2.3.

Strict Convexity and Uniform Convexity

of Spaces
2.3.1 Convex Set
Definition 2.20. A subset of a vector space is said to be convex if
implies that the set

is a subset of C, i.e. any line segment joining any two points of contained
in .
is called a closed segment, and are called the boundary points of the
segment and any other points of are called an interior points of .

Example 12.

a. Empty set and Singleton set are convex sets.

b. Lines, segments and disks are all convex sets.
c. Open and closed balls of a normed space is convex set.
. is not convex set. B/c the elements on a line segment between any two
points of does not lie in .
. The intersection of any convex sets is convex, but the union of convex
sets may not convex.

Verification:

Let 's are convex sets, for , then take , then

.
Now since all are convex,
and .

33
 ⋂

⋂ ⋂

Therefore the intersection of any convex set is convex.

To clarify; let set and are both convex sets, then
is convex set but is not convex set, b/c the elements on a
line segment between 0 and 1 is not belongs to .

2.3.2 Inner product space and Hilbert space

Definition 2.21. Let be a linear space over a field . An inner product on

is a scalar-valued function such that for all and

for all , we have

1. and if and only if ;

2. (The bar denotes complex conjugation.);

3. ;

4. .

The pair is called an inner product space. An inner product space is

also called pre-Hilbert space. If a space is complete then the inner product
space becomes complete inner product space.
An inner product space on defines a norm on given by
34
 √

Definition 2.22. (Hilbert space)

A vector space H is called a Hilbert space if:

i) H is an inner product space.

ii) H is complete with respect to the norm induced by the inner
product, i.e. every Cauchy sequence in H satisfies:
as

Simply Hilbert space is a complete inner product space.

Let be an inner product space, then the following properties hold.

1. [Cauchy-Schewarz inequality];

Note: If and are linearly dependent, i.e. for some , then

2. [Parallelogram
identity];

Note: If ( x and are orthogonal i.e. ) then we have

Pythagoras' theorem

Remark 2.2. For fixed the map defined by is a

linear functional with norm

‖ ‖

35
2.3.3 Best Approximation
Definition 2.23. Let be a normed linear space and let U be a subspace of
X or . Define the distance from to U by

i.e. , is any point .

Remark 2.3. If , then .

Definition 2.24. The set of all best approximations to from is denoted
by , and is metric projection or nearest point map
defined by

The set is called

i). Proximinal if each has a best approximation in U; i.e.,

;
ii). Chebyshev if each has a unique best approximation in .

Theorem 2.7. Let be an inner product space and be a non-empty

complete convex subset of . Then is a Chebyshev set.i.e.
.

Proof. By definition of inf such that

Claim 1. is a Cauchy sequence.

Now let and

36
 ‖ ‖

And

Then

Therefore, is Cauchy sequence in M.

Since is complete, the sequence is converges to a point in M.
i.e. such that

But

Therefore, the set which is nearest point from x .

Claim 2: This nearest point of is unique.
Suppose such that . Then

But (parallelogram identity), then

37
Now,

‖ ( )‖

‖ ‖

Therefore, the set M is Chebyshev set.

Lemma 2.2. In a normed space a set of best approximations to a given
point out of a subspace of is convex.

Proof. Let be the distance between to ,

Now, if or has a single element then is convex set.
If has more than one element, then let's take
.
Then let ,
Since is minimum distsnce from to .

38
Then from equation (21) and (22) we have
. Thus is convex.

Note: The set of best approximations is convex if is convex.

Definition 2.25. A subset of a normed space is said to be;
a). Uniformly convex: if such that with
and , we have

b). Strictly convex: if and only if with , we have

,

Or
with , we have for all
,
Or
with , we have for some constant .

Examples 12.Every Hilbert Space is Strictly Convex

Proof . Let x,y and . Let .

By the parallelogram equality we have

= 2( < 4

Hence

39
Example 13 . The space [ ] is not strictly convex.

Proof. Consider and defined as

[ ]

[ ]

Then [ ], and . We also have

However, ‖ + ‖ = ,t [ ]

This shows that [ ] is not strictly convex.

Example 14. Every Inner product space is uniformly convex.

Proof. Let , such that and such that
. Then we need to show .
Now by parallelogram law,

40
Thus √ such that

. is uniformly convex.

Note: Since every Hilbert space is Inner product space, so Hilbert space is
uniform convex space.

Example 15. spaces and are not uniformly convex.

Proof. For a real number , the - norm or -norm of is defined by

The -norm or maximum norm (or uniform norm) is the limit of the -
norms for . It turns out that this limit is equivalent to the following
definition:

Now take . Then

However, and there is no such that

. Thus, is not uniformly convex.
Similarly, if we take and
, then

Because is not uniformly convex since there is no

such that .

41
Remark 2.4. space is uniformly convex for .
Example 16. Every Inner product space is strictly convex.
Proof. Now, suppose a norm on a linear space is induced by an inner
product. Let and , then

Thus is strictly convex.

Note: [ and [ ] space are not strictly convex.
Lemma 2.3. Let be a subset of a normed space . If is uniformly
convex space then it is strictly convex.

Proof. Let is uniformly convex space,

s.t and s.t then
.

Now for s.t s.t

( )

So, every uniformly convex space is strictly convex. But the converse is not
hold.
Remark 2.5. If is finite dimensional, then is strictly convex if and only
if is uniformly convex.

42
Example 17. A norm of a Banach space is strictly convex.
To show this: A norm of a Banach space is always convex, i.e.,

[ ]

A number of Banach spaces do not have equality when , i.e.,

We use to denote the unit sphere on Banach

space . If with , then

which says that the unit sphere contains no line segments. This suggests
strict convexity of norm.

2.3.4 Convex Function

Definition 2.26. Let be a linear space and and be a function.

Then is called Convex if and only if for all and

all s.t :

And we obtained by replacing with the strict inequality <.

Explicitly, the map is called strictly convex if and only if for all real
and all such that x :

More over a function is strongly convex, if such that

43
 is convex.
If is twice continuously differentiable and the domain is the real line, then:

1. is convex if and only if for all .

2. Strictly convex if for all (note: this is sufficient, but

not necessary).

3. Strongly convex if and only if for all .

Example 18. i). The function is convex, strictly convex and

strongly convex over . , so is a convex function. It is
also strongly convex with strong convexity constant 2 and hence strictly
convex too.
ii). The function is strictly convex over . To show this,

And also for a twice differentiable function implies is

strictly convex.. (It is sufficient condition ). Here, . So,
for all . So, is strictly convex in .

Now suppose , . Then

44
So, is strictly convex over .
Remark 2.6. Strong convexity Strict convexity Convexity. But the
converse of neither implication is true.

2.3.5 Modulus of Convexity

Let be a Banach space. Then a function [ ] [ ] is called;
Modulus of convexity of if ], then

{ }

Note: From the above definition and parallelogram identity of norm we have
and , for all . Therefore, , for all

Proposition 2.3. A Banach space is uniformly convex if and only if

for all ].

Proof. Let be a uniformly convex Banach space. Then for , there

exists such that

‖ ‖

for all with and .

Therefore, from the definition of modulus of convexity and equation (23),
we have

{ ‖ ‖ }

45
Conversely, suppose is a Banach space with modulus of convexity
such that 0 for all ]. Let such that
with for fixed ]. By the modulus of convexity ,
we have

‖ ‖

which implies that

‖ ‖

where , which is independent of and . Therefore, is

uniformly convex.
Remark 2.7. A normed space is Uniformly convex: if ,

for all ].

2.4 Smoothness and uniform smoothness of spaces

Definition 2.27. A Banach space is said to be;
I). Smooth: if there exist unique functional such that

II). Uniformly smooth: if such that with

and then

46
Lemma 2.4. Let be a Banach space and be the dual of . Then

1. If is strictly convex, then is smooth.

2. If is smooth, then is strictly convex.

Proof. (1) Suppose is not smooth. There exist and with

such that

Now by triangular inequality, and

, then (i.e. ). Hence is not strictly
convex.
(2) Suppose is not strictly convex. There exist and with
such that .
Thus, two supporting hyperplanes pass through such that

Therefore, is not smooth.

Corollary 2.1. Therefore from the lemma there are strictly convex spaces
whose duals fail to be smooth. If is reflexive, then is strictly convex if
and only if is smooth and is smooth if and only if is strictly convex.

Example 19. A Hilbert space and space are reflexive

spaces, so, they satisfies the above corollary.

Remark 2.8. Every uniformly convex and every uniformly smooth Banach
space is reflexive.

47
2.4.1 Modulus of Smoothness
Definition 2.28. Let be a Banach space. Then a function is
said to be; Modulus of smoothness of if

{ }

{ }

Note: and for any .

Theorem 2.8. A normed space is uniformly smooth if and only if

Proof. If uniformly smooth and then, there exists such that

for every such that . This implies

for every . Since is arbitrary,

Conversely, let , and suppose there exists such that ,

for every . Let . Then with we have

and the space is uniformly smooth.

48
Proposition 2.4. Every uniformly smooth Banach space is smooth.
Proof. By contradiction, Suppose, that is not smooth. Then there exist
, and such that and
. Let such that and . For each
, we have

and it follows that

Hence is not uniformly smooth.

Example 20. space for is uniformly smooth.
Proposition 2.5. Let be a Banach space. Then
i). .
ii). is uniformly convex if and only if is uniformly smooth.

Proof. (i) We have for ,

49
The second assertion follows easily from the first. For example, if is
uniformly smooth then, for every , there exists a so that
. Hence, by (i), we get that i.e.
. This proves that is uniformly convex. The converse is
proved in a similar way.

2.5 Open Mapping Theorem and The Closed Graph

Theorem
2.5.1 Open mapping theorem
Definition: Let X and Y be topological space. A mapping
Is said to be open iff T maps open set of X into open sets
of Y.
2.5.2 Closed Graph theorem
The open mapping theorem is usually applied to a situation where
Y is a Banach space.
Definition: Let X and Y be a linear operator with domain D a
linear subspace of X. Then T is called a closed linear operator if
its graph

Is closed in the normed space XxY , where the two algebraic

operation of a vector space XxY are defined as usual , that is

For scalars and the norm on XxY is defined by

50
Theorem :(Closed graph theorem)

Let X and Y be Banach space and let be a closed linear

operator, where D is a linear subspace of X. Then if D is closed in
X , the operator T is bounded.

51
3. Strong, Weak and Weak* topologies

Let be a non empty set. Then and are indiscrete

topology and the discrete topology, respectively. If is any other topology
on , then . So, if and be two topologies on . Then we say
that is weaker than if . In this case, is stronger than .

Definition 3.1 (Weak Topology). Let be Banach space. The weak

topology on is the smallest topology such that is continuous, .
Open (closed) sets in this topology are called weakly open (weakly closed).

How to construct a weak topology

Let is a family of continuous functions, let is open in ,
then is open in X . We have such that . Take a
neighborhood . Basic neighborhood system for
in the weak topology is the collection of all sets of the form

Where is a finite collection of elements of . Then this collection

forms a weak topology.

Definition 3.2 (Hausdorff). A topological space is said to be

Hausdorff if for any two points with , there exists two disjoint
open sets and such that and .

Proposition 3.1. The weak topology is Hausdorff.

Proof. Let is a separates points

52
 . Let is a neighbourhood of and is a neighbourhood of
in , such that because is a Hausdorff space.Therefore, if
, and are weakly open sets, then
and hence weak topology is Hausdorff space.

Notation: Let is a sequence in . If in norm topology,

. If converges to in the weak topology, (weakly
convergent).

Proposition 3.2. Let be a Banach space and a sequence in .

(i.) .
(ii.) .
(iii.) , then is bounded and .
(iv.) in and in , then .

Proof. (i.) Suppose and . Let and be given.

Then is a weakly open set containing .
Choose such that for all . From this and definition of
weak topology we can conclude that . To prove the other way,
assume that for all .

Let be a weakly open set containing . Choose a and

such that

As for each , there exists such that

| | for all . Let { }. Then
for , we have

53
 | |

That is for all . Hence .

(ii.) Let by (i.) .
(iii.) Note that ( ) is bounded. Therefore, by the
uniform boundedness principle

(iv.) Note that by (a) and (c)

Therefore, .

Remark Proposition 3.2 (ii) shows that strong convergence implies

weak convergence. The converse, however, is false, i.e., weak
convergence does not always imply strong convergence. We give
the following example.
Example 21. Let be the set of sequences (real Hilbert space) with the

corresponding norm √ , where . Suppose

and let ⏟ such that .

By Riesz Representation Theorem, for each f

54
 , for some unique , and b B ’ ,

and so ∑

Hence, = 0. This implies

f( ) = →0 ,we see that →0 →∞.

However, { } does not converge strongly to zero since, for n m, because

and so is not convergent (not even Cauchy!).

Proposition 3.3. Let be a finite dimensional Banach space. Then the norm
and weak topology coincide.

Proof. Since weak topology contains fewer open sets than the strong
topology it is enough to show that every (strongly) open set is weakly open.

Let and strongly open with . Need to find

with for all
Let such that 。

Pick a basis in such that for all . Note that

are continuous linear functional on . Also

55
 ‖∑ ‖

∑ | |‖ ‖

∑ | |

Now take { }.Then is

weakly open and if , we have

Remark 3.1. Weakly open (resp. closed) sets are always open (resp. closed)
in the strong topology! If X is infinite-dimensional, the weak topology is
strictly smaller than the strong topology.

Definition 3.3 (Weak* Topology). The weak* topology is the smallest

topology on associated with the family , i.e., it is the smallest
topology on which makes all the maps , where
continuous.
OR

Let be a Banach space, be dual of . For all if

.

Proposition 3.4. Weak* convergent is weaker than weak convergent.(i.e.

strong topology)
Proof. Let we have that

56
 such that this is by definition .

Therefore, .

Proposition 3.6. When is finite-dimensional, then the three topologies

(strong, weak, weak*) on coincide.

Proposition 3.7. Let be a Banch space, be dual of . For all

, then such that .

Remark 3.2. Let be a Banch space, be dual of . For all

, then .

Theorem 3.1. The weak* topology is Hausdorff

Proof. Let such that . Let us take is a
neighborhood of and is a neighborhood of in or . Since both
the real line and complex plane are Hausdorff spaces, so .
Define two sets

and

So, and and since , hence

and are weak* open sets and disjoint.

57
3.2 MazurâĂŹs, AlaogluâĂŹs and Goldstine theorems.
Theorem 3.2 (Tychonoff's theorem). The product of any collection of
compact topological spaces is compact with respect to the product topology.

Theorem 3.3 (AlaogluâĂźs Theorem). Let be a Banach space. The unit

ball in is compact in the weak* topology.

Definition 3.7 (Totally bounded set). A subset A of a metric space is

totally bounded if with (so
each is within -distance from some ).

Definition 3.8 (Convex hull of a set). The convex hull of a set C is

{ ∑ }

Note that: Every compact set is totally bounded.

Proposition 3.8 (Compactness in complete metric spaces).

In a complete metric space subset is compact if and only if it is

closed and totally bounded.

Theorem 3.4 (MazurâĂźs Theorems). The closed convex hull of a

compact set of a Banach space is compact.

Theorem 3.5 (Goldstine Theorem). Let be a Banach space, then the

image of the closed unit ball under the canonical embedding into the
closed unit ball of the bidual space is a weak*-dense subset.

58
3.3 Fixed point theorem of Brouwer, Schauder, Tychnoff

Definition 3.9 (Fixed point). A fixed point of a mapping of a set

into itself is an which is mapped onto itself (is "kept fixed" by ),
that is,

the image coincides with .

Definition 3.10. A function between two topological spaces is a

homeomorphism if it has the following properties:

1. is a bijection (one-to-one and onto),

2. is continuous,

3. the inverse function is continuous ( is an open mapping).

3.3.1 Banach Contraction Principle

It is fundamental result in fixed point theory.

Theorem: let (X, d) be a complete metric space, and let T: be a

contraction mapping, i.e. there exists a constant [ such that

( )

59
 Example: Let [ ] with standard metric d | , and
define

by : =

Theorem 3.7 (Brouwer's fixed point theorem).

If the closed unit ball of , and is a continuous

map, then there is a point in such that .

Theorem 3.8. If is a Hilbert space, is a closed convex' nonempty

subset of , and , then there is a unique point in such that

Corollary 3.1. If is a nonempty compact convex subset of a finite

dimensional normed space and is a continuous function, then
there is a point in such that .

Proof. Since is isomorphic to either or , it is homeomorphic to

either or . So it suffices to assume that .

If , then the result is immediate from Brouwer's

Theorem. If is any compact convex subset of , let such that
.
Let be the function defined by the unique point in
such that

60
Then is continuous and for each in .
(In topological parlance, is a retract of B.) Hence is
continuous. By Brouwer's Theorem, there is an in such that
. Since . Hence and .

Schauder's Fixed Point Theorem is a generalization of the preceding

corollary to infinite dimensional spaces.

Definition 3.11. If is a normed space and , a function is

said to be compact if is continuous and clf(A) is compact whenever is a
bounded subset of .

Note that: If is itself a compact subset of , then every continuous

function from into is compact.

The following lemma will be needed in the proof of Schauder's Theorem.

Lemma 3.1. If is a compact subset of the normed space , and is

a finite subset of such that , define by

∑
∑

where if and if
. Then is a continuous function and

for all in .
Proof. Note that for each in and ∑ for
all in . So is well defined on .

61
The fact that is continuous follows from the fact that for each in
[ ] is continuous.
If , then

∑ [ ]
∑

If , then . Hence

∑
∑

This concludes the proof.

Theorem 3.9 (Schauder's fixed point theorem). Let be a closed bounded

convex subset of a normed space . If is a compact map such that
, then there is an in such that .

Proof. Let , so . For each positive integer let be a

finite subset of such that . For each let
as in the preceding lemma.

Now the definition of clearly implies that since

is convex; thus maps into . Also, Lemma (3.1) implies

Let be the linear span of the set and put . So is a

finite dimensional normed space, is a compact convex subset of , and

62
 is continuous.
By Corollary (3.1), there is a point in such that .
Now is a sequence in the compact set , so there is a point and a

subsequence { ( )} such that ( ) . Since ( ) ,

equation (24) implies

‖ ‖ ‖ ( ) ( )‖ ‖ ( ) ‖

‖ ( ) ‖

Thus . Since is continuous, ( ) .

Theorem 3.10 (Tychonoff's fixed point theorem). Let be a nonempty

compact convex subset of a locally convex topological linear space and
a continuous mapping. Then has a fixed point.

63