zettelkasten

Search IconIcon to open search
Dark ModeDark Mode

Relations

Based on 21-127 Concepts of Mathematics

H2 Defining Relation

A relation applies to a set and talks about how elements in that set are related.

Formally, let there be set XX and a relation RR on the set XX. Then RR is a declaration for each pair (a,b)X2(a, b) \in X^2 as to whether they are related or not. If aa is related to bb, we write aRba \mathrel{R} b, otherwise we write aba \\ \mathrel{\not R} \\ b.

H3 Relation extensionality axiom

This is similar to the Pure Math - Functions

. Basically, we consider relations to be equal iff they relate the same pairs of elements (duh). Symbolically, RR and SS on XX are equal iff a,bX,(aRbaSb)\forall a, b \in X, (a \mathrel{R} b \Leftrightarrow a \mathrel{S} b).

H3 Graph of relation

Let there be set XX and a relation RR on the set XX. The graph of RR is the set of pairs (a,b)X2(a, b) \in X^2 such that aRba \mathrel{R} b. That is, Gr(R)=(a,b)X2aRbX2\mathrm{Gr}(R) = { (a,b) \in X^2 \mid a \mathrel{R} b } \subseteq X^2
Relations are always well defined :)

Another way to define relation equality, therefore, is to say that they have the same graph:
R=SGr(R)=Gr(S)R = S \Leftrightarrow \mathrm{Gr}(R) = \mathrm{Gr}(S)
where RR and SS are relations on XX.

H3 Empty relation

We can have relation EE with Gr(E)=\mathrm{Gr}(E) = \varnothing. This is the same thing as saying nothing is related.

H2 Properties of relations

Let RR be a relation on XX. RR can have these properties

Reflexivity: RR is reflexive if xX,xRx\forall x \in X, x \mathrel{R} x. That is, every element is related to itself

Symmetry: RR is symmetric if a,bX,(aRb)(bRa)\forall a, b \in X, (a \mathrel{R} b) \Leftrightarrow (b \mathrel{R} a). That is, relation always go in two directions.

Antisymmetry: RR is antisymmetric if a,bX,(aRbbRa)(a=b)\forall a, b \in X, (a \mathrel{R} b \wedge b \mathrel{R} a) \Rightarrow (a = b). That is, things related in both direction are equal.

Transitivity: RR is transitive if a,b,cX,(aRbbRc)(aRc)\forall a,b,c \in X, (a \mathrel{R} b \wedge b \mathrel{R} c) \Rightarrow (a \mathrel{R} c). Think of it as we can somehow chain relations.

H3 Proving properties of relations

The usual strategy is just to show the relation satisfy a property.

H2 Equivalence Relation

An equivalence relation is a relation that is reflexive, symmetric, and transitive. (Note it could be antisymmetry too)

H3 Equivalence class

Given equivalence relation \sim on XX and element xXx \in X, we can create a set of all elements that xx is related to by \sim. We call this a equivalence class. This set [a][a]_{\sim} is defined by aXax{ a \in X \mid a \sim x }.

Notice that this means the same equivalence class could be represented using different element of XX, which we call representatives. Take the set 0,1,2,3{ 0, 1, 2, 3 } and the relation 2\equiv_{2}, then $[0]{\equiv{2}} = [2]{\equiv{2}} = { 0,2 }and and [1]{\equiv{2}} = [3]{\equiv{2}} = { 1,3 }$.

H3 Quotient

The quotient is defined as X/=UXaX,U=[a]X/{\sim} = { U \subseteq X \mid \exists a \in X, U = [a]_{\sim} } viz. the set of all equivalence classes.

Lemma: equivalence in the set XX \Leftrightarrow equality of equivalence class in X/X/{\sim}. Symbolically, $\forall a, b \in X, (a \sim b)\Leftrightarrow([a]{\sim} = [b]{\sim})$. (The proof should be straightforward)

Theorem: each equivalence relation \sim on XX corresponds with a unique partition of XX, namely S=X/\mathscr{S} = X/{\sim}. Proof omitted.

H2 Partial Order Relation

An partial order relation is a relation that is reflexive, not symmetric, antisymmetric, and transitive. (They behave like \leq or \subseteq, and their symbol often look like \leq or \subseteq).

H3 Bounds and mums

Let \preceq be a partial order relation on XX and AXA \subseteq X.

An upper bound for AA under \preceq is an element in XX that is “greater” than all elements in AA. That is, uXu \in X s.t. aA,au\forall a \in A, a \preceq u.

An lower bound for AA under \preceq is an element in XX that is “lower” than all elements in AA. That is, lXl \in X s.t. aA,la\forall a \in A, l \preceq a.

The supremum for AA under \preceq is the “least” upper bound. That is, sup(A)=sX\mathrm{sup}_{\preceq}(A) = s \in X such that (aA,as)(uX,(aA,au)(su))(\forall a \in A, a \preceq s) \wedge (\forall u \in X, (\forall a \in A, a \preceq u) \Rightarrow (s \preceq u)) viz. it’s an upper bound and it’s “less” than all other possible upper bounds.

The infimum for AA under \preceq is the “greatest” lower bound. That is, inf(A)=iX\mathrm{inf}_{\preceq}(A) = i \in X such that (aA,ia)(lX,(aA,la)(li))(\forall a \in A, i \preceq a) \wedge (\forall l \in X, (\forall a \in A, l \preceq a) \Rightarrow (l \preceq i)) viz. it’s a lower bound and it’s “greater” than all other possible lower bounds.

Useful: The supremum of X,Y{ X, Y } under \subseteq where X,YX, Y are sets is equal to XYX \cup Y. The infimum is XYX \cap Y.

H3 Existence of bounds and mums

Completeness axiom: for all subset XX of R\mathbb{R}, XX has an upper bound implies it has a supremum. Simile for infimum. Think about why.

Uniqueness theorem: if a infimum exists, it’s unique; if a supremum exists it’s unique. This comes straight out of definition of the mums and antisymmetry of the partial order relation.