You are on page 1of 6

Relations I

Margaret M. Fleck
7 November 2008
This lecture covers relations and basic properties of relations, which is
most of section 8.1 of Rosen.

Announcements

Model solutions for the second midterm will be posted as soon as we finish
doing makeup exams, which will probably be mid next week.

Relations

Suppose that A and B are sets. A relation from A to B is a subset of A B.


That is, its a set of ordered pairs (a, b) where a A and b B.
For example, suppose that A is a set of US states and B is the set of US
political parties. We could create a relation S that associates each state with
the parties of its senators. So S would contain the pair (Illinois, Democrat)
since both of our senators are democrats. And it contains (Utah, Republican)
because they have two Republican senators.
Notice that S isnt a function. Iowa has one senator of each party. So
S must contain both the pairs (Iowa, Democrat) and (Iowa, Republican). A
relation from A to B can associate an element of A with multiple elements of
B, or no elements of B. For example, we could extend our set A to include
the District of Columbia, even though it doesnt have a senator.
1

Relations are a very general construct, which unifies a wide range of


examples that probably dont look very similar to you. In fact, much of the
time, we will be looking at one or another special type of relation.

Functions and relations

Functions are one special case of relations. For example, suppose we want to
associate each state with its capital. We can describe this as a relation F from
the set of US states (A) to the set of US cities (C) So F is a set containing
pairs like (Massachusetts, Boston) and (Iowa, Des Moines). In this case, the
relation happens to associate each state with exactly one city. So we can also
describe C as a function from A to C.
You can formally define a function from A to B as a relation from A to
B such that each element of A is associated with exactly one element of B.
Or you can think of a relation as a generalized type of function, which
allows missing or multiple outputs for certain inputs. Many mathematical
applications use partial functions, which are functions that might not return a value for every input. Similarly, many programming languages let you
define procedures that do not return a value. Some programming languages
even let you define procedures that return multiple values.
If a relation is a function, its usually more convenient to define it as a
function and use function notation. So, if we observe them in the wild, mathematicians normally call something a relation only when it isnt a function,
or they are worried that it might not turn out to be a function.

Relations and databases

The facts stored in a computer database are also relations. For this application, we need to generalize our notion of relation to more than two sets. For
example, a simple registration database might contain 4-tuples like
(Jean Luc Picard, Math CS 173, Fall 2325, A-)

(James T. Kirk, CS 173, Spring 2248, B)


(Kathryn Janeway, Math 347, Fall 2345, A+)
So this database is a subset of A B C D, where A is the set of
students, B is the set of courses, C is the set of terms, and D is the set of
letter grades.
Database records dont look at all like functions, because theres often
no obvious way to identify one field as the input. The whole point of a
database is to let us easily switch perspective and organize the data according
to a different field, e.g. sort students by name or by grade.
We wont make much use of n-ary relations in this class. If you are curious
about database applications, look in section 8.2 of Rosen.

Relations on a single set

The final big group of relations are relations that associate two elements
of the same set. A relation R on a set A is a relation from A to A, or
equivalently, a subset of A A. That is, R is a set of pairs of elements from
A.
For example, suppose we let A = {2, 3, 4, 5, 6, 7} We can define a relation
R on A by (x, y) R if and only if |x y| 2. This relation contains the
pairs (3, 4) and (4, 6) and (6, 4), but not the pair (3, 6).
Another relation on A is the familiar = relation. It contains only pairs
whose two elements are identical, such as (5, 5) and (3, 3). Another similar
relation is congruence mod 3 (3 ). The 3 relation contains a wider range
of pairs, e.g. (4, 7) and (6, 3). Relations like these, which resemble equality,
are known as equivalence relations. Well get back to defining them precisely
in a couple lectures.
Order relations such as < and are also relations on the set A. For
example, the relation contains pairs like (5, 4) and (7, 2) and (5, 5).
No one said that a relation needs to involve finite sets or even onedimensional sets. We can define similar relations on infinite sets. For example, normal numerical is a relation on the set of integers or the set of
3

real numbers. Another example: we can define a relation T on the real plane
R2 in which (x, y) is related to (p, q) if and only if x2 + y 2 = p2 + q 2 . In other
words, two points are related if they are the same distance from the origin.

Random relations

No one said that a relation had to make sense or have any practical use.
For example, lets let A = {a, b, c, d}. We can select any random subset of
A A to be a relation. For example, R = {(a, a), (b, a), (c, d), (d, d)} is a
perfectly good relation on A. Wierdo relations like this are often easiest to
represent by making a table showing which pairs are in the set, or by drawing
a dot-and-arrow diagram of the relation. [See pictures pp. 520521 of Rosen.
Notice that we can do the dot-and-arrow diagram with either one copy of
the points or two copies.]
If a set A has n elements, how many possible relations are there on A?
A A contains n2 elements. A relation is just a subset of A A, and so there
2
are 2n relations on A. So a 3-element set has 29 = 512 possible relations.
Yup: most of these relations are of no practical use whatsoever. Most of
the time, you will be manipulating relations generated by some underlying
pattern and, thus, make more sense. Just be aware that these random guys
are also legitimate relations.

Properties of relations: reflexive

Familiar relations such as = and < have certain special properties which
make them especially useful, both in proofs and in practical applications.
Moreover, some relations seem to be structurally similar: similar properties
for apparently similar reasons. For example, and and divides. We can
make these intuitions concrete by classifying relations according to certain
key properties.
The commonly-used properties are: reflexive, irreflexive, symmetric, antisymmetric, and transitive. Relations that act like equality are all reflexive,
symmetric, and transitive. Relations that act like < are all irreflexive, anti4

symmetric, and transitive. And so forth.


A relation R on a set A is reflexive if every element is related to itself.
That is, for every x A, (x, x) R.
For example, normal equality (on the integers, or on the reals, or on sets
of integers, or whatever) is reflexive.
Another example: let B contain sets of UIUC students. So B might contain {Hanna, Melissa} and {John, Marco, Oscar} and so forth. Lets define a
relation R by saying that x and y are in R iff |x| = |y| (i.e x and y contain the
same number of people. R isnt equality, because two non-equal elements can
be related. For example, R contains ({Hanna, Melissa}, {Yunsook, Jaebum}).
But R is reflexive, because every set of students is related to itself.
Yet another example: the relation on B is reflexive, because every set
is a subset of itself.

Irreflexive

A relation R on a set A is irreflexive if every element is not related to itself.


That is, for every x A, (x, x) 6 R.
For example, the < relation on the integers is irreflexive, because no
integer is less than itself.
Notice that irreflexive is not the negation of reflexive. Lets look at the
definitions more closely:
reflexive: x A, (x, y) R
not reflexive: x A, (x, y) 6 R.
irreflexive: x A, (x, y) 6 R
So we can have a relation which is neither reflexive nor irreflexive, if some
elements are related to themselves and some arent. For example, lets define
a relation S on the integers by (x, y) S if and only if x + y = 0. (2, 2) isnt
in S, but (0, 0) is.
5

Symmetric and antisymmetric

Another important property of a relation is whether the order matters within


each pair. That is, if (x, y) is in R, is (y, x) always in R? A relation satisfying
this property is called symmetric.
Generalizations of equality are normally symmetric. For example, the
relation R on the integers defined by (x, y) R iff |x| = |y|. Or the relation
S on the real plane defined by ((x, y), (p, q)) S iff (x p)2 + (y q)2 25
(i.e. the two points are no more than 5 units apart).
A relation R is antisymmetric if (y, x) and (x, y) are never both in R,
except when x = y. This is typically true of order relations such as < and ,
or relations based on them. For example, the divides relation on the integers
is antisymmetric because n|m and m|n are only both true when m = n.
Formally, we can define them as:
symmetric: x, y A, (x, y) R (y, x) R
antisymmetric: x, y A with x 6= y, (x, y) R (y, x) 6 R
Even if we restrict our attention to pairs whose two coordinates arent
the same, symmetric and antisymmetric still arent opposites. Rather, they
represent two pure patterns of what might happen with the reversed pairs.
Most random relations and some real practical relations are mixed cases.
Specifically a relation is neither symmetric nor antisymmetric if (x, y)
and (y, x) are both in the relation for some values of x and y, but only one
is in the set for other values. For example, the loves relation is neither
symmetric nor antisymmetric for most large sets of people. There are some
pairs of people who both love each other. But there are also some people
who love another person, but their love isnt reciprocated.

You might also like