UNIT 1

Linear space:

A linear space is an additive abelian group with the property that any scalar𝛼 and any
vector x can be combined by an operation called scalar multiplication, then

i) 𝛼(𝑥 + 𝑦) = 𝛼𝑥 + 𝛼𝑦
ii) (𝛼 + 𝛽)𝑥 = 𝛼𝑥 + 𝛽𝑥
iii) (𝛼. 𝛽)𝑥 = 𝛼. (𝛽𝑥)
iv) 1.x=x
A linear space is a vector space.

The elements of a linear space are called vectors.

Norm:

A normed linear space is a linear space N in which each vector x, there corresponds a real
number ||x|| called norm of x such that,

i) ‖𝑥‖ ≥ 0 and ‖𝑥‖ = 0 iff x=0

ii) ‖𝑥 + 𝑦‖ ≤ ‖𝑥‖ + ‖𝑦‖
iii) ‖𝛼𝑥‖ = |𝛼|‖𝑥‖
Note:

The non-negative number ‖𝑥‖ is length of the vector x.

Banach space:

A complete normed linear space is called Banach space.

Some results of normed linear space:

Result 1:

Prove that |‖𝑥‖ − ‖𝑦‖| ≤ ‖𝑥 − 𝑦‖∀𝑥, 𝑦 ∈ 𝑁

Proof:

‖𝑥‖ = ‖𝑥 − 𝑦 + 𝑦‖

≤ ‖𝑥 − 𝑦‖ + ‖𝑦‖
 ‖𝑥‖ − ‖𝑦‖ ≤ ‖𝑥 − 𝑦‖ → (1)

From (1) interchanging x and y,

‖𝑦‖ − ‖𝑥‖ ≤ ‖𝑦 − 𝑥‖ = ‖𝑥 − 𝑦‖

-[‖𝑥‖ − ‖𝑦‖] ≤ ‖𝑥 − 𝑦‖ → (2)

From (1) & (2).

|‖𝑥‖ − ‖𝑦‖| ≤ ‖𝑥 − 𝑦‖

Hence proved.

Result 2:

Prove that a norm is a continuous function.

Proof:

We prove that

𝑥𝑛 → 𝑥 ⟹ ‖𝑥𝑛 ‖ → ‖𝑥‖

We know that |‖𝑥𝑛 ‖ − ‖𝑥‖| ≤ ‖𝑥𝑛 − 𝑥‖ → (1)

If 𝑥𝑛 → 𝑥 then ‖𝑥𝑛 − 𝑥‖ → 0 → (2)

From (1) & (2),

|‖𝑥𝑛 ‖ − ‖𝑥‖| → 0

Hence ‖𝑥𝑛 ‖ → ‖𝑥‖

Hence 𝑥𝑛 → 𝑥 ⟹ ‖𝑥𝑛 ‖ → ‖𝑥‖

∴ Norm is a continuous function.

Result 3:

Addition and scalar multiplication is jointly continuous.

Proof:

To prove that addition is jointly continuous.

i.e., 𝑥𝑛 → 𝑥 and 𝑦𝑛 → 𝑦
 ⟹ 𝑥𝑛 + 𝑦𝑛 → 𝑥 + 𝑦

Let 𝑥𝑛 → 𝑥 and 𝑦𝑛 → 𝑦

Hence ‖𝑥𝑛 − 𝑥‖ → 0 and ‖𝑦𝑛 − 𝑦‖ → 0

‖(𝑥𝑛 + 𝑦𝑛 ) − (𝑥 + 𝑦)‖ = ‖𝑥𝑛 − 𝑥 + 𝑦𝑛 − 𝑦‖

≤ ‖𝑥𝑛 − 𝑥‖ + ‖𝑦𝑛 + 𝑦‖

→0

Hence 𝑥𝑛 + 𝑦𝑛 → 𝑥 + 𝑦

∴Addition is jointly continuous.

To prove scalar multiplication is jointly continuous.

𝛼𝑛 → 𝛼and𝑥𝑛 → 𝑥

⟹ 𝛼𝑛 𝑥𝑛 → 𝛼𝑥

Let 𝛼𝑛 → 𝛼 and 𝑥𝑛 → 𝑥

Hence ‖𝑥𝑛 − 𝑥‖ → 0 and ‖𝛼𝑛 − 𝛼‖ → 0

‖𝛼𝑛 𝑥𝑛 − 𝛼𝑥‖ = ‖𝛼𝑛 𝑥𝑛 − 𝛼𝑥𝑛 + 𝛼𝑥𝑛 − 𝛼𝑥‖

= ‖(𝛼𝑛 − 𝛼)𝑥𝑛 + 𝛼(𝑥𝑛 − 𝑥)‖

≤ ‖(𝛼𝑛 − 𝛼)𝑥𝑛 ‖ + ‖𝛼(𝑥𝑛 − 𝑥)‖

≤ |(𝛼𝑛 − 𝛼)|‖𝑥𝑛 ‖ + |𝛼|‖(𝑥𝑛 − 𝑥)‖

→0

∴ 𝛼𝑛 𝑥𝑛 → 𝛼𝑥

Scalar multiplication is jointly continuous.

Note:

A normed linear space N is a metric space with respect to the metric d, then

𝑑(𝑥, 𝑦) = ‖𝑥 − 𝑦‖
Quotient space:

Let L be a linear space and M be a linear subspace of L.

Two elements𝑥1 , 𝑥2 in L are said to be equivalent modulo M if 𝑥1 − 𝑥2 in M.

i.e.,𝑥1 ≡ 𝑥2 (𝑚𝑜𝑑 𝑀)

The set of all equivalence classes is denoted by𝐿⁄𝑀. If [𝑥] = {𝑦: 𝑦 − 𝑥 ∈ 𝑀}

= {𝑥 + 𝑚: 𝑚 ∈ 𝑀}

i.e., [𝑥] is the set of all sums of x and elements of M.

the set [𝑥] is called the coset in M, determined by x and it is denoted by x+M. hence
𝐿⁄ = {𝑥 + 𝑀: 𝑥 ∈ 𝐿}
𝑀

The addition and scalar multiplication of 𝐿⁄𝑀 is defined as 𝑥 + 𝑀, 𝑦 + 𝑀 ∈ 𝐿⁄𝑀 and 𝛼

is scalar.

i.e., (x+M)+(y+M)=(x+y+M)

𝛼(𝑥 + 𝑀) = 𝛼𝑥 + 𝑀

𝐿⁄ is a linear space.
𝑀

Theorem:

Let M be a closed linear surface of a normed linear space N, if the norm of a cosetx+M in
the quotient space 𝑁⁄𝑀is defined by ‖𝑥 + 𝑀‖ = inf {‖𝑥 + 𝑚‖: 𝑚 ∈ 𝑀}

Then 𝑁⁄𝑀is a normed linear space.

Further, if N is a Banach Space, then so is 𝑁⁄𝑀.

Proof:

First, we prove that

𝑁⁄ = {𝑥 + 𝑀: 𝑥 ∈ 𝑀}is a normed linear space.

𝑀

Let x+M, y+M∈ 𝑁⁄𝑀 and 𝛼 be scalar.

 i) We prove that ‖𝑥 + 𝑀‖ > 0 and ‖𝑥 + 𝑀‖ = 0 iff x+M=M
i.e.,𝑥 ∈ 𝑀

by definition

‖𝑥 + 𝑀‖ = inf{‖𝑥 + 𝑚‖: 𝑚 ∈ 𝑀} ≥ 0

Also ‖𝑥 + 𝑀‖ = 0 ⟺ inf{‖𝑥 + 𝑚‖: 𝑚 ∈ 𝑀} = 0

⟺There exists a sequence {𝑚𝑛 } in M such that ‖𝑥 + 𝑚𝑛 ‖ → 0

⟺There exists a sequence {𝑚𝑛 } in M such that 𝑚𝑛 → {𝑥}

Since M is closed, it contains all its limit points, hence 𝑥 ∈ 𝑀.

𝑥+𝑀 =𝑀

Hence ‖𝑥 + 𝑀‖ = 0 ⟺ 𝑥 + 𝑀 = 𝑀 → (1)

To prove that

‖(𝑥 + 𝑀) + (𝑦 + 𝑀)‖ ≤ ‖𝑥 + 𝑀‖ + ‖𝑦 + 𝑀‖

Consider ‖(𝑥 + 𝑀) + (𝑦 + 𝑀)‖ = ‖𝑥 + 𝑦 + 𝑀‖

= inf{‖𝑥 + 𝑦 + 𝑚‖: 𝑚 ∈ 𝑀}

= inf{‖𝑥 + 𝑦 + 𝑚 + 𝑚′ ‖: 𝑚, 𝑚′ ∈ 𝑀}

≤ inf{‖𝑥 + 𝑚‖: 𝑚 ∈ 𝑀} + inf{‖𝑦 + 𝑚′‖: 𝑚′ ∈ 𝑀}

‖(𝑥 + 𝑀) + (𝑦 + 𝑀)‖ ≤ ‖(𝑥 + 𝑀)‖ + ‖(𝑦 + 𝑀)‖ (2)

To prove that ‖𝛼(𝑥 + 𝑀)‖ = |𝛼|‖𝑥 + 𝑀‖

‖𝛼(𝑥 + 𝑀)‖ = ‖𝛼𝑥 + 𝑀‖

= inf{‖𝛼𝑥 + 𝑀‖: 𝑚 ∈ 𝑀}

= inf{|𝛼|‖𝑥 + 𝑚‖: 𝑚 ∈ 𝑀}

= |𝛼| inf{‖𝑥 + 𝑚‖: 𝑚 ∈ 𝑀}

= |𝛼|‖𝑥 + 𝑀‖ → (3)

From (1), (2), (3),

𝐿⁄ is a normed linear space.

𝑀
We prove that,

If N is Banach space then

𝑁⁄ is also Banach space.

𝑀
i.e., if N is a Banach space then N is completed.

i.e., Every Cauchy sequence is convergent to N at a point in N.

To prove that 𝑁⁄𝑀 is complete.

Every Cauchy’s sequence is 𝑁⁄𝑀 is convergent subsequence.

By def. of Cauchy’s sequence {𝑆𝑛 + 𝑀}in 𝑁⁄𝑀.

Given 𝜀 > 0 there exist a positive integer K, such that ‖(𝑆𝑛 + 𝑀) − (𝑆𝑚 + 𝑀)‖ < 𝜀
whenever 𝑛, 𝑚 ≥ 𝐾.

Hence for 𝜀 = 1⁄2, there exists n, such that,‖(𝑆𝑛 + 𝑀) − (𝑆𝑚 + 𝑀)‖ < 𝜀
whenever 𝑛, 𝑚 ≥ 𝑛1 .
2
Let 𝑥1 = 𝑆𝑛1 , similarly 𝜀 = (1⁄2) , there exists 𝑛2 such that 𝑛2 ≥ 𝑛1 and ‖(𝑆𝑛 + 𝑀) −
2
(𝑆𝑚 + 𝑀)‖ < (1⁄2) whenever 𝑛, 𝑚 ≥ 𝑛2 .

Let 𝑥2 = 𝑆𝑛2

Since 𝑛2 ≥ 𝑛1 we get

‖(𝑥2 + 𝑀) − (𝑥1 + 𝑀)‖ < 1⁄2

Having chosen 𝑛1 , 𝑛2 , … 𝑛𝑘 and 𝑥1 , 𝑥2 , … 𝑥𝑘 .

Similarly we choose 𝑛𝑘+1 such that 𝑛𝑘+1 ≥ 𝑛𝑘 and ‖(𝑆𝑛 + 𝑀) − (𝑆𝑚 + 𝑀)‖ <
𝑘−1
(1⁄2)

Whenever 𝑛, 𝑚 ≥ 𝑛𝑘−1 ,

Let 𝑥𝑘+1 = 𝑆𝑛𝑘+1 then

1 𝑘
‖(𝑥𝑘+1 + 𝑚) − (𝑥𝑘 + 𝑚)‖ < ( )
2

Continue this process we get the sequence {𝑥𝑘 + 𝑀}of {𝑆𝑛 + 𝑀} such that
 1 𝑘
‖(𝑥𝑘+1 + 𝑚) − (𝑥𝑘 + 𝑚)‖ < ( )
2

Claim:

The subsequence convergence to an element 𝑁⁄𝑀

Let 𝑦1 ∈ 𝑥1 + 𝑀 then 𝑦1 ∈ 𝑥1 + 𝑚1 for some 𝑚1 ∈ 𝑀

i.e.,inf{‖𝑥2 − 𝑥1 + 𝑀‖: 𝑚 ∈ 𝑀 < 1⁄2}

Hence there exist 𝑚0 ∈ 𝑀 such that

‖𝑥1 − 𝑥2 + 𝑚0 ‖ < 1⁄2

Choose 𝑦2 = 𝑥2 + 𝑚1 − 𝑚0

Hence 𝑦2 = 𝑥2 + 𝑀 and ‖𝑦1 − 𝑦2 ‖ < 1⁄2

Similarly
2
‖𝑦2 − 𝑦3 ‖ < (1⁄2)

Continue this way we get a sequence {𝑦𝑛 }in N such that

𝑛
‖𝑦𝑛 − 𝑦𝑛+1 ‖ < (1⁄2)

Hence {𝑦𝑛 } is a Cauchy sequence in N and N is a complete space.

∴In this similar way ‖𝑦𝑛 − 𝑦𝑚 ‖ → 0 𝑎𝑠 ∞

∴ ‖𝑦𝑛 − 𝑦𝑚 ‖ → 0 as𝑛 → ∞

Hence {𝑥𝑛+𝑚 } converges in 𝑁⁄𝑀

Hence our claim is proved.

∴ 𝑁⁄𝑀is complete.

∴ 𝑁⁄𝑀is a Banach space if N is a Banach space.

Examples of Banach space:

i) The real linear space R and the complex linear C are simplest normed linear space
under the form,
‖𝑥‖ = |𝑥| ∀𝑥 ∈ 𝑅 𝑜𝑟 𝐶

Here ‖𝑥‖ is called absolute value of x. Here R & C are complete under the metrics
induced by the norm. Hence C and R are Banach space.

ii) Linear space 𝑅 𝑛 &𝐶 𝑛 of all n-tuples 𝑥 = {𝑥1 , 𝑥2 , … 𝑥𝑛 } of real and

complex numbers is a normed linear space under the norm
1⁄
2
‖𝑥‖ = [∑|𝑥𝑖 ] |2

iii) Let P be a real number such that 1 ≤ 𝑃 < ∞ defined 𝑙𝑝 𝑛 with the space of all n-tuples
𝑥 = {𝑥1 , 𝑥2 , … 𝑥𝑛 } of scalars then norm is defined by ‖𝑥‖𝑃 =
𝑛 1⁄
(∑𝑖=1|𝑥𝑖 |𝑃 ) 𝑃
iv) Let P be a real number such that 1 ≤ 𝑃 < ∞ defined 𝑙𝑃 be the spaces of all sequence
𝑥 = {𝑥1 , 𝑥2 , … 𝑥𝑛 , … } are scalar such that ∑∞ 𝑃
𝑛=1|𝑥𝑛 | < ∞. Then the norm is defined
by
1⁄
∞ 𝑃
‖𝑥𝑃 ‖ = (∑|𝑥𝑖 |𝑃 )
𝑖=1

Continuous linear transformation and the conjugate space:

Continuous linear transformation:

Let 𝑁 & 𝑁′ be normed linear space with the same scalar and let 𝜏 be a linear
transformation of 𝑁into 𝑁′, then T is said to be continuous.

If T is continuous mapping of the metric space N into the metric space 𝑁′.

i.e. 𝑥𝑛 → 𝑥 in N then 𝑇(𝑥𝑛 ) → 𝑇(𝑥) in 𝑁 ′ is called continuous linear

transformation.
Theorem:

Let 𝑁 & 𝑁′ be normed linear space and T be a linear transformation of 𝑁into 𝑁′,
then the following condition for T are all equivalent to one another,

a) T is continuous
b) T is continuous at the origin in the sense that 𝑥𝑛 → 𝑥 ⟹ 𝑇(𝑥𝑛 ) ⟶ 0
c) There exist a real number 𝑘 ≥ 0 with the property that ‖𝑇(𝑥)‖ ≤ 𝑘‖𝑥‖ for every
𝑥∈𝑁
d) If 𝑆 = {𝑥: ‖𝑥‖ ≤ 1} is a closed unit sphere in N then its image T(S) is a bounded
set in 𝑁′.
Proof:

We prove that

𝑎⟺𝑏⟺𝑐⟺𝑑

To prove 𝑎 ⟺ 𝑏

Assume that T is continuous on N.

Hence T is continuous at every point on N. Hence T is continuous at zero, 0 ∈ 𝑁

Conversely,

Assume that T is continuous at the origin.

i.e.𝑥𝑛 → 0 ⟹ 𝑇(𝑥𝑛 ) ⟶ 0

Let x be any point in N and there exist sequence {𝑥𝑛 } in N such that 𝑥𝑛 → 𝑥.

𝑥𝑛 − 𝑥 → 0 ⟹ 𝑇(𝑥𝑛 − 𝑥) ⟶ 0

⟹ 𝑇(𝑥𝑛 ) − 𝑇(𝑥) → 0

⟹ 𝑇(𝑥𝑛 ) ⟶ 𝑇(𝑥)

Hence 𝑥𝑛 ⟶ 𝑥 𝑖𝑛 𝑁 ⟹ 𝑇(𝑥𝑛 ) ⟶ 𝑇(𝑥) 𝑖𝑛 𝑁′

Hence T is continuous at x in N.

Since x is arbitrary point in N, this result is true for all x in N. Hence T is continuous.

∴ 𝑎⟺𝑏
To prove 𝑏 ⟺ 𝑐

Assume that T is continuous at origin

i.e. 𝑥𝑛 ⟶ 0 ⟹ 𝑇(𝑥𝑛 ) ⟶ 0, then there exist a real number k, such that

‖𝑇(𝑥)‖ ≤ 𝑘‖𝑥‖

Hence for any integer n, there exists𝑥𝑛 ∈ 𝑁, such that

‖𝑇(𝑥𝑛 )‖ ≥ 𝑛‖𝑥𝑛 ‖
𝑥𝑛
Let𝑦𝑛 = 𝑓𝑜𝑟 𝑒𝑎𝑐ℎ 𝑛.
𝑛‖𝑥𝑛 ‖

Then 𝑦𝑛 ⟶ 0 as 𝑛 ⟶ ∞
𝑥
But ‖𝑇(𝑦𝑛 )‖ = ‖𝑇 (𝑛‖𝑥𝑛 ‖)‖
𝑛

1
‖𝑇(𝑦𝑛 )‖ = ‖𝑇(𝑥𝑛 )‖ ≥ 1
𝑛‖𝑥𝑛 ‖

Hence 𝑇(𝑦𝑛 ) does not tends to 0 as 𝑛 ⟶ ∞

When 𝑦𝑛 ⟶ 0 as 𝑛 ⟶ ∞

⟹⟸that T is continuous at origin.

Hence there exist a real number k, such that

‖𝑇(𝑥)‖ ≤ 𝑘‖𝑥‖ ∀𝑥 ∈ 𝑁

Hence 𝑏 ⟹ 𝑐

To prove c⟹ 𝑏

Conversely there exist a real number k, such that ‖𝑇(𝑥)‖ ≤ 𝑘‖𝑥‖ ∀𝑥 ∈ 𝑁

Let 𝑥𝑛 ⟶ 0

∴ ‖𝑥𝑛 ‖ ⟶ 0

Hence ‖𝑇(𝑥)‖ ≤ 𝑘‖𝑥‖ ⟶ 0

But 𝑇(𝑥𝑛 ) ⟶ 0

𝑥𝑛 ⟶ 0
 ∴ T is continuous at origin.

∴𝑐⟹𝑏

Hence 𝑏 ⟺ 𝑐 proved.

To prove 𝑐 ⟺ 𝑑

Assume that there exist a real number 𝑘 ≥ 0 such that ‖𝑇(𝑥)‖ ≤ 𝑘‖𝑥‖ ∀𝑥 ∈
𝑁 → (1)

If 𝑥 ∈ 𝑆 then ‖𝑥‖ ≤ 1 → (2)

From (1) and (2)

‖𝑇(𝑥)‖ ≤ 𝑘for each x in S.

Hence T(S) is bounded by k,

Hence T(S) is a bounded set in 𝑁′

∴ 𝑐 ⟹ 𝑑is proved.

To prove 𝑑 ⟹ 𝑐

Assume that T(S) is bounded in 𝑁′ where 𝑆 = {𝑥: ‖𝑥‖ ≤ 1} is a closed unit

sphere in N.

Hence there exist a real number k such that

‖𝑇(𝑥)‖ ≤ 𝑘 ∀𝑥 ∈ 𝑆

Let y be any point in N.

𝑦
Let 𝑥 = ‖𝑦‖

𝑦
‖𝑥‖ = ‖ ‖
‖𝑦‖

1
= ‖𝑦‖ ‖𝑦‖

=1

Hence ‖𝑇(𝑥)‖ ≤ 𝑘
𝑦
‖𝑇 (‖‖𝑦‖‖)‖ ≤ 𝑘
 1
‖𝑇(𝑦)‖ ≤𝑘
‖𝑦‖

‖𝑇(𝑦)‖ ≤ 𝑘‖𝑦‖

Since y is arbitrary,

‖𝑇(𝑦)‖ ≤ 𝑘‖𝑥‖

𝑑⟹𝑐

From above,

𝑎⟺𝑏⟺𝑐⟺𝑑

Hence proved.

Bounded linear transformation:

If linear transformation T from normed linear space 𝑁 × 𝑁 ′ satisfies the condition there exist
a real number k≥ 0 then ∥ 𝑇(𝑥) ∥≤ 𝑘 𝑡𝑖𝑚𝑒𝑠 ∥ 𝑥 ∥ forevery x in N.

Then k is called bounded for T and T is called bounded linear transformation.

Definition:

The norm of bounded linear transformation is defined as ∥ 𝑇 ∥= sup{∥ 𝑇(𝑥) ∥: ∥ 𝑥 ∥≤ 1}

When 𝑁 ≠ 0 . Then

∥ 𝑇 ∥= sup{∥ 𝑇(𝑥) ∥: ∥ 𝑥 ∥= 1}

The set of all continuous linear transformation 𝑁 × 𝑁 ′ is denoted by B(𝑁, 𝑁 ′ ).

Theorem:

If 𝑁&𝑁 ′ are normed linear space then the set of all continuous linear transformation of 𝑁 × 𝑁 ′
(i.e) B(𝑁, 𝑁 ′ ) is a normed linear with respect to the point wise linear operation and the norm
defined as

∥ 𝑇 ∥= sup{∥ 𝑇(𝑥) ∥: ∥ 𝑥 ∥≤ 1}for T in B(𝑁, 𝑁 ′ )

Further if 𝑁 ′ is a banach space then B(𝑁, 𝑁 ′ ) is also a banach space.

Proof:

Claim 1:

B(𝑁, 𝑁 ′ ) is a normed linear space.

Let 𝑇 ∈B(𝑁, 𝑁 ′ ).

Hence,

∥ 𝑇 ∥= sup{∥ 𝑇(𝑥) ∥: ∥ 𝑥 ∥≤ 1} ≥ 0

Also, ∥ 𝑇 ∥= 0 iff sup{∥ 𝑇(𝑥) ∥: ∥ 𝑥 ∥≤ 1} = 0

Conversely, ∥ 𝑇(𝑥) ∥= 0

Hence, ∥ 𝑇(𝑥) ∥= 0
𝑦 ∥𝑦∥
If y in N and 𝑦 ≠ 0 then ∥ ∥𝑦∥ ∥= ∥𝑦∥=1

𝑦
Therefore ∥ 𝑇(∥𝑦∥ ) ∥= 0

Hence ∥ 𝑇(𝑦) ∥= 0

Hence ∥ 𝑇 ∥= 0 iff ∥ 𝑇(𝑥) ∥= 0 forall x in N

∥ 𝑇(𝑥) ∥= 0forall x in N

Implies T=0

Hence ∥ 𝑇 ∥≥ 0 and Hence ∥ 𝑇 ∥= 0 iff T=0

∥ 𝑇1 + 𝑇2 ∥= sup{∥ (𝑇1 + 𝑇2 ) 𝑥 ∥: ∥ 𝑥 ∥≤ 1}

≤∥ 𝑇1 ∥ +∥ 𝑇2 ∥

Let T∈B(𝑁, 𝑁 ′ ) and 𝛼 be any scalar .

∥ 𝛼𝑇 ∥= sup{∥ 𝛼𝑇(𝑥) ∥: ∥ 𝑥 ∥≤ 1}

= │𝛼│ ∥ 𝑇 ∥

Claim: 2

If 𝑁 ′ is a banach space then B(𝑁, 𝑁 ′ ) is also a banach space.

Let 𝑁 ′ be a banach space 𝑁 ′ is complete.

(i.e.) every Cauchy’s sequences in 𝑁 ′ converges in 𝑁 ′ .

Let {𝑇𝑛 } be a cauchy’s sequence in B(𝑁, 𝑁 ′ ).let x be an arbitrary vector in N &𝑥 ≠ 0.

Consider, ∥ 𝑇𝑚 (𝑥) − 𝑇𝑛 (𝑥) ∥≤∥ 𝑇𝑚 − 𝑇𝑛 ∥∥ 𝑥 ∥

𝜀
Since {𝑇𝑛 } is a cauchy’s sequence. Given 𝜀 > 0 for an integer no such that ∥ 𝑇𝑚 − 𝑇𝑛 ∥≤ ∥𝑥∥
whenever

m,n≥ 𝑛0 .

Hence {𝑇𝑛 (𝑥)} is a cauchy’s sequence . hence {𝑇𝑛 (𝑥)} is converges to 𝑁 ′ .

Claim 3:

T∈B(𝑁, 𝑁 ′ ) and 𝑇𝑛 → ∞ 𝑎𝑠 𝑛 → ∞ . we shall show that T is linear w.r.t addition & scalar
multiplication and T is bounded.

Now define T in N into 𝑁 ′ by T(x)= lim 𝑇𝑛 (𝑥)

𝑛→∞

𝑇(𝑥1 + 𝑥2 ) = lim 𝑇𝑛 (𝑥1 + 𝑥2 )

𝑛→∞

= lim 𝑇𝑛 (𝑥1 ) + lim 𝑇𝑛 (𝑥2 )

𝑛→∞ 𝑛→∞

=𝑇(𝑥1 ) + 𝑇(𝑥2 )

Hence T is linear w.r.t addition .

Now 𝑥 ∈ 𝑁 & 𝛼 is any scalar.

Such that ,

T(x)= lim 𝑇𝑛 (𝑥)

𝑛→∞

T(𝛼x)= lim 𝑇𝑛 (𝛼𝑥)

𝑛→∞

=𝛼 𝑇(𝑥)

Therefore T is linear w.r.t scalar multiplication.

Now, we prove that ,T is bounded .

T(x)= lim 𝑇𝑛 (𝑥)

𝑛→∞
∥T(x) ∥=∥ lim 𝑇𝑛 (𝑥) ∥
𝑛→∞

≤ sup ∥ 𝑇(𝑥) ∥∥ 𝑥 ∥

Let k=𝑠𝑢𝑝 ∥ 𝑇(𝑥) ∥∥ 𝑥 ∥

Therefore ,∥ 𝑇(𝑥) ∥≤∥ 𝑥 ∥

T is bounded.

Hence T is continuous bounded. Linear transformation from N into 𝑁 ′ . Hence T∈ B(𝑁, 𝑁 ′ ).

To prove 𝑇𝑛 → 𝑇 𝑎𝑠 𝑛 → ∞ (I.e) to prove that

∥ 𝑇𝑛 − 𝑇 ∥→ 0as 𝑛 → ∞ .

Since {𝑇𝑛 } is a cauchy’s sequence 𝜀 > 0 there exist a positive integer 𝑛0 . such that ∥ 𝑇𝑚 − 𝑇𝑛 ∥
≤𝜀

Whenever m,n≥ 𝑛0 .

Then

∥ 𝑇𝑚 (𝑥) − 𝑇𝑛 (𝑥) ∥=∥ (𝑇𝑚 − 𝑇𝑛 )𝑥 ∥

≤∥ 𝑇𝑚 − 𝑇𝑛 ∥∥ 𝑥 ∥

<𝜀

Now, m is fixed and n approaches to infinity. Therefore,

sup{∥ (𝑇𝑚 − 𝑇)𝑥 ∥: ∥ 𝑥 ∥≤ 1} < 𝜀whenever m≥ 𝑛0 .

The {𝑇𝑛 } is converges in B(𝑁, 𝑁 ′ ).

Every Cauchy’s sequences are B(𝑁, 𝑁 ′ ) is converges.

Therefore its complete.

B(𝑁, 𝑁 ′ ) is a Banach space.

Hence proved.

OPERATOR:

Let N be the normed linear space, a continuous linear transformation of N into itself is
called an operator on N.
If N is a Banach space then B(N). (i.e.)The normed liner space of operator.

DEFINITION:

Let N and 𝑁 ′ be normed linear space and isometric isomorphism of N into 𝑁 ′ is a one to
one linear transformation T of N into 𝑁 ′ . Such that

∥ 𝑇(𝑥) ∥=∥ 𝑇 ∥Forevery x in N.

Let N and 𝑁 ′ be a normed linear space N said to be isomertically isomorphic 𝑁 ′ if there exist
isomertic isomorphism of N into𝑁 ′ .

Prove that ,if T,𝑇 ′ ∈B(N) then ∥ 𝑇𝑇 ′ ∥≤∥ 𝑇 ∥ ∥ 𝑇 ′ ∥.

∥ 𝑇 ∥= sup{∥ 𝑇(𝑥) ∥: ∥ 𝑥 ∥≤ 1}

∥ 𝑇𝑇 ′ ∥= sup{∥ 𝑇𝑇 ′ (𝑥) ∥: ∥ 𝑥 ∥≤ 1}

≤ sup ∥ 𝑇 ∥∥ 𝑇 ′ (𝑥) ∥: ∥ 𝑥 ∥≤ 1

≤∥ 𝑇 ∥∥ 𝑇 ′ ∥

Hence,

∥ 𝑇𝑇 ′ ∥≤∥ 𝑇 ∥∥ 𝑇 ′ ∥

Hence proved.

Prove that multiplication is jointly continuous in the B(N) (ie) 𝑇𝑛 → 𝑇 &𝑇𝑛′ → 𝑇 then 𝑇 𝑇𝑛′ →
𝑇𝑇 ′ .

∥ 𝑇𝑛 𝑇𝑛′ − 𝑇𝑇 ′ ∥=∥ 𝑇𝑛 𝑇𝑛′ − 𝑇𝑛 𝑇 ′ + 𝑇𝑛 𝑇 ′ − 𝑇𝑇 ′

∥ (𝑎𝑑𝑑&𝑠𝑢𝑏 𝑇 ∩ 𝑇 ′ )

≤∥ 𝑇𝑛 ∥∥ 𝑇𝑛′ − 𝑇 ∥ +∥ 𝑇𝑛 – 𝑇 ∥ ∥ 𝑇𝑇 ′ ∥

=0 (𝑇𝑛 → 𝑇 &𝑇𝑛′ → 𝑇)

Hence, 𝑇 𝑇𝑛′ → 𝑇𝑇 ′

Hence proved.
If I is identity transformation of B(N) then show that ∥ 𝑥 ∥= 1.

∥ 𝑥 ∥= sup{∥ 𝐼(𝑥) ∥: ∥ 𝑥 ∥: ≤ 1}

=sup{∥ 𝑥 ∥: ∥ 𝑥 ∥: ≤ 1}

∥ 𝑥 ∥ =1

Hence proved.

HAHN BANACH:

CONJUGATE SPACE :(𝑵∗ )

Let N be the normed liner space set of all continuous linear transformation of N to R
or N to C. (ie) B(N,R), B(N,C) is called the conjugate space of N and its denoted by 𝑁 ∗ .

The element 𝑁 ∗ are called continuous liner function or simply function.

Note:

𝑁 ∗ is also banach space .

LEMMA:

Let M be a liner subspace of a normed liner space N let f be a function defined on M. If 𝑥𝑜 is

a vector not in M and if 𝑀𝑂 = 𝑀 + [𝑥𝑜 ] is a linear subspace spanned by M, and 𝑥𝑜 then f can be
extenned to a functional 𝑓𝑜

Defined on 𝑀𝑂 such that ∥ 𝑓𝑜 ∥=∥ 𝑓 ∥.

Proof:

Let M be a linear space and let N be a normed liner space.

Without loss of generality, assume that ∥ 𝑓 ∥= 1.

Since 𝑥𝑜 ∉ 𝑀 each vector y∈ 𝑀𝑂 = 𝑀 + [𝑥𝑜 ] is uniquely represented as y=𝑥 +

𝛼𝑥𝑜 , 𝑓𝑜 (𝑥𝑜 )=𝑥𝑜 with x in M and 𝛼 in R.

Now, 𝑓𝑜 (𝑦𝑜 )=( 𝑥 + 𝛼𝑥𝑜 )

=𝑓𝑜 ( 𝑥) + 𝑓𝑜 (𝛼𝑥𝑜 )

= f(x)+𝛼𝑥𝑜

Where 𝑥𝑜 is a real number.

Show that 𝑓𝑜 is a linear in 𝑀𝑂 .

let𝑥1, 𝑦1 ∈ 𝑀𝑂 𝑎𝑛𝑑 𝑥, 𝑦 ∈ 𝑀

such that

𝑥1 = 𝑥 + 𝛼𝑥𝑜 &𝑦1 =𝑦 + 𝛽𝑥𝑜

𝑓𝑜 (𝑥1 + 𝑦1 )=𝑓𝑜 (𝑥 + 𝛼𝑥𝑜 + 𝑦 + 𝛽𝑥𝑜 )

=f(x+y)+( 𝛼 + 𝛽) 𝑥𝑜

=𝑓𝑜 (𝑥1)+𝑓𝑜 (𝑦1 ) → ①

Let 𝑦 ∈ 𝑀𝑂 &k be any scalar

𝑓𝑜 (𝑘𝑦) = 𝑓𝑜 (𝑘(𝑥 + 𝛼𝑥𝑜 ) )

=𝑓𝑜 (𝑘𝑥 + 𝑘𝛼𝑥𝑜 )

= k 𝑓(𝑥) + 𝑘 𝛼𝑥𝑜

=k𝑓𝑜 (𝑦)→②

From ①&②

𝑓𝑜 is linear in 𝑀𝑂 .

2) 𝑓𝑜 extends from M to 𝑀𝑂 if y∈ 𝑀 then 𝛼 = 0 in the representation of y such that y=x.

𝑓𝑜 (𝑦)=𝑓 (𝑥) + 0 =𝑓 (𝑦)

Hence 𝑓𝑜 extends f from M to 𝑀𝑂 .

3) ∥ 𝑓𝑜 ∥=∥ 𝑓 ∥when = 0 , 𝑓𝑜 (𝑥)=𝑓 (𝑥) for all

x∈ 𝑀𝑂 .

Therefore, ∥ 𝑓𝑜 ∥=∥ 𝑓 ∥

Let 𝛼 ≠ 0, since 𝑀 ⊂ 𝑀𝑂

│𝑓𝑜 (𝑥)│
Sup { :𝑥 ∈ 𝑀, 𝑥 ≠ 0}
∥𝑥∥

│𝑓𝑜 (𝑥)│
≤ Sup { :𝑥 ∈ 𝑀𝑂 , 𝑥 ≠ 0}
∥𝑥∥

Since 𝑓𝑜 (𝑥)=𝑓 (𝑥) for all 𝑥 ∈ 𝑀

 │𝑓(𝑥)│ │𝑓𝑜 (𝑥)│
Sup { :𝑥 ∈ 𝑀, 𝑥 ≠ 0} ≤ Sup { ∈ 𝑀𝑂 , 𝑥 ≠ 0}
∥𝑥∥ ∥𝑥∥

Hence, ∥ 𝑓𝑜 ∥≤∥ 𝑓 ∥→ ③

Let 𝑥1 , 𝑥2 ∈ 𝑀

𝑓(𝑥2 ) − 𝑓(𝑥1 ) = 𝑓(𝑥2 − 𝑥1 )

=∥ 𝑥2 + 𝑥0 − (𝑥1 + 𝑥0 )∥

≤∥ 𝑥2 + 𝑥0 ∥ +∥ 𝑥1 + 𝑥0 ∥ −𝑓(𝑥1 )−∥ 𝑥1 + 𝑥0 ∥≤ −𝑓(𝑥2 ) ∥ 𝑥2 + 𝑥0 ∥ →④

Define two real number a&b by,

𝑎 = sup{−𝑓(𝑥)−∥ 𝑥 + 𝑥0 ∥: 𝑥 ∈ 𝑀}

𝑏 = sup{−𝑓(𝑥)+∥ 𝑥 + 𝑥0 ∥: 𝑥 ∈ 𝑀}

Equation ④ becomes ,𝑎 ≤ 𝑏

Choose a real number 𝛾𝑜 such that 𝑎 ≤ 𝛾𝑜 ≤ 𝑏

Then

−𝑓(𝑥)−∥ 𝑥 + 𝑥0 ∥≤ 𝛾𝑜 ≤ −𝑓(𝑥) ∥ 𝑥 + 𝑥0 ∥
𝑥
Put 𝑥 = 𝛼

𝑥 𝑥 𝑥 𝑥
−𝑓 ( ) −∥ + 𝑥0 ∥≤ 𝛾𝑜 ≤ −𝑓( ) ∥ + 𝑥0 ∥
𝛼 𝛼 𝛼 𝛼
│f(x)+ 𝛼𝑥0 │≤∥ 𝑥 + 𝛼𝑥0 ∥

Therefore, │𝑓𝑜 (x)+ 𝛼𝑥0 │≤∥ 𝑥 + 𝛼𝑥0 ∥

If z= 𝛾+ 𝛼𝑥0 ,𝑥0 ∈ 𝑀0

│𝑓0 (𝑧)│ ≤∥ 𝑧 ∥

Sup{│𝑓0 (𝑧)│: 𝑧 ∈ 𝑀0 , ∥ 𝑧 ∥≤ 1} ≤ 1

Hence ∥ 𝑓𝑜 ∥≤ 1 =∥ 𝑓 ∥→ ⑤

From ⑤&③
 ∥ 𝑓0 ∥=∥ 𝑓 ∥

Hence proved.

STATE AND PROVE HAHN BANACH THEOREM:

Let M be a linear subspace of a normed linear subspace N and let F be a functional defined on
M. Then f can be extended to a functional 𝑓0 defined on the whole space N such that

∥ 𝑓0 ∥=∥ 𝑓 ∥.

Proof:

By previous lemma, (statement only) f can be extended to functional 𝑓0 on 𝑀0 if 𝑀0 = 𝑁

then we get the hahn banach thrm,

Suppose 𝑀0 ≠ 𝑁 the repect the process of extension let p be the set of all order paris
(𝑀 , 𝑓 ′ )
′

When 𝑀′ is a subspacecontaningM and𝑓 ′ is the extension of f such that ∥ 𝑓 ′ ∥=∥ 𝑓 ∥

Since 𝑓0 in p can be partically order set .therefore its defined (𝑀′ , 𝑓 ′ )≤ (𝑀′′ , 𝑓 ′′ )

Iff𝑀′ ⊂ 𝑀′′ and 𝑓 ′ = 𝑓 ′′ on 𝑀′ .

Let s={(𝑚𝑖 , 𝑓𝑖 )} by a total ordered subset (let p be a set of all real function defined on non empty
set x and

f≤ 𝑔 implies f(𝑥) ≤ 𝑔(x) a partial ordered relation with the above property is called total order
relation).

Then , this chain has upper bound , therefore p has maximum element .

Claim:

̅̅̅=N)
M complement N (𝑀

Suppose ̅̅̅
𝑀 ≠N (i.e) ̅̅̅
𝑀 ⊂N.

By zorn’s lemma

[if x is a co set in which every total order subset has an upperbound then x has maximum
element]

̅̅̅ ,𝑓0 ) ≤ (𝑀
Therefore 𝑓0 can be extended to a functional 𝐹0 in 𝑓0 in such a way that (𝑀 ̅̅̅0 , 𝐹0 )

Where ̅̅̅
𝑀 0 is subspace spanned by ̅̅̅
𝑀 and 𝑥0 = 𝑁 − ̅̅̅
𝑀.
 ̅̅̅ ,𝑓0 ) is maximal that is 𝑀
Since(𝑀 ̅̅̅ = 𝑀
̅̅̅0 which is contradcation .

Hence ̅̅̅
𝑀 0 ≠ 𝑁its false

̅̅̅
𝑀0 =𝑁

Therefore 𝑓0 is extension of f on M such that

∥ 𝑓0 ∥=∥ 𝑓 ∥

Hence proved.

CONSEQUENCE OF HAHN BANACH SPACE:

If N is a normed liner space 𝑥0 is a non zero vector in N then there exist a functional 𝑓0 in 𝑁 ∗
such that 𝑓0 (𝑥0 ) =∥ 𝑥0 ∥ 𝑎𝑛𝑑 ∥ 𝑓0 ∥= 1.

Proof:

Let M={𝛼𝑥0 } be a linear subspace of M a spanned by 𝑥0 defined f on M as f(𝛼𝑥0 ) = 𝛼 ∥

𝑥0 ∥

Now, we show that , f is functional on M with ∥ 𝑓0 ∥= 1

Case(1);

To prove that f is linear.

Let 𝑥1 , 𝑥2 ∈ 𝑀 so that 𝑥1 = 𝛼1 𝑥0 &𝑥2 = 𝛼2 𝑥0 then

𝑓(𝑥1 + 𝑥2 ) = 𝑓(𝛼1 𝑥0 + 𝛼2 𝑥0 )

= 𝑓((𝛼1 + 𝛼2 )𝑥0 )

= (𝛼1 + 𝛼2 )𝑓(𝑥0 )

=(𝛼1 + 𝛼2 ) ∥ 𝑥0 ∥

= 𝑓(𝛼1 𝑥0 ) + 𝑓(𝛼2 𝑥0 )

= 𝑓( 𝑥1 + 𝑥2 )

Therefore f is linear with respect to addition .

Case(2);
 Let K be a real or complex scalar if x in M and defined x=𝛼𝑥0

𝑓(𝑘𝑥) = 𝑓(𝑘𝛼𝑥0 )

= 𝑘𝛼 ∥ 𝑥0 ∥

=k f(x)

Therefore f is linear .

Case(3);

To prove 𝑓(𝑥0 ) =∥ 𝑥0 ∥

Since 𝑥0 in M and 𝛼 = 1

𝑓(𝑥0 ) = 𝑓(1. 𝑥0 )

= 1𝑓(𝑥0 )

𝑓(𝑥0 ) =∥ 𝑥0 ∥

Case(4);

To prove f is bounded

For any x in M .

∥ 𝑓(𝑥0 ) ∥= │𝛼 ∥ 𝑥0 ∥ │

=∥ 𝛼𝑥0 ∥

=∥ 𝑥0 ∥→①

Therefore f is bounded.

To prove ∥ 𝑓 ∥= 1

Equation ① becomes ∥ 𝑓(𝑥) ∥=∥ 𝑥 ∥ x is arbitrary.

⇒Sup{∥ 𝑓(𝑥) ∥: ∥ 𝑥 ∥≤ 1}=1

⇒∥ 𝑓 ∥= 1

Hence f is a functional on M with ∥ 𝑓 ∥= 1also (𝑥0 ) =∥ 𝑥0 ∥ .

Hence hahnbanach theorem f can be extened to a functional 𝑓0 in 𝑁 ∗ with (𝑥0 ) =∥ 𝑥0 ∥ 𝑎𝑛𝑑 ∥

𝑓 ∥= 1.
Theorem:

Let M is closed linear subspace of a normed linear space N and 𝑥0 is a vector not in M then
there exist a functional 𝑓0 in 𝑁 ∗ such that 𝑓0 (𝑀) = 0𝑎𝑛𝑑𝑓0 (𝑥0 ) ≠ 0

Proof:

Since M is a closed linear subspace of normed linear space N. The quotient space N/M is a
normed linear space .

Consider, T:N→N/M:T(x)=x+M

Then T is bounded linear transformation with ∥ 𝑇 ∥≤ 1.

We get T is continuous linear transformation if 𝑚 ∈ 𝑀,

T(m)=m+M(zero element of N/M is M)

Hence T(m)=0

By hypothesis T(𝑥0 ) = 𝑥0 + 𝑀 which is not equal to M.

T(𝑥0 ) ≠ 𝑀

Therefore a function 𝑓 ∈ (𝑁/𝑀)∗ such that

𝑓(𝑥0 + 𝑀) =∥ 𝑥0 + 𝑀 ∥ 𝑎𝑛𝑑 ∥ 𝑓 ∥= 1.

Defined 𝑓0 on N as 𝑓0 (𝑥) = 𝑇(𝑓(𝑥))

Claim:1

𝑓0 is linear.

Let (x,y) in M and 𝛼, 𝛽 be any scalar then,

𝑓0 (𝛼𝑥 + 𝛽𝑦) = 𝑓(𝑇(𝛼𝑥 + 𝛽𝑦))

= 𝑓(𝛼(𝑥 + 𝑀) + 𝛽(𝑦 + 𝑀))

= 𝛼𝑓0 (𝑥) + 𝛽𝑓0 (𝑦)

Therefore 𝑓0 is linear.
Claim:2

𝑓0 is bounded.

│𝑓0 (𝑥)│ = │𝑓(𝑇(𝑥))│

≤∥ 𝑓 ∥∥ 𝑇(𝑥) ∥

≤∥ 𝑓 ∥∥ 𝑥 ∥

Hence 𝑓0 is bounded .

The claim 1&2 𝑓0 is continuous linear function on 𝑁 ∗

Therefore 𝑓0 ∈ 𝑁 ∗

Now ,

𝑓0 (𝑥) = 𝑓(𝑇(𝑥))

= 𝑓(𝑥0 + 𝑀) ≠ 𝑀

𝑓0 (𝑥) ≠ 0

There exist a function 𝑓0 ∈ 𝑁 ∗ such that 𝑓0 (𝑥) = 0 &𝑓0 (𝑥) ≠ 0.

Hence proved.

The open mapping theorem:

If B and B’ are banach spaces and if T is a continuous linear transformation of B and B’.
then the image of each open sphere centered on the origin in B contains an open sphere centered
on the origin in B’.

Proof:

Let 𝑆𝑟 and 𝑆𝑟 ′ denote open spheres centered at the origin and the radius r in B and B’

Consider

𝑆𝑟 = {𝑥є𝐵: ||𝑥|| < 𝑟}

 ||𝑥||
={xєB: <1} ={ryєB:||y||<1} 𝑆𝑟 = 𝑟𝑆1
𝑟

Hence T(𝑆𝑟 ) = 𝑇(𝑟𝑆1 ) = 𝑟𝑇(𝑆1 )

We prove that the lemma that T(𝑆𝑟 ) is open it sufficient to show that T(𝑆1 )contains some open
sphere 𝑆𝑟 ′ so that T(𝑆𝑟 ) is open.

Claim1:

An interior point 𝒚𝟎 of ̅̅̅̅̅̅̅̅̅̅

𝐓(𝑺𝒏𝟎 ) lies in 𝐓(𝑺𝒏𝟎 )

Let xєB then xє𝑆𝑟 for some positive real number r

B=⋃∞
𝑛=1 𝑆𝑛

Since T is linear and onto

B’=T(B)

=T(⋃∞
𝑛=1 𝑆𝑛 )

B’=⋃∞
𝑛=1 𝑇(𝑆𝑛 )

B’ is complete and B’ is of second category

ie) B’ could not written as the union of no were dense subsets of B’ hence the closure of at least
one set in the sequence must have non empty interior

let y be an interior point of ̅̅̅̅̅̅̅̅̅

T(𝑆𝑛0 ) then there exist an open set G such that y∈ 𝐺 ⊂ ̅̅̅̅̅̅̅̅̅
T(𝑆𝑛0 )
hence

y∈ ̅̅̅̅̅̅̅̅̅
T(𝑆𝑛0 )

therefore y is a limit point of T(𝑆𝑛0 )

hence neighborhood of G of y contains 𝑦0 of T(𝑆𝑛0 )

𝑦0 ∈ 𝐺 ⊂ ̅̅̅̅̅̅̅̅̅
T(𝑆𝑛0 )

𝑦0 is interior point of ̅̅̅̅̅̅̅̅̅

T(𝑆𝑛0 )

Therefore An interior point of 𝑦0 of ̅̅̅̅̅̅̅̅̅

T(𝑆𝑛0 ) lies in T(𝑆𝑛0 )itself.

Claim2:

To show that 0 is an interior point of ̅̅̅̅̅̅̅̅̅̅

𝐓(𝑺𝒏𝟎 ) – 𝒚𝟎
Consider a mapping 𝑓: 𝐵′ → 𝐵′

By f(y)=y-𝑦0

Then f is one to one and onto

Hence 𝑓 −1 exist

Therefore 𝑓 −1 = 𝐵′ → 𝐵′ by𝑓 −1 (y)=y+𝑦0

Let 𝑦𝑛 є𝐵 ′ 𝑎𝑛𝑑 𝑦𝑛 → 𝑦𝜖𝐵 ′

Then f(𝑦𝑛 ) = 𝑦𝑛 − 𝑦0 → 𝑦 − 𝑦0 =f(y)

Hence f is continuous on 𝐵 ′

Then 𝑓 −1 (𝑦𝑛 ) = 𝑦𝑛 + 𝑦0 → 𝑦 + 𝑦0 =𝑓 −1 (y)

Hence 𝑓 −1 is continuous on 𝐵 ′

Hence 𝑓 −1 is one to one map from 𝐵 ′ onto 𝐵 ′ and both f and 𝑓 −1 are continuous

Hence 𝑓 −1 is a homomorphism

From claim(1)

𝑦0 be an interior point of ̅̅̅̅̅̅̅̅̅

T(𝑆𝑛0 ) then there exist an open set G such that 𝑦0 ∈ 𝐺 ⊂ ̅̅̅̅̅̅̅̅̅
T(𝑆𝑛0 ) hence

f(𝑦0 ) є𝑓(𝐺) ⊂ ̅̅̅̅̅̅̅̅̅

T(𝑆𝑛0 )

that implies

𝑦0 − 𝑦0 = 0 є𝑓(𝐺) ⊂ ̅̅̅̅̅̅̅̅̅
T(𝑆𝑛0 )

Hence 𝑓 is a homomorphism

f(G) is an open set in 𝐵 ′

hence 0 is an interior point of ̅̅̅̅̅̅̅̅̅

T(𝑆𝑛0 ) – 𝑦0

claim3:

̅̅̅̅̅̅̅̅̅
T(𝑆𝑛0 ) – 𝑦0 = ̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
T(𝑆𝑛0 ) − 𝑦0 ⊂ 2𝑛0 ̅̅̅̅̅̅̅̅̅
T(𝑆𝑛0 )and𝑆𝜖 ′ ⊂ ̅̅̅̅̅̅̅̅
T(𝑆1 ) for some 𝜖 > 0

Let y∈ T(𝑆𝑛0 ) − 𝑦0 then there exist x∈ 𝑆𝑛0 such that y=T(x)-𝑦0

By claim1
 𝑦0 ∈ 𝑆𝑛0

𝑦0 = 𝑇(𝑥0 )𝑓𝑜𝑟 𝑠𝑜𝑚𝑒𝑥0 ∈ 𝑆𝑛0

y=T(x)-T(𝑥0 )

y=T(x-𝑥0 )

where x,𝑥0 ∈ 𝑆𝑛0 𝑎𝑛𝑑 ||𝑥|| ≤ 𝑛0 , ||𝑥0 || ≤ 𝑛0

|| x-𝑥0 || ≤ ||𝑥|| + ||𝑥0 || ≤ 2𝑛0

⟹ 𝑥 − 𝑥0 ∈ 2𝑛0

T(x-𝑥0 ) ∈ 𝑆2𝑛0

y∈ 𝑆2𝑛0

hence if T(𝑆𝑛0 ) − 𝑦0 then y∈ 𝑆2𝑛0

T(𝑆𝑛0 ) − 𝑦0 ⊂ T(𝑆2𝑛0 )=2𝑛0 𝑇(𝑆1 )

̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
T(𝑆𝑛0 ) − 𝑦0 ⊂ 2𝑛0 ̅̅̅̅̅̅̅̅̅
T(𝑆𝑛0 )------------------------------(1)

Since f is homomorphism

T(𝑆𝑛0 ))=̅̅̅̅̅̅̅̅̅̅̅̅̅
f(̅̅̅̅̅̅̅̅̅ f(T(𝑆𝑛0 )) --------------------------------------(2)

by definition of f

f(̅̅̅̅̅̅̅̅̅
T(𝑆𝑛0 ))=̅̅̅̅̅̅̅̅̅
T(𝑆𝑛0 ) – 𝑦0

̅̅̅̅̅̅̅̅̅̅̅̅̅̅ ̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
f (T(𝑆𝑛0 )) = T(𝑆𝑛0 ) − 𝑦0

From(1)&(2)

̅̅̅̅̅̅̅̅̅
T(𝑆𝑛0 ) – 𝑦0 = ̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
T(𝑆𝑛0 ) − 𝑦0 ---------------------------------------(3)

From(1)&(3)

̅̅̅̅̅̅̅̅̅
T(𝑆𝑛0 ) – 𝑦0 = ̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅̅
T(𝑆𝑛0 ) − 𝑦0 ⊂ 2𝑛0 ̅̅̅̅̅̅̅̅
T(𝑆1 )

Claim(2),

0 is an interior point of ̅̅̅̅̅̅̅̅̅

T(𝑆𝑛0 ) – 𝑦0
0 is an interior point of 2𝑛0 ̅̅̅̅̅̅̅̅
T(𝑆1 )

0 is an interior point of̅̅̅̅̅̅̅̅

T(𝑆1 )

𝑺′𝝐 ⊂ ̅̅̅̅̅̅̅̅
𝐓(𝑺𝟏 ) 𝒇𝒐𝒓 𝒔𝒐𝒎𝒆 𝒑𝒐𝒔𝒊𝒕𝒊𝒗𝒆 𝝐

Claim4:

𝑺′𝝐 ⊂ 𝐓(𝑺𝟑 ) yє𝑆𝜖′

⟹ ||𝑦|| < 𝜖 𝑎𝑛𝑑 yє̅̅̅̅̅̅̅̅

T(𝑆1 )

⟹ 𝑦 𝑖𝑠 𝑎 𝑙𝑖𝑚𝑖𝑡 𝑝𝑜𝑖𝑛𝑡 𝑜𝑓 T(𝑆1 )

∈
Therefore there exist 𝑦1 ∈ T(𝑆1 )such that ||y-𝑦1 || <
2

𝑦1 ∈ T(𝑆1 ) ⟹ 𝑦1 = T(𝑥1 )for some 𝑥1 ∈ 𝑆1 with ||𝑥1 || = 1

From claim(3)

𝑆𝜖′ ⊂ ̅̅̅̅̅̅̅̅
T(𝑆1 )

̅̅̅̅̅̅̅̅̅
𝑆∈′ ⊂ T (𝑆1 )
2 2

∈
Since ||y-𝑦1 || < 2

̅̅̅̅̅̅̅̅̅
y-𝑦1 ∈ 𝑆∈′ ⊂ T (𝑆1 )
2 2

that implies

y-𝑦1 is a limit point of T (𝑆1 )

2

𝜖
therefore𝑦2 ∈ T (𝑆1 ) such that|| y-𝑦1 − 𝑦2 || < 22
2

𝑦2 ∈ T (𝑆1 )
2

1
=>𝑦2 = 𝑇(𝑥2 )and ||𝑥2 || < 2

1
Continuing in this way, we get a sequence{𝑥𝑛 } 𝑖𝑛 𝐵 𝑤𝑖𝑡ℎ ||𝑥𝑛 || < 2𝑛−1 and

𝜖
||𝑦 − (𝑦1 + 𝑦2 + ⋯ + 𝑦𝑛 )|| <
2𝑛
Hence 𝑦 = log 𝑛→∞ (𝑦1 + 𝑦2 + ⋯ + 𝑦𝑛 )

𝑆𝑛 = (𝑥1 + 𝑥2 + ⋯ + 𝑥𝑛 )

||𝑆𝑛 || = ||𝑥1 + 𝑥2 + ⋯ + 𝑥𝑛 ||<2 for every n=1,2,3….for n>m

||𝑆𝑛 − 𝑆𝑚 || = ||𝑥𝑚+1 + 𝑥𝑚+2 + ⋯ + 𝑥𝑛 ||

1 1
= 2( 𝑚
) = 𝑚−1 → 0 𝑎𝑠 𝑛 → ∞
2 2
Hence {𝑆𝑛 } is a Cauchy sequence in B

Since B is complete

log 𝑛→∞ 𝑆𝑛 = 𝑥

||𝑥|| = ||log 𝑛→∞ 𝑆𝑛 ||

=log 𝑛→∞ ||𝑆𝑛 ||

<2<3

 𝑥 ∈ 𝑆3
Now 𝑦1 + 𝑦2 + ⋯ + 𝑦𝑛 = 𝑇(𝑥1 ) + 𝑇(𝑥2 ) +…. 𝑇(𝑥𝑛 )
=T(𝑆𝑛 )
Therefore T is continuous

log 𝑛→∞ 𝑆𝑛 = 𝑥 => 𝑇(𝑥 ) = log 𝑛→∞ 𝑇( 𝑆𝑛 ) = 𝑦

Y=T(x) where ||x||<3

Therefore 𝑦 ∈ 𝑇(𝑆3 )

ℎ𝑒𝑛𝑐𝑒 𝑆𝜖′ ⊂ T(𝑆3 )-----------------------(4)

From 𝑆𝑟 = 𝑟𝑆1

We get,
1 1
𝑆𝜖 ′=3 𝑆𝜖′ ⊂ 3 T(𝑆3 ) = T(𝑆1 )
3

Hence T(𝑆1 )contains an open sphere centered at the origin in 𝑩′

Hence T(𝑆𝑟 )contains an open sphere centered at the origin in B’

Hence the image of each sphere centered at the origin in B contains an open sphere
centered at the origin in B’

Hence the theorem

State and prove the open mapping theorem:

If B and B’ are banach space and if T is continuous linear transformation of B onto B’

then T is an open mapping.

Proof:

Let T is continuous linear transformation of B onto B’

To prove that T is an open

We have to prove that for every open set G in B

T(G) is open in B’

Let G be an open set in B

Let y∈ 𝑇(𝐺)

Hence y=T(x) for some xєG

Since G is open, there exist an open sphere S(x,r)⊂ 𝐺

ie) x+𝑆𝑟 ⊂ 𝐺

where𝑆𝑟 is the open sphere centered at the origin in B

there fore by the above lemma ,there exist an open sphere 𝑆𝜖 ′in B’ such that 𝑆𝜖 ′ ⊂ 𝑇(𝑆𝑟 )

therefore y+𝑆𝜖′ = 𝑦 + 𝑇(𝑆𝑟 )

= 𝑇(𝑥 + 𝑆𝑟 )

𝑆 ′(𝑦; 𝜖) ⊂ 𝑇(𝑥 + 𝑆𝑟 )

Hence
𝑆 ′(𝑦; 𝜖) ⊂ 𝑇(𝑆 (𝑥; 𝑟)) ⊂ 𝑇(𝐺)

Hence for every y∈ 𝑇(𝐺)

There exist an open sphere in B’ centered at y and contained in T(G)

Hence T(G) is open in B’

Hence T is an open mapping

Hence proved.