INTRODUCTION

In Functional Analysis and operator theory the notion of unbounded

operator provides an abstract framework for dealing with differential operators.

In constrast to bounded operators, unbounded operators on a given space

do not form an algebra , nor even a linear space, because each one is defined on
its own domain.

The theory of unbounded operators developed in the late 1920 ‘s and

early 1930 ‘s as part of developing a rigorous Mathematical framework
quantum mechanics. The theory ‘s development is due to John von Neumann
and Marshall stone. Von Neumann introduced using graphs to analyze
unbounded operators in 1936

In this thesis , Chapter -1 entitled as “ Preliminaries” we represent the

basic definitions which we needed for our study in next chapter

Chapter -2 entitled as “Unbounded Operators” it deals with definitions

and some related theorems.

Chapter -3 entitled as “The Sum Of Two Unbounded Linear

Operators” it states about sum of unbounded operators theorems.

Chapter -4 entitled as “Some Propertiesof Unbounded Operators With

Closed Range” in this we can see about theorems about closed range.

Chapter -5 entitled as “The Spectral Theorem” we can understand

about spectrum values from this chapter.

1
 CHAPTER-1

PRELIMINARIES

Definition 1.1:

A set X is said to be Metric space if with any two points p and q of X

there is associated a real number d(p,q) called distance from p to q such that

 d(p,q) >0 if p ≠q ; d(p,q)=0 iff p=q,

 d(p,q)=d(q,p)
 d(p,q)≤d(p,r)+d(r,q), for any p,q,r ϵ X

Any function with these properties is called distance function or a metric

Example 1.2:

Let d: R × R → R be a function defined by d(x, y) = | x – y |, x, y ∈ R. To

show that (R, d) is a metric space.

proof:
Since | x – y | ≥ 0 and | x – y | = 0 iff x = y, first condition follows
The other properties also follow
2.d(x, y) = | x – y |
= | y –x |
= d(y,x)
3. d(x, z) = | x – z |
= | x – y +y– z |
≤|x–y|+|y–z|

≤d(x,y) + d(y,z)

Thus d is a metric on R
Hence, (R, d) is a metric space.

2
Definition 1.3:
A sequence { pn} in a metric space X is said to be convergence if there
is a point p ϵ X with following property
For every ε > 0
there is an integer N such that n ≥ N implies that
d( pn , p ) < ε (here d denotes the distance in X)
We say that { pn} converges to p or p is the limit of { pn}
We write pn → p or
lim pn = p
n→∞

Definition 1.4:
A sequence { pn} in a metric space X is said to be Cauchy sequence if
for every ε > 0
there is an integer N such that n ≥ N implies that
d( pn , p m) < ε if n ≥ N∧m ≥ N

Definition 1.5:
A metric space (X ,d) is said to be complete if every Cauchy sequence in
X is converges.
Also, A metric space which is not complete is said to be incomplete.

Example 1.6:
Any distance metric is complete
Proof:
Let (M ,d) be a discrete metric space .
Let ( x n) be a Cauchy sequence in M.
Then there exists a positive integer n0 such that

3
 1
d( x n , x m) < 2 for all n,m ≥ n0

Since , d is the discrete metric distance between any two points is either 0 or 1.
∴ d ( x n , x m )=0 for all n , m≥ n0
∴ xn =x n =x ( say ) for all n≥ n0
0

∴d( x n , x ) = 0 for all n , ≥ n0

∴ ( xn )→ x

Hence , M is complete

Definition 1.7:
If X and Y are metric spaces with metric d and e respectively , then a
function F : X→Y is said to be continuous at
x 0 ϵ X if for every ε >0 , there is some δ>0 such that

e(F(x), F( x 0)) < ε for all x ϵ X satisfying

d( x , x 0 ) < δ .
Further,
F is said to be continuous on x if it is continuous at every point of X

Example 1.8:
Let ( M 1 , d 1) and ( M 2 ,d 2) be two metric spaces. Then any constant function
f : M 1 → M 2 is continuous

Proof:
Let f : M 1 → M 2 be given by f(x) = a where a ϵ M 2 is a fixed element.
Let x ϵ M 1 and ε > 0 be given .
Then for any δ >0 ,
f ( B ( x ,δ ) )= { a } ⊆ B ( a , ε )

= B ( f (x ), ε )
∴ f ( B ( x , δ )) ⊆ B ( f (x ), ε )

4
 ∴f is continuous at x.

Since x ϵ M 1 is arbitrary, f is continuous.

Definition 1.9:
If F : X→Y satisfies d ( x , x ' )=e ( F ( x ) , F ( x ' )) for all x , x ' ϵ X
Then F is called an isometry . If there is an isometry from X onto Y ,
Then X and Yare said to be isometric.

Definition 1.10:
Let (M,d) be a metric space. We say that a subset A of M is bounded if
there exists a real number k such that
d(x ,y)≤k for all x,yϵ A

Example 1.11:
Any finite subset A of a metric space (M,d) is bounded
Proof :
Let A be any finite subset of M.

If A= Φ

Then A is obviously bounded.

Let A ≠ Φ

Then {d(x ,y) / x ,yϵ A } is finite set of real numbers.

Let k = max { d(x ,y) / x ,yϵ A }

Clearly d(x,y) ≤ k for all x ,yϵ A

Therefore A is bounded.

5
Definition 1.12:

A subset E of X is said to be bounded if E⊂ U ( x , r) for some x ϵ X and r

¿ where U ( x , r) = { yϵ X :d (x , y)<r }

Definition 1.13:

A subset of x is said to be dense in X if É=X . This is the case if and only

if E∪ U (x , r)≠ ∅ for every xϵ X and every r¿ 0. In other words, E is dense in X if
every point of X is a limit point of E (or both) .

Definition 1.14:

A linear space or vector space over k is a non-empty set X along with a

function + : X × X → X , called addition and a function ∙ : K × X → X , called scalar
multiplication such that for all x ,y ,z ϵ X and k ,lϵ k , we have

x + y =y + x

x + ( y + z) =(x + y) + z

there exists 0ϵ X such that x +0 =x

there -xϵ X such that x + (-x) =0

k ∙ (x + y) =k ∙x + k ∙ y

(k + l) ∙x = k ∙x + l ∙ x

` kl∙ x=k ∙(l . x )

1∙ x=x

6
Definition 1.15:

Let X and Y are the linear spaces over K . A linear map from X to Y a
function F: X→Y such that

F( k 1 x 1+ k 2 x 2) = k 1 F( x¿¿ 1)+ k 2 F ( x ¿¿ 2)¿ ¿ for all x 1 , x 2 ϵ X and k 1 , k 2 ϵ K

Two important subspaces are associated with such a map.

Range space : The subspace

R(F) = { y ϵ Y : F ( x ) = y for some x ϵ X }

of Y is called the range space of F.

Zero space : The subspace

Z(F) = { x ϵ X :F ( x )=0}

of X is called the zero space of F. If Z(F) = X , we write F= 0.

Definition 1.16:

Let V ,W be two vector spaces both over same field F. The mappings
from V into W and such mappings are also called operators or
transformation.

An operator T: V → W is said to be

1.additive if it preserves the addition

T( v1 + v 2) = T( v1 ¿+T ¿ ) for v1 , v 2 ϵ V .

2. homogeneous if it preserves the multiplication

T(α v ) = α (Tv ) for α ϵ F , v ϵ V

7
The map T: V → W is called a linear transformation or operator if it is both
additive and homogeneous.

Note :

A mapping from a normed space X into a normed space Y is called an

operator.

Definition 1.17:

Let N be a linear space and let ‖∙‖: N → R be a function such that for

all x ,y ϵ N and for all α (may be real or complex) following conditions are
satisfied

 ‖x‖ ≥ 0
 ‖x‖ = 0 iff x =0
 ‖x + y‖ ≤‖x‖+‖ y‖
 ‖α x‖ ≤|α|‖x‖

Then the function ‖∙‖ is called a norm on N and ordered pair

( N , ‖∙‖) is called a normed linear space.

Example 1.18:

Let X = R , K= R

R is a linear space over R

Define ‖‖: R → R by ‖x‖ = |x| absolute value of x , x ϵ R

‖x‖ = |x| > 0

‖x‖> 0 ⇔ |x| =0

⇔ x=0

8
Let k ϵ R , x ϵ R

‖x + y‖ = |x + y|

≤|x|+| y|

= ‖x‖+‖ y‖

‖‖ is norm on R.

Therefore R is normed space .

Definition 1.19:

Let X be a linear space over K. An inner product on X is a function

⟨ , ⟩ : X × X → K such that for all x ,y ,z ϵ X kϵK , we have

 Positive definiteness : ⟨ x , x ⟩ ≥0∧⟨ x , x ⟩=0iff x =0

 Linearity in the first variable

⟨ x + y , z ⟩= ⟨ x , z ⟩ + ⟨ y , z ⟩ ∧⟨ kx , y ⟩=k ⟨ x , y ⟩

 Conjugate symmetry ⟨ y , x ⟩ =⟨ x ´, y ⟩

Note :

It follows that an inner product is conjugate linear in the second variable

⟨ x , y + z ⟩= ⟨ x , y ⟩+ ⟨ x , z ⟩ ∧⟨ x , ky ⟩ =ḱ ⟨ x , y ⟩

An inner product space is a linear space with norm on it .

Example 1.20:

Let X be a linear space over k

1
Define ⟨ , ⟩ : X → K by ‖x‖ = ⟨ x , x ⟩ 2 , x ϵ X

9
 1
‖x‖ = ⟨ x , x ⟩ 2 ≥ 0

1
And ⟨ x , x ⟩ 2 = 0

⟨ x , x ⟩=0

iff x = 0

Thus linearity and conjugate symmetry are also true .

Hence it defines a norm on X.

Definition 1.21:

A normed linear space X over K is called a Banach space if X is

complete in the metric space d(x ,y) = ‖x− y‖ induced by the norm ‖‖ .

Example 1.22:

 K nwith any of the norms ‖‖p, 1≤ p ≤ ∞ ,

 Finite dimensional normed spaces,
 l p, 1≤ p ≤ ∞ , with norm ‖‖p ,

Definition 1.23:

An inner product space which is complete in the norm induced by the

inner product is called a Hilbert space.

In other words,

A vector space V over the field F is a Hilbert space iff the

following two conditions hold

 There is an inner product on V

 Every Cauchy sequence with respect to the induced norm is
convergent.

10
Example 1.24:

Let H = K n.

For x=(x(1), .........x(n) ), y=(y(1), .........y(n) ) in H,

define
n
⟨ x , y ⟩ =∑ x ( j) y (´ j) .
j =1

Thus ⟨ , ⟩ is an inner product on H.

1
And ‖x‖2 =⟨ x , x ⟩ 2 =√ x ( 1 )2+ … … …+ x ( n )2 , xϵH .

∴ H is complete∈the metric d 2 ( x , y ) =‖x− y‖2.

Hence , H is a Hilbert space .

Theorem 1.25:

Let H 1∧H 2 be Hilbert spaces and A : H 1 → H 2 be linear.

A is continuous if and only if the graph

Gr(A) = {(x, A(x) ) : xϵ H 1} of a closed in the Hilbert space H 1 × H 2

with inner product

¿ = ⟨ x 1 , y 1 ⟩1 + ⟨ x 2 , y 2 ⟩ 2 .

Definition 1.26:

Let A ϵ B ( H ), the unique element B of B(H) which satisfies

⟨ A ( x ) , y ⟩ =⟨ x , B ( y ) ⟩

for all x ,yϵ H is called the adjoint of A. It is denoted by A¿.

11
Definition 1.27:

A Borel Set ia any set in a topological space that can be formed from
open sets or equivalently from closed sets through the operations of countable
union , countable intersection , and relative complement .

For a topological space X, the collection of all borel sets on X forms a

σ −¿algebra , known as the Borel Algebra Or Borel σ −¿Algebra.

Note :

The borel algebra on X is the smallest σ −¿algebra containing all open

sets or equivalently all closed sets.

12
 CHAPTER- 2

UNBOUNDED OPERATORS

2.1: Unbounded Operators

Definition 2.1:

Let x and y be a normed linear spaces and suppose T : x → y is a linear

map.

Define

║ Tx ║
║ T ║ op={ / x≠o}
║ x║

={ ║ T x ║ /║ x ║ ≤1 }

={ ║ T x ║ /║ x ║=1 }

if ║ T ║< ∝,

Then A is bounded linear operator

Definition 2.2 :

Let H be a Hilbert space An unbounded linear operator consists of a

dense linear subspace D (T) and a linear map T: D(T)→H. The linear subspace
D (T) is called the Domain of T

Range:

We define the range of T as

R (T) = → {T x / xϵD(T ) }

13
Graph:

The graph g(T ) of an operator T in H is the subspace of H × H that

consists of ordered pairs {x .T x }

g ( T ) =¿}

Definition 2.3:

Let H 1 , H 2 be Hilbert spaces and T : D ( T ) → H 2 be alinear operator T is said

to be densely defined operator if D(T) is dense linear subspace of H 1

Definition 2.4:

A closed operator in H is one whose graph is a closed subspace of H × H

That is, T is called closed operator, if the graph

g ( T ) =¿} is a closed subspace of H × H

We say that an operator T in H is closable if there is a closed operator S

in H such that T ⊂ S

Definition 2.5:

Let S.T be linear operators in H. When g ( T ) ⊂ g( s) we write T ⊂ S

Then S is an extension of T

Definition 2.6 :

Let T be an operator in H We define

T ¿ : D ( T ¿ ) → H by T ¿ =¿ x y
y

Which satisfies ⟨ T x , y ⟩ =⟨ x , T ¿ y ⟩ , x ϵ D(T )

T ¿ iscalled t h e adjoint of T .

14
We say that T is self – adjoint when T= T ¿

Theorem 2.7:

If S, T and ST are densely defined operators in H, Then

T ¿ S ¿ ⊂¿

If Sϵ B ( H ) T h en

T ¿ S ¿=(ST )¿

Proof:
Let y ϵ D(T ¿ ¿ ¿ S )∧xϵ D(ST )¿
¿

¿
Then S y ϵ D(T ¿¿ ¿)∧xϵ D (T ) ¿

So ⟨ T x , S y ⟩=⟨ x , T S y ⟩
¿ ¿ ¿

On The other hand, y ∈ D (S )

¿

So ⟨ ST x , y ⟩=⟨ T x , S y ⟩
¿

Hence ⟨ ST x , y ⟩ = ⟨ x , T ¿ S ¿ y ⟩

which implies that

¿
(ST) y = T ¿ S ¿ y for each yϵ D(T ¿ S ¿ ¿

Thus, T ¿ S ¿ ⊂ ¿

Suppose that S∈ B( H )

Hence, S¿ ϵ B ( H ) for w hic h D ( S ¿ ) =H

Let yϵ D((ST )¿¿ ¿)¿

for x ϵ D ( ST )

15
 ¿
⟨T x , S y ⟩=⟨ ST x , y ⟩=⟨ x ,( ST ) y ⟩
¿

This implies that S y ϵ D(T )

¿ ¿

and hence y ϵ D(T ¿ S ¿ ¿

Thus, we get D((ST )¿¿ ¿) CD (T ¿ S¿ ) ¿ ----------②

From ① &②

T ¿ S ¿=¿.

Definition 2.8:

If T is an operator in H we say that T is symmetric if

⟨ T x , y ⟩= ⟨ x , T y ⟩ , x , yϵD (T )

Theorem 2.9:

Let T be a densely defined operator in H. T is symmetric if only if T ⊂ T ¿

Proof:

Suppose that T is symmetric

Let ( y ,T y ) ϵ g(T ) for xϵ D(T )

¿ ⟨ T x , y ⟩∨¿∨⟨ x ,T y ⟩ │ ≤║ x ║║ T y ║

Hence, x → ⟨ T x , y ⟩ is continuous an D ( T ) ,

that is yϵ D(T ¿ )

For xϵ D ¿)

¿
We have ⟨ T x , y ⟩=⟨ x , T y ⟩ and on the other hand

⟨ T x , y ⟩=⟨ x , T y ⟩

16
 ¿
Thus, ⟨ x , T y ⟩=⟨ x ,T y ⟩ for all xϵ D (T)

Again, D (T) is dense in H

¿
we get T y =T y

that is , ¿,T y ¿ ϵ g(T ¿ ¿

Thus, g ( T ) ⊂ g ( T ¿ )

Hence, T⊂ T ¿

Conversly,

Suppose that g ( T ) ⊂ g(T )

¿

Let x , y ϵ D(T )
¿
We have ( y , T y ) ϵ g(T )

yϵ D ( T ¿ )∧T ¿ =T y
y

¿
Thus, ⟨ T x , y ⟩=⟨ x , T y ⟩=⟨ x ,T y ⟩

Hence, T is symmetric

Note:

If T is a symmetric operator in H then T is closable and T́ is symmetric

Definition 2.10:

An Operator T in H is said to be essentially self-adjoint when T is

densely defined symmetric and T́ (which is densely defined) is self -adjoint

17
2.2 Graphs:

For (a,b), (c,d) ϵ H × H define ⟨ (a,b), (c,d) ⟩ =⟨ a , c ⟩+⟨ b ,d ⟩

This is an inner product on H × H with which H × H is a Hilbert space.

We define V: H × H → H × H by

V (a,b) = (-b, a), for all (a,b) ϵ H × H

Which belongs to B ( H × H ). It is immediate that

VV ¿ =I ∧V ¿ V =I namely, V is unitary. As well V 2=−I whose if M is a

linear subspace of H × H then V 2 M =M.

Theorem 2.11:

Suppose that T is densely defined operator in H. It holds that

g(T ¿ ) = (Vϵ g(T )¿⟘

Proof:

Let T be a densely defined operator in H

for y, z.

{ y , z }ϵ g (T ¿ ¿

⟨ T x , y ⟩=( x . z ) for every x ϵ D (T)

⟨ { T x , x }, { y , z }⟩=0 for every x ϵ D (T)

{ y , z }ϵ [V g(T )¿⟘

18
 Hence,

g(T ¿ ) = (Vϵ g(T )¿⟘

Theorem 2.12:

If T is a densely defined operator in H Then T ¿ is a closed operator

Proof:

Let V g (T) is a linear subspace of H × H

The orthogonal complement of linear subspace of a Hilbert space is a

closed linear subspace of the Hilbert space.

By above Theorem, we have

g ¿ ¿) = [V g (T ¿ ¿ ¿⟘

Thus, g ¿ ¿) is a closed linear subspace of H × H

Hence, T ¿ is a closed operator

Note:

Let T be densely defined operator in H. If T is self – adjoint Then by

above Theorem T is itself a closed operator.

Theorem 2.13:

Suppose that T is a closed densely defined operator in H. Then

H × H=¿ (V g (T) ⨁ g ( T ) is an ort h ogonal direct ∑ ¿

¿

Proof:
19
Generally if M is a linear subspace of H × H

H × H = Ḿ ⨁ M ⟘ = Ḿ ⨁ ( Ḿ ¿⟘ is an orthogonal direct sum

For, M = V ( g ( T )

Since, g ( T ) is a closed linear subspace of H × H

So is M

Thus,

H × H=V g ( T ) ⨁ ¿

Since, ¿ g ( T ¿ )

We get H × H =V ( g ( T ) + ⨁ g ( T ¿ )

Hence, the Theorem

Theorem 2.14:

Suppose that T is a densely defined symmetric operator in H Then the

following statements are true

a) If D(T) =H, then T is self is self – adjoint and Tϵ B(H)

b) If T is self – adjoint and one- to- one then R(T) is dense in H and T −1is
self – adjoint
c) If R (T) is dense in H, Then T is one-to-one
d) If R(T) = H then T is self – adjoint and T −1 ϵB ( H )

Proof:

a) By assumption, T⊂ T ¿
If D(T) = H, it is is obvious that T=T ¿

20
 Hence, T is closed
∴ T is continuous

b) Suppose y ⟘ R (T)
Then x → ⟨ T x , y ⟩=0 iscontinuous in D (T)

Hence, y ϵ D(T )=D ( T )

¿

and ⟨ x , T y ⟩ = ⟨ T x , y ⟩=0 for all xϵ D(T )

Thus, T y =0
Since, T is assumed to be one-to-one
It follows that y=0
This proves that R(T) is dense in H
T −1is therefore densely defined with D (T −1)* exists
The relations g(T ¿¿−1) ¿ = V g (-T) and
V g(T ¿¿−1) ¿ = g (-T) are easily verified ------------ ①
¿

⇔ { b ,−a }ϵg (−T )

⇔ { a ,b }ϵ V g (−T )
Since, T is self-adjoint
⇒ T is closed
Hence, -T is closed
⇒T −1 is closed by ①
Now, We can use theorem 2.13 for T −1and –T then the orthogonal
decompositions are
H × H=v g (T ¿¿−1) ⨁ g ¿ ¿ ------------------- 2

And
H × H=v g (−T ) ⨁ g ( – T )

¿ g(T −1) ⨁ vg ¿) ---------------③

Consequently
g((T −1 ¿ ¿ ¿=¿ (T −1 ¿ ⟘ = g (T −1) -------------------- ④

Thus, we get (T −1 ¿ ¿ = T −1

21
 Hence, T −1 is self – adjoint
c) Suppose T x = 0
Then ⟨ x , T y , ⟩=⟨ T x , y ⟩= 0 for every y ϵ D(T )
Thus, x ⟘R (T) and Therefore x=0
d) Since, R (T)=H
By using (c) T is one to one and D ( T −1 )=H
if x ϵ H and y ϵ H
Then x=¿ T z , and y =T w for some z ϵ D(T )∧w ϵD(T ),
So that
⟨T −1 x , y ⟩=⟨ z , T w ⟩
= ⟨ T z,w ⟩
=⟨ x , T −1 y ⟩
Hence, T −1 is symmetric
(a) implies that T −1 is self-adjoint and bounded
(b) implies that T= ¿ ¿ is also self – adjoint
Hence, T −1 ϵ B(H )

Theorem 2.15:

If T is a closed densly defined operator in H. Then D ¿ ¿) is dense in H

and T ¿∗¿=T ¿

Proof:

Since, V is unitary and V 2=−I

Theorem 2.13 gives the orthogonal decomposition.

H × H=g ( T ) V g(T ¿ ) --------------------1

Suppose Z ⟘ D(T ¿¿ ¿)¿

Then ⟨ z , y ⟩=0

22
 ¿
and Therefore ⟨ { o , z },{−T y y }⟩=0 for all y ϵ D(T ¿¿ ¿)¿ -------------②

Thus, { o , z }ϵ ¿

which implies that z = T (0) =0

Consequently, D ( T ¿ ) is dense in H and T ¿∗¿ ¿ is defined

Again by 2.13

H × H=V g ( T ¿ ) ⊕ g ¿ ¿ --------------- ③

By ① and ②

g ¿ ¿ = g(T )

g ¿ ¿ = g(T )

So that (T ¿∗¿ ¿) = T

Theorem 2.16:

Suppose that T is a closed densely defined operator in H and let

Q¿ I + T ¿ T , with
¿
D ( Q )=D ( T ¿ T )=¿ { x ϵ D(T )/ T x ϵ D(T ) }

The following statements are true

1. Q :D(Q) → H is a bijection and there are B, Cϵ B(H ) with

║B║≤1, B ≥0, ║C║≤ 1 , C = TB and

B(I +T ¿ T ) C ( I + T ¿ T ¿ B=I
T ¿ T is self adjoint

2.Let T 0 be the restriction of T to D (T ¿ T ¿ . Then ( g T o ) is dense ∈g (T )

Proof:

23
 ¿
If xϵ D(Q) and Then T x ϵ D(T )

So that

⟨ x , x ⟩ +⟨T x ,T x⟩ = ⟨ x , x ⟩ + ⟨ x ,T T x⟩
¿

= ⟨ x , ax ⟩ -------------- ①

Therefore ║ x ║ 2 ║ ≤ ║ x ║ ║ Q x ║ , which shows That Q is one-to-one

By theorem 2.13 there corresponds to every h ϵ H a unique vector

B h ϵ D(T ) and a unique C h ϵD(T ) such that
¿

{ o ,h } = {−T Bh , B h } + { c h , T ¿ c h } --------------------②

It is clear that B and C are linear operators in H with domain H The

vectors on the right of ② are arthogonal to each other.

The definition of norm in H × H implies that

║h ║2 ≥║ B h║ 2+║ C h║ 2 (hϵH ) ---------------------③

So that

║ B║ ≤ 1and ║ C ║ ≤ 1

Consideration of the components in ② shows that C=TB

and that

h=B h+ T ¿ C h

¿ B h+T ¿ TB h

= QB h --------------- ④ for every h ϵH

Hence QB=I

In particular, B is one-to-one mapping of H onto D(Q)

24
If yϵD (Q) then y=B h for some h ϵ H

Hence, Qy=QBH=h

and BQy = B h =y

Thus, BQ ⊂ I

If h ϵH then h ϵQx for some xϵ D(Q)

So that

⟨ B h , h ⟩ = ⟨ BQ x , Qx ⟩

= ⟨ x ,Q x ⟩≥ 0 -------------------------⑤

By ①,Thus, B ≥ 0, B is self – adjoint

⇛Q is self- adjoint

So is T ¿ T =Q-I

This complete the proof of part (1)

Since, T is a closed operator g(T ) is closed subspace of H × H

Hence, g(T ) is a Hilbert space

Assume { z ,T z }ϵ g (T ) is orthogonal to g(T )

Then for every xϵ D(T T ) = D(Q)

¿

0=⟨ { z , T z }, { x , T x }⟩

= ⟨ z , x ⟩ + ⟨T z , T x ⟩

= ⟨ z , x ⟩+ ⟨ z ,T T x ⟩
¿

= ⟨ z ,Q x ⟩

25
But R ( Q )=H

Hence, z=0

This proves ②

Definition 2.17:

Let T be a symmetric operator in H. we say that T is maximally

symmetric if T⊂S and S being symmetric imply that S=T

Result:

Self –adjoint operators are maximally symmetric

Proof:

Suppose T is self-adjoint

S is symmetric and T⊂ S

By the definition of the adjoint, This inclusion implies obviously

S¿ ⊂ T ¿

Hence, S ⊂S ¿ ⊂T ¿ =T ⊂ S

Which proves that S=T ,

Theorem 2.18:

Suppose that T is a symmetric operator in H and let j be I or –i then

a) ║ T x + j x ║2 = ║ x ║ 2+║ T x ║2 for xϵ D(T )

b) T is closed if and only if R(T + jI ) is closed subset of H
c) T + jI is one-to-one
d) If R ( T + jI )=H then T is maximally symmetric

26
Proof:

Statement (a) follows from the identity

║ T x +ix ║2=║ x ║ 2+ ║ T x ║ 2+ ¿ ⟨ ix ,T x ⟩ + ⟨ T x , ix ⟩ ,

combined with the symmetry of T

By (a) ⟨ T +iI ⟩ x ↔ { x ,T x } is an isometric one-to-one correspondence an between

the range of T + iI and graph of T

This proves (b)

Next(c ) is also an immediate consequence of (a)

If R(T +iI )=H

and T 1is a proper extension of T

[, Th at is , D ( T ) is a proper⊂of D(T 1 )]

Then T 1+iI is a proper extension of T + iI which cannot be one-to-one

By (c) T 1is not symmetric

This prove (d)

Hence the Theorem

CHAPTER-3

THE SUM OF TWO UNBOUNDED LINEAR

OPERATORS

Introduction

27
 We start with some standard notions and results about linear operators on
a Hilbert space. All operators are assumed to be densely defined together with
any operation involving them or their adjoints. Bounded operators are assumed
to be defined on the whole Hilbert space.
1.If A and B are two unbounded operators with domains D(A) and D(B)
respectively, then B is called an extension of A, and we write A ⊂ B, if
D(A)⊂D(B)and if A and B coincide on D(A).
2.If A ⊂ B, then B¿⊂ A¿.
3.The product AB of two unbounded operators A and B is defined by
BA(x) = B(Ax) for x ∈ D(BA)
Where
D(BA) = {x ∈ D(A) : Ax ∈ D(B)}.
Note :
AB = BA, we mean that ABx = BAx for all x ∈ D(AB) = D(BA).

Definition 3.1 :
The unbounded operator A, defined on a Hilbert space H, is said to
be invertible if there exists an everywhere defined (i.e. on the whole of H)
bounded operator B such that
BA ⊂ AB = I
where I is the usual identity operator.

Definition 3.2:
An unbounded operator A is said to be
closed
if its graph is closed
self-adjoint

28
 if A = A¿(hence from known facts self-adjoint operators are automatically
closed);
normal
if it is closed and A A¿ = A¿A (this implies that D(A A¿) = D( A¿A)).

Lemma 3.3:
Let A be a densely defined operator. If A is invertible, then it is closed.
Remark:
When dealing with self-adjoint and normal operators, taking adjoints is
compulsory, so we list some straightforward results about the adjoint of the sum
and the product of unbounded operators.

Theorem 3.4:
Let A be an unbounded operator.
(1) (A + B)¿ = A¿ + B¿ if B is bounded, and (A + B)¿⊃ A¿ + B¿ if B is
unbounded.
(2) A + B is closed if A is assumed to be closed and if B is bounded.
(3) (BA)¿ = A¿B¿ if B is bounded.
(4) A¿B¿ ⊂ (BA)¿ for any unbounded B and if BA is densely defined.

Lemma 3.5:
The product AB (in this order) of two closed operators is closed if one
of the following occurs:
(1) A is invertible,
(2) B is bounded.

Lemma 3.6:
If A and B are densely defined and A is invertible with inverse

29
A−1in B(H), then (BA) = A B .
¿ ¿ ¿

Proof:
Since A is invertible, A A−1 = I
And hence
BA A−1 = B
⇒ ( A−1)¿ (BA)¿
⊂ [(BA) A−1]¿
⊂ B¿.
BA A−1⊂ B¿.
But ( A−1)¿ = (A¿ ¿−1 and so
A¿ (A¿ ¿−1 (BA)¿= (BA)¿
⊂ A¿B¿.
Since, we always have A¿B¿ ⊂ (BA)¿
Hence the lemma

Theorem 3.7:
Let A and B be two unbounded operators such that AB = BA. If
A (for instance) is invertible, B is closed and D(B A−1) ⊂ D(A), then A + B is
closed on D(B).
Proof:
First note that the domain of A + B is D(A) ∩ D(B).
But AB = BA
⇒ A−1B ⊂ B A−1
⇒ D(B) = D( A−1B) ⊂ D(B A−1) ⊂ D(A),
Hence the domain of A + B is D(B).
By Lemma 3.3 ,
A is automatically closed.
Then we have
A + B = A + BA A−1
30
 = A + AB A−1
= A(I + B A−1) (since D(B A−1) ⊂ D(A)).
Since A−1 is bounded (and B is closed), B A−1 is closed,
Hence by Theorem 3.5,
I + B A−1 is closed so that A(I + B A−1) is also closed by
Lemma 3.4, proving the closedness of A + B on D(B).

Remark :
The hypothesis D(B A−1) ⊂ D(A) cannot merely be dropped. As a
counterexample, let A be any invertible closed operator with domain
D(A)⊊ H where H is a complex Hilbert space.
Let B = −A. Then A + B = 0 on D(A) is notclosed.
Moreover, AB = BA but
D(B A−1) = D(A A−1)
= D(I) = H ⊄ D(A).

Theorem 3.8:
Let A and B be two unbounded self-adjoint operators such that B (for
instance) is also invertible. If AB = BA and
D(A B−1) ⊂ D(B),
Then A + B is self-adjoint on D(A).

Proof:
It is clear, D(A) = D( B−1A)
⊂ D(AB−1) ⊂ D(B),
that the domain of A + B is D(A).
We have
BA
⏟
B−1
¿ AB
+B ⊂ A + B

31
and hence
(I + B−1A)B ⊂ A + B.
So we have
(A + B¿¿ ⊂[(I + B−1A)B]¿
=B¿ (I + B−1A)¿ (by Lemma 3.6 since B is invertible)
=B¿ [I + ( B−1A)¿] (by Theorem 3.5)
=B¿ [I + A¿ ( B−1)¿] (since B−1 is bounded)
=B(I + A B−1) (since A and B are self-adjoint)
=B + BA B−1 (for D(A B−1) ⊂ D(B))
=B + AB B−1
=A + B.
We know the fact
A + B ⊂ (A + B)¿,
Hence , A + B is self-adjoint on D(A).

Proposition 3.9:
Let A and B be two unbounded self-adjoint operators such that B
(for instance) is also invertible. If AB = BA,
then A and B strongly commute.
Proof :
By Lemma 3.6, we may write
(AB)¿ = B¿A¿ = BA = AB,
i.e. AB is self-adjoint.
By the Devinatz-Nussbaum-von Neumann theorem,
A and B strongly commute.

Theorem 3.10:
Let A and B be two unbounded invertible operators such that

32
AB =BA. If D(A¿ (B¿ ¿−1) ⊂ D(B¿),
Then (A + B)¿ = A¿ + B¿.
Proof:
First, we always have
A¿ + B¿ ⊂ (A + B)¿.
Second, since A and B are both invertible, by Lemma 3.6,
we have
AB = BA
⇒ A¿B¿ = B¿A¿.
BA
⏟
Now write B−1 ¿ AB
+B ⊂ A + B

so that
(I + B−1A)B ⊂ A + B
and hence
(A + B)¿ ⊂[(I + B−1A)B]¿
=B¿ (I + B−1A)¿ (by Lemma 3.6 since B is invertible)
=B¿ [I + ( B−1A)¿] (by Theorem 3.5)
=B¿ [I + A¿ ( B−1)¿] (since B−1is bounded)
=B¿ + B¿ A¿ (B¿ ¿−1 (as D(A¿ (B¿ ¿−1 ⊂ D(B¿))
=B¿ + A¿B¿ (B¿ ¿−1 (for A¿B¿ = B¿A¿)
=A¿ + B¿.
Hence, the theorem.

Theorem 3.11:
Let A and B be two unbounded normal and invertible operators. Then
AB = BA ⇒ AB¿ = B¿A and BA¿ = A¿B.
Proof :
Since B is invertible, we may write
AB = BA

33
 ⇒ B−1A ⊂ A B−1.
Since B−1 is bounded and A is unbounded and normal,
By the classic Fuglede theorem we have
B−1A ⊂ A B−1
⇒ B−1A¿⊂ A¿ B−1.
Therefore,
A¿B ⊂ BA¿.
But B is invertible, then by Lemma 3.6 we may obtain
AB¿ ⊂ (BA¿)¿
⊂ (A¿B)¿ = B¿A.
Interchanging the roles of A and B, we shall get
B¿A ⊂ AB¿ and BA¿ ⊂ A¿B.
Thus,
B¿A = AB¿ and BA¿ = A¿B.
Hence , the theorem

Example 3.12
Let S be the unilateral shift on the Hilbert space l 2.
We may then easily show that both S+S¿ and S−S¿ are injective.
Hence, A = (S+S¿ ¿−1 and B = i(S−S¿ ¿−1are unbounded
self-adjoint such that AB = BA.

Nonetheless
A−1 B−1 ≠ B−1 A−1

Since, otherwise S and S¿would commute.

Theorem 3.13:

34
 Let A and B be two unbounded invertible normal operators with
domains D(A) and D(B) respectively. If AB = BA,
D(A) ⊂ D(BA¿) and
D(A B−1) =D(A(B¿ ¿−1) ⊂ D(B),
then A + B is normal on D(A).
Proof:
To prove that A + B is normal, we must say why A + B is closed and
show That (A + B)¿(A + B) = (A + B)(A + B)¿.
We know that A + B is closed
Also, since A and B are both normal,
We obviously have
D(A(B¿ ¿−1) ⊂ D(B) ⇒ D(A¿(B¿ ¿−1) ⊂ D(B¿)
so that Theorem 3.8, applies and yields
(A + B)¿ = A¿ + B¿.
Since AB = BA,
By above theorem implies that
AB¿ = B¿A and BA¿ = A¿B.
Since D(A) ⊂ D(BA¿),
we have
D(A) ⊂ D(A¿B) (⊂ D(B))
and
D(A) ⊂ D(BA¿)
= D(A¿B)
= D(AB)
= D(BA)
= D(B¿A).
All these domain inclusions allow us to have
1. A¿(A + B) = A¿A + A¿B and
B¿(A + B) = B¿A + B¿B.

35
 2. A(A¿ + B¿) = AA¿ + AB¿ and
B(A¿ + B¿) = BA¿ + BB¿.
Hence we may write
(A + B)¿(A + B) = A¿A + A¿B + B¿A + B¿B
and
(A + B)(A + B)¿ = AA¿ + AB¿ + BA¿ + BB¿.
Thus
(A + B)¿(A + B) = (A + B)(A + B)¿.

Corollary 3.14:
Let A and B be two unbounded invertible normal operators. If
BA =AB, then BA (and AB) is normal.

Remark:
By a result of Devinatz-Nussbaum , if A, B and N are normal
where N = AB = BA, then A and B strongly commute.

Corollary 3.15:
Let A and B be two unbounded invertible normal operators. If BA =
AB, then A and B strongly commute.

Proof:
We first note that BA is closed (or simply since BA
is invertible hence Lemma 3.3 applies),
Hence so is AB. Lemma 3.6 then gives us
(AB)¿ = B¿A¿.
By Theorem 3.12 we may then write

36
 (AB)¿AB = B¿A¿AB
= B¿AA¿B
= AB¿A¿B
= AB¿BA¿
= ABB¿A¿
= (AB)(AB)¿
This proves the corollary.

CHAPTER-4
SOME PROPERTIES OF UNBOUNDED OPERATORS
WITH CLOSED RANGE

Introduction
Let H 1 , H 2 be Hilbert spaces and T be a closed linear operator defined

37
on a dense subspace D(T) in H 1and taking values in H 2
In this note we establish the following:
(i) Range of T is closed if and only if 0 is not an accumulation point of
the spectrum σ (T ¿T) of T ¿T.

In addition, if H 1= H 2 and T is self adjoint, then

(ii) every isolated spectral value of T is an eigen value of T,
(iii) inf {||Tx|| : xϵ D(T)∩N(T) ┴, ||x|| = 1} = inf { |λ| : 0≠ λϵ σ(T)},

(iv) range of T is closed if and only if 0 is not an accumulation point of

the spectrum σ (T) of T,
(v) T is bounded if and only if σ (T) is bounded.

Definition 4.1:
An operator T ϵ L(H) is said to be positive if
〈 x, x 〉 ≥0 for all x ϵ D(T).

Definition 4.2:
Let T ϵ L(H). If T is one to one, then the Inverse of T is the linear
operator T −1 : R(T) →H defined in such a way that
T −1 Tx = x for all x ϵ D(T) and
T T −1 y = y for all y ϵ R(T).
Definition 4.3:
For T ϵ L(H), the resolvent of T is denoted
by ρ (T) and is defined as
ρ (T) = {λ ϵ C: T – λI is bijective and (T – λI )−1 ϵ B(H)}.

Definition 4.4:
For T ϵ L(H), the spectrum σ (T), approximate

38
spectrum σ a (T), and the eigen spectrumσ e (T) are defined by
σ (T) = C \ ρ (T),
σ a (T) = {λϵ C: T – λI is not bounded below},

σ e (T) = {λ ϵ C: T – λI is not one to one}

Result:
By Closed Graph Theorem , it follows that if T ϵ C(H), then
σ (T) = {λ ϵ C: T – λI is not bijective}

Definition 4.5:
Let T ϵ L( H 1 , H 2). The subspace C(T) = D(T)∩N(T) ┴ is called the
carrier of T. We denote the operator T}C(T ) ¿ byT 0 .

Definition 4.6:
Let T ϵ L( H 1 , H 2). The maximal generalized inverse of T, denoted by
T†, is a unique linear operator with domain
D(T†) = R(T) ⊕┴ R(T) ┴ and N(T†) = R(T) ┴
satisfying the following properties:
1. R(T) ⊆ D(T†)
2. R(T†) ⊆D(T)
3. T†Tx = P R(T´ † )x, for all x ϵ D(T)
4. TT†y = P R(T´ † )y, for all y ϵ D(T†).
In the above, P M denotes the orthogonal projection onto the closed subspace
M of Hilbert space.
It can be seen that for every yϵ R(T), T†y = T −1y.

Definition 4.7:
Let T ϵ C( H 1 , H 2). The reduced minimum
39
modulus of T is defined by γ (T) = inf {||Tx|| : x ϵ C(T), ||x|| = 1}.

Proposition 4.8:
For a densely defined T ϵ C( H 1 , H 2), the following
statements (1) to (8) are equivalent.
1. R(T) is closed.
2. R(T ¿)is closed.
3. T 0 := T ¿C (T) has a bounded inverse.
4.γ (T) > 0.
5. T† is bounded.
6.γ (T) = γ (T ¿))
7. R(T ¿T) is closed.
8. R(TT ¿)) is closed.

Proposition 4.9:
Let Tϵ C ( H 1 , H 2) be densely defined. Then we have the
following.
1
1. If R(T) is closed, then γ (T) = ¿∨T † ∨¿

2. γ (T ¿T) = γ (T¿2

Proof:

Proof of (1):
The proof is true for the case of bounded operators .
In the case of a closed operator T also, the proof is similar .
For the sake of completeness we observe that

40
 ¿∨T † y ∨¿
||T†|| = sup { ≠ ϵ
¿∨ y∨¿ : 0 y D(T†) }
¿∨T † y ∨¿
= sup { ≠ ϵ
¿∨ y∨¿ : 0 y R(T) }
{ ¿∨Tx∨¿ : 0≠ x ϵ C(T) }
¿∨x∨¿
= sup

= inf ({ ¿∨Tx∨¿
≠ ϵ
¿∨x∨¿ : 0 x C(T) }¿ −1

= γ (T¿ .
−1

Proof of (2):
If γ (T) = 0,
then by the equivalence of (1), (4) and (7) in Proposition4.8,
R(T) is not closed and γ (T ¿T) = 0.
Next, let us assume that γ (T) > 0.
Again, by the equivalence of (1), (4) and (5) in Proposition 4.8,
R(T) is closed and T† is bounded,
so that and T†¿T† is also bounded.
Hence by (1) and using the fact that (T ¿T)† = T†T ¿†
we have
1
γ (T ¿T) = ¿
¿∨(T T )† ∨¿
1
=
¿∨T † ∨¿2
=¿∨T †∨¿ 2
= (T¿2.
This completes the proof.

Proposition 4.10:
Let T ϵ C ( H 1 , H 2) be a densely defined operator.

41
Then
1. N(T) = R(T ¿) ┴
2. N(T ¿) = R(T) ┴
3. N(T ¿T) = N(T) and
4. R(T´ ¿ T ) = R(T
´ ¿)

Proposition 4.11:
Let T ϵ C ( H 1 , H 2) be densely defined.
Let T^ = (I +TT ¿ ¿−1,Ť = (I + T ¿T¿−1.
Then
1. T^ and Ť are everywhere defined, bounded and self adjoint.
2. T ¿ T^ and T Ť are everywhere defined bounded and positive operators.
3. T^ T ⊆ TŤ and Ť T ¿ ⊆ T ¿ T^ .
4. || I − Ť || ≤ 1.
5. R(I − Ť ) = R(T ¿T).

Theorem 4.12:
For T ϵ C(H) , we have the following.
1. If μ ϵ C and λ ϵ σ (T), then λ + μ ϵ σ (T + μI).
2. If α ϵ C and λ ϵ σ (T), then σλ ϵ σ (α T).
3. σ (T 2) = { λ 2: λ ϵ σ (T)}

Proof:
The proof of the first two statements are similar to the proof
of the corresponding assertions for bounded operators .

For the proof of (3), let λ ϵ C be such that λ 2 ϵ T 2.

Then T 2−λ2 I ¿−1 ϵ B( H )

By definition this means that there exists a unique operator S λ ϵ B(H)

42
such that S λ ( T 2−λ2 I ) =I }D (T ) ⊆ ( T 2−λI ) S λ =I ¿
2

That is,
S λ ( T −λI )=I }D (T ) ⊆ ( T −λI ) ( T + λI ) S λ =I ¿
2

From the last equality, it also follows that T −λI is onto.

Next we show thatT −λI is one to one.
For this, let xϵ D(T) such that (T −λI )x = 0,
that is, Tx = λ x.
Then we have Tx ϵ D(T) so that
x ϵ D(T 2).
Applying T to both sides of the equation Tx = λ x,
we get T 2 x=λ 2 x ,
that is,( T 2−λ 2 I ) x=0
Since T 2−λ2 I is injective,
we have x = 0.
Thus, T − λ I is one to one as well.
Consequently,
λ ∉ σ (T).

For the other way implication,

let λϵ C be such that λ 2 ϵσ (T 2).
We have to show that either λ ϵσ (T) or −λ ϵσ (T).
Suppose this is not true.
Then both T − λ I and T + λ I have bounded inverses.
First we show that N(T 2−λ2 I ) = {0}.
For this, let xϵ D(T 2) ⊆ D(T) such that (T 2−λ2 I ) x = 0,
that is, (T − λ I)(T + λ I)x = 0.
Now, by the injectivity of T − λ I and T + λ I .
It follows that x = 0.
It remains to show that T 2− λ I is onto.
43
For this, let yϵ H.
Since T − λ I is onto, there exists u ϵ D(T) ⊆ H
such that (T − λ I)u = y.
As T + λ I is onto, there exists x ϵ D(T)
Such that (T + λ I)x = u.
Hence (T − λ I)(T + λ I)x = y,
that is, (T 2 − λ I)x = y.
This completes the proof of (3).

A spectral characterization of closed range operators :

Proposition 4.13:
Let T ϵ L(H) be a positive operator. Then the following
results hold.
1. T† is positive.
2. σ (T) \ {0} = σ (T 0) \ {0}, where T 0 = T}C(T ) ¿ C(T)
3. σ (T†) \ {0} = σ ¿) \ {0}
4. σ (T) = σ a (T) , where σ a (T) is the approximate spectrum of T.
5. 0 ∉ σ (I + T). i.e., ( I +T )−1 ϵ B(H).
1
6. Let λ ϵ (0 , ∞). Then λ ϵ σ ( T ) ⇔ λ ϵ σ (T†).
1
7. If T −1 exists, then 0 ≠ λ ϵ σ ( T ) ⇔ λ ϵ σ (T −1).

Proof :
Proof of (1):
Let T ϵ L(H) be a positive operator.
Then T is a self adjoint operator
44
 Using the relation T†* = T*† ,
It is easy to show that T† is self adjoint.
Let y = Tu + v , where u ϵ C(T)
And v ϵ R(T¿⊥
This is possible because D(T†) = R(T¿ ⊕⊥ R(T¿⊥
Using (2) and (4) in Definition maximal generalized inverse ,
we have ,
⟨ T † y , y ⟩ = ⟨ T † y , Tu+ v ⟩
¿ ⟨ T † y , Tu ⟩
¿⟨T T† y,u⟩
¿¿
¿ ⟨ Tu , u ⟩ ≥ 0
Proof of (2):
Since T is self adjoint,
It is reducible by N(T)
That is, T(D(T)∩N(T) ⊆N(T)
and T(D(T)∩N(T¿⊥ ⊆N(T¿⊥

Therefore, σ ( T )=¿
That is., σ ( T )={0}∪σ (T 0)
Hence, σ (T) \ {0} = σ (T 0) \ {0}.

Proof of (3):
Since T† is self adjoint,
It is reducible by R(T¿⊥ .
Now, using the fact that T†}R ( T ) ¿ = T −1
0

(2) implies that, σ (T†) \ {0} = σ (T −1

0 ) \ {0}.

Proof of (4):

45
 By definition σ a(T) ⊆ σ (T), and
Since T is self adjoint, σ (T)⊆ R.
Now, suppose λ ∉ σ a (T).
Then there exists a positive number k such that
||Tx − λ x|| ≥ k||x||, for all x ϵ D(T).
This shows that R(T − λ I) is closed and T − λ I is one to one .
´ I ) = H,
As T − λ I is self adjoint, we also have R(T −λ
by (2) in Proposition 4.10.
Hence, λ ∉ σ (T) .

Proof of (5):
Since T is positive, I +T is positive.
Now by (4), σ (I +T) = σ a (I +T).
If 0ϵ σ (I + T),
then there exists a sequence ( x n) ϵ D(T), with || x n || = 1,
for each n such that ||(I + T) x n|| → 0 as n → ∞
Now using Cauchy-Schwartz inequality,
it follows that⟨ ( I +T ) x n , x n ⟩ → 0 as n → ∞.
That is⟨ T x n , x n ⟩ → -1
On the other hand, since T is positive ⟨ T x n , x n ⟩ ≥ 0 for all n.
This is a contradiction.

Proof of (6):
Let λϵ (0 ,∞ ) be such that λ ϵ σ (T).
Then by (4), there exists a sequence x n ϵ C(T) with || x n || = 1 for all n such that
||T x n − λx n || =||T 0 x n − c x n|| → 0 as n → ∞.
Let y n=T 0 (x n ) .
Since T 0 : C(T) →R(T) is bijective, x n=T −1
0 ( yn )

46
Thus ‖ y n −λ T −1
0 ( y n)‖→ 0.

Also for large valuesof n, || y n|| is close to | λ | and hence non zero.
n y
Now taking z n= y ,
‖ n‖
−1 1
we have ||T 0 z n− λ z n|| →0 as n → ∞.

This together with (3), we have

1
ϵ σ ¿ \ {0}) =σ (T†) \ {0}.
λ
1
If ϵ σ (T†),
λ
then by the above argument it follows that
1
ϵσ
1 (T††).
λ
But, it is known that T†† = T.
Thus, we have λ ϵ σ (T).

Proof of (7):
Since T −1 exists,
we have T −1 = T −1
0 = T†.
Hence, the resulst follows from (5).

Proposition 4.14:
Let T : H 1 → H 2 be a (possibly unbounded ) densely
defined, closed linear operator. Let T̆ =¿ and A = I − T̆ .
Then
λ ¿
1. σ ( A ) ={ 1+ λ : λ ϵ σ (T T )}

¿ μ
2. σ (T T )={ 1−μ : μ ϵ σ (A )}

Proof:

47
 By Theorem 4.12 and proposition 4.13, we have μ ϵ σ ( A )if and
only if there exists a
1 λ
λ ϵ σ (T ¿ T ) such that μ = 1 − =
λ+1 λ+1
Since I−A = (I+ T ¿ T ¿−1 is bijective, again by Theorem 4.12 and
Proposition 4.13,
λ ϵ σ (T ¿ T ) if and only if there exists a
1 μ
μϵ σ (A) such that λ = 1−μ -1 = 1−μ

This completes the proof.

Theorem 4.15:
Let Tϵ C ( H 1 , H 2) be densely defined. Then R(T) is closed
in H 2 if and only if there exists a r > 0 such that
σ (T ¿ T ) ⊆ {0}∪[r,∞ ).

Proof:
By Proposition 4.8 and Proposition 4.9 ,
R(T) is closed ⇔R(T ¿ T ) = R(A) is closed,
where A := I − (I + T ¿ T ¿−1 is a self-adjoint operator.
Since A is a bounded operator.
Hence by Theorem4.12,
R(A) is closed
⇔ 0 is not an accumulation point ofσ (A ¿ A)= σ ¿)

⇔ 0 is not an accumulation point of σ (A).

Now by above proposition , 0 is not an accumulation point of σ (A)

⇔ 0 is not an accumulation point of σ (T ¿ T ),

Equivalently , there exists an r > 0such that

σ (T ¿ T ) ⊆ {0}∪[r,∞ ).

48
Example 4.16:

Let H = l 2.
Define T on H by
T( x 1 , x 2 , x 3, . . . , x n, . . . , ) = ( x 1 , 2 x2 , 3 x3, . . . , nx n, . . . ,. )
whose domain is
∞

D(T) = {( x1 , x2 , x3 xn
, . . . , , . . . , )ϵ H : ∑
j=1
2
| j x j| < ∞ }

Clearly, T is unbounded and closed

Since T* = T.
And since C 00 ⊆D(T),
´ )= H.
wehave D(T
We claim that R(T) is closed.
Here N(T) = N(T*) = {0}.
Hence C(T) = D(T).
Also note that ||Tx|| ≥ ||x|| for all x ϵ D(T) = C(T).
Hence by Proposition 4.8, R(T) is closed.
Here γ (T) = 1.
Now σ e(T) ={n: n ϵ N}.
We claim that σ (T) = σ e (T).
Suppose λ ≠ n, for all nϵ N.
Then there exists an η> 0 such that | λ − n| ≥ η, for all n ϵ N.
Define S λ : H → H by
x1 xn
S λ ( x )=( ,……., )
λ−1 λ−n
Then S λ is bounded,
1
|| S λ|| ≤ η

49
And S λ is the inverse of λ I − T. Here d(T) = 1.
Similarly we can prove that σ (T T ) =σ (T 2 ) = {n2 :n ϵN }.
¿

CHAPTER -5

THE SPECTRAL THEOREM

Definition 5.1:

Let T be an operator in H. The resolvent set of T denoted by ρ ( T ) , is the

set of those λϵC scuh that T- λ I : D(T) → H is a bijection and

(T −λ I )−1 ϵ B(H ).

That is, λ ϵ ρ ( T ) if and only if there is some Sϵ B( H ) such that

S(T- λ I) ⊂ ( T −λ I ) S=I

We call R : ρ ( T ) → B( H ) defined by
50
 R(λ) = (T −λ I )−1 the resolvent of T.

Definition 5.2:

The spectrum of T is denoted as σ (T ) . It is defined as

σ ( T )=C }ρ( T ) ¿

It is a fact that ρ ( T ) is open, that σ ( T ) is closed, and that if σ ( T ) ≠ C then T is a

closed operator, that
R(z) - R(w) = (z - w)R(z)R(w); z , w ϵρ ( T )
and
dn R ( ) n+1
n
z =n ! R ( z ) , z ϵρ ( T ) .
dz
If T is a self-adjoint operator in H, one proves that σ ( T ) ⊂ R.

Definition 5.3:

Let (Ω , ℘) be a measurable space. A resolution of the identity is a function

E : ℘ →B(H)
satisfying:
1. E(∅ ) = 0, E(Ω ) = I.
2. For each a ϵ ℘ , E(a) is a s elf-adjoint projection.
3. E(a ∩b) = E(a)E(b).
4. If a ∩b =∅ then E(a ∪ b) = E(a) + E(b).
5. For each x , yϵ H, the function E x , y :℘→C defined by
E x , y (a) = ⟨ E ( a ) x , y ⟩ ,aϵ ℘

is a complex measure on ℘.

51
Definition 5.4:

Let {Di} be a countable collection of open discs that is a base for the
topology of C,
i.e., ∪ Di = C and for each i ,j and for z ϵ Di ∩ D j ,
there is some k such that x ϵ Dk ⊂ Di D j .
Let f :( Ω, ℘)→ C , BC be a measurable
function and let V be the union of those Di for which E( f −1( Di)) = 0.
Then E( f −1(V)) = 0.
The essential range of f is C\ V, and we say that f is
essentially bounded if the essential range of f is a bounded subset of C.

We define the essential supremum of f to be

‖f ‖∞=⁡{| λ|: λ ϵ C ¿ }

Definition 5.5:

A subset A of B(H) is said to be normal when ST = TS for all S,T ϵ A

and T ϵ A implies that T ¿ ϵ A.

Note :
If T ϵ B(H) is normal means that TT ¿= T ¿T .

Theorem 5.6:
If (Ω , ℘) is a measurable space and E : ℘ →B(H)
is a resolution of the identity, then there is a closed normal subalgebra A of
B(H) and a unique isometric *-isomorphism Ψ : L∞ ( E) → A such that

⟨ Ψ ( f ) x , y ⟩ =∫ fd E x, y , fϵ L∞ ( E ) , x , yϵ H .
Ω

Furthermore,

52
 2
‖Ψ ( f ) x‖ =∫|f|2 d E x , x , fϵ L ∞ ( E ) , xϵ H
Ω

∞
For fϵ L ( E ) , we define

∫ fdE = Ψ ( f )
Ω

For L∞ ( E ) , σ (Ψ ( f ) ) is equal to the essential range of f

The spectral theorem :

The following is the spectral theorem for self-adjoint operators
Theorem 5.7:
If T is a self-adjoint operator in H, then there is a unique resolution E
of the identity , on the borel subsets of the σ (T ), so that

⟨ Tx , y ⟩ = ∫ λ d E x , y ( λ ) xϵD ( T ) , yϵH -------------1

σ (T )

Proof :
Let U be t he Cayley transform of A ,
Le t Ω be the unit circle with a point 1 removed and let E' be the spectral
decomposition of U
Since I-U is one to one
E' ({1 }) = 0

Therefore (U x ,y) = ∫ λ d E' x, y ( λ ) xϵH , yϵH -----------2

Ω

Define

i(1+ λ)
f ( λ )= λ ϵ Ω ---------------------3
1−λ

And define Ψ ( f )as in a theorem with E' in place of E

( Ψ ( f ) x , y ¿ =∫ fd E' x , y xϵD ( T ) , yϵH ---------------4

Ω

53
Since f is real valued, Ψ ( f ) is self - adjoint

And since f ( λ ¿(1- λ ¿=i ( 1+ λ ) , the multiplication theorem gives

Ψ (f) (1-U ¿=i ( 1+U ) , ----------------5

In particular , (5) implies that

R(1-U ¿ ⊂ D(Ψ ( f ) )

By a theorem T(1-U ¿=i ( 1+U ) ---------------6

And D(T) = R(1-U ¿ ⊂ D(Ψ ( f ) )

Comparison of (5) and (6) shows that Ψ ( f ) is a self adjoint extention of the
self-adjiont operator T.

By using a theorem T = Ψ ( f )

Thus

⟨ Tx , y ⟩ =∫ fd E' x, y xϵD ( T ) , yϵH -----------7

Ω

Since , σ (T) is the essential range of f.

Thus , σ ( T ) ⊂(−∞ , ∞)

We know that f is one to one in Ω .

If we define E ( f ( ω ) )=E ' ( ω)-------------8

for every borel set ω ⊂ Ω . we obtain the desired resolution E which

converts ( 7) to(1)

just as (1) was derived from (2) by means of the Cayley transform,

54
(2) can be derived from (1) by using the inverse of the Cayley transform.

The uniqueness of the representation (2) leads therefore to the uniqueness of the
resolution E satisfies (1)

This compltes the proof.

CONCLUSION

Unbounded operators is an upcoming area in Functional

Analysis . Most of the definition and proof Unbounded operators
are given in this thesis. This assertion is subject to know about the
basic fundamentamental. The spectral theorem applies to self-adjiont
operators and moreover to normal operators but not to densely

55
defined operators in general, since in this case spectrum can be empty.
We can get deep knowledge about “Unbounded operators” in coming
years.

BIBILIOGRPHY

 Balmohan V Limaye : “ Functional Analysis” Ιιι Edition., New Age

International Publishers.
 S. Ponnusamy : “ Foundations Of Functional Analysis” .,
Crc Press,2002.
 Martin Schechter : “Principles Of Functional Analysis” Ιι Edition.,
American Mathematical Society Providence , Rhode Island.
 Walter Rudin : “Functional Analysis ” Ιι Edition., Tata Mcgraw-Hill
Education Private Limited, New Delhi.
 B.Choudhary And Sudarsan Nanda : “Functional Analysis With
Applications” John Wiley & Sons, Inc , 1989.

56
 Dr . Sudhir Kanna Pundir : “Functional Analysis” Cbs Pulishers &
Distributions Pvt.Ltd.
 “Unbounded Operators In A Hilbert Space And The trotter Product
Formula” Jordan Bell Jordan.Bell@Gmail.Com Department Of
Mathematics, University Of Toronto August 25, 2015
 “The Sum Of Two Unbounded Linear Operators:
Closedness, Self-Adjointness And Normality”
Mohammed Hichem Mortad Department Of Mathematics, University Of
Oran, B.P. 1524, El Menouar, Oran
 “ Some Properties Of Unbounded Operatorswith Closed Range”
S. H. Kulkarni, M. T. Nair And G. Ramesh(March 3, 2007)
 “The Spectral Theorem For Unbounded Normal Operators”
S. J. Bernau

57