Light Asymmetric Dark Matter on the Lattice: SU(2) Technicolor with Two Fundamental Flavors ♠ ∗ (cid:114) † (cid:114)‡ Randy Lewis , Claudio Pica , and Francesco Sannino ♠ Department of Physics and Astronomy, York University, Toronto, M3J 1P3, Canada and (cid:114) CP3-Origins & the Danish Institute for Advanced Study DIAS, University of Southern Denmark, Campusvej 55, DK-5230 Odense M, Denmark 2 1 0 2 n Abstract a J TheSU(2)gaugetheorywithtwomasslessDiracflavorsconstitutesthebuildingblockofseveral 3 1 models of Technicolor. Furthermore it has also been used as a template for the construction of a ] h naturallightasymmetric,ormixedtype,darkmattercandidate. Weuseexplicitlatticesimulations p - to confirm the pattern of chiral symmetry breaking by determining the Goldstone spectrum and p e h thereforeshowthatthedarkmattercandidatecan,defacto,beconstitutedbyacomplexGoldstone [ boson. We also determine the phenomenologically relevant spin-one and spin-zero isovector 2 v spectrumanddemonstratethatitiswellseparatedfromtheGoldstonespectrum. 3 1 5 3 9. Preprint: CP3-Origins-2011-28&DIAS-2011-15 0 1 1 : v i X r a ∗ [email protected] † [email protected] ‡ [email protected] 1 I. INTRODUCTION Thenatureofdarkmatter(DM)isanimportantprobleminmodernphysics. DMplays a key role in the formation of large structures and the evolution of the Universe. It is also widely expected to provide a link to physics beyond the Standard Model (SM). For ff these reasons much experimental, observational, and theoretical e ort has been devoted to shedding light on DM. It is popular to identify DM with Weakly Interacting Massive Particles(WIMPs). ManypropertiesofaWIMParenotconstrainedbyourcurrentknowledgeofDM,for example the WIMP may or may not be a stable particle [1] and it may or may not be identifiedwithitsantiparticle[2]. AsymmetricDMreferstoascenariowheretheWIMP’s antiparticle has been annihilated away, leaving the WIMP itself as the observed DM. AsymmetricDMcandidateswereputforwardin[2]astechnibaryons,in[3]asGoldstone bosons, and subsequently in many diverse forms [4–10]. There is also the possibility of mixed DM [11], i.e. having both a thermally-produced symmetric component and an asymmetricone. / Null results from several experiments, such as CDMS [12] and Xenon10 100 [13, 14], have placed stringent constraints on WIMP-nucleon cross sections. Interestingly DAMA [15] and CoGeNT [16] have both produced evidence for an annual modulation signature for DM, as expected due to the relative motion of the Earth with respect to the ff DMhalo. TheseresultssupportalightWIMPwithmassoforderafewGeV,whicho ers the attractive possibility of a common mechanism for baryogenesis and DM production. At first glance it seems that the WIMP-nucleon cross sections required by DAMA and CoGeNT have been excluded by CDMS and Xenon upon assuming spin-independent interactions between WIMPs and nuclei (with protons and neutrons coupling similarly to WIMPs), but a number of resolutions for this puzzle have been proposed in the litera- ture [17–22]. Interestingly, new results from the CRESST-II experiment report signals of lightDM[23]. A composite origin of DM is an intriguing possibility given that the bright side of the universe, constituted mostly by nucleons, is also composite. Thus a new strongly- coupled theory could be at the heart of DM. In this work we investigate, for the first time on the lattice, a technicolor-type extension of the SM expected to naturally yield a 2 light DM candidate, as introduced in [8] and used in [22] to reconcile the experimental observations. II. ALIGHTDARKMATTERTEMPLATE:SU(2)TECHNICOLORWITHTWODIRACFLA- VORS AsymmetricDMrequiresatleastonecomplexfieldwitha(nearly)conservedAbelian quantum number. It would be exciting if the theory predicting the existence of this state is also directly involvedwith the breaking of electroweak symmetry. Therefore a natural candidate is a technicolor model. Most of the states of the theory are much heavier than afewGeV,butifthereareGoldstonebosonsnotabsorbedbythelongitudinaldegreesof freedomoftheSMmassivegaugebosons,theybecomeprimarycandidatesforproducing anaturalhierarchybetweentheGeVandtheTeVscale. The minimal technicolor theory that breaks the electroweak symmetry and features a light DM state was constructed first in [24, 25]. This theory is SU(2) technicolor with two Dirac flavors, which has global symmetry SU(4) expected to break to Sp(4). Five Goldstone bosons are generated of which three are eaten by the SM gauge bosons and a complex one (a techni-diquark which is essentially a technibaryon of this SU(2) theory) remainsinthespectrum. ThisisourcandidateforthelightasymmetricDMparticle. The walking version of this model, known as Ultra Minimal Walking Technicolor (UMT), has been constructed in [8] and it contains, besides the fermions in the funda- mental representation, also a Dirac fermion in the adjoint. For the present exploratory work, we begin with Section III of [8] but without the adjoint fermion. Moreover, when implementing our lattice simulations we make two additional modifications: all elec- troweak interactions are omitted from the lattice simulations because that physics is well-understood by perturbative methods, and explicit technifermion masses are added becauselatticesimulationswithexactlymasslessfermionsarenotpractical. Theresulting Lagrangianisverysimple 1 L = − FaµνFaµν +U(iγµDµ −m)U+D(iγµDµ −m)D, (1) 4 where U and D are the two techniquark fields having a common bare mass m, Fa is the µν field strength, and Dµ is the covariant derivative. The Dirac and technicolor indices of U 3 andDarenotshownexplicitly. = Latticesimulationswillbeusedtoextrapolatetom 0,andinthatlimittheLagrangian hasaglobalSU(4)symmetrycorrespondingtothefourchiralfermionfields 1 1 1 1 U = (1−γ5)U , U = (1+γ5)U , D = (1−γ5)D, D = (1+γ5)D. (2) L R L R 2 2 2 2 (cid:44) For m 0, the SU(4) symmetry is explicitly broken to a remaining Sp(4) subgroup as follows. TheLagrangianfrom(1)canberewrittenas 1 m m (cid:16) (cid:17)† L = − FaµνFaµν +iUγµDµU+iDγµDµD+ QT(−iσ2)CEQ+ QT(−iσ2)CEQ , (3) 4 2 2 where U 0 0 1 0 L Q = −iσD2CLUT , E = −01 00 00 10 , (4) R −iσ2CDT 0 −1 0 0 R CisthechargeconjugationoperatoractingonDiracindices,andthePaulistructure−iσ2 isthestandardantisymmetrictensoractingoncolorindices. UnderaninfinitesimalSU(4) transformationdefinedby (cid:88)15 Q → 1+i αnTnQ, (5) n=1 theLagrangian(3)becomes im (cid:88)15 (cid:16) (cid:17) L → L+ αnQT(−iσ2)C ETn +TnTE Q+h.c., (6) 2 n=1 where Tn denotes the 15 generators of SU(4) and αn is a set of 15 constants. The only generatorsthatleavetheLagrangianinvariantarethosethatobey ETn +TnTE = 0 (7) which is precisely the definition of an Sp(4) Lie algebra. From this, it is straightforward toderivethetenSp(4)generatorsinaspecificrepresentation;seetheappendixof[8]. = For m 0 the Lagrangian retains the full SU(4) symmetry but, by analogy with the SU(3) theory of QCD, one might expect dynamical symmetry breaking associated with theappearanceofanonzerovacuumexpectationvalue, (cid:104) + (cid:105) (cid:44) . UU DD 0 (8) 4 Since this vacuum expectation value has the same structure as the terms containing m → in the Lagrangian, the dynamical breaking would also be SU(4) Sp(4). According to Noether’stheorem,thefivebrokengeneratorswouldbeaccompaniedbyfiveGoldstone bosons. Of course this suggestion of dynamical symmetry breaking must be checked nonper- turbatively using first-principles lattice simulations. As reported below, we have done so and our lattice simulations provide direct verification of this dynamical symmetry breaking. There have been several previous lattice studies of SU(2) gauge theory with fermions inthefundamentalrepresentation[26],mainlymotivatedbyinterestatnonzerochemical potential, but all of these studies relied on the staggered action where the number of = fermions is a multiple of 4. Our minimal technicolor theory requires N 2 and thus f weusetheWilsonactioninsteadofstaggeredfermions. Whenstudyingchiralsymmetry breaking scenarios with Wilson fermions attention must be paid to the presence, on a lattice,oftheunphysicalAokiphase. Forfixedgaugecoupling,theAokiphaseisentered as the quark mass is reduced. An analytic discussion of the Aoki phase symmetries for thistheoryisprovidedin[27],andthreelatticestudiesarealsoavailable[28–30]. Forour presentsimulations,weavoidtheAokiphaseandworkexclusivelyinthephysicalphase. There have been very few previous lattice results reported for SU(2) gauge theory with ff 2 fundamental fermions [30–32] and in each case the primary focus was on a di erent > action(eitherN 2oradjointrepresentationfermions). Ourworkisthefirstlatticestudy f focused on the mass spectrum of the two-color two-flavor theory, which is the familiar technicolortemplate. III. LATTICEHADRONOPERATORSANDEFFECTIVEFIELDTHEORY The creation operator for a meson is the Hermitian conjugate of its annihilation oper- ator,andasetoflocalannihilationoperatorsformesonsis O(Γ) ≡ U(x)ΓD(x), UD O(Γ) ≡ D(x)ΓU(x), DU (cid:18) (cid:19) O(Γ) ≡ √1 U(x)ΓU(x)±D(x)ΓD(x) , (9) UU±DD 2 5 where Γ is a chosen Dirac structure. In this work we consider Γ = 1, γ5, γµ, or γµγ5. In lattice simulations, meson masses are extracted from the time dependence of correlation functions,forexample (cid:88)(cid:88) (cid:16) (cid:17)† C(Γ) (t −t ) = O(Γ) (y) O(Γ) (x) x y UD UD UD (cid:126)x (cid:126)y (cid:88)(cid:88) (cid:104) (cid:105) = Tr ΓD(y)D(x)γ0Γ†γ0U(x)U(y) , (10) (cid:126)x (cid:126)y ··· where Tr[ ] denotes a trace over Dirac and color indices and we have dropped the vacuumexpectationvaluesforthepropagatorsasexplainedinAppendixA. Perhaps surprisingly, local diquark (i.e. baryon) correlation functions provide no new data in this theory. To understand why, notice that the available local diquark operators are O(Γ) ≡ UT(x)(−iσ2)CΓD(x), UD O(Γ) ≡ DT(x)(−iσ2)CΓU(x). (11) DU (Some operators containing UT···U or DT···D are identically zero.) The diquark corre- lationfunctionistherefore (cid:88)(cid:88) (cid:16) (cid:17)† C(Γ) (t −t ) = O(Γ) (y) O(Γ) (x) UD x y UD UD (cid:126)x (cid:126)y (cid:88)(cid:88) (cid:20) (cid:21) = Tr ΓD(y)D(x)γ0Γ†C†(−iσ2)†γ0TUT(x)UT(y)(−iσ2)C (12) (cid:126)x (cid:126)y andthiscanberewrittenbyusingtwopropertiesofthechargeconjugationoperator: one foraDiracmatrix, γµT = −CγµC† , (13) andtheotherfortheWilsonfermionmatrix, (cid:18) (cid:19)T C−1(−iσ2)−1 U(y)U(x) C(−iσ2) = U(x)U(y), (14) toarriveat (cid:88)(cid:88) (cid:104) (cid:105) C(Γ) (t −t ) = Tr ΓD(y)D(x)γ0Γ†γ0U(x)U(y) UD x y (cid:126)x (cid:126)y = C(Γ) (t −t ) (15) x y UD 6 Γ foranychoiceof . Thisconclusionmeansthatalatticesimulationwillfindnumerically- identicalcorrelationfunctions,andthusidenticalmasses,forthemesonanddiquark. We haveverifiedthisexplicitlyinthelatticesimulationsdescribedbelow. Noticethatthedegeneratepairshaveequalangularmomentumbutoppositeparities, forexample (cid:16) (cid:17) (cid:16) (cid:17) J O(Γ) = J O(Γ) , (16) UD UD (cid:16) (cid:17) (cid:16) (cid:17) P O(Γ) = −P O(Γ) . (17) UD UD Thisrelationshipfor JP hasalsobeenmentioned,forexample,in[33]. Wemustconclude thatthethreepseudoscalarmesonGoldstonebosonsareaccompaniedbytwoscalardiquark Goldstone bosons. This is in contrast to the identification made in [8], where it was ff assumed that all five Goldstones would be pseudoscalars. Despite this, the e ective Lagrangian in [8] remains unaltered because all scalar and pseudoscalar particles were retainedinthatworkwiththecorrectassignmentwithrespecttothebrokenandunbroken generatorsofthechiralsymmetrygroup1. ff Now that the five Goldstones have been identified, we can define an e ective theory wherein those five are the only fields. Indeed, our lattice simulations (discussed below) / indicate that the non-Goldstone scalar pseudoscalar hadrons are even heavier than the ff vectormesons,makingitquitenaturaltointegrateallnon-Goldstonesoutofthee ective theory. Theannihilationoperatorsareconvenientlycollectedintoacompactform, (cid:88)(cid:18) (cid:19) δL = QT(−iσ2)CΓTnQ Φ(Γ)n, (18) n whereΦ(Γ)nisagenericnameforthen’thparticlewithDiracstructureΓ. ThisδLrepresents theeffectiveLagrangiancouplingsforexternalfieldsΦ(Γ)n. Forthespecialcaseofthefive Goldstone bosons, we sum n only over the five broken generators and we must choose Γ = γ5. Theresultis δLG = QT(−iσ2)Cγ5GQ (19) G Π+ wherethematrix containsthefiveGoldstonebosons(i.e. thepseudoscalarmesons , 1InpracticetheonlychangeistheredefinitionofΠ(cid:101) andΠ(cid:101) attheunderlyingoperatorlevelwithΠ UD UD UD andΠ in[8]. UD 7 Π−,Π0 andthescalardiquarksΠ ,Π ): UD UD √ √ 0 2Π Π0 2Π+ √ UD √ G = i − 2ΠUD √0 2Π− √−Π0 . (20) 2 −Π0 − 2Π− 0 − 2Π √ √ UD − 2Π+ Π0 2Π 0 UD G The matrix is closed under Sp(4) transformations: the Goldstone bosons form a five- G dimensional representation of Sp(4). The matrix is also identical to M as defined by 4 equation(22)of[8]afterthenon-GoldstonefieldsareremovedfromM . 4 ff To summarize, an e ective field theory for the five Goldstone bosons is obtained by using the matrix G in place of M in [8]. The three pseudoscalar Goldstones (Π±, Π0) 4 Π areresponsibleforelectroweaksymmetrybreaking,andthetwoscalarGoldstones( , UD Π )aretheDMcandidateanditsantiparticle. UD IV. METHODSFORNUMERICALSIMULATIONS ThestandardWilsonaction, β (cid:88)(cid:18) 1 (cid:19) (cid:88) SW = 1− ReTrUµ(x)Uν(x+µˆ)Uµ†(x+νˆ)Uν†(x) +(4+m0) ψ¯(x)ψ(x) 2 2 x,µ,ν x 1 (cid:88)(cid:18) (cid:19) − ψ¯(x)(1−γµ)Uµ(x)ψ(x+µˆ)+ψ¯(x+µˆ)(1+γµ)Uµ†(x)ψ(x) , (21) 2 x,µ is used for this study of SU(2) gauge theory with two mass-degenerate fermions in the fundamentalrepresentation. ConfigurationsweregeneratedusingtheHMCalgorithmas implementedintheHiRepcode[34]. Atotalof12ensembleswerecreated,corresponding ff ff β tosixdi erentbarequarkmassesm foreachoftwodi erentgaugecouplings ;seeTable 0 β = . I for details. We note that [30] also contains some simulations of this theory at 20 and our findings for the pseudoscalar meson mass and PCAC quark mass are consistent with that paper. For other parameter choices, see [31] and [32]. In the present study, all lattices are L3 × T = 163 × 32 with periodic boundary conditions in each direction. Every ensemble contains 35 configurations separated by 20 unused configurations after aninitialthermalizationof320configurations. Figures1and2displaythefluctuationsin theaverageplaquettewithinselectedMarkovchains. 8 TABLEI:Numericalvaluesofthegaugecouplingandthebarequarkmassusedto generatethe12ensemblesofthisproject. β m 0 2.0 -0.85,-0.90,-0.94,-0.945,-0.947,-0.949 2.2 -0.60,-0.65,-0.68,-0.70,-0.72,-0.75 0.575 m = -0.949 0.57 0 0.565 e m = -0.94 ett 0 u 0.56 q a pl e g a 0.555 r e av m = -0.9 0 0.55 0.545 m = -0.85 0 0.54 0 200 400 600 800 1000 configuration number from an ordered start β = . FIG.1: Evolutionoftheaverageplaquetteforselectedbaremassesat 20. Our ensembleof35configurationsiscomprisedofthosenumbered320,340,360,...1000. Quark propagators are created from wall sources built of random U(1) phases at ev- ery lattice site on one chosen time step, t. Inversions are performed with the standard i BiCGstabalgorithm. Toreducestatisticalfluctuations,correlationfunctionsareaveraged ≤ ≤ over all lattice times 0 t 31. Correlation functions then depend on the time t cor- i f responding to the annihilation operator. This annihilation operator is local and summed 9 0.62 m = -0.75 0 0.615 m = -0.7 0 e ett 0.61 u q a pl m = -0.65 e 0 g a er 0.605 v a 0.6 m = -0.6 0 0.595 0 200 400 600 800 1000 configuration number from an ordered start β = . FIG.2: Evolutionoftheaverageplaquetteforselectedbaremassesat 22. Our ensembleof35configurationsiscomprisedofthosenumbered320,340,360,...1000. overallspatialsitestoproduceazero-momentumhadron. Ourmulti-statefits, (cid:88)n (cid:18) (cid:18) T(cid:19)(cid:19) = − , C(t) a cosh m t (22) j j 2 j=1 = − > useeverynonzerotimeseparationt t t 0,whichavoidsthesubjectivityofchoosing f i a fitting window. Pseudoscalar, vector and temporal-axial (A ) correlators are fitted to 4 = , , = three states, n 3 in (22); scalar and spatial-axial (A A A ) correlators use n 2. 1 2 3 Statistical uncertainties are produced from each 35-configuration ensemble by creating 150bootstrapensembles(having35configurationseach). ffi Isovector hadrons are su cient for most topics addressed in this work, but isoscalars are used to study a specific issue. Isoscalar operators require the computation of single- site propagators, i.e. quark propagators that begin and end on a single lattice site. In 10