BOUNDED LINEAR MAP

KRISHNA KUMAR. G
DEPARTMENT OF MATHEMATICS
UNIVERSITY OF KERALA

1. Continuity of linear maps

Theorem 1.1. Let X and Y be normed linear spaces.

(1) If X is finite dimensional, then every linear map from X to Y is continuous.
(2) If X is infinite dimensional and Y 6= 0, then there is a linear map from X to Y which
is discontinuous.

Proof. (1) Let X be finite dimensional and F : X → Y be linear. If X = {0}, then

there is nothing to prove. Let now X 6= {0} and {a1 , . . . , am } be a basis for X. For
a sequence (xn ) in X, let

xn = kn,1 a1 + · · · + kn,m am ,

where kn,j ∈ C, j = 1, . . . , m. If xn → x = k1 a1 + · · · + km am . Then kn,j → kj for

each j = 1, . . . , m, and

F (xn ) = kn,1 F (a1 ) + · · · + kn,m F (am )

→ k1 F (a1 ) + · · · + km F (am ) = F (x),

by the linearity of F and the continuity of addition and scalar multiplication of Y .

Thus every linear map from X to Y is continuous.
(2) Assume that X is infinite dimensional. Let {a1 , a2 , . . .} be an infinite linearity in-
dependent subset of X. Let xn = an /nkan k for n = 1, 2, . . .. Then {x1 , x2 , . . .} is
an infinite linearity independent subset of X and kxn k = 1/n for all n. Then there
exists a basis E of X such that L ⊆ B. Since Y 6= 0, consider 0 6= b ∈ Y , define
(
b, if x ∈ L
F (x) =
0, if x ∈ B, x ∈
/ L,
and extend F linearly to all of X. Since xn → 0 as n → ∞, but F (xn ) = b 6= 0 for
all n, F is not continuous. Thus there exists a linear map from X to Y which is not
continuous.
1
2 KRISHNA KUMAR. G

Theorem 1.2. Let X and Y be normed spaces and F : X → Y be a linear map. The
following conditions are equivalent.
(i) F is bounded on U (0, r)for some r > 0.
(ii) F is continuous at 0.
(iii) F is continuous on X.
(iv) F is uniformly continuous on X.
(v) kF (x)k ≤ αkxk for all x ∈ X and some α > 0.
(vi) The zero space Z(F ) of F is closed in X and the linear map F̃ : X/Z(F ) → Y defined
by F̃ (x + Z(F )) = F (x), x ∈ X, is continuous.

Proof. We show that (i) implies (v). The successive reverse implications are immediate. Let
kF (x)k ≤ β for all x ∈ U (0, r), r > 0. If x = 0, then F (x) = 0, and if x 6= 0, then since
rx/kxk is in U (0, r),  
kxk rx β
kF (x)k = F ≤ kxk.
r kxk r
Next, we show that F is continuous on X if and only if (vi) holds ((v) if and only if (vi)).
Let F be continuous on X. Then Z(F ) = F −1 {0} is closed in X, since {0} is closed in Y .
Hence X/Z(F ) is a normed space in the quotient norm. It can be easily seen that the map
F̃ : X/Z(F ) → Y given by F̃ (x + Z(F )) = F (x), x ∈ X, is well defined. By (v), there is
some α > 0 such that kF (x)k ≤ αkxk for all x ∈ X. Now let x ∈ X and z ∈ Z(F ). Then

kF̃ (x + Z(F ))k = kF̃ (x + z + Z(F ))k = kF (x + z)k ≤ αkx + zk.

Since this is true for every z ∈ Z(F ), we see that

kF̃ (x + Z(F ))k ≤ α inf{kx + zk : z ∈ Z(F )} = α|||x + Z(F )|||.

Hence F̃ is continuous. Conversely, assume that Z(F ) is closed and the linear map F̃ is
continuous. Then for some α > 0, we have

kF (x)k = F̃ (x + Z(F ))k ≤ α|||x + Z(F )||| ≤ αkxk, x ∈ X.

Hence F is continuous. 

Definition 1.3. Let X, Y be normed spaces and F : X → Y be linear. If

1. F is continuous;
 CONTINUOUS LINEAR MAPS 3

2. F is one-one;
3. F −1 : R(F ) → X is continuous;
then F is called a linear homeomorphism.

Corollary 1.4. A linear map F from a normed space X to a normed space Y is a homeo-
morphism if and only if there are α, β > 0 such that

βkxk ≤ kF (x)k ≤ αkxk

for all x ∈ X. In case there is a linear homeomorphism from X onto Y , then X is complete
if and only if Y is complete.

Proof. Let F be a linear map from X to Y and

βkxk ≤ kF (x)k ≤ αkxk

for some α, β > 0 and all x ∈ X. Then F is clearly injective and it is continuous. The
inverse map F −1 : R(F ) → X is automatically linear and kF −1 (y)k ≤ kyk/β for all y ∈ Y .
Hence F −1 is also continuous, showing that F is a homeomorphism.
Conversely, let F be a linear homeomorphism from X to Y . Then there exist α, γ > 0 such
that
kF (x)k ≤ αkxk and kF −1 (y)k ≤ γkyk
for all x ∈ X and y ∈ Y . Let x ∈ X and y = F (x), then x = F −1 (y), that is,
kxk
≤ kF (x)k ≤ αkxk
γ
for all x ∈ X, as desired (β = 1/γ).
Now suppose that there is a linear homeomorphism from X onto Y . Then (xn ) is a Cauchy
sequence in X (resp. (xn ) converges in X) if and only if (F (xn )) is a Cauchy sequence in Y
(resp. (f (xn )) converges in Y ). Hence X is complete if and only if Y is complete. 

Corollary 1.5. Let X and Y be normed spaces with X finite dimensional. Then every
bijective linear map from X to Y is a homeomorphism. All norms on X are equivalent and
X is complete in each. If the dimension of X is N , then there is a linear homeomorphism
from CN onto X.

Proof. Let X be finite dimensional and F : X → Y is linear. Then F is continuous (1)

of Theorem 1.1. If F is bijective, then Y is also finite dimensional and F −1 is linear and
4 KRISHNA KUMAR. G

continuous, that is, F is a homeomorphism.

If k · k and k · k0 are norms on a finite dimensional linear space X, then since the identity
map I : X → X is obviously linear and bijective, we have

βkxk ≤ kI(x)k0 = kxk0 ≤ αkxk

for all x ∈ X. Hence the norms k · k and k · k0 are equivalent.

If X has dimension N and {x1 , . . . , xN } be a basis of X. Define F : CN → X

F (α1 , . . . , αN ) = α1 x1 + · · · + αN xN .

Then F is a linear bijective homeomorphism between CN and X. 

Corollary 1.6. Let X and Y be normed spaces and F : X → Y be a linear map such that
the range R(F ) of F is finite dimensional. Then F is continuous if and only if the zero space
Z(F ) of F is closed in X.
In particular, a linear functional f on X is continuous if and only if Z(f ) is closed in X.

Proof. If R(F ) = {0}, then there is nothing to prove. Let R(F ) 6= {0} and {y1 , . . . , ym } be
a basis of R(F ). Let xj ∈ X be such that F (xj ) = yj , j = 1, . . . , m. Then

span{x1 + Z(F ), . . . , xm + Z(F )} = X/Z(F ).

This can be seen as follows. If x ∈ X and F (x) = k1 y1 + · · · + km ym , then x − k1 x1 − · · · −

km xm ∈ Z(F ). Hence

x + Z(F ) = (k1 x1 + · · · + km xm ) + Z(F )

= k1 (x1 + Z(F )) + · · · + km (xm + Z(F )).

In particular X/Z(F ) is finite dimensional.

If Z(F ) is closed in X, then X/Z(F ) is a finite dimensional normed space. Now Theorem
1.1 shows that the linear map F̃ : X/Z(F ) → Y given by F̃ (x + Z(F )) = F (x), x ∈ X is
continuous. By Theorem 1.2, F is continuous if and only if Z(F ) is closed in X.
If f is a linear functional on X, then Y = C and R(f ) is of dimension less than or equal to
1. Hence f is continuous if and only if Z(f ) is closed. 

Example 1.7. An important example of a discontinuous linear map is given by the

operation differentiation. Let X = C 0 [0, 1], the space of all continuously differentiable scalar
valued functions on [0, 1] and Y = C[0, 1], the space of all continuous scalar valued functions
 CONTINUOUS LINEAR MAPS 5

on [0, 1], both with the sup norms. For x ∈ X, let F (x) = x0 , the derivative of x on [0, 1].
Then F is linear but not continuous, since xn (t) = tn , t ∈ [0, 1], then kxn k∞ = 1 while
kF (xn )k = kx0n k = knxn−1 k∞ = n, so that the condition (v) of Theorem 1.2 is violated.
Note that Z(F ) consists of the set of all constant functions on [0, 1], which is closed in X,
and that R(F ) = Y , which is infinite dimensional. This shows that we cannot drop the
assumption of the finite dimensionality of R(F ) from Corollary 1.6.
Define f : X → C by f (x) = x0 (1), x ∈ X. Again, the linear functional f is not continuous.
By Corollary 1.6, Z(f ) = {x ∈ X : x0 (1) = 0} cannot be closed. This also follows directly by
considering zn (t) = t − tn /n, t ∈ [0, 1], and noting that zn → z where z(t) = t and zn ∈ Z(f )
but z ∈
/ Z(f ).

2. Bounded Linear Maps

A linear map from a normed space X to a normed space Y is continuous if and only
if it maps bounded set in X onto bounded sets in Y . Hence such a map is known as a
bounded linear map. The set of all such maps will be denoted by BL(X, Y ). A linear
map from X to itself is called an operator on X. We write BL(X) for the set BL(X, X)
of all bounded operators on X. Also, we write X 0 for the set BL(X, C) of all bounded
linear functionals on X.
If F ∈ BL(X, Y ) and F 6= 0 then F is not bounded on X in the usual sense of the word
’bounded’. However, we have
kF (x)k ≤ αkxk,
for all x ∈ X and some α > 0

Definition 2.1. A linear map F : X → Y is called bounded below if

βkxk ≤ kF (x)k,

for all x ∈ X and some β > 0.

Thus a linear map F is a homeomorphism if and only if it is bounded and bounded below.
It can be easily seen that BL(X, Y ) is a linear space under the pointwise operations. For
x ∈ X, we have
(F + G)(x) = F (x) + G(x).
(αF )(x) = α F (x).
6 KRISHNA KUMAR. G

Theorem 2.2. Let X, Y be normed spaces. For F ∈ BL(X, Y ), define

kF k = sup{kF (x)k : x ∈ X, kxk ≤ 1}.

Then k · k is a norm on BL(X, Y ), called the operator norm. For all x ∈ X, we have

kF (x)k ≤ kF kkxk.

Proof. Recall for F, G ∈ BL(X, Y ) and α ∈ C,

(F + G)(x) = F (x) + G(x), (αF )(x) = α F (x).

The zero element of BL(X, Y ) is denoted by “000 and is defined as

0(x) = 0 for every x ∈ X.

(1) From the definition, kF k ≥ 0 for every F ∈ BL(X, Y ).

Also k0k = 0.
Assume kF k = 0, then sup{kF (x)k : x ∈ X, kxk ≤ 1} = 0. We claim that F (x) = 0
for every x ∈ X. Since F (0) = 0 and for x 6= 0 we have x/kxk is of norm 1. Hence
x F (x)
F ( kxk )= kxk
= 0 implies F (x) = 0.
(2) From the definition, it follows that kF + Gk ≤ kF k + kGk.
(3) From the definition, if follows that kαF k = |α|kF k.
x
For x 6= 0, we have kxk
= 1 and
 
x
F ≤ kF k
kxk
implies

kF (x)k ≤ kF kkxk.

For x = 0, F (x) = 0 and kF (x)k ≤ kF kkxk holds. 

Theorem 2.3. Let X, Y be normed spaces and X 6= 0. For F ∈ BL(X, Y ), then

(1) kF k = inf{α > 0 : kF (x)k ≤ αkxk for all x ∈ X}.

(2) kF k = sup{kF (x)k : x ∈ X, kxk = 1}.
(3) kF k = sup{kF (x)k : x ∈ X, kxk < 1}.
 CONTINUOUS LINEAR MAPS 7

Proof. Denote

α0 = inf{α > 0 : kF (x)k ≤ αkxk for all x ∈ X}

β = sup{kF (x)k : x ∈ X, kxk = 1}

γ = sup{kF (x)k : x ∈ X, kxk < 1}

Then clearly β ≤ kF k and γ ≤ kF k. Consider a nonzero x ∈ X and 0 < r ≤ 1. Since F is

linear, we have
 
rx kxk kxk
kF (x)k = kF k ≤ sup{kF (z)k : z ∈ X, kzk = r} .
kxk r r
If r = 1, then kF (x)k ≤ βkxk. Hence α0 ≤ β. If r < 1, then
 
rx kxk kxk γkxk
kF (x)k = kF k ≤ sup{kF (z)k : z ∈ X, kzk < 1} = .
kxk r r r
Letting r → 1, we see that kF (x)k ≤ γkxk for all x ∈ X. Hence α0 ≤ γ.
Finally we show that kF k ≤ α0 . Consider α ≥ 0 such that kF (x)k ≤ αkxk for all x ∈ X.
Taking supremum over all x ∈ X with kxk ≤ 1, we obtain kF k ≤ α. Since α0 is infimum of
all such α0 s, we obtain kF k ≤ α0 .
Thus kF k ≤ α0 ≤ min{β, γ} ≤ kF k and the proof is complete. 

Department of Mathematics, University of Kerala, Kariavattom Campus, Thiruvanan-

thapuram, Kerala, India, 695581.
E-mail address: krishna.math@gmail.com