CHAPTER 1

Relations

1. Relations and Their Properties

1.1. Definition of a Relation.

Definition 1.1.1. A binary relation from a set A to a set B is a subset
R ⊆ A × B.
If (a, b) ∈ R we say a is Related to b by R.

A is the domain of R, and

B is the codomain of R.

If A = B, R is called a binary relation on the set A.

Notation.

• If (a, b) ∈ R, then we write aRb.

• If (a, b) 6∈ R, then we write a R

6 b.
Discussion

We recall the basic definitions and terminology associated with the concept of
a relation from a set A to a set B. You should review the notes from MAD 2104
whenever you wish to see more examples or more discussion on any of the topics we
review here.

A relation is a generalization of the concept of a function, so that much of the

terminology will be the same. Specifically, if f : A → B is a function from a set A to
a B, then the graph of f , graph(f ) = {(x, f (x))|x ∈ A} is a relation from A to B.
Recall, however, that a relation may differ from a function in two essential ways. If
R is an arbitrary relation from A to B, then

• it is possible to have both (a, b) ∈ R and (a, b0 ) ∈ R, where b0 6= b; that is,

an element a in A may be related to any number of elements of B; or
8
 1. RELATIONS AND THEIR PROPERTIES 9

• it is possible to have some element a in A that is not related to any element

in B at all.
Example 1.1.1. Suppose R ⊂ Z × Z+ is the relation defined by (a, b) ∈ R if
and only if a|b. (Recall that Z+ denotes the set of positive integers, and a|b, read “a
divides b”, means that b = na for some integer n.) Then R fails to be a function on
both counts listed above. Certainly, −2|2 and −2|4, so that −2 is related to more than
one positive integer. (In fact, it is related to infinitely many integers.) On the other
hand 06 | b for any positive integer b.
Remark 1.1.1. It is often desirable in computer science to relax the requirement
that a function be defined at every element in the domain. This has the beneficial
effect of reducing the number of sets that must be named and stored. In order not
to cause too much confusion with the usual mathematical definition of a function,
a relation such as this is called a partial function. That is, a partial function
f : A → B is a relation such that, for all a ∈ A, f (a) is a uniquely determined
1
element of B whenever it is defined. For example, the formula f (x) = 1−x 2 defines

a partial function f : R → R with domain R and codomain R: f is not defined at

x = −1 and x = 1, but, otherwise, f (x) is uniquely determined by the formula.
Exercise 1.1.1. Let n be a positive integer. How many binary relations are there
on a set A if |A| = n? [Hint: How many elements are there in |A × A|?]

1.2. Directed Graphs.

Definitions 1.2.1.
• A directed graph or a digraph D from A to B is a collection of vertices
V ⊆ A ∪ B and a collection of edges R ⊆ A × B.
• If there is an ordered pair e = (x, y) in R then there is an arc or edge from
x to y in D.
• The elements x and y are called the initial and terminal vertices of the
edge e = (x, y), respectively.

Discussion

A digraph can be a useful device for representing a relation, especially if the

relation isn’t “too large” or complicated. If R is a relation on a set A, we simplify
the digraph G representing R by having only one vertex for each a ∈ A. This results,
however, in the possibility of having loops, that is, edges from a vertex to itself, and
having more than one edge joining distinct vertices (but with opposite orientations).
A digraph for the relation R in Example 1.1.1 would be difficult to illustrate, and
impossible to draw completely, since it would require infinitely many vertices and
edges. We could draw a digraph for some finite subset of R.
 1. RELATIONS AND THEIR PROPERTIES 10

Example 1.2.1. Suppose R is the relation on {0, 1, 2, 3, 4, 5, 6} defined by mRn

if and only if m ≡ n(mod 3). The digraph that represents R is

0 1 2

3 6 4 5

1.3. Representing Relations with Matrices.

Definition 1.3.1. Let R be a relation from A = {a1 , a2 , . . . , am } to B = {b1 , b2 , . . . , bn }.
An m × n connection matrix MR = {mij } for R is defined by

1 if (ai , bj ) ∈ R,
mij =
0 otherwise.

1.4. Example 1.4.1.

Example 1.4.1. Let A = {a, b, c}, B = {e, f, g, h}, and R = {(a, e), (c, g)}.

then the connection matrix

 
1 0 0 0
MR = 0 0 0 0 .
0 0 1 0

Discussion

Recall the connection matrix for a finite relation is a method for representing a
relation using a matrix.
Remark 1.4.1. The ordering of the elements in A and B is important. If the
elements are rearranged the matrix would be different! If there is a natural order to
the elements of the sets (like numerical or alphabetical) you would expect to use this
order when creating connection matrices.

To find this matrix we may use a table as follows. First we set up a table labeling
the rows and columns with the vertices.

e f g h
a
b
c
 1. RELATIONS AND THEIR PROPERTIES 11

Since (a, e) ∈ R we put a 1 in the row a and column e and since (c, g) ∈ R we put
a 1 in row c and column g.

e f g h
a 1
b
c 1

Fill in the rest of the entries with 0’s. The matrix may then be read straight from
the table.

1.5. Inverse Relation.

Definition 1.5.1. Let R be a relation from A to B. Then R−1 = {(b, a)|(a, b) ∈
R} is a relation from B to A.

R−1 is called the inverse of the relation R.

Discussion

The inverse of a relation R is the relation obtained by simply reversing the ordered
pairs of R. The inverse of a relation is also called the converse of a relation.
Example 1.5.1. Let A = {a, b, c} and B = {1, 2, 3, 4} and let R = {(a, 1), (a, 2), (c, 4)}.
Then R−1 = {(1, a), (2, a), (4, a)}.
Exercise 1.5.1. Suppose A and B are sets and f : A → B is a function. The
graph of f , graph(f ) = {(x, f (x))|x ∈ A} is a relation from A to B.

(a) What is the inverse of this relation?

(b) Does f have to be invertible (as a function) for the inverse of this relation to
exist?
(c) If f is invertible, find the inverse of the relation graph(f ) in terms of the
inverse function f −1 .

1.6. Special Properties of Binary Relations. Definitions. Let A be a set,

and let R be a binary relation on A.

(1) R is reflexive if
∀x[(x ∈ A) → ((x, x) ∈ R)].
(2) R is irreflexive if
∀x[(x ∈ A) → ((x, x) 6∈ R)].
(3) R is symmetric if
∀x∀y[((x, y) ∈ R) → ((y, x) ∈ R)].
 1. RELATIONS AND THEIR PROPERTIES 12

(4) R is antisymmetric if
∀x∀y[((x, y) ∈ R ∧ (y, x) ∈ R) → (x = y)].
(5) R is asymmetric if
∀x∀y[((x, y) ∈ R) → ((y, x) 6∈ R)].
(6) R is transitive if
∀x∀y∀z[((x, y) ∈ R ∧ (y, z) ∈ R) → ((x, z) ∈ R)].

Discussion

The definition above recalls six special properties that a relation may (or may
not) satisfy. Notice that the definitions of reflexive and irreflexive relations are not
complementary. That is, a relation on a set may be both reflexive and irreflexive or
it may be neither. The same is true for the symmetric and antisymmetric properties,
as well as the symmetric and asymmetric properties.

The terms reflexive, symmetric, and transitive is generally consistent from author
to author. The rest are not as consistent in the literature
Exercise 1.6.1. Before reading further, find a relation on the set {a, b, c} that is
neither

(a) reflexive nor irreflexive.

(b) symmetric nor antisymmetric.
(c) symmetric nor asymmetric.

1.7. Examples of Relations and their Properties.

Example 1.7.1. Suppose A is the set of all residents of Florida and R is the
relation given by aRb if a and b have the same last four digits of their Social Security
Number. This relation is. . .

• reflexive
• not irreflexive
• symmetric
• not antisymmetric
• not asymmetric
• transitive
Example 1.7.2. Suppose T is the relation on the set of integers given by xT y if
2x − y = 1. This relation is. . .

• not reflexive
• not irreflexive
 1. RELATIONS AND THEIR PROPERTIES 13

• not symmetric
• antisymmetric
• not asymmetric
• not transitive
Example 1.7.3. Suppose A = {a, b, c, d} and R is the relation {(a, a)}. This
relation is. . .

• not reflexive
• not irreflexive
• symmetric
• antisymmetric
• not asymmetric
• transitive

Discussion

When proving a relation, R, on a set A has a particular property, the property

must be shown to hold for all appropriate combinations of members of the set. When
proving a relation R does not have a property, however, it is enough to give a coun-
terexample.
Example 1.7.4. Suppose T is the relation in Example 1.7.2 in Section 1.7. This
relation is

• not reflexive
Proof. 2 is an integer and 2 · 2 − 2 = 2 6= 1. This shows that ∀x[x ∈
Z → (x, x) ∈ T ] is not true. 
• not irreflexive
Proof. 1 is an integer and 2 · 1 − 1 = 1. This shows that ∀x[x ∈ Z →
(x, x) 6∈ T ] is not true. 
• not symmetric
Proof. Both 2 and 3 are integers, 2 · 2 − 3 = 1, and 2 · 3 − 2 = 4 6= 1.
6 2; that is, ∀x∀y[(x, y) ∈ Z → (y, x) ∈ T ] is not
This shows 2R3, but 3 R
true. 
• antisymmetric
Proof. Let m, n ∈ Z be such that (m, n) ∈ T and (n, m) ∈ T . By the
definition of T , this implies both equations 2m − n = 1 and 2n − m = 1
must hold. We may use the first equation to solve for n, n = 2m − 1, and
substitute this in for n in the second equation to get 2(2m − 1) − m = 1. We
 1. RELATIONS AND THEIR PROPERTIES 14

may use this equation to solve for m and we find m = 1. Now solve for n
and we get n = 1.
This shows that the only integers, m and n, such that both equations
2m − n = 1 and 2n − m = 1 hold are m = n = 1. This shows that
∀m∀n[((m, n) ∈ T ∧ (n, m) ∈ T ) → m = n]. 
• not asymmetric
Proof. 1 is an integer such that (1, 1) ∈ T . Thus ∀x∀y[((x, y) ∈ T →
(b, a) 6∈ T ] is not true (counterexample is a = b = 1). 
• not transitive
Proof. 2, 3, and 5 are integers such that (2, 3) ∈ T , (3, 5) ∈ T , but
(2, 5) 6∈ T . This shows ∀x∀y∀z[(x, y) ∈ T ∧ (y, z) ∈ T → (x, z) ∈ T ] is not
true. 
Exercise 1.7.1. Verify the assertions made about the relation in Example 1.7.1
in Section 1.7.
Exercise 1.7.2. Verify the assertions made about the relation in Example 1.7.3
in Section 1.7.
Exercise 1.7.3. Suppose R ⊂ Z+ × Z+ is the relation defined by (m, n) ∈ R if
and only if m|n. Prove that R is

(a) reflexive.
(b) not irreflexive.
(c) not symmetric.
(d) antisymmetric.
(e) not asymmetric.
(f ) transitive.

1.8. Theorem 1.8.1: Connection Matrices v.s. Properties.

Theorem 1.8.1. Let R be a binary relation on a set A and let MR be its connection
matrix. Then

• R is reflexive iff mii = 1 for all i.

• R is irreflexive iff mii = 0 for all i.
• R is symmetric iff M is a symmetric matrix.
• R is antisymmetric iff for each i 6= j, mij = 0 or mji = 0.
• R is asymmetric iff for every i and j, if mij = 1, then mji = 0.

Discussion

Connection matrices may be used to test if a finite relation has certain properties
and may be used to determine the composition of two finite relations.
 1. RELATIONS AND THEIR PROPERTIES 15

Example 1.8.1. Determine which of the properties: reflexive, irreflexive, symmet-

ric, antisymmetric, asymmetric, the relation on {a, b, c, d} represented by the following
matrix has.

 
0 0 1 0
1 0 0 0
 .
0 0 0 1
0 1 0 0

Solution. The relation is irreflexive, asymmstric and antisymmetric only.

Exercise 1.8.1. Determine which of the properties: reflexive, irreflexive, symmet-
ric, antisymmetric, asymmetric, the relation on {a, b, c, d} represented by the following
matrix has.

 
1 1 1 0
1 0 0 0
 .
1 0 0 1
0 0 1 0

1.9. Combining Relations. Suppose property P is one of the properties listed

in Section 1.6, and suppose R and S are relations on a set A, each having property
P . Then the following questions naturally arise.

(1) Does R (necessarily) have property P ?

(2) Does R ∪ S have property P ?
(3) Does R ∩ S have property P ?
(4) Does R − S have property P ?

Discussion

Notice that when we combine two relations using one of the binary set operations
we are combining sets of ordered pairs.

1.10. Example 1.10.1.

Example 1.10.1. Let R and S be transitive relations on a set A. Does it follow
that R ∪ S is transitive?

Solution. No. Here is a counterexample:

 1. RELATIONS AND THEIR PROPERTIES 16

A = {1, 2}, R = {(1, 2)}, S = {(2, 1)}

Therefore, R ∪ S = {(1, 2), (2, 1)}

Notice that R and S are both transitive (vacuously, since there are no two elements
satisfying the hypothesis of the conditions of the property). However R ∪ S is not
transitive. If it were it would have to have (1, 1) and (2, 2) in R ∪ S.

Discussion

The solution to Example 1.10.1 gives a counterexample to show that the union
of two transitive relations is not necessarily transitive. Note that you could find an
example of two transitive relations whose union is transitive. The question, however,
asks if the given property holds for two relations must it hold for the binary operation
of the two relations. This is a general question and to give the answer “yes” we must
know it is true for every possible pair of relations satisfying the property.

Here is another example:

Example 1.10.2. Suppose R and S are transitive relations on the set A. Is R ∩ S
transitive?

Solution. Yes.

Proof. Assume R and S are both transitive and suppose (a, b), (b, c) ∈ R ∩ S.
Then (a, b), (b, c) ∈ R and (a, b), (b, c) ∈ S. It is given that both R and S are
transitive, so (a, c) ∈ R and (a, c) ∈ S. Therefore (a, c) ∈ R ∩ S. This shows that for
arbitrary (a, b), (b, c) ∈ R ∩ S we have (a, c) ∈ R ∩ S. Thus R ∩ S is transitive. 

As it turns out, the intersection of any two relations satisfying one of the properties
in Section 1.6 also has that property. As the following exercise shows, sometimes even
more can be proved.
Exercise 1.10.1. Suppose R and S are relations on a set A.

(a) Prove that if R and S are reflexive, then so is R ∩ S.

(b) Prove that if R is irreflexive, then so is R ∩ S.
Exercise 1.10.2. Suppose R and S are relations on a set A.
 1. RELATIONS AND THEIR PROPERTIES 17

(a) Prove that if R and S are symmetric, then so is R ∩ S.

(b) Prove that if R is antisymmetric, then so is R ∩ S.
(c) Prove that if R is asymmetric, then so is R ∩ S.
Exercise 1.10.3. Suppose R and S are relations on a set A. Prove or disprove:

(a) If R and S are reflexive, then so is R ∪ S.

(b) If R and S are irreflexive, then so is R ∪ S.
Exercise 1.10.4. Suppose R and S are relations on a set A. Prove or disprove:

(a) If R and S are symmetric, then so is R ∪ S.

(b) If R and S are antisymmetric, then so is R ∪ S.
(c) If R and S are asymmetric, then so is R ∪ S.

1.11. Definition of Composition.

Definition 1.11.1.
(1) Let
• R be a relation from A to B, and
• S be a relation from B to C.
Then the composition of R with S, denoted S ◦ R, is the relation from A
to C defined by the following property:
(x, z) ∈ S ◦ R if and only if there is a y ∈ B such that (x, y) ∈ R and
(y, z) ∈ S.
(2) Let R be a binary relation on A. Then Rn is defined recursively as follows:
Basis: R1 = R
Recurrence: Rn+1 = Rn ◦ R, if n ≥ 1.

Discussion

The composition of two relations can be thought of as a generalization of the

composition of two functions, as the following exercise shows.
Exercise 1.11.1. Prove: If f : A → B and g : B → C are functions, then
graph(g ◦ f ) = graph(g) ◦ graph(f ).
Exercise 1.11.2. Prove the composition of relations is an associative operation.
Exercise 1.11.3. Suppose R is a relation on A. Using the previous exercise and
mathematical induction, prove that Rn ◦ R = R ◦ Rn .
Exercise 1.11.4. Prove an ordered pair (x, y) ∈ Rn if and only if, in the digraph
D of R, there is a directed path of length n from x to y.

Notice that if there is no element of B such that (a, b) ∈ R1 and (b, c) ∈ R2 for
some a ∈ A and c ∈ C, then the composition is empty.
 1. RELATIONS AND THEIR PROPERTIES 18

1.12. Example 1.12.1.

Example 1.12.1. Let A = {a, b, c}, B = {1, 2, 3, 4}, and
C = {I, II, III, IV }.

• If R = {(a, 4), (b, 1)} and

• S = {(1, II), (1, IV ), (2, I)}, then
• S ◦ R = {(b, II), (b, IV )}.

Discussion

It can help to consider the following type of diagram when discussing composition
of relations, such as the ones in Example 1.12.1 shown here.
a 1 I

2 II
b
3 III

c 4 IV
Example 1.12.2. If R and S are transitive binary relations on A, is R ◦ S tran-
sitive?

Solution. No. Here is a counterexample: Let

R = {(1, 2), (3, 4)}, and S = {(2, 3), (4, 1)}.
Both R and S are transitive (vacuously), but
R ◦ S = {(2, 4), (4, 2)}
is not transitive. (Why?)
Example 1.12.3. Suppose R is the relation on Z+ defined by aRb if and only if
a|b. Then R2 = R.
Exercise 1.12.1. Suppose R is the relation on the set of real numbers given by
xRy if and only if xy = 2.

(a) Describe the relation R2 .

(b) Describe the relation Rn , n ≥ 1.
Exercise 1.12.2. Suppose R and S are relations on a set A that are reflexive.
Prove or disprove the relation obtained by combining R and S in one of the following
ways is reflexive. (Recall: R ⊕ S = (R ∪ S) − (R ∩ S).)
 1. RELATIONS AND THEIR PROPERTIES 19

(a) R⊕S
(b) R−S
(c) R◦S
(d) R−1
(e) Rn , n ≥ 2
Exercise 1.12.3. Suppose R and S are relations on a set A that are symmetric.
Prove or disprove the relation obtained by combining R and S in one of the following
ways is symmetric.

(a) R⊕S
(b) R−S
(c) R◦S
(d) R−1
(e) Rn , n ≥ 2
Exercise 1.12.4. Suppose R and S are relations on a set A that are transitive.
Prove or disprove the relation obtained by combining R and S in one of the following
ways is transitive.

(a) R⊕S
(b) R−S
(c) R◦S
(d) R−1
(e) Rn , n ≥ 2

1.13. Theorem 1.13.1: Characterization of Transitive Relations.

Theorem 1.13.1. Let R be a binary relation on a set A. R is transitive if and
only if Rn ⊆ R, for n ≥ 1.

Proof. To prove (R transitive) → (Rn ⊆ R) we assume R is transitive and prove

n
R ⊆ R for n ≥ 1 by induction.

Basis Step, n = 1. R1 = R, so the statement is vacuously true when n = 1, since

1
R = R ⊆ R whether or not R is transitive.

Induction Step. Prove Rn ⊆ R → Rn+1 ⊆ R.

Assume Rn ⊆ R for some n ≥ 1. Suppose (x, y) ∈ Rn+1 . By definition, Rn+1 =

R ◦ R, so there must be some a ∈ A such that (x, a) ∈ R and (a, y) ∈ Rn . By the
n

induction hypothesis, Rn ⊆ R, so (a, y) ∈ R. Since R is transitive, (x, a), (a, y) ∈ R

implies (x, y) ∈ R. Since (x, y) was an arbitrary element of Rn+1 , we have shown
Rn+1 ⊆ R.
 1. RELATIONS AND THEIR PROPERTIES 20

We now prove the reverse implication: Rn ⊆ R, for n ≥ 1, implies R is transitive.

We prove this directly using only the hypothesis for n = 2.

Assume (x, y), (y, z) ∈ R. The definition of composition implies (x, z) ∈ R2 . But
R2 ⊆ R, so (x, z) ∈ R. Thus R is transitive.

Discussion

Theorem 1.13.1 gives an important characterization of the transitivity property.

Notice that, since the statement of the theorem was a property that was to be proven
for all positive integers, induction was a natural choice for the method of proof.
Exercise 1.13.1. Prove that a relation R on a set A is transitive if and only
if R2 ⊆ R. [Hint: Examine not only the statement, but the proof of the Theorem
1.13.1.]

1.14. Connection Matrices v.s. Composition. Recall: The boolean prod-

uct of an m × k matrix A = [aij ] and a k × n matrix B = [bij ], denoted A B = [cij ],
is defined by
cij = (ai1 ∧ b1j ) ∨ (ai2 ∧ b2j ) ∨ · · · ∨ (aik ∧ bkj ).
Theorem 1.14.1. Let X, Y , and Z be finite sets. Let R1 be a relation from X to
Y and R2 be a relation from Y to Z. If M1 is the connection matrix for R1 and M2
is the connection matrix for R2 , then M1 M2 is the connection matrix for R2 ◦ R1 .

We write MR2 ◦R1 = MR1 M R2 .

Corollary 1.14.1.1. MRn = (MR )n (the boolean nth power of MR ).
Exercise 1.14.1. Let the relation R on {a, b, c, d} be represented by the matrix
 
1 1 1 0
1 0 0 0
 .
1 0 0 1
0 0 1 0
Find the matrix that represents R3 .