EUROPEAN ORGANIZATION FOR NUCLEAR RESEARCH

CERNTH.6801/93

Masses, Mixings, Yukawa Couplings and their

arXiv:hep-ph/9302301v1 24 Feb 1993

Symmetries

Arcadi Santamaria)

CERN, Geneva, Switzerland

Abstract

We present a method to find the number of real and imaginary observ-

able parameters coming from the Yukawa sector in an arbitrary gauge theory.
The method leads naturally to a classification of Yukawa couplings accord-
ing to their symmetries and suggests a new parametrization of masses and
mixings that is useful to study the behaviour of Yukawa couplings under the
renormalization group. We apply it to some examples based on the Standard
Model with Yukawa couplings obeying various chiral symmetries. We also
show how our method of parameter counting can be used in some models with
an enlarged leptonic sector.

CERN-TH.6801/93
February 1993

*) On leave of absence from Departament de Fsica Teorica, Universitat de Valencia and

IFIC, Valencia, Spain.
 The most difficult part to understand of theories with spontaneously broken gauge
theories is, together with the Higgs potential, the Yukawa sector. Its origin is unclear,
and probably related to the origin of the spontaneous breaking of the symmetries itself.
It contains a large number of free parameters that must be adjusted by hand to obtain a
realistic spectrum of particles and mixings. From the 17 parameters needed to describe the
electro-weak theory, 13 come from the Yukawa sector. In addition, the Yukawa couplings
contain a lot of spurious parameters that cannot be observed because they can be removed
from the Lagrangian with an appropriate choice of phases and mixings for the fermion
fields. Thus, a simple question such as how many parameters are needed to describe
a theory with Yukawa couplings cannot be answered at first sight. We cannot even
say whether the couplings are real or not and so if the theory conserves CP or not.
This represents a great complication when one tries to understand mass matrices in the
Standard Model. In the last years this subject has received a lot of attention (for some
popular mass-matrix models, see for example [1, 2, 3, 4]). Usually people make ansatze
about a particular form of mass matrices and try to extract from them a reasonable
spectrum of masses and mixings. However, completely different forms of mass matrices
may be equivalent in their physical content. This has led to some confusion in the past.
Moreover, some symmetries actually present in the Yukawa sector may be hidden when
one does not work in the appropriate basis. For those reasons, an analysis based on
symmetry should be superior and could clarify some of the issues present in particular
forms of mass matrices. It would also allow a clear understanding of the number of
parameters needed to describe the model.
The standard method to calculate the number of parameters needed to describe the
Yukawa sector of a spontaneously broken gauge theory is just by construction. After
symmetry breaking one diagonalizes the mass matrices of all fields and after that one uses
the freedom in the phase definition of the fields to reabsorb as many phases as possible.
The remaining parameters in the Lagrangian are the physical parameters. As we know,
the method works very well for the Standard Model. However, as the complication of the
Yukawa sector increases, the method becomes intricate and can lead very easily to error;
the question of CP-violation, for example, can then be difficult to answer. In addition, in
many cases, phases and mixings can be moved from the gauge couplings to the Yukawa
sector or even included in a redefinition of the fields [5, 6], which leads to further confusion.
Here we propose a method, based on symmetry, to perform the counting before spon-
taneous symmetry breaking and without performing the explicit diagonalization of the
mass matrices. The method clarifies the structure of the manifold of the parameters that
are needed to describe the Yukawa sector of a gauge theory and suggests a parametriza-
tion that is useful when one wants to study the behaviour of the Yukawa couplings under
the renormalization group. Basically the method consists in the study of the symme-
tries of both the full and the Yukawa Lagrangians. The number of parameters needed to
describe the Yukawa sector comes from the balance between the number of parameters
contained in the Yukawa matrices and the number of symmetries broken by the Yukawa
couplings. As an example, we will start with the Standard Model with the most general
set of Yukawa couplings. We will then impose to the Standard Model a set of chiral
symmetries that could reduce the number of parameters and see how the method can be
used in those cases. Finally we will apply the method to more involved models based on
the same gauge group but different particle content.
The number of parameters needed to describe the Yukawa sector of a theory is ob-
viously the same before and after spontaneous symmetry breaking1 . Thus, we will start
with the hadronic part of the electro-weak Lagrangian before symmetry breaking

L = iQL D
6 QL + iuR D
6 uR + idR D
6 dR + (QL Yu uR + QL Yd dR + h.c.) . (1)

Here QL , uR and dR are the standard quark fields, left-handed doublet, u-singlet and
d-singlet respectively. If we assume n generations they are n-column vectors in genera-
tion space; Yu and Yd are the Yukawa couplings represented by n n complex arbitrary
matrices; 6D D where D is the Standard Model electro-weak covariant derivative.
Finally is the Higgs doublet and i2 .
If Yu = Yd = 0, the Lagrangian is obviously invariant under the following chiral
symmetries:
QL VQ QL u R Vu u R d R Vd d R , (2)
where, VQ , Vu , and Vd are n n unitary matrices acting on flavour space. The Yukawa
couplings break explicitly these symmetries, but in a very particular way. In fact, if we
let the Yukawa couplings transform as follows

Yu VQ Yu Vu Yd VQ Yd Vd , (3)

the Lagrangian of eq. (1) is still invariant under the combined action of the transformations
in eqs. (2) and (3). It is also easy to see that not all the symmetries in eq. (2) are broken by
the Yukawa couplings. Indeed if we choose VQ = Vu = Vd = ei , the Yukawa Lagrangian
remains invariant. This is nothing else than baryon number conservation.
Equation (3) defines an equivalence relation

(Yu , Yd ) (Yu , Yd ) = (VQ Yu Vu , VQ Yd Vd ) . (4)

The Lagrangians with couplings (Yu , Yd ) and couplings (Yu , Yd ) are completely equivalent.
Thus, counting how many parameters are needed to describe masses, mixings, and Yukawa
couplings, is the same as counting how many equivalent classes there are with respect to
the equivalence relation of eq. (4).
This problem has some similarity with the problem of spontaneous symmetry breaking
by the Higgs mechanism. There, one has a group G and a representation of Higgses
whose vacuum expectation value breaks the group G down into the group G G and
one wants to know the number of physical Higgses. As we know, the number of physical
Higgs degrees of freedom Nphys is

Nphys = N NGoldstone . (5)

Here N is the number of degrees of freedom in the Higgs representation and NGoldstone
is the number of Goldstone bosons that appear after spontaneous symmetry breaking. It
is equal to the number of broken generators of G, which is the number of generators of the
full group G minus the number of generators of the unbroken subgroup G (NGoldstone =
NG NG ). We have a similar situation here. A chiral symmetry G is broken explicitly
by the Yukawa sector Y to a group G . Then, only the broken part of G can be used to
absorb parameters from Y . Following eq. (5) we could write

NYphys = NY NG + NG . (6)
1
For the moment, we leave aside the case of spontaneous breakdown of CP, in which some parameters
can be moved from the Higgs potential to the Yukawa sector.
In the Standard Model, as follows from eq. (2), the group is G = U(n)Q U(n)u
U(n)d and the Yukawa couplings Y = (Yu , Yd) are two general n n complex matrices
transforming under the group G as eq. (4). Here the factors U(n) denote the full unitary
group U(n) = SU(n) U(1). As discussed previously, for the most general Yukawa
couplings, the only subgroup of G that leaves the Yukawa couplings invariant is a U(1)B
such that VQ = Vu = Vd = ei .
After eq. (6), counting of parameters is trivial. Taking into account that an arbitrary
complex matrix of dimension n contains n2 moduli and n2 phases and that a U(n) matrix
contains n(n 1)/2 moduli and n(n + 1)/2 phases, we find that Yphys can be expressed
in terms of
couplings and symmetries moduli phases
2
(Yu , Yd) 2n 2n2
U(n)Q U(n)u U(n)d 3n(n 1)/2 3n(n + 1)/2 . (7)
U(1)B 0 1
Yphys 2n + n(n 1)/2 (n 2)(n 1)/2

In particular if n = 3 we have 6 + 3 moduli and 1 phase. After symmetry breaking, some

of the moduli are related to masses and other to mixings or remaining Yukawa couplings.
To distinguish among them we can repeat the argumentation but using the completely
broken theory with all the heavy and Higgs bosons decoupled. In the case of the Standard
Model, what remains is a model with the standard fermion content but only QED (and
QCD if included in the analysis)2 . The resulting Lagrangian is

L = iuL D
6 uL + iuR D
6 uR + idL D
6 dL + idR D
6 dR + (uLMu uR + dL Md dR + h.c) . (8)

Here 6 D is the standard QED covariant derivative and Mu and Md are the quark mass
matrices related to the Yukawa couplings by the vacuum expectation value of the Higgs
scalar.
Except for the mass terms, the Lagrangian of eq. (8) is invariant under the following
symmetries

u L Vu L u L u R Vu R u R dL VdL dL dR VdR dR . (9)

Using similar arguments as above, but now taking into account that the full Lagrangian
has separate flavour conservation for u-type and d-type quarks, i.e. G = U(n)uR
U(n)uL U(n)dR U(n)dL and G = (U(1)u )n (U(1)d )n , we find that the Lagrangian (8)
gives rise to only 2n masses. Of course this is obvious since the symmetries in eq. (9) are
enough to diagonalize the two mass matrices completely. Thus, from the 2n + n(n 1)/2
moduli present in the Yukawa sector of the Standard Model, 2n correspond to masses and
n(n 1)/2 to mixings, as expected [7] (see also [8]).
Yukawa couplings related by eq. (4) are in the same class of equivalence. To find
a parametrization of the physical Yukawa sector ve have to characterize the equivalent
classes, and this can be done by taking one element of each equivalence class. We can use
a VQ and a Vu rotation to diagonalize the matrix Yu , VQ Yu Vu = Du . After that, since any
arbitrary complex matrix can be written as a Hermitian matrix times a unitary matrix,
we can use a Vd transformation to write Yd as a positive-definite Hermitian matrix. Thus
(Yu , Yd ) (Du , Hd ), with Du a positive-definite diagonal matrix and Hu a positive-definite
Hermitian matrix. But, Du is still invariant under transformations like Du KDu K ,
2
We have to keep all the exact symmetries. Fields with different charges never mix.
with K a diagonal matrix of phases, while Hd is not. This means that we can use K to
absorb phases from Hu . In fact we can chose K such that the next-to-diagonal diagonal
elements of Hd be real and positive. The only remaining symmetry is just U(1)B of baryon
number, as expected. Thus, without loss of generality we can represent the physical
Yukawa couplings of the Standard Model by a positive-definite diagonal matrix Du for
the u-type quarks and a positive-definite Hermitian matrix Hd with the next-to-diagonal
diagonal elements of Hd real and positive:

(Yu , Yd ) Yphys = (Du , Hd ) . (10)

One can easily check that the right-hand side of eq. (10) contains also 2n + n(n 1)/2
moduli and (n 1)(n 2)/2 phases. Of course, one can further write Hd in terms of a
Kobayashi-Maskawa-type matrix and a diagonal matrix of masses.
Since the full Lagrangian is invariant under the combined action of transformations
(2) and (3) it is easy to show that the renormalization group equations for the Yukawa
couplings must be covariant with respect to the set of transformations (3). This guarantees
that the renormalization group equations can be written only in terms of the parameters
in the right-hand side of eq. (10). This way we can obtain a set of renormalization group
equations with all the unphysical parameters removed3 .
The method we just explained for counting the physical parameters works very well
in the Standard Model with the most general Yukawa couplings. However, in the form
we have presented it, it is not suitable when some of the masses are zero or there are
some symmetries among them. We want to generalize the method to those cases. We will
impose that the only allowed reductions of parameters are those that are protected by
some symmetry. For instance, suppose we want to make the first n 1 d-type quarks (in
n generations) exactly massless. We could impose an additional chiral symmetry such as
!
U 0
QL QL uR uR dR dR , (11)
0 1

where U is a (n 1) (n 1) general unitary matrix acting only on the first n 1

generations of dR quarks. Clearly the only way to make the Yukawa Lagrangian invariant
under this symmetry is for

0 0 1
0 0 2



Yu = arbitrary Yd =
. (12)


0 0 n

Then, the Yukawa Lagrangian is only invariant under U(n 1) U(1)B . Nave counting
starting from the full group G = U(n)Q U(n)u U(n)d does not work in this case,
because the couplings in eq. (12) are not a representation of the full group G. Said in
another way, general G transformations do no preserve the form of the Yukawa matrices
in eq. (12) and, thus, we cannot use the full group to absorb parameters from the Yukawa
sector. It is not difficult to see that the only subgroup of G that preserves the form of the
Yukawa couplings in eq. (12) is G = U(n)Q U(n)u U(n 1) U(1)B . Now we can use
G to absorb parameters from the Yukawa sector and the counting of physical parameters
3
A similar approach has been followed by [9, 10, 11, 12]
comes as follows
couplings and symmetries moduli phases
(Yu , Yd ) n2 + n n2 + n
U(n)Q U(n)u U(n 1)d U(1)B (n 1)(3n 2)/2 n(3n + 1)/2 1 . (13)
U(n 1) U(1)B (n 1)(n 1)/2 (n 1)n/2 + 1
Yphys n + 1 + (n 1) 0

The model leads to n massive u-type quarks, 1 massive d-type quark and n 1 mixings.
There is no CP-violating phase. This result can be explicitly checked by full diagonaliza-
tion of the Yukawa sector.
Another extreme case can be obtained by imposing an axial U(1) symmetry on the
d-quark:

QL QL uR uR diR diR i = 2, , n d1R ei d1R . (14)

Then, the Yukawa couplings take the following form

0 1,1 1,n1
0 2,1 2,n1



Yu = arbitrary Yd =

. (15)



0 n,1 n,n1

It is easy to see that the group that preserves the form of these matrices is G = U(n)Q
U(n)u U (n 1)d U(1)B with the U (n 1) now acting on the 2, 3, , n generations
of dR quarks, while the only symmetries of the full Lagrangian are U(1)d1 U(1)B . Again
we can do the counting

couplings and symmetries moduli phases

(Yu , Yd ) n2 + n(n 1) n2 + n(n 1)
U(n)Q U(n)u U (n 1)d U(1)B (n 1)(3n 2)/2 n(3n + 1)/2 1
U(1)d1 U(1)B 0 2
Yphys n + (n 1) + n(n 1)/2 (n 1)(n 2)/2
(16)
This model just gives a massless d-quark. Phases and mixings are exactly the same as
in the Standard Model for any number of generations. This result is interesting because
the so-called mu = 0 solution of the Strong CP Problem (see for example [13, 14, 15] and
references therein) is based in a symmetry like ours. In our discussion we have obviated
the fact that chiral symmetries have in general anomalies. In the Standard Model, be-
cause of QCD instanton effects and because the chiral symmetries are anomalous, there
is an additional phase that cannot be removed. Chiral transformations move it from the
Yukawa sector to the QCD sector. This represents a problem because it gives a too large
contribution to the electric dipole moment of the neutron unless it is made unnaturally
small. If for instance mu = 0, there is an additional chiral symmetry (similar to the one
we have just studied) and the extra phase can be rotated away. However, imposing an
additional chiral symmetry is a very strong constraint on the mass matrices [see eq. (15)].
One has to redo all parameter counting and be sure that, there still is in the model an
observable CP-violating phase that could account for all observed CP-violating effects.
This is just what we did in our example.
 The examples discussed above show that the use of symmetry is a natural and basis-
independent way of choosing models for mass matrices. Of course, most of the symmetries
lead to some massless quarks and/or some vanishing mixings, but the symmetries could
be softly broken in order to generate a realistic pattern of masses and mixings. We
find it interesting to explore and classify all possible symmetries that can be imposed
on the Yukawa sector of the Standard Model according to the spectrum of masses and
mixings they generate. This is nothing but the classification of all subgroups of U(n)Q
U(n)u U(n)d according to the spectrum they produce. We think that this classification
is essential for any mass-matrix modelling.
Let us apply our method of parameter counting to a more complicated example. Let
us take an extension of the Standard Model including right-handed neutrinos with a
Majorana mass term. The leptonic part of the Lagrangian is
1
 
L = iL D
6 L + iR D 6 eR + L Y R + L Ye eR + Rc MR + h.c.
6 R + ieR D (17)
2
where L is the standard leptonic doublet, eR is the right-handed charged lepton singlet,
R is the right-handed neutrino, and Rc = CR T . The Yukawa couplings are the usual Ye
that give masses to the charged leptons, Y that give a Dirac mass term to neutrinos and
M is a complex symmetric Majorana mass matrix for right-handed neutrinos. It can be
put in by hand, as we did, because right-handed neutrinos are singlet, however, it could
be generated through the vacuum expectation value of some extra scalar singlet. The
model just described is nothing else than the see-saw model[16, 17, 18, 19] of neutrino
masses. If Y = Ye = M = 0 the Lagrangian is invariant under the following symmetries

L V L R V R eR Ve eR , (18)

V , V , and Ve are unitary matrices of dimension n for n generations of leptons. As in the

Standard Model, the Yukawa couplings break these symmetries explicitly. However, they
are broken in such a way that if we let the Yukawa couplings and the Majorana mass
transform in the following way

Ye V Ye Ve Y V Y V M V MV , (19)

the Lagrangian remains invariant. At difference with the Standard Model, we see that
the Yukawa couplings and the Majorana mass term completely break all the invariances
of eq. (18), including lepton-number conservation.
Equation (19) defines the following equivalence relation

(Ye , Y , M) (Ye , Y , M ) = (V Ye V , V Y Ve , V MV ) . (20)

Again the number of physical parameters needed to describe the model is given by the
difference between the number of parameters contained in the Yukawa coupling and the
number of generators broken by the couplings.
Now the group of symmetries of the kinetic and gauge part is G = U(n) U(n)e
U(n) and there is no unbroken subgroup. Counting of parameters is easy. Taking into
account that an arbitrary complex matrix of dimension n contains n2 moduli and n2
phases, that a symmetric complex matrix contains n(n + 1)/2 moduli and n(n + 1)/2
phases, and that a U(n) matrix contains n(n 1)/2 moduli and n(n + 1)/2 phases, we
find that Yphys can be written in terms of

couplings and symmetries moduli phases

2 2
(Ye , Y , M) 2n + n(n + 1)/2 n + n(n + 1)/2
U(n) U(n)e U(n) 3n(n 1)/2 3n(n + 1)/2 . (21)
0 0
Yphys 3n + n(n 1) n(n 1)

We obtain that even in the two-generation case there are two observable phases that
could violate CP. As we did in the Standard Model, to know how many of the physical
parameters are related to masses and how many to mixings, we have to decouple all kinds
of massive bosons. Then, all the mass matrices can be diagonalized, leading in general
to n massive charged leptons and 2n Majorana neutrinos. Thus, 3n of the moduli are
masses, and the rest correspond to mixings.
To find a parametrization of the Yphys, we can use the equivalence relation in eq. (20)
to reduce the number of parameters in the Yukawa couplings. Using the fact that an
arbitrary complex matrix can be written in a unique way as a unitary matrix times a
positive-definite Hermitian matrix4 and that a complex symmetric matrix can be written
in a unique way in terms of a positive-definite diagonal matrix and a unitary matrix
S = U T DU we can write the Yukawa couplings in the following form

(Ye , Y , M) Yphys = (He , H , Dm ) . (22)

Here He and H are positive-definite Hermitian matrices and Dm is a positive-definite

diagonal matrix. As we have exhausted all the symmetries in the group G to reduce the
Yukawa couplings to this form and there is no subgroup of G that leaves Yphys invariant,
the set of parameters in the right-hand side of eq. (22) is the physical set of parameters
needed to describe the theory. This means that we could start from the beginning without
loss of generality by choosing Y and Ye as positive-definite Hermitian matrices and M
as a positive definite diagonal matrix and all the parameters in those matrices would be
observable as can be easily checked by counting the number of parameters contained in
those couplings.
Finally to give a flavour of how these techniques can be used to study more exotic
models, we chose the model of refs. [20, 21, 22, 23] with the following Lagrangian,

L = iL D
6 L + iR D
6 R + ieR D
6 eR + isL D
6 sL
1
 
+ L Y R + L Ye eR + R Ms sL + h.c. , (23)
2
where sL is a singlet that carries the same lepton-number as the leptons and Ms is an
arbitrary complex matrix. The rest of the notation is the same as in the previous example.
In the Lagrangian of eq. (23), total lepton-number has been imposed as a global symmetry.
There are thus no Majorana mass terms for any of the singlet fermions. The group of
invariances of the non-Yukawa part of the Lagrangian is U(n) U(n)e U(n) U(n)s ,
where U(n)s is the new invariance related to the field sL . The only symmetry of the
Yukawa couplings is just U(1)lep of lepton-number conservation. Doing now the usual
4
We will assume that the matrices have non-zero determinant and there are no degeneracies. If there
is some degeneracy or zero eigenvalues, this must be imposed with some symmetry and would change the
counting of parameters.
counting we get

couplings and symmetries moduli phases

2
(Ye , Y , Ms ) 3n 3n2
U(n) U(n)e U(n) U(n)s 4n(n 1)/2 4n(n + 1)/2 . (24)
U(1)lep 0 1
2
Yphys 2n + n (n 1)2

Once the gauge symmetry is broken, the Yukawa couplings generate mass terms for the
leptons. It is easy to see that, as a consequence of lepton-number conservation, all neu-
trinos must be Dirac neutrinos or just massless. Then, from the full mass matrix of
neutrinos, n massless neutrinos and n massive Dirac neutrinos arise, hence only n masses
come from the neutrino mass matrix. The other n masses come from the charged lepton
sector. Thus, from all the moduli, 2n of them correspond to masses and the rest n2 cor-
respond to mixings. In addition there are (n 1)2 CP-violating phases that cannot be
removed. This result is in complete agreement with the results obtained in [24] after full
diagonalization of the mass matrices.
We have shown that a study of the chiral symmetries of the kinetic and gauge parts
of the Lagrangian and the Yukawa part of the Lagrangian before spontaneous symmetry
breaking allows us to compute, in a straightforward way, how many observable parameters
come from the Yukawa couplings in a general gauge theory. Basically, we have to know
which subgroup of chiral symmetries of the kinetic and gauge parts of the Lagrangian
preserves the form of the Yukawa couplings and which group leaves them invariant. The
number of parameters that can be absorbed from the Yukawa sector is just the difference
between the parameters needed to describe those groups. Then, the number of observable
parameters comes from the balance between the Yukawa couplings and the symmetries
broken by them. The method is useful when the diagonalization of the full set of mass
matrices becomes complicated. On the other hand it clarifies the analysis of observable
parameters. We explain the technique with several examples based on the Standard
Model, with Yukawa couplings obeying various chiral symmetries, and also with some
extensions of the Standard Model with enlarged fermion sector.
The study of the mass matrices and the gauge couplings is not general enough, because
some parameters such as phases, mixings, etc, can be moved from charged currents to
Yukawa couplings or even be included in a field definition [5, 6]. Only the analysis of
the full Lagrangian, and this is better done before symmetry breaking, can report the
number of observable parameters in a non-ambiguous way. This point of view suggests
a parametrization of Yukawa couplings in terms of invariants with respect to the chiral
symmetries of the kinetic and gauge parts of the Lagrangian that can be useful in the study
of the behaviour of Yukawa couplings under the renormalization group. We comment on
this parametrization in the case of the Standard Model and the see-saw model for neutrino
masses.
We find that a complete analysis and classification of all the chiral symmetries that
can be imposed on the Yukawa sector, according to which spectrum of masses and mixings
they originate is still missing. We think this analysis would be a good starting point for
mass matrix modeling.
Finally we want to remark that we only considered symmetries acting on fermions.
The analysis could be extended to symmetries involving fermions and scalars as well, like
the Peccei-Quinn [25, 26] symmetry or horizontal symmetries [27, 28, 29, 30]. We also
think that it can be extended to the spontaneous CP-violation case.
Acknowledgments
We thank F. Botella for helpful discussions on the subject of this paper and A. Pich for
a critical reading of the manuscript. This work has been supported in part by CICYT,
Spain, under grant AEN90-0040.

References
[1] H. Fritzsch, Nucl. Phys. B155 (1979) 189
[2] B. Stech, Phys. Lett. B130 (1983) 189
[3] M.V. Barnhill, Phys. Rev. D36 (1987) 192
[4] G. Giudice, Mod. Phys. Lett. A7 (1992) 2429
[5] J. Bernabeu and P. Pascual, Nucl. Phys. B228 (1983) 21
[6] J. Bernabeu, A. Pich and A. Santamaria, Z. Phys. C30 (1986) 213
[7] M. Kobayashi and K. Maskawa, Prog. Theor. Phys. 49 (1973) 652
[8] F.J. Botella and Ling-Lie Chau, Phys. Lett. 168B (1986) 97
[9] C. Jarlskog, Phys. Rev. Lett. 55 (1985) 1039
[10] K.S. Babu, Z. Phys. C35 (1987) 69
[11] M. Olechowski and S. Pokorski, Phys. Lett. B231 (1989) 165
[12] M. Olechowski and S. Pokorski, Phys. Lett. B257 (1991) 388
[13] J.E. Kim, Phys. Rep. 150 (1987) 1
[14] H.Y. Cheng, Phys. Rep. 158 (1988) 1
[15] R.D. Peccei, in CP-violation, ed. C. Jarlskog (World Scientific, Singapore, 1989)
[16] M. Gell-Mann, P. Ramond and R. Slansky, in Supergravity, eds. D. Freedman et al.
(North-Holland, Amsterdam, 1979)
[17] T. Yanagida, KEK lectures, ed. O. Swada et al. 1979
[18] R. Mohapatra and G. Senjanovic, Phys. Rev. Lett. 44 (1980) 912
[19] R. Mohapatra and G. Senjanovic, Phys. Rev. D23 (1981) 165
[20] J. Bjorken, K. Lane and S. Weinberg, Phys. Rev. D16 (1977) 1474
[21] T.P Cheng and L.F. Li, Phys. Rev. D16 (1977) 1425
[22] D. Wyler and L. Wolfenstein, Nucl. Phys. B218 (1983) 205
[23] J. Bernabeu, A. Santamaria, J. Vidal, A. Mendez and J. W.F. Valle, Phys. Lett.
B187 (1987) 303
[24] G.C. Branco, M. Rebelo and J.W.F. Valle, Phys. Lett. B225 (1989) 385
[25] R.D. Peccei and H.R. Quinn, Phys. Rev. Lett. 38 (1977) 1440
[26] R.D. Peccei and H.R. Quinn, Phys. Rev. D16 (1977) 1791
[27] R. Gatto, G. Morchio and F. Strocchi, Phys. Lett. 83B (1979) 348
[28] G. Segre and H.A. Weldon, Phys. Lett. 86B (1979) 291
[29] F. Wilczek and A. Zee, Phys. Rev. Lett. 42 (1979) 421
[30] R. Gatto, G. Morchio, G. Sartori and F. Strocchi, Nucl. Phys. B163 (1980) 221