Math 254 Course Notes

(Gautier) Cole Killian

February 10, 2020

This is McGill’s undergraduate Math 254, instructed by Axel Hundemer. The formal
name for this class is ”Honors Analysis 1”. You can find this and other course notes here:
https://colekillian.com/course-notes

Contents

1 Learning to Tex 2
√ √ √
1.1 ex. ∀x ≥ y ≥ 0 ∶ x − y ≥ x − y . . . . . . . . . . . . . . . . . . . . . . . . 4

2 Order Relations 4
2.1 Order Relations: . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4
2.1.1 Proof: . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5
2.2 Homework: Read 2.1 on your own . . . . . . . . . . . . . . . . . . . . . . . 6

3 Absolute Values 6
3.1 Discussing properties of absolute value . . . . . . . . . . . . . . . . . . . . . 6
3.2 Proof . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
3.3 The triangle inequality . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
3.4 Proof . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
3.5 Proof . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
3.6 Moving on. Absolute values are needed in the following definition: . . . . 7
3.7 Proof . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7

4 Supremum and Infimum 8

4.1 Examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8
4.2 Examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8

5 Lecture 10-02 8
5.1 Open and Closed Sets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8

6 Lecture 10-07 11

7 Limit laws 11

8 Monotone Sequences 16
8.1 Euler’s constant . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18
8.2 Subsequences . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19

9 10-28 19
9.1 Criterion for the divergence of sequences . . . . . . . . . . . . . . . . . . . 20

1
Cole Killian (February 10, 2020) Notes 2019-09-09

10 Tutorial 10-30 23
10.1 e . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 23

11 10-30 24
11.1 Properties of lim sup, lim inf . . . . . . . . . . . . . . . . . . . . . . . . . . . 27

12 11-06 28
12.1 Divergence to infinity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
12.2 Chapter 4: Limits of functions . . . . . . . . . . . . . . . . . . . . . . . . . . 29

13 Lecture 11-11 32
13.1 Limit Laws . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 35

14 Lecture 11-13 37
14.1 Limits and Inequalities . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
14.2 Criteria for non-existence of limits of functions . . . . . . . . . . . . . . . . 39
14.3 One-sided limits (Brief) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40
14.4 Chapter 5: Continuity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41

15 Lecture 11-18 41
15.1 Topological consequences of continuity . . . . . . . . . . . . . . . . . . . . . 46

16 Lecture 11-20 46
16.1 Preservation of compactness . . . . . . . . . . . . . . . . . . . . . . . . . . . 46

17 Lecture 11-25 46

18 Lecture 11-27 50
18.1 Application of Heine-Borel . . . . . . . . . . . . . . . . . . . . . . . . . . . . 50

19 Lecture 12-02 55

20 Lecture 12-03 57
√
20.1 Another method for proving that x is uniformly continuous on [0, ∞[. 58
20.2 Differentiation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59
20.3 Relationship Between Lipschitz Continuity and Differentiability . . . . . 61

21 Sequences 62

§1 Learning to Tex
Definition - The Cartesian Product: A × B of A and B = {(a, b) ∶ a ∈ A, b ∈ B}
i.e. R × R = R2 [0, 1] × [0, 2]
Definition - Functions in Calculus: Let D and E be sets; a function f ∶ D → E is a rule
that takes an input from D and assigns to it an output in E.
⎛w ⎞ p
⎛1⎞ ⎜ x ⎟⎛ ⎞
⎜2⎟ ⎜ ⎟⎜q ⎟
⎝3⎠ ⎜ y ⎟⎝ ⎠
⎝z ⎠ r

2
Cole Killian (February 10, 2020) Notes 2019-09-16

In modern math we define a function f ∶ D → E as a subset f of D × E s.t. ∀x ∈ D

there exists EXACTLY ONE y ∈ E s.t. (x, y) ∈ f . Functions are thus just sets, there is
thus just one fundamental concept (sets) we need to consider.
ex: f ∶ {−1, 0, 1} → {−1, 0, 1}
x ”maps to” x2
image vs. codomain
{(−1, 1), (0, 0), (1, 1)} = f
Definition - a function f ∶ D → E is called injective or one-to-one if x1 ≠ x2 ⇒ f (x1 ) ≠
f (x2 ) Everything gets mapped to its own unique point. Equivalently: f (x1 ) = f (x2 ) ⇒
x1 = x2
Definition - f ∶ D → E is called surjective or ”onto” if ∀y ∈ E ”there exists” x ∈ D ∶
f (x) = y
Definition - f ∶ D → E is called bijective if f is both injective and surjective
let f ∶ D → E, A ⊂ D then f (A) = {f (x) ∶ x ∈ A} ⊂ E is called the image of A under f
Definition - let f: D → E, B ⊂ E, then f −1 (B) = {x ∈ D ∶ f (x) ∈ B} ⊂ D is called the
inverse image of B under f
CAUTION: The inverse image f −1 (B) makes sense whether or not f is invertible!
ex: f ∶ {−1, 0, 1} → {−1, 0, 1}
x ”maps to” x2
image vs. codomain
{(−1, 1), (0, 0), (1, 1)} = f
note that f is NOT injective (because (f (1) = f (−1) and is thus not invertible. none
the less, inv. images make sense.
f 1 = {−1, 1} f 0 = {0} f −1 = {} = ∅
ex: let f ∶ D → E be bijective.
then f −1 ({y0 }) = {x0 } where f (x0 ) = y0
inv. function: f −1 (y0 ) = x0
inv. image: f −1 ({y0 }) = {x0 }
Theorem (i): let f ∶ D → E, A, B ⊂ D then (a) f (A ∪ B) = f (A) ∪ f (B)
(b) f (A ∩ B) ⊂ f (A) ∩ f (B)
(ii) letf ∶ D → E, A, B, ⊂ E then (a) f −1 (A ∪ B) = f −1 (A) ∪ f −1 (B)
(b) f −1 (A ∩ B) = f −1 (A) ∩ f −1 (B)
(ii)(a) will be shown in the tutorials (b) assign 1
we will prove (i):
(a) we have to show that the 2 sets f (A ∪ B) and f (A) ∪ f (B) are equal
Proof: let y ∈ f (A ∪ B) ⇒ ”there exists” x ∈ A ∪ B ∶ y = f (x) ⇒ ”there exists” x ∈
A ∶ y = f (x)v ”there exists” x ∈ B ∶ y = f (x) ⇒ y ∈ f (A)vy ∈ f (B) ⇒ y ∈ f (A) ∪ f (B) ⇒
f (A ∪ B) ⊂ f (A) ∪ f (B)
proof part 2:
let y ∈ f (A) ∪ f (B)
y ∈ f (A)vy ∈ f (B)
”there exists” x ∈ A ∶ y = f (x)v ”There exists” x ∈ B ∶ y = f (x)
”there exists” x ∈ A ∪ B ∶ y = f (x)
⇒ y ∈ f (A ∪ B)
f (A) ∪ f (B) ⊂ f (A ∪ B)
⇒ f (A ∪ B) = f (A) ∪ f (B)
Pitfall of proof in the wrong direction.

3
Cole Killian (February 10, 2020) Notes 2019-09-16
√ √ √
§1.1 ex. ∀x ≥ y ≥ 0 ∶ x−y ≥ x− y
√ √ √ √
⇔ x − y ≥ ( x − y)2 = x − 2 x y + y

§2 Order Relations
Def: A Relation on a set S is a subset of S × S
ex. (1) Equality: S set, x, y ∈ S. x ∼ (in relation with)y if x = y as a subset of S × S :
{(x, x) ∶ x ∈ S }
(2) ≤ on N
x ∼ y if x ≤ y ; {(x, y) ∶ x ≤ y}

§2.1 Order Relations:

Def: Let S be a set. A relation ∼ on S is called an order relation if it satisfies the following:
(1) ∀x ∈ S ∶ x ∼ x REFLEXIVITY
(2) ∀x, y ∈ S ∶ x ∼ y ∧ y ∼ x ⇒ x = y ANTI-SYMMETRY
(3) ∀x, y, z ∈ S ∶ x ∼ y and y ∼ z ⇒ x ∼ z TRANSITIVITY
ex.1 ”≤” on N or on R
Checking the axioms is straightforward
(1)
∀x ∈ R ∶ x ≤ x
(2)
∀x, y ∈ R ∶ x ≤ y ∧ y ≤ x ⇒ x = y
(3)
∀x, y, z ∈ R ∶ x ≤ y ∧ y ≤ z ⇒ x ≤ z
ex.2 Let S bea family of sets. Then ”⊂ ” is an order relation on S.
let A, B, C ∈ S

A ⊂ A ✓ REFLEXIVITY
A ⊂ B ∧ B ⊂ A ⇒ A = B ANTI-SYMMETRY
A ⊂ B ∧ B ⊂ C ⇒ A ⊂ C TRANSITIVITY

Cardinality of Sets
Finite Sets
Def: Let A be a finite set. Then ∣A∣ gives the number of elements of A; this is also called
the CARDINALITY of A.
Next we are going to learn to use functions to determine which of two sets is bigger.
Theorem: Let A, B be finite and non-empty sets. Then:
(a) ∣A∣ ≤ ∣B∣ IFF ∃f ∶ A → B s.t. f is injective.
(b) ∣A∣ ≥ ∣B∣ IFF ∃f ∶ A → B s.t. f is surjective.
(c) ∣A∣ = ∣B∣ IFF ∃f ∶ A → B s.t. f is bijective.

4
Cole Killian (February 10, 2020) Notes 2019-09-16

§2.1.1 Proof:
(a) ”⇒ ” let ∣A∣ ≤ ∣B∣. Let A = {a1 , . . . , an }, B = {b1 , . . . , bk } where n ≤ k. Define f ∶ A → B.
By: f (ai ) = bi ∀1 ≤ i ≤ n
This is a proper def since n ≤ k. Then f is injective by construction.
”⇐”. Assume ∃f ∶ A → B injective.
Let A = {a1 , . . . , an }. Then f (a1 ), . . . , f (an ). These are pairwise distinct and are
contained in B. Therefore B contains at least n elements. Therefore ∣A∣ ≤ ∣B∣.
(b) Exercise
” ⇒ ”if ∣A∣ = ∣B∣, f (ai ) = bi ∀1 ≤ i ≤ n is a bijection.
” ⇐ ” let f ∶ A → B be bijective especially, f isinjective
Furthermore, f −1 ∶ B → A is bijective and thus injective ⇒(a) ∣B∣ ≤ ∣A∣ ⇒ANTI-SYMMETRY
∣A∣ = ∣B∣

Generalizing for infinite sets.

def: let A, B be sets. We say that ∣A∣ ≤ ∣B∣ if ∃f ∶ A → B s.t. f is injective. (Note a bit of
cheating because in this definition we do not define Cardinality). (Note just because it
looks like less than or equal does not mean it shares all the properties of less than or
equal.)
”≤ ” is indeed an order relation on any fam ily of sets.
∣A∣ ≤ ∣A∣ because id ∶ A → A (The identity map) is injective REFLEXIVITY
let ∣A∣ ≤ ∣B∣ ∧ ∣B∣ ≤ ∣C∣ then ∃f ∶ A → B (injective) and ∃g ∶ B → C (injective) then
g ○ f ∶ A → C is injective as a composition of two injective maps (see tutorials). Thus
∣A∣ ≤ ∣C∣ which gives us transitivity TRANSITIVITY
ANTI-SYMMETRY. let ∣A∣ ≤ ∣B∣ ∧ ∣B∣ ≤ ∣A∣. ∃f ∶ A → B (injective) and ∃g ∶ B → A
(injective). This is the same as saying there is a surjective function from A to B. But this
doesn’t necesate bijectivity because they could be two different mappings. It is therefore
NOT intuitive that it follows from this that ∃h ∶ A → B bijective. This is true but NOT
easy to prove.
Theorem : Cantor, Bernstein, Schroder
If ∣A∣ ≤ ∣B∣ ∧ ∣B∣ ≤ ∣A∣ ⇒ ∣A∣ = ∣B∣ ANTI-SYMMETRY
Look up this proof on wikipedia it is very clever.
In other words, ”≤ ” is an order relation on any fam ily of sets.
Note: composition of injective maps is injective.
ex.1 Consider ∣N∣ and ∣N0 ∣
We have to find a bijective mapping f from one set to the other.
N = 1, 2, 3, . . .
N0 = 0, 1, 2, 3, . . .
Then n → n − 1 is a bijective map from N → N0 ⇒ ∣N∣ = ∣N0 ∣. But note that N is not
a proper subset of N0 . This cannot happen for ANY finite sets! (having both equal
cardinality and one being an improper subset of the other).
ex. ∣Z∣ = ∣N∣ ”ZAHL (I don’t know what this is here for)”
This example is a bit harder.
−3, −2, −1, 0, 1, 2, 3.
0 → 1, 1 → 2, −1 → 3, 2 → 4, −2 → 5, . . .
Consider f ∶ N → Z,
⎧
⎪
⎪n, if n is even
n → ⎨ 2 n−1
⎪
⎩− 2 , if n is odd
⎪
which is a bijection.

5
Cole Killian (February 10, 2020) Notes 2019-09-23

Review exercise: explicitely calculate f ∶ Z → N

Note: There are infinitely many levels of infinity.
The set of rational numbers: between any two numbers there are an infinite number of
numbers between the two. But it turns out that cardinality of rational numbers is the
same as natural numbers. Rational numbers are countable.
Real numbers are not countable. There is no bijection from real numbers to natural
numbers. He was proved that there cannot possibly be which was very hard.
Review: power set. power set changes level of infinity

Def: A set S is called countably infinite if ∣S∣ = ∣N∣

A set S is called countable if S is either finite or countably infinite.
ex. N, N0 , Z are all countably infinite.
Theorem: Let S be a subset of the natural numbers. Then S is countable (because natural
numbers are countable). Therefore it is either finite or countably infinite.
Proof: If S if finite, it is countable. Nothing more to show.
Now assume that S is countably infinite. S infinite ⇒ S ≠ ∅ ⇒ S has a least element
a1 .
S minus a1 is not empty. remove a2 , a3 .
A = {a1 , a2 , a3 , . . . } ⊂ S where a1 < a2 and pairwise distinct.
⇒A⊂S⊂N
(Actually A = S; Prove this! we don’t need it for this proof)
n → an is a bijection from N to A
⇒ ∣A∣ = ∣N∣

A ⊂ S ⊂ N ⇒ ∣A∣ ≤ ∣S∣ ≤ ∣N∣

⇒ ∣A∣ = ∣S∣ = ∣N∣ ⇒

A is countable (actually, countably infinite.)
Theorem let f ∶ N → S be surjective. Then S is countable. Proof; next class

§2.2 Homework: Read 2.1 on your own

§3 Absolute Values
Definition: Let x ∈ R, then
⎧
⎪
⎪x, x≥0
∣x∣ = ⎨
⎪
⎩−x,
⎪ x≤0
√
Note that ∣x∣ = x2

§3.1 Discussing properties of absolute value

Theorem
(a) ∀x, y ∈ R ∶ ∣x ∗ y∣ = ∣x∣ ∗ ∣y∣
(b) Let a > 0. Then ∣x∣ ≤ a ⇔ −a ≤ x ≤ a
(c) ∀x ∈ R ∶ −∣x∣ ≤ x ≤ ∣x∣
Review: Math notation so that I can confidently write it myself instead of copying from
the board. This will propably improve my retention.

6
Cole Killian (February 10, 2020) Notes 2019-09-23

§3.2 Proof
√ √ √ √
(a) ∣xy∣ = (xy)2 = x2 y 2 = x2 y 2 = ∣x∣∣y∣ ✓
(b) ”⇒ ” Let ∣x∣ ≤ a. First case: x ≥ 0. If this is true, then if follows that x = ∣x∣ ≤
a ⇒ x ≤ a and − a ≤ 0 ≤ x ⇒ −a ≤ x ≤ a✓. Second case: x < 0. If this is true then
−x = ∣x∣ ≤ a ⇒ x ≥ −a ⇒ −a ≤ x and x ≤ 0 ≤ a ⇒ x ≤ a ⇒ −a ≤ x ≤ a. Combining these
cases gives that −a ≤ x ≤ a in all cases.
(b) ”⇐ ”. Let −a ≤ x ≤ a ⇒ a ≥ −x ≥ −a ⇒ −a ≤ −x ≤ a. Because ∣x∣ = x or ∣x∣ = −x, it
follows that −a ≤ ∣x∣ ≤ a ⇒ ∣x∣ ≤ a
(c) Let a ≡ ∣x∣ ≥ 0, then ∣x∣ ≤ a = ∣x∣. Also, it follows from (b) that −a ≤ x ≤
a which can also be seen as − ∣x∣ ≤ x ≤ ∣x∣.

§3.3 The triangle inequality

About estimating absolute values of sums. Very important to analysis. Possibly most
important in all of mathematics.
∀x, y ∈ R ∶ ∣x + y∣ ≤ ∣x∣ + ∣y∣

§3.4 Proof
By Previous theorem part c we have −∣x∣ ≤ x ≤ ∣x∣ and − ∣y∣ ≤ y ≤ ∣y∣. The trick to the
proof involves adding these inequalities together.
This gives −(∣x∣ + ∣y∣)Let this be ”-a” ≤ x + y ≤ (∣x∣ + ∣y∣)Let this be ”a” . It follows from
previous theorem part b that ∣x + y∣ ≤ a = ∣x∣ + ∣y∣ ⇒ ∣x + y∣ ≤ ∣x∣ + ∣y∣. This theorem (the
triangle inequality) is used to find the upper bounds of sums.
Next theorem helps with lower bounds:
Theorem - ∀x, y ∈ R ∶ ∣x − y∣ ≥ ∣x∣ − ∣y∣ and ∣x − y∣ ≥ ∣y∣ − ∣x∣. This one is called the triangle
inequality for sums.

§3.5 Proof
∣x∣ = ∣x − y + y∣ ≤ ∣x − y∣ + ∣y∣ ⇒ ∣x∣ − ∣y∣ ≤ ∣x − y∣ ⇒ ∣x − y∣ ≥ ∣x∣ − ∣y∣ .
Interchange x and y (to avoid redoing the proof): ∣y−x∣ = ∣x−y∣ ≥ ∣y∣−∣x∣ ⇒ ∣x−y∣ ≥ ∣y∣−∣x∣
Remark: ∣x − y∣ ≥ ∣x∣ − ∣y∣ and ∣x − y∣ ≥ ∣y∣ − ∣x∣ can be combined to ⇒ ∣x − y∣ ≥ ∣∣x∣ − ∣y∣∣.
This final equation looks nice but can be hard to but into practice. It is normally easier
to pick the correct of the other two equations.
Theorem - Generalized Triangle Inequality. Let x1 , . . . , xn ∈ R, then ∣x1 + ⋅ ⋅ ⋅ + xn ∣ ≤
∣x1 ∣ + ⋅ ⋅ ⋅ + ∣xn ∣
Proof of this is on assignment 3.

§3.6 Moving on. Absolute values are needed in the following definition:
Definition:  neighborhood
Let  > 0 and let a ∈ R, the  neighborhood of a is defined as V (a) ≡ {x ∈ R ∶ ∣x − a∣ ≤ }
∣x − a∣ <  ⇔ − < x − a <  ⇔ a −  < x < a + .. This leads to V (a) =]a − , a + [
Theorem - if x ∈ V (a) for all  > 0, then x = a

§3.7 Proof
Assume that x ≠ a and find a contradiction.
First case: x > a. Let  = x − a > 0, then a +  = x ⇒ x ∋]a − , a + [= V (a)
Second case: x < a. Let  = a − x. Prove the rest yourself.

7
Cole Killian (February 10, 2020) Math 254 Notes

This theorem implies the following.

⋂ V (a) = {a}
>0

§4 Supremum and Infimum

Def: Let s ⊂ R, s ≠ ∅. We say that:
S is bounded from above if ∃u ∈ R such that ∀s ∈ S ∶ s ≤ u. Upper bound follows same
idea.

§4.1 Examples
(1) S = [0, 1[.
Then 1, 2, π, 1.5 are all upper bounds for S, and 0, -1, . . . are lower bounds for S.(
This answers my question about whether or not an upper or lower bound has to be right
at the bound. )
(2) A = [1, ∞[ is not bounded from above.
Definition: Let S ⊂ R, S ≠ ∅, S is bounded from above. s ∈ R is called the SUPREMUM
or Least upper bound of S. Symbolically: s = suprS if:
(1) s is an upper bound for S. (2) ∀t upper bounds of S, s ≤ t.
Similary for Infimum. Definition: Let S ⊂ R, S is bounded from below. A number
u ∈ R is called the infimum of S if u is a lower bound of S and ∀t lower bounds of S, u ≥ t

§4.2 Examples
S = [0, 1[. Claim that infS = 0. Proof: 0 is indeed lower bound of S ✓. Let v be any
lower bound for S. This lower bound cannot be positive because if it was 0 < v and so it
wouldn’t be a lower bound. ⇒ v ≤ 0 ⇒ 0 is the infimum of S.
No supremum in this case (THIS IS WHAT I THOUGHT INITIALLY BUT I WAS
WRONG). Claim: suprS = 1. Proof: 1 is an upperbound of S ✓. Let v be any upper
bound of S. If we assume that v is less than 1, we get contradiction that v is not an
upper bound of S. Therefore v ≥ 1. Therefore 1 = supr(S).
Questions: Given any non empty set S ⊂ R bounded from above, must there be a
supremum? Same idea of question for bounded below infimum. Complicated answers to
these questions. Postpone this to next class.

§5 Lecture 10-02
§5.1 Open and Closed Sets
Definition 5.1. Open interval does not contain any of its boundary points. Closed
interval contains all of its boundary points.

8
Cole Killian (February 10, 2020) Math 254 Notes

Theorem 5.2
Every open interval is open. This is not self evident because definition of open is
very specific.

Proof. Let I be an open interval. We need to show that I is always open.

1. Case: I = ]a, ∞]
Let x ∈ I be arbitrary. Let  = x − a. Then V (x) = ]x − , x + ] = ]a, 2x − a] ⊂
]a, ∞]. i.e. V (x) ⊂ ]a, ∞] ⇒ I is open.

2. Case: I = ]−∞, b]. Do yourself. Let x ∈ I be arbitrary. Let  = b − x. Then

V (x) = ]x − , x + [ = ]2x − b, b[ ⊆ ]−∞, b[. Therefore I is open.

3. Case: I = ]a, b[.

Let  = min {x − a, b − x} > 0.
Then V (x) = ]x − , x + [.
Note that x +  ≤ x + (b − x) = b and x −  ≥ x − (x − a) = a ⇒ ]x − , x + ] ⊂ ]a, b[
. Therefore I is open and therefore any open interval is open.

Theorem 5.3
Every closed interval is closed.

Proof. Let I be a closed interval. We need to show that R ∖ I is open.

1. Case: I = [a, ∞] ⇒ R ∖ I = ]−∞, a[ which as an open interval is open ⇒ I is

closed.

2. Case: I = [−∞, b]. do yourself. R ∖ I = ]b, ∞[ which is an open interval ⇒ I is

closed.

3. Case: I = [a, b] ⇒ R ∖ I = ]−∞, a] ∪ ]b, ∞]. Union of open with open is open
so I is closed. Therefore any closed interval is closed.

9
Cole Killian (February 10, 2020) Math 254 Notes

Theorem 5.4

a. Let J be an index set and let uj be open for all j ∈ J. Then

⋃ uj
j∈J

is open. ”Arbirary unions of open sets are open.

Proof. Let u = ⋃j∈J uj . Let x ∈ u be arbitrary ⇒ ∃j ∈ J such that x ∈ uj open
⇒ ∃ > 0 ∶ V (x) ⊂ uj ⊂ U . Can’t follow. Basically uses definition of union and
definition of openness.

b. Let u1 , . . . , un be open. Then

n
⋂ uj
j=1

is open. ”Finite intersections of open sets are open.

c. Finite unions of closed sets are closed. Let v1 , . . . , vn be closed, then ⋃nj=1 vj is
closed.
Proof. Let v1 , . . . , vn be closed, then R ∖ v1 , . . . , R ∖ vn are all open. ⇒ R ∖ v1 ∩
⋅ ⋅ ⋅ ∩ R ∖ vn is open. By demorgans law this equals R ∖ (v1 ∪ ⋅ ⋅ ⋅ ∪ vn ) is closed.
⇒ v1 ∪ ⋅ ⋅ ⋅ ∪ vn is closed. (because closed is comp lement of open)

d. Arbitrary intersections of closed sets are closed. Let J be an arbitraryIndex

set and let ∀j ∈ Jvj be closed, then ∩j∈J vj is closed.
Proof. Let vj be closed for all j ∈ J. ⇒ R ∖ vj is open. ⇒ ∪j∈J (R ∖ vj ) is open.
Demorgans law gives us that R ∖ ∩j∈J vj is open. Therefore ∩j=1 vj is closed.

Example 5.5
Every finite subset of R is closed.

Proof. Let’s first consider {x}. For some x ∈ R, {x} = ]x, x] is closed. Finite unions
of singleton sets are thus closed ⇒ all finite subsets of R are closed.

Example 5.6
S1 = { n1 ∶ n ∈ N} is NOT closed.

Proof. Assume it is closed. Then the comp lement u is open. we have 0 ∈ u, but
every  neighborhood V intersects S1 and is thus not contained in u ⇒ u is not
open.

Example 5.7
S2 = { n1 ∶ n ∈ N} ∪ {0} is closed.

Definition 5.8. Boundary.

Let S ⊂ R. A point x ∈ R is called a boundary point of S if every epsilon neighborhood
cenetered around x intersects both S and the comp lement of S.

10
Cole Killian (February 10, 2020) Math 254 Notes

A point x ∈ R is not a boundary point of S if ∃ > 0 ∶ V (x) ∩ S = ∅ ∨ V (x) ∩ (R ∖ S) = ∅.

Example 5.9
S = ]a, ∞] Claim:

§6 Lecture 10-07

Example 6.1

Definition 6.2. Let S ⊆ R. The interior S̊ ”S with dot on top” or int(S) is defined as:

S̊ = ⋃ U
U ⊂S, U open

Note that S̊ is open as a union of open sets. It is the largest open subset of S.

Definition 6.3. Let S ⊂ R. The closure S̃ of S is defined as:

S̃ = ⋂ V
V ⊃S, V closed

Note that S̃ is closed as an intersection of closed sets.

Theorem 6.4
Let S ⊂ R. Then S̊ = S ∖ δS

Proof. 1. ” ⊆ ”. Let x ∈ S̊ ⇒ ∃U open , U ⊂ S with x ∈ U .

Thus ∃ > 0s.t.V (x) ⊂ U ⊂ S.

Theorem 6.5
Let S ⊂ R. Then R ∖ S̃ = int(R ∖ S).

§7 Limit laws

11
Cole Killian (February 10, 2020) Math 254 Notes

Example 7.1

n
an =
4n
Show that lim(an ) = 0 Try using bernoulli but here it doesn’t help much.

4n = (1 + 3)n ≥ 1 + 3n
⇒ ∣an − 0∣ = n
4n ≤ n
1+3n → 1
3 ≠0

n
Unfortunately 1+3n does not converge to 0 so this estimate is too weak to be
useful. Note: This argument can be save (see next assignment).

Different approach: We’ll show that 4n ≥ n2 for all n ∈ N

Proof by Induction. .
n = 1: 41 = 4 ≥ 1 = 12
n → n + 1 ∶ Assume that 4n ≥ n2 , then

4n+1 = 4 ⋅ 4n ≥ 4 ⋅ n2 = 2n2 + n2 + n2 = 2n2 + (n + 1)2 + (n − 1)n − 2

= (2n2 − 2) + (n − 1)2 + (n + 1)2 ≥ (n + 1)2
⇒ 4n ≥ n2 ∀n ∈ N

Thus ∣an − 0∣ = 4nn ≤ nn2 ≤ 1

n →0
Therefore lim(an ) = 0

Theorem 7.2
Every convergent sequence is bounded.

Proof. Let (an ) be a sequence with lim(an ) = L, and let  = 1.

Then ∃N ∈ N ∀n ≥ N ∶ ∣an − L∣ <  = 1

⇒ ∣an ∣ = ∣(an − L) + L∣ ≤ ∣an − L∣ + ∣L∣ < 1 + ∣L∣ ∀n ≥ N

This proves that when n ≥ N , an is bounded.

Now let M = max{∣a1 ∣, ∣a2 ∣, . . . , ∣aN −1 ∣, 1 + ∣L∣}

Then ∣an ∣ ≤ M for all n ∈ N.

Remark 7.3. The convergence condition is essential. The sequence (n) = (1, 2, 3, . . . )
is unbounded.

12
Cole Killian (February 10, 2020) Math 254 Notes

Theorem 7.4
Let (an ), (bn ) be convergent sequences. Then (an + bn ) is convergent with lim(an +
bn ) = lim(an ) + lim(bn )

Proof. Let a = lim(an ), b = lim(bn ). Let  > 0.

∣an + bn − (a + b)∣ = ∣(an − a) + (bn − b)∣ ≤ ∣an − a∣ + ∣bn − b∣

Since lim(an ) = a, ∃N1 ∈ N ∀n ≥ N1 ∶ ∣an − a∣ < /2

Similarly, because lim(bn ) = b, ∃N2 ∈ N ∶ ∀n ≥ N2 ∶ ∣bn − b∣ < 2 .

Let N = max{N1 , N2 }. Then

 
∀n ≥ N ∶ ∣an − a∣ < ∧ ∣bn − b∣ <
2 2

Therefore
 
∣an + bn − (a + b)∣ = ∣(an − a) + (bn − b)∣ ≤ ∣an − a∣ + ∣bn − b∣ < + =  ∀n ≥ N
2 2

Thus (an + bn ) converges and lim(an + bn ) = a + b = lim(an ) + lim(bn )

This is supposed to be relatively simple.

Example 7.5

n+1 1 1
lim( ) = lim(1 + ) = lim(1) + lim( ) = 1 + 0 = 1
n n n

13
Cole Killian (February 10, 2020) Math 254 Notes

Theorem 7.6
Let (an ), (bn ) be convergent. Then (an bn ) converges and lim(an bn ) = lim(an ) ⋅
lim(bn )

Proof. Let a = lim(an ), b = lim(bn ). Let  > 0.

∣an bn − ab∣ = ∣an bn − abn + abn − ab∣

= ∣(an − a)bn + a(bn − b)∣
≤ ∣an − a∣∣bn ∣ + ∣a∣∣bn − b∣

Because (bn ) converges, (bn ) is bounded by a previous theorem. Thus ∃M1 > 0 such
that ∣bn ∣ ≤ M for all n ∈ N.

∣an bn − ab∣ ≤ M1 ⋅ ∣an − a∣ + ∣a∣ ⋅ ∣bn − b∣

Let M = max{M1 , ∣a∣}
≤ M ∣an − a∣ + M ∣bn − b∣ = M [∣an − a∣ + ∣bn − b∣]

Since lim(an ) = a, ∃N1 ∈ N ∀n ≥ N1 ∶ ∣an − a∣ < /2M

Similarly, because lim(bn ) = b, ∃N2 ∈ N ∶ ∀n ≥ N2 ∶ ∣bn − b∣ < 

2M .

Let N = max{N1 , N2 }. Then

 
∀n ≥ N ∶ ∣an − a∣ < ∧ ∣bn − b∣ <
2M 2M

Therefore
  
∣an bn − ab∣ ≤ M [∣an − a∣ + ∣bn − b∣] < M ( + )=M⋅ =  ∀n ≥ N
2M 2M M

Thus (an bn ) converges and lim(an bn ) = ab = lim(an ) ⋅ lim(bn )

This can be applied to finitely many sequences.

Example 7.7
lim( n1b ) = 0 for all k ∈ N

Proof. Because ( n1 ) converges to 0, lim( n1k ) = lim( n1 )⋯ lim( n1 ) = 0

Note 7.8. Special case where (bn ) is constant. i.e. bn = c for all n ∈ N. Let (an ) be
convergent with lim(an ) = a. Then lim(c ⋅ an ) = lim(c) ⋅ lim(an ) = c ⋅ lim(an )

14
Cole Killian (February 10, 2020) Math 254 Notes

Example 7.9

n−1 1 1 1
lim( ) = lim(1 − ) = lim(1 + (− )) = lim(1) + lim(− )
n n n n
1 1
= 1 + lim(−1 ⋅ ) = 1 + −1 ⋅ lim( ) = 1 + −1 ⋅ 0 = 1
n n

Theorem 7.10
In general, if (an ), (bn ) converges, then (an − bn ) converges and lim(an − bn ) =
lim(an ) − lim(bn )

Proof.

lim(an − bn ) = lim(an + (−bn )) = lim(an ) + lim(−bn ) = lim(an ) + −1 lim(bn ) = lim(an ) − lim(bn )

Theorem 7.11
Let (an ) be convergent with lim(an ) ≠ 0 and an ≠ 0 ∀n ∈ N. Then ( an1 ) converges
and lim( a1n ) = 1
lim(an )

Proof. Let lim(an ) = a, a ≠ 0. Let  > 0. Then

1 1 a − an ∣an − a∣ ∣an − a∣ 1
∣ − ∣=∣ ∣= < = ⋅ ∣an − a∣ = 0
an a an ⋅ a ∣an ∣ ⋅ ∣a∣ k∣a∣ k∣a∣

By conv. criterion, ( a1n ) converges to 1

a

Lemma 7.12
Let (an ) be convergent with an ≠ 0 ∀n ∈ N and lim(an ) = a ≠ 0. Then there
exists M > 0 such that ∣ a1n ∣ ≤ M ∀n ∈ N.

Proof. Let a = lim(an ) and  = 12 ∣a∣. Then ∃n ∈ N such that ∣an − a∣ <  = 21 ∣a∣ for
all n ≥ N , then ∣an ∣ = ∣a − (a − an )∣ ≥ ∣a∣ − ∣an − a∣ > ∣a∣ − 12 ∣a∣ = 12 ∣a∣ > 0 ∀n ≥ N
Let k = min{∣a1 ∣, ∣a2 ∣, . . . , ∣an−1 ∣, 12 ∣a∣} > 0, then ∣an ∣ > k > 0 ∀n ∈ N
1 1
⇒∣ ∣< =M ∀n ∈ N
an k

Theorem 7.13
Let (an ), (bn ) by convergent where ∀n ∈ N bn ≠ 0 and lim(bn ) ≠ 0. Then an
bn converges
lim(an )
and lim( abnn ) = lim(bn )

15
Cole Killian (February 10, 2020) Math 254 Notes

§8 Monotone Sequences
Recall 8.1. Monotone means increasing or decreasing in the non strict sense.

Theorem 8.2
Let (xn ) be a monotone sequence. Then (xn ) is convergent if and only if it is
bounded. This is useful because it is easier to check whether or not a sequence is
bounded than to check whether or not it is convergent.

Proof. Assume that (xn ) is increasing. We will show that (xn ) converges ot the
supremum.

What is the supremeum of a sequence. We take all the numbers and consider
it a set in R and then find the supremeum. x ∶= sup {x1 , x2 , x3 , . . . }.
´¹¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹¸ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¶
∶=S
Let  > 0, then x− is not an upper bound of S. Thus ∃N ∈ N such that x− < xN ≤ X
but (xn ) is increasing. We also have x −  < xN ≤ xN +1 ≤ xN +2 ≤ ⋅ ⋅ ⋅ ≤ x. i.e.
∀n ≥ N ∶ x −  < xn ≤ x

⇒ xn ∈]x − , x] for all n ≥ N ⊆]x − , x + [= V (x). i.e. ∀n ≥ N ∶ xn ∈ V (x).

Thus (xn ) converges to x ∶= sup{x1 , x2 , . . . }. The case that (xn ) is decreasing is left
as an exercise.

16
Cole Killian (February 10, 2020) Math 254 Notes

Example 8.3
x1 = 1, xn+1 = 21 xn + 2

Show that xn convreges and determine its limit. We will show that (xn ) is in-
creasing and bounded; by monotone convergence theorem, (xn ) converges. Lastly,
we will show that lim(xn ) = 4.

Proof. (xn ) is bounded from above by 4. We’ll show this using induction.

n = 1: 1≤4 ✓

n → n + 1: Assume that xn ≤ 4. Then xn+1 = 21 xn + 2 ≤ 1

2 ⋅4+2=4 ✓

Therefore (xn ) is bounded from above by 4.

Proof. Proving that (xn ) is increasing. Consider xn+1 −xn = 12 xn +2−xn = 2− 12 xn ≥ 0.

⇒ ∀n ∈ N xn+1 − xn ≥ 0
⇒ ∀n ∈ N xn+1 ≥ xn

i.e. (xn ) is increasing.

By showing that (xn ) is bounded from above and increasing, we know that (xn ) is
convregent by the monotone convergence theorem. Now to find where it converges.

Let x ∶= lim(xn ).
1
∀n ∈ Nxn+1 = xn + 2
2
1 1 1
⇒ lim(xn+1 ) = lim( xn + 2) = lim(xn ) + 2 = x + 2
2 2 2
1
⇒x= x+2
2
1
⇒ x=2⇒x=4
2

Note 8.4. It is essential for this argument that we knew in advance that (xn ) is
convergent.

We’ve now shown that lim(xn ) = 4.

Example 8.5
√
Exercise for the reader: x1 = 1. xn+1 = 2 + xn .

Prove that (xn ) converges to 2.

17
Cole Killian (February 10, 2020) Math 254 Notes

§8.1 Euler’s constant

Consider the squence xn = (1 + n1 )n and yn = (1 + n1 )n+1 .

We will show that (xn ) increases and that (yn ) decreases.

Proof. (xn ) is increasing. We have to show that ∀n ∈ N ∶ xn ≤ xn+1 . i.e. that

1 n 1 n
(1 + ) ≤ (1 + ) +1
n n+1
1 n+1 1
⇔ (1 + ) ≥ (1 + )n
n+1 n
√
1 1
⇔1+ ≥ (1 + )n
n+1

n+1 n

Recall the inequality of the algebraic and geometric mean. If a1 , a2 , . . . , an ≥ 0 , then

a1 + ⋯ + an √
≥ n a1 × ⋯ × an
n
Let a1 = ⋅ ⋅ ⋅ = an = 1 + 1
n and an+1 = 1. Then
√
√ 1 n
a1 × ⋯ × an × an+1 =
) (1 +
n+1 n+1

n
a1 + ⋯ + an + an+1 n(1 + n ) + 1 n + 1 + 1 n + 2
1
1
and = = = =1+
n+1 n+1 n+1 n+1 n+1
√
Thus, by AGM-inequality, 1 + n+1
1
≥ n+1 (1 + n1 )n .

Proof. Now to show that yn is decreasing. Similar strategy, but take inverse to reverse
inequality.

It follows from the above proofs that, Claim:

∀n, k ∈ N ∶ xn < yn

Definition 8.6.
1 n 1
e ∶= lim ((1 + ) ) = lim ((1 + )n+1 )
n n
In analysis 2, you’ll see that
∞
1
e= ∑
n=0 n!

From which it can be shown that e is irrational.

Estimates for e. Since (xn ) is increasing and (yn ) is decreasing, we have that ∀n ∈
N ∶ xn ≤ e ≤ yn .
⎧
⎪x6 ≥ 5 = 2.5
5 ⎪
<e<3⇐⎨ 2
2 ⎪
⎪y < 3
⎩ 5

18
Cole Killian (February 10, 2020) Math 254 Notes

§8.2 Subsequences
Definition 8.7. Let n1 < n2 < n3 < . . . be natural numbers and let (xn ) = (x1 , x2 , x3 , . . . )
be a sequence. Then (xnk ) = (xn1 , xn2 , xn3 , . . . ) is called a subsequence of (xn ).

Example 8.8
Let (x1 , x2 , x3 , . . . ) be a sequence. Then (x1 , x3 , x5 , x7 , . . . ) is called the subsequence
of odd indices; here nk = 2k − 1.

Likewise, (x2 , x4 , x6 , x8 , . . . ) is called the subsequence of even indices; here nk = 2k.

Theorem 8.9
Let (xn ) be convergent. Then every subsequence (xnk ) of (xn ) also converges to
the same limit.

Proof. Next class.

Example 8.10
Let 0 < a < 1 ; consider (an ). We will show that lim(an ) = 0. Note that (an )
is decreasing and is bounded from below. By monotone convergence theorem,
(an ) converges.

Let x ∶= lim(an ). Now consider the subsequence of even terms (a2n ). By

the theorem above, this subsequence converges and has the same limit. i.e.
lim(a2n ) = x.

On the other hand, we can rewrite this as

lim((an )2 ) = [lim(an )]2 = x2 = x

⇒ x2 − x = 0
⇒ x(x − 1) = 0

This means that either x = 0 or x = 1. But a3 < a2 < a1 = a < 1 ⇒ x < 1 ⇒ x = 0.

§9 10-28

19
Cole Killian (February 10, 2020) Math 254 Notes

Theorem 9.1
Let (xn ) be a convergent sequence, then every subsequence of (xn ) also converges
to the same limit. i.e. lim(xnk ) = lim(xn ).

Lemma 9.2
If n1 < n2 < n3 < . . . where nk ∈ N for all k, then nk ≥ k for all k ∈ N.

Proof. By induction.

k = 1 ∶ Base case where nk ≥ k.

k → k + 1 : Assume that nk ≥ k. Then

nk+1 > nk ≥ k ⇒ nk+1 > k ⇒ nk+1 ≥ k + 1

Thus nk ≥ k for all k ∈ N.

Proof. Let x ∶= lim(xn ). Let  > 0, then ∃N ∈ N ∀n ≥ N ∶ ∣xn − x∣ < .

Since nk ≥ k, by the lemma, we also have that ∣xnk − x∣ <  for all k ≥ N , since
nk ≥ k ≥ N .

Thus (xnk ) converges to x.

§9.1 Criterion for the divergence of sequences

Theorem 9.3 (1)

Let (xn ) be a sequence such that (xn ) has a subsequence (xnk ) that diverges.

Proof. If (xn ) were convergent, (xnk ) would converge, but it doesn’t. Thus (xn )
diverges.

Theorem 9.4
Let (xn ) be a sequence such that there exists two subsequences (xnk ) and (xnj )
that converge to different limits, then (xn ) diverges.

Proof. If (xn ) was convergent to x1 , then (xnk ) and (xnj ) would converge to x1 ;
but they don’t. Thus (xn ) diverges.

20
Cole Killian (February 10, 2020) Math 254 Notes

Example 9.5
xn = (−1)n . Consider the subsequences of the even and odd terms (x2n ) and (x2n−1 ).

x2n = (−1)2n = 12n = 1. i.e. (x2n ) is a constant sequence and lim(x2n ) = 1.

Similarly, x2n−1 = (−1)(−1)2n = −1. i.e. (x2n−1 ) is a constant sequence and

lim(x2n−1 ) = −1.

According to one of the criterion for the divergence of sequences theorems, (xn )
diverges.

Example 9.6
xn ∶ 1, 1, 2, 12 , 3, 31 , 4, 14 . Then x2n−1 ∶ 1, 2, 3, 4, . . . . Which diverges, thus (xn ) diverges.

Example 9.7
√
xn = n n ; Prove that (xn ) converges to 1.

1st step: (xn ) is eventually decreasing.

1
xn+1 (n + 1) n+1
= 1
xn nn
xn+1 n(n+1) 1 n + 1 n 1 1 1 3
⇒( ) = ⋅ = ⋅ (1 + )n ≤ ⋅ e < ≤ 1
xn n n n n n n

As long as n ≥ 3. Thus (xn ) is decreasing for all n ≥ 3.

Furthermore, (xn ) is bounded from below by 1. Thus (xn ) is bounded and eventu-
ally decreasing ⇒ (xn converges by monotone convergence theorem. Let x ∶= lim(xn ).

Second step: Show that x = 1.

Consider the subsequence (x2n ) of even terms.
√ √ √ √ √
x2n = 2n ⇒ x22n = 2n = 2 ⋅ n n = 2 ⋅ xn
2n n n n

Thus
√ √
lim(x22n ) = lim( 2 ⋅ xn ) = lim( 2) ⋅ lim(xn )
n n

´¹¹ ¹ ¹ ¹ ¹ ¹ ¹ ¸ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¶
=1
lim(x22n ) = (lim(x2n ))2
⇒ x2 = x ⇒ x − x = 0 ⇒ x(x − 1) = 0
2

⇒ x = 0 ∨ x = 1. but xn ≥ 1 ∀n ∈ N
⇒x=1

21
Cole Killian (February 10, 2020) Math 254 Notes

Theorem 9.8 (Bolzano - Weirstrass)

Let (xn ) be a bounded sequence. Then (xn ) has a convergent subsequence.

Proof. Since (xn ) is bounded, ∃µ > 0 such that xn ∈ [−M, M ] for all n ∈ N.
´¹¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹¸ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¶
=I1

Divide I1 into two subintervals of equal width. At least one of these subinter-
vals contains infinitely many terms of (xn ). Choose this one of these intervals and
call it I2 .

Divide I2 into 2 subintervals of equal width. At least one of them, called I3

contains infinitely many terms of (xn ). Etc...

We obtain an infinite sequence I1 ⊇ I2 ⊇ I3 ⊇ ⋯ of closed and bounded inter-

vals. By the nested interval property of R we know that the intersection over all of
these intervals is not empty. i.e. ∩n∈N In ≠ ∅.

Let x ∈ ∩n∈N In . We will now show that there exists a subsequence (xnk ) of (xn )
with lim(xnk ) = x.

Let n1 ∈ N be arbitrary. We know that xn1 ∈ I1 because all elements are in I1 .

I2 contains infinitely many terms of (xn ). Thus there exists n2 > n1 such that
xn2 ∈ I2 . The same goes for I3 ; etc...

We obtain n1 < n2 < n3 < . . . such that xnk ∈ Ik for all k ∈ N.

We also have that x ∈ Ik for all k ∈ N. This gives that ∣xnk − x∣ ≤ ∣Ik ∣ where
∣I1 ∣ = 2M, ∣I2 ∣ = M, ∣I3 ∣ = M
2 , ....

2M 4M 1
⇒ ∣Ik ∣ = k−1
= k ⇒ ∣xnk − x∣ ≤ 4M ⋅ ( )k
2 2 2
for all k ∈ N. By convergence criterion, lim(xnk ) = x ; especially, (xnk ) converges.
Corner stone of the proof is the nested interval property of R.

Definition 9.9. Let (xn ) be a sequence and let (xnk ) be a convergent subsequence. Let
x ∶= lim(xnk ). Then x is called an accumulation point or a subsequential limit (point) of
(xn ).

Example 9.10
xn = (−1)n . The accumulation points of (xn ) are +1 and −1.

Example 9.11
Let xn be an enumeration of Q. Every real number is an accumulation point because
Q is dense in R.

22
Cole Killian (February 10, 2020) Math 254 Notes

Theorem 9.12
Let (xn ) be a sequence. x ∈ R is an accumulation point of (xn ) iff ∀ > 0 ∶ V (x)
contains infinitely many terms of (xn ).

Proof.

(⇒) Let x be an accumulation point of (xn ). Thus there exists a subsequence (xnk )
of (xn ) with lim(xnk ) = x. Then ∃k ∈ N ∶ ∀k ≥ N xnk ∈ V (x). Thus V (x)
contains infinitely many terms of (xn ).

(⇐) Let x ∈ R be such that ∀ > 0 ∶ V (x) contains infinitely many terms of (xn ).
Let  ∶= 1. Then V1 (x) contains infinitely many terms of (xn ) .Let n1 ∈ N such
that xn1 ∈ V1 (x).

Let  ∶= 12 . Then V 1 (x) contains infinitely many terms of (xn ). Thus ∃nl > n1
2
such that xn2 ∈ V 1 (x).
2
⋮
 = k1 . Then V 1 (x) contains infinitely many terms of (xn ) thus ∃nk > nk−1
k
such that xnk ∈ V 1 (x)
k

Since n1 < n2 < n3 < . . . , we obtain a subsequence (xnk ) of (xn ) with

xnk ∈ V 1 (x). Now let  > 0 and let k > 1 ⇔ k1 <  ⇒ xnk , xnk+1 , xnk+2 , ⋅ ⋅ ⋅ ∈
k
V 1 (x) ⊆ V (x).
k

xnk ∈ V (x) ∀k ≥ K ⇒ xnk converges to x

§10 Tutorial 10-30

§10.1 e

Example 10.1

1.
1 −n
lim(1 − ) =e
n

2.
1 n 1 2n 1 1
(1 + ) = ((1 + ) ) 2 = (e) 2
2n 2n
Because (1 + 2n )
1
is a subsequence of (1 + n1 ) which converges to e.
n
3. (1 + n2 ) 2 is not a subsequence of (1 + n1 )n . It’s the other way around.

Let a > 0. Pick x1 > 0. Let xn+1 = 21 (xn + xn )

a
>0
√
Prove that xn → a.

23
Cole Killian (February 10, 2020) Math 254 Notes

§11 10-30

Theorem 11.1
A bounded sequence converges if and only if it has exactly one accumulation point.

Proof.

(⇒) Let (xn ) be convergent. x ∶= lim(xn ). Then every subsequence (xnk ) of (xn )
converges to x. Thus x is the only accumulation point of (xn ).

(⇐) Let (xn ) be a bounded sequence which has only one accumulation point x. We
will show that (xn ) converges to x. Assume that this is not the case.

Convergence: ∀ > 0, ∃N ∈ N ∶ ∀n ≥ N, ∣xn − x∣ < 

Negation: ∃ > 0 ∶ ∀N ∈ N, ∃n ≥ N ∶ ∣xn − x∣ ≥ 

Thus ∃ infinitely many n ∈ N such that ∣xn − x∣ ≥ 0 .

Let n1 < n2 < n3 < . . . such that ∀k ∈ N ∶ ∣xnk − x∣ ≥ 0 .

Consider the subsequence (xnk ) of (xn ) ⇒ (xnk ) is bounded because (xn ) is

bounded.

By Bolzano-weierstrass, (xnk ) has a convergent subsequence (xnkj ). Let

∼ x ∶= lim(xnkj ). Since it is a subsequence of (xn ) which has only one accumu-
lation point. It follows that ∼ x = x.

Thus lim(it) = x and ∀j ∈ N, ∣it − x∣ ≥ 0 CON T RADICT ION

Thus our assumption was wrong which proves that (xn ) converges to x.

24
Cole Killian (February 10, 2020) Math 254 Notes

Theorem 11.2
Let (xn ) be a bounded sequence and let A be the set of all accumulation points of
(xn ). Then A ≠ ∅ and A is compact (i.e. A is closed and bounded).

Proof. By BOLZANO-WEIERSTRASS, (xn ) has at least one convergent subse-

quence. Its limit is an accumulation point of (xn ) ⇒ A ≠ ∅.

A is bounded: (xn ) is bounded i.e. ∃M > 0 such that ∀n ∈ N, −M ≤ xn ≤ M .

Let x ∈ A be arbitrary. Then ∃ subsequence (xnk ) of (xn ) with x = lim(xnk ).

We have that ∀k ∈ N ∶ −M ≤ xnk ≤ M ⇒ −M ≤ x ≤ M .
⇒ x ∈ [−M, M ] for all accumulation points x of (xn ).
⇒ A ⊆ [−M, M ] ⇒ A is bounded.

A is closed: Let x ∈ R ∖ A i.e. x is not an accumulation point. Thus ∃ > 0 ∶ V (x)

contains at most finitely many terms of (xn ).
Let t ∈ V (x). V (x) is open. Thus ∃˜
 > 0 ∶ V˜(t) ⊆ V (x).
Thus V˜(t) contains at most finitely many terms of (xn ). Thus t is not an accumula-
tion point ⇒ no point in V (x) is an accumulation point of (xn ) ⇒ V (x) ⊆ R ∖ A.
Thus R ∖ A is open ⇒ A is closed.

We’ve just seen that the set of all accumulation points of a bounded sequence (xn ) is ≠ 0,
closed, and bounded.

Since A is bounded, it has a supremum and an infimum. Both sup and inf are boundary
points. A is closed so it contains sup and inf. Therefore sup(A) is the Maximum of
A and inf(A) is the minimum of A. i.e. sup(A) is an accumulation point of (xn ), the
greatest accumulation point of (xn ). Similarly inf(A) is the least accumulation point of
(xn ).

Definition 11.3.

1. Let (xn ) be a bounded sequence. Then the greatest accumulation point of (xn ) is
called the LIMES SUPERIOR of (xn ). In symbols: lim sup(xn ).

2. The least accumulation point of (xn ) is called the LIMES INFERIOR of (xn ). In
symbols: lim inf(xn ).

25
Cole Killian (February 10, 2020) Math 254 Notes

Theorem 11.4
Let (xn ) be a bounded sequence. Then (xn ) is convergent if and only if

lim inf(xn ) = lim sup(xn )

Proof.

(⇒) Let x ∶= lim(xn ). Then every subsequence (xnk ) of (xn ) converges to x.

⇒ A = {x} ⇒ lim inf(A) = x = lim sup(A)

(⇐) Assume that lim inf(xn ) = lim sup(xn ) ∶= x.

A = {x}

i.e. (xn ) has only one accumulation point. By previous theorem, (xn ) con-
verges.

26
Cole Killian (February 10, 2020) Math 254 Notes

Example 11.5

1.
xn = (−1)n
Accumulation points are −1 and 1 ⇒ lim inf(xn ) = −1 and lim sup = 1. Espe-
cially, (−1)n diverges because lim inf ≠ lim sup.

2. Let (xn ) be an enumeration of Q ∩ [a, b] where a < b. We’ll show that lim inf = a
and that lim sup = b.
Proof. Let x > b. Let  ∶= b − x > 0. Then ∀n ∈ N, xn ∉ V (x) ⇒ x is not an
accumulation point of (xn ).

Let x ∈ [a, b] and let  > 0; consider V (x) =]x − , x + [. By the density
of Q in R, V (x) contains infinitely many rational numbers, especially, V (xn )
contains infinitely many terms of (xn ) ⇒ x is an accumulation point of (xn ).

x = a: By density of Q in R, ]a, a + [ contains infinitely many terms of

(xn ) ⇒ a is an accumulation point of (xn ). Similarly for x = b.

Therefore A ∶= [a, b] ⇒ lim inf(xn ) = a and lim sup(xn ) = b.

3. Find all accumulation points of the following sequence.

1 1 1 1 1 1
xn ∶ 1, 1, , 1, , , 1, , , , . . .
2 2 3 2 3 4
Claim: A = {0} ∪ {1, 12 , 31 , 14 , . . . }
Proof. For every k ∈ N, the constant sequence n1 , n1 , n1 is a subsequence of (xn ).
Thus
1 1 1 1
= lim( , , , . . . ) ∈ A
n n n n
and 1, 12 , 31 , 14 , . . . is a subsequence of (xn ). Thus

1 1
0 = lim(1, , , . . . ) ∈ A
2 3
Now let x > 1,  ∶= x − 1 > 0. Then ∀n ∈ N ∶ xn ∉ V (x) ⇒ x ∉ A.

Similarly, x ∉ A for all x < 0. Let 0 < x < 1; x ∉ A. Then ∃n ∈ N ∶ 1

n+1 < x < n1 .

Let  ∶= min{x − 1 1
n+1 , n − x} > 0. Then 1
n+1 ∉ V (x) ∨ 1
n ∉ V (x)

⇒ xn ∉ V (x) ∀n ∈ N
x is not an accumulation point of (xn )

Thus A = {0} ∪ { n1 ∶ n ∈ N}

§11.1 Properties of lim sup, lim inf

27
Cole Killian (February 10, 2020) Math 254 Notes

Theorem 11.6
Let (xn ) be a bounded sequence and let  > 0. Then ∃N ∈ N ∀n ≥ N ∶ xn ∈
] lim inf(xn ), lim sup(xn ) + [. i.e. at most finitely many terms of (xn ) have the
property that xn > lim sup(xn ) +  or xn < lim inf(xn ) − 

Proof. assignment 8

Theorem 11.7
Let (xn ) be a bounded sequence. Then lim sup(xn ) = lim(sup{xk ∶ k ≥ n}) and
lim inf(xn ) = lim(inf{xk ∶ k ≥ n}).

Remark 11.8. It is not clear initially whether this is well defined. We’ll prove this.

Let yn ∶= sup{xk ∶ k ≥ n}. Then (yn ) is bounded because (xn ) is bounded.

Let A, B be bounded with A ⊆ B. Then sup(A) ≤ sup(B).

Note 11.9. {xk ∶ k ≥ n + 1} ⊆ {xk ∶ k ≥ n}.

Therefore sup{xk ∶ k ≥ n + 1} ≤ sup{xk ∶ k ≥ n}.

Therefore (yn ) is bounded and decreasing and therefore converges.

Thus lim(sup{xk ∶ x ≥ n}) exists. A similar argument applies to lim(inf{xk ∶ k ≥ n}).

Proof. Examination material. This is the cutoff for the midterm exam. Next week
coshy sequences. 3.4 in the textbook. Important: This doesn’t mean that you don’t
have to remember the stuff from before. If you don’t know stuff from before you will
be closed. I used open and closed todayand left it to you to know what open and
closed means. It did not contain interior and closure so that is midterm 2 material.
And you need to know what boundary sets are in order to make sense of these things
but I won’t ask a separate question on these things.

§12 11-06
§12.1 Divergence to infinity
Definition 12.1. Let (xn ) be a sequence. We say that (xn ) diverges to +∞ if

∀M > 0, ∃N ∈ N, ∀n ≥ N ∶ xn > M

In symbols:
lim(xn ) = +∞

(xn ) diverges to −∞ if

∀M > 0(∃N ∈ N)(∀ ≥ N ) ∶ xn < −M

In symbols:
lim(xn ) = −∞

28
Cole Killian (February 10, 2020) Math 254 Notes

Remark 12.2. If lim(xn ) = +∞ or lim(xn ) = −∞, then the sequence diverges. The limit
laws thus do NOT apply.

Example 12.3
√
lim(n2 ) = +∞. Let M > 0. Then n2 > M ⇔ n > M.
√
Let N > M . Then ∀n ≥ N ∶ n2 ≥ N 2 > M ⇒ n2 > M for all n ≥ M ⇒ (n2 )
diverges to +∞.

Example 12.4
Let a > 1. Show that lim(an ) = +∞.

Since a > 1, b ∶= a − 1 > 0. Then a = 1 + b and an = (1 + b)n . Applying bernoulli’s:

M
(1 + b)n ≥ 1 + nb > nb > M ⇔ n >
b

Let N > M
b . Then ∀n ≥ N , we know that an > nb ≥ N b > M . Thus an diverges to
+∞.

§12.2 Chapter 4: Limits of functions

Preparatory definition:

Definition 12.5 (In A). Let A ⊆ R. A sequence (xn ) is said to be in A if ∀n ∈ N ∶ xn ∈ A.

Definition 12.6 (Cluster point). Let A ⊆ R. A point x ∈ R is called a cluster point of

A if:
∀ > 0 ∶ V (x) ∖ {x} ∩A ≠ ∅
´¹¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¸¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¶
Punctured neighborhood

Note 12.7. Notation for punctered neighborhoods:

V∗ (x) ∶= V (x) ∖ {x}

i.e. x is a cluster point of A if ∀ > 0 ∶ V∗ (x) ∩ A ≠ ∅.

Remark 12.8. Cluster points of A are not necessarily elements of A.

Definition 12.9 (Isolated Point). Let A ⊆ R. x ∈ A is called an isolated point of A if

∃ > 0 ∶ V∗ (x) ∩ A = ∅.

i.e. x is the only element of A that is in V (x).

29
Cole Killian (February 10, 2020) Math 254 Notes

Example 12.10
S ∶= {0} ∪ { n1 ∶ n ∈ N}.

Claim: 0 is the only cluster point of S. All points 1

n ∶ n ∈ N are isolated points of S.

0 is a cluster point. Let  > 0. Then V (0) contains infinitely many numbers of the
form n1 because lim( n1 ) = 0. Thus 0 is a cluster point of S.

Let x ≠ 0. Then ∃ > 0 ∶ V∗ (x) ∩ S = ∅ (left as exercise). Especially, such  > 0 exists
for all x = n1 . Thus every n1 is an isolated point of S.

Example 12.11
Let A ∶= Q. Then every real number is a cluster point of A.

Proof. Let x ∈ R be arbitrary and let  > 0. Since Q is dense in R, V (x) contains
infinitely many rational numbers. Thus V∗ (x) contains at least one (in fact infinitely
many) rational numbers. i.e.

V∗ (x) ∩ A ≠ ∅ ⇒ x is a cluster point of A

Exercise 12.12. Let I be an interval. Then the set of all cluster points of I is I

Theorem 12.13
Let A ⊆ R. Then x ∈ R is a cluster point of A if and only if there exists a sequence
(xn ) in A ∖ {x} with lim(xn ) = x.

Proof.

(⇒) Let x be a cluster point of A.

Let  ∶= 1. Then V∗ (x) ∩ A ≠ ∅. Let x1 ∈ V1∗ (x) ∩ A.

Let  ∶= 21 . Then V∗ (x) ∩ A ≠ ∅. Let x2 ∈ V 1∗ (x) ∩ A.

2

We obtain a sequence (xn ) in A ∖ {x} with ∀n ∈ N ∶ xn ∈ V 1∗ (x) ∩ A.

n

Let  > 0. Let N > 1

 ⇔ 1
N < . Then

∀n ≥ N ∶ xn ∈ V 1∗ (x) ∩ A ⊆ V 1∗ (x) ⊆ V 1 (x) ⊆ V 1 (x) ⊆ V (x).

n n n N

i.e. ∀n ≥ N ∶ xn ∈ V (x) ⇒ (xn ) converges to x.

(⇐) Let (xn ) be a sequence in A ∖ {x} such that lim(xn ) = x. Let  > 0. Then
∃N ∈ N, ∀n ≥ N ∶ xn ∈ V (x). But since xn ∈ A ∖ {x}, xn ≠ x. This means that
xn ∈ V∗ (x) and xn ∈ A. Thus ∀n ≥ N ∶ xn ∈ V∗ (x)∩A. Thus v∗ (x)∩A ≠ ∅ ⇒ x
is a cluster point.

30
Cole Killian (February 10, 2020) Math 254 Notes

Theorem 12.14
Let A ⊆ R. Let x be a cluster point of A. Then x ∈ A.

Proof. Let x be a cluster point of A. By previous theorem, ∃(xn ) is A ∖ {x} such

that lim(xn ) = 0.

Since ∀n ∈ N ∶ xn ∈ A ∖ {x}. We have that ∀n ∈ N ∶ xn ∈ A ⊇ A ∖ {x}.

Since A is closed, lim(xn ) ∈ A (see assignment 6).

Definition 12.15 (The limit of a function: Sequential Definition).

Let f ∶ A ⊆ R → R. Let x0 ∈ R, we say that L is a limit of f as x → x0 . In symbols:

L = lim f (x)
x→x0

if for all sequences (xn ) in A ∖ {x0 } with lim(xn ) = x0 , we have that lim(f (xn )) = L.

Example 12.16
Let
x2
f ∶ R ∖ {0} → R, x →
∣x∣

Note that for x ≠ 0 we have that

x2
= ∣x∣
∣x∣
Claim: limx→0 f (x) = 0.

Let (xn ) be a sequence such that xn ≠ 0 for all n ∈ N and such that lim(xn ) = 0. We
need to show that (f (xn )) converges to 0. Note that f (xn ) = ∣xn ∣.

Let  > 0. Since lim(xn ) = 0, there exists (N ∈ N)(∀n ≥ N ) ∶ ∣xn − 0∣ = ∣xn ∣ < .

Thus ∀n ≥ N ∶ ∣∣xn ∣ − 0∣ = ∣∣xn ∣∣ = ∣xn ∣ <  ⇒ lim(f (xn )) = 0. Thus:

lim f (x) = 0
x→x0

Example 12.17
Let f ∶ R ∖ {0} → R where x → x1 . Let x0 ≠ 0. Show that

1
lim f (x) =
x→x0 x0

Proof. Let (xn ) be a sequence in R ∖ {0, x0 } with lim(xn ) = x0 . Then lim(f (xn )) =
lim( x1n ) = lim(x
1
n)
= x10 .

31
Cole Killian (February 10, 2020) Math 254 Notes

Example 12.18
Let f ∶ Z → R where x → 0. Let L ∈ R be arbitrary. Then

lim f (x) = L
x→0

Since 0 is an isolated point in Z, there doesn’t exist any sequence in Z ∖ {0} that
converges to 0. Thus all sequences (xn ) in Z ∖ {0} that converge to 0 hvae that
property that
lim f (x0 ) = L
x→0

Thus limx→0 f (x) = L for any L ∈ R.

Remark 12.19. This example shows that we should avoid isolated points when
considering limits.

Theorem 12.20
Let f ∶ A → R where x0 is a cluster point of A.

Then: if f has a limit as x approaches x0 , then this limit is uniquely determined.

Proof. Let L1 , L2 be limits of f as x approaches x0 . Then ∃(xn ) is A ∖ {x0 } with

lim(xn ) = x0 . Because f has a limit at x0 , lim(f (xn )) exists and L1 = lim(f (xn )) =
L2 .

§13 Lecture 11-11

Definition 13.1 (Weierstrass). The  definition of the limit of a function.

Let f ∶ A ⊆ R → R, and x0 ∈ R. We say that L is a limit of f as x approaches x0

if:
∀ > 0, ∃δ > 0, ∀x ∈ A ∶ 0 < ∣x − x0 ∣ < δ ⇒ ∣f (x) − L∣ < 

This can be rewritten in several ways:

1.
∀ > 0, ∃δ > 0 ∶ x ∈ Vδ∗ (x0 ) ∩ A ⇒ f (x) ∈ V (L)

2.
∀ > 0, ∃δ > 0 ∶ f (Vδ∗ (x0 ) ∩ A) ⊆ V (L)

32
Cole Killian (February 10, 2020) Math 254 Notes

Theorem 13.2
Let f ∶ A → R be a function. Let x0 ∈ R and L ∈ R. Then:

lim f (x) = L
x→x0

in the sequential sense if and only if this holds in the  − δ sense.

Proof.

1. ” − δ ⇒ Sequential”:

Let  > 0. Let δ > 0 be such that f (Vδ∗ (x0 ) ∩ A) ⊆ V (L).

Let (xn ) be a sequence in A ∖ {x0 } with lim(xn ) = x0 . Then ∃N ∈ N, ∀n ≥ N ∶

xn ∈ Vδ (x0 ).
We also have that xn ≠ x0 and xn ∈ A for all n ∈ N. This implies that

∀n ≥ N ∶ xn ∈ Vδ∗ (xn ) ∩ A
⇒ ∀n ≥ N ∶ f (xn ) ∈ V (L)
⇒ (f (xn )) converges toL

2. ”Sequential ⇒  − δ”:

Assume that the sequential definition holds but that there exists  > 0 for
which ulno δ > 0 exists that satisfies  − δ.

i.e. assume that f (Vδ∗ (x0 ) ∩ A) ⊆/ V (L) for all δ > 0. Especially:

δ=1∶ f (V1∗ (x0 ) ∩ A) ⊆/ V (L)

⇒ ∃x1 ∈ V1∗ (x0 ) ∩ A such that f (x1 ) ∉ V (L)

1
δ= ∶ f (V 1∗ (x0 ) ∩ A) ⊆/ V (L)
2 2

⇒ ∃x2 ∈ V 1 (x0 ) ∩ A such that f (x2 ) ∉ V (L)

∗
2

We then obtain a sequence (xn ) such that xn ∈ V 1∗ (x0 ) ∩ A but f (xn ) ∉ V (L).
n

Thus lim(xn ) = x0 but (f (xn )) does not converge to L. This contradicts

the sequential definition of limit.

Thus ∃δ > 0 such that f (Vδ∗ (x0 ) ∩ A) ⊆ V (L).

33
Cole Killian (February 10, 2020) Math 254 Notes

Example 13.3
Show that:

lim x2 = x20
x→x0

Solution.

1. Sequential:

Let (xn ) be a sequence in R ∖ {x0 } with lim(xn ) = x0 . Then lim(f (xn )) =

lim(x2n ) = [lim(xn )]2 = x20

2.  − δ:

Let  > 0. Let δ > 0 be arbitrary for now and assume that ∣x − x0 ∣ < δ.
Then

∣f (x) − f (x0 )∣ = ∣x2 − x20 ∣ = ∣x − x0 ∣ ⋅∣x + x0 ∣

´¹¹ ¹ ¹ ¹¸ ¹ ¹ ¹ ¹ ¶
<δ
⇒< ∣x + x0 ∣δ = ∣x − x0 + 2x0 ∣δ ≤ (∣x − x0 ∣ + 2∣x0 ∣)δ
< (δ + 2∣x0 ∣)δ < (δ + 2∣x0 ∣) ⋅ δ < 

Assume that δ < 1. Then ∣f (x) − f (x0 )∣ < (δ + 2∣x0 ∣)δ < (1 + 2∣x0 ∣)δ < 

Now let:

δ < min(1, )
1 + 2∣x0 ∣
Then if ∣x − x0 ∣ < δ, then ∣f (x) − f (x0 )∣ <  ⇒

lim x2 = x20
x→x0

34
Cole Killian (February 10, 2020) Math 254 Notes

Example 13.4

1
f ∶ R ∖ {0} → R, x →
x
Let x0 ∈ R ∖ {0}. Show that:
1 1
lim =
x→x0 x x0
Solution

Solution.

1. Sequential:

Let (xn ) be a sequence in R ∖ {0, x0 } with lim(xn ) = x0 . Then:

1 1 1
lim(f (xn )) = lim( )= =
xn lim(xn ) x0

2. With  − δ :

Let  > 0. Let δ > 0 be arbitrary for now. Let ∣x − x0 ∣ < δ. Then:
1 1 x0 − x
∣f (x) − f (x0 )∣ = ∣ − ∣ = ∣ ∣
x x0 xx0
∣x − x0 ∣ δ
= <
∣x∣∣x0 ∣ ∣x∣∣x0 ∣

Let δ < 12 ∣x0 ∣. Then for all x with ∣x − x0 ∣ < δ we have:

1 1
∣x∣ = ∣(x − x0 ) + x0 ∣ ≥ ∣x∣ − ∣x − x0 ∣ > ∣x0 ∣ − ∣x0 ∣ = ∣x0 ∣
2 2
i.e. ∣x∣ ≥ 21 ∣x0 ∣ Now:

δ δ 2δ
∣f (x) − f (x0 )∣ < ≤ 1 = 2 <
∣x∣∣x0 ∣ 2 ∣x0 ∣∣x0 ∣ x0
x20
⇔δ< ⋅
2
Let δ < min( 21 ∣x0 ∣, 21 x20 ). Then if ∣x − x0 ∣ < δ, we have that:

1 1
∣f (x) − f (x0 )∣ <  ⇒ lim =
x→x0 x x0

§13.1 Limit Laws

35
Cole Killian (February 10, 2020) Math 254 Notes

Theorem 13.5 (Limit of a Sum is the Sum of the Limits)

Let f, g ∶ A → R, and x0 be a cluster point of A. Assume that limx→x0 f (x) = L1 and
that limx→x0 g(x) = L2 .

Then

lim [(f + g)(x)] = lim [f (x) + g(x)] = L1 + L2

x→x0 x→x0
= lim f (x) + lim g(x)
x→x0 x→x0

i.e.
lim [(f + g)(x)] = lim f (x) + lim g(x)
x→x0 x→x0 x→x0

Proof. We’ll use the sequential criterion to prove this theorem. Let (xn ) be a
sequence in A ∖ {x0 } with lim(xn ) = x0 . Then

lim((f + g)(xn )) = lim(f (xn ) + g(xn ))

= lim(f (xn )) + lim(g(xn )) = L1 + L2 = lim f (x) + lim g(x)
x→x0 x→x0

Theorem 13.6 (Limit of a Product is the Product of the Limits)

Let f, g ∶ A → R and x0 be a cluster point of A. Assume that limx→x0 and limx→x0 g(x)
exist. Then:

lim [(f ⋅ g)(x)] = lim [f (x) ⋅ g(x)] = lim f (x) ⋅ lim g(x)
x→x0 x→x0 x→x0 x→x0

Proof. Let (xn ) be a sequence in A ∖ {x0 } with lim(xn ) = x0 . Then:

lim [(f ⋅ g)(x)] = lim(f (xn ) ⋅ g(xn )) = lim(f (xn )) ⋅ lim(g(xn )) = lim f (x) ⋅ lim g(x)
x→x0 x→x0 x→x0

Especially, let c ∈ R. Then

lim [c ⋅ f (x)] = c ⋅ lim f (x) Think of it as choosing g = c

x→x0 x→x0

Therefore:

lim [f (x) − g(x)] = lim [f (x) + (−1) ⋅ g(x)] = lim f (x) + lim[(−1)g(x)]
x→x0 x→x0 x→x0
= lim f (x) + (−1) lim g(x) = lim f (x) − lim g(x)
x→x0 x→x0 x→x0 x→x0
⇒ lim [f (x) − g(x)] = lim f (x) − lim g(x)
x→x0 x→x0 x→x0

36
Cole Killian (February 10, 2020) Math 254 Notes

Theorem 13.7
Let f, g ∶ A → R and x0 be a cluster point of A. Furthermore, let ∀x ∈ A, g(x) ≠ 0
and let limx→x0 f (x), limx→x0 g(x) exist where limx→x0 g(x) ≠ 0. Then:

f (x) limx→x0 f (x)

lim =
x→x0 g(x) limx→x0 g(x)

§14 Lecture 11-13

§14.1 Limits and Inequalities

Theorem 14.1 (Bounded Limit Theorem for Functions)

Let f ∶ A → R, and x0 bea cluster point of A. Assume that limx→x0 f (x) exists.

Furthermore, assume that ∃a, b ∈ R such that a ≤ f (x) ≤ b for all x ∈ A ∖ {x0 }.
Then a ≤ limx→x0 f (x) ≤ b.

Proof. Let limx→x0 f (x) = L. Then ∀(xn ) in A ∖ {x0 } with lim(xn ) = x0 , it holds
that lim(f (xn )) = L.

Since ∀n ∈ N ∶ xn ∈ A ∖ {x0 }, we have that

a ≤ f (xn ) ≤ b ⇒ a ≤ L = lim(f (xn )) ≤ b

®
Theorem from Chapter 3

⇒ a ≤ lim f (x) ≤ b
x→x0

37
Cole Killian (February 10, 2020) Math 254 Notes

Theorem 14.2 (Squeeze Theorem for Functions)

Let f, g, h ∶ A → R, and let x0 be a cluster point of A. Assume that

g(x) ≤ f (x) ≤ h(x)

For all x ∈ A ∖ {x0 }.

Furthermore, assume that

L ∶= lim g(x) = lim h(x)

x→x0 x→x0

Then the limit of f (x) as x → x0 exists and equals L.

Proof. Let (xn ) be a sequence in A∖{x0 } such that lim(xn ) = x0 . Then lim(g(xn )) =
L and lim(h(xn )) = L.

And since ∀n ∈ N ∶ xn ∈ A ∖ {x0 }, we know that

g(xn ) ≤ f (xn ) ≤ h(xn )

By the squeeze theorem for sequences it now follows that (f (xn ) converges to L.
Since this holds for any (xn ) in A ∖ {x0 } with lim(xn ) = x0 , it follows from sequence
criterion that

lim f (x) = L
x→x0

38
Cole Killian (February 10, 2020) Math 254 Notes

Example 14.3
Consider the following function :
1
f (x) ∶ R ∖ {0} where x → x ⋅ sin( )
x
Solution.
1 1
∣x ⋅ sin( )∣ = ∣x∣ ⋅ ∣ sin( )∣ ≤ ∣x∣
x x
1
⇒ −∣x∣ ≤ x sin( ) ≤ ∣x∣
x
for all x ∈ R ∖ {0}.

Note that

lim ∣x∣ = 0
x→x0
lim (−∣x∣) = − lim ∣x∣ = 0
x→x0 x→x0

Therefore, by squeeze theorem we have that

1 1
−∣x∣ ≤ x sin( ) ≤ ∣x∣ ⇒ lim (x sin( )) = 0
x ® x→x0 x
Squeeze Theorem

Example 14.4
Let f ∶ R+ → R and x → x3/2 . We want to find limx→0 x3/2 .

Restrict f to the interval [0, 1]. On this interval we have that

0 ≤ x ≤ x1/2
⇒ 0 ≤ x3/2 ≤ x

and limx→0 x = 0.

Therefore, by squeeze theorem,

0 ≤ x3/2 ≤ x ⇒ lim x3/2 = 0

® ® x→0
=0 =0

§14.2 Criteria for non-existence of limits of functions

39
Cole Killian (February 10, 2020) Math 254 Notes

Theorem 14.5 (Non-existence criteria where (f (xn )) diverges.)

Let f ∶ A → R and x0 be a cluster point of A. If ∃(xn ) in A ∖ {0} such that
lim(xn ) = x0 but such that lim(f (xn )) diverges, then limx→x0 f (x) DNE.

Proof. If limx→x0 f (x) would exist, then lim(f (xn ) = limx→x0 f (x) but f (xn )) di-
verges ⇒ limx→x0 f (x) DNE.

Theorem 14.6 (Non-existence criteria where (f (xn )) and (f (tn )) converge to different
limits)
Let f ∶ A → R and x0 be a cluster point of A. Assume that ∃(xn ), (tn ) in A ∖ {xn }
such that lim(xn ) = x0 = lim(tn ) and such that both (f (xn )) and (f (tn )) converge
but to different limits. Then limx→x0 f (x) does not exist.

Proof. Assume that limx→x0 f (x) = L. Then lim(f (xn )) = L = lim(f (tn )). Contra-
diction because lim(f (xn )) ≠ lim(f (tn )). Thus limx→x0 f (x) diverges.

Example 14.7
Let f ∶ R ∖ {0} → R and x → sin(1/x). Show that limx→0 f (x) DNE.

1. Solution using the 2-sequence criterion.

Choose (xn ) where xn ∶= πn 1

for all n ∈ N. Then f (xn ) = sin(πn) = 0 for
all n ∈ N. i.e. lim(f (xn )) = 0.

Now choose (tn ) where tn ∶= 1

π/2+2πn . Then f (tn ) = sin(π/2+2πn) = sin(π/2) = 1
for all n ∈ N.

⇒ lim(f (tn )) = 1 ≠ 0 = lim(f (xn ))

⇒ lim f (x) DNE
x→0

2. Solution using the 1-sequence criterion.

Let xn ∶= (2n−1)π/2
1
. Then lim(xn ) = 0 and f (xn ) = sin((2n − 1)π/2) = (−1)n
for all n ∈ N. i.e. (f (xn )) = ((−1)n ) which diverges!

⇒ lim f (x) DNE

x→0

§14.3 One-sided limits (Brief)

In calculus you’ve seen

lim f (x) and lim− f (x)

x→x0 + x→x0

How do we define these properly?

40
Cole Killian (February 10, 2020) Math 254 Notes

Definition 14.8 (Definition of limit from left and right). Let f ∶ A → R and x0 ∈ R.

lim f (x) ∶= f (x)

x→x+
0
∣A∩]x0 ,∞[
lim f (x) ∶= f (x)
x→x−
0
∣A∩]−∞,x0 [

Example 14.9
f ∶ R → R where x → ∣x∣. Determine limx→0+ f (x) and limx→0− f (x).

lim x = 0 ⇒ lim+ ∣x∣ = 0

x→0 x→x
lim x = 0 ⇒ lim− ∣x∣ = 0
x→0 x→x

Theorem 14.10 (Limit of function exists iff limits from left and right exists and are
equal)
Let f ∶ A → R and x0 be a cluster point of A. Then limx→x0 f (x) exists if and only
if limx→x+0 f (x) and limx→x−0 f (x) exist and are equal.

Proof. Assignment 11.

§14.4 Chapter 5: Continuity

Definition 14.11 (Defining a continuous function). Let f ∶ A → R and x0 ∈ A. We say
that f is continuous at x0 if

lim x → x0 f (x)

exists and is equal to f (x0 ). i.e limx→x0 f (x) = f (x0 ).

Remark 14.12. In the case that x0 is an isolated point, this definition should be read as
follows: f is continuous at x0 if it has a limit at x0 which equals f (x0 ). In other words,
all functions are continuous at all isolated points. Continuous is thus only interesting at
cluster points.

§15 Lecture 11-18

Definition of continuity: ∀ > 0 , ∃δ > 0 ∶ f (Vδ (x0 ) ∩ A) ⊆ V (f (x0 ))

Remark 15.1. Let x0 be an isolated point of A. Then any function f ∶ A → R is

continuous at x0 .

Proof. Let f ∶ A → R and let  > 0. Since x0 is an isolated point of A, ∃δ ∶ Vδ (x0 ) ∩ A =

{x0 }.

Then, f (Vδ (x0 ) ∩ A) = f ({x0 }) = {f (x0 )}. Thus f is continuous at x0 .

41
Cole Killian (February 10, 2020) Math 254 Notes

Theorem 15.2 (Algebraic Rules for Continuity)

Let f, g ∶ A → R and let x0 ∈ A be a cluster point of A. f, g is continuous at x0 , then:

(a) f + g is continuous at x0 .

(b) f ⋅ g is continuous at x0 .

(c) f − g is continuous at x0 .

(d) f /g is continuous at x0 if ∀x ∈ A, g(x) ≠ 0.

Proof.

(a) Let (xn ) be a sequence in A with lim(xn ) = x0 .

Since f and g are continuous at x0 , we have that lim(f (xn )) = f (x0 ) and
lim(g(xn )) = g(x0 ).

Thus,

lim((f + g)(x0 )) = lim(f (x0 ) + g(x0 ))

= lim(f (xn )) + lim(g(xn )) = f (x0 ) + g(x0 ) = (f + g)(x0 )
⇒ f + g is continuous at x0

Alternatively, we can use the limits of functions. f, g are continuous at x0 so

lim f (x) = f (x0 )

x→x0
lim g(x) = g(x0 )
x→x0

Thus

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

x→x0 x→x0
= lim f( x) + lim g(x) = f (x0 ) + g(x0 ) = (f + g)(x0 )
x→x0 x→x0
⇒ f + g is continous at x0

(b) Left as an exercise

(c) Left as an exercise

(d) Left as an exercise

42
Cole Killian (February 10, 2020) Math 254 Notes

Theorem 15.3 (Compositions of continuous functions)

Let f ∶ A → B, and g ∶ B → R where f (A) ⊆ B. Let x0 ∈ A, and let f be continuous
at x0 , and g is continuous at f (x0 ), then g ○ f is continuous at x0 .

Proof.

1. Proof with  − δ

Let  > 0. Because g is continuous at f (x0 ), we get that

∃ν > 0 such that g(Vν (f (x0 )) ∩ B) ⊆ V (g(f (x0 ))). (1)

And since f is continuous at x0 , we get that

∃δ > 0 such that f (Vδ (x0 ) ∩ A) ⊆ Vν (f (x0 )) (2)

Combining (1) and (2) we get that

(g ○ f )(Vδ (x0 ) ∩ A) = g(f (Vδ (x0 ) ∩ A) ⊆ g(Vν (f (x0 ) ∩ B) ⊆ V (g(f (x0 )))

⇒ (g ○ f )(Vδ (x0 ) ∩ A) ⊆ V ((g ○ f )(x0 ))

⇒ g ○ f is continuous at x0

2. Proof with sequential method

Let (xn ) be a sequence with lim(xn ) = x0 . Since f is continuous at x0 ,

we have that lim(f (xn )) = f (x0 ).

Because g is continuous at f (x0 ), we have that

lim(g(f (xn ))) = g(f (x0 ))

⇒ lim((g ○ f )(xn )) = (g ○ f )(x0 )

⇒ g ○ f is continuous at x0

43
Cole Killian (February 10, 2020) Math 254 Notes

Definition 15.4. A function f ∶ A → R is called continuous (on A ) if f is continuous at

all x0 ∈ A.

Example 15.5

1. x is continuous on R.

2. Because products of continuous functions are continuous, xn is continuous on

R for all n ∈ N.

Note also that if cn ∈ R, cn xn is continuous on R.

3. Since sums of continuous functions are continuous, every polynomial p(x) ∶=

a0 + a1 x + ⋯ + an xn is continuous on R.

4. Since quotients of continuous functions are continuous, wherever the denom-

P (x)
inator is non-zero, we have that all rational functions R(x) ∶= , P, Q
Q(x)
polynomials are continuous on R/N where N ∶= {x ∈ R ∶ Q(x) = 0}.
√ √ √
5. We’ve seen that lim x = x0 for all x0 ∈ R+0 . Thus is continuous on R+0 .
x→x0

6. sin and cos are continuous on R. See assignment 11.

44
Cole Killian (February 10, 2020) Math 254 Notes

Example 15.6 (Examples of discontinuous functions. sgn, Dirichlet, Thomae)

1.
⎧
⎪1, x>0
⎪
⎪
⎪
(x) ∶= ⎨0, x=0
⎪
⎪
⎪
⎪
⎩−1, x<0

(x)
1

−1

Let (xn ) be a sequence with xn > 0 for all n ∈ N and lim(xn ) = 0 (e.g. xn = 1/n.
Then (xn ) = 1 for all n ∈ N. Thus ((xn )) converges to 1.

But! (0) = 0 ≠ 1 = lim((xn )). Thus is discontinuous at 0.

2. Dirichlet’s Function. f ∶ [0, 1] → R where f is defined as follows:

⎧
⎪
⎪1, x∈Q
f (x) = ⎨
⎪
⎪ x∈R∖Q
⎩0,
Claim: f is discontinuous at all x0 ∈ [0, 1].
Proof. Proof by cases where x0 ∈ Q and x0 ∈ R ∖ Q :
a) Let x0 be rational. Because R ∖ Q is dense in R, we know that ∃(xn ) ∈
[0, 1] such that lim(xn ) = x0 and that ∀n ∈ N ∶ xn ∈ R ∖ Q.

Then ∀n ∈ N we have that f (xn ) = 0 ⇒ lim(f (xn )) = 0 ≠ 1 = f (x0 ).

b) Let x0 ∈ R ∖ Q. Because Q is dense in R, we know that ∃(xn ) ∈ [0, 1] with
lim(xn ) = x0 and ∀n ∈ N ∶ xn ∈ Q.

Then ∀n ∈ N ∶ f (xn ) = 1 ⇒ lim(f (xn )) = 1 ≠ 0 = f (x0 ).

3. Thomae’s Function Consider f ∶ [0, 1] → R such that

⎧
⎪
⎪1/q, x = n/q, gcd(n, q) = 1
f (x) = ⎨
⎪
⎪ x∈R∖Q
⎩0
Claim: f is continuous at all irrational numbers, but discontinuous at all
rational numbers.
45
Cole Killian (February 10, 2020) Math 254 Notes

§15.1 Topological consequences of continuity

Exercise.
1. Let I ⊆ R be an interval and let f ∶ I → R be continuous. Is f (I) an interval? (Yes,
we will see later)
2. If U ⊆ R is open and f ∶ U → R is continuous, is f (U ) open? (No. Find a
counterexample).
3. If V ⊆ R is closed, is f (V ) closed? (No)
4. If S ⊆ R is bounded, is f (S) bounded (No)
5. If C ⊆ R is compact (recall that this means closed and bounded), is f (C) compact?
Solution.
1. We will see later.
2. Let f ∶] − 1, 1[→ R where x → x2 . Then ] − 1, 1[ is open, but f (] − 1, 1[) = [0, 1[
which is not open.
3. f ∶ [1, ∞[→ R where x → 1/x. Then f ([1, ∞[) =]0, 1] which is not closed.
4. f ∶]0, 1] → R where x → 1/x. The domain of f is bounded. But (]0, 1]) = [1, ∞[ is
unbounded.
5.

§16 Lecture 11-20

§16.1 Preservation of compactness
We’ll need the following theorem:
A ⊆ R is closed iff every cauchy sequence in A has its limit in A.
Proof. Let A be closed and let (xn ) be a cauchy sequence in A. Assume that x0 ∶=

§17 Lecture 11-25

Definition 17.1. Let A ⊆ R and let c ∶= {Ui ∶ i ∈ I}, where I is an index set, Ui is open
for all i ∈ I.

Then c i scalled an open cover of A if A ⊆ Ui∈I Ui . i.e. every x ∈ A is contained.

If y ⊆ I such that {Uj ∶ j ∈ J}coloneqqϕ is still a cover of A, we say that ϕ′ is a
finite subcover of ϕ.

Example 17.2
Let A = [0, 1] and let ϕ ∶= {V1/2 (x) ∶ x ∈ [0, 1]}.

Then ϕ is an open cover of [0, 1] because

[0, 1] ⊆ ∪x∈[0,1] V1/2 (x) ∶ x ∈ [0, 1] ⊆ ]−1/2, 3/2[

46
Cole Killian (February 10, 2020) Math 254 Notes

47
Cole Killian (February 10, 2020) Math 254 Notes

Theorem 17.3 (Heine-Borel)

A ⊆ R is compact (closed and bounded) if and only if every open cover of A has a
finite subcover.

Proof.

⇒ Special Case: A is a closed and bounded interval [a, b] ∶= I0 . Assume that c

is an open cover of I0 that doesn’t have a finite subcover. Divide I0 into two
closed subintervals of equal width [a, c] and [c, b] where c = a+b
2 .

For at least one of these subintervals, ϕ does not have a finite subcover.
Otherwise, ϕ would have a finite subcover ϕ′ of [a, ϕ] and ϕ′′ of [ϕ, b]. Then
ϕ′ ∪ ϕ′′ would be a finite open cover of I0 , which doesn’t exist.

Let I1 be (one of) the subinterval(s) without finite subcover. Divide I1 into 2
closed subintervals of equal width. At least one of them doesn’t have A.

We obtain a nested sequence I0 ⊇ I1 ⊇ I2 ⊇ ⋯ of closed and bounded intervals.

Then
∩n∈N0 In ≠ ∅
by the nested interval property.

Let x0 ∈ ∩n∈N0 In . Then x0 ∈ I0 , thus ∃i ∈ I such that x0 ∈ Ui which is

open. Thus, ∃ > 0 ∶ V (x0 ) ⊆ Ui .

Claim: ∃n ∈ N0 ∶ In ⊆ V (x0 ).
Proof. ∣In ∣ = 1/2n ∣I0 ∣. Let n ∈ N0 such that 1/2n ∣I0 ∣ < .

Let x ∈ In be arbitrary. Then ∣ x − x0 ∣ ≤ 1/2n ∣I0 ∣ <  ⇒ x ∈ V (x0 ).

® ¯
∈In ∈In

⇒ In ⊆ V (x0 ). Now we have:

In ⊆ V (x0 ) ⊆ Ui

i.e. {Ui } covers In

ϕ has a finite (of length 1) subcover for In . CONTRADICTION.

⇒ ϕ does have a finite subcover.

General Case; A ⊆ R compact. ϕ open cover. Since A is bounded, ∃M > 0
such that A ⊆ [−M, M ]. Let U ∶= R/A which is open.
Consider ϕ′ ∶= ϕ ∩ {U }. Then ϕ′ covers R. Thus ϕ′ covers [−M, M ] which is
closed and bounded interval by special case.
By special case, ϕ′ has a finite subcover ϕ′′ . ϕ′′ may not be a subcover of ϕ
because ϕ′′ may contain U . However, if ϕ′′ should contain U , we can simply
remove it.
i.e. if U ∈ ϕ′′ , let ϕ′′′ = ϕ′′ /{U }. If U ∉ ϕ′′ , let ϕ′′′ ∶= ϕ′′ .
Since U = R/A, ϕ′′′ will still cover A. Thus we’ve obtained a finite subcover of
A.

48
Cole Killian (February 10, 2020) Math 254 Notes

Theorem 17.4
A ⊆ R is compact (closed and bounded) if and only if every open cover of A has a
finite subcover.

Proof.

⇐ Let A not be compact. We need to find an open cover of A without a finite

subcover. A not closed: assignment 12.

A unbounded
Let ϕ ∶= {Un ∶ n ∈ N} where Un ∶=] − n, n[. Then ϕ covers R and thus A.
Consider any finite subset m{Un1 , ⋯, Unk .

Remark 17.5. THe ”classical” definition of compacness is closed and bounded, however
this definition doesn’t generalize will beyond Rn since there isn’t even a notion of bound-
edness on general ”topological spaces” However, open covers still make perfect sense on
topological spaces. Thus, the def of compactness was revised to

Definition 17.6 (Modern definition of compactness). A is called compact if every open

cover of A has a finite subcover.

”Modern” heine borel becomes:

Definition 17.7. A ⊆ R is compact if and only if A is closed and bounded.

Applications of heine borel: It can often be useful to generalize ”local” properties of

functions to ”global” properties if the domain is compact.

Definition 17.8. f ∶ A → R is called locally bounded if ∀x0 ∈ A, ∃ > 0 ∶ f is bounded

on the domain V (x0 ).

Example 17.9
f ∶ ]0, ∞[ → R, x → 1/x.

f is bounded on any neighborhood about x0 that does not contain 0 is in its

boundary. Thus f is locally bounded, but not (globally) bounded!

However, this can’t happen if the domain is compact

49
Cole Killian (February 10, 2020) Math 254 Notes

Theorem 17.10
Let A ⊆ R be compact. f ∶ A → R be locall bounded. Then f is bounded (on A).

Proof. Let x ∈ A be arbitrary. f locally bounded ⇒ ∃x > 0 such that f is bounded
on interval Vx (x).

Then ϕ ∶= {Vx ∶ x ∈ A is an open cover of A. Since A is compact, ϕ has a fi-

nite subcover {Vx1 , ⋯, Vxn (xn )}.

On each of these n neighborhoods, f is bounded.

⇒ ∃M1 , ⋯, Mn ≥ 0

such that ∣f ∣(x) ≤ M1 , ⋯, ∣f ∣(x) ≤ Mn bounded on Vn (xn ).

Let M ∶= max{M1 , ⋯, Mn }. Then ∣f ∣(x) ≤ M, ⋯, ∣f ∣ ≤ M .

§18 Lecture 11-27

§18.1 Application of Heine-Borel

50
Cole Killian (February 10, 2020) Math 254 Notes

Theorem 18.1
Let A1 ⊇ A2 ⊇ A3 ⊇ ⋯ be a nested sequence of compact sets. Then

⋂ An ≠ ∅
n∈N

(This is by the nested interval property, but we are going to prove it using heine-borel)

Proof. ∀n ∈ N, let Un ∶= R ∖ An ⇒ ∀n ∈ NUn is open and U1 ⊆ U2 ⊆ U3 ⊆ ⋯

By de morgans law, we have that

⋃ Un = ⋃ R ∖ An = R ∖ ⋂ An
n∈N n∈N ® n∈N
De morgans

Now assume that ∩n∈N An = ∅. Then ∪n∈N Un = R ∖ ∅ = R.

i.e. The Un cover all of R and thus especially A1 . By heine-borel, this open
cover has a finite subcover.

{Un1 , . . . , Unk }, n1 < ⋅ ⋅ ⋅ < nk

k
⇒ A1 ⊆ ⋃ Uni = Un1 ∪ ⋯ ∪ Unk = Unk
i=1
⇒ A1 ⊆ Unk
⇒ Ank ⊆ A1 ⊆ Unk = R ∖ Ank
⇒ Ank ⊆ R ∖ Ank 
⇒ ⋂ An ≠ ∅
n∈N

Definition 18.2 (Uniform Continuity). Let’s recall the definition of continuity of f ∶

A → R:
(∀x0 ∈ A)(∀ > 0)(∃δ = δ(, x0 )) ∶ (∀x ∈ A)(∣x − x0 ∣ < δ ⇒ ∣f (x) − f (x0 )∣ < )

Note 18.3. In general, δ will depend on both  (unavoidable) and x0 .

It would be useful in many branches of analysis (e.g. Riemann integration) if δ would
only depend on  and not x0 .

i.e. we’d like to have this:

(∀x0 ∈ A)(∀ > 0)(∃δ = δ())(∀x ∈ A) ∶ (∣x − x0 ∣ < δ ⇒ ∣f (x) − f (x0 )∣ < )
≡
(∀ > 0)(∃ > 0)(∀x1 , x0 ∈ A) ∶ (∣x − x0 ∣ < δ ⇒ ∣f (x) − f (x0 )∣ < )
Since x0 is actually a variable, we’ll use µ instead and obtain:

f ∶ A ⊆ R → R is called uniformly continuous on A if

(∀ > 0)(∃δ > 0)(∀x, µ ∈ A) ∶ (∣x − µ∣ < δ ⇒ ∣f (x) − f (µ)∣ < )

51
Cole Killian (February 10, 2020) Math 254 Notes

Example 18.4
f ∶ R → R, x → x. Claim: f is uniformaly continuous.

Proof. Let  > 0 and let δ ∶= . Then ∀x, µ ∈ R, ∣x − µ∣ < δ =  ⇒ ∣f (x) − f (µ)∣ =
∣x − u∣ < 

Lemma 18.5
√ √ √
∀x, µ > 0 where x ≥ µ, we have that x − µ ≤ x − µ.

Proof.
√ √ √
x− µ≤ x−µ
√ √ √
⇔ ( x − µ)2 ≤ ( x − µ)2 = x − µ
√ √
⇔x−2 x µ+µ≤x−µ
√ √
⇔ 2µ − 2 x µ ≤ 0
√ √ √
⇔ 2 µ ( µ − x) ≤ 0 ✓
± ´¹¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹¸ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¶
≥0 ≤0

Because we only used equivalence statements, this final true statement proves that
√ √ √
x− µ≤ x−µ

Example 18.6
√
f ∶ R+0 = [0, ∞[→ R, x → x. Claim: f is uniformally continuous.
√
Remark 18.7. We did prove in chapter 4 that x is continuous on [0, ∞[. Back
then, the δ value we obtained did depend on both  and x !
√
However, this does not necessarily mean that is not uniformally continuous!
It might just mean that we need better estimates!

Proof. Let  > 0, let δ > 0 be arbitrary for now. Let x, µ ∈ [0, ∞[. We may assume
without loss of generality that x ≥ µ. Let ∣x − µ∣ = x − µ < δ. Then:
√ √ √ √ √ √
∣f (x) − f (µ)∣ = ∣ x − µ∣ = x − µ ≤ x − µ < δ < 
⇔ δ < 2

Note that δ is independent of x and µ !

With this uniform δ, we have

√
∣x − µ∣ < δ ⇒ ∣f (x) − f (µ)∣ <  ⇒ x

is uniform continuous on [0, ∞[.

How can we see whether a funciton is not uniformally continuous?

52
Cole Killian (February 10, 2020) Math 254 Notes

f ∶ A → R not continuous:

¬(∀ > 0)(∃δ > 0)(∀x, µ ∈ A) ∶ (∣x − µ∣ < δ ⇒ ∣f (x) − f (µ)∣ < )
≡ ¬(∀ > 0)(∃δ > 0)(∀x, µ ∈ A) ∶ (∣x − µ∣ ≥ δ ∨ ∣f (x) − f (µ)∣ < )
≡ (∃ > 0)(∀δ > 0)(∃x, µ ∈ A) ∶ (∣x − µ∣ < δ ∧ ∣f (x) − f (µ)∣ ≥ )

Recall 18.8. P ⇒ Q ≡ ¬P ∨ Q

Theorem 18.9 (2 sequence criterion for non-uniform continuity)

Let f ∶ A → R. Let 0 > 0 and let (xn ), (µn ) be sequences in A such that lim(xn −µn ) =
0 and ∣f (xn ) − f (µn )∣ ≥ 0 for all n ∈ N. Then f is not uniformally continuous on A.

Proof. Assume that f is uniform continuous. Then ∃δ > 0 such that ∀x, µ ∈ A ∶
∣x − µ∣ < δ ⇒ ∣f (x) − f (µ)∣ < 0 . (∗)

Now lim(xn −µn ) = 0. Then (∃N ∈ N)(∀n ≥ N ) ∶ ∣xn −µn ∣ < δ. Especially, ∣xn −µn ∣ < δ.

In (∗) ∶⇒ ∣f (xN ) − f (µN )∣ < 0 

Thus f is not uniformally continuous on A.

Example 18.10
f ∶ R → R, x → x2 .

Let xn ∶= n, un ∶= n + 1/n

Then ∣xn − µn ∣ = 1/n ⇒ lim(xn − µn ) = 0

But ∣f (xn ) − f (µn )∣ = ∣n2 − (n + 1/n)2 ∣ = ∣n2 − n2 − 2 − 1/n2 ∣ = 2 + 1/n2 > 2.

Let 0 ∶= 2. Then lim(xn − µn ) = 0, but ∀n ∈ N ∶ ∣f (xn ) − f (µn )∣ ≥ 0 .

⇒ x2 is not uniformally continuous on R.

53
Cole Killian (February 10, 2020) Math 254 Notes

Example 18.11
f ∶]0, ∞[→ R, x → 1/x

Let xn ∶= 1/n, µn ∶= 1/(n + 1).

Then, ∣xn − µn ∣ = ∣1/n − 1/(n + 1)∣ = ∣(x + 1 − x)/(n(n + 1))∣ = 1/(n(n + 1)) ≤ 1/n2 → 0.

By convergence criterion, lim(xn − µn ) = 0.

But, ∣f (xn ) − f (µn )∣ = ∣n − (n + 1)∣ = 1. Let 0 ∶= 1.

Then lim(xn − µn ) = 0. But ∣f (xn ) − f (µn )∣ ≥ 0 .

Therefore 1/x is not uniformally continuous on ]0, ∞[.

54
Cole Killian (February 10, 2020) Math 254 Notes

Theorem 18.12
Every continuous function on a compact domain is uniformally continuous.

Proof. Let f ∶ A → R, A be compact, and f continuous on A.

Let  > 0, then (∀x ∈ A)(∃δx > 0) ∶ (∣x − µ∣ < δx ⇒ ∣(f (x) − f (µ)∣ < /2)

Now consider the neighborhoods V(1/2)δx (x) for all x ∈ A.

Then ϕ ∶= {V(1/2)δx (x) ∶ x ∈ A} is an open cover of A. (Even just the centres

of these neighborhoods already cover A)

By Heine-Borel, ϕ has a finite subcover {V(1/2)δx1 , . . . , V(1/2)δxn } where x1 , . . . , xn ∈ A.

Let δ ∶= min{ 21 δx1 , . . . , 12 δxn } > 0.

We’ll prove that with this δ, we have that ∣x − µ∣ < δ ⇒ ∣f (x) − f (µ)∣ < .

Let x, µ ∈ A such that ∣x − µ∣ < δ. Since x ∈ A, ∃1 ≤ k ≤ n such that x ∈ V(1/2)δxk (xk )

⇒ ∣x − xk ∣ < 21 δxk < δxk

and
1
∣µ − xk ∣ = ∣(µ − x) + (x − xk )∣ ≤ ∣x − µ∣ + ∣x − xk ∣ < δ + δxk = δxk
2
⇒ x, µ ∈ Vδxk (xk )
⇒ ∣f (x) − f (µ)∣ = ∣(f (x) − f (xk )) + f (xk ) − f (µ))∣
≤ ∣f (x) − f (xk )∣ + ∣f (µ) − f (xk )∣ < 
´¹¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¸ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹¶ ´¹¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¸¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¶
≤/2 ≤/2

Because ∣x − xk ∣ < δxk and ∣µ − xk ∣ < δxk .

i.e. if ∣x − µ∣ < δ ⇒ ∣f (x) − f (µ)∣ <  ⇒ f is uniform continuous on A

Example 18.13
x2 is uniform continuous on all intervals [−a, a] where a > 0.

Example 18.14
1/x is uniform continuous on all intervals [a, 1] where 0 < a < 1.

§19 Lecture 12-02

55
Cole Killian (February 10, 2020) Math 254 Notes

Theorem 19.1
Let f ∶ A → R be uniformly continuous on A.

Let (xn ) be a cauchy sequence in A. Then (f (xn )) is also a cauchy sequence.

Proof. Let  > 0. Then ∃δ > 0 such that ∣x − µ∣ < δ ⇒ ∣f (x) − f (µ)∣ < 

(xn ) cauchy then ∃N ∈ N such that ∀n, m ≥ N ∶ ∣xn − xm ∣ < δ ⇒ ∣f (xn ) − f (xm )∣ < .

i.e. (∃N ∈ N)(∀n, m ≥ N ∶ ∣f (xn ) − f (xm )∣ <  ⇒ (f (xn ) is a cauchy sequence.

Remark 19.2. This result is, in general, false, if f is just continuous on A.

Example 19.3
f ∶ ]0, ∞[ → R, x → 1/x.

f is continuous but not uniformally continuous on ]0, ∞[.

Consider xn ∶= 1/n. Then (xn ) is a cauchy sequence but (f (xn )) = (n) which
diverges.

⇒ (f (xn )) is not a cauchy sequence

However: if f ∶ A → R is continuous, (xn ) is a convergent sequence in A such that

lim(xn ) ∈ A. Then:

lim(xn ) ∶= x ∈ A. Then f is continuous at x. Thus let lim(f (xn )) = f (x) be the

sequence of continuity. Especially, (f (xn )) is cauchy sequence in this case.

This can be turned into another criterion for non-uniform continuous functions.

Theorem 19.4 (One sequence criterion for a non-uniform continuous function)

Let f ∶ A → R. If (xn ) is cauchy sequence in A such that (f (xn )) is not cauchy, then
f is not uniformally continuous on A.

Example 19.5
f ∶ ]0, ∞[ → R, x → 1/x.
1
xn ∶=
n
cauchy but (f (xn )) = (n) is not cauchy.

⇒ f is not uniformly continuous on ]0, ∞[

56
Cole Killian (February 10, 2020) Math 254 Notes

Theorem 19.6
Let f ∶ A → R, A bounded, f a uniformly continuous on A, then f is bounded (i.e.
f (A) is bounded.

Proof. Assume that f is unbounded. Then ∀n ∈ N, ∃xn ∈ A ∶ ∣f (xn )∣ ≥ n.

Consider (xn ). Since A is bounded, (xn ) is bounded and thus has a convergent
subsequence (xnk ). Thus (xnk ) is cauchy ⇒ (f (xnk )) is cauchy and thus especially
bounded. But ∣f (xnk )∣ ≥ nk ≥ k for all k ∈ N.

This implies that f (xnk ) is unbounded. Contradiction!

Thus f is bounded.

Example 19.7
f ∶ ]0, 1[ → R, x → 1/x. Then f is unbounded on the bounded domain ]0, 1[ ⇒ f is
not continuous on ]0, 1[.

§20 Lecture 12-03

Lipschitz Continuous.

Example 20.1
√ √
Last class: x is not lipschitz on [0, ∞[, however x is lipschitz on [a, ∞[ for any
a > 0.

Proof. Let x, µ ∈ [a, ∞[. Then

√ √ √ √
√ √ ( x − µ)( x + µ)
∣ x − µ∣ = ∣ √ √ ∣
x+ u
1
≤ √ ∣x − µ∣
2 a
√
i.e. x is lipschitz continuous on [a, ∞[ with lipschitz constant k = 1
√
2 a

Example 20.2
Last class: x2 is lipschitz on ] − a, a[, a > 0.

However, x2 is not lipschitz on R.

Proof. x2 isn’t even uniformly continuous on R and thus cannot be lipschitz.

Definition 20.3 (Geometric interpretation of lipschitz continuous). Geometric interpre-

tation of lipschitz continuous:

57
Cole Killian (February 10, 2020) Math 254 Notes

f ∶ A → R is lipschitz if

∃k > 0 ∶ ∀x, µ ∈ A ∶ ∣f (x) − f (µ)∣ ≤ k ⋅ ∣x − µ∣

f (x) − f (µ)
if x ≠ µ ⇔ ∣ ∣ ≤k
x−µ
´¹¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¸ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹¶
Difference Quotient

i.e. f is lipschitz if and only if the average slope of f is bounded on A.

√
§20.1 Another method for proving that x is uniformly continuous on
[0, ∞[.

√
Idea: If x ≥ 1, x is lipschitz on [1, ∞[ and thus uniformly continuous. And: if
√
0 ≤ x ≤ 1 : x is uniformly continuous since it is continuous and [0, 1] is compact.
√
Q: if x is uniformly continuous on [0, 1] and [1, ∞[, does it follow that f is uniformly
continuous on [0, ∞[.

A: Yes; this requries proof!

58
Cole Killian (February 10, 2020) Math 254 Notes

Theorem 20.4
Let f be uniformly continuous on intervals I1 , I2 where I1 is closed on the right with
sup I1 = max I1 = b. And I2 is closed on the left with inf I2 = min I2 = b, then f is
uniformly continuous on I = I1 ∪ I2 .

Proof. Let  > 0, f uniformly continuous on I1 , thus ∃δ1 > 0 such that ∣x − µ∣ < δ1 ⇒
∣f (x) − f (µ)∣ < /2.

f is uniformly continuous on I2 . Thus ∃δ2 > 0 such that ∣x − µ∣ < δ2 ⇒ ∣f (x) − f (µ)∣ <
/2.

Let δ ∶= min{δ1 , δ2 }.

1. Case x, µ ∈ I1

∣x − µ∣ < δ ≤ δ1 ⇒ ∣f (x) − f (µ)∣ < /2 < 

2. Case x, µ ∈ I2

∣x − µ∣ < δ ≤ δ2 ⇒ ∣f (x) − f (µ)∣ < /2 < 

3. Case x ∈ I1 , µ ∈ I2

∣x − µ∣ < δ ⇒ ∣x − b∣δ ∧ ∣u − b∣ < δ

 
Thus ∣f (x) − f (b)∣ < and ∣f (µ) − f (b)∣ <
2 2
Now: ∣f (x) − f (µ)∣ = ∣[f (x) − f (b)] − [f (µ) − f (b)]∣
 
≤ ∣f (x) − f (b)∣ + f (µ) − f (b)∣ < + = 
2 2
i.e. ∣x − µ∣ < δ ⇒ ∣f (x) − f (µ)∣ < 
⇒ f is uniformly continous on I = I1 ∪ I2

√ √
Application: x is uniformly continuous on [0, 1] and [1, ∞[⇒ x is uniformly
continuous on [0, ∞[.

§20.2 Differentiation
Definition 20.5 (Differentiable Definition). Let f ∶ I → R, I be an interval, x0 ∈ I.

We say that f is differentiable at x0 , if

f (x) − f (x0 )
lim exists.
x→x0 x − x0
´¹¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¸¹¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¶
Difference Quotient

If the limit exists, we call its value the derivative of f at x0 , denoted by

df
f ′ (x0 ) =
(x0 )
dx
If f is differentiable at all x0 ∈ I, we say that f is differentiable on I.

59
Cole Killian (February 10, 2020) Math 254 Notes

Theorem 20.6 (Caratheodory Alternative Description of Differentiability)

Let f ∶ I → R, x0 ∈ I, then f is differentiable at x0 if and only if there exists a
function φ ∶ I → R continuous at x0 such that

∀x ∈ I f (x) = f (x0 ) + φ(x)(x − x0 )

If φ exists, it holds that φ(x0 ) = f ′ (x0 ).

Proof. ”⇒” Let f be differentiable at x0 . Let

⎧
⎪ f (x)−f (x0 )
⎪ , x ≠ x0
φ(x) ∶= ⎨ ′ x−x0
⎪
⎩f (x0 ),
⎪ x = x0

Then
f (x) − f (x0 )
lim φ(x) = lim = f ′ (x0 ) = φ(x0 )
x→x0 x→x0 x − x0
⇒ φ is continuous at x0

”⇐” Let φ ∶ I → R, continuous at x0 such that

f (x) = f (x0 ) + φ(x)(x − x0 )

f (x)−f (x0 )
Let x ≠ x0 . ⇒ φ(x) = x−x0

f (x)−f (x0 )
φ continuous at x0 ⇒ limx→x0 x−x0 exists and equals φ(x0 ) ⇒ f is differ-
entiable at x0 and f ′ (x0 ) = φ(x0 )

Applications: Differentiable implies continuous. i.e. if f ∶ I → R is differentiable at

x0 ∈ I, then f is continuous at x0 .

Proof. f differentiable at x0 ⇒ ∃φ ∶ I → R, continuous at x0 such that ∀x ∈ I,

f (x) = f (x0 ) + φ(x) ⋅ (x − x0 )
´¹¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¸¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¹ ¶
continuous at x0

Theorem 20.7 (Product Rule)

Let f, g ∶ I → R be differentiable at x0 . Then f ⋅ g is differentiable at x0 and
(f ⋅ g)′ (x0 ) = f ′ (x0 )g(x0 ) − f (x0 ) ⋅ g ′ (x0 ).

Proof. f, g differentiable at x0 ⇒ ∃φ, ψ ∶ I → R continuous at x0 such that

f (x) = f (x0 ) + φ(x)(x − x0 )

g(x) = g(x0 ) + ψ(x)(x − x0 )
⇒ (f ⋅ g)(x) = f (x) ⋅ g(x)
= f (x0 )g(x0 ) + f (x0 )ψ(x)(x − x0 ) + g(x0 )ψ(x)(x − x0 ) + φ(x)ψ(x)(x − x0 )2
⇒ (f ⋅ g)(x) = f (x0 )g(x0 ) + [f (x)g(x0 ) + f (x0 )ψ(x) + φ(x)ψ(x)(x − x0 )] ⋅ (x − x0 )

60
Cole Killian (February 10, 2020) Math 254 Notes

Theorem 20.8 (Chain Rule)

Let f ∶ I → R, f ∶ J → R, f (I) ⊆ J, x0 ∈ I, f differentiable at x0 , g differentiable at
y0 ∶= f (x0 ), then g ○ f is differentiable at x0 , and (g ○ f )′ (x0 ) = g ′ (f (x0 )) ⋅ f ′ (x)
f differentiable at x0 ⇒ ∃φ ∶ I → R, continuous at x0 such that f (x) = f (x0 ) +
φ(x)(x − x0 ).

g differentiable at y0 ⇒ ∃ψ ∶ J → R continuous at y0 such that g(y) = g(y0 ) +

ψ(y) ⋅ (y − y0 ). Therefore

g(f (x)) = g(f (x0 )) + ψ(f (x0 ) + φ(x)(x − x0 )) ⋅ [f (x0 ) + φ(x)(x − x0 ) − f (x0 )]
= g(f (x0 )) + ψ(f (x0 ) + φ(x)(x − x0 )) ⋅ φ(x) ⋅ (x − x0 ) ∶= Θ(x)

Then Θ is continuous at x0 as a composition of 2 continuous functions.

⇒ g ○ f is differentiable at x0

(g ○ f )′ (x0 ) = Θ(x0 )
= ψ(f (x0 ) + φ(x0 ) ⋅ 0) ⋅ φ(x0 )
= ψ(f (x0 )) ⋅ φ(x0 )
= ψ(y0 ) ⋅ φ(x0 )
= g ′ (y0 ) ⋅ f ′ (x0 )
= g ′ (f (x0 )) ⋅ f ′ (x0 )

§20.3 Relationship Between Lipschitz Continuity and Differentiability

Recall 20.9 (Mean Value Theorem). The mean value theorem. Let I = [a, b], f ∶ I → R
differentiable on ]a, b[ and continuous on the entire interval. Then there exists c ∈]a, b[
such that
f (b) − f (a)
f ′ (c) =
b−a

61
Cole Killian (February 10, 2020) Math 254 Notes

Theorem 20.10
Let f ∶ I → R be differentiable. Then f is lipschitz on I if and only if f ′ is bounded
on I.

Proof. ”⇒” Let f be lipschitz with lipchitz constant k.

f (x) − f (µ)
−k ≤ ≤k
x−µ

f (x) − f (µ)
⇒ −k ≤ lim ≤k
x→µ x−µ
⇒ −k ≤ f ′ (µ) ≤ k
⇒ ∣f ′ (µ)∣ ≤ k ∀ µ ∈ I
⇒ f ′ is bounded on I

”⇐” Assume that f ′ is bounded on I.

Let k > 0 such that ∣f ′ (x)∣ ≤ k for all x ∈ I.

Let x < µ, x, µ ∈ I. Apply mean value theorem to f on [x, µ] then ∃c ∈]x, µ[

such that
f (x) − f (µ) ∣f (x) − f (µ)∣
= f ′ (c) ⇒ = ∣f ′ (c)∣ ≤ k
x−µ ∣x − µ∣
⇒ ∣f (x) − f (µ)∣ ≤ k∣x − µ∣
⇒ f is lipschitz on I

§21 Sequences
Definition 21.1. Limit. xn → x if ∀ > 0, ∃k ∈ N such that ∣xn − x∣ < . ∀n ≥ K.

Example 21.2

2n
lim( )=2
n+1

Let  > 0. Compute (for any n ∈ N )

2n 2n − 2n − 2 2 2
∣ − 2∣ = ∣ ∣= <
n+1 n+1 n+1 n
By A.P, ∃k ∈ N such that K > 2 . Then ∀n ≥ K:

2n 2 2
∣ − 2∣ < ≤ < 
n+1 n k

62
Cole Killian (February 10, 2020) Math 254 Notes

Example 21.3

3n + 1 3
lim =
2n + 5 2
First, for any n ∈ N, we have that

3n + 1 3 6n + 2 − 6N − 15 13 106
∣ − ∣=∣ ∣= ≤
2n + 5 2 2(2n + 5) 4n + 10 n

Note: If unsure, use number much bigger i.e. 106 > 13.

106
Now, for any  > 0, by A.P, ∃k ∈ N such that k >  . Then, ∀n ≥ K:

3n + 1 3 106 106
∣ − ≤ ≤ <
2n + 5 2∣ n k

Example 21.4

n2 − 1 1
lim =
2n + 3 2
2

First, ∀n ∈ N,

n2 − 1 1 2n2 − 2 − 2n2 − 3 5 5
∣ − ∣ = ∣ ∣= 2 ≤ 2
2n + 3 2
2 2(2n + 3)
2 4n + 6 n
√
∀ > 0, ∃k ∈ N such that k > 5

Then, for any n ≥ k

n2 − 1 1 5 5
∣ − ∣≤ 2 ≤ 2 <
2n + 3 2 n
2 k

Example 21.5
√
n
lim =0
n+1
For any n ∈ N : √ √ √
n n n 1
∣ − 0∣ = ≤ =√
n+1 n+1 n n
So, ∀ > 0, let k ∈ N be such that k > 1
2
⇒ 2 > 1
k ⇒> √1 Then for any n ≥ k,
k
√
n 1 1
∣ − 0∣ ≤ √ ≤ √ < 
n+1 n k

Note:  > √1 ⇔ 2 > 1

k ⇔k> 1
2
k

63
Cole Killian (February 10, 2020) Math 254 Notes

Proposition 21.6
If xn → x, then ∣xn ∣ → ∣x∣.

Proof. Let  > 0 be arbitrary. We know that ∃k ∈ N such that ∣xn − x∣ <  ∀n ≥ K.

∣∣xn ∣ − ∣x∣∣ ≤ ∣xn − x∣ <  ∀n ≥ k

Side proof

Proof.

∣xn ∣ = ∣xn − x + x∣ ≤ ∣xn − x∣ + ∣x∣

⇒ ∣xn ∣ − ∣x∣ ≤ ∣xn − x∣
...

Proposition 21.7
If ∣xn ∣ → 0, then xn → 0.

Proof. Let  > 0. Then ∃k ∈ N such that

∣xn − 0∣ = ∣xn ∣ = ∣∣xn ∣ − 0∣ <  ∀n ≥ k

Exercise 21.8. Show that if a > 1, then 1

an → 0.

Proof. If a > 1, then a = 1 + r where r > 0.

an = (1 + r)n ≥ 1 + rn Bernoulli
1 1 1 1
⇒ ∣ n − 0∣ = n ≤ ≤
a a 1 + rn rn
For any  > 0, we can pick K ∈ N such that K > 1
r . Then ∀n ≥ k

1 1 1
∣ n
− 0∣ ≤ ≤ <
a rn rK

Exercise 21.9. Show that if a ∈ (−1, 1), then an → 0.

Proof. First, if a = 0, we are done.

If a > 0, pick b = a1 . an = 1
bn → 0.

If a < 0, then 0 < ∣a∣ < 1 ⇒ ∣a∣n → 0 ⇒ ∣an ∣ → 0 ⇒ an → 0

64
Cole Killian (February 10, 2020) Math 254 Notes

Note 21.10.
lim lim an,m ≠ lim lim an,m
m→∞ n→∞ n→∞ m→∞

Definition 21.11. Another definition of limit: We have xn → x if and only if for any
open set x ∈ U , ∀ > 0, ∃K ∈ N such that xn ∈ U for all n ≥ K.

(⇒) First, suppose xn → x. Let U ∋ x where U is open. We know that ∃ > 0

such that V (x) ⊆ U . This means that y ∈ R such that ∣x − y∣ <  ⇒ y ∈ U .

∃K ∈ N such that ∣xn − x∣ <  ∀n ≥ K. So, if n ≥ K, then ∣xn − x∣ <  ⇒ xn ∈ V (x) ⊆ U

(⇐) Fix  > 0. We know that V (x) is open. So, ∃K ∈ N such that xn ∈ V (x)∀n ≥ K ⇒
∣xn − x∣ <  ∀n ≥ K

Proposition 21.12
Let xn be a positive sequence. If lim...

65