Gauge Invariance and Weyl-Polymer Quantization

Lecture Notes in Physics 904
FrancoStrocchi
Gauge
Invariance and
Weyl-polymer
Quantization
Lecture Notes in Physics
Volume 904
Founding Editors
W. Beiglbck
J. Ehlers
K. Hepp
H. Weidenmller
Editorial Board
M. Bartelmann, Heidelberg, Germany
B.-G. Englert, Singapore, Singapore
P. Hnggi, Augsburg, Germany
M. Hjorth-Jensen, Oslo, Norway
R.A.L. Jones, Sheffield, UK
M. Lewenstein, Barcelona, Spain
H. von Lhneysen, Karlsruhe, Germany
J.-M. Raimond, Paris, France
A. Rubio, Donostia, San Sebastian, Spain
S. Theisen, Potsdam, Germany
D. Vollhardt, Augsburg, Germany
J.D. Wells, Ann Arbor, USA
G.P. Zank, Huntsville, USA
The Lecture Notes in Physics
The series Lecture Notes in Physics (LNP), founded in 1969, reports new devel-
opments in physics research and teaching-quickly and informally, but with a high
quality and the explicit aim to summarize and communicate current knowledge in
an accessible way. Books published in this series are conceived as bridging material
between advanced graduate textbooks and the forefront of research and to serve
three purposes:
to be a compact and modern up-to-date source of reference on a well-defined
topic
to serve as an accessible introduction to the field to postgraduate students and
nonspecialist researchers from related areas
to be a source of advanced teaching material for specialized seminars, courses
and schools
Both monographs and multi-author volumes will be considered for publication.
Edited volumes should, however, consist of a very limited number of contributions
only. Proceedings will not be considered for LNP.
Volumes published in LNP are disseminated both in print and in electronic for-
mats, the electronic archive being available at springerlink.com. The series content
is indexed, abstracted and referenced by many abstracting and information services,
bibliographic networks, subscription agencies, library networks, and consortia.
Proposals should be sent to a member of the Editorial Board, or directly to the
managing editor at Springer:
Christian Caron
Springer Heidelberg
Physics Editorial Department I
Tiergartenstrasse 17
69121 Heidelberg/Germany
[email protected]
More information about this series at http://www.springer.com/series/5304

Franco Strocchi
Gauge Invariance and

Weyl-polymer Quantization
123
Franco Strocchi
Dipto. Fisica
Scuola Normale Superiore
Pisa, Italy
ISSN 0075-8450 ISSN 1616-6361 (electronic)

Lecture Notes in Physics
ISBN 978-3-319-17694-9 ISBN 978-3-319-17695-6 (eBook)
DOI 10.1007/978-3-319-17695-6
Library of Congress Control Number: 2015953627
Springer Cham Heidelberg New York Dordrecht London

Springer International Publishing Switzerland 2016
This work is subject to copyright. All rights are reserved 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, express or implied, with respect to the material contained herein or for any
errors or omissions that may have been made.
Printed on acid-free paper
Springer International Publishing AG Switzerland is part of Springer Science+Business Media

(www.springer.com)
Contents
1 Heisenberg Quantization and Weyl Quantization .. . .. . . . . . . . . . . . . . . . . . . . . 1

1 Heisenberg and Weyl Quantizations . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . 1
2 * From Heisenberg to Weyl Quantization .. . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . 3
3 Stone-von Neumann Theorem and Schrdinger Quantum
Mechanics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . 6
2 Delocalization, Gauge Invariance and Non-regular Representations . . . 11
1 Delocalization and Gauge Invariance . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . 11
2 The Representation Defined by a Translationally Invariant State . . . . . . . . 13
3 Bloch Electron and Non-regular Quantization .. . . . . .. . . . . . . . . . . . . . . . . . . . . 15
4 Gauge Invariance and Non-regular Canonical Quantization . . . . . . . . . . . . . 20
4.1 Gauge Invariance and Superselection Rules . . .. . . . . . . . . . . . . . . . . . . . . 20
4.2 Gauge Invariance in the Two-Body Problem . .. . . . . . . . . . . . . . . . . . . . . 21
4.3 Non-regular Representations and Symmetry Breaking . . . . . . . . . . . . 22
4.4 Goldstone Theorem and Non-regular Representations .. . . . . . . . . . . . 24
4.5 Bloch Electron as a Gauge Model.. . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . 27
5 Quantum Hall Electron: Zak States . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . 29
3 Quantum Mechanical Gauge Models . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . 35
1 Quantum Particle on a Circle . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . 35
2 Jackiw Model of Gauss Law Constraint . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . 38
3 Christ-Lee Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . 41
4 A QM Model of QCD Structures and of Josephson Effect.. . . . . . . . . . . . . . 43
4 Non-regular Representations in Quantum Field Theory . . . . . . . . . . . . . . . . . 53
1 Quantum Field Algebras and Quantizations . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . 53
2 Massless Scalar Field in 1+1 Dimensions . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . 57
3 Temporal Gauge in QED . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . 60
4 Temporal Gauge in QCD: Chiral Symmetry Breaking .. . . . . . . . . . . . . . . . . . 64
4.1 Gauss Law and Gauge Transformations . . . . . . .. . . . . . . . . . . . . . . . . . . . . 64
4.2 Large Gauge Transformations .. . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . 68
4.3 Fermions and Chiral Symmetry Breaking . . . . .. . . . . . . . . . . . . . . . . . . . . 71
5 Abelian Chern-Simons Theory . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . 74
v
vi Contents
5 Diffeomorphism Invariance and Weyl Polymer Quantization . . . . . . . . . . . 77

1 Diffeomorphism Invariance and Weyl Polymer Quantization . . . . . . . . . . . 77
2 Quantum Mechanics on a Manifold and Diffeomorphism Invariance .. . 78
6 * A Generalization of the Stone-von Neumann Theorem.. . . . . . . . . . . . . . . . 85
1 Zak Algebra . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . 85
2 A Generalization of Stone-von Neumann Theorem .. . . . . . . . . . . . . . . . . . . . . 86
Bibliography . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . 91
Index . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . 95
Introduction
The standard formulation of Quantum Mechanics (QM) is based on Heisenberg-

Dirac canonical quantization of the Heisenberg canonical variables q; p. The Weyl
quantization, i.e. the canonical commutation relations formulated in terms of the
unitary exponentials of the canonical variables (Weyl operators), is usually ignored
in the textbooks of QM addressed to physicists. Its use in the mathematically minded
presentations of QM is usually motivated by the better behavior and mathematical
control of the unitary Weyl operators with respect to the Heisenberg canonical
variables, which are necessarily represented by unbounded operators.
The connection between the two quantizations is provided by the Stone-
von Neumann theorem, which states their equivalence under general regularity
conditions (Chap. 1). This may explain why, from a practical point of view, one
may feel satisfied with the Heisenberg-Dirac quantization. The point is that the
Dirac-Heisenberg quantization implicitly assumes that all the canonical variables
described observables and therefore the regularity of their exponentials (Weyl
operators) is required by their existence as (unbounded) operators in the Hilbert
representation space.
The regularity condition at the basis of the Stone-von Neumann theorem is stan-
dard in the mathematical analysis and classification of Lie group representations and
in the quantum mechanical case it amounts to consider the strongly (equivalently
weakly) continuous (unitary) representations of the Heisenberg group.
However, for a class of physically interesting systems, especially in connection
with quantum gauge theories, it has become apparent that the Dirac-Heisenberg
quantization is not compatible with a gauge invariant ground state, only the Weyl
quantization being allowed.
In these cases, the inequivalence of the two quantization methods arises by the
lack of regularity of the one-parameter groups generated by the Weyl operators, so
that the corresponding generators, i.e. the Heisenberg canonical variables, cannot be
defined as (self-adjoint) operators in the Hilbert space of states.
The physical reason at the basis of such a lack of regularity is that the QM
description of a class of physical systems involves canonical variables, not all of
which correspond to observable quantities; some of them are introduced for the
vii
viii Introduction
description of the states, namely of the representations of the algebra of observables,

for which purpose only their exponentials are needed to exist as well-defined
operators in the Hilbert space of states. Actually, the embedding of the observable
algebra A (of canonical variables) into a larger canonical algebra F , which contains
the intertwiners between the inequivalent representations of A, qualifies as a general
strategy for a complete description of the system.
Thus, what might at first sight look as an uninteresting singular, if not pathologi-
cal, case turns out to be crucial for the quantum description of physically interesting
systems.
For example, for a particle on a circle, or more generally in a periodic structure,
only the periodic functions of the position are observable and there is no compelling
physical reason for the existence of the position operator, which would require the
regularity of the corresponding Weyl operators, whose role is to act as intertwiners
between inequivalent representations of the observables algebra (starting from the
ground/vacuum state representation).
In general, this lack of regularity may be related to the existence of a gauge group.
Typically, one has the Weyl algebra AW generated by the exponentials of the full set
of canonical variables needed for the description of the states of the system, but only
a subalgebra A describes observables. Generically, A has a non-trivial center Z,
which generates transformations having the meaning of gauge symmetries (gauge
transformations). Thus, the algebra AW of canonical variables contains both gauge
dependent and gauge invariant (i.e. observable) variables.
Clearly, the regularity condition must be satisfied by the exponentials of the
observable variables, otherwise the representation is not physical, but there is no
physical reason for the regularity condition of the gauge dependent Weyl operators.
Actually, as it shall be discussed in these notes (Chap. 2), the representation of the
Weyl algebra by a gauge invariant ground state in general requires the non-regularity
of the gauge dependent Weyl operators and implies the impossibility of defining
the corresponding generators as well-defined operators in the corresponding Hilbert
space (non-regular representation of the Heisenberg group or of the Weyl algebra).
Relevant quantum mechanical examples of such a structure are the electron in
a periodic potential (Bloch electron), the Quantum Hall electron, the particle on a
circle, where the gauge transformations are, respectively, the lattice translations, the
magnetic translations and the rotations of 2.
The general mathematical structure of non-regular realizations of the canonical
commutation relations in Weyl form provides also a mathematically consistent
solution of the problem of the equivalence of the procedures which interchange
quantization and imposition of the constraints on the physical states, e.g. the
Gauss law (i.e. the gauge invariance) constraint in Gauge Quantum Field Theories
(GQFT).
A cheap widespread solution of the quantization problem, compatibly with
the existence of gauge invariant states, is to admit non-normalizable states. Non-
normalizable state vectors are often used in quantum mechanical calculations as
pioneered by Dirac, but their mathematical oddness is harmless, since they are
used as simplifying limiting extrapolations of well-defined normalizable vectors
Introduction ix
(e.g. the plane waves as the limit of narrower and narrower wave packets of
momentum). Much more problematic and actually mathematically inconsistent is
to consider quantizations built on a cyclic non-normalizable vector, typically the
ground state, because then all the so-obtained vectors are non-normalizable and all
transition amplitudes are divergent. As discussed in these notes (Chap. 3), a much
more satisfactory solution of this problem is Weyl quantization with non-regular
representations.
The non-regular Weyl representations for the quantization of systems with a
gauge symmetry exhibit the following characteristic structures, which play an
important role in the analysis of the vacuum structure in Quantum Chromodynamics
(QCD) and do not have a counterpart in the Dirac-Heisenberg canonical quantiza-
tion:
i) a gauge invariance constraint (typically a Gauss law constraint) in operator
form compatible with canonical Weyl quantization
ii) superselected charges defined by the center of the observable algebra
iii) gauge invariant ground states, defining inequivalent representations of the
observable algebra, labeled by the spectrum of the superselected charges (the
strict analog of the so-called sectors of QCD)
iv) absence of Goldstone states associated to the spontaneous breaking of
symmetries conjugated to the gauge transformations (like the chiral symmetry
in QCD).
Such features are not peculiar of QCD and also appear in all (finite dimensional)
QM models with a gauge symmetry generated by the center of the observable
algebra. In our opinion, the realization of such general structures and their very clear
and simple realization in (finite dimensional) QM mechanical models, fully under
control, discussed in Chap. 3, sheds light on the more difficult infinite dimensional
Gauge Quantum Field Theory models.
In particular, the occurrence and relevance of non-regular representations of field
algebras is exemplified by the massless scalar field in two spacetime dimensions
and by the (positive) realization of the temporal gauge in Quantum Electrodynamics
(QED) (Chap. 4).
Non-regular Weyl quantization also provides a strategy for a derivation of the
vacuum structure and chiral symmetry breaking in QCD in a more acceptable and
convincing mathematical setting, as discussed in Chap. 4, Sect. 4. In this respect,
these notes may be regarded as a supplement and a mathematical glossary to the
standard somewhat heuristic arguments about the vacuum structure in QCD and the
U.1/ axial symmetry breaking.
Non-regular representations arise also in quantizations of diffeomorphism covari-
ant theories with a diffeomorphism invariant ground state, as discussed in Chap. 5.
They play a crucial role in string quantization and in Loop Quantum Gravity, where
they have been advocated under the name of polymer quantizations. Their features
(like the occurrence of non-separable Hilbert spaces, the impossibility of defining
gauge dependent fields as Hilbert space operators, etc.) should not be regarded as
x Introduction
serious mathematical difficulties or physical oddnesses, the quantum mechanical

models (some of which very familiar) providing well-sounded prototypes of them.
The physical relevance of non-regular representations of the Heisenberg group
raises the problem of their classification, i.e. a generalization of the classical Stone-
von-Neumann theorem, which characterizes the regular ones.
Such a generalization may be obtained by exploiting the simple form of the
Gelfand spectrum of the maximal abelian subalgebra AZ of the Weyl algebra
generated by the pairs Ui .2=l/, Vi ./, i D 1; : : : d (d the space dimensions),
formally corresponding to the exponentials exp i.2=/qi , exp i pi . Such an
algebra is called the Zak algebra and its Gelfand spectrum is given by d copies of
the two-dimensional torus D .T2 /d .
Then, one may prove that all the representations of the Weyl algebra which are
spectrally multiplicity free as representations of its Zak algebra (a condition which
generalizes irreducibility) and are strongly measurable (a condition which replaces
regularity in non-separable spaces) are unitarily equivalent to a representation of
the Weyl algebra of the same form of the standard Schrdinger representation
on L2 .; d/, with d a (positive) translationally invariant Borel measure, which
reduces to the Lebesgue measure iff the regularity condition is satisfied (Chap. 6).
The conditions which yield such a classification are satisfied by all the non-
regular representations of physical interest discussed in Chaps. 25 below; thus,
they all have the above form, with a corresponding Borel measure on .T2 /d .
Technical parts, which may be skipped in a first reading, are marked with a *.
The content of these notes relies on results obtained with the collaboration of
Fabio Acerbi, Stefano Cavallaro, Jurg Lffelholz and Giovanni Morchio, to whom
I am deeply grateful. In particular, I feel greatly indebted to Giovanni Morchio for
his crucial role in such collaborations and for his contribution of ideas and technical
skill.
Chapter 1
Heisenberg Quantization and Weyl Quantization
1 Heisenberg and Weyl Quantizations
The standard formulation of quantum mechanics relies on the so-called canonical

quantization prescriptions at the basis of Dirac formulation.1 The starting point is the
identification of the canonical variables q; p, which in the classical case describe the
configurations of the system; then the quantization procedure amounts to replacing
the classical canonical Poisson brackets by commutators (in units in which D 1)
qi ; pj D iij ; qi ; qj D 0 D pi ; pj ; i; j D 1; : : : ; s; (1.1.1)
hereafter called canonical commutation relations (CCR) in Heisenberg form.

The next step in the formulation of quantum mechanics is the choice of a
representation of the canonical variables q; p as self-adjoint operators in the Hilbert
space H of states and the standard choice is the Schrdinger representation,
according to which H is given by the square integrable (wave) functions, H D
L2 .Rs /, with q the multiplication operator and p the generator of translations:
.q /.x/ D x .x/; .p /.x/ D i@x .x/;
for all in a suitable dense domain D (see below).2

The relevant physical and mathematical question is whether such a representation
covers all the physically interesting cases and more generally what is the status of
other possibly existing representations.
1
P.A.M. Dirac, The principles of Quantum mechanics, Oxford University Press 1986, Chap. IV.
2
The unitary equivalent so-called Heisenberg representation is obtained by (the unitary) Fourier
transform: .q Q /.k/ D i@k Q .k/, .p Q /.k/ D k Q .k/.
Springer International Publishing Switzerland 2016 1

F. Strocchi, Gauge Invariance and Weyl-polymer Quantization,
Lecture Notes in Physics 904, DOI 10.1007/978-3-319-17695-6_1
2 1 Heisenberg Quantization and Weyl Quantization
The implicit physical assumption underlying Heisenberg quantization is that the

canonical variables q; p, promoted to Hilbert space operators, have a direct physical
interpretation, i.e. they describe observable quantities. This implies that they must
be represented by (necessarily unbounded) self-adjoint operators. In general, one
cannot require that they have the same domain of self-adjointness and in order to
give a meaning to the commutation relations Eq. (1.1.1), q and p should at least
have a common dense domain D of essential self-adjointness, such that
qj D D; pj D D; qi pj pj qi D iij 1; on D: (1.1.2)
Without further conditions the commutation relations in the form of Eq. (1.1.2)
admit a plenty of representations which are not unitarily equivalent to the
Schrdinger representation. Thus, one is facing the mathematical and physical
problem of focusing the conditions which select the standard Schrdinger choice;
this will be the object of the present section.
To this purpose we recall the following basic concepts.
The (real) Heisenberg Lie algebra,, henceforth denoted by LsH is the Lie algebra
with generators Q Q i ; PQ i ; Z,
Q i D 1; : : : ; s, and Lie brackets Q
Q i ; PQ j D ij Z,
Q all other
Lie brackets vanishing. The relation with the canonical variables qi ; pi is given by
qj ! iQ Q j , pj ! iPQ j , 1 ! iZ. Q
The non-compact Heisenberg group Hs is obtained by the exponential map
Q Q ZQ
.; ; / D eQC PC
, ; 2 Rs , 2 R, so that the group law is
.; ; /. 0 ; 0 ; 0 / D . C 0 ; C 0 ; C 0 C 12 . 0 0 //:
The so defined Heisenberg group has a non-trivial center, generated by .0; 0; z/,
so that its irreducible representations are not faithful. Hence, it is convenient to
consider the reduced Heisenberg group, obtained by restricting z to be a real
number modulo 2; in the following for simplicity, the reduced group shall still
be referred to as the Heisenberg group.3 Clearly, it is generated by the elements
U./ .; 0; 0/, V./ .0; ; 0/, eiz 1 .0; 0; z/, z being a real number
modulo 2.
The Weyl algebra, also briefly called CCR algebra is the algebra generated by
the elements U./; V./, ; 2 Rs , with product rules provided by the group laws
of the Heisenberg group
U./ V./ D V./ U./ei ; (1.1.3)

U./U. 0 / D U. C 0 /; V./V. 0 / D V. C 0 /; (1.1.4)
3
For more details on the Heisenberg group, see G.B. Folland, Harmonic Analysis in Phase Space,
Princeton University Press 1989, esp. Chap. 1.
2 * From Heisenberg to Weyl Quantization 3
also called canonical commutation relations in Weyl form or briefly the Weyl
relations. Clearly, by the Weyl relations, all monomials of Us and Vs reduce to
products of the form UVeiz 1.
One may define a * operation by U./ D U./, V./ D V./, and there is
a unique norm jj jj, with jjU./jj D jjV./jj D 1 and satisfying jjA Ajj D jjAjj2 ,
8A.4 The norm closure of the Weyl algebra with such a norm defines the Weyl C -
algebra AW .5
Definition 1.1 A representation of a (real) Lie algebra L in a Hilbert space H is a
homomorphism of L into a set of linear operators in H having a common invariant
dense domain D. A representation is therefore identified by the pair .; D H/.
The representation is said to be self-adjoint if 8X 2 L, i.X/ is essentially self-
adjoint on D.
A representation of a Lie group G in a Hilbert space H is a homomorphism
of G into a set of bounded operators in H, i.e. .g1 g2 / D .g1 / .g2 /, .e/ D 1,
8g1 ; g2 2 G, e denoting the identity.
The representation of G is called unitary if 8g 2 G, .g/ is a unitary operator.
A unitary representation of G is called regular if the representatives of the one-
parameter subgroups of G are continuous in the group parameters with respect to
the strong Hilbert space topology.
A self-adjoint representation of the Heisenberg Lie algebra defines what shall
be called a Heisenberg quantization. A representation of the Weyl algebra
(equivalently a unitary representation of the Heisenberg group) defines a Weyl
quantization, which is called regular if so is the defining representation.6
A regular Weyl quantization defines a Heisenberg quantization, but the converse is
not true, without further (mathematical) conditions discussed in the next Section.
2 * From Heisenberg to Weyl Quantization
The quantization advocated by Weyl in his pioneering work on quantum mechanics

and adopted in the mathematical physics literature on canonical quantization, uses
the commutation relations in Weyl form and considers the unitary representations of
4
See J. Slawny, Comm. Math. Phys. 24, 151 (1971); J. Manuceau, M Sirugue, D. Testard and A.
Verbeure, Comm. Math. Phys. 32, 231 (1973).
5
We recall that a C -algebra A is a complex Banach algebra with an involution * such that the
norm satisfies jjA Ajj D jjAjj2 , 8A 2 A.
6
This means that for any pair of Hilbert space vectors ; 2 H , the matrix elements
F .; / .; .U./ V.// /, of the representatives of the group elements, are continuous
functions of ; .
the Heisenberg group.7 Clearly, Weyl strategy can be regarded as a regularized (and
actually more general) version of Heisenberg quantization, since it is formulated in
terms of the (bounded) Weyl variables U; V. Moreover, by Stone-Von-Neumann
(SvN) theorem, the requirement of regularity of the unitary representations of
the commutation relations in Weyl form, uniquely leads to Schrdinger quantum
mechanics (see below).
Furthermore one has
Theorem 2.1 Let U./; V./, ; 2 Rn be weakly continuous unitary groups
satisfying the Weyl relations (1.1.3), (1.1.4), in a separable Hilbert space H, then
there is a dense domain D of essential self-adjointness for the generators q; p
satisfying Eqs. (1.1.2)
The proof of Theorem 1.2 follows from Stone-von Neumann uniqueness theorem
on the unitary equivalence of all weakly continuous irreducible unitary representa-
tions of the Weyl relations (see below). A direct proof exploits the construction and
properties of the dense Grding domain D, defined as the linear space spanned by
the vectors
Z
W.f / d d f .; /U./ V./
2 2
with f of the form f .; / D n m ea. C / , m; n 2 N; a > 0. The argument
is essentially the same as for the proof of Stones theorem on weakly continuous
groups of unitary operators, by which weak continuity is equivalent to the condition
that the generators are represented by self-adjoint operators.8
Thus, strongly continuous unitary representations of the Heisenberg group define
self-adjoint representations of the Heisenberg Lie algebra. However, not all the
representations of the Heisenberg Lie algebra are obtained in this way. In fact,
even if the Heisenberg canonical variables q; p are represented by essentially self-
adjoint operators on a common invariant dense domain D, on which they satisfy
the canonical commutation relations (1.1.2), they need not to exponentiate to the
Heisenberg group, nor be equivalent to the Schrdinger representation.
An interesting physical question is to clarify the physical requirements which
lead from Heisenberg quantization to Schrdinger quantum mechanics. For this
purpose, as we shall see, two properties are at issue: the exponentiation of the
Heisenberg variables to unitary groups satisfying the Weyl relations and the weak
continuity of such unitary groups.
7
H. Weyl, The Theory of Groups and Quantum Mechanics, Dover 1931; W. Thirring, A Course in
Mathematical Physics, Vol. 3, Quantum Mechanics of Atoms and Molecules, Springer 1981. For
an extensive and excellent analysis of Weyl quantization see D.A. Dubin, M.A. Hennings and T.B.
Smith, Mathematical aspects of Weyl quantization and phase, World Scientific 2000.
8
See M. Reed and B. Simon, Methods of Modern Mathematical Physics, Vol. I,, Academic Press
1972, Sect. VIII.4, and Vol. II, Problem 30, p. 341.
2 * From Heisenberg to Weyl Quantization 5
From a mathematical point of view the property that the Heisenberg variables
q; p exponentiate to one-parameter weakly continuous unitary groups is equivalent
to their self-adjointness (by Stones theorem); therefore, it is implied by the physical
requirement that q; p describe (unbounded) observables and are therefore described
by self-adjoint operators in the Hilbert space of states.
The strong continuity is a standard condition in the theory of representations of
Lie groups and by von Neumann theorem, for representations in separable Hilbert
spaces, it is equivalent to the weak measurability of .U.//, .V.//.9 Therefore,
it is hard to think of a weaker condition, if the generators of the Heisenberg group
are required to be observable variables.
Given for granted that the Heisenberg variables must at least satisfy Eqs. (1.1.2),
the Weyl relations for the corresponding unitary groups may be derived by the
condition that on the common dense domain D also the quadratic Nelson operator
q2 C p2 associated to the Heisenberg Lie algebra is essentially self-adjoint. By
Nelson-Stinespring theorems this implies that on D all the polynomials of q or of p
are essentially self-adjoint on D, i.e. all such polynomials are uniquely defined as
selfadjoint operators.10 A physical motivation for such a condition is the selection
of those representations of the Heisenberg canonical variables which guarantee
the construction of at least the polynomial functions separately of q and of p,
as self-adjoint operators.11 Under such conditions the Weyl relations follow from
Eqs. (1.1.2) by the Rellich-Dixmier theorem.
Theorem 2.2 If q and p satisfy the Heisenberg relations in the form (1.1.2) and
also q2 C p2 is essentially self-adjoint of D, then the unitary groups defined by q and
p satisfy the Weyl relations.
Proof Under the same hypotheses Dixmier proved that the action of q and p is
unitarily equivalent to that of a direct sum of the Schrdinger representations and
for all of them the Weyl relations hold.12
9
See e.g. M. Reed and B. Simon, Methods of Modern Mathematical Physics, Vol. I, Academic
Press 1979, Chap. VIII, Sect. 4, Theorem VIII.9.
10
For the Nelson-Stinespring theorems see, e.g., A.O. Barut and R. Raczka, Theory of Group
Representations and Applications, World Scientific 1986, Chap. 12, Sect. 2.
11
One cannot require that all the elements of the enveloping algebra of the Heisenberg Lie algebra,
namely the polynomial algebra generated by the Heisenberg canonical variables q; p, are essentially
self-adjoint on D.
12
For the proof of Dixmier theorem and more generally for the discussion of mathematical
conditions on the generators of the Heisenberg Lie algebra, which ensure their exponentiation
to the Heisenberg group, see C.R. Putnam, Commutation Properties of Hilbert Space Operators
and Related Topics, Springer 1967, esp. Chap. IV; P.E.T Jorgensen and R.T. Moore, Operator
Commutation Relations, Reidel 1984.
3 Stone-von Neumann Theorem and Schrdinger Quantum

Mechanics
Weyl quantization allows for a simple formulation of the condition which leads to
Schrdinger quantum mechanics.
Definition 3.1 A unitary representation of a Lie group G in a Hilbert space H is
said to be irreducible if the only closed invariant subspaces are f0g and H.
Proposition 3.2 Given a unitary irreducible representation of a Lie group G
in a Hilbert space H, let AG denote the algebra generated by (complex) linear
combinations and products of the operators .g/, g 2 G. Then any vector 2 H is
cyclic for AG , i.e. AG D H.
Proof Clearly, denoting by * the Hilbert space adjoint one has AG D AG and H1
AG is a closed invariant subspace.
Theorem 3.3 (Stone-von Neumann Uniqueness Theorem) All unitary regular
irreducible representations of the canonical commutation relations in Weyl form
are unitarily equivalent to the Schrdinger representation:
.U./ /.x/ D eix .x/; .V./ /.x/ D .x C /; 2 L2 .Rn /: (1.3.1)
Proof Since, by the Weyl relations the monomials of elements of the form
W.; / D ei U./ V./ reduce to an element of the same form, a unitary
irreducible representation is completely determined, up to isometries, by the
expectations of the W.; /s on a (cyclic) vector . In fact, if ; 0 are two
unitary irreducible representations in H; H0 , respectively, and for a pair of (cyclic)
vectors 2 H, 0 2 H0 , one has .; .AG // D . 0 ; AG 0 /, then the mapping
U W .A/ ! 0 .A/ 0 , 8A 2 AG , and its inverse are densely defined and preserve
the scalar products. Then, U defines an isometry between the two representations.
Now, for any unitary regular representation , .W.; // is a continuous
bounded operator-valued function, so that the integral
Z
2 C 2 /=4
P .1=2/ d d e. .W.; // D P ; (1.3.2)
exists and defines a bounded operator. Furthermore, since

Z
2 C 2 /=4
e. .W.; // D d d ei.C / .W.; //P.W.; //;
P cannot vanish. Finally, by Gaussian integrations one proves that

2 C 2 /=4
P .W.; // P D e. P: (1.3.3)
3 Stone-von Neumann Theorem and Schrdinger Quantum Mechanics 7
Therefore, P is a non-trivial projection and there exists at least a vector 2 P H

2 2
such that .; W.; / / D e. C /=4 . Hence, all unitary irreducible regular
representation are isomorphic and it is not difficult to show that the Schrdinger
representation (1.3.1) is unitary, regular and irreducible.13
Stone-von Neumann theorem plays a very important role for the foundations of
quantum mechanics, since it clarifies the relation between Heisenberg quantization
and Schrdinger wave mechanics.
The regularity conditions at the basis of Stone-von Neumann theorem were
regarded so natural and harmless that the problem of a possible inequivalence
between Heisenberg and Schrdinger quantization is not even mentioned in most
textbooks on quantum mechanics. In fact, the problem of existence of inequivalent
representations of the canonical commutation relations has since then been regarded
to arise only in the case of infinite degrees of freedom.
Summarizing, the following relations emerge:
1. Heisenberg quantization at the basis of the standard presentation of quantum
mechanics requires the addition of some technical, yet very reasonable, condition
in order to lead to Schrdinger quantum mechanics (see e.g. Theorem 2.1).
From a physical point of view, Heisenberg quantization applies to systems
described by canonical variables which have the interpretation of observable
quantities. As we shall see in the following Chaps. 2, 3, this innocent looking
condition is not satisfied by a class of interesting physical systems, typically
those described by canonical variables with a gauge symmetry; in these cases, the
requirement of a gauge invariant ground state excludes Heisenberg quantization.
2. Weyl quantization is more general than Heisenberg quantization and, in fact,
applies also to canonical quantum systems with a gauge symmetry and/or with
strong delocalization (see the following Chapters). The regularity condition
provides the strict link between Weyl quantization and Schrdinger quantum
mechanics; such regularity may fail for the Weyl exponentials of gauge depen-
dent canonical variables, as it happens in the case of representations defined by a
gauge invariant state.
3. Gauge dependent canonical variables. The above structure suggests that an
important point for the quantization problem is the distinction between canonical
variables which have the meaning of observable quantities and those which are
instrumental for the description of the states, but whose corresponding operators
do not belong to the algebra of observables. This phenomenon is typically
associated to the presence of a gauge symmetry and the non-observable canonical
variables are not gauge invariant.
For example, the Weyl C -algebra generated by the canonical variables may be
taken as the C -algebra of observables in the case of N distinguishable particles, but
13
See, e.g., F. Strocchi, An Introduction to the Mathematical Structure of Quantum Mechanics, 2nd
expanded edition, World Scientific 2010, p. 63; hereafter this book will sometimes be referred to
as Strocchi (2010).
this is no longer the case for N identical particles, characterized by presence of the
gauge group of permutations.
These considerations suggest that the basic ingredient for description of quantum
systems is the C -algebra of observables and the quantization problem, i.e. the
identification of the quantum states of the system, reduces to the analysis of the
representations of its C -algebra of observables.14
Such a structure is already present, actually in a stronger form, in one of the
Dirac-von Neumann axioms, where it is assumed that the observables are described
by the set of self-adjoint operators in the Hilbert space H of vector states, so that
they generate the C -algebra of all bounded operators in H.
As discussed in the references of footnote 14, the C -algebraic structure of the
observables is enough for deriving the other Dirac-von Neumann axioms: (i) the
description of the states by Hilbert space vectors follows from the Gelfand-Naimark-
Segal representation theorem, (ii) the representation of the observables by Hilbert
space operators follows from Gelfand-Naimark theorem on the characterization of
abstract C -algebras, (iii) finally, by Stone-von Neumann theorem, Schrdinger
quantum mechanics follows from Heisenberg uncertainty relations codified by
the canonical commutation relations which define the non-abelian C -algebraic
structure of AW .
The formulation of quantum mechanics in terms of the algebra of observables
allows for a simple description of the so-called superselection rules, which in the
conventional formulation correspond to the existence of quantum numbers commut-
ing with all observables. In the C -algebraic formulation, superselection rules are
defined by the existence of inequivalent physically acceptable representations of the
observable algebra.
Therefore, in the cases in which the algebra of observables is given by the Weyl
algebra, so that the regularity of the representations of the Weyl algebra is physically
motivated and may be taken as a criterion of physical acceptability, Stone-von
Neumann uniqueness theorem implies that there are no superselection rules; this
is the case of quantum systems of N distinguishable particles.
In view of the relevance of the C -algebraic structure of the observables and in
order to provide a simple dictionary for the discussion of the following Chapters
we recall a few basic facts15 about representations of C -algebras, which in the
following shall always be assumed to be unital, i.e. to have an identity 1.
A state ! of a C algebra A is a positive linear functional on A, i.e. 8A; B 2 A,
!.A/ 2 C, !.A C B/ D !.A/ C !.B/, !.A A/ 0. A positive functional is
necessarily continuous: j!.A/j jjAjj !.1/. Without loss of generality, one can
assume that the states are normalized, i.e. !.1/ D 1.
14
For a presentation of quantum mechanics based on the physically motivated C -algebraic
structure of the observables, see F. Strocchi, An Introduction to the Mathematical Structure of
Quantum mechanics, 2nd expanded ed. World Scientific 2010; Eur. Phys. J. Plus (2012), 127:12.
15
For a more extended account see e.g. Strocchi (2010).
3 Stone-von Neumann Theorem and Schrdinger Quantum Mechanics 9
A state is pure if it cannot be written as a convex linear combination of two other

states.
A representation of a C -algebra A is a homomorphisms of A into a C -
subalgebra of the algebra B.H/ of bounded linear operators on a Hilbert space H,
i.e. .A C B/ D .A/ C .B/, .AB/ D .A/ .B/, .A/ D .A /. It follows
that jj.A/jj jjAjj.
A representation is faithful if it is an isomorphism.
Theorem 3.4 (GNS) Given a state ! on a C -algebra A, there exists a represen-
tation ! on a Hilbert space H! with the property that H! contains a cyclic vector
! , such that 8A 2 A,
!.A/ D .! ; ! .A/ ! /: (1.3.4)
Any other representation, !0 in a Hilbert space H!0 with a cyclic vector !0
satisfying Eq. (1.3.4), i.e. .!0 ; !0 .A/ !0 / D !.A/, is unitarily equivalent to ! .
Theorem 3.5 The GNS representation defined by a state ! is irreducible if and
only if ! is pure.
Theorem 3.6 (Gelfand-Naimark) A C -algebra A, with identity, is isomorphic to
a C -algebra of (bounded) operators in a Hilbert space.
A multiplicative linear functional m on an abelian (unital) C -algebra A, is a
homomorphism of A into the set C (of complex numbers), i.e. a mapping which
preserves all the algebraic relations:
m.AB/ D m.A/ m.B/; m.A C B/ D m.A/ C m.B/:
Clearly, if m is not the trivial homomorphism, m.1/ D 1.

The Gelfand spectrum of an abelian (unital) C -algebra A is the set .A/ of its
multiplicative linear functionals.
Theorem 3.7 (Gelfand) An abelian (unital) C -algebra A is isomorphic to the
C -algebra of continuous functions on a compact Hausdorff topological space,
which is its Gelfand spectrum .A/, with the topology induced by the weak
topology.
For a handy account of these notions see Strocchi (2010).
Chapter 2
Delocalization, Gauge Invariance
and Non-regular Representations
1 Delocalization and Gauge Invariance
As discussed in the previous chapter, the physical criterium of regularity of the

unitary representations of the Heisenberg group corresponds to the interpretation
of the generators q; p as (unbounded) observable variables, so that they must be
described by self-adjoint operators. This implies that the states of the system can
be described in terms of L2 wave functions on the spectrum of the position q (or,
equivalently, of the momentum p). This amounts to assume that the states of the
system have good localization properties. The aim of this section is to provide
evidence that this innocent looking assumption cannot be adopted for the quantum
description of some interesting physical systems.
The physical reason for the lack of regularity is that some generators of the
Heisenberg group may not correspond to observables and therefore they need not
be described by self-adjoint operators.
A typical example is when the configurations of the quantum systems cannot
have L2 localization. For example, for a quantum particle (typically an electron) in
a periodic potential (Bloch electron), by Bloch theorem the energy eigenstates, in
particular the ground state, are described by quasi periodic wave function, which
cannot belong to L2 .
This means that in the representations defined by the Bloch states the position
variable q does not exist as a (densely defined) self-adjoint operator and therefore
the corresponding representation of the Heisenberg group is not regular.
Another class of examples arises if one ask for a quantization of a system
with a ground state characterized by the property of being the eigenstate of a
canonical observable with continuous spectrum. Then, the unitary one-parameter
group generated by the conjugated canonical variable cannot be weakly continuous

12 2 Delocalization, Gauge Invariance and Non-regular Representations
and one has to consider non-regular representations (physically interesting models

shall be discussed below).1
Quite generally, in order to describe the set of physically relevant states of some
physical systems, it is convenient to introduce a canonical algebra AW larger than
the observable algebra A. Correspondingly, the Heisenberg group GH generated
by the larger algebra AW contains as a proper subgroup the group Gobs , called the
Heisenberg observable subgroup. generated by A.
Such a structure is particularly interesting if Gobs has a non-trivial center Z.
The physical origin and interpretation of such a structure is that the center Z of
Gobs defines gauge transformations, and the generators of Z have the meaning of
superselected charges, since they commute with the observable algebra.
The physical consequence is that the observable algebra has inequivalent rep-
resentations labeled by the spectrum of Z, called superselected sectors and the
variables conjugated to the elements of Z describe charge raising/lowering oper-
ators.
The physically motivated condition that one is interested in gauge invariant states
implies that they must be invariant under Z, i.e. be eigenstates of the generators
of Z. Hence, in such representations the one-parameter groups generated by the
variables conjugated to Z cannot be weakly continuous.
A very simple example is the two body problem, if one declares that one does not
observe the position of the center of mass. Then, the Heisenberg group generated
by the canonical variables q1 ; q2 ; p1 ; p2 contains as observable subgroup, Gobs , the
group generated by
q D q1 q2 ; p D .m2 p1 m1 p2 /=.m1 C m2 /; P D p1 C p2 :
The gauge invariant states are the eigenstates of P; in particular, the ground state of
the Hamiltonian
H D P2 =2M C p2 =2 C V.q1 q2 /; M m1 C m2 ; m1 m2 =M;
corresponds to the zero eigenvalue of P. Then, the GNS representation of the (twelve
dimensional) Heisenberg group defined by such ground state is not regular.
As discussed in the Introduction, the cheap widespread solution of admitting
non-normalizable states is mathematically unacceptable, since a non-normalizable
cyclic ground state implies that all transition amplitudes are divergent.
Actually, for the gauge invariant quantization of gauge models, as the examples
mentioned above, rather than considering the fake escape of non-normalizable state
1
For a mathematical analysis of non-regular representations of the Weyl algebra see the pioneering
paper by R. Baume, J. Manuceau, A. Pellet and M. Sirugue, Comm. Math Phys. 38, 29 (1974);
a systematic analysis from the point of view of gauge quantum models is given by F. Acerbi, G.
Morchio and F. Strocchi, Jour, Math. Phys. 34, 899 (1992); Lett. Math. Phys. 27, 1 (1993); Lett.
Math. Phys. 26, 13 (1992).
2 The Representation Defined by a Translationally Invariant State 13
vectors, one should better consider non-regular representations of the Heisenberg

group, or equivalently of the Weyl algebra. This leads to quantizations in which only
the exponentials of some canonical variables (hereafter called singular variables,
typically those which are not gauge invariant) can be represented, but not the
variables of which they are the formal exponentials.
Rather than working with mathematically inconsistent non-normalizable state
vectors one should use non-regular representations built on normalizable cyclic
vectors, which yield well defined expectations for the regular canonical variables
(typically gauge invariant variables) and (only) for the exponentials of the non-
regular ones.
The general lesson from the examples mentioned above is that the standard
canonical quantization in terms of the Heisenberg canonical variables q; p, may
not always be possible, since the canonical variables do not always have a direct
physical interpretation and their spectrum may not be observable. A safer and more
general quantization is that done in terms of the Weyl operators, whose existence is
related to the much milder requirement of existence of translations and boosts.
2 The Representation Defined by a Translationally Invariant

State
We consider for simplicity the one-dimensional case. The group of space transla-
tions , 2 R, is described by the one-parameter group V./, and a state !0 on
AW is translationally invariant if
!0 . .A// D !0 .V./AV./ / D !0 .A/; 8A 2 AW : (2.2.1)
An interesting problem is to characterize the GNS representation 0 defined by

!0 . As we shall see below, such a representation occurs in the quantization of several
physical systems and it is also of interest for analogies between quantum mechanical
and gauge field theory models. It is also the prototype of non- regular representations
of the Heisenberg group.
Proposition 2.1 The GNS representation 0 defined by a pure translationally
invariant state !0 is unitarily equivalent to the following representation
!0 .U./ V.// D 0; if 0; !0 .V.// D eipN ; pN 2 R: (2.2.2)
Thus, the one-parameter group U./ is non-regularly represented. The GNS

representation space H0 contains as representative of !0 a cyclic vector 0 such
that (denoting by the same symbols the elements of the Weyl algebra and their
representatives)
V./0 D eipN 0 ; .U./ 0 ; U. 0 / 0 / D 0; if 0 : (2.2.3)

The linear span D of the vectors U./0 , 2 R is dense in H0 , which is non-

separable.
The generator of the one-parameter group U./ does not exist, but nevertheless
a generic vector of D
X
A D A0 ; A D an U.n /; fan g 2 l2 ;
n2Z
P
may be represented by a wave function A .x/ D n2Z an ein x , with scalar product
given by the ergodic mean
X Z L
. A; A/ D jan j2 D lim .2L/1 dx N A .x/ A .x/: (2.2.4)
L!1 L
n2Z
The spectrum of V./ is a pure point spectrum.

Proof Quite generally, by Eq. (2.2.1) and the Weyl commutation relations one has,
8 2 R,
!0 .U./ V. /// D !0 .V./U./ V. /V./ / D !0 .U./ V. //ei : (2.2.5)
Thus one has
!0 .U./V. // D 0; unless D 0; !0 .V. // 0: (2.2.6)
Hence, the representative of the unitary group U./, 2 R, is not weakly

continuous and the representation is not regular.
In order to completely characterize the representation, we note that the translational
invariance of !0 implies that the operator T./ defined by
T./ A 0 D .A/ 0 ; T./ 0 D 0 ; 8A 2 AW ;
is a unitary operator and satisfies T./AT./ D .A/. In fact, the mapping T./
(together with its inverse) defined on the dense set AW 0 by
T./0 D 0 ; T./ A 0 D .A/0 ; 8A 2 AW ;
preserves the scalar product as a consequence of the invariance of !0 ,
.T./A 0 ; T./ B0 / D .0 ; .A B/ 0 / D !0 . .A B// D

D !0 .A B/ D .A0 ; B0 /:
Hence, T./ is unitary. Furthermore, by construction T./ V./ commutes with

AW (since both T./ and V./ generate the same automorphism), and by the
3 Bloch Electron and Non-regular Quantization 15
irreducibility of the representation (following from !0 being pure) T./ V./ D

ei./ 1. The group law requires ./ D pN , pN 2 R, and therefore V./0 D eiNp 0 :
Then, Eq. (2.2.6) implies Eqs. (2.2.2), (2.2.3).
The Weyl operators are represented in the following way:
.U./ /.x/ D eix .x/; .V./ /.x/ D .x C /:
The rest of the Proposition follows from a direct check that the cyclic vector
defined by the wave function 1 D 1 yields the same expectations as !0 ; therefore
the corresponding GNS representations are unitarily equivalent. Moreover, one has
V./ U./0 D ei.NpC/ U./ 0 :
3 Bloch Electron and Non-regular Quantization
It is a very good approximation to describe an electron in a periodic crystal by a

Schrdinger equation with a periodic potential; for simplicity we consider the one-
dimensional case.2 In this case the Hamiltonian is H D d2 =dx2 C W.x/, with the
potential satisfying the periodicity condition W.x C a/ D W.x/, for a suitable a.
The spectrum of H is purely continuous, so that the improper eigenvectors, and
in particular the improper ground state, are not described by square integrable
functions and therefore are not normalizable. On the other hand, much of the
wisdom on periodic structures in solids makes extensive use of such improper
states and in order to bypass the difficulties of non-normalizability it has becomes
standard to restrict the wave functions to an elementary unit cell. At the basis of this
prescription is the so called Floquet-Bloch theorem,3 according to which the energy
improper eigen-functions can be chosen of the form
k .x/ D eikx vkn .x/; vkn .x C a/ D vkn .x/; k 2 0; 2=a/; n 2 N;

n
(2.3.1)
(Bloch electrons). Such improper states do not play the mere role of limiting
extrapolations of well defined vectors (like the plane waves in the free case),
since all the physically relevant states used in the treatment of periodic structures
in solids belong to the cyclic representation defined by the ground state and the
2
For an excellent treatment of the Schrdinger operators with periodic potentials, see M. Reed and
B. Simon, Methods of Modern Mathematical Physics, Vol. IV, Academic Press 1978, Sect. XIII.16;
in particular pp. 287301 for the one-dimensional case. For an expository presentation of the one-
dimensional case, see e.g. A.A. Cottey, Am. J. Phys. 39, 1235 (1971). For the discussion of the
related physical problem see e.g. J.M. Ziman, Principles of the Theory of Solids, Cambridge Univ.
Press 1964, Chap. 1; N.W. Ashcroft, Solid State Physics, Saunders College Publ. 1976, p. 132141.
3
G. Floquet, Ann. cole Norm. Sup. 12, 47 (1883); W. Magnus and S. Winkler, Hills Equation,
Wiley 1966; F. Bloch, Z. Physik 52, 555 (1928).
latter corresponds to the improper wave function 00 .x/ D v00 .x/ (Eq. (2.3.1) with
k D 0; n D 0). This justifies to look for a mathematical control of the status of such
representations also in order to obtain well defined rules for computing transition
amplitudes etc.
Another motivation for such an investigation is to clarify the analogy drawn
between the vacua representations in quantum chromodynamics (QCD) and that
of the ground state of the Bloch electron.4
Proposition 3.1 5 Let W.x/ be a bounded measurable periodic potential, W.x/ D
W.x C a/, then there exists one and only one irreducible representation .; K/ of
the CCR algebra AW in which the Hamiltonian
H D p2 =2 C W.x/
is well defined, as a strong limit of elements of AW (on a dense domain), and has a
ground state 0 2 K.
Moreover, such a representation is independent of W, in the class mentioned
above, and it is the unique non-regular representation 0 in which the subgroup
V./, 2 R is regularly represented; its generator p has a discrete spectrum.
The Hilbert space K of 0 consists of the formal sums
X
.x/ D cn ein x ; fcn g 2 l2 .C/; x 2 R; n 2 R; (2.3.2)
n2Z
with scalar product given by the ergodic mean
X Z L
. ; /D 2
jcn j D lim .2L/ 1
dx N .x/ .x/: (2.3.3)
L!1 L
n2Z
The Weyl operators are represented by
.o .U.// /.x/ D eix .x/; .0 .V.// /.x/ D .x C /: (2.3.4)
The (orthogonal) decomposition of K over the spectrum of V.a/ is

X
KD H ; V.a/ H D ei H ; 2 0; 2/: (2.3.5)
20; 2/
4
R. Jackiw, Rev. Mod. Phys. 49, 681 (1977), Sect. III. G.
5
J. Lffeholz, G. Morchio and F. Strocchi, Lett. Math. Phys. 35, 251 (1995).
The spectrum of p in H is .p/jH D f2n=a C =a; n 2 Zg and the wave

functions 2 H are quasi periodic of the form
X
.x/ D ei x=a cn ei2nx=a : (2.3.6)
(Bloch waves). The unique ground state is a vector of H D0 .

Proof
1) The representation is non-regular for U./.
A bounded measurable potential belongs to the strong closure of AW and
therefore if H is a well defined operator as a strong limit of elements of AW , so
is H0 D p2 =2m. Hence, p2 is well defined and so is its square root p, which is the
generator of V./. Therefore, by Stones theorem V./ is weakly continuous,
i.e. regularly represented in H.
The Weyl commutation relations and the weak continuity of V./ imply
U./ p U./ D p , i.e. the spectrum .p/ of p is homogeneous. From
the irreducibility of 0 it follows that three cases are possible:
i) .p/ is absolutely continuous; hence, by irreducibility .p/ has no multiplic-
ity, so that U./ act as translations on .p/ and the absolute continuity of
the spectral measures of p with respect to the Lebesgue measure implies that
U./ is weakly continuous. Then, by the SvN theorem the representation is
equivalent to the Schrdinger representation and there is no ground state;
ii) .p/ is a pure point spectrum; then, if is a state corresponding to an
eigenvector of p with eigenvalue , one has, 8A 2 AW , .A V.// D
ei .A/; and
.U./V.// D .V. /U./ V./V. // D ei .U./ V.//;
i.e.
.U./V.// D 0; 8 0; .V.// D ei :
In conclusion, the representation is unitarily equivalent to the non-regular

representation of a translationally invariant state, Proposition 2.1. Then,
Eqs. (2.3.2), (2.3.3) hold;
iii) .p/ is purely singular, to be discussed later.
2) The spectrum of V.a/ and of the Hamiltonian.
In such Pa representation the potential can be written in the following form
W.x/ D n vn U.2n=a/, where vn are the coefficients of the Fourier expansion
of the periodic function W.x/ in 0; a and the series is strongly convergent on D.
For the analysis of the spectrum of H, since V.a/ D eipa commutes with H, it is
convenient to decompose K according to the spectrum of V.a/, which is purely
discrete and coincides with the whole circle, (Eq. (2.3.5)).
Then,
X X
.x/ D cn U.n / 1 D cn exp .in x/ 2 H ;
n n
implies V.a/ .x/ D ei .x/ and on the other hand, by the Weyl commutation
relations,
X X
V.a/ cn U.n / 1 D cn ein .xCa/ :
n n
Hence, one must have n D =a C 2n=a, n 2 Z and .x/ can be written as in
Eq. (2.3.6), i.e. .x/ is a quasi periodic function of the form of Eq. (2.3.1), with
k D =a.
3) The unique ground state belongs to H D0 .
Since V.a/ commutes with H, the subspaces H reduce K and in H the
Hamiltonian reduces to H D H0; C W.x/, H0; D p2 =2m. Since W.x/ is
bounded, it is a bounded operator in each H and therefore it is infinitesimally
smaller than H0; in the sense of Kato, (i.e. jjW jj ajjH0; jj C bjj jj,
with inf a D 0). Since the spectrum of H0; is discrete, so is the spectrum of H ;
this implies that ground states exist.
Moreover, the boundedness of W.x/ in H implies that eH has a strictly
positive kernel, i.e. eH .x/ > 0, 8 0. Now, if 0 2 H is the
ground state, it must have a non-vanishing projection 0; on at least one H ,
corresponding to inf .H /. By a generalized Perron-Frobenius theorem the
corresponding wave function may be chosen strictly positive
H 3 0; .x/ D ei' j 0; .x/j; ' 2 R; j 0; .x/j 0; a:e: (2.3.7)
Since j 0; .x/j is a periodic function, it belongs to H D0 and Eq. (2.3.7) is

consistent only for D 0. The ground state is unique because any other ground
state would be described by a positive wave function and therefore could not be
orthogonal to 0 .6
iii) The possibility of a representation characterized by a purely singular
.p/ is excluded by the following argument. As in the previous case, one
may decompose the Hilbert Rspace K of such a representation according to
the spectrum of V.a/, K D d
./ K ; with a measure d
./ now singular
with respect to the Lebesgue measure and translationally invariant because
6
For the permanence of a discrete spectrum under a bounded perturbation see M. Reed and
B. Simon, Methods of Modern Mathematical Physics, Vol. IV, Academic Press 1972, Theors.
XII.11, XII.13. For the generalized Perron-Frobenius theorem see J. Glimm and A. Jaffe, Quantum
Physics. A Functional Integral Point of View, Springer 1987, Sect. 3.3; for a simple account see F.
Strocchi, An Introduction to the Mathematical Structure of Quantum Mechanics, 2nd ed., World
Scientific 2008, Sect. 6.4.
Q
U./ U.=a/ maps K into K C . In each space K corresponding to such
a decomposition, the spectrum of p is discrete and coincides there with the
spectrum in the subspace H of the preceding case. Then, the representation
of the subalgebra generated by V./, 2 R and by U.2n=a/, n 2 Z, is
quasi equivalent to that in H . The Hamiltonian H is therefore defined in K
and has the same spectrum as in H ; by the same arguments if inf .H/ is
an eigenvalue the corresponding eigenvector must belong to K D0 , which must
therefore appear as a discrete component of K. Since the spectrum of p is purely
discrete in K D0 , by irreducibility it is purely discrete in K and has no singular
component there.
The above theorem allows to recover in a simple (mathematically rigorous) way
the basic features of the analysis of the energy spectrum in the case of periodic
potentials.7
a) Band structure. The energy spectrum fEn ./g is characterized by bands,
classified by the quantum number n 2 Z; within each band the energy levels
are functions of the parameter 2 0; 2=a/.
b) Description in terms of the elementary cell. Equation (2.3.6) defines an
isomorphism between H and L2 .0; a/; dx=a/, so that the scalar product in H
reduces to an L2 product with integration over the elementary cell 0; a. In this
identification, p is represented by the self adjoint extension of the differential
operator id=dx, corresponding to the boundary conditions .a/ D ei .0/.
The generic function 2 K can be expressed as a denumerable superposition
of k
X
.x/ D c.k / k .x/
k
and since . ; ' 0 / D 0 if 0 , the product (2.3.3) reduces to a sum of

products in L2 .0; a; dx=a/, i.e. in the elementary cell.
Q
The energy is a continuous function of , since U./ W H D0 ! H , so that
. ; H / D. 0; .H C p=m C 2 =2m/ 0 /:
7
M. Reed and B. Simon, Methods of Modern Mathematical Physics, Vol. IV, Academic Press,
Sect. XIII.16.
4 Gauge Invariance and Non-regular Canonical

Quantization
4.1 Gauge Invariance and Superselection Rules
A general case leading to non-regular representations is when (i) a quantum system

is described by canonical variables, generating a Heisenberg group GH , but only
a subset of them, and consequently only a subgroup Gobs GH , describes
observable quantities, (ii) Gobs is generated by a Heisenberg subgroup and by an
abelian subgroup G which commutes with Gobs . Then, G generates a group of
transformations g , g 2 G, which leave the observables pointwise invariant
g .A/ D A; 8A 2 Gobs ; 8g 2 G; (2.4.1)
i.e. G has the meaning of a gauge group,.

The elements of GH generate a C -algebra FW , called field algebra, and the
elements of Gobs generate a C -algebra A of observables, characterized by gauge
invariance, Eq. (2.4.1) (as discussed in Chap. 1, Sect. 3). A has a non-trivial center
Z generated by the elements of G. A representation of FW is physical if Gobs is
regularly represented.
In the irreducible representations of A, Z is represented by multiples of the
identity. The generators of G have the meaning of superselected charges and the
points of the spectrum .Z/ of Z label inequivalent representations .H ; / of
A, called sectors. Stone-von Neumann uniqueness theorem does not apply and
this can be traced back to the fact that, contrary to the Weyl C -algebra, A is not
simple.
By definition a gauge invariant state ! on FW satisfies !.g .F// D !.F/, 8F 2
FW and therefore, in the GNS representation ! of FW defined by !, the gauge
transformations are implemented by unitary operators U.g/ defined by (! denotes
the vector which represents !)
U.g/! D ! ; U.g/! .F/! D ! .g .F// ! ; 8F 2 FW :
Let V.g/ denote the element of G which defines g : g .F/ D V.g/ F V.g/1 ,
8F 2 FW ; then, ! .V.g//U.g/ commutes with FW and, in each irreducible
representation of FW , ! .V.g// U.g/ D ei.g/ 1. Hence, ! is an eigenvector of
! .V.g//, with eigenvalue ei.g/ ,
! .V.g//! D ! .V.g//U.g/! D ei.g/ ! : (2.4.2)

4 Gauge Invariance and Non-regular Canonical Quantization 21
Thus, the analysis of Sect. 2, applies, with the result that the GNS representation
! of FW , equivalently of GH , defined by a gauge invariant state ! is non-regular,
X
H! D H ;
2 .Z/
and the subspaces H carrying disjoint irreducible representations of A are proper

subspaces of the non-separable space H! .
Proposition 4.1 Let GH be the Heisenberg group defined by the set of canonical
variables fqi ; pi g, FW the corresponding canonical C -algebra, A FW the C -
subalgebra of observables and G the commutative group of gauge transformations,
defined by a subgroup G GH .
Then, the GNS representation of FW defined by a gauge invariant state is a
non-regular representation of FW , as well as of the Heisenberg group GH , and the
elements of G define superselection rules.
Relevant examples of such a structure are provided by quantum mechanical
models, in particular those exhibiting strong analogies with gauge quantum field
theories (see in particular the following chapter).
4.2 Gauge Invariance in the Two-Body Problem
The description of the quantum two-body problem is provided by the Weyl field
algebra FW generated by the exponentials u./; v./ of the center of mass canonical
variables Q; P, and by the exponentials U./, V./ of the relative canonical
variables q; p. The Hamiltonian has the form
H D P2 =2M C p2 =2 C V.q/ (2.4.3)
and, for the purpose of discussing the bound state spectrum and in particular
the lowest energy level, the position of the center of mass is irrelevant. It is
therefore natural to consider as observable C -algebra A the algebra generated by
the canonical variable q; p; P. Since the center of mass position is not observed, the
translations v./ of the center of mass have the meaning of gauge transformations.
Therefore, the representations of the canonical field algebra FW defined by a
gauge invariant state ! are characterized by the property that the vector ! , which
represents !, is an eigenvector of v./, equivalently of P. In particular, the lowest
energy state !0 must satisfy !0 .P2 / D 0, so that the corresponding vector 0
satisfies
0 D .0 ; P2 0 / D jjP 0 jj2 ; i.e. P 0 D 0: (2.4.4)

As discussed in Sect. 2, the representations of the canonical algebra FW defined

by gauge invariant states are non-regular; in fact, Eq. (2.4.4) is incompatible
with the canonical commutation relations in the Heisenberg form. It has been
suggested to bypass such incompatibility by allowing 0 to be non-normalizable.8
In our opinion, such a choice would have catastrophic consequence on the GNS
representation defined by such a ground state; by the cyclicity of 0 all vectors of
such a representation would be non-normalizable, all matrix elements (including the
ground state expectations of gauge invariant operators) would be divergent and one
could not extract finite results in a consistent mathematical way.
However, Eq. (2.4.4) is compatible with the CCR in Weyl form. Thus, a canonical
quantization is not forbidden, provided it is done in terms of the Weyl algebra, rather
than of the Heisenberg algebra; actually, it is uniquely determined and coincides
with the non-regular representation discussed in Sect. 2. The vector states are not
represented by square integrable functions on the spectrum of Q, but one can still
describe them by wave functions of the center of mass position by using a non-L2
scalar product (see Eq. (2.2.4)).
In our opinion, from a mathematical point of view, the non-regularity of the
representation is a much better price to pay, rather than living with non-normalizable
state vectors. The advantages of such a quantization is that the states are described
by normalizable vectors of a Hilbert space, the basic quantum mechanical rules are
not violated, the observable subalgebra A is regularly represented in the standard
way, the canonical variables which are not gauge invariant are non-regularly
represented, only their exponentials being well defined.
The quantization discussed above sheds light on the quantization of gauge field
theories, in particular on the quantization of the temporal gauge.9
4.3 Non-regular Representations and Symmetry Breaking
We briefly recall that, given a C -algebra A, an algebraic symmetry is an automor-

phism of A; given a state !, the symmetry is unbroken in the corresponding
representation space if is implemented by a unitary operator T./ there, i.e.
! ..A// D T./ ! .A/ T./ ; 8A 2 A: (2.4.5)
This means that the representation defined by the state ! , ! .A/ !..A// is
unitary equivalent to ! :
! .A/ D T./ ! .A/ T ./:
8
R. Jackiw, Topological Investigations of Quantized Gauge Theories, in S.B. Treiman, R. Jackiw,
B. Zumino and E. Witten, Current Algebra and Anomalies, World Scientific 1985.
9
J. Lffelholz, G. Morchio and F. Strocchi, J. Math. Phys. 44, 5095 (2003).
In this case, gives rise to a Wigner symmetry in H! , i.e. all transition amplitudes
are invariant. Otherwise, if there is no unitary operator which implements in H! ,
by Wigner theorem on symmetries at least one transition amplitude is not invariant
and the symmetry is said to be broken in H! . An algebraic symmetry is said to be
regular if it maps regular representations into regular ones.10
In the case of quantum systems described by the canonical Weyl algebra, any
regular algebraic symmetry is unbroken in any regular irreducible representation,
since, by Stone-von Neumann theorem, all such representations are unitarily
equivalent. Thus, the important phenomenon of symmetry breaking, in the strong
sense of a loss of symmetry as defined above, (not merely as the non-invariance
of the ground state) cannot appear in the case of Heisenberg quantization, more
generally in the case of regular Weyl quantization.
The situation drastically changes in the case of non-regular Weyl quantization.
A distinguished case is when one has the structure discussed in Sect. 4.1, namely
a canonical algebra FW and an observable (gauge invariant) subalgebra A, with a
non-trivial center Z A.
Clearly, any symmetry of A, defined by an element of FW , is implemented
by a unitary operator T./ in the non-regular representation of FW , defined by a
gauge invariant state ! , 2 .Z/.
However, if does not commute with the gauge group G, is broken in each
irreducible representation H of the observable subalgebra A, i.e. fails to define a
Wigner symmetry of the gauge invariant states of H D A ! , because T./ does
not leave H invariant.
In the regular irreducible representation, r of FW , the symmetry is unbroken
but the elements of Z have a continuous spectrum in Hr and there is no gauge
invariant (proper) state vector in Hr .
Proposition 4.2 Let FW denote the canonical field C -algebra defined by a
Heisenberg group GH , A the observable C -subalgebra, Z the non-trivial center
of A generated by the commutative subgroup G GH (gauge group), then
i) any algebraic symmetry of A, defined by an element of GH which does not
commute with G, is spontaneously broken in each irreducible representation of
A ( sector);
ii) in any representation of FW defined by a gauge invariant state !, the one-
parameter subgroups which do not commute with G are non-regularly repre-
sented, so that the corresponding generators cannot be defined as operators in
H! , only their exponentials exist.
For representations of A defined by a ground state !0 , (more generally by a state
! invariant under time translations), the non-invariance of !0 ,
< A > !0 .A/ !0 ..A//; for some A 2 A; (2.4.6)
10
For a discussion of the meaning and the mechanism of spontaneous symmetry breaking see: F.
Strocchi, Symmetry Breaking, 2nd ed., Springer 2008.
is still compatible with giving rise to a Wigner symmetry in the GNS representa-
tion space H!0 . In this case, if commutes with the dynamics, Eq. (2.4.6) implies
degeneracy of the ground state. This is what happens if (2.4.6) holds for defined
by an element of the field algebra FW which commutes also with the gauge group.
A one-parameter group , 2 R, of symmetries shall be called a continuous
symmetry. A symmetry is called internal if it commutes with the one-parameter
group t , t 2 R, of the time translations. In the following, the breaking of an internal
symmetry shall be called spontaneous symmetry breaking.
4.4 Goldstone Theorem and Non-regular Representations
The spontaneous breaking of a continuous symmetry in the quantum theory of

infinitely extended systems is usually accompanied by a strong constraint on the
energy spectrum; in fact, if the symmetry commutes with the dynamics (i.e. if the
Hamiltonian is symmetric) the Goldstone theorem predicts the absence of an energy
gap with respect to the ground state, in the channels related to the ground state by
the broken generators.11
Here, we investigate a possible quantum mechanical version of the Goldstone
theorem which mimics as closely as possible the formulation and proof for infinitely
extended quantum systems.
For this purpose, given a C -algebra A, a one-parameter group , 2 R of
automorphisms of A and a representation of A defined by a ground state !0 , we
consider :
i) the infinitesimal variation of a generic element F D .A/, A 2 A,
d . .A//
F D ..A// D jD0 ;
d
ii) the generation of the continuous symmetry by elements of the strong closure
.A/00 of .A/, in the sense that there is a sequence Qn D Qn 2 .A/00 ,
n D 1; ::, such that
F D i lim Qn ; F :
n!1
If Qn converges weakly to a self adjoint operator Q, then is implementable by

the unitary operator eiQ , the symmetry is not broken and < F > !0 .F/ 0
implies that !0 is not invariant, i.e. Q0 0. Furthermore, if commutes
with the time translations t , eiQ 0 is a family of degenerate ground
11
For a review and critical discussion of the Goldstone theorem see F. Strocchi, Symmetry Breaking,
2nd ed., Springer 2008, Chap. 15.
states. In this case, one gets a picture close to the standard heuristic formulation
of spontaneous breaking of a continuous symmetry, based on the following
oversimplified assumptions: (i) the continuous symmetry is generated by a
charge Q, in the sense that F D i Q; F , (ii) the Hamiltonian is symmetric,
i.e. Q; H D 0, (iii) < F > 0; the conclusion being that Q0 0 has zero
energy.
Closer to the infinite-dimensional case is the case in which there is no sequence
Qn , with the above property, which converges weakly to a self-adjoint operator
Q; then, if < F > 0, the symmetry is broken in the strong sense of loss of
symmetry and < F > is the strict analog of a symmetry breaking order parameter,
which characterizes symmetry breaking in quantum field theory or in many body
theory. Similarly, < F >D i limn!1 < Qn ; F > plays the role of the symmetry
breaking Ward identity.
If commutes with t , then, by the invariance of the ground state under t , one
has
lim < Qn .t/; F >D lim < t .Qn /; F >D

n!1 n!1
lim < Qn ; t .F/ >D id < .t .F// > =djD0 D id < .F/ > =djD0

n!1
D lim < Qn ; F > :

n!1
It is worthwhile to stress that such a time independence of the Ward identity holds
also in the more general case in which the symmetry does not commute with the
Hamiltonian, limn!1 Qn .t/; H D A 0, but < A; F >D 0 (in analogy with the
so-called anomaly occurring in the infinite-dimensional case). This is, e.g., the case
in which the Hamiltonian is invariant up to a time derivative which commutes with
F (see the example of the Bloch electron discussed below).
Theorem 4.3 Let , 2 R, be a one-parameter group of automorphisms of the
algebra A, t the one-parameter group of time translations and the representation
defined by a ground state !0 . If for some F 2 .A/,
< F > d < .F/ > =djD0 0;

< F >D i lim < Qn ; F >D i lim < Qn .t/; F >; (2.4.7)
n!1 n!1
for a suitable sequence of Qn D Qn , Qn .t/ t .Qn /, the limit being understood in
the sense of convergence of tempered distributions in the variable t, then there is no
energy gap above the ground state. Actually, there is a state (Goldstone-like state)
orthogonal to the ground state, with the ground state energy.
Proof It is enough to consider the case in which F D F , since if F D F1 C iF2 ,
Fi D Fi , i D 1; 2, by linearity the symmetry breaking condition must hold
for at least one Fi . Since the representation is defined by a ground state, .t/ is
implemented by a one-parameter group of unitary operators U.t/; t 2 R, with
the generator normalized so that the ground state has zero energy. Without loss
of generality one can assume that < Qn >D 0, since Eq. (2.4.7) holds also for
QQn Qn < Qn > : Then, one has
2 Im lim < Qn U.t/F >D i lim < Qn ; F > 0: (2.4.8)

n!1 n!1
The distributional convergence of Jn .t/ 2 Im < Qn U.t/F > and Eq. (2.4.7)
imply the following distributional convergence of the Fourier transforms JQn .!/
lim JQn .!/ D< F > .!/:

n!1
R
Then, by using the spectral representation U.t/ D ei!t dE.!/, one concludes that
the energy spectral measure contains a .!/.
The ground state !0 cannot be responsible for such a point spectrum, since its
contribution as intermediate state in the right hand side of Eq. (2.4.7) vanishes as
a consequence of < Qn >D 0; hence there is a zero energy eigenvector orthogonal
to the ground state vector.
Remarks A few remarks may be useful.
The statement that the infinitesimal variations under the symmetry transforma-
tions is given by a limit of commutators with charges Qn does not require that, in
the given ground state representation, is implemented by a weakly continuous
group of unitary operators.
In the infinite-dimensional cases of quantum field theory and of many body
theory, the generation of the symmetry is through the commutator of local charges,
typically the integrals of the charge density j0 .x; t/ of a conserved current j .x; t/
(@ j D 0):
Z
F D i lim QR ; F ; QR D dx j0 .x; t/
R!1 jxjR
and < F > 0 implies that the commutator QR ; F does not converge to the
commutator of a charge Q; therefore QR is not weakly convergent to a well defined
global charge Q. Hence, the generation of the symmetry can only be expected to
occur as a limit of commutators of (not weakly converging) local charges.
It is worthwhile to stress that the non-invariance of the ground state expectation
of a field F does not guarantee that one can write a corresponding Ward identity, a
crucial ingredient for the Goldstone theorem.
The interplay between gauge invariance and the breaking of a continuous
symmetry provides a mechanism for evading the conclusions of the Goldstone
theorem, i.e. for allowing an energy gap in the presence of symmetry breaking.
In fact, let us consider the case in which
i) the continuous symmetry is defined by a (one-parameter) subgroup of the
Heisenberg group GH , which does not commute with the gauge transformations,
ii) in the ground state irreducible representation of the observable algebra A

there is a gauge invariant operator F 2 .A/, which yields a non-symmetric
order parameter < F > 0, and
iii) the Hamiltonian is invariant under up to a time derivative which commutes
with F,
then the conclusions of the Goldstone theorem do not apply by the following
mechanism.
In the irreducible regular representation r of the field algebra FW , is
implemented by a (weakly continuous) group of unitary operators T./, all the
matrix elements are invariant, but there is no gauge invariant (proper) vector state
invariant under time translations. The symmetry gets broken by the direct integral
decomposition of Hr over the spectrum of Z, but one cannot write a symmetry
breaking Ward identity for the expectation on the gauge invariant ground state.
On the other side, in the representation of FW defined by a gauge invariant ground
state ! , the one-parameter group T./ is not regularly represented. Therefore its
generator cannot be defined as an operator in H!0 and ! .F/ 0 cannot be written
in terms of a limit of commutators of charges. In conclusion, the symmetry breaking
Ward identity cannot be written in terms of expectations on states.
Such a mechanism is active in several quantum mechanics gauge models
discussed in Chap. 3, as well as in the interesting case of chiral symmetry breaking
in QCD, as discussed in Chap. 4.
4.5 Bloch Electron as a Gauge Model
The field algebra FW is generated by the Weyl operators U./; V./, ; 2 R (we
keep considering the one-dimensional case).
The periodic structure of the system leads to consider as observable C -algebra
A the sub-algebra generated by V./ and by the periodic functions of the position
U.2n=a/, n 2 Z. The center Z of A is generated by the translations V.a/ and the
irreducible representations of A are defined by the subspaces H ( sectors).
The operators U.=a/; 2n intertwine between the inequivalent represen-
tations and C and the corresponding one-parameter group is non-regularly
represented in the representation of FW defined by the gauge invariant ground
state 0 .
The Bloch model has been discussed in order to clarify structures and mecha-
nisms argued to characterize Quantum Chromodynamics (QCD).12 For the analo-
gies and correspondences we remark that the lattice translations V.na/ play the
role of the large gauge transformations Tn and the sectors H correspond to the
12
R. Jackiw, Topological Investigations of Quantized Gauge Theories, in S.B. Treiman, R. Jackiw,
B. Zumino and E. Witten, Current Algebra and Anomalies, World Scientific 1985, p. 211359,
Sect. 3.5.
representations defined by the vacua, (here all the states D U./0 have
higher energy than 0 ). The transformations on FW defined by the one-parameter
Q
group U./ Q
U.=a/, 2 R: .F/ U./F Q
U./ 1
, 8F 2 FW , correspond to
the chiral transformations.
The transformations are implemented by unitary operators in the space H
carrying an irreducible representation of the (gauge dependent) field algebra FW ;
therefore they define Wigner symmetries there, but they do not leave the sectors
invariant and therefore they are not implemented by unitary operators there. The
corresponding symmetry is spontaneously broken in each sector.
An explicit symmetry breaking order parameter is provided by p or by V.na/,
since .p/ D p and . ; p / D 0: The Hamiltonian is invariant up to
a time derivative:
.H/ D H p=m:
The equations expected to hold in QCD are rigorously reproduced here
Q
U./ D C ; Q
V.na/U./V.na/ 1 Q
D ein U./:
In the QCD context, the last equation is usually written in terms of the chiral charge
Q5 , which is assumed to generate the chiral transformations,
Tn Q5 Tn1 D Q5 C n;
however, it should be stressed that the generator of the chiral transformations does
not exist, not only in the sectors, but not even in the large Hilbert space H, because
Q
U./ is non-regularly represented.
Thus, one cannot write a symmetry breaking Ward identity for the expectations
on states. The overlooking of this subtle point is at the basis of problems and
paradoxes affecting the use of Ward identities in the temporal gauge of QCD.
The Bloch model clearly displays the fact that the crucial ingredient for the
breaking of chiral symmetry with energy gap (the so-called U.1/ problem) is the
existence of a non-trivial center in the algebra of observables and its pointwise
instability under chiral transformations.13
13
For the realization and relevance of this structure see F. Strocchi, Selected Topics on the General
Properties of Quantum Field Theory, World Scientific 1994, Sect. 7.4 and refs. therein; G. Morchio
and F. Strocchi, J. Phys. A: Math. Theor. 40, 3173 (2007); Ann. Phys. 324, 2236 (2009).
5 Quantum Hall Electron: Zak States 29
5 Quantum Hall Electron: Zak States
The quantum mechanical behavior of an electron in a periodic lattice in the presence

of a constant magnetic field is particularly interesting also in connection with the
quantum Hall effect.
a) Bloch electron in a constant magnetic field
The Hamiltonian has the following form
2 e
HD C W.x/; i pi Ai ; i D 1; 2; 3; (2.5.1)
2M c
where M denotes the electron mass and W.x/ is a bounded measurable periodic
potential reflecting the lattice periodic structure in the x y plane.
We adopt the symmetric gauge, so that the electromagnetic potential Ai is given
by Ai D 12 "ijk Hj xk , we take the magnetic field H in the z-direction and consider the
motion in the xy-plane. For simplicity, we shall use units such that D 1 D M D
ejHj=c, so that the cyclotron frequency !c D ejHj=Mc and the magnetic length
l D .c=ejHj/1=2 are both equal to one. Then, one has
x D px y=2; y D py C x=2; y ; x D i: (2.5.2)
has the meaning of the (gauge invariant) velocity.

The lattice translations on the xy-plane are described by the operators
T.aj / eic aj ; j D 1; 2; c p C eA=c; (2.5.3)

c x ; c y D i; c ; D 0; (2.5.4)
where the vectors aj are the lattice basis. The operators T.aj /, also called magnetic
translations, commute with the Hamiltonian and satisfy the following commutation
relation
T.a1 / T.a2 / D T.a2 / T.a1 / ei.a1 x a2 y a1 y a2 x / : (2.5.5)
Thus, they commute if the lattice cell satisfies the rationality condition14
a1 x a2 y a1 y a2 x D 2k; k 2 Z: (2.5.6)
In the following for simplicity we shall consider a square lattice with unit lattice
spacing.
14
For a detailed excellent discussion of the magnetic translation group see E. Brown, Aspects of
Group Theory in Electron Dynamics, in Solid State Physics, F. Seitz et al. eds., Academic Press
1968, pp. 313408.
The abelian group generated by the Tj T.j /, with the condition of Eq. (2.5.6),
will be denoted by G. It plays the same role of the group of lattice translations
V.na/ of the Bloch electron without magnetic field (see Sect. 4.5). It may therefore
be given the meaning of an abelian gauge group.
In view of the above symmetry properties, it is convenient to describe the systems
in terms of the two pairs of canonical (independent) variables
q y ; p x I Q c x ; P c y : (2.5.7)
The corresponding Heisenberg group GH is generated by the exponentials
u./ ei q ; v./ ei p I U. / ei Q ; V./ eiP :
We denote by FW the corresponding field C -algebra. The elements u./, v./,

U.n/, V.2m/ generate the gauge invariant subgroup Gobs , which can be interpreted
as the observable subgroup. The corresponding C -algebra is denoted by A, with the
meaning of the C -algebra of observables. A has a non-trivial center Z generated
by the elements of the gauge group G, which play the role of the large gauge
transformations of QCD.
We start by considering the case of W D 0. In terms of the above canonical
variables one has
H0 D 12 .p2 C q2 / D Hosc C 12 Lz ; (2.5.8)

Hosc 12 p2x C p2y C 14 .x2 C y2 / D 14 .p2 C q2 C P2 C Q2 /; (2.5.9)
Lz x py y px D 12 .p2 C q2 / 12 .P2 C Q2 /: (2.5.10)
Lz is conserved, but it does not commute

p
with thepgauge group G generated by the
large gauge transformations T1 ei 2Q , T2 ei 2P .
The spectrum of H0 is the familiar quantum oscillator spectrum, each level being
now infinitely degenerate. For a very large magnetic field one may restrict the
attention to the first Landau level (LL) corresponding to the lowest energy states
of H. For the description of the degeneracy of the first LL one has many options.
One possibility, used for the discussion of Quantum Hall Effect, is to describe
such a degeneracy in terms of eigenstates of Lz or of Hosc . With such a choice, a
state (of the first LL) with maximum xy localization is
2 Cy2 /=4
0 0 .2/1=2 e.x ; Lz 0 0 D 0
and it is also the ground state of the harmonic oscillator Hamiltonian Hosc .Q; P/
2 2
2 .P C Q /.
1
A complete set of states for the first LL level if obtained by acting on 0 0 by the
magnetic translations
m n .x; y/ T.a1 /m T.a2 /n 0 0 .x; y/:

Since, apart from (xy dependent phases) the magnetic translations act as lattice
translations on the wave functions, the m n defined above are peaked at the lattice
points.
In this way, one gets a regular representation of the Heisenberg group GH which,
however, does not contain gauge invariant states, i.e. states invariant under the gauge
group G.
By an argument similar to that repeatedly used before (see e.g. Proposition 2.1)
a gauge invariant state ! defines a non-regular representation of the Heisenberg
group GH (or of the exponential field algebra FW ). Its representative cyclic vector
! is an eigenstate of Tj
Tj ! D eij ! ; j 2 0; 2/; j D 1; 2: (2.5.11)
Such states are the so-called Zak states15 !1 2

p
!1 2 .U. / V.// D ein1 eim2 ; if .; / D 2.n; m/
p
D 0; if .; / 2.Z; Z/: (2.5.12)
Clearly, the introduction of the periodic potential does not change such con-
clusions and actually strengthens the interpretation of the lattice translations as
the generators of a gauge group, with a picture which is very close to the case
of the Bloch electron without magnetic field. The first LL is not stable under the
application of u./; v./ and of the potential W.
As in case of zero magnetic field, the GNS representation space K defined by
a gauge invariant state has an orthogonal decomposition over the spectrum of the
generators of the gauge group
X
KD H1 ; 2 ; (2.5.13)
1 ; 2
each H1 ; 2 being the carrier of an irreducible representation of the gauge invariant
algebra of observables A.
The operators U. /, V./ intertwine between the sectors
p
U. / V./ H1 ;2 D H1 ;Q 2 CQ ; Q
.Q ; / 2.; /:
They play the same role of the U./ discussed in Sect. 4.5 and as such are the
analogs of the chiral transformations in QCD; they commute with the Hamiltonian
if the potential W (which plays the role of the fermion mass term in QCD) vanishes.
15
J. Zak, Phys. Rev. 168, 686 (1968).
They define Wigner symmetries in K which are broken in each irreducible repre-
sentation H1 ; 2 of the observable algebra. Such a breaking does not require the exis-
tence of Goldstone-like states by the same mechanism discussed in Sects. 4.3, 4.5.
The potential W is a periodic function of x D q P and y D Q p and since it
commutes with Tj it must be a function of them16 and therefore in each sector H1 ;2
it reduces to a periodic function of q 2 , Q 1 .
As in the case of vanishing magnetic field, in each sector W is infinitesimally
smaller than H0 12 .p2 C q2 / in the sense of Kato; therefore, in each sector, since
the spectrum of H0 is discrete, so is the spectrum of H D H0 C W, and, by a
generalized Perron-Frobenius theorem as in the case of zero magnetic field (Sect. 3,
Proposition 3.1), the ground state is unique. This implies that the spectrum of H is
discrete in K and there is a (possibly degenerate) ground state in K.
Particular instructive is the simple case of a periodic potential described by
p p p p
W D cos. 2 x/ cos. 2 y/ D cos. 2.q P// cos. 2.Q p//:
p p
In the sector H1 ;2 , W reduces to cos. 2q 2 / C cos. 2p 1 / and to first
order in the inf of the spectrum of H in H1 ;2 is
E0 .1 ; 2 / D 1
2 C e .cos.1 / C cos.2 //: (2.5.14)
Thus, for negative the minimum in K is obtained for i D 0, and for positive for
i D , i D 1; 2.
In conclusion one has
Proposition 5.1 A gauge invariant state !, i.e. a state invariant under magnetic
translations, T1 , T2 , defines an irreducible representation .H1 ;2 ; 1 ;2 / of the
observable algebra A, labeled by the eigenvalues 1 ; 2 of the magnetic translations
and non-regular representation .K; / of the Heisenberg group GH (and of the field
algebra FW ).
The (non-separable) Hilbert space K has an orthogonal decomposition over the
spectrum of the magnetic translations, Eq. (2.5.13).
The Hamiltonian H, Eq. (2.5.1), has a discrete spectrum in K and, at least for
small periodic potential, a unique ground state belonging to the 1 D 2 D 0
sector.
It it worthwhile to remark that in the representation defined by a gauge invariant
state the one-parameter groups of unitary operators U. /, V./ are not regularly
represented, so that Q; P cannot be defined as operators in K, only their exponentials
exists in K.17
16
J. Zak, Phys. Rev. 168, 686 (1968).
17
With such a proviso, some of the paradoxes raised in the literature (see, e.g., R. Ferrari, Int. Jour.
Mod. Phys. 12, 1105 (1998)) disappear. In particular, the derivatives with respect to the angles 1 ,
2 correspond to the momenta canonically conjugated to Q, P respectively, and therefore cannot
be defined in K.
In our opinion, as in the case of zero magnetic field (Sect. 3), the description
in terms of the representation given by the Hilbert space K clarifies the meaning
and the role of the boundary conditions used in the literature for the wave function
restricted to the primitive cell,18 as a substitute of Eq. (2.5.11); such boundary
conditions are unstable under the action of the unitary operators U. /, V./, which,
instead, are well defined in K. Thus, the description in terms of the states of K
already takes the infinite volume limit into account.
b) Quantum Hall electron
In the Quantum Hall Effect (QHE) each electron lives in a periodic lattice under
the influence of both a constant strong magnetic field, say in the z direction, and a
constant electric field E in the xy plane.
It is convenient to choose the symmetric temporal gauge for the vector potential
Ai D 12 "ijk Hj xk eEi ;
so that the motion of an electron in the xy plane is described by the following time
dependent Hamiltonian
Q 2 C W.x/;
H.t/ D 12 Q i pi eAi =c D i eEi :
(2.5.15)
By means of a Galilei transformation19 one can shift the dependence on the electric
field from the kinetic term to the periodic potential, obtaining the following new
Hamiltonian (by introducing the dual E of E, E1 E2 ; E2 E1 )
H 0 .t/ D 12 2 C W.x e.E C E t//: (2.5.16)
The only difference with respect to the case of zero electric field is that the periodic
potential has become time dependent.
Thus, most of the previous analysis applies. In particular, the algebraic structure
of the Heisenberg group GH , of the exponential field algebra FW , of the gauge group
G and of the gauge invariant or observable algebra A remain the same.
Gauge invariant states are analogously defined and the representations defined
by them have the same properties as in Proposition 5.1.
18
See, e.g. J. Zak, Phys. Rev. 168, 686 (1968); E. Brown, Aspects of Group Theory in Electron
Dynamics, in Solid State Physics, F. Seitz et al. eds., Academic Press 1968, pp. 313408; F.D.M.
Haldane, Phys. Rev. Lett. 55, 2095 (1985).
19
J. Belissard, Quantum systems periodically perturbed in time, in Fundamental aspects of
quantum theory, V. Gorini and A. Frigerio eds., Plenum Press 1986, p. 163171, and references
therein; R. Ferrari, Int. Jour. Mod. Phys. 12, 1105 (1998). For a comprehensive updated discussion
of the quantum Hall effect and of the physical principles underlying it see: S. Bieri and J.M.
Frhlich, Comptes Rendus Physique, 12, 332346 (2011).
Chapter 3
Quantum Mechanical Gauge Models
1 Quantum Particle on a Circle
The canonical variables which describe a quantum particle on a circle (of radius
R D 1) are the angle ' 2 0; 2/ and its conjugate momentum p. They are not
genuine Heisenberg variables, since one cannot write the corresponding canonical
commutation relations in Heisenberg form. In fact, '; p D i, would imply
'; eip D ieip , (the existence of eip , 8 2 R is given by the self-adjointness of
p), which yields jj'jj =2, 8, contrary to the constraint ' 2 0; 2 . One is then
led to consider a Weyl quantization based on the C -subalgebra A of the standard
Weyl (field) algebra FW , generated by U.n/ D ein' , n 2 Z, and V./ D eip , 2 R.
The canonical commutation relations read
U.n/ V./ D ein V./ U.n/: (3.1.1)
Actually, A may be characterized as the subalgebra of FW invariant under

the translations n of 2n, n 2 Z, which, therefore, get the meaning of gauge
transformations.
The structure is similar to that of the Bloch model and fits into the general
discussion of Heisenberg group GH , observable subgroup Gobs and gauge group G,
with corresponding C -algebras FW , A and a non-trivial center Z of A (see Chap. 2,
Sect. 4.1). A representation of A is regular if .V.// is a weakly continuous
group of unitary operators.
In each irreducible representation of A, the element V.2/ 2 Z is a multiple of
the identity, say ei , 2 0; 2/.
Theorem 1.1 For any given , all the irreducible regular representations of A
with .ei2p / D ei are unitary equivalent.
Proof We start by proving the Theorem for D 0. Any corresponding representa-
tion 0 satisfies

36 3 Quantum Mechanical Gauge Models
0 .W.0; // 0 .V.// D 0 .V. C 2//:
Then, the operator

Z 2
1
P .2/ d 0 .V.// D P ;
0
is well defined, satisfies
0 .V.// P D P
as a consequence of the periodicity of the operator valued function 0 .V.// and

cannot vanish because it would imply the vanishing of
Z 2
0 .U.n//PU.n// D .2/1 d ein 0 .V.//; 8n 2 Z;
0
i.e. the vanishing of the periodic unitary operator-valued function 0 .V.//.

Furthermore, one has
Z 2 Z 2
P0 .U.n/V.//P D .2/2 d d ein 0 .U.n/V. C C // D
0 0
Z 2
.2/1 dein 0 .U.n// P D n;0 P: (3.1.2)
0
Thus, P is a projection and the representation space H0 contains a (cyclic) vector
0 with the property
.0 ; 0 .U.n/ V.// 0 / D n;0 : (3.1.3)
By the GNS theorem (see Chap. 1, Sect. 3) all such representations are unitarily
equivalent.
For 0, we note that the automorphisms defined by
Q
.U.n// D U.n/; .V.// D ei V./; Q =2;
intertwines between and 0 : .A/ D 0 . .A//. Then, given two represen-

.1/ .2/
tations , , their unitary equivalence follows from that of the correspond-
.i/
ing 0 :
.1/ .1/ .2/ .2/

.A/ D 0 . .A// D U0 . .A//U 1 D U .A/U 1 :
The representations are the GNS representations defined by the states ! on A

defined by
1 Quantum Particle on a Circle 37
Q
! .U.n/V.// D n; 0 ei : (3.1.4)
The field algebra FW can be obtained as the enlargement of the observable

algebra A, such that the automorphisms become inner, i.e. are described by
Q Q 2 0; 1/, of the enlarged algebra. In fact, U./
(unitary) elements U./,
U.Q C n/ D U./ Q U.n/, 2 R and V./ generate the Heisenberg group, and
the corresponding C -algebra FW .
Proposition 1.2 In the representation of A, the Hamiltonian is well defined as
a strong limit of elements of A and given by
H D p2 =2m;
where p .p/. The spectrum of H is discrete:

Q 2 ; E0 .2m/1 ; n 2 Z:
En D E0 .n C / (3.1.5)
Each sector contains a lowest energy state, called vacuum, and the ground state
belongs to the D 0 sector.
Each vacuum defines a non-regular representation of the field algebra FW and
for 2 0; / it is given by
Q
! .U./ V.// D 0; if 0; ! .V.// D ei : (3.1.6)
Proof We start with the simpler case D 0. Then, Eq. (3.1.3) implies
.0 .V. / U.n// 0 ; 0 .V./ 1 0 / D 0, and since the linear span of the
vectors 0 .V. / U.n// 0, 2 R, n 2 Z, is dense,
0 .V./ 0 D 0 ; 0 .p/ 0 D 0; H0 0 D 0:
Furthermore, Eq. (3.1.1) gives
0 .p/ .V. / U.n// 0 D n .V. / U.n// 0;
so that the spectrum of H is given by Eq. (3.1.5), with Q D 0.

In the representation , the representative of the state ! defined by Eq. (3.1.4),
Q
satisfies .V.// D ei and therefore, for its extension to FW , one has,
8 2 R,
! .U./ V.// D ! .V. / U./V./V. // D ei ! .U./ V.//
and Eq. (3.1.6) follows (for 2 0; / the ! are the lowest energy states).
By Eq. (3.1.3) and the automorphism , the representation of A is the same
as that of the gauge invariant algebra of the Bloch model (Sect. 4.5) with a D 2.
The elements of H can be represented by quasi periodic wave functions as in
Eq. (2.3.6):
Q
X
.'/ D ei' an ein' ; fan g 2 l2 ;
n2Z
and the generator p of .V.// is the differential operator id=d' on quasi

periodic functions; the spectrum is well known and given by Eq. (3.1.5).
The strict correspondence with the Bloch model allows to draw similar analogies
with the structures argued to characterize the vacuum structure of quantum chro-
modynamics (QCD). The operators T n D V.2n/ play the role of the large gauge
transformations, the automorphisms play the role of the chiral transformations
and the vacua structure accounts for the chiral symmetry breaking with energy
gap (absence of Goldstone states), by the same mechanism discussed in Chap. 2,
Sect. 4.4, i.e. the impossibility of writing the symmetry breaking Ward identities in
terms of expectations on vacua.
2 Jackiw Model of Gauss Law Constraint
A characteristic feature of gauge theories is that one of the equations is the so-called
Gauss law constraint, e.g. in electrodynamics one has
G div E D 0; P D 0:
G
The time independent operator G is the so-called Gauss law operator.

The formulation of gauge theories, in particular the canonical formulation,
is conveniently done in terms of a gauge dependent field algebra FW and its
observable subalgebra A is characterized by its pointwise invariance under gauge
transformations, which are generated by the Gauss (law) operator G.
In abelian gauge theories, like electrodynamics, the gauge transformations are
related to an abelian Lie group, typically U.1/, and one has only one Gauss law
operator, which therefore belongs to the observable algebra. Thus, one has the same
structure discussed before, namely a field algebra FW , an observable subalgebra A
and a non-trivial center Z of A. Hence, the canonical quantization of such a structure
meets the same general problems discussed in the previous Chapter. In particular
the Gauss law constraint is incompatible with canonical commutation relations in
Heisenberg form. This has been the origin of several discussion and proposal in the
literature for overcoming this apparent difficulty at the basis of the quantization of
gauge theories.
2 Jackiw Model of Gauss Law Constraint 39
In his authoritative and excellent lectures on topological structures in gauge

theories1 Jackiw proposed quantum mechanical models for illustrating the role and
the implementation of the Gauss law constraint, implicitly suggesting the same
strategy advocated for the two-body problem he had discussed before.
As for the two-body problem, a mathematically more acceptable treatment of
such models is obtained by using non-regular representations rather than non-
normalizable (cyclic) vectors. This shall be argued below, by showing that Jackiw
models display the same general features discussed in Chap. 2, and therefore require
to adopt a non-regular Weyl quantization, the standard Heisenberg quantization
being mathematically precluded.
We discuss the following Jackiw model which mimics an interaction of two
particles with an electromagnetic potential A in 0 C 1 space-time dimensions and
in order to strengthen this interpretation we add the kinetic term for A. Thus, the
model is defined by the following Lagrangian:
L D 12 .Pq1 C eA/2 C 12 .Pq2 C eA/2 V.q1 q2 / C 12 AP 2 : (3.2.1)
The canonical variables are the two particle positions q1 ; q2 , the corresponding
conjugate momenta pi D qP i C eA, i D 1; 2, the electromagnetic potential A and
its conjugate momentum D qP E, playing the role of the electric field. The
Hamiltonian and the corresponding Hamilton equations are
H D 12 .p21 C p22 / e.p1 C p2 /A C 12 E2 C V.q1 q2 /;

qP i D pi eA; pP 1 D @V=@q1 D Pp2 ; qP D E; EP D e.p1 C p2 /:
The Lagrangian is invariant under the following gauge transformations .aCbt/ ,

a; b; t 2 R: .aCbt/ .qi / D qi C a C b t; .aCbt/ .A/ D A b=e. They imply the
invariance of the conjugate variables pi ; E as well as of the equations of motion; the
Hamiltonian is invariant up to the time derivative Eb=e.
The field algebra FW is generated by the Weyl operators U./, V./, u./, v./,
UA ./, VA ./, ; ; 2 R, corresponding to the exponentials of q .q1 q2 /; p
2 .p1 p2 /; Q 12 .q1 C q2 /; P p1 C p2 , A and E, respectively. The observable
1
subalgebra A is generated by (the exponentials of) q, p, P, E.

The gauge transformations a and b t are, respectively, generated by P and by
the time independent Gauss operator
G.t/ E.t/=e t.p1 C p2 / D E.0/=e; dG=dt D 0; (3.2.2)
(by the explicit appearance of the time variable G.t/ does not transform covariantly
under time translations). Clearly, the Gauss law constraint on the states in the form
1
R. Jackiw, Topological investigations of quantized gauge theories, in S.B. Treiman, R. Jackiw, B.
Zumino and E. Witten, Current Algebra and Anomalies, World Scientific 1985, pp. 211359.
G D 0, is incompatible with the canonical commutation relations in Heisenberg

form and therefore one must use Weyl quantization.
Proposition 2.1 Any state ! on the canonical field algebra FW satisfying the
Gauss law constraint !.FG/ D 0, 8F 2 FW , equivalently invariant under the
gauge transformations generated by the Gauss operator G, defines a non-regular
representation of FW :
!.u./ v./UA ./ VA . // D 0; if 0; or if 0: (3.2.3)
The one-parameter groups U./, V./, v./, VA . /, generated by q; p; P; E; are

regularly represented (so that the expectations of the gauge invariant variables are
well defined), but u./ and UA ./ are not, so that the corresponding gauge depen-
dent generators Q and A cannot be defined as operators in the GNS representation
defined by !.
Proof Let F.; ; ; / denote the operator in the expectation (3.2.3). For a gauge
invariant state !, the corresponding (GNS) representative vector ! must be an
eigenstate of G, then by using the canonical commutation relations in Weyl form,
one has (T.b/ ei b G ), 8b; t; 2 R,
!.F.; ; ; // D !.T.b/ F.; ; ; /T.b// D eib.tC/ !.F.; ; ; //
and Eq. (3.2.3) follows.

The Hamiltonian can also be written in the following form
H D 12 .p2 C P2 / C V.q/ ePA C 12 E2 H.q; p/ C H.P; A; E/:
The time evolution t of the observable variables q; p is the standard time evolution
of a particle in a potential V and for the gauge invariant variables P; E one has
t .P/ D P, t .E/ D E.0/ C e P t. Thus, as expected and needed, t maps the
observable algebra into itself.
For a gauge invariant state, the invariance under time translations requires
e2 t2 jjP ! jj2 C 2 e t .! ; E.0/ P ! / D 0;
i.e. P ! D 0. Then, if ! .A/ is irreducible, since the algebra generated by P; E is

abelian (and commutes with q; p), one has ! .P/ D 0, ! .E/ D e! .G/ D 1. In
such a representation the generator of time translations reduces to H.q; p/.
In any irreducible representation of the gauge invariant algebra the automor-
phisms defined by u./, UA ./ are broken by the same mechanism discussed in
Chap. 2, Sect. 4.4.
3 Christ-Lee Model 41
3 Christ-Lee Model
The problems of canonical quantization arising from the invariance of the

Lagrangian under time dependent gauge transformations have been discussed by
Christ and Lee2 in the analysis of the relation between the different gauges. In order
to clarify the nature of such problems they proposed a simple quantum mechanical
gauge model exhibiting the basic interplay between gauge invariance and canonical
quantization. The model has also been discussed by Jackiw in his lectures quoted
above.
In order to strengthen its relation with gauge theories we add a kinetic term for the
electromagnetic potential. Then, the model is defined by the following Lagrangian
(q2 q21 C q22 )
L D 12 .Pq21 C qP 22 / e.q1 qP 2 q2 qP 1 /A C 12 e2 A2 q2 C 12 AP 2 V.q/: (3.3.1)

p
By introducing the complex variables ' .q1 C i q2 /= 2, the Lagrangian takes the
following form
L D .Dt '/ Dt ' C 12 AP 2 V.' '/; Dt ' .@t i eA/';
which somewhat mimics the abelian Higgs-Kibble model in 0 +1 space-time

dimensions (with the term EP corresponding to the gauge fixing term 12 @ A ).
The Lagrangian is invariant under the following time dependent gauge transfor-
mations .aCb t/ ,
' ! eie .aCb t/ '; A ! A C b; a; b; t 2 R; : (3.3.2)
As a function of the canonical variables qi ; p1 D qP 1 C eAq2 , p2 D qP 2 eAq1 , A,

P the Hamiltonian reads
E D A,
H D 12 .p21 C p22 / C eA.q1 p2 q2 p1 / C 12 E2 C V.q/:
The operators
2q D ' '; j .q1 p2 q2 p1 / D i.Dt '/ ' ' Dt '
play, respectively, the role of a charge density and of a current density (with
Hamiltonian coupling e A j). It is easy to see that
dj dE
D i H; j D 0; D ej:
dt dt
2
N. Christ and T.D. Lee, Phys. Rev. D 22, 939 (1980); T.D. Lee, Particle Physics and Introduction
to Field Theory, Harwood Academic 1990, Chap. 18, Sect. 1A.
The gauge transformations a and b t are generated, respectively, by ej and by the

time independent Gauss operator
G.t/ D E.t/ C e j t D E.0/:
The field algebra FW is generated by the exponentials of the canonical variables

and a candidate for the observable algebra A is the algebra generated by (the
exponentials of) the gauge invariant operators p21 C p22 , E, , j, Dt ' Dt '. Thus,
as it happens in gauge quantum field theories, the observable algebra A is not
a subalgebra of the canonical (exponential) field algebra FW and therefore it is
convenient to introduce a larger field algebra, still denoted by FW , generated by
the exponentials of the canonical variables and by the exponentials of a set of their
gauge invariant polynomials, briefly called observable exponentials.
A representation of FW is regular if all the one-parameter groups defined
by such exponentials are weakly continuous, otherwise it is said to be non-
regular. A representation of FW is physical if the one-parameter groups defined
by the observable exponentials are weakly continuous, so that the corresponding
observable generators are well defined.
As it is typical in gauge field theories,3 the canonical Hamiltonian is gauge
P
invariant up to a time derivative, a t .H/ D H aE=e, and in fact the equations
of motion are gauge invariant.
The Gauss law constraint on the states, G D 0, is incompatible with canonical
commutations relations in Heisenberg form and, as before, one must use Weyl
quantization.4
Proposition 3.1 A gauge invariant state ! on FW defines a non-regular represen-
tation of FW , with the following properties:
i) the one-parameter group generated by the exponential of the electromagnetic
potential A is not weakly continuous and therefore A cannot be a well defined
operator in H!
ii) if ! is invariant under time translations and provides a physical representation
of FW , then ', ' cannot be applied to ! , and therefore ! cannot provide a
representation of the polynomial algebra generated by such fields.
Proof The first statement follows from the gauge invariance of ! and the canonical
commutation relations (in Weyl form) for the exponentials of A and G as in the proof
3
See, e.g., the canonical Hamiltonian for a complex scalar field interacting with the electromagnetic
field; G. Wentzel, Quantum Theory of Fields, Dover 2004, p. 68.
4
In the original form of the Christ-Lee model, without the term 12 AP 2 , the Lagrangian reduces to
L D 12 .Pr2 C r2 .P /2 / V.r/, where has the meaning of a gauge parameter. The role of
the Gauss law constraint is taken by the canonical momentum p q2 .P /, conjugated to
, and gauge invariance would require p D 0, incompatibly with the canonical commutation
relations in Heisenberg form. The strategy advocated above of adopting Weyl quantization provides
a mathematically acceptable treatment also in this case.
4 A QM Model of QCD Structures and of Josephson Effect 43
of Eq. (3.2.3) of Proposition 2.1. For the proof of the second statement we consider
the scalar product
W.t2 t1 / .'.0/ ! ; ei H .t2 t1 / '.0/ ! / D .'t1 ! ; 't2 ! /
and use the gauge invariance of !, i.e.
eib G.t/ ! D ei ! ; 2 R:
Then, by using the transformation properties of '.t/ under the time dependent gauge
transformations, Eq. (3.3.2), we get
W.t2 t1 / D eib .t2 t1 / W.t2 t1 /; 8b 2 R:
This implies W. / D 0, if 0; on the other hand, W.0/ cannot vanish, because

otherwise ', ' and the corresponding polynomial algebra would annihilate ! and
one would not get an acceptable representation of the gauge invariant algebra. Then,
one has W. / D 0, for 0, W.0/ 0, which imply that the unitary group of
time translations would not be weakly continuous, i.e. the Hamiltonian could not
be defined on H! . Hence, the existence of the Hamiltonian requires that ' and '
cannot be applied to ! and ! cannot provide a representation of the polynomial
algebra generated by '; ' .
4 A QM Model of QCD Structures and of Josephson Effect
We discuss a quantum mechanical gauge model exhibiting an interesting interplay

between gauge invariance and symmetry breaking, namely the mechanism by which
a symmetry generated by a gauge dependent operator gets broken in the irreducible
representations of the observable algebra compatibly with an energy gap.
The model is described by the canonical pairs '; p, with ' an angle, and A; E
'; p D i; A; E D i; (3.4.1)
with the following Hamiltonian
H D 12 .p eA/2 C 12 E2 M cos.' M /: (3.4.2)
The model can be considered as describing a quantum particle on a circle interacting

with an electromagnetic potential A, E D AP playing the role of the electric field.
The model also coincides with the model which describes the Josephson effect
taking into account inductance terms, ' being in this case the Josephson phase,
A the charge carried by the normal current (identified by AP D E).5
The model can also be interpreted as the bosonized version of the massive
Schwinger model in 0 C 1 dimensions, with M cos.' M / and M playing the role
of the fermion mass term and the mass angle, respectively. From this point of view
the model provides a non-perturbative realization and control of mechanisms which
have been argued to play a crucial role in Quantum Chromodynamics (QCD).6 We
briefly mention that the so called U.1/ and strong CP problem of QCD arise from
the fact that the QCD Lagrangian is invariant under an axial U.1/ symmetry which is
not realized in nature (absence of parity doublets) and for the spontaneous breaking
of which, without Goldstone bosons, one cannot invoke the Higgs mechanism.
The standard explanation is that the gauge invariant axial current associated to
the axial symmetry has an anomaly, but the non-trivial contribution of the anomaly
to the commutator with the order parameter has only been argued on the basis of
semiclassical approximations. On the other hand, as clarified by Bardeen on the
basis of perturbative renormalization (in local gauges),7 the time independent axial
U.1/ transformations define a time independent symmetry of the field algebra and of
its observable subalgebra and they are generated by the conserved gauge dependent
axial current. Thus, the Ward identities related to axial symmetry should be written
for such a current, and the absence of the corresponding Goldstone bosons requires
an explanation.
A very deep insight on the vacuum structure of QCD, with consequences on the
U.1/ problem, has been obtained by functional integral methods and semiclassical
approximations8: the topological structure of the smooth configurations of gauge
fields gives rise to the so called large gauge transformations, the spectrum of which
label the gauge invariant ground states or vacua. Since the set of such smooth
configurations are expected to have zero functional measure, a mathematical control
of this structure is at issue.
The above problems become even more relevant in the case of massive fermions
(M 0), since the invariance of QCD under CP requires the alignment of the
vacuum angle and the mass angle M , and the question arises of obtaining it
without fine tuning in a way stable under radiative corrections.
One of the interests of the model is that all the above problems are reproduced
and their solution is under mathematical control, shedding light on the expected
mechanisms of QCD.
5
S.M. Apenko, Phys. Lett. A 142, 277 (1989).
6
J. Lffelholz, G. Morchio and F. Strocchi, Ann. Phys. 250, 367 (1996).
7
W.A. Bardeen, Nucl. Phys. 75, 246 (1974).
8
G. t Hooft, Phys. Rev. Lett. 37, 8 (1976); R. Jackiw and C. Rebbi, Phys. Rev. Lett. 37, 172 (1976);
C.G. Callan, R.F. Dashen and D. Gross, Phys. Rev. Lett. B 63, 34 (1976). For an excellent review
see the quoted lectures by R. Jackiw in Current Algebra and Anomalies, 1985, pp. 211359. For a
mathematical control see Chapter 4, Sect. 4.
i) The observable algebra. Large gauge transformations

The meaning of ' as an angle requires to introduce only the periodic functions
of ', as for the particle on a circle; therefore, also for a better control of the
mathematical problem, we consider the C -algebra FW generated by ei' ; eip ,
ei A , eiE , ; ; 2 R, hereafter also called field algebra.
The gauge transformations
' ! '; p ! p C ; A ! A C =e; E ! E;
leave the canonical structure and the Hamiltonian invariant, but irreducible repre-
sentations of FW are obtained by fixing the gauge, namely by fixing the value
of the non-trivial center ZQ of FW , generated by ei2p , namely .ei2p / D ei2F 1,
F 2 0; 1/. In the following, for simplicity, we shall take F D 0. The following
group of gauge transformations (hereafter called large gauge transformations)
' ! '; p ! p C n; A ! A C n=e; E ! E; n 2 N; (3.4.3)
survive the gauge fixing and are implemented by the unitary operators
T n ein.'E=e/ ; H; T D 0: (3.4.4)
The exponential field algebra FW contains a gauge invariant subalgebra A, with

the physical interpretation of observable algebra, generated by the gauge invariant
variables ei' , P p eA, E. A has a non-trivial center Z, generated by the periodic
variable q ' E=e, equivalently by T; hence, the irreducible representations of
A are labeled by the points ei , 2 0; 2/, of the spectrum of T; the corresponding
Hilbert spaces H are called sectors.
The Hamiltonian takes a simple form in terms of the new canonical variables
Q E=e; P, eiq , and p:
H D 12 .P2 C m2 Q2 / M cos.Q C q M /; m2 e 2 : (3.4.5)
The quantum problem is to study the irreducible representations of A and the

corresponding spectrum of H there, i.e. with q D in Eq. (3.4.5). It is convenient
to discuss separately the massless and the massive case.
A. M = 0 (massless fermion)
ii) n-vacua, -vacua and symmetry breaking
In the discussion of the non-perturbative structures of QCD, the construction of
vacua is done in terms of the so called n-states n , which correspond to classical
field configurations of zero energy; however, their existence and status beyond the
semiclassical approximation is not completely under mathematical control and the
model provides a clear picture of the mechanism argued for QCD.
The lesson that the model teaches us is that there are two possible quantum
descriptions. One choice corresponds to the regular irreducible representation of
the field algebra FW : the Hilbert space is Hr D L2 .0; 2/ R; d' dA/, with states
described by wave functions .'; A/, on which the fields ', A act as multiplication
operators, and p; E as derivatives i@=@' (on periodic functions of '), and i@=@A,
respectively.
The spectrum of the large gauge transformation T D eiq is purely continuous in
Hr and therefore gauge invariant states (i.e. eigenstates of T) are not proper vectors
of Hr . The irreducible representations . ; H / of the gauge invariant (observable)
algebra can be obtained by decomposing Hr according to the spectrum of T
Z
Hr D d./ H ; 2 0; 2/; (3.4.6)
where H are improper subspaces of Hr .

This is the representation typically obtained by the perturbative expansion or by
the functional integral with functional measure D' DA. It is also the representation
which allows to work with the (gauge dependent) fields '; A and not only with their
exponentials (see below).
Another important feature of such a representation is its symmetry properties. In
the case M D 0, the Hamiltonian is invariant under the one-parameter group ,
2 R, of transformations
.'/ D ' C mod 2; (3.4.7)
all the other variables remaining unchanged, which will be called chiral transfor-
mations. The generator of such transformations is p, which satisfy a conservation
law dp=dt D 0, but it is not gauge invariant.
On the other hand the gauge invariant operator P D p eA generates the chiral
transformations on the fields '; A, at equal times, but not at any time, since
dP=dt D eE (3.4.8)
i.e. P is affected by an anomaly.

In the regular representation space Hr the Hamiltonian has the pure point
spectrum of the harmonic oscillator Ek D m.k C 12 /, k 2 N, with infinite
multiplicities, labeled by n 2 Z. Putting Hren H m=2 one has
Hren k; n D mkk; n ; eip k; n D ein k; n ; eiq k; n D k; nC1 : (3.4.9)
Thus, the ground states 0; n , also called n-vacua, are invariant under chiral
transformations and one has < F >0; n .0; n ; F 0; n / D 0, 8F 2 FW , since
< ei' >0; n D< eiq >0; n < eiQ >0; n D e1=4e < eiq >0; n D 0;
having used that the ground state of P2 C Q2 gives the expectation < ei. QCP/ >D
2 2
e. C /=4 and the last of Eqs. (3.4.9). This indicates that the n-vacua correlation
functions of the fields calculated either by the functional integral with the field
functional measure or by the perturbative expansion are chirally symmetric.
However, each n vacuum 0; n defines a reducible representation of the observ-
able subalgebra with an integral decomposition over the continuous spectrum of T,
Eq. (3.4.6).
Each sector has a lowest energy state 0 , called -vacuum, which by definition
is an eigenstate of T, T0 D ei 0 . Thus, an alternative to the regular (or
Heisenberg) quantization is to use from the start a gauge invariant state, typically
a vacuum ! . As we already learned from the quantum gauge models discussed
before, the representation of the exponential field algebra FW defined by a
vacuum is non-regular, namely the gauge dependent fields cannot be defined in
the corresponding representation space K. As repeatedly stressed before, this is the
mathematically consistent way of solving the conflict between gauge invariance and
canonical quantization.
A characteristic feature of the state space K is that the spectrum of T is a
purely point spectrum, so that eigenstates of T are proper vectors of K and in the
decomposition
X
KD H (3.4.10)
20;2/
H appear as proper subspaces. In K the standard formal equations get a rigorous

status:
X
0 D ein T n 0;0 ; eip 0 D 0C ; .0 ; 00 /K D ; 0 : (3.4.11)
n
Such equations display the mechanism of chiral symmetry breaking in the sectors
by the instability of the gauge invariant vacua.
The operator eip , which implements the time independent chiral transforma-
tions, is a well define operator with continuous spectrum in the regular represen-
tation space Hr of the exponential field algebra and it is weakly continuous in .
It has a pure point spectrum in K, but it does not define a one-parameter weakly
continuous group there, since
.0 ; eip 0 / D ; C D ;0 :
Therefore, the chiral symmetry breaking in each sector does not imply the
existence of a Goldstone state, because the analog of Eq. (2.4.7) cannot be written,
since the would be generator at all times p becomes a non-regular variable and does
not define an operator in H . As a matter of fact no Ward identity expressing the
breaking of the chiral symmetry can be written in H ; a Ward identity can be written
in Hr , where however the chiral symmetry is unbroken.
In each sector, the compatibility of chiral symmetry breaking and an energy gap
m > 0, in the subspace ei' 0 can be easily checked by computing (< > denotes the
expectation on 0 )
< P.t/; ei' > D ei < P.t/; eiQ > D ei cos.mt/ < eiQ > D
D ei cos.mt/e1=4e : (3.4.12)
It is worthwhile to note the essential singularity at e D 0, i.e. a severe lack of

analyticity.
Once it has been realized that the symmetry breaking Ward identity cannot be
written in the sector, and therefore one cannot get information on the energy
spectrum in this way, the following simple computation suggests a possible general
strategy for getting an estimate about the energy gap, in more general cases. Such
information can be obtained in terms of energy sum rules, by computing the
moments of the energy spectral measure in the symmetry breaking channel, here
Q . The calculation involves only equal time commutators:
< QH nC1 Q > D inC1 .dn =dtn /2 < PU.t/Q > jtD0 D
D inC1 .dn =dtn / < P.t/; Q.0/ > jtD0 :
In this simple model the computation of such energy moments yields

Z
.! 2 m2 / ! < Q0 ; dE.!/ Q0 >D 0;
i.e. the existence of a .! m/ in the energy spectral measure.

The overall picture emerging from the M D 0 case substantially agrees with
the conventional wisdom on QCD structures in the case of massless fermions. In
this case the lesson of the model is a mathematical control of delicate points and a
clarification of the mechanisms argued to hold in QCD.
A relevant message from the model is that, despite the anomaly of P (correspond-
ing to the gauge invariant axial current in QCD), chiral symmetry is a well defined
time independent symmetry of the field algebra and of the observable subalgebra.9
Its spontaneous breaking in each sector is due to the fact that the center of the
observable algebra is not pointwise invariant.10
9
For different statements in the literature, see e.g. S. Coleman, Aspects of Symmetry Cambridge
Univ. Press 1985, Chap. 7.
10
Such a mechanism of spontaneous symmetry breaking is compatible with an energy gap; relevant
examples are the breaking of Galilei symmetry in Coulomb systems, accompanied by the plasmon
energy gap, the breaking of the U.1/ symmetry in the BCS model of superconductivity. For a
general discussion of the mechanism see: G. Morchio and F. Strocchi, Removal of the infrared
cutoff, seizing of the vacuum and symmetry breaking in many body and in gauge theories, invited
talk at the IX Int. Conf. on Mathematical Physics, Swansea 1988, B. Simon et al eds., Adam Hilger
Publ. 1989, p. 490; F. Strocchi, Long range dynamics and spontaneous symmetry breaking in many
body systems, lectures at the Workshop on Fractals, Quasicrystals, Knots and Algebraic Quantum
mechanics, A. Amman et al. eds., Kluwer 1988.
Such a breaking is not accompanied by a Goldstone state, because first there

is no such a state in the large Hilbert space in which the fields are regularly
represented, since the symmetry is unbroken there. On the other hand, no genuine
symmetry breaking Ward identity can be written in each sector, because the time
independent generator p cannot be defined there. Even if i P.t/; ei' .ei' /,
one can nevertheless consider the expectation < P.t/; ei' > and show its time
dependence, as we did in Eq. (3.4.12), but such an expectation does not give a
genuine symmetry breaking Ward identity, since it is not directly related to the
spontaneous breaking of chiral symmetry.
With the insight so gained, we are in a better position for posing the much more
interesting and debated questions of the U.1/ and strong CP problems in the realistic
case of massive fermions.
B. M 0 (massive fermion)
The general question is what happens of the infinite ground state degeneracy when
the fermion mass term is switched on and in particular what happens of the
construction of the vacua in terms of the n-vacua.
A strictly related issue is the strong CP problem: the Hamiltonian H, Eq. (3.4.2),
is invariant under the following CP transformation
' ! ' C 2 M ; A ! A; (3.4.13)
crucially dependent on the mass angle M . As displayed by Eq. (3.4.5), in a sector

one has q D and the corresponding Hamiltonian is no longer invariant, except
in the special case D M . This equality requires a so called fine tuning, and in
QCD it has been argued that its validity at the tree level will not be stable under
radiative corrections. Thus, the experimental evidence that the strong interactions,
supposed to be described by QCD, are invariant under the CP symmetry defined by
the fermion mass term, requires an explanation.
iii) Hamiltonian spectrum and ground state
In the regular representation of the fields the Hamiltonian (3.4.2) has a purely
continuous spectrum, so that there is no proper ground state in Hr and the vacua
cannot be constructed as in Eq. (3.4.11).
Theorem 4.1 11 For any fixed F 2 0; 2/, let 0 denote the GNS representation
of FW defined by the following gauge invariant state
2 Ce2 2 /=4e
! .einq eip ei QCP / D ei e. ; if =2 2 Z;
D 0; otherwise: (3.4.14)
11
Then
i) the Hamiltonian H is well defined and gauge invariant states exists in the
corresponding Hilbert space K; moreover, for M 0 this is the only irreducible
representation of FW in which H has a ground state,
ii) the spectrum of T D eiq is discrete in K, which may therefore be decomposed as
in Eq. (3.4.10); in each proper subspace H the Hamiltonian has a pure point
spectrum, with no degeneracy and therefore a unique ground state ( vacuum),
iii) in K there is a lowest energy state corresponding to D M and therefore CP
invariant.
Proof The requirement that the Hamiltonian be well defined implies that so is also
P2 C e2 Q2 , since the mass term is a bounded perturbation. This condition selects
the Fock representation for the Weyl algebra A.Q; P/ generated by the canonical
variables P; Q.
Since T commutes with H, one can reduce the spectrum of H with respect to
the spectrum of T, i.e. discuss the spectrum E./ of H as a function of . Now,
one can prove (see below) that the function Inf E./ has only one minimum,
corresponding to D M . Therefore, the existence of a ground state requires that in
the decomposition of the representation space over the spectrum of T, of the form
of Eq. (3.4.6), the Hilbert space H DM should be a proper subspace, i.e. eiM should
be an eigenvalue of T.
Since eip acts transitively on the spectrum of q, the irreducibility of the repre-
sentation implies that the entire spectrum of q is a pure point spectrum, with no
multiplicity and K can be decomposed as in Eq. (3.4.10). Any vector of the proper
subspace H is an eigenvector of q and since 0 is a Fock representation of A.Q; P/,
one can find a vector 0 2 H , which gives the expectation (3.4.14) for n D 0 D .
Finally, since qeip 0 D . / 0 , for =2 Z, eip 0 is orthogonal to 0 and
Eq. (3.4.14) follows.
The uniqueness of the lowest energy state in each H follows from the strict
positivity of the kernel of eH in the variable Q and by a generalized Perron-
Frobenius argument.12 The corresponding energy eigenvalue E0 can be computed,
for fixed e > 0 and M small, to first order in M:
E0 ./ D 12 e Me1=4e cos. M /: (3.4.15)
The absolute minimum is attained for D M . It is worthwhile to note the essential

singularity at e D 0, signaling a non-perturbative effect.
Thus, the model suggests a possible dynamical solution of the strong CP
problem, the absolute ground state being 0DM . It is worthwhile to remark, that
an approach to the massive case which first chooses a sector and then adds the
12
J. Glimm and A. Jaffe, Quantum Physics. A Functional Integral Point of View, Springer 1987,
Sect. 3.3; for a simple account F. Strocchi, An Introduction to the Mathematical Structure of
Quantum Mechanics, World Scientific 2005, Sect. 6.4.
fermion mass perturbation would yield a CP non symmetric vacuum, if M .

In the analogy with the Bloch electron, this would correspond to choose a band
higher than the ground state band. Furthermore, the energy spectrum has an essential
singularity for vanishing gauge coupling constant.
Since no (local) observable can make transition between different sectors, it is
of some interest to investigate a mechanism for reaching the CP symmetric vacuum
if one starts from a metastable CP violating vacuum.
A possible selection of the absolute ground state emerges in a careful discussion
of the thermodynamical limit performed on the functional integral.13 In particular
the limit M ! 0 and the thermodynamical limit do not commute. The alignment of
to M in the thermodynamical limit occurs by the same mechanism by which in the
Heisenberg-Weiss model the mean field spin gets aligned to an external magnetic
field.14
13
14
G. Morchio and F. Strocchi, Boundary Terms, Long Range Effects, and Chiral Symmetry
Breaking, Lectures at the Schladming School 1990, Fields and Particles, H. Mitter and W.
Schweigen eds., Springer 1990, p. 171.
Chapter 4
Non-regular Representations in Quantum Field
Theory
1 Quantum Field Algebras and Quantizations
The traditional (historically the first) approach to field quantization1 mimics very
closely the standard quantum mechanical case, by realizing a field as a mechanical
system of infinite degrees of freedom and by adopting the canonical formalism and
quantization.
For example, for a real scalar field '.x; t/, by choosing a complete set ffn .x/g, n 2
N, of real orthonormal smooth functions, a real scalar field is completely identified
by a denumerable set of Lagrangian variables
Z
qn .t/ '. fn ; t/ d3 x fn .x/'.x; t/; qP n .t/ '.
P fn ; t/; (4.1.1)
since
X X
'.x; t/ D qn .t/ fn .x/; '.x;
P t/ D qP n .t/ fn .x/:
n n
A canonical formalism may be introduced by defining the canonical momenta

pi .t/ by
Z
pn .t/ D @L=@Pqn .t/ D d 3 x .x; t/ fn .x/; .x; t/ @L=@'.x;
P t/;
1
See G. Wentzel, Quantum Theory of Fields, Dover 1949; S. Schweber, An Introduction to
Relativistic Quantum Field Theory, Harper and Row 1961, Part Two.

54 4 Non-regular Representations in Quantum Field Theory
where
Z
L d 3 x L.'.x; t/; @i '.x; t/; '.x;
P t//
is the Lagrangian function written as an integral over the Lagrangian density L.

Then the Hamiltonian is defined by
X Z
HD pn .t/ qP n .t/ L D d3 x [email protected]/=@'.x/
P '.x/
P L.x//:
n
The canonical (classical) Poisson brackets read
fqi .t/; pj .t/g D i j ; (4.1.2)
all other Poisson brackets vanishing. They are equivalent to the following Poisson
brackets for the fields
f'.x; t/; .y; t/g D 3 .x y/; (4.1.3)
all other Poisson brackets vanishing.

Canonical quantization is obtained by Dirac prescription of replacing the Poisson
brackets by i times commutators. Thus, the Heisenberg Lie algebra and the
Heisenberg group become infinitely dimensional and the corresponding Weyl
algebra infinitely generated.
It should be mentioned that canonical quantization may be in conflict with a non-
trivial interaction. In fact, it requires that the fields allow sharp time restrictions, at
least as (tempered) distributions in the space variables and perturbative and non-
perturbative results indicate that generically this is not the case in 3+1 space-time
dimensions, except in the free case.2
This conflict arises whenever the so-called wave function renormalization con-
stant Z 1 is divergent as a consequence of short distance or ultraviolet (UV)
singularities. In fact, in this case canonical commutation relations may be required
to hold for the ill defined unrenormalized fields, whereas the canonical commutation
relations of the renormalized fields are proportional Z 1 .3 Thus, in order to safely
use canonical quantization one must introduce an UV cutoff.
2
For (irreducible) fermion fields, R.T. Powers (Comm. Math. Phys. 4, 145 (1967)) has shown that
canonical anticommutation relation are compatible only with a free theory in s C 1 dimensions
with s > 1, under very general conditions. No interaction theorems for the bosonic case have been
proved by K. Baumann (Jour. Math. Phys. 28, 697 (1987); Lecture at the Schladming School 1987,
in Recent Developments in Mathematical Physics, H. Mitter and L. Pittner eds., Springer 1987).
3
A.S. Wightman, Phys. Rev. 101, 105 (1956); S. Schweber, An Introduction to Relativistic
Quantum Field Theory, Harper and Row 1961, Sect. 7b; for a brief discussion see e.g. F. Strocchi,
An introduction to non-perturbative foundations of quantum field theory, Oxford Univ. Press 2013.
1 Quantum Field Algebras and Quantizations 55
For these reasons, following Wightman,4 a quantum field theory (QFT) is defined
in terms of a field *-algebra F , briefly called field algebra, generated by the
polynomials of the smeared fields
Z
'. f / D dx '.x/ f .x/; f 2 S.Rd /;
d denoting the space-time dimension and S.Rd / the C1 functions of fast decrease.
The field algebra is called local if it satisfies microcausality or locality; e.g., for a
hermitian scalar field '.x/, this means
'.x/; '.y/ D 0; if .x y/2 < 0: (4.1.4)
It will be understood that F will also include the sharp time restrictions of the
fields, whenever they exist as distributions in the space variables. The space-time
translations are assumed to define automorphisms a , a D .t; a/ 2 Rd , of F .
In the following, FW denotes the exponential field algebra formally generated
by the exponentials of the (smeared) fields '. f /; A and AW will denote the observ-
able (field) subalgebras of F and of FW , respectively. It should be mentioned that
the definition of the exponential of a smeared field requires its self-adjointness, a
property which is not obvious in QFT, in contrast with the quantum mechanical
case; for this purpose general conditions have been discussed in constructive QFT,
which guarantee the exponentiability of fields.5
Definition 1.1 A representation of a field algebra F is a homomorphism into a
*-algebra of operators acting on a common dense domain of a Hilbert space H.
By an easy adaptation of the GNS theorem, a positive linear functional ! on F
defines a representation of F in a Hilbert space H, with a cyclic vector ! such that
!.F/ D .! ; ! .F/ ! /; 8F 2 F :
Such a result in QFT is called the Wightman reconstruction theorem (see Streater
and Wightman book).
Definition 1.2 A quantum field theory is a representation of a field algebra
F defined by a positive linear functional W on F , invariant under space-time
translations (d the spacetime dimensions)
W.a .F// D W.F/; 8a 2 Rd ; 8F 2 F ;
4
R.F. Streater and A.S. Wightman, P C T, Spin and Statistics and All That, Benjamin 1980; R. Jost,
The General Theory of Quantized Fields, Am. Math. Soc. 1965.
5
J. Frhlich, Comm. Math. Phys.54, 135 (1977).
and satisfying the positive energy spectral condition

Z
dt eitE W.Ft .G// D 0; if E < 0; (4.1.5)
where t denotes the time translations.

A relativistic quantum field theory is defined by a local field algebra F ,
transforming covariantly under the Poincar transformations a; , and by a state
invariant under the Poincar group.
It should be remarked that the field algebra F involved for a formulation of a
quantum field theory does not need to consist of observable fields only. Therefore,
the locality of F and its covariance under the Lorentz transformations are not a must,
for a relativistic quantum mechanical interpretation of the theory; clearly, all what
is needed for an acceptable relativistic interpretation is that locality and covariance
are fulfilled by the observable subalgebra A. Such a structure characterizes gauge
quantum field theories and closely resembles the quantum mechanical structures
discussed in Chap. 2, Sect. 4.
Since F is not a C -algebra, just as it happens for the polynomial algebra
generated by the Heisenberg canonical variables, a positive linear functional W
provides a (natural) generalization of the concept of state, and, with a slight abuse
of language, shall be called a vacuum state on F , if it is invariant under space-time
translations and satisfies the positive energy spectral condition.
Definition 1.3 A vacuum state ! defines a regular representation of the expo-
nential field algebra FW if ! has a well defined (positive) extension to the field
algebra F ; otherwise, the representation defined by it is said to be a non-regular
representation.
Under general technical conditions the correlation function of a (smeared) field
'. f / can be obtained from a representation of FW if the one-parameter group
defined by the exponentials of '. f / is weakly continuous. A representation of FW is
physical if the representation of the observable subalgebra AW is regular, in order to
allow for the existence of the correlation functions of observable fields, as required
by a reasonable physical interpretation. However, non-regular representations of
the non-observable (exponential) field algebras are allowed, as in the quantum
mechanical examples discussed before.
As discussed in Chap. 2, Sect. 4, a general mechanism for the need of non-
regular representations of the field algebra FW is when i) the observable algebra A
is a proper subalgebra of the field algebra characterized by its pointwise invariance
under a group G of automorphisms of FW defined by elements of FW , and ii) one
considers representations defined by gauge invariant vacuum states.
2 Massless Scalar Field in 1+1 Dimensions 57
As we shall discuss below, the characteristic property of gauge quantum field

theories is the Gauss law6 and the Gauss law operator generates gauge transfor-
mations. Hence, the validity of the Gauss law constraint, as an operator equation
satisfied by the physical states, requires a non-regular representation of the field
algebra F , unless positivity of the vacuum functional is abandoned.
An explicit analysis shall be presented in Sect. 3 below for the QED case and
in Sect. 4 for the QCD case. As we shall see, the use of non-regular representations
will not only provide a mathematically acceptable formulation, but it will also prove
useful for the discussion and solution of structural problems.
2 Massless Scalar Field in 1+1 Dimensions
The quantum theory of a relativistic massless hermitian scalar field in two space-
time dimensions has attracted much attention in the last years, not only as a gauge
model, but also because it is one of the building blocks of very instructive two-
dimensional models, like the Schwinger model, which mimics very closely some
basic structure expected to characterize QCD, like quark confinement and chiral
symmetry breaking.7
The massless scalar field is also at the basis of the construction of two-
dimensional conformal models, which have received much attention in connection
with important developments of QFT8 as well as for string theory.9
At the classical level, the relativistic massless real scalar field '.x; t/ in 1+1
dimensions satisfies the free wave equation and the classical canonical Poisson
brackets
' D 0; f'.x; t/; @0 '.y; t/ g D .x y/; (4.2.1)
all the other Poisson brackets vanishing.

The quantization of such a system meets non-trivial problems. If one tries to
adopt canonical quantization, the so obtained field algebra does not admit a positive
linear functional !0 invariant under the Poincar group P (space-time translations
6
For a focusing of this property see e.g. F. Strocchi and A.S. Wightman, Jour. Math. Phys. 15,
2198 (1974); F. Strocchi, Gauss law in local quantum field theory, in Field Theory, Quantization
and Statistical Physics D. Reidel 1981, p. 227236; F. Strocchi, Elements of Quantum Mechanics
of Infinite Systems, World Scientific 1985, Part C, Chap. II; F. Strocchi, An Introduction to Non-
perturbative Foundations of Quantum Field Theory, Oxford Univ. Press 2013.
7
For an analysis of the Schwinger model which takes into account the delicate issue of the
quantization of the massless scalar field in two-dimensions, see G. Morchio, D. Pierotti and F.
Strocchi, Ann. Phys. 188, 317 (1988); F. Strocchi, Selected Topics on the General Properties of
Quantum Field Theory, World Scientific 1993, Chapter VII, Section 7.4.
8
See e.g. P. Furlan, G.M. Sotkov and I.T. Todorov, two-dimensional Quantum Field Theory, Riv.
Nuovo Cim. 12, 1 (1989).
9
B.F. Hatfield, Quantum Field Theory of Point Particles and Strings, Addison-Wesley 1992.
and Lorentz transformations). Actually, quite generally, the two point function of '
defined by the expectation of a Poincar invariant vacuum !0 violates positivity.10
Such a result has led to the statement that there is no quantum field theory for
the massless scalar field. Actually, as we shall see, the result means that there is no
regular representation of the exponential field algebra FW invariant under P, but a
Poincar invariant non-regular representation of FW exists.
For this purpose, we first note that, since the field is a free field, its sharp
time restrictions exist and FW may be assumed to contain the Weyl field algebra
generated by the exponentials U.h1 /, V.h2 / of '.h1 ; t/, .h2 ; t/ '.h
P 2 ; t/, hi 2
S.R/, respectively, satisfying the Weyl commutation relations
U.h1 / V.h2 / D V.h2 / U.h1 / ei.h1 ;h2 / ; (4.2.2)
where .h1 ; h2 / denotes the L2 .R; dx/ scalar product.

A (unique) Poincar invariant vacuum state exists for the subalgebra AW
generated by the exponentials of '. f /, with fQ.0/ D 0, briefly f 2 S0 .R2 /, defined
by the two point function
.@ '.x/ @
'.y// D @ @
.4/1 log.x y/2 C i".x0 y0 / ; (4.2.3)
(we recall that the elements of S0 .R/ may be written as the derivatives of elements
of S.R/). Such a state also satisfies the relativistic spectral condition, i.e. one has
8A; B 2 AW
Z
da eipa .A U.a/ B/ D 0; if p V C fpI p2 0; p0 0 g: (4.2.4)
Thus, AW qualifies as the observable subalgebra of FW .

AW can be characterized as the subalgebra of FW which is pointwise invariant
under the gauge transformations
W ' ! ' C ; 2 R:
Thus, the model shares the general structure discussed in Chap. 2, Sect. 4.
Proposition 2.1 Let be a positive (pure) vacuum state on AW , then
i) does not have a positive extension to the field algebra F , generated by '.g/,
g 2 S.R2 /, invariant under space-time translations and satisfying the positive
energy spectral condition,
10
A.S. Wightman, Introduction to some aspects of the relativistic dynamics of quantized fields,
in High Energy Electromagnetic Interactions and Field Theory, Cargse lectures in Theoretical
Physics (1964), M. Lvy ed., Gordon and Breach 1967, esp. Sect. I.4; for a general analysis of the
quantization in terms of a non-positive vacuum state, see G. Morchio, D. Pierotti and F. Strocchi,
Jour. Math. Phys. 31, 147 (1990).
2 Massless Scalar Field in 1+1 Dimensions 59
ii) has a unique positive extension ! to FW , invariant under gauge transforma-

tions and therefore given by
!.U.h/ V.g// D .U.h/ V.g//; if Q

h.0/ D 0;
D 0; otherwise: (4.2.5)
The representation of FW defined by the gauge invariant vacuum state ! has

the property that its representation space K is non-separable and has the following
orthogonal decomposition into disjoint irreducible representations of AW
X
KD H ; H D .AW / U.h/ 0; Q
h.0/ D ; (4.2.6)
2R
where 0 is the representative vector of !.

Proof The free wave equation in two space-time dimensions implies that the
Fourier transform of the two point function !.'.x/ '.y// defined by a translationally
invariant state ! is of the form
A.pC / .p / C B.p / .pC/; p p0 p1 ; A; B 2 S 0 .R/:
The energy spectral condition requires that A and B be of the form A.pC / D
a.pC /.pC /, B.p / D b.p /.p /, and the condition of providing an extension
of the unique two point function (4.2.3) implies that the distributions a.pC /, b.p /
satisfy
pC a.pC /.pC/ D 1; p b.p / .p / D 1
There are no positive distributions or measures which satisfy such equations except
.p /, which lead to a trivial two point function, (incompatibly with the non-
triviality of on AW ). Thus, does not have a positive extension to F .
A positive extension to FW is provided by the gauge invariant state defined by
Eq. (4.2.5). It is easy to see that gauge invariance uniquely selects such an extension.
In fact, a gauge invariant extension ! of to FW satisfies
Q
!.U.h/V.g// D !. .U.h/ V.g/// D ei h.0/ !.U.h/V.g///;
Q
i.e. Eq. (4.2.5) holds, since for h.0/ D 0, U.h/ V.g/ 2 AW and ! coincides with
there. One can actually show that the state ! defined by Eq. (4.2.5) is the only
positive extension of .11
11
See F. Acerbi, G. Morchio and F. Strocchi, Jour. Math. Phys. 34, 899 (1993); F. Acerbi, Master
thesis, SISSA Trieste, 1990.
The
R fields U.h/ have the interpretation of charged fields with charge
Q
dx h.x/ D h.0/:
.U.h// D ei U.h/:
The gauge invariance of ! implies that the gauge group is unbroken in K and the
decomposition of Eq. (4.2.6) follows.
In conclusion, the model reproduces most of the structural features discussed in
Chap. 2, for a general quantum mechanical model described by a canonical algebra.
Thus, one has the analog of the non-regular representations of the Heisenberg group,
which has now become infinite dimensional.
An important structure associated to the quantization of the massless scalar field
is obtained by considering the extension AF of the observable algebra AW generated
by the (unitary) operators
U.h/; V.g/; h 2 S.R/; g 2 @1 S.R/ fg 2 C1 .R/; @g 2 S.R/g:
The algebra AF contains anticommuting fields, which can be interpreted as

fermionic fields, and its construction in terms of a scalar (bosonic) field is called
fermion bosonization. The (positive) extension of to AF defines a non-regular
representation with an unbroken U.1/ group generated by the fermionic charges.12
It may be worthwhile to mention that the non-regular representation of AF
provides a simple mechanism for the proof of confinement in the bosonized
Schwinger model, since in this case the representation of the time translations is
not weakly continuous; this implies that in the charged sectors one cannot have a
finite energy and therefore the corresponding states are not physically realizable.13
3 Temporal Gauge in QED
For the analysis of non-perturbative aspects of QED, typically by the functional

integral approach, and in particular for the discussion of the existence of a symmetry
breaking order parameter in the Higgs phenomenon, the temporal gauge has been
widely used, but the textbook presentations of such a gauge have somewhat
neglected its peculiar mathematical features and in particular the mathematical
consistency of the proposed realizations.14
12
For an analysis of this structure, see F. Acerbi, G. Morchio and F. Strocchi, Lett. Math. Phys. 26,
13 (1992).
13
F. Acerbi, G. Morchio and F. Strocchi, Lett. Math. Phys. 27, 1 (1993).
14
See e.g. the comprehensive book by A. Bassetto, G. Nardelli and R. Soldati, Yang-Mills Theories
in Algebraic Non-covariant Gauges, World Scientific 1991.
3 Temporal Gauge in QED 61
The temporal gauge is defined by the gauge condition A0 D 0 (both in the free
as well as in the interacting case), without requiring the transversality of Ai . Thus,
in the free case it is similar to the Coulomb gauge, but longitudinal photons are
allowed. Manifest covariance is obviously lost, but, in contrast with the Coulomb
gauge, locality holds in the free case and has been argued to persist in the interacting
case.
The variation with respect to Ai of the gauge invariant Lagrangian
L D 14 F
F
C Lmatter ;
where Lmatter is the gauge invariant Lagrangian for the matter fields (for simplicity
taken to be Dirac fermions ; N ), typically obtained from the free Lagrangian by
the minimal coupling prescription, yields the following equations of motion
@20 Ai Ai C @i divA D ji ; (4.3.1)
where j is the conserved gauge invariant electromagnetic current constructed in

terms of the charged fields.
The canonical commutation relations for the vector potential
Ai .x; 0/; @0 Aj .y; 0/ D i i j .x y/; Ai .x; 0/; Aj .y; 0/ D 0; (4.3.2)
are clearly incompatible with the Gauss law, which in the free case reads
divE D 0; Ei D @0 Ai and such an incompatibility persists in the interacting case.
In fact, the conflict between quantization and the validity of the Gauss law
divE.x/ j0 .x/ D 0 can be argued quite generally, independently of the existence
of equal time restrictions of the fields and of a full quantum canonical structure.
Eqs. (4.3.1) and the current continuity equation imply that G.x; x0 / divE.x/j0 .x/
is time independent, so that
Z
G. f ; x0 / G. f ; h/; f 2 S.R3 /; h 2 S.R/; dt h.t/ D 1;
is a well (densely) defined operator and its equal time commutators with the fields
are well defined operator valued distributions, fixed by the condition, which can
be taken as part of the definition of the temporal gauge, that G generates the time
independent gauge transformations. Thus, quite generally, one has
Ai .x; t/; G.y; t/ D i @i .x y/ (4.3.3)
and, by the time independence of G. f ; t/
Q
Ai .g; h/; G. f ; t/ D i.@i g; f / h.0/;
where . ; / denotes the L2 .R3 ; dx/ scalar product.

By taking the vacuum expectation of the above equation, one gets a conflict with
the validity of the Gauss law on the vacuum:
G. f /0 D 0:
There are only two (mathematically consistent) alternative solutions15 of such

a conflict between Eq. (4.3.3) and the Gauss law, with peculiar (widely unnoticed)
mathematical properties.
One alternative, characterized by the requirement of existence of the two point
function (or of the propagator) of the vector potential Ai , requires to abandon
positivity and to weaken the Gauss law. In the second alternative, the vector potential
cannot be defined as a field operator, only its exponentials being definable.16
For the second (positive) alternative, it is convenient to introduce the following
field algebras (for simplicity, we consider the case in which the charged fields are
the Dirac fermion fields , ):
i) the field algebra F , generated by the smeared fields A. f / Ai . fi /, and
.g/; .g/, fi ; g 2 S.R4 /;
ii) the observable field subalgebra A characterized by its pointwise invariance
under time independent gauge transformations:
G. f /; A D 0; 8A 2 AI
iii) the exponential field algebra FW generated by exp iA. f /, and .g/, .g/,
f ; g 2 S.R4 /;
iv) the exponential longitudinal algebra FL generated by exp iG. f /, exp iA.@h/,
A.@h/ Ai .@i h/, h 2 S.R4 /.
The inevitable violation of positivity for the regular quantizations and the general
properties of the positive quantizations are stated by the following
Proposition 3.1 Let be a Poincar invariant positive vacuum state on the
observable subalgebra A, then one has
i) any extension of to the field algebra F violates positivity
ii) all the positive extensions ! to the exponential field algebra FW are invariant
under time independent gauge transformations, in particular, if ! denotes the
GNS cyclic vector which represents the state !, one has
G. f /! D 0: (4.3.4)
15
J. Lffelholz, G. Morchio and F. Strocchi, Jour. Math. Phys. 44, 5095 (2003), hereafter referred
to as LMS.
16
The non-regularity of the positive quantizations of the temporal gauge in the free case has been
discussed by D. Buchholz and K. Fredenhagen, Comm. Math. Phys. 84, 1 (1982); H. Grundling
and C.A. Hurst, Lett. Math. Phys. 15, 205 (1988); F. Acerbi, G. Morchio and F. Strocchi, J. Math.
Phys. 34, 899 (1993). The unique selection of the non-regular representation by the condition of
positivity of the energy spectrum has been proved in LMS.
3 Temporal Gauge in QED 63
iii) any positive extension ! of to the exponential longitudinal algebra FL is

non-regular, since it must satisfy
!.eiA.@h/ / D 0; if 0I
D 1; if D 0 (4.3.5)
iv) in the GNS representation defined by the a positive extension ! to FW , the

unitary operators U.a/ which implement the space translations are not strongly
continuous, and therefore their generator, the momentum, cannot be defined in
the whole representation space H! .
Proof
i) Since G. f / commutes with any element of A, by Theorem 4.5 of Streater and
Wightman book (referred to in footnote 4), if denotes the cyclic vector
which represents the state on A,
G. f / D c. f / ; c. f / 2 C;
and by the Lorentz invariance of one has c. f / D 0. Hence, by Schwarz

inequality, a positive extension ! to F should satisfy
!.G. f /F/ D 0; 8F 2 F ; (4.3.6)
which is incompatible with the !-expectation of Eq. (4.3.3). Hence, there is no

positive extension to F .
ii) Any positive extension to FW satisfies Eq. (4.3.6), with F replaced by any B 2
FW , so that Eq. (4.3.4) follows by the cyclicity of ! .
iii) Putting V. f / exp iG. f /, one has
R
d 4 xhf
!.eiA.@h/ / D !.V.f / eiA.@h/ V. f // D ei !.eiA.@h/ /

iv) Putting ha .x/ h.x C a/ one has
!.eiA.@h/ U.a/eiA.@h/ / D !.eiA.@h@ha / / D 0; if a 0I

D 1; if a D 0:
The invoked escape of a non-normalizable vacuum vector is mathematically

unacceptable because, if the vacuum is not normalizable, the divergent expectations
are not confined to those of the gauge dependent fields, but affect also the observable
fields.
As far as we know, the only mathematically acceptable solution is to keep the
normalizability of the vacuum and give up the well definiteness of the gauge depen-
dent fields A. f /, only their exponentials having well defined vacuum correlation
functions. This strategy has a well founded mathematical status and it is under
complete control for the quantization of canonical systems as discussed before.
In the free field case the positive non-regular representation of the exponential
field algebra FW is uniquely determined solely by the condition of positivity of the
energy spectrum.17
4 Temporal Gauge in QCD: Chiral Symmetry Breaking
For the discussion of the non-perturbative aspects of QCD, in particular the

vacuum structure, its relation with the topology of the gauge field configurations
and its role in chiral symmetry breaking, the temporal gauge has proved to be
particularly convenient,18 because the residual gauge group G consists of the time
independent gauge transformations and its non-trivial topology emerges clearly.
However, as noted before for the abelian (QED) case, the conflict between the
Gauss law constraint and canonical quantization raises problems of mathematical
consistency, which also have a non-trivial impact on the derivation of the general
structures leading to significant physical implications.
4.1 Gauss Law and Gauge Transformations
For simplicity, we start by considering the case with only vector fields (no fermion or
scalar field being present). Then, at the classical level the QCD Lagrangian density
reduces to the Yang-Mills form
X X
L D 14 Fa
F
a D 1
2 .E2a B2a /; (4.4.1)
a a
where in the temporal gauge, defined by A0a D 0,
P a;
Ea D A Ba D r a Aa 12 gfabc Ab Ac ; (4.4.2)
(a is a color index and fabc are the structure constants of the Lie algebra of the color
gauge group). The corresponding equations of motion, obtained by variations with
17
LMS, Proposition 3.3.
18
See the excellent lectures by R. Jackiw, Topological investigations of quantized gauge theories,
in S.B. Treiman, R. Jackiw, B. Zumino and E. Witten, Current Algebra and Anomalies, World
Scientific 1985, pp. 211359. This section relies on G. Morchio and F. Strocchi, Ann. Phys. 324,
2236 (2009), hereafter referred to as Ref. MS; for an expository account, see F. Strocchi, An
introduction to non-perturbative foundations of quantum field theory, Oxford Univ. Press 2013.
4 Temporal Gauge in QCD: Chiral Symmetry Breaking 65
respect to Aa ,
@t Ea D r Ba C gfabc Ab Bc .D B/a ; (4.4.3)
imply
@t Ga D 0; Ga r Ea C gfabc Ab Ec .D E/a : (4.4.4)
The operators Ga are called the Gauss law operators.

In the standard quantum version of the temporal gauge it is assumed that the fields
Aa and their powers can be defined and the quantization is given by the canonical
commutation relations. In particular one has the following commutation relations
i D Ea .x; t/; Ab .y; t/ D ab r .x y/ C gfabc Ac .x; t/ .x y/: (4.4.5)

i D Ea .x; t/; Eb .y; t/ D gfabc Ec .x; t/ .x y/:
They state that the Gauss operators generate the infinitesimal time independent
gauge transformations, , with the c-number gauge function a .x/ 2 S.R3 /
Ab .x/ D a b r a .x/ C gfabc Ab .x/ c .x//: (4.4.6)
Since the variables A0a are missing in the Lagrangian, one cannot exploit the
stationarity of the action with respect to them and therefore one does not get the
Gauss law Ga D 0. Actually, the Gauss law is incompatible with Eq. (4.4.5) and
therefore with canonical quantization and more crucially with the Gauss operator
being the generator of the time independent gauge transformations, Eq. (4.4.6).
A proposed solution of this conflict, widely adopted in the literature and in
textbook discussions of the temporal gauge, is to require the Gauss law constraint
as an operator equation on the (subspace of) physical states and, in particular, on the
vacuum state. However, such a solution is not free of paradoxes and mathematical
inconsistency. In fact, the vacuum expectation of Eq. (4.4.5) gives zero on the left
hand side and non-zero on the right hand side.
It has been proposed19 to cope with this paradox by admitting that the vacuum
vector is not normalizable. In our opinion, such a solution is not acceptable, because
it does not yield a representation of a field algebra containing both gauge dependent
and gauge independent fields, since the non-normalizability of the vacuum vector
would give divergent expectations also of the observable algebra (which in particular
contains the identity).
A mathematically acceptable solution for the Gauss law constraint is to adopt a
Weyl quantization and admit non-regular representations.
19
R. Jackiw, loc cit.; A. Bassetto, G. Nardelli and R. Soldati, Yang-Mills Theories in Algebraic
Non-covariant Gauges, World Scientific 1991.
To this purpose, we consider the local field algebra F generated by the

polynomials of the smeared fields
Z
Aia . f / D d 4 x Aia .x/ f .x/; f 2 S.R4 /;
transforming covariantly under the space time translations y , y 2 R4 , i.e.

y .Aia . f // D Aia . fy /; fy .x/ D f .x y/. We shall assume that by a suitable point
splitting procedure one can define as local space-time covariant fields belonging to
F the powers of Aa .x/ and its derivatives, like e.g. the Gauss operator Ga .g/, the
magnetic field Bia .g/, g 2 S.R4 /, etc.
For a simpler bookkeeping of the indices, it is convenient to introduce the
following notations:
i) T a denote the hermitian representation matrices of the Lie algebra of the gauge
group, normalized soP that Tr( T a T b / D a b ;
ii) the fields A .x/ D
i
a Aa .x/T are Lie
i a
P algebra valued distributions on Lie
algebra valued test functions f .x/ D a fai .x/ T a , fai 2 S.R4 /, (the sum over
i
repeated indices is understood) and

Z Z X
A.f/ d4 xTr A.x/ f.x/ D d4 x Aia .x/ fai .x/I
i; a
iii) G denotes the group of time independent gauge transformations U labeled

by space localized gauge functions U.x/ which are gauge group valued C1
functions differing from the identity only on a (space) compact set KU ; the
space-time support of U 1 is given by the cylinder CU KU R and
U .A.f// D A.UfU 1 / C U@U 1 .f/; (4.4.7)

Z X
U@U .f/ d4 x Tr
1
U.x/@i U 1 .x/ f i .x/ DU .f/
i
iv) U , 2 R denote the gauge functions corresponding to one-parameterP

subgroups of G; they are of the form U .x/ D ei g.x/ , with g.x/ D a ga .x/ T a
a Lie algebra valued function, infinitely differentiable and of compact support
(ga 2 D.R3 /). All gauge transformations of compact support in a neighborhood
of the identity, in the C1 topology, are of this form and generate the Gauss
subgroup G0 .
For the Weyl quantization one has to consider the exponential
P field algebra
FW generated by the unitary operators W.f/, with f i .x/ D a fai .x/ T a , fai 2 S.R4 /,
formally the exponentials eiA.f/ , and by the unitary operators V.U /, which represent
G0 , briefly called Gauss implementers, formally the exponentials of the Gauss
operators,
X
V.U / D ei G.g/ ; G.g/ D Ga .ga /; ga 2 D.R3 /: (4.4.8)
a
The operators V.U / transform covariantly under space translations. As it is

standard we assume that the dynamics t may be obtained as a limit of localized
dynamics tR defined by local gauge invariant Hamiltonians HR (e.g. by Hamiltonian
densities integrated over spheres of radius R), i.e. t .F/ D limR!1 tR .F/.
Then, the gauge invariance of the local Hamiltonians implies the time indepen-
dence of the V.U /
.d=dt/t .V.U // D i lim HR ; V.U / D 0:

R!1
A representation of FW also defines a representation of F , and in this case it is

called regular, if (the representatives of) the field exponentials W.f/, 2 R, define
weakly continuous one-parameter groups.
A vacuum state is invariant under spacetime translations and therefore in
the corresponding representation the space-time translations a , a 2 R4 , are
implemented by unitary operators U.a/.
A state ! on the exponential field algebra FW , in particular a vacuum state,
satisfies the Gauss law in exponential form if !.V.U // D 1, equivalently if its
representative vector ! , in the (GNS) Hilbert space H! defined by the expectations
!.FW /, satisfies
V.U / ! D ! ; 8 U :
Briefly, a vector state 2 H! satisfying V.U / D ; 8 U , is said to be

Gauss invariant and H0 H! denotes the subspace of Gauss invariant vectors.
An operator in H! is Gauss invariant if it commutes with all the V.U /.
Proposition 4.1 A vacuum state ! on the exponential field algebra FW , satisfying
the Gauss law, defines a non-regular representation of FW , since :
!.W.f// D 0; if f.x/ 0: (4.4.9)
The fields A, formally the generators of the W.f/, cannot be defined in the GNS
Hilbert space defined by the vacuum expectations and in particular the two point
function of the gauge potential does not exist, only (the vacuum expectations of)
the exponential functions (and of course the gauge invariant functions) of A can be
defined.
In the free case, the exponential field algebra becomes a Weyl field algebra
and Eqs. (4.4.9) uniquely determine its representation as a non-regular Weyl
quantization.
Proof For each f there is a one-parameter subgroup U such that U f U D f, and

exp iDU .f/ D exp .i U @ U .f// 1; therefore, by the Gauss invariance of !,
one has
!.W.f/ D !.V.U / W.f/ V.U / / D eiDU .f/ !.W.f//;
and Eqs. (4.4.9) follow.

Clearly, the one-parameter groups defined by W.f/ cannot be weakly continuous and
therefore the corresponding generators, i.e. the fields Aia . f / do not exist as operators
in the GNS Hilbert space H! defined by the expectations of FW on !. In particular
the gluon propagator does not exists. The free case can be worked out as for the
abelian case.20
The one-parameter groups V.U / are not assumed to be weakly continuous in ;
actually, continuity cannot hold if the global gauge group is simple and has rank at
least two (as in the case of color SU.3/) (see Ref. MS).
4.2 Large Gauge Transformations
By definition, Eq. (4.4.7), the gauge functions considered are localized and, there-
fore, they extend to the one-point compactification RP 3 of R3 , which is isomorphic
3
to the three-sphere S :
P 3
S3 ! G:
U.x/ W R
Such maps fall into disjoint homotopy classes labeled by winding numbers n
Z Z
2 1 3
n.U/ D .24 / d x " Tr .Ui .x/ Uj .x/ Uk .x//
ijk
d3 x nU .x/; (4.4.10)
where Ui .x/ U.x/1 @i U.x/.

The gauge transformations with n 0 are called large gauge transformations.
Those with zero winding number are called small; since they are contractible to
the identity, they are products of U.x/ which are close to the identity (in the C1
topology) and therefore are expressible as products of U . Clearly, all the gauge
transformations of this form have zero winding number.21
Thank to the space localization of the gauge functions the Gauss invariance of
the vacuum implies its invariance under the whole group G of (time independent)
20
J. Lffelholz, G. Morchio and F. Strocchi, Jour. Math. Phys. 44, 5095 (2003).
21
For the geometry of the gauge transformations see S. Coleman, Aspects of Symmetry, Cambridge
Univ. Press 1985, Chapter 7, Sect. 3; R. Jackiw, loc. cit.; T. Frankel, The Geometry of Physics. An
Introduction, 2nd ed. Cambridge Univ. Press 2004.
gauge transformations, (see Ref. MS). Then, the U are implementable by (time
independent) unitary operators V.U/.
The physically crucial question is the implications of the non-trivial topology of
the large gauge transformations on the classification of the physical states. Actually,
the non-triviality of the large gauge transformations on the physical space turns out
to be a rather subtle question, not fully appreciated in the standard treatment, and,
as we shall see below, the presence of fermions plays a crucial role.
An important ingredient is the exploitation of the properties of the so-called
topological current, formally defined by
C .x/ D .16 2 /1 "

Tr.F
.x/ A .x/ 23 A
.x/ A .x/ A .x//: (4.4.11)
@ C .x/ D .16 2 /1 Tr F
.x/ F
.x/ P;
where A D .0; Ai /, F
12 "
F . In the mathematical literature, for classical
fields, P is called the Pontryagin density and C the Chern-Simons secondary
characteristic class.
At the classical level, one can prove the following transformation law of C0 .x/
under gauge transformations U , defined by Eq. (4.4.7)
U .C0 .x// D C0 .x/ .8 2 /1 @i "ijk Tr.@j U.x/ U.x/1 Ak .x// C nU .x/:
(4.4.12)
Therefore, at the classical level the space integral of C0 .x; x0 / is invariant under
small transformations, but it gets shifted by n under gauge transformations with
winding number n.
In the quantum case, one meets non-trivial mathematical problems. First of
all, the formal expression in the right hand side of Eq. (4.4.11) requires a point
splitting regularization. It is very reasonable to assume that this can be done by
keeping the transformation properties of the formal expression under large gauge
transformations, Eq. (4.4.12).
The next problem is the space integral of C0 .x/. Even for conserved currents, the
space integral of the charge density is known to diverge and suitable regularizations
are needed, including a time smearing. Under some general conditions one may
obtain the convergence of a suitably regularized integral of the charge density of a
conserved current, in matrix elements on states with some localization22; but in the
general case of non-conserved currents the problem is seriously open.
For these reasons, one is led to consider a suitably regularized space integral of
C0 .x/
Z
0
C . fR R / d 4 x fR .x/ R .x0 / C0 .x/; (4.4.13)
22
B. Schroer and P. Stichel, Comm. Math. Phys. 3, 258 (1966); H.D. Maison, Nuovo Cim. 11A,
389 (1972); M. Requardt, Comm. Math. Phys. 50, 259 (1976); G. Morchio and F. Strocchi, Jour.
Math. Phys. 44, 5569 (2003).
where fR .x/ RD f .jxj=R/, f .x/ D 1, for jxj 1, D 0, for jxj 1 C ", R .x0 / D
.x0 =R/=R, dx0 .x0 / D .0/ Q D 1.
Actually, an even more serious problem arises in the quantum case, as a
consequence of the gauge dependence of C .x/, i.e. only its exponentials exist. To
this purpose, we consider the unitary operators V C . fR R /, formally the exponentials
exp i C0 . fR R /, with properties inferred from those of such exponentials. In partic-
ular, for any fR with fR D 1 on the space support KU of U, so that fR @j U D @j U,
@i fR @j U D 0, from the gauge transformations of C0 .x/, Eq. (4.4.12), one has that
U .V C . fR R // D einU V C . fR R /: (4.4.14)
Proposition 4.2 The operators V C . fR R /, 2 R, are not weakly continuous in

and therefore the field C0 . fR R / cannot be defined. Furthermore, for all Gauss
invariant vectors ; , one has
.; V C . fR R / / D 0: (4.4.15)
Proof In fact, if C0 . f /, f 2 D.R4 /, exists, by using the Gauss gauge invariance of

the vacuum state !, the vanishing of !.Ak / by rotational invariance and Eq. (4.4.12),
one has
Z
!.C0 . f // D !.V.U /C0 . f /V.U /1 / D !.C0 . f // C d3 xf .x; t/nU .x/:
Since for any f there is at least one U .x/ such that the integral on the right hand
side does not vanish, one gets a contradiction. Thus, only the exponential of C0 . f /
can be defined.
Moreover, given fR one can find a small gauge transformation U.x/, with
U.x/ D U1 .x/ U2 .x/; nU1 C nU2 D 0; nU1 0;

fR U2 D 1; fR U1 D U1 :
Then, @i fR U1 D 0, @i fR @j U D 0 Rand the second term on the right hand side of

Eq. (4.4.12) vanishes; furthermore d3 x nU .x/fR .x/ D nU1 . Hence, one has
.; V C . fR R / / D .; V.U/ V C . fR R / V.U/1 / D
D einU1 .; V C . fR R / /;
and therefore .; V C . fR R / / D 0.
Thus, if only vector fields are present, the topological current and its non
invariance under large gauge transformations cannot give rise to any relevant
structure on the physical states (as claimed in the literature).
4.3 Fermions and Chiral Symmetry Breaking
The situation changes substantially in the presence of massless fermions, since the
role of the topological current is taken by a conserved current; hence, there is a
symmetry associated to it and the crucial point is its relations with the large gauge
transformations and with their implementers.
In this case, the Lagrangian of Eq. (4.4.1) gets modified by the addition of the
(gauge invariant) fermion Lagrangian and the Gauss operators become
Ga D .D E/a j0a ; ja D ig N ta :
The time independent gauge transformations of the fermion fields in the fundamen-
tal representation of the gauge group are
U . .x// D U.x/ .x/: (4.4.16)
At the classical level, the Lagrangian is invariant under the one-parameter group of
chiral transformations , 2 R,
. / D e5 ; . N / D N e5 ; 5 D 5 ; .A/ D A: (4.4.17)
Correspondingly, there is a conserved current j5 D ig N 5 , the gauge invariant

fermion axial current.
In the quantum case, a gauge invariant point splitting regularization is needed for
the definition of j5 and this inevitably leads to an anomaly,23
@ j5 D 2@ C D 2P:
The conserved axial current is now the gauge dependent current
J5 .x/ D j5 .x/ C 2C ;
its conservation being equivalent to the anomaly equation for j5 . For the discussion
of the Weyl quantization, we take as local exponential field algebra FW the
algebra generated by the operators W.f/, by the local implementers of the gauge
transformations V.U/, by the gauge invariant bilinear functions of the fermion fields
and by the unitary operators V 5 . f /, formally given by exp i J05 . f /.
23
For a general review of this phenomenon, see R. Jackiw, Field Theoretical Investigations in
Current Algebra, in S.B. Treiman, R. Jackiw, B. Zumino and E. Witten, Current Algebra and
Anomalies, World Scientific 1985, pp. 81210.
As shown by Bardeen24 on the basis of perturbative renormalization in local

gauges, the chiral transformations are generated by the conserved gauge dependent
current J5 (not by j5 ); the continuity equation of J5 plays a crucial role in Bardeen
analysis.
In fact, since J5 is conserved, most of the standard wisdom is available25 ; in
particular, the commutation relations of VR5 ./ with the local fields are governed
by the canonical (anti)commutation relations and the extension to unequal times is
provided by the current conservation. Then, one has that
lim VR5 ./ F VR5 ./ D .F/; 8F 2 FW : (4.4.18)

R!1
It is important to stress that the limit R ! 1 exists thanks to locality, the limit
being actually reached for finite values of R; furthermore the limit preserves locality
and gauge invariance, i.e. it defines an automorphism of the local exponential
algebra FW .
Thus, contrary to statements appeared in the literature, the presence of the chiral
anomaly does not prevent the chiral symmetry from being a well defined time
independent automorphism of the observable algebra, which is identified as the
gauge invariant subalgebra of the field algebra FW .
Furthermore, the chiral symmetry is locally generated by the unitary operators
VR ./ V 5 . fR R /, the formal exponentials of J05 . fR R /, which act as local
5
implementers of on the local field algebra FW .

In conclusion, the above argument shows that: i) the chiral symmetry defines an
automorphism of the observable algebra ii) and it is locally generated by (gauge
dependent) unitary operators.
The loss of chiral symmetry is therefore a genuine phenomenon of spontaneous
symmetry breaking and the confrontation with the Goldstone theorem becomes
a crucial issue, the so-called U.1/ problem. The gauge dependence of the local
implementers does not allow to a priori dismiss the problem, since also non abelian
symmetries are generated by gauge dependent currents and the evasion of the
Goldstone theorem is not trivially solved by this property.
Actually, the crucial point is the interplay between large gauge transformations
and chiral symmetry. To this purpose we that the gauge transformations of FW are
given by Eqs. (4.4.7), (4.4.15). Therefore, since j50 is gauge invariant, Eq. (4.4.14)
implies that for R large enough, so that fR .x/ D 1 on the localization region of
Un .x/,
Un .VR5 .// D ei 2n VR5 ./: (4.4.19)
It is worthwhile to stress that Eq. (4.4.18) is merely a consequence of the localizabil-

ity of the large gauge transformations and that VR5 ./ has the same transformation
24
W.A. Bardeen, Nucl. Phys. B 75, 246 (1974).
25
G. Morchio and F. Strocchi, Jour. Phys. A: Math. Theor. 40, 3173 (2007).
properties under them as the formal exponential exp i J05 . fR R /. It codifies the
crucial consequence of the axial anomaly, namely that the implementers V.Un /
of the large gauge transformations have a non-trivial relation with the local
implementers of the chiral transformations.
The absence of parity doublets requires that the chiral symmetry be broken in
QCD and the U.1/ problem amounts to explaining the absence of the corresponding
Goldstone massless bosons. Now, as discussed above, one of the basic assumptions
of the Goldstone theorem, namely the existence of a one-parameter group of
automorphisms of the algebra of observables, which commute with space and time
translations is satisfied.
The second crucial property, needed the proof of the theorem,26 is the link
between the vacuum expectation of the infinitesimal variation of the symmetry
breaking order parameter and the local commutator of a conserved current
i.e.
d
< A > < .A/ >D0 D i lim < J05 . fR R /; A > : (4.4.20)
d R!1
Since the chiral automorphism is C1 in the group parameter , its generator is

well defined, but the problem is its relation with the formal generator of the one-
parameter group of unitary operators VR5 ./, which implement the symmetry on the
field algebra FW .
As a matter of fact, even if can be described by the action of the local
operators VR5 ./ on FW , Eq. (4.4.17), the non-regularity of the one-parameter unitary
group VR5 ./, prevents the existence of the corresponding generator J05 . fR R /, so that
one cannot write the symmetry breaking Ward identities and obtain the Goldstone
energy-momentum spectrum of < J05 .x/ A >.
Proposition 4.3 If ! is a Gauss invariant vacuum state and A is a gauge invariant
order parameter for chiral symmetry breaking, i.e.
!. .A// !.A/ 0;
(the standard candidate for A being N ), then the vacuum expectations

!.J05 . fR R / A/ cannot be defined and Eq. (4.4.19) does not hold.
Proof In fact, by exploiting the local properties of the gauge transformations Un
and its space translated one Una by a, one has that for any local operator F, for jaj
large enough, Un .F/ D U0 Una .F/ D U0 .F/, with U0 the Gauss transformation
labeled by Un .Una /1 . Then, the Gauss invariance of ! implies its gauge invariance.
Hence, if the expectations !.J05 . fR R / A/ are defined, by the invariance of ! under
gauge transformations and by Eq. (4.4.18), for R sufficiently large, one would get a
26
For a discussion of the role of this property, see e.g. F. Strocchi, Symmetry Breaking, 2nd ed.
Springer 2008, esp. Part II, Chapter 15.
contradiction
!.J05 . fR R / A/ D !.Un .J05 . fR R / A// D !.J05 . fR R / A/ C 2n !.A/:
The impossibility of writing expectations involving J5 on a gauge invariant

vacuum state, solves the the U.1/ problem and the problems raised by R.J. Crewther
in his analysis of chiral Ward identities.27
By exploiting the non-regular Weyl quantization one can prove that the non-
trivial topology of G gives rise to the vacuum structure and forces the chiral
symmetry breaking in each sector.28
It is worthwhile to remark that, for the evasion of the Goldstone theorem
discussed above, the occurrence of the so-called chiral anomaly (which is present
also in the abelian case) is not enough; the crucial ingredient is Eq. (4.4.18),
which directly implies the non-regularity of the unitary operators VR5 ./ and the
non-existence of the local charges J05 . fR R / generating the chiral symmetry, in
expectations on a gauge invariant vacuum state.
If V.Un / denote local implementers of the large gauge transformations,
Eq. (4.4.17) gives a relation of exactly the same form of the Weyl relation arising in
the description of a quantum particle on a circle (see Chap. 3, Sect. 1), with V.Un /
playing the role of ei 2np and VR5 ./ playing the role of formal exponential ei .
The emerging picture is also the same as in the quantum mechanical model of QCD
structures (see Chap. 3, Sect. 4). It is worthwhile to note that the Schwinger model
in the temporal gauge exactly reproduces the general results discussed above 28 .
5 Abelian Chern-Simons Theory
Another interesting example of the need of Weyl quantization is provided by the

abelian Chern-Simons free field theory29 defined by the following Lagrangian
L D .1=4/F
F
C .=4/ "
F
; F
D @ A
@
A ; (4.5.1)
with A a field in 2+1 spacetime dimensions. The equations of motion
@ F
C 12 "
F D 0; (4.5.2)
27
R.J. Crewther, Chiral Properties on Quantum Chromodynamics, in Field Theoretical Methods in
Particle Physics, W. Rhl ed., Reidel 1980, pp. 529590.
28
See G. Morchio and F. Strocchi, Ann. Phys. 324, 2236 (2009); F. Strocchi, An introduction to
non-perturbative foundations of quantum field theory, Oxford Univ. Press 2013.
29
F. Nill, A constructive approach to abelian Chern-Simons theory, in Theory of Elementary
Particles, G. Weigt ed., Inst. Hochenenergiephys. 1991, p. 78; Int. Jour. Mod. Phys. 6, 2159 (1992).
5 Abelian Chern-Simons Theory 75
imply . C 2 /F
D 0. They are invariant under the gauge transformations A !
A C @ , whereas the Lagrangian changes by a total derivative. It is convenient
to discuss the quantization in the temporal gauge, defined by A0 D 0.30 The Gauss
operator
G D @i i C 12 "i j @j Aj ; i D F0 i C 12 "i j Aj (4.5.3)
is independent of time. As a consequence of the canonical (equal time) commutation

relations
Ai .x/; j .y/ D ii j .x y/; (4.5.4)
all other equal time commutators vanishing, the Gauss operator generates the time
independent gauge transformations.
As for the cases discussed in Sects. 3, 4, the Gauss law must be imposed as a
condition of the physical states. We consider the local (exponential) field algebra
FW generated by the Weyl operators W.f/, f i .x/ 2 S.R3 /, formally the exponentials
of A.f/ D Ai . f i /, and by the unitary operators V.U /, U D eiG.g/ , g 2 D.R2 /,
formally the exponentials of the Gauss operator, representing the group G of time
independent gauge transformations
R
d 2 x fi @i g
V.U / W.f/V.U /1 D ei W.f/: (4.5.5)
The algebraic structure of FW is defined by the canonical (equal time) commutation

relations and by the time evolution generated by the Hamiltonian
Z
HD 1
2 d2 x E2 C B2 ; Ei D F0 i ; B D "i j @i Aj : (4.5.6)
A Gauss invariant vacuum state ! on FW satisfies the Gauss law !.V.U // D 1

and the physical vector states in the corresponding (GNS) Hilbert space H! are
characterized by the Gauss law
V.U / D : (4.5.7)
Similarly to the previous cases, we have

Proposition 5.1 A vacuum state ! on FW , satisfying the Gauss law defines a non-
regular representation of FW , since
!.W.f// D 0; if @i f i 0: (4.5.8)
30
S. Deser, R. Jackiv and S. Templeton, Ann. Phys. 140, 372 (1982).
Thus, the fields A.f/, cannot be defined in the Hilbert space H! , if @i f i 0 (non-
regular Weyl quantization).
Proof As before,
R
d 2 x fi @i g
!.W.f// D !.V.U / W.f/ V.U // D ei !.W.f//

The non-regularity of the representation defined by a vacuum state satisfying
the (suitably modified) Gauss law may be similarly proved in the presence of
(interacting) fermions, with interaction term
Lint D A j ; j D e N :
Clearly, the same argument of non-regular representation applies to the pure non-
abelian Chern-Simons theory in 2+1 dimension defined in the temporal gauge by the
Lagrangian L D 12 "i j AP ai Aaj .31
31
G.V.Dunne, R. Jackiw and C.A. Trugenberger, Ann. Phys. 194, 197 (1989).
Chapter 5
Diffeomorphism Invariance and Weyl Polymer
Quantization
1 Diffeomorphism Invariance and Weyl Polymer

Quantization
Especially in connection with Quantum Gravity, great attention has been paid to the
problem of combining diffeomorphism invariance and quantization.
According to Wigner theorem on quantum symmetries, diffeomorphism symme-
try, or equivalently diffeomorphism covariance, of a quantum theory requires the
implementation of the elements of the diffeomorphism group by unitary operators
in the Hilbert space of the (quantum) states.
Now, the diffeomorphism group of a manifold is infinite dimensional and the
representation of infinite dimensional groups is not yet under mathematical control;
this is one of the reasons of the relevance and difficulty of the problem.
For the infinite dimensional local gauge group the difficulty of quantization has
been overcome through the choice of the gauge fixing, which breaks the local gauge
symmetry and reduce it to the identity on the physical states characterized by the
fulfillment of a subsidiary condition; a crucial point is the independence of the
expectation values of the observables from the choice of the gauge fixing.
For the case of diffeomorphism symmetry such a route is made difficult by the
lack of a full control on the representations of the diffeomorphism group. To this
purpose it has been suggested, in particular in connection with Loop Quantum
Gravity (LQG) and String Theory (ST), to use a non-regular Weyl (also called
polymer) quantization1 and to guarantee the unitary implementation of the elements
1
For the vast literature on this subject we refer to the Bibliography of Publications related to
Classical Self-dual variables and Loop Quantum Gravity, see: A. Corichi and A. Hauser, arXiv:gr-
qc/0509039v2. For very recent reviews of LQG, see H. Sahlmann, Loop quantum gravity-a
short review, arXiv:1001.4188v1 [gr-qc]; S. Mercuri, Introduction to Loop Quantum Gravity,
arXiv:1001.1330v1 [gr-qc].

78 5 Diffeomorphism Invariance and Weyl Polymer Quantization
of the diffeomorphism group by the stronger condition that the Hilbert space of
the (quantum) states is the GNS representation space defined by a diffeomorphism
invariant (vacuum) state.
Such a proposal has raised both interest and discussions and, in our opinion, it
may be useful to focus the problem starting with the question of diffeomorphism
invariance and quantization in general. As one should expect on the basis of
quantum gauge models, the diffeomorphism invariance of the ground state leads
to a non-regular Weyl quantization. To get some insight, in the following we shall
start by considering the case of diffeomorphism covariance in quantum mechanics.
2 Quantum Mechanics on a Manifold and Diffeomorphism

Invariance
The problem of the formulation of quantum mechanics on a manifold M as

been largely discussed in the literature under suitable restrictions or choices and
recently a general approach which implements diffeomorphism covariance has been
proposed.2 We shall briefly review it, also because it shows that a diffeomorphism
covariant quantization may be obtained in a rather simple and natural way, once
the question of the independent degrees of freedom is properly settled. In fact, by
exploiting the relations induced on the Lie algebra of vector fields of the manifold
by multiplications by C1 functions, locally the number of independent vector fields
reduces to the dimension of the manifold.
i) Algebraic description of a manifold and its diffeomorphisms
The geometry of M is completely described by the algebra of the C1 real
functions on M and by the group Diff (M) of diffeomorphisms of M. At the
local level the relevant group is the subgroup Diff .M/ of the connected component
of the identity of Diff (M), generated by the one-parameter groups gv , 2 R,
v 2 Vect.M/ the Lie algebra of C1 vector fields of compact support.
Thus, as a basic geometrical structure one has
i) the algebra C1 .M/ generated by the real C1 functions of compact support and
by the constant functions on M and
ii) the space Vect (M/ of C1 real vector fields of compact support in M.
Vect(M// is a Lie algebra of derivations v 2 Vect(M// on C1 .M/, v W C1 .M/ 3
f ! v. f /, with Lie product fv; wg defined by
fv; wg. f / D v.w. f // w.v. f //: (5.2.1)
2
G. Morchio and F. Strocchi, Lett. Math. Phys. 82, 219 (2007).
2 Quantum Mechanics on a Manifold and Diffeomorphism Invariance 79
The action of Vect.M/ on C1 .M/ can also be interpreted as an extension of the

Lie product from Vect.M/ to C1 .M/ C Vect.M/ L.M/. Putting
f v; f g v. f /; f f ; g g 0; (5.2.2)
L.M/ gets the structure of a Lie algebra.

In order to discuss the representations of such an algebraic structure it is
convenient to introduce the analogs of the Weyl operators, namely the operators
W. f /, f 2 C1 .M/, formally the exponentials eif , and the operators V.v/, v 2
Vect.M/, formally the exponentials eiv . The operator V.v/, 2 R, can also be
labeled by the corresponding element of the one-parameter group g.v/ generated
by v; hence, with a slight abuse of notations we shall denote V.v/ also by
V.g.v//. In general, by V.g/ we shall denote the operator labeled by the generic
group element g 2 Diff.M/.
We introduce the following notations
fg.v/ .x/ f .g.v/1 x/ .v . f //.x/;

v .w/ g.v/.w/ D the adjoint action of Diff.M/ on Vect.M/:
Then, the analogs of the Weyl relations read
W. f / W.g/ D W. f C g/; V.v/ W. f / V.v/1 D W. fg.v/ /; (5.2.3)

V.v/ V.w/ V.v/1 D V.g.v/w/; V.g.v// V..g.w// D V.g.v/g.w//: (5.2.4)
The reality of C1 .M/ and of Vect.M/ induce the following involution on the
above defined Weyl operators
W. f / D W.f /; V.v/ D V.v/ (5.2.5)
and the polynomial algebra A generated by the Weyl operators becomes a *-algebra.
By construction, Eqs. (5.2.35.2.4), Diff(M) defines a group of *-automor-
phisms h , h 2 Diff(M) of A
h .W. f // D W. fh /; h .V.v// D V.hv/:
Since Vect.M/ is generated by an infinite number of linearly independent vector

fields, Diff.M/ is an infinite dimensional group and the algebra A is infinite-
dimensional. Thus, it is not easy to analyze the positive functionals on it and its
representations are not under complete mathematical control.
ii) Quantum mechanics with unitary implementation of Diff(M)

To overcome this difficulty, it has been suggested, also on the basis of physical
considerations,3 to take into account that different vector fields may be functionally
dependent through multiplication by elements of C1 .M/ . Indeed, as a module over
C1 .M/ , Vect.M/ is locally generated by n vector fields, with n the dimension of
M.
We recall that an algebra A is a module over a commutative ring C if there is a
product C A A, which is distributive in both factors and associative in the first,
i.e. 8f ; g 2 C; a; b 2 A,
f .a C b/ D f a C f b; . f C g/ a D f a C g a;
. fg/ a D f .g a/: (5.2.6)
If L is a (real) Lie algebra, with Lie product denoted by f ; g, which acts as

derivations on a (real) commutative algebra L0 , namely 8v 2 L; L0 3 f ! v. f /,
and L is a module over L0 with product satisfying
ff v; gg D f fv; gg; f .g v/ D . fg/ v;

fv; f wg D v. f / w C f fv; wg; (5.2.7)
then is a Lie-Rinehart (LR) product and the pair .L0 ; L/ is (called) a Lie-Rinehart
(LR) algebra. A LR algebra is said to have an identity if L0 has an identity 1,
satisfying 1 v D v; 8v 2 L.
A module structure of Vect(M) on C1 .M/ is realized by the product f v D w 2
Vect(M), where w is the vector field with components . f @=@xi /, if vi D @=@xi . It is
easy to check that the so-defined product satisfies Eqs. (5.2.6), (5.2.7)
Thus, the pair .C1 .M/; Vect M/ is a LR algebra with identity 1 given by
the function 1.x/ D 1; the crucial consequence is that, as a module over C1 .M/,
Vect .M/ is locally generated by n vector fields, with n the dimension of M, so that
the infinite dimensional Diff(M) and its Lie algebra may be described by a finite
number of generators. This means that, for any region O diffeomorphic to a disc,
one can find n vector fields pi , supp pi O, so that forPany v, with supp v O,
there exist n functions fi 2 C1 .M/ such that v D i fi pi . This means that
any vector v with supp v O, is functionally dependent on the n vector fields pi ,
through the LR product.
Since a quantum particle on a manifold has a finite number of degrees of freedom,
the reduction of the algebra A to finite dimensions is necessary for its quantum
mechanical use and the natural way is to take into account the Lie-Rinehart structure
of .C1 .M/; Vect M/.
3
Definition 2.1 A quantum particle regular representation of the algebra

A, generated by the Weyl operators W. f /, V.v/, Eqs. (5.2.35.2.5), is a homo-
morphism of A into the algebra of bounded operators in a Hilbert space H with
the following properties:
i) (regular representation) .W.f //, .V.v//, 2 R, define weakly continu-
ous one-parameter unitary groups and the algebra generated by the W. f / is not
trivially represented; the corresponding generators . f / Tf ; .v/ Tv exist
on a common dense domain D invariant under .A/,
ii) (Lie-Rinehart structure) the generators represent the Lie-Rinehart algebra
.C1 .M/; Vect M/ on D, i.e. the Lie algebra relations of the vector fields and of
their action over C1 .M/ , with the Lie-Rinehart product represented by
. f v/ D 12 . f /.v/ C .v/. f /: (5.2.8)
Clearly, a Hilbert space representation of A provides a quantization in which

the diffeomorphism group is implemented by unitary operators, i.e. one has a
diffeomorphism covariant quantum system.
Quantum particle regular representations of A have been shown to exists, to be
locally unitarily equivalent to the Schrdinger representation in L2 .M; d/, where
d is the Lebesgue measure, .W. f // act as multiplication operators and .V.v//
act in the following way, 8 2 L2 .M; d/,
.V.v// .x/ D .g.v/1 .x// J.g.v/; x/;

J.g.v/; x/ d.g.v/1 .x//=d.x/1=2 : (5.2.9)
Globally, such representations are in one-to-one correspondence to the unitary

irreducible representations of the first homotopy group of M.4
iii) Diffeomorphism invariant states
The algebra A is generated by the monomials W. f / V.v/, f 2 C1 .M/, v 2
Vect.M/ , since, as a consequence of the generalized Weyl relations, Eqs. (5.2.3
5.2.4), products of W. f /; V.v/ reduce to monomials of this form. Hence, a linear
functional ! on A is completely determined by the expectations !.W. f / V.v//.
A state ! on A is diffeomorphism invariant if
!.v .A// D !.A/; 8v 2 Vect.M/; 8A 2 A: (5.2.10)
By a standard argument, in the GNS representation defined by a diffeomorphism

invariant state !, Diff(M) is unitarily implemented by the operators U.v/
4
defined by
U.v/A! D v .A/ ! ; 8 A 2 A; U.v/ ! D ! ; (5.2.11)
with ! the representative vector of !.

Then, by Eqs. (5.2.35.2.4) the operators U.v/ V.v/1 commute with A and
therefore, if the representation is irreducible, they are multiple of the identity;
since Diff.M/ has no non-trivial one-dimensional representation, without loss of
generality, one may put V.v/! D ! , which implies 8f 2 C1 .M/; v; w 2
Vect.M/
!.W. f /V.v// D !.W. f // D !.W.g.w/. f //: (5.2.12)
Clearly, the requirement that a representation be defined by a diffeomorphism

invariant state is a much stronger requirement than the realization of diffeomorphism
symmetry.
More generally, it is worthwhile to stress the crucial difference, not sufficiently
emphasized in the literature, between the implementation of the diffeomorphism
group by unitary operators, the existence of a diffeomorphism invariant vector state,
the coincidence of the unitary implementers with the Weyl operators V.v/ and the
weak continuity of the latter.
Proposition 2.2 The GNS representation defined by a diffeomorphism invariant
state cannot be a regular quantum particle representation.
Proof In fact, the regularity of the representation implies the existence of the
generators and diffeomorphism invariance of the state ! gives
!.v; W. f // D 0; 8f 2 C1 .M/ ; 8v 2 Vect.M/:
This is incompatible with local equivalence to the Schrdinger representation,

Eq. (5.2.9 ), and therefore with regularity.
The next non-trivial question is the existence and the explicit realization of
diffeomorphism invariant states. For this purpose, we start by considering diffeo-
morphism invariant (hermitian) functionals, characterized by Eq. (5.2.10); then one
has to find the conditions which guarantee the property of positivity.
Proposition 2.3 A diffeomorphism invariant hermitian functional ! on A is posi-
tive iff j!.W. f //j 1, 8f 2 C1 .M/ .
Proof In fact, putting W. f ; v/ W. f / V.v/, one has
!..aW. f ; v/ C bW.g; w// .aW. f ; v/ C bW.g; w// D

D jaj2 C jbj2 C 2Re .a b !.W.g f //
and positivity follows.

On the other hand, if ! is positive, in the corresponding GNS representation ! ,

! .W. f // is a unitary operator and therefore j!.W. f /j D j.! ; W. f / ! /j 1.
For simplicity, we shall discuss the following two extreme cases
1) j!.W. f //j D 1, 8f 2 C1 .M/
2) !.W. f / D 0, if f is not a constant function, i.e. if f is not a multiple of the
identity, f const 1.
Proposition 2.4 The GNS representation space defined by a diffeomorphism invari-
ant state ! with
j!.W. f //j D 1; 8f 2 C1 .M/ ;
is one-dimensional.
Proof In fact, j!.W. f //j D 1 implies W. f /! D e. f / ! and since ! is
invariant under the V.v/, . f / must yield a representation of Diff(M). Since the
only one-dimensional representation is the trivial one, . f / must be a constant
independent of f and without loss of generality one can take W. f / ! D ! .
Hence, the generating monomials W. f ; v/ W. f / V.v/ leave ! invariant and
by the cyclicity of ! the Hilbert space H! is one-dimensional.
Proposition 2.5 The GNS representation defined by a diffeomorphism invariant
state ! with
!.W. f // D 0; if f const 1
has the following properties

a) ! is a cyclic vector for the algebra AW generated by the Weyl operators W. f /
and any two vectors W. f /! , W.g/! are orthogonal if f g const 1,
b) even if ! is diffeomorphism invariant, the one-parameter groups .V.g.v///
are not weakly continuous and therefore the corresponding generators do not
exist as operators in H!
Proof a) follows immediately from !.W. f // D 0, if f const 1. Moreover, one
has
..W. f //! ; .V.g.v/// .W. f // ! / D !.W. fg.v/ f //:
and the right hand side is equal to 1 if D 0 and vanishes otherwise. Hence, weak
continuity fails.
In connection with Loop Quantum Gravity (LQG) and String Theory (ST), it has
been suggested to use diffeomorphism invariant states with very strong invariance
properties, the so-called polymer states.5 Their analog for the algebra A is the state
!.W. f /V.v// D 0; if f const 1; or v 0:
The invariance under diffeomorphisms is obvious; the representation is highly

reducible and corresponds to a thermal state in the limit of infinite temperature.6
Both unitary groups .W.f //, .Vv// are not weakly continuous and the
corresponding generators do not exist.
An instructive example of diffeomorphism invariant state may be worked out in
the case of M D S1 , where the topology plays a non-trivial role.
5
Such a kind of states were proposed by A. Ashtekar, S. Fairhurst and J.L. Willis, Class. Quantum
Grav. 20, 1031 (2003); T. Thiemann, Class. Quantum Grav. 23 1923 (2006).
6
F. Acerbi, G. Morchio and F. Strocchi, Jour. Math. Phys. 34, 899 (1993).
Chapter 6
* A Generalization of the Stone-von Neumann
Theorem
1 Zak Algebra
The relevance of non-regular representations of the Heisenberg group (or of the

Weyl C -algebra AW ) raises the question of a possible classification of them,
which generalizes Stone-von Neumann (SvN) theorem. For this purpose, a possible
strategy is to consider a maximal abelian A subalgebra of AW , identify its Gelfand
spectrum .A/ and classify the realizations of such an abelian algebra in terms of
multiplication operators on L2 ..A/; d/, with d a (Borel) measure on .A/.
In the Schrdinger choice, such an abelian algebra is the C -algebra Aq
generated by the Weyl operators U./, 2 Rd , (d the space dimension), formally
corresponding to eiq , (equivalently one may consider the algebra Ap generated by
the V./, 2 Rd formally given by eip ). Aq is isomorphic to the algebra of almost
periodic functions f .x/, x 2 Rd ; by Gelfand theorem it is also isomorphic to the
algebra of continuous functions on the compact space .Aq /, whose description is
however not simple.1
For these reasons, it is convenient to consider the maximal abelian algebras
AZ ./ generated by the pair of elements Ui .2=/, Vi ./, i D 1; : : : d, for a fixed
positive (formally corresponding to exp i .2=/qi ; exp i pi ;, respectively).
Up to isomorphisms we can choose D 1; we shall call such an algebra the Zak
algebra and denote it by AZ .
For simplicity, in the following we shall consider the one-dimensional case; then
AZ is generated by only two elements, with spectra given by the unit circle, so that
1
J. Lffelholz, G. Morchio and F. Strocchi, Lett. Math. Phys. 35, 251 (1995).

86 6 * A Generalization of the Stone-von Neumann Theorem
its Gelfand spectrum is very simple

Proposition 1.1 The Gelfand spectrum of AZ is homeomorphic to the two dimen-
sional torus
T2 D f.1 ; 2 / 2 C2 I j1 j D j2 j D 1g: (6.1.1)
In the following we shall identify T2 with R=Z R=.2Z/ D 0; 1/ 0; 2/ and

a point .; / 2 T2 with the multiplicative linear functional
m .U.2/ V.1// D ei2 ei :
By the Gelfand-Naimark characterization of abelian C -algebras, AZ is isomorphic

to the algebra of continuous functions on T2 , the isomorphisms being given by the
Gelfand transform, A ! A. O
Proposition 1.2 The elements T.a; b/ U.a/ V.b/ eiab=2 , a; b 2 R define

automorphisms of AZ : a b .A/ D T.a; b/ A T.a; b/, A 2 AZ , which correspond
to translations of the Gelfand transforms.
The Schrdinger representation S is isometrically isomorphic to the following
representation Z in terms of L2 functions on (the Gelfand spectrum of AZ ) T2 :
8 2 L2 .T2 ; d d/,
.Z .W.a; 0///.; / D eiCa .. C a/ mod 1; /; (6.1.2)

.Z .W.0; b///.; / D eib .; . C b/ mod 2/: (6.1.3)
The isometry U W L2 .R/ 3 $ 2 L2 .T2 ; d d/ is given by

X
.U /.; / D .n C /ein ; (6.1.4)
n2Z
Z 2
.U 1 /.x/ D .2/1 d .x mod 1; / eix ; (6.1.5)
0
where x denotes the integer part of x. In the following, we shall analyze the
representations of AW in terms of representations of AZ .
2 A Generalization of Stone-von Neumann Theorem
For a generalization of SvN theorem which yields a classification of non regular

representations of the Weyl algebra (equivalently of non-regular unitary representa-
tion of the Heisenberg group), one has to find a weaker counterpart of irreducibility
and regularity.
2 A Generalization of Stone-von Neumann Theorem 87
As we shall see, a useful generalization is provided by the class of representations

which are spectrally multiplicity free (a generalization of irreducibility) and satisfy
a notion of strongly measurability, which applies also to non-separable spaces,
replaces regularity and it is satisfied by all non-regular representations of physical
interest discussed in the previous Chapters.
A representation of the Weyl algebra satisfying such conditions will be shown
to be unitarily equivalent to the representation given by Eqs. (6.1.2), (6.1.3) on
functions on T2 which are square integrable with respect to a (positive) trans-
lationally invariant measure , which reduces to the Lebesgue measure on T2 iff
the representation is regular, and otherwise is a general (positive) translationaly
invariant Borel measure on T2 .
To this purpose, we start by recalling that a Borel set in a topological space is
any set which can be obtained by taking countable unions, countable intersections
and complements of open sets. The family of Borel sets form a Borel -algebra. A
Borel measure is a measure defined on the Borel algebra and it is regular if for
any Borel set S it satisfies: .S/ D supC f.C/I C S; C compact and Borelg D
infO fO S; O openg: The Baire -algebra on a compact topological space is the
minimal -algebra of sets needed for the measurability of the continuous functions;
a Baire measure is a measure defined on the Baire -algebra.
Given a representation of a C -algebra A in a Hilbert space H, the Baire *-
algebra is the smallest C -algebra of operators in H which contains .A/ and the
limit of each weakly convergent monotone sequence.
As in the standard case, the strategy is to classify representations in which
the maximal abelian subalgebra AZ is represented by multiplicative operators on
L2 .T2 ; d/, with a Borel measure of T2 , so that and element A of AZ is
represented by the multiplication operator given by the continuous function defined
by its Gelfand transform A. O The advantage of the Zak algebra is that the space on
which such continuous functions are defined is the two-dimensional torus.
Thus, as a first step, one has to find a condition on the representation which
guarantees that its representation space is isomorphic to a space L2 .T2 ; d/, with
a Borel measure.
Definition 2.1 A representation of a (unital) abelian C -algebra A is spectrally
multiplicity free if there exists a positive measure on the Baire sets of (the
Gelfand spectrum) AO .A/ and an isometric isomorphism U of the representa-
O d/, such that U.A/U 1 is the multiplication operator
tion space H onto L2 .A;
O
by the Gelfand transform A.
A representation of a C -algebra A in H is non-degenerate iff H0 fx 2
H ; .A/x D 0; 8A 2 Ag D f0g.
Then, the property of being spectrally multiplicity free is in some way the gener-
alized counterpart of irreducibility and non-degeneracy for regular representations.
Proposition 2.2 A regular representation of the Weyl algebra AW is non-
degenerate and irreducible if and only if it is spectrally multiplicity free as a
representation of AZ .
Proof By an explicit control, one can see that the Schrdinger representation,
and therefore all the regular (non-degenerate) irreducible representations of AW ,
are spectrally multiplicity free for AZ . For the converse, spectral multiplicity free
implies non-degeneracy and by the Stone-von Neumann theorem the representation
is unitarily equivalent to a direct sum of Schrdinger representations. Multiplicity
free, implied by spectral multiplicity free, requires that such a sum contains only
one term.2
Furthermore, one has3
Theorem 2.3 Let be a representation of a (unital) abelian C -algebra A, then
the following statements are equivalent
i) the representation is spectrally multiplicity free
ii) for each vector x 2 H the projection on the cyclic subspace Hx D .A/x
belongs to the Baire *-algebra generated by .A/ (i.e. the smallest Baire *-
algebra in H containing .A/)
O such that x .A=S
iii) for every x 2 H , there exists a Baire subset Sx of A, O x / D 0,
?
and y .Sx / D 0, 8y 2 Hx , where x denotes the positive Baire measure on the
Gelfand spectrum AO defined by x via the Riesz-Markov representation theorem
(called the spectral measure of x).
The next problem is to find a generalized substitute of regularity. For this purpose,
we recall that, for representations in a separable Hilbert space, strong measurability
is equivalent to strong continuity and therefore to regularity. Now, there is a notion
of strong measurability in non-separable Hilbert spaces, which provides the needed
generalization of regularity.4
Definition 2.4 Let .X; M; / be a positive -finite measure space and H a Hilbert
space.
A function F W X ! H is said to be countably-valued if it assumes at most a
countable set of values in H, each value being taken on a measurable set. A function
F W X ! H is called measurable with respect to if there exists a sequence of
countably-valued functions converging -almost everywhere to it.
An operator valued function F W X 3 a ! F.a/ 2 B.H/ is strongly measurable
w.r.t. if, 8x 2 H, the vector-valued function F.a/x is measurable with respect to
.
Theorem 2.5 Let .X; M; / be a positive -finite measure space and H a Hilbert
space.
2
The detailed argument is given in S. Cavallaro, Ph. D Thesis, academic year 1996/97, ISAS,
Trieste, Chapter V, Sect. 3.
3
S. Cavallaro, Algebras and Representation Theory, 3, 175 (2000).
4
E. Hille and R.S. Phillips, Functional Analysis and Semigroups, Amer. Math. Soc. Colloq. Pub.
Vol. 31, New York 1948, Theors. 3.5.3, 3.5.5.
2 A Generalization of Stone-von Neumann Theorem 89
An operator valued function F W X ! B.H/ is strongly measurable w.r.t. to

iff
i) it is weakly measurable
ii) F.a/x; a 2 X; is -almost separably-valued for every x 2 H (i.e. there is a
-null measurable subset N X such that fF.a/xI a 2 X=Ng is separable.
The following Theorem provides a generalization of Stone-von Neumann theo-
rem by giving a characterization of all representations of the Weyl algebra which are
spectrally multiplicity free on AZ and satisfy a condition of strong measurability
Theorem 2.6 5
Let be a representation of the Weyl algebra AW with the following
properties
1) is spectrally multiplicity free as a representation of the Zak subalgebra AZ
2) the operator-valued function T2 ! .W.a; b// is strongly measurable w.r.t.
every positive spectral measure y , y 2 H .
Then, is an irreducible representation of AW and there exists a positive trans-
lationally invariant measure on the Borel -algebra of T2 , such that is unitarily
equivalent to the following representation on L2 .T2 ; d/, (see Eqs. (6.1.2), (6.1.3)),
..W.a; 0///.; / D eiCa .. C a/ mod 1; /; (6.2.1)

..W.0; b///.; / D eib .; . C b/ mod 2/; (6.2.2)
where C a denotes the integer part of C a.

Moreover, the measure can be written as the sum of a family of finite positive
Borel measures concentrated on disjoint sets.
A comment may be helpful concerning the strong measurability condition. In
general, for each y 2 H the set f.W.a; b// yI .a; b/ 2 T2 g is in general non-
separable, but by Theorem 2.6, strong measurability with respect to every positive
spectral measure x , implies that for every x there exists a Borel set N of T2 such
that x .N/ D 0 and f.W.a; b// yI .a; b/ 2 T2 =Ng is separable. This allows to make
local use of standard results which hold only in separable spaces (and for -finite
measures), even if H is non-separable.
It is worthwhile to remark that all the non-regular representations, arising in the
discussion of the physical problems discussed in the previous Chapters, are covered
by such a classification theorem.
The Schrdinger representation is given by the Lebesgue measure on T2 , with
representation space L2 .T2 ; .2/1 d; d/, see Eqs. (6.1.2)(6.1.5).
The representations appearing in the examples of Chaps. 2, Sect. 2 (translation-
ally invariant state), Sect. 3 (Bloch electron), and Chap. 3, Sect. 2 (Jackiw model),
5
S. Cavallaro, G. Morchio and F. Strocchi, Lett, Math. Phys. 47, 307 (1999). At the moment, we
cannot offer a simpler version of the rather technical proof presented there.
Sect. 3 (Christ-LeePmodel), Sect. 4 (QM model of QCD structures) correspond to

the measure D 20;2/ d , where d denotes for each 2 0; 2/ the one-
dimensional Lebesgue measure concentrated on the segment f.; /I 2 0; 1/g
fT2 g.
The representation defined by the Zak states in the discussion of the quantum
Hall electron (Chap. 2, Sect. 5), corresponds to the counting measure on T2 , with
representation space l2 .T2 /.
Since, as pointed out by G. Mackey,6 the Stone-von Neumann theorem can
be generalized in the direction of representations of locally compact groups, the
above classification of non-regular representations may provide an extension of the
standard strategy for the classification of the representations of such groups.
6
G. Mackey, Duke Math. J. 16, 313 (1949).
Bibliography
F. Acerbi, Master thesis, SISSA Trieste, 1990

F. Acerbi, G. Morchio, F. Strocchi, Jour. Math. Phys. 34, 899 (1993)
F. Acerbi, G. Morchio, F. Strocchi, Lett. Math. Phys. 27, 1 (1993); ibid 26, 13 (1992)
S.M. Apenko, Phys. Lett. A 142, 277 (1989)
N.W. Ashcroft, Solid State Physics (Saunders College Publishing, Fort Worth, 1976)
A. Ashtekar, S. Fairhust, J.L. Wills, Classical Quantum Gravity 20, 1031 (2003)
W.A. Bardeen, Nucl. Phys. 75, 246 (1974)
A.O. Barut, R. Raczka, Theory of Group Representations and Applications (World Scientific,
Singapore, 1986)
A. Bassetto, G. Nardelli, R. Soldati, Yang-Mills Theories in Algebraic Non-covariant Gauges
(World Scientific, Singapore, 1991)
K. Baumann (J. Math. Phys. 28, 697 (1987)); Lecture at the Schladming School 1987, in Recent
Developments in Mathematical Physics, ed. by H. Mitter, L. Pittner (Springer, New York, 1987)
R. Baume, J. Manuceau, A. Pellet, M. Sirugue, Commun. Math. Phys. 38, 29 (1974)
J. Belissard, Quantum systems periodically perturbed in time, in Fundamental Aspects of Quantum
Theory, ed. by V. Gorini, A. Frigerio (Plenum Press, New York, 1986)
F. Bloch, Z. Physik 52, 555 (1928)
E. Brown, Aspects of group theory in electron dynamics, in Solid State Physics, ed. by F. Seitz
et al. (Academic Press, New York, 1968)
D. Buchholz, K. Fredenhagen, Commun. Math. Phys. 84, 1 (1982)
C.G. Callan, R.F. Dashen, D. Gross, Phys. Rev. Lett. B 63, 34 (1976)
S. Cavallaro, G. Morchio, F. Strocchi, Lett. Math. Phys. 47, 307 (1999)
S. Cavallaro, Algebras Representation Theory 3, 175 (2000)
N. Christ, T.D. Lee, Phys. Rev. D 22, 939 (1980)
S. Coleman, Aspects of Symmetry (Cambridge University Press, Cambridge, 1985)
S. Coleman, The uses of instantons, in Aspects of Symmetry (Cambridge University Press,
Cambridge, 1985)
A. Corichi, A. Hauser, arXiv:gr-qc/0509039v2
A.A. Cottey, Am. J. Phys. 39, 1235 (1971)
S. Deser, R. Jackiv, S. Templeton, Ann. Phys. 140, 372 (1982)
P.A.M. Dirac, Can. J. Phys. 33, 650 (1955)
P.A.M. Dirac, The Principles of Quantum Mechanics (Oxford University Press, Oxford, 1986)
D.A. Dubin, M.A. Hennings, T.B. Smith, Mathematical Aspects of Weyl Quantization and Phase
(World Scientific, Singapore, 2000)
G.V. Dunne, R. Jackiw, C.A. Trugenberger, Ann. Phys. 194, 197 (1989)

Lecture Notes in Physics 904, DOI 10.1007/978-3-319-17695-6
92 Bibliography
R. Ferrari, Int. J. Mod. Phys. 12, 1105 (1998)

G. Floquet, Ann. cole Norm. Sup. 12, 47 (1883)
G.B. Folland, Harmonic Analysis in Phase Space (Princeton University Press, Princeton, 1989)
T. Frankel, The Geometry of Physics. An Introduction, 2nd edn. (Cambridge University Press,
Cambridge, 2004)
J. Frhlich, Commun. Math. Phys. 54, 135 (1977)
P. Furlan, G.M. Sotkov, I.T. Todorov, Two-dimensional quantum field theory. Riv. Nuovo Cim. 12,
1 (1989)
J. Glimm, A. Jaffe, Quantum Physics. A Functional Integral Point of View (Springer, New York,
1987)
H. Grundling, C.A. Hurst, Lett. Math. Phys. 15, 205 (1988)
F.D.M. Haldane, Phys. Rev. Lett. 55, 2095 (1985)
B.F. Hatfield, Quantum Field Theory of Point Particles and Strings (Addison-Wesley, Reading,
1992)
E. Hille, R.S. Phillips, Functional Analysis and Semigroups, Vol. 31 (Amer. Math. Soc. Colloq.
Pub., New York, 1948)
R. Jackiw, Rev. Mod. Phys. 49 , 681 (1977)
R. Jackiw, Topological effects on the physics of the standard model, in Lectures on QCD.
Foundations, ed. by F. Lenz et al. (Springer, New York, 1997)
R. Jackiw, Topological investigations of quantized gauge theories, in Current Algebra and
Anomalies, ed. by S.B. Treiman, R. Jackiw, B. Zumino, E. Witten (World Scientific, Singapore,
1985)
R. Jackiw, C. Rebbi, Phys. Rev. Lett. 37, 172 (1976)
P.E.T. Jorgensen, R.T. Moore, Operator Commutation Relations (Reidel, Dordrecht, 1984)
R. Jost, The General Theory of Quantized Fields (American Mathematical Society, Providence,
1965)
T.D. Lee, Particle Physics and Introduction to Field Theory (Harwood Academic, New York, 1990)
J. Lffeholz, G. Morchio, F. Strocchi, Lett. Math. Phys. 35, 251 (1995)
J. Lffelholz, G. Morchio, F. Strocchi, Ann. Phys. 250, 367 (1996)
J. Lffelholz, G. Morchio, F. Strocchi, J. Math. Phys. 44, 5095 (2003)
G. Mackey, Duke Math. J. 16, 313 (1949)
W. Magnus, S. Winkler, Hills Equation (Wiley, New York, 1966)
H.D. Maison, Nuovo Cim. 11A, 389 (1972)
J. Manuceau, M Sirugue, D. Testard, A. Verbeure, Commun. Math. Phys. 32, 231 (1973)
S. Mercuri, Introduction to loop quantum gravity, arXiv:1001.1330v1 [gr-qc]
G. Morchio, F. Strocchi, Removal of the infrared cutoff, seizing of the vacuum and symmetry
breaking in many body and in gauge theories, invited talk at the IX Int. Conf. on Mathematical
Physics, Swansea 1988, ed. by B. Simon et al. (Adam Hilger, Bristol, 1989)
G. Morchio, F. Strocchi, Boundary terms, long range effects, and chiral symmetry breaking,
Lectures at the Schladming School 1990, Fields and Particles, ed. by H. Mitter, W. Schweigen
(Springer, New York, 1990)
G. Morchio, F. Strocchi, J. Phys. A Math. Theor. 40, 3173 (2007)
G. Morchio, F. Strocchi, Lett. Math. Phys. 82, 219 (2007)
G. Morchio, F. Strocchi, Ann. Phys. 324, 2236 (2009)
F. Nill, A constructive approach to abelian Chern-Simons theory, in Theory of Elementary
Particles, ed. by G. Weigt ed. (Inst. Hochenenergiephys, 1991), p. 78; Int. J. Mod. Phys. 6,
2159 (1992)
R.T. Powers, Commun. Math. Phys. 4, 145 (1967)
C.R. Putnam, Commutation Properties of Hilbert Space Operators and Related Topics (Springer,
New York, 1967)
M. Reed, B. Simon, Methods of Modern Mathematical Physics, Vol. IV (Academic Press,
New York, 1972)
M. Reed, B. Simon, Methods of Modern Mathematical Physics, Vol. I, II (Academic Press,
New York, 1972, 1975)
Bibliography 93
M. Reed, B. Simon, Methods of Modern Mathematical Physics, Vol. IV (Academic Press,

New York, 1978)
M. Requardt, Commun. Math. Phys. 50, 259 (1976)
H. Sahlmann, Loop quantum gravity-a short review, arXiv:1001.4188v1 [gr-qc]
B. Schroer, P. Stichel, Commun. Math. Phys. 3, 258 (1966)
S. Schweber, An Introduction to Relativistic Quantum Field Theory (Harper and Row, New York,
1961)
J. Slawny, Commun. Math. Phys. 24, 151 (1971)
O. Steinmann, Perturbative Quantum Electrodynamics and Axiomatic Field Theory (Springer,
New York, 2000)
F. Strocchi, Gauss law in local quantum field theory, in Field Theory, Quantization and Statistical
Physics (D. Reidel, Dordrecht, 1981)
F. Strocchi, Elements of Quantum Mechanics of Infinite Systems (World Scientific, Singapore,
1985)
F. Strocchi, Long range dynamics and spontaneous symmetry breaking in many body systems,
lectures at the Workshop on Fractals, Quasicrystals, Knots and Algebraic Quantum mechanics,
ed. by A. Amman et al. (Kluwer, Boston, 1988)
F. Strocchi, Selected Topics on the General Properties of Quantum Field Theory (World Scientific,
Singapore, 1994)
F. Strocchi, An Introduction to Non-perturbative Foundations of Quantum Field Theory. Oxford
Univ. Press 2013
F. Strocchi, An Introduction to the Mathematical Structure of Quantum Mechanics (World
Scientific, Singapore, 2005)
F. Strocchi, Symmetry Breaking, 2nd edn. (Springer, New York, 2008)
F. Strocchi, A.S. Wightman, J. Math. Phys. 15, 2198 (1974)
R.F. Streater, A.S. Wightman, P C T, Spin and Statistics and All That (Benjamin, New York, 1980)
K. Symanzik, Lectures on Lagrangian Field Theory, DESY report T-71/1
G. t Hooft, Phys. Rev. Lett. 37, 8 (1976)
T. Thiemann, Classical Quantum Gravity 23, 1923 (2006)
W. Thirring, A Course in Mathematical Physics, Vol. 3, Quantum Mechanics of Atoms and
Molecules (Springer, New York, 1981)
G. Wentzel, Quantum Theory of Fields (Dover, New York, 2004)
H. Weyl, The Theory of Groups and Quantum Mechanics (Dover, New York, 1931)
A.S. Wightman, Phys. Rev. 101, 105 (1956)
A.S. Wightman, Introduction to some aspects of the relativistic dynamics of quantized fields, in
High Energy Electromagnetic Interactions and Field Theory, Cargse lectures in Theoretical
Physics (1964), ed. by M. Lvy (Gordon and Breach, New York, 1967)
J. Zak, Phys. Rev. 168, 686 (1968)
J.M. Ziman, Principles of the Theory of Solids (Cambridge University Press, Cambridge, 1964)
Index
Algebra Delocalization, 11
of observables, 8 Diff .M/, 78
unital, 8 Diffeomorphism
A observable algebra, 20 covariance, 77
AW Weyl algebra, 3 covariant quantum mechanics, 80
AZ Zak algebra, 85 group, 78
invariance, 77
invariant states, 81
Baire -algebra, 87 Dirac-von Neumann axioms, 8
Bloch electron, 15, 27
band structure, 19
ground state, 18 Exponential field algebra, 55, 66
in a magnetic field, 29
Bloch waves, 17
Borel -algebra, 87 Fermion bosonization, 60
Borel measure, 87 Fermionic fields, 60
FW exponential field algebra, 20, 55, 62
F field algebra, 55
C -algebra, 3 Field algebra, 20, 55, 62
Canonical commutation relations (CCR), 1 observable, 62
algebra, 2 Floquet-Bloch theorem, 15
Heisenberg form, 1
Weyl form, 3
Center, 30 Gauge invariance, 11, 20
of observable algebra, 20 Gauge invariant state, 20
of observables, 12 Gauge invariant subgroup, 30
Charged fields, 60 Gauge models, 35
Chern-Simons theory Gauge transformations, 12, 21, 35, 58
abelian, 74 large, 27
Chiral anomaly, 71 Gauss invariant state, 67
Chiral symmetry, 71 Gauss law, 57, 61, 62, 65, 67
Chiral symmetry breaking, 47 Chern-Simons theory, 75
Chiral transformations, 28, 46, 71 constraint, 38, 40
Christ-Lee model, 41 operator, 38, 65
C1 .M/, 78 QCD, 64

Lecture Notes in Physics 904, DOI 10.1007/978-3-319-17695-6
96 Index
Gauss operators, 66 QCD, 16, 30, 31, 38, 64

Chern-Simons theory, 75 QCD U.1/ problem
Gauss transformations, 66 model of, 44
Gelfand spectrum, 9 QCD vacuum structure, 64
Gelfand spectrum of Zak algebra, 86 model of, 43
G gauge group, 20 Quantization
GH Heisenberg group, 12, 20 Dirac-Heisenberg, 1
GNS representation, 9 Heisenberg, 2, 3
GNS theorem, 9 Weyl, 3
Gobs observable subgroup of GH , 12, 20 Quantum chromodynamics, 16, 27
Goldstone-like state, 25 Quantum field algebra, 53
Goldstone state Quantum field theory (QFT), 55
absence, 38 Quantum Hall electron, 29, 33
Goldstone theorem, 24 Quantum mechanics on a manifold, 79
quantum mechanical, 24
Regular representation, 3, 56
Heisenberg gauge subgroup, 20 Relativistic spectral condition, 58
Heisenberg group, 2, 12 Representation
Heisenberg Lie algebra, 2 of a C - algebra, 9
Heisenberg observable group, 12 of field algebra, 55
Heisenberg observable subgroup, 20 of a Lie algebra, 3
Higgs-Kibble model, 41 of a Lie group, 3
H sector, 17, 20 physical, 20, 42, 56
unitary, regular, 3
Jackiw model, 38
Schrdinger QM, 4
Landau levels, 30 Schrdinger representation, 1
Large gauge transformations, 30, 45, Schwinger model, massive, 44
68 sectors, 20, 27, 37
Lie-Rinehart algebra, 80 Spectral condition, 56
Lie-Rinehart product, 80 Spectral measure of x, 88
Loop quantum gravity, 77 Spectral multiplicity free, 87
State, 8
pure, 9
Magnetic translations, 29 Stone-von Neumann theorem, 6
Massless scalar field, 57 Strong CP problem, 50
Multiplicative linear functional, 9 Strong measurability, 87
Superselected charges, 12, 20
Superselection rules, 8, 20, 21
Non-regular quantization, 20 Symmetry, 22
Non-regular representation, 12, 13 broken, 23
n-vacua, 46 continuous, 24
regular, 23
unbroken, 22
Observable algebra, 20 Wigner, 23
Observable variables, 2, 5 Symmetry breaking, 22, 40
order parameter, 25
spontaneous, 24
Particle on a circle, 35 Ward identity, 25
Index 97
Temporal gauge Vect (M), 78

QCD, 64 Von Neumann theorem, 4
QED, 61
Topological current, 68, 69
Topology of gauge group, 68 Weyl algebra, 3
Translationally invariant state, 13 Weyl quantization, 3, 65
non-regular, 13, 67
Weyl relations, 3, 4
U.1/ problem Winding number, 68
solution, 74
Zak algebra, 85
Vacua, 16, 28, 37, 47 Zak states, 29, 31
Vacuum state, 56 Z center of A, 12, 20

Gauge Invariance and Weyl-Polymer Quantization

Uploaded by

Copyright:

Available Formats

Gauge Invariance and Weyl-Polymer Quantization

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

Gauge Invariance and Weyl-Polymer Quantization

Uploaded by

Copyright:

Available Formats

What is the Lecture Notes in Physics series about?

What is the Lecture Notes in Physics series about?

What are the purposes of books published in the LNP series?

What are the purposes of books published in the LNP series?

Lecture Notes in Physics 904

More information about this series at http://www.springer.com/series/5304

Gauge Invariance and

ISSN 0075-8450 ISSN 1616-6361 (electronic)

Springer Cham Heidelberg New York Dordrecht London

Printed on acid-free paper

Springer International Publishing AG Switzerland is part of Springer Science+Business Media

1 Heisenberg Quantization and Weyl Quantization .. . .. . . . . . . . . . . . . . . . . . . . . 1

5 Diffeomorphism Invariance and Weyl Polymer Quantization . . . . . . . . . . . 77

The standard formulation of Quantum Mechanics (QM) is based on Heisenberg-

description of the states, namely of the representations of the algebra of observables,

serious mathematical difficulties or physical oddnesses, the quantum mechanical

1 Heisenberg and Weyl Quantizations

The standard formulation of quantum mechanics relies on the so-called canonical

hereafter called canonical commutation relations (CCR) in Heisenberg form.

.q /.x/ D x .x/; .p /.x/ D i@x .x/;

for all in a suitable dense domain D (see below).2

Springer International Publishing Switzerland 2016 1

The implicit physical assumption underlying Heisenberg quantization is that the

U./ V./ D V./ U./ei ; (1.1.3)

2 * From Heisenberg to Weyl Quantization

The quantization advocated by Weyl in his pioneering work on quantum mechanics

3 Stone-von Neumann Theorem and Schrdinger Quantum

.U./ /.x/ D eix .x/; .V./ /.x/ D .x C /; 2 L2 .Rn /: (1.3.1)

exists and defines a bounded operator. Furthermore, since

P cannot vanish. Finally, by Gaussian integrations one proves that

Therefore, P is a non-trivial projection and there exists at least a vector 2 P H

A state is pure if it cannot be written as a convex linear combination of two other

!.A/ D .! ; ! .A/ ! /: (1.3.4)

m.AB/ D m.A/ m.B/; m.A C B/ D m.A/ C m.B/:

Clearly, if m is not the trivial homomorphism, m.1/ D 1.

1 Delocalization and Gauge Invariance

As discussed in the previous chapter, the physical criterium of regularity of the

Springer International Publishing Switzerland 2016 11

and one has to consider non-regular representations (physically interesting models

H D P2 =2M C p2 =2 C V.q1 q2 /; M m1 C m2 ;  m1 m2 =M;

vectors, one should better consider non-regular representations of the Heisenberg

2 The Representation Defined by a Translationally Invariant

!0 . .A// D !0 .V./AV./ / D !0 .A/; 8A 2 AW : (2.2.1)

An interesting problem is to characterize the GNS representation 0 defined by

!0 .U./ V.// D 0; if 0; !0 .V.// D eipN ; pN 2 R: (2.2.2)

Thus, the one-parameter group U./ is non-regularly represented. The GNS

V./0 D eipN 0 ; .U./ 0 ; U. 0 / 0 / D 0; if 0 : (2.2.3)

The linear span D of the vectors U./0 , 2 R is dense in H0 , which is non-

The spectrum of V./ is a pure point spectrum.

Thus one has

!0 .U./V. // D 0; unless D 0; !0 .V. // 0: (2.2.6)

Hence, the representative of the unitary group U./, 2 R, is not weakly

T./ A 0 D .A/ 0 ; T./ 0 D 0 ; 8A 2 AW ;

T./0 D 0 ; T./ A 0 D .A/0 ; 8A 2 AW ;

preserves the scalar product as a consequence of the invariance of !0 ,

.T./A 0 ; T./ B0 / D .0 ; .A B/ 0 / D !0 . .A B// D

Hence, T./ is unitary. Furthermore, by construction T./ V./ commutes with

irreducibility of the representation (following from !0 being pure) T./ V./ D

.U./ /.x/ D eix .x/; .V./ /.x/ D .x C /:

H D P2 =2M C p2 =2 C V.q1 q2 /; M m1 C m2 ; m1 m2 =M;

!0 .U./V. // D 0; unless D 0; !0 .V. // 0: (2.2.6)

The spectrum of p in H is .p/jH D f2n=a C =a; n 2 Zg and the wave

.U./V.// D .V. /U./ V./V. // D ei .U./ V.//;

H D P2 =2M C p2 =2 C V.q/ (2.4.3)

lim < Qn .t/; F >D lim < t .Qn /; F >D

D lim < Qn ; F > :

2 Im lim < Qn U.t/F >D i lim < Qn ; F > 0: (2.4.8)

x D px y=2; y D py C x=2; y ; x D i: (2.5.2)

T.aj / eic aj ; j D 1; 2; c p C eA=c; (2.5.3)

u./ ei q ; v./ ei p I U. / ei Q ; V./ eiP :