Lecture Notes in

Mathematics
Edited by A. Dold and B. Eckmann

936

S.M. Khaleelulla

Counterexamples in
Topological Vector Spaces

Springer-Verlag
Berlin Heidelberg New York 1982
Author
S.M. Khaleelulla
Department of Mathematics
Faculty of Science, King Abdulaziz University
P.O. Box 9028, Jeddah, Saudi Arabia

AMS Subject Classifications (1980): 46 A 05, 46 A 06, 46 A 07, 46 A 09,

46A14, 46A 25, 46A35, 46A40, 46B05, 46B15, 46B 30, 46C05,
46H 05, 46J 20

ISBN 3-540-11565-X Springer-Vertag Berlin Heidelberg New York

ISBN 0-887-11565-X Springer-Verlag New York Heidelberg Berlin

This work is subject to copyright. All rights are reserved, whether the whole or
part of the material is concerned, specifically those of translation, reprinting,
re-use of illustrations, broadcasting, reproduction by photocopying machine or
similar means, and storage in data banks. Under w 54 of the German Copyright
Law where copies are made for other than private use, a fee is payable to
"Verwertungsgesellschaft Wort", Munich.
9 by Springer-Verlag Berlin Heidelberg 1982
Printed in Germany
Printing and binding: Beltz Offsetdruck, Hemsbach/Bergstr.
2141/3140-543210
TO

PROFESSOR 6ALAL M. EL-SAYYAD

 PREFACE

During the last three decades much progress has been

made in the field of topological vector spaces. Many genera-

lizations have been introduced; this was, to a certain

extent, due to the curiosity of studying topological vector

spaces for which a known theorem of Functional analysis can

be proved. To justify that a class C I of topological vector

spaces is a proper generalization of another class C2 of

topological vector spaces, it is necessary to construct an

example of a topological vector space belonging to CI but

not to C 2 ; such an example is called a counterexample. In

this book the author has attempted to present such counter-

examples in topological vector spaces, ordered topological

vector spaces, topological bases and topological algebras.

The author makes no claim to completeness, obviously

because of the vastness of the subject. He makes no attempt

to give due recognition to the authorship of most of the

counterexamples presented in this book.

It is assumed that the reader is familiar with general

topology. The reader may refer s BE1 ~ for information

about general topology.

To facilitate the reading of this book, some funda-

mental concepts in vector spaces and ordered vector spaces

have been collected in the Chapter called 'Prerequisites'.

Thereafter each Chapter begins with an introduction which

presents the relevent definitions and statements of theorems

and propositions with references where their proofs can be

 VI

found. For some counterexamples which require long and

complicated proofs, only reference has been made to the

literature where they are available.

The books and papers are listed separately in the

bibliography at the end of the book. Any reference to a

book is indicated by writing B[ ] and to a paper by P [ ] .

The author would like to express his deep gratitude to

Professor T. Husain, McMaster University, Hamilton, Canada,

and Dr. I. Tweddle, University of Stirling, Stirling,

Scotland, who have given him both moral and material support

during the preparation of this book. The author wishes to

thank Mr. Mohammed Yousufuddin for typing the manuscript.

The author takes great pleasure in thanking the editors

and the staff of Springer's "Lecture Notes in Mathematics"

series for their keen interest in the publication of this

book.

S.M. Khaleelulla

Department of Mathematics
Faculty of Science
King Abdulaziz University
Jeddah, Saudi Arabia
 CONTENTS

PREREQUISITES

I TOPOLOGICAL V E C T O R SPACES 9

Introduction 9

I. A topology on a vector space, w h i c h is not compa-

tible w i t h the vector space structure. 12

2. A topological vector space w h i c h is not a locally

semi-convex space. 12

3. A locally b o u n d e d (and hence a locally semi-convex)

space w h i c h is not a locally convex space. 13

4. A locally convex space w h i c h is not a locally b o u n d e d

space. 14

5. A locally semi-convex space w h i c h is n e i t h e r locally

convex nor locally bounded.
- A m e t r i z a b l e topological vector space w h i c h is not
locally bounded. 14

6. A topolog~aivector space on w h i c h there exist no non-

trivial continuous linear functionals. 14

7. A topological vector space such that no finite-

dimensional subspace has a topological complement
in it. 15

8. Two closed subspaces of a topological vector space,

whose sum is not closed. 16

9. A topological vector space in w h i c h the convex

envelope of a p r e c o m p a c t set is not p r e c o m p a c t (not
even bounded). 16

i0. A bounded linear map from a topological vector space

to a topological vector space, w h i c h is not conti-
nuous. 17

II LOCALLY CONVEX SPACES 18

Introduction 18

i. A locally convex space w h i c h is not metrizable. 21

 Vlll

2. A metrizable topological vector space which is not

locally convex. 21

3. A sequentially complete locally convex space which

is not quasi-complete. 21

4. A quasi-complete locally convex space which is not

complete. 22

5. A complete locally convex space which is not

B-complete. 22

6. A complete locally convex space which is not metri-

zable. 23

7. A normed space (and hence a m e t r i z a b l e locally

convex space) which is not complete. 24

8. A locally convex space which contains a closed,

circled and convex set with no extreme points. 24

9. A topological vector space which contains a compact

convex set with no extreme points. 25

IO. A weakly compact set in a locally convex space,

whose weakly closed envelope is not weakly compact. 25

II. A bounded sequence in a topological vector space,

which is not convergent. 26

III SPECIAL CLASSES OF LOCALLY CONVEX SPACES 27

Introduction 27

I. An inner product (a pre-Hilbert) space which is not

a Hilbert space. 34

2. A generalized inner product space which is not an

inner product space. 34

3. A semi-inner product space which is not an inner

product space. 34

4. A generalized semi-inner product space which is

neither a semi-inner product space nor a genera-
lized inner product space. 35

5. A Banach space which is not a Hilbert space. 36

6. A Banach space which is not separable. 36

7. A Banach space which is not reflexive. 36

 IX

8. A Frechet space which is not a Banach space. 37

9. A t-polar space which is not B-complete.

- A t-polar space which is not bar~elled. 37

IO. A bar./led space which is not complete.

- A barrelled space which is not a Frechet space. 37

II. A barrelled space which is not metrizable.

A barrelled space which is not a Frechet space. 38

12. A Baire-like space which is not an unordered

Baire-like space. 38

13. A Baire-like space which is not a Baire space.

- A barrelled space which is not a Baire space. 38

14. A barrelled bornological space which is not the

inductive limit of Banach spaces. 39

15. A bornological space which is not metrizable. 40

16. A bornological space which is not barrelled. 40

17. A barrelled space which is not bornological. 40

18. A quasi-barrelled space which is neither barrelled

nor bornological. 41

19. A quasi-M-barrelled space which is not quasi-

barrelled. 41

20. A semi-bornological space which is not an S-bornolo-

gical space (and hence not a bornological space). 41

21. An S-bornological space which is not C-sequential

(and hence not a bornological space). 42

22. A C-sequential locally convex space which is not

S-bornological (and hence not bornological). 43

23. A Mackey space which is not quasi-barrelled. 44

24. A Mackey space which does not have property (S). 44

25. A Mackey space with property (S) but without

property (C). 44

26. A semi-reflexive space which is not reflexive.

- A Mackey and semi-reflexive space which is not
reflexive.
- A semi-reflexive space which is not quasi-barrelled.
 X

-- A complete locally convex space which is not

quasi-barrelled.
- A topological projective limit of barrelled spaces,
which is not quasi-barrelled. 45

27. A barrelled space which is not a Montel space. 45

28. A reflexive space which is not a Montel space. 46

29. A Frechet space which is not a Schwartz space. 46

30. A Schwartz space which is not a Montel space. 46

31. A Montel space which is not separable. 47

32. A Montel space (and hence a reflexive locally convex

space) which is not complete.
- A M ontel (and hence barrelled) space which is not
a Frechet space. 48

33. A distinguished space which is not semi-reflexive. 49

34. A Frechet space which is not distinguished.

-A barrelled space whose strong dual is not
barrelled (not even quasi-barrelled).
-A bornological space whose strong dual is not
bornological. 49

35. A distinguished space whose strong dual is not

separable. 50
36. A distinguished space whose strong dual is not
metrizable. 50

37. A distinguished space which is not quasi-barrelled.

- A semi-reflexive space which is not quasi-barrelled
- A Mackey space which is not quasi-barrelled.
- A semi-reflexive space whose strong dual is not
semi-reflexive. 51

38. A bornological space whose strong bidual is not

bornological. 51

39. An ( L B ) - s p a c e w h i c h i s n o t q u a s i - c o m p l e t e . 52

40. A locally convex space which is not reflexive (not

even semi-reflexive) but its strong dual is ref-
lexive. 53

41. A countably barrelled space which is not barrelled.

 •

- A countably quasi-barrelled space which is not

quasi-barrelled.
- A locally convex space C(X) of continuous func-
tions, which is not a Mackey space.
- A complete locally convex space which is not
barrelled. 53

42. A locally c o n v e x s p a c e C(X) o f c o n t i n u o u s functions

which is not countably barrelled. 54

43. A semi-reflexive countably barrelled space which is

not barrelled. 54

44. A countably quasi-barrelled (and hence o-quasi-

barrelled) space which is not o-barrelled.
- A countably quasi-barrelled space which is not
countably barrelled. 55

45. A o-barrelled space which is not a Mackey s p a c e . 55

45(a). A o-barrelled space which is not countably quasi-

barrelled (and hence n o t countably barrelled). 57

46. A Mackey space which is not o-quasi-barrelled. 57

47. A locally convex space which has property (C)~ but

is not o-barrelled. 58

48. A sequentially barrelled space which is not o-quasi-

barrelled (and hence not o-barrelled).
- A Mackey space which is sequentially barrelled
but not o-quasi-barrelled.
- A separable sequentially barrelled space which is
not barrelled.
- A sequentially barrelled space which has property
(S) but not property (C). 58

49. A sequentially barrelled space which does not have

property (S).
- A sequentially barrelled space which is not
o-barrelled. 59

50. A quasi-complete locally convex space which is not

sequentially barrelled. 59

51. A (DF)-space which is not countably barrelled. 60

52. A (DF)-space which is not quasi-barrelled. 60

 xII

53. A quasibarrelled (DF)-space which is not borno-

logical. 60

54. A locally topological space which is neither a

bornological space nor a (DF) space. 61

55. A k-quasi-barrelled space which is not k-barrelled. 62

56. An H-space which is not a distinguished space. 62

57. An H-space which is not metrizable. 63

58. An H - s p a c e whose strong dual is not separable. 63

OPEN PROBLEMS 63

IV SPECIAL CLASSES OF TOPOLOGICAL VECTOR SPACES 65

Introduction 65

1. A topological vector space in which the filter condi-

tion holds but not the closed neighbourhood condition 68

2. An N-S s p a c e which is not an L-W s p a c e . 69

3. A locally convex space C(X) of continuous functions,

which is barrelled and b o r n o l o g i c a l but not W-
barrelled. 69

4. An u l t r a b a r r e l which is not convex and which does

not have a defining sequence of convex sets. 69

5. An u l t r a b a r r e l l e d space which is net barrelled. 69

6. A barrelled space which is not ultrabarrelled. 70

7. An ~ ~ set which is not u-compact. 70

8. An u l t r a b a r r e l l e d space which is not non-meagre. 71

9. An u l t r a b o r n o l o g i c a l space which is not bornological. 71

10. A bornological space which is not ultrabornological. 72

11. An u l t r a b o r n o l o g i c a l space which is not ultra-

barrelled. 72

12. An u l t r a b a r r e l l e d space which is not ultraborno-

logical. 72

13. A quasi-ultrabarrelled space which is neither ultra-

barrelled nor ultrabornological. 73

14. A countably quasi-ultrabarrelled space which is not

countably ultrabarrelled. 73
 Xlll

15. A countably ultrabarrelled space which is not

ultrabarrelled.
- A countably quasi-ultrabarrelled space which is
not quasi-ultrabarrelled. 73

16. A countably barrelled space which is not countably

ultrabarrelled.
- A countably quasi-barrelled space which is not
countably quasi-ultrabarrelled. 73

17. A k-quasi-ultrabarrelled space which is not k-ultra-

barrelled. 74

18. A hyperbarrelled space which is not hyperborno-

logical. 75

19. A hyperbornological space which is not hyper-

barrelled. 75

20. A quasi-hyperbarrelled space which is neither hyper-

barrelled nor hyperbornological. 75

21. An jY-quasi-hyperbarrelled space which is not ~ -

hyperbarrelled. 75

22. A barrelled space which is not j~-hyperbarrelled. 75

V ORDERED TOPOLOGICAL VECTOR SPACES 77

Introduction 77

I. An ordered topological vector space with generating

cone which does not give open decomposition. 85

2. An ordered topological vector space with normal cone

but with a (topologically) bounded set which is not
order-bounded. 85

3. A cone in a topological vector space, which is not

normal. 86

4. An ordered topological vector space in which order

bounded sets are bounded but the cone is not normal.86

5. A cone in a topological vector space, which has no

interior points. 87

6. An element of a cone in a vector space, which is

an interior point for one topology but not for
another topology. 87
 XIV

7. A cone in a locally convex space, which is not a

b-cone. 88

8. A base of a cone in a topological vector space,

which is not closed. 89

9. An ordered normed space which is not an order-unit

normed space though its dual is a base normed space. 89

i0. An ordered topological vector space which is

complete but not order-complete. 90

Ii. An ordered topological vector space which is order-

complete but not complete. 90

12. An ordered topological vector space which is

complete and order-complete but not boundedly 90
order-complete.

13. An order-continuous linear functional on an

ordered topological vector space, which is not
continuous. 91

14. A continuous linear operator on an ordered topo-

logical vector space, which is not sequentially
order-continuous. 91

15. A positive linear functional on an ordered topo-

logical vector space, which is not continuous. 93

16. An ordered topological vector space on which

there exist no non-zero positive linear
functionals. 93

17. A topological vector lattice which has no non-

zero real lattice homomorphisms. 94

18. A topological vector space with lattice ordering

§
in which the map x+x is continuous for all x
but not uniformly continuous. 94

19. An ordered locally convex space with a positive

weakly convergent sequence which is not convergent. 95

20. An M-space which is not normable. 96

21. A pseudo-M-space which is not an M-space. 96

22. A topological vector lattice which is not a

pseudo-M-space. 97
 XV

23. The topology of a bornological locally convex

lattice which is not an order bound topology.
- A quasi-barrelled locally convex lattice which
is not order-quasi-barrelled. 97

24. An o r d e r -quasi-barrelled vector lattice which

is not barrelled. 98

25. A C.O.Q. vector lattice which is not order-quasi-

barrelled.
- An o r d e r - ( D F ) - v e c t o r lattice which is not order-
quasi-barrelled. 98

26. A C.O.Q. vector lattice which is not countably

barrelled.
- An order-(DF)-vector lattice which is not
countably barrelled. 99

27. A countably quasi-barrelled locally convex lattice

which is not a C.O.Q. vector lattice. 99

28. An order-quasi-barrelled (and hence a C.O.Q.)

vector lattice which is not an order-(DF)-vector
lattice, i00

29. An O.Q.U. vector lattice which is not ultra-

barrelled, i00

30. A quasi-ultrabarrelled topological vector lattice

which is not an O.Q.U. vector lattice. 1o0

31. An order-quasi-barrelled vector lattice which is

not an O.Q.U. vector lattice, i00

32. A countably O.Q.U. vector lattice which is not

countably ultrabarrelled, i01

33. A countably quasi-ultrabarrelled topological

vector lattice which is not a countably O.Q.U.
vector lattice, i01

34. A C.O.Q. vector lattice which is not a countably

O.Q.U. vector lattice. I01

VI HEREDITARY PROPERTIES 103

Introduction 103

i. A closed subspace of a reflexive space, ~hich is

not reflexive.
 XVI

- A closed subspace of a Montel space, which is

not Montel. 104

2. A closed subspace of a bornological space, which

is not bornological. 104

3. An infinite countable codimensional subspace of a

bornological space, which is not quasibarrelled.
- An infinite countable codimensional subspace of
a bornological space, which is not bornological. 104

4. A closed subspace of a barrelled space, which is

not countably quasi-barrelled.
- A closed subspace of a barrelled (quasi-barrelled,
countably barrelled or countably quasi-barrelled)
space, which is not a barrelled (quasi-barrelled,
countably barrelled or countably quasi-barrelled)
space. 105

5. A dense uncountable dimensional subspace of a

barrelled space, which is not barrelled. 105

6. A closed subspace of a (DF)-space which is not

a (DF)-space.
- A closed subspace of a barrelled (quasi-barrelled,
bornological) space, which is not barrelled (quasi-
barrelled, bornological).
- A closed subspace of a Montel space, which is
not Montel.
- A closed subspace of a countably quasi-barrelled
(countably barrelled) space which is not countably
quasi-barrelled (countably barrelled). 106

7. An infinite countable codimensional subspace of

a quasi-barrelled (DF) space, which is not a (DF)
space. 107

8. A closed subspace of a hyperbarrelled space,

which is not hyperbarrelled.
- A closed subspace of a quasi-hFperbarrelled (~ -
hyperbarrelled, ~ -quasi-hyperbarrelled) space
which is not quasi-hyperbarrelled (~-hype~barrelled,
-quasi-hyperbarrelled). 107

9. A closed subspace of an ultrabarrelled space, which

is not countabl y quasi-ultrabarrelled. 107
 XVII

- A closed subspace of an ultrabarrelled (quasi-

ultrabarrelled, countably ultrabarrelled, countably
quasi-ultrabarrelled) space which is not ultra-
barrelled (quasi-ultrabarrelled, countably ultra-
barrelled, countably quasi-ultrabarrelled). 107

I0. A lattice ideal in an order-quasi-barrelled vector

lattice, which is not order-quasi-barrelled.
- A lattice ideal in a C.O.Q. vector lattice,
which is not a C.O.Q. vector lattice.
- A lattice ideal in an O.Q.U. vector lattice
which is not an O.Q.U. vector lattice.
- A lattice ideal in a countably O.Q.U. vector
lattice, which is not a countably O.Q.U. vector
lattice. 108

ii. A complete locally convex space whose quotient is

not sequentially complete.
- A complete (quasi-complete, sequentially complete)
space whose quotient is not complete (quasi-
complete, sequentially complete). 108

12. A quotient of a Montel space, which is not semi-

reflexive.
- A Montel (reflexive, semi-reflexive) space
whose quotient is not a ~ontel (reflexive, semi-
reflexive) space. 108

13. A quotient of a Frechet Nontel space, which is

not reflexive.
- A Frechet Montel space whose quotient is not a
Montel space.
- A reflexive Fr~chet space whose quotient is not
reflexive. 108

14. A product of B-complete spaces which is not

B-complete. 109

15. An arbitrary direct sum of B-complete spaces,

which is not B-complete. 109

VII TOPOLOGICAL BASES 110

Introduction 110

I. A separable Banach space which has no basis. 115

 XVIII

2. A Banach space w i t h a basis, whose dual space does

not have a basis. 115

3. A Banach space w h i c h has no u n c o n d i t i o n a l basis. 115

4. A Banach space w i t h a basis which is not u n c o n d i -

tional. 116

5. A Banach space w i t h an u n c o n d i t i o n a l basis which

is not b o u n d e d l y complete. 117

6. A Banach space w i t h a basis which is not absolutely

convergent. 118

7. A Banach space w i t h a basis which is not a normal

basis. 119

8. A Banach space whose dual space has a normal basis

which is not a retro-basis. 120

9. A Banach space w i t h a Besselian basis which is not

a Hilbertian basis. 120

i0. A Banach space w i t h a Hilbertian basis which is

not a Besselian basis. 120

ii. A Banach space w i t h a basis which is not a monotonic

basis. 121

12. A Banach space with a sub-symmetric basis which is

not a symmetric basis. 121

13. A Banach space w i t h o u t a sub-symmetric basis. 122

14. An E - c o m p l e t e biorthogonal system in a Banach

space, which is not a basis. 122

15. A normed space with a basis which is not a Schauder

basis. 122

16. A normed space w i t h a Schauder basis which is

neither an (e)-Schauder basis nor a (b)-Schauder
basis. 123

17. A Banach space whose dual has a weak basis but no

basis.
- A Banach space whose dual has a weak -Schauder
basis which is not a (Schauder) basis. 124

18. A Banach space whose dual space has a basis which

is not a weak basis. 125
 XIX

19. A Banach space whose dual space has a weak basis

which is not a weak Schauder basis. 126

20. A separable locally convex space which has no

basis. 128

21. A basis in a locally convex space, which is not a

Schauder basis. 128

22. A complete, metrizable and separable (non-locally

convex) topological vector space which has no
basis. 130

23. A generalized basis in a non-separable Banach

space, which is not a Markushevich basis. 130

24. A Markushevich basis in a Fr~chet space, which

is not a Schauder basis. 130

25. A maximal biorthogonal system in a Fr~chet space

which is not a generalized basis. 131

26. An extended unconditional basis in a countably

barrelled space, which is not an extended
unconditional Schauder basis. 132

27. The isomorphism theorem fails if the domain or

the range space is not barrelled. 132

28. The isomorphism theorem does not hold for

generalized basis even if the domain and the
range spaces are complete and barrelled. 133

29. A vector space with two compatible locally convex

topologies such that there is a Schauder basis
for one topology, which is not a Schauder basis
for the other topology. 133

VIII TOPOLOGICAL ALGEBRAS 137

Introduction 137

I. An algebra which cannot be made into a Banach

algebra. 142

2. A Banach algebra which has no radical. 142

3. A Banach algebra with a closed ideal which is

not an intersection of m a x i m ~ l regular ideals. 142

4. A Banach algebra with an approximate identity

 XX

which is not an identity. 143

5. An A -algebra which is not a B -algebra. 143

6. An A -algebra which is not symmetric. 144

7. A Frechet algebra which is not a Banach algebra.

- A Q-algebra which is not a Banach algebra. 144

8. A Frechet algebra which is not a locally

m-convex algebra.
- A Fr~chet algebra which is not a Banach algebra.
- A locally convex algebra which is not a locally
m-convex a l g e b r a . 145

9. A locally m-convex algebra which is not metri-

zable.
- A locally m-convex algebra which is a Q-algebra
but not a normed algebra. 145

i0. A Frechet algebra which has closed ideals but

not closed maximal ideals. 146

II. A Frechet algebra which does not have the Wiener

property.

- A Fr~chet algebra which is not a locally

m-convex algebra. 147

12. A semi-simple locally m-convex Fr~chet algebra

which is a projective limit of Banach algebras
which are not semi-simple. 147

13. ~[-singular elements of a locally m-convex

Frechet algebra, which are not topological
divisors of zero. 149

14. A locally m-convex Fr~chet algebra which has

neither topological divisors of zero nor J~-
singular elements. 150

iS. An m-barrelled algebra which is not barrelled. 150

16. A countably m-barrelled algebra which is not

m-barrelled. 151

17. A complete p.i.b, algebra which is not a P-

algebra. 151

18. A metrizable p.i.b, algebra which is neither

a P-algebra nor an.m-bornological algebra. 152
 XXJ

19. The Gelfand map which is continuous for a locally

convex algebra which is not m-barrelled. 152

20. A GB*-algebra which is not a locally m-convex

algebra. 153
21. A GB*-algebra on which there are no non-trivial
multiplicative linear functionals. 153
22. A Pseudo-complete locally convex algebra which
is not sequentially complete. 154
23. An A-convex algebra which is not a locally m-
convex algebra. 155

24. A p-normed (locally bounded) algebra which is

not a normed algebra. 156

25. A locally m-semi-convex algebra which is not a

locally m-convex algebra. 157

OPEN PROBLEMS 157

BIBLIOGRAPHY 158

INDEX 170
 PREREQUISITES

VECTOR SPACES AND O R D E R E D VECTOR SPACES

A nonempty set E is c a l l e d a vector space over a

field ~ if

(a) E is an additive abelian group, and

(b) for every ~ ~ and xEE, there is d e f i n e d an

element ~x in E subject to the f o l l o w i n g

conditions:

(b 1) ~(x+y) = ~x+~y

(b 2) (~+~)x = ~x+~x

(b a) ~(~x) = (~B) x

and (b~) lx = x

for all a, 8 e ~, x,y ~ E and 1 the unit element of

under m u l t i p l i c a t i o n .

If ~ is the field ~(~) of real (complex) numbers~

the v e c t o r space E is called a real (complex) vector

space.

Throughout this book, we deal w i t h only real or

complex vector spaces, and we use O t o denote the zero

element of ~ as well as that of a v e c t o r space.

PROPOSITION I. If E is a v e c t o r space over

(a) sO = 0 for all ~ E ~ ;

(b) Ox = 0 for x~E ;

(c) (-~)x = -(~x) for ~ E ~, x c E ;

and (d) ~ x = O, x ~ O, implies that ~ = 0

A vector space E w i t h a multiplication (that is, if

x,y ~E, then x y , y x EE) is c a l l e d an algebra.

 If E is a v e c t o r space and F a n o n e m p t y subset of

E, then F is c a l l e d a vector subspace (or simply, sub-

space) of E if, u n d e r the o p e r a t i o n s of E, F itself forms

a vector space over the field ~. If x , ..... , x ~ E,

n 1 n

then X ~i xi '
i=l
~i ~ K' is c a l l e d a linear combination of x , .... , x . A
1 n

subset B of a v e c t o r space E is c a l l e d linearly indepen-

dent if B # 0 or {0) and no element of B is a linear

combination of any finite subset of other elements of B.

A maximal linearly independent subset of a v e c t o r space

is c a l l e d a Hamel basis (or v e c t o r basis). Every vector

space has a Hamel basis and any two Hamel bases of a

vector space have the same cardinal number. The cardinal

number of a Hamel basis of a v e c t o r space is c a l l e d its

dimens ion.

If F is a s u b s p a c e of a v e c t o r space E over the

field K, the q u o t i e n t space of E by F is a v e c t o r space

E/F over K where, for x + F, x + F ~ E/F and ~ ~ K ,

1 2

(i) (x + F) + (x + F) = (x + x ) + F
I 2 I 2

and (ii) ~(x + F) = ~x + F

An a r b i t r a r y product E = ~ E of v e c t o r spaces E

is a v e c t o r space w h e r e addition and scalar m u l t i p l i c a t i o n

are d e f i n e d as c o o r d i n a t e w i s e addition and scalar m u l t i -

plication.

If {E } is a family of v e c t o r spaces and

F = X E = {x = {x ) ; x = 0 for all ~ except for a

finite subset of I} , then F is a v e c t o r space, called

direct sum of {E } , where addition and m u l t i p l i c a t i o n

are d e f i n e d as above.

A map f of a v e c t o r space E into another vector space

F is said to be linear if

f(ax + By) ~ af(x) + ~f(y)

for all x , y r E and a, ~ ~ ~ . If F= ~, then f is c a l l e d

a linear functional.

A subset A of a v e c t o r space E is said to be (i)

circled (or b a l a n c e d ) if aAcA for e v e r y a r E such that

l~l : 1 ,

(ii) absorbing if for every x r E there is an a > 0 such

that x r ~ A for all X r E with I~] ! a , (iii) convex

if x,y r A and 0 < X < 1 imply that Xx + (i - X)y r A ,

and (iv) semiconvex if A + A c ~ A for some ~ > 0.

Let E be a v e c t o r space. A map p:E + ~+ is c a l l e d

a semi-norm if

(a) p(x+y) < p(x) + p(y) for all x , y E E ,

and (b) p(~x) = IXI p(x) for all x e E and ~ r ~ .

Clearly p(O) = O. If p(x) = 0 implies x = O, then

p is c a l l e d a norm on E and is d e n o t e d by II " ]I

p is c a l l e d a k-semi-norm if (b) is r e p l a c e d by the

following:

(b') p(Xx) = IXl k p(x), 0 < k < I, x r E and ~ e ~ 9

Clearly p(O) = 0 . If p(x) = 0 implies x = O, then p

is c a l l e d a k-norm.

p is c a l l e d a quasi-semi-norm if (a) is r e p l a c e d by

the following:
 (a') There is a n u m b e r b > 1 for w h i c h

p(x+y) < b(p(x) + p(y)) for all x , y e E .

The smallest value of b for w h i c h (a') is s a t i s f i e d

is r e f e r r e d to as the m u l t i p l i e r of p.

Clearly p(O) = 0 . If p(x) = 0 implies x = O, then

p is c a l l e d a quasi-norm.

THEOREM i. If q is a q u a s i - s e m i - n o r m on a v e c t o r

space E with multiplier b and k < log 2 , then there is a

2b
k-semi-norm p on E e q u i v a l e n t to q.

The inner product (. , .) in a v e c t o r space E is a

map E x E + ~ satisfying the following conditions:

(a) (x,x) > 0 for all x e E ;

(b) (x,x) = 0 iff x = 0 ;

(c) (x,y) = (y,x) for all x, y e E ;

(d) (~x + ~y,z) = ~(x,z) + ~(y,z) for all x , y e E

and ~, ~ ~ I~ .

The semi-inner product [. , .] in a v e c t o r space E is

a map E • + ~ satisfying the following conditions:

(al) ~x+y,z] = Ix,z] + [y,z] , x,y,z ~ E

(b2) [Xx,y] = X[ x,y] , x,y r E, ~. r K ;

(c,) Ix,x] > 0 for x ~ O, x ~ E ;

(d,) I [ x,y] [ < ~x,x] ~ [y,y] ~ , x,y e E .

All vector spaces in w h a t follows are o v e r the field

of real numbers.

A vector space E is c a l l e d an o r d e r e d vector space if

it is e q u i p p e d with a reflexive, transitive and a n t i s y m -

metric relation < satisfying the following conditions:

 (a) x < y implies x + z < y + z for all x,y,z E E;

and (b) x < y implies ~x < Xy for all x,y a E and

~ e~,~ > O.

The set C = {x s E ; x > 0} is c a l l e d the p o s i t i v e

cone (or simply, cone) in an o r d e r e d vector space E. It

satisfies the f o l l o w i n g conditions:

(al) C + C ~ C ;

(b~) XC ~ C for ~ >0 ;

(c3) c n (-c) = {o}

On the other hand, if C is a subset of a v e c t o r space

E satisfying (al), (b2) and (c3) then x < y if and only if

(henceforth abbreviated to iff) y - x c C defines an order

relation < on E for w h i c h E becomes an o r d e r e d vector

space w i t h C as p o s i t i v e cone.

We w r i t e (E,C) to d e n o t e an o r d e r e d v e c t o r space E

with the p o s i t i v e cone C.

A subset C of E s a t i s y i n g (al) and (bl) is called a

wedge.

The o r d e r - i n t e r v a l between two elements x and y of

an o r d e r e d vector space is the set {t r E; x < t < y}

which is d e n o t e d by E x , y ] . A subset B of E is said to be

order-bounded if there exist x,y in E such that B E [ x , y ] -

A subset A of E is said to be m a j o r i z e d (minorized) if

there is an element t in E such that t > a (respectively,

t < a) for all a ~ A. If every pair x,y in A is m a j o r i z e d

(minorized), then A is said to be d i r e c t e d < (respectively,

directed > ).
 The p o s i t i v e cone C in an o r d e r e d vector space E is

generating if E = C - C. The p o s i t i v e cone C is g e n e r a t i n g

iff E is d i r e c t e d < . An element e in (E,C) is called an

order-unit if the order interval ~-e,e~ is absorbing, that

is, if e c C and for each x E E there exists ~> 0 such

that -~ e < x < ~ e. If (E,C) contains an order-unit,

then C is generating. A net {e~ ; ~ eA, ~ } i n (E,C) is

called an a p p r o x i m a t e order-unit if the f o l l o w i n g condi-

tions are satisfied:

(i) e~ ~ C for each ~;

(ii) for any pair of elements ~ ,~ in the d i r e c t e d

] 2

set A w i t h ~ < ~ , we have ex1 < e~ ;

1-- 2 -- 2

(iii) for each x in E there exist X ~ h and a real

number a > 0 such that -X ex ~ x ~ a e~ .

Clearly each o r d e r - u n i t is a p p r o x i m a t e order-unit.

An o r d e r e d vector space is c a l l e d A r c h i m e d e a n (almost

Archimedean) if x < 0 w h e n e v e r ~ x < y for some y ~ C and

all a >0 (respectively, if x = 0 w h e n e v e r -~ y < x < ay

for some y ~ C and all ~ > 0). Every Archimedean ordered

vector space is almost A r c h i m e d e a n . A set A in an o r d e r e d

vector space (E,C) is said to be decomposable if for each

a in A there exist a , a in A n C such that a=a a - a a

1 2 1 1 2 2

for some a , ~ > 0 with ~ + ~ = 1. A subset A of an

1 2 -- 1 z

ordered vector space (E,C) is said to be full if

A = (z ~ E ; x < z < y for x,y e A} .

A subset S of the positive cone C in an ordered

vector space is said to be exhausting if for each x c C,

there are s ~ S and ~ > 0 such that x < ~s 9 A nonempty

convex subset B of the positive cone C in an ordered vector

space (E,C) is a base for C if each x E C, x # O, has a

unique representation of the form x = ~b , b r B, ~ > O.

x is an extremal point of the cone C if each point of the

order interval ~O,x~ is a p o s i t i v e scalar m u l t i p l e of x.

A linear map T from an ordered vector space (E,C)

into another brdered vector space (F,K) is positive if

Tx E K w h e n e v e r x ~ C. A linear functional on an ordered

vector space (E,C) is p o s i t i v e if Tx > 0 w h e n e v e r x E C.

Let (E,C) be an ordered vector space. Let A be a

subset of E satisfying the following properties:

(S) x > a for all a E A ;

l

(S) y > x whenever y > a for all a ~ A.

2 --

Then x is called the supremum of A and is w r i t t e n as

x = sup(A). Dually we can define the infimum of A.

If sup ~x,y~ , written as x v y, or inf (x,y) ,

written as x ^ y, of every pair x , y of elements of E

exists in E, then E is called a vector lattice.

+
x = sup ~x,O) , x -= (-x) +, [x] = sup (x, -x) are

respectively called the positive part, negative part and

absolute value of x in a vector lattice E. It follows that

+
x = x - x so that the p o s i t i v e cone in a vector lattice

is always generating.

In a vector lattice (E,C) the following property,

called the d e c o m p o s i t i o n property, is always satisfied:

EO,x~ + ~O,y~ -- ~0, x + y ~ , x,y r C.

An o r d e r e d vector space (a vector lattice)is order-

complete ( o-order complete ) if every d i r e c t e d subset D, <--of

E that is m a j o r i z e d in E has supremum that belongs to E

(if the supremum of every countable majorized subset of E

exists in E). An o r d e r - c o m p l e t e vector space (E,C) is a

vector lattice iff C is generating. An ordered vector space

(E,C) is o r d e r - s e p a r a b l e if every subset B of E that

has a supremum in E contains a countable subset B I such

that sup(B) = sup(B1).

A net {x } in a vector lattice E decreases to x ~ E

0

if x0 = inf ~x } and x~ _> x B whenever B _> ~ . A net

{x } in E o r d e r - c o n v e r g e s to x E E if {x } is an order-

bounded subset of E and there is a net ~y~} that d e c r e a s e s

to 0 such that Ix - x I < y~ for all ~. A linear map T

from a vector lattice E into a vector lattice F is order-

continuous if the net (Tx } o r d e r - c o n v e r g e s to 0 in F

whenever ~x a} is a net that order-converges to 0 in E.

Instead of nets, if sequences are considered, then we call

it as s e q u e n t i a l l y order-continuous.

A linear map from a vector lattiGe (E,C) into another

vector lattice (F,K) is called a lattice homomorphism if

it preserves lattice operations v and ^

A subset A of a vector lattice (E,C) is called solid

if Ixl ~IYl, Y ~ A implies x ~ A. A solid subspace of E

is called a lattice ideal. A subspace F of a vector lattice

(E,C) is called a sublattice if for every pair x,y ~ E

the s u p r e m u m and infimum of x and y in E lies in F. Every

lattice ideal is a sublattice. A lattice ideal I in an

order-complete vector lattice is called a band in E if I

cont,ains the s u p r e m u m of every subset of I that is m a j o r -

ized in E.
 CHAPTER I

TOPOLOGICAL VECTOR SPACES

Introduction

A topological space E which is also a vector space

over the field K of reals or complexes is called a

topological vector space if

(TVS!) the map (x,y) § x+y from E~E into E is

continuous, and

(TVS2) the map (~, x) § I x from E x E into E is

continuous.

A topology on a vector space is said to be compatible

if the axioms (TVSI) and (TVS2) are satisfied.

REMARK I. We shall u s u a l l y deal w i t h the H a u s d o r f f

topological vector spaces in the sequel.

THEOREM I. Let E be a topological vector space.

(a) For each x ~ E and ~ E ~, X ~ O, the m a p

0
x + ~ x + x is a h o m e o m o r p h i s m of E onto itself.
0

(b) For any subset A of E and any basis ~ of the

neighbourhood filter at O,

where A is the closure of A.

(c) If A is an open subset of E and B any subset of

E, then A + B is an open subset of E.

(d) If A is a closed subset and B a compact subset

of E, then A + B is a closed subset of E.

 10

(e) If A is a circled subset of E, so is its

closure X .

THEOREM 2. Let E be a topological vector space. Then

there exists a neighbourhood basis ~ of O in E such that

(N1) each U in ~ is closed, circled and absorbing,

and (N2) for each U in ~ , there is a V in ~ with V + V c U.

Conversely, if E is a vector space and ~ is a filter

basis satisfying (NI) and (N2), then there is a unique

topology u on E which makes it a topological vector space

and ~ is a neighbourhood basis at O.

A topological vector space (E,u) is metrizable if

there is a metric on E whose open balls form a basis. A

topological vector space E is metrizable iff there is a

countable neighbourhood basis at O. These neighbourhoods

can be so chosen as to satisfy (NI) and (N2) of Theorem 2.

A subset B of a topological vector space E is called

(i) bounded if it is absorbed by every neighbourhood of

O in E, (ii) totally bounded (precompact) it for each

neighbourhood V of O in E, there is a finite subset B0 in B

such that B ~ B 0 + V.

Every totally bounded subset of a topological vector

space E is bounded.

A topological vector space E is said to be (i) com-

plete if every Cauchy filter is convergent, (ii) quasi-

complete if every closed and bounded subset of E is

complete, and (iii) sequentially complete (or, semi-

complete) if every Cauchy sequence in E converges.

Completeness=> quasircompleteness=> sequential

 11

completeness. (Cf. B [ 3 ~ )

If M is a closed subspace and N a finite dimensional

subspace of a topological vector space E, then M+N is

closed in E. If M is a closed subspace of finite codimen-

sion (that is, E/M is of finite dimension), then E = M@N

for every algebraic complementary subspace N of M.

THEOREM 3. Every n-dimensional topological vector

space E over the field K is topologically isomorphic to K n

with its natural topology. (Cf. B [ 3 ~ )

THEOREM 4. A topological vector space is finite dimen-

sional iff it is locally compact. (Cf. B [31])

THEOREM 5. There is a one-to-one correspondence

between the kernals of the continuous non-zero linear

functionals on a topological vector space E and the closed

hyperplanes in E. (Cf. B [ 3 ~ )

A topological vector space E is called a semiconvex

space if it has a neighbourhood basis at O consisting of

semiconvex sets. E is semiconvex iff there is a family

{p }of (continuous) ks -semi-norms (O<k <i) such that the

sets {xcE ; p~(x)<l} form a neighbourhood basis at O.

A topological vector space E is called locally bounded

if it has a bounded neighbourhood at O. E is locally boun-

ded iff its topology is given by a k-norm (O<k<l). Clearly

a locally bounded space is semi-convex. A (Hausdorff)

locally bounded space is metrizable. A product of an

infinite family of locally bounded spaces # {O} is not

locally bounded (B [31], page 30).

 12

A topological vector space E is called a locally

convex space if it has a n e i g h b o u r h o o d basis at O consis-

ting of convex neighbourhoods. These neighbourhoods can be

chosen to be closed and circled. Since there is a one-to-

one c o r r e s p o n d e n c e between the sets of all closed, circled

and convex subsets containing O as their interior point

and the set of all continuous semi-norms in a topological

vector space, it follows that a locally convex topology

can also be defined by a set of continuous semi-norms on E.

Clearly a locally convex space is a locally semi-

convex space.

I. A topology on a vector space, which is not c o m p a t i b l e

with the vector space structure.

The discrete topology on a vector space E ~ (O~ is not

compatible with its v e c t o r space structure.

2. A topological vector space which is not a locally

semi-convex space.

(i) Let E be the vector space of all m e a s u r a b l e real

functions on the closed interval ~O,~. Define the follow-

ing m e t r i c on E:

1
d(f,g) = f If(x) -g(x)[ dx, f,gcE .
0 l+lf(x) -g(x][

Then E, w i t h the metric topology defined b y d, is a topo-

logical vector space which is not a locally semi-convex

space.

(ii) Let p = { p i } be a s e q u e n c e of positive numbers,

O<Pi<l f o r all i, decreasing t o O. L e t E be t h e set of all

sequences x = (xi) of real. numbers such that

 13

n=l
I xil spi < ~
Definite the metric

d(x,y) = F Ixi- yilSPi

i=l

Then E, w i t h the metric topology defined b y d, is a topo-

logical vector space which is not a locally semi-convex
space.

3. Alocally bounded (and hence a locally semi-convex)

space which is not a locally convex space.

(i) Let E =~P[a,b], a,b~R, O<p<l, be the space of

all equivalence classes of measurable functions f(t) on
[a,b] with

fb If(t) I Pdt
a
<~ 9

Define

qp(f) = (~b if(t) ip dt~/p

Then E, equipped with the topology induced by the quasi-

norm qp, is a locally bounded space but not locally convex.

(ii) Similarly the space Zp, O<p<l, of sequences

x=(Xn) , ~ Ixn[P< ~, is a locally bounded space for the
n=l

topology induced by qp, where

IxnlP) 1/p
qp = ( n= 1
But i t is not locally convex.

(iii) Let H p, O<p<l, denote the vector space of

functions f(z) of the complex variable z, which are
analytic in the interior of the unit circle and satisfy
 14

sup {f If(rel@)IP d@ ; O<r<l)< ~.

0

Define

qp(f) = sup {Ap(r;f) ; Ojr<l}

where
1 2~ I/p
Ap(r;f) = (--~9 (lf(rei~l p d@)

Then (H p, qp) is a locally bounded space which is not a

locally convex space.

4. A locally convex space which is not a locally bounded

space.

~N, equipped with the product topology, is a locally

convex space which is not locally bounded.

5. A locally semi-convex space which is neither locally

convex nor locally bounded.

Let s O<pjl, be defined as in =~= 3(ii). Then the

space
E = ~Ix Z~ x ~ I/3 x

equipped w i t h the product topology i s a l o c a l l y semi-

convex space; but E i s not l o c a l l y bounded, because the

product of an infinite family of topological vector spaces

can never be locally bounded. Clearly E is not locally

convex.

6. A topological vector space on which there exist no

non-trivial continuous linear functionals.

Consider the topological vector space ~P [a,b~ ,

O<p<l, as defined in ~= 3(i). We show that every continuous

linear functional on ]P vanishes identically. Suppose

# O is a continuous linear functional on I P. Then

 15

]r = 1 for some foe~ p. For a <s<b, we put

(I) fo(t) if a<t<s

f (t) = (
s O if t>s

Define

(2) (1)
fs (t) = fo(t) fs (t).

Now, qp(f~z)), where

(I) s
qp(fs ) = (~ If~ lp dt~/p

increases continuously from 0 to q~(f0) so that there is

an so with
(,) (2)

Since [@ ( f o ) l = 1 ,
(i)
[r ( f s o ) 1 Z I , i = 1 or z.
Let
(i)
f1(t) = 2f (t) for this i.
s0
Then I~ (fl)I >i with
p_I/p
qp(fl) = Z qp(fO) 9

By induction, we define a sequence {fn } with ] # (fn)I_>l ,

and
n(p-1)/p
qp(fn ) = 2 qp(fO) ~ O

contradicting the continuity of ~ .

7. A topological vector space such that no finite-

dimensional suhspace has a topological complement in it.

Let E = ~P[a,b], O<p<l, be as in ~# 3(i). Let H be a

 16

finite dimensional subspace of E and HI a one-dimensional

subspace of H. Then, by Theorem 3, HI has a topological

complement H2 in H under the induced topology so that

H -- HI@ H2. If G w e r e a topological complement of H, then

H2@ G would be a topological complement of HI and so a

closed hyperplane of E which is impossible by Theorem 5 and

4/~ 6.

8. Two closed subspaces of a topological vector space,

whose sum is not closed.

Let E be a Hilbert space with an orthonormal basis

(en) n> 0 9

Let an= e2n and

= + i
bn e2n (~-~-~) e2n+1

for every n>0; let A (respectively B) be the closed vector

subspace of E generated by the an(respectively bn). Show

that:

(a) A n B = {0}; hence the sum A+B is direct (algebraically).

(b) The direct sum A+B is not a topological direct sum

(consider in that subspace the sequence of points b n- a n

and consider projections of A • B).

(c) The subspace A+B of E is dense but not closed in E

(show that the point ~ (b n- an) does not belong to A+B).

0

(Cf. Page 123 of Dieudonne's "Foundations of Modern Analy-

sis", Academic Press, Inc., New York, 1960).

9. A topological vector space in which the convex

envelope of a precompact set is not precompact (not even

bounded).

l
Consider ~2 with the usual norm. Let e n be the element
 17

having 1 in the n th coordinate and zeros elsewhere. The

-!
set consisting of 0 and the points n 2e n is compact whose

convex envelope is unbounded, because the sequence of

convex combinations

1 n -!
Yn n Z r 2e r
r=l

is unbounded.

i0. A bounded linear map from a topological vector space

to a topological vector space, which is not continuous.

Let ( E , I [ . [ I ) be an infinite-dimensional Banach

space. Then the identity map: (E,~ ( E , E ' ) ) + (E,II.II) is

bounded but not continuous.

 CHAPTER 2

LOCALLY CONVEX SPACES

Introduction

We have already introduced the definition of a locally

convex space in Chapter I. Now we introduce some proper-

ties and results of locally convex spaces which will be

used in the sequel.

A locally convex space E is called B-complete if

every linear, continuous and almost open map of E onto any

locally convex space is open. (A linear map f from E to F

is almost open if, for each neighbourhood V of O in E,

f(v) contains a neighbourhood of O in F). A complete and

metrizable locally convex space is called a Fr@chet space.

A locally convex space whose topology is generated by a

norm If'If is called a normed space. A complete nermed

space is called a Banach space. A vector space equipped

with an inner product is called an inner product (or, pre-

Hilbert) space. A complete inner product space is called a

Hilbert space.

Hilbert space ~ > Banach space ~ > Fr~chet space.

A vector space equipped with a semi-inner product is

called a semi-inner product space. A semi-inner product

space can be made into a normed space and a normed space

can be made into a semi-inner product space. Every inner

product space is a semi-inner product space. (Cf. P [ 2 ~ )

A vector space E is called a generalized inner

product space (generalized semi-inner product space) if

 19

(i) there is a subspace M of E which is an i n n e r

product space (respectively, a semi-inner product space)

and ( i i ) there is a nonempty set ~ of linear operators on

E such that (a) ~ E ~ M and (b) Tx = 0 f o r all T in

implies x = 0 .

Every generalized inner product space is a generalized

semi-inner product space. (Cf. p [7~ , p [6~ )

Let ~(E,F) denote the vector space of all continuous

linear maps f r o m a t o p o l o g i c a l vector space E into a topo-

logical vector s p a c e F. L e t ~ be a c l a s s of subsets o f E.

We c a n d e f i n e a topology in ~(E,F) of uniform convergence

over the sets of G as f o l l o w s : L e t ~ be a n e i g h b o u r h o o d

basis a t 0 i n F. The f a m i l y {M(S,V) ; ScG, Vr where

M(S,V) = {fcZ(E,F) ; f ( S ) =V },

is a neighbourhood basis at 0 in ~(E,F) for a unique

translation-invariant topology called the ~-topology.

L~(E,F) is a topological vector s p a c e iff f(S) is bounded

in F f o r e a c h S~G and f r In addition, if F is

locally c o n v e x so i s
~(E,F). If~consists of bounded
G
subsets of E such that u S is total in E (that is, the

linear envelope
o f uS i s d e n s e i n E) and F i s a l o c a l l y
6
convex space, then ~(E,F) is a Hausdorff locally convex
@
space. If Gconsists of all finite (compact, precompact,

bounded.) subsets o f E, t h e n the G-topology is called the

topology of simple convergence (respectively, the topology

of compact convergence, the topology of precompact conver-

gence, the topology of uniform convergence on b o u n d e d s e t s ) .

Let E and F be a pair of vector spaces over K. Let r

 20

be a bilinear functional on E~F satisfying the following

separation axioms:

(D) r (x,y) = O for all ygF implies x = O.

(DI) r (x,y) = O for all xgE implies y = O.

Then E and F are said to form a dual pair or dual

system and we write <E, F>. r is called the canonical

bilinear functional of the duality and is usually denoted

by

(x,y) § <x,y>

Now, let E be a locally convex space and E' the

(topological) dual of E, that is, the vector space of all

continuous linear functionals on E. Clearly

E'c E* ~ KE

w h e r e E* i s the algebraic dual o f E, that is, the vector

space of all linear functionals on E, and K E is the

product space.

The coarsest locally convex topology for which the

map x + <x,x'> , for each x'~E', is continuous is called

the weak topology on E; it is denoted by ~(E,E') or w.

Similarly we can define the so called week topology

~(E',E) or W* on E' ~(E',E) is precisely the topology of

simple convergence on E' which, in turn, is induced from

the product topology of K E.

If ~ c o n s i s t s of all bounded subsets of E, the ~-

topology on E' is denoted by B(E',E) and is called the

strong topology on E' Similarly we can define the strong

topology B(E,E') on E.

L e t A be a s u b s e t of a lo,cally convex space. Then

 21

A ~ = {x'~E' ; Re<x,x'><l for all xEE)

is called the polar of A. The bipolar A ~176of A is defined

by

A ~176 = {xcE ; Re <x,x'> <I for all x'~A ~

THEOREM 1 . Let A,B and {Aa}as I be subsets of a

locally convex space E.

(a) ( ~ A ) ~ = ~1 AO , ~K, ~ ~ O.

(b) If AcB, then A~ ~B ~ and A ~ 1 7~6 B oo

(c) A c A ~176
.

(d) A ~ = A ~ 1 7 6.1 7 6

(e) A ~ is convex, ~(E',E) - closed and contains 0.

If A is circled, so is A~

(f) If A is a subspace of E, then A ~ is a ~(E',E)-

closed subspace of E' (Cf. B ~31~ )

i. A locally convex space which is not metrizable.

Let I be an uncountable set of indices and ~ the real

line. Then the product RI is a locally convex space which

is not metrizable, because the topological product of

metrizable locally convex spaces is metrizable iff the

product has finitely or countably many factors.

2. A metrizable topological vector space which is not

locally convex.

The spaces considered in =~ 3(i), (ii), (iii) in

Chapter 1 are metrizable topological vector spaces which

are not locally convex.

3. A sequentially complete locally convex space which is

not quasi-complete.
 22

(i) Let w d be the topological product of d, d>~0 '

copies of the field E, where K is equipped with the

natural topology. Let H be the subspace of w d consisting

of all vectors x = (xa) with only countably many non-zero

coordinates x . H is dense in Wd, and indeed every element

of w d is a closure point of H; H is, therefore, not quasi-

complete, although it is sequentially complete.

1
(ii) 2 = ~ = {X = (Xn) , Z IXnl<m},
n=

with its weak topology, is sequentially complete. It is

not quasi-complete because if x" is an element of 2" not

generated by an element of 2, then x" is the o(E",E')-

limit of a bounded directed family {x i} in E, and then the

(weakly) closed convex envelope in 2 of {x i} is (weakly)

bounded and (weakly) closed but not (weakly) complete.

4. A quasi-complete locally convex space which is not

complete.

Let E be an infinite-dimensional Banach space. Then

its dual E' is weakly quasi-complete but not weakly

complete. (if 2' were weakly complete, E would be finite-

dimensional).

S. A complete locally convex space which is not B-

complete.

Let 2 be an infinite-dimensional Bnach space and

let E w denote 2 equipped with the finest locally convex

 23

topology w. Let

i : E -~ E
w u

be the identity map. Then i is linear, one-to-one and onto.

Since u w, i is continuous, and since E u is barrelled,

(see Chapter 3 for definition) i is almost open (because)

a linear map from a locally convex space onto a b a r r e l l e d

space is almost open). We show that i is not open : If it

were, any circled, convex and absorbing subset of E would

be a u - n e i g h b o u r h o o d of O. In particular, any linear func-

tional on E w o u l d be u-continuous. This is false: Take any

infinite sequence ~tX n) in E that is linearly independent

and assume that x n = 1 for each n. The sequence (Xn) may be

extended into an algebraic base for E and a linear func-

tional on E may be defined so as to take a r b i t r a r i l y

preassigned values at elements of the base. In particular,

there is a linear functional f on E such that f(Xn) = Xn,

n = 1,2, .... Evidently this f is not u-continuous.

6. A complete locally convex space w h i c h is not metri-

zable.

(i) The locally convex space ~I of ~f= 1 is complete

since each ~ is. But it is not m e t r i z a b l e as was shown

there.

(ii) Consider the vector space ~ = ~(~) of all

finite sequences equipped with the finest locally convex

topology w. ~ is the strict inductive limit {See Chapter

3 for definition) of a strictly increasing sequence of

finite-dimensional Euclidean spaces ~ n (n > i) w h i c h are

Fr~chet spaces. Since a locally convex space with the

finest locally convex topology is complete, ~ is complete.

 24

But ~ is not metrizable: It it were, it w o u l d then be a

Frechet space and hence of the second category. But ~ is a

countable union of non-dense subsets (because the identity

map ~ n § ~n+l is "into" for each n), and therefore of

the first category which is a contradiction.

7. A normed space (and hence a m e t r i z a b l e locally convex

space) w h i c h is not complete.

(i) The vect space ~ = R(~) of all finite sequences

with the supremum norm topology is a normed space which is

not complete (See Chapter 3, =/~ 16).

(ii) Let E = C(I) be the Banach space of continuous

functions on the closed interval I = EO,~ , equipped with

the supremum norm. Let F be the subspace of E consisting

of functions f that vanish in a n e i g h b o u r h o o d (depending

on f) of t = O, under the relative topology (See Chapter 6

for definition). Then F is a n o r m e d space w h i c h is not

complete (See Chapter 5, =~ 23).

8. A locally convex space w h i c h contains a closed,

circled and convex set with no extreme points.

(i) Let B denote the closed unit ball of the Banach

space c o of sequences converging to O. Then B does not

have extreme points : If x = (Xn) is in co and

[[x[[ = sup [Xnl = i,

n

then we may replace some coordinate Xk, w i t h ]Xk[<l ,

a l t e r n a t e l y by Xk+ r and x k- r with a sufficiently small.

x, then, lies b e t w e e n the two points of B obtained in this

way.
 25

(ii) Let ] ~ [a,bJ , a,br be the vector space of all

the equivalence classes of measurable functions f(t) on

the closed interval [a,b] , with

b
r If(x) l dx <
a

~[a,bj is a Banach space under the norm

b
llfll = I If(x)[ dx.
a

The unit ball B of ~l[a,b] has no extreme points: Let

b
s If(x) l dx = l, f Ea,b]
a

We determine c in such a way that

b
I
S If(x)[ dx =
a

Let
2f(x) i f xr [ a , c )
ft(x) = {
0 i f xe [ c , b ]
and

0 if xE[a,c)
f2(x) = {
2f(x) if x~[c,b~
Then f is the middle point of the segment ~fl, f ~ w h o s e

end points both belong to B.

9. A topological vector space w h i c h contains a compact

convex set with no extreme points.

Cf. Roberts, J.W.: A compact convex set with no

extreme points, Studia Mathematica, T. LX(1977), 255-264.

I0. A w e a k l y compact set in a locally convex space, whose

weakly closed convex hull is not w e a k l y compact.

 26

Let r = ~(~) be the space of finite sequences with

the usual supremum norm. Define

fn(X) = xn, x = (Xn)E ~.

Then the set consisting of the points 22nf n together with

0 is w e a k l y compact, but is not strongly bounded. Thus,

its w -closed convex hull is not w -compact.

ii. A bounded sequence in a topological vector space,

which is not convergent.

The sequence 0,I,0,I,0,I, ~ in ~ is a bounded

sequence w h i c h is not convergent.

 CHAPTER 3

SPECIAL CLASSES OF LOCALLY CONVEX SPACES

Introduction

The concepts of Banach space, Hilbert space, inner

(semi-inner) product space and generalized inner (semi-

inner) product space are defined in the introduction to

Chapter 2. In the present Chapter, we introduce some more

concepts which are needed in the sequel. As we have

remarked in the introduction to Chapter I, we usually deal

with Hausdorff topologies.

A topological vector space E is called a Baire space

if it cannot be written as the union of an increasing

sequence of nowhere dense sets. (A subset A of E is called

nowhere dense if its closure A has empty interior). A

locally convex space E is called a Baire-like space if it

is not the union of an increasing sequence of nowhere

dense, circled and convex sets. A complete metrizable

locally convex space is called a Fr~chet space. A Banach

space is a Fr~chet space; a Fr~chet space is a Baire

locally convex space; a Baire locally convex space is a

Baire-like space.

A locally convex space E is called an unordered

Baire-like space if it is not the union of a sequence of

nowhere dense, circled and convex sets. An unordered

Baire-like space is Baire-like.

Let {E } c I be a family of locally convex spaces, E a

vector space and f a linear map from E~ into E, for each

~. Let E = ~f ( E ) . The finest locally convex topology u

 28

for which each fa is continuous is called the inductive

limit of {E }as I with respect to the maps f If I =

each fn is the identity map and the inductive limit topo-

logy on E induces the same topology as that of E n, then

(E,u) is called the strict inductive limit of {En}.

The (strict) inductive limit of a properly increasing

sequence of Banach (Fr~chet) spaces is called the (strict)

(LB)-space (respectively, (LF)-space).

A (Hausdorff) locally convex space E is t-polar if a

subspace M of E' is weakly closed whenever M n B 0 is weakly

closed for every barrel B in E.

A subset B of a locally convex space E is said to be

bornivorous if it absorbs all bounded subsets of E. A

closed, circled, convex and absorbing subset A of E is

called a barrel. A locally convex space E is called a

barrelled (quasi-barrelled) space if each barrel (borni-

vorous barrel) in E is a neighbourhood of O. A Fr~chet

space is barrelled. A barrelled space is quasi-barrelled.

THEOREM I. A sequentially complete quasibarrelled

space is barrelled. (Cf. B [ i ~ )

THEOREM 2. Let E be a locally convex space. The

following statements are equivalent:

(a) E is barrelled (quasi-barrelled).

(b) Each o(E',E)-bounded (B(E',E)-bounded) subset of

E' is equicontinuous.

(c) Each lower semi-continuous (bounded lower semi-

continuous) semi-norm on E is continuous. (Cf. B [ i ~ )

 29

A locally convex space E is called bornological if

each circled, convex and bornivorous set in E is a neigh-

bourhood of O. A bornological space is quasi-barrelled. A

metrizable locally convex space is bornological.

THEOREM 3. (a) A locally convex space E is bornolo-

gical iff each bounded semi-norm on E is continuous.

(b) A locally convex space E is bornological iff

each bounded linear map of E into any locally convex space

F is continuous. (Cf. B [ I ~ or B [ 3 ~ or B[20])

PROPOSITION i. The strong dual of a bornological

space is complete. (Cf. B [ I ~ or B [ 3 ~ or B[20])

COROLLARY i. Let E be a metrizable locally convex

space. The following statements are equivalent:

(a) (E', B(E',E)) is bornological.

(b) (E', B(E',E)) is quasi-barrelled.

(c) (E', 8(E',E)) is barrelled.

Let C(X) denote the locally convex space of continuous

functions on a (Hausdorff) completely regular space X,

equipped with compact-open topology.

THEOREM 4. C(X) is barrelled iff for every noncompact

closed subset A of X, there exists some f in C(X) which is

unbounded on A. (Cf. P ~ 6 ~ )

THEOREM 5. C(X) is bornological iff X is a Q-space

(real-compact space). (Cf. P [85]).

REMARK. For the definition of a Q-space (real-

compact space), see BE18 ] or P[85]

 30

A locally convex space E is semi-bornological if

every bounded linear functional on E is con-

tinuous. E is sequentially bornological (in short, S-

bornological) if each convex and bornivorous set B in E

is a sequential neighbourhood of O (that is, every sequence

converging to 0 belongs to B eventually).

THEOREM 6. Let E be a locally convex space. The

following statements are equivalent:

(a) E is S-bornological.

(b) Every circled, convex and bornivorous set in E

is a sequential neighbourhood of O.

(c) Every bounded semi-norm on E is sequentially

continuous. (Cf. pLs~)

A locally convex space E is said to be convex-

sequential (in short, C-sequential) if every convex sequen-

tially open subset of E is open.

THEOREM 7. Let E be a locally convex space. The

following statements are equivalent:

(a) E is C-sequential.

(b) Every convex, circled and sequentially open

subset of E is open.

(c) Every sequentially continuous semi-norm on E is

continuous.

(d) Every convex, circled and sequential neighbourhood

of O in E is a neighbourhood of O.

(e) Every convex sequential neighbourhood of O in E

is a neighbourhood of O. (Cf. P [88] )

 31

THEOREM 8. A locally convex space is bornological

iff it is both C-sequential and S-bornological. (Cf.

P ).

The G - t o p o l o g y on a locally convex space (E,u) is

called the Mackey topology if G i s the family of all

circled, convex and ~(E',E)-compact subsets of E', and is

denoted by T(E,E'). If u : ~(E,E'), (E,u) is called a

Mackey space.

THEOREM 9. A locally convex space E is a Mackey

space iff each convex and o(E',E)-relatively compact sub-

set of E' is equicontinuous. (Cf. B ~ 0 ] )

A quasi-barrelled space is a Mackey space.

A locally convex space E is called quasi-M-barrelled

if, in E', every circled, convex and ~(E',E)-bounded set

is ~(E',E)-relatively compact (equivalently, if the

strong bidual induces the Mackey topology on E). A quasi-

barrelled space is quasi-M-barrelled. E is said to have

property (S) if (E', ~(E',E) is sequentially complete and

property (C) if every ~(E',E)-bounded subset of E' is

~(E',E)-relatively countably compact. If (E~) '= E (algeb-

raically), where ~ is the strong topology B(E',E), then

Y
E is called a semi-reflexive space. If (E~)8 = E (topolo-

gically), then E is called a reflexive space. A reflexive

space is clearly semi-reflexive.

THEOREM iO. (a) A locally convex space E is semi-

reflexive iff each ~(E,E')-closed and bounded set is

o(E,E')-compact.

(b) A locally convex space E is reflexive iff E is

 32

barrelled and each ~(E,E')-bounded set in E is ~(E,E')-

relatively compact. (Cf. B E 3 ~ )

A locally convex space E is a Montel space if E is

barrelled and each bounded set in E is relatively compact.

A Montel space is reflexive. E is called a Schwartz space

if for every closed, circled and convex neighbourhood U

of O in E, there exists a neighbourhoed V of O such that

for every ~>O, the set V can be covered by finitely many

translates of ~U.

THEOREM ii. A locally convex space E is a Schwartz

space iff the following condition is satisfied: (Cf. BE1 ~ )

(S) Every bounded subset of E is totally bounded and

for every closed, circled and convex neighbourhood U of O,

there is a neighbourhood V of O such that for every ~>O,

we can find a bounded subset A of E such that V C ~ U + A.

A locally convex space E is called a distinguished

space if each ~(E",E')-bounded subset of its strong bidual

E" is contained in the ~(E", E') closure of some bounded

subset of E (equivalently, if for each ~(E",E')-bounded

subset B of E", there is a bounded subset A of E such that

B ~ A ~176 where A~176 the polar of A~with respect to the dual

pair <E',E">).Normed and semi-reflexive space s are distin-

guished.
THEOREM 12. A locally convex space E is distinguished

iff (E', B(E',E)) is barrelled. (Cf. B ~ I ~ or B ~20~ or

COROLLARY 2. A metrizable locally convex space is

! T
distinguished iff (E , B(E ,E)) is bornological.
 33

A locally convex space E is countably barrelled

(countably quasi-barrelled) if each ~(E',E) - bounded

(~(E',E) - bounded) subset of E', which is the countable

union of equicontinuous subsets of E', is itself equi-

continuous (equivalently, if each barrel (bornivorous

barrel) which is the countable intersection of circled,

convex and closed neighbourhoods of 0 is a neighbourhood of

0). A barrelled (quasi-barrelled) space is countably

barrelled (countably quasi-barrelled). A countably

barrelled space is countably quasi-barrelled. A countably

quasibarrelled space which has fundamental sequence of

bounded sets is called a (DF)-space. A normed space is a

(DF)-space. The strong dual of a metrizable locally convex

space is a (DF)-space. A locally convex space E is a-

barrelled (a-quasi-barrelled) if each o(E',E)-bounded

(B(E',E)-bounded) sequence in E' is equicontinuous. A

countably barrelled (countably quasi-barrelled) space is

a-barrelled (a-quasi-barrelled). A o-barrelled space is

a-quasi-barrelled. E is sequentially barrelled if each

a(E', E)-convergent sequence in E' is equicontinuous. A

a-barrelled space is sequentially barrelled. E is an H-

space if its strong dual is countably barrelled. Distin-

guished spaces and metrizable locally convex spaces are H-

spaces. E is k-barrelled (k-quasi-barrelled) if the inter-

section of a sequence {Vn} of circled, convex and closed

neighbourhoods of 0 is a barrel (bornivorous barrel)

implies that o knVn is a neighbourhood of O. A k-

n=l
barrelled space is clearly k-quasi-barrelled.

THEOREM 13. Let X be a completely regular Hausdorff

 B4

space. C(X) is countably barrelled iff every C(X)-pseudo-

compact subset of X which is the closure of a countable

union of compact sets is actually compact. (Cf. B [ I ~ )

i. An inner product (a pre-Hilbert) space which is not a

Hilbert space.

Consider the space C E a , ~ , a,bE~, of continuous

functions on the closed interval [ a , ~ . Define

b
(f,g) = I f(t). g(t) dt.
a

Then C[a,b~ becomes an inner product space which is not

complete.

2. A generalized inner product space which is not an

inner product space.

Let E = C (~) be the space of all continuous real-

valued functions on ~. Let M be the subspace consisting of

all square integrable functions in E. We define the inner

product in M as follows:

(x,y) = f x(t) y(t) dt.

R

We denote by z the family of all projections P(I) defined

by
(P(I)(x)) (t) = fi(t) x(t)

where, for every compact interval I, we choose an Uhryson

function fI with compact support which is 1 on I. Then E

is a generalized inner product space which is not an inner

product space.

3. A semi-inner product space which is not an inner

product space.
 35

Let ~ (~), 2!p<~, be the vector space of all the

equivalence classes of measurable functions f(t) on ~, with

f If(t) IPdt<~ .

Define
i
/ glfl p-I sgn(f) dt
I]fll p-z
p

where

]Ifl[p = (flf(t) lp dt~/p

Then [g,f] is a semi-inner product. But it is not an inner

product.

4. A generalized semi-inner product space which is

neither a semi-inner product space nor a generalized inner

product space.

Let E be the space of all measurable functions on ~.

Let M = ~P (~), 2~p<~ , be as defined in ~ 3, with the

semi-inner product in M as defined there. Let T be the

family of operators EP-I(I) such that for all x,ycE and

for any scalars a and 8 ,

E p-I (I) (~x + BY) (t)

= X (t) (]x(t)l p-2 + ly(t)l p-2) (~x(t) + By(t))

I

where I is the finite non-degenerate interval and Xi(t )

is the characteristic function of I. Then it is easy to

check that E is a generalized semi-inner product space

which is neither a generalized inner product space nor a

semi-inner product space.

 36

5. A Banach space w h i c h is not a Hilbert space.

(i) The s p a c e ~P[o,~, l<p<~ , p ~ 2, of equivalence

classes of pth power summable functions on t h e closed

interval [0,9 with t h e norm

1
[[f[] = (~ [ f ( t ) [ p dt~/p, f~]P [O,~

is a Banach space which is not a Hilbert space.

(ii) The s p a c e s c and c o o f a l l convergent sequences

and a l l sequences converging to O respectively are Banach

spaces which are not Hilbert spaces.

6. A Banach space w h i c h is not separable.

The space s of bounded sequences with the s u p r e m u m

norm topology is a Banach space. Two vectors in s whose

coordinates are equal to +i and -I are always distance 2

apart. Since there is a c o n t i n u u m of these, the set of

these vectors is not separable, and so s is not separable.

(If a normed space is separable, so is every subset of it).

7. A Banach space w h i c h is not reflexive.

(i) The space c o of all sequence converging to O,

equipped with the s u p r e m u m norm topology, is a B a n a c h space

which is not reflexive. For, the dual of c o is

~1 = {x = (x n) ; z ]Xnl < ~ }
n=l

with t h e norm

IIxll = ZlXnJ ,

and t h e dual o f Zl i s the Banach space t~ as d e f i n e d in

=#~ 6. T h u s , ~ is the bidual of c o , which is clearly

 37

larger than c o

(ii) From the fact that c o is not reflexive, it

follows that ~ and ~ are not reflexive.

8. A Fr~chet space which is not a Banach space.

Consider E = ~ ~n with the product topology,

n=l

where each ~n is a copy of the real line equipped with the

usual topology. Then E is a Fr6chet space, but not a Banach

space because a product of Banach spaces is a Banach space

iff it is a product of a finite number of Banach spaces.

9. A t-polar space which is not B-complete.

Let E = ~, the space of finite sequences, equipped

with the supremum norm topology. Then E'= Z:. (E', o(E',E))
is t-polar: For, the weakly bounded subsets of E"= E are

exactly those which are bounded in the norm topology of E.

Furthermore, if L is a subspace of E, whose intersection

with the unit sphere is weakly closed (and hence closed),

then L itself is closed, and therefore also closed in the

weak topology ~(E,E'). (E',~(E',E)) is not B-complete: The

equicontinuous subsets of E = E" are all of finite-dimen-

sion, and hence any subspace L of E satisfies the condition

that L flu 0 is weakly closed for all neighbourhoods U of O

in E'.

IO. A barrelled space which is not complete.

Let E be a separable infinite-dimensional Banach space

(eg. E = Zl). Then E contains a dense subspace M of count-

able infinite codimension, which is a Baire space. M is a

 38

barrelled space which is not complete.

ii. A barrelled space which is not metrizable.

Let E = ~, the space of finite sequences, equipped

with the finest locally convex topology v. Then (E,v) is a

barrelled space. Also it is complete (BE1 ~ , page 67). We

show that it is not metrizable: It is easy to see that E is

the strict inductive limit of a strictly increasing

sequence of finite-dimensional Euclidean spaces R n which

are Fr~chet spaces. Now suppose E is metrizable, then E is

a Fr~chet space and hence of the second category. But,

since the identity map Rn + R n+1 is "into" for each n, E

is a countable union of non-dense subsets and hence of

the first category which is a contradiction.

12. A Baire-like space which is not an unordered Baire-

like space.

Let w be the space of all real sequences with the

product topology and let E = {(ap)aW; ap= O whenever

p~(nk), for some (nk) with lim (nk/k) = O}. Then E is a

dense vector subspace of w and that every countable

! !

o(E ,E)-bounded subset of E is equicontinuous, so that

E is o-barrelled. Since E has its weak topology, E is

Baire-like (Cf. Todd, A.R. and Saxon, S.A. : A product of

locally convex spaces, Math. Ann., 206(1973), 23-34). But

E is clearly the union of the closed, proper (and hence

nowhere dense) subspaces Fn= {(ap)~E; an= O}, and thus E

is not unordered Baire-like.

13. A Baire-like space which is not a Baire space.

 39

Let {en} be the u n i t v e c t o r s in s If the scalar sequence

{bn}es has infinitely many non-zero entries, t h e n the span E

in s of {e } U{ ~ b. e } , w h e r e n. r a n g e s over all sub-

n i=l l ni l
s e q u e n c e s of the s e q u e n c e { 1 , 2 , . . . } , is d e n s e in s and so its

dual is m = s . If we a s s u m e that some pointwise bounded

subset B of E' = m is n o t n o r m - b o u n d e d , then we may use a

"sliding hump" argument (Cf. P ~7~) to o b t a i n a sequence

{hk}C B and a s u b s e q u e n c e {n i} of {1,2,... } s u c h that

I hk(l~
I'= b.l eni) I + ~

contradicting the f a c t that B is o(E',E)-bounded. Thus, E is

barrelled. Let fk be t h a t u n i q u e member of E' such that

fk(ej ) = ~kj ' k,j = 1,2 ....

We c h o o s e the sequence {bi}es so t h a t the n u m b e r of zero

entries between the n th and (n + i) th n o n - z e r o entries is

a strictly increasing function of n (n = 1 , 2 , . . . ) . Then, we

readily see t h a t every finite linear combination of the

sort d e f i n i n g E has a zero entry, and thus

E = u fk ( {0 }).
k=l

-i
But it is c l e a r that each fk ({0 }) is a c l o s e d proper

l-condimensional subspace in E, and h e n c e is n o w h e r e dense

in E. So, E is n o t a B a i r e space. But it is a B a i r e - l i k e

space, being a normed barrelled space.

14. A barrelled bornological space which is n o t the

inductive limit of B a n a c h spaces.

Cf. Valdivia, M. : A class of b o r n o l o g i c a l barrelled

 40

spaces w h i c h are not u l t r a b o r n o l o g i c a l , Math. Ann., 194

(1971), 43-51.

REMARK. The i n d u c t i v e limit of B a n a c h spaces is also

called ultrabornological. But we u s e this name for d i f f e r e n t

concept in C h a p t e r 4.

15. A bornological space w h i c h is not m e t r i z a b l e .

Let E = H R t, w h e r e R t is a copy of the real line

tE E0,1~
~. S i n c e E is an u n c o u n t a b l e product of m e t r i z a b l e spaces,

it is not m e t r i z a b l e (A p r o d u c t of m e t r i z a b l e locally convex

spaces is m e t r i z a b l e iff it is a c o u n t a b l e product). However,

)70
E is b o r n o l o g i c a l , since the c a r d i n a l i t y 2 of E0,1~ is

smaller than the s m a l l e s t strongly inaccessible cardinal.

(A c a r d i n a l do is s t r o n g l y inaccessible if (i) do > JY0,

(ii) ~{ d ; ~A}<d0 whenever cardinal of A<d0 and d < do for

all ~eA, (iii) d<d 0 implies 2d< d o . See BE20 ~, page 392).

16. A bornological space w h i c h is not b a r r e l l e d .

The space ~ of f i n i t e sequences equipped w i t h the

supremum norm topology is a n o r m e d space and h e n c e b o r n o -

logical. But it is not b a r r e l l e d . For, the s e q u e n c e {fn}

of c o n t i n u o u s linear functionals, defined by

fn(X) = xn , x = (xn) e~ ,

is w e a k l y b o u n d e d but not e q u i c o n t i n u o u s .

17. A barrelled space w h i c h is not b o r n o l o g i c a l .

Let W(~2) be the space of all o r d i n a l s less than the

initial ordinal ~2 of the f o u r t h class w i t h the i n t e r v a l

topology and let X be the s u b s p a c e of W(~2) whose elements

 41

are not ~ 0 - 1 i m i t s . T h e n the space X is ~ 1 - a d d i t i v e and is

has
not a Q-space, since i t / n o c o m p l e t e s t r u c t u r e s . F u r t h e r m o r e ,

by the n o r m a l i t y and the ~ 1 - a d d i t i v i t y of X, a n y c l o s e d and

bounded s u b s e t of X is f i n i t e and h e n c e satisfies the

condition of T h e o r e m 4. Thus C(X) is a b a r r e l l e d space w h i c h

is not b o r n o l o g i c a l .

18. A quasi-barrelled space w h i c h is n e i t h e r barrelled nor

bornological.

Let E be a b a r r e l l e d space w h i c h is n o t b o r n o l o g i c a l

and F a b o r n o l o g i c a l space w h i c h is not barrelled. T h e n the

product ExF is c l e a r l y a quasi-barrelled space. However,

E• is n e i t h e r barrelled nor b o r n o l o g i c a l : Let B be a

bornivorous convex set in E, w h i c h is n o t a n e i g h b o u r h o o d

of 0 in E. T h e n BxF is a b o r n i v o r o u s convex set in E•

which is not a n e i g h b o u r h o o d of 0 in ExF. This shows that

ExF is not b o r n o l o g i c a l . Next, let B I be a b a r r e l in F,

which is not a n e i g h b o u r h o o d of 0 in F. T h e n ExB I is a

barrel in ExF, which is not a n e i g h b o u r h o o d of 0 in ExF.

This shows that ExF is not b a r r e l l e d .

19. A quasi-M-barrelled space w h i c h is not q u a s i - b a r r e l l e d .

The r e q u i r e m e n t of b e i n g q u a s i - M - b a r r e l l e d depends

o n l y on the d u a l s y s t e m and so, w e a k e n i n g the t o p o l o g y of

quasi-M-barrelled space w i t h o u t affecting the dual, would

still leave it q u a s i - M - b a r r e l l e d . Let E be an i n f i n i t e -

!
dimensional Banach space. Then (E, ~ ( E , E )) is a q u a s i - M -

barrelled space w h i c h is not q u a s i - b a r r e l l e d '

20. A semi-bornological space w h i c h is not an S - b o r n o l o g i c a l

space (and h e n c e n o t a b o r n o l o g i c a l space).

 42

2
Consider ~2 = {x = (Xn) ; Z IXnl < ~}.
n=l

Then (~2, o(~2, ~2)) is a semi-bornological space which

is not S-bornological : Let u be the norm topology on s

Since the space (~2,u) is normable, the space

(s o(Z2, s is semi-bornological (B[3~, page 1 9 0 ) . We

show t h a t (~2, 0(s s is not S-bornological. For t h i s ,

we must f i n d a convex, circled and 0(s 2, s
set in s which is not a o(s 2 , s neighbourhood

of 0 in ~2 . Consider the unit ball

B = {xe~ 2 ; llx]I <i}

B is a convex, circled and u-neighbourhood of 0 in ~2

Consequently B is a convex, circled and u-bornivorous set.

Since the o(~ 2, ~2)-bounded subsets of Z2 are the same as

the u-bounded subsets of Z2 , B is a convex, circled and

0(~ 2, ~2)-bornivorous set in Z2 . We show that B is not a

o(~ 2, ~2)-sequential neighbourhood of 0 in ~2 . Consider

the sequence (en) of unit vectors in ~2 . e n is the

sequence in ~ with 1 in the n th place and 0 in all other

places. The sequence (en) is weak convergent to the zero

sequence in ~, that is, (en) is o(~ 2, ~ ) - c o n v e r g e n t to 0

in s . Since each element e n of this sequence has norm 1

and is not in B, (en) is not ultimately in B. Clearly, B

is not a o(~ 2, Z2)-sequential neighbourhood of 0 in ~2.

21. An S-bornological space which is not C-sequential

(ant hence not a bornological space).

The space (~i, o ( ~ , ~)) is S-bornological: We show

that (~i (~i ~)) is a braked space

 43

(E is a braked space if given any sequence {Xn} in E, which

converges to 0, there exists a sequence of positive real

numbers {Xn} such that %n § + ~ and the sequence {Xn Xn} in

E converges to 0). But then, it is S-bornological, because

a braked space is S-bornological (P[8~, page 279). Let u

be the norm topology on ~1. Let {x (n)} be a sequence in ~
converging to 0 in the weak topology o(Z 1, ~=). Then {x (n)}

is u-convergent to O, because weak convergence and norm

convergence of sequences in ~i are the same. Since (Zl, u)

is normable, it is braked. Consequently there exists a

sequence of positive real numbers {%n} such that %n § +

and the sequence {Xn x(n)} is u-convergent to O. Clearly

{~n x(n)} is o(~ I, ~ ) - c o n v e r g e n t to O. Hence (~I,o(~I,~))

is a braked space and hence an S-bornological space.

(zl, o(Zl, ~=)) is not C-sequential because the unit ball

i
B = {xc~ ; llxll<l}

is a convex, circled and o(~ I, ~= )-sequentially open

subset of Zl which is not o(~ I, s

22. A C-sequential locally convex space which is not S-

bornological (and hence not bornological).

Consider the space (Z2, o(~2, ~2)cs) which is the C-

sequential locally convex space generated by the locally

convex space (~2, o(~2, ~2)). We show that (~2, o(~2,~2)cs)

is not S-bornological. For this we must find a convex,

circled and o(~ 2, ~2)c s -bornivorous set in Z2 which is

not a o(~2,~ 2 )cs-Sequential neighbourhood of 0 in Z2.

As shown in ~ 20, the unit ball
 44

B = {xeZ 2 ; IIxlI<1

is a convex, circled and ~(s s s set in s

Since (~2, ~(s ~2)) and (s o(~ z, ~2)cs) have the same

convergent sequences, they have the same bounded sets.

This is because of the sequential characterization of

bounded sets. A set A in a topological vector space is

bounded iff given any sequence {x (n)} in A an@ given any

sequence {ln} of positive real numbers such that lim ~ = 0,

n~+~ n
we h a v e lim %nXn = O. Consequently B is a convex, circled
n~+~
and o(~z, ~2)cs_bornivorou s set in ~2. Of course, B is not

a ~(s s neighbourhood of 0 in ~2. The

sequence {en} of unit vectors in s is o(Z2,s

gent to 0 in ~2, since it is ~(s s to 0 in

~2. However, every term en of this sequence has norm 1

and is, therefore, not in B.

23. A Hackey space which is not quasi-barrelled.

Cf. ~ 26 , and ~- 37.

24. A Mackey space which does not have property (S).

The locally convex space (Zl, : (Zl, Co)) , where

:(Z I, Co) is the Mackey topology, is a Hackey space which

does not have property (S), because (Co, :(c 0 , Zl)) is

not sequentially complete.

25. A Hackey space with property (S) but without property

(C).

The Mackey space ( ~ , 9 ( ~ , ~:)) has property (S)

(Cf. B [ ~ , Page 374). To show that it does not have

 45

property (C), we p r o c e e d as follows: Let

B = {e n , n = 1,2,...}

be the canonical Schauder basis of ~i where e is the

' n

sequence of all zeros except the n th coordinate w h i c h is i.

B is ~(~i, ~)-bounded, but has no ~(Z~, Z~)-accumulation

point in Zl. For, suppose y~Zl. There exists a sequence

~E~of the form

= (0, ., 0,2,2, .)

such that I <~, Y > 1<i. Thus,

<~, e n- y > = <~, en > - <~, y > >I

holds true for n big enough that < ~, e n > = 2. Then

e n- yE{~} ~

for atmost finitely m a n y nr Therefore y is not a

a(~ I, ~ ) - a c c u m u l a t i o n point of B.

26. A semi-reflexive space w h i c h is not reflexive.

Let E be a n o n - r e f l e x i v e Frechet space (for example,

T !
E = co). Then the M a c k e y space (E , T (E ,E)) is a semi-

reflexive space w h i c h is not reflexive, because its

strong dual E is not reflexive, (The strong dual of a

reflexive space is reflexive).

27. A barrelled space w h i c h is not a Montel space.

An i n f i n i t e - d i m e n s i o n a l Banach space is clearly a

barrelled space. However, it is not a Montel space, because

a normed Montel space is locally compact and hence finite-

dimensional.
 46

28. A reflexive space which is not a Montel space.

Any infinite-dimensional reflexive Banach space is an

example of a reflexive space which is not a Montel space.

(For example, consider the space s I< p<~).

29. A Frechet space which is not a Schwartz space.

Let E be any infinite-dimensional Banach space. As

shown in ~ 27, E is not a Montel space. Clearly E is a

Frechet space. Since a Frechet Schwartz space is a Montel

space, it is clear that E is not a Schwartz space.

30. A Schwartz space which is not a Montel space.

Let E be any infinite-dimensional vector space with

algebraic dual E* = F. We show that (E, O(E,F)) is a

Schwartz space. For this, we show that it satisfies the

condition (S) of Theorem II: Every bounded subset of E is

relatively compact in (F , o(F ,F)') and thus precompact in

E. Let

U = {x ; I <x' Yk > I ~ ' l<k<n}

be a neighbourhood of O, where we may suppose that the

vectors Yk are linearly independent. Let M be the subspace

of F generated by the Yk (l<k<n), N an algebraic supplement

of M and {zz}~e I a basis of N. We choose V = U and

A = {x ; l<x, yk>I~E, l<k<n,

l<x, zs s }

for the set A which appears in the condition (S) of

Theorem ~I, indpendently of e. In the first place, A is

bounded because, if
 47

n
y = ~ nky k + S ~zgcF,
k=1 gel

We h a v e
n
[< x , y >1<~( E Ink[ + E I~$1).
- k=l ~.EI

By an e l e m e n t a r y algebraic consideration, there exists a

basis {x k ; l < k < n } i n N f o r w h i c h

< Xk, y~> = O if k ~

and

< Xk, yk > = I.

For e a c h xEU, set

n
u = ~ < x , Y k > x k-
k=2

Then, u~A, since

I< u, Yk >I = I< x,Yk >I

and
u, z~> = O.

Finally

x - u~U
since

< x - u, Yk > = < x ' Y k ~- ~ x ' Y k > = O.

Thus,

Ucc~U + A.

Hence (E,~[E,F)) is a S c h w a r t z space by T h e o r e m ii. But it

is not barrelled (See =/P 47).

31. A Montel space which is not separable.

 48

The space ~d' d> ~0 ,of all sequences with d non-zero

coordinates, equipped with the finest locally convex topology,

is a M o n t e l space which is n o t separable.

32. A Montel space (and h e n c e a reflexive locally convex

space) which is n o t c o m p l e t e .

Consider the following notations:

{d n} = the s e q u e n c e of c a r d i n a l n u m b e r s such that

~u d
do = 2 and d = 2 n-l,
n

{~n} = H R ,
eeA
n

{~n } = @R
eeA ~ '
n

where each R n is the r e a l line and A n is an i n d e x set of

power d n. ~n,0 = {(xa)~Wn ; xa = 0 except for c o u n t a b l e a},

{e~} = a Hamel base of Wn, 0

{e2} = the b a s e of ~n' that is, ena = (xs) such that

xa 1 and x8 0 for ~ ~ a.

(e2} = a Hamel base of ~n-1 such that for a n y X + 0 ,

X s ~n_1 , the set (e~ ; <ina ' X> # 0} is u n c o u n t a b l e .

Sn the linear operator on ~ n , 0 to ~n s u c h that Sn ena ena

T the linear operator on ~n to ~ n _ i S u c h t h a t T n e na = e na "

n

En = tT e n- i
n

E the v e c t o r s u b s p a c e of ~ n x Wn+1 generated by

n
n n+l
(( e~ , e )}

E = K E
n
n
 49

F = @n ~n "

(E, ~(E,F)) is a M o n t e l space which is not complete

(Cf. P~0] for d e t a i l s ) .

33. A distinguished space which is not semi-reflexive.

Let E be a n o n - r e f l e x i v e Banach space (for e x a m p l e ,

E=c0). Since E is a n o r m e d space, it is d i s t i n g u i s h e d . But

E is n o t semi-reflexive, because a locally convex space is

reflexive iff it is s e m i - r e f l e x i v e and q u a s i - b a r r e l l e d .

34. A Frechet space which is not d i s t i n g u i s h e d .

Let E be the v e c t o r space of all n u m e r i c a l double

sequences x = (xi~)J such that for each naN

(n)
Pn(X) = laij xijI< +
i,j

where
j for i< n
(n)
a. . ~
13 1 for i > n .

The semi-norms {pn } g e n e r a t e a locally convex topology under

which E is a F r e c h e t space. The d u a l E' can be i d e n t i f i e d

with the space of d o u b l e sequences u = (uij) such that

(n)
luijl~ c aij for all i,j and suitable c>0 , nan . Let

Bn = U~ where U n = {x ; Pn(X)<_ i} , and W the convex circled

hull of ~ ~n Bn . Then W is b o r n i v o r o u s in E' and W c o n t a i n s

no element ueE' with the p r o p e r t y that, for e a c h i, t h e r e

exists j with luijl~ 2. We show t h a t W can c o n t a i n no s t r o n g

!

neighbourhood of 0 : For each strong neighbourhood B ~ in E ,

B a bounded subset of E, there exists a sequence P = (Pn)

of s t r i c t l y positive numbers such that FnPnBnCB ~ . Now

 50

define elements u ( n ) E E ' so t h a t

u! n ) = 0 for (i j) ~ (n, k n)
13

and
(n)
U =i
n,k n

where k n is c h o s e n so t h a t

n+l (n)
2 u e Pn Bn

For e a c h g i v e n p , the s e q u e n c e w i t h g e n e r a l t e r m
N (n)
s = 2 E u is a w e a k C a u c h y s e q u e n c e in E' and h e n c e
N 1
c o n v e r g e n t to seE' Now, s eB ~ for all Ne N. H e n c e seB ~
N
but s%W. It f o l l o w s that W does not contain B~ Thus, we

have shown that (E', 8(E',E)) is n o t b o r n o l o g i c a l and h e n c e

not quasibarrelled.

35. A distinguished space whose strong dual is n o t

separable.

The Banach space s is a d i s t i n g u i s h e d space. But its

strong dual s is not separable.

36. A distinguished space whose strong dual is n o t m e t r i -

zable.

The s p a c e w of all sequences equipped with the n o r m a l

topology is a F r ~ c h e t space, but its strong dual

(w, 8(w x , w)) = (r 8(~, w)) is a n o n - m e t r i z a b l e barrelled

space, where ~ is the space of finite sequences. (To a

X
sequence space l, we d e f i n e I to be the sequence space

consisting of all sequences u = (u i) for w h i c h the scalar

product ux = E u.x. converge absolutely for all x e l

i=l ii
For example w = ~, w . Now let I D ~. Then I and l•
 51

form a dual pair. Define the seminorms

Oo

Pu(X) = Z luil Ixil , uEl x

i=l

on I. Then the family {pu} of s e m i - n o r m s defines a locally

convex topology on I called normal topology. See B[20~ ,

page 407).

37. A distinguished space w h i c h is not q u a s i - b a r r e l l e d .

Let F be a n o n - r e f l e x i v e Banach space (for example,

F = Co , the B a n a c h space of s e q u e n c e s converging to 0).

Let E be the M a c k e y space (F', T(F', F)). T h e n E' = F and

the strong topology on E' is the n o r m topology, because the

~(F' , F)-bounded sets are norm-bounded. Thus, E is semi-

reflexive but not r e f l e x i v e so that E is not q u a s i - b a r r e l l e d .

(A l o c a l l y convex space is r e f l e x i v e iff it is semi-

reflexive, and q u a s i - b a r r e l l e d ) . Since E is semi-reflexive,

it is d i s t i n g u i s h e d .

38. A bornological space w h o s e strong bidual is not

bornological.

Let A be an infinite set and {fi } a s e q u e n c e of p o s i t i v e

functions on A such that i<j implies fi ~ fj T h e n we

obtain a bornological space E as the i n d u c t i v e limit of

Banach spaces E i whose unit sphere is {f ; If(1) I~fi(1),leA} 9

We show that E" is not b o r n o l o g i c a l when functions fi are

adequately chosen. Let F be the t o t a l i t y of f u n c t i o n s g on

A such that

Z f. (I) Ig(1)I< + ~ , i = i, 2, ....

IEA ~

Then F can be c o n s i d e r e d as a s u b s p a c e of E' (as f u n c t i o n a l s

f + Z f(1)g(1)) and also as the d u a l of a subspace E0 of E

lea
 52

which consists of all functions feE w h o s e values are 0

except for a f i n i t e number of I. If r eE' is c o n t i n u o u s on

E by the o r d e r topology, then ~eF. So, there is a c o n t i n u o u s

projection of E' o n t o F. Then there is a l s o a continuous

projection of E" o n t o F' and h e n c e F' is a h o m e o m o r p h i c image

of E". Therefore, it is s u f f i c i e n t , for our p u r p o s e , to c h o o s e

fi so as to m a k e F' non-bornological. But this has already

been found in 2 = 34 - - A is the set of p a i r s (n, m) of

positive integers and

m if n< i
f (n, m) = {
1
1 if n > i

39. An (LB)-space which is n o t q u a s i - c o m p l e t e .

Let E 0 = c o , with the e l e m e n t s written as d o u b l e

sequences x = (xij). x e E 0 iff i,lim

lj+~ x i j I = 0 " Clearly E0 is

a Banch space under the n o r m I Ixl I= sup Ixijl Let

i,j
j if i < n
a (n) {
13
1 if i > n .

Let E be the space of all d o u b l e sequences with

n

lim Ixijl = 0
i,j§ a!~ )
13
and n o r m
= sup Txi I
llXIln (n)
i,j aij

E n is o b t a i n e d f r o m E 0 by m a k i n g a diagonal transformation

and so it is t o p o l o g i c a l l y isomorphic to c o The embedding

En-l+ E n is c o n t i n u o u s . The topological inductive limit

E = lim En, therefore exists, provided that the h u l l topology

u on E is H a u s d o r f f , and E is t h e n an (LB)-space. If w e p u t

a.. = k for all i and j, the n o r m

13
 53

ixijl
llxil~ = sup
aij

is w e a k e r than llxl In on E n , and so it d e f i n e s a

Hausdorff topology on E w h i c h is w e a k e r than u. Thus,

u is a l s o Hausdorff. E' c a n be identified with the

space of a l l d o u b l e sequences u = (uij) for w h i c h

(n)
n = i, 2, ...
i , j = l l U i j I a .13 < ~,

Let e = (i, i, ... ). L e t B be the set of a l l v e c t o r s

e (n) = (e . . . . . e, 0, 0, ...), e occuring in n places,

n = i, 2, ... B is c o n t a i n e d in E, and is w e a k l y

bounded. Since e (n) E E n - E n _ l , B is n o t c o n t a i n e d in

a n y o n e of the spaces E If E w e r e quasi-complete, the

n
closed circled convex cover of B, w h i c h would be c o m p l e t e ,

would have to lie in s o m e E -- because every circled

n
convex bounded and complete subset of an (LB)-space

(E, u) = ~(En, u n) is a b o u n d e d subset of s o m e (E k, u k)

[2~ w afortiori, the same would b e t r u e of B.

40. A locally convex space which is n o t r e f l e x i v e (not

even semi-reflexive) but its strong dual is r e f l e x i v e .

L e t E be a d e n s e proper subspace of an i n f i n i t e -

dimensional reflexive Banach space with the relative norm

topology. Then E' is r e f l e x i v e while E is n o t semi-

reflexive.

41. A countably barrelled space which is n o t b a r r e l l e d .

 (i) It is a w e l l - k n o w n fact (Cf. B ~ or B D5] )

that the strong dual of a m e t r i z a b l e locally convex space

is a c o m p l e t e (DF)-space ; a (DF)-space is c o u n t a b l y

quasi-barrelled and a c o m p l e t e countably quasibarrelled

spa~e is c o u n t a b l y barrelled. But it need not be quasi-

barrelled (and so neither barrelled nor bornoligical) as

shown in ~/= 34.

(ii) Let W be the space of o r d i n a l s less than the

first u n c o u n t a b l e ordinal. Then the space C(W) of all

continuous (real or complex) functions on W, e q u i p p e d with

the c o m p a c t - o p e n topology is a c o u n t a b l y barrelled space

by T h e o r e m 13, b e c a u s e the c l o s u r e of a c o u n t a b l e union of

compact subsets of W is compact. But C(W) is not barrelled,

since W is p s e u d o - c o m p a c t but not compact.

42. A locally convex space C(X) of c o n t i n u o u s functions

which is not c o u n t a b l y barrelled.

Let W be the space of o r d i n a l s less than or equal

to the first u n c o u n t a b l e ordinal, and let T be the

Tychonoff plank. Then

co

T = U (W x{n}) .
n=l

Clearly, each W x{n} is c o m p a c t and T is p s e u d o - c o m p a c t

but not compact. Hence C(T) is not c o u n t a b l y barrelled.

43. A semi-reflexive countably barrelled space w h i c h is

 55

not barrelled.

Let h be an uncountable set. If E is the direct sum

~(h) and E' is the subspace of the product ~h consisting

of all x = (~k)kr A for which atmost countably many ~k

are non-zero, < E,E'> is a dual pair. Given an atmost

countable non-empty subset T of A and a family (ak)Xr z

of real numbers, the set

{(~)k~h ; l~l~aX for ~E~ ,

~k = O otherwise }

is o(E', E)-compact~ circled and convex. The topology on E

of uniform convergence on all such sets is a topology of

the dual pair < E,E'> , under which E is countably

barrelled. Also (E, ~(E,E')) is semi-reflexive.

(E, T(E,E')) is not barrelled, since

~h

is o(E',E)-closed and bounded but is not o(E',E)-compact.

44. A countably quasi-barrelled (and hence o-quasi-

barrelled) space which is not o-barrelled.

The space ~ of ~/= 16 is a quasi-barrelled space and

hence countably quasi-barrelled. But it is not o-barrelled,

because a quasi-barrelled o-barrelled space is barrelled

and ~ is not barrelled as shown in =~ 16.

45. A a-barrelled space which is not a Mackey space.

Let X be an uncountable vector space, and we denote

 56

by ~ ( r e s p e c t i v e l y , ~ ) the vector space of real-or complex-

valued functions on X which are different from 0 for a

finite (respectively, an infinite countable) number of

elements of X. Then the spaces ~ and ~ form a dual pair

separating points which permits us to consider the weak

space F associated to ~ . We denote by E the space

equipped with the topology generated by the family of semi-

norms

p~ (f) = sup {I~(f) I ; ~e~}, feS,

where ~ varies over the countable family of bounded sub-

sets of F. We note that for each ~, there is a countable

subset ~ of X such that ps(f) = 0 for each feE which is

zero on % . We can immediately deduce that each neighbour-

hood of 0 in E contains a nontrivial subspace of E. We show

that E is ~-barrelled. For this, it is sufficient to show

! !

that F = (E , ~(E , E)). Now, if ~ is a continuous linear

functional on E, there exists a ~ such that

! % (.).

We,then, have ~(f) = 0 for each fEE, which is zero on % .

The conclusion now follows. Next, we show that E is not a

Mackey space. For this, it is sufficient to prove that

@ = {feE; If(x) l < i, for each xeX}

is a neighbourhood of 0 equipped with the Mackey topology,

for @ contains no nontrivial subspace of E. Now, if # is

a linear functional on E which is not continuous, there

exists E>O such that Ir l~a for an infinite set of points

xeX. The conclusion now follows from the theorem of

bipolars since @ is a barrel in E.

 57

45(a).A o-barrelled space which is not countably quasi-

barrelled (and hence not countably barrelled).

Let the space X of -~ 45 be, in addition, a countable

union of uncountable sets X n which are increasing and such

that X-X n is uncountable for some n sN. We define E I as

the space ~ endowed with the topology given by the system

of semi-norms obtained by filtering those of E and the

semi-norms

pn(f) = sup {If(x) l ; XeXn}, fE~ .

We show that El is o-barrelled: Note that F is also equal

T !

to (El, o(EI,E)). Infact, for each n ~ , we have

@ c b n = {feEl ; Pn (f)jl} ,

and so each linear functional bounded on b n is bounded on

@, and hence continuous on E. The conclusion then follows

from the following result: If G is o-barrelled, it is o-

barrelled for each system P of semi-norms lying between

those of G and those of Mackey space associated to G. (To

!

prove this, we observe that each bounded sequence of (Gp ,

O(Gp , ~)) is a bounded sequence of (G , o(G , G)) and

hence is equicontinuous in G and, a f o r t i o r i , in ~ ) . Now

we show that E I is not countably quasi-barrelled. Denote by

Xn ( n ~ ) the set of characteristic functions of points of

X n. Each Xn is clearly equicontinuous on E~ and X =u X

t !
is bounded in (El, B(EI , E)) because X is included in O ~

that is, in a compact, circled and convex subset of

! !

(E I , o(E1 , E)). However, X is not equicontinuous on El.

46. A Mackey space which is not o-quasi-barrelled.

The Mackey space E = ( ~ , ~(~, ~i)) has property (S)

 58

but not property (C) as shown in ~, 25. Since a u-barrelled

space has property (C], E is not c-barrelled. If E were o-

! !
quasi-barrelled, then, since (E , ~(E , E)) is sequentially

complete, E would be u-barrelled which is not true.

47. A locally convex space which has property (C), but is

not o-barrelled.

Let E be an infinite-dimensional vector space with

algebraic dual E . Then (E, o(E, E )) has property (C),

because every o(E , E)-bounded subset of E is relatively

o(E ,E)-compact. To show that (E, o(E,E )) is not o-

barrelled, let B be any Hamel basis of E. Let B0 be a

countably infinite subset of B. For xeB0, let fx be that

linear functional in E such that

fx(X) = 1

and

fx(y) = O, y~B-{x}

The s e t

A = {fx ; x~B0}

is a countable o(E',E)-bounded subset of E*. A ~ contains

no finite-codimensional subspace of E. Then A is not equi-

continuous and so (E,o(E,E*)) is not u-barrelled.

48. A sequentially barrelled space which is not u-quasi-

barrelled (and hence not u-barrelled).

It i s known t h a t the space ~ is a perfect sequence

s p a c e and ~1 i t s K~the d u a l . Then (Z~, r ( ~ ~, ~1)) is

sequentially barrelled, b e c a u s e a KOthe f u n c t i o n space A

is sequentially barrelled i n t h e Mackey t o p o l o g y T(A,A ) -

 59

In a KSthe function space A, each Cauchy sequence for the

weak topology is weakly convergent P[l~--and hence the

same is true of a perfect sequence space. But (s ,T(s ,s ))

is not a-quasibarrelled as shown in 4 = 46.

49. A sequentially barrelled space which does not have

property (S).

(s T(s Co)) is a sequentially barrelled space,

because a perfect sequence space with Mackey topology is
sequentially barrelled (P~). Clearly, it does not have
property (S).

50. A quasi-complete locally convex space which is not

sequentially barrelled.

Let w be the space of all sequences which is a perfect

sequence space. Let mbe the set of all sequences (ai)

such that a.= 0 or 1 and

1

1 n
-- Z a.-~ 0 as n +o~.
n i=I x

Define

w0 = {ax = (a i xi) ; a = (ai]r , x = (xi)ew}

Denote by lol(w *, w) the normal topology on W* . T h e n we

have

(A) : B(w , w0) = ~(w , w) ;

(B) : ~(w , w0) i I~I (w , Wo)i~(w , w0) ;

Hence a subset of w is o(w , wo)-bounded iff it is

I~l(w , w~

(C) : gvery I~l(w , w0) - bounded subset of w is

lol (w*, w)-bounded. (For proofs, see P[9~).

Let ~ be the space of all finite sequences. Then

 60

o(~,w)-bounded subsets of r are finite-dimensional; hence

by (B) and (C) the o(~,w0)-bounded subsets of ~ are also

finite-dimensional. Thus (r T(r w0)) is quasi-complete.

Now (w 0 , o(w0, r is a barrelled metrizable space with

completion w and bidual w by (A). Hence (# , T(~ , w0))

is not sequentially barrelled (because, if E is a m e t r i z a b l e

locally convex space with (E', T(E', E)) sequentially

barrelled, then E is c o m p l e t e ~ E 9 2 ] ).

51. A (DF)-space which is not countably barrelled.

The space ~ of ~ 16 is a normed space and hence a

(DF) space. But it is not o-barrelled as shown in ~ 44,

and hence it is not countably barrelled either.

52. A (DF) space which is not quasi-barrelled.

Cf. ~ 41 (i).

53. A quasi-barrelled (DF)-space which is not bornological.

This counterexample is based on the following result:

(*) Let I be an echelon space defined by the increasing

system (k) , k = 1,2,..., such that every (k)__ Ira(k)}

~ n
n=l
is a sequence of strictly positive numbers.
Suppose that, for every positive integer p, there is a
strictly increasing sequence {pn }~ of positive integers
n=l
such that
(k) (i)
is a bounded sequence for
(I) {aP n / aP n ~ = 1

k = 1,2,..., p .

(p+1) (i)
(2) lim (apn / a ) =
n§ Pn

Then there is a dense subspace E in (Ix, o(Ix,l)) such that

 61

(E,B(I• is a n o n - b o r n o l o g i c a l quasi-barrelled space.

This r e s u l t as well as the c o u n t e r - e x a m p l e that follows

are due to M . V a l d i v i a (Math. Z. 136 (1974), 249-251). For

information about echelon space I and c o - e c h e l o n space I x,

See B [2~ , page 419. For the example, we proceed as

follows:

We choose a sequence a (k) = i~a(k)}

n ~ defined in the
n=l
following way. Given a p o s i t i v e integer n, we write
h-i
n = (2m-I)2 , m and h being p o s i t i v e integers. If h<k,
h-1
we put a (k)= (2m - 1) 2 If h>k, then a (k)= i. Obviously,
n n
the system (k) , k=l,2,..., is increasing. Given a positve

integer p, we put Pn = (2n - i) 2 p, n = 1,2, . . . . Then

a (k)= 1 , k=l,2 ..... p and a (p+l) = (2n - i)2 p. Hence,

Pn Pn
~k) (1) (p+1) (1)
apn / apn i, k = 1 , 2 , . . . , p and n§ / apn ) = ~
= (apn

and, therefore, conditions (i) and (2) of the r e s u l t (*) are

satisfied.
^
Now, we a p p l y ( * ) t o obtain a space (E, 8(I , l)) w h i c h

is q u a s i b a r r e l l e d and n o n - b o r n o l o g i c a l . Since there is a

X
countable fundamental system of b o u n d e d sets in (E, 8(I ,I)),

(E, 8(I • I)) is a (DF)-space.

54. A locally topological space w h i c h is neither a borno-

logical space nor a (DF)-space.

Cf. Adasch, N., Ernst, B. and Keim, D.: Topological

vector spaces, Lecture Notes in Mathematics, 639 (1978),

Springer-Verlag, Berlin-Heidelberg-New York.

(A locally topological space w h i c h is a common genera-

lization of b o r n o l o g i c a l and (DF)-spaces is d e f i n e d as

follows: Let B be the family of all b o u n d e d and c i r c l e d

sets in a l o c a l l y convex space (E, u) such that

 62

B,, B2eB=>BInB2, BI + B2~B 9 A linear map f: E+F, F a

locally convex space, is called locally continuous if all

restrictions f/B,BCB , are continuous at 0 in the topology

induced by u on B. (E,u) is called a locally topological

space if all locally continuous maps from E into any

locally convex space F are continuous).

55. A k-quasibarrelled space which is not k-barrelled.

Let E be the vector space of complex sequences with

only a finite number of n o n - z e r o elements, equipped with

the topology of pointwise covergence. Then E is m e t r i z a b l e

and so q u a s i b a r r e l l e d which imp~es that E is k-quasi-

barrelled. We now show that E is not k - b a r r e l l e d for any

k>l. Consider

U n = {xcE ; ]xj[~l, l<j<n}

Each U n is a circled, convex and closed n e i g h b o u r h o o d of

0 in E. Their intersection is a barrel in E. Let

U(k) = N {k n Un} , khl

n=1

Let
= en a(n)
an e , = (al, a2,..., an,O,O, ... ).

Then B = (a (n) 9 n
, = 1,2 , ,- .} is a bounded subset of E

that is not absorbed by U(k) for any k>l. Hence E is not

k-barrelled for any k>l.

56. An H-space which is not a d i s t i n g u i s h e d space.

Cf. ~= 34.
 63

57. An H-space which is not metrizable.

Let r be the space of all finite sequences and u the

normal topology on ~. Then it is the same as the locally

convex direct sum topology on ~ and hence (~, u) is not

metrizable (See ~ Ii). But (~, u) is a distinguished

space, because (~, 8(r162 = (w, 8(w,r is a Frechet

space. Hence (~, u) is an H-space.

58. An H-space whose strong dual is not separable.

cf. ~ 34.

OPEN PROBLEMS

I. Do the concepts of Baire and unordered Baire-like

locally convex spaces coincide?

2. Is every countable-codimensional subspace of a Baire

space of the same kind? (This is satisfied by unordered

Baire-like spaces -- Math. Ann. 206(1973), 23-34. So, the

invalidity of this problem would distinguish between

Baire and unordered Baire-like spaces).

3. (I.Tweddle): Is there a locally convex space E such

!
that (E, T(E, E )) is countably barrelled but not barrelled?

4. Is there a bornological space whose strong dual is

not countably barrelled?

5. (Levin and Saxon P [ 5 ~ ) : Is a Mackey space with

property (C) always a a-barrelled space?

6. If F is a g-barrelled space, is it necessarily

countably barrelled under the topology of uniform conver-
T
gence on the ~(F ,F)-bounded separable sets?

7. (Crothendieck): Is there a barrelled (DF) space

which is not bornological?

 CHAPTER 4

SPECIAL CLASSES OF TOPOLOGICAL VECTOR SPACES

Introduction

A topological space X which satisfies the condition of

Theorem 4 in the introduction to Chapter 3 is referred to

as an N-S space. A topological space X is said to be an

L-W space if for every decreasing sequence {F n} of closed

and noncompact subsets of X for which nFn# 0, there exists

fr unbounded on each F n. C(X) is a barrelled space

whenever X is an L-W space.

A topological vector space E is W-barrelled if every

closed, circled and absorbing set in E is a neighbourhood

of O.

THEOREM I. Let X be a topological space which is

first countable at a point p which has no compact neigh-

bourhood. Then X is not an L-W space. (Cf. P [ 5 ~ )

THEOREM 2. If C(X) is W-barrelled, then X is an L-W

space. (Cf. PISS)

A closed and circled subset B of a topological vector

space is called an ultrabarrel (a bornivorous ultrabarrel)

if there exists a sequence {B n} of closed, circled and

absorbing (closed, circled and bornivorous) subsets of E

such that BI+ BI ~_B and Bn+1+ Bn+ r ~ B n for all n~l. The

sequence {Bn} is called a defining sequence for B. B is

called a suprabarrel (bornivorous suprabarrel) if the

closedness of B and Bn, for each n, is dropped from the

above definition.
 66

A topological vector space E is called ultrabarrelled

(quasi-ultrabarrelled, ultrabornological) if each ultra-

barrel (respectively, bornivorous ultrabarrel, bornivorous

suprabarrel) in E is a neighbourhood of O. Ultrabarrelled

and ultrabornological spaces are quasi-ultrabarrelled. A

locally convex ultrabarrelled (quasi-ultrabarrelled, ultra-

bornological) space is barrelled (quasibarrelled, borno-

logical).

If (E,u) is a topological vector space, we w r i t e u ~176

to denote the finest locally convex topology coarser than

u. If (E,u) is metrizable, so is (E,u ~176and so, if

! T
< E, E > is a dual pair, u ~ 1 7 6= T ( E , E ).

PROPOSITION i. If (E,u) is a metrizable topological

! !
vector space with dual E , < Eu, E > is a dual pair and

(E,u ~176 is ultrabarrelled, then u = u ~176 (Cf. P [ 7 ~ )

For each integer k~l, let {V~ n) ; n = O,1,2 ...}

be a sequence of closed and circled neighbourhoods of O

in a topological vector space E such that

Vk(n+1) + Vk(n+l) =_ Vk(n)

for all n. If, for each n,

V (n) = o Vk(n)
k=l

is absorbing (bornivorous), then V (~ is an ultrbarrel (a

bornivorous ultrabarrel) in E with {V (n) ; n = 1,2...} as

a defining sequence. V (~ is called an ultrabarrel (a

bornivorous ultrabarrel) of type (~). A topological vector

space E is called a countably ultrabarrelled (countably

 67

quasi-ultrabarrelled) space if each ultrabarrel (borni-

vorous ultrabarrel) of type (a) is a neighbourhood of 0

in E. Every ultrabarrelled (quasi-ultrabarrelled) space

is countably ultrabarrelled (countably quasi-ultrabarrelled)

and every countably ultrabarrelled space is countably

quasi-ultrabarrelled. A locally convex space which is

countably ultrabarrelled (countably quasi-ultrabarrelled)

is countably barrelled (countably quasibarrelled).

A topological vector space E is called k-ultrabar-

co

relled (k-quasi-ultrabarrelled) if N k j Vj (~ is a
j=1
oo

neighbourhood of 0 whenever V (~ = n V! ~ is an ultra-

j=l J
barrel (bornivorous ultrabarrel) of type (a). A k-ultra-

barrelled space is k-quasi-ultrabarrelled. A locally convex

space which is k-ultrabarrelled (k-quasi-ultrabarrelled)

is k-barrelled (k-quasi-barrelled).

A locally semiconvex space E is called hyperbarrelled

(quasi-hyperbarrelled) if every closed, circled, semi-

convex and absorbing (bornivorous) set in E is a neighbour-

hood of O. E is hyperbornological if every circled, semi-

convex and absorbing set in E is a neighbourhood of 0.

Clearly hyperbarrelled and hyperbornological spaces are

quasi-hyperbarrelled. E is called ~-hyperbarrelled (~-

quasi-hyperbarrelled) if every closed, circled, semi-

convex and absorbing (bornivorous) set V in E is a

neighbourhood of 0 whenever it satisfies the following

condition : V = n V , where, for some ~>0, each V is

closed, circled and h-convex neighbourhood of 0 and the

cardinal of r is ~. (V is ~-conv~x if V a + V c ~V~).

 68

A hyperbarrelled (quasi-hyperbarrelled, ~-hyperbarrelled)

space is %-hyperbarrelled (%-quasi-hyperbarrelled, ~-

quasi-hyperbarrelled) for every ~.

THEOREM 3. If {fn } is a pointwise convergent sequence

of continuous linear msps from an ~0-hyperbarrelled space

E into a locally semiconvex space F, then its limit map

is continuous. If, in addition, F is sequentially complete,

{fn } is necessarily pointwise convergent if it is point-

wise bounded on a set which is everywhere dense in E.

(Of. P [ 3 ~ )

Let f be a linear map from a topological vector

space E into another topological vector space F. We say

that the filter condition holds if for any Cauchy filter

basis ~ on E such that f ( ~ ) is convergent to a point of

f(E), it follows that ~ is convergent to an element in E.
If E is complete, the filter condition holds. Let E
be a topological vector space under each of two topo-
logies u and v. We say that the closed neighbourhood

condition holds if there is a basis u of u-neighbourhoods

of 0 which are v-closed.

THEOREM 4. Let (F,v) be a locally convex space. Then

F is barrelled iff the only convex topologies with bases

of v-closed neighbourhoods of 0 are those coarser than v.

(Cf. B [ I ~ )

I. A topological vector space in which the filter condi-

tion holds but not the closed neighbourhood condition.

! *

Let (E,u) be a Banach space with E # E , (Any

infinite dimensional Banach space satisfies this condition).

 69

Let v = T(E,E ). Then (E,v) is complete and so the filter

condition holds; but the closed neighbourhood condition

does not hold by Theorem 4, because v is strictly finer

than u.

2. An N-S space which is not an L-W space.

The metric space Q of all rational numbers is clearly

an N-S space. But it is not an L-W space in view of Theo-

rem i.

3. A locally convex space C(X) of continuous functions

which is barrelled and bornological but not W-barrelled.

Let X = Q be the metric space of all rational numbers.

Since it is a Lindelof space, it is realcompact (Q-space)

so that C(X) is bornological. Further, since X is an N-S

space, C(X) is a barrelled space. But C(X) is not W-

barrelled in view of Theorem 2.

4. An ultrabarrel which is not convex and which does not

have a defining sequence of convex sets.

Let E be a complete locally bounded space which is

not locally convex (say, E = ~ LO,~, O<p<l). Let B be a

closed, circled and bounded neighbourhood of O. Then B is

an ultrabarrel with {%n B} as a defining sequence for some

sequence {%n } of positive real numbers. But B and %n B

are not convex.

5. An ultrabarrelled space which is not barrelled.

The t o p o l o g i c a l vector space E = [PEo,~, O<p<l,

(See C h a p t e r 1, ~ 3(i)) i s c o m p l e t e and m e t r i z a b l e and
 70

hence an ultrabarrelled space. Clearly it is not barrelled.

6. A barrelled space which is not ultrabarrelled.

Consider the sequence space

1 oo

2 = {X = (Xn) ; E I Xnl ~ <~o}.

n = 1

Let

Let u: be the topology with a basis of neighbourhoods of

0 formed by the sets {x ; ]Ixl] ~ r Then (s is a

complete and metrizable topological vector space and so

1
ultrabarrelled. Hence (6:, u ~176 is barrelled. Now, let

Ilxlll =ZlXnl.
Then

llxll, Jlxlll ,
I
~ is a subspace o f ~z and t h e topology i n d u c e d on ~ by

t h e norm t o p o l o g y o f ~1 i s coarser t h a n u. The s p a c e ~

of all finite sequences is dense i n gz u n d e r t h e norm

! !
topology, and s o , since O c g2 g2 i s dense i n gl under the
1
norm t o p o l o g y . Hence t h e dual of ~ under t h e norm t o p o l o g y
1
is ~ and t h e norm t o p o l o g y is thus ~(~, ~). The d u a l
1 1
of (g~, u) is also g~ and so u ~176 T(~ ~, g~). By proposi-
1
tion 1, it follows that (~2, u00) is not ultra-barrelled.

7. An u ~176 compact set which is not u-compact.

!
We construct a subset of s which is u-precompact

but whose circled and convex envelope is not u-precompact,

 71

and deduce that a u~176 set is not u-compact. Let

xll = (1,0,0, ... )

xal = (0,~,0, ... )

x22 = (0,0,~,0, ...)

x31 = (0,0,0,I/3,0, ...)

x32 = (0,0,0,0,1/3,0, ...)

x~3 = (0,0,0,0,0,1/3,0, ...)

and in general let Xnm (l<m<n) be the element with all

terms zero except the (~ n(n-l)+m) th term which is I/n.

Let

B = {Xnm ; n,m = 1,2, ...}

Then
1
II Xnmll ~ =

and Xnm O, regarding (Xnm) as the sequence x11, x21, X22,

x~l, .... Hence B is u-precompact. Define the sequence

(yn) by
1 n
Yn - n E Xnm
m=l

so that Yn is the element with the first 1 n(n-l) terms

zero, then n terms each equal to ~2 and the remaining terms

zero. Then Yn belongs to the circled and convex envelope of
B; if n~k,
IIYn Ykll~ = 4

and so (yn) is not u-compact. Hence the circled and convex

_ 1
envelope of B is not u-precompact. Also llynl I n

and

A = {0} O {Yn ; n = 1,2, ...}

is u~176 but certainly A is not u-compact.

8. An ultrabarrelled space which is not non-meagre.

 72

The space ~ of finite (real) sequences, under the

finest compatible topology u is ultrabarrelled. Since

is a countable union of finite-dimensional spaces, it is

meagre.

9. An ultrabornological space which is not bornological.

The s p a c e [P [0,13, O<p<l, as d e f i n e d in C h a p t e r 1,

--//~ 3(i), is a complete and metrizable topological vector

space and hence it is ultrabornological. Clearly it is

not bornological.

I0. A bornological space which is not ultrabornological.

Let E be a vector space of uncountable dimension.

Then the finest linear topology s is strictly finer than

T(E,E*) (P[48], Theorem 3.1), so that the identity map of

(E, T(E,E*)) into (E,s) is not continuous; but it is boun-

ded, since every z(E,E*)-bounded subset is contained in a

finite-dimensional subspace of E. Hence (E, ~(E,E*)) is

not ultrabornological thought it is bornological.

Ii. An ultrabornological space which is not ultrabarrelled.

Let E be a countably dimensional non-locally convex

metrizable topological vector space. Since it is metrizable,

it is ultra-bornological. Since the finest linear topology

on a countably dimensional vector space is locally convex,

any ultrabarrelled topology on a countably dimensional

vector space is necessarily locally convex. Hence E is not

ultrabarrelled.

12. An ultrabarrelled space which is not ultrabornologi-

cal.
 73

Let E be an incomplete Hansdorff inductive limit of a

sequence of Banach spaces, E the completion of E and xeE-E.

Then the subspace E l of E spanned by E and x is ultra-

barrelled (P~6]). But El is not bornological (P~, page

155) and hence not ultrabornological.

13. A quasi-ultrabarrelled space which is neither ultra-

barrelled nor ultra-bornological.

Let E I be an ultrabarrelled space which is not ultra-

bornological and E 2 an ultrabornological space which is

not ultrabarrelled. Then the product of E I and E 2 is quasi-

ultrabarrelled which is neither ultrabarrelled nor ultra-

bornological.

14. A countably quasi-ultrabarrelled space which is not

countably ultrabarrelled.

The normed space E of Chapter 3, ~# 16 is not count-

ably barrelled (Chapter 3, ~= 44) and hence not countably

ultrabarrelled either. However, it is quasi-ultrabarrelled,

being a normed space, and hence countably quasi-ultra-

barrelled.

15. A countably ultrabarrelled space which is not ultra-

barrelled.

The s t r o n g d u a l o f a m e t r i z a b l e locally convex space

is c o u n t a b l y u l t r a b a r r e l l e d (P[3~, Proposition 3.1) and
so t h e space c o n s i d e r e d on page 435 i n B[20] is countably
ultrabarrelled. But, as shown there, it is not barrelled

and hence not ultrabarrelled either.

16. A countably barrelled space which i s n o t c o u n t a b l y

 74

ultrabarrelled.

Let (E,u) be the complete metrizable topological

vector space as defined in ~ 6. If ~(E,E*) is the finest

locally convex topology on E, then (E, ~(E,E*)) is

barrelled and hence countably barrelled. We show that it

is not countably quasi-ultrabarrelled: For each integer

n and element x = (xl, x2, ...) in E, let

tn(X) = (xl, x2 . . . . , Xn, O,O,...).

Then (tn) is a sequence of continuous linear maps from

(E,u ~176 into (E,u) such that, for each x in E, tn(X )

converges to x in (E,u). Clearly, each t n is a continuous

map from (E, ~(E,E*)) into (E,u). Moreover, {tn} is

uniformly bounded on the ~(E,E*)-bounded subsets of E. For,

any T(E,E*)-bounded subset A of E is contained in some

finite-dimensional subspace E 0 of E, and the restrictions

of {tn} to E 0 must be equicontinuous, implying that

Utn(A) is u-bounded. However, since the identity map from

(E, T(E,E*)) onto (E,u) is not continuous, (t n} is not an

equicontinuous set of maps from (E, ~(E,E*)) into (E,u).

Hence, by (P[37~, Theorem 3.2), (E, T(E,E*)) is not count-

ably quasi-ultrabarrelled and, a f o r t i o r i , not countably

ultra-barrelled.

17. A k-quasi-ultrabarrelled space which is not k-ultra-

barrelled.

The space E of complex sequences with only a finite

number of non-zero elements equipped with the topology of

pointwise convergence is k-quasi-ultrabarrelled, being

metrizable. But it is not a k-ultrabarrelled space,

 75

because it is locally convex but not k-barrelled (Chapter

3, ~ SS).

18. A hyperbarrelled space which is not hyperbornological.

The space E1as defined in # 12 is locally convex ultra-

barrelled and hence hyperbarrelled. Since a convex hyper-

bornological space is bornological and since Ells not

bornological, it follows that E1is not hyperbornological.

19. A hyperbornological space which is not hyper-

barrelled.

Let E be a countably infinite dimensional normed

space. Then E is not barrelled and so not hyperbarrelled

either. However, since E is metrizable, it is hyperborno-

logical.

20. A quasi-hyperbarrelled space which is neither hyper-

barrelled nor hyperbornological.

Let E be a hyperbarrelled space which is not hyper-

bornological and F a hyperbornological space which is not

hyperbarrelled. Then E• is a quasi-hyperbarrelled space

which is neither hyperbarrelled nor hyperbornological.

21. An ~-quasi-hyperbarrelled space which is not jy-

hyperbarrelled.

The space E = ~ of Chapter 3, ~ 16, is, for each

~ > ~0, an ~-quasi-hyperbarrelled space which is not

~-hyperbarrelled.

22. A barrelled space which is not ~0-hyperbarrelled.

 76

Let (E,u) and (E,u ~176 be as in ~= 6. Then (tn} , as

defined in ~ 16, is a sequence of continuous linear maps

from (E,u ~176 into (E,u) such that, for each x in E,tn(X )

converges to x in (E,u). Since the identity map from

(E,u ~176 into (E,u) is not continuous, it follows from

Theorem 3 that the barrelled space (E,u ~176 is not JYo-

hyperbarrelled.
 CHAPTER 5

ORDERED TOPOLOGICAL VECTOR SPACES

Introduction

All vector spaces in this Chapter are over the field

of real numbers.

An ordered vector space (E,C) which is also a topo-

logical vector space is called an ordered topological

vector space; we denote it by (E,C,u) where u is the

topology. The positive cone C in an ordered topological

vector space (E,C,u) is called normal for the topology u

if there is a neighbourhood basis B at 0 in (E,C,u)

consisting of full sets. The members of ~ can be chosen

to be circled. If u is a locally convex topology, we can

assume that the members of ~ are convex.

THEOREM i. Let (E,C,u) be an ordered topological

vector space. The following statements are equivalent;

(a) C is normal.

(b) There is a neighbourhood basis B = {V} at O such

that O~x~y, y~V implies x~V.

(c) For any two nets {x } and {y~} in (E,C,u), if

O~x ~y~ for all ~ and if {y } converges to O for u,

then {x } converges to O for u. (Cf. BE2 q )

PROPOSITION i. Let (E,C,u) be an ordered topologi-

cal space with normal cone C. Then each order-bounded

subset of E is u-bounded. (Cf. B [ 2 ~ )

THEOREM 2. Let (E,C,u) be an ordered locally convex

 78

space. Then the following statements are equivalent:

(a) C is normal.

(bl) There is a family {p } of seminorms giving the

topology u such that O<x~y implies p~(x) ! p~(y) for all

~(equivalently, p~(t+s)~ p (t) for all t, s~C and all

(cf. B [263

COROLLARY i. If C is a normal cone in an ordered

locally convex space (E,C,u), so is C .

THEOREM 3. Let (E, If'If) be a space ordered by

the positive cone C. The following statements are equiva-

lent:

(a) C is normal for the norm topology.

(b) There i s an equivalent norm ll'll on E such

that llxll~ ! llYlll whenever O<x<y.

(c) There is a constant %>0 such that %Ilxll~llyll

whenever O<_x<_y.

(d) There is a constant %>0 such that

II x + ylI~% max {IIxll , IIyII} for all x,yeC.

(e) The set {IlxII; o<x<y, IIyll~l} is bounded

above. (Cf. B[26])

Let ~ be the saturated class of all u-bounded

subsets of an ordered locally convex space (E,C,u) such

that E = US. The cone C in E is called a b-cone (strict

So%
b-cone) if the sets {SnC - SnC ; S e ~ } ( r e s p e c t i v e l y ,

{SnC - SoC ; S e ~ } ) form a fundamental system for

We say that the cone C in an ordered topological vector

space (E,C,u) gives an open decomposition in E if each
 79

VflC - V tiC is an u-neighbourhood of O whenever V is an

u-neighbourhood of O.

THEOREM 4. Let u be a metric vector topology in an

ordered vector space E such that C is generating and

complete. If C gives an open decomposition in (E,u), then

(E,u) is complete. (Cf. B [ 3 ~ )

An ordered topological vector space (E,C,u) is

boundedly order-complete if every u-bounded and directed

(j) subset has a supremum.

An ordered topological vector (ordered locally

convex) space (E,C,u) which is a vector lattice is called

a topological vector lattice (locally convex lattice) if

there is a neighbourhood basis of O for u consisting of

solid sets. Since the convex hull of a solid set in a

vector lattice is solid, there is a neighbourhood basis

of 0 for u consisting of convex and solid sets in a

locally convex lattice and the topology is referred to as

locally solid topology.

A normed space (Banach space) (E, II'II) which is a

vector lattice is called a normed vector lattice (Banach

lattice) if Ixl~lylimplies llx[ljllyll

PROPOSITION 2. Let (E,C,u) be an ordered topologi-

cal vector lattice which is a vector lattice. The follow-

ing statements are equivalent:

(i) (x,y) § ~ y is a continuous map of ExE into E.

(ii) (x,y) § xvy is a continuous map of ExE into E.

+
(iii) x § x is a.continuous map on E.
 80

(iv) x + x-is a continuous map on E.

(v) x § Ixl is a continuous map on E. (Cf. B [ 2 ~ )

If the cone in E is normal, the continuity of each

map in (iii), (iv) and (v) is equivalent to its continuity

at O. The maps defined in (i) through (v) are referred to

as lattice operations.

THEOREM 5. A topological vector space (E,u) which

is a vector lattice is a topological vector lattice iff

the cone in E is normal and the lattice operations are

continuous. (Cf. B [ 2 ~ )

THEOREM 6. Let (E,u) be a locally convex space which

is a vector lattice. The following statements are equiva-

lent.

(i) (E,u) is a locally convex lattice.

(ii) For any nets {x } and {y } in E,

if Ixal~fya) for all a a n d xa§ for u,

then y § f o r u.

(iii) There is a family {Pc a~I} o f s e m i n o r m s on E

generating u such that Ixl~lyl implies

Pa(x)~Pa (y) for all a. (Cf. B [ 2 ~ )

A vector sublattice M in a topological vector lattice

(E,C,u) is a topological vector lattice in the relative

topology. If M is a lattice ideal in E, then the quotient

E/M is a topological vector lattice for the quotient

topology. A product of topological vector lattices is a

topological vector lattice.

An o r d e r e d normed space (E,C,II-II) is called an

 81

order-unit normed space if there is an o r d e r - u n i t e such

the norm II'll is the gauge of [ - e , ~ . (E,C, II'll) is

called an approximate order-unit normed space if there

is an approximate order-unit {ex , ~eA ,~} in C such

that the norm II'[I is the gauge of the circled and

convex set

sA = U{[e , ];

A n order-unit normed space is clearly an approxi-

mate order-unit normed space.

THEOREM 7. Let (E,C, II']]) be an ordered normed space.

The following statements are equivalent.

(i) E is an a p p r o x i m a t e order-unit normed s p a c e .

(ii) The o p e n u n i t ball B = {xEE ; llxlI<l} is solid

and d i r e c t e d upwards. (Cf. B[3~)

Let (E,C) be an o r d e r e d vector space, where C is

generating and l e t B be a b a s e o f C. The c i r c l e d and

convex envelope F(B) of B is absorbing i n E. The g a u g e

of P(B) is called a base semi-norm on E. If the seminorm

is infact a norm then it is referred to as a base norm.

An ordered normed space (E,C, ll'II) is called a base

normed space if C is generating and if there is a base B

of C such that the norm II'II is the gauge of P(B).

PROPOSITION 3. If an ordered normed space (E,C,II-II)

is an approximate order-unit normed space, then

(E',C', I I'll) is a base normed space. (Cf. B[39])

The order-bound topology u0 on an ordered vector

space is the finest locally convex topology u for which

 82

every order-bounded set is u-bounded.

PROPOSITION 4. A neighbourhood basis at O for the

order-bound topology u0 on an ordered vector space is

given by the family of all circled convex and order-borni-

vorous (a set which absorbs all order-bounded sets) sets.

(Cf. B [ 2 ~ )

A locally convex space E which is a vector lattice is

called a pseudo-M-space if it has a neighbourhood basis at

0 consisting of sublattices. A locally convex lattice which

is a pseudo-M-space is called an M-space.

A locally convex lattice (E,C,u) is called order-

quasibarrelled if each order-bornivorous barrel in E is a

neighbourhood of O.

Let (E,C,u) be a locally convex lattice. For any

x~C, we define

px(f) = sup,{[<y,f > ]; y e [ - x , x ~ }

for all feE'. Then t h e f a m i l y {Px ; xr defines a locally

solid topology on E' denoted by ~ (E',E).

THEOREM 8. Let (E,C,u) be a locally convex lattice.

The following statements are equivalent.

(a) (E,C,u) is order-quasibarrelled.

(b) Each ~ (E',E)-bounded subset of E' is equiconti-

nuous. (Cf. B [ I ~ or B [ 3 ~ )

A locally convex lattice (E,C,u) is called a countably

order-quasi-barrelled vector lattice (abbreviated to C.O.Q.

vector lattice) if each order-bornivorous barrel which is

 83

the countable intersection of closed, circled and c o n v e x

neighbourhoods of 0 is itself a neighbourhood o f 0 i n E.

Every order-quasi-barrelled vector lattice is a C.O.Q.

vector lattice. A countably barrelled locally convex

lattice is a C.O.Q. vector lattice and a C . 0 . Q . vector

lattice is a countably quasibarrelled locally convex

lattice.

THEOREM 9. Let (E,C,u) be a l o c a l l y convex lattice.

The f o l l o w i n g statements are equivalent.

(a) (E,C,u) is a C.O.Q. vector lattice.

(b) Each solid barrel which is the countable inter-

section of closed, convex and solid neighbourhoods of 0 is

itself a neighbourhood of 0 in E.

(c) Each ~s(E',E)-bounded subset of E' which is the

countable union of equicontinuous subsets of E' is itself

equicontinuous. (Cf. B~I~)

A.C.O.Q. vector lattice which has a fundamental

sequence of order-bounded subsets is called an o r d e r - ( D F )

vector lattice.

A C.O.Q. vector lattice with an o r d e r - u n i t is an

order-(DF) vector lattice. In particular an o r d e r - u n i t -

normed v e c t o r lattice is an o r d e r - ( D F ) vector lattice.

Let (E,C,u) be a t o p o l o g i c a l vector lattice. If, in

the definition o f an u l t r a b a r r e l (Chapter 4, Introduction),

e a c h Bn i s circled and o r d e r - b o r n i v o r o u s (solid and

absorbing), then B is said t o be o r d e r - b o r n i ~ o r o u s ultra-

barrel (solid ultrabarrel, if B is solid). (E,C,u) is

 84

called an order-quasi-ultrabarrelled vector lattice

(abbreviated to O.Q.U. vector lattice) if each order-

bornivorous ultrabarrel in E is a neighbourhood of O.

THEOREM iO. A topological vector lattice (E,C,u) is

an O.Q.U. vector lattice iff each solid ultrabarrel in E

is a neighbourhood of 0. (Cf. B [ I ~ )

An ultrabarrelled topological vector lattice is an

O.Q.U. vector lattice and an O.Q.U. vector lattice is a

quasi-ultrabarrelled topological vector lattice. A locally

convex O.Q.U. vector lattice is order-quasibarrelled. An

order-unit-normed vector lattice is an O.Q.U. vector

lattice.

Let (E,C,u) be a topological vector lattice. If, in

the definition of an ultrabarrel of type ( ~ ) (Chap-

ter 4, Introduction), V (n) is order-bornivorous then V ~

is called an order-bornivorous ultrabarrel of type (~) in

(E,C,u). On the other hand, if each V~ n) is closed and

solid, then V 0 is called a solid ultrabarrel of type (~)

in (E,C,u). (E,C,u) is called a countably O.Q.U. vector

lattice if each order-bornivorous ultrabarrel of type (e)

in (E,C,u) is a neighbourhood of O.

THEOREM II. A topological vector lattice (E,C,u) is

a countably O.Q.U. vector lattice iff each solid ultra-

barrel of type (~) is a neighbourhood of 0 in (E,C,u).

(Cf. P[4~ )

THEOREM 12. Let (E,C,u) be a c o u n t a b l y O.Q.U. v e c t o r

lattice and ( F , K , v ) any t o p o l o g i c a l vector lattice. If

{fn } is a pointwise bounded sequence o f c o n t i n u o u s l a t t i c e
 85

homomorphisms of E into F, then {fn} is equicontinuous.

(Of. P ~ 5 ] )

An O.Q.U. vector lattice is a countably O.Q.U. vector

lattice. A countably ultrabarrelled topological vector

lattice is a countably O.Q.U. vector lattice which is a

countably quasi-ultrabarrelled topological vector lattice.

A locally convex lattice which is also a countably O.Q.U.

vector lattice is a C.O.Q. vector lattice.

I. An ordered topological vector space with generating

cone which does not give open decomposition.

Consider the Banach space E = C[0,1] of all continuous

(real-valued) functions on the closed interval [0,1],

equipped with the usual supremum norm. Let C be the cone

in E consisting of all non-negative and convex functions

and !et F = C - C By the Stone-Weierstrass theorem , F

is dense in the Banach space E. Also, since any function

in C must be differentiable on ~,l] except at, atmost, a

countable number of points, any function f in E which is

not differentiable at an uncountable subset of ~,l] is not

in F. Therefore, F is a proper dense subspace of E, and F

is a non-complete normed space in the relative topology.

Since C is a generating and norm-complete cone in F, it

follows from Theorem 4 that C does not give an open decom-

position.

2. An ordered topological vector space with normal cone

but with a {topologically) bounded set which is not order-

bounded.

Let (S,Z,~) be an arbitrary measure space. Consider

 86

the space E = ~P (S,E,~), pzl, equipped with the norm

[]fl] = {f[f(t)] p d~(t)} 1/p

The cone of functions in E that are n o n - n e g a t i v e on S

~-almost everywhere is normal for the norm topology.

However, the unit ball

B = {fee ; ]]f]]51}

is n o r m - b o u n d e d but not order-bounded.

3. A cone in a topological vector space, w h i c h is not

normal.

Let E = s (or co or ~ ), equipped with the supremum

norm and Ps the cone consisting of sequences in E having

all partial sums non-negative. We show that Ps is not

normal in E. Let

x n = e~ + ... + e n - n e n + ,

and

Yn = nen+1

where e n is the element in E having 1 in the n th place and

zeros elsewhere. Then

Xn ' Yn ~Ps ' IlXn]l = [lYn]] = n

and

[IXn § YnI[ = 1

Hence Ps is not normal in E, in view of T h e o r e m 3.

4. An ordered topological vector space in w h i c h order-

bounded sets are b o u n d e d but the cone is not normal.

Let E = ~ be the space of finite sequences equipped

 87

with supremum norm and ordered by the cone Ps as defined in

=/~ 3. As shown there, Ps is not normal. Now,

let a = (al, . ' ak) E Ps' and

Ar al + + ar , (l<r<s),

A = max A
r

If x = (Xn) and O!x!a , then

0< ~r!A

for all reN, where

~l" XI+ + Xr

Hence,

IXr I = I~r - ~r-,l<_ A

for all r, that is, Ilxll! A.

5. A cone in a topological vector space, which has no

interior points.

(i) The positive cone C defined by

C = {x = (Xn)Ss ; x n ! 0 for all n}

in s which is equipped with the usual norm, has no

interior points. For, if it had one, it would be an order-

unit in s s does not have order-units.

(ii) Similarly the positive cone in c o does not have

interior points.

6. An element of a cone in a vector space which is an

interior point for one topology but not for another

topology.
 88

el is an interior point of the cone Ps(See ~ 3) in

s with respect to the usual norm. However, el is not

interior point of P with respect to the supremum norm p

S

on s : for, if

1 (e + + )
xn = ~ 1 en+ 1 ,

then
_ 1
P(Xn) n
and

el - x n ~ Ps

7. A cone in a locally convex space, which is not a

b-cone.

Let E = C0[O, ~ be the space of continuous (real-

valued) functions on the closed interval [O,~ that vanish

at O, equipped with the norm

llfll sup Tf(x) 1 ;

and ordered by the cone K of non-negative and convex

functions in E. Then K is not a b-cone : To show this, let

f in E be defined by

f(x) xe[o,
and let

O<c<~

Then

I If - (g-h) I I<e , g,neE

implies that

g(t)>f(t)-e, te[O, i~

In particular,
 89

g(4g2)>E ;
hence, since g is c o n v e x a n d g(0) = O, it f o l l o w s that

g(1) = g(1) - g(O) > g(4E 2) - g(O) > i_~

I - 0 -- 4g 2 - 0 4~

and so

IIgII~ i
4~

Hence K is n o t a b-cone in E.

8. A base of a c o n e in a t o p o l o g i c a l vector space,

which is not closed.

Consider E = 4, the space of finite (real) sequences

equipped with supremum norm and ordered by the cone

K = {x = (Xn)eE ; x n sO for all n}

Then the set

B = {X = ( X n ) e K ; Zx n = 1}

is a b a s e for B. It is n o t closed, since

(n) = el + + en
x cB
n

and

x (n) ~ 0 .

9. An ordered normed space which is n o t an o r d e r - u n i t

normed space though its d u a l is a b a s e normed space.

The s p a c e co of sequences converging to O, equipped

with supremum norm, is a normed space ordered by the

positive cone of non-negative sequences. It is not an

order-unit-normed space since it does not have an order-

 90

unit.

Since the open unit ball B = {xec 0 ; llxlI<l} is

solid and d i r e c t e d upwards, co is an approximate order-

unit-normed space by T h e o r e m 7 and hence its dual s is a

base n o r m e d space by p r o p o s i t i o n 5.

iO. An o r d e r e d topological vector space which is c o m p l e t e

but not order-complete.

(i) The Banach l a t t i c e C[O,~ of continuous functions

on t h e c l o s e d interval [O,~, e q u i p p e d w i t h t h e supremum

norm and o r d e r e d by t h e p o s i t i v e cone, is clearly complete.

But i t is not order-complete.

(ii) The Banach l a t t i c e c of convergent sequences is

complete but not order-complete.

Ii. An ordered topological vector space w h i c h is order-

complete but not complete.

The normed vector lattice r of finite sequences,

equipped with the supremum n o r m and ordered by the positive

cone, is o r d e r - c o m p l e t e but not complete.

12. An o r d e r e d topological vector space which is c o m p l e t e

and o r d e r - c o m p l e t e but not boundedly order-complete.

The Banach l a t t i c e Co o f s e q u e n c e s c o n v e r g i n g t o O,

e q u i p p e d w i t h t h e supremum norm and o r d e r e d by t h e p o s i t i v e

cone, is order-complete. Clearly it is complete, Now l e t

e n = (I, i . . . . , 1,0,O, .)ec 0

where 1 appears in n places, and let

B = {e n ; n ~ }
 91

Then B is increasing and norm-bounded. But, clearly, B

does not have a supremum in c o . Hence c o is not boundedly

order-complete.

13. An o r d e r - c o n t i n u o u s linear functional on an ordered

topological vector space, w h i c h is not continuous.

Consider the normed vector lattice r as defined in

5~ ii. The topological dual of ~ can be identified with ~i.

However every sequence u = (Un) of real numbers determines

an o r d e r - c o n t i n u o u s linear functional fu on r defined by

fu(X) = Z XnU n ; x = (Xn)e r .

n=1

(See B [ 2 ~ , page 56). C o n s e q u e n t l y there exist order-

continuous linear functionals on r that are not norm-

continuous.

14. A continuous linear operator on an ordered topological

vector space, w h i c h is not sequentially order-continuous.

Consider the space

s = {X = (Xn) ; Z [Xn 12 <~, X n e~}

n=l

equipped w i t h the norm

[IxI[ = ( Z [Xnl2) ~ , x = (Xn)es

n=l

and ordered by the positive cone

K = {x = (Xn)Ss ," x > 0 for all n}

n,--

Consider the infinite matrix T = (amn) defined by

1
ifm~n
m-n
amn "~ {
0 ifm=n .
 92

Then T is a matrix transformation on s into itself

(B[I~, page 236). T is continuous for the usual norm on

~2. We show that T is not order-bounded which implies that

it is not sequentially order-complete, because the vector

space of order-bounded linear operators on ~2 is equal to

that of sequentially order-continuous linear operators on

~2 ( B [ 2 ~ , page 52,53). If x = (Xm) and y = (ym) are the

elements of ~2 defined by

1
Xm = Ym - (re>l),
logm

and

Xl = Yl = Y2 ,

then
co co

(.) z z [amnlXm Yn = + ~ "

m=i n-- 1

Define z = (Zm)e~2 by
oo

Zm = E lamn I Yn
n=l

and set

(sign
Un ( m ) = amn) Yn

Then

u (m) _- (un(m)) e [-Y,Y]

for all m, and

oo
= Z a u (m) = (T u(m)) m
Zm mn n
n = I

Suppose T is order-bounded. Then there is a u>O such that

Tu(m) c [-U,U]
 93

for all positive integers m. It follows that

- < u,v > < < Tu (m) v > < <u,v>

for all v>O in ~2; in particular, if we take v to be the

unit vector e (m) , we obtain the r e l a t i o n

zm = ]Zml = [(Tu(m))ml~ um

for each m. But then

Z ~ l a m n [ Yn Xm = ~ Zm Xm
m=l n=l m=l

< ~ u a < co
- m m
m=l

since x,ue~ 2 This contradicts (,).

Hence T is not order-bounded.

15. A positive linear functional on an ordered topological

vector space, which is not continuous.

Consider the n o r m e d vector lattice ~ as defined in

=i~ ii. If x = (Xn)e ~ , the linear functional f defined by

f(x) = Zx n is p o s i t i v e but not continuous.

16. An ordered topological vector space on which there

exist no n o n - z e r o positive linear functionals.

Consider the complete metrizable topological vector

lattice L~ ~a,b], a,be~, O<p<l, as defined in chapter I,

3(i). As we have seen in Chapter I, ~Y= 6, there exist

no n o n - z e r o continuous linear functionals on ~P [a,b]. But

every p o s i t i v e linear functional on a complete metrizable

topological vector lattice is continuous ( B [ 2 ~ , page 88).

 94

Hence there exist no non-zero positive linear functionals

on [P[a,b], O<p<l.

17. A topological vector lattice w h i c h has no n o n - z e r o

real l ~ t i c e homomorphisms.

Consider the Banach lattice E = ~[O,~ with the

usual norm and order. Every positive linear functional on

E is continuous (B[2~, page 88). Let u be a real lattice

homomorphism and u(f) = 1 for some f with llfll = I. We

can write f in the form g1+ g~ where

IIg ll = Ilgill = I
and
!
inf {g~ , gi} = 0 .

We have u(fl) = 1 for fl = gl or g~ Repeating the con-

struction, we obtain a sequence of elements {fn} such that

llfnI[ = 2 -n , U(fn) = 1

contradicting the c o n t i n u i t y of f.

18. A topological vector space with lattice ordering in

+
w h i c h the map x § x is continuous for all x but not

uniformly continuous.

Consider c o with the usual n o r m and the order given by

the cone Ps as defined in ~ 3. As shown there, Ps is not

§
normal. Hence the map x§ is not u n i f o r m l y continuous.

Now we show that it is continuous at any point x = x n in c0.

Take c>O. There exists N such that IXr[~for r>N. Take

y=(yn) in co such that

E
fly -
 95

Let

Xr X1 + + X r ,

Yr = Y, + + Yr' (r>l)

X o -- O, Yo = 0 .

Then
§ +
x = (~r) , y = (~r)
where

= X + _ X +
C~r r r-I '

8 r = Y +r - Y +r-1 ' ( r --
>l) "

For r~N, we have

r
]Y+r - Xr+[ <-I Yr - X r l-< Z I Ys - Xsl_< c
S=I

and

]8 r - arl_<2e

For r > N,

I (~ rl-I< X r l -< ~
and

ISrl~_lYrl~_~ § I l y - ~11 ~_ 2~

and so

]~r - ~r ]E 3~ .

Hence,

§ +
[ly - x II~_ 3 ~ .

19. An o r d e r e d locally convex space with a positive weakly

convergent sequence which is not convergent.

 96

Consider the normed space ~ of bounded sequences,

equipped with the supremum n o r m , and ordered by the cone

Ps a s d e f i n e d i n =A~ 3. Let en denote the element in s

having 1 in the n th coordinate and zeros elsewhere. Then

the sequence {en} is a decreasing sequence which converges

to 0 in t h e weak t o p o l o g y but not in the norm t o p o l o g y .

20. An M-space which is not n o r m a b l e .

The space of all real-valued functionals on an infi-

nite set X, equipped with the topology of p o i n t w i s e

convergence is an M - s p a c e which is not n o r m a b l e .

21. A Pseudo-M-space which is n o t an M - s p a c e .

Consider the n o r m e d space E = ~ or co, equipped with

the supremum norm and o r d e r e d by the cone Ps as d e f i n e d in

3. G i v e n x = (x n) and y = (yn), let

Xn = x l + " + Xn '

=
Yn Yl + " + Yn '

z n = (X n v Yn ) - (Xn_ I ~ Y n - 1 ) , n~2

and

zl = X ! v YI

It is e a s i l y seen that

iz i!IXnl v lYnl

for all n so t h a t

IIzll llxll ilYll, z = (Zn).

Also

Zl + + z n = X n v Yn
 97

so that z = xvy. Thus, E is a pseudo-M-space. Since Ps is

not normal as shown in ~ 3, E is not an M-space.

22. A topological vector lattice which is not a pseudo-M-

space.

The Banach lattice ~I with the usual norm and order is

not a pseudo-M-space, because any sublattice containing

the unit ball is unbounded.

23. The topology of a bornological locally convex lattice

which is not an order bound topology.

Consider the Banach lattice E = C[O,IJ as defined in

lO(i). Let F be the subspace of E consisting of all

elements f in E which vanish in a neighbourhood (depending

on f) of t = O, K the positive cone in F. Then F is a

normed vector lattice under the induced topology and so a

bornological space. Let

V = (f~F ; If( ) ! ~ ' n = 1,2, .t

Clearly V is closed, convex and solid.

If f~F, let af, O<af <1, be such that f(t) = 0 for all

t lO, a f l . zf
_1
= af l l f l l ,

then ~x~V a n d so V is absorbing. We now s h o w t h a t V is not

a neighbourhood of O. For this, it is sufficient to show

that for any natural number r~l, there is an frEF with

IIfrIl~ such that x r ~ V ; it follows, then, that O is

not an interior point of V and so V is not a neighbourhood

of O. Consider A = [O,-~r ~ and B = [ ~ + i ' 1 ] in [O,i]

 98

By Urysohn's lemma, there is a continuous real-valued

1
function fr on [0,i] with range in [0, ~ ] s u c h that

fr(t ) = { O if teA

s if teB .
r

Clearly freF and

1
l ifrli_< ? 9

On the other hand, since

i i > I
fr ( ~ ) = r r+l

it follows that fr ~ V. Thus, V is not a neighbourhood of

O. Since a solid set is absorbing iff it is order-borni-

vorous, we conclude that the topology on F is not an order

bound topology.

24. An order-quasibarrelled vector lattice which is not

barrelled.

of. ~ 26.

25. A C.O.Q. vector lattice which is not order-quasi-

barrelled.

The locally convex space C(W) as defined in Chapter 3,

~ 41(ii), ordered by the positive cone

K = {feC(W) ; f(t) > O for all teW}

is a locally convex lattice. We have seen in Chapter 3~

~ 41(ii), that C(W) is countably barrelled; hence it is

a C.O.Q. vector lattice. But it is not order-quasibarrelled

because it is not quasibarrelled as shown in Chapter 3,

~ 41(ii).
 99

26. A C.O.Q. vector lattice which is not countably

barrelled.

Consider the Banach lattice E = ~ of bounded sequen-

ces, equipped with the supremum norm and the usual point-

wise ordering. Let e n be the sequence in E having 1 in the

n th coordinate and zeros elsewhere, F 0 the subspace of E

generated by the e n (F 0 consists of all finite sequences),

e the sequence having 1 in every coordinate and

F = {x + Xe ; xcF o , ~}

Then with the induced norm and order, F is an order-unit-

normed vector lattice with order-unit e. It is easy to see

that an order-unit-norm topology is an order bound topology

and that each vector lattice with an order bound topology

is order-quasibarrelled. Thus F is order-quasibarrelled.

Now, define fn on F by

f n ( X + Xe) = xn , x = (Xn)eF 0

Then {fn} is a pointwise bounded sequence of continuous

linear functionals but {fn} is not equicontinuous. Hence

F is not countably barrelled.

27. A countably quasibarrelled locally convex lattice

which is not a C.O.Q. vector lattice.

The normed vector lattice F as defined in =~ 23 is

quasibarrelled and hence countably quasibarrelled. But it

is not a C.O.Q. vector lattice. For, if it were, then,

being metrizable, it would be order-quasibarrelled, because

a quasibarrelled C.O.Q. vector lattice is order-quasi-

barrelled (B[I~ or P[28]). But it is not an order-quasi-

 100

barrelled vector lattice.

28. An order-quasibarrelled (hence a C.O.Q.) vector

lattice which is not an order-(DF)-vector lattice.

Consider the Banach lattice ~: with the usual norm

and pointwise ordering. Clearly ~i is an order-quasi-

barrelled vector lattice. We show that it does not have a

fundamental sequence of order-bounded sets. Suppose it has.

Then as (~,~i) is metrizable. Since (~, us (~,~i)) is

complete, it follows that (~, as(~ ~, ~i)) is complete and

metrizable and hence a s (Z~, ~i) coincides with the

supremum norm topology of ~ , because the two topologies

are comparable. Thus, as(~ ~, ~i) is normable. Also the

positive cone in ~i is generating and a s (~i, ~ ) - c l o s e d .

But then ~i has an order-unit (B[2~), which is not true.

29. An O.Q.U. vector lattice which is not ultrabarrelled.

The normed vector lattice F, as defined in =~ 26, is

not barrelled and hence not ultrabarrelled, because a

locally convex space which is ultrabarrelled is barrelled.

But it is an O.Q.U. vector lattice because it is an order-

unit-normed space.

30. A quasi-ultrabarrelled topological vector lattice

which is not an O.Q.U. vector lattice.

The normed vector lattice F, as defined in ~ 23, is

not an O.Q.U. vector lattice, because it is not order-

quasibarrelled as can be seen in @# 23. However, since a

normed space is quasi-ultrabarrelled, so is F.

31. An order-quasibarrelled vector lattice which is not an

 101

O.Q.U. vector lattice.

cf. ~ 34.

32. A countably O.Q.U. vector lattice which is not

countably ultrabarrelled.

The order-unit-normed vector lattice F, as in ~= 26,

is an O.Q.U. vector lattice, as shown in ~= 29, and hence

countably O.Q.U. vector lattice. But F is not countably

barrelled, as shown in =~= 26, and hence not countably

ultrabarrelled either. Note that a locally convex space

which is countably ultrabarrelled is countably barrelled.

33. A countably quasi-ultrabarrelled topological vector

lattice which is not a countably O.Q.U. vector lattice.

The normed vector lattice F, as defined in ~ 23, is

quasi-ultrabarrelled, being a normed space, and hence

countably quasi-ultrabarrelled. Since it is not a C.O.Q.

vector lattice, as shown in ~ 27, it is not a countably

O.Q.U. vector lattice either. Note that a locally convex

space which is countably O.Q.U. space is a C.O.Q. space.

34. A C.O.Q. vector lattice which is not a countably

O.Q.U. vector lattice.

I
The space (~, u), as defined in Chapter 4, #~ 6, is

infact a complete and metrizable topological vector lattice

for the order induced by the cone

K = {x = (xi)s~2 ; x i ~ O for all i}

I
As we have seen there, (Z~, u ~~ barrelled and hence
 102

order-quasibarrelled which, in turn, implies that it is a

!
C.O.Q. vector lattice. Now, for each x = (xi)e~2 , define

tn(X ) = (x~, , x n , 0,0, . ).

Then each t n is a continuous lattice h o m o m o r p h i s m of

! ! !
(s K, u ~ 1 7 6 i n t o (s K, u) and for each xes 2

fn(X) + x under u. But {t n} is not e q u i c o n t i n u o u s as

!
observed in Chapter 4, ~ 16, and hence (s K, u ~176 is not

a countably O.Q.U. vector lattice by T h e o r e m 12.

 CHAPTER 6

HEREDITARY PROPERTIES

Introduction

Let E be a vector space. A nonempty subset M of E is

called a (vector) subspace of E if M + M ~ M and A M c M . If E

is a topological vector space, then a subspace M of E is a

topological vector space under the induced topology.

Let M be a subspace of a vector space E. Consider

G = {x + M ; xeE such that x+M = y+M

iff x - y~M}

In G, define the addition operation as

(x+M) + (y+M) = x + y + M

and scalar m u l t i p l i c a t i o n by

X(x+M) = ~ x + M .

Then G becomes a vector space called quotient vector space

which is denoted by E/M. Now let E be a topological vector

space, M a subspace of E and ~ the map of E onto E/M

defined by r = x+M. ~ is called the canonical or quoti-

ent map. The quotient topology on E/M is defined to be the

finest topolog F on E/M for w h i c h r is continuous. E/M,

equipped with this topology, is called a quotient space.

Let {E } be a family of v e c t o r spaces and E = ~Ea the

cartesian product. E becomes a vector space w i t h pointwise

addition and scalar m u l t i p l i c a t i o n . If {E } is a family of

topological vector spaces, then E = ~E is also a topolo-

gical vector space with the basis of n e i g h b o u r h o o d s

U = ~Wa , when finitely m a n y W are circled n e i g h b o u r h o o d s

U of 0 in E a and all the other W a = E~.

 104

I. A closed subspace of a reflexivespace, which is not

reflexive.

Let w denote the space of all sequences x = (Xn) and

the space of finite sequences. We denote by ~w the

(topological) countable direct sum of copies of w, because

all the vectors of this direct sum are obtained when in

each vector (xl, . Xn, 0,0, .) of ~ the non-zero

x i are replaced by arbitrary n o n - z e r o vectors u i of w and

the zeros are replaced by the zero vector 0 of w. Similarly

we denote by w~ the (topological) countable product of

copies of ~. Let E = ~w + w~ . Then its dual is E'=w~+~w.

Since the locally convex direct sum of Montel spaces is a

Montel space, E' is a Montel space and hence reflexive.

Let

M = {(x,x) ; x~wnw~ = ~}.

Then N = M i consists of all (x,-x), x ~ , in E' = w~ + ~w.

N is a closed subspace of E', w h i c h is not reflexive in

the relative topology.

2. A closed subspace of a b o r n o l o g i c a l space, w h i c h is

not bornological.

Clearly E' = w~ + ~w is b o r n o l o g i c a l (because the

locally convex direct sum of bornological spaces is borno-

logical).

The closed subspace N, as defined in ~ I, is not

bornological; if it were, then, being complete, it w o u l d be

barrelled. But then, being semi-reflexive, it w o u l d be

reflexive w h i c h is not true.

3. An infinite countable codimensional subspace of a

 105

bornological space, which is not quasibarrelled.

Cf. Valdivia, M. : Some examples on quasibarrelled

spaces, Ann. Inst. Fourier, Grenoble, 22, 2(1972), 21-26.

4. A closed subspace of a barrelled space which is not

countably quasibarrelled.

Since any Hausdorff locally convex space is a closed

subspace of some barrelled space (P[4@), it is now suffi-

cient to give an example of a Hausdorff locally convex space

which is not countably quasibarrelled. Let u be the supre-

mum norm on the space co of sequences convergent to O, and

v the associated weak topology on co. For each n, let

gn : (E,v) + (E,u)

be defined by

gn(X) = (xz, x~, .,Xn, O,O, .).

Then {gn } is a sequence of continuous linear maps from

(E,v) into (E,u) such that, for each x in E, gn(X)

converges to x in (E,u). Moreover, {gn } is uniformly

bounded on bounded sets, for if B is the unit ball in (E,u),

the union over n of gn(B) is contained in B. But {gn } is

not equi-continuous since v is strictly coarser than u and

so (E,v) is not countably quasibarrelled.

5. A dense uncountable dimensional subspace of a barrel-

led space, which is not barrelled.

Let s be the space of real sequences with the product

topology. Let m be the dense subspace of s consisting Of

the bounded sequences. Then m is not barrelled in the

relative topology because the barrel {x = (Xn)em ;

 106

x <i, neN} is not a n e i g h b o u r h o o d of 0. Of c o u r s e m has

n --
uncountable dimension, because a countable codimensional

subspace of a b a r r e l l e d space is b a r e l l e d .

6. A closed subspace of a (DF)-space, which is not a (DF)-

space.

Let E be the v e c t o r space (over R) of all d o u b l e

sequences x = (xij) such that for each neN,

(n)
Pn(X) = ~ aij Ixijl <
i,j

where
(n) jn for i<n
a, , ~ i
z3 .n
l for i>n

Equipped with the topology generated by the semi-norms {pn }, E

is a F r ~ c h e t Montel space. The d u a l E' of E can be i d e n t i f i e d

with the space of all d o u b l e s e q u e n c e s u = (uij) such that

(n)
luijl< c aij for all i, j and s u i t a b l e c>0 , ne~ (the c a n o n i c a l

bilinear functional being (x, u)+< x, u> = 7 xij uij). E a c h

i,j
xeE defines a summable family {xij ; (i,j)e N x N} . L e t T be

the linear map from E which sends each x = (xij)eE to the

vector
co oo

TX = ( ~ xil , ~ Xi2 , ... ).

i=l i=l

It f o l l o w s that

Z { Z xij {< Z Z { X i j { < ~

j=l i=l --

and so T x e s x . T is a c o n t i n u o u s linear map of E o n t o a dense

subspace of s The adjoint T' m a p s each element ~eZ ~ to the

element T'~eE'. The subspace of all T'~ is w e a k l y closed in E'

T' is thus a weakly continuous one-one linear map from s into

E' w i t h a weakly closed image space. It f o l l o w s that T is a

topological homomorphism of E o n t o ZX(Schaefer D~ , page 160).

 107

Thus E/N, N = (xEE ; T(x) = 0} , is isomorphic with ~I. Let

be the c a n o n i c a l m a p of E onto E/N . Every closed bounded subset

B of E is compact, and so its image ~(B) is compact. But there

exist b o u n d e d sets, which are not c o m p a c t in E/N ~ ~i ; not

every b o u n d e d subset of E/N is c o n t a i n e d in the closure of the

image ~(B) of a b o u n d e d set B. So, the G - t o p o l o g y on the closed

subspace H=T'(Z~, w h e r e ~ is the family of all r e l a t i v e l y compact

subsets of E/N, is strictly finer than 8(E',E). H is not quasi-

barrelled for this topology. Since the strong dual of a F r e c h e t

space is a (DF)-space, (E',8(E',E)) is a (DF)-space. But the

closed subspace H is not a (DF)-space, because a separable (DF)-

space is quasibarrelled. (We note that H is separable).

7. An infinite countable codimensional subspace of a quasi-

barrelled (DF) space, which is not a (DF) space.

Cf. V a l d i v i a ' s paper m e n t i o n e d in ~& 3.

8. A closed subspace of a h y p e r b a r r e l l e d space, which is

not h y p e r b a r r e l l e d .

Since a Hausdorff locally semiconvex space is a closed

subspace of some H a u s d o r f f hyperbarrelled space, (Cf. P ~ 4 ~ ) , ~ 3

is an e x a m p l e of a closed subspace of a h y p e r b a r r e l l e d space w h i c h

is not ~ - q u a s i - h y p e r b a r r e l l e d , afortiori, not h y p e r b a r r e l l e d .

9. A closed subspace of an u l t r a b a r r e l l e d space, which is

not c o u n t a b l y quasi-ultrabarrelled.

The space (E, T(E,E*)) as d e f i n e d in Chapter 4, ~=- 16, is

not c o u n t a b l y quasi-ultrabarrelled, afortiori, not u l t r a -

barrelled. Since (E, T(E,E*)) is H a u s d o r f f and complete, it can

be e m b e d d e d as a closed subspace of a p r o d u c t G of B a n a c h spaces.

Since G is of the second category in itself, it is u l t r a b a r r e l l e d

while the closed subspace (E, T(E,E*)) is not.

10. A lattice ideal in an o r d e r - q u a s i - b a r r e l l e d vector

 108

lattice, which is not order-quasibarrelled.

The subspace F of E as defined in Chapter 5, =~ 23 is

infact a lattice ideal in E. But F is not order-quasi-

barrelled.

ii. A complete locally convex space whose quotient is not

sequentially complete.

The locally convex space E = Cw + w~ , as defined in

~ i, is complete. We show that the quotient space E/M,

where M is the closed subspace as defined in ~= I, is not

sequentially complete. Given a sequence x (n)= (n (n), z (n))

in E, the cosets ~(n) with respect to M form a Cauchy

sequence if the sequence is almost constant in each co-

ordinate. In particular, if we take the sequence

n n
~(n) = ( Z elk , -Z elk )
i,k=z i,k=1

then ~(n) is a Cauchy sequence. But this has no limit in

E/M;for,there is no x = (n , z)~E for which ~ _ ~(n)

converges to O.

12. A quotient of a Montel space, which is not semi-

reflexive.

The space E = ~w + w~ is a Montel space. The quotient

space E/M as in =~ II is not semi-reflexive since it is

not weakly sequentially complete. Note that a locally

convex space is semi-reflexive iff it is weakly quasi-

complete.

13. A quotient of a Frechet Montel space, which is not

reflexive.
 109

The locally convex space E, as d e f i n e d in ~= 6, is a

Frechet Montel space. But the q u o t i e n t space E/N, as d e f i n e d

there, is not r e f l e x i v e because E/N is t o p o l o g i c a l l y isomorphic

to ii and i* is not reflexive.

14. A product of B - c o m p l e t e spaces which is not B-complete.

Let E be a p r o d u c t of B a n a c h spaces w h i c h are, of

course, B-complete. Suppose E is B-complete. Then, since every

complete locally convex space is a c l o s e d subspace of a p r o d u c t

of B a n a c h spaces and since every closed subspace of a B-

complete space is B-complete, it follows that every complete

locally convex space is B - c o m p l e t e which is not true as shown

in C h a p t e r 2, 5~ 5.

15. An a r b i t r a r y direct sum of B - c o m p l e t e spaces, which is

not B-complete.

Take E w = Z R e in ~ 5 of C h a p t e r 2, where R e is a
eeI
copy of the real line. A F r ~ c h e t space is B - c o m p l e t e and so

each R e is B-complete. However, E w is not B - c o m p l e t e as

shown in C h a p t e r 2, 9 & 5.
 CHAPTER 7

TOPOLOGICAL BASES

Introduction

In this chapter we present some counter-examples in

topological bases.

One of the celebrated queries of the subject was the

following:

Does there exist a basis for every separable Banach

space? This query appeared in Banach's Monograph B [~ and

remained unsolved till the appearance of Enflo's paper

P [2~ where the query has been answered in the negative

by a counter-example which is the first counter-example

presented in this chapter. We, first, collect some defini-

tions and results pertaining to the subject matter of this

chapter.

Let E be an infinite-dimensional Banach space. Let

the series Z xi be convergent in E in which case we write

i=I

x = Z xi
i=i

Z X. is called unconditionally convergent if for

i=l 1
every permutation r of the integers the series Z Xr(i)
i
converges i n E.

THEOREM I. Let E be a Banach space and Zx i a conver-

1
gent series i n E. The f o l l o w i n g statements are equivalent.

(a) Z xi is unconditionally convergent.

i
(b) z x. is unordered convergent (that is,
i i
 111

~im Z x. = x, where ~ is the set of all finite subsets

~er ie~ i
of N, directed by 9 ).

(c)
Z x i is subseries convergent (that is, if for
i
every increasing sequence {ni} of integers, the series

Z x converges to some element of E ).

i ni

(d)
Z x i is bounded multiplier convergent (that is,
i
if for each bounded sequence {el} in ~, the series ~ei xi
1
is convergent to some element of E). (Cf. B [ 2 ~ ) .

A series Z x. in a Banach space E is said to be

i= I 1
absolutely convergent if iz=IIlxill is convergent.

Every absolutely convergent series is unconditionally

convergent. The converse is not true for all infinite-

dimensional Banach spaces P[l~

A (topological) basis for an infinite dimensional

Banach space E is a sequence {x i} in E such that to every

xeE, there corresponds a unique sequence {ai } in K for

n
which x = l~mZ ~i x. in the strong topology of E. If
i= 1 1
E is finite-dimensional, it is just the Hamel basis. If

the limit is taken in the weak topology of E, then we call

{x i} the weak basis for E,

The elements of the sequence {ai } which depend line-

arly on x are called the coefficient functionals of the

basis {xi}. The uniqueness of {~i } implies that every

element in {x i} is non-zero. A basis which has continuous

coefficient functionals is called a Schauder basis. Every

basis for a Banach space (even for a Frechet space) is a

Schauder basis.
 112

A basis {xi} for a Banach space E is said to be

boundedly complete if for each sequence {~i } in K with

SUPn I[i<neiXiZ II< ~, there exists an x in E such that

x = lim Z a i x.
n i<n 1

Let {xi} and {fi} be sequences in a Banach space E and

its dual E' respectively. (xi, fi) is called a biorthogonal

system for E if fi(xj) = ~ij

THEOREM 2. {xi} is a (weak) basis for E iff there is

a sequence {fi} in E' such that (xi, fi ) is a biorthogonal

n
system for E and Z fi(x)xi converges strongly
i=l
~espectively, weakly) to x for each x in E. (Cf. B [ 2 ~ )

COROLLARY I. (weak basis theorem) {x i} is a basis

for the Banach space E iff it is a weak basis for E.

A basis (xi, fi) for a Banach space E is said to be

unconditional if for all x in E, the series Z fi(x)x i

i=I
converges unconditionally. It is said to be absolutely

convergent if Z fi(x)x i converges absolutely for every

x in E. It is called normal if IlxiI[ = l[fill = i,

i = 1,2, . . . . It is called E-complete biorthogonal

system if the sequence {xi} is complete in E in the sense

n
that the set of all finite linear combinations Z ~i xi '
i=i
~ieK, n = 1,2, , is dense in E.

A basis {x.} of a Banach space E is

1

(i) monotone if

n n+m
II z ixill -< II z ixi II
I=i i=l
 113

for all finite sequences of scalars ~, , .

an+m

(ii) symmetric if

k
sup sup l l.z ~i fi (x? xa(i) ll< ~'
~cp J~11 1 1=1

I!k<~
where { f i } a r e t h e c o e f f i c i e n t functionals and P i s t h e
set of all permutations of N;

(iii) subsymmetric if it is unconditional and for

every increasing sequence of positive integers {ik} , the

basis {Xik} o f t h e s p a c e s p a n n e d by t h e s e q u e n c e {Xik} i s

equivalent to the basis {x i} ;

(iv) Besselian if Z laiI2<~ whenever Z ~ixi is

i=i i=I
convergent; and

(v) Hilbertian if Z ~ixi is convergent whenever

i=l

Z l a i 12 <~.
i=1

I f ~ i s a Banach s p a c e , a basis { f i } o f E' is called

retro-basis if, for the coefficient functionals {~i } in
E" , { r } c : ( E ) where ~ i s t h e c a n o n i c a l map o f E i n t o

Let E be a topological vector space, E' its dual and

I any index set. Let {x~}%el and {fl}%el be families

of elements in E and E' respectively. (x% , f% ) is

called a biorthogonal system if f% ( x ) = ~EB It is

maximal with respect to E if there is no biorthogonal

system which contains (x% , f% ) properly. (x% , f% ) is

 114

called a generalized basis for E if rE(x) = O, XsI,

implies that x = 0 for all x in E. It is called an extended

Markushevich basis for E if {xx} is total in E; in parti-

cular, if I is countable, then (xx, f~) is called a Mark-

ushevich basis for E.

As in the case of Banach space, a sequence {xi} in a

topological vector space E is a basis for E if for each x

in E, there is a unique sequence {~i } in K such that

x = lira Z eixi
n i<n

in the topology of E. Each expansion coefficient ai

defines a linear functional fi' fi (x) = ai ' on E. These

coefficient functionals fi' however, need not be continuous

(see ~ 21). If the coefficient functionals {fi } are

continuous, then {xi} is called a Schauder basis.

Let (xi, fi) be a Schauder basis for a topological

vector space E. Let

n
Sn(X) = Z fi(x) x i , x~E, n = 1,2,
i=:

{x i} is called

(i) an (e)-Schauder basis of E if {s n} is an equi-

continuous subset of ~(E), th~ space of continuous linear

maps of E into itself;

(ii) a (b)-Schauder basis of E if {s n} is a bounded

subset of L(E), where ~(E) is equipped with the topology

of uniform convergence on bounded subsets of E.

If E is a normed space, then (e)-Schauder bases and

 115

(b)-Schauder bases are equivalent.

For a topological vector space the following implica-

tions hold:

Schauder basis
4
Markuschevich basis
Extended Markushevich basis

Generalized basis

Maximal orthogonal system.

L e t E and F b e two t o p o l o g i c a l vector spaces. A

sequence {xi} in E is said to be similar to a sequence

{yi} in F if for all sequences {~i} in K, Z ~ixi

i=*

converges (in E) iff I eiy~ converges (in F).

i=l

THEOREM 3. (Isomorphism Theorem) Let E and F be

barrelled spaces and (xi, fi) and (Yi' gi ) Schauder bases

in E and F respectively. Then (xi, fi ) is similar to

(Yi' gi ) iff there is a topological isomorphism T:E § F

such that Tx i = Yi for all i~l . (Cf. B [ i ~ )

i. A separable Banach space which has no basis.

cf. p [2q
2. A Banach space with a basis, whose dual does not

have a basis.

Consider the Banach space E = s It has a basis,

namely (8i} where 8i = (6ij)j= i ' ~ij being the Kronecker

delta. But the dual of E is ~ which is not separable and

does not have a basis.

3. A Banach space which has no unconditional basis.

 116

Consider the Banach space E = C [0,~ of continuous

functions on the closed interval [O,~ with the s u p r e m u m

norm topology. Since E' is not separable, it does not

have a basis. Now, since E' is w e a k l y sequentially com-

plete, E has no u n c o n d i t i o n a l basis in view of the

following: If (Xn, fn) is an u n c o n d i t i o n a l basis in a

Banach space X and if X' is w e a k l y sequentially complete,

then {fn} is an u n c o n d i t i o n a l basis in X'.

4. A Banach space with a basis w h i c h is not uncondi-

tional.

Consider the Banch space c o of sequences converging

to O, equipped w i t h the supremum norm topology. Let

i if j < i
X- -
={
z3
0 if j > i .

Let {xij } = x i . We show that {xi} is a basis for c o .

Let {6ij} = 6 i , where ~ij is the Kronecker delta. Then

the sequence {fi} in s given by

fi = 6i - 6i+i '

is b i o r t h o g o n a l to {x i} . Since the closure of the linear

span of {x i} is c o , and since

sup l[Tn~[[ = sup II z fi(a)xi[ I

n n i<n

= sup ]] Z (a i - ~i+l)Xil]
n i<n

= sup sup {l~j - "nl ; j<n

n
 117

2 sup lanl = 2 I]~l[ ,

n

it follows that {xi} is a basis for c o in view of the

following: If (xi, fi ) is a biorthogonal system for a

Banach space E such that sup If(TnX) I< ~, xeE, feE', then

{x i} is a basis for the closure of the linear span of

{x i} and {fi } is a basis for the closure of the linear span

of {fi } . Now, we show that {xi} is not unconditional. Let

(-i) i+l
= ECo
i

We then take the subseries of all odd terms in the expan-

sion of ~ : With i as odd integer,

7- f. ( a ) x i = 7 (a i - a i + ~ ) x i
i=I i i:I

{ T, (-i) a. )
i:2j_1 i

co
-1
= { ~ i }
i=2j-1

We conclude that the series expansion for a is not sub-

series convergent (because the series 1 + { + -13 +

is divergent) and hence not unconditional convergent.

5. A Banach space with a n unconditional basis which is

not boundedly complete.

Consider the Banach space co Let~ ={ai } c c o

Then m = lim Z ~i 5.
n i<n i
!

In view of the equation a i = fi(a) where fi e c o = s is

given by fi = 6i ' and Theorem 2, {~i } is a basis for c o

 118

The basis is unconditional, since the definition of

supremum norm in co implies that the expansion for a is

subseries convergent, and hence unconditionally convergent.

Now, the sequence a = (I,I, ) in K is such that

sup [l Z a i ~i[I = 1 .
n i<n

But Z ai ~i is not convergent in co

i<n

6. A Banach space with a basis which is not absolutely

convergent.

Consider the Banach space E = ~P, l<p<=. The bior-

;

thogonal sequence to {~i } is {fi } , where fit~p = ~q

( ~i + ql = i) is given by fi = 6i . Since for a = {ai } t~p'

fi(a) = a i, Z fi(a)6 i converges to a, and so {6i } is a

i<n
basis in ~P Now, let {ni} be any increasing sequence of

integers. Since ~P is complete and

Z fn. (c~) 1 ( T IC~nllP) I/p

I [nit (iN) i 6nil = nit (~q) "

< ( z ]aiIP) 1/p

- it (~)

for e a c h a i n zP a n d e a c h f i n i t e subset ( N ) of integers,

the expansion for a is subseries convergent. Thus, {6i }

is an u n c o n d i t i o n a l basis b y T h e o r e m 1. To show t h a t {6i}

is not absolutely convergent, let p' be such that l<p'<p .

Then
-p'/p
a = {i } t~ p
 119

But

i<n
I llfi (~) ~il[ = z fail
i<n
= z
i<n
i-p'/p

n
> f t -p'/P dt
-- 1

-I 1-p,/p
= (1- 2p ' ) (n -1)

which diverges as n § ~.

7. A Banach space with a basis which is not a normal

basis.

Consider the Banach space E = C [0,~ as defined in

3. The sequence

x o(t) = 1 , x l(t) : t ,

r2s zs )
, ~ ,
i O if t, v2-]~--/T

x
2s
(t) : 1 if t =
k
2+s
r 2s 2s l and
linear in L---wl-r- ,
2~T* 2-V -u
2s , 2s

(s = 1,2, ...,2 k ; k = O,1,2, ...) constitutes a basis for

C [O,i] . But the sequence of coefficient functionals

fo(x) = x(O), f1(s) = x(1) - x(O) ,

f (x) = x r, 2s
~ ) r 2s
- I ~ , ~
) 2s
-'+~ (Z-!T
~)
2k+i

(x~c [o,1] ; s 1,2 .... , 2 k ; k : 0,1,2, ...)

is such that
 120

llfoll : i, llfnll : 2,(n = 1,2, ... 3.

8. A Banach space whose dual space has a normal basis

which is not a retro-basis.

Let E be the real vector space of all real sequences

x = (Xn) with

lim xn = 0
n

and
n
l lxll = sup {i=IZ (Xkzi_1- Xk2i)2 + (Xkzn+1)2}< ~

where the supremum is taken over all positive integers n

and finite increasing sequences of positive integers

kl , ... , k2n+l E is a Banach space. Define

6 i = (Sij); = i

Then {8 i} is a basis in E. Let {fi } be the corresponding

coefficient functionals. Define a sequence {gi } in E' by

gz = fz ' gn = fn-1 -fn (n = 2,3, ... ).

Then {gi } is a normal basis for E', but not a retrobasis

for E' (For a proof, the interested reader is referred

to B [33], p a g e 2833.

9. A Banach space with a Besselian basis which is not a

Hilbertian basis.

The unit vector basis 6 i = (Sij)j= I of the Banach

space ~l is a Besselian basis which is not a Hilbertian

basis.

iO. A Banach space with a Hilbertian basis which is not

 121

a Besselian basis.

The u n i t vector basis ~i = (~ij) of the Banach space

Co i s Hilbertian but not Besselian.

ii. A Banach space with a basis which is not a monotonic

basis.

Consider the Banach space E = C [O,1] as defined in

~/~3.
Define

Xi(t ) = ti , i = 1,2, ... , t e[O,l] 9

We can select an infinite subsequence {Xik} of {x i} ,

which is a basic sequence in E. Then the (closed) subspace

E 0 spanned by the subsequence {Xik} has no monotonic basis.

(For a proof, the interested reader is referred to B [33],

pages 241 - 248).

12. A Banach space with a sub-symmetric basis which is

not a symmetric basis.

Consider the Banach space w of all real sequences

x = (Xn) equipped with the norm

l]x[[ = sup Z I Xn i
i=I .... [ ; {ni}c~}<
i
where

= {{ni}e ~ ; nl<n2 }

The unit vectors {~i} , ~i = (6ij form a sub-

symmetric basis of w which is not symmetric. (For a proof,

see B [3~, page $83).

 122

13. A Banach space without a subsymmetric basis.

The Banach space ]__p_ Eo,l], l<p<~ , p ~ 2, does not

have subsymmetric basis (B [33] , page 563).

14. An E-complete biorthogonal system in a Banach space,

which is not a basis 9

Let E be the space of all continuous functions on

having period 2z and equipped with the norm

iIxll = sup {Ix(t)l ; tell }.

Let

Xo ( t ) = 1

x2i_l (t) = sin(it)

x2i(t) = cos(it), tell, i = 1,2, ...

and

f2i(x) = 1 ~x(s) cos(is) ds ,

f2i+z (x) = 1 f~x(s) sin(i+l)s ds, xeE, i = 1 2, 9 9

Then (xi, fi) is an E-complete biorthogonal system, since

{x i} is complete in E. However, from the existence of con-

tinuous functions whose Fourier series are not uniformly

convergent, it follows that {x i} is not a basis of E.

15. A normed space with a basis which is not a Schauder

basis.

Consider the Banach space E = C [O,i~ as defined in

@r 3 and let
xi(t ) = t i-I , t E [ O,i ] , i = 1,2, ..,

Then {xi} is w-linearly independent(that is, Z ei x'= 0

i=l l '
 123

{ ~i } c ~ , implies a i = O, i = 1,2, ...) but not

minimal (that is, x i does not belong to the subspace

spanned by {x i , ..., xi_1 ' xi+1...} by B [ 2 5 ( a ~ ,

Chapter III~ 3, Theorem 2). Also the span of the sequence

{xi} is E itself.

Let
oo co

M = { r ai xi EE ; {ai} c • , Z ai xi converges }
i=I i=l

Then M is a normed space under the induced topology from E.

{x i} is a basis for M but not a Schauder basis for M.

Infact, if there is a sequence {$i } c M' such that

$i(xj ) = ~ij ' i,j = 1 , 2 , ... ,

then since M = E , we can expand each $i' by continuity,

to an f.sE' and we h a v e
1

fi(xj) = 6ij ' i,j = 1,2, ...

which contradicts the assumption that {x i} is not minimal.

16. A normed space with a Schauder basis which is neither

an (e)-Schauder basis nor a (b)-Schauder basis.

Consider t h e s p a c e E o f ~= 14. Let (xi, f i ) b e t h e E-

complete biorthogonal system such that {x i} is not a basis
o f E as i n ~ 14. Consider the subspace

n
M = {xeE ; llm Tn(X) = lim Z fi(x) x i = x } ,
n n i=l

equipped with the induced norm topology. Then {x i} is a

Schauder basis for M, but not a (b)-Schauder basis for M.

Infact, if

sup { l I T 'n i l ; l<n<~}<

 124

where T'n is T n r e s t r i c t e d to M, then, since M = E, we

have

sup iIITnll ; fin< ~ } <

Hence there is a c o n s t a n t K>I such that

l[Tn[[~ K, n = 1.2, .

Now f o r every finite linear coinbination

In
p= ~ ~. x.
j=1 ] )

and all i>in, we have

n In
T n (p) = Z Z
fi ( j=l aj xj) x.
i=l i

n In
Z Z aj ~ij x.
i=l j=l i

In
= F. ~j X. = p
j=: J

so that liin T n ( X ) = x for all x in E. But this contra-

n

dicts the fact that {xi} is not a basis for E. Since an

(e)-Schauder basis is e q u i v a l e n t to a ( b ) - S c h a u d e r basis

for a norined space, it follows that {x i} is not a (b)-

Schauder basis for M.

17. A Banach space w h o s e dual has a weak* basis but no

basis.

Consider the Banach space E = ~l Then the unit

vectors in E' = %~ c o n s t i t u t e a weak* basis for E'

(because a sequence {fi } in a conjugate Banach space E'

is a weak* Schauder basis for E' if and only if E has a

basis ix i} whose coefficient functionals are {fi } ); but

E' = ~ has no basis since it is not separable.

 125

18. A Banach space whose dual space has a basis which is

not a weak* basis.

Consider the Banach space E = c o , the space of

sequences converging to O, equipped with the supremum norm

topology.

Define the sequence {hi} in E' = s by

hl = fl, h i = 6i_ I ~i ' i = 2,3,

where 6 i = (~ij) is the unit vector basis of ~i The

sequence ~ = {i , , i, O,O, . } ,

i occuring i times and i = 1,2, , is a basis for E.

(This follows from the following result: Let {xi} be a

bounded basis in a Banach space E and {~i } a sequence of

scalars such that ei ~ O, i = 1,2, .... The sequence

{yi } in E defined by

i
Yi = Z a n x n , i = 1,2,
n=l

is a basis for E if and only if the sequence

lJYi§

is bounded). Identifying canonically Z I with E' , the

sequence {hi+ I} is obviously the sequence of coefficient

functionals for the basis {yi }of E, whence {hi+ I} is a

basis for the closed subspace of E' spanned by {hi+ I} ,

because if [xi, fi) is a basis for a Banach space E, then

{fi } is a basis for the closed subspace [f~ of E'

spanned by {fi } and

f = E f(x n) fn ' fe [fi ])"

n=l

NOW, h I ~ [hi+1~ , since for the functional

 126

oo

~0[f) = Z ~ n ' f = (~i)~E' , we have

n--I

~o(hz) = i, ~o(hi+z) = O, i = 1,2, ... , and {h i } is

complete in E' , since the relations

~EE", r = O, i = 1,2, .

imply ~=O. Consequently {h i } is a basis for E' Further-

more, although every fEE' has a weak* expansion

oo

f(x) = Z a n hn(X ), xcE,

n=l

this weak* expansion is not unique, since by

i
hi(x ) = Z hn(X) = fi(x), xsE, i = 2,3, ...
n=2

we have

hz(x ) - Z hn(X) = O, x~E.

n=2

Thus, {hi} is not a weak* basis for E'

19. A Banach space w h o s e dual space has a weak* basis

which is not a weak* Schauder basis.

Consider the Banach space E = c o and define the

sequence {fi} in E' = ~ by

(*) fz = 81, fi = (-I)i+I~ i + 8 i i = 2,3, ...

where {6 i} is the unit vector basis of ~z

Then {fi } is a basis for E' in the norm t o p o l o g y and a

weak* basis for E'. To show that it is not a weak*

Schauder basis, we p r o c e e d as follows: Let {~i }=E'' be

the c o e f f i c i e n t functionals for the basis {~i } of E'

Define
 127

(**) X (f) = r (f) + m =2z (-I) m Cm(f) ,

fEE', Xi = $i ' i = 2,3, ....

Then (fi' Xi) is a biorthogonal system and we have, for

fEE', i = 1,2,
i oo

F Xn(f) fn = l~1(f) + Z (-i) m Cm(f)]61

n= 1 m=2
i
+Z ~m(f) I(-I)m+161 + ~m I
m = 2

Z Cm(f)6m + I ~
m= I m=i+ I
(-l)m Cm(f) l~ i

oo
Hence, since Z lCm(f) I< co (feE' = s
m = 1

for every e>O and feE', there exists an integer N>O (depen-

ding on e and f) such that

i i
II Z Xm (f) f m - Z ~m(f) 6ml l<e ,
m=l m=l

for i>N. Hence

co

f = Xm (f] fm ' feE'

m=l

From this, it follows that {f.} is a basis for E' in the

1

norm topology. Furthermore, it follows from the above

that every feE' also has a weak* expansion

co

f(x) = Z Xm(f ) f (x) xeE .

m= I m '

We now show that this expansion is unique. Let

oo

(***) Z a m fm (x) = O, xEE,

m= 1

for a sequence {~i } of scalars. Since, for x = b i

(i = 1,2, ...), where {b i} is the unit vector basis {~i }

 128

of E = co , we have, by (,) ,

fm (bl) = (-l)m+z' fm (bi) = ~mi '

(i = 2,3, ; m = 1,2, ).

and by (***) it follows that

Z (-i) m+1 = O, a = = ai = = 0 ,
m 2
m=l

whence ~. = 0 (i = 1,2, ). Thus { f i } i s weak* b a s i s

l

for E' On t h e otherhand, b y (**) Xl ~ g(E), where ~ is

the canonical map o f E i n t o E', and X i i s n o t weak* c o n -

tinuous. Thus { f . } is n o t a weak* S c h a u d e r b a s i s for E'.

1

20. A separable locally convex space which has no basis.

! T
Consider the locally convex space (s , a(Z , s )),

~ = Z~ It is s e p a r a b l e since the sequence {fi } in

s defined by

fi (x) = ~i ' x = (~i) i = 1,2,

is total on s , whence its w * - h l o s e d linear hull is = .

Suppose it has a weak* basis {fi } . Then, to e v e r y f in

s there is a u n i q u e sequence {a i } such that

n
f = lim Z a.f.
n i=I 1 1
y
in the topology o(~ , s ). But then the series c o n v e r -
s
ges w e a k l y to f. Thus {fi } is a w e a k basis for = and

hence a basis for But then the B a n a c h space = is

separable in the strong topology and so s is s e p a r a b l e in

the strong topology which is not true.

21. A basis in a l o c a l l y convex space, which is not a

Schauder basis.
 129

Let E be the space of all r e a l - v a l u e d functions expan-

dable as absolutely summable power series on the interval

[0,i ) w i t h the topology of u n i f o r m convergence on compact

subsets of [0,i). Let {xi} be a sequence in E defined by

xi(t) = t 1, i : 0,I, . . . .

Then {xi} is a basis for E. (This follows from the follow-

oo

ing result : If two power series Z a n (z - z0) n and

n=O
O0

(z - z0) n converge on a n e i g h b o u r h o o d of zo
Bn
n:O
absolutely to the same sum, then a n : Bn for all n). For

each polynomial P in t, the c o e f f i c i e n t functional fl ,

determined by the expansion c o e f f i c i e n t a,, is given by

PCt) - P ( O )
f, (P) : lim
t§ t

Since the p o l y n o m i a l s are dense in the Banach space

C [0,~ , we can choose a polynomial z n in E for each n

such that
I
sup { [ Z n ( t ) - (1-nt)[ ; te[0,1/~}s ~

and

sup { [ Z n ( t ) ] ; t e ( ~ , 1 )}<

Define

Yn(t) : ft Zn(t' ) dt' , te[O,l )

0

Then Yn is a polynomial. Therefore

Yn ( t ) 1 I + ]I -
f1(yn) : lim - Zn(O)e [ i - H ' n '
t+O t

(n = 1,2, .), so that

lim fz(yn) = 1 o

But lim Yn = O, since

n
 130

sup { l Y n ( t ] [ ; t r oft[ Z n ( t ) [ dt

< 1 1 3
-- n-~ +
n 2n

22. A complete, metrizable and separable (non-locally

convex) topological vector space which has no basis.

Consider the complete, metrizable and separable (non-

locally convex) topological vector space E : [_P [o,1] ,

O<p<l, as d e f i n e d in Chapter 1, W/= 6. It is separable.

Since E' = { 0 }, as shown i n C h a p t e r 1, q~ 6, E h a s no

Schauder basis and h e n c e no b a s i s .

23. A generalized basis in a non-separable Banach space,

which is not a Markushevich basis.

Let E be the Banach space of all bounded real-valued

functions on an infinite set I, under the supremum norm.

For each i~I, define x i as the function assuming the value

1 at i and 0 elsewhere. The point functions fi(x) = x(i)

are all continuous and {fi} is biorthogonal to {xi}

Since fi(x) = O implies x = O for all x in E, it follows

that {xi} is a (countable) generalized basis in E. It is

easy to see that E is non-separable and {xi} is not total

in E. That {xi} is not a Markushevich basis in now clear.

24. A Markushevich basis in a Fr~chet space, which is

not a Schauder basis.

Let E be the Fr~chet space of all functions analytic

on the open unit disc Izl< l, equipped with the topology

via the metric of uniform convergence on compact sets. We

set
 131

fn (z) = (z - a ) n, n = O,1,2 .

where a is any complex number satisfying the inequality

O<laI< i. These functions form a Markushevich basis in E,

the corresponding biorthogonal sequence of continuous

linear functionals being given by

f(n) (a)
Cn(f) = n~ , n = O,1,2 .

On the other hand, {fn} is not a Schauder basis, since

the series E (z-a)n corresponding to the functions

n=O (l-a) n+1
- l

f(z) = (l-z) diverges for z outside the circle

I z -a I = I1 - a I

25. A maximal biorthogonal system in a Frechet space,

which is not a generalized basis.

Let E be the Frechet space as defined in ~= 24. We

set

f0 (Z) = l,

fl (Z) = i + z ,

9 . 9 , 9 9 9 , 9 9 9 9 9 ,

fn(Z) = l+z+ .... + zn ,

9 ~ 9 9 9 9 9 , 9 . 9 9 9 .

Define r on E by

f(n+1) (IO)
Cn (f) = f(nJ (0) _ , n = O,I, . . . .
n' (n+1) :

Then (fn' Cn ) form a biorthogonal system over E. Since

{fn } is total in E, {r } is the only possible sequence of

coefficient functionals. But the ~n(f) = O, n = O,i,..., does

not imply that f = O ; infact the condition Cn(f) = O,

 132

n = O, I, is necessary and sufficient for f to be of

the form

f(z) = C(1 + z + z2 + + zn + . )

where C is an arbitrary constant. Thus, {fn } is not a

generalized basis in E. Now, if (fn' ~n ) is not maximal,

then there exists a continuous linear functional annihila-

ting {fn} but not vanishing identically. But this is

impossible, since every continuous linear functional ~ on

E has the representation

r = z f(o] hn
n=O n!

for a suitably chosen sequence {hn} o f c o m p l e x n u m b e r s .

26. An e x t e n d e d unconditional basis in a countably

barrelled space, which is not an extended unconditional

Schauder basis.

Let E and E' be as in Chapter 3, g/= 43. Choose kocA 9

Define

Yk
by
1 if U=X or >=Xo-
=

0 otherwise .

Then {yl} is an extended unconditional basis in E. But if

x = (~k)~E, ~cA

then

x = Z ~ y~ + (~ - Z ~ )
l~X0 X0 lfk0 ~ Yl0 '

and the coefficient functional of Yl0 clearly does not

belong to E'.

27. The isomorphism theorem fails if the domain or the

 133

range space is not barrelled.

(i) L e t E be a B a n a c h s p a c e w i t h a Schauder basis

(x n , fn ). L e t F be E e q u i p p e d w i t h t h e weak t o p o l o g y . Then

{x n} i n E and {Xn} i n F a r e similar, b u t E and F a r e n o t

homeomorphic.

(ii) Let (Xn, fn) denote the unit vector basis in

E = ~i. Let F be ~i considered as a dense subspace of co

and let (Yn' gn ) denote the unit vector basis of F. Then

{x n} and {yn } are similar, but E and F are not isomorphic.

28. The isomorphism theorem does not hold for generalized

basis even if the domain and the range spaces are complete
and barrelled.

L e t E be an i n f i n i t e dimensional Banach space with a

generalized basis (xk , fk ) , and l e t F be E e q u i p p e d w i t h

the finest locally convex topology. Then E and F a r e com-

plete barrelled spaces and (xk , fk ) i s a generalized

basis i n F. Clearly {xk} i n E and {xk} in F are similar.

But, since F is not metrizable, E and F a r e n o t h o m e o m o r -

phic.

29. A vector space with two c o m p a t i b l e locally convex

topologies such that there is a Schauder basis for one

topology, which is not a Schauder basis for the other

topology.

Let r denote the space of finite sequences and e t h e

sequence with every coordinate equal to I. Define

= {x = ( x i ) ; (xi) is eventually constant}

={~e + x ; ~ i s a scalar, xr162

Clearly the KDthe dual kx is tl and the sequence (e (n))

is a ~(k, XX)-Schauder basis for k , where e (n) is the

sequence with 1 in the n th place and zeros elsewhere. We

now show that (e (n)) is not a T(X, ~)-Schauder basis for

X 9 Let

u (n) = e (n) _ e (n+l)

be considered as an element of X• Then,

although for each m,

lim < e (n) u (m) > = < e, u (m) >

n

the convergence is not uniform with respect to m. Infact,

< e (n), u (n) > - < e, u (n) > = 1

for all n. Hence we need only to show that the convex

circled hull A of the set {u (n)} is w e a k l y r e l a t i v e l y

compact. Let S be the closed unit ball in ~i Then

A = { a = (ai) ; a = Z Bj u(J) B= (Bj)~SOQ }

j=1

Hence A is a c o o r d i n a t e w i s e bounded subset of the space w

of all sequences, and since (w, o(w,~)) is a Montel space,

A is a ( w , r compact. Furthermore, if ae + xck

and acA, then

(*) < ae + x,a > = a < e,a > + < x, a >

co
= a Z B < e u (j) > + < x, a >
j_-~ J ,

= < X, a >

Therefore, if (a ( v ) ) is a net in A, there is a cofinal

subset w h i c h we denote by ( a (~)) itself, and an element

 135

a(~162 such that if ae +x.r , in view of (.) ,

lira < ae + x, a (~) > = lira < x, a (v) >

= < x, a (~ >

How a l l that we n e e d t o show i s that ~~ ~*and that

(*) remains valid if a is replaced by a (~ for this means

that a (~ is a o(~, X) - l i m i t point of the original net.

Since A c 2S, a(~)e2S for all ~. Since a (~ is the coordi-

natewiselimit of the net ( a (~)),

a (~ e 2S c ~z= ~•

Furthermore, we m a y write

r (x~) u ( j ) (,o) (,o))e Srldp

a (') = Z B. B = (Bj
j=1 J

Hence
(~) (,) (,) (,)
ai = Bi - Bi_ l , ( Bo = 0 )

and s o

(,) ('o) (',,)

Bi = aI + . + ai

and s i n c e ( a (v)) is coordinatewise convergent to a (~

we may d e f i n e
(o) (.) (o) (o)
Bi = lira Bi = aI + . + a

Therefore, (B(~)) is a net in S, which is coordinatewise

(o)
convergent to B (o) = ( ~i )
and so
(0)
B r S c~ 1
Hence
 136

(o)
lim 8i = 0
i

Therefore,

< ae + x, a(~ = a < e, a (~ > + < x, a (~ >

oo ~ (o)
a Z a. ~ + < x, a >
i=l

(o) >
= ~ lira 6 (~ + < x, a
i x

= < x, a (~ >

Hence a (~ satisfies (.)

 CHAPTER 8

TOPOLOGICAL ALGEBRAS

Introduction

A subset S of an algebra is called a left (right)

ideal of A if S is a vector subspace of A such that xy~S

(respectively yxeS) for all xeA and yeS. S is called a

two-sided ideal if it is both left and right ideal. S is

called maximal if it is different from A and is not pro-

perly contained in any ideal of the same type except A.

A left ideal L in an algebra A is regular (or modular) if

there exists e~A such that A(l-e) e L ; in other words, the

element e is a right identity for A modulo L. Similarly a

right ideal R is regular (or modular) if there exists a

left identity for A modulo R. A two-sided ideal I is

regular if it is regular both as a left and as a right

ideal. The intersection of all the maximal ideals in A is

called a radical. If the radical R is {0} , then A is

called semi-simple. A subset S in an algebra A is called

idempotent if S S ~ S and m-convex if it is idempotent and

convex. The circle operation in A is defined by x o y =

x+y-xy. An element of A which has a left (right) inverse

relative to the circle operation is said to be left

(right) quasi-regular. If r is both left and right quasi-

regular, then it is called quasi-regular.

The set of elements of an algebra whose inverses

exist in A, is denoted by G(A).

A normed space over K which is also an algebra is

called a normed algebra if ][xy]lj][x[[ ]]Y]I If the

 138

underlying space is a Banach space, then it is called a

Banach algebra. It is commutative if xy = yx for all

x,yeA. An element e of a Banach algebra is called identity

if lleI[ = i. A family {e ; ~cI } of elements of a

Banach algebra A where I is a directed set, is called an

approximate identity if lle II~ 1 for each ~ and

lim e x = lim x e = x for each xeA.

A Banach algebra A is called a Banach * - a l g e b r a if it

has an involution, that is, if there exists a mapping x§

of A into itself with the following properties:

(a) (x+y)* = x* + y* ;

(b) (Xx)* = ~x* ;

(c) (xy)* = y* x* ;

(d) x** = x

A B*-algebra is a Banach *-algebra which satisfies

llx* xll = llxll 2 An A* -algebra is a Banach * - a l g e b r a

in w h i c h there is defined a second norm IIIxIII which

satisfies, in addition to the m u l t i p l i c a t i v e condition

[I[xYlII~ lllxlll I[lyIII , the B * - c o n d i t i o n IIIx*xl]l =

II[xl[[ 2 The second n o r m is called an auxiliary norm. A

Banach *-algebra A is called symmetric if every element

of the form -x* x is q u a s i - r e g u l a r in A.

A locally convex space A w h i c h is also an algebra is

called a locally convex algebra if the ring m u l t i p l i c a t i o n

in A is separately continuous. A complete m e t r i z a b l e

locally convex algebra is called a Frechet algebra. A

locally convex algebra A is called locally m - c o n v e x if

 139

there is a basis of neighbourhoods at 0 consisting of m-

convex and circled sets. The set of all non-zero multipli-

cative linear functionals of A is denoted by m~(A) and

continuous members of m ~(A) by m(A) If m(A) is non-empty,

we topologize it as follows: Let A' have the weak topology

and let m(A) have the relative topology it inherits as a

subset of A'. With this topology, m(A) is a completely

regular Hausdorff space. We define a map G:A+c(m(A)) by

G(x) (f) = f(x) for each xEA and fcm(A). G is called the

Gelfand map.

An element x of a locally m-convex algebra A is

called a left topological divisor of zero if inf {p~(xy) ;

p~(y) = i} = O. Similarly we can define right and two-

sided topological divisors of O.

Let ~ be a class of locally m-convex algebras and

Ae~. Let xeA. Then x is said to be E - s i n g u l a r if it is

singular (~hat is, the inverse of x does not exist) in any

superalgebra BmA belonging to the c l a s s y .

A locally m-convex algebra A is called a P-algebra

if {x ; xn§ is a neighbourhood of O. A locally convex

algebra A is a Q-algebra if the set of quasi-regular

elements of A is open in A. A is a Q-algebra if and only

if the set of quasi-regular elements of A has an interior.

A subset S of a locally m-convex algebra A is said to

be m-bounded if for some %>0, kS is contained in a bounded

and idempotent set. A locally m-convex algebra A is said

to be p.i.b, if for all x~A, (x} is m-bounded. Any normed

algebra is p.i.b.. A locally mrconvex algebra is called

 140

m-barrelled if every m-barrel (that is, idempotent barrel)

is a neighbourhood of O. A locally m-convex algebra is

called countably m-barrelled if every m-barrel which is

the countable intersection of circled, convex and closed

neighbourhoods of 0 is itself a neighbourhood of O.

Let A be a locally convex algebra. An element xeA is

said to be bounded if for some non-zero complex number I,

the set {(Ix) n ," n = 1,2, ...} is a bounded subset of A.

The set of all bounded elements of A is denoted by A U

We write ~,to denote the family of all subsets S of A such

that S is circled, convex, bounded, closed, eeS and S2c_S.

For each S in ~i , let A(S) denote the subalgebra of A

generated by S. Then,

A(S) = {Ix ; I~C , xES}

and

I lxl Is = inf {x>o ; xeIS} , xeA(S)

defines a norm on A(S) which makes A(S) a normed algebra.

A locally convex algebra A is pseudo-complete if each of

the normed algebras A(S), SeBI, is a Banach algebra. If A

is sequentially complete, then it is pseudo-complete.

PROPOSITION I. If 9~ contains a basic sub-division

92 (i.e. for every B1e 91there is some B2 ~ 92 such that

Bic B2) such that A(S) is a Banach algebra for every Se92,

then A is pseudo-complete. (C~, P [i])

A locally convex algebra A with a continuous involu-

tion x § x* is called a locally convex *-algebra. With

identity e, it is said to be symmetric if, for every xeA,

e+x*x has a bounded inverse (that is, it has an inverse

 141

belonging to A ).

A pseudo-complete locally convex *-algebra A with

identity e is called a GB*-algebra if A is symmetric and

~* = {Sc~ ; S = S*} has a greatest number.

Let A be a locally convex space which is also an

algebra. A subset V of A is said to be left (multiplica-

tively) absorbing if aV is absorbed by V for every a in

A. It is right (multiplicatively) absorving if Va is

absorbed by V for every a in A. It is (multiplicatively)

absorbing (m-absorbing) if it is bo~h left and right

absorbing. Let p and q be two members of the family P of

semi-norms generating the topology of A. p is said to

absorb q if there exists a positive real number M such

that q(x)J Mp(x) for every x in A. The semi-norms p and q

are said to be conjugate if they are mutually absorbing.

The left-translate ap (right-translate pa ) of a semi-norm

pep by the element a of A is defined by aP(X) = p(ax)

{respectively, Pa(X) = p(xa)} for x in A. A semi-norm p is

left (right) absorbing if it absorbs all of its left

(right) translates, and absorbing if it is both left and

right absorbing. A is called an A-convex (absorbing convex)

algebra if there exists a family P of absorbing semi-

norms defining the topology of A. Every locally m-convex

algebra is an A-convex algebra.

A p-normed space A which is also an algebra is called

a p-normed algebra (locally bounded algebra) if the p-norm

[I'I[ , O<pjl. satisfies the following:

Ilxyll~ Ilxll Ilyll

 142

A locally semi-convex space which is also an algebra

is called a locally m-semi-convex algebra if its topology

is generated by a family {qa} of k -seminorms satisfying

the following:

q~(xy)j q~(x) q (y).

i. An algebra which can not be made into a Banach algebra.

Consider the algebra A = ~ [ O , ~ of infinitely diffe-

rentiable complex functions on [ O , ~ There is no norm on

A which makes it a Banach algebra. For, otherwise there

would exist, for any integer m>O, a number cm such that

sup {If(m)(x)] ; xeEo,~}j Cmllfll

for all lEA. But we can construct a function f satisfying

[f(m)(O)l~ mc m

for all m.

2. A Banach algebra which has no radical.

(i) The Banach algebra ~ m [ O , l ] of complex functions

m times continuously differentiable on [O,i] with the usual

pointwise multiplication and norm defined by

m
II f[[ Z -~, sup [f(P)(x)[ ,
p=O
has no radical.

(ii) The Banach algebra C(X) of continuous functions

on a compact space X with the usual supremum norm and

pointwise multiplication has no radical.

3. A Banach algebra with a closed ideal which is not an

 143

intersection of maximal r e g u l a r ideals.

Let G be a non-compact locally compact abelian group.

Then the Banach algebra ~ ( G ) with convolution as multi-

plication contains a closed ideal which is not an inter-

section of maximal regular ideals. (See P E 6 ~ )

4. A Banach algebra with an approximate indentity which

is not an identity.

(i) Let G be a non-discrete topological group. The

Banach algebra ~ ( G ) with convolution as the multiplication

does not have an identity, because ~(G) will have an

identity if and only if G is discrete. But ~ ( G ) does

always contain an approximate identity, namely the family

{e V ; V~u }where u is the set of all compact neighbour-

hoods of the identity in G and eV is any non-negative real

function on G which vanishes outside V and for which

fev(t ) dt = I.

The partial ordering Vz<V2 in u is defined by the inclusion

V2cVI We then have

lim (eV * f)=lim (f ,ev) = f

V V

for every f in L~(G).

(ii) Any B*-algebra is also an example of a Banach

space with approximate identity consisting of hermitian

elements, which is not an identity. (See P E 8 ~ ) .

5. An A*-algebra which is not a B*-algebra.

Consider the Banach algebra [~(G), where G is a

locally compact topological group and convolution is the

 144

multiplication. With an involution x+x* defined by

x (t) : A ( t -1) x ( t - 1 ) , teG,

L~(G) i s an A * - a l g e b r a which i s n o t a B * - a l g e b r a .

6. An A*-algebra which is not symmetric.

Let G be the group of all complex 2• matrices (aij)

with determinant equal to I. Let u be the subgroup con-
sisting of unitary matrices in G and denote by M the
subalgebra of L~I(G) consisting of all f in ~(G) such that

f6 = f~ = f' 6cu , where f6(x) = f(6-1x),

f6(x) =f(x6-:),xEG.

Then N is a closed and commutative *-subalgebra of ~ ( G ) .

Furthermore, M contains quasi-inverses, that is, if feM
and f is quasi-regular in ~(G), then it is quasi-regular
in M. But M is not symmetric so that [!(G) is not symmetric.

7. A Frechet algebra which is not a Banach algebra.

Consider the a l g e b r a ~ [ O , ~ as defined in ~ I, with

seminorms

Pn(t) = sup {If(n)(x)I ; xe [ 0 , ~ }

We replace this system by an equivalent system of semi-

norms given by

qn(f) = 2n sup { 2 i} sup { I f(i)(x)i }.

O<i<n O<x<l

We then have
qn (f)
supl f(i) (x) l!
X 2n+1

and so it is easy to see that

 145

qn(fg) < qn (f) qn(g)

Hence~[O,l] is a locally m-convex and Frechet algebra.

But it is not a Banach algebra as shown in ~/]= i.

REMARK. This is also an example of a Q-algebra which

is not a Banach algebra.

8. A Fr~chet algebra which is not a locally m-convex

algebra.

Let ~ denote the set of equivalence classes of

measurable functions on [ O , ~ such that

pn(f) = (fllf(x) l n dx) 1 / n <~ , n = 1 , 2 ,

0

Then [~ is a Fr~chet algebra with pointwise multiplication

and t h e topology generated by the semi-norms {pn } We

show t h a t ~ is not a locally m- c o n v e x a l g e b r a . For this,

it is sufficient to prove that the inverse is not conti-

nuous in [~, because a locally m-convex algebra is an

algebra with a continuous inverse. We can find a sequence

of real numbers an> O, an+ ~ such that

fn(X) = a n ~A(X) + i,

where
1 1 I+I
A= =

and
1 if x~A
~A(X) : {
0 if x~A,

is divergent in ~ . On the otherhand fn ~ G ( ~ ) and fn-1+ e.

9. A locally m-convex algebra which is not metrizable.

 146

Let~ denote the space of all infinitely differen-

tiable functions with compact support in R n. If G is an

arbitrary compact subset of R n, we denote b Y ~ G the sub-

space o f ~ c o n s i s t i n g of those functions i n ' w i t h compact

support in G. If k = {k~, ..., k n} is an arbitrary set

of n non-negative integers, define the differential

operator D k by
Ikl
a f
Dkf -

- k
a x~ ~ xa ... ~n x
n

where

[k[ = k I + + k n.

The family of semi-norms {p~m) ; m = i, 2, ... } ,

defined by

p~m) (f) = sup {IDkf(x)[ ; xed, O~[kl~ m } ,

generates a topology u G under which ~ G is a Fr~chet space.

The so called Schwartz topology u on ~ is the

inductive limit topology o n ' w i t h respect to the family

{r •
; r = 1,2, ... }
G
of subspaces of ~ where

Sr = {x = (Xs) eR n ; lxsl~r, s = 1,2, .,n }

(o6, u) is a locally m-convex algebra with the point-

wise multiplication, but is not metrizable.

IO. A Fr~chet algebra which has closed ideals but not

closed maximal ideals.

Consider the Fr~chet algebra A = ~ as defined in

=I= 8. It has closed ideals; for example

 147

is a closed ideal in [~ . But ~ does not have closed

maximal ideals in view of the Gelfand-Mazur theorem for

Fr6chet algebras, since if M is a maximal ideal, then A/M

is a field, and if I is a closed ideal, then A/I is a

Frechet algebra.

ii. A Frechet algebra which does not have the Wiener

property.

A locally convex algebra A is said to have the Wiener

property if for every fem(A), f(x) ~ O, we have xeG(A).

Let
i+
s = n s
p>1

where

s = { x = (Xn) ; I I Xnl p <~}"

n=O

It is a Frechet algebra with convolution as multiplication

and with the semi-norms

Qp(x) = ( n~OI Xnl P) I/p

I+
But s does not have the Wiener property. (Cf. B [41]).

12. A semi-simple locally m-convex Frechet algebra which

is a projective limit of Banach algebras which are not

semi-simple.

Consider the commutative and semi-simple Banach

algebra A0 = s with identify e, norm

llxll = ~ Ixnl
O
 148

and convolution as multiplication. Define

llXllo = Zo 2n {Xnl.
Then {Ixll is a continuous submultiplicative norm on A 0
0

such that the completion of A 0 in the norm llxl] is not

0
a semi-simple algebra. Now we define A as an algebra of

sequences x = ($n) , ~nsA0 , such that

Pi(X) = sup {{[~:{I , , , '

(i = 1,2, 9 ).

Then A is a semi-simple locally m-convex Frechet algebra

with the semi-norms {pi} and pointwise multiplication.

Let {qi} be an increasing system of continuous submulti-

plicative semi-norms on A equivalent to {pi} There are

constants C I and C 2 and integers i and m such that

PZ (x) < C I qi(x)_< C 2 pro(x)

for all x in A. Since p1(x) is a norm on A, qi(x) is also

a norm on A. Let A i be the completion of A in this norm.

Let x 0 be a non-zero element of A 0 such that

n 1In
lira l{Xo{{ = o.
0
Let

x = (0, , xo, Xo , )cA ,

where the zeros appear in i places. We have

qi(xn) < C2 Pm(X n) _ Cz l]x~l{

- CI CI 0

So,
i/n C2 n 1/n
lira lqi(mn) l = lira I ~ {{xo{{o{ = 0
 149

and x # O. Hence A. is not a semi-simple algebra.

1

13. ~ - s i n g u l a r elements of a locally m - c o n v e x Frechet

algebra, w h i c h are not topological divisors of zero.

Consider the locally m-convex Frechet algebra ~ of

power s e r i e s

x = X x n tn
0
with semi-norms

n
Pn(X) = Z Ixi]
i=x
and with Cauchy m u l t i p l i c a t i o n of the series, that is,

xy =
z = E zn t n

where
k
zk = X xi Yk-i "
i=0

Writing

x(t) = Zx tn
n

Xo(t ) = t is a generator, and

m(A) = m # ( A )

since A is a Q-algebra. But x 0- le is invertible in~ for

any I # 0 and ~(x ) = {0} . If f~m(A), then f(x0) = 0 and

so there is only one element in m #(A) given by f(x)=x(O)

for any xe~. Thus there is only one m a x i m a l ideal in ~ .

Let us denote it by M. Since

Gfl) = ~ - M,

the only topological divisors of zero can be in M. But

M = x 0 ~ and so if there are topological divisors of zero

in M, then x0 is also one such divisor. But x0 is not a

topological divisor of zero i n ~ , since for any ye~ ,

Pn(Y x0) = Pn-*(Y) , n ~ i.

Thus,~K has no topological divisors of zero. On the other

hand,
M = rad<A)c tad(B)

for any m-convex superalgebra B of A, and so any element

xeM must be J~-singular.

14. A locally m-convex Frechet algebra which has neither

topological divisors of zero nor N - s i n g u l a r elements.

The locally m-convex Frechet algebra e of all entire

functions of one complex v a r i a ~ e with pointwise multipli-

cation and semi-norms

Pn(X) = sup {x(t) ; ]t[~ n }

fulfills the requirement.

15. An m-barrelled algebra which is not barrelled.

Consider the algebra A of all polynomials X without

constant term, equipped with the strongest locally m-

convex topology. Let ~ be a rational number with O<e<l,

and let V(~) be the circled and convex envelope of

{a n X n ," n is positive integer} . Then the family of all

such V(e) is a neighbourhood basis at 0 for the strongest

locally m-convex topology on A. It is clear that A is an

m-barrelled algebra. To show that it is not barrelled,

let S be the circled and convex envelope of {2-n2x n ;

n = 1,2, ...}

S is absorbing and so S is a barrel. If 8>O,

82-n2x n ~ -S ;
 151

for, if
I/n
O<a <2-n (B- i)

Then

ILB2 - n 2 Xn + V{~.)l flS = 0 9

For any a> O,

lim 2-n 2 - n = O, n+= .

So, for large n,

n -n 2
>2

that is,
n
a = Ba, n 2 -n2 , Ba,n >I 9

Hence, if V(a) c g , for suitably large n we have

n Xn 2-n2xnev(~) c g
= ~,n

which is a contradiction. Hence S is not a neighbourhood

of O.

16. A countably m - b a r r e l l e d algebra which is not m-

barrelled.

The locally convex space C(W), as defined in Chapter

3, ~f= 41 (ii) is infact a locally m-convex algebra with

the usual pointwise multiplication. It is a countably m-

barrelled algebra becuase it is countably barrelled. But

it is not m - b a r r e l l e d because the set

B = {fcC(W) ; If{x) l! I, x~W}

is closed, circled, m-convex and absorbing but not a

neighbourhood of O.

17. A complete p.i.b, algebra which is not a P-algebra.

The locally convex algebra C(W), as defined in

 152

Chapter 3, ~ 41(ii) is complete because W is locally

compact, p.i.b, because W is pseudo-compact, but not a P-

algebra because W is not compact.

18. A metrizable p.i.b, algebra which is neighter a P-

algebra nor an m-bornological algebra.

Let X be a non-compact completely regular space. Let

E be the algebra of all continuous real-valued functions

on X with compact supports, equipped with the compact-open

topology u. Let v be the topology defined by the uniform

norm. Then

B = {fee ; If(x) l_< 1 for all xsX }

is bounded and idempotent for u and v, and also every

bounded and idempotent set for either topology is contained

in B. Hence the m-bounded sets for u and v coincide. By

the hypothesis on X, v is strictly stronger than u so

that (E,u) is not m-bornological, because (E,u) is m-

bornological if and only if no strictly stronger locally

m-convex topology on E can have the same m-bounded sets.

Clearly (E,u) is p.i.b.. But it is not a P-algebra, for,

given any compact set K in X, there exists feE such that

f = 0 on K, but f(x)> 1 for some xeX ; hence the sequence

{x n} does not converge to O. Now let

co

X= O K
1 n '

where {Kn} i s a fundamental sequence of compact subsets

of X. Then E is metrizable.

19. The Gelfand map which is continuous for a locally

convex algebra which is not m-barrelled.

 153

The algebra C(~) of complex-valued continuous func-

tions on ~, equipped with compact-open topology, is a

locally m-convex algebra. The subspace Cb(R ) consisting of

bounded functions in C(R) is not m-barrelled, since the

set

B = {feC b(R) ; If( x )I~ I, xe~}

is an m-barrel but not a neighbourhood of O. But the

Gelfand map G is continuous.

20. A GB*-algebra which is not a locally m-convex algebra.

We have seen in ~/= 8 that A = ~is not a locally

m-convex algebra. We now show that it is a GB*-algebra.

Clearly it is a complete (and hence pseudo-complete)

locally convex *-algebra, the involution being complex

conjugation. A 0 is simply the subalgebra ~ of essentially

bounded measurable functions, and ~* has the greatest

member

B 0 = {f~A ; ess sup If(x) l~ i}

For any f in A,
_1 2 _1
(e + f* f) = (e + I fl ) eA0

so t h a t A is symmetric. Thus A = ]~ i s a GB*-algebra.

21. A GB*-algebra on w h i c h t h e r e a r e no n o n - t r i v i a l

multiplicative linear functionals.

Consider the GB* - a l g e b r a A = i~as defined i n =/~ 8.

L e t r be a n o n - z e r o multiplicative linear functional on A.

Then r is also multiplicative restricted to the subalgebra

C [O,E . Thus, there is some p o i n t x0~[O, E such that

r = f(x0) for every f in C [O,~-r can not annihilate

 154

C [O,l],since C [O,Q contains the identity of A. But

there is certainly some c o n t i n u o u s extended complex-valued

function heA such that

h(xo) = =, h(x)>O for all x.

But then there is some fn in C[O,~ , n = 1,2,

such that fn(X) = 1 and

n fn(X) ~ h(x) for all x.

Thus

r n = 1,2,

which is a contradiction.

22. A pseudo-complete locally convex algebra which is

not sequentially complete.

Let A be the algebra of all complex polynomials P

and let A be given the topology u of uniform convergence

on the compact subsets of ~+ . Then A0 consists of the

constant functions and B, has a greatest member, namely

the set of all constant functions not exceeding unity in

absolute value. Thus A is pseudo-complete, by Proposition

i. Now, let the sequence {Pn} in A be defined by

n (_l)r+1 2r+i
Pn(X) = z x
r=O (2r + i) !

Then Pn(X) converges to sin(x) uniformly on compact subsets

of ~+. Let

B : {P~A ; I P( x) l~ eX
Then B is u-bounded. Also

e-X[Pn(X) - sin(x) l.O

~*
uniformly on ; it is thus clear that {Pn } is Cauchy in

the sense of Mackey but does not converge to an element of

 155

A. Thus A is not Mackey-complete which implies that A is

not sequentially complete.

23. An A-convex algebra which is not a locally m-convex

algebra.

(i) Let Cb(~) be as in =~= 19. We d e n o t e by C:(~)

the s e t of s t r i c t l y positive continuous real-valued func-
tions on E which vanish at infinity. The family of semi-
norms {p~ ; ~EC~(E)} generates a locally convex t o p o l o g y
u on Cb(~ ) where
pc(f) = sup {If(x) r l ; xE~} , fECb(~ )

The space ( C b ( ~ ) , u) is A-convex and each p~ fails to be

submultiplicative. Suppose (Cb(~), u) is locally m-convex

and let Q be a set of submultiplicative semi-norms which

define u. We may assume that

max (ql , . qn)e Q and %qi EQ

whenever

ql, qnEQ and ~_>i.

+ [1)
Thus, for ~ECo( j there exists qeq and ~ C + ( t l ) such

that

(*) V(~) =V(q) =~f(r

where

V($) = {f ; pc(f) _< I}

Since V ( ~ ) % V ( ~ ) , ~<~ (pointwise).

Let Og~ with

0<@< rain (I, M(~)),

where N($) is the maximum of I$I . Then, for some xE~, it

follows that

(x) = e>~(x) and sn~ ~(x)

 156

where n is a positive integer. Consider the function f(y)

defined by

y- x + 1 if x-i < y<x

f(y) = - y + x + 1 if x<y < x+l

(y)
f 0 ~ (y)otherwise.
Then f is well-defined, since ~sC$(~) and f e C b ( R ) . But

P~(f) = 1 and

pr fn(x ) r = @-n ~(x)> 1

which contradicts (*) since q is sub-multiplicative. Hence

(Cb(~), u) is not locally m-convex.

(ii) Consider the algebra C[O,~ of all continuous

complex-valued functions on the closed interval [O,~ . A

norm [I'II is defined on this algebra by

I lfll = sup { I f ( x ) ,(x) I ; x~ [o,13}

where
x if O<x<
r = {
l-x if ~<x<l .

Then ( c [ o , ~ , II-II) is a normed space which is A-

convex but not locally m-convex.

24. A p-normed (locally bounded) algebra which is not a

normed algebra.

The algebra ~P, O<p<l, of all complex two-sided

sequences x = (Xn)~ with the p-norm

llxll = ~ IXnl p <

n ~ I~
 157

and convolution as multiplication is a p-normed algebra

but not a normed algebra.

25. A locally m-semi-convex algebra which is not a

locally m-convex algebra.

Consider the algebra s , O~q<l , of all complex

two-sided sequences x = (Xn)~ with convolution as multi-

plication.

Define

[]XIIp = z
n=_~ Ix n I p

for every p satisfying q<p<l Then ~q+ is a locally m-

semi-convex algebra for the family of p-semi-norms

{[I'[Ip} , q<p<l, which is neither locally m-convex nor

locally bounded.

OPEN PROBLEMS:

I. (Michael B [ 2 ~ ): If A is a complete locally m-convex

a lgebra~is R(A) = n{closed, regular, maximal, right

ideals in A} closed?

2. (Michael B [23]) : Is every multiplicative linear

functional on a commutative Frechet algebra conti-

nuous?
 BIBLIOGRAPHY

BOOKS

[i] BANACH, S. : Theorie des operations linearies,

Warsaw,(1932).

[2] BIRKHOFF, G. : Lattice theory, Amer. Math. Soc.

Colloq. Publ. Vol. 25 (Third Printing), (1973).

Is] BOURBAKI, N. : Elements de mathematique, Livre V,

Espaces Vectoriels topologiques, Hermann, Paris,
(1953, 1955).

[4] DAY, M.M. : Normed linear spaces, Springer-Verlag,

Berlin, (1958).

[s] DUNFORD, N. and SCHWARTZ, J. : Linear operators,

Vol. I, Interscience, New York, (1958).

[6] EDWARDS, R.E. : Functional analysis, Theory and

applications, Holt, Rinehart and Winston,
New York, (1965).

[7] GFLBAUM, B.R. and OLMSTED, J.M.H. : Counterexamp~es

in analysis, Holden-Day, Inc. San Francisco,
(1964).

[8] GILLMAN, L. and JERISON, M. : Rings of continuous

functions, Van Nostran, Princeton, (1960).

[9] GROTHENDIECK, A.: Products tensoriels topologiques

et spaces nucleaires, Mem. Amer. Math. Soc. 16
(1955).

[9(a)]GROTHENDIECK, A.: Topological vector spaces (English

edition), Gordon and Breach, Sc. Publishers,
New York - London - Paris, (1973).

[lO] GUICHARDET, A.: Special topics in topological

algebras, Gordon and Breach, Sc. Publishers,
New York-London-Paris, (1968).
 159

[11] HALMOS, P.R.: Introduction to Hilbert space and the

Theory of Spectral Multiplicity, Chelsea, New
York, (1951).

[12] HARDY, G.H.; LITTLEWOOD, J.E. and POLYA, G: Inequa-

lities, Camb. Univ. Press, (1934).

E13] HORVATH,J.: Topological vector spaces and distri-

butions, Vol. I, Addison-Wesley, Reading,
Mass., (1966).

[14] HUSAIN, T." The open mapping and closed graph

theorems in topological vector spaces, Oxford
Mathematical Monographs, Clarendon Press,
Oxford, (1965).

[is] HUSAIN, T. and KHALEELULLA , S.M.: Barrelledness

in topological and ordered vector spaces,
Lecture Notes in Mathematics, Springer-Verlag,
Berlin-Heidelberg - New York, (1978).

[16] JAMESON, G.: Ordered linear spaces, Lecture Notes

in Mathematics, No. 141, Springer-Verlag,
Berlin-Heidelberg, New York, (1970).

[17"] KELLEY, J.L., NAMIOKA, I and Co-authors.: Linear

topological spaces, Van Nostrand, Princeto n ,
(1963).

[18] KELLEY, J.L.: General Topology; Van Nostrand,

(1955).

[19] KHALEELULLA, S.M.: Semi-inner product spaces and

algebras, Shafco Publ., Hassan-5732Ol, India,
(1981).

[2o] KOTHE, G.: T o p o l o g i c a l v e c t o r spaces I ( E n g l i s h

edition), S p r i n g e r - V e r l a g , New York, (1969).

[.21] LOOMIS, L . H . : An i n t r o d u c t i o n to a b s t r a c t harmonic

analysis, Ven Nostrand, (1953).

[22] MARTI, J.: Introduction to the theory of bases,

Springer Tracts in Natural Philosophy, Vol. 18,
New York,(1969).
 160

[23] MICHAEL, E.A.: Locally multiplicatively convex

topological algebras, Mem. Amer. Math. Soc.
No. ii (Third printing), (1971).

[24] NAIMARK, M.A.: Normed rings, Noordhoff, (1959).

[25] NAMIOKA, I.: Partially ordered linear topological

spaces, Mem Amer. Math. Soc. No. 24, (1957).

[25(a)]NATASON, I.P.: Constructive Theory of functions,

(English edition), New York, (1964-65).

[26] PERESSINI, A.L.: Ordered topological vector spaces,

Harper and Row Publ.,New York, (1967).

[2q RICKART, C.E.: Banach algebras, Van Nostrand,

Princeton, (1960).

[28] RIESZ, F. and Sz.-NAGY.: Functional analysis

(English edition), Blackie, London, (1956).

[29] ROBERTSON, A.P. and ROBERTSON, W.J.: Topological

vector spaces, Camb. Tracts in Math., Camb.
Univ. Press, England, (Second edition), (1973).

[3o] RUDIN, W.: Functional analysis, McGraw-Hill, Inc.

(1973).

[31] SCHAEFER, H.S.: Topological vector spaces, Springer-

Verlag, New York-Heidelberg-Berlin, (Third
Printing), (1971).

[32] SCHWARTZ, L.: Th~orie des distributions, Her~ann,

Paris, (1950, 1951).

[33] SINGER, I.: Bases in Banach spaces, Vol. I,

Springer-Verlag, New York-Heidelberg-Berlin,
(1970).

[34] TAYLOR, A.E.: Introduction to functional analysis,

Wiley, New York, (1958).

[35] TREVES, F.: Topological vector spaces, distributions

and Kernels, Acad. Press, New York, (1967).
 161

[36] WAELBROECK, L.: Topological vector spaces and

algebras, Lecture Notes in Mathematics, No. 230,
Springer-Verlag, New York-Heidelberg-Berlin,
(1971).

[373 WILANSKY, A.: Functional analysis, Blaisdell,

New York, (1964).

bs] WILANSKY, A.: Topics in functional analysis, Lecture

Notes in Mathematics, No. 45, Springer-Verlag,
New York-Heidelberg-Berlin,(1967).

[39] WONG, Y.-C. and NG, K.-F.: Partially ordered

topological vector spaces, Oxford Mathematical
Monographs, Clarendon Press, Oxford, (1973).

Bo] YOSIDA,K.:Functional analysis, Springer-~erlag,

Berlin-Gottingen-Heidelberg, (1965).

[41] ZELAZKO,W.: Selected topics in topological algebras,

Lecture Notes Series No. 31, Matematisk Insti-
tute, Aarhus Universitet, (1971).

[42] IVES, R.S. : Semicon~exity and locally bounded

spaces, Ph.D. Thesis, University of Washington,
(1957).

PAPERS

[i] ALLAN, G.R.: A spectral theory for locally convex

algebras, Proc. London Math. Soc.(3), 15, (1965),
399-421.

[2] ALLAN, G.R.: On a class of locally convex algebras,

Froc. London Math. Soc. (3), 17, (1967), 91-114.

[3] ARENS, R.F.: The space ~ and convex topological

rings, Bull, Amer. Math. Soc. 52, (1946),
931-935.

[4] ARENS, R.F.: Linear topological division algebras,

Bull. Amer. Math. Soc. 53, (1947), 623-630.
 162

Is] ARSOVE, M.G.: Similar bases and isomorphisms in

Frechet spaces, Math. Analan 135, (1958),
283-293.

[6] ARSOVE, M.G. and EDWARDS, R.E.: Generalized bases

in topological linear spaces, Studia Math. 19,
(1960), 95-113.

[7] BOURGIN, D.G.: Linear topological spaces, Amer. J.

Math. 65, (1943), 637-659.

[8] CHILANA, A.K.: Some special operators and new

classes of locally convex spaces, Proc. Camb.
Phil. Soc. 71, (1972), 475-489.

[9] CHILANA, A.K. and KAUSHIK, V.: Examples in linear

topological spaces, J. London Math. Soc. (2),
8, (1974), 231-238.

[10] COCHRAN, A.C., KEOWN, R. and WILLI~4S, C.R.: On a

class of topological algebras, Pacific J.
Math., Vol. 34, No. I, (1970), 17-25.

[11] COLLINS, H.S.: Completeness and compactness in

linear topological spaces, Trans. Amer. Math.
Soc. 79, (1955), 256-280.

[12] DAY, M.M.: The spaces G with O<p<l, Bull. Amer.

Math. Soc. 46, (1940), 816-823.

[13] DEAN, D.W.: Schauder decompositions in (m), Proc.

Amer. Math. Soc. 18, (1967), 619-623.

[14] De WILDE, M. and HOUET, C. : On increasing

sequences of absolutely convex sets in locally
convex spaces, Math. Ann. 192 (1971), 257-261.

[15] DIEUDONNE, J.: Sur les espaces de KDthe, J. Analyse

Math., i, (1951), 81-115.

[16] DIEUDONNE, J.: Sur les espaces de Montel metrizable,

C.R. Acad. Sci. Paris. 238 (1954), 194-195.
 163

[17] DIEUDONNE, J.: On biorthogonal systems, Michigan

Math. J., 2, (1954), 7-20.

[18] DIEUDONNE, J. and SCHWARTZ, L.: La dualite dans les

espaces (F) et (LF), Ann. Inst. Fourier,
Grenoble, I, (1950), 61-101.

[19] DWORETZKY, A. and ROGERS, C.A.: Absolute and uncon-

ditional convergence in normed linear spaces,
Proc. Nat. Acad. Sci. U.S.A. 36 (1950), 192-197.

112o] ELLIS, A.J.: The duality of partially ordered normed

linear spaces, J. London Math. Soc. 39 (1964),
730-744.

F2q ENFLO, P.: A counterexample to the approximation

problem in Banach spaces, Acta Math. 130, (1973),
309-317.

[22] GELBAUM, B.R.: Expansions in Banach spaces, Duke

Math. J. 17, (1950), 187-196.

[23] GELFAND, I.M. and NAIMARK, M.A.: Normed rings with

their involutions and their representations,
Izvestiya Akad. Nauk S.S.S.R. Ser. Mat., 12,
(1948), 445-480.

[24] GILES, J.L.: Classes of semi-inner product spaces,

Trans. Amer. Math. Soc. 129, (1967), 436-446.

[283 GROTHENDIECK, A.: Sur les espaces (F) and (DF),

Summa Bras. Math. 3, (1954), 57-123.

[263 HUSAIN, T . : S - s p a c e s and t h e open mapping t h e o r e m ,

P a c i f i c J . Math. 12, (1962), 253-271.

[2 7] HUSAIN, T . : Two new c l a s s e s o f l o c a l l y convex

s p a c e s , Math. A n n . , 166, (1966), 289-299.

[28] HUSAIN, T. and KHALEELULLA, S.M." Countably order-

quasi-barrelled vector lattices and spaces,
Mathematicae Japonica, Vol. 20, (1975), 3-15.
 164

[29] HUSAIN, T. and KHALEELULLA, S.M.: Order-quasi-ultra-

barrelled vector lattices and spaces, Periodica
Math. Hungarica, Vol. 6(4), (1975), 363-371.

[3o] HUSAIN, T. and KHALEELULLA, S.M.: On countably, ~-

and sequentially barrelled spaces, Canad. Math.
Bull. Vol. 18(3), (1975), 431-432.

[31] HYERS, D.H.: Locally bounded linear topological

spaces, Rev. Ci. (lima), 41, (1939), 555-574.

[32] HYERS, D.H.: Linear topological spaces, Bull. Amer.

Math. Soc. 51, (1945), 1-24.

[33] IYAHEN, S.O.: Some remarks on countably barrelled

and countably quasi-barrelled spaces, Proc.
Edinburgh Math. Soc. (2), 15, (1966/67),
295-296.

[34] IYAHEN, S.O.: Semiconvex spaces, Glasgow Math. J.9~,

(1968), 111-118.

[35] IYAHEN, S.O.: Semiconvex spaces II, Glasgow Math. J.

iO, (1968), 103-105.

[36] IYAHEN, S.O.: On certain classes of linear topolo-

gical spaces, Proc. London Math. Soc. 18,
(1968), 285-307.

[37] IYAHEN, S.O.: On certain classes of linear topologi-

cal spaces II, J. London Math. Soc. (2), 3,
(1971), 609-617.

[38] JAMES, R.C.: Bases and reflexivity of Banach spaces,

Ann. of Math. (2), 52, (1950), 518-527.

[39] JAMESON, G.J.O.: Topological M-spaces, Math. Z.,

103, (1968), 139-150.

[4o] JONES, O.T. and RETHERFORD, J.R.: On similar bases

in barrelled spaces, Proc. Amer. Math. Soc. 18,
(1967), 677-680.
 165

[41] KAPLANSKY, I.: Topological rings, Amer. J. Math.

69, (1947), 153-183.

[42] KAPLANSKY, I.: Topological rings, Bull, Amer. Math.

Soc. 45, (1948), 809-826.

[43] KARLIN, S.: Unconditional convergence in Banach

spaces, Bull. Amer. Math. Soc. 54, (1948),
148-152.

[44] KARLIN, S.: Bases in Banach spaces, Duke Math. J.

15, (1948), 971-985.

[45] KHALEELULLA, S.M.: Countably O.Q.U. vector lattices

and spaces, Tamkang J. Math. Vol. 6, No.I,
(1975), 43-51.

[46] KHALEELULLA, S.M.: Countably m-barrelled algebras,

Tamkang J. Math. Vol. 6, No. 2, (1975), 185-190.

[47] KHALEELULLA, S.M.: Husain spaces, Tamkang J. Math.

Vol. 6, No. 2, (1975), 115-119.

[48] KLEE, V . L . : Convex sets in linear spaces III, Duke

Math. J. 20 ( 1 9 5 3 ) , 105-111.

[49] KOMURA, Y . :On l i n e a r t o p o l o g i c a l spaces, Kumamoto

J. Sc. Series A, 5, (1962), 148-157.

[5o] KOMURA, Y.: Some examples on linear topological

spaces, Math. Ann. 153, (1964), 150-162.

[51] KREIN, M. and MILMAN, I.: On extreme points of

regular convex sets, Stud. Math. 9, (1940),
133-138.

[52] KUCZMA, M.E.: On a problem of E. Michael concerning

topological divisors of zero, Coil. Math. 19,
(1968), 295-299.

[53] LANDSBERG, M.: Pseudonormen in der theorie der

linearen topologischen raume, Math. Nachr. 14,
(1955), 29-38.
 166

Es43 LANDSBERG, M.: Lineare topologische raume die nicht

lokelkonvex sind, Math. Z., 65, (1956), 104-112.

[ss] LAVERELL, W.D.: Between barrelled and W-barrelled

for C(X), Indiana Univ. Math. J., Vol. 22, No.i,
(1972), 25-31.

Eso3 LEVIN, M. and SAXON, S.: A note on the inheritance

of properties of locally convex spaces by sub-
spaces of countable codimension, Proc. Amer.
Math. Soc. 29, No. i, (1971), 97-102.

Es7] LIVINGTON, A.E.: The space H p, O<p<l, is not norm-

able, Pacific J. Math. 3, (1953), 613-616.

E58] LUMER, G.: Semi-inner product spaces, Trans. Amer.

Math. Soc. iO0, (1961), 29-43.

[593 MACKEY, G.W.: On infinite-dimensional linear spaces,

Trans. Amer. Math. Soc. 57, (1945), 155-207.

~60] MACKEY, G.W.: On c o n v e x t o p o l o g i c a l l i n e a r s p a c e s ,

T r a n s . Amer. Math. Soc. 60, ( 1 9 4 6 ) , 519-537.

E61"] MAHOWALD, M. and GOULD, G.: Quasibarrelled locally

convex spaces, Proc. Amer. Math. Soc. 2,
(1960), 811-816.

E62~ MALLIAVIN, P.: Impossibilite dela synthese spectrale

sur groups abeliens noncompact, Faculte des
Sciences de Paris, Seminaire d' analyse (P.
Lelong), 1958-1959.

MALLIOS, A.: On the spectra of topological algebras,

J. of Functional analysis, 3, (1969), 301-309.

E64 MARKUSHEVICH, A.: Sur les bases ( an sens large )

dans les espaces linearies, Doklady Akad, Nauk,
S.S.S.R., 41, (1943), 227-229.

E65] McARTHER, C.W. and RETHERFORD, J.R.: Uniform and

equicontinuous Schauder bases of subspaces,
Canad. J. Math., 17, (1965), 207-212.
 167

[.663 MORRIS, P.D. and WULBERT, D.E.: Functional repre-

sentation of topological algebras, Pacific J.
Math. Vol. 22, No. 2, (1967), 323-337.

[673 NACHBIN, L.: Topological vector spaces of continuous

functions, Proc. Nat. Acad. Sci. Wash., 40,
(1954), 471-474.

[68] NAKANO, H.: Linear topologies on semi-ordered

linear spaces, J. Foc. Sci. Hokkaido Univ. 12,
(1953), 87-104.

[69] NATH, B.: Topologies on generalized semi-inner

product spaces, Compositio Math., Vol. 23,
Fasc. 3, (1971), 309-316.

E7o] NG, K.F.: Solid sets in ordered topological vector

spaces, Proc. London Math. Soc. (3), 22, (1971),
106-120.

12713 PERSSON, A.: A remark on the closed graph theorem

in locally convex spaces, Math. Scand., 19,
(1966), 54-58.

1723 PRUGOVECKI, E.: Topologies on generalized inner

product spaces, Canad, J. Math., 21, (1969),
158-169.

[733 PTAK, V.: On complete topological linear spaces,

Ceh. Math. Zur., 3(78), ~1953), 301-364.

[743 ROBERTSON, W.: Completion of topological vector

spaces, Proc. London Math. Soc. 8, (1958),
242-257.

E7s3 ROLEWICZ, R.: Example of semisimple m-convex B0 -

algebra which is not a projective limit of
semi-simple B-algebras, Bull. Acad. Polon. Sci.
Ser. Math. Astronem Phys., Ii, (1963), 459-462.

[76] RUTHERFORD, J.R.: Two counterexamples to a con-

jecture of S. Karlin, Bull. Acad. Polon. Sci.,
16, (1968), 293-295.
 168

[773 SAXON, S.A.: Nuclear and product spaces, Baire-like

spaces and the strongest locally convex topology,
Math. Ann., 197, (1972), 87-106.

[78] SAXON, S. and LEVIN, M.: Every countable-codimen-

sional subspace of a barrelled space is barrel-
led, Proc. Amer. Math. Soc., 29, No.l, (1971),
91-96.

[79] SCHAEFER, H.H.: On the completeness of topological

vector lattices, Mich. Math. J., 7, (1960),
303-309.

[8o] SCHAUDER, J.: Zur theorie Abbildungun in Funktional-

raumen, Math. Z., 26, (1927), 47-65.

[813 SCHAUDER, J.: Eine Eigenschaft des Haarschen ortho-

gonal system Math. Z., 28, (1928), 317-320.

[82] SCHMETS, J.: Independence des propri~t~s de

tonnelage et d~valuabilit~ affaiblis, Bull.
Soc. Roy. Sc., Liege, 42 e annee, n ~ 3-4, (1973),
104-108.

[83] SEGAR, I.E.: The group algebra of a locally compact

group. Trans. Amer. Math. Soc. 61, (1947),
69-105.

[84] SEGAR, I.E.: Irreducible representation of operator

algebras, Bull. Amer. Math. Soc. 53, (1947),
73-88.

[85] SHIROTA, T.: On locally convex vector spaces of

continuous functions, Proc. Jap. Acad. 30,
(1954), 294-299.

[86] SIMONS, S.: Boundedness in linear topological spaces,

Trans. Amer. Math. Soc.,Vol. 113, No. i, (1964),
169-180.

['87] SINGER, I.: On the basis problem in topological

linear spaces, Rev. Roumeine Math. Pures et
Appl. iO, (1965), 453-457..
 169

[88] SNIPES, R.F.: C-sequential and S-bornological

topological vector spaces, Math. Ann. 202,
(1973), 273-283.

[89] TWEDDLE, I.: Some remarks on o-barrelled spaces,

(unpublished).

[9o] WARNER, S.i Inductive limits of normed algebras.

Trans. Amer. Math. Soc. 82 (1956), 190-216.

[91] WARNER, S.: The topology of compact convergence on

continuous function spaces, Duke Math. J., 25,
(1958), 265-282.

[9 2] WEBB, J.H.: Sequential convergence in locally convex

spaces, proc. Camb. Phil. Soc. 64, (1968),
341-364.

[93] WESTON, J.D.: The principle of equicontinuity for

topological vector spaces, Proc. Durham Phil.
Soc., 13, (1957), 1-5.

[94] WONG, Y.-C.: Order-infrabarrelled Riesz spaces,

Math. Ann., 183, (1969), 17-32.

ZELAZKO, W.: On the locally bounded and m-convex

topological algebras, Stud. Math., 19, (1960),
333-356.
 INDEX

Absolutely convergent basis 112

Absolutely convergent series iii

Absolute value 7

Absorbing 3

A-convex algebra 141

~-hyperbarrelled space 67
~-quasihyperbarrelled space 67
Algebra 1

Algebraic dual 20

Almost Archimedean 6

Almost open 18

Approximate identity 138

Approximate order-unit 6

Approximate order-unit normed space 81

Archimedean 6

A*-algebra 138

Auxiliary norm 138

Baire space 27

Baire-like space 27

Balanced 3

Banach algebra 138

Banach lattice 79

Banach space 18

Banach *-algebra 138

Band 8

Barrel 28

Barrelled space 28

Base for positive cone

 171

Basis ill

Base norm 81

Base normed space 81

Base semi-norm 81

B-complete 18

b-cone 78
Besselian basis 113

Biorthogonal system ll2

Bipolar 21

Bornivorous 28

Bornivorous suprabarrel 66

Bornivorous ultrabarrel 66

Bornivorous ultrabarrel of type (~) 66

Bornological space 29

Bounded multiplier convergent iii

Bounded set i0

Boundedly complete basis 112

Boundedly order-complete 79

(b)-Schauder basis 114

B*-algebra 138

Canonical bilinear functional 2O

Cartesian product space 103

Circled 3

Circle operation 137

Closed neighbourhood condition 68

Compatible 9

Complete I0

Complex vector space 1

Cone 5

Convex 3
 172

Coefficient functionals iii

C.O.Q. vector lattice 82

Countably barrelled space 33

Countably m-barrelled algebra 140

Countably O.Q.U. vector lattice 84

Countably quasibarrelled space 33

Countably quasi-ultrabarrelled space 66

Countably ultrabarrelled space 66

C-sequential space 30

Decomposable 6

Decomposition property 7

(DF) space 33

Dimension 2

Direct sum 2

Directed set 5

Distinguished space 32

Dual 20

E-complete biorthogonal system 112

(e)-Schauder basis 114

Exhausting 6

Extended Markushevich basis 114

Extremal point 7

Filter condition 68

Fr~chet algebra 138

Fr~chet space 18

Full 6

GB*-algebra 141

Gelfand map 139

Generalized basis 114
 173

Generalized inner product space 18

Generalized semi-inner product space 18

G e n e r a t i n g cone 6

Hamel b a s i s 2

Hilbertian basis 113

Hilbert space 18

H-space 33

Hyperbarrelled space 67

Hyperbornological space 67

Hyper-quasibarrelled space 67

Idempotent 137

Identity 138

Inductive limit 28

Infimum 7

Inner product 4

Inner produce space 18

I s o m o r p h i s m Theorem f o r b a s e s 115

k-barrelled space 33

k-norm 3

k-quasibarrelled space 33

k-quasiultrabarrelled space 67

k-semi-norm 3

k-ultrabarrelled space 67

Lattice homomorphism 8

Lattice ideal 8

Lattice operations 8, 82

(LB)-space 28

Left ideal 137

Left topological divisor of zero 139

 174

(LF)-space 28

Linear combination 2

Linear functional 3

Linear map 3

Linearly independent 2

Locally bounded space ii

Locally convex algebra 138

Locally convex lattice 79

Locally convex space 12

Locally convex *-algebra 140

Locally m-convex algebra 138

Locally m-semiconvex algebra 142

Locally semi-convex space ii

Locally solid topology 79

Locally topological space 55

L-W space 65

Mackey space 31

Majorized 5

Markushevich basis 114

Maximal biorthogonal system 113

Maximal ideal 137

m-barrelled space 140

m-bounded 139

m-convex 137

Metrizable i0

Minorized 5

Modular ideal 137

Monotone basis 112

Monte1 s p a c e 32

~[-singular 139
 175

M-space 82

Negative part 7

Norm 3

Normal basis 112

Normal cone 77

Normal topology 47
137
Normed algebra

Normed space 18
79
Normed vector lattice

Nowhere dense 27

N-S space 65

u-linearly independent 122

Open decomposition 78
84
O.Q.U. vector lattice
82
Order-bornivorous
83
Order-bornivorous ultrabarrel
84
Order-bornivorous ultrabarrel of type (~)

Order-bound topology 81

Order-bounded 5

Order-complete 7

Order-continuous 8

Order-convergence 8
83
Order-(DF) vector lattice

Ordered locally convex space 77

Ordered topological vector space 77

Ordered vector space 5

Order-interval 5

Order-quasibarrelled space 82

Order-separable 8

Order-unit 6
 176

Order-unit normed s p a c e 81

P-algebra 139
p.i.b, algebra 139

p-normed a l g e b r a 141

Polar 21

Positive cone 5

Positive linear functional 7

Positive linear map 7

Positive part 7

Precompact i0

Pre-Hilbert space 18

Product 2

Property ( C ) 31

Property (S) 31

Pseudo-complete 140

Pseudo-M-space 82

Q-algebra 139

Q-space 29

Quasi-barrelled space 28

Quasi-complete i0

Quasi-hyperbarrelled space 67

Quasi-M-barrelled space 31

Quasi-norm 4

Quasi-regular 137

Quasi-semi-norm 3

Quasi-ultrabarrelled space 66

Q u o t i e n t map 103

Quotient space 2,103

103
Quotient topology
Radical 137
 177

Real-compact 29

Real v e c t o r space 1

Reflexive 31

Regular ideal 137

Retro-basis 113

Right ideal 137

Right topological divisor o f zero 139

S - b o r n o l o g i c a l space 3O

Schauder b a s i s iii

Schwartz space 32

S e m i - b o r n o l o g i c a l space 30

Semi-complete I0

Semi-convex 3

Semi-inner product 4

S e m i - i n n e r p r o d u c t space 18

Semi-norm 3

Semi-reflexive 31

Semi-simple 137

Sequentially barrelled space 33

Sequentially complete i0

Sequentially order-continuous 8

o-barrelled space 33

u-order complete 7

~-quasibarrelled space 33

Similar bases 115

Singular 139

Solid set 8

Solid ultrabarrel 83

Solid ultrabarrel of type (e) 84

Sublattice 8
 178

Subseries convergent iii

Subspace 1

Subsymmetric basis 113

Suprabarrel 65

Supremum 7

C-topology 19

Strict b-cone 78

Strict inductive limit 28

Strong topology 20

Symmetric Banach *-algebra 138

Symmetric locally convex *-algebra 140

Symmetric basis 113

Topological divisor of zero 139

Topological dual 20

Topological vector lattice 79

Topological vector space 9

t-polar space 28

Topology of compact convergence 19

Topology of precompact convergence 19

Topology of simple convergence 19

Topology of uniform convergence on bounded sets 19

Total 19

Totally bounded iO

Two-sided ideal 137

Ultrabarrel 65

Ultrabarrel of type (~) 66

Ultrabarrelled space 66

Ultrabornological space 66

Unconditional basis 112

Unconditional convergent series ii0

 179

Unit vector basis 120

Unordered Baire-like space 27

Unordered convergent series llO

Vector space 1

Vector lattice 7

W-barrelled space 65

Weak basis lll

Weak basis theorem ll2

Weak topology 20

Weak* topology 20

Wedge 5

Wiener property 147