In this section we shall introduce a tool which is very useful in mathematics because it provides a way of reducing a complicated situation to a simpler one.
An equivalence relation is a relation which is reflexive, symmetric and transitive.
Thus we have already seen two examples of equivalence relations:
‘‘congruence modulo ’’ on the set of integers;
‘‘being in the same college’’ on the set of MATH111 students.
Let us take the first of these. Given any , we may consider the subset of consisting of numbers which are congruent to modulo . For example, if we obtain the subset , while if instead we obtain the subset . This prompts the following definition.
Given an equivalence relation on a non-empty set , for any we call the set the equivalence class of with respect to ; we denote it by .
Thus for the relation of congruence modulo on the set , we have
We note several things about each of these equivalence classes:
is an element of the equivalence class of (e.g., );
if is any element of the equivalence class of , then the equivalence classes of and are the same (e.g., , and );
if is any element of not in the equivalence class of , then the equivalence classes of and are disjoint (e.g., , and
These also hold for the ‘‘college’’ relation on the set of MATH111 students, as is easy to see; in fact we shall prove that they are true for equivalence classes in general.
Let be an equivalence relation on a non-empty set . Then for each pair , we have the following:
if then
if then .
Proof. (i) Since is reflexive, and so
(ii) Suppose that , so that ; as is symmetric, we also have
If we take , then ; as is transitive, from we deduce and so — thus .
On the other hand if we take , then as is transitive, from we deduce and so — thus Therefore as required.
(iii) We prove this by contraposition. Suppose that , and take We then have and ; from the former we have because is symmetric, and then using the latter as well we have since is transitive. Hence , and the result is proved.
Let be an equivalence relation on a non-empty set . Then each element of lies in a unique equivalence class.
Proof. Let . Then , so lies in at least one equivalence class. Now suppose that is any other equivalence class containing . Then by Proposition 6.2.3(ii), and thus is the unique equivalence class containing .
Motivated by the above results, we make the following definition.
A partition of a set is a collection of non-empty subsets of such that each element of lies in exactly one of the subsets.
We may think of a partition of as a way of dividing up into pieces. We have shown that the equivalence classes corresponding to any equivalence relation on form a partition of ; the pieces into which is divided are the equivalence classes. In fact given any partition of a non-empty set we may define a relation on by stipulating that if and lie in the same subset of (as in the ‘‘college’’ example, above). An easy check shows that is an equivalence relation, whose equivalence classes are exactly the subsets making up the partition. Thus the concepts of ‘‘equivalence relation’’ and ‘‘partition’’ are very closely linked; each gives rise to the other in a natural way.
Another consequence of our results is that we can make the following definition.
Given an equivalence relation on a non-empty set , any element of lying in a given equivalence class is called a representative of that class.
From Proposition 6.2.3(i), we see that is a representative of ; from (ii) and (iii) we see that is a representative of if and only if .
For the equivalence relation of congruence modulo on the set , we see that it is easy to obtain the full set of equivalence classes (or congruence classes, as they are called in this case). We already know that every integer is congruent to exactly one of , , , and , and the corresponding congruence classes are
Note that , while any two classes are disjoint.
For the ‘‘college’’ equivalence relation above, the equivalence classes are the sets
where is Bowland, Cartmel, County, Furness, Fylde, Grizedale, Lonsdale or Pendle; their union is the set of all MATH111 students, and any two are disjoint.
Thus if we have an equivalence relation on a set , we may consider the set of equivalence classes; often this is ‘‘simpler’’ than itself, because it has fewer elements (each equivalence class may contain many elements of ). We shall see later that sometimes the equivalence classes provide enough detail for the task at hand, allowing us to ignore much that is irrelevant and concentrate on what is essential.
We conclude this section by mentioning two equivalence relations which exist on any non-empty set . The first is the relation defined by for all , in which any two elements of are related; here there is a single equivalence class consisting of the whole of . The second is the relation of equality, defined by if ; this time each equivalence class consists of a single element of . These two relations are not very useful, as the first oversimplifies the structure of to a single class while the second does not simplify at all; interesting equivalence relations lie between these two extremes.