Symmetries and renormalisation in two-Higgs-doublet models

# Symmetries and renormalisation in two-Higgs-doublet models

M. Maniatis Fakultät für Physik, Universität Bielefeld, 33615 Bielefeld, Germany    O. Nachtmann Institut für Theoretische Physik, Philosophenweg 16, 69120 Heidelberg, Germany
###### Abstract

We discuss the classification of symmetries and the corresponding symmetry groups in the two-Higgs-doublet model (THDM). We give an easily useable method how to determine the symmetry class and corresponding symmetry group of a given THDM Higgs potential. One of the symmetry classes corresponds to a Higgs potential with several simultaneous generalised CP symmetries. Extending the CP symmetry of this class to the Yukawa sector in a straightforward way, the so-called maximally-CP-symmetric model (MCPM) is obtained. We study the evolution of the quartic Higgs-potential parameters under a change of renormalisation point. Finally we compute the so called oblique parameters , , and , in the MCPM and we identify large regions of viable parameter space with respect to electroweak precision measurements. We present the corresponding allowed regions for the masses of the physical Higgs bosons. Reasonable ranges for these masses, up to several hundred GeV, are obtained which should make the (extra) Higgs bosons detectable in LHC experiments.

preprint: BI-TP 2011/14

## I Introduction

In today’s particle physics one of the main hunting grounds of theorists and experimentalists alike are scalars. In the Standard Model (SM) we have as scalar one Higgs-boson doublet field, playing an essential role. It is supposed to be responsible for electroweak symmetry breaking thereby giving mass to the  and bosons as well as to quarks and leptons. However, more complicated Higgs sectors are by no means excluded experimentally. On the contrary, there are good theoretical reasons for more than one Higgs-boson doublet field. Extended Higgs sectors are, for instance, required in supersymmetric models; see for instance Fayet:1974fj (); Fayet:1974pd (); Inoue:1982pi (); Inoue:1983pp (); Flores:1982pr (); Gunion:1984yn (), and in many models trying to solve the so called strong CP problem Peccei:1977hh (); Peccei:1977ur ().

One simple extension of the SM scalar sector has two Higgs-boson doublet fields. This two-Higgs-doublet model (THDM) has been studied extensively in the literature; see Lee:1973iz (); Deshpande:1977rw (); Georgi:1978xz (); Haber:1978jt (); Donoghue:1978cj (); Golowich:1978nh (); Hall:1981bc (); Haber:1993an (); Cvetic:1993cy (); Botella:1994cs (); Lavoura:1994fv (); Lavoura:1994yu (); Velhinho:1994vh (); Bernreuther:1998rx (); Davidson:2005cw (); Ginzburg:2004vp (); Barbieri:2005kf (); Nagel:2004sw (); Maniatis:2006fs (); Nishi:2006tg (); Ivanov:2006yq (); Maniatis:2007vn (); Ferreira:2009wh (); Mahmoudi:2009zx (); Grzadkowski:2010dj () and references therein. In our group we have, in particular, emphasised the usefulness of gauge-invariant bilinears for studying properties of THDMs and we have introduced a special THDM, the maximally CP-symmetric model (MCPM) which may give some understanding of the family structure and the fermion mass hierarchies observed in Nature Maniatis:2007de (). Predictions of the MCPM for high-energy proton–antiproton and proton–proton collisions were presented in Maniatis:2009vp (); Maniatis:2009by (); Maniatis:2010sb ().

THDM’s with additional symmetries were studied in Ivanov:2007de (); Ma:2009ax (); Ferreira:2010hy (). A review of the relation between the usual field formalism and the geometric picture for THDMs working with field bilinears was given in Ferreira:2010yh ().

In the present work we make some remarks concerning symmetries and the corresponding groups for THDMs. We discuss the renormalisation procedure in view of the symmetry constraints on the potential parameters. As an explicit example we treat the renormalisation of the dimension-four couplings in the MCPM. Finally we calculate the so called oblique parameters , ,  Peskin:1990zt () for the MCPM. Comparing with electroweak precision data we derive restrictions on the masses of the (extra compared to the SM) Higgs bosons for the MCPM.

## Ii The bilinear formalism

We consider models with the particle content as in the SM but with two Higgs-boson doublets

 φi(x)=(φ+i(x)φ0i(x)), (1)

where . Both doublets are assigned weak hypercharge . We use the conventions for kinematics etc. as in Maniatis:2007de (). The most general gauge invariant and renormalisable potential of the THDM may be written in terms of fields as Haber:1993an ()

 V(φ1,φ2)=+m211(φ†1φ1)+m222(φ†2φ2)−m212(φ†1φ2)−(m212)∗(φ†2φ1)+λ12(φ†1φ1)2+λ22(φ†2φ2)2+λ3(φ†1φ1)(φ†2φ2)+λ4(φ†1φ2)(φ†2φ1)+12[λ5(φ†1φ2)2+λ∗5(φ†2φ1)2]+[λ6(φ†1φ2)+λ∗6(φ†2φ1)](φ†1φ1)+[λ7(φ†1φ2)+λ∗7(φ†2φ1)](φ†2φ2), (2)

with , , real, , complex. To study the properties of the Higgs potential, it is convenient to write it in terms of field bilinears Nagel:2004sw (); Maniatis:2006fs (); Nishi:2006tg (); Ivanov:2006yq (). In Nagel:2004sw (); Maniatis:2006fs () a one-to-one correspondence of bilinear gauge-invariant expressions with a Minkowski-type four vector was revealed leading to a simple geometric interpretation. We arrange the fields  of (1) in a matrix

 ϕ(x)=(φ+1(x)φ01(x)φ+2(x)φ02(x)) (3)

and define the hermitian, positive semi definite, matrix

 K––(x):=ϕ(x)ϕ†(x)=(φ†1(x)φ1(x)φ†2(x)φ1(x)φ†1(x)φ2(x)φ†2(x)φ2(x)). (4)

 K––(x)=12(K0(x)\mathbbm12+K(x)σ) (5)

with Pauli matrices . In this way one defines the real bilinears

 K0(x)=φ†1φ1+φ†2φ2,K1(x)=φ†1φ2+φ†2φ1,K2(x)=iφ†2φ1−iφ†1φ2,K3(x)=φ†1φ1−φ†2φ2. (6)

We have

 K0(x)≥0,(K0(x))2−(K(x))2≥0. (7)

In terms of these bilinears the general THDM potential (2) can be written in the simple form

 V(φ1,φ2)=ξ0K0(x)+ξTK(x)+η00K20(x)+2K0(x)ηTK(x)+KT(x)EK(x), (8)

with and parameters , , three-component vectors , and the matrix . All parameters in (8) are real. The translation from the conventional parameters to the bilinear parameters is

 ξ0=12(m211+m222),ξ=12⎛⎜ ⎜⎝−2Re(m212)2Im(m212)m211−m222⎞⎟ ⎟⎠,η00=18(λ1+λ2)+14λ3,η=14⎛⎜ ⎜⎝Re(λ6+λ7)−Im(λ6+λ7)12(λ1−λ2)⎞⎟ ⎟⎠,E=14⎛⎜ ⎜⎝λ4+Re(λ5)−Im(λ5)Re(λ6−λ7)−Im(λ5)λ4−Re(λ5)−Im(λ6−λ7)Re(λ6−λ7)−Im(λ6−λ7)12(λ1+λ2)−λ3⎞⎟ ⎟⎠. (9)

We also define the   matrix of the parameters corresponding to the potential terms quadratic in the bilinears,

 ~E=(η00ηTηE)=⎛⎜ ⎜ ⎜ ⎜⎝η00η01η02η03η01η11η12η13η02η12η22η23η03η13η23η33⎞⎟ ⎟ ⎟ ⎟⎠. (10)

Since both Higgs doublets carry the same quantum numbers we may also consider the unitarily mixed fields

 (φ1(x)φ2(x))→(φ′1(x)φ′2(x))=U(φ1(x)φ2(x)), (11)

with . For the bilinears a basis, or Higgs-family, transformation (11) of the fields corresponds to a SO(3) rotation given by

 K0(x)→K′0(x)=K0(x),K(x)→K′(x)=R(U)K(x). (12)

Here is obtained from

 U†σaU=Rab(U)σb. (13)

We note that every proper rotation matrix  is a rotation about an axis and can be represented, in a suitable basis, as

 Rα=⎛⎜⎝cos(α)−sin(α)0sin(α)cos(α)0001⎞⎟⎠, (14)

where  is the angle of rotation.

We shall also consider generalized CP (GCP) transformations Lee:1966ik (); Ecker:1981wv (); Ecker:1983hz (); Bernabeu:1986fc (); Ecker:1987qp (); Neufeld:1987wa (); Lavoura:1994fv (); Botella:1994cs (), where

 φi(x)→Uijφ∗j(x′),i,j=1,2,x=(x0,x),x′=(x0,−x) (15)

with . Note that the ordinary CP transformation is the special case of in (15). In space the generalized CP transformations (15) correspond to the improper rotations Maniatis:2007vn (); Ferreira:2009wh ()

 K0(x)→K0(x′),K(x)→K′(x′)=¯R(U)K(x′). (16)

Here

 ¯R(U)=R(U)¯R2 (17)

with  the matrix for reflection on the 1–3 plane. We define the matrices () for the reflections on the coordinate planes in  space as

 ¯R1=diag(−1,1,1),¯R2=diag(1,−1,1),¯R3=diag(1,1,−1). (18)

Here and in the following proper rotation matrices will be denoted by , , etc., improper rotation matrices by , , , etc. By a suitable basis choice we can always arrange that the improper rotation matrix  has the form

 ¯Rα=⎛⎜⎝cos(α)−sin(α)0sin(α)cos(α)000−1⎞⎟⎠,with 0≤α≤π. (19)

Note that for we get the GCP transformation corresponding to a reflection on the 1–2 plane in  space () accompanied by the space-time transformation . A basis transformation (12) exchanging the 2 and 3 axes in  space shows that this is equivalent to the standard CP transformation where  in (16). For more details on GCPs in THDMs see Maniatis:2007vn (); Ferreira:2009wh (); Ferreira:2010hy ().

Finally we recall from Maniatis:2006fs () that a transformation (12) in space with always corresponds to a field transformation (11) which is unique up to gauge transformations. Similarly, a -space transformation (16) with , , always corresponds to a GCP transformation (15) of the fields which is unique up to gauge transformations.

## Iii Symmetry classes and symmetry groups

The general THDM potential has 14 parameters; see (9). Considering only the scalar sector we can make a basis change as in (11), (12) to diagonalise , thereby reducing the number of parameters to 11. One may want to further reduce this number by imposing symmetries. This can be Higgs-family or GCP symmetries. A Higgs-family transformation (11), (12) is a symmetry of the potential if and only if the parameters (9) satisfy

 R(U)ξ=ξ,R(U)η=η,R(U)ERT(U)=E. (20)

A GCP transformation (15), (16) is a symmetry if and only if

 ¯R(U)ξ=ξ,¯R(U)η=η,¯R(U)E¯RT(U)=E. (21)

In Ivanov:2007de () the possible symmetry classes of THDMs were derived, however, only potentials which are stable in the strong sense were considered. Here we define, as in Maniatis:2006fs (), a potential to be stable in the strong sense if stability is guaranteed by the quartic field terms alone and in the weak sense if it is guaranteed only after inclusion of the quadratic field terms in (2) respectively (8). A potential being bounded from below but having directions in field space where it does not grow indefinitely for the fields going to infinity has only marginal stability. In all other cases the potential is unstable. In Ferreira:2010hy () the symmetry classes of the THDMs were further studied and also softly broken symmetries were considered.

We give in Table 1 the maximal symmetry group for each symmetry class and the corresponding constraints on the potential (8). Note that in Table 1 the classes are defined to be mutually exclusive, that is, we assign a THDM to a certain class if it has the corresponding group  (up to trivial equivalences) as symmetry group and not a bigger one. If the parameters of a THDM potential are not satisfying any of the constraints of Table 1, the theory has no symmetry group except the trivial one, that is, the unit transformation. In appendix A we present a derivation of these symmetry classes and groups where, as mentioned above, we do not use any assumptions on the stability of the potential (2), (8). The methods explained in apppendix A also give an easy practical recipe for finding out if a THDM potential has a symmetry and which one this is. The symmetry relations as given in Table 1 will be used in section IV for the discussion of the renormalisation in specific THDMs.

We emphasize that in Table 1 we give the exact conditions for the parameters of the scalar potential to have the symmetry group  as listed and not a bigger one. The elements of  give the corresponding transformations in space. For proper rotations these are Higgs basis transformations; see eq. (12), for improper rotations these are generalized CP transformations; see eq. (16). Of course, a group of a symmetry class may contain the groups of other classes as subgroups, as is obvious from Table 1. For instance, the group O(3) contains all other groups as subgroups and, clearly, the potential of the SO(3) symmetry class has all other symmetries as well. The numbering of the eigenvalues of  in Table 1 is - without loss of generality - chosen conveniently, in order to give the same invariance group and not an equivalent one for all subclasses of one class. For the cases of degenerate eigenvalues of  it is understood that a convenient choice of basis in the degenerate subspaces gives the groups as listed. Other choices of bases give equivalent groups.

In Table 1 we have listed subclasses for , U(1), and CP1. These are distinguished by the degeneracies of the eigenvalues  and for the CP1 case also by relations for and . These subclasses of a class correspond to the same symmetry group and therefore lead to no new symmetry classes. Under renormalisation only the groups will be preserved. That is, the subclasses of one class will not be invariant under renormalisation but will mix among each other. Considering the theory of the two Higgs-boson doublets alone the renormalisation of the potential parameters can not lead from one symmetry class to another one. If we start, for instance, with a theory of the CP2 class where all are different we can not come by renormalisation to the CP3 or SO(3) classes where two, respectively all three, of the ’s are equal. We shall elaborate on this point below in section IV in connection with the renormalisation in the MCPM which is a complete theory including fermions and bosons.

The elements of the various symmetry groups in Table 1 are listed according to their action in  space; see (12), (16). For completeness we list in appendix A also the corresponding transformations for the fields.

## Iv Renormalisation of the dimension four couplings in the MCPM

In this section we consider the renormalisation-group equations (RGEs) for the dimension four couplings in the maximally CP symmetric model (MCPM) as constructed and studied in Maniatis:2007de (); Maniatis:2009vp (); Maniatis:2009by (); Maniatis:2010sb (). In the MCPM the Higgs potential parameters (9), in a diagonal basis of the matrix , have to fulfill

 ξ=0,η=0,E=diag(μ1,μ2,μ3). (22)

In conventional notation of the Higgs potential (2) this corresponds to the constraints

 m212=0,m211=m222,λ1=λ2,Im(λ5)=0,λ6=λ7=0. (23)

Without loss of generality we can assume

 μ1≥μ2≥μ3. (24)

From Table 1 we see that the Higgs potential satisfying (22) can be in the symmetry classes CP2, CP3, or SO(3). As shown in Maniatis:2007de (), stability, the correcte electroweak symmetry breaking (EWSB), and absence of zero-mass charged Higgs bosons require and are guaranteed by

 η00>0,μi+η00>0for i=1,2,3,ξ0<0,μ3<0. (25)

In the MCPM there are five physical Higgs bosons, three neutral ones, , , , and a charged pair, . Their squared masses in terms of the model parameters are, at tree level,

 m2ρ′=2v20(η00+μ3),m2h′=2v20(μ1−μ3),m2h′′=2v20(μ2−μ3),m2H±=2v20(−μ3). (26)

Here

 v0=√−ξ0η00+μ3≈246 GeV (27)

is the standard vacuum-expectation value. Requiring now also absence of zero-mass neutral Higgs bosons and absence of mass degeneracy between and leads to

 μ1>μ2>μ3, (28)

replacing the weaker condition (24). From Table 1 we see that we are dealing now with potentials in the CP2 symmetry class with the corresponding symmetry group  as listed there. The main point of the MCPM is that the symmetry group of the CP2 class is required to be respected also by the complete Lagrangian, including the fermions, the gauge-boson, and the Yukawa sectors. It was shown in Maniatis:2007de () that with this requirement a coupling of the two Higgs-boson doublets to only one fermion family is not possible with non-vanishing Yukawa couplings. However, with a coupling of the two Higgs-boson doublets to two fermion families this is indeed possible with one fermion family acquiring masses and the other remaining massless. With a third fermion family kept uncoupled to the Higgs-boson doublets this model gives very roughly what we observe in Nature: two rather light fermion families and one very heavy (the third) fermion family.

The complete Lagrangian of the MCPM is recalled in App. B. The parameters of the MCPM are as follows.

• Higgs potential parameters

 ξ0,η00,μ1,μ2,μ3. (29)
• Yukawa sector coupling constants

 cτ,ct,cb (30)

related to the third-fermion-family masses

 mτ=cτv0√2,mt=ctv0√2,mb=cbv0√2. (31)
• Gauge couplings

 g1,g2,g3 (32)

of the gauge groups , , and , respectively.

Let us now proceed and consider the one-loop RGEs in this model. The one-loop RGEs for the couplings of the dimension-four terms in any renormalisable gauge theory are given in Cheng:1973nv (). The RGEs given there apply to the deep Euclidean region where coupling terms of dimension two can be neglected. Also shifts of scalar fields to give them zero vacuum expectation value after EWSB are irrelevant there. For the quartic Higgs-potential couplings including the and gauge interactions with couplings and , respectively, taking also the Yukawa couplings (68) into account we find for the MCPM from the results of Cheng:1973nv ()

 8π2dλ1dt=6λ21+2λ23+2λ3λ4+λ24+λ25−λ1(32g21+92g22)+38g41+34g21g22+98g42+2λ1(c2τ+c2b+c2t)−2(c4τ+c4b+c4t),8π2dλ3dt=2λ1(3λ3+λ4)+2λ23+λ24+λ25−λ3(32g21+92g22)+38g41−34g21g22+98g42+λ3(c2τ+c2b+c2t),8π2dλ4dt=2λ1λ4+4λ3λ4+2λ24+4λ25−λ4(32g21+92g22)+32g21g22+λ4(c2τ+c2b+c2t),8π2dλ5dt=λ5(2λ1+4λ3+6λ4)−λ5(32g21+92g22)+λ5(c2τ+c2b+c2t),dλ2dt=dλ1dt,dλ6dt=dλ7dt=0. (33)

Here with the mass scale of the renormalisation point and  a convenient reference scale, for instance,  TeV. The RGEs of the ’s can easily be translated to space. For the generic THDM Higgs potential this was done in Ma:2009ax (). In the case of the MCPM we have to extend these RGEs by including the Yukawa interactions (68). From the RGEs for the parameters of the generic Higgs potential as given in Ma:2009ax () we can check that the diagonality of the matrix  and , see (22), are preserved under one-loop renormalisation in the MCPM. This must be so, since this is guaranteed by the symmetry group  of the CP2 class; see Table 1. Here we find from (9), (10), and (33),

 8π2dη00dt=4η200+η00(η11+η22+η33)+η211+η222+η233−η00(32g21+92g22)+34g41+94g42+(32η00+12η33)(c2τ+c2b+c2t)−12(c4τ+c4b+c4t),8π2dη11dt=η11(3η00+3η11−η22−η33−32g21−92g22+c2τ+c2b+c2t)+32g21g22,8π2dη22dt=η22(3η00−η11+3η22−η33−32g21−92g22+c2τ+c2b+c2t)+32g21g22,8π2dη33dt=η33(3η00−η11−η22+3η33−32g21−92g22)+32g21g22+(12η00+32η33)(c2τ+c2b+c2t)−12(c4τ+c4b+c4t). (34)

As mentioned above these RGEs apply in the deep Euclidean region.

Let us now discuss the evolution of the differences of the eigenvalues of :

 μ1−μ2,μ2−μ3. (35)

From (34) we find

 (36)

Suppose now that we start at TeV, corresponding to , with the conditions (28). We have then, in particular,

 [μ1(t)−μ2(t)]∣∣t=0>0. (37)

From (36) we can see that the one loop RGEs preserve this property as long as all couplings stay finite. Indeed, suppose that for we have

 ∣∣3(η00+μ1+μ2)−μ3−32g21−92g22+c2τ+c2b+c2t∣∣≤8π2C, (38)

where is a constant. We get then from (36)

 8π2ddtln(μ1−μ2)=3(η00+μ1+μ2)−μ3−32g21−92g22+c2τ+c2b+c2t, (39)
 −C≤ddtln(μ1−μ2)≤C, (40)
 e−Ct≤μ1(t)−μ2(t)μ1(0)−μ2(0)≤eCt,for 0≤t≤t1. (41)

Thus, stays positive for . A similar argument applies for the evolution to negative  values. Hence, can not change sign as long as the theory parameters stay finite.

The analogous result for  can not be derived in the same way from (36). This is due to the terms not proporotional to on the r.h.s of (36). But in the pure scalar theory, that is, if we set and we can again derive the analogue of (41).

We conclude that the one loop RGEs preserve but, in the full theory, not necessarily If now for some -value we have we have for the Higgs potential a higher symmetry, here the CP3 symmetry, where two eigenvalues of  are equal; see Table 1. But for the full theory this CP3 symmetry is not realised. Thus, in the full theory the RGEs can lead to renormalisation scales  where the Higgs potential alone shows a higher symmetry than the full theory. Of course, only the symmetry of the full theory is relevant for physics.

## V Oblique parameters in the MCPM

The oblique parameters , , and denote certain combinations of self-energies of the electroweak gauge bosons with respect to any new contributions compared to the SM Peskin:1990zt (). In any model beyond the SM the oblique parameters can be computed and compared to the electroweak precision data Nakamura:2010zzi () which require:

 S=0.01±0.10,T=0.03±0.11,U=0.06±0.10. (42)

For the case of the general THDM the oblique parameters have been computed in Froggatt:1991qw (); Haber:2010bw ().

We shall now derive the predictions for the oblique parameters in the MCPM. In the MCPM the Yukawa couplings are completely fixed and the only free parameters we encounter in the calculation of the oblique parameters are the Higgs-boson masses , , , and . Here and are the CP-even and is the CP-odd Higgs boson and denotes the pair of charged Higgs bosons. In Figure 1 we show the contour plots for the 1- (dark) and 2- (bright) deviations of the oblique parameters from the electroweak precision data (42) in the  plane. The mass of the SM-like Higgs-boson  is fixed to  125 GeV. The charged-Higgs-boson mass is set to different values in the range of 125-500 GeV in the various plots. In Figure 2 we show analogous plots but for a mass of the SM-like Higgs boson  of  170 GeV. Note that we have always in the MCPM which is the reason that there are no allowed regions of parameter space above the diagonal of equal masses in Figures 1 and 2

We see from Figures 1 and 2 that there are large regions for the masses of the Higgs bosons , , and where the electroweak constraints (42) are satisfied. The allowed regions for these masses, up to several hundred GeV, are very reasonable. The CP odd extra Higgs boson  could even be below 100 GeV in mass. But then it would be necessary to study all other experimental constraints for such a low-mass boson. Furthermore, we see from Figures 1 and 2 that with increasing masses of  and also the allowed domains for the masses of the Higgs bosons and shift to higher mass values.

## Vi Conclusions

In this paper we started with briefly reviewing the bilinear formalism which turns out to be quite powerful for the study of the THDM. We have discussed the classification of the possible symmetry classes without any assumption on the stability type of the THDM potential. We have given a practical and easily usable method how to determine the symmetry class of a given THDM Higgs potential. We have defined the symmetry classes to be mutually exclusive; see Table 1. We have also given the symmetry group  corresponding to each symmetry class. We have focussed on one of these symmetry classes, denoted by CP2, in some detail. The CP2 symmetric THDM has a number of simultaneous CP invariances. As shown in Maniatis:2007de () the extension of the CP symmetries of the potential to the Yukawa interactions leads in a straightforward way to the so-called maximally CP-symmetric model (MCPM). In this model the Yukawa couplings are completely fixed. We have studied the renormalisation-group equations of the quartic Higgs-potential parameters in this model. We have found that the symmetries of this model are preserved by the RGEs, as it has to be.

The MCPM has a hierarchy of quartic couplings . We have shown that considering the theory of the Higgs bosons alone this hierarchy of quartic couplings turn out to be stable against renormalisation group evolution. However, taking the Yukawa couplings into account is stable but not necessarily . Reaching at a certain renormalisation scale would elevate the CP2 symmetry of the Higgs potential to a CP3 symmetry. But, of course, this does not imply that the full MCPM which includes fermions and gauge bosons has a higher symmetry than CP2 at this renormalisation scale.

Eventually, we have computed the oblique parameters in the MCPM. We find for large parameter space agreement with the electroweak precision measurements. In particular we have presented the 1- and 2- contours of valid regions in the mass plane for different choices for the charged-Higgs-boson mass and for SM-like Higgs-boson masses of 125 GeV and 170 GeV, respectively. The allowed regions for the masses of the Higgs bosons are in a reasonable range; see Figures 1 and 2. These Higgs bosons with masses below 500 GeV should therefore be detectable in the LHC experiments. As shown in Maniatis:2009vp (); Maniatis:2009by (); Maniatis:2010sb () in the MCPM these Higgs bosons have characteristic production and decay properties giving clear experimental signatures.

## Appendix A Derivation of symmetry classes

In this appendix we give a recipe which allows an easy identification of the symmetry class of any given THDM potential (8).

The first step is to diagonalise  by a basis transformation (12). We get then

 E=diag(μ1,μ2,μ3). (43)

Since  is symmetric a diagonalisation is always possible. Therefore we work in the following in the diagonal basis and consider  and  (9) in this basis. Now we have to distinguish three cases for the ’s.

• , , all different.

Then we see from (20) and (21) that only diagonal O(3) matrices or may lead to symmetries, that is, we have to consider

 ¯R=¯Rj(j=1,2,3;see (???)),¯R=¯R1¯R2¯R3=−\mathbbm13,R=¯Ri¯Rjwith i≠j. (44)

Now we can easily check the conditions for , from (20) and (21). We can have the following cases:

• .

With none of the matrices from (44) we can fulfill the symmetry relations in (20). In this case the potential has only the trivial symmetry group .

• Exactly one pair fulfills where .

Without loss of generality we can set , , . Clearly, from (20), (21) we have and nothing else as symmetry transformation. We get the symmetry group

 \mathbbmG={\mathbbm13,¯R2} (45)

which characterises the CP1 symmetry class; see Table 1, the first subclass of CP1.

• Exactly two pairs fulfill where .

Without loss of generality we can set , . From (20), (21) and (44) we see that the invariance group is

 \mathbbmG={\mathbbm13,¯R1,¯R2,¯R1¯R2}. (46)

We get the symmetry group characterising the symmetry class; see Table 1, the first subclass of .

• and .

Here we find from (20), (21) and (44) as symmetry group

 \mathbbmG={\mathbbm13,¯R1,¯R2,¯R3,¯R1¯R2,¯R2¯R3,¯R1¯R3,¯R1¯R2¯R3=−\mathbbm13}. (47)

This characterises the CP2 symmetry class.

• Exactly two eigenvalues of are equal.

Without loss of generality we set

 μ1=μ2≠μ3. (48)

From  (20), (21) we see that  allows now as invariances

 R2θ =⎛⎜⎝cos(2θ)−sin(2θ)0sin(2θ)cos(2θ)0001⎞⎟⎠0≤θ<π, (49) ¯R =¯Rj,j=2,3, (50) ¯R =R2θ¯R3, (51) ¯R =R2θ¯R2. (52)

Note that is included in (52) for : .

Now we consider again all possibilities for  and .

• and .

The first relation implies that neither  (50) nor any  (51) can lead to a symmetry. The second relation implies that and are linearly independent. Therefore, neither  (49) nor any  (52) can lead to a symmetry and we have here only the trivial invariance group

 \mathbbmG={\mathbbm13}. (53)
• , ,