CHAPTER 1: Sets and Relations

1 Binary Relations

In this chapter we are going to define relation formally. A relation in everyday life shows an association
of objects of a set with objects of other sets (or the same set) such as John owns a BMW, Jim has a green
Audi, etc. The essence of relation is these associations. A collection of these individual associations is a
relation, such as the ownership relation between people and automobiles. To represent these individual
associations, a set of "related" objects, such as John and a BMW, can be used. However, simple sets
such as {John, BMW} are not sufficient here. The order of the objects must also be taken into account,
because John owns a BMW but the BMW does not own John, and simple sets do not deal with orders.
Thus sets with an order on its members are needed to describe a relation. Here the concept of ordered
pair is going to be defined first. A relation is then defined as a set of ordered pairs.

Definition 1 (Ordered pair) An ordered pair is a list of a pair of objects with an order associated with
them. If objects are represented by x and y, then we write an ordered pair as (x, y) or (y, x). In general
(x, y) is different from (y, x).

Two ordered pairs (x, y) and (x0 , y0 ) are equal if and only if x = x0 and y = y0 . For example, if the ordered
pair (x, y) is equal to (1, 2), then x = 1, and y = 2. The ordered pair (1, 2) is not equal to the ordered pair
(2, 1).

Definition 2 (Cartesian product) A Cartesian product of two non-empty sets X and Y , denoted as X ×
Y , is defined as the set of all ordered pairs (x, y) where x is an element of X and y is an element of Y .
That is,
X ×Y := {(x, y)| x ∈ X and y ∈ Y }.

Definition 3 (Binary relation) Let X and Y be two non-empty sets. A subset R of X × Y is called a
binary relation from X to Y . In other words, a binary relation from a set X to a set Y is a set of ordered
pairs (x, y) where x is an element of X and y is an element of Y .

Example 1 If X = {1, 2, 3} and Y = {4, 5}, then {(1, 4), (2, 5), (3, 5)}, for example, is a binary relation
from X to Y . However, {(1, 1)} is not a binary relation from X to Y because 1 is not in Y . 

If X = Y , i.e., if R is a relation from X to X, we simply say that it is a relation on X. In other words, R is

a relation on X if and only if R ⊆ X 2 . If (x, y) ∈ R, then we think of R as associating the object x with the
object y, and if {(x, y), (y, x)} ∩ R = ∅, then we understand that there is no connection between x and y
as envisaged by R. Conventionally, we write xRy instead of (x, y) ∈ R.

1
2 Preorders and Equivalence Relations

Definition 4 (Preorder) A binary relation % on a non-empty set X is called a preorder if it has the
following properties:

(i) Reflexivity: x % x for all x ∈ X,

(iii) Transitivity: x % y and y % z =⇒ x % z for all x, y, z ∈ X.

Definition 5 (Equivalence relation) A binary relation ∼ on a non-empty set X is called an equivalence

relation if it has the following properties:

(i) Reflexivity: x ∼ x for all x ∈ X,

(ii) Symmetry: x ∼ y =⇒ y ∼ x for all x, y ∈ X,

(iii) Transitivity: x ∼ y and y ∼ z =⇒ x ∼ z for all x, y, z ∈ X.

For any x ∈ X, the equivalence class of x is defined as the set

[x] := {y ∈ X| y ∼ x}.

Definition 6 (Partition) A partition of a non-empty set X is a class {Xi } of non-empty subsets of X such
that (i) i Xi = X, and (ii) Xi ∩ X j = ∅ for any two Xi and X j in {Xi }. The Xi ’s are called the partition
S

sets.

If X = {1, 2, 3, 4, 5}, then {{1, 2, 3}, {4, 5}} and {{1, 2, 5}, {3, 4}} are two different partitions of X.
If X = R, then it can be partitioned into the infinitely many closed-open intervals of the form [n, n + 1)
where n is an integer.
We now show that a given equivalence relation ∼ on X determines a natural partition of X. Let the
relation ∼ on X satisfies reflexivity, symmetry and transitivity. We show that all distinct equivalence
classes form a partition of X. By reflexivity, x ∈ [x] for each element x ∈ X, so each equivalence set is
non-empty and their union is X. It remains to show that any two equivalence sets [x1 ] and [x2 ] are either
disjoint or identical. We prove by showing that if [x1 ] and [x2 ] are not disjoint then they must be identical.
Suppose that z ∈ [x1 ] ∩ [x2 ]. Then z ∼ x1 and z ∼ x2 , and by symmetry, x1 ∼ z. Let y ∈ [x1 ] which implies
that y ∼ x1 . Since y ∼ x1 and x1 ∼ z, transitivity implies that y ∼ z. Then, again by transitivity, y ∼ z
and z ∼ x2 implies that y ∼ x2 , so that y ∈ [x2 ]. Hence, [x1 ] ⊆ [x2 ]. By a similar logic, one can show that
[x2 ] ⊆ [x1 ]. From this we can conclude that [x1 ] = [x2 ].
The converse statement is somewhat trivial. Consider a partitions {Xi } of a set X, and let R be a
relation on X defined by, for any two elements x and y of X,

xRy if and only if x and y belong to the same element Xi of {Xi }.

We think of the elements of X being distributed into a number of sets, each of which is a member of a
given partition. Every element of X is exactly in one set. Notice that every element x of X is in the same

2
set as itself. For any x and y of X, if x and y are in the same set, so are y and x. Finally, For any x, y and z
of X, if x and y are in the same set, and y and z are also, so are x and z. Thus, the relation R is reflexive,
symmetric and transitive, and hence is an equivalence relation. Thus, there is no real distinction between
partitions and equivalence relations.

Example 2 (The Indifference Set) Let x = (x1 , x2 ) be the quantities of two goods 1 (apple) and 2 (or-
ange), and x is an element of X ⊆ R2+ , which is called the consumption set. Define the preference
relations % on X of a consumer as follows. We read x % y as “x is at least as good as y”. From %, we can
derive two other important relations on X:

(a) The strict preference relation, , defined by x  y ⇐⇒ x % y but y  x and read “x is (strictly)
preferred to y”.

(d) The indifference relation, ∼, defined by x ∼ y ⇐⇒ x % y and y % x and read “x is indifferent to y”.

The relation % is rational if (i) it is complete, i.e., for all x, y ∈ X, we have that x % y or y % x (or both),
and (ii) it is transitive, i.e., for all x, y, z ∈ X, if x % y and y % z, then x % z. It is easy to prove the
following property.

Lemma 1 If % is rational, then the indifference relation ∼ is an equivalence relation, but the strict
preference relation  is not.

For a consumption bundle x ∈ X, the set [x] = {y ∈ X|y ∼ x} is called the indifference set of x. Notice
that ∼ induces a partition of X and vice versa. 

3 Partial Orders

Definition 7 (Partial order) A binary relation % on a non-empty set X is called a partial order if it has
the following properties:

(i) Reflexivity: x % x for all x ∈ X,

(ii) Antisymmetry: x0 % x00 and x00 % x0 imply x0 = x00 for all x0 , x00 ∈ X,

(iii) Transitivity: x0 % x00 and x00 % x000 imply x0 % x000 for all x0 , x00 and x000 ∈ X.

Definition 8 (Partially ordered set) A partially ordered set (written “poset”) is a set X on which there
is a binary relation % that is reflexive, antisymmetric and transitive.

Definition 9 Two elements x0 and x00 of a partially ordered set are ordered if either x0 % x00 or x00 % x0 ;
otherwise x0 and x00 are unordered.

Definition 10 (Chain) A partially ordered set is a chain if it does not contain an unordered pair of
elements.

Example 3 The following are partially ordered sets.

3
 (a) The real line R with usual ordering relation ≥ on the real numbers is a poset.
(b) Given any two vectors x0 , x00 in the m-dimensional Eucledian space Rm = {x = (x1 , . . . , xm ) | xi ∈ R
for all i = 1, . . . , m}, x0 ≥ x00 if xi0 ≥ xi00 for all i = 1, . . . , m. Rm with the vector ordering relation ≥
is a poset.
(c) The power set, P(X), of a set X is the set of all subsets of X. The power set P(X) with the set
inclusion ordering relation ⊇ is a poset. If X 0 and X 00 are distinct subsets of X with X 0 ⊆ X 00 , then
X 0 ⊂ X 00 .
(d) The lexicographic ordering relation lex on Rm is such that x0 lex x00 in Rm if either x0 = x00 or
there is some i0 with 1 ≤ i0 ≤ m, xi0 = xi00 for each i with 1 ≤ i < i0 , and xi00 < xi000 . The set Rm with
the ordering relation lex is a poset, indeed, is a chain. 

Definition 11 (Bound of a set and bounded set) Suppose that X is a partially ordered set with respect
to %, and X 0 is a subset of X. If u ∈ X such that u % x for every x ∈ X 0 , then u is an upper bound of X 0 .
Similarly, if l ∈ X such that l - x for every x ∈ X 0 , then l is a lower bound of X 0 . If a subset X 0 of a poset
X has an upper (lower) bound, then X 0 is bounded above (below). In case X 0 has both an upper bound
and a lower bound, then X 0 is a bounded set.

First, observe that a subset X 0 of a poset X may be unbounded. Consider X = R, the set of real numbers,
and X 0 = Z = {. . . , −2, −1, 0, 1, 2, . . .}, the set of integers. Then Z is an unbounded set in R. On the
other hand, if X 0 = N = {1, 2, . . .}, the set of natural numbers, then X 0 is bounded below by 1. Next,
let X = R and X 0 = {5, 10, 15}. Then both 15 and 20 are upper bounds of X 0 . Now let X = R and
X 0 = [a, b). The real number b is an upper bound of X 0 which is not contained in X 0 . Two important
observations emerge. First, a subset X 0 of a poset X may or may not contain its upper (lower) bounds (if
they exist). This gives rise to the notion of a greatest element and a least element.

Definition 12 (Greatest and least elements) Suppose that X is a partially ordered set with respect to
%, and X 0 is a subset of X. If x0 ∈ X 0 such that x0 % x for every x ∈ X 0 , then x0 is a greatest element of X 0 .
Similarly, if z0 ∈ X 0 such that z0 - x for every x ∈ X 0 , then z0 is a least element of X 0 .

A greatest (least) element of X 0 is trivially an upper (lower) bound of X 0 . For example, if X = R and
X 0 = {5, 10, 15}, then 2 is not a least element of X 0 , whereas 5 is. Notice that a subset X 0 of a poset X
can have at most one greatest (least) element. Next, a subset X 0 of a poset X may have many different
upper and lower bounds. This gives rise to the notion of a supremum and an infimum.

Definition 13 (Supremum and infimum) Suppose that X is a partially ordered set with respect to %,
and X 0 is a subset of X. An element of X is a supremum or a least upper bound of X 0 (with respect to X),
denoted supX (X 0 ), if it is the case that (a) supX (X 0 ) % x for all x ∈ X 0 , and (b) for any u ∈ X such that
u % x for all x ∈ X 0 , it holds that supX (X 0 ) - u. Similarly, an element of X is an infimum or a greatest
lower bound of X 0 (with respect to X), denoted infX (X 0 ), if it is the case that (a) infX (X 0 ) - x for all
x ∈ X 0 , and (b) for any l ∈ X such that l - x for all x ∈ X 0 , it holds that infX (X 0 ) % u.

The above definition says that if the set of upper (lower) bounds of X 0 has a least (greatest) element, then
this least (greatest) element is the supremum (infimum) of X 0 . Two important points must be kept in mind.
First, a supremum or an infimum of a set may not exist. If it exists, then it must be unique. Second, one
must be clear about the underlying set in expressing sup(X 0 ) or inf(X 0 ). Consider the following example.

4
Example 4 Suppose X = R, Y = [0, 1) ∪ {2}, and X 0 = [0, 1). Then supX (X 0 ) = 1 6= 2 = supY (X 0 ). 

Notice also that, if a subset X 0 of a poset X contains supX (X 0 ), then the supremum is the greatest element.
Similar property holds for the infimum. Following is an important result concerning the set of real
numbers.

Lemma 2 (The supremum property) Every non-empty set of real numbers that is bounded above has
a supremum. This supremum is a real number.

Proof. Let X ⊆ R be a non-empty subset of real numbers, and let U := {u ∈ R | u ≥ x for all x ∈ X},
the set of upper bounds of X. By assumption, U is non-empty. By the axiom of completeness,1 there
exists a real number α such that

x ≤ α ≤ u, for all x ∈ X and u ∈ U.

Because x ≤ α for all x ∈ X, α is an upper bound of X. This implies that α ∈ U with the property that
α ≤ u for all u ∈ U, and hence, α is a supremum of X. 
The following definition introduces the notion of a maximal (minimal) element, which is, in general,
different from a greatest (least) element.

Definition 14 (Maximal and minimal element) If x0 is in X 0 and there does not exist any x00 ∈ X 0 with
x00  x0 (x0  x00 ), then x0 is a maximal (minimal) element of X 0 .

A greatest (least) element is a maximal (minimal) element. Thus, it is a stronger notion that maximal
(minimal) element. A poset can may have any number of maximal (minimal) element. For example,
In the fence a1 > b1 < a2 > b2 < a3 > b3 < . . ., all the ai ’s are maximal, and all the bi ’s are minimal.
Consider another example. Let X be the set with at least two elements, and S = {{x} | x ∈ X} be the
subset of P(X) consisting of singletons. Take the usual set inclusion ordering, i.e., for any two sets X
and Y in P(X), X - Y if and only if X ⊆ Y . This is the discrete poset – no two elements are comparable.
Thus, every element {x} ∈ S is maximal and minimal, and for any x0 and x00 , neither {x0 } ⊂ {x00 } nor
{x00 } ⊂ {x0 }. Hence, we can conclude that distinct maximal (minimal) elements are unordered.

Example 5 (Consumer theory) In economics, we relax the axiom of antisymmetry, using preorders
instead of partial orders. Let X ⊆ Rn+ . Preferences of a consumer are represented by a preorder % (as
in Example 2). The preference relations are never assumed to be antisymmetric. In this context, for any
B ⊂ X, we call x ∈ B a maximal element if x % y for all y ∈ B, and it is interpreted as the consumption
bundle that is not dominated in the sense that x ≺ y for any y ∈ B. The notion of greatest element for
a preference preorder would be that of most preferred choice. That is, some x ∈ B with y ∈ B implying
that x  y. An obvious application is to the definition of demand correspondence. An element p of Rn++
is called a price system which maps every consumption bundle x into its market value p · x ∈ R+ . An
income level m is an element of R+ . The budget correspondence B is a correspondence from Rn++ × R+
to X which is given by
B(p, m) = {x ∈ X | p · x ≤ m}.
1 This
axiom says that, if L and H are two non-empty subsets of real numbers with the property that for all l ∈ L and for all
h ∈ H we have l ≤ h, then there is a real number α such that l ≤ α ≤ h for all l ∈ L and for all h ∈ H.

5
The demand correspondence maps any price p and any level of income m into the set of maximal ele-
ments (with respect to %) of B(p, m), which is given by

D(p, m) = {x ∈ X | x is a maximal element of B(p, m)}.

It is called demand correspondence because the theory predicts that for given p and m, the rational choice
of a consumer x∗ will be some element of D(p, m). 

4 Functions

4.1 Definition and Properties

A function from a set X to a set Y is a rule that assigns to each element x of X a unique element y of Y .
We say that y is the image of x under f , and write y = f (x). Conversely, x is an element of the preimage
or inverse image of y, written x ∈ f −1 (y). Formally,

Definition 15 (Function) Let X and Y be two sets. A function f from X to Y , written f : X −→ Y , is a

relation from X to Y with the property that for each x ∈ X there exists a unique element y ∈ Y such that
(x, y) ∈ f .

The underlying set X is called the domain of f , and f (X) is its range. If A ⊆ X, then its image under f
is given by
f (A) := {y ∈ Y | there exists x ∈ A such that y = f (x)} = ∪x∈A { f (x)}.
Given a subset B of Y , its inverse image is given by

f −1 (B) := {x ∈ X | f (x) ∈ B}.

If f : X −→ Y and f (X) = Y , then f is surjective or onto. A function f is injective or one-to-one if for any
two distinct elements x1 , x2 ∈ X, we have f (x1 ) 6= f (x2 ). A function is bijective if it is both one-to-one
and onto, i.e., if each element y ∈ Y has a unique inverse image in X. Let f : X −→ Y and g : Y −→ Z.
Then their composition, g ◦ f is a function from X to Z and is defined by (g ◦ f )(x) = g( f (x)). We have
(h ◦ g) ◦ f = h ◦ (g ◦ f ) = h ◦ g ◦ f , i.e., the composition is associative.

Theorem 1 Let f : X −→ Y be a function and B = {Bi | i ∈ I} a family of subsets of Y . Then

(a) f −1 (∪i∈I Bi ) = ∪i∈I f −1 (Bi ),

(b) f −1 (∩i∈I Bi ) = ∩i∈I f −1 (Bi ).

Proof. The proof is left as an exercise. 

Definition 16 (Increasing function) A function f : X −→ Y is (strictly) increasing if x0 > x00 in X implies

that f (x0 ) (>) ≥ f (x00 ).

6
4.2 Representation of Binary Relations

Definition 17 A function f : X −→ R is a utility function, where X ⊆ Rm

+ , representing preference rela-
tion % if, for all x, y ∈ X, x % y ⇐⇒ f (x) ≥ f (y).

Proposition 1 A preference relation % can be represented by a utility function only if it is rational.

Proof. To prove this proposition, we must show that if there is a utility function that represents
the preference relation %, then % must be complete and transitive. First, because f (·) is a real valued
function defined on X, it must be that for any x, y ∈ X, either f (x) ≥ f (y) or f (y) ≥ f (x). Because f (·)
represents %, this implies either that x % y or that y % x. Hence, % must be complete. Next, Suppose
that x % y and y % z. Because f (·) represents %, it must be that f (x) ≥ f (y) and f (y) ≥ f (z). Therefore,
f (x) ≥ f (z), which implies that x % z, and hence % is transitive. 

4.3 Examples and Graphical Representation

Let f : X −→ Y be a function, and consider the subset of X ×Y

G f := {(x, y) ∈ X ×Y | y = f (x)}. (1)

The above set is called the graph of the function f . In what follows we consider real valued functions
f : X −→ R, where X ⊆ R2 , that assign to each x = (x1 , x2 ) ∈ X a number f (x1 , x2 ) in R.

Example 6 Following are the examples of real valued functions from R2 to R.

(i) f (x1 , x2 ) = α1 x1 + α2 x2 (Linear),

(ii) f (x1 , x2 ) = Ax1α1 x2α2 (Cobb-Douglas),

(iii) f (x1 , x2 ) = min{α1 x1 , α2 x2 } (Leontief),

i 1
−ρ − ρ
h
−ρ
(iv) f (x1 , x2 ) = A α1 x1 + α2 x2 (Constant elasticity of substitution). 

Given a real valued function f : X −→ R, consider the set I(x) := {y ∈ R2 | f (x) = f (y)}. Thus, if x ∼ y
then the function f (.) is constant on the set I(x). Let this value be α. The graph of I(x) defines an
indifference curve or a level curve at the level α. The graphical representation of functions from R2 to
R is often difficult since the graphs are three dimensional. We represent such functions by their level
curves. How such level curves are drawn is a topic of discussion of Chapter 3.

5 Correspondences

A correspondence ϕ from X to Y is a function that to each element x in X assigns a non-empty subset

ϕ(x) of Y . Hence, a correspondence from X to Y , denoted ϕ : X →→ Y , is a function from X to P(Y ).
Notice that a function associates with every element x ∈ X an element y ∈ Y . In saying this we are not

7
making the distinction between the element y in Y and the subset {y} of Y which consists of one point
only. Given a correspondence ϕ : X −→ P(Y ), consider the subset Gϕ of X ×Y defined by

Gϕ := {(x, y) ∈ X ×Y | y ∈ ϕ(x)}. (2)

The set Gϕ is called the graph of ϕ. Conversely, if G is a subset of X ×Y such that for every x ∈ X the
set of points y ∈ Y with (x, y) ∈ G is non-empty, then G is the graph of a correspondence, i.e.,

ϕ(x) := {y ∈ Y | (x, y) ∈ G}. (3)

We now introduce the notion of inverse image(s) of a correspondence. Let ϕ : X →→ Y be a correspon-

dence, where X ⊆ Rn and Y ⊆ Rm , and W be any subset of Y . Define the upper inverse of W under ϕ,
denoted ϕ+−1 (W ), by
ϕ+−1 (W ) := {x ∈ X | ϕ(x) ⊆ W },
and the lower inverse of W under ϕ, denoted ϕ −1 (W ), by

ϕ −1 (W ) := {x ∈ X | ϕ(x) ∩W 6= ∅}.

Example 7 (Best response correspondence) Consider the following two-player normal form game Γ =
hN, {Si }i∈N , {ui (·)}i∈N i, where N = {1, 2} is the set of players, S1 = {s1 , s2 , s3 } and S2 = {t1 , t2 , t3 , t4 }
are the strategy sets of players 1 and 2 respectively, and ui (s, t) is the payoff function of player i which is
given as follows. A strategy t ∗ of player 2 is a best response against a strategy s of player 1 if u2 (s, t ∗ ) ≥

Table 1: Payoff matrix

t1 t2 t3 t4
s1 1, 1 1, 1 0, 0 0, 0
s2 0, 0 2, 2 2, 2 0, 0
s3 1, 0 1, 0 0, 0 3, 3

u2 (s, t) for all t ∈ S2 . In a similar fashion one can define the best response of player 1. Notice that the
best response of player 2 is a correspondence ϕ as there is no unique optimal strategy against s1 and s2 .
The best response correspondence of player 2 is given by ϕ(s1 ) = {t1 , t2 }, ϕ(s2 ) = {t2 , t3 }, ϕ(s3 ) = {t4 }.
In this example, the upper and lower inverses of the set W = {t2 , t3 } ⊆ S2 are given by ϕ+−1 (W ) = {s2 }
and ϕ −1 (W ) = {s1 , s2 }.