1.

NOTIONS FROM SETS THEORY

SET: A set is a collection, or a family, or ensemble of objects. Those objects are called
ELEMENTS or MEMBERS of the set.

EMPTY SET: The empty set is the set that doesn`t have elements. It is denoted by ∅ .

EQUALITY OF SETS: 𝐴 = 𝐵 ⇔ ∀𝑥, (𝑥 ∈ 𝐴 ⇔ 𝑥 ∈ 𝐵)

Ways to describe a set:

Extension: Naming all the elements of the set.

Contraction: Naming the characteristics that all the elements of the set, and just them, have
in common.

1.1 CLASIFICATION

FINITE SET: A set is finite if it has only finitely many elements.

INFINITE SET: A set is infinite if it is not finite.

Observation: ℕ is an infinite set.

Proof: Assume otherwise that ℕ is a finite set.

ℕ 𝑖𝑠 𝑎 𝑓𝑖𝑛𝑖𝑡𝑒 𝑠𝑒𝑡.
⇒ ∃𝑚 𝑚 = 𝑡𝑕𝑒 𝑔𝑟𝑒𝑎𝑡𝑒𝑠 𝑒𝑙𝑒𝑚𝑒𝑛𝑡 𝑜𝑓 ℕ ⇒
ℕ 𝑖𝑠 𝑎𝑛 𝑜𝑟𝑑𝑒𝑟𝑒𝑑 𝑠𝑒𝑡
𝑚∈ℕ ⇒𝑚+1 ∈ℕ
⇒ ⇒ 𝑚 + 1 ≤ 𝑚 (𝑎𝑏𝑠𝑢𝑟𝑑)
𝑚 ≥ 𝑗, ∀𝑗 ∈ ℕ


NOTE: If S is an infinite set of numbers, there might not be the greatest element.

1.2 SUBSET

SUBSET: A is a subset of B iff for ∀𝑥, (𝑥 ∈ 𝐴 ⇒ 𝑥 ∈ 𝐵)

DENOTATION: 𝐴 ⊆ 𝐵 VAIN DIAGRAM:

PROPPER SUBSET: A is a proper subset of B iff 𝐴 ⊆ 𝐵 and 𝐴 ≠ 𝐵

Proprieties (HW): The inclusion is a partially ordered relation

1. Reflexivity: 𝐴 ⊆ 𝐴, ∀ 𝐴 𝑠𝑒𝑡
2. Transitivity: 𝐴 ⊆ 𝐵, 𝐵 ⊆ 𝐶 ⇒ 𝐴 ⊆ 𝐶 ⇒ 𝐴 ⊆ 𝐶
3. Antisymmetry: 𝐴 ⊆ 𝐵, 𝐵 ⊆ 𝐴 ⇔ 𝐴 = 𝐵

Proof:
1. Directly by definition of inclusion.
2. 𝑥∈𝐴 𝑥∈𝐵 𝑥∈𝐶
𝐴⊆𝐵 𝐵⊆𝐶
3. Directly by definition id equality of sets.


1.3 OPERATIONS OF SETS

UNION: Let A, B be any 2 sets. Define the union 𝐴 ∪ 𝐵 = 𝑥 𝑥 ∈ 𝐴 ∨ 𝑥 ∈ 𝐵

INTERSECTION: Let A, B be any 2 sets. Define the intersection 𝐴 ∩ 𝐵 = 𝑥 𝑥 ∈ 𝐴 ∧ 𝑥 ∈ 𝐵

ARBITRARIE UNION: Let 𝐴𝑖 𝑖 ∈ 𝐼 𝑏𝑒 𝑎 𝑓𝑎𝑚𝑖𝑙𝑦 𝑜𝑓 𝑠𝑒𝑡𝑠.

.𝑊𝑒 𝑑𝑒𝑓𝑖𝑛𝑒 𝑖∈𝐼 𝐴𝑖 = 𝑥 ∃𝑖 ∈ 𝐼, 𝑥 ∈ 𝐴𝑖

ARBITRARIE INTERSECTION: : Let 𝐴𝑖 𝑖 ∈ 𝐼 𝑏𝑒 𝑎 𝑓𝑎𝑚𝑖𝑙𝑦 𝑜𝑓 𝑠𝑒𝑡𝑠.

.𝑊𝑒 𝑑𝑒𝑓𝑖𝑛𝑒 𝑖∈𝐼 𝐴𝑖 = 𝑥 ∀𝑖 ∈ 𝐼, 𝑥 ∈ 𝐴𝑖

Observation (Práctico): 𝐴 ∪ 𝐴 = 𝐴 ∩ 𝐴 = 𝐴 ∪ ∅ = 𝐴
Observation (Práctico): 𝐴 ∩ ∅ = ∅

MINESS: A, B any two sets. Define 𝐵 − 𝐴 = 𝐵 ∖ 𝐴 = 𝑥 𝑥 ∈ 𝐵 ∧ 𝑥 ∉ 𝐴

Observation: 𝐵 ∖ 𝐵 = ∅

Proof: Assume otherwise that 𝐵 ∖ 𝐵 ≠ ∅ ⇒

⇒ ∃𝑥 𝑥 ∈ 𝐵 ∖ 𝐵 ⇒ ∃𝑥 𝑥 ∈ 𝐵 ∧ 𝑥 ∉ 𝐵 (𝑎𝑏𝑠𝑢𝑟𝑑) 

Observation: 𝐵 ∖ 𝐴 ∩ (𝐴 ∖ 𝐵) = ∅

Proof: Assume otherwise that 𝐵 ∖ 𝐴 ∩ 𝐴 ∖ 𝐵 ≠ ∅ ⇒

⇒ ∃ 𝑥 𝑥 ∈ 𝐵 ∖ 𝐴 ∩ (𝐴 ∖ 𝐵) ⇒ ∃ 𝑥 𝑥 ∈ 𝐵 ∖ 𝐴 ∧ 𝑥 ∈ (𝐴 ∖ 𝐵) ⇒
⇒ ∃𝑥 𝑥 ∈ 𝐵 ∧ 𝑥 ∉ 𝐴 ∧ 𝑥 ∈ 𝐴 ∧ 𝑥 ∉ 𝐵 ⇒
⇒ ∃ 𝑥 𝑥 ∈ 𝐴 ∧ 𝑥 ∉ 𝐴 (absurd) 

COMPLEMENT: Let X be the universe set. ∀𝐴 ⊆ 𝑋, 𝐴𝐶 = 𝑋 − 𝐴

Proprieties: (DE MORGAN’S LAWS)

1. 𝐵∖ 𝑖∈𝐼𝐴𝑖 = 𝑖∈𝐼 𝐵 ∖ 𝐴𝑖
2. 𝐵∖ 𝑖∈𝐼𝐴𝑖 = 𝑖∈𝐼 𝐵 ∖ 𝐴𝑖
3. 𝑖∈𝐼 𝐴𝑖 𝐶 = 𝑖∈𝐼 𝐴𝑖 𝐶
4. 𝑖∈𝐼 𝐴𝑖 𝐶 = 𝑖∈𝐼 𝐴𝑖 𝐶

Proof:

1. 𝑥 ∈ 𝐵 ∖ 𝑖∈𝐼 𝐴𝑖 ⇔ 𝑥 ∈ 𝐵 ∧ 𝑥 ∉ 𝑖∈𝐼 𝐴𝑖 ⇔
⇔ 𝑥 ∈ 𝐵 ∧ ∃ 𝑖 ∈ 𝐼 𝑥 ∉ 𝐴𝑖 ⇔ ∃ 𝑖 ∈ 𝐼 𝑥 ∈ 𝐵 ∧ 𝑥 ∉ 𝐴𝑖 ⇔
⇔ ∃ 𝑖 ∈ 𝐼 𝑥 ∈ (𝐵 ∖ 𝐴𝑖 ) ⇔ 𝑥 ∈ 𝑖∈𝐼 𝐵 ∖ 𝐴𝑖

2. 𝑥 ∈ 𝐵 ∖ 𝑖∈𝐼 𝐴𝑖 ⇔ 𝑥 ∈ 𝐵 ∧ 𝑥 ∉ 𝑖∈𝐼 𝐴𝑖 ⇔
⇔ 𝑥 ∈ 𝐵 ∧ ∀ 𝑖 ∈ 𝐼 𝑥 ∉ 𝐴𝑖 ⇔ ∀ 𝑖 ∈ 𝐼 𝑥 ∈ 𝐵 ∧ 𝑥 ∉ 𝐴𝑖 ⇔
⇔ ∀ 𝑖 ∈ 𝐼 𝑥 ∈ (𝐵 ∖ 𝐴𝑖 ) ⇔ 𝑥 ∈ 𝑖∈𝐼 𝐵 ∖ 𝐴𝑖

3. Trivial taking B as the universe set on 1

4. Trivial taking B as the universe set on 2



Observation (Práctico): ∀𝐴 𝑠𝑒𝑡, (𝐴𝐶 )𝐶 = 𝐴

CARTESIAN PRODUCT: Let A and B any two sets. 𝐴 × 𝐵 = 𝑎, 𝑏 ; 𝑎 ∈ 𝐴 ∧ 𝑏 ∈ 𝐵

Observation: ∅ × 𝐵 = 𝐵 × ∅ = ∅

Proof:

1) ∅ × 𝐵 = ∅: Assume otherwise that ∅ × 𝐵 ≠ ∅ ⇒

⇒ ∃ (𝑥, 𝑦) (𝑥, 𝑦) ∈ ∅ × 𝐵 ⇒ ∃ 𝑥 𝑥 ∈ ∅ (𝑎𝑏𝑠𝑢𝑟𝑑)

2) 𝐵 × ∅ = ∅: Assume otherwise that 𝐵 × ∅ ≠ ∅ ⇒

⇒ ∃ (𝑥, 𝑦) (𝑥, 𝑦) ∈ 𝐵 × ∅ ⇒ ∃ 𝑦 𝑦 ∈ ∅ (𝑎𝑏𝑠𝑢𝑟𝑑)

∴ 𝐹𝑟𝑜𝑚 1) 𝑎𝑛𝑑 2): ∅ × 𝐵 = ∅ = 𝐵 × ∅ 

2. NOTIONS FROM FUNCTIONS THEORY
FUNCTION or MAP: Let A and B be sets. A function or map 𝑓: 𝐴 → 𝐵 is a rule that assigns a
unique element of B to each element of A.

We call A the DOMAIN of the function f and B the TARGET or SQUACE of the function f.

2.1 CLASIFICATION

ONE TO ONE or INJECTIVE: Given a function 𝑓: 𝐴 → 𝐵, we say that f is injective or 1-1 iff
∀𝑥, 𝑦 ∈ 𝐴, (𝑥 ≠ 𝑦 ⇒ 𝑓 𝑥 ≠ 𝑓 𝑦 )

NOTE: Because the inverse of a implication, it is the same if we prove that ∀𝑥, 𝑦 ∈
𝐴, (𝑓 𝑥 = 𝑓(𝑦) ⇒ 𝑥 = 𝑦).

ONTO or SURJECTIVE: Given a function 𝑓: 𝐴 → 𝐵, we say that f is surjective or onto iff

∀𝑦 ∈ 𝐵, ∃𝑥 ∈ 𝐴; 𝑓 𝑥 = 𝑦

BIJECTIVE: Given a function 𝑓: 𝐴 → 𝐵, we say that f is bijective iff 𝑓 𝑖𝑠 1-1 and onto.

2.2 IMAGE AND PREIMAGE SETS

IMAGE SET: Given a function 𝑓: 𝑋 → 𝑌, and a set A ∕ 𝐴 ⊂ 𝑋. We define the image set of A under
f to be 𝑓 𝐴 = 𝑓(𝑥) ∈ 𝑌 ∕ 𝑥 ∈ 𝐴

In other words: 𝑓 𝐴 is the collection of mages or values of elements of A.

PREIMAGE SET: Given a function 𝑓: 𝑋 → 𝑌, and a set B ∕ 𝐵 ⊂ 𝑌. We define the preimage set of
B under f to be 𝑓 −1 𝐵 = 𝑥 ∈ 𝑋 ∕ 𝑓(𝑥) ∈ 𝐵

In other words: 𝑓 𝐵 is the collection of preimages of elements of B.

Proprieties: Let have a function 𝑓: 𝑋 → 𝑌 𝑎𝑛𝑑 𝑡𝑕𝑒 𝑠𝑒𝑡𝑠 𝐴, 𝐵 ⊂ 𝑋 𝑎𝑛𝑑 𝐶 ⊂ 𝑌.

1. 𝑓 𝐴∪𝐵 = 𝑓 𝐴 ∪𝑓 𝐵
2. 𝑓 𝐴∩𝐵 ⊂ 𝑓 𝐴 ∩𝑓 𝐵
3. 𝐴 ⊂ 𝑓 −1 (𝑓(𝐴))
4. 𝑓(𝑓 −1 (𝐶)) ⊂ 𝐶

Proof:

1) 𝑦 ∈ 𝑓 𝐴 ∪ 𝐵 ⇔ ∃𝑥, 𝑥 ∈ 𝐴 ∪ 𝐵 ∧ 𝑓 𝑥 = 𝑦 ⇔
⇔ ∃𝑥, 𝑥 ∈ 𝐴 ∨ 𝑥 ∈ 𝐵 ∧ 𝑓 𝑥 = 𝑦 ⇔
⇔ ∃𝑥, 𝑥 ∈ 𝐴 ∧ 𝑓 𝑥 = 𝑦 ∨ (𝑥 ∈ 𝐵 ∧ 𝑓 𝑥 = 𝑦 ) ⇔
⇔ 𝑦 ∈ 𝑓 𝐴 ∨ 𝑦 ∈ 𝑓 𝐵 ⇔ 𝑦 ∈ 𝑓 𝐴) ∪ 𝑓(𝐵

2) 𝑦 ∈ 𝑓 𝐴 ∩ 𝐵 ⇒ ∃𝑥, 𝑥 ∈ 𝐴 ∩ 𝐵 ∧ 𝑓 𝑥 = 𝑦 ⇒
⇒ ∃𝑥, 𝑥 ∈ 𝐴 ∧ 𝑥 ∈ 𝐵 ∧ 𝑓 𝑥 = 𝑦 ⇒
⇒ ∃𝑥, 𝑥 ∈ 𝐴 ∧ 𝑓 𝑥 = 𝑦 ∧ (𝑥 ∈ 𝐵 ∧ 𝑓 𝑥 = 𝑦 ) ⇒
⇒ 𝑦 ∈ 𝑓 𝐴 ∧ 𝑦 ∈ 𝑓 𝐵 ⇒ 𝑦 ∈ 𝑓 𝐴) ∧ 𝑓(𝐵

3) 𝑥 ∈ 𝐴 ⇒ 𝑓 𝑥 ∈ 𝑓 𝐴 ⇒ 𝑥 ∈ 𝑓 −1 (𝑓(𝐴))

4) 𝑦 ∈ 𝑓 𝑓 −1 𝐶 ⇒ ∃𝑥, 𝑥 ∈ 𝑓 −1 𝐶 ∧ 𝑓 𝑥 = 𝑦 ⇒
⇒ ∃𝑥, 𝑓(𝑥) ∈ 𝐶 ∧ 𝑓 𝑥 = 𝑦 ⇒ 𝑦 ∈ 𝐶


Propriety (HW): Let have a function 𝑓: 𝑋 → 𝑌 𝑎𝑛𝑑 𝑡𝑕𝑒 𝑠𝑒𝑡𝑠 𝐴, 𝐵 ⊂ 𝑋. 𝑓 𝑖𝑠 𝑖𝑛𝑗𝑒𝑐𝑡𝑖𝑣𝑒 ⇒

⇒𝑓 𝐴∩𝐵 = 𝑓 𝐴 ∩𝑓 𝐵

Proof: 𝑦∈𝑓 𝐴 ∩𝑓 𝐵 ⇒𝑦∈𝑓 𝐴 ∧𝑦∈𝑓 𝐵 ⇒

⇒ ∃𝑥 ∈ 𝐴; 𝑓 𝑥 = 𝑦 ∧ ∃𝑧 ∈ 𝐵; 𝑓 𝑧 = 𝑦
⇒ ∃𝑥, 𝑥 ∈ 𝐴 ∧ 𝑥 ∈ 𝐵 ∧ 𝑓 𝑥 = 𝑦 ⇒
𝑓 𝑖𝑠 1 − 1 𝑎𝑛𝑑 𝑓 𝑥 = 𝑓 𝑧 ⇒ 𝑥 = 𝑧
⇒ ∃𝑥, 𝑥 ∈ 𝐴 ∩ 𝐵 ∧ 𝑓 𝑥 = 𝑦 ⇒ 𝑦 ∈ 𝑓 𝐴 ∩ 𝐵

∴ 𝑓 𝐴 ∩𝑓 𝐵 ⊂ 𝑓 𝐴∩𝐵
⇒𝑓 𝐴 ∩𝑓 𝐵 =𝑓 𝐴∩𝐵
𝐵𝑒𝑐𝑎𝑢𝑠𝑒 𝑡𝑕𝑒𝑜𝑟𝑒𝑚 𝑏𝑒𝑓𝑜𝑟𝑒, 𝑝𝑎𝑟𝑡 2: 𝑓 𝐴 ∩ 𝐵 ⊂ 𝑓 𝐴 ∩ 𝑓 𝐵


NOTA MÍA: Que 𝑓 𝑥 ∈ 𝑓 𝐴 𝑛𝑜 𝑠𝑖𝑔𝑛𝑖𝑓𝑖𝑐𝑎 𝑞𝑢𝑒 𝑥 ∈ 𝐴. Solamente si es inyectiva

Observation (Práctico): Let have a function 𝑓: 𝑋 → 𝑌 𝑎𝑛𝑑 𝑡𝑕𝑒 𝑠𝑒𝑡 𝐴 ⊂ 𝑋.

𝑓 𝑖𝑠 𝑖𝑛𝑗𝑒𝑐𝑡𝑖𝑣𝑒 ⇔ 𝐴 = 𝑓 −1 (𝑓(𝐴))

Observation (Práctico): Let have a function 𝑓: 𝑋 → 𝑌 𝑎𝑛𝑑 𝑡𝑕𝑒 𝑠𝑒𝑡 𝐶 ⊂ 𝑌.

𝑓 𝑖𝑠 𝑠𝑢𝑣𝑗𝑒𝑐𝑡𝑖𝑣𝑒 ⇔ 𝑓 𝑓 −1 𝐶 = 𝐶
Proprieties: Let have a function 𝑓: 𝑋 → 𝑌 𝑎𝑛𝑑 𝑡𝑕𝑒 𝑠𝑒𝑡𝑠 𝐶, 𝐷 ⊂ 𝑌.

1. 𝑓 −1 𝐶 ∪ 𝐷 = 𝑓 −1 𝐶 ∪ 𝑓 −1 𝐷
2. 𝑓 −1 𝐶 ∩ 𝐷 = 𝑓 −1 𝐶 ∩ 𝑓 −1 𝐷

Proof:

1) 𝑥 ∈ 𝑓 −1 𝐶 ∪ 𝐷 ⇔ 𝑓 𝑥 ∈ 𝐶 ∪ 𝐷 ⇔ 𝑓 𝑥 ∈ 𝐶 ∨ 𝑓 𝑥 ∈ 𝐷 ⇔
⇔ 𝑥 ∈ 𝑓 −1 𝐶 ∨ 𝑥 ∈ 𝑓 −1 𝐷 ⇔ 𝑥 ∈ 𝑓 −1 𝐶 ∪ 𝑓 −1 𝐷

2) 𝑥 ∈ 𝑓 −1 𝐶 ∩ 𝐷 ⇔ 𝑓 𝑥 ∈ 𝐶 ∩ 𝐷 ⇔ 𝑓 𝑥 ∈ 𝐶 ∧ 𝑓 𝑥 ∈ 𝐷 ⇔
⇔ 𝑥 ∈ 𝑓 −1 𝐶 ∧ 𝑥 ∈ 𝑓 −1 𝐷 ⇔ 𝑥 ∈ 𝑓 −1 𝐶 ∩ 𝑓 −1 𝐷


2.3 COMPOSITION OF FUNCTIONS

COMPOSITION OF FUNCTIONS: Given the function𝑠 𝑓: 𝑋 → 𝑌 𝑎𝑛𝑑 𝑔: 𝑌 → 𝑍, we define the

composition of g and f as the function 𝑔 ∘ 𝑓: 𝑋 → 𝑍 ∕ 𝑔 ∘ 𝑓 𝑥 = 𝑔 𝑓(𝑥)

Propriety: (ASSOCIATIVITY)
Let have the function𝑠 𝑓: 𝑋 → 𝑌 𝑎𝑛𝑑 𝑔: 𝑌 → 𝑍, 𝑕 ∘ 𝑔 ∘ 𝑓 = 𝑕 ∘ (𝑔 ∘ 𝑓)

Proof:

Notice that both functions have domain X and same target set Z, so it’s enough to show that
𝑕 ∘ 𝑔 ∘ 𝑓 𝑥 = 𝑕 ∘ 𝑔 ∘ 𝑓 𝑥 , ∀𝑥 ∈ 𝑋

𝑕∘𝑔 ∘𝑓 𝑥 = 𝑕∘𝑔 𝑓 𝑥 =𝑕 𝑔 𝑓 𝑥 = 𝑕 𝑔∘𝑓 𝑥 =

= 𝑕∘ 𝑔∘𝑓 𝑥 
3. THE REAL NUMBERS
𝐑𝐄𝐀𝐋 𝐍𝐔𝐌𝐁𝐄𝐑 𝐀𝐗𝐈𝐎𝐌: We define the real number system to be a set
ℝ 𝑎𝑛𝑑 𝑡𝑤𝑜 𝑜𝑝𝑒𝑟𝑎𝑡𝑖𝑜𝑛𝑠 𝑑𝑒𝑓𝑖𝑛𝑒𝑑 𝑖𝑛 𝑖𝑡 + 𝑎𝑛𝑑 ∙ 𝑐𝑎𝑙𝑙𝑒𝑑 𝑎𝑑𝑑𝑖𝑡𝑖𝑜𝑛 𝑎𝑛𝑑 𝑚𝑢𝑙𝑡𝑖𝑝𝑙𝑖𝑐𝑎𝑡𝑖𝑜𝑛, 𝑠𝑢𝑐𝑕 𝑎𝑠
𝑏𝑜𝑡𝑕 𝑕𝑎𝑣𝑒 𝑡𝑕𝑒 𝑓𝑜𝑙𝑙𝑜𝑤𝑖𝑛𝑔 𝑝𝑟𝑜𝑝𝑒𝑟𝑡𝑖𝑒𝑠:

COMMUTATIVITY: 𝑎 + 𝑏 = 𝑏 + 𝑎, ∀𝑎, 𝑏 ∈ ℝ
𝑎 ⋅ 𝑏 = 𝑏 ⋅ 𝑎, ∀𝑎, 𝑏 ∈ ℝ
ASOCIATIVITY: 𝑎 + 𝑏 + 𝑐 = 𝑎 + 𝑏 + 𝑐 , ∀𝑎, 𝑏, 𝑐 ∈ ℝ
𝑎 ⋅ 𝑏 ⋅ 𝑐 = 𝑎 ⋅ 𝑏 ⋅ 𝑐 , ∀𝑎, 𝑏, 𝑐 ∈ ℝ
DISTRIBUTIVITY: 𝑎 ⋅ 𝑏 + 𝑐 = 𝑎 ⋅ 𝑏 + 𝑎 ⋅ 𝑐, ∀𝑎, 𝑏, 𝑐 ∈ ℝ
EXISTENCE OF NEUTRAL ∃0 ∈ ℝ, ∀𝑎 ∈ ℝ; 𝑎 + 0 = 𝑎
ELEMENTS: ∃1 ∈ ℝ, 1 ≠ 0, ∀𝑎 ∈ ℝ; 𝑎 ⋅ 1 = 𝑎
EXISTENCE OF INVERSES: ∀𝑎 ∈ ℝ, ∃ −𝑎 ∈ ℝ; 𝑎 + −𝑎 = 0
∀𝑎 ∈ ℝ, 𝑎 ≠ 0, ∃ 𝑎−1 ∈ ℝ; 𝑎 ∙ 𝑎−1 = 1

Proprieties:
1. 𝑎 ∙ 0 = 0, ∀𝑎 ∈ ℝ
2. 𝑎 ∙ 𝑏 = 0 ⇔ 𝑎 = 0 ∨ 𝑏 = 0
3. – 𝑎 = (−1) ⋅ 𝑎, ∀𝑎 ∈ ℝ
4. – −𝑎 = 𝑎, ∀𝑎 ∈ ℝ
5. (𝑎−1 )−1 = 𝑎, ∀𝑎 ∈ ℝ

NOTATION (Def. of the symbol “-“): ∀𝑎, 𝑏 ∈ ℝ, we write: 𝑏 + −𝑎 𝑎𝑠 𝑏 − 𝑎

3.1 ORDER

AXIOM OF ORDER IN ℝ: There is a subset ℝ+ = ℝ+ ⊂ ℝ such that:

1. ∀𝑎, 𝑏 ∈ ℝ+, 𝑎 + 𝑏 ∈ ℝ+ 𝑎𝑛𝑑 𝑎 ∙ 𝑏 ∈ ℝ+
2. ∀𝑎 ∈ ℝ, 𝑜𝑛𝑒 𝑎𝑛𝑑 𝑜𝑛𝑙𝑦 𝑜𝑛𝑒 𝑜𝑓 𝑡𝑕𝑒 𝑓𝑜𝑙𝑙𝑜𝑤𝑖𝑛𝑔 𝑠𝑡𝑎𝑡𝑒𝑚𝑒𝑛𝑠 𝑎𝑟𝑒 𝑡𝑟𝑢𝑒:
i. 𝑎 ∈ ℝ+
ii. 𝑎 = 0
iii. −𝑎 ∈ ℝ+

POSITIVE AND NEGATIVE NUMBERS: We call ℝ+ 𝑡𝑕𝑒 𝑠𝑒𝑡 𝑜𝑓 𝑡𝑕𝑒 Positive Numbers; and we
call Negative Numbers to the set ℝ− = ℝ− = 𝑥 ∈ ℝ/−𝑥 ∈ ℝ+

Observation: ∀𝑎, 𝑏 ∈ ℝ−, 𝑎 + 𝑏 ∈ ℝ−

Properties: (MULTIPLICATION RULES)

a b a.b
1. ∈ ℝ+ ∈ ℝ+ ∈ ℝ+
2. ∈ ℝ+ ∈ ℝ− ∈ ℝ−
3. ∈ ℝ− ∈ ℝ− ∈ ℝ+
GREATER OR LESS: Given 𝑎, 𝑏 ∈ ℝ , we say that b is greater than a or a is less than b ⇔
𝑏 − 𝑎 ∈ ℝ+

Observation: 𝑎 ∈ ℝ+ ⇔ 𝑎 > 0

Property: (TRICOTOMY)
𝐺𝑖𝑣𝑒𝑛 𝑎, 𝑏 ∈ ℝ, 𝑜𝑛𝑒 𝑎𝑛𝑑 𝑜𝑛𝑙𝑦 𝑜𝑛𝑒 𝑜𝑓 𝑡𝑕𝑒 𝑓𝑜𝑙𝑙𝑜𝑤𝑖𝑛𝑔 𝑠𝑡𝑎𝑡𝑒𝑚𝑒𝑛𝑠 𝑖𝑠 𝑡𝑟𝑢𝑒:
i. 𝑎<𝑏
ii. 𝑎=𝑏
iii. 𝑎>𝑏

Properties: 𝑎, 𝑏, 𝑐, 𝑑 ∈ ℝ
1. Transitivity: 𝑎 > 𝑏 𝑎𝑛𝑑 𝑏 > 𝑐 ⇒ 𝑎 > 𝑐
2. 𝑎 > 𝑏 𝑎𝑛𝑑 𝑐 ≥ 𝑑 ⇒ 𝑎 + 𝑐 > 𝑏 + 𝑑
3. 𝑎, 𝑏, 𝑐, 𝑑 ∈ ℝ+ . 𝑎 > 𝑏 𝑎𝑛𝑑 𝑐 ≥ 𝑑 ⇒ 𝑎 ∙ 𝑏 > 𝑏 ∙ 𝑑

Property: ∀𝑎 ∈ ℝ, 𝑎2 ≥ 0 𝑎𝑛𝑑 𝑎2 = 0 ⇔ 𝑎 = 0

Properties: 𝑎, 𝑏 ∈ ℝ+
1
1. ∈ ℝ+
𝑎
1 1
2. 𝑎<𝑏 ⇒ <
𝑏 𝑎

3.2 EXPONENTIATION and ABSOLUTE VALUE

EXPONENTIATION: Given 𝑎 ∈ ℝ, 𝑎𝑛𝑑 𝑛 ∈ ℤ, we define 𝑎𝑛 = 𝑎. 𝑎. 𝑎. 𝑎 … . 𝑎. If 𝑎 ≠ 0 we define

𝑛 𝑡𝑖𝑚𝑒𝑠
1
𝑎0 = 1 𝑎𝑛𝑑 𝑎−𝑛 =
𝑎𝑛

Properties: 𝑎, 𝑏 ∈ ℝ 𝑎𝑛𝑑 𝑛, 𝑚 ∈ ℤ
1. 𝑎𝑛 . 𝑎𝑚 = 𝑎𝑛+𝑚
2. 𝑎𝑛 𝑚 = 𝑎𝑛.𝑚
3. 𝑎. 𝑏 𝑛 = 𝑎𝑛 . 𝑏𝑛

ABSOLUTE VALUE: Given 𝑎 ∈ ℝ we define absolute value of a as:

i. 𝑎 = 𝑎, ∀𝑎 ∈ ℝ+ ∪ 0
ii. 𝑎 = −𝑎, ∀𝑎 ∈ ℝ−

Properties: 𝑎, 𝑏 ∈ ℝ 𝑎𝑛𝑑 𝑛, 𝑚 ∈ ℤ
1. ∀𝑎 ∈ ℝ, 𝑎 = −𝑎
2. ∀𝑎 ∈ ℝ, 𝑎 2 = 𝑎2
3. ∀𝑎 ∈ ℝ, 𝑎 ≥ 0 𝑎𝑛𝑑 𝑎 = 0 ⇔ 𝑎 = 0
4. ∀𝑎. 𝑏 ∈ ℝ, 𝑎. 𝑏 = 𝑎 . 𝑏
5. 𝑇𝑟𝑖𝑎𝑛𝑔𝑢𝑙𝑎𝑟 𝑑𝑒𝑠𝑖𝑔𝑢𝑎𝑙𝑖𝑡𝑦: ∀𝑎. 𝑏 ∈ ℝ, 𝑎 ± 𝑏 ≤ 𝑎 + 𝑏
6. ∀𝑎. 𝑏 ∈ ℝ, 𝑎 ± 𝑏 ≥ 𝑎 − 𝑏
3.3 RATIONAL NUMBERS
𝑎
RATIONAL NUMBERS: We call rational numbers to the set ℚ = 𝑠𝑢𝑐𝑕 𝑎𝑠 𝑎 ∈ ℤ, 𝑏 ∈ ℤ ∧
𝑏
𝑏≠0

Observation (HW): (DENSITY OF ℚ IN ℝ) ∀𝑎, 𝑏 ∈ ℝ, 𝑎 < 𝑏 , ∃𝑞 ∈ ℚ / 𝑎 < 𝑞 < 𝑏

Observation (HW): (DENSITY OF ℝ/ℚ IN ℝ) ∀𝑎, 𝑏 ∈ ℝ, 𝑎 < 𝑏 , ∃𝑞 ∈ ℝ − ℚ / 𝑎 < 𝑞 < 𝑏

Theorem: 𝑎 ∈ ℚ, 𝑏 ∈ ℝ − ℚ

1. 𝑎 ± 𝑏 ∈ ℝ − ℚ
2. 𝑖𝑓 𝑎 ≠ 0, 𝑎. 𝑏 ∈ ℝ − ℚ
𝑎
3. 𝑖𝑓 𝑎 ≠ 0, ∈ ℝ − ℚ
𝑏
𝑏
4. 𝑖𝑓 a ≠ 0, ∈ ℝ − ℚ
𝑎
Proof:

1. 𝐴𝑠𝑠𝑢𝑚𝑒 𝑜𝑡𝑕𝑒𝑟𝑤𝑖𝑠𝑒 𝑡𝑕𝑎𝑡 𝑎 ± 𝑏 ∉ ℝ − ℚ ⇒ 𝑎±𝑏 ∈ℚ

⇒ 𝑎 ± 𝑏 + −𝑎 ∈ ℚ ⇒
𝑎 ∈ ℚ ⇒ −𝑎 ∈ ℚ
𝐴𝑠𝑠𝑜𝑐 .
±𝑏 ∈ ℚ ⇒ 𝑏 ∈ ℚ (𝑎𝑏𝑠𝑢𝑟𝑑)
𝐶𝑜𝑚𝑚 .

2. 𝐴𝑠𝑠𝑢𝑚𝑒 𝑜𝑡𝑕𝑒𝑟𝑤𝑖𝑠𝑒 𝑡𝑕𝑎𝑡 𝑎. 𝑏 ∉ ℝ − ℚ ⇒ 𝑎. 𝑏 ∈ ℚ 1

1 ⇒ 𝑎. 𝑏 . ∈ℚ ⇒
𝑎 ∈ ℚ , 𝑎 ≠ 0 𝑎𝑛𝑑 1 ∈ ℚ ⇒ ∈ ℚ 𝑎
𝑎
𝐴𝑠𝑠𝑜𝑐 .
𝑏 ∈ ℚ (𝑎𝑏𝑠𝑢𝑟𝑑)
𝐶𝑜𝑚𝑚 .
𝑎 𝑎
3. 𝐴𝑠𝑠𝑢𝑚𝑒 𝑜𝑡𝑕𝑒𝑟𝑤𝑖𝑠𝑒 𝑡𝑕𝑎𝑡 ∉ ℝ−ℚ ⇒ ∈ℚ 𝑎 1
𝑏 𝑏
1 ⇒ . ∈ℚ ⇒
𝑏 𝑎
𝑎 ∈ ℚ, 𝑎 ≠ 0 𝑎𝑛𝑑 1 ∈ ℚ ⇒ ∈ ℚ
𝑎
⇒ 𝑏 ∈ ℚ (𝑎𝑏𝑠𝑢𝑟𝑑)
𝑏 𝑏
4. 𝐴𝑠𝑠𝑢𝑚𝑒 𝑜𝑡𝑕𝑒𝑟𝑤𝑖𝑠𝑒 𝑡𝑕𝑎𝑡 ∉ ℝ − ℚ ⇒ ∈ℚ ⇒ 𝑏 . 𝑎 ∈ℚ ⇒
𝑎 𝑎 𝑎
𝑎∈ℚ
⇒ 𝑏 ∈ ℚ (𝑎𝑏𝑠𝑢𝑟𝑑)

3.4 COMPLETENESS OF ℝ

BOUNDING

BOUNDED FROM ABOVE: Let 𝐴 ⊂ ℝ 𝑏𝑒 𝑠𝑢𝑐𝑕 𝑎𝑠 𝐴 ≠ ∅. We say A is bounded from above iff
∃𝑘 ∈ ℝ / ∀𝑎 ∈ 𝐴, 𝑎 ≤ 𝑘.
In that case we say that b is an UPPER BOUND of A.

NOTATION: A is bdd from above and b is an ub of A

Observation: 𝐴 ⊂ ℝ , 𝐴 ≠ ∅. ∃𝑘/𝑘 𝑖𝑠 𝑎𝑛 𝑢𝑏 𝑜𝑓 𝐴 ⇒ ∀𝑥; 𝑥 ≥ 𝑏, 𝑥 𝑖𝑠 an ub of A

BOUNDED FROM BELOW: Let 𝐴 ⊂ ℝ 𝑏𝑒 𝑠𝑢𝑐𝑕 𝑎𝑠 𝐴 ≠ ∅. We say A is bounded from below iff
∃𝑘 ∈ ℝ / ∀𝑎 ∈ 𝐴, 𝑎 ≥ 𝑘.
In that case we say that b is an LOWER BOUND of A.

NOTATION: A is bdd from below and b is a lb of A

Observation: 𝐴 ⊂ ℝ , 𝐴 ≠ ∅. ∃𝑘/𝑘 𝑖𝑠 𝑎𝑛 𝑙𝑏 𝑜𝑓 𝐴 ⇒ ∀𝑥; 𝑥 ≤ 𝑏, 𝑥 𝑖𝑠 an lb of A

BOUNDED: Let 𝐴 ⊂ ℝ 𝑏𝑒 𝑠𝑢𝑐𝑕 𝑎𝑠 𝐴 ≠ ∅. We say A is bdd iff A is bdd from above and from
below.

Observation (mia): 𝐴 ⊂ 𝐵
1. A is not bdd from above ⇒ B is not bdd from above
2. A is not bdd from below ⇒ B is not bdd from below

Proof:
1. Assume otherwise that B is bdd from above ⇒ ∃𝑘 ∈ ℝ / ∀𝑏 ∈ 𝐵, 𝑏 ≤ 𝑘 ⇒
⇒ ∃𝑘 ∈ ℝ / ∀𝑎 ∈ 𝐴, 𝑎 ≤ 𝑘 ⇒ A is bdd from above. (absurd)
2. Assume otherwise that B is bdd from below ⇒ ∃𝑘 ∈ ℝ / ∀𝑏 ∈ 𝐵, 𝑏 ≥ 𝑘 ⇒
⇒ ∃𝑘 ∈ ℝ / ∀𝑎 ∈ 𝐴, 𝑎 ≥ 𝑘 ⇒ A is bdd from below. (absurd)

Observation (mia): ℕ is not bdd from above

Proof: Assume otherwise that ℕ is bdd from above ⇒ ∃𝑘 ∈ ℝ / ∀𝑛 ∈ ℕ, 𝑛 ≤ 𝑘

𝑙∈ℕ
Let’s take 𝑙 = 𝑘 + 1 ⇒ (𝑎𝑏𝑠𝑢𝑟𝑑) 
𝑙>𝑘

Observation (mia): ℤ is not bounded from above or below

Proof: ℤ 𝑖𝑠 𝑛𝑜𝑡 𝑏𝑑𝑑 𝑓𝑟𝑜𝑚 𝑎𝑏𝑜𝑣𝑒 𝑏𝑒𝑐𝑎𝑠𝑢𝑒 ℕ ⊂ ℤ 𝑎𝑛𝑑 ℕ 𝑖𝑠 𝑛𝑜𝑡 𝑏𝑑𝑑 𝑓𝑟𝑜𝑚 𝑎𝑏𝑜𝑣𝑒.

Let’s consider ℤ− = −𝑛 /𝑛 ∈ ℕ . Because definition of ℤ, ℤ− ⊂ ℤ.

Assume otherwise that ℤ− is bdd from below ⇒ ∃𝑘 ∈ ℝ/∀𝑛 ∈ ℕ, −𝑛 ≥ 𝑘 ⇒
⇒ ∃𝑘 ∈ ℝ/∀𝑛 ∈ ℕ, 𝑛 ≤ 𝑘 ⇒ ℕ 𝑖𝑠 𝑏𝑑𝑑 𝑓𝑟𝑜𝑚 𝑎𝑏𝑜𝑣𝑒. (𝑎𝑏𝑠𝑢𝑟𝑑) 
Observation: ℚ is not bdd from above or below

Proof: ℤ ⊂ ℚ and ℤ is not bdd from above or below ⇒ ℚ is not bdd from above or below.


SUPREMUS AND INFIMUS

LEAST UPPER BOUND or SUPREMUS: Let 𝐴 ⊂ ℝ / 𝐴 ≠ ∅. A is bounded from above. We say

𝑎 ∈ ℝ is the least upper bound of A iff:
1. a is an ub of A
2. ∀𝑐 𝑢𝑏 𝑜𝑓 𝐴, 𝑐 ≥ 𝑎
NOTATION: a= 𝑙𝑢𝑏 𝐴 = sup⁡ (𝐴)

Theorem: (N&SC of SUPREMUS)

1. ∀𝑥 ∈ 𝐴, 𝑥 ≤ 𝑎
𝐴 ⊂ ℝ / 𝐴 ≠ ∅. 𝑎 = sup 𝐴 ⇔
2. ∀𝜀 > 0, ∃𝑥 ∈ 𝐴 , 𝑥 > 𝑎 − 𝜀
Proof:

⇒ :
𝑎 = sup 𝐴 𝑎 𝑖𝑠 𝑎𝑛 𝑢𝑏 𝑜𝑓 𝐴 ∀𝑥 ∈ 𝐴, 𝑥 ≤ 𝑎 (1)
𝑑𝑒𝑓 . 𝑑𝑒𝑓 .
𝐴𝑠𝑠𝑢𝑚𝑒 𝑜𝑡𝑕𝑒𝑟𝑤𝑖𝑠𝑒 𝑡𝑕𝑎𝑡 ∃𝜀 > 0, ∀𝑥 ∈ 𝑎, 𝑥 ≤ 𝑎 − 𝜀 ⇒ ∃𝜀 > 0, 𝑎 − 𝜀 𝑖𝑠 𝑎𝑛 𝑢𝑏 𝑜𝑓 𝐴
𝑎 = sup 𝐴 ∀𝑐 𝑢𝑏 𝑜𝑓 𝐴, 𝑐 ≥ 𝑎 ⇒
𝑑𝑒𝑓 .
⇒ ∃𝜀 > 0, 𝑎 − 𝜀 ≥ 𝑎 (𝑎𝑏𝑠𝑢𝑟𝑑) ∴ ∀𝜀 > 0, ∃𝑥 ∈ 𝐴 , 𝑥 > 𝑎 − 𝜀 (2)

⇐ :
Because 1) ∀𝑥 ∈ 𝐴, 𝑥 ≤ 𝑎 𝑎 𝑖𝑠 𝑎𝑛 𝑢𝑏 𝑜𝑓 𝐴
𝑑𝑒𝑓 .
𝐴𝑠𝑠𝑢𝑚𝑒 𝑜𝑡𝑕𝑒𝑟𝑤𝑖𝑠𝑒 𝑡𝑕𝑎𝑡 ∃𝑐 𝑢𝑏 𝑜𝑓 𝐴 ; 𝑐 < 𝑎
𝐵𝑒𝑐𝑎𝑢𝑠𝑒 2) ∀𝜀 > 0, ∃𝑥 ∈ 𝐴 , 𝑥 > 𝑎 − 𝜀
⇒ ∃𝑥 ∈ 𝐴 , 𝑥 > 𝑎 − 𝑎 − 𝑐 = 𝑐 (𝑎𝑏𝑠𝑢𝑟𝑑)
𝑐 < 𝑎 ⇒ 𝑎 − 𝑐 > 0. 𝐿𝑒𝑡𝑠 𝑡𝑎𝑘𝑒 𝜀 = 𝑎 − 𝑐


MAXIMUM: Let 𝐴 ⊂ ℝ / 𝐴 ≠ ∅. We say 𝑐 is the maximum of A iff c ∈ A and ∀𝑎 ∈ 𝐴, 𝑐 ≥ 𝑎

NOTATION: c= 𝑚𝑎𝑥 𝐴

Observation (HW): sup 𝐴 ∈ 𝐴 ⇒ sup 𝐴 = max⁡

(𝐴)

Proof: 𝐵𝑒𝑐𝑎𝑠𝑢𝑒 𝑑𝑒𝑓𝑖𝑛𝑖𝑡𝑖𝑜𝑛: ∀𝑥 ∈ 𝐴, 𝑥 ≤ sup 𝐴 𝑎𝑛𝑑 𝑏𝑒𝑐𝑎𝑢𝑠𝑒 𝑕𝑦𝑝𝑜𝑡𝑕𝑒𝑠𝑖𝑠 sup 𝐴 ∈ 𝐴 ⇒

⇒ sup 𝐴 = max⁡ (𝐴) 

𝐴 𝑖𝑠 𝑏𝑑𝑑 𝑓𝑟𝑜𝑚 𝑎𝑏𝑜𝑣𝑒

Observation (HW): ∃𝑥/𝑥 = 𝑚𝑎𝑥(𝐴) ⇒
sup 𝐴 = 𝑥

Proof: ∃𝑥/𝑥 = 𝑚𝑎𝑥 𝐴 ⇒ ∃𝑥, ∀𝑎 ∈ 𝐴, 𝑎 ≤ 𝑥 ⇒ 𝐴 𝑖𝑠 𝑏𝑑𝑑 𝑓𝑟𝑜𝑚 𝑎𝑏𝑜𝑣𝑒

∃𝑥, ∀𝑎 ∈ 𝐴, 𝑎 ≤ 𝑥
𝑥 = 𝑚𝑎𝑥 𝐴 ⇒ sup 𝐴 = 𝑥
𝑥 ∈ 𝐴 ⇒ ∀𝜀 > 0, ∃𝑥 ∈ 𝐴, 𝑥 > 𝑥 − 𝜀 𝑑𝑒𝑓 .

GREATEST LOWER BOUND or INFIMUS: Let 𝐴 ⊂ ℝ / 𝐴 ≠ ∅. A is bounded from below. We say
𝑎 ∈ ℝ is the greatest lower bound of A iff:
1. a is an lower bound of A
2. ∀𝑐 𝑙𝑜𝑤𝑒𝑟 𝑏𝑜𝑢𝑛𝑑 𝑜𝑓 𝐴, 𝑐 ≤ 𝑎
NOTATION: a= 𝑙𝑢𝑏 𝐴 = sup⁡ (𝐴)

Theorem: (N&SC of INFIMUS)

1. ∀𝑥 ∈ 𝐴, 𝑥 ≥ 𝑎
𝐴 ⊂ ℝ / 𝐴 ≠ ∅. 𝑎 = inf 𝐴 ⇔
2. ∀𝜀 > 0, ∃𝑥 ∈ 𝐴 /𝑥 < 𝑎 + 𝜀

Proof: Idem to the supremus. 

MINIMUM: Let 𝐴 ⊂ ℝ / 𝐴 ≠ ∅. We say 𝑐 is the minimum of A iff c ∈ A and ∀𝑎 ∈ 𝐴, 𝑐 ≤ 𝑎

NOTATION: c= 𝑚𝑖𝑛 𝐴

Observation: inf 𝐴 ∈ 𝐴 ⇒ inf 𝐴 = min⁡

(𝐴)

Proof: Idem to the supremus. 

𝐴 𝑖𝑠 𝑏𝑑𝑑 𝑓𝑟𝑜𝑚 𝑏𝑒𝑙𝑜𝑤

Observation: ∃𝑥/𝑥 = 𝑚𝑖𝑛(𝐴) ⇒
inf 𝐴 = 𝑥

Proof: Idem to the supremus. 

Theorem: 𝐴, 𝐵 𝑎𝑟𝑒 𝑏𝑜𝑢𝑛𝑑𝑒𝑑 𝑠𝑒𝑡𝑠.

1. 𝑖𝑓 𝛼 < 0, sup 𝛼𝐴 = 𝛼. inf⁡

(𝐴) 4. 𝑖𝑓 𝛼 < 0, inf 𝛼𝐴 = 𝛼. sup⁡(𝐴)
2. 𝑖𝑓 𝛼 > 0, sup 𝛼𝐴 = 𝛼. sup⁡(𝐴) 5. 𝑖𝑓 𝛼 > 0, inf 𝛼𝐴 = 𝛼. inf⁡
(𝐴)
3. sup 𝐴 + 𝐵 = sup 𝐴 + sup⁡ (𝐵) 6. inf 𝐴 + 𝐵 = inf 𝐴 + inf⁡ (𝐵)

Proof:
1. ∀𝑎 ∈ 𝐴, 𝑎 ≥ inf 𝐴 ∀𝑎 ∈ 𝐴, 𝛼. 𝑎 ≤ α. inf 𝐴 ∀𝑥 ∈ 𝛼𝐴, 𝑥 ≤ α. inf 𝐴 (1)
𝛼<0 𝑑𝑒𝑓 .
𝛼𝐴
𝜀 𝜀
∀𝜀 > 0, ∃𝑎 ∈ 𝐴 , 𝑎 < inf 𝐴 + ⇒ ∀𝜀 > 0, ∃𝑎 ∈ 𝐴 , 𝛼𝑎 > 𝛼 inf 𝐴 − ⇒
−𝛼 𝛼
⇒ ∀𝜀 > 0, ∃𝑥 ∈ 𝛼. 𝐴 , 𝑥 > 𝛼. inf 𝐴 − 𝜀 (2)
From (1) and (2): sup 𝛼𝐴 = 𝛼. inf⁡ (𝐴)

2. ∀𝑎 ∈ 𝐴, 𝑎 ≤ sup 𝐴 ∀𝑎 ∈ 𝐴, 𝛼. 𝑎 ≤ α. sup 𝐴 ∀𝑥 ∈ 𝛼𝐴, 𝑥 ≤ α. sup 𝐴 (1)

𝛼>0 𝑑𝑒𝑓 .
𝛼𝐴
𝜀 𝜀
∀𝜀 > 0, ∃𝑎 ∈ 𝐴 , 𝑎 > sup 𝐴 − ⇒ ∀𝜀 > 0, ∃𝑎 ∈ 𝐴 , 𝛼. 𝑎 > 𝛼. sup 𝐴 − ⇒
𝛼 𝛼
⇒ ∀𝜀 > 0, ∃𝑥 ∈ 𝛼. 𝐴 , 𝑥 > 𝛼. sup 𝐴 − 𝜀 (2)
From (1) and (2): sup 𝛼𝐴 = 𝛼. sup⁡ (𝐴)

3. ∀𝑎 ∈ 𝐴, 𝑎 ≤ sup 𝐴 𝑎𝑛𝑑 ∀𝑏 ∈ 𝐵, 𝑏 ≤ sup 𝐵 ⇒

⇒ ∀𝑎 ∈ 𝐴, ∀𝑏 ∈ 𝐵, 𝑎 + 𝑏 ≤ sup 𝐴 + sup 𝐵 ⇒
⇒ ∀𝑥 ∈ 𝐴 + 𝐵, 𝑥 ≤ sup 𝐴 + sup 𝐵 (1)
 𝜀
∀𝜀 > 0, ∃𝑎 ∈ 𝐴 , 𝑎 > sup 𝐴 − 2
𝜀 ⇒
∀𝜀 > 0, ∃𝑏 ∈ 𝐵 , 𝑏 > sup 𝐵 − 2
⇒ ∀𝜀 > 0, ∃𝑎 ∈ 𝐴 , ∃𝑏 ∈ 𝐵, 𝑎 + 𝑏 > sup 𝐴 + sup⁡(𝐵) − 𝜀 ⇒
⇒ ∀𝜀 > 0, ∃𝑥 ∈ 𝐴 + 𝐵 , 𝑥 > sup 𝐴 + sup⁡ (𝐵) − 𝜀 (2)
From (1) and (2): sup 𝛼𝐴 = 𝛼. sup⁡ (𝐴)

4. Analog to 1
5. Analog to 2
6. Analog to 3


Theorem: (COMPLETENESS Of ℝ)

𝐴 ⊂ ℝ, 𝐴 ≠ ∅ 𝑎𝑛𝑑 𝐴 𝑖𝑠 𝑏𝑑𝑑 𝑓𝑟𝑜𝑚 𝑎𝑏𝑜𝑣𝑒 ⇒ ∃𝑥 ∈ ℝ / 𝑥 = sup⁡

(𝐴)

Corollary: 𝐵 ⊂ ℝ, 𝐵 ≠ ∅ 𝑎𝑛𝑑 𝐵 𝑖𝑠 𝑏𝑑𝑑 𝑓𝑟𝑜𝑚 𝑏𝑒𝑙𝑜𝑤 ⇒ ∃𝑥 ∈ ℝ / 𝑥 = inf⁡

(𝐵)

Proof: Consider the set 𝐴 = −𝐵 = −𝑥/𝑥 ∈ 𝐵

Claim 1: A is bdd from above.

Proof: A is bdd from above ⇔ ∃𝑀 ∈ ℝ / ∀𝑦 ∈ 𝐴, 𝑀 ≥ 𝑦

Pick up any 𝑦 ∈ 𝐴 ⇒ 𝑦 = −𝑥, 𝑡𝑜 𝑠𝑜𝑚𝑒 𝑥 ∈ 𝐵

𝐵 𝑖𝑠 𝑏𝑑𝑑 𝑓𝑟𝑜𝑚 𝑏𝑒𝑙𝑜𝑤 ⇒ ∃𝑘 ∈ ℝ / ∀𝑏 ∈ 𝐵, 𝑏 ≥ 𝑘
⇒ ∃𝑘 ∈ ℝ / −𝑥 ≤ −𝑘 ⇒
𝑥∈𝐵

⇒ ∃𝑀 = −𝑘 ∈ ℝ / 𝑦 ≤ 𝑀 ∴ 𝐴 𝑖𝑠 𝑏𝑑𝑑 𝑓𝑟𝑜𝑚 𝑎𝑏𝑜𝑣𝑒

Claim 2: ∃𝑎 ∈ ℝ/ 𝑎 = sup 𝐴

Proof: 𝐴 𝑖𝑠 𝑏𝑑𝑑 𝑓𝑟𝑜𝑚 𝑎𝑏𝑜𝑣𝑒

𝐵⊂ℝ⇒𝐴⊂ℝ ∃𝑎 ∈ ℝ/ 𝑎 = sup 𝐴
𝐶𝑜𝑚𝑝𝑙𝑒𝑡𝑒𝑛𝑒𝑠𝑠
𝐵≠∅⇒𝐴≠∅

Claim 2: ∃𝑏 ∈ ℝ/ 𝑏 = inf 𝐵

Proof: 𝑎 = sup 𝐴 ⇒ −𝑎 = inf⁡

(−𝐴)
⇒ −𝑎 = inf⁡
(𝐵)
𝐴 = −𝐵 ⇒ 𝐵 = −𝐴

∴ 𝐶𝑎𝑙𝑙 𝑏 = −𝑎: ∃𝑏 ∈ ℝ / 𝑏 = inf⁡

(𝐵)

4. METRIC SPACES
𝐌𝐄𝐓𝐑𝐈𝐂 𝐒𝐏𝐀𝐂𝐄: A metric space is an ordered pair (𝐸, 𝑑) where E is a nonempty set and
𝑑: 𝐸 × 𝐸 → ℝ is a function that satisfies:
i. 𝑑 𝑥, 𝑦 ≥ 0, ∀𝑥, 𝑦 ∈ 𝐸
ii. 𝑑 𝑥, 𝑦 = 0 ⇔ 𝑥 = 𝑦
iii. 𝑑 𝑥, 𝑦 = 𝑑(𝑦, 𝑥), ∀𝑥, 𝑦 ∈ 𝐸
iv. Triangular inequality: 𝑑 𝑥, 𝑦 ≤ 𝑑 𝑥, 𝑧 + 𝑑 𝑧, 𝑦 , ∀𝑥, 𝑦, 𝑧 ∈ 𝐸

The function d is called 𝐌𝐄𝐓𝐑𝐈𝐂 or DISTANCE.

Theorem: Given a metric space 𝐸, 𝑑 , 𝑡𝑕𝑒𝑛 𝑑 𝑥, 𝑦 − 𝑑(𝑧, 𝑦) ≤ 𝑑(𝑥, 𝑧) , ∀𝑥, 𝑦, 𝑧 ∈ 𝐸

Proof:
𝑑 𝑥, 𝑦 − 𝑑(𝑧, 𝑦) ≤ 𝑑 𝑥, 𝑧 ⇔ −𝑑 𝑥, 𝑧 ≤ 𝑑 𝑥, 𝑦 − 𝑑(𝑧, 𝑦) ≤ 𝑑 𝑥, 𝑧 ⇔
⇔ 𝑑 𝑧, 𝑦 − 𝑑 𝑥, 𝑧 ≤ 𝑑 𝑥, 𝑦 ≤ 𝑑 𝑧, 𝑦 + 𝑑 𝑥, 𝑧 (3)

Since d is a metric, d satisfies ⊿ −Ineq. Then 𝑑 𝑥, 𝑦 ≤ 𝑑 𝑧, 𝑦 + 𝑑 𝑥, 𝑧 , ∀𝑥, 𝑦, 𝑧 ∈ 𝐸 (1)

Since d is a metric, d satisfies ⊿ −Ineq. Then 𝑑 𝑧, 𝑦 ≤ 𝑑 𝑧, 𝑥 + 𝑑 𝑥, 𝑦 , ∀𝑥, 𝑦, 𝑧 ∈ 𝐸 ⇒
⇒ 𝑑 𝑧, 𝑦 − 𝑑 𝑧, 𝑥 ≤ 𝑑 𝑥, 𝑦 , ∀𝑥, 𝑦, 𝑧 ∈ 𝐸
⇒ 𝑑 𝑧, 𝑦 − 𝑑 𝑥, 𝑧 ≤ 𝑑 𝑥, 𝑦 , ∀𝑥, 𝑦, 𝑧 ∈ 𝐸 (2)
𝑑 𝑖𝑠 𝑎 𝑚𝑒𝑡𝑟𝑖𝑐 ⇒ 𝑑 𝑧, 𝑥 = 𝑑 𝑥, 𝑧

From (1) and (2): 𝑑 𝑧, 𝑦 − 𝑑 𝑥, 𝑧 ≤ 𝑑 𝑥, 𝑦 ≤ 𝑑 𝑧, 𝑦 + 𝑑 𝑥, 𝑧

(3)
⇒ 𝑑 𝑥, 𝑦 − 𝑑(𝑧, 𝑦) ≤ 𝑑 𝑥, 𝑧


4.1 METRIC SUB-SPACES

Theorem (HW): Let (𝐸, 𝑑) be any metric space. 𝐴 ⊂ 𝐸, 𝐴 ≠ ∅ ⇒ 𝐴, 𝑑 𝑖𝑠 𝑎 𝑚𝑒𝑡𝑟𝑖𝑐 𝑠𝑝𝑎𝑐𝑒

Proof: Directly because d is metric. 

𝐌𝐄𝐓𝐑𝐈𝐂 𝐒𝐔𝐁 − 𝐒𝐏𝐀𝐂𝐄: Given 𝐸, 𝑑 a metric space and 𝐴 ⊂ 𝐸, 𝐴 ≠ ∅; 𝑡𝑕𝑒 𝑚𝑒𝑡𝑟𝑖𝑐 𝑠𝑝𝑎𝑐𝑒
𝐴, 𝑑 𝑖𝑠 𝑐𝑎𝑙𝑙𝑒𝑑 𝑎 𝑚𝑒𝑡𝑟𝑖𝑐 𝑠𝑢𝑏 − 𝑠𝑝𝑎𝑐𝑒 𝑜𝑓 (𝐸, 𝑑).
4.2 SOME METRIC SPACES

𝑫𝒊𝒔𝒄𝒓𝒆𝒕𝒆 𝒎𝒆𝒕𝒓𝒊𝒄 𝒔𝒑𝒂𝒄𝒆𝒔

Observation (HW): (DISCRETE METRIC)

1, 𝑖𝑓 𝑥 ≠ 𝑦
(𝐸, 𝑑) with 𝑑: 𝐸 × 𝐸 → ℝ ; 𝑑 𝑥, 𝑦 = , is a metric space.
0, 𝑖𝑓 𝑥 = 𝑦

Proof:
i. 𝒅 𝒙, 𝒚 ≥ 𝟎, ∀𝒙, 𝒚 ∈ 𝑬:
𝐼𝑓 𝑥 ≠ 𝑦 ⇒ 𝑑 𝑥, 𝑦 = 1 ≥ 0
⇒ ∀𝑥, 𝑦 ∈ 𝐸, 𝑑(𝑥, 𝑦) ≥ 0
𝐼𝑓 𝑥 = 𝑦 ⇒ 𝑑 𝑥, 𝑦 = 0 ≥ 0

ii. 𝒅 𝒙, 𝒚 = 𝟎 ⇔ 𝒙 = 𝒚: By the definition of the function.

iii. 𝒅 𝒙, 𝒚 = 𝒅(𝒚, 𝒙), ∀𝒙, 𝒚 ∈ 𝑬: By the definition of the function.

iv. Triangular inequality: 𝒅 𝒙, 𝒚 ≤ 𝒅 𝒙, 𝒛 + 𝒅 𝒛, 𝒚 , ∀𝒙, 𝒚, 𝒛 ∈ 𝑬

Case 1: 𝑥 = 𝑦 = 𝑧 𝑑 𝑥, 𝑦 = 0 = 0 + 0 = 𝑑 𝑥, 𝑧 + 𝑑(𝑧, 𝑦)
Case 2: 𝑥 = 𝑦 ≠ 𝑧 𝑑 𝑥, 𝑦 = 0 ≤ 2 = 1 + 1 = 𝑑 𝑥, 𝑧 + 𝑑(𝑧, 𝑦)
Case 3: 𝑥 ≠ 𝑦 = 𝑧 𝑑 𝑥, 𝑦 = 1 = 1 + 0 = 𝑑 𝑥, 𝑧 + 𝑑(𝑧, 𝑦)
Case 4: 𝑥 = 𝑧 ≠ 𝑦 𝑑 𝑥, 𝑦 = 1 = 0 + 1 = 𝑑 𝑥, 𝑧 + 𝑑(𝑧, 𝑦)
Case 5: 𝑥 ≠ 𝑦, 𝑦 ≠ 𝑧 𝑎𝑛𝑑 𝑥 ≠ 𝑧 𝑑 𝑥, 𝑦 = 1 ≤ 2 = 1 + 1 = 𝑑 𝑥, 𝑧 + 𝑑(𝑧, 𝑦)

𝒏 𝑫𝒊𝒆𝒎𝒏𝒔𝒊𝒐𝒏𝒂𝒍 𝑬𝒖𝒄𝒍𝒊𝒅𝒆𝒂𝒏 𝒔𝒑𝒂𝒄𝒆𝒔

Observation: (USUAL DISTANCE IN ℝ)

(ℝ, 𝑑) with 𝑑: ℝ × ℝ → ℝ ; 𝑑 𝑥, 𝑦 = 𝑥 − 𝑦 , is a metric space.

Proof:
1) 𝑑 𝑥, 𝑦 = 𝑥 − 𝑦 ≥ 0, ∀𝑥, 𝑦 ∈ ℝ
2) 𝑑 𝑥, 𝑦 = 0 ⇔ 𝑥 − 𝑦 = 0 ⇔ 𝑥 − 𝑦 = 0 ⇔ 𝑥 = 𝑦
3) 𝑑 𝑥, 𝑦 = 𝑥 − 𝑦 = − 𝑥 − 𝑦 = 𝑦 − 𝑥 = 𝑑 𝑦, 𝑥 , ∀𝑥, 𝑦 ∈ ℝ
4) 𝑑 𝑥, 𝑦 = 𝑥 − 𝑦 = 𝑥 − 𝑧 + 𝑧 − 𝑦 = 𝑥 − 𝑧 + (𝑧 − 𝑦) ≤𝑙𝑜𝑜𝑘 𝑛𝑜𝑡𝑒
≤ 𝑥 − 𝑧 + 𝑧 − 𝑦 = 𝑑 𝑥, 𝑧 + 𝑑(𝑧, 𝑦), ∀𝑥, 𝑦 ∈ ℝ

Note: ∀𝑎, 𝑏 ∈ ℝ, 𝑎 + 𝑏 ≤ 𝑎 + 𝑏

∀𝑎 ∈ ℝ, − 𝑎 ≤ 𝑎 ≤ 𝑎
⇒ ∀𝑎, 𝑏 ∈ ℝ, − 𝑎 + 𝑏 ≤ 𝑎+𝑏 ≤ 𝑎 + 𝑏 ⇒
∀𝑏 ∈ ℝ, − 𝑏 ≤ 𝑏 ≤ 𝑏
⇒ 𝑎+𝑏 ≤ 𝑎 + 𝑏 

Lema: (SCHWARZ INEQUALITY)

𝑖=𝑛 𝑖=𝑛 𝑖=𝑛
.∀𝑛 ≥ 1, ∀𝑥, 𝑦 ∈ ℝ𝑛 , 𝑖=1 𝑥𝑖 𝑦𝑖 ≤ 𝑖=1 𝑥𝑖 2 𝑖=1 𝑦𝑖 2
Proof:
Case 1: 𝑥 ≠ 0, … ,0 and 𝑦 ≠ 0, … ,0
Let s and t be any two real numbers.

. 𝑖=𝑛
𝑖=1 𝑠. 𝑥𝑖 + 𝑡. 𝑦𝑖
2 ≥ 0 ⇒ 𝑖=𝑛 𝑠 2 . 𝑥 2 + 𝑡 2 . 𝑦 2 + 2𝑠𝑡𝑥 𝑦 ≥ 0 ⇒
𝑖=1 𝑖 𝑖 𝑖 𝑖
.⇒ 𝑠 2 . 𝑖=1 𝑥𝑖 2 + 𝑡 2 . 𝑖=𝑛
𝑖=𝑛 2 𝑖=𝑛
𝑖=1 𝑦𝑖 + 2𝑠𝑡. 𝑖=1 𝑥𝑖 𝑦𝑖 ≥ 0 (1)
𝑖=𝑛
𝑠= 𝑖=1 𝑦𝑖 2
𝑖=𝑛 2 𝑖=𝑛 𝑖=𝑛 𝑖=𝑛 𝑖=𝑛
.𝐼𝑓 2. 𝑖=1 𝑦𝑖 . 𝑖=1 𝑥𝑖 2 + 2 𝑖=1 𝑦𝑖 2 𝑖=1 𝑥𝑖 2 𝑖=1 𝑥𝑖 𝑦𝑖 ≥ 0 ⇒
(1)
𝑖=𝑛 2
𝑡= 𝑖=1 𝑥𝑖

𝑖=𝑛 𝑖=𝑛 𝑖=𝑛 𝑖=𝑛 𝑖=𝑛

.⇒ 𝑖=1 𝑦𝑖 2 . 𝑖=1 𝑥𝑖 2 ≥ − 𝑖=1 𝑦𝑖 2 𝑖=1 𝑥𝑖 2 𝑖=1 𝑥𝑖 𝑦𝑖 ⇒

𝑖=𝑛 𝑖=𝑛 𝑖=𝑛 𝑖=𝑛 𝑖=𝑛

⇒− 𝑖=1 𝑦𝑖 2 . 𝑖=1 𝑥𝑖 2 ≤ 𝑖=1 𝑦𝑖 2 𝑖=1 𝑥𝑖 2 𝑖=1 𝑥𝑖 𝑦𝑖 (i)

𝑖=𝑛
𝑠= 𝑖=1 𝑦𝑖 2
𝑖=𝑛 𝑖=𝑛 𝑖=𝑛 𝑖=𝑛 2 𝑖=𝑛
.𝐼𝑓 2. 𝑖=1 𝑦𝑖 2 . 𝑖=1 𝑥𝑖 2 + 2 𝑖=1 𝑦𝑖 2 𝑖=1 𝑥𝑖 𝑖=1 𝑥𝑖 𝑦𝑖 ≥ 0 ⇒
(1)
𝑖=𝑛 2
𝑡=− 𝑖=1 𝑥𝑖

𝑖=𝑛 𝑖=𝑛 𝑖=𝑛 𝑖=𝑛 𝑖=𝑛

.⇒ 𝑖=1 𝑦𝑖 2 . 𝑖=1 𝑥𝑖 2 ≥ 𝑖=1 𝑦𝑖 2 𝑖=1 𝑥𝑖 2 𝑖=1 𝑥𝑖 𝑦𝑖 (ii)

𝑖=𝑛 𝑖=𝑛 𝑖=𝑛 𝑖=𝑛 2 𝑖=𝑛 𝑖=𝑛 𝑖=𝑛

From i and ii: − 𝑖=1 𝑦𝑖 2 . 𝑖=1 𝑥𝑖 2 ≤ 𝑖=1 𝑦𝑖 2 𝑖=1 𝑥𝑖 𝑖=1 𝑥𝑖 𝑦𝑖 ≤ 𝑖=1 𝑦𝑖 2 . 𝑖=1 𝑥𝑖 2 ⇒
𝑖=𝑛 𝑖=𝑛 𝑖=𝑛 𝑖=𝑛
– 𝑦𝑖 2 . 𝑥𝑖2 𝑖=𝑛 𝑦𝑖 2 . 𝑥𝑖2
. 𝑖=1 𝑖=1
≤ 𝑖=1 𝑥𝑖 𝑦𝑖 ≤ 𝑖=1 𝑖=1
⇒
∗ 𝑖=𝑛 2 𝑖=𝑛 2 𝑖=𝑛 2 𝑖=𝑛
𝑖=1 𝑦𝑖 𝑖=1 𝑥𝑖 𝑖=1 𝑦𝑖 𝑖=1 𝑥𝑖2

𝑖=𝑛 𝑖=𝑛 𝑖=𝑛 𝑖=𝑛 𝑖=𝑛

.⇒ − 𝑖=1 𝑦𝑖 2 𝑖=1 𝑥𝑖 2 ≤ 𝑖=1 𝑥𝑖 𝑦𝑖 ≤ 𝑖=1 𝑦𝑖 2 𝑖=1 𝑥𝑖 2 ⇒

𝑖=𝑛 𝑖=𝑛 𝑖=𝑛 2

.⇒ 𝑖=1 𝑥𝑖 𝑦𝑖 ≤ 𝑖=1 𝑦𝑖 2 𝑖=1 𝑥𝑖

𝑥 ≠ 0, … ,0 𝑖=𝑛 𝑖=𝑛
. ∗ ⇒ 𝑖=1 𝑦𝑖 2 𝑖=1 𝑥𝑖 2 > 0
𝑦 ≠ (0, … ,0)

Case 2: 𝑥 ≠ 0, … ,0
𝑖=𝑛 𝑖=𝑛 𝑖=𝑛 𝑖=𝑛 𝑖=𝑛
. 𝑖=1 𝑥𝑖 𝑦𝑖 = 𝑖=1 0 =0= 𝑖=1 𝑦𝑖 2 . 0 = 𝑖=1 𝑦𝑖 2 𝑖=1 𝑥𝑖 2

Case 3: 𝑦 ≠ 0, … ,0
𝑖=𝑛 𝑖=𝑛 𝑖=𝑛 𝑖=𝑛 𝑖=𝑛 2
. 𝑖=1 𝑥𝑖 𝑦𝑖 = 𝑖=1 0 = 0 = 0. 𝑖=1 𝑥𝑖 2 = 𝑖=1 𝑦𝑖 2 𝑖=1 𝑥𝑖


𝑖=𝑛 𝑖=𝑛 𝑖=𝑛 2

Corollary: .∀𝑛 ≥ 1, ∀𝑥, 𝑦 ∈ ℝ𝑛 , 𝑖=1 (𝑥𝑖 + 𝑦𝑖 )2 ≤ 𝑖=1 𝑥𝑖 2 + 𝑖=1 𝑦𝑖

𝑖=𝑛 𝑖=𝑛 𝑖=𝑛 𝑖=𝑛

Proof: 𝑖=1 (𝑥𝑖 + 𝑦𝑖 )2 = 𝑖=1 𝑥𝑖 2 + 𝑖=1 𝑦𝑖 2 + 2. 𝑖=1 𝑥𝑖 𝑦𝑖 ≤
 2 2
𝑖=𝑛 𝑖=𝑛 𝑖=𝑛
.≤ 𝑖=1 𝑥𝑖 2 + 𝑖=1 𝑦𝑖 2 + 2. 𝑖=1 𝑥𝑖 𝑦𝑖 ≤𝑆𝑐𝑕𝑢𝑎𝑟𝑡𝑧
2 2
𝑖=𝑛 𝑖=𝑛 𝑖=𝑛 𝑖=𝑛
.≤ 𝑖=1 𝑥𝑖 2 + 𝑖=1 𝑦𝑖 2 + 2. 𝑖=1 𝑥𝑖 2 𝑖=1 𝑦𝑖 2 =
2
𝑖=𝑛 𝑖=𝑛
.= 𝑖=1 𝑥𝑖 2 + 𝑖=1 𝑦𝑖 2 ⇒

𝑖=𝑛 𝑖=𝑛 𝑖=𝑛

.⇒ 𝑖=1 (𝑥𝑖 + 𝑦𝑖 )2 ≤ 𝑖=1 𝑥𝑖 2 + 𝑖=1 𝑦𝑖 2 

Preposition: (USUAL DISTANCE IN ℝ𝒏)

𝑖=𝑛
(ℝ𝑛 , 𝑑) with 𝑑: ℝ𝑛 × ℝ𝑛 → ℝ ; 𝑑 𝑥, 𝑦 = 𝑖=1 (𝑥𝑖 − 𝑦𝑖 )2 , is a metric space .∀𝑛 ≥ 1.

Proof:

i. 𝒅 𝒙, 𝒚 ≥ 𝟎, ∀𝒙, 𝒚 ∈ ℝ𝑛 :
𝑖=𝑛
.𝑑 𝑥, 𝑦 = 𝑖=1 (𝑥𝑖 − 𝑦𝑖 )2 ≥ 0, , ∀𝑥, 𝑦 ∈ ℝ𝑛

ii. 𝒅 𝒙, 𝒚 = 𝟎 ⇔ 𝒙 = 𝒚: By the definition of the function.

𝑖=𝑛 𝑖=𝑛
.𝑑 𝑥, 𝑦 = 0 ⇔ 𝑖=1 (𝑥𝑖 − 𝑦𝑖 )2 = 0 ⇔ 𝑖=1 (𝑥𝑖 − 𝑦𝑖 )2 = 0 ⇔ (𝑥𝑖 − 𝑦𝑖 )2 = 0 ⇔
≥0
⇔ 𝑥𝑖 − 𝑦𝑖 = 0 ⇔ 𝑥𝑖 = 𝑦𝑖

iii. 𝒅 𝒙, 𝒚 = 𝒅(𝒚, 𝒙), ∀𝒙, 𝒚 ∈ ℝ𝑛 : By the definition of the function.

𝑖=𝑛 𝑖=𝑛
.𝑑 𝑥, 𝑦 = 𝑖=1 (𝑥𝑖 − 𝑦𝑖 )2 = 𝑖=1 (𝑦𝑖 − 𝑥𝑖 )2 = 𝑑(𝑦, 𝑥)

iv. Triangular inequality: 𝒅 𝒙, 𝒚 ≤ 𝒅 𝒙, 𝒛 + 𝒅 𝒛, 𝒚 , ∀𝒙, 𝒚, 𝒛 ∈ ℝ𝑛

𝑖=𝑛 𝑖=𝑛 𝑖=𝑛

.𝑑 𝑥, 𝑦 ≤ 𝑑 𝑥, 𝑧 + 𝑑 𝑧, 𝑦 ⇔ 𝑖=1 (𝑥𝑖 − 𝑦𝑖 )2 ≤ 𝑖=1 (𝑥𝑖 − 𝑧𝑖 )2 + 𝑖=1 (𝑧𝑖 − 𝑦𝑖 )2

Let 𝑎𝑖 = 𝑥𝑖 − 𝑧𝑖 and 𝑏𝑖 = 𝑧𝑖 − 𝑦𝑖
Because corollary: 𝑖=𝑛
𝑖=1 (𝑎𝑖 + 𝑏𝑖 )2 ≤ 𝑖=𝑛
𝑖=1 𝑎𝑖 2 + 𝑖=𝑛
𝑖=1 𝑏𝑖 2 ⇒

𝑖=𝑛 𝑖=𝑛 𝑖=𝑛

.⇒ 𝑖=1 (𝑥𝑖 − 𝑧𝑖 + 𝑧𝑖 − 𝑦𝑖 )2 ≤ 𝑖=1 (𝑥𝑖 − 𝑧𝑖 )2 + 𝑖=1 (𝑧𝑖 − 𝑦𝑖 )2 ⇒

𝑖=𝑛 𝑖=𝑛 𝑖=𝑛

.⇒ 𝑖=1 (𝑥𝑖 − 𝑦𝑖 )2 ≤ 𝑖=1 (𝑥𝑖 − 𝑧𝑖 )2 + 𝑖=1 (𝑧𝑖 − 𝑦𝑖 )2


𝐧 𝐃𝐈𝐌𝐄𝐍𝐓𝐈𝐎𝐍𝐀𝐋 𝐄𝐔𝐂𝐋𝐈𝐃𝐄𝐀𝐍 𝐒𝐏𝐀𝐂𝐄: .∀𝑛 ≥ 1 (ℝ𝑛 , 𝑑) with 𝑑: ℝ𝑛 × ℝ𝑛 → ℝ ; 𝑑 𝑥, 𝑦 =

𝑖=𝑛
𝑖=1 (𝑥𝑖 − 𝑦𝑖 )2, is called n Dimensional Euclidean Space, denoted by 𝐸 𝑛 .
𝑶𝒕𝒉𝒆𝒓 𝒎𝒆𝒕𝒓𝒊𝒄 𝒊𝒏 ℝ𝒏

Observation (Práctico):
𝑖=𝑛
(ℝ𝒏 , 𝑑) is a metric space, with 𝑑: ℝ𝒏 × ℝ𝒏 → ℝ ; 𝑑 𝑥, 𝑦 = 𝑖=1 𝑥𝑖 − 𝑦𝑖 .

Corollary (Quiz 3): ℝ2 , 𝑑 𝑤𝑖𝑡𝑕 𝑑 𝑥, 𝑦 = 𝑥1 − 𝑦1 + 𝑥2 − 𝑦2 𝑖𝑠 𝑎 𝑚𝑒𝑡𝑟𝑖𝑐 𝑠𝑝𝑎𝑐𝑒.

𝑨 𝒎𝒆𝒕𝒓𝒊𝒄 𝒊𝒏 𝒕𝒉𝒆 𝒔𝒑𝒂𝒄𝒆 𝒐𝒇 𝒕𝒉𝒆 𝑩𝒅𝒅 𝒔𝒆𝒒𝒖𝒆𝒏𝒄𝒆𝒔

Observation (HW): (THE SUP METRIC)

(𝐸, 𝑑) is a metric space, with 𝐸 = 𝑥1 , … , 𝑥𝑛 , … 𝑏𝑑𝑑 /𝑥𝑖 ∈ ℝ, ∀𝑖 ∈ ℕ∗ 𝑎𝑛𝑑 𝑑: 𝐸 × 𝐸 → ℝ ;
𝑑 𝑥, 𝑦 = 𝑠𝑢𝑝𝑖∈ℕ∗ 𝑥𝑖 − 𝑦𝑖 .

𝑪𝒂𝒓𝒕𝒆𝒔𝒊𝒂𝒏 𝑷𝒓𝒐𝒅𝒖𝒄𝒕 𝒐𝒇 𝒎𝒆𝒕𝒓𝒊𝒄 𝒔𝒑𝒂𝒄𝒆𝒔

Observation (HW & Práctico): (THE SUM DISTANCE)

𝐿𝑒𝑡 (𝐸1 , 𝑑1 ) and (𝐸2 , 𝑑2 ) be metric spaces. 𝐸, 𝑑 with 𝐸 = 𝐸1 × 𝐸2 𝑎𝑛𝑑 𝑑: 𝐸 × 𝐸 → ℝ ;
𝑑 𝑥, 𝑦 = 𝑑1 𝑥1 + 𝑦1 + 𝑑2 (𝑥2 + 𝑦2 ) 𝑖𝑠 𝑎 𝑚𝑒𝑡𝑟𝑖𝑐 𝑠𝑝𝑎𝑐𝑒.

Observation (Práctico): (THE MAX DISTNACE)

𝐿𝑒𝑡 (𝐸1 , 𝑑1 ) and (𝐸2 , 𝑑2 ) be metric spaces. 𝐸, 𝑑 with 𝐸 = 𝐸1 × 𝐸2 𝑎𝑛𝑑 𝑑: 𝐸 × 𝐸 → ℝ ;
𝑑 𝑥, 𝑦 = 𝑚𝑎𝑥 𝑑1 𝑥1 + 𝑦1 , 𝑑2 (𝑥2 + 𝑦2 ) 𝑖𝑠 𝑎 𝑚𝑒𝑡𝑟𝑖𝑐 𝑠𝑝𝑎𝑐𝑒.

Corollary (Quiz 3): ℝ2 , 𝑑 𝑤𝑖𝑡𝑕 𝑑 𝑥, 𝑦 = 𝑚𝑎𝑥 𝑥1 − 𝑦1 , 𝑥1 − 𝑦1 𝑖𝑠 𝑎 𝑚𝑒𝑡𝑟𝑖𝑐 𝑠𝑝𝑎𝑐𝑒.

5. TOPOLOGY IN METRIC SPACES
𝐎𝐏𝐄𝐍 𝐁𝐀𝐋𝐋: Given a metric space 𝐸, 𝑑 , 𝑝 ∈ 𝐸 𝑎𝑛𝑑 𝑟 ∈ ℝ+ we call open ball centered at p
with radius r to the set 𝐵 𝑝, 𝑟 = 𝑥 ∈ 𝐸 / 𝑑(𝑥, 𝑝) < 𝑟

Observation: Given a metric space 𝐸, 𝑑 , 𝐵 𝑝, 𝑟 ≠ ∅

Proof: d is a metric ⇒ 𝑑 𝑝, 𝑝 = 0 < 𝑟 ⇒ 𝑝 ∈ 𝐵 𝑝, 𝑟 ⇒ 𝐵 𝑝, 𝑟 ≠ ∅ 

𝐂𝐋𝐎𝐒𝐄𝐃 𝐁𝐀𝐋𝐋: Given a metric space 𝐸, 𝑑 , 𝑝 ∈ 𝐸 𝑎𝑛𝑑 𝑟 ∈ ℝ+ we call closed ball centered at p
with radius r to the set 𝐵 𝑝, 𝑟 = 𝑥 ∈ 𝐸 / 𝑑(𝑥, 𝑝) ≤ 𝑟

Observation: Given a metric space 𝐸, 𝑑 , 𝐵 𝑝, 𝑟 ≠ ∅

Proof: d is a metric ⇒ 𝑑 𝑝, 𝑝 = 0 ≤ 𝑟 ⇒ 𝑝 ∈ 𝐵 𝑝, 𝑟 ⇒ 𝐵 𝑝, 𝑟 ≠ ∅ 

𝐎𝐏𝐄𝐍 𝐒𝐄𝐓: Given a metric space 𝐸, 𝑑 𝑎𝑛𝑑 𝑎 𝑠𝑒𝑡 𝐴 ⊂ 𝐸 (𝐴 𝑐𝑜𝑢𝑙𝑑 𝑏𝑒 𝑒𝑚𝑝𝑡𝑦). We say that A is
open in 𝐸, 𝑑 ⇔ ∀𝑝 ∈ 𝐴, ∃𝐵 𝑝, 𝑟 ⊂ 𝐴

𝐂𝐋𝐎𝐒𝐄𝐃 𝐒𝐄𝐓: Given a metric space 𝐸, 𝑑 𝑎𝑛𝑑 𝑎 𝑠𝑒𝑡 𝐴 ⊂ 𝐸 (𝐴 𝑐𝑜𝑢𝑙𝑑 𝑏𝑒 𝑒𝑚𝑝𝑡𝑦). We say that A
is closed in 𝐸, 𝑑 ⇔ 𝐴𝐶 𝑖𝑠 𝑜𝑝𝑒𝑛 𝑖𝑛 (𝐸, 𝑑)

Theorem: Given a metric space 𝐸, 𝑑 , 𝑎𝑛𝑦 𝑜𝑝𝑒𝑛 𝑏𝑎𝑙𝑙 𝑖𝑠 𝑜𝑝𝑒𝑛

Proof: Pick up any open ball, say 𝐵 𝑝, 𝑟 𝑓𝑜𝑟 𝑠𝑜𝑚𝑒 𝑝 ∈ 𝐸 𝑎𝑛𝑑 𝑟 ∈ ℝ+.

Because definition, 𝐵 𝑝, 𝑟 is open ⇔ ∀𝑢 ∈ 𝐵 𝑝, 𝑟 , ∃𝐵 𝑢, 𝑠 ⊂ 𝐵 𝑝, 𝑟

Define 𝑠 = 𝑟 − 𝑑(𝑝, 𝑢)

𝑢 ∈ 𝐵 𝑝, 𝑟 𝑑 𝑝, 𝑢 < 𝑟 ⇒ 𝑠 = 𝑟 − 𝑑(𝑝, 𝑢) > 0 (1)

𝑑𝑒𝑓 .

𝐼𝑓 𝑥 ∈ 𝐵 𝑢, 𝑠 ⇒ 𝑑 𝑥, 𝑢 < 𝑠 = 𝑟 − 𝑑 𝑢, 𝑝 ⇒ 𝑑 𝑥, 𝑢 + 𝑑 𝑢, 𝑝 < 𝑟
𝑥, 𝑢, 𝑝 ∈ 𝐸 𝑑 𝑥, 𝑝 ≤ 𝑑 𝑥, 𝑢 + 𝑑 𝑢, 𝑝 ⇒ 𝑑 𝑥, 𝑝 < 𝑟 ⇒
⊿−𝐼𝑛𝑒𝑞𝑢𝑎𝑙𝑖𝑡𝑦
⇒ 𝑥 ∈ 𝐵 𝑝, 𝑟
∴ 𝐵 𝑢, 𝑠 ⊂ 𝐵 𝑝, 𝑟 (2)

From (1) and (2): ⇔ ∀𝑢 ∈ 𝐵 𝑝, 𝑟 , ∃𝑠 = 𝑟 − 𝑑 𝑝, 𝑢 > 0/𝐵 𝑢, 𝑠 ⊂ 𝐵 𝑝, 𝑟 ⇒

⇒ 𝐵 𝑝, 𝑟 is open


Theorem: Given a metric space 𝐸, 𝑑 , 𝑎𝑛𝑦 𝑐𝑙𝑜𝑠𝑒𝑑 𝑏𝑎𝑙𝑙 𝑖𝑠 𝑐𝑙𝑜𝑠𝑒𝑑

Proof: Pick up any closed ball, say 𝐵 𝑝, 𝑟 𝑓𝑜𝑟 𝑠𝑜𝑚𝑒 𝑝 ∈ 𝐸 𝑎𝑛𝑑 𝑟 ∈ ℝ+.
𝐶
Because definition, 𝐵 𝑝, 𝑟 is closed ⇔ 𝐵 𝑝, 𝑟 is open ⇔
 𝐶 𝐶
⇔ ∀𝑢 ∈ 𝐵 𝑝, 𝑟 , ∃𝐵 𝑢, 𝑠 ⊂ 𝐵 𝑝, 𝑟
Define 𝑠 = 𝑑 𝑝, 𝑢 − 𝑟
𝐶
𝑢 ∈ 𝐵 𝑝, 𝑟 𝑑 𝑝, 𝑢 > 𝑟 ⇒ 𝑠 = 𝑑 𝑝, 𝑢 − 𝑟 > 0 (1)
𝑑𝑒𝑓 .

𝐼𝑓 𝑥 ∈ 𝐵 𝑢, 𝑠 ⇒ 𝑑 𝑥, 𝑢 < 𝑠 = 𝑑 𝑢, 𝑝 − 𝑟 ⇒ 𝑑 𝑢, 𝑝 − 𝑑 𝑢, 𝑥 > 𝑟
𝑥, 𝑢, 𝑝 ∈ 𝐸 𝑑 𝑝, 𝑢 ≤ 𝑑 𝑥, 𝑝 + 𝑑 𝑢, 𝑥 ⇒ 𝑑 𝑝, 𝑢 − 𝑑 𝑢, 𝑥 ≤ 𝑑 𝑥, 𝑝
⊿−𝐼𝑛𝑒𝑞𝑢𝑎𝑙𝑖𝑡𝑦
𝐶
⇒ 𝑑 𝑥, 𝑝 > 𝑟 ⇒ 𝑥 ∉ 𝐵 𝑝, 𝑟 ⇒ 𝑥 ∈ 𝐵 𝑝, 𝑟
𝐶
∴ 𝐵 𝑢, 𝑠 ⊂ 𝐵 𝑝, 𝑟 (2)

𝐶 𝐶
From (1) and (2): ⇔ ∀𝑢 ∈ 𝐵 𝑝, 𝑟 , ∃𝑠 = 𝑑 𝑝, 𝑢 − 𝑟 > 0/𝐵 𝑢, 𝑠 ⊂ 𝐵 𝑝, 𝑟 ⇒
𝐶
⇒ 𝐵 𝑝, 𝑟 is open ⇒ 𝐵 𝑝, 𝑟 is closed.


Theorem: Given a metric space 𝐸, 𝑑 , 𝑡𝑕𝑒𝑛:

1. ∅ is open
2. 𝐸 𝑖𝑠 𝑜𝑝𝑒𝑛
3. 𝑈1 , … , 𝑈𝑛 ⊂ 𝐸 𝑎𝑟𝑒 𝑜𝑝𝑒𝑛 ⇒ 𝑖=𝑛
𝑖=1 𝑈𝑖 is open.
4. 𝐹𝑜𝑟 𝑎𝑛𝑦 𝑓𝑎𝑚𝑖𝑙𝑦 𝑜𝑓 𝑜𝑝𝑒𝑛 𝑠𝑒𝑡𝑠 𝑈𝜆 𝜆 ∈ 𝐼 ⊂ 𝐸 ⇒ 𝜆∈𝐼 𝑈𝜆 is open.

Proof:

1. Following directly from the definition.

2. Pick up any 𝑝 ∈ 𝐸 ⇒ 𝐵(𝑝, 1) ⊂ 𝐸. ∴ ∀𝑝 ∈ 𝐸, ∃𝐵 𝑝, 1 ⊂ 𝐸 ⇒ 𝐸 𝑖𝑠 𝑜𝑝𝑒𝑛

𝑃𝑖𝑐𝑘 𝑢𝑝 𝑝 ∈ 𝑖=𝑛
3. 𝑖=1 𝑈𝑖 ⇒ 𝑝 ∈ 𝑈𝑖 , ∀𝑖 = 1, … , 𝑛
. ⇒ ∀𝑖 = 1, … , 𝑛, ∃𝐵 𝑝, 𝑟𝑖 ⊂ 𝑈𝑖
𝑈𝑖 𝑖𝑠 𝑜𝑝𝑒𝑛 ∀𝑖 = 1, … , 𝑛
𝑟>0
Define 𝑟 = min 𝑟𝑖 /𝑖 = 1, … , 𝑛 ⇒
𝑟 ≤ 𝑟𝑖 , ∀𝑖 = 1, … , 𝑛 ⇒
.⇒ 𝐵 𝑝, 𝑟 ⊂ 𝐵 𝑝, 𝑟𝑖 ⊂ 𝑈𝑖 , ∀𝑖 = 1, … , 𝑛 ⇒ 𝐵 𝑝, 𝑟 ⊂ 𝑖=𝑛 𝑖=1 𝑈𝑖
∴ ∀𝑝 ∈ 𝑖=𝑛 𝑈
𝑖=1 𝑖 , ∃𝐵 𝑝, 𝑟 ⊂ 𝑖=𝑛
𝑈
𝑖=1 𝑖 ⇒ 𝑖=𝑛
𝑈
𝑖=1 𝑖 is open

4.
𝑃𝑖𝑐𝑘 𝑢𝑝 𝑎𝑛𝑦 𝑝𝑜𝑖𝑛𝑡 𝑝 ∈ 𝜆∈𝐼 𝑈𝜆 ⇒ ∃𝜆 ∈ 𝐼 ; 𝑝 ∈ 𝑈𝜆
. ⇒ ∃𝐵 𝑝, 𝑟 ⊂ 𝑈𝜆 ⇒
𝑈𝜆 𝑖𝑠 𝑜𝑝𝑒𝑛, ∀𝜆 ∈ 𝐼
⇒ ∃𝐵 𝑝, 𝑟 ⊂ 𝑈𝜆 𝜆∈𝐼 𝑈𝜆 𝐸 ⇒ 𝜆∈𝐼 𝑈𝜆 is open.


NOTE: Given a metric space 𝐸, 𝑑 , any collection of subsets of E satisfying 1-4 above is called a
TOPOLOGY on E.
Theorem: Given a metric space 𝐸, 𝑑 , 𝑡𝑕𝑒𝑛:

1. ∅ is closed
2. 𝐸 𝑖𝑠 𝑐𝑙𝑜𝑠𝑒𝑑
𝑖=𝑛
3. 𝑈1 , … , 𝑈𝑛 ⊂ 𝐸 𝑎𝑟𝑒 𝑐𝑙𝑜𝑠𝑒𝑑 ⇒ 𝑖=1 𝑈𝑖 is closed.
4. 𝐹𝑜𝑟 𝑎𝑛𝑦 𝑓𝑎𝑚𝑖𝑙𝑦 𝑜𝑓 𝑐𝑙𝑜𝑠𝑒𝑑 𝑠𝑒𝑡𝑠 𝑈𝜆 𝜆 ∈ 𝐼 ⊂ 𝐸 ⇒ 𝜆∈𝐼 𝑈𝜆 is closed.

Proof:

1. By definition, ∅ is closed iff ∅𝐶 = 𝐸 𝑖𝑠 𝑜𝑝𝑒𝑛, 𝑤𝑕𝑎𝑡 𝑖𝑠 𝑡𝑟𝑢𝑒 𝑏𝑦 𝑡𝑕𝑒𝑜𝑟𝑒𝑚 𝑏𝑒𝑓𝑜𝑟𝑒

2. By definition E is closed iff 𝐸 𝐶 = ∅ 𝑖𝑠 𝑜𝑝𝑒𝑛, 𝑤𝑕𝑎𝑡 𝑖𝑠 𝑡𝑟𝑢𝑒 𝑏𝑦 𝑡𝑕𝑒𝑜𝑟𝑒𝑚 𝑏𝑒𝑓𝑜𝑟𝑒

𝐶 𝑖=𝑛 𝑐
3. 𝑈𝑖 𝑖𝑠 𝑐𝑙𝑜𝑠𝑒𝑑 ∀𝑖 = 1, … , 𝑛 ⇒ 𝑈𝑖 𝑖𝑠 𝑜𝑝𝑒𝑛 ∀𝑖 = 1, … , 𝑛 𝑖=1 𝑈𝑖 𝑖𝑠 𝑜𝑝𝑒𝑛
𝑇𝑕.
. 𝑏𝑒𝑓𝑜𝑟𝑒 ⇒
𝑖=𝑛 𝐶 𝑖=𝑛 𝑐
𝐵𝑒𝑐𝑎𝑢𝑠𝑒 𝑀𝑜𝑟𝑔𝑎𝑛𝑠: 𝑖=1 𝑈𝑖 = 𝑖=1 𝑈𝑖
𝑖=𝑛 𝐶 𝑖=𝑛
⇒ 𝑖=1 𝑈𝑖 is open ⇒ 𝑖=1 𝑈𝑖 is closed

4. 𝑈𝜆 𝑖𝑠 𝑐𝑙𝑜𝑠𝑒𝑑 ∀𝜆 ∈ 𝐼 ⇒ 𝑈𝜆 𝐶 𝑖𝑠 𝑜𝑝𝑒𝑛 ∀𝜆 ∈ 𝐼 𝑈𝜆 𝐶 𝑖𝑠 𝑜𝑝𝑒𝑛

𝜆∈𝐼
. 𝑇𝑕.𝑏𝑒𝑓𝑜𝑟𝑒 ⇒
𝐵𝑒𝑐𝑎𝑢𝑠𝑒 𝑀𝑜𝑟𝑔𝑎𝑛𝑠: 𝜆∈𝐼 𝑈𝜆 𝐶 = 𝜆∈𝐼 𝑈𝜆 𝐶

⇒ 𝐶 is open ⇒
𝜆∈𝐼 𝑈𝜆 𝜆∈𝐼 𝑈𝜆 is closed


Theorem: Given a metric space 𝐸, 𝑑 , 𝑎𝑛𝑦 𝑓𝑖𝑛𝑖𝑡𝑒 𝑠𝑒𝑡 𝑖𝑠 𝑐𝑙𝑜𝑠𝑒𝑑.

Proof: Let A be a finite set.

Case 1: 𝐴 = ∅
Proved directly from theorem before.

Case 2: A≠ ∅
A is finite, so 𝐴 = 𝑎1 , … , 𝑎𝑛

Pick up any 𝑝 ∈ 𝐴𝐶 and define 𝑟 = min 𝑑 𝑝, 𝑎𝑖 / 𝑖 = 1, … , 𝑛

𝑝 ∈ 𝐴𝐶 ⇒ 𝑝 ∉ 𝐴 ⇒ 𝑝 ≠ 𝑎𝑖 , ∀𝑖 = 1, … , 𝑛 ⇒ 𝑑 𝑝, 𝑎𝑖 > 0, ∀𝑖 = 1, … , 𝑛
⇒ 𝑟 > 0 (1)
𝑟 = min 𝑑 𝑝, 𝑎𝑖 / 𝑖 = 1, … , 𝑛 ⇒ 𝑟 ∈ 𝑑 𝑝, 𝑎𝑖 / 𝑖 = 1, … , 𝑛

Assume otherwise that 𝐵 𝑝, 𝑟 ⊄ 𝐴𝐶 ⇒ ∃𝑥 ∈ 𝐵 𝑝, 𝑟 / 𝑥 ∉ 𝐴𝐶 ⇒ ∃𝑥 ∈ 𝐵 𝑝, 𝑟 / 𝑥 ∈ 𝐴 ⇒

⇒ ∃𝑗 = 1, … , 𝑛 / 𝑎𝑗 ∈ 𝐵 𝑝, 𝑟 ⇒ ∃𝑗 = 1, … , 𝑛 / 𝑑(𝑝, 𝑎𝑗 ) < 𝑟 = min 𝑑 𝑝, 𝑎𝑖 / 𝑖 = 1, … , 𝑛
(absurd) ∴ 𝐵 𝑝, 𝑟 ⊂ 𝐴𝐶 (2)

From 1 𝑎𝑛𝑑 2 : ∀𝑝 ∈ 𝐴𝐶 , ∃ 𝐵 𝑝, 𝑟 ⊂ 𝐴𝐶 ⇒ 𝐴𝐶 𝑖𝑠 𝑜𝑝𝑒𝑛 ⇒ 𝐴 𝑖𝑠 𝑐𝑙𝑜𝑠𝑒𝑑. 

Theorem: 𝐸, 𝑑 𝑚. 𝑒. 𝐴 ⊂ ℝ 𝑖𝑠 𝑏𝑑𝑑 𝑓𝑟𝑜𝑚 𝑎𝑏𝑜𝑣𝑒 𝑎𝑛𝑑 𝐴 𝑖𝑠 𝑐𝑙𝑜𝑠𝑒𝑑 ⇒ ∃max⁡
(𝐴)

Proof:

A⊂ ℝ is bounded from above, so by completeness ∃sup⁡

(𝐴) ∈ ℝ.

𝐴𝑠𝑠𝑢𝑚𝑒 𝑜𝑡𝑕𝑒𝑟𝑤𝑖𝑠𝑒 𝑡𝑕𝑎𝑡 sup 𝐴 ∉ 𝐴 ⇒ sup⁡ (𝐴) ∈ 𝐴𝐶

𝐶 ⇒ ∃𝐵 sup 𝐴 , 𝑟 ⊂ 𝐴𝐶 ⇒
𝐴 𝑖𝑠 𝑐𝑙𝑜𝑠𝑒𝑑 ⇒ 𝐴 𝑖𝑠 𝑜𝑝𝑒𝑛
⇒ ∃𝑟 ∈ ℝ+/𝐵 sup 𝐴 , 𝑟 ⊂ 𝐴𝐶
𝑟 𝑥 ∈ 𝐵 sup 𝐴 , 𝑟 ⇒ ∃𝑥 ∈ 𝐴𝐶 /𝑥 < sup 𝐴 (absurd)
𝑇𝑎𝑘𝑒 𝑥 = sup 𝐴 − ⇒
2 𝑥 < sup 𝐴

∴ sup 𝐴 ∈ 𝐴 ⇒ sup 𝐴 = max⁡

(𝐴)


Theorem (Quiz 6): 𝐸, 𝑑 𝑚. 𝑒. 𝐴 ⊂ ℝ 𝑖𝑠 𝑏𝑑𝑑 𝑓𝑟𝑜𝑚 𝑏𝑒𝑙𝑜𝑤 𝑎𝑛𝑑 𝐴 𝑖𝑠 𝑐𝑙𝑜𝑠𝑒𝑑 ⇒ ∃min⁡

(𝐴)

Proof:

A⊂ ℝ is bounded from below, so by corollary completeness ∃inf⁡

(𝐴) ∈ ℝ.

𝐴𝑠𝑠𝑢𝑚𝑒 𝑜𝑡𝑕𝑒𝑟𝑤𝑖𝑠𝑒 𝑡𝑕𝑎𝑡 inf 𝐴 ∉ 𝐴 ⇒ inf⁡(𝐴) ∈ 𝐴𝐶

𝐶 ⇒ ∃𝐵 inf 𝐴 , 𝑟 ⊂ 𝐴𝐶 ⇒
𝐴 𝑖𝑠 𝑐𝑙𝑜𝑠𝑒𝑑 ⇒ 𝐴 𝑖𝑠 𝑜𝑝𝑒𝑛
⇒ ∃𝑟 ∈ ℝ+/𝐵 inf 𝐴 , 𝑟 ⊂ 𝐴𝐶
𝑟 𝑥 ∈ 𝐵 inf 𝐴 , 𝑟 ⇒ ∃𝑥 ∈ 𝐴𝐶 /𝑥 > inf 𝐴 (absurd)
𝑇𝑎𝑘𝑒 𝑥 = inf 𝐴 + ⇒
2 𝑥 > inf 𝐴

∴ inf 𝐴 ∈ 𝐴 ⇒ inf 𝐴 = min⁡

(𝐴)


5.1 BALLS IN SOME METRIC SPACES

𝟏 𝑫𝒊𝒎𝒆𝒏𝒔𝒊𝒐𝒏𝒂𝒍 𝑬𝒖𝒄𝒍𝒊𝒅𝒆𝒂𝒏 𝒔𝒑𝒂𝒄𝒆 (𝑬 = 𝑬𝟏)

𝐵 𝑝, 𝑟 = (𝑝 − 𝑟, 𝑝 + 𝑟)
and 𝐵 𝑝, 𝑟 = 𝑝 − 𝑟, 𝑝 − 𝑟

Proof:

𝑥 ∈ 𝐵 𝑝, 𝑟 ⇔ 𝑑 𝑥, 𝑝 < 𝑟 ⇔ 𝑥 − 𝑝 < 𝑟 ⇔
⇔ −𝑟 < 𝑥 − 𝑝 < 𝑟 ⇔ 𝑝 − 𝑟 < 𝑥 < 𝑝 + 𝑟 ⇔
⇔ 𝑥 ∈ (𝑝 − 𝑟, 𝑝 + 𝑟)

𝑥 ∈ 𝐵 𝑝, 𝑟 ⇔ 𝑑 𝑥, 𝑝 ≤ 𝑟 ⇔ 𝑥 − 𝑝 ≤ 𝑟 ⇔
⇔ −𝑟 ≤ 𝑥 − 𝑝 ≤ 𝑟 ⇔ 𝑝 − 𝑟 ≤ 𝑥 ≤ 𝑝 + 𝑟 ⇔
⇔ 𝑥 ∈ 𝑝 − 𝑟, 𝑝 + 𝑟
𝟐 𝑫𝒊𝒎𝒆𝒏𝒔𝒊𝒐𝒏𝒂𝒍 𝑬𝒖𝒄𝒍𝒊𝒅𝒆𝒂𝒏 𝒔𝒑𝒂𝒄𝒆 (𝑬 = 𝑬𝟐 )

𝐵 𝑝, 𝑟 = 𝑑𝑖𝑠𝑐 𝑐𝑒𝑛𝑡𝑒𝑟𝑒𝑑 𝑎𝑡 𝑝 𝑤𝑖𝑡𝑕 𝑟𝑎𝑑𝑖𝑢𝑠 𝑟 and

𝐵 𝑝, 𝑟 = 𝑑𝑖𝑠𝑐 𝑐𝑒𝑛𝑡𝑒𝑟𝑒𝑑 𝑎𝑡 𝑝 𝑤𝑖𝑡𝑕 𝑟𝑎𝑑𝑖𝑢𝑠 𝑟 𝑎𝑛𝑑 𝑖𝑡𝑠 𝑏𝑜𝑟𝑑𝑒𝑟.

Proof:

𝑥 ∈ 𝐵 𝑝, 𝑟 ⇔ 𝑑 𝑥, 𝑝 < 𝑟 ⇔
2 2
⇔ 𝑥 − 𝑥𝑝 + 𝑦 − 𝑦𝑝 <𝑟⇔
2 2
⇔ 𝑥 − 𝑥𝑝 + 𝑦 − 𝑦𝑝 < 𝑟2

2 2 2 2
𝑥 ∈ 𝐵 𝑝, 𝑟 ⇔ 𝑑 𝑥, 𝑝 ≤ 𝑟 ⇔ 𝑥 − 𝑥𝑝 + 𝑦 − 𝑦𝑝 ≤ 𝑟 ⇔ 𝑥 − 𝑥𝑝 + 𝑦 − 𝑦𝑝 ≤ 𝑟2

𝟑 𝑫𝒊𝒎𝒆𝒏𝒔𝒊𝒐𝒏𝒂𝒍 𝑬𝒖𝒄𝒍𝒊𝒅𝒆𝒂𝒏 𝒔𝒑𝒂𝒄𝒆 (𝑬 = 𝑬𝟑 ) (HW)

𝐵 𝑝, 𝑟 = 𝑒𝑠𝑓𝑒𝑟𝑒 𝑐𝑒𝑛𝑡𝑒𝑟𝑒𝑑 𝑎𝑡 𝑝 𝑤𝑖𝑡𝑕 𝑟𝑎𝑑𝑖𝑢𝑠 𝑟 and

𝐵 𝑝, 𝑟 = 𝑒𝑠𝑓𝑒𝑟𝑒 𝑐𝑒𝑛𝑡𝑒𝑟𝑒𝑑 𝑎𝑡 𝑝 𝑤𝑖𝑡𝑕 𝑟𝑎𝑑𝑖𝑢𝑠 𝑟 𝑎𝑛𝑑 𝑖𝑡𝑠 𝑏𝑜𝑟𝑑𝑒𝑟.

Proof:

2 2 2
𝑥 ∈ 𝐵 𝑝, 𝑟 ⇔ 𝑑 𝑥, 𝑝 < 𝑟 ⇔ 𝑥 − 𝑥𝑝 + 𝑦 − 𝑦𝑝 + 𝑧 − 𝑧𝑝 <𝑟⇔
2 2 2
⇔ 𝑥 − 𝑥𝑝 + 𝑦 − 𝑦𝑝 + 𝑧 − 𝑧𝑝 < 𝑟2

2 2 2
𝑥 ∈ 𝐵 𝑝, 𝑟 ⇔ 𝑑 𝑥, 𝑝 ≤ 𝑟 ⇔ 𝑥 − 𝑥𝑝 + 𝑦 − 𝑦𝑝 + 𝑧 − 𝑧𝑝 ≤𝑟⇔
2 2 2
⇔ 𝑥 − 𝑥𝑝 + 𝑦 − 𝑦𝑝 + 𝑧 − 𝑧𝑝 ≤ 𝑟2

𝑫𝒊𝒔𝒄𝒓𝒆𝒕𝒆 𝒎𝒆𝒕𝒓𝒊𝒄 𝒔𝒑𝒂𝒄𝒆𝒔 (HW)

1, 𝑖𝑓 𝑥 ≠ 𝑦
Let E be any subset and 𝑑: 𝑑 𝑥, 𝑦 =
0, 𝑖𝑓 𝑥 = 𝑦

𝑝 , 𝑖𝑓 𝑟 ≤ 1 𝑝 , 𝑖𝑓 𝑟 < 1
𝐵 𝑝, 𝑟 = and 𝐵 𝑝, 𝑟 =
𝐸, 𝑖𝑓 𝑟 > 1 𝐸, 𝑖𝑓 𝑟 ≥ 1

Proof: Directly by definition of d.

Observation (Quiz 4): 𝐸, 𝑑 𝑤𝑖𝑡𝑕 𝑑 𝑡𝑕𝑒 𝑑𝑖𝑠𝑐𝑟𝑒𝑡𝑒 𝑚𝑒𝑡𝑟𝑖𝑐. 𝑇𝑕𝑒𝑛, 𝑎𝑛𝑦 𝑠𝑢𝑏𝑠𝑒𝑡 𝑜𝑓 𝐸 𝑖𝑠 𝑜𝑝𝑒𝑛.
6. SEQUENCES
6.1 CONVERGENCY

𝐂𝐎𝐍𝐕𝐄𝐑𝐆𝐄𝐍𝐂𝐘 𝐎𝐅 𝐀 𝐒𝐄𝐐𝐔𝐄𝐍𝐂𝐄: Given a metric space 𝐸, 𝑑 , 𝑎𝑛𝑑 𝑝𝑛 / 𝑛 ∈ ℕ∗ 𝑜𝑟 𝑝𝑛 a

sequence of elements of E. We say that 𝑝𝑛 converges to 𝑝 ∈ 𝐸 lim𝑛→+∞ 𝑝𝑛 = 𝑝 iff ∀𝜀 >
0, ∃𝑁 ∈ ℕ / ∀𝑛 ≥ 𝑁, 𝑑(𝑝, 𝑝𝑛 ) < 𝜀.

Theorem: 𝐸, 𝑑 𝑚. 𝑒. 𝑝𝑛 𝑠𝑒𝑞𝑢𝑒𝑛𝑐𝑒 𝑖𝑛 𝐸 / 𝑝𝑙 = 𝑝𝑘 , ∀𝑙, 𝑘 ≥ 𝑀 ∈ ℕ ⇒ 𝑝𝑛 𝑐𝑜𝑛𝑣𝑒𝑟𝑔𝑒𝑠

Proof:
𝐶𝑎𝑙𝑙 𝑝 = 𝑝𝑘 , 𝑡𝑜 𝑠𝑜𝑚𝑒 𝑘 ≥ 𝑀
⇒ 𝑝𝑛 = 𝑝, ∀𝑛 ≥ 𝑀 ⇒ 𝑑 𝑝𝑛 , 𝑝 = 0, ∀𝑛 ≥ 𝑀
𝐵𝑦 𝑕𝑦𝑝𝑜𝑡𝑕𝑒𝑠𝑖𝑠: 𝑝𝑙 = 𝑝𝑘 , ∀𝑙, 𝑘 ≥ 𝑀

∴ ∀𝜀 > 0, ∃𝑀 ∈ ℕ / ∀𝑛 ≥ 𝑀, 𝑑(𝑝, 𝑝𝑛 ) = 0 < 𝜀 ⇒ converges to 𝑝 

Theorem: (A PROPERTY WITH THE DISCRETE METRIC)

𝐸, 𝑑 𝑤𝑖𝑡𝑕 𝑑 𝑡𝑕𝑒 𝑑𝑖𝑠𝑐𝑟𝑒𝑡𝑒 𝑚𝑒𝑡𝑟𝑖𝑐. 𝑝𝑛 𝑐𝑜𝑛𝑣𝑒𝑟𝑔𝑒𝑠 𝑖𝑛 𝐸 ⇒ ∃𝑀 ∈ ℕ/ 𝑝𝑙 = 𝑝𝑘 , ∀𝑙, 𝑘 ≥ 𝑀

Proof:
𝐵𝑦 𝑕𝑦𝑝𝑜𝑡𝑕𝑒𝑠𝑖𝑠: 𝑝𝑛 𝑐𝑜𝑛𝑣𝑒𝑟𝑔𝑒𝑠 𝑖𝑛 𝐸. 𝐶𝑎𝑙𝑙 lim 𝑝𝑛 = 𝑝 ⇒
1
⇒ ∀𝜀 > 0, ∃𝑀 ∈ ℕ/∀𝑛 ≥ 𝑀, 𝑑 𝑝, 𝑝𝑛 < 𝜀 1 ∃𝑀 ∈ ℕ/∀𝑛 ≥ 𝑀, 𝑑 𝑝, 𝑝𝑛 < ⇒
𝜀= 2
2
⇒ ∃𝑀 ∈ ℕ/∀𝑛 ≥ 𝑀, 𝑑 𝑝, 𝑝𝑛 = 0 ⇒ ∃𝑀 ∈ ℕ/∀𝑛 ≥ 𝑀, 𝑝 = 𝑝𝑛 ⇒
⇒ ∃𝑀 ∈ ℕ/∀𝑛, 𝑘 ≥ 𝑀, 𝑝𝑘 = 𝑝𝑛 

Theorem: (UNICITY OF THE LIMIT)

𝐸, 𝑑 𝑚. 𝑒. 𝑝𝑛 𝑐𝑜𝑛𝑣𝑒𝑟𝑔𝑒𝑠 𝑖𝑛 𝐸 𝑡𝑜 𝑝 ⇒ 𝑝 𝑖𝑠 𝑢𝑛𝑖𝑞𝑢𝑒

Proof: 𝐵𝑦 𝑕𝑦𝑝𝑜𝑡𝑕𝑒𝑠𝑖𝑠: 𝑝𝑛 𝑐𝑜𝑛𝑣𝑒𝑟𝑔𝑒𝑠 𝑖𝑛 𝐸. 𝐶𝑎𝑙𝑙 lim 𝑝𝑛 = 𝑝 𝑎𝑛𝑑 lim 𝑝𝑛 = 𝑚

𝜀
𝑃𝑖𝑐𝑘 𝑢𝑝 𝑎𝑛𝑦 𝜀 > 0 ⇒ >0
2
𝜀
lim 𝑝𝑛 = 𝑝 ⇒ ∃𝑁 ∈ ℕ/∀𝑛 ≥ 𝑁, 𝑑 𝑝, 𝑝𝑛 <
2
𝜀
lim 𝑝𝑛 = 𝑚 ⇒ ∃𝑀 ∈ ℕ/∀𝑛 ≥ 𝑀, 𝑑 𝑚, 𝑝𝑛 < ⇒
2
𝑇𝑎𝑘𝑒 𝑅 = 𝑚𝑎𝑥 𝑁, 𝑀
𝜀 𝜀
⇒ ∃𝑅 ∈ ℕ/∀𝑛 ≥ 𝑅, 𝑑 𝑝, 𝑝𝑛 + 𝑑 𝑚, 𝑝𝑛 < + = 𝜀
2 2 ⇒ 𝑑 𝑝, 𝑚 < 𝜀
𝐵𝑦 𝑡𝑟𝑖𝑎𝑛𝑔𝑢𝑙𝑎𝑟 𝐼𝑛𝑞𝑢𝑎𝑙𝑖𝑡𝑦: 𝑑 𝑝, 𝑚 ≤ 𝑑 𝑝, 𝑝𝑛 + 𝑑 𝑚, 𝑝𝑛

∴ ∀𝜀 > 0, 𝑑 𝑝, 𝑚 < 𝜀 ⇒ 𝑑 𝑝, 𝑚 = 0 ⇒ 𝑚 = 𝑝 ∴ 𝑡𝑕𝑒 𝑙𝑖𝑚𝑖𝑡 𝑖𝑠 𝑢𝑛𝑖𝑞𝑢𝑒.


𝐁𝐎𝐔𝐍𝐃𝐄𝐃 𝐒𝐄𝐓: Given 𝐸, 𝑑 𝑚𝑒𝑡𝑟𝑖𝑐 𝑠𝑝𝑎𝑐𝑒. 𝐴 ⊂ 𝐸 𝑖𝑠 𝑏𝑜𝑢𝑛𝑑𝑒𝑑 𝑖𝑓𝑓 ∃𝐵(𝑂, 𝑟)/𝐴 ⊂ 𝐵(𝑂, 𝑟)

Theorem: 𝐸, 𝑑 𝑚. 𝑒. 𝑝𝑛 𝑐𝑜𝑛𝑣𝑒𝑟𝑔𝑒𝑠 𝑖𝑛 𝐸 ⇒ 𝑡𝑕𝑒 𝑠𝑒𝑡 𝑝𝑛 ; 𝑛 ∈ ℕ 𝑖𝑠 𝑏𝑑𝑑.

Proof: 𝐵𝑦 𝑕𝑦𝑝𝑜𝑡𝑕𝑒𝑠𝑖𝑠: 𝑝𝑛 𝑐𝑜𝑛𝑣𝑒𝑟𝑔𝑒𝑠 𝑖𝑛 𝐸. 𝐶𝑎𝑙𝑙 lim 𝑝𝑛 = 𝑝 ⇒

⇒ ∀𝜀 > 0, ∃𝑁 ∈ ℕ/∀𝑛 ≥ 𝑁, 𝑑 𝑝, 𝑝𝑛 < 𝜀. 𝑇𝑎𝑘𝑒 𝜀0 > 0 𝑎 𝑓𝑖𝑥𝑒𝑑 𝑛𝑢𝑚𝑏𝑒𝑟 ⇒
⇒ ∃𝑁 ∈ ℕ/∀𝑛 ≥ 𝑁, 𝑑 𝑝, 𝑝𝑛 < 𝜀0 (i)

Consider the set 𝐷 = 𝑑 𝑝, 𝑝𝑛 , 𝑤𝑖𝑡𝑕 𝑛 < 𝑁 . Since D is a finite set of real numbers, D has
maximum. Call 𝛼 = max 𝐷 ⇒ ∀𝑥 ∈ 𝐷, 𝑥 ≤ 𝛼 ⇒ ∀𝑛 < 𝑁, 𝑑 𝑝, 𝑝𝑛 ≤ 𝛼 (ii)

𝐵𝑦 𝑖 : ∀𝑛 ≥ 𝑁, 𝑑 𝑝, 𝑝𝑛 < 𝜀0 ≤ 𝛽 < 𝛽 + 1
Take 𝛽 = 𝑚𝑎𝑥 𝛼, 𝜀0 ⇒ ⇒
𝐵𝑦 𝑖𝑖 : ∀𝑛 < 𝑁, 𝑑 𝑝, 𝑝𝑛 ≤ 𝛼 ≤ 𝛽 < 𝛽 + 1

⇒ ∀𝑛 ∈ ℕ, 𝑑 𝑝, 𝑝𝑛 < 𝛽 + 1 ⇒ ∃𝐵 𝑝, 𝛽 + 1 𝑠𝑢𝑐𝑕 𝑡𝑕𝑎𝑡 ∀𝑛 ∈ ℕ, 𝑝𝑛 ∈ 𝐵 𝑝, 𝛽 + 1 ⇒

⇒ ∃𝐵 𝑝, 𝛽 + 1 𝑠𝑢𝑐𝑕 𝑡𝑕𝑎𝑡 𝑝𝑛 ; 𝑛 ∈ ℕ ⊂ 𝐵 𝑝, 𝛽 + 1 ⇒ 𝑝𝑛 ; 𝑛 ∈ ℕ is bdd.


Theorem: (ARITMETIC PROPERTIES OF CONVERGENT SEQUENCES in ℝ)

𝐸, 𝑑 = 𝐸1 . 𝑎𝑛 𝑎𝑛𝑑 𝑏𝑛 𝑎𝑟𝑒 𝑐𝑜𝑛𝑣𝑒𝑟𝑔𝑒𝑛𝑡 𝑖𝑛 𝐸 𝑡𝑜 𝐴 𝑎𝑛𝑑 𝐵, 𝑟𝑒𝑠𝑝𝑒𝑐𝑡𝑖𝑣𝑒𝑙𝑦. 𝑐 ∈ ℝ. 𝑇𝑕𝑒𝑛:

1. lim 𝑐 . 𝑎𝑛 = 𝑐. 𝐴
2. lim 𝑎𝑛 + 𝑏𝑛 = 𝐴 + 𝐵
3. lim 𝑎𝑛 − 𝑏𝑛 = 𝐴 − 𝐵
4. lim 𝑎𝑛 . 𝑏𝑛 = 𝐴. 𝐵
1 1
5. If 𝐵 ≠ 0: lim =
𝑏𝑛 𝐵
𝑎𝑛 𝐴
6. If 𝐵 ≠ 0: lim =
𝑏𝑛 𝐵

Proof:

1. Analysis: lim 𝑐. 𝑎𝑛 = 𝑐. 𝐴 ⇔ ∀𝜀 > 0, ∃𝑁 ∈ ℕ/∀𝑛 ≥ 𝑁, 𝑐. 𝑎𝑛 − 𝑐. 𝐴 < 𝜀

𝑐. 𝑎𝑛 − 𝑐. 𝐴 = 𝑐 . 𝑎𝑛 − 𝐴

Proof: Case 1: 𝑐 = 0
𝑐 = 0 ⇒ c. 𝑎𝑛 = 0, ∀𝑛 ∈ ℕ ⇒ ∀𝜀 > 0, ∃𝑁 ∈ ℕ/∀𝑛 ≥ 𝑁, 𝑑 𝑐. 𝑎𝑛 , 0 < 𝜀 ⇒
⇒ ∀𝜀 > 0, ∃𝑁 ∈ ℕ/∀𝑛 ≥ 𝑁, 𝑑 𝑐. 𝑎𝑛 , 0. 𝐴 < 𝜀 ⇒ ∀𝜀 > 0, ∃𝑁 ∈ ℕ/∀𝑛 ≥ 𝑁, 𝑑 𝑐. 𝑎𝑛 , 𝑐. 𝐴 < 𝜀 ⇒
⇒ lim 𝑐 . 𝑎𝑛 = 𝑐. 𝐴

Case 2: 𝑐 ≠ 0
𝜀
𝑃𝑖𝑐𝑘 𝑢𝑝 𝑎𝑛𝑦 𝜀 > 0 ⇒ >0 𝜀
𝑐 ⇒ ∃𝑁 ∈ ℕ/∀𝑛 ≥ 𝑁, 𝑎𝑛 − 𝐴 < ⇒
lim 𝑎𝑛 = 𝐴 𝑐
𝜀
⇒ ∃𝑁 ∈ ℕ/∀𝑛 ≥ 𝑁, 𝑐 . 𝑎𝑛 − 𝐴 < . 𝑐 =𝜀
𝑐
∴ ∀𝜀 > 0, ∃𝑁 ∈ ℕ/∀𝑛 ≥ 𝑁, 𝑐. 𝑎𝑛 − 𝑐. 𝐴 < 𝜀 ⇒ lim 𝑐 . 𝑎𝑛 = 𝑐. 𝐴
 2. Analysis: lim 𝑎𝑛 + 𝑏𝑛 = 𝐴 + 𝐵 ⇔ ∀𝜀 > 0, ∃𝑁 ∈ ℕ/∀𝑛 ≥ 𝑁, 𝑎𝑛 + 𝑏𝑛 − (𝐴 + 𝐵) < 𝜀

𝑎𝑛 + 𝑏𝑛 − (𝐴 + 𝐵) = (𝑎𝑛 − 𝐴) + (𝑏𝑛 − 𝐵) ≤ 𝑎𝑛 − 𝐴 + 𝑏𝑛 − 𝐵

Proof:
𝜀
lim 𝑎𝑛 = 𝐴 ⇒ ∃𝑁 ∈ ℕ/∀𝑛 ≥ 𝑁, 𝑎𝑛 − 𝐴 <
2
𝜀 ⇒
lim 𝑏𝑛 = 𝐵 ⇒ ∃𝑀 ∈ ℕ/∀𝑛 ≥ 𝑀, 𝑏𝑛 − 𝐵 <
2
𝑇𝑎𝑘𝑒 𝑅 = 𝑚𝑎𝑥 𝑁, 𝑀
𝜀 𝜀
⇒ ∃𝑅 ∈ ℕ/∀𝑛 ≥ 𝑁, 𝑎𝑛 − 𝐴 + 𝑏𝑛 − 𝐵 < + = 𝜀
2 2
𝐵𝑦 𝑇𝑟𝑖𝑎𝑛𝑔𝑢𝑙𝑎𝑟 𝐼𝑛𝑒𝑞𝑢𝑎𝑙𝑖𝑡𝑦: 𝑎𝑛 + 𝑏𝑛 − (𝐴 + 𝐵) = (𝑎𝑛 − 𝐴) + (𝑏𝑛 − 𝐵) ≤ ⇒
≤ 𝑎𝑛 − 𝐴 + 𝑏𝑛 − 𝐵
⇒ ∃𝑅 ∈ ℕ/∀𝑛 ≥ 𝑁, 𝑎𝑛 + 𝑏𝑛 − 𝐴 + 𝐵 < 𝜀
∴ lim 𝑎𝑛 + 𝑏𝑛 = 𝐴 + 𝐵

3. Is a direct consequence of 1 and 2.

4. Analysis: lim 𝑎𝑛 . 𝑏𝑛 = 𝐴. 𝐵 ⇔ ∀𝜀 > 0, ∃𝑁 ∈ ℕ/∀𝑛 ≥ 𝑁, 𝑎𝑛 . 𝑏𝑛 − 𝐴. 𝐵 < 𝜀

𝑎𝑛 . 𝑏𝑛 − 𝐴. 𝐵 = 𝑎𝑛 . 𝑏𝑛 − 𝑎𝑛 . 𝐵 + 𝑎𝑛 . 𝐵 − 𝐴. 𝐵 = 𝑎𝑛 . 𝑏𝑛 − 𝐵 − 𝐵. (𝑎𝑛 − 𝐴) ≤⊿−𝐼𝑛𝑒𝑞 .
≤ 𝑎𝑛 . 𝑏𝑛 − 𝐵 + 𝐵. (𝑎𝑛 − 𝐴) = 𝑎𝑛 . 𝑏𝑛 − 𝐵 + 𝐵 . 𝑎𝑛 − 𝐴

Proof: 𝑃𝑖𝑐𝑘 𝑢𝑝 𝑎𝑛𝑦 𝜀 > 0

𝐵𝑦 𝑕𝑦𝑝𝑜𝑡𝑕𝑒𝑠𝑖𝑠:
𝑎𝑛 𝑖𝑠 𝑐𝑜𝑛𝑣𝑒𝑟𝑔𝑒𝑛𝑡 𝑎𝑛 /𝑛 ∈ ℕ 𝑖𝑠 𝑏𝑑𝑑. ⇒ ∃𝑘 ∈ ℝ+/∀𝑛 ∈ ℕ, 𝑎𝑛 < 𝑘
𝐵𝑦 𝑇𝑕.
𝑏𝑒𝑓𝑜𝑟𝑒
𝜀 ⇒
𝜀>0 ⇒ >0 𝜀
2𝑘 ⇒ ∃𝐻 ∈ ℕ/∀𝑛 ≥ 𝐻, 𝑏𝑛 − 𝐵 <
lim 𝑏𝑛 = 𝐵 2𝑘

ℕ 𝜀 𝜀
⇒ ∃𝐻 ∈ ≥ 𝐻, 𝑎𝑛 . 𝑏𝑛 − 𝐵 < .𝑘 = (i)
∀𝑛 2𝑘 2

Case 1: 𝐵 = 0

𝜀
𝐵𝑒𝑐𝑎𝑢𝑠𝑒 i : ∃𝐻 ∈ ℕ/∀𝑛 ≥ 𝐻, 𝑎𝑛 . 𝑏𝑛 − 𝐵 <
2 ⇒
𝐵𝑦 𝑎𝑛𝑎𝑙𝑦𝑠𝑖𝑠: 𝑎𝑛 . 𝑏𝑛 − 𝐴. 𝐵 ≤ 𝑎𝑛 . 𝑏𝑛 − 𝐵 + 𝐵 . 𝑎𝑛 − 𝐴 = 𝑎𝑛 . 𝑏𝑛 − 𝐵
=0
𝜀
⇒ ∃𝐻 ∈ ℕ/∀𝑛 ≥ 𝐻, 𝑎𝑛 . 𝑏𝑛 − 𝐴. 𝐵 < <𝜀
2
𝜀
∴ ∀𝜀 > 0, ∃𝐻 ∈ ℕ/∀𝑛 ≥ 𝐻, 𝑎𝑛 . 𝑏𝑛 − 𝐴. 𝐵 < < 𝜀 ⇒ lim 𝑎𝑛 . 𝑏𝑛 = 𝐴. 𝐵
2
Case 2: 𝐵 ≠ 0

𝜀
𝜀 > 0 𝑎𝑛𝑑 𝐵 ≠ 0 ⇒ >0 𝜀
2𝐵 ⇒ ∃𝑅 ∈ ℕ/∀𝑛 ≥ 𝑅, 𝑎𝑛 − 𝐴 < ⇒
lim 𝑎𝑛 = 𝐴 2𝐵
𝜀 𝜀
⇒ ∃𝑅 ∈ ℕ/∀𝑛 ≥ 𝑅, 𝐵 . 𝑎𝑛 − 𝐴 < . 𝐵 =
2𝐵 2
𝜀 𝜀 ⇒
𝐵𝑒𝑐𝑎𝑢𝑠𝑒 𝑖 : ∃𝐻 ∈ ℕ/∀𝑛 ≥ 𝐻, 𝑎𝑛 . 𝑏𝑛 − 𝐵 < .𝑘 =
2𝑘 2
𝑇𝑎𝑘𝑒 𝑁 = 𝑚𝑎𝑥 𝐻, 𝑅
𝜀 𝜀
⇒ ∃𝑁 ∈ ℕ/∀𝑛 ≥ 𝑁, 𝑎𝑛 . 𝑏𝑛 − 𝐵 + 𝐵 . 𝑎𝑛 − 𝐴 < + =𝜀
2 2 ⇒
𝐵𝑦 𝑎𝑛𝑎𝑙𝑦𝑠𝑖𝑠: 𝑎𝑛 . 𝑏𝑛 − 𝐴. 𝐵 ≤ 𝑎𝑛 . 𝑏𝑛 − 𝐵 + 𝐵 . 𝑎𝑛 − 𝐴

⇒ ∃𝑁 ∈ ℕ/∀𝑛 ≥ 𝑁, 𝑎𝑛 . 𝑏𝑛 − 𝐴. 𝐵 < 𝜀
∴ ∀𝜀 > 0, ∃𝑁 ∈ ℕ/∀𝑛 ≥ 𝑁, 𝑎𝑛 . 𝑏𝑛 − 𝐴. 𝐵 < 𝜀 ⇒ lim 𝑎𝑛 . 𝑏𝑛 = 𝐴. 𝐵

1 1 1 1
5. Analysis: lim = ⇔ ∀𝜀 > 0, ∃𝑁 ∈ ℕ/∀𝑛 ≥ 𝑁, − <𝜀
𝑏𝑛 𝐵 𝑏𝑛 𝐵

1 1 𝐵 − 𝑏𝑛 𝐵 − 𝑏𝑛 1
− = = = 𝑏𝑛 − 𝐵 .
𝑏𝑛 𝐵 𝑏𝑛 . 𝐵 𝐵. 𝑏𝑛 𝐵. 𝑏𝑛

Proof: 𝑃𝑖𝑐𝑘 𝑢𝑝 𝑎𝑛𝑦 𝜀 > 0

𝐵2
lim 𝑏𝑛 = 𝐵 lim 𝐵 . 𝑏𝑛 = 𝐵 2 ∃𝑁 ∈ ℕ/∀𝑛 ≥ 𝑁, 𝐵. 𝑏𝑛 − 𝐵 2 < ⇒
𝑃𝑟𝑜𝑝 .1 𝐵2 2
𝜀= >0 𝑖𝑛 𝑡𝑕𝑒
2
𝑑𝑒𝑓 .𝑜𝑓 𝑙𝑖𝑚𝑖𝑡
𝐵2 𝐵2
⇒ ∃𝑁 ∈ ℕ/∀𝑛 ≥ 𝑁, 𝐵 2 − < 𝐵. 𝑏𝑛 ⇒ ∃𝑁 ∈ ℕ/∀𝑛 ≥ 𝑁, . .0 < < 𝐵. 𝑏𝑛 ⇒
2 2
1 2 1 2
⇒ ∃𝑁 ∈ ℕ/∀𝑛 ≥ 𝑁, . .0 < < ⇒ ∃𝑁 ∈ ℕ/∀𝑛 ≥ 𝑁, . . < 2
𝐵. 𝑏𝑛 𝐵 2 𝐵. 𝑏𝑛 𝐵
lim 𝑏𝑛 = 𝐵
𝐵2 . 𝜀 ⇒
𝐵2 . 𝜀 ⇒ ∃𝐻 ∈ ℕ/∀𝑛 ≥ 𝐻, 𝑏𝑛 − 𝐵 <
𝜀 > 0 𝑎𝑛𝑑 𝐵 ≠ 0 ⇒ >0 2
2
𝑇𝑎𝑘𝑒 𝑅 = 𝑚𝑎𝑥 𝐻, 𝑁

1 𝐵 2 .𝜀 2
⇒ ∃𝑅 ∈ ℕ/∀𝑛 ≥ 𝑅, 𝑏𝑛 − 𝐵 . < . =𝜀
𝐵.𝑏 𝑛 2 𝐵2 1 1
1 1 1
⇒ ∃𝑅 ∈ ℕ/∀𝑛 ≥ 𝑅, − <𝜀
𝑏𝑛 𝐵
− = 𝑏𝑛 − 𝐵 .
𝑏𝑛 𝐵 𝐵.𝑏 𝑛

1 1 1 1
∴ ∀𝜀 > 0, ∃𝑅 ∈ ℕ/∀𝑛 ≥ 𝑅, − < 𝜀 ⇒ lim =
𝑏𝑛 𝐵 𝑏𝑛 𝐵

6. Is a direct consequence of 4 and 5.


6.2 MONOTONIC SEQUENCES

𝐈𝐍𝐂𝐑𝐄𝐀𝐒𝐈𝐍𝐆 𝐀𝐍𝐃 𝐃𝐄𝐂𝐑𝐄𝐀𝐒𝐈𝐍𝐆 𝐒𝐄𝐐𝐔𝐄𝐍𝐂𝐄𝐒: Given 𝑝𝑛 a sequence of real numbers. We

say that 𝑝𝑛 is:
INCREASING 𝑁𝑜𝑡𝑎𝑡𝑖𝑜𝑛: ↑ iff 𝑝𝑛 ≤ 𝑝𝑛+1 , ∀𝑛 ∈ ℕ
DECREASING 𝑁𝑜𝑡𝑎𝑡𝑖𝑜𝑛: ↓ iff 𝑝𝑛 ≥ 𝑝𝑛+1 , ∀𝑛 ∈ ℕ

𝐌𝐎𝐍𝐎𝐓𝐎𝐍𝐈𝐂 𝐒𝐄𝐐𝐔𝐄𝐍𝐂𝐄𝐒: Given 𝑝𝑛 a sequence of real numbers. We say that 𝑝𝑛 is

monotonic iff 𝑝𝑛 is increasing or decreasing.

Theorem: Any bdd and monotonic sequence of real numbers converges.

𝑝𝑛 𝑖𝑠 𝑚𝑜𝑛𝑜𝑡𝑜𝑛𝑖𝑐 𝑎𝑛𝑑 𝑝𝑛 /𝑛 ∈ ℕ 𝑏𝑑𝑑 ⇒ 𝑝𝑛 converges.

𝑝𝑛 𝑖𝑠 𝑖𝑛𝑐𝑟𝑒𝑎𝑠𝑖𝑛𝑔: lim 𝑝𝑛 = sup 𝑝𝑛 /𝑛 ∈ ℕ

Therefore:
𝑝𝑛 𝑖𝑠 𝑑𝑒𝑐𝑟𝑒𝑎𝑠𝑖𝑛𝑔: lim 𝑝𝑛 = inf 𝑝𝑛 /𝑛 ∈ ℕ

Proof: Case 1: 𝑝𝑛 𝑖𝑠 𝑖𝑛𝑐𝑟𝑒𝑎𝑠𝑖𝑛𝑔

𝑝𝑛 /𝑛 ∈ ℕ ⊂ ℝ 𝑏𝑑𝑑 ∃𝑝 ∈ ℝ/𝑝 = sup 𝑝𝑛 /𝑛 ∈ ℕ ⇒

𝐶𝑜𝑚𝑝𝑙𝑒𝑡𝑒𝑛𝑒𝑠𝑠
𝑖𝑛 ℝ
𝐶𝑁&𝑆 𝑠𝑢𝑝: ∀𝜀 > 0, ∃𝑁 ∈ ℕ/ 𝑝 − 𝜀 < 𝑝𝑁
⇒
𝐷𝑒𝑓. 𝑠𝑢𝑝: ∀𝑛 ∈ ℕ, 𝑝𝑛 ≤ 𝑝 ⇒
𝑝𝑛 𝑖𝑠 𝑖𝑛𝑐𝑟𝑒𝑎𝑠𝑖𝑛𝑔 ⇒ ∀𝑛 ∈ ℕ, 𝑝𝑛 ≤ 𝑝𝑛+1 ⇒ ∀𝑛 ≥ 𝑁, 𝑝𝑁 ≤ 𝑝𝑛

⇒ ∀𝜀 > 0, ∃𝑁 ∈ ℕ/∀𝑛 ≥ 𝑁, 𝑝 − 𝜀 < 𝑝𝑁 ≤ 𝑝𝑛 ≤ 𝑝 < 𝑝 + 𝜀 ⇒

⇒ ∀𝜀 > 0, ∃𝑁 ∈ ℕ/∀𝑛 ≥ 𝑁, 𝑝 − 𝜀 < 𝑝𝑛 < 𝑝 + 𝜀 ⇒ lim 𝑝𝑛 = 𝑝 = sup 𝑝𝑛 /𝑛 ∈ ℕ

Case 2: 𝑝𝑛 𝑖𝑠 𝑑𝑒𝑐𝑟𝑒𝑎𝑠𝑖𝑛𝑔

𝑝𝑛 /𝑛 ∈ ℕ ⊂ ℝ 𝑏𝑑𝑑 ∃𝑝 ∈ ℝ/𝑝 = inf 𝑝𝑛 /𝑛 ∈ ℕ ⇒

𝐶𝑜𝑟𝑜𝑙𝑎𝑟𝑖𝑜 𝐶𝑜𝑚𝑝𝑙𝑒𝑡𝑒𝑛𝑒𝑠𝑠
𝑖𝑛 ℝ (∗)
𝐶𝑁&𝑆 𝑖𝑛𝑓: ∀𝜀 > 0, ∃𝑁 ∈ ℕ/ 𝑝𝑁 < 𝑝 + 𝜀
⇒
𝐷𝑒𝑓. 𝑖𝑛𝑓: ∀𝑛 ∈ ℕ, 𝑝 ≤ 𝑝𝑛 ⇒
𝑝𝑛 𝑖𝑠 𝑑𝑒𝑐𝑟𝑒𝑎𝑠𝑖𝑛𝑔 ⇒ ∀𝑛 ∈ ℕ, 𝑝𝑛+1 ≤ 𝑝𝑛 ⇒ ∀𝑛 ≥ 𝑁, 𝑝𝑛 ≤ 𝑝𝑁

⇒ ∀𝜀 > 0, ∃𝑁 ∈ ℕ/∀𝑛 ≥ 𝑁, 𝑝 − 𝜀 < 𝑝 ≤ 𝑝𝑛 ≤ 𝑝𝑁 < 𝑝 + 𝜀 ⇒

⇒ ∀𝜀 > 0, ∃𝑁 ∈ ℕ/∀𝑛 ≥ 𝑁, 𝑝 − 𝜀 < 𝑝𝑛 < 𝑝 + 𝜀 ⇒ lim 𝑝𝑛 = 𝑝 = inf 𝑝𝑛 /𝑛 ∈ ℕ


6.3 SUB SEQUENCES

𝐒𝐔𝐁𝐒𝐄𝐐𝐔𝐄𝐍𝐂𝐄𝐒: Given 𝑝𝑛 a sequence. We call subsequence to any ordered infinitely subset

of elements of 𝑝𝑛 .
Lemma: 𝑝𝑚 𝑘 𝑎 𝑠𝑢𝑏𝑠𝑒𝑞𝑢𝑒𝑛𝑐𝑒 𝑜𝑓 𝑝𝑛 ⇒ ∀𝑘 ∈ ℕ, 𝑚𝑘 ≥ 𝑘

Proof:

INDUCTIVE BASE: 𝑘 = 1: 𝑚1 ∈ ℕ ⇒ 𝑚1 ≥ 1 = 𝑘

INDUCTIVE STEP: 𝐻𝐼 𝑚𝑘 ≥ 𝑘 ⇒ 𝑇𝐼 𝑚𝑘+1 ≥ 𝑘 + 1

𝐵𝑒𝑐𝑎𝑢𝑠𝑒 𝑑𝑒𝑓𝑖𝑛𝑖𝑡𝑖𝑜𝑛 𝑜𝑓 𝑠𝑢𝑏𝑠𝑒𝑞𝑢𝑒𝑛𝑐𝑒: 𝑚𝑘+1 > 𝑚𝑘 𝑘

⇒ 𝑚𝑘+1 > 𝑘 𝑚𝑘+1 ≥ 𝑘 + 1
𝐵𝑒𝑐𝑎𝑢𝑠𝑒 𝐻𝑦𝑝𝑜𝑡𝑕𝑒𝑠𝑖𝑠: 𝑚𝑘 ≥ 𝑘 𝑚 𝑘+1 ∈ℕ
𝑘∈ℕ


Theorem: Any subsequence of a convergent sequence converges to the same limit.

Proof: Consider a sequence 𝑝𝑛 such that lim⁡𝑝𝑛 = 𝑝. Pick up any subsequence 𝑝𝑚 𝑘 of 𝑝𝑛 .

lim⁡𝑝𝑛 = 𝑝 ⇒ ∀𝜀 > 0, ∃𝑁 ∈ ℕ / ∀𝑛 ≥ 𝑁, 𝑑(𝑝, 𝑝𝑛 ) < 𝜀

𝑝𝑚 𝑘 𝑠𝑢𝑏𝑠𝑒𝑞𝑢𝑒𝑛𝑐𝑒 𝑜𝑓 𝑝𝑛 ∀𝑘 ∈ ℕ, 𝑚𝑘 ≥ 𝑘 ⇒
𝐿𝑒𝑚𝑚𝑎

⇒ ∀𝜀 > 0, ∃𝑁 ∈ ℕ / ∀𝑚𝑘 , 𝑘 ≥ 𝑁, 𝑑(𝑝, 𝑝𝑚 𝑘 ) < 𝜀 ⇒ ∀𝜀 > 0, ∃𝑁 ∈ ℕ / ∀𝑚𝑘 ≥ 𝑁, 𝑑(𝑝, 𝑝𝑚 𝑘 ) < 𝜀 ⇒

⇒ lim 𝑝𝑚 𝑘 = 𝑝


6.4 TOPOLOGY AND SEQUENCES

Theorem: 𝐸, 𝑑 𝑚. 𝑒. 𝐴 ⊂ 𝐸 𝑖𝑠 𝑐𝑙𝑜𝑠𝑒𝑑 ⇔ ∀ 𝑝𝑛 ⊂ 𝐴, ( 𝑝𝑛 𝑐𝑜𝑛𝑣𝑒𝑟𝑔𝑒𝑛𝑡 ⇒ lim 𝑝𝑛 ∈ 𝐴)

Proof:

⇒ :
Assume otherwise that ∃ 𝑝𝑛 ⊂ 𝐴 𝑠𝑢𝑐𝑕 𝑡𝑕𝑎𝑡 𝑝𝑛 𝑐𝑜𝑛𝑣𝑒𝑟𝑔𝑒𝑛𝑡 𝑎𝑛𝑑 lim 𝑝𝑛 = 𝑝 ∉ 𝐴

𝐵𝑦 𝑕𝑦𝑝𝑜𝑡𝑕𝑒𝑠𝑖𝑠: 𝐴 𝑖𝑠 𝑐𝑙𝑜𝑠𝑒𝑑 ⇒ 𝐴𝐶 𝑖𝑠 𝑜𝑝𝑒𝑛 ⇒ ∀𝑥 ∈ 𝐴𝐶 , ∃𝐵 𝑥, 𝑟 ⊂ 𝐴𝐶

⇒
𝑝 ∉ 𝐴 ⇒ 𝑝 ∈ 𝐴𝐶
⇒ ∃𝐵 𝑝, 𝑟 ⊂ 𝐴 𝐶 𝐵 𝑝, 𝑟
⇒ ∃ ∈ ℕ, 𝑝𝑛 ∉ 𝐵 𝑝, 𝑟 ⇒
𝑝𝑛 ⊂ 𝐴 ⇒ ∀𝑛 ∈ ℕ, 𝑝𝑛 ∈ 𝐴 ∀𝑛
⇒ ∃𝑟 > 0/∀𝑛 ∈ ℕ, 𝑝𝑛 ∉ 𝐵 𝑝, 𝑟 ⇒ lim 𝑝𝑛 ≠ 𝑝 (absurd)

∴ ∀ 𝑝𝑛 ⊂ 𝐴, ( 𝑝𝑛 𝑐𝑜𝑛𝑣𝑒𝑟𝑔𝑒𝑛𝑡 ⇒ lim 𝑝𝑛 ∈ 𝐴)

⇐ :
Assume otherwise that A is not closed ⇒ 𝐴𝐶 𝑖𝑠 𝑛𝑜𝑡 𝑜𝑝𝑒𝑛 ⇒ ∃𝑥 ∈ 𝐴𝐶 /∀𝑟 > 0, 𝐵 𝑥, 𝑟 ⊄ 𝐴𝐶 ⇒
⇒ ∃𝑥 ∈ 𝐴𝐶 /∀𝑟 > 0, 𝐵 𝑥, 𝑟 ∩ 𝐴 ≠ ∅ 1
1 ⇒ ∃𝑥 ∈ 𝐴𝐶 /∀𝑛 ∈ ℕ, 𝐵 𝑥, ∩ 𝐴 ≠ ∅ ⇒
𝑇𝑎𝑘𝑒 𝑟 = 𝑛 𝑛
1
⇒ ∃𝑥 ∈ 𝐴𝐶 /∀𝑛 ∈ ℕ, ∃ 𝑗𝑛 ∈ 𝐵 𝑥, ∩ 𝐴
𝑛
Claim: 𝑗𝑛 𝑐𝑜𝑛𝑣𝑒𝑟𝑔𝑒𝑠 𝑎𝑛𝑑 lim 𝑗𝑛 = 𝑥⁡
⁡⁡

1
Proof: ℕ 𝑖𝑠 𝑛𝑜𝑡 𝑏𝑑𝑑 𝑓𝑟𝑜𝑚 𝑎𝑏𝑜𝑣𝑒 ⇒ ∀𝜀 > 0, ∃𝑁 ∈ ℕ/ < 𝑁 ⇒
𝜀
1 1
⇒ ∀𝜀 > 0, ∃𝑁 ∈ ℕ/∀𝑛 ≥ 𝑁, < 𝑛 ⇒ ∀𝜀 > 0, ∃𝑁 ∈ 𝑁/∀𝑛 ≥ 𝑁, < 𝜀
𝜀 𝑛 ⇒
1 1 1
∀𝑛 ∈ ℕ, 𝑗𝑛 ∈ 𝐵 𝑥, ∩ 𝐴 ⇒ ∀𝑛 ∈ ℕ, 𝑗𝑛 ∈ 𝐵 𝑥, ⇒ ∀𝑛 ∈ ℕ, 𝑑(𝑗𝑛 , 𝑥) <
𝑛 𝑛 𝑛
1
⇒ ∀𝜀 > 0, ∃𝑁 ∈ 𝑁/∀𝑛 ≥ 𝑁, 𝑑(𝑗𝑛 , 𝑥) < < 𝜀 ⇒ ∀𝜀 > 0, ∃𝑁 ∈ 𝑁/∀𝑛 ≥ 𝑁, 𝑑(𝑗𝑛 , 𝑥) < 𝜀 ⇒
𝑛
⇒ lim 𝑗𝑛 = 𝑥⁡
.

1
∴ ∃𝑥 ∈ 𝐴𝐶 , ∃ 𝑗𝑛 /∀𝑛 ∈ ℕ, 𝑗𝑛 ∈ 𝐵 𝑥, ∩ 𝐴 𝑎𝑛𝑑 lim 𝑗𝑛 = 𝑥⁡⇒
𝑛
⇒ ∃ 𝑗𝑛 ⊂∩ 𝐴 / lim 𝑗𝑛 = 𝑥 ∈ 𝐴𝐶 (absurd by hypothesis)

∴ 𝐴 𝑖𝑠 𝑐𝑙𝑜𝑠𝑒𝑑


Corollary (HW): ℚ is not open and not closed.

Proof:
1. ℚ is not open

Assume otherwise that ℚ 𝑖𝑠 𝑜𝑝𝑒𝑛 ⇒ ℚ𝐶 = 𝕀 𝑖𝑠 𝑐𝑙𝑜𝑠𝑒𝑑 ⇒

⇒ ∀ 𝑝𝑛 ⊂ 𝕀, ( 𝑝𝑛 𝑐𝑜𝑛𝑣𝑒𝑟𝑔𝑒𝑛𝑡 ⇒ lim 𝑝𝑛 ∈ 𝕀)
2
Consider the sequence 𝑝𝑛 / 𝑝𝑛 = , ∀𝑛 ∈ ℕ∗
𝑛

1) Claim: 𝑝𝑛 ∈ 𝕀, ∀𝑛 ∈ ℕ∗
𝑊𝑒 ′ 𝑣𝑒 𝑝𝑟𝑜𝑣𝑒𝑑: 2 ∈ 𝕀 1 2
1 1 ⇒ 2. ∈ 𝕀, ∀𝑛 ∈ ℕ∗ ⇒ ∈ 𝕀, ∀𝑛 ∈ ℕ∗ ⇒ 𝑝𝑛 ∈ 𝕀, ∀𝑛 ∈ ℕ∗
∀𝑛 ∈ ℕ∗ , ∈ ℚ 𝑎𝑛𝑑 ≠0 𝑛 𝑛
𝑛 𝑛

2) Claim: lim⁡𝑝𝑛 = 0
1 1
lim 2 = 2 𝑎𝑛𝑑 lim = 0 ⇒ lim 2 . = 2. 0 = 0 ⇒ lim⁡𝑝𝑛 = 0
𝑛 𝑛

∴ From 1 and 2: ∃ 𝑝𝑛 ⊂ 𝕀, 𝑝𝑛 𝑐𝑜𝑛𝑣𝑒𝑟𝑔𝑒𝑛𝑡 𝑎𝑛𝑑 lim 𝑝𝑛 ∉ 𝕀 (absurd)

∴ ℚ is not open.

2. ℚ is not closed

Assume otherwise that ℚ 𝑖𝑠 𝑐𝑙𝑜𝑠𝑒𝑑 ⇒ ∀ 𝑝𝑛 ⊂ ℚ , ( 𝑝𝑛 𝑐𝑜𝑛𝑣𝑒𝑟𝑔𝑒𝑛𝑡 ⇒ lim 𝑝𝑛 ∈ ℚ )

2.10 𝑛
Consider the sequence 𝑝𝑛 / 𝑝𝑛 = , ∀𝑛 ∈ ℕ∗
10 𝑛
NOTE: 𝑎 = 𝑤𝑕𝑜𝑙𝑒 𝑝𝑎𝑟𝑡 𝑜𝑓 𝑎, ∀𝑎 ∈ ℝ

1) Claim: 𝑝𝑛 ∈ ℚ , ∀𝑛 ∈ ℕ∗
𝐵𝑦 𝑑𝑒𝑓𝑖𝑛𝑖𝑡𝑖𝑜𝑛 𝑜𝑑 𝑤𝑕𝑜𝑙𝑒 𝑝𝑎𝑟𝑡, 2. 10𝑛 ∈ ℤ ⊂ ℚ, ∀𝑛 ∈ ℕ∗
𝑆𝑖𝑛𝑐𝑒 𝑡𝑕𝑒 𝑝𝑟𝑜𝑑𝑢𝑐𝑡 𝑖𝑠 𝑐𝑙𝑜𝑠𝑒 𝑖𝑛 ℚ 𝑎𝑛𝑑 10 ∈ ℤ ⊂ ℚ, 10𝑛 ∈ ℚ , ∀𝑛 ∈ ℕ∗ 1 ⇒
𝑛 ∗ ⇒ 𝑛 ∈ ℚ , ∀𝑛 ∈ ℕ∗
1 ∈ ℚ 𝑎𝑛𝑑 10 ≠ 0 , ∀𝑛 ∈ ℕ 10
2. 10𝑛 ∗ ∗
⇒ ∈ ℚ , ∀𝑛 ∈ ℕ ⇒ 𝑝𝑛 ∈ ℚ , ∀𝑛 ∈ ℕ
10𝑛

2) Claim: lim⁡𝑝𝑛 = 2

∴ From 1 and 2: ∃ 𝑝𝑛 ⊂ ℚ , 𝑝𝑛 𝑐𝑜𝑛𝑣𝑒𝑟𝑔𝑒𝑛𝑡 𝑎𝑛𝑑 lim 𝑝𝑛 ∉ ℚ (absurd)

∴ ℚ is not closed.

Corollary (HW): 𝕀 = ℝ ∖ ℚ is not open and not closed.

Proof:

1. 𝕀 is not open

Assume otherwise that 𝕀 𝑖𝑠 𝑜𝑝𝑒𝑛 ⇒ 𝕀𝐶 = ℚ𝐶 𝑖𝑠 𝑐𝑙𝑜𝑠𝑒𝑑 (absurd)

2. 𝕀 is not closed

Assume otherwise that 𝕀 𝑖𝑠 𝑐𝑙𝑜𝑠𝑒𝑑 ⇒ 𝕀𝐶 = ℚ𝐶 𝑖𝑠 𝑜𝑝𝑒𝑛 (absurd)



6.4 INTERIOR, CLAUSURE AND BOUNDERY

𝐈𝐍𝐓𝐄𝐑𝐈𝐎𝐑 𝐏𝐎𝐈𝐍𝐓: Given a metric space 𝐸, 𝑑 , 𝑎𝑛𝑑 𝑎 𝑠𝑒𝑡 𝑆 ⊂ 𝐸. We say that 𝑝 ∈ 𝑆 is an

interior point of S iff ∃𝐵 𝑝, 𝑟 ⊂ 𝑆.

𝐈𝐍𝐓𝐄𝐑𝐈𝐎𝐑 𝐨𝐟 𝐚 𝐒𝐄𝐓: Given a metric space 𝐸, 𝑑 , 𝑎𝑛𝑑 𝑎 𝑠𝑒𝑡 𝑆 ⊂ 𝐸. The set of all the interior
points of S is called interior of S, and denoted 𝑖𝑛𝑡(𝑆).

Observation (Mio): 𝐸, 𝑑 𝑚. 𝑒. ∀𝑆 ⊂ 𝐸, 𝑖𝑛𝑡 𝑆 ⊂ 𝑆

Observation (Práctico): 𝐸, 𝑑 𝑚. 𝑒. ∀𝑆 ⊂ 𝐸, 𝑖𝑛𝑡 𝑆 𝑖𝑠 𝑜𝑝𝑒𝑛

Observation (Práctico): 𝐸, 𝑑 𝑚. 𝑒. 𝑆 ⊂ 𝐸. ∀𝐴 ⊂ 𝑆, (𝐴 𝑖𝑠 𝑜𝑝𝑒𝑛 ⇒ 𝐴 ⊂ 𝑖𝑛𝑡 𝑆 )

𝐂𝐋𝐎𝐒𝐔𝐑𝐄 𝐒𝐄𝐓: Given a metric space 𝐸, 𝑑 , 𝑎𝑛𝑑 𝑎 𝑠𝑒𝑡 𝑆 ⊂ 𝐸. We define the closure of S as the
intersection of all closed subsets of E that contain S. The closure set of S is denoted by 𝑆.

Observation (Práctico): 𝐸, 𝑑 𝑚. 𝑒. ∀𝑆 ⊂ 𝐸, 𝑆 ⊂ 𝑆

Observation (Práctico): 𝐸, 𝑑 𝑚. 𝑒. 𝑆 ⊂ 𝐸. 𝑆 𝑖𝑠 𝑐𝑙𝑜𝑠𝑒𝑑 ⇔ 𝑆 = 𝑆

Observation (Práctico): 𝐸, 𝑑 𝑚. 𝑒. 𝑆 ⊂ 𝐸
𝑆 = lim 𝑎𝑛 / 𝑎𝑛 ⊂ 𝑆 𝑎𝑛𝑑 𝑎𝑛 𝑐𝑜𝑛𝑣𝑒𝑟𝑔𝑒𝑛𝑡 𝑖𝑛 𝐸
Theorem (Práctico): 𝐸, 𝑑 𝑚. 𝑒. 𝑆 ⊂ 𝐸. ∀𝑝 ∈ 𝐸, 𝑝 ∈ 𝑆 ⇔ ∀𝐵(𝑝, 𝑟) ∩ 𝑆 ≠ ∅ ⇔ 𝑝 ∉ 𝑖𝑛𝑡(𝑆 𝐶 )

𝐁𝐎𝐔𝐍𝐃𝐄𝐑𝐘 𝐨𝐟 𝐚 𝐒𝐄𝐓: Given a metric space 𝐸, 𝑑 , 𝑎𝑛𝑑 𝑎 𝑠𝑒𝑡 𝑆 ⊂ 𝐸. The boundary of S s

defined to be 𝑆 ∩ 𝑆 𝐶 . The boundary set of S is denoted by b(𝑆).

Observation (Práctico): 𝐸, 𝑑 𝑚. 𝑒. ∀𝑆 ⊂ 𝐸, 𝐸 = 𝑖𝑛𝑡(𝑆) ⊔ 𝑖𝑛𝑡(𝑆 𝐶 ) ⊔ 𝑏(𝑆)

Observation (Práctico): 𝐸, 𝑑 𝑚. 𝑒. 𝑆 ⊂ 𝐸. 𝑆 𝑖𝑠 𝑐𝑙𝑜𝑠𝑒𝑑 ⇔ 𝑏(𝑆) ⊂ 𝑆

Observation (Práctico): 𝐸, 𝑑 𝑚. 𝑒. 𝑆 ⊂ 𝐸. 𝑆 𝑖𝑠 𝑜𝑝𝑒𝑛 ⇔ 𝑏 𝑆 ∩ 𝑆 = ∅

7. COMPLETENESS
𝐂𝐀𝐔𝐂𝐇𝐘 𝐒𝐄𝐐𝐔𝐄𝐍𝐂𝐄: Given a metric space 𝐸, 𝑑 , 𝑎𝑛𝑑 𝑝𝑛 a sequence of elements of E. We
say that 𝑝𝑛 is Cauchy iff ∀𝜀 > 0, ∃𝑁 ∈ ℕ / ∀𝑛, 𝑚 ≥ 𝑁, 𝑑(𝑝𝑛 , 𝑝𝑚 ) < 𝜀.

Theorem: 𝐸, 𝑑 = 𝐸1 . 𝑎𝑛 𝑎𝑛𝑑 𝑏𝑛 𝑎𝑟𝑒 𝐶𝑎𝑢𝑐𝑕𝑦 𝑖𝑛 𝐸 . 𝑐 ∈ ℝ. 𝑇𝑕𝑒𝑛:

1. 𝑐. 𝑎𝑛 𝑖𝑠 𝐶𝑎𝑢𝑐𝑕𝑦
2. 𝑎𝑛 + 𝑏𝑛 𝑖𝑠 𝐶𝑎𝑢𝑐𝑕𝑦
3. 𝑎𝑛 − 𝑏𝑛 𝑖𝑠 𝐶𝑎𝑢𝑐𝑕𝑦
4. 𝑎𝑛 . 𝑏𝑛 𝑖𝑠 𝐶𝑎𝑢𝑐𝑕𝑦

Proof:

1. Analysis: 𝑐. 𝑎𝑛 𝑖𝑠 𝐶𝑎𝑢𝑐𝑕𝑦 ⇔ ∀𝜀 > 0, ∃𝑁 ∈ ℕ / ∀𝑛, 𝑚 ≥ 𝑁, 𝑐. 𝑎𝑛 − 𝑐. 𝑎𝑚 < 𝜀

𝑐. 𝑎𝑛 − 𝑐. 𝑎𝑚 = 𝑐 . 𝑎𝑛 − 𝑎𝑚

Proof: Case 1: 𝑐 = 0
𝑐 = 0 ⇒ c. 𝑎𝑛 = 0, ∀𝑛 ∈ ℕ ⇒ ∀𝜀 > 0, ∃𝑁 ∈ ℕ/∀𝑛 ≥ 𝑁, 𝑐. 𝑎𝑛 − 𝑐. 𝑎𝑚 < 𝜀 ⇒
⇒ 𝑐. 𝑎𝑛 𝑖𝑠 𝐶𝑎𝑢𝑐𝑕𝑦

Case 2: 𝑐 ≠ 0
𝜀
𝑃𝑖𝑐𝑘 𝑢𝑝 𝑎𝑛𝑦 𝜀 > 0 ⇒ >0 𝜀
𝑐 ⇒ ∃𝑁 ∈ ℕ/∀𝑛, 𝑚 ≥ 𝑁, 𝑎𝑛 − 𝑎𝑚 < ⇒
𝑐
𝑎𝑛 𝑖𝑠 𝐶𝑎𝑢𝑐𝑕𝑦
𝜀
⇒ ∃𝑁 ∈ ℕ/∀𝑛, 𝑚 ≥ 𝑁, 𝑐 . 𝑎𝑛 − 𝑎𝑚 < . 𝑐 =𝜀
𝑐
∴ ∀𝜀 > 0, ∃𝑁 ∈ ℕ/∀𝑛, 𝑚 ≥ 𝑁, 𝑐. 𝑎𝑛 − 𝑐. 𝑎𝑚 < 𝜀 ⇒ 𝑐. 𝑎𝑛 𝑖𝑠 𝐶𝑎𝑢𝑐𝑕𝑦

2. Analysis: 𝑎𝑛 + 𝑏𝑛 𝑖𝑠 𝐶𝑎𝑢𝑐𝑕𝑦 ⇔ ∀𝜀 > 0, ∃𝑁 ∈ ℕ / ∀𝑛, 𝑚 ≥ 𝑁, 𝑎𝑛 + 𝑏𝑛 − (𝑎𝑚 + 𝑏𝑚 ) < 𝜀

𝑎𝑛 + 𝑏𝑛 − (𝑎𝑚 + 𝑏𝑚 ) = (𝑎𝑛 − 𝑎𝑚 ) + (𝑏𝑛 − 𝑏𝑚 ) ≤ 𝑎𝑛 − 𝑎𝑚 + 𝑏𝑛 − 𝑏𝑚
𝜀
Proof: 𝑃𝑖𝑐𝑘 𝑢𝑝 𝑎𝑛𝑦 𝜀 > 0 ⇒ > 0
2
𝜀
>0 𝜀
2 ⇒ ∃𝑁1 ∈ ℕ/∀𝑛, 𝑚 ≥ 𝑁1 , 𝑎𝑛 − 𝑎𝑚 <
𝑎𝑛 𝑖𝑠 𝐶𝑎𝑢𝑐𝑕𝑦 2
𝜀 ⇒
>0 𝜀
2 ⇒ ∃𝑁2 ∈ ℕ/∀𝑛, 𝑚 ≥ 𝑁2 , 𝑏𝑛 − 𝑏𝑚 <
𝑏𝑛 𝑖𝑠 𝐶𝑎𝑢𝑐𝑕𝑦 2
𝐶𝑕𝑜𝑜𝑠𝑒 𝑁 = max 𝑁1 , 𝑁2
𝜀 𝜀
⇒ ∃𝑁 ∈ ℕ/∀𝑛, 𝑚 ≥ 𝑁, 𝑎𝑛 − 𝑎𝑚 + 𝑏𝑛 − 𝑏𝑚 < + = 𝜀
2 2 ⇒
𝐵𝑦 ⊿ − 𝐼𝑛𝑒𝑞.: 𝑎𝑛 + 𝑏𝑛 − (𝑎𝑚 + 𝑏𝑚 ) = (𝑎𝑛 − 𝑎𝑚 ) + (𝑏𝑛 − 𝑏𝑚 ) ≤ 𝑎𝑛 − 𝑎𝑚 + 𝑏𝑛 − 𝑏𝑚

⇒ ∃𝑁 ∈ ℕ/∀𝑛, 𝑚 ≥ 𝑁, 𝑎𝑛 + 𝑏𝑛 − (𝑎𝑚 + 𝑏𝑚 ) < 𝜀

∴ ∀𝜀 > 0, ∃𝑁 ∈ ℕ/∀𝑛, 𝑚 ≥ 𝑁, 𝑎𝑛 + 𝑏𝑛 − (𝑎𝑚 + 𝑏𝑚 ) < 𝜀 ⇒ 𝑎𝑛 + 𝑏𝑛 𝑖𝑠 𝐶𝑎𝑢𝑐𝑕𝑦

3. Is a direct consequence of 1 and 2.

4. Analysis: 𝑎𝑛 . 𝑏𝑛 𝑖𝑠 𝐶𝑎𝑢𝑐𝑕𝑦 ⇔ ∀𝜀 > 0, ∃𝑁 ∈ ℕ/∀𝑛, 𝑚 ≥ 𝑁, 𝑎𝑛 . 𝑏𝑛 − 𝑎𝑚 . 𝑏𝑚 < 𝜀

𝑎𝑛 . 𝑏𝑛 − 𝑎𝑚 . 𝑏𝑚 = 𝑎𝑛 . 𝑏𝑛 − 𝑎𝑛 . 𝑏𝑚 + 𝑎𝑛 . 𝑏𝑚 − 𝑎𝑚 . 𝑏𝑚 = 𝑎𝑛 . 𝑏𝑛 − 𝐵 − 𝐵. (𝑎𝑛 − 𝐴) ≤
≤ 𝑎𝑛 . 𝑏𝑛 − 𝑏𝑚 + 𝑏𝑚 . (𝑎𝑛 − 𝐴) = 𝑎𝑛 . 𝑏𝑛 − 𝑏𝑚 + 𝑏𝑚 . 𝑎𝑛 − 𝑎𝑚

Proof: 𝑃𝑖𝑐𝑘 𝑢𝑝 𝑎𝑛𝑦 𝜀 > 0

𝑎𝑛 𝑖𝑠 𝐶𝑎𝑢𝑐𝑕𝑦 𝑎𝑛 /𝑛 ∈ ℕ 𝑖𝑠 𝑏𝑑𝑑. ⇒ ∃𝑘 ∈ ℝ+/∀𝑛 ∈ ℕ, 𝑎𝑛 < 𝑘

𝐵𝑦 𝑇𝑕.
𝑏𝑒𝑓𝑜𝑟𝑒
𝑏𝑛 𝑖𝑠 𝐶𝑎𝑢𝑐𝑕𝑦 ⇒
𝜀
𝜀 ⇒ ∃𝐻 ∈ ℕ/∀𝑛, 𝑚 ≥ 𝐻, 𝑏𝑛 − 𝑏𝑚 <
𝜀>0⇒ >0 2𝑘
2𝑘
𝜀 𝜀
⇒ ∃𝐻 ∈ ℕ/∀𝑛, 𝑚 ≥ 𝐻, 𝑎𝑛 . 𝑏𝑛 − 𝑏𝑚 < .𝑘 = (i)
2𝑘 2

𝑏𝑛 𝑖𝑠 𝐶𝑎𝑢𝑐𝑕𝑦 𝑏𝑛 /𝑛 ∈ ℕ 𝑖𝑠 𝑏𝑑𝑑. ⇒ ∃𝑞 ∈ ℝ+ /∀𝑛 ∈ ℕ, 𝑏𝑛 < 𝑞

𝐵𝑦 𝑇𝑕.
𝑏𝑒𝑓𝑜𝑟𝑒
𝑎𝑛 𝑖𝑠 𝐶𝑎𝑢𝑐𝑕𝑦 ⇒
𝜀
𝜀 ⇒ ∃𝑅 ∈ ℕ/∀𝑛, 𝑚 ≥ 𝑅, 𝑎𝑛 − 𝑎𝑚 <
𝜀>0⇒ >0 2𝑞
2𝑞

𝜀 𝜀
⇒ ∃𝑅 ∈ ℕ/∀𝑛, 𝑚 ≥ 𝑅, 𝑏𝑛 . 𝑎𝑛 − 𝑎𝑚 < .𝑞 = (ii)
2𝑞 2

𝑇𝑎𝑘𝑒 𝑁 = 𝑚𝑎𝑥 𝐻, 𝑅
𝑖 𝑎𝑛𝑑 (𝑖𝑖)

𝜀 𝜀
⇒ ∃𝑁 ∈ ℕ/∀𝑛, 𝑚 ≥ 𝑁, 𝑎𝑛 . 𝑏𝑛 − 𝑏𝑚 + 𝑏𝑛 . 𝑎𝑛 − 𝑎𝑚 < + =𝜀
2 2 ⇒
𝐵𝑦 ⊿ − 𝐼𝑛𝑒𝑞.: 𝑎𝑛 . 𝑏𝑛 − 𝑎𝑚 . 𝑏𝑚 ≤ 𝑎𝑛 . 𝑏𝑛 − 𝑏𝑚 + 𝑏𝑚 . 𝑎𝑛 − 𝑎𝑚

⇒ ∃𝑁 ∈ ℕ/∀𝑛, 𝑚 ≥ 𝑁, 𝑎𝑛 . 𝑏𝑛 − 𝑎𝑚 . 𝑏𝑚 < 𝜀

∴ ∀𝜀 > 0, ∃𝑁 ∈ ℕ/∀𝑛, 𝑚 ≥ 𝑁, 𝑎𝑛 . 𝑏𝑛 − 𝑎𝑚 . 𝑏𝑚 < 𝜀 ⇒ 𝑎𝑛 . 𝑏𝑛 𝑖𝑠 𝐶𝑎𝑢𝑐𝑕𝑦



Theorem: Any convergent sequence is Cauchy.

𝐸, 𝑑 𝑒. 𝑚. 𝑝𝑛 ⊂ 𝐸 / lim 𝑝𝑛 = 𝑝 ∈ 𝐸 ⇒ 𝑝𝑛 𝑖𝑠 𝐶𝑎𝑢𝑐𝑕𝑦

Proof: 𝑝𝑛 𝑖𝑠 𝐶𝑎𝑢𝑐𝑕𝑦 ⇔ ∀𝜀 > 0, ∃𝑁 ∈ ℕ / ∀𝑛, 𝑚 ≥ 𝑁, 𝑑(𝑝𝑛 , 𝑝𝑚 ) < 𝜀

𝜀
𝑃𝑖𝑐𝑘 𝑢𝑝 𝑎𝑛𝑦 𝜀 > 0 ⇒ >0 𝜀
2 ⇒ ∃𝑁 ∈ ℕ / ∀𝑛 ≥ 𝑁, 𝑑(𝑝𝑛 , 𝑝) < ⇒
lim 𝑝𝑛 = 𝑝 2
𝜀
𝑑 𝑝𝑛 , 𝑝 <
⇒ ∃𝑁 ∈ ℕ / ∀𝑛, 𝑚 ≥ 𝑁, 2
𝜀 ⇒ ∃𝑁 ∈ ℕ / ∀𝑛, 𝑚 ≥ 𝑁, 𝑑 𝑝𝑛 , 𝑝 + 𝑑 𝑝𝑚 , 𝑝 < 𝜀 ⇒
𝑑 𝑝𝑚 , 𝑝 <
2
𝐵𝑦 ⊿ − 𝐼𝑛𝑒𝑞𝑢𝑎𝑙𝑖𝑡𝑦: 𝑑 𝑝𝑛 , 𝑝𝑚 ≤ 𝑑 𝑝𝑛 , 𝑝 + 𝑑 𝑝𝑚 , 𝑝
⇒ ∃𝑁 ∈ ℕ / ∀𝑛, 𝑚 ≥ 𝑁, 𝑑 𝑝𝑛 , 𝑝𝑚 < 𝜀

∴ ∀𝜀 > 0, ∃𝑁 ∈ ℕ/∀𝑛, 𝑚 ≥ 𝑁, 𝑑 𝑝𝑛 , 𝑝𝑚 < 𝜀 ⇒ 𝑝𝑛 𝑖𝑠 𝐶𝑎𝑢𝑐𝑕𝑦 

Theorem: Any Cauchy sequence is bounded.

𝐸, 𝑑 𝑒. 𝑚. 𝑝𝑛 𝑖𝑠 𝐶𝑎𝑢𝑐𝑕𝑦 ⇒ 𝑝𝑛 𝑖𝑠 𝑏𝑑𝑑

Proof: 𝑝𝑛 𝑖𝑠 𝐶𝑎𝑢𝑐𝑕𝑦 ∃𝑁 ∈ ℕ / ∀𝑛, 𝑚 ≥ 𝑁, 𝑑(𝑝𝑛 , 𝑝𝑚 ) < 1

𝜀=1 𝑖𝑛 𝑡𝑕𝑒 𝑑𝑒𝑓𝑖𝑛𝑖𝑡𝑖𝑜𝑛 𝑁≥𝑁
⇒ ∃𝑁 ∈ ℕ / ∀𝑛 ≥ 𝑁, 𝑑(𝑝𝑛 , 𝑝𝑁 ) < 1

Take 𝑟 = max 1 , 𝑑 𝑝𝑛 , 𝑝𝑁 ; 𝑛 ≤ 𝑁 − 1

max 1 , 𝑑 𝑝𝑛 , 𝑝𝑁 ; 𝑛 ≤ 𝑁 − 1 ≥ 1 > 0 ⇒ 𝑟 > 0 ⇒ 𝑟 + 1 > 0

Claim: ∀𝑛 ∈ ℕ, 𝑝𝑛 ∈ 𝐵(𝑝𝑁 , 𝑟 + 1)

Proof: Pick up any 𝑛 ∈ ℕ ⇒ 𝑛 ≤ 𝑁 − 1 𝑜𝑟 𝑛 ≥ 𝑁

If 𝑛 ≤ 𝑁 − 1 ⇒ 𝑑 𝑝𝑛 , 𝑝𝑁 ∈ 1 , 𝑑 𝑝𝑛 , 𝑝𝑁 ; 𝑛 ≤ 𝑁 − 1 ⇒
⇒ 𝑑 𝑝𝑛 , 𝑝𝑁 ≤ max 1 , 𝑑 𝑝𝑛 , 𝑝𝑁 ; 𝑛 ≤ 𝑁 − 1 = 𝑟 < 𝑟 + 1 ⇒ 𝑝𝑛 ∈ 𝐵(𝑝𝑁 , 𝑟 + 1)

𝐼𝑓 𝑛 ≥ 𝑁 ⇒ 𝑑 𝑝𝑛 , 𝑝𝑁 ≤ 1
⇒
1 ∈ 1 , 𝑑 𝑝𝑛 , 𝑝𝑁 ; 𝑛 ≤ 𝑁 − 1 ⇒ 1 ≤ max 1 , 𝑑 𝑝𝑛 , 𝑝𝑁 ; 𝑛 ≤ 𝑁 − 1 = 𝑟 < 𝑟 + 1
⇒ 𝑑 𝑝𝑛 , 𝑝𝑁 < 𝑟 + 1 ⇒ 𝑝𝑛 ∈ 𝐵(𝑝𝑁 , 𝑟 + 1)

∴ ∃𝐵 𝑝𝑁 , 𝑟 + 1 / 𝑝𝑛 ⊂ 𝐵 𝑝𝑁 , 𝑟 + 1 ⇒ 𝑝𝑛 𝑖𝑠 𝑏𝑑𝑑 

Theorem: Any subsequence of a Cauchy sequence is Cauchy.

Let 𝑝𝑛 be a Cauchy sequence and 𝑝𝑛 𝑘 𝑎 𝑠𝑢𝑏𝑠𝑒𝑞𝑢𝑒𝑛𝑐𝑒 𝑜𝑓 𝑝𝑛 ⇒ 𝑝𝑛 𝑘 𝑖𝑠 𝐶𝑎𝑢𝑐𝑕𝑦

𝐏𝐫𝐨𝐨𝐟: 𝑝𝑛 𝑖𝑠 𝐶𝑎𝑢𝑐𝑕𝑦 ⇒ ∀𝜀 > 0, ∃𝑁 ∈ ℕ / ∀𝑛, 𝑚 ≥ 𝑁, 𝑑(𝑝𝑛 , 𝑝𝑚 ) < 𝜀

𝑝𝑛 𝑘 ≥ 𝑘 ≥ 𝑁 ⇒
𝐵𝑒𝑐𝑎𝑢𝑠𝑒 𝐿𝑒𝑚𝑚𝑎: ∀𝑘, 𝑙 ≥ 𝑁,
𝑝𝑛 𝑙 ≥ 𝑙 ≥ 𝑁
⇒ ∀𝜀 > 0, ∃𝑁 ∈ ℕ / ∀𝑘, 𝑙 ≥ 𝑁, 𝑑(𝑝𝑛 𝑘 , 𝑝𝑛 𝑙 ) < 𝜀 ⇒ 𝑝𝑛 𝑘 𝑖𝑠 𝐶𝑎𝑢𝑐𝑕𝑦 

Theorem: Let 𝑎𝑛 be a Cauchy sequence and 𝑎𝑛 𝑘 𝑎 𝑠𝑢𝑏𝑠𝑒𝑞𝑢𝑒𝑛𝑐𝑒 𝑜𝑓 𝑎𝑛 . If 𝑎𝑛 𝑘

converges to A ⇒ 𝑎𝑛 converges to A

Proof: lim𝑎𝑛 = 𝐴 ⁡⇔ ∀𝜀 > 0, ∃𝑁 ∈ ℕ / ∀𝑛 ≥ 𝑁, 𝑑(𝑎𝑛 , 𝑎) < 𝜀

𝜀
𝑃𝑖𝑐𝑘 𝑢𝑝 𝑎𝑛𝑦 𝜀 > 0 ⇒ >0
2
 𝜀
>0 𝜀
2 ⇒ ∃𝑁1 ∈ ℕ / ∀𝑛, 𝑚 ≥ 𝑁1 , 𝑑(𝑎𝑛 , 𝑎𝑚 ) <
𝑎𝑛 𝑖𝑠 𝐶𝑎𝑢𝑐𝑕𝑦 2
𝜀
>0 𝜀 𝑚 𝑘 ≥𝑘≥𝑁2
2 ⇒ ∃𝑁2 ∈ ℕ / ∀𝑘 ≥ 𝑁2 , 𝑑(𝑎𝑚 𝑘 , 𝐴) <
lim 𝑎𝑚 𝑘 = 𝐴 2
𝐶𝑕𝑜𝑜𝑠𝑒 𝑁 = max 𝑁1 , 𝑁2

𝜀 𝜀
⇒ ∃𝑁 ∈ ℕ / ∀𝑛, 𝑘 ≥ 𝑁, 𝑑 𝑎𝑛 , 𝑎𝑚 𝑘 + 𝑑 𝑎𝑚 𝑘 , 𝐴 < + =𝜀
2 2 ⇒
𝐵𝑦 ⊿ − 𝐼𝑛𝑒𝑞𝑢𝑎𝑙𝑖𝑡𝑦: 𝑑 𝑎𝑛 , 𝐴 ≤ 𝑑 𝑎𝑛 , 𝑎𝑚 𝑘 + 𝑑 𝑎𝑚 𝑘 , 𝐴
⇒ ∃𝑁 ∈ ℕ / ∀𝑛 ≥ 𝑁, 𝑑 𝑎𝑛 , 𝐴 < 𝜀
∴ ∀𝜀 > 0, ∃𝑁 ∈ ℕ / ∀𝑛 ≥ 𝑁, 𝑑 𝑎𝑛 , 𝐴 < 𝜀 ⇒ 𝑎𝑛 𝑐𝑜𝑛𝑣𝑒𝑟𝑔𝑒𝑠𝑡𝑜 𝐴 

𝐂𝐎𝐌𝐏𝐋𝐄𝐓𝐄 𝐌𝐄𝐓𝐑𝐈𝐂 𝐒𝐏𝐀𝐂𝐄: We say that a metric space 𝐸, 𝑑 is complete iff ∀ 𝑝𝑛 ⊂ 𝐸,

𝑝𝑛 𝐶𝑎𝑢𝑐𝑕𝑦 ⇒ ∃𝑝 ∈ 𝐸/ lim⁡𝑝𝑛 = 𝑝

Theorem: Given a complete metric space 𝐸, 𝑑 𝑎𝑛𝑑 𝑎 𝑠𝑒𝑡 𝐴 ≠ ∅ 𝑎𝑛𝑑 𝐴 ⊂ 𝐸.

A is closed ⇔ 𝑡𝑕𝑒 𝑚. 𝑒. 𝐴, 𝑑 𝑖𝑠 𝑐𝑜𝑚𝑝𝑙𝑒𝑡𝑒
Proof:

⇒ : 𝐴, 𝑑 𝑖𝑠 𝑐𝑜𝑚𝑝𝑙𝑒𝑡𝑒 ⇔ ∀ 𝑝𝑛 ⊂ 𝐴, 𝑝𝑛 𝐶𝑎𝑢𝑐𝑕𝑦 ⇒ ∃𝑝 ∈ 𝐴/ lim⁡𝑝𝑛 = 𝑝

Let 𝑝𝑛 be a Cauchy sequence of 𝐴, 𝑑

𝑝𝑛 ⊂ 𝐴 ⊂ 𝐸
𝑝𝑛 𝑖𝑠 𝐶𝑎𝑢𝑐𝑕𝑦 𝑖𝑛 𝐴 ⇒ 𝑝𝑛 𝑖𝑠 𝐶𝑎𝑢𝑐𝑕𝑦 𝑖𝑛 (𝐸, 𝑑) ∃𝑝 ∈ 𝐸/ lim⁡𝑝𝑛 = 𝑝 ⇒
𝐸,𝑑 𝑖𝑠
𝐴 𝑎𝑛𝑑 𝐸 𝑕𝑎𝑣𝑒 𝑠𝑎𝑚𝑒 𝑚𝑒𝑡𝑟𝑖𝑐 𝑐𝑜𝑚𝑝𝑙𝑒𝑡𝑒

⇒ 𝑝𝑛 𝑖𝑠 𝑎 𝑐𝑜𝑛𝑣𝑒𝑟𝑔𝑒𝑛𝑡 𝑠𝑒𝑞𝑢𝑒𝑛𝑐𝑒 𝑎𝑛𝑑 lim⁡𝑝𝑛 = 𝑝

𝐴 𝑖𝑠 𝑐𝑙𝑜𝑠𝑒𝑑 ⇒ ∀ 𝑝𝑛 ⊂ 𝐴, ( 𝑝𝑛 𝑐𝑜𝑛𝑣𝑒𝑟𝑔𝑒𝑛𝑡 ⇒ lim 𝑝𝑛 ∈ 𝐴) ⇒ lim 𝑝𝑛 = 𝑝 ∈ 𝐴
𝐵𝑦 𝑕𝑦𝑝𝑜𝑡𝑕𝑒𝑠𝑖𝑠: 𝑝𝑛 ⊂ 𝐴

∴ ∀ 𝑝𝑛 ⊂ 𝐴, 𝑝𝑛 𝐶𝑎𝑢𝑐𝑕𝑦 ⇒ ∃𝑝 ∈ 𝐴/ lim⁡𝑝𝑛 = 𝑝 ⇒ 𝐴, 𝑑 𝑖𝑠 𝑐𝑜𝑚𝑝𝑙𝑒𝑡𝑒

⇐ : Assume otherwise that A is not closed

𝑡𝑕.𝑏𝑒𝑓𝑜𝑟𝑒
⇒ ∃ 𝑝𝑛 ⊂ 𝐴, ( 𝑝𝑛 𝑐𝑜𝑛𝑣𝑒𝑟𝑔𝑒𝑛𝑡 𝑎𝑛𝑑 lim 𝑝𝑛 ∉ 𝐴)
⇒ ∃ 𝑝𝑛 ⊂ 𝐴, ( 𝑝𝑛 𝐶𝑎𝑢𝑐𝑕𝑦 𝑎𝑛𝑑 lim 𝑝𝑛 ∉ 𝐴) ⇒
𝑝𝑛 𝑐𝑜𝑛𝑣𝑒𝑟𝑔𝑒𝑛𝑡 ⇒ 𝑝𝑛 𝑖𝑠 𝐶𝑎𝑢𝑐𝑕𝑦
⇒ 𝐴, 𝑑 𝑖𝑠 𝑛𝑜𝑡 𝑐𝑜𝑚𝑝𝑙𝑒𝑡𝑒 (absurd)


Theorem: (ℝ IS COMPLETE)

∀𝐴 ⊂ ℝ , 𝐴 ≠ ∅ 𝑎𝑛𝑑 𝐴 𝑖𝑠 𝑏𝑑𝑑 𝑓𝑟𝑜𝑚 𝑎𝑏𝑜𝑣𝑒 ⇒ ∃𝑎 = sup 𝐴 ∈ ℝ ⇔

⇔ ∀ 𝐶𝑎𝑢𝑐𝑕𝑦 𝑝𝑛 ⊂ ℝ, ∃𝑝 ∈ ℝ/ lim⁡𝑝𝑛 = 𝑝

Proof:

⇒ : Let 𝑎𝑛 be a Cauchy sequence of 𝔼1 .

Consider 𝑆 = 𝑥 ∈ ℝ 𝑥 ≤ 𝑎𝑛 𝑓𝑜𝑟 𝑎𝑛 𝑖𝑛𝑓𝑖𝑛𝑖𝑡𝑒 𝑛𝑢𝑚𝑏𝑒𝑟 𝑜𝑓 𝑛 ∈ ℕ

Claim 1: ∃𝑎 = sup 𝑆 ∈ ℝ

Proof:
𝑎𝑛 is Cauchy ⇒ 𝑎𝑛 𝑖𝑠 𝑏𝑑𝑑 ⇒ ∃𝑕 ∈ ℝ ∀𝑛 ∈ ℕ, 𝑕 ≤ 𝑎𝑛 ⇒ ∃𝑕 ∈ ℝ 𝑕 ≤ 𝑎𝑛 for an infinite
number of 𝑛 ∈ ℕ ⇒ 𝑕 ∈ 𝑆 ⇒ 𝑆 ≠ ∅ (𝑖)

𝑎𝑛 𝑖𝑠 𝐶𝑎𝑢𝑐𝑕𝑦 ⇒ 𝑎𝑛 𝑖𝑠 𝑏𝑑𝑑 ⇒ ∃𝑘 ∈ ℝ ∀𝑛 ∈ ℕ, 𝑎𝑛 ≤ 𝑘
⇒ ∃𝑘 ∈ ℝ ∀𝑥 ∈ 𝑆, 𝑥 ≤ 𝑘 ⇒
𝐵𝑦 𝑑𝑒𝑓. 𝑜𝑓 𝑆: ∀𝑥 ∈ 𝑆, 𝑥 ≤ 𝑎𝑛 ∀𝑛 ∈ ℕ′ ⊂ ℕ
⇒ 𝑘 𝑖𝑠 𝑎𝑛 𝑢𝑏 𝑜𝑓 𝑆 ⇒ 𝑆 𝑖𝑠 𝑏𝑑𝑑 𝑓𝑟𝑜𝑚 𝑎𝑏𝑜𝑣𝑒 (𝑖𝑖)

By definition of S: 𝑆 ⊂ ℝ (𝑖𝑖𝑖)

From 𝑖 , 𝑖𝑖 𝑎𝑛𝑑 𝑖𝑖𝑖 ∃𝑎 = sup 𝑆 ∈ ℝ 

𝐻𝑦𝑝𝑜𝑡 𝑕𝑒𝑠𝑖𝑠

Lemma Claim 2: 𝑖𝑓 𝑥 ∈ 𝑆 ⇒ ∀𝑦 ∈ ℝ, (𝑦 ≤ 𝑥 ⇒ 𝑦 ∈ 𝑆)

Proof: 𝑥 ∈ 𝑆 ⇒ 𝑥 ≤ 𝑎𝑛 𝑓𝑜𝑟 𝑎𝑛 𝑖𝑛𝑓𝑖𝑛𝑖𝑡𝑒 𝑛𝑢𝑚𝑏𝑒𝑟 𝑜𝑓 𝑛 ∈ ℕ

⇒
𝑦≤𝑥
⇒ 𝑦 ≤ 𝑎𝑛 𝑓𝑜𝑟 𝑎𝑛 𝑖𝑛𝑓𝑖𝑛𝑖𝑡𝑒 𝑛𝑢𝑚𝑏𝑒𝑟 𝑜𝑓 𝑛 ∈ ℕ ⇒ 𝑦 ∈ 𝑆 

Claim 2: lim 𝑎𝑛 = 𝑎

Proof: lim 𝑎𝑛 = 𝑎 ⇔ ∀𝜀 > 0, ∃𝑁 ∈ ℕ / ∀𝑛 ≥ 𝑁, 𝑎𝑛 − 𝑎 < 𝜀

𝜀
Pick up any 𝜀 > 0 ⇒ >0
2

𝜀
>0 𝜀
2 ⇒ ∃𝑁 ∈ ℕ / ∀𝑛, 𝑚 ≥ 𝑁, 𝑎𝑛 − 𝑎𝑚 < (𝑖)
𝑎𝑛 𝑖𝑠 𝐶𝑎𝑢𝑐𝑕𝑦 2

𝜀 𝜀
𝜀 ∃𝑥 ∈ 𝑆 / 𝑎 − <𝑥≤𝑎 𝑎− ∈𝑆
>0 2 𝐿𝑒𝑚𝑚𝑎 2
2 ⇒ 𝐶𝑙𝑎𝑖𝑚 2
𝑎 = sup 𝑆 𝜀
𝑎+ ∉𝑆
2
 𝜀 𝜀
𝑎− ∈ 𝑆 ⇒ 𝑎 − ≤ 𝑎𝑛 , 𝑓𝑜𝑟 𝑎𝑛 𝑖𝑛𝑓𝑖𝑛𝑖𝑡𝑒 𝑛𝑢𝑚𝑏𝑒𝑟 𝑜𝑓𝑛 ∈ ℕ
2 2 ⇒
𝜀 𝜀
𝑎 + ∉ 𝑆 ⇒ 𝑎 + ≤ 𝑎𝑛 , 𝑓𝑜𝑟 𝑎 𝑓𝑖𝑛𝑖𝑡𝑒 𝑛𝑢𝑚𝑏𝑒𝑟 𝑜𝑓𝑛 ∈ ℕ
2 2

𝜀 𝜀
⇒ 𝑇𝑕𝑒𝑟𝑒 𝑒𝑥𝑖𝑠𝑡 𝑖𝑛𝑓𝑖𝑛𝑖𝑡𝑒𝑙𝑦 𝑒𝑙𝑒𝑚𝑒𝑛𝑡𝑠 𝑎𝑛 𝑖𝑛 𝑡𝑕𝑒 𝑖𝑛𝑡𝑒𝑟𝑣𝑎𝑙 𝑎 − , 𝑎 + ⇒
2 2
𝜀 𝜀 𝜀
⇒ ∃𝑚 ≥ 𝑁 /𝑎𝑚 ∈ 𝑎 − , 𝑎 + ⇒ ∃𝑚 ≥ 𝑁 / 𝑎𝑚 − 𝑎 < (𝑖𝑖)
2 2 2
𝜀 𝜀
𝐹𝑟𝑜𝑚 𝑖 𝑎𝑛𝑑 𝑖𝑖 : ∃𝑁 ∈ ℕ/∃𝑚 ≥ 𝑁, ∀𝑛 ≥ 𝑁, 𝑎𝑛 − 𝑎𝑚 + 𝑎𝑚 − 𝑎 < + =𝜀
2 2 ⇒
𝐵𝑦 ∆ − 𝐼𝑛𝑒𝑞.: 𝑎𝑛 − 𝑎 ≤ 𝑎𝑛 − 𝑎𝑚 + 𝑎𝑚 − 𝑎

⇒ ∃𝑁 ∈ ℕ/∀𝑛 ≥ 𝑁, 𝑎𝑛 − 𝑎 < 𝜀
∴ ∀𝜀 > 0, ∃𝑁 ∈ ℕ / ∀𝑛 ≥ 𝑁, 𝑎𝑛 − 𝑎 < 𝜀 ⇒ lim 𝑎𝑛 = 𝑎 

∴ ∀ 𝐶𝑎𝑢𝑐𝑕𝑦 𝑝𝑛 ⊂ ℝ, ∃𝑝 ∈ ℝ/ lim⁡𝑝𝑛 = 𝑝

⇐ :
Theorem: (𝔼𝒏 IS COMPLETE)

𝐶𝑜𝑛𝑠𝑖𝑑𝑒𝑟 𝑡𝑕𝑒 𝑚𝑒𝑡𝑟𝑖𝑐 𝑠𝑝𝑎𝑐𝑒𝑠 𝔼j . Then, ∀𝑗 ∈ ℕ, 𝑗 ≥ 2, 𝔼𝑗 𝑖𝑠 𝑐𝑜𝑚𝑝𝑙𝑒𝑡𝑒.

Proof: 𝔼𝑗 𝑖𝑠 𝑐𝑜𝑚𝑝𝑙𝑒𝑡𝑒 ⇔ ∀ 𝑝𝑛 ⊂ 𝔼𝑗 , 𝑝𝑛 𝐶𝑎𝑢𝑐𝑕𝑦 ⇒ ∃𝑝 ∈ 𝔼 𝑗 / lim⁡𝑝𝑛 = 𝑝

Pick up any Cauchy sequence 𝑝𝑛 ⊂ 𝔼𝑗

𝑝𝑛 ⊂ 𝔼𝑗 ⇒ ∀𝑛 ∈ ℕ, 𝑝𝑛 = 𝑎1,𝑛 , … , 𝑎𝑗 ,𝑛 ∈ 𝔼𝑗 ⇒
⇒ ∀𝑛 ∈ ℕ, 𝑝𝑛 = 𝑎1,𝑛 , … , 𝑎𝑗 ,𝑛 ∀𝑖 = 1, … , 𝑗 , 𝑎𝑖,𝑛 ∈ 𝔼1 = ℝ

𝑝𝑛 𝑖𝑠 𝐶𝑎𝑢𝑐𝑕𝑦 ⇒ ∀𝜀 > 0, ∃𝑁 ∈ ℕ / ∀𝑛, 𝑚 ≥ 𝑁, 𝑑(𝑝𝑛 , 𝑝𝑚 ) < 𝜀 ⇒

2 2
⇒ ∀𝜀 > 0, ∃𝑁 ∈ ℕ / ∀𝑛, 𝑚 ≥ 𝑁, 𝑎1,𝑛 − 𝑎1,𝑚 + ⋯ . 𝑎𝑗 ,𝑛 − 𝑎𝑗 ,𝑚 <𝜀
2 2 2
𝑎1,𝑛 − 𝑎1,𝑚 = 𝑎1,𝑛 − 𝑎1,𝑚 ≤ 𝑎1,𝑛 − 𝑎1,𝑚 + ⋯ . 𝑎𝑗 ,𝑚 ⇒
⋮
2 2 2
𝑎𝑗 ,𝑛 − 𝑎𝑗 ,𝑚 = 𝑎𝑗 ,𝑛 − 𝑎𝑗 ,𝑚 ≤ 𝑎1,𝑛 − 𝑎1,𝑚 + ⋯ . 𝑎𝑗 ,𝑚

∀𝜀 > 0, ∃𝑁 ∈ ℕ / ∀𝑛, 𝑚 ≥ 𝑁, 𝑎1,𝑛 − 𝑎1,𝑚 < 𝜀

⇒ ⋮ ⇒
∀𝜀 > 0, ∃𝑁 ∈ ℕ / ∀𝑛, 𝑚 ≥ 𝑁, 𝑎𝑗 ,𝑛 − 𝑎𝑗 ,𝑚 < 𝜀

⇒ 𝐸𝑎𝑐𝑕 𝑜𝑛𝑒 𝑜𝑓 𝑡𝑕𝑒 𝑟𝑒𝑎𝑙 𝑠𝑒𝑞𝑢𝑒𝑛𝑐𝑒𝑠 𝑎1,𝑛 , … , 𝑎𝑗 ,𝑛 𝑎𝑟𝑒 𝐶𝑎𝑢𝑐𝑕𝑦. 𝑆𝑖𝑛𝑐𝑒 𝑤𝑒 𝑗𝑎𝑣𝑒 𝑝𝑟𝑜𝑣𝑒𝑑 ℝ 𝑖𝑠
complete: ∃𝐴1 , … , 𝐴𝑗 ∈ ℝ lim 𝑎1,𝑛 = 𝐴1 , … , lim 𝑎𝑗 ,𝑛 = 𝐴𝑗

Call 𝐴 = 𝐴1 , … , 𝐴𝑗 ∈ 𝔼𝑗

Claim: lim⁡𝑝𝑛 = 𝐴

2 2
Proof: lim 𝑝𝑛 = 𝐴 ⇔ ∀𝜀 > 0, ∃𝑅 ∈ ℕ / ∀𝑛 ≥ 𝑁, 𝑎1,𝑛 − 𝐴1 + ⋯ + 𝑎𝑗 ,𝑛 − 𝐴𝑗 <𝜀

𝑃𝑖𝑐𝑘 𝑢𝑝 𝑎𝑛𝑦 𝜀 > 0 𝜀

⇒ >0
𝑗≥2⇒ 𝑗>0 𝑗 𝜀
⇒ ∀𝑖 = 1, … , 𝑗 , ∃𝑅𝑖 ∈ ℕ/ ∀𝑛 ≥ 𝑁, 𝑎𝑖,𝑛 − 𝐴𝑖 <
𝑗 ⇒
∀𝑖 = 1, … , 𝑗 , lim 𝑎𝑖,𝑛 = 𝐴𝑖
𝑇𝑎𝑘𝑒 𝑅 = max 𝑅𝑖 / 𝑖 = 1, … , 𝑗
𝜀
⇒ ∃𝑅 ∈ ℕ/∀𝑖 = 1, … , 𝑗 , ∀𝑛 ≥ 𝑁, 𝑎𝑖,𝑛 − 𝐴𝑖 <
𝑗
2 2 ⇒
2 2 𝜀 𝜀 𝜀2 𝜀2
𝑎1,𝑛 − 𝐴1 + ⋯ + 𝑎𝑗 ,𝑛 − 𝐴𝑗 < +⋯+ = +⋯+ = 𝜀2 = 𝜀 = 𝜀
𝑗 𝑗 𝑗 𝑗
 2 2
⇒ ∃𝑅 ∈ ℕ / ∀𝑛 ≥ 𝑁, 𝑎1,𝑛 − 𝐴1 + ⋯ + 𝑎𝑗 ,𝑛 − 𝐴𝑗 <𝜀
2 2
∴ ∀𝜀 > 0, ∃𝑅 ∈ ℕ/∀𝑛 ≥ 𝑁, 𝑎1,𝑛 − 𝐴1 + ⋯ + 𝑎𝑗 ,𝑛 − 𝐴𝑗 < 𝜀 ⇒ lim 𝑝𝑛 = 𝐴 

∴ ∀ 𝑝𝑛 ⊂ 𝔼𝑗 , 𝑝𝑛 𝐶𝑎𝑢𝑐𝑕𝑦 ⇒ ∃𝑝 ∈ 𝔼𝑗 / lim⁡𝑝𝑛 = 𝑝 ⇒ 𝔼𝑗 𝑖𝑠 𝑐𝑜𝑚𝑝𝑙𝑒𝑡𝑒 

Theorem (Quiz 7):

𝐶𝑜𝑛𝑠𝑖𝑑𝑒𝑟 𝐴 ≠ ∅ 𝑎𝑛𝑑 𝛿 𝑡𝑕𝑒 𝑑𝑖𝑠𝑐𝑟𝑒𝑡𝑒 𝑚𝑒𝑡𝑟𝑖𝑐 𝑖𝑛 𝐴. 𝑇𝑕𝑒𝑛, 𝐴, 𝛿 𝑖𝑠 𝑎 𝑐𝑜𝑚𝑝𝑙𝑒𝑡𝑒 𝑚𝑒𝑡𝑟𝑖𝑐 𝑠𝑝𝑎𝑐𝑒.

Proof: 𝐴 𝑖𝑠 𝑐𝑜𝑚𝑝𝑙𝑒𝑡𝑒 ⇔ ∀ 𝑝𝑛 ⊂ 𝐴, 𝑝𝑛 𝐶𝑎𝑢𝑐𝑕𝑦 ⇒ ∃𝑝 ∈ 𝐴/ lim⁡𝑝𝑛 = 𝑝

Pick up any Cauchy sequence 𝑝𝑛 ⊂ 𝐴

𝑝𝑛 𝐶𝑎𝑢𝑐𝑕𝑦 ⇒ ∀𝜀 > 0, ∃𝑁 ∈ ℕ/∀𝑛, 𝑚 ≥ 𝑁, 𝛿(𝑝𝑛 , 𝑝𝑚 ) < 𝜀

𝜀=1/2
1
⇒ ∃𝑁 ∈ ℕ/∀𝑛, 𝑚 ≥ 𝑁, 𝛿(𝑝𝑛 , 𝑝𝑚 ) < ⇒ ∃𝑁 ∈ ℕ/∀𝑛, 𝑚 ≥ 𝑁, 𝛿(𝑝𝑛 , 𝑝𝑚 ) = 0 ⇒
2
⇒ ∃𝑁 ∈ ℕ/∀𝑛, 𝑚 ≥ 𝑁, 𝑝𝑛 = 𝑝𝑚
⇒ ∃𝑁 ∈ ℕ/∀𝑛 ≥ 𝑁, 𝑝𝑛 = 𝑝𝑘 ∈ 𝐴 𝑏𝑒𝑐𝑎𝑢𝑠𝑒 𝑝𝑛 ⊂ 𝐴 (𝑖)
𝑇𝑎𝑘𝑒 𝑘 ≥ 𝑁, 𝑘 𝑓𝑖𝑥𝑒𝑑

Claim: lim 𝑝𝑛 = 𝑝𝑘

Proof: Pick up any 𝜀 > 0.

Because 𝑖 : ∃𝑁 ∈ ℕ/∀𝑛 ≥ 𝑁, 𝑝𝑛 = 𝑝𝑘 ⇒ ∃𝑁 ∈ ℕ/∀𝑛 ≥ 𝑁, 𝛿(𝑝𝑛 , 𝑝𝑘 ) = 0 < 𝜀
∴ ∀𝜀 > 0, 𝑁 ∈ ℕ/∀𝑛 ≥ 𝑁, 𝛿(𝑝𝑛 , 𝑝𝑘 ) < 𝜀 ⇒ lim 𝑝𝑛 = 𝑝𝑘

∴ 𝑝𝑛 ⊂ 𝐴, 𝑝𝑛 𝐶𝑎𝑢𝑐𝑕𝑦 ⇒ ∃𝑝 ∈ 𝐴/ lim⁡𝑝𝑛 = 𝑝 ⇒ 𝐴 𝑖𝑠 𝑐𝑜𝑚𝑝𝑙𝑒𝑡𝑒 

8. COMPACT SETS
𝐎𝐏𝐄𝐍 𝐂𝐎𝐕𝐄𝐑: Given a metric space 𝐸, 𝑑 , 𝑎𝑛𝑑 𝐴 ⊂ 𝐸 𝐴 𝑐𝑎𝑛 𝑏𝑒 ∅ , we say that a collection
𝑈𝜆 / 𝜆 ∈ 𝐼 𝑜𝑓 𝑜𝑝𝑒𝑛 𝑠𝑒𝑡𝑠 𝑜𝑓 𝐸 𝑖𝑠 𝑎𝑛 𝑜𝑝𝑒𝑛 𝑐𝑜𝑣𝑒𝑟 𝑜𝑓 𝐴 iff 𝐴 ⊂ 𝜆∈𝐼 𝑈𝜆 .

𝐅𝐈𝐍𝐈𝐓𝐄 𝐎𝐏𝐄𝐍 𝐂𝐎𝐕𝐄𝐑: Given a metric space 𝐸, 𝑑 , 𝑎𝑛𝑑 𝐴 ⊂ 𝐸 𝐴 𝑐𝑎𝑛 𝑏𝑒 ∅ , we say that a
collection 𝑈𝜆 / 𝜆 ∈ 𝐼 𝑜𝑓 𝑜𝑝𝑒𝑛 𝑠𝑒𝑡𝑠 𝑜𝑓 𝐸 𝑖𝑠 𝑎 𝑓𝑖𝑛𝑖𝑡𝑒 𝑜𝑝𝑒𝑛 𝑐𝑜𝑣𝑒𝑟 𝑜𝑓 𝐴 iff 𝐴 ⊂ 𝜆∈𝐼 𝑈𝜆 𝑎𝑛𝑑 I is
a finite set.

𝐒𝐔𝐁 − 𝐎𝐏𝐄𝐍 𝐂𝐎𝐕𝐄𝐑: Given a metric space 𝐸, 𝑑 , 𝐴 ⊂ 𝐸, 𝑎𝑛𝑑 𝑈𝜆 / 𝜆 ∈ 𝐼 is an open cover of

A. 𝐺𝑖𝑣𝑒𝑛 𝐽 ⊂ 𝐼, we say that 𝑈𝜆 / 𝜆 ∈ 𝐽 is a sub-open cover of A iff 𝐴 ⊂ 𝜆∈𝐽 𝑈𝜆 .

𝐂𝐎𝐌𝐏𝐀𝐂𝐓 𝐒𝐄𝐓: Given a metric space 𝐸, 𝑑 𝑎𝑛𝑑 𝐴 ⊂ 𝐸 𝐴 𝑐𝑎𝑛 𝑏𝑒 ∅ . We say A is compact

(𝑐𝑝𝑡) iff for any open cover of A, there exist a finite sub-open cover of A .

Observation: ℝ = 𝔼1 𝑖𝑠 𝑛𝑜 𝑐𝑜𝑚𝑝𝑎𝑐𝑡

Proof: Consider 𝑈𝜆 = (𝜆, 𝜆 + 2)/ 𝜆 ∈ ℤ

Claim: 𝑈𝜆 = (𝜆, 𝜆 + 2)/ 𝜆 ∈ ℤ is an open cover of ℝ

Proof: Trivial

Claim: 𝐿𝑒𝑡 𝑈𝑛 1 , … , 𝑈𝑛 𝑘 𝑏𝑒 𝑎𝑛𝑦 𝑓𝑖𝑛𝑖𝑡𝑒 𝑐𝑜𝑙𝑙𝑒𝑐𝑡𝑖𝑜𝑛 𝑜𝑓 𝑒𝑙𝑒𝑚𝑒𝑛𝑡𝑠 𝑜𝑓 𝑈𝜆 = (𝜆, 𝜆 + 2)/ 𝜆 ∈ ℤ .

𝑇𝑎𝑘𝑒 𝑀 = 𝑚𝑎𝑥 𝑛1 + 2 , … . , 𝑛𝑘 + 2 ∈ ℝ. 𝑇𝑕𝑒𝑛, 𝑀 ∉ 𝑖=𝑘 𝑖=1 𝑈𝑛 𝑖

Proof: .
𝑖=𝑘
𝑆𝑢𝑝𝑝𝑜𝑠𝑒 𝑡𝑕𝑎𝑡 𝑀 ∈ 𝑖=1 𝑈𝑛 𝑖
⇒ ∃𝑖 = 1, … , 𝑘/ 𝑀 ∈ 𝑈𝑛 𝑖 ⇒ ∃𝑖 = 1, … , 𝑘/ 𝑀 < 𝑛𝑖 + 2
𝑎𝑏𝑠𝑢𝑟𝑑𝑜
𝑀 = 𝑚𝑎𝑥 𝑛1 + 2 , … . , 𝑛𝑘 + 2 ⇒ 𝑀 ≥ 𝑛𝑖 + 2, ∀𝑖 = 1, … , 𝑘


Lemma: Given a metric space 𝐸, 𝑑 . ∅ 𝑖𝑠 𝑐𝑜𝑚𝑝𝑎𝑐𝑡.

Proof: Pick up any open cover 𝑈𝜆 / 𝜆 ∈ 𝐼 of ∅. Take any 𝜆 ∈ 𝐼. 𝑆𝑖𝑛𝑐𝑒 ∅ ⊂ 𝑈𝜆 , 𝑈𝜆 is a finite

sub-open cover of ∅. 𝑇𝑕𝑒𝑛, ∅ 𝑖𝑠 𝑐𝑜𝑚𝑝𝑎𝑐𝑡.


Theorem: Given a metric space 𝐸, 𝑑 . 𝐴 ⊂ 𝐸 𝑖𝑠 𝑐𝑜𝑚𝑝𝑎𝑐𝑡 ⇒ 𝐴 𝑖𝑠 𝑏𝑑𝑑.

Proof:

Case 1: 𝐴 = ∅ Trivial.

Case 2: 𝐴 ≠ ∅ Consider 𝐵(𝑎, 1)/𝑎 ∈ 𝐴

𝐴 𝑖𝑠 𝑐𝑜𝑚𝑝𝑎𝑐𝑡 ⇒ ∀ 𝑜𝑝𝑒𝑛 𝑐𝑜𝑣𝑒𝑟 𝑜𝑓 𝐴, ∃ 𝑓𝑖𝑛𝑖𝑡𝑒 𝑠𝑢𝑏 − 𝑜𝑝𝑒𝑛 𝑐𝑜𝑣𝑒𝑟 𝑜𝑓 𝐴

⇒
𝐵(𝑎, 1)/𝑎 ∈ 𝐴 𝑖𝑠 𝑎𝑛 𝑜𝑝𝑒𝑛 𝑐𝑜𝑣𝑒𝑟 𝑜𝑓 𝐴 (∗)
⇒ ∃ 𝑎1 , … , 𝑎𝑘 ∈ 𝐴/ 𝐵(𝑎𝑖 , 1)/𝑖 = 1, … , 𝑘 𝑖𝑠 𝑎 𝑓𝑖𝑛𝑖𝑡𝑒 𝑠𝑢𝑏 − 𝑜𝑝𝑒𝑛 𝑐𝑜𝑣𝑒𝑟 𝑜𝑓 𝐴 ⇒
⇒ ∃ 𝑎1 , … , 𝑎𝑘 ∈ 𝐴/𝐴 ⊂ 𝑖=𝑘𝑖=1 𝐵(𝑎𝑖 , 1) (i)

Pick up any 𝑝 ∈ 𝐴. Define 𝑟 = max 𝑑(𝑝, 𝑎𝑖 )/𝑖 = 1, … , 𝑘

Claim: 𝐴 ⊂ 𝐵(𝑝, 𝑟 + 1)

Proof: 𝑃𝑖𝑐𝑘 𝑢𝑝 𝑎𝑛𝑦 𝑎 ∈ 𝐴 ∃𝑗 = 1, … , 𝑘 / 𝑎 ∈ 𝐵 𝑎𝑗 , 1 ⇒ 𝑑(𝑎, 𝑎𝑗 ) < 1

(𝑖)
⇒
𝑟 = max 𝑑(𝑝, 𝑎𝑖 )/𝑖 = 1, … , 𝑘 𝑑 𝑝, 𝑎𝑗 ≤ 𝑟
𝑗 ∈ 1,…,𝑘
⇒ 𝑑 𝑎, 𝑎𝑗 + 𝑑 𝑝, 𝑎𝑗 < 1 + 𝑟
⇒ 𝑑 𝑎, 𝑝 < 1 + 𝑟 ⇒ 𝑎 ∈ 𝐵(𝑝, 𝑟 + 1) 
𝐵𝑦 ⊿ − 𝐼𝑛𝑒𝑞. : 𝑑 𝑎, 𝑝 ≤ 𝑑 𝑎, 𝑎𝑗 + 𝑑(𝑝, 𝑎𝑗 )

∴ 𝐴 ⊂ 𝐵 𝑝, 𝑟 + 1 ⇒ 𝐴 𝑖𝑠 𝑏𝑑𝑑.

Claim (*): 𝐵(𝑎, 1)/𝑎 ∈ 𝐴 is an open cover of A.

Proof: Since any open ball is open, ∀𝑎 ∈ 𝐴, 𝐵 𝑎, 1 𝑖𝑠 𝑜𝑝𝑒𝑛. Then, 𝐵(𝑎, 1)/𝑎 ∈ 𝐴 is a collection
of open sets.

Pick up any 𝑎 ∈ 𝐴 ⇒ 𝑎 ∈ 𝐵 𝑎, 1 ⇒ 𝑎 ∈ 𝑎∈𝐴 𝐵 𝑎, 1

∴ 𝐵(𝑎, 1)/𝑎 ∈ 𝐴 is an open cover of A. 



Theorem: Given a metric space 𝐸, 𝑑 . 𝐴 ⊂ 𝐸 𝑖𝑠 𝑓𝑖𝑛𝑖𝑡𝑒 ⇒ 𝐴 𝑖𝑠 𝑐𝑜𝑚𝑝𝑎𝑐𝑡

Proof:

Case 1: 𝐴 = ∅ It is true by lemma.

Case 2: 𝐴 ≠ ∅ 𝐴 = 𝑎1 , … , 𝑎𝑛 .

Pick up any open cover 𝑈𝜆 / 𝜆 ∈ 𝐼 of A.

. 𝑈𝜆 / 𝜆 ∈ 𝐼 𝑖𝑠 𝑎𝑛 𝑜𝑝𝑒𝑛 𝑐𝑜𝑣𝑒𝑟 𝑜𝑓 𝐴 ⇒ 𝐴 ⊂ 𝜆∈𝐼 𝑈𝜆 ⇒ ∀𝑖 = 1, … , 𝑘, 𝑎𝑖 ∈ 𝜆∈𝐼 𝑈𝜆 ⇒

⇒ ∀𝑖 = 1, … , 𝑘, ∃𝜆𝑖 ∈ 𝐼 / 𝑎𝑖 ∈ 𝑈𝜆 𝑖 ⇒ ∃𝑈𝜆 1 , … , 𝑈𝜆 1 ∈ 𝑈𝜆 / 𝜆 ∈ 𝐼 /𝐴 ⊂ 𝑖=𝑘
𝑖=1 𝑈𝜆 𝑖
⇒ Uλ 1 , … , Uλ 1 is a finite sub-open cover of A

∴ For any open cover of A, there exist a finite sub-open cover of A ⇒ 𝐴 𝑖𝑠 𝑐𝑜𝑚𝑝𝑎𝑐𝑡 .


Theorem (HW): Given a m.e. 𝐸, 𝑑 . 𝑝𝑛 ⊂ 𝐸 / lim 𝑝𝑛 = 𝑝 ∈ 𝐸 ⇒ 𝑝𝑛 ∪ 𝑝 𝑖𝑠 𝑐𝑜𝑚𝑝𝑎𝑐𝑡

Proof: Pick up any open cover 𝑈𝜆 / 𝜆 ∈ 𝐼 of 𝑝𝑛 ∪ 𝑝

. 𝑈𝜆 / 𝜆 ∈ 𝐼 𝑖𝑠 𝑎𝑛 𝑜𝑝𝑒𝑛 𝑐𝑜𝑣𝑒𝑟 𝑜𝑓 𝑝𝑛 ∪ 𝑝 ⇒ 𝑝𝑛 ∪ 𝑝 ⊂ 𝜆∈𝐼 𝑈𝜆 ⇒

 ∀𝑛 ∈ ℕ, ∃𝜆𝑛 ∈ 𝐼 /𝑝𝑛 ∈ 𝑈𝜆 𝑛 (𝑖)
⇒
∃𝜆0 ∈ 𝐼 /𝑝 ∈ 𝑈𝜆 0 (𝑖𝑖)

𝑝 ∈ 𝑈𝜆 0 𝑎𝑛𝑑 𝑈𝜆 0 𝑖𝑠 𝑜𝑝𝑒𝑛 ⇒ ∃𝐵(𝑝, 𝑟)/𝐵(𝑝, 𝑟) ⊂ 𝑈𝜆 0 (𝑖𝑖𝑖)

lim 𝑝𝑛 = 𝑝
⇒ ∃𝑁 ∈ ℕ / ∀𝑛 ≥ 𝑁, 𝑝𝑛 ∈ 𝐵 𝑝, 𝑟 (𝑖𝑣)
𝑟>0

Consider 𝑈𝜆 0 , 𝑈𝜆 1 , … , 𝑈𝜆 𝑁 −1 ⊂ 𝑈𝜆 / 𝜆 ∈ 𝐼

𝑖=𝑁−1
Claim: 𝑝𝑛 ∪ 𝑝 ⊂ 𝑖=0 𝑈𝜆 𝑖

Proof: Pick up any 𝑥 ∈ 𝑝𝑛 ∪ 𝑝

𝑖=𝑁−1
Case 1: If 𝑥 = 𝑝 𝑥 ∈ 𝑈𝜆 0 ⇒ 𝑥 ∈ 𝑖=0 𝑈𝜆 𝑖
(𝑖𝑖 )
𝑖=𝑁−1
Case 2: If 𝑥 = 𝑝𝑛 / 𝑛 ≥ 𝑁 𝑥 ∈ 𝐵 𝑝, 𝑟 𝑥 ∈ 𝑈𝜆 0 ⇒ 𝑥 ∈ 𝑖=0 𝑈𝜆 𝑖
(𝑖𝑣) (𝑖𝑖𝑖 )
𝑖=𝑁−1
Case 3: If 𝑥 = 𝑝𝑛 /𝑛 ≤ 𝑁 − 1 𝑥 ∈ 𝑈𝜆 𝑛 / 𝑛 ≤ 𝑁 − 1 ⇒ 𝑥 ∈ 𝑖=0 𝑈𝜆 𝑖
𝑖

𝑖=𝑁−1 𝑖=𝑁−1
.∴ ∀𝑥, 𝑥 ∈ 𝑝𝑛 ∪ 𝑝 ⇒ 𝑥 ∈ 𝑖=0 𝑈𝜆 𝑖 ⇒ 𝑝𝑛 ∪ 𝑝 ⊂ 𝑖=0 𝑈𝜆 𝑖

Then, 𝑈𝜆 0 , 𝑈𝜆 1 , … , 𝑈𝜆 𝑁 −1 𝑖𝑠 𝑎 𝑠𝑢𝑏 − 𝑜𝑝𝑒𝑛 𝑐𝑜𝑣𝑒𝑟 𝑜𝑓 𝑝𝑛 ∪ 𝑝 ⇒ 𝑝𝑛 ∪ 𝑝 𝑖𝑠 𝑐𝑜𝑚𝑝𝑎𝑐𝑡.



Theorem: Given a compact m.e. 𝐸, 𝑑 . 𝐴 ⊂ 𝐸. 𝐴 𝑖𝑠 𝑐𝑙𝑜𝑠𝑒𝑑 ⇒ 𝐴 𝑖𝑠 𝑐𝑜𝑚𝑝𝑎𝑐𝑡

Proof: Pick up any open cover 𝑈𝜆 / 𝜆 ∈ 𝐼 of 𝐴

𝐴 𝑖𝑠 𝑐𝑙𝑜𝑠𝑒𝑑 ⇒ 𝐴𝐶 𝑖𝑠 𝑜𝑝𝑒𝑛.

𝑈𝜆 / 𝜆 ∈ 𝐼 is an open cover of 𝐴 ⇒ 𝐴 ⊂ 𝜆∈𝐼 𝑈𝜆 ⇒ 𝐴 ∪ 𝐴𝐶 ⊂ 𝜆∈𝐼 𝑈𝜆 ∪ 𝐴𝐶 ⇒

⇒𝐸⊂ 𝐶
𝜆∈𝐼 𝑈𝜆 ∪ 𝐴
𝐶
𝐴 𝑖𝑠 𝑐𝑙𝑜𝑠𝑒𝑑 ⇒ 𝐴𝐶 𝑖𝑠 𝑜𝑝𝑒𝑛 ⇒ 𝑈𝜆 / 𝜆 ∈ 𝐼 ∪ 𝐴 𝑖𝑠 𝑎𝑛 𝑜𝑝𝑒𝑛 𝑐𝑜𝑣𝑒𝑟 𝑜𝑓 𝐸.
𝑈𝜆 𝑖𝑠 𝑜𝑝𝑒𝑛, ∀𝜆 ∈ 𝐼

Then, since E is compact, there exist a finite sub-cover of 𝑈𝜆 / 𝜆 ∈ 𝐼 ∪ 𝐴𝐶 of E ⇒

𝑖=𝑘 𝐶
⇒ ∃ 𝑈𝜆 1 , … , 𝑈𝜆 𝑘 ∈ 𝑈𝜆 / 𝜆 ∈ 𝐼 / 𝐸 ⊂ 𝑖=1 𝑈𝜆 𝑖 ∪ 𝐴 𝐴⊂𝐸
𝑖=𝑘 𝑖=𝑘
⇒ 𝐴⊂ 𝑈𝜆 𝑖 ∪ 𝐴𝐶 𝐴⊂ 𝑈𝜆 𝑖
𝑖=1 𝐴∩𝐴𝐶 =∅ 𝑖=1

Then, 𝑈𝜆 1 , … , 𝑈𝜆 𝑘 𝑖𝑠 𝑎 𝑠𝑢𝑏 − 𝑜𝑝𝑒𝑛 𝑐𝑜𝑣𝑒𝑟 𝑜𝑓 𝐴 ⇒ 𝐴 𝑖𝑠 𝑐𝑜𝑚𝑝𝑎𝑐𝑡.


Theorem: (NESTED PROPERTY)

Given a compact m.e. 𝐸, 𝑑 . 𝑆𝑛 (𝑛≥1) is a sequence of non-empty closed subsets of E such that
𝑆1 ⊃ 𝑆2 ⊃ ⋯ ⊃ 𝑆𝑛 ⊃ ⋯. Then, +∞𝑛=1 𝑆𝑛 ≠ ∅

+∞
Proof: Assume otherwise that , 𝑛=1 𝑆𝑛 =∅

𝐵𝑦 𝑕𝑦𝑝𝑜𝑡𝑕𝑒𝑠𝑖𝑠: ∀𝑛 ∈ ℕ, 𝑆𝑛 𝑖𝑠 𝑐𝑙𝑜𝑠𝑒𝑑 ⇒ ∀𝑛 ∈ ℕ, 𝑆𝑛 𝐶 𝑖𝑠 𝑜𝑝𝑒𝑛.

𝐶 ⇒.
𝐵𝑒𝑐𝑎𝑢𝑠𝑒 𝑀𝑜𝑟𝑔𝑎𝑛′ 𝑠 𝐿𝑎𝑤: +∞ 𝑛=1 𝑆𝑛 =
+∞
𝑛=1 𝑆𝑛
𝐶 = ∅𝐶 = 𝐸

⇒ {𝑆𝑛 𝐶 } 𝑖𝑠 𝑎𝑛 𝑜𝑝𝑒𝑛 𝑐𝑜𝑣𝑒𝑟 𝑜𝑓 𝐸

⇒ ∃ 𝑆𝑛 1 𝐶 , 𝑆𝑛 2 𝐶 , … , 𝑆𝑛 𝑘 𝐶 𝑎 𝑓𝑖𝑛𝑖𝑡𝑒 𝑠𝑢𝑏 − 𝑐𝑜𝑣𝑒𝑟 𝑜𝑓 𝐸
𝐸 𝑖𝑠 𝑐𝑜𝑚𝑝𝑎𝑐𝑡
𝑤𝑖𝑡𝑕 𝑛1 < 𝑛2 < ⋯ < 𝑛𝑘

𝑆𝑛 1 𝐶 , … , 𝑆𝑛 𝑘 𝐶 𝑎 𝑓𝑖𝑛𝑖𝑡𝑒 𝑠𝑢𝑏 − 𝑐𝑜𝑣𝑒𝑟 𝑜𝑓 𝐸 ⇒ 𝐸 ⊂ 𝑖=𝑘

𝑖=1 𝑆𝑛 𝑖
𝐶
⇒𝐸= 𝑖=𝑘
𝑖=1 𝑆𝑛 𝑖
𝐶

𝐶 𝐶 𝐶 ⇒.
𝑆1 ⊃ ⋯ ⊃ 𝑆𝑛 ⊃ ⋯ 𝑆𝑛 1 ⊃ ⋯ ⊃ 𝑆𝑛 𝑘 ⇒ 𝑆𝑛 1 ⊂ ⋯ ⊂ 𝑆𝑛 𝑘 ⇒ 𝑖=𝑘
𝑖=1 𝑆𝑛 𝑖 = 𝑆𝑛 𝑘 𝐶
𝑛 1 <⋯<𝑛 𝑘

⇒ 𝐸 = 𝑆𝑛 𝑘 𝐶 ⇒ ∅ = 𝑆𝑛 𝑘 (𝑎𝑏𝑠𝑢𝑟𝑑)


𝐂𝐋𝐔𝐒𝐓𝐄𝐑 𝐎𝐑 𝐒𝐓𝐈𝐂𝐊𝐘 𝐏𝐎𝐈𝐍𝐓: Given a metric space 𝐸, 𝑑 , 𝑎𝑛𝑑 𝑆 ⊂ 𝐸 𝑠 𝑐𝑎𝑛 𝑏𝑒 ∅ . We say

that 𝑝 ∈ 𝐸 is a cluster point of S ⇔ ∀𝐵 𝑝, 𝑟 , 𝐵 𝑝, 𝑟 contains infinitely many elements of S ⇔
⇔ ∀𝐵 𝑝, 𝑟 , 𝐵 ∗ 𝑝, 𝑟 : = 𝐵 𝑝, 𝑟 − 𝑝 ∩ 𝑆 ≠ ∅.

Theorem: 𝐸, 𝑑 𝑐𝑜𝑚𝑝𝑎𝑐𝑡 𝑚. 𝑒. 𝐴 ⊂ 𝐸 𝑖𝑠 𝑎𝑛 𝑖𝑛𝑓𝑖𝑛𝑖𝑡𝑒 𝑠𝑒𝑡 ⇒ 𝐴 𝑕𝑎𝑠 𝑎𝑡 𝑙𝑒𝑎𝑠𝑡 𝑜𝑛𝑒 𝑐𝑙𝑢𝑠𝑡𝑒𝑟 𝑝𝑜𝑖𝑛𝑡.

Proof: Suppose that A does not have a cluster point ⇒

⇒ ∀𝑥 ∈ 𝐸, ∃𝐵 𝑥, 𝑟 𝑠𝑢𝑐𝑕 𝑡𝑕𝑎𝑡 𝐵 𝑥, 𝑟𝑥 ∩ 𝐴 𝑐𝑜𝑛𝑡𝑎𝑖𝑛𝑠 𝑓𝑖𝑛𝑖𝑡𝑒 𝑒𝑙𝑒𝑚𝑒𝑛𝑡𝑠 (𝑖)

Consider the collection 𝐵(𝑥, 𝑟𝑥 )/𝑥 ∈ 𝐸

𝑆𝑖𝑛𝑐𝑒 𝐵(𝑥, 𝑟𝑥 )/𝑥 ∈ 𝐸 𝑖𝑠 𝑎𝑛 𝑜𝑝𝑒𝑛 𝑐𝑜𝑣𝑒𝑟 𝑜𝑓 𝐸 𝑎𝑛𝑑 𝑒 𝑖𝑠 𝑐𝑜𝑚𝑝𝑎𝑐𝑡, ∃𝑎 𝑓𝑖𝑛𝑖𝑡𝑒 𝑠𝑢𝑏𝑐𝑜𝑣𝑒𝑟 𝑜𝑓 𝐸 ⇒
𝑖=𝑘
𝑖=𝑘 𝑖=𝑘
⇒ ∃𝑥1 , … , 𝑥𝑘 ∈ 𝐸/𝐸 ⊂ 𝐵 𝑥𝑖 , 𝑟𝑥 𝑖 ⇒𝐴⊂ 𝐵 𝑥𝑖 , 𝑟𝑥 𝑖 ⇒ 𝐴 ∩ 𝐵 𝑥𝑖 , 𝑟𝑥 𝑖 = 𝐴 ⇒
𝑖=1
𝑖=1 𝑖=1
𝐴⊂𝐸
𝑖=𝑘
⇒ 𝐵 𝑥𝑖 , 𝑟𝑥 𝑖 ∩ 𝐴 = 𝐴 ⇒ 𝐴 𝑖𝑠 𝑓𝑖𝑛𝑖𝑡𝑒 (𝑎𝑏𝑠𝑢𝑟𝑑)
𝑖=1
𝐹𝑖𝑛𝑖𝑡𝑒 𝑠𝑒𝑡 𝑏𝑒𝑐𝑎𝑢𝑠𝑒 (𝑖)

NOTE: On the one hand all elements of a are contained in these k open balls. On the other hand
each open ball contains finitely many elements of A. Because we have only finitely many (k)of
such open balls, A is a finite set.

Corollary 1: 𝐸, 𝑑 𝑐𝑜𝑚𝑝𝑎𝑐𝑡 𝑚. 𝑒. 𝑝𝑛 ⊂ 𝐸 𝑠𝑒𝑞𝑢𝑒𝑛𝑐𝑒 ⇒ 𝑝𝑛 has a convergent subsequence.

Proof: Pick up any sequence 𝑝𝑛 ⊂ 𝐸 . Call 𝐴 = 𝑝𝑛 ; 𝑛 ∈ ℕ

Case 1: A is finite

A is finite. Say 𝐴 = 𝑥1 , 𝑥2 , … , 𝑥𝑘 . Since 𝐴 = 𝑝𝑛 ; 𝑛 ∈ ℕ and 𝑝𝑛 is an infinite sequence, then

∃𝑖 = 1, … , 𝑘 / 𝑥𝑖 appears infinitely many times in the sequence 𝑝𝑛 .

Consider the subsequence 𝑝𝑚 𝑛 such that 𝑝𝑚 𝑛 = 𝑥𝑖 , ∀𝑛 ∈ ℕ.

Claim: 𝑝𝑚 𝑘 𝑖𝑠 𝑐𝑜𝑛𝑣𝑒𝑟𝑔𝑒𝑛𝑡

Proof: Pick up any 𝜀 > 0. ∀𝑛 ≥ 1, 𝑝𝑚 𝑛 − 𝑥𝑖 = 𝑥 − 𝑥𝑖 = 0 < 𝜀.

∴ ∀𝜀 > 0, ∃1 ∈ ℕ / ∀𝑛 ≥ 1, 𝑝𝑚 𝑛 − 𝑥𝑖 < 𝜀 ⇒ 𝑝𝑚 𝑘 𝑐𝑜𝑛𝑣𝑒𝑟𝑔𝑒𝑠 𝑡𝑜 𝑥𝑖 . 

Case 2: A is infinite

A is infinit.e Then A is an infinite subset of E and E is a compact m.e.

𝑇𝑕.𝑏𝑒𝑓𝑜𝑟𝑒
⇒ ∃𝑝 ∈ 𝐸/𝑝 𝑖𝑠 𝑎 𝑐𝑙𝑢𝑠𝑡𝑒𝑟 𝑝𝑜𝑖𝑛𝑡 𝑜𝑓 𝐴 ⇒ ∀𝐵 𝑝, 𝑟 contains infinitely many elements of A ⇒
1 1
⇒ ∀𝐵 𝑝, contains infinitely many elements of A ⇒ ∀𝑛 ∈ ℕ, ∃𝑝𝑚 𝑛 ∈ 𝐵 𝑝, , such
𝑛 𝑛
tha𝑡 𝑚𝑛 < 𝑚𝑛+1 .

Consider the subsequence 𝑝𝑚 𝑘

Claim: 𝑝𝑚 𝑘 𝑖𝑠 𝑐𝑜𝑛𝑣𝑒𝑟𝑔𝑒𝑛𝑡

1
Proof: Pick up any 𝜀 > 0. Choose 𝑁 > (N exist because ℕ 𝑖𝑠 𝑛𝑜𝑡 𝑏𝑜𝑢𝑛𝑑𝑒𝑑)
𝜀
1 1
𝐵𝑦 𝑑𝑒𝑓𝑖𝑛𝑖𝑡𝑖𝑜𝑛 𝑜𝑓 𝑝𝑚 𝑘 , ∀𝑛 ∈ ℕ, 𝑝𝑚 𝑛 ∈ 𝐵 𝑝, ⇒ ∀𝑛 ∈ ℕ, 𝑑(𝑝𝑚 𝑛 , 𝑝) <
𝑛 𝑛 ⇒
1 1
∀𝑛 ≥ 𝑁, ≤
𝑛 𝑁

∀𝑛 ≥ 1, 𝑝𝑚 𝑛 − 𝑥𝑖 = 𝑥 − 𝑥𝑖 = 0 < 𝜀.
∴ ∀𝜀 > 0, ∃1 ∈ ℕ / ∀𝑛 ≥ 1, 𝑝𝑚 𝑛 − 𝑥𝑖 < 𝜀 ⇒ 𝑝𝑚 𝑘 𝑐𝑜𝑛𝑣𝑒𝑟𝑔𝑒𝑠 𝑡𝑜 𝑝. 

∴ ∀ 𝑝𝑛 ⊂ 𝐸 𝑠𝑒𝑞𝑢𝑒𝑛𝑐𝑒 , 𝑝𝑛 has a convergent subsequence. 

Corollary 2: 𝐴𝑛𝑦 𝑐𝑜𝑚𝑝𝑎𝑐𝑡 𝑚. 𝑒. 𝐸 𝑖𝑠 𝑐𝑜𝑚𝑝𝑙𝑒𝑡𝑒.

Proof: Pick up any 𝑝𝑛 𝐶𝑎𝑢𝑐𝑕𝑦 𝑠𝑒𝑞𝑢𝑒𝑛𝑐𝑒 𝑜𝑓 𝐸.

𝑝𝑛 𝑖𝑠 𝑎 𝑠𝑒𝑞𝑢𝑒𝑛𝑐𝑒 𝑝𝑛 has a convergent subsequence 𝑐𝑎𝑙𝑙 lim 𝑝𝑚 𝑘 = 𝑝 ∈ 𝐸
𝑇𝑕𝑒𝑜𝑟𝑒𝑚 𝑇𝑕.𝐶𝑕.7
⇒ 𝑝𝑛 → 𝑝

∴ ∀ 𝑝𝑛 ⊂ 𝐸, 𝑝𝑛 𝐶𝑎𝑢𝑐𝑕𝑦 ⇒ ∃𝑝 ∈ 𝐸/ lim⁡𝑝𝑛 = 𝑝 ⇒ E is complete. 

Corollary 3: 𝐴𝑛𝑦 𝑐𝑜𝑚𝑝𝑎𝑐𝑡 𝑠𝑢𝑏𝑠𝑒𝑡 𝐴 𝑜𝑓 𝑎 𝑚. 𝑒. 𝐸 𝑖𝑠 𝑐𝑙𝑜𝑠𝑒𝑑

Proof: Pick up any 𝑝𝑛 convergent sequence of A. Call 𝐶𝑎𝑙𝑙 𝑝 = lim⁡𝑝𝑛

𝐴 ⊂ 𝐸 ⇒ 𝐴, 𝑑 𝑖𝑠 𝑎 𝑚𝑒𝑡𝑟𝑖𝑐 𝑠𝑝𝑎𝑐𝑒
⇒ 𝐴, 𝑑 𝑖𝑠 𝑎 𝑐𝑜𝑚𝑝𝑎𝑐𝑡 𝑚. 𝑒. 𝐴 𝑖𝑠 𝑐𝑜𝑚𝑝𝑙𝑒𝑡𝑒
𝐵𝑦 𝑕𝑦𝑝𝑜𝑡𝑕𝑒𝑠𝑖𝑠: 𝐴 𝑖𝑠 𝑐𝑜𝑚𝑝𝑎𝑐𝑡 𝐶𝑜𝑟𝑜𝑙𝑙𝑎𝑟𝑦 2 ⇒
𝑝𝑛 ⊂ 𝐴 𝑐𝑜𝑛𝑣𝑒𝑟𝑔𝑒𝑛𝑡 ⇒ 𝑝𝑛 ⊂ 𝐴 𝑖𝑠 𝐶𝑎𝑢𝑐𝑕𝑦
⇒ ∃𝑙 ∈ 𝐴 / lim 𝑝𝑛 = 𝑙
𝑝=𝑙∈𝐴
𝑝 = lim⁡𝑝𝑛 𝑈𝑛𝑖𝑐𝑖𝑡𝑦 𝑜𝑓 𝑡𝑕𝑒 𝑙𝑖𝑚𝑖𝑡

∴ ∀ 𝑝𝑛 ⊂ 𝐴, 𝑝𝑛 𝑐𝑜𝑛𝑣𝑒𝑟𝑔𝑒𝑛𝑡 ⇒ lim 𝑝𝑛 ∈ 𝐴 A is closed. 

𝑇𝑕.𝐶𝑕.6

𝑖=𝑘
Lemma: A⊂ 𝔼𝑛 . A is bdd. ⇒ ∀𝜀 > 0, ∃𝑥1 , … , 𝑥𝑘 ∈ 𝔼𝑛 /𝐴 ⊂ 𝑖=1 𝐵 𝑥𝑖 , 𝜀

In other words: 𝐹𝑜𝑟 𝑎𝑛𝑦 𝜀 > 0, there exist finitely many closed balls with radio 𝜀, whose union
contains A.

Proof: The proof is analog to every 𝑛 ∈ ℕ. We will write an idea of it to 𝑛 = 2

A is bdd ⇒ ∃𝐵 𝑝, 𝑟 𝑠𝑢𝑐𝑕 𝑡𝑕𝑎𝑡 𝐴 ⊂ 𝐵(𝑝, 𝑟). Say 𝑝(𝑥𝑝 , 𝑦𝑝 )

Call 𝑎 = 𝑥𝑝 − 𝑟, 𝑏 = 𝑥𝑝 + 𝑟, 𝑐 = 𝑦𝑝 − 𝑟 𝑎𝑛𝑑 𝑑 = 𝑦𝑝 + 𝑟
Then, A is included in the area limited by 𝑥 = 𝑎 𝑎𝑛𝑑 𝑥 = 𝑏 𝑎𝑛𝑑 𝑦 = 𝑐 𝑎𝑛𝑑 𝑦 = 𝑑
𝜀
Pick up any 𝜀 > 0. Divide the intervals 𝑎, 𝑏 𝑎𝑛𝑑 (𝑐, 𝑑) in equal intervals.
2𝑟

Consider every closed ball with radius 𝜀 and centered in every intersection of the net.
𝜀 2
Then, there exist +1 𝑚𝑎𝑛𝑦 𝑐𝑙𝑜𝑠𝑒𝑑 𝑏𝑎𝑙𝑙𝑠 𝑤𝑖𝑡𝑕 𝑟𝑎𝑑𝑖𝑢𝑠 𝜀, 𝑤𝑕𝑜𝑠𝑒 𝑢𝑛𝑖𝑜𝑛 𝑐𝑜𝑛𝑡𝑎𝑖𝑛𝑠 𝐴. 
2𝑟

Theorem: (HAHN - BANACH)

𝐿𝑒𝑡 𝐴 ⊂ 𝔼𝑛 .
A is compact ⇔ A is closed and bounded.

Proof:

⇒ : Directly by theorems proved before.

⇐ : Assume otherwise that A is not compact ⇒ ∃ 𝑈𝜆 ; 𝜆 ∈ 𝐼 open cover of A / 𝑈𝜆 ; 𝜆 ∈ 𝐼

does not have a finite sub-cover of A.

𝑖=𝑘 1 𝑖=𝑘 1
A is bdd 𝐿𝑒𝑚𝑚𝑎 ∃𝑥1,1 , … , 𝑥𝑘,1 ∈ 𝔼𝑛 /𝐴 ⊂ 𝑖=1 𝐵 𝑥𝑖,1 , ⇒𝐴∩ 𝑖=1 𝐵 𝑥𝑖,1 , =𝐴 ⇒
2 2
1
𝜀=
2
𝑖=𝑘 1 1 1
⇒ 𝐴 ∩ 𝐵 𝑥𝑖,1 , = 𝐴 ⇒ 𝐴 = 𝐴 ∩ 𝐵 𝑥1 , ∪ … ∪ 𝐴 ∩ 𝐵 𝑥𝑘 , ⇒
𝑖=1 2 2 2
1
⇒ ∃𝑖 = 1, … , 𝑘 / 𝐴 ∩ 𝐵 𝑥𝑖,1 , cannot be covered by a finite sub-cover of 𝑈𝜆 ; 𝜆 ∈ 𝐼 .
2
1
Call 𝐴 ∩ 𝐵 𝑥𝑖,1 , = 𝐴1 .
2
Notice that:
1. 𝐴1 ≠ ∅ Because otherwise it will be covered by a finite sub-cover of 𝑈𝜆 ; 𝜆 ∈ 𝐼 .
1
2. 𝐴1 is closed. Because A is closed, 𝐵 𝑥𝑖,1 , , 𝑡𝑕𝑒𝑛 𝑡𝑕𝑒 𝑖𝑛𝑡𝑒𝑟𝑠𝑒𝑐𝑡𝑖𝑜𝑛 𝑖𝑠 𝑐𝑙𝑜𝑠𝑒𝑑.
2
1 1
3. ∀𝑥, 𝑦 ∈ 𝐴1 , 𝑑 𝑥, 𝑦 ≤ 1 Because by ∆ − 𝐼𝑛𝑒𝑞.: 𝑑 𝑥, 𝑦 ≤ 𝑑 𝑥, 𝑥𝑖 + 𝑑 𝑥𝑖 , 𝑦 ≤ + = 1
2 2
4. 𝐴1 is bdd Because is included in a ball
5. 𝐴1 cannot be covered by a finite sub-cover of 𝑈𝜆 ; 𝜆 ∈ 𝐼

1
Apply the lemma to 𝐴1 with 𝜀 = …
4
1
𝐴1 = 𝐴 ∩ 𝐵 𝑥𝑖,1 ,
2
By repeating this process we get: 𝐴𝑛 : 1
such that:
𝐴𝑛 = 𝐴𝑛−1 ∩ 𝐵 𝑥𝑖,𝑛 ,
2𝑛
1. 𝐴𝑛 ≠ ∅, ∀𝑛 ∈ ℕ
2. 𝐴𝑛 is closed, ∀𝑛 ∈ ℕ
1
3. ∀𝑥, 𝑦 ∈ 𝐴1 , 𝑑 𝑥, 𝑦 ≤ , ∀𝑛 ∈ ℕ
𝑛
4. 𝐴𝑛 cannot be covered by a finite sub-cover of 𝑈𝜆 ; 𝜆 ∈ 𝐼
5. 𝐴1 ⊃ 𝐴2 ⊃ ⋯ ⊃ 𝐴𝑛 …

𝐴𝑛 ≠ ∅, ∀𝑛 ∈ ℕ ⇒ ∀𝑛 ∈ ℕ, ∃𝑝𝑛 ∈ 𝐴𝑛 . Consider the sequence 𝑝𝑛

Claim: 𝑝𝑛 is Cauchy

1
Proof: Pick up any 𝜀 > 0. 𝐶𝑕𝑜𝑜𝑠𝑒 𝑁 >
𝜀

𝑝𝑛 ∈ 𝐴𝑛 ⊂ 𝐴𝑁 ⇒ 𝑝𝑛 ∈ 𝐴𝑁 1
𝑃𝑖𝑐𝑘 𝑢𝑝 𝑎𝑛𝑦 𝑚, 𝑛 ≥ 𝑁 ⇒ #5 ⇒ 𝑑 𝑝𝑛 , 𝑝𝑚 ≤
𝑝𝑚 ∈ 𝐴𝑚 ⊂ 𝐴𝑁 ⇒ 𝑝𝑚 ∈ 𝐴𝑁 #3 𝑁 ⇒ 𝑑 𝑝 ,𝑝 < 𝜀
𝑛 𝑚
1 1
𝑁> ⇒ <𝜀
𝜀 𝑁
∴ ∀𝜀 > 0, ∃𝑁 ∈ ℕ / ∀𝑚, 𝑛 ≥ 𝑁, 𝑑 𝑝𝑛 , 𝑝𝑚 < 𝜀 ⇒ 𝑝𝑛 is Cauchy

𝔼𝑛 𝑖𝑠 𝑐𝑜𝑚𝑝𝑙𝑒𝑡𝑒 𝑎𝑛𝑑 𝑝𝑛 ⊂ 𝐴 ⊂ 𝔼𝑛 𝑖𝑠 𝐶𝑎𝑢𝑐𝑕𝑦 ⇒ ∃𝑝 ∈ 𝔼𝑛 / lim 𝑝𝑛 = 𝑝⁡

⇒𝑝∈𝐴⇒
𝐴 𝑖𝑠 𝑐𝑙𝑜𝑠𝑒𝑑
⇒ ∃𝜆0 ∈ 𝐼 /𝑝 ∈ 𝑈𝜆 0
⇒ ∃𝐵 𝑝, 𝑟 ⊂ 𝑈𝜆 0 (2)
𝑈𝜆 0 𝑖𝑠 𝑜𝑝𝑒𝑛
𝑟 𝑟
lim 𝑝𝑛 = 𝑝 𝑎𝑛𝑑 > 0 ⇒ ∃𝑁 ∈ ℕ/∀𝑛 ≥ 𝑁, 𝑝𝑛 ∈ 𝐵 𝑝, (1)
2 2
𝑟
𝑛0 > 𝑁 ⇒ 𝑑(𝑝, 𝑝𝑛 0 ) <
2 1 2
𝑇𝑎𝑘𝑒 𝑎𝑛𝑦 𝑛0 > 𝑚𝑎𝑥 , 𝑁 ⇒ 2 1 𝑟 ⇒
𝑟 𝑛0 > ⇒ <
𝑟 𝑛0 2
𝑟 𝑟
⇒ ∀𝑥 ∈ 𝐴𝑛 0 ∶ 𝑑 𝑝, 𝑥 ≤ 𝑑 𝑝, 𝑝𝑛 0 + 𝑑 𝑥, 𝑝𝑛 0 < + = 𝑟 ⇒ ∀𝑥 ∈ 𝐴𝑛 0 , 𝑥 ∈ 𝐵 𝑝, 𝑟 ⇒
2 2
⇒ 𝐴𝑛 0 ⊂ 𝐵 𝑝, 𝑟 ⊂ 𝑈𝜆 0 . Then, 𝐴𝑛 0 can be covered by a finite sub-cover. (absurd)

9. LIMIT AND CONTINUITY OD FUNCTIONS
𝐋𝐈𝐌𝐈𝐓 𝐎𝐅 𝐅𝐔𝐍𝐂𝐓𝐈𝐎𝐍𝐒: 𝐸1 , 𝑑1 𝑎𝑛𝑑 𝐸2 , 2 are two metric spaces . We say that a function
𝑓: 𝐸1 → 𝐸2 converges to 𝑄 ∈ 𝐸2 when x goes to A
𝐴 ∈ 𝐸1 𝑎𝑛𝑑 𝐴 𝑖𝑠 𝑎 𝑐𝑙𝑢𝑠𝑡𝑒𝑟 𝑝𝑜𝑖𝑛𝑡 𝑜𝑓 𝐸1 ⇔ ∀𝜀 > 0, ∃𝛿 > 0/ ∀𝑥 ∈ 𝐸1 , 0 < 𝑑1 (𝑥, 𝐴) < 𝛿 ⇒
𝑑2 (𝑓 𝑥 , 𝑄) < 𝜀

Notation: lim𝑥→𝐴 𝑓(𝑥) = 𝑄

NOTE: lim𝑥 →𝐴 𝑓(𝑥) = 𝑄 ⇔ ∀𝜀 > 0, ∃𝛿 > 0/ ∀𝑥, 𝑥 ∈ 𝐵 ∗ (𝐴, 𝛿) ⇒ 𝑓(𝑥) ∈ 𝐵(𝑄, 𝜀)

𝐏𝐔𝐍𝐓𝐔𝐀𝐋 𝐂𝐎𝐍𝐓𝐈𝐍𝐔𝐈𝐓𝐘: 𝐸1 , 𝑑1 𝑎𝑛𝑑 𝐸2 , 2 are two metric spaces . We say that a function
𝑓: 𝐸1 → 𝐸2 is continuous in 𝐴 ∈ 𝐸1 ⇔ lim𝑥 →𝐴 𝑓(𝑥) = 𝑓(𝐴)

𝐂𝐎𝐍𝐓𝐈𝐍𝐔𝐈𝐓𝐘 𝐎𝐅 𝐀 𝐅𝐔𝐍𝐂𝐓𝐈𝐎𝐍: 𝐸1 , 𝑑1 𝑎𝑛𝑑 𝐸2 , 2 are two metric spaces . We say that a

function 𝑓: 𝐸1 → 𝐸2 iff f is continuous in every 𝑥 ∈ 𝐸1 .

Theorem: (LIMIT OF A CONSTANT FUNCTION)

𝐸1 , 𝑑1 𝑎𝑛𝑑 𝐸2 , 2 𝑚𝑒𝑡𝑟𝑖𝑐 𝑠𝑝𝑎𝑐𝑒𝑠. 𝑓: 𝐸1 → 𝐸2 /𝑓(𝑥) = 𝑘 ⇒ ∀𝐴 ∈ 𝐸1 , lim𝑥→𝐴 𝑓(𝑥) = 𝑘

Proof: Pick up any 𝐴 ∈ 𝐸1 𝑎𝑛𝑑 𝜀 > 0. Choose 𝛿 = 1.

Pick up any ∈ 𝐸1 such that 0 < 𝑑1 𝑥, 𝐴 < 1 ⇒ 𝑑2 𝑓 𝑥 , 𝑘 = 𝑑2 𝑘, 𝑘 = 0 < 𝜀
∴ ∀𝜀 > 0, ∃𝛿 > 0/ ∀𝑥 ∈ 𝐸1 , 0 < 𝑑1 𝑥, 𝐴 < 1 ⇒ 𝑑2 𝑓 𝑥 , 𝑘 < 𝜀
∴ ∀𝐴 ∈ 𝐸1 , lim𝑥→𝐴 𝑓(𝑥) = 𝑘 

Corollary: 𝐴𝑛𝑦 𝑐𝑜𝑛𝑠𝑡𝑎𝑛𝑡 𝑓𝑢𝑛𝑐𝑡𝑖𝑜𝑛 𝑖𝑠 𝑐𝑜𝑛𝑡𝑖𝑛𝑢𝑜𝑢𝑠.

Proof: Directly from theorem. 

Theorem: (PASAJE THEOREM)

𝐸1 , 𝑑1 𝑎𝑛𝑑 𝐸2 , 2 𝑚𝑒𝑡𝑟𝑖𝑐 𝑠𝑝𝑎𝑐𝑒𝑠. 𝑓: 𝐸1 → 𝐸2 . 𝐴 ∈ 𝐸1 𝑎𝑛𝑑 𝑄 ∈ 𝐸2
lim 𝑓(𝑥) = 𝑄 ⇔ ∀ 𝑝𝑛 ⊂ 𝐸1 − 𝐴 , 𝑝𝑛 → 𝐴 ⇒ 𝑓 𝑝𝑛 →𝑄
𝑥 →𝐴
Proof:

⇒ : Pick up any sequence 𝑝𝑛 ⊂ 𝐸1 − 𝐴 , such that 𝑝𝑛 → 𝐴

Pick up any 𝜀 > 0

lim 𝑓(𝑥) = 𝑄 𝑎𝑛𝑑 𝜀 > 0 ⇒ ∃𝛿 > 0/∀𝑥 ∈ 𝐸1 , 0 < 𝑑1 𝑥, 𝐴 < 𝛿 ⇒ 𝑑2 𝑓 𝑥 , 𝑄 < 𝜀 (1)
𝑥→𝐴
𝑝𝑛 → 𝐴 𝑎𝑛𝑑 𝛿 > 0 ⇒ ∃𝑁 ∈ ℕ/∀𝑛 ≥ 𝑁, 𝑑 𝑝𝑛 , 𝐴 < 𝛿
⇒
𝑝𝑛 ⊂ 𝐸1 − 𝐴 ⇒ 𝑝𝑛 ≠ 𝐴, ∀𝑛 ∈ ℕ ⇒ 0 < 𝑑 𝑝𝑛 , 𝐴 , ∀𝑛 ∈ ℕ
⇒ ∃𝑁 ∈ ℕ/∀𝑛 ≥ 𝑁, 0 < 𝑑(𝑝𝑛 , 𝐴) < 𝛿 ∃𝑁 ∈ ℕ/∀𝑛 ≥ 𝑁, 𝑑2 (𝑓 𝑝𝑛 , 𝑄) < 𝜀
(1)
∴ ∀𝜀 > 0, ∃𝑁 ∈ ℕ/∀𝑛 ≥ 𝑁, 𝑑2 (𝑓 𝑝𝑛 , 𝑄) < 𝜀 ⇒ 𝑓 𝑝𝑛 →𝑄

∴ ∀ 𝑝𝑛 ⊂ 𝐸1 − 𝐴 , 𝑝𝑛 → 𝐴 ⇒ 𝑓 𝑝𝑛 →𝑄
 ⇐ : Assume otherwise that lim𝑥→𝐴 𝑓(𝑥) ≠ 𝑄 ⇒
⇒ ∃𝜀 > 0, ∀𝛿 > 0, ∃𝑥 ∈ 𝐸1 such that 0 < 𝑑1 𝑥, 𝐴 < 𝛿 and 𝑑2 𝑓 𝑥 , 𝑄 > 𝜀 ⇒
1
⇒ ∃𝜀 > 0, ∀𝑛 ∈ ℕ, ∃𝑥𝑛 ∈ 𝐸1 such that 0 < 𝑑1 𝑥𝑛 , 𝐴 < and 𝑑2 𝑓 𝑥𝑛 , 𝑄 > 𝜀 (1)
𝑛
Consider the sequence 𝑥𝑛

Claim 1: 𝑥𝑛 ⊂ 𝐸1 − 𝐴

Proof: ∀𝑛 ∈ ℕ, 𝑥𝑛 ∈ 𝐸1 and 0 < 𝑑1 𝑥, 𝐴 ⇒ ∀𝑛 ∈ ℕ, 𝑥𝑛 ∈ 𝐸1 and 𝑥𝑛 ≠ 𝐴 ⇒ 𝑥𝑛 ⊂ 𝐸1 − 𝐴 

Claim 2: 𝑥𝑛 converges to A

1 1 1
Proof: Pick up any 𝜀 > 0. Choose > . Then, ∀𝑛 ≥ 𝑁, 0 < 𝑑1 𝑥𝑛 , 𝐴 < < < 𝜀
𝜀 𝑛 𝑁
∴ ∀𝜀 > 0, ∃ 𝑁 ∈ ℕ / ∀𝑛 ≥ 𝑁, 0 < 𝑑1 𝑥𝑛 , 𝐴 < 𝜀 ⇒ 𝑥𝑛 converges to A 

Claim 3: 𝑓(𝑥𝑛 ) does not converge to Q

Proof: By (1): ∃𝜀 > 0, ∀𝑛 ∈ ℕ, 𝑑2 𝑓 𝑥𝑛 , 𝑄 > 𝜀 ⇒ 𝑓(𝑥𝑛 ) does not converge to Q 

∴ ∃ 𝑥𝑛 ⊂ 𝐸1 − 𝐴 𝑠𝑢𝑐𝑕 𝑡𝑕𝑎𝑡 𝑥𝑛 converges to A and 𝑓(𝑥𝑛 ) does not converge to Q (absurd)

∴ lim𝑥 →𝐴 𝑓(𝑥) = 𝑄 

Theorem: (ARITHMETIC PROPERTIES OF THE LIMIT)

𝐸1 , 𝑑1 𝑎𝑛𝑑 𝐸2 , 2 𝑚𝑒𝑡𝑟𝑖𝑐 𝑠𝑝𝑎𝑐𝑒𝑠. 𝑓, 𝑔: 𝐸1 → 𝔼1 .

𝑖) lim𝑥→𝐴 𝛼. 𝑓 (𝑥) = 𝛼. 𝐹
lim𝑥→𝐴 𝑓(𝑥) = 𝐹 𝑖𝑖) lim𝑥→𝐴 𝑓 ± 𝑔 (𝑥) = 𝐹 ± 𝐺
⇒ 𝑖𝑖𝑖) lim𝑥→𝐴 𝑓. 𝑔 (𝑥) = 𝐹. 𝐺 .
lim𝑥→𝐴 𝑔(𝑥) = 𝐺
𝑓 𝐹
𝑖𝑣) 𝑖𝑓 𝐺 ≠ 0: lim𝑥→𝐴 (𝑥) =
𝑔 𝐺

Proof: Pick up any sequence 𝑝𝑛 ⊂ 𝐸1 − 𝐴 such that 𝑝𝑛 → 𝐴 (1)

𝛼. 𝑓 𝑝𝑛 → 𝛼. 𝐹 (𝑖)
lim 𝑓(𝑥) = 𝐹 𝑓 𝑝𝑛 →𝐹 𝑓 ± 𝑔 𝑝𝑛 →𝐹±𝐺 (𝑖𝑖)
𝑥→𝐴 1 𝑎𝑛𝑑
𝑃𝑎𝑠𝑎𝑗𝑒
𝑓. 𝑔 𝑝𝑛 → 𝐹. 𝐺 (𝑖𝑖𝑖)
lim 𝑔(𝑥) = 𝐺 𝑔 𝑝𝑛 →𝐺 𝐴𝑟𝑖𝑡 𝑕𝑚𝑒𝑡𝑖𝑐 𝑝𝑟𝑜𝑝𝑒𝑟𝑡𝑖𝑒𝑠
𝑥→𝐴 1 𝑎𝑛𝑑 𝑜𝑓 𝑓 𝐹
𝑃𝑎𝑠𝑎𝑗𝑒 𝑙𝑖𝑚𝑖𝑡𝑒 𝑜𝑓 𝑠𝑒𝑞𝑢𝑒𝑛𝑐𝑒𝑠 𝑆𝑖𝑛𝑐𝑒 𝐺 ≠ 0 𝑝𝑛 → (𝑖𝑣)
𝑔 𝐺
∴ 𝐹𝑟𝑜𝑚 𝑃𝑎𝑠𝑎𝑗𝑒 𝑎𝑛𝑑 …
𝑖 ∀ 𝑝𝑛 ⊂ 𝐸1 − 𝐴 , 𝑝𝑛 → 𝐴 ⇒ 𝛼. 𝑓 𝑝𝑛 → 𝛼. 𝐹 ⇒ lim 𝛼. 𝑓 (𝑥) = 𝛼. 𝐹
𝑥→𝐴
𝑖𝑖 ∀ 𝑝𝑛 ⊂ 𝐸1 − 𝐴 , 𝑝𝑛 → 𝐴 ⇒ 𝑓 ± 𝑔 𝑝𝑛 → 𝐹 ± 𝐺 ⇒ lim 𝑓 ± 𝑔 (𝑥) = 𝐹 ± 𝐺
𝑥→𝐴
𝑖𝑖𝑖 ∀ 𝑝𝑛 ⊂ 𝐸1 − 𝐴 , 𝑝𝑛 → 𝐴 ⇒ 𝑓. 𝑔 𝑝𝑛 → 𝐹. 𝐺 ⇒ lim 𝑓. 𝑔 (𝑥) = 𝐹. 𝐺
𝑥 →𝐴
𝑓 𝐹 𝑓 𝐹
𝑖𝑣 ∀ 𝑝𝑛 ⊂ 𝐸1 − 𝐴 , 𝑝𝑛 → 𝐴 ⇒ 𝑝𝑛 → ⇒ lim𝑥→𝐴 (𝑥) = 
𝑔 𝐺 𝑔 𝐺
Theorem: 𝐿𝑒𝑡 𝐸, 𝑑 𝑏𝑒 𝑎𝑛𝑦 𝑚. 𝑒. 𝐶𝑜𝑛𝑠𝑖𝑑𝑒𝑟 𝑓: 𝐸 → 𝔼𝑛 𝑛 ≥ 1 / 𝑓(𝑥) = 𝑓1 𝑥 , … , 𝑓𝑛 (𝑥) .
Notice that to be f a function, 𝑓𝑖 : 𝐸 → 𝔼1 must be a function, ∀𝑖 = 1, … , 𝑛.

𝑓 𝑖𝑠 𝑐𝑜𝑛𝑡𝑖𝑛𝑢𝑜𝑢𝑠 𝑎𝑡 𝐴 ∈ 𝐸 ⇔ 𝑓𝑖 : 𝐸 → 𝔼1 𝑖𝑠 𝑐𝑜𝑛𝑡𝑖𝑛𝑢𝑜𝑢𝑠 𝑎𝑡 𝐴 ∈ 𝐸, ∀𝑖 = 1, … , 𝑛

Proof:

⇒ : Pick up any 𝜀 > 0

𝑓 𝑖𝑠 𝑐𝑜𝑛𝑡. 𝑎𝑡 𝐴 ∈ 𝐸 ⇒
⇒ lim 𝑓(𝑥) = 𝑓(𝐴) ⇒ ∃𝛿 > 0/ ∀𝑥 ∈ 𝐸1 , 0 < 𝑑 𝑥, 𝐴 < 𝛿 ⇒ 𝑑 𝑓 𝑥 , 𝑓 𝐴 <𝜀 (1)
𝑥 →𝐴

∀𝑖 = 1, … , 𝑛:
2 2 2
𝑓𝑖 𝑥 − 𝑓𝑖 𝐴 = 𝑓𝑖 𝑥 − 𝑓𝑖 𝐴 < 𝑓1 𝑥 − 𝑓1 𝐴 + ⋯ + 𝑓𝑛 𝑥 − 𝑓𝑛 (𝐴) =
= 𝑑 𝑓1 𝑥 , … , 𝑓𝑛 (𝑥) , 𝑓1 𝐴 , … , 𝑓𝑛 (𝐴) = 𝑑 𝑓 𝑥 , 𝑓 𝐴 ⇒
⇒ ∀𝑖 = 1, … , 𝑛: 𝑓𝑖 𝑥 − 𝑓𝑖 𝐴 < 𝑑 𝑓 𝑥 , 𝑓 𝐴 (2)

From 1 𝑎𝑛𝑑 2 : ∀𝑖 = 1, … , 𝑛, ∃𝛿 > 0/ ∀𝑥 ∈ 𝐸1 , 0 < 𝑑 𝑥, 𝐴 < 𝛿 ⇒ 𝑓𝑖 𝑥 − 𝑓𝑖 𝐴 < 𝜀

∴ ∀𝑖 = 1, … , 𝑛: ∀𝜀 > 0, ∃𝛿 > 0/ ∀𝑥 ∈ 𝐸1 , 0 < 𝑑 𝑥, 𝐴 < 𝛿 ⇒ 𝑓𝑖 𝑥 − 𝑓𝑖 𝐴 < 𝜀 ⇒
⇒ ∀𝑖 = 1, … , 𝑛: lim 𝑓𝑖 (𝑥) = 𝑓𝑖 𝐴 ⇒ 𝑓𝑖 : 𝐸 → 𝔼1 𝑖𝑠 𝑐𝑜𝑛𝑡𝑖𝑛𝑢𝑜𝑢𝑠 𝑎𝑡 𝐴 ∈ 𝐸, ∀𝑖 = 1, … , 𝑛
𝑥→𝐴

⇐ : Pick up any 𝜀 > 0

𝜀
𝜀>0 ⇒ >0
𝑛 ⇒
𝑓𝑖 𝑖𝑠 𝑐𝑜𝑛𝑡𝑖𝑛𝑢𝑜𝑢𝑠 𝑎𝑡 𝐴 ∈ 𝐸, ∀𝑖 = 1, … , 𝑛 ⇒ ∀𝑖 = 1, … , 𝑛: lim 𝑓𝑖 (𝑥) = 𝑓𝑖 𝐴
𝑥→𝐴

𝜀
⇒ ∀𝑖 = 1, … , 𝑛: ∃𝛿𝑖 > 0/ ∀𝑥 ∈ 𝐸1 , 0 < 𝑑 𝑥, 𝐴 < 𝛿𝑖 ⇒ 𝑓𝑖 𝑥 − 𝑓𝑖 𝐴 <
𝑛

𝑇𝑎𝑘𝑒 𝛿 = 𝑚𝑖𝑛 𝛿𝑖 / 𝑖 = 1, … , 𝑛 > 0 (𝑏𝑒𝑐𝑎𝑠𝑢𝑒 𝑏𝑦 𝑑𝑒𝑓. 𝑡𝑕𝑒 𝑚𝑖𝑛𝑖𝑚𝑢𝑚 𝑏𝑒𝑙𝑜𝑛𝑔𝑠 𝑡𝑜 𝑡𝑕𝑒 𝑠𝑒𝑡)

𝑇𝑎𝑘𝑒 𝛿 = 𝑚𝑖𝑛 𝛿𝑖 / 𝑖 = 1, … , 𝑛 ⇒ 𝛿 ≤ 𝛿𝑖 , ∀𝑖 = 1, … , 𝑛 ⇒
⇒ ∀𝑥 ∈ 𝐸1 , 𝑖𝑓 0 < 𝑑 𝑥, 𝐴 < 𝛿 ⇒ 0 < 𝑑 𝑥, 𝐴 < 𝛿𝑖 , ∀𝑖 = 1, … , 𝑛
(1)
𝜀
⇒ 𝑓𝑖 𝑥 − 𝑓𝑖 𝐴 < , ∀𝑖 = 1, … , 𝑛 ⇒
𝑛
𝜀 2 𝜀 2
2 2
⇒ 𝑑 𝑓 𝑥 ,𝑓 𝐴 = 𝑓1 𝑥 − 𝑓1 𝐴 + ⋯ + 𝑓𝑛 𝑥 − 𝑓𝑛 𝐴 < + ⋯+ =
𝑛 𝑛

𝑛. 𝜀 2
= = 𝜀 2 =𝜀>0 𝜀
𝑛

∴ ∀𝜀 > 0, ∃𝛿 > 0/ ∀𝑥 ∈ 𝐸1 , 0 < 𝑑 𝑥, 𝐴 < 𝛿 ⇒ 𝑑 𝑓 𝑥 , 𝑓 𝐴 < 𝜀 ⇒ lim 𝑓(𝑥) = 𝑓(𝐴) ⇒

𝑥→𝐴
⇒ 𝑓 𝑖𝑠 𝑐𝑜𝑛𝑡𝑖𝑛𝑢𝑜𝑢𝑠 𝑎𝑡 𝐴 ∈ 𝐸.

Theorem: (PINCHING OR SANDWICH)

𝐿𝑒𝑡 𝐸, 𝑑 𝑏𝑒 𝑎𝑛𝑦 𝑚. 𝑒. 𝐶𝑜𝑛𝑠𝑖𝑑𝑒𝑟 𝑓, 𝑔, 𝑕: 𝐸 → 𝔼1 𝑠𝑢𝑐𝑕 𝑡𝑕𝑎𝑡 𝑓 𝑥 ≤ 𝑔 𝑥 ≤ 𝑕 𝑥 , ∀𝑥 ∈ 𝐸 − 𝐴 .

lim 𝑓(𝑥) = 𝐿 = lim 𝑕(𝑥) ⇒ lim 𝑔(𝑥) 𝑒𝑥𝑖𝑠𝑡𝑠 𝑎𝑛𝑑 lim 𝑔(𝑥) = L
𝑥→𝐴 𝑥 →𝐴 𝑥→𝐴 𝑥→𝐴

Proof: Pick up any 𝜀 > 0

lim 𝑓(𝑥) = 𝐿 ⇒ ∃𝛿1 > 0/ ∀𝑥 ∈ 𝐸1 , 0 < 𝑑 𝑥, 𝐴 < 𝛿1 ⇒ 𝑓 𝑥 − 𝐿 < 𝜀 ⇒

𝑥→𝐴
⇒ ∃𝛿1 > 0/ ∀𝑥 ∈ 𝐸1 , 0 < 𝑑 𝑥, 𝐴 < 𝛿1 ⇒ −𝜀 + 𝐿 < 𝑓 𝑥 < 𝜀 + 𝐿 (1)
lim 𝑕(𝑥) = 𝐿 ⇒ ∃𝛿2 > 0/ ∀𝑥 ∈ 𝐸1 , 0 < 𝑑 𝑥, 𝐴 < 𝛿2 ⇒ 𝑔 𝑥 − 𝐿 < 𝜀 ⇒
𝑥→𝐴
⇒ ∃𝛿2 > 0/ ∀𝑥 ∈ 𝐸1 , 0 < 𝑑 𝑥, 𝐴 < 𝛿2 ⇒ −𝜀 + 𝐿 < 𝑕 𝑥 < 𝜀 + 𝐿 (2)

Take 𝛿 = 𝑚𝑖𝑛 𝛿1 , 𝛿2 > 0 (𝑏𝑒𝑐𝑎𝑠𝑢𝑠𝑒 𝑡𝑕𝑒 𝑚𝑖𝑛𝑖𝑚𝑢𝑚 𝑜𝑓 𝑎 𝑠𝑒𝑡 𝑏𝑒𝑙𝑜𝑛𝑔𝑠 𝑡𝑜 𝑡𝑕𝑒 𝑠𝑒𝑡)

Pick up any 𝑥 ∈ 𝐸1 such that 0 < 𝑑 𝑥, 𝐴 < 𝛿

0 < 𝑑 𝑥, 𝐴 < 𝛿1 −𝜀 + 𝐿 < 𝑓 𝑥
𝛿 = 𝑚𝑖𝑛 𝛿1 , 𝛿2 ⇒ 𝛿 ≤ 𝛿1 𝑎𝑛𝑑 𝛿 ≤ 𝛿2 (1)
⇒
0 < 𝑑 𝑥, 𝐴 < 𝛿 0 < 𝑑 𝑥, 𝐴 < 𝛿2 𝑕 𝑥 < 𝜀+𝐿 ⇒
(2)
𝐵𝑦 𝑕𝑦𝑝𝑜𝑡𝑕𝑒𝑠𝑖𝑠: 𝑓 𝑥 ≤ 𝑔 𝑥 ≤ 𝑕 𝑥 , ∀𝑥 ∈ 𝐸 − 𝐴
⇒ −𝜀 + 𝐿 < 𝑓 𝑥 ≤ 𝑔 𝑥 ≤ 𝑕 𝑥 < 𝜀 + 𝐿 ⇒ −𝜀 + 𝐿 < 𝑔 𝑥 < 𝜀 + 𝐿 ⇒ 𝑔 𝑥 − 𝐿 < 𝜀

∴ ∀𝜀 > 0, ∃𝛿 > 0/ ∀𝑥 ∈ 𝐸1 , 0 < 𝑑 𝑥, 𝐴 < 𝛿 ⇒ 𝑔 𝑥 − 𝐿 < 𝜀 ⇒ lim𝑥→𝐴 𝑔(𝑥) = 𝐿 

Theorem: 𝐿𝑒𝑡 𝐸, 𝑑 𝑏𝑒 𝑎𝑛𝑦 𝑚. 𝑒. 𝑓: 𝐸 → 𝔼1 . 𝑇𝑕𝑒𝑛:

lim 𝑓(𝑥) = 0 ⇔ lim 𝑓(𝑥) = 0

𝑥→𝐴 𝑥→𝐴
Proof:

⇒ : Pick up any 𝜀 > 0

lim 𝑓(𝑥) = 0 ⇒ ∃𝛿 > 0/ ∀𝑥 ∈ 𝐸1 , 0 < 𝑑 𝑥, 𝐴 < 𝛿 ⇒ 𝑓 𝑥 − 0 < 𝜀

𝑥→𝐴
⇒
∀𝑥 ∈ 𝐸1 , 𝑓 𝑥 − 0 = 𝑓 𝑥 = 𝑓 𝑥 = 𝑓 𝑥 − 0
⇒ ∃𝛿 > 0/ ∀𝑥 ∈ 𝐸1 , 0 < 𝑑 𝑥, 𝐴 < 𝛿 ⇒ 𝑓 𝑥 − 0 < 𝜀

∴ ∀𝜀 > 0, ∃𝛿 > 0/ ∀𝑥 ∈ 𝐸1 , 0 < 𝑑 𝑥, 𝐴 < 𝛿 ⇒ 𝑓 𝑥 − 0 < 𝜀 ⇒ lim𝑥→𝐴 𝑓 𝑥 =0

⇐ : Pick up any 𝜀 > 0

lim 𝑓 𝑥 = 0 ⇒ ∃𝛿 > 0/ ∀𝑥 ∈ 𝐸1 , 0 < 𝑑 𝑥, 𝐴 < 𝛿 ⇒ 𝑓 𝑥 − 0 < 𝜀

𝑥→𝐴 ⇒
∀𝑥 ∈ 𝐸1 , 𝑓 𝑥 − 0 = 𝑓 𝑥 = 𝑓 𝑥 = 𝑓 𝑥 − 0
⇒ ∃𝛿 > 0/ ∀𝑥 ∈ 𝐸1 , 0 < 𝑑 𝑥, 𝐴 < 𝛿 ⇒ 𝑓 𝑥 − 0 < 𝜀

∴ ∀𝜀 > 0, ∃𝛿 > 0/ ∀𝑥 ∈ 𝐸1 , 0 < 𝑑 𝑥, 𝐴 < 𝛿 ⇒ 𝑓 𝑥 − 0 < 𝜀 ⇒ lim𝑥→𝐴 𝑓(𝑥) = 0


Theorem (HW): 𝐿𝑒𝑡 𝐸, 𝑑 𝑏𝑒 𝑎𝑛𝑦 𝑚. 𝑒. 𝑓: 𝐸 → 𝔼1 . 𝑇𝑕𝑒𝑛:

lim 𝑓(𝑥) = 1 𝑜𝑟 − 1 ⇒ lim 𝑓(𝑥) = 1

𝑥→𝐴 𝑥→𝐴
Proof:

Case 1: lim𝑥→𝐴 𝑓 𝑥 = 1

Pick up any 𝜀 > 0

1 1
lim 𝑓(𝑥) = 1 ∃𝛿1 > 0/ ∀𝑥 ∈ 𝐸1 , 0 < 𝑑 𝑥, 𝐴 < 𝛿1 ⇒ − + 1 < 𝑓(𝑥) < + 1 ⇒
𝑥→𝐴
𝜀=
1 2 2
2 0<
⇒ ∃𝛿1 > 0/ ∀𝑥 ∈ 𝐸1 , 0 < 𝑑 𝑥, 𝐴 < 𝛿1 ⇒ 0 < 𝑓 𝑥 (1)

lim 𝑓(𝑥) = 1 ⇒ ∃𝛿2 > 0/ ∀𝑥 ∈ 𝐸1 , 0 < 𝑑 𝑥, 𝐴 < 𝛿2 ⇒ −𝜀 + 1 < 𝑓 𝑥 < 𝜀 + 1 (2)

𝑥→𝐴

Take 𝛿 = 𝑚𝑖𝑛 𝛿1 , 𝛿2 > 0 (𝑏𝑒𝑐𝑎𝑠𝑢𝑠𝑒 𝑡𝑕𝑒 𝑚𝑖𝑛𝑖𝑚𝑢𝑚 𝑜𝑓 𝑎 𝑠𝑒𝑡 𝑏𝑒𝑙𝑜𝑛𝑔𝑠 𝑡𝑜 𝑡𝑕𝑒 𝑠𝑒𝑡)

Pick up any 𝑥 ∈ 𝐸1 such that 0 < 𝑑 𝑥, 𝐴 < 𝛿

0 < 𝑑 𝑥, 𝐴 < 𝛿1 0 < 𝑓 𝑥 ⇒ 𝑓(𝑥) = 𝑓(𝑥)
𝛿 = 𝑚𝑖𝑛 𝛿1 , 𝛿2 ⇒ 𝛿 ≤ 𝛿1 , 𝛿 ≤ 𝛿2 (1)
⇒ ⇒
0 < 𝑑 𝑥, 𝐴 < 𝛿 0 < 𝑑 𝑥, 𝐴 < 𝛿2 −𝜀 + 1 < 𝑓 𝑥 < 𝜀 + 1
(2)
⇒ −𝜀 + 1 < 𝑓 𝑥 <𝜀+1

∴ ∀𝜀 > 0, ∃𝛿 > 0/ ∀𝑥 ∈ 𝐸1 , 0 < 𝑑 𝑥, 𝐴 < 𝛿 ⇒ 𝑓 𝑥 − 1 < 𝜀 ⇒ lim𝑥→𝐴 𝑓 𝑥 =1

Case 2: lim𝑥→𝐴 𝑓 𝑥 = −1

Pick up any 𝜀 > 0

1 1
lim 𝑓(𝑥) = 1 ∃𝛿1 > 0/ ∀𝑥 ∈ 𝐸1 , 0 < 𝑑 𝑥, 𝐴 < 𝛿1 ⇒ − − 1 < 𝑓 𝑥 < − 1 ⇒
𝑥→𝐴
𝜀=
1 2 2
2 <0
⇒ ∃𝛿1 > 0/ ∀𝑥 ∈ 𝐸1 , 0 < 𝑑 𝑥, 𝐴 < 𝛿1 ⇒ 𝑓 𝑥 < 0 (1)

lim 𝑓 𝑥 = 1 ⇒ ∃𝛿2 > 0/ ∀𝑥 ∈ 𝐸1 , 0 < 𝑑 𝑥, 𝐴 < 𝛿2 ⇒ −𝜀 − 1 < 𝑓 𝑥 < 𝜀 − 1 ⇒

𝑥→𝐴
∃𝛿2 > 0/ ∀𝑥 ∈ 𝐸1 , 0 < 𝑑 𝑥, 𝐴 < 𝛿2 ⇒ 𝜀 + 1 > −𝑓 𝑥 > −𝜀 + 1 (2)

Take 𝛿 = 𝑚𝑖𝑛 𝛿1 , 𝛿2 > 0 (𝑏𝑒𝑐𝑎𝑠𝑢𝑠𝑒 𝑡𝑕𝑒 𝑚𝑖𝑛𝑖𝑚𝑢𝑚 𝑜𝑓 𝑎 𝑠𝑒𝑡 𝑏𝑒𝑙𝑜𝑛𝑔𝑠 𝑡𝑜 𝑡𝑕𝑒 𝑠𝑒𝑡)

Pick up any 𝑥 ∈ 𝐸1 such that 0 < 𝑑 𝑥, 𝐴 < 𝛿

0 < 𝑑 𝑥, 𝐴 < 𝛿1 0 > 𝑓 𝑥 ⇒ 𝑓(𝑥) = −𝑓(𝑥)
𝛿 = 𝑚𝑖𝑛 𝛿1 , 𝛿2 ⇒ 𝛿 ≤ 𝛿1 , 𝛿 ≤ 𝛿2 (1)
⇒ ⇒
0 < 𝑑 𝑥, 𝐴 < 𝛿 0 < 𝑑 𝑥, 𝐴 < 𝛿2 𝜀 + 1 > −𝑓 𝑥 > −𝜀 + 1
(2)
⇒ −𝜀 + 1 < 𝑓 𝑥 <𝜀+1

∴ ∀𝜀 > 0, ∃𝛿 > 0/ ∀𝑥 ∈ 𝐸1 , 0 < 𝑑 𝑥, 𝐴 < 𝛿 ⇒ 𝑓 𝑥 − 1 < 𝜀 ⇒ lim𝑥→𝐴 𝑓 𝑥 =1


Theorem: 𝐿𝑒𝑡 𝐸1 , 𝑑1 𝑎𝑛𝑑 𝐸2 , 𝑑2 𝑏𝑒 𝑎𝑛𝑦 𝑚. 𝑒. 𝑓: 𝐸1 → 𝐸2 . 𝑇𝑕𝑒𝑛:

𝑓 𝑖𝑠 𝑐𝑜𝑛𝑡𝑖𝑛𝑢𝑜𝑢𝑠 𝑜𝑣𝑒𝑟 𝐸1 ⇔ ∀𝑈 ⊂ 𝐸2 , 𝑈 𝑖𝑠 𝑜𝑝𝑒𝑛 ⇒ 𝑓 −1 𝑈 𝑖𝑠 𝑜𝑝𝑒𝑛 𝑖𝑛 𝐸1

Proof:

⇒ : Pick up any 𝑈 ⊂ 𝐸2 such that U is open. Consider 𝑓 −1 𝑈 . Pick up any 𝐴 ∈ 𝑓 −1 𝑈 .

𝐴 ∈ 𝑓 −1 𝑈 ⇒ 𝑓(𝐴) ∈ 𝑈
⇒ ∃𝐵 𝑓 𝐴 , 𝜀 𝑠𝑢𝑐𝑕 𝑡𝑕𝑎𝑡 𝐵 𝑓 𝐴 , 𝜀 ⊂ 𝑈 (1)
𝑈 𝑖𝑠 𝑜𝑝𝑒𝑛

𝑓 𝑖𝑠 𝑐𝑜𝑛𝑡𝑖𝑛𝑢𝑜𝑢𝑠 𝑜𝑣𝑒𝑟 𝐸1 ⇒ 𝑓 𝑖𝑠 𝑐𝑜𝑛𝑡𝑖𝑛𝑢𝑜𝑢𝑠 𝑖𝑛 𝐴 ⇒ lim 𝑓 𝑥 = 𝑓 𝐴 ⇒

𝑥→𝐴
⇒ ∃𝛿 > 0/ ∀𝑥 ∈ 𝐸1 , 0 < 𝑑 𝑥, 𝐴 < 𝛿 ⇒ 𝑓(𝑥) ∈ 𝐵(𝑓 𝐴 , 𝜀)
⇒ ∃𝐵 𝐴, 𝛿 / 𝑓 𝐵 𝐴, 𝛿 ⊂ 𝐵(𝑓 𝐴 , 𝜀) ⇒
𝑓(𝐴) ∈ 𝐵(𝑓 𝐴 , 𝜀)
∃𝐵 𝐴, 𝛿 / 𝑓 𝐵 𝐴, 𝛿 ⊂ 𝑈 ⇒ ∃𝐵 𝐴, 𝛿 / 𝐵 𝐴, 𝛿 ⊂ 𝑓 −1 𝑈
(1)

∴ ∀𝐴 ∈ 𝑓 −1 𝑈 , ∃𝐵 𝐴, 𝛿 / 𝐵 𝐴, 𝛿 ⊂ 𝑓 −1 𝑈 ⇒ 𝑓 −1 𝑈 𝑖𝑠 𝑜𝑝𝑒𝑛.

⇐ : Pick up any 𝜀 > 0 . Pick up any 𝐴 ∈ 𝐸1

𝐵 𝑓 𝐴 , 𝜀 𝑖𝑠 𝑜𝑝𝑒𝑛 𝑓 −1 𝐵 𝑓 𝐴 , 𝜀 𝑖𝑠 𝑜𝑝𝑒𝑛
𝐻𝑦𝑝𝑜𝑡 𝑕𝑒𝑠𝑖𝑠
⇒ ∃𝐵 𝐴, 𝛿 ⊂ 𝑓 −1 𝐵 𝑓 𝐴 , 𝜀 ⇒
𝑓 𝐴 ∈ 𝐵 𝑓 𝐴 , 𝜀 ⇒ 𝐴 ∈ 𝑓 −1 𝐵 𝑓 𝐴 , 𝜀
⇒ ∃𝐵 𝐴, 𝛿 / 𝑓 𝐵 𝐴, 𝛿 ⊂ 𝐵 𝑓 𝐴 , 𝜀

∴ ∀𝐴 ∈ 𝐸1 , ∀𝜀 > 0, ∃𝛿 > 0/ ∀𝑥 ∈ 𝐸1 , 𝑥 ∈ 𝐵 𝐴, 𝛿 ⇒ 𝑓(𝑥) ∈ 𝐵 𝑓 𝐴 , 𝜀 ⇒

⇒ ∀𝐴 ∈ 𝐸1 , ∀𝜀 > 0, ∃𝛿 > 0/ ∀𝑥 ∈ 𝐸1 , 0 < 𝑑 𝑥, 𝐴 < 𝛿 ⇒ 𝑓(𝑥) ∈ 𝐵 𝑓 𝐴 , 𝜀 ⇒
⇒ ∀𝐴 ∈ 𝐸1 , lim 𝑓 𝑥 = 𝑓 𝐴 ⇒ 𝑓 𝑖𝑠 𝑐𝑜𝑛𝑡𝑖𝑛𝑢𝑜𝑢𝑠 𝑜𝑣𝑒𝑟 𝐸1
𝑥→𝐴


Corollary: 𝐿𝑒𝑡 𝐸1 , 𝑑1 , 𝐸2 , 𝑑2 𝑎𝑛𝑑 𝐸3 , 𝑑3 𝑏𝑒 𝑎𝑛𝑦 𝑚. 𝑒.

𝑓 𝑔
𝐸1 → 𝐸2 → 𝐸3
𝑓 𝑖𝑠 𝑐𝑜𝑛𝑡𝑖𝑛𝑢𝑜𝑢𝑠 𝑜𝑣𝑒𝑟 𝐸1 ⇒ 𝑔 ∘ 𝑓 𝑖𝑠 𝑐𝑜𝑛𝑡𝑖𝑛𝑢𝑜𝑢𝑠 𝑜𝑣𝑒𝑟 𝐸1
𝑔 𝑖𝑠 𝑐𝑜𝑛𝑡𝑖𝑛𝑢𝑜𝑢𝑠 𝑜𝑣𝑒𝑟 𝐸2

Proof: Pick up any 𝑈 ⊂ 𝐸3 such that U is open. Since g is continuous, 𝑔 −1 (𝑈) is open. Then,
since f is continuous, 𝑓 −1 𝑔 −1 (𝑈) is open. Notice that 𝑓 −1 𝑔−1 (𝑈) = 𝑔 ∘ 𝑓 −1 (𝑈).

∴ ∀𝑈 ⊂ 𝐸3 , 𝑈 𝑖𝑠 𝑜𝑝𝑒𝑛 ⇒ 𝑔 ∘ 𝑓 −1 𝑈 𝑖𝑠 𝑜𝑝𝑒𝑛 𝑖𝑛 𝐸1 ⇒ 𝑔 ∘ 𝑓 is continuous over 𝐸1 

Corollary: 𝐿𝑒𝑡 𝐸𝑖 , 𝑑𝑖 𝑏𝑒 𝑎𝑛𝑦 𝑚. 𝑒. , ∀𝑖 = 1, … , 𝑛, 𝑛 + 1

𝑓1 𝑓2 𝑓3 𝑓𝑛 −1 𝑓𝑛
𝐸1 → 𝐸2 → 𝐸3 → … 𝐸𝑛 → 𝐸𝑛+1 ⇒ 𝑓𝑛 ∘ … ∘ 𝑓1 𝑖𝑠 𝑐𝑜𝑛𝑡𝑖𝑛𝑢𝑜𝑢𝑠 𝑜𝑣𝑒𝑟 𝐸1
𝑓𝑖 𝑖𝑠 𝑐𝑜𝑛𝑡𝑖𝑛𝑢𝑜𝑢𝑠 𝑜𝑣𝑒𝑟 𝐸𝑖 , ∀𝑖 = 1, … , 𝑛
Proof:
INDUCTIVE BASE: 𝑛 = 2 Proved by corollary above.

INDUCTIVE STEP: If 𝑓𝑛−1 ∘ … ∘ 𝑓1 𝑖𝑠 𝑐𝑜𝑛𝑡𝑖𝑛𝑢𝑜𝑢𝑠 ⇒ 𝑓𝑛 ∘ … ∘ 𝑓1 𝑖𝑠 𝑐𝑜𝑛𝑡𝑖𝑛𝑢𝑜𝑢𝑠

Proof: Pick up any 𝑈 ⊂ 𝐸𝑛+1 such that U is open. Since 𝑓𝑛 is continuous, 𝑓𝑛 −1 (𝑈) is open. Then,
since 𝑓𝑛−1 ∘ … ∘ 𝑓1 is continuous, 𝑓𝑛−1 ∘ … ∘ 𝑓1 −1 𝑓𝑛 −1 (𝑈) 𝑖𝑠 𝑜𝑝𝑒𝑛. Notice that 𝑓𝑛−1 ∘ … ∘
𝑓1 −1 𝑓𝑛 −1 (𝑈) = 𝑓𝑛 ∘ … ∘ 𝑓1 −1 (𝑈).

∴ ∀𝑈 ⊂ 𝐸3 , 𝑈 𝑖𝑠 𝑜𝑝𝑒𝑛 ⇒ 𝑓𝑛 ∘ … ∘ 𝑓1 −1 𝑈 𝑖𝑠 𝑜𝑝𝑒𝑛 𝑖𝑛 𝐸1 ⇒ 𝑓𝑛 ∘ … ∘ 𝑓1 is continuous



Corollary: 𝐿𝑒𝑡 𝐸1 , 𝑑1 , 𝐸2 , 𝑑2 𝑎𝑛𝑑 𝐸3 , 𝑑3 𝑏𝑒 𝑎𝑛𝑦 𝑚. 𝑒.

𝑓 𝑔
𝐸1 → 𝐸2 → 𝐸3
lim 𝑓(𝑥) = 𝑘 (𝑒𝑥𝑖𝑠𝑡𝑠) ⇒ lim 𝑔 𝑓(𝑥) = 𝑔 lim 𝑓(𝑥)
𝑥→𝐴 𝑥→𝐴 𝑥→𝐴
𝑔 𝑖𝑠 𝑐𝑜𝑛𝑡𝑖𝑛𝑢𝑜𝑢𝑠 𝑎𝑡 𝑘 ∈ 𝐸2

Proof: Pick up any 𝜀 > 0.

𝑔 𝑖𝑠 𝑐𝑜𝑛𝑡𝑖𝑛𝑢𝑜𝑢𝑠 𝑎𝑡 𝑘 ⇒ lim 𝑔 𝑥 = 𝑔 𝑘
𝑥 →𝑘 ⇒
𝜀>0
⇒ ∃𝛿1 > 0/ ∀𝑥 ∈ 𝐸2 , 0 < 𝑑 𝑥, 𝑘 < 𝛿1 ⇒ 𝑑 𝑔 𝑥 , 𝑔 𝑘 < 𝜀 (1)
lim 𝑓(𝑥) = 𝑘
𝑥→𝐴 ⇒ ∃𝛿2 > 0/ ∀𝑥 ∈ 𝐸1 , 0 < 𝑑 𝑥, 𝐴 < 𝛿2 ⇒ 𝑑 𝑓 𝑥 , 𝑘 < 𝛿1 (2)
𝛿1 > 0

Take any 𝑥 ∈ 𝐸1 , 𝑠𝑢𝑐𝑕 𝑡𝑕𝑎𝑡 0 < 𝑑 𝑥, 𝐴 < 𝛿2 .

0 < 𝑑 𝑥, 𝐴 < 𝛿2 𝑑 𝑓 𝑥 , 𝑘 < 𝛿1 𝑑 𝑔 𝑓(𝑥) , 𝑔 𝑘 <𝜀
(2) (1)

∴ ∀𝜀 > 0, ∃𝛿2 > 0/ ∀𝑥 ∈ 𝐸1 , 0 < 𝑑 𝑥, 𝐴 < 𝛿2 ⇒ 𝑑 𝑔 𝑓(𝑥) , 𝑔 𝑘 <𝜀 ⇒

⇒ lim 𝑔 𝑓(𝑥) = 𝑔(𝑘)
𝑥→𝐴
⇒ lim 𝑔 𝑓(𝑥) = 𝑔 lim 𝑓(𝑥)
lim 𝑓(𝑥) = 𝑘 𝑥→𝐴 𝑥→𝐴
𝑥→𝐴


9.1 CONTINUOUS FUNCTIONS IN COMPACT SUBSETS

Theorem: 𝐿𝑒𝑡 𝑓: 𝐸1 → 𝐸2 𝑏𝑒 𝑐𝑜𝑛𝑡𝑖𝑛𝑢𝑜𝑢𝑠 𝑜𝑣𝑒𝑟 𝐸1 . 𝐴 ⊂ 𝐸1 . 𝐴 𝑖𝑠 𝑐𝑜𝑚𝑝𝑎𝑐𝑡 ⇒ 𝑓 𝐴 𝑖𝑠 𝑐𝑜𝑚𝑝𝑎𝑐𝑡

Proof: Take any 𝑈𝜆 ; 𝜆 ∈ 𝐼 ⊂ 𝐸2 open cover of 𝑓(𝐴), and consider the set
𝐵 = 𝑓 −1 (𝑈𝜆 ); 𝜆 ∈ 𝐼 ⊂ 𝐸1 .

Claim 1: B is an open cover of A.

Proof: By hypothesis f is continuous. Also, by construction, ∀𝜆 ∈ 𝐼, 𝑈𝜆 is open. Then, by CN&S:

∀𝜆 ∈ 𝐼, 𝑓 −1 (𝑈𝜆 ) is an open set. Then B is a collection of open sets (1)
𝑃𝑖𝑐𝑘 𝑢𝑝 𝑎𝑛𝑦 𝑥 ∈ 𝐴 ⇒ 𝑓 𝑥 ∈ 𝑓 𝐴
⇒ ∃𝜆0 ∈ 𝐼/𝑓(𝑥) ∈ 𝑈𝜆 0 ⇒ 𝑥 ∈ 𝑓 −1 (𝑈𝜆 0 ) ⇒ 𝑥 ∈ 𝜆∈𝐼 𝑓
−1 (𝑈 ).
𝜆
𝑈𝜆 ; 𝜆 ∈ 𝐼 𝑖𝑠 𝑜𝑝𝑒𝑛 𝑐𝑜𝑣𝑒𝑟 𝑜𝑓 𝑓(𝐴)
∴ 𝐴 ⊂ 𝜆∈𝐼 𝑓 −1 (𝑈𝜆 ). (2)

From (1) and (2): B is an open cover of A.

𝐵 = 𝑓 −1 (𝑈𝜆 ); 𝜆 ∈ 𝐼 𝑖𝑠 𝑎𝑛 𝑜𝑝𝑒𝑛 𝑐𝑜𝑣𝑒𝑟 𝑜𝑓 𝐴 𝑖=𝑘 −1

⇒ ∃𝜆1 , … , 𝜆𝑘 ∈ 𝐼, 𝑠𝑢𝑐𝑕 𝑡𝑕𝑎𝑡 𝐴 ⊂ 𝑖=1 𝑓 (𝑈𝜆 𝑖 ).
𝐵𝑦 𝑕𝑦𝑝𝑜𝑡𝑕𝑒𝑠𝑖𝑠: 𝐴 𝑖𝑠 𝑐𝑜𝑚𝑝𝑎𝑐𝑡

𝑖=𝑘
Claim 2: 𝑓(𝐴) ⊂ 𝑖=1 𝑈𝜆 𝑖

Proof: 𝑃𝑖𝑐𝑘 𝑢𝑝 𝑎𝑛𝑦 𝑦 ∈ 𝑓 𝐴 ⇒ ∃𝑥 ∈ 𝐴/ 𝑓(𝑥) = 𝑦

⇒ 𝑦 ∈ 𝑈𝜆 𝑗 
𝑥 ∈ 𝐴 𝑎𝑛𝑑𝐴 ⊂ 𝑖=𝑘 −1 −1
𝑖=1 𝑓 (𝑈𝜆 𝑖 ) ⇒ ∃𝑗 = 1, … , 𝑘/𝑥 ∈ 𝑓 (𝑈𝜆 𝑗 ) ⇒ 𝑓(𝑥) ∈ 𝑈𝜆 𝑗

∴ 𝑓𝑜𝑟 𝑎𝑛𝑦 𝑜𝑝𝑒𝑛 𝑐𝑜𝑣𝑒𝑟 𝑜𝑓 𝑓 𝐴 , 𝑡𝑕𝑒𝑟𝑒 𝑒𝑥𝑖𝑠𝑡 𝑎 𝑓𝑖𝑛𝑖𝑡𝑒 𝑠𝑢𝑏𝑜𝑝𝑒𝑛 𝑐𝑜𝑣𝑒𝑟 𝑜𝑓 𝑓 𝐴 .

𝑇𝑕𝑒𝑛 𝑓 𝐴 𝑖𝑠 𝑐𝑜𝑚𝑝𝑎𝑐𝑡. 

Corollary: 𝐿𝑒𝑡 𝑓: 𝐸1 → 𝐸2 𝑏𝑒 𝑐𝑜𝑛𝑡𝑖𝑛𝑢𝑜𝑢𝑠 𝑜𝑣𝑒𝑟 𝐸1 . 𝐴 ⊂ 𝐸1 . 𝐴 𝑖𝑠 𝑐𝑜𝑚𝑝𝑎𝑐𝑡 ⇒ 𝑓 𝐴 𝑖𝑠 𝑏𝑑𝑑

Proof: By theorem before, 𝑓(𝐴) is compact, and since we’ve proved that any compact set is
bdd, 𝑓(𝐴) is bdd. 

Corollary: 𝐿𝑒𝑡 𝑓: 𝐸1 → 𝔼1 𝑏𝑒 𝑐𝑜𝑛𝑡𝑖𝑛𝑢𝑜𝑢𝑠 . 𝐴 ⊂ 𝐸1 . 𝐴 𝑖𝑠 𝑐𝑜𝑚𝑝𝑎𝑐𝑡 ⇒

∃𝑎 ∈ 𝐴 𝑠𝑢𝑐𝑕 𝑡𝑕𝑎𝑡 𝑓 𝑎 = max 𝑓(𝐴)
⇒
∃𝑏 ∈ 𝐴 𝑠𝑢𝑐𝑕 𝑡𝑕𝑎𝑡 𝑓 𝑏 = min 𝑓(𝐴)

Proof: By theorem before, 𝑓 𝐴 is compact. Apply Hahn-Banach theorem to 𝑓 𝐴 ⊂ 𝔼1 . Then

𝑓 𝐴 is closed and bdd. Then, by theorem, 𝑓 𝐴 𝑕𝑎𝑠𝑚𝑖𝑛𝑖𝑚𝑢𝑚 𝑎𝑛𝑑 𝑚𝑎𝑥𝑖𝑚𝑢𝑚 𝑒𝑙𝑒𝑚𝑒𝑛𝑡𝑠.

Call 𝑘 = max 𝑓(𝐴) ⇒ 𝑘 ∈ 𝑓(𝐴) ⇒ ∃𝑎 ∈ 𝐴 𝑠𝑢𝑐𝑕 𝑡𝑕𝑎𝑡 𝑓 𝑎 = 𝑘 = max 𝑓(𝐴)

Call 𝑕 = min 𝑓(𝐴) ⇒ 𝑕 ∈ 𝑓(𝐴) ⇒ ∃𝑏 ∈ 𝐴 𝑠𝑢𝑐𝑕 𝑡𝑕𝑎𝑡 𝑓 𝑏 = 𝑕 = min 𝑓(𝐴)


9.2 DERIVATIVES

𝐏𝐔𝐍𝐓𝐔𝐀𝐋 𝐃𝐄𝐑𝐈𝐕𝐀𝐓𝐈𝐎𝐕𝐄 𝐎𝐅 𝐀 𝐅𝐔𝐍𝐂𝐓𝐈𝐎𝐍: 𝐿𝑒𝑡 𝑓: 𝔼1 → 𝔼1 be a function. 𝐴 ∈ 𝔼1 . We say

𝑓 𝐴+𝑕 −𝑓(𝐴)
that f is derivable at A iff lim𝑕→0 exists.
𝑕
In such case we say that limit is the first derivative of f at A and we notate it as 𝑓 ′ (𝐴)

Theorem (HW): (ARITHMETIC PROPERTIES OF THE DERIVATIVES)

𝑖) 𝛼. 𝑓 ′ 𝐴 = 𝛼. 𝑓 ′ 𝐴
𝑓, 𝑔: 𝔼1 → 𝔼1 .
⇒ 𝑖𝑖) 𝑓 ± 𝑔 ′ 𝐴 = 𝑓 ′ 𝐴 ± 𝑔 ′ 𝐴 .
𝑓 𝑎𝑛𝑑 𝑔 𝑑𝑒𝑟𝑖𝑣𝑎𝑏𝑙𝑒𝑠 𝑎𝑡 𝐴
𝑖𝑖𝑖) 𝑓. 𝑔 ′ 𝐴 = 𝑓 ′ 𝐴 𝑔 𝐴 + 𝑔′(𝐴)𝑓(𝐴)
Proof:

′ 𝛼.𝑓 𝐴+𝑕 − 𝛼.𝑓 𝐴 𝛼 .𝑓 𝐴+𝑕 −𝛼.𝑓 𝐴 𝑓 𝐴+𝑕 −𝑓 𝐴

.𝑖) 𝛼. 𝑓 𝐴 = lim = lim𝑕→0 = lim𝑕→0 𝛼. =
𝑕→0 𝑕 𝑕 𝑕
𝑓 𝐴+𝑕 −𝑓(𝐴)
.= 𝛼. lim𝑕→0 = 𝛼. 𝑓 ′ 𝐴
𝑕

′ 𝑓±𝑔 𝐴+𝑕 − 𝑓±𝑔 𝐴 𝑓 𝐴+𝑕 ±𝑔 𝐴+𝑕 − 𝑓 𝐴 ±𝑔(𝐴)

.𝑖𝑖) 𝑓 ± 𝑔 𝐴 = lim𝑕→0 = lim𝑕→0 =
𝑕 𝑕
𝑓 𝐴+𝑕 −𝑓 𝐴 𝑔 𝐴+𝑕 −𝑔 𝐴 𝑓 𝐴+𝑕 −𝑓(𝐴) 𝑔 𝐴+𝑕 −𝑔(𝐴)
lim𝑕→0 ± = lim𝑕→0 ± lim𝑕→0 = 𝑓 ′ 𝐴 ± 𝑔′ 𝐴
𝑕 𝑕 𝑕 𝑕

′ 𝑓.𝑔 𝐴+𝑕 − 𝑓.𝑔 𝐴 𝑓 𝐴+𝑕 .𝑔 𝐴+𝑕 −𝑓 𝐴 .𝑔(𝐴)

.𝑖𝑖𝑖) 𝑓. 𝑔 𝐴 = lim𝑕→0 = lim𝑕→0 =
𝑕 𝑕
𝑓 𝐴+𝑕 .𝑔 𝐴+𝑕 +𝑓 𝐴+𝑕 .𝑔 𝐴 −𝑓 𝐴+𝑕 .𝑔(𝐴)−𝑓 𝐴 .𝑔(𝐴)
lim𝑕→0 =
𝑕
𝑓 𝐴+𝑕 𝑔 𝐴+𝑕 −𝑔 𝐴 +𝑔(𝐴) 𝑓 𝐴+𝑕 −𝑓 𝐴
lim𝑕→0 =
𝑕
𝑔 𝐴+𝑕 −𝑔(𝐴) 𝑓 𝐴+𝑕 −𝑓(𝐴)
lim𝑕→0 𝑓(𝐴 + 𝑕) + lim𝑕→0 𝑔(𝐴) =
𝑕 𝑕
𝑔 𝐴 + 𝑕 − 𝑔(𝐴) 𝑓 𝐴 + 𝑕 − 𝑓(𝐴)
lim 𝑓(𝐴 + 𝑕) . lim + 𝑔 𝐴 . lim = 𝑓 ′ 𝐴 𝑔 𝐴 + 𝑔′(𝐴)𝑓(𝐴)
𝑕→0 𝑕 𝑕→0 𝑕
𝑕 →0
