CHAPTER – 2

BANACH ALGEBRA

DEFINITION 2.1

A Banach algebra A is a complex Banach space which is

also an algebra with the identity and in which the multiplicative structure
is related to the norm as follows
(i) ‖xy‖ ≤ ‖x‖‖ y‖

(ii) ‖e‖=1

THEOREM 2.1

In any Banach algebra A, the multiplication is jointly

continuous.
ie) If x n x and y n y then x n y n xy .

PROOF

Let x n x
Given   0, there exists a positive integer n1 such that
ε
‖x n x‖< 2‖ y‖ for all n ≥ n1.

Since x nis a convergent sequence. We have,

“ Every convergent sequence is bounded sequence.”
Given   0, there exists a positive integer n such that
 x n  ≤ M for all n where M is a constant.
Since yn y , given   0, there exists a positive integer n2 such
 1
that‖ y n − y‖ ≤ 2 M for all n ≥ n2.

Let n 0=max {n 1 , n2 }.
Now ‖x n y n−xy‖ = ‖x n y n−x n y+ x n y−xy‖
=‖x n ( y n− y ) + y ( x n−x )‖

≤ ‖x n‖‖ y n− y‖+‖ y‖‖ x n−x‖

1 1
≤M 2M
+ ‖ y‖ 2‖ y‖
❑+❑
< 2 2 =

Therefore ‖x n y n−xy‖ < 

Hence x n y n xy as n   .
Therefore In any Banach algebra A, the multiplication is jointly
continuous.

DEFINITION 2.2

Let A be a Banach algebra with an identity element e. The

elements of A having inverses are called Regular elements or units of A.

The Regular elements are often called invertible elements
or non – singular elements.
Let G denote the set of all regular elements. Since e is

regular, e ϵ G and G is multiplicative group.

DEFINITION 2.3

The elements of the Banach algebra A which are not regular

are called Singular. Let S be the set of all singular elements.

Since 0 is singular, 0 ϵ S.

NOTE 2.1

From DEFINITION 2.2 and 2.3, it is obvious that the

complement of G is S.

THEOREM 2.2

Let A be a Banach algebra with identity. If ‖e−x‖ < 1,

then x is regular and the inverse of x is given by the formula
❑

= e + ∑ (e−x ) .
−1 n
x
n =1

PROOF

Let us take ‖e−x‖ = r. By hypothesis, r  1.

Using this hypothesis we have ‖(e−x)n‖ ≤ ‖e−x‖n = r n .
❑

Hence the series ∑ (e−x ) converges absolutely in the norm.

n

n =1

ie) sequence of partial sums forms a Cauchy sequence in A.

Since A is complete, the sequence of partial sums converges to an element
of A.
❑

Let us define y = e + ∑ (e−x ) .

n

n =1

Now, y – xy = ( e – x ) y
❑

= ( e – x ) + ∑ (e−x )
n

n =2

= ∑ (e−x )
n

n =1

=y–e
Hence, xy = e. Similarly, yx = e.
Therefore y is the inverse of x.
❑

Therefore x = e + ∑ (e−x ) .
−1 n

n =1

Hence the proof.

THEOREM 2.3

The set G is an open set and therefore S is a closed set.

PROOF

Let G be the set of all regular elements.

To prove that G is open.
It is enough to prove that for any x є G, there exist an open sphere
S ( x,  )  G.
Since e є G,

Let us consider the open sphere S (e, 1 ) = { x :‖x−e‖< 1 }.

By theorem 2.2,
For each element x in S ( e, 1) is regular,
we have S ( e, 1)  G.
If x is regular, we have x x−1 = e є S ( e, 1)
Since the multiplication is continuous,
there exist an  neighbourhood S ( x,  ) of x such that
S ( x,  ) x−1  S ( e, 1)  G,
where S ( x,  ) x−1 ={ y x−1 : y є S (x ,) }.
Hence for any y є S ( x , ) we have y x−1 є G.
So that y x−1 is regular, there exists z є G such that
 ( y x−1) z = z ( y x−1) = e
y( x−1z ) = ( x−1z )y = e
Hence y is regular whose inverse is x
−1
є G.
Therefore y є G.
Since x є G is arbitrary, every open sphere S ( x,  ) contains only points
like y from G.
Hence every point y є G is the centre of some open sphere contained in G.
Therefore G is open.
Let S be the set of all singular elements and it is the complement of G.
Therefore S is closed set.
Hence the proof.

THEOREM 2.4
The mapping x → x−1 of G into G is continuous and is
therefore a Homeomorphism of G onto itself.
PROOF

Let x 0 є G and
Let{ x n } be a sequence in G such that xn → x0 as n
To prove x → x−1 is continuous.
It is enough to prove x−1 −1
n → xo

n −x n ‖ = ‖x n ( x 0−x n ) x 0 ‖
Now, ‖x −1 −1 −1 −1

≤ ‖x −1
n ‖‖ x 0−x n‖‖x 0 ‖
−1

Hence ‖x −1
n −x 0 ‖ ≤ ‖x n ‖‖ x 0−x n‖‖x 0 ‖
−1 −1 −1
(2.1)
Using the hypothesis,
we find the upper bound for each term on the right hand side of (2.1)
Since x n → x 0,
Given   0 there exists a positive integer n 0
1
such that n ≥ n0 , ‖x n−x 0‖ < 2‖ x−1‖ (2.2)
0

1
where we have taken  = 2‖ x−1‖
0

Now ‖(e−x−1
0 x n )‖ = ‖x 0 (x 0−x n )‖
−1

≤‖x −1
0 ‖‖ x 0−x n‖ (2.3)

1
Using (2.2) in (2.3), we get ‖(e−x−1
0 x n )‖ <
2
(2.4)

So by THEOREM 2.2 ,
−1
x0 xn is regular and its inverse is given by
−1 −1 −1
x n x 0 =( x 0 x n)
∞

= e + ∑ (e−x−1
0 x n)
n

n =1

∞
n
Thus ‖x −1
n x 0‖ ≤ 1 + ∑ ‖( e−x−1
0 x n )‖
n =1

1
≤ 1−‖e−x−1 x ‖
0 n

≤2 by (4)
This gives ‖x −1
n x 0‖ < 2

So that we have ‖x −1
n ‖ = ‖x n x 0 x 0 ‖
−1 −1

≤ ‖x −1
n x 0‖‖x 0 ‖
−1

= 2 ‖x −1
0 ‖ (2.5)
From (2.1) and (2.5) we get,
‖x −1
n −x 0 ‖ ≤ 2‖x 0 ‖‖ x 0−x n‖‖x 0 ‖
−1 −1 −1

 0 as n
n −x o  0 as n 
Hence x−1 −1
Hence x−1 −1
n → xo

Therefore x → x−1 of G into G is continuous.

Hence the mapping x→ x
−1
is a continuous open mapping of G onto
Itself.
Therefore it is a homeomorphism of G onto itself.
Hence the proof.

THEOREM 2.5
Let x є A and λ be a scalar. Then if |λ| > ‖x‖, (λe – x ) is
❑

regular element and ( λe – x)−1 = ∑ λ

−n n−1
x
n =1

PROOF

Let x є A and λ be a scalar

x
Now, (λe – x ) = λ (e - λ
)

To prove (λe – x ) is regular.

x
ie) To prove λ (e - λ
) is regular.

Now, ‖ ( )‖ = ‖ ‖ < 1 by the hypothesis

e− e−
x
λ
x
λ

x
Hence by THEOREM 2.2, (e - λ ) is regular.

So taking inverses on both sides of (1),

We obtain
−1
−1 −1 x
(λe – x) =λ (e− )
λ

Since ‖ ( )‖
e− e –
x
λ
<1

By THEOREM 2.2, we obtain

 [ { ( )} ]
∞ n
x
(λe – x)−1 =λ−1 e + ∑ e− e –
n=1 λ

[ ( )]
∞ n
x
= λ−1 e+ ∑
n =1 λ

Taking e = x 0
We can rewrite the sum on the right hand side as
∞
(λe – x)−1 =∑ λ−n x−n
n=1

Hence the proof.

DEFINITION 2.4
An element z in a Banach algebra A is called a topological
divisor of zero if there exists a sequence { z n } in A such that ‖z n‖ = 1
and either z z n  0 or z n z  0 .

NOTE 2.2
It is clear that every divisor of zero is also a topological
divisor of zero. We denote the set of all topological divisors of zero
by Z.

THEOREM 2.6
Z is a subset of S.
PROOF

Let Z be the set of all topological divisors of zero.

Let S be the set of all singular elements.
Let z be an element of Z and
Let { z n } be a sequence such that ‖z n‖ = 1 and z z n 0
To prove z is in S.
Suppose z is not in S then z is in G.
By the joint continuity of multiplication
We would have z−1( z z n ) = z n  0
Which is a contradiction to ‖z n‖ = 1
Therefore z is in S
Therefore z is the subset of S
Hence the proof.

THEOREM 2.7
The boundary of S is a subset of Z.
PROOF

Let S be the set of all singular elements and

Let Z be the set of all topological divisors of zero.
Since S is closed.
Therefore S is the boundary consists of all points in S which are
limits of convergent sequences in G. where G is the set of all regular
elements.
If z is in S and there exists a sequence {r n } in G such that r n  z,
then z is in Z.
Now r n−1z – 1 = r n−1( z - r n )
Therefore { r n−1 } is unbounded
Otherwise, ‖r n−1 z−1‖ < 1 for some n
So that r n−1 z and therefore z = r n (r n−1 z ) would be regular.
Since { r n−1 } is unbounded
We may assume that ‖r n−1‖  
−1
rn
If z n = then ‖z n‖ = 1
‖r n−1‖
−1
z rn
And z z n =
‖r n−1‖
−1
1+(z−r n)r n
=
‖r n−1‖
1
= ‖r −1‖ + ( z - r n ) z n
n

0
Therefore the boundary of S is a subset of Z.
Hence the proof.

NOTE 2.3
The topological divisors of zero in A are permanently
Singular.
 CHAPTER – 3

THE SPECTRUM

DEFINITION 3.1
Let B be a Banach algebra with identity e and Let x є B.
The spectrum of x denoted byσ (x) is the subset of the complex plane
defined by
σ (x) = { λ є C : ( x− λe ) is singular }

DEFINITION 3.2
The Resolvent of x is a function defined on ρ (x) with
values in A. That is if λ є ρ (x), then the Resolvent of x is
defined as
x(λ) = ( x−λe )−1

NOTE 3.1
ρ (x) is defined as the compliment of σ (x).

NOTE 3.2
If λ , μ є ρ (x), then we have
x(λ) – x(μ) = ( λ – μ ) x(λ) x(μ)
this equation is called the resolvent equation.

THEOREM 3.1 ( Liouville’s theorem for analytic

vector valued functions )
STATEMENT
Let B be a Banach space and let x(λ) be defined as x : C  B
Where λ  x(λ). If x(λ) is an entire function ( analytic throughout C )
and if there is a constant M > 0 such that ‖x (λ)‖ ≤ M for all λ є C
then x(λ) must be a constant.
PROOF

Let B be a Banach space and

x(λ) be defined as as x : C  B where λ  x(λ).
Given x(λ) is an entire function and if there is a constant M > 0
Such that ‖x (λ)‖ ≤ M for all λ є C
If f є B¿
Then f { x ( λ)} is an entire function on C to C
We have |f { x( λ)}| ≤ ‖f ‖‖x ( λ)‖
≤ ‖f ‖ M
Therefore f is bounded entire function
By classical liouville’s theorem,
|f { x( λ)}| must be a constant
Let α and β be the two complex numbers in C
We have f { x ( α ) } = f { x ( β ) }
Since f is linear
Therefore f { x (α )} - f { x (β )} = 0
¿
Since f is an arbitrary element of B

By corollary
“ If all functionals vanish on a given vector, then the vector must
be zero”
Therefore x (α ) - x (β) = 0
Therefore x (α ) = x (β)
Therefore x (λ) must be a constant
Hence the proof.

THEOREM 3.2
σ (x) is a compact subset of the complex plane C
PROOF

Let C be a complex plane

To prove σ (x) is a compact in C
First we prove σ (x) is a closed subset of the complex plane.
c
Let [ σ (x)] be the complement of σ (x)
c
To prove [ σ (x)] is open
ie) To prove σ (x) is closed
let λ 0  [ σ (x)]
c

Then xλ =
0 x - λ 0e is regular
Therefore x λ є G 0

Since G is open, If there exist a open sphere S ( x λ , ) ⊆ G 0

The function λ  x – λe from C to the Banach algebra A is a continuous

function.
If λ n  λ0 implies x - λ n e  x - λ 0e
Hence for any given neighbourhood S ( x λ , ) of x λ , there exist a
0 0

neighbourhood S ( λ 0 , ) of λ 0 such that

xλ є S ( x λ , δ )  G for all λ є S ( λ 0 , δ ).
0

Hence for all such λ, xλ єG

Therefore all the points of S ( λ 0 , δ ) are regular
c
Hence given an arbitrary point λ0 є [ σ (x)] , there exist an open sphere
S ( λ 0 , δ )  [ σ (x)]
c

c
Therefore [ σ (x)] is open
Therefore σ (x) is closed
Next we prove σ ( x )is a subset of the closed disc { z :|z|≤‖ x‖}in
the complex plane
Let us assume that λ is complex number such that |λ| > ‖x‖

‖ x‖
Then λ < 1

Therefore ‖ ( )‖ < 1
e− e−
x
λ

By lemma ,
“ Let A be a Banach algebra with identity then if ‖e−x‖≤ 1
Then x is regular.”

Therefore ( e− xλ ) is regular
Therefore ( x−λe ) is regular
Which is contradiction to ( x−λe ) is singular
Therefore our assumption is wrong
Therefore |λ| ≤ ‖x‖
Therefore σ (x)  S[ 0 ,‖x‖]
Therefore σ (x) is a subset of the closed disc in the complex plane C.
Therefore σ (x) is closed and bounded subset of the complex plane.
Therefore σ (x) is compact.
Hence the proof.

THEOREM 3.3
If A is a Banach algebra and x є A then σ (x)
is non empty.
PROOF

Let A be a Banach algebra.

Let x є A
Let f be a element of A¿ where A
¿
is a conjugate space of A.
¿
ie) f є A

Let us define the function f(λ) = f{ x ( λ) }

Where λ belongs to the resolvent set ρ (x)
To prove σ (x) ≠ ϕ
Since f is continuous and x( λ) is also continuous, f(λ) is a continuous
complex valued function defined on the resolvent set ρ (x).
since f is linear, we have from the resolvent equation.
f[ x ( λ)] - f[ x ( μ)] = ( λ - μ ¿ f[ x ( λ ) x(μ)]
f ( λ )−f (μ)
λ−μ
= f[ x ( λ ) x(μ)]

Taking limit on both sides

f ( λ )−f ( μ) f [ x ( λ ) x ( μ) ]
lim
λ−μ
= lim
λμ
λμ
 f ( λ )−f ( μ)
lim
λ−μ
= f(x ( μ ))2
λμ

Therefore f(λ) has derivative at each point of A of the resolvent set ρ (x).
By the definition of f(λ), we have
|f ( λ)| = |f [ x ( λ) ]|
≤ ‖f ‖‖x ( λ)‖
A , ‖f ‖
¿
Since f є is bounded.
Therefore, f(λ)  0 as λ  
Let us assume that σ (x) ≠ ϕ
So that ρ (x) is the entire complex plane.
Hence f(λ) is an entire function in the whole complex plane
such that f(λ)  0 as λ  
Therefore f( λ ¿ is a bounded entire function.
By Liouville’s theorem from complex analysis,
f(λ) must be a constant.
Since f(λ)  0 as λ  
therefore f(λ) = 0 for all λ є ρ (x)
ie) f[ x ( λ)] = 0 for all λ є ρ (x)
¿
since f є A is arbitrary
we have, If all functionals vanish on a given vector, then the vector must
be zero.
Therefore x(λ) = 0 for all λ є ρ (x)
But x(λ) = ( x−λe )−1 so that ( x−λe )−1=0
Now, e =( x−λe ) ( x−λe )−1
e=0
Which is contradiction to our assumption
Therefore σ (x) ≠ ϕ for all x є A
Therefore σ (x) is non empty.
DEFINITION 3.3
A Division algebra is an algebra with identity in which each
non – zero element is regular.

THEOREM 3.4
If A is a division algebra, then it equals the set of all scalar
multiples of the identity.
PROOF

Given A is a division algebra.

Let e be the identity element.
Let x є A
To prove x = λe for some scalar λ
Suppose x≠ λe for every λ
Therefore x - λe≠ 0 for every λ
Hence ( x−λe ) is regular for every λ and x є A
Therefore σ (x) ≠ ϕ
Which is a contradiction to σ (x) ≠ ϕ
Therefore x = λe for some scalar λ
Therefore x is equals to the set of all scalar multiples of the identity.
Hence the proof.

COROLLARY 3.4.1
Any Banach algebra A which is a division algebra equals c
in the sense of isometric isomorphism.
PROOF

Given A is a division algebra

Therefore by the theorem, for every x є A is a unique scalar multiple
of the identity.
So that x = λe
Now, consider the mapping f : A  C by f(x) = λ
Therefore f( λe) = e
Therefore f is one to one
If x≠ y then λ≠ μ where y  μ
If λ = μ then λe = μe, x = y where y =μe
To prove f is Linear.
f( x + y ) = f( λe + μe )
= f(( λ + μ ) e)
=λ+μ
= f(x) + f(y)
f(αx ¿ = f( α λe )
= α f( λe )
=αλ
= α f(x) for all α is a scalar
Therefore f is linear
For any scalar λ the vector λe maps into λ
So that f is onto
Hence f is an isomorphism
Since f is easily seen to be structure preserving.
It is isometric for ‖x− y‖ = |f ( x ) −f ( y )|
Now, |f ( x ) −f ( y )| = |λ−μ| (3.1)
But ‖x− y‖ = ‖λe−μe‖
= |λ−μ|‖e‖
‖x− y‖ = |λ−μ| (3.2)
From (3.1) and (3.2) we get
‖x− y‖=,|f ( x )−f ( y )|

Therefore A isometrically isomorphic to C

Therefore A = C
Hence the proof.

THEOREM 3.5
If 0 is the only topological dvisor of zero in A then A = C.
PROOF

Let 0 be the only topological divisor of zero in A

To prove A = C
Let x be any element of A
We know that σ (x) is non empty, then it has boundary point λ ( say )
Now we prove that ( x−λe ) is a boundary point of the sets S of all
singular elements.
Since λ is a boundary point of σ (x)
Then there exists a sequence ( λ n) in σ (x) such that λ n  λ
Note that each of ( x - λne ) is singular.
So that each of ( x−λe ) is a boundary point of S
By theorem 3.4
Therefore x−λe is a topological divisor of zero.
Since 0 is a topological divisor of zero.
Therefore x−λe = 0
Therefore x=λe
By corollary 3.4.1,
A = C in the sence of isometric isomorphism.
Therefore A = C
Hence the proof.

THEOREM 3.6
If the norm in A satisfies the inequality ‖xy‖≥ k ‖ x‖‖ y‖
For some positive constant k, then A = C.
PROOF

Given the norm in A satisfies the inequality

‖xy‖≥ k ‖ x‖‖ y‖ for some positive constant k.

To prove A = C
By theorem 3.5, we get
“ 0 is the only topological divisor of zero in A then A = C “
Let z be another topological divisor of zero in A
This implies there exists a sequemce ( x n ) such that ‖x n‖ = 1 and
z z n or z n z  0 as n  
by given hypothesis, we get
‖z z n‖ ≥ k ‖z‖ ‖z n‖
When n   , we get k ≤ 0
Which is a contradiction to k is a positive constant
Our assumption is wrong
Therefore 0 is the only topological divisor of zero then A = C.
Therefore A = C
Hence the proof.

THEOREM 3.7
If A is a Banach sub algebra of a Banach algebra A,
'
Then the
spectra of an element x in A with respect to A and A
'
are related as follows
(i) σ A (x) ⊆ σ A (x)
'

(ii)Each boundary point of σ A (x) is also a boundary point of

σ A (x).'

PROOF

(i) Given A is Banach sub algebra of a Banach algebra A

'

Let A  A
'

Let x є A
If ( x−λe ) is singular in A
'

Therefore ( x−λe ) is also singular in A

Assume ( x−λe ) is not singular in A
Therefore ( x−λe ) is regular in A
Since x є A
'

Also ( x−λe ) is regular in A

'

Which is a contradiction to ( x−λe ) is singular in A

'

Hence ( x−λe ) is singular in A

'

Therefore ( x−λe ) is also singular in A

Therefore σ A (x) ⊆ σ A (x)
'
 (ii) Let λ be a boundary point of σ A (x)
Then ( x−λe )is a boundary point of the set of singular elements
in A
We have “ The boundary of S is a subset of Z ”
Therefore ( x−λe ) is a topological divisor of zero in A
( x−λe )is also a topological divisor of zero in A
'

Therefore ( x−λe ) is singular in A

'

Therefore ( x−λe ) is singular in A

'
and λ є σ A (x)
'

Since σ A (x) ⊆ σ A (x)

'

Therefore λ is a boundary point of σ A (x) '

Each boundary point of σ A (x) is also a boundary point of σ A (x).

'

Hence the proof.

THE FORMULA FOR SPECTRAL RADIUS

DEFINITION 3.4
Let x be an element in our general banach algebra A, and
consider Its spectral radius r(x), which is defined by
r(x) = sup {| λ|: λ є σ A ( x ) }

THEOREM 3.8
Let A be a complex banach algebra with identity and x є A.
Let p(t) be a polynomial with constant complex coefficients. Then
[ p(x )] = p[ σ (x)].
PROOF
 Let p(t) be a polynomial with constant complex coefficients.
Let p(x) є A
So that σ[ p(x )] is well defined and p[ σ (x)] = { p ( λ ) : λ є σ (x)}
Taking up the case of general nth degree polynomial,
Let us prove the theorem when p(t) = a 0 and p(t) = at.
Now σ[ p(x )] = σ(a 0 e ¿
= { λ : ( a0 e−λe ) is singular }
When λ = a 0
⇒ a0 e−a0 e=0 so that it is singular and λ= a 0 is the only value for
Which (a 0 e− λe ¿ is singular.
Hence we get σ[ p ( x ) ] = a 0, a≠ 0.
When we put p(t) = at
To prove σ(ax) = a σ(x)
If a = 0
The result is true.
If a≠ 0 then λ є σ(ax) ⇔ ax – λe is singular
λ
ie) x - a
e is singular
λ
⇒
a
є σ(x)
⇒ λ є a σ (x)

Therefore σ(ax) = a σ(x)

To prove the theorem in the general case, consider the polynomial p(t)
With the leading coefficient 1.
Let p(t) = a 0+ a1 t+. . . .+ an−1 tn−1 +t n
Now consider p(t) – λ for λ є C
Since C is algebraically closed we can factorize p(t) – λ
So that p(t) – λ = (t - b 1 ) (t - b 2) . . . . (t -b n) (3.3)
p(x) – λe = ( x - b 1e ) ( x - b 2 e ) . . . . ( x - b ne ) (3.4)
since λ є σ[ p(x )], one of the factors in the right hand side of (3.3) is
singular.
Let it be x - b je.
Since ( x - b je ) is singular, b j є σ(x).
Therefore p(b j ¿ є { p ( λ ) : λ є σ (x)} = p[ σ (x)] (3.5)
Using b j in (3.3), we get p(b j ¿ – λ = 0 or p(b j ¿ = λ (3.6)
From (3.5) and (3.6), we get λ є p[ σ (x)]
Thus we have proved that if λ є σ[ p ( x ) ], then λ є p[ σ (x)].
Hence we get σ[ p ( x ) ]  p[ σ (x)] (3.7)
To complete the proof, we prove the reverse inclusion.
Let λ є p[ σ (x)], then by the definition of p[ σ (x)], there exists a c j є σ(x)
such that λ є p(c j ¿ .
Now consider the polynomial,
p(t) – λ = (t - c 1 ) (t - c 2) . . . . (t -c j) . . . . (t -c n)
where c 1 , c 2 , . . . . ,c n are the roots b 1 , b2 , . . . . , b n in (3.3) in some order
or other.
From the above equation, we get
p(x) – λe = (x - c 1 e ) (x - c 2 e) . . . . (x -c j e ) . . . . (x -c n e) (3.8)
If λ є σ[ p ( x ) ], then p(x) – λe is regular so that we have
[ p(x )– λe ][ p (x) – λe ] = e
−1

= [ p(x )– λe ] [( x−c 1 e )(x−c 2 e). . . .(x−c n e)] (x -c j e )

−1
(3.9)
Therefore (3.9) shows that (x -c j e ) has a left inverse
Similarly it shows that (x -c j e ) has a right inverse
Therefore (x -c j e ) is regular.
 Which is a contradiction to c j є σ(x)
Therefore λ є σ [ p ( x ) ]
Therefore p[ σ (x)]  σ [ p ( x ) ] (3.10)
From (3.7) and (3.10) we get
p[ σ (x)] = σ [ p ( x ) ]
Hence the theorem.

COROLLARY 3.8.1
n
For any x є A, σ( x n) = [ σ (x)]
PROOF

Let p(x) = x n.
Hence from the above theorem
σ [ p ( x ) ] =σ (x )=¿ p[ σ (x)] = [ σ (x)] .
n n

n
Thus we have σ( x n) = [ σ (x)] .
Hence the proof.

THEOREM 3.9
1

If x is an element of a Banach algebra, then r(x) = lim ‖x n‖n .

n

PROOF

n
First we prove that r( x n) = [ r (x )]
Now r( x n) = sup {| λ|: λ є σ ( x n ) }
Using the corollary of theorem 1 in the above
r( x n) = sup {|λ|: λ є [ σ ( x)] }
n
 n
Since λ є [ σ (x )] , there is a μ є σ(x) such that |λ| = |μ|n
Hence r( x n) = sup {|μ|n :μ є σ ( x ) }
= sup {|μ|: μ є σ (x)n }
n
= [ r (x )]
n
Therefore we have esatablished that r( x n) = [ r (x )] (3.11)
Also we have r( x n) ≤ ‖x n‖ (3.12)
Using (3.11) and (3.12), we get
1
[ r (x )] ≤ ‖x n‖ or r(x) ≤ ‖x n‖n
n
(3.13)

Taking the limit inferior on both sides as n   in (3.13), we get

1

r(x) ≤ lim ‖x ‖
n n
(3.14)
n

If |λ| > ‖x‖, then the function x(λ) = (x−λe )−1 is regular.

( )
−1
x
Hence we have x(λ) = ( x−λe ) = λ −1 −1
λ
−e

( ) [ ]
−1 ∞
x xn
But λ
−1
−e =- λ−1 e+ ∑
λ n =1 λn
¿
Let f є A be functional on A.

[ ( )]
∞
xn
Then f[ x ( λ)] = - λ
−1
f (e)+∑ f n
n=1 λ

[ ]
∞

=- λ−1 f (e)+∑ f ( x n ) λ−n (3.15)

n=1

For all λ >‖x‖.

Further f[ x ( λ)] is analytic outside σ(x) and hence for |λ| > r(x)
Equation (3.15) is its Laurent series expansion which converges for
|λ| > r(x).

Hence sup f ( x n ) λ−n is bounded for every f є A

¿
and |λ| > r(x)
By theorem we conclude that to each λ with λ > r(x), there corresponds a
M(λ) such that
‖λ−n x n‖ ≤ M(λ) for n = 1, 2, 3, . . . (3.16)
Now multiplying both sides of (3.16) by |λ|n and taking the nthroot we get
1 1
‖x ‖ |λ|[ M ( λ)] n , for n = 1, 2, 3, . . . . and
n n ≤

[ ]≤
1

Hence we get sup lim ‖ x n‖n |λ| ≤ r(x) (3.17)

n

[ ] [ ]
1 1

From (3.14) and (3.17), we prove that sup lim ‖ x n‖n ≤ inf lim ‖ x n‖n
n n

Since in any situation limit inferior is less than or equal to limit superior.
1

Therefore r(x) = lim ‖x ‖ n n

Hence the proof.

 CHAPTER 4

IDEALS IN A BANACH ALGEBRA

DEFINTION 4.1
A non – empty subset I of a banach algebra A is called
an ideal if
(i) I is a subspace of A
(ii) For any x є A and y є I, xy є I

NOTE 4.1
The singleton set { 0 } and the whole algebra A are trivially
Two Ideals. An ideal I≠ A is called a proper ideal.

NOTE 4.2
In an algebra with identity, e cannot be in any proper ideal.
If I were a proper ideal and if e є I, then for every x є A, x = ex є I
So that, A  I.
Since I  A is obvious.
Hence A  I.

DEFINITION 4.2
The Radical of A is denoted by R is the intersection of all
Its Maximal ideals. That is R = ∩ { M : M є ∆ }, where ∆ is the set all
Maximal ideals in A.
NOTE 4.3
The radical of A is the intersection of all the maximal left
ideals.
R = ∩ MLI
The radical of A is the intersection of all the maximal right
ideals.
R = ∩ MRI

DEFINITION 4.3
An element x in A is Regular has meant that there exists an
element y such that xy = yx = 1.
An element x is said to be left regular if there exists an element
y such that yx = 1
If x is not left regular, then it is called left singular.
The term right regular and right singular are defined.

LEMMA 4.1
If r is an element of R, then 1 – r is left regular.
PROOF

Given r is an element of R.
To prove that 1 – r is left regular.
We assume that 1 – r is left singular, so that
L = A (1 – r) = { x−xr : x є A } is a proper left ideal which
Contains 1 – r.
By zorn’s lemma we know that
Any proper left ideal can be imbedded in a maximal left ideal.
Therefore we can imbed L in a maximal left ideal M. which of course

Also contains 1 – r.
Since r is in R, it is also in M and
Therefore 1 = (1 – r) + r is in M.
This implies that M = A
Which is a contradiction .
Therefore 1 – r is left regular.
Hence the proof.

LEMMA 4.2
If r is an element of R, Then 1 – r is regular.
PROOF

Given r is an element of R.
To prove 1 – r is regular.
By lemma 4.1
There exists an element s such that s (1 – r) = 1, so s is right
regular and s = 1 – (-s) r.
The fact that R is a left ideal implies that (-s) r is in R along with r,
And another application of the above lemma shows that 1 – (-s) r = s
Is left regular.
Since s is both left regular and right regular, it is regular with inverse
1 – r.
Therefore 1 – r is also regular.
Hence the proof.
LEMMA 4.3
If r is an element of R, then 1 – xr is regular for every x.
PROOF

Given r is an element of R.
Let R is a left ideal, so xr is in R.
Therefore by lemma 4.1 and lemma 4.2.
Therefore 1 – xr is regular for every x.
Hence the proof.

LEMMA 4.4
If r is an element of A with the property that 1 – xr is regular
For every x, then r is in R.
PROOF

Given r is an element of A with the property that 1 – xr is

Regular for every x.
To prove r is in R.
We assume that r is not in R, so that r is not in some maximal left ideal M.
It is easy to see that the set
M + Ar = { m+ xr :m є M ∧x є A } is a left ideal which contains
Both M and r, so M + Ar = A and m + xr = 1 for some m and x.
It now follows that 1 – xr = m is a regular element in M,
and this is impossible, for no proper ideal can contain any regular
element.
The effect of the above lemma is to establish the equality of two sets.
∩ MLI = { r :1−xr is regular for every x } (4.1)
Precisely the same arguments, when applied to maximal right ideals
∩ MRI= {r :1−rx is regular for every x } (4.2)
We now prove that all four of these sets are the same by showing that the
Two sets on the right of (4.1) and (4.2) are equal to one another.
By symmetry, it evidently suffices to prove r is in R.
Hence the proof.

LEMMA 4.5
If 1 – xr is regular, then 1 – rx is also regular.
PROOF

We assume that 1 – xr is regular with inverse s =(1−xr )−1

This means, of course that (1 –xr) s = s (1 – xr) = 1
We need to show that (1 – rx) (1 + rsx) = (1 +rsx)(1 – rx) = 1
ie) to show that 1 – rx is regular with inverse 1 + rsx.
S =(1−xr )−1
= 1 + xr + ( xr )2+ ….
And (1−rx)−1 = 1 + rx +(rx )2+ (rx )3+ ….
= 1 + rx + rx rx + rx rx rx + ….
= 1 + r (1 + xr + xr xr + ….) x
= 1 + rsx
Therefore 1 – rx is also regular.
Hence the proof.

THEOREM 4.1
If I is a proper ideal in an algebra A, then its closure Ī is
also a Proper ideal in A.
PROOF

Given, I is a proper ideal in an algebra A.

To show that Ī is an algebra A
We have to prove x , y є Ī and z є A implies x + y є Ī and zx є Ī
Since x , y є Ī , there exist sequences.
( x n ) , ( y n ) in I such that x n → x and y n → y .
By the continuity of addition and multiplication,
xn + yn → x + y and z n x n  zx
Therefore x + y є Ī and zx є Ī for z є A.
Hence Ī is an ideal.
Now, To show that Ī is a proper ideal.
If x є I, We prove that it is singular.
If x є I is not singular, We have to show that I =A
Which is a contradiction.
Since I is a proper ideal in A.
For, if x−1 є A and e = x−1 x є I
Which means I = A which is against the hypothesis.
Hence x є I implies x is singular.
Hence I  Gc  S which is the set of all singular elements of A.
Since S is a proper closed subset of A, Ī  S.
So that Ī is a proper ideal.
Hence the proof.

DEFINITION 4.4
A proper ideal M of A is said to be a Maximal ideal if it is not
properly contained in any proper ideal of A.

THEOREM 4. 2
If I is a proper ideal of A, then there exists a maximal
ideal M Such that I  M.
PROOF

Let I be a proper ideal of A.

Let I be the set of all proper ideals of A that contains the given ideal I.
With respect to the set inclusion, I is a partially ordered set.
Let T = { I α : α є ∧ } be any totally ordered subset of I

Then ∪ { I α : α є ∧ } is an upper bound for T containing I.

Since the ordering is a total ordering, this upper bound is an ideal
So that this ideal belongs to I.

Further it is a proper ideal because if ∪ { I α : α є ∧ } = A,

Then every totally ordered subset T has a upper bound in I.

Hence by zorn’s lemma, I has a maximal element M ⊃ I.

M is a ideal. To see this let x,y є M. then there exists α, β є ∧
Such that x є I α,y є I β .
If I α I β , then x + y є I β  M.
Let x є M, y є A.
There is α є ∧ such that x є I α.
Then xy є I α M.
Hence M is an ideal.
Hence the proof.
COROLLARY 4.2.1
If x is a singular element of A, then there exists a maximal
Ideal that contains x.
PROOF

Let x є A be a singular element.

Let us consider the set I = (x) = { yx : y є A }.
Let x 1, x 2 є (x)
Then there exist y 1, y 2 є A such that x 1= y 1x , x 2 = y 2x.
Now x 1+ x2 =( y 1 + y 2 ¿ x.
Since y1+ y2 є A , x 1+ x2 є (x)
In a similar manner if z є A, z x 1 є (x).
This proves that (x) is an ideal.
It is a proper ideal
For if it is not proper, Let (x) = A.
Since A contains identity e and e є (x) also, for some y є A, e = yx
Which is contradiction to x is singular.
This proves that (x) is a proper ideal.
Hence by the above theorem there exists a maximal ideal containing
(x) to which the singular element x belongs.

NOTE 4.4
It is obvious that x is regular if and only if it is no ideal.

NOTE 4.5
 Any maximal ideal M in A is always closed. If M is a maximal,
if it is proper.

THEOREM 4.3
If x is a singular element of A, then there exists a maximal
Ideal that contain x.
PROOF
Let x є A be a singular element.
Let us consider the set I = (x) = { yx : y є A }
Let x 1 , x 2є (x)
Then there exists y 1 , y 2є A such that x 1= y 1x , x 2 = y 2x
Now, x 1+ x2 = ¿ ¿) x
Since y1+ y2 є A, x 1+ x2 є (x)
In a similar manner if z є A, z x 1 є (x)
This proves that (x) is an ideal
It is a proper ideal.
For, if it is not proper.
Let (x) = A
Since A contains the identity e and e є (x) also,
For some y є A, e = yx contradicting x is singular
This proves that (x) is a proper ideal.
We have “ If I is a proper ideal of A then there exists a maximal ideal M
such that I  M.”
Therefore there exists a maximal ideal containing (x) to with the singular
element x belongs.
Hence the proof.
NOTE 4.6
From the above theorem it is obvious that x is regular if it is
In no ideal.

THEOREM 4.4
If A is a commutative Banach algebra with identity and M is a
A
Maximal ideal in A then M .

PROOF

Given A is a commutative Banach algebra with identity and

M is a maximal ideal in A.
A
We have to that every non – zero element in M
has an inverse with

respect to multiplication.
Let x + M ≠ 0 which means x ∉ M
Let I be ideal generated by M ∪ { x }. Then
I = { y + zx : y ∈ M , z ∈ A }
Since x ∈ I and x ∉ M, I contains M properly.
Since M is a maximal ideal in A, M cannot be considered in I.
So that I = A
Since e ∈ A, there must be a element y0 ∈ M and Z 0 ∈ A such that
e= y0 + z 0x
hence e + M = ( y 0 + z 0x) + M
= ¿ + M)+ ¿ ¿x + M)
Since y0 ∈ M. We have,
¿ + M)+ ¿ ¿x + M) = M + ¿ ¿x + M)
So that we get
(e+ M) = ¿ ¿x + M)
=¿ ¿ + M) (x + M)
This proves that the inverse of (x + M) is ¿ ¿ + M)
A
So that M
is a field.
A
That is M is commutative division ring.

Hence the proof.

THEOREM 4.5
A
If I is a proper closed ideal of A, then the quotient algebra I

is a Banach algebra.
PROOF

Given I is a proper closed ideal of A.

A
We know that I
is non trivial complex Banach space with respect to the

Quotient norm ‖x + M ‖ = inf {‖ x+i‖:i ∈ I }

A
To introduce the algebraic structure in I
,
A
We shall introduce the multiplication operation in I as follows
A
If x + I , y + I ∈ I
, then ( x + I ) ( y + I ) = xy + I
A
To prove that I is a Banach algebra,

(i) ‖(x+ I )( y + I )‖ ≤ ‖x + I ‖ ‖ y + I ‖
(ii) ‖e+ I ‖=1

Now,
‖(x+ I )( y + I )‖=‖ xy+ I ‖
 = inf {‖ xy+ i:i∈ I‖}
≤ inf {‖( x +i1 )( y+ i2 ) :i1 ,i2 ∈ I‖}
= inf {‖x+ i1 : i1 ∈ I ‖} inf{‖ y +i2 :i 2 ∈ I‖}
= ‖x + I ‖ ‖ y + I ‖
This proves that ‖(x+ I )( y + I )‖ ≤ ‖x + I ‖ ‖ y + I ‖
A
Note that I is an algebra with identity element e +1.

Hence ‖e+ I ‖ = inf {‖e+i :i∈ I‖}

≤ ‖e‖

=1

So that ‖e+ I ‖ ≤ 1 (4.3)

A
Since ( e + I ) is the multiplicatice identity for I
,

We have ‖(e+ I )‖ = ‖(e+ I )2‖

≤ ‖I +e‖2

≥1

Which shows that 1 ≤ ‖e+ I ‖ (4.4)

From (4.3) and (4.4), ‖e+ I ‖ = 1

A
Therefore I
is a Banach algebra.

Hence the proof.

DEFINITION 4.5

The algebra A is called semi – simple, if the radical consists of

The vector 0 alone.

In otherwords the radical equals to zero ideal { 0 }.

THEOREM 4.6

A
If A is a Banach algebra, then R is a semi – simple Banach

algebra.

PROOF

Given A is a Banach algebra.

Let R is the radical of A.

A
Since R is closed ideal in A, R
is a Banach algebra.

A
To prove that R is a semi – simple Banach algebra.

A
Now, consider the natural homomorphism x  x + R of A onto R , which
A
is a one – to – one correspondence between A and R .

The kernel of the homomorphism is the ideal containing the 0 vector

alone.

A
So that the intersection of the maximal ideals in R
is also the ideal

containing the zero vector alone under this homomorphism.

A
Hence R
is semi – simple.

Hence the proof.