In this work we will discuss a subject that is still of extreme importance in today’s theoretical physics. The issue is gauge invariance, one of the main ingredients of Standard Model. One path to explore the subject is through the constraint analysis, since it is well known that first-class constraints system, in Dirac’s constraints classification,1 is a gauge invariant one. Hence, when we have a second-class constrained system, the gauge invariance is lost and it is very convenient to recover this gauge invariance. Thus, we have to convert this second-class into a first−class one. The literature offers several ways to do that. In this work we will accomplish the task through the so−called Gauge Unfixing (GU) conversion technique.
As an example of what we have just said about constraints, let us consider the Abelian pure Chern-Simons (CS) theory, which is a mixed constrained system, where one of its four constraints must be, let us say, “converted,” in order to be a first−class one. Then, after this step, we would have well established algebras of two first-class constraints and two second−class constraints.
One of the “conversion” methods, the well known BFT formalism,2–11 which enlarges the phase space variables through the introduction of the so−called Wess Zumino (WZ) fields, has been used since its first design, to embed the CS theory.12 As a result, the authors demonstrate many important features.
In another paper,13 the authors have also employed the BFT formalism to analyze a non−Abelian version of the CS theory. In this paper, the authors suggest two methods that overcome the problem of embedding mixed constrained systems. In an alternative view of the BFT formalism, there is the GU method which embeds second-class constrained systems. It was introduced by Mitra & Rajaraman14,15 and developed by Vytheeswaran.16,17
The GU formalism considers part of the whole group of second−class constraints as being the gauge symmetry generators. The remaining constraints are now the gauge fixing terms. The corresponding second−class Hamiltonian must be adapted, i.e., modified, in such a convenient way in order to satisfy the first−class algebra together with the constraints that were chosen at the beginning as being the gauge symmetry generators. The GU method has a classy property which prevents us from extending the phase space with extra variables. Some time ago, one of us has provided the constraint literature with an alternative procedure concerning the GU formalism and applied it to the CS theory.18 The objective was to redefine the original phase space variables of a certain constrained system, without introducing any WZ terms, in order to be gauge invariant fields. After that, functions of these gauge invariant fields, which will be gauge invariant quantities, were constructed. This so called “extended” GU formalism begins with a kind of mixed constrained system, which was the CS theory, at that occasion, and, applying the technique, it was obtained a first−class system which was written just in terms of the original phase space variables with many new features. As many important constrained systems have only two second−class constraints, then, in principle, the formalism was introduced only for systems with two second−class constrains without any loss of generality.
In this work we will use this extended GU constraint conversion method to explore the O(N) nonlinear sigma model (NLSM). We have some experience with the NLSM19 and the motivation to work with it is based on the fact that, although the NLSM were first introduced in high energy physics in the context of chiral symmetry breaking, NLSM also plays an underlying role in condensed matter issues, where it appears naturally as effective field theories depicting the low energy long−wave−length limit of several microscopic models.
Having said that and, in order to clarify the exposition of the subject, this paper is organized as follows: in Section 2, we give a short review of the usual GU formalism. In Section 3, we present our formalism. In Section 4, we apply our procedure to the CS theory. In Section 5, we make our concluding remarks.
Let us study a second−class constrained system described by its correspondent Hamiltonian which has, for example, two second-class constraints T1
andT2
. The main idea of GU formalism is to convert a second−class system into a first−class one by selecting one of the two second−class constraints to be the gauge symmetry generator, i.e., this constraint will be “defined” ad hoc as being first−class. The other constraint will be discarded since a new first−class Hamiltonian will be constructed. However, since we have two constraints, the next step is to build another conversion procedure with the second constraint that was discarded. Now this second constraint will be the chosen one, and the first constraint will be discarded. To sum up, we have two cases in this GU formalism, namely, two ways to obtain gauge invariance. This will be clear in a moment.
The idea is to understand the original non invariant gauge theory as being a gauge fixed version of the gauge invariant system. If we choose T1
as the symmetries generator, then the second−class Hamiltonian have to be modified in order to satisfy a first−class algebra. To accomplish the task, both the new and gauge invariant Hamiltonian can be constructed through a power series of T2
in order to not generate any new constraints, of course. Hence, with this procedure obligation well established, we can write conveniently that
˜H=H+T2{H,T1}+12!T22{{H,T1},T1}+13!T32{{{H,T1},T1},T1}+...,
(1)
where it can be shown that
{˜H,T1}=0
(i.e., there are no secondary or any new constraints) and
T1
must satisfy the first-class algebra
{T1,T1}=0
. In this way this final system was shown precisely to be a first−class one, and consequently, gauge invariant.
The O(N) nonlinear sigma model is described by the Lagrangian density
L=12∂μ∅a∂μ∅a+12λ(∅a∅a−1) (2)
where the
μ=0, 1
and a is an index related to the O(N) symmetry group, and the corresponding canonical Hamiltonian density is given by
H=12πaπa−12∂i∅a∂i∅a−12λ(∅a∅a−1) (3)
The second−class constraints of the system in Eq. (1), in Dirac’s constraints classification, are
T1(x)= ∅a(x)∅a(x)− 1
and T2(x)= ∅a(x)πa(x)
(4)
The second−class constraint algebra is
{T1(x), T2(y)}= 2 ∅a ∅a δ(x−y )
(5)
which shows clearly that the Poisson brackets of both constraints is not zero and consequently the system is not gauge invariant.
The infinitesimal gauge transformations generated by symmetry generator
T1
are
δ ∅a= ∈ { ∅a(x), T1(y) }=0 (6)
and
δ πa= ε {πa(x), T1(y)}= −2 ε ∅a δ (x−y)
(7)
The gauge invariant field
∅a
will now be constructed by using a known Taylor expansion in series of powers of
T2
, namely,
˜∅a= ∅a+ ba1 T2+ ba2T22+ ⋯+ba2Tn2
(8)
From the invariance condition,
δ˜∅a=0
, we can calculate all the set of correction terms
bn
. For the linear correction term in order of
T2
, we have that
δ ˜∅a=0 ⇒ ∈ { ˜∅a(x), T1(y)}= 0
(9)
Substituting Eq. (8) into the last equation, and using the algebra in Eq. (5) to equate the
ban
terms, we can easily arrive the result that shows that
ba1= ba2= ⋯= ban= 0
(10)
Namely, due to this last result, all the correction terms
ban
where
n≥1
are zero. Therefore, the gauge invariant field
˜∅a
is
˜∅a=∅a (11)
And, by using Eq. (6), it is readily to show thatδ˜∅a=0
. The gauge invariant field ˜πa
is also constructed via Taylor series in powers of T2
˜πa= πa+ ca1T2+ ca2T22+ ⋯+canTn2
(12)
From the invariance condition
δ˜πa=0
, we can work out all the correction terms
can
. For the linear correction term in order of
T2
, we have that
δπa+ ca1 ∅b δπb=0
(13)
and consequently,
ca1= − ∅a∅2
(14)
For the quadratic term, we can write that
ca2=0
, since
δca1=ε{ca1,˜T}=0
. Thereby, using this result, we can say that all the correction terms
can
with
n≥2
is zero. Hence, the gauge invariant field
˜πa
is
˜πa=πa−∅a∅bπb∅2
(15)
where, by using Eq. (7), It is direct to demonstrate that
δ˜πa=0
The gauge invariant Hamiltonian, written only in terms of the original phase space variables, is obtained by substituting
∅a
by
˜∅a
, and
πa
by
˜πa
, Eqs. (11) and (15), respectively, into the canonical Hamiltonian, Eq.(3), as follows
˜H=12 πaπa−12 (∅π)2∅2 − 12 ∂i∅a∂i∅a
(16)
From Eqs. (2) and (3) we have that
δ˜Hcδπb=˙∅b
(17)
And substituting this result into Eq. (15), we can write that
πb−∅aπa∅2∅b=˙∅b
(18)
Multiplying this last equation by
∅b
, we can see that
∅a˙∅a=0
(19)
which is not a constraint considering the new Hamiltonian. Hence, using this constraint to construct the new Lagrangian we know that
˜L=˜π˙˜∅−˜H
(20)
Using Esq. (16) and (19) we obtain that
˜L=12∂μ∅a∂μ∅a
(21)
And the gauge invariant condition is
δ∅a=0
(22)
Now, let us follow the extended GU technique in order to consider the infinitesimal gauge transformations generated by symmetry generator
T2
given by Eq. (4), i.e.,
T2=∅a(x)πa(x)
. The gauge transformations are
δ∅a(x)=ε{∅a(x),T2(y)}=ε∅a(y) δ(x−y)
(23)
The gauge transformations are
δ∅a(x)=ε{∅a(x),T2(y)}=ε∅a(y) δ(x−y)
and
δπa(x)=ε{πa(x),T2(y)}=−επa(y)δ(x−y)
(24)
In order to construct gauge invariant fields, we can write as
˜∅a=∅a(1−12∅2−1∅2−18(∅2−1)2∅4+⋯)
(25)
Hence, this last result in Eq. (25) suggest that we can write the last equation in the following convenient form,
˜∅a=f(∅2)∅a
(26)
From invariance condition
δ˜∅a(x)=0
we can obtain
f(∅2)
, which is
f(∅2)=1√∅2
(27)
Substituting it in Eq. (26) we have that
˜∅a=1√∅2∅a
(28)
By making the same procedure for
˜πa
we have that
˜πa=πa(1+12∅2−1∅2+38(∅2−1)2∅4+⋯)
(29)
which, analogously, suggests that
˜πa
can be written such as
˜πa=g(∅2)πa (30)
from the invariance condition
δ˜πa=0
we can obtain
g(∅2)
g(∅2a)=√∅2
(31)
which, substituting into Eq. (6), we have that
˜πa=√∅2 πa
(32)
Therefore, the gauge invariant Hamiltonian is written as
˜H=12˜πa˜πa−12∂i˜∅a∂i˜∅a
(33)
where ˜∅a and ˜πa
are given by Eqs. (5) and (8) respectively.