Generalized B Algebras and Applications Maria Fragoulopoulou Atsushi Inoue Martin Weigt Ioannis Zarakas Online Ebook Texxtbook Full Chapter PDF

Generalized B Algebras and
Applications Maria Fragoulopoulou

Atsushi Inoue Martin Weigt Ioannis
Zarakas
Visit to download the full and correct content document:
https://ebookmeta.com/product/generalized-b-algebras-and-applications-maria-fragou
lopoulou-atsushi-inoue-martin-weigt-ioannis-zarakas/
More products digital (pdf, epub, mobi) instant
download maybe you interests ...
Tomita s Lectures on Observable Algebras in Hilbert

Space Atsushi Inoue
https://ebookmeta.com/product/tomita-s-lectures-on-observable-
algebras-in-hilbert-space-atsushi-inoue/
Algebra and Applications 1: Non-associative Algebras

and Categories 1st Edition Makhlouf
https://ebookmeta.com/product/algebra-and-applications-1-non-
associative-algebras-and-categories-1st-edition-makhlouf/
Algebra and Applications 2 Combinatorial Algebra and

Hopf Algebras 1st Edition Makhlouf Abdenacer
https://ebookmeta.com/product/algebra-and-
applications-2-combinatorial-algebra-and-hopf-algebras-1st-
edition-makhlouf-abdenacer/
Classical Hopf Algebras and Their Applications Algebra

and Applications 29 Pierre Cartier Frédéric Patras
https://ebookmeta.com/product/classical-hopf-algebras-and-their-
applications-algebra-and-applications-29-pierre-cartier-frederic-
patras/
Pro Apache NetBeans: Building Applications on the Rich
Client Platform 1st Edition Ioannis Kostaras
https://ebookmeta.com/product/pro-apache-netbeans-building-
applications-on-the-rich-client-platform-1st-edition-ioannis-
kostaras/
Code Clone Analysis Research Tools and Practices

Katsuro Inoue
https://ebookmeta.com/product/code-clone-analysis-research-tools-
and-practices-katsuro-inoue/
Introduction to Yokai Studies and On Kokkuri Inoue

Enry■
https://ebookmeta.com/product/introduction-to-yokai-studies-and-
on-kokkuri-inoue-enryo/
Theory of Groups and Symmetries Representations of

Groups and Lie Algebras Applications 1st Edition Alexey
P. Isaev
https://ebookmeta.com/product/theory-of-groups-and-symmetries-
representations-of-groups-and-lie-algebras-applications-1st-
edition-alexey-p-isaev/
Applications of Nanomaterials in Energy Systems 2nd

Edition Eleftheria C Pyrgioti Ioannis F Gonos Diaa
Eldin A Mansour
https://ebookmeta.com/product/applications-of-nanomaterials-in-
energy-systems-2nd-edition-eleftheria-c-pyrgioti-ioannis-f-gonos-
diaa-eldin-a-mansour/
Lecture Notes in Mathematics 2298
Maria Fragoulopoulou
Atsushi Inoue
Martin Weigt
Ioannis Zarakas
Generalized
B*-Algebras and
Applications
Lecture Notes in Mathematics
Volume 2298
Editors-in-Chief
Jean-Michel Morel, CMLA, ENS, Cachan, France
Bernard Teissier, IMJ-PRG, Paris, France
Series Editors
Karin Baur, University of Leeds, Leeds, UK
Michel Brion, UGA, Grenoble, France
Alessio Figalli, ETH Zurich, Zurich, Switzerland
Annette Huber, Albert Ludwig University, Freiburg, Germany
Davar Khoshnevisan, The University of Utah, Salt Lake City, UT, USA
Ioannis Kontoyiannis, University of Cambridge, Cambridge, UK
Angela Kunoth, University of Cologne, Cologne, Germany
László Székelyhidi , Institute of Mathematics, Leipzig University, Leipzig,
Germany
Ariane Mézard, IMJ-PRG, Paris, France
Mark Podolskij, University of Luxembourg, Esch-sur-Alzette, Luxembourg
Sylvia Serfaty, NYU Courant, New York, NY, USA
Gabriele Vezzosi, UniFI, Florence, Italy
Anna Wienhard, Ruprecht Karl University, Heidelberg, Germany
This series reports on new developments in all areas of mathematics and their
applications - quickly, informally and at a high level. Mathematical texts analysing
new developments in modelling and numerical simulation are welcome. The type of
material considered for publication includes:
1. Research monographs
2. Lectures on a new field or presentations of a new angle in a classical field
3. Summer schools and intensive courses on topics of current research.
Texts which are out of print but still in demand may also be considered if they fall
within these categories. The timeliness of a manuscript is sometimes more important
than its form, which may be preliminary or tentative.
Titles from this series are indexed by Scopus, Web of Science, Mathematical
Reviews, and zbMATH.
Maria Fragoulopoulou • Atsushi Inoue •
Martin Weigt • Ioannis Zarakas
Generalized B*-Algebras
and Applications
Maria Fragoulopoulou Atsushi Inoue
Department of Mathematics Department of Applied Mathematics
National and Kapodistrian Fukuoka University
University of Athens Fukuoka, Japan
Athens, Greece
Martin Weigt Ioannis Zarakas

Department of Mathematics and Applied Department of Mathematics
Mathematics Hellenic Army Academy
Nelson Mandela University Vari, Greece
Port Elizabeth, South Africa
ISSN 0075-8434 ISSN 1617-9692 (electronic)

Lecture Notes in Mathematics
ISBN 978-3-030-96432-0 ISBN 978-3-030-96433-7 (eBook)
https://doi.org/10.1007/978-3-030-96433-7
Mathematics Subject Classification: 46H20, 46H35, 46K10, 47L60, 46H30, 46K05, 46L60
© The Editor(s) (if applicable) and The Author(s), under exclusive license to Springer Nature Switzerland
AG 2022
This work is subject to copyright. All rights are solely and exclusively licensed by the Publisher, whether
the whole or part of the material is concerned, specifically the rights of translation, reprinting, reuse
of illustrations, recitation, broadcasting, reproduction on microfilms or in any other physical way, and
transmission or information storage and retrieval, electronic adaptation, computer software, or by similar
or dissimilar methodology now known or hereafter developed.
The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication
does not imply, even in the absence of a specific statement, that such names are exempt from the relevant
protective laws and regulations and therefore free for general use.
The publisher, the authors and the editors are safe to assume that the advice and information in this book
are believed to be true and accurate at the date of publication. Neither the publisher nor the authors or
the editors give a warranty, expressed or implied, with respect to the material contained herein or for any
errors or omissions that may have been made. The publisher remains neutral with regard to jurisdictional
claims in published maps and institutional affiliations.
This Springer imprint is published by the registered company Springer Nature Switzerland AG
The registered company address is: Gewerbestrasse 11, 6330 Cham, Switzerland
see [G.S. Kirk, J.E. Raven, The Presocratic
Philosophers (Cambridge Univ. Press,1957/
with corrections, Cambridge, 1973), pp.
236–237]
Contemporaneously with these philo-

sophers [sc. Leucippus and Democri-
tus], and before them, the Pythagore-
ans, as they are called, devoted them-
selves to mathematics; they were the
first to advance this study, and hav-
ing been brought up in it they thought
its principles were the principles of all
things.
W. Jaeger (ed.)
Aristotelis Metaphysica
book 1, 985b 23-26
see [W. Jaeger (ed.), Aristotelis Metaphysica
(Oxford Univ. Press, Oxford, 1957), pp. 13,14]
To the fond memory of the late Professor
G.R. ALLAN
and to Professor
P.G. DIXON
with much respect
Preface
In 1967, G.R. Allan initiated and studied a class of locally convex algebras with
continuous involution, called GB∗ -algebras (an abbreviation for “Generalized B ∗ -
algebras”). The structure of a GB∗ -algebra A[τ ] is defined by a certain collection
B∗A of subsets of the underlying locally convex ∗-algebra A[τ ], which, being
partially ordered by inclusion, attains a maximal member, denoted by B0 . Among
the first results, Allan showed an algebraic Gelfand–Naimark type theorem for
commutative GB∗ -algebras. Namely, he proved that a commutative GB∗ -algebra
A[τ ] is algebraically ∗-isomorphic to a ∗-algebra of extended-complex valued
continuous functions on a compact Hausdorff space M0 . The latter is, in fact,
the Gelfand space of the C ∗ -subalgebra A[B0 ] := {λx : λ ∈ C, x ∈ B0 } of
A[τ ], endowed with the gauge function of B0 . This C ∗ -algebra is the key tool for
investigating the structure of a GB∗ -algebra.
In 1970, P.G. Dixon extended Allan’s definition of a GB∗ -algebra to include
topological ∗-algebras that are not locally convex, thus enriching the set of examples
of GB∗ -algebras. Dixon then showed that the locally convex noncommutative GB∗ -
algebras are realized by closed operators on a Hilbert space. Namely, he gave a
noncommutative algebraic Gelfand–Naimark type theorem for GB∗ -algebras, which
determines them among unbounded operator algebras.
Typical examples of GB∗ -algebras are C ∗ -algebras, pro-C ∗-algebras (i.e., inverse
limits of C ∗ -algebras), C ∗ -like locally convex ∗-algebras initiated by A. Inoue and
K.-D. Kürsten, the Arens algebra
Lω [0, 1] = ∩ Lp [0, 1]
1≤p<∞
(Allan) equipped with the topology of the Lp -norms, 1 ≤ p < ∞ and the algebra
M[0, 1] of all measurable functions on [0, 1] (modulo equality a.e.), endowed with
the topology of convergence in measure, which is not necessarily locally convex
(Dixon).
Thus, GB∗ -algebras generalize the celebrated C ∗ -algebras, which consist entirely
of bounded operators. As noted, a GB∗ -algebra A[τ ] consists mainly of unbounded
ix
x Preface
operators. The bounded operators that it may contain are all concentrated in the
C ∗ -subalgebra A[B0 ] of A[τ ], mentioned above.
The main body of the present monograph is based on the theory of GB∗ -algebras
as developed by Allan and Dixon. In addition, it is augmented by related results
of, among others, S.J. Bhatt, W. Kunze, G. Lassner, M. Oudadess, K. Schmüdgen,
A.W. Wood, and the authors.
It is very well known that our physical world mostly consists of unbounded
operators. Among them, the most well-known are the Hamiltonian operator H rep-
resenting the observable energy, and the operators P , Q representing the observable
momentum and observable position, respectively. The following algebraic equations
involving the previous operators
P2 mω2 Q2
[P , Q] = P Q − QP = −i h̄I, H = + ,
2m 2
correspond to the one-dimensional harmonic oscillator, where i is the imaginary
unit, h̄ is the Planck’s constant, I is the identity operator, and m, ω the mass
and frequency of the oscillator, respectively. According to quantum mechanics, a
physical observable is represented by a self-adjoint linear operator, while certain
algebraic relations, as before, correspond to a physical system, whose mathematical
image is an operator ∗-algebra in an inner product space. In the case of the one-
dimensional harmonic oscillator, the respective ∗-algebra is the one generated by
the essentially self-adjoint operators, H, P , Q.
The journey to the quantum mechanical relation P Q − QP = −i h̄I , known as
the Heisenberg (or canonical) commutation relation, was not straightforward. More-
over, the conclusion that the quantum mechanical observables are not represented
by numbers but by operators is considered to be one of the greatest achievements of
science.
From the above it is clear and quite natural why scientists were led to the study
of unbounded operator algebras, among them being the GB∗ -algebras. This is the
reason why the latter have been investigated, in different directions, by various
authors.
Athens, Greece Maria Fragoulopoulou

Fukuoka, Japan Atsushi Inoue
Port Elizabeth, South Africa Martin Weigt
Vari, Greece Ioannis Zarakas
Contents
1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 1
2 A Spectral Theory for Locally Convex Algebras . . . . .. . . . . . . . . . . . . . . . . . . . 9
2.1 Basic Definitions and Notation .. . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 9
2.2 The Set of Bounded Elements. Radius of Boundedness . . . . . . . . . . . . . . 11
2.3 Spectrum and Spectral Radius .. . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 20
2.4 A Functional Calculus . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 30
2.5 The Carrier Space . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 32
3 Generalized B*-Algebras: Functional Representation Theory . . . . . . . . . 39
3.1 Hermitian and Symmetric Locally Convex *-Algebras . . . . . . . . . . . . . . . 39
3.2 Some Results on C*-Algebras .. . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 45
3.3 GB*-Algebras . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 51
3.4 Commutative GB*-Algebras: Functional Representation Theory . . . . 62
3.5 C*-Like Locally Convex *-Algebras as GB*-Algebras .. . . . . . . . . . . . . . 70
4 Commutative Generalized B*-Algebras: Functional Calculus
and Equivalent Topologies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 79
4.1 Functional Calculus .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 79
4.2 Positive Linear Functionals .. . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 83
4.3 Equivalent Topologies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 91
4.4 A *-Algebra of Functions with no GB*-Topology . . . . . . . . . . . . . . . . . . . . 99
5 Extended C*-Algebras and Extended W*-Algebras .. . . . . . . . . . . . . . . . . . . . 103
5.1 O∗ -Algebras and Unbounded *-Representations ... . . . . . . . . . . . . . . . . . . . 103
5.2 Uniform Topologies on O∗ -Algebras . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 106
5.3 Extended C*-Algebras .. . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 114
5.4 Left Extended W*-Algebras of Unbounded Hilbert Algebras .. . . . . . . 116
6 Generalized B*-Algebras: Unbounded *-Representation Theory .. . . . . 127
6.1 A Functional Calculus . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 127
6.2 Positive Elements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 129
xi
xii Contents
6.3 *-Representations of GB*-Algebras . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 137

6.4 Ogasawara’s Commutativity Condition in GB*-Algebras . . . . . . . . . . . . 154
7 Applications I: Miscellanea . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 157
7.1 The Completion of a C*-Algebra Under a Locally
Convex Topology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 157
7.2 Continuity of Positive Linear Functionals of GB*-Algebras . . . . . . . . . 167
7.3 Every GB*-Algebra has a Bounded Approximate Identity . . . . . . . . . . . 170
7.4 Inverse Limits and Quotients of GB*-Algebras . . .. . . . . . . . . . . . . . . . . . . . 172
7.5 A Vidav-Palmer Theorem for GB*-Algebras .. . . . .. . . . . . . . . . . . . . . . . . . . 178
8 Applications II: Tensor Products .. . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 211
8.1 Prerequisites: Examples . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 211
8.2 B∗ -Collections in Tensor Product Locally Convex *-Algebras . . . . . . 215
8.3 Tensor Product GB*-Algebras . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 218
8.4 GB*-Nuclearity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 223
8.5 Some Applications on Tensor Product GB*-Algebras . . . . . . . . . . . . . . . . 226
Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 229
Index . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 235
Chapter 1
Introduction
A Generalized B ∗ -algebra (for short GB∗ -algebra) is a symmetric pseudo-complete

locally convex ∗-algebra A[τ ], with an identity element, whose strucure is de-
termined by a collection B∗A of certain subsets of A that enjoy algebraic and
topological properties (Definition 3.3.2).
The aim of the book in hands, is to exhibit the results of the fathers of the
subject G.R. Allan and P.G. Dixon, as carried out in [3–5, 48–50]. Related results
by other authors are also included, providing useful information and indicating how
far the investigation of GB∗ -algebras can reach. There is a small number of new
results presented in Sects. 6.3 and 6.4. There are results that are not included in the
monograph, although they are connected with GB∗ -algebras; some of them can be
found in [19, 20, 38, 44, 56, 121, 153–155].
As was remarked in the Preface, GB∗ -algebras are realized by unbounded
operators acting on a dense subspace of a Hilbert space (see e.g., Theorems 6.3.5
and 6.3.11). For the most known unbounded linear operators P , Q, H mentioned in
Preface, the reader is referred to [29, 57, 108, 138, 142].
In this regard, notice that until the theory of quantum was recognized, it passed
from various stages, that could not be interpreted by the views of classical physics of
that time. This was first stated, in 1900, by the German physicist Max Planck, aiming
to clarify the laws of the ‘thermic radiation’ that are taken experimentally, but cannot
be interpreted theoretically through classical physics. For blending the theoretical
computations, with the experimental data, Planck stated the theory of quantum
according to which the light, and more general, the electromagnetic radiation is
emitted and absorbed not continuously, but in elementary amounts, which he named
quantum of light. Later Einstein used the name photon instead of quantum.
The quantum theory interpreted various phenomena like photoelectric, fluo-
rescence, etc. From the end of 1924 to the beginning of 1925 Heisenberg and
Schrödinger introduced independently equivalent, although seemingly not similar,
interpretations for the empirical quantization laws of Bohr and Sommerfeld. These
laws were based on the matrix, respectively wave mechanics. Later on, it turned out
© The Author(s), under exclusive license to Springer Nature Switzerland AG 2022 1

M. Fragoulopoulou et al., Generalized B*-Algebras and Applications, Lecture Notes
in Mathematics 2298, https://doi.org/10.1007/978-3-030-96433-7_1
2 1 Introduction
that these two approaches are equivalent and correspond to two different realizations
of the momentum and position operators P , Q of quantum mechanics. For a very
recent algebraic approach to quantum theories, see [138, Chapter 1].
But, GB∗ -algebras occur also, among the so-called unbounded Hilbert algebras
[78–83, 157], that are very important for the Tomita Takesaki theory for unbounded
operator algebras developed in [85], by the second named author; on the other hand,
they contribute to the rising of the so-called EW∗ -algebras; see Chap. 5, as well
[49, 83, 85], that also play a decisive role in the aforementioned Tomita Takesaki
theory.
All the preceding combined with the desire of each one of us to make the
structure of GB∗ -algebras more familiar to a wider spectrum of researchers, gave
us the motive and impetus for attempting the writing of this monograph.
The whole essay is divided in seven chapters. Chapter 2 offers a preparatory stage
for the introduction and basic properties of GB∗ -algebras. Namely, Chap. 2 gives a
general theory of locally convex algebras with separately continuous multiplication,
aiming to a proper definition of spectrum for elements of this sort of algebras and to
the investigation of its properties, in comparison to the usual theory of spectrum in
the classical case. A motivation for this comes from the spectral theory of a closed
operator T on a Banach space E, where the spectrum of T is given by those complex
numbers λ, such that the operator λI − T has no bounded inverse (see, for instance,
[145]), where I is the restriction of the identity operator of E on the domain of T .
Thus, a proper definition of a bounded element (Allan) in a locally convex algebra
should be given first. This was done in Definition 2.2.1 and its choice is confirmed
from the theory that rises from it, based on the definition of the spectrum of an
element (see Sect. 2.3), as it can be seen from Theorem 2.3.7 and its corollaries. In
Theorem 2.3.13, the new concept of spectrum is related with the usual concept of
the spectrum of an element, when the considered locally convex algebra A[τ ] has
continuous inversion. When A[τ ] is also pseudo-complete (a weaker notion than
completeness, that plays an essential role in the whole theory of GB∗ -algebras),
then Allan bounded elements are just characterized by the boundedness of the usual
spectrum (Corollary 2.3.14). Similar ideas to the previous ones were considered
earlier by L. Waelbroeck [148], in 1957, but in a more specific framework and
under the assumption of quasi-completeness and commutativity of the topological
algebras involved. Moreover, a comparable definition to that of G.R. Allan for
a bounded element, but in an m-convex algebra, was given by S. Warner [149,
p. 197, Definition 3], in 1956. Finally, in Sects. 2.4 and 2.5 a functional calculus
for a pseudo-complete locally convex algebra, respectively the carrier space of a
commutative pseudo-complete locally convex algebra A[τ ] are discussed, where
in both cases the set A0 of (Allan-)bounded elements of A[τ ] plays an important
role; in the second case A0 coincides with A[B0 ], therefore it is a C ∗ -algebra (cf.
Lemma 3.3.7(ii) and Theorem 3.3.9(i)), although, in the general case, A0 is not even
a subspace. In this regard, an interesting result is given by Corollary 6.4.5.
Chapter 3 deals with the basic theory of GB∗ -algebras. More precisely, Sect. 3.1
concerns hermitian and symmetric locally convex ∗-algebras (see Definition 3.1.6),
where the respective concepts are given by using, in fact, (Allan) bounded elements
1 Introduction 3
(cf. Definitions 2.2.1 and 2.3.1). Note that the classical definitions of hermitic-
ity and symmetry are completely algebraic, given through the usual spectrum
sp(x) of an element x in a ∗-algebra A (see discussion before Definition 3.1.3).
Every symmetric locally convex ∗-algebra is “classically” symmetric and as in
the algebraic case, every symmetric pseudo-complete locally convex ∗-algebra is
hermitian (Corollary 3.1.7); symmetry and pseudo-completeness are among the
main ingredients of a GB∗ -algebra (Definition 3.3.2). In Sect. 3.2, another of the
main ingredients of a GB∗ -algebra A[τ ], the collection B∗A arises, in an attempt
of characterizing the C ∗ -condition in a normed ∗-algebra A[ · ] by using not
the properties of the given norm · , but the properties of A[ · ], as a locally
convex ∗-algebra (see, e.g., Theorem 3.2.9). Sect. 3.3 discusses and comments on
the definitions of a GB∗ -algebra given by Allan and Dixon (cf., e.g., Remark 3.3.4
and Definitions 3.3.5, 3.3.6). At the same time, it unfolds the fundamental structure
of such an algebra and presents several examples. Section 3.4 gives a functional
representation of a commutative GB∗ -algebra A[τ ], which is an algebraic analogue
of the commutative Gelfand–Naimark theorem for C ∗ -algebras, due to G.R. Allan.
The decisive role of the C ∗ -subalgebra A[B0 ] of A[τ ] plays an essential role, in
this regard (Theorem 3.4.9). In Sect. 3.5, the subclass of GB∗ -algebras, called C ∗ -
like locally convex algebras, is discussed; this was introduced by A. Inoue and
K.-D. Kürsten, in 2002 (cf. [88]). The main result in this section states that every
C ∗ -like locally convex algebra A[τ ] is a GB∗ -algebra, with B0 the unit ball of the
“bounded part” Ab of A[τ ] (see Theorem 3.5.3 and the indicated by discussion
in Remark 3.5.7(3)). An important fact in this direction is the comparison of the
∗-subalgebras A[B0 ], D(p ) of A[τ ] in the case of a GB∗ -algebra A[τ ] and their
coincidence with Ab , when A[τ ] is a C ∗ -like locally convex ∗-algebra. For this (cf.
Corollary 3.5.4 and Proposition 3.5.8, together with all results between them).
Chapter 4 considers various aspects of the theory of commutative GB∗ algebras.
Namely, Sect. 4.1 treats a functional calculus analogous to that of commutative
C ∗ -algebras (cf., for instance, Theorem 4.1.2(Allan)). Such a deal is natural, after
an algebraic commutative Gelfand–Naimark type theorem has been proved for
GB∗ -algebras (Theorem 3.4.9). The preceding functional calculus is extended to
noncommutative GB∗ -algebras, in Chap. 6 (Theorem 6.1.3 (Dixon)). In Sect. 4.2,
it is proved that every commutative GB∗ -algebra admits an abundance of positive
linear functionals that separate its points (Corollary 4.2.6, Allan). Unlike to the
C ∗ -algebras theory, there is only a ‘partial analogue’ of the last result in the
noncommutative case, due to Dixon (Theorem 6.3.4). It is well known that every C ∗ -
algebra has a unique C ∗ -norm. Section 4.3 investigates whether a similar situation
happens for GB∗ -algebras. In this regard, one defines an equivalence between
two GB∗ -topologies, in virtue of the B∗ collections corresponding to the given
topologies; this means that the C ∗ -algebras generated by the maximal elements of
the B∗ -collections, under consideration, coincide. It is proved that any two locally
convex ∗-algebra topologies, that make a commutative ∗-algebra with identity a
GB∗ -algebra, are equivalent in the preceding sense (Corollary 4.3.10). This result
of G.R. Allan was extended later to the noncommutative case by P.G. Dixon
(see Corollary 6.3.7). Furthermore, it is shown that on every commutative GB∗ -
4 1 Introduction
algebra A[τ ] there is a finest locally convex ∗-algebra topology “equivalent” to τ

(Theorem 4.3.13). The last Sect. 4.4 presents an example of P.G. Dixon, according
to which there is a ∗-algebra of functions with no GB∗ -topology.
In Chap. 5, GB∗ -algebras of closable operators in a Hilbert space are discussed.
Section 5.1 introduces the basic definitions and properties of unbounded operator
algebras called O∗ -algebras and unbounded ∗-representations of a ∗-algebra. The
notion of closedness (resp. self-adjointness and integrability) of an O∗ -algebra
is defined and studied in analogy with the notion of a closed (resp. self-adjoint)
operator. Section 5.2 introduces several locally convex topologies on O∗ -algebras.
The uniform topologies τu , τ u and τ∗u and the quasi-uniform topologies τqu and
∗ , which are generalizations of the operator-norm topology, in the bounded
τqu
case, are defined and they are different to one another (see (5.2.3)). Moreover,
other topologies called weak, strong, strong*, quasi-strong and quasi-strong* are
defined and the relations among all the preceding topologies are investigated
(see Proposition 5.2.1(1)). Section 5.3, deals with a symmetric O∗ -algebra called
extended C ∗ -algebra (resp. extended W ∗ -algebra), whose bounded part is a C ∗ -
algebra (resp. a von Neumann algebra). In this regard, it is proved that a closed
O∗ -algebra M is an extended C ∗ -algebra, if and only if, M[τu ] is a locally
convex GB∗ -algebra of Dixon (Theorem 5.3.4). In Sect. 5.4, the concept of an
unbounded Hilbert algebra, which is a generalization of a standard Hilbert algebra,
is introduced. Given a Hilbert algebra A0 , we use the noncommutative integration
theory of I.E. Segal to define the noncommutative Arens algebra Lω2 (A0 ), which
is an unbounded Hilbert algebra, maximal among all unbounded Hilbert algebras
containing A0 (Theorem 5.4.4). In case A0 has an identity element, then Lω2 (A0 )
is a Fréchet GB∗ -algebra equipped with the locally convex topology defined by the
family of Lp norms { · p }2≤p<∞ (Theorem 5.4.10). Furthermore, an extended
W ∗ -algebra is constructed by the left regular representation of the unbounded
Hilbert algebra Lω2 (A0 ) and the left regular representation of the Hilbert algebra
A0 (Theorem 5.4.5). Note that the unbounded Hilbert algebras play an important
role in the Tomita–Takesaki theory in algebras of unbounded operators (see [85]).
In Chap. 6, our aim is to present a noncommutative Gelfand–Naimark type
theorem for GB∗ -algebras. For this purpose, Sect. 6.1 discusses a noncommutative
functional calculus (Theorem 6.1.3 (Dixon)) similar to the commutative one
discussed in Sect. 4.1. Section 6.2 introduces positive elements in a GB∗ -algebra
A[τ ]. Denoting by A+ the set of all positive elements in A[τ ], it is proved that
A+ is a convex cone (Theorem 6.2.5), τ -closed under a certain condition; the
latter is fulfilled by a barrelled and bornological topology T defined on A, finer
than τ (see Theorems 6.2.6, 6.2.11 and 6.2.13). In Sect. 6.3, after discussing
the algebraic version of the Gelfand–Naimark–Segal construction and defining the
direct sum ∗-representation of a family of closed ∗-representations, we prove that
every GB∗ -algebra has enough positive linear functionals in order to separate its
points (Theorem 6.3.4). All these lead to an ‘algebraic’ extension of the celebrated
noncommutative Gelfand–Naimark theorem for C ∗ -algebras in the context of GB∗ -
algebras (Dixon); see Theorem 6.3.5. In this regard, a natural question is whether
one can obtain an extension of the latter result, where the constructed faithful ∗-
1 Introduction 5
representation is bicontinuous on its image. This is achieved by employing the so

called AO∗ -algebras, introduced and studied by G. Lassner in [103], K. Schmüdgen
[133] and others. In this regard, see Theorem 6.3.11. Among the biproducts of the
latter theorem, Corollary 6.3.16 characterizes an AO∗ -algebra A[τ ] in the context of
Fréchet GB∗ -algebras by the fact that each self-adjoint continuous linear functional
on A[τ ] is expressed as a difference of two (continuous) positive linear functionals
on A[τ ]. The last Sect. 6.4 deals with an analogue of Ogasawara’s commutativity
condition theorem in C ∗ -algebras, within the context of GB∗ -algebras. The corre-
sponding result is due to M. Oudadess [118] (see Theorem 6.4.1). Two byproducts
of Theorem 6.4.3, that are also due to M. Oudadess and give a more general version
of Ogasawara’s commutativity condition theorem, are Corollaries 6.4.4 and 6.4.5
that provide interesting new results for a GB∗ -algebra.
The last two chapters are devoted to applications of the developed theory in
the previous five chapters. Thus, in Chap. 7 and, in particular, in Sect. 7.1, we
investigate the structure and the representation theory of the completion of a given
C ∗ -algebra A[ · ] with identity, with respect to a locally convex ∗-algebra
topology τ , coarser than the topology of the given C ∗ -norm and such that the
multiplication in A[τ ] is jointly continuous. Such a study started in 2006, with
[12] and continued by [62], in 2007. If we denote by A[τ ] the aforementioned
∗
completion, it was proved that this is a GB -algebra over the closed unit ball U(A)
of A[ · ], if and only if, U(A) is τ -closed (see Corollary 7.1.3). Even more,
] is a GB∗ -algebra, if and only if, there exists a faithful (τ − τs ∗ )-continuous
A[τ
∗-representation π of A[τ ], such that τ ≺ rπ , where τs ∗ is the strong∗ topology
on L (D(π)) and rπ an unbounded C ∗ -seminorm in A[τ
† ] (see Theorem 7.1.7). If
the multiplication on A is not jointly continuous with respect to τ , then A[τ ] is
not necessarily a locally convex ∗-algebra, but it has the structure of a partial ∗-
algebra. The investigation of structure and representation theory of A[τ ], in this
case, was studied in [12, Section 3] and [62, Section 3]. Partial ∗-algebras consist
of unbounded operators and admit a product operation that is not defined for all
pairs of elements. For a complete account in their theory and a rich literature, see
[7]. Section 7.2 studies continuity of positive linear functionals on a GB∗ -algebra.
Unlike to the C ∗ -case, positive linear functionals on a GB∗ -algebra are not always
continuous (see Theorem 7.2.1, its corollaries and Example 7.2.4). In Sect. 7.3, it
is proved that as in the C ∗ -case, every GB∗ -algebra admits a bounded approximate
identity (Theorem 7.3.1, Bhatt). In Sect. 7.4, it is shown that the inverse limit of an
inverse system of GB∗ -algebras is again a GB∗ -algebra (cf. Corollary 7.4.7, Kunze).
Moreover, every closed ideal of a GB∗ -algebra A[τ ] is a ∗-ideal and the quotient
algebra A/I , under the quotient topology, is a GB∗ -algebra with underlying C ∗ -
algebra the quotient A[B0 ]/(I ∩ A[B0 ]) (Theorem 7.4.8, Kunze); recall that A[B0 ]
is the underlying C ∗ -algebra of A[τ ]. Finally, Sect. 7.5 deals with various types
of the well known Vidav–Palmer theorem for GB∗ -algebras, due to A.W. Wood
[156] (see, for example, Theorems 7.5.43, 7.5.49, 7.5.51). For stating such results, a
proper definition of the numerical range of an element in a locally convex ∗-algebra
is needed and its basic properties are investigated. At the same time an equivalent
variant of the usual definition of a GB∗ -algebra has been used, given by A.W. Wood,
6 1 Introduction
for reaching the obtained generalizations of a non-normed Vidav–Palmer theorem

(see, for instance, Theorem 7.5.49 and Theorem 7.5.51).
Applications carry on in Chap. 8, which exclusively deals with tensor products
of GB∗ -algebras. Such a matter was first considered in [64]. For other sources of
topological tensor products of unbounded operator algebras, see [1, 65, 71]. There
is a physical justification for using tensor products. They are viewed to describe two
quantum systems as one joint system [2], while the physical significance of tensor
products always depends on the applications, which may involve wave functions,
spin states, oscillators etc.; see e.g., [29, 70].
In Sect. 8.1, a background material is presented concerning standard notions
and known results on topological tensor products. Moreover, some examples of
GB∗ -algebras are exhibited that are realized as tensor products of GB∗ -algebras.
The Example 8.1.4 leads us to Definition 8.2.5 of the tensor product of two GB∗ -
algebras. Section 8.2 deals with B∗ collections (see Definitions 3.3.1, 3.3.2) in the
case of topological tensor products of (pseudo-complete) locally convex ∗-algebras.
In Sect. 8.3, we testify Definition 8.2.5 of the GB∗ -tensor product in several cases.
For instance, if ε is the injective locally convex tensor topology, Corollary 8.3.6
shows that if A1 [τ1 ], A2 [τ2 ] are complete locally convex ∗-algebras with identities,
such that either A1 or A2 is commutative, then A1 ⊗ A2 is a tensor product GB∗ -
ε
algebra, if and only if, A1 [τ1 ] and A2 [τ2 ] are GB∗ -algebras. In fact, commutativity
is needed only in the ‘if’ part. Justifications of the usage of commutativity are given
at the introduction of Sect. 8.3. Section 8.4, touches nuclear GB∗ -algebras, where
some examples are presented and a few characterizations are mentioned together
with some questions on this interesting matter. For some further information on this
topic, see [152]. The final Sect. 8.5 provides some applications, concerning mainly
existence and properties of (unbounded) ∗-representations on tensor product GB∗ -
algebras.
We have tried to make this book as self-contained as possible and have tried to
give rather detailed proofs of the results, for the convenience of the readers. In cases
that some proofs are omitted, full references are given.
For the reading of this book, some familiarity is assumed with general topology
[97, 100], functional analysis [73, 74, 91, 113, 123, 128, 129, 131], Banach and
Banach ∗-algebras [32, 43, 72, 115, 120, 127], C ∗ -algebras and von Neumann
algebras [45, 46, 52, 114, 122, 130, 144], basic theory of topological (*-)algebras
[15, 60, 72, 111, 112, 136, 158] and basic theory of bounded and unbounded
operators [7, Chapter 1], [66, pp. 237–248], [95, 114, 136, 144]. Our suggestions
are only indicative, since almost all mentioned fields have to present a very rich
literature.
The last 31 years, seven books have been published on unbounded operator
algebras and their representation theory. The readers, who are interested in being
more acquainted with unbounded operator theory and its applications, may chrono-
logically be referred to [7, 53, 66, 85, 136, 138] and [86].
Finally, we would like to express our cordial thanks to all of our colleagues and
collabotators, with whom either we had useful discussions on the topics of the
1 Introduction 7
present book, or we were receivers of their important comments, having them as

listeners in our seminars, on relevant topics.
Last, but not least, we want to address our heartfelt thanks to our colleague
Professor Emerita Maria Papathanassiou (who is also an archeologist), for providing
us with the ancient Greek text of Aristotle, as well with its translation in English (see
front matter).
Chapter 2
A Spectral Theory for Locally Convex
Algebras
Chapter 2 acts as a preparative stage to the basic definitions and fundamentals

of our main theme, which is the development of the theory of GB∗ -algebras, as
algebras of unbounded operators. More precisely, the notion of a ‘bounded element’
(Definition 2.2.1) in a locally convex algebra is introduced (Allan), as well the
concept of spectrum of an element (Definition 2.3.1). In addition, the carrier space
of a commutative complete locally convex algebra is defined (cf. Sect. 2.5) and a
Gelfand-type theory is developed.
2.1 Basic Definitions and Notation
All the algebras considered throughout this book are complex, linear, associative. If
an algebra A has an identity element, this will be denoted by e. We emphasize that
all topological spaces are considered Hausdorff, unless indicated otherwise.
In this section we exhibit the basic definitions and basic notations that will be
used in this book.
The symbols N, R, C stand for the natural, real and complex numbers, respec-
tively. The imaginary unit is denoted by i and the conjugate of a complex number
α by α.
Let X be a topological space with two topologies τ, τ . In order to indicate that τ
is coarser than τ , or equivalently τ is finer than τ , we shall write τ ≺ τ . In order
to indicate that τ , τ are equivalent, in the usual sense, we shall write τ ≈ τ ; see
also Definition 4.3.1 and Remark 4.3.2.
If X[τ ] is a topological space endowed with a topology τ and S a subset of X,
τ
the closure of S will be denoted by S or for distinction, by S .
If A is an algebra with identity, denote by GA , the group of the invertible
elements of A.
© The Author(s), under exclusive license to Springer Nature Switzerland AG 2022 9

M. Fragoulopoulou et al., Generalized B*-Algebras and Applications, Lecture Notes
in Mathematics 2298, https://doi.org/10.1007/978-3-030-96433-7_2
10 2 A Spectral Theory for Locally Convex Algebras
If A is an algebra without identity, the symbol A1 will stand for its unitization.
That is, A1 = A ⊕ C, with linear operations defined coordinatewise and multipli-
cation by
(x, α)(y, β) := (xy + αy + βx, αβ), ∀ x, y ∈ A and α, β ∈ C.
For the identity of the unitization A1 of A, we shall use the symbol e1 ≡ (0, 1).
Take again an algebra A without identity. If x is in A, employing the circle
operation ◦, we shall say that x is quasi-invertible, if there are elements y, z ∈ A
with
x ◦ y = 0 = z ◦ x, where x ◦ y := x + y + xy
(alternatively, you can use in the previous equality −xy instead of +xy). If x is
quasi-invertible in A, then y = z is unique, it is called the quasi-inverse of x and
it is denoted by x ◦ [60, 111]. It is easily seen, that x ∈ A is quasi-invertible with
quasi-inverse x ◦ , if and only if, e + x is invertible in the unitization A1 of A with
inverse e + x ◦ . Readily, the same is true if A has an identity. The symbol GA , will
q
stand for the group of quasi-invertible elements of A.

We first give some definitions and terminology needed in what follows. For the
basic theory of topological vector spaces and general topological algebras with or
without involution, the reader is referred, respectively to [74, 113, 128, 131] and
[15, 60, 72, 111, 112].
Definition 2.1.1 By a topological algebra A[τ ] we shall mean a topological vector
space, which is also an algebra, such that the ring multiplication is separately
continuous. A[τ ] is said to be a locally convex algebra, if it is a topological algebra,
whose underlying topological vector space is a locally convex space. In this case,
the topology τ of A may be described by an upwards directed family of (nonzero)
seminorms, or otherwise by a saturated family of seminorms [74, p. 96]. Such a
family will be usually denoted by = {p} or {pν }ν∈ , with an upwards directed
index set; sometimes, for distinction, we shall use the symbol A instead of just .
When each seminorm pν , ν ∈ is also submultiplicative (briefly, m-seminorm)
i.e.,
pν (xy) ≤ pν (x)pν (y), ∀ x, y ∈ A and ν ∈ ,
then A[τ ] is called an m-convex algebra, where m-convex is an abbreviation of

“multiplicatively-convex” (the term ‘m-convex’ was used in [158], but other authors
have used instead, the term ‘locally m-convex’; see, e.g., [15, 37, 111, 112]). A
seminorm p with the property p(x) = 0, if and only if, x = 0, is called a norm
and it is usually denoted by · . An m-seminorm, which is also a norm, is called
algebra norm. An algebra A endowed with an algebra norm · , is called normed
algebra and it is denoted by A[ · ]. A complete normed algebra is said to be a
Banach algebra.
2.2 The Set of Bounded Elements. Radius of Boundedness 11
Concerning Definition 2.1.1, we shall (almost) always refer to (and/or

to A ), as a defining family of seminorms for the topology τ . The same will
apply, when we speak for a family of ∗-seminorms, too.
If A[τ ] is a topological algebra without identity, its unitization A1 endowed with

the product topology, denoted by τ1 , is also a topological algebra, for which we
shall use the symbol A1 [τ1 ]. In the case, when A[ · ] is a normed (resp. Banach)
algebra, its unitization is denoted by A1 [ · 1 ], where (x, λ)1 := x + |λ|, for
all (x, λ) ∈ A1 ≡ A × C.
Note that since we are working on algebras, we shall use, without

any particular mention, the term norm, instead of the term algebra norm.
Moreover, we shall always assume that the seminorms in , as above, will
be mutually non-equivalent, in the usual sense of equivalence between two
seminorms.
2.2 The Set of Bounded Elements. Radius of Boundedness
Definition 2.2.1 (Allan) Let A[τ ] be a locally convex algebra. An element x of A

is called (Allan-) bounded and for simplicity just bounded, if there exists a nonzero
complex number λ, such that the set {(λx)n : n ∈ N} is a bounded subset of A[τ ].
For some historical comments on the concept introduced in Definition 2.2.1, see
Introduction, just before the description of the results of Chap. 3. The set of all
bounded elements of A[τ ] will be denoted by A0 . It is straightforward to see that
every element of a normed algebra is bounded in the previous sense. Also if A[τ ]
has an identity e, then e is bounded.
Definition 2.2.2 For a locally convex algebra A[τ ], let (B0 )A denote the collection
of all subsets B of A, which fulfill the following properties:
(1) B is absolutely convex and B 2 ⊂ B;
(2) B is bounded and closed.
In case no confusion arises as to which algebra the previous collection of sets is
regarded, we use the symbol B0 . For every B ∈ B0 , we use A[B] to denote the
subalgebra of A generated by B. Based on the properties that each set B ∈ B0
possesses, we have that

A[B] = λx : λ ∈ C, x ∈ B . (2.2.1)
It is easily proven that for every B ∈ B0 the Minkowski functional (or gauge
function) · B on A[B], that is, the function

xB = inf t > 0 : x ∈ tB , x ∈ A[B], (2.2.2)
defines a norm with which A[B] becomes a normed algebra. If not explicitly stated
otherwise, A[B] will always be assumed to carry this norm topology.
We note that the topology τ , which A[B] carries as a subalgebra of A, is weaker
than ·B . Indeed let (xn )n∈N be a sequence in A[B], such that xn → x, with respect
to · B . Then, since B is τ -bounded, for each 0-neighbourhood in A[τ ], say U ,
there is an ε > 0, such that B ⊂ εU . Due to convergence of the sequence (xn ) to
x, with respect to · B , we have that there is n0 ∈ N, such that xn − xB < 1ε ,
for every n ∈ N, n ≥ n0 . Hence, xn − x ∈ 1ε B, n ≥ n0 ; thus, xn − x ∈ U , for all
n ≥ n0 . Therefore, xn → x with respect to τ .
The intimate relation between the set A0 , of all bounded elements in A, and the
normed algebras A[B], B ∈ B0 is given by the following proposition, for which
the next definition is helpful.
Definition 2.2.3 A subcollection B1 of B0 is called basic if for every B ∈ B0 ,
there is some B1 ∈ B1 , such that B ⊂ B1 .
Proposition 2.2.4 Let A[τ
] be a locally convex
algebra and B1 a basic subcollec-
tion of B0 . Then, A0 = A[B] : B ∈ B1 .
Proof Let x ∈ A[B], for some B ∈ B1 . Then, for λ > xB we have that λ1 x ∈ B.

Since B 2 ⊂ B, by induction we have B n ⊂ B, for n ∈ N, so that ( λ1 x)n : n ∈
1
N ⊂ B. Therefore, the subset ( λ x)n : n ∈ N of A is bounded, since B is
bounded and so x ∈ A0 .
inverse direction, if x ∈ A0 and λ ∈ C, λ = 0, then the set S ≡ (λx) :
In the n
n ∈ N is bounded in A[τ ]. It is easy to show that the closure of the absolutely

convex hull of S belongs to B0 . Therefore, there is some B ∈ B1 , such that S ⊂ B
and so x ∈ A[B].

If for a locally convex algebra A[τ ], all the normed subalgebras A[B], B ∈ B0
are Banach algebras, then A[τ ] is called pseudo-complete .
Proposition 2.2.5 If A[τ ] is sequentially complete, then A[τ ] is pseudo-complete.
Proof Let B ∈ B0 and (xn )n∈N ⊂ A[B], such that the sequence (xn )n∈N is · B -
Cauchy. Since τ is weaker than · B on A[B], (xn )n∈N is τ -Cauchy and therefore,
from the assumption of A[τ ] being sequentially complete, we have that there is an
x ∈ A, such that xn → x, with respect to τ . Let ε > 0. Since (xn )n∈N is · B -
Cauchy, there is a n0 ∈ N, with xm − xn B < ε4 , for all m, n ≥ n0 . Then, since
xm −xn0 → x −xn0 , with respect to τ and B is τ -closed, we have that x −xn0 ∈ ε4 B
and thus x ∈ A[B]. So, for
ε ε ε
m ≥ n0 , xm − x = xm − xn0 + xn0 − x ∈ B + B ⊂ B,
4 4 2
since B is convex. Therefore, xm − xB < ε, for all m ≥ n0 . Hence, A[B] is
sequentially complete, therefore a Banach algebra.

Proposition 2.2.6 If B0 contains a basic subcollection B1 , such that A[B] is a
Banach algebra for every B ∈ B1 , then A[τ ] is pseudo-complete.
Proof Let B ∈ B0 and (xn )n∈N be a ·B -Cauchy sequence in A[B]. There is some
B1 ∈ B1 , such that B ⊂ B1 and hence xB1 ≤ xB , for x ∈ A[B]. Therefore,
(xn )n∈N is a ·B1 -Cauchy sequence in A[B1 ], hence there is an element x ∈ A[B1 ],
such that xn → x with respect to · B1 . Following similar arguments to those of
the proof of Proposition 2.2.5 we have that x ∈ A[B] and xn → x with respect to
· B . Therefore, A[B] is a Banach algebra and since B is an arbitrary element of
B0 , we have that A[τ ] is pseudo-complete.

In the example that follows we are going to show that the converse of Propo-
sition 2.2.5 is not valid. Towards this direction, the notion of convergence in the
sense of Mackey will be proven useful (cf. [74, Chapter 3, §5] and/or [131, Chapter
IV, 3.]). We recall that in a topological vector space E[τ ], a sequence (xn )n∈N is a
Mackey Cauchy sequence if there is some sequence (εn )n∈N of positive numbers,
which tends to 0 and a bounded subset B of E, such that
∀ n ∈ N, xn − xm ∈ εn B, for m > n.
Furthermore, the sequence (xn )n∈N is Mackey convergent to an element x ∈ E, if

there is some sequence (εn )n∈N of positive numbers, which tends to 0 and a bounded
subset B of E, such that
xn − x ∈ εn B, ∀ n ∈ N.
The topological vector space E[τ ] is Mackey complete, if every Mackey Cauchy
sequence in E is Mackey convergent. We note that if a locally convex space E[τ ] is
metrizable and sequentially complete, then it is also Mackey complete.
Indeed, a metrizable locally convex space E[τ ] is bornological (see Defini-
tion 4.3.14(2) in Sect. 4.2 and [131, p. 61, 8.1]). But then, E[τ ] becomes a
Mackey space (ibid., p. 132, 3.4). This means that the topology τ is the Mackey
topology τ (E, E ), E the dual of E[τ ]. Namely, τ (E, E ) is the topology of
uniform convergence on the absolutely convex σ (E , E)-compact subsets of E
(cf. Definitition 4.3.11 and (4.3.12) in Sect. 4.3). But, a metrizable and sequentially
complete locally convex space is complete, hence E[τ ] is Mackey complete.
The example that follows illustrates a locally convex algebra, which is metrizable
and pseudo-complete, but not Mackey complete and so a fortiori not sequentially
complete.
Example 2.2.7 Let A be the algebra of all polynomials in one variable with complex
coefficients. We endow A with the topology τ of uniform convergence on compact
subsets of the positive real line R+ . Then, clearly A[τ ] is metrizable. Moreover,
A0 is the set of all constant functions. Yet, the family (B)A of subsets of A has a
greatest member, say B0 , which consists of all constant functions of absolute value
not exceeding 1. Therefore, A[B0 ] is a Banach algebra (see (2.2.1) and (2.2.2)) and
so from Proposition 2.2.6 we have that A[τ ] is pseudo-complete.
Consider a sequence of polynomials (Pn )n∈N in A, such that

n
x 2r+1
Pn (x) = (−1)r+1 . (2.2.3)
(2r + 1)!
r=0
Then, Pn → sin(x), with respect to τ and the set

B = P ∈ A : |P (x)| ≤ ex , ∀ x ∈ R+
is clearly τ -bounded. Since,
|Pn (x) − sin(x)|e−x → 0, uniformly on R+ ,
we deduce that (Pn )n∈N is a Mackey Cauchy sequence. Nevertheless, (Pn )n∈N is not
convergent to any element of A. Hence, A[τ ] is not Mackey complete.
Note that this last implication is based on the fact that if a sequence (xn )n∈N in a
topological vector space E[τ ] is Mackey convergent to an element, say x ∈ E, then
xn → x, with respect to τ . Indeed, since xn → x in the sense of Mackey, there is a
sequence of positive real numbers (εn )n∈N , which tends to 0 and a bounded subset
V of E, such that xn − x ∈ εn V , for all n ∈ N. Let U be a 0-neighbourhood in E.
There is ε > 0, such that V ⊂ εU , hence εn V ⊂ εn εU, n ∈ N. Since εn → 0,
there is n0 ∈ N, such that εn < ε+1 1
, for every n ≥ n0 . Suppose, without loss of
generality, that U is also balanced. Then, we have that εn V ⊂ εn εU ⊂ U , for every
n ≥ n0 . Therefore, xn − x ∈ U , for all n ≥ n0 , thus xn → x with respect to τ .
Proposition 2.2.8
(1) If A[τ ] is a pseudo-complete algebra and B is a closed subalgebra of A, then
B[τB ] is also pseudo-complete.
(2) If A[τ ] has no identity, then its unitization A1 [τ1 ] is pseudo-complete, if and
only if, A[τ ] is pseudo-complete.
(3) If A[τ ] has an identity, say e, then the collection B : B ∈ B0 , e ∈ B is a
basic subcollection of B0 .
Proof
(1) Let (B0 )A , (B0 )B be the corresponding collections of subsets for A[τ ] and
B[τ B ] respectively. Given the fact that B is a closed subalgebra of A it is
straightforward that (B0 )B ⊂ (B0 )A . Hence, the result follows.
(2) If A1 [τ1 ] is pseudo-complete, then from (1) we have that A[τ ] is pseudo-
complete. In the inverse direction let A[τ ] be pseudo-complete. For B ∈
(B0 )A1 we will show that A1 [B] is a Banach algebra with respect to the norm
· B . We first note that if (λ, x) ∈ B, then λ ∈ D = {z ∈ C : |z| ≤ 1}: indeed
since B 2 ⊂ B, by induction we have that (λ, x)n = (λn , xn ) ∈ B, for all n ∈ N,

where xn is an element of A. Since B is bounded the fact that (λn , xn ) ∈ B is
possible only for those complex numbers λ that belong to D.
Consider now B1 = 13 D + 13 B. It is easy to show that B1 ∈ (B0 )A1 . Let
B2 = {x ∈ A : (λ, x) ∈ B, for some λ ∈ C}. Then, 13 B2 ⊂ B1 ∩ A. If C
denotes the closed absolutely convex hull of 13 B2 , then C ⊂ B1 ∩ A and C ∈
(B0 )A . Hence, by the assumption of pseudo-completeness for A[τ ] we have
that A[C] is a Banach algebra. Let A[C]1 [ · 1C ] be the unitization of A[C],
with norm (λ, x)1C = |λ| + xC , for λ ∈ C, x ∈ A[C]. If Be ≡ U (A[C]1 )
is the closed unit ball of A[C]1 , with respect to the norm · 1C , it can easily be
shown that 16 B ⊂ Be and therefore A1 [B] ⊂ A[C]1 . From the latter inclusion
we have that, on A1 [B], the topology of the norm · 1C is weaker than that of
· B . Indeed, let us consider a sequence (xn )n∈N ⊂ A1 [B], such that xn → x,
with respect to · B and ε > 0. Then,
1 1
∃ m0 ∈ N : < ε and n0 ∈ N with xn − xB < , ∀ n ≥ n0 .
m0 6m0
Therefore,
1 1
xn − x ∈ Be , ∀ n ≥ n0 , so that xn − x1C ≤ < ε, for n ≥ n0 .
m0 m0
Hence, xn → x, with respect to the norm · 1C on A[C]1 .

Since τ ≺ · C on A[C] we have that the norm topology · 1C on A[C]1 is
stronger than the product topology induced on A[C]1 by the product topology
from A[C] × C. Based on this relation between the two topologies and since
B ∈ (B0 )A1 is closed, with respect to the product topology on A1 , we have that
B is · 1C -closed in A[C]1 . Aiming to show that A1 [B] is a Banach algebra, let
(xn )n∈N be a · B -Cauchy sequence in A1 [B]. Since · 1C ≺ · B on A1 [B],
we have that (xn )n∈N is a ·1C -Cauchy sequence in A[C]1 . Since A[C]1 [·1C ]
is a Banach algebra, there exists an x ∈ A[C]1 , such that xn → x, with respect
to · 1C . Let ε > 0. Then,
ε
∃ n0 ∈ N : xn − xm B < , ∀ m, n ≥ n0 .
4
Hence, xn − xn0 ∈ ε4 B, for all n ≥ n0 . Given that xn − xn0 → x − xn0 with

respect to ·1C and that B is ·1C -closed, we have that x−xn0 ∈ ε4 B. Therefore,
ε ε ε
xn − x = (xn − xn0 ) + (xn0 − x) ∈ B + B ⊂ B, ∀ n ≥ n0 .
4 4 2
So, xn − xB < ε, for all n ≥ n0 , i.e. xn → x with respect · B . Therefore,
A1 [B] is a Banach algebra and thus A1 [τ1 ] is pseudo-complete.
(3) For a subset B ∈ B0 let S be the closed absolutely convex hull of B ∪ {e}.
Then it is clear that S ∈ B0 , B ⊂ S and e ∈ S. The result then immediately
follows.

If A is a locally convex algebra with identity e, we shall denote the basic

subcollection of Proposition 2.2.8(3) by (B)A (or simply by B if no confusion
arises). In what follows, for simplicity of notation, we retain the same symbol,
i.e., B, to refer also to the collection of subsets of Definition 2.2.2, for the
case where A does not have an identity.
Theorem 2.2.10 below, will be proven very helpful in various arguments in

the sections of this chapter. In the course of the proof of Theorem 2.2.10, the
following general result, i.e. Proposition 2.2.9, is used, which we record for sake
of completeness. For the proof of Proposition 2.2.9, the reader is referred to [131,
§4.2, p. 83, Theorem].
Proposition 2.2.9 Let E, F be locally convex spaces, such that E is barrelled,
or let E, F be topological vector spaces, such that E is a Baire space. Then,
every bounded subset H of L(E, F ) with respect to the topology of pointwise
convergence on E, where L(E, F ) denotes all continuous linear maps on E into
F , is equicontinuous.
Theorem 2.2.10 If A[τ ] is a commutative and pseudo-complete locally convex
algebra, then B is outer-directed by inclusion, that is for B, C ∈ B, there is F ∈ B,
such that B ∪ C ⊂ F .
Proof Let us first consider the case where A has an identity e. For B, C ∈ B and
for every b ∈ B consider the linear map Lb : A[C] → A, such that Lb (a) =
ba, a ∈ A[C]. From the separate continuity of multiplication in A[τ ] we have
that Lb is continuous. Moreover, since B is bounded we have that the set =
{Lb : b ∈ B} is bounded with respect to the topology of pointwise convergence on
A[C]. Hence, since A[C] is a Banach algebra and A[τ ] a locally convex algebra,
from Proposition 2.2.9 we have that is equicontinuous. Therefore, if V is a 0-
neighbourhood in A[τ ] there is a 0-neighbourhood U in A[C], such that Lb (U ) ⊂
V , for every b ∈ B. Since C is a bounded subset of A[C] there is ε > 0, such that
C ⊂ εU . Thus, for every b ∈ B, we have that bC ⊂ Lb (εU ) ⊂ εV , that is BC is
a bounded subset of A[τ ]. Furthermore, since A is commutative, (BC)2 = B 2 C 2 ,
hence (BC)2 ⊂ BC. Therefore, the closed absolutely convex hull of BC belongs to
B0 . Therefore, by Proposition 2.2.8, there is an F ∈ B, such that BC ⊂ F . Since
e ∈ B ∩ C we conclude that B ∪ C ⊂ BC ⊂ F .
In case A has no identity, for B, C ∈ B we consider B1 , C1 to be the closed
absolutely convex hulls of the sets B ∪ {(λ, 0) : λ ∈ D} and C ∪ {(λ, 0) : λ ∈ D}
respectively, where D denotes the closed unit disk in C. Then B1 , C1 ∈ (B0 )A1
and so, as seen above, there is an F ∈ (B)A1 , such that B1 ∪ C1 ⊂ F . Therefore,

B ∪ C ⊂ F ∩ A, where F ∩ A ∈ (B)A .

Corollary 2.2.11 If A[τ ] is a commutative and pseudo-complete locally convex
algebra, then A0 is a subalgebra of A.
Proof From Proposition 2.2.4 we have that A0 = ∪{A[B] : B ∈ B}. Hence,
for x, y ∈ A0 there are B, C ∈ B such that x ∈ A[B] and y ∈ A[C]. By
Theorem 2.2.10, there is an F ∈ B such that B ∪ C ⊂ F . Therefore, x + y, xy ∈
A[F ] and thus x + y, xy ∈ A0 .

Remark 2.2.12 Note that if A[τ ] is not commutative, then A0 need not even
be a linear subspace. But if A[τ ] is pseudo-complete and x, y ∈ A, such that
xy = yx, then taking the closed subalgebra B of A[τ ] generated by x, y, B is
a commutative pseudo-complete locally convex algebra (see Proposition 2.2.8(i)),
therefore Corollary 2.2.11 implies that xy, x +y ∈ B0 , consequently xy, x +y ∈ A0
too.
Corollary 2.2.13 Let A[τ ] be a commutative and pseudo-complete locally convex
algebra and B0 = ∪{B : B ∈ B}. Then, B0 is absolutely convex, B02 ⊂ B0
and B0 is absorbent in A0 . Hence, the Minkowski functional · B0 of B0 is a
submultiplicative seminorm on A0 .
Proof Let 0 = x ∈ A0 . Then there is B ∈ B such that x ∈ A[B]. Hence, x = λy,
for some λ ∈ C, y ∈ B. Since B is balanced we have that |λ| 1
x = |λ|
λ
y ∈ B ⊂ B0 .
Therefore, B0 is absorbent in A0 . The properties that B0 ⊂ B0 and that of B0 being
2
absolutely convex result from the fact that every B ∈ B enjoys the same properties
and from Theorem 2.2.10.

For a locally convex algebra A[τ ] a useful tool is the notion of the radius of
boundedness for an element x ∈ A. The radius of boundedness β(·) of x is defined
by the relation
1
β(x) := inf λ > 0 : ( x)n : n ∈ N is bounded ,
λ
where inf ∅ = +∞. In Sect. 2.3 we are going to see the relation between the spectral
radius and the radius of boundedness of an element in a locally convex algebra. In
the proposition, which follows some basic properties of β(x) are listed. The proofs
of these properties are obvious, so that are omitted.
Proposition 2.2.14 Let A[τ ] be a locally convex algebra and x ∈ A. Then, the
following hold:
(1) β(x) ≥ 0 and β(λx) = |λ|β(x), for λ ∈ C, with the convention that 0 · ∞ = 0.
(2) β(x) < +∞, if and only if, x ∈ A0 .
(3) β(x) = inf λ > 0 : ( λ1 x)n → 0, for n → ∞ .
(4) For x ∈ A0 , if λ ∈ C, such that |λ| > β(x), then ( λ1 x)n → 0, for n → ∞. If
0 < |λ| < β(x), then the set {( λ1 x)n : n ∈ N} is unbounded.
Proposition 2.2.15 Let A[τ ] be a locally convex algebra.
Then, the
restriction of β
to A0 is the Minkowski functional · B0 of B0 = B: B∈B .

Proof Let x ∈ A0 and λ > 0, such that the set S = ( λ1 x)n : n ∈ N is bounded.
Then the closed absolutely convex hull of S, say B, belongs to B. Thus, B ⊂ B0
and therefore λ1 x ∈ B0 . Hence, xB0 ≤ β(x).
For the reverse inequality, let μ > 0 such that x ∈ μB0 . Then, there is some
B ∈ B, such that μ1 x ∈ B. Since B 2 ⊂ B, by induction we have that ( μ1 x)n :

n ∈ N ⊂ B, hence ( μ1 x)n : n ∈ N is bounded. Therefore,
1
μ > 0 : x ∈ μB0 ⊂ μ > 0 : ( x)n : n ∈ N is bounded ,
μ
from which follows that β(x) ≤ xB0 .

Based on Corollary 2.2.13 and on Proposition 2.2.15 we have the following
result.
Corollary 2.2.16 Let A[τ ] be a commutative and pseudo-complete locally convex
algebra. Then, the restriction of β to A0 is a submultiplicative seminorm.
Corollary
2.2.17 Let A[τ ] be a locally convex algebra and x ∈ A0 . Then, β(x) =
inf xB : B ∈ B, x ∈ A[B] .
Proof Let α > 0, such that x ∈ αB0 . Then, there is B ∈ B, such that x ∈ αB.
Hence,

inf α > 0 : x ∈ αB0 ≥ inf xB : B ∈ B, x ∈ A[B]
and thus from Proposition 2.2.15 we have that

β(x) ≥ inf xB : B ∈ B, x ∈ A[B] .
if B ∈ B with
On the other hand, x ∈ A[B], it is clear that for every ε > 0 :
xB + ε ≥ inf α > 0 : x ∈ αB0 . Therefore, inf xB : B ∈ B, x ∈ A[B] ≥
β(x).

The next proposition provides us with two other relations with which the radius of
boundedness of an element can be expressed. For a locally convex algebra A[τ ], let
A denote the topological dual of A and let τ be a family of seminorms defining
the locally convex topology τ of A. Consider the following formulae

1
β (x) = sup lim sup|f (x n )| n : f ∈ A and
n→∞
1
β (x) = sup lim sup|p(x n )| n : p ∈ τ , x ∈ A.
n→∞
Proposition 2.2.18 Let A[τ ] be a locally convex algebra, and x ∈ A. Then,
β(x) = β (x) = β (x).
Proof Let x ∈ / A0 . Then, β(x) = +∞ and hence, trivially, |f (x)| ≤ β(x), for
every f ∈ A .
Consider now f ∈ A and suppose that x ∈ A0 and λ > 0 with λ > β(x).
Then, from Proposition 2.2.14(4), we have ( λ1 x)n → 0, for n → ∞. Hence, there
1
is n0 ∈ N, such that |f ( λ1n x n )| ≤ 1, ∀n ≥ n0 . Therefore, lim sup|f (x n )| n ≤ λ, for
n→∞
each f ∈ A and λ > β(x). Thus, β (x) ≤ β(x), for all x ∈ A. The same arguments
as above hold if in place of the functional f we have a seminorm p ∈ τ . Hence,
β (x) ≤ β(x), x ∈ A.
Next we show that β(x) ≤ β (x). For β (x) = +∞ this is trivial, so we suppose
that β (x) < +∞. In this case, for λ > 0 with λ > β (x), there is some n0 ∈ N,
such that for any
1 n
f ∈ A , |f ( x) | < 1, ∀ n ≥ n0 .
λ
Therefore, the set {( λ1 x)n : n ∈ N} is weakly bounded and thus τ -bounded by [128,
p. 67, Theorem 1]. Hence, λ ≥ β(x) and so β(x) ≤ β (x).
The result will be proven once we show that β (x) ≤ β (x). Indeed, if f ∈ A ,
then from the continuity of f there is some p ∈ τ and M > 0, such that
1 1 1
|f (x n )| n ≤ M n p(x n ) n , ∀ n ∈ N.
Therefore,
1 1
lim sup|f (x n )| n ≤ lim sup|p(x n )| n ≤ β (x),
n→∞ n→∞
so β (x) ≤ β (x).

2.3 Spectrum and Spectral Radius
The spectrum of an element x of an arbitrary algebra A, will be denoted by sp(x)

or by spA (x), when more than one algebra is involved; more precisely, if A has an
identity e
/ GA }.
sp(x) := {λ ∈ C : λe − x ∈
If A has no identity, then
spA (x) := spA1 (x).
Throughout this book C∗ will denote the one point compactification of C. A

partial algebraic structure is defined on C∗ as follows:
∞ + λ = ∞, λ ∈ C; ∞ · λ = ∞, λ ∈ C∗ \{0}; and ∞ = ∞.
Now, in case of a locally convex algebra A[τ ], G.R. Allan [4, Definition (3.1)]
defined a new spectrum of an element x ∈ A as follows:
Definition 2.3.1 Let A[τ ] be a locally convex algebra with an identity e. The
spectrum of an element x ∈ A, denoted by σ (x) (or by σA (x), when more than
one algebra is involved), is the subset of C∗ defined by

σA (x) = λ ∈ C : λe − x has no inverse in A0 ∪ ∞ ⇔ x ∈
/ A0 .
In case A has no identity then σA (x) := σA1 (x, 0).

Concerning the latter, since from Proposition 2.2.8(2), A[τ ] is pseudo-complete, if
and only if, A1 [τ1 ] is pseudo-complete, we can assume without loss of generality
that the locally convex algebra A[τ ] has an identity e.
The complement of σA (x) in C∗ is denoted by ρA (x) (or simply by ρ(x), if no
confusion arises) and is called the resolvent set of x.
If A[τ ] is a locally convex ∗-algebra, it is clear that

σA (x ∗ ) = λ : λ ∈ σA (x) , ∀ x ∈ A.
Proposition 2.3.2 Let A[τ ] be a locally convex algebra and x ∈ A. If C denotes a

maximal commutative subalgebra that contains x, then e ∈ C and σA (x) = σC (x).
Also if A[τ ] is pseudo-complete, then C[τC ] is pseudo-complete.
Proof Clearly the facts that e ∈ C and that C is closed follow from the assumption of
C being a maximal commutative subset of A. Therefore, from Proposition 2.2.8(1)
we have that C[τC ] is pseudo-complete if A[τ ] is supposed to be pseudo-complete.
/ σA (x) and λ = ∞, then (λe − x)−1 ∈ A0 . Since x ∈ C and C is maximal
If λ ∈
commutative it is straightforward that (λe − x)−1 ∈ C. So, (λe − x)−1 is a bounded
2.3 Spectrum and Spectral Radius 21
element in C[τC ]. Hence, λ ∈

/ σC (x). In case ∞ ∈
/ σA (x) then x ∈ A0 . Thus, x is
a bounded element in C, hence ∞ ∈ / σC (x). Therefore, σC (x) ⊂ σA (x). The other
inclusion is trivial.

Towards the development of a functional calculus for a pseudo-complete locally
convex algebra that we are going to describe in Sect. 2.4, the definition of the
following function is crucial.
Definition 2.3.3 Let A[τ ] be a locally convex algebra and x ∈ A. The resolvent of
x is the function R(x) defined by R(x)λ ≡ Rλ (x) = (λe − x)−1 , for all λ, such
that the inverse of λe − x is defined. When there is no danger of confusion with
respect to which element x the resolvent function is being considered, the symbol
Rλ is used instead of Rλ (x).
The resolvent of x is thus a function on a subset of C∗ , which takes values in A. We
recall that a function f on a subset of C∗ that takes values in a locally convex space
E[τ ] is holomorphic on some open set G of C if the limit lim f (λ)−fλ−λ0
(λ0 )
exists for
λ→λ0
every λ0 ∈ G. Moreover, f is holomorphic at ∞ if the limit lim f ( λ1 ) exists and the
λ→0
function
⎧
⎨f ( 1 ), if λ = 0
λ
g(λ) =
⎩ lim f ( λ1 ), if λ = 0
λ→0
is holomorphic in some neighbourhood of 0. In order to obtain some useful results

concerning the resolvent Rλ we are going to use the weak topology on A. The
reason behind this choice can be seen by the following direct observation: if f is
a holomorphic function that takes values in any locally convex algebra A[τ ] and
x ∈ A , then the function λ → x (f (λ)) is a complex valued holomorphic function.
For any locally convex algebra A[τ ] we have the following result concerning the
weak topology σ (A, A ) on A.
Lemma 2.3.4 Let A[τ ] be a locally convex algebra. Then, A is also a locally
convex algebra with respect to σ (A, A ).
Proof It suffices to show that the multiplication in A is separately continuous with
respect to σ (A, A ). Let f ∈ A , y ∈ A. Consider the maps fy , f y from A into
C, such that fy (x) = f (yx), f y (x) = f (xy), x ∈ A. Then, from the separate
continuity of multiplication in A, we have that fy , f y ∈ A , from which the result
is easily derived.

Lemma 2.3.5 Let A[τ ] be a locally convex algebra and x ∈ A. Suppose that R(x)
is weakly holomorphic at μ (= ∞). Then, R(x) has weak derivatives R (n) (x), n ∈
(n)
N, at μ, given by Rμ = (−1)n n!Rμn+1 , n ∈ N.
Proof We note that for λ, μ in the domain of R(x) we have that
Rλ − Rμ = (λe − x)−1 − (μe − x)−1

= (λe − x)−1 (μe − x) − (λe − x) (μe − x)−1
= −(λ − μ)Rλ Rμ .
Therefore, based on Lemma 2.3.4 and the assumption for R(x) being weakly
holomorphic, hence weakly continuous at μ we have that
Rλ − Rμ
= −Rλ Rμ → −Rμ2 , with respect to σ (A, A ).
λ−μ λ→μ
Hence, the result follows for n = 1. Let us suppose that the result holds for all
n = 1, 2, . . . , m. Then,
(m) (m)
Rλ − Rμ
= (−1)m m!(λ − μ)−1 (Rλm+1 − Rμm+1 )
λ−μ
= (−1)m m!(λ − μ)−1 (Rλ − Rμ )(Rλm + Rλm−1 Rμ + · · · + Rμm )

m
= (−1)m+1 m! Rλr+1 Rμm+1−r .
r=0
We show that R r+1 (x) is weakly continuous at μ, for r = 0, 1, . . . , m. Towards this

direction, given any f ∈ A consider the complex valued function ϕ(λ) = f (Rλ ),
which is holomorphic at μ by the assumption of the lemma. By the inductive
hypothesis and taking into account the continuity of f we have that ϕ (r) (λ) =
f (Rλ(r) ) for λ in a neighbourhood of μ and for r = 0, 1, . . . , m. The function ϕ (r)
is necessarily continuous at μ for r = 0, 1, . . . , m. Therefore, R r+1 (x) is weakly
continuous at μ for r = 0, 1, . . . , m. Hence, we have that
(m) (m)
Rλ − Rμ
→ (−1)m+1 (m + 1)!Rμm+2 , for λ → μ, with respect to σ (A, A ).
λ−μ
Thus, the result of the lemma is established by induction.

Lemma 2.3.6 Let A[τ ] be a locally convex algebra and x ∈ A, such that R(x) is
holomorphic at ∞ with respect to the weak topology σ (A, A ). If
⎧
⎨R 1 , λ = 0,
Sλ = λ
⎩ lim R 1 , λ=0
λ→0 λ
then, in some neighbourhood of 0, Sλ has weak derivatives of all orders given by

Sλ(n) = n!x n−1 (e − λx)−(n+1) , for n ∈ N and Sλ = λ(e − λx)−1 .
−1 −1
Proof For λ = 0 we have that Sλ = R 1 = λ1 e − x = λ e − λx . Let
λ
l be the weak limit of R 1 , as λ → 0 (note that the existence of l is ensured due
λ
to the assumption of R(x) being weakly holomorphic at ∞). Then, (e − λx)Sλ =
Sλ − x(λSλ ) and the right hand side of the previous equation will tend weakly to l,
as λ → 0, based on Lemma 2.3.4. On the other hand, we have that
1 −1
(e − λx)Sλ = (e − λx) e−x = λe → 0 as λ → 0.
λ
−1
Therefore, l = 0; hence, Sλ = λ e − λx holds true for λ = 0 too.
Now from the assumption that R(x) is weakly holomorphic at ∞ we have that
S(x) is weakly holomorphic in some 0-neighbourhood, say N (similar to R(x), the
notation S(x) denotes the function S(x)(λ) ≡ Sλ ). Then, we have that
1 1
Sλ − Sμ = R 1 − R 1 = − − R 1 R 1 , for λ = 0 = μ in N.
λ μ λ μ λ μ
Hence, for the weak first derivative of S(x), based on Lemma 2.3.4, we have that
Sλ − Sμ 1 1
−1 2 −2
Sμ = lim = 2 Sμ2 = 2 μ e − μx = e − μx .
λ→μ λ − μ μ μ
Therefore, the formula for the weak derivatives of S(x) in some 0-neighbourhood
holds for n = 1. Suppose the result holds for n = 1, . . . , m. Then we have that
Sλ(m) − Sμ(m) m!x m−1 1 m+1 1 m+1

= R − R
λ−μ λ − μ λm+1 λ 1
μm+1 μ 1
m!x m−1 1 1 m+1 1 m+1 m+1

= − R + R − R
λ−μ λm+1 μm+1 λ 1
μm+1 λ 1 1
μ
m!x m−1 μm + μm−1 λ + · · · + λm m+1

= μ−λ R1
λ−μ μm+1 λm+1 λ
1 1 1 1 −1
+ m!x m−1 − − R m+1
1 − R m+1
1
μm+1 λμ λ μ λ μ
−(m + 1)! m−1 m+1 1 1

→ x R 1 + m!x m−1
−
λ→μ μm+2 μ μm+1 μ2

· − (m + 1)R m+2
1
μ
x m−1 m+1 1 x m−1 m+1

= (m + 1)! R R 1 − e = (m + 1)! R xR 1
μm+2 μ1 μ μ μm+2 μ1 μ
1 −(m+2)
= (m + 1)!x m R m+2 = (m + 1)!x m e − μx .
μm+2 μ1
Therefore, the result follows by induction.

Theorem 2.3.7 Let A[τ ] be a locally convex algebra and x ∈ A. Let R(x) be the
resolvent of x. Then, the following hold:
(1) If R(x) is weakly holomorphic at μ, μ ∈ C∗ , then μ ∈ ρ(x).
(2) If μ ∈ ρ(x), then there is a neighbourhood N of μ and B ∈ B, such that, for
every λ ∈ N ∩ ρ(x), Rλ ∈ A[B]. Also, for μ = ∞, R(x) is differentiable at μ
relative to ρ(x) in the sense of norm convergence in A[B].
(3) If A[τ ] is pseudo-complete and μ ∈ ρ(x), then the neighbourhood N and the
set B ∈ B as in (2) can be chosen, such that N ⊂ ρ(x), Rλ ∈ A[B], for all
λ ∈ N and R(x) is holomorphic at μ in the sense of norm convergence in A[B].
Proof (1) Let R(x) be weakly holomorphic at μ = ∞. Then, there exists δ > 0,
such that R(x) is weakly holomorphic at λ ∈ C, for |λ−μ| < δ. If f ∈ A , consider
the function φ(λ) = f (Rλ ), λ ∈ C. Note that, φ is holomorphic at λ, for every λ,
such that |λ − μ| < δ. From Lemma 2.3.5 we have that φ (n) (μ) = f (Rμ(n) ) =
(−1)n n!f (Rμn+1 ), n ∈ N. Then, the Taylor expansion of φ about μ is given by
+∞

φ(λ) = (−1)k f (Rμk+1 )(λ − μ)k , for |λ − μ| < δ.
k=0
1
Therefore, from Cauchy’s radius-of-convergence formula lim sup|f (Rμn )| n ≤ 1
δ.
n
Hence, from Proposition 2.2.18 we have that β(Rμ ) ≤ 1δ . Thus, Rμ ∈ A0 (see
Proposition 2.2.14(2)), so μ ∈ ρ(x).
In case now R(x) is weakly holomorphic at ∞, we have
⎧
⎨R 1 , λ = 0
Sλ = λ
is weakly holomorphic at 0.
⎩ lim R 1 , λ=0
λ→0 λ
So, for f ∈ A , the function φ(λ) = f (Sλ ) is holomorphic at λ, |λ| < δ, for some
δ > 0. From Lemma 2.3.6 we have that φ (n) (0) = f (n!x n−1 ). Therefore, the Taylor
expansion of φ about 0 is given by
+∞

φ(λ) = f (x k−1 )λk , for |λ| < δ.
k=0
1
Hence, we have that lim sup|f (x n )| n ≤ 1
δ. Consequently, by Proposition 2.2.18,
n
β(x) ≤ 1δ , thus x ∈ A0 and so ∞ ∈ ρ(x).
For (2) and (3), we consider two cases below.
• Case μ = ∞. Let μ ∈ ρ(x), μ = ∞. Then, Rμ exists and Rμ ∈ A0 . From
Proposition 2.2.4 and the comment in , before Proposition 2.2.9, we have that
there is B ∈ B, such that Rμ ∈ A[B]. Clearly Rμ B > 0. Let λ ∈ C, with
|λ − μ| < Rμ1 B . Moreover, let sn denote the n-th partial sum of the series
Rμ − Rμ2 (λ − μ) + Rμ3 (λ − μ)2 − · · · (2.3.4)
For m > n we have that

m
sn − sm B ≤ Rμ B Rμ kB |λ − μ|k → 0, for n, m → +∞.
k=n+1
Hence, (sn )n∈N forms a Cauchy sequence in A[B]. Since τ ≺ · B on A[B], for
every seminorm p ∈ τ , there is Cp > 0, such that xB ≤ Cp p(x), x ∈ A[B].

Therefore, for every p ∈ τ and λ, such that λ ∈ N = z ∈ C : |z−μ| < Rμ1 B
we have
p (λe − x)sn − e = p (λ − μ)sn + (μe − x)sn − e

= p e + (−1)n (λ − μ)n+1 Rμn+1 − e (2.3.5)
≤ Cp |λ − μ|n+1 Rμ n+1

B → 0, for n → +∞.
So, lim (λe − x)sn = e and similarly lim sn (λe − x) = e, with respect to
n n
τ . Therefore, for λ ∈ N ∩ ρ(x), we have that sn → Rλ , with respect to τ .
Furthermore, since (sn )n∈N is a · B -Cauchy sequence in A[B] there is M > 0,
such that sn B ≤ M for all n ∈ N. Hence, M 1
sn ∈ B, for all n ∈ N. Then,
since M sn → M Rλ , with respect to τ , and given that B is τ -closed, we conclude
1 1
that Rλ ∈ A[B]. Then, we have that sn → Rλ , with respect to · B : indeed, by

analogous computations as those of (2.3.5), we have that for λ ∈ N, (λe−x)sn −
eB = |λ−μ|n+1 Rμ n+1 B → 0, as n → ∞. Therefore, (λe−x)(sn −Rλ )B →
0, for n → ∞. Hence,
sn − Rλ B = Rλ (λe − x)(sn − Rλ )B ≤ Rλ B (λe − x)(sn − Rλ )B → 0,
as n → ∞. It is clear then from the series in (2.3.4) that R is differentiable at μ

relative to ρ(x) in the sense of norm convergence in A[B].
If, in addition, A[τ ] is pseudo-complete, then A[B] is a Banach algebra.

Hence, for λ ∈ N, the series in (2.3.4) is necessarily convergent, so Rλ exists
and belongs to A[B]. Therefore, λ ∈ ρ(x), so that N ⊂ ρ(x).
• Case μ = ∞. Suppose that ∞ ∈ ρ(x). Then, x ∈ A0 , so there is some B ∈ B,
such that x ∈ A[B]. Let N = {λ ∈ C∗ : |λ| > xB }. For λ ∈ N ∩ ρ(x),
using similar arguments just as in the previous case, we have
that−n Rλ ∈ A[B].
Moreover, for λ = ∞, λ ∈ N ∩ ρ(x), we have that Rλ = +∞ n=1 λ x n−1 , with
respect to norm convergence in A[B]. If A[τ ] is pseudo-complete, then Rλ exists
and belongs to A[B] for λ ∈ N. Moreover, for λ ∈ N, λ = ∞, R is holomorphic
at λ and
+∞
1
Rλ B ≤ λ−n x n−1 B = .
|λ| − xB
n=1
So, Rλ B → 0, for λ → ∞. It is then easily deduced that R is holomorphic at

∞.

Based on the previous proposition we can now establish the following corollaries.
Corollary 2.3.8 Let A[τ ] be a locally convex algebra and x ∈ A. Then, σ (x) = ∅.
If A[τ ] is pseudo-complete, then σ (x) is closed.
Proof Let us suppose that σ (x) = ∅. Then from Theorem 2.3.7(2) we have that R
is holomorphic on the whole complex plane C. Also x ∈ A0 since ∞ ∈ ρ(x). Thus,
by using the same argument just as at the end of the proof of Theorem 2.3.7, we
have that Rλ → 0, for λ → ∞. Then from Liouville’s theorem, as this is applied
in locally convex spaces, we have that Rλ = 0, a contradiction. Moreover, in case
A[τ ] is pseudo-complete, from Theorem 2.3.7(3) we obtain that σ (x) is closed.

The following corollary extends to the locally convex case the Gelfand–Mazur
theorem.
Corollary 2.3.9 Let A[τ ] be a locally convex algebra, such that, for every x ∈ A,
there is some nonzero λ ∈ C, such that (λx)n → 0. If moreover A is a division
algebra, then A is topologically isomorphic to C.
Proof We consider the map φ : C → A[τ ] : λ → λe. It is clear that φ is a
topological isomorphism onto its image. Based on the assumption that for every
x ∈ A there is a nonzero λ ∈ C, such that (λx)n → 0, we derive that A = A0 .
Then, by Corollary 2.3.8, we have that σ (x)∩C = ∅. For λ ∈ σ (x)∩C, the element
λe − x has no inverse in A0 = A. Hence, since A is a division algebra, λe = x and
thus φ is onto. Therefore, the result follows.

Lemma 2.3.10 Let A[τ ] be a pseudo-complete locally convex algebra, x ∈ A and
K a compact subset of ρ(x). Then, there is some B ∈ B, such that Rλ ∈ A[B], for
every λ ∈ K.
Proof Let C be a maximal commutative subalgebra of A that contains x. Being

maximal commutative subalgebra, C is closed and e ∈ C. Since C is a closed subal-
gebra of A[τ ], C[τ C ] is pseudo-complete by Proposition 2.2.8(1). Furthermore, by
Proposition 2.3.2, we have that ρ(x) ≡ ρA (x) = ρC (x), hence for each λ ∈ ρ(x),
Rλ ∈ C. Let now λ ∈ K. From Theorem 2.3.7(3) there is a neighbourhood Nλ
of λ and a set B ∈ B, such that Rμ ∈ A[B] ∩ C, with μ ∈ Nλ . Since K is
compact there are finitely many points λ1 , . . . , λn ∈ K, such that the corresponding
neighbourhoods Nλ1 , . . . , Nλn cover K. Let B1 , . . . , Bn be the respective subsets in
B, such that Rμ ∈ A[Bi ] ∩ C, for i = 1, . . . , n, where μ ∈ Nλi . Since C is closed
and e ∈ C, it is straightforward to show that B ∩ C ∈ BC , for every B ∈ B. Hence,
taking into account pseudo-completeness of C, we have that A[B ∩ C] = A[B] ∩ C
is a Banach algebra with respect to · B , for every B ∈ B. Therefore, by using
analogous arguments as those in the proof of Theorem 2.2.10, it can be shown that
the family {B ∩C : B ∈ B} is outer-directed by inclusion. So, there is some B ∈ B,
such that
(B1 ∩ C) ∪ (B2 ∩ C) ∪ · · · ∪ (Bn ∩ C) ⊂ B ∩ C.
Therefore, Bi ∩ C ⊂ B ∩ C, for i = 1, . . . , n. Hence, Rλ ∈ A[B] ∩ C, for every

λ ∈ K.

If A is an arbitrary algebra and x ∈ A, the spectral radius of x will be denoted
by r(x), or for distinction by rA (x), with

r(x) := sup |λ| : λ ∈ sp(x) .
In the case of a locally convex algebra A[τ ], considering σ (x) in the place
of sp(x), we use the same symbols and the same formula for the spectral
radius of an element x in A[τ ], with the convention |∞| = +∞.
The relation between the spectral radius and the radius of boundedness, in a
locally convex algebra, is given by the following
Theorem 2.3.11 Let A[τ ] be a locally convex algebra and x ∈ A. Then, β(x) ≤
r(x). In case A[τ ] is pseudo-complete, then β(x) = r(x).
Proof Let r(x) < +∞, for otherwise the inequality is trivial. Then, ∞ ∈
/ σ (x)
and so x ∈ A0 . Hence, from Proposition 2.2.4, there is some B ∈ B, such that
x ∈ A[B]. As in the proof of Theorem 2.3.7(2) we have that if
N = {μ ∈ C : |μ| > xB }, then Rλ = λ−1 e + λ−2 x + · · · , for λ ∈ ρ(x) ∩ N,

with respect to norm convergence in A[B]. Then, for f ∈ A the function φ(λ) =
f (Rλ ) is written as follows:
φ(λ) = λ−1 f (e) + λ−2 f (x) + · · · , for λ ∈ N ∩ ρ(x). (2.3.6)
Moreover, by Theorem 2.3.7, φ is holomorphic at λ ∈ C with |λ| > r(x). So, φ has
a Laurent expansion in that region, which must coincide with the series in (2.3.6).
1
Thus, we have that lim sup|f (x n )| n ≤ r(x) and so by Proposition 2.2.18, β(x) ≤
n
r(x), x ∈ A.
Let us assume now that A[τ ] is pseudo-complete. We show that r(x) ≤ β(x),
x ∈ A. Suppose that β(x) < +∞, for otherwise the inequality is trivial. Then,
by Proposition 2.2.14(2), x ∈ A0 . Let λ ∈ C, such that |λ| > β(x). By
Corollary 2.2.17, there exists B ∈ B, such that x ∈ A[B] and |λ| > xB . Then,
as in the proof of Theorem 2.3.7(3), we conclude that Rλ exists in A[B] and so
λ ∈ ρ(x). Therefore, for every μ ∈ σ (x), we have that |μ| ≤ β(x), from which the
result follows.

In case of a pseudo-complete locally convex algebra A[τ ], the following result
provides us with a relation between the spectrum σ (x) of an element x ∈ A0 and
the spectra σA[B] (x) = spA[B] (x), for those B ∈ B, such that x ∈ A[B].
Proposition 2.3.12 Let A[τ ] be a pseudo-complete locally convex algebra and x ∈
A0 . Then, the following hold:

(1) σA (x) = σA[B] (x) : B ∈ B, x ∈ A[B] ;
(2) rA (x) = inf rA[B] (x) : B ∈ B, x ∈ A[B] .
Proof For the proof of both claims let C denote a maximal commutative subalgebra
of A containing x.
(1) Since A[B] ⊂ A, for every B ∈ B, we have that

σA (x) ⊂ σA[B] (x) : B ∈ B, x ∈ A[B] .
Now let λ(= ∞), such that λ ∈/ σA (x). Then, Rλ ∈ A0 , so there is a B1 ∈ B, such
that Rλ ∈ A[B1 ] ∩ C (Proposition 2.2.4). By the outer-directedness of {B ∩ C : B ∈
B} (see Theorem 2.2.10) we have that there exists a B ∈ B, such that Rλ ∈ A[B]∩C
and x ∈ A[B]. Hence, λ ∈ / σA[B] (x) and so the inverse inclusion follows.
(2) From (1) we have that

rA (x) ≤ inf rA[B] (x) : B ∈ B, x ∈ A[B] .

Consider μ ∈ C, such that μ > rA (x) and let K = λ ∈ C∗ : |λ| ≥ μ . Then, K
is a compact subset of ρ(x) and so from Lemma 2.3.10 we have that there is some
B ∈ B, such that Rλ ∈ A[B], for all λ ∈ K. By using the same argument as in (1)
we can assume that x ∈ A[B]. Then, we have that rA[B] (x) ≤ μ. Hence,

inf rA[B] (x) : B ∈ B, x ∈ A[B] ≤ μ, ∀ μ > rA (x),
from which the inverse inequality follows.

For a locally convex algebra A[τ ] in which inversion is continuous on the invertible
elements of A, the relation between σ (x) and sp(x), x ∈ A (for the respective
definitions, see beginning of Sect. 2.3) is described by the following result. In the
notation of the following theorem sp(x) stands for the closure of sp(x) in C∗ .
Theorem 2.3.13 Let A[τ ] be a locally convex algebra with continuous inversion
and let x ∈ A. Then, sp(x) ⊂ σ (x) ⊂ sp(x). Moreover, if A[τ ] is pseudo-complete,
then σ (x) = sp(x).
Proof It is clear from the respective definitions that sp(x) ⊂ σ (x). For the second
inclusion, assume that sp(x) = C∗ , for otherwise the result is trivial and let μ ∈ C,
such that μ ∈/ sp(x). Then, there is some δ > 0, such that if λ ∈ C with |λ − μ| < δ,
then λ ∈/ sp(x). Under the assumption of continuity of inversion on the invertible
elements of A, we have that R(x) is continuous at μ. Then, since
Rλ − Rμ = −(λ − μ)Rλ Rμ
(see beginning of the proof of Lemma 2.3.5), the function R(x) is differentiable at
μ, for every finite point μ in the complement of sp(x) in C∗ .
Now if ∞ ∈ / sp(x), then there is M > 0, such that if λ = ∞ and |λ| > M, then
λ∈/ sp(x). For these λ’s we then have that
Rλ = λ−1 (e − λ−1 x)−1 → 0, as λ → ∞.
It follows that R is holomorphic at ∞. Therefore, R is holomorphic on C∗ \ sp(x).

Thus, from Theorem 2.3.7(1) we have that σ (x) ⊂ sp(x).
In case A[τ ] is pseudo-complete, then by Corollary 2.3.8, σ (x) is closed. Hence,
sp(x) ⊂ σ (x), thus the desired equality follows.

Corollary 2.3.14 Let A[τ ] be a pseudo-complete locally convex algebra with
continuous inversion. Then, x ∈ A0 , if and only if, sp(x) is bounded.
Proof By Theorem 2.3.13, sp(x) = σ (x). Let x ∈ A0 and assume that sp(x) is not
bounded. Then, there exists a sequence (λn )n∈N in sp(x), such that λn → ∞. So,
∞ ∈ sp(x) = σ (x), hence x ∈ / A0 , a contradiction. For the reverse implication,
assume that sp(x) is bounded. Then, sp(x) is bounded [131, p. 25, 5.1]. So, ∞ ∈ /
sp(x) = σ (x) and thus x ∈ A0 .

2.4 A Functional Calculus
In this section we develop a functional calculus for a pseudo-complete locally

convex algebra. Let A be a pseudo-complete locally convex algebra and x ∈ A. Let
us denote with Fx the set of all complex-valued functions, which are holomorphic
on the spectrum σ (x) of x, hence on some neighbourhood of σ (x). By Fx we denote
the quotient set of Fx by the equivalence relation ∼, which is given as follows: for
f, g ∈ Fx , f ∼ g, if and only if, f equals g on some neighbourhood of σ (x). With
the algebraic operations defined pointwise on suitable neighbourhoods of σ (x), it
follows that Fx is an algebra.
We recall that in case A is a Banach algebra, it is known that for f ∈ Fx and γ a
closed rectifiable Jordan curve, such that its interior domain, say D, contains σ (x)
and f is holomorphic on D and continuous on D ∪ γ , then the formula

1
f (x) = f (λ)Rλ (x)dλ
2πi γ
defines a homomorphism f → f (x) of Fx into A, which enjoys certain properties

(see [115, Chapter III, §11.6, Theorem 7]).
An extension of this result can be established for the case where A[τ ] is a pseudo-
complete locally convex algebra. Towards this direction we first show the following
Proposition 2.4.1 Let A[τ ] be a pseudo-complete locally convex algebra and x ∈
A with ρ(x) = ∅. Let γ be any rectifiable Jordan arc in ρ(x) ∩ C and f a complex-
is some B ∈ B, such that
valued function, which is holomorphic on γ . Then, there
Rλ ∈ A[B], for all λ ∈ γ . Furthermore, the integral γ f (λ)Rλ (x)dλ exists, in the
sense of norm convergence in A[B] and its value is an element of A[B].
Proof Since the set γ is a compact subset of ρ(x), by Lemma 2.3.10 we have that
there is some B ∈ B, such that Rλ ∈ A[B], for all λ ∈ γ . Therefore, the function
λ → f (λ)Rλ is a holomorphic function on γ which takes values in the Banach
algebra A[B]. Then the result follows from the holomorphic functional calculus for
Banach algebras (see [115, Chapter I, § 4.7, Theorem I]).

The following definition is a slight variant of [145, Definition, p. 193].
Definition 2.4.2 A subset D of C∗ is called a Cauchy domain if it fulfills the
following conditions:
(1) D is open;
(2) D has a finite number of components, whose closures are pairwise disjoint;
(3) the boundary ∂D of D is a subset of C, which consists of a finite number of
closed rectifiable Jordan curves no two of which intersect.
The next result can be obtained with the use of Proposition 2.4.1 and by following
very similar arguments to those developed in [145, Theorem 4.1] and so its proof is
omitted.
2.4 A Functional Calculus 31
Lemma 2.4.3 Let A[τ ] be a pseudo-complete locally convex algebra and x ∈ A,

such that ρ(x) = ∅. Then for any f ∈ Fx there is a Cauchy domain D, such that
(i) σ (x) ⊂ D;
⊂ (f ), where (f ) denotes the domain of f . Furthermore, the integral
(ii) D
∂D f (λ)Rλ (x)dλ defines an element of A0 , which is independent of the choice
of the Cauchy domain D satisfying (i) and (ii).
We are now in position to establish a functional calculus for a pseudo-compete
locally convex algebra.
Theorem 2.4.4 Let A[τ ] be a pseudo-complete locally convex algebra and x ∈ A.
Then, there is a homomorphism f → f (x) from Fx into A0 , which is given by the
following formulae:

(1) if x ∈ A0 , then f (x) = 2πi
1
∂D f (λ)Rλ (x)dλ, whereD is as in Lemma 2.4.3;
/ A0 and ρ(x) = ∅, then f (x) = f (∞)e + 2πi
(2) if x ∈ 1
∂D f (λ)Rλ (x)dλ, where
D is as before;
(3) if ρ(x) = ∅, then Fx contains only constant functions. If f (λ) ≡ c, then f (x) =
ce.
Furthermore, for all cases, if C is a maximal commutative subalgebra of A, which
contains x, then f (x) ∈ A0 ∩ C.
If u0 , u1 denote the complex functions u0 (λ) ≡ 1, u1 (λ) ≡ λ, then in case (1),
u1 ∈ Fx and u1 (x) = x and in all cases (1)–(3), u0 ∈ Fx and u0 (x) = e.
Proof For the proof of (1) and (2) are used standard arguments that follow from
similar arguments to those in the proof of [145, Theorem 4.3].
For (3), we have that σ (x) = C∗ . Then, any f ∈ Fx is holomorphic on the
whole complex plane and at ∞. Therefore, by Liouville’s theorem as is applied to
vector-valued functions (see [115, Chapter I, §3.12]), we have that Fx consists only
of constant functions.
Moreover, with respect to the fact that f (x) ∈ C ∩ A0 , where C is a maximal
commutative subalgebra of A containing x, this follows directly from the fact that
e ∈ C and that (λe − x)−1 commutes with all elements of C, for every λ ∈ ∂D.
Now if u1 (λ) = λ and x is in A0 , then taking into account Proposition 2.3.12(1)
we can choose D, such that its boundary ∂D is a circle of radius greater than xB ,
where B is some element of B with x ∈ A[B]. Then, for λ ∈ ∂D we have that
−1 −1
(λe − x)−1 = e − λ−1 x λ = λ−1 e + λ−2 x + · · ·
Hence,

1 1
u1 (x) = λ(λe − x)−1 dλ = e + λ−1 x + λ−2 x 2 + · · · dλ = x.
2πi ∂D 2πi ∂D
The statement about u0 follows from similar considerations.

2.5 The Carrier Space
Consider a commutative and pseudo-complete locally convex algebra A[τ ] with

an identity e. Recall that B denotes the collection of subsets of A just as the one
described in Proposition 2.2.8(3). It will be convenient for the purposes of this
section to denote the elements of B by {Bα : α ∈ }, where is an index set.
For α, β ∈ , we use the notation, α ≤ β if Bα ⊂ Bβ . Then, by Theorem 2.2.10 the
index set is outer-directed by the ordering ≤. To simplify the notation, for each
α ∈ , we denote by Aα the Banach algebras A[Bα ]. Let Mα denote the set of all
nonzero multiplicative linear functionals on Aα endowed with the weak* topology
σ (Mα , Aα ).
The respective set of all nonzero multiplicative linear functionals on A0 , denoted
by M0 , and endowed with the weak* topology σ (M0 , A0 ) is called the carrier
space (or Gelfand space, or maximal ideal space) of A0 .
Proposition 2.5.1 There exists a natural homeomorphism, say j , of the carrier
space M0 with the projective limit lim Mα , where j (ϕ) = (ϕα )α∈ with ϕα (x) =
← −
α∈
ϕ(x), x ∈ Aα , ϕ ∈ M0 , α ∈ .
Proof For α, β ∈ with α ≤ β consider the map παβ : Mβ → Mα , such that
παβ (ϕβ ) = ϕβ |Aα . The maps παβ are well-defined, that is ϕβ |Aα is not zero, since
e ∈ Aα . It is also clear that παβ are weak ∗-continuous maps, such that παα is
the identity map on Mα , for all α ∈ and παβ ◦ πβγ = παγ for α ≤ β ≤ γ
in . Therefore, the spaces {Mα : α ∈ } form a projective system. Since for
each α ∈ , Aα is a commutative Banach algebra with identity, Mα is a non-
empty compact Hausdorff space. Hence, the projective limit lim Mα is a non-empty
←−
α∈
compact Hausdorff space (see [35, Chapter I, §9, No. 6, Proposition 8]). The fact
that the map j is a homeomorphism is then easily checked.

A direct implication of the previous proposition is that the carrier space

M0 of A0 is a non-empty compact Hausdorff space.
Lemma 2.5.2 Let A[τ ] be a commutative and pseudo-complete locally convex

algebra and x ∈ A0 . Then, x is invertible in A0 , if and only if, ϕ(x) = 0, for
every ϕ ∈ M0 .
Proof The forward implication is immediate since ϕ(e) = 1, for all ϕ ∈ M0 . For
the inverse implication let x ∈ A0 and suppose that ϕ(x) = 0, for every ϕ ∈ M0 . We
show that there is α ∈ , such that x ∈ Aα and ϕα (x) = 0, for all ϕα ∈ Mα . From
a well-known result in the theory of Banach algebras it will then follow that x is
invertible in Aα and thus in A. Let us suppose to the contrary that, for every α ∈ ,
such that x ∈ Aα , there is ϕα ∈ Mα with ϕα (x) = 0. Let us fix an index δ ∈ , such
2.5 The Carrier Space 33
that x ∈ Aδ and for every α ≥ δ, let Nα denote the set {ϕα ∈ Mα : ϕα (x) = 0}.
From the assumption we have made, Nα = ∅, for each α ≥ δ. Also Nα , α ≥ δ, is
closed, hence a compact subspace of Mα . Since παβ (Nβ ) ⊂ Nα , for all β ≥ α ≥ δ,
we have that {Nα }α≥δ forms a projective system, whose projective limit lim Nα is a
←−
α≥δ
non-empty compact Hausdorff space. Let {ψα }α≥δ be an element in limNα . Define
← −
α≥δ
{ϕα }α∈ ∈ α∈ Mα as follows

ψα , if α ≥ δ
ϕα =
παβ (ψβ ), otherwise, for some β ≥ α, β ≥ δ.
It is clear that ϕα , α ∈ , is well-defined since if α δ and β, β ∈ , such that

β, β ≥ α, then there exists β ∈ with β ≥ β, β and so
παβ (ψβ ) = παβ (πβ β (ψβ )) = παβ (ψβ ) = παβ (πββ (ψβ )) = παβ (ψβ ).
Moreover, {ϕα }α∈ ∈ lim Mα as can easily be verified. So, based on Proposi-
←−
α∈
tion 2.5.1 there is ϕ ∈ M0 , such that ϕ(x) = ϕδ (x) = ψδ (x) = 0, a contradiction.
Thus, the result follows.

Theorem 2.5.3 Let A[τ ] be a commutative and pseudo-complete locally convex
algebra and x an element in A0 . Then, σ (x) = {ϕ(x) : ϕ ∈ M0 }.
Proof Since x ∈ A0 we have that ∞ ∈ / σ (x). Then, λ ∈ C belongs to σ (x), if
and only if, λe − x has no inverse in A0 . From Lemma 2.5.2 this is equivalent to
ϕ(λe − x) = 0, for some ϕ ∈ M0 , that is ϕ(x) = λ. Hence, the result follows.
The next result describes the unique extension of any functional ϕ ∈ M0 to a bigger
set, namely to Aρ := {x ∈ A : ρ(x) = ∅}.
Proposition 2.5.4 Let A[τ ] be a commutative and pseudo-complete locally convex
algebra. Then, to each functional ϕ ∈ M0 corresponds a unique C∗ -valued function
ϕ on Aρ , such that the following hold:
(1) ϕ is an extension of ϕ.
The C∗ -valued function ϕ on Aρ is a ‘partial character’ of A, in the following
sense:
(2) ϕ (λx) = λϕ (x), λ ∈ C, x ∈ Aρ , with the convention that 0 · ∞ = 0;
(3) ϕ (x1 + x2 ) = ϕ (x1 ) + ϕ (x2 ), provided that x1 , x2 , x1 + x2 ∈ Aρ and
ϕ (x1 ), ϕ (x2 ) are not both ∞;
(4) ϕ (x1 x2 ) = ϕ (x1 )ϕ (x2 ), provided that x1 , x 2 , x1 x2 ∈ Aρ and ϕ (x1 ), ϕ (x2 )
are not 0, ∞ in some order.
Proof
(1) Let x ∈ Aρ and μ ∈ ρ(x), such that μ = ∞. Let y = (μe − x)−1 ∈ A0
and consider ϕ ∈ M0 . If an extension, say ϕ , of ϕ to Aρ satisfying properties
(2)–(4) is possible, then provided that ϕ(y) = 0, ϕ must satisfy the relation
ϕ (μe − x)ϕ(y) = ϕ(e) = 1 ⇒ ϕ (x) = μ − ϕ(y)−1 . (2.5.7)
If ϕ(y) = 0, then by (3),(4) we have that ϕ (x) = ∞. So, the equality ϕ (x) =
μ − ϕ(y)−1 holds in any case and it is considered as the definition of ϕ .
The definition of ϕ is independent from the choice of μ ∈ C ∩ ρ(x). Indeed
let μ1 , μ2 ∈ C∩ρ(x). If either of the elements ϕ (μ1 e−x)−1 , ϕ (μ2 e−x)−1
is 0, then the other must be 0 also, as it follows from the relation Rμ1 − Rμ2 =
(μ2 −μ1 )Rμ1 Rμ2 (see beginning of the proof of Lemma 2.3.5). If both elements
are not 0, then
1 1 ϕ(Rμ1 − Rμ2 )
− =−
ϕ (μ1 e − x)−1 ϕ (μ2 e − x)−1 ϕ(Rμ1 )ϕ(Rμ2 )
(μ1 − μ2 )ϕ(Rμ1 Rμ2 )
=
ϕ(Rμ1 )ϕ(Rμ2 )
= μ1 − μ2 ,
hence μ1 − ϕ((μ e−x)1

−1 ) = μ2 − ϕ((μ e−x)−1 ) . The previous argumentation
1
1 2
establishes also the uniqueness of any possible extension of ϕ to Aρ , satisfying
conditions (2)–(4).
Moreover, ϕ is an extension of ϕ. Indeed if x ∈ A0 , then from ϕ(μe −
x)ϕ (μe − x)−1 = 1 we have that ϕ (μe − x)−1 = 0, therefore
1 ϕ μ(μe − x)−1 − e
ϕ (x) = μ − =
ϕ (μe − x)−1 ϕ((μe − x)−1 )
ϕ x(μe − x)−1
= = ϕ(x).
ϕ((μe − x)−1 )
(2) Let x ∈ Aρ and μ ∈ ρ(x) ∩ C, such that ϕ (μe − x)−1 = 0. If λ ∈ C, λ = 0,

then λμ ∈ ρ(λx), hence

1 1
ϕ (λx) = λμ − = λ μ − = λϕ (x).
ϕ((λμe − λx)−1 ) ϕ((μe − x)−1 )
If λ = 0, then clearly λϕ (x) = 0 = ϕ (λx). By using similar arguments for the

case ϕ (μe − x)−1 = 0, we derive the result.
2.5 The Carrier Space 35
(3) Let x1 , x2 , x1 + x2 ∈ Aρ , such that ϕ (x1 ), ϕ (x2 ) are not both ∞. Consider
μ1 ∈ ρ(x1 ) ∩ C, μ2 ∈ ρ(x2 ) ∩ C, λ ∈ ρ(x1 + x2 ) ∩ C. We have that

ϕ Rλ (x1 + x2 ) (λ − μ1 − μ2 )ϕ Rμ1 (x1 ) ϕ Rμ2 (x2 )

+ ϕ Rμ1 (x1 ) + ϕ Rμ2 (x2 )
= ϕ Rλ (x1 + x2 ) ϕ Rμ1 (x1 ) (λ − μ1 − μ2 )e
(2.5.8)
+ μ2 e − x2 + μ1 e − x1 Rμ2 (x2 )
= ϕ Rλ (x1 + x2 ) λe − (x1 + x2 ) Rμ1 (x1 )Rμ2 (x2 )
= ϕ Rμ1 (x1 ) ϕ Rμ2 (x2 ) ,
where the last to one equality is derived due to commutativity of A.

Therefore, if both ϕ Rμ2 (x2 ) , ϕ Rμ1 (x1 ) are different from 0, then
by (2.5.8) we have that ϕ Rλ (x1 + x2 ) = 0 and
1 1 1
= λ − μ1 − μ2 + + ,
ϕ Rλ (x1 + x2 ) ϕ Rμ1 (x1 ) ϕ Rμ2 (x2 )
so that (2.5.7) implies ϕ (x1 + x2 ) = ϕ (x1 ) + ϕ (x2 ).

Also if one of ϕ(Rμ1 (x1 )) or ϕ(Rμ2 (x2 )) is 0, then from (2.5.8) we have that
ϕ Rλ (x1 + x2 ) = 0. Thus, ϕ (x1 + x2 ) = ∞ = ϕ (x1 ) + ϕ (x2 ).
(4) Let x1 , x2 , x1 x2 ∈ Aρ . Consider μ1 ∈ ρ(x1 ) ∩ C, μ2 ∈ ρ(x2 ) ∩ C and λ ∈
ρ(x1 x2 ) ∩ C. Then, we have

ϕ Rλ (x1 x2 ) (λ − μ1 μ2 )ϕ Rμ1 (x1 ) ϕ Rμ2 (x2 )

+ μ1 ϕ Rμ1 (x1 ) + μ2 ϕ Rμ2 (x2 ) − 1

= ϕ Rλ (x1 x2 ) λϕ Rμ1 (x1 )Rμ2 (x2 )

− ϕ (μ1 Rμ1 (x1 ) − e)(μ2 Rμ2 (x2 ) − e)

= ϕ Rλ (x1 x2 ) λϕ Rμ1 (x1 )Rμ2 (x2 ) − ϕ Rμ1 (x1 )x1 x2 Rμ2 (x2 )

= ϕ Rλ (x1 x2 ) ϕ (λe − x1 x2 )Rμ1 (x1 )Rμ2 (x2 )
= ϕ Rμ1 (x1 ) ϕ Rμ2 (x2 ) .

(2.5.9)
Suppose that ϕ Rμ1 (x1 ) = 0 and ϕ Rμ2 (x2 ) = 0. The previous two relations,
due to the very definition of ϕ , result equivalently in that ϕ (x1 ) = ∞ and
Another random document with
no related content on Scribd:
Sir William Berkeley, Governor of Virginia, had an especial interest
in the mountains and the country beyond. As agent in America for
the Hudson’s Bay Company he hoped to help break the French
monopoly on the hinterland trade by exploiting the western territory
from Virginia. As early as 1669 he sent out John Lederer, a German,
who ranged the eastern slopes of the Blue Ridge, north and south,
for many miles. Even before that, Abraham Wood and other traders
with commissions from the governor had bartered far to the
southwest among the headwaters of Carolina coastal rivers.
They transported their wares on pack horses, 150 to 200 pounds
on each animal, making twenty miles or more a day on their journeys
when forage was plentiful. With guns, powder and shot as prime
trade goods they visited tribes who had previously bartered with the
Spaniards of Florida. They took hatchets, kettles, iron tools, colorful
blankets and a variety of trinkets to villages never before visited by
white men. On these occasions the appearance of that strange
animal, the horse, strung with tinkling bells and packing unbelievable
wealth on his back, created more awe among the savages than the
bearded white man himself.
In 1671 the Virginians crossed the southerly ridges into the New
River valley, and in another two years young Gabriel Arthur opened
commerce with the Cherokees in the terminal hills of the
Appalachians. He and his partner, James Needham, had some
extraordinary experiences. Needham, a much older man of some
experience in the Indian trade, was murdered by the savages on this
venture. Arthur himself escaped burning only through the
intervention of a Cherokee chief who, during the midst of the torture,
adopted him into the tribe.
The Cherokee chief dressed and armed Arthur as a brave and
sent him out with raiding war parties. In the first such instance, the
Virginian seems to have joined willingly enough in a murderous
surprise attack on a Spanish mission settlement in West Florida. In
another, he helped slaughter some sleeping native villagers one
night in the vicinity of Port Royal, South Carolina, on the promise of
the Cherokees that no Englishmen in those parts would be harmed
during the raid. Arthur later said he could tell that one English family
was celebrating Christmas when his war party crept by their hut.
In still another instance, Arthur went all the way to the banks of the
Ohio with his Cherokee chief to attack a Shawnee village. There he
was badly wounded and captured, but released with some reverence
when he scrubbed himself and exhibited his white skin to the
amazed savages. After making his way back to the country of his
Cherokee friends, the young Virginian finally returned to his own kind
on the James River, richly laden with furs and trade treaties.
Henry Woodward, Carolina’s resourceful pioneer, found evidence
of the Virginians’ trade on the backside of Lord Ashley’s proprietary
in 1674. Woodward, who saved the fledgling colony at Charles
Towne from bankruptcy by developing a trade in pelts and skins with
the hinterland savages, visited the palisaded village of the Westoes
that year. There, high up the Savannah River, he found the natives
already “well provided with arms, ammunition, tradeing cloath &
other trade from ye northward for which at set times of ye year they
truck drest deare skins furrs & young Indian slaves.”
Governor Berkeley’s traders were indeed carrying on a highly
profitable commerce. So much so, that in the interests of those
profits, it was claimed, the governor permitted favored hinterland
tribes to pillage Virginia tobacco planters with impunity. In any case
Berkeley, who operated gainfully in his capacity as a British fur
factor, did not respond with enough enthusiasm to the planters’
demands for protection, and a civil war resulted in 1676 that set back
the colony’s economy by years. The rebellion was led by a fiery,
twenty-nine-year-old patriot named Nathaniel Bacon. Before he died
suddenly of a camp malady, Bacon chased the governor across the
Chesapeake Bay to the Eastern Shore and burned Jamestown, the
capital of Virginia, to the ground. With Bacon’s death the revolt
collapsed and twenty-three prominent insurgent leaders were
hanged by the governor in an orgy of personal revenge.
But, if Governor Berkeley had won the war over the fur trade, it
was a merchant at the Falls of the James River who prospered most.
There, at his store, William Byrd maintained a fine stock of calico,
red coats, beads, knives, guns and Barbadian brandy for the pack-
traders who sought out beaver pelts among remote Indian villages in
the interior. So successful was Byrd that by the early 1680’s he
dominated the hinterland trading paths of Virginia and Carolina.
From this commerce he created the fortune that bought enough
slaves and tobacco lands to promote his family to a position among
the wealthiest in the colony, while the great hogsheads of pelts that
he shipped yearly down the James River to England contributed in
no small way to the support of Britain’s growing empire.
Henry Woodward and his Carolinians driving straight west avoided
the trading paths of the Virginians, as well as the Appalachian
Mountains, to invade the preserves of Spanish Florida. This took
them to the headwaters of rivers emptying into the Gulf of Mexico, to
the villages of the Creeks, where the Spaniards had previously
monopolized the trade in deerskins and Indian slaves. The
Carolinians diverted much of this profitable commerce to newly
located Charleston. Thousands upon thousands of deerskins were
shipped yearly to England, to be manufactured into a variety of
articles. Hundreds of Indian slaves were supplied to New England
and Virginia, and to Barbados where the rate of mortality on the hot
sugar plantations insured a steady demand.
Spanish resistance in the south, the extinction of deer and the
elimination of whole tribes of Indians who succumbed to slavery,
kept the Charleston traders pushing ever toward the unknown west,
across the headwaters of the Chattahoochee and the Alabama and
into the valley of the Tennessee River. Before the turn of the century
they had reached the lands of the Chickasaw Indians bordering on
the Mississippi River, where their bright trade goods soon brought in
all the available deer in those parts. There, they were busily helping
the Chickasaws make war on their neighbors, the Choctaws, to
procure slaves in lieu of the skins, when the French arrived.
French forts and a French alliance with the Choctaws halted this
English advance into the lower valley of the Mississippi. Even so, the
Carolina traders had pushed the English frontier farther west, by
hundreds of miles, than any other colonials would do during the next
half century.
North of Virginia in the latter part of the seventeenth century the
two major areas of the fur trade among the English colonies were
New England and New York.
The New England trade, exhausting itself, was on the decline. It
had been blocked from expansion by national and political barriers in
the west and by the hostility of the French in the north. Raids and
counter raids, with the Indians used as allies on both sides, kept the
borders between the French and the New Englanders alive with
savage horrors. And, because of the prolonged hostilities in Europe
these conditions would continue into the next century, until 1763,
long after competition for pelts was no longer a controlling motive in
that area.
The main fur trade of the colonies in the north after the fall of New
Netherland was New York’s hinterland traffic, that which had been
inherited from the Dutch. All wilderness paths led to Albany, even
those made by the coureurs des bois and their copper-hued families
packing their illegal furs to the Hudson when they could not do
business with their own countrymen at Montreal. In 1679, it was said,
there were over 500 of these French renegades living among the
Indians. And to Albany, of course, came not only the beaver of the
Five Nations but the peltries of vassal tribes deep in the hinterland
for whom the Iroquois acted as middlemen.
The Five Nations were jealous enough of their trade and
sovereignty to visit swift vengeance on any vassals who tried to deal
direct with the white men, as happened to the Illinois in 1680 when
those distant natives sold their pelts to La Salle. For the same
reason they also tried to keep white pack traders from pushing
farther into the west, where they might exchange their wares direct
with the less sophisticated natives. It was a losing battle however.
By the turn of the century, traders from New York, Pennsylvania

and Virginia were working the Appalachian passes for beaver and
otter. In another twenty years many were squeezing through the
more northerly gaps into the valley of the Ohio River. By then, the
pressure of immigrant families upon the land east of the mountains
had commenced in earnest. Palatine farmers were flowing up the
valley of the Mohawk in great numbers, and land-hungry Ulster
Scots were scrambling through the Susquehanna valley and
southward up the Shenandoah.
The fur trader as usual had searched out the country. Then, while
he was still exploiting it for his own purposes, he had to make way
for the farmer. The two could never blend, not after the frontier
began to roll westward. Farmers spoiled the trade. The pioneer
traders could only move on to more fertile trading grounds, to open
new territory which itself would later be taken up by farmers.
Of course, Indian titles had to be extinguished before settlers
could legally move into the lands opened up by the fur traders. Some
tribes were a bit troublesome about this detail. The Delaware kicked
up an especially bloody fuss on the Pennsylvania frontier. They had
more than a suspicion that they had been swindled by the “Walking
Purchase.”
When William Penn, the founding proprietor of the Quaker colony,
bought land from the Delaware tribe, the extent of the purchase was
limited to the distance a man could go in 1¹⁄₂ days. But, when the
time came in 1737 for Penn’s son to measure this off, he did not
have it walked off as the Indians had presumed it would be done. To
cover the distance, the Quaker employed trained white athletes,
runners! It was even suspected that the white runners may have
used horses concealed along the route, that is, after they were out of
sight of the Indians who panted along behind them full of Penn’s
rum, according to some accounts.
Things settled down rather quickly however when James Logan,
that astute Pennsylvanian who guided the Indian policy of the colony,
treated generously with the Iroquois to keep the Delaware in line.
The Delaware, in fear of their fierce overlords in the north, vacated
most of their lands east of the mountains and joined the equally
unhappy Shawnee in the upper Ohio valley. There they listened
malevolently to French traders and soldiers who promised a red-
handed revenge.
So successful was the Indian policy in general, however, that on
the outbreak of King George’s War in 1744 between England and
France, the Iroquois were cajoled into granting the English practically
all the Ohio valley and sealing the bargain with an alliance to help
protect the property against the French who were already there. In
fact, commissioners from Virginia, Maryland and Pennsylvania,
meeting with an Iroquois delegation around the council fire at
Lancaster, obtained “a Deed recognizing the King’s right to all the
Lands that are, or shall be, by his Majesty’s appointment, in the
Colony of Virginia.”
As far as the Virginians were concerned those lands by ancient
charter stretched all the way to the South Sea, wherever that was,
although they were willing to settle for the Ohio valley for the time
being. Nor did the Quaker colony seriously dispute Virginia’s claim at
the time, even though nearly all of the fur traders beyond the
mountains, who were now aggressively competing with the French,
were Pennsylvanians.
Chief among these was George Croghan. He had not arrived in
Pennsylvania from Dublin until 1741, but he was established in trade
on the Ohio River well before King George’s War. By 1746 he had a
number of storehouses on Lake Erie itself. From the bustling base of
his operations in the 1740’s near Harris Ferry (Harrisburg) on the
Susquehanna, and later from Aughwick farther west, he and his
various partners directed effective attacks on French trade in the
Ohio valley.
Together with his brother-in-law William Trent, and Andrew
Montour, Barney Curran and John Fraser, Croghan controlled fort-
like storehouses about the forks of the Ohio, up the Allegheny and
the Youghiogheny, on the south shore of Lake Erie, at the forks of
the Muskingum, and even on the Scioto and Miami Rivers. From
these trading posts, all of which developed into rude settlements of
sorts, the Pennsylvanians distributed rum, gunpowder, lead and
flints, as well as calicoes, ribbons, colored stockings, kettles, axes,
bells, whistles and looking-glasses. In return, they collected a fine
variety of pelts and skins—beaver, raccoon, otter, muskrat, mink,
fisher, fox, deer, elk, and bear.
Croghan’s pack traders, at times possibly numbering twenty-five
men and driving a hundred or more mules altogether, followed the
Ohio down to the falls and worked the streams that fed it. They were
trading and fighting in what is now West Virginia and eastern
Kentucky almost a quarter of a century before Daniel Boone. They
bartered under the very guns of French forts, engaging in bloody
skirmishes with the French and Indians and on occasion being taken
as captives to Montreal and even to France.
Croghan had his English competitors too. There were, for
instance, the five Lowrey brothers, as aggressive and as rugged a lot
of rivals as might have been found on any fur frontier. But all the
Pennsylvanians were as one in their persistent encroachment on the
French. Backed by factors in Philadelphia and Lancaster, including
Shippen and Lawrence and the firm of Levy, Franks and Simon, both
of which specialized in the Indian trade and in turn received credit
from wealthy merchants of London and Bristol, these intrepid
frontiersmen stubbornly picked away at the French trade.
In one respect the Englishmen were fortunate. During King
George’s War the French had trouble getting sufficient trade goods,
and many Indians with whom they had been trading became
contemptuous of them. It is said that, on one occasion, when a
Frenchman only offered a single charge of powder for a beaver skin,
the Indian with whom he was bartering “took up his Hatchet, and
knock’d him on the head, and killed him upon the Spot.” Croghan
and his Pennsylvanians took full advantage of the temporary French
embarrassment, building up their annual business in pelts to a value
of some 40,000 pounds sterling.
It was the prospect of a share in this lucrative trade that motivated
some wealthy Virginians, among them Thomas Lee and the
Washingtons, who conceived the Ohio Company after the Treaty of
Lancaster. While acting as a vehicle to establish England’s claim
west of the mountains, the company as it was finally organized
promised future dividends from land development in those parts. But
there was the immediate prospect of rich gains from the fur trade,
and little time was lost in lining up experienced Indian traders for the
project.
Thomas Cresap, a clever Yorkshireman, who operated a trading
post in the mountains near the junction of the North and South
Branches of the Potomac River, became an organizing member of
the Ohio Company. So lavishly hospitable was Cresap to the Indians
and others with whom he did business that he was known to them as
“Big Spoon,” but to his Pennsylvania trading competitors he was an
undercutting Marylander not above committing murder for a beaver
skin. Certain it is that he had once been carted off in irons, after
some “rascality” on the Maryland-Pennsylvania border, to spend a
year in prison at Philadelphia.
In any case the aristocratic tidewater Virginians counted “Colonel
Cressup” a key member of their Ohio Company. The Marylander’s
trading paths already led to the Youghiogheny, the Monongahela,
and the Ohio. So, the fort-like establishment he maintained on
Virginia’s northwestern frontier served as a convenient base from
which the company commenced its well-financed operations in the
Ohio valley.
Employed by the company were some former associates of
George Croghan. Among them were Andrew Montour, a colorful half-
breed of coureur des bois stock, and Croghan’s brother-in-law
William Trent. With the aid of experienced men like these, Thomas
Cresap was soon proving his worth to his tidewater partners and to
the British Empire.
The threat was too obvious to be ignored by the French. They laid
plans to push the English back over the mountains. Already, a
French army detachment, using a traders’ portage between the
eastern end of Lake Erie and Lake Chautauqua, had gone down to
the Ohio via the Allegheny River, planting lead plates along both
streams as a warning to trespassers. Already, French-led and
French-inspired Indian raids had taken the lives of English traders,
as well as those of their native hosts. In one case a prominent Indian
chief allied with the English had been boiled and eaten by some
Ottawas led by a French half-breed, all without discouraging the
Englishmen it seemed. Now, in 1753, a French army of 1,000 men
headed down the Allegheny from Canada, to begin building a line of
forts along the line of the previously planted lead plates.
Forts were built first at present-day Erie on the lake and at the
head of French Creek to secure a portage. Then, John Frazer’s
trading post at Venango on the Allegheny was taken and converted
to a fortification. There the French troops, bogged down with
sickness, dug in for the winter.
That is where Major George Washington found them when he
carried a note from the Governor of Virginia to their commanding
officer suggesting that they all retire promptly to Canada. This, the
Frenchmen said, they had no intention of doing. In the spring they
would push on, down the Allegheny, to the strategic Forks of the
Ohio.
Even as the French army was building canoes that winter for its
advance on the Ohio, Thomas Cresap and William Trent were
supervising the construction of an English fort at the forks of the
river. They now represented the Governor of Virginia and the King of
England, as well as the Ohio Company, all of whom were one and
the same as far as the Ohio valley was concerned. In fact Trent, the
fur trader in the employ of the Ohio Company, had been
commissioned a captain by Governor Dinwiddie of Virginia to
command the new fort and oppose the French.
But when spring and the French came to the Forks of the Ohio,
Captain Trent was absent. He said he was looking for recruits; some
suspected he was ferrying his beaver to a safer spot. The ensign in
command yielded the English fort in the face of overwhelming odds,
and the French built an impressive citadel in its place, which they
named Fort DuQuesne to honor the Canadian governor of that
name.
Washington, now Colonel Washington, who was advancing from
Virginia with his militia to Trent’s support, was much too late. He was
forced to content himself with palisading a defensive position along
the road at Great Meadows. There at Fort Necessity, as he called it,
he warded off as best he could the large number of French troops
who came out to engage him.
That summer of 1754 Washington surrendered. When he led his
militiamen back over the Alleghenies, the Frenchmen had
succeeded in their purpose. The English were out of the Ohio valley.
However, the American phase of the Seven Years’ War had
commenced—two years before it was officially declared in Europe.
The critical contest known on this continent as the French and Indian
War was under way, and the very next year General Edward
Braddock arrived with his British regulars to direct the campaign.
The strategic plan decided upon encompassed a four-fold attack
upon the French at DuQuesne, Crown Point, Niagara, and in Nova
Scotia. General Braddock himself assumed the DuQuesne
assignment, the most important immediate objective. But he failed
on this mission, his abortive attempt to reach the Forks of the Ohio
ending in the disastrous rout of his troops and his own death.
It was not until 1758 that General John Forbes forced the
evacuation of the fort at the Forks of the Ohio. The French then
abandoned the entire valley. Fort DuQuesne became Fort Pitt and
the English were in control of the Ohio River.
The war was savagely fought out on all fronts in America. Other
French citadels fell—Louisburg, Frontenac, Niagara, Ticonderoga,
and Crown Point. Eventually, Quebec and Montreal, those ancient
fortresses on the St. Lawrence River, capitulated to the British. Then,
in 1763, by the Treaty of Paris, France ceded Canada and all her
territory east of the Mississippi River to Great Britain, except for one
small plot encompassing New Orleans. Spain likewise ceded Florida.
The English flanks no longer needed protection. The way west
was open and the frontier was boundless!
Settlers spilled through the gaps of the Appalachians, into Ohio,
Kentucky, Illinois. And the fur traders, making way for them as they
pressed upon their trading grounds, pushed on, ever westward,
across the plains after the turn of the century to the Rocky Mountains
and the coastal rivers of the Pacific.
But the era of the early fur trader, typified by the white trader and
the Indian hunter, had come to an end. As the frontier began rolling
across the great plains of America, the white man became trapper as
well as trader. When he took over the function of the Indian, who had
formerly caught the beaver, a whole new conception of the fur trade
in America was born. A new era commenced—that of the fur trapper.
The fur trader of early America had played out his historically
important role.
Bibliography and Acknowledgements
Adams, Charles Francis, Three Episodes in Massachusetts
History, Cambridge, 1892.
Adventures of Marco Polo, ed. by Richard J. Walsh, New York,
1948.
Adventurers of Purse and Person, Virginia, 1607-1625, compiled
and ed. by Anna Jester and Martha Woodruff Hiden, Princeton,
1956.
Alvord, Clarence Walworth, and Bidgood, Lee, The First
Explorations of the Trans-Allegheny Region by the Virginians, 1650-
1674, Cleveland, 1912.
Andrews, Charles M., The Colonial Period of American History, 4
vols., New Haven, 1934-38.
Arber, Edward, The Story of the Pilgrim Fathers, Boston, 1897.
Archives of Maryland, ed. by William Hand Browne, Baltimore,
1883-.
Arnold, Samuel Greene, History of the State of Rhode Island,
1636-1700, 2 vols., New York, 1874.
Bachrach, Max, Fur, New York, 1953.
Bacon, Janet Ruth, The Voyage of the Argonauts, London, 1925.
Baxter, James Phinney, Pioneers of New France in New England,
Albany, 1894.
Beauchamp, William M., A History of the New York Iroquois, in
New York State Museum Bulletin No. 78, Albany, 1905.
Biggar, H. P., The Early Trading Companies of New France,
Ottawa, 1901.
——, The Precursors of Jacques Cartier, 1497-1534, Ottawa,
1911.
Bicknell, Thomas Williams, History of the State of Rhode Island
and Providence Plantation, New York, 1920.
Book of Ser Marco Polo, 2 vols., trans. and ed. by Colonel Sir
Henry Yule, London, 1903.
Bozman, John Leeds, A Sketch of the History of Maryland During
the First Three Years, Baltimore, 1811.
Bradford, William, Of Plymouth Plantation, 1620-1647, ed. by
Samuel Eliot Morison, New York, 1952.
Browne, William Hand, Maryland, Boston, 1890.
Bruce, Philip Alexander, Economic History of Virginia in the
Seventeenth Century, 2 vols., New York, 1895.
Buffington, Arthur Howard, New England and the Western Fur
Trade, 1629-1675, in Publications of the Colonial Society of
Massachusetts, Vol. XVIII, Boston, 1916.
Calendar of Virginia State Papers and Other Manuscripts, ed. by
William P. Palmer, Richmond, 1875.
Champlain, Samuel de, Narrative of a Voyage to the West Indies
and Mexico in 1599-1602, London, 1859.
Channing, Edward, A History of the United States, New York,
1909.
Chapin, Howard Millar, The Trading Post of Roger Williams with
Those of John Wilcox and Richard Smith, in Publications of the
Society of Colonial Wars in the State of Rhode Island and
Providence Plantations, Providence, 1933.
Coleman, R. V., Liberty and Property, New York, 1951.
Cox, John Lyman, Governor William Pynchon, in Historical
Publications of the Society of Colonial Wars in the Commonwealth of
Pennsylvania, Philadelphia, 1950.
Documents Relating to the Colonial History of the State of New
York, ed. by Brodhead and O’Callaghan, New York, 1856.
Dorr, Henry C., The Narragansetts, in Collections of the Rhode
Island Society, Vol. VII.
Doyle, J. A., The English in America: The Puritan Colonies,
London, 1887.
Davidson, Robert L. D., War Comes to Quaker Pennsylvania, New
York, 1957.
Davis, W. W. H., The History of Bucks County, Pennsylvania,
Doylestown, 1876.
Felt, Joseph B., The Customs of New England, Boston, 1853.
Fiske, John, The Dutch and Quaker Colonies in America, 2 vols.,
Boston, 1899.
——, New France and New England, Boston and New York, 1902.
——, Old Virginia and Her Neighbors, 2 vols., Cambridge, 1900.
Frobisher, Sir Martin, Three Voyages in Search of a Passage to
Cathia and India by the Northwest, 1576-8, London, 1867.
Fur Facts and Figures, A Survey of the United States Fur Industry,
U. S. Department of Commerce, Washington, 1958.
Furs, Summary of Information, U. S. Department of Commerce,
B.S.B. 6, Washington, 1957.
Galbraith, John S., The Hudson’s Bay Company, As an Imperial
Factor, 1821-1869, Berkeley and Los Angeles, 1957.
Gardiner, J. Warren, Roger Williams, The Pioneer of Narragansett,
in Narragansett Historical Register, Vol. II.
Hakluyt, Richard, Divers Voyages Touching the Discovery of
America, London, 1850.
——, The Principal Navigations, Voyages, Traffiques and
Discoveries of the English Nation, 12 vols., Glasgow, 1903-05.
Hale, Nathaniel C., Virginia Venturer, A Historical Biography of
William Claiborne, 1600-1677, Richmond, 1951.
Hawkins, J. H., History of the Worshipful Company of Feltmakers
of London, London, 1917.
Holand, Hjalmar R., Explorations in America Before Columbus,
New York, 1956.
Hunt, George T., The Wars of the Iroquois, Madison, 1940.
Innes, Harold A., The Fur Trade in Canada, New Haven, 1930.
Janvier, Thomas A., The Dutch Founding of New York, New York,
1903.
——, Henry Hudson, New York, 1909.
Jastrow, Morris, The Civilization of Babylonia and Assyria,
Philadelphia, 1915.
Jesuit Relations and Allied Documents, ed. by Reuben Gold
Thwaites, Cleveland, 1896-1901.
Johnson, Amandus, The Swedish Settlements on the Delaware,
1638-1664, 2 vols., New York, 1911.
Kegley, F. B., Kegley’s Virginia Frontier, Roanoke, 1938.
Koopman, Harry Lyman, The Narragansett Country, Providence,
1927.
Lamb, Harold, New Found World, New York, 1955.
Latane, John H., The Early Relations Between Maryland and
Virginia, in Johns Hopkins University Studies, 13th Series, No. 3,
Baltimore, 1895.
Laut, Agnes Christina, The Fur Trade of America, New York, 1921.
Leithauser, Joachim G., Worlds Beyond the Horizon, New York,
1955.
Lewis, Clifford M. and Loomie, Albert J., The Spanish Jesuit
Mission in Virginia, Chapel Hill, 1953.
Massachusetts Historical Collections, Series 3, Vol. VIII, 1843;
Series 4, Vols. II, III, VI, and IX, 1856.
Moloney, Francis X., The Fur Trade in New England, 1620-1676,
Cambridge, 1931.
Narratives of Early Maryland, 1634-1684, ed. by Clayton Colman
Hall, New York, 1910.
Narratives of Early Pennsylvania, West New Jersey, and
Delaware, 1630-1670, ed. by Albert Cook Myers, New York, 1912.
Narratives of Early Virginia, 1606-1625, ed. by Lyon Gardiner
Tyler, New York, 1907.
New Voyages To America, ed. by R. G. Thwaites, Chicago, 1905.
Nicholson, J. P., Fort Beversrede, in Pennsylvania Magazine, Vol.
XV, Philadelphia, 1891.
Norse Discovery of America, A Compilation of Translations and
Deductions, pub. by Norroena Society, ed. by Rasmus B. Anderson,
New York, 1907.
O’Callaghan, E. B., History of New Netherland, New York, 1846.
Palsits, Victor Hugo, The Founding of New Amsterdam in 1626, in
Proceedings of the American Antiquarian Society, NS Vol. XXXIV,
1924, Worcester, 1925.
Parkman, Frances, Pioneers of France in the New World, 2 vols.,
Boston, 1897.
Pohl, Frederick J., The Lost Discovery, New York, 1952.
Pomfret, John E., The Province of West New Jersey, 1609-1702,
Princeton, 1956.
Powys, Llewelyn, Henry Hudson, New York, 1928.
Purchas, Samuel, Purchas His Pilgrims, 20 vols., Glasgow, 1905-
07.
Randall, Daniel R., A Puritan Colony in Maryland, in Johns
Hopkins University Studies in Historical and Political Science, 4th
Series, No. 6, Baltimore, 1886.
Records of the Colony of Rhode Island and Providence
Plantations, Vol. I, ed. by J. R. Bartlett, Providence, 1856.
Records of the Virginia Company of London, 4 vols., ed. by Susan
Myra Kingsbury, Washington; Vols. I and II, 1906; Vol. III, 1933; Vol.
IV, 1935.
Relations of John Verrazano, in Collections of the New York
Historical Society, Vol. I, New York, 1809.
Sailors Narratives of Voyages along the New England Coast,
1524-1624, with Notes by George Parker Winship, Boston, 1905.
Scharf, J. Thomas, History of Maryland from the Earliest Period to
the Present Day, Baltimore, 1879.
Semmes, Raphael, Captains and Mariners of Early Maryland,
Baltimore, 1937.
Shaftesbury Papers, in Collections of the Historical Society of
South Carolina, Vol. V, ed. by Langdon Cheves, Charleston, 1897.
Smith, Bradford, Captain John Smith, Philadelphia, 1953.
Smith, John, The Generall Historie of Virginia, New England, and
the Summer Isles, from the London Edition of 1629, republished in 2
volumes at Richmond, 1819.
Statutes at Large of Virginia, Vol. I, 1606-1659, ed. by William
Waller Hening, New York, 1823.
Steiner, Bernard C., Beginnings of Maryland, 1631-1639,
Baltimore, 1903.
Summers, Lewis Preston, Annals of Southwest Virginia, 1769-
1800, Abingdon, 1929.
——, Maryland During the English Civil Wars, in Johns Hopkins
University Studies; Series 24, Nos. 11-12; Series 25, Nos. 4-5;
Baltimore, 1906-07.
Sturgis, Samuel Booth, Manners and Customs of Swedish
Forefathers in New Sweden, 1638-1655, in Historical Publications of
the Society of Colonial Wars in the Commonwealth of Pennsylvania,
Philadelphia, 1957.
Thwaites, Reuben Gold, Father Marquette, New York, 1902.
Tracts and Other Papers relating principally to the Origin,
Settlement and Progress of the Colonies in North America From the
Discovery of the Country to the Year 1776, 4 volumes, collected by
Peter Force, Washington, 1836-46.
Vestal, Stanley, King of the Fur Traders, Boston, 1940.
Virginia Reader, A Treasury of Writings From the First Voyages to
the Present, ed. by Francis C. Rosenberger, New York, 1948.
Volwiler, Albert T., George Croghan and the Westward Movement,
1741-1782, Cleveland, 1926.
Voyages of Peter Esprit Radisson: his travels and experiences
Among the North American Indians: 1652-1684, translated by
Gideon D. Scull, Boston, 1885.
Voyages of Samuel de Champlain, 1567-1635, 3 vols., translated
by Charles Pomeroy Otis, Boston, 1880.
Voyages of Samuel de Champlain, 1604-1618, ed. by W. C. Grant,
in Original Narratives of Early American History, New York, 1907.
Wade, Herbert Treadwell, A Brief History of the Colonial Wars of
America, in Publications of the Society of Colonial Wars in the State
of New York, New York, 1948.
Wallmeyer, Bruno, Fur-Bearing Animals, Deutsch-Englische
Ausgabe, Frankfurt am Main, 1951.
Weeden, William B., Economic and Social History of New
England, 1620-1789, 2 vols., Cambridge, 1890.
Weslager, C. A., An Early American Name Puzzle, in Names, Vol.
II, No. 4, 1954.
——, Delaware’s Forgotten River, Wilmington, 1947.
——, Red Men on the Brandywine, Wilmington, 1953.
——, Robert Evelyn’s Indian Tribes and Place-Names of New
Albion, in Bul. 9, The Archeological Society of New Jersey, 1954.
Wilcox, R. Turner, The Mode in Furs, New York, 1951.
Williamson, William D., History of the State of Maine, Vol. I, New
York, 1940.
Willison, George F., Saints and Strangers, New York, 1945.
Willson, Beckles, The Great Company, Toronto, 1899.
Winthrop, John, The History of New England, 2 vols., Boston,
1825.
Zimmern, Helen, The Hansa Towns, New York, 1889.
INDEX
A
Abnaki Indians, 103
Acadia, 107, 110, 112
Acadian Peninsula, 39
Accomac, 115, 126
Accomac Indians, 120
Africa, 4, 14, 26, 135
Africa (ship), 126
Agawam, 169
Alabama River, 196
Albany, 70, 73, 137, 141, 177, 191, 197
Albemarle Sound, 52
Alexander, Pope, 14
Alexander, Sir William, 108, 109, 110, 126
Algonkin Indians, 33, 34, 40, 42
Algonquin Indians, 34, 59, 60, 73, 74, 117, 119, 129, 138, 170,
175, 176, 190
Allegheny River, 199, 201
Alleghenies, 202
Allerton, Isaac, 103, 104, 105, 111, 157
American Revolution, 185
Amerigo’s Land, 25

Generalized B Algebras and Applications Maria Fragoulopoulou Atsushi Inoue Martin Weigt Ioannis Zarakas Online Ebook Texxtbook Full Chapter PDF

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Generalized B Algebras and Applications Maria Fragoulopoulou Atsushi Inoue Martin Weigt Ioannis Zarakas Online Ebook Texxtbook Full Chapter PDF

Uploaded by

Copyright:

Available Formats

Generalized B Algebras and

Applications Maria Fragoulopoulou

Tomita s Lectures on Observable Algebras in Hilbert

Algebra and Applications 1: Non-associative Algebras

Algebra and Applications 2 Combinatorial Algebra and

Classical Hopf Algebras and Their Applications Algebra

Code Clone Analysis Research Tools and Practices

Introduction to Yokai Studies and On Kokkuri Inoue

Theory of Groups and Symmetries Representations of

Applications of Nanomaterials in Energy Systems 2nd

Martin Weigt Ioannis Zarakas

ISSN 0075-8434 ISSN 1617-9692 (electronic)

Contemporaneously with these philo-

Athens, Greece Maria Fragoulopoulou

6.3 *-Representations of GB*-Algebras . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 137

A Generalized B ∗ -algebra (for short GB∗ -algebra) is a symmetric pseudo-complete

© The Author(s), under exclusive license to Springer Nature Switzerland AG 2022 1

algebra A[τ ] there is a finest locally convex ∗-algebra topology “equivalent” to τ

representation is bicontinuous on its image. This is achieved by employing the so

for reaching the obtained generalizations of a non-normed Vidav–Palmer theorem

present book, or we were receivers of their important comments, having them as

Chapter 2 acts as a preparative stage to the basic definitions and fundamentals

2.1 Basic Definitions and Notation

© The Author(s), under exclusive license to Springer Nature Switzerland AG 2022 9

(x, α)(y, β) := (xy + αy + βx, αβ), ∀ x, y ∈ A and α, β ∈ C.

stand for the group of quasi-invertible elements of A.

pν (xy) ≤ pν (x)pν (y), ∀ x, y ∈ A and ν ∈ ,

then A[τ ] is called an m-convex algebra, where m-convex is an abbreviation of

Concerning Definition 2.1.1, we shall (almost) always refer to (and/or

If A[τ ] is a topological algebra without identity, its unitization A1 endowed with

Note that since we are working on algebras, we shall use, without

2.2 The Set of Bounded Elements. Radius of Boundedness

Definition 2.2.1 (Allan) Let A[τ ] be a locally convex algebra. An element x of A

n ∈ N is bounded in A[τ ]. It is easy to show that the closure of the absolutely

Furthermore, the sequence (xn )n∈N is Mackey convergent to an element x ∈ E, if

Then, Pn → sin(x), with respect to τ and the set

is clearly τ -bounded. Since,

|Pn (x) − sin(x)|e−x → 0, uniformly on R+ ,

since B 2 ⊂ B, by induction we have that (λ, x)n = (λn , xn ) ∈ B, for all n ∈ N,

Hence, xn → x, with respect to the norm  · 1C on A[C]1 .

Hence, xn − xn0 ∈ ε4 B, for all n ≥ n0 . Given that xn − xn0 → x − xn0 with

If A is a locally convex algebra with identity e, we shall denote the basic

Theorem 2.2.10 below, will be proven very helpful in various arguments in

and so, as seen above, there is an F ∈ (B)A1 , such that B1 ∪ C1 ⊂ F . Therefore,

from which follows that β(x) ≤ xB0 . 

and thus from Proposition 2.2.15 we have that

the locally convex topology τ of A. Consider the following formulae

Proposition 2.2.18 Let A[τ ] be a locally convex algebra, and x ∈ A. Then,

β(x) = β (x) = β (x).

2.3 Spectrum and Spectral Radius

The spectrum of an element x of an arbitrary algebra A, will be denoted by sp(x)

If A has no identity, then

spA (x) := spA1 (x).

Throughout this book C∗ will denote the one point compactification of C. A

In case A has no identity then σA (x) := σA1 (x, 0).

Proposition 2.3.2 Let A[τ ] be a locally convex algebra and x ∈ A. If C denotes a

element in C[τC ]. Hence, λ ∈

is holomorphic in some neighbourhood of 0. In order to obtain some useful results

Proof We note that for λ, μ in the domain of R(x) we have that

Rλ − Rμ = (λe − x)−1 − (μe − x)−1

We show that R r+1 (x) is weakly continuous at μ, for r = 0, 1, . . . , m. Towards this

Thus, the result of the lemma is established by induction. 

then, in some neighbourhood of 0, Sλ has weak derivatives of all orders given by

6.3 -Representations of GB-Algebras . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 137

pν (xy) ≤ pν (x)pν (y), ∀ x, y ∈ A and ν ∈ ,

Hence, xn → x, with respect to the norm · 1C on A[C]1 .

from which follows that β(x) ≤ xB0 .

Thus, the result of the lemma is established by induction.

Sλ(m) − Sμ(m) m!x m−1 1 m+1 1 m+1

m!x m−1 1 1 m+1 1 m+1 m+1

−(m + 1)! m−1 m+1 1 1

x m−1 m+1 1 x m−1 m+1

Therefore, the result follows by induction.

≤ Cp |λ − μ|n+1 Rμ n+1

that Rλ ∈ A[B]. Then, we have that sn → Rλ , with respect to · B : indeed, by

sn − Rλ B = Rλ (λe − x)(sn − Rλ )B ≤ Rλ B (λe − x)(sn − Rλ )B → 0,

So, Rλ B → 0, for λ → ∞. It is then easily deduced that R is holomorphic at

N = {μ ∈ C : |μ| > xB }, then Rλ = λ−1 e + λ−2 x + · · · , for λ ∈ ρ(x) ∩ N,

from which the inverse inequality follows.

defines a homomorphism f → f (x) of Fx into A, which enjoys certain properties

The statement about u0 follows from similar considerations.

It is clear that ϕα , α ∈ , is well-defined since if α δ and β, β ∈ , such that

(2) Let x ∈ Aρ and μ ∈ ρ(x) ∩ C, such that ϕ (μe − x)−1 = 0. If λ ∈ C, λ = 0,