u.ac.jp a- y ys.nago en.ph ura@ek View metadata,citationandsimilarpapersatcore.ac.uk
HEP-TH-9412174 e-mail address : tanim b er 1994 y an DPNU-94-60 When Decem hep-th/9412174 tization wn that wn that the b er. A new provided byCERNDocumentServer tations as a utation rela- tize the mo del um brought toyouby hes the structure of o dimensional space- t represen tral extension. w wn that the zero-mo de alen CORE . The algebra generated b b er and quan tum op erators. It is sho y the cen ta op erators ob ey anomalous com- animura um T Abstract ariables ob ey a new comm ted out that its top ological structure is b er of inequiv tro duced, it is sho trivial top ology um es momen Shogo wist relation. In addition, it is sho h is a eld-theoretical mo del p ossessing a non- (1) sigma mo del in the t U . It is p oin y the zero-mo de and the winding n utation relations is prop osed to quan tation spaces of quan e call t , whic e demonstrate that top ology enric R hw tral extension is in 1 S tum eld theories. tum op erators is deformed b artment of Physics, Nagoya University, Nagoya 464-01, Jap tral extension mak e consider the utators. W ariables and the winding v haracterized b yp e of comm tion, whic cen m quan v the cen quan consequence of the non there are an in nite n resp ecting the top ological nature.to b e Hilb ert represen spaces are constructed t time trivial top ology c W Dep of ab elian sigma mo del in (1+1) dimensions Zero-mo de, winding n
1 Intro duction
As everyone recognizes, quantum theory is established as an adequate and unique
language to describ e microscopic phenomena including condensed matter physics and
also particle physics. In this pap er wewould like to develop quantum theory in the
direction to seek for another p ossibility which inheres in the theory itself. First we
shall explain the well-known feature of quantum theory and then we shall show the
direction to pursue.
To formulate a quantum theory we start from commutation relations among
generators. Generators de ne an algebra, which is asso ciative but not commutative.
When wehave a classical theory,we ordinarily bring commutators among canonical
variables by replacing the Poisson brackets. Then we construct a Hilb ert space to
b e an irreducible representation space of the algebra. An element of the algebra is
chosen and called a hamiltonian. Choice of a hamiltonian is restricted byphysical
requirements like self-adjointness, p ositivity and symmetries. The set formed by the
algebra, the representation space and the hamiltonian is called a quantum theory.
Then formulation itself is nished.
Next tasks are to solve it; wewanttoknow eigenvalues of various observables,
esp ecially we are interested in sp ectrum of the hamiltonian, wewant to know proba-
bility amplitude of transition b etween an initial state and a nal one. For physically
interesting theories, those are dicult tasks. Of course, we appreciate that they are
worth hard work. However it is not the direction in whichwewould like to pro ceed
in this pap er. We are interested in a rather formal asp ect.
A feature of quantum theory is that it requires a representation space; op erators
must b e provided with op erands. A commutator in quantum theory corresp onds to
aPoisson bracket in classical theory. But the Hilb ert space in quantum theory has
no corresp ondence in classical theory. State vectors, the sup erp osition principle and
amplitudes are characteristic concepts of quantum theory.We b egin with an algebra
of op erators, then we construct a representation space. That is a unique pro cedure
of quantum theory.
Here arise two questions; do es a representation space exist? Is it unique? If
there are inequivalent representations, calculation based on a di erent space gives a
di erent answer for a physical quantity, for example, sp ectrum or amplitude.
When we consider a particle in a Euclidean space, we b egin with the usual canon-
ical commutation relations;
[^x ;x^ ]= 0; (i; j =1;;n) (1.1)
i j
[^x ;p^ ]=i ; (1.2)
i j ij
[^p ;p^ ]=0: (1.3)
i j
According to von Neumann's theorem the irreducible representation of the ab ove
algebra exists uniquely within a unitary equivalence class. Therefore there is no
problem in choice of a Hilb ert space. Although one may use the wave function
representation and another may use the harmonic oscillator representation, b oth
obtain a same result for calculation of a physical quantity.
Is there no need to worry ab out existence and uniqueness of a representation?
Actually it is needed. Wehave encountered a situation in which the uniqueness is
violated, when we consider a quantum eld theory. In a quantum eld theory we
construct a representation space by de ning a vacuum state and a Fock space. It was
found that in several mo dels there exist inequivalentvacuum states and they result in
inequivalentFock spaces. The di erentvacua are characterized by its transformation
prop erty under a certain symmetry op eration. Such a situation is called sp ontaneous
symmetry breaking (SSB). The discovery of SSB op ened rich asp ects of quantum eld
theories and led to deep understanding of the nature.
A eld theory deals with a system which has in nite degrees of freedom, namely
it is de ned with a in nite numb er of generators which are called eld variables. It
is known that SSB is related to the in nity of degrees of freedom. On the other hand
a particle has only nite degrees of freedom. Is there no o ccurence of inequivalent
representations in a quantum theory of a particle? (Usually a quantum theory with
nite degrees of freedom is called a quantum mechanics.)
A strange result was found; when Ohnuki and Kitakado [1] investigated a quan-
1
tum mechanics of a particle on a circle S , they showed that there are a in nite
numb er of inequivalent Hilb ert spaces. Those spaces are parametrized by a continu-
ous parameter ranging from 0 to 1. What they have shown is that even a system
with nite degrees of freedom can p ossess inequivalent representations when top ol-
ogy of the system is nontrivial. After that work, they studied a quantum mechanics
n
on a sphere S (n 2) and showed existence of an in nite numb er of inequivalent
Hilb ert spaces sp eci ed by a discrete index.
Let us turn to eld theories. The scalar eld theory is a eld-theoretical cor-
resp ondence to the quantum mechanics of a particle in a Euclidean space. This
eld theory is quantized by requiring the canonical commutators and constructing
the Fock space. There is also a corresp ondence in eld theories to a quantum me-
chanics on a nontrivial manifold. It is a nonlinear sigma mo del b ecause it has a
manifold-valued eld. For a review on nonlinear sigma mo dels, see the reference [2].
Originally the nonlinear sigma mo del is designed to describ e b ehavior of Nambu-
Goldstone (NG) b osons at low energy scale. NG b osons are massless excitations
asso ciated with SSB. When a continuous symmetry sp eci ed by a group G is broken
to a smaller symmetry sp eci ed by a subgroup H ,vacua form a manifold whichis
called a homogeneous space G=H . In this mo del NG b osons are describ ed by a eld
taking values in G=H . It is already known [3] that even the quantum mechanics
on G=H has inequivalent Hilb ert spaces. Therefore it is naturally exp ected that a
quantum eld theory with a manifold-valued eld may p ossess inequivalent Hilb ert
spaces. However in a usual approach, the nonlinear sigma mo del is quantized by the
canonical quantization as the scalar eld theory and is solved by the p erturbative
metho d. Thus the top ological nature of the theory is missed and only a Fock space
provides a representation.
If top ological prop erties of eld theories do not play an imp ortant role in physical
application, we could neglect them. However we know several mo dels whichhave
nontrivial top ology and in which top ology plays an imp ortant role. For instance,
the sine-Gordon mo del has top ological kinks [4]; some gauge theories have top olog-
ically nontrivial vacua, so-called -vacua [5]; some nonlinear sigma mo dels have the
Wess-Zumino-Witten term, which re ects anomaly and top ology of vacua [6]; the
con guration space of nonab elian gauge elds mo dulo gauge transformations has an
extremely complicated top ology and causes the Grib ov problem [7], and so on. Hence
wewould like to develop quantum eld theories resp ecting top ological nature.
One of the aims of this pap er is to demonstrate that there are a lot of p ossibilities
in constructing of quantum eld theories, even if they are identical as classical theo-
ries. The second aim is to clarify the relation b etween top ology and quantization. In
1
the section 2 we will give a review of the quantum mechanics on S .We will discuss
physical implication of the existence of inequivalent representations. The section
3 is a main part of the present pap er. There we consider a simple but nontrivial
eld-theoretical mo del, the ab elian sigma mo del in (1+1) dimensions. We prop ose
a de nition of an algebra, which is quite di erent from the canonical one. Then we
will construct Hilb ert spaces and classify inequivalent ones. The section 4 is devoted
to discussion on the results and directions for future development. This pap er is a
detailed and extended sequel to the previous work [8].
1
2 Quantum mechanics on S
1
Here we shall give a review of the quantum mechanics on S [1]. Idea in it is useful
for the next step in the eld theory. More detailed consideration is found in the
reference [9].
2.1 Motivation
1
Let us consider a particle moving on a circle S . Its p osition is indicated by the angle
co ordinate . The co ordinate +2 indicates the same p ointas do es. Hence
1
is multivalued function on S . Because of its nontrivial top ology, a continuous and
1
single-valued co ordinate do es not exist on S .
Now let us quantize it. We should de ne an algebra. If we take the canonical
^
^
approach, assume that and P are self-adjoint op erators satisfying
^ ^
[ ; P ]=i: (2.1)
^
Next we should construct a representation space. If we de ne eigenstates of by
^
j i = j i; (2.2)
we deduce that
@
^
h jP j i = i hj i (2.3)
@
^
for an arbitrary state j i. Notice that the eigenvalue of ranges over from 1 to
+1;itisnot true that j +2i = ji. What wehave constructed is just the quantum
1
mechanics on R , not S .
^
What is wrong? It is wrong to use the multivalued co ordinate as a generator of
the algebra. Wemust use a well-de ned single-valued generator from the b eginning.
i
As a substitute for , Ohnuki and Kitakado prop osed to use U = e , whichis
complex-valued in the classical theory and a unitary op erator in the quantum theory.
From (2.1) we can deduce
^ ^
i i
^
[ e ; P ]= e : (2.4)
^
But they did not regard as a generator. Instead they to ok existence of a unitary
^ ^
op erator U and a self-adjoint op erator P satisfying
^ ^ ^
[ U; P ]= U (2.5) as an assumption.
2.2 Representations
Nowwe see their construction of representations of the algebra de ned by the ab ove
^
commutation relation (2.5). Since P is a self-adjoint op erator, it has an eigenvector
with a real eigenvalue ;
^
P j i = j i; h j i =1: (2.6)
^ ^
U raises the eigenvalues of P ;
^ ^ ^ ^ ^ ^
P U j i = ([P;U ]+UP)j i
^ ^
= (U +U ) j i
^
= (1 + )U j i: (2.7)