Arxiv:1503.07521V4 [Gr-Qc]

Home , Claudia de Rham

arXiv:1503.07521v4 [gr-qc] 4 Jul 2015 ΛD)mdlpoie eygo tt observational to fit good very a provides model (ΛCDM) Introduction r ecie olna neatoso h gravitational the of the- This interactions 5]. nonlinear [4, describes instability has ghost ory bigravity) this avoid [3]. (or to theory instability shown been bimetric ghost particular suf- (BD) a to Boulware-Deser Recently general, the in found, from were fer theories fields nonlinear spin-2 2], [1, interacting history theo- of long “bimetric” a have and gravity of “massive” ries While fields spin-2 interactions. additional include and distances. to is of theories range field wide classical a over GR mimic must ob- to gravity be able of to theory be order modified In any viable, problems. servationally theoretical its despite matter data, dark Λ-cold GR-based To the problems. puzzle, matter the dark compound and cosmologi- energy, the dark with constant, is cal associated GR is particular, In and GR. nonrenormalizable is and physics Model There seek Standard the to field. beyond inter- motivation spin-2 observational gravitational massless and the a theoretical to introducing General by this forces. actions mat- electroweak extends describing and (GR) strong 1, the relativity and as 1/2, well 0, as ter spins with fields contains ¶ ∗ § ‡ † ahrAkrami, Yashar [email protected];Crepnigauthor Corresponding [email protected]; [email protected] [email protected] [email protected] [email protected] aua osblt o xedn h e fknown of set the extending for possibility natural A — Terprso ydahhv engreatly been have death my exaggerated.” of reports “The erc Twain Metrics efrhrso hti hslmttehlct- oei olo no is mode helicity-0 the limit this in that show further We iha ffciecsooia osatwihi,remarkabl is, which constant cosmological effective an with iu okhssgetdta t omlgclsltosar solutions cosmological its that suggested has work vious cl.Ti rvdsalt-ieepninhsoywihis constant. which cosmological history the expansion for late-time value a th technically-natural provides limit, This this In scale. times. early unobservably to back pushed be hwta ytkn h lnkms o h eodmetric, second the for mass Planck the taking by that show scales. perturbations cosmological for order in scale electroweak —TeSadr oe fpril physics particle of Model Standard The .— iercter ecie rvttoa neatosi t in interactions gravitational describes theory Bimetric ,2, 1, ∗ 4 ntttfu hoeicePyi,Eidgenössische Techn Physik, für Theoretische Institut ..Hassan, S.F. 3 eateto hsc n h sa li ete Stockholm Centre, Klein Oskar the and Physics of Department 1 2 5 odt,KHRylIsiueo ehooyadSokomUn Stockholm and Technology of Institute Royal KTH Nordita, ntttfu hoeicePyi,Ruprecht-Karls-Unive Physik, für Theoretische Institut AT,Cnr o ahmtclSine,Uiest fCam of University Sciences, Mathematical for Centre DAMTP, laoaUiest etr E169 tchl,Sweden Stockholm, 91 106 SE Center, University AlbaNova iercgaiyi omlgclyviable cosmologically is gravity Bimetric ofagPuiSrse2,89 ¨rc,Switzerland Zürich, 8093 27, Wolfgang-Pauli-Strasse olgtlsakn2,S-09 tchl,Sweden Stockholm, SE-10691 23, Roslagstullsbacken ,3, 1, hlspewg1,610Hiebr,Germany Heidelberg, 69120 16, Philosophenweg ibroc d,CmrdeC30A UK 0WA, CB3 Cambridge Rd., Wilberforce † rn Könnig,Frank ,2, 1, ‡ nnsSchmidt-May, Angnis 1.Teeoho ceeaini e ytems scale mass the by set is acceleration of epoch The 41]. es 1,1] ned irvt a ensont pos- to shown been has bigravity Indeed, 13]. [12, verse bebcgon ouin i ntefiiebac [16–19]. vi- branch Most finite the 41]. on 34, lie [33, solutions sectors background tensor ghosts able and contain 32], scalar to the [19, found both been backgrounds in have sensible perturbations the have but can models branch aldtefiieadifiiebranches. infinite and finite the called view. of This from point attractive setup). theoretical especially gravity a fields massive spin-2 for the interacting [42] in makes Ref. analysis (see explicit Model Standard an the in symmetry the chiral in recovered is limit that symmetry diffeomorphism energy, extra vacuum small a Unlike m interactions. spin-2 the of o hc h bec fteB hs ttenonlinear the 9–11]. at [5, ghost Refs. BD in the established of was level absence [6–8] the mas- background) which flat (a nondynamical for gravity a extension massive on an of field spin-2 is theory sive It ghost-free earlier field. an spin-2 of additional an with metric a akrud aeas ensuidetniey[28– extensively studied been also have backgrounds cal aso itr,ee nteasneo aumenergy vacuum of absence ex- cosmic the the in [14–20]. of even observations history, match pansion can which lutions Uni- the can- Friedmann-Lemaˆıtre-Robertson-Walker so- of a (FLRW) sess expansion therefore accelerated is the theory explain Bimetric to didate distances. large at 2 1 hr satidbac otiigbucn ouin,btt but [41]. solutions, pathologies have bouncing containing to branch tend third a is There tbeFR ouin onteiti asv rvt [21–27 gravity massive in exist not do solutions FLRW Stable omlgclsltosleo n ftobranches, two of one on lie solutions Cosmological nldn pn2itrcin oie GR, modifies interactions spin-2 Including spoetdfo ag unu orcin u oan to due corrections quantum large from protected is efind We m ob tbeb i-agnucleosynthesis. big-bang by stable be to eeial lge yisaiiis We instabilities. by plagued generically e epeec fa xr pn2fil.Pre- field. spin-2 extra an of presence he ,dtrie ytesi- interaction spin-2 the by determined y, 1 M xrml ls oΛD,btwt a with but ΛCDM, to close extremely hoyapoce eea relativity general approaches theory e → f ierprubtosaon hs cosmologi- these around perturbations Linear grsrnl-ope tlwenergy low at strongly-coupled nger ob ml,teeisaiiiscan instabilities these small, be to , ,js sfrinmse r rtce by protected are masses fermion as just 0, sh ohcueZürich Hochschule ische M f sta Heidelbergrsität hudb olre hnthe than larger no be should ,4, 1, University § iversity bridge n dmR Solomon R. Adam and NORDITA-2015-31 2 h infinite- The ne alia inter ,2 5, 2, 1, hese m ]. ¶ , 2

While these avoid the aforementioned ghosts, they con- In the above, m is a mass scale and βn are dimension- tain a scalar instability at early times [29, 32, 33] that in- less interaction parameters. β0 and β4 parameterize the validates the use of linear perturbation theory and could vacuum energies in the two sectors. Guided by the ab- potentially rule these models out. For parameter values sence of ghosts and the weak equivalence principle, we 4 thought to be favored by data, this instability was found take the matter sector to be coupled to gµν . Then the to be present until recent times (i.e., a similar time to the vacuum-energy contributions from the matter sector Lm onset of cosmic acceleration) and thus seemed to spoil the are captured in β0. We can interpret gµν as the space- predictivity of bimetric cosmology. time metric used for measuring distance and time, while In this Letter we study a physically well-motivated re- fµν is an additional symmetric tensor that mixes nontriv- gion in the parameter space of bimetric theory that has ially with gravity. As we discuss further below, the two been missed in earlier work due to a ubiquitous choice metrics do not correspond to the spin-2 mass eigenstates of parameter rescaling. We demonstrate how in this re- but each contain both massive and massless components. gion the instability problem in the finite branch can be Even before fitting to observational data, the parameters resolved while the model still provides late-time acceler- in the bimetric action are subject to several theoretical ation in agreement with observations. constraints. For instance, the squared mass of the mas- Our search for viable bimetric cosmologies will be sive spin-2 field needs to be positive, it must not violate guided by the precise agreement of GR with data on all the Higuchi bound [59, 60], and ghost modes should be scales, which motivates us to study models of modified absent. gravity which are close to their GR limit. Often this In terms of the Einstein tensor, Gµν , the equations of limit is dismayingly trivial; if a theory of modified grav- motion for the two metrics take the form ity is meant to produce late-time self-acceleration in the 1 G (g)+ m2V g = T , (5) absence of a cosmological constant degenerate with vac- µν µν M 2 µν uum energy, then we would expect that self-acceleration Pl 2 2 f to disappear as the theory approaches GR. We will see, α Gµν (f)+ m Vµν =0, (6) however, that there exists a GR limit of bigravity which (g,f) retains its self-acceleration, leading to a GR-like universe where Vµν are determined by varying the interaction with an effective cosmological constant produced purely potential, V . Taking the divergence of eq. (5) and using by the spin-2 interactions. the Bianchi identity leads to the Bianchi constraint, µ V g =0. (7) Bimetric gravity.— The ghost-free action for bigravity ∇(g) µν containing metrics gµν and fµν is given by [4, 43] The analogous equation for fµν carries no additional in- formation due to the general covariance of the action. M 2 M 2 S = d4x Pl √gR(g) f fR(f) Finally, note that the action (1) has a status similar 2 2 Z h − − p to Proca theory on curved backgrounds. It is therefore 2 2 X expected to require an analogue of the Higgs mechanism, + m MPl√g V ( )+ √g m (g, Φi) . (1) L i with new degrees of freedom, in order to have improved quantum behavior. The search for a ghost-free Higgs Here M and M are the Planck masses for g and f , Pl f µν µν mechanism for gravity is still in progress [61]. respectively, and we will frequently refer to their ratio, The GR limit.— When bigravity is linearized around M ¯ 2 5 α f . (2) proportional backgrounds fµν = c g¯µν with constant c, ≡ M Pl 1 gµν =g ¯µν + δgµν , (8) The potential V (X) is constructed from the elementary MPl symmetric polynomials en(X) of the eigenvalues of the 2 c fµν = c g¯µν + δfµν , (9) matrix X g−1f, defined by Mf ≡ p µ α µα the canonically-normalized perturbations can be diago- X αX ν g fαν , (3) ≡ nalized into massless modes δGµν and massive modes δM as [4, 62] and has the form [8, 43],3 µν δGµν (δgµν + cαδfµν ) , (10) 3 ∝ X X δMµν (δfµν cαδgµν ) . (11) √g V ( )= √gβ0 + √g βnen( )+ fβ4. (4) ∝ − nX=1 p

4 More general matter couplings not constrained by these require- ments have been studied in Refs. [20, 44–58]. 3 This is a generalization of the massive-gravity potential [8] (to 5 These correspond to Einstein spaces and, for nonvanishing α, which it reduces for fµν = ηµν and a restricted set of βn) given solve the ﬁeld equations only in vacuum. A quartic equation in Ref. [43]. determines c = c(βn, α). 3

Notice that when α 0 (or M M ), the massless the f metric to exactly follow the g metric (both at the → Pl ≫ f state aligns with δgµν , i.e., up to normalization, background and perturbative levels), while the g sector remains weakly coupled. δG δg + (α2). (12) µν → µν O Strong-coupling scales.— We now argue that at en-

Because gµν is the physical metric, this suggests that ergy scales relevant to cosmology, this model avoids α 0 is the general-relativity limit of bigravity.6 We known strong-coupling issues, sometimes contrary to in- will→ see below that the nonlinear field equations indeed tuition gained from massive gravity. reduce to Einstein’s equations for α = 0 and that the There are several strong-coupling scales one might ex- limit is continuous. Thus gµν is close to a GR solution pect to arise. At an energy scale k, the f sector has an for sufficiently small values of α. We therefore identify effective coupling k/Mf , as can be seen from expanding MPl with the measured physical Planck mass whenever the Einstein-Hilbert action in δfµν /Mf , just as in GR. α 1, holding it fixed while making M smaller. In- Then, for small but nonzero α, which is the case of in- ≪ f terestingly, in the bimetric setup a large physical Planck terest here, one might worry that perturbations of fµν mass is correlated with the fact that gravity is approx- with momentum k become strongly coupled at low scales k M . However, we have seen that in the limit of imated well by a massless field. In other words, when ∼ f bimetric theory is close to GR, the gravitational force is infinite strong coupling, Mf = 0, fµν becomes nondy- naturally weak. namical and is entirely determined in terms of gµν , while The GR limit can be directly realized at the nonlinear the gµν equation is degenerate with GR and its pertur- level [64, 65]. The metric potentials satisfy the identity bations remain weakly coupled. Due to the continuity of the limit, we expect that, for small enough α, strong- µα g µα f µ coupling effects will continue to not affect the g sector, √gg Vαν + ff Vαν = √gVδ ν , (13) p even when perturbations of fµν are strongly coupled at where V is the potential in the action (1). For Mf = 0, relatively small energy scales. In practice, however, since f the fµν equation (6) gives Vµν = 0, an algebraic con- the measured value of MPl is very large, even reasonably straint on fµν . Then, using the above identity, the gµν high values of Mf can still lead to small α. In cosmo- equation (5) becomes logical applications, all observable perturbations satisfy k/M 1 for M 100H 10−31 eV, roughly the f ≪ f ≫ 0 ∼ 2 1 scale at which linear cosmological perturbation theory Gµν (g)+ m Vgµν = 2 Tµν . (14) MPl breaks down at recent times, so that perturbations of fµν remain weakly coupled in any case. Since Tµν is conserved, taking the divergence gives Another potentially-problematic scale is associated with the helicity-0 mode of the massive graviton. In mas- ∂ V =0. (15) µ sive gravity, this mode becomes strongly coupled at the scale [72, 73] We see that eq. (14) is the Einstein equation for gµν with cosmological constant m2V . Remarkably, because 2 1/3 V depends on f and all the β , this effective cosmologi- Λ3 m MPl , (16) µν n ≡ cal constant is generically not simply the vacuum energy from matter loops (which is parameterized by β0). Even where m is defined to coincide with the Fierz-Pauli mass in the GR limit, the impact of the spin-2 interactions [1] on flat backgrounds. This scale is rather small, −13 −1 −33 remains and bigravity’s self-acceleration survives. Λ3 10 eV (1000 km) for m H0 10 eV, ∼ ∼ ∼ ∼ It is straightforward to see that, unlike the m 0 and severely restricts the applicability of massive gravity limit, the α 0 limit is not affected by the van Dam-→ [74]. The same scale also appears in the decoupling-limit Veltman-Zakharov→ (vDVZ) discontinuity [66, 67]. The analysis of bimetric theory [37], where m is now the pa- cause of this discontinuity is the Bianchi constraint (7) rameter in front of the potential in the action (1). In which constrains the solutions even when m = 0. On the limit α 0, the f sector approaches massive gravity → the contrary, when α 0, the Bianchi constraint simply [65] and one might worry that the strong-coupling prob- reduces to eq. (15) and→ is automatically satisfied. lem persists or becomes worse with the emergence of an f even lower scale (m2M )1/3. This is not the case. In The conditions Vµν = 0 and ∂µV = 0 determine fµν f 2 the bimetric context, the scale defined in eq. (16) is not algebraically in terms of gµν , generically as fµν = c gµν . physical, since m2 is degenerate with the β . The physi- In the limit Mf = 0, the f sector is infinitely strongly n coupled.7 Due to the nontrivial potential, this causes cally relevant strong-coupling scale must be defined with respect to the bimetric Fierz-Pauli mass [62],

2 2 1 2 3 m = m +1 cβ1 +2c β2 + c β3 , (17) 6 FP c2α2 See Ref. [63] for an early discussion of such a limit. 7 Strongly-coupled gravity in the context of GR has been studied, for instance, in Refs. [68–71] and has been argued to allow for a which is only defined around proportional backgrounds, simplified quantum-mechanical treatment. f = c2g . In the massive-gravity limit, α , the µν µν → ∞ 4 helicity-0 mode is mostly contained in g with a strong- with K Y˙ /Y . The final ingredient is the Bianchi con- coupling scale straint (7),≡ which yields

2 2 1/3 (HX Ky) β1 +2β2y + β3y =0. (26) Λ3 mFPMPl , (18) − ≡ Taking the first or second term of eq. (26) to vanish se- consistent with eq. (16) for appropriately restricted pa- lects the so-called dynamical or algebraic branches, re- rameters. However, in the GR limit, α 0, the helicity-0 → spectively. Perturbations in the algebraic branch are mode resides mostly in f, where the strong-coupling scale pathological [29], so we will consider the dynamical is branch in which the f-metric lapse is fixed, 2 1/3 1/3 m M Ky Λ˜ m2 M Pl (β ) , (19) X = . (27) 3 ≡ FP f → α O n H which is no longer small. Note that for solutions that Inserting this into the fµν equation (25) transforms it admit this limit, c becomes independent of α. We can into an “alternate” Friedmann equation, also consider the α 0 limit of eq. (18), to verify that → 2 2 2 β1 2 the small part of the helicity-0 mode in g is not strongly 3α H = m +3β2 +3β3y + β4y . (28) y coupled, We take at least two of the βn for n 1 to be nonzero in m2M 1/3 order to ensure the existence of interesting≥ solutions in Λ Pl (β ) . (20) 3 → α2 O n the GR limit α 0. The solutions to eq. (28) in the GR limit are always→ on the “finite” branch, i.e., y evolves

This is even higher than Λ˜ 3. Therefore the strong- from 0 to a finite late-time value. The perturbations coupling issues with the helicity-0 mode are alleviated, on this branch are healthy except for a scalar instability, rather than exacerbated, when α 0. which we discuss below. → Equation (28) has two features which are useful for our Cosmology.— We now proceed to apply the above argu- purposes. First, in the limit α 0 it tends to a poly- ments to the particular example of a homogeneous and nomial constraint that leads to→ a constant solution for isotropic universe. We will take both metrics to be of the 8 y, so that the potential term in the Friedmann equation diagonal FLRW form [14–16], (23) becomes a cosmological constant. This provides an µ ν 2 2 i j explicit example of the statement above that as α 0, gµν dx dx = dt + a (t)δij dx dx , (21) → − the theory approaches general relativity with an effective µ ν 2 2 2 i j fµν dx dx = X (t)dt + Y (t)δij dx dx , (22) cosmological constant (even with β0 = 0). Recall that − even though the theory approaches GR in this limit, the where we can freely choose the cosmic-time coordinate for bigravity interactions survive in the form of this constant. gµν (g00 = 1) because of general covariance. Because The other useful feature is that, because eq. (28) does not − matter couples minimally to gµν , this choice is physical, involve ρ, it can be used to rephrase the potential term and a(t) corresponds to the scale factor inferred from in eq. (23) in terms of the Hubble rate. This will allow us observations. We furthermore take the matter source to to determine the time-dependence of the potential term µ 9 be a perfect fluid, T ν = diag( ρ,p,p,p). The g-metric order by order in α. equation (5) leads to the Friedmann− equation, The effective cosmological constant.— Let us illus- 2 ρ 2 2 3 trate the new viable bimetric cosmologies qualitatively 3H = 2 + m β0 +3β1y +3β2y + β3y , (23) 10 M by selecting the model with β0 = β3 = β4 = 0, which Pl we will refer to as the β1β2 model. The Friedmann and where the Hubble rate is defined as H a/a˙ and the “alternate” Friedmann equations (23) and (28) are ≡ ratio of the scale factors is ρ 3H2 = +3m2 β y + β y2 , (29) Y M 2 1 2 y . (24) Pl ≡ a 2 2 2 β1 3α H = m +3β2 . (30) The analogous equation for the f metric is y

m2 β β β 3K2 = X2 1 +3 2 +3 3 + β , (25) α2 y3 y2 y 4 9 One can also combine eqs. (23) and (28) to obtain a quartic equation for y involving ρ [14–17, 31], but this is more cumbersome as it involves higher powers of y than eq. (28) does. 10 Since we are interested in ﬁnding self-accelerating solutions in 8 See Ref. [75] and the references therein for other possible metrics the absence of vacuum energy, we will set β0 = 0 herein, but in bimetric cosmology. emphasize that this is not necessary. 5

We can use eq. (30) to eliminate y in eq. (29). It is 2 Model Λeff (α → 0) O(α ) correction 2 2 2 instructive to work in the GR limit where eq. (30) gives 2 β1 2 2 β1 2 ρ β1 2 β1, β2 =6 0 − m − 2 α 2 − m 3 β2 9 β2 2MPl 3β2 3/2 3/2 α→0 1 β1 8 β1 2 β1 2 ρ 8β1 2 β1, β3 =6 0 m α 2 − m y . (31) 3√3 √ β3 β3 3M 9√ 3β3 3 β − Pl − −−−→− 2 4 3 2 4 3 / 3 / β1 2 β1 2 ρ β1 2 β1, β4 =6 0 3 3 m − − α 2 + 3 3 m √ β4 β4 MPl √ β4 The α 0 limit is nonsingular only if both β1 and β2 −3 2 3 − → β2 2 β2 2 ρ 2β2 2 are nonzero. Plugging this into eq. (29) we obtain β2, β3 =6 0 2 2 m − 2 α 2 + 2 m β3 β3 MPl β3 2 2 β2 2 β2 2 ρ 9β2 2 2 4 6 − 2 − 2 β , β = 0 9 β4 m 3 β4 α M β4 m 2 ρ 2 β1 2 Pl 3H = 2 m . (32) MPl − 3 β2 TABLE I. The effective cosmological constant and lowest- The effective cosmological constant is order corrections (which are time-dependent through ρ) for a variety of two-parameter models. We have chosen solution 2 branches which lead to positive Λeff for appropriate signs of 2 β1 2 Λeff = m . (33) the βn, and generally take β1 ≥ 0 based on viability conditions −3 β2 [19]. The β3, β4 =6 0 model does not possess a finite-branch solution [19]. Late-time acceleration requires β2 < 0. When we are not exactly in the GR limit, we should consider corrections to eq. (32), where Φ represents any of the scalar perturbation vari- 2 4 2 ρ β1 m 2 2 2 ables, N ln a, and we have taken the limit k aH 3H = + 3α H 2β2m ≡ ≫ 2 2 2 2 2 where k is the comoving wavenumber. The eigenfrequen- MPl 3 (H α β2m ) − − cies ω were presented for particular models in Ref. [32], ρ 2 β2 α2β2 = 1 m2 1 H2 + (α4). (34) where it was found that all models with viable back- 2 2 2 MPl − 3 β2 − 3β2 O grounds have ω < 0 at early times, revealing a gradient instability that only ends at a very low redshift. Using This expansion is valid as long as the formulation of the linearized equations of motion pre- 2 sented in Ref. [33], we can write the eigenfrequencies for 2 β2m < general βn and α in the compact form [41] H 2 . (35) ∼ α 2 2 2 ′ Rearranging and again keeping terms up to (α ), we aH 2 β1 +4β2y +3β3y y O ω =1+ 2 find a standard Friedmann equation with a time-varying k 3y (β1 +2β2y + β3y ) effective cosmological constant given by 2 2 2 ′2 1+ α y β1 β3y y − , (38) 2 2 2 3α2y3 ρ˜(1 + w) 2 β1 2 2 β1 2 ρ β1 2 4 − Λeff = m 2 α 2 m + (α ). −3 β2 − 9 β2 2MPl − 3β2 O (36) whereρ ˜ ρ/m2M 2 and primes denote d/d ln a. ≡ Pl Because matter is coupled minimally to gµν , it will have We apply this to the β1β2 model. Assuming a universe the standard behavior ρ a−3(1+w), where w = p/ρ is dominated by dust (w = 0), ω2 crosses zero when11 the equation-of-state parameter,∼ allowing ρ to stand in for time. This captures the first hint of the dynamical 2 2 2 2 5 2 2 2 2 4 18α β2 α β1 +4β2 y +9α β1 α β1 + 10β2 y dark energy that is typical of bigravity [16–20]. +48α2β2β y3 +6β 2α2β2 β2 y2 6β2β y β3 =0 . These results generalize easily to other parameter com- 1 2 2 1 − 2 − 1 2 − 1 binations. We list the effective cosmological constant (39) up to (α2) for all the two-parameter models (setting O β0 = 0) in table I. We remind the reader that, in order Solving this for y, we can then use eq. (30) to determine for the α 0 limit to be well-behaved, at least two of the value of Hubble rate at the transition era, before → the βn parameters (excluding the vacuum energy contri- which the gradient instability is present and after which bution, β0) must be nonzero. it vanishes. While this solution is too complicated to write down explicitly, in the limit α 0 the leading- Exorcising the instability.— The stability of cosmo- → logical perturbations in bigravity was investigated in Ref. [32] by determining the full solutions to the linearized Einstein equations in the subhorizon régime. The perturbations were shown to obey a WKB solution given 11 We have used eqs. (29) and (30) and their derivatives to solve for 2 by y′ and ρ in eq. (38) in terms of βn and y [31]. Note that ω = 0 does not imply strong coupling because, while the gradient terms Φ eiωN , (37) vanish, the kinetic terms remain nonzero. ∼ 6

A natural requirement would be to push the instability Era of transition to stability H⋆ α Mf 16 17 outside the range of the effective field theory, i.e., above BBN 10− eV 10− 100 GeV either the cut-off scale where new physics must enter, or ˜ 2 1/3 3 31 3 Λ3 = m MPl/α 10− eV 10− 10− eV the strong-coupling scale where tree-level unitary breaks 13 55 27 GUT-scale inflation 10 GeV 10− 10− eV down.13 The cut-off scale in massive and bimetric gravity 19 61 33 MPl 10 GeV 10− 10− eV is not known. The strong-coupling scale, to the extent it is understood, was discussed above. Here we focus on ob- TABLE II. The values of α and Mf for a few choices of the servational constraints. It is natural to demand that the era at which perturbations become stable. instability lie beyond some important cosmic era which we can indirectly probe, such as big-bang nucleosynthesis (BBN) or inflation. Both of these possibilities are then order term is remarkably simple,12 likely to be observationally safe as long as the Universe is

2 decelerating (e.g., is radiation-dominated) after inflation, 2 β2m 0 because the instability is only a problem for subhorizon H⋆ = + (α ), (40) ±√3α2 O modes with large k/aH, and during a decelerating epoch modes with fixed comoving wavelength always become where H⋆ is defined as the Hubble rate at the time when 2 smaller with respect to the horizon. Consider, as an ex- ω = 0, i.e., after which the gradient instability is absent. ample, that the transition to stability occurs between We pick the negative branch of eq. (40) for physical rea- 2 inflation and BBN. During that period, modes will grow sons, i.e., so that H⋆ > 0 given that β2 < 0. We have rapidly on small scales, but those will be far, far smaller checked explicitly that by solving for y with this value of than the modes relevant for the cosmic microwave back- H and plugging it into ω2, all terms up to (α2) vanish. O ground or large-scale structure. One might worry that in- Interestingly, eq. (40) is the same as the condition (35) flation’s ability to set initial conditions is spoiled in this for the small-α expansion of the background solution to scenario (assuming that the linear theory is even valid be valid. Therefore, simply by pushing the instability during inflation, which is not guaranteed due to the argu- back to early times, one gets late-time bimetric dynam- ments above). However, the instability should be absent ics that can be described as perturbative corrections to during inflation; notice from eq. (38) that ω2 generically GR, except for the effective cosmological constant which becomes large and positive for w close to 1.14 Therefore remains nonperturbative. This is nontrivial; while we ex- the instability would not affect the generation− of primor- pect everything to reduce to GR at late times when we dial perturbations during inflation. If the instability later can expand in αH/√βnm, there could in principle have appears with the onset of radiation domination, it would been earlier times during which perturbations were stable only affect small scales which are irrelevant for present- but still fundamentally different than in GR. day cosmology. We can rewrite eq. (40) in more physical terms as If the instability ends at the time of BBN, Mf can be as high as about 100 GeV, far larger than the wavenum- √ 2 2 3 3 β2 2 bers probed by cosmological observations. We remind H⋆ = 2 HΛ, (41) − 2α β1 the reader that for such a “large” Mf , perturbations in the Einstein-Hilbert term for fµν remain weakly-coupled where HΛ is the far-future value of H and should be for all observationally-relevant k. comparable to the present Hubble rate, H . For β 0 1 While analytic results like eq. (40) cannot be obtained β , this implies simply | | ∼ | 2| for most of the other two-parameter models, we have checked that in each case the relevant behavior, H⋆ H0 15 ∼ H⋆ . (42) H /α, holds. The values given in table II are therefore ∼ α Λ fairly model-independent. We see that as we approach the GR limit, the smaller The other pathology that is typical of massive and bi- one takes the f-metric Planck mass, the earlier in time metric gravity, the Higuchi ghost, is not present in these bigravity’s gradient instability is cured. Our goal is to make this era so early as to be effectively unobservable. One has a variety of choices for the scale where the insta- 13 bility sets in; the values of α and Mf for various choices These two are not always the same, and may not be in massive and bimetric gravity [76, 77]. are summarized in table II. 14 This depends on the exact βn parameters and the evolution of y. Background viability requires β1 > 0 [19], so as long as β3 ≤ 0, at the very least the last term in eq. (38) is large and manifestly positive. 12 15 While eq. (40) only holds exactly in the presence of dust, w = 0, Specifically, this holds in the models with β1 =6 0. The gradient for other reasonable equations of state, such as radiation (w = instability is absent from the β2β3 and β2β4 models at early 1/3), it will only be modified by an O(1) factor. Since we will be times [32]. These were shown in Ref. [19] to have problematic using this analysis only to make order-of-magnitude estimates, background behavior at early times, but these again can be made the exact factors are unimportant. unobservably early in the GR limit. 7

models. There is a simple condition for the absence of well-motivated but heretofore underexplored region of pa- this ghost, dρ/dy < 0 [35, 36] (see also Refs. [33, 37]). rameter space in bimetric gravity can lead to cosmolog- Because for normal matter ρ is always decreasing with ical solutions which are observationally viable and close time, this amounts to demanding that y be increasing. to general relativity, with an effective cosmological con- In the “finite-branch” solutions which we are considering, stant that is set by the spin-2 interaction scale m. In y evolves monotonically from 0 at early times to a fixed this limit, obtained by taking a small f-metric Planck positive value at late times, and so the Higuchi bound is mass, the gradient instability that seems to generically always satisfied [41]. plague bimetric models at late times is relegated to the very early Universe, where it can be either made unob- Parameter rescalings.— We have presented a physi- servable or pushed outside the régime of validity of the cally well-motivated region of bimetric parameter space, effective theory. This instability had been considered in near the GR limit, in which observable cosmological per- previous work to make bimetric cosmologies nonpredic- turbations are stable and yet self-acceleration remains. tive even at late times. Furthermore, in this limit the One is naturally led to ask how this has been missed by theory avoids the usual low-scale strong-coupling issue the many previous studies of bimetric cosmology. The is- that affects the helicity-0 sector in the massive-gravity sue lies in a rescaling which leaves the action (1) invariant limit. [28, 62], What is encouraging is that the one property of bi- 1 f Ω2f , β β ,M ΩM , (43) gravity which survives in the small-α limit is its cosmo- µν → µν n → Ωn n f → f logically most useful feature, the technically-natural dark and hence gives rise to a redundant parameter. It has energy scale. In other words, the effective cosmological become common to let α play this role and perform the constant of bigravity in a region close to GR is not just rescaling Ω = 1/α such that α is set to unity. While the vacuum-energy contribution and can give rise to self- our results do not invalidate this rescaling, they do show acceleration in its absence. that it picks out a particular region of parameter space The model we have presented is expected to be ex- which may not capture all physically-meaningful situa- tremely close to GR at all but very high energy scales. tions. In particular, the α 0 limit, in which the theory In particular the Newtonian limit is well-behaved; unlike approaches GR—the behavior→ at the heart of our remov- m2 0, which suffers from the vDVZ discontinuity, the → ing the gradient instability—would look extremely odd GR limit α 0 is completely smooth because all the → after this rescaling: the βn would not only be very large, helicity states of the massive spin-2 mode decouple from 16 but each βn+1 would be parametrically larger than βn. matter. Note also that massive gravity does not possess Therefore, studies which set α to unity could in principle such a continuous GR limit. have found the GR-like solutions which we study here, It is worth emphasizing that the α 0 limit brings → but only by looking at what would have appeared to be bimetric theory arbitrarily close to GR even for a large a highly unnatural and tuned set of parameters, even value of the spin-2 mass scale, m H0. The presence ≫ though they have a simple and sensible physical explana- of heavy spin-2 fields in the Universe is therefore not tion. Without performing this rescaling, we can simply excluded as long as their self-interaction scale (set by take the nonzero β to be (1) and consider that we are Mf ) is sufficiently small compared to MPl. In this case, n O in the small-Mf régime. however, the βn parameters need to be highly tuned for It is clear that in phenomenological studies of bigrav- the effective cosmological constant small enough to be ity, α must not automatically be set to unity. When compatible with observations.17 Note however that, since working with a two-βn model, perhaps a more sensible the βn are protected against loop corrections [42, 78, 79], rescaling would be one such that the two βn are equal this tuning does not violate technical naturalness. to each other (up to a possible sign). They can further Finally we comment on the potential observable signa- be absorbed into m2. In this case, the free parameters tures of this theory. While at low energies, corresponding are effectively the spin-2 interaction scale, m2, and the to recent cosmological epochs, this limit of bigravity is ex- f-metric Planck mass, Mf . Their effects decouple nicely: tremely close to GR, there may be observable effects at Mf controls the earliness of the instability, while m sets early times when the effects of strong coupling become the acceleration scale. Alternatively, one may consider important. In this case, given by H >H⋆, the small-α that the rescaling (43) simply tells us that rather differ- approximation breaks down and modified-gravity effects ent regions of parameter space happen to have the same must be taken into account. This may be particularly solutions, and therefore not perform any rescaling a pri- important for inflation, which will see such effects unless ori at all. Mf is extraordinarily small. A better understanding of Summary and discussion.— We have shown that a

17 Indeed, without this tuning of the βn, the interaction term would 16 We can recast this as a large m2, but there would remain a lead to acceleration at an unacceptably early epoch. This sce- specific tuning among the βn of the form βn/βn+1 ∼ ǫ, where ǫ nario is related to the findings of Ref. [36], where it was shown is the value of α before the rescaling. that the instability becomes negligible for large values of m. 8 strong coupling in the fµν sector will therefore point the Edvard Mörtsell, Luigi Pilo, Marit Sandstad, and Sergey way towards tests of this important region of bimetric Sibiryakov for useful discussions. We thank the “Ex- parameter space, since at this point it is not clear how tended Theories of Gravity” workshop at Nordita for pro- to perform computations in the strong-coupling regime. viding a stimulating atmosphere during the completion There may also be effects related to the Vainshtein mech- of this work. Y.A. and F.K. acknowledge support from anism [80, 81]. We conclude that the closeness of this DFG through the project TRR33 “The Dark Universe.” theory to GR is both a blessing and a curse: while it is F.K. is also supported by the Graduate College “Astro- behind the exorcism of the gradient instability and brings physics of Fundamental Probes of Gravity.” The work the theory in excellent agreement with experiments, it of A.S.M. is supported by ERC grant No. 615203 un- presents a serious observational challenge if it is to be der the FP7 and the Swiss National Science Foundation compared against GR. It is nevertheless encouraging that through the NCCR SwissMAP. A.R.S. acknowledges sup- this “GR-adjacent” bigravity naturally explains cosmic port from the STFC. acceleration while avoiding the instabilities that plague other bimetric models, and therefore merits serious con- sideration.

ACKNOWLEDGMENTS

We are grateful to Luca Amendola, Jonas Enander, Matteo Fasiello, Emir Gümrük¸cüo˘glu, Nima Khosravi,

[1] M. Fierz and W. Pauli, “On relativistic wave equations mulation,” Phys.Lett. B715, 335–339 (2012), for particles of arbitrary spin in an electromagnetic field,” arXiv:1203.5283 [hep-th]. Proc.Roy.Soc.Lond. A173, 211–232 (1939). [12] Adam G. Riess et al. (Supernova Search [2] C.J. Isham, Abdus Salam, and J.A. Strathdee, “F- Team), “Observational evidence from supernovae dominance of gravity,” Phys.Rev. D3, 867–873 (1971). for an accelerating universe and a cosmologi- [3] D.G. Boulware and Stanley Deser, “Can gravitation have cal constant,” Astron.J. 116, 1009–1038 (1998), a finite range?” Phys.Rev. D6, 3368–3382 (1972). arXiv:astro-ph/9805201 [astro-ph]. [4] S.F. Hassan and Rachel A. Rosen, “Bimet- [13] S. Perlmutter et al. (Supernova Cosmology Project), ric Gravity from Ghost-free Massive Gravity,” “Measurements of Omega and Lambda from 42 high JHEP 1202, 126 (2012), arXiv:1109.3515 [hep-th]. redshift supernovae,” Astrophys.J. 517, 565–586 (1999), [5] S.F. Hassan and Rachel A. Rosen, “Confirmation of the arXiv:astro-ph/9812133 [astro-ph]. Secondary Constraint and Absence of Ghost in Massive [14] Mikhail S. Volkov, “Cosmological solutions Gravity and Bimetric Gravity,” JHEP 1204, 123 (2012), with massive gravitons in the bigravity theory,” arXiv:1111.2070 [hep-th]. JHEP 1201, 035 (2012), arXiv:1110.6153 [hep-th]. [6] Paolo Creminelli, Alberto Nicolis, Michele Papucci, [15] D. Comelli, M. Crisostomi, F. Nesti, and L. Pilo, and Enrico Trincherini, “Ghosts in massive gravity,” “FRW Cosmology in Ghost Free Massive Gravity,” JHEP 0509, 003 (2005), arXiv:hep-th/0505147 [hep-th]. JHEP 1203, 067 (2012), arXiv:1111.1983 [hep-th]. [7] Claudia de Rham and Gregory Gabadadze, [16] Mikael von Strauss, Angnis Schmidt-May, Jonas “Generalization of the Fierz-Pauli Ac- Enander, Edvard Mörtsell, and S.F. Hassan, tion,” Phys.Rev. D82, 044020 (2010), “Cosmological Solutions in Bimetric Gravity and arXiv:1007.0443 [hep-th]. their Observational Tests,” JCAP 1203, 042 (2012), [8] Claudia de Rham, Gregory Gabadadze, and arXiv:1111.1655 [gr-qc]. Andrew J. Tolley, “Resummation of Massive [17] Yashar Akrami, Tomi S. Koivisto, and Marit Sand- Gravity,” Phys.Rev.Lett. 106, 231101 (2011), stad, “Accelerated expansion from ghost-free bigrav- arXiv:1011.1232 [hep-th]. ity: a statistical analysis with improved generality,” [9] S.F. Hassan and Rachel A. Rosen, “Resolv- JHEP 1303, 099 (2013), arXiv:1209.0457 [astro-ph.CO]. ing the Ghost Problem in non-Linear Mas- [18] Yashar Akrami, Tomi S. Koivisto, and Marit Sandstad, sive Gravity,” Phys.Rev.Lett. 108, 041101 (2012), “Cosmological constraints on ghost-free bigravity: back- arXiv:1106.3344 [hep-th]. ground dynamics and late-time acceleration,” (2013), [10] S.F. Hassan, Rachel A. Rosen, and Angnis Schmidt- arXiv:1302.5268 [astro-ph.CO]. May, “Ghost-free Massive Gravity with a Gen- [19] Frank Könnig, Aashay Patil, and Luca Amendola, “Vi- eral Reference Metric,” JHEP 1202, 026 (2012), able cosmological solutions in massive bimetric gravity,” arXiv:1109.3230 [hep-th]. JCAP 1403, 029 (2014), arXiv:1312.3208 [astro-ph.CO]. [11] S.F. Hassan, Angnis Schmidt-May, and Mikael [20] Jonas Enander, Adam R. Solomon, Yashar Akrami, von Strauss, “Proof of Consistency of Nonlin- and Edvard Mortsell, “Cosmic expansion histories in ear Massive Gravity in the Stúckelberg For- massive bigravity with symmetric matter coupling,” 9

JCAP 01, 006 (2015), arXiv:1409.2860 [astro-ph.CO]. [38] Jonas Enander, Yashar Akrami, Edvard [21] G. D’Amico, C. de Rham, S. Dubovsky, Mörtsell, Malin Renneby, and Adam R. G. Gabadadze, D. Pirtskhalava, et al., “Mas- Solomon, “Integrated Sachs-Wolfe effect in massive Cosmologies,” Phys.Rev. D84, 124046 (2011), sive bigravity,” Phys.Rev. D91, 084046 (2015), arXiv:1108.5231 [hep-th]. arXiv:1501.02140 [astro-ph.CO]. [22] A. Emir Gümrük¸cüo˘glu, Chunshan Lin, and [39] Luca Amendola, Frank Knnig, Matteo Martinelli, Shinji Mukohyama, “Open FRW universes and Valeria Pettorino, and Miguel Zumalacarregui, self-acceleration from nonlinear massive gravity,” “Surfing gravitational waves: can bigravity survive JCAP 1111, 030 (2011), arXiv:1109.3845 [hep-th]. growing tensor modes?” JCAP 1505, 052 (2015), [23] A. Emir Gümrük¸cüo˘glu, Chunshan Lin, and Shinji arXiv:1503.02490 [astro-ph.CO]. Mukohyama, “Cosmological perturbations of self- [40] Matthew Johnson and Alexandra Terrana, “Tensor accelerating universe in nonlinear massive gravity,” Modes in Bigravity: Primordial to Present,” (2015), JCAP 1203, 006 (2012), arXiv:1111.4107 [hep-th]. arXiv:1503.05560 [astro-ph.CO]. [24] Babak Vakili and Nima Khosravi, “Classical and quan- [41] Frank Könnig, “Higuchi Ghosts and Gra- tum massive cosmology for the open FRW universe,” dient Instabilities in Bimetric Grav- Phys.Rev. D85, 083529 (2012), arXiv:1204.1456 [gr-qc]. ity,” Phys.Rev. D91, 104019 (2015), [25] Antonio De Felice, A. Emir Gümrük¸cüo˘glu, and arXiv:1503.07436 [astro-ph.CO]. Shinji Mukohyama, “Massive gravity: nonlin- [42] Claudia de Rham, Lavinia Heisenberg, and ear instability of the homogeneous and isotropic Raquel H. Ribeiro, “Quantum Corrections in universe,” Phys.Rev.Lett. 109, 171101 (2012), Massive Gravity,” Phys.Rev. D88, 084058 (2013), arXiv:1206.2080 [hep-th]. arXiv:1307.7169 [hep-th]. [26] Matteo Fasiello and Andrew J. Tolley, “Cosmological per- [43] S.F. Hassan and Rachel A. Rosen, “On Non-Linear turbations in Massive Gravity and the Higuchi bound,” Actions for Massive Gravity,” JHEP 1107, 009 (2011), JCAP 1211, 035 (2012), arXiv:1206.3852 [hep-th]. arXiv:1103.6055 [hep-th]. [27] Antonio De Felice, A. Emir Gümrük¸cüo˘glu, Chun- [44] Yashar Akrami, Tomi S. Koivisto, David F. Mota, shan Lin, and Shinji Mukohyama, “Nonlinear sta- and Marit Sandstad, “Bimetric gravity doubly cou- bility of cosmological solutions in massive gravity,” pled to matter: theory and cosmological implications,” JCAP 1305, 035 (2013), arXiv:1303.4154 [hep-th]. JCAP 1310, 046 (2013), arXiv:1306.0004 [hep-th]. [28] Marcus Berg, Igor Buchberger, Jonas Enander, Ed- [45] Nicola Tamanini, Emmanuel N. Saridakis, and Tomi S. vard Mörtsell, and Stefan Sjörs, “Growth Histories Koivisto, “The Cosmology of Interacting Spin-2 Fields,” in Bimetric Massive Gravity,” JCAP 1212, 021 (2012), JCAP 1402, 015 (2014), arXiv:1307.5984 [hep-th]. arXiv:1206.3496 [gr-qc]. [46] Yashar Akrami, Tomi S. Koivisto, and Adam R. [29] D. Comelli, M. Crisostomi, and L. Pilo, “Perturbations Solomon, “The nature of spacetime in bigravity: in Massive Gravity Cosmology,” JHEP 1206, 085 (2012), two metrics or none?” Gen.Rel.Grav. 47, 1838 (2014), arXiv:1202.1986 [hep-th]. arXiv:1404.0006 [gr-qc]. [30] Frank Könnig and Luca Amendola, “In- [47] Yasuho Yamashita, Antonio De Felice, and stability in a minimal bimetric grav- Takahiro Tanaka, “Appearance of Boulware- ity model,” Phys.Rev. D90, 044030 (2014), Deser ghost in bigravity with doubly cou- arXiv:1402.1988 [astro-ph.CO]. pled matter,” Int.J.Mod.Phys. D23, 3003 (2014), [31] Adam R. Solomon, Yashar Akrami, and arXiv:1408.0487 [hep-th]. Tomi S. Koivisto, “Linear growth of structure [48] Claudia de Rham, Lavinia Heisenberg, and Raquel H. in massive bigravity,” JCAP 1410, 066 (2014), Ribeiro, “On couplings to matter in massive (bi- arXiv:1404.4061 [astro-ph.CO]. )gravity,” Class.Quant.Grav. 32, 035022 (2015), [32] Frank Könnig, Yashar Akrami, Luca Amendola, arXiv:1408.1678 [hep-th]. Mariele Motta, and Adam R. Solomon, “Sta- [49] Claudia de Rham, Lavinia Heisenberg, and Raquel H. ble and unstable cosmological models in bimet- Ribeiro, “Ghosts & Matter Couplings in Massive ric massive gravity,” Phys.Rev. D90, 124014 (2014), (bi-&multi-)Gravity,” Phys.Rev. D90, 124042 (2014), arXiv:1407.4331 [astro-ph.CO]. arXiv:1409.3834 [hep-th]. [33] Macarena Lagos and Pedro G. Ferreira, “Cos- [50] Johannes Noller and Scott Melville, “The cou- mological perturbations in massive bigravity,” pling to matter in Massive, Bi- and Multi-Gravity,” JCAP 1412, 026 (2014), arXiv:1410.0207 [gr-qc]. JCAP 1501, 003 (2014), arXiv:1408.5131 [hep-th]. [34] Giulia Cusin, Ruth Durrer, Pietro Guarato, and Mariele [51] Adam R. Solomon, Jonas Enander, Yashar Akrami, Motta, “Gravitational waves in bigravity cosmology,” Tomi S. Koivisto, Frank Könnig, et al., “Cos- JCAP 1505, 030 (2015), arXiv:1412.5979 [astro-ph.CO]. mological viability of massive gravity with gener- [35] Yasuho Yamashita and Takahiro Tanaka, “Mapping alized matter coupling,” JCAP 1504, 027 (2015), the ghost free bigravity into braneworld setup,” arXiv:1409.8300 [astro-ph.CO]. JCAP 1406, 004 (2014), arXiv:1401.4336 [hep-th]. [52] Angnis Schmidt-May, “Mass eigenstates in bimetric [36] Antonio De Felice, A. Emir Gümrük¸cüo˘glu, Shinji theory with matter coupling,” JCAP 1501, 039 (2015), Mukohyama, Norihiro Tanahashi, and Takahiro arXiv:1409.3146 [gr-qc]. Tanaka, “Viable cosmology in bimetric theory,” [53] Lavinia Heisenberg, “Quantum corrections in JCAP 1406, 037 (2014), arXiv:1404.0008 [hep-th]. massive bigravity and new effective composite [37] Matteo Fasiello and Andrew J. Tolley, “Cosmological metrics,” Class.Quant.Grav. 32, 105011 (2015), Stability Bound in Massive Gravity and Bigravity,” arXiv:1410.4239 [hep-th]. JCAP 1312, 002 (2013), arXiv:1308.1647 [hep-th]. [54] A. Emir Gümrük¸cüo˘glu, Lavinia Heisenberg, and 10

Shinji Mukohyama, “Cosmological perturbations graviton mass,” JETP Lett. 12, 312 (1970). in massive gravity with doubly coupled matter,” [68] Marc Henneaux, Martin Pilati, and Claudio Teitelboim, JCAP 1502, 022 (2015), arXiv:1409.7260 [hep-th]. “Explicit Solution for the Zero Signature (Strong Cou- [55] Johannes Noller, “On Consistent Kinetic and Deriva- pling) Limit of the Propagation Amplitude in Quantum tive Interactions for Gravitons,” JCAP 1504, 025 (2015), Gravity,” Phys.Lett. B110, 123 (1982). arXiv:1409.7692 [hep-th]. [69] G. Francisco and M. Pilati, “Strong Coupling Quan- [56] A. Emir Gümrük¸cüo˘glu, Lavinia Heisenberg, Shinji tum Gravity. 3. Quasiclassical Approximation,” Mukohyama, and Norihiro Tanahashi, “Cosmol- Phys.Rev. D31, 241 (1985). ogy in bimetric theory with an effective com- [70] D.S. Salopek, “Hamilton-Jacobi solutions posite coupling to matter,” JCAP 1504, 008 (2015), for strongly coupled gravity and mat- arXiv:1501.02790 [hep-th]. ter,” Class.Quant.Grav. 15, 1185–1206 (1998), [57] Denis Comelli, Marco Crisostomi, Kazuya Koyama, arXiv:gr-qc/9802025 [gr-qc]. Luigi Pilo, and Gianmassimo Tasinato, “Cos- [71] Max Niedermaier, “The dynamics of strong coupling mology of bigravity with doubly coupled matter,” gravity,” Class.Quant.Grav. 32, 015008 (2015). JCAP 1504, 026 (2015), arXiv:1501.00864 [hep-th]. [72] Nima Arkani-Hamed, Howard Georgi, and [58] Kurt Hinterbichler and Rachel A. Rosen, “A Note on Matthew D. Schwartz, “Effective field the- Ghost-Free Matter Couplings in Massive Gravity and ory for massive gravitons and gravity in the- Multi-Gravity,” (2015), arXiv:1503.06796 [hep-th]. ory space,” Annals Phys. 305, 96–118 (2003), [59] Atsushi Higuchi, “Forbidden Mass Range for arXiv:hep-th/0210184 [hep-th]. Spin-2 Field Theory in De Sitter Space-time,” [73] Matthew D. Schwartz, “Constructing gravita- Nucl.Phys. B282, 397 (1987). tional dimensions,” Phys.Rev. D68, 024029 (2003), [60] A. Higuchi, “Massive Symmetric Tensor Field in arXiv:hep-th/0303114 [hep-th]. Space-times With a Positive Cosmological Constant,” [74] Clare Burrage, Nemanja Kaloper, and Antonio Padilla, Nucl.Phys. B325, 745–765 (1989). “Strong Coupling and Bounds on the Spin-2 Mass [61] Garrett Goon, Kurt Hinterbichler, Austin Joyce, and in Massive Gravity,” Phys.Rev.Lett. 111, 021802 (2013), Mark Trodden, “Einstein Gravity, Massive Gravity, arXiv:1211.6001 [hep-th]. Multi-Gravity and Nonlinear Realizations,” (2014), [75] Henrik Nersisyan, Yashar Akrami, and Luca Amendola, arXiv:1412.6098 [hep-th]. “Consistent metric combinations in cosmology of massive [62] S.F. Hassan, Angnis Schmidt-May, and Mikael von bigravity,” (2015), arXiv:1502.03988 [gr-qc]. Strauss, “On Consistent Theories of Massive Spin-2 [76] Ufuk Aydemir, Mohamed M. Anber, and Fields Coupled to Gravity,” JHEP 1305, 086 (2013), John F. Donoghue, “Self-healing of unitar- arXiv:1208.1515 [hep-th]. ity in effective field theories and the onset of [63] Z. Berezhiani, D. Comelli, F. Nesti, and new physics,” Phys.Rev. D86, 014025 (2012), L. Pilo, “Spontaneous Lorentz Breaking and Mas- arXiv:1203.5153 [hep-ph]. sive Gravity,” Phys.Rev.Lett. 99, 131101 (2007), [77] Claudia de Rham, “Massive Gravity,” arXiv:hep-th/0703264 [HEP-TH]. Living Rev.Rel. 17, 7 (2014), arXiv:1401.4173 [hep-th]. [64] Valentina Baccetti, Prado Martin-Moruno, and [78] Kurt Hinterbichler, Mark Trodden, and Daniel Matt Visser, “Massive gravity from bimetric Wesley, “Multi-field galileons and higher co- gravity,” Class.Quant.Grav. 30, 015004 (2013), dimension branes,” Phys.Rev. D82, 124018 (2010), arXiv:1205.2158 [gr-qc]. arXiv:1008.1305 [hep-th]. [65] S.F. Hassan, Angnis Schmidt-May, and [79] Claudia de Rham, Gregory Gabadadze, Lavinia Heisen- Mikael von Strauss, “Particular Solutions berg, and David Pirtskhalava, “Nonrenormalization in Bimetric Theory and Their Implica- and naturalness in a class of scalar-tensor theories,” tions,” Int.J.Mod.Phys. D23, 1443002 (2014), Phys.Rev. D87, 085017 (2013), arXiv:1212.4128. arXiv:1407.2772 [hep-th]. [80] A.I. Vainshtein, “To the problem of nonvanishing gravi- [66] H. van Dam and M.J.G. Veltman, “Massive tation mass,” Phys.Lett. B39, 393–394 (1972). and massless Yang-Mills and gravitational fields,” [81] Eugeny Babichev and Cédric Deffayet, Nucl.Phys. B22, 397–411 (1970). “An introduction to the Vainshtein mecha- [67] V.I. Zakharov, “Linearized gravitation theory and the nism,” Class.Quant.Grav. 30, 184001 (2013), arXiv:1304.7240 [gr-qc].