95

2.8 Linear Functionals

Definition 4.10 (Linear Functional). A linear functional f is a linear

operator with domain in a vector space X and range in the scalar field K of
X; where
• K = R if X is real,
• K = C if X is complex.
Thus, a linear functional f : D(f ) ⊂ X → K, satisfies the following:
For any x, y ∈ D(f ) and α, β ∈ K,

f (αx + βy) = αf (x) + βf (y).

Definition 4.11 (Bounded Linear Functional). A linear functional

f : D(f ) ⊂ X → K

is called bounded if there exists a real number c ≥ 0 such that for all x ∈
D(f ),

|f (x)| ≤ c∥x∥.

The norm of f is given by

|f (x)|
∥f ∥ = sup ,
x∈D(f ) ∥x∥
x̸=0

or equivalently,

∥f ∥ = sup |f (x)|.
x∈D(f )
∥x∥=1
96

Note: For a bounded linear functional, we always have

|f (x)| ≤ ∥f ∥∥x∥ for all x ∈ D(f ).

Theorem 4.12 ( Continuity and Boundedness). A linear functional

f with domain D(f ) in a normed space is continuous if and only if f is
bounded.

Examples.

Example 4.9 (Norm). The norm ∥ · ∥ : X → R on a normed space

(X, ∥ · ∥) is a functional on X which is not linear.

Example 4.10 (Dot Product). The familiar dot product with one factor
kept fixed defines a functional f : R3 → R by

f (x) = x · a, for x = (ξ1, ξ2, ξ3) ∈ R3,

where a = (a1, a2, a3) ∈ R3 is fixed. f is linear and bounded. By the

Cauchy–Schwarz inequality, we obtain
v v
3
X
u 3 u 3
uX uX
|f (x)| = |x · a| ≤ |ξj aj | ≤ t 2
|ξj | t |aj |2 = ∥x∥∥a∥.
j=1 j=1 j=1

Thus, by taking supremum over all x ∈ R3 with norm 1, we obtain

∥f ∥ ≤ ∥a∥.
 97

On the other hand, by the boundedness of f , we have

|f (x)| ≤ ∥f ∥∥x∥ for all x ∈ R3

|f (x)|
=⇒ ∥f ∥ ≥ for all 0 ̸= x ∈ R3.
∥x∥

By taking x = a,

|f (a)|
∥f ∥ ≥
∥a∥
∥a∥2
=
∥a∥
= ∥a∥.

Thus, we obtain ∥f ∥ = ∥a∥.

Example 4.11 (Space l2). We can obtain a linear functional f on the

Hilbert space by choosing a fixed a = (aj ) ∈ l2 and setting

∞
X
f (x) = ξj aj for x = (ξj ).
j=1
98

This series converges absolutely (for well definedness of f ) and f is bounded,

since by the Cauchy–Schwarz inequality, we have
∞
X
|f (x)| = ξj aj
j=1
∞
X
≤ |ξj aj |
j=1
v v
∞
u∞
uX uX
u
≤t |ξj |2t |aj |2
j=1 j=1

= ∥x∥∥a∥.

Thus, by taking supremum over all x ∈ l2 with norm 1, we obtain

∥f ∥ ≤ ∥a∥.

On the other hand, by the boundedness of f , we have

|f (x)| ≤ ∥f ∥∥x∥ for all x ∈ l2

|f (x)|
=⇒ ∥f ∥ ≥ for all 0 ̸= x ∈ l2.
∥x∥
By taking x = a,
|f (a)|
∥f ∥ ≥
∥a∥
∥a∥2
=
∥a∥
= ∥a∥.

Thus, we obtain ∥f ∥ = ∥a∥.

 99

Example 4.12 (Space C[a, b]). An important functional on C[a, b] is ob-

tained if we choose a fixed t0 ∈ [a, b] and set

f (x) = x(t0) for x ∈ C[a, b].

Clearly f is linear. Moreover, f is bounded and has norm ∥f ∥ = 1. In fact,

we have

|f (x)| = |x(t0)| ≤ max |x(t)| = ∥x∥∞.

t∈[a,b]

Thus, by taking supremum over all x ∈ C[a, b] with norm 1, we obtain

∥f ∥ ≤ 1.

On the other hand, by the boundedness of f , we have

|f (x)| ≤ ∥f ∥∥x∥∞ for all x ∈ C[a, b]

|f (x)|
=⇒ ∥f ∥ ≥ for all 0 ̸= x ∈ C[a, b].
∥x∥∞
By taking x = 1 (constant function), we have ∥x∥∞ = 1 and
|f (1)|
∥f ∥ ≥ = 1.
∥x∥∞
Thus, we obtain ∥f ∥ = 1.

Example 4.13 (Definite Integral). The definite integral is a functional

on the space C[a, b]. Define
Z b
g(x) = x(t) dt, x ∈ C[a, b].
a
100

Clearly, g is linear. We claim that g is bounded, with norm ∥g∥ = b − a. In

fact,
Z b
|g(x)| = x(t) dt
a
Z b
≤ |x(t)| dt
a
Z b
≤ max |x(t)| dt
a t∈[a,b]

= (b − a) max |x(t)|
t∈[a,b]

= (b − a)∥x∥∞.

This shows that g is bounded. Moreover, by taking supremum over all

x ∈ C[a, b] with norm 1, we obtain

∥g∥ ≤ b − a.

On the other hand, by the boundedness of g, we have

|g(x)| ≤ ∥g∥∥x∥∞ for all x ∈ C[a, b]

|g(x)|
=⇒ ∥g∥ ≥ for all 0 ̸= x ∈ C[a, b].
∥x∥∞
By taking x = 1 (constant function), we have ∥x∥∞ = 1 and
|g(1)| b − a
∥g∥ ≥ = =b−a
∥x∥∞ 1
Thus, we obtain, ∥g∥ = b − a.
 101

Definition 4.12 (Algebraic Dual Space). The set of all linear func-
tionals defined on a vector space X can also form a vector space. This space
is denoted by X ∗ and is called the algebraic dual space of X. Its algebraic
operations of vector space are defined in a natural way as follows:
The sum f1 + f2 of two functionals f1 and f2 is the functional s whose value
at every x ∈ X is

s(x) = (f1 + f2)(x) = f1(x) + f2(x).

The product αf of a scalar α and a functional f is the functional p whose

value at x ∈ X is

p(x) = (αf )(x) = αf (x).

Definition 4.13 (The Second Algebraic Dual Space). Consider the

algebraic dual (X ∗)∗ of X ∗, whose elements are the linear functionals defined
on X ∗. We denote (X ∗)∗ by X ∗∗ and call it the second algebraic dual
space of X. Thus

X ∗ = {f | f : X → K is linear}

X ∗∗ = {g | g : X ∗ → K is linear}.

Importance: Why do we consider X ∗∗? The point is that we can obtain

an interesting and important relation between X and X ∗∗.
We choose the notations:
102

Space General element Value at a point

X x –
X∗ f f (x)
X ∗∗ g g(f )

We can obtain g ∈ X ∗∗, which is a linear functional defined on X ∗, by

choosing a fixed x ∈ X and setting

g(f ) = gx(f ) = f (x) (x ∈ X fixed, f ∈ X ∗ variable). (35)

The subscript x is a reminder that we got g by the use of a certain x ∈ X.

Claim: gx, as defined by (35), is linear.
For f1, f2 ∈ X ∗ and scalars α, β, by definition, we have

gx(αf1 + βf2) = (αf1 + βf2)(x)

= (αf1)(x) + (βf2)(x)

= αf1(x) + βf2(x)

= αgx(f1) + βgx(f2).

This shows that gx is a linear functional of X ∗. Hence, by definition of X ∗∗,

gx is an element of X ∗∗.

Definition 4.14 (The Canonical Mapping). To each x ∈ X, there

corresponds a gx ∈ X ∗∗. This defines a mapping

C : X → X ∗∗

x 7→ gx.
 103

That is, (Cx)(f ) = gx(f ). C is called the canonical mapping (or natural
mapping) of X into X ∗∗.

Claim: C is linear.
Since its domain is a vector space, for any x, y ∈ X, scalars α, β ∈ K, and
f ∈ X ∗, we have

(C(αx + βy))(f ) = gαx+βy (f )

= f (αx + βy)

= αf (x) + βf (y) (as f ∈ X ∗ is linear)

= αgx(f ) + βgy (f ) (by definition of gx and gy )

= α(Cx)(f ) + β(Cy)(f ).

This implies that C(αx + βy) = αCx + βCy.

Hence, C is linear. Thus, C is also called the canonical embedding of
X into X ∗∗. □
To understand and motivate the term “embedding,” we explain the concept
of isomorphism, which is of general interest.
By definition, this is a bijective mapping of X onto Y which preserves the
structure.

Definition 4.15 (Isomorphism of Metric Spaces). An isomorphism

T of a metric space X = (X, dX ) onto a metric space Y = (Y, dY ) is a
bijective mapping that preserves distance, that is, for all x, y ∈ X,

dY (T x, T y) = dX (x, y).
104

Definition 4.16 (Isomorphism of Vector Spaces). Let X and Y be

vector spaces over the same field. An isomorphism T of X onto Y is a
bijective mapping that preserves the two algebraic operations of vector spaces.
Specifically, for all x, y ∈ X and scalars α,

T (x + y) = T x + T y, T (αx) = αT x.

Thus, T : X → Y is a bijective linear operator. The vector spaces X and Y

are called isomorphic, and X is said to be isomorphic with Y .
For normed spaces, an isomorphism is a vector space isomorphism that also
preserves norms.

Definition 4.17 (Canonical Mapping and Embedding). The canon-

ical mapping C : X → X ∗∗ is defined by C(x) = gx, where gx is the
element of X ∗∗ defined by

gx(f ) = f (x), for f ∈ X ∗.

The canonical mapping C is linear and injective (we prove in next section).
Hence, C is an isomorphism of X onto the range R(C) ⊂ X ∗∗.
If X is isomorphic to a subspace of a vector space Y , we say that X is
embeddable in Y . Therefore, X is embeddable in X ∗∗, and C is also
called the canonical embedding of X into X ∗∗.

Definition 4.18 (Reflexivity). If C is surjective (hence bijective), so that

R(C) = X ∗∗, then X is said to be algebraically reflexive.