ≠ 0 for all n such that (1 / bn ) diverges.
1 1
Let bn := . Then, = n is a divergent sequence.
n bn

(d) an unbounded sequence (an ) and a convergent sequence (bn ) with (an - bn ) bounded.
Proof.
This request is impossible.
If (an - bn ) is a bounded sequence, then there exists M1 ∈ N, such that for all n ∈ N,
|a n - b n | ≤ M 1 .

By theorem 2.3.3, convergent sequences are bounded. So, (bn ) is a bounded sequence.
Hence, there exists M2 ∈ N such that |bn | ≤ M2 for all n ∈ N.
We have:

|a n | = |( a n - b n ) + b n |
≤ |( a n - b n ) | + |b n | Triangle Inequality
≤ M1 + M2

Thus, (an ) must be a bounded sequence.

(e) two sequences (an ) and (bn ) where (an bn ) and (an ) converge but (bn ) does not.
Proof.
1 1
Consider (an ) = and (bn ) = n. The sequences (an bn ) = 1 and (an ) = converge but
n n
(bn ) = n does not.

Exercise 2.3.8. Let (xn ) → x and let p(x) be a polynomial.

(a) Show p(xn ) → p(x).
Proof.
Let p(x) be the polynomial :

p(x) = a0 + a1 x + a2 x 2 + a3 x 3 +…+ am x m

The image sequence p(xn ) is defined as :

p(xn ) = a0 + a1 (xn ) + a2 (xn ) 2 + a3 (xn ) 3 +…+ am (xn ) m

Taking limits on both sides, we have:

lim p(xn ) = lim a0 + a1 (xn ) + a2 (xn ) 2 + a3 (xn ) 3 +…+ am (xn ) m

= lim a0 + a1 lim xn + a2 lim xn ⋅ lim xn +… Algebraic Limit Theorem
= a0 + a1 x + a2 x 2 +…+ am x m
= p( x)

(b) Find an example of a function f(x) and a convergent sequence (xn ) → x where the
sequence f(xn ) converges, but not to f(x).
Consider the function

f ( x) =
x2 if x ≠ 0
1 if x = 0
and let (xn ) be the sequence

1
xn =
n
1
Then, (xn ) → 0, and f(xn ) = , so f(xn ) → 0, but f(0) = 1.
n2
Exercise 2.3.9. (a) Let (an ) be a bounded (not necessarily convergent) sequence, and
assume lim bn = 0. Show that lim(an bn ) = 0. Why are we not allowed to use the Algebraic
Limit Theorem to prove this?
Proof.
(a) We are interested to prove that the distance |an bn | can be made as small as we please.
Pick an arbitrary 𝜖 > 0. Since (an ) is a bounded sequence there exist M ∈ N for all n ∈ N
such that |an | ≤ M.
As (bn ) → 0, there exists N ∈ N, such that for all n ≥ N, we have:
𝜖
|b n | <
M
Together, for all n ≥ N, we have:
𝜖
|a n b n | < M ⋅ =𝜖
M
Thus, (an bn ) → 0.
The Algebraic Limit Theorem is applicable, if and only if, the sequences (an ) and (bn ) are
both convergent.
(b) Can we conclude anything about the convergence of (an bn ) if we assume that (bn )
converges to some nonzero limit b?
No, we cannot conclude anything about the convergence of (an bn ). Consider the sequence
1
(an ) = (-1) n , (bn ) = 1 + . Here, (an ) is a bounded sequence. (bn ) → 1. But, (an bn ) is
n
divergent.
(c) Use (a) to prove theorem 2.3.3. part (iii) for the case when a = 0.
Proof.
We are interested to prove that lim an bn = ab where a = 0.
We have that lim an = 0. As (bn ) is a convergent sequence, and convergent sequences are
bounded, so (bn ) is bounded. From (a) it follows that lim an bn = 0.
Exercise 2.3.10. Consider the following list of conjectures. Provide a short proof for those
that are true and a counterexample for any that are false.
(a) If lim(an - bn ) = 0, then lim an = lim bn .
1
This proposition is false. Consider (an ) = n and (bn ) = n - . Then, lim(an - bn ) = 0, but
n
both (an ) and (bn ) are divergent sequences.
(b) If (bn ) → b, then |bn | → |b|.
This proposition is true.
We are interested to make the distance ||bn | - |b|| as small as we please.
Pick an arbitrary 𝜖 > 0. We are interested to prove that:

||bn | - |b|| < 𝜖

We can strengthen the condition we wish to prove by replacing the left-hand side of the
inequality by its upper bound. We know that, ||bn | - |b|| ≤ |bn - b|. Thus, we shall prove
that:

| b n - b| < 𝜖
Since (bn ) → b, there exists N ∈ N such that for all n ≥ N, we have:

| b n - b| < 𝜖
Consequently, |bn | → |b|.
(c) If (an ) → a and (bn - an ) → 0, then (bn ) → a.
Proof.
This proposition is true.
By the Algebraic Limit Theorem,

lim bn = lim (an + (bn - an ))

= lim an + lim (bn - an )
= a+0
=a
Exercise 2.3.11. (Cesaro Means).
(a) Show that if (xn ) is a convergent sequence, then the sequence given by the averages:

x1 + x2 +…+ xn
yn =
n
also converges to the same limit.
Proof.
Let (xn ) → x. We are interested to make the distance |(x1 + x2 +…+ xn ) / n - x| as small
as we please. Pick an arbitrary 𝜖 > 0.
We are interested to prove that:

x1 + x2 +…+ xn
-x <𝜖
n
(x1 + x2 +…+ xn ) - nx
<𝜖
n
Consider the expression:
 (x1 + x2 +…+ xn ) - nx 1
= |(x1 - x) + (x2 - x) + (x3 - x) +…+ (xn - x)|
n n
1
≤ (|(x1 - x)| + |(x2 - x)| +…+ |xl - x| +…+ |xn - x|)
n
n
∑ l | x i - x| ∑ | x - x|
i=1 i=l+1 i
≤ +
n n

Let 𝜖 > 0. Since, (xn ) → x, we can pick l such that for all k > l, |xl - x| < 𝜖 / 2. Now that
l is fixed, pick N large enough so that for all n > N, we can make the first term:

∑ l | x i - x|
i=1 𝜖
<
n 2

It follows that for all n > N:

l n
∑ | x - x| ∑ | x i - x|
i=1 i i=l+1 𝜖 (n - l)(𝜖 / 2)
+ ≤ + <𝜖
n n 2 n

Exercise 2.4.1. (a) Prove that the sequence defined by x1 = 3 and

1
xn+1 =
4 - xn
converges.
Proof.
Claim. (xn ) is a bounded sequence.
We prove using induction that (xn ) is a bounded sequence.
(1) Base case : 0 < x1 ≤ 3
(2) We assume that 0 < xn ≤ 3.
(3) We are interested to prove that 0 < xn+1 ≤ 3.
Clearly,
1 1
xn+1 = ≤ =1<3
4 - xn 4 - 3
And,

1 1 1
xn+1 = ≥ = >0
4 - xn 4 - 0 4
By the principle of mathematical induction, 0 < xn ≤ 3 for all n ∈ N.
Claim. (xn ) is a monotonically decreasing sequence.
1
(1) Base case: x1 = 3 and x2 = . Thus, x1 ≥ x2 .
3

(2) We assume that xn ≥ xn+1 .

(3) We are interested to prove that xn+1 ≥ xn+2 .
We have:

1 1
xn+1 - xn+2 = -
4 - xn 4 - xn+1
xn - xn+1
=
(4 - xn )(4 - xn+1 )
1
> (xn - xn+1 ) { x n > 0 ∀n ∈ N }
16
>0
Consequently, by the principle of mathematical induction, xn - xn+1 ≥ 0 for all n ∈ N.
Thus, (xn ) is a monotonically decreasing and bounded sequence. By the Monotone
Convergence Theorem (MCT), (xn ) is a convergent sequence.
(b) Now that we know that lim xn exists, explain why lim xn+1 must also exist and equal
the same value.
Pick an arbitrary 𝜖 > 0. Since (xn ) converges to x, there exists N(𝜖) ∈ N, such that for all
n ≥ N, we have

x - 𝜖 < xn < x + 𝜖

Define N' = N + 1. If yn = xn+1 , then for all n ≥ N', we have:

x - 𝜖 < yn < x + 𝜖
So, (yn ) → x.
(c) Taking limits on both sides, we have:

1
x =
4-x
x( 4 - x) = 1
4x - x 2 = 1
x 2 - 4x = -1
x 2 - 4x + 1 = 0
2
( x - 2) 2 - 3 =0
x = 2± 3

Since, (xn ) ≤ 3, by the Order Limit Theorem, x < 3. So, x = 2 - 3.

Exercise 2.4.2. (a) Consider the recursively defined sequence y1 = 1,

yn+1 = 3 - yn

and set y = lim yn . Because (yn ) and (yn+1 ) have the same limit, taking the limit across
the recursive equation gives y = 3 - y. Solving for y we conclude that lim yn = 3 / 2. What
is wrong with this statement?
Proof.
This is incorrect, as we do not whether the sequence (yn ) is convergent. Looking at the first
few terms of the sequence:

(yn ) = (1, 2, 1, 2, 1, 2, … )

it is indeed a divergent sequence.

1
(b) This time set (y1 ) and (yn+1 ) = 3 - . Can the strategy in (a) be applied to compute
yn
the limit of this sequence?
Proof.
The first few terms of the sequence are :

5 13 34
( y n ) = 1, 2, , , , …
2 5 13
Claim. The sequence (yn ) is bounded. 1 ≤ yn < 3 for all n ∈ N.
Base Case: P(1) is true. 1 ≤ y1 < 3.
Hypotheses : Assume that P(n) is true. We assume that 1 ≤ yn < 3.
Inductive step : We are interested to prove that 1 ≤ yn+1 < 3.
We have:

1
yn+1 = 3 - ≥ 3-1 = 2
yn

Also,

1 1 8
yn+1 = 3 - < 3- =
yn 3 3
Consequently:
 1 ≤ yn+1 < 3

Thus, by the principle of mathematical induction 1 ≤ yn < 3 for all n ∈ N.

Claim. The sequence (yn ) is monotonically increasing. yn ≤ yn+1 for all n ∈ N.
Proof.
Base Case: y1 ≤ y2 .
Hypotheses : Assume that yn ≤ yn+1 , that is yn+1 - yn ≥ 0.
Inductive step: Consider

1 1
yn+2 - yn+1 = 3- - 3-
yn+1 yn
1 1
= -
yn yn+1
yn+1 - yn
=
yn yn+1
≥0 { y n > 0 ∀n ∈ N ,
yn+1 - yn ≥ 0 }
By the principle of mathematical induction (yn ) is a monotonically increasing sequence.
By the Monotone convergence theorem, (yn ) is convergent and lim yn exists. The strategy
in (a) can be applied to compute the limit of this sequence. Let y = lim yn . We have:

1
y = 3-
y
y 2 = 3y - 1
y 2 - 3y + 1 = 0
3± 9-4⋅1⋅1 3± 5
y = =
2⋅1 2
 3+ 5
Since (yn ) ≥ 1, by the order limit theorem y ≥ 1. So, y = .
2

Exercise. 2.4.3. (a) Show that :

2, 2+ 2, 2+ 2+ 2, …
converges and find the limit.
Proof.
This sequence is recursively defined as :
yn+1 = 2 + yn

(a) Claim. The sequence (yn ) is bounded. 0 < yn < 2.

Base Case. P(1) is true. 0 < y1 < 2.
Hypotheses. We assume that 0 < yn < 2.
Inductive Step. We are interested to prove that:

0 < yn+1 < 2

We have:

yn+1 = 2 + yn
> 2 { yn > 0 }
>0
Moreover,

yn+1 = 2 + yn
< 2+2
=2
Thus, by the priniciple of mathematical induction, 0 < yn < 2 for all n ∈ N.
Claim. The sequence (yn ) is monotonically increasing. yn ≤ yn+1 for all n ∈ N.
Base Case. y1 ≤ y2 .
Hypotheses. We assume that yn ≤ yn+1 .
Inductive step. We have:

yn ≤ yn+1
(2 + yn ) ≤ (2 + yn+1 )
(2 + yn ) 2 ≤ (2 + yn+1 ) 2
yn2+1 ≤ yn2+2
yn+1 ≤ yn+2 { As yn > 0 ∀n ∈ N,
we can take square roots on both sides. }
Consequently, the sequence (yn ) is a monotonically increasing sequence.
By the Monotone Convergence Theorem, (yn ) is a convergent sequence. Let lim yn = y.
Then,

y = 2+y
y2 = 2+y
y2 - y - 2 = 0
y 2 - 2y + y - 2 = 0
y( y - 2) + 1( y - 2) = 0
(y - 2)(y + 1) = 0
Since y > 0, y = 2.
(b) Does the sequence

2, 2 2, 2 2 2, …
converge? If so find the limit.
Solution.
This sequence can be recursively defined as:

yn+1 = 2yn

Let's prove that this sequence is bounded. Our claim is that 2 ≤ yn < 2 for all n ∈ N.

Base case. 2 ≤ y 1 < 2.

Hypotheses. We assume that, 2 ≤ y n < 2.

Inductive step. We are interested to prove that 2 ≤ yn+1 < 2.

We have:

yn+1 = 2yn

≥ 2 2
≥ 2⋅1 = 2
Also,

yn+1 ≤ 2⋅2
=2
Thus, by the principle of mathematical induction, the sequence (yn ) is bounded and

2 ≤ yn < 2
for all n ∈ N.
Let's prove that this sequence is monotonically increasing. Our claim is that yn ≤ yn+1 for
all n ∈ N.
Base Case. We have y1 ≤ y2 .
Hypotheses. We assume that yn ≤ yn+1 .
Inductive step. We are interested to prove that yn+1 ≤ yn+2 . We have:

yn+2 = 2yn+1
> yn+1 ⋅ yn+1
= yn+1
By the principle of mathematical induction, (yn ) is a monotonically increasing sequence.
Since, (yn ) is a bounded and monotonically increasing sequence, by the Monotone
Convergence Theorem, (yn ) is a convergent sequence. Let lim yn = y. Then we have:

y= 2y
y y- 2 = 0
∴y=2
Exercise 2.4.4. (a) In section 1.4, we used the Axiom of Completeness(AoC) to prove the
Archimedean property of R. Show that the Monotone Convergence Theorem can also be
used to prove the Archimedean property without making any use of the AoC.
Proof.
Archimedean Property of Real Numbers. Given any real number x ∈ R, there exists a
natural number n ∈ N, such that x < n.
Proof.
We proceed by contradiction. We are given a real number x ∈ R. Assume that for all
natural numbers n ∈ N, we have n ≤ x. Thus, N is a bounded set. Moreover, (xn ) = n is a
monotonically increasing sequence. Hence, by the Montone Convergence Theorem, (xn ) = n
is a convergent sequence. Let l = lim xn .
By the definition of convergence, for all 𝜖 > 0, there exists N ∈ N, such that for all n ≥ N,
|xn - l| < 𝜖. Pick 𝜖0 = 1. Then,

l-1 < N < l+1

But, N + 2 and all the successive natural numbers do not belong to this 𝜖-neighbourhood.
This is a contradiction.
Hence our initial assumption is false.
(b) Use the Monotone Convergence Theorem to supply a proof for the Nested Interval
Property that doesn't make use of AoC.
These two results suggest that we could have used the Monotone Convergence Theorem in
place of AoC as our starting axiom for building a proper theory of real numbers.
Proof.
1 2
Exercise 2.4.5. (Calculating Square roots). Let x1 = 2 and define xn+1 = xn + .
2 xn

(a) Show that xn2 is always greater than or equal to 2, and then use this to prove that
xn - xn+1 ≥ 0. Conclude that lim xn = 2.

(b) Modify the sequence (xn ) so that it converges to c.

Proof.
(a)
Let us prove that (xn ) is a bounded sequence. We claim that 1 ≤ xn ≤ 2. P(1) is true.
x1 = 2. We assume that P(n) is true. So, 1 ≤ xn ≤ 2. We have:

1 2
xn+1 = xn +
2 xn
1 2
≥ 1+ =1
2 2
And,
 1 2
xn+1 = xn +
2 xn
1 2
≤ 2+ =2
2 1
By the principle of mathematical induction, 1 ≤ xn ≤ 2 for all n ∈ N.
We are interested to prove that (xn ) is a monotonically decreasing sequence. Our claim is
xn - xn+1 ≥ 2 for all n ∈ N.

We are also interested to prove that xn2 ≥ 2. We will induct on n.

Base Case. x12 = 4. So, x12 ≥ 2
Hypotheses. Our claim is that xn2 ≥ 2.
Inductive step. We have:
2xn+1 xn = xn2 + 2
xn2 - 2xn xn+1 + 2 = 0

We know that xn is real and 1 ≤ xn ≤ 2. For this equation to have real roots, we must
have b 2 - 4ac ≥ 0 or 4xn2+1 - 4(2) ≥ 0. Consequently, xn2+1 ≥ 2.
By the principle of mathematical induction, xn2 ≥ 2 for all natural numbers n ∈ N.
1 2 1 2 3
Base Case. x2 = x1 + = 2+ = . So, x1 ≥ x2 .
2 x1 2 2 2
Hypotheses. We assume that xn - xn+1 ≥ 0.
Inductive step. We have:
 1 2
xn+1 - xn+2 = xn+1 - xn+1 +
2 xn+1
xn+1 1
= -
2 xn+1
xn2+1 - 2
=
2xn+1
≥0 xn2+1 ≥ 2, 1 ≤ xn ≤ 2

Thus, (xn ) is a bounded and monotonically decreasing sequence. Consequently, by the

Montone Convergence theorem, (xn ) is a convergent sequence. Let lim xn = x. Then,

1 2
x =x+
2 x
2
2x = x +
x
2
x =
x
2
x =2
x = 2 since 1 ≤ x ≤ 2
(b) Modify the sequence (xn ) so that it converges to c.

1 c
Let xn+1 = xn + . This sequence converges to c.
2 xn
Exercise 2.4.6. (Arithmetic-Geometric Mean).
(a) Explain why xy ≤ (x + y) / 2 for any two positive real numbers x and y. (The
geometric mean is always less than or equal to the arithmetic mean).
Proof.
We have:

( x - y) 2 ≥ 0
for any two real numbers x and y. Thus,
 (x + y) 2 - 4xy ≥0
2
x+y
≥ xy
2
x+y
≥ xy
2
(b) Now let 0 ≤ x1 ≤ y1 and define

xn + yn
xn+1 = xn yn  and  yn+1 =
2
Show that lim xn and lim yn both exist and are equal.
Proof.
We claim that 0 ≤ xn ≤ yn for all n ∈ N. P(1) is true. 0 ≤ x1 ≤ y1 . We assume that P(n)
is true. We are interested to prove that 0 ≤ xn+1 ≤ yn+1 .
We have that:

GM ≤ AM
Thus,

(x n + y n )
xn yn ≤
2
∴ xn+1 ≤ yn+1
Moreover, since xn and yn are non-negative, xn+1 ≥ 0 and so 0 ≤ xn+1 ≤ yn+1 . Thus, by
the principle of mathematical induction,

0 ≤ xn ≤ yn  ∀n ∈ N

Also, consider the sequence (yn ).

 xn + yn
yn+1 =
2
yn + yn
≤ = yn
2
Consequently, (yn ) is a monotonically decreasing and bounded below by 0. By the Montone
Convergence Theorem, lim yn exists.
Furthermore, if we consider the sequence (xn ):

xn+1 = xn yn
≥ xn ⋅ xn = xn
So, (xn ) is a monotonically increasing sequence and bounded above by y1 . By the Montone
Convergence Theorem, lim xn exists.
Consider the expression yn+1 = (xn + yn ) / 2. Taking limits on both sides, we have:

1
lim yn+1 = lim xn + lim yn
2
1
lim yn = lim xn + lim yn since lim yn = lim yn+1
2
1 1
lim yn = lim xn
2 2
lim yn = lim xn
Exercise 2.4.7. Let (an ) be a bounded sequence.
(a) Prove that the sequence defined by yn = sup{ ak : k ≥ n } converges.
Proof.
We are given that (an ) is a bounded sequence. Consider the set:

A = { an : n ∈ N }
By Axiom Of Completeness (AoC), A has an infimum and supremum. Thus, sup A and
inf A exist.
We propose that (yn ) is a decreasing sequence. We have

yn = sup { ak : k ≥ n }
= max an , sup { ak : k ≥ n + 1 }
≥ sup { ak : k ≥ n + 1 }
= yn+1
By the principle of mathematical induction, yn ≥ yn+1 for all n ∈ N.
Thus, (yn ) is a monotonically decreasing sequence.
Moreover, define (zn ) = inf{ ak : k ≥ n }. We know that,
inf{ ak : k ≥ n } ≥ inf{ ak : k ≥ 1 } = inf A. Thus, we can write:

sup A = y1 ≥ … ≥ yn ≥ … ≥ zn ≥ … ≥ z1 = inf A

Thus, the sequence (yn ) is bounded by inf A. By the Montone Convergence Theorem, (yn )
is a convergent sequence and nlim
→∞
yn exists.

(b) The limit infimum of (an ) or lim inf an is defined by:

lim inf an = lim zn

Since, (zn ) is a montonically increasing and bounded sequence, by the Monotone
Convergence Theorem, (zn ) is a convergent sequence.
(c) Prove that lim inf an ≤ lim sup an for every bounded sequence, and give and example
of a sequence for which the inequality is strict.
Proof.
We have that zn ≤ yn for all n ∈ N. Since both (yn ) and (zn ) are convergent sequences, by
the Order Limit Theorem,

lim zn ≤ lim yn
lim inf an ≤ lim sup an
Consider the sequence

1
( a n ) = ( - 1) n 1 +
n
We observe that:

lim sup an =1
lim inf an = -1
Thus, for this sequence, lim inf an < lim sup an .
(d) Show that lim inf an = lim sup an if and only lim an exists. In this case, all three
share the same value.
Proof.
⟹ direction.

We are given that lim inf an = lim sup an . We are interested to prove that lim an exists.
Let lim inf an = lim sup an = a. We know that:

inf{ ak : k ≥ n } ≤ an ≤ sup { ak : k ≥ n }
zn ≤ an ≤ yn
Since, lim zn = lim yn , by the Squeeze Theorem, lim an exists and lim an = l
⟸ direction.
( - 1) n
For the sake of an example consider the sequence (an ) = . By the Algebraic limit
n
theorem,
1
lim an = lim(-1) n ⋅ lim = lim(-1) n ⋅ 0 = 0.
n
 0.6

0.4

0.2

0
Y Axis

−0.2

−0.4

−0.6

−0.8

−1
X Axis

Pick an arbitrary 𝜖 > 0. Since (an ) → a, there exists N ∈ N, such that for all n > N:

an ∈ (a - 𝜖 / 2, a + 𝜖 / 2)
Since yN = supk≥N
{ ak : k ∈ N }, we must have yN ∈ [a - 𝜖 / 2, a + 𝜖 / 2] ⊆ (a - 𝜖, a + 𝜖). In
fact, for all n > N, yn ∈ (a - 𝜖, a + 𝜖).
Also, since zN = infk≥N { ak : k ∈ N }, we must have
zN ∈ [a - 𝜖 / 2, a + 𝜖 / 2] ⊆ (a - 𝜖, a + 𝜖). In fact, for all n > N, zn ∈ (a - 𝜖, a + 𝜖).

Consequently, lim yn = a and lim zn = a. And we have:

lim yn = lim zn = lim an = a

□

Exercise 2.4.8. For each series, find an explicit formula for the sequence of the partial sums
and determine if the series converges.
 ∞
1
(a) ∑ n
n=1 2

Proof.
Let (sn ) be the sequence of partial sums. We have:

k 1
sk = ∑ n=1
2n
n
1 1 - ( 1 / 2)
= ⋅
2 1 - ( 1 / 2)
1
= 1- n
2
The sequence (sn ) is monotonically increasing sequence and bounded above by 1.
Consequently, the infinite series is convergent.
∞
1
(b) ∑
n=1
n ( n + 1)

Proof.
Let (sn ) be the sequence of partial sums. We have:

∞ 1
sk = ∑ n=1
n ( n + 1)
1 1 1
= + +…+
1⋅2 2⋅3 k ( k + 1)
1 1 1 1 1 1
= - + - +…+ -
1 2 2 3 k k+1
1 1 1 1 1 1
= - + - +…+ -
1 2 2 3 k k+1
1
= 1-
k+1
This is monotonically increasing sequence bounded above by 1. Hence, the infinite series is
convergent.
∞
n+1
(c) ∑ log
n=1
n

Proof.
Again this can be written as a telescopic sum. Let (sn ) be the sequence of partial sums.
Then,

k n+1
sk = ∑ n=1 log
n
2 3 k+1
= log + log +…+ log
1 2 k
= log 2 - log 1 + log 3 - log 2 +…+ log k + 1 - log k
= log k + 1
However, (sk ) is unbounded and hence it is a divergent series.
Exercise 2.4.9. Complete the proof of the Cauchy condensation test theorem 2.4.6. by
∞ ∞
showing that if the series ∑ 2nb 2n diverges, then so does ∑ bn . Example 2.4.5. may be a
n=0 n=1
useful reference.
Proof.
( ⟸ direction)

We know that bn ≥ 0 and decreasing. Let (sn ) be the sequence of partial sums of the
∞
infinite series ∑ bn . We have:
n=0
 s2 k = b1 + b2 + b3 +…+ b2 k
= b1 + b2 + (b3 + b4 ) + (b5 + b6 + b7 + b8 ) +…+ (b2 k-1 +1 +…+ b2 k )
≥ b1 + b2 + (b4 + b4 ) + (b8 + b8 + b8 + b8 ) +…+ 2 k-1 b2 k
1
= b1 + 2b2 + 4b4 + 8b8 +…+ 2 k b2 k
2
b1 k
= + ∑ n=0 2 n b2 n
2
k ∞
Since ∑ 2 n
b2 n is unbounded, ∑ bn is unbounded and divergent.
n=0 n=0

Exercise 2.5.1. Give an example of each of the following, or argue that such a request is
impossible.
(a) A sequence that has a subsequence that is bounded but contains no subsequence that
converges.
Proof.
This request is impossible. Assume that a sequence (an ) contains a bounded subsequence
(an ). By the Bolzanno Weierstrass Theorem, every bounded sequence has atleast one
k

convergent subsequence. So, (an ) has atleast one subsequence (an ) that is convergent.
k ki

Since (an ) is a subsequence of the original sequence (an ), the given request is impossible.
ki

(b) A sequence that does not contain 0 or 1 as a term but contains subsequences
converging to each of these values.
Proof.
Consider the sequence:

1 1 1 2 1 3
(a n ) = , , , , , ,…
2 2 3 3 4 4
1 1
where the odd term is given by a2n-1 = and the even term is given by a2n = 1 - .
n n

(c) A sequence that contains subsequences converging to every point in the infinite set
{ 1, 1 / 2, 1 / 3, 1 / 4, 1 / 5, … }.
Proof.
Consider the sequence:

1 1 1 1 1 1
1, 1, , 1, , , 1, , , , …
2 2 3 2 3 4

(d) A sequence that contains subsequences converging to every point in the infinite set
{ 1, 1 / 2, 1 / 3, 1 / 4, 1 / 5, … } and no subsequences converging to points outside of this set.

Solution.
This request is impossible. Consider the sequence
1 1 1 1 1 1
1, 1, , 1, , , 1, , , , …
2 2 3 2 3 4
This contains a subsequence that converges to 0.
Exercise 2.5.2. Decide whether the following propositions are true or false, providing a
short justification for each conclusion.
(a) If every proper subsequence of (xn ) converges, then (xn ) converges as well.
Solution.
Consider the tail subsequence (xn )n∞=2 . Let x = lim{ xn }n∞=2 . Pick an arbitrary 𝜖 > 0. There
exists N ∈ N, such that for all n > N, xn ∈ (x - 𝜖, x + 𝜖). Consequently, (xn ) is a
convergent sequence.
(b) If (xn ) contains a divergent subsequence, then (xn ) diverges.
Solution.
This proposition is true. We proceed by contradiction.
We are given that (xn ) contains a divergent subsequence. Assume that (xn ) is a convergent
sequence.
By Theorem 2.5.2, every subsequence of (xn ) converges to the same limit as the original
sequence. This contradicts the fact that, (xn ) contains a divergent subsequence. Hence, our
initial assumption is false. (xn ) is a divergent sequence.
(c) If (xn ) is bounded and diverges, then there exist two subsequences of (xn ) that converge
to different limits.
Proof.
This proposition is true.
We are given that (xn ) is a bounded sequence and diverges. We proceed by contradiction.
Assume that all subsequences of (xn ) converge to the same limit l. Then the tail
subsequence { xn }n∞=2 converges to l. But this implies that (xn ) is a convergent sequence
and (xn ) → l. This contradicts the fact that (xn ) is a divergent sequence.
Hence, our initial assumption is false. There exist two subsequences of (xn ) that converge
to two different limits.
(d) If (xn ) is monotone and contains a convergent subsequence, then (xn ) converges.
Proof.
This proposition is true.
Let (xn ) be a subsequence of (xn ). Since (xn ) is convergent, it is bounded. Thus, ∃M > 0
k k

for all k ∈ N, such that |xn | ≤ M. Since, (xn ) is monotone, we must have:
k

xnk ≤ xnk +1 ≤ xnk +2 ≤ … ≤ xnk+1

or

xnk ≥ xnk +1 ≥ xnk +2 ≥ … ≥ xnk+1

Since, xn and xn belong to the closed interval [-M, M], we must have that all
k k+1

intermediate terms xn +1 , … , xn -1 of the sequence (xn ) also lie in the closed interval
k k+1

[-M, M]. So, this is true for all of (xn ). Consequently, (xn ) is a bounded sequence.

By the Montone Convergence Theorem, (xn ) is convergent.

Exercise 2.5.3. (a) Prove that if an infinite series converges, then the associative property
holds. Assume a1 + a2 + a3 + a4 + a5 +… converges to a limit L (that is the sequence of
partial sums (sn ) → L). Show that any regrouping of the terms

(a1 + a2 +…+ an1 ) + (an1 +1 +…+ an2 ) + (an2 +1 +…+ an3 ) +…

leads to a series that also converges to L.
Proof.
The sequence of partial sums (sn ) → L. By Theorem 2.5.2, every subsequence of a
convergent equence converges to the same limit as the original sequence. Therefore, the
subsequence

(s n 1 , s n 2 , s n 3 , … )
converges to L. That is the sequence:

(a1 +…+ an1 )

(a1 +…+ an1 ) + (an1 +1 +…+ an2 )
(a1 +…+ an1 ) + (an1 +1 +…+ an2 ) + (an2 +1 +…+ an3 )
⋮

Consequently, any regrouping of the terms of the infinite series leads to a series that also
converges to L.
(b) Compare this result to the example discussed at the end of section 2.1 where infinite
addition was shown not be associative. Why doesn't our proof in (a) apply to this example?
Proof.
Our proof in (a) applies only to an infinite series if it is convergent. The example at the end
of section 2.1 is a divergent series. Consider the sequence of partial sums (sn ):

(-1, 0, -1, 0, -1, 0, … )

which as we know is divergent.
Example 2.5.4. The Bolzano-Weierstrass Theorem is extremely important and so is the
strategy employed in the proof. To gain some more experience with this technique, assume
that the Nested Interval Property (NIP) is true and use it to provide a proof of the Axiom
of Completeness. To prevent the argument from being circular, assume also that
1 / 2 n → 0. (Why precisely is this last assumption needed to avoid circularity)?

Proof.
Let S be a subset of R bounded above by M. Our claim is that sup S exists.
x = sup S
I3
a1 b1 = M

I2

I1

If S is a finite subset of R, then sup S = max{ x : x ∈ S }.

Let's assume that S is an infinite set. Let I1 = [a1 , b1 ] be a closed interval such that
b1 = M, a1 < b1 with [a1 , b1 ] containing an infinite number of terms of S. We bisect I 1
into two intervals L1 = [a1 , (a1 + b1 ) / 2] and R1 = [(a1 + b1 ) / 2, b1 ].
We define I2 = L1 if R1 ∩ S = ∅ else I2 = R1 . In general,

I k+1 =
Lk if Rk ∩ S = ∅
Rk otherwise
Since
 I 1 ⊇ I 2 ⊇ I 3 ⊇ … ⊇ I k ⊇ I k+1 ⊇ …
∞
by the Nested Interval Property (NIP) there exists an element s ∈ ⋂ Ik .
k=1

Our claim is that s = sup S. Since (1 / 2) n → 0, for all 𝜖 > 0, there exists N such that for
l(I 1 )
all n > N, l(In ) = < 𝜖. Thus, lim an = lim bn = s.
2n

(1) s is an upper bound for S.

It is clear that all bn 's are an upper bound S. By the Order Limit Theorem, x ≤ s for all
x ∈ S.

(2) Ik ∩ S ≠ ∅ for all k ∈ N. Thus, for all 𝜖 > 0, ∃x ∈ Ik ∩ S such that s - 𝜖 < x.
Consequently, by lemma 1.3.8, s = sup S.
Exercise 2.5.5. Assume that (an ) is a bounded sequence with the property that every
convergent subsequence of (an ) converges to the same limit a ∈ R. Show that (an ) must
converge to a.
Proof.
We are given that (an ) is a bounded sequence with the property that every convergent
subsequence (an ) converges to the same limit a ∈ R. We proceed by contradiction.
Assume that (an ) is a divergent sequence.
From 2.5.2 (c), since (an ) is a bounded and divergent sequence, there exist atleast two
subsequences of (an ) that converge to different limits. This contradicts the fact that every
convergent subsequence of (an ) converges to the same limit a ∈ R.
Hence, our initial assumption is false. (an ) must be a convergent sequence.
Exercise 2.5.6. Use a similar strategy to the one in example 2.5.3. to show that b 1/n exists
for all b ≥ 0 and find the value of the limit.
Solution.
Assume that 0 ≤ b < 1. Then, since
 b ≤ b 1/2 ≤ b 1/3 ≤ b 1/4 ≤ …
(bn ) is a monotonically increasing sequence and bounded by 1, (bn ) is a convergent
1/2
sequence. Let lim bn = l. Since, lim b2n = lim b 1/n = l, we have that l = l. So,
l = 1.

Assume that b ≥ 1. Then, since

b ≥ b 1/2 ≥ b 1/3 ≥ b 1/4 ≥ …

b 1/n is a monotonically decreasing sequence bounded below by 1. Again, b 1/n → 1.

Exercise 2.5.7. Extend the result proved in example 2.5.3 to the case |b| < 1, that is show
that lim b n = 0 if and only if -1 < b < 1.
Proof.
( ⟹ ) direction.
Assume that lim b n = 0. Then, lim|b n | = 0. We proceed by contradiction. Assume that
|b| ≥ 1. Then:

1 ≤ | b| ≤ | b| 2 ≤ | b| 3 ≤ … ≤ | b| n ≤ …

Since | b| n is convergent, by the Order Limit Theorem:

1 ≤ lim|b| n

This contradicts the fact that lim|b| n = 0.

( ⟸ ) direction.
Assume that -1 < b < 0. Pick an arbitrary 𝜖 > 0. We pick N ∈ N such that,

| b| N < 𝜖
that is :
 N log|b| < 𝜖
𝜖
∴N >
log|b|
Then, for all n > N, we have:

b n ∈ (-𝜖, 𝜖)
Consequently, lim b n = 0.
If b = 0, then the constant sequence (0, 0, 0, … ) converges to 0.
Exercise 2.6.1. Suppy a proof for the Theorem 2.6.2.
Every convergent sequence is a Cauchy sequence.
Proof.
Let (xn ) be a convergent sequence. Assume that (xn ) → x. Pick an arbitrary 𝜖 > 0. There
exists N ∈ N, for all n ≥ N, the distance

| x n - x| < 𝜖 / 2

Now, consider the expression |xn - xm | where m, n ≥ N. We have:

| x n - x m | = | ( x n - x) - ( x m - x) |
≤ | x n - x| + | x m - x| Triangle Inequality
𝜖 𝜖
< + =𝜖
2 2
Thus, (xn ) is a Cauchy sequence.
Exercise 2.6.2. Give an example of each of the following or argue that such a request is
impossible.
(a) A Cauchy sequence that is not monotone.
Proof.
 - 1) n
Consider an = ( . This is a convergent sequence and hence Cauchy. Moreover, it is not
n
monotone.
(b) A Cauchy sequence with an unbounded subsequence.
Solution.
This request is impossible. By Lemma 2.6.3, Cauchy sequences are bounded. Thus, there
exists M > 0, for all n ∈ N, such that |xn | ≤ M. Let (xn ) be any arbitrary subsequence of
k

(xn ). Then it follows that, |xn | ≤ M for all k ∈ N.

k

Thus, all subsequences of a Cauchy sequence are bounded.

(c) A divergent montone sequence with a Cauchy subsequence.
Solution.
This request is impossible. A divergent monotone sequence cannot contain a Cauchy
subsequence. We proceed by contradiction.
Let (xn ) be a divergent monotone sequence. Assume that, there exists a Cauchy
subsequence (xn ) of (xn ).
k

We know, from theorem 2.6.4, that if a sequence is convergent ⟺ the sequence is

Cauchy. So, Cauchy sequences are convergent. Thus, if a sequence is divergent, it is not
Cauchy. Thus, (xn ) is not Cauchy.
Carefully negating the definition of a Cauchy sequence, we have that ∃𝜖0 > 0, for all
N ∈ N, such that for some n > m ≥ N, it follows that |xn - xm | ≥ 𝜖0 .

We know that (xn ) is Cauchy. Pick 𝜖 = 𝜖0 . There exists C ∈ N, such that for all
k

nk+1 > nk ≥ C, it follows that |xn - xn | < 𝜖0 .

k+1 k

Since (xn ) is monotone, it follows that all the intermediate terms

{ xn : n ∈ N, nk < n < nk+1 } lie between xn and xn on the real line. Consequently, the
k k+1

distance amongst them must be smaller than 𝜖0 . Thus, we conclude that, for all
n > m ≥ C, |x n - x m | < 𝜖 0 .
This is a contradiction. Our initial assumption is false.
(d) An unbounded sequence containing a subsequence that is Cauchy.
Consider the sequence formed by juxtaposing the terms of the sequence
(an ) = (0, 0, 0, 0, … ) which is Cauchy and (bn ) = (1, 2, 3, 4, 5, … ) which is unbounded.
The shuffle sequence (cn ) = (0, 1, 0, 2, 0, 3, … ) is unbounded and contains a Cauchy
subsequence.
Exercise 2.6.3. If (xn ) and (yn ) are Cauchy sequences, then one easy way to prove that
(xn + yn ) is Cauchy is to use the Cauchy criterion. By theorem 2.6.4, (xn ) and (yn ) must
be convergent, and the Algebraic Limit Theorem the implies that (xn + yn ) is convergent
and hence Cauchy.
(a) Give a direct argument that (xn + yn ) is a Cauchy sequence that does not use the
Cauchy criterion or the Algebraic Limit Theorem.
Proof.
Pick an arbitrary 𝜖 > 0.
Since (xn ) is a Cauchy sequence, there exists N1 > 0, such that for all n > m ≥ N1 , we
have |xn - xm | < 𝜖 / 2.
Since (yn ) is a Cauchy sequence, there exists N2 > 0, such that for all n > m ≥ N2 , we
have |yn - ym | < 𝜖 / 2.
Consider two arbitrary terms zm , zn of the sum sequence (zn ) = (xn + yn ), such that
m, n ≥ N = max{ N1 , N2 }. We have:

|z n - z m | = |x n + y n - ( x m + y m ) |
≤ |x n - x m | + |y n - y m | Triangle Inequality
𝜖 𝜖
< + =𝜖
2 2
Thus, by definition, (zn ) is a Cauchy sequence.
(b) Do the same for the product (xn yn ).
Proof.
Pick an arbitrary 𝜖 > 0.
Consider two arbitrary terms zm , zn of the product sequence (zn ) = (xn yn ). We have:

|z n - z m | = |x n y n - x m y m |
= |x n y n - x m y n + x m y n - x m y m |
≤ |x n y n - x m y n | + |x m y n - x m y m | Triangle Inequality
= |yn ||xn - xm | + |xm ||yn - ym |
Since (xn ) and (yn ) are Cauchy sequences and Cauchy sequences are bounded, it follows
that (xn ) and (yn ) are bounded. There exists M1 > 0, for all n ∈ N, such that |xn | ≤ M1 .
There exists M2 > 0, for all n ∈ N, such that |yn | ≤ M2 .
Since (xn ) is a Cauchy sequence, there exists N1 > 0, such that for all n > m ≥ N1 , we
have |xn - xm | < 𝜖 / 2M2 .
Since (yn ) is a Cauchy sequence, there exists N2 > 0, such that for all n > m ≥ N2 , we
have |yn - ym | < 𝜖 / 2M1 .
Thus, for all n > m ≥ N = max{ N1 , N2 }, we can write:

|zn - zm | ≤ |yn ||xn - xm | + |xm ||yn - ym |

𝜖 𝜖
< M2 ⋅ + M1 ⋅ =𝜖
2M2 2M1
Consequently, (zn ) is Cauchy.
Exercise 2.6.4. Let (an ) and (bn ) be Cauchy sequences. Decide whether each of the
following sequences is a Cauchy sequence justifying each conclusion.
(a) cn = |an - bn |.
Proof.
If (an ) and (bn ) are Cauchy sequences, by the Cauchy criterion, (an ) and (bn ) are
convergent. Applying the algebraic limit theorem, (an - bn ) is a convergent sequence.
Moreover, if (xn ) is a convergent sequence, then from Exercise 2.3.10, |xn | is a convergent
sequence. Thus, |an - bn | is a convergent sequence, and there Cauchy.
(b) cn = (-1) n an .
This conclusion is false. Consider the counterexample:
1
c n = ( - 1) n 1 + where an = 1 + 1 . (an ) is Cauchy, but (cn ) is not a Cauchy sequence.
n n

(c) cn = [[an ]] where [[x]] is the greatest integer less than or equal to x.
Proof.
( - 1) n
This proposition is false. Consider the sequence an = .
n

1 1 1 1
(an ) = -1, , - , , - , …
2 3 4 5
so (an ) → 0 whilst

(cn ) = (-1, 0, -1, 0, -1, 0, … )

so (cn ) is not a Cauchy sequence.
Exercise 2.6.5. Consider the following (invented) definition: A sequence (sn ) is pseudo-
Cauchy if for all 𝜖 > 0, there exists an N such that if n ≥ N, then |sn+1 - sn | < 𝜖.
Decide for which one of the following two propositions is actually true. Supply a proof for
the valid statement and a counterexample for the other.
(i) Pseudo-Cauchy sequences are bounded.
Solution.
This proposition is false.
n
1
Consider the sequence of partial sums sn = ∑ . This is a pseudo-cauchy sequence. The
k=1
n
difference between successive terms can be made as small as possible. However, the sum
keeps increasing ever-so slowly, and does not stop even if you took one million, one billion
or one trillion terms.

(ii) If (xn ) and (yn ) are pseudo-Cauchy, then (xn + yn ) is Pseudo-Cauchy as well.
This proposition is true.
Pick an arbitrary 𝜖 > 0.
Since (xn ) is pseudo-cauchy, ∃N1 such that for all n ≥ N1 , we have:
 |xn+1 - xn | < 𝜖 / 2

Since (yn ) is pseudo-cauchy, ∃N2 such that for all n ≥ N2 , we have:

|yn+1 - yn | < 𝜖 / 2
Let N = max{ N1 , N2 }.
Then, for all n ≥ N, we have:

|(xn+1 + yn+1 ) - (xn + yn )| = |(xn+1 - xn ) + (yn+1 - yn )|

≤ |(xn+1 - xn )| + |(yn+1 - yn )| Triangle Inequality
𝜖 𝜖
< + =𝜖
2 2
Thus, (xn + yn ) is pseudo-Cauchy.
Exercise 2.7.1. Proving the Alternating Series Test (Theorem 2.7.7) amounts to show that
the sequence of partial sums

s n = a1 - a2 + a3 -…± an
converges. Different characterizations of completeness lead to different proofs.
(a) Prove the Alternating Series Test by showing that (sn ) is a Cauchy sequence.
Proof.
Firstly, we have a1 ≥ an for all n ≥ 1. Taking limits on both sides, by the Order Limit
theorem, a1 ≥ lim an = 0. Similarly, we can conclude
(a) Consider the distance |sn - sm |. We are interested to make this distance as small as we
please.

|sn - sm | = |(-1) m am+1 +…+ (-1) n an |

(b) Consider the sequence of partial sums (s1 , s2 , s3 , s4 , … )
Let S = { s1 , s2 , s3 , … }.

s0 = 0 s2 s4 … … s5 s3 s1
The sequence of partial sums (s n )

Define I1 = [0, s1 ]. We bisect the interval I1 into two halves L1 and R1 . We let:

I2 =
R1 if R1 ∩ S contains an infinite number of points
L1 otherwise

In general, we let:

I k+1 =
Rk if Rk ∩ S contains an infinite number of points of S
Lk otherwise
∞
We have Ik ⊇ Ik+1 . By the Nested Interval Property, there exists s ∈ ⋂ Ik . Our claim is
k=1
that (sn ) → s. Pick an arbitrary 𝜖 > 0.
We can pick K such that:

s1
l(I K ) = <𝜖
2 K-1
Then, for all k > K, since s ∈ Ik , we must have that |sk - s| < 𝜖. Thus, (sk ) → s.
(c) We find that:
 s 2k+2 = s2k + (a2k+1 - a2k+2 )
≥ s2k since (a2k+1 - a2k+2 ) ≥ 0
And,
s 2k+3 = s2k+1 - a2k+2 + a2k+3
= s2k+1 - (a2k+2 - a2k+3 )
≤ s2k+1 since (a2k+2 - a2k+3 ) ≥ 0
Thus, the subsequence (s2n ) is a monotonically increasing sequence. And (s2n+1 ) is a
monotonically decreasing sequence. Moreover,
s 2n = s2n+1 - a2n+1
≤ s2n+1
And therefore we conclude:

0 ≤ s2 ≤ s4 ≤ s6 ≤ … ≤ s2n ≤ s2n+1 ≤ … ≤ s5 ≤ s3 ≤ s1

Since (s2n ) is monotonically increasing and bounded above by all (s2n+1 )'s, by the
Monotone Convergence Theorem (MCT), (s2n ) is a convergent sequence.
Since (s2n+1 ) is monotonically increasing and bounded below by all (s2n )'s, by the
Monotone Convergence Theorem (MCT), (s2n+1 ) is a convergent sequence.
The limits of (s2n ) and (s2n+1 ) could be different. We will prove that the limits are same.
Let s = lim s2n . Pick an arbitrary 𝜖 > 0. Consider the 𝜖 - neighbourhood (s - 𝜖, s + 𝜖).
There exists N1 ∈ N, such that for all n ≥ N1 , s2n ∈ (s - 𝜖, s + 𝜖).
There exists N2 ∈ N, such that for all n ≥ N2 , |a2n+1 | < 𝜖 / 2
Let N = max{ N1 , N2 }. Then, 2N + 1 ≥ 2N1 + 1 > 2N1 and 2N + 1 ≥ 2N2 + 1.
Consider the distance:
 |s2N+1 - s| = |s2N+1 - s2N + s2N - s|
≤ |s2N+1 - s2N | + |s2N - s|
= |a2N+1 | + |s2N - s|
𝜖 𝜖
< + =𝜖
2 2

So, s2N and s2N+1 belong to (s - 𝜖, s + 𝜖). Since,

s - 𝜖 < s 2N ≤ s 2N+2 ≤ … ≤ s 2N+3 ≤ s 2N+1 < s + 𝜖

we have that for all n ≥ N, sn ∈ (s - 𝜖, s + 𝜖). Consequently, (sn ) → s.

Exercise 2.7.2. Decide whether each of the following series converges or diverges:
∞
1
(a) ∑
n=1 2n + n

We have:
1 1
0≤ ≤
2n + n 2n
∞ ∞
Since the infinite series∑ 1n converges, by the comparision test, ∑ n
1
also converges.
n=1 2 n=1 2 + n

∞
(b) ∑ sin(2n) .
n=1 n

We have:

|sin n| 1
0≤ 2
≤
n n2
∞ ∞
1
Since ∑ p
converges if p > 1, by the comparision test, ∑ |sin2n| converges. By the
n=1 n n=1 n
 ∞
sin n
absolute convergence test, ∑ 2 also converges.
n=1 n

3 4 5 6 7
(c) 1 - + - + - +…
4 6 8 10 12

Solution.
∞
The general term of this infinite series is ak = ∑ (-1) k+1 k + 1 .
k=1
2k

∞ ∞
We know that, if ∑ an converges, an → 0. Consequently, if lim an ≠ 0, then ∑ an
n=1 n=1
n+1
diverges. This is called the nth term test. Clearly, both lim(-1) n and lim are non-
2n
zero, and hence by the nth term test, this series diverges.
1 1 1 1 1 1 1 1
(d) 1 + - + + - + + - +…
2 3 4 5 6 7 8 9

Solution.
We have:
 1 1 1 1 1 1 1 1
S = 1+ - + + - + + - +…
2 3 4 5 6 7 8 9
1 1 1 1 1 1 1 1
> 1 + - + + - + + - +…
3 3 4 6 6 7 9 9
1 1 1 1 1 1 1
= 1+ + + + + + + +…
4 7 9 12 15 18 21
1 1 1 1 1 1
> 1+ + + + + + +…
9 9 9 18 18 18
1 1 1 1
= 1+ + + + +…
3 6 9 12
1 1 1 1
= 1 + 1 + + + +…
3 2 3 4
Since the harmonic series is unbounded, S is unbounded and divergent.
1 1 1 1 1 1 1
(e) 1 - + - + - + - +…
22 3 42 5 62 7 82

Solution.
Let (an ) be the sequence of terms

1 1 1 1 1
1, , , , , ,…
22 3 42 5 62
Clearly, a1 ≥ a2 ≥ a3 ≥ a4 ≥ … . And (an ) → 0. Therefore by the alternating series test,
∞
∑ (-1) n+1 an is convergent.
n=1

Exercise 2.7.3. (a) Provide the details for the proof of the Comparision Test using the
Cauchy Criterion for Series.
(b) Give another proof for the Comparison test, this time using the Monotone Convergence
Theorem.
Proof.
(a) We have that 0 ≤ ak ≤ bk for all k ∈ N.
 ∞
(i) Suppose that ∑ bk is convergent.
k=0
∞
Since, ∑ bk converges, by the Cauchy Criterion, for all 𝜖 > 0, ∃N ∈ N, such that, for all
k=0
m, n ≥ N, we have:

|bm+1 +…+ bn | < 𝜖

Since, 0 ≤ ak ≤ bk , it follows that:

|am+1 +…+ an | ≤ |bm+1 +…+ bn | < 𝜖

∞
Consequently, by the Cauchy criterion, ∑ ak converges.
n=1
∞
(ii) Next, suppose that ∑ ak is divergent.
k=0

There exists 𝜖0 > 0, for all N ∈ N, such that for some m, n ≥ N, we have:

|am+1 +…+ an | ≥ 𝜖0

Since, |bm+1 +…+ bn | ≥ |am+1 +…+ an |, it follows that, for some m, n ≥ N:

|bm+1 +…+ bn | ≥ |am+1 +…+ an | ≥ 𝜖0

∞
Consequently, by the Cauchy criterion, ∑ bk diverges.
n=1

(b) Give another proof of the Comparison Test, this time using the Monotone Convergence
Theorem.
Proof.
(1) We have that 0 ≤ ak ≤ bk . Let (𝛼n ) be the sequence of partial sums of the infinite
∞ ∞
series ∑ ak and let (𝛽n ) be the sequence of partial sums of the infinite series ∑ bk .
k=0 k=0

We know that (𝛽n ) is convergent. Let 𝛽 = lim 𝛽n . We have:

𝛼n ≤ 𝛽n ≤ 𝛽

since bn ≥ 0 for all n ∈ N.

Thus, the sequence (𝛼n ) is monotonically increasing and bounded by 𝛽. By the Monotone
Convergence Theorem, the sequence of partial sums (𝛼n ) is convergent.
(2) Suppose that (𝛼n ) is divergent. Since, (𝛼n ) is monotone and divergent, it is unbounded.
Thus, (𝛽n ) is unbounded and therefore, (𝛽n ) is divergent.
Exercise 2.7.4. Give an example of each or explain why the request is impossible
referencing the proper theorem(s).

(a) Two series ∑ xn and ∑ yn that both diverge but where ∑ xn yn converges.
Solution.
1
Consider ∑ (-1) n+1 and ∑ . Both these series diverge, but the alternating harmonic
n
n+1
-1
series ∑ ( ) converges.
n

(b) A convergent series ∑ xn and a bounded sequence (yn ) such that ∑ xn yn diverges.
-1) n+1
Consider the infinite series ∑ ( and the sequence (yn ) = (-1) n+1 . The alternating
n
n+1
( - 1)
harmonic series ∑ is convergent and the sequence (yn ) = (-1) n+1 is bounded in
n
(-1) 2n+2 1
[-1, 1]. But, the infinite series ∑ =∑ diverges.
n n
(c) Two sequences (xn ) and (yn ) where ∑ xn and ∑ (xn + yn ) both converge but ∑ yn
diverges.
This request is impossible.

∑ y n = ∑ (x n + y n ) - x n
= ∑ (x n + y n ) - ∑ x n

By the Algebraic Limit Theorem for infinite series, if ∑ (xn + yn ) → A and ∑ xn → B,

then ∑ yn → A - B.

(d) A sequence (xn ) satisfying 0 ≤ xn ≤ 1 / n where ∑ (-1) n xn diverges.

Exercise 2.7.5. Now that we have proved the basic facts about the Geometric series,
supply a proof for the Corollary 2.4.7.
Proof.
∞
Corollary 2.4.7. The series ∑ 1 / n p converges if and only if p > 1.
n=1

Proof.
⟹ direction.
1
Let bn = .
np
∞ ∞
By the Cauchy condensation test, since ∑ bn converges, we have that ∑ 2 n b2 also
n

n=0 n=0
converges.
Thus,
 S = b1 + 2b2 + 2 2 b4 + 2 3 b8 +…
1 2 22 23 24
= + + + + +…
1p 2p 4p 8p 16 p
1 1 1
= 1+ + + +…
p-1 p-1 2 p-1 3
2 2 2
1
But the latter is a geometric series which converges if and only if |r| < 1. Since >0
2 p-1
1
for all p ∈ R, we must have 0 < p-1 < 1. Thus, 2 p-1 > 1 so p - 1 > 0 and therefore
2
p > 1.

⟸ direction.
This direction should be trivial.
Exercise 2.7.6. Let's say that a series subverges if the sequence partial sums contains a
subsequence that converges. Consider this (invented) definition for a moment, and then
decide which of the following statements are valid propositions about the subvergent series.

(a) If (an ) is bounded, then ∑ an subverges.

Proof.
This proposition is false.
Consider the constant sequence (an ) = (1, 1, 1, 1, … ). (an ) is a bounded sequence.
Consider the sequence of partial sums of ∑ an .

(1, 2, 3, 4, … )
No subsequence of of the partial sums is convergent.
(b) All convergent series are subvergent.
This proposition is true.
Let (sn ) be the sequence of partial sums of the infinite series ∑ an . Since (sn ) is
convergent, so is (sn )n∞=2 . Thus, (sn ) is subvergent.

(c) If ∑ |an | subverges, then ∑ an subverges as well.

Let (sn ) be the sequence of partial sums of the absolute value series ∑ |an | and let (tn ) be
the sequence of partial sums of the series ∑ an .
Since (sn ) subverges, there exists a subsequence (sn ) of (sn ) that converges.
k

Pick an arbitrary 𝜖 > 0. By the Cauchy criterion, there exists N > 0, such that for all
m > l ≥ N, we have:

|s n m - s n l | < 𝜖
Thus,

||anl +1 | +…+ |anm || < 𝜖

But,
||anl +1 | +…+ |anm || = |anl +1 | + … + |anm |

From the triangle inequality, we know that,

|anl +1 +…+ anm | ≤ |anl +1 | + … + |anm | < 𝜖

Thus, for m > l ≥ N,

|t n m - t n l | < 𝜖

Thus, (tn ) is Cauchy. Thus, ∑ an is subvergent.

k

(d) If ∑ an subverges, then (an ) has a convergent subsequence.

This proposition is false.
Consider the sequence

(an ) = (1, -1, 2, -2, 3, -3, 4, -4, … )

The sequence of partial sums (sn ) of the infinite series ∑ an is:

(sn ) = (1, 0, 2, 0, 3, 0, 4, 0, … )
The sequence of partial sums (sn ) has a convergent subsequence, (0, 0, 0, 0, … ). Hence,
the infinite series ∑ an is subvergent. But, (an ) has no convergent subsequence.

Exercise 2.7.7. (a) Show that if an > 0 and lim nan = l, with l ≠ 0, then the series ∑ an
diverges.
Firstly, since nan > 0, by the order limit theorem, l ≥ 0. We are given that l ≠ 0, so l > 0.
Pick 𝜖 = l / 2. Since nan → l, there exists N ∈ N, such that for all n ≥ N:

l l
l- < nan < l +
2 2
that is:

l 3l
0< < nan <
2 2

We multiply the above inequality throughout by 1 / n. Since n ≠ 0, and 1 / n > 0, we have:

l
0< < an
2n
∞ ∞
l
Now, ∑ diverges and therefore by the comparison test, ∑ an is divergent. Thus,
n=N
2n n=N
∞
∑ an is divergent.
n=1

(b) Assume that an > 0 and lim n 2 an exists. Show that ∑ an converges.
Let lim n 2 an = l. Pick 𝜖 = |l| / 2. There exists N ∈ N, such that for all n ≥ N, we have:

|l| 3|l|
0≤ ≤ n 2 an ≤
2 2
Multiplying throughout by 1 / n 2 :
|l| 3|l|
0≤ ≤ an ≤
2n 2 2n 2

∞ ∞
3|l|
Since ∑ 2
is convergent, by the comparison test, ∑ an is also convergent.
n=N 2n n=N
∞
Consequently, ∑ an converges.
n=1

Exercise 2.7.8. Consider each of the following propositions. Provide short proofs for those
that are true and counterexamples for any that are not.

(a) If ∑ (an ) converges absolutely, then ∑ an2 also converges absolutely.

Solution.
Pick an arbitrary 𝜖 > 0. Since ∑ an converges absolutely, there exists N ∈ N, such that,
for all n > m ≥ N:

|am+1 | +…+ |an | < 𝜖

Squaring both sides,

(|am+1 | +…+ |an |) 2 < 𝜖

We have:

|am+1 | 2 +…+ |an | 2 ≤ (|am+1 | +…+ |an |) 2

So, for all n > m ≥ N, it follows that:

 |am+1 | 2 +…+ |an | 2 < 𝜖

Consequently, ∑ an2 converges absolutely.

(b) If ∑ an converges and (bn ) converges, then ∑ an bn converges.

Solution.
This proposition is false.
-1) n+1 -1) n+1
As a counterexample, let an = ( and bn = ( . And let an ' = 1
. Since
n n n
a'1 ≥ a'2 ≥ … and (an ') → 0, by the alternating series test ∑ (-1) n+1 an ' is convergent.
1
Thus, ∑ an is convergent and (bn ) is a convergent sequence. But, ∑ an bn = ∑ is the
n
harmonic series and we know this is divergent.

(c) If ∑ an converges conditionally, then ∑ n 2 an diverges.

Solution.

Let's proceed by contradiction. We are given that ∑ an is convergent, but it is not

absolutely convergent. Assume that ∑ n 2 an converges.

Since ∑ n 2 an converges, by theorem 2.7.3, lim n 2 an → 0. Pick 𝜖 = 1. There exists

N ∈ N, such that, for all n ≥ N, we have:

-1 < n 2 a n < 1
that is,

0 ≤ n 2 |a n | < 1
which implies:
 1
0 ≤ |a n | <
n2
∞ ∞
1
By the comparison test, since ∑ is convergent, it follows that ∑ |an | is convergent,
n=N n2 n=N
∞
so adding a finite number N of terms to this, ∑ |an | should also be convergent. Thus,
n=1

∑ an is absolutely convergent. This is a contradiction.

Hence, our initial assumption is false. ∑ n 2 an diverges.

∞
Exercise 2.7.9. Given a series ∑ an with an ≠ 0, the ratio Test states that if (an ) satisfies:
n=1

an+1
lim =r<1
an
then the series converges absolutely.
(a) Let r' satisfy r < r' < 1. Explain why there exists an N such that n ≥ N implies that
|an+1 | ≤ |an |r'.

Proof.
Pick 𝜖 = (r' - r). There exists N ∈ N, such that for all n ≥ N, we have

an+1
r - (r' - r) < < r + (r' - r)
an
that is,

an+1
0≤ < r' < 1
an
or |an+1 | < |an |r'.
(b) Why does |aN |∑ (r') n converge?
Proof.
Since |r'| < 1, the geometric series |aN |∑ (r') n is convergent.

(c) Now, show that ∑ |an | converges, and conclude that ∑ an converges.
Solution.
∞
Consider the series ∑ |an |. Let (sn ) be the sequence of partial sums of this infinite series.
n=1
Since the terms of this series are non-negative, (sn ) is a monotonically increasing sequence.
Moreover, let n ≥ N, then we can write:

sn = |a1 | +…+ |aN | + |aN+1 | +…+ |an |

< |a1 | +…+ |aN | + |aN |r' + |aN |(r') 2 +…+ |aN |(r') n-1
∞
< |a1 | +…+ |aN |∑ n=0 (r') n
|a N |
= |a1 | +…+ |aN-1 | +
1 - r'
Thus, the sequence (sn ) has an upper bound that we found above. By the Monotone
convergence Theorem (MCT), (sn ) is a convergent sequence. So, ∑ |an | converges. By the
absolute convergence test,∑ an converges.

[Abbott 2.7.11] Find examples of two series ∑ an and ∑ bn both of which diverge but
for which ∑ min{ an , bn } converges. To make it more challenging, produce examples where
(an ) and (bn ) are strictly positive and decreasing.

Proof.
Let (an ) be the sequence (0, 1, 0, 1, 0, 1, … ) and (bn ) be the sequence
 ∞ ∞
1 1
1, 0, , 0, , 0, … . Both ∑ an and ∑ bn diverge. But, min{ an , bn } = 0 and thus,
2 3 n=1 n=1
∞
∑ min{ an , bn } converges.
n=1

This would also work, for example, with:

1
(an ) = 1, 1, ,
32
1 1
(bn ) = (1, 2 , ,
2 2

[Abbott 2.7.12] (Summation-by-parts). Let (xn ) and (yn ) be sequences, let

s n = x1 + x2 +… xn and set s 0 = 0. Use the observation that xj = s j - s j-1 to verify the
formula

n n
∑ xj yj = sn yn+1 - sm-1 ym + ∑ sj (yj - yj+1 )
j=m j=m

Proof.
We can simplify the expression on the RHS as follows:
 ∑ n sj (yj - yj+1 ) = sm ym - sm ym+1 + sm+1 ym+1
j=m

-s m+1 ym+2 +…+ s n yn - s n yn+1

= sm ym + ym+1 (sm+1 - sm ) + ym+2 (sm+2 - sm+1 )
+ …+ yn (sn - sn-1 ) - sn yn+1
= sm ym - sn yn+1 + ym+1 xm+1 + ym+2 xm+2 +…+ yn xn
n
= sm ym - sn yn+1 + ∑ j=m+1 xj yj
∑ n sj (yj - yj+1 ) - sm-1 ym = sm ym - sm-1 ym - sn yn+1 + ∑ n xy
j=m j=m+1 j j
n
= ym (sm - sm-1 ) - sn yn+1 + ∑ j=m+1 xj yj
n n
∑ s (y - yj+1 ) + s n yn+1 - s m-1 ym = xm ym + ∑ j=m+1 xj yj
j=m j j
n n
∑ s (y - yj+1 ) + s n yn+1 - s m-1 ym = ∑ j=m xj yj
j=m j j

∞
[Abbott 2.7.13] (Abel's Test). Abel's test for convergence states that if the series ∑ xk
k=1
converges and if (yk ) is a sequence satisfying:

y1 ≥ y2 ≥ y3 ≥ … ≥ 0
∞
then the series ∑ xk yk converges.
k=1
(a) Use exercise 2.7.12 to show that:
n n
∑ xk yk = sn yn+1 + ∑ sk (yk - yk+1 )
k=1 k=1

where sn = x1 +…+ xn .
Proof.
By the formula for summation by parts, we have:
 ∑ n xk yk = sn yn+1 - s0 y1 + ∑ n sk (yk - yk+1 )
k=1 k=1
n
= sn yn+1 - 0 ⋅ y1 + ∑ k=1 sk (yk - yk+1 )
n
= sn yn+1 + ∑ k=1 sk (yk - yk+1 )
∞
(b) Use the comparison test to argue that: ∑ sk (yk - yk+1 ) converges absolutely, and show
k=1
how this leads directly to a proof of Abel's test.
Proof.
∞
Since ∑ xk converges, (sk ) is a convergent sequence and hence it is bounded. There exists
k=1
M>0 for all k ∈ N, such that |sk | ≤ M. Thus,

|sk (yk - yk+1 )| ≤ M|yk - yk+1 | ≤ M(yk - yk+1 ) { ∵ yk ≥ yk+1 }

We know that (yk - yk+1 ) ≥ 0. Let (tk ) be the sequence of partial sums of the infinite series
∞
∑ (yk - yk+1 ). Since (tk ) is a monotonically-increasing sequence and
k=1

tk = (y1 - y2 ) + (y2 - y3 ) +…+ (yk - yk+1 )

= y1 - y2 + y2 - y3 + y3 +…+ yk - yk+1
= y1 - yk+1
≤ y1
it is bounded by y1 , by the Montone convergence theorem, tk is a convergent series. By the
∞
Algebraic limit theorem for infinite series, M ⋅ ∑ (yk - yk+1 ) is a convergent series.
k=1
∞
Hence, by the comparison test, ∑ |sk (yk - yk+1 )| is a convergent series.
k=1

∞
By the Absolute convergence test, ∑ sk (yk - yk+1 ) is a convergent series.
k=1
Passing to the limits, we have:

n n
lim n → ∞ ∑ k=1 xk yk = lim n → ∞ sn yn+1 + ∑ k=1 sk (yk - yk+1 )

Note that, (yn ) is bounded by [0, y1 ] and is a monotonically decreasing sequence. Hence,
by MCT, it is a convergent sequence. (sn ) is also given to be a convergent sequence. Hence,
the limit of the right hand side of the expression can be written as:
n n
lim s n yn+1 + ∑ s k (yk - yk+1 ) = lim sn yn+1 + lim ∑ sk (yk - yk+1 )
n→∞ n→∞ n→∞
k=1 k=1

Since both these limits exist, the limit on the right hand side exists. Thus, the product
∞
series ∑ xk yk converges. □
k=1

[Abbott 2.7.14] (Dirichlet's Test) Dirichlet's test for convergence states that, if the
∞
partial sums of ∑ xk are bounded (but not necessarily convergent), and if (yk ) is a
k=1
∞
sequence satisfying y1 ≥ y2 ≥ y3 ≥ … ≥ 0 with lim yk = 0, then the series ∑ xk yk
k=1
converges.
(a) Point out how the hypothesis of Dirichlet's Test differs from that of Abel's test in
Exercise 2.7.13, but show that essentially the same strategy can be used to provide a proof.
Proof.
n
Since the sequence of partial sums sn = ∑ xk is bounded, there exists M > 0, such that
k=1
|sn | ≤ M for all n ∈ N. We can use the same strategy as in part (a) of exercise 2.7.13 to
show that the infinite product series is convergent.
(b) Show how the Alternating Series Test (Theorem 2.7.7) can be derived as a special case
of the Dirichlet's test.
Proof.
Let xk = (-1) k+1 and yk = ak , such that a1 ≥ a2 ≥ a3 ≥ … ≥ 0 and lim ak = 0. Since
∞
the partial sums (sk ) of the infinite series ∑ (-1) k+1 are bounded, by the Dirichlet test,
k=1
∞
∑ (-1) k+1 ak is convergent.
k=1

Exercise 3.2.1. (a) Where in the proof of theorem 3.2.3 part (ii) does the assumption that
the collection of open sets be finite get used?
Theorem 3.2.3 part (ii) The intersection of a finite collection { Oi : 1 ≤ i ≤ N, N ∈ N } of
open sets is open.
Proof. This assumption is used to find a candidate 𝜖-neighbourhood for the point
N
x∈ ⋂ Oi . We chose 𝜖 = min{ 𝜖1 , … , 𝜖N }. It would not be possible to choose such a
i=1
candidate should the collection of open sets be countably infinite or uncountable. Consider
1
the case we have a countable collection of open sets O1 , O2 , … and where 𝜖i = . Then,
n
𝜖i > 0, and inf{ 𝜖i : i ∈ N } = 0. So, we would be unable to choose an 𝜖.

(b) Give an example of a countable collection of open sets { O1 , O2 , O3 , … } whose

∞
intersection ⋂ On is closed, not empty and not all of R.
n=1

Proof.
∞
Consider On = - 1 , 1 . We have: ⋂ On = { 0 } which is closed, not empty and not all of
n n n=1
R.

Exercise 3.2.2. Let

 2
A = ( - 1) n + : n = 1, 2, 3, …  and B = { x ∈ Q : 0 < x < 1 }
n
Answer the following questions for each set.
(a) What are the limit points?
(i) Enumerating the first few points of A, we have:

2 2 2 2
1, 2, -1 + , 1 + , - 1 + , 1 + , …
3 4 5 6
L = { -1, 1 } are the limit points of A.
(ii) [0, 1] are the set of all limit points of B. To see this, let y ∈ [0, 1] be an arbitrary point.
Pick an arbitrary 𝜖 > 0 and consider the punctured neighbourhood (y - 𝜖, y + 𝜖) - { y }.
Since Q is dense in R, there exists a rational number x ∈ Q, such that y < x < y + 𝜖.
Thus, (V𝜖 (y) - { y }) ∩ B ≠ ∅. Consequently y is a limit point for B. Since y was arbitrary,
[0, 1] is the set of all limit points of B.

(b) Is the set open? Closed?

(i) A is not open. To see this, take the point x = 1. This is a boundary point of A.
Consider the 𝜖-neighbourhood of 1: V𝜖 (1) = (1 - 𝜖, 1 + 𝜖). V𝜖 (1) ⊈ A, for all 𝜖 > 0.
Since L ⊆ A, A is not closed.
(ii) B is not open. Let x be an arbitrary point of B. Consider the 𝜖 - neighbourhood of x :
(x - 𝜖, x + 𝜖). Since Q is dense in R, we know that there exists a rational number r ∈ Q
such that:

x-𝜖- 2 < r < x+𝜖- 2

So:

x-𝜖 < r+ 2 < x+𝜖

So, there always exists an irrational number in V𝜖 (x). Consequently, for all 𝜖 > 0.
V 𝜖 ( x) ⊈ B .

0 is a limit point of B, that doesn't belong to B. So, B is not closed.

(c) Does the set contain any isolated points?

(i) The set A contains an infinite number of isolated points. For n = 2m, pick
1 1 2 1
𝜖= - = = , then A2 , A4 , A6 , A8 , … are isolated
2m 2m + 2 2m(2m + 2) 2m(m + 1)
points.
(ii) All the points in B are limit points. Hence, B has no isolated points.
(d)
(i) cl(A) = A ∪ { -1 }
(ii) cl(B) = [0, 1]
Exercise 3.2.3. Decide whether the following sets are open, closed or neither. If a set is not
open, find a point in the set for which there is no 𝜖-neighbourhood contained in the set. If
a set is not closed, find a limit point that is not contained in the set.
(a) Q.
Q is neither open nor closed.
First, let x ∈ Q, and consider any 𝜖-neighbourhood, V𝜖 (x) of the point x. There exists an
irrational number between any two real numbers, so there exists y ∈ I, such that
x - 𝜖 < y < x + 𝜖. Consequently, ∀𝜖 > 0, ∃x, such that V𝜖 (x) ⊈ Q.

1 2
Next, consider the point 2 ∈ I and the sequence xn+1 = xn + with x0 = 1. The
2 xn
sequence (xn ) ⊆ Q, and (xn ) → 2. Thus, 2 is a limit point for Q, that does not belong
to Q.
(b) N.
Let n ∈ N. Let 𝜖 = 1. Then, V1 (n) = (n - 1, n + 1) ⊈ N. Consequently, N is not open.
N has no limit points. Therefore, N is closed.
(c) { x ∈ R : x ≠ 0 }
R - { 0 } is open. Let y ∈ R - { 0 }. If y > 0, then pick 𝜖 = y / 2. (y / 2, 3y / 2) ⊆ R - { 0 }. If
y < 0, pick 𝜖 = |y| / 2. Then, (y - |y| / 2, y + |y| / 2) ⊆ R - { 0 }.

R - {0} is not closed. 0 is a limit point for R - { 0 }, because for all 𝜖 > 0, (-𝜖, 𝜖) intersects
the set in atleast one point other than 0. And 0 does not belong to the set.

(d) 1 + 1 / 4 + 1 / 9 +…+ 1 / n 2 : n ∈ N

This set is not open. Let (sn ) be the sequence of partial sums of the infinite series ∑ 12 .
n
Then, the set consists of S = { s1 , s2 , s3 , … , sn , … }. Since, an > 0, sn > 0 and (sn ) is
monotonically increasing. Pick 𝜖 = min{ sn - sn-1 , sn+1 - sn }. Then, V𝜖 (sn ) ⊈ S.
1 1
This set is not closed. We know that, ∑ is convergent, if and only if p > 1. So, ∑
n p
n2
1 𝜋2 𝜋2
is convergent and in fact, ∑ 2 = . Thus, the sequence (sn ) → . So, 𝜋 2 / 6 is a
n 6 6
limit point for S and does not belong to S.
(e) { 1 + 1 / 2 + 1 / 3 +…+ 1 / n : n ∈ N }.
This set is not open. Again let (sn ) be the sequence of partial sums of the infinite series
∑ 1 . Then, the set S = { s1 , s2 , … , sn }. The rest of the argument is similar to part(d).
n

1
S is closed. The harmonic series ∑ is divergent. Thus, S only has isolated points and no
n
limit points.
Exercise 3.2.4. Let A be nonempty and bounded above so that s = sup A exists.
(a) Show that s ∈ cl(A).
Proof.
1
By definition, ∀𝜖 > 0, ∃a ∈ A, such that s - 𝜖 < a ≤ s. Take 𝜖 = for all n ∈ N. Then,
n
1 1
we can construct a sequence (an ) ⊆ A, such that s - < an ≤ s. If we pick N > , then
n 𝜖
for all n ≥ N, s - 𝜖 < an < s + 𝜖. So, (an ) → s. Thus, s is limit point for A. Consequently,
s ∈ cl(A).

(b) Can an open set contain its supremum?

No, an open set cannot contain its supremum. Let O be an open set and let s = sup O.
Then, s is a limit point of O. We proceed by contradiction. Assume that s ∈ O. There
exists V𝜖 (s) such that V𝜖 (s) ⊆ O, so V𝜖 (s) ∩ O C = ∅.
0 0 0

Pick an arbitrary 𝜖 > 0. And let u be any point such that s - 𝜖 < s < u < s + 𝜖. Since,
x ≤ s for all x ∈ O, we must have that u ∉ O, that is u ∈ O C . Consequently, ∀𝜖 > 0,
V 𝜖 ( s ) - { s } ∩ O C ≠ ∅.

This is a contradiction. Hence, our initial assumption is false. s ∉ O.

Exercise 3.2.5. Prove Theorem 3.2.8.
Theorem 3.2.8. A set F ⊆ R is closed, if and only if every Cauchy sequence contained in F
has a limit that is also an element of F.
Proof.
⟹ direction.

We are given that F is closed. We proceed by contradiction.

Assume that, there exists a Cauchy sequence (xn ) contained in F, such that (xn ) → x, with
x ∉ F. By Theorem 3.2.5, since (xn ) ⊆ F and xn ≠ x, x is a limit point of F.

But this is a contradiction. Since F is closed, x must belong to F. Consequently, our initial
assumption is false.
∀Cauchy sequences contained in F, their limit is also an element of F.

⟸ direction.

We are given that, every Cauchy sequence contained in F has a limit that is also an element
of F.
We proceed by contradiction. Assume that F is not closed. Then, there exists a limit point
x of F, such that x ∉ F. Since, x is a limit point of F, there exists a sequence (xn ) ⊆ F,
such that xn ≠ x, and (xn ) → x. This is a contradiction.
Hence our initial assumption is false.
Exercise 3.2.6. Decide whether the following statements are true or false. Provide
counterexamples for those that are false, and supply proofs for those that are true.
(a) An open set that contains every rational number must necessarily be all of R.
Proof.
This proposition is true. Let O be any open set containing R. We proceed by contradiction.
Assume that there exists x ∈ R, such that x ∉ O.
(b) The Nested Interval Property remains true if the term closed interval is replaced by
closed set.
Proof.
This proposition is false.
Counterexample.
Consider the set defined by Ik = { n ≥ k : n ∈ N } for k = 1, 2, 3, … . I1 , I2 , … are nested
∞
closed sets. However, ⋂ In = ∅.
n=1

(c) Every non-empty open set contains a rational number.

Proof.
This proposition is true.
Let O be a non-empty open set. Let x ∈ O. Then, since x is an interior point, there exists
𝜖0 > 0, such that V𝜖 (x) ⊆ O. Since, Q is dense in R, there exists a rational number
0

r ∈ Q, such that x - 𝜖 < r < x + 𝜖. Hence, r ∈ O.

(d) Every bounded infinite closed set contains a rational number.

Proof.
This proposition is false.
Counterexample.
Consider the set

1
S= 2+ :n ∈ N ∪ 2
n

S is a bounded set. 2 ≤ x ≤ 2 + 1 for all x ∈ S. S is infinite and contains all its limit
points, so S is closed. But, S does not have a rational number.
(e) The Cantor set is closed.
Proof.
Since

1 2 1 2 7 8
C = [0, 1] - , ∪ , ∪ , ∪…
3 3 9 9 9 9
it is the complementation of the set
1 2 1 2 7 8
, ∪ , ∪ , ∪…
3 3 9 9 9 9
(which is an open set) with respect to [0, 1]. Thus, C is a closed set.
Exercise 3.2.7. Given A ⊆ R, let L be the set of all limit points of A.
(a) Show that the set L is closed.
Proof.
Let x be an arbitrary limit point of L.
Pick an arbitrary 𝜖 > 0. Since x is a limit point of L, (x - 𝜖, x + 𝜖) intersects L in some
point l other than x. Thus, ∃l ∈ L, such that x - 𝜖 < l < x + 𝜖.

l - ( x - 𝜖) ( x + 𝜖) - l | l - x|
Since, l ∈ L, l is a limit point for A. Pick 𝜉 = min , , . Then,
2 2 2
V𝜉 (l) = (l - 𝜉, l + 𝜉) intersects A for in some point other than l and x.
Thus, V𝜖 (x) intersects A in some point other than x. Since 𝜖 was arbitrary, this is true for
all 𝜖 > 0. Consequently, x is a limit point for A. Thus, x belongs to L. Since x was an
arbitrary limit point of L and belongs to L, L is closed.
(b) Argue that if x is a limit point of A ∪ L, then x is a limit point of A. Use this
observation to furnish a proof for theorem 3.2.12.
Proof.
Let x be an arbitrary limit point of A ∪ L.
Pick an arbitrary 𝜖 > 0. There exists y ∈ A ∪ L, y ≠ x, such that y ∈ V𝜖(x) ∩ (A ∪ L).
This implies that either y ∈ V𝜖(x) ∩ A or y ∈ V𝜖(x) ∩ L or both.
If y ∈ V𝜖 (x) ∩ A, then since 𝜖 was arbitrary, this must be true for all 𝜖 > 0. Thus, x is a
limit point of A.
If y ∈ V𝜖 (x) ∩ L, then y is a limit point of A. We can construct a tiny 𝜉-neighbourhood
inside V𝜖 (x), determined by the rule:
 y - ( x - 𝜖) ( x + 𝜖) - y | y - x|
𝜉 = min , ,
2 2 2
Since y is a limit point of A, ∃z, such that z ≠ x and z ≠ y with z ∈ V𝜉 (y) ∩ A.
But this implies that, ∃z, with z ≠ x, such that z ∈ V𝜖 (x) ∩ A. Since 𝜖 was arbitrary, this
is true for 𝜖 > 0. So, x is a limit point of A.
In both cases, x is a limit point of A.
If x is a limit point of A, x belongs to L and thus x ∈ A ∪ L. Since, x was arbitrary, A ∪ L
is closed.
[Abbott 3.2.8] Assume that A is an open set B is a closed set. Determine if the following
sets are definitely open, definitely closed, both or neither.
(a) cl(A ∪ B)
We know, that the closure of any set S is closed, so cl(A ∪ B) is definitely closed.
(b) A ⧵ B = { x ∈ A : x ∉ B }
We have A ⧵ B = A ∩ B C . B C is an open set A ∩ B C is definitely open, since the
intersection of a finite collection of open sets is open.
C
(c) AC ∪ B .
C
AC ∪ B is closed, since the finite union of closed sets is closed. Thus, AC ∪ B is
definitely open.
(d) (A ∩ B) ∪ AC ∩ B .
We can simplify this as,

A ∪ AC ∩ B = B
Thus, it is a closed set.
(e) cl(A) C ∩ cl AC

AC is closed, so cl AC = AC and so,

cl(A) C ∩ A C = (cl(A) ∪ A) C = (cl(A)) C

Thus, it is definitely open.
[Abbott 1.2.5] (De-Morgan's Laws) Let A and B be subsets of R.
(a) If x ∈ (A ∩ B) c , explain why x ∈ A C ∪ B C . This shows that (A ∩ B) C ⊆ A C ∪ B C .
Proof.
Let x ∈ (A ∩ B) c .
Intuitively, x does not belong to both A and B. This is possible if and only if, x does not
belong to atleast one of A or B. So, either x ∉ A or x ∉ B. So, x ∈ A C or x ∈ B C . Thus,
x ∈ AC ∪ BC .

Formally:
x ∈ ( A ∩ B) C
⟺ ⌐((x ∈ A) ∧ (x ∈ B))
⟺ ⌐ ( x ∈ A ) ∨ ⌐ ( x ∈ B)
⟺ x ∈ AC ∨ x ∈ BC
⟺ x ∈ AC ∪ BC

(b) Prove the reverse inclusion (A ∩ B) c ⊇ A C ∪ B C and conclude that

( A ∩ B) C = A C ∪ B C .

Let x ∈ AC ∪ BC .
Intuitively, x belongs atleast one of A C , B C . Thus, x does not belong to atleast one of A,
B. Naturally, x cannot belong to both A and B. So, x ∉ (A ∩ B). Thus, x ∈ (A ∩ B) C .
Formally:
x ∈ AC ∪ BC
⟺ x ∈ AC ∨ x ∈ BC

⟺ ⌐ ⌐ x ∈ AC ∧ ⌐ x ∈ BC
⟺ ⌐((x ∈ A) ∧ (x ∈ B)
⟺ ⌐(x ∈ (A ∩ B))
⟺ x ∈ ( A ∩ B) C

(c) Show that (A ∪ B) C = A C ∩ B C by demonstrating inclusion both ways.

Let x ∈ (A ∪ B) C .
Intuitively, x is an element of (A ∪ B) C . So, x belongs neither to A nor to B. So, x must
simultaneously be an element of both A C and B C . Consequently, x ∈ A C ∩ B C .
Formally:
x ∈ ( A ∪ B) C
⟺ ⌐(x ∈ (A ∪ B))
⟺ ⌐((x ∈ A) ∨ (x ∈ B))
⟺ ⌐ ( x ∈ A ) ∨ ⌐ ( x ∈ B)
⟺ x ∈ AC ∨ x ∈ BC
⟺ x ∈ AC ∪ BC
Let x ∈ AC ∩ BC .

Intuitively, x belongs to both A C , B C . So, x does not belong to both A and B. Thus,
x ∈ ( A ∩ B) C .
 x ∈ AC ∩ BC
⟺ x ∈ AC ∧ x ∈ BC
⟺ ⌐(⌐ x ∈ A C ∨ ⌐ x ∈ B C
⟺ ⌐((x ∈ A) ∨ (x ∈ B))
⟺ ⌐(x ∈ (A ∪ B))
⟺ x ∈ ( A ∪ B) C

[Abbott 3.2.9] A proof for De Morgan's Laws in the case of two sets is outlined in the
exercise 1.2.5. The general argument is similar.
(a) Given a collection of sets { E𝜆 : 𝜆 ∈ 𝛬 }, show that :
c C
⋃ E𝜆 = ⋂ E𝜆C  and  ⋂ E𝜆 = ⋃ E𝜆C
𝜆∈𝛬 𝜆∈𝛬 𝜆∈𝛬 𝜆∈𝛬

Proof.
This proof is similar to the above exercise. So, we do not repeat it.
(b) Now, provide the details for the proof of the theorem 3.2.14.
(i) The union of a finite collection of closed sets is closed.
Let O1 , O2 , … , ON be a finite collection of open sets. Let
N
O= ⋂ Oi
i=1

then from theorem 3.2.3, the intersection of a finite collection of open sets is open. Taking
complementation on both sides,
 N
O = C ⋃ Oic
i=1

we must have that the union of a finite collection of closed sets is closed.
(ii) The intersection of an arbitrary collection of closed sets is closed.
Let { O𝜆 : 𝜆 ∈ 𝛬 } be an arbitrary collection of open sets. From theorem 3.2.3, we know that:

O= ⋃ O𝜆
𝜆∈𝛬

the union of an abitrary collection of open sets is open. Taking complementation on both
sides, we have that:

OC = ⋂ O𝜆C
𝜆∈𝛬

Since O is closed and each of the are closed, we must have that the intersection of an
C
O𝜆C
arbitrary collection of closed sets is closed.
□
[Abbott 3.2.10] Only one of the following three descriptions can be realized. Provide an
example that illustrates the viable description and explain why the other two cannot exist.
Proof.
(i) A countable set contained in [0, 1] with no limit points.
This is not viable.
Let S = { xn : n ∈ N } be a countable set. Thus, (xn ) is an arbitrary sequence in [0, 1]. By
the Bolzanno Weierstrass theorem, every bounded sequence has atleast one convergent
subsequence. So, there exists (xn ) ⊆ (xn ) such that (xn ) → x. So, x is a limit point of S.
k k

(ii) A countable set contained in [0, 1] with no isolated points.

Consider Q ∩ [0, 1]. Let x be an arbitrary rational number in (0, 1). ∀𝜖 > 0, V𝜖 (x)
intersects Q in atleast one point other than x. So, all points in this set are limit points.
Therefore, Q ∩ [0, 1] has no isolated points.
(iii) A set with an uncountable number of isolated points.
This is not viable.
Let S be such a set, that is uncountable, there are no holes and the set is complete under
the limiting operation. For all x ∈ S, there exists a cauchy sequence (xn ) ⊆ S, such that
( x n ) → x.

Thus, every point of S is a limit point. S cannot have isolated points.

[Abbott 3.2.11] (a) Prove that cl(A ∪ B) = cl(A) ∪ cl(B).
Proof.
( ⟹ direction).
Let x ∈ cl(A ∪ B). Let L be the set of limit points of A ∪ B. Since,
cl(A ∪ B) = (A ∪ B) ∪ L, then atleast one of x ∈ (A ∪ B) or x ∈ L holds true.

If x ∈ (A ∪ B), then x ∈ cl(A) ∪ cl(B).

Suppose that x ∈ L. That is x is the limit point of A ∪ B. By definition, there exists a
sequence (xn ) ⊆ (A ∪ B) such that xn ≠ x and (xn ) → x. There exists a subset
(xn ) ⊆ (xn ), such that atleast one of (xn ) ⊆ A or (xn ) ⊆ B holds. Every subsequence of a
k k k

convergent sequence, converges to the same limit as the original sequence. So, (xn ) → x.
k

Thus, x is either a limit point of A or x is a limit point of B. Consequently,

x ∈ cl(A) ∪ cl(B).

( ⟸ direction).
Let x ∈ cl(A) ∪ cl(B).
x belongs to atleast one of cl(A) or cl(B). Suppose x ∈ cl(A). Let cl(A) = A ∪ LA .
If x belongs to A, then x ∈ (A ∪ B) and so x ∈ cl(A ∪ B) and we are done.
If x belongs to LA , then x is a limit point of A. So, there exists a sequence (xn ) ⊆ A, such
that xn ≠ x and (xn ) → x. Therefore, (xn ) ⊆ A ∪ B. So, x is also a limit point of A ∪ B.
Thus, x ∈ cl(A ∪ B).
We can similarly argue for the set B.
Consequently x ∈ cl(A ∪ B).
(b) Does this result extend to infinite unions of sets?
No, this result does not extend to infinite unions of sets.
Counterexample.
∞
Let An = 1 , where n ∈ N. Then, An is compact for all n ∈ N. But, ⋃ An is not
n n=1
compact, since the limit point 0 does not belong to the infinite union.
[Abbott 3.3.1] Show that if K is compact and nonempty, then sup K and inf K both exist
and are elements of K.
Proof.
By the Heine Borel Theorem, K is a closed and bounded set. Since K is bounded, by the
Axiom Of Completeness (AoC), both inf K and sup K exist.
We are interested to prove that sup K is an element of K.
1
By definition, for all 𝜖 > 0, there exists x ∈ K, such that s - 𝜖 < x ≤ s < s + 𝜖. Let 𝜖 = .
n
Then, we can construct a sequence (xn ) in K. Given any arbitrary 𝜖 > 0, if we choose
1
N > , then for all n ≥ N, xn ∈ (s - 𝜖, s + 𝜖). So, (xn ) → s. Thus, s is a limit point of K.
𝜖
Since K is closed, s belongs to K.
We can argue similarly for inf K ∈ K.
[Abbott 3.3.2.] Decide which of the following sets are compact. For those that are not
compact, show how definition 3.3.1. breaks down. In other words, give an example of a
sequence contained in the given set, that does not possess a subsequence converging to a
limit in the set.
(a) N.
N is not compact. Consider (xn ) = n, then no subsequence of (xn ) does converges to a
limit in the set. N is unbounded.
(b) Q ∩ [0, 1].
The set of rational numbers in [0, 1] is not compact. Consider the sequence (xn ) defined
1 2
recursively as xn+1 = xn + with x1 = 1. (xn ) ⊆ Q and (xn ) → 2, which does not
2 xn
belong to Q. So, clearly, not all limit points belong to the set.
(c) The Cantor Set.
The Cantor Set C is compact. It is closed and since C ⊆ [0, 1] it is bounded.

(d) 1 + 1 / 2 2 + 1 / 3 2 +…+ 1 / n 2 : n ∈ N .

The limit point 𝜋 2 / 6 does not belong to the set. So, it is not compact.
(e) { 1, 1 / 2, 2 / 3, 3 / 4, 4 / 5, … }
n
The general term is . By the Algebraic Limit Theorem,
n+1
n n+1-1 1
lim = lim = lim 1 - = 1.
n→∞ n + 1 n→∞ n+1 n→∞ n+1

The set is closed and bounded. Hence, it is compact.

[Abbott 3.3.3.] Prove the converse of Theorem 3.3.4 by showing that if a set K ⊆ R is
closed and bounded, then it is compact.
Proof.
Let K be a closed and bounded set.
Let (xn ) be an arbitrary sequence in K. As K is bounded, the sequence (xn ) is bounded. By
the Bolzanno Weierstrass Theorem, there exists a converges subsequence (xn ) of (xn ). Let
k

(xn ) → l. Since K is closed, l ∈ K.

k

As (xn ) was arbitrary, it follows that, every sequence (xn ) in K has a subsequence that
converges to a limit that is also in K. Hence, K is compact by definition.
[Abbot 3.3.4] Assume K is compact and F is closed. Decide if the following sets are
definitely compact, definitely closed, both, or neither.
(a) K ∩ F.
Proof.
Since K is closed, K ∩ F is closed. Moreover, since K is bounded, K ∩ F ⊆ K and therefore
bounded. So, K ∩ F is definitely compact.
(b) cl FC ∪ KC

Proof.
Since F C and K C are both open sets, F C ∪ K C is an open set. The closure of an open set is
closed.
(c) K ⧵ F.
Proof.
K⧵F is neither compact nor closed.
(d) cl K ∩ FC

Since K ∩ F C ⊆ K, it is a bounded set. Let x be an arbitrary limit point of the set. Since
K ∩ F C is bounded, it's limit point x must lie within the bounds. Thus, cl K ∩ F C is both
closed and bounded. Hence, it is compact.
[Abbott 3.3.5] Decide whether the following propositions are true or false. If the claim is
valid, supply a short proof, and if the claim is false, provide a counterexample.
(a) The arbitrary intersection of compact sets is compact.
Proof.
This claim is true.
The arbitrary intersection of closed sets is closed. Moreover, if the sets are bounded, the
intersection of the sets is also bounded. Consequently, the arbitrary intersection of compact
sets is compact.
(b) The arbitrary union of compact sets is compact.
Proof.
This proposition is false.

We define x1 = 1, with xn+1 = 1 xn +

2
.
2 xn

Let K1 = { x1 } and define :

Kn = { xn }
Then,
∞
⋃ Kn
n=1

does not contain the limit point 2 and is therefore not compact.

(c)Let A be arbitrary, and let K be compact. Then, the intersection A ∩ K is compact.

Proof.
This proposition is false.
Counterexample.
Let A = (0, 1) and K = [0, 1] then A ∩ K = (0, 1) which is not compact.
(d) If F1 ⊇ F2 ⊇ F3 ⊇ F4 ⊇ F5 ⊇ … is a nested sequence of nonempty closed sets, then
∞
the intersection ⋂ Fn ≠ ∅.
n=1
This proposition is false.
Counterexample.
∞
Consider the set ⋂ [n, ∞). Each of [n, ∞) is a closed set. But, their intersection is empty.
n=1

1
As yet another example, let (sn ) be the sequence of partial sums of the infinite series ∑ .
n
And consider the following closed sets:

F1 = { s1 , s2 , s3 , … }
F2 = { s2 , s3 , s4 , … }
F3 = { s3 , s4 , s5 , … }
⋮
Fn = { sn , sn+1 , sn+2 , … }
All of these sets are closed, because they do not have a limit point. And F1 ⊇ F2 ⊇ ….
But, their intersection is empty.
[Abbott 3.3.6] This exercise is meant to illustrate the point made in the opening
paragraph to the section 3.3. Verify that the following three statements are true if every
blank is filled in with the word finite. Which are true if every blank is filled in with the word
compact? Which are true if every blank is filled in with the word closed?
(a) Every _______ set has a maximum.
Solution.
Every finite set has a maximum.
Every compact set is bounded, so it has a supremum and further sup K is a limit point of
the set, so it belongs to the set. Thus, every compact set has a maximum.
(b) If A and B are _____, then A + B = { a + b : a ∈ A, b ∈ B } is also ______.
Solution.
If A and B are finite, then A + B has at most # of elements of A times # of elements of
B, so A + B is finite.

If A and B is bounded, then there exists M1 > 0, such that ∀a ∈ A, |a| ≤ M1 and there
exists M2 > 0, such that ∀b ∈ B, |b| ≤ M2 . Consequently, |a + b| ≤ |a| + |b| ≤ M1 + M2 ,
So, A + B is also bounded.

1
Consider A = n+ :n ∈ N and B = { -n : n ∈ N, n ≠ 2 }. Both A and B have no limit
n
points so they are closed, but A + B has a limit point 0 that does not belong to the set
A + B, and hence it is not closed.

Suppose that A and B are compact. Let c be a limit point of A + B. There exists a
sequence (cn ) ⊆ A + B such that cn ≠ c, with (cn ) → c. Every subsequence of a convergent
sequence also approached the limit point c.
Since cn ∈ A + B, we can write (cn ) as the sum of the sequences (an ) ⊆ A and (bn ) ⊆ B.
So:

cn = an + bn

Now, since A is compact, by the Heine Borel Theorem, there is a subsequence (an ) of (an )
k

that converges to a limit in A. Let lim an = a,

k

Now, corresponding subsequence terms in B can be expressed as:

 b nk = c nk - a nk

Taking limits on both sides, we have:

lim bnk = lim(cnk - ank )

Since both lim cn and lim an exist, we are allowed to apply the Algebraic Limit Theorem,
k k

and thus:

lim bnk = lim cnk - lim ank = c - a

Consequently, (bn ) is a convergent subsequence. Let b = lim bn . Since B is compact,

k k

b ∈ B. Thus, there exists a ∈ A, b ∈ B, such that c = a + b. Consequently, c ∈ A + B.

Therefore, A + B is compact.
(c) If { An : n ∈ N } is a collection of ______ sets with the property that every finite
∞
subcollection has a non-empty intersection, then ⋂ An is non-empty as well.
n=1

Solution.
If { An : 1 ≤ n ≤ N } is a collection of finite sets with the property that every finite
N
subcollection has a non-empty intersection, then ⋂ An is non-empty as well.
n=1

Consider An = [n, ∞). Each of the An 's are closed. The intersection of any finite
∞
subcollection of sets is non-empty, but ⋂ An is empty.
n=1

Let { An : n ∈ N } be a collection of compact sets. Since the arbitrary intersection of

compact sets is compact, consider the sequence of sets
 B1 = A1
B2 = A1 ∩ A2
B3 = A1 ∩ A2 ∩ A3
⋮ ⋮
n
Bn = ⋂ i=1 Ai
Since, any finite subcollection has a non-empty intersection, each of the Bi 's are non-
empty and compact. Moreover,

B1 ⊇ B2 ⊇ B3 ⊇ B4 ⊇ …
∞ ∞
By the Nested Compact set property, ⋂ Bi = ⋂ Ai is non-empty.
n=1 n=1

[Abbott 3.3.7] As some more evidence of the surprising nature of the Cantor set, follow
these steps to show that the sum C + C = { x + y : x, y ∈ C } is equal to the closed interval
[0, 2]. (Keep in mind that C has zero length and contains no intervals.)

Because C ⊆ [0, 1], C + C ⊆ [0, 2], so we only need to prove the reverse inclusion
[0, 2] ⊆ { x + y : x, y ∈ C }. Thus, given s ∈ [0, 2], we must find two elements x, y ∈ C,
satisfying x + y = s.
(a) Show that there exists x1 , y1 ∈ C1 , for which x1 + y1 = s. Show in general that, for
arbitrary n ∈ N, we can always find xn , yn ∈ Cn for which xn + yn = s.
Solution.
Consider an arbitrary s ∈ [0, 2].
Consider the straight-line x + y = s. Let us represent the regions [0, 1 / 3] × [0, 1 / 3],
[0, 1 / 3] × [1 / 3, 2 / 3], [2 / 3, 1] × [0, 1 / 3] and [2 / 3, 1] × [2 / 3, 1] graphically. For all
s ∈ [0, 2], the straight-line x + y = s will pass through atleast one of these four squares.

2 1 1
If s ∈ 0, , then we can pick x1 ∈ 0, and y1 ∈ 0, .
3 3 3
 2 4 2 1
If s ∈ , , then we can pick x1 ∈ ,1 and y1 ∈ 0, .
3 3 3 3

4 2 2
If s ∈ ,2 , then we can pick x1 ∈ ,1 and y1 ∈ ,1 .
3 3 3

x+y = s
s

Thus, there exists x1 , y1 ∈ C1 such that x1 + y1 = s.

For n = 2, each square is further subdivided into 4 square regions.
In general, for n = k, there are 4 k squares. By symmetry, the line x + y = s is bound to
pass through atleast one of the 4 k squares. Thus, there exists xn , yn ∈ Cn such that
xn + yn = s for all k ∈ N.

(b) Keeping in mind that the sequences (xn ) and (yn ) do not necessarily converge, show
how they can nevertheless be used to produce the desired x and y in C satisfying x + y = s.
Since (xn ) ⊆ [0, 1], it is a bounded sequence. By the Bolzanno Weierstrass Theorem, the
sequence (xn ) has atleast one convergent subsequence (xn ). Let lim xn = x.
k k

Since

y n k = (x n k + y n k ) - x n k

we have that:
 lim(ynk ) = lim(xnk + ynk ) - lim xnk = s - x = y

Thus, (yn ) is a convergent sequence.

k

Since (xn ) ⊆ Cn and (yn ) ⊆ Cn for all n ≥ n1 , it follows that these subsequences lie in
k k
∞ ∞
the infinite intersection ⋂ Cn and thus in ⋂ Cn . Since, this is a closed and bounded set,
n=n 1 n=1
∞
their limit points x, y ∈ ⋂ Cn = C.
n=1

[Abbott 3.3.8] Let K and L be non-empty compact sets, and define

d = inf{ |x - y| : x ∈ K, y ∈ L }

This turns out to be a reasonable definition for the distance between K and L.
(a) If K and L are disjoint show that d > 0 and that d = |x0 - y0 | for some x0 in K and
y 0 ∈ L.

Proof.
Let

A = { | x - y| : x ∈ K , y ∈ L }

Since |x - y| ≥ 0, A is bounded below. Thus, inf A exists. We are interested to prove that
A is compact and therefore d = inf A exists and further belongs to A.

Let a be an arbitrary limit point of A. Thus, there exists a sequence (an ) = |xn - yn | ⊆ A,
such that (an ) = |xn - yn | → a. Since K is bounded, by the Bolzanno Weierstrass theorem,
there exists a convergent sequence (xn ) ⊆ K, whose limit point also belongs to K.
k

We now define the sequence yn such that |xn

k k - y nk | = a nk . Since, yn satisfies this
k
equation, it must be a subsequence of (yn ). Since both (an ) and (xn ) are convergent,
k k

(yn ) is convergent. Since, lim xn ∈ K and lim yn ∈ L, we must have that a ∈ A.

k k k

Consequently, inf A ∈ A.
So, there exists x0 ∈ K and y0 ∈ L such that d = |x0 - y0 |.
(b) Show that it's possible to have d = 0 if we assume only that the disjoint sets K and L
are closed.
[Abbott 3.4.1] If P is a perfect set and K is compact, is the intersection P ∩ K always
compact? Always perfect?
Proof.
Since P is closed and K is closed and bounded, and the intersection of closed sets is closed,
P ∩ K is closed. Further P ∩ K ⊆ K, so it is bounded and therefore always compact.

This is false. Consider P = [0, 1] and K = { 1 }. P ∩ K = { 1 } which is not a perfect set.

[Abbott 3.4.2] Does there exist a perfect set consisting of only rational numbers?
No, the rational numbers Q or any subset of it is not perfect. Q is not closed and hence it
is not perfect.
[Abbott 3.4.3] Review the portion of the proof given in example 3.4.2. and follow these
steps to complete the argument.
(a) Because x ∈ C1 , argue that there exists an x1 ∈ C ∩ C1 with x1 ≠ x satisfying
| x - x 1 | ≤ 1 / 3.

Solution.
∞
Let x ∈ C be arbitrary. Since C = ⋂ Cn , x ∈ Cn for all n ∈ N.
n=1

We have established that x ∈ C1 . Suppose that x ∈ [a1 , b1 ] ⊆ C1 . We choose

 x1 =
a1 if x ≤ (a1 + b1 ) / 2
b1 otherwise

That is, we let x1 to be the left-hand enpoint of the closed interval, if x is smaller than
than the mid-point, otherwise we let x1 be the right-hand endpoint. Since the length of the
1 1
interval l(I1 ) = , we have that |x - x1 | ≤ .
3 3

Since, endpoints of intervals are never excluded in the construction of the Cantor set,
x1 ∈ C ∩ C1 .

(b) Finish the proof

Similarly, we know that x ∈ C2 . Suppose that x ∈ [a2 , b2 ] ⊆ C2 . Then, we choose

x2 =
if x ≤ (a2 + b2 ) / 2
a2
otherwise
b2
In general, if x ∈ [an , bn ] ⊆ Cn , then we choose

xn =
an if x ≤ (an + bn ) / 2
bn otherwise

1 1
Since l(In ) = n
, | x n - x| ≤ . Pick an arbitrary 𝜖 > 0. Pick N, such that
3 3n

1
<𝜖
3N
or

log(1 / 𝜖)
N>
log 3
Then, for all n ≥ N,

| x n - x| < 𝜖
Consequently, (xn ) → x.
Thus, all points in the Cantor Set C are limit points.
[Abbott 3.4.4] Repeat the Cantor construction from section 3.1 starting with the interval
[0, 1]. This time, however, remove the open middle fourth from each component.

(a) Is the resulting set compact? Perfect?

Proof.
We have:

3 5
C1 ' = 0, ∪ ,1
8 8
9 15 24 40 49 55
C2 ' = 0, ∪ , ∪ , ∪ ,1
64 64 64 64 64 64
⋮
∞
Since ⋂ Cn ' is closed and bounded, it is a compact set.
n=1

∞
Let x ∈ C' be an arbitrary point. Since x ∈ ⋂ Cn ', x ∈ Cn ' for all n ∈ N. Again, we can
n=1
construct a sequence (xn ) ⊆ C' such that (xn ) → x.
Pick an arbitrary 𝜖 > 0.
We can choose N such that
N
3
<𝜖
8

Then, for all n ≥ N, |xn - x| < 𝜖.

Thus, every point is a limit point of C'. Consequently, C' is closed and has no isolated
points. Therefore, C' is perfect.
(b) Using the algorithms from section 3.1, compute the length and dimension of this Cantor-
like set.
Proof.
The length of the middle 1 / 4th Cantor set is given as follows:
n-1
∞ 1 3
l C (1/4) = 1 - ∑ n=1 2 n-1
⋅
4 8
n-1
1 ∞ 3
= 1 - ∑ n=1
4 4
1 1
= 1- ⋅
4 1- 3
4
=0
If we magnify the Cantor set C (1/4) by a factor of 8 / 3, this results in [0, 8 / 3]. Removing
the open middle 1 / 4th results in two intervals [0, 1] and [5 / 3, 8 / 3] which is where we
started in the original construction, except that we now stand to produce an additional copy
of C (1/4) in the interval [5 / 3, 8 / 3]. Magnifying the Cantor set C (1/4) by a factor of 8 / 3
yields two copies of the original set.
Thus,

( 8 / 3) x = 2
Or,

log(8 / 3)
x=
log 2

[Abbott 3.4.5] Let A and B be non-empty subsets of R. Show that if there exists disjoint
open sets U and V with A ⊆ U and B ⊆ V, then A and B are separated.
Solution.
We are given that U and V are disjoint open sets. We proceed by conradiction.
Let a be an arbitrary limit point of A. Assume that a ∈ B.
Since, B ⊆ V, a is an interior point of V. Consequently, there exists 𝜖0 > 0, such that
V𝜖 (a) = (a - 𝜖0 , a + 𝜖0 ) ⊆ V. Since U and V are disjoint open sets, V𝜖 (a) ∩ A = ∅ and
0 0

therefore (V𝜖 (a) - { a }) ∩ A = ∅.

0

But, a is the limit point of a. For all 𝜖 > 0, it follows that (V𝜖 (a) - { a }) ∩ A ≠ ∅. This is a
contradiction. Hence, our initial assumption is false. a ∉ B.
Since a was arbitrary, this holds true for all limit points of A. So, cl(A) ∩ B = ∅. We can
similarly argue that A ∩ cl(B) = ∅.
Therefore, A and B are separated.
[Abbott 3.4.6] Prove theorem 3.4.6.
Theorem 3.4.6. A set E ⊆ R is connected, if and only if, for all nonempty disjoint sets A
and B satisfying E = A ∪ B, there always exists a convergent sequence (xn ) → x with (xn )
contained in one of A or B, and x an element of the other.
Proof.
⟹ direction.

We are given that E is connected. We proceed by contradiction.

Assume that there exist non-empty disjoint sets A, B satisfying E = A ∪ B, such that for all
convergent sequences (xn ) → x, both ((xn ) ⊆ A) ∧ (x ∈ B) and ((xn ) ⊆ B) ∧ (x ∈ A) are
false.
Let a be a limit point of A. Then, ∃(an ) ⊆ A, such that (an ) → a. From above, it follows
that a ∉ B. Since, a is a limit point of A, we have: a ∈ cl(A), a ∉ B. Since a was arbitrary,
we must have: cl(A) ∩ B = ∅.
We can similarly argue that A ∩ cl(B) = ∅.
Since E is connected, by definition, for all non-empty disjoint pairs of sets C, D satisfying
E = C ∪ D, atleast one of cl(C) ∩ D or C ∩ cl(D) is non-empty.
This is a contradiction. Hence, our initial assumption is false.
⟸ direction.
We are given that, for all non-empty disjoint pairs A, B satisfying E = A ∪ B, there always
exists a convergent sequence (xn ) → x, such that atleast one of the following holds
(i) ((xn ) ⊆ A) ∧ (x ∈ B)
(ii) ((xn ) ⊆ B) ∧ (x ∈ A)
It follows, that atleast one of the following holds:
(i) there exists a limit point a of A, such that a ∈ B. Since, a ∈ cl(A), cl(A) ∩ B ≠ ∅
(ii) there exists a limit point b of B, such that b ∈ A. Since b ∈ cl(B), A ∩ cl(B) ≠ ∅
Thus, by definition E is connected.
[Abbott 4.2.1] (a) Supply details for how corollary 4.2.4 part (ii) follows from the
Sequential criterion for functional limits in theorem 4.2.3 and the Algebraic Limit Theorem
for sequences proved in chapter 2.
Proof.
We are interested to prove that, given xlim
→c
f(x) = L and lim g(x) = M,
x→c

lim [f(x) + g(x)] = L + M

x→c

Let (xn ) be an arbitrary sequence, satisfying xn ≠ c, with (xn ) → c.

By the sequential criterion for functional limits, it follows that lim f(xn ) = L and
lim g(xn ) = M.

Therefore,

lim [f(xn ) + g(xn )] = lim f(xn ) + lim g(xn ) Algebraic limit theorem for sequences
= L+M
Since, (xn ) was arbitrary, this is true for all sequences (xn ) ⊆ A, with xn ≠ c and (xn ) → c.
Thus, by the Sequential Criterion for functional limits,

lim [f(x) + g(x)] = L + M

x→c

(b) Now, write another proof of corollary 4.2.4 part (ii) directly from the definition 4.2.1
without using the sequential criterion in theorem 4.2.3.
Proof.
We are given that xlim
→c
f(x) = L and lim g(x) = M. Since:
x→c

Pick an arbitrary 𝜖 > 0.

There exists 𝛿1 > 0, such that for all x ∈ (c - 𝛿1 , c + 𝛿1 ), we have | f(x) - L| < 𝜖 / 2.
There exists 𝛿2 > 0, such that for all x ∈ (c - 𝛿 2 , c + 𝛿 2 ), we have |g(x) - M| < 𝜖 / 2.
Pick 𝛿 = min{ 𝛿1 , 𝛿2 }. Then, for all x ∈ (c - 𝛿, c + 𝛿), we have:

| f(x) + g(x) - (L + M)| = | f(x) - L + g(x) - M |

≤ | f ( x) - L | + | g( x) - M |
𝜖 𝜖
< + =𝜖
2 2

Consequently, xlim
→c
f ( x) + g( x) = L + M .

(c) Repeat (a) and (b) for corollary 4.2.4 part (iii).
Proof.
(i) We are given that lim f(x) = L and lim g(x) = M.
x→c x→c

Let (xn ) be an arbitrary sequence, satisfying xn ≠ c, such that (xn ) → c.

By the Sequential criterion for Functional Limits, lim f(xn ) → L and lim g(xn ) → M.
We have:

lim f(xn ) ⋅ g(xn ) = lim f(xn ) ⋅ lim g(xn ) Algebraic limit theorem for sequences
= L⋅M

Since (xn ) was an arbitrary sequence, this is true for all sequences (xn ) → c, with xn ≠ c.
By the sequential criterion for functional limits, lim f(x)g(x) = LM.
x→c

(ii) We are interested to make the distance |f(x)g(x) - LM| as small as we please. Pick an
arbitrary 𝜖 > 0.
Let us explore the expression |f(x)g(x) - LM|. We have:

|f(x)g(x) - LM| = |f(x)g(x) - Lg(x) + Lg(x) - LM|

≤ |g(x)||f(x) - L| + |L||g(x) - M|

If we replace |g(x)| by its upper bound, we will strengthen the condition the we wish to
prove.
Pick 𝜖 = 1. There exists 𝛿1 > 0, such that for all |x - c| < 𝛿1 , we have :

| g( x) - M | < 1
or

| g( x) | < | M | + 1

There exists 𝛿2 > 0, such that for all |x - c| < 𝛿2 , we have:

𝜖
| f ( x) - L | <
2( | M | + 1)
There exists 𝛿3 > 0 such that for all |x - c| < 𝛿3 , we have:
𝜖
| g( x) - M | <
2|L|
Let 𝛿 = min{ 𝛿1 , 𝛿2 , 𝛿3 }. Then, for all |x - c| < 𝛿, it follows that:

|f(x)g(x) - LM| ≤ |g(x)||f(x) - L| + |L||g(x) - M|

𝜖 𝜖
< ( | M | + 1) + | L|
2( | M | + 1) 2|L|
=𝜖
Consequently, lim f(x)g(x) = LM.
x→c

[Abbott 4.2.2] For each stated limit, find the largest possible 𝛿-neighbourhood that is a
proper response to the given 𝜖-challenge.
(a) lim (5x - 6) = 9 where 𝜖 = 1.
x→3

Proof.
We are interested to make the distance |(5x - 6) - 9| < 1. We have:

|(5x - 6) - 9| < 1
|5x - 15| < 1
1
|x - 3| <
5
Thus, the largest 𝛿 - neighbourhood that is a proper response to the given 𝜖-challenge is
1 1
3- ,3+ .
5 5

(b) lim x = 2, where 𝜖 = 1.

x→4

Proof.

We are interested to make the distance | x - 2| < 𝜖. We have:

 | x - 2| < 𝜖
| x + 2|
| x-2 |× <𝜖
| x + 2|
x-4
<𝜖
x+2

Since 𝜖 = 1, we would like to have, |x - 4| < | x + 2|.

If we replace | x + 2| by its lower bound, we can strengthen the condition we wish to

prove.

If 𝜖 = 1, there exists 𝛿 > 0, such that for all |x - 4| < 𝛿, we have 1 < x < 3.
Consequently, 3 < x + 2 < 5. Thus, 3 is a lower bound for x + 2.
Therefore, let x be such that |x - 4| < 3. Then, x ∈ (1, 7).
(c) lim [[x]] = 3, where 𝜖 = 1.
x→𝜋

Proof.
We are interested to make the distance |[[x]] - 3| < 1.
We have:

-1 < [[x]] - 3 < 1

⟺ 2 < [[x]] < 4
⟺ [[x]] = 3
If we pick 𝛿 = 𝜋 - 3, then for all x ∈ V𝛿 (𝜋), it follows that [[x]] = 3 or |[[x]] - 3| < 1.
(d) lim [[x]] = 3, where 𝜖 = 0.01.
x→𝜋

We are interested to make the distance |[[x]] - 3| < 0.01.

We have:

-0.01 < [[x]] - 3 < 0.01

⟺ 2.99 < [[x]] < 3.01
⟺ [[x]] = 3
Again if we pick 𝛿 = 𝜋 - 3, then for all x ∈ V𝛿 (𝜋), it follows that [[x] = 3 or
|[[x]] - 3| < 0.01.

[Abbott 4.2.3] Review the definition of Thomae's function t(x) from section 4.1.
(a) Construct three different sequences (xn ), (yn ) and (zn ), each of which converges to 1
without using the number 1 as a term in the sequence.
Proof.
The Thomae's function is given by:

1 if x = 0
t ( x) = 1/n if x = m / n ∈ Q ⧵ { 0 }
0 if x ∉ Q
1 2
Let xn = 1 - , yn = 1 - , n ≥ 2, and zn be the sequence defined as:
n n

1
1- if n = 2m, m∈N
n
zn =
2
1- if n = 2m + 1, m ∈ N, m ≥ 1
n

(b) Now, compute lim t(xn ), lim t(yn ) and lim t(zn ).
Proof.
n-1 1
Since xn = , t(xn ) = . Thus, t(xn ) → 0.
n n
Since yn is irrational, t(yn ) = 0. Thus, t(yn ) is the constant zero sequence and approaches
0.

Since t(zn ) is the shuffle sequence consisting of the terms of t(xn ) and t(yn ) juxtaposed
next to each other, t(zn ) → 0.
(c) Make an educated conjecture for the limit lim t(x) and use the definition 4.2.1B to
x→1
verify the claim.
Our claim is that xlim
→1
t ( x) = 0.

We proceed by contradiction. Assume that there exists 𝜖0 > 0 for all 𝛿 > 0 such that for
atleast some x ∈ (1 - 𝛿, 1 + 𝛿) different from 1, we have |t(x)| ≥ 𝜖0 .
1
By the Archimedean property, there exists N ∈ N, such that < 𝜖0 .
N

1 1 1
Consider 𝛿 = . There exists x ∈ 1 - , 1 + different from 1 such that
2N 2N 2N
t ( x) ∉ ( - 𝜖 0 , 𝜖 0 ) .

Such an x must necessarily have a denominator greater than 2N. But then,
1
t(x) ∈ 0, and so, t(x) ∈ (-𝜖0 , 𝜖0 ). This is a contradiction. Hence, our initial
2N
assumption is false.
[Abbott 4.2.4] Consider the reasonable but erroneous claim that:

lim 1 / [[x]] = 1 / 10
x → 10

Solution.
(a) Find the largest 𝛿 that represents a proper response to the challenge of 𝜖 = 1 / 2.

Proof.
Suppose that:

1 1 1
- <
[[x]] 10 2
Then,

1 1 1 1 1
- < < +
10 2 [[x]] 10 2
or,

2 1 3
- < <
5 [[x]] 5

2 1 5 2
Consider - < . If [[x]] < 0, then - > [[x]]. If [[x]] > 0, then - < [[x]].
5 [[x]] 2 5

1 3 5 5
Consider < . If [[x]] < 0, then [[x]] < . If [[x]] > 0, then [[x]] > .
[[x]] 5 3 3

5 5
Consequently, it follows that [[x]] < - or [[x]] > .
2 3

Therefore, x < -2 or x ≥ 2. That is x - 10 < -12 or x - 10 ≥ -8.

Thus, if |x - 10| ≤ 8, the inequality is satisfied. Hence, 𝛿𝜖=1/2 = 8 is the largest 𝛿 response
to the given 𝜖-challenge.
(b) Find the largest 𝛿 that represents a proper response to 𝜖 = 1 / 50.
Suppose that:

1 1 1 1 1
- < < +
10 50 [[x]] 10 50
Thus,

2 1 3
< <
25 [[x]] 25

So, [[x]] > 0. Thus,

25
0 < [[x]] <
2

and

25
[[x]] >
3

Consequently,
1 ≤ x < 13
and

x≥7
For both these conditions to be true simultaneously, we must have:

7 ≤ x < 13
or

-3 ≤ x - 10 < 3
Thus, the inequality is satisfied for all x such that:

|x - 10| < 3
So, the largest 𝛿𝜖=1/50 response to the given 𝜖-challenge is 3.
(c) Find the largest 𝜖-challenge for which there is no suitable 𝛿 response possible.
Proof.
 1
I think this is 𝜖0 = .
90

[Abott 4.2.5.] Use definition 4.2.1 to supply a proper proof for the following limit
statements.
(a) lim (3x + 4) = 10.
x→2

Proof.
We are interested to make the distance |(3x + 4) - 10| as small as we please.
Pick an arbitrary 𝜖 > 0. Let's explore the inequality:

|(3x + 4) - 10| < 𝜖

|3x - 6| < 𝜖
𝜖
|x - 2| <
3

𝜖
If we choose 𝛿 = , then |x - 2| < 𝛿 implies that |(3x + 4) - 10| < 𝜖. Consequently,
3
lim (3x + 4) = 10.
x→2

(b) lim x 3 = 0.
x→0

We are interested to make the distance |x 3 | as small as we please.

Pick an arbitrary 𝜖 > 0. Let's explore the inequality:

|x 3 | < 𝜖
∴ -𝜖 < x 3 < 𝜖
∴ -∛𝜖 < x < ∛𝜖
|x| < ∛𝜖
Pick 𝛿 = 𝜖 1/3 . Then, |x| < 𝛿 implies that |x 3 | < 𝜖.
Since 𝜖 was arbitrary, this is true for all 𝜖 > 0. Consequently, lim x 3 = 0.
x→0

(c) lim x 2 + x - 1 = 5.
x→2

We are interested to make the distance | x 2 + x - 1 - 5| as small as we please.

Pick an arbitrary 𝜖 > 0. Let's explore the inequality:

| x 2 + x - 1 - 5| < 𝜖
|x 2 + x - 6| <𝜖
|x 2 + 3x - 2x - 6| <𝜖
| x( x + 3) - 2( x + 3) | <𝜖
|(x - 2)(x + 3)| <𝜖
Assume that 𝛿 < 1. Then |x - 2| < 𝛿 implies that x ∈ (1, 3). Consequently, 4 < x + 3 < 6.
So, |x + 3| < 6. If in the above inequality, we replace |x + 3| by its upper bound, we are
strengthening the condition we wish to prove.
So, we would like to therefore prove:
𝜖
|x - 2| <
6
If we choose 𝛿 = min{ 1, 𝜖 / 6 }, then |x - 2| < 𝛿 implies that | x 2 + x - 1 - 5| < 𝜖.

Consequently, lim x 2 + x - 1 = 6.
x→2

(d) lim 1 / x = 1 / 3.
x→3

1 1
We are interested to make the distance - as small as we please.
x 3

Pick an arbitrary 𝜖 > 0. Let's explore the inequality

 1 1
- <𝜖
x 3
|x - 3|
<𝜖
3|x|
8 1 17 1
Assume that 𝛿 < . Then < x < . So, |x| > . If we replace |x| by it's lower bound
3 3 3 3
in the above inequality, we strengthen the condition, we wish to prove. Thus, our claim is:

|x - 3| < 𝜖
8 1 1
If we choose 𝛿 = min ,𝜖 , then |x - 3| < 𝛿 implies - < 𝜖.
3 x 3

1 1
Since 𝜖 was arbitrary, this holds true for all 𝜖 > 0. Consequently, xlim = .
→3 x 3

[Abbott 4.2.6] Decide if the following claims are true or false and give short justifications
for each conclusion.
(a) If a particular 𝛿 has been constructed as a suitable response to a particular 𝜖-challenge,
then any small positive 𝛿 will also suffice.
Proof.
This proposition is true.
Suppose a particular 𝛿-neighbourhood has been constructed in response to a particular 𝜖-
challenge. Then, for all x ∈ (c - 𝛿, c + 𝛿), we have f(x) ∈ (L - 𝜖, L + 𝜖).
If 0 < 𝜉 < 𝛿, then V𝜉 (c) ⊆ V𝛿 (c). Consequently, for all x ∈ (c - 𝜉, c + 𝜉), it follows that
f(x) ∈ (L - 𝜖, L + 𝜖).

(b) If lim f(x) = L and a happens to be in the domain of f, then L = f(a).

x→a

This proposition is false.

Consider f(x) defined piecewise as:

f ( x) =
x if x > 0
1 if x = 0
lim f(x) = lim (x) = 0. But, f(0) = 1.
x→0 x→0

(c) lim f(x) = L, then lim 3[f(x) - 2] 2 = 3(L - 2) 2 .

x→a x→a

This proposition is true.

lim x → a 3[f(x) - 2] 2 = 3 ⋅ lim x → a[f(x) - 2] ⋅ lim x → a[f(x) - 2] { Algebraic Limit theorem

for functional limits }
= 3 ⋅ ( L - 2) ⋅ ( L - 2)
= 3( L - 2) 2
(d) xlim
→a
f(x) = 0, then lim f(x)g(x) = 0 for any function g (with domain equal to the
x→a
domain of f)
This proposition is false.
Let f(x) = x - a be a function defined for all R - { a }. Then, xlim
→a
f(x) = 0. And let
1
g( x) = , x ≠ a. We have:
x-a

1
lim f(x) ⋅ g(x) = lim (x - a) ⋅ = lim (1) = 1
x→a x→a ( x - a) x → a

[Abbott 4.2.7] Let g : A → R and assume that f is a bounded function on A in the sense
that there exists M > 0 satisfying |f(x)| ≤ M for all x ∈ A. Show that if lim g(x) = 0,
x→c
then xlim
→c
g(x)f(x) = 0 as well.

Proof.
We are interested to make the distance |g(x)f(x)| as small as we please.
Pick an arbitrary 𝜖 > 0.
There exists 𝛿 > 0, such that for all x ∈ (c - 𝛿, c + 𝛿), it follows that:
𝜖
| g( x) | <
M
Since, f(x) is bounded, there exists M > 0, for all x ∈ A, such that

| f ( x) | ≤ M
Consequently, for all x ∈ (c - 𝛿, c + 𝛿), we have:
𝜖
|g(x)f(x)| = |g(x)||f(x)| < ⋅M = 𝜖
M
Thus, lim g(x)f(x) = 0.
x→c

[Abbott 4.2.8] Compute each limit or state that it does not exist. Use the tools developed
in this section to justify each conclusion.
|x - 2|
(a) lim .
x → 2 ( x - 2)

Proof.
This limit does not exist.
1
Consider the sequence (xn ) defined by xn = 2 + . Now, lim xn = 2. We have:
n

|xn - 2|
lim f(xn ) = lim n → ∞
( x n - 2)
( x n - 2)
= lim since xn ≥ 2
( x n - 2)
= lim 1
=1
 1
Consider the sequence (yn ) defined by yn = 2 - . Now, lim yn = 2. We have:
n

|yn - 2|
lim f(yn ) = lim n → ∞
( y n - 2)
- ( y n - 2)
= lim since yn ≤ 2
( y n - 2)
= lim (-1)
= -1
Therefore, there exists two sequences (xn ) and (yn ) such that lim xn = lim yn , but
|x - 2|
lim f(xn ) ≠ lim f(yn ). Consequently, lim does not exist.
x→2 ( x - 2)

|x - 2|
(b) xlim
→ 7/4 ( x - 2)

Proof.
Our claim is xlim
→ 7/4
f ( x) = - 1.

We are interested to make the distance |f(x) - (-1)| as small as we please.

Pick an arbitrary 𝜖 > 0.
Let's explore the inequality:

| ( x - 2) |
- ( - 1) < 𝜖
( x - 2)
In other words:

|x - 2| + (x - 2)
<𝜖
( x - 2)
 1 7 1 3
Assume that 𝛿 < . Then, x - < implies that x ∈ ,2 . Consequently, x < 2 and
4 4 4 2
so |x - 2| = -(x - 2). Thus, we have:

0
< 𝜖,   { x ≠ 2 }
( x - 2)
Since 𝜖 > 0, this is vacuously true.
1
Thus, a 𝛿 < is a suitable response to any given 𝜖-challenge.
4
|x - 2|
Therefore, xlim = - 1.
→ 7/4 ( x - 2)

(c) lim (-1) [[1/x]] .

x→0

Proof.
This limit does not exist.
1 1
Let (xn ) be a sequence defined by xn = , n ∈ N. We have, (xn ) → 0. Now, = 2n
2n xn
and
f(xn ) = (-1) [[1/xn ]] = (-1) 2n = 1
So, lim f(xn ) = 1.
1
Let (yn ) be a sequence defined by yn = , n ∈ N. We have, (yn ) → 0. And
2n + 1

f(yn ) = (-1) [[1/yn ]] = (-1) 2n+1 = (-1)

So, lim f(yn ) = - 1.

Thus, ∃(xn ), (yn ) such that lim xn = lim yn = 0, but lim f(xn ) ≠ lim f(yn ).
(d) lim ∛x(-1) [[1/x]] .
x→0
Our claim is lim ∛x(-1) [[1/x]] = 0.
x→0

Since, |(-1) [[1/x]] | ≤ 1, given an arbitrary 𝜖 > 0, we can pick 𝛿 = 𝜖 3 . Then, for all |x| < 𝛿,
it follows that |∛x||(-1) [[1/x]] | < ∛𝜖 3 ⋅ 1 = 𝜖, since h(t) = ∛t is a monotonically increasing
function.
[Abbott 4.2.9] (Infinite Limits.) The statement lim 1 / x 2 = ∞ certainly makes intuitive
x→0
sense. To construct a rigorous definition in the challenge response style of the definition
4.2.1., for an infinite limit statement of this form, we replace the arbitrarily small 𝜖 > 0
challenge with an (arbitrarily large) M > 0 challenge:
Definition: lim f(x) = ∞ means that for all M > 0 we can find a 𝛿 > 0 such that
x→c
whenever 0 < |x - c| < 𝛿, it follows that f(x) > M.
(a) Show that lim 1 / x 2 = ∞ in the sense described in the previous definition.
x→0

Proof.
1
Pick an arbitrary M > 0. We are interested to make > M. Let's explore this inequality.
x2

1
>M
x2
1
x2 <
M
1
| x| <
M
1 1
Pick 𝛿 = , then |x| < 𝛿 implies that > M. Since M > 0 was arbitrary, this is true
M x2
for all M > 0. Consequently,

1
lim =∞
x→0 x2
(b) Now construct a definition for the statement xlim
→∞
f(x) = L. Show that lim 1 / x = 0.
x→∞

Proof.
Definition. xlim
→∞
f(x) = L means that for all 𝜖 > 0, there exists M > 0, such that for all
|x| > M, it follows that |f(x) - L| < 𝜖.

1
We are interested to make the distance < 𝜖. Assume that M > 1, then |x| > M
x
implies that |x| > 1. Therefore,

1
| x| >
𝜖
1 1
If we pick M > max 1, , then |x| > M implies that < 𝜖. Consequently,
𝜖 x
1
lim = 0.
x→∞ x

(c) What would a rigorous definition for xlim

→∞
f(x) = ∞ look like? Give an example of such
a limit.
Solution.
Definition. For all M > 0, there exists N > 0, such that for all x > N, it follows that
f ( x) > M .

[Abbott 4.2.10] Right and left limits. Introductory calculus courses typically refer to the
right-hand limit of a function as the limit obtained by letting x approach a from the right-
hand side
(a) Give a proper definition in the style of the definition 4.2.1 for the right-hand and left-
hand limit statements:

lim f(x) = L   and   lim f(x) = M

x → a+ x → a-

Proof.
lim f(x) = L means that for all 𝜖 > 0, there exists 𝛿 > 0, such that for all x ∈ (a, a + 𝛿),
x → a+
it follows that f(x) ∈ (L - 𝜖, L + 𝜖).
lim f(x) = M means that for all 𝜖 > 0, there exists 𝛿 > 0, such that for all x ∈ (a - 𝛿, a),
x → a-
it follows that f(x) ∈ (L - 𝜖, L + 𝜖).
(b) Prove that lim f(x) = L if and only if both the right and the left-handside limits equal
x→a
L.

⟹ direction.
We are given xlim
→a
f ( x) = L .

Pick an arbitrary 𝜖 > 0.

There exists 𝛿 > 0, such that for all x ∈ (a - 𝛿, a + 𝛿), f(x) ∈ (L - 𝜖, L + 𝜖).
Let t be an arbitrary point in (a - 𝛿, a). Since |t - a| < 𝛿, f(t) ∈ (L - 𝜖, L + 𝜖).
As t is arbitrary, this is true for all t ∈ (a − 𝛿, a). Since 𝜖 was arbitrary, this is true for all
𝜖 > 0.

Consequently, lim f(x) = L.

x → a-

Let s be an arbitrary point (a, a + 𝛿). Since |s - a| < 𝛿, f(s) ∈ (L - 𝜖, L + 𝜖).

As s is arbitrary, this is true for all s ∈ (a, a + 𝛿). Since 𝜖 was arbitrary, this is true for all
𝜖 > 0.

Consequently, lim f(x) = L

x → a+

⟸ direction.

We are given that xlim

→ a-
f(x) = lim f(x) = L.
x → a+
Pick an arbitrary 𝜖 > 0.
There exists 𝛿1 > 0, such that for all x ∈ (a - 𝛿1 , a), f(x) ∈ (L - 𝜖, L + 𝜖).
There exists 𝛿2 > 0, such that for all x ∈ (a, a + 𝛿2 ), f(x) ∈ (L - 𝜖, L + 𝜖).
Pick 𝛿 = min{ 𝛿1 , 𝛿2 }. Then, for all x ∈ (a - 𝛿, a + 𝛿), f(x) ∈ (L - 𝜖, L + 𝜖).
Since 𝜖 was arbitrary, this is true for all 𝜖 > 0.
Consequently, lim f(x) = L.
x→a

[Abbott 4.2.11] (Squeeze Theorem.) Let f, g and h satisfy f(x) ≤ g(x) ≤ h(x) for all x
in some common domain A. If xlim
→c
f(x) = L and lim h(x) = L at some point c of A, show
x→c
that lim g(x) = L as well.
x→c

Proof.
Pick an arbitrary 𝜖 > 0.
There exists 𝛿1 > 0, such that for all x ∈ (c - 𝛿1 , c + 𝛿1 ), f(x) ∈ (L - 𝜖, L + 𝜖).
There exists 𝛿2 > 0, such that for all x ∈ (c - 𝛿2 , c + 𝛿2 ), h(x) ∈ (L - 𝜖, L + 𝜖).
Let 𝛿 = min{ 𝛿1 , 𝛿2 }.
But since f(x) ≤ g(x) ≤ h(x), for all x ∈ (c - 𝛿, c + 𝛿), we have that
L - 𝜖 < f ( x) ≤ g( x) ≤ h( x) < L + 𝜖.

Since 𝜖 was arbitrary, this holds true for all 𝜖 > 0. Consequently, xlim
→c
g( x) = L .
[Abbott 4.3.1] Let g(x) = ∛x.
(a) Prove that g is continuous at c = 0.
Proof.
We are interested to make the distance |g(x) - g(0)| as small as we please.
Pick an arbitrary 𝜖 > 0.
Let's explore the inequality |g(x)| < 𝜖.

| g( x) | < 𝜖
|∛x| < 𝜖
| x| < 𝜖 3
Pick 𝛿 = 𝜖 3 . Then, for all |x| < 𝛿, it follows that |g(x)| < 𝜖. Since 𝜖 was arbitrary, this is
true for all 𝜖 > 0. Consequently, g(x) is continuous at c = 0.
(b) Prove that g is continuous at a point c ≠ 0. (The identity
a 3 - b 3 = (a - b) a 2 + ab + b 2 will be helpful).

Proof.
We are interested to make the distance |g(x) - g(c)| as small as we please.
Consider the inequality |g(x) - g(c)| < 𝜖.

| g( x) - g( c ) | < 𝜖
|∛x - ∛c| < 𝜖
|x 2/3 + x 1/3 c 1/3 + c 2/3 |
|∛x - ∛c| × <𝜖
|x 2/3 + x 1/3 c 1/3 + c 2/3 |
| x - c|
<𝜖
|x 2/3 + x 1/3 c 1/3 + c 2/3 |

Now, if we complete the square in the denominator, we have:

 2 2
c 1/3 c 2/3 3c 2/3 c 1/3 3c 1/3
x 2/3 + x 1/3 c 1/3 + c 2/3 = x 2/3 + 2 ⋅ x 1/3 ⋅ + + = x 1/3 + +
2 4 4 2 2
Since the sum of squares is always positive, we can write:

| x - c|
2
<𝜖
1/3 2 1/3
c 3c
x 1/3 + 2
+ 2
2
Moreover, since x 1/3 + c 1/3 / 2 ≥ 0, we can prove the stronger condition:

| x - c|
< 𝜖,  { c ≠ 0 }
3 2/3
4
c
3 2/3
Pick 𝛿 = c 𝜖. Then, |x - c| < 𝛿 implies that |g(x) - g(c)| < 𝜖.
4

Consequently, g(x) is continuous at c ≠ 0.

[Abbott 4.3.2.] To gain a deeper understanding of the relationship between 𝜖 and 𝛿 in the
definition of continuity, let's explore some modest variations of Definition 4.3.1. In all of
these, let f be a function defined on all of R.
(a) Let's say that f is onetinuous at c if for all 𝜖 > 0 we can choose 𝛿 = 1 and it follows
that |f(x) - f(c)| < 𝜖 whenever |x - c| < 𝛿. Find an example of a function that is
onetinuous on all of R. Find an example of a function that is onetinuous on all of R.
Proof.
Consider the constant function f(x) = k.
Pick an arbitrary 𝜖 > 0. Consider |f(x) - f(c)| < 𝜖. We have:

|k - k | < 𝜖

But, this is vacuously true, irrespective of the 𝜖-challenge. Hence, we can pick 𝛿 = 1
response.
Consequently, f(x) = k is onetinuous.
(b)Let's say that f is equaltinuous at c, if for all 𝜖 > 0 we can choose 𝛿 = 𝜖 and it follows
that |f(x) - f(c)| < 𝜖 whenever |x - c| < 𝛿. Find an example of a function that is
equaltinuous on R, that is no where onetinuous, or explain why there is no such function.
Proof.
Consider the linear function f(x) = x. Let c be an arbitrary point.
Claim. f(x) is equaltinuous.
Pick an arbitrary 𝜖 > 0. Consider |f(x) - f(c)| < 𝜖. We have:

| f ( x) - f ( c ) | < 𝜖
| x - c| < 𝜖
Pick 𝛿 = 𝜖. Then for all |x - c| < 𝛿, it follows that |f(x) - f(c)| < 𝜖.
Claim. f(x) is nowhere ontinuous.
1
Let 𝜖0 = . Let's explore the inequality |f(x) - f(c)| ≥ 𝜖0 . We have:
2
1
| x - c| ≥
2
Since Q is dense in R, we can pick x ∈ Q satisfying:
1 1
c - 1 < x ≤ c -  or  c + ≤ x < c + 1
2 2
Thus, |x - c| < 1.
So, there exists V𝜖 (f(c)) for V1 (c), such that for atleast some x ∈ V1 (c), it follows that
0

f(x) ∉ V𝜖 (f(c)).
0

There exists an 𝜖-challenge for which a 𝛿-response with 𝛿 = 1 is not suitable.

Since c was arbitrary, f is nowhere onetinuous.
(c) Let's say f is lesstinuous at c if for all 𝜖 > 0, we can choose 0 < 𝛿 < 𝜖 and it follows
that |f(x) - f(c)| < 𝜖 whenever |x - c| < 𝛿. Find an example of a function that is
lesstinuous on R, that is no where equaltinuous, or explain why there is no such function.
Proof.
Consider f(x) = 2x.
Claim. f(x) is lesstinuous.
Let's explore the inequality f(x) = 2x. We have:

| f ( x) - f ( c ) | < 𝜖
|2x - 2c| < 𝜖
𝜖
| x - c| <
2
If we pick 𝛿 = 𝜖 / 2, then for all x satisfying |x - c| < 𝛿, it follows that |f(x) - f(c)| < 𝜖.
Claim. f(x) is nowhere equaltinuous.
1
Let 𝜖0 = . Now, 𝛿 = 𝜖0 .
2

We can pick x belonging to the set:

(c - 𝛿, c - 𝜖0 / 2) ∪ (c + 𝜖0 / 2, c + 𝛿)

Clearly, |f(x) - f(c)| ≥ 𝜖0 . So, there exists 𝜖0 for 𝛿 = 𝜖0 , such that there for atleast some
x ∈ V𝛿 (c), it follows that |f(x) - f(c)| ≥ 𝜖0 .

There exists an 𝜖-challenge for which a 𝛿-response with 𝛿 = 𝜖0 is not suitable.

(d) Is every lesstinuous function continuous? Is every continuous function lesstinuous?
Explain.
Proof.
Every lesstinuous function is function, this follows from the definition of lesstinuous
functions.
All continuous functions are not lesstinuous.
Consider f(x) = x / 2. Pick an arbitrary 𝜖 > 0. Let's explore the inequality |f(x) - f(c)| < 𝜖.

| f ( x) - f ( c ) | < 𝜖
|x / 2 - c / 2| < 𝜖
|x - c| < 2𝜖
If we choose 𝛿 = 2𝜖, then for all x ∈ (c - 𝛿, c + 𝛿), it follows that f(x) ∈ (f(c) - 𝜖, f(c) + 𝜖).
Consequently, f is continuous but not lesstinuous.
[Abbott 4.3.3.] (a) Supply a proof for the theorem 4.3.9. using the 𝜖 - 𝛿 characterization
of continuity.
Proof.
We are given that f is continuous at c ∈ A and g is continuous at f(c) ∈ B.
We are interested to make the distance |g(f(x)) - g(f(c))| as small as we please.
Pick an arbitrary 𝜖 > 0. Let's explore the inequality |g(f(x)) - g(f(c))| < 𝜖.
Since g is continuous at f(c) ∈ f(A), there exists 𝜉 > 0, such that for all f(x) ∈ f(A)
satisfying |f(x) - f(c)| < 𝜉, it follows that |g(f(x) - g(f(c))| < 𝜖.
Since f is continuous at c, there exists 𝛿 > 0, such that for all x ∈ A, satisfying
|x - c| < 𝛿, it follows that |f(x) - f(c)| < 𝜉.

Since 𝜖 was arbitrary, this is true for all 𝜖 > 0.

Consequently, for all 𝜖 > 0, there exists 𝛿 > 0, such that for all x ∈ V𝛿 (c), it follows that
g(f(x)) ∈ V𝜖 (g(f(c)). Consequently, g(f(x)) is continuous at c.

(b) Give another proof of this theorem usng the sequential characterization of continuity.
Let (xn ) be an arbitrary sequence, with (xn ) ⊆ A, such that (xn ) → c.
Since f is continuous at c, it follows that the image sequence f(xn ) → f(c).
Now, f(xn ) ⊆ f(A). Since, g is continuous at f(c), the image sequence under g of f(xn ),
g(f(xn )) approaches g(f(c)).

Since (xn ) was an arbitrary sequence in A, this must be true for all sequences (xn ) ⊆ A,
with (xn ) → c.
Consequently, g(f(x)) is continuous at c.
[Abbott 4.3.4] Assume f and g are defined on all of R and that lim f(x) = q and
x→p
lim g(x) = r.
x→q

(a) Give an example to show that it may not be true that

lim g(f(x)) = r
x→p

Proof.
Define:

f ( x) = 0
and

g( x) =
x if x ≠ 0
1 if x = 0

We have xlim
→0
f(x) = 0 and lim g(x) = 0. But, lim g(f(x)) = lim g(0) = lim (1) = 1.
x→0 x→0 x→0 x→0

(b) Show that the results in (a) does follow if we assume that f and g are continuous.
Proof.
For all sequences (tn ), such that (tn ) → p, since f is continuous at p, the image sequence
f(t n ) → f(p). Consequently, q = f(p).
For all sequences (yn ), such that (yn ) → q, since g is continuous at q, g(yn ) → g(q).
Consequently, r = g(q).
Let (xn ) be an arbitrary sequence such that (xn ) → p.
Since f is continuous at p, the image sequence f(xn ) → f(p).
Since g is continuous at f(p), the image sequence g(f(xn )) → g(f(p)) = g(q) = r.
As (xn ) is arbitrary, this is true for all sequences (xn ) → p.
Consequently, xlim
→p
g(f(x)) = r.

(c) Does the result in (a) hold if we only assume f is continuous? How about if we only
assume that g is continuous?
Proof.
No, the result in (a) does not hold if we only assume f is continuous.
[Abbott 4.3.5] Show using Definition 4.3.1 that is c is an isolated point of A ⊆ R, then
f : A → R, then f : A → R is continuous at c.

Proof.
Since c is an isolated point, there exists 𝛿0 > 0, such that V𝛿 (c) ∩ A = { c }.
0

Pick an arbitrary 𝜖 > 0. For any given 𝜖-challenge, we always choose the above 𝛿0 as the
response.
For all x ∈ V𝛿 (c) (and x ∈ A), we must necessarily have x = c. Consequently, the distance
0

| f ( x) - f ( c ) | = | f ( c ) - f ( c ) | = 0 < 𝜖.

By definition 4.3.1, f is continuous at c.

[Abbott 4.3.6] Provide an example of each or explain why the request is impossible.
(a) Two functions f and g, neither of which is continuous at 0 but such that f(x)g(x) and
f(x) + g(x) are continuous at 0.

Consider

f ( x) =
x if x ≠ 0
1 if x = 0
and

1
g( x) =
if x ≠ 0
x
1 if x = 0
We have:

f ( x) g( x) = 1
Both f(x) and g(x) are not continuous at x = 0. But the product f(x)g(x) is the constant
function that maps all x → 1. So, f(x)g(x) is continuous at x = 0.
Further, consider

1
f ( x) =
if x ≠ 0
x
0 if x = 0

1
g( x) =
- if x ≠ 0
x
0 if x = 0
Again both f(x) and g(x) are not continuous at x = 0. But, the sum f(x) + g(x) is the
constant function that maps all x → 0. So, f(x) + g(x) is continuous at x = 0.
(b) A function f(x) continuous at 0 and g(x) not continuous at 0 such that f(x) + g(x) is
continuous at 0.
Proof.
This request is impossible.
Since g(x) = [f(x) + g(x)] - f(x), and f(x) + g(x) as well as f(x) are continuous, by the
Algebraic continuity theorem, g(x) must be continuous at 0.
(c) A function f(x) continuous at 0 and g(x) not continous at 0 such that f(x)g(x) is
continuous at 0.
Proof.
1
Consider f(x) = x and g(x) = . f(x) is continuous at c = 0, whilst g(x) is not continuous
x
at c = 0. The product f(x)g(x) = 1 is continuous at c = 0.
1
(d) A function f(x) not continuous at 0 such that f(x) + is continuous at 0.
f ( x)

Proof.
(e) A function f(x) not continuous at 0 such that [f(x)] 3 is continuous at 0.
This request is impossible.
Let g(x) = ∛x. [f(x)] 3 is continuous at x = 0. g(x) is continuous everywhere and therefore,
it is continuous at [f(0)] 3 .
The theorem 4.3.9 says that if f is continuous at c ∈ A and g is continuous at
f(c) ∈ f(A), then g(f(x)) is continuous at c.

So, g(f(x)) = ∛[f(x)] 3 = f(x) must be continuous at 0.

[Abbott 4.3.7] (a) Referring to the proper theorems, give a formal argument that
Dirichlet's function from section 4.1 is nowhere-continuous on R.
Proof.
The Dirichlet's function is defined as:
 f ( x) =
1 if x ∈ Q
0 if x ∉ Q

Let c ∈ Q be an arbitrary rational point.

1
Define (xn ) = c + . We have, (xn ) ⊆ Q, with (xn ) → c. The image sequence f(xn ) is the
n
constant sequence (1, 1, … ). Thus, f(xn ) → 1.

2
Define (yn ) = c + . We have (yn ) ⊆ I, with (yn ) → c. The image sequence f(yn ) is the
n
constant sequence (0, 0, 0, … ). Thus, f(yn ) → 0.
Consequently, lim f(xn ) ≠ lim f(yn ). So, f is not continuous at any rational point.
Let d ∈ I be an arbitrary irrational point.
Since Q is dense in R :
We can pick a rational number x1 , satisfying d - 1 < x1 < d + 1. We can pick the rational
1 1
number x2 satisfying d - < x2 < d + . In general, let xn ∈ Q, be such that,
2 2
1 1
d - < xn < d + .
n n

1
Pick an arbitrary 𝜖 > 0. If we pick N > , then for all n ≥ N, xn ∈ (d - 𝜖, d + 𝜖). Thus,
𝜖
(xn ) ⊆ Q with (xn ) → d.

The image sequence f(xn ) is the constant sequence (1, 1, 1, … ). So, f(xn ) → 1.

2
Define yn = d + . Since the irrationals are closed under addition, (yn ) ⊆ I and
n
(yn ) → d. The image sequence f(yn ) = (0, 0, 0, 0, … ). Thus, f(yn ) → 0.
Consequently, lim f(xn ) ≠ lim f(yn ). So, f is not continuous at any irrational point.
(b) Review the definition of Thomae's function in section 4.1 and demonstrate that it fails
to be continuous at every rational point.
Proof.
The Thomae's function t(x) is defined as:

1 if x = 0
1
t ( x) = if x = m ∈ Q - { 0 } is in the lowest terms with n > 0
n n
0 if x ∉ Q

2
Let c ∈ Q be an arbitrary rational point. Consider the sequence xn = c + . The image
n
sequence t(xn ) is the constant zero sequence (0, 0, 0, … ). So, t(xn ) → 0. But, t(c) ≠ 0.
(c) Use the characterization of continuity in Theorem 4.3.2 (iii) to show that Thomae's
function is continuous at every irrational point in R.
Proof.
Since the Thomae's function t(x) is periodic with a frequency 1, and repeats itself between
any two integers, it suffices to show that it is continuous at an irrational point c, c ∉ Q,
0 < c < 1.

We proceed by contradiction.
Assume that t(x) is discontinous at c. Now, t(c) = 0.
Carefully negating the definition of the continuity of a function, we find that, there exists an
𝜖0 > 0, for all 𝛿 > 0, such that for atleast some x satisying |x - c| < 𝛿, it follows that
t ( x) ≥ 𝜖 0 .
 1
By the Archimedean property, there exists N ∈ N, such that < 𝜖0 .
N

1
Since t(x) is a rational number of the form , n ∈ N, t(x) must belong to the set of finite
n
numbers:

1 1 1
t ( x) ∈ , , …, ,1
N-1 N-2 2
Thus, x must belong to the finite set S:

k
S= : k, l ∈ N, 1 ≤ l ≤ N - 1, k < l
l
But, we assumed that, for all 𝛿 > 0, there exists x satisfying |x - c| < 𝛿, such that
1
t(x) ≥ 𝜖0 . If we choose 𝛿n = , we can construct a sequence (xn ) → c.
n

However, x belongs to the finite set S. A sequence whose terms are the elements of a finite
set is either divergent or converges to an element from the set. Therein, lies our
contradiction.
[Abbott 4.3.8] Decide if the following claims are true or false, providing either a short
proof or counterexample to justify each conclusion. Assume throughout that g is defined
and continuous on all of R.
(a) If g(x) ≥ 0 for all x < 1, then g(1) ≥ 0 as well.
Proof.
Since g is continuous, by the sequential characterization of continuity, for all sequences
( x n ) → 1, g( x n ) → g( 1) .

Consider the sequence (an ) defined by:

1
an = 1 -
n
Since an < 1, g(an ) ≥ 0.
We know that, g(an ) → g(1). By the order limit theorem, lim g(an ) ≥ 0, so g(1) ≥ 0.
(b) If g(r) = 0 for all r ∈ Q, then g(x) = 0 for all x ∈ R.
Proof.
Let c be any arbitrary rational point. For all sequences (xn ) → c, g(xn ) → g(c).
1
Define an = c + . Since g(an ) is the constant zero sequence, g(an ) → 0. Thus, g(c) = 0.
n

Let d be an arbitrary irrational point. Since Q is dense in R, we can construct a rational

sequence (bn ) → d. Since g(bn ) is the constant zero sequence, g(bn ) → 0. Thus, g(d) = 0.
So, g(x) = 0 for all x ∈ R.
(c) If g(x0 ) > 0 for a single point x0 ∈ R, then g(x) is in fact strictly positive for
uncountably many points.
Since g(x) is continuous at x0 , for all 𝜖 > 0, there exists 𝛿 > 0, such that for all x satisfying
| x - x0 | < 𝛿, it follows that |g(x) - g(x0 )| < 𝜖.

g( x 0 )
Pick 𝜖 = > 0. Then, there exists 𝛿𝜖 , such that for all x satisfying |x - x0 | < 𝛿𝜖 , it
2
g( x ) 3g(x0 )
follows that 0 < g(x) < . Thus, g(x) is strictly positive.
2 2

Since, the interval (x0 - 𝛿𝜖 , x0 + 𝛿𝜖 ) consists of uncountably many points, and g(x) is
defined and everywhere continuous, g(x) is strictly positive for uncountably many points.
[Abbott 4.3.9] Assume that h : R → R is continuous on R and let K = { x : h(x) = 0 }.
Show that K is a closed set.
Proof.
Let a be a limit point of K. We are interested to prove that a ∈ K. By definition, there
exists (an ) ⊆ K, such that an ≠ a, with (an ) → a.
Since h is a continuous function, h(an ) → h(a). But, h(an ) = 0 for all n ∈ N. Consequently,
h(an ) → 0. Thus, h(a) = 0. Consequently, a ∈ K.

Since a was arbitrary, this is true of all the limit points of K.

So, K is closed.
[Abbott 4.3.10] Observe that if a and b are real numbers, then

1
max{ a, b } = [(a + b) + |a - b|]
2
(a) Show that if f1 , f2 , … , fn are continuous functions, then

g(x) = max{ f1 (x), f2 (x), … , fn (x)}

is a continuous function.
Proof.
We proceed by mathematical induction. First let's prove that g2 (x) = max{ f1 (x), f2 (x)} is
continuous for the case n = 2. We have:

1
g 2 ( x) = [ f 1 ( x) + f 2 ( x) + | f 1 ( x) - f 2 ( x) | ]
2
By the Algebraic continuity theorem, if f1 (x) and f2 (x) are continuous functions,
f1 (x) + f2 (x) is also continuous.

Let's prove that h(x) = |f1 (x) - f2 (x)| is also continuous. I use a direct argument to prove
this. However, notice that h(x) = l(f1 (x) - f2 (x)) where l(x) = |x|, and since |x| is
continuous, and the composition of continuous functions is continuous, it follows that
|f1 (x) - f2 (x)| is continuous.

Nevertheless, suppose we are interested to make the distance

||f1 (x) - f2 (x)| - |f1 (c) - f2 (c)| as small as we please.

Let's explore the inequality:

 ||f1 (x) - f2 (x)| - |f1 (c) - f2 (c)| <𝜖
Replacing ||f1 (x) - f2 (x)| - |f1 (c) - f2 (c)| by its upper bound will strengthen the condition
we wish to prove.
Note that:

||a| - |b|| ≤ |a - b|

Short proof.

| a| = | a - b + b |
≤ | a - b| + | b|
| a| - | b | ≤ | a - b |
And

| b | = | b - a + a|
≤ | a - b | + | a|
| b | - | a| ≤ | a - b |
Consequently,

- | a - b | ≤ | a| - | b | ≤ | a - b |

Thus, the distance ||a| - |b|| ≤ |a - b|.

Consequently,
||f1 (x) - f2 (x)| - |f1 (c) - f2 (c)| ≤ |f1 (x) - f2 (x) - (f1 (c) - f2 (c))|
= |f1 (x) - f1 (c) - (f2 (x) - f2 (c))|
≤ | f 1 ( x) - f 1 ( c ) | + | f 2 ( x) - f 2 ( c ) |
Thus, our claim is :
| f 1 ( x) - f 1 ( c ) | + | f 2 ( x) - f 2 ( c ) | < 𝜖

There exists 𝛿1 > 0, such that for all x satisfying |x - c| < 𝛿1 , it follows that
| f 1 ( x) - f 1 ( c ) | < 𝜖 / 2.
There exists 𝛿2 > 0, such that for all x satisfying |x - c| < 𝛿2 , it follows that
| f 2 ( x) - f 2 ( c ) | < 𝜖 / 2.

Pick 𝛿 = min{ 𝛿1 , 𝛿2 }. Then, for all x satisfying |x - c| < 𝛿, it follows that:

||f1 (x) - f2 (x)| - |f1 (c) - f2 (c)| ≤ |f1 (x) - f1 (c)| + |f2 (x) - f2 (c)|
𝜖 𝜖
< + =𝜖
2 2
So, |f1 (x) - f2 (x)| is continuous at c. Since c was arbitrary, |f1 (x) - f2 (x)| is continuous
on R.
Thus, g2 (x) is continuous.
Assume that gn-1 (x) = max{ f1 (x), … , fn-1 (x)} is continuous.
We are interested to prove that gn (x) is continuous. Because,

gn (x) = max{ gn-1 (x), fn (x)}

and both gn-1 and fn are continuous, we can argue as above that, gn is continuous. By the
principle of mathematical induction, this is true for all n ∈ N.
(b) Let's explore whether the result in (a) extends to the infinite case. For each n ∈ N,
define fn on R by:

f n ( x) =
1 if |x| ≥ 1 / n
n | x| if |x| < 1 / n

Now explicitly compute h(x) = sup{ f1 (x), f2 (x), f3 (x), … }.

Proof.
Note that, all fn (x) are continuous on R.
We are only interested in the behavior of these functions and h(x) in [-1, 1], since
fn (x) = 1 for all |x| ≥ 1.
Let (xn ) ⊆ (0, 1] be an arbitrary sequence such that (xn ) → 0.
Since 0 < xn < 1, by the Archimedean property, for all n ∈ N, there exists M ∈ N, such
1
that < xn . By definition, fM+1 (xn ) = 1 and in fact, fm (xn ) = 1 for all m ≥ M. So,
M
h(xn ) = 1. Consequently, h(xn ) is the constant sequence (1, 1, 1, … ). Thus, h(xn ) → 1.

Since (xn ) was arbitrary, this is true for all sequences (xn ) → 0. Thus, the functional limit
lim h(x) = 1. But, h(0) = 0. Therefore, h has a jump discontinuity at c = 0.
x→0

Hence, the result in (a) does not extend to the infinite case.
[Abbott 4.3.11] (Contraction Mapping Theorem.) Let f be a function defined on all of R,
and assume that there is a constant c such that 0 < c < 1 and

| f ( x) - f ( y) | ≤ c | x - y|

for all x, y ∈ R.
(a) Show that: f is continuous on R.
Proof.
Let x0 be an arbitrary but fixed point. Our claim is that f is continuous at x0 . We would
like to make the distance |f(x) - f(x0 )| as small as we please.
Pick an arbitrary 𝜖 > 0.
We would like to prove that:

| f ( x) - f ( x 0 ) | < 𝜖

But, |f(x) - f(x0 )| ≤ c|x - x0 |. Replacing |f(x) - f(x0 )| by its upper bound strengthens
the inequality we wish to prove.
Our claim therefore is:
 𝜖
|x - x0 | < ,  { 0 < c < 1 }
c
𝜖
Pick 𝛿 = . Then for all x satisfying |x - x0 | < 𝛿, it follows that
c

| f ( x) - f ( x 0 ) | ≤ c | x - x 0 | < c ⋅ ( 𝜖 / c ) = 𝜖
Since x0 was arbitrary, this must be true of all x ∈ R. So, f is continuous on R.
(b) Pick some point y1 ∈ R and construct the sequence

( y 1 , f( y 1 ) , f( f( y 1 ) , … )

In general, if yn+1 = f(yn ), show that the resulting sequence (yn ) is a Cauchy sequence.
Hence, we may let y = lim yn .
Proof.
Consider the distance |yn - ym |. We are interested to make this distance as small as
possible.
Let's explore the expresson |yn - ym |. We have:

|yn - ym | = |yn - yn-1 + yn-1 - yn-2 +…+ ym+1 - ym |

≤ |yn - yn-1 | + |yn-1 - yn-2 | +…+ |ym+1 - ym |
= |f(yn-1 ) - f(yn-2 )| + |f(yn-2 ) - f(yn-3 )| +…+ |f(ym ) - f(ym-1 )|
≤ c|yn-1 - yn-2 | + c|yn-2 - yn-3 | +…+ c|ym - ym-1 |
≤ c n-2 |y2 - y1 | + c n-3 |y2 - y1 | +…+ c m-1 |y2 - y1 |
= c m-1 |y2 - y1 | 1 + c + c 2 +…+ c (n-m+1)
c m-1
≤ |y 2 - y 1 | since { |c| < 1 }
1-c
Pick an arbitrary 𝜖 > 0. We are interested to prove that |yn - ym | < 𝜖. Replacing |yn - ym |
by its upper bound strengthens the condition we wish to prove.
Hence our claim is:
 c m-1
⋅ |y 2 - y 1 | < 𝜖
1-c
or

𝜖( 1 - c )
c m-1 <
|y 2 - y 1 |
Taking log on both sides:

𝜖( 1 - c )
(m - 1)log c <
|y 2 - y 1 |

that is,

𝜖( 1 - c )
( m - 1) > ,  since log c < 0
|y2 - y1 |log c
𝜖( 1 - c )
If we pick M > 1 + , then for all n > m ≥ M, it follows that |yn - ym | < 𝜖.
|y2 - y1 |log c
Consequently, (yn ) is a Cauchy sequence.
(c) Prove that y is a fixed point of f (that is f(y) = y) and that it is unique in this regard.
Proof.
We have: y = lim yn . Since f is a continuous function, f(yn ) → f(y). But, f(yn ) = yn+1 .
So, f(yn ) → y. Consequently, f(y) = y.
(d) Finally, prove that if x is any arbitrary point in R, then the sequence
(x, f(x), f(f(x)), … ) converges to y defined in (b).

Proof.
Let x be an arbitrary point in R. Let (xn ) be the sequence given by:

(xn ) = (x, f(x), f(f(x)), … )

Consider the distance |xn - y|. We would like to make this distance as small as we please.
We have:

|xn - y| = |f(xn-1 ) - f(y)|

≤ c|xn-1 - y|
= c|f(xn-2 ) - f(y)|
≤ c 2 |xn-2 - y|
≤ c n-1 |x - y|
Given any arbitrary 𝜖-challenge, we can pick N such that,

c N-1 |x - y| < 𝜖
𝜖
that is, c N-1 <
| x - y|
𝜖
or N-1 >
|x - y|log c
It follows that |xn - y| < 𝜖. Consequently, lim xn = y.
[Abbott 4.3.12] Let F ⊆ R be a nonempty closed set and define
g(x) = inf{ |x - a| : a ∈ F }. Show that g is continuous on all of R and g(x) ≠ 0 for all x ∉ F.

Proof.
[Abbott 4.4.1] (a) Show that f(x) = x 3 is continuous on all of R.
Proof.
Let c be an arbitrary point in R. We are interested to make the distance |f(x) - f(c)| as
small as we please.
Pick an arbitrary 𝜖 > 0.
Case I. c = 0.
Since f(c) = 0, if we pick 𝛿 = ∛𝜖, then for all |x| < 𝛿, we have |f(x)| < 𝜖.
Case II. c ≠ 0.
We have:

| f ( x) - f ( c ) | = | x 3 - c 3 |
= |x - c||x 2 + cx + c 2 |
2
2
c 3
= | x - c| x + + c
2 2
2
2
c 3c
= | x - c| x + +
2 2
If we assume that 𝛿 < 1, then x ∈ (c - 1, c + 1). So, x ≤ c + 1. Therefore,
2 2 2
2
c 3c 3c + 2 3c
x+ + ≤ +
2 2 2 2

Hence, we try to prove the stronger condition:

2 2
3c + 2 3c
| x - c| + ≤𝜖
2 2

2
Since c ≠ 0, the expression {(3c + 2) / 2) 2 + 3c / 2 } is strictly positive. Hence, we can
pick

𝜖
𝛿 = min 1, 2
2
3c+2 3c
2
+ 2

Then, for all x satisfying |x - c| < 𝛿, it follows that |f(x) - f(c)| < 𝜖.
Since c was arbitrary, f is continuous on all of R.
(b) Argue, using theorem 4.4.5, that f is not uniformly continuous on R.
Proof.
1 1
Consider the sequence xn = n + and yn = n. Let 𝜖0 = 1. Then, (xn - yn ) = and
n n
(xn - yn ) = 0. But,

3
1
|f( x n ) - f( y n ) | = n+ - n3
n
1 1 1
= n 3 + 3n 2 ⋅ + 3n ⋅ 2 + 3 - n 3
n n n
3 1
= 3n + +
n n3
3 1
= 3n + +
n n3
≥ 3n
≥ 𝜖0
Consequently, by the sequential criterion for the absence of uniform continuity, f is not
uniformly continuous.
(c) Show that f is uniformly continuous on any bounded subset of R.
Proof.
Let S ⊆ R be any bounded subset of R. Since S is bounded, there exists M > 0 for all
x ∈ S, such that |x| ≤ M.

We are interested to make the distance |f(x) - f(y)| as small as we please. Pick an
arbitrary 𝜖 > 0.
 | f ( x) - f ( y) | = | x 3 - y 3 |
≤ |x - y||x 2 + xy + y 2 |
≤ |x - y||x 2 | + |x||y| + |y 2 |
≤ | x - y| M 2 + M 2 + M 2
= |x - y| ⋅ 3M 2
𝜖
Pick 𝛿 = . Then, for all x, y ∈ S satisfying |x - y| < 𝛿, it follows that
3M 2
| f ( x) - f ( y) | < 𝜖.

[Abbott 4.4.2] (a) Is f(x) = 1 / x uniformly continuous on (0, 1)?

Proof.
1 1
Let xn = 2 and yn = 2 , with n ≥ 2. Both (xn ) and (yn ) are contained in (0, 1). Let
n n +n
𝜖0 = 1. Then (xn - yn ) → 0. Let's explore the expression |f(xn ) - f(yn )|.

|f( x n ) - f( y n ) | = n 2 + n - n 2
=n
≥ 1 = 𝜖0
Consequently, by the sequential criterion for the absence of uniform continuity, f is not
uniformly continuous on (0, 1).

(b) Is g(x) = x2 + 1 uniformly continuous on (0, 1)?

We are interested to make the distance |g(x) - g(y)| as small as we please. Pick an
arbitrary 𝜖 > 0.
 | g( x) - g( y) | = | x 2 + 1 - y 2 + 1|
x2 + 1 + y2 + 1
= x2 + 1 - y2 + 1 ×
x2 + 1 + y2 + 1
| x2 + 1 - y2 + 1 |
=
x2 + 1 + y2 + 1
x2 - y2
=
x2 + 1 + y2 + 1
| x - y || x + y |
=
x2 + 1 + y2 + 1
|x - y| ⋅ 2
< {0 < x < 1}
1+1 {0 < y < 1}
= |x - y|
Therefore, we can try to prove the stronger condition |x - y| < 𝜖. Pick 𝛿 = 𝜖. Then, for all
x, y ∈ (0, 1) satisfying |x - y| < 𝛿, it follows that |f(x) - f(y)| < 𝜖.

(c) Is h(x) = x sin(1 / x) uniformly continuous on (0, 1)?

Proof.
f ( x) = x is continuous on (0, 1). We can show that g(x) = sin(1 / x) is continuous on
(0, 1).

Claim. sin x is continuous on (0, 1]

We can write:
 x+c x-c
|sin x - sin c| ≤ 2 cos sin
2 2
x-c
≤ 2 sin since |cos x| ≤ 1
2
3 5

=2
x-c
-
1 x-c
+
1 x-c
-…
Taylor's series
2 3! 2 5! 2 expansion

1 1
Pick 𝛿 < . Then |x - c| < implies that
2 2

4 2
x-c x-c
<
2 2
8 6
x-c x-c
<
2 2
Thus, the term inside the curly brackets is strictly non-negative and has a lower bound 0.
Also Therefore, we can try to prove the stronger condition

x-c
|sin x - sin c| ≤ 2
2
= | x - c|
1
Pick 𝛿 = min ,𝜖 . Then, for all x satisfying |x - c| < 𝛿, it follows that
2
|sin x - sin c| < 𝜖. Consequently, sin x is continuous on (0, 1].

1
Since is continuous on (0, 1] and the composition of continuous functions is continuous,
x
provided they are well-defined, sin(1 / x) is continuous on (0, 1].
By Algebraic continuity theorem, x sin(1 / x) is continuous on (0, 1].
We define a new function:
 1
h( x) =
x sin if x ≠ 0
x
0 if x = 0
Claim. h(x) is continuous on [0, 1].
We already know that h(x) is continuous on (0, 1]. Since, |x sin(1 / x)| ≤ |x|, if we choose
𝛿 = 𝜖, for all |x| < 𝛿, it follows that |h(x)| < 𝜖. Consequently, h(x) is continuous at 0.

So, h(x) is continuous on [0, 1].

By the theorem 4.4.7, a function that is continuous on a compact set K is uniformly
continuous on K. Consequently, h is uniformly continuous on [0, 1]. Thus, h is uniformly
continuous on (0, 1).
[Abbott 4.4.3] Show that f(x) = 1 / x 2 is uniformly continuous on the set [1, ∞) but not
on the set (0, 1].
Proof.
We are interested to make the distance |f(x) - f(y)| as small as we please.
We have:
 1 1
| f ( x) - f ( y) | = -
2
x y2
x2 - y2
=
x2y2
| x - y | | x + y|
=
x2y2
| x - y| | x + y|
= ⋅
xy xy
x+y
≤ | x - y| { x, y ≥ 1 }
xy
1 1
= | x - y| +
x y
1 1
≤ | x - y| + { x, y ≥ 1 }
1 1
= 2|x - y|
𝜖
Let 𝜖 > 0 be arbitrary. If we pick 𝛿 = , then for all x, y ∈ [1, ∞) satisfying |x - y| < 𝛿, it
2
follows that
𝜖
|f(x) - f(y)| ≤ 2|x - y| < 2 ⋅ =𝜖
2
Consequently, f is uniformly continuous on [1, ∞).
1 1
Moreover, let xn = 2 and yn = 2 . Then, (xn ) and (yn ) are contained in (0, 1] Let
n n +n
𝜖0 = 1. Then, (xn - yn ) → 0, but

2
|f( x n ) - f( y n ) | = n 2 + n - n 4
= n 4 + 2n 3 + n 2 - n 4
= 2n 3 + n 2
≥ 𝜖0
Consequently, f is not uniformly continuous on (0, 1].
[Abbott 4.4.4] Decide whether each of the following statements is true or false, justifying
each conclusion.
(a) If f is continuous on [a, b] with f(x) > 0 for all a ≤ x ≤ b, then 1 / f is bounded on
[a, b] (meaning 1 / f has bounded range).

Since K = [a, b] is a compact set, by the extreme value theorem, there exists x0 , x1 ∈ [a, b]
such that f(x0 ) ≤ f(x) ≤ f(x1 ) for all x ∈ [a, b]. Further, since f(x0 ) and f(x1 ) belong to
f([a, b]), we must have f(x0 ) ≠ 0 and f(x1 ) ≠ 0. Consequently, it follows that:

1 1 1
≤ ≤
f ( x 1 ) f ( x) f ( x 0 )
for all x ∈ [a, b].
Thus, 1 / f is bounded.
(b) If f is uniformly continuous on a bounded set A, then f(A) is bounded.
Proof.
We are given that f is uniformly continuous on a bounded set A.
We proceed by contradiction. Assume that f(A) is unbounded. So, for all M ∈ N, there
exists f(x) ∈ f(A), such that |f(x)| > N. Pick N = 1, 2, 3, … . Then, there exists N ∈ N
such that (f(xn )) → ∞.
Since (xn ) ⊆ A and A is bounded, by the Bolzanno Weierstrass Theorem, there exists a
convergent subsequence (xn ) ⊆ (xn ). Let lim xn = x. Since, f is uniformly continuous on
k k

A, it implies that f is continuous on A. Therefore, for all sequences (an ) → a contained in

A , f ( a n ) → f ( a) .

Thus, f(xn ) → f(x). Pick 𝜖 = 1. Then, there exists K > 0, such

k

Alternative proof.
Consider the function h(x) defined on the closure of A, cl(A) as :
 limx → a f(x) where if a is a limit point of A
h( x) =
f ( x) if x ∈ A

We would like prove that (i) h is well-defined on cl(A) (ii) h is continuous on cl(A).
Claim. h is well-defined on any limit point of A.
Let a be an arbitrary limit point of A. Let (xn ) be any arbitrary sequence that converges to
a. (We know that there is atleast one such sequence.)

Pick an arbitrary 𝜖 > 0.

Since f is uniformly continuous on A, there exists 𝛿(𝜖) > 0, such that for all x, y ∈ A
satisfying |x - y| < 𝛿, it follows that |f(x) - f(y)| < 𝜖.
Since (xn ) is a Cauchy sequence, there exists N ∈ N, such that for all n > m ≥ N, it
follows that |xn - xm | < 𝛿.
Consequently, for all n > m ≥ N, it follows |f(xn ) - f(xm )| < 𝜖.
Since 𝜖 is arbitrary, this is true for all 𝜖 > 0. Consequently, (f(xn )) is a Cauchy sequence
and therefore it is convergent. Thus, lim f(xn ) exists.
Since (xn ) was arbitrary, this must be true of all sequences (xn ) → a. So, lim f(x) exists.
x→a
Hence, h is well-defined at all limit points.
Claim. h is continuous on cl(A).
Let c be an arbitrary limit point of A. We proceed by contradiction. Assume that h is not
continuous at c. Then, there exists a sequence (xn ) ⊆ cl(A) such that (xn ) → c but
lim h(xn ) ≠ h(c).

If (xn ) ⊆ A, then since h is continuous on A, h(xn ) → h(c) and we have the desired
contradiction.
(c) If f is defined on R and f(K) is compact whenever K is compact, then f is continuous
on R.
Proof.
This proposition is false. Consider the function f defined as:

3 / 2x if 0 ≤ x ≤ 1 / 3
f ( x) = 0 if 1 / 3 < x < 2 / 3 or x ∉ [0, 1]
3 / 2( x - 2 / 3) if 2 / 3 ≤ x ≤ 1

Since f is continuous on [0, 1 / 3] and [1 / 3, 2 / 3] we know that f([0, 1 / 3] and f[2 / 3, 1])
is compact. The same logic applies if you take any compact subset of the union of these
sets. Moreover, f maps (1 / 3, 2 / 3) to 0. Thus, whenever K is compact f(K) is compact,
but f is not continuous.
[Abbott 4.4.5] Assume that g is defined on an open interval (a, c) and it is known to be
uniformly continuous on (a, b] and [b, c) where a < b < c. Prove that g is uniformly
continuous on (a, c).
Direct Proof.
Pick an arbitrary 𝜖 > 0.
1. There exists 𝛿1 (𝜖) > 0, such that for all x, y ∈ (a, b] satisfying |x - y| < 𝛿1 , it follows
that |g(x) - g(y)| < 𝜖 / 2.
2. There exists 𝛿2 (𝜖) > 0, such that for all x, y ∈ (a, b] satisfying |x - y| < 𝛿2 , it follows
that |g(x) - g(y)| < 𝜖 / 2.
Let 𝛿 = min{ 𝛿1 , 𝛿2 }.
Let x, y ∈ (a, c) be any two arbitrary points satisfying |x - y| < 𝛿.
If x, y ∈ (a, b], then (1) applies and we are done.
If x, y ∈ [b, c), then (2) applies and we are done.
If x ∈ (a, b] and y ∈ [b, c), then |x - y| < 𝛿 implies that:
a) |x - b| < 𝛿
b) |b - y| < 𝛿
Consequently, it follows that |g(x) - g(b)| < 𝜖 / 2 and |g(b) - g(y)| < 𝜖 / 2. Now,

| g( x) - g( y) | = | g( x) - g( b ) + g( b ) - g( y) |
≤ | g( x) - g( b ) | + | g( b ) - g( y) |
𝜖 𝜖
= + =𝜖
2 2
Since x, y were arbitrary points in (a, c), satisfying |x - y| < 𝛿, this is true for all of such
points.
Consequently, g is uniformly continuous on (a, c).
Alternative Proof.
We extend g(x) to the points x = a and x = c. Define the function:

lim g(x)
x→a
if x = a
h( x) = g( x) if x ∈ (a, c)
lim g(x)
x→c
if x = c
Let's prove that h is well-defined and continuous on [a, c].
Let (xn ) be an arbitrary sequence in (a, c) approaching a.
Pick an arbitrary 𝜖 > 0.
Since g is uniformly continuous on (a, c), there exists 𝛿(𝜖), such that for all x, y ∈ (a, c)
satisfying |x - y| < 𝛿, it follows that |g(x) - g(y)| < 𝜖.
Since (xn ) is Cauchy, there exists N ∈ N, such that for all n > m ≥ N, we have

|x n - x m | < 𝛿
Consequently, for all n > m ≥ N, the terms of the image sequence (g(xn )) satisfy:
 | g( x n ) - g( x m ) | < 𝜖
Thus, (g(xn )) is Cauchy and therefore lim g(xn ) exists.
Since, (xn ) was an arbitrary sequence, this must be true of all sequences (xn ) → a. So,
lim g(x) exists.
x→a

Next, we prove that h is continuous at a.

Let's proceed by contradiction. Assume that h is not continuous at a. Then, there exists a
sequence (xn ) ⊆ dom(h) such that (xn ) → a, but h(xn ) does not converge to h(a).
1. If (xn ) is the constant sequence (a, a, a, … ), then h(xn ) → h(a). This is a contradiction.
2. If (xn ) has an infinite number of terms equal to c, then (xn ) would not converge to a.
So, such a possibility is ruled out.
3. The only other possibility is that (i) (xn ) ⊆ (a, c) or (ii) (xn ) has a subsequence
(xn ) → a contained in (a, c).
k

But, by construction h(a) = lim g(x). For all sequences (an ) ⊆ (a, c) such that (an ) → a,
x→a
h(an ) = g(an ) approaches h(a).

Thus, h(xn ) → h(a) (for both cases 3(i) and 3(ii)). This is a contradiction.
Hence, our initial assumption is false. h is continuous at a.
We can similarly argue that h is well-defined and continuous at c.
Thus, h is continuous on the compact set [a, c]. So, h is uniformly continuous on [a, c].
Consequently, g is uniformly continuous on (a, c).
[Abbott 4.4.6] Give an example of each of the following, or state that such a request is
impossible. For any that are impossible, supply a short explanation for why this is the case.
(a) A comtinuous function f : (0, 1) → R and a Cauchy sequence (xn ) such that f(xn ) is
not a Cauchy sequence.
Proof.
Consider the function f defined as:

f ( x) =
0 if x = 0
1/x if x ∈ (0, 1]
1
Consider the sequence xn = , with n ≥ 2. (xn ) ⊆ (0, 1). (xn ) → 0 so (xn ) is a Cauchy
n
sequence, but f(xn ) = n, so f(xn ) is unbounded and therefore not Cauchy.
(b) A uniformly continuous function f : (0, 1) → R and a Cauchy sequence (xn ) such that
f(xn ) is not a Cauchy sequence.

Proof.
This request is impossible.
Assume that f is uniformly continuous on (0, 1).
Pick an arbitrary 𝜖 > 0.
Since f is uniformly continuous over (0, 1), there exists 𝛿(𝜖) > 0, such that for all
x, y ∈ (0, 1) satisfying |x - y| < 𝛿, it follows that |f(x) - f(y)| < 𝜖.

Let (xn ) be an arbitrary Cauchy sequence in (0, 1). Since (xn ) is Cauchy, there exists
n > m ≥ N, such that |xn - xm | < 𝛿.

Thus, for all n > m ≥ N, it follows that |f(xn ) - f(xm )| < 𝜖.

Consequently, f(xn ) is Cauchy.
(c) A continuous function f : [0, ∞) → R and a Cauchy sequence (xn ) such that f(xn ) is
not a Cauchy sequence.
This request is impossible.
Let (xn ) be any arbitrary Cauchy sequence in [0, ∞). Let a = lim xn . Since, f is continuous
on [0, ∞), f(xn ) → f(a). Consequently, f(xn ) is Cauchy.

[Abbott 4.4.7] Prove that f(x) = x is uniformly continuous on [0, ∞).

Proof.

The function f(x) = x is continuous on [0, 1]. A continuous function on a compact set K
is uniformly continuous. So, f is uniformly continuous on [0, 1].

Also, the function x is Lipschitz on [1, ∞), since, for all x ≠ y ∈ [1, ∞):

f ( x) - f ( y) x- y
=
x-y x-y
1
=
x+ y
1
≤
2
Since Lipschitz continuity implies uniform continuity, f is uniformly continuous on [1, ∞).
By the previous exercise, f is uniformly continuous on the whole half-line from 0.
[Abbott 4.4.8] Give an example of each of the following, or provide a short argument for
why the request is impossible.
(a) A continuous function defined on [0, 1] with the range (0, 1).
Proof.
This request is impossible.
By the property on the preservation of compact sets, if f is continuous function on a
compact set K, then f(K) is compact. So, f([0, 1]) must be compact.
(b) A continuous function defined on (0, 1) with range [0, 1].
Proof.
Define

0 if 0 < x < 1 / 4
f ( x) = 2x if 1 / 4 ≤ x ≤ 3 / 4
0 if 3 / 4 ≤ x < 1

(c) A continuous function defined on (0, 1] with range (0, 1).

Proof.
This request is impossible. Consider c > 0. The image of
[Abbott 4.4.9] (Lipschitz Functions) A function f : A → R is called Lipschitz if there
exists a bound M > 0 such that:

f ( x) - f ( y)
≤M
x-y
for all x ≠ y ∈ A. Geometrically speaking, a function f is Lipschitz if there is a uniform
bound on the magnitude of the slopes of lines drawn through any two points on the graph
of f.
(a) Show that if f : A → R is Lipschitz, then it is uniformly continuous on A.
Proof.
Pick an arbitrary 𝜖 > 0.

| f ( x) - f ( y) | ≤ M | x - y|

𝜖
Pick 𝛿 = . Then, for all x ≠ y ∈ A satisfying |x - y| < 𝛿, it follows that
M
| f ( x) - f ( y) | < 𝜖.

Consequently, f is uniformly continuous on A.

(b) Is the converse statement true? Are all uniformly continuous functions necessarily
Lipschitz?
Proof.
The converse statement is false.
Counterexample.

Consider f(x) = x. f is uniformly continuous on [0, 1].

We are interested to make the slope of the secant lines as large as we please. Let M > 0 be
an arbitrary large number.
1
If we pick y = 0 and x <
( M + 1) 2

f ( x) - f ( y) 1
=
x-y x+ y
> M+1
Consequently, ∀M > 0, there exists x ≠ y ∈ A, such that

f ( x) - f ( y)
>M
x-y
Hence, f is not Lipschitz on [0, 1].
[Abbott 4.4.10] Assume that f and g are uniformly continuous functions defined on a
common domain A. Which of the following combinations are uniformly continuous on A:

f ( x)
f(x) + g(x),  f(x)g(x),   ,   f(g(x))
g( x)
(Assume that the quotient and the composition are properly defined and atleast continuous)
Proof.
(1) We are interested to make the distance |f(x) + g(x) - (f(y) + g(y))| as small as we
please.
Pick an arbitrary 𝜖 > 0.
There exists 𝛿1 (𝜖) > 0, such that for all x, y ∈ A satisfying |x - y| < 𝛿1 , it follows that
| f ( x) - f ( y) | < 𝜖 / 2.

There exists 𝛿2 (𝜖) > 0, such that for all x, y ∈ A satisfying |x - y| < 𝛿2 , it follows that
| g( x) - g( y) | < 𝜖 / 2.

Let 𝛿 = min{ 𝛿1 , 𝛿2 }. Then, for all x, y ∈ A satisfying |x - y| < 𝛿, it follows that:

|f(x) + g(x) - (f(y) + g(y))| ≤ |f(x) - f(y)| + |g(x) - g(y)|

𝜖 𝜖
< + =𝜖
2 2
Consequently, f + g is uniformly continuous.
(2) Consider f(x) = x and g(x) = x and let A = [0, ∞). Both f and g are uniformly
continuous on [0, ∞).
Let 𝜖 > 0 be arbitrary. Pick 𝛿 = 𝜖. Then, for all x, y ∈ A satisfying |x - y| < 𝛿, it follows
that |f(x) - f(y)| < 𝜖.
fg(x) = f(x)g(x) = x 2is not uniformly continuous on [0, ∞). Let 𝜖0 = 1. Consider the two
1
sequences xn = n and yn = n + . |xn - yn | → 0 but |f(xn ) - f(yn )| ≥ 𝜖0 .
n

(3) Consider the constant function f(x) = 1 and g(x) = x. Both f(x) and g(x) are
f ( x) 1
uniformly continuous on (0, 1). But, = is not uniformly continuous on (0, 1).
g( x) x

(4) We are interested to make the distance |f(g(x)) - f(g(y))| as small as we please.
Pick an arbitrary 𝜖 > 0.
There exists 𝜉(𝜖), such that for all s, t ∈ g(A), satisfying |s - t| < 𝜉, it follows that
| f ( s ) - f ( t ) | < 𝜖.

There exists 𝛿(𝜉), such that for all x, y ∈ A, satisfying |x - y| < 𝛿, it follows that
| g( x) - g( y) | < 𝜉 .

Hence, for all x, y ∈ A, |x - y| < 𝛿 implies that |g(x) - g(y)| < 𝜉 which in turn implies that
|f(g(x)) - f(g(y))| < 𝜖.

Consequently, f(g(x)) is uniformly continuous on A.

[Abbott 4.4.11] (Topological characterization of continuity). Let g be defined on all of
R. If B is a subset of R, define the set g -1 (B) by:

g -1 (B) = { x ∈ R : g(x) ∈ B }

Show that g is continuous if and only if g -1 (O) is open whenever O ⊆ R is an open set.
Proof.
⟹ direction.
Assume that g is continuous. Let O be an arbitrary open set. We are interested to prove
that g -1 (O) is open.
Let c be an arbitrary point in g -1 (O). Thus, g(c) ∈ O. Since O is an open set, ∃𝜖 > 0,
such that V𝜖 (g(c)) ⊆ O.
Since g is well-defined on R and continuous at c, ∃V𝛿 (c), such that for all x ∈ V𝛿 (c), it
follows that g(x) ∈ V𝜖 (g(c)) ⊆ O.
But, we know that for all g(x) ∈ O, x ∈ g -1 (O). Therefore, V𝛿 (c) ⊆ g -1 (O).
So, there exists a 𝛿-neighbourhood around the point c, such that V𝛿 (c) ⊆ g -1 (O). Since, c
was arbitrary point, this must be true of all points in the set. Thus, g -1 (O) is open.
⟸ direction.
We are given that, for all open sets O ⊆ R, the pre-image g -1 (O) is open. We are
interested to prove that g is continuous.
Let c be an arbitrary point in dom(g). g(c) is the image of c under g.
Pick an arbitrary 𝜖 > 0 and consider the 𝜖 - neighbourhood around g(c):

V𝜖 (g(c)) = (g(c) - 𝜖, g(c) + 𝜖)

Since V𝜖 (g(c)) is an open interval, it is an open set. Let's refer to this set as O. We know,
that if O ⊆ R is open, then g -1 (O) is open. Hence,

g -1 (O) = { x : g(x) ∈ V𝜖 (g(c))}

is open.
Now, c ∈ g -1 (O), since g(c) ∈ V𝜖 (g(c)). Thus, there exists 𝛿(𝜖, c), such that
V𝛿 (c) ⊆ g -1 (O).

Hence, we have found that there exists V𝛿(𝜖,c) (c) such that for all x ∈ V𝛿 (c),
g(x) ∈ V𝜖 (g(c)). Since 𝜖 was arbitrary, this is true for all 𝜖 > 0. Consequently, g is
continuous at c ∈ R.
Since c was an arbitrary point, g is continuous on R.
[Abbott 4.4.12] Review exercise 4.4.11, and then determine which of the following
statements is true about a continuous function defined on R:
(a) f -1 (B) is finite whenever B is finite.
Proof.
This is false. Consider the constant function f(x) = 1. Let B = { 1 }. f -1 (B) = R.
(b) f -1 (K) is compact whenever K is compact.
Proof.
This is false.
Consider f(x) = 1. Then, K = { 1 } which is compact, but f -1 (K) = R is unbounded and
therefore not compact.
(c) f -1 (A) is bounded whenever A is bounded.
This is false.
(d) f -1 (F) is closed whenever F is closed.
This is true.
Let x be arbitrary limit point of f -1 (F). There exists a sequence (xn ) ⊆ f -1 (F), such that
lim xn = x. By construction, it follows that f(xn ) ∈ F. As f is continuous at x,
lim f(xn ) = f(x). Since F is closed, f(x) ∈ F. Thus, x ∈ f -1 (F).

Since x was abitrary, this must be true for all limit points of f -1 (F). Thus, f -1 (F) is closed.
[Abbott 4.4.13] (Continuous Extension Theorem.)
(a) Show that a uniformly continuous function preserves Cauchy sequences; that is, if
f : A → R is uniformly continuous and (xn ) ⊆ A is a Cauchy sequence, then show f(xn ) is
a Cauchy sequence.
Proof.
Pick an arbitrary 𝜖 > 0.
There exists 𝛿 > 0, such that for all x, y ∈ A satisfying |x - y| < 𝛿, it follows that
| f ( x) - f ( y) | < 𝜖.

Let (xn ) be an arbitrary Cauchy sequence. Then, there exists N(𝛿), such that for all
n > m ≥ N, it follows that |xn - xm | < 𝛿. But, this implies that |f(xn ) - f(xm )| < 𝜖 for all
n > m ≥ N. Consequently, (f(xn )) is a Cauchy sequence.
(b) Let g be a continuous function on the open interval (a, b). Prove that g is uniformly
continous on (a, b) if and only if it is possible to define values g(a) and g(b) at the
endpoints s that the extended function g is continuous on [a, b]. (In the forward direction,
first produce candidates for g(a) and g(b) and then show the extended g is continuous.)
Proof.
We define :

lim g(x)
x→a
if x = a
h( x) = g( x) if x ∈ (a, b)
lim g(x)
x→b
if x = b
as the extended version of g.
Pick an arbitrary 𝜖 > 0.
Since g is uniformly continuous on (a, b), there exists 𝛿(𝜖) > 0, such that for all
x, y ∈ (a, b) satisfying |x - y| < 𝛿, it follows that |g(x) - g(y)| < 𝜖.

a is a limit point of (a, b). Consider any arbitrary sequence (an ) ⊆ (a, b), such that
(an ) → a. Since (an ) is a Cauchy sequence, there exists N(𝛿) ∈ N, such that for all
n > m ≥ N, we have |xn - xm | < 𝛿.

But, this implies that, for all n > m ≥ N, the distance |g(xn ) - g(xm )| < 𝜖. Since (an ) was
an arbitrary sequence approaching a, this must be true of all sequences approaching a.
Thus, xlim
→a
g(x) exists.

Claim. The extended version of g is continuous.

We proceed by contradiction.
Assume that g is not continuous. Then, there exists a sequence (an ) ⊆ dom(g) such that,
(an ) → a, but lim g(an ) ≠ g(a).

If an = a for any n ≥ N, N ∈ N, it follows that the tail of the image sequence (g(an )) is
the constant sequence (g(a)) which approaches g(a). This is a contradiction.
If (an ) consists of an infinite number of terms different from a, then there exists a
subseqeuence (an ) ⊆ (an ), with an ∈ (a, b), that is an ≠ a such that lim an = a.
k k k k

Pick an arbitrary 𝜖 > 0. There exists 𝛿(𝜖) > 0, such that for all x, y ∈ (a, b) satisfying
|x - y| < 𝛿, it follows that |g(x) - g(y)| < 𝜖.

Since lim an = a, there exists K ∈ N, such that for all nk > K, |an
k k
- a| < 𝛿 . But, this
implies that for all nk > K, |g(an ) - g(a)| < 𝜖.
k

Since 𝜖 was arbitrary, this must be true for all 𝜖 > 0. Consequently, lim g(an ) = g(a). This
k

is again a contradiction.
Hence, our initial assumption is false. g is continuous on [a, b]. Since, g is continuous on a
compact set [a, b], g is uniformly continuous on [a, b]. Consequently, g is uniformly
continuous on (a, b).
[Abbott 4.5.1] Show how the Intemediate Value Theorem follows as a corollary to
Theorem 4.5.2.
Proof.
Let f : [a, b] → R be a continuous function. Define E = (a, b). Since, E is a connected set
and f is continuous, by the preservation of connected sets property, f(E) is connected.
Thus, for all L satisfying f(a) < L < f(B), L ∈ f(E). But, this implies that there exists
c ∈ E = (a, b) such that f(c) = L.

[Abbott 4.5.2] Provide an example of each of the following, or explain why the request is
impossible.
(a) A continuous function defined on an open interval with the range equal to the closed
interval.
Proof.
Let f : (0, 1) → R with range [0, 1] be the continuous function defined as:
 0 0 < x < 1/4
f ( x) = 2( x - 1 / 4) 1 / 4 ≤ x ≤ 3 / 4
1 3/4 ≤ x < 1

(b) A continuous function defined on a closed interval with range equal to an open interval.
Proof.
This request is impossible. Let f : [a, b] → R be a continuous function. Since [a, b] is closed
and bounded, it is compact. Since f is continuous on f, by the property on the preservation
of compact sets, the range f([a, b]) is compact - hence it is closed and bounded.
Moreover, since f is continuous on [a, b], by the property on the preservation of connected
sets, f([a, b]) is connected and it is therefore an interval in R. Consequently, f([a, b]) is a
closed interval.
(c) A continuous function defined on an open interval with range equal to an unbounded
closed set different from R.
Proof.
Consider f : (0, 1) → R defined as :

1 1
0<x≤
f ( x) = x 2
1
2 ≤x<1
2
The range of f is the unbounded closed set [2, ∞). f is continuous on (0, 1).
(d) A continuous function defined on all of R with range equal to Q.
Let f be continuous on all of R. Then, by the theorem on the preservation of connected
sets, since R is connected, f(R) must be connected.

But, Q is not a connected set. For example, consider A = -∞, 2 ∩ Q and

B= 2, ∞ ∩ Q. Then, A ∪ B = Q. And the cl(A) will have all its limit points lesser
than or equal to 2. Thus, cl(A) ∩ B = ∅ and A ∩ cl(B) = ∅. Hence Q is disconnected.

Consequently, f cannot be continuous.

[Abbott 4.5.3] A function f is increasing on A if f(x) ≤ f(y) for all x < y in A. Show
that if f is increasing on [a, b] and satisfies the intermediate value property (definition
4.5.3), then f is continuous on [a, b].
Proof.
Let c be an arbitrary point in [a, b]. We are interested to prove that f is continuous at c.
Let (xn ) be any arbitrary sequence approaching c.

[Abbott 4.5.5] (a) Finish the proof of the Intermediate Value Theorem using the Axiom of
Completeness started previously.
Intermediate Value Theorem. Let f be a continuous function on a closed and bounded
interval [a, b]. f : [a, b] → R. If L is a real number satisfying f(a) < L < f(b), then there
exists c ∈ (a, b), such that f(c) = L.
Proof.
To simplify matters, we first considered the special case satisfying f(a) < 0 < f(b) and
show that f(c) = 0 for some c ∈ (a, b).

We let

K = { x ∈ [ a, b ] : f ( x ) ≤ 0 }

Since K is bounded by [a, b], by the Axiom of completeness, it has a supremum. Let
c = sup K.

By the trichotomy property of the reals, exactly one of the following holds:
1) f(c) < 0
2) f(c) > 0
3) f(c) = 0
(i) Assume that f(c) < 0. Then, by construction c ∈ K.
But, c = sup K. So, it is an upper bound for K. Thus, f(c) ≥ 0. This is a contradiction.
Hence our assumption is false.
(ii) Assume that f(c) > 0. Then, by construction c ∈ [a, b] - K. Since the complementation
of a closed set is open, [a, b] - K is an open set. So, there exists V𝜉 (c) such that
V 𝜉 ( c ) ⊆ [ a, b ] - K C .

For all c - 𝜉 < x < c, we have x ∉ K. But, c = sup K. So, there exists x0 ∈ K, such that
c - 𝜉 < x0 < c. This is a contradiction. Hence our assumption is false.

Thus, we are left with f(c) = 0.

(b) Finish the proof of the Intermediate Value Theorem using the Nested Interval Property
started previously.
Proof.
∞
By the Nested Interval Property, there exists x ∈ ⋂ In . Clearly, x ∈ [a, b].
n=1

Moreover, lim an = lim bn = x. Since f is continuous at x, f(an ) → f(x). Clearly,

f(an ) < 0 by construction, and thus by the order limit theorem f(x) ≤ 0. Also, f(bn ) > 0
by construction, and thus by the order limit theorem, f(x) ≥ 0. Since, both of these are
simultaneously true, it must be the case that f(x) = 0.
Also, x cannot be any of the end-points, for otherwise f(a) = 0 or f(b) = 0. This
contradicts the fact that f(a) < 0 < f(b).
So, there exists x ∈ (a, b) such that f(x) = 0.
[Abbott 5.2.1] Supply proofs for parts (i) and (ii) of theorem 5.2.4.
Proof.
(i) We are interested to prove that (f + g)'(c) = f'(c) + g'(c).
By definition,

f(x) + g(x) - (f(c) + g(c))

(f + g)'(c) = lim
x→c x-c

We can write the difference quotient as :

f(x) + g(x) - (f(c) + g(c)) f ( x) - f ( c ) g( x) - g( c )

= +
x-c x-c x-c

Now, by the algebraic limit theorem for functional limits,

f ( x) - f ( c ) g( x) - g( c ) f ( x) - f ( c ) g( x) - g( c )
lim + = lim + lim = f'(c) + g'(c)
x→c x-c x-c x→c x-c x→c x-c

Consequently,

(f + g)'(c) = f'(c) + g'(c)

(ii) We are interested to prove that (kf(c))' = kf'(c).

By definition,

kf(x) - kf(c) f ( x) - f ( c )
(kf(c))' = lim = k lim = kf'(c)
x→c x-c x→c x-c

[Abbott 5.2.2] Exactly one of the following requests is impossible. Decide which it is, and
provide examples for the other three. In each case, let's assume the functions are defined on
all of R.
(a) Functions f and g not differentiable at zero but where fg is differentiable at zero.
Proof.
Consider f(x) = |x| and

g( x) =
x sin(1 / x) if x ≠ 0
0 if x = 0

Both f and g are not differentiable at zero. Now consider:

f ( x) g( x) - f ( 0) g( 0)
(fg)'(0) = lim x → 0
x
|x| ⋅ x sin(1 / x)
= lim x → 0
x
= lim x → 0|x| ⋅ sin(1 / x)
Both the left hand and the right hand limits exists and are equal. Hence, the above limit
exists and fg is differentiable at x = 0.
(b) A function f not differentiable at zero and a function g differentiable at zero where fg
is differentiable at zero.
Proof.
Consider f(x) = sin(1 / x) and g(x) = x 2 . f(x)g(x) = x 2 sin(1 / x) is differentiable at
c = 0, since

x 2 sin(1 / x) - 0
(fg)'(0) = lim = lim x sin 1 / x = 0
x→0 x x→0

(c) A function f not differentiable at zero and a function g differentiable at zero where
f + g is differentiable at zero.

Proof.
This request is impossible.
 f'(c) = ((f + g) - g)'(c)
= (f + g)'(c) - g'(c) Algebraic Differentiability theorem
Thus, if f + g is differentiable at zero and g is differentiable at zero, f must be
differentiable at zero.
(d) A function f differentiable at zero but not differentiable at any other point.
Proof.
Consider the function:

f ( x) =
x2 if x ∈ Q
0 if x ∉ Q
We have:

f ( x) - f ( 0)
f'(0) = lim
x→0 x-0
We are interested to show that this functional limit exists. Our claim is that f'(0) = 0.
Pick an arbitrary 𝜖 > 0. Let's explore the expression:

f ( x) - f ( 0) f ( x) - f ( 0) | f ( x) |
-0 = =
x-0 x | x|
Since f(x) ≥ 0 for all x ∈ R, we have |f(x)| = f(x) ≤ x 2 = |x 2 |. Consequently,

| x| < 𝜖

Pick 𝛿 = 𝜖. For all |x| < 𝛿, it follows that |x| < 𝜖. Consequently, f'(0) = 0.
Let's prove that f(x) is not continuous for c ≠ 0.
c2
Let c ∈ Q. Consider the sequences xn = c + 1 and yn = c + 1
with 𝜖0 = .
n 2n 2
|xn - yn | → 0. But,
1 1 2c 1 c2
|f( x n ) - f( y n ) | = c2 +2⋅c⋅ + 2
-0 = c + + ≥ = 𝜖0
n n2 n n2 2
Thus, f is not continuous at any rational point c.
Moreover, let t ∈ I be any irrational point. By the density of rationals in R, there exists a
1
rational sequence (xn ) ∈ Q, such that xn ≠ t with (xn ) → t. Also, let (yn ) = t + be an
n
irrational sequence converging to t. Again, we can similarly, prove that f is not continuous
at an irrational point.
Since, continuity is necessary condition for differentiability, f is not differentiable at c ≠ 0.
[Abbott 5.2.3] (a) Use definition 5.2.1 to produce the proper formula for the derivative of
h( x) = 1 / x.

Proof.
Let c be an arbitrary point, with c ≠ 0. We are interested to find h'(c).
By definition 5.2.1.,

h( x) - h( c ) ( 1 / x) - ( 1 / c )
h'(c) = lim x → c = lim x → c
x-c x-c
( x - c) 1
= lim x → c - ⋅
(x - c) cx
( x - c) 1
= -lim x → c ⋅ {x ≠ c}
(x - c) cx

= -lim x → c
1 1
= - 2
By algebgraic limit theorem
cx c for functional limits
(b) Combine the result in part (a) with the Chain rule (theorem 5.2.5) to supply a proof for
part (iv) of theorem 5.2.4.
Proof.
 1 1
Let w(t) = . Then, w(v(x)) = . By the Chain Rule:
t v( x)

'
1 v'(c)
[w(v(c))]' = = w'(v(c)) ⋅ v'(c) = -
v( c ) (v(c)) 2

By the product rule:

' '
1 1 1
u( c ) ⋅ = u'(c) ⋅ + ⋅ u( c )
v( c ) v( c ) v( c )
u'(c) v'(c)u(c)
= -
v( c ) v( c ) 2
v(c)u'(c) - v'(c)u(c)
=
v( c ) 2
(c) Supply a direct proof of the theorem 5.2.4 (iv) by algebraically manipulating the
difference quotient for (f / g) in a style similar to the proof of theorem 5.2.4 (iii).
Proof.
By definition:

' f (x ) f (c )
f ( c) g (x )
- g (c )
= lim x → c
g( c ) x-c
f ( x) g( c ) - f ( c ) g( x)
= lim x → c
g(x)g(c)(x - c)
f ( x) g( c ) - f ( c ) g( c ) + f ( c ) g( c ) - f ( c ) g( x)
= lim x → c
g(x)g(c)(x - c)
-[f(c)g(x) - f(c)g(c)] + [f(x)g(c) - f(c)g(c)] 1
= lim x → c ⋅
( x - c) g( x) g( c )
1 -[g(x) - g(c)] f ( x) - f ( c ) 1
= lim x → c f(c) ⋅ + lim x → c g(c) ⋅ ⋅ lim x → c
g( c ) ( x - c) x-c g( x)
 1
By the Algebraic continuity theorem, if g is continuous at c and g(c) ≠ 0, then is
g( x)
continuous at c. Thus,

'
f ( c) 1 1
= { -f(c)g'(c) + g(c)f'(c)} ⋅
g( c ) g( c ) g( c )
g(c)f'(c) - f(c)g'(c)
=
(g(c)) 2
[Abbott 5.2.4] Follow these steps to provide a slightly modified proof of the Chain Rule.
(a) Show that a function h : A → R is differentiable at a ∈ A if and only if there exists a
function l : A → R which is continuous at a and satisfies:

h(x) - h(a) = l(x)(x - a) for all x ∈ A

Proof.
⟹ direction.

We are given that h is differentiable at a ∈ A. By definition,

h ( x ) - h ( a)
h'(a) = lim
x→a ( x - a)

Since, x ≠ a, we can define a new function l(x) to be the difference quotient:

h ( x ) - h ( a)
l ( x) =
if x ∈ A, x ≠ a
,
x-a
h'(a) if x = a
Since h(x) is defined for all x ∈ A, l(x) is defined for all A.
h ( x ) - h ( a)
Since xlim l(x) = lim = h'(a) = l(a), l is continuous at a.
→a x→a x-a

Hence, there exists a function l : A → R which is continuous and satisfies:

 h(x) - h(a) = l(x)(x - a),  ∀x ∈ a
⟸ direction.
We are given that the function l(x) satisfies:

h(x) - h(a) = l(x)(x - a),  ∀x ∈ A

where h is a function from A into R.

We are interested to prove that h is differentiable at a.

Since, l is continuous at a, xlim
→a
l ( x ) = l ( a) .
If x ≠ a:

h ( x ) - h ( a)
l ( x) =
x-a
Consequently,

h ( x ) - h ( a)
lim l(x) = l(a) = lim
x→a x→a x-a
Therefore, the limit on the right hand side of the above expression exists and h'(a) = l(a).
Consequently, h is differentiable at a ∈ A.
(b) Use this criterion for differentiability (in both directions) to prove theorem 5.2.5.
Proof.
We are interested to prove that [g(f(c))]' = g'(f(c)) ⋅ f'(c).
Since f is differentiable at c ∈ A, there exists a function l(x) for all x ∈ A, such that:

f(x) - f(c) = l(x)(x - c)

Notice that, if x ≠ c,
 f ( x) - f ( c )
l ( x) =
x-c

The differentiability of f at c ⟹ the continuity of l at c. So, passing to the limits, we

have:

f ( x) - f ( c )
lim l(x) = l(c) = lim = f'(c)
x→c x→c x-c

Since g is differentiable at f(c) ∈ f(A) ⊆ B, there exists a function m(y) for all y ∈ f(A),
such that:

g(y) - g(f(c)) = m(y)(y - f(c))

If y ≠ f(c),

g(y) - g(f(c))
m ( y) =
y - f ( c)

The differentiability of g at f(c) implies the continuity of m at f(c). So, passing to the
limits, we have:

g(y) - g(f(c))
lim m(y) = m(f(c)) = lim = g'(f(c))
y → f(c ) y → f(c ) y - f ( c)

Since, y ∈ f(A), y must be the image of x under f.

By Theorem 4.3.9, if f is continuous at c and m is continuous at f(c), then m(f( ⋅ )) is
continuous at c. The composition of continuous functions is continuous.
Consider the product of the functions:

m(f(x)) ⋅ f(x)

By the Algebraic Continuity theorem, if m(f( ⋅ )) is continuous at c and l is continuous at c,

then the product m(f(x)) ⋅ l(x) is also continuous at c.
By the definition of continuity:

lim x → c m(f(x)) ⋅ l(x) = m(f(c)) ⋅ l(c)

= g'(f(c)) ⋅ f'(c)
On the other hand, we can write:

g(f(x)) - g(f(c)) f ( x) - f ( c )
lim x → c m(f(x)) ⋅ l(x) = lim x → c ⋅ lim x → c
f ( x) - f ( c ) x-c
g(f(x)) - g(f(c))
= lim x → c
x-c
= [g(f(c))]'
This completes the proof.
[Abbott 5.2.5] Let

f a ( x) =
if x > 0
xa
if x ≤ 0
0
(a) For which values of a, is f continuous at zero?
Proof.
Case I. Let a be any real number, where a > 0.
We are interested to make the distance |fa (x)| as small as we please. We have:

|x a | < 𝜖
⟺ | x| a < 𝜖
⟺ |x| < 𝜖 1/a

Pick 𝛿 = 𝜖 1/a . Then, for all |x - 0| < 𝛿, it follows that |fa (x) - fa (0)| < 𝜖.
So, f is continuous for all a > 0.
Let's prove that f is not continuous for a ≤ 0.
Case II. Let a = 0. Then,

f 0 ( x) =
1 if x > 0
0 if x ≤ 0
f0 has a jump discontinuity at x = 0.
Case III. Consider f-a (x) where a > 0.

1
f-a (x) =
if x > 0
xa
0 if x ≤ 0
Pick 𝜖0 = 1. Let 𝛿 > 0 be arbitrary.
1
By the Archimedean property, there exists N ∈ N, such that < 𝛿.
N

1
Pick an arbitrary x ∈ 0, . Thus, |x| < 𝛿.
N

Now,
1 1
> N ⟹ a > N a ,  { N ≥ 1 }
x x
Therefore, there exists atleast some x satisfying |x| < 𝛿, such that

|f-a (x) - f-a (0)| ≥ 𝜖0

Since 𝛿 was arbitrary, it follows that: there exists 𝜖0 > 0, for all 𝛿 > 0, such that for atleast
some x satisfying |x| < 𝛿, we have |f-a (x) - f-a (0)| ≥ 0.
Thus, f-a (x) is discontinuous at x = 0.
(b) For which values of a is f differentiable at zero? In this case, is the derivative function
continuous?
Using the definition of the derivative, we have:
 f a ( x) - f a ( 0)
f'a (0) = lim
x→0 x-0
The right hand limit in the above case is:

f a ( x) - f a ( 0) xa
lim+ = lim+ = lim+ x a-1
x→0 x-0 x→0 x x→0

The left hand limit in the above case is:

f a ( x) - f a ( 0) 0
lim = lim =0
x → 0- x-0 x → 0- x

As seen earlier, if x > 0, then lim x a-1 = 0 if and only if a - 1 > 0. Consequently, fa (x) is
x→0
differentiable for all values of a > 1.
(c) For which values of a is f twice differentiable?
We know that,

f a ' ( x) =
ax a-1 if x > 0
0 if x ≤ 0

Using the definition of the derivative, we have:

f a ' ( x) - f a ' ( 0)
f''a (x) = lim
x→0 x-0
The right-hand limit in the above case is:

ax a-1 - 0
lim = lim ax a-2
x → 0+ x x → 0+

The left-hand limit in the above case is:

 0
lim =0
x → 0- x

If x > 0, then lim ax a-2 = 0 if and only if a - 2 > 0. Consequently, fa (x) is twice
x→0
differentiable for all values of a > 2.
[Abbott 5.2.6] Let g be defined on an interval A, and let c ∈ A.
(a) Explain why g'(c) in the definition 5.2.1 could have been given by:

g( c + h) - g( c )
g'(c) = lim
h→0 h

Proof.
We know that g is a function of x. Let x be a function f of the single-variable h.

x = f ( h) = c + h

Then, we can write g as:

g(x) = g(f(h))

When x = c, h = 0. By the chain rule:

g'(c) = [g(f(0))]' = g'(f(0)) ⋅ f'(0)

g(f(h)) - g(f(0)) f ( h) - f ( 0)
= lim c+h → c ⋅ lim h → 0
f ( h) - f ( 0) h-0
g(f(h)) - g(f(0)) f ( h) - f ( 0)
= lim h → 0 ⋅ lim h → 0
f ( h) - f ( 0) h-0
   since lim h = lim (c + h) - lim c
g(f(h) - g(f(0))
= lim h → 0
h
g( c + h) - g( c )
= lim h → 0
h
(b) Assume that A is open. If g is differentiable at c ∈ A, show that :

g( c + h) - g( c - h)
g'(c) = lim
h→0 2h
Proof.
We introduce two intermediate variables t = t(h) = c + h and u = u(h) = c - h.
We may write:

g(t) = g(t(h)) = g(c + h)

and

g(u) = g(u(h)) = g(c - h)

When t = c, h = 0. Also, when u = c, we have h = 0. Thus,

 1
g'(c) = [g'(c) + g'(c)]
2
1
= [{ g(t(0)}' + { g(u(0)}']
2
1
= [g'(t(0)) ⋅ t'(0) + g'(u(0)) ⋅ u'(0)]
2
1 g(t(h)) - g(t(0))
= [lim t(h) → t(0) ⋅1
2 t ( h) - t ( 0)
g(u(h)) - g(u(0))
+lim u(h) → u(0) ⋅ (-1)]
u ( h) - u ( 0)
1 g(t(h)) - g(t(0))
= [lim h → 0
2 ( c + h) - c
g(u(h)) - g(u(0))
-lim h → 0 ]
( c - h) - c
1 g( c + h) - g( c ) g( c - h) - g( c )
= lim h → 0 - lim h → 0
2 h-0 -h
1 g( c + h) - g( c ) g( c ) - g( c - h)
= lim h → 0 - lim h → 0
2 h h

[Abbott 5.2.7] Let

g a ( x) =
if x ≠ 0
x a sin(1 / x)
0 if x = 0
Find a particular (potentially non-integer) value for a so that:
(a) ga is differentiable on R, but such that ga ' is unbounded on [0, 1].
Proof.
Consider a = 3 / 2. We have:
At points different from zero, we can use the familiar rules of differentiation to find:
 1
g3/2 '(x) = (3 / 2)x 1/2 sin(1 / x) - cos(1 / x)
x
Let's investigate if g3/2 is differentiable at zero. By definition:

g3/2 (x) - g3/2 (0) x 3/2 sin(1 / x)

g3/2 '(0) = lim = lim = lim x sin(1 / x)
x→0 x-0 x→0 x x→0

Let's find out the above functional limit. We are interested to make the distance
| x sin(1 / x)| as small as we please. Our claim is that:

| x sin(1 / x)| < 𝜖

If we replace |sin(1 / x)| by its upper-bound, we strengthen the condition we wish to prove.
Since |sin(1 / x)| < 1, we shall try to prove that:

| x| < 𝜖

Squaring on both sides:

2
| x| 2 = | x | = | x| < 𝜖 2

Pick 𝛿 = 𝜖 2 . Then, for all x ∈ [0, ∞) satisfying |x - 0| < 𝛿, it follows that

| x ⋅ sin(1 / x)| < 𝜖 ⋅ 1 = 𝜖

Consequently, xlim
→0
x ⋅ sin(1 / x) = 0. Thus:

1
(3 / 2)x 1/2 sin(1 / x) - cos(1 / x) if x ≠ 0
g3/2 '(x) = x
0 if x = 0
Moreover, g3/2 '(x) is bounded on [0, 1] because of the factor 1 / x.

(b) ga is differentiable on R with ga ' continuous but not differentiable at zero.

Consider a = 5 / 2. At points different from zero, we can use the familiar rules of
differentiation to find:

1
g5/2 '(x) = (5 / 2)x 3/2 sin(1 / x) - x 5/2 cos(1 / x) ⋅ -
x2
= (5 / 2)x 3/2 sin(1 / x) - x cos(1 / x)

Consider the point c = 0. By definition:

g5/2 (x) - g5/2 (0)

g5/2 '(0) = lim x → 0
x-0
5/2 ⋅ sin(1 / x)
x
= lim x → 0 = lim x → 0 x 3/2 sin(1 / x)
x
Let's find out the above functional limit.
We are interested to make the distance |x 3/2 sin(1 / x)| as small as we please. Let's explore
the condition:

|x 3/2 sin(1 / x)| < 𝜖

If we replace the the quantity |sin 1 / x| by its upper bound, we strengthen the condition we
wish to prove.

|x 3/2 | < 𝜖

That is:

|x| < 𝜖 2/3

Pick 𝛿 = 𝜖 2/3 . Then, for all x ∈ [0, ∞) satisfying |x| < 𝛿, it follows that:

|x 3/2 sin(1 / x)| < 𝜖

Consequently, lim x 3/2 sin(1 / x) = 0.
x→0

Thus, g5/2 (x) is differentiable on R. Moreover:

g5/2 '(x) = (5 / 2)x 3/2 sin(1 / x) - x cos(1 / x) if x ≠ 0

0 if x = 0
Let's prove that g5/2 '(x) is continuous at zero.

We know that, both (5 / 2)x 3/2 sin(1 / x) and x cos(1 / x) are continuous at zero. By the
Algebraic continuity theorem, their algebraic difference is also continuous at zero.
We are interested to prove that g5/2 '(x) is not differentiable at zero. By definition, we have:

g5/2 '(x) - g5/2 '(0)

g5/2 ''(0) = lim x → 0
x-0
(5 / 2)x 3/2 sin(1 / x) - x 1/2 cos(1 / x)
= lim x → 0
x
1
= lim x → 0(5 / 2)x 1/2 sin(1 / x) - ⋅ cos(1 / x)
x
1 1
Now, let (xn ) = and (yn ) = be two sequences in [0, 1]. Clearly,
2n𝜋 (2n + 1)𝜋
lim xn = lim yn = 0, but lim f(xn ) ≠ lim f(yn ). Consequently, the functional limit
1
lim cos(1 / x) does not exist. g5/2 '(x) is not differentiable at zero.
x→0
x

(c) ga is differentiable on R and ga ' is differentiable on R, but such that ga '' is not
continuous at zero.
Consider a = 7 / 2. At points different from zero, we can apply the familiar rules of
differentiation to find:

g7/2 '(x) = x 5/2 sin(1 / x) - x 3/2 cos(1 / x)

Let's find if g7/2 (x) is differentiable at zero. We have:

 g7/2 (x) - g7/2 (0)
g7/2 '(0) = lim = lim x 3/2 sin(1 / x) - x 1/2 cos(1 / x)
x→0 x-0 x→0

We have:

-x 3/2 ≤ x 3/2 sin(1 / x) ≤ x 3/2

for all x ∈ R. By the Squeeze theorem, since lim -x 3/2 = lim x 3/2 = 0, it follows that
x→0 x→0
lim x sin(1 / x) = 0. Similarly, lim x cos(1 / x) = 0.
3/2 1/2
x→0 x→0

Thus, by the Algebraic Limit theorem,

g7/2 '(0) = lim x 3/2 sin(1 / x) - lim x 1/2 cos(1 / x) = 0

x→0 x→0

Consequently, g7/2 (x) is differentiable on R.

We have:

g7/2 '(x) =
if x ≠ 0
x 5/2 sin(1 / x) - x 3/2 cos(1 / x)
0 if x = 0
From the earlier discussion in part (b), g7/2 '(x) is differentiable on R, but g7/2 ''(x) is not
continuous at zero.
[Abbott 5.2.8] Review the definition of uniform continuity (Definition 4.4.4). Given a
differentiable function f : A → R, let's say that f is uniformly differentiable on A, if, given
𝜖 > 0, there exists a 𝛿 > 0 such that:

f ( x) - f ( y)
- f'(y) < 𝜖 whenever 0 < |x - y| < 𝛿
x-y
(a) Is f(x) = x 2 uniformly differentiable on R? How about g(x) = x 3 ?
Proof.
f ( x) - f ( y)
Let's explore the expression - f'(y) . We have:
x-y
 f ( x) - f ( y) x2 - y2
- f'(y) = - 2y
x-y x-y
= |x - y|

Pick 𝛿 = 𝜖. Then, for all x, y ∈ A satisfying |x - y| < 𝛿, it follows

f ( x) - f ( y)
- f'(y) < 𝜖
x-y
Consequently, f(x) = x 2 is uniformly differentiable on R.

(b) Show that if a function is uniformly differentiable on an interval A, then the derivative
must be continuous on A.
Proof.
Since the function f is uniformly differentiable on A, for all 𝜖 > 0, there exists 𝛿 > 0, such
that for all x, y ∈ A, satisfying |x - y| < 𝛿, it follows that

f ( x) - f ( y)
- f'(y) < 𝜖
x-y

Let c be an arbitrary fixed point in the interval A. Therefore, for all 𝜖 > 0, there exists
𝛿 > 0, such that for all x satisfying |y - c| < 𝛿, we must have:

f ( y) - f ( c )
- f'(y) < 𝜖
y-c
By definition, it means that:

f ( y) - f ( c )
lim f'(y) - =0
y→c y-c
f ( y) - f ( c )
Since f'(c) = lim , the limit lim f'(y) exists and further:
y→c y-c y→c
 lim f'(y) = f'(c)
y→c

Consequently, f'(y) is continuous at y = c.

(c) Is there a theorem analogous to theorem 4.4.7 for differentiation? Are functions that are
differentiable on a closed interval [a, b] necessarily uniformly differentiable?
The contrapositive of part (b) is : if the derivative function f'(x) is not continuous on A, it
is not uniformly differentiable. Thus, f(x) = x sin(1 / x) is differentiable on [0, 1] but it is
not uniformly differentiable, since f'(x) is not continuous at 0.
[Abbott 5.2.9] Decide whether each conjecture is true or false. Provide an argument for
those that are true and a counterexample for each one that is false.
(a) If f' exists on an interval and is not constant, then f' must take on some irrational
values.
Proof.
Suppose f : [a, b] → R and f' is not constant. Then, f'(a) ≠ f'(b). There exists an
irrational number 𝛼 ∈ I between any two reals. So, there exists 𝛼 ∈ I, such that
f'(a) < 𝛼 < f'(b) or f'(a) > 𝛼 > f'(b).

Since f is differentiable on [a, b], there exists c ∈ (a, b), such that f'(c) = 𝛼. Thus, f' must
take on some irrational value.
(b) If f' exists on an open interval and there is some point c where f'(c) > 0, then there
exists a 𝛿-neighbourhood V𝛿 (c) in which f'(x) > 0 for all x ∈ V𝛿 (c).
Proof.
This proposition is false.
Consider the function:
 x 1
f ( x) =
+ x 2 sin if x ≠ 0
2 x
0 if x = 0
The graph of f(x) is:
0.2

0.1

-0.5 -0.4 -0.3 -0.2 -0.1 0 0.1 0.2 0.3 0.4 0.5 0.6

-0.1

-0.2

-0.3

The derivative of this function is:

1 1 1
+ 2x sin - cos if x ≠ 0
f'(x) = 2 x x
1
if x = 0
2

The graph of the derivative function f'(x) is:

 1

-2 -1 0 1 2 3

1 1 1 1 1 1 3
Consider xn = , , , , … . f' = - 1 / 2 whilst f' = .
𝜋 2𝜋 3𝜋 4𝜋 (2n + 1)𝜋 2n𝜋 2
Thus, if we take any arbitrary 𝛿-neighbourhood of the point zero, V𝛿 (0), we will find both
positive and negative values of f'(x).
(c) If f is differentiable on an interval containing zero and if xlim
→0
f'(x) = L, then it must
be that L = f'(0).
[Abbott 5.2.10] Recall that a function f : (a. b) → R is increasing on (a, b) if f(x) ≤ f(y)
whenever x < y in (a, b). A familiar mantra from Calculus is that a differentiable function is
increasing if its derivative is positive, but this statement requires some sharpeneing in order
to be completely accurate.
Show that the function:

g( x) =
if x ≠ 0
x / 2 + x 2 sin(1 / x)
0 if x = 0
is differentiable on R and satisfies g'(0) > 0. Now prove that g is not increasing over any
open interval containing 0.
Proof.
For all other points different from zero, we can differentiate g(x) using the familiar rules of
differentiation to find:

1 1 1
g'(x) = + 2x sin - cos ,  { x ≠ 0 }
2 x x
By definition,

(x / 2) + x 2 sin(1 / x) 1 1 1
g'(0) = lim = lim + x sin =
x→0 x x→0 2 x 2

Consider the sequence

1 1 1 1
, , , ,…
3𝜋 / 2 5𝜋 / 2 7𝜋 / 2 9𝜋 / 2

We find that the image sequence has the following order relation:

1 1 1 1
g <g ,g <g ,…
3𝜋 / 2 5𝜋 / 2 7𝜋 / 2 9𝜋 / 2

Since the above seqeuence approaches zero, any open interval containing zero must contain
the tail of this sequence. Hence, g is not increasing on any interval containing 0.
[Abbott 5.2.11] Assume that g is differentiable on [a, b] and satisfies g'(a) < 0 < g'(b).
(a) Show that there exists a point x ∈ (a, b) where g(a) > g(x), and a point y ∈ (a, b)
where g(y) < g(b).
Proof.
By definition:

g ( x ) - g ( a)
g'(a) = lim
x→a x-a
|g'(a)|
Pick 𝜖 = . Then, there exists 𝛿 > 0, such that for all x ∈ (a, b) satisfying |x - a| < 𝛿,
2
we have:

|g'(a)| g(x) - g(a) |g'(a)|

- < - g'(a) <
2 x-a 2
Since g'(a) < 0, |g'(a)| = - g'(a), so we have:

g'(a) g ( x ) - g ( a) g'(a)
< - g'(a) < -
2 x-a 2
Thus,

3g'(a) g(x) - g(a) g'(a)

< <
2 x-a 2

Thus, there exists x ∈ [a, b] such that

g(x) - g(a) g'(a)

< <0
x-a 2
Since x - a ≥ 0, ∃x ∈ [a, b] such that

g ( x ) - g ( a) < 0
or

g ( a) > g ( x )

In a similar fashion, we can prove that there exists y ∈ (a, b) such that

g( y) < g( b )

(b) Now, complete the proof of Darboux's theorem started earlier.

Proof.
We perform the following construction:
Using (a) there exists x1 ∈ (a, b) such that g(a) > g(x1 ). There exists y1 ∈ (x1 , b) such that
g( y 1 ) < g( b ) . Define I1 = [x1 , y1 ].
Again using (a), there exists x2 ∈ (x1 , y1 ) such that g(x2 ) > g(x1 ). There exists
y2 ∈ (x2 , y1 ) such that g(y2 ) < g(y1 ). Define I 2 = [x2 , y2 ].

∞
By the Nested Interval Property, there exists an element c ∈ ⋂ In such that (xn ) → c and
n=1
(yn ) → c. Since g is continuous, g(xn ) → g(c) and g(yn ) → g(c).

[Abbott 5.3.1] Recall from the exercise 4.4.9 that a function f : A → R is Lipschitz on A
if there exists an M > 0 such that:

f ( x) - f ( y)
≤M
x-y
for all x ≠ y in A.
(a) Show that if f is differentiable on a closed interval [a, b] and if f' is continuous on
[a, b] then f is Lipschitz on [a, b].

Proof.
Let x ≠ y be any two arbitrary points in [a, b]. Since f is differentiable on [x, y] (or [y, x] if
y < x), by the Mean Value Theorem, there exists c ∈ (x, y) such that:

f ( x) - f ( y)
= f'(c)
x-y
Since f' is continuous on a closed interval [a, b], by the extreme value theorem, f' attains a
maxima and a minima on [a, b], that is, there exists x0 , x1 such that f(x0 ) ≤ f(x) ≤ f(x1 )
for all x ∈ [a, b]. Let M = f(x1 ). Thus, f'(c) ≤ M.
Since x, y were arbitrary points, this must be true for all x, y ∈ [a, b]. Consequently, f is
Lipschitz continuous on [a, b].
(b) Review the definition of a contractive function in the exercise 4.3.11. If we add the
assumption that |f'(x)| < 1 on [a, b], does it follow that f is contractive on this set?
Proof.
We have: |f(x) - f(y)| ≤ M|x - y|. Since f'(x) < 1 on [a, b], it follows that
|f(x) - f(y)| < |x - y|. Thus, f is contractive on [a, b].

[Abbott 5.3.2] Let f be differentiable on an interval A. If f'(x) ≠ 0 on A, show that f is

one-to-one on A. Provide an example to show that the converse statement need not be true.
Proof.
Let x ≠ y be any two arbitrary points contained in A. Since f is differentiable on (x, y),
there exists a point x < c < y or x > c > y such that :

f ( x) - f ( y)
= f'(c)
x-y
Thus,

f(x) - f(y) = f'(c)(x - y)

Since f'(c) ≠ 0 and (x - y) ≠ 0, it follows that f(x) ≠ f(y). Thus, x ≠ y ⟹ f(x) ≠ f(y)
for all x, y ∈ A. Consequently, f is one-to-one on A.
Consider f(x) = x 3 . f is one-to-one on any interval of the real-line, but f'(x) = 3x 2 is zero
at the point x = 0.
[Abbott 5.3.3] Let h be a differentiable function defined on the inteval [0, 3], and assume
that h(0) = 1, h(1) = 2 and h(3) = 2.
(a) Argue that there exists a point d ∈ [0, 3], where h(d) = d.
Proof.
Consider g(x) = h(x) - x. By the Algebraic differentiability theorem, g is also differentiable
on [a, b].
We have:
 g( 0) = h( 0) - 0 = 1
g( 1) = h( 1) - 1 = 1
g( 3) = h( 3) - 3 = - 1
Since h is continuous on [1, 3], by the Intermediate Value Theorem, there exists c ∈ (1, 3)
such that g(c) = 0. Consequently, there exists c ∈ (1, 3), such that h(c) = c.
(b) Argue that at some point c, we have h'(c) = 1 / 3.
Proof.
Since h is differentiable on (1, 3), there exists c ∈ (0, 3) such that

h( 3) - h( 0) 2-1 1
= = = h'(c)
3-0 3 3
(c) Argue that h'(x) = 1 / 4 at some point in the domain.
Proof.
Since h is continuous on [0, 1] and differentiable on the interval (0, 1), by the Mean Value
Theorem, there exists c ∈ (0, 1) such that

h( 1) - h( 0)
h'(c) = =1
1
Since h is continuous on [1, 3] and differentiable on the interval (1, 3), by the Mean Value
Theorem, there exists t ∈ (1, 3) such that

h( 3) - h( 1)
h'(t) = =0
3-1

Since h is differentiable on [c, t] by the Darboux's theorem, the derivative function satisfies
1
the intermediate value property. If h'(t) < < h'(c), then there exists x ∈ (c, t) such that
4
1
h'(x) = .
4
[Abbott 5.3.4] Let f be differentiable on an interval A containing zero, and assume (xn ) is
a sequence in A with (xn ) → 0 and xn ≠ 0.
(a) If f(xn ) = 0 for all n ∈ N, show that f(0) = 0 and f'(0) = 0.
Proof.
Since f is differentiable at zero, f is also continuous at zero. Consequently,
lim f(x) = f(0). Thus, for all sequences (xn ) → 0, with xn ≠ 0, it follows that
x→0
f(xn ) → f(0). But, f(xn ) is the constant sequence (0, 0, 0, … ). Thus, f(0) = 0.

By definition:

f ( x) - f ( 0)
f'(0) = lim
x→0 x-0
Since f is differentiable at zero, for all sequences (t n ) → 0 with tn ≠ 0, the sequence of
difference quotients

f ( t n ) - f ( 0)
d( t n ) =
tn - 0
approaches f'(0). So, it must hold for the sequence (xn ) ⊆ A as well. We have:

f ( x n ) - f ( 0) 0-0
d( x n ) = = =0
xn - 0 xn
So, d(xn ) is the constant sequence (0, 0, 0, … ) which approaches 0. But, we know that
d(xn ) → f'(0). Consequently, f'(0) = 0.

(b) Add the assumption that f is twice-differentiable at zero and show that f''(0) = 0 as
well.
Proof.
By definition:
 f'(x) - f'(0)
f''(0) = lim
x→0 x-0
Since f' is differentiable at zero, it follows that for all sequences (xn ) → 0, with xn ≠ 0, the
sequence of the difference quotients:

f'(xn ) - f'(0)
y( x n ) =
xn - 0
approaches f''(0). But, f'(xn ) = 0 for all n ∈ N since f(xn ) = 0. Consequently, y(xn ) is
the constant zero sequence and thus f''(0) = 0.
[Abbott 5.3.5] (a) Supply the details for the proof of Cauchy's Generalized Mean Value
Theorem (Theorem 5.3.5).
Proof.
Let f and g be continuous on the closed interval [a, b] and differentiable on the open
interval (a, b). Define

h(x) = [f(b) - f(a)]g(x) - [g(b) - g(a)]f(x)

We have:

h ( a) = f ( b ) g ( a) - f ( a) g ( a) - g ( b ) f ( a) + f ( a) g ( a) = f ( b ) g ( a) - g ( b ) f ( a)
h ( b ) = f ( b ) g ( b ) - f ( a) g ( b ) - g ( b ) f ( b ) + g ( a) f ( b ) = g ( a) f ( b ) - f ( a) g ( b )
By the Algebraic differntiability theorem, h is differentiable on the open interval (a, b).
Applying the Mean Value Theorem, there exists c ∈ (a, b), such that:

h ( b ) - h ( a)
h'(c) = =0
b-a
Differentiating h(x) using the familiar rules of differentiation, we get:

h'(x) = [f(b) - f(a)]g'(x) - [g(b) - g(a)]f'(x)

Thus,
 h'(c) = [f(b) - f(a)]g'(c) - [g(b) - g(a)]f'(c)

To conclude, there exists a point c ∈ (a, b) such that:

[f(b) - f(a)]g'(c) = [g(b) - g(a)]f'(c)

(b) Give a graphical interpretation of the Generalized Mean Value Theorem analogous to
the one given for the Mean Value Theorem at the beginning of section 5.3 (Consider f and
g as parametric equations for a curve).

Proof.
Let F : R → R 2 be any curve in whose parameteric equation is given by (x(t), y(t)). Suppose
that a particle in motion in the 2D-plane according to this parametric curve. Then, x'(c) is
the x-component of the velocity at time t = c and y'(c) is the y-component of the velocity
at time t = c. The velocity vector at time t = c is v(c) = (x'(c), y'(c)), the magnitude of
the velocity (speed) is given by | v(c)| = (x'(c)) 2 + (y'(c)) 2 , whilst the direction of
velocity vector is given by:

y'(c)
tan 𝜃 =
x'(c)
Thus, the graphical interepretation of the Generalized Mean Value theorem, implies, that
there exists an instant c ∈ (a, b) during which the direction of motion of the particle is
parallel to the displacment vector between the points (x(a), y(a)) and (x(b), y(b)).
[Abbott 5.3.6] (a) Let g : [0, a] → R be differentiable, g(0) = 0 and |g'(x)| ≤ M for all
x ∈ [0, a]. Show that |g(x)| ≤ Mx for all x ∈ [0, a].

Proof.
Let x be an arbitrary point in (0, a]. Since g differentiable on the open interval (0, x), we
can apply the mean value theorem. Thus, there exists c ∈ (0, x), such that:

g( x) - g( 0) g( x)
| g'(c)| = = ≤M
x-0 x
Since x > 0, we can write:

| g(x)| ≤ Mx

Since x was an arbitrary point in (0, a], this must be true for all x ∈ (0, a]. Since, g(0) = 0,
it is also true for x = 0.
(b) Let h : [0, a] → R be twice differentiable, h'(0) = h(0) = 0 and | h''(x)| ≤ M for all
x ∈ [0, a]. Show that | h(x)| ≤ Mx 2 / 2 for all x ∈ [0, a].

Proof.
Using part (a), we have that |h'(x)| ≤ Mx for all x ∈ [0, a].
Again as before, let x be an arbitrary point in (0, a]. Since h(x) and g(x) = x 2 / 2 are
differentiable on (0, x) by the Generalized Mean Value Theorem, there exists c ∈ (0, x) such
that:

h'(c) h( x) - h( 0)
=
c x2 / 2 - 0
Thus,

h( x) |h'(c)| Mc
= ≤ = M,  { ∵ c > 0 }
x2 / 2 c c
Therefore,

Mx 2
| h( x) | ≤
2
Since, x was an arbitrary point in (0, a], this must be true for all x ∈ (0, a]. This is also true
for x = 0.
(c) Conjecture and prove and analogous result for a function that is differentiable three
times on [0, a].
Proof.
Let h : [0, a] → R be a function that is differentiable three times on [0, a] and assume that
h''(0) = h'(0) = h(0) = 0 and | h'''(x)| ≤ M for all x ∈ [0, a]. Our claim is that
|h(x)| ≤ Mx 3 / 6 for all x ∈ [0, a].

Using (b), we know that |h'(x)| ≤ Mx 2 / 2 for all x ∈ [0, a].

As before, if x is an arbitrary point in (0, a], applying generalized MVT to (0, x), there
exists a point c ∈ (0, x) such that:

h( x) - h( 0) h'(c)
=
x3 / 3 - 0 c2
Therefore,

h( x) Mc 2 / 2 M
≤ =
x3 / 3 c2 2
Consequently,

Mx 3
| h( x) | ≤
6
Since x was an arbitrary point in (0, a], this is true for all x ∈ (0, a].
[Abbott 5.4.1] Define

h( x) = | x|

on the interval [-1, 1] and extend the definition of h to all of R by requiring that
h(x + 2) = h(x). This results in a periodic sawtooth function.

Sketch a graph of (1 / 2)h(2x) on [-2, 3]. Give a qualitative description of the functions:

1
h n ( x) = n
h 2nx
2
as n gets larger.
Now define:
∞ ∞
1
g( x) = ∑ h n ( x) = ∑ n
h 2nx
n=0 n=0 2
The claim is that g(x) is continuous on all of R, but fails to be differentiable at any point.
Proof.
The graph of h1 (x) can be deduced by computing it's values at a few points:

x h(2x) (1 / 2)h(2x)
-2 h( - 4) = 0 0
-3 / 2 h( - 3) = 1 1/2
-1 h( - 2) = 0 0
-1 / 2 h( - 1) = 1 1/2
0 h( 0) = 0 0
1/2 h( 1) = 1 1/2
1 h( 2) = 0 0
3/2 h( 3) = 1 1/2
2 h( 4) = 0 0

Thus, h1 (x) has height (1 / 2) and a period L = 1. It has a corner at all points of the form
a 1
x = , where a ∈ Z. h2 (x) has height (1 / 4) and a period L = . It has a corner at all
2 2
a
points of the form x = 2 , where a ∈ Z.
2
[Abbott 5.4.2] Fix x ∈ R. Argue that the series
∞
1
∑ n
h 2nx
n=0 2
converges and thus g(x) is properly defined.
Proof.
 ∞
1
Let (sn ) be the sequence of partial sums of the infinite series ∑ n
h 2nx . Since
n=0 2
1
h 2 n x ≥ 0, the sequence of partial sums is monotonically increasing. Moreover,
n
2

k 1
sk = ∑ n=0 h 2nx
n
2
1
k
≤ ∑ n=0 n
since h 2nx ≤ 1
2
∞ 1 1
≤ ∑ n=0 =
2n 1-1/2
=2
Consequently, the sequence (sn ) is bounded. By the Monotone Convergence Theorem, (sn )
is a convergent sequence. Thus, g(x) is properly defined.
[Abbott 5.4.3] Taking the continuity of h(x) as given, reference the proper theorems from
chapter 4 that imply that the finite sum:
m
1
g m ( x) = ∑ n
h 2nx
n=0 2

is continuous on R.
Proof.
Let c be an arbitrary point in R. Let n ∈ N be arbitrary.
We know that 2 n x is continuous c and h( ⋅ ) is continuous at c. Since, the composition of
continuous functions is continuous, it follows that h 2 n x is continuous at c.
By the Algebraic Continuity Theorem, the finite sum:
m
1
g m ( x) = ∑ n
h 2nx
n=0 2
is also continuous at c. Since, c was arbitrary, this must be true for all points x ∈ R.
[Abbott 5.4.4] As the graph in figure 5.7 suggests, the structure of g(x) is quite intricate.
Answer the following questions, assuming that g(x) is indeed continuous.
(a) How do we know that g attains a maximum value M on [0, 2]? What is this value?
Proof.
The function g(x) repeats itself with period L = 2. Moreover, since g is continuous on the
closed and bounded interval [0, 2], by the extreme value theorem, there exists xmin and
xmax in [0, 2], such that:

g(xmin ) ≤ g(x) ≤ g(xmax )

for all x ∈ [0, 2]. Thus, the above points are global extrema.

(b) Let D be the set of points in [0, 2] where g attains its maximum. That is :

D = { x ∈ [0, 2] : g(x) = M }.
Find one point in D.
[Abbott 6.2.1] Let
nx
f n ( x) =
1 + nx 2

(a) Find the pointwise limit of (fn ) for all x ∈ (0, ∞).
Proof.
For a fixed x ∈ R, we have:
 nx
lim n → ∞ fn (x) = lim n → ∞
1 + nx 2
x
= lim n → ∞
1
n
+ x2
x 1
= =
x2 x
(b) Is the convergence uniform on (0, ∞)?
Proof.
The definition of uniform convergence is :

(∀𝜖 > 0)(∃N ∈ N)(∀x ∈ A)(∀n ≥ N)(|fn (x) - f(x)| < 𝜖)

Carefully negating the definition of uniform convergence:
(∃𝜖0 > 0)(∀N ∈ N)(∃xn ∈ A)(∃n ≥ N)(|fn (x) - f(x)| ≥ 𝜖0 )

Note. The point x ∈ A where the distance |fn (x) - f(x)| exceeds 𝜖0 can be different for
each N.
Let's explore the distance |fn (x) - f(x)|. We have:

nx 1
| f n ( x) - f ( x) | = -
1 + nx 2 x

nx 2 - 1 + nx 2
=
x 1 + nx 2
1 1
= - =
2
x 1 + nx x 1 + nx 2
1
=
x 1 + nx 2
Let 𝜖0 = 1. Let us choose 0 < x < 1.
 1 1
>
x 1 + nx 2 x( 1 + n )
Let's explore the inequality:
1
≥ 𝜖0
x( 1 + n )
We are interested to make

1
≥1
x( 1 + n )
⟹ x( 1 + n ) ≤1
1
⟹ xn ≤
n+1
1
Consequently, we can choose xn ≤ .
n+1

(fn ) does not converge uniformly to f on (0, ∞).

(c) Is the convergence uniform on (0, 1)?

Proof.
No, the convergence is not uniform on (0, 1) as reasoned above.
(d) Is the convergence uniform on (1, ∞)?
Let us explore the expression |fn (x) - f(x)| :

1
| f n ( x) - f ( x) | =
x 1 + nx 2
1
≤ { ∵ x ≥ 1}
1+n

1
We can choose N > . Then, |fn (x) - f(x)| < 𝜖 for all n ≥ N and for all x ∈ (1, ∞).
𝜖
[Abbott 6.2.2] (a) Define a sequence of functions on R by:

f n ( x) =
1 if x = 1, 1 , 1 , … , 1
2 3 n
0

and let f be the pointwise limit of the function fn . Is each fn continuous at zero? Does
fn → f uniformly on R? Is f continuous at zero?

Proof.
Point-wise convergence:
The definition of pointwise convergence is:

(∀𝜖 > 0)(∀x ∈ A)(∃N ∈ N)(∀n ≥ N)(|fn (x) - f(x)| < 𝜖)

Define

1
f ( x) =
1 if x = , ∀n ∈ N
n
0 otherwise

Let S be the set of rational numbers of the form:

1
S= :m ∈ N
m

We are interested to prove that fn → f pointwise. For x > 1 or x < 0 or x ∉ S, fn (x) = 0

for all n ∈ N. Thus, what is of interest to us, is the set S.
1
Assume that x ∈ S and let x = be fixed, where M is some natural number.
M

Pick an arbitrary 𝜖 > 0. We have: f(x) = 1. We are interested to make the distance
 1 1
|fn (x) - f(x)| as small as we please. If we choose n ≥ M, it follows that ≤ and
n M
f n ( x) = 1. Thus, |fn (x) - f(x)| = 0 < 𝜖.
Consequently, the suitable response to the given 𝜖-challenge is to pick N ≥ M.
Continuity of fn at zero.
Pick an arbitrary 𝜖 > 0. Pick an arbitrary fixed fN (x). We are interested to prove that
fN (x) is continuous at zero.

1 1 1
By definition, fN (x) = 1 for x = 1, , … , . Thus, if we choose 𝛿 = , for all
2 N N+1
x ∈ V𝛿 (0), it follows that |fN (x) - fN (0)| = 0 < 𝜖. Consequently, fN (x) is continuous at
c = 0. Since, N was arbitrary to begin with, this must be true for all N ∈ N.

Uniform convergence on R.
1 1
Let 𝜖0 = . We pick xn ≥ . Clearly, fn (xn ) = 0 and f(xn ) = 1, so
2 n+1
|fn (xn ) - f(xn )| ≥ 𝜖0 . Hence, fn is not uniformly convergent on R.

Is f continuous at zero?
The uniform convergence of the sequence of functions (fn ) is a necessary condition for the
limit function f to be continuous. The absence of uniform convergence implies that f is not
continuous at zero.
(b) Repeat this exercise using the sequence of functions

g n ( x) =
x if x = 1, 1 , 1 , … , 1
2 3 n
0 otherwise
Point-wise convergence:
Define
 g( x) =
x if x = 1 , n ∈ N
n
0 otherwise

Let S be the set of rational numbers of the form:

1
S= :m ∈ N
m
gn converges pointwise to g for all x ∉ S. Hence, let's investigate what happens if x is an
1
arbitrary fixed element of S. Let x = .
M

Again as before, if we pick n ≥ M + 1, then gn (x) = x and g(x) = x, so

| g n ( x) - g( x) | = 0 < 𝜖.

Hence, the suitable response to the given 𝜖-challenge is N ≥ M + 1.

Continuity of gn at zero.
Pick an arbitrary 𝜖 > 0. Pick an arbitrary fixed gN (x). We are interested to prove that
gN (x) is continuous at zero. By definition:

1 1 1
g N ( x) =
x if x = 1, , , … ,
2 3 N
0 otherwise

1
If we choose 𝛿 < , then for all x ∈ R satisfying |x| < 𝛿, we have
N
|gN (x) - gN (0)| = 0 < 𝜖. Consequently, gN (x) is continuous at zero. Since, N was abitrary,
this must be true for all n ∈ N.
Uniform convergence on R.
Pick an arbitrary 𝜖 > 0. By the Archimedean property there exists a natural number K ∈ N,
1
such that < 𝜖. Consider what happens when we sample fK , fK+1 at the points
K
 1 1
x= , , ….
K K+1

1 1 1 1
gK = gK =0 gK =0
K K K+1 K+2
1 1 1 1 1
gK+1 = gK+1 = gK+1 =0
K K K+1 K+1 K+2
1 1 1 1 1 1
g = g = g =
K K K+1 K+1 K+2 K+2
Then, for all x ∈ R, for k ≥ K, it follows that |gk (x) - g(x)| < 𝜖.
Is g continuous at zero?
The sequence functions (gn ) are continuous at zero and (gn ) converges uniformly to g. By
the Continuous Limit Theorem, g is continuous at zero.
(c) Repeat the exercise once more with the sequence

1 if x = 1
n
h n ( x) =
x if x = 1, 1 , … , 1
2 n-1
0 otherwise

Proof.
Point-wise convergence:
Define

h( x) =
x if x = 1 : n ∈ N
n
0 otherwise
 1
Pick an arbitrary 𝜖 > 0. Let x ∈ R be a fixed arbitrary real number of the form , where
m
1
m ∈ N. Let N ∈ N be such that N > m. Then, hn (x) = x = for all n ≥ N.
m
Consequently, |hn (x) - h(x)| = 0 < 𝜖 for all n ≥ N.
Thus, the sequence of functions (hn ) converge pointwise to h.
Continuity of hn at zero.
1
Let 𝜖 > 0 be arbitrary. Pick 𝛿 < . Then, for all x ∈ V𝛿 (0), we have
n
|hn (x) - hn (0)| = 0 < 𝜖. Consequently, hn is continuous at zero.

Uniform convergence on R.
1
Let 𝜖0 = .
2

1 1
Case I. N = 1. There exists x1 = , such that |h1 (x1 ) - h(x1 )| = ≥ 𝜖0 .
2 2

1 1 1
Case II. N ≥ 2. Pick xn = . We have |hN (xn ) - h(xn )| = 1 - ≥ = 𝜖0 .
N N 2

Consequently,

(∃𝜖0 > 0)(∀N ∈ N)(∃xn ∈ A)(∃n ≥ N)(|hn (xn ) - h(xn )| ≥ 𝜖0 )

Hence, (hn ) does not converge uniformly to h.

[Kaczor&Nowak 3.1.1] Prove that a sequence of functions { fn } defined on A is uniformly
convergent on B ⊂ A to f : B → R if and only if the sequence of numbers { dn }, where:

dn = sup{ |fn - f(x)| : x ∈ B },  n ∈ N

converges to zero.
Proof.
⟹ direction.

Pick an arbitrary 𝜖 > 0. (fn ) converges to f uniformly, so there exists N𝜖 such that for all
0

x ∈ B, |fn (x) - f(x)| < 𝜖 for all n ≥ N𝜖 . Thus, the set

D N 𝜖 = { | f N 𝜖 - f ( x) | : x ∈ B }

has lower bound 0 and upper bound 𝜖. This applies to all the sets DN +1 ,DN +2 ,DN +3 ,....
𝜖 𝜖 𝜖

This can also be seen if we graph fn and f. fn is always within an 𝜖-band of f. So, the
distance between fn and f is bounded.
By the least upper bound property, the supremum of Dn exists for all n ≥ N𝜖 . Let
dn = sup Dn .

Since dn is a limit point of the set Dn , by the order limit theorem, 0 ≤ dn ≤ 𝜖. Again, this
holds for all n ≥ N𝜖 .
Consequently, it follows that (dn ) → 0.
⟸ direction.
Pick an arbitrary 𝜖 > 0.
Since (dn ) → 0, there exists N ∈ N such that

-𝜖 < 0 ≤ d n < 𝜖

Since dn is an upper bound for the set Dn :

-𝜖 < 0 ≤ |fn (x) - f(x)| ≤ sup { |fn (x) - f(x)| : x ∈ B } < 𝜖

∀ x ∈ B, ∀ n ≥ N .

Thus, (fn ) converges uniformly to f.

[Abbott 6.2.3] For each n ∈ N and x ∈ [0, ∞), let:

1
x 1 x≥
g n ( x) =   and  hn (x) = n
1 + xn
nx if 0 ≤ x < 1
n
Answer the following questions for the sequences (gn ) and (hn ):
(a) Find the pointwise limit on [0, ∞).
Proof.
The pointwise limit of gn (x) is given by:
Case I. 0 < x < 1.
We have:

x
lim n → ∞ gn (x) = lim n → ∞
1 + xn
lim n → ∞ x
=
lim n → ∞ 1 + lim n → ∞ x n
x
=   ∵ (x) n → 0, if 0 < x < 1
1+0
=x

Case II. x = 1.
We have:

1 1
lim gn (x) = lim =
n→∞ n→∞ 1 + 1n 2

Case III. x > 1.

We have:

x
lim n → ∞ gn (x) = lim n → ∞
1 + xn
x -n+1
= lim n → ∞
x -n + 1
0
= =0
0+1
Thus, the sequence of functions (gn ) converges pointwise to g defined by:

x if 0 ≤ x < 1
1
g( x) = if x = 1
2
0 if x > 1

The pointwise limit of hn (x) is given by:

Case I. If x > 0.
Pick an arbitrary 𝜖 > 0. By the Archimedean property, there exists N ∈ N, such that
1
< x. For all n ≥ N, the distance |hn (x) - h(x)| = 0 < 𝜖.
N

Case II. If x = 0.
The sequence (hn (0)) = (0, 0, 0, 0, 0, … ) converges to 0.
Thus, the sequence of functions (hn ) converges pointwise to h defined by:
1 if x > 0
h( x) =
0 if x = 0
(b) Explain how we know that the convergence cannot be uniform on [0, ∞).
The sequence of functions (gn ):
Let x ∈ (0, 1). Let's explore the expression |gn (x) - g(x)|. We have:

x
| g n ( x) - g( x) | = -x
1 + xn
x - x - x n+1
=
1 + xn
x n+1
=
1 + xn
x n+1
= n
∵ x n+1 > 0 and 1 + x n > 0
1+x
x n+1 x 2n
> >
2 2
1
Let 𝜖0 = . Our claim is:
4

x 2n 1
>
2 22
1
x 2n >
2
2n
1
xn >
2
Thus,

(∃𝜖0 > 0)(∀N ∈ N)(∃xn ∈ [0, ∞))(∃n ≥ N)(|gn (xn ) - g(xn )| ≥ 𝜖0 )

So, (gn ) does not converge uniformly to g.
The sequence of functions (hn ):
This is actually evident from the graph of h(x). If we draw an 𝜖 band around h, for each
N ∈ N, there exists points xn ∈ (0, 1), such that |hn (xn ) - h(xn )| ≥ 𝜖.

1 1 1 1
Let 𝜖0 = . For all N ∈ N, pick xN ∈ 0, . Then, |hN (xN ) - h(xN )| ≥ 1 - = ≥ 𝜖0 .
4 2N 2 2
Again,
(∃𝜖0 > 0)(∀N ∈ N)(∃xn ∈ [0, ∞))(∃n ≥ N)(|hn (xn ) - h(xn )| ≥ 𝜖0 )

(c) Choose a smaller set over which the convergence is uniform and supply an argument to
show that this is indeed the case.
Proof.
Using a computer algebra system to plot (gn ), I think that (gn ) should converge uniformly
1
g( x) = x in the interval 0, .
2

0.6

0.4

0.2

0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6

Mathematically,
 x
| g n ( x) - g( x) | = -x
1 + xn
x n+1 x n+1
= = ∵ x n+1 ≥ 0 and 1 + x n > 0
1 + xn 1 + xn
(1 / 2) n+1
< { ∵ 0 ≤ x ≤ 1 / 2}
1 + 0n
n+1
1
=
2
Pick an arbitray 𝜖 > 0. Our claim is that:

1
<𝜖
2 n+1
Thus:

1
2 n+1 >
𝜖
which implies that

1
( n + 1) > log(1 / 𝜖)
2
Thus,

(∀𝜖 > 0)(∃N ∈ N)(∀x ∈ [0, 1 / 2])(∀n ≥ N)(|gn (x) - g(x)| < 𝜖)

Consequently, (gn ) converges uniformly to g on [0, 1 / 2]. Actually, it converges uniformly to

g on any closed interval [0, a] where a < 1, as well as any interval of the form [a, ∞) where
a > 1.

1
Similarly, (hn ) converges uniformly to h on ,1 . Let 𝜖 > 0 be arbitrary. Pick N ≥ 2.
2
1
Then, for all x ∈ ,1 and for all n ≥ N, |hn (x) - h(x)| = 0 < 𝜖.
2
[Kaczor&Nowak 3.1.2] Assume that (fn ) converges uniformly to f on A and (gn )
converges uniformly to g on A. Show that (fn + gn ) converge uniformly to (f + g) on A. Is
it true that (fn ⋅ gn ) converge uniformly (f. g)?
Proof.
Pick an arbitrary 𝜖 > 0.
𝜖
There exists N1 ∈ N for all x ∈ A, such that for all n ≥ N, |fn (x) - f(x)| < .
2

𝜖
There exists N2 ∈ N for all x ∈ A, such that for all n ≥ N2 , |gn (x) - g(x)| < .
2

We can write:

|(fn (x) + gn (x)) - (f(x) + g(x)| = |fn (x) - f(x) + gn (x) - g(x)|
≤ | f n ( x) - f ( x) | + | g n ( x) - g( x) |
Let N = max{ N1 , N2 }. Then for all n ≥ N, it follows that:

|(fn (x) + gn (x)) - (f(x) + g(x)| ≤ |fn (x) - f(x)| + |gn (x) - g(x)|
𝜖 𝜖
= + =𝜖
2 2
Counterexample.

[Abbott 6.2.4] Review exercise 5.2.8 which includes the definition for a uniformly
differentiable function. Use the results discussed in the section 6.2 to show that if f is
uniformly differentiable, then f' is continuous.
Proof.
Define the sequence of functions
 1
f x + n - f ( x)
g n ( x) =
1
n

and let

g(x) = f'(x)

We can show that (gn ) converges uniformly on A to g.

Pick an arbitrary 𝜖 > 0. By definition of uniform differentiability, there exists 𝛿 > 0 (where
𝛿 is only a function of 𝜖), such that for all x, y ∈ A, satisfying |x - y| < 𝛿, it follows that:

f ( y) - f ( x)
- f'(x) < 𝜖
y-x
1
By the Archimedean property, there exists N ∈ N, such that < 𝛿. But, this means, that
N
1
for all n ≥ N, the point y = x + satisfies |y - x| < 𝛿.
n

Thus, there exists N ∈ N, such that for all x ∈ A and n ≥ N, it follows that:

f ( x + 1 / n ) - f ( x)
- f'(x) < 𝜖
1/n
Note that N is a function of 𝛿, which in turn depends only on 𝜖. Hence, the sequence of
functions (gn ) converge uniformly on A to g.
Now,
Uniform differentiability of f ⟹ Differentiability of f ⟹ Continuity of f

1
For a fixed n, since x + is continuous and f is continuous, f(x + 1 / n) is also
n
continuous. By algebraic continuity theorem,
 f ( x + 1 / n ) - f ( x)
g n ( x) =
( 1 / n)
is continous.
Let c ∈ A be an arbitrary fixed point.
Each of the gn are continuous at c ∈ A and (gn ) converges uniformly to g on A. By the
continuous limit theorem, g is continuous at c ∈ A. Since c was arbitrary to begin with, g is
continuous on the whole of A.
[Abbott 6.2.5] Using the Cauchy Criterion for convergent sequences of real numbers,
supply a proof for Theorem 6.2.5. (First define a candidate for f(x) and then argue that
(fn ) → f uniformly).

Proof.
⟹ direction.

We are given that a sequence of functions (fn ) defined on the set A ⊆ R converges
uniformly on A to f.
Pick an arbitrary 𝜖 > 0. There exists N ∈ N for all x ∈ A, such that for all k ≥ N:

| f k ( x) - f ( x) | < 𝜖 / 2
Let n, m ≥ N be arbitrary. We have:

|fn (x) - fm (x)| = |(fn (x) - f(x)) - (fm (x) - f(x))|

≤ |(fn (x) - f(x))| + |(fm (x) - f(x))|
𝜖 𝜖
= + =𝜖
2 2
Since n, m ≥ N were arbitrary to begin with, this must be true ∀n, m ≥ N.
⟸ direction.
We are given that, for all 𝜖 > 0, ∃N ∈ N, such that for all x ∈ A and for all n, m ≥ N, we
have:
 | f n ( x) - f m ( x) | < 𝜖

Fix t ∈ A. Then it follows that, (fn (t)) is a Cauchy sequence. By the Cauchy criterion for
convergence of real numbers, (fn (t)) is a convergent sequence and nlim
→∞
fn (t) exists.

Since t was arbitrary, this must be true for all t ∈ A. Consequently, we define:

f(x) = lim fn (x)

n→∞

Now, fix n ≥ N and x ∈ A be arbitrary. Consider the sequence:

|fn (x) - fN (x)|, |fn (x) - fN+1 (x)|, … , |fn (x) - fm (x)|, …

Pick an arbitrary 𝜖 > 0.

By the uniform Cauchy condition, ∃N(𝜖) ∈ N, such that for all m ≥ N,
am = |fn (x) - fm (x)| < 𝜖. Thus, lim am exists.
m→∞

Since each of the terms am is strictly smaller than 𝜖, by the Order Limit theorem,
lim am ≤ 𝜖.

We have:
lim am = lim m → ∞|fn (x) - fm (x)|
= |lim m → ∞ fn (x) - lim m → ∞ fm (x)| Since lim |bn | = |lim bn |
= | f n ( x) - f ( x) |

Consequently, |fn (x) - f(x)| ≤ 𝜖. Since n ≥ N and x ∈ A were arbitrary to begin with, this
must be true for all x ∈ A and n ≥ N.
By definition, (fn ) converges uniformly on A to f.
[Abbott 6.2.6] Assume that fn → f on the set A. Theorem 6.2.6 is an example of a
typical question which asks whether a trait possessed by each fn is inherited by the limit
function. Provide an example to show that all of the following propositions are false if the
convergence is only assumed to pointwise on A. Then, go back and decide which are true
under the stronger hypothesis of uniform convergence.
(a) If each fn is uniformly continuous, then f is uniformly continous.
Proof.
Let fn (x) = x n for x ∈ [0, 1]. Since fn (x) is continuous on a compact set [0, 1], it is
uniformly continuous on [0, 1].
Define

f ( x) =
if 0 ≤ x < 1
0
if x = 1
1
Pick x ∈ [0, 1). We know that, if |x| < 1, then x n → 0. If x = 1, then x n is the
constant sequence (1, 1, 1, … ) and converges to 1. Consequently, fn converges pointwise
to f.
Since f is not continuous at c = 1, f is not uniformly continuous on [0, 1].

Claim. If fn ⏪A⏫ f and if each fn is uniformly continuous, f is uniformly continuous.

By definition of uniform continuity:

(∀𝜖 > 0)(∃𝛿(𝜖) > 0)(∀x, y ∈ A)(∀|x - y| < 𝛿)(|fn (x) - fn (y)| < 𝜖)
By definition of uniform convergence:

(∀𝜖 > 0)(∃N(𝜖) ∈ N)(∀x ∈ A)(∀n ≥ N)(|fn (x) - f(x)| < 𝜖)

Pick an arbitrary 𝜖 > 0.

There exists N(𝜖 / 3) such that for all x ∈ A, |fN (x) - f(x)| < 𝜖 / 3.
There exists 𝛿(𝜖 / 3) such that for all x, y ∈ A, satisfying |x - y| < 𝛿,
| f N ( x) - f N ( y) | < 𝜖 / 3.
We can write for all x, y ∈ A satisfying |x - y| < 𝛿:

| f ( x) - f ( y) | = | f ( x) - f N ( x) + f N ( x) - f N ( y) + f N ( y) - f ( y) |
≤ | f ( x) - f N ( x) | + | f N ( x) - f N ( y) | + | f N ( y) - f ( y) |
< 𝜖/3+𝜖/3+𝜖/3 = 𝜖
Consequently, f is uniformly continuous on A.
(b) If each fn is bounded, then f is bounded.
Proof.

Claim. If fn ⏪A⏫ f and each fn is bounded, f is bouned.

By the definition of uniform convergence:

(∀𝜖 > 0)(∃N(𝜖) ∈ N)(∀x ∈ A)(∀n ≥ N)(|fn (x) - f(x)| < 𝜖)

Pick 𝜖 = 1. Then, there exists N ∈ N such that for all x ∈ A,

| f ( x) | ≤ | f N ( x) | + 1

Since fN (x) is bounded, ∃M > 0, ∀x ∈ A, such that |fN (x)| ≤ M. Consequently, for all
x ∈ A,

| f ( x) | ≤ M + 1
Thus, f(x) is bounded.
[Abbott 6.2.7] Let f be uniformly continuous on all of R and define a sequence of
1
functions by fn (x) = f x + . Show that fn → f uniformly. Give an example to show
n
that this proposition fails if f is only assumed to be continuous and not uniformly
continuous on R.
Proof.
Pick an arbitrary 𝜖 > 0.
Since f is uniformly continuous on R, ∃𝛿0 (𝜖) > 0, such that for all |x - y| < 𝛿,
| f ( x) - f ( y) | < 𝜖.

Let y be any arbitrary point defined by the following linear function:

1
y = x+
n
1
where n ∈ N. Now, we can choose y to be arbitrarily close to x. Pick N > . Then, for all
𝛿0
n ≥ N, |x - y| < 𝛿. But, this implies, that for all n ≥ N, |f(x + 1 / n) - f(x)| < 𝜖|.

Consequently,

(∀𝜖 > 0)(∃N(𝜖) ∈ N)(n ≥ N)(|fn (x) - f(x)| < 𝜖)

Thus, fn converges uniformly on R to f.

Carefully negating the definition of uniform convergence, we have:

(∃𝜖0 > 0)(∀N ∈ N)(∃n ≥ N)(∃xn ∈ A)(|fn (xn ) - f(x)| ≥ 𝜖0 )

Let f(x) = x 2 . We know that f is continuous on R, but it is not uniformly continuous (the
derivative is not bounded).

1 1
Fix 𝜖0 = . Let's explore the expression: f x+ - f ( x) . We have:
2 n
 2
1 1
f x+ - f ( x) = x+ - x2
n n
2x 1
= x2 + + - x2
n n2
2x 1
= x2 + + 2 - x2
n n
2x 1
= +
n n2
2x
≥
n
Let's explore the inequality:

2x 1
≥ 𝜖0 =
n 2
N
Pick xN ≥ . Then,
4

1 2xN 1 2( N / 4) 1 1 1
f x+ - f ( x) = + = + = +
N N N2 N N2 2 N2
Thus, fn (x) does not converge uniformly to f(x).
[Abbott 6.2.8] Let gn be a sequence of continuous functions that converges uniformly to g
1 1
on a compact set K. If g(x) ≠ 0 on K, show that converges uniformly on K to .
gn g

Proof.
1 1
Consider the expression - . We have:
g n ( x) g( x)

1 1 | g n ( x) - g( x) |
- =
g n ( x) g( x) |gn (x)||g(x)|
Now, (gn ) ⏪K⏫ g. Since, each gn is continuous on K, by the continuous limit theorem, g is
continuous on K. By the extreme value theorem, g attains a minima m and maxima M on
K, such that:

m ≤ g( x) ≤ M

for all x ∈ K. Since g(x) ≠ 0 for all x ∈ K, either g is strictly positive or g is strictly
negative, and it does not intersect the x-axis.
Let L = min(|m|, |M|). We have, L > 0 and |g(x)| ≥ L.
L
Pick 𝜖 = . As (gn ) ⏪K⏫ g, there exists N1 (L / 2) such that for all x ∈ K and for all n ≥ N1 ,
2
we have:

L
||gn (x)| - |g(x)|| ≤ |gn (x) - g(x)| <
2

So, for all x ∈ K and n ≥ N1 ,

L
||gn (x)| - |g(x)|| <
2
L L
⟺- < | g n ( x) | - | g( x) | <
2 2
L L
⟺ | g( x) | - < | g n ( x) | < | g( x) | +
2 2
This implies, (∀x ∈ K)(∀n ≥ N1 ) :

L
| g n ( x) | > | g( x) | -
2
L L
> L- =
2 2
Finally, since (gn ) ⏪K⏫ g, there exists N2 (𝜖) such that, for all x ∈ K and ∀n ≥ N2 :
 L2
| g n ( x) - g( x) | < 𝜖 ⋅
2
Choose N = max{ N1 , N2 }. Then, for all x ∈ K and n ≥ N,

1 1 | g n ( x) - g( x) | 𝜖 L2 / 2
- = < =𝜖
g n ( x) g( x) |gn (x)||g(x)| L2 / 2
1 K 1
Thus, .
gn ⏪⏫ g

[Abbott 6.2.10] This exercise and the next explore partial converses of the Continuous
Limit Theorem. Assume that fn → f pointwise on [a, b] and the limit function f is
continuous on [a, b]. If each fn is increasing (but not necessarily continuous), show that
fn → f uniformly.

Proof.
Define the sequence of functions

g n ( x) = f n ( x) - f ( x)

Fix x ∈ [a, b].

We have:

lim gn (x) = lim fn (x) - lim f(x) = 0

n→∞ n→∞ n→∞

Thus, gn (x) converges pointwise to the constant zero function g(x) = 0 on [a, b].
Moreover, eachgn (x) is increasing.
Pick an arbitrary 𝜖 > 0.
Since gn (a) → 0, there exists N1 (𝜖, a) ∈ N, such that for all n ≥ N1 , gn (a) ∈ (-𝜖, 𝜖).
Since gn (b) → 0, there exists N2 (𝜖, b) ∈ N, such that for all n ≥ N2 , gn (b) ∈ (-𝜖, 𝜖).
Let N = max{ N1 , N2 }. Let x ∈ [a, b] be an arbitrary point.
Since a ≤ x ≤ b, gn (a) ≤ gn (x) ≤ gn (b). Thus, for all n ≥ N, gn (x) ∈ (-𝜖, 𝜖). Since x was
arbitrary, this is true for all x ∈ [a, b].
Consequently,

(∀𝜖 > 0)(∃N(𝜖, a, b) ∈ N)(∀n ≥ N)(∀x ∈ [a, b])(|fn (x) - f(x)| < 𝜖)

Thus, fn converges uniformly to f on [a, b].

[Abbott 6.2.11] (Dini's Theorem). Assume that fn → f pointwise on a compact set K
and assume that for each x ∈ K the sequence fn (x) is increasing. Follow these steps to
show that if fn and f are continuous on K, then the convergence is uniform.
(a) Set gn = f - fn and translate the preceding hypotheses into statements about the
sequence (gn ).
Proof.
We have: gn → g pointwise, where g(x) = 0 the constantly zero function on a compact set
K, and for each x ∈ K, the sequence (gn (x)) is decreasing.

Since both fn and f are continuous on K, gn is continuous on K.

(b) Let 𝜖 > 0 and define Kn = { x ∈ K : gn (x) ≥ 𝜖 }. Argue that K1 ⊇ K2 ⊇ K3 ⊇ … and
use this observation to finish the argument.
[Abbott 6.3.1] Consider the sequence of functions defined by:

xn
g n ( x) =
n
(a) Show that (gn ) converges uniformly on [0, 1] and find g = lim gn . Show that g is
differentiable and compute g'(x) for all x ∈ [0, 1].
Proof.
Define g as the constantly zero function:
 g( x) = 0

Our claim is that gn converges uniformly on [0, 1] to g. Our claim is that:

| g n ( x) - g( x) | < 𝜖
xn
<𝜖
n
We can strengthen the condition we wish to prove by replacing the LHS by its upper bound.
xn
Since ≤ 1 , we are interested to prove:
n n

1
<𝜖
n
1 1
Choose N > . Then, for all n ≥ N, < 𝜖. Consequently,
𝜖 n
(∀x ∈ [0, 1])(∀n ≥ N)(|gn (x) - g(x)| < 𝜖). Thus, gn converges uniformly on [0, 1] to g.

The constantly zero function g is differentiable for all x ∈ [0, 1] and g'(x) = 0 for all
x ∈ [0, 1].

(b) Now show that (g'n ) converges on [0, 1]. Is the convergence uniform? Set h = lim gn '
and compare h and g'. Are they the same?
Proof.
We have:

gn '(x) = x n-1
Define h as:

h( x) =
0 if 0 ≤ x < 1
1 if x = 1
1
gn ' converges pointwise to h. The convergence is not uniform. Pick 𝜖0 = . Let
2
 1
xn ∈ ,1 . We are interested to make :
2
1
x n-1 ≥
2
If we replace the LHS by its lower bound, we can strengthen the condition we wish to
prove. Since x n-1 ≥ x n , our claim is:

1
xn ≥
2
So, we choose:
(1/n)
1
x≥
2
Hence,
(∃𝜖0 > 0)(∀N ∈ N)(∃xn ∈ [0, 1])(∃n ≥ N)(|gn '(xn ) - h(xn )| ≥ 𝜖0 )

Consequently, gn ' converges NOT uniformly on [0, 1] to h.

h and g' are not the same.
[Abbott 6.3.2] Consider the sequence of functions :

1
h n ( x) = x2 +
n
(a) Compute the pointwise limit of (hn ) and then prove that the convergence is uniform on
R.

Proof.

Fix x ∈ R. We know that, if lim an = a, then lim an = lim an = a. Thus:

 1
lim n → ∞ hn (x) = lim n → ∞ x2 +
n
1
= lim n → ∞ x 2 +
n
(1/2)
1 2
= lim n → ∞ x + lim n → ∞
n
= x2
= | x|
Consider the expression:

1 1
| h n ( x) - h m ( x) | = x2 + - x2 +
n m
1 1
x2 + n - x2 + m
=
1 1
x2 + + x2 +
n m
1 1
-m
=
n the denominator
1 1
is non-negative
x2 + + x2 +
n m
1 1
n
-m
≤ ∵ x2 ≥ 0
1 1
+
n m
1 1 1 1
- +
n m n m
=
1 1
+
n m

1 1
= -
n m
1
Pick an arbitrary 𝜖 > 0. Since → 0, and convergent sequences are Cauchy, there exists
n
N ( 𝜖) > 0, such that for all n > m ≥ N,
1 1
- <𝜖
n m
Consequently, by Cauchy criterion for uniform convergence of a sequence of functions, (hn )
converges uniformly on R to h.
(b) Note that each hn is differentiable. Show that g(x) = lim hn '(x) exists for all x and
explain how we can be certain that the convergence is not uniform on any neighbourhood
of zero.
Proof.
By Chain rule of differentiation, we have:
x
h n ' ( x) =
1
x2 +
n
Moreover,
x x
lim hn '(x) = lim =
n→∞
1 | x|
x2 +
n
If x > 0, lim hn '(x) = 1. If x < 0, then lim hn '(x) = - 1. Define:

h'(x) =
1 if x > 0
-1 if x < 0
h'(x) is not defined at x = 0.
We are given that the sequence of functions (hn ) that converge pointwise to h and are
differentiable on any neighbourhood of zero.
By the Differentiable Limit Theorem, if (hn ') converges uniformly on any neighbourhood of
zero to g, then g = h'.
By the contrapositive of the Differentiable Limit Theorem, since h' is not defined at x = 0,
so h' is not differentiable on any neighbourhood containing 0, so it implies (hn ') does NOT
converge uniformly to h on any neighbourhood containing zero.
[Abbott 6.3.3] Consider the sequence of functions
x
f n ( x) =
1 + nx 2

(a) Find the points on R where each fn (x) attains its maximum and minimum value. Use
this to prove that (fn ) converges uniformly on R. What is the limit function?
Proof.
We have:
1 + nx 2 (1) - (x)(2nx)
f n ' ( x) = 2
1 + nx 2
1 + nx 2 - 2nx 2
= 2
1 + nx 2
1 - nx 2
= 2
1 + nx 2
f n ' ( x) attains a maxima/minima at the points :

1 1
x0 = - , x1 =
n n
We have:

1 1
f n (x 0 ) = - ,  fn (x1 ) =
2 n 2 n
The limit function f:
Firstly, fix x = 0. Then, (fn (x)) is the constant zero sequence (0, 0, 0, 0, … ) which
converges to 0.
Next, assume that x ≠ 0. We have:

x
lim fn (x) = lim n → ∞
1 + nx 2
( x / n)
= lim n → ∞
( 1 / n) + x 2
=0
Consequently, the limit function f is the constantly zero function f(x) = 0.
Uniform convergence to f:
Pick an arbitrary 𝜖 > 0.
We have:

x
| f n ( x) - f ( x) | =
1 + nx 2

1 1 1
≤ - ≤ f n ( x) ≤
2 n 2 n 2 n
1
Pick N > 2 . Then, for all n ≥ N and for all x ∈ R, |fn (x) - f(x)| < 𝜖. Consequently,
4𝜖
(fn ) converges uniformly on R to f.

(b) Let f = lim fn . Compute fn '(x) and find all the values of x for which
f'(x) = lim fn '(x).

Proof.
We have:

1 - nx 2
f n ' ( x) =
2
1 + nx 2

Fix x ≠ 0. Then,
 1 - nx 2
lim n → ∞ fn '(x) = lim n→∞ 2
1 + nx 2
1 x2
-
n2 n
= lim n → ∞
2
1
n
+ x2

=0
If x = 0, then fn '(x) = 1, so lim fn '(x) = 1.
n→∞

Also, since the limit function f is the constant zero function f(x) = 0, its derivative
f'(x) = 0.

Thus, f'(x) = lim fn '(x) for all values of x except x = 0.

[Abbott 6.3.4] Let
sin(nx)
h n ( x) =
n
Show that hn → 0 uniformly on R but that the sequence of derivatives (hn ') diverges for
every x ∈ R.
Proof.
Since |sin nx| ≤ 1, we have:

1
|hn (x) - 0| ≤
n

1
If we pick N > 2 , then for all n ≥ N and for all x ∈ R, we have:
𝜖

|hn (x) - 0| < 𝜖

Thus, (hn ) converges uniformly on R to the constantly zero function h(x) = 0.
By the familiar rules of differentiation,

h n ' ( x) = n cos nx
Now, h'(x) = 0.

If x = 0, clearly n is a divergent sequence. If x is a multiple of 𝜋 / 2, then the

subsequence of even terms (h2 ', h4 ', … ) is divergent, so the sequence is divergent.
Fix x ∈ R, such that x ≠ 0 and x is not a multiple of 𝜋 / 2. Consequently, sin x ≠ 0.
We can show that cos nx is not a convergent sequence.
We proceed by contradiction. Let un = cos nx. Then, un+2 = cos(n + 2)x. Assume that
(un ) is a convergent sequence. Then, lim(un - un+2 ) = 0. We have:

(un - un+2 ) = cos nx - cos(n + 2)x

= 2 sin(n + 1)x ⋅ sin x
Since lim(un - un+2 ) = 0, this implies that lim sin(n + 1)x approaches zero {since
sin x ≠ 0}, which implies that lim sin nx = 0. But, with the addition formula for since, we
have:

sin(n + 1)x = sin nx cos x + cos nx sin x

lim sin(n + 1)x = lim (sin nx cos x) + lim (cos nx sin x)
= lim cos nx
So, both lim nx = 0 and lim cos nx = 0. But, we know that:

sin 2 nx + cos 2 nx = 1
Passing to limits on both sides, we have a contradiction. Hence, out initial assumption is
false. un = cos nx is not a convergent sequence.
[Abbott 6.3.5] Let
 nx + x 2
g n ( x) =
2n
and set g(x) = lim gn (x). Show that g is differentiable in two ways:
(a) Compute g(x) by algebraically taking the limit as n → ∞ and then find g'(x).
Proof.
Fix x ∈ R. We have:
nx + x 2
lim n → ∞ gn (x) = lim n → ∞
2n
x + x2 / n
= lim n → ∞
2
x2
lim n → ∞ x + lim n → ∞ n
=
lim n → ∞ 2
x
=
2
x
Thus, the limit function g(x) = . The derivative of the limit function is:
2

1
g'(x) =
2
(b) Compute gn '(x) for each n ∈ N for each n ∈ N and show that the sequence of
derivatives (gn ') converges uniformly on every interval [-M, M]. Use the theorem 6.3.3 to
conclude that g'(x) = lim gn '(x).
Proof.
Fix n ∈ N. By the familiar rules of differentiation:

1 1 x
g n ' ( x) = (n + 2x) = +
2n 2 2n
We are interested to prove that gn '(x) converges uniformly on any bounded interval
1
[-M, M] to the constant function h(x) = .
2
Pick an arbitrary 𝜖 > 0. We are interested to make the distance |gn '(x) - h(x)| smaller than
𝜖.

We have:
x
| g n ' ( x) - h( x) | =
2n
M
≤
2n
M
If we pick N > , then for all n ≥ N, and for all x ∈ [-M, M], |gn '(x) - g'(x)| < 𝜖.
2𝜖

Consequently, the sequence of the derivatives (gn ') converges uniformly on [-M, M] to the
1
constant function h(x) = lim gn '(x) = .
2

We find that the sequence of functions (gn ) converge pointwise on the closed interval
[-M, M] to g and are differentiable. Since (gn ') converges uniformly on [-M, M] to h, by
the Differentiable Limit Theorem, it follows that lim gn ' = h = g' on [-M, M].
(c) Repeat parts (a) and (b) for the sequence fn (x) = nx 2 + 1 / (2n + x).

Proof.
Pointwise convergence of fn :
Fix x ∈ R. We have:
nx 2 + 1
lim n → ∞ fn (x) = lim n → ∞
2n + x
x2 + 1 / n
= lim n → ∞
( 2 + x / n)
2
x
=
2
x2
Let f(x) = .
2

Uniform convergence of fn ' on any bounded interval [-M, M]:

Our claim is that the sequence of derivatives (fn ') converges uniformly on [-M, M] to
g( x) = x.

By the familiar rules of differentiation:

d nx 2 + 1
f n ' ( x) =
dx 2n + x
(2n + x)(2nx) - nx 2 + 1 (1)
=
(2n + x) 2
nx 2 + 4n 2 x - 1
=
(2n + x) 2
We have:

nx 2 + 4n 2 - 1
f n ' ( x) - x = -x
(2n + x) 2
nx 2 + 4n 2 x - 1 - x 4n 2 + 4nx + x 2
=
(2n + x) 2
nx 2 + 4n 2 x - 1 - 4n 2 x - 4nx 2 - x 3
=
(2n + x) 2
x 3 + 3nx 2 + 1
=-
(2n + x) 2
Pick an arbitrary 𝜖 > 0.
We have:

|x 3 + 3nx 2 + 1|
| f n ' ( x) - x| =
(2n + x) 2
Since x ∈ [-M, M], we have:
(2n + x) 2 ≥ (2n - M) 2
and
 |x 3 + 3nx 2 + 1| ≤ |x 3 | + |3nx 2 | + 1 ≤ M 3 + 3nM 2 + 1

1
Pick N1 > max ,M . Then, for all n ≥ N1 , since
M2

n > M ⟹ nM 2 > M 3 ⟹ 3nM 2 > M 3  { M ∈ (0, ∞)}

Also,
nM 2 > 1
Thus,

|x 3 + 3nx 2 + 1| ≤ M 3 + 3nM 2 + 1 ≤ 3nM 2 + 3nM 2 + nM 2 = 7nM 2

Since, M > 0, 2n > 2n - M. Consequently:

1 1
<
(2n) 2 (2n - M) 2

|x 3 + 3nx 2 + 1| 7nM 2 7M 2
| f n ' ( x) - x| = ≤ =
(2n + x) 2 4n 2 4n
7M 2
Pick N2 > . Then, for all n ≥ N2 , the |fn '(x) - x| < 𝜖.
4𝜖

Thus, a suitable response to the given 𝜖-challenge is N = max{ N1 , N2 }.

[Abbott 6.3.6] Provide an example or explain why the request is impossible. Let's take the
domain of the functions to be all of R.
(a) A sequence (fn ) of nowhere differentiable functions with fn → f uniformly and f
everywhere differentiable.
Proof.
Consider the sequence of functions (tn ) defined by:
 1
t n ( x) =
if x ∉ Q
n
if x ∈ Q
0
t n (x) is nowhere continuous and hence nowhere differentiable. However, (t n ) converges
uniformly on R to the constantly zero function t(x) = 0, which is differentiable.
(b) A sequence (fn ) of differentiable functions such that (fn ') converges uniformly but the
original sequence (fn ) does not converge for any x ∈ R.
Proof.

(c) A sequence (fn ) of differentiable functions such that both (fn ) and (fn ') converge
uniformly but f = lim fn is not differentiable at some point.
Proof.
This proposition is false. The sequence of functions (fn ) converges pointwise to f and are
differentiable. Moreover,the sequence of the derivative functions (fn ') converges uniformly
on R. By the Differentiable Limit Theorem, lim fn ' = f' for all x ∈ R.
[Abbott 6.3.7] Use the Mean Value Theorem to supply a proof for the Theorem 6.3.2. To
get started, observe that the triangle inequality implies that, for any x ∈ [a, b] and
m, n ∈ N ,

|fn (x) - fm (x)| ≤ |(fn (x) - fm (x)) - (fn (x0 ) - fm (x0 )| + |fn (x0 ) - fm (x0 )|
Proof.
Pick an arbitrary 𝜖 > 0.
Fix x ∈ [a, b] and consider the closed interval [x, x0 ] ([x0 , x] if x0 < x).
Since fn - fm is continuous on [x0 , x] and differentiable on (x0 , x), by the Mean Value
Theorem(MVT), there exists 𝛼 ∈ (x, x0 ), such that:
 |(fn (x) - fm (x)) - (fn (x0 ) - fm (x0 ))| = |fn '(𝛼) - fm '(𝛼)| \ cdot|x-x0 |
≤ |fn '(𝛼) - fm '(𝛼)| \ cdot|b-a|

Since (fn ') converges uniformly on [a, b], there exists N1 ∈ N such that for all
n > m ≥ N1 and for all x ∈ [a, b],

𝜖
|fn '(x) - fm '(x)| <
2|b - a|

Since (fn (x0 )) is convergent, by the Cauchy criterion for sequences of real numbers, there
exists N2 (x0 , 𝜖), such that for all n > m ≥ N2 ,

𝜖
|fn (x0 ) - fm (x0 )| <
2

If we pick N = max{N1 , N2 }, then for all n > m ≥ N and for all x ∈ [a, b],

|fn (x) - fm (x)| < 𝜖

Note that, neither N1 nor N2 are functions of x, so our choice of N works regardless of
what x is. This closes the proof.
[Abbott 6.4.1] Supply the details for the proof of the Weierstrass M-Test (Corollary 6.4.5).
Proof.
Pick an arbitrary 𝜖 > 0.
∞
We are given, that the infinite series ∑ Mn converges. By the Cauchy Criterion for the
n=1
convergence of an infinite series of real numbers, it follows that, there exists N(𝜖), such
that for all n > m ≥ N, we have:
 |Mm+1 + Mm+2 +…+ Mn | < 𝜖

Pick an arbitrary x ∈ A. For all n > m ≥ N, we have:

|fm+1 (x) + fm+2 (x) +…+ fn (x)| ≤ |fm+1 (x)| + |fm+2 (x)| +…+ |fn (x)|
≤ Mm+1 + Mm+2 +…+ Mn
= |Mm+1 + Mm+2 +…+ Mn | { M n > 0 ∀n ∈ N }
<𝜖
Since x was arbitrary, this is true for all x ∈ A.
Thus:

(∀𝜖 > 0)(∃N(𝜖) ∈ N)(∀x ∈ A)(∀n > m ≥ N)(|fm+1 (x) +…+ fn (x)| < 𝜖)

∞
By the Cauchy criterion for the uniform convergence of infinite series, ∑ fn converges.
n=1

[Abbott 6.4.2] Decide whether each proposition is true or false, providing a short
justification or counterexample as appropriate.
∞
(a) If ∑ gn converges uniformly, then (gn ) converges uniformly to zero.
n=1

Proof.
This proposition is true.
∞
We are given that ∑ gn converges uniformly.
n=1

Pick an arbitrary 𝜖 > 0.

By the Cauchy criterion for the uniform convergence of infinite series, there exists
N ∈ N(𝜖), such that for all m, n ≥ N and for all x ∈ A, we have:
 |gm+1 (x) +…+ gn (x)| < 𝜖
Hence, for all n ≥ N and ∀x ∈ A, it follows that:

| g n ( x) | < 𝜖
By the definition of uniform convergence, the sequence of functions (gn ) converges
uniformly to the constantly zero function g(x) = 0.
∞ ∞
(b) If 0 ≤ fn (x) ≤ gn (x) and ∑ gn converges uniformly, then ∑ fn converges uniformly.
n=1 n=1

Proof.
This proposition is true.
We have for all m, n ∈ N and for all x ∈ A:

|fm+1 (x) +…+ fn (x)| = fm+1 (x) +…+ fn (x) since fn (x) ≥ 0 ∀n ∈ N
≤ gm+1 (x) +…+ gn (x)
≤ |gm+1 (x) +…+ gn (x)|
Pick an arbitrary 𝜖 > 0. There exists N(𝜖) ∈ N, such that for all x ∈ A and for all
n > m ≥ N, it follows that:

|fm+1 (x) +…+ fn (x)| ≤ |gm+1 (x) +…+ gn (x)| < 𝜖

∞
Consequently, ∑ fn converges uniformly.
n=1

∞
(c) If ∑ fn converges uniformly on A, then there exists constants Mn such that
n=1
∞
|fn (x)| ≤ Mn for all x ∈ A and ∑ Mn converges.
n=1

Proof.
The converse of the Weierstrass M-Test is:
∞ ∞
For all uniformly convergent series ∑ fn (x), (∃Mn ) (|fn (x)| ≤ Mn ) ∧ ∑ Mn converges .
n=1 n=1

We are interested to prove, that this proposition is false.

∞
We are interested to show that there exists a uniformly convergent series ∑ fn (x) such
n=1
that there does not exist a sequence (Mn ) such that the properties (|fn (x)| ≤ Mn ) and
∞
∑ Mn converges are both true.
n=1

Consider the sequence of constant functions :

( - 1) n
f n ( x) =
n
∞
( - 1) n
We know that, ∑ converges uniformly on R. For all sequences Mn satisfying,
n=1
n
|fn (x)| ≤ Mn , we must have:

1
Mn ≥
n
∞ ∞
1
But, ∑ is a divergent series. By the comparision test, ∑ Mn diverges.
n=1
n n=1

[Abbott 6.4.3] (a) Show that:

∞ cos 2 n x
g( x) = ∑
n=0 2n
is continuous on all of R.
Proof.
Since, (∃Mn ) such that

cos 2 n x 1
| g n ( x) | = ≤ = Mn
2n 2n
∞ ∞
and ∑ Mn converges, by the Weierstrass M-Test, ∑ gn converges uniformly on R.
n=1 n=1

∞
Since each gn (x) is continuous on R, by the term-by-term Continuity theorem, ∑ gn (x) is
n=1
continuous on R.
(b) The function g was cited in section 5.4 as an example of a continuous nowhere
differentiable function. What happens if we try to use the theorem 6.4.3 to explore whether
g is differentiable?

Proof.
Let

cos 2 n x
g n ( x) =
2n
So,

gn '(x) = - sin 2 n x
Since g is nowhere differentiable, by the contrapositive of the term-by-term differentiability
theorem, we have that:

If g is not differentiable on A, then either ∑ gn (x) converges NOT pointwise for all x ∈ A,
or ∑ gn '(x) converges NOT uniformly on A.
 ∞
Since ∑ gn converges uniformly on R, the only possibility is that ∑ gn ' = ∑ sin 2nx
n=1
does not converge uniformly on R. We can actually prove this result.

If the infinite series ∑ hn is uniformly convergent on A, it uniformly convergent on S,

(∀S ⊆ A).

If (∃S ∈ A), where ∑ hn converges NOT uniformly, then ∑ hn is not uniformly

convergent on A.
𝜋
Consider the point x0 = . Then,
3

(gn '(x0 )) = (-sin 2𝜋 / 3, -sin 4𝜋 / 3, -sin 8𝜋 / 3, -sin 16𝜋 / 3, -sin 32𝜋 / 3, … )

3 3 3 3
= - , ,- , ,…
2 2 2 2

Thus, gn '(x0 ) is a divergent sequence. By the nth term test, ∑ gn '(x0 ) does not converge
pointwise at x0 = 𝜋 / 3. Consequently, ∑ gn '(x) does not converge uniformly on any
interval [a, b] containing x0 = 𝜋 / 3. So, ∑ gn ' does not converge uniformly on R.
[Abbott 6.4.4] Define:
∞
x 2n
g( x) = ∑
n=0 1 + x 2n
Find the values of x where the series converges and show that we get a continuous function
on this set.
Proof.
Let x0 be a fixed point in (-1, 1). Then,
 x02n
0 ≤ g n (x 0 ) = ≤ x02n
1 + x02n
∞ ∞
x02n
We know that, ∑ x02n is a convergent series. Hence, by the comparison test ∑ 2n
is
n=1 n=1 1 + x 0
∞
a convergent series. Hence, ∑ gn (x) converges pointwise on (-1, 1).
n=1

Let x belong to any compact interval [a, b] such that -1 < a < b < 1. Let
c = max{ |a|, |b| }.

We have:

x 2n
g n ( x) = 2n
≤ x 2n ≤ c 2n   since x 2n ≥ 0
1+x
Define Mn = c 2n . Then,
∞
1
∑ Mn =
n=1 1 - c2
∞
x 2n
By the Weierstrass M-test, the infinite series ∑ 2n
converges uniformly on [a, b].
n=1 1 + x
∞
Since ∑ gn converges uniformly on [a, b] to g, and each gn is continuous on [a, b], by the
n=1
term-by-term continuity theorem, g is continuous on [a, b].
Let's show that the radius of convergence does not exceed 1.
Assume that |x0 | = 1. Let sk (x) be the sequence of partial sums of the infinite series
∞
∑ gn (x). We have:
n=1

k k
1 k
s k (x 0 ) = ∑ g n (x 0 ) = ∑ =
n=1 n=1
2 2
 ∞
We know that this is a divergent sequence. Hence, ∑ gn (x) does not converge at x0 = 1.
n=1

∞
Assume that |x0 | > 1. We know that, if ∑ an converges, then (an ) → 0. We have:
n=1

(x0 ) 2n 1 1
lim = lim = =1
1 + (x0 ) 2n (1 / x0 ) 2n + 1 0+1
∞
Thus, ∑ gn (x) does not converge for |x| > 1.
n=1

[Kaczor&Nowak 3.2.1] Find where the following series converges pointwise:

∞
1
(a) ∑ , x ≠ 1.
n=1 1 + xn

Proof.
Case I. Fix x = x0 and assume that |x0 | < 1. Then, -1 < x0 < 1. We have:

1
lim n → ∞ fn (x0 ) = lim n → ∞
1 + x0n
1
=
1 + lim n → ∞ x0n
1
= =1
1+0
∞ ∞
We know that if ∑ an converges, then (an ) → 0. Thus, ∑ fn (x) does not converge for
n=1 n=1
| x| < 1.

Case II. If x0 = - 1, then fn (x0 ) is an unbounded sequence and hence diverges.

1
Case III. Assume that x0 > 1 or x0 < - 1. If x0 > 1, then it follows that 0 < < 1.
x0
 1 1
Moreover, if x0 < - 1, then -1 < < 0. Consequently, we must have -1 < < 1. We
x0 x0
have:

1 1
0 < f n (x 0 ) = < = g n (x 0 )
1 + x0n x0n
∞ ∞
1
Since ∑ n
is a geometric series with 0 < 1 < 1, it follows that the series ∑ gn (x0 )
n=1 x 0
x0 n=1
∞
is convergent. By the comparison test, ∑ fn (x) also converges for |x| > 1.
n=1

∞
xn
(b) ∑ , x ≠ - 1.
n=1 1 + xn

Proof.
Case I. Fix x = x0 . Let's quickly dispose off the case when |x0 | > 1. We have:

1 1
lim fn (x0 ) = lim = =1
1
n 0+1
x0
+1

∞ ∞
We know that, if the infinite series ∑ an converges, then (an ) → 0. Thus, ∑ fn (x) does
n=1 n=1
not converge for |x0 | > 1.
1
Case II. Also, if x0 = 1, then lim fn (x0 ) = . Thus, fn (x0 ) does not converge for x0 = 1.
2

Case III. Assume that 0 < x0 < 1. Since x0 > 0, x0n > 0 and thus 1 + x0n > 1.
Consequently,
 x0n
0 < f n (x 0 ) = < x0n = gn (x0 )
1 + x0n
∞ ∞
We know that, ∑ gn (x0 ) is a convergent series. Hence, by the comparison test, ∑ fn (x0 )
n=1 n=1
converges for 0 < x0 < 1.
Case IV. If x0 = 0, the series converges to 0.
Case V. Now, assume that -1 < x0 < 0. We have:

an+1 x0n+1 1 + x0n 1 + x0n

= ⋅ = |x 0 | ⋅
an x0n 1 + x0n+1 1 + x0n+1
So,

an+1 1 + x0n
lim = |x0 | ⋅ lim
an 1 + x0n+1
Since lim |an | = |lim an |, we have:

an+1 1 + lim x0n

lim = |x 0 | ⋅ = |x 0 | < 1
an 1 + lim x0n+1
∞
Hence, by the ratio test, ∑ an converges absolutely over -1 < x0 < 0.
n=1

∞
2n + xn
(c) ∑ , x ≠ -1
n=1
n n
1+3 x 3

Proof.
Case I. Fix x = x0 . Assume that x0 > 1. Then,
 2 n + x0n 2 n + x0n 2 n + x0n 1 1
0 < f n (x 0 ) = ≤ = = +
1 + 3 n x0n 1 + 2 n x0n 2 n x0n x0n 2n
∞
1 1
Since (1 / x0 ) < 1, it follows that ∑ n
+ converges.
n=1 x 0 2n
∞
By the comparison test, ∑ fn (x0 ) converges.
n=1
Case II. Assume that .
[Abbott 6.4.5] (a) Prove that :
∞
xn x2 x3 x4
h( x) = ∑ = x+ + + +…
n=1 n2 4 9 16
is continuous on [-1, 1].
Proof.
Assume that x ∈ [-1, 1]. We have:

xn 1
| h n ( x) | = 2
≤ ,  since |x| n < 1
n n2
∞
1
Let Mn = . We know that, ∑ Mn is a convergent series. By the Weierstrass M-Test,
n2 n=1
∞
∑ hn (x) converges uniformly on [-1, 1]. Further, since each hn (x) is continuous on
n=1
[-1, 1], by the term-by-term continuity theorem, h(x) is continuous on [-1, 1].

(b) The series

∞
x n
x2 x3 x4
f ( x) = ∑ = x + + + +…
n=1
n 2 3 4
converges for every x in the half-open interval [-1, 1) but does not converge when x = 1.
For a fixed x0 ∈ (-1, 1) explain how we can still use the Weierstrass M-Test to prove that
f is continuous at x0 .
Proof.
Let x0 be an arbitrary point in (-1, 1) and let c < 1 be such that the interval (-1, c)

|x 0 | n
|f n ( x 0 ) | =
n
≤ |x 0 | n
Define:

M n = |x 0 | n
∞ ∞
We know that, ∑ Mn converges, since |x0 | < 1. Hence, by the Weierstrass M-Test, ∑ fn
n=1 n=1
converges uniformly at x0 . Since each fn (x) is continuous at x0 , by the Term-by-term
continuity theorem, f(x) is continuous at x0 .
[Abbott 6.4.6] Let

1 1 1 1 1
f ( x) = - + - + -…
x x+1 x+2 x+3 x+4

Show that f is defined for all x > 0. Is f continuous on (0, ∞)? How about differentiable?
Proof.
Well-definedness of f.
Define

( - 1) n
f n ( x) =
x+n
Then,
 ∞
f ( x) = ∑ f n ( x)
n=0

Let x0 be an arbitrary point, such that x0 > 0. Fix x = x0 .

We have:

1 1 1
≥ ≥ ≥ … ≥0
x0 x0 + 1 x0 + 2
Moreover,

1
lim =0
x0 + n
∞
By the Alternating Series Test for convergence, ∑ fn (x) converges pointwise on x > 0.
n=0
Continuity of f.
Let [a, ∞) be any interval such that a > 0. Let us pair of the terms of f and write:

1 1 1 1 1 1
f ( x) = - + - + - +…
x x+1 x+2 x+3 x+4 x+5
Infinite addition is associative, as long as we have conditional convergence.
Define:

1 1 1
g n ( x) = - =
(x + 2n) (x + 2n + 1) (x + 2n)(x + 2n + 1)

Then,
∞
f ( x) = ∑ g n ( x)
n=0

1
Now, g0 (x) = = M0 . Moreover,
a( a + 1 )
 1 1
0 ≤ g n ( x) ≤ ≤ = Mn
(2n)(2n + 1) 4n 2
∞ ∞
for all n ≥ 1. Since ∑ Mn converges, by the Weierstrass M-Test, ∑ gn (x) converges
n=0 n=0
uniformly on [a, ∞) for any a > 0.
Since each gn (x) is continuous for x > 0, by the Term-by-term continuity theorem, f(x) is
continuous on [a, ∞), where a > 0. Thus, f is continuous on (0, ∞).
Differentiability of f.
We have:

n 1 (-1) n+1
f n ' ( x) = ( - 1) ⋅ - =
( x + n) 2 ( x + n) 2
Again let [a, ∞) be an interval such that a > 0 is arbitrary. Now,

1
| f 0 ' ( x) | ≤ = M0
a2
and

1
| f n ' ( x) | ≤ = Mn
n2

∞
Since ∑ Mn converges, by the Weierstrass M-Test, ∑ fn ' is uniformly convergent on
n=1

[a, ∞). Since each fn is differentiable, and ∑ fn (x) converges pointwise on [a, ∞), by the
∞
Term-by-Term differentiability theorem, f(x) = ∑ fn (x) is differentiable and further
n=0
∞
f'(x) = ∑ fn '(x).
n=0

[Abbott 6.4.7] Let

 ∞
sin kx
f ( x) = ∑
k=1 k3
(a) Show that f(x) is differentiable and the derivative f'(x) is continuous.
Proof.
Let us recall that, the infinite series
∞
1
∑
n=1 np
converges if and only if p > 1. If p > 1, we have:
Consider the Cauchy series:
∞
∑ 2 n b2 n = b1 + 2b2 + 4b4 +…
n=0

which is:
∞ ∞ n
1 1
∑2 n
⋅ =∑
n=0 2 n
p
n=0 2 p-1
1
The latter series converges if and only if p-1
< 1, that is 2 p-1 > 1, which means
2
∞
p-1 > 0 or p > 1. By the Cauchy condensation test, if the series ∑ 2 n b2 converges, it n

n=0
∞
implies that ∑ bn converges.
n=1

Uniform Convergence of fk (x).

We have:
 sin kx 1
0 ≤ |f k ( x 0 ) | = ≤ = Mk
k3 k3
∞ ∞
We know that ∑ Mk converges and hence by the Weierstrass M-Test, ∑ fk converges
k=1 k=1
uniformly on R.
Uniform Convergence of fk '(x).
We have:

cos kx
f k ' ( x) =
k2
And,

cos kx 1
0 ≤ | f k ' ( x) | = ≤ = Ck
k2 k2
∞
We know that, ∑ Ck is convergent. Hence, by the Weierstrass M-Test, ∑ fk ' converges
k=1
uniformly on R.

Since ∑ fk ' converges uniformly on R, ∑ fk converges uniformly on R and each fk is

differentiable, by the term by term differentiability theorem, f(x) is differentiable and
further f'(x) = ∑ fk '(x) for all x ∈ R.
Moreover, since each fk ' is continuous on R, by the term-by-term continuity theorem,
f'(x) = ∑ fk '(x) is continuous on R.

(b) Can we determine if f is twice-differentiable?

Proof.
We have:
 sin kx
fk ''(x) = -
k

It is not possible to determine if f is twice differentiable.

[Abbott 6.4.8] Consider the function
∞
sin(x / k)
f ( x) = ∑
k=1
k
Where is f defined? Continuous? Differentiable? Twice-differentiable?
Proof.
Well-definedness.
Fix x = x0 and consider the absolute value series:
∞
sin(x0 / k)
∑
k=1
k
We have that:

sin(x0 / k) |x 0 / k | |x 0 |
0≤ ≤ =
k k k2
∞
|x 0 | sin(x0 / k)
Since ∑ is a convergent series, by the comparision test, ∑ is
k2 k=1
k
∞
sin(x0 / k)
convergent. By the Absolute convergence test, ∑ also converges.
k=1
k

Consequently, f is defined on R.
Uniform Convergence and continuity.
Let x ∈ [-1, 1]. Define Mk as:
 sin(x / k) |x / k | 1
0≤ ≤ ≤ 2
k k k

Since ∑ Mk is a convergent series, by the Weierstrass M-Test, ∑ fk converges uniformly

on [-1, 1]. Since each fk is continuous on [-1, 1], by the term-by-term continuity theorem,
f is continuous on [-1, 1].

Differentiability.
We have:

cos x / k
f k ' ( x) =
k2
Define Ck as :

cos x / k 1
0≤ ≤
k2 k2

Since ∑ Ck is convergent, by the Weierstrass M-Test, ∑ fk ' is uniformly convergent on

R. ∑ fk is convergent pointwise on atleast one point in [-1, 1]. Each fk is differentiable
on R. Hence, by the term-by-term differentiability theorem, f is differentiable on R and
f'(x) = ∑ fk '(x) for all x ∈ R. In turn, this implies f is continuous on R.

Twice Differentiability.
We have:

sin x / k
fk ''(x) = -
k3
Define Dk as :

sin x / k 1
0≤ - ≤
k3 k3
Since ∑ Dk is convergent, by the Weierstrass M-Test, ∑ fk '' is uniformly convergent on
R. ∑ fk ' is convergent pointwise on atleast one point in R. Each fk ' is differentiable on
R. Hence, by the term-by-term differentiability theorem, f is twice-differentiable on R.

In fact, f is infinitely differentiable.

[Abbott 6.4.9] Let
∞
1
h( x) = ∑
n=1 x2 + n2
(a) Show that h is a continuous function defined on all of R.
Proof.
Since x 2 ≥ 0, we have that:

1 1
0 < h n ( x) = ≤ = Mn
x2 + n2 n2

By the Weierstrass M-Test, since ∑ Mn converges, it follows that ∑ hn converges

uniformly on R. Since each hn (x) is continuous on R, by the term-by-term continuity
theorem, h is continuous on R.
(b) Is h differentiable? If so, is the derivative function h' continuous?
Proof.
We have:

2x
h n ' ( x) = - 2
x2 + n2
Power Series.
It is time to put some mathematical teeth into our understanding of functions expressed in
the form of a power series; that is functions of the form:
∞
f(x) = ∑ an x n = a0 + a1 x + a2 x 2 + a3 x 3 +…
n=0

The first order of business is to determine the points x ∈ R for which the resulting series on
the right-hand side converges. This set certainly contains x = 0 and as the next result
demonstrates it takes a very predictable form.
∞
Theorem 6.5.1. If a power series ∑ an x n converges at some point x0 ∈ R, then it
n=1
converges absolutely for any x satisfying |x| < |x0 |.
Proof.
∞
If ∑ an x0n converges, then an x0n → 0. Since convergent sequences are bounded, an x0n
n=0
is a boundd sequence. Let M > 0 and satisfy |an x0n | ≤ M for all n ∈ N. If x ∈ R satisfies
|x| < |x0 | then:

n n
n x x
|a n x | = |an x0n | ⋅ ≤M
x0 x0
But notice that:

∞ n
∑M x
n=0
x0
x
is a geometric series with the ratio < 1 and so it converges. By the Comparison Test,
x0
∞
∑ an x n converges absolutely. If a series converges absolutely, then it also converges
n=0
 ∞
conditionally. Hence, ∑ an x n converges for all |x| < |x0 |. □
n=0
The main implication of theorem 6.5.1 is that the set of points for which a given power
series converges must necessarily be { 0 }, R or a bounded interval centered around x = 0.
Because of the strict inequality in theorem 6.5.1, there is some ambiguity about the
endpoints of the interval, and it is possible that the set of convergent points may be of the
form (-R, R), [-R, R), (-R, R] or [-R, R].
The value of R is referred to as the radius of convergence of a power series, and it is
customary to assign R the value 0 or ∞ to represent the set { 0 } or R respectively. Some of
the standard devices for computing the radius of convergence of a power series are explored
in the exercises. Of more interest to us here, is the properties of functions defined in this
way. Are they continuous? Are they differentiable? If so, can we differentiate the series term-
by-term? What happens at the endpoints?
Establishing Uniform Convergence.
The positive answers to the preceding questions, and the userfulness of power series in
general, are largely due to the fact that they converge uniformly on compact sets contained
in their domain of convergent points. As we are about to see, a complete proof of this fact
requires a fairly delicate argument attributed to the Norwegian mathematician Niels Henrik
Abel. A significant amount of progress, however, can be made with the Weierstrass M-Test.
∞
Theorem 6.5.2. If a power series ∑ an x n converges absolutely at a point x0 , then it
n=0
converges uniformly on the closed interval [-c, c], where c = |x0 |.
Proof.
Consider x ∈ [-|x0 |, |x0 |]. Then, |x| ≤ |x0 |. We have:

0 ≤ f n ( x) = | a n x n | = | a n | ⋅ | x| n ≤ | a n | ⋅ | x 0 | n = M n
∞ ∞
Since ∑ Mn is convergent, by the Weierstrass M-Test, it follows that ∑ fn (x) converges
n=1 n=1
uniformly for all x ∈ [-c, c] where c = |x0 |.
 ∞
Thus, ∑ an x n converges absolutely for all |x| < |x0 |.
n=1

For many applications, theorem 6.5.2 is good enough. For instance, because any
x ∈ (-R, R) is contained in the interior of a closed interval [-c, c] ⊆ (-R, R), it follows that
a power series that is convergent on an open interval is necessarily continuous on this
interval.
But, what happens if we know that a series converges at an endpoint of its interval of
convergence? Does the good behavior of the series on (-R, R) necessarily extend to the
endpointx = R?
If the convergence of the series at x = R is absolute, then we can again rely on theorem
6.5.2, to conclude that the series converges uniformly on the set [-R, R]. The remaining
open question question is what happens if a series converges conditionally at a point x = R.
We may still use Theorem 6.5.1. to conclude that we have pointwise convergence on the
interval (-R, R], but more work is needed to establish that we have uniform convergence on
compact sets containing x = R.
Abel's Theorem.
∞
We should remark that if the power series g(x) = ∑ an x n converges conditionally at
n=0
x = R, then it is possible for it to diverge at x = -R. As a counterexample, let:

( - 1) n n
g n ( x) = x
n
∞
∑ gn (x) converges at x = 1, but diverges at x = -1. To keep our attention fixed on the
n=1
convergent endpoint, we will prove uniform convergence on [0, R]. The first step in the
argument is an estimate that should be compared to Abel's test for the convergence of
series developed back in chapter 2. (Exercise 2.7.13).
∞
Lemma 6.5.3 (Abel's Lemma). Let bn satisfy b1 ≥ b2 ≥ b3 ≥ … ≥ 0 and ∑ an be a
n=1
series for which the partial sums are bounded. In other words, assume that there exists
A > 0 such that

|a1 + a2 +…+ an | ≤ A
for all n ∈ N. Then, for all n ∈ N, we have:

|a1 b1 + a2 b2 +…+ an bn | ≤ Ab1

Proof.
Let sn = a1 + a2 +…+ an . Using the summation-by-parts formula derived in the exercise
2.7.12, we can write:
n n
∑ ak bk = sn bn+1 + ∑ sk (bk - bk+1 )
k=1 k=1

Thus,

∑ n ak bk n
= sn bn+1 + ∑ k=1 sk (bk - bk+1 )
k=1

n
≤ | sn | |bn+1 | + ∑ k=1 sk (bk - bk+1 ) Triangle Inequality
n
≤ | sn | |bn+1 | + ∑ k=1 |sk ||(bk - bk+1 )|
n
≤ Abn+1 + ∑ k=1 A(bk - bk+1 )

= Abn+1 + (Ab1 - Ab2 + Ab2 - Ab3 +…+ Abn - Abn+1 ) Telescopic Sum
= Ab1
∞
Hence the sequence of partial sums of the product series ∑ an bn is bounded. □
n=1

It is worth observing that if A were an upper bound on the partial sums of ∑ |an | (the
absolute value series), then the proof of lemma 6.5.3 would be a simple exercise in the
triangle inequality. The point of the matter is that because we are only assuming
conditional convergence, the triangle inequality is not going to be of any use in proving
Abel's theorem, but we are now in possession of an inequality we can use in its place.
 ∞
Theorem 6.5.4 (Abel's Theorem.) Let g(x) = ∑ an x n be a power series that converges
n=0
at the point x = R > 0. Then, the series converges uniformly on the interval [0, R]. A
similar result holds if the series converges at x = -R.
Proof.
To set the stage for an application of lemma 6.5.3, we first write:
∞ ∞
xn
g( x) = ∑ a n x = ∑ a n R n
n

n=0 n=0 Rn
Let 𝜖 > 0. By the Cauchy Criterion for the Uniform Convergence of the series (Theorem
6.4.4), we will be done, if we can produce an N such that for all n > m ≥ N and for all
x ∈ [0, R], we have:

|gm+1 (x) +…+ gn (x)| < 𝜖

that is:

x m+1 xn
am+1 R m+1 +…+ an R n <𝜖 (1)
R m+1 Rn
∞
Because we are assuming that ∑ an R n converges, the Cauchy criterion for an infinite series
n=1
of real numbers implies that:
𝜖
(∃N(𝜖) ∈ N)(∀n > m ≥ N) |am+1 R m+1 + an R n | <
2
Now, fix x ∈ [0, R] and fix the index m ∈ N.
∞
Consider the partial sums of ∑ am+1 R m+1 . The sequence of partial sums is bounded by
j=m+1
𝜖 / 2. Moreover,
 x m+1 xn
≥ … ≥ ≥0
R m+1 Rn
form a non-negative monotonically decreasing sequence.
Hence, applying Abel's lemma, we have:

n m+1
x m+1 xn 𝜖 x
∑ am+1 R m+1
+…+ an R n
< ⋅ <𝜖
k=m+1 R m+1 Rn 2 R
Since m ≥ N was arbitrary and x ∈ [0, R] was arbitrary, this holds true for all n > m ≥ N
and for all x ∈ [0, R].
Consequently, by the Cauchy criterion for the uniform convergence of series of functions,
∞
∑ an x n converges uniformly on the interval [0, R]. □
n=0

The Success of the Power Series.

An economical way to summarize the conclusions of theorem 6.5.2 and Abel's theorem is
with the following statement:
Theorem 6.5.5. If a power series converges pointwise on the set A ⊆ R, then it converges
uniformly on any compact set K ⊆ A.
Proof.
Let K ⊆ A be an arbitrary compact set in A. A compact set contains both it's maximum
x1 and a minimum x0 , which by the hypothesis must be in A. Since the power series
∞
∑ an x n converges at x = x1 , by the Abel's theorem, the series converges uniformly on
n=1
[x0 , x1 ]. Hence, it converges uniformly on K. Since K was arbitrary, the series converges
uniformly ∀K ⊆ A, where K is a compact set. □
 ∞
Let c be an arbitrary point in the domain of convergence, A, of the series ∑ an x n . We can
n=1
always construct a compact interval containing c. Hence, the power series is uniformly
convergent on this compact interval. By the term-by-term continuity theorem,
∞
g( x) = ∑ a n x n is continuous on this compact interval and therefore at c. Since c was
n=1
arbitrary, g is continuous for all c ∈ A. Thus, a power series is continuous at every point at
which it converges.
To make an argument for differentiation, we would like to appeal to the Term-by-Term
differentiability theorem. However, this result has a more involved set of hypotheses. In
∞
order to conclude that a power series ∑ an x n is differentiable, and that term-by-term
n=0
differentiation is allowed, we need to know beforehand that the differentiated series
∑ nan x n-1 converges uniformly.
∞
Theorem 6.5.6. If ∑ an x n converges for all x ∈ (-R, R), then the differentiated series
n=0
∞
∑ nan x n-1 converges at each x ∈ (-R, R) as well. Consequently, the convergence is
n=1
uniform on compact sets contained in (-R, R).
Proof.
Exercise 6.5.5. □

We should point out that it is possible