Logo

Sets

Fundamentals

We define first what a set is:

(Sets)

We define set A to be any unordered collection of objects. If xx is an object, we say that x is an element of A or xAx \in A if x lies in the collection. Otherwise xAx \in A

We start with some axioms:

  1. (Sets are objects) If AA is a set, then AA is also an object. A side track about "Pure Set Theory" - This theory states that everything in the mathematical universe is a set. We can write 0 as or an empty set, 1 can be written as 0 and 2 as 0,1 and so on. Terence Tao argues that they are the 'cardinalities of the set.'

  2. (Equality of sets) Two sets A and B are equal, A = B, iff every element of A is an element of B. A = B, if and only if every element of xx of A also belongs to B, and every element yy of B belongs to A.

  3. (Empty set) There exists a set \emptyset known as the empty set, which contains no elements. xx \notin \emptyset

(Partial Order)

If AB,BCACA \subseteq B, B \subseteq C \Rightarrow A \subseteq C

Proof. If xAx \in A, then xBx \in B, If xBx \in B, then xCx \in C, Then xAx \in A, then xCx \in C

Thus, ACA \subseteq C

Let AA be a non-empty set. Then there exists an object xx such that xAx \in A
Proof. Proving by contradiction, Suppose there is no object xx that belongs to A. For all xx, we have xAx \notin A. We know from Axiom 3, that xx \notin \emptyset

For the statement,

xAxx \in A \Leftrightarrow x \in \emptyset

Is false both ways, which gives us the result true, which is a contradiction.

Thus we also prove that \emptyset is unique.

  1. (Singleton sets and pair sets) If aa is an object, then there exists a set {a}\{a\} whose only element is aa, i.e. for every object yy, we have y{a}y \in \{a\} if and only if y=ay=a; we refer to {a}\{a\} as the singleton set whose element is aa. Furthermore if aa and bb are objects, then there exists a set {a,b}\{a,b\} if and only if y=ay=a or y=by=b, we refer to this set as the pair set formed by aa and bb.

  2. (Single choice) aa is an object, {a}\{a\}. y{a}y \in \{a\}, y = a

  3. (Pairwise Union) AB={x:xA or xB}A \cup B = \{x : x \in A \text{ or } x \in B\}

A(BC)=(AB)CA \cup (B \cup C) = (A \cup B) \cup C
Proof. Taking the left hand side, We have xAx \in A or x(BC)x \in (B \cup C). If we look to the right hand side, we have x(AB)x \in (A \cup B) or xCx \in C If we break the statement down further. We have xAx \in A or xBx \in B or xCx \in C, and on the right xAx \in A or xBx \in B or xCx\in C

The two statements are equivalent.

  1. (Axiom Of Specification) A, xAx \in A, let P(x) be a property pertaining to xx. Then there exists a set called {xA,P(x)is true}\{x \in A, P(x) \text{is true}\} whose elements are precisely the elements xx in A for which P(x)P(x) is true.

  2. (Replacement) Let A be a set, for any object xAx \in A, and any object yy, suppose we have a statement P(x,y)P(x,y) pertaining to xx and yy, such that for each xAx \in A there is at most one yy for which P(x,y)P(x,y) is true. Then there exists a set {y:P(x,y)\{y : P(x,y) is true for some xA}x \in A\}

  3. (Infinity) There exists a set N\mathbb{N}, whose elements are called natural numbers, as well as an object 00 in N\mathbb{N}, and an object n++n++ assigned to every natural number nNn \in \mathbb{N} such that the Peano axioms hold.

  4. Russel's Paradox (Axiom Of Universal Specification) Suppose for every xx we have a property P(x)P(x) pertaining to xx, Then there exists a set {x:P(x)\{x : P(x) is true}\} such that for every object yy:

y{x:P(x)is true}P(y)is true.y \in \{ x : P(x) \text{is true} \} \Leftrightarrow P(y) \text{is true.}

This is quite a handy axiom, we can even find all the axioms through this axiom, but there's an issue.

Let's say we defined the property P(x)x is a set, and xxP(x) \Leftrightarrow \text{x is a set, and } x \notin x And we defined a set Ω\Omega for which this property is true. A paradox is created, where we don't know whether Ω\Omega belongs in this set or not, it's simultaneously true and untrue. If Ω\Omega is not in the set, then it is a set that is not in its own set, thereby being contained in the set. If Ω\Omega is in the set, then it's a set that is in itself, therefore it cannot be in the set. To resolve this paradox we have, the next axiom.

There is a similar paradox known as the "Grellig-Nelson paradox" with the words heterological and autological, the word autological means that the word is an example of itself, for example, the word 'word' is a word or 'pronounceable' is pronounceable or a 'noun' is a noun. The word heterological means that the word is not an example of itself. The word 'triangle' is clearly not a \triangle and so on. The question arises, is heterological autological or heterological?

(Deriving everything from the axiom of universal specification)

The Axiom Of Universal Specification implies the third axiom onwards.

Proof. Proving for each axiom:

  1. For the axiom of null sets,

    We can construct null sets in many ways. We can define the set such that:

{x:xx}\{ x : x \ne x \}

A property that is not true for any given object. This allows us to then say that, for P(y)P(y) is true, where yϕy \in \phi then that means yyy \neq y which is not true. The third axiom is true.

  1. For the axiom of single choice,

    We define the property P(x)P(x) such that x=ax = a,

    Then for y{x:P(x)is true}y \in \{x : P(x) \text{is true}\}, then yAy \in A

  2. For the axiom of pair sets,

    We define the property similarly P(x)P(x) such that x=ax = a or x=bx = b

    For the axiom of pairwise unions,

    We take P(x)P(x) to be xAx \in A or xBx \in B For the axiom of specification,

    We define a property Q(x):xA,P(x)Q(x) : x \in A, P(x) is true.

    Then we have,

y{x:xA,Q(x)is true}y \in \{x : x\in A, Q(x) \text{is true}\}

This means that when Q(y)Q(y) is true, yAy \in A P(y)P(y) is true

  1. For the axiom of replacement, We define Q(y):xA,P(x,y)istrueQ(y): x \in A, P(x,y) is true

    z{y:Q(y) is true}Q(z)is true or P(x,z)is truez \in \{y: Q(y) \text{ is true} \}\Leftrightarrow Q(z) \text{is true or } P(x,z) \text{is true}
  2. For the axiom of infinity, We take the property P(x)P(x) that xx is a natural number.

  1. (Axiom Of Regularity) If A is a non-empty set, then there is at least one element xx of A which is either not a set or is disjoint from A. How exactly does this mitigate the issues that come in from Russel's Paradox?

(A set cannot contain itself.)

Show that if A is a set then AAA \notin A

Proof. We take the set A and form a singleton set, {A}\{A\}

A, if x{A}x=Ax \in \{A\} \Rightarrow x = A

A{A}=ϕA \cap \{A\} = \phi

AA\Rightarrow A \notin A

(One of a pair sets cannot contain the other)

If A and B are two sets then either ABA \notin B or BAB \notin A

Proof. Let's say we have A,BA,B such that x{A,B}x=Ax \in \{A,B\} \Rightarrow x = A or x=Bx = B

x = A,

A{A,B}=ϕA \cap \{A,B\} = \phi BA{A,B}B \notin A \cup \{A,B\} B{A,B}B \in \{A,B\}

So if A{A,B}=ϕA \cap \{A,B\} = \phi Then we know that,

BAB \notin A

Because if it were then we'd haveA{A,B}=BA \cap \{A,B\} = B

(Intersection)
AB={x:xA and xB}A \cap B = \{x: x \in A \text{ and } x \in B \}
(Existence of universal set)

The axiom of universal specification is equivalent to an axiom postullating the existence of a universal set consisting of all objects (for all objects xx we have xΩx \in \Omega). Conversely if a universal set exists, then the axiom of universal specification is true

Proof. We can say from the axiom of universal specification,

Ω={x:x is an object}\Omega = \{ x : x \text{ is an object}\}

Conversely,

Assuming the universal set exists, we can write

{xΩ,P(x) is true }\{x \in \Omega, P(x) \text{ is true }\}

Since every object belongs to the universal set, this gives us the axiom of universal specificiation

(Replacement's use case)

Show that the axiom of replacement implies the axiom of specification

Proof. We define the property P(x,y)P(x,y) to be y=xy=x Thus, we get the axiom of specification. Since y=xy=x we are saying something about P(x)P(x)
(Proper Subsets)

Define a proper subset of a set AA to be a subset BB of AA with BAB \neq A. Let AA be a non-empty set. Show that AA does not have any non-empty proper subsets if and only if AA is of the form A={x}A = \{x\} for some object xx.

Proof. Proving the converse statement is relatively straightforward,

Given that, A={x}A = \{x\}

We must prove that there are no non-empty proper subsets of AA.

A proper subset of AA is defined as AB,ABA \subseteq B, A \neq B,

Suppose there is a non-empty proper subset BB of AA

yByA={x}y=x\begin{aligned} y \in B & \Rightarrow y \in A = \{x\} \\ & \Rightarrow y = x \end{aligned}

Which would mean,

B=A or B=ϕ\begin{aligned} B & = A \text{ or } \\ B & = \phi \\ \end{aligned}

Which is a contradiction.

For the statement, If A does not have any non-empty proper subsets, then A is of the form A={x}A = \{x\},

Proving by contradiction,

Suppose that A does not have any non-empty proper subsets and A is not singleton.

This means that the only proper subset of AA has to be ϕ\phi.

But we know from Axiom 4, that all sets are built off of singleton sets, which would be a non-empty proper subset of that set AA.

Thus the statement is true, proven by contradiction

("")

Suppose that AA,BB, AA', BB' are sets such that AAA\prime \subseteq A and BBB' \subseteq B. Show that ABABA' \cup B \subseteq A \cup B and ABABA' \cap B \subseteq A \cap B

Proof. 1. ABABA' \cup B \subseteq A \cup B
This means xAorxBx \in A' or x \in B

We already know that AAA' \subseteq A or if xAx \in A', then xAx \in A. This means, we can say

ABAB=xAorxBABAB=AB\begin{aligned} A' \cup B \subseteq A \cup B & = x \in A or x \in B \\ A' \cup B \subseteq A \cup B & = A \cup B \\ \end{aligned}

When we take out the Axiom Of Universal Specification, we have a set of axioms known as "Zermelo- Fraenkel Set Theory".

We will discuss another axiom, known as the "Axiom of choice" in upcoming sessions, which allows us to talk about unions and intersections of sets that aren't countable.

Functions

(Function)

Let A,BA,B be sets and let P(x,y)P(x,y) be a property pertaining to an object xXx \in X and an object yYy \in Y such that for every xXx \in X, there is exactly one yYy \in Y for which P(x,y)P(x,y) is true. Then we define the function f:XYf:X \rightarrow Y defined by P on the domain XX and the codomain to be the object which, given any input xXx \in X, assigns an output f(x)Yf(x) \in Y defined to be the unique object f(x)Yf(x) \in Y for which P(x,f(x))P(x, f(x)) is true. Thus for any xXx \in X and yYy \in Y

y=f(x)P(x,y)y = f(x) \Leftrightarrow P(x,y)
(One-to-one Function)

A function ff is one-to-one (or injective) if different elements map to different elements:

xxf(x)f(x)x \ne x' \Rightarrow f(x) \ne f(x')
(Onto Functions)

A function ff is onto if every element in YY comes from applying ff to some element in XX:

For every yYy \in Y there exists xXx \in X such that f(x)=yf(x) = y

These functions are extremely important for modelling the real world. We represent bodies as a set of points. And the way we model the real world is by making these sets have a one-to-one mapping with the Euclidean space.

(Bijective functions)

Functions f:XYf: X \rightarrow Y which are both one-one and onto are called bijective.

(Composition Of Functions)

Let f:XYf: X \rightarrow Y and g:YZg: Y \rightarrow Z be two functions such that the codomain fo f is the same set as the domain of gg. We then define the composition gof:XZg o f: X \rightarrow Z of the two functions gg and ff to be the function defined explicitly by the formula

(gof)(x):=g(f(x))(g o f)(x) := g(f(x))

If the codomain of ff does not match the domain of gg, we leave the composition gofg o f undefined.

(Equality Of Functions)

Two functions are said to be equal when the functions have the same domain, codomain and the functions have the same output for all xDomainx \in \text{Domain}

(Inverse)

Let f:XYf: X \rightarrow Y be a bijective function and let f1:YXf^{-1}: Y \rightarrow X be its inverse. Verify the cancellation laws f1(f(x))=xf^{-1}(f(x)) = x for all xXx \in X and f(f1(y))=yf(f^{-1}(y))= y for all yYy \in Y Conclude that f1f^{-1} is also invertible and has ff as its inverse thus (f1)1=f{(f^{-1})}^{-1} = f

(Images)

If fXYf X \rightarrow Y is a function from XX to YY, and S is a subset of XX, we define f(S)f(S) to be the set

f(S):={f(x):xS};f(S) := \{f(x): x \in S\};
(Inverse images)

If U Is a subset of YY, we define the set f1(U)f^{-1}(U) to be the set,

f1(U):={xX:f(x)U}tobef^{-1}(U) := \{x \in X: f(x) \in U\}to be

This leads to the introduction of a new Axiom.

The Power set Axiom.

There exists a set YXY^X such that all the functions ff that map from XX to YY are contained within it.

(Union)

Let A be a set, all of whose elements are themselves sets, then there exists a set UAUA whose elements are precisely those objects which are elements of the elements of A, thos for objects xx

xA(xS for some SA)x \in \bigcup A \Leftrightarrow (x \in S \text{ for some } S \in A)

Similarly, intersections can be described as,

(Intersection)
xA(xS for all SA)x \in \bigcap A \Leftrightarrow (x \in S \text{ for all } S \in A)

Cartesian Products

(Ordered Pair)

If xx and yy are any objects, we define the ordered pair (x,y)(x,y) to be a new object, consisting of xx as its first component and yy as its second component.

(Cartesian Product)

If XX and YY are sets, then we define the Cartesian product X×YX \times Y to be the collection of ordered pairs, whose first component lies in XX and second component lies in YY, thus,

X×Y=(x,y):xX,yYX \times Y = (x,y): x \in X, y \in Y
(Ordered n-tuple)

An ordered n-typle is a collection of objects xix_i, one for every natural number between ii and nn; we refer to xix_i as the ithi^{th} component of the n-tuple \prod

CC BY-SA 4.0 Adithya Nair.
Last modified: March 04, 2025.
Website built with Franklin.jl and the Julia programming language.