Chapter 1

PRELIMINARIES

1.1 Introduction:

Functional analysis is the branch of mathematics where vector spaces and

operators on them are in focus. In linear algebra, the discussion is about fi-
nite dimensional vector spaces over any field of scalars. The functions are
linear mappings which can be viewed as matrices with scalar entries. If the
functions are mappings from a vector space to itself, the functions are called
operators and they are represented by square matrices. In functional analy-
sis, the vector spaces are in general infinite dimensional and not all operators
on them can be represented by matrices. Hence the theory becomes more
complicated, but nonetheless there are many similarities.

Functional analysis has its origin in ordinary and partial differential equa-
tions, and in the beginning of the 20th century it started to form a discipline
of its own via integral equations. However, for a long time there were doubts
wether the mathematical theory was rich enough. Despite the efforts of
many prominent mathematicians, it was not sure if there were sufficiently
many functionals to support a good theory, and it was not until 1920 that
the question was finally settled with the celebrated HahnBanach theorem.

Seen from the modern point of view, functional analysis can be consid-
ered as a generalization of linear algebra. However, from a historical point of
view, the theory of linear algebra was not developed enough to provide a ba-
sis for functional analysis at its time of creation. Thus, to study the history
of functional analysis we need to investigate which concepts of mathematics
that needed to be completed in order to get a theory rigorous enough to
support it. Those concepts turn out to be functions, limits and set theory.

1
 For a long time, the definition of a function was due to Euler in his In-
troduction in Analysis Infinite term from 1748 which read: A function of a
variable quantity is an analytic expression composed in any way whatsoever
of the variable quantity and numbers or constant quantities.. For the pur-
pose of this report, it is enough to say that the entire focus of this definition
is on the function itself, and the properties of this particular function. What
lead to the success of functional analysis was that the focus was lifted from
the function, and shifted to the algebraic properties of sets of functions The
algebraization of analysis. The process of algebraization led mathematicians
to study sets of functions where the functions are nothing more than abstract
points in the set.

At the same time as the theory is very concrete and applicable to phys-
ical problems, it can be presented in a very abstract way. Some proofs and
results are significantly simplified by introducing the axiom of choice, Zorns
lemma or the Baire category theorem some of the most abstract concepts in
set theory. The main theorems are

1. The Hahn-Banach theorem by Hans Hahn and Stephan Banach,

which states that there are sufficiently many continuous functionals on every
normed space to make the theory of dual spaces and adjoint operators inter-
esting.

2. The uniform boundedness principle or the Banach-Steinhaus theorem

by Banach and Hugo Steinhaus, which states that for any family of continu-
ous linear operators on a Banach space, pointwise boundedness is equivalent
to boundedness in the operator norm.

3. The open mapping theorem or the Banach-Schauder theorem by Ba-

nach and Juliusz Pawel Schauder, which classifies the open mappings between
two Banach spaces.

To a certain extent, functional analysis can be described as infinite-

dimensional linear algebra combined with analysis, in order to make sense of
ideas such as convergence and continuity. It follows that we will make exten-
sive use of these topics, so in this chapter we briefly recall and summarize the
various ideas and results which are fundamental to the study of functional
analysis. We must stress, however, that this chapter only attempts to review

2
the material and establish the notation that we will use. We do not attempt
to motivate or explain this material, and any reader who has not met this
material before should consult an appropriate textbook for more information.

Section 1.2 discusses the basic results from linear algebra that will be re-
quired. The material here is quite standard although, in general, we do not
make any assumptions about finite-dimensionality except where absolutely
necessary. Section 1.3 discusses the basic ideas of metric spaces. Metric
spaces are the appropriate setting in which to discuss basic analytical con-
cepts such as convergence of sequences and continuity of functions. The
ideas are a natural extension of the usual concepts encountered in elemen-
tary courses in real analysis. In general metric spaces no other structure is
imposed beyond a metric, which is used to discuss convergence and continu-
ity. However, the essence of functional analysis is to consider vector spaces
(usually infinite-dimensional) which are metric spaces and to study the in-
terplay between the algebraic and metric structures of the spaces, especially
when the spaces are complete metric spaces.

An important technical tool in the theory is Lebesgue integration. This

is because many important vector spaces consist of sets of integrable func-
tions. In order for desirable metric space properties, such as completeness, to
hold in these spaces it is necessary to use Lebesgue integration rather than
the simpler Riemann integration usually discussed in elementary analysis
courses. Of the three topics discussed in this introductory chapter, Lebesgue
integration is undoubtedly the most technically difficult and the one which
the prospective student is most likely to have not encountered before. In this
book we will avoid arcane details of Lebesgue integration theory. The basic
results which will be needed are described in Section 1.4, without any proofs.
For the reader who is unfamiliar with Lebesgue integration and who does
not wish to embark on a prolonged study of the theory, it will be sufficient
to accept that an integration process exists which applies to a broad class
of Lebesgue integrable functions and has the properties described in Section
1.4, most of which are obvious extensions of corresponding properties of the
Riemann integral.

This project will introduce the methods of linear functional analysis. Our
basic goal here is to perform analysis on infinite-dimension vector spaces.
Because not all of our linear algebra properties hold for infinite spaces, we

3
extend ideas like the length of a vector into an abstract setting: the norm.
Much like in real and complex analysis, we will give additional structures to
certain spaces to yield special properties. Through different examples, we
will gain an understanding of normed and inner product spaces, bounded
linear operators, compact metric space.

1.2 Linear Algebra :

Here, we will recall some essential concepts from Linear Algebra for use
throughout the project. We will be working with the sets R, C and N. Note
that R and C are both algebraic fields, and when giving examples, theorems,
or definitions that apply equally to RR and C, we will simply use F to denote
either set.

Definition 1.2.1 : A vector space over F is a non-empty set V with two

operations, vector addition and scalar multiplication, where vector addition
is a function from V × V → V defined by (x,y) = x + y, and scalar multipli-
cation is a function from F × V → V defined by (α , x) = αx. The following
properties hold for any α, β ∈ F and any x , y , z ∈ V .

(i) x + y = y + x, x + (y + z) = (x + y) + z;

(ii) there exists a unique 0 ∈ V such that for all x ∈ V , x + 0 = 0 + x = x;

(iii) for all x ∈ V , there exists a unique -x ∈ V such that x + (-x) = (-x)
+ x = 0;

(iv) 1x = x, α(βx) = (αβ)x;

(v) α(x + y) = αx + αy, (α + β)x = αx + βx.

Note that if F = R, then V is a real vector space, and if F = C, then V is

a complex vector space. For our purposes, we will usually just use the term
vector space as most results about vector spaces will apply equally well to
both the real and complex case.

Definition 1.2.2 : Let V be a vector space, and U a non-empty subset

of V . Then U is a linear subspace of V if U is a vector space under the same

4
vector addition and scalar multiplication as V .

Rather than proving Definition 1.2.1 (i)-(v) for U, we can use the sub-
space test:
Let U be a non-empty subset of a vector space V . For all α ∈ F and x , y
∈ U , if

(i) x + y ∈ U ;

(ii) αx ∈ U

then U is a subspace of V . Note that the set {0} ∈ V is a linear subspace.

Definition 1.2.3 : Let V be a vector space, v = {v1 , v2 , ........., vk } ⊂

V , k ≥ 1 be a finite set, and A be an arbitrary non-empty subset of V .

(i) A linear combination of v is any vector x of the form:

x = α1 v1 + α2 v2 + ....... + αk vk ∈ V ;
for any scalars α1 , α2 , ........., αk .

(ii) The set v is linearly independent if:

α1 , α2 , ........., αk = 0 ⇒ α1 = α= ...... = αk = 0
(iii) The span of A (denoted SpA) is the set of all linear combinations of
all finite subsets of A. In other words, SpA is the intersection of all linear
subspaces of V containing A. By this definition, SpA is the smallest linear
subspace of V that still contains A.

(iv) Let v be linearly independent, and Spv = V . Then v is a basis of

V . When V has such a finite basis with k elements, every other basis of V
will also have k elements. We say that V is k-dimensional and write dimV
= k.

(v) The set Fk is a vector space over F, and the set of vectors
ê1 = (1, 0, 0, ....., 0), ê2 = (0, 1, 0, ......., 0)..........êk = (0, 0, 0, ......, 1)

5
 forms a basis for Fk , known as the standard basis.

Definition 1.2.4 : Let V and W be vector spaces over F. Then the

Cartesian product V × W is a vector space, where vector addition and scalar
multiplication are defined as follows: For any α ∈ F and (x1 , y1 ), (x2 , y2 ) ∈
V × W.

(x1 , y1 ) + (x2 , y2 ) = (x1 + x2 , y1 + y2 ) and α(x1 , y1 ) = (αx1 , αy1 ):

Definition 1.2.5 : Let V and W be vector spaces over F. Then a function

T : V → W is a linear transformation if, for all α, β ∈ F and x , y ∈ V ,

T (αx + βy) = αT (x) + βT (y):

We define a set of all linear transformations T : V→ W, denoted by L(V,W),

which is also a vector space. If V = W, then we will shorten L(V , V) to
just L(V ). A notable element of L(V) is the identity transformation on V ,
defined by IV (x) = x for x ∈ V .

1.3 Metric Space:

Metric spaces are an abstract setting for the discussion of concepts from
analysis (such as convergence and continuity). The basic tool used here is the
notion of a distance function, or a metric, which we will define in this section.

Definition 1.3.1 : A metric on a set M is a function d : M × M → R

such that for all x , y , z ∈ M:

(i) d(x , y)≥ 0;

(ii) d(x , y) = 0 ⇔ x = y;

(iii) d(x , y) = d(y , x);

(iv) d( , z)≤ d(x , y) + d(y , z) (the triangle inequality).

If d is a metric on M, then we called the pair (M , d) a metric space.

Unless the set M consists of a single point, it can have more than one metric,

6
and we often say the metric space M” rather than (M,d).

Example 1.3.2 : For an integer k≥1, the function d : Fk × Fk → R

defined by
k
! 12
X
d(x, y) = |xj − yj |2
j=1

is the standard metric on the set Fk . Thus we call Fk a metric space.

Definition 1.3.3 : A sequence {xn } in a metric space M is convergent

if for some x ∈ M and every  > 0, there exists an N ∈ N such that

d(xn , x) <  for all n≥ N .

If this is true, we write lim xn = x. or xn → x

n→∞

Definition 1.3.4 : A sequence {xn } in a metric space M is a Cauchy

sequence if for every  > 0 , there exists an N ∈ N such that

d(xm , xn ) <  for all m,n ≥ N.

Definition 1.3.5 : Let (M,d) be a metric space and A ⊂ M.

(i) A is bounded if there exists a number B > 0 such that d(x,y) < B for
all x,y ∈ A.

(ii) A is open if for all x ∈ A, there exists an  < 0 such that Bx () =
{y ∈ M : d(x, y) < } ⊂ A. We call Bx () the open ball with radius centered
at x.

(iii) A is closed if the set M \ A is open.

(iv) A point x ∈ M is a closure point of A if for every  > 0, there exists

a y ∈ A with d(x,y) (we can also say, if there exists an {yn } ⊂ A such that
yn → x).

(v) The closure of A, denoted A, is the set of all closure points in A.

7
 (vi) A is dense in M if A = M.

Theorem 1.3.6 : Let (M,d) be a metric space and A ⊂ M.

(i) A is closed and is the smallest closed set that contains A.

(ii) A is closed, if and only if A = A.

(iii) A is closed if and only if the following implication holds: if {xn } is a

sequence in A and x ∈ M such that xn → x, then x ∈ A.

(iv) x ∈ A if and only if inf{d(x, y) : y ∈ A} = 0.

(v) A is dense if and only if for any element x ∈ M, there exists a sequence
{yn } in A such that yn → x

We can extend the idea of a continuous function from real analysis to

describe general metric spaces.

Definition 1.3.7 : Let (M,dM ) and (N,dN ) be metric spaces, and f : M

→ N.

(i) f is continuous at a point x ∈ M if for all  >0, there exists a δ > 0

such that, for all y ∈ M,

dM (x, y) < δ ⇒ dN (f (x), f (y)) < 

(ii) f is continuous on M if it is continuous at each point of M. (iii) f is

uniformly continuous on M if for all  > 0, there exists a δ > 0 such that, for
all x,y ∈ M,

dM (x, y) < δ ⇒ dN (f (x), f (y)) < 

Definition 1.3.8: A metric space (M,d) is complete if every Cauchy

sequence in (M,d) is convergent. Furthermore, a set A ⊂ M is complete in
(M,d) if every Cauchy sequence in A converges to an element of A.

Definition 1.3.9: Let (M;,d) be a metric space. A set A ⊂ M is compact

if every sequence in A contains a subsequence converging to an element of

8
A. A set A ⊂ M is relatively compact if A is compact.

Theorem 1.3.10: Let (M,d) be a metric space and A ⊂ M. Then:

(i) if A is complete, then A is closed;

(ii) if M is complete, then A is complete if and only if A is closed;

(iii) if A is compact, then A is closed and bounded;

(iv) every closed and bounded subset of Fk is compact (Bolzano-Weirstrass

theorem).

1.4 Lebesgue Integration:

So far we have seen only the uniform metric on C[a, b], but there are other
Rb
metrics that are defined by integrals. If a f (x)dx is the usual Riemann
integral of f ∈ C[a, b], then for 1 ≤ p < ∞, the function dp : C[a, b] × C[a, b] →
R defined by
Z b  p1
dp (f, g) = |f (x) − g(x)|p dx
a

is a metric on mathcalC[a, b]. However, (mathcalC[a, b], dp ) is not a complete

metric space, due to some drawbacks of the Riemann integral. Thus, we will
introduce here the Lebesgue Integral, which is more powerful and works on
a wider class of functions. We will look at metric spaces which are complete
using the metric dp , and will introduce the notion of a measure of a set (AKA
the length;” for example, the bounded interval [a,b] has length b-a). Note
that many sets such as R have infinite measure,” and so we introduce the
+
extended sets of real numbers: R = R ∪ {−∞, ∞} and R = [0, ∞) ∪ {∞}.
The standard algebraic operations apply, with the products 0.∞ = 0.(-∞) =
0, and the operations ∞−∞ and ∞/∞ forbidden (like dividing by zero in R).
P
Definition 1.4.1: A σ-algebra is a collection, , of subsets of X satis-
fying:

9
 P
(i) ∅,X ∈ ;
P P
(ii) S ∈ ⇒X S∈ ;

, n = 1, 2, ......., ⇒ ∪∞
P P
(iii) Sn ∈ n=1 Sn ∈ )
P
A set S ∈ is said to be measurable.
P
Definition 1.4.2: Let X be a set and be a σ-algebra of subsets of X.
P +
A function µ : → R is a measure if it has the following properties:

(i) µ(∅) = 0;
P
(ii) µ is countably additive. In other words, if Sj ∈ , j = 1,2,.... are
pairwise disjoint sets, then
∞
! ∞
[ X
µ Sj = µ(Sj )
j=1 j=1
P
The triple (X, , µ) is called a measure space.
P P
Definition 1.4.3: Let (X, , µ) be a measure space. Then a set S∈
with µ(S) = 0 is said to have measure zero. Also, a given property P(x) is
said to hold almost everywhere if the set {x : P (x) is f alse} has measure
zero.
P
Definition 1.4.4: Let X = N, c be the class of all subsets of N,Pand
for any S ⊂ N, define µc (S) to be thePnumber of elements of S. Then c is
a σ-algebra and µc is a measure on c calledPthe counting measure on N.
Note that the only set of measure zero in (N, c , µc ) is the empty set.
P
Definition
P 1.4.5: Let X = R, L be a σ-algebra,
P and µL be a measure
P L , such that any finite interval I = [a,b] ∈ L and µL (I) = `(I). Then
on
L is called the Lebesgue measure and the sets in are Lebesgue measurable.

P
Note that the sets of measure P zero in (R, L , µL ) are the sets A with
the following property: for any >0, there exists a sequence of intervals
Ij ⊂ R, j = 1, 2,......., such that

10
 S∞ P∞
A⊂ j=1 Ij and j=1 `(Ij ) < 

Definition 1.4.6: A function φ : X → R is simple if it has the form

φ = kj=1 αj X Sj
P

P
for some k ∈ N where αj ∈ R and Sj ∈ , j = 1,......, k. If is non-negative
and simple, then the integral of over X with respect to µ is defined by
Z k
X
φdµ = αj µ(Sj )
X j=1

+
We do allow µ(Sj ) = ∞ here, and use the algebraic rules of R from the
beginning of this section to evaluate (since φ is non-negative, we will not
encounter any problems like ∞ − ∞). The value of the integral may be ∞.
P
Definition 1.4.7: Let (X, , µ) be a measure space. Then a function f
: X → R is measurable if, for every α ∈ R,
P
{x ∈ X : f (x) > α} ∈

Theorem 1.4.8: Let 1 ≤ p ≤ ∞. Then the metric space Lp (X) is com-

plete, and in particular, the sequence space `p is complete.

11
 Chapter 2
LINEAR FUNCTIONAL ANALYSIS

2.1 Linear Functionals:

Definition 2.1.1: A real linear functional is a mapping l(v) : V → R

that is linear with respect to its argument v ∈ V. That is, it must satisfy the
properties
l(u + v) = l(u) + l(v)
l(αv) = αl(v)
for all u, v ∈ V and α ∈ R.

These are the same linearity properties used in the definition of a linear
mapping; indeed, a real linear functional is simply a special case of a linear
mapping where the range space is R. Complex-valued linear functionals can
also be defined. Well almost always consider only real linear functionals, and
will often simply call them linear functionals. Here are some examples.
R1
1. l(u) = 0 u(x)dx is a Linear Functional on the space of integrable
functions on [0,1].

2. l(u) = u( 14 ) is a Linear Functional on the space C 0 of continuous func-

tions on [0,1].

3. l(u) = maxx u(x) is not a Linear Functional on C 0

4. Multiplication by a 1-by-N matrix is a Linear Functional on RN .

Matrix Representation Of Linear Functionals 2.1.2 :

If V is finite dimensional, then upon choosing a basis for V any l : V → R

can be represented as a 1 × dim(V) matrix. You can construct the matrix

12
by computing the action of l on each basis vector.

Example
R1 2.1.3: We will find the matrix representation for the functional
l(u) = −1 u(x)dx defined on P 2 , the space of quadratic polynomials. Any
vector Ru(x) ∈ P 2 can be written as a Linear Combination of basis vectors,
2
u(x) = i=0 ui φi (x). Apply l to this vector and find
2
X Z 1
l(u) = ui φi (x)dx
i=0 −1

R1
Define the 1×3 matrix A with elements A1j = −1 φj (x)dx . Then l(u) = Au
where u is the vector of coefficients in the representation of u in the basis
{φi }. Notice that the matrix A depends on the basis used. For example, if
we use the Vandermonde basis {1, x, x2 } we have
2
A=[2 0 3
]

whereas if we use the Legendre basis {1, x, 32 (x2 − 1)} we have

A=[2 0 0]

Notice also that multiplication of a 1× N matrix A with a vector u is equiv-

alent to the Euclidean inner product between the first (and only) row of A
and u. In finite dimensions, then, we can define c = row1 (A) and write the
action of l as

l(u) = cT u = c.u = uT c.

Again, the elements of c and u will depend on the basis used.

2.2 Bilinear Functionals:

Bilinear functionals can then be defined in terms of linear functionals: A

real bilinear functional maps an ordered pair of vectors to the reals, that is
a real linear functional with respect to each argument.

Definition 2.2.1: A real bilinear functional is a mapping a(u, v) : (u ∈

U, v ∈ V )→ R obeying the properties

13
 a(u + w, v) = a(u, v) + a(w, v) ∀u,w ∈ U, v ∈ V
a(u, v + w) = a(u, v) + a(u,w), ∀ u ∈ U, v,w ∈ V
a(au, v) = aa(u, v) ∀α ∈ R, u ∈ U, v ∈ V
a(u, av) = aa(u, v) ∀α ∈ R, u ∈ U, v ∈ V.
Note that these are two of the properties defining a real inner product. There-
fore, every real inner product is a real bilinear functional; however, it is not
the case that every bilinear functional is an inner product.

Definition 2.2.2: A symmetric bilinear functional is a bilinear func-

tional such that a(u, v) = a(v, u).
R1
Examples of symmetric BFs include a(u, v) = 0
u0 (x)v 0 (x)dx and a(x, y) =
xT Ay where A is any symmetric matrix.

Definition 2.2.3: A positive definite bilinear functional is a bilinear

functional such that a(u, u)≥0 for all u, with the equality obtained only
when u = 0.

A real inner product is a symmetric positive definite bilinear functional.

Symmetric positive definite appears often enough to have its own acronym:
SPD.

Matrix Representations Of Bilinear Functionals 2.2.4 :

When working in finite dimensions we can represent the arguments u ∈

U and v ∈ V in bases. Let {ψj }N M
j=1 be a basis for V and {φi }i=1 be a basis for
U. Applying the bilinear functional to u and v and making use of bilinearity
gives us
M X
X N
a(u, v) = vj ui a(φi , ψj ).
i=1 j=1

Defining the M×N matrix A with elements Aij = a(φi , ψj ), we recognize

that
a(u, v) = uT Av.
As with linear functionals, the matrix representation will depend on the
bases used. If the same bases are used for u and v, and if the functional a is

14
symmetric, then its matrix representation will be symmetric. Notice that a
symmetric functional can be represented by a non-symmetric matrix if dif-
ferent bases are chosen for U and V.

Positive Definiteness And Eigenvalues 2.2.5:

A positive definite bilinear functional has a(u, u)≥0 ∀u 6= 0. Assuming a

matrix representation, this condition becomes

uT Au ≥ 0 ∀u 6= 0

When a is also symmetric a symmetric matrix representation is possible.

Recall that a symmetric matrix has real eigenvalues P
l and orthogonal eigen-
vectors K. Write u as a LC of the eigenvectors, u = j uj Kj . In this basis,
the inequality above is then
X X
ui KiT uj AKj ≥ 0 ∀u 6= 0
i j

which after using AK =λK and the orthogonality of the eigenvectors can be
reduced to X
λi u2i ≥ 0 ∀u 6= 0
i

This will be true iff we have λi > 0 for all i. Thus, a positive definite bilinear
functional has positive definite eigenvalues.

2.3 Bounded Linear Functional:

Defination 2.3.1: A Bounded linear functional f is a bounded linear

operator with range in the scalar field of the normed space X in which the
domain D(f ) lies.

Thus there exists a real number c such that for all x ∈ D(f ),

|f (x)| ≤ ckxk

15
Furthermore, the norm of f is,
!
f (x)
kf k = sup
x∈D(f ),x6=0 kxk
!
kf k = sup |f (x)|
x∈D(f ),kxk=1

|f (x)| < kf kkxk

Bounded, Linear Functionals on C[a, b]:

Definition (Variation) 2.3.2:. Let f : [a, b] → mathbbR. We define

the variation of f as
( n )
X
V ar f = sup |f (ti ) − f (ti−1 )|
i=1

where
a = t0 < t1 < < tn = b
is a partition of [a, b].

Definition (Bounded Variation) 2.3.3:. We say f is of bounded vari-

ation and write f ∈ BV [a,b] if
V arf < ∞
Note BV [a,b] is a vector space. If we define the norm
kf k = |f (a)| + V arf, ∀ f ∈ BV [a, b]
then BV [a,b] is a normed space.

Definition (Riemann-Stieltjes Integral) 2.3.4: Let x ∈ C[a,b] and f

∈ BV [a,b] and Pn be a partition of [a,b]. Define
ηPn = max{t1 − t0 , ..., tn − tn1 }
Consider S defined by
n
X
S(Pn ) = x(ti )[f (ti ) − f (ti1 )]
i=1

16
Then ∃ I ∈ R s.t. ∀  > 0 ∃ δ > 0 s.t.

ηPn < δ ⇒ |IS(Pn )| ≤ .

We call I the Riemann-Stieltjes integral of x with respect to f and write

Z b
I= x(t)df (t).
a

Theorem (Reisz - B.L.F.s on C[a, b]) 2.3.5: Every b.l.f. g on C[a,b]

can be represented by a Riemann-Stieltjes integral
Z b
g(x) = x(t)df (t)
a

and
kgk = V ar(f ).

2.4 Convex Functional:

Definition 2.4.1: A real valued functional p(h) in H is convex if

•CF 1 p(f + g) ≤ p(f ) + p(g)

•CF 2 p(αf ) = kαkp(f )

for all f,g ∈ H and for any complex number α.

It follows from this definition that a convex functional does not take
negative values; for,

0 = p(f − f ) ≤ p(f ) + p(−f ) = 2p(f )

Example 2.4.2: The absolute values |(x)| of a linear functional, and also
norm khk of an element.

17
 Chapter 3
REPRESENTATION OF BOUNDED LINEAR FUNCTIONALS
IN C(X)

3.1 Introduction:

Here in this chapter we restrict ourselves to proving the version of the

Riesz Representation Theorem which asserts that ’positive linear functionals
come from measures’. Thus what we call the Riesz Representation Theorem
is stated in three parts - as theorems 3.2.1, 3.3.3, 3.4.1 corresponding to the
compact metric, compact Hausdorff, and locally compact Hausdorff cases of
the theorem.

3.2 The Compact Metric Case:

Let X be a compact metric space. The symbol BX will denote the Borel
σ-algebra of X; i.e., BX is the smallest σ-algebra of subsets of X which con-
tains all compact sets.

Lemma 3.2.1: Let X be a compact Hausdorff space. Then the follow-

ing conditions on a linear functional τ : C(X) −→ C are equivalent:

(a) τ is positive in the sense of the statement of the previous theorem;

(b) τ is a bounded linear functional and kτ k = kτ (1)k, where we write 1

for the constant function identically equal to one.

Proposition 3.2.2: Every compact metric case is a continuous image of

the (compact metric) space 2N , the Cartesian product of countably infinitely
many copies of a two point space.
proof: The proof relies on two facts:

18
 (i) a compact metric space is totally bounded, i,e,. for any ∈> 0 it is
possible to cover X by finitely many sets with diamater less than ∈; and

(ii) Cantors intersection theorem which states that the intersection, in a

compact metric space, of a decreasing sequence of closed sets with diameters
converging to zero is a singleton set. Indeed, by (i) above, we may find closed
sets Bk : 1 ≤ k ≤ n1 each with diameter at most one, whose union is X. Next,
since Bj is compact, we can find closed sets Bj,k : 1 ≤ k ≤ m0j of diameter
at most 21 whose union is Bj ; choosing n2 = max1≤j≤n1 m0j , and defining

Bj,k = Bj,m0j for m0j ≤ k ≤ n2 . In other words, we may assume that in the la-
beling of Bj,k , the range of the second index k is over the finite index set 1, , n2
and independent of the first index j. Using (i) repeatedly, an easy induction
argument shows that we may, for each q, find a positive integer nq and closed
sets Bj1 ,j2 ,,jq : 1 ≤ ji ≤ ni of diameter at most 1q such that (a) X = ∪nj=1


1
Bj ,
 
nq
and (b) Bj1 ,j2 ,,jq−1 = ∪jq =1 Bj1 ,j2 ,···,jq for each j1 , j2 , · · · , jq−1 .

If j = (j1 , j2 , , jq · · ·), appeal to (ii) above to find that ∩∞

q=1 Bj1 ,j2 ,···,jq =
∞
!
Y
f (j) for a uniquely determined function f : 1, 2, · · · , nq → X. The
q=1
hypotheses ensure that if i and j agree in the first q co-ordinates, then f (i)
and f (j) are at a distance of at most 2q from one another. This shows that
the function f is continuous. Finally (a) and (b) of the last paragraph ensure
that the function f maps onto X. Finally, we may clearly assume, without
loss of generality, that nq = 2mq . Then ! it is clear that there is a (continuous)
∞
Y
surjection from 2N to 1, 2, · · · , 2mq . Combining these two maps we get
q=1
the desired surjection to X.

Theorem 3.2.3: If τ : C(X) −→ C is a linear functional which is positive

in the sense of assuming non-negative real values on non-negative real-valued
continuous functions, then there exists a unique finite positive measure µ
defined on the Borel σ-algebra BX such that
Z
τ (f ) = f dµ.

proof: Let us consider the case when X = 2N . Now suppose that τ is a

19
positive linear functional on C(2N ), which is assumed as in normalised form
so that τ (1) = 1. Let

Πn] (j1 , j2 , · · · , jn ) = (j1 , j2 , · · · , jn )

denote the projection πn] : 2N −→ 2n onto the first n co-ordinates (where we

have written 2n to denote the product of n copies of the two point space.)
−1
Let A = πn] (E) : E ⊂ 2n , n ∈ N. Then A is a base for the topology of 2N ,
and all the members of A are open and compact: and A is an algebra of sets
which generates the Borel σ-algebra B2 . In particular the functions IA are
continuous for each A ∈ A. Hence, we may define

µ(A) = τ (IA )∀A ∈ A

The linearity of τ clearly implies that µ is a finitely additive set function

on A. Since members of A are compact and open, no element of A can be
expressed as a countable union of pairwise disjoint non-empty members of A.
In other words, we see that any finitely additive set function on A is auto-
matically countably additive, and hence it extends
RR to a probability measure
µ defined on all of BX . If we define τµ (f ) = dµ, then we see from equa-
tion (3.2.1) that τ and τµ agree on any continuous function f which factors
through some πn] ; since such functions are dense in C(2N ), we may conclude
that τ = τµ . In fact, equation (3.2.1) determines µ uniquely since a finite
measure on B is uniquely determined by its restriction to any ’algebra of sets’
which generates B as a σ-algebra. Thus the theorem is true for X = 2N .

Suppose now that X is a general compact metric space and that τ is

a positive linear functional on C(X). We may assume that kτ k = τ (1) =
1. Proposition 3.2.2 guarantees the existence of a continuous surjection p :
2N −→ X. Then it is easy to see that the map p∗ : C(X) −→ C(2N ) defined
by p∗ (f ) = f ◦ p is an isometric positivity preserving homomorphism of
algebras. In particular, we may regard C(X) as a subspace of C(2N ) via p∗ .
It follows from the Hahn-banach theorem that there exists a bounded linear
functional τ̃ on C(2N ) such that

τ̃ (p∗ (f )) = τ (f )∀f ∈ C(X)

and
kτ̃ k = kτ k

20
 = τ (1)
= τ̃ (p∗ (1))
|τ̃ k = τ̃ (1).
Now deducing Lemma 3.2.1 that τ̃ is a positive linear functional on C(2N ).
Concluding from the already proved special case of the theorem for 2N that
there exists Ra positive (in fact probability) measure ν defined on B2N such
that τ̃ (g) = 2N gdν. Hence, we see from the ’change of variable formula’ that

τ (f ) = τ̃ (p∗ (f ))
Z
= p∗ (f )dν
2N
Z
= f ◦ pdν
2N
Z
= f d(ν ◦ p−1 )
X
−1
hence with µ = ν ◦ p , we see that τ is indeed given by integration against
µ.
To complete the proof, we only need to establish uniqueness. For this,
we first assert that if τ is given by integration against µ, then

µ(K) = inf τ (f ) : IK ≤ f ∈ C(X)

for every compact K ⊂ X.

Since Z Z
IK ≤ f ⇒ µ(K) = IK dµ ≤ f dµ = τ (f )

it is clear that µ(K) is no greater than the infimum.

Conversely, since compact subsets of X are Gδ sets, we can find a de-

creasing sequence of open sets Un such that K = ∩Un . (For instance, we can
choose Un = x ∈ X : d(x, K) < n1 .) We can find continuous fn : X −→ [0, 1]
such that IK ≤ f ≤ IUn . Then it is clear that fn (x) −→ IK (x)∀x ∈ X; and
since 0 ≤ fn ≤ 1∀n, it follows that
Z Z
µ(K) = IK dµ = lim fn dµ

21
and in particular, µ(K) is no smaller than the infimum .

Finally, since any finite measure on a compact metric space is determined

by its values on compact sets, we see that τ indeed determines µ uniquely.

3.3 The General Compact Hausdorff Case:

Lemma 3.3.1:(a) The following conditions on a closed subset A of X

are equivalent

(i) A is a Gσ set i.e., there exists a sequence Un of open sets such that
A = ∩n Un ;

(ii) there exists a continuous function f : X −→ [0, 1] such that A =

−1
f (0).

Lemma 3.3.2: The following conditions on a subset E ⊂ X are equiva-

lent:

(i) A is a Baire set.

(ii) there exists a continuous function F : X −→ Y from X into a com-

pact metric space Y and a Baire-equivalently Borel-set E ∈ BY such that
A = F −1 (E).

proof:(ii) ⇒ (i) : Suppose F : X → Y is a continuous function from X

into a compact metric space Y . Let C = E ∈ BY : F −1 (E) is a Baire set.
First notice that if K is any closed set in Y , then K is also a Gδ set, and
consequently F−1 (K) is a compact Gδ set and hence a Baire set in X; hence
C contains all closed sets in Y . Since the definition shows that C is clearly a
σ-algebra, it follows that C = BY . (i) ⇒ (ii) : Let B denote the collection of
those subsets A ⊂ X which are inverse images, under continuous functions
into Y where Y = [0, 1]C for some countable set C - of Borel sets in Y. We
check now that B is closed under countable unions. Suppose An = Fn−1 (En )
for some En ∈ BYn , where Fn : X → Y n = [0, 1]Cn (with Cn some countable
set) is continuous; we may assume, without loss of generality, that the index

22
 Q disjoint. Then, let C = ∪n Cn , so we may identify
sets Cn are pairwise
C
Y = [0, 1] with n Yn . Writing πn for the natural projection mapping of Y
onto Yn . Defining F : X → Y by requiring that πn ◦ F = Fn , and defining
En0 = πn−1 (En ). The definitions show that F is continuous, En0 ∈ BY , so that
∪n En0 ∈ BY and

∪n An = ∪n F −1 (En0 ) = F −1 (∪n En0 );

so B is indeed closed under countable unions. Since it is trivially closed un-
der complementation, it follows that B is a σ-algebra of sets. So, in order to
complete the proof of the lemma, it remains only to prove that B contains
all compact Gδ sets; but this is guaranteed by Lemma 3.3.1.

Theorem 3.3.3: If τ : C(X) −→ C is a positive linear functional, then

there exists a unique finite positive µ defined on the Baire σ-algebra BX such
that Z
τ (f ) = f dµ.

proof:Let

C = (Y, π) : π is a continuous map of X onto a metric space Y.

For (Y, π), (Y1 , π1 ), (Y2 , π2 ) ∈ C, let us write

Bπ = π −1 (E) : E ∈ BY , and

(Y1 , π1 ) ≤ (Y2 , π2 ) ⇔ ∃ψ : Y2 −→ Y1 such thatπ1 = ψ ◦ π2 .

The proof involves a series of assertions.

(1) If (Y, π) ∈ C, there exists a unique measure µY defined on By such

that
Z
gdµY = τ (g ◦ π, ∀g ∈ C(Y ))
Y
(2) If (Y1 , π! ), (Y2 , π2 ), (Y2 , π2 ) ∈ C, and if (Y1 , π1 ) ≤ (Y2 , π2 ), so that
there exists a continuous map ψ : Y2 −→ Y1 such that π1 = ψ ◦ π2 then
µY1 = ψ(µY2 ) (meaning that µY1 (E) = µY2 (ψ −1 (E))), ∀E ∈ BY1 .

23
 (3) If (Yn , πn ) : n = 1, 2, · · · ⊂ C, then there exists (Y, pi) ∈ C such that
(Yn , πn ) ≤ (Y < π)∀n.

Finally, it is a direct consequence of Lemma 3.3.2 that BX = ∪(Y,π)inC Bπ .

The above assertions show that there exists a unique well-defined and count-
ably additive set-function µ on BX with the property that µ(π −1 (E)) =
µ
R Y (E) whenever R E ∈ BY , (Y, π) ∈ C. In other words, π∗ (µ) = µY so
X
g ◦ πdµ = y gdµY = τ (g ◦ π)∀g ∈ C(Y ). Now if f ∈ C(X), setting
Y = f (X), π = f , defining g ∈ C(Y ) by g(z) = z∀z ∈ Y , and deducing from
the previous sentence that indeed
Z
f dµ = τ f ,
X

as required.
As for uniqueness, it is seen, exactly as in the compact metric case, that
equation (3.2.4) is valid for every compact Gδ set K.

3.4 The Locally Compact Case:

Here let us assume that X0 is a locally compact Hausdorff space which can
be written as

Cc (X0 ) = f : X0 −→ C : f continuous, supp(f )compact,

where we write ’supp(f)’ to denote the support of f , i.e., the closure of

x ∈ X0 : f (x) 6= 0. As before, we shall let BX0 denote the σ-algebra gener-
ated by compact Gδ subsets of X0 .

Let us call a positive measure µ defined on BX0 inner regular if: (i)
µ(K) < ∞ for every compact Gδ subset K of X0 , and (ii) µ(E) = sup µ(K) :
K a compact Gδ subset of E.

Theorem 3.4.1: If τ : Cc (X0 ) −→ C is a linear functional which is

positive-meaning Cc (X0 ) 3 f ≥ 0 ⇒ τ (f ) > 0, then there exists a unique
positive inner regular measure µ defined on BX0 such that
Z
τ (f ) = f dµ∀f ∈ Cc (X0 ).

24
Proof:Let K be a compact Gδ subset of X 0 , and suppose K = ∩n Un ,, with
Un open. We may, and do, assume without loss of generality that Un has
compact closure and that Un+1 ⊂ Un . For each n, pick a continuous function
φn : X → [0, 1] such that 1Un+1 ≤ φn ≤ 1Un ,
The construction implies that
0 ≤ φn+1 = φn φn+1 ≤ φn
Assertion: Suppose now that f ∈ C(K) and f ≥ 0. Suppose f˜ ∈ Cc (X0 ) is
a non-negative extension of f - i.e., f˜|K = f . Then limn→∞ → τ (f˜φn ) exists
and this limit depends only on f and is independent of the choices of any of
Un , φn , f˜.

If g is another continuous non-negative function with compact support

which extends f , and if ∈> 0, then Un ∩ x ∈ X0 : |f˜(x) − g(x)| ≥∈ is seen
to be a decreasing sequence of compact sets whose intersection is empty. So
there exists an n such that
x ∈ Un ⇒ |f˜(x) − g(x) <∈ |,
it follows from the positivity of τ that
|τ (f˜φk ) − τ (gφk )| ≤∈ τ (φk )
for all large k. Since lim τ (φn ) exists (by virtue of the conclusion of the pre-
vious paragraph applied to f = 1k , with f˜ = 1U1 (say)) this establishes that
the limit of the Assertion is indeed independent of f˜.

Suppose (Vn , ψn )n is an alternative choice to (Un , φn )n in the sense that

(i) Vn is a sequence of open sets such that K = ∩n Vn ,

(ii) ψn ∈ Cc (X0 ), ψn : X0 −→ [0, 1]

Then, it may be seen, exactly as above, that if ∈> 0 is arbitrary, then

supx∈X0 |f˜(x)(φn (x) − ψn (x))| <∈ for large n, and hence deduced from posi-
tivity of τ that
lim τ (f˜φn ) = lim τ (f˜ψn ),
n→∞ n→∞

25
thereby completing the proof of the assertion.

It is clear that there exists a linear functional τK on C(K) such that

τK (f ) = lim τ (f˜φn )
n

for any non-negative f (and f˜ and φk s as above). Since τK is a positive

functional on C(K), we may deduce from Theorem 3.3.3 that there exists a
unique finite positive measure µK defined on BK such that
Z
τK (f ) = f dµK ∀f ∈ C(K).
K

Notice that the collection A of compact Gδ sets is a directed set with respect
to the ordering defined by

K ≤ L ⇔ K ⊂ Int(L)

where Int(L) denotes the interior of L.

We wish now to show that the family µK : K ∈ A is consistent in the

sense that
K, L ∈ A, K ≤ L ⇒ µL |K = µK .
First notice that it follows from the definitions that if g ∈ Cc (X0 ) and if
supp(g) ≤ L, then Z
gdµL = τ (g).
L
Suppose now that K, L ∈ A and K ≤ L. We may find a sequence Un , φn as
above, such that K = ∩n Un and U1 ⊂ L. Now, if f ∈ C(K) is arbitrary, and
if f˜ is any extension of f to a compactly supported continuous function with
supp(f˜) ≤ L, and the dominated convergence, for instance, that
Z
f dµK = lim τ (f˜φn )
K n
Z
= lim (f˜φn )dµL
n L
Z
= 1K f dµL

26
 Z
= f d(µL )|K.
K
Finally, for any set E ∈ BX0 , notice that E ∩ K ∈ BK ∀K ∈ A and
that µK (E ∩ K) : K ∈ A is a non-decreasing net of real numbers which must
converge to a number - call it µ(E) - in [0, ∞]. Note that

µ(K) = sup µK (E ∩ K)∀E ∈ BK

K∈A

is the measure on BX0 whose existence is asserted

We wish to say that this µ `
by theorem. Suppose E = ∞ n=1 En where E, En ∈ BX0 . Then

µ(E) = sup µK (E ∩ K)
k∈A

∞
a
= sup ( En ∩ K)
K∈A
n=1
∞
X
= sup µK (En ∩ K)
K∈A n=1

N
X
= sup sup µK (En ∩ K)
K∈A N n=1
N
X
= sup sup µK (En ∩ K)
N K∈A n=1

N
X
= sup lim µK (En ∩ K)
N K∈A n=1

N
X
= sup lim µK (En ∩ K)
N K∈A
n=1
N
X
= sup µ(En )
N
n=1
∞
X
= µ(En ).
n=1

27
and hence µ is indeed countably additive, i.e., a measure defined on BX0 .

Now, if f ∈ Cc (X0 ), if L ∈ A satisfies supp(f ) ≤ L, then

Z Z
τ (f ) = f dµL = f dµ.
L X0

Finally, ifK is any compact Gδ , then it is clear that

µ(K) = µK (K) < ∞.

Let us defer, for a moment, the proof of inner regularity of our µ, and verify,
instead, that there is at most one inner regular measure µ related to a τ as
in the theorem. By inner regularity, it suffices to observe that

µ(K) = inf τ (f ) : 1K ≤ Cc (X0 )

for all compact Gδ subsets K.

Finally,if E ∈ BX0 is arbitrary, we have by definition of our µ and corollary

(3.4.3).
µ(E) = sup µ(E ∩ K)
K∈A

= sup sup µ(C)

K∈A,C⊂E∩K, C compact Gδ

so indeed
µ(E) = sup µ(C).
C⊂E,C compact Gδ

The proof of the theorem is complete, modulo that of the following proposi-
tion and its corollary.

Proposition 3.4.2: Suppose µ is a finite positive measure defined on the

Borel σ-algebra BX of a compact metric space. Then, for any Borel set
E ∈ BX , we have:

µ(E) = supµ(K) : K ⊂ E andK is compact

= inf µ(U ) : E ⊂ U and U is open .

Corollary 3.4.3: Any finite positive measure defined on the Baire σ-
algebra BX of a compact Hausdorff space X is inner regular. proof: By

28
Lemma 3.3.2, if A is any Baire subset of X, then there exists a compact
matric space Y , a Borel subset E ⊂ Y and a continuous map f : X → Y
such that A = f −1 (E). Applying Proposition 3.4.2 to the measure ν = µ◦f −1
defined on BY , to find compact sets Cn ⊂ E such that ν(E) = supn ν(Cn ).
Noticing that Kn = f −1 (Cn ) is a closed, hence compact. Clearly Gδ subset
of X such that Kn ⊂ A and concluding that

µ(A) = ν(E)

= sup ν(Cn )
n

= sup µ(Kn ).
n

29
 Chapter 4

LINEAR OPERATORS

4.1 Continuous Linear Transformation:

Since we’ve covered some properties of normed spaces, we can now talk
about functions mapping from one normed space into another. The sim-
plest to work with of these maps will be linear (recall the definition of a
linear transformation: a function T : V → W such that for all α, β ∈ F and
x, y ∈ V , T(αx + βy) =αT (x) + βT (y)), and the most important will be con-
tinuous. We will give examples, as well as look at the space of all continuous
linear transformations. First however, we will give alternate characteriza-
tions of continuity for linear transformations.

Theorem 4.1.1: Let X and Y be normed linear spaces and T : X → Y

be a linear transformation. Then the following are equivalent:

(i) T is uniformly continuous;

(ii) T is continuous;

(iii) T is continuous at 0;

(iv) there exists a real number k > 0 such that kT (x)k ≤ k whenever x
∈ X and kxk ≤ 1;

(v) there exists a real number k > 0 such that kT (x)k ≤ kxk for all x ∈ X.

Example 4.1.2: Define T : C[0,1] → F by T(f) = f(0). Then if f ∈ C[0,1],

we get
kT (f )k = |f (0)| ≤ sup{|f (x)| : x ∈ [0, 1]} = kf k :

30
So using part (v) of Theorem 4.1.1 with k = 1, we see that T is continu-
ous, and also that this was a much easier way of proving continuity than the
epsilon-delta methods that we are used to.

Theorem 4.1.3: Let X be a denite-dimensional normed space, Y be any

normed linear space, and T : X → Y be a linear transformation. Then T is
continuous.

Definition 4.1.4: Let X and Y be normed linear spaces and T : X →

Y be a linear transformation. T is bounded if there exists a real number k
> 0 such that kT (x)k ≤ kkxk for all x ∈ X.

Notice that this definition is part (v) of Theorem 4.1.1, hence we can
use the words continuous and bounded interchangeably when talking about
linear transformations. However, this use of the word bounded is different
than our conventional definition for functions from R → R. For example,
the linear transformation T : R → R defined by T(x) = x is bounded by
Definition 4.1.4, but obviously not bounded in the usual sense.

Definition 4.1.5: Let X and Y be normed linear spaces. Then the set
of all continuous linear transformations from X to Y is denoted B(X; Y ).
Elements of this set are called bounded linear operators (or sometimes just
operators).

Lemma 4.1.6: Let X and Y be normed linear spaces, and S,T ∈ B(X,Y)
with kS(x)k ≤ k1 kxk and kT (x)k ≤ k2 kxk for all x ∈ X and some real num-
bers k1 , k2 > 0. Let λ ∈ F. Then
(i) k(S + T )(x)k ≤ (k1 + k2 )kxk for all x ∈ X;
(ii) k(λS)(x)k ≤ |λ|k1 kxk for all x ∈ X;
(iii) B(X,Y) is a linear subspace of L(X,Y), and so B(X,Y) is a vector space.

Proof.
(i) For x ∈ X,
k(S + T )(x)k = kS(x) + T (x)k ≤ kS(x)k + kT (x)k ≤ k1 kxk + k2 kxk =
(k1 + k2 )kxk:
(ii) For x ∈ X,
k(λS)(x)k = |λ|kS(x)k ≤ |λ|k1 kxk

31
: (iii) First, B(X; Y ) is not empty because we can always define We see that
(S + T) and (λS) are bounded by Theorem 4.1.1 (v), and so are elements
of B(X,Y). Thus parts (i) and (ii) satisfy the subspace test, so B(X,Y) is a
linear subspace of L(X,Y) (and hence is a vector space).

For the next definition, recall that if X and Y are normed spaces then the
Cartesian product X × Y is also a normed space.

Definition 4.1.7: If X and Y are normed spaces and T : X → Y is a

linear transformation, the graph of T is the linear subspace of X × Y defined
by
G(T ) = {(x, Tx ) : x ∈ X} :
4.2 The Space B(X,Y):

We just saw that B(X,Y) is a vector space, so we will now look to define
a norm. It is important to note that proofs here may have three different
norms from different spaces in the same equation. However, we will use the
symbol k.k as usual for all three norms, as it will be easy to determine which
norm we are referring to based on the space that the element is from. We
should also note that for a normed linear space X, the set of bounded linear
operators from X to X will simply be denoted B(X).

Lemma 4.2.1: Let X and Y be normed spaces and define k.k : B(X,Y)
→ R by kT k = sup{kT (x)k : kxk ≤ 1}. Then k.k is a norm on B(X,Y), so
B(X,Y) is a normed space.

Proof. Before checking the axioms of a norm for k.k, we must note the
following consequence of Theorem 4.1.1:

sup{kT (x)k : kkk ≤ 1} = inf {k : kT (x)k ≤ kkxk∀x ∈ X};

and so kT (y)k ≤ sup{kT (x)k : kxk ≤ 1}kyk∀y ∈ X :

Now we can proceed checking the axioms of a norm:

(i) Obviously, kT k = sup{kT (x)k : kxk ≤ 1} ≥ 0 for all T ∈ B(X,Y).

32
 (ii) Note that the zero transformation is defined by T(x) = 0 for all x ∈
X. Then
kT k = 0 ⇐⇒ kT xk = 0∀x ∈ X
⇐⇒ T x = 0∀x ∈ X
⇐⇒ T is the zero transf ormation :
(iii) Since kT (x)k ≤ kT kkxk, we know that k(λT )(x)k ≤ |λ|kT kkxk for all
x ∈ X by Lemma 4.1.6 (ii), hence kλT k ≤ |λ|kT k. If λ = 0, then clearly
kαT k = |λ|kT k and we are done. So assume λ 6= 0. Then

kT k = kλ−1 λT k ≤ |λ−1 |kλT k ≤ |λ|−1 |λ|kT k = kT k :

So we get kT k = |λ|−1 kλT k, and multiplying both sides by |λ|−1 gives

kλT k = |λ|kT k :

(iv) By the construction of kT k, we see that the triangle inequality holds as

follows:
k(S + T )(x)k ≤ kS(x)kkT (x)k
≤ kSkkxk + kT kkxk
= (kSk + kT k)kxk :
So kS + T k ≤ kSk + kT k: Therefore k.k is a norm, known as the norm of T.

Lemma 4.2.2: Let X be a normed linear space with W a dense sub-

space of X, and let Y be a Banach Space with S ∈ B(X,Y). If x ∈ X and
{xn }, {yn } are sequences in W such that limn→∞ xn = limn→∞ yn = x then
{S(xn )}and{S(yn )} both converge, and limn→∞ S(xn ) = limn→∞ S(yn ) = x

Proof. Since xn → x, {xn } is a Cauchy sequence, and because we have

kS(xn ) − S(xm )k = kS(xn − xm )k ≤ kSkkxn − xm k;

we see that {S(xn )} is also a Cauchy sequence. Now since Y is a Banach

space, it is complete, hence {S(xn )} converges. Furthermore, because of the
condition limn→∞ xn = limn→∞ yn = x we must have limn→∞ (xn − yn ) = 0
Now because

kS(xn ) − S(yn )k = kS(xn − xm )k ≤ kSkkxn − xm k;

33
we get limx→∞ (S(xn ) − S(yn )) = 0, hence limn→∞ S(xn ) = limn→∞ S(yn ).

In the next theorem, we explore what conditions are required for B(X,Y)
to be a Banach Space.

Theorem 4.2.3: If X is a normed linear space and Y is a Banach space,

then B(X,Y) is also a Banach Space.

Definition 4.2.4: Let X, Y , and Z be normed linear spaces, T ∈ B(X,Y )

and S ∈ B(Y,Z). Then the composition of S and T, called the product of S
and T, will be denoted by ST.

Given the conditions from the definition, if X, Y , and Z are not all the
same, the fact that we can de
ne ST does not necessarily mean that we can define TS. However, if X
= Y = Z, both products will be defined, but generally TS 6= ST. Another
notation point to make is that if X is a normed space and T ∈ B(X), the
product TT will be denoted T 2 , and the product of T with itself n times will
be denoted T n .

4.3 Isometries, Isomorphisms, and Inverses:

In this section we discuss different classifications that may be given to an

operator.

Definition 4.3.1: Let X and Y be normed linear spaces and T ∈ L(X,Y).

If kT (x)k = kxk for all x ∈ X, then T is called an isometry.

One can see it is easy to find the norm of this type of operator. Note that
on every normed space, there is at least one isometry, as we see next.

Example 4.3.2: If X is a normed space and I is the identity linear trans-

formation on X, then I is an isometry since for all x ∈ X, I(x) = x. Hence
kI(x)k = kx
.

Definition 4.3.3: If X and Y are normed linear spaces and T is an isometry

from X onto Y , then T is called an isometric isomorphism and X and Y are

34
isometrically isomorphic.

If two spaces are isometrically isomorphic, we can think of them as hav-

ing essentially the same structure concerning vector space and norm prop-
erties. Next we will try and generalize the idea of an inverse to an infinite-
dimensional setting.

Definition 4.3.4: Let X and Y be normed linear spaces. An operator

T ∈ B(X,Y) is invertible if there exists an S ∈ B(Y,X) such that ST = IX
and TS = IY . We say that S is the inverse of T and denote it by T −1 .

Definition 4.3.5: Let X and Y be normed linear spaces. If there exists

an invertible operator T ∈ B(X,Y), then X and Y are isomorphic, and we
say T is an isomorphism.

It is apparent that if T ∈ B(X,Y) is an isomorphism, then T −1 ∈ B(X,Y)

is also an isomorphism. Also, spaces that are isomorphic are similar in struc-
ture, as we see next.

Lemma 4.3.6: If two normed linear spaces X and Y are isomorphic, then:

(i) dimX < ∞ ⇐⇒ dimY < ∞ (in which case, dimX = dimY );

(ii) X is separable ⇐⇒ Y is separable;

(iii) X is complete ⇐⇒ Y is complete.

Lemma 4.3.7: If X and Y are normed linear spaces and T ∈ B(X,Y) is

invertible, then for all x ∈ X,

kTx k ≥ kT −1 k−1 kxk :

Proof. For all x ∈ X we have kxk = kT −1 (Tx )k ≤ kT −1 kkTx k, hence multi-

plying both sides by the inverse of T −1 k gives kTx k ≥ kT −1 k−1 kxk as desired.

35