I. Cotaescu - "Canonical quantization of the covariant fields: the Dirac field on the de Sitter spacetime"

Canonical quantization of the covariant fields:
the Dirac field on the de Sitter spacetime
Ion I. Cotaescu
Abstract
The properties of the covariant fields on the de Sitter spacetimes are
investigated focusing on the isometry generators and Casimir operators in
order to establish the equivalence among the covariant representations (CR)
and the unitary irreducuble ones (UIR) of the de Sitter isometry group. For
the Dirac field it is shown that the spinor CR transforming the Dirac field under
de Sitter isometries is equivalent to a direct sum of two UIRs of the Sp(2, 2)
group transforming alike the particle and antiparticle field operators in momentum
representation (rep.). Their basis generators and Casimir operators are written
down finding that these reps. are equivalent to a UIR from the principal series
whose canonical labels are determined by the fermion mass and spin.
Pacs: 04.20.Cv, 04.62.+v, 11.30.-j
arXiv:1602.06810
Keywords: Sitter isometries; Dirac fermions; covariant rep.; unitary rep.; basis generators;
Casimir operators; conserved observables; canonical quantization.
1

Contents
1. Introduction 4
2. Covariant quantum fields 9
Natural and local frames . . . . . . . . . . . . . . . . . . . . . . . . . . 9
Covariant fields on curved spacetimes . . . . . . . . . . . . . . . . . 11
Lagrangian formalism . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17
Canonical quantization . . . . . . . . . . . . . . . . . . . . . . . . . . . 21
3. Covariant fields in special relativity 24
Generators of manifest CRs . . . . . . . . . . . . . . . . . . . . . . . . 24
Wigner’s induced UIRs . . . . . . . . . . . . . . . . . . . . . . . . . . . 28
2

4. Covariant ﬁelds on de Sitter spacetime 34
de Sitter isometries and Killing vectors . . . . . . . . . . . . . . . . . 34
Generators of induced CRs . . . . . . . . . . . . . . . . . . . . . . . . 41
Casimir operators . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43
5. The Dirac ﬁeld on de Sitter spacetimes 48
Invariants of the spinor CR . . . . . . . . . . . . . . . . . . . . . . . . 48
Invariants of UIRs in momentum rep. . . . . . . . . . . . . . . . . . . 53
CR-UIR equivalence . . . . . . . . . . . . . . . . . . . . . . . . . . . . 58
6. Concluding remarks 61
Appendix 63
A: Finite-dimensional reps. of the sl(2, C) algebra . . . . . . . . . . . . . . . . 63
B: Some properties of Hankel functions . . . . . . . . . . . . . . . . . . . . . . 65
3

1. Introduction
Our world is governed by four fundamental interactions that can be investigated at two
levels:
- the classical level of continuous matter neutral or electrically charged,
- the quantum level of interacting quantum ﬁelds whose quanta are the elementary
particles, fermions of spin 1/2 and gauge bosons of spin 1.
The actual physical evidences (up to the scale of the LHC energies ∼ 10−20
m) give
the following picture
CLASSICAL QUANTUM
Gravity NO DATA
Electromagnetism Electromagnetism
NO DATA Nuclear weak
NO DATA Nuclear strong
At this scale we can adopt the semi-classical picture of quantum ﬁelds in the
presence of the classical gravity of curved spacetimes without torsion.
The principal measured quantities are the conserved ones corresponding to symmetries
(via Noether theorem). These are the mass, the spin and different charges.
4

Figure 1: Symmetries determining conserved quantities
5

It is known that in special relativity the isometries play a crucial role in quantizing free
fields since the principal particle properties, the mass and spin, are eigenvalues of the
Casimir operators of the Poincaré group. Then it is natural to ask what happens in
the case of the curved spacetimes - how the mass and spin can be defined by the
isometry invariants. The answer could be found taking into account that:
1. The theory of quantum fields with spin on curved (1 + 3)-dimensional local-
Minkowskian manifolds (M, g), having the Minkowski flat model (M0, η), can be
correctly constructed only in orthogonal (non-holonomic) local frames [1, 2].
2. The entire theory must be invariant under the orthogonal transformations of the local
frames, i. e. gauge transformations that form the gauge group G(η) = SO(1, 3)
which is the isometry group of the flat model M0.
3. Since the transformations of the isometry group I(M) can change this gauge
we proposed to enlarge the concept of isometry considering external symmetry
transformations that preserve not only the metric but the gauge too [3, 4]. The group
of external symmetry S(M) is isomorphic to the universal covering group of the
isometry group I(M).
6

4. The quantum fields transform according to the covariant reps. (CRs) of the group
S(M) which are induced by the (non-unitary) finite-dimensional reps. of the group
Spin(1, 3) ∼ SL(2, C), i. e. the universal covering group of the gauge group
G(η) = L↑
+ ⊂ SO(1, 3) [3]-[5].
5. The conserved observables are the generators of the CRs, i. e. the differential
operators produced by the Killing vectors (associated to isometries) according to the
generalized Carter and McLenagan formula [25, 3, 4].
The main purpose of the present talk is to present the general theory of the CRs
pointing out the role of their generators in canonical quantization.
The examples we give are the well-known case of special relativity as well as the CRs
on the (1+3)-dimensional de Sitter spacetime where the specific SO(1, 4) isometries
generate conserved observables with a well-defined physical meaning [20] that allow us
to perform the canonical quantization just as in special relativity.
We must stress that in the case of the de Sitter spacetime our theory of induced CRs,
we use here, is equivalent to that proposed by Nachtmann [16] many years ago, and is
completely different from other approaches [13]-[15] that are using linear reps. of the
universal covering group of the de Sitter isometry group [9, 10].
7

Figure 2: The external symmetry combines the isometries with suitable gauge
transformations preserving thus the metric and the relative positions of the local frames
with respect to the natural ones.
8

2. Covariant quantum fields
Natural and local frames
Let (M, g) be a (1+3)-dimensional local-Minkowskian spacetime equipped with local
frames {x; e} formed by a local chart (or natural frame) {x} and a non-holonomic
orthogonal frame {e}.
The coordinates xµ
of the local chart are labelled by natural indices µ, ν, ... =
0, 1, 2, 3. The orthogonal frames are defined by the vector fields,
eˆα = eµ
ˆα∂µ , (1)
while the corresponding coframes are defined by the 1-forms
ωˆα
= êˆα
µdxµ
. (2)
These are called tetrad fields, are labelled by local indices, ˆα, ...ˆµ, ˆν, ... = 0, 1, 2, 3
and obey the usual duality relations
êˆµ
α eα
ˆν = δˆµ
ˆν , êˆµ
α eβ
ˆµ = δβ
α , (3)
9

and the orthonormalization conditions,
eˆµ · eˆν
def
= gαβeα
ˆµeβ
ˆν = ηˆµˆν , ωˆµ
· ωˆν def
= gαβ
êˆµ
αêˆν
β = ηˆµˆν
, (4)
where η = diag(1, −1, −1, −1) is the metric of the Minkowski model (M0, η) of
(M, g). Then the line element can be written as
ds2
= ηˆα ˆβωˆα
ω
ˆβ
= gµνdxµ
dxν
. (5)
which means that gµν = ηˆα ˆβ êˆα
µê
ˆβ
ν . This is the metric tensor which raises or lowers the
natural indices while for the local ones we have to use the flat metric η.
The vector fields eˆν satisfy the commutation rules
[eˆµ, eˆν] = eα
ˆµeβ
ˆν (êˆσ
α,β − êˆσ
β,α)ˆ∂ˆσ = C ··ˆσ
ˆµˆν·
ˆ∂ˆσ (6)
defining the Cartan coefficients which help us to write the conecttion coefficients in
local frames as
ˆΓˆσ
ˆµˆν = eα
ˆµeβ
ˆν (êˆσ
γΓγ
αβ − êˆσ
β,α) =
1
2
ηˆσˆλ
(Cˆµˆνˆλ + Cˆλˆµˆν + Cˆλˆν ˆµ) . (7)
We specify that this connection is often called spin connection (and denoted by Ωˆσ
ˆµˆν) but
it is the same as the natural one. The notation Γ stands for the usual Christoffel symbols
representing the connection coefficients in natural frames.
10

Covariant ﬁelds on curved spacetimes
The metric η remains invariant under the transformations of the group O(1, 3) which
includes as subgroup the gauge group G(η) = L↑
+, whose universal covering group
is Spin(1, 3) = SL(2, C). In the usual covariant parametrization, with the real
parameters, ˜ωˆα ˆβ
= −˜ω
ˆβ ˆα
, the transformations
A(˜ω) = exp −
i
2
˜ωˆα ˆβ
Sˆα ˆβ ∈ SL(2, C) (8)
depend on the covariant basis-generators Sˆα ˆβ of the sl(2, C) Lie algebra which satisfy
[Sˆµˆν, Sˆσˆτ] = i(ηˆµˆτ Sˆνˆσ − ηˆµˆσ Sˆνˆτ + ηˆνˆσ Sˆµˆτ − ηˆνˆτ Sˆµˆσ) . (9)
The matrix elements in local frames of the SO(1, 3) transformation associated to A(˜ω)
through the canonical homomorphism can be expanded as
Λˆµ ·
· ˆν(˜ω) = δˆµ
ˆν + ˜ωˆµ ·
· ˆν + · · · ∈ SO(1, 3) (10)
We denote by I = A(0) ∈ SL(2, C) and 1 = Λ(0) ∈ SO(1, 3) the identity
transformations of these groups.
11

The covariant fields, ψ(ρ) : M → V(ρ), are locally defined over M with values in the
vector spaces V(ρ) carrying the finite-dimensional non-unitary reps. ρ of the group
SL(2, C) (briefly presented in the Appendix A). In general, these representations are
reducible as arbitray sums of irreducible ones, (j1, j2). For example, the vector field
transforms according to the irreducible rep. ρv = (1/2, 1/2) while for the Dirac field we
use the reducible rep. ρs = (1/2, 0) ⊕ (0, 1/2).
The covariant derivatives of the field ψ(ρ) in local frames (or natural ones),
D
(ρ)
ˆα = eµ
ˆαD(ρ)
µ = eµ
ˆα∂µ +
i
2
ρ(S
ˆβ ·
· ˆγ ) ˆΓˆγ
ˆα ˆβ
, (11)
assure the covariance of the whole theory under the (point-dependent) gauge
transformations,
ω(x) → Λ[A(x)]ω(x) (12)
ψ(ρ)(x) → ρ[A(x)]ψ(ρ)(x), (13)
produced by the sections A : M → SL(2, C) of the spin fiber bundle.
Note that in the case of the vector and tensor fields the local derivatives coincide with the
covariant ones acting as DµT···ν···
= ∂µT···ν···
· · · + Γν
µαT···α···
+ · · · . This means
that the local frames are needful only in the case of the field with half integer spin.
12

(M, g) may have isometries, x → x = φg(x), given by the (non-linear) rep. g → φg
of the isometry group I(M) with the composition rule φg◦φg = φgg , ∀g, g ∈ I(M).
Then we denote by id = φe the identity function, corresponding to the unit e ∈ I(M),
and deduce φ−1
g = φg−1. In a given parametrization, g = g(ξ) (with e = g(0)), the
isometries
x → x = φg(ξ)(x) = x + ξa
ka(x) + ... (14)
lay out the Killing vectors ka = ∂ξa
φg(ξ)|ξ=0 associated to the parameters ξa
(a, b, ... = 1, 2...N).
The isometries may change the relative position of the local frames. For this reason we
proposed the theory of external symmetry [3] where the combined transformations
(Ag, φg) are able to correct the position of the local frames preserving thus not only
the metric but the gauge too, i. e.
⇒ ω(x) → ω (x ) def
= ω[φg(x)] = Λ[Ag(x)]ω(x) . (15)
Hereby, we deduce [3],
Λˆα ·
· ˆβ
[Ag(x)] = ˆeˆα
µ[φg(x)]
∂φµ
g(x)
∂xν
eν
ˆβ
(x) , (16)
13

assuming, in addition, that Ag=e(x) = 1. We obtain thus the desired transformation
laws under isometries,
(Ag, φg) :
e(x) → e (x ) = e[φg(x)] ,
ψ(ρ)(x) → ψ(ρ)(x ) = ρ[Ag(x)]ψ(ρ)(x) . (17)
that preserve the gauge.
The set of combined transformations (Ag, φg) form the group of external symmetry,
denoted by S(M). This is isomorphic with the universal covering group of I(M). The
multiplication rule is defined as
(Ag , φg ) ∗ (Ag, φg) def
= ((Ag ◦ φg) × Ag, φg ◦ φg) = (Ag g, φg g) , (18)
such that the unit element is (Ae, φe) = (I, id) while the inverse of the element
(Ag, φg) reads (Ag, φg)−1
= (Ag−1 ◦ φg−1, φg−1). For other mathematical details
see Ref. [3].
In a given parametrization, g = g(ξ), for small values of ξa
, the SL(2, C) parameters
of Ag(ξ)(x) ≡ A[˜ωξ(x)] can be expanded as ˜ωˆα ˆβ
ξ (x) = ξa
Ωˆα ˆβ
a (x) + · · · where
Ωˆα ˆβ
a ≡
∂˜ωˆα ˆβ
ξ
∂ξa
|ξ=0
= êˆα
µ kµ
a,ν + êˆα
ν,µkµ
a eν
ˆλ
η
ˆλ ˆβ
(19)
14

are skew-symmetric functions, Ωˆα ˆβ
a = −Ω
ˆβ ˆα
a , only when ka are Killing vectors [3].
The last of Eqs. (17) defines the CRs induced by the finite-dimensional rep., ρ, of the
group SL(2, C). These are operator-valued reps., T(ρ)
: (Ag, φg) → T
(ρ)
g , of the
group S(M) whose covariant transformations,
⇒ (T(ρ)
g ψ(ρ))[φg(x)] = ρ[Ag(x)]ψ(ρ)(x) , (20)
leave the field equation invariant since their basis-generators [3],
⇒ X(ρ)
a = i∂ξaT
(ρ)
g(ξ)|ξ=0
= −ikµ
a ∂µ +
1
2
Ωˆα ˆβ
a ρ(Sˆα ˆβ) , (21)
commute with the operator of the field equation. These satisfy the commutation rules
[X(ρ)
a , X
(ρ)
b ] = icabcX(ρ)
c (22)
where cabc are the structure constants of the algebras s(M) ∼ i(M) - they are
the basis-generators of a CR of the s(M) algebra induced by the rep. ρ of the
spin(1, 3) = sl(2, C) algebra.
15

These generators can be put in (general relativistic) covariant form either in non-
holonomic frames [3],
⇒ X(ρ)
a = −ikµ
a D(ρ)
µ +
1
2
ka µ;ν eµ
ˆα eν
ˆβ
ρ(Sˆα ˆβ
) , (23)
or even in holonomic ones [4], generalizing thus the formula given by Carter and
McLenaghan for the Dirac ﬁeld [25].
The generators (21) have, in general, point-dependent spin terms which do not commute
with the orbital parts. However, there are tetrad-gauges in which at least the generators
of a subgroup H ⊂ I(M) may have point-independent spin terms commuting with
the orbital parts. Then we say that the restriction to H of the CR T(ρ)
is manifest
covariant [3].
Obviously, if H = I(M) then the whole rep. T(ρ)
is manifest covariant. In particular,
the linear CRs on the Minkowski spacetime have this property. This gives rise to the so
called Lorentz covariance which, according to our theory, is universal for any (1 + 3)-
dimensional local-Minkowskian manifold.
16

Lagrangian formalism
In the Lagrangian theory the Lagrangian densities must be invariant positively defined
quantities. Since the finite-dimensional reps. ρ of the SL(2, C) group are non-unitary,
we must use the (generalized) Dirac conjugation, ψ(ρ) = ψ+
ρ γ(ρ), where the matrix
γ(ρ) = γ+
(ρ) = γ−1
(ρ) satisfies ρ(A) = γ(ρ)ρ(A)+
γ(ρ) = ρ(A−1
). Then the form
ψ(ρ)ψ(ρ) is invariant under the gauge transformations. In general, the Dirac conjugation
can be defined for the reducible reps. of the form ρ = ...(j1, j2) ⊕ (j2, j1)....
The covariant equations of the free fields can be derived from actions of the form
S[ψ(ρ), ψ(ρ)] =
∆
d4
x
√
g L(ψ(ρ), ψ(ρ);µ, ψ(ρ), ψ(ρ);µ) , g = |det gµν| , (24)
depending on the field ψ(ρ), its Dirac adjoint ψ(ρ) and their corresponding covariant
derivatives ψ(ρ);µ = D
(ρ)
µ ψ(ρ) and ψ(ρ);µ = D
(ρ)
µ ψ(ρ) defined by the rep. ρ of the
group SL(2, C).
17

The action S is extremal if the covariant ﬁelds satisfy the Euler-Lagrange equations
∂L
∂ψ(ρ)
−
1
√
g
∂µ
∂(
√
g L)
∂ψ(ρ),µ
= 0 ,
∂L
∂ψ(ρ)
−
1
√
g
∂µ
∂(
√
g L)
∂ψ(ρ),µ
= 0 . (25)
Any transformation ψ(ρ) → ψ(ρ) = ψ(ρ) + δψ(ρ) leaving the action invariant,
S[ψ(ρ), ψ(ρ)] = S[ψ(ρ), ψ(ρ)], is a symmetry transformation. The Noether theorem
shows that each symmetry transformation gives rise to the current
Θµ
∝ δψ(ρ)
∂L
∂ψ(ρ),µ
+
∂L
∂ψ(ρ),µ
δψ(ρ) (26)
which is conserved in the sense that Θµ
;µ = 0.
In the case of isometries we have δψ(ρ) = −iξa
X
(ρ)
a ψ(ρ). Consequently, each
isometry of parameter ξa
give rise to the corresponding conserved current
Θµ
a = i X
(ρ)
a ψ(ρ)
∂L
∂ψ(ρ),µ
−
∂L
∂ψ(ρ),µ
X(ρ)
a ψ(ρ) , a = 1, 2...N . (27)
18

Then we may deﬁne the relativistic scalar product , as
⇒ ψ, ψ = i
∂∆
dσµ
√
g ψ
∂L
∂ψ ,µ
−
∂L
∂ψ,µ
ψ , (28)
such that the conserved quantities (charges) can be represented as expectation values
of isometry generators,
⇒ Ca =
∂∆
dσµ
√
g Θµ
a = ψ(ρ), X(ρ)
a ψ(ρ) , (29)
Notice that the operators X are self-adjoint with respect to this scalar product, i. e.
Xψ, ψ = ψ, Xψ .
From the algebra freely generated by the isometry generators we may select the sets
of commuting operators {A1, A2, ...An} determining the fundamental solutions of
particles, Uα ∈ F+
, and antiparticles, Vα ∈ F−
, that depend on the set of the
corresponding eigenvalues α = {a1, a2, ...an} spanning a discrete or continuous
spectra of the common eigenvalue problems
AiUα = aiUα , AiVα = −aiVα , i = 1, 2...n . (30)
19

After determining the fundamental solutions we may write the mode expansion
⇒ ψ(ρ)(x) =


α∈Σd
+
α∈Σc
d(α)

 [Uα(x)a(α) + Vα(x)b∗
(α)] , (31)
where we sum over the discrete part (Σd) and integrate over the continuous part (Σc) of
the spectrum Σ = Σd ∪ Σc.
The fundamental solutions are orthogonal with respect to the relativistic scalar product
and can be normalized such that
Uα, Uα = ± Vα, Vα = δ(α, α ) =
δα,α if α, α ∈ Σd
δ(α − α ) if α, α ∈ Σc
(32)
Uα, Vα = Vα, Uα = 0 , (33)
where the sign + arises for fermions while the sign − is obtained for bosons.
20

Canonical quantization
The theory get a physical meaning only after performing the second quantization
postulating canonical non-vanishing rules (with the notation [x, y]± = xy ± yx) as
a(α), a†
(α )
±
= b(α), b†
(α )
±
= δ(α, α ) . (34)
Then the fields ψ(ρ) become quantum fields (with b†
instead of b∗
) while the conserved
quantities (29) become one-particle operators,
⇒ Ca → X(ρ)
a =: ψ(ρ), X(ρ)
a ψ(ρ) : (35)
calculated respecting the normal ordering of the operator products [26]. Now the one
particle operators X
(ρ)
a are the basis generators of a rep. of the algebra s(M) with
values in operator algebra. In a similar manner one can define the generators of the
internal symmetries as for example the charge one-particle operator Q =: ψ, ψ :.
Thus we obtain a reach operator algebra formed by field operators and the one-particle
ones which have the obvious properties
[X, ψ(x)] = −(Xψ)(x) , [X, Y] =: ψ, ([X, Y ]ψ) : . (36)
21

In general, if the one-particle operator X does not mix among themselves the subspaces
of fundamental solutions it can be expanded as
X = : ψ, Xψ := X(+)
+ X(−)
=
α∈Σ α ∈Σ
˜X(+)
(α, α )a†
(α)a(α ) + ˜X(−)
(α, α )b†
(α)b(α ) , (37)
where
˜X(+)
(α, α ) = Uα, XUα , ˜X(−)
(α, α ) = Vα, XVα . (38)
When there are differential operators ˜X(±)
acting on the continuous variables of the set
{α} such that ˜X(±)
(α, α ) = δ(α, α ) ˜X(±)
we say that ˜X(±)
are the operators of
the rep. {α} (in the sense of the relativistic QM).
We stress that all the isometry generators have this property such that the corresponding
operators ˜X
(±)
a are the basis generators of the isometry transformations of the ﬁeld
operators a and b . However, the algebraic relations (34) remain invariant only if a and
b transform according to UIRs of the isometry group.
A crucial problem is now the equivalence between the CR transformimg the
covariant ﬁeld ψ(ρ) and the set of UIRs transforming the particle and antiparticle
operators a and b. This will be referred here as the CR-UIR equivalence.
22

Figure 3: The CR-UIR equivalence of the reps. of the group S(M) and its algebra s(M)
23

3. Covariant fields in special relativity
The problem of CR-UIR equivalence is successfully solved in special relativity thanks to
the Wigner theory of induced reps. of the Poincaré group.
On the Minkowski spacetime, (M0, η), the fields ψ(ρ) transform under isometries
according to manifest CRs in inertial (local) frames defined by eµ
ν = êµ
ν = δµ
ν .
Generators of manifest CRs
The isometries are just the transformations x → x = Λ[A(ω)]x − a of the Poincaré
group I(M0) = P↑
+ = T(4) L↑
+ [24] whose universal covering group is S(M0) =
˜P↑
+ = T(4) SL(2, C). The manifest CRs, T(ρ)
: (A, a) → T
(ρ)
A,a, of the S(M0)
group have the transformation rules
⇒ (T
(ρ)
A,aψ(ρ))(x) = ρ(A)ψ(ρ) Λ(A)−1
(x + a) , (39)
24

and the well-known basis-generators of the s(M0) algebra,
ˆPµ ≡ ˆX
(ρ)
(µ) = i∂µ , (40)
ˆJ(ρ)
µν ≡ ˆX
(ρ)
(µν) = i(ηµαxα
∂ν − ηναxα
∂µ) + S(ρ)
µν , (41)
which have point-independent spin parts denoted by S
(ρ)
ˆµˆν instead of ρ(Sˆµˆν). Hereby, it is
convenient to denote the energy operator as ˆH = ˆP0 and write the sl(2, C) generators,
ˆJ
(ρ)
i =
1
2
εijk
ˆJ
(ρ)
jk = −iεijkxj
∂k + S
(ρ)
i , S
(ρ)
i =
1
2
εijkS
(ρ)
jk , (42)
ˆK
(ρ)
i = ˆJ
(ρ)
0i = i(xi
∂t + t∂i) + S
(ρ)
0i , i, j, k... = 1, 2, 3 , (43)
denoting S2
= SiSi and S2
0 = S0iS0i. Thus we lay out the standard basis of the
s(M0) algebra, { ˆH, ˆPi, ˆJ
(ρ)
i , ˆK
(ρ)
i }.
The invariants of the manifest covariant ﬁelds are the eigenvalues of the Casimir
operators of the reps. T(ρ)
that read
ˆC1 = ˆPµ
ˆPµ
, ˆC
(ρ)
2 = −ηµν
ˆW(ρ) µ ˆW(ρ) ν
, (44)
25

where the components of the Pauli-Lubanski operator [24],
ˆW(ρ) µ
= −
1
2
εµναβ ˆPν
ˆJ
(ρ)
αβ , (45)
are defined by the skew-symmetric tensor with ε0123
= −ε0123 = −1. Thus we obtain,
ˆW
(ρ)
0 = ˆJ
(ρ)
i
ˆPi = S
(ρ)
i
ˆPi , ˆW
(ρ)
i = ˆH ˆJ
(ρ)
i + εijk
ˆK
(ρ)
j
ˆPk . (46)
The first invariant (44a) gives the mass condition, ˆP2
ψ(ρ) = m2
ψ(ρ), fixing the orbit
in the momentum spaces on which the fundamental solutions are defined. The second
invariant is less relevant for the CRs since its form in configurations is quite complicated
ˆC
(ρ)
2 = −(S(ρ)
)2
∂2
t + 2(iS
(ρ)
0k − εijkS
(ρ)
i S
(ρ)
0j )∂k∂t
− (S0
(ρ)
)2
∆ − (S
(ρ)
i S
(ρ)
j + S
(ρ)
0i S
(ρ)
0j )∂i∂j . (47)
Consequently, we may study its action in the momentum reps. where it selects the
induced Wigner UIRs equivalent with the CR. Nevertheless, for fields with unique spin
with ρ = ρ(s) = (s, 0) ⊕ (0, s) we obtain in the rest frame where Pi ∼ 0 that
ˆC
ρ(s)
2 = m2
s(s + 1) (48)
26

since then S
ρ(s)
0i = ±iS
ρ(s)
i .
In the Poincaré algebra we find the complete system of commuting operators
{ ˆH, ˆP1, ˆP2, ˆP3} defining the momentum rep.. The fundamental solutions are common
eigenfunctions of this system such that any covariant quantum field can be written as
ψ(ρ)(x) = d3
p
sσ
Up,sσ(x)asσ(p) + Vp,sσ(x)b†
sσ(p) (49)
where asσ and bsσ are the field operators of a particle and antiparticle of spin s and
polarization σ while the fundamental solutions have the form
Up,sσ(x) =
1
(2π)
3
2
usσ(p)e−iEt+ip·x
, Vp,sσ(x) =
1
(2π)
3
2
vsσ(p)eiEt−ip·x
. (50)
The vectors usσ(p) and vsσ(p) have to be determined by the concrete form of the field
equation and relativistic scalar product. However, when the field equations are linear we
can postulate the orthonormalization relations
usσ(p)us σ (p) = vsσ(p)vs σ (p) = δss δσσ , (51)
usσ(p)vs σ (p) = vsσ(p)us σ (p) = 0 . (52)
27

that guarantee the separation of the particle and antiparticle sectors.
For the massive ﬁelds of mass m the momentum spans the orbit Ωm = {p | p2
= m2
}
which means that p0 = ±E where E = m2 + p2. The solutions U are considered
of positive frequencies having p0 = E while for the negative frequency ones, V , we
must take p0 = −E. In this manner the general rule (30) of separating the particle and
antiparticle modes becomes
HUp,sσ = EUp,sσ , HVp,sσ = −EVp,sσ , (53)
PUp,sσ = p Up,sσ , PVp,sσ = −p Vp,sσ . (54)
Wigner’s induced UIRs
The Wigner theory of the induced UIRs is based on the fact that the orbits in
momentum space may be built by using Lorentz transformations [6, 7]. In the case
of massive particles we discuss here, any p ∈ Ωm can be obtained applying a boost
transformation Lp ∈ L↑
+ to the representative momentum ˚p = (m, 0, 0, 0) such that
p = Lp ˚p.
28

The rotations that leave ˚p invariant, R˚p = ˚p, form the stable group SO(3) ⊂ L↑
+
whose universal covering group SU(2) is called the little group associated to the
representative momentum ˚p.
We observe that the boosts Lp are deﬁned up to a rotation since LpR ˚p = Lp ˚p.
Therefore, these span the homogeneous space L↑
+/SO(3). The corresponding
transformations of the SL(2, C) group are denoted by Ap ∈ SL(2, C)/SU(2)
assuming that these satisfy Λ(Ap) = Lp and A˚p = 1 ∈ SL(2, C).
In applications one prefers to chose genuine Lorentz transformatios Ap = e−iαni
S0i
with
α = atrctanhp
E and ni
= pi
p with p = |p|. In the spinor rep. ρs = (1
2, 0) ⊕ (0, 1
2) of
the Dirac theory one ﬁnds [27]
ρs(Ap) =
E + m + γ0
γi
pi
2m(E + m)
. (55)
where γµ
denote the Dirac matrices. The corresponding transformations of the L↑
+
group, Lp = Λ(Ap), have the matrix elements
(Lp)0 ·
· 0 =
E
m
, (Lp)0 ·
· i = (Lp)i ·
· 0 =
pi
m
, (Lp)i ·
· j = δij +
pi
pj
m(E + m)
. (56)
29

Furthermore, we look for the transformations in momentum rep. generated by the CR
under consideration. After a little calculation we obtain
s σ
us σ (p)(TA,aas σ )(p) =
sσ
ρ(A)usσ p asσ p e−ia·p
(57)
s σ
vs σ (p)(TA,ab†
s σ )(p) =
sσ
ρ(A)vsσ p b†
sσ p eia·p
(58)
where a · p = Ea0
− p · a and p = Λ(A)−1
p.
Focusing on the ﬁrst equation, we introduce the Wigner mode functions
usσ(p) = ρ(Ap)˚usσ (59)
where the vectors ˚usσ ∈ V(ρ) are independent on p and satisfy ˚usσ˚us σ =
usσ(p)us σ (p) = δss δσσ according to Eq. (51). We obtain thus the transformation
rule of the Wigner reps. induced by the subgroup T(4) SU(2) that read [6, 8]
⇒ (TA,aasσ)(p) =
σ
Ds
σσ (A, p)asσ (p )eia·p
(60)
where
⇒ Ds
σσ (A, p) = ˚usσρ[W(A, p)]˚usσ , W(A, p) = A−1
p AAp (61)
30

The Wigner transformations W(A, p) = A−1
p AAp is of the little group SU(2) since
one can verify that Λ[W(A, p)] = L−1
p Λ(A)Lp ∈ SO(3) leaving invariant the
representative momentum ˚p.
Therefore the matrices Ds
realize the UIR of spin (s) of the little group SU(2) that
induces the Wigner UIR (60) denoted by (s, ±m) [8].
Note that the role of the vectors ˚usσ is to select the spin content of the CR determining
the Wigner UIRs whose direct sum is equivalent to the CR T(ρ)
.
A similar procedure can be applied for the antiparticle but selecting the normalized
vectors ˚vsσ ∈ V(ρ) such that ˚vsσ˚vs σ = δss δσσ and ˚vsσρ[W(A, p)]˚vsσ =
Ds
σσ (A, p)∗
since the operators a and b must transform alike under isometries [24].
Moreover, from Eq. (52) we deduce that the vectors ˚u and ˚v must be orthogonal,
˚usσ˚vs σ = ˚vsσ˚us σ = 0.
The conclusion is that the CRs are equivalent to direct sums of Wigner UIRs with an
arbitrary spin content deﬁned by the vectors ˚usσ and ˚vsσ. For each spin s we meet the
31

UIR (±m, s) in the space Vs ⊂ V(ρ) of the linear UIR of the group SU(2) generated
by the matrices S
(s)
i .
The transformation (60) allows us to derive the generators of the UIRs in momentum
rep. (denoted by tilde) that are differential operator acting alike on the operators asσ(p)
and bsσ(p) seen as functions of p. Thus for each UIR (s, ±m) we can write down the
basis generators
˜J
(s)
i = −iεijkpj
∂pk + S
(s)
i , (62)
˜K
(s)
i = iE∂pi −
pi
2E
+
1
E + m
εijkpj
S
(s)
k . (63)
With their help we derive the components of the Pauli-Lubanski operator
˜W
(s)
0 = p · S(s)
, ˜W
(s)
i = mS
(s)
i +
pi
E + m
p · S(s)
, (64)
and we recover the well-known result [8]
⇒ ˜C1 = m2
, ˜C
(s)
2 = m2
(S(s)
)2
∼ m2
s(s + 1) (65)
32

Finally we stress that the Wigner theory determine completely the form of the covariant
fields without using field equations. Thus in special relativity we have two symmetric
equivalent procedures:
(i) to start with the covariant field equation that gives the form of the covariant field
determining its CR, or
(ii) to construct the Wigner covariant field and then to derive its field equation [24]. The
typical example is the Dirac field in Minkowski spacetime [8, 27].
33

4. Covariant fields on de Sitter spacetime
The Wigner theory works only in local-Minkowskian manifold whose isometry group has
a similar structure as the Poincaré one having an Abelian normal subgroup T(4).
Unfortunately the Abelian group T(3)P of the de Sitter isometry group is not a normal (or
invariant) subgroup such that the we must study of the de Sitter CRs in the configuration
space following to consider the UIRs in momentum representation after the field is
determined by a concrete field equation.
de Sitter isometries and Killing vectors
Let (M, g) be the de Sitter spacetime defined as the hyperboloid of radius 1/ω in
the five-dimensional flat spacetime (M5
, η5
) of coordinates zA
(labeled by the indices
A, B, ... = 0, 1, 2, 3, 4) and metric η5
= diag(1, −1, −1, −1, −1).
34

The local charts {x} can be introduced on (M, g) giving the set of functions zA
(x)
which solve the hyperboloid equation,
η5
ABzA
(x)zB
(x) = −
1
ω2
. (66)
Here we use the chart {t, x} with the conformal time t and Cartesian spaces
coordinates xi
deﬁned by
z0
(x) = −
1
2ω2t
1 − ω2
(t2
− x2
)
zi
(x) = −
1
ωt
xi
, (67)
z4
(x) = −
1
2ω2t
1 + ω2
(t2
− x2
)
This chart covers the expanding part of M for t ∈ (−∞, 0) and x ∈ R3
while the
collapsing part is covered by a similar chart with t > 0. Both these charts have the
conformal ﬂat line element,
ds2
= η5
ABdzA
(x)dzB
(x) =
1
ω2t2
dt2
− dx2
. (68)
35

Figure 4: de Sitter spacetime.
36

In addition, we consider the local frames {t, x; e} of the diagonal gauge,
e0
0 = −ωt , ei
j = −δi
j ωt , ˆe0
0 = −
1
ωt
, ˆei
j = −δi
j
1
ωt
. (69)
The gauge group G(η5
) = SO(1, 4) is the isometry group of M, since its
transformations, z → gz, g ∈ SO(1, 4), leave the equation (66) invariant. Its universal
covering group Spin(η5
) = Sp(2, 2) is not involved directly in our construction since
the spinor CRs are induced by the spinor representations of its subgroup SL(2, C).
Therefore, we can restrict ourselves to the group SO(1, 4) for which we adopt the
parametrization
g(ξ) = exp −
i
2
ξAB
SAB ∈ SO(1, 4) (70)
with skew-symmetric parameters, ξAB
= −ξBA
, and the covariant generators SAB
of the fundamental representation of the so(1, 4) algebra carried by M5
. These
generators have the matrix elements,
(SAB)C ·
· D = i δC
A ηBD − δC
B ηAD . (71)
37

The principal so(1, 4) basis-generators with physical meaning [20] are the energy H =
ωS04, angular momentum Jk = 1
2εkijSij, Lorentz boosts Ki = S0i, and the Runge-
Lenz-type vector Ri = Si4. In addition, it is convenient to introduce the momentum
Pi = −ω(Ri + Ki) and its dual Qi = ω(Ri − Ki) which are nilpotent matrices
(i. e. (Pi)3
= (Qi)3
= 0) generating two Abelian three-dimensional subgroups,
T(3)P and respectively T(3)Q. All these generators may form different bases of the
algebra so(1, 4) as, for example, the basis {H, Pi, Qi, Ji} or the Poincar´e-type one,
{H, Pi, Ji, Ki}. We note that the four-dimensional restriction of the so(1, 3) subalegra
generate the vector representation of the group L↑
+.
Using these generators we can derive the SO(1, 4) isometries, φg, deﬁned as
z[φg(x)] = g z(x). (72)
The transformations g ∈ SO(3) ⊂ SO(4, 1) generated by Ji, are simple rotations
of zi
and xi
which transform alike since this symmetry is global. The transformations
generated by H,
exp(−iξH) :
z0
→ z0
cosh α − z4
sinh α
zi
→ zi
z4
→ −z0
sinh α + z4
cosh α
(73)
38

whith α = ωξ, produce the dilatations t → t eα
and xi
→ xi
eα
, while the T(3)P
transformations
exp(−iξi
Pi) :
z0
→ z0
+ ω ξ · z + 1
2 ω2
ξ
2
(z0
+ z4
)
zi
→ zi
+ ω ξi
(z0
+ z4
)
z4
→ z4
− ω ξ · z − 1
2 ω2
ξ
2
(z0
+ z4
)
(74)
give rise to the space translations xi
→ xi
+ ξi
at ﬁxed t. More interesting are the
T(3)Q transformations generated by Qi/ω,
exp(−iξi
Qi/ω) :
z0
→ z0
− ξ · z + 1
2 ξ
2
(z0
− z4
)
zi
→ zi
− ξi
(z0
− z4
)
z4
→ z4
− ξ · z + 1
2 ξ
2
(z0
− z4
)
(75)
which lead to the isometries
t →
t
1 − 2ω ξ · x − ω2ξ
2
(t2 − x2)
(76)
xi
→
xi
+ ωξi
(t2
− x2
)
1 − 2ω ξ · x − ω2ξ
2
(t2 − x2)
. (77)
39

We observe that z0
+ z4
= − 1
ω2t is invariant under translations (74), ﬁxing the value of
t, while z0
− z4
= t2
−x2
t is left unchanged by the t(3)Q transformations (75).
The orbital basis-generators of the natural representation of the s(M) algebra (carried
by the space of the scalar functions over M5
) have the standard form
L5
AB = i η5
ACzC
∂B − η5
BCzC
∂A = −iKC
(AB)∂C (78)
which allows us to derive the corresponding Killing vectors of (M, g), k(AB), using the
identities k(AB)µdxµ
= K(AB)CdzC
. Thus we obtain the following components of the
Killing vectors:
k0
(04) = t , ki
(04) = xi
, k0
(0i) = k0
(4i) = ωtxi
(79)
kj
(0i) = ωxi
xj
+ δj
i
1
2ω
[ω2
(t2
− x2
) − 1] (80)
kj
(4i) = ωxi
xj
+ δj
i
1
2ω
[ω2
(t2
− x2
) + 1] (81)
kk
(ij) = δk
j xi
− δk
i xj
. (82)
40

Generators of induced CRs
In the covariant parametrization of the sp(2, 2) algebra adopted here, the generators
X
(ρ)
(AB) corresponding to the Killing vectors k(AB) result from equation (21) and the
functions (19) with the new labels a → (AB). Then we have
H = ωX
(ρ)
(04) = −iω(t∂t + xi
∂i) , (83)
J
(ρ)
i =
1
2
εijkX
(ρ)
(jk) = −iεijkxj
∂k + S
(ρ)
i , S
(ρ)
i =
1
2
εijkS
(ρ)
jk , (84)
K
(ρ)
i = X
(ρ)
(0i) = xi
H +
i
2ω
[1 + ω2
(x2
− t2
)]∂i − ωtS
(ρ)
0i + ωS
(ρ)
ij xj
, (85)
R
(ρ)
i = X
(ρ)
(i4) = −K
(ρ)
i +
1
ω
i∂i . (86)
where H is the energy (or Hamiltonian), J total angular momentum , K generators
of the Lorentz boosts , and R is a Runge-Lenz type vector. These generators form
the basis {H, J
(ρ)
i , K
(ρ)
i , R
(ρ)
i } of the covariant rep. of the sp(2, 2) algebra with the
following commutation rules:
41

J
(ρ)
i , J
(ρ)
j = iεijkJ
(ρ)
k , J
(ρ)
i , R
(ρ)
j = iεijkR
(ρ)
k , (87)
J
(ρ)
i , K
(ρ)
j = iεijkK
(ρ)
k , R
(ρ)
i , R
(ρ)
j = iεijkJ
(ρ)
k , (88)
K
(ρ)
i , K
(ρ)
j = −iεijkJ
(ρ)
k , R
(ρ)
i , K
(ρ)
j =
i
ω
δijH , (89)
and
H, J
(ρ)
i = 0 , H, K
(ρ)
i = iωR
(ρ)
i , H, R
(ρ)
i = iωK
(ρ)
i . (90)
In some applications it is useful to replace the operators K(ρ)
and R(ρ)
by the Abelian
ones, i. e. the momentum operator P and its dual Q(ρ)
, whose components are deﬁned
as
Pi = −ω(R
(ρ)
i + K
(ρ)
i ) = −i∂i , Q
(ρ)
i = ω(R
(ρ)
i − K
(ρ)
i ) , (91)
obtaining the new basis {H, Pi, Q
(ρ)
i , J
(ρ)
i }.
The last two bases bring together the conserved energy (83) and momentum (91a) which
are the only genuine orbital operators, independent on ρ. What is speciﬁc for the de
Sitter symmetry is that these operators can not be put simultaneously in diagonal form
since they do not commute to each other.
42

Casimir operators
The first invariant of the CR T(ρ)
is the quadratic Casimir operator
C
(ρ)
1 = − ω21
2
X
(ρ)
(AB)X(ρ) (AB)
(92)
= H2
+ 3iωH − Q(ρ)
· P − ω2
J(ρ)
· J(ρ)
. (93)
After a few manipulation we obtain its definitive expression
C
(ρ)
1 = EKG + 2iωe−ωt
S
(ρ)
0i ∂i − ω2
(S(ρ)
)2
, (94)
depending on the Klein-Gordon operator EKG = −∂2
t − 3 ω∂t + e−2ωt
∆.
The second Casimir operator, C
(ρ)
2 = −η5
ABW(ρ) A
W(ρ) B
, is written with the help of
the five-dimensional vector-operator W(ρ)
whose components read [13]
W(ρ) A
=
1
8
ω εABCDE
X
(ρ)
(BC)X
(ρ)
(DE) , (95)
where ε01234
= 1 and the factor ω assures the correct flat limit. After a little calculation
we obtain the concrete form of these components,
43

W
(ρ)
0 = ω J(ρ)
· R(ρ)
, (96)
W
(ρ)
i = H J
(ρ)
i + ω εijkK
(ρ)
j R
(ρ)
k , (97)
W
(ρ)
4 = −ω J(ρ)
· K(ρ)
, (98)
which indicate that W(ρ)
plays an important role in theories with spin, similar to that of
the Pauli-Lubanski operator (45) of the Poincar´e symmetry. For example, the helicity
operator is now W
(ρ)
0 − W
(ρ)
4 = S
(ρ)
i Pi.
Then by using the components (96)-(98) we are faced with a complicated calculation but
which can be performed using algebraic codes on computer. Thus we obtain the closed
form of the second Casimir operator,
C
(ρ)
2 = −ω2
(S(ρ)
)2
(t2
∂2
t − 2t∂t + 2) + 2ω2
t2
(iS
(ρ)
0k − εijkS
(ρ)
i S
(ρ)
0j )∂k∂t
+ωt (S0
(ρ)
)2
∆ − (S
(ρ)
i S
(ρ)
j + S
(ρ)
0i S
(ρ)
0j )∂i∂j
−2iω2
t(S
(ρ)
i S
(ρ)
k S
(ρ)
0i + S
(ρ)
0k )∂k . (99)
44

In the case of fields with unique spin s we must select the reps. ρ(s) = (s, 0)⊕(0, s),
for which we have to replace S
ρ(s)
0i = ± iS
ρ(s)
i in equation (99) finding the remarkable
identity
C
ρ(s)
2 = C
ρ(s)
1 (Sρ(s)
)2
− 2ω2
(Sρ(s)
)2
+ ω2
[(Sρ(s)
)2
]2
. (100)
It is interesting to look for the invariants of the particles at rest in the chart {t, x}. These
have the vanishing momentum (Pi ∼ 0) so that H acts as i∂t and, therefore, it can
be put in diagonal form its eigenvalues being just the rest energies, E0. Then, for each
subspace Vs ⊂ V(ρ) of given spin, s, we obtain the eigenvalues of the first Casimir
operator,
C
ρ(s)
1 ∼ E2
0 + 3iωE0 − ω2
s(s + 1) , (101)
using Eq. (94) while those of the second Casimir operator,
C
ρ(s)
2 ∼ s(s + 1)(E2
0 + 3iωE0 − 2ω2
) , (102)
result from equation (99). These eigenvalues are real numbers so that the rest energies,
E0 = E0 − 3iω
2 , must be complex numbers whose imaginary parts are due to the
decay produced by the de Sitter expansion.
45

The above results indicate that the CRs are reducible to direct sums of UIRs of the
principal series [9]. These are labeled by two weights, (p, q), with p = s while q is a
solution of the equation q(1 − q) = 1
ω2 ( E0)2
+ 1
4.
In the flat limit we recover the Poincaré generators. We observe that the generators (84)
are independent on ω having the same form as in the Minkowski case, J
(ρ)
k = ˆJ
(ρ)
k . The
other generators have the limits
lim
ω→0
H = ˆH = i∂t , lim
ω→0
(ωR
(ρ)
i ) = − ˆPi = i∂i , lim
ω→0
K
(ρ)
i = ˆK
(ρ)
i , (103)
which means that the basis {H, Pi, J
(ρ)
i , K
(ρ)
i } of the algebra s(M) = sp(2, 2) tends
to the basis { ˆH, ˆPi, ˆJ
(ρ)
i , ˆK
(ρ)
i } of the s(M0) algebra when ω → 0. Moreover, the
Pauli-Lubanski operator (45) is the flat limit of the five-dimensional vector-operator (95)
since
lim
ω→0
W
(ρ)
0 = ˆW
(ρ)
0 , lim
ω→0
W
(ρ)
i = ˆW
(ρ)
i , lim
ω→0
W
(ρ)
4 = 0 . (104)
Under such circumstances the limits of our invariants read
lim
ω→0
C
(ρ)
1 = ˆC1 = ˆP2
, lim
ω→0
C
(ρ)
2 = ˆC
(ρ)
2 , (105)
indicating that their physical meaning may be related to the mass and spin of the matter
fields in a similar manner as in special relativity.
46

Minkowski de Sitter
CRs manifest SL(2, C) CRs induced by SL(2, C)
H = i∂t H = −iω(t∂t + xi
∂i)
Pi
= −i∂i Pi
= −i∂i
J
(ρ)
i = −iεijkxj
∂k + S
(ρ)
i J
(ρ)
i = −iεijkxj
∂k + S
(ρ)
i
K
(ρ)
i = i(xi
∂t + t∂i) + S
(ρ)
0i K
(ρ)
i = xi
H + i
2ω[1 + ω2
(x2
− t2
)]∂i
−ωtS
(ρ)
0i + ωS
(ρ)
ij xj
UIR ˜P↑
+ = T(4) SL(2, C) UIR Spin(1, 4) = Sp(2, 2)
C1 = m2
C
ρ(s)
1 = M2
+ 9
4 ω2
− ω2
s(s + 1)
C
ρ(s)
2 = m2
s(s + 1) C
ρ(s)
2 = M2
+ 1
4 ω2
s(s + 1)
where m = E0 where M = E0
Scalar field s = 0 M = m2 − 9
4ω2 C1 = m2
C2 = 0
Dirac field s = 1
2 M = m C1 = m2
+ 3
2 ω2
C2 = 3
4 m2
+ 1
4 ω2
Proca field s = 1 M = m2 − 1
4ω2 C1 = m2
C2 = 2m2
47

5. The Dirac field on de Sitter spacetimes
In the absence of a strong theory like the Wigner one in the flat case we must study
the CR-UIR equivalence resorting to the covariant field equations able to give us the
structure of the covariant field. Then, bearing in mind that the de Sitter UIRs are well-
studied [9, 10], we can establish the CR-UIR equivalence by studying the CR and UIR
Casimir operators in configurations and momentum rep..
In what follows we concentrate on the Dirac equation on the de Sitter spacetime since
this is the only equation on this background giving the natural rest energy E0 = m
[20].
Invariants of the spinor CR
In the frame {t, x; e} introduced above the free Dirac equation takes the form [18],
(ED − m)ψ(x) = −iωt γ0
∂t + γi
∂i +
3iω
2
γ0
− m ψ(x) = 0 , (106)
48

depending on the point-independent Dirac matrices γˆµ
that satisfy {γˆα
, γ
ˆβ
} = 2ηˆα ˆβ
giving rise to the basis-generators S(ρs) ˆα ˆβ
= i
4[γˆα
, γ
ˆβ
] of the spinor rep. ρs = ρ(1
2) =
(1
2, 0) ⊗ (0, 1
2) of the group ˆG = SL(2, C) that induces the spinor CR [3, 18, 19].
Eq. (106) can be analytically solved either in momentum or energy bases with correct
orthonormalization and completeness properties [18, 19] with respect to the relativistic
scalar product
ψ, ψ = d3
x (−ωt)−3
ψ(t, x)γ0
ψ (t, x) . (107)
The mode expansion in the spin-momentum rep. [19],
ψ(t, x) = d3
p
σ
Up,σ(x)a(p, σ) + Vp,σ(x)b†
(p, σ) , (108)
is written in terms of the ﬁeld operators, a and b (satisfying canonical anti-commutation
rules), and the particle and antiparticle fundamental spinors of momentum p (with p =
|p|) and polarization σ = ±1
2,
Up,σ(t, x ) =
1
(2π)
3
2
up,σ(t)eip·x
, Vp,σ(t, x ) =
1
(2π)
3
2
vp,σ(t)e−ip·x
(109)
49

whose time-dependent terms have the form [19, 21]
up,σ(t) =
i
2
πp
ω
1
2
(ωt)2 e
1
2πµ
H
(1)
ν− (−pt) ξσ
e−1
2πµ
H
(1)
ν+ (−pt) σ·p
p ξσ
, (110)
vp,σ(t) =
i
2
πp
ω
1
2
(ωt)2 e−1
2πµ
H
(2)
ν− (−pt) σ·p
p ησ
e
1
2πµ
H
(2)
ν+ (−pt) ησ
, (111)
in the standard rep. of the Dirac matrices (with diagonal γ0
) and a ﬁxed vacuum of
the Bounch-Davies type [21]. Obviously, the notation σi stands for the Pauli matrices
while the point-independent Pauli spinors ξσ and ησ = iσ2(ξσ)∗
are normalized as
ξ+
σ ξσ = η+
σ ησ = δσσ [19]. The terms giving the time modulation depend on the
Hankel functions H
(1,2)
ν± of indices
ν± =
1
2
± iµ , µ =
m
ω
. (112)
Based on their properties (presented in Appendix B) we deduce
u+
p,σ(t)up,σ(t) = v+
p,σ(t)vp,σ(t) = (−ωt)3
(113)
50

obtaining the ortonormalization relations [18]
Up,σ, Up ,σ = Vp,σ, Vp ,σ = δσσ δ3
(p − p ) , (114)
Up,σ, Vp ,σ = Vp,σ, Up ,σ = 0 , (115)
that yield the useful inversion formulas, a(p, σ) = Up,σ, ψ and b(p, σ) = ψ, Vp,σ .
Moreover, it is not hard to verify that these spinors are charge-conjugated to each other,
Vp,σ = (Up,σ)c
= C(Up,σ)T
, C = iγ2
γ0
, (116)
and represent a complete system of solutions in the sense that [18]
d3
p
σ
Up,σ(t, x)U+
p,σ(t, x ) + Vp,σ(t, x)V +
p,σ(t, x ) = e−3ωt
δ3
(x − x ) .
(117)
The Dirac ﬁeld transforms under isometries x → x = φg(x) (with g ∈ I(M))
according to the CR Tg : ψ(x) → (Tgψ)(x ) = Ag(x)ψ(x) whose generators are
given by Eqs (83) - (86) where now ρ = ρs. Then, according to equations (94) and
(106) we obtain the identity
C
(ρs)
1 = E2
D +
3
2
ω2
14×4 ∼ m2
+
3
2
ω2
. (118)
51

This result and equation (101) yield the rest energy of the Dirac field,
E0 = −
3iω
2
± m , (119)
which has a natural simple form where the decay (first) term is added to the usual rest
energy of special relativity. A similar result can be obtained by solving the Dirac equation
with vanishing momentum.
The second invariant results from equations (100) and (118) if we take into account that
(S(ρs)
)2
= 3
4 14×4. Thus we find
C
(ρs)
2 =
3
4
E2
D +
3
16
ω2
14×4 ∼
3
4
m2
+
1
4
ω2
= ω2
s(s + 1)ν+ν− , (120)
where s = 1
2 is the spin and ν± = 1
2 ±im
ω are the indices of the Hankel functions giving
the time modulation of the Dirac spinors of the momentum basis [18].
These invariants define the UIRs that in the flat limit become Wigner’s UIRs (±m, 1
2)
since
lim
ω→0
C
(ρs)
1 ∼ m2
, lim
ω→0
C
(ρs)
2 ∼
3
4
m2
. (121)
52

Invariants of UIRs in momentum rep.
The above inversion formulas allow us to write the transformation rules in momentum
rep. as
(Tga)(p, σ) = Up,σ, [ρs(Ag)ψ] ◦ φ−1
g , (122)
(Tgb)(p, σ) = [ρs(Ag)ψ] ◦ φ−1
g , Vp,σ , (123)
but, unfortunately, these scalar product are complicated integrals that cannot be solved.
Therefore, we must restrict ourselves to study the corresponding Lie algebras focusing
on the basis generators in momentum rep..
Any self-adjoint generator X of the spinor rep. of the group S(M) gives rise to a
conserved one-particle operator of the QFT,
X =: ψ, Xψ := X(+)
+X(−)
= d3
p α†
(p) ˜X(+)
α(p) + β†
(p) ˜X(−)
β(p) ,
(124)
53

calculated respecting the normal ordering of the operator products [26]. The operators
˜X(±)
are the generators of CRs in momentum rep. acting on the operator valued Pauli
spinors,
α(p) =
a(p, 1
2)
a(p, −1
2)
, β(p) =
b(p, 1
2)
b(p, −1
2)
. (125)
As observed in Ref. [16], the straightforward method for ﬁnding the structure of these
operators is to evaluate the entire expression (124) by using the form (108) where the
ﬁeld operators a and b satisfy the canonical anti-commutation rules [16, 18].
For this purpose we consider several identities written with the notation ∂pi
= ∂
∂pi
as
H Up,σ(t, x) = −iω pi
∂pi +
3
2
Up,σ(t, x) ,
H Vp,σ(t, x) = −iω pi
∂pi +
3
2
Vp,σ(t, x) ,
that help us to eliminate some multiplicative operators and the time derivative when
we inverse the Fourier transform. Furthermore, by applying the Green theorem and
54

calculating some terms on computer we find two identical reps. whose basis
generators read, ˜P
(±)
i = ˜Pi = pi and
˜H(±)
= ω ˜X
(±)
(04) = iω pi∂pi
+
3
2
(126)
˜J
(±)
i =
1
2
εijk
˜X
(±)
(jk) = −iεijkpj∂pk
+
1
2
σi (127)
˜K
(±)
i = ˜X
(±)
(0i) = i ˜H(±)
∂pi
+
ω
2
pi∆p − pi
p2
+ m2
2ωp2
+
1
2
εijk iω∂pj
− pj
m
2p2
σk (128)
˜R
(±)
i = ˜X
(±)
(i4) = − ˜K
(±)
i −
1
ω
˜Pi , (129)
where ∆p = ∂pi
∂pi
. These basis generators satisfy the specific sp(2, 2) commutation
rules of the form (87)-(90). Moreover, it is not difficult to verify that these are Hermitian
operators with respect to the scalar products of the momentum rep.
α, α = d3
p α†
(p)˜α(p) , β, β = d3
pβ†
(p) ˜β(p) . (130)
55

Therefore, we can conclude that these operators generate a pair of unitary reps. of the
group S(M).
Starting with the Pauli-Lubanski operator,
˜W
(±)
0 =
ω
4
(σ · p)∆p +
ων−
2
σ · ∂p +
im
2p2
(σ · p) p · ∂p
+
m2
− p2
+ 2iωm
4p2ω
σ · p , (131)
˜W
(±)
i =
i
2
(σ · p)∂pi
−
iν−
2p2
σi −
m
2ωp2
(σ · p)pi , (132)
˜W
(±)
4 = ˜W
(±)
0 +
1
2ω
σ · p , (133)
we calculate on computer the following Casimir operators,
˜C
(±)
1 = ω2
[−s(s + 1) − (q + 1)(q − 2)] = m2
+
3ω2
2
, (134)
˜C
(±)
2 = ω2
[−s(s + 1)q(q − 1)] = ω2
s(s + 1)ν+ν− =
3
4
m2
+
ω2
4
,(135)
recovering thus the results (118) and (120) obtained in conﬁgurations.
56

Dirac ﬁeld on Minkowski Dirac ﬁeld on de Sitter
UIR (1
2, ±m) of ˜P↑
+ (1
2, ν±) of Sp(2, 2)
Pi
= pi
Pi
= pi
˜H = E = m2 + p2 ˜H(±)
= iω pi∂pi
+ 3
2
˜J
(±)
i = −iεijkpj∂pk
+ 1
2σi
˜J
(±)
i = −iεijkpj∂pk
+ 1
2σi
˜K
(±)
i = iE∂pi − pi
2E
˜K
(±)
i = i ˜H(±)
∂pi
+ ω
2 pi∆p − pi
p2
+m2
2ωp2
+ 1
2(E+m) εijkpj
σk +1
2εijk iω∂pj
− pj
m
2p2 σk
˜W
(±)
0 = 1
2 p · σ ˜W
(±)
0 = ω
4 (σ · p)∆p + ων−
2 σ · ∂p
+im
2p2(σ · p) p · ∂p + m2
−p2
+2iωm
4p2ω σ · p
˜W
(±)
i = 1
2 mσi + pi
2(E+m)σ · p ˜W
(±)
i = i
2(σ · p)∂pi
− iν−
2p2 σi − m
2ωp2(σ · p)pi
˜W
(±)
4 = ˜W
(±)
0 + 1
2ω σ · p
˜C
(±)
1 = m2 ˜C
(±)
1 = m2
+ 3
2 ω2
˜C
(±)
2 = 3
4 m2 ˜C
(±)
2 = 3
4 m2
+ 1
4 ω2
57

CR-UIR equivalence
The above result shows that the identical spinor reps. we obtained here are UIRs of the
principal series corresponding to the canonical labels (s, q) with s = 1
2 and q = ν±. In
other words the spinor CR of the Dirac theory is equivalent with the orthogonal sum of
the equivalent UIRs of the particle and antiparticle sectors.
This suggests that the UIRs (s, ν±) of the group S(M) = Sp(2, 2) can be seen
as being analogous to the Wigner ones (s, ±m) of the Dirac theory in Minkowski
spacetime.
In general, the above equivalent spinor UIRs may not coincide since the expressions
of their basis generators are strongly dependent on the arbitrary phase factors of the
fundamental spinors whether these depend on p. Thus if we change
Up,σ → eiχ+
(p)
Up,σ , Vp,σ → e−iχ−
(p)
Vp,σ , (136)
with χ±
(p) ∈ R, performing simultaneously the associated transformations,
α(p) → e−iχ+
(p)
α(p) , β(p) → e−iχ−
(p)
β(p) , (137)
58

that preserves the form of ψ, we find that the operators ˜Pi keep their forms while the
other generators are changing, e. g. the Hamiltonian operators transform as, ˜H(±)
→
˜H(±)
+ pi
∂piχ±
(p). Obviously, these transformations are nothing other than unitary
transformations among equivalent UIRs. Note that thanks to this mechanism one can
fix suitable phases for determining desired forms of the basis generators keeping thus
under control the flat and rest limits of these operators in the Dirac [19] or scalar [16, 28]
field theory on M.
At the level of QFT, the operators {X(AB)}, given by Eq. (124) where we introduce the
differential operators (126) -(129), generate a reducible operator valued CR which can
be decomposed as the orthogonal sum of CRs - generated by {X
(+)
(AB)} and {X
(−)
(AB)}
- that are equivalent between themselves and equivalent to the UIRs (1
2, ν±) of the
sp(2, 2) algebra. These one-particle operators are the principal conserved quantities of
the Dirac theory corresponding to the de Sitter isometries via Noether theorem.
It is remarkable that in our formalism we have ˜X
(+)
AB = ˜X
(−)
AB which means that the
particle and antiparticle sectors bring similar contributions such that we can say that
these quantities are additive, e. g., the energy of a many particle system is the sum of
the individual energies of particles and antiparticles.
59

Other important conserved one-particle operators are the components of the Pauli -
Lubanski operator,
WA = W
(+)
A + W
(−)
A = d3
p α†
(p) ˜W
(+)
A α(p) + β†
(p) ˜W
(−)
A β(p) , (138)
as given by Eqs. (131)-(133).
The Casimir operators of QFT have to be calculated according to Eqs. (92) and (99) but
by using the one-particle operators X(AB) and WA instead of ˜X(AB) and ˜WA. We
obtain the following one-particle contributions
C1 = m2
+
3
2
ω2
N + · · · , C2 =
3
4
m2
+
1
4
ω2
N + · · · , (139)
where N = N(+)
+ N(−)
is the usual operator of the total number of particles and
antiparticles.
Thus the additivity holds for the entire theory of the spacetime symmetries in contrast
with the conserved charges of the internal symmetries that take different values for
particles and antiparticles as, for example, the charge operator corresponding to the
U(1)em gauge symmetry [21] that reads Q = q(N(+)
− N(−)
).
60

6. Concluding remarks
The principal conclusion is that the QFT on the de Sitter background has similar features
as in the flat case. Thus the covariant quantum fields transforming according to CRs
induced by the reps. of the group ˆG = SL(2, C) that must be equivalent to orthogonal
sums of UIRs of the group S(M) = Sp(2, 2) whose specific invariants depend only
on particle masses and spins.
The example is the spinor CR of the Dirac theory that is induced by the linear rep.
(1
2, 0) ⊗ (0, 1
2) of the group ˆG but is equivalent to the orthogonal sum of two equivalent
UIRs of the group S(M) labelled by (1
2, ν±).
Thus at least in the case of the Dirac field we recover a similar conjuncture as in the
Wigner theory of the induced reps. of the Poincaré group in special relativity.
However, the principal difference is that the transformations of the Wigner UIRs can be
written in closed forms while in our case this cannot be done because of the technical
difficulties in solving the integrals (122) and (123). For this reason we were forced to
restrict ourselves to study only the reps. of the corresponding algebras.
61

This is not an impediment since physically speaking we are interested to know the
properties of the basis generators (in configurations or momentum rep.) since these give
rise to the conserved observables (i. e. the one-particle operators) of QFT, associated
to the de Sitter isometries.
It is remarkable that the particle and antiparticle sectors of these operators bring
additive contributions since the particle and antiparticle operators transform alike under
isometries just as in special relativity.
Notice that this result was obtained by Nachtmann [16] for the scalar UIRs but this is less
relevant as long as the generators of the scalar rep. depend only on m2
. Now we see
that the generators of the spinor rep. which have spin terms depending on m preserve
this property such that we can conclude that all the one-particle operators corresponding
to the de Sitter isometries are additive, regardless the spin.
The principal problem that remains unsolved here is how to build on the de Sitter
manifolds a Wigner type theory able to define the structure of the covariant fields without
using field equations.
62

Appendix
A: Finite-dimensional reps. of the sl(2, C) algebra
The standard basis of the sl(2, C) algebra is formed by the generators J = (J1, J2, J3) and
K = (K1, K2, K3) that satisfy [30, 8]
[Ji, Jj] = iεijkJk , [Ji, Kj] = iεijkKk , [Ki, Kj] = −iεijkJk , (140)
having the Casimir operators c1 = iJ · K and c2 = J2
− K2
. The linear combinations Ai =
1
2 (Ji + iKi) and Bi = 1
2 (Ji − iKi) form two independent su(2) algebras satisfying
[Ai, Aj] = iεijkAk , [Bi, Bj] = iεijkBk , [Ai, Bj] = 0 . (141)
Consequently, any ﬁnite-dimensional irreducible rep. (IR) τ = (j1, j2) is carried by the space of
the direct product (j1)⊗(j2) of the UIRs (j1) and (j2) of the su(2) algebras (Ai) and respectively
(Bi). These IRs are labeled either by the su(2) labels (j1, j2) or giving the values of the Casimir
operators c1 = j1(j1 + 1) − j2(j2 + 1) and c2 = 2[j1(j1 + 1) + j2(j2 + 1)].
The fundamental reps. deﬁning the sl(2, C) algebra are either the IR (1
2, 0) generated by
{1
2σi, −i
2σi} or the IR (0, 1
2) whose generators are {1
2σi, i
2σi}. Their direct sum form the spinor
63

IR ρs = (1
2, 0) ⊕ (0, 1
2) of the Dirac theory. In applications it is convenient to consider ρs as the
fundamental rep. since here invariant forms can be defined using the Dirac conjugation.
The spin basis of the IR τ can be constructed as the direct product,
|τ, sσ =
λ1+λ2=σ
Csσ
j1λ1,j2λ2
|j1, λ1 ⊗ |j2, λ2 , (142)
of su(2) canonical bases where the Clebsh-Gordan coefficients [8] give the spin content of the
IR τ, i. e. s = j1 + j2, j1 + j2 − 1, ..., |j1 − j2|. Note that for integer values of spin we can resort
to the tensor bases constructed as direct products of the vector bases {e1, e2, e3} of the vector
IR (j = 1) (which satisfy |1, ±1 = 1√
2
(e1 ± ie2) and |1, 0 = e3).
Given the IR τ = (j1, j2) we say that its adjoint IR is ˙τ = (j2, j1) and observe that these have
the same spin content while their generators are related as J( ˙τ)
= J(τ)
and K( ˙τ)
= −K(τ)
. On
the other hand, the operators A and B are Hermitian since we use UIRs of the su(2) algebra.
Consequently, we have J+
= J and K+
= −K for any finite-dimensional IR of the sl(2, C)
algebra, such that we can write
(J(τ)
)+
= J(τ)
, (K(τ)
)+
= K( ˙τ)
. (143)
64

Hereby we conclude that invariant forms can be constructed only when we use reducible reps.
ρ = · · · τ1 ⊕ τ2 · · · ˙τ1 ⊕ ˙τ2 · · · containing only pairs of adjoint reps.. Then the matrix γ(ρ) may
be constructed with the matrix elements
τ1, s1σ1|γ(ρ)|τ2, s2σ2 = δτ1 ˙τ2δs1s2δσ1σ2 . (144)
Note that the canonical basis {τ, jλ } deﬁnes the chiral rep. while a new basis in which γ(ρ)
becomes diagonal gives the so called standard rep.. This terminology comes from the Dirac
theory where γ(ρs) = γ0
is the Dirac matrix that may have these reps. [27].
B: Some properties of Hankel functions
According to the general properties of the Hankel functions [31], we deduce that those used here,
H
(1,2)
ν± (z), with ν± = 1
2 ± iµ and z ∈ R, are related among themselves through [H
(1,2)
ν± (z)]∗
=
H
(2,1)
ν (z) and satisfy the identities
e±πk
H(1)
ν (z)H(2)
ν±
(z) + e πk
H(1)
ν±
(z)H(2)
ν (z) =
4
πz
. (145)
65

References
[1] R. M. Wald, General Relativity, (Univ. of Chicago Press, Chicago and London, 1984)
[2] H. B. Lawson Jr. and M.-L. Michaelson, Spin Geometry (Princeton Univ. Press. Princeton,
1989).
[3] I. I. Cot˘aescu, J. Phys. A: Math. Gen. 33, 9177 (2000).
[4] I. I. Cot˘aescu, Europhys. Lett. 86, 20003 (2009).
[5] I. I. Cot˘aescu, Mod. Phys. Lett. A 28 (2013) 13500033-1
[6] E. Wigner, Ann. Math. 40 149 (1939).
[7] G. Mackey, Ann. Math. 44, 101 (1942).
[8] W.-K. Tung, Group Theory in Physics (World Sci., Philadelphia, 1984).
[9] J. Dixmier, Bull. Soc. Math. France 89, 9 (1961).
[10] B. Takahashi, Bull. Soc. Math. France 91, 289 (1963).
[11] B. Allen and T. Jacobson, Commun. Math. Phys. 103, 669 (1986).
[12] N. C. Tsamis and R. P. Woodard, J. Math. Phys. 48, 052306 (2007).
[13] J.-P. Gazeau and M.V. Takook, J. Math. Phys. 41, 5920 (2000) 5920. Gaz
66

[14] P. Bartesaghi, J.-P. Gazeau, U. Moschella and M. V. Takook, Class. Quantum. Grav. 18
(2001) 4373.
[15] T. Garidi, J.-P. Gazeau and M. Takook, J.Math.Phys. 44, 3838 (2003).
[16] O. Nachtmann, Commun. Math. Phys. 6, 1 (1967).
[17] I. I. Cot˘aescu Mod. Phys. Lett. A 28, 1350033 (2013).
[18] I. I. Cot˘aescu, Phys. Rev. D 65, 084008 (2002).
[19] I. I. Cot˘aescu, Mod. Phys. Lett. A 26, 1613 (2011).
[20] I. I. Cot˘aescu, GRG 43, 1639 (2011).
[21] I. I. Cot˘aescu and C. Crucean, Phys. Rev. D 87, 044016 (2013).
[22] I. I. Cota˘escu and D.-M. B˘alt˘at¸eanu, Mod. Phis. Lett. A 30, 1550208 (2014).
[23] N. D. Birrel and P. C. W. Davies, Quantum Fields in Curved Space (Cambridge University
Press, Cambridge 1982).
[24] S. Weinberg, Phys. Rev. 133, B1318 (1964)
[25] B. Carter and R. G. McLenaghan, Phys. Rev. D 19, 1093 (1979).
[26] S. Drell and J. D. Bjorken, Relativistic Quantum Fields (Me Graw-Hill Book Co., New York
1965).
67

[27] B. Thaller, The Dirac Equation (Springer Verlag, Berlin Heidelberg, 1992).
[28] I. I. Cot˘aescu and G. Pascu, Mod. Phys. Lett. A 28, 1350160 (2013).
[29] G. B¨orner and H. P. D¨urr, Nuovo Cimenta LXIV A, 669 (1969).
[30] M. A. Naimark, Linear reps. of the Lorentz Group (Pergamon Press, Oxford, 1964).
[31] I. S. Gradshtein and I. M. Ryzhik, Table of Integrals, Series and Products (Academic
Press Inc., San Diego 1980).
68

I. Cotaescu - "Canonical quantization of the covariant fields: the Dirac field on the de Sitter spacetime"

More Related Content

What's hot (20)

Viewers also liked (17)

Similar to I. Cotaescu - "Canonical quantization of the covariant fields: the Dirac field on the de Sitter spacetime" (20)

More from SEENET-MTP (20)

Recently uploaded (20)

I. Cotaescu - "Canonical quantization of the covariant fields: the Dirac field on the de Sitter spacetime"