Contents

1 Normed Spaces. Banach Spaces. 2

1.1 Vector Space. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2
1.2 Normed Space. Banach Space. . . . . . . . . . . . . . . . . . . . . . . . . 10
1.3 Further Properties of Normed Spaces. . . . . . . . . . . . . . . . . . . . . 16
1.4 Finite Dimensional Normed Spaces and Subspaces. . . . . . . . . . . . . 26
1.5 Linear Operators. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
1.6 Bounded and Continuous Linear Operators. . . . . . . . . . . . . . . . . 38
1.7 Linear Functionals. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 46
1.8 Linear Operators and Functionals on Finite Dimensional Spaces. . . . . . 55
1 Normed Spaces. Banach Spaces.
1.1 Vector Space.
Definition 1.1.
1. An arbitrary subset M of a vector space X is said to be linearly independent if
every non-empty finite subset of M is linearly independent.
2. A vector space X is said to be finite dimensional if there is a positive integer
n such that X contains a linearly independent set of n vectors whereas any set of
n + 1 or more vectors of X is linearly dependent. n is called the dimension of X,
written n = dim X.
3. If X is any vector space, not necessarily finite dimensional, and B is a linearly
independent subset of X which spans X, then B is called a basis (or Hamel
basis) of X.
• Hence if B is a basis for X, then every nonzero x 2 X has a unique repre-
sentation as a linear combination of (finitely many!) elements of B with
nonzero scalars as coefficients.

Theorem 1.2. Let X be an n-dimensional vector space. Then any proper subspace Y of
X has dimension less than n.

1. Show that the set of all real numbers, with the usual addition and multiplication,
constitutes a one-dimensional real vector space, and the set of all complex numbers
constitutes a one-dimensional complex vector space.

Solution: The usual addition on R and C are commutative and associative, while
scalar multiplication on R and C are also associative and distributive. For R, the
zero vector is 0R = 0 2 R, the identity scalar is 1R = 1 2 R, and the additive
inverse is x for any x 2 R. For C, the zero vector is 0C = 0 + 0i 2 C, the
identity scalar is 1C = 1 + 0i 2 C and the additive inverse is z for all z 2 C.

2. Prove that 0x = 0, ↵0 = 0 and ( 1)x = x.

Solution:
0x = (0 + 0)x = 0x + 0x =) 0 = 0x + ( (0x))
= 0x + 0x + ( (0x))
= 0x + 0 = 0x.
↵0 = ↵(0 + 0) = ↵0 + ↵0 =) 0 = ↵0 + ( (↵0))
= ↵0 + ↵0 + ( (↵0))
= ↵0 + 0 = ↵0.
( 1)x = ( 1(1))x = 1(1x) = x.

Page 2
3. Describe the span of M = {(1, 1, 1), (0, 0, 2)} in R3 .

Solution: The span of M is

8 2 3 2 3 9
< 1 0 =
4
span M = ↵ 1 + 5 4 5
0 : ↵, 2 R
: ;
1 2
82 3 9
< ↵ =
= 4 ↵ 5 : ↵, 2 R .
: ;
↵+2

We see that span M corresponds to the plane x = y on R3 .

4. Which of the following subsets of R3 constitute a subspace of R3 ? Here, x =

(⇠1 , ⇠2 , ⇠3 ).
(a) All x with ⇠1 = ⇠2 and ⇠3 = 0.

Solution: For any x, y 2 W and any ↵, 2 R,

2 3 2 3 2 3
⇠1 ⌘1 ↵⇠1 + ⌘1
↵x + y = ↵ 4⇠2 5 + 4⌘2 5 = 4↵⇠2 + ⌘2 5 2 W.
0 0 0

(b) All x with ⇠1 = ⇠2 + 1.

2 3 2 3
2 3
Solution: Choose x1 = 1 2 W , x2 = 25 2 W , then
4 5 4
0 0
2 3 2 3 2 3
2 3 5
x1 + x2 = 1 + 2 = 3 5 2
4 5 4 5 4 /W
0 0 0

since 5 6= 3 + 1.

(c) All x with positive ⇠1 , ⇠2 , ⇠3 .

Solution: Choose ↵ = 1, then for any x 2 W , ↵x 2

/ W.

(d) All x with ⇠1 ⇠2 + ⇠3 = k.

Solution: For any x, y 2 W ,

2 3
⇠ 1 + ⌘1
x + y = 4 ⇠ 2 + ⌘2 5 .
⇠ 3 + ⌘3

Page 3
 Since

⇠ 1 + ⌘1 (⇠2 + ⌘2 ) + (⇠3 + ⌘3 ) = (⇠1 ⇠2 + ⇠3 ) + (⌘1 ⌘2 + ⌘3 ) = 2k.

we see that W is a subspace of R3 if and only if k = 0.

5. Show that {x1 , . . . , xn }, where xj (t) = tj , is a linearly independent set in the space
C[a, b].

Solution: This is a simple consequence of Fundamental Theorem of Alge-

bra. Fix a finite n > 1. Suppose that for all t 2 [a, b], we have
n
X n
X
j
j xj (t) = jt = 0.
j=1 j=1

Suppose n 6= 0. Fundamental Theorem of Algebra states that any polynomials

with degree n can have at most n real roots. Since the equation above is true for
n
X
j
all t 2 [a, b], and the set of points in the interval [a, b] is uncountable, j t has
j=1
to be the zero polynomial. Since n 1 was arbitrary (but finite), this shows that
any non-empty finite subset of {xj }j2⇤ , ⇤ a countable/uncountable indexing set,
is linearly independent.

6. Show that in an n-dimensional vector space X, the representation of any x as a linear

combination of a given basis vectors e1 , . . . , en is unique.

Solution: Let X is an n-dimensional vector space, with a basis {e1 , . . . , en }.

Suppose any x 2 X has a representation as a linear combination of the basis
vectors, we claim that the representation is unique. Indeed, if x 2 X has two
representations

x = ↵1 e1 + . . . + ↵n en = 1 e1 + ... + n en .

subtracting them gives

n
X
(↵1 1 )e1 + . . . + (↵n n )en = (↵j j )ej = 0.
j=1

Since {e1 , . . . , en } is a basis of X, by definition it is linearly independent, which

implies that ↵j j = 0 for all j = 1, . . . , n, i.e. the representation is unique.

7. Let {e1 , . . . , en } be a basis for a complex vector space X. Find a basis for X regarded
as a real vector space. What is the dimension of X in either case?

Page 4
 Solution: A basis for X regarded as a real vector space is {e1 , . . . , en , ie1 , . . . , ien }.
The dimension of X is n as a complex vector space and 2n as a real vector space.

8. If M is a linearly dependent set in a complex vector space X, is M linearly dependent

in X, regarded as a real vector space?



1 i
Solution: No. Let X = C2 , with K = C, and consider x = and y = .
i 1
{x, y} is a linearly dependent set in X since ix = y. Now suppose K = R, and
 
↵+ i 0 + 0i
↵x + y = =0= .
↵i 0 + 0i

Since ↵, can only be real numbers, we see that (↵, ) = (0, 0) is the only solution
to the equation. Hence {x, y} is a linearly independent set in X = C2 over R.

9. On a fixed interval [a, b] ⇢ R, consider the set X consisting of all polynomials with
real coefficients and of degree not exceeding a given n, and the polynomial x = 0 (for
which a degree is not defined in the usual discussion of degree).
(a) Show that X, with the usual addition and the usual multiplication by real num-
bers, is a real vector space of dimension n + 1. Find a basis for X.

Solution: Let X be the set given in the problem. It is clear that X is

a real vector space. Indeed, for any P, Q 2 X, with deg(P ), deg(Q)  n,
deg(P + Q)  n and deg(↵P )  n for any ↵ 2 R. A similar argument from
Problem 5 shows that {1, t, t2 , . . . , tn } is a linearly independent set in X, and
since {1, t, t2 , . . . , tn } spans X, it is a basis of X and X has dimension n + 1.

(b) Show that we can obtain a complex vector space X̃ in a similar fashion if we let
those coefficients be complex. Is X a subspace of X̃?

Solution: No. Consider P (t) = t 2 X, choose ↵ = i, then ↵P (t) = it 2

/ X.

10. If Y and Z are subspaces of a vector space X, show that Y \ Z is a subspace of X,

but Y [ Z need not be one. Give examples.

Solution: Let Y and Z be subspaces of a vector space X. Take any x, y 2 Y \Z,

note that x, y are both elements of Y and Z. For any ↵, 2 K, ↵x + y 2 Y
(since Y is a subspace of X) and ↵x+ y 2 Z (since Z is a subspace of X). Hence
↵x + y 2 Y \ Z.

Page 5
 ⇢ ⇢
↵ 0
For the second part, consider Y = : ↵ 2 R and Z = : ↵ 2 R . It
0 ↵
 subspaces of R , but Y [ Z is not a
2
can be (easily) deduced that Y and Z are
1 0 1
subspace of R2 . To see this, choose x = and y = , then x + y = 2
/
0 1 1
Y [ Z.

11. If M 6= ? is any subset of a vector space X, show that span M is a subspace of X.

Solution: This is immediate since a (scalar) field K is closed under addition and
sums of two finite sums remain finite.

12. (a) Show that the set of all real two-rowed square matrices forms a vector space X.
What is the zero vector in X?

Solution: This follows from Problem 1 and the definition of matrix addition
 prove that R is a vector space over R
and matrix scalar multiplication: we
0 0
or C. The zero vector in X is 0 = .
0 0

(b) Determine dim X. Find a basis for X.

Solution: We claim that dim X = 4. To prove this, consider the following

four vectors in X
   
1 0 0 1 0 0 0 0
e1 = e2 = e3 = e4 = .
0 0 0 0 1 0 0 1

↵ ↵
Suppose ↵1 e1 + . . . + ↵4 e4 = 0 = 1 2 , we have ↵1 = ↵2 = ↵3 = ↵4 = 0,
↵3 ↵4
i.e. {e1 , e2 , e3 , e4 } is a linearly independent set in X. However,
 any set of 5
a b
or more vectors of X is linearly dependent, since any x = 2 X can be
c d
written as a linear combination of {e1 , e2 , e3 , e4 }, i.e. x = ae1 +be2 +ce3 +de4 .
Hence, a basis for X is {e1 , e2 , e3 , e4 }.

(c) Give examples of subspaces of X.

⇢
↵ 0
Solution: An example is W = :↵2R .
0 0

(d) Do the symmetric matrices x 2 X form a subspace of X?

Page 6
  
a1 b 1 a b
Solution: Yes. Consider any symmetric matrices x = ,y = 2 2 .
b1 d 1 b2 d 2
For any ↵, 2 R (or C),

↵a1 + a2 ↵b1 + b2
↵x + y =
↵b1 + b2 ↵d1 + d2

which is a symmetric matrix.

(e) Do the singular matrices x 2 X form a subspace of X?

 
1 1 3 2
Solution: No. To see this, consider x = and y = ; both
1 1 6 4
x and y are singular matrices since they have zero determinant. However,
4 3
x+y = is not a singular matrix since det(x + y) = 20 21 = 1 6= 0.
7 5

13. (Product) Show that the Cartesian product X = X1 ⇥ X2 of two vector spaces over
the same field becomes a vector space if we define the two algebraic operations by

(x1 , x2 ) + (y1 , y2 ) = (x1 + y1 , x2 + y2 ),

↵(x1 , x2 ) = (↵x1 , ↵x2 ).

Solution: This is a simple exercise. We first verify vector addition:

(x1 , x2 ) + (y1 , y2 ) = (x1 + y1 , x2 + y2 )

= (y1 + x1 , y2 + x2 )
= (y1 , y2 ) + (x1 , x2 ).
⇣ ⌘
(x1 , x2 ) + [(y1 , y2 ) + (z1 , z2 )] = x1 + (y1 + z1 ), x2 + (y2 + z2 )
⇣ ⌘
= (x1 + y1 ) + z1 , (x2 + y2 ) + z2
= (x1 + y1 , x2 + y2 ) + (z1 , z2 )
h i
= (x1 , x2 ) + (y1 , y2 ) + (z1 , z2 ).
(x1 , x2 ) = (x1 + 0, x2 + 0)
= (x1 , x2 ) + (0, 0).
(0, 0) = (x1 + ( x1 ), y1 + ( y1 ))
= (x1 , y1 ) + ( x1 , y1 ).

Next, we verify scalar vector multiplication:

(x1 , x2 ) = (1K x1 , 1K x2 )
= 1K (x1 , x2 ).

Page 7
 h i
↵ (x1 , x2 ) = ↵( x1 , x2 )
⇣ ⌘
= ↵( x1 ), ↵( x2 )
⇣ ⌘
= (↵ )x1 , (↵ )x2
= (↵ )(x1 , x2 ).
h i
↵ (x1 , x2 ) + (y1 , y2 ) = ↵(x1 + y1 , x2 + y2 )
⇣ ⌘
= ↵(x1 + y1 ), ↵(x2 + y2 )
= (↵x1 + ↵y1 , ↵x2 + ↵y2 )
= (↵x1 , ↵y1 ) + (↵x2 , ↵y2 )
= ↵(x1 , y1 ) + ↵(x2 , y2 ).
⇣ ⌘
(↵ + )(x1 , x2 ) = (↵ + )x1 , (↵ + )x2
= (↵x1 + x1 , ↵x2 + x2 )
= (↵x1 , ↵x2 ) + ( x1 , x2 )
= ↵(x1 , x2 ) + (x1 , x2 ).

14. (Quotient space, codimension) Let Y be a subspace of a vector space X. The

coset of an element x 2 X with respect to Y is denoted by x + Y and is defined to
be the set
x + Y = {x : v = x + y, y 2 Y }.
(a) Show that the distinct cosets form a partition of X.

Solution:

(b) Show that under algebraic operations defined by

(w + Y ) + (x + Y ) = (w + x) + Y
↵(x + Y ) = ↵x + Y

these cosets constitute the elements of a vector space. This space is called the
quotient space (or sometimes factor space) of X by Y (or modulo Y ) and is
denoted by X/Y . Its dimension is called the codimension of Y and is denoted
by codim Y , that is,
codim Y = dim (X/Y ).

Solution:

15. Let X = R3 and Y = {(⇠1 , 0, 0) : ⇠1 2 R}. Find X/Y , X/X, X/{0}.

Page 8
Solution: First, X/X = {x + X : x 2 X}; since x + X 2 X for any x 2 X,
we see that X/X = {0}. Next, X/{0} = {x + 0 : x 2 X} = {x : x 2 X} = X.
For X/Y , we are able to deduce (geometrically) that elements of X/Y are lines
parallel to the ⇠1 -axis. More precisely, by definition,X/Y = {x + Y : x 2 X}; for
a fixed x0 = (⇠10 , ⇠20 , ⇠30 ),

x0 + Y = {(⇠10 , ⇠20 , ⇠30 ) + (0, 0, ⇠3 ) : ⇠3 2 R}

= {(⇠ 0 , ⇠ 0 , ⇠˜3 ) : ⇠˜3 2 R}.
1 2

which corresponds to a line parallel to ⇠1 -axis.

Page 9
1.2 Normed Space. Banach Space.
Definition 1.3. A norm on a (real or complex) vector space X is a real-valued function
on X whose value at an x 2 X is denoted by kxk and which has the properties

(N1) kxk 0.

(N2) kxk = 0 () x = 0.

(N3) k↵xk = |↵|kxk.

(N4) kx + yk  kxk + kyk. (Triangle inequality)

Here, x and y are arbitrary vectors in X and ↵ is any scalar.

• A norm on X defines a metric d on X which is given by

d(x, y) = kx yk , x, y, 2 X

and is called the metric induced by the norm.

• The norm is continuous, that is, x 7! kxk is a continuous mapping of (X, k·k) into R.

Theorem 1.4. A metric d induced by a norm on a normed space X satisfies

(a) d(x + a, y + a) = d(x, y).

(b) d(↵x, ↵y) = |↵|d(x, y).

for all x, y, a 2 X and every scalar ↵.

• This theorem illustrates an important fact: Every metric on a vector space might
not necessarily be obtained from a norm.

• A counterexample is the space s consisting of all (bounded or unbounded) sequences

of complex numbers with a metric d defined by
X1
1 |⇠j ⌘j |
d(x, y) = .
j=1
2j 1 + |⇠j ⌘j |

Page 10
1. Show that the norm kxk of x is the distance from x to 0.

Solution: kxk = kx yk = d(x, 0), which is precisely the distance from x to

y=0
0.

2. Verify that the usual length of a vector in the plane or in three dimensional space
has the properties (N1) to (N4) of a norm.

p
Solution: For all x 2 R3 , define kxk = x21 + x22 + x23 . (N1) to (N3) are obvious.
(N4) is an easy consequence of the Cauchy-Schwarz inequality for sums.
More precisely, for x = (x1 , x2 , x3 ) and y = (y1 , y2 , y3 ) we have:

kx + yk2 = (x1 + y1 )2 + (x2 + y2 )2 + (x3 + y3 )2

= (x21 + x22 + x23 ) + (y12 + y22 + y32 ) + 2(x1 y1 + x2 y2 + x3 y3 )
 kxk2 + kyk2 + 2kxkkyk = (kxk + kyk)2 .

Taking square root on both sides yields (N4).

3. Prove (N4) implies kyk kxk  ky xk.

Solution: From triangle inequality of norm we have the following two inequali-
ties:

kxk = kx y + yk  kx yk + kxk
=) kxk kyk  kx yk.
kyk = ky x + xk  ky xk + kxk
=) kyk kxk  kx yk.

Combining these two yields the desired inequality.

4. Show that we may replace (N2) by kxk = 0 =) x = 0 without altering the concept
of a norm. Show that nonnegativity of a norm also follows from (N3) and (N4).

Solution: For any x 2 X,

kxk = kx + x xk  kx + xk + k xk = 2kxk + kxk = 3kxk

=) 0  2kxk =) 0  kxk.

n
!1/2
X p
2
5. Show that kxk = |⇠j | = |⇠1 |2 + . . . + |⇠n |2 defines a norm.
j=1

Page 11
 Solution: (N1) to (N3) are obvious. (N4) follows from the Cauchy-Schwarz
inequality for sums, the proof is similar to that in Problem 3.

6. Let X be the vector space of all ordered pairs x = (⇠1 , ⇠2 ), y = (⌘1 , ⌘2 ), · · · of real
numbers. Show that norms on X are defined by
kxk1 = |⇠1 | + |⇠2 |
kxk2 = (⇠12 + ⇠22 )1/2
kxk1 = max{|⇠1 |, |⇠2 |}.

Solution: (N1) to (N3) are obvious for each of them. To verify (N4), for x =
(⇠1 , ⇠2 ) and y = (⌘1 , ⌘2 ),

kx + yk1 = |⇠1 + ⌘1 | + |⇠2 + ⌘2 |

 |⇠1 | + |⌘1 | + |⇠2 | + |⌘2 | = kxk1 + kyk1 .

kx + yk22 = (⇠1 + ⌘1 )2 + (⇠2 + ⌘2 )2

= (⇠12 + ⇠22 ) + (⌘12 + ⌘22 ) + 2(⇠1 ⌘1 + ⇠2 ⌘2 )
 kxk2 + kyk2 + 2kxk2 kyk2 = (kxk2 + kyk2 )2
=) kx + yk2  kxk2 + kyk2 .

kx + yk1 = max{|⇠1 + ⌘1 |, |⇠2 + ⌘2 |}

 max{|⇠1 | + |⌘1 |, |⇠2 | + |⌘2 |}
 max{|⇠1 |, |⇠2 |} + max{|⌘1 |, |⌘2 |} = kxk1 + kyk1 .

where we use the inequality |a|  max{|a|, |b|} for any a, b 2 R.

1
!1/p
X
7. Verify that kxk = |⇠j |p satisfies (N1) to (N4).
j=1

Solution: (N1) to (N3) are obvious. (N4) follows from Minkowski inequality
for sums. More precisely, for x = (⇠j ) and y = (⌘j ),

1
! p1 1
! p1 1
! p1
X X X
kx + yk = |⇠j + ⌘j |p  |⇠k |p + |⌘m |p .
j=1 k=1 m=1

8. There are several norms of practical importance on the vector space of ordered n-
tuples of numbers, notably those defined by
kxk1 = |⇠1 | + |⇠2 | + . . . + |⇠n |

Page 12
 ⇣ ⌘1/p
p p p
kxkp = |⇠1 | + |⇠2 | + . . . + |⇠n | (1 < p < +1)
kxk1 = max{|⇠1 |, . . . , |⇠n |}.
In each case, verify that (N1) to (N4) are satisfied.

Solution: This is a generalisation of Problem 6, with the proof being almost

identicall. The only thing that di↵ers is we use Minkowski inequality for
sums to prove (N4) for k · kp .

9. Verify that kxk = max |x(t)| defines a norm on the space C[a, b].
t2[a,b]

Solution: (N1) and (N2) are clear, as we readily see. For (N3), for any scalars
↵ we have:
k↵xk = max |↵x(t)| = |↵| max |x(t)| = |↵|kxk.
t2[a,b] t2[a,b]

Finally, for (N4),

kx + yk = max |x(t) + y(t)|  max |x(t)| + max |y(t)| = kxk + kyk.

t2[a,b] t2[a,b] t2[a,b]

10. (Unit Sphere) The sphere

S1 (0) = {x 2 X : kxk = 1}.
in a normed space X is called the unit sphere. Show that for the norms in Problem
6 and for the norm defined by kxk4 = (⇠14 + ⇠24 )1/4 , the unit spheres look as shown in
figure.

Solution: Refer to Kreyszig, page 65.

11. (Convex set, segment) A subset A of a vector space X is said to be convex if

x, y 2 A implies
M = {z 2 X : z = ↵x + (1 ↵)y, 0  ↵  1} ⇢ A.
M is called a closed segment with boundary points x and y; any other z 2 M is called
an interior point of M . Show that the closed unit ball in a normed space X is convex.

Solution: Choose any x, y 2 B̃1 (0), then for any 0  ↵  1,

k↵x + (1 ↵)yk  ↵kxk + (1 ↵)kyk  ↵ + 1 ↵ = 1.

This shows that the closed unit ball in X is convex.

Page 13
12. Using Problem 11, show that
⇣p p ⌘2
(x) = |⇠1 | + |⇠2 |

does not define a norm on the vector space of all ordered pairs x = (⇠1 , ⇠2 ) of real
numbers. Sketch the curve (x) = 1.

Solution: Problem 11 shows that if is a norm, then the closed unit ball in a
normed space X = (X, ) is convex. Choose x = (1, 0) and y = (0, 1), x, y are
elements of the closed unit ball in (X, ) since (x) = (y) = 1. However, if we
choose ↵ = 0.5,
⇣p p ⌘2 p
(0.5x + 0.5y) = |0.5| + |0.5| = (2 0.5)2 = 2 > 1.

This shows that for 0.5x + 0.5y is not an element of the closed unit ball in (X, ),
and contrapositive of result from Problem 11 shows that (x) does not define a
norm on X.

13. Show that the discrete metric on a vector space X 6= {0} cannot be obtained from a
norm.

Solution: Consider a discrete metric space X 6= {0}. Choose distinct x, y 2 X,

for ↵ = 2, d(2x, 2y) = 1 but |2|d(x, y) = 2. The statement then follows from
theorem.

14. If d is a metric on a vector space X 6= {0} which is obtained from a norm, and d˜ is
defined by
˜ x) = 0,
d(x, ˜ y) = d(x, y) + 1 (x 6= y),
d(x,
show that d˜ cannot be obtained from a norm.

Solution: Consider a metric space X 6= {0}. Choose any x 2 X, for ↵ = 2,

˜
d(2x, ˜ x) = 2(d(x, x) + 1) = 2. The statement
2x) = d(2x, 2x) + 1 = 1 but |2|d(x,
then follows from theorem.

15. (Bounded set) Show that a subset M in a normed space X is bounded if and only
if there is a positive number c such that kxk  c for every x 2 M .

Solution: Suppose a subset M in a normed space X is bounded. By definition,

the diameter (M ) of M is finite, i.e.

(M ) = sup d(x, y) = sup kx yk < 1.

x,y2M x,y2M

Page 14
Fix an y 2 M , then for any x 2 M ,

kxk = kx y + yk  kx yk + kyk  (M ) + kyk < 1.

Choosing c = (M ) + kyk yields the desired result. Conversely, suppose there

exists an c > 0 such that kxk  c for all x 2 M . Then for any x, y 2 M ,

d(x, y) = kx yk  kxk + kyk  2c.

Taking supremum over x, y 2 M on both sides, we obtain (M )  2c < 1. This

shows that M (in a normed space X) is bounded.

Page 15
1.3 Further Properties of Normed Spaces.
Definition 1.5. A subspace Y of a normed space X is a subspace of X considered as a
vector space, with the norm obtained by restricting the norm on X to the subset Y . This
norm on Y is said to be induced by the norm on X.

Theorem 1.6. A subspace Y of a Banach space X is complete if and only if the set Y
is closed in X.

Definition 1.7.

1. If (xk ) is a sequence in a normed space X, we can associate with (xk ) the sequence
(sn ) of partial sums
sn = x 1 + x 2 + . . . + x n
where n = 1, 2, . . .. If (sn ) is convergent, say, sn ! s as n ! 1, then the
X 1
infinite series xk is said to converge or to be convergent, s is called the
k=1
sum of the infinite series and we write
1
X
s= xk = x1 + x2 + . . . .
k=1

1
X
2. If kx1 k+kx2 k+. . . converges, the series xk is said to be absolutely convergent.
k=1

3. If a normed space X contains a sequence (en ) with the property that for every x 2 X,
there is a unique sequence of scalars (↵n ) such that

kx (↵1 e1 + . . . + ↵n en )k ! 0 as n ! 1.
1
X
then (en ) is called a Schauder basis for X. The series ↵j ej which has the
j=1
sum x is called the expansion of x with respect to (en ), and we write
1
X
x= ↵j ej .
j=1

• If X has a Schauder basis, then it is separable. The converse is however not

generally true.

Theorem 1.8. Let X = (X, k · k) be a normed space. There exists a Banach space X̂
and an isometry A from X onto a subspace W of X̂ which is dense in X̂. The space X̂
is unique, except for isometries.

Page 16
1. Show that c ⇢ l1 is a vector subspace of l1 and so is c0 , the space of all sequences
of scalars converging to zero.

Solution: The space c consists of all convergent sequences x = (⇠j ) of complex

numbers. Choose any x = (⇠j ), y = (⌘j ) 2 c ⇢ l1 , with limit ⇠, ⌘ 2 C respectively.
For fixed scalars ↵, , the result is trivial if they are zero, so suppose not. Given
any " > 0, there exists N1 , N2 2 N such that
"
|⇠j ⇠| < for all j > N1 .
2|↵|
"
|⌘j ⌘| < for all j > N2 .
2| |

Choose N = max{N1 , N2 }, then for all j > N we have that

|↵⇠j + ⌘j ↵⇠ ⌘| = |↵(⇠j ⇠) + (⌘j ⌘)|

 |↵||⇠j ⇠| + | |k⌘j ⌘k
" "
< |↵| +| | = ".
2|↵| 2| |

This shows that the sequence ↵x + y = (↵⇠j + ⌘j ) is convergent, hence x 2 c.

Since ↵, were arbitrary scalar, this proves that c is a subspace of l1 . By
replacing ⇠ = ⌘ = 0 as limit, the same argument also shows that c0 is a subspace
of l1 .

2. Show that c0 in Problem 1 is a closed subspace of l1 , so that c0 is complete.

Solution: Consider any x = (⇠j ) 2 c¯0 , the closure of c. There exists xn = (⇠jn ) 2
c0 such that xn ! x in l1 . Hence, given any " > 0, there exists an N 2 N such
that for all n N and all j we have
"
|⇠jn ⇠j |  kxn xk < .
2
in particular, for n = N and all j. Since xN 2 c0 , its terms ⇠jN form a convergent
sequence with limit 0. Thus there exists an N1 2 N such that for all j N1 we
have
"
|⇠jN | < .
2
The triangle inequality now yields for all j N1 the following inequality:
" "
|⇠j |  |⇠j ⇠jN | + |⇠jN | < + = ".
2 2
This shows that the sequence x = (⇠j ) is convergent with limit 0. Hence, x 2 c0 .
Since x 2 c¯0 was arbitrary, this proves closedness of c0 in l1 .

Page 17
3. In l1 , let Y be the subset of all sequences with only finitely many nonzero terms.
Show that Y is a subspace of l1 but not a closed subspace.

Solution: Consider any x = (⇠j ), y = (⌘j ) 2 Y ⇢ l1 , there exists Nx , Ny 2 N

such that ⇠j = 0 for all j > Nx and ⌘j = 0 for all j > Ny . Thus for any scalars
↵, , ↵⇠j + ⌘j = 0 for all j > N = max{Nx , Ny }, and ↵x + y 2 Y . This shows
that Y is a subspace of l1 . However, Y is not a closed subspace. Indeed, consider
a sequence xn = (⇠jn ) 2 Y defined by
8
<1 if j  n,
n
⇠j = j
:
0 if j > n.
✓ ◆
1
Let x = (⇠j ) = , then xn ! x in l1 since
j
1
kxn xkl1 = sup |⇠j | = !0 as n ! 1.
j>n n+1

but x 2
/ Y since x has infinitely many nonzero terms.

4. (Continuity of vector space operations) Show that in a normed space X, vector

addition and multiplication by scalars are continuous operation with respect to the
norm; that is, the mappings defined by (x, y) 7! x+y and (↵, x) 7! ↵x are continuous.

Solution: Consider any pair of points (x0 , y0 ) 2 X ⇥ X. Given any " > 0, choose
"
1 = 2 = > 0. Then for all x satisfying kx x0 k < 1 and all y satisfying
2
ky y0 k < 2 ,

kx + y (x0 + y0 )k  kx x0 k + ky y0 k < 1 + 2 = ".

Since (x0 , y0 ) 2 X ⇥ X was arbitrary, the mapping defined by (x, y) 7! x + y is

continuous with respect to the norm.

Choose any scalar ↵0 . Consider any nonzero x0 2 X. Given any " > 0, choose
" "
1 = > 0 and 2 > 0 such that ( 1 + |↵0 |) 2 = . Then for all ↵ satisfying
2kx0 k 2
k↵ ↵0 k < 1 and all x satisfying kx x0 k < 2 ,

k↵x ↵0 x0 k = k↵x ↵x0 + ↵x0 ↵0 x0 k

 |↵|kx x0 k + |↵ ↵0 |kx0 k
⇣ ⌘
 |↵ ↵0 | + |↵0 | kx x0 k + |↵ ↵0 |kx0 k
<( 1 + |↵0 |) 2 + 1 kx0 k = ".

Page 18
 "
If x0 = 0 2 X, choose 1 = 1 > 0 and 2 = > 0. Then for all ↵ satisfying
1 + |↵0 |
|↵ ↵0 | < 1 and all x satisfying kxk < 2 ,
⇣ ⌘
k↵xk = |↵|kxk  |↵ ↵0 | + |↵0 | kxk
⇣ ⌘
< 1 + |↵0 | 2
⇣ ⌘ "
= 1 + |↵0 | ⇠ = ".
1+
⇠ ⇠⇠ |↵⇠0|

Since ↵0 and x0 were arbitrary scalars and vectors in K and X, the mapping
defined by (↵, x) 7! ↵x is continuous with respect to the norm.

5. Show that xn ! x and yn ! y implies xn + yn ! x + y. Show that ↵n ! ↵

and xn ! x implies ↵n xn ! ↵x.

Solution: If xn ! x and yn ! y, then

kxn + yn x yk  kxn xk + kyn yk ! 0 as n ! 1.

If ↵n ! ↵ and xn ! x, then

k↵n xn ↵xk = k↵n xn ↵n x + ↵n x ↵xk

= k↵n (xn x) + (↵n ↵)xk
 |↵n |kxn xk + |↵n ↵|kxk
 C kxn xk + |↵n ↵| kxk
| {z } | {z }
!0 !0

where we use the fact that convergent sequences are bounded for the last inequal-
ity.

6. Show that the closure Ȳ of a subspace Y of a normed space X is again a vector

subspace.

Solution: If x, y 2 Ȳ , there exists sequences xn , yn 2 Y such that xn ! x and

yn ! y. Then for any scalars ↵, ,

k↵xn + yn (↵x + y)k  |↵|kxn xk + | |kyn yk ! 0 as n ! 1.

This shows that the sequence (↵xn + yn ) 2 Y converges to ↵x + y, which

implies that ↵x + y 2 Ȳ .

Page 19
7. (Absolute convergence) Show that convergence of ky1 k + ky2 k + ky3 k + . . . may
not imply convergence of y1 + y2 + y3 + . . ..

Solution: Let Y be the set of all sequences in l1 with only finitely many nonzero
terms, which is a normed space. Consider the sequence (yn ) = (⌘jn ) 2 Y defined
by 8
< 1 if j = n,
n
⌘j = j 2
:
0 if j 6= n.
X1
1 1
Then kyn k = 2 and 2
< 1. However, y1 + y2 + y3 + . . . ! y, where
n n
✓ ◆ n=1
1
y= , since
n2
n
X 1
yj y = ! 0 as n ! 1.
j=1
(n + 1)2

but y 2
/ Y.

8. If in a normed space X, absolute convergence of any series always implies convergence

of that series, show that X is complete.

Solution: Choose (xn ) be any Cauchy sequence in X. Given any k 2 N, there

1
exists Nk 2 N such that for all m, n Nk , kxm xn k < k ; by construction, (Nk )
2
is an increasing sequence. Consider the sequence (yk ) defined by yk = xNk+1 xNk .
Then 1 1 1
X X X 1
kyk k = kxNk+1 xNk k < < 1.
k=1 k=1 k=1
2k
1
X
This shows that the series yk is absolute convergent, which is also convergent
k=1
by assumption. Thus,
1
X n
X
yk = lim yk
n!1
k=1 k=1
n
X
= lim xNk+1 x Nk
n!1
k=1
= lim xNn+1 xN1 < 1.
n!1

Hence, (xNn+1 ) is a convergent subsequence of (xn ), and since (xn ) is a Cauchy

sequence, (xn ) is convergent. Since (xn ) was an arbitrary Cauchy sequence, X is
complete.

Page 20
 9. Show that in a Banach space, an absolutely convergent series is convergent.

1
X
Solution: Let kxk k be any absolutely convergent series in a Banach space
k=1
X. Since a Banach space is a complete normed space, it suffices to show that the
sequence (sn ) of partial sums sn = x1 + x2 + . . . + xn is Cauchy. Given any " > 0,
there exists an N 2 N such that
1
X
kxk k < ".
k=N +1

For any m > n > N ,

ksm sn k = kxn+1 + xn+2 + . . . + xm k

 kxn+1 k + kxn+2 k + . . . + kxm k
Xm
= kxk k
k=n+1
X1
 kxk k
k=n+1
X1
 kxk k < ".
k=N +1

This shows that (sn ) is Cauchy and the desired result follows from completeness
of X.

10. (Schauder basis) Show that if a normed space has a Schauder basis, it is separable.

Solution: Suppose X has a Schauder basis (en ). Given any x 2 X, there exists
a unique sequence of scalars ( n ) 2 K such that

kx ( 1 e1 + . . . + n en )k !0 as n ! 1.
en
Consider the sequence (fn ) ⇢ X defined by fn = . Note that kfn k = 1 for all
ken k
n 1 and (fn ) is a Schauder basis for X; indeed, if we choose µj = j kej k 2 K,
then
n
X n
X
x µj f j = x j ej !0 as n ! 1.
j=1 j=1

In particular, for any x 2 X, given any " > 0, there exists an N 2 N such that
n
X "
x µj f j < for all n > N.
j=1
2

Page 21
 Define M to be the set
( n
)
X
M= ✓j fj : ✓j 2 K̃, n 2 N .
j=1

where K̃ is a countable dense subset of K. Since µj 2 K, given any " > 0, there
"
exists an ✓j 2 K̃ such that |µj ✓j | < for all j = 1, . . . , n. Then
2n
n
X n
X n
X n
X
x ✓j f j  x µj f j + µj f j ✓j f j
j=1 j=1 j=1 j=1
n
X
"
< + |µj ✓j |kfj k
2 j=1
n
" " X◆◆
< + 1
2 2⇢ ◆
n j=1
◆
" "
= + = ".
2 2
This shows that there exists an y 2 M in any "-neighbourhood of x. Since x 2 X
was arbitrary, M is a countable dense subset of X and X is separable.

11. Show that (en ), where en = ( nj ), is a Schauder basis for lp , where 1  p < +1.

Solution: Let x = (⇠j ) be any sequence in lp , we have

1
! p1
X
|⇠j |p !0 as n ! 1.
j=n+1

Now choose a sequence of scalars ( n) 2 C defined by j = ⇠j ,

1
! p1
X
kx ( 1 e1 + . . . + n en )k = |⇠j |p !0 as n ! 1.
j=n+1

This shows that (en ) = ( nj ) is a Schauder basis for lp . Uniqueness?

12. (Seminorm) A seminorm on a vector space X is a mapping p : X ! R satisfying

(N1), (N3), (N4). (Some authors call this a pseudonorm.) Show that

p(0) = 0, |p(y) p(x)|  p(y x).

(Hence if p(x) = 0 =) x = 0, then p is a norm.)

Page 22
 Solution: Using (N3),

p(0) = p(0x) = 0p(x) = 0.

Using (N4), for any x, y 2 X,

p(y)  p(y x) + p(x).

p(x)  p(x y) + p(y) = p(y x) + p(y).
=) |p(y) p(x)|  p(y x).

13. Show that in Problem 12, the elements x 2 X such that p(x) = 0 form a subspace
N of X and a norm on X/N is defined by kx̂k0 = p(x), where x 2 x̂ and x̂ 2 X/N .

Solution: Let N be the set consisting of all elements x 2 X such that p(x) = 0.
For any x, y 2 N and scalars ↵, ,

0  p(↵x + y)  p(↵x) + p( y) = |↵|p(x) + | |p(y) = 0.

This shows that N is a subspace of X. Consider kx̂k0 = p(x), where x 2 x̂ and

x̂ 2 X/N . We start by showing k · k0 is well-defined. Indeed, for any u, v 2 x̂,
there exists nu , nv 2 N such that u = x + nu and v = x + nv . Since N is a
subspace of X,

0  |p(u) p(v)|  |p(u v)| = |p(nu nv )| = 0.

• kx̂k0 = p(x) 0.

• Suppose x̂ = 0̂ = N , then kx̂k0 = p(x) = 0. Now suppose kx̂k0 = 0, then

p(x) = 0 =) x 2 N =) x̂ = 0̂. Thus, (N2) is satisfied.

• For any nonzero scalars ↵, any y 2 ↵x̂ can be written y = ↵x + n for some
n 2 N . Thus,
⇣ n⌘
k↵x̂k0 = p(↵x + n) = |↵|p x +
↵
= |↵|kx̂k0 .

If ↵ = 0, then 0x̂ = N and k0x̂k0 = 0 by definition of N .

• Lastly, for any x̂, ŷ 2 X/N ,

kx̂ + ŷk0 = p(x + y)  p(x) + p(y)

= kx̂k0 + kŷk0 .

Thus, (N4) is satisfied.

Page 23
14. (Quotient space) Let Y be a closed subspace of a normed space (X, k · k). Show
that a norm k · k0 on X/Y is defined by

kx̂k0 = inf kxk.

x2x̂

where x̂ 2 X/Y , that is, x̂ is any coset of Y .

Solution: Define kx̂k0 as above. Also, recall that X/Y = {x̂ = x + Y : x 2 X}

and its algebraic operations are defined by

û + v̂ = (u + Y ) + (v + Y ) = (u + v) + Y = u[
+ v.
↵û = ↵(u + Y ) = ↵u + Y = ↵ cu.

• (N1) is obvious.

• If x̂ = 0̂ = Y , then kx̂k0 =0 since 0 2 Y . Conversely, suppose kx̂k0 =

inf kxk = 0. Properties of infimum gives that there exists a minimising
x2x̂
sequence (xn ) 2 x̂ such that kxn k0 ! 0, with limit x = 0. Since Y is
closed, any x̂ 2 X/Y is closed, this implies that 0 2 x̂, and x̂ = 0̂. Thus,
(N2) is satisfied.

• For any nonzero scalars ↵,

k↵x̂k0 = inf k↵x + yk

y2Y
y
= |↵| inf x +
y2Y ↵
= |↵| inf kx + yk
y2Y

= |↵|kx̂k0 .

c 0 = k0̂k0 = 0 = 0kx̂k0 . Thus, (N3) is satisfie.d

If ↵ = 0, then k0x̂k0 = k0xk

• For any û, v̂ 2 X/Y ,

kû + v̂k0 = inf ku + y1 + v + y2 k

y1 ,y2 2Y

 inf ku + y1 k + kv + y2 k
y1 ,y2 2Y

= inf ku + y1 k + inf kv + y2 k
y1 2Y y2 2Y

= kûk0 + kv̂k0 .

Page 24
15. (Product of normed spaces) If (X1 , k· k1 ) and (X2 , k· k2 ) are normed spaces, show
that the product vector space X = X1 ⇥ X2 becomes a normed space if we define
n o
kxk = max kx1 k1 , kx2 k2 , where x = (x1 , x2 ).

Solution: (N1) to (N3) are obvious. To verify (N4), for x = (x1 , x2 ), y =

(y1 , y2 ) 2 X1 ⇥ X2 ,
n o
kx + yk = max kx1 + y1 k1 , kx2 + y2 k2
n o
 max kx1 k1 + ky1 k1 , kx2 k2 + ky2 k2
n o n o
 max kx1 k1 , kx2 k2 + max ky1 k1 , ky2 k2
= kxk + kyk.

where we use the inequality |a|  max{|a|, |b|} for any a, b 2 R.

Page 25
1.4 Finite Dimensional Normed Spaces and Subspaces.
Lemma 1.9. Let {x1 , . . . , xn } be a linearly independent set of vectors in a normed space
X (of any dimension). There exists a number c > 0 such that for every choice of scalars
↵1 , . . . , ↵n we have
⇣ ⌘
k↵1 x1 + . . . + ↵n xn k c |↵1 | + . . . + |↵n | .

• Roughly speaking, it states that in the case of linear independence of vectors, we

cannot find a linear combination that involves large scalars but represents a small
vector.

Theorem 1.10. Every finite dimensional subspace Y of a normed space X is complete.

In particular, every finite dimensional normed space is complete.

Theorem 1.11. Every finite dimensional subspace Y of a normed space X is closed in X.

Definition 1.12. A norm k · k on a vector space X is said to be equivalent to a norm

k · k0 on X if there are positive constants a and b such that for all x 2 X we have

akxk0  kxk  bkxk0 .

• Equivalent norms on X define the same topology for X.

Theorem 1.13. On a finite dimensional vector space X, any norm k · k is equivalent to

any other norm k · k0 .

1. Given examples of subspaces of l1 and l2 which are not closed.

Solution:

2. What is the largest possible c in (1) if

(a) X = R2 and x1 = (1, 0), x2 = (0, 1),

Solution:

(b) X = R3 and x1 = (1, 0, 0), x2 = (0, 1, 0), x3 = (0, 0, 1).

Solution:

3. Show that in the definition of equivalance of norms, the axioms of an equivalence

relation hold.

Page 26
 Solution: We say that k·k on X is equivalent to k·k0 on X, denoted by k·k ⇠ k·k0
if there exists positive constants a, b > 0 such that for all x 2 X we have

akxk0  kxk  bkxk0 .

• Reflexivity is immediate.

• Suppose k · k ⇠ k · k0 . There exists a, b > 0 such that for all x 2 X we have

1 1
akxk0  kxk  bkxk0 =) kxk  kxk0  kxk.
b a
This shows that k · k0 ⇠ k · k, and symmetry is shown.

• Suppose k · k ⇠ k · k0 and k · k0 ⇠ k · k1 . There exists a, b, c, d > 0 such that

for all x 2 X we have

akxk0  kxk  bkxk0 .

ckxk1  kxk0  dkxk1 .

On one hand,
kxk  bkxk0  bdkxk1 .
On the other hand,
kxk akxk0 ackxk1 .
Combining them yields for all x 2 X

ackxk1  kxk0  bdkxk1 .

This shows tht k · k ⇠ k · k1 , and transitivity is shown.

4. Show that equivalent norms on a vector space X induce the same topology for X.

Solution: Suppose k · k and k · k0 are equivalent norms on a vector space X.

There exists positive constants a, b > 0 such that for all x 2 X we have

akxk0  kxk  bkxk0 .

To show that they induce the same topology for X, we want to show that the
open sets in (X, k · k) and (X, k · k0 ) are the same. Consider the identity map

I : (X, k · k) ! (X, k · k0 ).

Let x0 be any point in X. Given any " > 0, choose = a" > 0. For all x satisfying
kx x0 k < , we have
1 ⇢
a"
kx x0 k0  kx x0 k < = ".
a ⇢
a

Page 27
 Since x0 2 X is arbitrary, this shows that I is continuous. Hence, if M ⇢ X
is open in (X, k · k0 ), its preimage M again is also open in (X, k · k). Similarly,
consider the identity map

I˜: (X, k · k0 ) ! (X, k · k).

"
Let x0 be any point in X. Given any " > 0, choose = > 0. For all x satisfying
b
kx x0 k0 < , we have

b"
kx x0 k  bkx x 0 k0 < = ".
b
Since x0 2 X is arbitrary, this shows that I˜ is continuous. Hence, if M ⇢ X is
open in (X, k · k), its preimage M is also open in (X, k · k0 ).

Remark : The converse is also true, i.e. if two norms k · k and k · k0 on X give
the same topology, then they are equivalent norms on X.

5. If k · k and k · k0 are equivalent norms on X, show that the Cauchy sequences in

(X, k · k) and (X, k · k0 ) are the same.

Solution: Suppose k · k on X is equivalent to k · k0 on X. There exists positive

constants a, b > 0 such that for all x 2 X we have

akxk0  kxk  bkxk0 .

Let (xn ) be any Cauchy sequence in (X, k · k). Given any " > 0, there exists
N1 2 N such that

kxm xn k < a" for all m, n > N1 .

which implies
1 ⇢
a"
kxm xn k0  kxm xn k < = ". for all m, n > N1 .
a ⇢
a
This shows that (xn ) is also a Cauchy sequence in (X, k · k0 ). Conversely, let (xn )
be any Cauchy sequence in (X, k · k0 ). Given any " > 0, there exists N2 2 N such
that
"
kxm xn k0 < for all m, n > N1 .
b
which implies

b"
kxm xn k  bkxm x n k0 < =" for all m, n > N2 .
b
This shows that (xn ) is also a Cauchy sequence in (X, k · k).

Page 28
6. Theorem 2.4.5 implies that k · k2 and k · k1 are equivalent. Give a direct proof of this
fact.

Solution: Let X = Rn , and x = (⇠j ) be any element of X. On one hand,

⇣ ⌘2
kxk21 = max |⇠j |  |⇠1 |2 + . . . + |⇠n |2 = kxk22 .
j=1,...,n

Taking square roots of both sides yields kxk1  kxk2 . On the other hand,
⇣ ⌘
kxk22 = |⇠1 |2 + . . . + |⇠n |2  n max |⇠j |2 = nkxk21 .
j=1,...,n

p
Taking square roots of both sides yields kxk2  nkxk1 . Hence, combining these
inequalities gives for all x 2 X
p
kxk1  kxk2  nkxk1 .

7. Let k · k2 be as in Problem 8, section 2.2, and let k · k be any norm on that vector
space, call it X. Show directly that there is a b > 0 such that kxk  bkxk2 for all x.

Solution: Let X = Rn , and {e1 , . . . , en } be the standard basis of X defined by

⇠jn = jn . Any x = (⇠j ) in X has a unique representation x = ⇠1 e1 + . . . + ⇠2 e2 .
Thus,

kxk = k⇠1 e1 + . . . + ⇠n en k  |⇠1 |ke1 k + . . . + |⇠n |ken k

Xn
= |⇠j |kej k
j=1

n
! 12 n
! 12
X X
 |⇠j |2 kej k2
j=1 j=1

= bkxk2 .

where we use Cauchy-Schwarz inequality for sums in the last inequality.

Since x 2 X was arbitrary, the result follows.

8. Show that the norms k · k1 and k · k2 satisfy

1
p kxk1  kxk2  kxk1 .
n

Page 29
Solution: Let X = Rn , and x = (⇠j ) be any element of X. Note that if x = 0,
the inequality is trivial since kxk1 = kxk2 = 0 by definition of a norm. So, pick
any nonzero x 2 Rn . Using Cauchy-Schwarz inequality for sums,

n n
! 12 n
! 12
X X X p
kxk1 = |⇠j |  12 |⇠j |2 = nkxk2 .
j=1 j=1 j=1

⇠j
On the other hand, since kxk2 6= 0, define y = (⌘j ), where ⌘j = . Then
kxk2

n
! 12 n
! 12
X 1 X
kyk2 = |⌘j |2 = 2
|⇠j |2
j=1
kxk2 j=1
1
= kxk2 = 1.
kxk2
n
! n
X 1 X
kyk1 = |⌘j | = |⇠j |
j=1
kxk2 j=1
kxk1
= .
kxk2

and
n
X  n
X
kyk22 = |⌘j |  2
max |⌘i | |⌘j |  kyk1
i=1,...,n
j=1 j=1
kxk1
=) 1  kyk1 =
kxk2
=) kxk2  kxk1 .

To justfiy the second inequality on the first line, note that it suffices to prove that
|⌘i |  1 for all i = 1, . . . , n, or equivalently, |⇠i |  kxk2 for all i = 1, . . . , n. From
the definition of k · k2 ,
n
X
kxk22 = |⇠j |2 |⇠i |2 for all i = 1, . . . , n.
j=1

Taking square roots of both sides yields kxk2 |⇠i | for all i = 1, . . . , n.

Remark : Alternatively,
n
!2 n
! n
!
X X X
kxk21 = |⇠j | = |⇠j |2 + |⇠i ||⇠j |
j=1 j=1 i6=j
n
!
X
|⇠j |2 = kxk22 .
j=1

Page 30
 9. If two norms k · k and k · k0 on a vector space X are equivalent, show that

kxn xk ! 0 () kxn xk0 ! 0.

Solution: Suppose two norms k · k and k · k0 on a vector space X are equivalent,

there exists positive constant a, b > 0 such that for all x 2 X we have

akxk0  kxk  bkxk0 .

If kxn xk ! 0, then
1
kxn xk0  kxn xk ! 0 as n ! 0.
a
Conversely, if kxn xk0 ! 0, then

kxn xk  bkxn xk0 ! 0 as n ! 0.

10. Show that all complex m ⇥ n matrices A = (↵jk ) with fixed m and n constitute an
mn-dimensional vector space Z. Show that all norms on Z are equivalent. What
would be the analogues of k · k1 , k · k2 and k · k1 for the present space Z?

Solution:

Page 31
1.5 Linear Operators.
Definition 1.14. A linear operator T is an operator such that
(a) the domain D(T ) of T is a vector space and the range R(T ) lies in a vector space
over the same field,

(b) for all x, y 2 D(T ) and scalars ↵, ,

T (↵x + y) = ↵T (x) + T (y).

• Note that the above formula expresses the fact that a linear operator T is a ho-
momorphism of a vector space (its domain) into another vector space, that is, T
preserves the two operations of a vector space.

• On the LHS, we first apply a vector space operation (addition or scalar multiplica-
tion) and then map the resulting vector into Y , whereas on the RHS we first map
x and y into Y and then perform the vector space operations in Y , the outcome
being the same.

Theorem 1.15. Let T be a linear operator. Then:

(a) The range R(T ) is a vector space.

(b) If dim D(T ) = n < 1, then dim R(T )  n.

(c) The null space N (T ) is a vector space.

• An immediate consequence of part (b) is worth noting: Linear operators preserve

linear dependence.

Theorem 1.16. Let X, Y be a vector spaces, both real or both complex. Let T : D(T ) !
Y be a linear operator with domain D(T ) ⇢ X and range R(T ) ⇢ Y . Then:
(a) The inverse T 1 : R(T ) ! D(T ) exists if and only if T x = 0 =) x = 0.
1
(b) If T exists, it is a linear operator.
1
(c) If dim D(T ) = n < 1 and T exists, then dim R(T ) = dim D(T ).

Theorem 1.17. Let T : X ! Y and S : Y ! Z be bijective linear operators, where

X, Y, Z are vector spaces. Then the inverse (ST ) 1 : Z ! X of the composition ST
exists, and
(ST ) 1 = T 1 S 1 .

Page 32
1. Show that the identity operator, the zero operator and the di↵erentiation operator
(on polynomials) are linear.

Solution: For any scalars ↵, and x, y 2 X,

IX (↵x + y) = ↵x + y = ↵IX x + IX y.
0(↵x + y) = 0 = ↵0x + 0y.
T (↵x(t) + y(t)) = (↵x(t) + y(t))0
= ↵x0 (t) + y 0 (t) = ↵T x(t) + T y(t).

2. Show that the operators T1 , . . . , T4 from R2 into R2 defined by

T1 : (⇠1 , ⇠2 ) 7! (⇠1 , 0)
T2 : (⇠1 , ⇠2 ) 7! (0, ⇠2 )
T3 : (⇠1 , ⇠2 ) 7! (⇠2 , ⇠1 )
T4 : (⇠1 , ⇠2 ) 7! ( ⇠1 , ⇠2 )
respectively, are linear, and interpret these operators geometrically.

Solution: Denote x = (⇠1 , ⇠2 ) and y = (⌘1 , ⌘2 ). For any scalars ↵, ,

T1 (↵x + y) = (↵⇠1 + ⌘1 , 0)
= ↵(⇠1 , 0) + (⌘1 , 0) = ↵T1 (x) + T1 (y).
T2 (↵x + y) = (0, ↵⇠2 + ⌘2 )
= ↵(0, ⇠2 ) + (0, ⌘2 ) = ↵T2 (x) + T2 (y).
T3 (↵x + y) = (↵⇠2 + ⌘2 , ↵⇠1 + ⌘1 )
= (↵⇠2 , ↵⇠1 ) + ( ⌘2 , ⌘1 )
= ↵(⇠2 , ⇠1 ) + (⌘2 , ⌘1 ) = ↵T3 (x) + T3 (y).
⇣ ⌘
T4 (↵x + y) = (↵⇠1 + ⌘1 ), (↵⇠2 + ⌘2 )
= (↵ ⇠1 , ↵ ⇠2 ) + ( ⌘1 , ⌘2 )
= ↵( ⇠1 , ⇠2 ) + ( ⌘1 , ⌘2 ) = ↵T4 (x) + T4 (y).

T1 and T2 are both projection to x-axis and y-axis respectively, while T4 is a

scaling transformation. T3 first rotates the vector 90 anti-clockwise about the
origin, then reflects across the y-axis.

3. What are the domain, range and null space of T1 , T2 , T3 in Problem 2?

Solution: The domain of T1 , T2 , T3 is R2 , and the range is the x-axis for T1 , the
y-axis for T2 and R2 for T3 . The null space is the line ⇠1 = 0 for T1 , the line ⇠2 = 0
for T2 and the origin (0, 0) for T3 .

Page 33
4. What is the null space of T4 in Problem 2? Of T1 and T2 in 2.6-7? Of T in 2.6-4?

Solution: The null space of T4 is R2 if = 0 and the origin (0, 0) if 6= 0. Fix

a vector a = (a1 , a2 , a3 ) 2 R3 . Consider the linear operators T1 and T2 defined by
2 3 2 3 2 3
⇠1 a1 ⇠ 2 a3 ⇠ 3 a2
T 1 x = x ⇥ a = 4 ⇠ 2 5 ⇥ 4 a2 5 = 4 ⇠ 3 a1 ⇠ 1 a3 5 .
⇠3 a3 ⇠ 1 a2 ⇠ 2 a1
T 2 x = x · a = ⇠ 1 a1 + ⇠ 2 a2 + ⇠ 3 a3 .

i.e. T1 and T2 are the cross product and the dot product with the fixed vector a
respectively. The null space of T1 is any scalar multiple of the vector a, while the
null space of T2 is the plane ⇠1 a1 + ⇠2 a2 + ⇠3 a3 = 0 in R3 . For the di↵erentiation
operator, the null space is any constant functions x(t) for t 2 [a, b].

5. Let T : X ! Y be a linear operator.

(a) Show that the image of a subspace V of X is a vector space.

Solution: Denote the image of a subspace V of X under T by Im(V ), it

suffices to show that Im(V ) is a subspace of X. Choose any y1 , y2 2 Im(V ),
there exists x1 , x2 2 V such that T x1 = y1 and T x2 = y2 . For any scalars
↵, ,
↵y1 + y2 = ↵T x1 + T x2 = T (↵x1 + x2 ).
This shows that ↵y1 + y2 2 Im(V ) since ↵x1 + x2 2 V due to V being a
subspace of X.

(b) Show that the inverse image of a subspace W of Y is a vector space.

Solution: Denote the inverse image of a subspace W of Y under T by

PIm(W ), it suffices to show that PIm(W ) is a subspace of Y . Choose any
x1 , x2 2 PIm(W ), there exists y1 , y2 2 W such that T x1 = y1 and T x2 = y2 .
For any scalars ↵, ,

T (↵x1 + x2 ) = ↵T x1 + T x2 = ↵y1 + y2 .

This shows that ↵x1 + x2 2 PIm(W ) since ↵y1 + y2 2 W due to W being

a subspace of Y .

6. If the composition of two linear operators exists, show that it is linear.

Solution: Consider any two linear operators T : X ! Y , S : Y ! Z. For any

x, y 2 X and scalars ↵, ,
h i
ST (↵x + y) = S(↵T x + T y) by linearity of T.

Page 34
 h i
= ↵(ST )x + (ST )y by linearity of S.

7. (Commutativity) Let X be any vector space and S : X ! X and T : X ! X

any operators. S and T are said to commute if ST = T S, that is, (ST )x = (T S)x
for all x 2 X. Do T1 and T3 in Problem 2 commute?

Solution: No. Choose x = (1, 2), then

(T1 T3 )(1, 2) = T1 (2, 1) = (2, 0).

(T3 T1 )(1, 2) = T3 (1, 0) = (0, 1).

8. Write the operators in Problem 2 using 2 ⇥ 2 matrices.

Solution:
   
1 0 0 0 0 1 0
T1 = T2 = T3 = T4 = .
0 0 0 1 1 0 0

9. In 2.6-8, write y = Ax in terms of components, show that T is linear and give

examples.

n
X
Solution: For any j = 1, . . . , r, we have that ⌘j = ajk ⇠k = aj1 ⇠1 + . . . + ajn ⇠n .
k=1
To see that T is linear, for any j = 1, . . . , r,
⇣ ⌘ n
X
A(↵x + y) = ajk (↵⇠k + ⌘k )
j
k=1
Xn n
X
=↵ ajk ⇠k + ajk ⌘k = ↵(Ax)j + (Ay)j .
k=1 k=1

10. Formulate the condition in 2.6-10(a) in terms of the null space of T .

Solution: Let X, Y be vector spaces, both real or both complex. Let T : D(T ) !
Y be a linear operator with domain D(T ) ⇢ X and R(T ) ⇢ Y . The inverse
T 1 : R(T ) ! D(T ) exists if and only if the null space of T , N (T ) = {0}.

Page 35
11. Let X be the vector space of all complex 2 ⇥ 2 matrices and define T : X ! X by
T x = bx, where b 2 X is fixed and bx denotes the usual product of matrices. Show
that T is linear. Under what condition does T 1 exists?

Solution: For any x, y 2 X and scalars ↵, ,

 
b1 b2 ↵⇠1 + ⌘1 ↵⇠2 + ⌘2
T (↵x + y) =
b3 b4 ↵⇠3 + ⌘3 ↵⇠4 + ⌘4
 ⇢  
b1 b2 ⇠ ⇠2 ⌘1 ⌘2
= ↵ 1 +
b3 b4 ⇠3 ⇠4 ⌘3 ⌘4
= ↵bx + by = ↵T x + T y.
1
This shows that T is linear. T exists if and only if b is a non-singular 2 ⇥ 2
complex matrix.

12. Does the inverse of T in 2.6-4 exist?

Solution: The inverse of the di↵erentiation operator T does not exist because
N (T ) 6= {0}, the zero function.

13. Let T : D(T ) ! Y be a linear operator whose inverse exists. If {x1 , . . . , xn } is

a linearly independent set in D(T ), show that the set {T x1 , . . . , T xn } is linearly
independent.

Solution: Let T : D(T ) ! Y be a linear operator whose inverse exists. Suppose

↵ 1 T x1 + . . . + ↵ n T xn = 0 Y .

By linearity of T , the equation above is equivalent to

T (↵1 x1 + . . . + ↵n xn ) = 0Y .
1
Since T exists, we must have “ T x = 0Y =) x = 0X ”. Thus,

↵ 1 x1 + . . . + ↵ n xn = 0X .

but {x1 , . . . , xn } is a linearly independent set in D(T ), so this gives ↵1 = . . . =

↵n = 0.

14. Let T : X ! Y be a linear operator and dim X = dim Y = n < 1. Show that
R(T ) = Y if and only if T 1 exists.

Page 36
 Solution: Suppose R(T ) = Y , by definition, for all y 2 Y , there exists x 2 X
such that T x = y, i.e. T is surjective. We now show that T is injective. Since
dim (X) = dim (Y ) = n < 1, the Rank-Nullity Theorem gives us

dim(R(T )) + dim(N (T )) = dim(X).

but by assumption, R(T ) = Y , so

dim(Y ) + dim(N (T )) = dim(X).

=) dim(N (T )) = dim(X) dim(Y ) = 0.
=) N (T ) = {0X }.

This shows that T is injective. Indeed, suppose for any x1 , x2 , we have T x1 = T x2 .

By linearity of T , T x1 T x2 = T (x1 x2 ) = 0Y =) x1 x2 = 0X =) x1 = x2 .
Since T is both injective and surjective, we conclude that the inverse of T , T 1 ,
exists.

Conversely, suppose T 1 exists. From Problem 10, this means that N (T ) =

{0X } =) dim(N (T )) = 0. Invoking the Rank-Nullity Theorem gives

dim(R(T )) + dim(N (T )) = dim(X).

=) dim(R(T )) = dim(X) = n.

This implies that R(T ) = Y since any proper subspace W of Y has dimension
less than n.

15. Consider the vector space X of all real-valued functions which are defined on R and
have derivatives of all orders everywhere on R. Define T : X ! X by y(t) = T x(t) =
x0 (t). Show that R(T ) is all of X but T 1 does not exist. Compare with Problem 14
and comment.

Rt
Solution: For any y(t) 2 R(T ), define x(t) = 1 y(s) ds 2 X; Fundamental
Theorem of Calculus gives that x0 (t) = T x(t) = y(t). On the other hand, T 1
does not exist since the null space of T consists of every constant functions on
R. However, it doesn’t contradict Problem 14 since X is an infinite-dimensional
vector space.

Page 37
1.6 Bounded and Continuous Linear Operators.
Definition 1.18.

1. Let X and Y be normed spaces and T : D(T ) ! Y a linear operator, where D(T ) ⇢
X. The operator T is said to be bounded if there is a nonnegative number C such
that for all x 2 D(T ), kT xk  Ckxk.

• This also shows that a bounded linear operator maps bounded sets in D(T )
onto bounded sets in Y .

2. The norm of a bounded linear operator T is defined as

kT xk
kT k = sup .
x2D(T ) kxk
x6=0

• This is the smallest possible C for all nonzero x 2 D(T ).

• With C = kT k, we have the inequality kT xk  kT kkxk.
• If D(T ) = {0}, we define kT k = 0.

Lemma 1.19. Let T be a bounded linear operator. Then:

(a) An alternative formula for the norm of T is

kT k = sup kT xk.
x2D(T )
kxk=1

(b) kT k is a norm.

Theorem 1.20. If a normed space X is finite dimensional, then every linear operator
on X is bounded.

Theorem 1.21. Let T : D(T ) ! Y be a linear operator, where D(T ) ⇢ X and X, Y

are normed spaces.
(a) T is continuous if and only if T is bounded.

(b) If T is continuous at a single point, it is continuous.

Corollary 1.22. Let T be a bounded linear operator. Then:

(a) xn ! x [where xn , x 2 D(T )] implies T xn ! T x.

(b) The null space N (T ) is closed.

• It is worth noting that the range of a bounded linear operator may not be closed.

Page 38
Definition 1.23.

1. Two operators T1 and T2 are defined to be equal, written T1 = T2 if they have the
same domain D(T1 ) = D(T2 ) and if T1 x = T2 x for all x 2 D(T1 ) = D(T2 ).

2. The restriction of an operator T : D(T ) ! Y to a subset B ⇢ D(T ) is denoted

by T |B and is the operator defined by T |B : B ! Y , satisfying

T |B x = T x for all x 2 B.

3. The extension of an operator T : D(T ) ! Y to a superset M D(T ) is an

operator T̃ : M ! Y such that T̃ |D(T ) = T , that is, T̃ x = T x for all x 2 D(T ).
[Hence T is the restriction of T̃ to D(T ).]

• If D(T ) is a proper subset of M , then a given T has many extensions; of prac-

tical interest are those extensions which preserve linearity or boundedness.

Theorem 1.24 (Bounded linear extension).

Let T : D(T ) ! Y be a bounded linear operator, where D(T ) lies in a normed space
X and Y is a Banach space. Then T has an extension T̃ : D(T ) ! Y , where T̃ is a
bounded linear operator with norm kT̃ k = kT k.

• The theorem concerns an extension of a bounded linear operator T to the closure

D(T ) of the domain such that the extended operator is again bounded and linear,
and even has the same norm.

• This includes the case of an extension from a dense set in a normed space X to all
of X.

• It also includes the case of an extension from a normed space X to its completion.

1. Prove kT1 T2 k  kT1 kkT2 k and kT n k  kT kn (n 2 N) for bounded linear operators

T2 : X ! Y , T1 : Y ! Z and T : X ! X, where X, Y, Z are normed spaces.

Solution: Using boundedness of T1 and T2 ,

k(T1 T2 )xk = kT1 (T2 x)k  kT1 kkT2 xk  kT1 kkT2 kkxk.

The first inequality follows by taking supremum over all x of norm 1. A similar
argument also shows the second inequality.

2. Let X and Y be normed spaces. Show that a linear operator T : X ! Y is bounded

if and only if T maps bounded sets in X into bounded sets in Y .

Page 39
 Solution: Suppose T : X ! Y is bounded, there exists an C > 0 such that
kT xk  Ckxk for all x 2 X. Take any bounded subset A of X, there exists
MA > 0 such that kxk  MA for all x 2 A. For any x 2 A,

kT xk  Ckxk  CMA .

This shows that T maps bounded sets in X into bounded sets in Y .

Conversely, suppose a linear operator T : X ! Y maps bounded sets in X into

bounded sets in Y . This means that for any fixed R > 0, there exists a constant
MR > 0 such that kxk  R =) kT xk  MR . We now take any nonzero y 2 X
and set
y
x=R =) kxk = R.
kyk
Thus,
✓ ◆
R R
kT yk = T y = kT zk  MR .
kyk kyk
MR
=) kT yk  kyk.
R
where we crucially used the linearity of T . Rearranging and taking supremum
over all y of norm 1 shows that T is bounded.

3. If T 6= 0 is a bounded linear operator, show that for any x 2 D(T ) such that kxk < 1
we have the strict inequality kT xk < kT k.

Solution: We have kT xk  kT kkxk < kT k.

4. Let T : D(T ) ! Y be a linear operator, where D(T ) ⇢ X and X, Y are normed

spaces. Show that if T is continuous at a single point, it is continuous on D(T ).

Solution: Suppose T is continuous at an arbitrary x0 2 D(T ). This means that

given any " > 0, there exists a > 0 such that kT x T x0 k  " for all x 2 D(T )
satisfying kx x0 k  . Fix an y0 2 D(T ), and set
y y0
x x0 = =) kx x0 k = .
ky y0 k
Since T is linear, for any y 2 D(T ) satisfying ky y0 k  ,
✓ ◆
kT (y y0 )k = T (y y0 ) = kT (x x0 )k  "
ky y0 k ky y0 k
ky y0 k "
=) kT (y y0 )k  "  = ".

This shows that T is continuous at y0 . Since y0 2 D(T ) is arbitrary, the statement

follows.

Page 40
5. Show that the operator T : l1 ! l1 defined by y = (⌘j ) = T x, ⌘j = ⇠j /j, x = (⇠j ),
is linear and bounded.

Solution: For any x, z 2 l1 and scalars ↵, ,

✓ ◆ ✓ ◆ ✓ ◆
⇠j j ⇠j j
T (↵x + z) = ↵ + =↵ + = ↵T x + T z.
j j j j

For any x = (⇠j ) 2 l1 ,

⇠j
 |⇠j |  sup |⇠j | = kxk.
j j2N

Taking supremum over j 2 N on both sides yields kT xk  kxk. We conclude that

T is a bounded linear operator.

6. (Range) Show that the range R(T ) of a bounded linear operator T : X ! Y need
not be closed in Y .

Solution: Define (xn ) to be a sequence in the space l1 , where xn = (⇠jn ) and

(p
j if j  n,
⇠jn =
0 if j > n.

Consider the operator T : l1 ! l1 in Problem 5. Then T xn = yn = (⌘jn ), where

8
⇠jn< p1 if j  n,
⌘jn = = j
j :
0 if j > n.

We now have our sequence (yn ) 2 R(T ) ⇢ l1 . We claim that it converges to y in

1
l1 , where y is a sequence in l1 defined as y = (⌘j ), ⌘j = p . Indeed,
j
1
kyn ykl1 = sup |⌘jn ⌘j | = p !0 as n ! 1.
j2N n+1
1
However, y 2
/ R(T ). Indeed, if there exists
p an x 2 l such that T x = y, then
x must be the sequence (⇠j ), with ⇠j = j, which is clearly not in the space l1 .
Hence, R(T ) is not closed in l1 .

7. (Inverse operator) Let T be a bounded linear operator from a normed space X

onto a normed space Y . If there is a positive b such that

kT xk bkxk for all x 2 X,

1
show that then T :Y ! X exists and is bounded.

Page 41
 Solution: We first show that T is injective, and therefore the inverse T 1 exists
since T is bijective (T is surjective by assumption). Indeed, choose any x1 , x2 2 X
and suppose x1 6= x2 , then kx1 x2 k > 0. Since T is linear,

kT x1 T x2 k = kT (x1 x2 )k  bkx1 x2 k > 0 =) T x1 6= T x2 .

We are left to show there exists an C > 0 such that kT 1 yk  Ckyk for all
y 2 Y . Since T is surjective, for any y 2 Y , there exists x 2 X such that y = T x;
existence of T 1 then implies x = T 1 y. Thus,

1 1 1 1
kT (T (y))k bkT yk =) kT yk  kyk.
b
1
where C = > 0.
b

1
8. Show that the inverse T : R(T ) ! X of a bounded linear operator T : X !Y
need not be bounded.

Solution: Consider the operator T : l1 ! l1 defined by y = T x = (⌘j ), ⌘j =

⇠j /j, x = (⇠j ). We shown in Problem 6 that T is a bounded linear operator. We
first show that T is injective. For any x1 , x2 2 l1 , suppose T x1 = T x2 . For any
j 2 N,

⇠j1 ⇠j2 1
(T x1 )j = (T x2 )j =) = =) ⇠j1 = ⇠j2 since 6= 0.
j j j
This shows that x1 = x2 and T is injective. Thus, there exists an inverse
T 1 : R(T ) ! l1 defined by x = T 1 y = (⇠j ), ⇠j = j⌘j , y = (⌘j ). Let’s
verify that this is indeed the inverse operator.
✓ ◆ ✓ ◆
1 1 1 ⇠j ⇠j
T (T x) = T y = T = j = (⇠j ) = x.
j j
✓ ◆
1 j⌘j
T (T y) = T x = T ((j⌘j )) = = (⌘j ) = y.
j
1
We claim that T 1 is not bounded. Indeed, let yn = ( jn )j=1 , where jn is the
Kronecker delta function. Then kyn k = 1 and

1 kT 1 yn k
kT yn k = k(j jn )k = n =) = n.
kyn k

Since n 2 N is arbitrary, this shows that there is no fixed number C > 0 such
kT 1 yn k
that  C, i.e. T 1 is not bounded.
kyn k

Page 42
 9. Let T : C[0, 1] ! C[0, 1] be defined by
Z t
y(t) = x(s) ds.
0

1 1
Find R(T ) and T : R(T ) ! C[0, 1]. Is T linear and bounded?

Solution: First, Fundamental Theorem of Calculus yields

R(T ) = {y(t) 2 C[0, 1] : y(t) 2 C 1 [0, 1], y(0) = 0} ⇢ C[0, 1].

Next, we show that T is injective and thus the inverse T 1 : R(T ) ! C[0, 1]
exists. Indeed, suppose for any x1 , x2 2 C[0, 1], T x1 = T x2 . Then
Z t Z t
T x1 = T x2 =) x1 (s) ds = x2 (s) ds
0 0
Z th i
=) x1 (s) x2 (s) ds = 0
0
=) x(s) = y(s) for all s 2 [0, t].

where the last implication follows from x y being a continuous function in

[0, t] ⇢ [0, 1]. The inverse operator T 1 is defined by T 1 y(t) = y 0 (t), i.e. T 1
is the di↵erentiation operator. Since di↵erentiation is a linear operation, so is
T 1 . However, T 1 is not bounded. Indeed, let yn (t) = tn , where n 2 N. Then
kyn k = 1 and
kT 1 yn k
kT 1 yn k = kntn 1 k = n =) = n.
kyn k
Since n 2 N is arbitrary, this shows that there is no fixed number C > 0 such
kT 1 yn k
that  C, i.e. T 1 is not bounded.
kyn k

10. On C[0, 1] define S and T

Z 1
Sx(t) = y(t) = t x(s) ds T x(t) = y(t) = tx(t).
0

respectively. Do S and T commute? Find kSk, kT k, kST k and kT Sk.

Solution: They do not commute. Take x(t) = t 2 C[0, 1]. Then

Z 1
2 t
(ST )x(t) = S(t ) = t s2 ds = .
0 3
✓ ◆ 2
t t
(T S)x(t) = T = .
2 2

Page 43
11. Let X be the normed space of all bounded real-valued functions on R with norm
defined by
kxk = sup |x(t)|,
t2R

and let T : X ! X defined by

y(t) = T x(t) = x(t )
where > 0 is a constant. (This ia model of a delay line, which is an electric device
whose output y is a delayed version of the input x, the time delay being .) Is T
linear? Bounded?

Solution: For any x, z 2 X and scalars ↵, ,

T (↵x + z) = ↵x(t ) + z(t ) = ↵T x + T z.

This shows that T is linear. T is bounded since

kT xk = sup |x(t )| = kxk.

t2R

12. (Matrices) We know that an r ⇥ n matrix A = (↵jk ) defines a linear operator from
the vector space X of all ordered n-tuples of numbers into the vector space Y of all
ordered r-tuples of numbers. Suppose that any norm k · k1 is given on X and any
norm k · k2 is given on Y . Remember from Problem 10, Section 2.4, that there are
various norms on the space Z of all those matrices (r and n fixed). A norm k · k on
Z is said to be compatible with k · k1 and k · k2 if
kAxk2  kAkkxk1 .
Show that the norm defined by
kAxk2
kAk = sup
x2X kxk1
x6=0

is compatible with k · k1 and k · k2 . This norm is often called the natural norm defined
by k · k1 and kcdotk2 . IF we choose kxk1 = maxj |⇠j | and kyk2 = maxj |⌘j |, show that
the natural norm is n
X
kAk = max |↵jk |.
j
k=1

Solution:

13. Show that in 2.7-7 with r = n, a compatible norm is defined by

n X n
! 12
X
2
kAk = ↵jk ,
j=1 k=1

Page 44
 but for n > 1 this is not the natural norm defined by the Euclidean norm on Rn .

Solution:

14. If in Problem 12, we choose

n
X r
X
kxk1 = |⇠k | kyk2 = |⌘j |,
k=1 j=1

show that a compatible norm is defined by

r
X
kAk = max |↵jk |.
k
j=1

Solution:

15. Show that for r = n, the norm in Problem 14 is the natural norm corresponding to
k · k1 and k · k2 as defined in that problem.

Solution:

Page 45
1.7 Linear Functionals.
Definition 1.25.
1. A linear functional f is a linear operator with domain in a vector space X and
range in the scalar field K of X; thus, f : D(f ) ! K, where K = R if X is real
and K = C if X is complex.

2. A bounded linear functional f is a bounded linear operator with range in the

scalar field of the normed space X in which the domain D(f ) lies. Thus there exists
a nonnegative number C such that for all x 2 D(f ), |f (x)|  Ckxk. Furthermore,
the norm of f is
|f (x)|
kf k = sup = sup |f (x)|.
x2D(f ) kxk x2D(f )
x6=0 kxk=1

• As before, we have that |f (x)|  kf kkxk.

Theorem 1.26. A linear functional f with domain D(f ) in a normed space X is con-
tinuous if and only if f is bounded.

Definition 1.27.
1. The set of all linear functionals defined on a vector space X can itself be made
into a vector space. This space is denoted by X ⇤ and is called the algebraic dual
space of X. Its algebraic operations of vector space are defined in a natural way
as follows.
(a) The sum f1 + f2 of two functionals f1 and f2 is the functional s whose value
at every x 2 X is

s(x) = (f1 + f2 )(x) = f1 (x) + f2 (x).

(b) The product ↵f of a scalar ↵ and a functional f is the functional p whose value
at every x 2 X is
p(x) = (↵f )(x) = ↵f (x).

2. We may also consider the algebraic dual (X ⇤ )⇤ of X ⇤ , whose elements are the lin-
ear functionals defined on X ⇤ . We denote (X ⇤ )⇤ by X ⇤⇤ and call it the second
algebraic dual space of X.

• We can obtain an interesting and important relation between X and X ⇤⇤ , as

follows. We can obtain a g 2 X ⇤⇤ , which is a linear functional defined on X ⇤ ,
by choosing a fixed x 2 X and setting

g(f ) = gx (f ) = f (x) where x 2 X fixed, f 2 X ⇤ variable.

• gx is linear. Indeed,

gx (↵f1 + f2 ) = (↵f1 + f2 )(x) = ↵f1 (x) + f2 (x) = ↵gx (f1 ) + gx (f2 ).

Hence, gx is an element of X ⇤ ⇤, by the definition of X ⇤⇤ .

Page 46
 • To each x 2 X there corresponds a gx 2 X ⇤⇤ . This defines a mapping
C : X ! X ⇤⇤ , C : x 7! gx ; C is called the canonical mapping of X into
X ⇤⇤ . C is linear since its domain is a vector space and we have
(C(↵x + y))(f ) = g↵x+ y (f )
= f (↵x + y)
= ↵f (x) + f (y)
= ↵gx (f ) + gy (f )
= ↵(Cx)(f ) + (Cy)(f ).
C is also called the canonical embedding of X into X ⇤⇤ .
3. An metric space isomorphism T of a metric space X = (X, d) onto a metric
˜ is a bijective mapping which preserves distance, that is, for all
space X̃ = (X̃, d)
˜
x, y 2 X, d(T x, T y) = d(x, y). X̃ is then called isomorphic with X.
4. An vector space isomorphism T of a vector space X onto a vector space X̃ over
the same field is a bijective mapping which preserves the two algebraic operations
of vector space; thus, for all x, y 2 X and scalars ↵,
T (x + y) = T x + T y and T (↵x) = ↵T x,
that is, T : X ! X̃ is a bijective linear operator. X̃ is then called isomorphic
with X, and X and X̃ are called isomorphic vector spaces.
5. If X is isomorphic with a subspace of a vector space Y , we say that X is embed-
dable in Y .
• It can be shown that the canonical mapping C is injective. Since C is linear,
it is a vector space isomorphism of X onto the range R(C) ⇢ X ⇤⇤ .
• Since R(C) is a subspace of X ⇤⇤ , X is embeddable in X ⇤⇤ , and C is also called
the canonical embedding of X into X ⇤⇤ .
• If C is surjective (hence bijective), so that R(C) = X ⇤⇤ , then X is said to be
algebraically reflexive.

1. Show that the functionals in 2.8-7 and 2.8-8 are linear.

Solution: Choose a fixed t0 2 J = [a, b] and set f1 (x) = x(t0 ), where x 2 C[a, b].
For any x, y 2 C[a, b] and scalars ↵, ,

f1 (↵x + y) = (↵x + y)(t0 ) = ↵x(t0 ) + y(t0 ) = ↵f1 (x) + f1 (y).

1
X
2
Choose a fixed a = (aj ) 2 l and set f (x) = ⇠j ↵j , where x = (⇠j ) 2 l2 . For
j=1
2
any x = (⇠j ), y = (⌘j ) 2 l and scalars ↵, ,
1
X 1
X 1
X
f (↵x + y) = (↵⇠j + ⌘j )↵j = ↵ .⇠j ↵j + ⌘j ↵j = ↵f (x) + f (y).
j=1 j=1 j=1

Page 47
 Note we can split the infinite sum because the two infinite sums are convergent
by Cauchy-Schwarz inequality.

2. Show that the functionals defined on C[a, b] by

Z b
f1 (x) = x(t)y0 (t) dt (y0 2 C[a, b] fixed.)
a
f2 (x) = ↵x(a) + x(b) (↵, fixed.)

are linear and bounded.

Solution: For any x, y 2 C[a, b] and scalars , ,

Z bh i
f1 ( x + y) = x(t) + y(t) y0 (t) dt
a
Z b Z b
= x(t)y0 (t) dt + y(t)y0 (t) dt.
a a
= f1 (x) + f1 (y).
f2 ( x + y) = ↵( x + y)(a) + ( x + y)(b)
= ↵( x(a) + y(a)) + ( x(b) + y(b))
= (↵x(a) + x(b)) + (↵y(a) + y(b))
= f2 (x) + f2 (y).

To show that f1 and f2 are bounded, for any x 2 C[a, b],

Z b Z b
|f1 (x)| = x(t)y0 (t) dt  max |x(t)| y0 (t) dt
a t2[a,b] a
✓Z b ◆
= y0 (t) dt kxk.
a
|f2 (x)| = |↵x(a) + x(b)|  ↵ max |x(t)| + max |x(t)|
t2[a,b] t2[a,b]

= (↵ + )kxk.

3. Find the norm of the linear functional f defined on C[ 1, 1] by

Z 0 Z 1
f (x) = x(t) dt x(t) dt.
1 0

Page 48
 Solution:
Z 0 Z 1
|f (x)| = x(t) dt x(t) dt
1 0
Z 0 Z 1
 x(t) dt + x(t) dt
1 0
Z 0 Z 1
 kxk dt + kxk dt
1 0
= 2kxk

Taking the supremum over all x of norm 1, we obtain kf k  2. To get kf k 2,

we choose the particular x(t) 2 C[ 1, 1] defined by
8
> 1
>
> 2t 2 if 1t ,
>
> 2
<
1 1
x(t) = 2t if t ,
>
> 2 2
>
> 1
>
: 2t + 2 if  t  1.
2
Note that kxk = 1 and

|f (x)|
kf k  = |f (x)| = 2.
kxk

4. Show that for J = [a, b],

f1 (x) = max x(t) f2 (x) = min x(t)

t2J t2J

define functionals on C[a, b]. Are they linear? Bounded?

Solution: Both f1 and f2 define functionals on C[a, b] since continuous func-

tions attains its maximum and minimum on closed interval. They are not linear.
t a (t a)
Choose x(t) = and y(t) = . Then
b a b a

f1 (x + y) = 0 but f1 (x) + f1 (y) = 1 + 0 = 1.

f2 (x + y) = 0 but f2 (x) + f2 (y) = 0 1 = 1.

They are, however, bounded, since for any x 2 C[a, b],

|f1 (x)| = max x(t)  max |x(t)| = kxk.

t2J t2J
|f2 (x)| = min x(t)  max |x(t)| = kxk.
t2J t2J

Page 49
5. Show that on any sequence space X we can define a linear functional f by setting
f (x) = ⇠n (n fixed), where x = (⇠j ). Is f bounded if X = l1 ?

Solution: For any x = (⇠j ), y = (⌘j ) 2 X and scalars ↵, ,

f (↵x + y) = ↵⇠n + ⌘n = ↵f (x) + f (y).

f is bounded if X = l1 . Indeed, for any x 2 l1 ,

|f (x)| = |⇠n |  sup |⇠j | = kxk.

j2N

Remark : In fact, we can show that kf k = 1. Taking supremum over all x of

norm 1 on previous equation yields kf k  1. To get kf k 1, we choose the
particular x = (⇠j ) = ( jn ), note that kxk = 1 and

|f (x)|
kf k = |f (x)| = 1.
kxk

6. (Space C 1 [a, b]) The space C 1 [a, b] is the normed space of all continuously di↵er-
entiable functions on J = [a, b] with norm defined by
kxk = max |x(t)| + max |x0 (t)|.
t2J t2J

(a) Show that the axioms of a norm are satisfied.

Solution: (N1) and (N3) are obvious. For (N2), if x(t) ⌘ 0, then kxk = 0.
On the other hand, if kxk = 0, since both max |x(t)| and max |x0 (t)| are
t2J t2J
nonnegative, we must have |x(t)| = 0 and |x0 (t)| = 0 for all t 2 [a, b] which
implies x(t) ⌘ 0. Finally, (N4) follows from

kx + yk = max |x(t) + y(t)| + max |x0 (t) + y 0 (t)|

t2J t2J
 max |x(t)| + max |y(t)| + max |x0 (t)| + max |y 0 (t)|
t2J t2J t2J t2J
= kxk + kyk.

a+b
(b) Show that f (x) = x0 (c), c = , defines a bounded linear functional on
2
C 1 [a, b].

Solution: For any x, y 2 C 1 [a, b] and scalars ↵, ,

f (↵x + y) = ↵x0 (c) + y 0 (c) = ↵f (x) + f (y).
To see that f is bounded,
|f (x)| = |x0 (c)|  max |x0 (t)|  kxk.
t2J

Page 50
 (c) Show that f is not bounded, considered as functional on the subspace of C[a, b]
which consists of all continuously di↵erentiable functions.

Solution:

7. If f is a bounded linear functional on a complex normed space, is f¯ bounded? Linear?

(The bar denotes the complex conjugate.)

Solution: f¯ is bounded since |f (x)| = |f¯(x)|, but it is not linear since for any
x 2 X and complex numbers ↵, f (↵x) = ↵f (x) = ↵ ¯ f (x) 6= ↵f (x).

8. (Null space) The null space N (M ⇤ ) of a set M ⇤ ⇢ X ⇤ is defined to be the set of all
x 2 X such that f (x) = 0 for all f 2 M ⇤ . Show that N (M ⇤ ) is a vector space.

Solution: Since X is a vector space, it suffices to show that N (M ⇤ ) is a subspace

of X. Note that all element of M ⇤ are linear functionals. For any x, y 2 N (M ⇤ ),
we have f (x) = f (y) = 0 for all f 2 M ⇤ . Then for any f 2 M ⇤ and scalars ↵, ,
h i
f (↵x + y) = ↵f (x) + f (y) = 0. by linearity of f 2 M ⇤

This shows that ↵x + y 2 N (M ⇤ ) and the statement follows.

9. Let f 6= 0 be any linear functional on a vector space X and x0 any fixed element of
X \ N (f ), where N (f ) is the null space of f . Show that any x 2 X has a unique
representation x = ↵x0 + y, where y 2 N (f ).

Solution: Let f 6= 0 be any linear functional on X and x0 any fixed element

of X \ N (f ). We claim that for any x 2 X, there exists a scalar ↵ such that
x = ↵x0 + y, where y 2 N (f ). First, applying f on both sides yields

f (x) = f (↵x0 + y) = ↵f (x0 ) + f (y) =) f (y) = f (x) ↵f (x0 ).

f (x)
By choosing ↵ = (which is well-defined since f (x0 ) 6= 0), we see that
f (x0 )

f (x)
f (y) = f (x) f (x0 ) = 0 =) y 2 N (f ).
f (x0 )

To show uniqueness, suppose x has two representations x = ↵1 x0 +y1 = ↵2 x0 +y2 ,

where ↵1 , ↵2 are scalars and y1 , y2 2 N (f ). Subtracting both representations
yields
(↵1 ↵2 )x0 = y2 y1 .
Applying f on both sides gives

f ((↵1 ↵2 )x0 ) = f (y2 y1 )

Page 51
 (↵1 ↵2 )f (x0 ) = f (y2 ) f (y1 ) = 0

by linearity of f and y1 , y2 2 N (f ). Since f (x0 ) 6= 0, we must have ↵1 ↵2 = 0.

This also implies y1 y2 = 0.

10. Show that in Problem 9, two elements x1 , x2 2 X belong to the same element of the
quotient space X/N (f ) if and only if f (x1 ) = f (x2 ). Show that codim N (f ) = 1.

Solution: Suppose two elements x1 , x2 2 X belong to the same element of the

quotient space X/N (f ). This means that there exists an x 2 X and y1 , y2 2 N (f )
such that
x1 = x + y1 and x2 = x + y2 .
Substracting these equations and applying f yields

x1 x2 = y1 y2 =) f (x1 x2 ) = f (y1 y2 )
=) f (x1 ) f (x2 ) = f (y1 ) f (y2 ) = 0.

where we use the linearity of f and y1 , y2 2 N (f ). Conversely, suppose f (x1 )

f (x2 ) = 0; linearity of f gives

f (x1 x2 ) = 0 =) x1 x2 2 N (f ).

This means that there exists y 2 N (f ) such that x1 x2 = 0 + y, which implies

that x1 , x2 2 X must belong to the same coset of X/N (f ).

Codimension of N (f ) is defined to be the dimension of the quotient space X/N (f ).

Choose any x̂ 2 X/N (f ), there exists an x 2 X such that x̂ = x + N (f ). Since
f 6= 0, there exists an x0 2 X \ N (f ) such that f (x0 ) 6= 0. Looking at Problem 9,
we deduce that x̂ has a unique representation x̂ = ↵x0 + N (f ) = ↵(x0 + N (f )).
This shows that x0 + N (f ) is a basis for X/N (f ) and codim N (f ) = 1.

11. Show that two linear functionals f1 6= 0 and f2 6= 0 which are defined on the same
vector space and have the same null space are proportional.

Solution: Let x, x0 2 X and consider z = xf1 (x0 ) x0 f1 (x). Clearly, f1 (z) =

0 =) z 2 N (f1 ) = N (f2 ). Thus,

0 = f2 (z) = f2 (x)f1 (x0 ) f2 (x0 )f1 (x).

Since f1 6= 0, there exists some x0 2 X \ N (f1 ) such that f1 (x0 ) 6= 0; we also have
f2 (x0 ) 6= 0 since N (f1 ) = N (f2 ). Hence, for such an x0 , we have
f2 (x0 )
f2 (x) = f1 (x).
f1 (x0 )
Since x 2 X is arbitrary, the result follows.

Page 52
12. (Hyperplane) If Y is a subspace of a vector space X and codimY = 1, then every
element of X/Y is called a hyperplane parallel to Y . Show that for any linear func-
tional f 6= 0 on X, the set H1 = {x 2 X : f (x) = 1} is a hyperplane parallel to the
null space N (f ) of f .

Solution: Since f 6= 0 on X, H1 is not empty. Fix an x0 2 H1 , and consider

the coset x0 + N (f ). Note that this is well-defined irrespective of elements in H1 .
Indeed, for any y 2 H1 , y 6= x0 , y x0 2 N (f ) since f (y x0 ) = f (y) f (x0 ) =
1 1 = 0; this shows that x + N (f ) = y + N (f ) for any x, y 2 H1 .

• For any x 2 x0 + N (f ), there exists an y 2 N (f ) such that x = x0 + y.

Since f is linear, f (x) = f (x0 + y) = f (x0 ) + f (y) = 1 =) x 2 H1 . This
shows that x0 + N (f ) ⇢ H1 .

• For any x 2 H1 , x = x + 0 = x + x0 x0 = x0 + (x x0 ) 2 x0 + N (f ) since

f (x x0 ) = f (x) f (x0 ) = 1 1 = 0. This shows that H1 ⇢ x0 + N (f ).

Finally, combining the two set inequality gives H1 = x0 +N (f ) and the statement
follows.

13. If Y is a subspace of a vector space X and f is a linear functional on X such that

f (Y ) is not the whole scalar field of X, show that f (y) = 0 for all y 2 Y .

Solution: The statement is trivial if f is the zero functional, so suppose f 6= 0.

Suppose, by contradiction, that f (y) 6= 0 for all y 2 Y , then there exists an
y0 2 Y such that f (y0 ) = ↵ for some nonzero ↵ 2 K. Since Y is a subspace of a
vector space X, y0 2 Y for all 2 K. By linearity of f ,

f ( y0 ) = f (y0 ) = ↵ 2 f (Y ).

Since 2 K is arbitrary, this implies that f (Y ) = K; this is a contradiction to

the assumption that f (Y ) 6= K. Hence, by proof of contradiction, f (y) = 0 for
all y 2 Y .

14. Show that the norm kf k of a bounded linear functional f 6= 0 on a normed space X
can be interpreted geometrically as the reciprocal of the distance d˜ = inf{kxk : f (x) =
1k of the hyperplane H1 = {x 2 X : f (x) = 1} from the origin.

Solution:

15. (Half space) Let f 6= 0 be a bounded linear functional on a real normed space
X. Then for any scalar c we have a hyperplane Hc = {x 2 X : f (x) = c}, and Hc
determines the two half spaces

Xc1 = {x 2 X : f (x)  c} and Xc2 = {x 2 X : f (x) c}.

Page 53
Show that the closed unit ball lies in Xc1 , where c = kf k, but for no " > 0, the half
space Xc1 with c = kf k " contains that ball.

Solution:

Page 54
1.8 Linear Operators and Functionals on Finite Dimensional
Spaces.
• A linear operator T : X ! Y determines a unique matrix representing T with
respect to a given basis for X and a given basis for Y , where the vectors of each of
the bases are assumed to be arranged in a fixed order. Conversely, any matrix with
r rows and n columns determines a linear operator which it represents with respect
to given bases for X and Y .

• Let us now turn to linear functionals on X, where P dim X = n and {e1 , . . . , en }

⇤
is a basis for X. For every f 2 X and every x = ⇠j ej 2 X, we have
n
! n n
X X X
f (x) = f ⇠j ej = ⇠j f (ej ) = ⇠j ↵j .
j=1 j=1 j=1

where ↵j = f (ej ) for j = 1, . . . , n. We see that f is uniquely determined by its val-

ues ↵j ath the n basis vectors of X. Conversely, every n-tuple of scalars ↵1 , . . . , ↵n
determines a linear functional on X.

Theorem 1.28. Let X be an n-dimensional vector space and E = {e1 , . . . , en } a basis

for X. Then F = {f1 , . . . , fn } given by fk (ej ) = jk is a basis for the algebraic dual X ⇤
of X, and dim X ⇤ = dim X = n.

• {f1 , . . . , fn } is called the dual basis of the basis {e1 , . . . , en } for X.

Lemma 1.29. Let X be a finite dimensional vector space. If x0 2 X has the property
that f (x0 ) = 0 for all f 2 X ⇤ , then x0 = 0.

Theorem 1.30. A finite dimensional vector space is algebraically reflexive.

1. Determine the null space of the operator T : R3 ! R2 represented by


1 3 2
.
2 1 0

Solution: Performing Gaussian elimination on the matrix yields

 
1 3 2 0 1 3 2 0
! .
2 1 0 0 0 7 4 0

This yields a solution of the form (⇠1 , ⇠2 , ⇠3 ) = t( 2, 4, 7) where t 2 R is a free

variable. Hence, the null space of T is the span of ( 2, 4, 7).

Page 55
2. Let T : R3 ! R3 be defined by (⇠1 , ⇠2 , ⇠3 ) 7! (⇠1 , ⇠2 , ⇠1 ⇠2 ). Find R(T ), N (T )
and a matrix which represents T .

Solution: Consider the standard basis for X, given by e1 = (1, 0, 0), e2 = (0, 1, 0),
e3 = (0, 0, 1). The matrix representing T with respect to {e1 , e2 , e3 } is
2 3
1 0 0
A=4 0 1 05 .
1 1 0

The range of T , R(T ) is the plane ⇠1 + ⇠2 + ⇠3 = 0. The null space of T , N (T ) is

span of (0, 0, 1).

3. Find the dual basis of the basis {(1, 0, 0), (0, 1, 0), (0, 0, 1)} for R3 .

Solution: Consider a basis {e1 , e2 , e3 } of R3 defined by ej = (⇠nj ) = jn for

X3
j = 1, 2, 3. Given any x = (⌘j ) in R3 , let fk (x) = ↵jk ⌘j be the dual basis
j=1
of {e1 , e2 , e3 }. From the definition of a dual basis, we require that fk (ej ) = jk .
More precisely, for f1 , we require that

f1 (e1 ) = ↵11 = 1.
f1 (e2 ) = ↵21 = 0.
f1 (e3 ) = ↵31 = 0.

which implies that f1 (x) = ⌘1 . Repeating the same computation for f2 and f3 ,
we find that f2 (x) = ⌘2 and f3 (x) = ⌘3 . Hence,

f1 = (1, 0, 0) f2 = (0, 1, 0) f3 = (0, 0, 1).

4. Let {f1 , f2 , f3 } be the dual basis of {e1 , e2 , e3 } for R3 , where e1 = (1, 1, 1), e2 =
(1, 1, 1), e3 = (1, 1, 1). Find f1 (x), f2 (x), f3 (x), where x = (1, 0, 0).

Solution: Note that we can write x as

2 3 2 3 2 3 2 3
1 1 1 1
4 5 14 5 4 5 14 5 1 1
x= 0 = 1 +0 1 + 1 = e1 + e3 .
2 2 2 2
0 1 1 1

Thus, using the defintiion of a dual basis fk (ej ) = jk , we have

1 1 1
f1 (x) = f1 (e1 ) + f1 (e3 ) = .
2 2 2

Page 56
 1 1
f2 (x) = f2 (e1 ) + f2 (e3 ) = 0.
2 2
1 1 1
f3 (x) = f3 (e1 ) + f3 (e3 ) = .
2 2 2
where we use linearity of f1 , f2 , f3 .

5. If f is a linear functional on an n-dimensional vector space X, what dimension can

the null space N (f ) have?

Solution: The Rank-Nullity theorem states that

dim (N (f )) = dim (X) dim (R(f )) = n dim (R(f )).

If f is the zero functional, then N (f ) = X and N (f ) has dimension n; if f is

not the zero functional, then R(f ) = K which has dimension 1, so N (f ) has
dimension n 1. Hence, N (f ) has dimension n or n 1.

6. Find a basis for the null space of the functional f defined on R3 by f (x) = ⇠1 +⇠2 ⇠3 ,
where x = (⇠1 , ⇠2 , ⇠3 ).

Solution: Let x = (⇠1 , ⇠2 , ⇠3 ) be any point in the null space of f , they must
satisfy the relation ⇠1 + ⇠2 ⇠3 = 0. Thus,
2 3 2 3 2 3 2 3
⇠1 ⇠1 1 0
4
x = ⇠2 = 5 4 ⇠2 5 = ⇠1 0 + ⇠2 15 .
4 5 4
⇠3 ⇠1 + ⇠2 1 1

Hence, a basis for N (f ) is given by {(1, 0, 1), (0, 1, 1)}.

7. Same task as in Problem 6, if f (x) = ↵1 ⇠1 + ↵2 ⇠2 + ↵3 ⇠3 , where ↵1 6= 0.

Solution: Let x = (⇠1 , ⇠2 , ⇠3 ) be any point in N (f ), they must satisfy the relation
↵2 ↵3
↵1 ⇠1 + ↵2 ⇠2 + ↵3 ⇠3 = 0 () ⇠1 = ⇠2 ⇠3 .
↵1 ↵1
Rewriting x using this relation yields
2 3 2 ↵2 ↵3 3 2 3 2 3
⇠1 ⇠2 ⇠3 ↵2 ↵3
6 ↵1 ↵1 7 ⇠2 4 ⇠3 4
x = 4⇠2 5 = 4 ⇠2 5= ↵1 5 + 0 5.
↵1 ↵1
⇠3 ⇠3 0 ↵1

Hence, a basis for N (f ) is given by {( ↵2 , ↵1 , 0), ( ↵3 , 0, ↵1 )}.

Page 57
8. If Z is an (n 1)-dimensional subspace of an n-dimensional vector space X, show
that Z is the null space of a suitable linear functional f on X, which is uniquely
determined to within a scalar multiple.

Solution: Let X be an n-dimensional vector space, and Z an (n 1)-dimensional

subspace of X. Choose a basis A = {z1 , . . . , zn 1 } of Z, here we can obtain a basis
B = {z1 , . . . , zn 1 , zn } of X, where B is obtained by extending A using sifting
method. Any z 2 Z can be written uniquely as z = ↵1 z1 + . . . + ↵n 1 zn 1 . If we
want to find f 2 X ⇤ such that N (f ) = Z, using linearity of f this translates to
f (z) = f (↵1 z1 + . . . + ↵n 1 zn 1 ) = ↵1 f (z1 ) + . . . + ↵n 1 f (zn 1 ) = 0.
Since this must be true for all z 2 Z, it enforces the condition f (zj ) = 0 for all
j = 1, . . . , n 1. A similar argument shows that f (zn ) 6= 0, otherwise N (f ) =
X 6= Z. Thus, the functional we are looking for must satisfy the following two
conditions:
(a) f (zj ) = 0 for all j = 1, . . . , n 1.
(b) f (zn ) 6= 0.

Consider a linear functional f : X ! K defined by f (zj ) = jn for all j =

1, . . . , n. We claim that N (f ) = Z. Indeed,
• Choose any z 2 Z, there exists a unique sequence of scalars ( j ) such that
z = 1 z1 + . . . + n 1 zn 1 . Using linearity of f , we have
n 1
! n 1
X X
f (z) = f j zj = j f (zj ) = 0.
j=1 j=1

This shows that Z ⇢ N (f ).

• Choose any x 2 X \ Z, there exists a unique sequence of scalars ( j ) such
that x = 1 z1 + . . . + n zn and n 6= 0. (otherwise x 2 Z.) Using linearity
of f , we have
n
! n
X X
f (x) = f j zj = j f (zj ) = n f (zn ) 6= 0.
j=1 j=1

This shows that (X \ Z) 6⇢ N (f ) or equivalently N (f ) ⇢ Z.

Hence, Z ⇢ N (f ) and N (f ) ⇢ Z implies Z = N (f ).

We are left to show that f is uniquely determined up to scalar multiple. Let

be any nonzero scalars and consider the linear functional f . Any x 2 X can be
written uniquely as x = ↵1 z1 + . . . + ↵n zn . Using linearity of f , we have
n
! n
!
X X
( f )(x) = f (x) = f ↵j zj = ↵j f (zj ) = ↵n f (zn ).
j=1 j=1

If ↵n = 0, then x 2 Z and ( f )(x) = 0; if ↵n 6= 0, then x 2 X \ Z and

( f )(x) = ↵n 6= 0.

Page 58
 9. Let X be the vector space of all real polynomials of a real variable and of degree less
than a given n, together with the polynomial x = 0 (whose degree is left undefined
in the usual discussion of degree). Let f (x) = x(k) (a), the value of the kth derivative
(k fixed) of x 2 X at a fixed a 2 R. Show that f is a linear functional on X.

Solution: This follows from the algebra rules of di↵erentiation. More precisely,
for any x, y 2 X and scalars ↵, ,

f (↵x + y) = (↵x + y)(k) (a) = ↵x(k) (a) + y (k) (a) = ↵f (x) + f (y).

10. Let Z be a proper subspace of an n-dimensional vector space X, and let x0 2 X \ Z.

Show that there is a linear functional f on X such that f (x0 ) = 1 and f (x) = 0 for
all x 2 Z.

Solution: Fix an nonzero x0 2 X \ Z, note that x0 = 0 then such a functional

doesn’t exist since f (x0 ) = 0. Let X be an n-dimensional vector space, and Z be
an m-dimensional subspace of X, with m < n. Choose a basis A = {z1 , . . . , zm } of
Z, here we can obtain a basis B = {z1 , . . . , zm , zm+1 , . . . , zn } of X with zm+1 = x0 ,
where B is obtained by extending A using sifting method. Any x 2 X can be
written uniquely as

x = ↵1 z1 + . . . + ↵m zm + ↵m+1 x0 + ↵m+2 zm+2 + . . . + ↵n zn .

Consider the linear functional f : X ! K defined by f (x) = ↵m+1 + . . . + ↵n ,

where ↵j is the j-th scalar of x with respect to the basis B for all j = m+1, . . . , n.
We claim that f (x0 ) = 1 and f (Z) = 0. Indeed,

• If x = x0 , then ↵m+1 = 1 and ↵j = 0 for all j 6= m + 1.

• If x 2 Z, then ↵j = 0 for all j = m + 1, . . . , n.

• If x 2 X \ Z, at least one of {↵m+1 , . . . , ↵n } is non-zero.

Remark : The functional f we constructed here is more tight, in the sense that
N (f ) = Z. If we only require that Z ⇢ N (f ), then f (x) = ↵m+1 will do the job.

11. If x and y are di↵erent vectors in a finite dimensional vector space X, show that
there is a linear functional f on X such that f (x) 6= f (y).

Solution: Let X be an n-dimensional vector space and {e1 , . . . , en } a basis of

X. Any x, y 2 X can be written uniquely as
n
X n
X
x= ↵j ej and y= j ej .
j=1 j=1

Page 59
 Since x 6= y, there exists at least one j0 2 {1, . . . , n} such that ↵j0 6= j0 . Consider
the linear functional f : X ! K defined by f (ej ) = jj0 . Using linearity of f ,
n
! n
X X
f (x) = f ↵j ej = ↵j f (ej ) = ↵j0 .
j=1 j=1
n
! n
X X
f (y) = f j ej = j f (ej ) = j0 .
j=1 j=1

Clearly, f (x) 6= f (y).

12. If f1 , . . . , fp are linear functionals on an n-dimensional vector space X, where p < n,

show that there is a vector x 6= 0 in X such that f1 (x) = 0, . . . , fp (x) = 0. What
consequences does this result have with respect to linear equations?

Solution:

13. (Linear extension) Let Z be a proper subspace of an n-dimensional vector space

X, and let f be a linear functional on Z. Show that f can be extended linearly to
X, that is, there is a a linear functional f˜ on X such that f˜|Z = f .

Solution:

14. Let the functional f on R2 be defined by f (x) = 4⇠1 3⇠2 , where x = (⇠1 , ⇠2 ). Regard
R2 as the subspace of R3 given by ⇠3 = 0. Determine all linear extensions f˜ of f from
R2 to R3 .

Solution:

15. Let Z ⇢ R3 be the subspace represented by ⇠2 = 0 and let f on Z be defined by

⇠1 ⇠3
f (x) = . Find a linear extension f˜ of f to R3 such that f˜(x0 ) = k (a given
2
constant), where x0 = (1, 1, 1). Is f˜ unique?

Solution:

Page 60