Strong CP violation: problem or blessing?

Goran Senjanovi´c1, 2 and Vladimir Tello1 1International Centre for Theoretical Physics, Trieste, Italy 2Tsung-Dao Lee Institute & Department of Physics and Astronomy, SKLPPC, Shanghai Jiao Tong University, 800 Dongchuan Rd., Minhang, Shanghai 200240, China (Dated: June 15, 2020) We readdress the issue of strong CP violation both in the and in the Minimal Left-Right Symmetric Model and try to clear the confusion that seems to still pervade the field. We argue that the smallness of strong CP violation, while harmless and basically decoupled from the rest of physics in the SM, in the context of LR symmetry provides a blessing by helping to narrow down the parameter space of the theory and connecting apparently uncorrelated physical quantities. In particular, in the context of left-right symmetry being , it either points to the suppression of leptonic CP violation noticed before, or it leads to relatively light right-handed neutrinos, potentially accessible at the next hadron collider. The latter, more natural in view of complex Yukawa couplings, goes hand in hand with the smallness of lepton flavor violation and enhances the possibility of observing neutrinoless double beta decay.

I. STRONG CP VIOLATION IN THE We are rushing ahead of ourselves though, first things STANDARD MODEL first. The experimental limit from the electric dipole mo- ment of the implies a strong constraint [1] Strong CP violation provides a beautiful interplay be- θ¯ 10−10, (3) tween strong and electro-weak interactions. Its measure . manifestly illustrates this point with a similar limit arising form mercury atoms [2]. This has led to the proclamation of the strong CP problem. It ¯ θ = θ + arg det(MuMd), (1) is a strange proclamation for there is nothing that tells us what the value of θ¯ should be. It does not get renor- where in obvious notation we have up and down quark malised, being a non-perturbative phenomenon, and is mass matrices and θ is a vacuum selection parameter of thus self-protected. In this sense, it is similar to the the QCD defined by (a, b = 1, ...8 is the color index) small electron Yukawa coupling, although unlike in the latter case there is no protective symmetry. 1 θ The argument often does not stop here though. Some- L = − Ga Gµν,a + µναβGa Ga . (2) QCD 4 µν 16π2 µν αβ times, even in the context of the SM, one separates the two contributions in (1) and talks about the apparent Although we have separated the two contributions, from weak-interaction part arg det(MuMd). The argument θ and quark mass matrices, in the Standard Model the then goes as follows. We should expect this parameter to physical parameter is only θ¯, since neither θ nor quark be much larger due to the CP violation mass matrices can be measured on their own and the in the KM mixing matrix, and worse, its smallness will axial anomaly connects the two. Furthermore, it is im- not be respected in perturbation theory - hence the prob- portant to stress that θ is not an ordinary coupling, but lem of . This has been repeated often since its µναβ a a rather a boundary condition since  Gµν Gαβ is a sur- first conception more than forty years ago, in spite of be- face term. ing wrong. As shown in [3], the radiatively induced finite The separation becomes useful in a theory beyond the value of arg det(MuMd) appears at the three-loop level Standard Model that can address the issue of the her- and is vanishingly small, some nine orders of magnitude arXiv:2004.04036v2 [hep-ph] 11 Jun 2020 miticity of quark mass matrices. An example of such a bellow the experimental limit. theory is the Left-Right Symmetric Model of the spon- This should not come a surprise at all, it could be taneous breakdown of parity in weak interactions, the guessed (with some education) without doing the com- principal topic of our work. In order to set the stage, putation. First of all, the weak CP violation invariant and prepare the reader for a relatively painless discus- parameter is not just the phase δ, rather the tiny prod- −5 sion of our conclusions, we need to critically readdress uct θ1θ2θ3δ ' 10 , where θi are the CKM mixing angles. the question of strong CP violation in the SM. Namely, A quick glance suffices to see the further loop and quark it is generally considered to be the problem - and if this mass suppression - it should have been obvious that there was true, one would conclude the same for the LR sym- could not have been a strong CP problem in the SM. The metric model. We wish to convince the reader that the arg det(MuMd) is structurally protected in the SM to an prevailing wisdom is not correct, and that in reality this extraordinary degree. so-called problem can be a blessing in disguise for a the- Although reassuring, strictly speaking this has no ory beyond the SM - although strictly speaking in the physical meaning since arg det(MuMd) is not a physi- SM it is neither of the two. cal parameter in the SM. All one can do is to compute 2 the pure electro-weak part of the electric dipole of nuclei, true for the PQ mechanism that leads to the existence of which in the SM are extremely small, allowing, at least in the [8], among others a possible source of dark mat- principle, to measure the effect of θ¯ in the future [4]. In ter. What the axion field does is to provide the screening any case, the real task, whether in the SM or any theory of θ¯, but once again cannot say anything per se about the beyond, is to untangle the weak and strong contributions the value of θ¯ or better, the value of the electric dipole to the electric dipole moments. moment of nuclei (or atoms). We can conclude - ironically - that the only problem In the discrete symmetry approach one can obtain use- with the strong CP violation in the SM is that it is not ful constraints on the parameter space of a model in ques- a problem at all, for it offers no help in further reducing tion, unlike in the SM. Of particular interest is the left- the parameter space of the theory. More precisely, the right symmetric theory [9], put forward in order to have a strong CP violation does not constrain any of the physical dynamical theory of parity violation in weak interactions. parameters of the SM, besides of course θ¯ itself. Although This theory led to the existence to the right-handed (RH) this is recognised by the experts, it is often said that neutrino and neutrino mass, long before it was suggested one would like to know the tree-level value of θ¯, or at by experiment. In recent years, the modern version of least why it is so small, something that the SM cannot this theory, the Minimal Left-Right Symmetric Model address. Strangely enough, one equally often calls this (MLRSM) [10] has been shown to be a self-contained, issue a problem - and when you argue that there is a predictive theory of neutrino mass [11, 12], potentially problem, you must look for a solution. accessible at the LHC and it is thus of imperative impor- Before we do that, we wish to emphasise that nothing tance to study the consequences of strong CP violation we said is new or original, but we believe that it was in this theory. The situation depends critically whether crucial to put this issue in the proper light in order to one uses (generalised) parity P or (generalised) charge understand better its implications. If at this point we conjugation C as the LR symmetry, and must be treated succeeded in having the reader share our outlook on the separately. strong CP violation in the SM, she is likely to accept The case of P was discussed at length in [13] but we our findings in what follows. If not, if she persists in the believe it is still worth revisiting it in order to have a belief that the problem is genuine, and that one must clear grasp of the phenomenological implications. Just ’explain’ why θ¯ is so small, our conclusions will definitely like in the PQ approach, in this case one may tempted appear as unjustifiably euphoric. We continue though in to speak of having a theory of strong CP violation. In the spirit that there is nothing to explain, the smallness the PQ case, θ¯ variable becomes dynamical, the vacuum of strong CP violation is just another of small numbers expectation value of the axion filed - up to the explicit that plays a useful role in phenomenology, just the way PQ symmetry breaking terms that must (and can) be as- the smallness of flavour violation in neutral currents leads sumed small. In the MLRSM it is parity breaking that to small charm quark mass. is rendered dynamical - again up to the explicit P break- ing terms that must (and can) be assumed small, and this allows a unique opportunity to separate the θ and Strong CP: blessing for the BSM arg det(MuMd) contributions. Not surprisingly then, certain CP violating parameters The so-called “solutions” to this non-existing problem end up being tiny. Their stability requires either real neu- can be divided in two categories: (i) attribute parity P [5] trino Dirac Yukawa couplings (hard to keep without pro- and/or CP violation [6] to spontaneous breaking and (ii) tective symmetry) [14] - or the right-handed neutrinos promote θ¯ to a dynamical variable as in the case of the have to be somewhat light compared to the LR break- Peccei-Quinn mechanism (PQ) [7]. In both cases, how- ing scale. In the latter case, no fine-tuning is needed ever, one trades the smallness of the tree-level θ¯ of the whatsoever since the obtained limits are technically nat- SM for the smallness of parameters that break the sym- ural. Ironically - contrary from what is usually preached metries assumed in the above approaches. In the latter - strong CP violation helps in reducing the parameter case one has to assume, ad-hoc, that the Yukawa cou- space of the theory. Just as in the SM, it is not a prob- plings which violate the PQ symmetry must be ridicu- lem at all, but unlike in the SM case, it plays a useful lously small - and this is surely far less natural than the role in pinning down the theory. smallness of θ¯ itself. In the former case, similarly, one This leads to important correlations with other phys- must believe in basically exact discrete symmetries. But ical process, such as lepton number and lepton flavour then, what is wrong with simply proclaiming that θ¯ is violation. We should stress that the main contribution small in the SM? In this sense, basically all the proposed of our paper is to point out these important physical models based on these ideas fare no better than the SM connections which shed light on the strong CP violation in explaining the smallness of strong CP violation. in the MLRSM. The beauty of the strong CP issue lies This said, they may offer something new. Although in the fact that one’s conclusions depend crucial on the one may not know why θ¯ is so small, one may correlate it interpretation of the question and thus require great con- to other physical quantities. In other words, the strong ceptual care, which we believe is equally important as the CP violation can becomes dynamical. This is especially technical aspect itself. We have basically nothing to add 3 to the latter, and in all honesty, not so much even to the right SU(2) triplets ∆L(3, 1, 2) and ∆R(1, 3, 2), where former aspect, and an expert may be disappointed by the numbers in brackets denote the representation con- what follows - still, we believe that our discussion may tent under (4). These scalars have the following form help the reader get a clearer picture of this fundamental √ ! issue. δ+ / 2 δ++ ∆ = L,R L,R√ (8) On the other hand, the case C is equivalent to the SM L,R 0 + δL,R −δ / 2 situation where one can say nothing about the strong CP L,R violation per se, simply because the only physical param- The essential point is the spontaneous breaking of LR eter is θ¯, non-perturbative and thus insensitive to weak 0 0 symmetry achieved through vL = hδLi = 0, vR = hδRi= 6 corrections. Just as in the SM, all one can do is to com- 0. The vev vR is responsible for the masses of all non-SM pute the weak part of different electric dipole moments. particles, in particular the heavy gauge WR and We discuss this in the central sectionIII of this paper, ZR, and the RH neutrinos N, leaving only the SM gauge where both cases are addressed. symmetry. Before we ask the reader to follow our discussion, Next, the SM gauge symmetry breaking is achieved by though, in the next section we first review the salient the SU(2)L × SU(2)R bi-doublet Φ(2, 2, 0), which con- features of the MLRSM in order to ease her pain. Our tains two SM doublets conclusions are then left for the sectionIV, but it is worth  0  emphasising once again that we are not after explaining ∗ φi Φ = [φ1, iσ2φ2] , φi = − , i = 1, 2. (9) why the strong CP violation is so small for it would end φi up being a tautology as in the most self-proclaimed “ex- with the most general vev given by planations”. Instead, we simply wish to correlate this beautiful issue with other physical phenomena, which we hΦi = v diag(cos β, − sin βe−ia) (10) believe is what physics is all about. This is the same at- titude that [13] took correctly, but the confusion persists The quark Yukawa couplings take the following form when it comes to issues of naturalness and our aim is to q q ∗ once for all clear this up, at least in the context of the LY = −qL Y1 Φ − Y2 σ2Φ σ2) qR + h.c., (11) MLRSM, and to show that the resulting constraints are a blessing. while their leptonic analogs are given by ` ` ∗ LY = − `L(Y1 Φ − Y2 σ2Φ σ2)`R (12) II. MINIMAL LEFT-RIGHT SYMMETRIC 1 T T  − `LYLiσ2∆L`L + `RYRiσ2∆R`R + h.c. MODEL 2 Under parity, consistent with (6), one has The MLRSM is based on the following gauge symmetry † group ∆L ↔ ∆R, Φ → Φ (13)

GLR = SU(2)L × SU(2)R × U(1)B−L (4) which imply Hermitian Dirac type couplings of the bi- doublet (both for and leptons) and the same left augmented by a discrete LR symmetry which plays the and right couplings for the Higgs triplets role of left-right symmetry by interchanging left and right † SU(2) groups, and can be either P or C. We discuss Y1,2 = Y1,2,YL = YR. (14) only the central features of the MLRSM, a reader in need of a more in-depth review of the theory would benefit Since the bi-doublet vev is complex in general, see (10), from [15]. the quark mass matrices do not have to be hermitian. The sector is given by the LH and RH doublets An essential comment is called for. We do not claim that the above relations are exact, for it makes no sense  ν   u  to speak of exact global symmetries. After all, unlike in ` = , q = . (5) L,R e L,R d the case of gauge invariance, one can move continuously L,R L,R away from a point of symmetry without losing anything which transform under P as - a point of symmetry stops being special, period. All we mean is that the spontaneous breaking of left-right sym- metry dominates over the explicit one, just as one must `L ↔ `R, qL ↔ qR, (6) argue in the case of PQ symmetry that the explicit break- while under C one has ing is small. We should stress that the explicit breaking, even extremely tiny suppressed by the Planck scale [16] ∗ ∗ `L ↔ `R, qL ↔ qR, (7) plays an important role in providing a way out of the do- main wall problem [17] associated with the spontaneous The original GLR symmetry is broken down to the SM breaking of an exact discrete symmetry. True, the do- gauge symmetry at the large scale vR by utilizing left and main wall problem would be in principle also solved [18] 4 by the non-restoration of parity at high temperature [19], lepton flavor violation [15]. Moreover, the original LR but there is no proof that this happens due to subtleties symmetry allows to disentangle the seesaw by predicting of higher order effects in the temperature expansion [20]. MD from Mν and MN , as shown originally in [11]. The bottom line is that the small explicit breaking of par- A small digression. Since the Yukawa couplings are ity is actually welcome, and it is the only guarantee that complex in general, CP is broken explicitly. There is there is no domain wall problem unless inflation takes also induced CP violation through the bi-doublet vev, place below the scale of LR symmetry, not the case in measured by the small parameter sat2β, with sat2β . the MLRSM we are discussing here (we would have to go 2mb/mt [24]. When it vanishes the LR symmetry in the beyond the minimal model, but doing that would open quark mass sector is maintained, and quark mass matri- Pandora’s box of possibilities and the predictions would ces become hermitian in the case of parity. Thus sat2β be lost). will play a crucial role in controlling strong CP viola- To be precise, we are then only asking that this theory tion. In particular, as we will see, in the case of parity ¯ be self-contained and complete when it comes to fermion sat2β = 0 is consistent with θ = 0. We are getting ahead masses - including the neutrino one - which determines of ourselves though, first things first. the precision of the validity of (14). With this cleared, we can now proceed. Under charge conjugation, consistent with (7), one has III. STRONG CP IN MLRSM

∆ ↔ ∆∗ , Φ → ΦT (15) L R We are now ready to discuss the constraints that the so that the Yukawa couplings of the bi-doublet are sym- strong CP violation imply for the MLRSM. Since the sit- metric (both for quarks and leptons) and the left and uation is quite different in the cases of parity and charge right triplet Yukawas are complex conjugates of each conjugation as LR symmetries, we discuss them sepa- other rately. We need to clarify some issues though before we plunge into our discussion. More often than not, the T ∗ Y1,2 = Y1,2,YL = YR. (16) smallness of physical parameters has been treated as a problem in the modern day high energy physics. This In this case, the quark mass matrices are symmetric and by itself however does not make sense. After all, how ∗ thus one has VR = VL before rotating VL into the CKM do we probe the theories of natural phenomena? Before form by getting rid of unphysical phases. In the physical facing experiment, the parameter space of a theory in basis, one would have question is usually too big and needs to be restricted, thus the smallness of some parameters should rather be V = K V ∗ K , (17) R u CKM d taken as a gift. Take for example flavor conservation in neutral currents - it requires the charm quark to lie much where Ku,d are diagonal matrices of phases. The SM Higgs doublet h and the new heavy scalar below the electro-weak scale. The heavy top quark re- doublet H are given by quires tiny mixings with the second and especially first generation and so on and so forth. −ia ia h = cβφ1 + e sβφ2,H = −e sβφ1 + cβφ2, (18) Now, the experiment has started to put constraints on the MLRSM parameter space by requiring that the new where c ≡ cos , s ≡ sin hereafter. The new doublet β β β β scale vR be large. LHC has already set a rigorous limit H mediates tree level flavor violation in the K and B- from through the di-jet decay of WR, MWR & 4TeV [25], meson sectors, which imply mH & 20 TeV [21] and take and the low energy constraints from K and B meson it out of the LHC reach. physics, as we said require mH & 20TeV. For this rea- In the neutrino sector the situation is more involved son, not much can be learned directly about the complex since the RH neutrinos become very heavy. It is conve- Higgs potential couplings and the constraints from the nient to work with their charge conjugate states NL = strong CP violation help a lot as we show below. Con- T ∗ Cν¯R , which from (12) implies MN = vRYR. This leads trary to the prevailing wisdom, the smallness of strong to the seesaw expression [10, 22] for light neutrino masses CP violation is not a problem whatsoever. In any case, 1 we are used to small couplings in the SM - think about M = −M T M (19) the electron mass. The only thing one could worry ν D M D N about is the technical naturalness of small parameters, ` ia ` i.e. whether it amounts to large fine tunings. No such where MD = YD v = (cβY1 + e sβY2 ) v. The physical meaning of this is profound since one can produce N 0s problem emerges in the MLRSM as we will see. at hadron colliders through the Drell-Yan production of WR and actually measure MN . The golden channel is the so-called KS process [23], consisting of both same and A. Strong CP and P opposite sign charged lepton pairs accompanied by jets, which probes directly the Majorana nature of RH neu- Parity plays a special role in the discussion of the trinos while observing directly both lepton number and strong CP violation, for it allows - unlike in the SM or 5 the PQ case - to speak separately about the individual The question is whether this is stable in perturbation terms in θ¯ [5]. It is simply required by consistency of the theory due to the P and CP breaking. It is easy to see theory since the Dirac Yukawa couplings must be hermi- that in the quark sector the answer is positive since small tian, see (14) which then requires θ to vanish separately sat2β guarantees near hermiticity of quark mass matri- in the P symmetry limit. The exact limit is extreme, as ces, and parity breaking comes only from vR that enters we argued before - all that is needed that the explicit de- sub-dominantly at higher loops. Moreover, the KM CP viations are smaller than the spontaneous breaking. Of violation enters also at high orders and is negligible as course, one could argue that θ is zero due to the imposed we argued in the Introduction. parity symmetry, but that would be misleading. It was noticed by Kuchimanchi [14], however, that lep- After all, a small explicit breaking serves to get of rid tonic CP violation affects the above conclusions. The ar- of dangerous domain walls as we discussed above - and gument goes as following. The physical measure of strong strictly speaking saying that θ is small due to P is a tau- CP violation, the sat2β parameter, is directly computable tology, just like saying say that proton is stable due to from the CP violating term in the Higgs potential (see number conservation. Even worse, θ is a vacuum e.g. [27] for a detailed discussion) selection parameter - so why should vacuum respect the symmetry that we like to impose [26]? Another way of Im α2 sat2β = −4 , (21) arguing is that θ is not the usual dimensionless coupling, α3 but rather a kind of initial or boundary condition, since it accompanies a surface term. In short, while setting cou- where the α2,3 couplings correspond to the following plings to zero by a global symmetry is in a sense a tautol- terms of the Higgs potential ogy (proton is stable because baryon number is conserved  ˜ †  † † † say), arguing that parity implies θ = 0 is simply wrong. α2Tr(ΦΦ ) + h.c. Tr(∆R∆R) + α3Tr(Φ Φ∆R∆R). If you wish, θ is another source of the spontaneous break- (22) ing of parity. It should be stressed that α3 cannot be small for MWR . The bottom line is that, even if P is not exact to start 20TeV, since it is responsible for the mass of the heavy with, its explicit breaking being small is perfectly con- scalar doublet H which must be rather heavy, mH & sistent (just as it is consistent to have small explicit PQ 20 TeV [21] due to its direct mediation of flavor violation. symmetry breaking). And when it comes to domain walls It is not surprising then that leptons matter since ∆R 2 2 couples directly to the RH neutrino N. The imaginary it may be as small as MR/MP l, where MR and MP l and the LR symmetry breaking and the Planck scale respec- part of α2 is turned on at the one-loop level, and by using tively, which for the LHC (or the next hadron collider) (20) and (21) it gives [14] reach M ' 10 TeV is far below the experimental limit R 1 m M on θ¯. All we need to make sure is that the effects of the ¯ t † ` ` P l θloop ' 2 Tr(YRYR[Y1 ,Y2 ]) ln (23) spontaneous symmetry breaking of parity do induce an 16π mb vR unacceptably large value of θ¯, or better to say, the electric which, barring fine-tuning with the tree level, implies an dipole moment of neutron. As we show now, everything important bound from (3) is under control, and the consistency of the smallness of strong CP violation actually leads to physically interest- Im Tr(Y †Y [Y `,Y `]) ' Im Tr(Y †Y [Y ,Y ]) 10−11 ing correlations. Thus, unlike in the SM, the smallness R R 1 2 R R D ` . (24) of θ¯ provides useful constraints on the theory. where Y and Y are neutrino Dirac and charged lepton Now, prior to symmetry breaking quarks are massless D ` Yukawa couplings, respectively. and so clearly there is no strong CP violation. What Let us inspect carefully what the above limit really im- happens after the symmetry breaking? The physical pa- plies. Obviously, the leptonic Yukawa couplings could be rameter is θ¯, so besides the value of θ, it is therefore quite real (which cries for symmetry protection) and al- controlled by arg det(M M ), which is turn determined u d leviate the situation [14]. This would have an important by a small parameter s t that measures P and CP a 2β consequence of suppressing the leptonic electric dipole breaking in the quark sector. A careful study was made moments, since the leading contribution is proportional in [13] who find (assuming no fine-tuning between θ and to the imaginary part of the product of W − W mix- arg det(M M )) L R u d ing ξ ' M 2 /M 2 t (c + is ) and Y [11]. Since LR WL WR 2β a a D ¯ mt the imaginary part of of ξLR is negligible, being directly θ ' sat2β , (20) proportional to θ¯, real Y implies tiny leptonic electric mb D dipole moments. implying a practically vanishing sat2β (assuming of However, as we argue now, it turns out that the Dirac course that we do not fine-tune it against θ, but that Yukawa couplings do not have to be real. This is a sur- would be against the very idea of LR symmetry). This prising and fundamental result, since even if the Dirac has important phenomenological implications and forces Yukawas were complex as their quark counterparts - a the new scale to be large, MWR & 20TeV, pushing WR natural scenario - the situation would be actually wel- out of the LHC reach. come. Namely, smaller YD means smaller mN , which 6 helps explaining the smallness of lepton number viola- real Dirac Yukawa matrix, which in turn implies small tion, see e.g. [15]. We can actually be more precise about leptonic EDM or, more naturally, small N masses as we it. have shown. In either case, the smallness of strong CP The crucial point is that in the MLRSM the seesaw can violation makes the theory more predictive. be untangled. In particular for the case of the Hermitian Dirac mass matrix MD, relevant here, MD - and thus in turn YD - can be determined from the knowledge of B. Strong CP and C: connection lost Mν and MN [28]. For illustration, without any loss of generality, assume equal left and right-handed leptonic If the LR symmetry was charge conjugation C, the sit- mixing matrices VL = VR which then gives [28] uation would be less restrictive since at the tree level θ¯ is undetermined, just as in the SM. Namely, in this case the g √ † YD = i VL mν mN V , (25) quark mass matrices are symmetric, the arg det(M M ) 2M L u d W is arbitrary and so is θ itself since it does not break C. where VL is the PMNS mixing matrix. The situation is similar to the SM one, the only physical parameter is θ¯, and, unlike in the P case, one has no right Since YR = gMN /MWR and using (25), and by taking the largest value y` = yτ , one can rewrite (24) as an to separate the strong and weak contributions. In other estimate words, it makes no sense to speak of radiative corrections to θ¯, as it makes no sense to do it either in the SM. M 5 M 10−16M 2 M 4 (26) One could still pretend to study the loop contribu- N ν . W WR tion to arg det(MuMd), as done in [3]. We know that where we ignore the details of leptonic mixings without in the SM these corrections are negligible, so we would any real loss of generality, since they are not small. This have to bring in the physics of RH gauge bosons and new implies the following limit on the mass of N by taking scalars into the game. It is easy to see that the dominant −10 Mν ' 10 GeV contribution would appear at the one-loop level through WL − WR mixing as in Fig. 1. −6/5 4/5 MN . 5 × 10 (MWR /GeV) GeV. (27) One could go on and put constraints on the three con- tributions in the above equation, coming from the KM Thus, the RH neutrinos must be relatively light com- phase, the Ku,d phases and the complex WL−WR mixing. pared to WR, just like most charged of the SM It would be wrong though. Once again, arg det(MuMd) is compared to WL. This fits nicely with their possible im- not a physical parameter and it cannot be computed, pe- pact on neutrinoless double beta decay and the limits riod. Instead, one could use the Fig. 1 as a typical contri- from lepton number violation. The lower vR, the smaller bution to the electric dipole of a quark, but it is subdom- then MN and in turn MD, making it easier to account for inant contribution to the EDM of the neutron. Namely, the smallness of strong CP violation. Of course, this is it is suppressed by a small light quark mass which goes not very strict since the imaginary part of Dirac Yukawas away with the strong interactions. The leading contri- could be somewhat small. Increasing the scale vR poses bution comes the chiral loops, and for recent discussions no real problem, all that is needed smaller YR in order to in the context of the MLRSM see e.g. [13, 31]. Simply comply with the strong CP violation limits. on dimensional grounds one expects for the EDM of the To get a feel for the scales, let us exemplify (27) with neutron MWR ' 10 TeV which gives 1 ∗ dn ' e mN GF Im VLudV ξLR (29) MN . 500 GeV (28) 16π2 Rud in the right ballpark for the observation of the KS pro- where ξ ' M 2 /M 2 t (c + is ) and m ' GeV LR WL WR 2β a a N 2 −21 cess. is the nucleon mass, so that mN GF /16π ' 10 cm. How natural are the above constraints, in a technical Then, experimental limits on the neutron EDM, dn . sense of the word? Since fermion masses are protected 10−26 e · cm, imply by chiral symmetries, there is absolutely no problem and ∗ −5 since most of the SM Yukawas are small, there is also Im VLudVRud ξLR . 10 . (30) nothing unnatural in an intuitive sense. It may happen that the N’s are too light to be seen at In view of the smallness of the mixing ξLR there is basi- colliders. In particular if the lightest N is the warm dark cally no limit on the VR elements. The difference between matter [29]. Still in this case, MN ends up being com- the P and C case is really striking. pletely determined [30] with mN ' KeV − GeV, making In short, independently of the details of the LR sym- Dirac Yukawa couplings much smaller, so that the strong metry implemented, be it P or C, one cannot predict CP violation bound is automatically satisfied. the value of θ¯ in the MLRSM, just as one cannot do it Needless to say, there is always the contribution due in the PQ case or basically any other theory, including to the KM phase, but that is negligible as in the SM. the SM. In the case of P one obtains useful constraints Thus, in a nut-shell, the smallness of θ¯ implies either a needed to keep θ¯ under control, but its value itself can 7

model is all about small Yukawa couplings of quarks and charged leptons. It may appear that the constraints persist even for large LR breaking scale, in contradiction with the situa- tion in the SM. After all, the limit on sat2β is indepen-

dent of MWR . The point though is that sat2β appears only through the new physics of heavy WR and thus ef- fectively disappears in the SM limit - in other words, its smallness stops having physical meaning. Everything is consistent as it should be. FIG. 1. Typical diagram for the one loop contribution to θ¯ in case C. As we argue inFig the text,I this has has no physical Alternatively, as noticed in [14], there is a less natu- meaning whatsoever - one must and can compute only the ral possibility of real neutrino Dirac Yukawa couplings, direct electric dipole moments. which implies an important prediction of small leptonic EDM. In any case, the smallness of strong CP violation plays a gratifying role of reducing the parameter space and implying important phenomenological predictions. be determined only from experiment. What is needed is The constraints in question play particularly impor- to measure different electric dipole moments in order to tant role for the MLRSM accessible at the LHC or next untangle the weak and strong contributions [33]. hadron collider energies, but why would that ever hap- pen? The answer lies in a deep connection with a neutri- noless double beta decay whose observation may signal IV. SUMMARY AND OUTLOOK the contribution of new physics if neutrino mass is not sufficiently large to do the job. Or if for example one We have revisited the issue of strong CP violation in could measure the polarization of electrons and find out the MLRSM with the conclusion that the smallness of that they are right handed, in which case the SM neu- θ¯ helps constrain a number of unknown parameters of trino contribution would be ruled out. In this case WR the theory. There is nothing technically unnatural that could not be too heavy [32], making the case for its man- takes place, which provides yet another example besides ifestation if not at the LHC, surely at the new collider. the SM, where there is no strong CP problem. All that We close with a remark regarding the choice of the LR one learns is that the spontaneous CP violating parame- symmetry in defining the MLRSM. The original choice ter sat2β has to be tiny, implying a very small imaginary was parity, a logical assumption in view of LH fermions part of a particular quartic coupling in the Higgs po- becoming RH under parity. The trouble is that charge tential. This we believe is a fundamental point which conjugation is also a possibility, maybe less intuitive, but illustrates nicely the fact that strong CP violation is a nonetheless a LR symmetry since LH fermion become not a problem per se at all. On the contrary, it may even LH anti-fermions, or complex conjugates of RH fermions. be a benediction for a well defined self-contained theory An apparent advantage of the latter choice is the predic- such as the one in question. tion of symmetric quark and lepton mass matrices, which The situation depends critically on whether the LR predicts same LH and RH quark mixing angles, and the symmetry is parity or charge conjugation. In the latter form of neutrino Dirac mass matrix and untangle the case, the question of θ¯ cannot be addressed perturba- seesaw [11]. tively, just as in the SM - all one can do is to compute Furthermore, C becomes a gauge symmetry in the the various electric dipole moments. There have been SO(10) grand unified theory, which adds to its appeal. studies before and we have stayed away from it here. However, the same SO(10) theory predicts astronomi- In the former case, the situation is different. Having P cally large LR breaking scale and thus this arguments allows to address separately the weak and strong contri- loses its physical relevancy. (By arbitrarily adding many butions to θ¯ and to treat it perturbatively. The bottom new particles, one can of course lower arbitrarily the in- line is that the smallness of sat2β is maintained in pertur- termediate LR scale, but this in turns implies giving un bation theory as long as the RH neutrinos are fairly light on all predictions - this should not be considered a grand mN . 500 GeV for a LR breaking scale around TeV pos- unified theory anymore, or a theory at all, for that mat- sibly reachable at the next hadron collider (small sat2β ter.) takes the MLRSM out of the LHC reach). This at the It turns out that one can untangle the seesaw also same time accounts automatically for the smallness of in the case P [12] and moreover, one can predict both LFV and provides a possibility that the RH physics dom- the RH quark mixing angles and phases with a single inates the neutrinoless double beta decay over the usual unknown parameter [24]. The possibility of controlling contribution of LH neutrino Majorana mass. As vR in- strong CP violation with the emerging correlations dis- creases, this limit relaxes but still, the essential prediction cussed in this work make a strong case for the choice is the Yukawa couplings of N 0s, must be small. This of of parity in defining the left-right symmetric theory, the course is both technically and intuitively natural, the SM way it all started. The crucial thing to remember is that 8 we do not require parity to be exact, as long as its spon- Acknowledgments taneous breaking dominates over the explicit one. The reason lies in the original motivation of the left-right sym- G.S. acknowledges ongoing debates with Gia Dvali on metric theory of having the breakdown of parity originate the question of strong CP violation over the years. He dynamically, as is the case with gauge symmetries. This wishes to thank Xiao-Gang He and other members of the is in complete analogy with the axion picture where one TD Lee institute in Shanghai for their warm hospitality demands that the explicit breaking of PQ symmetry be in the course of this work. We have benefited from fruitful smaller that the spontaneous one. discussions with Kaladi Babu, Alessio Maiezza, Fabrizio Nesti and Juan Carlos Vasquez. We are grateful to Babu and Dvali for helping us improve the manuscript.

[1] V. Baluni, “CP Violating Effects in QCD,” Phys. Rev. (1978). doi:10.1103/PhysRevLett.40.279 D 19 (1979), 2227-2230 doi:10.1103/PhysRevD.19.2227 [9] J.C. Pati, A. Salam, “Lepton Number As The Fourth R. J. Crewther, P. Di Vecchia, G. Veneziano and E. Wit- Color,” Phys. Rev. D 10 (1974) 275; ten, “Chiral Estimate of the R. N. Mohapatra, J.C. Pati, “A ’Natural’ Left-Right of the Neutron in ,” Phys. Symmetry, Phys. Rev. D 11 (1975) 2558; Lett. 88B (1979) 123 Erratum: [Phys. Lett. 91B (1980) G. Senjanovi´c,R. N. Mohapatra, “Exact Left-Right Sym- 487]. doi:10.1016/0370-2693(80)91025-4, 10.1016/0370- metry And Spontaneous Violation Of Parity,” Phys. Rev. 2693(79)90128-X D 12 (1975) 1502; [2] B. Graner, Y. Chen, E. Lindahl and B. Heckel, “Re- G. Senjanovi´c, “Spontaneous Breakdown Of Parity In A duced Limit on the Permanent Electric Dipole Moment Class Of Gauge Theories,” Nucl. Phys. B 153 (1979) 334. of Hg199,” Phys. Rev. Lett. 116 (2016) no.16, 161601 [10] P. Minkowski, “Mu → E Gamma At A Rate Of One Out doi:10.1103/PhysRevLett.116.161601 [arXiv:1601.04339 Of 1-Billion Muon Decays?,” Phys. Lett. B 67 (1977) [physics.atom-ph]]. 421; [3] J. R. Ellis and M. K. Gaillard, “Strong and Weak CP Vio- R. N. Mohapatra and G. Senjanovi´c,“Neutrino Mass and lation,” Nucl. Phys. B 150, 141 (1979). doi:10.1016/0550- Spontaneous Parity Violation,” Phys. Rev. Lett. 44, 912 3213(79)90297-9 (1980). doi:10.1103/PhysRevLett.44.912 [4] T. Chupp, P. Fierlinger, M. Ramsey-Musolf and J. Singh, R. N. Mohapatra, G. Senjanovi´c,“Neutrino Masses And “Electric dipole moments of atoms, molecules, nuclei, Mixings In Gauge Models With Spontaneous Parity Vi- and particles,” Rev. Mod. Phys. 91 (2019) no.1, 015001 olation,” Phys. Rev. D23 (1981) 165. doi:10.1103/RevModPhys.91.015001 [arXiv:1710.02504 [11] M. Nemevˇsek, G. Senjanovi´c and V. Tello, “Con- [physics.atom-ph]]. necting Dirac and Majorana Neutrino Mass N. Yamanaka, B. Sahoo, N. Yoshinaga, T. Sato, K. Asahi Matrices in the Minimal Left-Right Symmetric and B. Das, “Probing exotic phenomena at the interface Model,” Phys. Rev. Lett. 110 (2013) no.15, 151802 of nuclear and with the electric dipole doi:10.1103/PhysRevLett.110.151802 [arXiv:1211.2837 moments of diamagnetic atoms: A unique window to [hep-ph]]. hadronic and semi-leptonic CP violation,” Eur. Phys. [12] For a recent discussion and references see G. Senjanovi´c J. A 53 (2017) no.3, 54 doi:10.1140/epja/i2017-12237-2 and V. Tello, “Parity and the origin of neutrino mass,” [arXiv:1703.01570 [hep-ph]]. arXiv:1912.13060 [hep-ph]. N. Yamanaka and E. Hiyama, “Standard model G. Senjanovi´c and V. Tello, “Disentangling the see- contribution to the electric dipole moment of the saw mechanism in the minimal left-right symmetric deuteron,3H, and3He nuclei,” JHEP 02 (2016), 067 model,” Phys. Rev. D 100, no. 11, 115031 (2019) doi:10.1007/JHEP02(2016)067 [arXiv:1512.03013 [hep- doi:10.1103/PhysRevD.100.115031 [arXiv:1812.03790 ph]]. [hep-ph]]. [5] M. A. B. Beg and H.-S. Tsao, “Strong P, T Noninvari- [13] A. Maiezza and M. Nemevˇsek, “Strong P invariance, ances in a Superweak Theory,” Phys. Rev. Lett. 41, 278 neutron electric dipole moment, and minimal left-right (1978). doi:10.1103/PhysRevLett.41.278. parity at LHC,” Phys. Rev. D 90, no. 9, 095002 (2014) R. N. Mohapatra and G. Senjanovi´c,“Natural Suppres- doi:10.1103/PhysRevD.90.095002 [arXiv:1407.3678 [hep- sion of Strong p and t Noninvariance,” Phys. Lett. 79B, ph]]. 283 (1978). doi:10.1016/0370-2693(78)90243-5 [14] R. Kuchimanchi, “Leptonic CP problem in left-right sym- [6] H. Georgi, “A Model of Soft CP Violation,” Hadronic J. metric model,” Phys. Rev. D 91, no. 7, 071901 (2015) 1, 155 (1978). doi:10.1103/PhysRevD.91.071901 [arXiv:1408.6382 [hep- [7] R. D. Peccei and H. R. Quinn, “Constraints Imposed by ph]] CP Conservation in the Presence of Instantons,” Phys. [15] V. Tello, “Connections between the high and low energy Rev. D 16, 1791 (1977). doi:10.1103/PhysRevD.16.1791 violation of Lepton and Flavor numbers in the minimal [8] S. Weinberg, “A New Light ?,” Phys. Rev. Lett. left-right symmetric model,” 40, 223 (1978). doi:10.1103/PhysRevLett.40.223 [16] B. Rai and G. Senjanovi´c, “Gravity and domain F. Wilczek, “Problem of Strong P and T Invariance in wall problem,” Phys. Rev. D 49, 2729 (1994) the Presence of Instantons,” Phys. Rev. Lett. 40, 279 doi:10.1103/PhysRevD.49.2729 [hep-ph/9301240]. 9

[17] Y. B. Zeldovich, I. Y. Kobzarev and L. B. Okun, “Cos- doi:10.1103/PhysRevD.94.095023 [arXiv:1502.05704 mological Consequences of the Spontaneous Breakdown [hep-ph]]. of Discrete Symmetry,” Zh. Eksp. Teor. Fiz. 67 (1974) 3 [25] G. Aad et al. [ATLAS Collaboration], “Search for new [Sov. Phys. JETP 40 (1974) 1]. resonances in mass distributions√ of jet pairs using 139 [18] G. R. Dvali and G. Senjanovi´c, “Is there a domain fb−1 of pp collisions at s = 13 TeV with the ATLAS wall problem?,” Phys. Rev. Lett. 74, 5178 (1995) detector,” arXiv:1910.08447 [hep-ex]. doi:10.1103/PhysRevLett.74.5178 [hep-ph/9501387]. [26] This point of view has been advocated by G. Dvali to G.S, G. R. Dvali, A. Melfo and G. Senjanovi´c, “Non- see e.g. G. Dvali, “Three-form gauging of axion symme- restoration of spontaneously broken P and CP at tries and gravity,” hep-th/0507215. high temperature,” Phys. Rev. D 54, 7857 (1996) [27] A. Maiezza, G. Senjanovi´c and J. C. Vasquez, doi:10.1103/PhysRevD.54.7857 [hep-ph/9601376]. “Higgs sector of the minimal left-right symmetric The use of symmetry non-restoration to get rid of do- theory,” Phys. Rev. D 95, no. 9, 095004 (2017) main walls resulting from the spontaneous breakdown of doi:10.1103/PhysRevD.95.095004 [arXiv:1612.09146 CP symmetry was discussed already in V. A. Kuzmin, [hep-ph]]. M. E. Shaposhnikov and I. I. Tkachev, “Matter - An- [28] G. Senjanovi´cand V. Tello, “Probing Seesaw with Parity timatter Domains in the Universe: A Solution of the Restoration,” Phys. Rev. Lett. 119 (2017) no.20, 201803 Vacuum Walls Problem,” Phys. Lett. 105B, 167 (1981). doi:10.1103/PhysRevLett.119.201803 [arXiv:1612.05503 doi:10.1016/0370-2693(81)91013-3 [hep-ph]]. V. A. Kuzmin, I. I. Tkachev and M. E. Shaposhnikov, G. Senjanovi´c and V. Tello, “Disentangling See- “Are There Domains of Antimatter in the Universe? (In saw in the Minimal Left-Right Symmetric Model,” Russian),” Pisma Zh. Eksp. Teor. Fiz. 33, 557 (1981). arXiv:1812.03790 [hep-ph]. [19] R. N. Mohapatra and G. Senjanovi´c,“Broken Symme- [29] S. Dodelson and L. M. Widrow, “Sterile-neutrinos tries at High Temperature,” Phys. Rev. D 20, 3390 as dark matter,” Phys. Rev. Lett. 72, 17 (1994) (1979). doi:10.1103/PhysRevD.20.3390 doi:10.1103/PhysRevLett.72.17 [hep-ph/9303287]. R. N. Mohapatra and G. Senjanovi´c,“High Temperature [30] M. Nemevˇsek, G. Senjanovi´c and Y. Zhang, “Warm Behavior of Gauge Theories,” Phys. Lett. 89B (1979) 57. Dark Matter in Low Scale Left-Right Theory,” JCAP doi:10.1016/0370-2693(79)90075-3 1207 (2012) 006 doi:10.1088/1475-7516/2012/07/006 The original observation of symmetry breaking at high [arXiv:1205.0844 [hep-ph]]. temperature can be found in S. Weinberg, “Gauge and F. Bezrukov, H. Hettmansperger and M. Lindner, “keV Global Symmetries at High Temperature,” Phys. Rev. D sterile neutrino Dark Matter in gauge extensions of 9, 3357 (1974). doi:10.1103/PhysRevD.9.3357 the Standard Model,” Phys. Rev. D 81, 085032 (2010) [20] G. Bimonte and G. Lozano, “On Symmetry nonrestora- doi:10.1103/PhysRevD.81.085032 [arXiv:0912.4415 [hep- tion at high temperature,” Phys. Lett. B 366, 248 (1996) ph]]. doi:10.1016/0370-2693(95)01395-4 [hep-th/9507079]. [31] S. Bertolini, A. Maiezza and F. Nesti, “Kaon CP viola- [21] tion and neutron EDM in the minimal left-right symmet- [21] See e.g. S. Bertolini, A. Maiezza and F. Nesti, “Present ric model,” arXiv:1911.09472 [hep-ph]. and Future K and B Meson Mixing Constraints on [32] M. Nemevˇsek, F. Nesti, G. Senjanovi´c and V. Tello, TeV Scale Left-Right Symmetry,” Phys. Rev. D 89, “Neutrinoless Double Beta Decay: Low Left-Right Sym- no. 9, 095028 (2014) doi:10.1103/PhysRevD.89.095028 metry Scale?,” arXiv:1112.3061 [hep-ph]. [arXiv:1403.7112 [hep-ph]]. V. Tello, M. Nemevˇsek, F. Nesti, G. Senjanovi´c and [22] S. Glashow, Quarks and leptons, Carg`ese1979, ed. M. F. Vissani, “Left-Right Symmetry: from LHC to L´evy(Plenum, NY, 1980); Neutrinoless Double Beta Decay,” Phys. Rev. Lett. M. Gell-Mann et al., Supergravity Stony Brook workshop, 106, 151801 (2011) doi:10.1103/PhysRevLett.106.151801 New York, 1979, ed. P. Van Niewenhuizen, D. Freeman [arXiv:1011.3522 [hep-ph]]. (North Holland, Amsterdam, 1980). [33] We have been informed that this has been done by [23] W.-Y. Keung, G. Senjanovi´c,“Majorana Neutrinos And M. Ramsey-Musolf and J. C. Vasquez, to appear. We The Production Of The Right-handed Charged Gauge thank them for sharing their results with us. Boson,” Phys. Rev. Lett. 50 (1983) 1427. [24] G. Senjanovi´c and V. Tello, “Right Handed Quark Mixing in Left-Right Symmetric The- ory,” Phys. Rev. Lett. 114, no. 7, 071801 (2015) doi:10.1103/PhysRevLett.114.071801 [arXiv:1408.3835 [hep-ph]]. G. Senjanovi´c and V. Tello, “Restoration of Par- ity and the Right-Handed Analog of the CKM Matrix,” Phys. Rev. D 94 (2016) no.9, 095023