arXiv:math-ph/0108022v1 26 Aug 2001

Clifford Algebras and Lorentz Group

V. V. Varlamov

Abstract
Finitedimensional representations of the proper orthochronous Lorentz group
are studied in terms of spinor representations of the Cliord algebras. The Cliord
algebras are understood as an algebraic covering of a full system of the nite
dimensional representations of the Lorentz group. Spacetime discrete symmetries
P , T and P T , represented by fundamental automorphisms of the Cliord algebras, are dened on all the representation spaces. Real, complex, quaternionic and
octonionic representations of the Lorentz group are considered. Physical elds of
the dierent types are formulated within such representations. The AtiyahBott
Shapiro periodicity is dened on the Lorentz group. It is shown that modulo 2
and modulo 8 periodicities of the Cliord algebras allow to take a new look at the
de BroglieJordan neutrino theory of light and the Gell-MannNeemann eightfold
way in particle physics. On the representation spaces the charge conjugation C is
represented by a pseudoautomorphism of the complex Cliord algebra. Quotient
representations of the Lorentz group are introduced. It is shown that quotient representations are the most suitable for description of the massless physical elds. By
way of example, neutrino eld is described via the simplest quotient representation.
WeylHestenes equations for neutrino eld are given.

Key words: Clifford algebras, Lorentz group, finitedimensional representations, discrete symmetries, AtiyahBottShapiro periodicity, charge conjugation, quotient representations, neutrino field, WeylHestenes equations.
1998 Physics and Astronomy Classification Scheme: 02.10.Tq, 11.30.Er, 11.30.Cp
2000 Mathematics Subject Classification: 15A66, 15A90, 20645

Introduction

Importance of discrete transformations is wellknown, many textbooks on quantum theory began with description of the discrete symmetries, and famous L
udersPauli CP T
Theorem is a keystone of quantum field theory. However, usual practice of definition
of the discrete symmetries from the analysis of relativistic wave equations does not give
a full and consistent theory of the discrete transformations. In the standard approach,
except a well studied case of the spin j = 1/2 (Dirac equation), a situation with the
Department of Mathematics, Siberia State University of Industry, Kirova 42, Novokuznetsk 654007,
Russia

discrete symmetries remains vague for the fields of higher spin j > 1/2. It is obvious
that a main reason of this is an absence of a fully adequate formalism for description of
higherspin fields (all widely accepted higherspin formalisms such as RaritaSchwinger
approach [51], BargmannWigner [6] and GelfandYaglom [27] multispinor theories, and
also JoosWeinberg 2(2j+1)component formalism [36, 66] have many intrinsic contradictions and difficulties). The first attempt of going out from this situation was initiated by
Gelfand, Minlos and Shapiro in 1958 [29]. In the GelfandMinlosShapiro approach the
discrete symmetries are represented by outer involutory automorphisms of the Lorentz
group (there are also other realizations of the discrete symmetries via the outer automorphisms [45, 40, 58]). At present the GelfandMinlosShapiro ideas have been found
further development in the works of Buchbinder, Gitman and Shelepin [14, 30], where
the discrete symmetries are represented by both outer and inner automorphisms of the
Poincare group.
In 1957, Shirokov pointed out [57] that an universal covering of the inhomogeneous
Lorentz group has eight inequivalent realizations. Later on, in the eighties this idea
was applied to a general orthogonal group O(p, q) by Dabrowski [18]. As known, the
orthogonal group O(p, q) of the real space Rp,q is represented by the semidirect product of
a connected component O0 (p, q) and a discrete subgroup {1, P, T, P T }. Further, a double
covering of the orthogonal group O(p, q) is a CliffordLipschitz group Pin(p, q) which
is completely constructed within a Clifford algebra Cp,q . In accordance with squares of
elements of the discrete subgroup (a = P 2 , b = T 2 , c = (P T )2 ) there exist eight double
coverings (Dabrowski groups [18]) of the orthogonal group defining by the signatures
(a, b, c), where a, b, c {, +}. Such in brief is a standard description scheme of the
discrete transformations. However, in this scheme there is one essential flaw. Namely, the
CliffordLipschitz group is an intrinsic notion of the algebra Cp,q (a set of the all invertible
elements of Cp,q ), whereas the discrete subgroup is introduced into the standard scheme
in an external way, and the choice of the signature (a, b, c) of the discrete subgroup is not
determined by the signature of the space Rp,q . Moreover, it is suggest by default that
for any signature (p, q) of the vector space there exist the all eight kinds of the discrete
subgroups. It is obvious that a consistent description of the double coverings of O(p, q)
in terms of the CliffordLipschitz groups Pin(p, q) Cp,q can be obtained only in the
case when the discrete subgroup {1, P, T, P T } is also defined within the algebra Cp,q .
Such a description has been given in the works [60, 61, 62], where the discrete symmetries
are represented by fundamental automorphisms of the Clifford algebras. Moreover, this
description allows to incorporate the GelfandMinlosShapiro automorphism theory into
ShirokovDabrowski scheme and further to unite them on the basis of the Clifford algebras
theory.
In the present paper such an unification is given. First of all, Clifford algebras are
understood as algebraic coverings of finitedimensional representations of the proper
Lorentz group G+ . In the section 2 Clifford algebras Cn over the field F = C are associated
with complex finitedimensional representations C of the group G+ . It allows to define
a new class of the finitedimensional representations of G+ (quotient representations)
corresponded to the type n 1 (mod 2) of the algebras Cn . In its turn, representation
spaces of C are the spinspaces S2n/2 or the minimal left ideals of the algebras Cn . In virtue
of this in the section 3 the discrete symmetries representing by spinor representations

of the fundamental automorphisms of Cn are defined for both complex and real finite
dimensional representations of the group G+ .
A full system M = M+ M of the finitedimensional representations of the group G+
allows to define in the section 4 the AtiyahBottShapiro periodicity [4] on the Lorentz
group. In case of the field F = C we have modulo 2 periodicity on the representations
C, C C, that allows to take a new look at the de BroglieJordan neutrino theory of
light [11, 37]. In its turn, over the field F = R we have on the system M the modulo
8 periodicity which relates with octonionic representations of the Lorentz group and the
G
unaydinG
ursey construction of the quark structure in terms of an octonion algebra O
[31]. In essence, the modulo 8 periodicity on the system M gives an another realization
of the wellknown Gell-MannNeemann eightfold way [26]. It should be noted here that
a first attempt in this direction was initiated by Coquereaux in 1982 [17].
Other important discrete symmetry is the charge conjugation C. In contrast with the
transformations P , T , P T the operation C is not spacetime discrete symmetry. This
transformation is firstly appearred on the representation spaces of the Lorentz group
and its nature is strongly different from other discrete symmetries. By this reason in the
section 5 the charge conjugation C is represented by a pseudoautomorphism A A which
is not fundamental automorphism of the Clifford algebra. All spinor representations of
the pseudoautomorphism A A are given in Theorem 3.
Quotient representations of the group G+ compose the second half M of the full
system M and correspond to the types n 1 (mod 2) (F = C) and p q 1, 5 (mod 8)
(F = R). An explicit form of the quotient representations is given in the section 6 (Theorem 4). In the section 7 the first simplest physical field (neutrino field), corresponded
to a fundamental representation C1,0 of the group G+ , is studied within a quotient repre 0,1
sentation C1,0
. Such a description of the neutrino was firstly given in the work
c Cc
[60], but in [60] this description looks like an exotic case, whereas in the present paper
it is a direct consequence of all mathematical background developed in the previous sections. It is shown also that the neutrino field (1/2, 0) (0, 1/2) can be defined in terms
of a DiracHestenes spinor field [32, 33], and the wave function of this field satisfies the
WeylHestenes equations (massless DiracHestenes equations).

Finite-dimensional representations of the Lorentz

group and complex Clifford algebras

It is wellknown [56, 23, 7] that representations of the Lorentz group play a fundamental role in the quantum field theory. Physical fields are defined in terms of finite
dimensional irreducible representations of the Lorentz group O(1, 3) O(3, 1) (correspondingly, Poincare group O(1, 3) T (4), where T (4) is a subgroup of fourdimensional
translations). It should be noted that in accordance with [47] any finitedimensional
irreducible representation of the proper Lorentz group G+ = O0 (1, 3) O0 (3, 1)
SL(2; C)/Z2 is equivalent to some spinor representation. Moreover, spinor representations exhaust in essence all the finitedimensional representations of the group G+ . This
fact we will widely use below.
Let us consider in brief the basic facts concerning the theory of spinor representa-

tions of the Lorentz group. The initial point of this theory is a correspondence between
transformations of the proper Lorentz group and complex matrices of the second order.
Indeed, follows to [29] let us compare the Hermitian matrix of the second order


x0 + x3 x1 ix2
X=
(1)
x1 + ix2 x0 x3
to the vector v of the Minkowski spacetime R1,3 with coordinates x0 , x1 , x2 , x3 . At this
point det X = x20 x21 x22 x23 = S 2 (x). The correspondence between matrices X and
vectors v is onetoone and linear. Any linear transformation X = aXa in a space of the
matrices X may be considered as a linear transformation ga in R1,3 , where a is a complex
matrix of the second
order
with det a = 1. The correspondence a ga possesses following


1 0
properties: 1)
e (identity element); 2) ga1 ga2 = ga1 a2 (composition); 3) two
0 1
different matrices a1 and a2 correspond to one and the same transformation ga1 = ga2 only
in the case a1 = a2 . Since every complex matrix is defined by eight real numbers, then
from the requirement det a = 1 it follow two conditions Re det a = 1 and Im det a = 0.
These conditions leave six independent parameters, that coincides with parameter number
of the proper Lorentz group.
Further, a set of all complex matrices of the second order forms a full matrix algebra
M2 (C) that is isomorphic to a biquaternion algebra C2 . In its turn, Pauli matrices








1 0
0 i
0 1
1 0
.
(2)
, 3 =
, 2 =
, 1 =
0 =
0 1
i 0
1 0
0 1
form one from a great number of isomorphic spinbasis of the algebra C2 (by this reason
in physics the algebra C2 C+
1,3 C3,0 is called Pauli algebra). Using the basis (2) we
can write the matrix (1) in the form
X = x .

(3)

The Hermitian matrix (3) is correspond to a spintensor (1, 1) X with following coordinates

x0 = +(1/ 2)( 1 1 + 2 2),

x1 = +(1/ 2)( 1 2 + 2 1 ),

x3 = +(1/ 2)( 1 1 2 2 ),
(4)
x2 = (i/ 2)( 1 2 2 1),
where and are correspondingly coordinates of spinors and cospinors of spinspaces
S2 and S 2 . Linear transformations of vectors (spinors and cospinors) of the spinspaces
S2 and S 2 have the form
1

= 1 + 2 ,
2
= 1 + 2 ,



=

2 ,
=
1 +

2
2 ,
= 
1 +

.
=

Transformations (5) form the group SL(2; C), since M2 (C) and







SL(2; C) =
C2 : det
= 1 Spin+ (1, 3).

(5)

The expressions (4) and (5) compose a base of the 2spinor van der Waerden formalism
[64, 54], in which the spaces S2 and S 2 are called correspondingly spaces of undotted
and dotted spinors. The each of the spaces S2 and S 2 is homeomorphic to an extended
complex plane C representing an absolute (the set of infinitely distant points) of a
Lobatchevskii space S 1,2 . At this point a group of fractional linear transformations of the
plane C is isomorphic to a motion group of S 1,2 [53]. Besides, in accordance with [39]
the Lobatchevskii space S 1,2 is an absolute of the Minkowski world R1,3 and, therefore,
the group of fractional linear transformations of the plane C (motion group of S 1,2 )
twice covers a rotation group of the spacetime R1,3 , that is the proper Lorentz group.
Theorem 1. Let C2 be a biquaternion algebra and let i be a canonical spinor representations (Pauli matrices) of the units of C2 , then 2k tensor products of the k matrices
i form a basis of the full matrix algebra M2k (C), which is a spinor representation of a
complex Clifford algebra C2k . The set containing 2k + 1 tensor products of the k matrices
i is homomorphically mapped onto a set consisting of the same 2k tensor products and
forming a basis of the spinor representation of a quotient algebra C2k .
Proof. As a basis of the spinor representation of the algebra C2 we take the Pauli matrices
(2). This choice is explained by physical applications only (from mathematical viewpoint
the choice of the spinbasis for C2 is not important). Let us compose now 2k 2k dimensional
matrices:
E1
E2
E3
....
Ek
Ek+1
Ek+2
....
E2k

=
=
=
...
=
=
=
...
=

1 0 0 0 0 ,
3 1 0 0 0 ,
3 3 1 0 0 ,
.....................
3 3 3 1 ,
2 0 0 ,
3 2 0 0 ,
.....................
3 3 3 2 .

(6)

Since i2 = 0 , then for a square of any matrix from the set (6) we have
E2i = 0 0 0 ,

i = 1, 2, . . . , 2k,

(7)

where the product 0 0 0 equals to 2k dimensional unit matrix. Further,

Ei Ej = Ej Ei,

i < j; i, j = 1, 2, . . . 2k.

(8)

Indeed, when i = 1 and j = 3 we obtain

E1 E3 = 1 3 3 1 0 0 ,
E3 E1 = 3 1 3 1 0 0 .
Thus, the equalities (7) and (8) show that the matrices of the set (6) satisfy the multiplication law of the Clifford algebra. Moreover, let us show that a set of the matrices

E1 1 E2 2 . . . E2k2k , where each of the indices 1 , 2 , . . . , 2k takes either of the two values 0
or 1, consists of 22k matrices. At this point these matrices form a basis of the full 2k
dimensional matrix algebra (spinor representation of C2k ). Indeed, in virtue of i1 2 = 3
from (6) it follows
Nj = Ej Ek+j = 0 0 3 0 0 ,
j = 1, 2, . . . k,

(9)

here the matrix 3 occurs in the jth position. Further, since the tensor product 0
0 of the unit matrices of the second order is also unit matrix E0 of the 2k order,
then we can write
1
(E0 Nj ) = 0 0 Q++ 0 0 ,
2
1
(E0 + Nj ) = 0 0 Q 0 0 ,
=
2

Zj++ =
Zj

where the matrices Q++ and Q occur in the jth position and have the following form




0 0
1 0
.
,
0 1
0 0
From (9) it follows
N1 N2 . . . Nj = 3 3 3 0 0 ,
where on the left side we have j matrices 3 . In virtue of this equality we obtain
Lj = N1 . . . Nj1 Ej = 0 0 1 0 0 ,
Lk+j = N1 . . . Nj1 Ek+j = 0 0 2 0 0 .
here the matrices 1 and 2 occur in the jth position, j = 1, 2, . . . , k. Thus,
1
(Lj + iLk+j ) = 0 0 Q+ 0 0 ,
2
1
(Lj iLk+j ) = 0 0 Q+ 0 0 ,
=
2

Zj+ =
Zj+

where the matrices Q+ and Q+ also occur in the jth position and have correspondingly
the following form




0 0
0 1
.
,
1 0
0 0
Therefore, a matrix
k
Y
s r
(Zj j j ) = Qs1 r1 Qs2 r2 Qsk rk ,
j

sj = , rj =

(10)

has unit elements at the intersection of (s1 , s2 , . . . , sk )th row and (r1 , r2 , . . . , rk )th column, the other elements are equal to zero. In virtue of (7) and (8) each of the matris r
ces Lj , Lk+j and, therefore, each of the matrices Zj j j , is represented by a linear comQ
s r
bination of the matrices E1 1 E2 2 E2k2k . Hence it follows that the matrices kj (Zj j j )
and, therefore, all the 2k dimensional matrices, are represented by such the linear combinations. Thus, 2k matrices E1 , . . . , E2k generate a group consisting of the products
E1 1 E2 2 E2k2k , and an enveloped algebra of this group is a full 2k dimensional matrix
algebra.
The following part of this Theorem tells that the full matrix algebra, forming by the
tensor products (6), is a spinor representation of the algebra C2k . Let us prove this part
on the several examples. First of all, in accordance with (6) tensor products of the k = 2
order are

0
0
1
0

 

0 0 0 1
1 0
0
1

E1 = 1 0 =
1 0 0 0 ,
0
1
1 0
0 1 0 0

0
1
0
0

 

1 0 0
0
0 1
1 0
,
=

E2 = 3 1 =

0 0 0 1
1 0
0 1
0 0 1 0

0
0
i
0

 

0 0 0 i
0 i
1 0

E3 = 2 0 =

=
i 0 0 0 ,
i 0
0 1
0 i 0 0

0
i
0
0

 

i 0 0 0
1 0
0 i

(11)
E4 = 3 2 =

=
0 0 0 i .
0 1
i 0
0 0 i 0

It is easy to see that E2i = E0 and Ei Ej = Ej Ei (i, j = 1, . . . , 4). The set of the matrices
E1 1 E2 2 E4 4 , where each of the indices 1 , 2 , 3 , 4 takes either of the two values 0
or 1, consists of 24 = 16 matrices. At this point these matrices form a basis of the full
4dimensional matrix algebra. Thus, we can define a onetoone correspondence between
sixteen matrices E1 1 E2 2 E4 4 and sixteen basis elements ei1 ei2 . . . eik of the Dirac algebra
C4 . Therefore, the matrices (11) form a basis of the spinor representation of C4 . Moreover,
from (11) it follows that C4 C2 C2 , that is, the Dirac algebra is a tensor product of
the two Pauli algebras.
Analogously, when k = 3 from (6) we obtain following products (matrices of the eighth
order):
E1 = 1 0 0 ,
E4 = 2 0 0 ,

E2 = 3 1 0 ,
E5 = 3 2 0 ,

E3 = 3 3 1 ,
E6 = 3 3 2 ,

The set of the matrices E1 1 E2 2 E6 6 , consisting of 26 = 64 matrices, forms a basis of

the full 8dimensional matrix algebra, which is isomorphic to a spinor representation of

the algebra C6 C2 C2 C2 . Generalizing we obtain that 2k tensor products of the k

Pauli matrices form a basis of the spinor representation of the complex Clifford algebra
C2k C2 C2 C2 .
{z
}
|

(12)

E2k+1 = 3 3 3 ,
|
{z
}

(13)

k times

Let us consider now a case of odd dimensions. When n = 2k + 1 we add to the set of
the tensor products (6) a matrix

k times

which satisfying the conditions

E22k+1 = E0 , E2k+1 Ei = Ei E2k+1 ,

i = 1, 2, . . . , k.
It is obvious that a product E1 E2 E2k E2k+1 commutes with all the products of the form
E1 1 E2 2 E2k2k . Further, let
(
1 , if k 0 (mod 2),
=
i , if k 1 (mod 2).
Then a product
U = E1 E2 E2k E2k+1

2k+1
,
satisfies the condition U2 = E0 . Let P be a set of all 22k+1 matrices E1 1 E2 2 E2k2k E2k+1
where j equals to 0 or 1, j = 1, 2, . . . , 2k + 1. Let us divide the set P into two subsets
by the following manner:

P = P1 + P0 ,

(14)

where the subset P0 contains products with the matrix E2k+1 , and P1 contains products
without the matrix E2k+1 . Therefore, products E1 E2 2 E2k2k P1 form a full 2k
dimensional matrix algebra. Further, when we multiply the matrices from the subset P0
by the matrix U the factors E2k+1 are mutually annihilate. Thus, matrices of the set UP0
are also belong to the 2k dimensional matrix algebra. Let us denote UP0 via P2 . Taking
into account that U2 = E0 we obtain P0 = UP2 and
P = P1 + P2 ,
where P1 , P2 M2k . Let
: P1 + UP2 P1 + P2

(15)

be an homomorphic mapping of the set (14) containing the matrix E2k+1 onto the set
P1 + P2 which does not contain this matrix. The mapping preserves addition and
multiplication operations. Indeed, let P = P1 + UP2 and Q = Q1 + UQ2 , then
P + Q = P1 + UP2 + Q1 + UQ2 P1 + P2 + Q1 + Q2 ,

PQ = (P1 + UP2 )(Q1 + UQ2 ) = (P1 Q1 + P2 Q2 ) + U(P1 Q2 + P2 Q1 )

(P1 Q1 + P2 Q2 ) + (P1 Q2 + P2 Q1 ) = (P1 + P2 )(Q1 + Q2 ),

(16)

that is, an image of the products equals to the product of factor images in the same order.
In particular, when P = U we obtain P1 = 0, P2 = E0 and
U E0 .

(17)

In such a way, at the mapping all the matrices of the form E1 1 E2 2 E2k2k
UE1 1 E2 2 E2k2k are mapped into zero. Therefore, a kernel of the homomorphism is
defined by an expression Ker = {P1 UP1 }. We obtain the homomorphic mapping of
2k+1
onto the full matrix algebra M2k .
the set of all the 22k+1 matrices E1 1 E2 2 E2k2k E2k+1
In the result of this mapping we have a quotient algebra M2k P/ Ker . As noted
previously, 2k tensor products of the k Pauli matrices (or other k matrices defining spinor
representation of the biquaternion algebra C2 ) form a basis of the spinor representation
of the algebra C2k . It is easy to see that there exists onetoone correspondence between 22k+1 matrices of the set P and basis elements of the odddimensional Clifford
algebra C2k+1 . It is wellknown [52, 49] that C2k+1 is isomorphic to a direct sum of
two evendimensional subalgebras: C2k+1 C2k C2k . Moreover, there exists an homomorphic mapping : C2k+1 C2k , in the result of which we have a quotient algebra

C2k C2k+1 / Ker , where Ker = {A1 A1} is a kernel of the homomorphism , A1
is an arbitrary element of the subalgebra C2k , is a volume element of the algebra C2k+1 .
It is easy to see that the homomorphisms and have a similar structure. Thus, hence
it immediately follows an isomorphism C2k M2k . Therefore, a basis of the matrix
quotient algebra M2k is also a basis of the spinor representation of the Clifford quotient
algebra C2k , that proves the latter assertion of the theorem.
Let us consider now spintensor representations of the proper Lorentz group O0 (1, 3)
SL(2; C)/Z2 Spin+ (1, 3)/Z2 and their relations with the complex Clifford algebras.
From each complex Clifford algebra Cn = C Cp,q (n = p + q) we obtain a spinspace
S2n/2 , which is a complexification of the minimal left ideal of the algebra Cp,q : S2n/2 =
C Ip,q = C Cp,q epq , where epq is a primitive idempotent of the algebra Cp,q . Further,
a spinspace corresponding the Pauli algebra C2 has a form S2 = C I2,0 = C C2,0e20
or S2 = C I1,1 = C C1,1 e11 (C I0,2 = C C0,2 e02 ). Therefore, the tensor product
(12) of the k algebras C2 induces a tensor product of the k spinspaces S2 :
S2 S2 S2 = S2k .

Vectors of the spinspace S2k (or elements of the minimal left ideal of C2k ) are spintensors
of the following form
X
1 2 k =
1 2 k ,
(18)

where summation is produced on all the index collections (1 . . . k ), i = 1, 2. In virtue

of (5) for each spinor i from (18) we have a transformation rule i = ii i . Therefore,
in general case we obtain
X

1 2 k
(19)

=
11 22 kk 1 2 k .

A representation (19) is called undotted spintensor representation of the proper Lorentz

group of the rank k.

Further, let C2 be a biquaternion algebra, the coefficients of which are complex con

jugate. Let us show that the algebra C2 is obtained from C2 under action of the autoe Indeed, in virtue of an isomorphism
morphism A A or antiautomorphism A A.
C2 C3,0 a general element
0

A = a e0 +

3
X

a ei +

i=1

3 X
3
X

aij eij + a123 e123

i=1 j=1

of the algebra C3,0 can be written in the form

A = (a0 + a123 )e0 + (a1 + a23 )e1 + (a2 + a31 )e2 + (a3 + a12 )e3 ,

(20)

where = e123 . Since belongs to a center of the algebra C3,0 (commutes with all
the basis elements) and 2 = 1, then we can to suppose i. The action of the
automorphism on the homogeneous element A of a degree k is defined by a formula
A = (1)k A. In accordance with this the action of the automorphism A A , where
A is the element (20), has a form
A A = (a0 a123 )e0 (a1 a23 )e1 (a2 a31 )e2 (a3 a12 )e3 .

(21)

e
Therefore, : C2 C2 . Correspondingly, the action of the antiautomorphism A A
k(k1)
e = (1) 2 A.
on the homogeneous element A of a degree k is defined by a formula A
Thus, for the element (20) we obtain
e = (a0 a123 )e0 + (a1 a23 )e1 + (a2 a31 )e2 + (a3 a12 )e3 ,
A A

(22)

that is, e : C2 C2 . This allows to define an algebraic analog of the Wigners rep

resentation doubling: C2 C2 . Further, from (20) it follows that A = A1 + A2 =

(a0 e0 + a1 e1 + a2 e2 + a3 e3 ) + (a123 e0 + a23 e1 + a31 e2 + a12 e3 ). In general case, by virtue
of an isomorphism C2k Cp,q , where Cp,q is a real Clifford algebra with a division
ring K C, p q 3, 7 (mod 8), we have for a general element of Cp,q an expression
A = A1 + A2 , here 2 = e212...p+q = 1 and, therefore, i. Thus, from C2k under
action of the automorphism A A we obtain a general algebraic doubling

C2k C2k .

(23)

Correspondingly, a tensor product C2 C2 C2 C2r of r algebras C2 induces

a tensor product of r spinspaces S 2 :
S 2 S 2 S 2 = S 2r .
Te vectors of the spinspace S 2r have the form
X
1 2 r =
1 2 r ,

(24)

where the each cospinor i from (24) in virtue of (5) is transformed by the rule i =

ii i . Therefore,
X

1 2 r

=
11 22 rr 1 2 r .
(25)
A representation (25) is called a dotted spintensor representation of the proper Lorentz
group of the rank r.

In general case we have a tensor product of k algebras C2 and r algebras C2 :

C2 C2 C2 C2 C2 C2 C2k C2r ,
which induces a spinspace
S2 S2 S2 S 2 S 2 S 2 = S2k+r
with the vectors
1 2 k 1 2 r =

1 2 k 1 2 r .

(26)

In this case we have a natural unification of the representations (19) and (25):
X


1 2 k 1 2 r
=
11 22 kk 11 22 rr 1 2 k 1 2 r .

(27)

So, a representation (27) is called a spintensor representation of the proper Lorentz group
of the rank (k, r).
In general case, the representations defining by the formulas (19), (25) and (27), are
reducible, that is there exists possibility of decomposition of the initial spinspace S2k+r
(correspondingly, spinspaces S2k and S2r ) into a direct sum of invariant (with respect to
transformations of the group G+ ) spinspaces S21 S22 S2s , where 1 +2 +. . .+s =
k + r.
Further, an important notion of the physical field is closely related with finite
dimensional representations of the proper Lorentz group G+ . In accordance with
Wigner interpretation [67], an elementary particle is described by some irreducible
finitedimensional representation of the Poincare group. The double covering of the
proper Poincare group is isomorphic to a semidirect product SL(2; C) T (4), or
Spin+ (1, 3) T (4), where T (4) is the subgroup of fourdimensional translations. Let
(x) be a physical field, then at the transformations (a, ) of the proper Poincare group
the field (x) is transformed by a following rule

m
u(x) =

C () (1 (x a)),

(28)

where a T (4), G+ , is a Lorentz transformation, and C is a representation of

the group G+ in the space S2k+r . Since the group T (4) is Abelian, then all its representations are onedimensional. Thus, all the finitedimensional representations of the proper
Poincare group in essence are equivalent to the representations C of the group G+ . If

the representation C is reducible, then the space S2k+r is decomposed into a direct sum
of irreducible subspaces, that is, it is possible to choose in S2k+r such a basis, in which
all the matrices C take a blockdiagonal form. Then the field (x) is reduced to some
number of the fields corresponding to obtained irreducible representations of the group
G+ , each of which is transformed independently from the other, and the field (x) in
this case is a collection of the fields with more simple structure. It is obvious that these
more simple fields correspond to irreducible representations C. As known [47, 29, 56], a
system of irreducible finitedimensional representations of the group G+ is realized in the
space Sym(k,r) S2k+r of symmetric spintensors. The dimensionality of Sym(k,r) is equal
to (k + 1)(r + 1). A representation of the group G+ by such spintensors is irreducible and

denoted by the symbol Cj,j , where 2j = k, 2j = r, numbers j and j defining the spin
are integer or halfinteger. Then the field (x) transforming by the formula (28) is, in
general case, a field of the type (j, j ). In such a way, all the physical fields are reduced
to the fields of this type, the mathematical structure of which requires a knowledge of

representation matrices Cj,j

. As a rule, in physics there are two basic types of the fields:
1) The field of type (j, 0). The structure of this field (or the field (0, j)) is described by the

representation Cj,0 (C0,j ), which is realized in the space Sym(k,0) S2k (Sym(0,r) S2r ).
At this point in accordance with Theorem 1 the algebra C2k C2 C2 C2

(correspondingly, C2 C2 C2 C2 ) is associated with the field of type (j, 0)

(correspondingly, (0, j )) The trivial case j = 0 corresponds to a PauliWeisskopf field
describing the scalar particles. In other particular case, when j = j = 1/2 we have a
Weyl field describing the neutrino. At this point the antineutrino is described by a fundamental representation C1/2,0 = of the group G+ and the algebra C2 related with this
representation (Theorem 1). Correspondingly, the neutrino is described by a conjugated

representation C0,1/2 and the algebra C2 . In relation with this, it is hardly too much to
say that the neutrino field is a more fundamental physical field, that is a kind of the basic
building block, from which other physical fields built by means of direct sum or tensor
product.
2) The field of type (j, 0) (0, j). The structure of this field admits a space inversion and,
therefore, in accordance with a Wigners doubling [68] is described by a representation
Cj,0 C0,j of the group G+ . This representation is realized in the space Sym(k,k) S22k .
In accordance with (23) the Clifford algebra related with this representation is a direct

sum C2k C2k C2 C2 C2 C2 C2 C2 . In the simplest case j = 1/2 we

have bispinor (electronpositron) Dirac field (1/2, 0) (0, 1/2) with the algebra C2 C2 .
It should be noted that the Dirac algebra C4 considered as a tensor product C2 C2 (or

C2 C2 ) in accordance with (18) (or (26)) gives rise to spintensors 1 2 (or 1 1 ), but it
contradicts with the usual definition of the Dirac bispinor as a pair ( 1 , 1 ). Therefore,

the Clifford algebra associated with the Dirac field is C2 C2 , and a spinspace of this
sum in virtue of unique decomposition S2 S 2 = S4 (S4 is a spinspace of C4 ) allows to
define matrices in the Weyl basis. The case j = 1 corresponds to Maxwell fields (1, 0)

and (0, 1) with the algebras C2 C2 and C2 C2 . At this point the electromagnetic field

is defined by complex linear combinations F = E iH, F = E + iH (Helmholtz representation). Besides, the algebra related with Maxwell field is a tensor product of the two
algebras C2 describing the neutrino fields. In this connection it is of interest to recall a
neutrino theory of light was proposed by de Broglie and Jordan [11, 37]. In the de Broglie
Jordan neutrino theory of light electromagnetic field is constructed from the two neutrino
fields (for more details and related papers see [22]). Traditionally, physicists attempt to
describe electromagnetic field in the framework of (1, 0) (0, 1) representation (see old
works [41, 54, 56] and recent developments based on the JoosWeinberg formalism [36, 66]
and its relation with a BargmannWightmanWigner type quantum field theory [3, 20]).
However, Weinbergs equations (or Weinberglike equations) for electromagnetic field obtained within the subspace Sym(k,r) with dimension 2(2j + 1) have acausal (tachionic)
solutions [2]. Electromagnetic field in terms of a quotient representation (1, 0) (0, 1) in
the full representation space S2k+r will be considered in separate paper.
In this connection it should be noted two important circumstances related with irreducible representations of the group G+ and complex Clifford algebras associated with
these representations. The first circumstance relates with the Wigner interpretation of
an elementary particle. Namely, a relation between finitedimensional representations of
the proper Lorentz group and comlex Clifford algebras (Theorem 1) allows to essentially
extend the Wigner interpretation by means of the use of an extraordinary rich and universal structure of the Clifford algebras at the study of spacetime (and also intrinsic)
symmetries of elementary particles. The second circumstance relates with the spin. Usually, the Clifford algebra is associated with a halfinteger spin corresponding to fermionic
fields, socalled matter fields, whilst the fields with an integer spin (bosonic fields) are
eliminated from an algebraic description. However, such a nonsymmetric situation is
invalid, since the fields with integer spin have a natural description in terms of spintensor

representations of the proper Lorentz group with even rank and algebras C2k and C2k
associated with these representations, where k is even (for example, Maxwell field). In
this connection it should be noted that generalized statistics in terms of Clifford algebras
have been recently proposed by Finkelstein and collaborators [25, 8].
As known, complex Clifford algebras Cn are modulo 2 periodic [4] and, therefore, there
exist two types of Cn : n 0 (mod 2) and n 1 (mod 2). Let us consider these two types
in the form of following series:
C2

C4
C3

C5

C2k

C2k+1

Let us consider the decomposition C2k+1 C2k C2k in more details. This decomposition
may be represented by a following scheme
C2k+1

J
J


J
^

C2k

C2k

Here central idempotents

+ =

1 + e1 e2 e2k+1
,
2

where

1 e1 e2 e2k+1
,
2

(
1, if k 0 (mod 2),
=
i, if k 1 (mod 2)

satisfy the relations (+ )2 = + , ( )2 = , + = 0. Thus, we have a decomposition

of the initial algebra C2k+1 into a direct sum of two mutually annihilating simple ideals:
C2k+1 12 (1 + )C2k+1 21 (1 )C2k+1. Each of the ideals C2k+1 is isomorphic to the
subalgebra C2k C2k+1 . In accordance with Chisholm and Farwell [16] the idempotents
+ and can be identified with helicity projection operators which distinguish left
and right handed spinors. The ChisholmFarwell notation for we will widely use
below.Therefore, in virtue of the isomorphism C2k+1 C2k C2k and the homomorphic
mapping : C2k+1 C2k the second series (type n 1 (mod 2)) is replaced by a sequence
of the quotient algebras C2k , that is,
C2

C4

C2

C4

C2k

C2k

Representations corresponded these two series of Cn (n 0, 1 (mod 2)) form a full system
M = M0 M1 of finitedimensional representations of the proper Lorentz group G+ .
All the physical fields used in quantum field theory and related representations of

the group G+ Cj,0 (C2k ), Cj,0 C0,j (C2k C2k ) are constructed from the upper series
(type n 0 (mod 2)). Whilst the lower series (type n 1 (mod 2)) is not considered
in physics as yet. In accordance with Theorem 1 we have an isomorphism C2k M2k .
Therefore, the quotient algebra C2k induces a spinspace S2k that is a space of a quotient

representation Cj,0 of the group G+ . Analogously, a quotient representation C0,j is

realised in the space S 2r which induced by the quotient algebra C2r . In general case, we

have a quotient representation Cj,j defined by a tensor product C2k C2r . Thus, the
complex type n 1 (mod 2) corresponds to a full system of irreducible finitedimensional

quotient representations Cj,j of the proper Lorentz group. Therefore, until now in physics
only one half (n 0 (mod 2)) of all possible finitedimensional representations of the
Lorentz group has been used.
Let us consider now a full system of physical fields with different types. First of all,
the field
(j, 0) = (1/2, 0) (1/2, 0) (1/2, 0)

(29)

in accordance with Theorem 1 is a tensor product of the k fields of type (1/2, 0), each of
which corresponds to the fundamental representation C1/2,0 = of the group G+ and the
biquaternion algebra C2 related with fundamental representation. In its turn, the field
(0, j ) = (0, 1/2) (0, 1/2) (0, 1/2)

(30)

is a tensor product of the r fundamental fields of the type (0, 1/2), each of which corre

sponds to a conjugated representation C0,1/2 = and the conjugated algebra C2 obtained

in accordance with (21)(22) under action of the automorphism A A (space invere (time reversal). The numbers
sion), or under action of the antiautomorphism A A

j and j are integer (bosonic fields) if in the products (29) and (30) there are k, r 0
(mod 2) (1/2, 0) (or (0, 1/2)) factors, and the numbers j and j are halfinteger (fermionic
fields) if in the products (29)(30) there are k, r 1 (mod 2) factors. Further, the field
O
(j, j ) = (1/2, 0) (1/2, 0) (1/2, 0)
(0, 1/2) (0, 1/2) (0, 1/2) (31)
is a tensor product of the fields (29) and (30). As consequence of the doubling (23) we
have the field of type (j, 0) (0, j):
M
(j, 0) (0, j) = (1/2, 0) (1/2, 0) (1/2, 0)
(0, 1/2) (0, 1/2) (0, 1/2)

In general, all the fields (29)(31) describe multiparticle states. The decompositions of
these multiparticle states into single states provided in the full representation space S2k+r ,
where Sym(k,r) and Sym(k,k) with dimensions (2j +1)(2j +1) and 2(2j +1) are subspaces of
S2k+r (for example, the ClebshGordan decomposition of two spin 1/2 particles into singlet
and triplet: (1/2, 1/2) = (1/2, 0) (0, 1/2) = (0, 0) (1, 0)). In the papers [34, 19, 59]
a multiparticle state is described in the framework of a tensor product C3,0 C3,0
C3,0. It is easy to see that in virtue of the isomorphism C3,0 C2 the tensor
product of the algebras C3,0 is isomorphic to the product (12). Therefore, the Holland
approach naturally incorporates into a more general scheme considered here. Finally, for

the type n 1 (mod 2) we have quotient representations Cj,j of the group G+ . The
physical fields corresponding to the quotient representations are constructed like the fields
(29)-(31). Due to the decomposition Cn Cn1 Cn1 (n 1 (mod 2)) we have a field
[
(j, 0) (j, 0) = (1/2, 0) (1/2, 0) (1/2, 0) (1/2, 0) (1/2, 0) (1/2, 0),
(32)
and also we have fields (0, j) (0, j) and (j, 0) (0, j) if the quotient algebras Cn1 admit
space inversion or time reversal. The field
[
(j, 0) (0, j) = (1/2, 0) (1/2, 0) (1/2, 0) (0, 1/2) (0, 1/2) (0, 1/2)
(33)
is analogous to the field (j, 0) (0, j), but, in general, the field (j, 0) (0, j) has a quantity
of violated discrete symmetries. An explicit form of the quotient representations and their
relations with discrete symmetries will be explored in the following sections.

Discrete symmetries on the representation spaces

of the Lorentz group

Since all the physical fields are defined in terms of finitedimensional representations of
the group G+ , then a construction of the discrete symmetries (space inversion P , time

reversal T and combination P T ) on the representation spaces of the Lorentz group has a
primary importance.
In the recent paper [60] it has been shown that the space inversion P , time reversal T
and full reflection P T correspond to fundamental automorphisms A A (involution),
e (reversion) and A A
f (conjugation) of the Clifford algebra C. Moreover, there
AA
exists an isomorphism between a discrete subgroup {1, P, T, P T } Z2 Z2 (P 2 = T 2 =
(P T )2 = 1, P T = T P ) of the orthogonal group O(p, q) and an automorphism group
Aut(C) = {Id, , e , e
}:
Id

Id

Id

Id

Id

e
e

PT

PT

PT

PT

Id

PT

PT

(34)

Further, in the case P 2 = T 2 = (P T )2 = 1 and P T = T P there is an isomorphism

between the group {1, P, T, P T } and automorphism group Aut(C) = {I, W, E, C}. Spinor
representations of the fundamental automorphisms of the algebras Cn was first obtained
by Rashevskii in 1955 [52]: 1) Involution: A = WAW1 , where W is a matrix of the
e =
automorphism (matrix representation of the volume element ); 2) Reversion: A
T 1
EA E , where E is a matrix of the antiautomorphism e satisfying the conditions EiE
m(m1)
EETi = 0 and ET = (1) 2 E, here Ei = (ei ) are matrix representations of the units
f = CAT C1 , where C = EWT is a matrix of the
of the algebra C; 3) Conjugation: A
m(m+1)
antiautomorphism e
satisfying the conditions CETi + Ei C = 0 and CT = (1) 2 C.
So, for the Dirac algebra C4 in the canonical basis there exists a standard (Wigner)
representation P = 0 and T = 1 3 [9], therefore, {1, P, T, P T } = {1, 0 , 1 3 , 01 3 }.
On the other hand, the automorphism group of the algebra C4 for basis has a form
Aut(C4 ) = {I, W, E, C} = {I, 01 2 3 , 1 3 , 02 }. In [60] it is shown that {1, P, T, P T } =
{1, 0, 1 3 , 0 1 3 } Aut(C4 ) Z4 , where Z4 is a complex group with the signature
(+, , ).
In general case, according to Theorem 1 a space of the finitedimensional representation of the group SL(2; C) is a spinspace S2k+r , or a minimal left ideal of the algebra

C2k C2r . Therefore, in the spinor representation the fundamental automorphisms of

the algebra C2k C2r (all spinor representations of the fundamental automorphisms have
been found in [61]) by virtue of the isomorphism (34) induce discrete transformations on
the representation spaces (spinspaces) of the Lorentz group.

Theorem 2. 1) The field F = C. The tensor products C2 C2 C2 , C2 C2 C2 ,

C2 C2 C2 C2 C2 C2 of the Pauli algebra C2 correspond to finitedimensional

representations Cl0
+l1 1,0 , C0,l0 l1 +1 , Cl0 +l1
1,l0 l1 +1 of the proper Lorentz
group G+ , where


r r
kr k+r
k k
(l0 , l1 ) = 2 , 2 + 1 , (l0 , l1 ) = 2 , 2 + 1 , (l0 , l1 ) = 2 , 2 + 1 , and spinspaces S2k ,

S2r , S2k+r are representation spaces of the group G+ , C2 C1,0 (C2 C0,1 ) is a

fundamental representation of G+ . Then in a spinor representation of the fundamental

automorphisms of the algebra Cn for the matrix W of the automorphism A A (space
e (time reinversion) and also for the matrices E and C of the antiautomorphisms A A
f (full reflection) the following permutation relations with infinitesimal
versal) and A A
operators of the group G+ take place:
[W, A23 ] = [W, A13 ]= [W, A12 ] = 0,
[W, H+ ] = [W, H ] = [W, H3 ] = 0,

{W, B1 } = {W, B2 } = {W, B3 } = 0

{W, F+ }= {W, F } = {W, F3 } = 0.

(35)
(36)

[E, A23 ] = [E, A13 ]= [E, A12 ] = 0,

[C, A23 ] = [C, A13 ]= [C, A12 ] = 0,
[E, H+ ] = [E, H ] = [E, H3 ] = 0,
[C, H+ ] = [C, H ] = [C, H3 ] = 0,

{E, B1 } = {E, B2 } = {E, B3 }= 0,

[C, B1 ] = [C, B2 ] = [C, B3 ] = 0,
{E, F+ }= {E, F } = {E, F3 } = 0,
[C, F+ ] = [C, F ] = [C, F3 ] = 0.

(37)
(38)
(39)
(40)

[E, A23 ] = [E, A13 ]= [E, A12 ] = 0,

[C, A23 ] = [C, A13 ]= [C, A12 ] = 0,
[E, H+ ] = [E, H ] = [E, H3 ] = 0,
[C, H+ ] = [C, H ] = [C, H3 ] = 0,

[E, B1 ] = [E, B2 ] = [E, B3 ] = 0,

{C, B1 } = {C, B2 ] = {C, B3 ] = 0,
[E, F+ ] = [E, F ] = [E, F3 ] = 0,
{C, F+ }= {C, F } = {C, F3 } = 0.

(41)
(42)
(43)
(44)

[E, A23 ] = 0,
[C, A23 ] = 0,

{E, A13 }= {E, A12 } = 0,

{C, A13 }= {C, A12 } = 0,

[E, B1 ] = 0, {E, B2 } = {E, B3 }= 0,

{C, B1 } = 0, [C, B2 ] = [C, B3 ] = 0.

(45)
(46)

[E, A23 ] = 0,
[C, A23 ] = 0,

{E, A13 }= {E, A12 } = 0,

{C, A13 }= {C, A12 } = 0,

{E, B1 } = 0, [E, B2 ] = [E, B3 ] = 0,

[C, B1 ] = 0, {C, B2 } = [C, B3 } = 0.

(47)
(48)

[E, B1 ] = [E, B2 ] = 0,
{C, B1 } = {C, B2 } = 0,
[E, F+ ] = [E, F ] = 0,
{C, F+ }= {C, F } = 0,

(49)
(50)
(51)
(52)

{E, A23 } = {E, A13 }= 0, [E, A12 ] = 0,

{C, A23 } = {C, A13 }= 0, [C, A12 ] = 0,
{E, H+ } = {E, H } = 0, [E, H3 ] = 0,
{C, H+ } = {C, H } = 0, [C, H3 ] = 0,

{E, B3 }= 0,
[C, B3 ] = 0,
{E, F3 } = 0,
[C, F3 ] = 0.

{E, A23 } = 0, [E, A13 ]= 0, {E, A12 } = 0, [E, B1 ] = 0, {E, B2 } = 0, [E, B3 ] = 0, (53)
{C, A23 } = 0, [C, A13 ]= 0, {C, A12 } = 0, {C, B1 } = 0, [C, B2 ] = 0, {C, B3 } = 0. (54)
{E, A23 } = 0, [E, A13 ]= 0, {E, A12 } = 0, {E, B1 } = 0, [E, B2 ] = 0, {E, B3 }= 0, (55)
{C, A23 } = 0, [C, A13 ]= 0, {C, A12 } = 0, [C, B1 ] = 0, {C, B2 } = 0, [C, B3 ] = 0. (56)
{E, A23 } = {E, A13 }= 0, [E, A12 ] = 0,
{C, A23 } = {C, A13 }= 0, [C, A12 ] = 0,
{E, H+ } = {E, H } = 0, [E, H3 ] = 0,
{C, H+ } = {C, H } = 0, [C, H3 ] = 0,

{E, B1 } = {E, B2 } = 0,
[C, B1 ] = [C, B2 ] = 0,
{E, F+ }= {E, F } = 0,
[C, F+ ] = [C, F ] = 0,

[E, B3 ] = 0,
{C, B3 }= 0,
[E, F3 ] = 0,
{C, F3 } = 0.

(57)
(58)
(59)
(60)

where A23 , A13 , A12 are infinitesimal operators of a subgroup of threedimensional rotations, B1 , B2 , B3 are infinitesimal operators of hyperbolic rotations.
2) The field F = R. The factorization Csi ,tj Csi ,tj Csi ,tj of the real Clifford
algebra Cp,q corresponds

to a real finitedimensional representation of the group G+ with
p+q
a pair (l0 , l1 ) = 4 , 0 , that is equivalent to a representation of the subgroup SO(3) of
threedimensional rotations (B1 = B2 = B3 = 0). Then there exist two classes of real
0
representations Rl0,2
of the group G+ corresponding to the algebras Cp,q with a division
ring K R, p q 0, 2 (mod 8), and also there exist two classes of quaternionic repre0
sentations Hl4,6
of G+ corresponding to the algebras Cp,q with a ring K H, p q 4, 6
0
(mod 8). For the real representations Rl0,2
operators of the discrete subgroup of G+ defining by the matrices W, E, C of the fundamental automorphisms of Cp,q with p q 0, 2
(mod 8) are always commute with all the infinitesimal operators of the representation. In
0
following relations hold:
turn, for the quaternionic representations Hl4,6
[W, A23 ] = [W, A13 ] = [W, A12 ] = 0,
[E, A23 ] = [E, A13 ]= [E, A12 ] = 0,
[E, H+ ] = [E, H ] = [E, H3 ] = 0,
[E, A23 ] = 0,

{E, A13 } = {E, A12 } = 0,

{E, A23 } = {E, A13 }= 0, [E, A12 ] = 0,

{E, H+ } = {E, H } = 0, [E, H3 ] = 0,

[W, H+ ] = [W, H ] = [W, H3 ] = 0.

(61)

[C, A23 ]= [C, A13 ] = [C, A12 ]= 0,

[C, H+ ] = [C, H ] = [C, H3 ] = 0.

(62)
(63)

[C, A23 ] = 0,

{C, A13 } = {C, A12 } = 0. (64)

{C, A23 } = {C, A13 }= 0, [C, A12 ] = 0, (65)

{C, H+ } = {C, H } = 0, [C, H3 ] = 0. (66)

{E, A23 } = 0, [E, A13 ] = 0, {E, A12 } = 0, {C, A23 } = 0, [C, A13 ] = 0, {C, A12 } = 0.
(67)
Proof. 1) Complex representations.
As noted previously, a full representation space of the finitedimensional representation
of the proper Lorentz group G+ is defined in terms of the minimal left ideal of the algebra
C2 C2 C2 C2k . Indeed, in virtue of an isomorphism
C2k Cp,q ,

(68)

where Cp,q is a Clifford algebra over the field F = R with a division ring K C, pq 3, 7
(mod 8), we have for the minimal left ideal of C2k an expression S = Cp,q f , here
1
1
1
f = (1 e1 ) (1 e2 ) (1 et )
2
2
2
is a primitive idempotent of the algebra Cp,q [42], and e1 , e2 , . . . , et are commuting
elements with square 1 of the canonical basis of Cp,q generating a group of order 2t . The
values of t are defined by a formula t = q rqp , where ri are the RadonHurwitz numbers
[50, 35], values of which form a cycle of the period 8: ri+8 = ri + 4. The values of all ri
are

i
ri

0 1 2
0 1 2

3 4
2 3

5 6 7
3 3 3
p+q1

The dimension of the minimal left ideal S is equal to 2k = 2 2 . Therefore, for the each
finitedimensional representation of the group G+ we have 2t copies of the spinspace S2k
(full representation space). It should be noted that not all these copies are equivalent to
each other, some of them give rise to different reflection groups (see [61]).
In general, all the finitedimensional representations of group G+ in the spinspace S2k
are reducible. Therefore, there exists a decomposition of the spinspace S2k S2 S2
S2 into a direct sum of invariant subspaces Sym(kj ,0) of symmetric spintensors with
dimensions (kj + 1):
S2k = Sym(k1 ,0) Sym(k2 ,0) Symks ,0) ,
where k1 + k2 + . . . + ks = k, kj Z. At this point there exists an orthonormal basis with
matrices of the form
0

At
1/2

At

..

A
t

,
0

Bt

1/2

..

.
Bts

where for the matrices Ai1 , Aj2 , Aj3 , B1j , B2j , B3j (matrices of the infinitesimal operators of
G+ ) in accordance with GelfandNaimark formulas [29, 47]
A23 l,m =
A13 l,m =

ip
ip
(l + m + 1)(l m)l,m+1
(l + m)(l m + 1)l,m1 ,
2
2

1p
1p
(l + m)(l m + 1)l,m1
(l + m + 1)(l m)l,m+1 ,
2
2
A12 l,m = iml,m ,

(69)

(70)

(71)

i p
i p
B1 l,m = Cl (l m)(l m 1)l1,m+1 + Al (l m)(l + m + 1)l,m+1
2
2
p
i
i p
Cl+1 (l + m + 1)(l + m + 2)l+1,m+1 + Cl (l + m)(l + m 1)l1,m1 +
2
2
p
i p
i
Al (l + m)(l m + 1)l,m1 + Cl+1 (l m + 1)(l m + 2)l+1,m1 , (72)
2
2

1 p
1 p
B2 l,m = Cl (l + m)(l + m 1)l1,m1 Al (l + m)(l m + 1)l,m1
2
2
p
1 p
1
Cl+1 (l m + 1)(l m + 2)l+1,m1 Cl (l m)(l m 1)l1,m+1 +
2
2
p
1 p
1
Al (l m)(l + m + 1)l,m+1 Cl+1 (l + m + 1)(l + m + 2)l+1,m+1 , (73)
2
2
p

B3 l,m = iCl l2 m2 l1,m + iAl ml,m + iCl+1 (l + 1)2 m2 l+1,m,

il0 l1
,
Al =
l(l + 1)

i
Cl =
l

(l2 l02 )(l2 l12 )

4l2 1

(74)

(75)

m = l, l + 1, . . . , l 1, l
l = l0 , l0 + 1, . . .
we have

Aj23

lj +1
0
...
lj +1
0
lj +2 . . .

0
...
lj +2
i

= . . . . . . . . . . . . . . . . . . . .
2
. . . . . . . . . . . . . . . . . . . .
0
0
0
...
0
0
0
...

0
0
0
0

0
0

. . . . . .

. . . . . .

0 lj
lj 0

(76)

Aj13

lj +1
0
...
0
0
lj +1
0
lj +2 . . .
0
0

0
.
.
.
0
0
l
+2
j

1
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
=

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

0
0
0
...
0
lj
0
0
0
. . . lj 0

(77)

Aj12

ilj
0
0
...
0
0
0 i(lj 1)
0
...
0
0

0
0
i(lj 2) . . .
0
0

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
=

. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

0
0
0
. . . i(lj 1) 0
0
0
0
...
0
ilj

(78)

lj +1
0
...
lj +1
0
lj +2 . . .

0
lj +2
0
...
i
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
...
B1j = Aj
2
. . . . . . . . . . . . . . . . . . . .

0
0
0
...
0
0
0
...

lj +1
0
lj +1
0
lj +2

0
lj +2
1
j

B2 = Aj . . . . . . . . . . . . . . . . . . .
2
. . . . . . . . . . . . . . . . . . .
0
0
0
0
0
0

0
0
0
0

0
0

. . . . . .

. . . . . .

0 lj
lj 0

... 0
0
... 0
0

... 0
0

. . . . . . . . . . .

. . . . . . . . . . .

. . . 0 lj
. . . lj
0

ilj
0
0
...
0
0
0 i(lj 1)
0
...
0
0

0
0
i(l

2)
.
.
.
0
0
j

1
j
. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .
B3 = Aj

.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.
.

0
0
0
. . . i(lj 1) 0
0
0
0
...
0
ilj

(79)

(80)

(81)

p
where m = (lj + m)(lj m + 1). The formulas (69)(75) define a finitedimensional
representation of the group G+ when l12 = (l0 + p)2 , p is some natural number, l0 is an
integer or halfinteger number, l1 is an arbitrary complex number. In the case l12 6= (l0 +p)2
we have an infinitedimensional representation of G+ . We will deal below only with
the finitedimensional representations, because these representations are most useful in
physics.
The relation between the numbers l0 , l1 and the number k of the factors C2 in the
product C2 C2 C2 is given by a following formula


k k
, +1 ,
(l0 , l1 ) =
2 2
whence it immediately follows that k = l0 +l1 1. Thus, we have a complex representation
Cl0 +l1 1,0 of the proper Lorentz group G+ in the spinspace S2k .
Let us calculate now infinitesimal operators of the fundamental
representation C1,0 of
G+ . The representation C1,0 is defined by a pair (l0 , l1 ) = 12 , 23 . In accordance with

(69)(75) from (76)(81) we obtain





i 0 1/2
i 0 1
1/2
,
A23 =
=
2 1/2 0
2 1 0




1
1 0 1
0
1/2
1/2
A13 =
,
=
0
2 1/2
2 1 0


1 i 0
1/2
A12 =
,
2 0 i




1 0 1
i
0 1/2
1/2
,
=
B1 = A1/2
1/2
0
2
2 1 0




1
1 0 i
0 1/2
1/2
B2 = A1/2
,
=
1/2
0
2
2 i 0




i
1 1 0
1 0
1/2
B3 = A1/2
.
=
0 1
2
2 0 1

(82)
(83)
(84)
(85)
(86)
(87)

The operators (82)(87) satisfy the relations

[A23 , A13 ] = A12 ,
[A13 , A12 ] = A23 , [A12 , A23 ]= A13 ,
[B1 , B2 ] = A12 ,
[B2 , B3 ] = A23 , [B3 , B1 ] = A13 ,
[A23 , B1 ] = 0,
[A13 , B2 ] = 0, [A12 , B3 ]
= 0,
[A23 , B2 ] = B3 ,
[A23 , B3 ] = B2 ,
[A13 , B3 ] = B1 ,
[A13 , B1 ] = B3 ,
[A12 , B1 ] = B2 ,
[A12 , B2 ] = B1 .

(88)

From (85)(87) it is easy to see that there is an equivalence between infinitesimal operators
1/2
Bi and Pauli matrices:
1/2

B1

1
= 1 ,
2

1/2

B2

1
= 2 ,
2

1
= 3 .
2

(89)

1
1/2
A12 = 1 2 .
2

(90)

1/2

B3

In its turn, from (82)(84) it follows

1
1/2
A23 = 2 3 ,
2

1
1/2
A13 = 1 2 ,
2

It is obvious that this equivalence takes place also for high dimensions, that is, there exists
an equivalence between infinitesimal operators (76)(81) and tensor products of the Pauli
matrices. In such a way, let us suppose that
1
1
1
Aj23 Ea Eb , Aj13 Ec Eb , Aj12 Ec Ea ,
2
2
2
1
1
1
j
j
j
B2 Ea ,
B3 Eb ,
B1 Ec ,
2
2
2

(91)
(92)

where Ei (i = a, b, c) are kdimensional matrices (the tensor products (6)) and c < a < b.
It is easy to verify that the operators (91)(92) satisfy the relations (88). Indeed, for the
commutator [A23 , A13 ] we obtain
1
[A23 , A13 ] = A23 A13 A13 A23 = Ea EbEc Eb
4
1
1
1
1
Ec Eb Ea Eb = Ea Ec + Ec Ea = Ec Ea = A12
4
4
4
2
and so on. Therefore, the operator set (91)(92) isomorphically defines the set of infinitesimal operators of the group G+ .
In accordance with GelfandYaglom approach [27] (see also [29, 47]) an operation of
space inversion P commutes with all the operators Aik and anticommutes with all the
operators Bi :
P A23 P 1 = A23 , P A13 P 1 = A13 , P A12 P 1 = A12 ,
P B1 P 1 = B1 , P B2 P 1 = B2 , P B3 P 1 = B3 .

(93)

Let us consider permutation conditions of the operators (91)(92) with the matrix
W of the automorphism A A (space inversion). Since W = E1 E2 En is a volume
element of Cn , then Ea , Eb , Ec W, E2i = I, W2 = I at n 0 (mod 4) and W2 = I at
n 2 (mod 4). Therefore, for the operator A23 12 Ea Eb we obtain
1
A23 W = (1)a+b2 E1 E2 Ea1 Ea+1 Eb1 Eb+1 En ,
2
1
WA23 = (1)2nab E1 E2 Ea1 Ea+1 Eb1 Eb+1 En ,
2
whence it immediately follows a comparison a + b 2 2n a b (mod 2) or 2(a + b)
2(n+1) (mod 2). Thus, W and A23 are always commute. It is easy to verify that analogous
conditions take place for the operators A13 , A12 (except the case n = 2). Further, for
B1 12 Ec we obtain
1
B1 W = (1)c1 E1 E2 Ec1 Ec+1 En ,
2
1
WB1 = (1)nc E1 E2 Ec1Ec+1 En ,
2
that is, c 1 n c (mod 2) or n 2c 1 (mod 2). Therefore, the matrix W always
anticommute with B1 (correspondingly with B1 , B2 ), since n 0 (mod 2). Thus, in full
accordance with GelfandYaglom relations (93) we have1
WA23 W1 = A23 , WA13 W1 = A13 , WA12 W1 = A12 ,
WB1 W1 = B1 , WB2 W1 = B2 , WB3 W1 = B3 ,
1

(94)

Except the case of the fundamental representation C1,0 for which the automorphism group is
Aut+ (C2 ) = {I, W, E, C} = {0 , 3 , 1 , i2 } D4 /Z2 . It is easy to verify that the matrix W 3 does
not satisfy the relations (94). Therefore, in case of C1,0 we have an anomalous behaviour of the parity
transformation W P . This fact will be explained further within quotient representations.

where the matrix W of the automorphism A A is an element of an Abelian automorphism group Aut (Cn ) Z2 Z2 with the signature (+, +, +) at n 0 (mod 4) and also
a nonAbelian automorphism group Aut+ (Cn ) Q4 /Z2 with the signature (, , ) at
n 2 (mod 4), here Z2 Z2 is a GaussKlein group and Q4 is a quaternionic group (see
Theorem 9 in [60]).
Let us consider now permutation conditions of the operators (91)(92) with the matrix
e (time reversal). Over the field F = C the matrix E
E of the antiautomorphism A A
has two forms (Theorem 9 in [60]): 1) E = E1 E2 Em at m 1 (mod 2), the group
Aut+ (Cn ) Q4 /Z2 ; 2) E = Em+1 Em+2 En at m 0 (mod 2), the group Aut+ (Cn )
Z2 Z2 . Obviously, in both cases W = E1 E2 Em Em+1 Em+2 En , where the matrices
Ei are symmetric for 1 < i m and skewsymmetric for m < i n.
e m 0 (mod 2). Let us assume
So, let E = Em+1 Em+2 En be a matrix of A A,
1
that Ea , Eb , Ec E, then for the operator A23 2 Ea Eb we obtain
1
A23 E = (1)b+a2 Em+1 Em+2 Ea1 Ea+1 Eb1 Eb+1 En ,
2
1
EA23 = (1)2mab Em+1 Em+2 Ea1 Ea+1 Eb1 Eb+1 En ,
2

(95)

that is, b + a 2 2m a b (mod 2) and in this case the matrix E commutes with A23
(correspondingly with A13 , A12 ). For the operator B1 we have (analogously for B2 and
B3 ):
1
B1 E = (1)c1 Em+1 Em+2 Ec1Ec+1 En ,
2
mc 1
Em+1 Em+2 Ec1Ec+1 En ,
EB1 = (1)
2

(96)

that is, m 2c1 (mod 2) and, therefore, the matrix E in this case always anticommutes
with Bi , since m 0 (mod 2). Thus,
EA23 E1 = A23 , EA13 E1 = A13 , EA12 E1 = A12 ,
EB1 E1 = B1 , EB2 E1 = B2 , EB3 E1 = B3 .

(97)

Let us assume now that Ea , Eb , Ec 6 E, then

A23 E = (1)2m EA23 ,

(98)

that is, in this case E and Aik are always commute. For the hyperbolic operators we
obtain
Bi E = (1)m EBi

(i = 1, 2, 3)

(99)

and since m 0 (mod 2), then E and Bi are also commute. Therefore,
[E, A23 ] = 0, [E, A13 ] = 0, [E, A12 ] = 0, [E, B1 ] = 0, [E, B2 ] = 0, [E, B3 ] = 0.

(100)

Assume now that Ea , Eb E and Ec 6 E. Then in accordance with (95) and (99) the
matrix E commutes with A23 and B1 , and according to (96) anticommutes with B2 and
B3 . For the operator A13 we find
1
A13 E = (1)b1 Ec Em+1 Em+2 Eb1 Eb+1 En ,
2
2mb 1
EA23 = (1)
Ec Em+1 Em+2 Eb1 Eb+1 En ,
2

(101)

that is, 2m 2b 1 (mod 2) and, therefore, in this case E always anticommutes with A13
and correspondingly with the operator A12 which has the analogous structure. Thus,
[E, A23 ] = 0, {E, A13 } = 0, {E, A12 } = 0, [E, B1 ] = 0, {E, B2 } = 0, {E, B3 } = 0.
(102)
If we take Ea E, Eb , Ec 6 E, then for the operator A23 it follows that
1
A23 E = (1)a2 Eb Em+1 Em+2 Ea1 Ea+1 En ,
2
1
EA23 = (1)2ma1 Eb Em+1 Em+2 Ea1 Ea+1 En ,
2

(103)

that is, a 2 2m a 1 (mod 2) or 2m 2a 1 (mod 2). Therefore, E anticommutes

with A23 . Further, in virtue of (98) E commutes with A13 and anticommutes with A12
in virtue of (101). Correspondingly, from (99) and (96) it follows that E commutes with
B1 , B3 and anticommutes with B2 . Thus,
{E, A23 } = 0, [E, A13 ] = 0, {E, A12 } = 0, [E, B1 ] = 0, {E, B2 } = 0, [E, B3 ] = 0.
(104)
Cyclic permutations of the indices in Ei , Ej E, Ek 6 E and Ei E, Ej , Ek 6 E,
i, j, k = {a, b, c}, give the following relations
{E, A23 } = 0, {E, A13 } = 0, [E, A12 ] = 0,
{E, B1 } = 0, {E, B2 } = 0, [E, B3 ] = 0,

Ea , Ec E, Eb 6 E.

(105)

{E, A23 } = 0, [E, A13 ] = 0, {E, A12 } = 0,

{E, B1 } = 0, [E, B2 ] = 0, {E, B3 } = 0,

Eb , Ec E, Ea 6 E.

(106)

{E, A23 } = 0, {E, A13 } = 0, [E, A12 ] = 0,

[E, B1 ] = 0,
[E, B2 ] = 0,
{E, B3 } = 0,

Eb E, Ea , Ec 6 E.

(107)

[E, A23 ] = 0, {E, A13 } = 0, {E, A12 } = 0,

{E, B1 } = 0, [E, B2 ] = 0,
[E, B3 ] = 0,

Ec E, Ea , Eb 6 E.

(108)

Let us consider now the matrix E of the group Aut+ (Cn ) Q4 /Z2 . In this case
E = E1 E2 Em , m 1 (mod 2). At Ea , Eb , Ec E from (95) it follows that 2(a + b)

2(m + 1) (mod 2), therefore, in this case E always commutes with Aik . In its turn, from
(96) it follows that E always commutes with Bi , since m 1 (mod 2). Thus, we have the
relations (100). At Ea , Eb, Ec 6 E (except the cases n = 2 and n = 4) from (98) it follows
that E always commutes with Aik , and from (99) it follows that E always anticommutes
with Bi . Therefore, for the case Ea , Eb , Ec 6 E we have the relations (97). Analogously,
at Ea , Eb E, Ec 6 E in accordance with (95) and (96) E commutes with A23 and B1 , B3 ,
and in accordance with (101) and (99) anticommutes with A13 , A12 and B1 . Therefore,
for this case we have the relations (108). At Ea E, Eb , Ec 6 E from (103) it follows that
E anticommutes with A23 . In virtue of (98) E commutes with A13 and anticommutes with
A12 in virtue of (101). Correspondingly, from (99) and (96) it follows that E anticommutes
with B1 , B3 and commutes with B2 . Therefore, for the case Ea E, Eb , Ec 6 E we have
the relations (106). Further, cyclic permutations of the indices in Ei , Ej E, Ek 6 E and
Ei E, Ej , Ek 6 E, i, j, k = {a, b, c}, give for Ea , Ec E, Eb 6 E the relations (107), for
Eb , Ec E, Ea 6 E the relations (104), for Eb E, Ea , Ec 6 E the relations (105) and for
Ec E, Ea , Eb 6 E the relations (102).
Let us consider now the permutation conditions of the operators (91)(92) with the
f (full reflection). Over the field F = C the
matrix C of the antiautomorphism A A
matrix C has two different forms (Theorem 9 in [60]): 1) C = E1 E2 Em at m 0
(mod 2), the group Aut (Cn ) Z2 Z2 ; 2) C = Em+1 Em+2 En at m 1 (mod 2),
f , m 1
the group Aut+ (Cn ) Q4 /Z2 . So, let C = E1 E2 Em be a matrix of A A
e m 0
(mod 2), then by analogy with the matrix E = Em+1 Em+2 En of A A,
(mod 2), we have for C the relations of the form (97)(108). In its turn, the matrix
C = Em+1 Em+2 En is analogous to E = E1 E2 Em , m 1 (mod 2), therefore, in this
case we have also the relations (97)(108) for C.
Now we have all possible combinations of permutation relations between the matrices
of infinitesimal operators (91)(92) of the proper Lorentz group G+ and matrices of the
fundamental automorphisms of the complex Clifford algebra Cn associated with the complex representation Cl0 +l1 1,0 of G+ . It is obvious that the relations (94) take place for
any representation Cl0 +l1 1,0 of the group G+ . Further, if Cl0 +l1 1,0 with 2(l0 + l1 1) 0
(mod 4) and if Aut (Cn ) Z2 Z2 with E = Em+1 Em+2 En , C = E1 E2 Em , m 0
(mod 2), then at Ea , Eb, Ec E (Ea , Eb , Ec 6 C) we have the relations (97) for E and the
relations of the form (100) for C and, therefore, we have the relations (37) and (38) of the
present Theorem. Correspondingly, at Ea , Eb , Ec 6 E (Ea , Eb , Ec C) we have the relations
(100) for E and the relations of the form (97) for C, that is, the relations (41) and (42) of
Theorem. At Ea , Eb E, Ec 6 E (Ea , Eb 6 C, Ec C) we obtain the relations (102) for E
and the relations of the form (108) for C ((45)(46) in Theorem) and so on. Analogously,
if Cl0 +l1 1,0 with 2(l0 +l1 1) 2 (mod 4) and if Aut+ (Cn ) Q4 /Z2 with E = E1 E2 Em ,
C = Em+1 Em+2 En , m 1 (mod 2), then at Ea , Eb, Ec E (Ea , Eb, Ec 6 C) we have the
relations (100) for E and the relations of the form (97) for C (relations (38) and (37) in
Theorem) and so on.
Further, let us consider the following combinations of Aik and Bi (rising and lowering
operators):
H+ = iA23 A13 , H = iA23 + A13 , H3 = iA13 ,
F+ = iB1 B2 ,
F = iB1 + B2 ,
F3 = iB3 ,

(109)

satisfying in virtue of (88) the relations

[H+ , H3 ] = H+ , [H , H3 ] = H , [H+ , H ] = 2H3 ,
[H+ , F+ ] = [H , F ] = [H3 , F3 ] = 0,
[F+ , F3 ] = H+ , [F , F3 ] = H , [F+ , F ] = 2H3 ,
[H+ , F3 ] = F+ , [H , F3 ] = F ,
[H , F+ ] = [H+ , F ] = 2F3 ,
[F+ , H3 ] = F+ , [F , H3 ] = F .

(110)

It is easy to see that for W from (94) and (109) we have always the relations (35). Further,
for Aut (Cn ) Z2 Z2 at Ea , Eb, Ec E from (97) and (100) in virtue of (109) it follow
the relations (39) and (40). Analogously, at Ea , Eb, Ec 6 E from (100) and (97) we obtain
the relations (43) and (44). In contrast with this, at Ea , Eb E, Ec 6 E the combinations
(102) and (108) do not form permutation relations with operators H+,,3 and F+,,3,
since E and C commute with A23 and anticommute with A13 , and B1 commutes with
C and anticommutes with E (inverse relations take place for B2 ). Other two relations
(57)(60) and (49)(52) for Aut (Cn ) Z2 Z2 correspond to Ea , Ec E, Eb 6 E and
Eb E, Ea , Ec 6 E. It is easy to see that in such a way for the group Aut+ (Cn ) Q4 /Z2
we obtain the relations (43)(44), (39)(40), (57)(60) and (49)(52) correspondingly for
Ea , Eb , Ec E, Ea , Eb , Ec 6 E, Ea , Ec E, Eb 6 E and Eb E, Ea , Ec 6 E.
In accordance with [29] a representation conjugated to Cl0 +l1 1,0 is defined by a pair
 r r

(l0 , l1 ) = , + 1 ,
2 2
that is, this representation has a form C0,l0 l1 +1 . In its turn, a representation conjugated
to fundamental representation C1,0 is C0,1 . Let us find infinitesimal operators of the
representation C0,1 . At l0 = 1/2 and l1 = 3/2 from (69)(75) and (76)(81) we obtain






i 0 1
1 0 1
1 i 0
1/2
1/2
1/2
A23 =
, A13 =
, A12 =
,
2 1 0
2 1 0
2 0 i






1 0 1
1 0 i
i i 0
1/2
1/2
1/2
B1
=
,
B2
=
, B3
=
.
2 1 0
2 i 0
2 0 i

Or
1/2

A23

1/2

B1

1
1
1
1/2
1/2
= 2 3 , A13 = 1 3 , A12 = 1 2 ,
2
2
2
1
1
1
1/2
1/2
B2
= 2 ,
B3
= 3 .
= 1 ,
2
2
2

(111)

It is easy to see that operators (111) differ from (89)(90) only in the sign at the operators
Bi of the hyperbolic rotations. This result is a direct consequence of the wellknown
definition of the group SL(2; C) as a complexification of the special unimodular group
SU(2) (see [63]). Indeed, the group SL(2; C) has six parameters a1 , a2 , a3 , ia1 , ia2 , ia3 ,
where a1 , a2 , a3 SU(2). It is easy to verify that operators (111) satisfy the relations

(88). Therefore, as in case of the the representation Cl0 +l1 1,0 infinitesimal operators of
the conjugated representation C0,l0 l1 +1 are defined as follows

1
1
1
Aj23 Ea Eb , Aj13 Ec Eb , Aj12 Ec Ea ,
2
2
2

1
1
1
B1j Ec ,
B2j Ea ,
B3j Eb ,
2
2
2

(112)

where Ea , Eb, Ec are tensor products of the form (6). It is not difficult to verify that
operators (112) satisfy the relations (88), and their linear combinations H+,,3 , F+,,3
satisfy the relations (110). Since the structure of the operators (112) is analogous to the
structure of the operators (91)(92), then all the permutation conditions between the operators of discrete symmetries and operators (91)(92) of the representation Cl0 +l1 1,0 are
valid also for the conjugated representation C0,l0 l1 +1 and, obviously, for a representation
Cl0 +l1 1,l0 l1 +1 .
2) Real representations.
As known [29], if an irreducible representation of the proper Lorentz group G+ is defined
by the pair (l0 , l1 ), then a conjugated representation is also irreducible and defined by a
pair (l0 , l1 ). Hence it follows that the irreducible representation is equivalent to its
conjugated representation only in case when this representation is defined by a pair (0, l1 )
or (l0 , 0), that is, either of the two numbers l0 and l1 is equal to zero. We assume that

l1 = 0. In its turn, for the complex Clifford algebra Cn (Cn ) associated with the representation Cl0 +l1 1,0 (C0,l0 l1 +1 ) of the group G+ the equivalence of the representation to its

conjugated representation induces a relation Cn = Cn , which, obviously, is fulfilled only in

case when the algebra Cn = C Cp,q is reduced into its real subalgebra Cp,q (p + q = n).
Thus, a restriction of the complex representation Cl0 +l1 1,0 (or C0,l0 l1 +1 ) of the group G+
onto a real representation, (l0 , l1 ) (l0 , 0), induces a restriction Cn Cp,q . Further,
over the field F = R at p + q 0 (mod 2) there are four types of real algebras Cp,q : two
types p q 0, 2 (mod 8) with a real division ring K R and two types p q 4, 6
(mod 8) with a quaternionic division ring K H. Thus, we have four classes of the real
representations of the group G+ :
Rl00
Rl20
Hl40
Hl60

Cp,q ,
Cp,q ,
Cp,q ,
Cp,q ,

pq
pq
pq
pq

0
2
4
6

(mod
(mod
(mod
(mod

8),
8),
8),
8),

K R,
K R,
K H,
K H,

(113)

We will call the representations Hl40 and Hl60 are quaternionic representations of the group
G+ . It is not difficult to see that for the real representations with the pair (l0 , 0) all the
coefficients Al = il0 l1 /l(l + 1) are equal to zero, since l1 = 0. Therefore, all the infinitesimal operators B1j , B2j , B3j (see formulas (79)(81)) of hyperbolic rotations are also equal
to zero. Hence it follows that the restriction (l0 , l1 ) (l0 , 0) induces a restriction of the
group G+ onto its subgroup SO(3) of threedimensional rotations. Thus, real representations with the pair (l0 , 0) are representations of the subgroup SO(3). This result directly
follows from the complexification of SU(2) which is equivalent to SL(2; C). Indeed, the

parameters a1 , a2 , a3 compose a real part of SL(2; C) which under complex conjugation remain unaltered, whereas the parameters ia1 , ia2 , ia3 of the complex part of SL(2; C)
under complex conjugation are mutually annihilate for the representations with the pairs
(l0 , l1 ) and (l0 , l1 ).
Let us find a relation of the number l0 with dimension of the real algebra Cp,q . If
p + q 0 (mod 2) and 2 = e212...p+q = 1, then Cp,q is called positive (Cp,q > 0 at
p q 0, 4 (mod 8)) and correspondingly negative if 2 = 1 (Cp,q < 0 at p q 2, 6
(mod 8)). Further, in accordance with Karoubi Theorem [38, Prop. 3.16] it follows that
if C(V, Q) > 0, and dim V is even, then C(V V , Q Q ) C(V, Q) C(V , Q ), and
also if C(V, Q) < 0, and dim V is even, then C(V V , Q Q ) C(V, Q) C(V , Q ),
where V is a vector space associated with Cp,q , Q is a quadratic form of V . Using the
Karoubi Theorem we obtain for the algebra Cp,q a following factorization
Cp,q Csi,tj Csi,tj Csi,tj
|
{z
}

(114)

r times

where si , tj {0, 1, 2}. For example, there are two different factorizations C1,1 C0,2
and C1,1 C2,0 for the spacetime algebra C1,3 and Majorana algebra C3,1. It is obvious
that l0 = r/2 and n = 2r = p + q = 4l0 , therefore, l0 = (p + q)/4.
So, we begin with the representation of the class Rl00 . In accordance with Theorem 4
in [61] for the algebra Cp,q of the type p q 0 (mod 8) a matrix of the automorphism
A A has a form W = E1 E2 Ep+q and W2 = I. It is obvious that for E2a = E2b = E2c = I
permutation conditions of the matrix W with Aik are analogous to (94), that is, W always
commutes with the operators Aik of the subgroup SO(3). It is sufficient to consider
permutation conditions of W with the operators Aik of SO(3) only, since Bi = 0 for the
real representations. In this case the relations (88) take a form
[A23 , A13 ] = A12 ,

[A13 , A12 ] = A23 ,

[A12 , A23 ] = A13 .

(115)

Assume now that E2a = E2b = E2c = I, then it is easy to verify that operators
1
A23 Ea Eb ,
2

1
A13 Ec Eb ,
2

1
A12 Ec Ea
2

(116)

satisfy the relations (115) and commute with the matrix W of A A . It is easy
to see that at E2i = I, E2j = E2k = I and E2i = E2j = I, E2k = I (i, j, k = {a, b, c})
the operators Aik do not satisfy the relations (115). Therefore, there exist only two
possibilities E2a = E2b = E2c = I and E2a = E2b = E2c = I corresponding to the operators
(91) and (116), respectively.
Further, for the type p q 0 (mod 8) (p = q = m) at E = Ep+1 Ep+2 Ep+q and C =
E1 E2 Ep there exist Abelian groups Aut (Cp,q ) Z2 Z2 with the signature (+, +, +)
and Aut (Cp,q ) Z4 with (+, , ) correspondingly at p, q 0 (mod 4) and p, q 2
(mod 4), and also at E = E1 E2 Ep and C = Ep+1Ep+2 Ep+q there exist nonAbelian
groups Aut+ (Cp,q ) D4 /Z2 with (+, , +) and Aut+ (Cp,q ) D4 /Z2 with (+, +, )
correspondingly at p, q 3 (mod 4) and p, q 1 (mod 4) (Theorem 4 in [61]). Besides,
for the algebras C8t,0 of the type p q 0 (mod 8), t = 1, 2, . . . , the matrices E I,
C E1 E2 Ep and W form an Abelian group Aut (Cp,0) Z2 Z2 . Correspondingly,

for the algebras C0,8t of the type p q 0 (mod 8) the matrices E E1 E2 Eq , C I

and W also form the group Aut (C0,q ) Z2 Z2 .
e and A A
f ,
So, let E = Em+1 Em+2 E2m and C = E1 E2 Em be matrices of A A
p = q = m, m 0 (mod 2). Then for the operators (91) we obtain
Aik E = (1)2m EAik ,

1
Aik C = (1)i+j2 (i)(j)E1 E2 Ei1 Ei+1 Ej1Ej+1 Em ,
2
1
CAik = (1)2mij (i)(j)E1 E2 Ei1 Ei+1 Ej1 Ej+1 Em ,
2

(117)

(118)

since Ea , Eb , Ec 6 E and a function (n) = (m n) has a form

(
1 if n 0
(n) =
+1 if n > 0
Analogously, for the operators (116) we find
1
Aik E = (1)i+j2 (i)(j)Em+1 Em+2 Ei1 Ei+1 Ej1 Ej+1 E2m ,
2
2mij 1
(i)(j)Em+1 Em+2 Ei1 Ei+1 Ej1Ej+1 E2m , (119)
EAik = (1)
2
Aik C = (1)2m CAik .

(120)

It is easy to see that in both cases the matrices E and C always commute with the operators
Aik . Therefore, for the type p q 0 (mod 8) at p, q 0 (mod 4) and p, q 2 (mod 4)
the elements of the Abelian groups Aut (Cp,q ) Z2 Z2 and Aut (Cp,q ) Z4 with
(+, , ) are always commute with infinitesimal operators Aik of SO(3). Correspondingly,
for the algebra C8t,0 the elements of Aut (C8t,0 ) Z2 Z2 are also commute with
Aik . The analogous statement takes place for other degenerate case Aut (C0,8t ) Z2
Z2 . In the case of nonAbelian groups Aut+ (Cp,q ) D4 /Z2 (signatures (+, , +) and
(+, +, )) we obtain the same permutation conditions as (117)(120), that is, in this case
the matrices of the fundamental automorphisms always commute with Aik . Thus, for the
real representation of the class Rl00 operators of the discrete subgroup always commute
with all the infinitesimal operators of SO(3).
Further, for the real representation of the class Rl20 , type p q 2 (mod 8), at
E = Ep+1Ep+2 Ep+q and C = E1 E2 Ep there exist Abelian groups Aut (Cp,q ) Z4
with (, , +) and Aut (Cp,q ) Z4 with (, +, ) correspondingly at p 0 (mod 4),
q 2 (mod 4) and p 2 (mod 4), q 0 (mod 4), and also at E = E1 E2 Ep and C =
Ep+1Ep+2 Ep+q there exist nonAbelian groups Aut+ (Cp,q ) Q4 /Z2 with (, , ) and
Aut+ (Cp,q ) D4 /Z2 with (, +, +) correspondingly at p 3 (mod 4), q 1 (mod 4)
and p 1 (mod 4), q 3 (mod 4) (Theorem 4 in [61]). So, for the Abelian groups at
Ea , Eb , Ec 6 E (operators (91)) we obtain
Aik E = (1)2q EAik ,

(121)

1
Aik C = (1)i+j2 (i)(j)E1 E2 Ei1 Ei+1 Ej1Ej+1 Ep ,
2
1
CAik = (1)2pij (i)(j)E1 E2 Ei1 Ei+1 Ej1Ej+1 Ep .
2

(122)

Correspondingly, for the operators (116) (Ea , Eb , Ec E) we have

1
Aik E = (1)i+j2 (i)(j)Ep+1Ep+2 Ei1 Ei+1 Ej1 Ej+1 Ep+q ,
2
1
EAik = (1)2qij (i)(j)Ep+1Ep+2 Ei1 Ei+1 Ej1 Ej+1 Ep+q , (123)
2
Aik C = (1)2p CAik .

(124)

The analogous relations take place for the nonAbelian groups. From (121)(124) it is
easy to see that the matrices E and C always commute with Aik . Therefore, for the real
representation of the class Rl20 operators of the discrete subgroup always commute with
all the infinitesimal operators of SO(3).
Let us consider now quaternionic representations. Quaternionic representations of the
classes Hl40 and Hl60 , types p q 4, 6 (mod 8), in virtue of the more wide ring K H
have a more complicated structure of the reflection groups than in the case of K R.
Indeed, if E = Ej1 Ej2 Ejk is a product of k skewsymmetric matrices (among which l
matrices have +-square and t matrices have -square) and C = Ei1 Ei2 Eip+qk is a
product of p + q k symmetric matrices (among which there are h +-squares and g squares), then at k 0 (mod 2) for the type pq 4 (mod 8) there exist Abelian groups
Aut (Cp,q ) Z2 Z2 and Aut (Cp,q ) Z4 with (+, , ) if correspondingly lt, hg
0, 1, 4, 5 (mod 8) and l t, h g 2, 3, 6, 7 (mod 8), and also at k 1 (mod 2) for the
type p q 6 (mod 8) there exist Aut (Cp,q ) Z4 with (, +, ) and Aut (Cp,q ) Z4
with (, , +) if correspondingly l t 0, 1, 4, 5 (mod 8), h g 2, 3, 6, 7 (mod 8) and
l t 2, 3, 6, 7 (mod 8), h g 0, 1, 4, 5 (mod 8). Inversely, if E = Ei1 Ei2 Eip+qk
and C = Ej1 Ej2 Ejk , then at k 1 (mod 8) for the type p q 4 (mod 8) there
exist nonAbelian groups Aut+ (Cp,q ) D4 /Z2 with (+, , +) and Aut+ (Cp,q ) D4 /Z2
with (+, +, ) if correspondingly h g 2, 3, 6, 7 (mod 8), l t 0, 1, 4, 5 (mod 8)
and h g 0, 1, 4, 5 (mod 8), l t 2, 3, 6, 7 (mod 8), and also at k 1 (mod 2)
for the type p q 6 (mod 8) there exist Aut+ (Cp,q ) Q4 /Z2 with (, , ) and
Aut+ (Cp,q ) D4 /Z2 with (, +, +) if correspondingly h g, l t 2, 3, 6, 7 (mod 8)
and h g, l t 0, 1, 4, 5 (mod 8) (see Theorem 4 in [61]).
e and
So, let E = Ej1 Ej2 Ejk and C = Ei1 Ei2 Eip+qk be the matrices of A A
f , k 0 (mod 2). Assume that Ea , Eb , Ec E, that is, all the matrices Ei in the
AA
operators (91) or (116) are skewsymmetric. Then for the operators (91) and (116) we
obtain
1
A23 E = (1)b+a2 (ja )(jb )Ej1 Ej2 Eja1 Eja+1 Ejb1 Ejb+1 Ejk ,
2
1
EA23 = (1)2kab (ja )(jb )Ej1 Ej2 Eja1 Eja+1 Ejb1 Ejb+1 Ejk , (125)
2

A23 C = (1)2(p+qk) CA23 ,

(126)

that is, E and C commute with A23 (correspondingly with A13 , A12 ). It is easy to see
that relations (125) and (126) are analogous to the relations (95) and (98) for the field
F = C. Therefore, from (125) and (126) we obtain the relations (62) of Theorem. Further,
assume that Ea , Eb, Ec 6 E, that is, all the matrices Ei in the operators (91) and (116) are
symmetric. Then
A23 E = (1)2k EA23 ,

(127)

1
A23 C = (1)b+a2 (ia )(ib )Ei1 Ei2 Eia1 Eia+1 Eib1 Eib+1 Eip+qk ,
2
1
(128)
CA23 = (1)2(p+qk)ab (ia )(ib )Ei1 Ei2 Eia1 Eia+1 Eib1 Eib+1 Eip+qk
2
and analogous relations take place for A13 , A12 . It is easy to verify that from (127) and
(128) we obtain the same relations (62), since (62) are relations (37) or (41) at Bi = 0.
Therefore, over the ring K H (quasicomplex case) the elements of Abelian reflection
0
groups of the quaternionic representations Hl4,6
satisfy the relations (37)(60) over the field
F = C at Bi = 0. Indeed, at Ea , Eb E, Ec 6 E and Ec E, Ea , Eb 6 E we have relations
(64) which are particular cases of (45) at Bi = 0 and so on. It is easy to verify that the
same relations take place for nonAbelian reflection groups at E = Ei1 Ei2 Eip+qk and
C = Ej1 Ej2 Ejk , k 1 (mod 2).
Remark. Theorem exhausts all possible permutation relations between transformations
P, T, P T and infinitesimal operators of the group G+ . The relations (35)(36) take place
always, that is, at any n 0 (mod 2) (except the case n = 2). In turn, the relations
(37)(60) are divided into two classes. The first class contains relations with operators
H+,,3 and F+,,3 (the relations (37)(40), (41)(44), (49)(52), (57)(60)). The second
class does not contain the relations with H+,,3, F+,,3 (the relations (45)(46), (47)(48),
(53)(54), (55)(56). Besides, in accordance with [29] for the transformation T there are
only two possibilities T = P and T = P (both these cases correspond to relation (37)).
However, from other relations it follows that T 6= P , as it should be take place in
general case. The exceptional case n = 2 corresponds to neutrino field and further it will
be explored in the following sections within quotient representations of the group G+ .
Permutations relations with respect to symmetric subspaces Sym(k,r) can be obtained by
similar manner.

AtiyahBottShapiro periodicity on the Lorentz

group

In accordance with the section 2 the finitedimensional representations C, C, C C related

with the algebras C2k , C2r , C2k C2k of the type n 0 (mod 2) and the quotient

representations C, C, C C ( C C) related with the quotient algebras C2k , C2r ,

C2k C2k corresponding to the type n 1 (mod 2) form a full system M = M0 M1

of the finitedimensional representations of the proper Lorentz group G+ . This extension
of the Lorentz group allows to apply AtiyahBottShapiro periodicity [4] on the system
M, and also it allows to define on M some cyclic relations that give rise to interesting
applications in particle physics (in the spirit of Gell-MannNeeman eightfold way [26]).
As known, ABSperiodicity based on the Z2 graded structure of the Clifford algebra
C. Indeed, let C+ (correspondingly C ) be a set consisting of all even (correspondingly
odd) elements of the algebra C. The set C+ is a subalgebra of C. It is obvious that
C = C+ C , and also C+ C+ C+ , C+ C C , C C+ C , C C
C+ . A degree deg a of the even (correspondingly odd) element a C is equal to 0
(correspondingly 1). Let A and B be the two associative Z2 graded algebras over the
field F; then a multiplication of homogeneous elements a A and b B in a graded
is defined as follows: (a b)(a b ) = (1)deg bdeg a aa bb . The
tensor product AB
graded tensor product of the two graded central simple algebras is also graded central
simple [65, Theorem 2]. Over the field F = C the Clifford algebra Cn is central simple
if n 0 (mod 2). It is known that for the Clifford algebras with odd dimensionality
+
there is a following isomorphism C+
n+1 Cn [52]. Thus, Cn+1 is central simple algebra.
Further, in accordance with Chevalley Theorem [15] for the graded tensor product there is
n Cn +n . Two algebras Cn and Cn are said to be of the same
an isomorphism Cn C

class if n n (mod 2). The graded central simple Clifford algebras over the field F = C
form two similarity classes, which, as it is easy to see, coincide with the two types of the
algebras Cn : n 0, 1 (mod 2). The set of these 2 types (classes) forms a BrauerWall
group BWC [65, 42] that is isomorphic to a cyclic group Z2 . Thus, the algebra Cn is an
element of the BrauerWall group BWC Z2 , and a group operation is the graded tensor
Coming back to representations of the group G+ we see that in virtue of
product .
identifications Cn C (n 0 (mod 2)) and Cn C C (n 1 (mod 2)) a group action
of BWC Z2 can be transferred onto the system M = M0 M1 . Indeed, a cyclic structure
h
of the group BWC Z2 is defined by a transition C+
n Cn , where the type of the algebra
Cn is defined by a formula n = h+2r, here h {0, 1}, r Z [12, 13]. Therefore, the action
+

of BWC Z2 on M is defined by a transition C C, where Cl0 +l1 1,0 Cl0 +l1 2,0 when
+

+
C M0 (h = 1) and C = Cl0 +l1 1,0 Cl0 +l1 1,0 Cl0 +l1 1,0 when C M1 (h = 0),
dim C = h + 2r (dim C = l0 + l1 1 if C M0 and dim C = 2(l0 + l1 1) if C M1 ).
+
For example, in virtue of C+
2 C1 a transition C2 C2 (C1 C2 ) induces on the
+

system M a transition C1,0 C1,0 that in virtue of C1,0 C0,0 is equivalent to C0,0 C1,0
(C0,0 is onedimensional representation of G+ ) and, therefore, h = 1. In its turn, a
+
1,0
transition C+
C1,0 ) C1,0 C1,0 or
3 C3 induces on the system M a transition (C
1,0
C C1,0 C1,0 and, therefore, h = 0. In such a way, we see that a cyclic structure of
the group BWC Z2 induces on the system M modulo 2 periodic relations which can be
explicitly showed on the Trautmannlike diagram (spinorial clock [12, 13], see also [61]):

n 1 (mod 2)
CC

0
1

C
n 0 (mod 2)
Fig.1 The action of the BrauerWall group BWC Z2 on
the full system M = M0 M1 of complex nitedimensional
representations C of the proper Lorentz group G+ .

It is obvious that a group structure over Cn , defined by the group BWC Z2 , immediately
relates with a modulo 2 periodicity of the complex Clifford algebras [4, 38]: Cn+2

Cn C2 . Therefore, we have the following relations for C and C:

Cl0 +l1 ,0 Cl0 +l1 1,0 C1,0 ,
C0,l0 l1 C0,l0 l1 +1 C0,1

and correspondingly for C C

Cl0 +l1 ,l0 l1 Cl0 +l1 1,l0 l1 +1 C1,1 .
Thus, the action of BWC Z2 form a cycle of the period 2 on the system M, where
the basic 2-period factor is the fundamental representation C1,0 (C0,1 ) of the group G+ .
This cyclic structure is intimately related with de BroglieJordan neutrino theory of light
and, moreover, this structure is a natural generalization of BJtheory. Indeed, in the
simplest case we obtain two relations C2,0 C1,0 C1,0 and C0,2 C0,1 C0,1 , which,
as it is easy to see, correspond two helicity states of the photon (left and righthanded
polarization)2 . In such a way, subsequent rotations of the representation 2-cycle give all
other higher spin physical fields and follows to de Broglie and Jordan this structure should
be called as neutrino theory of everything.
Further, when we consider restriction of the complex representation C onto real representations R and H, that corresponds to restriction of the group G+ onto its subgroup
SO(3)3, we come to a more highgraded modulo 8 periodicity over the field F = R. Indeed, the Clifford algebra Cp,q is central simple if p q 6 1, 5 (mod 8). It is known
that for the Clifford algebra with odd dimensionality, the isomorphisms are as follows:
+
+
+
C+
p,q+1 Cp,q and Cp+1,q Cq,p [52, 49]. Thus, Cp,q+1 and Cp+1,q are central simple
algebras. Further, in accordance with Chevalley Theorem [15] for the graded tensor prod p,q Cp+p,q+q . Two algebras Cp,q and Cp,q are
uct there is an isomorphism Cp,q C
2

Such a description corresponds to HelmholtzSilberstein representation of the electromagnetic eld

as the complex linear combinations F = E + iH, F = E iH that form a basis of the Majorana
Oppenheimer quantum electrodynamics [48, 55, 24, 21] (see also recent development on this subject
based on the JoosWeinberg and BargmannWigner formalisms [20]).
3
Physical feilds dened within such representations describe neutral particles, or particles at rest such
as atomic nuclei.

said to be of the same class if p + q p + q (mod 8). The graded central simple Clifford
algebras over the field F = R form eight similarity classes, which, as it is easy to see,
coincide with the eight types of the algebras Cp,q . The set of these 8 types (classes) forms
a BrauerWall group BWR [65, 42] that is isomorphic to a cyclic group Z8 . Therefore, in
virtue of identifications (113) a group action of BWR Z8 can be transferred onto the
system M = M+ M . In its turn, a cyclic structure of the group BWR Z8 is defined
h
by a transition C+
p,q Cp,q , where the type of the algebra Cp,q is defined by a formula
q p = h + 8r, here h {0, . . . , 7}, r Z [12, 13]. Thus, the action of BWR Z8 on M is
+
 0

h
0
0
0
0
0
0
, Hl4,6
, Cl3,7
, Rl0,2
Rl0,2
, Hl4,6
Hl4,6
,
defined by a transition Dl0 Dl0 , where Dl0 = Rl0,2

+
+
r1
and Dr/2 D 2 when D M+ and Dr/2 Dr/2
Dr/2 when D M , r is a number of tensor products in (114). Therefore, a cyclic structure of the group BWR Z8
induces on the system M modulo 8 periodic relations which can be explicitly showed on
the following diagram (the round on the diagram is realized by an hourhand):
p q 0 (mod 8)
p q 1 (mod 8)
l0 l0
0
R
R
Rl0,2
0
0,2 0

1
7

U
p q 2 (mod 8)
l0
R
C l70 p q 7 (mod 8)
2

6
2

l0
l0
H
p q 3 (mod 8)C3
6 p q 6 (mod 8)

5
3

0
0
H40 Hl4,6
Hl4,6
p q 4 (mod 8)
p q 5 (mod 8)
Fig.2 The action of the BrauerWall group BWR Z8 on the
full system M = M+ M of real representations D of the
proper Lorentz group G+ , l0 = p+q
4 .

Further, it is wellknown that a group structure over Cp,q , defined by BWR Z8 , immediately relates with the AtiyahBottShapiro periodicity [4]. In accordance with [4], the
Clifford algebra over the field F = R is modulo 8 periodic: Cp+8,q Cp,q C8,0 (Cp,q+8
Cp,q C0,8 ). Therefore, we have a following relation
Dl0 +2 Dl0 R20 ,
since R20 C8,0 (C0,8) and in virtue of Karoubi Theorem from (114) it follows that
C8,0 C2,0 C0,2 C0,2 C2,0 (C0,8 C0,2 C2,0 C2,0 C0,2 )4 , therefore, r = 4,
4

The minimal left ideal of C8,0 is equal to S16 and in virtue of the real ring K R is dened
within the full matrix algebra M16 (R). At rst glance, from the factorization of C8,0 it follows that
M2 (R) H H M2 (R) 6 M16 (R), but it is wrong, since there is an isomorphism H H M4 (R) (see
Appendix B in [10]).

l0 = r/2 = 2. On the other hand, in terms of minimal left ideal the modulo 8 periodicity
looks like
Sn+8 Sn S16 .
In virtue of the mapping 8,0 : C8,0 M2 (O) [44] (see also excellent review [5]) the latter
relation can be written in the form
Sn+8 Sn O2 ,
where O is an octonion algebra. Since the algebra C8,0 C0,8 admits an octonionic
representation, then in virtue of the modulo 8 periodicity the octonionic representations
can be defined for all high dimensions and, therefore, on the system M = M+ M we
have a relation
Dl0 +2 Dl0 O,
where O is an octonionic representation of the group G+ (O R20 ). Thus, the action of
BWR Z8 form a cycle of the period 8 on the system M. This is intimately related with
an octonionic structure. In 1973, G
unaydin and G
ursey showed that an automorphism
group of the algebra O is isomorphic to an exceptional Lie group G2 that contains SU(3)
as a subgroup [31]. The G
unaydinG
ursey construction allows to incorporate the quark
phenomenology into a general algebraic framework. Moreover, this construction allows
to define the quark structure on the system M within octonionic representations of the
proper Lorentz group G+ . It is obvious that within such a framework the quark structure
cannot be considered as a fundamental physical structure underlieing of the world (as it
suggested by QCD). This is fully derivative structure firstly appearred in 8-dimension and
further reproduced into high dimensions by the round of 8-cycle generated by the group
BWR Z8 from 8 ad infinitum (growth of quarks flavors with increase of energy). One
can say that such a description, included very powerful algebraic tools, opens an another
way of understanding of the Gell-MannNeemann eightfold way in particle physics.

Pseudoautomorphism A A and charge conjugation

As noted previously, an extraction of the minimal left ideal of the complex algebra Cn
C2 C2 C2 induces a space of the finitedimensional spintensor representation
of the group G+ . Besides, the algebra Cn is associated with a complex vector space
Cn . Let n = p + q, then an extraction operation of the real subspace Rp,q in Cn forms
the foundation of definition of the discrete transformation known in physics as a charge
conjugation C. Indeed, let {e1 , . . . , en } be an orthobasis in the space Cn , e2i = 1. Let
us remain the first p vectors of this basis unchanged, and other q vectors multiply by the
factor i. Then the basis
{e1 , . . . , ep , iep+1, . . . , iep+q }

(129)

allows to extract the subspace Rp,q in Cn . Namely, for the vectors Rp,q we take the vectors
of Cn which decompose on the basis (129) with real coefficients. In such a way we obtain
a real vector space Rp,q endowed (in general case) with a nondegenerate quadratic form
Q(x) = x21 + x22 + . . . + x2p x2p+1 x2p+2 . . . x2p+q ,
where x1 , . . . , xp+q are coordinates of the vector x in the basis (129). It is easy to see
that the extraction of Rp,q in Cn induces an extraction of a real subalgebra Cp,q in Cn .
Therefore, any element A Cn can be unambiguously represented in the form
A = A1 + iA2 ,
where A1 , A2 Cp,q . The onetoone mapping
A A = A1 iA2

(130)

transforms the algebra Cn into itself with preservation of addition and multiplication
operations for the elements A; the operation of multiplication of the element A by the
number transforms to an operation of multiplication by the complex conjugate number.
Any mapping of Cn satisfying these conditions is called a pseudoautomorphism. Thus,
the extraction of the subspace Rp,q in the space Cn induces in the algebra Cn a pseudoautomorphism A A [52].
Let us consider a spinor representation of the pseudoautomorphism A A of the
algebra Cn when n 0 (mod 2). In the spinor representation the every element A Cn
should be represented by some matrix A, and the pseudoautomorphism (130) takes a form
of the pseudoautomorphism of the full matrix algebra M2n/2 :
A A.
On the other hand, a transformation replacing the matrix A by the complex conjugate
is also some pseudoautomorphism of the algebra M2n/2 . The composimatrix, A A,
tion of the two pseudoautomorpisms A A and A A, A A A, is an internal
automorphism A A of the full matrix algebra M2n/2 :
1 ,
A = A

(131)

where is a matrix of the pseudoautomorphism A A in the spinor representation.

The sufficient condition for definition of the pseudoautomorphism A A is a choice of
1 transfers into itself the
the matrix in such a way that the transformation A A
matrices E1 , . . . , Ep , iEp+1, . . . , iEp+q (the matrices of the spinbasis of Cp,q ), that is,
Ei Ei = E i 1

(i = 1, . . . , p + q).

(132)

Theorem 3. Let Cn be a complex Clifford algebra when n 0 (mod 2) and let Cp,q Cn
be its subalgebra with a real division ring K R when p q 0, 2 (mod 8) and quaternionic division ring K H when p q 4, 6 (mod 8), n = p + q. Then in dependence

on the division ring structure of the real subalgebra Cp,q the matrix of the pseudoautomorphism A A has the following form:

1) K R, p q 0, 2 (mod 8).
The matrix for any spinor representation over the ring K R is proportional to the
unit matrix.

2) K H, p q 4, 6 (mod 8).
= E1 E2 Ea when a 0 (mod 2) and = E1 E2 Eb when b 1 (mod 2),
where a complex matrices Et and b real matrices Es form a basis of the spinor representation of the algebra Cp,q over the ring K H, a + b = p + q, 0 < t a, 0 < s b. At
this point
=

I if a, b 0, 1 (mod 4),
= I if a, b 2, 3 (mod 4),

where I is the unit matrix.

Proof. The algebra Cn (n 0 (mod 2), n = p + q) in virtue of Cn = C Cp,q and
definition of the division ring K f Cp,q f (f is a primitive idempotent of the algebra
Cp,q ) has four different real subalgebras: p q 0, 2 (mod 8) for the real division ring
K R and p q 4, 6 (mod 8) for the quaternionic division ring K H.

1) K R.
Since for the types p q 0, 2 (mod 8) there is an isomorphism Cp,q M p+q
(R)
2 2
(WedderburnArtin Theorem), then all the matrices Ei of the spinbasis of Cp,q are real
and E i = Ei . Therefore, in this case the condition (132) can be written as follows
Ei Ei = Ei 1 ,

whence Ei = Ei . Thus, for the algebras Cp,q of the types p q 0, 2 (mod 8) the
matrix of the pseudoautomorphism A A commutes with all the matrices Ei. It is
easy to see that I.

2) K H.
In turn, for the quaternionic types p q 4, 6 (mod 8) there is an isomorphism Cp,q
(H). Therefore, among the matrices of the spinbasis of the algebra Cp,q there are
M p+q
2 2
matrices E satisfying the condition E = E . Let a be a quantity of the complex
matrices, then the spinbasis of Cp,q is divided into two subsets. The first subset {E t =
Et } contains complex matrices, 0 < t a, and the second subset {E s = Es } contains
real matrices, 0 < s p + q a. In accordance with a spinbasis structure of the algebra
Cp,q M p+q
(H) the condition (132) can be written as follows
2
2

Et Et = Et 1 ,

Es Es = Es 1 .

Whence
Et = Et ,

Es = Es .

(133)

Thus, for the quaternionic types p q 4, 6 (mod 8) the matrix of the pseudoautomorphism A A anticommutes with a complex part of the spinbasis of Cp,q and commutes
with a real part of the same spinbasis. From (133) it follows that a structure of the
matrix is analogous to the structure of the matrices E and C of the antiautomorphisms
e and A A
f , correspondingly (see Theorem 4 in [61]), that is, the matrix of
AA
the pseudoautomorphism A A of the algebra Cn is a product of only complex matrices,
or only real matrices of the spinbasis of the subalgebra Cp,q .
So, let 0 < a < p + q and let = E1 E2 Ea be a matrix of A A, then
permutation conditions of the matrix with the matrices Es of the real part (0 < s
p + q a) and with the matrices Et of the complex part (0 < t a) have the form
Es = (1)a Es ,

(134)

Et = (1)at (t )E1 E2 Et1 Et+1 Ea ,

Et = (1)t1 (t )E1 E2 Et1 Et+1 Ea ,

(135)

that is, when a 0 (mod 2) the matrix commutes with the real part and anticommutes
with the complex part of the spinbasis of Cp,q . Correspondingly, when a 1 (mod 2) the
matrix anticommutes with the real part and commutes with the complex part. Further,
let = E1 E2 Ep+qa be a product of the real matrices, then
Es = (1)p+qas (s )E1 E2 Es1 Es+1 Ep+qa ,
Es = (1)s1 (s )E1 E2 Es1 Es+1 Ep+qa ,

(136)

Et = (1)p+qa Et ,

(137)

that is, when p + q a 0 (mod 2) the matrix anticommutes with the real part
and commutes with the complex part of the spinbasis of Cp,q . Correspondingly, when
p + q a 1 (mod 2) the matrix commutes with the real part and anticommutes with
the complex part.
The comparison of the conditions (134)(135) with the condition (133) shows that the
matrix = E1 E2 Ea exists only at a 0 (mod 2), that is, is a product of the
complex matrices Et of the even number. In its turn, a comparison of (136)(137) with
(133) shows that the matrix = E1 E2 Ep+qa exists only at p + q a 1 (mod 2),
that is, is a product of the real matrices Es of the odd number.
Let = E1 E2 Ep+qa be a product of the
Let us calculate now the product .
= and
= 2 . Therefore,
p + q a real matrices. Since E s = Es , then
= (E1 E2 Ep+qa )2 = (1)

(p+qa)(p+qa1)
2

I.

(138)

Further, let = E1 E2 Ea be a product of the a complex matrices. Then E t = Et

= (1)a = , since a 0 (mod 2). Therefore,
and
= (E1 E2 Ea )2 = (1)

a(a1)
2

I.

(139)

Let p + q a = b be a quantity of the real matrices Es of the spinbasis of Cp,q , then

p + q = a + b. Since p + q is always even number for the quaternionic types p q 4, 6
(mod 8), then a and b are simultaneously even or odd numbers. Thus, from (138) and
(139) it follows
(
I, if a, b 0, 1 (mod 4),
=

I, if a, b 2, 3 (mod 4),
which required to be proved.
In the present form of quantum field theory complex fields correspond to charged
particles. Thus, the extraction of the subalgebra Cp,q with the real ring K R in Cn ,
p q 0, 2 (mod 8), corresponds to physical fields describing truly neutral particles such
as photon and neutral mesons ( 0 , 0 , 0 , 0, 0 , K 0 ). In turn, the subalgebras Cp,q with
the ring K H, p q 4, 6 (mod 8) correspond to charged or neutral fields.
As known [28], the charge conjugation C should be satisfied the following requirement
CI ik = I ik C,

(140)

where I ik are infinitesimal operators of the group G+ 5 . This requirement is necessary

for the definition of the operation C on the representation spaces of C. Let us find
permutation conditions of the matrix with I ik defined by the relations (91)(92). It is
obvious that in the case of K R the matrix I commutes with all the operators I ik
and, therefore, the relations (140) hold. In the case of K H and = E1 E2 Ea
it is easy to verify that at Ea , Eb , Ec (all Ea , Eb, Ec are complex matrices) the matrix
commutes with Aik and anticommutes with Bi (permutation conditions in this case
are analogous to (95) and (96)). In its turn, when Ea , Eb, Ec 6 (all Ea , Eb , Ec are real
matrices) the matrix commutes with all the operators I ik . It is easy to see that all
other cases given by the cyclic permutations Ei , Ej , Ek 6 and Ei , Ej , Ek 6
(i, j, k {a, b, c}) do not satisfy the relations (140). For example, at Ea , Eb and
Ec 6 the matrix commutes with A23 and B1 and anticommutes with A13 , A12 , B2 , B3 .
Further, in the case of = E1 E2 Eb it is not difficult to see that only at Ea , Eb, Ec
(all Ea , Eb, Ec are real matrices) the matrix commutes with all I ik . Therefore, in both
cases = E1 E2 Ea and = E1 E2 Eb the relations (140) hold when all the
matrices Ea , Eb, Ec belonging to (91)(92) are real.
When we restrict the complex representation C (charged particles) of G+ to real representation R (truly neutral particles) and H (neutral particles) we see that in this case
the charge conjugation is reduced to an identical transformation I for R and to a particle
antiparticle conjugation C for H. Moreover, as follows from Theorem 2 for the real
representations Bi = 0 and, therefore, the relations (140) take a form
C Aik = Aik C .

(141)

Over the ring K R the relations (141) hold identically. It is easy to verify that over the
ring K H for the matrix = E1 E2 Ea the relations (141) hold at Ea , Eb , Ec and
5

The requirement CP = P C presented also in the GelfandYaglom work [28] is superuous, since the
inverse relation CP = P C is valid in BWWtype quantum eld theories [3].

Ea , Eb , Ec 6 . The same result takes place for the matrix = E1 E2 Eb . All other
cases given by the cyclic permutations do not satisfy the relations (141). Therefore, in both
cases the relations (141) hold when all the matrices Ea , Eb , Ec in (91) (correspondingly in
(116)) are complex or real. Let us consider the action of the pseudoautomorphism A A
on the spinors (5) (vectors of the fundamental representation of the group G+ ). The

matrix allows to compare to each spinor its conjugated spinor by the following
rule

= ,

(142)

= 1 or
here = ( ) . In accordance with Theorem 3 for the matrix we have
= 1 , where 1 = . Then a twice conjugated spinor looks like

= = ( ) = ( ) = .
Therefore, the twice conjugated spinor coincides with the initial spinor in the case of the
real subalgebra of C2 with the ring K R (the algebras C1,1 and C2,0 ), and also in the
case of K H (the algebra C0,2 H) at ab 0, 1 (mod 4). Since for the algebra C0,2
H we have always a b 0 (mod 4), then a property of the reciprocal conjugacy of the
spinors ( = 1, 2) is an invariant fact for the fundamental representation of the group
G+ (this property is very important in physics, since this is an algebraic expression of the
requirement C 2 = 1). Further, since the vector (spintensor) of the finitedimensional
P
representation of the group G+ is defined by the tensor product 1 2 k = 1 2
k , then its conjugated spintensor takes a form
X
1 2 k
(143)
=
11 22 kk 1 2 k ,

It is obvious that the condition of reciprocal conjugacy 1 2 k = 1 2 k is also fulfilled

= 1 (all the matrices i are
for (143), since for each matrix ii in (143) we have
i
defined for the algebra C2 ).
Let us define now permutation conditions of the matrix of the pseudoautomorphism
A A (charge conjugation) with the matrix W of the automorphism A A (space
inversion). First of all, in accordance with Theorem 3 in the case of Cp,q with the real
ring K R (types p q 0, 2 (mod 8)) the matrix is proportional to the unit matrix
and, therefore, commutes with the matrix W. In the case of K H (types p q 4, 6
(mod 8)) from Theorem 3 it follows two possibilities: = E1 E2 Ea is a product of a
complex matrices at a 0 (mod 2) and = E1 E2 Eb is a product of b real matrices
at b 1 (mod 2). Since a + b = p + q, then the matrix W can be represented by the
product E1 E2 Ea E1 E2 Eb . Then for = E1 E2 Ea we have
W = (1)
W = (1)

a(a1)
2

(1 )(2 ) (a )E1 E2 Eb ,

a(a1)
+ba
2

(1 )(2 ) (a )E1 E2 Eb .

Hence it follows that at ab 0 (mod 2) the matrices and W always commute, since
a 0 (mod 2). Taking = E1 E2 Eb we obtain following conditions:
W = (1)
W = (1)

b(b1)
+ab
2
b(b1)
2

(1 )(2 ) (b )E1 E2 Ea ,

(1 )(2 ) (b )E1 E2 Ea .

Hence it follows that ab 1 (mod 2), since in this case b 1 (mod 2), and p + q = a + b
is even number, a is odd number. Therefore, at ab 1 (mod 2) the matrices and W
always anticommute.
It should be noted one important feature related with the anticommutation of the
matrices and W, W = W, that corresponds to relation CP = P C. The latter
relation holds for BargmannWightmannWigner type quantum field theories in which
bosons and antibosons have mutually opposite intrinsic parities [3]. Thus, in this case the
matrix of the operator C is a product of real matrices of odd number.

Quotient representations of the Lorentz group

e be the automorphisms of
Theorem 4. 1) F = C. Let A A, A A , A A
the odddimensional complex Clifford algebra Cn+1 (n + 1 1, 3 (mod 4)) corresponding
the discrete transformations C, P, T (charge conjugation, space inversion, time reversal)
and let Cn be a quotient algebra obtained in the result of the homomorphic mapping
: Cn+1 Cn . Then over the field F = C in dependence on the structure of Cn all the
quotient representations of the Lorentz group are divided in the following six classes:
1) Cla01+l1 1,0 : {T, C I},
2) Cla02+l1 1,0 : {T, C},

3) Clb0 +l1 1,0 : {T, CP, CP T },

4) Clc0 +l1 1,0 : {P T, C, CP T },
5) Cld01+l1 1,0 : {P T, CP IP, CT IT },

6) Cld02+l1 1,0 : {P T, CP, CT }.

2) F = R. Real quotient representations are divided into four different classes:

7) Rle01 : {T, C I, CT IT },

8) Rle02 : {T, CP IP, CP T IP T },

9) Hlf01 : {T, C C , CT C T },

10) Hlf02 : {T, CP C P, CP T C P T }.

Proof. 1) Complex representations.
Before we proceed to find an explicit form of the quotient representations C it is necessary
to consider in details a structure of the quotient algebras Cn obtaining in the result of
the homomorphic mapping : Cn+1 Cn . The structure of the quotient algebra Cn
e AA
f , A A of
depends on the transfer of the automorphisms A A , A A,
the algebra Cn+1 under action of the homomorphism onto its subalgebra Cn . As noted
previously (see conclusion of Theorem 1), the homomorphisms and have an analogous
texture. The action of the homomorphism is defined as follows
: A1 + A2 A1 + A2 ,

where A1 , A2 Cn , = e12n+1 , and

(
1, if n + 1 1 (mod 4),
=
i, if n + 1 3 (mod 4);
so that ()2 = 1. At this point 1 and the quotient algebra has a form

Cn Cn+1 / Ker ,

where Ker = {A1 A1} is a kernel of the homomorphism .

e from Cn+1 into Cn it is necessary
For the transfer of the antiautomorphism A A
that
f
= .

(144)

Indeed, since under action of the elements 1 and are equally mapped into the unit,
then transformed elements e
1 and f
are also should be mapped into 1, but e
1 = 1 1,
n(n1)
2
and f
= 1 in virtue of
e = (1)
, whence
(
,
if n + 1 1 (mod 4);

e=
(145)
, if n + 1 3 (mod 4).

e is transTherefore, under action of the homomorphism the antiautomorphism A A

ferred from Cn+1 into Cn only at n 0 (mod 4).
In its turn, for the transfer of the automorphism A A it is necessary that () =
. However, since the element is odd and = (1)n+1 , then we have always
= .

(146)

Thus, the automorphism A A is never transferred from Cn+1 into Cn .

f from Cn+1 into Cn it is
Further, for the transfer of the antiautomorphism A A
necessary that
] = .
()

(147)

It is easy to see that the condition (147) is satisfied only at n + 1 3 (mod 4), since in
this case from the second equality of (145) and (146) it follows
] =
f = = .
()

(148)

f is transTherefore, under action of the homomorphism the antiautomorphism A A

ferred from Cn+1 into Cn only at n 2 (mod 4).
Let n + 1 = p + q. Defining in Cn+1 the basis {e1 , . . . , ep , iep+1 , . . . , iep+q } we extract
the real subalgebra Cp,q , where at p q 3, 7 (mod 8) we have a complex division ring
K C, and at p q 1 (mod 8) and p q 5 (mod 8) correspondingly a double real
division ring K R R and a double quaternionic division ring K H H. The product

e1 e2 ep iep+1 iep+q = iq Cn+1 sets a volume element of the real subalgebra Cp,q .
At this point we have a condition (iq ) = iq , that is, (i)q = iq , whence
= (1)q .

(149)

When q is even, from (149) it follows = and, therefore, the pseudoautomorphism

A A is transferred at q 0 (mod 2), and since p + q is odd number, then we have
always p 1 (mod 2). In more detail, at n + 1 3 (mod 4) the pseudoautomorphism
A A is transferred from Cn+1 into Cn if the real subalgebra Cp,q possesses the complex
ring K C, p q 3, 7 (mod 8), and is not transferred ( = , q 1 (mod 2), p 0
(mod 2)) in the case of Cp,q with double rings K R R and K H H, p q 1, 5
(mod 8). In its turn, at n+ 1 1 (mod 4) the pseudoautomorphism A A is transferred
from Cn+1 into Cn if the subalgebra Cp,q has the type p q 1, 5 (mod 8) and is not
transferred in the case of Cp,q with p q 3, 7 (mod 8). Besides, in virtue of (146) at
n + 1 3 (mod 4) with p q 1, 5 (mod 8) and at n + 1 1 (mod 4) with p q 3, 7
(mod 8) a pseudoautomorphism A A (a composition of the pseudoautomorphism
A A with the automorphism A A ) is transferred from Cn+1 into Cn , since
= .
Further, in virtue of the second equality of (145) at n + 1 3 (mod 4) with p q 1, 5
e (a composition of the pseudoautomorphism
(mod 8) a pseudoantiautomorphism A A
e is transferred from Cn+1 into Cn , since
A A with the antiautomorphism A A)
f
= .

f (a composition of the pseudoautomorphism

Finally, a pseudoantiautomorphism A A
f ), corresponded to CP T transformation, is
A A with the antiautomorphism A A
transferred from Cn+1 into Cn at n + 1 3 (mod 4) and Cp,q with p q 3, 7 (mod 8),
since in this case in virtue of (148) and (149) we have
]
) = .
(
Also at n + 1 1 (mod 4) and q 1 (mod 2) we obtain
]
) = (
]
) = () = ,
(
f is transferred at n + 1 1 (mod 4) and Cp,q with
therefore, the transformation A A
p q 3, 7 (mod 8).
The conditions for the transfer of the fundamental automorphisms of the algebra Cn+1
into its subalgebra Cn under action of the homomorphism allow to define in evident way
an explicit form of the quotient algebras Cn .
1) The quotient algebra Cn , n 0 (mod 4).

e and
As noted previously, in the case n + 1 1 (mod 4) the antiautomorphism A A
pseudoautomorphism A A are transferred from Cn+1 into Cn if the subalgebra Cp,q

Cn+1 possesses the double rings K R R, K H H (p q 1, 5 (mod 8)), and also

f are transferred
the pseudoautomorphism A A and pseudoantiautomorpism A A
if Cp,q has the complex ring K C (p q 3, 7 (mod 8)). It is easy to see that in
dependence on the type of Cp,q the structure of the quotient algebras Cn of this type is
divided into two different classes:
e and
a) The class of quotient algebras Cn containing the antiautomorphism A A
pseudoautomorphism A A. It is obvious that in dependence on a division ring structure
of the subalgebra Cp,q Cn+1 this class is divided into two subclasses:
e A A at Cp,q with the ring K R R, p q 1 (mod 8).
a1 ) Cn with A A,

e A A at Cp,q with the ring K H H, p q 5 (mod 8).

a2 ) Cn with A A,

e A A ,
b) The class of quotient algebras Cn containing the transformations A A,
f if the subalgebra Cp,q Cn+1 has the complex ring K C, p q 3, 7 (mod 8).
AA

2) The quotient algebra Cn , n 2 (mod 4).

f , pseudoautomorphism
In the case n + 1 3 (mod 4) the antiautomorphism A A
f are transferred from Cn+1 into Cn if the
A A and pseudoantiautomorphism A A
subalgebra Cp,q Cn+1 possesses the complex ring K C (p q 3, 7 (mod 8)), and
e are transalso the pseudoautomorphism A A and pseudoantiautomorphism A A
ferred if Cp,q has the double rings K R R, K H H (p q 1, 5 (mod 8)). In
dependence on the type of Cp,q Cn+1 all the quotient algebras Cn of this type are
divided into following two classes:
f , A A,
c) The class of quotient algebras Cn containing the transformations A A
f if the subalgebra Cp,q has the ring K C, p q 3, 7 (mod 8).
AA

f , pseud) The class of quotient algebras Cn containing the antiautomorphism A A

e At this point, in dependence
doautomorphism A A and pseudoautomorphism A A.
on the division ring structure of Cp,q we have two subclasses

e at Cp,q with the ring K R R, p q 1

f , A A and A A
d1 ) Cn with A A
(mod 8).
f , A A and A A
e at Cp,q with the ring K H H, p q 5
d2 ) Cn with A A
(mod 8).

Thus, we have 6 different classes of the quotient algebras Cn . Further, in accordance with
[60] the automorphism A A corresponds to space inversion P , the antiautomorphisms
e and A A
f set correspondingly time reversal T and full reflection P T , and the
AA
pseudoautomorphism A A corresponds to charge conjugation C. Taking into account
this relation and Theorem 1 we come to classification presented in Theorem for complex
quotient representations.
2) Real representations.

Let us define real quotient representations of the group G+ . First of all, in the case of types
p q 3, 7 (mod 8) we have the isomorphism (68) and, therefore, these representations
are equivalent to complex representations considered in the section 3. Further, when
p q 1, 5 (mod 8) we have the real algebras Cp,q with the rings K R R, K H H
and, therefore, there exist homomorphic mappings : Cp,q Cp,q1, : Cp,q Cq,p1.
In this case the quotient algebra has a form

Cp,q1 Cp,q / Ker

or

Cq,p1 Cp,q / Ker ,

where Ker = {A1 A1} is a kernel of , since in accordance with

(
1 if p q 2, 3, 6, 7 (mod 8),
2 =
+1 if p q 0, 1, 4, 5 (mod 8)
at p q 1, 5 (mod 8) we have always 2 = 1 and, therefore, = 1. Thus, for the
e from Cp,q into Cp,q1 (Cq,p1) it is necessary
transfer of the antiautomorphism A A
that

e=

(p+q)(p+q+1)
2

In virtue of the relation

e = (1)
(
+

e=

we obtain

if p q 1, 5 (mod 8),
if p q 3, 7 (mod 8).

(150)

e is transTherefore, for the algebras over the field F = R the antiautomorphism A A

ferred at the mappings Cp,q Cp,q1, Cp,q Cq,p1, where p q 1, 5 (mod 8).
In its turn, for the transfer of the automorphism A A it is necessary that = .
However, since the element is odd and = (1)p+q , then we have always
= .

(151)

Thus, the automorphism A A is never transferred from Cp,q into Cp,q1 (Cq,p1)
f it is necessary that
Further, for the transfer of the antiautomorphism A A
f = .

From (150) and (151) for the types p q 1, 5 (mod 8) we obtain

f = = .

(152)

f is never
Therefore, under action of the homomorphism the antiautomorphism A A
transferred from Cp,q into Cp,q1 (Cq,p1).

As noted previously, for the real representations of G+ the pseudoautomorphism

0
A A is reduced into identical transformation I for Rl0,2
and to particleantiparticle
l

0
conjugation C for H4,6 . The volume element of Cp,q (types p q 1, 5 (mod 8)) can
be represented by the product e1 e2 ep ep+1 ep+2 ep+q , where ep+j = iep+j , e2j = 1,
(ep+j )2 = 1. Therefore, for the transfer of A A from Cp,q into Cp,q1 (Cq,p1) we
have a condition
= ,
and in accordance with (149) it follows that the pseudoautomorphism A A is transferred at q 0 (mod 2). Further, in virtue of the relation (151) the pseudoautomorphism
A A is transferred at q 1 (mod 2), since in this case we have
= .

e is transferred at
Also from (150) it follows that the pseudoantiautomorphism A A
p q 1, 5 (mod 8) and q 0 (mod 2), since

e = .

f (CP T transformation) in virtue of (152)

Finally, the pseudoantiautomorphism A A
and (149) is transferred from Cp,q into Cp,q1 (Cq,p1) at p q 1, 5 (mod 8) and q 1
(mod 2).
Now we are in a position that allows to classify the real quotient algebras Cp,q1
( Cq,p1).
1) The quotient algebra Cp,q1 ( Cq,p1, p q 1 (mod 8).
In this case the initial algebra Cp,q has the double real division ring K R R and its
subalgebras Cp,q1 and Cq,p1 are of the type p q 0 (mod 8) or p q 2 (mod 8)
with the ring K R. Therefore, in accordance with Theorem 3 for all such quotient
algebras the pseudoautomorphism A A is equivalent to the identical transformation
e in this case is transferred into Cp,q1 (Cq,p1) at any
I. The antiautomorphism A A
p q 1 (mod 8). Further, in dependence on the number q we have two different classes
of the quotient algebras of this type:
e A A, A A,
e p q 1 (mod 8), q 0 (mod 2).
e1 ) Cp,q1 ( Cq,p1) with A A,

f , pq 1 (mod 8), q 1 (mod 2).

e A A , A A
e2 ) Cp,q1 ( Cq,p1) with A A,

2) The quotient algebras Cp,q1 ( Cq,p1), p q 5 (mod 8).

In this case the initial algebra Cp,q has the double quaternionic division ring K H H
and its subalgebras Cp,q1 and Cq,p1 are of the type p q 4 (mod 8) or p q 6
(mod 8) with the ring K H. Therefore, in this case the pseudoautomorphism A A
is equivalent to the particleantiparticle conjugation C . As in the previous case the
e is transferred at any p q 5 (mod 8). For this type in
antiautomorphism A A
dependence on the number q there are two different classes:
e p q 5 (mod 8), q 0 (mod 2).
e A A, A A,
f1 ) Cp,q1 ( Cq,p1) with A A,

e A A , A A
f , pq 5 (mod 8), q 1 (mod 2).
f2 ) Cp,q1 ( Cq,p1) with A A,

Quotient representation Cc0,1 and neutrino field

Analysing the quotient representations of the group G+ presented in Theorem 4, we see

that only a repsesentation of the class c at j = (l0 + l1 1)/2 = 1/2 is adequate for
description of the neutrino field. This representation admits full reflection P T , charge
conjugation C and CP T transformation (space inversion P is not defined). In contrast
with this, the first three classes a1 , a2 and b are unsuitable for description of neutrino, since
in this case j is an integer number, n 0 (mod 4) (bosonic fields). In turn, the classes d1
and d2 admit CT transformation that in accordance with CP T Theorem is equivalent
to space inversion P , which, as known, is a forbidden operation for the neutrino field.
So, we have an homomorphic mapping : C3 C2 , where C3 is a simplest Clifford
algebra of the type n + 1 3 (mod 4). In accordance with Theorem 4 under action of
f , A A and A A
f are
the homomorphism : C3 C2 the transformations A A
transferred from C3 into C2 . At this point, the real subalgebra C3,0 C3 has the complex
ring K C, p q 3 (mod 8), and, therefore, the matrix of the pseudoautomorphism
A A is not unit, that according to Theorem 3 corresponds to charged or neutral fields.
In accordance with general scheme presented in the section 2 a decomposition of the
algebra C3 looks like
C3

J
+
J


J
^

C2

C2

Here central idempotents

=

1 ie1 e2 e3
,
2

+ =

1 + ie1 e2 e3
2

(153)

in accordance with [16] can be identified with helicity projection operators. In such a

way, we have two helicity states describing by the quotient algebras C2 and C2 , and a

full neutrinoantineutrino algebra is C2 C2 (cf. electronpositron algebra C2 C2 ).

Let C3 be an algebraic spinor of the form
= a0 + a1 e1 + a2 e2 + a3 e3 + a12 e1 e2 + a13 e1 e3 + a23 e2 e3 + a123 e1 e2 e3 .

(154)

Then it is easy to verify that spinors

1
+ = + = (1 + ie1 e2 e3 ),
2

1
= = (1 ie1 e2 e3 )
2

(155)

are mutually orthogonal, + = 0, since + = 0, and also + C2 , C2 .

Further, it is obvious that a spinspace of the algebra C2 C2 is S2 S 2 . It should be

noted here that structures of the spinspaces S2 S 2 and S2 S 2 are different. Indeed6 ,

00 01
 



00, 0 0  01, 0 1 
2 = 10 11
.
S2 S 2 = 
,
S

S
2

0 0 0 1
10, 1 0
11, 1 1
1 0 1 1

Since spinor representations of the quotient algebras C2 and C2 are defined in terms

of Pauli matrices i , then the algebraic spinors + C2 and C2 correspond to

spinors i S2 and i S 2 (i = 0, 1). Hence it immediately follow Weyl equations





= 0,
= 0.
(156)

+
x0
x
x0
x
Therefore, twocomponent Weyl theory can be naturally formulated within quotient
0,1
representation C1,0
of the group G+ . Further, in virtue of an isomorphism
c Cc
+
C2 C3,0 C1,3 (C1,3 is the spacetime algebra) the spinor field of the quotient representation C0,1
( C1,0
c
c ) can be expressed via the DiracHestenes spinor field (x) C3,0
[32, 33, 43]. Indeed, the DiracHestenes spinor is represented by a following biquaternion
number
= a0 + a01 0 1 + a02 0 2 + a03 0 3 +
a12 1 2 + a13 1 3 + a23 2 3 + a0123 0 1 2 3 ,
or using basis








I 0
0 1
0 2
0 3
0 =
, 1 =
, 2 =
, 3 =
,
0 I
1 0
2 0
3 0
we can write (157) in the matrix form

1 2
2 1
=
3 4
4 3
where

1 = a0 ia12 ,

2 = a13 ia23 ,

3 4
4 3
,
1 2
2 1

3 = a03 ia0123 ,

(157)

(158)

(159)

4 = a01 + ia02 .

From (154)(155) and (157) it is easy to see that spinors + and are algebraically
equivalent to the spinor C2 C3,0 . Further, since C+
1,3 , then actions of the
e and A A
f on the field are equivalent. On the other hand,
antiautomorphisms A A
in accordance with FeynmanStueckelberg interpretation, time reversal for the chiral field
is equivalent to charge conjugation (particles reversed in time are antiparticles). Thus,
for the field C0,1
we have C T and, therefore, this field is CP invariant.
c
6

See also [1].

The spinor (157) (or (159)) satisfies the DiracHestenes equation

2 1

mc
0 = 0,
~

(160)

where = x is the Dirac operator. Let us show that a massless DiracHestenes

equation
2 1 = 0

(161)

describes the neutrino field. Indeed, the matrix 0 1 2 3 = 5 commutes with all the
elements of the biquaternion (157) and, therefore, 5 is equivalent to the volume element
= e1 e2 e3 of the biquaternion algebra C3,0 . In such a way, we see that idempotents
P+ =

1 + 5
,
2

P =

1 5
2

cover the central idempotents (153). Further, from (161) we obtain

P

2 1 = Pm p 2 1 = 0,

x
x

that is, there are two separated equations for = P 2 1 :

where

= 0,
x

(162)

1 3
2 4
3 1 4 2
1
i 2 4 1 3 4 2
3 1

= (1 5 )21 =

2
2 1 + 2 2 4 3 + 1 4 + 2
2 + 4 1 + 3 4 + 2 3 1

Therefore, each of the functions + and contains only four independent components
and in the split form we have




5 6 7 8
1
2
3
4
+

=
, =
,
1 2 3 4
5 6 7 8
where









i 3 1
i 2 + 4
i 4 2
i 1 3
, 3 =
,
, 2 =
, 4 =
1 =
2 2 4
2 1 3
2 4 2
2 3 + 1








i 3 + 1
i 2 4
i 4 + 2
i 1 + 3
, 7 =
.
, 6 =
, 8 =
5 =
2 2 + 4
2 1 + 3
2 4 + 2
2 3 1

Thus, in the basis we obtain from (162)






i = 0,
i+4 = 0,

+
x0
x
x0
x

(i = 1, 2, 3, 4)

These equations are equivalent to Weyl equations (156) and, therefore, should be called
WeylHestenes equations for neutrino field.

Acknowledgments
I am grateful to Prof. J.S.R. Chisholm for sending me his interesting works.

References
[1] R. Ablamowicz, Spinor Representations of Clifford Algebras: A Symbolic Approach, CPC
Thematic Issue Computer Algebra in Physics Research, Physics Communications 115,
510535 (1998).
[2] D.V. Ahluwalia, D.J. Ernst, Paradoxical kinematic acasuality in Weinbergs equations for
massless particles of arbitrary spin, Mod. Phys. Lett. A7, 19671974 (1992).
[3] D.V. Ahluwalia, M.B. Johnson, T. Goldman, A Bargmann-Wightman-Wigner Type Quantum Field Theory, Phys. Lett. B316(1), 102108 (1993).
[4] M.F. Atiyah, R. Bott, A. Shapiro, Clifford modules, Topology, 3, (Suppl. 1), 3-38, (1964).
[5] J.C. Baez, The Octonions, preprint math.RA/0105155 (2001).
[6] V. Bargmann, E.P. Wigner, Group theoretical discussion of relativistic wave equations,
Proc. Nat. Acad. USA 34, 211223 (1948).
[7] A.O. Barut, R. Raczka, Theory of Group Representations and Applications (PWN,
Warszawa, 1977).
[8] J. Baugh, D.R. Finkelstein, A. Galiautdinov, H. Saller, Clifford algebra as quantum language, J. Math. Phys. 42, 14891500 (2001).
[9] V.B. Berestetskii, E.M. Lifshitz, L.P. Pitaevskii, Quantum Electrodynamics, Course of
Theoretical Physics Vol. 4, 2nd Edition (Pergamon Press, Oxford, 1982).
[10] M. Berg, C. DeWitt-Morette, S. Gwo, E. Kramer, The Pin Groups in Physics: C, P , and
T , preprint math-ph/0012006 (2000).
[11] L. de Broglie, Compt. Rend. 195, 862 (1932).
[12] P. Budinich, A. Trautman, An introduction to the spinorial chessboard, J. Geometry and
Physics 4, 363390 (1987).
[13] P. Budinich, A. Trautman, The Spinorial Chessboard (Springer, Berlin, 1988).
[14] I.L. Buchbinder, D.M. Gitman, A.L. Shelepin, Discrete symmetries as automorphisms of
proper Poincare group, preprint hep-th/0010035 (2000).
[15] C. Chevalley, The construction and study of certain important algebras, Publications of
Mathematical Society of Japan No 1 (Herald Printing, Tokyo, 1955).
[16] J.S.R. Chisholm, R.S. Farwell, Properties of Clifford Algebras for Fundamental Particles,
in Cliord (Geometric) Algebras, ed. W. Baylis (Birkhauser, 1996), pp. 365388.

[17] R. Coquereaux, Modulo 8 periodicity of Clifford Algebras and particle physics, Phys. Lett.
115B, 389 (1982).
[18] L. Dabrowski, Group Actions on Spinors (Bibliopolis, Naples, 1988).
[19] C. Doran, A. Lasenby, S. Gull, States and Operators in Spacetime Algebra, Found. Phys.
23(9), 1239 (1993).
[20] V.V. Dvoeglazov, The Weinberg Formalism and a New Look at the Electromagnetic Theory,
in The Enigmatic Photon. Vol. IV (Eds.: M. Evans, J.P. Vigier, S. Roy and G. Hunter)
p. 305353, Series Fundamental Theories of Physics. Vol. 90 (Ed. A. van der Merwe) Dordrecht, Kluwer Academic Publishers 1997.
[21] V.V. Dvoeglazov, Essay on the NonMaxwellian Theories of Electromagnetism, Hadronic
J. Suppl. 12, 241288 (1997).
[22] V.V. Dvoeglazov, Speculations on the Neutrino Theory of Light, Annales de la Fondation
de Louis de Brogle 24, No. 1-4, pp.111128 (1999).
[23] J.P. Elliot, P.G. Dawber, Symmetry in Physics (Macmillan Press, London 1979).
[24] S. Esposito, Covariant Majorana Formulation of Electrodynamics, Found. Phys. 28, 231
244 (1998).
[25] D. Finkelstein, A. Galiautdinov, Cliffordons, preprint hepth/0005039 (2000).
[26] M. Gell-Mann, Y. Neeman, The Eightfold Way (Benjamin, 1964).
[27] I.M. Gelfand, A.M. Yaglom, General relativisticinvariant equations and infinite
dimensional representations of the Lorentz group, Zh. Ehksp. Teor. Fiz. 18, 703733 (1948).
[28] I.M. Gelfand, A.M. Yaglom, Charge conjugation for general relativisticinvariant equations, Zh. Ehksp. Teor. Fiz. 18, 11051111 (1948).
[29] I.M. Gelfand, R.A. Minlos, Z.Ya. Shapiro, Representations of the Rotation and Lorentz
Groups and their Applications (Pergamon Press, Oxford, 1963).
[30] D.M. Gitman, A.L. Shelepin, Fields on the Poincare Group: Arbitrary Spin Description
and Relativistic Wave Equations, Int. J. Theor. Phys. 40(3), 603684 (2001).
[31] M. G
unaydin, F. G
ursey, Quark Structure and Octonions, J. Math. Phys. 14, 16511667
(1973).
[32] D. Hestenes, SpaceTime Algebra (Gordon & Breach, New York, 1966).
[33] D. Hestenes, Real spinor fields, J. Math. Phys. 8, 798-808, (1967).
[34] P.R. Holland, Causal interpretation of a system of the two spin1/2 particles, Phys. Rep.
169, 293327 (1988).
[35] A. Hurwitz, Uber die Komposition der quadratischen Formen, Math. Ann. 88, 125 (1923).
[36] H. Joos, Zur darstellungstheorie der inhomogenen Lorentzgrouppe als grundlade quantenmechanische kinematick, Fortschr. Phys. 10, 65 (1962).

[37] P. Jordan, Z. Phys. 93, 434 (1935).

[38] M. Karoubi, K-Theory. An Introduction (Springer-Verlag, Berlin, 1979).
[39] A.P. Kotelnikov, In memorial Lobatchevskii 2, 3766 (Kazan, 1927).
[40] T.K. Kuo, Internal-symmetry groups and their automorphisms, Phys. Rev. D4, 36203637
(1971).
[41] O. Laport, G.E. Uhlenbeck, Application of spinor analysis to the Maxwell and Dirac equations, Phys. Rev. 37, 13801397 (1931).
[42] P. Lounesto, Scalar Products of Spinors and an Extension of Brauer-Wall Groups, Found.
Phys. 11, 721740 (1981).
[43] P. Lounesto, Clifford algebras and Hestenes spinors, Found. Phys. 23, 1203-1237 (1993).
[44] C.A. Manogue, J. Schray, Octonionic representations of Clifford algebras and triality,
Found. Phys. 26, 1770 (1996).
[45] L. Michel, Invariance in quantum mechanics and group extension, in Group Theoretical
Concepts and Methods in Elementary Particle Physics, pages 135200 (Gordon & Breach,
New York, 1964).
[46] H. Minkowski, Raum and Zeit, Phys. Zs. 10, 104 (1909).
[47] M.A. Naimark, Linear Representations of the Lorentz Group (Pergamon, London, 1964).
[48] J.R. Oppenheimer, Phys. Rev. 38, 725 (1931).
[49] I.R. Porteous, Topological Geometry (van Nostrand, London, 1969).
[50] J. Radon, Lineare Scharen orthogonaler Matrizen, Abh. Math. Seminar Hamburg 1, 124
(1922).
[51] W. Rarita, J. Schwinger, On a theory of particles with half-integral spin, Phys. Rev. 60, 61
(1941).
[52] P.K. Rashevskii, The Theory of Spinors, (in Russian) Uspekhi Mat. Nauk 10, 3110 (1955);
English translation in Amer. Math. Soc. Transl. (Ser. 2) 6, 1 (1957).
[53] B.A. Rozenfeld, NonEuclidean Geometries (Moscow, 1955) [in Russian].
[54] Yu.B. Rumer, Spinorial Analysis (Moscow, 1936) [in Russian].
[55] Yu.B. Rumer, A.I. Fet, Polarization operator of quantized fields, Zh. Ehksp. Theor. Fiz.
55, 13901392 (1968).
[56] Yu.B. Rumer, A.I. Fet, Group Theory and Quantized Fields (Moscow, 1977) [in Russian].
[57] Yu.M. Shirokov, A group theoretical consideration of the first principles of relativistic quantum mechanics. I, Zh. Ehksp. Teor. Fiz. 33, 861872 (1957).

[58] Z.K. Silagadze, On the internal parity of antiparticles, Sov. J. Nucl. Phys. 55, 392396
(1992).
[59] S. Somaroo, A. Lasenby, C. Doran, Geometric Algebra and the causal approach to multiparticle quantum mechanics, J. Math. Phys. 40, 33273340 (1999).
[60] V.V. Varlamov, Fundamental Automorphisms of Clifford Algebras and an Extension of
Dabrowski Pin Groups, Hadronic J. 22, 497535 (1999).
[61] V.V. Varlamov, Discrete Symmetries and Clifford Algebras, Int. J. Theor. Phys. 40(4),
769805 (2001).
[62] V.V. Varlamov, Clifford Algebras and Discrete Transformations of Spacetime, Proceedings
of Third Siberian Conference on Mathematical Problems of Spacetime Physics, Novosibirsk,
2022 June 2000. (Institute of Mathematics Publ., Novosibirsk, pp. 97135, 2001).
[63] N.Ya. Vilenkin, Special Functions and the Theory of Group Representations (AMS, Providence, 1968).
[64] B.L. van der Waerden, Spinoranalyse, Nachr. d. Ces. d. Wiss. G
ottingen, 100109 (1929).
[65] C.T.C. Wall, Graded Brauer Groups, J. reine und angew. Math. 213, 187199 (1964).
[66] S. Weinberg, Feinman rules for any spin I & II & III, Phys. Rev. 133B, 13181332 &
134B, 882896 (1964) & 181B, 18931899 (1969).
[67] E.P. Wigner, On unitary representations of the inhomogeneous Lorentz group, Ann. Math.
40, 149204 (1939).
[68] E.P. Wigner, Unitary Representations of the Inhomogeneous Lorentz Group Including Reflections in Group Theoretical Concepts and Methods in Elementary Particle Physics, Ed.
F. G
ursey (Gordon & Breach, New York, 1964).