Professional Documents
Culture Documents
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 Online Ebook Texxtbook Full Chapter PDF
https://ebookmeta.com/product/tomita-s-lectures-on-observable-
algebras-in-hilbert-space-atsushi-inoue/
https://ebookmeta.com/product/algebra-and-applications-1-non-
associative-algebras-and-categories-1st-edition-makhlouf/
https://ebookmeta.com/product/algebra-and-
applications-2-combinatorial-algebra-and-hopf-algebras-1st-
edition-makhlouf-abdenacer/
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/
https://ebookmeta.com/product/code-clone-analysis-research-tools-
and-practices-katsuro-inoue/
https://ebookmeta.com/product/introduction-to-yokai-studies-and-
on-kokkuri-inoue-enryo/
https://ebookmeta.com/product/theory-of-groups-and-symmetries-
representations-of-groups-and-lie-algebras-applications-1st-
edition-alexey-p-isaev/
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
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]
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.
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
Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 229
Index . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . 235
Chapter 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
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.
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
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
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
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.
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
14 2 A Spectral Theory for Locally Convex Algebras
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
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
2.2 The Set of Bounded Elements. Radius of Boundedness 15
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
ε
∃ n0 ∈ N : xn − xm B < , ∀ m, n ≥ n0 .
4
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.
16 2 A Spectral Theory for Locally Convex Algebras
(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.
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 → ∞ .
18 2 A Spectral Theory for Locally Convex Algebras
(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
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
2.2 The Set of Bounded Elements. Radius of Boundedness 19
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).
20 2 A Spectral Theory for Locally Convex Algebras
/ GA }.
sp(x) := {λ ∈ C : λe − x ∈
∞ + 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 .
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
(m) (m)
Rλ − Rμ
→ (−1)m+1 (m + 1)!Rμm+2 , for λ → μ, with respect to σ (A, A ).
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
1 1 1 1 −1
+ m!x m−1 − − R m+1
1 − R m+1
1
μm+1 λμ λ μ λ μ
1 −(m+2)
= (m + 1)!x m R m+2 = (m + 1)!x m e − μx .
μm+2 μ1
+∞
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
2.3 Spectrum and Spectral Radius 25
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
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
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
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
with respect to norm convergence in A[B]. Then, for f ∈ A the function φ(λ) =
f (Rλ ) is written as follows:
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)
2.3 Spectrum and Spectral Radius 29
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),
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
Hence,
1 1
u1 (x) = λ(λe − x)−1 dλ = e + λ−1 x + λ−2 x 2 + · · · dλ = x.
2πi ∂D 2πi ∂D
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.
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.
34 2 A Spectral Theory for Locally Convex Algebras
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
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 ,
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 )
(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 ) ,
1 1 1
= λ − μ1 − μ2 + + ,
ϕ Rλ (x1 + x2 ) ϕ Rμ1 (x1 ) ϕ Rμ2 (x2 )
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.