Banach Contraction Principle

It is a fundamental result in fixed point theory and metric space

analysis. It provides sufficient condition under which a mapping has a
unique fixed point.
Theorem :(Banach Contraction Mapping)
Let (X,d) be a complete metric space and let T: be a mapping , if
there exist a constant such that:
………………(1)
then
1. T has a unique fixed point x* i.e., T(x*) = x*
2. For any defined by = T(x)
converges to x*.

Proof: We construct a sequence ( ) and show that it is Cauchy, so that it

converges in the complete space X, and then we prove that its limit x is a fixed
point of T and T has no further fixed points. This is the idea of the proof.

We choose any and define the "iterative sequence" ( ) by

, …, ,..(2)

We show that ( ) is Cauchy. By (1) and (2),

 Hence by the triangle inequality and the formula for the sum of a
geometric progression we obtain for n > m

Since 0< a < 1, in the numerator we have < 1. Consequently,

On the right, 0< a < 1 and is fixed,

Since X is complete, ( ) converges, say, . i.e. taking the limit

an in = T(x), we have:
*= =
=
The limit of this the fixed point of mapping T.
Uniqueness of the fixed point
Suppose * are two fixed points of T. Then:

( )

( )

Therefore the mapping T has a unique fixed x* and the sequence (xn)
defined by successive iteration of T converges to x*
Example: Let with the usual metric .

Define by: . Show that T is a contraction mapping

and find its fixed point using the iterative algorithm
Solution : 1. Verify the contraction property.
For

Thus , T has a contraction mapping.

Next determine fixed point

T(x) =x , .

2. Algorithm (to find the fixed point)

Start with an initial guess
Let arbitrary starting point) compute the first few terms of the
sequence.
The sequence is

The sequence converges to zero ( ), which satisfies T(x*)= x*

Thus, the fixed point is x* = 0, and the sequence converges to zero.
Nonexpansive Mapping
A mapping T: on a metric space (X, d) is said to be nonexpansive
if:
,then
fixed points may exist if certain condition (e.g. Compactness or convex)
holds or may not exist for a nonexpansive mapping.

Example : Let and show that is it

nonexpansive or not?
Solution: For any

T is a nonexpansive, but not a contraction.

Contractive Mapping
A contractive mapping is a broader concept than a Banach contraction. It
satisfies:

( ) ,

Then a fixed point may exist if a certain condition (e.g., Compactness or

completeness of X) holds.
Example: Let with √

Show that is it contractive mapping? If it is? Find fixed point?

Solution: for √ √
The fixed point is x*= 1, since T(1)=1.
4 Operator Theory
4.1 Elementary Properties and Example in Hilbert
Space
Proposition 4.1. Let and be Hilbert spaces and a linear
transformation. The following statements are equivalent.
(a) is continuous.
(b) is continuous at 0 .
(c) is continuous at some point.
(d) There is a constant such that , for all .

Proof. We already proved that and in

Proposition (2.1). Let us show that : The definition of continuity at 0
implies that

So there is a such that . Let , since

every Hilbert space is vector space. So, ( ) .

If ‖ ‖ ,
 ( )

‖ ( )‖

‖ ( )‖

Letting with .

Therefore, (d) holds.

Definition 4.1. If , then

{ }

Also, is called the norm of and a linear transformation

with finite norm is called bounded. Let be the set of bounded linear
transformations from into . For .

Note that: all the bounded linear functionals on .

Proposition 4.2. (a) If and , then , and
 .
(b) If and , then and .
(c) If and , then and .

Proof. (a) Let and , then

and

Now,

and

Therefore, is bounded linear operator and hence .

And assume that

(b) Let and .

Hence is linear operator.

And since is bounded (i.e. ), we have

Therefore, is bounded linear operator on and hence .

And

(c) If , then . Hence, if and so let

and

Hence is linear operator on .

Therefore, and .

Note: Define a metric on , then is a metric

space with a metric .

Definition 4.2. If and are Hilbert spaces, a function is a

sesquilinear form (sesquilinear functional) if for in in , and in ,

1. Linearity in the first argument

(a) ;
2. Conjugate linearity in the second argument
(b) ‾ ‾ .

Definition 4.3. Sesquilinear form is bounded if there is a constant such that

for all in and in . The constant is called a bound for .

Theorem 4.1 (Riesz Representation Theorem). Let be a Hilbert space, then for
every continuous linear functional there exist unique such that

Theorem 4.2. If is a bounded sesquilinear form with bound , then

there are unique operators in and in such that

for all in and in and .

Proof. For each in , define by .
Let and , then
 ‾ (

And

Therefore, is linear and bounded. By the Riesz Representation Theorem there is

a unique vector in such that

and since , by Cauchy-Schwarz inequality, then

Suppose , then by applying the definition of norm on equation (27) we

have,

And

From equation (28) and (29), we have

 ,

Therefore,

Let . Then from equation (26) we have,

.

In order to show the uniqueness. If and

, then

Thus, is unique. And from equation (30) and , we have,

Similarly it is possible to show that there is a unique which satisfies

Therefore, , for all in and for all in .

4.2 Adjoint of Operators
Definition 4.4 (Adjoint of Operator). If , then the unique operator
in satisfying equation (25) is called the adjoint of and is denoted by
and defined by the property that for all

Definition 4.5. If , then

(a) is hermitian or self-adjoint if ;
(b) is unitary, if is bijective and ;
(c) is normal if .

Note: Self-adjoint or unitary operator both imply normal operator but not
conversely.
Proposition 4.3. Let be a bounded linear operator on a Hilbert space .
Then is self-adjoint if and only if is real for all .

Proof. , Let then is real for all

. And Conversely, since is real for all
implies and therefore . Hence is
self-adjoint.

Definition 4.6. Let and be Hilbert spaces with their respective

inner products. An isomorphism between and is a map , such that:

1. A is a linear operator

2. A is surjective

3.
These three requirements may be summarized by stating that A be a surjective
isometry.
Proposition 4.4. If , then is an isomorphism if and only if is
invertible and .

Proof. Suppose that: is an isomorphism. That is:

Then from the definition of the adjoint, we have:

So, from Inner Product is Sesquilinear, we have:

Setting , we have:

So from the positiveness of the inner product, we have:

so:

for all . So:

where is the identity map on . Since every isomorphism map in a Hilbert
Space is bijection, then we have that: is a bijection. So: is invertible. Let
be the inverse of . Then, we have:

So: is invertible with .

(Conversely) Suppose that: is invertible and since a mapping is bijective if and
only if has an inverse. So is surjective with . By the definition of an
isomorphism, we aim to show that:

for each . Let . Then, we have:

Hence is an isometry.
Therefore, is a surjective isometry, which is an isomorphism.
Proposition 4.5. Let be a Hilbert space. Then,

1. If , then .

2. If , then .

3. If and , then ‾ .

4. If , then .
 5. If , then .

Proof. 1. For all we have, , then

2. For all we have,

Then by definition of norm for , we have

And since by (1) and if we substitute instead of from

equation (34), we have

Then from equation (34) and equation (35) we have

Next, we have . But also, from equation (33), we have

. Taking the supremum over all unit vectors, we obtain

Consequently . Therefore,
 3. Let and , then

Therefore, ‾ .
4. For all , we have

and so .
5. For all , we have

and so it follows that .

4.3 Compact Operators

Definition 4.7 (Relatively Compact). Let be a metric space; is relatively
compact in , if ‾ (i.e. Closure set of ) is compact in .

Definition 4.8. A linear operator is compact if (ball ) has compact

closure in . The set of compact operators from into is denoted by
, and .

Definition 4.9 (Operators of Finite Rank). The image or range of a bounded

linear operator , written as Im , which is the subspace
 . If Im is finite dimensional, then is called an operator of finite
rank and

Definition 4.10 (Invertible operators). A bounded linear operator is

called invertible if there exist an operator such that

The operator is called the inverse of and moreover has at most one
inverse.
Definition 4.11 (Kernel of an Operator). The kernel of a bounded linear operator
: , written as Ker , is the closed subspace .
is called injective or one-one if ker .
is injective if and only if whenever . Thus, if is injective, then
the equation has at most one solution.