Chapter 1

Some Basic Background

In this chapter we want to recall some important basic results from Functional Analysis most of which were already covered in the Real Analysis
course Math607/608 and can be found in the textbooks [Fol] and [Roy].

1.1

Normed Linear Spaces, Banach Spaces

All our vectors spaces will be vector spaces over the real field R or the
complex field C. In the case that the field is undetermined we denote it by
K.
Definition 1.1.1. [Normed linear spaces]
Let X be a vector space over K, with K = R or K = C. A semi norm on
X is a function k k : X ! [0, 1) satisfying the following properties for all
x, y 2 X and 2 K
1. kx + yk kxk + kyk (triangle inequality) and
2. k xk = | | kxk (homogeneity),
and we call a semi norm k k a norm if it also satisfies
3. kxk = 0 () x = 0, for all x 2 X.
In that case we call (X, k k), or simply X, a normed space. Sometimes we
might denote the norm on X by k kX to distinguish it from some other
norm k kY defined on some other space Y .
For a normed space (X, k k) the sets
BX = {x 2 X : kxk 1} and SX = {x 2 X : kxk = 1}
are called the unit ball and the unit sphere of X, respectively.
5

CHAPTER 1. SOME BASIC BACKGROUND

Note that a norm k k on a vector space defines a metric d(, ) by
d(x, y) = kx

yk,

x, y 2 X,

and this metric defines a topology on X, also called the strong topology.
Definition 1.1.2. [Banach Spaces]
A normed space which is complete, i.e. in which every Cauchy sequence
converges, is called a Banach space.
To verify that a certain norm defines a complete space it is enough, and
sometimes easier to verify that absolutely converging series are converging:
Proposition 1.1.3. Assume that
space so that for all
P X is a normed linearP
sequencesP(xn ) X for which
kxkn < 1, the series
xn converges (i.e.
limn!1 nj=1 xj exists in X).
Then X is complete.
Proposition 1.1.4. A subspace of a Banach space is a Banach space if and
only if it is closed.
Proposition 1.1.5. [Completion of normed spaces]
so that:
If X is a normed space, then there is a Banach space X
meaning that I :
There is an isometric embedding I from X into X,

X ! X is linear and kI(x)k = kxk, for x 2 X, so that the image of X under

I is dense in X.
is unique up to isometries, meaning that whenever Y is
Moreover X
a Banach space for which there is an isometric embedding J : X ! Y ,
! Y (i.e. a linear
with dense image, then there is an isometry J : X

so that
bijection between X and Y for which kJ(
x)k = k
xk for all x
2 X),
J I(x) = J(x) for all x 2 X.
is called the completion of X.
The space X
Let us recall some examples of Banach spaces.
Examples 1.1.6. Let (, , ) be a measure space, and let 1 p < 1,
then put
Z
n
o
Lp () := f : ! K mble :
|f |p d(x) < 1 .

For p = 1 we put

L1 () := f : ! K mble : 9 C ({! 2 : |f (!)| > C}) = 0 .

1.1. NORMED LINEAR SPACES, BANACH SPACES

Then Lp () is a vector space, and the map
k kp : Lp () ! R,

f 7!

|f (!)| d(!)

1/p

if 1 p < 1, and
k k1 : L1 () ! R,

f 7! inf{C > 0 : ({! 2 : |f (!)| > C}) = 0},

if p = 1, is a seminorm on Lp ().
For f, g 2 Lp () define the equivalence relation by
f g : () f (!) = g(!) for -almost all ! 2 .
Define Lp () to be the quotient space Lp ()/ . Then k kp is well defined and a norm on Lp (), and turns Lp () into a Banach space. Although,
strictly speaking, elements of Lp () are not functions but equivalence classes
of functions, we treat the elements of Lp () as functions, by picking a representative out of each equivalence class. Equality then means almost
everywhere equality.
If A R, or A Rd , d 2 N, and is the Lebesgue measure on A we
write Lp (A) instead of Lp (). If is a set and is the counting measure on
we write `p ( ) instead of Lp (). Thus
n
`p ( ) = x() :

! K : kxkp =

X
2

|x |p

1/p

o
< 1 , if 1 p < 1,

and
`1 ( ) = x() :

! K : kxk1 = sup |x | < 1 .

2

If = N we write `p instead of `p (N) and if = {1, 2 . . . n}, for some n 2 N

we write `np instead of `p ({1, 2 . . . n}).
The set
c0 = {(xn : n 2 N) K : lim xn = 0}
n!1

is a linear closed subspace of `1 , and, thus, it is also a Banach space (with

k k1 ).
More generally, let S be a (topological) Hausdor space, then
Cb (S) = {f : S ! K continuous and bounded}

CHAPTER 1. SOME BASIC BACKGROUND

is a closed subspace of `1 (S), and, thus, Cb (S) is a Banach space. If K is a

compact space we will write C(K) instead of Cb (K). If S is locally compact
then
C0 (S) = f : S ! K continuous and {|f |

c} is compact for all c > 0

is a closed subspace of Cb (S), and, thus, it is a Banach space.

Let (, ) be a measurable space and assume first that K = R. Recall
that a finite signed measure on (, ) is a map : ! R so that (;) = 0,
and so that
(

1
[

n=1

En ) =

1
X

n=1

(En ), whenever (En ) is pairwise disjoint.

The Jordan Decomposition Theorem says that such a signed measure can
be uniquely written as the dierence of two positive finite measure + and
.
If we let
kkv = + () + () =

sup

(A)

(B),

A,B2,disjoint

then k kv is a norm, the variation norm, on

M () = MR () := { : ! R : signed measure},
which turns M () into a real Banach space.
If K = C, we define
M () = MC (M ) = + i : , 2 MR () ,
and define for + i 2 MC ()
k + ikv =

kk2v + kk2v .

Then MC () is a complex Banach space.

Assume S is a topological space and BS is the sigma-algebra of Borel sets,
i.e. the -algebra generated by the open subsets of S. We call a (positive)
measure on BS a Radon measure if
1) (A) = inf{(U ) : U S open and A U } for all A 2 BS , (outer
regularity)
2) (U ) = sup{(C) : C S compact and C U } for all U S, (inner
regularity on open sets) and

1.1. NORMED LINEAR SPACES, BANACH SPACES

3) it is finite on all compact subsets of S.

If K = R a signed Radon measure is the dierence of two finite positive
Radon measure, and, as before, if K = C then + i, where and ]nu are
two real valued Radon measures, is a signed Radon measure
We denote the set of all signed Radon measures by M (S). Then M (S)
is a closed linear subspace of M (BS ).
It can be shown (cf. [Fol, Proposition 7.5]) that a -finite Radon measure
is inner regular on all Borel sets.
Proposition 1.1.7. [Fol, Theorem 7.8]
Let X be a locally compact space for which all open subsets are -compact
(i.e. a countable union of compact sets). Then every Borel measure which
is bounded on compact sets is a Radon Measure.
There are many ways to combine Banach spaces to new spaces.
Proposition 1.1.8. [Complemented sums of Banach spaces]
If Xi is a Banach space for all i 2 I, I some index set, and 1 p 1,
we let
n
o
X
:=
(x
)
:
x
2
X
,
for
i
2
I,
and
(kx
k
:
i
2
I)
2
`
(I)
.
i `p
i i2I
i
i
i
p
i2I
We put for x 2

i2I

Xi

`p

kxkp := (kxi k : i 2 I)

Then k k is a norm on
We call
i2I Xi
Moreover,
i2I Xi c0

`p

i2I

8
< P

p
i2I kxi kXi

: sup kx k
i Xi
i2I

Xi

and

`p

1/p

i2I

Xi

if 1 p < 1,
if p = 1.
`p

is a Banach space.

the `p sum of the Xi , i 2 I.

:= (xi )i2I 2

i2I Xi `1

is a closed linear subspace of

i2I

: 8c > 0

Xi

`1

{i 2 I : kxi k

c} is finite

, and, thus also a Banach space.

If all the spaces Xi are the same spaces in Proposition 1.1.8, say Xi = X,
for i 2 I we write `p (I,X), and c0 (I,X), instead of
i2I Xi `p or
i2I Xi c0 ,
respectively. We write `p (X), and c0 (X) instead of `p (N, X) and c0 (N, X),
respectively, and `np (X), instead of `p ({1, 2 . . . n}, X), for n 2 N.

10

CHAPTER 1. SOME BASIC BACKGROUND

Note that if I is finite then for any norm k k on RI , the norm topology
on ( Xi )kk does not depend on k k. By i2I Xi we mean therefore the
norm product space, which is, up to isomorphy unique, for example in this
case ( Xi )`1 ( Xi )`1 .
If X and Y are Banach space we often call the product space X Y also
X Y.

Exercises:
1.

Prove Proposition 1.1.4.

2.

Prove Proposition 1.1.5.

Hint: let
= (xn ) : (xn ) is a Cauchy sequence in X}/ ,
X
where (xn ) (yn ) : () limn!1 kxn

appropriate norm on X.

yn k = 0. The define the

3.

Let S be a topological space. Find a condition on a Borel measure

on S, which ensures that Cb (S) is a closed subspace of L1 ().

4.

Let 1 p 1, and let Xi be a Banach space for each i 2 I, where I

is some index set. Show that the norm k kp introduced in Proposition
1.1.8 on
i2I Xi `p turns this space into a complete normed space
(only show completeness).

5. Let f 2 Lp [0, 1] for some p > 1. Show that

lim kf kr = kf k1 .

r!1+

1.2. OPERATORS ON BANACH SPACES, DUAL SPACES

1.2

11

Operators on Banach Spaces, Dual Spaces

If X and Y are two normed linear spaces, then for a linear map (we also say
linear operator) T : X ! Y the following are equivalent:
a) T is continuous,
b) T is continuous at 0,
c) T is bounded , i.e. kT k = supx2BX kT (x)k < 1.
In this case k k is a norm on
L(X, Y ) = {T : X ! Y linear and bounded}
which turns L(X, Y ) into a Banach space if Y is a Banach space, and we
observe that
kT (x)k kT k kxk for all T 2 L(X, Y ) and x 2 X.
We call a bounded linear operator T : X ! Y an isomorphic embedding if
there is a number c > 0, so that ckxk kT (x)k. This is equivalent to saying
that the image T (X) of T is a closed subspace of Y and T has an inverse
T 1 : T (X) ! Y which is also bounded.
An isomorphic embedding which is onto (we say also surjective) is called
an isomorphy between X and Y . If kT (x)k = kxk for all x 2 X we call T
an isometric embedding, and call it an isometry between X and Y if T is
surjective.
If there is an isometry between two spaces X and Y we write X ' Y .
In that case X and Y can be identified for our purposes. If there is an
isomorphism T : X ! Y with kT k kT 1 k c, for some number c 1 we
write X c Y and we write X Y if there is a c 1 so that X c Y .
If X and Y are two Banach spaces which are isomorphic (for example if
both spaces are finite dimensional and have the same dimension), we define
dBM (X, Y ) = inf{kT k kT

k : T : X ! Y, T isomorphism},

and call it the Banach Mazur distance between X and Y . Note that always
dBM (X, Y ) 1.
Remark. If (X, k k) is a finite dimensional Banach space over K, K = R
or K = C, and its dimension is n 2 N we can after passing to an isometric

12

CHAPTER 1. SOME BASIC BACKGROUND

image, assume that X = Kn . Indeed, let x1 , x2 , . . . xn be a basis of X, and

consider on Kn the norm given by:
k(a1 , a2 , . . . , an )kX =

n
X
j=1

aj xj , for (a1 , a2 , . . . an ) 2 Kn .

Then
I : Kn ! X,

(a1 , a2 , . . . , an ) 7!

n
X

aj xj ,

j=1

is an isometry. Therefore we can always assume that X = (Kn , k kX ). This

means BX is a closed and bounded subset of Kn , which by the Theorem
of Bolzano-Weierstra, means that BX is compact. In Theorem 1.5.4 we
will deduce the converse and prove that a Banach space X, for which BX is
compact, must be finite dimensional.
Definition 1.2.1. [Dual space of X]
If Y = K and X is a normed linear space over K, then we call L(X, K) the
dual space of X and denote it by X .
If x 2 X we often use h, i to denote the action of x on X, i.e. we
write hx , xi instead of x (x).
Theorem 1.2.2. [Representation of some Dual spaces]
1. Assume that 1 p < 1 and 1 < q 1 with p1 + 1q = 1, and assume
that (, , ) is a measure space without atoms of infinite measure.
Then the following map is a well defined isometry between Lp () and
Lq ().
: Lq () !

Lp (),

h (g), f i =

(g)(f ) :=

f ()g() d(),

for g 2 Lq (), and f 2 Lp ().

2. Assume that S is a locally compact Hausdor space, then the map
Z
: M (S) ! C0 (S), h (), f i = ()(f ) :=
f () d()
S

for 2 M (S) and f 2 C0 (S),

is an isometry between M (S) and C0 (S).

1.2. OPERATORS ON BANACH SPACES, DUAL SPACES

13

Remark. If p = 1 and q = 1 then the map

in Theorem 1.2.2 part
(1) is still an isometric embedding, but in general (i.e. if Lp () is infinite
dimensional) not onto.
Example 1.2.3. c0 ' `1 (by Theorem 1.2.2 part (2)) and `1 ' `1 (by
Theorem 1.2.2 part (1)) .
Exercises
1.

and Y be completions
Let X and Y be normed linear spaces, and X
of X and Y , respectively (recall Proposition 1.1.5). Then every T 2
Y ),
L(X, Y ) can be extended in a unique way to an element T in L(X,
and kTk = kT k.

2.

A Banach space X is called strictly convex if for any x, y 2 SX , x 6= y

kx + yk < 2.

Prove that c0 and `1 are not strictly convex, but that they can be
given equivalent norms with which they are strictly convex.
Recall that two norms k k and ||| ||| on the some linear space X are
equivalent if the identity I : (X, k k) ! (X, ||| |||) is an isomorphism.
3. A Banach space X is called uniform convex if for for every " > 0 there
is a so that:
If x, y 2 SX with kx

yk > " then kx + yk < 2

Prove that `p , 1 < p < 1 are uniform convex but `1 and c0 do not
have this property.