This action might not be possible to undo. Are you sure you want to continue?
Alexander C. R. Belton
Copyright c _ Alexander C. R. Belton 2004, 2006
Hyperlinked and revised edition
All rights reserved
The right of Alexander Belton to be identiﬁed as the author
of this work has been asserted by him in accordance with
the Copyright, Designs and Patents Act 1988.
Contents
Contents i
Introduction iii
1 Normed Spaces 3
1.1 Basic Deﬁnitions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3
1.2 Subspaces and Quotient Spaces . . . . . . . . . . . . . . . . . . . . . . 4
1.3 Completions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6
1.4 Direct Sums . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7
1.5 Initial Topologies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9
1.6 Nets . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10
1.7 Exercises 1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14
2 Linear Operators 17
2.1 Preliminaries . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
2.2 Completeness of B(X, Y ) . . . . . . . . . . . . . . . . . . . . . . . . . . 18
2.3 Extension of Linear Operators . . . . . . . . . . . . . . . . . . . . . . . 18
2.4 The Baire Category Theorem . . . . . . . . . . . . . . . . . . . . . . . 19
2.5 The OpenMapping Theorem . . . . . . . . . . . . . . . . . . . . . . . 21
2.6 Exercises 2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22
2.7 The ClosedGraph Theorem . . . . . . . . . . . . . . . . . . . . . . . . 26
2.8 The Principle of Uniform Boundedness . . . . . . . . . . . . . . . . . . 27
2.9 The Strong Operator Topology . . . . . . . . . . . . . . . . . . . . . . 27
2.10 Exercises 3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
3 Dual Spaces 31
3.1 Initial Deﬁnitions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
3.2 The Weak Topology . . . . . . . . . . . . . . . . . . . . . . . . . . . . 31
3.3 Zorn’s Lemma . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 32
3.4 The HahnBanach Theorem . . . . . . . . . . . . . . . . . . . . . . . . 32
3.5 The Weak Operator Topology . . . . . . . . . . . . . . . . . . . . . . . 36
3.6 Adjoint Operators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36
3.7 The Weak* Topology . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
3.8 Exercises 4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 37
3.9 Tychonov’s Theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39
i
ii Contents
3.10 The BanachAlaoglu Theorem . . . . . . . . . . . . . . . . . . . . . . . 40
3.11 Topological Vector Spaces . . . . . . . . . . . . . . . . . . . . . . . . . 41
3.12 The KreinMilman Theorem . . . . . . . . . . . . . . . . . . . . . . . . 45
3.13 Exercises 5 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 47
4 Normed Algebras 53
4.1 Quotient Algebras . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 54
4.2 Unitization . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 55
4.3 Approximate Identities . . . . . . . . . . . . . . . . . . . . . . . . . . . 56
4.4 Completion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57
5 Invertibility 59
5.1 The Spectrum and Resolvent . . . . . . . . . . . . . . . . . . . . . . . . 60
5.2 The GelfandMazur Theorem . . . . . . . . . . . . . . . . . . . . . . . 61
5.3 The SpectralRadius Formula . . . . . . . . . . . . . . . . . . . . . . . 61
5.4 Exercises 6 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 63
6 Characters and Maximal Ideals 65
6.1 Characters and the Spectrum . . . . . . . . . . . . . . . . . . . . . . . 66
6.2 The Gelfand Topology . . . . . . . . . . . . . . . . . . . . . . . . . . . 67
6.3 The Representation Theorem . . . . . . . . . . . . . . . . . . . . . . . 68
6.4 Examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 68
6.5 Exercises 7 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72
A Tychonov via Nets 75
A.1 Exercises A . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78
Solutions to Exercises 79
Exercises 1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 79
Exercises 2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 83
Exercises 3 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88
Exercises 4 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92
Exercises 5 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 97
Exercises 6 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105
Exercises 7 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109
Exercises A . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 116
Bibliography 117
Index 119
Introduction
These notes are an expanded version of a set written for a course given to ﬁnalyear
undergraduates at the University of Oxford.
A thorough understanding of the Oxford thirdyear b4 analysis course (an introduction
to Banach and Hilbert spaces) or its equivalent is a prerequisite for this material. We
use [24] as a compendium of results from that series of lectures. (Numbers in square
brackets refer to items in the bibliography.)
The author acknowledges his debt to all those from whom he has learnt functional
analysis, especially Professor D. A. Edwards, Dr G. R. Allen and Dr J. M. Lindsay. The
students attending the course were very helpful, especially Mr A. Evseev, Mr L. Taitz
and Ms P. Iley.
This document was typeset using L
A
T
E
X2
ε
with Peter Wilson’s memoir class and the
/
/
oL
A
T
E
X andX
Y
pic packages. The index was produced with the aid of the MakeIndex
program.
Alexander C. R. Belton
Lady Margaret Hall
Oxford
20th August 2004
This edition contains a few additional exercises and the electronic version is equipped
with hyperlinks, thanks to the hyperref package of Sebastian Rahtz and Heiko Oberdiek.
ACRB
University College
Cork
30th September 2006
Convention
Throughout these notes we follow the Diracformalism convention that inner products
on complex vector spaces are conjugate linear in the ﬁrst argument and linear in the
second, in contrast to many Oxford courses.
iii
Spaces
1
One Normed Spaces
Throughout, the scalar ﬁeld of a vector space will be denoted by F and will be either
the real numbers R or the complex numbers C.
Basic Deﬁnitions
Deﬁnition 1.1. A norm on a vector space X is a function
 : X →R
+
:= [0, ∞); x → x
that satisﬁes, for all x, y ∈ X and α ∈ F,
(i) x = 0 if and only if x = 0 (faithfulness),
(ii) αx = [α[ x (homogeneity)
and (iii) x + y x +y (subadditivity).
A seminorm on X is a function p: X →R
+
that satisﬁes (ii) and (iii) above.
Deﬁnition 1.2. A normed vector space is a vector space X with a norm  ; if necessary
we will denote the norm on the space X by 
X
. We will sometimes use the term normed
space as an abbreviation.
Deﬁnition 1.3. A Banach space is a normed vector space (E,  ) that is complete,
i.e., every Cauchy sequence in E is convergent, where E is equipped with the metric
d(x, y) := x −y.
Deﬁnition 1.4. Let (x
n
)
n1
be a sequence in the normed vector space X. The series
∞
n=1
x
n
is convergent if there exists x ∈ X such that
_
n
k=1
x
k
_
n1
is convergent to x,
and the series is said to have sum x. The series is absolutely convergent if
∞
n=1
x
n
 is
convergent.
Theorem 1.5. (Banach’s Criterion) A normed vector space X is complete if and
only if every absolutely convergent series in X is convergent.
Proof
This is a b4 result: see [24, Theorem 1.2.9].
3
4 Normed Spaces
Subspaces and Quotient Spaces
Deﬁnition 1.6. A subspace of a vector space X is a subset M ⊆ X that is closed under
vector addition and scalar multiplication: M + M ⊆ M and αM ⊆ M for all α ∈ F,
where
A + B := ¦a + b : a ∈ A, b ∈ B¦ and αA := ¦αa : a ∈ A¦ ∀ A, B ⊆ X, α ∈ F.
Example 1.7. Let (X, T) be a topological space and let (E,  
E
) be a Banach space
over F. The set of continuous, Evalued functions on X forms an vector space, denoted
by C(X, E), where the algebraic operations are deﬁned pointwise: if f, g ∈ C(X, E) and
α ∈ F then
(f + g)(x) := f(x) + g(x) and (αf)(x) := αf(x) ∀ x ∈ X.
The subspace of bounded functions
C
b
(X, E) :=
_
f ∈ C(X, E)
¸
¸
f
∞
< ∞
_
,
where
 
∞
: C
b
(X, E) →R
+
; f → sup
_
f(x)
E
: x ∈ X
_
,
is a Banach space, with supremum norm  
∞
(see Theorem 1.36). If X is compact
then every continuous, Evalued function is bounded, hence C(X, E) = C
b
(X, E) in this
case. If E = C (the most common case of interest) we use the abbreviations C(X) and
C
b
(X).
Proposition 1.8. A subspace of a Banach space is closed if and only if it is complete.
Proof
See [24, Theorem 1.2.10].
Deﬁnition 1.9. Given a vector space X with a subspace M, the quotient space X/M
is the set
X/M :=
_
[x] := x + M
¸
¸
x ∈ X
_
, where x + M := ¦x + m : m ∈ M¦,
equipped with the vectorspace operations
[x] + [y] := [x + y] and α[x] := [αx] ∀ x, y ∈ X, α ∈ F.
(It is a standard result of linear algebra that this deﬁnes a vector space; for a full
discussion see [7, Appendix A.4].) The dimension of X/M is the codimension of M (in
X).
Theorem 1.10. Let X be a normed vector space with a subspace M and let
_
_
[x]
_
_
X/M
:= inf
_
x −m : m ∈ M
_
∀ [x] ∈ X/M.
1.2. Subspaces and Quotient Spaces 5
This deﬁnes a seminorm on X/M, which is a norm if and only if M is closed, called
the quotient seminorm (or quotient norm) on X/M. If E is a Banach space and M is a
closed subspace of E then
_
E/M,  
E/M
_
is a Banach space.
Proof
Clearly
_
_
[x]
_
_
X/M
= 0 if and only if d(x, M) = 0, which holds if and only if x ∈
¯
M.
Hence
_
_
_
_
X/M
is faithful if and only if M is closed.
If α ∈ F and x ∈ X then
_
_
α[x]
_
_
X/M
=
_
_
[αx]
_
_
X/M
= inf
_
αx −m : m ∈ M
_
= inf
_
[α[ x −n : n ∈ M
_
= [α[
_
_
[x]
_
_
X/M
,
using the fact that α
−1
M = M if α ,= 0 (because M is a subspace).
For subadditivity, let x, y ∈ X and note that
_
_
[x] + [y]
_
_
X/M
=
_
_
[x + y]
_
_
X/M
x + y −(m+ n) x −m +y −n
for all m, n ∈ M. Taking the inﬁmum over such m and n gives the result.
We prove the ﬁnal claim in Proposition 2.15 as a consequence of the openmapping
theorem; see also Exercise 1.2.
Example 1.11. Let I be a subinterval of R and let p ∈ [1, ∞). The vector space of
Lebesguemeasurable functions on I that are pintegrable is denoted by L
p
(I):
L
p
(I) :=
_
f : I →C
¸
¸
f is measurable and f
p
< ∞
_
,
with vectorspace operations deﬁned pointwise and
f
p
:=
_
_
I
[f(x)[
p
dx
_
1/p
.
(Note that
[f + g[
p
_
[f[ +[g[
_
p
_
2 max¦[f[, [g[¦
_
p
= 2
p
max¦[f[
p
, [g[
p
¦ 2
p
_
[f[
p
+[g[
p
_
,
so L
p
(I) is closed under addition; it is simple to verify that L
p
(I) is a vector space.)
The map f → f
p
is a seminorm, but not a norm; the subadditivity of  
p
is known
as Minkowski’s inequality (see [17, Theorem 28.19] for its proof).
If N := ¦f ∈ L
p
(I) : f
p
= 0¦ then L
p
(I) := L
p
(I)/N is a Banach space, with
norm [f] →
_
_
[f]
_
_
p
:= f
p
. (A function lies in N if and only if it is zero almost
everywhere.) As is usual practise in functional analysis, we shall frequently blur the
distinction between f and [f]. (Discussion of L
p
(R) may be found in [17, Chapter 28]
and [26, Chapter 7]; the generalisation from R to a subinterval I is trivial.)
Example 1.12. Let I be a subinterval of R and let L
∞
(I) denote the vector space of
Lebesguemeasurable functions on I that are essentially bounded:
L
∞
(I) :=
_
f : I →C
¸
¸
f is measurable and f
∞
< ∞¦,
6 Normed Spaces
with vectorspace operations as usual and
f
∞
:= inf¦M : [f(x)[ M almost everywhere¦.
(It is not diﬃcult to show that f
∞
= sup¦[f(x)[ : x ∈ I ¸ N¦ for some null set N
which may, of course, depend on f.)
As in the previous example, f → f
∞
is a seminorm, N = ¦f ∈ L
∞
(I) : f
∞
= 0¦
consists of those functions that are zero almost everywhere and L
∞
(I) := L
∞
/N is a
Banach space with respect to the norm [f] →
_
_
[f]
_
_
∞
:= f
∞
.
Although it may seem that we have two diﬀerent meanings for f
∞
, the above and that
in Example 1.7, they coincide if f is continuous.
Example 1.13. Let Ω be an open subset of the complex plane C and let
H
b
(Ω) := ¦f : Ω →C [ f is bounded and holomorphic in Ω¦.
Equipped with the supremum norm on Ω, H
b
(Ω) is a Banach space. (Completeness is
most easily established via Morera’s theorem [16, Theorem 5.6].)
Completions
Recall that a map f : X → Y between metric spaces (X, d
X
) and (Y, d
Y
) is an isometry
if d
Y
_
f(x
1
), f(x
2
)
_
= d
X
(x
1
, x
2
) for all x
1
, x
2
∈ X, and an isometric isomorphism
between normed vector spaces is an invertible linear isometry (the inverse of which is
automatically linear and isometric).
Theorem 1.14. If X is a normed vector space then there exists a Banach space
˜
X and a
linear isometry i : X →
˜
X such that i(X) is dense in
˜
X. The pair
_
˜
X, i
_
is a completion
of X, and is unique in the following sense: if (Y, i) and (Z, j) are completions of X then
there exists an isometric isomorphism k: Y → Z such that k
◦
i = j.
Proof
We defer this until we have developed more machinery; see Propositions 3.11 and
2.6.
As we have uniqueness, we talk about the completion of a normed vector space.
The process of completing a given space may often be simpliﬁed by realising it as a dense
subspace of some known Banach space. The following examples demonstrate this.
Example 1.15. If (X, T) is a topological space and (E,  ) a Banach space then
C
b
(X, E) contains two subspaces worthy of note:
(i) C
0
(X, E), the continuous, Evalued functions on X that vanish at inﬁnity (i.e.,
those f ∈ C(X, E) such that, for all ε > 0, the set ¦x ∈ X : f(x) ε¦ is
compact);
1.4. Direct Sums 7
(ii) C
00
(X, E), the continuous, Evalued functions on X with compact support (i.e.,
those f ∈ C(X, E) such that supp f := ¦x ∈ X : f(x) ,= 0¦ is compact).
The ﬁrst is a closed subspace of C
b
(X, E) (the proof of this is Exercise 2.1) and if X
is Hausdorﬀ and locally compact (for all x ∈ X there exists U ∈ T such that x ∈ U
and
¯
U is compact) then C
00
(X, E) is dense in C
0
(X, E). Hence the latter space is the
completion of the former; for the proof of this claim see Proposition 2.18.
Example 1.16. It is immediate that C[0, 1] is a subspace of L
1
[0, 1] (because of the
inequality f
1
f
∞
). Furthermore, since step functions can be approximated arbi
trarily well by continuous functions (with respect to the  
1
norm), C[0, 1] is dense in
L
1
[0, 1]; more accurately, its image under the map f → [f] is. Hence the completion of
C[0, 1] (with respect to  
1
) is L
1
[0, 1].
Direct Sums
Throughout this section
_
E
a
: a ∈ A
_
denotes a family of Banach spaces with common
scalar ﬁeld F.
Deﬁnition 1.17. Let
a∈A
E
a
denote the algebraic direct sum of the spaces E
a
, i.e.,
a∈A
E
a
:=
_
x := (x
a
)
a∈A
¸
¸
x
a
= 0 for all but ﬁnitely many a ∈ A
_
⊆
a∈A
E
a
.
This is a vector space, with the vectorspace operations deﬁned pointwise:
x + y := (x
a
+ y
a
)
a∈A
and αx := (αx
a
)
a∈A
∀ x, y ∈
a∈A
E
a
, α ∈ F.
Theorem 1.18. The set
a∈A
E
a
:=
_
x = (x
a
)
a∈A
: x
∞
< ∞
_
⊆
a∈A
E
a
,
equipped with vectorspace operations deﬁned pointwise and norm
 
∞
: x → sup
_
x
a
 : a ∈ A
_
,
is a Banach space, the direct product of the Banach spaces E
a
.
Proof
The only thing that is not immediate is the proof of completeness; this follows the
pattern of the b4 proof that ℓ
∞
is complete [24, Example 1.3.4] so we omit it.
For other types of direct sum we need the following deﬁnition.
Deﬁnition 1.19. (Uncountable Sums) If A is an arbitrary set let
a∈A
x
a
:= sup
_
a∈A
0
x
a
: A
0
is a ﬁnite subset of A
_
8 Normed Spaces
for any collection (x
a
)
a∈A
of nonnegative real numbers. It is easy to show that this
agrees with the usual deﬁnition if A is countable. (We are looking at the net of ﬁnite
sums
_
a∈A
0
x
a
_
A
0
∈A
in R
+
, where A is the aggregate of all ﬁnite subsets of A, ordered
by inclusion; see Deﬁnition 1.39.)
Theorem 1.20. For p ∈ [1, ∞) and x ∈
a∈A
E
a
let
x
p
:=
_
a∈A
x
a

p
_
1/p
.
Then
(p)
a∈A
E
a
:= ¦x ∈
a∈A
E
a
: x
p
< ∞¦
is a subspace of
a∈A
E
a
and
_
(p)
a∈A
E
a
,  
p
_
is a Banach space, the pnorm direct sum
of the Banach spaces E
a
.
Proof
Let A
0
be a ﬁnite subset of A; the inequality
a∈A
0
x
a
+ y
a

p
a∈A
0
2
p
_
x
a

p
+y
a

p
_
2
p
_
x
p
p
+y
p
p
_
∀ x, y ∈
(p)
a∈A
E
a
,
which may be proved as in Example 1.11, shows that
(p)
a∈A
E
a
is a subspace of
a∈A
E
a
.
Subadditivity of  
p
on
(p)
a∈A
E
a
follows from Minkowski’s inequality on C
n
, and
this can be obtained by applying the integral version of Minkowski’s inequality in Ex
ample 1.11 to suitable step functions.
To see that we have completeness, let (x
(n)
)
n1
be a Cauchy sequence in
(p)
a∈A
E
a
.
Since x
(n)
a
−x
(m)
a
 x
(n)
−x
(m)

p
, we have that x
a
:= lim
n→∞
x
(n)
a
exists for all a ∈ A.
If A
0
⊆ A is ﬁnite then
a∈A
0
x
a

p
= lim
n→∞
a∈A
0
x
(n)
a

p
lim
n→∞
x
(n)

p
p
;
this last limit exists because
_
x
(n)

_
n1
is Cauchy:
¸
¸
x
(n)

p
−x
(m)

p
¸
¸
x
(n)
−x
(m)

p
.
This bound shows that x ∈
(p)
a∈A
E
a
; it remains to prove that x
(n)
→ x. Note ﬁrst that
if A
0
⊆ A is ﬁnite and m, n ∈ N then
a∈A
0
x
(n)
a
−x
(m)
a

p
x
(n)
−x
(m)

p
p
. Let ε > 0
and suppose that n
0
∈ N is such that x
(n)
−x
(m)

p
< ε for all m, n n
0
. Then
a∈A
0
x
a
−x
(m)
a

p
= lim
n→∞
a∈A
0
x
(n)
a
−x
(m)
a

p
ε
p
∀ m n
0
,
so x −x
(m)

p
ε for all m n
0
, and this gives the result.
Proposition 1.21. For all p ∈ [1, ∞) the algebraic direct sum
a∈A
E
a
is dense in the
pnorm direct sum
(p)
a∈A
E
a
.
1.5. Initial Topologies 9
Proof
This is a simple consequence of the fact that if x ∈
(p)
a∈A
E
a
then x
a
,= 0 for only
countably many a ∈ A: see Exercise 1.7.
Deﬁnition 1.22. The previous proposition motives the deﬁnition of
(∞)
a∈A
E
a
as the
completion of
a∈A
E
a
with respect to  
∞
; this space is the direct sum of the Banach
spaces E
a
. Clearly
(∞)
a∈A
E
a
⊆
a∈A
E
a
, but the inclusion may be strict.
Example 1.23. (Sequence Spaces) Let A = N := ¦1, 2, 3, . . .¦ and take E
a
= F for
all a ∈ A. Then the algebraic direct sum
a∈A
E
a
= c
00
:=
_
x = (x
n
)
n∈N
: ∃ N ∈ N such that x
N+1
= x
N+2
= = 0
_
,
the pnorm direct sum
(p)
a∈A
E
a
=
_
ℓ
p
if p ∈ [1, ∞),
c
0
if p = ∞,
and the direct product
a∈A
E
a
= ℓ
∞
.
In general, if A is any set and E
a
= F for all a ∈ A then we deﬁne c
00
(A), c
0
(A) and
ℓ
p
(A) in this manner.
Initial Topologies
Deﬁnition 1.24. Let X be a set and F be collection of functions on X, such that
f : X → Y
f
, where (Y
f
, S
f
) is a topological space, for all f ∈ F. The initial topology
generated by F, denoted by T
F
, is the coarsest topology such that each function f ∈ F is
continuous. (Older books call T
F
the weak topology generated by F: the adjective ‘weak’
is tremendously overworked in functional analysis so we prefer the modern term.) It is
clear that T
F
is the intersection of all topologies on X that contain
_
f∈F
f
−1
(S
f
) = ¦f
−1
(U) : f ∈ F, U ∈ S
f
¦.
In fact, every element of T
F
is the arbitrary union of sets of the form
n
i=1
f
−1
i
(U
i
) (n ∈ N, f
1
, . . . , f
n
∈ F, U
1
∈ S
f
1
, . . . , U
n
∈ S
fn
); (1.1)
these sets are a basis for this topology. (To see this, note that every set of this form
lies in T
F
, and that the collection of arbitrary unions of these sets is a topology (cf.
Exercise 1.4).)
Proposition 1.25. Let F be a collection of functions as in Deﬁnition 1.24 and let
(Z, U) be a topological space. A function g : (Z, U) → (X, T
F
) is continuous if and only
if f
◦
g : (Z, U) → (Y
f
, S
f
) is continuous for all f ∈ F.
10 Normed Spaces
Proof
As the composition of continuous functions is continuous, one implication is imme
diate. For the converse, suppose that f
◦
g is continuous for all f ∈ F and let U ∈ T
F
.
We may assume that U =
n
i=1
f
−1
i
(U
i
) for f
1
, . . . , f
n
∈ F and U
1
∈ S
f
1
, . . . , U
n
∈ S
fn
,
and then
g
−1
(U) =
n
i=1
g
−1
_
f
−1
i
(U
i
)
_
=
n
i=1
(f
i
◦
g)
−1
(U
i
) ∈ U,
as required.
The above may remind the reader of a result concerning the product topology, which is
an initial topology (that generated by the coordinate projections). We shall see other
examples of initial topologies later.
Proposition 1.26. Let F be a collection of functions as in Deﬁnition 1.24, such that
(Y
f
, S
f
) is Hausdorﬀ for all f ∈ F. The initial topology T
F
is Hausdorﬀ if F separates
points: for all x, y ∈ Y such that x ,= y there exists f ∈ F such that f(x) ,= f(y).
Proof
Let x, y ∈ X be distinct and suppose that f ∈ F is such that f(x) ,= f(y). Since S
f
is Hausdorﬀ there exist disjoint sets U, V ∈ S
f
such that f(x) ∈ U and f(y) ∈ V . Then
f
−1
(U), f
−1
(V ) ∈ T
F
are such that
x ∈ f
−1
(U), y ∈ f
−1
(V ) and f
−1
(U) ∩ f
−1
(V ) = f
−1
(U ∩ V ) = ∅.
Hence T
F
is Hausdorﬀ, as claimed.
Nets
In a metric space (X, d) it is readily proven that, given a set M ⊆ X, the element x ∈
¯
M
if and only if there exists a sequence (x
n
)
n1
⊆ M such that x
n
→ x. Hence closed sets
(those such that M =
¯
M ) may be characterised by means of sequences, and therefore
so can the topology generated by the metric d.
If X = R is equipped with the cocountable topology (which consists of the empty
set and the complement of each countable subset of R) then M = R ¸ ¦0¦ has closure
¯
M = R but there is no sequence (x
n
)
n1
⊆ M such that x
n
→ 0. This shows that the
result of the previous paragraph does not hold for general topological spaces (and that
the cocountable topology on R is nonmetrizable).
However, arguments with sequences are often very natural and easy to follow, whereas
arguments involving open sets can sometimes appear rather opaque. Is it possible to
replace the notion of sequence with some generalisation which allows a version of the
result above? The answer is, of course, yes.
Deﬁnition 1.27. Let A be a set. A preorder on A is a binary relation that satisﬁes,
for all a, b, c ∈ A,
(i) a a (reﬂexivity)
and (ii) a b and b c imply that a c (transitivity);
1.6. Nets 11
we say that A is ordered by . For convenience we write b a if and only if a b. A
directed set (A, ) is a set A and a preorder on A with the following property: for all
a, b ∈ A there exists c ∈ A such that a c and b c; the element c is called an upper
bound or majorant for a and b. A net in a set X is a directed set (A, ) and a function
x: A → X; this is denoted by (x
a
)
a∈A
(the preorder being understood).
Example 1.28. If N is equipped with the usual order then it is a directed set, and
the net (x
n
)
n∈N
is the same thing as the sequence (x
n
)
n1
.
Example 1.29. Let f be a continuous function on the interval [0, 1], and let A be the
collection of (realvalued) step functions φ on [0, 1] that are bounded above by f: a
step function φ ∈ A if φ(x) f(x) for all x ∈ [0, 1] (where is the usual order on
R). Order A by saying that φ ψ if and only if φ(x) ψ(x) for all x ∈ [0, 1]; since
φ ∨ ψ: x → max¦φ(x), ψ(x)¦ is a step function if φ and ψ are, the pair (A, ) forms a
directed set.
Deﬁnition 1.30. Let (X, T) be a topological space. A net (x
a
)
a∈A
in X is convergent
if there exists x ∈ X such that, for all U ∈ T with x ∈ X, there exists a
0
∈ A such that
x
a
∈ U for all a a
0
; the element x is the limit of this net, and we write x
a
→ x or
lim
a∈A
x
a
= x. (This latter notation is a slight abuse as, in general, limits need not be
unique.)
Proposition 1.31. A net in a Hausdorﬀ topological space has at most one limit.
Proof
This may be proved in the same manner as the corresponding result for sequences
[22, Proposition 4.2.2].
Example 1.32. A net (x
n
)
n∈N
in a topological space (X, T) converges to x if and only
if the sequence (x
n
)
n1
converges to x; the proof of this is immediate. Hence nets are
generalisations of sequences.
Example 1.33. If (A, ) is the directed set of Example 1.29 and
_
φ denotes the integral
of the step function φ then (
_
φ)
φ∈A
is a net in R that converges to
_
f. (To see this,
note ﬁrst that
_
f = sup¦
_
φ : φ ∈ A¦: see [17, ¸9.6].)
We now give some theorems that illustrate how nets can be used to answer topological
questions.
Theorem 1.34. Let (X, T) be a topological space and suppose that M ⊆ X. Then M
is closed if and only if lim
a∈A
x
a
∈ M for all convergent nets (x
a
)
a∈A
⊆ M.
Proof
Suppose M is closed and (x
a
)
a∈A
is a net in M, such that x
a
→ x for some x ∈ X. If
x / ∈ M then X ¸ M is an open set containing x, so it contains some elements of (x
a
)
a∈A
,
contrary to hypothesis. Hence lim
a∈A
x
a
∈ M.
Conversely, let x ∈
¯
M and let A = ¦U ∈ T : x ∈ U¦ denote the collection of open
sets containing x; this forms a directed set if ordered by reverse inclusion, i.e., A B
12 Normed Spaces
if and only if A ⊇ B. If A ∈ A then A ∩ M ,= ∅, for otherwise X ¸ A is a closed set
containing M and so must contain x. Hence we may choose x
A
∈ A∩ M for all A ∈ A,
and the net (x
A
)
A∈A
converges to x: note that if U ∈ T contains x then x
A
∈ U for all
A U. This gives the result.
The following proposition shows why nets and initial topologies work so well together.
Proposition 1.35. Let (X, T) and (Y, S) be topological spaces. A function f : X → Y
is continuous if and only if f(x
a
) → f(x) for every net (x
a
)
a∈A
such that x
a
→ x.
Proof
Let f be continuous, suppose that (x
a
)
a∈A
is convergent to x and let U ∈ S contain
f(x). Then V = f
−1
(U) ∈ T contains x, so there exists a
0
∈ A such that x
a
∈ V for all
a a
0
. Hence f(x
a
) ∈ U for all a a
0
, as required.
Conversely, let U ∈ S and suppose for contradiction that V = f
−1
(U) is not open.
Then there exists x ∈ V ¸ V
◦
, and no open set containing x is contained in V : for all
A ∈ A := ¦W ∈ T : x ∈ W¦ there exists x
A
∈ A ∩ (X ¸ V ). If A is ordered by reverse
inclusion then (x
A
)
A∈A
is a net converging to x, so f(x
A
) → f(x). In particular there
exists A
0
∈ A such that f(x
A
0
) ∈ U, and x
A
0
∈ V , the desired contradiction.
This proposition allows us to give a simple proof of the completeness of C
b
(X, E).
Theorem 1.36. If X is a topological space and E is a Banach space then C
b
(X, E) is
a Banach space with respect to the supremum norm  
∞
.
Proof
We prove only completeness; everything else is trivial. Let (f
n
)
n1
be a Cauchy
sequence in C
b
(X, E) and note that f
n
(x) −f
m
(x) f
n
−f
m

∞
for all x ∈ X, so
f : X → E; x → lim
n→∞
f
n
(x)
is well deﬁned. If ε > 0 then there exists n
0
∈ N such that f
n
−f
m

∞
< ε if m, n n
0
,
so
f(x) −f
m
(x) = lim
n→∞
f
n
(x) −f
m
(x) ε ∀ m n
0
, x ∈ X
whence f −f
m

∞
→ 0 as m → ∞ and also f
∞
f −f
n
0

∞
+f
n
0

∞
ε+f
n
0

∞
,
i.e., f is bounded. Finally, if x
a
→ x then choose a
0
∈ A such that f
n
0
(x
a
)−f
n
0
(x) < ε
for all a a
0
and note that
f(x
a
) − f(x) f(x
a
) − f
n
0
(x
a
) + f
n
0
(x
a
) − f
n
0
(x) + f
n
0
(x) − f(x) < 3ε
for all a a
0
. This shows that f(x
a
) → f(x), so f is continuous.
Proposition 1.37. Let F be a collection of functions as in Deﬁnition 1.24, and suppose
that (x
a
)
a∈A
is a net in X. Then x
a
→ x in (X, T
F
) if and only if f(x
a
) → f(x) in
(Y
f
, S
f
) for all f ∈ F.
Proof
As T
F
makes each f ∈ F continuous, half of the proof follows from Proposition 1.35.
For the other half, suppose that (x
a
)
a∈A
⊆ X and x ∈ X are such that f(x
a
) → f(x)
1.6. Nets 13
for all f ∈ F. Let U ∈ T
F
be such that x ∈ U; it is suﬃcient to consider U of the form
(1.1) in Deﬁnition 1.24. Note that x ∈ f
−1
i
(U
i
) for i = 1, . . . , n, so f
i
(x) ∈ U
i
and there
exists a
i
∈ A such that f
i
(x
a
) ∈ U
i
for all a a
i
. If a
0
∈ A is such that a
i
a
0
for each
i then x
a
∈
n
i=1
f
−1
i
(U
i
) = U for all a a
0
, as required.
Example 1.38. Let Ω ⊆ C be open and let
ι
K
: C(Ω) → C(K); f → f[
K
be the restriction map to K, where K ⊆ Ω. If
F = ¦ι
K
: K is a compact subset of Ω¦
and each C(K) has the supremum norm then T
F
is the topology of locally uniform
convergence:
f
n
→ f ⇐⇒ f
n
[
K
→ f[
K
uniformly on K for all compact K ⊆ Ω.
Nets allow us to give a proper treatment of summability, which coincides with the ad
hoc method used in Deﬁnition 1.19.
Deﬁnition 1.39. Let X be a normed vector space. A family (x
a
)
a∈A
⊆ X is summable
if the net of partial sums
_
a∈A
0
x
a
_
A
0
∈A
is convergent, where A is the collection of
ﬁnite subsets of A, ordered by inclusion. If (x
a
)
a∈A
is summable then
a∈A
x
a
denotes
the limit of the net of partial sums.
Example 1.40. If (x
a
)
a∈A
is a family of nonnegative real numbers then this deﬁnition
agrees with Deﬁnition 1.19 (Exercise 1.6). If (z
n
)
n∈N
is a family of complex numbers
then it is summable if and only if
∞
n=1
z
n
is absolutely convergent (see Exercises 1.7
and 1.8). Note also that if (x
a
)
a∈A
is a family of vectors in a Banach space then
a∈A
x
a
 < ∞ =⇒
a∈A
x
a
is convergent,
i.e., absolute summability implies summability (Exercise 1.7).
Nets were introduced by Moore and Smith in [13] – the theory is also called MooreSmith
convergence, especially in older references – and they were applied to general topological
spaces by Garrett Birkhoﬀ [2]. As to the choice of nomenclature, the reader may care
to reﬂect upon the following, taken from [12, Third footnote on p.3]:
[J. L.] Kelley writes me that [“net”] was suggested by Norman Steenrod
in a conversation between Kelley, Steenrod and Paul Halmos. Kelley’s own
inclination was to the name “way”; the analogue of a subsequence would then
be a “subway”!
An aged but excellent introduction to nets is [9, Chapter 2]; McShane’s article [12] is a
very pleasant introductory exposition. Pedersen [14] refers to the viewpoints of topology
in terms of nets and of open sets as dynamic and static, respectively.
14 Normed Spaces
Exercises 1
Exercise 1.1. Let X be a normed vector space and let M be a closed subspace of X.
Prove that
π¦y ∈ X : y −x < ε¦ =
_
[y] ∈ X/M :
_
_
[y] −[x]
_
_
< ε
_
∀ x ∈ X, ε > 0,
where π: X → X/M; x → [x] is the natural map from X onto X/M (the quotient map).
Deduce that the quotient norm yields the quotient topology on X/M given by
Q := ¦U ⊆ X/M : π
−1
(U) ∈ T¦,
where T denotes the norm topology on X, and that the quotient map is open (i.e., sends
open sets to open sets). Prove also that the quotient map is linear and continuous.
Exercise 1.2. Prove directly that if E is a Banach space and M is a closed subspace of
E then the quotient space
_
E/M,  
E/M
_
is complete. [Use Banach’s criterion.]
Exercise 1.3. Let M and N be subspaces of the normed space X. Prove that if M is
ﬁnite dimensional and N is closed then M +N is closed. [Recall that ﬁnitedimensional
subspaces of normed spaces are closed [24, Corollary 1.2.18] and use the quotient map.]
Exercise 1.4. Prove that if ¦A
i
j
: i ∈ I, j ∈ J¦ and ¦B
k
l
: k ∈ K, l ∈ L¦ are families of
sets, where the index sets I, J, K and L are arbitrary, then
_
_
i∈I
j∈J
A
i
j
_
∩
_
_
k∈K
l∈L
B
k
l
_
=
_
(i,k)∈I×K
(j,l)∈J×L
A
i
j
∩ B
k
l
.
What does this have to do with initial topologies?
Exercise 1.5. Prove that if T
F
is the initial topology on X generated by a collection
of functions F and Y ⊆ X then T
F
[
Y
, the relative initial topology on Y , is the initial
topology generated by F[
Y
= ¦f[
Y
: f ∈ F¦, the restrictions of the functions in F to Y .
Exercise 1.6. Let (x
a
)
a∈A
be a family of nonnegative real numbers and let A denote
the collection of ﬁnite subsets of A. Prove that (x
a
)
a∈A
is summable (with sum α) if and
only if β = sup
_
a∈A
0
x
a
: A
0
∈ A
_
< ∞ and in this case α = β.
Exercise 1.7. Let E be a Banach space and let (x
a
)
a∈A
a family of vectors in E. Prove
that if
a∈A
x
a
 is convergent then S := ¦a ∈ A : x
a
,= 0¦ is countable. [Consider the
sets S
n
:= ¦a ∈ A : x
a
 > 1/n¦ for n ∈ N.] Deduce that (x
a
)
a∈A
is summable with sum
a∈A
x
a
=
_
a∈S
x
a
if S is ﬁnite,
∞
j=1
x
a
j
if S is inﬁnite,
where (if S is inﬁnite) j → a
j
is a bijection between N and S.
Exercise 1.8. Prove that a family of complex numbers (z
a
)
a∈A
is summable if and only
if
_
[z
a
[
_
a∈A
is summable. [Consider real and imaginary parts to reduce to the real case
and then consider positive and negative parts.]
1.7. Exercises 1 15
Exercise 1.9. Find a Hilbert space H and a countable family of vectors (x
n
)
n∈N
in H
that is summable but not absolutely summable (i.e.,
_
x
n

_
n∈N
is not summable).
Exercise 1.10. Prove the converse to Proposition 1.31, that in a space with a non
Hausdorﬀ topology there exists a net that converges to two distinct points. [Take two
points that cannot be separated by open sets and deﬁne a net that converges to both of
them.]
Exercise 1.11. A sequence in a normed vector space that is convergent is necessarily
bounded. Is the same true for nets?
Two Linear Operators
Preliminaries
Let X be a normed vector space; for all r ∈ R
+
let X
r
denote the closed ball in X with
radius r and centre the origin:
X
r
:=
_
x ∈ X : x r
_
.
Deﬁnition 2.1. A bounded (linear) operator from X to Y is a linear transformation
T : X → Y such that the operator norm T is ﬁnite, where
T := inf¦M ∈ R
+
: Tx Mx for all x ∈ X¦
= sup¦Tx : x ∈ X
1
¦
= sup
_
Tx : x ∈ X, x = 1
_
.
Proposition 2.2. Let T : X → Y be a linear transformation. The following statements
are equivalent:
(i) T is a bounded linear operator;
(ii) T is uniformly continuous;
(iii) T is continuous;
(iv) T is continuous at 0;
(v) T(X
1
) is bounded: T(X
1
) ⊆ Y
r
for some r ∈ R
+
.
Proof
The implications (i) ⇒ (ii) ⇒ (iii) ⇒ (iv) ⇒ (v) ⇒ (i) are immediate.
We denote the collection of bounded linear operators from X to Y by B(X, Y ) (or B(X)
if X = Y ). Note that B(X, Y ) is a normed vector space, where
(T + S)x := Tx + Sx and (αT)x := αTx ∀ S, T ∈ B(X, Y ), x ∈ X, α ∈ F
and the norm on B(X, Y ) is the operator norm.
Theorem 2.3. If T : X → Y be a linear transformation then
(i) the kernel ker T := ¦x ∈ X : Tx = 0¦ is a subspace of X;
17
18 Linear Operators
(ii) the image imT := ¦Tx : x ∈ X¦ is a subspace of Y ;
(iii) X/ ker T
∼
= imT via the linear transformation
˜
T : X/ ker T → imT; [x] → Tx.
If X and Y are normed spaces and T ∈ B(X, Y ) then ker T is closed, X/ ker T is a
normed space and
˜
T is continuous, with
_
_ ˜
T
_
_
= T.
Proof
The algebraic facts are immediately veriﬁed, as is the fact that ker T = T
−1
¦0¦ is
closed if T is continuous. If
π: X → X/ ker T; x → [x]
is the quotient map, the inequality π(x) =
_
_
[x]
_
_
x implies that π 1 and so
T
_
_ ˜
T
_
_
π
_
_ ˜
T
_
_
(because
˜
T
◦
π = T). Conversely, let x ∈ X, ε > 0 and choose
y ∈ ker T such that x −y <
_
_
[x]
_
_
+ ε. Then
_
_ ˜
T[x]
_
_
= Tx = T(x −y) T x −y T
__
_
[x]
_
_
+ ε
_
,
and since this holds for all ε > 0 and x ∈ X we have the result.
Completeness of B(X, Y )
Proposition 2.4. If X ,= ¦0¦ then the normed vector space B(X, Y ) is a Banach space
if and only if Y is a Banach space.
Proof
The fact that the completeness of Y entails the completeness of B(X, Y ) is a result
from b4 [24, Exercise 1.5.17(ii)]; a leisurely proof may be found in [10, Theorem 2.10
2] and a more concise version in [14, Theorem 2.1.4]. We prove the converse as an
application of the HahnBanach theorem: see Proposition 3.12.
Extension of Linear Operators
Often it is easiest to deﬁne a linear operator on some dense subspace of a Banach space,
and extend it to the whole space “by continuity”: the following theorem explains the
meaning of this.
Theorem 2.5. (BLT) Let X
0
be a dense subspace of a normed vector space X and
let T
0
∈ B(X
0
, Y ), where Y is a Banach space. There exists a unique T ∈ B(X, Y ) such
that T[
X
0
= T
0
, and such satisﬁes T = T
0
.
Proof
Existence is a b4 result [24, Exercise 1.5.17(iii)]; for a proof see [10, Theorem 2.711] or
[14, Theorem 2.1.11]. For uniqueness, note that if S[
X
0
= T
0
= T[
X
0
for S, T ∈ B(X, Y )
then (S −T)[
X
0
= 0 and so S −T = 0 by continuity: if x ∈ X let (x
n
)
n1
⊆ X
0
be such
that x
n
→ x and note that (S −T)x = lim
n→∞
(S −T)x
n
= 0.
2.4. The Baire Category Theorem 19
Uniqueness of Completions
We are now in the position to prove that the completion of a normed space is unique.
Proposition 2.6. Let (Y, i) and (Z, j) be completions of the normed space X. There
exists an isometric isomorphism k: Y → Z such that the following diagram commutes.
X
i
j
Y
k
Z
Proof
Note that
k
0
: i(X) → Z; i(x) → j(x)
is a welldeﬁned linear isometry from a dense subspace of Y onto a dense subspace of
Z, viz j(X). Hence, by Theorem 2.5, k
0
extends to k ∈ B(Y, Z) and k is an isometry:
if y ∈ Y then there exists (x
n
)
n1
⊆ X such that i(x
n
) → y, but k and x → x are
continuous, so
k(y) = lim
n→∞
_
_
k
_
i(x
n
)
__
_
= lim
n→∞
_
_
k
0
_
i(x
n
)
__
_
= lim
n→∞
i(x
n
) = y.
Furthermore, k(Y ) is closed in Z: to see this, let
_
k(y
n
)
_
n1
be a convergent sequence
(with y
n
∈ Y ) and note that it is Cauchy, so (y
n
)
n1
is Cauchy (k being an isometry) and
hence convergent, to y ∈ Y ; the continuity of k gives k(y
n
) → k(y) ∈ k(Y ), as required.
As k(Y ) contains j(X), it contains its closure, Z, and therefore k is surjective.
The Baire Category Theorem
Deﬁnition 2.7. A subset of a metric space is said to be meagre or of the ﬁrst category
if it is a countable union of closed sets with empty interior; a set is nonmeagre or of the
second category otherwise. (Recall that a set is nowhere dense if its closure has empty
interior.)
We introduce some notation (which we learned from Dr J. M. Lindsay) that will be used
in the proof of the next theorem. If (X, d) is a metric space then let
B
X
r
(x) := ¦y ∈ X : d(x, y) < r¦ (the open ball of radius r and centre x)
and
B
X
r
[x] := ¦y ∈ X : d(x, y) r¦ (the closed ball of radius r and centre x)
for all x ∈ X and r ∈ (0, ∞); if the space is clear from the context then we abbreviate
these to B
ε
(x) and B
ε
[x] respectively. If X is a normed space then
B
X
r
(x) = x + rB
X
1
(0) and B
X
r
[x] = x + X
r
= x + rX
1
;
20 Linear Operators
furthermore, B
X
r
(x) = B
X
r
[x].
Recall that G is said to be dense in the topological space (X, T) if and only if
¯
G = X.
This is equivalent to the condition that G ∩ U ,= ∅ for all U ∈ T ¸ ¦∅¦ (because
¯
G = X
if and only if (X ¸ G)
◦
= X ¸
¯
G = ∅). If T is given by a metric d then this condition is
equivalent to requiring that G∩ B
ε
(x) ,= ∅ for all x ∈ X and ε > 0.
Theorem 2.8. (Baire) If (G
n
)
n1
is a sequence of open, dense subsets of the complete
metric space (X, d) then the intersection
n1
G
n
is dense in X.
Proof
Let x
0
∈ X and r
0
> 0; it suﬃces to prove that B
r
0
(x
0
) ∩
n1
G
n
,= ∅. To do this
we construct sequences (r
n
)
n1
⊆ (0, ∞) and (x
n
)
n1
⊆ X such that
r
0
> r
1
> r
2
> → 0 (2.1)
as n → ∞ and
B
rn
[x
n
] ⊆ B
r
n−1
(x
n−1
) ∩ G
n
∀ n 1. (2.2)
Given (2.2), if n > m 1 then B
rn
[x
n
] ⊆ B
r
n−1
(x
n−1
) ⊆ B
r
n−1
[x
n−1
] ⊆ ⊆ B
rm
(x
m
);
in particular,
d(x
n
, x
m
) < max¦r
n
, r
m
¦ ∀ m, n 1,
which, if (2.1) holds, shows that (x
n
)
n1
is Cauchy, and so convergent. As x
n
∈ B
rm
[x
m
]
for all n m 1,
lim
n→∞
x
n
∈
n1
B
rn
[x
n
] ⊆ B
r
0
(x
0
) ∩
n1
G
n
and we are done. (This last step shows why we must consider both closed and open
balls.)
To see why we may ﬁnd sequences satisfying (2.1) and (2.2), consider ﬁrst the case
n = 1. Note that B
r
0
(x
0
) ∩ G
1
is open (being the intersection of two open sets) and
nonempty (as G
1
is dense in X). Hence there exists x
1
∈ X and s ∈ (0, r
0
) such that
B
s
(x
1
) ⊆ B
r
0
(x
0
) ∩ G
1
; taking r
1
∈
_
0, min¦s, 1¦
_
shows that (2.2) holds in the case
n = 1. This argument may be repeated for all n (each time ensuring that r
n
< 1/n)
which completes the proof.
An alternative formulation of Baire’s category theorem as follows, which states that a
complete metric space is nonmeagre.
Theorem 2.9. Let (F
n
)
n1
be a sequence of closed subsets of the complete metric
space (X, d). If X =
n1
F
n
then there exists n
0
∈ N such that F
n
0
has nonempty
interior.
Proof
Suppose for contradiction that F
◦
n
= ∅ for all n 1, and let G
n
= X ¸ F
n
. Then
¯
G
n
= X ¸ F
n
= X ¸ F
◦
n
= X
and so G
n
is an open, dense subset of X, whence
n1
G
n
is dense, by Theorem 2.8, but
n1
G
n
=
n1
X ¸ F
n
= X ¸
n1
F
n
= ∅.
2.5. The OpenMapping Theorem 21
The OpenMapping Theorem
Recall that a function between topological spaces is open if the image of every open set
is open.
Proposition 2.10. Let X and Y be normed spaces; a linear transformation T : X → Y
is open if and only if T(X
1
) contains Y
ε
for some ε > 0.
Proof
If T is open then T
_
B
X
1
(0)
_
is an open subset of Y that contains T0 = 0, and so
contains B
Y
2ε
(0) for some ε > 0. Hence Y
ε
⊆ B
Y
2ε
(0) ⊆ T
_
B
X
1
(0)
_
⊆ T(X
1
).
Conversely, suppose there exists ε > 0 such that Y
ε
⊆ T(X
1
). Let U be an open
subset of X; if u ∈ U then B
X
2δ
(u) ⊆ U for some δ > 0. Hence
T(U) ⊇ T
_
u + δB
X
2
(0)
_
⊇ Tu + δT(X
1
) ⊇ Tu + δY
ε
⊇ B
Y
δε
(Tu),
which shows that T(U) is open, as required.
The OpenMapping Lemma
Deﬁnition 2.11. Let (X, d) be a metric space and suppose that A, B ⊆ X; we say that
A is kdense in B if for all b ∈ B there exists a ∈ A such that d(b, a) k. Equivalently,
B ⊆
a∈A
B
k
[a]. (If A is dense in B then A is kdense in B for all k > 0; if A is dense
in B and B is kdense in C then A is k
′
dense in C for any k
′
> k.)
Lemma 2.12. (OpenMapping Lemma) Let E be a Banach space, Y a normed
space and T ∈ B(E, Y ). If there exist r > 0 and k ∈ (0, 1) such that T(E
r
) is kdense
in Y
1
then
(i) for all y ∈ Y there exists x ∈ E such that x
r
1 −k
y and Tx = y, so T is
surjective,
(ii) T is an open mapping
and (iii) Y is complete.
Proof
To prove (i) let y ∈ Y ; without loss of generality we may take y ∈ Y
1
(the case y = 0
is trivial and otherwise we may replace y by y/y ) so there exists x
0
∈ E
r
such that
y −Tx
0
 k.
As k
−1
(y −Tx
0
) ∈ Y
1
there exists x
1
∈ E
r
such that
k
−1
(y −Tx
0
) −Tx
1
 k ⇐⇒ y −T(x
0
+ kx
1
) k
2
.
Continuing in this way we see that there exists a sequence (x
l
)
l0
⊆ E
r
such that
_
_
y −T
n
l=0
k
l
x
l
_
_
k
n+1
∀ n 0. (2.3)
22 Linear Operators
Let x :=
∞
l=0
k
l
x
l
; note that this series is absolutely convergent, so convergent, and
(2.3) shows that Tx = y. The inequality x r/(1 −k) follows from the deﬁnition of
x and the fact that the series
∞
l=0
k
l
has sum 1/(1 −k).
For (ii), note that (i) implies that T(X
1
) ⊇ Y
(1−k)/r
: if y ∈ Y satisﬁes y (1−k)/r
then there exists x ∈ X such that Tx = y and x ry/(1 − k) 1. Hence T is
open, by Proposition 2.10.
Finally, let
˜
Y denote the completion of Y and regard Y as a dense subspace of
˜
Y ,
so that T is a bounded operator from E into
˜
Y . The hypotheses of the theorem hold
with Y replaced by
˜
Y , i.e., T(E
r
) is k
′
dense in
˜
Y
1
(where k
′
∈ (k, 1), say k
′
= (k +1)/2:
to prove this, use the density of Y
1
in
˜
Y
1
) and so by (i) we must have that T(E) =
˜
Y .
Since T(E) ⊆ Y this gives the equality Y =
˜
Y , showing that Y is complete.
Theorem 2.13. (OpenMapping Theorem) If T ∈ B(E, F) is surjective, where E
and F are Banach spaces, then T is an open map.
Proof
Since T is surjective, F =
n1
T(E
n
) and so, by Theorem 2.9, there exists n
0
∈ N
such that the closure of T(E
n
0
) has nonempty interior: let y
0
∈ F and ε > 0 be such
that B
F
ε
(y
0
) ⊆ T(E
n
0
). If y ∈ F
1
then
y =
1
2ε
_
(y
0
+εy)−(y
0
−εy)
_
∈
1
2ε
_
B
F
ε
[y
0
]−B
F
ε
[y
0
]
_
⊆
1
2ε
_
T(E
n
0
)−T(E
n
0
)
_
⊆ T(E
n
0
/ε
),
where A − B := ¦a − b : a ∈ A, b ∈ B¦ for all A, B ⊆ E. (For the last inclusion, let
(x
n
)
n1
, (y
n
)
n1
⊆ E
n
0
be such that (Tx
n
)
n1
and (Ty
n
)
n1
are convergent, and note
that
1
2ε
_
lim
n→∞
Tx
n
− lim
n→∞
Ty
n
_
= lim
n→∞
T
_
1
2ε
(x
n
−y
n
)
_
∈ T(E
n
0
/ε
)
since
_
_
(x
n
− y
n
)/(2ε) (n
0
+ n
0
)/(2ε) = n
0
/ε.) This shows that T(E
n
0
/ε
) is dense in
F
1
, and the result follows by Lemma 2.12.
Corollary 2.14. If T ∈ B(E, F) is a bijection between Banach spaces E and F then
the inverse T
−1
∈ B(F, E).
Proof
This is immediate.
Exercises 2
Exercise 2.1. Let X be a topological space and E a Banach space; recall that C
b
(X, E),
the space of Evalued, bounded, continuous functions on X, is complete with respect to
the norm
f → f
∞
:= sup¦f(x)
E
: x ∈ X¦.
Prove that C
0
(X, E), the continuous, Evalued functions on X that vanish at inﬁnity
(i.e., those f ∈ C(X, E) such that ¦x ∈ X : f(x)
E
ε¦ is compact for all ε > 0) is a
closed subspace of C
b
(X, E).
2.6. Exercises 2 23
Exercise 2.2. Let (X, T) be a Hausdorﬀ, locally compact space and let ∞ denote a
point not in X. Show that
˙
T := T ∪ ¦U ⊆
˙
X : ∞ ∈ U, X ¸ U is compact¦
is a Hausdorﬀ, compact topology on
˙
X := X ∪¦∞¦. [(
˙
X,
˙
T) is the onepoint compacti
ﬁcation of (X, T).] Prove that there is a natural correspondence between C
0
(X, E) and
¦f ∈ C(
˙
X, E) : f(∞) = 0¦.
Exercise 2.3. Let X be a separable normed space. Prove that X
1
is separable (in the
norm topology). Prove that any separable Banach space E is isometrically isomorphic
to a quotient space of ℓ
1
. [Consider the map x →
∞
n=1
x
n
e
n
for suitable (e
n
)
n1
⊆ E
1
.]
Exercise 2.4. Prove that no inﬁnitedimensional Banach space E has a countable Hamel
basis (where a Hamel basis is a linearly independent set S such that every vector in E
is a ﬁnite linear combination of elements of S).
Exercise 2.5. Let T : X → Y be a linear transformation from the normed space X
onto the ﬁnitedimensional normed space Y . Prove that T is continuous if and only if
ker T is closed and that if T is continuous then T is open. [Recall that all norms on a
ﬁnitedimensional space are equivalent.]
Exercise 2.6. Let X = C([0, 1], R) denote the Banach space of continuous, realvalued
functions on the unit interval and for all k ∈ N let
D
k
:= ¦f ∈ X : there exists t ∈ [0, 1] such that [f(s)−f(t)[ k[s−t[ for all s ∈ [0, 1]¦.
Prove that D
k
is closed. [You may ﬁnd the BolzanoWeierstrass theorem useful.] Prove
further that D
k
is nowhere dense. [Consider suitable piecewiselinear functions.] Deduce
that there exist continuous functions on [0, 1] that are diﬀerentiable at no point in (0, 1).
Exercise 2.7. Let H be an inﬁnitedimensional, separable Hilbert space. Prove that
B(H)
1
is not separable in the norm topology. [Take an orthonormal basis ¦e
1
, e
2
, . . .¦ of
H, deﬁne suitable projection operators on its linear span and extend these by continuity.
Recall the b4 proof that ℓ
∞
is closed; you may assume that 2
N
, the set of all subsets of
N, is uncountable.]
Completeness of Quotient Spaces
The openmapping theorem provides a quick proof of the following proposition.
Proposition 2.15. Let M be a closed subspace of the Banach space E. The quotient
space
_
E/M,  
E/M
_
is complete, i.e., a Banach space.
Proof
If [x] ∈ (E/M)
1
then, by deﬁnition of the quotient norm, there exists m ∈ M such
that x − m ∈ E
2
(any number greater than 1 will do) and so (E/M)
1
⊆ π(E
2
), where
π: x → [x] is the quotient map. Since π is a bounded linear operator, the result follows
by Lemma 2.12.
24 Linear Operators
Urysohn’s Lemma
To prove certain facts about C(X), the continuous functions on a compact, Hausdorﬀ
space (which is the most important of all commutative Banach algebras) we need a result
from analytic topology.
Deﬁnition 2.16. A topological space X is normal if every pair of disjoint, closed sets
can be separated by open sets: if C, D ⊆ X are closed and disjoint there there exist
disjoint, open sets U, V ⊆ X such that C ⊆ U and D ⊆ V . Equivalently, X is normal
if for every open set W and closed set C such that C ⊆ W ⊆ X there exists an open set
U such that C ⊆ U ⊆
¯
U ⊆ W. [To see the equivalence, let D = X ¸ W.]
Some authors require the additional condition that all singleton sets to be closed for
a topology to be normal; we follow [3], [9] and [22], but the deﬁnition in [21] includes this
and [14] insists on the seemingly stronger requirement that normal spaces be Hausdorﬀ;
in fact, in a normal space the conditions that singletons are closed and the Hausdorﬀ
property are equivalent.
It is an easy exercise [22, Exercise 5.10.17] to prove that a compact, Hausdorﬀ space
is normal (for a proof see [3, Lemma 6.1], [14, Theorem 1.6.6] or [21, Theorem 27.A]).
The following lemma yields the fact that compact, Hausdorﬀ spaces (indeed, Hausdorﬀ
spaces that are normal) have suﬃcient continuous functions to separate points.
Lemma 2.17. (Urysohn) Let X be a normal space and let C, D ⊆ X be disjoint and
closed. There exists a continuous function f : X → [0, 1] such that f[
C
= 0 and f[
D
= 1.
Proof
Let U
1
= X ¸ D; by normality there exists an open set U
1/2
such that C ⊆ U
1/2
and
¯
U
1/2
⊆ U
1
, and then open sets U
1/4
and U
3/4
such that
C ⊆ U
1/4
⊆
¯
U
1/4
⊆ U
1/2
and
¯
U
1/2
⊆ U
3/4
⊆
¯
U
3/4
⊆ U
1
.
Continuing in this manner we ﬁnd a family of open sets ¦U
m2
−n : 1 m 2
n
, n 1¦
such that C ⊆ U
r
⊆
¯
U
r
⊆ U
s
if r < s. (Throughout this proof the letters r, s and t
refer to dyadic rationals in (0, 1], i.e., numbers of the form m2
−n
, where n, m ∈ N and
1 m 2
n
; these are dense in [0, 1].) We set
f : X → [0, 1]; x →
_
1 if x ∈ D,
inf¦r : x ∈ U
r
¦ if x / ∈ D.
It is immediate that f[
D
= 1, and f[
C
= 0 since C ⊆ U
r
for all r; it remains to prove
that f is continuous.
If a ∈ (0, 1] then f(x) < a if and only if inf¦r : x ∈ U
r
¦ < a, which holds exactly
when x ∈ U
r
for some r < a, and hence f
−1
[0, a) =
r<a
U
r
is open. If b ∈ [0, 1) then
inf¦r : x ∈ U
r
¦ b if and only if for all r > b there exists s < r such that x ∈ U
s
, which
gives that
f
−1
[0, b] =
r>b
_
s<r
U
s
⊆
t>b
¯
U
t
.
2.6. Exercises 2 25
This inclusion is actually an equality; let x ∈
¯
U
t
for all t > b and suppose that r > b.
There exist s, t such that r > s > t > b and so x ∈
¯
U
t
⊆ U
s
, whence x ∈
r>b
s<r
U
s
,
as claimed. This shows that f
−1
[0, b] is closed, and so f
−1
(b, 1] is open, for all b ∈ [0, 1).
As ¦(0, a], (b, 1] : a ∈ (0, 1], b ∈ [0, 1)¦ is a subbase for the usual topology on [0, 1] we
have the result.
Proposition 2.18. Let X be a locally compact, Hausdorﬀ space and let E be a Banach
space. The space C
00
(X, E) of compactly supported, continuous, Evalued functions on
X is dense in C
0
(X, E), the space of continuous, Evalued functions on X that vanish
at inﬁnity.
Proof
Let ε > 0 and f ∈ C
0
(X, E); by deﬁnition, K = ¦x ∈ X : f(x) ε¦ is compact.
For all x ∈ K let U
x
be an open set containing x and with compact closure; these
exist by local compactness. As K is compact, K ⊆
n
i=1
U
x
i
for x
1
, . . . , x
n
∈ K, and
L =
n
i=1
¯
U
x
i
is a compact set such that K ⊆ L
◦
.
By Urysohn’s lemma there exists a continuous function g : L → [0, 1] such that
g[
K
= 1 and g[
L\L
◦ = 0; extend g to X by deﬁning
h: X → [0, 1]; x →
_
g(x) if x ∈ L,
0 if x ∈ X ¸ L.
Then h has compact support and is continuous: if C ⊆ [0, 1] is closed then g
−1
(C) is
closed in L, so in X, and h
−1
(C) equals g
−1
(C) (if 0 / ∈ C) or g
−1
(C)∪(X¸L
◦
) (if 0 ∈ C).
Hence fh ∈ C
00
(X, E), and fh−f
∞
< ε, as required: if x ∈ K then f(x)h(x) = f(x),
and if x ∈ X ¸ K then f(x) < ε, so f(x)h(x) −f(x) = (1 −h(x))f(x) < ε.
The following theorem can be deduced from Urysohn’s lemma directly, or with an ap
plication of the openmapping lemma. It is a theorem of HahnBanach type, but applies
to continuous functions on normal spaces.
Theorem 2.19. (Tietze) Let X be a normal space and let Y be a closed subset of X.
If f is a continuous, bounded, realvalued function on Y then there exists a continuous,
bounded, realvalued function F on X such that F[
Y
= f and F
∞
= f
∞
.
Proof
Let
T : C
b
(X, R) → C
b
(Y, R); f → f[
Y
be the restriction map and note that T is continuous. Let f ∈ C
b
(Y, R) be such that
f
∞
1, and let C = f
−1
[−1, −1/3] and D = f
−1
[1/3, 1]. These are closed subsets of
Y , so of X, and by Urysohn’s lemma there exists g ∈ C
b
(X, R) such that g
∞
1/3,
g[
C
= −1/3 and g[
D
= 1/3. Hence Tg − f
∞
2/3, and so T satisﬁes the conditions
of the openmapping lemma: T(C
b
(X, R)
1/3
) is 2/3dense in C
b
(Y, R)
1
. In particular, T
is surjective, so there exists F ∈ C
b
(X, R) such that T(F) = f and
f
∞
F
∞
1/3
1 −2/3
f
∞
= f
∞
,
26 Linear Operators
as required.
It is an exercise to extend the above to unbounded realvalued functions, and to complex
valued functions; see Exercise 3.3.
The ClosedGraph Theorem
Recall that if (X, T) and (Y, S) are topological spaces then the product topology on XY
has basis
¦U V : U ∈ T, V ∈ S¦.
If X and Y are normed spaces then this topology is given by the product norm,
 
X×Y
: X Y →R
+
; (x, y) → x
X
+y
Y
.
Theorem 2.20. (ClosedGraph Theorem) Let E, F be Banach spaces. A linear
transformation T : E → F is bounded if and only if the graph of T,
G(T) := ¦(x, Tx) : x ∈ E¦ ⊆ E F,
is closed (with respect to the product topology on E F ).
Proof
It is a standard (and simple) result from pointset topology that any continuous
function with values in a Hausdorﬀ space has closed graph; this is often set as an exercise
[21, Exercise 26.6], [22, Exercise 4.3.3] and a proof may be found in [19, Proposition 2.14].
Now suppose that G(T) is closed and note that G(T) is a subspace of E F, so a
Banach space with respect to the product norm. Let
π
1
: G(T) → E; (x, Tx) → x;
this linear transformation is normdecreasing (so continuous) and bijective, so by the
openmapping theorem π
−1
1
is bounded. Furthermore
π
2
: G(T) → F; (x, Tx) → Tx
is continuous, hence T = π
2
◦
π
−1
1
is bounded, as required.
The closedgraph theorem is often used in the following manner. A priori, to show that a
linear transformation between Banach spaces is continuous we must show that if x
n
→ x
then Tx
n
→ Tx, for any sequence (x
n
)
n1
. The closedgraph theorem means that we
need only show that the graph of T contains its limit points, i.e., if (x
n
)
n1
is such that
x
n
→ x and Tx
n
→ y then Tx = y. In this case we have control over (x
n
)
n1
and also
over (Tx
n
)
n1
, which is a considerable improvement. An application of this idea occurs
in the solution to Exercise 4.8.
2.8. The Principle of Uniform Boundedness 27
The Principle of Uniform Boundedness
We employ the closedgraph theorem to give a proof of the principle of uniform bound
edness, also known as the BanachSteinhaus theorem.
Theorem 2.21. (BanachSteinhaus) Let E be a Banach space, Y a normed space
and suppose that ¦T
a
: a ∈ A¦ ⊆ B(E, Y ). If ¦T
a
x : a ∈ A¦ ⊆ Y is bounded, for all
x ∈ E, then ¦T
a
 : a ∈ A¦ is bounded.
Proof
Note ﬁrst that we may replace Y by its completion, so without loss of generality we
assume that Y is a Banach space. Let Y
a
:= Y for all a ∈ A and let Z :=
a∈A
Y
a
be
their direct product. Deﬁne
T : E → Z; x → (T
a
x)
a∈A
and note that the pointwise boundedness of the T
a
ensures that T is well deﬁned. For
all b ∈ A let
π
b
: Z → Y
b
; (y
a
)
a∈A
→ y
b
and observe that this map is linear and normdecreasing. Let (x
n
, Tx
n
)
n1
⊆ G(T) be
convergent, say x
n
→ x and Tx
n
→ y. For all a ∈ A we have that
π
a
y = lim
n→∞
π
a
Tx
n
= lim
n→∞
T
a
x
n
= T
a
x = π
a
Tx,
by the continuity of π
a
and T
a
, which shows that y = Tx. Hence T has closed graph, so
is bounded, and
T
a
x = π
a
Tx Tx T x ∀ x ∈ E,
which shows that T
a
 T for all a ∈ A.
The Strong Operator Topology
Let H be a Hilbert space with orthonormal basis ¦e
1
, e
2
, . . .¦ and deﬁne orthogonal
projections P
n
∈ B(H) by setting
P
n
: H → H; x →
n
k=1
¸e
k
, x)e
k
.
It is easy to see that P
n
x → x as n → ∞ for all x ∈ H, by the Parseval equality, but as
P
m
−P
n
 1 for all m ,= m it cannot be the case that P
n
−I → 0 as n → ∞. This
example highlights the utility of a weaker sense of convergence for operators.
Deﬁnition 2.22. Let X and Y be normed spaces; the initial topology on B(X, Y )
generated by the family of maps ¦T → Tx : x ∈ X¦ (where Y is equipped with its norm
topology) is called the strong operator topology.
28 Linear Operators
Since
Tx T x ∀ x ∈ X, T ∈ B(X, Y ),
we see that norm convergence implies strong operator convergence: a net (T
a
)
a∈A
in
B(X, Y ) is convergent to T in the norm topology if and only if for all ε > 0 there
exists a
0
∈ A such that T
a
− T < ε for all a a
0
, and similarly for strong operator
convergence. Hence sets that are strong operator closed are also norm closed, and so the
strong operator topology is coarser that the norm topology on B(X, Y ).
Exercises 3
Exercise 3.1. Let H be a separable Hilbert space with orthonormal basis ¦e
1
, e
2
, . . .¦.
For n 1 let P
n
denote the orthogonal projection onto Fe
1
+ + Fe
n
; prove that
P
n
TP
n
x → Tx as n → ∞ for all T ∈ B(H) and x ∈ H. Deduce that B(H) is separable
in the strong operator topology.
Exercise 3.2. Prove that if E is a Banach space with respect to two diﬀerent norms then
they are either equivalent or noncomparable (i.e., neither is coarser than the other).
Exercise 3.3. Prove the following extension of Tietze’s theorem to complexvalued
functions: if X is a normal space, Y a closed subset of X and f ∈ C
b
(Y ) then there
exists F ∈ C
b
(X) such that F[
Y
= f and F
∞
= f
∞
. Prove also that Tietze’s
theorem applies to unbounded, realvalued functions: if X and Y are as above and
f : Y →R is continuous then there exists F : X →R such that F[
Y
= f.
Exercise 3.4. Let E be a Banach space, Y a normed vector space and suppose that
(T
n
)
n1
⊆ B(E, Y ) is such that lim
n→∞
T
n
x exists for all x ∈ E. Prove that there exists
T ∈ B(E, Y ) such that T
n
→ T in the strong operator topology. What can be said
about the norm of T?
Exercise 3.5. Let x = (x
n
)
n1
be a sequence of complex numbers such that the series
∞
n=1
x
n
y
n
is convergent for all y ∈ c
0
. Prove that x ∈ ℓ
1
. [Consider the mappings
f
n
: y →
n
j=1
x
j
y
j
.]
Exercise 3.6. Let E be a Banach space with closed subspaces F and G such that
E = F ⊕G (i.e., every element of E can be expressed uniquely as the sum of an element
of F and an element of G). Deﬁne P
F
and P
G
by setting
P
F
: E → E; f + g → f and P
G
: E → E; f + g → g ∀ f ∈ F, g ∈ G.
Prove that P
F
and P
G
are bounded linear operators such that P
2
F
= P
F
, P
2
G
= P
G
and
P
F
P
G
= P
G
P
F
= 0.
Exercise 3.7. Find a Banach space E with closed subspaces F and G such that E =
F ⊕G and
P : E → E; f + g → f ∀ f ∈ F, g ∈ G
has norm strictly greater than one. [Let E = R
3
with the norm (x
1
, x
2
, x
3
) =
max¦[x
1
[, [x
2
[, [x
3
[¦.]
2.10. Exercises 3 29
Exercise 3.8. Let E be a Banach space with closed subspaces F and G such that
F ∩ G = ¦0¦. Prove that F ⊕G is closed if and only if there exists C > 0 such that
f Cf + g ∀ f ∈ F, g ∈ G.
Deduce that F ⊕G is closed if and only if
c := inf¦f −g : f ∈ F, g ∈ G, f = g = 1¦ > 0.
Three Dual Spaces
Initial Deﬁnitions
Deﬁnition 3.1. Let X be a vector space over the ﬁeld F. A linear functional on X is a
linear map φ: X →F. The set of all linear functionals on X is a vector space denoted by
X
′
, the algebraic dual space of X, where the vectorspace structure is deﬁned pointwise:
(φ + ψ)(x) := φ(x) + ψ(x) and (αφ)(x) := αφ(x) ∀ φ, ψ ∈ X
′
, α ∈ F.
If X is a normed space then X
′
contains X
∗
:= B(X, F), the topological dual space of
X. An element of X
∗
is said to be a bounded linear functional on X.
Our notation for the algebraic and topological dual spaces is the opposite of that adopted
in [10].
If X is inﬁnite dimensional then the inclusion of X
∗
in X
′
is proper; if X is ﬁnite
dimensional then the spaces coincide. Being interested primarily in analysis, henceforth
the term dual space will refer to the topological dual.
Example 3.2. Recall that
(c
0
)
∗
∼
= ℓ
1
and (ℓ
p
)
∗
∼
= ℓ
q
if p ∈ [1, ∞) and 1/p+1/q = 1, where
∼
= denotes isometric isomorphism. More generally,
if I is a subinterval of R (or a σﬁnite measure space)
L
p
(I)
∗
∼
= L
q
(I)
for the same pairs p and q; the isomorphism is analogous to the ℓ
p
case: g ∈ L
q
(I) yields
an element of L
p
(I)
∗
via f →
_
I
g(t)f(t) dt. Proof that every element of L
p
(I)
∗
arises
this way requires the RadonNikod´ ym theorem [18, Theorem 6.16].
Example 3.3. The RieszFr´echet theorem implies that H
∗ ∼
= H for any Hilbert space
H.
The Weak Topology
Deﬁnition 3.4. Any normed space X gains a natural topology from its dual space, its
weak topology. This is the initial topology generated by X
∗
, i.e., the coarsest topology
31
32 Dual Spaces
to make each map φ ∈ X
∗
continuous. The weak topology on X is denoted by σ(X, X
∗
).
(The letter σ is used here because the German word for weak is schwach.)
It is by no means clear that an inﬁnitedimensional space has any continuous functionals,
but the HahnBanach theorem guarantees a plentiful supply (enough to ensure that
the weak topology is Hausdorﬀ). In order to prove the HahnBanach theorem in full
generality, we need a version of the Axiom of Choice.
Zorn’s Lemma
Deﬁnition 3.5. A partial order on a set A is a preorder (see Deﬁnition 1.27) that is
antisymmetric: for all a, b ∈ A,
a b and b a imply that a = b.
Let A be a set with a partial order . A chain C in A is a subset of A such that, for all
a, b ∈ C, either a b or b a. An upper bound or majorant for B ⊆ A is an element
a ∈ A such that b a for all b ∈ B. An element a ∈ A is maximal if a b implies that
a = b for all b ∈ X.
Lemma 3.6. (Zorn) Let A be a nonempty set with a partial order . If every chain
in A has an upper bound then A has a maximal element.
Proof
We take the lemma as axiomatic; it is equivalent to the Axiom of Choice. (For a
proof of this, see [14, Theorem 1.1.6].)
The HahnBanach Theorem
Deﬁnition 3.7. Let X be a real vector space. A sublinear functional on X is a function
p: X →R such that, for all x, y ∈ X and α ∈ R
+
,
(i) p(αx) = αp(x) (positive homogeneity)
and (ii) p(x + y) p(x) + p(y) (subadditivity).
Theorem 3.8. (HahnBanach) Let p be a sublinear functional on the real vector
space X and suppose that M is a subspace of X. If f ∈ M
′
is a linear functional that
satisﬁes f(m) p(m) for all m ∈ M (p is a majorant for f) then there exists a linear
functional F ∈ X
′
such that F[
M
= f and F(x) p(x) for all x ∈ X.
Proof
The proof of the HahnBanach theorem falls naturally into two parts. The ﬁrst
involves showing that f has a “onedimensional extension”, i.e., f extends to N, where
N has codimension one in M: this should be familiar from the b4 course. The second
part is an application of Zorn’s lemma (which is necessary only if X is nonseparable;
otherwise simple induction will suﬃce). Throughout we assume (as we may) that M is
a proper subspace of X.
3.4. The HahnBanach Theorem 33
Choose a vector x
0
∈ X ¸ M, let N := M +Rx
0
and for all γ ∈ R deﬁne
f
γ
: N →R; m+ αx
0
→ f(m) + αγ.
This is a sound deﬁnition because N is the direct sum of M and Rx
0
, and furthermore
f
γ
[
M
= f. It remains to show that we may ﬁnd γ ∈ R such that f
γ
(x) p(x) for all
x ∈ N, i.e.,
f(m) + αγ p(m + αx
0
) ∀ m ∈ M, α ∈ R.
By positive homogeneity of p and linearity of f, the above inequality will be satisﬁed if
f(m) + γ p(m + x
0
) ∀ m ∈ M and f(m) −γ p(m−x
0
) ∀ m ∈ M.
Hence we wish to ﬁnd γ ∈ R such that
f(m) −p(m−x
0
) γ −f(n) + p(n + x
0
) ∀ m, n ∈ M
but, by the subadditivity of p and linearity of f,
−f(n) + p(n + x
0
) −f(m) + p(m−x
0
) −f(n + m) + p(n + m) 0,
so such γ exists.
Let
S =
_
(g, N) : N is a subspace of X, M ⊆ N, g ∈ N
′
, g[
M
= f, g(x) p(x) ∀ x ∈ N
_
denote all suitable extensions of f; the previous part shows that this set is nonempty.
Deﬁne on S by saying that
(g, N) (h, P) ⇐⇒ N ⊆ P and h[
N
= g
and note that this is a partial order. Let ¦(g
a
, N
a
) : a ∈ A¦ be a chain in (S, ), let
N :=
a∈A
N
a
and deﬁne
g : N →R; x → g
a
(x) if x ∈ N
a
.
It is a simple exercise to see that (g, N) ∈ S and (g
a
, N
a
) (g, N) for all a ∈ A. By
Zorn’s lemma we conclude that there exists (h, P) ∈ S that is maximal for ; if we can
show that P = X then we are done. Suppose otherwise; then P is a proper subspace
of X and there exists a proper extension of h, by the ﬁrst part of this proof. This
contradicts the maximality of (h, P).
Theorem 3.9. (BohnenblustSobczyk) Let p be a seminorm on the vector space
X and suppose that M is a subspace of X. If φ ∈ M
′
is a linear functional such that
[φ(m)[ p(m) for all m ∈ M (φ is dominated by p) then there exists a linear functional
Φ ∈ X
′
that extends φ (i.e., Φ[
M
= φ) and is dominated by p (i.e., [Φ(x)[ p(x) for all
x ∈ X).
34 Dual Spaces
Proof
Suppose ﬁrst that X is a real vector space. Note that a seminorm is a sublinear
functional, so we may apply the HahnBanach theorem to obtain Φ ∈ X
′
such that
Φ[
M
= φ and Φ(x) p(x) for all x ∈ X, but also
−Φ(x) p(−x) = p(x) ∀ x ∈ X
by the homogeneity of the seminorm p. Hence [Φ(x)[ p(x) for all x ∈ X, as required.
Now suppose that X is a complex vector space. We may regard it as a real vector
space and apply the ﬁrst part of this proof to obtain a reallinear functional F on X
that extends Re φ and is dominated by p. Deﬁne Φ by
Φ: X →C; x → F(x) −iF(ix).
It is clear that Φ is additive, and if a, b ∈ R then
Φ((a + ib)x) = F((a + ib)x) −iF(i(a + ib)x)
= F(ax) + F(ibx) −iF(iax) −iF(−bx)
= (a + ib)F(x) −i(a + ib)F(ix) = (a + ib)Φ(x) ∀ x ∈ X,
so Φ ∈ X
′
. Note also that
Φ(m) = F(m) −iF(im) = Re φ(m) −i Re iφ(m)
= Re φ(m) + i Imφ(m) = φ(m) ∀ m ∈ M,
so Φ[
M
= φ. Finally, let x ∈ X and choose α ∈ T := ¦z ∈ C : [z[ = 1¦ such that
αΦ(x) ∈ R
+
. Then
[Φ(x)[ = [αΦ(x)[ = Re αΦ(x) = Re Φ(αx) = F(αx) p(αx) = p(x)
and so Φ is dominated by p, as required.
The Dual Space Separates Points
Theorem 3.10. Let X be a normed space. For all x ∈ X¸¦0¦ there exists φ ∈ X
∗
such
that φ(x) = x and φ = 1.
Proof
Let M = Fx := ¦αx : α ∈ F¦ and deﬁne f : M → F; αx → αx. Note that
[f(αx)[ = [α[ x = αx and so f is dominated on M by  . By Theorem 3.9 there
exists a linear functional φ: X →F such that φ[
M
= f (in particular, φ(x) = x ) and
[φ(y)[ y for all y ∈ X (so that φ 1). Combining these observations gives the
result.
This proves that the weak topology on X is Hausdorﬀ: if x, y ∈ X are such that x ,= y
then there exists φ ∈ X
∗
such that φ(x − y) = x − y , = 0; the claim follows by
Proposition 1.26.
3.4. The HahnBanach Theorem 35
Existence of Completions
Recall that X
∗∗
= (X
∗
)
∗
is the bidual or second dual of the normed space X. For all
x ∈ X deﬁne a linear functional ˆ x on X
∗
by setting ˆ x(φ) = φ(x) and note that
[ˆ x(φ)[ = [φ(x)[ φ x
so that ˆ x ∈ X
∗∗
with ˆ x x. Theorem 3.10 shows that the map
Γ: X → X
∗∗
; x → ˆ x
is an isometry, called the canonical embedding of X into its bidual. If the canonical
embedding is surjective then X is said to be reﬂexive.
Proposition 3.11. If X is a normed vector space then it has a completion: there exists
a Banach space
˜
X and a linear isometry i : X →
˜
X such that i(X) is dense in
˜
X.
Proof
Let
˜
X := Γ(X) be the closure in the bidual X
∗∗
of the image of X under the
canonical embedding. As X
∗∗
is complete (being the dual of a normed space) and closed
subspaces of Banach spaces are complete (Proposition 1.8),
_
˜
X,  
X
∗∗[
˜
X
_
is a Banach
space containing X as a dense subspace. Let i : X →
˜
X; x → ˆ x; it is immediate from
the previous remarks that i is a linear isometry, as required.
There is another, more pedestrian way of ﬁnding the completion of a normed space (or
any metric space) which mimics the way that the real numbers may be constructed as a
collection of equivalence classes of sequences of rational numbers: see [3, pp. 34–35] or
[22, Theorem 11.2.2]; it is not diﬃcult to check that the completion inherits the structure
of a normed space [3, Theorem 2.7].
Y is Complete if B(X, Y ) is Complete
Proposition 3.12. If X and Y are normed vector spaces with X ,= ¦0¦ and B(X, Y )
complete then Y is complete.
Proof
Let (y
n
)
n1
⊆ Y be a Cauchy sequence and let x
0
∈ X be a unit vector. By
Theorem 3.10 there exists a linear functional φ ∈ X
∗
such that φ = 1 = φ(x
0
). For
n 1 deﬁne T
n
∈ B(X, Y ) by setting T
n
x = φ(x)y
n
and note that T
n
x
0
= y
n
. Then
(T
n
−T
m
)x = [φ(x)[ y
n
−y
m
 y
n
−y
m
 x ∀ x ∈ X
and so T
n
− T
m
 y
n
− y
m
, which shows that (T
n
)
n1
is a Cauchy sequence in
B(X, Y ). Let T be the limit of this sequence and conclude by noting that
lim
n→∞
y
n
= lim
n→∞
T
n
x
0
= Tx
0
.
36 Dual Spaces
Vectorvalued Holomorphic Functions
Deﬁnition 3.13. Let X be a complex normed space and let U be an open subset of
C. A function f : U → X is weakly holomorphic if φ
◦
f : U → C is holomorphic for all
φ ∈ X
∗
, and is strongly holomorphic if
lim
h→0
f(z + h) −f(z)
h
exists ∀ z ∈ U,
where this limit is taken with respect to the norm topology on X. Note that every
strongly holomorphic function is weakly holomorphic.
Theorem 3.14. (Liouville) Let f : C → X be a weakly holomorphic function into the
complex normed space X. If f is bounded, i.e., there exists r ∈ R
+
such that f(z) r
for all z ∈ C (equivalently, f(C) ⊆ X
r
) then f is constant.
Proof
For all φ ∈ X
∗
we have that φ
◦
f : C → C is bounded and holomorphic everywhere,
and so constant, by the classical Liouville’s theorem [16, ¸ 5.2]. Hence
φ(f(z)) = φ(f(0)) ∀z ∈ C, φ ∈ X
∗
.
As X
∗
separates points in X we must have that f(z) = f(0) for all z ∈ C, i.e., f is
constant.
The proof holds if the holomorphic function f is only required to be weakly bounded,
i.e., for all φ ∈ X
∗
there exists r
φ
∈ R
+
with [φ(f(z))[ r
φ
for all z ∈ C. This is not a
generalisation, however, as a subset of a normed space is weakly bounded if and only if
it is norm bounded (Exercise 4.6).
The Weak Operator Topology
Deﬁnition 3.15. Let X, Y be normed spaces; the intial topology on B(X, Y ) generated
by the collection of maps ¦T → φ(Tx) : x ∈ X, φ ∈ Y
∗
¦ (where F is equipped with its
usual topology) is called the weak operator topology.
In the same manner as we compared strong operator and norm convergence, the fact
that
[φ(Tx)[ φ Tx ∀ x ∈ X, φ ∈ Y
∗
, T ∈ B(X, Y )
shows that strong operator convergence implies weak operator convergence; thus sets
that are weak operator closed are also strong operator closed, and so the weak operator
topology is coarser than the strong operator topology.
Adjoint Operators
Theorem 3.16. Let X, Y be normed vector spaces and let T ∈ B(X, Y ). There exists
T
∗
∈ B(Y
∗
, X
∗
) such that
φ(Tx) = (T
∗
φ)(x) ∀ x ∈ X, φ ∈ Y
∗
.
3.7. The Weak* Topology 37
Furthermore T
∗
 = T.
Proof
This is a b4 result [24, Theorem 2.2.13].
Deﬁnition 3.17. If M is a subspace of the normed space X and N is a subspace of X
∗
then
M
⊥
:= ¦φ ∈ X
∗
: φ(x) = 0 for all x ∈ M¦
is the annihilator of M and
⊥
N := ¦x ∈ X : φ(x) = 0 for all φ ∈ N¦
is the preannihilator of N.
Theorem 3.18. Let X and Y be normed spaces and let T ∈ B(X, Y ). Then
ker T =
⊥
(imT
∗
) and ker T
∗
= (imT)
⊥
.
Proof
Note that, since Y
∗
separates points in Y ,
ker T = ¦x ∈ X : Tx = 0¦ = ¦x ∈ X : φ(Tx) = 0 for all φ ∈ Y
∗
¦
= ¦x ∈ X : (T
∗
φ)(x) = 0 for all φ ∈ Y
∗
¦ =
⊥
(imT
∗
).
The other identity can be established in the same manner.
The Weak* Topology
Deﬁnition 3.19. Let X be a normed vector space. The weak* topology on X
∗
is the
initial topology generated by the maps
ˆ x: X
∗
→F; φ → φ(x) (x ∈ X),
i.e., the coarsest topology to make these maps continuous. The weak* topology on X
∗
is denoted by σ(X
∗
, X).
Note that the weak* topology is Hausdorﬀ, by Proposition 1.26: if φ, ψ ∈ X
∗
are distinct
then there exists x ∈ X such that φ(x) ,= ψ(x), and so the map ˆ x separates these points.
Note also that φ → φ(x) ∈ X
∗∗
for all x ∈ X, so the weak* topology σ(X
∗
, X) is even
coarser than the weak topology σ(X
∗
, X
∗∗
): there are fewer functions required to be
continuous.
Exercises 4
Exercise 4.1. A closed subspace M of the normed space X is complemented in X if there
exists a closed subspace N such that M ⊕N = X, i.e., M +N = X and M ∩ N = ¦0¦.
Prove that M is complemented in X if M is ﬁnite dimensional. [Start by considering a
38 Dual Spaces
basis of M
∗
.] Prove also that M is complemented in X if M has ﬁnite codimension, i.e.,
dimX/M < ∞.
Exercise 4.2. Let M be a ﬁnitedimensional subspace of the normed space X and let N
be a closed subspace of X such that X = M ⊕N. Prove that if φ
0
is a linear functional
on M then
φ: M ⊕N →F; m+ n → φ
0
(m) ∀ m ∈ M, n ∈ N
is an element of the dual space X
∗
.
Exercise 4.3. Prove that a normed vector space X is separable if its dual X
∗
is. [You
may assume that if M is a nonempty subspace of X and x
0
∈ X ¸
¯
M then there exists
φ ∈ X
∗
such that φ[
M
= 0 and φ(x
0
) = 1.] Find a separable Banach space E such that
E
∗
is not separable. [Proof of (non)separability is not required.] Prove that a reﬂexive
Banach space E is separable if and only if E
∗
is.
Exercise 4.4. Prove that a Banach space E is reﬂexive if and only its dual E
∗
is
reﬂexive.
Exercise 4.5. Prove that any inﬁnitedimensional normed space has a discontinuous
linear functional deﬁned on it.
Exercise 4.6. Let A be a subset of the normed vector space X. Prove that A is norm
bounded (there exists r ∈ R
+
such that a r for all a ∈ A) if and only if it is weakly
bounded (for all φ ∈ X
∗
there exists r
φ
∈ R
+
such that [φ(a)[ r
φ
for all a ∈ A). [Use
the principle of uniform boundedness and the canonical embedding Γ: x → ˆ x.] Deduce
that a weakly holomorphic function is (strongly) continuous. [Cauchy’s integral formula
may be useful.]
Exercise 4.7. Let H be a Hilbert space. Prove that the adjoint T → T
∗
is continuous
with respect to the weak operator topology on B(H), but not necessarily with respect
to the strong operator topology. [For the latter claim, consider the operators T
n
∈ B(ℓ
2
)
such that T
n
x = ¸e
1
, x)e
n
for all x ∈ ℓ
2
, where ¦e
k
: k 1¦ is the standard orthonormal
basis of ℓ
2
.]
Exercise 4.8. Let E and F be Banach spaces. Show that if T : E → F and S: F
∗
→ E
∗
are linear transformations that satisfy
φ(Tx) = (Sφ)(x) ∀ x ∈ E, φ ∈ F
∗
then S and T are bounded, with S = T
∗
. [Use the closedgraph theorem.]
Exercise 4.9. Let E and F be Banach spaces and suppose that T ∈ B(E, F) has closed
range, i.e., imT is closed in F. Prove that imT
∗
= (ker T)
⊥
(where
M
⊥
:= ¦φ ∈ E
∗
: φ(x) = 0 for all x ∈ M¦
is the annihilator of the subspace M ⊆ E).
Exercise 4.10. Let E = c
0
, so that E
∗
= ℓ
1
and E
∗∗
= ℓ
∞
. Prove that x →
∞
n=1
x
n
is
weakly continuous on ℓ
1
but is not weak* continuous.
3.9. Tychonov’s Theorem 39
Exercise 4.11. Prove that a compact metric space is separable. Prove that if X is a
separable normed space then X
∗
1
, the closed unit ball of the dual space X
∗
, is metrizable
when equipped with the weak* topology. [Let (x
n
)
n1
⊆ X
1
be dense in X
1
and consider
d(φ, ψ) :=
∞
n=1
2
−n
[φ(x
n
)−ψ(x
n
)[.] Deduce that X
∗
is separable in the weak* topology.
Exercise 4.12. Let X and Y be normed spaces and for all x ∈ X and y ∈ Y let
x ⊗y : B(X, Y
∗
) →F; T → (Tx)(y).
Prove that x ⊗y ∈ B(X, Y
∗
)
∗
, with x ⊗y = x y, and that the mapping
X Y → B(X, Y
∗
); (x, y) → x ⊗y
is bilinear. If Z is the closed linear span of ¦x ⊗y : x ∈ X, y ∈ Y ¦ in B(X, Y
∗
)
∗
, prove
that
j : B(X, Y
∗
) → Z
∗
; j(T)z = z(T)
is an isometric isomorphism. [For surjectivity, let φ ∈ Z
∗
and consider φ
x
: y → φ(x⊗y).]
Tychonov’s Theorem
Deﬁnition 3.20. Let
_
(X
a
, T
a
) : a ∈ A
_
be a collection of topological spaces. Their
topological product is (X, T), where
X =
a∈A
X
a
:= ¦(x
a
)
a∈A
: x
a
∈ X
a
∀ a ∈ A¦
is the Cartesian product of the sets X
a
and T =
a∈A
T
a
is the product topology, i.e., the
initial topology generated by the projection maps
π
b
: X → X
b
; (x
a
)
a∈A
→ x
b
(b ∈ A).
The fundamental fact about the product topology is that is preserves compactness; this
is Tychonov’s theorem. The proof of Tychonov’s theorem is not particularly simple; to
avoid clutter, we remind ourselves of some minor points from basic topology.
Let (X, T) be a topological space.
(i) The space (X, T) is compact if and only if every collection of closed subsets of
X with the ﬁniteintersection property has nonempty intersection; a collection F
of subsets of X has the ﬁniteintersection property if F
1
∩ . . . ∩ F
n
,= ∅ for all
F
1
, . . . , F
n
∈ F. [For a proof of this, see [21, Theorem 21.D] or [3, p.116].]
(ii) If A ⊆ X then x ∈
¯
A if and only if every open set containing x meets A. [Otherwise
x ∈ (X ¸ A)
◦
= X ¸
¯
A.]
(iii) If X is the product of ¦(X
a
, T
a
) : a ∈ A¦ then every set in T is the union of sets of
the form
n
i=1
π
−1
a
i
(U
i
) (n ∈ N, a
1
, . . . , a
n
∈ A, U
1
∈ T
a
1
, . . . , U
n
∈ T
an
);
these sets form a base for T.
40 Dual Spaces
Theorem 3.21. (Tychonov) Let ¦(X
a
, T
a
) : a ∈ A¦ be a collection of compact topo
logical spaces. The product space (X, T) is compact.
Proof
Let F be a family of closed subsets of X with the ﬁniteintersection property. By
Zorn’s lemma we may ﬁnd a maximal family H of (not necessarily closed) subsets of X
such that H contains F and has the ﬁniteintersection property. Note that H is closed
under ﬁnite intersections: if A
1
, . . . , A
n
∈ H then A = A
1
∩ . . . ∩ A
n
∈ H, as otherwise
H∪ ¦A¦ strictly contains H, contains F and has the ﬁniteintersection property, which
contradicts the maximality of H. Furthermore, if A ⊆ X is such that A∩ H ,= ∅ for all
H ∈ H then A ∈ H; otherwise considering H∪ ¦A¦ leads to the same contradiction.
Let a ∈ A and note that ¦π
a
(H) : H ∈ H¦ has the ﬁniteintersection property
(because f(
b∈B
S
b
) ⊆
b∈B
f(S
b
) for any function f and collection of sets ¦S
b
¦), so
there exists x
a
∈
H∈H
π
a
(H) by the compactness of X
a
. We complete the proof by
showing that x := (x
a
)
a∈A
∈
H∈H
H, which suﬃces to show that
F∈F
F ,= ∅.
Any open set containing x contains a set of the form U =
a∈A
0
π
−1
a
(U
a
), where
U
a
∈ T
a
contains x
a
and A
0
is a ﬁnite subset of A. If a ∈ A
0
then x
a
∈ U
a
∩ π
a
(H), so
π
−1
a
(U
a
) ∩H ,= ∅, for all H ∈ H. Hence π
−1
a
(U
a
) ∈ H for all a ∈ A
0
, and so U ∈ H, as H
is closed under ﬁnite intersections. As H has the ﬁniteintersection property, U
H ,= ∅
for all H ∈ H, and therefore x ∈
H∈H
¯
H, as required.
The BanachAlaoglu Theorem
Theorem 3.22. (BanachAlaoglu) If X is a normed space then X
∗
1
, the closed unit
ball of X
∗
, is compact in the weak* topology.
Proof
By Tychonov’s theorem, the space K =
x∈X
F
x
is compact when equipped with
the product topology. The map
F : X
∗
1
→ K; φ →
_
φ(x)
_
x∈X
is well deﬁned (since [φ(x)[ φ x x ) and injective, so F
−1
is well deﬁned on
F(X
∗
1
). This map is weak* continuous, by Proposition 1.25, since ˆ x[
X
∗
1
◦
F
−1
= π
x
[
F(X
∗
1
)
for all x ∈ X, and X
∗
1
= F
−1
(F(X
∗
1
)) is weak* compact if F(X
∗
1
) is closed in K. To
prove this, by Theorem 1.34 it suﬃces to take a net (φ
a
)
a∈A
in X
∗
1
such that
_
F(φ
a
)
_
a∈A
has limit f = (f
x
)
x∈X
∈ K and show that φ: x → f
x
∈ X
∗
1
. Note that F(φ
a
) → f if and
only if φ
a
(x) → f
x
, by Proposition 1.37, and so
f
x
+ αf
y
= lim
a∈A
φ
a
(x) + αlim
a∈A
φ
a
(y) = lim
a∈A
φ
a
(x + αy) = f
x+αy
∀ x, y ∈ X, α ∈ F.
Hence φ: x → f
x
∈ X
′
, and f
x
∈ F
x
for all x ∈ X implies that [φ(x)[ = [f
x
[ x, so
φ 1, as required. (The linearity result used here follows from continuity of addition
and multiplication in F.)
3.11. Topological Vector Spaces 41
Characterisation of Normed Vector Spaces
The following theorem reduces the study of normed vector spaces to the study of sub
spaces of a particular type of Banach space, the collection of continuous functions on
a compact, Hausdorﬀ space. In particular, all Banach spaces are isomorphic to closed
subspaces of C(K) for some compact, Hausdorﬀ space K.
Theorem 3.23. Let (X,  ) be a normed space. There exists a compact, Hausdorﬀ
space K and a linear isometry i : X → C(K) such that X is isometrically isomorphic to
i(X), a subspace of C(K), which is closed if and only if X is complete.
Proof
Let K = X
∗
1
be the closed unit ball of X
∗
, equipped with the (restriction of the) weak*
topology. (This is Hausdorﬀ, being a subspace of the Hausdorﬀ space
_
X
∗
, σ(X
∗
, X)
_
.)
Deﬁne i : X → C(K) by setting i(x) = ˆ x[
K
, i.e.,
i(x): K →F; φ → φ(x);
this map is continuous by the deﬁnition of the weak* topology, and i is clearly linear.
Furthermore,
i(x)
∞
= sup¦[i(x)(φ)[ : φ ∈ K¦ = sup¦[φ(x)[ : φ ∈ K¦ = x,
by Theorem 3.10. Hence i is an isometry, so i(X) is closed if and only if X is complete:
a subspace of a Banach space is closed if and only if it is complete (Proposition 1.8) and
isometries preserve Cauchy sequences.
The previous theorem is the starting point which motivates the theory of operator spaces:
every Banach space is isometrically isomorphic to a closed subspace of some C(K),
which is the paradigm example of a commutative C
∗
algebra. An operator space is a
closed subspace of some B(H), where H is a Hilbert space; B(H) is the natural non
commutative generalisation of C(K). This is a very active area of current research (see
[5] or [15]).
Topological Vector Spaces
Deﬁnition 3.24. Let X be a vector space. A set of linear functionals A ⊆ X
′
is
separating if for all x ∈ X ¸ ¦0¦ there exists φ ∈ A such that φ(x) ,= 0. If M ⊆ X
′
is a
separating subspace then σ(X, M) := T
M
is the initial topology on X generated by M;
this topology is Hausdorﬀ by Proposition 1.26.
The weak and weak* topologies are deﬁned in this fashion.
Proposition 3.25. Let X be a vector space with separating subspace M ⊆ X
′
. A linear
functional φ ∈ X
′
is σ(X, M)continuous if and only if φ ∈ M.
42 Dual Spaces
Proof
See Exercise 5.1.
Proposition 3.26. Let X be a vector space over F and let M ⊆ X
′
be a separating
subspace. The functions
F X → X; (α, x) → αx and X X → X; (x, y) → x + y
are continuous (where X is equipped with the topology σ(X, M) and F has its usual
topology).
Proof
The function (α, x) → αx is continuous if (α, x) → φ(αx) is continuous for all φ ∈ M,
by Proposition 1.25, and this function is continuous if φ(α
a
x
a
) → φ(αx) for any net
(α
a
, x
a
)
a∈A
⊆ FX such that (α
a
, x
a
) → (α, x), by Proposition 1.35. If (α
a
, x
a
) → (α, x)
in F X then (by the deﬁnition of the product topology and Proposition 1.37) α
a
→ α
and x
a
→ x, i.e., φ(x
a
) → φ(x) for all φ ∈ M. Continuity of multiplication in F yields
φ(α
a
x
a
) = α
a
φ(x
a
) → αφ(x) = φ(αx) ∀φ ∈ M,
as required. The proof for the other function is similar (and depends upon the continuity
of addition in F).
Deﬁnition 3.27. A topological vector space is a vector space X equipped with a Haus
dorﬀ topology such that the maps
X X → X; (x, y) → x + y
and
F X → X; (α, x) → αx
are continuous.
Any normed space, or vector space equipped with the topology given by a separating
subspace of linear functionals, is a topological vector space.
Lemma 3.28. Let X be a topological vector space. A linear functional φ ∈ X
′
is
continuous if and only if [φ[
−1
[0, 1) contains an open set containing 0.
Proof
One implication is immediate from the deﬁnitions. For the converse, suppose that
U ⊆ X is an open set containing 0 and such that [φ(u)[ < 1 for all u ∈ U. Let A ⊆ F
be open and let x ∈ φ
−1
(A); there exists ε
x
> 0 such that B
x
:= B
F
εx
_
φ(x)
_
⊆ A. As
[φ(y) −φ(x)[ = [φ(y −x)[ < ε
x
∀ y ∈ x + ε
x
U,
we see that
φ
−1
(A) ⊇ φ
−1
(B
x
) ⊇ x + ε
x
U
and so φ
−1
(A) =
x∈φ
−1
(A)
(x+ε
x
U) is open, as required. (The fact that x+ε
x
U is open
follows as the maps y → y + x and y → ε
x
y are homeomorphisms of X to itself.)
3.11. Topological Vector Spaces 43
Separation
Deﬁnition 3.29. A subset C of a vector space is convex if tC + (1 − t)C ⊆ C for all
t ∈ (0, 1), i.e., tx + (1 − t)y ∈ C for all x, y ∈ C and t ∈ (0, 1). [Geometrically, this
condition states that every line segment with endpoints in C lies in C. It is immediate
that linear transformations preserve convexity.]
Lemma 3.30. Let X be a real topological vector space. If C is a convex, open set in
X that contains the origin then the map
µ
C
: X →R
+
; x → inf¦t ∈ R
+
: x ∈ tC¦
(called the gauge or Minkowski functional of C) is a sublinear functional on X such that
C = µ
−1
C
[0, 1) = ¦x ∈ X : µ
C
(x) < 1¦.
Proof
If x ∈ X then m
x
: R → X; t → tx is continuous, so m
−1
x
(C) is open and contains 0.
Hence there exists δ > 0 such that tx ∈ C if [t[ < δ, i.e., x ∈ sC if [s[ > δ
−1
. This shows
that µ
C
is well deﬁned.
Let x, y ∈ X and suppose that ε > 0; we may ﬁnd s, t > 0 such that s < µ
C
(x) +ε,
t < µ
C
(y) + ε and x ∈ sC, y ∈ tC. Then
x + y ∈ sC + tC = (s + t)
_
s
s + t
C +
t
s + t
C
_
⊆ (s + t)C,
by the convexity of C, and µ
C
(x + y) s + t < µ
C
(x) + µ
C
(y) + 2ε. Since ε > 0 is
arbitrary we have the subadditivity of µ
C
.
Positive homogeneity is immediate: if s > 0 then x ∈ tC if and only if sx ∈ stC,
hence µ
C
(sx) = sµ
C
(x) for all x ∈ X and s ∈ R
+
.
Finally, if x ∈ C then (1 + ε)x ∈ C for some ε > 0 (because m
−1
x
(C) is open and
contains 1) and so x ∈ (1 + ε)
−1
C, which yields µ
C
(x) < 1. Conversely, if µ
C
(x) < 1
then x ∈ tC for some t < 1, whence t
−1
x ∈ C and so x = (1 −t)0 +t(t
−1
x) ∈ C because
0 ∈ C and C is convex.
Deﬁnition 3.31. A topological vector space is locally convex if every open set containing
the origin contains a convex open set containing the origin. A normed space, or a topo
logical vector space with topology given by a separating subspace of linear functionals,
is locally convex (Exercise 5.8).
Theorem 3.32. Let X be a topological vector space and let A, B be nonempty, disjoint,
convex subsets of X.
(i) If A is open then there exists a continuous linear functional φ ∈ X
′
and s ∈ R such
that
Re φ(x) < s Re φ(y) ∀ x ∈ A, y ∈ B.
(ii) If X is locally convex, x
0
∈ X ¸ B and B is closed then there exists a continuous
linear functional φ ∈ X
′
such that
Re φ(x
0
) < inf¦Re φ(y) : y ∈ B¦.
44 Dual Spaces
Proof
First, note that we may assume that the scalar ﬁeld F = R, working as in the last part
of the proof of Theorem 3.9: if φ is a continuous, reallinear functional on X satisfying
(i) or (ii) then Φ: x → φ(x) − iφ(ix) is an continuous element of X
′
with the same
property (since Re Φ = φ).
For (i), let a
0
∈ A, b
0
∈ B and consider C = A−B +(b
0
−a
0
); it contains 0, is open
(being the union of translates of the open set A) and is convex (this is immediate upon
checking the deﬁnition). Hence the gauge µ
C
is sublinear on X.
Since A and B are disjoint, x
0
:= b
0
−a
0
/ ∈ C, and therefore µ
C
(x
0
) 1. Let N = Rx
0
and deﬁne φ
0
on N by setting φ
0
(tx
0
) = t for all t ∈ R. Then φ
0
µ
C
on N, and so, by
Theorem 3.8, there exists a (reallinear) functional φ such that φ(x
0
) = φ
0
(x
0
) = 1 and
φ µ
C
. If x ∈ C then φ(x) µ
C
(x) < 1 and if x ∈ −C then φ(x) = −φ(−x) > −1.
Hence [φ(x)[ < 1 on C ∩ (−C); since C ∩ (−C) is an open set containing the origin, φ
is continuous (by Lemma 3.28).
If a ∈ A, b ∈ B then a −b + x
0
∈ C and so
φ(a) −φ(b) + 1 = φ(a −b + x
0
) µ
C
(a −b + x
0
) < 1,
hence φ(a) < φ(b). Then φ(A) and φ(B) are disjoint, convex subsets of R and φ(A) is
open (see Exercise 5.3), so taking s = sup φ(A) gives the result.
For (ii), note that (X ¸ B) − x
0
is open and contains 0, so by local convexity there
exists a convex, open set U such that 0 ∈ U ⊆ (X¸B)−x
0
. Then A := x
0
+U is convex,
open and contained in X ¸ B; the result follows by (i).
Corollary 3.33. If X is a locally convex topological vector space and x ∈ X ¸ ¦0¦ then
there exists a continuous linear functional φ ∈ X
′
such that φ(x) ,= 0. In other words,
the topological dual of X separates points.
Proof
Let B = ¦0¦, x
0
= x and apply Theorem 3.32(ii).
Example 3.34. Let 0 < p < 1,
L
p
[0, 1] :=
_
f : [0, 1] →C
¸
¸
f is measurable and d(f, 0) < ∞
_
,
where
d(f, g) :=
_
1
0
[f(t) −g(t)[
p
dt,
and let L
p
[0, 1] = L
p
[0, 1]/N, where N := ¦f ∈ L
p
[0, 1] : d(f, 0) = 0¦ is the subspace of
functions zero almost everywhere. The map ([f], [g]) → d(f, g) is a metric on L
p
[0, 1]
which makes it a topological vector space (with algebraic operations deﬁned pointwise).
However, this topology is not locally convex and the only continuous linear functional
on L
p
[0, 1] is the zero functional (Exercise 5.6),
3.12. The KreinMilman Theorem 45
A consequence of Proposition 3.25 is the fact that the collections of normcontinuous
and weakly continuous linear functionals on a normed space coincide; combined with
the separation theorem this yields the following.
Corollary 3.35. A convex subset of a normed space is norm closed if and only if it is
weakly closed.
Proof
Since the weak topology is coarser than the norm topology, we need only consider a
nonempty, normclosed, convex subset C of the normed space X. If x ,∈ C then there
exists ε > 0 such that B
ε
(x) ∩ C = ∅; applying Theorem 3.32(i) with A = B
ε
(x) and
B = C yields φ ∈ X
∗
and s ∈ R such that
¦y ∈ X : Re φ(y) < s¦ ∩ C = ∅.
As (Re φ)
−1
(−∞, s) is weakly open and contains x, it follows that C is weakly closed
and we have the result.
In fact, the previous proposition holds for all locally convex topological vector spaces:
the proper generalisation of the weak topology is the initial topology generated by all
the continuous linear functionals (Exercise 5.9).
The KreinMilman Theorem
Deﬁnition 3.36. Let X be a vector space. A face of a convex set C is a nonempty,
convex subset F ⊆ C such that if t ∈ (0, 1) and x, y ∈ C satisfy tx + (1 −t)y ∈ F then
x and y ∈ F. An extreme point of a convex set C is a onepoint face, i.e., an element of
C that cannot be expressed as a nontrivial convex combination of elements of C. [We
blur the distinction between an extreme point and the singleton set containing it.] The
extremal boundary of C is the set of its extreme points, denoted by ∂
e
C
In geometrical terms, a face F is a subset of the convex set C such that if f ∈ F and ℓ is
a line through f then ℓ ∩C ⊆ F. An extreme point is a point of C that is not contained
in the interior of any line segment in C.
[Some pictures would go well here.]
Deﬁnition 3.37. If X is a vector space and A ⊆ X then
cnv A :=
_
n
i=1
α
i
x
i
: n ∈ N, α
1
, . . . , α
n
∈ R
+
,
n
i=1
α
i
= 1, x
1
, . . . , x
n
∈ A
_
is the convex hull of A. If X is a topological vector space then the closed convex hull of
A, denoted by cnv A, is the closure of cnv A.
Proposition 3.38. If X is a vector space and A ⊆ X then cnv A is the smallest convex
set containing A. If X is a topological vector space then cnv A is the smallest closed,
convex set containing A.
46 Dual Spaces
Proof
The convexity of cnv A is readily veriﬁed. Let B be any convex set containing A; we
claim that cnv A ⊆ B, i.e., for all n ∈ N,
n
i=1
α
i
x
i
∈ B if x
1
, . . . , x
n
∈ A and α
1
, . . . , α
n
∈ R
+
with
n
i=1
α
i
= 1. (3.1)
To see this, we proceed by induction: the cases n = 1 and 2 are immediate, so suppose
that n > 2 and that (3.1) holds for sums containing n − 1 terms. Without loss of
generality α
n
,= 1 and
n
i=1
α
i
x
i
= (1 −α
n
)
_
n−1
i=1
α
i
1 −α
n
x
i
_
+ α
n
x
n
∈ B
by the inductive hypothesis and convexity. This proves the ﬁrst statement.
If B is a closed, convex set containing A then cnv A ⊆ B (since B is convex) and
hence cnv A ⊆
¯
B = B (taking closures). It remains to prove that cnv A is convex; let
x, y ∈ cnv A and choose a net (x
p
, y
p
)
p∈P
⊆ cnv A cnv A such that (x
p
, y
p
) → (x, y)
(recall that the closure of a product is the product of the closures). By continuity of
scalar multiplication and vector addition,
tx + (1 −t)y = t lim
p∈P
x
p
+ (1 −t) lim
p∈P
y
p
= lim
p∈P
tx
p
+ (1 −t)y
p
∈ cnv A
: =
cnv A
for all t ∈ (0, 1), as required.
Lemma 3.39. Let X be a topological vector space and let C be a nonempty, compact,
convex subset of X. If φ ∈ X
′
is continuous then
F :=
_
x ∈ C : Re φ(x) = min¦Re φ(y) : y ∈ C¦
_
is a closed face of C.
Proof
The set F is nonempty (since C is compact and x → Re φ(x) is continuous), closed
(since it is the preimage of a point under the continuous function Re φ) and convex (since
Re φ is reallinear). Furthermore, if t ∈ (0, 1) and x, y ∈ C are such that tx+(1−t)y ∈ F
then
min
C
Re φ = Re φ(tx + (1 −t)y) = t Re φ(x) + (1 −t) Re φ(y) min
C
Re φ. (3.2)
(The notation min
C
means that the minimum is taken over the set C.) If x is not in F
then Re φ(x) > min
C
Re φ, which gives a strict inequality in (3.2), a contradiction, and
similarly for y. Hence x, y ∈ F, as required.
Theorem 3.40. (KreinMilman) Let X be a locally convex topological vector space
and let C be a nonempty, compact, convex subset of X. Then C = cnv ∂
e
C, i.e., C is
the closed convex hull of its extreme points. (In particular, ∂
e
C is nonempty.)
3.13. Exercises 5 47
Proof
Let F denote the collection of closed faces of C; it is an exercise to verify that
(F, ⊇) is a nonempty, partially ordered set, such that every chain in F has an upper
bound. Applying Zorn’s lemma we obtain a maximal element of F, i.e., a closed face
F
0
such that if F
1
∈ F satisﬁes F
0
⊇ F
1
then F
1
= F
0
. We claim that F
0
∈ ∂
e
C; to
see this, suppose otherwise, so that there exist distinct x, y ∈ F
0
. By Corollary 3.33
there exists a continuous φ ∈ X
′
such that φ(x) ,= φ(y), and without loss of generality
Re φ(x) < Re φ(y) (else we replace φ by one of −φ, iφ or −iφ). Let
F
1
=
_
z ∈ F
0
: Re φ(z) = min¦Re φ(w) : w ∈ F
0
¦
_
;
this is a proper subset of F
0
(since y / ∈ F
1
) and a closed face of F
0
, by Lemma 3.39, and
so a closed face of C: a face of a face of C is itself a face of C. This contradicts the
minimality of F
0
and so ∂
e
C ,= ∅.
It is immediate that C ⊇ cnv ∂
e
C, so to complete the proof suppose for contradiction
that x ∈ C ¸ cnv ∂
e
C. By Theorem 3.32(ii) (applied to x and cnv ∂
e
C) we may ﬁnd a
continuous ψ ∈ X
′
such that Re ψ(x) < min¦Re ψ(y) : y ∈ cnv ∂
e
C¦. Let
F =
_
z ∈ C : Re ψ(z) = min¦Re ψ(w) : w ∈ C¦
_
;
this is a closed face of C, by Lemma 3.39, and applying the ﬁrst part of this proof, with
F in place of C, yields z ∈ ∂
e
F ⊆ ∂
e
C. Hence
min
C
Re ψ = Re ψ(z) > Re ψ(x) min
C
Re ψ,
the desired contradiction.
Exercises 5
Exercise 5.1. Prove that if X is a vector space with separating subspace M ⊆ X
′
and
φ ∈ X
′
is a linear functional that is σ(X, M)continuous then there exist φ
1
, . . . , φ
n
∈ M
such that
[φ(x)[ max
1in
[φ
i
(x)[ ∀ x ∈ X.
Deduce that
n
i=1
ker φ
i
⊆ ker φ and that there exists f ∈ (F
n
)
∗
such that
f
_
φ
1
(x), . . . , φ
n
(x)
_
= φ(x) ∀ x ∈ X.
Conclude that φ ∈ M.
Exercise 5.2. Let X be an inﬁnitedimensional normed space and let V ⊆ X be a
weakly open set containing the origin. Show that V contains a closed subspace of ﬁnite
codimension in X. Deduce that the weak topology on X is strictly coarser than the
norm topology.
Exercise 5.3. Let X be a topological vector space. Prove that every φ ∈ X
′
¸ ¦0¦ is
open. [Note that m
x
: α → αx is continuous for all x ∈ X and that there exists x
0
∈ X
such that φ(x
0
) = 1.]
48 Dual Spaces
Exercise 5.4. Suppose that X is a vector space equipped with a topology that makes
vector addition and scalar multiplication, i.e., the maps
X X → X; (x, y) → x + y and F X → X; (α, x) → αx,
continuous. Show that if this topology is such that singleton sets are closed (i.e., ¦x¦ is
closed for all x ∈ X) then the topology is Hausdorﬀ (so X is a topological vector space).
Exercise 5.5. Let X be a topological vector space. Prove that every open set containing
the origin contains a nonempty open set which is balanced: a set B is balanced if λb ∈ B
for all b ∈ B and λ ∈ F
1
. [Balanced sets are in some ways analogous to open balls about
the origin in normed spaces.] Deduce that if C ⊆ X is compact and does not contain
the origin then there exist disjoint open sets A, B ⊆ X such that C ⊆ A and B is a
balanced set containing 0. Show that a balanced set is connected and give an example
to show that a balanced set need not be convex.
Exercise 5.6. Let p ∈ (0, 1),
L
p
[0, 1] :=
_
f : [0, 1] →C
¸
¸
f is measurable and ∆(f) < ∞
_
,
where ∆(f) :=
_
1
0
[f(x)[
p
dx, and let L
p
[0, 1] := L
p
[0, 1]/N, where
N :=
_
f : [0, 1] →C
¸
¸
f is measurable and zero almost everywhere
_
.
Prove that d([f], [g]) := ∆(f −g) is a metric on L
p
[0, 1] and that L
p
[0, 1] is a topological
vector space (when equipped with this topology). Prove further that L
p
[0, 1] has no
convex, open sets other than ∅ and L
p
[0, 1]. Deduce that the only continuous linear
functional on L
p
[0, 1] is the zero functional.
Exercise 5.7. Let X be a topological vector space over F and let M be a ﬁnite
dimensional subspace of X. Prove that M is linearly homeomorphic to F
n
, where n
is the dimension of M. Prove also that M is closed in X.
Exercise 5.8. Prove that a topological vector space with topology given by a separating
family of linear functionals is locally convex.
Exercise 5.9. Suppose that X is a locally convex topological vector space and M is
the collection of continuous linear functionals on X. Prove that a convex subset of X is
closed (with respect to the original topology) if and only if it is closed with respect to
σ(X, M), the initial topology generated by M. Need this hold if X is not locally convex?
Exercise 5.10. Let X be a locally convex topological vector space. Show that if N
is a nonempty subspace of X and x
0
∈ X ¸
¯
N then there exists a continuous linear
functional φ ∈ X
′
such that φ[
N
= 0 and φ(x
0
) = 1. [Use the separation theorem.]
Exercise 5.11. Let X be a topological vector space and suppose V is an open set
containing 0. Prove there exists an open set U containing 0 such that U + U ⊆ V .
Deduce or prove otherwise that if A ⊆ B ⊆ X, where A is compact and B is open, then
there exists an open set U ⊆ X such that 0 ∈ U and A+ U ⊆ B.
3.13. Exercises 5 49
Exercise 5.12. Suppose that X is a topological vector space such that the continuous
elements of X
′
separate points. Prove that given disjoint, nonempty, compact, convex
A, B ⊆ X there exists a continuous φ ∈ X
′
such that
sup
x∈A
Re φ(x) < inf
x∈B
Re φ(x).
[Consider X equipped with the topology σ(X, M), where M is the collection of contin
uous linear functionals on X.] Deduce that Theorem 3.40 is true for topological spaces
with continuous dual spaces that separate points.
Exercise 5.13. Let X be a topological vector space and suppose that C is a nonempty,
compact, convex subset of X. Prove that (F, ⊇), the collection of closed faces of C
ordered by reverse inclusion, is a nonempty, partially ordered set such that every chain
in F has an upper bound.
Exercise 5.14. Prove that the closed unit ball of c
0
has no extreme points.
Exercise 5.15. Let H be a Hilbert space. Prove that every unit vector in H is an
extreme point of the closed unit ball H
1
. [Note that 1 is an extreme point of F
1
.]
Deduce that every isometry in B(H) is an extreme point of the closed unit ball B(H)
1
.
Exercise 5.16. Let C be a convex subset of a topological vector space X. Prove that if
x ∈ C and y ∈ C
◦
, the interior of C, then tx+(1 −t)y ∈ C
◦
for all t ∈ [0, 1). Prove also
that the interior C
◦
and the extremal boundary ∂
e
C are disjoint (as long as X ,= ¦0¦).
Exercise 5.17. Let C ⊆ R
n
be compact and convex. Prove that every element of C can
be written as a convex combination of at most (n+1) elements of ∂
e
C. [Use induction.]
Algebras
51
Four Normed Algebras
We start with some purely algebraic deﬁnitions. All the algebras that we consider will
have scalar ﬁeld R or C.
Deﬁnition 4.1. An algebra (more correctly, an associative algebra) is a vector space A
equipped with a bilinear map (called multiplication)
: AA → A; (a, b) → ab
that is associative:
(ab)c = a(bc) ∀ a, b, c ∈ A.
An algebra is commutative if its multiplication is, i.e., ab = ba for all a, b ∈ A. An
algebra is unital if there exists 1 ∈ A such that 1a = a1 = a for all a ∈ A; such an
element is unique if it exists. A subalgebra of the algebra A is a subspace B that is closed
under multiplication: ab ∈ A for all a, b ∈ B (more brieﬂy, B
2
⊆ B).
Example 4.2. A ﬁeld is a unital algebra over itself. (If the underlying scalar ﬁeld F
of an algebra needs to be mentioned explicitly we refer to an algebra over F.) More
generally, the collection of n n matrices over a ﬁeld F forms a unital algebra, denoted
by M
n
(F), when equipped with the usual multiplication:
_
a
i
j
_
n
i,j=1
_
b
j
k
_
n
j,k=1
=
_
n
j=1
a
i
j
b
j
k
_
n
i,k=1
.
The examples above are ﬁnitedimensional (the dimension of an algebra is its dimension
as a vector space) and every ﬁnitedimensional algebra A over F is isomorphic to a
subalgebra of M
n
(F) (with n = dimA if A is unital, or n = 1 +dimA otherwise), where
homomorphism and isomorphism are deﬁned in the usual manner: see Deﬁnition 4.7
Example 4.3.
If X is a topological space then the set of complexvalued, continuous functions on X
is an algebra over C, denoted by C(X), with the algebraic operations deﬁned pointwise:
if f, g ∈ C(X) and α ∈ C then
(f + αg)(x) := f(x) + αg(x) and (fg)(x) := f(x)g(x) ∀ x ∈ X.
This algebra has three important subalgebras:
53
54 Normed Algebras
(i) C
00
(X), the continuous functions on X with compact support;
(ii) C
0
(X), the continuous functions on X that vanish at inﬁnity;
(iii) C
b
(X), the bounded, continuous functions on X.
We have
C
00
(X) ⊆ C
0
(X) ⊆ C
b
(X) ⊆ C(X),
with equality if and only if X is compact. Note that C
b
(X) and C(X) are unital, but
C
00
(X) and C
0
(X) are unital only if X is compact.
Deﬁnition 4.4. A normed algebra is a normed space that is also an associative algebra,
such that the norm is submultiplicative: ab a b for all a, b ∈ A. A Banach
algebra is a complete normed algebra, i.e., a normed algebra that is also a Banach space
(with respect to its norm). A normed algebra is unital if it is a unital algebra and
1 = 1.
Note that the submultiplicativity of the norm means that multiplication in normed
algebras is jointly continuous: if a
n
→ a and b
n
→ b then (a
n
)
n1
is bounded and
a
n
b
n
−ab = a
n
(b
n
−b) + (a
n
−a)b a
n
 b
n
−b +a
n
−a b
sup
_
a
n

_
b
n
−b +b a
n
−a → 0
as n → ∞.
Example 4.5. If X is a Banach space then
_
B(X),  
_
is a unital Banach algebra,
where   is the operator norm; this example generalises Example 4.2, as if X is ﬁnite
dimensional then B(X) is isomorphic to M
n
(F) (where n = dimX).
Example 4.6. Let X be a locally compact, Hausdorﬀ space. When equipped with the
supremum norm
 
∞
: f → f
∞
:= sup¦[f(x)[ : x ∈ X¦,
C
b
(X) is a unital Banach algebra, C
0
(X) is a closed subalgebra of C
b
(X) (so a Banach
algebra in its own right) and C
00
(X) is a dense subalgebra of C
0
(X). [This follows from
Proposition 2.18.]
Quotient Algebras
Deﬁnition 4.7. Let A and B be algebras over the ﬁeld F. An algebra homomorphism
is an Flinear map φ: A → B such that φ(ab) = φ(a)φ(b) for all a, b ∈ A. An algebra
isomorphism is a bijective algebra homomorphism (which implies that φ
−1
is an algebra
homomorphism from B to A). The kernel of an algebra homomorphism φ: A → B is
ker φ := ¦a ∈ A : φ(a) = 0¦ and the image is imφ := ¦φ(a) : a ∈ A¦.
4.2. Unitization 55
The obvious fundamental theorem about homomorphisms is true; in order to state it we
need to deﬁne the concept of a quotient algebra.
Deﬁnition 4.8. Let A be an algebra. An ideal of the algebra A is a subspace I such
that ab, ba ∈ I for all a ∈ A and b ∈ I (i.e., AI, IA ⊆ I). If I is an ideal of A then the
quotient space A/I is an algebra, called the quotient algebra of A by I, when equipped
with the multiplication
(a + I)(b + I) = ab + I ∀ a, b ∈ A,
and A/I is unital if A is unital.
Theorem 4.9. Let A be an algebra and φ: A → B an algebra homomorphism. Then
ker φ is an ideal of A, imφ is a subalgebra of B and A/ ker φ
∼
= imφ, via
˜
φ: A → imφ; [a] := a + ker φ → φ(a).
Proof
This is trivial.
Proposition 4.10. If A is a normed algebra and I is a closed, proper ideal then A/I is
a normed algebra when equipped with the quotient norm, unital if A is unital; if A is a
Banach algebra then so is A/I.
Proof
To see that the quotient norm is submultiplicative, note that if a, b ∈ A then
_
_
[a]
_
_
A/I
_
_
[b]
_
_
A/I
= inf
_
a −x : x ∈ I
_
inf
_
b −y : y ∈ I
_
inf
_
ab −ay −xb + xy : x, y ∈ I
_
inf
_
ab −x : x ∈ I
_
=
_
_
[ab]
_
_
A/I
.
In particular, if A is unital then
[1]
A/I
= [1]
2

A/I
[1]
2
A/I
so [1]
A/I
1 (since I is proper, 1 / ∈ I and so [1] ,= [0], whence [1] > 0). We have
also that [1]
A/I
1
A
= 1 and therefore [1]
A/I
= 1. Everything else follows from
Theorem 1.10.
Unitization
In many cases an algebra has a unit; there are some situation, however, when none exists
but it would be useful to act as though one did.
Deﬁnition 4.11. If A is an algebra over the ﬁeld F then A
u
is the unitization of A,
deﬁned by setting A
u
= A⊕F and
(a, α)(b, β) = (ab + αb + βa, αβ) ∀ a, b ∈ A, α, β ∈ F.
56 Normed Algebras
The algebra A is an ideal of A
u
. The unitization A
u
is commutative if and only if A is
commutative.
If A is a normed algebra over R or C then A
u
is a normed algebra, where
 
A
u : A
u
→R
+
; (a, α) → a
A
+[α[.
The unitization A
u
is complete if and only if A is complete, and A is a closed ideal of
A
u
.
Example 4.12. Let L
1
(R) denote the space of (equivalence classes of) complexvalued,
Lebesgueintegrable functions on the real line, with norm
 
1
: L
1
(R) →R
+
; f →
_
R
[f[.
This is a commutative Banach algebra when equipped with the convolution product:
f ⋆ g : R →R; t →
_
R
f(t −s)g(s) ds.
(By the theorems of Fubini and Tonelli, if f and g are integrable then this integral exists
almost everywhere and deﬁnes an element of L
1
(R).) This algebra lacks a unit; it is easy
to see that L
1
(R)
u
is isomorphic to the algebra given by adjoining the Dirac measure δ
0
:
by deﬁnition
(f ⋆ δ
0
)(t) = “
_
R
f(t −s)δ
0
(s) ds ” = f(t) ∀ t ∈ R
and δ
0
⋆ δ
0
= δ
0
.
Approximate Identities
Deﬁnition 4.13. Let A be a Banach algebra. An approximate identity for A is a net
(e
λ
)
λ∈Λ
⊆ A
1
such that
lim
λ
e
λ
a = a = a lim
λ
e
λ
∀ a ∈ A.
Example 4.14. The sequence (e
n
)
n1
is an approximate identity for L
1
(R), where
e
n
(x) =
n
√
2π
e
−
1
2
n
2
x
2
= ne
1
(nx) ∀ x ∈ R.
For a proof of this, see [17, Section 33.13 and Theorem 33.14] (but note that k
λ
(x) should
equal
_
λ
2π
e
−
1
2
λx
2
). Another possibility is (h
n
)
n1
, where
h
n
(x) =
1
π
n
1 +n
2
x
2
= nh
1
(nx) ∀ x ∈ R :
see [18, Section 9.7 and Theorem 9.10].
4.4. Completion 57
Example 4.15. Let A = L
1
[−π, π], which we can identify with the closed subalgebra
of L
1
(R) consisting of 2πperiodic functions. This has approximate identity (P
r
)
r∈[0,1)
,
where [0, 1) is directed in the usual manner and
P
r
(t) =
1
2π
∞
n=−∞
r
n
e
int
=
1
2π
1 −r
2
1 −2r cos t + r
2
=
1
2π
Re
1 +re
it
1 −re
it
.
[The function P is the Poisson kernel ; it occurs in the theory of harmonic functions on
the unit disc.] To prove that this is an approximate identity, note that if f : t → e
ikt
(where k ∈ Z) then
(P
r
⋆ f)(t) =
_
π
−π
1
2π
∞
n=−∞
r
n
e
in(t−s)
e
iks
ds =
∞
n=−∞
r
n
e
int
2π
_
π
−π
e
i(k−n)s
ds = r
k
f(t),
so (P
r
⋆ f) − f
1
= (1 − r
k
)f
1
→ 0 as r → 1−. The linear span of ¦e
ikt
: k ∈ Z¦
(the trigonometric polynomials) is dense in L
1
[−π, π] (an easy consequence of Fej´er’s
theorem [16, Theorem 30.4]) so a simple ε/2 argument gives this result for general f.
In fact, Fej´er’s theorem implies that if f is continuous and ε > 0 then there exists a
trigonometric polynomial p: t →
N
n=−N
a
n
e
int
such that f −p
∞
< ε/2, so that
[(P
r
⋆f)(t)−f(t)[ [(P
r
⋆f)(t)−p(t)[+[p(t)−f(t)[
_
N
n=−N
1−r
n
_
f
∞
+p−f
∞
< ε
if r is near enough to 1. Hence P
r
⋆ f → f uniformly in this case. [Compare this proof
to that given in [16, Section 10.36(2)].]
Completion
We have the notion of completion as for a normed space; an isometric isomorphism is
now required to be multiplicative, of course.
Theorem 4.16. If A is a normed algebra then there exists a Banach algebra
¯
A and
an isometric homomorphism i : A →
¯
A such that i(A) is a dense subalgebra of
¯
A, with
¯
A unital if A is; (
¯
A, i) is called a completion of A. Furthermore, if (B, j) and (C, k)
are completions of A then there exists an isometric isomorphism l : B → C such that
l
◦
j = k.
Proof
Let (
¯
A, i) be the completion of A considered as a normed space (see Theorem 1.14).
Deﬁne a product on
¯
A by setting
ab := lim
n→∞
i(a
n
b
n
) ∀ a, b ∈
¯
A
if (a
n
)
n1
, (b
n
)
∞
n=1
⊆ A are such that i(a
n
) → a and i(b
n
) → b as n → ∞; this limit
exists (as (i(a
n
b
n
))
∞
n=1
is Cauchy), is independent of the choice of sequences (a
n
)
∞
n=1
and
58 Normed Algebras
(b
n
)
∞
n=1
and the new product agrees with the old on i(A), i.e., i(a)i(b) = i(ab) for all
a, b ∈ A. (To see the ﬁrst claim, note that
i(a
n
b
n
) −i(a
m
b
m
) = i(a
n
b
n
−a
m
b
m
)
= a
n
b
n
−a
m
b
m

a
n
 b
n
−b
m
 +a
n
−a
m
 b
m

= i(a
n
) i(b
n
) −i(b
m
) +i(a
n
) −i(a
m
) i(b
m
)
→ 0
as m, n → ∞; proof of uniqueness is similar.) It is easy to verify that this product makes
¯
A an associative algebra (which is unital if A is) and the norm on
¯
A is submultiplicative
because
i(a
n
)i(b
n
) = i(a
n
b
n
) = a
n
b
n
 a
n
 b
n
 = i(a
n
) i(b
n
).
Hence
¯
A is a Banach algebra with dense subalgebra i(A).
If (B, j) and (C, k) are completions of A then there exists an isometric linear bijection
l : B → C such that l
◦
j = k, by Proposition 2.6. Furthermore, if a, b ∈ A then
l
_
j(a)j(b)
_
= l
_
j(ab)
_
= k(ab) = k(a)k(b) = l
_
j(a)
_
l
_
j(b)
_
,
which shows that l is multiplicative on j(A). Since j(A) is dense in B and multiplication
is jointly continuous, l is multiplicative on B and so is an isometric isomorphism.
Five Invertibility
From now on, A denotes a unital Banach algebra over C.
Deﬁnition 5.1. An element a ∈ A is invertible if there exists b ∈ A such that ab =
ba = 1. [If such an inverse exists, it is unique.] The collection of invertible elements
in A is denoted G(A); this is a group. [This last claim is obvious once we recall that
(ab
−1
)
−1
= ba
−1
if a and b are invertible.]
Proposition 5.2. Let a ∈ A be such that a < 1. Then 1 −a ∈ G(A) and
(1 −a)
−1
=
∞
n=0
a
n
, (5.1)
where this series converges in the norm topology. Hence G(A) is an open subset of A;
furthermore a → a
−1
is a homeomorphism of G(A).
Proof
Since a < 1, the Neumann series (5.1) is absolutely convergent, so convergent.
Furthermore,
lim
n→∞
_
n
j=0
a
j
_
(1 −a) = (1 −a) lim
n→∞
n
j=0
a
j
= lim
n→∞
1 −a
n+1
= 1,
so the sum of this series is an inverse for (1 −a), as claimed.
If a ∈ G(A) let b ∈ A and note that
b = a −(a −b) = a
_
1 −a
−1
(a −b)
_
∈ G(A)
if a
−1
(a −b) < 1, e.g., if b ∈ B
A
a
−1
−1
(a); this shows that G(A) is open.
Finally, if h ∈ A is such that h <
1
2
a
−1

−1
then a
−1
h <
1
2
, which implies that
a + h = a(1 +a
−1
h) ∈ G(A) and
(a + h)
−1
−a
−1
= a
−1
_
(1 +a
−1
h)
−1
−1
_
= a
−1
∞
n=1
(−a
−1
h)
n
has norm at most
a
−1
 a
−1
h/(1 −a
−1
h) 2a
−1

2
h
(using the fact that 
∞
n=1
x
n
 x/(1 −x) if x < 1). Hence (a +h)
−1
→ a
−1
as
h → 0, as required.
59
60 Invertibility
The Spectrum and Resolvent
Deﬁnition 5.3. For all a ∈ A the spectrum of a is
σ(a) := ¦λ ∈ C : λ1 −a / ∈ G(A)¦.
The resolvent set of a is ρ(a) := C ¸ σ(a), the complement of the spectrum of a. The
resolvent of a is the function deﬁned on ρ(a) by λ → r
λ
(a) := (λ1 −a)
−1
.
Theorem 5.4. The resolvent satisﬁes the resolvent equation,
r
λ
(a) −r
µ
(a) = −(λ −µ)r
λ
(a)r
µ
(a) ∀ λ, µ ∈ ρ(a),
and is strongly holomorphic on ρ(a), which is an open set. The spectrum is a compact,
nonempty subset of C
a
.
Proof
For the ﬁrst claim, note that if λ, µ ∈ ρ(a) then
(λ1 −a)
−1
−(µ1 −a)
−1
= (λ1 −a)
−1
(1 −(λ1 −a)(µ1 −a)
−1
)
= (λ1 −a)
−1
_
(µ1 −a) −(λ1 −a)
_
(µ1 −a)
−1
= −(λ −µ)(λ1 −a)
−1
(µ1 −a)
−1
.
From this equation and Proposition 5.2 it follows that
r
λ
(a) −r
µ
(a)
λ −µ
= −r
λ
(a)r
µ
(a) → −r
µ
(a)
2
as λ → µ: the resolvent is strongly holomorphic on ρ(a) (recall Deﬁnition 3.13). To
see that this set is open, let f : C → A; λ → λ1 − a and note that f is continuous, so
f
−1
(G(A)) = ρ(a) is open.
If a ∈ A and [λ[ > a then λ
−1
a < 1, so by Proposition 5.2 we have that
λ1 − a = λ(1 − λ
−1
a) is invertible. Hence σ(a) ⊆ ¦λ ∈ C : [λ[ a¦. By the
HeineBorel theorem and the fact that σ(a) = C ¸ ρ(a) is closed, σ(a) is compact.
Suppose for contradiction that σ(a) is empty, so that the resolvent is a strongly
holomorphic function deﬁned on the whole of C. If [λ[ > a then
r
λ
(a) = λ
−1
(1 −λ
−1
a)
−1
=
∞
n=0
λ
−(n+1)
a
n
(5.2)
and so
r
λ
(a) [λ[
−1
/
_
1 −λ
−1
a
_
=
_
[λ[ −a
_
−1
;
this shows that r
λ
(a) → 0 as λ → ∞. In particular the resolvent is bounded, so constant
(by Theorem 3.14) and therefore equal to 0; this is the desired contradiction.
5.2. The GelfandMazur Theorem 61
The GelfandMazur Theorem
It is rather surprising that a theorem with as short a proof as the following has such
important consequences. Note that commutativity forms no part of the hypotheses but
is part of the conclusion.
Theorem 5.5. (GelfandMazur) If A is a unital Banach algebra over C in which
every nonzero element is invertible then A is isometrically isomorphic to C.
Proof
Let a ∈ A; since σ(a) is nonempty and
λ ∈ σ(a) ⇔ λ1 −a / ∈ G(A) ⇔ λ1 −a = 0 ⇔ a = λ1,
we see that for all a ∈ A there exists λ
a
∈ C such that a = λ
a
1. The map a → λ
a
is the
desired isomorphism; note that a = λ
a
1 = [λ
a
[.
The SpectralRadius Formula
Deﬁnition 5.6. The spectral radius of a ∈ A is the radius of the smallest disc about
the origin that contains the spectrum of a:
ν(a) := inf¦r 0 : σ(a) ⊆ C
r
¦ = sup¦[λ[ : λ ∈ σ(a)¦.
We recall the spectral mapping theorem for polynomials; it is stated in b4 for bounded
operators on a Banach space but its proof holds in any Banach algebra
Theorem 5.7. Let a ∈ A and suppose that p(z) ∈ C[z] is a complex polynomial. Then
σ
_
p(a)
_
= p
_
σ(a)
_
, i.e.,
σ
_
p(a)
_
:= ¦λ ∈ C : λ1 −p(a) / ∈ G(A)¦ = ¦p(λ) : λ1 −a / ∈ G(A)¦.
Proof
This follows the same pattern as the proof which may be found in Dr VincentSmith’s
b4 notes ([25, Theorem 5.2.11]).
The following theorem gives the BeurlingGelfand spectralradius formula. (According
to [11, p.525], Beurling led Sweden’s eﬀort to crack the Enigma code during the second
world war.)
Theorem 5.8. If a ∈ A then the sequence
_
a
n

1/n
_
n1
is convergent and
ν(a) = inf
n1
a
n

1/n
= lim
n→∞
a
n

1/n
.
Proof
Firstly, the spectral mapping theorem for polynomials implies that
ν(a)
n
= ν(a
n
) a
n
 ∀ n 1. (5.3)
62 Invertibility
Next, f : λ → (λ1−a)
−1
is continuous on ρ(a), so M
r
:= sup
_
(λ1−a)
−1
 : [λ[ = r
_
< ∞
for all r > ν(a). Let φ ∈ A
∗
and note that f is strongly holomorphic, so φ
◦
f is
holomorphic, on S = ¦λ ∈ C : [λ[ > ν(a)¦, with
(φ
◦
f)(λ) =
∞
n=0
φ(a
n
)λ
−(n+1)
_
[λ[ > a
_
by (5.2), so for all λ ∈ S, by the uniqueness of Laurent series. Furthermore, by the
integral formula for Laurent coeﬃcients,
[φ(a
n
)[ =
¸
¸
¸
¸
1
2πi
_
{λ:λ=r}
(φ
◦
f)(λ)λ
n
dλ
¸
¸
¸
¸
r
n+1
φM
r
and so a
n
 r
n+1
M
r
(by Theorem 3.10) for all r > ν(a). This and (5.3) yield
ν(a) a
n

1/n
ν(rM
r
)
1/n
→ r
as n → ∞; the result follows. (If α a
n
for all n and a
n
→ α then α = inf a
n
; that
a
n

1/n
→ ν(a) is immediate.)
The following theorem allows us to restrict our attention to commutative algebras when
considered certain questions about the spectrum. Zorn’s lemma allows us to prove that
any commutative set in A is contained in a maximal commutative subalgebra.
Theorem 5.9. Let B be a maximal commutative subalgebra of A. If a ∈ B then
σ
B
(a) := ¦λ ∈ C : λ1 −a / ∈ G(B)¦ = σ
A
(a)
_
= ¦λ ∈ C : λ1 −a / ∈ G(A)¦
_
.
Proof
Clearly G(B) ⊆ G(A), so we need to prove that if λ1 −a is invertible in A then it is
invertible in B. Suppose that λ1 −a ∈ G(A), let b = (λ1 −a)
−1
and let c ∈ B; note that
bc = bc(λ1 −a)b = b(λ1 −a)cb = cb
and so b ∈ B (else B is not maximal) as required.
Example 5.10. If S is a subset of the unital Banach algebra A then
S
c
:= ¦a ∈ A : sa = as for all s ∈ S¦
is the commutant (or centraliser) of S. It is readily veriﬁed that S
c
is a unital, closed
subalgebra of A, that if S ⊆ T ⊆ A then T
c
⊆ S
c
and that S is commutative if and only if
S ⊆ S
c
. Furthermore, if S is commutative then the double commutant (or bicommutant)
S
cc
= (S
c
)
c
is a commutative subalgebra of A containing S. It is an exercise to prove
that a commutative subalgebra B is maximal if and only if B = B
c
, and that the
proof of Theorem 5.9 goes through with the weaker hypothesis that B is a commutative
subalgebra of A containing ¦a¦
cc
.
5.4. Exercises 6 63
Exercises 6
Exercise 6.1. Prove that the vector space L
1
(R) is a commutative Banach algebra
when equipped with the convolution product and the norm f
1
:=
_
[f[. [To show
associativity, use the convolution theorem and the inversion theorem for the Fourier
transform.] Prove further that this algebra is not unital. [You may assume that the
functions f
n
: x → (n/
√
2π) exp(−
1
2
n
2
x
2
) are such that f
n
⋆ g −g
1
→ 0 as n → ∞ for
all g ∈ L
1
(R).]
Exercise 6.2. Let X be a Hausdorﬀ, locally compact space. Prove that C
0
(X)
u
, the
unitization of the algebra of continuous functions on X that vanish at inﬁnity, is topo
logically isomorphic to C(
˙
X), the algebra of continuous functions on
˙
X, the onepoint
compactiﬁcation of X.
Exercise 6.3. Let A = C[z] denote the unital algebra of complex polynomials and let
p := sup¦[p(α)[ : [α[ 1¦ for all p ∈ A. Show that (A,  ) is a unital, normed
algebra which is not complete. [For the latter statement, consider invertibility and the
polynomials p
n
(z) = 1 +z/n.]
Exercise 6.4. Let A be a (nonunital) Banach algebra such that every element is nil
potent (i.e., for all a ∈ A there exists n ∈ N such that a
n
= 0). Prove that A is
uniformly nilpotent: there exists N ∈ N such that a
N
= 0 for all a ∈ A. [Consider the
decomposition A =
n∈N
¦a ∈ A : a
n
= 0¦.]
Exercise 6.5. Let A be a unital Banach algebra over C and let e
a
:=
∞
n=0
a
n
/n! for all
a ∈ A. Prove that e
a+b
= e
a
e
b
if a and b commute. Deduce that e
a
is invertible. Prove
further that f : λ → e
λa
is holomorphic everywhere, with f
′
(λ) = af(λ) = f(λ)a, for all
a ∈ A.
Exercise 6.6. Let A be a unital Banach algebra over C and let a, b ∈ A. Use the
identity (ab)
n
= a(ba)
n−1
b to prove that ab and ba have the same spectral radius.
Exercise 6.7. Let A be a unital Banach algebra over C. Suppose that there exists
K > 0 such that a Kν(a) for all a ∈ A, where ν(a) denotes the spectral radius of a.
Prove that A is commutative. [Let a, b ∈ A and consider the function g : λ → e
λa
be
−λa
.]
Exercise 6.8. Let A be a Banach algebra and suppose that (x
p
)
p∈P
, (y
q
)
q∈Q
⊆ A are
absolutely summable. Prove that
p∈P
q∈Q
x
p
y
q
=
(p,q)∈P×Q
x
p
y
q
=
q∈Q
p∈P
x
p
y
q
.
Six Characters and Maximal Ideals
Proposition 6.1. Let A be a unital algebra. If φ: A →C is a nonzero algebra homo
morphism then φ(1) = 1 and φ(a) ,= 0 for all a ∈ G(A). Furthermore, if A is a Banach
algebra then φ ∈ A
∗
with φ = 1.
Proof
Note that φ(a) = φ(1a) = φ(1)φ(a) for all a ∈ A; it cannot be the case that φ(a) = 0
for all a ∈ A (as φ ,= 0) and so φ(1) = 1. Thus 1 = φ(1) = φ(aa
−1
) = φ(a)φ(a
−1
) for all
a ∈ G(A), whence φ(a) ,= 0 if a is invertible.
To see the statement about the norm of φ, note that φ(1) = 1 (so φ 1) and
suppose for contradiction that there exists a ∈ A such that a 1 and [φ(a)[ > 1. Let
b = φ(a)
−1
a, so that b < 1 and 1 −b ∈ G(A), but φ(1 −b) = 1 −φ(a)
−1
φ(a) = 0, the
desired contradiction.
From now on, A is a commutative unital Banach algebra over C, unless otherwise
speciﬁed.
Deﬁnition 6.2. A character of A is a nonzero algebra homomorphism from A to C.
The collection of all characters of A is denoted by Φ(A). A maximal ideal of A is an
ideal I that is proper (I ,= A) and maximal with respect to inclusion: if J is an ideal
such that J ⊇ I then either J = A or J = I.
Proposition 6.3. Every proper ideal of A contains no invertible element of A and is
contained in a maximal ideal. A maximal ideal of A is closed.
Proof
If I is an ideal of A such that I ∩ G(A) ,= ∅ then 1 = a
−1
a ∈ I for some a ∈ G(A)
and so I = A, i.e., I is not proper.
To see the next claim, let I be a proper ideal of A, let F denote the collection of
proper ideals of A that contain I, preordered by inclusion, and apply Zorn’s lemma (note
that no proper ideal contains an invertible element of A, so neither does the union of a
chain of such).
Finally, let I be a maximal ideal; since the closure of an ideal is an ideal (an easy
exercise), either
¯
I = A or
¯
I = I; if the former then
¯
I ∩ G(A) ,= ∅, but I ⊆ A ¸ G(A)
implies that
¯
I ⊆ A ¸ G(A) since G(A) is open. Hence I =
¯
I, as claimed.
65
66 Characters and Maximal Ideals
Before we can prove the fundamental connexion between characters and maximal ideals,
we need a fact from algebra.
Lemma 6.4. Let A be a unital commutative algebra (or even a commutative ring with
identity) and suppose that I is an ideal of A. Then A/I is a ﬁeld if and only if I is
maximal.
Proof
Suppose that I is maximal and let a ∈ A ¸ I. Then aA + I is an ideal in A that
properly contains I, so aA + I = A. Hence there exists b ∈ A and c ∈ I such that
ab + c = 1, whence [a][b] = [1] (where [a] = a + I et cetera) and A/I is a ﬁeld.
If I is not maximal then there exists a ∈ A¸ I such that aA+I ,= A, so 1 / ∈ aA+I
and there exists no b ∈ A such that [a][b] = [1]. Hence [a] is not invertible and A/I is
not a ﬁeld.
Theorem 6.5. The map φ → ker φ is a bijection between Φ(A) and the set of all
maximal ideals of A.
Proof
Since A/ ker φ
∼
= imφ = C, A/ ker φ is a ﬁeld for all φ ∈ Φ(A) and hence ker φ is
maximal, by Lemma 6.4.
Suppose that φ, ψ ∈ Φ(A) are such that ker φ = ker ψ. For all a ∈ A we have that
φ(a − φ(a)1) = 0, so a − φ(a)1 ∈ ker φ = ker ψ and 0 = ψ(a − φ(a)1) = ψ(a) − φ(a).
Hence φ = ψ and φ → ker φ is injective.
Finally, if I is a maximal ideal of A then A/I is a ﬁeld (by Lemma 6.4) and a Banach
algebra (with respect to its quotient norm), so A/I
∼
= C, by the GelfandMazur theorem;
let i : A/I →C denote this isomorphism. Then φ = i
◦
π is the desired character, where
π: A → A/I is the quotient map.
Characters and the Spectrum
Recall that A is a commutative, unital Banach algebra over C, unless otherwise stated.
Corollary 6.6. Let a ∈ A. Then
(i) a ∈ G(A) if and only if φ(a) ,= 0 for all φ ∈ Φ(A);
(ii) σ(a) = ¦φ(a) : φ ∈ Φ(A)¦;
(iii) ν(a) = sup¦[φ(a)[ : φ ∈ Φ(A)¦.
Proof
For (i), note that
a ∈ G(A) ⇔ A = Aa
⇔ Aa is not contained in a maximal ideal of A
⇔ Aa ,⊆ ker φ ∀ φ ∈ Φ(A)
⇔ φ(a) ,= 0 ∀ φ ∈ Φ(A).
6.2. The Gelfand Topology 67
The other two claims are immediate.
Corollary 6.7. If A is a (not necessarily commutative) unital Banach algebra and
a, b ∈ A commute, i.e., ab = ba, then
ν(a + b) ν(a) + ν(b) and ν(ab) ν(a)ν(b).
Proof
Let B be a maximal commutative subalgebra containing a and b. By Theorem 5.9,
σ
B
(a + b) = σ
A
(a + b) et cetera, and so
ν(a + b) = sup¦[φ(a + b)[ : φ ∈ Φ(B)¦
sup¦[φ(a)[ : φ ∈ Φ(B)¦ + sup¦[φ(b)[ : φ ∈ Φ(B)¦ = ν(a) + ν(b).
The other claim is proved in the same manner.
Recall that an element a of a ring is said to be nilpotent if a
n
= 0 for some n ∈ N. If
a ∈ A is nilpotent then σ(a) = 0, by the spectral mapping theorem. More generally, we
have the following deﬁnition.
Deﬁnition 6.8. An element a ∈ A is quasinilpotent if ν(a) = 0. The set of all quasi
nilpotent elements in A is the Jacobson radical of A, denoted by J(A). An algebra is
semisimple if J(A) = ¦0¦.
Proposition 6.9. The Jacobson radical of A is an ideal; in fact,
J(A) =
φ∈Φ(A)
ker φ,
the intersection of all maximal ideals in A.
Proof
Since
ν(a) = sup¦[φ(a)[ : φ ∈ Φ(A)¦,
we have that ν(a) = 0 if and only if a ∈ ker φ for all φ ∈ Φ(A).
The Gelfand Topology
Lemma 6.10. The character space Φ(A) is a compact, Hausdorﬀ space when equipped
with the Gelfand topology, i.e., the restriction of σ(A
∗
, A), the weak* topology on A
∗
,
to Φ(A). Equivalently, it is the coarsest topology to make the maps ˆ a[
Φ(A)
: φ → φ(a)
continuous (for all a ∈ A).
Proof
Recall that A
∗
1
is a compact, Hausdorﬀ space, by Theorem 3.22, and
Φ(A) = ¦φ ∈ A
∗
1
: φ(1) = 1, φ(ab) = φ(a)φ(b) ∀ a, b ∈ A¦
= A
∗
1
∩
ˆ
1
−1
(1) ∩
a,b∈A
_
´
ab −
ˆ
a
ˆ
b
_
−1
(0)
68 Characters and Maximal Ideals
is a closed subset of A
∗
1
(since ˆ a: A
∗
→C; φ → φ(a) is continuous for all a ∈ A). [Note
that the condition φ(1) = 1 is necessary to rule out the zero homomorphism.] This gives
the result.
The Representation Theorem
From now on it is more convenient to let
ˆ a: Φ(A) →C; φ → φ(a).
That is, ˆ a is the restriction to Φ(A) of the map φ → φ(a) on A
∗
.
Theorem 6.11. If A is a commutative unital Banach algebra then the Gelfand transform
ˆ : A → C
_
Φ(A)
_
; a → ˆ a
is a normdecreasing homomorphism. Its kernel is J(A) and its image
ˆ
A is a subalgebra
of C
_
Φ(A)
_
that separates the points of Φ(A).
Proof
The Gelfand transform is a homomorphism because characters are; for example,
´
ab(φ) = φ(ab) = φ(a)φ(b) =
_
ˆ
a
ˆ
b
_
(φ) ∀ φ ∈ Φ(A), a, b ∈ A.
Furthermore,
ˆ a
∞
= sup¦[φ(a)[ : φ ∈ Φ(A)¦ = ν(a) a ∀ a ∈ A,
so a → ˆ a is normdecreasing; this calculation also shows that the kernel is as claimed.
Finally, if φ, ψ ∈ Φ(A) are such that ˆ a(φ) = ˆ a(ψ) for all a ∈ A then φ = ψ (by deﬁnition),
so
ˆ
A separates the points of Φ(A).
Gelfand theory reaches its peak when the algebra is equipped with an involution, i.e., a
conjugatelinear map a → a
∗
such that
(ab)
∗
= b
∗
a
∗
and (a
∗
)
∗
= a ∀ a, b ∈ A.
If the involution satisﬁes a
∗
a = a
2
for all a ∈ A then we have a C
∗
algebra: for the
theory of such, see [8, Chapter 4 et seq.] or [19, Chapter 11 et seq.].
Examples
Example 6.12. If X is a compact, Hausdorﬀ space and A = C(X) is the algebra of
continuous functions on X then Φ(A) = ¦ǫ
x
: x ∈ X¦, where
ǫ
x
: C(X) →C; φ → φ(x)
6.4. Examples 69
is the evaluation homomorphism at x.
Example 6.13. If D = B
C
1
(0) = ¦z ∈ C : [z[ < 1¦ and
A(D) = ¦f ∈ C(
¯
D) : f[
D
is holomorphic¦
is the disc algebra then A(D) is a Banach algebra (when equipped with the supremum
norm: recall that uniform limits of holomorphic functions are holomorphic). In this
case the character space of A(D) is again just the set of evaluation homomorphisms,
¦ǫ
z
: z ∈
¯
D¦.
Example 6.14. Suppose that n ∈ N and let Z/nZ denote the quotient group of integers
with addition modulo n. The ﬁnitedimensional Banach space
ℓ
1
(Z/nZ) =
_
(x
j
)
j∈Z
: x
j
= x
n+j
∀ j ∈ Z
_
=
_
(x
[j]
)
[j]∈Z/nZ
_
becomes a commutative, unital Banach algebra when equipped with the convolution
product
(x ⋆ y)
[j]
=
[k]∈Z/nZ
x
[k]
y
[j−k]
∀ x, y ∈ ℓ
1
(Z/nZ).
Let
δ := [j] →
_
1 [j] = [1],
0 [j] ,= [1]
and note that
x =
[j]∈Z/nZ
x
[j]
δ
j
∀ x ∈ ℓ
1
(Z/nZ).
In particular, any φ ∈ Φ
_
ℓ
1
(Z/nZ)
_
is determined by λ = φ(δ) and λ is an nth root of
unity as λ
n
= φ(δ
n
) = φ(1) = 1. Conversely, each λ ∈ ¦ω
j
: j = 0, 1, . . . , n − 1¦, where
ω = exp(2πi/n), corresponds to a character, via
x →
[j]∈Z/nZ
x
[j]
λ
j
.
Hence the Gelfand theory of ℓ
1
(Z/nZ) corresponds to the theory of the discrete Fourier
transform.
Before investigating the Gelfand theory of L
1
(R) we need a couple of preliminary results.
(The fact that L
1
(R)
∗
= L
∞
(R) gives a simple proof of the following; as this has not
been established the lemma is proved directly.)
Lemma 6.15. If f ∈ L
1
(R
2
) and φ ∈ L
1
(R)
∗
then
φ
_
r →
_
R
f(r, s) ds
_
=
_
R
φ
_
r → f(r, s)
_
ds.
Proof
Note ﬁrst that Fubini’s theorem gives that f(, s): r → f(r, s) ∈ L
1
(R) for almost
every s ∈ R and that
_
R
f(, s) ds: r →
_
R
f(r, s) ds ∈ L
1
(R), so the quantities above are
70 Characters and Maximal Ideals
well deﬁned; measurability of s → φ
_
f(, s)
_
will follow from the below. If f = χ
A×B
(where A, B ⊆ R are bounded intervals) then s → φ
_
f(, s)
_
= φ(χ
A
)χ
B
and
φ
_
_
R
f(, s) ds
_
= φ(χ
A
_
R
χ
B
(s) ds) =
_
R
φ(χ
A
)χ
B
(s) ds =
_
R
φ
_
f(, s)
_
ds,
as claimed; linearity gives the result for all f ∈ L
step
(R
2
). For general f ∈ L
1
(R
2
) take
(f
n
)
n1
⊆ L
step
(R
2
) such that f
n
→ f in L
1
(R
2
). Note that
_
R
f
n
(, s) ds →
_
R
f(, s) ds
in L
1
(R) as
_
R
¸
¸
_
R
f
n
(r, s) ds −
_
R
f(r, s) ds
¸
¸
dr
_
R
_
R
[f
n
(r, s) −f(r, s)[ ds dr = f
n
−f
1
,
so
φ
_
_
R
f(, s) ds
_
= lim
n→∞
φ
_
_
R
f
n
(, s) ds
_
= lim
n→∞
_
R
φ
_
f
n
(, s)
_
ds =
_
R
φ
_
f(, s)
_
ds.
The last equality holds as
¸
¸
¸
_
R
φ
_
f
n
(, s)
_
ds −
_
R
φ
_
f(, s)
_
ds
¸
¸
¸
_
R
φ (f
n
−f)(, s)
1
ds = φ f
n
−f
1
.
(This calculation also shows that s → φ
_
f
n
(, s)
_
→ s → φ
_
f(, s)
_
in L
1
(R), so has
there is a subsequence such that φ
_
f
n
k
(, s)
_
→ φ
_
f(, s)
_
for almost every s ∈ R. In
particular, s → φ
_
f(, s)
_
is the almosteverywhere limit of a sequence of measurable
functions, so is measurable.)
Lemma 6.16. If χ: R → C is a continuous, bounded function such that χ(0) = 1 and
χ(s + t) = χ(s)χ(t) for all s, t ∈ R then there exists α ∈ R such that χ(t) = e
iαt
for all
t ∈ R.
Proof
Since χ(0) = 1 and χ is continuous, there exists r > 0 such that c =
_
r
0
χ(x) dx ,= 0.
Hence
cχ(t) =
_
r
0
χ(x + t) dx =
_
t+r
t
χ(y) dy (t ∈ R)
and χ is diﬀerentiable; diﬀerentiating the equation χ(s +t) = χ(s)χ(t) with respect to s
at 0 yields χ
′
(t) = χ
′
(0)χ(t) and so χ(t) = exp(dt), where d = χ
′
(0). Since χ is bounded
we must have that d is purely imaginary, as claimed.
Example 6.17. Recall that L
1
(R) is a nonunital Banach algebra when equipped with
the convolution product; let A = L
1
(R)
u
denote its unitization. If φ ∈ Φ(A) then either
ker φ = L
1
(R) (i.e., φ(α1 + f) = α for all α ∈ C and f ∈ L
1
(R) ) or there exists
f ∈ L
1
(R) such that φ(f) = 1. Suppose that the latter holds; as C
00
(R) is dense in
L
1
(R) we may assume that f is continuous and has compact support. (If g ∈ C
00
(R) is
such that f − g
1
1/2 then [φ(g)[ [φ(f)[ − [φ(f − g)[ 1 − φ f − g
1
1/2;
now replace f by g/φ(g).) If g ∈ L
1
(R) then Lemma 6.15 gives that
φ(g) = φ(f)φ(g) = φ(f ⋆ g) = φ
_
s →
_
R
f
t
(s)g(t) dt
_
=
_
R
φ(f
t
)g(t) dt,
6.4. Examples 71
where f
t
(s) = f(s −t) for all s, t ∈ R. Deﬁne
χ: R →C; t → φ(f
t
)
and note that, since f
t

1
= f
1
for all t ∈ R, [χ(t)[ φ f
t

1
= f
1
, i.e., χ is
bounded. Furthermore, χ(0) = φ(f
0
) = φ(f) = 1 and
[χ(t + h) −χ(t)[ = [φ(f
t+h
−f
t
)[ f
t+h
−f
t

1
= f
h
−f
1
→ 0
as h → 0; this follows from the continuous form of the dominatedconvergence theorem.
Note also that, if g, h ∈ L
1
(R),
_
g
s+t
⋆ h
_
(r) =
_
R
g
_
r −p −(s + t)
_
h(p) dp =
_
R
g(r −q −s)h(q −t) dq = (g
s
⋆ h
t
)(r)
and so
χ(s + t) = φ(f
s+t
) = φ(f
s+t
)φ(f) = φ(f
s+t
⋆ f) = φ(f
s
⋆ f
t
) = φ(f
s
)φ(f
t
) = χ(s)χ(t)
for all s, t ∈ R. By Lemma 6.16 we must have that χ(t) = e
−ist
for all t ∈ R, where
s ∈ R is such that e
−is
= χ(1). To see that s ∈ R is independent of the choice of f, let
g ∈ L
1
(R) be such that φ(g) = 1 and note that
χ(t) = φ(f
t
)φ(g) = φ(f
t
⋆ g) = φ(f ⋆ g
t
) = φ(f)φ(g
t
) = φ(g
t
) ∀ t ∈ R.
Hence
φ(g) =
_
R
g(t)e
−ist
dt ∀ g ∈ L
1
(R)
and the Gelfand transform corresponds the classical Fourier transform for L
1
(R); the
fact that α1 + f → α +
_
R
f(t)e
−ist
dt is a character for all s ∈ R is an immediate
consequence of Fubini’s theorem.
Example 6.18. If L
1
(R
+
) is equipped with the convolution product
(f ⋆ g)(t) =
_
t
0
f(t −s)g(s) ds ∀ f, g ∈ L
1
(R
+
)
then it becomes a Banach algebra. The Gelfand theory here corresponds to the Laplace
transform.
Example 6.19. Let T = ∂D = ¦z ∈ C : [z[ = 1¦ and A = ¦f ∈ C(
¯
D) : f[
T
∈ A(D)[
T
¦:
A consists of those continuous functions on the closed unit disc
¯
D that agree on the unit
circle T with a continuous function on
¯
D that is holomorphic in D. The character space
of A is homeomorphic to the sphere S
2
:= ¦(x, y, z) ∈ R
3
: x
2
+ y
2
+ z
2
= 1¦.
72 Characters and Maximal Ideals
Exercises 7
Exercise 7.1. Let A = C(X), where X is a compact, Hausdorﬀ space. Prove that the
map ǫ: X → Φ(A); x → ǫ
x
is a homeomorphism, where ǫ
x
(f) = f(x) for all x ∈ X and
f ∈ C(X).
Exercise 7.2. Prove that if A is a unital Banach algebra generated by a single element
(i.e., there exists a ∈ A such that ¦p(a) : p(z) ∈ C[z]¦ is dense in A) then Φ(A) is
homeomorphic to σ(a). [Consider φ → φ(a).] Deduce that Φ(A) is homeomorphic to
¯
D := ¦z ∈ C : [z[ 1¦ if A = A(D) is the disc algebra.
Exercise 7.3. Let A be a unital Banach algebra that is generated by one element,
a, and let λ ,∈ σ(a). Show there exists a sequence of polynomials (p
n
)
n1
such that
p
n
(z) → (λ − z)
−1
uniformly for all z ∈ σ(a). [Hint: (λ1 − a)
−1
∈ A.] Deduce that the
complement of σ(a) is connected. [Prove that if C is a bounded, maximally connected
component of C¸ σ(a) then C is open and then employ the maximummodulus theorem
[16, Theorem 5.20].]
Exercise 7.4. Let A be a commutative, unital Banach algebra. Prove that the Gelfand
transform on A is isometric if and only if a
2
 = a
2
for all a ∈ A.
Exercise 7.5. Let A be a Banach algebra and B a semisimple, commutative, unital
Banach algebra. Prove that if φ: A → B is a homomorphism then φ is continuous. [Use
the closedgraph theorem.]
Exercise 7.6. Let A = C
1
[0, 1], equipped with the norm f := f
∞
+ f
′

∞
. Prove
that A is a semisimple, commutative, unital Banach algebra and ﬁnd its character space.
Prove that I = ¦f ∈ A : f(0) = f
′
(0) = 0¦ is a closed ideal in A such that A/I is a
twodimensional algebra with onedimensional radical. What do you notice about A and
A/I?
Exercise 7.7. Prove that the Banach space ℓ
1
(Z) is a commutative, unital Banach
algebra when equipped with the multiplication
a ⋆ b: Z →C; n →
m∈Z
a
m
b
n−m
_
a, b ∈ ℓ
1
(Z)
_
.
[You may assume that
m∈Z
n∈Z
a
m
b
n
=
n∈Z
m∈Z
a
m
b
n
for all a, b ∈ ℓ
1
(Z).]
Exercise 7.8. Let δ ∈ ℓ
1
(Z) be such that δ
1
= 1 and δ
n
= 0 if n ,= 1. Prove that a =
n∈Z
a
n
δ
n
for all a ∈ ℓ
1
(Z). Deduce that the character space of ℓ
1
(Z) is homeomorphic
to T := ¦z ∈ C : [z[ = 1¦ and (with this identiﬁcation) the Gelfand transform on ℓ
1
(Z)
is the map
Γ: ℓ
1
(Z) → C(T); Γ(a)(λ) =
n∈Z
a
n
λ
n
∀ λ ∈ T, a ∈ ℓ
1
(Z).
Exercise 7.9. Let f : T →C be continuous and such that
n∈Z
[
ˆ
f(n)[ < ∞, where
ˆ
f(n) :=
1
2π
_
π
−π
f(e
it
)e
−int
dt (n ∈ Z).
6.5. Exercises 7 73
Prove that if f(z) ,= 0 for all z ∈ T then g = 1/f : T → C; z → 1/f(z) satisﬁes
n∈Z
[ˆ g(n)[ < ∞. [This result is known as Wiener’s lemma.]
Exercise 7.10. Let T = ∂D = ¦z ∈ C : [z[ = 1¦ and A = ¦f ∈ C(
¯
D) : f[
T
∈ A(D)[
T
¦:
A consists of those continuous functions on the closed unit disc
¯
D that agree on the unit
circle T with a continuous function on
¯
D that is holomorphic in D.
[A corollary of the maximummodulus theorem [16, Theorem 5.20] will be useful: if
f ∈ A(D) then f
∞
:= sup¦[f(z)[ : [z[ 1¦ = sup¦[f(z)[ : [z[ = 1¦ =: f[
T

∞
.]
(i) Show that A is a Banach algebra when equipped with the supremum norm.
(ii) Prove that I = ¦f ∈ A : f[
T
= 0¦ is a closed ideal in A and that A = A(D) ⊕ I.
[Consider a suitable map j : A → A(D).]
(iii) Prove that
i : I → C
0
(D); f → f[
D
is an isometric isomorphism. Deduce that I
u
is topologically isomorphic to C(
˙
D),
where
˙
D is the onepoint compactiﬁcation of D.
(iv) Prove that if φ ∈ Φ(A) is such that ker φ ⊇ I then φ =
˜
φ
◦
j, where
˜
φ ∈ Φ(A(D)).
Deduce that φ = ε
z
◦
j for some z ∈
¯
D (where ε
z
: A(D) →C; f → f(z) ).
(v) Let
¯
D
1
and
¯
D
2
be two copies of the unit disc and let S
2
=
¯
D
1
∪
¯
D
2
_
∼ be the sphere
obtained by identifying each point on T
1
= ∂D
1
with the corresponding point on
T
2
= ∂D
2
. Deﬁne
T : S
2
→ Φ(A); z →
_
ε
z
◦
j (z ∈
¯
D
1
),
ε
z
(z ∈
¯
D
2
)
and prove that this is a welldeﬁned, continuous injection.
(vi) Prove that Φ(A) is homeomorphic to the sphere S
2
.
A Tychonov via Nets
Compactness in metric spaces can be characterised by the behaviour of sequences (viz
every sequence having a convergent subsequence: the BolzanoWeierstrass property).
Furthermore, a proof of Tychonov’s theorem for countable products of metric spaces can
be given using sequences, together with construction of a diagonal subsequence (of a
sequence of sequences!).
The astute reader may suspect that compactness in general topological spaces is
equivalent to a property involving nets and that a proof of Tychonov’s theorem can be
given using subnets, whatever they may be. This suspicion is well founded; ﬁrst we need
to deﬁne a subnet.
Deﬁnition A.1. Let (A, ) be a directed set and (x
a
)
a∈A
a net. A subnet of (x
a
)
a∈A
is
a net (y
b
)
b∈B
(where (B, ≪) is a directed set) and a map j : B → A such that
(i) y
b
= x
j(b)
for all b ∈ B;
(ii) for all a
0
∈ A there exists b
0
∈ B such that if b ≫ b
0
then j(b) a
0
.
Deﬁnition A.2. A net (x
a
)
a∈A
is eventually in a set S if there exists a
0
∈ A such that
x
a
∈ S for all a a
0
. (A net in a topological space converges to a point if it is eventually
in every open set containing that point.) A net (x
a
)
a∈A
is frequently in S if for all a
0
∈ A
there exists a ∈ A such that a a
0
and x
a
∈ A. (Note that a net is not frequently in a
set if it is eventually in its complement et cetera.) If X is a topological space then x ∈ X
is an accumulation point for a net if that net is frequently in every open set containing
x.
Lemma A.3. Let (x
a
)
a∈A
be a net in X and let F be a nonempty family of subsets of
X that forms a directed set under reverse inclusion, such that (x
a
)
a∈A
is frequently in S
for all S ∈ F. There is a subnet of (x
a
)
a∈A
that is eventually in every S ∈ F.
Proof
Let
B = ¦(c, S) ∈ AF : x
c
∈ S¦,
ordered by setting
(c, S) ≪ (d, T) ⇐⇒ c d and S ⊇ T.
75
76 Tychonov via Nets
If (c, S), (d, T) ∈ B then there exist f ∈ A and W ∈ F such that c, d f and
S, T ⊇ W. Since (x
a
)
a∈A
is frequently in W, there exists g f such that x
g
∈ W, and
so (c, S), (d, T) ≪ (g, W) ∈ B. This shows that (B, ≪) is a directed set; we claim that
deﬁning
y
(c,S)
= x
c
∀ (c, S) ∈ B and j : B → A; (c, S) → c
makes (y
b
)
b∈B
a subnet of (x
a
)
a∈A
that is eventually in S for all S ∈ F.
Let S ∈ F and a
0
∈ A; there exists c ∈ A such that c a
0
and x
c
∈ S, so if
b
0
:= (c, S) then b = (d, T) ≫ b
0
implies that j(b) a
0
(we have a subnet) and that
y
b
= x
d
∈ T ⊆ S (so (y
b
)
b∈B
is eventually in S). This gives the result.
Proposition A.4. Let (x
a
)
a∈A
be a net in the topological space X. The point x ∈ X is
an accumulation point of (x
a
)
a∈A
if and only if there is a subnet of (x
a
)
a∈A
that converges
to x.
Proof
Let (y
b
)
b∈B
be a subnet of (x
a
)
a∈A
that converges to x, let U be an open set containing
x and let a
0
∈ A. There exists b
0
∈ B such that j(b) a
0
for all b ≫ b
0
(from the
deﬁnition of a subnet) and there exists b
1
∈ B such that y
b
∈ U for all b ≫ b
1
(as (y
b
)
b∈B
converges to x). If b ≫ b
0
and b ≫ b
1
then x
j(b)
= y
b
∈ U and j(b) a
0
, as required.
Conversely, let x be an accumulation point of (x
a
)
a∈A
and let F = ¦U ∈ T : x ∈ U¦.
Then (x
a
)
a∈A
has a subnet that is eventually in every element of F, i.e., converges to x,
by Lemma A.3.
From here we can prove the BolzanoWeierstrass theorem in its full generality.
Theorem A.5. A topological space X is compact if and only if every net in X has a
convergent subnet.
Proof
Suppose that every net in X has a convergent subnet, and so an accumulation point,
by Proposition A.4. Suppose further that X has an open cover F with no ﬁnite subcover,
and let A denote the collection of ﬁnite subsets of F. Ordered by inclusion, A is a
directed set; we deﬁne a net (x
a
)
a∈A
by choosing x
a
/ ∈
U∈a
U for all a ∈ A. Let x be an
accumulation point of (x
a
)
a∈A
; since F is a cover, there exists U
0
∈ F such that x ∈ U
0
.
The net (x
a
)
a∈A
is frequently in U
0
(as this set contains an accumulation point) and so
there exists a ∈ A such that a ⊇ ¦U
0
¦ and x
a
∈ U
0
. Since x
a
/ ∈
U∈a
U ⊇ U
0
, this is a
contradiction.
Conversely, suppose that X is compact and let (x
a
)
a∈A
be a net with no accumulation
points: for all x ∈ X there exists an open set U
x
containing x and a
x
∈ A such that
x
b
/ ∈ U
x
for all b a
x
. The sets ¦U
x
: x ∈ X¦ form an open cover of X, so there is a
ﬁnite subcover ¦U
x
1
, . . . , U
xn
¦, but if a ∈ A is such that a a
x
i
for i = 1, . . . , n then
x
a
/ ∈
n
i=1
U
x
i
= X. This contradiction gives the result.
Deﬁnition A.6. A net in X is universal if, for every S ⊆ X, the net is eventually in S
or eventually in X ¸ S.
Tychonov via Nets 77
A universal net can be thought of as being ‘maximally reﬁned’; the ﬁrst part of the next
proposition makes this idea rigorous.
Proposition A.7. A universal net converges to its accumulation points. The image of
a universal net by any function is universal.
Proof
If x is an accumulation point of the universal net (x
a
)
a∈A
then let U be an open
set containing x. Since (x
a
)
a∈A
is frequently in U, it is eventually in U (since it cannot
eventually be in its complement). Hence (x
a
)
a∈A
converges to x.
If (x
a
)
a∈A
is a universal net in X and f : X → Y then let S ⊆ Y . The net (x
a
)
a∈A
is
eventually in either f
−1
(S) or X ¸ f
−1
(S) = f
−1
(Y ¸ S), hence
_
f(x
a
)
_
a∈A
is eventually
in either S or Y ¸ S, as required. (Note that f(f
−1
(B)) = B ∩f(X) for any B ⊆ Y .)
The next lemma is the most technically involved; the proof involves construction of what
a Bourbakiste would call a ﬁlter (indeed, an ultraﬁlter).
Lemma A.8. Every net has a universal subnet.
Proof
Let (x
a
)
a∈A
be a net in X and let
C :=
_
U = ¦U ⊆ X¦ : U ∈ U ⇒ (x
a
) is frequently in U; U, V ∈ U ⇒ U ∩ V ∈ U
_
,
ordered by inclusion. By Zorn’s lemma, C has a maximal element, U
0
, and Lemma A.3
implies that (x
a
)
a∈A
has a subnet (y
b
)
b∈B
which is eventually in every U ∈ U
0
.
If S ⊆ X is such that (y
b
)
b∈B
is not eventually in X¸S then (y
b
)
b∈B
is frequently in S
and so (x
a
)
a∈A
is frequently in U ∩S for all U ∈ U
0
: by the construction in Lemma A.3,
(y
b
)
b∈B
= (y
(a,U)
)
(a,U)∈B
and given (a, U) ∈ B there exists (b, V ) ∈ B such that b a,
V ⊆ U and x
b
= y
(b,V )
∈ S, so as x
b
∈ V (by construction) x
b
∈ V ∩ S ⊆ U ∩ S. Hence,
by maximality, S ∈ U
0
(as U
0
∪¦S, S ∩U : U ∈ U
0
¦ ∈ C) and so (y
b
)
b∈B
is eventually in
S. This shows that (y
b
)
b∈B
is universal, as required.
Corollary A.9. A space is compact if and only if every universal net is convergent.
Proof
By the BolzanoWeierstrass theorem (Theorem A.5), if the space is compact then
every universal net has a convergent subnet and so an accumulation point (by Proposi
tion A.4). Since universal nets converge to their accumulation points (Proposition A.7),
the universal net is convergent.
Conversely, if every universal net is convergent then, as every net has a universal
subnet (Lemma A.8), every net has a convergent subnet. Theorem A.5 gives the result.
We can now present a proof of Tychonov’s theorem that has a beautiful simplicity (the
dust having been swept under the rug that is Corollary A.9).
Alternative Proof of Tychonov’s Theorem
Let (X, T) be the product of the compact spaces ¦(X
b
, T
b
) : b ∈ B¦ and suppose
that (x
a
)
a∈A
is a universal net in X. Since the image of a universal net is universal
78 Tychonov via Nets
(Proposition A.7), (π
b
(x
a
))
a∈A
is universal in X
b
, so convergent (by Corollary A.9), for
all b ∈ B. Hence x
a
→ x, where x
b
:= lim
a∈A
π
b
(x
a
) (by a property of initial topologies),
and X is compact, by Corollary A.9 again.
Exercises A
Exercise A.1. Let X = ℓ
∞
and for all n ∈ N deﬁne δ
n
∈ X
∗
by setting δ
n
_
(x
k
)
k1
_
=
x
n
. Prove that (δ
n
)
n1
has no weak*convergent subsequence but that (δ
n
)
n1
has a
weak*convergent subnet.
Solutions to Exercises
Solutions to Exercises 1
Exercise 1.1. Let X be a normed vector space and let M be a closed subspace of X.
Prove that
π¦y ∈ X : y −x < ε¦ =
_
[y] ∈ X/M :
_
_
[y] −[x]
_
_
< ε
_
∀ x ∈ X, ε > 0,
where π: X → X/M; x → [x] is the natural map from X onto X/M (the quotient map).
Let
B
X
ε
(x) := ¦y ∈ X : x −y < ε¦
denote the open ball in X with centre x and radius ε. If y ∈ B
X
ε
(x) then
_
_
[y] −[x]
_
_
=
_
_
[y −x]
_
_
y −x < ε
and π(y) ∈ B
X/M
ε
_
[x]
_
. If y ∈ X is such that
_
_
[y] −[x]
_
_
< ε then there exists m ∈ M
such that y −x −m < ε, whence [y] = π(y −m) ∈ π
_
B
X
ε
(x)
_
.
Deduce that the quotient norm yields the quotient topology on X/M given by
Q := ¦U ⊆ X/M : π
−1
(U) ∈ T¦,
where T denotes the norm topology on X, and that the quotient map is open (i.e., sends
open sets to open sets).
Let T
·
denote the topology on X/M given by the quotient norm. If U ∈ Q then
π
−1
(U) ∈ T and so π
−1
(U) =
x∈U
B
X
εx
(x), whence
U = π
_
π
−1
(U)
_
=
_
x∈U
π
_
B
X
εx
(x)
_
=
_
x∈U
B
X/M
εx
_
[x]
_
∈ T
·
.
(The ﬁrst equality holds because π is surjective.) Conversely, since π
−1
_
π(A)
_
= A+M
for all A ⊆ X,
π
−1
_
B
X/M
ε
_
[x]
_
_
= π
−1
_
π
_
B
X
ε
(x)
_
_
= B
X
ε
(x) + M =
_
m∈M
B
X
ε
(x + m) ∈ T.
Hence B
X/M
ε
_
[x]
_
∈ Q and we have the ﬁrst result. The quotient map is open by the
ﬁrst part of this exercise.
79
80 Solutions to Exercises
Prove also that the quotient map is linear and continuous.
Linearity is immediate and continuity follows from the fact that
_
_
[x]
_
_
x for all
x ∈ X.
Exercise 1.2. Prove directly that if E is a Banach space and M is a closed subspace of
E then the quotient space
_
E/M,  
E/M
_
is complete.
Let
_
[x
n
]
_
n1
be a sequence in E/M with
∞
n=1
_
_
[x
n
]
_
_
convergent. For all n 1 there
exists m
n
∈ M such that x
n
−m
n

_
_
[x
n
]
_
_
+2
−n
, by deﬁnition of the quotient norm,
and by comparison
∞
n=1
x
n
−m
n
 is convergent. Hence
∞
n=1
x
n
−m
n
converges, by
the completeness of E, and, as the quotient map π: E → E/M; x → [x] is continuous,
so does π(
∞
n=1
x
n
−m
n
) =
∞
n=1
[x
n
].
Exercise 1.3. Let M and N be subspaces of the normed space X. Prove that if M is
ﬁnite dimensional and N is closed then M + N is closed.
Note that M + N = π
−1
_
π(M)
_
if π: X → X/N is the quotient map, and that if
¦x
1
, . . . , x
n
¦ is a basis for M then ¦π(x
1
), . . . , π(x
n
)¦ is a spanning set for π(M), so
π(M) is ﬁnitedimensional and therefore closed. Hence M + N is the preimage of a
closed set under a continuous map, so is itself closed.
Exercise 1.4. Prove that if ¦A
i
j
: i ∈ I, j ∈ J¦ and ¦B
k
l
: k ∈ K, l ∈ L¦ are families of
sets, where the index sets I, J, K and L are arbitrary, then
_
_
i∈I
j∈J
A
i
j
_
∩
_
_
k∈K
l∈L
B
k
l
_
=
_
(i,k)∈I×K
(j,l)∈J×L
A
i
j
∩ B
k
l
.
We have that
x ∈
_
_
i∈I
j∈J
A
i
j
_
∩
_
_
k∈K
l∈L
B
k
l
_
⇐⇒ ∃ i
0
∈ I, k
0
∈ K such that x ∈ A
i
0
j
and x ∈ B
k
0
l
∀ j ∈ J, l ∈ L
⇐⇒ ∃ (i
0
, k
0
) ∈ I K such that x ∈ A
i
0
J
∩ B
k
0
l
∀ (j, l) ∈ J L
⇐⇒ x ∈
_
(i,k)∈I×K
(j,l)∈J×L
A
i
j
∩ B
k
l
,
as claimed.
What does this have to do with initial topologies?
A consequence of this is that the collection of arbitrary unions of ﬁnite intersections of
elements of a subbase is itself closed under ﬁnite intersections, and so initial topologies
are as claimed in Deﬁnition 1.24.
Exercise 1.5. Prove that if T
F
is the initial topology on X generated by a collection
of functions F and Y ⊆ X then T
F
[
Y
, the relative initial topology on Y , is the initial
topology generated by F[
Y
= ¦f[
Y
: f ∈ F¦, the restrictions of the functions in F to Y .
Solutions to Exercises 81
Since f[
−1
Y
(U) = f
−1
(U) ∩ Y , this is immediate.
Exercise 1.6. Let (x
a
)
a∈A
be a family of nonnegative real numbers and let A denote
the collection of ﬁnite subsets of A. Prove that (x
a
)
a∈A
is summable (with sum α) if and
only if β = sup
_
a∈A
0
x
a
: A
0
∈ A
_
< ∞ and in this case α = β.
If
a∈A
0
x
a
→ α then let ε > 0 and choose A
1
∈ A such that
¸
¸
a∈A
0
x
a
−α
¸
¸
< ε for
all A
0
∈ A with A
0
⊇ A
1
. If A
2
∈ A then
a∈A
2
x
a
a∈A
1
∪A
2
x
a
¸
¸
a∈A
1
∪A
2
x
a
−α
¸
¸
+[α[ < [α[ + ε,
so sup
_
a∈A
0
x
a
: A
0
∈ A
_
< ∞.
Conversely, if β = sup
_
a∈A
0
x
a
: A
0
∈ A
_
< ∞then let ε > 0 and choose A
1
∈ A
such that
a∈A
1
x
a
> β −ε. Then
β + ε >
a∈A
0
x
a
a∈A
1
x
a
> β −ε
for all A
0
∈ A such that A
0
⊇ A
1
; hence
a∈A
0
x
a
→ β. Since R is Hausdorﬀ we must
have α = β.
Exercise 1.7. Let E be a Banach space and let (x
a
)
a∈A
a family of vectors in E. Prove
that if
a∈A
x
a
 is convergent then S := ¦a ∈ A : x
a
,= 0¦ is countable.
Let S
n
:= ¦a ∈ A : x
a
 > 1/n¦ for n ∈ N; applying the previous exercise to
a∈A
x
a
 we see that β = sup
_
a∈A
0
x
a
 : A
0
∈ A
_
< ∞. If S
n
is inﬁnite for some
n ∈ N then let a
1
, . . . , a
m
be distinct elements of S
n
, where m > nβ, and note that
m
j=1
x
a
j
 > m/n > β, a contradiction. Hence S =
∞
n=1
S
n
is a countable union of
ﬁnite sets, so countable.
Deduce that (x
a
)
a∈A
is summable with sum
a∈A
x
a
=
_
a∈S
x
a
if S is ﬁnite,
∞
j=1
x
a
j
if S is inﬁnite,
where (if S is inﬁnite) j → a
j
is a bijection between N and S.
If S is ﬁnite then
a∈A
0
x
a
−
a∈S
x
a
= 0 for all A
0
∈ A such that A
0
⊇ S, hence
a∈A
0
x
a
→
a∈S
x
a
. If S is countably inﬁnite, j → a
j
is as above and m, n ∈ N are
such that m > n then
_
_
m
j=1
x
a
j
−
n
j=1
x
a
j
_
_
m
j=n+1
x
a
j
 β −
n
j=1
x
a
j
. (⋆)
Let ε > 0 and choose A
0
∈ A such that
a∈A
0
x
a
 > β −ε; if n
0
:= max¦j : a
j
∈ A
0
¦
then
n
j=1
x
a
j

a∈A
0
x
a
 > β −ε for all n n
0
, so from (⋆) we have that
_
_
m
j=1
x
a
j
−
n
j=1
x
a
j
_
_
< β −(β −ε) = ε ∀ m, n n
0
.
82 Solutions to Exercises
Thus
_
n
j=1
x
a
j
_
n1
is Cauchy, so convergent to x ∈ E, say. If n n
0
is such that
_
_
n
j=1
x
a
j
−x
_
_
< ε then, for all A
0
∈ A such that A
0
⊇ ¦a
1
, . . . , a
n
¦,
_
_
a∈A
0
x
a
−x
_
_
_
_
a∈A
0
\{a
1
,...,an}
x
a
_
_
+
_
_
n
j=1
x
a
j
−x
_
_
<
a∈A
0
x
a
 −
n
j=1
x
a
 + ε
< β −(β −ε) + ε = 2ε.
Hence
a∈A
0
x
a
→ x, as claimed.
Exercise 1.8. Prove that a family of complex numbers (z
a
)
a∈A
is summable if and only
if
_
[z
a
[
_
a∈A
is summable.
Note ﬁrst that, since
max
_
[ Re z[, [ Imz[
_
[z[ [ Re z[ +[ Imz[ ∀ z ∈ C,
_
[z
a
[
_
a∈A
is summable if and only if
_
[ Re z
a
[
_
a∈A
and
_
[ Imz
a
¸
¸
)
a∈A
are summable, by Ex
ercise 1.6. Note also that (z
a
)
a∈A
is summable if and only if
_
Re z
a
_
a∈A
and
_
Imz
a
_
a∈A
are summable, because if A
0
is a ﬁnite subset of A then
max
_
¸
¸
a∈A
0
Re z
a
−Re z
¸
¸
,
¸
¸
a∈A
0
Imz
a
−Imz
¸
¸
_
¸
¸
a∈A
0
z
a
−z
¸
¸
¸
¸
a∈A
0
Re z
a
−Re z
¸
¸
+
¸
¸
a∈A
0
Imz
a
−Imz
¸
¸
.
Hence it suﬃces to prove the real case of this proposition.
Given (x
a
)
a∈A
⊆ R let x
+
a
= max¦x
a
, 0¦ and x
−
a
= −min¦x
a
, 0¦ for all a ∈ A; note
that
a∈A
0
[x
a
[ =
a∈A
0
x
+
a
+
a∈A
0
x
−
a
∀ A
0
∈ A,
so
_
[x
a
[
_
a∈A
is summable if and only if (x
+
a
)
a∈A
and (x
−
a
)
a∈A
are summable. Since
(y
a
)
a∈A
is summable if and only if (−y
a
)
a∈A
is, and x
a
= x
+
a
− x
−
a
, it suﬃces to prove
that the summability of (x
a
)
a∈A
implies that of (x
+
a
)
a∈A
.
For this we suppose otherwise; let
a∈A
x
a
= x, take ε > 0 and choose A
1
∈ A such
that
¸
¸
a∈A
0
x
a
−x
¸
¸
< ε for all A
0
∈ A that contain A
1
. Set A
2
= ¦a ∈ A
1
: x
−
a
> 0¦
and let A
3
∈ A be such that
a∈A
3
x
+
a
>
a∈A
2
x
−
a
+[x[ + ε;
this exists as
a∈A
x
+
a
is not convergent. If A
0
= ¦a ∈ A
3
: x
+
a
> 0¦ ∪ A
1
then
a∈A
3
x
+
a
a∈A
0
x
+
a
¸
¸
a∈A
0
x
+
a
−x
−
a
−x
¸
¸
+
¸
¸
a∈A
0
x
−
a
+ x
¸
¸
< ε +
a∈A
2
x
−
a
+[x[,
the desired contradiction.
Solutions to Exercises 83
Exercise 1.9. Find a Hilbert space H and a countable family of vectors (x
n
)
n∈N
in H
that is summable but not absolutely summable (i.e.,
_
x
n

_
n∈N
is not summable).
Let H = ℓ
2
(N) and let x
n
= e
n
/n for all n ∈ N, where e
n
is the standard basis vector,
with 1 in nth position and 0 elsewhere. As the harmonic series is divergent,
sup
_
n∈N
0
x
n
 : N
0
is a ﬁnite subset of N
_
sup
_
k
n=1
1
n
: k ∈ N
_
= ∞,
and Exercise 1.6 gives that
_
x
n

_
n∈N
is not summable. However, (x
n
)
n∈N
has sum
x := (1, 1/2, 1/3, . . .): for ε > 0 take k ∈ N such that
∞
n=k+1
1/n
2
< ε
2
and note that

n∈N
0
x
n
−x < ε for every ﬁnite subset N
0
⊆ N that contains ¦1, . . . , k¦.
Exercise 1.10. Prove the converse to Proposition 1.31, that in a space with a non
Hausdorﬀ topology there exists a net that converges to two distinct points.
Suppose that T is not a Hausdorﬀ topology on the set X: there exist distinct points
x, y ∈ X such that every pair of open sets (U, V ) with x ∈ U and y ∈ V satisﬁes
U ∩V ,= ∅. Let A ⊆ T T denote the aggregate of such pairs, with preorder deﬁned
by setting
(A, B) (C, D) ⇐⇒ A ⊇ C and B ⊇ D.
This is a directed set: if (A, B), (C, D) ∈ A then (A ∩ C, B ∩ D) ∈ T T is an upper
bound for (A, B) and (C, D), and since x ∈ A∩C and y ∈ B∩D, (A∩C, B∩D) ∈ A.
For all (A, B) ∈ A let x
(A,B)
∈ A ∩ B; we claim that the net (x
(A,B)
)
(A,B)∈A
converges
to x and to y. If U ∈ T is such that x ∈ U then x
(A,B)
∈ U for all (A, B) ∈ A such that
A ⊆ U, hence x
(A,B)
∈ U for all (A, B) (U, X) and x
(A,B)
→ x. This same argument
works for y and so we have the result.
Exercise 1.11. A sequence in a normed vector space that is convergent is necessarily
bounded. Is the same true for nets?
For a simple counterexample, consider Z directed by its usual order and deﬁne a net
(x
n
)
n∈Z
by setting x
n
= n if n 0 and x
n
= 1/n if n > 0.
Solutions to Exercises 2
Exercise 2.1. Let X be a topological space and E a Banach space; recall that C
b
(X, E),
the space of Evalued, bounded, continuous functions on X, is complete with respect to
the norm
f → f
∞
:= sup¦f(x)
E
: x ∈ X¦.
Prove that C
0
(X, E), the continuous, Evalued functions on X that vanish at inﬁnity
(i.e., those f ∈ C(X, E) such that ¦x ∈ X : f(x)
E
ε¦ is compact for all ε > 0) is a
closed subspace of C
b
(X, E).
84 Solutions to Exercises
Note that any function f ∈ C
0
(X, E) is bounded, as f is continuous on the compact
set ¦x ∈ X : f(x)
E
1¦ and so bounded there. Let (f
n
)
n1
⊆ C
0
(X, E) converge
to f; if ε > 0 and n ∈ N are such that f
n
−f
∞
< ε/2 then
f
n
(x) f(x) −f
n
(x) −f(x) > f(x) −
1
2
ε,
so ¦x ∈ X : f(x) ε¦ is a closed subset of the compact set ¦x ∈ X : f
n
(x) ε/2¦
and therefore is itself compact. Hence C
0
(X, E) is closed.
If f, g ∈ C
0
(X, E) then K = ¦x ∈ X : f(x) ε/2¦ and L = ¦x ∈ X : g(x)
ε/2¦ are compact, and if x ,∈ K ∪ L then
f(x) + g(x) f(x) +g(x) <
1
2
ε +
1
2
ε = ε,
whence ¦x ∈ X : (f + g)(x) ε¦ ⊆ K ∪ L is compact. If α ∈ F then either α = 0,
in which case αf = 0 ∈ C
0
(X, E) trivially, or ¦x ∈ X : αf ε¦ = ¦x ∈ X : f
ε/[α[¦ is compact for all ε > 0. This shows that C
0
(X, E) is a subspace of C
b
(X, E).
Exercise 2.2. Let (X, T) be a Hausdorﬀ, locally compact space and let ∞ denote a
point not in X. Show that
˙
T := T ∪ ¦U ⊆
˙
X : ∞ ∈ U, X ¸ U is compact¦
is a Hausdorﬀ, compact topology on
˙
X := X ∪ ¦∞¦.
It is routine to verify that
˙
T is a topology; recall that compact sets are closed under
ﬁnite unions and arbitrary intersections, and that compact sets are closed in Hausdorﬀ
spaces. If C ⊆
˙
T is an open cover of
˙
X then there exists U ∈ C such that ∞ ∈ U,
and since X ¸ U is compact and has open cover C ¸ ¦U¦, this has a ﬁnite subcover
C
0
. Hence C
0
∪ ¦U¦ is a ﬁnite subcover of the cover C, showing that
˙
T is compact.
Finally, since T is Hausdorﬀ it suﬃces to take x ∈ X and prove that there exist open
sets separating ∞ and x. As T is locally compact there exists U ∈ T such that x ∈ U
and
¯
U is compact, whence U and ¦∞¦ ∪ (X ¸
¯
U) are elements of
˙
T as required.
Prove that there is a natural correspondence between C
0
(X, E) and ¦f ∈ C(
˙
X, E) :
f(∞) = 0¦.
For f ∈ C
0
(X, E) let
˙
f :
˙
X → E; x →
_
f(x) if x ∈ X,
0 if x = ∞.
Let U ⊆ E be open; if 0 ,∈ U then
˙
f
−1
(U) = f
−1
(U) ∈ T ⊆
˙
T, and if 0 ∈ U then
without lost of generality U = B
E
ε
(0) (as we may write U as the union of a set of this
form and the open set U ¸ ¦0¦ = (X ¸ ¦0¦) ∩ U). Since
˙
f
−1
_
B
E
ε
(0)
_
= ¦∞¦∪¦x ∈ X : f(x) < ε¦ = ¦∞¦∪
_
X¸¦x ∈ X : f(x) ε¦
_
∈
˙
T,
we see that
˙
f is continuous.
Solutions to Exercises 85
Conversely, let
˙
f ∈ C(
˙
X, E) be such that
˙
f(∞) = 0, and let f =
˙
f[
X
. Since the
relative topology
˙
T
X
equals T, f is continuous, and if ε > 0 then
¦x ∈ X : f(x) < ε¦ = ¦x ∈
˙
X : 
˙
f(x) < ε¦ ∩ X =
˙
f
−1
_
B
E
ε
(0)
_
∩ X = X ¸ K
for some compact set K ⊆ X, as required.
Exercise 2.3. Let X be a separable normed space. Prove that X
1
is separable (in the
norm topology).
Let S be a countable, dense set in X and note that ¦qs : q ∈ Q, s ∈ S¦ is countable
(being the image of the countable set Q S under the mapping (q, s) → qs). Hence
we may assume, without loss of generality, that S is closed under multiplication by
rationals.
Let x ∈ X
1
and choose (y
n
)
n1
⊆ S such that x −y
n
 1/n for all n 1. Then
z
n
:= ny
n
/(n + 1) ∈ S for all n 1, z
n
→ x by the algebra of limits and
z
n

_
_
n
n + 1
(y
n
−x)
_
_
+
_
_
n
n + 1
x
_
_
1
n + 1
+
n
n + 1
x 1,
as required.
Prove that any separable Banach space E is isometrically isomorphic to a quotient space
of ℓ
1
.
Let (e
n
)
n1
⊆ E
1
be dense; such exists by the ﬁrst part of the question. Deﬁne
T : ℓ
1
→ E; x → Tx :=
∞
n=1
x
n
e
n
and note that T ∈ B(ℓ
1
, E); linearity is obvious, as is the absolute convergence of Tx
(and the fact that T 1) because
∞
n=1
x
n
e
n

∞
n=1
[x
n
[ = x
1
.
Since T
_
(ℓ
1
)
1
_
⊇ ¦e
n
: n 1¦, which is dense in E
1
, the openmapping lemma gives
that T is surjective. Furthermore, this set is kdense in E
1
for all k ∈ (0, 1), so if
y ∈ E there exists x
k
∈ ℓ
1
such that y = Tx
k
and x
k
 Tx
k
/(1 −k). The identity
Tx
k
= Tx
k
′ for all k, k
′
∈ (0, 1) gives that
_
_
[x
k
]
_
_
x
k

Tx
k

1 −k
=
_
_ ˜
T[x
k
]
_
_
1 −k
∀ k ∈ (0, 1),
where [x] = x +ker T and
˜
T : ℓ
1
/ ker T → imT = E is a bijection such that Tx =
˜
T[x]
for all x ∈ ℓ
1
. This shows that
_
_
[y]
_
_
_
_ ˜
T[y]
_
_
for all y ∈ ℓ
1
, and the opposite inequality
follows from the fact
_
_ ˜
T
_
_
= T 1. Hence ℓ
1
/ ker T is isometrically isomorphic to
imT = E, as required.
86 Solutions to Exercises
Exercise 2.4. Prove that no inﬁnitedimensional Banach space E has a countable Hamel
basis (where a Hamel basis is a linearly independent set S such that every vector in E
is a ﬁnite linear combination of elements of S).
Suppose for contradiction that S = ¦e
1
, e
2
, . . .¦ is a countable Hamel basis for the
Banach space E, let F
n
= Fe
1
+ Fe
n
and note that E =
∞
n=1
F
n
. Each F
n
is closed
(being ﬁnitedimensional) and has empty interior: if U ⊆ F
n
is open and nonempty
then it contains B
E
ε
(u) for some u ∈ U and ε > 0, but then u +
1
2
εe
n+1

−1
e
n+1
∈
U ⊆ F
n
, whence e
n+1
∈ F
n
, contradicting linear independence. This shows that E is a
countable union of nowhere dense sets, a contradiction to the Baire category theorem.
Exercise 2.5. Let T : X → Y be a linear transformation from the normed space X onto
the ﬁnitedimensional normed space Y . Prove that T is continuous if and only if ker T
is closed and that if T is continuous then T is open.
It is immediate that if T is continuous then ker T = T
−1
¦0¦ is closed. If ker T is closed
then X/ ker T is a normed space (with respect to the quotient norm) and the quotient
map π: X → X/ ker T is open and continuous. Furthermore,
˜
T : X/ ker T → imT = Y
is a linear bijection between ﬁnitedimensional normed spaces, so is a homeomorphism.
(Recall than a linear transformation between normed space is continuous if its domain
is ﬁnite dimensional: if S: Y → Z is a linear transformation and dimY < ∞ then
the norm y → y
Y
+ Sy
Z
is equivalent to  
Y
, so there exists M > 0 such that
Sy
Z
y
Y
+ Sy
Z
My
Y
for all y ∈ Y .) As T =
˜
T
◦
π and both these maps
are open and continuous, so is T.
Exercise 2.6. Let X = C([0, 1], R) denote the Banach space of continuous, realvalued
functions on the unit interval and for all k ∈ N let
D
k
:= ¦f ∈ X : there exists t ∈ [0, 1] such that [f(s)−f(t)[ k[s−t[ for all s ∈ [0, 1]¦.
Prove that D
k
is closed.
Let (f
n
)
n1
⊆ D
k
be convergent, with limit f, and for each f
n
let t
n
be as in the
deﬁnition of D
k
. Then (t
n
)
n1
is a bounded sequence, so (by the BolzanoWeierstrass
theorem, passing to a subsequence if necessary) we may assume that t
n
→ t ∈ [0, 1].
Let ε > 0 and n
0
∈ N be such that [f(t
n
) −f(t)[ < ε for all n n
0
; for such n,
[f(s) −f(t)[ [f(s) −f
n
(s)[ +[f
n
(s) −f
n
(t
n
)[ +[f
n
(t
n
) −f(t
n
)[ +[f(t
n
) −f(t)[
< 2f −f
n

∞
+ k[s −t
n
[ + ε
→ k[s −t[ + ε
as n → ∞, and since this holds for all ε > 0 we see that f ∈ D
k
.
Prove further that D
k
is nowhere dense.
Let f ∈ X and let ε > 0; since f is uniformly continuous on [0, 1] we may ﬁnd δ > 0
such that [s−t[ < δ implies that [f(s)−f(t)[ < ε/2. Choose 0 = t
0
< t
1
< < t
n
= 1
Solutions to Exercises 87
such that t
i
−t
i−1
< δ for i = 1, 2, . . . , n and let
g : [0, 1] →R; t →
t
i
−t
t
i
−t
i−1
f(t
i−1
) +
t −t
i−1
t
i
−t
i−1
f(t
i
) (if t ∈ [t
i−1
, t
i
], i = 1, 2, . . . , n).
The function g is piecewiselinear (so continuous) and if t ∈ [t
i−1
, t
i
] then
[f(t) −g(t)[
t
i
−t
t
i
−t
i−1
[f(t) −f(t
i−1
)[ +
t −t
i−1
t
i
−t
i−1
[f(t) −f(t
i
)[ <
ε
2
,
so f −g
∞
< ε/2; furthermore,
M := inf¦m ∈ R
+
: [g(s) −g(t)[ m[s −t[ for all s, t ∈ [0, 1]¦
= sup¦[g(s) −g(t)[/[s −t[ : s, t ∈ [0, 1], s ,= t¦
< ∞,
as if 0 s < t 1 then either s, t ∈ [t
i−1
, t
i
], so
_
g(t) −g(s)
_
/(t −s) = g
′
_
(t + s)/2
_
,
or s ∈ [t
i−1
, t
i
] and t ∈ [t
j−1
, t
j
] for j > i, so
¸
¸
¸
¸
g(t) −g(s)
t −s
¸
¸
¸
¸
¸
¸
¸
¸
g(t) −g(t
j−1
)
t −t
j−1
¸
¸
¸
¸
+
¸
¸
¸
¸
g(t
j−1
) −g(t
j−2
)
t
j−1
−t
j−2
¸
¸
¸
¸
+ +
¸
¸
¸
¸
g(t
i
) −g(s)
t
i
−s
¸
¸
¸
¸
(since t −s t −t
j−1
, t −s t
i
−s and t −s t
k
−t
k−1
if k lies between j and i).
Next, deﬁne the sawtooth function (draw a picture)
h: R →R; t → ε[
1
2
−(t −n)[ (t ∈ [n, n + 1], n ∈ Z)
and let g
m
: [0, 1] →R; x → g(x) +h(mx) for m ∈ N. Since h
∞
ε/2 we have that
f −g
m

∞
< ε, so if m > (M + k)/ε, t ∈ [0, 1] and s is suﬃciently near to t then
¸
¸
¸
¸
g
m
(s) −g
m
(t)
t −s
¸
¸
¸
¸
m
¸
¸
¸
¸
h(ms) −h(mt)
ms −mt
¸
¸
¸
¸
−
¸
¸
¸
¸
g(s) −g(t)
s −t
¸
¸
¸
¸
>
M + k
ε
ε −M = k,
which shows that f lies in the closure of X ¸ D
k
. Thus D
k
has empty interior and is
nowhere dense.
Deduce that there exist continuous functions on [0, 1] that are diﬀerentiable at no point
in (0, 1).
If f ∈ X is diﬀerentiable at t ∈ (0, 1) then
g : [0, 1] →R; s →
_
_
_
f(s) −f(t)
s −t
s ,= t,
f
′
(t) s = t
is continuous on [0, 1], so bounded. Hence [f(s) −f(t)[ g
∞
[t −s[ for all s ∈ [0, 1],
and f ∈ D
k
for all k g
∞
. Since X ,=
∞
k=1
D
k
(by the Baire category theorem)
the result follows.
88 Solutions to Exercises
Exercise 2.7. Let H be an inﬁnitedimensional, separable Hilbert space. Prove that
B(H)
1
is not separable in the norm topology.
Let ¦e
1
, e
2
, . . .¦ be an orthornormal basis for H, let H
0
= lin¦e
1
, e
2
, . . .¦ denote the
linear span of this basis and for each subset A ⊆ N deﬁne
P
0
(A): H
0
→ H
0
;
n∈N
α
n
e
n
→
n∈A
α
n
e
n
.
(In H
0
only ﬁnitely many coeﬃcients in
n∈N
α
n
e
n
are nonzero, so this map is well
deﬁned.) It is immediate that P
0
(A) 1, so we may extend P
0
(A) to P(A) ∈
B(H) such that P(A) = P
0
(A)[
H
0
and P(A) 1. If A, B ⊆ N are distinct then
P(A) − P(B) 1: let n ∈ (A ¸ B) ∪ (B ¸ A) and consider
_
_
_
P(A) − P(B)
_
e
n
_
_
.
Suppose that S ⊆ B(H)
1
is dense and let S
A
∈ S be such that S
A
−P(A) < 1/2 for
all A ⊆ N; if A, B ⊆ N are distinct then
1 P(A) −P(B) P(A) −S
A
 +S
A
−S
B
 +S
B
−P(B) < 1 +S
A
−S
B
,
hence A → S
A
is injective and thus S is uncountable.
Solutions to Exercises 3
Exercise 3.1. Let H be a separable Hilbert space with orthonormal basis ¦e
1
, e
2
, . . .¦.
For n 1 let P
n
denote the orthogonal projection onto Fe
1
+ + Fe
n
; prove that
P
n
TP
n
x → Tx as n → ∞ for all T ∈ B(H) and x ∈ H.
Note that
P
n
TP
n
x −Tx P
n
T (P
n
−I)x +(P
n
−I)Tx
T (P
n
−I)x +(P
n
−I)Tx,
so it suﬃces to prove that P
n
y → y for all y ∈ H. To see this, note that Parseval’s
equality gives that
(P
n
−I)y
2
= ¸(P
n
−I)y, (P
n
−I)y) = y
2
−
n
k=1
[¸e
k
, y)[
2
→ 0
as n → ∞.
Deduce that B(H) is separable in the strong operator topology.
Note that
n1
P
n
B(H)P
n
is strong operator dense in B(H) and that a countable union
of countable sets is countable, so it suﬃces to prove that, for all n 1, P
n
B(H)P
n
con
tains a countable, strongoperatordense set. Next, note that P
n
B(H)P
n
is isomorphic
to F
n
2
and all norm topologies on ﬁnitedimensional spaces coincide, so P
n
B(H)P
n
is
norm separable. As the norm topology is ﬁner than the strong operator topology the
result follows.
Solutions to Exercises 89
Exercise 3.2. Prove that if E is a Banach space with respect to two diﬀerent norms then
they are either equivalent or noncomparable (i.e., neither is coarser than the other).
If the norms   and  
′
are comparable then, without loss of generality, the topology
T
·
generated by the ﬁrst is ﬁner than T
·
′ , the topology generated by the second. In
particular, the identity map is continuous from (E, T
·
) to (E, T
·
′ ) (as T
·
′ ⊆ T
·
).
Since every continuous, linear bijection between Banach spaces has continuous inverse,
we have the result.
Exercise 3.3. Prove the following extension of Tietze’s theorem to complexvalued
functions: if X is a normal space, Y a closed subset of X and f ∈ C
b
(Y ) then there
exists F ∈ C
b
(X) such that F[
Y
= f and F
∞
= f
∞
.
Apply Tietze’s theorem to obtain g, h ∈ C
b
(X, R) such that g[
Y
= Re f and h[
Y
=
Imf. Let k = g + ih, so that k[
Y
= f and let
l : C →C; z →
_
z if [z[ f
∞
,
zf
∞
/[z[ if [z[ > f
∞
.
Then F = l
◦
k is as required: if y ∈ Y then F(y) = l(k(y)) = l(f(y)) = f(y) and
F
∞
= l
∞
f
∞
.
Prove also that Tietze’s theorem applies to unbounded, realvalued functions: if X and
Y are as above and f : Y → R is continuous then there exists F : X → R such that
F[
Y
= f.
Let g = arctan
◦
f, so that g takes values in (−π/2, π/2); by Tietze’s theorem there
exists G ∈ C
b
(X) such that G
∞
π/2 and G[
Y
= g. The set C = G
−1
¦±π/2¦
is closed and, by Urysohn’s lemma, there exists H ∈ C
b
(X) such that H
∞
1,
H[
C
= 0 and H[
Y
= 1. The function F = tan
◦
GH is as required: [GH[ < π/2 and
F(y) = tan(G(y)H(y)) = tan(arctan f(y)) = f(y) ∀ y ∈ Y.
Exercise 3.4. Let E be a Banach space, Y a normed vector space and suppose that
(T
n
)
n1
⊆ B(E, Y ) is such that lim
n→∞
T
n
x exists for all x ∈ E. Prove that there exists
T ∈ B(E, Y ) such that T
n
→ T in the strong operator topology.
It is immediate that
T : E → Y ; x → lim
n→∞
T
n
x
is a linear operator, by the continuity of vector addition and scalar multiplication in
a normed space. As convergent sequences in normed spaces are bounded, ¦T
n
x :
n 1¦ is bounded for all x ∈ E, and, as E is complete, the principle of uniform
boundedness implies that M := sup¦T
n
 : n 1¦ is ﬁnite. Hence T M, because
Tx = lim
n→∞
T
n
x Mx ∀ x ∈ E.
What can be said about the norm of T?
90 Solutions to Exercises
The working above shows that
_
T
n

_
n1
is a bounded sequence, so there exists a
subsequence
_
T
n
k

_
k1
such that T
n
k
 → lim
n→∞
T
n
 := sup
n1
inf
mn
T
m
. Since
a subsequence of a convergent sequence converges to the same limit,
Tx = lim
k→∞
T
n
k
x lim
k→∞
T
n
k
 x → lim
n→∞
T
n
 x ∀ x ∈ E.
Hence T lim
n→∞
T
n
.
Exercise 3.5. Let x = (x
n
)
n1
be a sequence of complex numbers such that the series
∞
n=1
x
n
y
n
is convergent for all y ∈ c
0
. Prove that x ∈ ℓ
1
.
For n 1 deﬁne the linear operator
f
n
: c
0
→C; y →
n
j=1
x
j
y
j
.
Since [f
n
(y)[
n
j=1
[x
j
[ [y
j
[ y
∞
n
j=1
[x
j
[ it follows that f
n
is bounded, with
f
n

n
j=1
[x
j
[. This is actually an equality: let y
j
∈ T := ¦α ∈ C : [α[ = 1¦
be such that y
j
x
j
= [x
j
[ and note that z = (y
1
, . . . , y
n
, 0, 0, . . .) ∈ (c
0
)
1
is such that
[f
n
(z)[ =
n
j=1
[x
j
[. For all y ∈ c
0
the sequence (f
n
(y))
n1
is convergent, so
sup¦[f
n
(y)[ : n 1¦ < ∞
and, since c
0
is a Banach space, the principle of uniform boundedness implies that
x
1
=
∞
n=1
[x
n
[ = sup¦f
n
 : n 1¦ < ∞.
Exercise 3.6. Let E be a Banach space with closed subspaces F and G such that
E = F ⊕G (i.e., every element of E can be expressed uniquely as the sum of an element
of F and an element of G). Deﬁne P
F
and P
G
by setting
P
F
: E → E; f + g → f and P
G
: E → E; f + g → g ∀ f ∈ F, g ∈ G.
Prove that P
F
and P
G
are bounded linear operators such that P
2
F
= P
F
, P
2
G
= P
G
and
P
F
P
G
= P
G
P
F
= 0.
The algebraic facts are easily veriﬁed; we prove only that P
F
is bounded, by applying
the closedgraph theorem. Let (x
n
)
n1
⊆ E be such that x
n
→ x and P
F
x
n
→ y; since
P
F
x
n
∈ F for all n and F is closed, y ∈ F, and similarly x−y = lim
n→∞
x
n
−P
F
x
n
∈ G.
Thus x = y + (x −y) ∈ F + G and by uniqueness P
F
x = y, as required.
Exercise 3.7. Find a Banach space E with closed subspaces F and G such that E =
F ⊕G and
P : E → E; f + g → f ∀ f ∈ F, g ∈ G
has norm strictly greater than one.
Solutions to Exercises 91
Let E = R
3
equipped with the norm x = max¦[x
1
[, [x
2
[, [x
3
[¦ (writing vectors in
bold and denoting their coordinates in the obvious manner) and let
f : R
3
→R; (x
1
, x
2
, x
3
) → 2x
1
+ 2x
2
−3x
3
,
F := ker f and G := Ru, where u := (1, 1, 1); since x = (x − f(x)u) + f(x)u for all
x ∈ R
3
and f(u) = 1, E, F and G are as required. Suppose for contradiction that
P = 1 and let v := Pu; note that v = Pu P u = 1, so [v
i
[ 1 for
i = 1, 2, 3.
If x := (1, −1, 1) and y := x + t(u − Pu), where t ∈ R is chosen so that y ∈ F,
whence y = Px, then
y ∈ F ⇐⇒ x + tu ∈ F ⇐⇒ t = 3,
by direct calculation, so y = (4, 2, 4) − 3Pu = (4 − 3v
1
, 2 − 3v
2
, 4 − 3v
3
). Now,
y = Px 1, so [4 −3v
1
[ 1 and v
1
∈ [1, 5/2]; since [v
1
[ 1, v
1
= 1 = v
3
and, as
y ∈ F, v
2
= 1/2. Thus v = Pu = (1, 1/2, 1).
Now let z := (−1, 1, 1) and w := z+s(u−Pu), where s ∈ R is chosen so that w ∈ F
and so w = Pz. As w = (−1, 1 + s/2, 1), it follows that s = 3 and w = (−1, 5/2, 1),
contradicting the fact that w = Pz 1.
[This example is due to Goodner [6].]
Exercise 3.8. Let E be a Banach space with closed subspaces F and G such that
F ∩ G = ¦0¦. Prove that F ⊕G is closed if and only if there exists C > 0 such that
f Cf + g ∀ f ∈ F, g ∈ G.
If F ⊕ G is closed then the projection map P
F
: F ⊕ G → F; f + g → f is bounded
(by Exercise 3.6). Hence
f = P
F
(f + g) P
F
 f + g ∀ f ∈ F, g ∈ G,
as required.
Conversely, suppose that such C > 0 exists and let (x
n
)
n1
⊆ F ⊕ G be such that
x
n
→ x for some x ∈ E. Let x
n
= f
n
+ g
n
for all n 1, where f
n
∈ F and g
n
∈ G,
and note that
f
n
−f
m
 C(f
n
−f
m
) + (g
n
−g
m
) = Cx
n
−x
m
,
so (f
n
)
n1
is Cauchy and hence convergent, say f
n
→ f ∈ F. Furthermore, since G is
closed,
lim
n→∞
g
n
= lim
n→∞
x
n
−f
n
= x −f ∈ G
and x = f + (x −f) ∈ F ⊕G, as required.
Deduce that F ⊕G is closed if and only if
c := inf¦f −g : f ∈ F, g ∈ G, f = g = 1¦ > 0.
92 Solutions to Exercises
If F ⊕ G is closed then, by the previous part, there exists C > 0 such that f
Cf +g for all f ∈ F and g ∈ G. Hence (replacing g by −g) f −g C
−1
f = C
−1
if f ∈ F, g ∈ G and f = g = 1, whence c C
−1
> 0.
Conversely, if F ⊕G is not closed then there exists no such C, so for all n 1 there
exist f
n
∈ F and g
n
∈ G such that f
n
 > nf
n
+g
n
; replacing f
n
by f
n
/f
n
 and g
n
by g
n
/f
n
 we may assume that f
n
 = 1 and f
n
+ g
n
 < 1/n. Since
¸
¸
1 −g
n

¸
¸
=
¸
¸
f
n
 − −g
n

¸
¸
f
n
+ g
n
 < 1/n
we see that
_
_
f
n
+g
n

−1
g
n
_
_
f
n
+ g
n
 +
_
_
−g
n
+g
n

−1
g
n
_
_
<
1
n
+
¸
¸
−1 +g
n

−1
¸
¸
g
n
 <
2
n
;
thus c < 2/n for all n 1 and so c = 0, as claimed.
Solutions to Exercises 4
Exercise 4.1. A closed subspace M of the normed space X is complemented in X if there
exists a closed subspace N such that M ⊕N = X, i.e., M +N = X and M ∩ N = ¦0¦.
Prove that M is complemented in X if M is ﬁnite dimensional.
Let ¦x
1
, . . . , x
n
¦ be a basis for M, let ¦φ
1
, . . . , φ
n
¦ ⊆ M
∗
be the dual basis, use the
HahnBanach theorem to extend φ
i
to
˜
φ
i
∈ X
∗
and then let N =
n
i=1
ker
˜
φ
i
. The fact
that N is closed is immediate and if x ∈ M ∩ N then φ
i
(x) = 0 for all i, so x = 0.
Finally, note that x −
n
i=1
˜
φ
i
(x)x
i
∈ N for all x ∈ X.
Prove also that M is complemented in X if M has ﬁnite codimension, i.e., dimX/M <
∞.
Let X/M have basis
_
[x
1
], . . . , [x
n
]
_
and suppose that N is the vector space spanned
by ¦x
1
, . . . , x
n
¦. Then N is closed (being ﬁnitedimensional), X = π
−1
(X/M) =
π
−1
_
π(N)
_
= M + N (where π: X → X/M is the quotient map) and if
n
i=1
α
i
x
i
∈
M ∩ N then [0] =
n
i=1
α
i
[x
i
], whence α
1
= = α
n
= 0 by linear independence of
_
[x
1
], . . . , [x
n
]
_
.
Exercise 4.2. Let M be a ﬁnitedimensional subspace of the normed space X and let N
be a closed subspace of X such that X = M ⊕N. Prove that if φ
0
is a linear functional
on M then
φ: M ⊕N →F; m+ n → φ
0
(m) ∀ m ∈ M, n ∈ N
is an element of the dual space X
∗
.
Note that ker φ = N + ker φ
0
is closed, as N is closed and ker φ
0
is ﬁnitedimensional.
The result follows from the fact that surjective linear transformations with ﬁnite
dimensional range and closed kernel are continuous (Exercise 2.5).
Exercise 4.3. Prove that a normed vector space X is separable if its dual X
∗
is.
Solutions to Exercises 93
Let (φ
n
)
n1
be dense in X
∗
, let (x
n
)
n1
⊆ X
1
be such that [φ
n
(x
n
)[ φ
n
/2 for all
n 1 and let M = Q−lin¦x
n
¦ (or (Q+ iQ) −lin¦x
n
¦ if F = C). If
¯
M ,= X then let
x
0
∈ X ¸
¯
M; by a corollary to the separation theorem there exists φ ∈ X
∗
such that
φ[
M
= 0 and φ(x
0
) = 1. Let (φ
n
k
)
k1
converge to φ and note that
1
2
φ
n
k
 [φ
n
k
(x
n
k
)[ [φ
n
k
(x
n
k
)−φ(x
n
k
)[+[φ(x
n
k
)[ φ
n
k
−φ x
n
k
 φ
n
k
−φ → 0
as k → ∞, implying that φ = 0. This contradiction gives the result.
Find a separable Banach space E such that E
∗
is not separable.
Note that the separable Banach space ℓ
1
has nonseparable dual space ℓ
∞
.
Prove that a reﬂexive Banach space E is separable if and only if E
∗
is.
The ﬁrst part gives one implication and the other follows immediately by applying the
ﬁrst part with E
∗
in place of E; note that E and E
∗∗
are isometrically isomorphic.
Exercise 4.4. Prove that a Banach space E is reﬂexive if and only its dual E
∗
is
reﬂexive.
If E is reﬂexive then the canonical embedding Γ: E → E
∗∗
is an isometric isomorphism:
for all Z ∈ E
∗∗∗
we wish to ﬁnd φ ∈ E
∗
such that
Z(Ψ) = Ψ(φ) ∀ Ψ ∈ E
∗∗
.
Consider φ = Z
◦
Γ; it is immediate that φ ∈ E
∗
, and if Ψ ∈ E
∗∗
then Ψ = Γ(x) for
some x ∈ E, so
Ψ(Z
◦
Γ) = Γ(x)(Z
◦
Γ) = Z(Γ(x)) = Z(Ψ),
as required.
Conversely, suppose that E
∗
is reﬂexive but E is not, so that Γ(E) is a proper
subspace of E
∗∗
(which is closed because E is complete). Exercise 4.2 and Theorem 3.9
give Z ∈ E
∗∗∗
such that Z[
Γ(E)
= 0 but Z ,= 0; as E
∗
is reﬂexive, Z = Γ
∗
(φ) for some
φ ∈ E
∗
(where Γ
∗
: E
∗
→ E
∗∗∗
is the canonical embedding) but then
0 = Z(Γ(x)) = Γ
∗
(φ)(Γ(x)) = Γ(x)(φ) = φ(x) ∀ x ∈ E,
contradicting the fact that Z = Γ
∗
(φ) ,= 0.
Exercise 4.5. Prove that any inﬁnitedimensional normed space has a discontinuous
linear functional deﬁned on it.
First use Zorn’s lemma to prove that the space X has a Hamel basis ¦e
a
: a ∈ A¦
with e
a
 = 1 for all a ∈ A. Deﬁne φ by choosing an inﬁnite set of distinct elements
¦a
n
: n ∈ N¦ and setting
φ
_
a∈A
λ
a
e
a
_
=
∞
n=1
nλ
an
∀ x =
a∈A
λ
a
e
a
∈ X;
it is easy to verify that φ is a welldeﬁned linear functional on X and since φ(e
an
) =
n → ∞ as n → ∞, φ is not bounded.
94 Solutions to Exercises
Exercise 4.6. Let A be a subset of the normed vector space X. Prove that A is norm
bounded (there exists r ∈ R
+
such that a r for all a ∈ A) if and only if it is weakly
bounded (for all φ ∈ X
∗
there exists r
φ
∈ R
+
such that [φ(a)[ r
φ
for all a ∈ A).
If A is norm bounded then there exists r ∈ R
+
such that a r for all a ∈ A.
If φ ∈ X
∗
then [φ(a)[ φ a φr for all a ∈ A, showing that A is weakly
bounded. Conversely, if A is weakly bounded then ¦φ(a) : a ∈ A¦ = ¦ˆ a(φ) : a ∈ A¦ is
bounded for all φ ∈ X
∗
, and the principle of uniform boundedness yields boundedness
of ¦ˆ a : a ∈ A¦. Since ˆ a = a we see that A is norm bounded.
Deduce that a weakly holomorphic function is (strongly) continuous.
Suppose that f : U → X be weakly holomorphic, where U is an open subset of C and
X is a complex normed space. Let a ∈ U and choose ε > 0 such that B
C
2ε
(a) ⊆ U; if
φ ∈ X
∗
then, by Cauchy’s integral formula,
¸
¸
¸
(φ
◦
f)(z) −(φ
◦
f)(a)
z −a
¸
¸
¸ =
¸
¸
¸
1
2πi
_
γ
(φ
◦
f)(w)
(w −z)(w −a)
dw
¸
¸
¸
M
φ
ε/2
∀ z ∈ B
C
ε/2
(a) ¸ ¦a¦,
where γ : [0, 2π] →C; t → a + εe
it
and M
φ
= sup
_
[(φ
◦
f)(w)[ : [w −a[ = ε
_
. Hence
_
f(z) −f(a)
z −a
: 0 < [z −a[ < ε/2
_
is weakly bounded, so norm bounded: there exists r ∈ R
+
such that
f(z) −f(a) r[z −a[ ∀ z ∈ B
C
ε/2
(a),
whence f(z) → f(a) in X as z → a.
[This is the ﬁrst step in proving Dunford’s theorem, that weakly holomorphic functions
are strongly holomorphic [19, Theorem 3.31].]
Exercise 4.7. Let H be a Hilbert space. Prove that the adjoint T → T
∗
is continuous
with respect to the weak operator topology on B(H), but not necessarily with respect
to the strong operator topology.
Note that T
a
→ T in the weak operator topology on B(H) if and only if ¸x, T
a
y) →
¸x, Ty) for all x, y ∈ H. Since ¸x, Sy) = ¸S
∗
x, y) = ¸y, S
∗
x) and z → ¯ z is continuous
on C we have the ﬁrst claim. For the second, let ¦e
n
: n ∈ N¦ denote the standard
orthonormal basis of ℓ
2
and let T
n
= [e
n
)¸e
1
[ for all n 1, i.e.,
T
n
: ℓ
2
→ ℓ
2
; x → ¸e
1
, x)e
n
.
It is readily veriﬁed that T
n
∈ B(H) (with T
n
 1) and T
n
e
1
 = 1. Furthermore,
T
∗
n
= [e
1
)¸e
n
[ and T
∗
n
x = [¸e
n
, x)[ → 0 as n → ∞ (by Parseval’s equality) so T
∗
n
→ 0
in the strong operator topology. This gives the result.
Exercise 4.8. Let E and F be Banach spaces. Show that if T : E → F and S: F
∗
→ E
∗
are linear transformations that satisfy
φ(Tx) = (Sφ)(x) ∀ x ∈ E, φ ∈ F
∗
(⋆)
then S and T are bounded, with S = T
∗
.
Solutions to Exercises 95
Let (x, y) be a limit point of G(T), the graph of T, and suppose that (x
n
)
n1
⊆ E is
such that x
n
→ x and Tx
n
→ y. For all φ ∈ F
∗
we have that Sφ ∈ E
∗
and so
φ(y) = lim
n→∞
φ(Tx
n
) = lim
n→∞
(Sφ)(x
n
) = (Sφ)(x) = φ(Tx),
hence Tx = y (since F
∗
separates points in F); by the closedgraph theorem T ∈
B(E, F). Furthermore, T
∗
∈ B(F
∗
, E
∗
) is such that S = T
∗
, by (⋆), as E certainly
separates points in E
∗
.
[This is the HellingerToeplitz theorem.]
Exercise 4.9. Let E and F be Banach spaces and suppose that T ∈ B(E, F) has closed
range, i.e., imT is closed in F. Prove that imT
∗
= (ker T)
⊥
(where
M
⊥
:= ¦φ ∈ E
∗
: φ(x) = 0 for all x ∈ M¦
is the annihilator of the subspace M ⊆ E).
If φ ∈ F
∗
then
(T
∗
φ)(x) = φ(Tx) = φ(0) = 0 ∀ x ∈ ker T,
so that imT
∗
⊆ (ker T)
⊥
. Conversely, suppose that ψ ∈ (ker T)
⊥
, i.e., ψ ∈ E
∗
satisﬁes
ψ(x) = 0 for all x ∈ ker T, and deﬁne
θ
0
: E/ ker T →F; [x] → ψ(x).
This is a good deﬁnition (because ker T ⊆ ker ψ) and
[θ
0
[x][ = [ψ(x)[ = [ψ(x + m)[ ψ x + m ∀ m ∈ ker T,
so [θ
0
[x][ ψ [x] for all x ∈ E and θ
0
is continuous. Since imT is closed, the
bounded linear operator
˜
T : E/ ker T → imT has continuous inverse (by the open
mapping theorem); extending θ
0
◦
˜
T
−1
to θ ∈ F
∗
by the HahnBanach theorem we see
that
(T
∗
θ)(x) = θ(Tx) = θ
0
[x] = ψ(x) ∀ x ∈ E
so ψ ∈ imT
∗
and the result follows.
Exercise 4.10. Let E = c
0
, so that E
∗
= ℓ
1
and E
∗∗
= ℓ
∞
. Prove that x →
∞
n=1
x
n
is
weakly continuous on ℓ
1
but is not weak* continuous.
The weak topology on ℓ
1
is such that ˆ y : x →
∞
n=1
y
n
x
n
is continuous for all y ∈ ℓ
∞
,
so if 1 := (1, 1, . . .) then
ˆ
1: x →
∞
n=1
x
n
is weakly continuous. If
_
x
(n)
_
n1
⊆ ℓ
1
is
deﬁned by setting x
(n)
n
= 1 and x
(n)
k
= 0 if k ,= n then
ˆ
1
_
x
(n)
_
= 1 for all n 1 but
ˆ y
_
x
(n)
_
= y
n
→ 0 as n → ∞ for all y ∈ c
0
, i.e., x
(n)
→ 0 in the weak* topology. Hence
ˆ
1 is not weak* continuous.
Exercise 4.11. Prove that a compact metric space is separable.
96 Solutions to Exercises
Let (X, d) be a compact metric space. For n ∈ N the set ¦B
1/n
(x) : x ∈ X¦ is an
open cover of X, so there exist x
(n)
1
, . . . , x
(n)
mn
such that X =
mn
k=1
B
1/n
_
x
(n)
k
_
. We claim
that S =
n∈N
_
x
(n)
1
, . . . , x
(n)
mn
_
is countable (being a countable union of ﬁnite sets) and
dense in X. For the latter claim, let ε > 0 and x ∈ X; there exists n ∈ N such that
ε > 1/n and some k ∈ ¦1, . . . , m
n
¦ such that x ∈ B
1/n
_
x
(n)
k
_
, whence d
_
x, x
(n)
k
_
< ε
and S ∩ B
ε
(x) ,= ∅. The result follows.
Prove that if X is a separable normed space then X
∗
1
, the closed unit ball of the dual
space X
∗
, is metrizable when equipped with the weak* topology.
Let (x
n
)
n1
⊆ X
1
be dense in X
1
and deﬁne
d: X
∗
1
X
∗
1
→R
+
; (φ, ψ) →
∞
n=1
2
−n
[φ(x
n
) −ψ(x
n
)[. (†)
Note that the series is convergent (by comparison with
∞
n=1
2
−n
φ −ψ) so d is well
deﬁned. Symmetry and the triangle inequality are immediate and if d(φ, ψ) = 0 then
(φ −ψ)(x
n
) = 0 for all n 1, whence φ = ψ. Hence d is a metric on X
∗
1
; it remains to
prove that T
d
= σ(X
∗
, X)[
X
∗
1
. Note ﬁrst that (φ, ψ) → [φ(x
n
) − ψ(x
n
)[ = [ˆ x
n
(φ −ψ)[
is a continuous function on X
∗
1
X
∗
1
(where each factor is equipped with the weak*
topology) and the series (†) is uniformly convergent on this set, so continuous. In
particular, the balls
¦ψ ∈ X
∗
1
: d(ψ, φ) < ε¦ (φ ∈ X
∗
1
, ε > 0)
are σ(X
∗
, X)[
X
∗
1
open, so T
d
⊆ σ(X
∗
, X)[
X
∗
1
. As T
d
is Hausdorﬀ and σ(X
∗
, X)[
X
∗
1
is compact, these topologies are equal. (Recall that a continuous bijection from a
compact space to a Hausdorﬀ space is a homeomorphism.)
Deduce that X
∗
is separable in the weak* topology.
This follows because X
∗
n
is separable for all n ∈ N (being the image of the weak*
separable space X
∗
1
under the homeomorphism x → nx) and X
∗
=
n∈N
X
∗
n
.
Exercise 4.12. Let X and Y be normed spaces and for all x ∈ X and y ∈ Y let
x ⊗y : B(X, Y
∗
) →F; T → (Tx)(y).
Prove that x ⊗y ∈ B(X, Y
∗
)
∗
, with x ⊗y = x y, and that the mapping
X Y → B(X, Y
∗
); (x, y) → x ⊗y
is bilinear.
It is straightforward to see that x ⊗ y ∈ B(X, Y
∗
) with x ⊗ y x y; for the
reverse inequality, suppose that x and y are nonzero and let φ ∈ X
∗
and ψ ∈ Y
∗
be
such that φ = ψ = 1, φ(x) = x and ψ(y) = y. (These exist by Theorem 3.10).
If T := z → φ(z)ψ then T ∈ B(X, Y
∗
), with T φ ψ = 1, and (x ⊗ y)(T) =
φ(x)ψ(y) = x y, whence x⊗y x y, as required. Veriﬁcation of bilinearity
is routine.
Solutions to Exercises 97
If Z is the closed linear span of ¦x ⊗y : x ∈ X, y ∈ Y ¦ in B(X, Y
∗
)
∗
, prove that
j : B(X, Y
∗
) → Z
∗
; j(T)z = z(T)
is an isometric isomorphism.
Clearly, j is a linear map and j(T) T for all T ∈ B(X, Y
∗
). For the reverse
inequality, let ε > 0 and choose x ∈ X such that x = 1 and Tx > T − ε/2,
then choose y ∈ Y such that y = 1 and (Tx)(y) > Tx − ε/2. It follows that
x ⊗y = 1 and (x ⊗y)T = (Tx)(y) > T −ε, whence j(T) T. Finally,
let φ ∈ Z
∗
and let φ
x
: Y →F; y → φ(x⊗y). Since [φ(x⊗y)[ φ x y, φ
x
∈ Y
∗
,
and if T := x → φ
x
then T φ, so T ∈ B(X, Y
∗
) and j(T) = φ; hence j is
surjective.
[This idea, which allows one to put a weak* topology on B(X, Y
∗
), was used by Arveson
in [1] and dates back to Schatten [20, Theorem 3.2].]
Solutions to Exercises 5
Exercise 5.1. Prove that if X is a vector space with separating subspace M ⊆ X
′
and
φ ∈ X
′
is a linear functional that is σ(X, M)continuous then there exist φ
1
, . . . , φ
n
∈ M
such that
[φ(x)[ max
1in
[φ
i
(x)[ ∀ x ∈ X.
By the deﬁnition of σ(X, M),
[φ[
−1
[0, 1) ⊇
n
i=1
[φ
i
[
−1
[0, ε
i
),
where φ
1
, . . . , φ
n
∈ M and ε
1
, . . . , ε
n
> 0. Replacing φ
i
by φ
i
/ε
i
if necessary, with
out loss of generality ε
1
= = ε
n
= 1. If x ∈ X is such that α := [φ(x)[ >
max
1in
[φ
i
(x)[ then [φ(α
−1
x)[ = 1 but
max
1in
[φ
i
(α
−1
x)[ = max
1in
[φ
i
(x)[/[φ(x)[ < 1,
a contradiction.
Deduce that
n
i=1
ker φ
i
⊆ ker φ and that there exists f ∈ (F
n
)
∗
such that
f
_
φ
1
(x), . . . , φ
n
(x)
_
= φ(x) ∀ x ∈ X.
The ﬁrst deduction is immediate; for the next, deﬁne Φ: X →F
n
by setting
Φ(x) :=
_
φ
1
(x), . . . , φ
n
(x)
_
∀ x ∈ X
and note that ker Φ =
n
i=1
ker φ
i
⊆ ker φ, so that if Φ(x) = Φ(y) then φ(x) = φ(y).
Hence
f
0
: Φ(X) →F;
_
φ
1
(x), . . . , φ
n
(x)
_
→ φ(x)
is well deﬁned, and we extend f
0
to f ∈ (F
n
)
∗
by setting f[
Φ(X)
= f
0
and f[
Φ(X)
⊥ = 0.
98 Solutions to Exercises
Conclude that φ ∈ M.
Let f(0, . . . , 1, . . . , 0) = α
i
(where the 1 is in the ith place) so that
φ(x) = f
_
φ
1
(x), . . . , φ
n
(x)
_
=
n
i=1
φ
i
(x)α
i
=
_
n
i=1
α
i
φ
i
_
(x) ∀ x ∈ X,
i.e., φ =
n
i=1
α
i
φ
i
∈ M.
Exercise 5.2. Let X be an inﬁnitedimensional normed space and let V ⊆ X be a
weakly open set containing the origin. Show that V contains a closed subspace of ﬁnite
codimension in X.
Since V is a weakly open set containing 0 there exist φ
1
, . . . , φ
n
∈ X
∗
such that V
contains
n
j=1
[φ
j
[
−1
[0, 1). The linear transformation
Φ: X →F
n
; x →
_
φ
1
(x), . . . , φ
n
(x)
_
has kernel ker Φ =
n
i=1
ker φ
i
⊆ V , which is closed because ker φ
i
is closed for i =
1, . . . , n. Since X/ ker Φ
∼
= imΦ ⊆ F
n
, ker Φ has ﬁnite codimension, as desired.
Deduce that the weak topology on X is strictly coarser than the norm topology.
Note that the open unit ball B
X
1
(0) is open with respect to the norm topology on X, but
it contains no subspace of X other than ¦0¦, which does not have ﬁnite codimension.
Hence σ(X, X
∗
) has fewer open sets than the norm topology.
(To see that these topologies are comparable, note that x → φ(x) is norm continuous
for all φ ∈ X
∗
, so φ
−1
(U) is open (with respect to the norm topology) for all open
U ⊆ F.)
Exercise 5.3. Let X be a topological vector space. Prove that every φ ∈ X
′
¸ ¦0¦ is
open.
Let x
0
∈ X be such that φ(x
0
) = 1 and let A ⊆ X be open. If x ∈ A then A − x is
an open set containing 0 (by the continuity of vector addition) and so m
−1
x
0
(A − x) is
an open set containing 0, where m
x
0
: F → X; α → αx
0
(by the continuity of scalar
multiplication). Hence there exists δ > 0 such that αx
0
∈ A − x if [α[ < δ, and so
α +φ(x) = φ(αx
0
+x) ∈ φ(A) for such α. This shows that φ(A) is open and gives the
result.
Exercise 5.4. Suppose that X is a vector space equipped with a topology that makes
vector addition and scalar multiplication, i.e., the maps
X X → X; (x, y) → x + y and F X → X; (α, x) → αx,
continuous. Show that if this topology is such that singleton sets are closed (i.e., ¦x¦ is
closed for all x ∈ X) then the topology is Hausdorﬀ (so X is a topological vector space).
Solutions to Exercises 99
Note that, for ﬁxed y ∈ X, the map x → x+y is a homeomorphism of X with itself, so
it suﬃces take x ∈ X¸¦0¦ and ﬁnd an open set U such that 0 ∈ U and U∩(x+U) = ∅.
Note that the map
X X → X; (y, z) → y −z
is continuous and so ¦(y, z) ∈ XX : y−z ,= x¦ is an open subset of XX containing
(0, 0). Hence there exist open sets V , W ⊆ X such that 0 ∈ V ∩ W and x ,∈ V −W;
taking U = V ∩ W gives the result.
Exercise 5.5. Let X be a topological vector space. Prove that every open set containing
the origin contains a nonempty open set which is balanced: a set B is balanced if λb ∈ B
for all b ∈ B and λ ∈ F
1
.
Let U ⊆ X be an open set containing 0. As m : F X → X; (λ, x) → λx is
continuous, m
−1
(U) is an open set in F X which contains (0, 0); by the deﬁnition of
the product topology, there exist ε > 0 and an open set V ⊆ X such that 0 ∈ V and
B
F
ε
(0) V ⊆ m
−1
(U). It is easily veriﬁed that m(B
F
ε
(0) V ) = ¦λv : [λ[ < ε, v ∈ V ¦
is balanced, contained in U and open, since it is the union of λV over all λ ∈ F such
that 0 < [λ[ < ε and the map v → λv is a homeomorphism for all λ ,= 0.
Deduce that if C ⊆ X is compact and does not contain the origin then there exist disjoint
open sets A, B ⊆ X such that C ⊆ A and B is a balanced set containing 0.
As X is Hausdorﬀ, for all x ∈ C there exist disjoint open sets A
x
, B
x
⊆ X such that
x ∈ A
x
and 0 ∈ B
x
; by the ﬁrst part of this exercise, B
x
may be taken to be balanced.
Since C is compact, there exist x
1
, . . . , x
n
∈ C such that A := A
x
1
∪ ∪ A
xn
⊆ X,
and if B := B
x
1
∩ ∩ B
xn
then A and B are as required.
Show that a balanced set is connected and give an example to show that a balanced set
need not be convex.
For any point x in a balanced set B, the path t → (1 − t)x connects x to the origin;
thus B is path connected, so connected. The crossshaped set ¦(x, 0), (0, y) : x, y ∈
[−1, 1]¦ ⊆ R
2
is balanced but not convex.
Exercise 5.6. Let p ∈ (0, 1),
L
p
[0, 1] :=
_
f : [0, 1] →C
¸
¸
f is measurable and ∆(f) < ∞
_
,
where ∆(f) :=
_
1
0
[f(x)[
p
dx, and let L
p
[0, 1] := L
p
[0, 1]/N, where
N :=
_
f : [0, 1] →C
¸
¸
f is measurable and zero almost everywhere
_
.
Prove that d([f], [g]) := ∆(f −g) is a metric on L
p
[0, 1] and that L
p
[0, 1] is a topological
vector space (when equipped with this topology).
The meanvalue theorem may be used to establish the inequality (1 + z)
p
− z
p
1
for all z 0, from which it follows that (x + y)
p
x
p
+ y
p
for all x, y 0. This is
enough to see that L
p
[0, 1] is closed under sums and so is a vector space; it also gives
100 Solutions to Exercises
the triangle inequality for d. The other two requirements of a metric are obviously
satisﬁed and the inequalities
d([f + g], [f
a
+ g
a
]) d([f], [f
a
]) + d([g], [g
a
])
and
d(λf, λ
a
f
a
) [λ −λ
a
[
p
_
1
0
[f(t)[
p
dt +[λ
a
[
p
d([f], [f
a
])
show that vector addition and scalar multiplication are suitably continuous.
Prove further that L
p
[0, 1] has no convex, open sets other than ∅ and L
p
[0, 1].
Let V ,= ∅ be convex and open; without loss of generality 0 ∈ V and so there exists
ε > 0 such that ¦f ∈ L
p
[0, 1] : d(f, 0) < ε¦ ⊆ V . Let f ∈ L
p
[0, 1] and choose
n ∈ N such that n
p−1
d(f, 0) < ε. By the intermediatevalue theorem, there exist
0 = x
0
< x
1
< < x
n
= 1 such that
_
x
i
x
i−1
[f(t)[
p
dt = d(f, 0)/n (i = 1, . . . , n);
let g
i,n
:= nf on [x
i−1
, x
i
) and 0 elsewhere, so that f = (g
1,n
+ + g
n,n
)/n and
d(g
i,n
, 0) = n
p
_
x
i
x
i−1
[f(t)[
p
dt = n
p−1
d(f, 0) < ε
for all i. It follows immediately that f ∈ V and V = L
p
[0, 1].
Deduce that the only continuous linear functional on L
p
[0, 1] is the zero functional.
If φ ∈ L
p
[0, 1]
′
is continuous then φ
−1
_
B
C
ε
(0)
_
is convex, open and nonempty, for all
ε > 0. Hence φ(L
p
[0, 1]) ⊆ B
C
ε
(0) for all ε > 0 and so φ = 0.
[This argument is (essentially) due to Tychonov [23], who worked with the sequence
space ¦(x
n
)
n∈N
:
∞
n=1
[x
n
[
1/2
< ∞¦.]
Exercise 5.7. Let X be a topological vector space over F and let M be a ﬁnite
dimensional subspace of X. Prove that M is linearly homeomorphic to F
n
, where n
is the dimension of M.
Let ¦e
1
, . . . , e
n
¦ be a basis for M and let
T : F
n
→ M; (λ
1
, . . . , λ
n
) → λ
1
e
1
+ + λ
n
e
n
.
It is immediate that T is a linear bijection which is continuous, because scalar multi
plication and vector addition are; it remains to show that T is open. Let
S := ¦(λ
1
, . . . , λ
n
) : [λ
1
[
2
+ +[λ
n
[
2
= 1¦
be the unit sphere in F
n
; this is compact and therefore so is its image, T(S). By
Exercise 5.5, there exist disjoint open sets U, V ⊆ M such that 0 ∈ U, which is
Solutions to Exercises 101
balanced, and T(S) ⊆ V . It follows that U ⊆ T(B
F
1
(0)): otherwise, there exists
u ∈ U such that u = T(λ
1
, . . . , λ
n
) with α := [λ
1
[
2
+ + [λ
n
[
2
1, but then
α
−1/2
(λ
1
, . . . , λ
n
) ∈ S and α
−1/2
u ∈ T(S) ∩ U, a contradiction. Hence if W ⊆ F
n
is
open and w ∈ W then there exists ε > 0 such that B
F
ε
(0) ⊆ W −w, so
Tw + εU ⊆ Tw + εT(B
F
1
(0)) ⊆ T(w + B
F
ε
(0)) ⊆ T(W);
since Tw + εU is open and contains Tw, the set T(W) is open and the result follows.
Prove also that M is closed in X.
Let T be the map above, now considered to have codomain X; clearly T is (still)
continuous and the same working as above yields an open set U ⊆ X such that 0 ∈ U
and U ∩M ⊆ T(B
F
1
(0)). Let x be in the closure of M and note that tx → 0 as t → 0+,
by the continuity of scalar multiplication, so there exists r > 0 such that x ∈ rU.
Hence
x ∈ rU ∩ M ⊆ rU ∩ M ⊆ rT(B
F
1
(0)) ⊆ T((F
n
)
r
) = T((F
n
)
r
) ⊆ M,
since (F
n
)
r
is compact and T is continuous. (The ﬁrst inclusion holds as rU is open.)
Exercise 5.8. Prove that a topological vector space with topology given by a separating
family of linear functionals is locally convex.
Let X be such a topological vector space, with separating family M. If U is an open
subset of X containing the origin, there exist φ
1
, . . . , φ
n
∈ M and ε
1
, . . . , ε
n
> 0 such
that
n
i=1
φ
−1
i
_
(B
F
ε
i
(0)
_
⊆ U,
so it suﬃces to prove that φ
−1
(B
F
ε
(0)) is convex for any φ ∈ M and ε > 0. To see this,
let t ∈ (0, 1) and suppose u and v ∈ X are such that [φ(u)[ < ε and [φ(v)[ < ε. Then
[φ(tu + (1 −t)v)[ t[φ(u)[ + (1 −t)[φ(v)[ < ε, as required.
Exercise 5.9. Suppose that X is a locally convex topological vector space and M is
the collection of continuous linear functionals on X. Prove that a convex subset of X is
closed (with respect to the original topology) if and only if it is closed with respect to
σ(X, M), the initial topology generated by M.
As σ(X, M) is the coarsest topology to make every element of M continuous, by
deﬁnition, σ(X, M) is coarser than T
X
, the original topology on X; in particular,
every σ(X, M)closed set is T
X
closed. Conversely, if B ⊆ X is convex and T
X

closed then let x
0
∈ X ¸ B; by Theorem 3.32(ii), there exists φ ∈ M such that
Re φ(x
0
) < inf
y∈B
Re φ(y). Thus B is σ(X, M)closed, as required.
Need this hold if X is not locally convex?
No: if X = L
p
[0, 1] for some p ∈ (0, 1) then, by Exercise 5.6, F = ¦0¦ and T
F
= ¦0, X¦,
the trivial topology. However, the onedimensional subspace of X consisting of the
102 Solutions to Exercises
constant functions is closed, convex, nonempty and not the whole of X. (For the ﬁrst
of these claims, either use Exercise 5.7 or prove it directly: if the sequence of constant
functions (λ
n
1)
n1
converges to f ∈ X then it is Cauchy, but d(λ
n
1, λ
m
1) = [λ
n
−λ
n
[
p
and so λ
n
→ λ for some λ ∈ C. Furthermore, d(f, λ1) d(f, λ
n
1) +d(λ
n
1, λ) → 0, so
f = λ1, as required.)
Exercise 5.10. Let X be a locally convex topological vector space. Show that if N
is a nonempty subspace of X and x
0
∈ X ¸
¯
N then there exists a continuous linear
functional φ ∈ X
′
such that φ[
N
= 0 and φ(x
0
) = 1.
Let B =
¯
N; by the separation theorem (Theorem 3.32) there exists a continuous
φ
0
∈ X
′
such that
Re φ
0
(x
0
) < inf¦Re φ
0
(y) : y ∈ B¦.
In particular, φ
0
_
¦x
0
¦
_
∩φ
0
(B) = ∅, and as φ
0
(B) is a subspace of φ
0
(X) = F we must
have that φ
0
(B) = ¦0¦. Hence φ
0
(x
0
) ,= 0 and φ: x → φ
0
(x)/φ
0
(x
0
) is as required.
Exercise 5.11. Let X be a topological vector space and suppose V is an open set
containing 0. Prove there exists an open set U containing 0 such that U + U ⊆ V .
Since a : X X → X; (x, y) → x +y is continuous and 0 + 0 = 0 ∈ V , the preimage
a
−1
(V ) is an open set containing (0, 0); hence there exist open sets W
1
, W
2
⊆ X such
that 0 ∈ W
1
∩ W
2
and W
1
+ W
2
⊆ V . Letting U := W
1
∩ W
2
establishes the claim.
Deduce or prove otherwise that if A ⊆ B ⊆ X, where A is compact and B is open, then
there exists an open set U ⊆ X such that 0 ∈ U and A+ U ⊆ B.
If x ∈ A then B − x is an open set containing 0 so, by the ﬁrst part, there exists an
open set U
x
which contains 0 and satisﬁes U
x
+U
x
⊆ B −x. As x +U
x
is open for all
x ∈ A, compactness yields x
1
, . . . , x
n
∈ A such that A ⊆ (x
1
+U
x
1
) ∪ ∪(x
n
+U
xn
);
if U := U
x
1
∩ ∩ U
xn
then U is an open set containing 0 and
A+ U ⊆
n
_
i=1
(x
i
+ U
x
i
+ U) ⊆
n
_
i=1
(x
i
+ U
x
i
+ U
x
i
) ⊆ B,
as required.
The Appendix provides the means for an alternative proof. Suppose, for contradiction,
that if U ⊆ X is open and contains 0 then there exist x
U
∈ A and y
U
∈ U such that
x
U
+ y
U
,∈ B. Ordering such U by reverse inclusion, (y
U
) converges to 0 and (x
U
) is
contained in the compact set A, so has a convergent subnet, by Theorem A.5. Hence
(x
U
+ y
U
) has a convergent subnet, with limit z in A, but x
U
+ y
U
∈ X ¸ B for all U,
which is closed, so z ∈ X ¸ B ⊆ X ¸ A. This contradiction gives the result.
Exercise 5.12. Suppose that X is a topological vector space such that the continuous
elements of X
′
separate points. Prove that given disjoint, nonempty, compact, convex
A, B ⊆ X there exists a continuous φ ∈ X
′
such that
sup
x∈A
Re φ(x) < inf
x∈B
Re φ(x).
Solutions to Exercises 103
Since σ(X, M) is coarser than the original topology on X, the sets A and B are
σ(X, M)compact; furthermore, σ(X, M) is Hausdorﬀ, because M separates points,
(Proposition 1.26) and both vector addition and scalar multiplication are σ(X, M)
continuous, by Propositions 1.25, 1.35 and 1.37 together with the continuity of addition
and multiplication in F. Hence X is a topological vector space when equipped with
σ(X, M); we work with this topology from now on, which has the beneﬁt of being
locally convex, by Exercise 5.8, and also gives rise to the same collection of continuous
linear functionals, M, by Exercise 5.1. Applying Exercise 5.11 to A and X ¸ B yields
an open set U containing 0 such that A + U ⊆ X ¸ B, whence A + U and B are
disjoint. Moreover, the set U may be taken to be convex, so A + U is convex and
Theorem 3.32(i) yields φ ∈ M and s ∈ R such that Re φ(x) < s inf
y∈B
Re φ(y) for
all x ∈ A+ U; the result follows.
Deduce that Theorem 3.40 is true for topological spaces with continuous dual spaces
that separate points.
The proof given in Section 3.12 applies with the following changes: Corollary 3.33 is
not required, since the continuous dual space is assumed to separate points, and the
result just established is a generalisation of Theorem 3.32(ii).
Exercise 5.13. Let X be a topological vector space and suppose that C is a nonempty,
compact, convex subset of X. Prove that (F, ⊇), the collection of closed faces of C
ordered by reverse inclusion, is a nonempty, partially ordered set such that every chain
in F has an upper bound.
Since C is trivially a face of itself, F is nonempty, and reverse inclusion is easily seen
to be reﬂexive, transitive and antisymmetric, i.e., a partial order. If C is a chain in
F then consider F, the intersection of all the elements of C; it is immediate that this
is an upper bound for C if it is a face. The set F is closed and also nonempty: C is
compact, so every collection of closed subsets of C with the ﬁniteintersection property
has nonempty intersection and C has this property because it is a chain. If x, y ∈ C
and t ∈ (0, 1) are such that tx + (1 −t)y ∈ F then tx + (1 −t)y lies in every element
of C, so x and y lie in every element of C, as these elements are faces. Hence x and y
are in F and F is a face.
Exercise 5.14. Prove that the closed unit ball of c
0
has no extreme points.
Let x ∈
_
c
0
_
1
and note that there exists n
0
∈ N such that [x
n
0
[ <
1
2
. Deﬁne y and
z ∈ c
0
by setting y
n
= z
n
= x
n
for all n ,= n
0
, y
n
0
= x
n
0
−
1
2
and z
n
0
= x
n
0
+
1
2
. It is
immediate that y, z ∈
_
c
0
_
1
and x =
1
2
(y +z), showing that x is not an extreme point
of
_
c
0
_
1
.
Exercise 5.15. Let H be a Hilbert space. Prove that every unit vector in H is an
extreme point of the closed unit ball H
1
.
Suppose that x is a unit vector in H and let y, z ∈ H
1
and t ∈ (0, 1) be such that
104 Solutions to Exercises
x = ty + (1 −t)z. Then
1 = x
2
= t¸y, x) + (1 −t)¸z, x),
which implies that t =
1
2
and ¸y, x) = ¸z, x) = 1 (because 1 is an extreme point in F
1
).
As 1 = [¸y, x)[ y x 1 we have equality in the CauchySchwarz inequality and
so x and y are linearly dependent; since ¸y, x) = 1 we must have y = x. Hence x is an
extreme point of H
1
.
Deduce that every isometry in B(H) is an extreme point of the closed unit ball B(H)
1
.
Let V ∈ B(H) be a isometry; it is immediate that V  = 1 and so V ∈ B(H)
1
. Let
S, T ∈ B(H)
1
and t ∈ (0, 1) be such that V = tS + (1 − t)T and let x ∈ H be a unit
vector. Then V x = x = 1, so V x is a unit vector, and Sx, Tx 1, so the
identity V x = tSx + (1 −t)Tx implies that Sx = Tx = V x, by the ﬁrst part. As this
holds for all unit vectors in H we have that S = T = V , so V is an extreme point of
B(H)
1
, as claimed.
Exercise 5.16. Let C be a convex subset of a topological vector space X. Prove that
if x ∈ C and y ∈ C
◦
, the interior of C, then tx + (1 −t)y ∈ C
◦
for all t ∈ [0, 1).
Since y ∈ C
◦
, there exists an open set U such that y ∈ U ⊆ C; if t ∈ [0, 1) then
tx + (1 − t)U is open (since z → z + tx and z → (1 − t)z are homeomorphisms) and
contained in C (since C is convex).
[It follows that C is the sequential closure of C
◦
, if this set is nonempty: for every
x ∈ C there exists a sequence (x
n
)
n1
⊆ C
◦
such that x
n
→ x.]
Prove also that the interior C
◦
and the extremal boundary ∂
e
C are disjoint (as long as
X ,= ¦0¦).
Suppose for contradiction that x ∈ C
◦
∩∂
e
C and let U be an open set with x ∈ U ⊆ C.
Let u ∈ X be a nonzero vector and note that function f : R → X; λ → x + λu is
continuous, so f
−1
(U) is an open set containing 0. Hence there exists δ > 0 such that
x ±δu ∈ U, but then, since x ∈ ∂
e
C,
x =
1
2
(x + δu) +
1
2
(x −δu) =⇒ x + δu = x −δu =⇒ u = 0;
this contradiction give the result.
Exercise 5.17. Let C ⊆ R
n
be compact and convex. Prove that every element of C
can be written as a convex combination of at most (n + 1) elements of ∂
e
C.
If n = 1 then C is connected, so is a compact interval and the claim is clear. Now
suppose the result holds for some n and let C ⊆ R
n+1
be compact and convex. If every
set of n+1 points in C is linearly dependent then lin C is at most ndimensional (since
C is a spanning set which contains no more than n linearly independent points) so
C ⊆ lin C and the result follows by the inductive hypothesis. Otherwise, there exists
a linearly independent subset of C containing n +1 points; the convex hull of this set,
Solutions to Exercises 105
being homeomorphic to the standard n + 1simplex
¦(λ
1
, . . . , λ
n+1
) ∈ R
n+1
: λ
1
, . . . , λ
n+1
0,
n+1
i=1
λ
i
= 1¦,
has nonempty interior, and therefore so does C. Suppose now that x ∈ C ¸ C
◦
;
Theorem 3.32(i) gives a continuous linear functional f such that f(c) < f(x) for all
c ∈ C
◦
and so f(c) f(x) for all c ∈ C, by Exercise 5.16. If
D := ¦c ∈ C : f(c) = f(x)¦ = (x + ker f) ∩ C
then D is a face of C which lies in an ndimensional hyperplane, so the result follows by
the inductive hypothesis; recall that ∂
e
D ⊆ ∂
e
C. Finally, if x ∈ C
◦
then let y ∈ ∂
e
C
and consider the line L through x and y; this is a closed set and so C ∩ L is compact
and convex, so is an interval. As y is an extreme point of C, it must be one endpoint of
this interval; let z be the other and note that z ∈ C¸C
◦
, since C
◦
∩L ⊆ (C∩L)
◦
(where
the second interior is taken with respect to the relative topology on C ∩ L). Hence x
is an interior point of the interval, so there exists t ∈ (0, 1) such that x = ty +(1 −t)z;
furthermore, by the previous working, z is a convex combination of at most n points
in ∂
e
C and the result follows.
[This result is due to Carath´eodory [4].]
Solutions to Exercises 6
Exercise 6.1. Prove that the vector space L
1
(R) is a commutative Banach algebra when
equipped with the convolution product and the norm f
1
:=
_
[f[.
We take it for granted that L
1
(R) is a normed space and that (f, g) → f ⋆ g is a
welldeﬁned, bilinear map. It remains to prove that this multiplication is commutative
and associative and that L
1
(R) is complete.
For commutativity and associativity note that the Fourier transform f →
ˆ
f is
injective on L
1
(R), that
f ⋆ g =
ˆ
f
ˆ
g =
ˆ
g
ˆ
f =
g ⋆ f
and that
f ⋆ (g ⋆ h) =
ˆ
f
g ⋆ h =
ˆ
f
_
ˆ
g
ˆ
h
_
=
_
ˆ
f
ˆ
g
_
ˆ
h =
f ⋆ g
ˆ
h =
(f ⋆ g) ⋆ h
by the convolution theorem. (This theorem, which asserts the existence of f ⋆ g, is an
easy application of the theorems of Fubini and Tonelli; see [17, ¸¸ 26.15–26.16]).
For the proof of completeness we use Banach’s criterion. If (f
k
)
k1
⊆ L
1
(R)
is such that
∞
k=1
_
[f
k
[ < ∞ then the monotoneconvergence theorem applied to
(
n
k=1
[f
k
[)
n1
yields the convergence almost everywhere of
∞
k=1
[f
k
(x)[ to an inte
grable function and the fact that
∞
k=1
_
[f
k
[ =
_
∞
k=1
[f
k
[. Hence f(x) =
∞
k=1
f
k
(x)
is (absolutely) convergent almost everywhere, and applying the dominatedconvergence
theorem to (
n
k=1
f
k
)
n1
(with dominating function
∞
k=1
[f
k
[ ) gives that f ∈ L
1
(R),
as required.
106 Solutions to Exercises
Prove further that this algebra is not unital.
For n 1 let
f
n
: R →R; x →
n
√
2π
e
−
1
2
n
2
x
2
and note that
_
f
n
= 1. The continuous version of the dominatedconvergence theorem
gives that g
t
−g
1
→ 0 as t → 0 for all g ∈ C
00
(R), where g
t
: s → g(s−t). As C
00
(R)
is dense in L
1
(R) it follows that g
t
− g
1
→ 0 as t → 0 for all g ∈ L
1
(R); note that
t → g
t
maps L
1
(R) to itself, that g
t
−h
t

1
= g −h
1
and that
g
t
−g
1
g
t
−h
t

1
+h
t
−h
1
+h −g
1
= 2h −g
1
+h
t
−h
1
.
If g ∈ L
1
(R) then
g ⋆ f
n
−g
1
=
_
R
¸
¸
_
R
g(t −s)f
n
(s) −g(t)f
n
(s) ds
¸
¸
dt
_
R
_
R
[g
s
−g[(t) dt nf
1
(ns) ds =
_
R
g
r/n
−g
1
f
1
(r) dr → 0
as n → ∞, by the dominatedconvergence theorem: the map t → g
t
−g
1
is bounded
and f
1
∈ L
1
(R).
Suppose that there exists 1 ∈ L
1
(R) such that 1 ⋆ f = f for all f ∈ L
1
(R).
Then 1 ⋆ f
n
→ 1 as n → ∞ (in L
1
(R) ) but (1 ⋆ f
n
)(x) = f
n
(x) → 0 as n → ∞ almost
everywhere, which implies that 1(x) = 0 almost everywhere. This is clearly impossible.
Exercise 6.2. Let X be a Hausdorﬀ, locally compact space. Prove that C
0
(X)
u
, the
unitization of the algebra of continuous functions on X that vanish at inﬁnity, is topo
logically isomorphic to C(
˙
X), the algebra of continuous functions on
˙
X, the onepoint
compactiﬁcation of X.
Recall from the solution of Exercise 2.2 that the map f → f[
X
is a bijection between
I := ¦f ∈ C(
˙
X) : f(∞) = 0¦ and C
0
(X). Since f − f(∞)1 ∈ I for all f ∈ C(
˙
X) the
map
C(
˙
X) → C
0
(X)
u
; f → (f −f(∞)1)[
X
+ f(∞)1
is a bijection, and it is readily veriﬁed that it is an algebra homomorphism. Finally,
(f −f(∞)1)[
X

∞
+[f(∞)[ f[
X

∞
+ 2[f(∞)[ 3f
∞
so this map is a continuous bijection between Banach spaces, hence a homeomorphism
(by Theorem 2.14).
Exercise 6.3. Let A = C[z] denote the unital algebra of complex polynomials and let
p := sup¦[p(α)[ : [α[ 1¦ for all p ∈ A. Show that (A,  ) is a unital, normed
algebra which is not complete.
Proving that   is a submultiplicative norm on A is trivial.
Consideration of degree shows that G(A) = (C ¸ ¦0¦)1, so p
n
∈ A ¸ G(A) for all
n 1, where p
n
(z) := 1 +z/n. If A is complete then G(A) is open, but p
n
−1 → 0
as n → ∞ and 1 ∈ G(A), so A¸ G(A) is not closed. This gives the result.
Solutions to Exercises 107
Exercise 6.4. Let A be a (nonunital) Banach algebra such that every element is nil
potent (i.e., for all a ∈ A there exists n ∈ N such that a
n
= 0). Prove that A is uniformly
nilpotent: there exists N ∈ N such that a
N
= 0 for all a ∈ A.
Let E
n
= ¦a ∈ A : a
n
= 0¦ for all n ∈ N; these sets are closed and A =
n∈N
E
n
so,
by the Baire category theorem, E
◦
N
,= ∅ for some N ∈ N. Hence there exists a
0
∈ A
and ε > 0 such that a −a
0
 < ε implies that a
N
= 0. Let b ∈ A¸ ¦0¦ and note that
p(t) := (a
0
+ tb)
N
= a
N
0
+ + t
N
b
N
= 0 ∀ t ∈
_
−ε/b, ε/b
_
because (a
0
+ tb) − a
0
 = [t[b < ε. Hence 0 = p
(N)
(0) = N! b
N
, where p
(N)
is the
Nth derivative of p, and so b
N
= 0, as required.
Exercise 6.5. Let A be a unital Banach algebra over C and let e
a
:=
∞
n=0
a
n
/n! for
all a ∈ A. Prove that e
a+b
= e
a
e
b
if a and b commute.
Note ﬁrst that e
a
converges absolutely, so converges. As
(a + b)
n
=
n
k=0
_
n
k
_
a
k
b
n−k
=
1
n!
n
k=0
a
k
k!
b
n−k
(n −k)!
if a, b ∈ A commute, it suﬃces to prove that if a =
∞
n=0
a
n
and b =
∞
n=0
b
n
are
absolutely convergent in A then ab =
∞
n=0
c
n
, where c
n
=
n
k=0
a
k
b
n−k
.
To see this, note that
n
k=0
k
l=0
a
l
b
k−l
= a
0
b
0
+ (a
0
b
1
+ a
1
b
0
) + + (a
0
b
n
+ + a
n
b
0
)
= a
0
n
l=0
b
l
+ a
1
n−1
l=0
b
l
+ + a
n
b
0
=
_
n
k=0
a
k
_
b +
n
k=0
a
k
_
n−k
l=0
b
l
−b
_
so it suﬃces to prove that r
n
=
n
k=0
a
k
d
n−k
=
n
k=0
a
n−k
d
k
→ 0 as n → ∞, where
d
k
=
k
l=0
b
l
−b. Let ε > 0 and choose N such that d
n
 < ε for all n N; if n > N
then
_
_
_
n
k=0
a
n−k
d
k
_
_
_
_
_
_
N
k=0
a
n−k
d
k
_
_
_ +
n
k=N+1
a
n−k
 d
k

_
_
_
N
k=0
a
n−k
d
k
_
_
_ + ε
∞
k=1
a
k
 → ε
∞
k=1
a
k

as n → ∞. As ε > 0 is arbitrary we have the result.
Deduce that e
a
is invertible.
Since e
0
= 1 we have that e
a
e
−a
= 1 = e
−a
e
a
, i.e., (e
a
)
−1
= e
−a
, for all a ∈ A.
Prove further that f : λ → e
λa
is holomorphic everywhere, with f
′
(λ) = af(λ) = f(λ)a,
for all a ∈ A.
108 Solutions to Exercises
Let a ∈ A, λ ∈ C and h ∈ C ¸ ¦0¦. Then
e
(λ+h)a
−e
λa
h
−ae
λa
=
e
ha
−1 −ha
h
e
λa
→ 0
as h → 0, because
_
_
_
_
e
ha
−1 −ha
h
_
_
_
_
∞
n=2
[h[
n−1
a
n
n!
=
e
h a
−1
[h[
−a
and
d
dz
e
za
¸
¸
¸
¸
z=0
= a.
Exercise 6.6. Let A be a unital Banach algebra over C and let a, b ∈ A. Use the
identity (ab)
n
= a(ba)
n−1
b to prove that ab and ba have the same spectral radius.
Note that
(ab)
n

1/n
a
1/n
(ba)
n−1

1/n
b
1/n
= a
1/n
_
(ba)
n−1

1/(n−1)
_
(n−1)/n
b
1/n
→ ν(ba)
and so ν(ab) ν(ba). Exchanging the rˆoles of a and b gives the result.
Exercise 6.7. Let A be a unital Banach algebra over C. Suppose that there exists
K > 0 such that a Kν(a) for all a ∈ A, where ν(a) denotes the spectral radius of
a. Prove that A is commutative.
The function g : C → A; λ → e
λa
be
−λa
is holomorphic everywhere and bounded,
because
g(λ) Kν(e
λa
be
−λa
) = Kν(be
λa
e
−λa
) = Kν(b).
Hence g is constant, so 0 = g
′
(0) = ag(0) −g(0)a = ab −ba, as required.
Exercise 6.8. Let A be a Banach algebra and suppose that (x
p
)
p∈P
, (y
q
)
q∈Q
⊆ A are
absolutely summable. Prove that
p∈P
q∈Q
x
p
y
q
=
(p,q)∈P×Q
x
p
y
q
=
q∈Q
p∈P
x
p
y
q
.
We prove only the ﬁrst equality; the second follows by symmetry. Let ε > 0 and note
that (x
p
y
q
)
(p,q)∈P×Q
is absolutely summable (as x
p
y
q
 x
p
 y
q
) so there exist
ﬁnite sets R
0
⊆ P, S
0
⊆ Q and T
0
⊆ P Q such that
_
_
p∈R
x
p
−
p∈P
x
p
_
_
<
ε
1 +y
1
for all ﬁnite R ⊆ P such that R ⊇ R
0
(where y
1
=
q∈Q
y
q
 )
_
_
q∈S
y
q
−
q∈Q
y
q
_
_
<
ε
1 +x
1
for all ﬁnite S ⊆ Q such that S ⊇ S
0
,
Solutions to Exercises 109
(where x
1
=
p∈P
x
p
 ) and
_
_
(p,q)∈T
x
p
y
q
−
(p,q)∈P×Q
x
p
y
q
_
_
< ε for all ﬁnite T ⊆ P Q such that T ⊇ T
0
.
If R ⊆ P and S ⊆ Q are ﬁnite sets such that T = R S ⊇ T
0
∪ (R
0
S
0
) then
_
_
(p,q)∈P×Q
x
p
y
q
−
p∈P
x
p
q∈Q
y
q
_
_
_
_
(p,q)∈P×Q
x
p
y
q
−
(p,q)∈T
x
p
y
q
_
_
+
p∈R
x
p

_
_
q∈S
y
q
−
q∈Q
y
q
_
_
+
_
_
p∈R
x
p
−
p∈P
x
p
_
_
y
1
< ε + ε + ε.
As ε > 0 is arbitrary, we have the result.
Solutions to Exercises 7
Exercise 7.1. Let A = C(X), where X is a compact, Hausdorﬀ space. Prove that the
map ǫ: X → Φ(A); x → ǫ
x
is a homeomorphism, where ǫ
x
(f) = f(x) for all x ∈ X and
f ∈ C(X).
Note that ǫ is continuous if and only if
ˆ
f
◦
ǫ: x → ǫ
x
(f) = f(x) is continuous for all
f ∈ A, by deﬁnition of the Gelfand topology. Furthermore, if x, y ∈ X are distinct
then there exists f ∈ A such that f(x) ,= f(y), i.e., ǫ
x
(f) ,= ǫ
y
(f), by Urysohn’s lemma.
As X is compact and Φ(A) is Hausdorﬀ, ǫ is a homeomorphism between X and ǫ(X);
it remains to prove that ǫ is surjective.
If φ ∈ Φ(A) ¸ ǫ(X) then for all x ∈ X there exists f ∈ A such that f ∈ ker φ but
f ,∈ ker ǫ
x
(i.e., f(x) ,= 0). By compactness there exist f
1
, . . . , f
n
∈ ker φ such that
n
i=1
¦x ∈ X : f
i
(x) ,= 0¦ = X, but then f :=
n
i=1
¯
f
i
f
i
∈ ker φ and f(x) > 0 for all
x ∈ X, whence ker φ = A, a contradiction.
Exercise 7.2. Prove that if A is a unital Banach algebra generated by a single element
(i.e., there exists a ∈ A such that ¦p(a) : p(z) ∈ C[z]¦ is dense in A) then Φ(A) is
homeomorphic to σ(a).
Note that ˆ a: Φ(A) → σ(a) is a continuous function from a compact space onto a
Hausdorﬀ space. If φ, ψ ∈ Φ(A) are such that ˆ a(φ) = ˆ a(ψ) then φ(p(a)) = ψ(p(a)) for
any complex polynomial p (as φ(1) = 1 and φ(a
n
) = φ(a)
n
for all n ∈ N so φ(p(a)) =
p(φ(a)) = p(ˆ a(φ)) ). Hence φ = ψ, by continuity and the density of ¦p(a) : p(z) ∈ C[z]¦
in A, so ˆ a is a homeomorphism.
Deduce that Φ(A) is homeomorphic to
¯
D := ¦z ∈ C : [z[ 1¦ if A = A(D) is the disc
algebra.
110 Solutions to Exercises
If f ∈ A(D) then f is uniformly continuous, so if ε > 0 there exists r ∈ (0, 1) such
that [f(z) − f(rz)[ < ε/2 for all z ∈
¯
D (because [z − rz[ 1 − r). Since z → f(rz)
is holomorphic on ¦z ∈ C : [z[ < r
−1
¦, its Taylor series is uniformly convergent on
¯
D,
so there exists a polynomial p such that [p(z) − f(rz)[ < ε/2 for all z ∈
¯
D. Hence
f −p
∞
< ε and polynomials in id:
¯
D →C; z → z are dense in A.
Exercise 7.3. Let A be a unital Banach algebra that is generated by one element,
a, and let λ ,∈ σ(a). Show there exists a sequence of polynomials (p
n
)
n1
such that
p
n
(z) → (λ −z)
−1
uniformly for all z ∈ σ(a).
Since λ ,∈ σ(a), the inverse (λ1 −a)
−1
is an element of A, so there exists a sequence of
polynomials (p
n
)
n1
such that p
n
(a) → (λ1 −a)
−1
. If z ∈ σ(a) then, by Exercise 7.2,
there exists φ ∈ Φ(A) such that φ(a) = z; since φ is a unital algebra homomorphism,
it follows that
[p
n
(z) −(λ −z)
−1
[ = [φ(p
n
(a) −(λ1 −a)
−1
)[ p
n
(a) −(λ1 −a)
−1

and therefore p
n
(z) → (λ −z)
−1
uniformly on σ(a) as n → ∞.
Deduce that the complement of σ(a) is connected.
Suppose for contradiction that C ¸ σ(a) has a maximally connected component, C,
which is bounded (and nonempty); ﬁx λ ∈ C and choose (p
n
)
n1
as above. Note
ﬁrst that ∂C := C ¸ C
◦
, the (topological) boundary of C, is contained in σ(a): if
x ∈ C ∩ (C ¸ σ(a)) then there exist a sequence (x
n
)
n1
⊆ C such that x
n
→ x and
ε > 0 such that B
C
ε
(x)∩σ(a) = ∅, as σ(a) is closed. Hence B
C
ε
(x)∩C ,= ∅ and therefore
B
C
ε
(x) ⊆ C, as C is maximally connected, so x ∈ C
◦
. It follows that C is a bounded
region in C (i.e., an open, connected set): if x ∈ C then x ,∈ σ(a), so x ,∈ ∂C and thus
x ∈ C
◦
.
By the maximummodulus theorem [16, Theorem 5.20],
sup¦[p
n
(z) −p
m
(z)[ : z ∈ C¦ = sup¦[p
n
(z) −p
m
(z)[ : z ∈ ∂C¦
sup¦[p
n
(z) −p
m
(z)[ : z ∈ σ(a)¦ → 0
as m, n → ∞, so (p
n
)
n1
is uniformly convergent on C and its limit, f, is continuous
there and holomorphic in C. Then g : z → (λ − z)f(z) − 1 is holomorphic in C and
g(z) = 0 for all z ∈ ∂C, so g ≡ 0, by the identity theorem, contradicting the fact that
g(λ) = −1.
Exercise 7.4. Let A be a commutative, unital Banach algebra. Prove that the Gelfand
transform on A is isometric if and only if a
2
 = a
2
for all a ∈ A.
If a
2
 = a
2
then a
2
n
 = a
2
n
for all n ∈ N, and so
ˆ a
∞
= ν(a) = lim
n→∞
a
n

1/n
= lim
n→∞
a
2
n

1/2
n
= a.
Solutions to Exercises 111
Conversely, if ν(a) = ˆ a
∞
= a then there exists λ ∈ σ(a) such that [λ[ = a.
Hence λ
2
∈ σ(a
2
) (by the spectralmapping theorem for polynomials) and so
a
2
 a
2
= [λ[
2
= [λ
2
[ ν(a
2
) a
2
.
Exercise 7.5. Let A be a Banach algebra and B a semisimple, commutative, unital
Banach algebra. Prove that if φ: A → B is a homomorphism then φ is continuous.
By the closedgraph theorem, it suﬃces to take a sequence (a
n
)
n1
⊆ A such that
a
n
→ a and φ(a
n
) → b and prove that b = φ(a). Let ψ ∈ Φ(B) and note that ψ
and ψ
◦
φ are continuous, by Proposition 6.1; note that φ extends to A
u
by setting
φ(α1 + a) = α + φ(a) so we may assume without loss of generality that A is unital.
Hence
ψ(b) = lim
n→∞
ψ(φ(a
n
)) = ψ(φ(a))
for all ψ ∈ Φ(B), and therefore b −φ(a) ∈ J(B), which gives the result.
Exercise 7.6. Let A = C
1
[0, 1], equipped with the norm f := f
∞
+ f
′

∞
. Prove
that A is a semisimple, commutative, unital Banach algebra and ﬁnd its character space.
To prove completeness, let (f
n
)
n1
⊆ A be Cauchy, which implies that (f
n
)
n1
and
(f
′
n
)
n1
are Cauchy, so convergent, sequences in C[0, 1]. Let f = lim
n→∞
f
n
, g =
lim
n→∞
f
′
n
and note that
_
t
0
g =
_
t
0
lim
n→∞
f
′
n
= lim
n→∞
_
t
0
f
′
n
= lim
n→∞
f
n
(t) −f
n
(0) = f(t) −f(0),
so f
′
(t) = g(t) for all t ∈ [0, 1] and
f −f
n
 = f −f
n

∞
+f
′
−f
′
n

∞
= f −f
n

∞
+g −f
′
n

∞
→ 0,
as required.
To see that A is semisimple, note that Φ(A) = ¦ε
x
: x ∈ X¦ (which may be proved
as for C(X)), so J(A) =
x∈X
¦f ∈ A : f(x) = 0¦ = ¦0¦.
Prove that I = ¦f ∈ A : f(0) = f
′
(0) = 0¦ is a closed ideal in A such that A/I is a
twodimensional algebra with onedimensional radical.
Note that
j : C
1
[0, 1] → M
2
(C); f →
_
f(0) f
′
(0)
0 f(0)
_
is a continuous algebra homomorphism with kernel I and 2dimensional image. Every
linear functional on this image must have the form
φ:
_
a b
0 a
_
→ λa + µb ∀ a, b ∈ C,
where λ, µ ∈ C. If φ(1) = 1 then λ = 1, and multiplicativity forces µ = 0, so A/I has
one maximal ideal (viz. id +I).
112 Solutions to Exercises
What do you notice about A and A/I?
This example shows that a quotient of a semisimple algebra need not be semisimple.
Exercise 7.7. Prove that the Banach space ℓ
1
(Z) is a commutative, unital Banach
algebra when equipped with the multiplication
a ⋆ b: Z →C; n →
m∈Z
a
m
b
n−m
_
a, b ∈ ℓ
1
(Z)
_
.
Note that
n∈Z
[b
n
[ < ∞ implies that (b
n
)
n∈Z
is bounded, and so if a, b ∈ ℓ
1
(Z) then
m∈Z
a
m
b
n−m
is (absolutely) summable for all n ∈ Z. Furthermore,
a
1
b
1
=
m∈Z
[a
m
[
n∈Z
[b
n
[ =
m∈Z
n∈Z
[a
m
[ [b
n
[
=
m∈Z
p∈Z
[a
m
[ [b
p−m
[
p∈Z
[(a ⋆ b)
p
[ = a ⋆ b
1
which shows that a ⋆ b ∈ ℓ
1
(Z) and  
1
is submultiplicative on ℓ
1
(Z). (To see that
the reversal of order is valid, note that
r∈R
s∈S
x
r
y
s
=
(r,s)∈R×S
x
r
y
s
for any
absolutely summable (x
r
)
r∈R
, (y
s
)
s∈S
in a Banach algebra.) Furthermore,
(a ⋆ b)
n
=
m∈Z
a
m
b
n−m
=
p∈Z
b
p
a
n−p
= (b ⋆ a)
n
so ⋆ is commutative; bilinearity is readily veriﬁed, and if a, b, c ∈ ℓ
1
(Z) and p ∈ Z then
_
(a ⋆ b) ⋆ c
_
p
=
n∈Z
_
m∈Z
a
m
b
n−m
_
c
p−n
=
m∈Z
a
m
n∈Z
b
n−m
c
p−n
=
m∈Z
a
m
r∈Z
b
r
c
p−m−r
=
m∈Z
a
m
(b ⋆ c)
p−m
=
_
a ⋆ (b ⋆ c)
_
p
.
If 1 ∈ ℓ
1
(Z) is deﬁned by setting 1
0
= 1 and 1
n
= 0 for all n ∈ Z ¸ ¦0¦ then 1 ∈ ℓ
1
(Z),
1
1
= 1 and
(1 ⋆ a)
n
=
m∈Z
1
m
a
n−m
= a
n
∀ n ∈ Z, a ∈ ℓ
1
(Z).
Exercise 7.8. Let δ ∈ ℓ
1
(Z) be such that δ
1
= 1 and δ
n
= 0 if n ,= 1. Prove that
a =
n∈Z
a
n
δ
n
for all a ∈ ℓ
1
(Z).
If ε > 0 choose a ﬁnite set A
0
⊆ Z such that
n∈A
0
[a
n
[ > a
1
−ε and note that
a −
n∈A
a
n
δ
n

1
= sup
_
n∈B\A
[a
n
[ : B is a ﬁnite subset of Z
_
= a
1
−
n∈A
[a
n
[ < ε
for any ﬁnite set A ⊆ Z such that A ⊇ A
0
.
Solutions to Exercises 113
Deduce that the character space of ℓ
1
(Z) is homeomorphic to T := ¦z ∈ C : [z[ = 1¦
and (with this identiﬁcation) the Gelfand transform on ℓ
1
(Z) is the map
Γ: ℓ
1
(Z) → C(T); Γ(a)(λ) =
n∈Z
a
n
λ
n
∀ λ ∈ T, a ∈ ℓ
1
(Z).
Let φ ∈ Φ(ℓ
1
(Z)) and note that φ(δ
n
) = φ(δ)
n
for all n ∈ Z, so
φ(a) =
n∈Z
a
n
φ(δ
n
) =
n∈Z
a
n
λ
n
∀ a ∈ ℓ
1
(Z),
where λ = φ(δ); as φ is continuous, if x
s
→ x then φ(x
s
) → φ(x). Furthermore, λ ∈ T
because
[φ(δ)[ φ δ
1
= 1 and [φ(δ)
−1
[ = [φ(δ
−1
)[ φ δ
−1

1
= 1.
Conversely, if λ ∈ T then φ: a →
n∈Z
a
n
λ
n
deﬁnes a character. To see this, note that
this series is absolutely summable, φ is linear and bounded (because [φ(a)[ a
1
)
and
φ(δ
m
⋆ δ
n
) = φ(δ
m+n
) = λ
m+n
= λ
m
λ
n
= φ(δ
m
)φ(δ
n
) ∀ m, n ∈ Z;
it follows by continuity that φ(a)φ(b) = φ(a ⋆ b) for all a, b ∈ ℓ
1
(Z). Hence
ˆ
δ : Φ(ℓ
1
(Z)) →T; φ → φ(δ)
is a continuous bijection from a compact space to a Hausdorﬀ space, so a homeomor
phism.
Finally, note that f ∈ C
_
Φ(ℓ
1
(Z))
_
corresponds to f
◦
ˆ
δ
−1
∈ C(T) and so the
Gelfand map becomes
Γ: ℓ
1
(Z) → C(T); a → ˆ a
◦
ˆ
δ
−1
,
i.e.,
Γ(a)(λ) =
_
ˆ a
◦
ˆ
δ
−1
_
(λ) = ˆ a
_
ˆ
δ
−1
(λ)
_
=
ˆ
δ
−1
(λ)(a) =
n∈Z
a
n
λ
n
.
Exercise 7.9. Let f : T →C be continuous and such that
n∈Z
[
ˆ
f(n)[ < ∞, where
ˆ
f(n) :=
1
2π
_
π
−π
f(e
it
)e
−int
dt (n ∈ Z).
Prove that if f(z) ,= 0 for all z ∈ T then g = 1/f : T → C; z → 1/f(z) satisﬁes
n∈Z
[ˆ g(n)[ < ∞.
Exercise 7.8 implies that A = Γ
_
ℓ
1
(Z)
_
consists of those g ∈ C(T) such that
g(z) =
n∈Z
a
n
z
n
(z ∈ T) and
n∈Z
[a
n
[ < ∞.
114 Solutions to Exercises
Since
n∈Z
a
n
z
n
is uniformly convergent on T for such g we see that
ˆ g(m) =
n∈Z
a
n
2π
_
π
−π
e
int
e
−imt
dt = a
m
(m ∈ Z)
and so A consists of those g ∈ C(T) such that
n∈Z
[ˆ g(n)[ < ∞. In particular, f ∈ A
and the condition f(z) ,= 0 for all z ∈ T is equivalent to the fact that φ(f) ,= 0 for all
φ ∈ Φ(A), i.e., 0 ,∈ σ(f). Hence 1/f ∈ A, as required.
Exercise 7.10. Let T = ∂D = ¦z ∈ C : [z[ = 1¦ and A = ¦f ∈ C(
¯
D) : f[
T
∈ A(D)[
T
¦:
A consists of those continuous functions on the closed unit disc
¯
D that agree on the unit
circle T with a continuous function on
¯
D that is holomorphic in D.
[A corollary of the maximummodulus theorem [16, Theorem 5.20] will be useful: if
f ∈ A(D) then f
∞
:= sup¦[f(z)[ : [z[ 1¦ = sup¦[f(z)[ : [z[ = 1¦ =: f[
T

∞
.]
(i) Show that A is a Banach algebra when equipped with the supremum norm.
Let (f
n
)
n1
⊆ A be such that f
n
→ f ∈ C(
¯
D). For n 1 let g
n
∈ A(D) be such
that f
n
[
T
= g
n
[
T
and note that
g
n
−g
m

∞
= (g
n
−g
m
)[
T

∞
= (f
n
−f
m
)[
T

∞
f
n
−f
m

∞
→ 0
as m, n → ∞, so (g
n
)
n1
converges to g ∈ A(D). Since
g[
T
= lim
n→∞
g
n
[
T
= lim
n→∞
f
n
[
T
= f[
T
,
we see that f ∈ A, as required.
(ii) Prove that I = ¦f ∈ A : f[
T
= 0¦ is a closed ideal in A and that A = A(D) ⊕I.
Suppose that f ∈ A; if g, h ∈ A(D) are such that f[
T
= g[
T
= h[
T
then g−h
∞
=
(g − h)[
T

∞
= 0 and so g = h. Thus to every f ∈ A there exists a unique
j(f) ∈ A(D) such that f[
T
= j(f)[
T
; the map
j : A → A(D); f → j(f)
is a normdecreasing algebra homomorphism (as j(f)
∞
= f[
T

∞
f
∞
). In
particular, I = ¦f ∈ A : f[
T
= 0¦ = ker j is a closed ideal. Furthermore, if f ∈ A
then f −j(f) ∈ I so f ∈ I +A(D) and if f ∈ A(D) ∩I then f
∞
= f[
T

∞
= 0;
thus A = A(D) ⊕I.
(iii) Prove that
i : I → C
0
(D); f → f[
D
is an isometric isomorphism. Deduce that I
u
is topologically isomorphic to C(
˙
D),
where
˙
D is the onepoint compactiﬁcation of D.
Solutions to Exercises 115
If f ∈ I and ε > 0 then
¦z ∈ D : [f(z)[ ε¦ = ¦z ∈
¯
D : [f(z)[ ε¦
is a closed subset of
¯
D and is therefore compact; hence f[
D
∈ C
0
(D). Since
f[
D

∞
= f
∞
the map i is isometric; it remains to prove that i is surjective.
Let f ∈ C
0
(D) and deﬁne g :
¯
D → C by setting g[
D
= f and g[
T
= 0. Let U ⊆ C
be open; if 0 ,∈ U then g
−1
(U) = f
−1
(U) is an open subset of D, and therefore of
¯
D, so suppose that 0 ∈ U. Then g
−1
(U) = f
−1
(U) ∪ T and, since 0 ∈ U, there
exists ε > 0 such that B
C
ε
(0) ⊆ U. Thus f
−1
(U) ⊇ f
−1
_
B
C
ε
(0)
_
and f
−1
_
B
C
ε
(0)
_
has compact complement, hence
D ¸ f
−1
(U) ⊆ D ¸ f
−1
_
B
C
ε
(0)
_
⊆ B
C
r
[0]
for some r ∈ (0, 1). From this we see that f
−1
(U) ⊇ D ¸ B
C
r
[0] and therefore
g
−1
(U) = f
−1
(U) ∪
_
¯
D ¸ B
C
r
[0]
_
is open. This shows that g is continuous, as
required.
The deduction follows immediately from Exercise 6.2.
(iv) Prove that if φ ∈ Φ(A) is such that ker φ ⊇ I then φ =
˜
φ
◦
j, where
˜
φ ∈ Φ(A(D)).
Deduce that φ = ε
z
◦
j for some z ∈
¯
D (where ε
z
: A(D) →C; f → f(z) ).
Note that if f, g ∈ A are such that j(f) = j(g) then f − g ∈ I so φ(f) = φ(g).
Hence
˜
φ: A(D) →C; j(f) → φ(f)
is well deﬁned. It is immediate that
˜
φ is an algebra homomorphism such that
φ =
˜
φ
◦
j, and since φ is nonzero, so is
˜
φ. The deduction is immediate, as
Φ(A(D)) = ¦ε
z
: z ∈
¯
D¦ by the solution to Exercise 7.2.
(v) Let
¯
D
1
and
¯
D
2
be two copies of the unit disc and let S
2
=
¯
D
1
∪
¯
D
2
_
∼ be the sphere
obtained by identifying each point on T
1
= ∂D
1
with the corresponding point on
T
2
= ∂D
2
. Deﬁne
T : S
2
→ Φ(A); z →
_
ε
z
◦
j (z ∈
¯
D
1
),
ε
z
(z ∈
¯
D
2
)
and prove that this is a welldeﬁned, continuous injection.
If z ∈ T
1
∪ T
2
then ε
z
(j(f)) = j(f)(z) = f(z) = ε
z
(f) for all f ∈ A, so this is
a good deﬁnition. As Φ(A) has the weak* topology, T is continuous if
ˆ
f
◦
T is
continuous for all f ∈ A, and since
_
ˆ
f
◦
T
_
(z) =
_
j(f)(z) (z ∈
¯
D
1
),
f(z) (z ∈
¯
D
2
),
this follows because f[
T
= j(f)[
T
. Since the identity function lies in A(D), ε
z
,= ε
w
and ε
z
◦
j ,= ε
w
◦
j for all distinct z, w ∈
¯
D. Finally, to see that ε
z
,= ε
w
◦
j for all
z, w ∈ D, note that there exists f ∈ I such that f(z) = 1 (by Urysohn’s lemma)
and so ε
z
(f) = 1 ,= 0 = (ε
w
◦
j)(f). Hence T is an injection.
116 Solutions to Exercises
(vi) Prove that Φ(A) is homeomorphic to the sphere S
2
.
If φ ∈ Φ(A) is such that ker φ ⊇ I then φ = ε
z
◦
j for some z ∈
¯
D. Otherwise
φ[
I
,= 0; note that
˙
φ: I
u
→C; f + α1 → φ(f) + α
is a character of I
u
∼
= C(
˙
D) and so has the form ε
z
for some z ∈
˙
D. Since
φ[
I
=
˙
φ[
I
is nonzero, z ,= ∞ and φ[
I
= ε
z
for some z ∈ D. Choose f ∈ I such
that f(z) = 1 (such exists by Urysohn’s lemma) and note that f(g −φ(g)1) ∈ I
for all g ∈ A, so
g(z) −φ(g) = ε
z
_
f(g −φ(g)1)
_
= φ
_
f(g −φ(g)1)
_
= φ(f)φ(g −φ(g)1) = 0,
i.e., φ(g) = ε
z
(g). Hence T is surjective and we have the result (since Φ(A) and
S
2
are both compact, Hausdorﬀ spaces).
Solutions to Exercises A
Exercise A.1. Let X = ℓ
∞
and for all n ∈ N deﬁne δ
n
∈ X
∗
by setting δ
n
_
(x
k
)
k1
_
=
x
n
. Prove that (δ
n
)
n1
has no weak*convergent subsequence but that (δ
n
)
n1
has a
weak*convergent subnet.
Suppose for contradiction that (δ
n
) has a weak*convergent subsequence, say (δ
n
k
)
k1
.
Deﬁne x ∈ ℓ
∞
by setting
x
l
=
_
_
_
−1 l = n
2k−1
,
1 l = n
2k
,
0 otherwise.
Then
_
δ
n
k
(x
l
)
l1
_
k1
= (−1, 1, −1, 1, . . .) is not convergent, the desired contradiction.
For the second claim, note that (δ
n
)
n1
⊆ X
∗
1
, which is weak* compact by the
BanachAlaoglu theorem (Theorem 3.22). Hence the existence of a convergent subnet
follows from the (generalised) BolzanoWeierstrass theorem, Theorem A.5.
Bibliography
[1] W. Arveson, Subalgebras of C
∗
algebras, Acta Mathematica 123, 141–224 (1969).
[Cited on page 97.]
[2] G. Birkhoff, MooreSmith convergence in general topology, Annals of Mathe
matics 38, 39–56 (1937). [Cited on page 13.]
[3] B. Bollob´ as, Linear analysis, corrected printing, Cambridge University Press,
Cambridge, 1992. [Cited on pages 24, 35 and 39.]
[4] C. Carath´eodory,
¨
Uber den Variabilit¨atsbereich der Fourierschen Konstanten
von positiven harmonischen Funktionen, Rendiconti del Circolo Mathematico di
Palermo. Serie I 32, 193–217 (1911). [Cited on page 105.]
[5] E. Effros & Z.J. Ruan, Operator spaces, London Mathematical Society Mono
graphs 23, Oxford University Press, Oxford, 2000. [Cited on page 41.]
[6] D. B. Goodner, A projection of norm one may not exist, Mathematics Magazine
73, 334–335 (1964). [Cited on page 91.]
[7] K. M. Hoffman & R. A. Kunze, Linear algebra, second edition, Prentice Hall,
Inc., New Jersey, 1971. [Cited on page 4.]
[8] R. V. Kadison & J. R. Ringrose, Fundamentals of the theory of operator alge
bras. Volume I: elementary theory, Graduate Studies in Mathematics 15, American
Mathematical Society, Providence, Rhode Island, 1997. [Cited on page 68.]
[9] J. L. Kelley, General topology, D. Van Nostrand Co., Inc., Princeton, New Jersey,
1955. [Cited on pages 13 and 24.]
[10] E. Kreyszig, Introductory functional analysis with applications, John Wiley &
Sons, Inc., New York, 1978. [Cited on pages 18 and 31.]
[11] P. Lax, Functional analysis, John Wiley & Sons, Inc., New York, 2002. [Cited on
page 61.]
[12] E. McShane, Partial orderings and MooreSmith limits, American Mathematical
Monthly 59, 111 (1952). [Cited on page 13.]
117
118 Solutions to Exercises
[13] E. H. Moore & H. L. Smith, A general theory of limits, American Journal of
Mathematics 44, 102–121 (1922). [Cited on page 13.]
[14] G. K. Pedersen, Analysis now, revised printing, Graduate Texts in Mathematics
118, SpringerVerlag, New York, 1995. [Cited on pages 13, 18, 24 and 32.]
[15] G. Pisier, Introduction to Operator Space Theory, London Mathematical Society
Lecture Noes Series 294, Cambridge University Press, Cambridge, 2003. [Cited on
page 41.]
[16] H. A. Priestley, Introduction to complex analysis, revised edition, Oxford Uni
versity Press, Oxford, 1990. [Cited on pages 6, 36, 57, 72, 73, 110 and 114.]
[17] H. A. Priestley, Introduction to integration, Oxford University Press, Oxford,
1997. [Cited on pages 5, 11, 56 and 105.]
[18] W. Rudin, Real and complex analysis, third edition, McGrawHill, Inc., New York,
1987. [Cited on pages 31 and 56.]
[19] W. Rudin, Functional analysis, second edition, McGrawHill, Inc., New York, 1991.
[Cited on pages 26, 68 and 94.]
[20] R. Schatten, A theory of crossspaces, Annals of Mathematics Studies 26, Prince
ton University Press, Princeton, 1950. [Cited on page 97.]
[21] G. F. Simmons, Introduction to topology and modern analysis, McGrawHill, Inc.,
New York, 1963. [Cited on pages 24, 26 and 39.]
[22] W. A. Sutherland, Introduction to metric and topological spaces, corrected print
ing, Oxford University Press, Oxford, 1981. [Cited on pages 11, 24, 26 and 35.]
[23] A. Tychonoff, Ein Fixpunktsatz, Mathematische Annalen 111, 767776 (1935).
[Cited on page 100.]
[24] G. F. VincentSmith, B4: Analysis, Mathematical Institute notes, University of
Oxford, Oxford, 1991. [Cited on pages iii, 3, 4, 7, 14, 18 and 37.]
[25] G. F. VincentSmith, An elementary analytical proof of the spectral radius for
mula, Expositiones Mathematicae 16, 383–384 (1998). [Cited on page 61.]
[26] A. J. Weir, Lebesgue integration and measure, Cambridge University Press, Cam
bridge, 1973. [Cited on page 5.]
Index
Numbers set in italic type refer to the page where the term is ﬁrst deﬁned.
A
accumulation point . . . . 75
algebra . . . . . . . . . . . . . 53
associative . . . . . . . . 53
Banach . . . . . . . . . . 54
C
∗
. . . . . . . . . . . . . . 68
normed . . . . . . . . . . 54
quotient . . . . . . . . . . 55
algebra homomorphism . 54
algebra isomorphism . . . 54
annihilator . . . . . . . . . . 37
pre . . . . . . . . . . . . . 37
approximate identity . . . 56
B
B(X) . . . . . . . . . . . . . . 17
B(X, Y ) . . . . . . . . . . . . 17
completeness of . 18, 35
Baire category theorem . 20
balanced . . . . . . . . . . . . 48
ball
closed . . . . . . . . . . . 19
open . . . . . . . . . . . . 19
Banach
isomorphism theorem 22
space . . . . . . . . . . . . . 3
Banach algebra . . . . . . . 54
Banach’s criterion . . . . . . 3
BanachAlaoglu theorem 40
BanachSteinhaus
theorem . . . . . . . . . . 27
bicommutant . . . . . . . . 62
bidual . . . . . . . . . . . . . 35
BLT theorem . . . . . . . . 18
BohnenblustSobczyk
theorem . . . . . . . . . . 33
bound
upper . . . . . . . . . . . 32
boundary
extremal . . . . . . . . . 45
bounded
essentially . . . . . . . . . 5
norm . . . . . . . . . . . . 38
weakly . . . . . . . . . . . 38
C
C
∗
algebra . . . . . . . . . . 68
C
0
(X, E) . . . . . . . . . . . . 6
C
00
(X, E) . . . . . . . . . . . . 7
canonical embedding . . . 35
category
ﬁrst . . . . . . . . . . . . . 19
second . . . . . . . . . . . 19
centraliser . . . . . . . . . . 62
chain . . . . . . . . . . . . . . 32
character . . . . . . . . . . . 65
closed ball . . . . . . . . . . 19
closed convex hull . . . . . 45
closedgraph theorem . . 26
cnv A . . . . . . . . . . . . . . 45
cnv A . . . . . . . . . . . . . . 45
cocountable topology . . 10
codimension . . . . . . . . . . 4
commutant . . . . . . . . . . 62
double . . . . . . . . . . . 62
compact
locally . . . . . . . . . . . . 7
compact support . . . . . . . 7
compactiﬁcation
onepoint . . . . . . . . . 23
complemented subspace . 37
completion . . . . . . . . 6, 57
existence of . . . . . . . 35
uniqueness of . . . . . . 19
convergence
locally uniform . . . . . 13
of a net . . . . . . . . . . 11
convex . . . . . . . . . . . . . 43
locally . . . . . . . . . . . 43
convex hull . . . . . . . . . . 45
convolution . . . . . . . . . . 56
D
∂
e
C . . . . . . . . . . . . . . . 45
dense . . . . . . . . . . . . . . 20
nowhere . . . . . . . . . . 19
direct product . . . . . . . . . 7
direct sum . . . . . . . . . . . 9
algebraic . . . . . . . . . . 7
pnorm . . . . . . . . . . . 8
directed set . . . . . . . . . . 11
disc algebra . . . . . . . . . 69
dominated . . . . . . . . . . 33
double commutant . . . . 62
dual
second . . . . . . . . . . . 35
dual space . . . . . . . . . . 31
algebraic . . . . . . . . . 31
topological . . . . . . . . 31
E
element
maximal . . . . . . . . . 32
119
120 Index
essentially bounded . . . . . 5
evaluation homomorphism 69
eventually . . . . . . . . . . . 75
extremal boundary . . . . 45
extreme point . . . . . . . . 45
F
face . . . . . . . . . . . . . . . 45
ﬁniteintersection
property . . . . . . . . . 39
ﬁrst category . . . . . . . . 19
frequently . . . . . . . . . . . 75
function
open . . . . . . . . . . . . 21
functional
linear . . . . . . . . . . . . 31
linear, bounded . . . . 31
Minkowski . . . . . . . . 43
sublinear . . . . . . . . . 32
G
G(T) . . . . . . . . . . . . . . 26
gauge . . . . . . . . . . . . . . 43
Gelfand topology . . . . . 67
Gelfand transform . . . . . 68
graph . . . . . . . . . . . . . . 26
H
HahnBanach theorem . . 32
holomorphic
strongly . . . . . . . . . . 36
weakly . . . . . . . . . . . 36
homomorphism
algebra . . . . . . . . . . 54
evaluation . . . . . . . . 69
hull
convex . . . . . . . . . . . 45
convex, closed . . . . . 45
I
ideal . . . . . . . . . . . . . . . 55
maximal . . . . . . . . . 65
identity
approximate . . . . . . . 56
image . . . . . . . . . . . 18, 54
initial topology . . . . . . . . 9
invertible . . . . . . . . . . . 59
isometric isomorphism . . . 6
isometry . . . . . . . . . . . . . 6
isomorphism
algebra . . . . . . . . . . 54
isometric . . . . . . . . . . 6
J
Jacobson radical . . . . . . 67
K
kdense . . . . . . . . . . . . 21
kernel . . . . . . . . . . . 17, 54
Poisson . . . . . . . . . . 57
KreinMilman theorem . 46
L
limit . . . . . . . . . . . . . . . 11
Liouville’s theorem . . . . 36
locally compact . . . . . . . . 7
locally convex . . . . . . . . 43
locally uniform
convergence . . . . . . . 13
M
majorant . . . . . . . . 11, 32
map
quotient . . . . . . . . . . 14
maximal element . . . . . . 32
maximal ideal . . . . . . . . 65
meagre . . . . . . . . . . . . . 19
Minkowski functional . . 43
Minkowski’s inequality . . 5
N
net . . . . . . . . . . . . . . . . 11
convergent . . . . . . . . 11
universal . . . . . . . . . 76
nonmeagre . . . . . . . . . . 19
norm . . . . . . . . . . . . . . . 3
operator . . . . . . . . . 17
product . . . . . . . . . . 26
quotient . . . . . . . . . . . 5
supremum . . . . . . . . . 4
norm bounded . . . . . . . 38
normal . . . . . . . . . . . . . 24
normed algebra . . . . . . . 54
normed space . . . . . . . . . 3
normed vector space . . . . 3
nowhere dense . . . . . . . 19
O
onepoint
compactiﬁcation . . . . 23
open ball . . . . . . . . . . . 19
open function . . . . . . . . 21
openmapping lemma . . 21
openmapping theorem . 22
operator
linear, bounded . . . . 17
operator norm . . . . . . . 17
operator space . . . . . . . 41
order
partial . . . . . . . . . . . 32
pre . . . . . . . . . . . . . 10
P
pintegrable . . . . . . . . . . 5
partial order . . . . . . . . . 32
point
accumulation . . . . . . 75
extreme . . . . . . . . . . 45
Poisson kernel . . . . . . . . 57
preannihilator . . . . . . . 37
preorder . . . . . . . . . . . . 10
product
direct . . . . . . . . . . . . 7
topological . . . . . . . . 39
product norm . . . . . . . . 26
product topology . . 26, 39
Q
quasinilpotent . . . . . . . . 67
quotient algebra . . . . . . 55
quotient map . . . . . . . . 14
quotient norm . . . . . . . . . 5
quotient seminorm . . . . . 5
quotient space . . . . . . . . 4
completeness of . . . . 23
quotient topology . . . . . 14
R
radical
Jacobson . . . . . . . . . 67
radius
spectral . . . . . . . . . . 61
reﬂexive . . . . . . . . . . . . 35
resolvent . . . . . . . . . . . 60
resolvent set . . . . . . . . . 60
Index 121
S
σ(X, M) . . . . . . . . . . . . 41
σ(X, X
∗
) . . . . . . . . . . . 32
σ(X
∗
, X) . . . . . . . . . . . 37
second category . . . . . . 19
second dual . . . . . . . . . 35
seminorm . . . . . . . . . . . . 3
quotient . . . . . . . . . . . 5
semisimple . . . . . . . . . . 67
separates points . . . . . . 10
separating . . . . . . . . . . 41
sequence spaces . . . . . . . 9
series
absolutely convergent . 3
convergent . . . . . . . . . 3
sum of . . . . . . . . . . . . 3
set
balanced . . . . . . . . . 48
directed . . . . . . . . . . 11
space
dual, algebraic . . . . . 31
dual, topological . . . . 31
normed . . . . . . . . . . . 3
operator . . . . . . . . . 41
quotient . . . . . . . . . . . 4
vector, normed . . . . . . 3
vector, topological . . 42
spectral radius . . . . . . . 61
spectrum . . . . . . . . . . . 60
strong operator topology 27
subalgebra . . . . . . . . . . 53
submultiplicative . . . . . 54
subnet . . . . . . . . . . . . . 75
subspace . . . . . . . . . . . . . 4
complemented . . . . . 37
sum
direct . . . . . . . . . . . . 9
direct, pnorm . . . . . . 8
direct, algebraic . . . . . 7
summable . . . . . . . . . . . 13
support
compact . . . . . . . . . . . 7
supremum norm . . . . . . . 4
T
Tietze’s theorem . . . . . . 25
topological product . . . . 39
topological vector space . 42
topology
cocountable . . . . . . . 10
Gelfand . . . . . . . . . . 67
initial . . . . . . . . . . . . 9
operator, strong . . . . 27
operator, weak . . . . . 36
product . . . . . . . . . . 26
quotient . . . . . . . . . . 14
weak . . . . . . . . . . 9, 31
weak* . . . . . . . . . . . 37
trigonometric polynomials 57
Tychonov’s theorem . . . 40
U
unitization . . . . . . . . . . 55
universal net . . . . . . . . . 76
upper bound . . . . . . 11, 32
Urysohn’s lemma . . . . . 24
V
vanish at inﬁnity . . . . . . . 6
W
weak operator topology . 36
weak topology . . . . . 9, 31
weak* topology . . . . . . . 37
weakly bounded . . . . . . 38
Wiener’s lemma . . . . . . 73
X
X
r
. . . . . . . . . . . . . . . . 17
X
∗
. . . . . . . . . . . . . . . . 31
Z
Zorn’s lemma . . . . . . . . 32
Copyright c Alexander C. R. Belton 2004, 2006 Hyperlinked and revised edition All rights reserved The right of Alexander Belton to be identiﬁed as the author of this work has been asserted by him in accordance with the Copyright, Designs and Patents Act 1988.
Contents
Contents Introduction 1 Normed Spaces
1.1 1.2 1.3 1.4 1.5 1.6 1.7 2.1 2.2 2.3 2.4 2.5 2.6 2.7 2.8 2.9 2.10 3.1 3.2 3.3 3.4 3.5 3.6 3.7 3.8 3.9 Basic Deﬁnitions . . . . . . . . Subspaces and Quotient Spaces Completions . . . . . . . . . . . Direct Sums . . . . . . . . . . . Initial Topologies . . . . . . . . Nets . . . . . . . . . . . . . . . Exercises 1 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
i iii 3 3 4 6 7 9 10 14 17 17 18 18 19 21 22 26 27 27 28 31 31 31 32 32 36 36 37 37 39
2 Linear Operators
Preliminaries . . . . . . . . . . . . . . Completeness of B(X, Y ) . . . . . . . . Extension of Linear Operators . . . . . The Baire Category Theorem . . . . . The OpenMapping Theorem . . . . . Exercises 2 . . . . . . . . . . . . . . . . The ClosedGraph Theorem . . . . . . The Principle of Uniform Boundedness The Strong Operator Topology . . . . Exercises 3 . . . . . . . . . . . . . . . . Initial Deﬁnitions . . . . . . . The Weak Topology . . . . . Zorn’s Lemma . . . . . . . . . The HahnBanach Theorem . The Weak Operator Topology Adjoint Operators . . . . . . . The Weak* Topology . . . . . Exercises 4 . . . . . . . . . . . Tychonov’s Theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
3 Dual Spaces
i
. . . . .5 The BanachAlaoglu Theorem Topological Vector Spaces . . . . Bibliography Index . . . . . . . . . . . . The KreinMilman Theorem . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .4 6. . . . . . . . . . Unitization . . . . . . . . . . . . . . . . . . . . . . . . . . . . .3 5. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .4 6. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Quotient Algebras . . . . . . . . . . . . . . . . . . . . . . Contents . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Exercises 7 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .4 5. The Representation Theorem Examples . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6 Characters and Maximal Ideals A Tychonov via Nets A. . . . . . . . . . . . . . . . . . . . . . . . .2 4. . . . . . . . . . . . . .3 4. . . . . . . . . . . . . . . . . . . . . . . . . . . . . .1 5. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .2 5. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .3 6. . . . . . . . . . . . . . . . . . . . . . . . Exercises 5 . .13 4. . . . . . . . . . . . . . . . . . . . . . . . Characters and the Spectrum The Gelfand Topology . . . . . . . . . . . . . . . .12 3. . . . .1 4. . . . . . . . .1 Exercises A . . . . . . . . . . . . . . . . . . . . . . . . The GelfandMazur Theorem The SpectralRadius Formula Exercises 6 . . . . . . . . . . . . . . . . . . . . . .ii 3. . . . . . . . . . . . 40 41 45 47 53 54 55 56 57 59 60 61 61 63 65 66 67 68 68 72 75 78 79 79 83 88 92 97 105 109 116 117 119 4 Normed Algebras 5 Invertibility The Spectrum and Resolvent .1 6. . . . . . . . . . . . . . . . . . . . . . . . . .11 3. . . . . . . . . . . . .10 3. . Solutions to Exercises Exercises Exercises Exercises Exercises Exercises Exercises Exercises Exercises 1 2 3 4 5 6 7 A . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .2 6. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Approximate Identities Completion . . . . . .
We use [24] as a compendium of results from that series of lectures.Introduction These notes are an expanded version of a set written for a course given to ﬁnalyear undergraduates at the University of Oxford. R. Mr L. Lindsay.) The author acknowledges his debt to all those from whom he has learnt functional analysis. A This document was typeset using LTEX 2ε with Peter Wilson’s memoir class and the A AMSL TEX and X pic packages. especially Professor D. Alexander C. Taitz and Ms P. A thorough understanding of the Oxford thirdyear b4 analysis course (an introduction to Banach and Hilbert spaces) or its equivalent is a prerequisite for this material. The index was produced with the aid of the MakeIndex Y program. Belton Lady Margaret Hall Oxford 20th August 2004 This edition contains a few additional exercises and the electronic version is equipped with hyperlinks. Dr G. Edwards. (Numbers in square brackets refer to items in the bibliography. Iley. A. ACRB University College Cork 30th September 2006 Convention Throughout these notes we follow the Diracformalism convention that inner products on complex vector spaces are conjugate linear in the ﬁrst argument and linear in the second. The students attending the course were very helpful. M. Evseev. in contrast to many Oxford courses. iii . Allen and Dr J. R. thanks to the hyperref package of Sebastian Rahtz and Heiko Oberdiek. especially Mr A.
.
Spaces 1 .
.
The series ∞ n n=1 xn is convergent if there exists x ∈ X such that k=1 xk n 1 is convergent to x. i. · ) that is complete. if necessary we will denote the norm on the space X by · X .1. ∞). A norm on a vector space X is a function · : X → R+ := [0.2.3. A Banach space is a normed vector space (E. (homogeneity) (subadditivity). and the series is said to have sum x. Deﬁnition 1. We will sometimes use the term normed space as an abbreviation. Deﬁnition 1. (Banach’s Criterion) A normed vector space X is complete if and only if every absolutely convergent series in X is convergent. Theorem 1. The series is absolutely convergent if ∞ xn is n=1 convergent. every Cauchy sequence in E is convergent.One Normed Spaces Throughout. Basic Deﬁnitions Deﬁnition 1. x → x that satisﬁes. for all x.4. y ∈ X and α ∈ F. and A seminorm on X is a function p : X → R+ that satisﬁes (ii) and (iii) above.5.9]. Theorem 1. 3 . the scalar ﬁeld of a vector space will be denoted by F and will be either the real numbers R or the complex numbers C.e. where E is equipped with the metric d(x. Proof This is a b4 result: see [24.. (i) (ii) (iii) x = 0 if and only if x = 0 αx = α x x+y x + y (faithfulness). A normed vector space is a vector space X with a norm · . Let (xn )n 1 be a sequence in the normed vector space X.2. Deﬁnition 1. y) := x − y .
A subspace of a Banach space is closed if and only if it is complete.4 Normed Spaces Subspaces and Quotient Spaces Deﬁnition 1.10]. denoted by C(X. E).) The dimension of X/M is the codimension of M (in X).36). T) be a topological space and let (E. Deﬁnition 1. B ⊆ X. Evalued functions on X forms an vector space. equipped with the vectorspace operations [x] + [y] := [x + y] and α[x] := [αx] ∀ x.2. Theorem 1. E) = Cb (X. g ∈ C(X. E Cb (X. E) := f ∈ C(X. Appendix A.6. for a full discussion see [7. the quotient space X/M is the set X/M := [x] := x + M x ∈ X . Proposition 1. where the algebraic operations are deﬁned pointwise: if f . Theorem 1.7. b ∈ B} and αA := {αa : a ∈ A} ∀ A. where x + M := {x + m : m ∈ M}. hence C(X. (It is a standard result of linear algebra that this deﬁnes a vector space.8. If E = C (the most common case of interest) we use the abbreviations C(X) and Cb (X).4]. α ∈ F. If X is compact then every continuous. Proof See [24. is a Banach space. Let (X. Evalued function is bounded. f → sup f (x) :x∈X . where A + B := {a + b : a ∈ A. y ∈ X. Let X be a normed vector space with a subspace M and let [x] X/M := inf x−m :m∈M ∀ [x] ∈ X/M. E) in this case. α ∈ F. E) and α ∈ F then (f + g)(x) := f (x) + g(x) and (αf )(x) := αf (x) The subspace of bounded functions Cb (X. Example 1. · E ) be a Banach space over F. A subspace of a vector space X is a subset M ⊆ X that is closed under vector addition and scalar multiplication: M + M ⊆ M and αM ⊆ M for all α ∈ F.9. with supremum norm · ∞ (see Theorem 1. .10. E) where · ∞: ∀ x ∈ X. E) → R+ . Given a vector space X with a subspace M. f ∞ <∞ . The set of continuous.
the generalisation from R to a subinterval I is trivial. called the quotient seminorm (or quotient norm) on X/M. We prove the ﬁnal claim in Proposition 2.2. Chapter 7]. see also Exercise 1.2.15 as a consequence of the openmapping theorem.19] for its proof).11. 5 Proof ¯ Clearly [x] X/M = 0 if and only if d(x. Taking the inﬁmum over such m and n gives the result. (Discussion of Lp (R) may be found in [17. If N := {f ∈ Lp (I) : f p = 0} then Lp (I) := Lp (I)/N is a Banach space. · E/M is a Banach space. Let I be a subinterval of R and let L∞ (I) denote the vector space of Lebesguemeasurable functions on I that are essentially bounded : L∞ (I) := f : I → C f is measurable and f ∞ < ∞}. so Lp (I) is closed under addition. Example 1. M) = 0. the subadditivity of · p is known as Minkowski’s inequality (see [17. but not a norm. Theorem 28. .1. 2 max{f . which is a norm if and only if M is closed. with norm [f ] → [f ] p := f p . n ∈ M. let x. := I f (x)p dx .) Example 1. Chapter 28] and [26. which holds if and only if x ∈ M . gp} 2p f p + gp . = [x + y] X/M x + y − (m + n) x−m + y−n for all m. Let I be a subinterval of R and let p ∈ [1. If E is a Banach space and M is a closed subspace of E then E/M. If α ∈ F and x ∈ X then α[x] X/M = [αx] X/M = inf αx − m : m ∈ M X/M = inf α x − n : n ∈ M = α [x] using the fact that α−1 M = M if α = 0 (because M is a subspace). we shall frequently blur the distinction between f and [f ]. it is simple to verify that Lp (I) is a vector space. For subadditivity. The vector space of Lebesguemeasurable functions on I that are pintegrable is denoted by Lp (I): Lp (I) := f : I → C f is measurable and f with vectorspace operations deﬁned pointwise and f (Note that f + gp f  + g p p 1/p p <∞ . (A function lies in N if and only if it is zero almost everywhere. g} p = 2p max{f p .) The map f → f p is a seminorm. y ∈ X and note that [x] + [y] X/M . Hence · X/M is faithful if and only if M is closed. ∞).12.) As is usual practise in functional analysis. Subspaces and Quotient Spaces This deﬁnes a seminorm on X/M.
of course. Theorem 5. E) contains two subspaces worthy of note: · ) a Banach space then (i) C0 (X. E).14. Cb (X. The process of completing a given space may often be simpliﬁed by realising it as a dense subspace of some known Banach space.6].) Completions Recall that a map f : X → Y between metric spaces (X. ∞. i is a completion of X. x2 ∈ X. N = {f ∈ L∞ (I) : f ∞ = 0} consists of those functions that are zero almost everywhere and L∞ (I) := L∞ /N is a Banach space with respect to the norm [f ] → [f ] ∞ := f ∞ .) As in the previous example. Although it may seem that we have two diﬀerent meanings for f in Example 1. Example 1. The pair X.15. the above and that Example 1.e.13. dX ) and (Y. and an isometric isomorphism between normed vector spaces is an invertible linear isometry (the inverse of which is automatically linear and isometric). for all ε > 0. and is unique in the following sense: if (Y. If (X. As we have uniqueness.6 with vectorspace operations as usual and f ∞ Normed Spaces := inf{M : f (x) M almost everywhere}. (It is not diﬃcult to show that f ∞ = sup{f (x) : x ∈ I \ N} for some null set N which may. x2 ) for all x1 . If X is a normed vector space then there exists a Banach space X and a ˜ ˜ ˜ linear isometry i : X → X such that i(X) is dense in X.7. j) are completions of X then there exists an isometric isomorphism k : Y → Z such that k ◦ i = j. the set {x ∈ X : f (x) ε} is compact). . Equipped with the supremum norm on Ω. Proof We defer this until we have developed more machinery. ˜ Theorem 1. Let Ω be an open subset of the complex plane C and let Hb (Ω) := {f : Ω → C  f is bounded and holomorphic in Ω}.6. depend on f . i) and (Z.11 and 2. The following examples demonstrate this. see Propositions 3. dY ) is an isometry if dY f (x1 ). they coincide if f is continuous. Evalued functions on X that vanish at inﬁnity (i.. f → f ∞ is a seminorm. we talk about the completion of a normed vector space. (Completeness is most easily established via Morera’s theorem [16. the continuous. Hb (Ω) is a Banach space. those f ∈ C(X. E) such that. T) is a topological space and (E. f (x2 ) = dX (x1 .
1. It is immediate that C[0. those f ∈ C(X. since step functions can be approximated arbitrarily well by continuous functions (with respect to the · 1 norm). Furthermore. Evalued functions on X with compact support (i. := x := (xa )a∈A xa = 0 for all but ﬁnitely many a ∈ A ⊆ This is a vector space. α ∈ F. for the proof of this claim see Proposition 2. 1] is dense in L1 [0. Deﬁnition 1.e..3. 1]. C[0. this follows the pattern of the b4 proof that ℓ∞ is complete [24. The ﬁrst is a closed subspace of Cb (X. 1] (because of the inequality f 1 f ∞ ). (Uncountable Sums) If A is an arbitrary set let xa := sup a∈A a∈A0 xa : A0 is a ﬁnite subset of A . the direct product of the Banach spaces Ea . 1] is a subspace of L1 [0.. 1].4.4] so we omit it.17. i.16. E) such that supp f := {x ∈ X : f (x) = 0} is compact). 1] (with respect to · 1 ) is L1 [0.19. 7 Direct Sums Throughout this section Ea : a ∈ A denotes a family of Banach spaces with common scalar ﬁeld F. Example 1.1) and if X is Hausdorﬀ and locally compact (for all x ∈ X there exists U ∈ T such that x ∈ U ¯ and U is compact) then C00 (X. the continuous. more accurately. := x = (xa )a∈A : x ∞ <∞ ⊆ equipped with vectorspace operations deﬁned pointwise and norm · ∞: × a∈A Ea . Proof The only thing that is not immediate is the proof of completeness. its image under the map f → [f ] is.18. x → sup xa : a ∈ A . Example 1. Hence the latter space is the completion of the former. E). E). y ∈ a∈A Ea . For other types of direct sum we need the following deﬁnition. with the vectorspace operations deﬁned pointwise: x + y := (xa + ya )a∈A Theorem 1.18. The set a∈A Ea × a∈A Ea . E) (the proof of this is Exercise 2. Direct Sums (ii) C00 (X.e. Hence the completion of C[0. Let a∈A Ea a∈A Ea denote the algebraic direct sum of the spaces Ea . is a Banach space. Deﬁnition 1. E) is dense in C0 (X. and αx := (αxa )a∈A ∀ x.
it remains to prove that x(n) → x. (n) (m) (n) Since xa − xa x(n) − x(m) p . ordered by inclusion. the inequality xa + ya a∈A0 p a∈A0 (p) 2p xa p + ya p 2p x p p + y p p ∀ x.11 to suitable step functions. (p) Subadditivity of · p on a∈A Ea follows from Minkowski’s inequality on Cn . It is easy to show that this agrees with the usual deﬁnition if A is countable. n ∈ N then a∈A0 xa − xa p x(n) − x(m) p . let (x(n) )n 1 be a Cauchy sequence in a∈A Ea . so x − x(m) p n0 . Let ε > 0 p (n) (m) and suppose that n0 ∈ N is such that x − x n0 . := {x ∈ (p) a∈A Ea . (p) To see that we have completeness.21. is dense in the . see Deﬁnition 1. and this can be obtained by applying the integral version of Minkowski’s inequality in Example 1. If A0 ⊆ A is ﬁnite then xa a∈A0 p = lim n→∞ x(n) a a∈A0 p n→∞ lim x(n) p . ∞) the algebraic direct sum (p) pnorm direct sum a∈A Ea . a∈A Ea : x p < ∞} is a subspace of a∈A Ea and of the Banach spaces Ea . For all p ∈ [1. n a∈A0 x(n) − x(m) p . where A is the aggregate of all ﬁnite subsets of A. xa − x(m) a ε for all m p = lim n→∞ a∈A0 x(n) − x(m) a a p εp ∀m n0 . the pnorm direct sum Proof Let A0 be a ﬁnite subset of A. p p− this last limit exists because x(n) (p) n 1 is Cauchy: x(n) x(m) p This bound shows that x ∈ a∈A Ea . shows that a∈A Ea is a subspace of a∈A Ea . y ∈ (p) a∈A Ea . which may be proved as in Example 1. ∞) and x ∈ x Then (p) a∈A Ea p a∈A Ea let p 1/p := a∈A xa . · p is a Banach space. (We are looking at the net of ﬁnite + sums a∈A0 xa A0 ∈A in R . and this gives the result. For p ∈ [1.20.11.) Theorem 1. we have that xa := limn→∞ xa exists for all a ∈ A.8 Normed Spaces for any collection (xa )a∈A of nonnegative real numbers. a∈A Ea Proposition 1.39. Then p < ε for all m. Note ﬁrst that (n) (m) if A0 ⊆ A is ﬁnite and m.
Let F be a collection of functions as in Deﬁnition 1. the pnorm direct sum (p) a∈A Ea = and the direct product a∈A Ea = ℓ∞ . .24. (To see this. . Example 1. . . f1 . U) → (X. . . if A is any set and Ea = F for all a ∈ A then we deﬁne c00 (A). and that the collection of arbitrary unions of these sets is a topology (cf. Sf ) is continuous for all f ∈ F . Un ∈ Sfn ). Let X be a set and F be collection of functions on X. The initial topology generated by F . note that every set of this form lies in TF . In general. Initial Topologies Proof This is a simple consequence of the fact that if x ∈ countably many a ∈ A: see Exercise 1. U ∈ Sf }. Then the algebraic direct sum a∈A Ea = c00 := x = (xn )n∈N : ∃ N ∈ N such that xN +1 = xN +2 = · · · = 0 . . for all f ∈ F . U) → (Yf . where (Yf . . this space is the direct sum of the Banach (∞) spaces Ea .22. .1) these sets are a basis for this topology.1. ∞). 9 (p) a∈A Ea then xa = 0 for only (∞) Deﬁnition 1.23. .7. every element of TF is the arbitrary union of sets of the form n fi−1 (Ui ) i=1 (n ∈ N. c0 if p = ∞. ℓp if p ∈ [1.) Proposition 1. . Initial Topologies Deﬁnition 1.5. 3. denoted by TF . (Older books call TF the weak topology generated by F : the adjective ‘weak’ is tremendously overworked in functional analysis so we prefer the modern term. The previous proposition motives the deﬁnition of a∈A Ea as the completion of a∈A Ea with respect to · ∞ . A function g : (Z. Sf ) is a topological space. fn ∈ F.25. is the coarsest topology such that each function f ∈ F is continuous.4).} and take Ea = F for all a ∈ A. (Sequence Spaces) Let A = N := {1. but the inclusion may be strict. . c0 (A) and p ℓ (A) in this manner. Exercise 1.24 and let (Z. f ∈F In fact.) It is clear that TF is the intersection of all topologies on X that contain f −1 (Sf ) = {f −1 (U) : f ∈ F. 2. U) be a topological space. such that f : X → Yf . (1. U1 ∈ Sf1 . Clearly a∈A Ea ⊆ a∈A Ea . TF ) is continuous if and only if f ◦ g : (Z.
. d) it is readily proven that. c ∈ A. Since Sf is Hausdorﬀ there exist disjoint sets U. y ∈ X be distinct and suppose that f ∈ F is such that f (x) = f (y). given a set M ⊆ X. of course. as claimed. A preorder for all a. V ∈ Sf such that f (x) ∈ U and f (y) ∈ V .24. i=1 and then n n g −1 (U) = i=1 g −1 fi−1 (Ui ) = i=1 (fi ◦ g)−1 (Ui ) ∈ U. yes. such that (Yf . whereas arguments involving open sets can sometimes appear rather opaque. c imply that a . . .26. fn ∈ F and U1 ∈ Sf1 . We shall see other examples of initial topologies later. Proof Let x. . y ∈ f −1 (V ) and f −1 (U) ∩ f −1 (V ) = f −1 (U ∩ V ) = ∅. and therefore so can the topology generated by the metric d. . Proposition 1. which is an initial topology (that generated by the coordinate projections). If X = R is equipped with the cocountable topology (which consists of the empty set and the complement of each countable subset of R) then M = R \ {0} has closure ¯ M = R but there is no sequence (xn )n 1 ⊆ M such that xn → 0. as required. Is it possible to replace the notion of sequence with some generalisation which allows a version of the result above? The answer is. The initial topology TF is Hausdorﬀ if F separates points: for all x. For the converse. Let F be a collection of functions as in Deﬁnition 1. (reﬂexivity) c (transitivity). b. . We may assume that U = n fi−1 (Ui ) for f1 . Nets ¯ In a metric space (X. the element x ∈ M if and only if there exists a sequence (xn )n 1 ⊆ M such that xn → x. suppose that f ◦ g is continuous for all f ∈ F and let U ∈ TF . . Hence TF is Hausdorﬀ. (i) a a and (ii) a b and b on A is a binary relation that satisﬁes. However. Let A be a set. one implication is immediate. Un ∈ Sfn . Deﬁnition 1. f −1 (V ) ∈ TF are such that x ∈ f −1 (U). This shows that the result of the previous paragraph does not hold for general topological spaces (and that the cocountable topology on R is nonmetrizable).10 Normed Spaces Proof As the composition of continuous functions is continuous. Hence closed sets ¯ (those such that M = M ) may be characterised by means of sequences. . Sf ) is Hausdorﬀ for all f ∈ F .27. Then f −1 (U). arguments with sequences are often very natural and easy to follow. The above may remind the reader of a result concerning the product topology. y ∈ Y such that x = y there exists f ∈ F such that f (x) = f (y).
Let (X. let x ∈ M and let A = {U ∈ T : x ∈ U} denote the collection of open sets containing x. this is denoted by (xa )a∈A (the preorder being understood). b ∈ A there exists c ∈ A such that a c and b c. and we write xa → x or lima∈A xa = x. the proof of this is immediate. (To see this. 11 Proof Suppose M is closed and (xa )a∈A is a net in M. there exists a0 ∈ A such that xa ∈ U for all a a0 . 1]. A net (xn )n∈N in a topological space (X. and the net (xn )n∈N is the same thing as the sequence (xn )n 1 . Let f be a continuous function on the interval [0. i.6]. Deﬁnition 1.31. so it contains some elements of (xa )a∈A .32. T) be a topological space.2. the element c is called an upper bound or majorant for a and b. for all U ∈ T with x ∈ X. Let (X. §9. Hence lima∈A xa ∈ M. and let A be the collection of (realvalued) step functions φ on [0.29.33. ) forms a directed set. 1] that are bounded above by f : a step function φ ∈ A if φ(x) f (x) for all x ∈ [0. the element x is the limit of this net. ψ(x)} is a step function if φ and ψ are. such that xa → x for some x ∈ X.29 and φ denotes the integral of the step function φ then ( φ)φ∈A is a net in R that converges to f . A B . T) be a topological space and suppose that M ⊆ X. Theorem 1. ) is a set A and a preorder on A with the following property: for all a. Nets we say that A is ordered by . Example 1. Order A by saying that φ ψ if and only if φ(x) ψ(x) for all x ∈ [0.e.) We now give some theorems that illustrate how nets can be used to answer topological questions.1.6. A directed set (A. If x ∈ M then X \ M is an open set containing x.30. Proof This may be proved in the same manner as the corresponding result for sequences [22. ) is the directed set of Example 1. since φ ∨ ψ : x → max{φ(x). T) converges to x if and only if the sequence (xn )n 1 converges to x. A net in a Hausdorﬀ topological space has at most one limit..28. If (A. in general. limits need not be unique. ) and a function x : A → X. If N is equipped with the usual order then it is a directed set. this forms a directed set if ordered by reverse inclusion. For convenience we write b a if and only if a b. / contrary to hypothesis. Example 1. note ﬁrst that f = sup{ φ : φ ∈ A}: see [17. Proposition 4. Example 1.34. A net (xa )a∈A in X is convergent if there exists x ∈ X such that. Example 1. (This latter notation is a slight abuse as. ¯ Conversely. Hence nets are generalisations of sequences.) Proposition 1. the pair (A. 1] (where is the usual order on R). 1].2]. Then M is closed if and only if lima∈A xa ∈ M for all convergent nets (xa )a∈A ⊆ M. A net in a set X is a directed set (A.
so f (xA ) → f (x).37. Let F be a collection of functions as in Deﬁnition 1. Proof Let f be continuous. let U ∈ S and suppose for contradiction that V = f −1 (U) is not open. E) is a Banach space with respect to the supremum norm · ∞ . If ε > 0 then there exists n0 ∈ N such that fn − fm ∞ < ε if m. For the other half.24. TF ) if and only if f (xa ) → f (x) in (Yf . x ∈ X n→∞ n0 . This proposition allows us to give a simple proof of the completeness of Cb (X.. if xa → x then choose a0 ∈ A such that fn0 (xa )−fn0 (x) < ε for all a a0 and note that f (xa ) − f (x) for all a f (xa ) − fn0 (xa ) + fn0 (xa ) − fn0 (x) + fn0 (x) − f (x) < 3ε a0 . the desired contradiction. i. and the net (xA )A∈A converges to x: note that if U ∈ T contains x then xA ∈ U for all A U. Proof We prove only completeness. f is bounded. The following proposition shows why nets and initial topologies work so well together. x → lim fn (x) n→∞ is well deﬁned.12 Normed Spaces if and only if A ⊇ B. and xA0 ∈ V . E). Let (X. Finally. so there exists a0 ∈ A such that xa ∈ V for all a a0 . as required. and suppose that (xa )a∈A is a net in X.35. If A is ordered by reverse inclusion then (xA )A∈A is a net converging to x. E) and note that fn (x) − fm (x) fn − fm ∞ for all x ∈ X. A function f : X → Y is continuous if and only if f (xa ) → f (x) for every net (xa )a∈A such that xa → x. Let (fn )n 1 be a Cauchy sequence in Cb (X. This gives the result. suppose that (xa )a∈A ⊆ X and x ∈ X are such that f (xa ) → f (x) Proposition 1. Conversely. n so f (x) − fm (x) = lim fn (x) − fm (x) ε ∀ m n0 . If A ∈ A then A ∩ M = ∅. . If X is a topological space and E is a Banach space then Cb (X. half of the proof follows from Proposition 1. everything else is trivial. Then there exists x ∈ V \ V ◦ . so f : X → E. for otherwise X \ A is a closed set containing M and so must contain x. S) be topological spaces. Theorem 1. Sf ) for all f ∈ F .36. suppose that (xa )a∈A is convergent to x and let U ∈ S contain f (x). and no open set containing x is contained in V : for all A ∈ A := {W ∈ T : x ∈ W } there exists xA ∈ A ∩ (X \ V ). This shows that f (xa ) → f (x). Then V = f −1 (U) ∈ T contains x.35. so f is continuous.e. whence f − fm ∞ → 0 as m → ∞ and also f ∞ f − fn0 ∞ + fn0 ∞ ε + fn0 ∞ . T) and (Y. Then xa → x in (X. Hence f (xa ) ∈ U for all a a0 . In particular there exists A0 ∈ A such that f (xA0 ) ∈ U. Hence we may choose xA ∈ A ∩ M for all A ∈ A. Proof As TF makes each f ∈ F continuous. Proposition 1.
n. . If (zn )n∈N is a family of complex numbers then it is summable if and only if ∞ zn is absolutely convergent (see Exercises 1. A family (xa )a∈A ⊆ X is summable if the net of partial sums a∈A0 xa A0 ∈A is convergent. Nets allow us to give a proper treatment of summability. If (xa )a∈A is summable then a∈A xa denotes the limit of the net of partial sums. Deﬁnition 1. a∈A 13 a∈A i. If (xa )a∈A is a family of nonnegative real numbers then this deﬁnition agrees with Deﬁnition 1.6. Pedersen [14] refers to the viewpoints of topology in terms of nets and of open sets as dynamic and static. which coincides with the ad hoc method used in Deﬁnition 1. where K ⊆ Ω. as required. Nets were introduced by Moore and Smith in [13] – the theory is also called MooreSmith convergence. . Let X be a normed vector space. Note that x ∈ fi−1 (Ui ) for i = 1. If a0 ∈ A is such that ai a0 for each i then xa ∈ n fi−1 (Ui ) = U for all a a0 . respectively. Let Ω ⊆ C be open and let ιK : C(Ω) → C(K). If F = {ιK : K is a compact subset of Ω} and each C(K) has the supremum norm then TF is the topology of locally uniform convergence: fn → f ⇐⇒ fn K → f K uniformly on K for all compact K ⊆ Ω. Example 1.19 (Exercise 1. absolute summability implies summability (Exercise 1.40. .1. Third footnote on p. the reader may care to reﬂect upon the following.19. the analogue of a subsequence would then be a “subway”! An aged but excellent introduction to nets is [9..6).7 n=1 and 1.8).39.38. so fi (x) ∈ Ui and there exists ai ∈ A such that fi (xa ) ∈ Ui for all a ai . where A is the collection of ﬁnite subsets of A.7). McShane’s article [12] is a very pleasant introductory exposition. . . As to the choice of nomenclature. taken from [12.e. it is suﬃcient to consider U of the form (1.24. Chapter 2]. Kelley’s own inclination was to the name “way”. Let U ∈ TF be such that x ∈ U.3]: [J. especially in older references – and they were applied to general topological spaces by Garrett Birkhoﬀ [2]. i=1 Example 1. ordered by inclusion. f → f K be the restriction map to K. L. Nets for all f ∈ F .] Kelley writes me that [“net”] was suggested by Norman Steenrod in a conversation between Kelley. Note also that if (xa )a∈A is a family of vectors in a Banach space then xa < ∞ =⇒ xa is convergent. Steenrod and Paul Halmos.1) in Deﬁnition 1.
sends open sets to open sets). Exercise 1. Let M and N be subspaces of the normed space X. where (if S is inﬁnite) j → aj is a bijection between N and S. the restrictions of the functions in F to Y .l)∈J×L i∈I j∈J Ai ∩ Blk . is the initial topology generated by F Y = {f Y : f ∈ F }. Let X be a normed vector space and let M be a closed subspace of X. [Use Banach’s criterion. Prove also that the quotient map is linear and continuous.k)∈I×K (j. Let (xa )a∈A be a family of nonnegative real numbers and let A denote the collection of ﬁnite subsets of A.] Exercise 1.] Deduce that (xa )a∈A is summable with sum a∈A xa = a∈S xa ∞ j=1 xaj if S is ﬁnite. K and L are arbitrary. where the index sets I. Prove that if TF is the initial topology on X generated by a collection of functions F and Y ⊆ X then TF Y .] .] Exercise 1. where T denotes the norm topology on X. Deduce that the quotient norm yields the quotient topology on X/M given by Q := {U ⊆ X/M : π −1 (U) ∈ T}. Exercise 1.4. Exercise 1. the relative initial topology on Y . j ∈ J} and {Blk : k ∈ K. [Recall that ﬁnitedimensional subspaces of normed spaces are closed [24. Prove that if a∈A xa is convergent then S := {a ∈ A : xa = 0} is countable.5. Exercise 1. Prove directly that if E is a Banach space and M is a closed subspace of E then the quotient space E/M.3.6.. l ∈ L} are families of j sets. [Consider real and imaginary parts to reduce to the real case and then consider positive and negative parts. ε > 0. · E/M is complete. Corollary 1. x → [x] is the natural map from X onto X/M (the quotient map).1. j What does this have to do with initial topologies? Exercise 1. if S is inﬁnite.7. Let E be a Banach space and let (xa )a∈A a family of vectors in E.8.18] and use the quotient map. [Consider the sets Sn := {a ∈ A : xa > 1/n} for n ∈ N. J. Prove that if M is ﬁnite dimensional and N is closed then M + N is closed. Prove that if {Ai : i ∈ I. Prove that (xa )a∈A is summable (with sum α) if and only if β = sup a∈A0 xa : A0 ∈ A < ∞ and in this case α = β. Prove that a family of complex numbers (za )a∈A is summable if and only if za  a∈A is summable. Prove that π{y ∈ X : y − x < ε} = [y] ∈ X/M : [y] − [x] < ε ∀ x ∈ X. where π : X → X/M. and that the quotient map is open (i.2.2. then Ai ∩ j Blk = k∈K l∈L (i.e.14 Normed Spaces Exercises 1 Exercise 1.
Prove the converse to Proposition 1.11. [Take two points that cannot be separated by open sets and deﬁne a net that converges to both of them. Exercise 1. A sequence in a normed vector space that is convergent is necessarily bounded. Exercises 1 Exercise 1.e.9. that in a space with a nonHausdorﬀ topology there exists a net that converges to two distinct points.] Exercise 1..1.10.31. Is the same true for nets? 15 . xn n∈N is not summable). Find a Hilbert space H and a countable family of vectors (xn )n∈N in H that is summable but not absolutely summable (i.7.
.
A bounded (linear) operator from X to Y is a linear transformation T : X → Y such that the operator norm T is ﬁnite. x = 1 . We denote the collection of bounded linear operators from X to Y by B(X. Proposition 2. where T := inf{M ∈ R+ : T x M x for all x ∈ X} = sup{ T x : x ∈ X1 } = sup T x : x ∈ X. for all r ∈ R+ let Xr denote the closed ball in X with radius r and centre the origin: Xr := x ∈ X : x r . Theorem 2. T is continuous.Two Linear Operators Preliminaries Let X be a normed vector space. T is continuous at 0.1. Y ). The following statements are equivalent: (i) (ii) (iii) (iv) (v) T is a bounded linear operator. Y ) is the operator norm. T is uniformly continuous. Note that B(X. α ∈ F . Y ) is a normed vector space. 17 ∀ S. If T : X → Y be a linear transformation then (i) the kernel ker T := {x ∈ X : T x = 0} is a subspace of X. where (T + S)x := T x + Sx and (αT )x := αT x and the norm on B(X. Deﬁnition 2. x ∈ X. Let T : X → Y be a linear transformation.3. T (X1 ) is bounded: T (X1 ) ⊆ Yr for some r ∈ R+ . Proof The implications (i) ⇒ (ii) ⇒ (iii) ⇒ (iv) ⇒ (v) ⇒ (i) are immediate.2. T ∈ B(X. Y ) (or B(X) if X = Y ).
X/ ker T is a ˜ ˜ normed space and T is continuous.1. Proof The algebraic facts are immediately veriﬁed.1. Proof Existence is a b4 result [24. Exercise 1.711] or [14.18 (ii) the image im T := {T x : x ∈ X} is a subspace of Y .12. There exists a unique T ∈ B(X. Proof The fact that the completeness of Y entails the completeness of B(X. Conversely. Y ) Proposition 2. Y ) is a Banach space if and only if Y is a Banach space. Theorem 2. Y ) then (S − T )X0 = 0 and so S − T = 0 by continuity: if x ∈ X let (xn )n 1 ⊆ X0 be such that xn → x and note that (S − T )x = limn→∞ (S − T )xn = 0.5. ε > 0 and choose ˜ T T T y ∈ ker T such that x − y < [x] + ε. (BLT) Let X0 be a dense subspace of a normed vector space X and let T0 ∈ B(X0 .5. . for a proof see [10. a leisurely proof may be found in [10. Y ) then ker T is closed. If π : X → X/ ker T .11]. Theorem 2. Y ) is a result from b4 [24. ˜ T : X/ ker T → im T . Theorem 2.4. note that if SX0 = T0 = T X0 for S. and extend it to the whole space “by continuity”: the following theorem explains the meaning of this. For uniqueness. T ∈ B(X. Then ˜ T [x] = T x = T (x − y) T x−y T [x] + ε . Completeness of B(X. as is the fact that ker T = T −1 {0} is closed if T is continuous. [x] → T x. Extension of Linear Operators Often it is easiest to deﬁne a linear operator on some dense subspace of a Banach space.102] and a more concise version in [14. Y ) such that T X0 = T0 . Theorem 2. x → [x] is the quotient map. the inequality π(x) = [x] x implies that π 1 and so ˜ π ˜ (because T ◦ π = T ). and since this holds for all ε > 0 and x ∈ X we have the result.17(iii)].17(ii)]. with T = T . Y ). let x ∈ X. We prove the converse as an application of the HahnBanach theorem: see Proposition 3. Theorem 2. If X = {0} then the normed vector space B(X. and such satisﬁes T = T0 . Exercise 1. where Y is a Banach space.4]. Linear Operators (iii) X/ ker T ∼ im T via the linear transformation = If X and Y are normed spaces and T ∈ B(X.5.
) We introduce some notation (which we learned from Dr J. j) be completions of the normed space X. ∞). The Baire Category Theorem Deﬁnition 2.2. let k(yn ) n 1 be a convergent sequence (with yn ∈ Y ) and note that it is Cauchy. d) is a metric space then let X Br (x) := {y ∈ X : d(x. it contains its closure.4. M. a set is nonmeagre or of the second category otherwise. and therefore k is surjective.7. . The Baire Category Theorem Uniqueness of Completions We are now in the position to prove that the completion of a normed space is unique. If X is a normed space then X X Br (x) = x + rB1 (0) and X Br [x] = x + Xr = x + rX1 . k(Y ) is closed in Z: to see this. if the space is clear from the context then we abbreviate these to Bε (x) and Bε [x] respectively. Z. Lindsay) that will be used in the proof of the next theorem. X j i 19 Y k Z Proof Note that is a welldeﬁned linear isometry from a dense subspace of Y onto a dense subspace of Z. Proposition 2. Let (Y. There exists an isometric isomorphism k : Y → Z such that the following diagram commutes. A subset of a metric space is said to be meagre or of the ﬁrst category if it is a countable union of closed sets with empty interior. so (yn )n 1 is Cauchy (k being an isometry) and hence convergent. the continuity of k gives k(yn ) → k(y) ∈ k(Y ).5. (Recall that a set is nowhere dense if its closure has empty interior. by Theorem 2. Z) and k is an isometry: if y ∈ Y then there exists (xn )n 1 ⊆ X such that i(xn ) → y. If (X. as required. viz j(X). As k(Y ) contains j(X). y) r} (the closed ball of radius r and centre x) for all x ∈ X and r ∈ (0. to y ∈ Y . Hence. i) and (Z.6. k0 extends to k ∈ B(Y. i(x) → j(x) = lim k0 i(xn ) n→∞ = lim i(xn ) = y . y) < r} (the open ball of radius r and centre x) and X Br [x] := {y ∈ X : d(x. so k(y) = lim k i(xn ) n→∞ k0 : i(X) → Z. but k and x → x are continuous. n→∞ Furthermore.
if (2. To do this we construct sequences (rn )n 1 ⊆ (0. ∞) and (xn )n 1 ⊆ X such that r 0 > r1 > r2 > · · · → 0 as n → ∞ and Brn [xn ] ⊆ Brn−1 (xn−1 ) ∩ Gn Given (2.20 X X furthermore. Theorem 2. rm } ∀ m. Hence there exists x1 ∈ X and s ∈ (0. if n > m in particular. Note that Br0 (x0 ) ∩ G1 is open (being the intersection of two open sets) and nonempty (as G1 is dense in X). lim xn ∈ Brn [xn ] ⊆ Br0 (x0 ) ∩ Gn n→∞ n 1 n 1 and we are done.8. 1. d) then the intersection n 1 Gn is dense in X. xm ) < max{rn .2). If X = n 1 Fn then there exists n0 ∈ N such that Fn0 has nonempty interior. and let Gn = X \ Fn . (This last step shows why we must consider both closed and open balls. Proof ◦ Suppose for contradiction that Fn = ∅ for all n and so Gn is an open.) To see why we may ﬁnd sequences satisfying (2. This argument may be repeated for all n (each time ensuring that rn < 1/n) which completes the proof. ∀n 1.2) holds in the case n = 1.1) which. ¯ This is equivalent to the condition that G ∩ U = ∅ for all U ∈ T \ {∅} (because G = X ¯ if and only if (X \ G)◦ = X \ G = ∅). As xn ∈ Brm [xm ] for all n m 1. r0) such that Bs (x1 ) ⊆ Br0 (x0 ) ∩ G1 . Let (Fn )n 1 be a sequence of closed subsets of the complete metric space (X. n 1. but ◦ ¯ Gn = X \ Fn = X \ Fn = X n 1 .2). whence n 1 Gn = n 1 X \ Fn = X \ n 1 Fn = ∅. min{s. (2. consider ﬁrst the case n = 1. Br (x) = Br [x]. (Baire) If (Gn )n 1 is a sequence of open.1) holds. shows that (xn )n 1 is Cauchy. by Theorem 2.9. (2. dense subsets of the complete metric space (X. d). which states that a complete metric space is nonmeagre. An alternative formulation of Baire’s category theorem as follows. and so convergent.8. it suﬃces to prove that Br0 (x0 ) ∩ n 1 Gn = ∅.2) 1 then Brn [xn ] ⊆ Brn−1 (xn−1 ) ⊆ Brn−1 [xn−1 ] ⊆ · · · ⊆ Brm (xm ). If T is given by a metric d then this condition is equivalent to requiring that G ∩ Bε (x) = ∅ for all x ∈ X and ε > 0. Theorem 2. Linear Operators ¯ Recall that G is said to be dense in the topological space (X. T) if and only if G = X. dense subset of X. taking r1 ∈ 0. Proof Let x0 ∈ X and r0 > 0.1) and (2. Then Gn is dense. 1} shows that (2. d(xn .
if A is dense in B and B is kdense in C then A is k ′ dense in C for any k ′ > k. as required. .) Lemma 2. (ii) T is an open mapping and (iii) Y is complete. 1) such that T (Er ) is kdense in Y1 then r (i) for all y ∈ Y there exists x ∈ E such that x y and T x = y. suppose there exists ε > 0 such that Yε ⊆ T (X1 ).5.2. B ⊆ a∈A Bk [a]. Let (X. The OpenMapping Lemma Deﬁnition 2. so T is 1−k surjective. a) k. Continuing in this way we see that there exists a sequence (xl )l y−T k l xl l=0 ⊆ Er such that (2. Y ).12. Y a normed space and T ∈ B(E. Proposition 2.11. Hence Yε ⊆ B2ε (0) ⊆ T B1 (0) ⊆ T (X1 ). (OpenMapping Lemma) Let E be a Banach space. The OpenMapping Theorem 21 The OpenMapping Theorem Recall that a function between topological spaces is open if the image of every open set is open. Let U be an open X subset of X. a linear transformation T : X → Y is open if and only if T (X1 ) contains Yε for some ε > 0. (If A is dense in B then A is kdense in B for all k > 0. Let X and Y be normed spaces.3) k n+1 ∀n 0. which shows that T (U) is open. If there exist r > 0 and k ∈ (0. Conversely. As k −1 (y − T x0 ) ∈ Y1 there exists x1 ∈ Er such that k ⇐⇒ y − T (x0 + kx1 ) 0 k2. if u ∈ U then B2δ (u) ⊆ U for some δ > 0. d) be a metric space and suppose that A. Hence X Y T (U) ⊇ T u + δB2 (0) ⊇ T u + δT (X1 ) ⊇ T u + δYε ⊇ Bδε (T u).10. we say that A is kdense in B if for all b ∈ B there exists a ∈ A such that d(b. Proof X If T is open then T B1 (0) is an open subset of Y that contains T 0 = 0. without loss of generality we may take y ∈ Y1 (the case y = 0 is trivial and otherwise we may replace y by y/ y ) so there exists x0 ∈ Er such that y − T x0 k −1 (y − T x0 ) − T x1 n k. B ⊆ X. Equivalently. and so Y Y X contains B2ε (0) for some ε > 0. Proof To prove (i) let y ∈ Y .
1).. E). use the density of Y1 in Y1 ) and so by (i) we must have that T (E) = Y . F ) is a bijection between Banach spaces E and F then the inverse T −1 ∈ B(F. say k ′ = (k + 1)/2: ˜ with Y replaced by Y ˜ ˜ to prove this. note that this series is absolutely convergent. by Theorem 2. note that (i) implies that T (X1 ) ⊇ Y(1−k)/r : if y ∈ Y satisﬁes y (1 −k)/r then there exists x ∈ X such that T x = y and x r y /(1 − k) 1. E). The hypotheses of the theorem hold ˜ . . (OpenMapping Theorem) If T ∈ B(E. is complete with respect to the norm f → f ∞ := sup{ f (x) E : x ∈ X}.13.12. Proof Since T is surjective. F = n 1 T (En ) and so. Corollary 2. let Y denote the completion of Y and regard Y as a dense subspace of Y . where E and F are Banach spaces. B ⊆ E.) This shows that T (En0 /ε ) is dense in F1 .e. If y ∈ F1 then y= 1 1 1 F F (y0 +εy)−(y0 −εy) ∈ Bε [y0 ]−Bε [y0 ] ⊆ T (En0 )−T (En0 ) ⊆ T (En0 /ε ). ˜ so that T is a bounded operator from E into Y . and l=0 (2. bounded. there exists n0 ∈ N such that the closure of T (En0 ) has nonempty interior: let y0 ∈ F and ε > 0 be such F that Bε (y0 ) ⊆ T (En0 ).. ˜ ˜ Finally.3) shows that T x = y. showing that Y is complete. recall that Cb (X. F ) is surjective. T (Er ) is k ′ dense in Y1 (where k ′ ∈ (k. the continuous.9. Exercises 2 Exercise 2. Proof This is immediate. b ∈ B} for all A.1.e.22 Linear Operators Let x := ∞ k l xl . 2ε 2ε 2ε where A − B := {a − b : a ∈ A.14. and note that 1 1 lim T xn − lim T yn = lim T (xn − yn ) ∈ T (En0 /ε ) n→∞ n→∞ 2ε n→∞ 2ε since (xn − yn )/(2ε) (n0 + n0 )/(2ε) = n0 /ε. ˜ Since T (E) ⊆ Y this gives the equality Y = Y . Theorem 2.10. continuous functions on X. then T is an open map. let (xn )n 1 . (For the last inclusion. l=0 For (ii). Let X be a topological space and E a Banach space. the space of Evalued. The inequality x r/(1 − k) follows from the deﬁnition of x and the fact that the series ∞ k l has sum 1/(1 − k). so convergent. E). Evalued functions on X that vanish at inﬁnity (i. by Proposition 2. E). E) such that {x ∈ X : f (x) E ε} is compact for all ε > 0) is a closed subspace of Cb (X. i. Hence T is open. and the result follows by Lemma 2. those f ∈ C(X. If T ∈ B(E. (yn )n 1 ⊆ En0 be such that (T xn )n 1 and (T yn )n 1 are convergent. Prove that C0 (X.
realvalued functions on the unit interval and for all k ∈ N let Dk := {f ∈ X : there exists t ∈ [0. Prove that any separable Banach space E is isometrically isomorphic to a quotient space of ℓ1 . Let (X. [(X.6. 1]}.15. Exercise 2. R) denote the Banach space of continuous.] Exercise 2. the set of all subsets of N. [You may ﬁnd the BolzanoWeierstrass theorem useful. the result follows by Lemma 2. 1). Prove that T is continuous if and only if ker T is closed and that if T is continuous then T is open. Exercises 2 Exercise 2. compact topology on X := X ∪ {∞}. Let T : X → Y be a linear transformation from the normed space X onto the ﬁnitedimensional normed space Y .] Deduce that there exist continuous functions on [0. locally compact space and let ∞ denote a point not in X. Since π is a bounded linear operator. . [Take an orthonormal basis {e1 . 1] that are diﬀerentiable at no point in (0.12. Prove that X1 is separable (in the norm topology).] Completeness of Quotient Spaces The openmapping theorem provides a quick proof of the following proposition. 1]. T) is the onepoint compactiﬁcation of (X.] Prove that there is a natural correspondence between C0 (X. separable Hilbert space. e2 . · E/M is complete. T). Proposition 2. [Recall that all norms on a ﬁnitedimensional space are equivalent.6. 23 Exercise 2.7. The quotient space E/M. a Banach space.e. E) and ˙ {f ∈ C(X.2.} of H.3. deﬁne suitable projection operators on its linear span and extend these by continuity. Recall the b4 proof that ℓ∞ is closed. Let H be an inﬁnitedimensional. Let X = C([0. Prove that B(H)1 is not separable in the norm topology.. T) be a Hausdorﬀ. by deﬁnition of the quotient norm. Prove that Dk is closed. . there exists m ∈ M such that x − m ∈ E2 (any number greater than 1 will do) and so (E/M)1 ⊆ π(E2 ). Let M be a closed subspace of the Banach space E. . [Consider the map x → ∞ xn en for suitable (en )n 1 ⊆ E1 . Proof If [x] ∈ (E/M)1 then. you may assume that 2N .] Prove further that Dk is nowhere dense. E) : f (∞) = 0}. [Consider suitable piecewiselinear functions. i. is uncountable. X \ U is compact} ˙ ˙ ˙ is a Hausdorﬀ. Show that ˙ ˙ T := T ∪ {U ⊆ X : ∞ ∈ U. where π : x → [x] is the quotient map. . Let X be a separable normed space.2. Prove that no inﬁnitedimensional Banach space E has a countable Hamel basis (where a Hamel basis is a linearly independent set S such that every vector in E is a ﬁnite linear combination of elements of S).4.] n=1 Exercise 2. 1] such that f (s)−f (t) ks−t for all s ∈ [0.5. Exercise 2.
Lemma 2. / It is immediate that f D = 1.10. 1) then inf{r : x ∈ Ur } b if and only if for all r > b there exists s < r such that x ∈ Us . b] = Us ⊆ Ut . Hausdorﬀ space (which is the most important of all commutative Banach algebras) we need a result from analytic topology. we follow [3]. It is an easy exercise [22. (Throughout this proof the letters r. inf{r : x ∈ Ur } if x ∈ D.A]). Theorem 27.1]. [To see the equivalence. Equivalently. Continuing in this manner we ﬁnd a family of open sets {Um2−n : 1 m 2n .. [9] and [22]. these are dense in [0. Hausdorﬀ spaces that are normal) have suﬃcient continuous functions to separate points. i. There exists a continuous function f : X → [0.16. [14. (Urysohn) Let X be a normal space and let C. n 1} ¯ such that C ⊆ Ur ⊆ Ur ⊆ Us if r < s. Hausdorﬀ spaces (indeed. and then open sets U1/4 and U3/4 such that U ¯ C ⊆ U1/4 ⊆ U1/4 ⊆ U1/2 and ¯ ¯ U1/2 ⊆ U3/4 ⊆ U3/4 ⊆ U1 . If a ∈ (0. Exercise 5.) We set f : X → [0. Theorem 1.e. m ∈ N and 1 m 2n .24 Urysohn’s Lemma Linear Operators To prove certain facts about C(X). it remains to prove that f is continuous. 1] then f (x) < a if and only if inf{r : x ∈ Ur } < a. numbers of the form m2−n .] Some authors require the additional condition that all singleton sets to be closed for a topology to be normal.6. D ⊆ X are closed and disjoint there there exist disjoint. A topological space X is normal if every pair of disjoint. V ⊆ X such that C ⊆ U and D ⊆ V . in fact. Proof Let U1 = X \ D. which gives that ¯ f −1 [0. 1] such that f C = 0 and f D = 1. Hausdorﬀ space is normal (for a proof see [3. by normality there exists an open set U1/2 such that C ⊆ U1/2 and ¯1/2 ⊆ U1 .17. Deﬁnition 2. D ⊆ X be disjoint and closed. 1]. X is normal if for every open set W and closed set C such that C ⊆ W ⊆ X there exists an open set ¯ U such that C ⊆ U ⊆ U ⊆ W . but the deﬁnition in [21] includes this and [14] insists on the seemingly stronger requirement that normal spaces be Hausdorﬀ. closed sets can be separated by open sets: if C. s and t refer to dyadic rationals in (0. If b ∈ [0. a) = r<a Ur is open. open sets U. x → 1 if x ∈ D. Lemma 6. 1]. The following lemma yields the fact that compact. in a normal space the conditions that singletons are closed and the Hausdorﬀ property are equivalent.17] to prove that a compact.6] or [21. the continuous functions on a compact. 1]. r>b s<r t>b . and f C = 0 since C ⊆ Ur for all r. where n. let D = X \ W . and hence f −1 [0. which holds exactly when x ∈ Ur for some r < a.
these exist by local compactness. E). by deﬁnition. Let X be a locally compact. xn ∈ K. and if x ∈ X \ K then f (x) < ε. For all x ∈ K let Ux be an open set containing x and with compact closure. and h−1 (C) equals g −1 (C) (if 0 ∈ C) or g −1(C)∪(X \L◦ ) (if 0 ∈ C). and so f −1 (b. This shows that f [0. 1] we have the result. Proposition 2. E). (b. Then h has compact support and is continuous: if C ⊆ [0. Hence T g − f ∞ 2/3. the space of continuous. These are closed subsets of Y . for all b ∈ [0. R)1/3 ) is 2/3dense in Cb (Y. Proof Let T : Cb (X. Let f ∈ Cb (Y. R)1 . realvalued function F on X such that F Y = f and F ∞ = f ∞ . −1 as claimed. K = {x ∈ X : f (x) ε} is compact. realvalued function on Y then there exists a continuous. . 25 0 if x ∈ X \ L. Theorem 2. E). and i=1 ¯ L = n Uxi is a compact set such that K ⊆ L◦ . whence x ∈ r>b s<r Us . i=1 By Urysohn’s lemma there exists a continuous function g : L → [0. and by Urysohn’s lemma there exists g ∈ Cb (X. R) → Cb (Y.6. 1] such that gK = 1 and gL\L◦ = 0. . T is surjective. 1). so of X. t such that r > s > t > b and so x ∈ Ut ⊆ Us . and let C = f −1 [−1. continuous. as required: if x ∈ K then f (x)h(x) = f (x). b ∈ [0. If f is a continuous. R) be such that f ∞ 1. 1]. R) such that T (F ) = f and f ∞ F ∞ 1/3 f 1 − 2/3 ∞ = f ∞. As {(0. b] is closed. x → g(x) if x ∈ L. ¯ There exist s. f → f Y be the restriction map and note that T is continuous. As K is compact.2. but applies to continuous functions on normal spaces. bounded. K ⊆ n Uxi for x1 . bounded. R). 1]. a]. 1]. (Tietze) Let X be a normal space and let Y be a closed subset of X. Exercises 2 ¯ This inclusion is actually an equality. extend g to X by deﬁning h : X → [0. 1)} is a subbase for the usual topology on [0. Hausdorﬀ space and let E be a Banach space. Evalued functions on X that vanish at inﬁnity. Proof Let ε > 0 and f ∈ C0 (X. . The space C00 (X. In particular. . so in X. 1] is closed then g −1 (C) is closed in L. It is a theorem of HahnBanach type. E) of compactly supported. / Hence f h ∈ C00 (X. .18. −1/3] and D = f −1 [1/3. let x ∈ Ut for all t > b and suppose that r > b. R) such that g ∞ 1/3. 1] is open. The following theorem can be deduced from Urysohn’s lemma directly. and f h − f ∞ < ε. so there exists F ∈ Cb (X. 1] : a ∈ (0. Evalued functions on X is dense in C0 (X. and so T satisﬁes the conditions of the openmapping lemma: T (Cb (X. or with an application of the openmapping lemma.19. so f (x)h(x) − f (x) = (1 − h(x)) f (x) < ε. gC = −1/3 and gD = 1/3.
and to complexvalued functions. [22.e.26 as required. Furthermore π2 : G(T ) → F . In this case we have control over (xn )n 1 and also over (T xn )n 1 . hence T = π2 ◦ π1 is bounded. which is a considerable improvement. If X and Y are normed spaces then this topology is given by the product norm. (x. A priori. Exercise 26. this is often set as an exercise [21. Proof It is a standard (and simple) result from pointset topology that any continuous function with values in a Hausdorﬀ space has closed graph. so by the −1 openmapping theorem π1 is bounded. y) → x X + y Y. (x. F be Banach spaces. Linear Operators It is an exercise to extend the above to unbounded realvalued functions.14]. . Theorem 2. see Exercise 3. Now suppose that G(T ) is closed and note that G(T ) is a subspace of E × F . T x) → T x −1 is continuous. this linear transformation is normdecreasing (so continuous) and bijective. T x) → x. as required. Let π1 : G(T ) → E.3] and a proof may be found in [19.. Exercise 4. for any sequence (xn )n 1 . so a Banach space with respect to the product norm. V ∈ S}. (ClosedGraph Theorem) Let E.3. if (xn )n 1 is such that xn → x and T xn → y then T x = y.8. A linear transformation T : E → F is bounded if and only if the graph of T . The closedgraph theorem is often used in the following manner. i. T) and (Y. T x) : x ∈ E} ⊆ E × F.3. Proposition 2. G(T ) := {(x.6]. to show that a linear transformation between Banach spaces is continuous we must show that if xn → x then T xn → T x. · X×Y : X × Y → R+ . (x. is closed (with respect to the product topology on E × F ). S) are topological spaces then the product topology on X ×Y has basis {U × V : U ∈ T. The ClosedGraph Theorem Recall that if (X.20. The closedgraph theorem means that we need only show that the graph of T contains its limit points. An application of this idea occurs in the solution to Exercise 4.
21. e2 . (ya )a∈A → yb and observe that this map is linear and normdecreasing. Y a normed space and suppose that {Ta : a ∈ A} ⊆ B(E.} and deﬁne orthogonal projections Pn ∈ B(H) by setting n Pn : H → H.8. x → ek . The Principle of Uniform Boundedness 27 The Principle of Uniform Boundedness We employ the closedgraph theorem to give a proof of the principle of uniform boundedness. For all a ∈ A we have that πa y = lim πa T xn = lim Ta xn = Ta x = πa T x. . say xn → x and T xn → y. T for all a ∈ A. for all x ∈ E. This example highlights the utility of a weaker sense of convergence for operators.22. the initial topology on B(X. Y ) generated by the family of maps {T → T x : x ∈ X} (where Y is equipped with its norm topology) is called the strong operator topology. by the Parseval equality. n→∞ n→∞ 1 ⊆ G(T ) be by the continuity of πa and Ta . Let (xn . If {Ta x : a ∈ A} ⊆ Y is bounded. Let X and Y be normed spaces. For all b ∈ A let πb : Z → Yb . x → (Ta x)a∈A and note that the pointwise boundedness of the Ta ensures that T is well deﬁned. The Strong Operator Topology Let H be a Hilbert space with orthonormal basis {e1 . and Ta x = πa T x which shows that Ta Tx T x ∀ x ∈ E. Let Ya := Y for all a ∈ A and let Z := a∈A Ya be their direct product. which shows that y = T x. Theorem 2. (BanachSteinhaus) Let E be a Banach space. . but as Pm − Pn 1 for all m = m it cannot be the case that Pn − I → 0 as n → ∞. Y ). Hence T has closed graph. so is bounded. Proof Note ﬁrst that we may replace Y by its completion.2. x ek . T xn )n convergent. Deﬁnition 2. Deﬁne T : E → Z. . so without loss of generality we assume that Y is a Banach space. k=1 It is easy to see that Pn x → x as n → ∞ for all x ∈ H. then { Ta : a ∈ A} is bounded. . also known as the BanachSteinhaus theorem.
and so the strong operator topology is coarser that the norm topology on B(X.. and similarly for strong operator convergence. realvalued functions: if X and Y are as above and f : Y → R is continuous then there exists F : X → R such that F Y = f . Linear Operators we see that norm convergence implies strong operator convergence: a net (Ta )a∈A in B(X. Prove that x ∈ ℓ .] [Let E = R3 with the norm (x1 . Exercise 3. Prove the following extension of Tietze’s theorem to complexvalued functions: if X is a normal space. Exercise 3. Y a normed vector space and suppose that (Tn )n 1 ⊆ B(E. Prove also that Tietze’s theorem applies to unbounded.}. Prove that there exists T ∈ B(E. f + g → f and PG : E → E. Deduce that B(H) is separable in the strong operator topology.e. Find a Banach space E with closed subspaces F and G such that E = F ⊕ G and P : E → E.2. Y ) is such that limn→∞ Tn x exists for all x ∈ E. PG = PG and PF PG = PG PF = 0. Hence sets that are strong operator closed are also norm closed.28 Since Tx T x ∀ x ∈ X. Y ) such that Tn → T in the strong operator topology. every element of E can be expressed uniquely as the sum of an element of F and an element of G). Y ). Y a closed subset of X and f ∈ Cb (Y ) then there exists F ∈ Cb (X) such that F Y = f and F ∞ = f ∞ . Let H be a separable Hilbert space with orthonormal basis {e1 . x3 ) = . e2 .5. x2 . x3 }. Exercises 3 Exercise 3. g ∈ G has norm strictly greater than one. .4. f + g → f ∀ f ∈ F.e. 2 2 Prove that PF and PG are bounded linear operators such that PF = PF . [Consider the mappings n fn : y → j=1 xj yj . What can be said about the norm of T ? Exercise 3.. Let E be a Banach space.7.6. Deﬁne PF and PG by setting PF : E → E. T ∈ B(X. max{x1 . Prove that if E is a Banach space with respect to two diﬀerent norms then they are either equivalent or noncomparable (i. Exercise 3. . Let E be a Banach space with closed subspaces F and G such that E = F ⊕ G (i. Y ) is convergent to T in the norm topology if and only if for all ε > 0 there exists a0 ∈ A such that Ta − T < ε for all a a0 . Let x = (xn )n 1 be a sequence of complex numbers such that the series ∞ 1 n=1 xn yn is convergent for all y ∈ c0 . prove that Pn T Pn x → T x as n → ∞ for all T ∈ B(H) and x ∈ H. Y ). x2 . g ∈ G.] Exercise 3.1. For n 1 let Pn denote the orthogonal projection onto Fe1 + · · · + Fen . f + g → g ∀ f ∈ F. neither is coarser than the other). Exercise 3. .3.
Exercises 3 Exercise 3.10. Prove that F ⊕ G is closed if and only if there exists C > 0 such that f C f +g ∀ f ∈ F. Let E be a Banach space with closed subspaces F and G such that F ∩ G = {0}. 29 Deduce that F ⊕ G is closed if and only if c := inf{ f − g : f ∈ F. f = g = 1} > 0. .8. g ∈ G.2. g ∈ G.
.
if X is ﬁnite dimensional then the spaces coincide. A linear functional on X is a linear map φ : X → F. y Example 3. Recall that (c0 )∗ ∼ ℓ1 and (ℓp )∗ ∼ ℓq = = if p ∈ [1. ∀ φ. The RieszFr´chet theorem implies that H ∗ ∼ H for any Hilbert space e = H. where the vectorspace structure is deﬁned pointwise: (φ + ψ)(x) := φ(x) + ψ(x) and (αφ)(x) := αφ(x) If X is a normed space then X ′ contains X ∗ := B(X.2. the algebraic dual space of X. An element of X ∗ is said to be a bounded linear functional on X.e. ψ ∈ X ′ .4. = if I is a subinterval of R (or a σﬁnite measure space) Lp (I)∗ ∼ Lq (I) = for the same pairs p and q. More generally.1. If X is inﬁnite dimensional then the inclusion of X ∗ in X ′ is proper. F). α ∈ F. the topological dual space of X.. Proof that every element of Lp (I)∗ arises this way requires the RadonNikod´m theorem [18.3. Being interested primarily in analysis. The Weak Topology Deﬁnition 3. the isomorphism is analogous to the ℓp case: g ∈ Lq (I) yields an element of Lp (I)∗ via f → I g(t)f (t) dt. where ∼ denotes isometric isomorphism. Let X be a vector space over the ﬁeld F.16]. Theorem 6. The set of all linear functionals on X is a vector space denoted by X ′ . Example 3. Any normed space X gains a natural topology from its dual space. its weak topology. Our notation for the algebraic and topological dual spaces is the opposite of that adopted in [10]. i. henceforth the term dual space will refer to the topological dual. ∞) and 1/p+1/q = 1. the coarsest topology 31 . This is the initial topology generated by X ∗ .Three Dual Spaces Initial Deﬁnitions Deﬁnition 3.
6]. but the HahnBanach theorem guarantees a plentiful supply (enough to ensure that the weak topology is Hausdorﬀ).7. If f ∈ M ′ is a linear functional that satisﬁes f (m) p(m) for all m ∈ M (p is a majorant for f ) then there exists a linear functional F ∈ X ′ such that F M = f and F (x) p(x) for all x ∈ X. The weak topology on X is denoted by σ(X. see [14. i.1. for all x.) The HahnBanach Theorem Deﬁnition 3. (For a proof of this. The ﬁrst involves showing that f has a “onedimensional extension”. Theorem 1. a b and b (see Deﬁnition 1. Zorn’s Lemma Deﬁnition 3. Throughout we assume (as we may) that M is a proper subspace of X. it is equivalent to the Axiom of Choice.5.32 Dual Spaces to make each map φ ∈ X ∗ continuous. The second part is an application of Zorn’s lemma (which is necessary only if X is nonseparable. either a b or b a.e. A chain C in A is a subset of A such that.) It is by no means clear that an inﬁnitedimensional space has any continuous functionals. In order to prove the HahnBanach theorem in full generality. b ∈ A. where N has codimension one in M: this should be familiar from the b4 course. b ∈ C. Let A be a set with a partial order . (Zorn) Let A be a nonempty set with a partial order in A has an upper bound then A has a maximal element. X ∗ ). . Let X be a real vector space. If every chain Proof We take the lemma as axiomatic. Proof The proof of the HahnBanach theorem falls naturally into two parts. otherwise simple induction will suﬃce). An element a ∈ A is maximal if a b implies that a = b for all b ∈ X. An upper bound or majorant for B ⊆ A is an element a ∈ A such that b a for all b ∈ B. A partial order on a set A is a preorder antisymmetric: for all a.6. we need a version of the Axiom of Choice. y ∈ X and α ∈ R+ . A sublinear functional on X is a function p : X → R such that. Theorem 3.27) that is a imply that a = b.. (HahnBanach) Let p be a sublinear functional on the real vector space X and suppose that M is a subspace of X. for all a. f extends to N. (The letter σ is used here because the German word for weak is schwach.8. . and (i) (ii) p(αx) = αp(x) (positive homogeneity) p(x + y) p(x) + p(y) (subadditivity). Lemma 3.
By positive homogeneity of p and linearity of f . g ∈ N ′ . This contradicts the maximality of (h. 33 Hence we wish to ﬁnd γ ∈ R such that f (m) − p(m − x0 ) γ −f (n) + p(n + x0 ) ∀ m. N) for all a ∈ A. and furthermore fγ M = f . x → ga (x) if x ∈ Na . P ) ∈ S that is maximal for . It is a simple exercise to see that (g. by the ﬁrst part of this proof.9. by the subadditivity of p and linearity of f . Φ(x) p(x) for all x ∈ X). . It remains to show that we may ﬁnd γ ∈ R such that fγ (x) p(x) for all x ∈ N.. then P is a proper subspace of X and there exists a proper extension of h. m + αx0 → f (m) + αγ. n ∈ M but. if we can show that P = X then we are done.4. If φ ∈ M ′ is a linear functional such that φ(m) p(m) for all m ∈ M (φ is dominated by p) then there exists a linear functional Φ ∈ X ′ that extends φ (i. Na ) (g. Na ) : a ∈ A} be a chain in (S. N := a∈A Na and deﬁne g : N → R. This is a sound deﬁnition because N is the direct sum of M and Rx0 . The HahnBanach Theorem Choose a vector x0 ∈ X \ M. let and note that this is a partial order. ΦM = φ) and is dominated by p (i. the previous part shows that this set is nonempty. By Zorn’s lemma we conclude that there exists (h. α ∈ R. N) ∈ S and (ga .3. −f (n) + p(n + x0 ) − f (m) + p(m − x0 ) so such γ exists. the above inequality will be satisﬁed if f (m) + γ p(m + x0 ) ∀ m ∈ M and f (m) − γ p(m − x0 ) ∀ m ∈ M. g(x) p(x) ∀ x ∈ N −f (n + m) + p(n + m) 0. f (m) + αγ p(m + αx0 ) ∀ m ∈ M. Deﬁne on S by saying that (g. gM = f. i.e. P ).. Theorem 3. Suppose otherwise. denote all suitable extensions of f . let N := M + Rx0 and for all γ ∈ R deﬁne fγ : N → R. M ⊆ N. Let {(ga .e.e. P ) ⇐⇒ N ⊆ P and hN = g ). N) (h. Let S = (g.. N) : N is a subspace of X. (BohnenblustSobczyk) Let p be a seminorm on the vector space X and suppose that M is a subspace of X.
Combining these observations gives the result. Deﬁne Φ by Φ : X → C. so we may apply the HahnBanach theorem to obtain Φ ∈ X ′ such that ΦM = φ and Φ(x) p(x) for all x ∈ X. Then Φ(x) = αΦ(x) = Re αΦ(x) = Re Φ(αx) = F (αx) and so Φ is dominated by p.34 Dual Spaces Proof Suppose ﬁrst that X is a real vector space. p(αx) = p(x) . but also −Φ(x) p(−x) = p(x) ∀ x ∈ X by the homogeneity of the seminorm p. This proves that the weak topology on X is Hausdorﬀ: if x.26. Finally. as required. The Dual Space Separates Points Theorem 3. αx → α x . φ(x) = x ) and φ(y) y for all y ∈ X (so that φ 1). Let X be a normed space. and if a. b ∈ R then Φ((a + ib)x) = F ((a + ib)x) − iF (i(a + ib)x) = F (ax) + F (ibx) − iF (iax) − iF (−bx) = (a + ib)F (x) − i(a + ib)F (ix) = (a + ib)Φ(x) ∀ x ∈ X. Hence Φ(x) p(x) for all x ∈ X. By Theorem 3. y ∈ X are such that x = y then there exists φ ∈ X ∗ such that φ(x − y) = x − y = 0. Note that a seminorm is a sublinear functional. let x ∈ X and choose α ∈ T := {z ∈ C : z = 1} such that αΦ(x) ∈ R+ . Note also that Φ(m) = F (m) − iF (im) = Re φ(m) − i Re iφ(m) = Re φ(m) + i Im φ(m) = φ(m) ∀ m ∈ M. the claim follows by Proposition 1. as required. For all x ∈ X \ {0} there exists φ ∈ X ∗ such that φ(x) = x and φ = 1. Proof Let M = Fx := {αx : α ∈ F} and deﬁne f : M → F. It is clear that Φ is additive. so Φ ∈ X ′ . so ΦM = φ. We may regard it as a real vector space and apply the ﬁrst part of this proof to obtain a reallinear functional F on X that extends Re φ and is dominated by p. Now suppose that X is a complex vector space.10.9 there exists a linear functional φ : X → F such that φM = f (in particular. Note that f (αx) = α x = αx and so f is dominated on M by · . x → F (x) − iF (ix).
If X is a normed vector space then it has a completion: there exists ˜ ˜ ˜ a Banach space X and a linear isometry i : X → X such that i(X) is dense in X. Y ) complete then Y is complete. it is not diﬃcult to check that the completion inherits the structure of a normed space [3. Proof Let (yn )n 1 ⊆ Y be a Cauchy sequence and let x0 ∈ X be a unit vector. Y ) is Complete Proposition 3. As X ∗∗ is complete (being the dual of a normed space) and closed ˜ subspaces of Banach spaces are complete (Proposition 1. pp.8). as required. more pedestrian way of ﬁnding the completion of a normed space (or any metric space) which mimics the way that the real numbers may be constructed as a collection of equivalence classes of sequences of rational numbers: see [3. For n 1 deﬁne Tn ∈ B(X. Theorem 2.2]. If the canonical embedding is surjective then X is said to be reﬂexive. Y ) by setting Tn x = φ(x)yn and note that Tn x0 = yn .2.12. ˆ the previous remarks that i is a linear isometry.10 there exists a linear functional φ ∈ X ∗ such that φ = 1 = φ(x0 ). Let T be the limit of this sequence and conclude by noting that n→∞ lim yn = lim Tn x0 = T x0 . Y ). Y is Complete if B(X. For all x ∈ X deﬁne a linear functional x on X ∗ by setting x(φ) = φ(x) and note that ˆ ˆ ˆ(φ) = φ(x) x so that x ∈ X ∗∗ with x ˆ ˆ φ x 35 x .4. Proof ˜ Let X := Γ(X) be the closure in the bidual X ∗∗ of the image of X under the canonical embedding. n→∞ . There is another. Let i : X → X. X. Theorem 11. called the canonical embedding of X into its bidual. Theorem 3. Proposition 3. x → x ˆ is an isometry.10 shows that the map Γ : X → X ∗∗ . Then (Tn − Tm )x = φ(x) yn − ym yn − ym x ∀x ∈ X and so Tn − Tm yn − ym .3.7]. By Theorem 3. 34–35] or [22.11. The HahnBanach Theorem Existence of Completions Recall that X ∗∗ = (X ∗ )∗ is the bidual or second dual of the normed space X. · X ∗∗ X is a Banach ˜ ˜ x → x. it is immediate from space containing X as a dense subspace. which shows that (Tn )n 1 is a Cauchy sequence in B(X. If X and Y are normed vector spaces with X = {0} and B(X.
This is not a generalisation. and is strongly holomorphic if Theorem 3. Proof For all φ ∈ X ∗ we have that φ ◦ f : C → C is bounded and holomorphic everywhere. Y ). T ∈ B(X. h→0 h where this limit is taken with respect to the norm topology on X. f (C) ⊆ Xr ) then f is constant. as a subset of a normed space is weakly bounded if and only if it is norm bounded (Exercise 4. Let X be a complex normed space and let U be an open subset of C. Y ) shows that strong operator convergence implies weak operator convergence. Note that every strongly holomorphic function is weakly holomorphic. Y be normed vector spaces and let T ∈ B(X. Y be normed spaces.14.e. i. φ ∈ X ∗ . φ ∈ Y ∗ . however. φ ∈ Y ∗ } (where F is equipped with its usual topology) is called the weak operator topology. ..13. Let X. As X ∗ separates points in X we must have that f (z) = f (0) for all z ∈ C. Let X. i. the intial topology on B(X. lim Deﬁnition 3. the fact that φ(T x) φ Tx ∀ x ∈ X. Adjoint Operators Theorem 3. § 5. Hence φ(f (z)) = φ(f (0)) ∀z ∈ C. If f is bounded. The proof holds if the holomorphic function f is only required to be weakly bounded.e. and so constant. there exists r ∈ R+ such that f (z) r for all z ∈ C (equivalently. and so the weak operator topology is coarser than the strong operator topology. (Liouville) Let f : C → X be a weakly holomorphic function into the complex normed space X. There exists T ∗ ∈ B(Y ∗ .16. thus sets that are weak operator closed are also strong operator closed.36 Vectorvalued Holomorphic Functions Dual Spaces f (z + h) − f (z) exists ∀ z ∈ U. Y ) generated by the collection of maps {T → φ(T x) : x ∈ X..6).15. i. φ ∈ Y ∗ . The Weak Operator Topology Deﬁnition 3. for all φ ∈ X ∗ there exists rφ ∈ R+ with φ(f (z)) rφ for all z ∈ C. X ∗ ) such that φ(T x) = (T ∗ φ)(x) ∀ x ∈ X.2]. f is constant. by the classical Liouville’s theorem [16. In the same manner as we compared strong operator and norm convergence..e. A function f : U → X is weakly holomorphic if φ ◦ f : U → C is holomorphic for all φ ∈ X ∗ .
⊥ (im T ∗ ). Theorem 3. ˆ ∗∗ Note also that φ → φ(x) ∈ X for all x ∈ X.. ker T = {x ∈ X : T x = 0} = {x ∈ X : φ(T x) = 0 for all φ ∈ Y ∗ } = {x ∈ X : (T ∗ φ)(x) = 0 for all φ ∈ Y ∗ } = The other identity can be established in the same manner.19. Y ). i. Then ker T = ⊥ (im T ∗ ) and ker T ∗ = (im T )⊥ . Prove that M is complemented in X if M is ﬁnite dimensional. ψ ∈ X ∗ are distinct then there exists x ∈ X such that φ(x) = ψ(x). by Proposition 1. If M is a subspace of the normed space X and N is a subspace of X ∗ then M ⊥ := {φ ∈ X ∗ : φ(x) = 0 for all x ∈ M} is the annihilator of M and ⊥ 37 N := {x ∈ X : φ(x) = 0 for all φ ∈ N} is the preannihilator of N.7. since Y ∗ separates points in Y . the coarsest topology to make these maps continuous. The weak* topology on X ∗ is denoted by σ(X ∗ . X) is even coarser than the weak topology σ(X ∗ . X ∗∗ ): there are fewer functions required to be continuous.1. Note that the weak* topology is Hausdorﬀ. Deﬁnition 3. X).17.e. Let X and Y be normed spaces and let T ∈ B(X. φ → φ(x) ˆ (x ∈ X).2. A closed subspace M of the normed space X is complemented in X if there exists a closed subspace N such that M ⊕ N = X. M + N = X and M ∩ N = {0}.26: if φ. The Weak* Topology Deﬁnition 3.3. Theorem 2. Exercises 4 Exercise 4. The weak* topology on X ∗ is the initial topology generated by the maps x : X ∗ → F. [Start by considering a . Proof This is a b4 result [24. Let X be a normed vector space. so the weak* topology σ(X ∗ . i.. Proof Note that.e. The Weak* Topology Furthermore T ∗ = T . and so the map x separates these points.13].18.
Let H be a Hilbert space. dim X/M < ∞. Prove that A is norm bounded (there exists r ∈ R+ such that a r for all a ∈ A) if and only if it is weakly ∗ + bounded (for all φ ∈ X there exists rφ ∈ R such that φ(a) rφ for all a ∈ A). [For the latter claim. with S = T ∗ . Exercise 4.e.5.7. Let M be a ﬁnitedimensional subspace of the normed space X and let N be a closed subspace of X such that X = M ⊕ N. F ) has closed range. [You ¯ may assume that if M is a nonempty subspace of X and x0 ∈ X \ M then there exists ∗ φ ∈ X such that φM = 0 and φ(x0 ) = 1. Prove that the adjoint T → T ∗ is continuous with respect to the weak operator topology on B(H).. ∞ n=1 xn is . so that E ∗ = ℓ1 and E ∗∗ = ℓ∞ . Let E and F be Banach spaces and suppose that T ∈ B(E. Exercise 4. i. Let A be a subset of the normed vector space X. Prove that a Banach space E is reﬂexive if and only its dual E ∗ is Exercise 4. im T is closed in F .] Exercise 4. n ∈ N is an element of the dual space X ∗ . m + n → φ0 (m) ∀ m ∈ M. i.8. [Use the principle of uniform boundedness and the canonical embedding Γ : x → x. Prove that a normed vector space X is separable if its dual X ∗ is. x en for all x ∈ ℓ2 . φ ∈ F ∗ then S and T are bounded.10.] Deduce ˆ that a weakly holomorphic function is (strongly) continuous. Exercise 4.e.6. but not necessarily with respect to the strong operator topology. Prove that if φ0 is a linear functional on M then φ : M ⊕ N → F.38 Dual Spaces basis of M ∗ .] Find a separable Banach space E such that E ∗ is not separable. Prove that x → weakly continuous on ℓ1 but is not weak* continuous. Show that if T : E → F and S : F ∗ → E ∗ are linear transformations that satisfy φ(T x) = (Sφ)(x) ∀ x ∈ E.] Prove also that M is complemented in X if M has ﬁnite codimension. Prove that im T ∗ = (ker T )⊥ (where M ⊥ := {φ ∈ E ∗ : φ(x) = 0 for all x ∈ M} is the annihilator of the subspace M ⊆ E). Let E and F be Banach spaces. Prove that any inﬁnitedimensional normed space has a discontinuous linear functional deﬁned on it.] Exercise 4.2.9. reﬂexive..4. Exercise 4. [Use the closedgraph theorem. [Proof of (non)separability is not required. consider the operators Tn ∈ B(ℓ2 ) such that Tn x = e1 . Exercise 4. where {ek : k 1} is the standard orthonormal basis of ℓ2 .] Prove that a reﬂexive Banach space E is separable if and only if E ∗ is. [Cauchy’s integral formula may be useful.] Exercise 4.3. Let E = c0 .
[For surjectivity. p. . [Let (xn )n 1 ⊆ X1 be dense in X1 and consider d(φ. (xa )a∈A → xb (b ∈ A). U1 ∈ Ta1 . . . this is Tychonov’s theorem.11. with x ⊗ y = x y . y) → x ⊗ y is bilinear. the closed unit ball of the dual space X ∗ . . y ∈ Y } in B(X.] (iii) If X is the product of {(Xa . . j(T )z = z(T ) 39 is an isometric isomorphism. Prove that if X is a ∗ separable normed space then X1 .e. Fn ∈ F. Ta ) : a ∈ A be a collection of topological spaces. Y ∗ ) → Z ∗ . and that the mapping X × Y → B(X. (i) The space (X. . we remind ourselves of some minor points from basic topology. . If Z is the closed linear span of {x ⊗ y : x ∈ X.3. Their topological product is (X. T).] Deduce that X ∗ is separable in the weak* topology. × a∈A Xa := {(xa )a∈A : xa ∈ Xa ∀ a ∈ A} The fundamental fact about the product topology is that is preserves compactness. i. n=1 Exercise 4. Prove that a compact metric space is separable. Y ∗ )∗ .. to avoid clutter. Y ∗ )∗ . let φ ∈ Z ∗ and consider φx : y → φ(x⊗y). Un ∈ Tan ). T) is compact if and only if every collection of closed subsets of X with the ﬁniteintersection property has nonempty intersection. Tychonov’s Theorem Exercise 4.] ¯ (ii) If A ⊆ X then x ∈ A if and only if every open set containing x meets A. . . . . Ta ) : a ∈ A} then every set in T is the union of sets of the form n −1 πai (Ui ) i=1 (n ∈ N. prove that j : B(X. the initial topology generated by the projection maps πb : X → Xb . see [21. . a collection F of subsets of X has the ﬁniteintersection property if F1 ∩ . [For a proof of this.9. Prove that x ⊗ y ∈ B(X. .D] or [3. an ∈ A. [Otherwise ◦ ¯ x ∈ (X \ A) = X \ A. where X= is the Cartesian product of the sets Xa and T = a∈A Ta is the product topology. (x. these sets form a base for T.20. Let X and Y be normed spaces and for all x ∈ X and y ∈ Y let x ⊗ y : B(X. Theorem 21. T → (T x)(y). Y ∗ ). Let (Xa . .116]. The proof of Tychonov’s theorem is not particularly simple. Y ∗ ) → F. a1 . is metrizable when equipped with the weak* topology. T) be a topological space. ∩ Fn = ∅ for all F1 . Let (X.] Tychonov’s Theorem Deﬁnition 3.12. ψ) := ∞ 2−n φ(xn )−ψ(xn ). .
by Theorem 1. which contradicts the maximality of H. An ∈ H then A = A1 ∩ . As H has the ﬁniteintersection property. and so U ∈ H. Note that H is closed under ﬁnite intersections: if A1 . so F −1 is well deﬁned on ∗ ∗ F (X1 ). contains F and has the ﬁniteintersection property. the space K = the product topology.22.40 Dual Spaces Theorem 3.25. where Ua ∈ Ta contains xa and A0 is a ﬁnite subset of A. as required. Let a ∈ A and note that {πa (H) : H ∈ H} has the ﬁniteintersection property (because f ( b∈B Sb ) ⊆ b∈B f (Sb ) for any function f and collection of sets {Sb }). y ∈ X.34 it suﬃces to take a net (φa )a∈A in X1 such that F (φa ) a∈A ∗ has limit f = (fx )x∈X ∈ K and show that φ : x → fx ∈ X1 . To ∗ prove this. If a ∈ A0 then xa ∈ Ua ∩ πa (H). . the closed unit ball of X ∗ . U H = ∅ ¯ for all H ∈ H. by Proposition 1. if A ⊆ X is such that A ∩ H = ∅ for all H ∈ H then A ∈ H.37. (The linearity result used here follows from continuity of addition and multiplication in F. as required.) . By Zorn’s lemma we may ﬁnd a maximal family H of (not necessarily closed) subsets of X such that H contains F and has the ﬁniteintersection property. Furthermore. as H is closed under ﬁnite intersections. so φ 1.21. by Proposition 1. This map is weak* continuous. (Tychonov) Let {(Xa . . We complete the proof by showing that x := (xa )a∈A ∈ H∈H H. and X1 = F (F (X1 )) is weak* compact if F (X1 ) is closed in K. . otherwise considering H ∪ {A} leads to the same contradiction. which suﬃces to show that F ∈F F = ∅. Hence πa (Ua ) ∈ H for all a ∈ A0 . . The product space (X. for all H ∈ H. The BanachAlaoglu Theorem ∗ Theorem 3. Hence φ : x → fx ∈ X ′ . and so fx + αfy = lim φa (x) + α lim φa (y) = lim φa (x + αy) = fx+αy a∈A a∈A a∈A ∀ x. Note that F (φa ) → f if and only if φa (x) → fx . φ → φ(x) x∈X is well deﬁned (since φ(x) φ x x ) and injective. Ta ) : a ∈ A} be a collection of compact topological spaces. so there exists xa ∈ H∈H πa (H) by the compactness of Xa . The map × x∈X F x is compact when equipped with ∗ F : X1 → K. (BanachAlaoglu) If X is a normed space then X1 . Proof By Tychonov’s theorem. and therefore x ∈ H∈H H. so −1 −1 πa (Ua ) ∩ H = ∅. T) is compact. is compact in the weak* topology. −1 Any open set containing x contains a set of the form U = a∈A0 πa (Ua ). and fx ∈ F x for all x ∈ X implies that φ(x) = fx  x . α ∈ F. Proof Let F be a family of closed subsets of X with the ﬁniteintersection property. since xX1 ◦ F −1 = πx F (X1 ) ˆ ∗ ∗ −1 ∗ ∗ for all x ∈ X. . ∩ An ∈ H. as otherwise H ∪ {A} strictly contains H. .
a subspace of C(K). If M ⊆ X ′ is a separating subspace then σ(X. i(x) ∞ 41 = sup{i(x)(φ) : φ ∈ K} = sup{φ(x) : φ ∈ K} = x . ˆ i(x) : K → F. There exists a compact. B(H) is the natural noncommutative generalisation of C(K)..24. so i(X) is closed if and only if X is complete: a subspace of a Banach space is closed if and only if it is complete (Proposition 1. (This is Hausdorﬀ. Hence i is an isometry.e. An operator space is a closed subspace of some B(H). Topological Vector Spaces Characterisation of Normed Vector Spaces The following theorem reduces the study of normed vector spaces to the study of subspaces of a particular type of Banach space. . Proof ∗ Let K = X1 be the closed unit ball of X ∗ . The weak and weak* topologies are deﬁned in this fashion. Topological Vector Spaces Deﬁnition 3.8) and isometries preserve Cauchy sequences. Let X be a vector space. Furthermore.23.11. equipped with the (restriction of the) weak* topology.25. M)continuous if and only if φ ∈ M. A set of linear functionals A ⊆ X ′ is separating if for all x ∈ X \ {0} there exists φ ∈ A such that φ(x) = 0. Let (X. σ(X ∗ . the collection of continuous functions on a compact. In particular.) Deﬁne i : X → C(K) by setting i(x) = xK . all Banach spaces are isomorphic to closed subspaces of C(K) for some compact. Hausdorﬀ space K and a linear isometry i : X → C(K) such that X is isometrically isomorphic to i(X). φ → φ(x). this map is continuous by the deﬁnition of the weak* topology. which is closed if and only if X is complete. Let X be a vector space with separating subspace M ⊆ X ′ . A linear functional φ ∈ X ′ is σ(X. The previous theorem is the starting point which motivates the theory of operator spaces: every Banach space is isometrically isomorphic to a closed subspace of some C(K). and i is clearly linear. Hausdorﬀ space. X) . which is the paradigm example of a commutative C ∗ algebra. this topology is Hausdorﬀ by Proposition 1. M) := TM is the initial topology on X generated by M. by Theorem 3. This is a very active area of current research (see [5] or [15]). Theorem 3. i.26. being a subspace of the Hausdorﬀ space X ∗ . Hausdorﬀ space K. · ) be a normed space. where H is a Hilbert space.10.3. Proposition 3.
e. The proof for the other function is similar (and depends upon the continuity of addition in F). Continuity of multiplication in F yields φ(αa xa ) = αa φ(xa ) → αφ(x) = φ(αx) ∀φ ∈ M. y) → x + y are continuous (where X is equipped with the topology σ(X. As φ(y) − φ(x) = φ(y − x) < εx ∀ y ∈ x + εx U. (The fact that x + εx U is open follows as the maps y → y + x and y → εx y are homeomorphisms of X to itself. (x. Proof One implication is immediate from the deﬁnitions. A linear functional φ ∈ X ′ is continuous if and only if φ−1 [0.42 Proof See Exercise 5.35.27. Lemma 3. x). there exists εx > 0 such that Bx := Bεx φ(x) ⊆ A. (x. F × X → X. For the converse. x) → φ(αx) is continuous for all φ ∈ M. x) → αx and X × X → X. Let X be a vector space over F and let M ⊆ X ′ be a separating subspace. we see that and so φ−1 (A) = x∈φ−1 (A) (x + εx U) is open. The functions F × X → X. (α. by Proposition 1. φ(xa ) → φ(x) for all φ ∈ M. Dual Spaces Proposition 3. x) → αx Any normed space. M) and F has its usual topology). or vector space equipped with the topology given by a separating subspace of linear functionals.37) αa → α and xa → x. (α. as required.28. x) in F × X then (by the deﬁnition of the product topology and Proposition 1. A topological vector space is a vector space X equipped with a Hausdorﬀ topology such that the maps X × X → X. Let X be a topological vector space. suppose that U ⊆ X is an open set containing 0 and such that φ(u) < 1 for all u ∈ U.) φ−1 (A) ⊇ φ−1 (Bx ) ⊇ x + εx U . by Proposition 1.. 1) contains an open set containing 0. y) → x + y and are continuous. xa )a∈A ⊆ F×X such that (αa . as required. xa ) → (α.1. xa ) → (α. Proof The function (α.26.25. x) → αx is continuous if (α. If (αa . is a topological vector space. Let A ⊆ F F be open and let x ∈ φ−1 (A). and this function is continuous if φ(αa xa ) → φ(αx) for any net (αa . i. Deﬁnition 3.
B be nonempty. C 43 Proof If x ∈ X then mx : R → X. or a topological vector space with topology given by a separating subspace of linear functionals.11. open set in X that contains the origin then the map µC : X → R+ . s+t s+t by the convexity of C. A subset C of a vector space is convex if tC + (1 − t)C ⊆ C for all t ∈ (0.30.. whence t−1 x ∈ C and so x = (1 − t)0 + t(t−1 x) ∈ C because 0 ∈ C and C is convex. [Geometrically. t → tx is continuous.8). y ∈ C and t ∈ (0. Since ε > 0 is arbitrary we have the subadditivity of µC . 1). x ∈ sC if s > δ −1 . Let X be a real topological vector space. if µC (x) < 1 then x ∈ tC for some t < 1. t < µC (y) + ε and x ∈ sC. Conversely. (ii) If X is locally convex. this condition states that every line segment with endpoints in C lies in C. t > 0 such that s < µC (x) + ε. Let X be a topological vector space and let A. . convex subsets of X.e. if x ∈ C then (1 + ε)x ∈ C for some ε > 0 (because m−1 (C) is open and x contains 1) and so x ∈ (1 + ε)−1 C. we may ﬁnd s.. It is immediate that linear transformations preserve convexity.] Lemma 3.29. 1). is locally convex (Exercise 5. Topological Vector Spaces Separation Deﬁnition 3. This shows that µC is well deﬁned. i. (i) If A is open then there exists a continuous linear functional φ ∈ X ′ and s ∈ R such that Re φ(x) < s Re φ(y) ∀ x ∈ A. Then x + y ∈ sC + tC = (s + t) t s C+ C ⊆ (s + t)C. and µC (x + y) s + t < µC (x) + µC (y) + 2ε. Theorem 3. Deﬁnition 3. Let x. Finally. 1) = {x ∈ X : µC (x) < 1}. so m−1 (C) is open and contains 0. tx + (1 − t)y ∈ C for all x. x Hence there exists δ > 0 such that tx ∈ C if t < δ.e. If C is a convex.3. hence µC (sx) = sµC (x) for all x ∈ X and s ∈ R+ . y ∈ B. x → inf{t ∈ R+ : x ∈ tC} (called the gauge or Minkowski functional of C) is a sublinear functional on X such that C = µ−1 [0. disjoint. y ∈ X and suppose that ε > 0. i. A normed space. y ∈ tC. Positive homogeneity is immediate: if s > 0 then x ∈ tC if and only if sx ∈ stC. A topological vector space is locally convex if every open set containing the origin contains a convex open set containing the origin. which yields µC (x) < 1. x0 ∈ X \ B and B is closed then there exists a continuous linear functional φ ∈ X ′ such that Re φ(x0 ) < inf{Re φ(y) : y ∈ B}.31.32.
convex subsets of R and φ(A) is open (see Exercise 5. 0) < ∞ .34. note that (X \ B) − x0 is open and contains 0. by Theorem 3. Proof Let B = {0}. Example 3. since C ∩ (−C) is an open set containing the origin.28). For (ii). Then φ0 µC on N. Let 0 < p < 1. there exists a (reallinear) functional φ such that φ(x0 ) = φ0 (x0 ) = 1 and φ µC .9: if φ is a continuous. b ∈ B then a − b + x0 ∈ C and so φ(a) − φ(b) + 1 = φ(a − b + x0 ) µC (a − b + x0 ) < 1. b0 ∈ B and consider C = A − B + (b0 − a0 ). x0 = x and apply Theorem 3. Let N = Rx0 / and deﬁne φ0 on N by setting φ0 (tx0 ) = t for all t ∈ R. In other words. where N := {f ∈ Lp [0. 1] = Lp [0. is open (being the union of translates of the open set A) and is convex (this is immediate upon checking the deﬁnition). Hence φ(x) < 1 on C ∩ (−C). and so. [g]) → d(f.32(ii). 1] is the zero functional (Exercise 5. Then A := x0 + U is convex. 1] → C f is measurable and d(f.44 Dual Spaces Proof First. Lp [0. 1] := f : [0. Then φ(A) and φ(B) are disjoint. However. 1] : d(f. hence φ(a) < φ(b). the result follows by (i). so taking s = sup φ(A) gives the result. Corollary 3. and let Lp [0. g) is a metric on Lp [0.8. 0) = 0} is the subspace of functions zero almost everywhere. this topology is not locally convex and the only continuous linear functional on Lp [0. working as in the last part of the proof of Theorem 3. note that we may assume that the scalar ﬁeld F = R. If a ∈ A.3). φ is continuous (by Lemma 3. where 1 d(f. . The map ([f ]. open and contained in X \ B. For (i). 1] which makes it a topological vector space (with algebraic operations deﬁned pointwise). and therefore µC (x0 ) 1. Since A and B are disjoint. reallinear functional on X satisfying (i) or (ii) then Φ : x → φ(x) − iφ(ix) is an continuous element of X ′ with the same property (since Re Φ = φ).6). 1]/N. If X is a locally convex topological vector space and x ∈ X \ {0} then there exists a continuous linear functional φ ∈ X ′ such that φ(x) = 0. g) := 0 f (t) − g(t)p dt. it contains 0. so by local convexity there exists a convex. open set U such that 0 ∈ U ⊆ (X \ B) − x0 . the topological dual of X separates points. x0 := b0 −a0 ∈ C. let a0 ∈ A. Hence the gauge µC is sublinear on X. If x ∈ C then φ(x) µC (x) < 1 and if x ∈ −C then φ(x) = −φ(−x) > −1.33.
32(i) with A = Bε (x) and B = C yields φ ∈ X ∗ and s ∈ R such that {y ∈ X : Re φ(y) < s} ∩ C = ∅. . If X is a topological vector space then the closed convex hull of A. .e. Corollary 3. Proof Since the weak topology is coarser than the norm topology. y ∈ C satisfy tx + (1 − t)y ∈ F then x and y ∈ F . Proposition 3. the previous proposition holds for all locally convex topological vector spaces: the proper generalisation of the weak topology is the initial topology generated by all the continuous linear functionals (Exercise 5. An extreme point is a point of C that is not contained in the interior of any line segment in C.25 is the fact that the collections of normcontinuous and weakly continuous linear functionals on a normed space coincide. . As (Re φ)−1 (−∞. denoted by cnv A. convex subset C of the normed space X. . The KreinMilman Theorem A consequence of Proposition 3. .3. convex set containing A. If X is a topological vector space then cnv A is the smallest closed. [We blur the distinction between an extreme point and the singleton set containing it. i. combined with the separation theorem this yields the following. convex subset F ⊆ C such that if t ∈ (0.37..9). In fact. α1 . xn ∈ A is the convex hull of A.12. s) is weakly open and contains x.36. [Some pictures would go well here.] The extremal boundary of C is the set of its extreme points. is the closure of cnv A.38. an element of C that cannot be expressed as a nontrivial convex combination of elements of C. 45 The KreinMilman Theorem Deﬁnition 3. + i=1 αi = 1. denoted by ∂e C In geometrical terms. x1 . a face F is a subset of the convex set C such that if f ∈ F and ℓ is a line through f then ℓ ∩ C ⊆ F . applying Theorem 3. we need only consider a nonempty. A convex subset of a normed space is norm closed if and only if it is weakly closed. . If X is a vector space and A ⊆ X then cnv A is the smallest convex set containing A. . 1) and x. αn ∈ R .] Deﬁnition 3. Let X be a vector space. normclosed. A face of a convex set C is a nonempty. An extreme point of a convex set C is a onepoint face. If x ∈ C then there exists ε > 0 such that Bε (x) ∩ C = ∅.35. If X is a vector space and A ⊆ X then n n cnv A := i=1 αi xi : n ∈ N. . it follows that C is weakly closed and we have the result. .
Without loss of generality αn = 1 and n n−1 i=1 αi xi = (1 − αn ) i=1 αi xi + αn xn ∈ B 1 − αn by the inductive hypothesis and convexity. Hence x. if t ∈ (0. This proves the ﬁrst statement. and similarly for y. Let B be any convex set containing A. . . Proof The set F is nonempty (since C is compact and x → Re φ(x) is continuous).. 1). Furthermore. ∂e C is nonempty. .39. If B is a closed.) := tx + (1 − t)y = t lim xp + (1 − t) lim yp = lim txp + (1 − t)yp ∈ cnv A cnv A (3. convex subset of X. If φ ∈ X ′ is continuous then F := x ∈ C : Re φ(x) = min{Re φ(y) : y ∈ C} is a closed face of C. (In particular. . compact. closed (since it is the preimage of a point under the continuous function Re φ) and convex (since Re φ is reallinear). Then C = cnv ∂e C.2) . i. n n i=1 αi xi ∈ B if x1 .2). Lemma 3. we claim that cnv A ⊆ B. a contradiction. .. yp ) → (x.40. By continuity of scalar multiplication and vector addition. which gives a strict inequality in (3. we proceed by induction: the cases n = 1 and 2 are immediate. 1) and x. for all n ∈ N. as required. Theorem 3. compact. C is the closed convex hull of its extreme points. .) If x is not in F then Re φ(x) > minC Re φ. as required. It remains to prove that cnv A is convex. i=1 (3. p∈P p∈P p∈P for all t ∈ (0. (KreinMilman) Let X be a locally convex topological vector space and let C be a nonempty.1) holds for sums containing n − 1 terms. αn ∈ R with + αi = 1. yp )p∈P ⊆ cnv A × cnv A such that (xp . let x. y ∈ F .1) To see this. i.e. Let X be a topological vector space and let C be a nonempty.46 Dual Spaces Proof The convexity of cnv A is readily veriﬁed. convex subset of X. y) (recall that the closure of a product is the product of the closures). convex set containing A then cnv A ⊆ B (since B is convex) and ¯ hence cnv A ⊆ B = B (taking closures). so suppose that n > 2 and that (3. y ∈ cnv A and choose a net (xp . C (The notation min means that the minimum is taken over the set C.e. y ∈ C are such that tx+(1−t)y ∈ F then min Re φ = Re φ(tx + (1 − t)y) = t Re φ(x) + (1 − t) Re φ(y) C C min Re φ. xn ∈ A and α1 . . .
We claim that F0 ∈ ∂e C. a closed face F0 such that if F1 ∈ F satisﬁes F0 ⊇ F1 then F1 = F0 . . with F in place of C. Prove that if X is a vector space with separating subspace M ⊆ X ′ and φ ∈ X ′ is a linear functional that is σ(X. 1 i n Deduce that n i=1 ker φi ⊆ ker φ and that there exists f ∈ (Fn )∗ such that f φ1 (x). . and applying the ﬁrst part of this proof. iφ or −iφ). φn ∈ M such that φ(x) max φi (x) ∀ x ∈ X. Exercise 5. . Exercises 5 Exercise 5. .32(ii) (applied to x and cnv ∂e C) we may ﬁnd a continuous ψ ∈ X ′ such that Re ψ(x) < min{Re ψ(y) : y ∈ cnv ∂e C}. Let X be an inﬁnitedimensional normed space and let V ⊆ X be a weakly open set containing the origin. Exercise 5. this is a proper subset of F0 (since y ∈ F1 ) and a closed face of F0 . to see this.3. Applying Zorn’s lemma we obtain a maximal element of F. . This contradicts the minimality of F0 and so ∂e C = ∅. i.33 there exists a continuous φ ∈ X ′ such that φ(x) = φ(y).3. φn (x) = φ(x) ∀ x ∈ X. by Lemma 3. Show that V contains a closed subspace of ﬁnite codimension in X. Hence min Re ψ = Re ψ(z) > Re ψ(x) C 47 min Re ψ. Exercises 5 Proof Let F denote the collection of closed faces of C. . this is a closed face of C.. so to complete the proof suppose for contradiction that x ∈ C \ cnv ∂e C. so that there exist distinct x. . [Note that mx : α → αx is continuous for all x ∈ X and that there exists x0 ∈ X such that φ(x0 ) = 1. Let X be a topological vector space. yields z ∈ ∂e F ⊆ ∂e C. partially ordered set. and without loss of generality Re φ(x) < Re φ(y) (else we replace φ by one of −φ. By Corollary 3. . it is an exercise to verify that (F. by Lemma 3. . y ∈ F0 . Deduce that the weak topology on X is strictly coarser than the norm topology.] Conclude that φ ∈ M. By Theorem 3. Let F = z ∈ C : Re ψ(z) = min{Re ψ(w) : w ∈ C} .1. such that every chain in F has an upper bound.39. suppose otherwise. Prove that every φ ∈ X ′ \ {0} is open. Let F1 = z ∈ F0 : Re φ(z) = min{Re φ(w) : w ∈ F0 } .e.13. It is immediate that C ⊇ cnv ∂e C.2. ⊇) is a nonempty.39. M)continuous then there exist φ1 . C the desired contradiction. and / so a closed face of C: a face of a face of C is itself a face of C.
1] is a topological vector space (when equipped with this topology). Deduce that the only continuous linear functional on Lp [0. 1). 1] has no convex. open sets other than ∅ and Lp [0. 1] → C f is measurable and zero almost everywhere .7. Exercise 5. Prove that d([f ]. Lp [0.5. Exercise 5. B ⊆ X such that C ⊆ A and B is a balanced set containing 0. Exercise 5. the maps X × X → X. 1] := Lp [0. [Balanced sets are in some ways analogous to open balls about the origin in normed spaces. Suppose that X is a locally convex topological vector space and M is the collection of continuous linear functionals on X. i.9. {x} is closed for all x ∈ X) then the topology is Hausdorﬀ (so X is a topological vector space). Let X be a topological vector space. where ∆(f ) := 1 0 f (x)p dx.11. Need this hold if X is not locally convex? Exercise 5. Exercise 5.10. Prove there exists an open set U containing 0 such that U + U ⊆ V . 1]. y) → x + y and F × X → X. 1] → C f is measurable and ∆(f ) < ∞ . Let p ∈ (0. continuous. Prove also that M is closed in X. Suppose that X is a vector space equipped with a topology that makes vector addition and scalar multiplication. Exercise 5. Prove that a convex subset of X is closed (with respect to the original topology) if and only if it is closed with respect to σ(X. 1] is the zero functional. M). Let X be a topological vector space and suppose V is an open set containing 0.] Deduce that if C ⊆ X is compact and does not contain the origin then there exist disjoint open sets A. 1] and that Lp [0. and let Lp [0. the initial topology generated by M.e. (α. Show that if N ¯ is a nonempty subspace of X and x0 ∈ X \ N then there exists a continuous linear ′ functional φ ∈ X such that φN = 0 and φ(x0 ) = 1.6. Prove that M is linearly homeomorphic to Fn .] Exercise 5. Show that a balanced set is connected and give an example to show that a balanced set need not be convex. 1] := f : [0. [g]) := ∆(f − g) is a metric on Lp [0. then there exists an open set U ⊆ X such that 0 ∈ U and A + U ⊆ B. x) → αx.48 Dual Spaces Exercise 5..4. where n is the dimension of M. Deduce or prove otherwise that if A ⊆ B ⊆ X. Let X be a topological vector space over F and let M be a ﬁnitedimensional subspace of X. where N := f : [0.. [Use the separation theorem. Prove further that Lp [0. Let X be a locally convex topological vector space. Prove that a topological vector space with topology given by a separating family of linear functionals is locally convex.8. Prove that every open set containing the origin contains a nonempty open set which is balanced : a set B is balanced if λb ∈ B for all b ∈ B and λ ∈ F1 . . (x. 1]/N. Show that if this topology is such that singleton sets are closed (i. where A is compact and B is open.e.
15. where M is the collection of continuous linear functionals on X. Exercise 5.12. Let H be a Hilbert space. M). 1).] Deduce that Theorem 3. convex A. Suppose that X is a topological vector space such that the continuous elements of X ′ separate points. the collection of closed faces of C ordered by reverse inclusion. Exercises 5 Exercise 5. partially ordered set such that every chain in F has an upper bound. Prove that every element of C can be written as a convex combination of at most (n + 1) elements of ∂e C.13. [Note that 1 is an extreme point of F1 . nonempty. Exercise 5. B ⊆ X there exists a continuous φ ∈ X ′ such that sup Re φ(x) < inf Re φ(x). [Use induction.14. x∈A x∈B 49 [Consider X equipped with the topology σ(X. Prove that (F. is a nonempty. then tx + (1 − t)y ∈ C ◦ for all t ∈ [0. Exercise 5. Prove also that the interior C ◦ and the extremal boundary ∂e C are disjoint (as long as X = {0}). Let X be a topological vector space and suppose that C is a nonempty.17. the interior of C. convex subset of X. Prove that if x ∈ C and y ∈ C ◦ . Prove that every unit vector in H is an extreme point of the closed unit ball H1 . Let C ⊆ Rn be compact and convex.13. Prove that given disjoint.] Deduce that every isometry in B(H) is an extreme point of the closed unit ball B(H)1 . Exercise 5. compact. Let C be a convex subset of a topological vector space X.3.40 is true for topological spaces with continuous dual spaces that separate points. Prove that the closed unit ball of c0 has no extreme points. Exercise 5. ⊇). compact.] .16.
.
Algebras 51 .
.
g ∈ C(X) and α ∈ C then (f + αg)(x) := f (x) + αg(x) and (f g)(x) := f (x)g(x) This algebra has three important subalgebras: ∀ x ∈ X. denoted by C(X). A subalgebra of the algebra A is a subspace B that is closed under multiplication: ab ∈ A for all a. An algebra (more correctly.k=1 . or n = 1 + dim A otherwise). when equipped with the usual multiplication: ai j n i. b) → ab that is associative: (ab)c = a(bc) An algebra is commutative if its multiplication is.7 Example 4. i. the collection of n × n matrices over a ﬁeld F forms a unital algebra. (If the underlying scalar ﬁeld F of an algebra needs to be mentioned explicitly we refer to an algebra over F. = n i j n j=1 aj bk i. All the algebras that we consider will have scalar ﬁeld R or C. 53 .. b ∈ A.2. b ∈ B (more brieﬂy.k=1 ∀ a. denoted by Mn (F).Four Normed Algebras We start with some purely algebraic deﬁnitions.e. continuous functions on X is an algebra over C. An algebra is unital if there exists 1 ∈ A such that 1a = a1 = a for all a ∈ A. Deﬁnition 4. Example 4. B 2 ⊆ B). b. with the algebraic operations deﬁned pointwise: if f . c ∈ A. ab = ba for all a.1.j=1 j bk n j. A ﬁeld is a unital algebra over itself. an associative algebra) is a vector space A equipped with a bilinear map (called multiplication) · : A × A → A.) More generally.3. such an element is unique if it exists. If X is a topological space then the set of complexvalued. The examples above are ﬁnitedimensional (the dimension of an algebra is its dimension as a vector space) and every ﬁnitedimensional algebra A over F is isomorphic to a subalgebra of Mn (F) (with n = dim A if A is unital. (a. where homomorphism and isomorphism are deﬁned in the usual manner: see Deﬁnition 4.
b ∈ A. a normed algebra that is also a Banach space (with respect to its norm). A Banach algebra is a complete normed algebra.e. An algebra isomorphism is a bijective algebra homomorphism (which implies that φ−1 is an algebra homomorphism from B to A). We have C00 (X) ⊆ C0 (X) ⊆ Cb (X) ⊆ C(X).54 Normed Algebras (i) C00 (X). i. Let X be a locally compact. The kernel of an algebra homomorphism φ : A → B is ker φ := {a ∈ A : φ(a) = 0} and the image is im φ := {φ(a) : a ∈ A}. (ii) C0 (X). C0 (X) is a closed subalgebra of Cb (X) (so a Banach algebra in its own right) and C00 (X) is a dense subalgebra of C0 (X). Let A and B be algebras over the ﬁeld F.6. Example 4. Deﬁnition 4. [This follows from Proposition 2.5. A normed algebra is a normed space that is also an associative algebra. A normed algebra is unital if it is a unital algebra and 1 = 1. Example 4.. where · is the operator norm.] Quotient Algebras Deﬁnition 4. .4. Note that the submultiplicativity of the norm means that multiplication in normed algebras is jointly continuous: if an → a and bn → b then (an )n 1 is bounded and an bn − ab = an (bn − b) + (an − a)b as n → ∞. (iii) Cb (X). Note that Cb (X) and C(X) are unital. this example generalises Example 4. When equipped with the supremum norm · ∞: an sup an bn − b + an − a b an − a → 0 bn − b + b f→ f ∞ := sup{f (x) : x ∈ X}. An algebra homomorphism is an Flinear map φ : A → B such that φ(ab) = φ(a)φ(b) for all a.7. such that the norm is submultiplicative: ab a b for all a. the continuous functions on X that vanish at inﬁnity. as if X is ﬁnitedimensional then B(X) is isomorphic to Mn (F) (where n = dim X). but C00 (X) and C0 (X) are unital only if X is compact.2. b ∈ A. the bounded. the continuous functions on X with compact support. Hausdorﬀ space. · is a unital Banach algebra. If X is a Banach space then B(X).18. continuous functions on X. with equality if and only if X is compact. Cb (X) is a unital Banach algebra.
Proof This is trivial. b ∈ A. if A is unital then [1] A/I A/I [1] so [1] A/I 1 (since I is proper.9.11. α. Proof To see that the quotient norm is submultiplicative. If I is an ideal of A then the quotient space A/I is an algebra. Theorem 4. We have / also that [1] A/I 1 A = 1 and therefore [1] A/I = 1. when none exists but it would be useful to act as though one did.e. proper ideal then A/I is a normed algebra when equipped with the quotient norm. . Then ∼ ker φ is an ideal of A. Everything else follows from Theorem 1. if A is a Banach algebra then so is A/I. If A is an algebra over the ﬁeld F then Au is the unitization of A. im φ is a subalgebra of B and A/ ker φ = im φ. If A is a normed algebra and I is a closed. called the quotient algebra of A by I.4. note that if a. b ∈ A then [a] A/I 55 ∀ a. α)(b. Let A be an algebra. however. deﬁned by setting Au = A ⊕ F and (a. when equipped with the multiplication (a + I)(b + I) = ab + I and A/I is unital if A is unital. Deﬁnition 4. 1 ∈ I and so [1] = [0].8. [a] := a + ker φ → φ(a). b ∈ A. y ∈ I inf = [1]2 ab − x : x ∈ I = [ab] 2 A/I A/I . In particular. β) = (ab + αb + βa. whence [1] > 0).10. Deﬁnition 4.10. Unitization In many cases an algebra has a unit. ba ∈ I for all a ∈ A and b ∈ I (i. AI. in order to state it we need to deﬁne the concept of a quotient algebra.. there are some situation. Unitization The obvious fundamental theorem about homomorphisms is true. unital if A is unital. αβ) ∀ a. Proposition 4. [b] A/I = inf inf a − x : x ∈ I inf b − y : y ∈ I ab − ay − xb + xy : x. Let A be an algebra and φ : A → B an algebra homomorphism.2. IA ⊆ I). β ∈ F. via ˜ φ : A → im φ. An ideal of the algebra A is a subspace I such that ab.
Let L1 (R) denote the space of (equivalence classes of) complexvalued.56 Normed Algebras The algebra A is an ideal of Au .13 and Theorem 33.10]. Lebesgueintegrable functions on the real line. The sequence (en )n 1 is an approximate identity for L1 (R).7 and Theorem 9. (a. it is easy u to see that L1 (R) is isomorphic to the algebra given by adjoining the Dirac measure δ0 : by deﬁnition (f ⋆ δ0 )(t) = “ R f (t − s)δ0 (s) ds ” = f (t) ∀t ∈ R and δ0 ⋆ δ0 = δ0 . f → R f .14. n 1 2 2 en (x) = √ e− 2 n x = ne1 (nx) 2π λ − 1 λx2 e 2 ). Let A be a Banach algebra. .12. An approximate identity for A is a net (eλ )λ∈Λ ⊆ A1 such that lim eλ a = a = a lim eλ λ λ ∀ a ∈ A. Example 4. Section 9. 2π For a proof of this. where hn (x) = 1 n = nh1 (nx) π 1 + n2 x2 ∀x ∈ R : see [18. Approximate Identities Deﬁnition 4. α) → a A + α.14] (but note that kλ (x) should equal Another possibility is (hn )n 1 . The unitization Au is complete if and only if A is complete. If A is a normed algebra over R or C then Au is a normed algebra. Section 33. R (By the theorems of Fubini and Tonelli. This is a commutative Banach algebra when equipped with the convolution product: f ⋆ g : R → R. where ∀ x ∈ R.13. see [17. where · Au : Au → R+ . with norm · 1: L1 (R) → R+ . Example 4. The unitization Au is commutative if and only if A is commutative. and A is a closed ideal of Au . if f and g are integrable then this integral exists almost everywhere and deﬁnes an element of L1 (R). t → f (t − s)g(s) ds.) This algebra lacks a unit.
of course. b ∈ A if (an )n 1 . so that n=−N an e N (Pr ⋆f )(t)−f (t) (Pr ⋆f )(t)−p(t)+p(t)−f (t) 1−r n n=−N f ∞+ p−f ∞ <ε if r is near enough to 1. Section 10. Hence Pr ⋆ f → f uniformly in this case. Proof ¯ Let (A. Furthermore. j) and (C.14).4. ¯ Theorem 4. 1) is directed in the usual manner and Pr (t) = 1 2π ∞ 57 r n eint = n=−∞ 1 1 − r2 1 1 + reit = Re . note that if f : t → eikt (where k ∈ Z) then π (Pr ⋆ f )(t) = −π 1 r n eint r n ein(t−s) eiks ds = 2π n=−∞ 2π n=−∞ ∞ ∞ π ei(k−n)s ds = r k f (t). ¯ Deﬁne a product on A by setting ab := lim i(an bn ) n→∞ ¯ ∀ a. Fej´r’s theorem implies that if f is continuous and ε > 0 then there exists a e int trigonometric polynomial p : t → N such that f − p ∞ < ε/2.] Completion We have the notion of completion as for a normed space. i) be the completion of A considered as a normed space (see Theorem 1. π]. −π so (Pr ⋆ f ) − f 1 = (1 − r k ) f 1 → 0 as r → 1−.] To prove that this is an approximate identity. where [0. π] (an easy consequence of Fej´r’s e theorem [16. (A.4]) so a simple ε/2 argument gives this result for general f . The linear span of {eikt : k ∈ Z} (the trigonometric polynomials) is dense in L1 [−π.4.16. which we can identify with the closed subalgebra of L1 (R) consisting of 2πperiodic functions. 2π 1 − 2r cos t + r 2 2π 1 − reit [The function P is the Poisson kernel .1) . [Compare this proof to that given in [16.36(2)]. with ¯ an isometric homomorphism i : A → A ¯ ¯ A unital if A is. Let A = L1 [−π. an isometric isomorphism is now required to be multiplicative. k) are completions of A then there exists an isometric isomorphism l : B → C such that l ◦ j = k. i) is called a completion of A. it occurs in the theory of harmonic functions on the unit disc. This has approximate identity (Pr )r∈[0. In fact. If A is a normed algebra then there exists a Banach algebra A and ¯ such that i(A) is a dense subalgebra of A. Theorem 30. Completion Example 4. is independent of the choice of sequences (an )∞ and n=1 n=1 .15. this limit n=1 exists (as (i(an bn ))∞ is Cauchy). if (B. (bn )∞ ⊆ A are such that i(an ) → a and i(bn ) → b as n → ∞.
58
Normed Algebras (bn )∞ and the new product agrees with the old on i(A), i.e., i(a)i(b) = i(ab) for all n=1 a, b ∈ A. (To see the ﬁrst claim, note that i(an bn ) − i(am bm ) = = = →0 i(an bn − am bm ) an bn − am bm an bn − bm + an − am bm i(an ) i(bn ) − i(bm ) + i(an ) − i(am )
i(bm )
as m, n → ∞; proof of uniqueness is similar.) It is easy to verify that this product makes ¯ ¯ A an associative algebra (which is unital if A is) and the norm on A is submultiplicative because i(an )i(bn ) = i(an bn ) = an bn an bn = i(an ) i(bn ) .
¯ Hence A is a Banach algebra with dense subalgebra i(A). If (B, j) and (C, k) are completions of A then there exists an isometric linear bijection l : B → C such that l ◦ j = k, by Proposition 2.6. Furthermore, if a, b ∈ A then l j(a)j(b) = l j(ab) = k(ab) = k(a)k(b) = l j(a) l j(b) , which shows that l is multiplicative on j(A). Since j(A) is dense in B and multiplication is jointly continuous, l is multiplicative on B and so is an isometric isomorphism.
Five
Invertibility
From now on, A denotes a unital Banach algebra over C. Deﬁnition 5.1. An element a ∈ A is invertible if there exists b ∈ A such that ab = ba = 1. [If such an inverse exists, it is unique.] The collection of invertible elements in A is denoted G(A); this is a group. [This last claim is obvious once we recall that (ab−1 )−1 = ba−1 if a and b are invertible.] Proposition 5.2. Let a ∈ A be such that a < 1. Then 1 − a ∈ G(A) and
∞
(1 − a)
−1
=
n=0
an ,
(5.1)
where this series converges in the norm topology. Hence G(A) is an open subset of A; furthermore a → a−1 is a homeomorphism of G(A). Proof Since a < 1, the Neumann series (5.1) is absolutely convergent, so convergent. Furthermore,
n n→∞ n
lim
j=0
aj (1 − a) = (1 − a) lim
n→∞
j=0
aj = lim 1 − an+1 = 1,
n→∞
so the sum of this series is an inverse for (1 − a), as claimed. If a ∈ G(A) let b ∈ A and note that
if a−1 (a − b) < 1, e.g., if b ∈ B A −1 −1 (a); this shows that G(A) is open. a 1 Finally, if h ∈ A is such that h < 1 a−1 −1 then a−1 h < 2 , which implies that 2 a + h = a(1 + a−1 h) ∈ G(A) and
∞
b = a − (a − b) = a 1 − a−1 (a − b) ∈ G(A)
(a + h) has norm at most
−1
−a
−1
=a
−1
(1 + a h)
−1
−1
−1 = a
−1 n=1 2
(−a−1 h)n
a−1 (using the fact that h → 0, as required.
∞ n=1
x
n
a−1h /(1 − a−1 h )
2 a−1
h
x /(1 − x ) if x < 1). Hence (a + h)−1 → a−1 as
59
60
Invertibility
The Spectrum and Resolvent
Deﬁnition 5.3. For all a ∈ A the spectrum of a is σ(a) := {λ ∈ C : λ1 − a ∈ G(A)}. / The resolvent set of a is ρ(a) := C \ σ(a), the complement of the spectrum of a. The resolvent of a is the function deﬁned on ρ(a) by λ → rλ (a) := (λ1 − a)−1 . Theorem 5.4. The resolvent satisﬁes the resolvent equation, rλ (a) − rµ (a) = −(λ − µ)rλ (a)rµ (a) ∀ λ, µ ∈ ρ(a),
and is strongly holomorphic on ρ(a), which is an open set. The spectrum is a compact, nonempty subset of C a . Proof For the ﬁrst claim, note that if λ, µ ∈ ρ(a) then (λ1 − a)−1 − (µ1 − a)−1 = (λ1 − a)−1 (1 − (λ1 − a)(µ1 − a)−1 ) = (λ1 − a)−1 (µ1 − a) − (λ1 − a) (µ1 − a)−1 = −(λ − µ)(λ1 − a)−1 (µ1 − a)−1 .
From this equation and Proposition 5.2 it follows that rλ (a) − rµ (a) = −rλ (a)rµ (a) → −rµ (a)2 λ−µ as λ → µ: the resolvent is strongly holomorphic on ρ(a) (recall Deﬁnition 3.13). To see that this set is open, let f : C → A; λ → λ1 − a and note that f is continuous, so f −1 (G(A)) = ρ(a) is open. If a ∈ A and λ > a then λ−1 a < 1, so by Proposition 5.2 we have that λ1 − a = λ(1 − λ−1 a) is invertible. Hence σ(a) ⊆ {λ ∈ C : λ a }. By the HeineBorel theorem and the fact that σ(a) = C \ ρ(a) is closed, σ(a) is compact. Suppose for contradiction that σ(a) is empty, so that the resolvent is a strongly holomorphic function deﬁned on the whole of C. If λ > a then
∞
rλ (a) = λ−1 (1 − λ−1 a)−1 = and so rλ (a) λ−1/ 1 − λ−1 a
λ−(n+1) an
n=0
(5.2)
= λ − a
−1
;
this shows that rλ (a) → 0 as λ → ∞. In particular the resolvent is bounded, so constant (by Theorem 3.14) and therefore equal to 0; this is the desired contradiction.
11]). Theorem 5. Proof Firstly. (GelfandMazur) If A is a unital Banach algebra over C in which every nonzero element is invertible then A is isometrically isomorphic to C.2.8. note that a = λa 1 = λa . We recall the spectral mapping theorem for polynomials. (5. The map a → λa is the desired isomorphism. Let a ∈ A and suppose that p(z) ∈ C[z] is a complex polynomial. Beurling led Sweden’s eﬀort to crack the Enigma code during the second world war. it is stated in b4 for bounded operators on a Banach space but its proof holds in any Banach algebra Theorem 5. The SpectralRadius Formula Deﬁnition 5.. The following theorem gives the BeurlingGelfand spectralradius formula.) Theorem 5. Theorem 5. Then σ p(a) = p σ(a) .6. Note that commutativity forms no part of the hypotheses but is part of the conclusion. The spectral radius of a ∈ A is the radius of the smallest disc about the origin that contains the spectrum of a: ν(a) := inf{r 0 : σ(a) ⊆ Cr } = sup{λ : λ ∈ σ(a)}. Proof Let a ∈ A.3) . / / Proof This follows the same pattern as the proof which may be found in Dr VincentSmith’s b4 notes ([25. (According to [11.e. since σ(a) is nonempty and λ ∈ σ(a) ⇔ λ1 − a ∈ G(A) ⇔ λ1 − a = 0 ⇔ a = λ1.525]. p. The GelfandMazur Theorem 61 The GelfandMazur Theorem It is rather surprising that a theorem with as short a proof as the following has such important consequences. If a ∈ A then the sequence ν(a) = inf an n 1 1/n an 1/n n 1 is convergent and 1/n = lim an n→∞ .5.5. the spectral mapping theorem for polynomials implies that ν(a)n = ν(an ) an ∀n 1.7. σ p(a) := {λ ∈ C : λ1 − p(a) ∈ G(A)} = {p(λ) : λ1 − a ∈ G(A)}. i. / we see that for all a ∈ A there exists λa ∈ C such that a = λa 1.2.
let b = (λ1 − a)−1 and let c ∈ B. so Mr := sup (λ1−a)−1 : λ = r < ∞ for all r > ν(a). by the uniqueness of Laurent series.10) for all r > ν(a). if S is commutative then the double commutant (or bicommutant) S cc = (S c )c is a commutative subalgebra of A containing S. so we need to prove that if λ1 − a is invertible in A then it is invertible in B. so for all λ ∈ S. It is an exercise to prove that a commutative subalgebra B is maximal if and only if B = B c . If a ∈ B then / / σB (a) := {λ ∈ C : λ1 − a ∈ G(B)} = σA (a) = {λ ∈ C : λ1 − a ∈ G(A)} .10. It is readily veriﬁed that S c is a unital. Let B be a maximal commutative subalgebra of A. (If α an 1/n → ν(a) is immediate. that if S ⊆ T ⊆ A then T c ⊆ S c and that S is commutative if and only if S ⊆ S c . note that bc = bc(λ1 − a)b = b(λ1 − a)cb = cb and so b ∈ B (else B is not maximal) as required. . that The following theorem allows us to restrict our attention to commutative algebras when considered certain questions about the spectrum. This and (5. by the integral formula for Laurent coeﬃcients.) an for all n and an → α then α = inf an .62 Invertibility Next. Proof Clearly G(B) ⊆ G(A). Suppose that λ1 − a ∈ G(A). Furthermore.9 goes through with the weaker hypothesis that B is a commutative subalgebra of A containing {a}cc . so φ ◦ f is holomorphic. If S is a subset of the unital Banach algebra A then S c := {a ∈ A : sa = as for all s ∈ S} is the commutant (or centraliser ) of S. Theorem 5.3) yield ν(a) an 1/n ν(rMr )1/n → r as n → ∞. with ∞ (φ ◦ f )(λ) = n=0 φ(an )λ−(n+1) λ > a by (5.9. closed subalgebra of A. the result follows. f : λ → (λ1−a)−1 is continuous on ρ(a). Let φ ∈ A∗ and note that f is strongly holomorphic. and that the proof of Theorem 5. Furthermore. φ(an ) = and so an 1 2πi (φ ◦ f )(λ)λn dλ {λ:λ=r} r n+1 φ Mr r n+1 Mr (by Theorem 3. on S = {λ ∈ C : λ > ν(a)}. Zorn’s lemma allows us to prove that any commutative set in A is contained in a maximal commutative subalgebra.2). Example 5.
Let A be a unital Banach algebra over C and let a. . where ν(a) denotes the spectral radius of a. Prove that xp yq = p∈P q∈Q (p. Prove that A is commutative. Deduce that ea is invertible.3. [Let a.. the onepoint compactiﬁcation of X.4. is topo˙ ˙ logically isomorphic to C(X).] Prove further that this algebra is not unital.] ∞ Exercise 6. Show that (A.q)∈P ×Q xp yq = q∈Q p∈P xp yq . Suppose that there exists K > 0 such that a Kν(a) for all a ∈ A. consider invertibility and the polynomials pn (z) = 1 + z/n. Let A be a Banach algebra and suppose that (xp )p∈P . [To show associativity.5. Prove further that f : λ → eλa is holomorphic everywhere. Let A be a unital Banach algebra over C. use the convolution theorem and the inversion theorem for the Fourier transform.5. with f ′ (λ) = af (λ) = f (λ)a. Prove that ea+b = ea eb if a and b commute.6.7. Use the identity (ab)n = a(ba)n−1 b to prove that ab and ba have the same spectral radius. for all a ∈ A there exists n ∈ N such that an = 0). b ∈ A. the unitization of the algebra of continuous functions on X that vanish at inﬁnity. Prove that C0 (X)u . Let X be a Hausdorﬀ. Exercises 6 63 Exercises 6 Exercise 6. [Consider the decomposition A = n∈N {a ∈ A : an = 0}. b ∈ A and consider the function g : λ → eλa be−λa . Let A be a unital Banach algebra over C and let ea := n=0 an /n! for all a ∈ A. (yq )q∈Q ⊆ A are absolutely summable. locally compact space. Exercise 6. [For the latter statement. Let A = C[z] denote the unital algebra of complex polynomials and let p := sup{p(α) : α 1} for all p ∈ A. Exercise 6.] Exercise 6. Prove that the vector space L1 (R) is a commutative Banach algebra when equipped with the convolution product and the norm f 1 := f . normed algebra which is not complete.] Exercise 6.e.8.2.] Exercise 6. Prove that A is uniformly nilpotent: there exists N ∈ N such that aN = 0 for all a ∈ A. the algebra of continuous functions on X. Exercise 6. Let A be a (nonunital) Banach algebra such that every element is nilpotent (i. [You may assume that the √ functions fn : x → (n/ 2π) exp(− 1 n2 x2 ) are such that fn ⋆ g − g 1 → 0 as n → ∞ for 2 all g ∈ L1 (R).1. for all a ∈ A.4. · ) is a unital.
.
and apply Zorn’s lemma (note that no proper ideal contains an invertible element of A. preordered by inclusion. if the former then I ∩ G(A) = ∅. so that b < 1 and 1 − b ∈ G(A). but I ⊆ A \ G(A) ¯ ⊆ A \ G(A) since G(A) is open. To see the statement about the norm of φ. Furthermore. as claimed. I is not proper.3. the desired contradiction. it cannot be the case that φ(a) = 0 for all a ∈ A (as φ = 0) and so φ(1) = 1. The collection of all characters of A is denoted by Φ(A). Every proper ideal of A contains no invertible element of A and is contained in a maximal ideal. Let A be a unital algebra. Let b = φ(a)−1 a. since the closure of an ideal is an ideal (an easy ¯ ¯ ¯ exercise). so neither does the union of a chain of such). Proposition 6. Proof If I is an ideal of A such that I ∩ G(A) = ∅ then 1 = a−1 a ∈ I for some a ∈ G(A) and so I = A. Hence I = I. A maximal ideal of A is an ideal I that is proper (I = A) and maximal with respect to inclusion: if J is an ideal such that J ⊇ I then either J = A or J = I. Proof Note that φ(a) = φ(1a) = φ(1)φ(a) for all a ∈ A. If φ : A → C is a nonzero algebra homomorphism then φ(1) = 1 and φ(a) = 0 for all a ∈ G(A). either I = A or I = I. Deﬁnition 6. if A is a Banach algebra then φ ∈ A∗ with φ = 1. A maximal ideal of A is closed. let I be a maximal ideal. whence φ(a) = 0 if a is invertible. Finally. let I be a proper ideal of A.2.1.Six Characters and Maximal Ideals Proposition 6.. i. unless otherwise speciﬁed. ¯ implies that I 65 . note that φ(1) = 1 (so φ 1) and suppose for contradiction that there exists a ∈ A such that a 1 and φ(a) > 1.e. A is a commutative unital Banach algebra over C. Thus 1 = φ(1) = φ(aa−1 ) = φ(a)φ(a−1 ) for all a ∈ G(A). To see the next claim. From now on. A character of A is a nonzero algebra homomorphism from A to C. let F denote the collection of proper ideals of A that contain I. but φ(1 − b) = 1 − φ(a)−1 φ(a) = 0.
whence [a][b] = [1] (where [a] = a + I et cetera) and A/I is a ﬁeld. The map φ → ker φ is a bijection between Φ(A) and the set of all maximal ideals of A. If I is not maximal then there exists a ∈ A \ I such that aA + I = A. Then aA + I is an ideal in A that properly contains I. Proof Suppose that I is maximal and let a ∈ A \ I. For all a ∈ A we have that φ(a − φ(a)1) = 0. Hence [a] is not invertible and A/I is not a ﬁeld. (i) a ∈ G(A) if and only if φ(a) = 0 for all φ ∈ Φ(A). by the GelfandMazur theorem. unital Banach algebra over C. Suppose that φ. (iii) ν(a) = sup{φ(a) : φ ∈ Φ(A)}. so A/I ∼ C. Finally. Corollary 6. Then (ii) σ(a) = {φ(a) : φ ∈ Φ(A)}. . Then φ = i ◦ π is the desired character. we need a fact from algebra. Let a ∈ A.4. where π : A → A/I is the quotient map. by Lemma 6. Proof Since A/ ker φ ∼ im φ = C. A/ ker φ is a ﬁeld for all φ ∈ Φ(A) and hence ker φ is = maximal. Proof For (i). unless otherwise stated.4) and a Banach algebra (with respect to its quotient norm). so 1 ∈ aA + I / and there exists no b ∈ A such that [a][b] = [1].66 Characters and Maximal Ideals Before we can prove the fundamental connexion between characters and maximal ideals. Lemma 6. if I is a maximal ideal of A then A/I is a ﬁeld (by Lemma 6. note that a ∈ G(A) ⇔ ⇔ ⇔ ⇔ A = Aa Aa is not contained in a maximal ideal of A Aa ⊆ ker φ ∀ φ ∈ Φ(A) φ(a) = 0 ∀ φ ∈ Φ(A). Theorem 6. Hence φ = ψ and φ → ker φ is injective. Hence there exists b ∈ A and c ∈ I such that ab + c = 1. so a − φ(a)1 ∈ ker φ = ker ψ and 0 = ψ(a − φ(a)1) = ψ(a) − φ(a). so aA + I = A.4. ψ ∈ Φ(A) are such that ker φ = ker ψ. Characters and the Spectrum Recall that A is a commutative. Then A/I is a ﬁeld if and only if I is maximal. = let i : A/I → C denote this isomorphism.6. Let A be a unital commutative algebra (or even a commutative ring with identity) and suppose that I is an ideal of A.5.
An algebra is semisimple if J(A) = {0}.e.b∈A . The Jacobson radical of A is an ideal. Hausdorﬀ space. by the spectral mapping theorem..10.22. and so ν(a + b) = sup{φ(a + b) : φ ∈ Φ(B)} sup{φ(a) : φ ∈ Φ(B)} + sup{φ(b) : φ ∈ Φ(B)} = ν(a) + ν(b). Proof Since we have that ν(a) = 0 if and only if a ∈ ker φ for all φ ∈ Φ(A). b ∈ A} 1 (0) a. ν(a) = sup{φ(a) : φ ∈ Φ(A)}. the intersection of all maximal ideals in A. φ(ab) = φ(a)φ(b) ∀ a. By Theorem 5. The Gelfand Topology Lemma 6. denoted by J(A). If a ∈ A is nilpotent then σ(a) = 0. ab = ba. i. Proof Let B be a maximal commutative subalgebra containing a and b. Recall that an element a of a ring is said to be nilpotent if an = 0 for some n ∈ N. and 1 = A∗ ∩ ˆ−1 (1) ∩ 1 1 ˆb ab − aˆ −1 Φ(A) = {φ ∈ A∗ : φ(1) = 1. by Theorem 3. σB (a + b) = σA (a + b) et cetera.6. then ν(a + b) ν(a) + ν(b) and ν(ab) ν(a)ν(b). we have the following deﬁnition. The other claim is proved in the same manner. Hausdorﬀ space when equipped with the Gelfand topology. Corollary 6. The character space Φ(A) is a compact. More generally. b ∈ A commute. If A is a (not necessarily commutative) unital Banach algebra and a.8. i. ˆ to Φ(A). in fact. An element a ∈ A is quasinilpotent if ν(a) = 0.e.7. Proposition 6. The set of all quasinilpotent elements in A is the Jacobson radical of A. Equivalently. The Gelfand Topology The other two claims are immediate.9.2. Deﬁnition 6. Proof Recall that A∗ is a compact.. it is the coarsest topology to make the maps aΦ(A) : φ → φ(a) continuous (for all a ∈ A).9. A). J(A) = φ∈Φ(A) 67 ker φ. the weak* topology on A∗ . the restriction of σ(A∗ .
Its kernel is J(A) and its image A is a subalgebra of C Φ(A) that separates the points of Φ(A).11. a ˆ ∞ ∀ φ ∈ Φ(A). i. Proof The Gelfand transform is a homomorphism because characters are. Chapter 4 et seq.]. φ → φ(a). ˆ Theorem 6.68 Characters and Maximal Ideals is a closed subset of A∗ (since a : A∗ → C. ˆ That is. φ → φ(x) . so A Gelfand theory reaches its peak when the algebra is equipped with an involution.. a is the restriction to Φ(A) of the map φ → φ(a) on A∗ . Chapter 11 et seq. ˆb ab(φ) = φ(ab) = φ(a)φ(b) = aˆ (φ) Furthermore. The Representation Theorem From now on it is more convenient to let a : Φ(A) → C. [Note ˆ 1 that the condition φ(1) = 1 is necessary to rule out the zero homomorphism. Examples Example 6.e. If X is a compact.12. Hausdorﬀ space and A = C(X) is the algebra of continuous functions on X then Φ(A) = {ǫx : x ∈ X}. If A is a commutative unital Banach algebra then the Gelfand transform ˆ: A → C Φ(A) . this calculation also shows that the kernel is as claimed. b ∈ A. where ǫx : C(X) → C. φ → φ(a) is continuous for all a ∈ A). ˆ ˆ ˆ separates the points of Φ(A). ˆ Finally. if φ.] This gives the result. so a → a is normdecreasing. a. a conjugatelinear map a → a∗ such that (ab)∗ = b∗ a∗ and (a∗ )∗ = a ∀ a.] or [19. see [8. ψ ∈ Φ(A) are such that a(φ) = a(ψ) for all a ∈ A then φ = ψ (by deﬁnition). for example. a → a · ˆ ˆ is a normdecreasing homomorphism. b ∈ A. = sup{φ(a) : φ ∈ Φ(A)} = ν(a) a ∀ a ∈ A. If the involution satisﬁes a∗ a = a 2 for all a ∈ A then we have a C ∗ algebra: for the theory of such.
0 [j] = [1] ∀ x ∈ ℓ1 (Z/nZ). ¯ {ǫz : z ∈ D}. s) ds : r → R f (r. Before investigating the Gelfand theory of L1 (R) we need a couple of preliminary results. . y ∈ ℓ1 (Z/nZ). [k]∈Z/nZ Let δ := [j] → and note that x= [j]∈Z/nZ 1 [j] = [1]. x[j] δ j In particular. . s) ds. [j]∈Z/nZ Hence the Gelfand theory of ℓ1 (Z/nZ) corresponds to the theory of the discrete Fourier transform. each λ ∈ {ω j : j = 0. . so the quantities above are . as this has not been established the lemma is proved directly. C Example 6. s) ∈ L1 (R) for almost every s ∈ R and that R f (·. The ﬁnitedimensional Banach space ℓ1 (Z/nZ) = (xj )j∈Z : xj = xn+j ∀ j ∈ Z = (x[j] )[j]∈Z/nZ becomes a commutative. any φ ∈ Φ ℓ1 (Z/nZ) is determined by λ = φ(δ) and λ is an nth root of unity as λn = φ(δ n ) = φ(1) = 1. Suppose that n ∈ N and let Z/nZ denote the quotient group of integers with addition modulo n. Example 6. (The fact that L1 (R)∗ = L∞ (R) gives a simple proof of the following. via x→ x[j] λj . s) ds = R R φ r → f (r. .6. n − 1}. where ω = exp(2πi/n). s) ds ∈ L1 (R). Examples is the evaluation homomorphism at x.13. unital Banach algebra when equipped with the convolution product (x ⋆ y)[j] = x[k] y[j−k] ∀ x. s) : r → f (r. In this case the character space of A(D) is again just the set of evaluation homomorphisms. 1. Proof Note ﬁrst that Fubini’s theorem gives that f (·. If f ∈ L1 (R2 ) and φ ∈ L1 (R)∗ then φ r→ f (r.14.) Lemma 6.15. corresponds to a character. If D = B1 (0) = {z ∈ C : z < 1} and 69 ¯ A(D) = {f ∈ C(D) : f D is holomorphic} is the disc algebra then A(D) is a Banach algebra (when equipped with the supremum norm: recall that uniform limits of holomorphic functions are holomorphic).4. Conversely.
B ⊆ R are bounded intervals) then s → φ f (·. as C00 (R) is dense in L1 (R) we may assume that f is continuous and has compact support. s) ds = φ(χA R χB (s) ds) = R φ(χA )χB (s) ds = R φ f (·. Suppose that the latter holds. s → φ f (·. s) → φ f (·. Since χ is bounded we must have that d is purely imaginary. If χ : R → C is a continuous. s) ds − f (r. let A = L1 (R)u denote its unitization. Note that R fn (·. as claimed. diﬀerentiating the equation χ(s + t) = χ(s)χ(t) with respect to s at 0 yields χ′ (t) = χ′ (0)χ(t) and so χ(t) = exp(dt).. so is measurable.15 gives that φ(g) = φ(f )φ(g) = φ(f ⋆ g) = φ s → ft (s)g(t) dt = R R φ(ft )g(t) dt. (This calculation also shows that s → φ fn (·. measurability of s → φ f (·.70 Characters and Maximal Ideals well deﬁned. χ(x + t) dx = t χ(y) dy (t ∈ R) and χ is diﬀerentiable. so φ R f (·. bounded function such that χ(0) = 1 and χ(s + t) = χ(s)χ(t) for all s.) Lemma 6.17. φ(α1 + f ) = α for all α ∈ C and f ∈ L1 (R) ) or there exists f ∈ L1 (R) such that φ(f ) = 1. s) ds − φ f (·.e. there exists r > 0 such that c = Hence r t+r cχ(t) = 0 r 0 χ(x) dx = 0. s) is the almosteverywhere limit of a sequence of measurable functions. Example 6.) If g ∈ L1 (R) then Lemma 6. In particular. s) will follow from the below. s) ds R R φ R (fn − f )(·. s) in L1 (R). s) ds = lim n→∞ φ fn (·. If f = χA×B (where A.16. (If g ∈ C00 (R) is such that f − g 1 1/2 then φ(g) φ(f ) − φ(f − g) 1 − φ f − g 1 1/2. Proof Since χ(0) = 1 and χ is continuous. s) ds → R f (·. . s) − f (r. so has there is a subsequence such that φ fnk (·. s) ds. now replace f by g/φ(g). s) 1 ds = φ fn − f 1. where d = χ′ (0). s) ds dr R R R R R fn (r. For general f ∈ L1 (R2 ) take (fn )n 1 ⊆ Lstep (R2 ) such that fn → f in L1 (R2 ). s) → s → φ f (·. The last equality holds as φ fn (·. linearity gives the result for all f ∈ Lstep (R2 ). Recall that L1 (R) is a nonunital Banach algebra when equipped with the convolution product. s) ds. s) = φ(χA )χB and φ R f (·. s) ds in L1 (R) as fn (r. If φ ∈ Φ(A) then either ker φ = L1 (R) (i. s) for almost every s ∈ R. s) ds = R R φ f (·. as claimed. t ∈ R then there exists α ∈ R such that χ(t) = eiαt for all t ∈ R. s) ds = lim φ n→∞ R fn (·. s) ds dr = fn − f 1.
6.4. Examples where ft (s) = f (s − t) for all s, t ∈ R. Deﬁne χ : R → C; t → φ(ft ) and note that, since ft 1 = f 1 for all t ∈ R, χ(t) bounded. Furthermore, χ(0) = φ(f0 ) = φ(f ) = 1 and χ(t + h) − χ(t) = φ(ft+h − ft ) ft+h − ft
1
71
φ
ft
1
= f 1, i.e., χ is
= fh − f
1
→0
as h → 0; this follows from the continuous form of the dominatedconvergence theorem. Note also that, if g, h ∈ L1 (R), gs+t ⋆ h (r) =
R
g r − p − (s + t) h(p) dp =
R
g(r − q − s)h(q − t) dq = (gs ⋆ ht )(r)
and so χ(s + t) = φ(fs+t ) = φ(fs+t)φ(f ) = φ(fs+t ⋆ f ) = φ(fs ⋆ ft ) = φ(fs )φ(ft ) = χ(s)χ(t) for all s, t ∈ R. By Lemma 6.16 we must have that χ(t) = e−ist for all t ∈ R, where s ∈ R is such that e−is = χ(1). To see that s ∈ R is independent of the choice of f , let g ∈ L1 (R) be such that φ(g) = 1 and note that χ(t) = φ(ft )φ(g) = φ(ft ⋆ g) = φ(f ⋆ gt ) = φ(f )φ(gt ) = φ(gt ) Hence φ(g) =
R
∀ t ∈ R.
g(t)e−ist dt
∀ g ∈ L1 (R)
and the Gelfand transform corresponds the classical Fourier transform for L1 (R); the fact that α1 + f → α + R f (t)e−ist dt is a character for all s ∈ R is an immediate consequence of Fubini’s theorem. Example 6.18. If L1 (R+ ) is equipped with the convolution product
t
(f ⋆ g)(t) =
0
f (t − s)g(s) ds
∀ f, g ∈ L1 (R+ )
then it becomes a Banach algebra. The Gelfand theory here corresponds to the Laplace transform. ¯ Example 6.19. Let T = ∂D = {z ∈ C : z = 1} and A = {f ∈ C(D) : f T ∈ A(D)T }: ¯ that agree on the unit A consists of those continuous functions on the closed unit disc D ¯ circle T with a continuous function on D that is holomorphic in D. The character space 2 := of A is homeomorphic to the sphere S {(x, y, z) ∈ R3 : x2 + y 2 + z 2 = 1}.
72
Characters and Maximal Ideals
Exercises 7
Exercise 7.1. Let A = C(X), where X is a compact, Hausdorﬀ space. Prove that the map ǫ : X → Φ(A); x → ǫx is a homeomorphism, where ǫx (f ) = f (x) for all x ∈ X and f ∈ C(X).
Exercise 7.2. Prove that if A is a unital Banach algebra generated by a single element (i.e., there exists a ∈ A such that {p(a) : p(z) ∈ C[z]} is dense in A) then Φ(A) is homeomorphic to σ(a). [Consider φ → φ(a).] Deduce that Φ(A) is homeomorphic to ¯ D := {z ∈ C : z 1} if A = A(D) is the disc algebra.
Exercise 7.3. Let A be a unital Banach algebra that is generated by one element, a, and let λ ∈ σ(a). Show there exists a sequence of polynomials (pn )n 1 such that pn (z) → (λ − z)−1 uniformly for all z ∈ σ(a). [Hint: (λ1 − a)−1 ∈ A.] Deduce that the complement of σ(a) is connected. [Prove that if C is a bounded, maximally connected component of C \ σ(a) then C is open and then employ the maximummodulus theorem [16, Theorem 5.20].] Exercise 7.4. Let A be a commutative, unital Banach algebra. Prove that the Gelfand transform on A is isometric if and only if a2 = a 2 for all a ∈ A.
Exercise 7.5. Let A be a Banach algebra and B a semisimple, commutative, unital Banach algebra. Prove that if φ : A → B is a homomorphism then φ is continuous. [Use the closedgraph theorem.] Exercise 7.6. Let A = C 1 [0, 1], equipped with the norm f := f ∞ + f ′ ∞ . Prove that A is a semisimple, commutative, unital Banach algebra and ﬁnd its character space. Prove that I = {f ∈ A : f (0) = f ′ (0) = 0} is a closed ideal in A such that A/I is a twodimensional algebra with onedimensional radical. What do you notice about A and A/I? Exercise 7.7. Prove that the Banach space ℓ1 (Z) is a commutative, unital Banach algebra when equipped with the multiplication a ⋆ b : Z → C; n → [You may assume that
m∈Z
am bn−m
m∈Z
a, b ∈ ℓ1 (Z) .
m∈Z
n∈Z am bn
=
n∈Z
Exercise 7.8. Let δ ∈ ℓ1 (Z) be such that δ1 = 1 and δn = 0 if n = 1. Prove that a = n 1 1 n∈Z an δ for all a ∈ ℓ (Z). Deduce that the character space of ℓ (Z) is homeomorphic to T := {z ∈ C : z = 1} and (with this identiﬁcation) the Gelfand transform on ℓ1 (Z) is the map Γ : ℓ1 (Z) → C(T); Γ(a)(λ) = an λn
n∈Z
am bn for all a, b ∈ ℓ1 (Z).]
∀ λ ∈ T, a ∈ ℓ1 (Z).
n∈Z
Exercise 7.9. Let f : T → C be continuous and such that 1 ˆ f (n) := 2π
π
ˆ f (n) < ∞, where
f (eit )e−int dt
−π
(n ∈ Z).
6.5. Exercises 7 Prove that if f (z) = 0 for all z ∈ T then g = 1/f : T → C; z → 1/f (z) satisﬁes g n∈Z ˆ(n) < ∞. [This result is known as Wiener’s lemma.] ¯ Exercise 7.10. Let T = ∂D = {z ∈ C : z = 1} and A = {f ∈ C(D) : f T ∈ A(D)T }: ¯ that agree on the unit A consists of those continuous functions on the closed unit disc D ¯ circle T with a continuous function on D that is holomorphic in D. [A corollary of the maximummodulus theorem [16, Theorem 5.20] will be useful: if f ∈ A(D) then f ∞ := sup{f (z) : z 1} = sup{f (z) : z = 1} =: f T ∞ .] (i) Show that A is a Banach algebra when equipped with the supremum norm. (ii) Prove that I = {f ∈ A : f T = 0} is a closed ideal in A and that A = A(D) ⊕ I. [Consider a suitable map j : A → A(D).] i : I → C0 (D); f → f D
73
(iii) Prove that
˙ is an isometric isomorphism. Deduce that I u is topologically isomorphic to C(D), ˙ is the onepoint compactiﬁcation of D. where D ˜ ˜ (iv) Prove that if φ ∈ Φ(A) is such that ker φ ⊇ I then φ = φ ◦ j, where φ ∈ Φ(A(D)). ¯ Deduce that φ = εz ◦ j for some z ∈ D (where εz : A(D) → C; f → f (z) ). ¯ ¯ ¯ ¯ (v) Let D1 and D2 be two copies of the unit disc and let S 2 = D1 ∪ D2 ∼ be the sphere obtained by identifying each point on T1 = ∂D1 with the corresponding point on T2 = ∂D2 . Deﬁne T : S 2 → Φ(A); z → ¯ εz ◦ j (z ∈ D1 ), ¯ εz (z ∈ D2 )
and prove that this is a welldeﬁned, continuous injection. (vi) Prove that Φ(A) is homeomorphic to the sphere S 2 .
.
ordered by setting (c. Let (xa )a∈A be a net in X and let F be a nonempty family of subsets of X that forms a directed set under reverse inclusion. together with construction of a diagonal subsequence (of a sequence of sequences!).) A net (xa )a∈A is frequently in S if for all a0 ∈ A there exists a ∈ A such that a a0 and xa ∈ A. (Note that a net is not frequently in a set if it is eventually in its complement et cetera. (A net in a topological space converges to a point if it is eventually in every open set containing that point. Proof Let B = {(c. Lemma A. There is a subnet of (xa )a∈A that is eventually in every S ∈ F. a proof of Tychonov’s theorem for countable products of metric spaces can be given using sequences. Deﬁnition A. Let (A. S) ≪ (d. T ) ⇐⇒ 75 c d and S ⊇ T. The astute reader may suspect that compactness in general topological spaces is equivalent to a property involving nets and that a proof of Tychonov’s theorem can be given using subnets.2. ≪) is a directed set) and a map j : B → A such that (ii) for all a0 ∈ A there exists b0 ∈ B such that if b ≫ b0 then j(b) (i) yb = xj(b) for all b ∈ B. Furthermore.1. ﬁrst we need to deﬁne a subnet. This suspicion is well founded. Deﬁnition A. a0 . ) be a directed set and (xa )a∈A a net. A net (xa )a∈A is eventually in a set S if there exists a0 ∈ A such that xa ∈ S for all a a0 . whatever they may be.) If X is a topological space then x ∈ X is an accumulation point for a net if that net is frequently in every open set containing x. A subnet of (xa )a∈A is a net (yb)b∈B (where (B. such that (xa )a∈A is frequently in S for all S ∈ F.A Tychonov via Nets Compactness in metric spaces can be characterised by the behaviour of sequences (viz every sequence having a convergent subsequence: the BolzanoWeierstrass property). . S) ∈ A × F : xc ∈ S}.3.
converges to x. (d. for every S ⊆ X. From here we can prove the BolzanoWeierstrass theorem in its full generality. we deﬁne a net (xa )a∈A by choosing xa ∈ U ∈a U for all a ∈ A.5. d f and S.. .4. S) → c makes (yb )b∈B a subnet of (xa )a∈A that is eventually in S for all S ∈ F. Uxn }. Let x be an / accumulation point of (xa )a∈A . If b ≫ b0 and b ≫ b1 then xj(b) = yb ∈ U and j(b) a0 . S). there exists g f such that xg ∈ W . i.4. so there is a / ﬁnite subcover {Ux1 . / i=1 Deﬁnition A. as required. T ) ∈ B then there exist f ∈ A and W ∈ F such that c. let U be an open set containing x and let a0 ∈ A. suppose that X is compact and let (xa )a∈A be a net with no accumulation points: for all x ∈ X there exists an open set Ux containing x and ax ∈ A such that xb ∈ Ux for all b ax . since F is a cover. S) ∈ B and j : B → A. Theorem A. . . but if a ∈ A is such that a axi for i = 1. Then (xa )a∈A has a subnet that is eventually in every element of F. and let A denote the collection of ﬁnite subsets of F. ≪) is a directed set. . The net (xa )a∈A is frequently in U0 (as this set contains an accumulation point) and so there exists a ∈ A such that a ⊇ {U0 } and xa ∈ U0 .e. T ) ≫ b0 implies that j(b) a0 (we have a subnet) and that yb = xd ∈ T ⊆ S (so (yb )b∈B is eventually in S). n then xa ∈ n Uxi = X. (c. . . The point x ∈ X is an accumulation point of (xa )a∈A if and only if there is a subnet of (xa )a∈A that converges to x. and so an accumulation point. there exists U0 ∈ F such that x ∈ U0 . Conversely. so if b0 := (c. Since (xa )a∈A is frequently in W . the net is eventually in S or eventually in X \ S. W ) ∈ B.6. there exists c ∈ A such that c a0 and xc ∈ S. let x be an accumulation point of (xa )a∈A and let F = {U ∈ T : x ∈ U}. A is a directed set. . Let S ∈ F and a0 ∈ A. Conversely. The sets {Ux : x ∈ X} form an open cover of X. A topological space X is compact if and only if every net in X has a convergent subnet.76 Tychonov via Nets If (c. . this is a / contradiction. Since xa ∈ U ∈a U ⊇ U0 . T ⊇ W . T ) ≪ (g. Let (xa )a∈A be a net in the topological space X.S) = xc ∀ (c. There exists b0 ∈ B such that j(b) a0 for all b ≫ b0 (from the deﬁnition of a subnet) and there exists b1 ∈ B such that yb ∈ U for all b ≫ b1 (as (yb )b∈B converges to x). This gives the result. This contradiction gives the result. Proof Suppose that every net in X has a convergent subnet. This shows that (B. by Lemma A. A net in X is universal if. S) then b = (d. Proposition A. . Proof Let (yb )b∈B be a subnet of (xa )a∈A that converges to x. (d. by Proposition A. and so (c. we claim that deﬁning y(c. S). Suppose further that X has an open cover F with no ﬁnite subcover.3. Ordered by inclusion.
V ∈ U ⇒ U ∩ V ∈ U . Conversely.7.7). Hence (xa )a∈A converges to x.U )∈B and given (a.U ) )(a. U0 .9. (yb )b∈B = (y(a. Alternative Proof of Tychonov’s Theorem Let (X. Hence. The image of a universal net by any function is universal. if the space is compact then every universal net has a convergent subnet and so an accumulation point (by Proposition A. Every net has a universal subnet. If (xa )a∈A is a universal net in X and f : X → Y then let S ⊆ Y . (Note that f (f −1 (B)) = B ∩ f (X) for any B ⊆ Y .3. an ultraﬁlter ). and Lemma A. U) ∈ B there exists (b. A universal net converges to its accumulation points. it is eventually in U (since it cannot eventually be in its complement). Since the image of a universal net is universal 77 . the universal net is convergent. Since universal nets converge to their accumulation points (Proposition A. Proof By the BolzanoWeierstrass theorem (Theorem A. hence f (xa ) a∈A is eventually in either S or Y \ S. A space is compact if and only if every universal net is convergent. Since (xa )a∈A is frequently in U. S ∩ U : U ∈ U0 } ∈ C) and so (yb )b∈B is eventually in S. as required.5). Tb ) : b ∈ B} and suppose that (xa )a∈A is a universal net in X. C has a maximal element. the proof involves construction of what a Bourbakiste would call a ﬁlter (indeed. We can now present a proof of Tychonov’s theorem that has a beautiful simplicity (the dust having been swept under the rug that is Corollary A.) The next lemma is the most technically involved.Tychonov via Nets A universal net can be thought of as being ‘maximally reﬁned’. Lemma A. Theorem A. ordered by inclusion. Proof Let (xa )a∈A be a net in X and let C := U = {U ⊆ X} : U ∈ U ⇒ (xa ) is frequently in U.8. V ) ∈ B such that b a. U.5 gives the result. By Zorn’s lemma.8). by maximality. V ⊆ U and xb = y(b. the ﬁrst part of the next proposition makes this idea rigorous. if every universal net is convergent then.4). Proof If x is an accumulation point of the universal net (xa )a∈A then let U be an open set containing x. If S ⊆ X is such that (yb )b∈B is not eventually in X \ S then (yb )b∈B is frequently in S and so (xa )a∈A is frequently in U ∩ S for all U ∈ U0 : by the construction in Lemma A.V ) ∈ S. so as xb ∈ V (by construction) xb ∈ V ∩ S ⊆ U ∩ S. The net (xa )a∈A is eventually in either f −1 (S) or X \ f −1 (S) = f −1 (Y \ S). as required. S ∈ U0 (as U0 ∪ {S. T) be the product of the compact spaces {(Xb . This shows that (yb )b∈B is universal. as every net has a universal subnet (Lemma A. Corollary A. every net has a convergent subnet.3 implies that (xa )a∈A has a subnet (yb )b∈B which is eventually in every U ∈ U0 .9). Proposition A.
for all b ∈ B.9).1. by Corollary A.7). Exercises A Exercise A. and X is compact. . where xb := lima∈A πb (xa ) (by a property of initial topologies).9 again. Prove that (δn )n 1 has no weak*convergent subsequence but that (δn )n 1 has a weak*convergent subnet. Hence xa → x. (πb (xa ))a∈A is universal in Xb . so convergent (by Corollary A.78 Tychonov via Nets (Proposition A. Let X = ℓ∞ and for all n ∈ N deﬁne δn ∈ X ∗ by setting δn (xk )k 1 = xn .
sends open sets to open sets). [x] ∈ Q and we have the ﬁrst result. [y] − [x] = [y − x] X/M y−x <ε Deduce that the quotient norm yields the quotient topology on X/M given by Q := {U ⊆ X/M : π −1 (U) ∈ T}. where T denotes the norm topology on X. The quotient map is open by the Hence Bε ﬁrst part of this exercise. x → [x] is the natural map from X onto X/M (the quotient map). Bεx (The ﬁrst equality holds because π is surjective.1. If y ∈ Bε (x) then X Bε (x) := {y ∈ X : x − y < ε} ∀ x ∈ X. whence U = π π −1 (U) = x∈U X π Bεx (x) = x∈U X/M [x] ∈ T · .) Conversely. Let X be a normed vector space and let M be a closed subspace of X. Let T · denote the topology on X/M given by the quotient norm. whence [y] = π(y − m) ∈ π Bε (x) .Solutions to Exercises Solutions to Exercises 1 Exercise 1. If y ∈ X is such that [y] − [x] < ε then there exists m ∈ M X such that y − x − m < ε. where π : X → X/M. 79 .. and π(y) ∈ Bε [x] . ε > 0. If U ∈ Q then X π −1 (U) ∈ T and so π −1 (U) = x∈U Bεx (x). Prove that π{y ∈ X : y − x < ε} = [y] ∈ X/M : [y] − [x] < ε Let X denote the open ball in X with centre x and radius ε. and that the quotient map is open (i. since π −1 π(A) = A+M for all A ⊆ X. X/M [x] π −1 Bε X = π −1 π Bε (x) X = Bε (x) + M = m∈M X/M X Bε (x + m) ∈ T.e.
Solutions to Exercises Linearity is immediate and continuity follows from the fact that [x] x ∈ X. . ⇐⇒ ⇐⇒ x∈ ∃ (i0 . . . l ∈ L} are families of j sets. n=1 n=1 Exercise 1. Let M and N be subspaces of the normed space X. l) ∈ J × L ∀ j ∈ J. . J. K and L are arbitrary. as the quotient map π : E → E/M. Exercise 1. the restrictions of the functions in F to Y . x → [x] is continuous. xn } is a basis for M then {π(x1 ). Prove directly that if E is a Banach space and M is a closed subspace of E then the quotient space E/M. and. is the initial topology generated by F Y = {f Y : f ∈ F }. Hence ∞ xn − mn converges. Let [xn ] n 1 be a sequence in E/M with ∞ [xn ] convergent. ∩ Blk0 ∀ (j. k0 ) ∈ I × K such that x ∈ Ai j (i. Prove that if {Ai : i ∈ I.k)∈I×K (j.l)∈J×L Ai ∩ Blk . . by n=1 n=1 the completeness of E. and so initial topologies are as claimed in Deﬁnition 1. Exercise 1.k)∈I×K (j. and by comparison ∞ xn − mn is convergent. k0 ∈ K such that x ∈ Ai0 and x ∈ Blk0 j i AJ0 ∩ Blk . and that if {x1 . For all n 1 there n=1 exists mn ∈ M such that xn −mn [xn ] +2−n . · E/M is complete. l ∈ L What does this have to do with initial topologies? A consequence of this is that the collection of arbitrary unions of ﬁnite intersections of elements of a subbase is itself closed under ﬁnite intersections. Note that M + N = π −1 π(M) if π : X → X/N is the quotient map. where the index sets I. Prove that if TF is the initial topology on X generated by a collection of functions F and Y ⊆ X then TF Y . . so is itself closed. . Hence M + N is the preimage of a closed set under a continuous map. . . so does π( ∞ xn − mn ) = ∞ [xn ]. Prove that if M is ﬁnite dimensional and N is closed then M + N is closed.l)∈J×L ∃ i0 ∈ I. j ∈ J} and {Blk : k ∈ K.80 Prove also that the quotient map is linear and continuous. by deﬁnition of the quotient norm. x for all Exercise 1.3. π(xn )} is a spanning set for π(M).4. j We have that x∈ i∈I j∈J Ai ∩ j Blk k∈K l∈L ⇐⇒ as claimed.5. the relative initial topology on Y . then i∈I j∈J Ai ∩ j Blk = k∈K l∈L (i.2. so π(M) is ﬁnitedimensional and therefore closed.24.
hence a∈A0 xa → a∈S xa . j=1 (⋆) Let ε > 0 and choose A0 ∈ A such that a∈A0 xa > β − ε. this is immediate. . If A2 ∈ A then xa a∈A2 a∈A1 ∪A2 a∈A0 81 xa − α < ε for xa a∈A1 ∪A2 xa − α + α < α + ε. Deduce that (xa )a∈A is summable with sum a∈A a∈S xa ∞ j=1 xaj xa = if S is ﬁnite. so countable. if n0 := max{j : aj ∈ A0 } then n n0 . Conversely. Y Exercise 1.Solutions to Exercises Since f −1 (U) = f −1 (U) ∩ Y . If a∈A0 xa → α then let ε > 0 and choose A1 ∈ A such that all A0 ∈ A with A0 ⊇ A1 . a contradiction. Prove that if a∈A xa is convergent then S := {a ∈ A : xa = 0} is countable. Prove that (xa )a∈A is summable (with sum α) if and only if β = sup a∈A0 xa : A0 ∈ A < ∞ and in this case α = β. Hence S = n=1 Sn is a countable union of ﬁnite sets. and note that m ∞ j=1 xaj > m/n > β. Let Sn := {a ∈ A : xa > 1/n} for n ∈ N.6. am be distinct elements of Sn . where m > nβ. if β = sup a∈A0 xa : A0 ∈ A < ∞ then let ε > 0 and choose A1 ∈ A such that a∈A1 xa > β − ε. Let (xa )a∈A be a family of nonnegative real numbers and let A denote the collection of ﬁnite subsets of A. If Sn is inﬁnite for some n ∈ N then let a1 . . n ∈ N are such that m > n then m n m n j=1 xaj − xaj j=1 j=n+1 xaj β− xaj . j → aj is as above and m. where (if S is inﬁnite) j → aj is a bijection between N and S. n n0 . so from (⋆) we have that j=1 xaj a∈A0 xa > β − ε for all n m n j=1 xaj − j=1 xaj < β − (β − ε) = ε ∀ m. hence have α = β.7. Since R is Hausdorﬀ we must Exercise 1. if S is inﬁnite. Then β+ε> a∈A0 xa a∈A1 a∈A0 xa > β − ε for all A0 ∈ A such that A0 ⊇ A1 . Let E be a Banach space and let (xa )a∈A a family of vectors in E. . If S is ﬁnite then a∈A0 xa − a∈S xa = 0 for all A0 ∈ A such that A0 ⊇ S. If S is countably inﬁnite. applying the previous exercise to a∈A xa we see that β = sup a∈A0 xa : A0 ∈ A < ∞. so sup a∈A0 xa : A0 ∈ A < ∞. . xa → β. .
note a a that xa  = x+ + x− ∀ A0 ∈ A. za  a∈A is summable if and only if  Re za  a∈A and  Im za )a∈A are summable. Exercise 1. since max  Re z.  Im z z  Re z +  Im z ∀ z ∈ C. as claimed. If A0 = {a ∈ A3 : x+ > 0} ∪ A1 then a a∈A0 x+ a a∈A0 x+ − x− − x + a a x− + x < ε + a a∈A0 a∈A2 x− + x. Note also that (za )a∈A is summable if and only if Re za a∈A and Im za a∈A are summable. so convergent to x ∈ E.. a∈A0 Im za − Im z za − z a∈A0 a∈A0 Re za − Re z + a∈A0 Im za − Im z .an } n j=1 xaj − x a∈A0 xa − xa + ε j=1 < β − (β − ε) + ε = 2ε.. by Exercise 1. a the desired contradiction. a For this we suppose otherwise. 0} for all a ∈ A. take ε > 0 and choose A1 ∈ A such − that a∈A0 xa − x < ε for all A0 ∈ A that contain A1 . Given (xa )a∈A ⊆ R let x+ = max{xa .. . Prove that a family of complex numbers (za )a∈A is summable if and only if za  a∈A is summable. an }. a a a∈A0 a∈A0 a∈A0 + so xa  a∈A is summable if and only if (xa )a∈A and (x− )a∈A are summable. 0} and x− = − min{xa . . Hence it suﬃces to prove the real case of this proposition. . Set A2 = {a ∈ A1 : xa > 0} and let A3 ∈ A be such that x+ > a a∈A3 a∈A2 x− + x + ε. because if A0 is a ﬁnite subset of A then max a∈A0 Re za − Re z . If n n0 is such that − x < ε then. and xa = x+ − x− . n a∈A0 n j=1 xaj xa − x < xa + a∈A0 \{a1 . say. let a∈A xa = x.. . Since a (ya )a∈A is summable if and only if (−ya )a∈A is. a this exists as x+ a a∈A3 + a∈A xa is not convergent. Note ﬁrst that.6. it suﬃces to prove a a that the summability of (xa )a∈A implies that of (x+ )a∈A .82 Thus n j=1 xaj n 1 Solutions to Exercises is Cauchy. Hence a∈A0 xa → x. .8. for all A0 ∈ A such that A0 ⊇ {a1 .
B) → x. k}. B) ∈ A let x(A.Solutions to Exercises Exercise 1. continuous functions on X. B ∩ D) ∈ A. E). D).B) )(A. .9. D) ⇐⇒ A ⊇ C and B ⊇ D.. Exercise 1. . Prove the converse to Proposition 1.): for ε > 0 take k ∈ N such that ∞ n=k+1 1/n < ε and note that n∈N0 xn − x < ε for every ﬁnite subset N0 ⊆ N that contains {1. As the harmonic series is divergent. (C. 1/2. E) such that {x ∈ X : f (x) E ε} is compact for all ε > 0) is a closed subspace of Cb (X. xn n∈N is not summable).6 gives that xn n∈N is not summable. we claim that the net (x(A.1. those f ∈ C(X. Find a Hilbert space H and a countable family of vectors (xn )n∈N in H that is summable but not absolutely summable (i.11. y ∈ X such that every pair of open sets (U. Solutions to Exercises 2 Exercise 2. B) (U. V ) with x ∈ U and y ∈ V satisﬁes U ∩ V = ∅. . with 1 in nth position and 0 elsewhere. Let X be a topological space and E a Banach space. .. recall that Cb (X. Prove that C0 (X.10. A sequence in a normed vector space that is convergent is necessarily bounded.31. B) ∈ A such that A ⊆ U. B) and (C. . where en is the standard basis vector.e. Let H = ℓ2 (N) and let xn = en /n for all n ∈ N. . with preorder deﬁned by setting (A.e. . (xn )n∈N has sum 2 2 x := (1. Exercise 1. that in a space with a nonHausdorﬀ topology there exists a net that converges to two distinct points. If U ∈ T is such that x ∈ U then x(A.B)∈A converges to x and to y. k 83 sup n∈N0 xn : N0 is a ﬁnite subset of N sup n=1 1 : k ∈ N = ∞. is complete with respect to the norm f → f ∞ := sup{ f (x) E : x ∈ X}. and since x ∈ A ∩ C and y ∈ B ∩ D. B) (C. . hence x(A. This same argument works for y and so we have the result.B) ∈ U for all (A. B ∩ D) ∈ T × T is an upper bound for (A. the continuous. the space of Evalued. This is a directed set: if (A. E). 1/3.B) ∈ A ∩ B. (A ∩ C. X) and x(A.B) ∈ U for all (A. Is the same true for nets? For a simple counterexample. For all (A. E). bounded. However. consider Z directed by its usual order and deﬁne a net (xn )n∈Z by setting xn = n if n 0 and xn = 1/n if n > 0. n and Exercise 1. Suppose that T is not a Hausdorﬀ topology on the set X: there exist distinct points x. Evalued functions on X that vanish at inﬁnity (i. B). D) ∈ A then (A ∩ C. Let A ⊆ T × T denote the aggregate of such pairs.
since T is Hausdorﬀ it suﬃces to take x ∈ X and prove that there exist open sets separating ∞ and x. if ε > 0 and n ∈ N are such that fn − f ∞ < ε/2 then fn (x) f (x) − fn (x) − f (x) > f (x) − 1 ε. we see that f˙ is continuous. E) converge to f . E) let ˙ f˙ : X → E. Exercise 2. ˙ Let U ⊆ E be open. or {x ∈ X : αf ε} = {x ∈ X : f ε/α} is compact for all ε > 0. E) : f (∞) = 0}. If f . compact topology on X := X ∪ {∞}. x → f (x) if x ∈ X. 2 2 whence {x ∈ X : (f + g)(x) ε} ⊆ K ∪ L is compact. ˙ Prove that there is a natural correspondence between C0 (X.2. For f ∈ C0 (X. If C ⊆ T is an open cover of X then there exists U ∈ C such that ∞ ∈ U. Let (fn )n 1 ⊆ C0 (X. If α ∈ F then either α = 0. g ∈ C0 (X. Let (X. showing that T is compact. and that compact sets are closed in Hausdorﬀ ˙ ˙ spaces. 2 so {x ∈ X : f (x) ε} is a closed subset of the compact set {x ∈ X : fn (x) ε/2} and therefore is itself compact. E) is bounded. This shows that C0 (X. Finally. E). 0 if x = ∞. recall that compact sets are closed under ﬁnite unions and arbitrary intersections. Hence C0 (X. whence U and {∞} ∪ (X \ U) are elements of T as required. T) be a Hausdorﬀ. in which case αf = 0 ∈ C0 (X. E) then K = {x ∈ X : f (x) ε/2} and L = {x ∈ X : g(x) ε/2} are compact. ˙ It is routine to verify that T is a topology. E) and {f ∈ C(X. E) trivially. as f is continuous on the compact set {x ∈ X : f (x) E 1} and so bounded there. . E) is a subspace of Cb (X. and since X \ U is compact and has open cover C \ {U}. and if x ∈ K ∪ L then f (x) + g(x) f (x) + g(x) < 1 ε + 1 ε = ε. Hence C0 ∪ {U} is a ﬁnite subcover of the cover C. locally compact space and let ∞ denote a point not in X. As T is locally compact there exists U ∈ T such that x ∈ U ˙ ¯ ¯ and U is compact. Show that ˙ ˙ T := T ∪ {U ⊆ X : ∞ ∈ U. Since E f˙−1 Bε (0) = {∞} ∪ {x ∈ X : f (x) < ε} = {∞} ∪ X \ {x ∈ X : f (x) ˙ ε} ∈ T. if 0 ∈ U then f˙−1 (U) = f −1 (U) ∈ T ⊆ T.84 Solutions to Exercises Note that any function f ∈ C0 (X. E) is closed. X \ U is compact} ˙ is a Hausdorﬀ. and if 0 ∈ U then E without lost of generality U = Bε (0) (as we may write U as the union of a set of this form and the open set U \ {0} = (X \ {0}) ∩ U). this has a ﬁnite subcover ˙ C0 .
k ∈ (0. Hence ℓ1 / ker T is isometrically isomorphic to im T = E. Hence we may assume.3. s ∈ S} is countable (being the image of the countable set Q × S under the mapping (q. s) → qs). which is dense in E1 . This shows that [y] T [y] for all y ∈ ℓ . linearity is obvious. without loss of generality. Since the ˙ relative topology TX equals T. this set is kdense in E1 for all k ∈ (0. as is the absolute convergence of T x (and the fact that T 1) because ∞ ∞ xn en n=1 n=1 xn  = x 1 . dense set in X and note that {qs : q ∈ Q. 1). Let S be a countable. that S is closed under multiplication by rationals. the openmapping lemma gives that T is surjective. zn → x by the algebra of limits and zn zn as required. 1). let f˙ ∈ C(X. and if ε > 0 then E ˙ ˙ {x ∈ X : f (x) < ε} = {x ∈ X : f(x) < ε} ∩ X = f˙−1 Bε (0) ∩ X = X \ K 85 for some compact set K ⊆ X. E). ˜ ˜ where [x] = x + ker T and T : ℓ1 / ker T → im T = E is a bijection such that T x = T [x] 1 1 ˜ for all x ∈ ℓ . such exists by the ﬁrst part of the question. ⊆ E1 be dense. so if y ∈ E there exists xk ∈ ℓ1 such that y = T xk and xk T xk /(1 − k). and the opposite inequality ˜ follows from the fact T = T 1. Then := nyn /(n + 1) ∈ S for all n 1. Let (en )n 1 n n (yn − x) + x n+1 n+1 1 n + x n+1 n+1 1. Prove that X1 is separable (in the norm topology). x → T x := 1 xn en n=1 and note that T ∈ B(ℓ1 . Let x ∈ X1 and choose (yn )n 1 ⊆ S such that x − yn 1/n for all n 1. Since T (ℓ1 )1 ⊇ {en : n 1}. and let f = fX . Prove that any separable Banach space E is isometrically isomorphic to a quotient space of ℓ1 . Furthermore. 1) gives that [xk ] xk ˜ T [xk ] T xk = 1−k 1−k ∀ k ∈ (0. The identity ′ T xk = T xk′ for all k. E) be such that f(∞) = 0. as required. Deﬁne ∞ T : ℓ → E. . f is continuous. Exercise 2.Solutions to Exercises ˙ ˙ ˙ Conversely. Let X be a separable normed space. as required.
Prove that no inﬁnitedimensional Banach space E has a countable Hamel basis (where a Hamel basis is a linearly independent set S such that every vector in E is a ﬁnite linear combination of elements of S). Exercise 2.} is a countable Hamel basis for the ∞ Banach space E. as n → ∞. Each Fn is closed (being ﬁnitedimensional) and has empty interior: if U ⊆ Fn is open and nonempty E then it contains Bε (u) for some u ∈ U and ε > 0. e2 . It is immediate that if T is continuous then ker T = T −1 {0} is closed. Let f ∈ X and let ε > 0. 1]. Let (fn )n 1 ⊆ Dk be convergent. so is T . a contradiction to the Baire category theorem. 1]. so is a homeomorphism. realvalued functions on the unit interval and for all k ∈ N let Dk := {f ∈ X : there exists t ∈ [0.5. and for each fn let tn be as in the deﬁnition of Dk . . let Fn = Fe1 + · · · Fen and note that E = n=1 Fn . If ker T is closed then X/ ker T is a normed space (with respect to the quotient norm) and the quotient ˜ map π : X → X/ ker T is open and continuous. but then u + 1 ε en+1 −1 en+1 ∈ 2 U ⊆ Fn . and since this holds for all ε > 0 we see that f ∈ Dk . whence en+1 ∈ Fn . for such n. Then (tn )n 1 is a bounded sequence. T : X/ ker T → im T = Y is a linear bijection between ﬁnitedimensional normed spaces. Let T : X → Y be a linear transformation from the normed space X onto the ﬁnitedimensional normed space Y . Prove that T is continuous if and only if ker T is closed and that if T is continuous then T is open. f (s) − f (t) f (s) − fn (s) + fn (s) − fn (tn ) + fn (tn ) − f (tn ) + f (tn ) − f (t) < 2 f − fn ∞ + ks − tn  + ε → ks − t + ε ks−t for all s ∈ [0. . (Recall than a linear transformation between normed space is continuous if its domain is ﬁnite dimensional: if S : Y → Z is a linear transformation and dim Y < ∞ then the norm y → y Y + Sy Z is equivalent to · Y . Prove further that Dk is nowhere dense. This shows that E is a countable union of nowhere dense sets. . Let ε > 0 and n0 ∈ N be such that f (tn ) − f (t) < ε for all n n0 . Let X = C([0. 1] we may ﬁnd δ > 0 such that s−t < δ implies that f (s)−f (t) < ε/2. Exercise 2.86 Solutions to Exercises Exercise 2. so (by the BolzanoWeierstrass theorem. R) denote the Banach space of continuous. passing to a subsequence if necessary) we may assume that tn → t ∈ [0. with limit f . since f is uniformly continuous on [0. so there exists M > 0 such that ˜ Sy Z y Y + Sy Z M y Y for all y ∈ Y .4. contradicting linear independence. Furthermore. Choose 0 = t0 < t1 < · · · < tn = 1 .) As T = T ◦ π and both these maps are open and continuous. 1] such that f (s)−f (t) Prove that Dk is closed. 1]}.6. Suppose for contradiction that S = {e1 .
furthermore. . or s ∈ [ti−1 . s = t} < ∞. n and let g : [0. Next. . . 1]} = sup{g(s) − g(t)/s − t : s. . t − s ti − s and t − s tk − tk−1 if k lies between j and i). ti ]. . n). t ∈ [0. 2. Deduce that there exist continuous functions on [0. ti ] and t ∈ [tj−1 . so g(t) − g(s) t−s g(t) − g(tj−1) g(tj−1) − g(tj−2) g(ti ) − g(s) + +···+ t − tj−1 tj−1 − tj−2 ti − s (since t − s t − tj−1 . ti ]. s → s−t ′ f (t) s=t is continuous on [0. as if 0 s < t 1 then either s. i = 1. 1] → R. n ∈ Z) and let gm : [0. deﬁne the sawtooth function (draw a picture) 1 h : R → R. Since h ∞ ε/2 we have that f − gm ∞ < ε.Solutions to Exercises such that ti − ti−1 < δ for i = 1. 1] → R. 1] that are diﬀerentiable at no point in (0. tj ] for j > i. 1]. Hence f (s) − f (t) g ∞ t − s for all s ∈ [0. . x → g(x) + h(mx) for m ∈ N. g : [0. 2. ε m which shows that f lies in the closure of X \ Dk . If f ∈ X is diﬀerentiable at t ∈ (0. 1). . n + 1]. 1) then f (s) − f (t) s = t. M := inf{m ∈ R+ : g(s) − g(t) ms − t for all s. so if m > (M + k)/ε. 1] and s is suﬃciently near to t then gm (s) − gm (t) t−s h(ms) − h(mt) g(s) − g(t) − ms − mt s−t M +k > ε − M = k. Thus Dk has empty interior and is nowhere dense. 1]. t ∈ [0. . ti ] then f (t) − g(t) so f − g ∞ < ε/2. 1] → R. and f ∈ Dk for all k g ∞ . Since X = ∞ Dk (by the Baire category theorem) k=1 the result follows. so g(t) − g(s) /(t − s) = g ′ (t + s)/2 . t ∈ [ti−1 . t → t − ti−1 ti − t f (ti−1 ) + f (ti ) (if t ∈ [ti−1 . . ti − ti−1 ti − ti−1 ti − t t − ti−1 ε f (t) − f (ti−1 ) + f (t) − f (ti ) < . ti − ti−1 ti − ti−1 2 87 The function g is piecewiselinear (so continuous) and if t ∈ [ti−1 . 1]. so bounded. t → ε 2 − (t − n) (t ∈ [n. t ∈ [0.
n∈N αn en → αn en . e2 . .1. y 2 → 0 Deduce that B(H) is separable in the strong operator topology. . To see this. if A. Suppose that S ⊆ B(H)1 is dense and let SA ∈ S be such that SA − P (A) < 1/2 for all A ⊆ N. . Note that n 1 Pn B(H)Pn is strong operator dense in B(H) and that a countable union of countable sets is countable. 2 = (Pn − I)y. so this map is well deﬁned. (Pn − I)y = y 2 − k=1  ek . Pn B(H)Pn contains a countable. separable Hilbert space. Note that Pn T Pn x − T x Pn T (Pn − I)x + (Pn − I)T x T (Pn − I)x + (Pn − I)T x . . . Let {e1 . hence A → SA is injective and thus S is uncountable.} denote the linear span of this basis and for each subset A ⊆ N deﬁne P0 (A) : H0 → H0 . If A. Let H be an inﬁnitedimensional. note that Pn B(H)Pn is isomorphic 2 to F n and all norm topologies on ﬁnitedimensional spaces coincide. B ⊆ N are distinct then P (A) − P (B) 1: let n ∈ (A \ B) ∪ (B \ A) and consider P (A) − P (B) en . . strongoperatordense set. n∈A (In H0 only ﬁnitely many coeﬃcients in n∈N αn en are nonzero. prove that Pn T Pn x → T x as n → ∞ for all T ∈ B(H) and x ∈ H. so it suﬃces to prove that.} be an orthornormal basis for H.) It is immediate that P0 (A) 1. . Let H be a separable Hilbert space with orthonormal basis {e1 . let H0 = lin{e1 .7. Solutions to Exercises 3 Exercise 3. Prove that B(H)1 is not separable in the norm topology. . . so we may extend P0 (A) to P (A) ∈ B(H) such that P (A) = P0 (A)H0 and P (A) 1.}. so it suﬃces to prove that Pn y → y for all y ∈ H. Next. B ⊆ N are distinct then 1 P (A) − P (B) P (A) − SA + SA − SB + SB − P (B) < 1 + SA − SB . so Pn B(H)Pn is norm separable. For n 1 let Pn denote the orthogonal projection onto Fe1 + · · · + Fen . e2 . for all n 1. e2 .88 Solutions to Exercises Exercise 2. As the norm topology is ﬁner than the strong operator topology the result follows. . note that Parseval’s equality gives that n (Pn − I)y as n → ∞.
. Prove also that Tietze’s theorem applies to unbounded.4. so that g takes values in (−π/2. The set C = G−1 {±π/2} is closed and. h ∈ Cb (X. as E is complete. the principle of uniform M. Hence T T x = lim Tn x n→∞ M x ∀ x ∈ E. there exists H ∈ Cb (X) such that H ∞ 1. by the continuity of vector addition and scalar multiplication in a normed space. by Urysohn’s lemma. π/2). Prove that if E is a Banach space with respect to two diﬀerent norms then they are either equivalent or noncomparable (i. Apply Tietze’s theorem to obtain g. Prove the following extension of Tietze’s theorem to complexvalued functions: if X is a normal space.3. In particular. the topology T · generated by the ﬁrst is ﬁner than T · ′ . Exercise 3. without loss of generality. ∞.Solutions to Exercises Exercise 3. T · ) to (E. neither is coarser than the other). Exercise 3. Y a closed subset of X and f ∈ Cb (Y ) then there exists F ∈ Cb (X) such that F Y = f and F ∞ = f ∞ . { Tn x : n 1} is bounded for all x ∈ E. Y ) such that Tn → T in the strong operator topology. because boundedness implies that M := sup{ Tn : n 1} is ﬁnite. As convergent sequences in normed spaces are bounded. Since every continuous. if z > f Then F = l ◦ k is as required: if y ∈ Y then F (y) = l(k(y)) = l(f (y)) = f (y) and F ∞= l ∞ f ∞. realvalued functions: if X and Y are as above and f : Y → R is continuous then there exists F : X → R such that F Y = f . the identity map is continuous from (E. linear bijection between Banach spaces has continuous inverse.2. If the norms · and · ′ are comparable then. the topology generated by the second. x → lim Tn x n→∞ is a linear operator. z → z z f ∞ /z 89 if z f ∞. R) such that gY = Re f and hY = Im f . T · ′ ) (as T · ′ ⊆ T · ). HC = 0 and HY = 1. The function F = tan ◦ GH is as required: GH < π/2 and F (y) = tan(G(y)H(y)) = tan(arctan f (y)) = f (y) ∀ y ∈ Y. by Tietze’s theorem there exists G ∈ Cb (X) such that G ∞ π/2 and GY = g. Let k = g + ih. so that kY = f and let l : C → C. Prove that there exists T ∈ B(E. Y ) is such that limn→∞ Tn x exists for all x ∈ E.e. Let g = arctan ◦ f . Let E be a Banach space. What can be said about the norm of T ? . It is immediate that T : E → Y . and. we have the result. Y a normed vector space and suppose that (Tn )n 1 ⊆ B(E.
j=1 n n Since fn (y) y ∞ j=1 xj  it follows that fn is bounded. Find a Banach space E with closed subspaces F and G such that E = F ⊕ G and P : E → E. 0.90 Solutions to Exercises The working above shows that Tn n 1 is a bounded sequence. f + g → f ∀ f ∈ F. PG = PG and PF PG = PG PF = 0. since c0 is a Banach space. . as required. Let E be a Banach space with closed subspaces F and G such that E = F ⊕ G (i. T x = lim Tnk x k→∞ k→∞ lim Tnk x → lim Tn n→∞ x ∀ x ∈ E. so j=1 sup{fn (y) : n 1} < ∞ and. For all y ∈ c0 the sequence (fn (y))n 1 is convergent. y → xj yj . 2 2 Prove that PF and PG are bounded linear operators such that PF = PF . y ∈ F . so there exists a subsequence Tnk k 1 such that Tnk → limn→∞ Tn := supn 1 inf m n Tm . Thus x = y + (x − y) ∈ F + G and by uniqueness PF x = y. Exercise 3. Deﬁne PF and PG by setting PF : E → E. yn .7. 0. For n 1 deﬁne the linear operator n fn : c0 → C. Since a subsequence of a convergent sequence converges to the same limit. f + g → f and PG : E → E. This is actually an equality: let yj ∈ T := {α ∈ C : α = 1} be such that yj xj = xj  and note that z = (y1 . . . Let x = (xn )n 1 be a sequence of complex numbers such that the series ∞ 1 n=1 xn yn is convergent for all y ∈ c0 . . since PF xn ∈ F for all n and F is closed. every element of E can be expressed uniquely as the sum of an element of F and an element of G).6.5. .e. . Prove that x ∈ ℓ . Exercise 3. we prove only that PF is bounded. g ∈ G. g ∈ G has norm strictly greater than one. Hence T limn→∞ Tn . . the principle of uniform boundedness implies that ∞ x 1 = n=1 xn  = sup{ fn : n 1} < ∞.. The algebraic facts are easily veriﬁed. Let (xn )n 1 ⊆ E be such that xn → x and PF xn → y. . and similarly x−y = limn→∞ xn −PF xn ∈ G. Exercise 3. f + g → g ∀ f ∈ F.) ∈ (c0 )1 is such that fn (z) = n xj . by applying the closedgraph theorem. with j=1 xj  yj  n fn j=1 xj .
1). 1/2. where fn ∈ F and gn ∈ G. 2. g ∈ G. 2. x2 . F and G are as required. whence y = P x. 1 C (fn − fm ) + (gn − gm ) = C xn − xm . g ∈ G. Conversely. Prove that F ⊕ G is closed if and only if there exists C > 0 such that f C f +g ∀ f ∈ F. Now let z := (−1. note that v = P u P u = 1. 2 − 3v2 . Hence f = PF (f + g) PF f +g ∀ f ∈ F. [This example is due to Goodner [6].Solutions to Exercises Let E = R3 equipped with the norm x = max{x1 . where u := (1. 91 If F ⊕ G is closed then the projection map PF : F ⊕ G → F . E. v2 = 1/2. x3 } (writing vectors in bold and denoting their coordinates in the obvious manner) and let f : R3 → R. Let xn = fn + gn for all n 1. say fn → f ∈ F . 1) and y := x + t(u − P u).6). suppose that such C > 0 exists and let (xn )n 1 ⊆ F ⊕ G be such that xn → x for some x ∈ E. x2 . so 4 − 3v1  1 and v1 ∈ [1. (x1 . 5/2. If x := (1. and note that fn − fm so (fn )n closed. since x = (x − f (x)u) + f (x)u for all x ∈ R3 and f (u) = 1. 1) and w := z+s(u−P u). where t ∈ R is chosen so that y ∈ F . f + g → f is bounded (by Exercise 3. 1). 1. as required. as required. As w = (−1. f = g = 1} > 0. 4) − 3P u = (4 − 3v1 . F := ker f and G := Ru. v1 = 1 = v3 and. Let E be a Banach space with closed subspaces F and G such that F ∩ G = {0}. as y ∈ F . c := inf{ f − g : f ∈ F. where s ∈ R is chosen so that w ∈ F and so w = P z. since v1  1. Thus v = P u = (1. then y ∈ F ⇐⇒ x + tu ∈ F ⇐⇒ t = 3.] Exercise 3. so y = (4. so vi  1 for i = 1. 1). Furthermore. g ∈ G.8. it follows that s = 3 and w = (−1. Suppose for contradiction that P = 1 and let v := P u. by direct calculation. y = Px 1. . is Cauchy and hence convergent. −1. 1. Now. contradicting the fact that w = P z 1. 1). 3. 5/2]. x3 ) → 2x1 + 2x2 − 3x3 . 4 − 3v3 ). since G is n→∞ lim gn = lim xn − fn = x − f ∈ G n→∞ Deduce that F ⊕ G is closed if and only if and x = f + (x − f ) ∈ F ⊕ G. 1 + s/2.
whence α1 = · · · = αn = 0 by linear independence of i=1 [x1 ]. i. . Exercise 4. . m + n → φ0 (m) ∀ m ∈ M. The fact i=1 that N is closed is immediate and if x ∈ M ∩ N then φi (x) = 0 for all i. . Prove that M is complemented in X if M is ﬁnite dimensional. M + N = X and M ∩ N = {0}. Hence (replacing g by −g) f −g C −1 f = C −1 −1 if f ∈ F . . . as claimed. i. by the previous part.5). Then N is closed (being ﬁnitedimensional). . Let M be a ﬁnitedimensional subspace of the normed space X and let N be a closed subspace of X such that X = M ⊕ N. n ∈ N is an element of the dual space X ∗ . so for all n 1 there exist fn ∈ F and gn ∈ G such that fn > n fn + gn . The result follows from the fact that surjective linear transformations with ﬁnitedimensional range and closed kernel are continuous (Exercise 2. xn } be a basis for M. so x = 0. Let {x1 . use the ˜ ˜ HahnBanach theorem to extend φi to φi ∈ X ∗ and then let N = n ker φi. let {φ1 . as N is closed and ker φ0 is ﬁnitedimensional. dim X/M < ∞. .e. X = π −1 (X/M) = π −1 π(N) = M + N (where π : X → X/M is the quotient map) and if n αi xi ∈ i=1 M ∩ N then [0] = n αi [xi ]. Let X/M have basis [x1 ].2.. g ∈ G and f = g = 1. ˜ Finally. Solutions to Exercises 4 Exercise 4. Exercise 4. . [xn ] . . i=1 Prove also that M is complemented in X if M has ﬁnite codimension.e. xn }. there exists C > 0 such that f C f +g for all f ∈ F and g ∈ G. if F ⊕ G is not closed then there exists no such C.3.1. . replacing fn by fn / fn and gn by gn / fn we may assume that fn = 1 and fn + gn < 1/n. . whence c C > 0. . . . . . .92 Solutions to Exercises If F ⊕ G is closed then. . note that x − n φi (x)xi ∈ N for all x ∈ X. Since 1 − gn we see that fn + gn −1 = fn − − gn fn + gn < 1/n gn fn + gn + −gn + gn −1 gn < 1 + −1 + gn n −1 gn < 2 . φn } ⊆ M ∗ be the dual basis. . Prove that a normed vector space X is separable if its dual X ∗ is. Note that ker φ = N + ker φ0 is closed. . . n thus c < 2/n for all n 1 and so c = 0. Prove that if φ0 is a linear functional on M then φ : M ⊕ N → F.. [xn ] and suppose that N is the vector space spanned by {x1 . Conversely. A closed subspace M of the normed space X is complemented in X if there exists a closed subspace N such that M ⊕ N = X.
contradicting the fact that Z = Γ∗ (φ) = 0.5.4. Z = Γ∗ (φ) for some φ ∈ E ∗ (where Γ∗ : E ∗ → E ∗∗∗ is the canonical embedding) but then 0 = Z(Γ(x)) = Γ∗ (φ)(Γ(x)) = Γ(x)(φ) = φ(x) ∀ x ∈ E. so Ψ(Z ◦ Γ) = Γ(x)(Z ◦ Γ) = Z(Γ(x)) = Z(Ψ). Prove that any inﬁnitedimensional normed space has a discontinuous linear functional deﬁned on it. Prove that a Banach space E is reﬂexive if and only its dual E ∗ is If E is reﬂexive then the canonical embedding Γ : E → E ∗∗ is an isometric isomorphism: for all Z ∈ E ∗∗∗ we wish to ﬁnd φ ∈ E ∗ such that Z(Ψ) = Ψ(φ) ∀ Ψ ∈ E ∗∗ .Solutions to Exercises Let (φn )n 1 be dense in X ∗ . Exercise 4. First use Zorn’s lemma to prove that the space X has a Hamel basis {ea : a ∈ A} with ea = 1 for all a ∈ A. Deﬁne φ by choosing an inﬁnite set of distinct elements {an : n ∈ N} and setting ∞ φ a∈A λa ea = n=1 nλan ∀x = a∈A λa ea ∈ X. and if Ψ ∈ E ∗∗ then Ψ = Γ(x) for some x ∈ E. as required. Conversely.9 give Z ∈ E ∗∗∗ such that ZΓ(E) = 0 but Z = 0. reﬂexive.2 and Theorem 3. Exercise 4. Let (φnk )k 1 converge to φ and note that 1 2 93 φ nk φnk (xnk ) φnk (xnk )−φ(xnk )+φ(xnk ) φnk −φ xnk φnk −φ → 0 Find a separable Banach space E such that E ∗ is not separable. φ is not bounded. as k → ∞. implying that φ = 0. it is easy to verify that φ is a welldeﬁned linear functional on X and since φ(ean ) = n → ∞ as n → ∞. as E ∗ is reﬂexive. . by a corollary to the separation theorem there exists φ ∈ X ∗ such that φM = 0 and φ(x0 ) = 1. Note that the separable Banach space ℓ1 has nonseparable dual space ℓ∞ . Prove that a reﬂexive Banach space E is separable if and only if E ∗ is. so that Γ(E) is a proper subspace of E ∗∗ (which is closed because E is complete). let (xn )n 1 ⊆ X1 be such that φn (xn ) φn /2 for all ¯ = X then let n 1 and let M = Q − lin{xn } (or (Q + iQ) − lin{xn } if F = C). The ﬁrst part gives one implication and the other follows immediately by applying the ﬁrst part with E ∗ in place of E. This contradiction gives the result. it is immediate that φ ∈ E ∗ . Exercise 4. note that E and E ∗∗ are isometrically isomorphic. Consider φ = Z ◦ Γ. suppose that E ∗ is reﬂexive but E is not. If M ¯ x0 ∈ X \ M .
Note that Ta → T in the weak operator topology on B(H) if and only if x. 1 (φ ◦ f )(z) − (φ ◦ f )(a) = z−a 2πi γ Deduce that a weakly holomorphic function is (strongly) continuous. 2π] → C.31]. Let H be a Hilbert space. If φ ∈ X ∗ then φ(a) φ a φ r for all a ∈ A. x → e1 . ∗ ∗ ∗ Tn = e1 en  and Tn x =  en . For the second.] Exercise 4. y = y. t → a + εeit and Mφ = sup (φ ◦ f )(w) : w − a = ε . Let A be a subset of the normed vector space X. f (z) − f (a) : 0 < z − a < ε/2 z−a is weakly bounded. so norm bounded: there exists r ∈ R+ such that f (z) − f (a) rz − a C ∀ z ∈ Bε/2 (a). Hence (φ ◦ f )(w) dw (w − z)(w − a) Mφ ε/2 C ∀ z ∈ Bε/2 (a) \ {a}. by Cauchy’s integral formula.94 Solutions to Exercises Exercise 4. y ∈ H. where U is an open subset of C and C X is a complex normed space. whence f (z) → f (a) in X as z → a. Show that if T : E → F and S : F ∗ → E ∗ are linear transformations that satisfy φ(T x) = (Sφ)(x) then S and T are bounded. If A is norm bounded then there exists r ∈ R+ such that a r for all a ∈ A. Theorem 3. Let a ∈ U and choose ε > 0 such that B2ε (a) ⊆ U. if A is weakly bounded then {φ(a) : a ∈ A} = {ˆ(φ) : a ∈ A} is a bounded for all φ ∈ X ∗ . x  → 0 as n → ∞ (by Parseval’s equality) so Tn → 0 in the strong operator topology. [This is the ﬁrst step in proving Dunford’s theorem. It is readily veriﬁed that Tn ∈ B(H) (with Tn 1) and Tn e1 = 1. that weakly holomorphic functions are strongly holomorphic [19. Furthermore. This gives the result. ˆ ˆ Suppose that f : U → X be weakly holomorphic.7.6. Sy = S ∗ x. φ ∈ F ∗ (⋆) Tn : ℓ2 → ℓ2 . Prove that A is norm bounded (there exists r ∈ R+ such that a r for all a ∈ A) if and only if it is weakly bounded (for all φ ∈ X ∗ there exists rφ ∈ R+ such that φ(a) rφ for all a ∈ A).8. Since a = a we see that A is norm bounded. T y for all x. i.. showing that A is weakly bounded. Ta y → x. Let E and F be Banach spaces. Exercise 4. with S = T ∗ . x en .e. Conversely. . Since x. Prove that the adjoint T → T ∗ is continuous with respect to the weak operator topology on B(H). S ∗x and z → z is continuous ¯ on C we have the ﬁrst claim. let {en : n ∈ N} denote the standard orthonormal basis of ℓ2 and let Tn = en e1  for all n 1. ∀ x ∈ E. but not necessarily with respect to the strong operator topology. if ∗ φ ∈ X then. where γ : [0. and the principle of uniform boundedness yields boundedness of { a : a ∈ A}.
ψ ∈ E ∗ satisﬁes ψ(x) = 0 for all x ∈ ker T . ∞ n=1 xn is The weak topology on ℓ1 is such that y : x → ∞ yn xn is continuous for all y ∈ ℓ∞ . E ∗ ) is such that S = T ∗ . and suppose that (xn )n 1 ⊆ E is such that xn → x and T xn → y. n→∞ n→∞ 95 hence T x = y (since F ∗ separates points in F ). Prove that x → weakly continuous on ℓ1 but is not weak* continuous. and deﬁne θ0 : E/ ker T → F. by the closedgraph theorem T ∈ B(E.) then ˆ : x → ∞ xn is weakly continuous. 1 Exercise 4. so θ0 [x] ψ [x] for all x ∈ E and θ0 is continuous. If x(n) n 1 ⊆ ℓ1 is 1 n=1 (n) (n) ˆ deﬁned by setting xn = 1 and xk = 0 if k = n then 1 x(n) = 1 for all n 1 but y x(n) = yn → 0 as n → ∞ for all y ∈ c0 . i. x(n) → 0 in the weak* topology. . 1.. extending θ0 ◦ T −1 to θ ∈ F ∗ by the HahnBanach theorem we see that (T ∗ θ)(x) = θ(T x) = θ0 [x] = ψ(x) ∀x ∈ E so ψ ∈ im T ∗ and the result follows. Hence ˆ ˆ is not weak* continuous. Furthermore. so that im T ∗ ⊆ (ker T )⊥ . so that E ∗ = ℓ1 and E ∗∗ = ℓ∞ . the graph of T .Solutions to Exercises Let (x. the ˜ bounded linear operator T : E/ ker T → im T has continuous inverse (by the open˜ mapping theorem). Conversely.e.. i.e. .9. . im T is closed in F . suppose that ψ ∈ (ker T )⊥ .. Since im T is closed. F ) has closed range. Let E and F be Banach spaces and suppose that T ∈ B(E.11. For all φ ∈ F ∗ we have that Sφ ∈ E ∗ and so φ(y) = lim φ(T xn ) = lim (Sφ)(xn ) = (Sφ)(x) = φ(T x). [x] → ψ(x). Prove that im T ∗ = (ker T )⊥ (where M ⊥ := {φ ∈ E ∗ : φ(x) = 0 for all x ∈ M} is the annihilator of the subspace M ⊆ E).e. Prove that a compact metric space is separable.10. [This is the HellingerToeplitz theorem. T ∗ ∈ B(F ∗ .] Exercise 4. Exercise 4. Let E = c0 . by (⋆). . as E certainly separates points in E ∗ . F ). ˆ n=1 so if 1 := (1. This is a good deﬁnition (because ker T ⊆ ker ψ) and θ0 [x] = ψ(x) = ψ(x + m) ψ x+m ∀ m ∈ ker T. y) be a limit point of G(T ). If φ ∈ F ∗ then (T ∗ φ)(x) = φ(T x) = φ(0) = 0 ∀ x ∈ ker T. i.
96
Solutions to Exercises Let (X, d) be a compact metric space. For n ∈ N the set {B1/n (x) : x ∈ X} is an (n) (n) (n) open cover of X, so there exist x1 , . . . , xmn such that X = mn B1/n xk . We claim k=1 (n) (n) that S = n∈N x1 , . . . , xmn is countable (being a countable union of ﬁnite sets) and dense in X. For the latter claim, let ε > 0 and x ∈ X; there exists n ∈ N such that (n) (n) ε > 1/n and some k ∈ {1, . . . , mn } such that x ∈ B1/n xk , whence d x, xk < ε and S ∩ Bε (x) = ∅. The result follows.
∗ Prove that if X is a separable normed space then X1 , the closed unit ball of the dual space X ∗ , is metrizable when equipped with the weak* topology.
Let (xn )n
1
⊆ X1 be dense in X1 and deﬁne
∗ ∗ d : X1 × X1 → R+ ; (φ, ψ) →
∞
n=1
2−n φ(xn ) − ψ(xn ).
(†)
Note that the series is convergent (by comparison with ∞ 2−n φ − ψ ) so d is well n=1 deﬁned. Symmetry and the triangle inequality are immediate and if d(φ, ψ) = 0 then ∗ (φ − ψ)(xn ) = 0 for all n 1, whence φ = ψ. Hence d is a metric on X1 ; it remains to ∗ ∗ prove that Td = σ(X , X)X1 . Note ﬁrst that (φ, ψ) → φ(xn ) − ψ(xn ) = ˆn (φ − ψ) x ∗ ∗ is a continuous function on X1 × X1 (where each factor is equipped with the weak* topology) and the series (†) is uniformly convergent on this set, so continuous. In particular, the balls
∗ {ψ ∈ X1 : d(ψ, φ) < ε} ∗ (φ ∈ X1 , ε > 0)
∗ ∗ ∗ are σ(X ∗ , X)X1 open, so Td ⊆ σ(X ∗ , X)X1 . As Td is Hausdorﬀ and σ(X ∗ , X)X1 is compact, these topologies are equal. (Recall that a continuous bijection from a compact space to a Hausdorﬀ space is a homeomorphism.)
Deduce that X ∗ is separable in the weak* topology.
∗ This follows because Xn is separable for all n ∈ N (being the image of the weak*∗ ∗ separable space X1 under the homeomorphism x → nx) and X ∗ = n∈N Xn .
Exercise 4.12. Let X and Y be normed spaces and for all x ∈ X and y ∈ Y let x ⊗ y : B(X, Y ∗ ) → F; T → (T x)(y). Prove that x ⊗ y ∈ B(X, Y ∗ )∗ , with x ⊗ y = x is bilinear. It is straightforward to see that x ⊗ y ∈ B(X, Y ∗ ) with x ⊗ y x y ; for the ∗ reverse inequality, suppose that x and y are nonzero and let φ ∈ X and ψ ∈ Y ∗ be such that φ = ψ = 1, φ(x) = x and ψ(y) = y . (These exist by Theorem 3.10). If T := z → φ(z)ψ then T ∈ B(X, Y ∗ ), with T φ ψ = 1, and (x ⊗ y)(T ) = φ(x)ψ(y) = x y , whence x ⊗ y x y , as required. Veriﬁcation of bilinearity is routine. y , and that the mapping X × Y → B(X, Y ∗ ); (x, y) → x ⊗ y
Solutions to Exercises If Z is the closed linear span of {x ⊗ y : x ∈ X, y ∈ Y } in B(X, Y ∗ )∗ , prove that j : B(X, Y ∗ ) → Z ∗ ; j(T )z = z(T ) is an isometric isomorphism. Clearly, j is a linear map and j(T ) T for all T ∈ B(X, Y ∗ ). For the reverse inequality, let ε > 0 and choose x ∈ X such that x = 1 and T x > T − ε/2, then choose y ∈ Y such that y = 1 and (T x)(y) > T x − ε/2. It follows that x ⊗ y = 1 and (x ⊗ y)T = (T x)(y) > T − ε, whence j(T ) T . Finally, ∗ let φ ∈ Z and let φx : Y → F; y → φ(x ⊗ y). Since φ(x ⊗ y) φ x y , φx ∈ Y ∗ , and if T := x → φx then T φ , so T ∈ B(X, Y ∗ ) and j(T ) = φ; hence j is surjective. [This idea, which allows one to put a weak* topology on B(X, Y ∗ ), was used by Arveson in [1] and dates back to Schatten [20, Theorem 3.2].]
97
Solutions to Exercises 5
Exercise 5.1. Prove that if X is a vector space with separating subspace M ⊆ X ′ and φ ∈ X ′ is a linear functional that is σ(X, M)continuous then there exist φ1 , . . . , φn ∈ M such that φ(x) max φi (x) ∀ x ∈ X.
1 i n
By the deﬁnition of σ(X, M),
n
φ [0, 1) ⊇
−1
i=1
φi−1 [0, εi ),
where φ1 , . . . , φn ∈ M and ε1 , . . . , εn > 0. Replacing φi by φi /εi if necessary, without loss of generality ε1 = · · · = εn = 1. If x ∈ X is such that α := φ(x) > max1 i n φi(x) then φ(α−1x) = 1 but
1 i n
max φi (α−1 x) = max φi(x)/φ(x) < 1,
1 i n
a contradiction. Deduce that
n i=1
ker φi ⊆ ker φ and that there exists f ∈ (Fn )∗ such that f φ1 (x), . . . , φn (x) = φ(x) ∀ x ∈ X. ∀x ∈ X
The ﬁrst deduction is immediate; for the next, deﬁne Φ : X → Fn by setting Φ(x) := φ1 (x), . . . , φn (x) and note that ker Φ = n ker φi ⊆ ker φ, so that if Φ(x) = Φ(y) then φ(x) = φ(y). i=1 Hence f0 : Φ(X) → F; φ1 (x), . . . , φn (x) → φ(x) is well deﬁned, and we extend f0 to f ∈ (Fn )∗ by setting f Φ(X) = f0 and f Φ(X)⊥ = 0.
98 Conclude that φ ∈ M.
Solutions to Exercises
Let f (0, . . . , 1, . . . , 0) = αi (where the 1 is in the ith place) so that
n n
φ(x) = f φ1 (x), . . . , φn (x) =
i=1
φi (x)αi =
i=1
αi φi (x)
∀ x ∈ X,
i.e., φ =
n i=1
αi φi ∈ M.
Exercise 5.2. Let X be an inﬁnitedimensional normed space and let V ⊆ X be a weakly open set containing the origin. Show that V contains a closed subspace of ﬁnite codimension in X. Since V is a weakly open set containing 0 there exist φ1 , . . . , φn ∈ X ∗ such that V contains n φj −1 [0, 1). The linear transformation j=1 Φ : X → Fn ; x → φ1 (x), . . . , φn (x) has kernel ker Φ = n ker φi ⊆ V , which is closed because ker φi is closed for i = i=1 1, . . . , n. Since X/ ker Φ ∼ im Φ ⊆ Fn , ker Φ has ﬁnite codimension, as desired. =
Deduce that the weak topology on X is strictly coarser than the norm topology.
X Note that the open unit ball B1 (0) is open with respect to the norm topology on X, but it contains no subspace of X other than {0}, which does not have ﬁnite codimension. Hence σ(X, X ∗ ) has fewer open sets than the norm topology.
(To see that these topologies are comparable, note that x → φ(x) is norm continuous for all φ ∈ X ∗ , so φ−1 (U) is open (with respect to the norm topology) for all open U ⊆ F.) Exercise 5.3. Let X be a topological vector space. Prove that every φ ∈ X ′ \ {0} is open. Let x0 ∈ X be such that φ(x0 ) = 1 and let A ⊆ X be open. If x ∈ A then A − x is an open set containing 0 (by the continuity of vector addition) and so m−1 (A − x) is x0 an open set containing 0, where mx0 : F → X; α → αx0 (by the continuity of scalar multiplication). Hence there exists δ > 0 such that αx0 ∈ A − x if α < δ, and so α + φ(x) = φ(αx0 + x) ∈ φ(A) for such α. This shows that φ(A) is open and gives the result. Exercise 5.4. Suppose that X is a vector space equipped with a topology that makes vector addition and scalar multiplication, i.e., the maps X × X → X; (x, y) → x + y and F × X → X; (α, x) → αx,
continuous. Show that if this topology is such that singleton sets are closed (i.e., {x} is closed for all x ∈ X) then the topology is Hausdorﬀ (so X is a topological vector space).
. Bx may be taken to be balanced. Let p ∈ (0. Exercise 5. y 0. 99 Deduce that if C ⊆ X is compact and does not contain the origin then there exist disjoint open sets A. As m : F × X → X. y ∈ [−1. by the deﬁnition of the product topology. 0). 1] → C f is measurable and ∆(f ) < ∞ . B ⊆ X such that C ⊆ A and B is a balanced set containing 0. [g]) := ∆(f − g) is a metric on Lp [0. for all x ∈ C there exist disjoint open sets Ax . 1] := f : [0.Solutions to Exercises Note that. y) : x. 1). and if B := Bx1 ∩ · · · ∩ Bxn then A and B are as required. 1] and that Lp [0. x) → λx is continuous. . so it suﬃces take x ∈ X \{0} and ﬁnd an open set U such that 0 ∈ U and U ∩(x+U) = ∅. z) → y − z is continuous and so {(y. it also gives . Hence there exist open sets V . The crossshaped set {(x. 1]} ⊆ R2 is balanced but not convex. by the ﬁrst part of this exercise. . the path t → (1 − t)x connects x to the origin. Since C is compact. contained in U and open. (y. 1] is a topological vector space (when equipped with this topology).5. v ∈ V } is balanced. The meanvalue theorem may be used to establish the inequality (1 + z)p − z p 1 for all z 0. W ⊆ X such that 0 ∈ V ∩ W and x ∈ V − W .6. Lp [0. 1] is closed under sums and so is a vector space. for ﬁxed y ∈ X. there exist x1 . Bx ⊆ X such that x ∈ Ax and 0 ∈ Bx . Prove that every open set containing the origin contains a nonempty open set which is balanced : a set B is balanced if λb ∈ B for all b ∈ B and λ ∈ F1 . For any point x in a balanced set B. Note that the map X × X → X. xn ∈ C such that A := Ax1 ∪ · · · ∪ Axn ⊆ X. 1] := Lp [0. This is enough to see that Lp [0. Exercise 5. (0. It is easily veriﬁed that m(Bε (0) × V ) = {λv : λ < ε. where ∆(f ) := 1 0 f (x)p dx. 0). since it is the union of λV over all λ ∈ F such that 0 < λ < ε and the map v → λv is a homeomorphism for all λ = 0. As X is Hausdorﬀ. the map x → x + y is a homeomorphism of X with itself. 1]/N. where N := f : [0. 0). and let Lp [0. thus B is path connected. (λ. Prove that d([f ]. Let U ⊆ X be an open set containing 0. Show that a balanced set is connected and give an example to show that a balanced set need not be convex. there exist ε > 0 and an open set V ⊆ X such that 0 ∈ V and F F Bε (0) × V ⊆ m−1 (U). . from which it follows that (x + y)p xp + y p for all x. Let X be a topological vector space. m−1 (U) is an open set in F × X which contains (0. so connected. 1] → C f is measurable and zero almost everywhere . taking U = V ∩ W gives the result. z) ∈ X ×X : y −z = x} is an open subset of X ×X containing (0.
Let {e1 . It follows immediately that f ∈ V and V = Lp [0. for all i. By the intermediatevalue theorem.5. . 0)/n (i = 1. .100 Solutions to Exercises the triangle inequality for d. . 0) = np xi xi−1 f (t)p dt = np−1 d(f. (λ1 .7. λn ) → λ1 e1 + · · · + λn en . 1]′ is continuous then φ−1 Bε (0) is convex. Prove further that Lp [0. [fa ]) + d([g]. 1]. . 1] has no convex.] n=1 Exercise 5. . Let f ∈ Lp [0. 0) < ε} ⊆ V . T (S). . [fa ]) show that vector addition and scalar multiplication are suitably continuous. 1] and choose n ∈ N such that np−1 d(f. it remains to show that T is open. Let S := {(λ1 . The other two requirements of a metric are obviously satisﬁed and the inequalities d([f + g]. . 1]) ⊆ Bε (0) for all ε > 0 and so φ = 0. By Exercise 5. 1]. this is compact and therefore so is its image. It is immediate that T is a linear bijection which is continuous. [ga]) 1 f (t)p dt + λa p d([f ]. 0) < ε Deduce that the only continuous linear functional on Lp [0. there exist 0 = x0 < x1 < · · · < xn = 1 such that xi xi−1 f (t)p dt = d(f. . where n is the dimension of M. n). . . so that f = (g1. V ⊆ M such that 0 ∈ U. who worked with the sequence space {(xn )n∈N : ∞ xn 1/2 < ∞}. open and nonempty. Let V = ∅ be convex and open. Prove that M is linearly homeomorphic to Fn . [fa + ga ]) and d(λf. 1] is the zero functional. λa fa ) λ − λa  p 0 d([f ].n )/n and d(gi.n := nf on [xi−1 . .n + · · · + gn. [This argument is (essentially) due to Tychonov [23]. . xi ) and 0 elsewhere. Hence φ(Lp [0. let gi. Let X be a topological vector space over F and let M be a ﬁnitedimensional subspace of X. . . λn ) : λ1 2 + · · · + λn 2 = 1} be the unit sphere in Fn . without loss of generality 0 ∈ V and so there exists ε > 0 such that {f ∈ Lp [0. which is . . there exist disjoint open sets U.n . C If φ ∈ Lp [0. 1] : d(f. . 0) < ε. because scalar multiplication and vector addition are. open sets other than ∅ and Lp [0. for all C ε > 0. en } be a basis for M and let T : Fn → M.
101 since T w + εU is open and contains T w. . Prove that a topological vector space with topology given by a separating family of linear functionals is locally convex. However.) Exercise 5. by Exercise 5. . Suppose that X is a locally convex topological vector space and M is the collection of continuous linear functionals on X. there exists u ∈ U such that u = T (λ1 . Exercise 5.9. . σ(X. To see this. (The ﬁrst inclusion holds as rU is open. Let x be in the closure of M and note that tx → 0 as t → 0+.Solutions to Exercises F balanced. the initial topology generated by M. 1) and suppose u and v ∈ X are such that φ(u) < ε and φ(v) < ε. . Prove also that M is closed in X. φn ∈ M and ε1 . . As σ(X. as required. . .32(ii). 1] for some p ∈ (0. . Let T be the map above. there exist φ1 .8. there exists φ ∈ M such that Re φ(x0 ) < inf y∈B Re φ(y). the set T (W ) is open and the result follows. the onedimensional subspace of X consisting of the . F = {0} and TF = {0. . so there exists r > 0 such that x ∈ rU. clearly T is (still) continuous and the same working as above yields an open set U ⊆ X such that 0 ∈ U F and U ∩ M ⊆ T (B1 (0)). as required.6. . and T (S) ⊆ V . by Theorem 3. λn ) ∈ S and α−1/2 u ∈ T (S) ∩ U. . now considered to have codomain X. . M). . a contradiction. M)closed. M) is the coarsest topology to make every element of M continuous. . but then α−1/2 (λ1 . 1) then. Thus B is σ(X. εn > 0 such that n i=1 −1 F φ−1 (Bεi (0) ⊆ U. Need this hold if X is not locally convex? No: if X = Lp [0. Then φ(tu + (1 − t)v) tφ(u) + (1 − t)φ(v) < ε. so F F T w + εU ⊆ T w + εT (B1 (0)) ⊆ T (w + Bε (0)) ⊆ T (W ). . λn ) with α := λ1 2 + · · · + λn 2 1. Prove that a convex subset of X is closed (with respect to the original topology) if and only if it is closed with respect to σ(X. Hence F x ∈ rU ∩ M ⊆ rU ∩ M ⊆ rT (B1 (0)) ⊆ T ((Fn )r ) = T ((Fn )r ) ⊆ M. since (Fn )r is compact and T is continuous. with separating family M. by the continuity of scalar multiplication. every σ(X. by deﬁnition. i F so it suﬃces to prove that φ (Bε (0)) is convex for any φ ∈ M and ε > 0. If U is an open subset of X containing the origin. It follows that U ⊆ T (B1 (0)): otherwise. Hence if W ⊆ Fn is F open and w ∈ W then there exists ε > 0 such that Bε (0) ⊆ W − w. the trivial topology. the original topology on X. . Conversely. M)closed set is TX closed. M) is coarser than TX . in particular. if B ⊆ X is convex and TX closed then let x0 ∈ X \ B. Let X be such a topological vector space. X}. let t ∈ (0.
Letting U := W1 ∩ W2 establishes the claim. convex A. which is closed. λn 1) + d(λn 1. compactness yields x1 .5. where A is compact and B is open.11.32) there exists a continuous φ0 ∈ X ′ such that Re φ0 (x0 ) < inf{Re φ0 (y) : y ∈ B}. so has a convergent subnet. y) → x + y is continuous and 0 + 0 = 0 ∈ V . (yU ) converges to 0 and (xU ) is contained in the compact set A. Suppose. then there exists an open set U ⊆ X such that 0 ∈ U and A + U ⊆ B. either use Exercise 5. . φ0 {x0 } ∩ φ0 (B) = ∅. Hence φ0 (x0 ) = 0 and φ : x → φ0 (x)/φ0 (x0 ) is as required.12. ¯ Let B = N . Exercise 5.) Exercise 5. hence there exist open sets W1 . x∈A x∈B . This contradiction gives the result. . .102 Solutions to Exercises constant functions is closed. (For the ﬁrst of these claims. nonempty and not the whole of X. nonempty. by the ﬁrst part. W2 ⊆ X such that 0 ∈ W1 ∩ W2 and W1 + W2 ⊆ V . as required. but xU + yU ∈ X \ B for all U. Deduce or prove otherwise that if A ⊆ B ⊆ X.10. As x + Ux is open for all x ∈ A. (x. if U := Ux1 ∩ · · · ∩ Uxn then U is an open set containing 0 and n n A+U ⊆ as required. so f = λ1. compact. Suppose that X is a topological vector space such that the continuous elements of X ′ separate points. λm 1) = λn − λn p and so λn → λ for some λ ∈ C. B ⊆ X there exists a continuous φ ∈ X ′ such that sup Re φ(x) < inf Re φ(x). λ) → 0. Since a : X × X → X. Let X be a locally convex topological vector space. so z ∈ X \ B ⊆ X \ A. xn ∈ A such that A ⊆ (x1 + Ux1 ) ∪ · · · ∪ (xn + Uxn ). 0). The Appendix provides the means for an alternative proof. that if U ⊆ X is open and contains 0 then there exist xU ∈ A and yU ∈ U such that xU + yU ∈ B. i=1 (xi + Uxi + U) ⊆ i=1 (xi + Uxi + Uxi ) ⊆ B. . convex. the preimage a−1 (V ) is an open set containing (0. In particular. Exercise 5. Show that if N ¯ is a nonempty subspace of X and x0 ∈ X \ N then there exists a continuous linear ′ functional φ ∈ X such that φN = 0 and φ(x0 ) = 1. by the separation theorem (Theorem 3. Prove that given disjoint. with limit z in A.7 or prove it directly: if the sequence of constant functions (λn 1)n 1 converges to f ∈ X then it is Cauchy. Hence (xU + yU ) has a convergent subnet. but d(λn 1. there exists an open set Ux which contains 0 and satisﬁes Ux + Ux ⊆ B − x. If x ∈ A then B − x is an open set containing 0 so. λ1) d(f. d(f. Ordering such U by reverse inclusion. Prove there exists an open set U containing 0 such that U + U ⊆ V . Furthermore. Let X be a topological vector space and suppose V is an open set containing 0. and as φ0 (B) is a subspace of φ0 (X) = F we must have that φ0 (B) = {0}. for contradiction. by Theorem A.
by Exercise 5. 1) are such that tx + (1 − t)y ∈ F then tx + (1 − t)y lies in every element of C.32(i) yields φ ∈ M and s ∈ R such that Re φ(x) < s inf y∈B Re φ(y) for all x ∈ A + U. σ(X. If C is a chain in F then consider F . It is 2 2 immediate that y. The proof given in Section 3. showing that x is not an extreme point 2 of c0 1 . so x and y lie in every element of C.8. by Exercise 5. Deduce that Theorem 3. Deﬁne y and z ∈ c0 by setting yn = zn = xn for all n = n0 . Prove that (F. M) is Hausdorﬀ. the intersection of all the elements of C. transitive and antisymmetric. Moreover. Prove that every unit vector in H is an extreme point of the closed unit ball H1 .35 and 1. z ∈ H1 and t ∈ (0. 1) be such that . y ∈ C and t ∈ (0. F is nonempty. convex subset of X. partially ordered set such that every chain in F has an upper bound. so A + U is convex and Theorem 3. because M separates points. and reverse inclusion is easily seen to be reﬂexive. furthermore. as these elements are faces. The set F is closed and also nonempty: C is compact. we work with this topology from now on. M). z ∈ c0 1 and x = 1 (y + z). M)compact.1. whence A + U and B are disjoint. and the result just established is a generalisation of Theorem 3.32(ii). the set U may be taken to be convex. so every collection of closed subsets of C with the ﬁniteintersection property has nonempty intersection and C has this property because it is a chain. by Propositions 1. compact. since the continuous dual space is assumed to separate points. M)continuous. Suppose that x is a unit vector in H and let y.26) and both vector addition and scalar multiplication are σ(X.15. a partial order. M) is coarser than the original topology on X. is a nonempty. the result follows.37 together with the continuity of addition and multiplication in F.11 to A and X \ B yields an open set U containing 0 such that A + U ⊆ X \ B.33 is not required. 1 Let x ∈ c0 1 and note that there exists n0 ∈ N such that xn0  < 2 .25.12 applies with the following changes: Corollary 3. the collection of closed faces of C ordered by reverse inclusion. yn0 = xn0 − 1 and zn0 = xn0 + 1 .13. i. (Proposition 1.Solutions to Exercises Since σ(X.. Hence x and y are in F and F is a face. ⊇).40 is true for topological spaces with continuous dual spaces that separate points.e. and also gives rise to the same collection of continuous linear functionals. Exercise 5. which has the beneﬁt of being locally convex.14. Prove that the closed unit ball of c0 has no extreme points. it is immediate that this is an upper bound for C if it is a face. the sets A and B are σ(X. If x. Let H be a Hilbert space. 1. Exercise 5. Let X be a topological vector space and suppose that C is a nonempty. Hence X is a topological vector space when equipped with σ(X. Since C is trivially a face of itself. Applying Exercise 5. 103 Exercise 5. M.
. λ → x + λu is continuous. Exercise 5. If n = 1 then C is connected. and Sx . which implies that t = 1 and y. x + (1 − t) z. 1). x . so f −1 (U) is an open set containing 0. Since y ∈ C ◦ . T x 1. so is a compact interval and the claim is clear. Prove that every element of C can be written as a convex combination of at most (n + 1) elements of ∂e C. but then. Now suppose the result holds for some n and let C ⊆ Rn+1 be compact and convex. If every set of n + 1 points in C is linearly dependent then lin C is at most ndimensional (since C is a spanning set which contains no more than n linearly independent points) so C ⊆ lin C and the result follows by the inductive hypothesis. Otherwise. since x ∈ ∂e C. Then V x = x = 1. 1) then tx + (1 − t)U is open (since z → z + tx and z → (1 − t)z are homeomorphisms) and contained in C (since C is convex). there exists an open set U such that y ∈ U ⊆ C. Let S. x = z. 2 2 this contradiction give the result. As this holds for all unit vectors in H we have that S = T = V . if this set is nonempty: for every x ∈ C there exists a sequence (xn )n 1 ⊆ C ◦ such that xn → x. then tx + (1 − t)y ∈ C ◦ for all t ∈ [0. Exercise 5. 1) be such that V = tS + (1 − t)T and let x ∈ H be a unit vector. Hence there exists δ > 0 such that x ± δu ∈ U.17. there exists a linearly independent subset of C containing n + 1 points. Prove that if x ∈ C and y ∈ C ◦ . by the ﬁrst part. Then 1= x 2 Solutions to Exercises = t y. Let C be a convex subset of a topological vector space X. so V is an extreme point of B(H)1 . x  y x 1 we have equality in the CauchySchwarz inequality and so x and y are linearly dependent. so the identity V x = tSx + (1 − t)T x implies that Sx = T x = V x. Suppose for contradiction that x ∈ C ◦ ∩∂e C and let U be an open set with x ∈ U ⊆ C. it is immediate that V = 1 and so V ∈ B(H)1 . x = 1 we must have y = x. 2 As 1 =  y. the interior of C. the convex hull of this set. x = 1 (x + δu) + 1 (x − δu) =⇒ x + δu = x − δu =⇒ u = 0. Let u ∈ X be a nonzero vector and note that function f : R → X.104 x = ty + (1 − t)z. T ∈ B(H)1 and t ∈ (0. Let V ∈ B(H) be a isometry.16. x = 1 (because 1 is an extreme point in F1 ). Deduce that every isometry in B(H) is an extreme point of the closed unit ball B(H)1 . Hence x is an extreme point of H1 . so V x is a unit vector. Let C ⊆ Rn be compact and convex. since y. if t ∈ [0. [It follows that C is the sequential closure of C ◦ . as claimed.] Prove also that the interior C ◦ and the extremal boundary ∂e C are disjoint (as long as X = {0}).
For the proof of completeness we use Banach’s criterion. . furthermore. ˆ For commutativity and associativity note that the Fourier transform f → f is 1 injective on L (R).32(i) gives a continuous linear functional f such that f (c) < f (x) for all c ∈ C ◦ and so f (c) f (x) for all c ∈ C. . so the result follows by the inductive hypothesis. that ˆˆ ˆˆ f ⋆ g = f g = gf = g ⋆ f and that ˆ ˆ ˆˆ ˆˆ ˆ ˆ f ⋆ (g ⋆ h) = f g ⋆ h = f g h = f g h = f ⋆ g h = (f ⋆ g) ⋆ h by the convolution theorem. has nonempty interior. g) → f ⋆ g is a welldeﬁned. since C ◦ ∩L ⊆ (C ∩L)◦ (where the second interior is taken with respect to the relative topology on C ∩ L). λn+1) ∈ R n+1 : λ1 .Solutions to Exercises being homeomorphic to the standard n + 1simplex n+1 105 {(λ1 . this is a closed set and so C ∩ L is compact and convex.16. . (This theorem. If D := {c ∈ C : f (c) = f (x)} = (x + ker f ) ∩ C then D is a face of C which lies in an ndimensional hyperplane. see [17. . λn+1 0. We take it for granted that L1 (R) is a normed space and that (f. . it must be one endpoint of this interval. It remains to prove that this multiplication is commutative and associative and that L1 (R) is complete. As y is an extreme point of C. i=1 λi = 1}. bilinear map. if x ∈ C ◦ then let y ∈ ∂e C and consider the line L through x and y. by the previous working.1. which asserts the existence of f ⋆ g. let z be the other and note that z ∈ C \C ◦ . is an easy application of the theorems of Fubini and Tonelli. k=1 k=1 as required.15–26. . Hence f (x) = ∞ fk (x) k=1 k=1 k=1 is (absolutely) convergent almost everywhere. and therefore so does C. by Exercise 5. . If (fk )k 1 ⊆ L1 (R) ∞ is such that fk  < ∞ then the monotoneconvergence theorem applied to k=1 n ( k=1 fk )n 1 yields the convergence almost everywhere of ∞ fk (x) to an intek=1 ∞ grable function and the fact that ∞ fk  = fk .] e Solutions to Exercises 6 Exercise 6. Hence x is an interior point of the interval. so is an interval. [This result is due to Carath´odory [4]. . z is a convex combination of at most n points in ∂e C and the result follows. Theorem 3. . Prove that the vector space L1 (R) is a commutative Banach algebra when equipped with the convolution product and the norm f 1 := f . so there exists t ∈ (0.16]). Suppose now that x ∈ C \ C ◦ . and applying the dominatedconvergence theorem to ( n fk )n 1 (with dominating function ∞ fk  ) gives that f ∈ L1 (R). §§ 26. 1) such that x = ty + (1 − t)z. recall that ∂e D ⊆ ∂e C. Finally.
Suppose that there exists 1 ∈ L1 (R) such that 1 ⋆ f = f for all f ∈ L1 (R). Consideration of degree shows that G(A) = (C \ {0})1. Exercise 6. hence a homeomorphism (by Theorem 2. (f − f (∞)1)X ∞ + f (∞) f X ∞ + 2f (∞) 3 f ∞ so this map is a continuous bijection between Banach spaces. note that t → gt maps L1 (R) to itself. The continuous version of the dominatedconvergence theorem gives that gt − g 1 → 0 as t → 0 for all g ∈ C00 (R). but pn − 1 → 0 as n → ∞ and 1 ∈ G(A). locally compact space.14). so pn ∈ A \ G(A) for all n 1. Let A = C[z] denote the unital algebra of complex polynomials and let p := sup{p(α) : α 1} for all p ∈ A. Exercise 6. If A is complete then G(A) is open. Let X be a Hausdorﬀ. This gives the result. This is clearly impossible. · ) is a unital. Then 1 ⋆ fn → 1 as n → ∞ (in L1 (R) ) but (1 ⋆ fn )(x) = fn (x) → 0 as n → ∞ almost everywhere.3.2 that the map f → f X is a bijection between ˙ ˙ I := {f ∈ C(X) : f (∞) = 0} and C0 (X).2. where pn (z) := 1 + z/n. As C00 (R) is dense in L1 (R) it follows that gt − g 1 → 0 as t → 0 for all g ∈ L1 (R). .106 Prove further that this algebra is not unital. Prove that C0 (X)u . the algebra of continuous functions on X. the unitization of the algebra of continuous functions on X that vanish at inﬁnity. and it is readily veriﬁed that it is an algebra homomorphism. x → √ e− 2 n x 2π and note that fn = 1. = R R g(t − s)fn (s) − g(t)fn (s) ds dt R R gs − g(t) dt nf1 (ns) ds = R gr/n − g 1 f1 (r) dr → 0 as n → ∞. Recall from the solution of Exercise 2. the onepoint compactiﬁcation of X. where gt : s → g(s − t). is topo˙ ˙ logically isomorphic to C(X). that gt − ht 1 = g − h 1 and that For n 1 let If g ∈ L1 (R) then g ⋆ fn − g 1 gt − g 1 gt − ht 1 + ht − h 1 + h−g 1 = 2 h−g 1 + ht − h 1 . which implies that 1(x) = 0 almost everywhere. Finally. Show that (A. f → (f − f (∞)1)X + f (∞)1 is a bijection. by the dominatedconvergence theorem: the map t → gt − g 1 is bounded and f1 ∈ L1 (R). Solutions to Exercises 1 2 2 n fn : R → R. Since f − f (∞)1 ∈ I for all f ∈ C(X) the map ˙ C(X) → C0 (X)u . so A \ G(A) is not closed. Proving that · is a submultiplicative norm on A is trivial. normed algebra which is not complete.
n=0 k=0 To see this. where k=0 k=0 dk = k bl − b. for all a ∈ A there exists n ∈ N such that an = 0). ◦ by the Baire category theorem. these sets are closed and A = n∈N En so. Let A be a unital Banach algebra over C and let ea := all a ∈ A. as required..e. as n → ∞. Prove that A is uniformly nilpotent: there exists N ∈ N such that aN = 0 for all a ∈ A.Solutions to Exercises Exercise 6. and so bN = 0.4. As n ∞ n n=0 a /n! 107 for (a + b) = k=0 n 1 n k n−k a b = n! k n k=0 ak bn−k k! (n − k)! ∞ n=0 bn if a. Since e0 = 1 we have that ea e−a = 1 = e−a ea . it suﬃces to prove that if a = ∞ an and b = n=0 absolutely convergent in A then ab = ∞ cn . note that n k are k=0 l=0 al bk−l = a0 b0 + (a0 b1 + a1 b0 ) + · · · + (a0 bn + · · · + an b0 ) n n−1 n n n−k = a0 l=0 bl + a1 l=0 bl + · · · + an b0 = ak b + k=0 k=0 ak l=0 bl − b so it suﬃces to prove that rn = n ak dn−k = n an−k dk → 0 as n → ∞. for all a ∈ A. Exercise 6. (ea )−1 = e−a . b ∈ A commute. EN = ∅ for some N ∈ N. if n > N l=0 then n N n an−k dk k=0 k=0 N an−k dk + k=N +1 ∞ an−k ak k=1 dk ∞ an−k dk + ε k=0 → ε ak k=1 Deduce that ea is invertible. As ε > 0 is arbitrary we have the result. ε/ b because (a0 + tb) − a0 = t b < ε. Let ε > 0 and choose N such that dn < ε for all n N. for all a ∈ A.5. Let En = {a ∈ A : an = 0} for all n ∈ N. . Hence there exists a0 ∈ A and ε > 0 such that a − a0 < ε implies that aN = 0. i. Let A be a (nonunital) Banach algebra such that every element is nilpotent (i. Let b ∈ A \ {0} and note that p(t) := (a0 + tb)N = aN + · · · + tN bN = 0 0 ∀ t ∈ −ε/ b . Prove further that f : λ → eλa is holomorphic everywhere. Hence 0 = p(N ) (0) = N! bN . Prove that ea+b = ea eb if a and b commute. so converges. with f ′ (λ) = af (λ) = f (λ)a. where cn = n ak bn−k . Note ﬁrst that ea converges absolutely.e. where p(N ) is the Nth derivative of p..
b ∈ A. where ν(a) denotes the spectral radius of a. as required.8. Let A be a Banach algebra and suppose that (xp )p∈P .108 Let a ∈ A.7. Prove that xp yq = p∈P q∈Q (p. Exercise 6. so 0 = g ′ (0) = ag(0) − g(0)a = ab − ba. Prove that A is commutative. We prove only the ﬁrst equality. Then Solutions to Exercises eha − 1 − ha λa e(λ+h)a − eλa − aeλa = e →0 h h as h → 0.q)∈P ×Q xp yq = q∈Q p∈P xp yq . the second follows by symmetry. The function g : C → A.6. S0 ⊆ Q and T0 ⊆ P × Q such that ε xp − xp < for all ﬁnite R ⊆ P such that R ⊇ R0 1+ y 1 p∈R p∈P (where y 1 = yq − q∈Q yq ) yq < ε 1+ x 1 q∈S q∈Q for all ﬁnite S ⊆ Q such that S ⊇ S0 . Use the identity (ab)n = a(ba)n−1 b to prove that ab and ba have the same spectral radius. λ ∈ C and h ∈ C \ {0}. Hence g is constant. because g(λ) Kν(eλa be−λa ) = Kν(beλa e−λa ) = Kν(b). Exchanging the rˆles of a and b gives the result.q)∈P ×Q is absolutely summable (as xp yq xp yq ) so there exist ﬁnite sets R0 ⊆ P . . Exercise 6. Let A be a unital Banach algebra over C and let a. Suppose that there exists K > 0 such that a Kν(a) for all a ∈ A. λ → eλa be−λa is holomorphic everywhere and bounded. Note that (ab)n 1/n a = a 1/n 1/n (ba)n−1 (ba)n−1 1/n b 1/n 1/(n−1) (n−1)/n b 1/n → ν(ba) and so ν(ab) ν(ba). (yq )q∈Q ⊆ A are absolutely summable. Let ε > 0 and note that (xp yq )(p. Let A be a unital Banach algebra over C. because eha − 1 − ha h and d z e dz a z=0 ∞ n=2 hn−1 a n! n = eh a h −1 − a = a . o Exercise 6.
. As ε > 0 is arbitrary. if x. i. y ∈ X are distinct then there exists f ∈ A such that f (x) = f (y). by deﬁnition of the Gelfand topology. Furthermore.e. If R ⊆ P and S ⊆ Q are ﬁnite sets such that T = R × S ⊇ T0 ∪ (R0 × S0 ) then (p. If φ ∈ Φ(A) \ ǫ(X) then for all x ∈ X there exists f ∈ A such that f ∈ ker φ but f ∈ ker ǫx (i.. .q)∈P ×Q xp yq − (p. . so a is a homeomorphism.q)∈T + p∈R xp q∈S yq − yq + q∈Q p∈R xp − xp p∈P y 1 < ε + ε + ε. ǫx (f ) = ǫy (f ). ˆ Note that ǫ is continuous if and only if f ◦ ǫ : x → ǫx (f ) = f (x) is continuous for all f ∈ A. Solutions to Exercises 7 Exercise 7.2. . it remains to prove that ǫ is surjective. Hausdorﬀ space. Prove that the map ǫ : X → Φ(A). there exists a ∈ A such that {p(a) : p(z) ∈ C[z]} is dense in A) then Φ(A) is homeomorphic to σ(a).Solutions to Exercises (where x = xp ) and xp yq < ε (p. x → ǫx is a homeomorphism.e. By compactness there exist f1 . As X is compact and Φ(A) is Hausdorﬀ. Hence φ = ψ.. Prove that if A is a unital Banach algebra generated by a single element (i. a contradiction. by continuity and the density of {p(a) : p(z) ∈ C[z]} a ˆ in A. by Urysohn’s lemma.q)∈P ×Q 109 1 p∈P (p. we have the result.1. Exercise 7. but then f := i=1 fi fi ∈ ker φ and f (x) > 0 for all x ∈ X.q)∈T xp yq − for all ﬁnite T ⊆ P × Q such that T ⊇ T0 . f (x) = 0).q)∈P ×Q xp yq − xp p∈P q∈Q yq xp yq (p. If φ. ¯ Deduce that Φ(A) is homeomorphic to D := {z ∈ C : z 1} if A = A(D) is the disc algebra. Note that a : Φ(A) → σ(a) is a continuous function from a compact space onto a ˆ ˆ ˆ Hausdorﬀ space. . Let A = C(X). . whence ker φ = A. ǫ is a homeomorphism between X and ǫ(X).e. where ǫx (f ) = f (x) for all x ∈ X and f ∈ C(X). fn ∈ ker φ such that n n ¯ i=1 {x ∈ X : fi (x) = 0} = X. ψ ∈ Φ(A) are such that a(φ) = a(ψ) then φ(p(a)) = ψ(p(a)) for n any complex polynomial p (as φ(1) = 1 and φ(a ) = φ(a)n for all n ∈ N so φ(p(a)) = p(φ(a)) = p(ˆ(φ)) ). where X is a compact.
Exercise 7. Then g : z → (λ − z)f (z) − 1 is holomorphic in C and g(z) = 0 for all z ∈ ∂C. a. by Exercise 7. as σ(a) is closed. ﬁx λ ∈ C and choose (pn )n 1 as above. is contained in σ(a): if x ∈ C ∩ (C \ σ(a)) then there exist a sequence (xn )n 1 ⊆ C such that xn → x and C C ε > 0 such that Bε (x)∩σ(a) = ∅. 1) such ¯ that f (z) − f (rz) < ε/2 for all z ∈ D (because z − rz 1 − r). the (topological) boundary of C. contradicting the fact that g(λ) = −1.. By the maximummodulus theorem [16. f . It follows that C is a bounded region in C (i. and so 1/n = ν(a) = lim an n→∞ = lim a2 n→∞ n 1/2n = a . so g ≡ 0. so there exists a sequence of polynomials (pn )n 1 such that pn (a) → (λ1 − a)−1 . Suppose for contradiction that C \ σ(a) has a maximally connected component. Let A be a commutative. .4. If z ∈ σ(a) then. connected set): if x ∈ C then x ∈ σ(a). ¯ f − p ∞ < ε and polynomials in id : D → C. ¯ Hence so there exists a polynomial p such that p(z) − f (rz) < ε/2 for all z ∈ D. so if ε > 0 there exists r ∈ (0.110 Solutions to Exercises If f ∈ A(D) then f is uniformly continuous.2.20]. its Taylor series is uniformly convergent on D. so x ∈ C . If a2 = a 2 then a2 a ˆ ∞ n = a 2n for all n ∈ N. Hence Bε (x)∩C = ∅ and therefore C ◦ Bε (x) ⊆ C.3. there exists φ ∈ Φ(A) such that φ(a) = z. it follows that pn (z) − (λ − z)−1  = φ(pn (a) − (λ1 − a)−1 ) Deduce that the complement of σ(a) is connected. n → ∞. is continuous there and holomorphic in C. the inverse (λ1 − a)−1 is an element of A. z → z are dense in A. so x ∈ ∂C and thus x ∈ C ◦. Show there exists a sequence of polynomials (pn )n 1 such that pn (z) → (λ − z)−1 uniformly for all z ∈ σ(a). as C is maximally connected. Note ﬁrst that ∂C := C \ C ◦ . sup{pn (z) − pm (z) : z ∈ C} = sup{pn (z) − pm (z) : z ∈ ∂C} sup{pn (z) − pm (z) : z ∈ σ(a)} → 0 as m. an open. Since λ ∈ σ(a). Since z → f (rz) −1 ¯ is holomorphic on {z ∈ C : z < r }. C. pn (a) − (λ1 − a)−1 and therefore pn (z) → (λ − z)−1 uniformly on σ(a) as n → ∞. Let A be a unital Banach algebra that is generated by one element. Theorem 5. so (pn )n 1 is uniformly convergent on C and its limit. and let λ ∈ σ(a). Exercise 7. unital Banach algebra. Prove that the Gelfand transform on A is isometric if and only if a2 = a 2 for all a ∈ A. by the identity theorem.e. which is bounded (and nonempty). since φ is a unital algebra homomorphism.
Prove that A is a semisimple. Exercise 7. so convergent. Every linear functional on this image must have the form φ: a b 0 a → λa + µb ∀ a. unital Banach algebra and ﬁnd its character space. sequences in C[0. 1] and f − fn = f − fn ∞ ′ + f ′ − fn ∞ = f − fn ∞ ′ + g − fn ∞ → 0.6. let (fn )n 1 ⊆ A be Cauchy. Let A be a Banach algebra and B a semisimple. Prove that I = {f ∈ A : f (0) = f ′ (0) = 0} is a closed ideal in A such that A/I is a twodimensional algebra with onedimensional radical. Prove that if φ : A → B is a homomorphism then φ is continuous. ˆ 2 2 Hence λ ∈ σ(a ) (by the spectralmapping theorem for polynomials) and so a2 a 2 111 = λ2 = λ2  ν(a2 ) a2 . Let A = C 1 [0. f → f (0) f ′ (0) 0 f (0) as required. commutative.5. 1]. where λ. unital Banach algebra. note that Φ(A) = {εx : x ∈ X} (which may be proved as for C(X)). Hence ψ(b) = lim ψ(φ(an )) = ψ(φ(a)) n→∞ for all ψ ∈ Φ(B). 1]. so J(A) = x∈X {f ∈ A : f (x) = 0} = {0}. so A/I has one maximal ideal (viz. Exercise 7. To see that A is semisimple.1. By the closedgraph theorem. n→∞ g= 0 0 n→∞ so f ′ (t) = g(t) for all t ∈ [0. and therefore b − φ(a) ∈ J(B). it suﬃces to take a sequence (an )n 1 ⊆ A such that an → a and φ(an ) → b and prove that b = φ(a). b ∈ C. µ ∈ C. note that φ extends to Au by setting φ(α1 + a) = α + φ(a) so we may assume without loss of generality that A is unital. 1] → M2 (C). If φ(1) = 1 then λ = 1. Let ψ ∈ Φ(B) and note that ψ and ψ ◦ φ are continuous. g = ′ limn→∞ fn and note that t t t ′ lim fn = lim n→∞ 0 ′ fn = lim fn (t) − fn (0) = f (t) − f (0). which gives the result. and multiplicativity forces µ = 0.Solutions to Exercises Conversely. commutative. which implies that (fn )n 1 and ′ (fn )n 1 are Cauchy. is a continuous algebra homomorphism with kernel I and 2dimensional image. if ν(a) = a ∞ = a then there exists λ ∈ σ(a) such that λ = a . id + I). equipped with the norm f := f ∞ + f ′ ∞ . Let f = limn→∞ fn . To prove completeness. . by Proposition 6. Note that j : C 1 [0.
b. Furthermore. a 1 b 1 = m∈Z am  n∈Z bn  = = m∈Z n∈Z am  bn  am  bp−m p∈Z m∈Z p∈Z (a ⋆ b)p  = a⋆b 1 which shows that a ⋆ b ∈ ℓ1 (Z) and · 1 is submultiplicative on ℓ1 (Z).s)∈R×S xr ys for any absolutely summable (xr )r∈R . If ε > 0 choose a ﬁnite set A0 ⊆ Z such that a− an δ n n∈A 1 n∈A0 an  > a 1 − ε and note that 1 = sup n∈B\A an  : B is a ﬁnite subset of Z = a − n∈A an  < ε for any ﬁnite set A ⊆ Z such that A ⊇ A0 . a ∈ ℓ1 (Z). (To see that the reversal of order is valid. note that r∈R s∈S xr ys = (r. Note that n∈Z bn  < ∞ implies that (bn )n∈Z is bounded. Prove that a = n∈Z an δ n for all a ∈ ℓ1 (Z).7.8.) Furthermore. and so if a. bilinearity is readily veriﬁed. 1 1 = 1 and (1 ⋆ a)n = m∈Z 1m an−m = an ∀ n ∈ Z. . c ∈ ℓ1 (Z) and p ∈ Z then (a ⋆ b) ⋆ c p = n∈Z m∈Z am bn−m cp−n = m∈Z am n∈Z bn−m cp−n a ⋆ (b ⋆ c) p . Exercise 7.112 What do you notice about A and A/I? Solutions to Exercises This example shows that a quotient of a semisimple algebra need not be semisimple. Prove that the Banach space ℓ1 (Z) is a commutative. Exercise 7. (ys )s∈S in a Banach algebra. b ∈ ℓ1 (Z) . = m∈Z am r∈Z br cp−m−r = m∈Z am (b ⋆ c)p−m = If 1 ∈ ℓ1 (Z) is deﬁned by setting 10 = 1 and 1n = 0 for all n ∈ Z \ {0} then 1 ∈ ℓ1 (Z). (a ⋆ b)n = m∈Z am bn−m = p∈Z bp an−p = (b ⋆ a)n so ⋆ is commutative. unital Banach algebra when equipped with the multiplication a ⋆ b : Z → C. n → am bn−m m∈Z a. Let δ ∈ ℓ1 (Z) be such that δ1 = 1 and δn = 0 if n = 1. b ∈ ℓ1 (Z) then m∈Z am bn−m is (absolutely) summable for all n ∈ Z. and if a.
ˆ f (n) < ∞. note that this series is absolutely summable.9. Let f : T → C be continuous and such that 1 ˆ f (n) := 2π π n∈Z f (eit )e−int dt −π (n ∈ Z). Hence ˆ δ : Φ(ℓ1 (Z)) → T. z → 1/f (z) satisﬁes g n∈Z ˆ(n) < ∞. Γ(a)(λ) = an λn n∈Z 113 ∀ λ ∈ T. n ∈ Z. a → a ◦ δ −1 . where Exercise 7. if xs → x then φ(xs ) → φ(x). λ ∈ T because φ(δ) φ δ 1 =1 and φ(δ)−1  = φ(δ −1) φ δ −1 1 = 1. note that f ∈ C Φ(ℓ1 (Z)) corresponds to f ◦ δ −1 ∈ C(T) and so the Gelfand map becomes Γ : ℓ1 (Z) → C(T). so φ(a) = n∈Z an φ(δ n ) = n∈Z an λn ∀ a ∈ ℓ1 (Z). where λ = φ(δ). To see this. Conversely. ˆ ˆ i. . φ → φ(δ) is a continuous bijection from a compact space to a Hausdorﬀ space. ˆ Γ(a)(λ) = a ◦ δ −1 (λ) = a δ −1 (λ) = δ −1 (λ)(a) = ˆ ˆ ˆ ˆ n∈Z an λn . Prove that if f (z) = 0 for all z ∈ T then g = 1/f : T → C. as φ is continuous. if λ ∈ T then φ : a → n∈Z an λn deﬁnes a character. it follows by continuity that φ(a)φ(b) = φ(a ⋆ b) for all a. so a homeomorphism.. Exercise 7. ˆ Finally. b ∈ ℓ1 (Z). a ∈ ℓ1 (Z).8 implies that A = Γ ℓ1 (Z) consists of those g ∈ C(T) such that g(z) = n∈Z an z n (z ∈ T) and n∈Z an  < ∞.e. φ is linear and bounded (because φ(a) a 1) and φ(δ m ⋆ δ n ) = φ(δ m+n ) = λm+n = λm λn = φ(δ m )φ(δ n ) ∀ m.Solutions to Exercises Deduce that the character space of ℓ1 (Z) is homeomorphic to T := {z ∈ C : z = 1} and (with this identiﬁcation) the Gelfand transform on ℓ1 (Z) is the map Γ : ℓ1 (Z) → C(T). Let φ ∈ Φ(ℓ1 (Z)) and note that φ(δ n ) = φ(δ)n for all n ∈ Z. Furthermore.
˙ where D is the onepoint compactiﬁcation of D. f → j(f ) is a normdecreasing algebra homomorphism (as j(f ) ∞ = f T ∞ f ∞ ). For n that fn T = gn T and note that gn − gm ∞ 1 let gn ∈ A(D) be such fn − fm ∞ = (gn − gm )T 1 ∞ = (fn − fm )T ∞ →0 as m. (iii) Prove that ˙ is an isometric isomorphism. Thus to every f ∈ A there exists a unique j(f ) ∈ A(D) such that f T = j(f )T .e. Hence 1/f ∈ A.. i. 0 ∈ σ(f ). if g. Suppose that f ∈ A. I = {f ∈ A : f T = 0} = ker j is a closed ideal. Let T = ∂D = {z ∈ C : z = 1} and A = {f ∈ C(D) : f T ∈ A(D)T }: ¯ that agree on the unit A consists of those continuous functions on the closed unit disc D ¯ circle T with a continuous function on D that is holomorphic in D. if f ∈ A then f − j(f ) ∈ I so f ∈ I + A(D) and if f ∈ A(D) ∩ I then f ∞ = f T ∞ = 0.114 Since Solutions to Exercises an z n is uniformly convergent on T for such g we see that g (m) = ˆ n∈Z n∈Z an 2π π eint e−imt dt = am −π (m ∈ Z) and so A consists of those g ∈ C(T) such that n∈Z ˆ(n) < ∞. In particular. n → ∞. (ii) Prove that I = {f ∈ A : f T = 0} is a closed ideal in A and that A = A(D) ⊕ I. f ∈ A g and the condition f (z) = 0 for all z ∈ T is equivalent to the fact that φ(f ) = 0 for all φ ∈ Φ(A). we see that f ∈ A. as required.20] will be useful: if f ∈ A(D) then f ∞ := sup{f (z) : z 1} = sup{f (z) : z = 1} =: f T ∞ .10. i : I → C0 (D). ¯ Let (fn )n 1 ⊆ A be such that fn → f ∈ C(D). Since n→∞ n→∞ gT = lim gn T = lim fn T = f T . Furthermore. [A corollary of the maximummodulus theorem [16. Deduce that I u is topologically isomorphic to C(D). In particular.] (i) Show that A is a Banach algebra when equipped with the supremum norm. as required. Theorem 5. so (gn )n converges to g ∈ A(D). h ∈ A(D) are such that f T = gT = hT then g−h ∞ = (g − h)T ∞ = 0 and so g = h. thus A = A(D) ⊕ I. the map j : A → A(D). ¯ Exercise 7. f → f D .
¯ f (z) (z ∈ D2 ). w ∈ D. since 0 ∈ U. T is continuous if f ◦ T is continuous for all f ∈ A. T : S 2 → Φ(A). and since ˆ f ◦ T (z) = ¯ j(f )(z) (z ∈ D1 ). 1). This shows that g is continuous. and since φ is nonzero. The deduction follows immediately from Exercise 6. z → ¯ εz (z ∈ D2 ) and prove that this is a welldeﬁned. From this we see that f −1 (U) ⊇ D \ Br [0] and therefore C ¯ g −1(U) = f −1 (U) ∪ D \ Br [0] is open. g ∈ A are such that j(f ) = j(g) then f − g ∈ I so φ(f ) = φ(g). ¯ Deduce that φ = εz ◦ j for some z ∈ D (where εz : A(D) → C. Hence T is an injection. w ∈ D. as ¯ by the solution to Exercise 7. ˜ ˜ (iv) Prove that if φ ∈ Φ(A) is such that ker φ ⊇ I then φ = φ ◦ j. hence f D ∈ C0 (D). where φ ∈ Φ(A(D)). this follows because f T = j(f )T . and therefore of ¯ D. Since the identity function lies in A(D).2. Hence ˜ φ : A(D) → C. note that there exists f ∈ I such that f (z) = 1 (by Urysohn’s lemma) and so εz (f ) = 1 = 0 = (εw ◦ j)(f ). Finally. it remains to prove that i is surjective. Φ(A(D)) = {εz : z ∈ D} ¯ ¯ ¯ ¯ (v) Let D1 and D2 be two copies of the unit disc and let S 2 = D1 ∪ D2 ∼ be the sphere obtained by identifying each point on T1 = ∂D1 with the corresponding point on T2 = ∂D2 . Then g −1(U) = f −1 (U) ∪ T and. f → f (z) ). ¯ Let f ∈ C0 (D) and deﬁne g : D → C by setting gD = f and gT = 0.Solutions to Exercises If f ∈ I and ε > 0 then 115 ¯ is a closed subset of D and is therefore compact. Since f D ∞ = f ∞ the map i is isometric. Note that if f .2. As Φ(A) has the weak* topology. . continuous injection. if 0 ∈ U then g −1(U) = f −1 (U) is an open subset of D. If z ∈ T1 ∪ T2 then εz (j(f )) = j(f )(z) = f (z) = εz (f ) for all f ∈ A. The deduction is immediate. as required. there C C C exists ε > 0 such that Bε (0) ⊆ U. so this is ˆ a good deﬁnition. It is immediate that φ is an algebra homomorphism such that ˜ ˜ φ = φ ◦ j. hence C C D \ f −1 (U) ⊆ D \ f −1 Bε (0) ⊆ Br [0] {z ∈ D : f (z) ¯ ε} = {z ∈ D : f (z) ε} C for some r ∈ (0. to see that εz = εw ◦ j for all z. so suppose that 0 ∈ U. Thus f −1 (U) ⊇ f −1 Bε (0) and f −1 Bε (0) has compact complement. so is φ. Let U ⊆ C be open. Deﬁne ¯ εz ◦ j (z ∈ D1 ). εz = εw ¯ and εz ◦ j = εw ◦ j for all distinct z. j(f ) → φ(f ) ˜ is well deﬁned.
z = ∞ and φI = εz for some z ∈ D. Choose f ∈ I such that f (z) = 1 (such exists by Urysohn’s lemma) and note that f (g − φ(g)1) ∈ I for all g ∈ A. Hence T is surjective and we have the result (since Φ(A) and S 2 are both compact. Deﬁne x ∈ ℓ∞ by setting −1 l = n2k−1 . . Otherwise φI = 0. φ(g) = εz (g). Hence the existence of a convergent subnet follows from the (generalised) BolzanoWeierstrass theorem. Suppose for contradiction that (δn ) has a weak*convergent subsequence. Solutions to Exercises ¯ If φ ∈ Φ(A) is such that ker φ ⊇ I then φ = εz ◦ j for some z ∈ D. 1.1. note that ˙ φ : I u → C. Let X = ℓ∞ and for all n ∈ N deﬁne δn ∈ X ∗ by setting δn (xk )k 1 = xn . Prove that (δn )n 1 has no weak*convergent subsequence but that (δn )n 1 has a weak*convergent subnet.e.22). say (δnk )k 1 . Hausdorﬀ spaces).) is not convergent. 1. ∗ For the second claim. . Theorem A. i. . which is weak* compact by the BanachAlaoglu theorem (Theorem 3. Since = ˙ φI = φI is nonzero.116 (vi) Prove that Φ(A) is homeomorphic to the sphere S 2 . −1. Solutions to Exercises A Exercise A. the desired contradiction. note that (δn )n 1 ⊆ X1 . 1 l = n2k . Then δnk (xl )l 1 k 1 = (−1. f + α1 → φ(f ) + α ˙ ˙ is a character of I u ∼ C(D) and so has the form εz for some z ∈ D. so g(z) − φ(g) = εz f (g − φ(g)1) = φ f (g − φ(g)1) = φ(f )φ(g − φ(g)1) = 0. . xl = 0 otherwise..5.
Partial orderings and MooreSmith limits. 2002. 111 (1952). Birkhoff. Serie I 32. Introductory functional analysis with applications. [Cited on page 105.] [5] E.] [9] J.. Providence.] [10] E. [Cited on page 4. New Jersey. 2000. [Cited on pages 18 and 31. Ringrose. Inc. [Cited on page 91.] 117 .] [12] E. [Cited on page 97. Acta Mathematica 123. Uber den Variabilit¨tsbereich der Fourierschen Konstanten e a von positiven harmonischen Funktionen. Linear algebra. Subalgebras of C ∗ algebras. John Wiley & Sons. London Mathematical Society Monographs 23. Rendiconti del Circolo Mathematico di Palermo. R.] [7] K. Kreyszig.Bibliography [1] W. second edition. [Cited on page 13. John Wiley & Sons. Cambridge University Press. Graduate Studies in Mathematics 15. Inc. New York. McShane. Inc.. A. Van Nostrand Co. Annals of Mathematics 38. 35 and 39. Oxford. 193–217 (1911). MooreSmith convergence in general topology. [Cited on page 61.] [2] G. [Cited on page 13. Cambridge. Bollobas. Ruan.. Oxford University Press. Princeton.] ¨ [4] C.] [8] R. A projection of norm one may not exist. Linear analysis. Lax.] [11] P. Functional analysis.. 1955. Kadison & J. [Cited on pages 13 and 24. V. Kelley. Goodner. New Jersey.. 1992. [Cited on page 41. Hoffman & R. Effros & Z. D. Kunze.J. Rhode Island. 1971. B. corrected printing. Prentice Hall. Arveson. 334–335 (1964). Inc. New York. Mathematics Magazine 73. L. 1978. 39–56 (1937). Carath´odory. [Cited on pages 24.] [6] D. Volume I: elementary theory.] ´ [3] B. Fundamentals of the theory of operator algebras. American Mathematical Society. [Cited on page 68. General topology. 1997. 141–224 (1969). American Mathematical Monthly 59. Operator spaces. M.
767776 (1935). Simmons. [Cited on page 5. 24. 4. F.] [20] R. London Mathematical Society Lecture Noes Series 294. Introduction to Operator Space Theory. [Cited on pages 6. 110 and 114. 73. Cambridge University Press. New York.] . SpringerVerlag. 24 and 32. Analysis now. K. [Cited on pages 24. 1987. Expositiones Mathematicae 16. Oxford University Press. University of Oxford.] [25] G. Introduction to topology and modern analysis. Introduction to metric and topological spaces. Princeton University Press. 14. Oxford. A. A. 11. McGrawHill. [Cited on pages 5. Princeton.118 Solutions to Exercises [13] E. 1990. 1997. 3. Cambridge. VincentSmith. Oxford University Press. Real and complex analysis. A general theory of limits. Functional analysis. Inc. Mathematische Annalen 111. McGrawHill. F. Moore & H. [Cited on pages 26. Weir. 18 and 37. 18. Oxford. Cambridge. 1991. Priestley. third edition. B4: Analysis. McGrawHill. [Cited on page 61.] [21] G. 2003. Schatten. Annals of Mathematics Studies 26. [Cited on pages 13. corrected printing.] [26] A. Oxford. Rudin.] [18] W. J.] [17] H. 383–384 (1998). Pedersen. 68 and 94. 57.] [19] W. Introduction to integration.. Mathematical Institute notes.] [15] G.] [24] G. New York. [Cited on pages iii. Smith..] [22] W. [Cited on page 97. An elementary analytical proof of the spectral radius formula. F. Graduate Texts in Mathematics 118. Tychonoff. Inc. 1991. revised edition. 102–121 (1922). 7. [Cited on page 13.] [14] G. second edition. Rudin. New York. Lebesgue integration and measure. 1995. 1973. 56 and 105. VincentSmith. Sutherland. 1950. 26 and 39. Ein Fixpunktsatz. [Cited on page 41.] [23] A. Priestley. Oxford. 1963.] [16] H. [Cited on page 100. 1981. [Cited on pages 11. 72. American Journal of Mathematics 44. A theory of crossspaces. A. H. Introduction to complex analysis.. L. New York. [Cited on pages 31 and 56. Inc. revised printing. Oxford University Press. 36. 26 and 35. Pisier. Cambridge University Press.
. . . . . . . . . . . . . 7 ∂e C . 17 B(X. . . . . 33 . . . . . . . 48 ball closed . . . chain . . . . . . . . . . 9 . . . . . closed convex hull . . . . norm . 75 53 53 54 68 54 55 54 54 37 37 56 BohnenblustSobczyk theorem . . . . . . compactiﬁcation onepoint . . . . . . second . . . . . . .. . . . . . . 35 dense . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . normed . . . canonical embedding category ﬁrst . quotient . . . . . . . . . . . 38 B B(X) . . . . 10 dual space . . . . . . 19 direct sum . . C∗ . commutant . . . . . . . . . . . . . . . . . . . . . . 35 Baire category theorem . . . . . . . . . . . annihilator . . . double . . . nowhere . . . C C ∗ algebra . . . . . . . . 19 direct product . . . . . . . . . . . . .. . . . C00 (X. 18. . . . . . . . . . 3 Banach algebra . . . 4 algebraic . . . 38 . . . . . . . . . closed ball . . . . . . . . 7 . 3 BanachAlaoglu theorem 40 BanachSteinhaus theorem . . . . . . 23 37 57 35 19 13 11 43 43 45 56 45 20 19 . compact support . 26 double commutant 45 dual 45 second . . . . . . . . . . . . . . . . . . 7 maximal . . . 19 disc algebra . . . . . centraliser . . . . . . . . . . 19 Banach isomorphism theorem 22 space . . . . . . . . . . . . . . . . . . . . convex . . . . . 54 Banach’s criterion . . . . Y ) . . . . .. . . . . . . . . . . . . closedgraph theorem cnv A . . . . . . 7 . . 68 . . . . . . . . . approximate identity . . . . . . . . . . algebra homomorphism algebra isomorphism . . . . . . . . . . . . . . 5 . . 32 . . .Index Numbers set in italic type refer to the page where the term is ﬁrst deﬁned. existence of . . . . . . 20 balanced . . . . . . . . . . . . . . . . bound upper . . . . . . . . . . . . weakly . . . . . . . . . 27 bicommutant . . . . . . . A accumulation point . 62 E . . . . 45 dominated . . C0 (X. . . . 17 completeness of . . . . . . . . . . . . . . . . . . . . of a net . . convergence locally uniform . . . . . . . . . . . 7 element . . . . . . . . . . uniqueness of . . . . . . . . . . convex hull . . . . . . . . . . . . . . . . . . . . cocountable topology codimension . . . . . . . . . . . . . . 19 open . . . . . . . . . . . . . . . . . . . . . 6 D . . 62 algebraic . . . pre. . . 45 . cnv A . . . . . . . . . . . . . . compact locally . . . . . . . . . . . .. . . . . . . . . . . 65 directed set . . . . . . . . boundary extremal . . . . . . . . completion . . . . . . . 32 119 . . convolution . . . . . . . . . . E) . . . .. bounded essentially . . . . . . . . Banach . .. . . . . . . . . . . . . . . . . 32 pnorm . . . . . . . . 62 topological . associative . . . . . . . . . . . . . . . . . . . complemented subspace . . . algebra . . 18 . . . .. . . . . . . . locally . . 35 BLT theorem . . . . . . . character . . . . . 8 11 69 33 62 35 31 31 31 . . . . . . 6. . . . E) . . . . . . . 62 bidual . . . .
. . . .. . . . . . . 3 resolvent . . . 32 topological . . operator . . . . . . . .. . . . . . . . . . .. . . 17 17. 6 open ball . . . . . . . . invertible . . . . . . F face . . . hull convex . . . . . 37 preorder . . 38 Jacobson . . . . 57 11. . . . . . . . 6 nowhere dense . weakly . . . . . 55 quotient . . 6 normed vector space . . . . . . . 60 . . . . . . . . . . initial topology . . . 46 order partial . . 54 net . . . . . . . . . . . . . . . 24 radius . . . . . . . . 21 L functional limit . . . homomorphism algebra . . . . evaluation . . . . .120 essentially bounded . . 11 P . . . . . . . . . . linear. 57 operator space . . . 31 locally compact .. . . . . . . . 7 . Gelfand Gelfand graph . . . . 13 extreme . . . . . . . 41 . . . . 23 . 43 partial order . . . . . . . . . . . .. . .. . . . 4 . . . . . .. 21 . . 5 . 54 spectral . 75 . 7 pintegrable . 9 normed algebra . . . . . . . 43 product topology . . . . . . . . . . . . 5 . . . . . 26. ﬁrst category . . 35 . . . . 36 . . . . . . universal . 11 quotient algebra . . 32 locally uniform convergence . 76 quotient seminorm . . . . . . . . 5 isomorphism evaluation homomorphism 69 algebra . closed . . . . . . . . . 5 R . . . . . 39 . . . . . . . Minkowski . . . . 21 openmapping theorem . . . . . . . . . . . . . . . . 45 norm . . . . . 19 resolvent set . . . .. . .. . . . . maximal ideal . . . . . . . . . . . . . . . . . . . . . . . . . . 23 . 56 18. . identity approximate . 26 maximal element . . . . . . . . . . . . 14 quotient norm . G G(T ) . . . . 32 preannihilator . 10 . 45 Poisson kernel . . . . . . eventually . . . . . . . . . . 17 . . . . . . . . . . . . . . . 68 map quotient . . . . . . 21 linear. . 32 point accumulation . . 75 isometric . . 45 nonmeagre . sublinear . . . . . 39 . . . . . . . . convex. . . . . . . . . . . .. . . . . . . .. . 3 completeness of . .. . . . 11 quotient map . . . . . . . . . . . . . . . . . . . . 19 quotient space . .. . .. . . . . . . . . .. . . . . . . . . .. . . . 61 . . . . property . 59 normed space . . . . . 26 M 43 67 majorant . . . . . 5 Q quasinilpotent . . . . bounded . . . . norm bounded . . . . .. . . . . 67 openmapping lemma . . . . 54 onepoint compactiﬁcation . . . . . . . . . 31 Liouville’s theorem . . . . . . . . 75 KreinMilman theorem function open . . . . 3 reﬂexive . . . . . . . 4 radical . . . . . . 19 open function . . . . . . . . . . . . . . . . . . . . . . . . . 19 Poisson . . .. . . . bounded . . . 67 . . . . 17 quotient topology . 45 J extreme point . .. . . . . . . . . . . . . . . . . . . . . . .. 60 . . . . . . . . . . . . . . . . . . 54 operator norm . . . .. . . . . O Index H HahnBanach theorem holomorphic strongly . . . . . . . . topology transform . . . . . . . 39 kernel . Minkowski functional . . . . 32 pre . . 10 . . . . . 65 supremum . 36 Minkowski’s inequality . .. . . . frequently . . 36 N . gauge . . . 54 normal . . . . . . . .. . . . . . . product . . isometric isomorphism isometry . . image . I ideal . 26 . linear . . . . . .. . . 5 . . . . . . . . . . . . 69 convergent . . . . .. . . extremal boundary . .. . . . . . . maximal . . . . . 55 . 19 . . . . . . . . 67 . 43 locally convex . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . 14 . . . 45 Jacobson radical . . . . . . . 65 product norm . . . . 14 product direct . . . . 22 operator . . . . . . . . . . . . . . . . . 26 . . . . 32 meagre . .. . . . . . . . . 45 K ﬁniteintersection kdense . . . . .
3 42 . . . . . 17 initial . . . . . . sum direct . . . . seminorm . . . 39 weakly bounded . . . . . . . . . . . 36 T weak topology . . . . . . . 121 41 32 37 19 35 . . . sequence spaces . . . . . . . . . . . . . . . . . normed . . . . . . . . . . . . . . . . . 8 Urysohn’s lemma . σ(X. topological . . . . . . . . . . strong operator topology subalgebra . . topological . . second category . . . . . . . . . . . . . . . . . operator . . σ(X ∗ . . 37 53 trigonometric polynomials 57 54 Tychonov’s theorem . . . . direct. . . separates points . . . . . . . . . . . algebraic . . . 3 41 . . . . . . . . 4 weak operator topology . . . . . vector. 24 . . . 25 weak* topology . . . 67 Xr . . sum of . . . . . . 10 X Gelfand . . . . . . . weak . . . X ∗ ) . . . . . . . . . . . . . . . . . . 26 Zorn’s lemma . 76 . . . . . . . . . 36 product . 32 . . . . . . 27 Z operator. . . . . set balanced . 6 . . 4 . vector. 38 topological vector space . . X) . . direct. 9. . . . . . . . . . . . . . . . . . 3 48 11 31 31 . 3 . . . . . . 7 V 13 vanish at inﬁnity . . . subspace . . . . . second dual . 61 quotient . . . . . . . . . . series absolutely convergent convergent . . 9 . . . 40 75 U . . . . . . . . . algebraic . quotient . space dual. semisimple . 9. . . . . pnorm . support compact . . . subnet . complemented . . . . . . . . . . . . . . . . . . 32 spectral radius . . . . . . . . . . directed . . . . . . . . . . . . . . . . . . . 11. .Index S σ(X. . . strong . . . . . . . . . . . . . . . . . . . . . . . . . 31 operator. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . summable . spectrum . . . . . . . . . . . . . . . . . . . . . normed . . . . . . . 4 37 unitization . . . quotient . . . supremum norm . . . . . . 31 27 weak* . . . . . . 3 . . . . . . 5 67 10 41 . 3 . . . M ) . . 37 topological product . 55 universal net . 31 Tietze’s theorem . . . . . . . . . . . . 9 X ∗ . . . . . . . . 42 Wiener’s lemma . . dual. . . . . . . . . . . . . . . . . . 9 upper bound . . . 73 topology cocountable . . . . 14 60 weak . . . . . . submultiplicative . 7 W . . separating . .