COMPSCI 230: Discrete Mathematics for Computer Science February 11, 2019

Lecture 9
Lecturer: Debmalya Panigrahi Scribe: Kevin Sun

1 Overview
In this lecture, we study a special class of relations on a set known as equivalence relations. We
give examples and then prove a connection between equivalence relations and partitions of a set.

2 Equivalence Relations
Definition 1. An equivalence relation is a relation that is reflexive, symmetric, and transitive.
We shall now give some examples of equivalence relations. Recall that if R is a relation, then
the statements “xRy”, “( x, y) ∈ R”, and “x relates to y” (but not “y relates to x”) are all equivalent.
Example 1: Consider the relation R = {( x, y) : x = y} on the set Z+ . To show that R is an
equivalence relation, we must show that R is reflexive, symmetric, and transitive:
• For all x ∈ Z+ , the equality x = x is true, so xRx. This proves R is reflexive.

• If xRy, then x = y, which means y = x, so yRx. This proves R is symmetric.

• If xRy and yRz, then x = y and y = z, which means x = z, so xRz. This proves R is transitive.
Similarly, we can show that if S is any set, then the relation {( A, B) : A = B} on the set 2S is also an
equivalence relation.
Example 2: Consider the sets A = {1, 2, a, b} and B = { a, b} (so B ⊂ A). Recall that if S is a set,
then the power set of S is denoted by 2S , and its elements are all possible subsets of S. Now consider
the relation R on 2 A (i.e., R is a subset of 2 A × 2 A ), defined as follows:

( X, Y ) ∈ R if and only if X ∩ B = Y ∩ B.

In other words, X relates to Y if they contain the same subset of letters. For instance, under this
relation, ∅ relates to exactly four elements of 2 A : ∅, {1}, {2}, and {1, 2}.
We now show that R is an equivalence relation, which requires showing that R is reflexive,
symmetric, and transitive. This proof (and many proofs of equivalence relations) is very similar to
the proof in Example 1.
• For all X ∈ 2S , we have X ∩ B = X ∩ B, so R is reflexive.

• If ( X, Y ) ∈ R then X ∩ B = Y ∩ B, which means Y ∩ B = X ∩ B. This implies (Y, X ) ∈ R, so

R is symmetric.

• If ( X, Y ) ∈ R and (Y, Z ) ∈ R, then X ∩ B = Y ∩ B and Y ∩ B = Z ∩ B. Together, these

equalities imply X ∩ B = Z ∩ B, which means ( X, Z ) ∈ R, so R is transitive.

9-1
Example 3: Now let’s slightly modify the definition of the relation from Example 2. Now, X relates
to Y if and only if X and Y contain the same number of letters, that is, | X ∩ B| = |Y ∩ B|. It is
straightforward to verify that this new relation is also an equivalence relation.
But note that this modified equivalence relation is indeed different from the one in Example 2.
Consider X = {1, a} and Y = {2, b}: here, X and Y share the same number of letters (one), but not
the same set of letters ({ a} and {b}, respectively). Therefore, X does not relate to Y according to the
relation defined in Example 2, but X relates to Y in the current example. In general, a set can have
many different equivalence relations.

3 Partitions
One of the reasons that equivalence relations are interesting is because of their connection to
partitions. To motivate this connection, let’s consider the set A = {1, 2, 3} and define a relation R
on 2 A as follows: ( X, Y ) ∈ R if and only if | X | = |Y |. We shall explicitly draw this relation below:
for every element of 2 A , and if X relates to Y, we’ll draw an arrow from X to Y.

{1} {1, 2}

∅ {2} {1, 3} A

{3} {2, 3}

Figure 1: The equivalence relation R = {( X, Y ) : | X | = |Y |} on A = {1, 2, 3}, where an arrow from

S1 to S2 indicates (S1 , S2 ) ∈ R.

In Fig. 1, each circle represents an element of 2 A (i.e., a subset of A). Notice that based on the
arrows, the elements of 2 A are grouped into four distinct clusters. Within each cluster, all of the
elements relate to each other (including themselves, because R is reflexive), and if two elements are
in different clusters, then neither relates to the other.
The structure of these clusters is no coincidence: if S is a set and R is an equivalence relation on
S, then R induces a clustering of this form, and this kind of clustering is known as a partition.

Definition 2. Let P be a set containing subsets of S, so P is a subset of 2S . Then P is a partition of S if P

satisfies the following properties:

1. The union of the elements of P is equal to S.

2. If S1 and S2 are elements of P and S1 ∩ S2 6= ∅, then S1 = S2 .

If P is a partition, then each element of P is called a block of P, so each block of P is a subset of S.

9-2
 Informally, the first property ensures that every element of A is covered by a block, and the
second property ensures that no two blocks overlap. An alternative definition is the following: if P
is a collection of subsets of A, then P is a partition of A if every element of A is in exactly one subset
in P. At the end of the lecture, we will formally state and prove the claim that every equivalence
relation induces a partition, but for now, we continue exploring this idea via examples.
In Fig. 1, R induces the following partition on 2 A :
   
{∅}, {1}, {2}, {3} , {1, 2}, {1, 3}, {2, 3} , { A}.

This example may be somewhat confusing, because the elements of the set we partition (2 A )
are actually subsets of some other set (A). Nonetheless, 2 A has 8 elements, and this partition
assigns each element to blocks of size 1, 3, 3, and 1. We now give more examples that are more
straightforward.
Example 1: Let R = {( a, b) : a = b} be a relation on Z+ ; it’s easy to verify that R is an equivalence
relation. Now let P be the partition induced by R. Then for all n ∈ Z+ , the only element to which n
relates is n itself. Thus, every block of the partition contains exactly one positive integer, and there
are infinitely many blocks.
Example 2: Let R2 = {( a, b) : a ≡ b (mod 2)} be a relation on Z. (We say that a ≡ b (mod k ) if
a − b is divisible by k.) Again, it’s easy to verify that R is an equivalence relation. Notice that if
( x, y) ∈ R, then x − y is even, which means x and y are either both even or both odd. Thus, the
partition induced by R2 has two blocks: the set of even integers, and the set of odd integers.
Similarly, if R16 = {( a, b) : a ≡ b (mod 16)} is a relation on Z, then R16 is also an equivalence
relation, so R16 induces a partition. The elements of each block all have the same remainder after
dividing by 16, and there are 16 possible remainders (0, 1, . . . , 15), so this partition has 16 blocks.

To begin formalizing the connection between equivalence relations and partitions, we now state
a useful definition regarding equivalence relations.

Definition 3. Suppose R is an equivalence relation on a set S, and x is any element of S. The equivalence
class of x, denoted by [ x ], is the set {y ∈ S : xRy}.

Example 1: In Fig. 1, recall that 2 A is being partitioned, and 2 A has 8 elements. The corresponding
equivalence classes are the following:

[∅] = {∅}
 
[{1}] = [{2}] = [{3}] = {1}, {2}, {3}
 
[{1, 2}] = [{1, 3}] = [{2, 3}] = {1, 2}, {1, 3}, {2, 3}

[ A ] = { A }.

Notice that the equivalence classes form a partition—this is the connection between equivalence
relations and partitions that we will prove below.

9-3
Example 2: Recall the relation R16 = {( a, b) : a ≡ b (mod 16)} on Z. Some of the equivalences
classes are given below:

[0] = {. . . , −32, −16, 0, 16, 32, . . .},

[1] = {. . . , −31, −15, 1, 17, 33, . . .},
..
.
[15] = {. . . , −33, −17, −1, 15, 31 . . .}.

We are now ready to state and prove a theorem that formally describes the fundamental
relationship between equivalence relations and partitions. Note that throughout this lecture, we
have already seen that an equivalence relation induces a partition, but now we shall formally prove
this phenomenon.

Theorem 1. If R is an equivalence relation on a set S, then the equivalence classes of R partition S.

Proof. Let P denote the subsets described in the theorem, so P = {[ x ] : x ∈ S}. Recall that P is a
partition if the following properties hold:

1. The union of the elements of P is equal to S.

2. If S1 and S2 are elements of P and S1 ∩ S2 6= ∅, then S1 = S2 .

Let’s begin with the first property: let s be any element of S. Since R is reflexive, we have sRs,
which means s ∈ [s]. Since [s] is an element of P, s is covered by an element of P, so P satisfies the
first property.
To prove the second property, we start by letting S1 and S2 be elements of P such that S1 ∩ S2 6=
∅. By the definition of P, there exist a, b such that S1 = [ a] and S2 = [b]. Since S1 ∩ S2 6= ∅, there
exists z such that z ∈ [ a] ∩ [b]. Thus, aRz and bRz. Since R is symmetric, this means zRa and zRb.
Furthermore, since R is transitive, this means aRb and bRa.
We will now prove that [ a] = [b], and as usual, we begin by showing [ a] ⊆ [b]: let x be any
element of [ a]. Then aRx, and combined with bRa and transitivity of R, we have bRx. Thus, x ∈ [b],
so [ a] ⊆ [b].
Similarly, let y be any element of [b]. Then bRy, and combined with aRb and transitivity of R,
this means aRy. Thus, y ∈ [ a], so [b] ⊆ [ a]. Since [ a] ⊆ [b] and [b] ⊆ [ a], we must have [ a] = [b], as
desired.

4 Summary
In this lecture, we learned about a special class of relations known as equivalence relations. We saw
multiple examples, and ended by proving that the equivalence classes of an equivalence relation
form a partition of the set that the relation is defined on.

9-4