Arxiv:2005.14571V2 [Gr-Qc] 5 Aug 2020 ∗∗ Nete Rm,Wieteei Nogigdbt Sto As Work Debate [ Ongoing Can Physical an One Is Is One There Scalar Gravity

arXiv:2005.14571v2 [gr-qc] 5 Aug 2020 ∗∗ nete rm,wieteei nogigdbt sto as work debate [ ongoing can physical an One is is one there scalar gravity. which while the to frame, where either coupled frame in minimally Einstein is the reparametriza- a to field field perform move scalar always and a matter and can tion metric to one the course, coupled of rescaling Of minimally frame). but non-minimally (Jordan gravity generally to is scalar-tensor inflaton coupled of the ob- class the the where to for theories belong predictions also same They the give servables. they seem- class why equivalence though is same the which even to belong theories, different, very these ingly Incidentally, [ Starobinsky gion. [ most inflation the namely Higgs of minimal Two models, essentially models. popular and inflationary space many parameter excluded available the strained elsaa ed hs nrydniydie h near- the drives density expansion. energy exponential whose fields more scalar or sup- one real by with action formulated Einstein-Hilbert usually the is structure plementing Inflation large-scale observe. subsequent we the that for generate be- seeds which to the inhomogeneities came able primordial is the preserve era ex- and Universe. inflationary Sitter the of this quasi-de moments Furthermore, early a very features postulating the These during by pansion explained scales homogeneous. and be large flat can at is that Universe indicate the (CMB) radiation ground ¶ ∗ § ‡ † [email protected] antonio.racioppi@kbfi.ee [email protected] [email protected] alexandros.karam@kbfi.ee [email protected] eety h lnkstliemsin[ mission satellite Planck the Recently, back- microwave cosmic the of observations Recent 3 nttt fTertclPyisUiest fWolw pl Wroclaw, of University Physics Theoretical of Institute n aaiifraim,adas ootieamto o os to how observabl same method the a produce outline that to formulations eq both yields also in model and a models formalisms, when conditions Palatini specific and the an elucidate transformations to conformal us under neat invariant be remain can observables which both inflationary encompasses that fact which the o employing importan principle depend classification becomes generally a – describe predictions Palatini observational or and metric dynamics – gravity of principle ihasaa ednnmnmlyculdt uvtr,teu the curvature, to coupled non-minimally field scalar a With 2 ainlIsiueo hmclPyisadBohsc,Rä Biophysics, and Physics Chemical of Institute National h ercadPltn omlto fsaa-esrtheor scalar-tensor of formulation Palatini and metric the I. 1 nttt fPyis nvriyo at,W swli1 50 1, Ostwaldi W. Tartu, of University Physics, of Institute 4 hsc eatet nvriyo onia R410Ioan GR–45110 Ioannina, of University Department, Physics Introduction 7 – neo Lykkas, Angelos 3 28 – arJärv,Laur 6 .T icmetteissue, the circumvent To ]. qiaec fifltoaymdl between models inflationary of Equivalence tl i ntealwdre- allowed the in lie still ] 1, ∗ 4, lxnrsKaram, Alexandros 2 n non- and ] § 1 a con- has ] noi Racioppi, Antonio in ewe h w aitoa rnilshsbeen has [ principles in variational studied predic- two recently ac- the the general in between more difference for the tions true inflation, longer Regarding no tions. is this but coupled, e st lrf h iuto n ootieamto of method a pa- outline to current and the situation of the same clarify aim the to The is to per observables. hence the and for potential values invariant same lead can the parametrizations to actions and fundamental formulations many different there as in degeneracy, down, a pinned in- remain effectively the will when be even can However, potential models. variant viable of narrow range considerably will the that precision high at servables rm-nain prahwsdvlpdi [ in developed was approach frame-invariant a IO[ PICO scheme pro- classification [ can the of formalisms extend two and the results when similar duce cases the on [ however focus we (see infla- concerned when Palatini is and results tion metric different the give that generally shown formulations was it studies vious an for in equation linear is field them. Lagrangian same of whose the both action to to respect lead with approaches action Both the vary to and variables has Pala- independent one the be the to and assumed [ in are formalism freedom connection However, order of first or Levi-Civita. degree tini the dynamical is only connection met- the the is formalism, metric ric employed the the In of principle. that variational is theories coupled non-minimally these of terms in then observables invariants. and inflationary reparametrization of the field rescaling express scalar Weyl and conformal metric re- the the that under quantities invariant define any main can from starting one that, theory, is scalar-tensor method this of advantage The [ inflation slow-roll to applied fruitfully then xeddt eae hoisadfruain [ formulations and theories related to extended nte su htaie hnoei neetdin interested is one when arises that issue Another uuesaemsin LTBR [ (LITEBIRD missions space Future 2, 33 † 2, aaBra9 026Wolw Poland Wroclaw, 50-206 9, Borna Maxa . oicuePltn models. Palatini include to ] yepesdi em fcranquantities certain of terms in expressed ly lkadrKozak, Aleksander ¶ ercadPltn oeso inflation, of models Palatini and metric 80 hscoc.I h rsn ae we paper present the In choice. this n clrfil eentos hsallows This redefinitions. field scalar d iaetpeoeooyi h metric the in phenomenology uivalent n agsSaal Margus and n ae ieec,a h field the as difference, a makes and t es. )poiet eemn h nainr ob- inflationary the determine to promise ]) aa1,113Tlin Estonia Tallinn, 10143 10, vala dryn emtyadvariational and geometry nderlying seaial osrc different construct ystematically 1 at,Estonia Tartu, 411 37 ia Greece nina, , 38 3, , 1, ‡ 42 ∗∗ 40 – , 77 49 41 .I oto h pre- the of most In ]. , ,temti n the and metric the ], ies 50 R ) nti article this In ]). n sminimally is and 78 ,PXE[ PIXIE ], 32 – 34 35 29 and ] – – 79 39 31 ], ]. ], 2

how to explore and reconstruct such equivalent actions We refer to the set of , , , σ as the model func- in a systematic way. In the end, some actions in a given tions. By considering a{A WeylB V rescaling} of metric (re- equivalence class would be better motivated from the ferred later as a change of frame) and scalar field redef- theoretical point of view, while the degeneracy could inition (referred later as a reparametrization) be also broken by some observations of noninflationary 2¯γ(Φ)¯ physics. gµν =e g¯µν , (4a) The paper is organized as follows. In the next sec- Φ= f¯(Φ)¯ , (4b) tion we adopt the approach of invariants to study general scalar-tensor theories in both metric and Palatini the action functional (1) preserves its structure (up to formalisms. In section III we focus on inflation and ex- the boundary term) if the functions , , and σ trans- press the slow-roll parameters and inflationary observ- form as [81] A B V ables in terms of the invariant potential and its deriva- ¯ tives. Then, in section IV we determine under which ¯(Φ)=e¯ 2¯γ(Φ) f¯(Φ)¯ , (5a) A A conditions the metric and Palatini formalisms can gen- 2¯γ(Φ)¯ 2 ¯(Φ)=e¯ f¯′ (f¯(Φ))¯ erate the same slow-roll parameters when one starts B B 2¯γ(Φ)¯ 2 from the same action and study some examples. Con- 6 δj Γe (¯γ ′) f¯(Φ)¯ γ¯ ′f¯′ ′ , versely, starting from the same invariant potential in − A − A (5b) section V we explore the reconstruction of the corre- h i ¯ sponding metric and Palatini actions. We summarize ¯(Φ)=e¯ 4¯γ(Φ) f¯(Φ)¯ , (5c) our results and conclude in section VI. Finally, we in- V V σ¯(Φ)¯ = σ f¯(Φ)¯ +¯γ(Φ)¯ , (5d) clude an appendix A where we illustrate how an additional independent (conformal) transformation of the where prime denotes a derivative with respect to the connection enlarges the general Palatini action, but a scalar field. The Jordan frame is defined by the con- suitable choice neutralizes the effect, a point that has dition σ(Φ) = 0. For what follows we omit the matter not received much attention in the literature so far. part of the action and take m = 0, since our interest is now on the scalar non-minimallyS coupled to gravity which will be identified with the inflaton field. II. Action and invariant quantities By a straightforward calculation it is possible to make sure, that in every spacetime point the numerical value Regardless the gravity formulation, the action for of the quantities [29] general scalar-tensor theory can be written as1 [81], e2σ(Φ) 1 1 m(Φ) = , (6) = d4x √ g (Φ)R (Φ) ( Φ)2 (Φ) I (Φ) S − 2A − 2B ∇ −V A Z (Φ) 2σ(Φ) (Φ) = V 2 , (7) + m[e gµν ,χm] , (1) IV ( (Φ)) S A 2 where we used Planck units MPl = 1 and metric signa- (Φ) 3 ′(Φ) ture ( , +, +, +). The Ricci scalar R = gµν R [Γ, ∂Γ] Φ(Φ) = dΦ B + δj Γ A (8) − µν I s (Φ) 2 (Φ) is a function of the metric tensor gµν and the connection Z A A Γ. The choice of the gravity formulation is reflected on remain invariant, i.e. ¯i(Φ)¯ = i(Φ). In a similar vein, the expression of Γ in eq. (1)[43]: we may introduce anI invariantI metricg ˆ = g , µν A µν λ λ which is unaffected by the conformal transformation Γαβ = αβ + (2) (4a). One can see that the invariant field Φ has a λ λ λ I (1 δj Γ) δα∂βω(Φ) + δβ ∂αω(Φ) gαβ∂ ω(Φ) , different dependence on the model functions when one − − g considers the metric (we use the notation Φ) or Palatini where formalism (denoted as Γ). Still, in bothI formalisms we IΦ ω (Φ) = ln (Φ) , (3) may take the quantity Φ as an invariant description A of the scalar degree of freedomI in the theory [29, 37]. λ p αβ is the Levi-Civita connection, δjk is the Kronecker Negative values for the expression under the square root delta{ } and j = g stands for the metric case while j =Γ in eq. (8) suggest that the scalar field is a ghost, while for the Palatini one. identically constant Φ indicates that the scalar is not dynamical. In the metricI formulation this occurs only 2 ′(Φ) 3 (A ) when (Φ) = 2 (Φ) , while in the Palatini for B − A 1 (Φ) = 0. In both cases the theory is equivalent to The most general Palatini action contains also additional terms B due to the non-metricity of the theory [37]. However, it is general relativity plus a cosmological constant. A mul- possible to show that the action can always be cast in the form tiscalar generalization of the integrand in eq. (8) plays of eq. (1). For the interested reader the concerning details are the role of the invariant volume form on the space of given in Appendix A. scalar fields, hence here has a natural interpretation IΦ 3

as an invariant “distance” in the 1-dimensional space of substitution into we express the PSRPs in terms of IV the scalar field [32, 35]. ( Φ). By inverting eq. (8) we may switch to use as the IVTheI tensor-to-scalar ratio r, the scalar spectral in- IΦ basic variable instead of Φ, and employing the invariant dex ns and the amplitude of the scalar power spectrum metricg ˆµν we can rewrite the action (1) in terms of As are some of the inflationary observable quantities invariant quantities only [29], posing strict constraints on the parameter space of the inflationary models. These are usually computed in the 4 1 µν 1 2 slow-roll approximation and, up to first order in PSRPs, ˆ = d x gˆ gˆ Rµν [Γ, gˆµν ] ( ˆ Φ) S − 2 − 2 ∇I − IV they read as follows [33, 34]: Z p + m[ m gˆµν , ψ] . (9) S I d ln 2 r =8 IV , (12) An arbitrary scalar-tensor theory with four free func- d IΦ tions (1) can therefore be cast by the two transforma- d ln 2 1 d2 V V tions of frame change and reparametrization (4) into ns =1 3 I +2 I2 , (13) − d Φ d the action (9) endowed by two functions that carry in- I IV IΦ 2 variant meaning. The quantity m( Φ) characterizes d ln − I I A = V V . (14) the coupling of gravity to matter fields. For constant s I 2 I 12π d Φ m the theory is equivalent to general relativity with a I minimallyI coupled scalar field, otherwise the scalar field Note that all of the above observables are calculated participates in mediating the gravitational interaction at horizon exit, Φ = ∗ . The number of e-foldings, I IΦ and the effective gravitational “constant” starts to vary characterizing the duration of inflation, is given by according to the scalar field value. The quantity ( Φ) ∗ is the invariant scalar potential. In the case of inflationIV I IΦ 1 d ( Φ) − where the matter fields can be neglected, the physics of N = ( Φ) IV I d Φ , (15) IV I d Φ I the model is encoded by the invariant potential alone Zend I [33]. The form of the invariant action (9) coincides with IΦ end the usual Einstein frame action, a circumstance which where Φ and Φ∗ are the field values at the end and will help us to write down the inflationary parameters start ofI inflation,I respectively. in terms of the invariants in the next section. The invariant formalism can be applied in a straightforward way to any model that can be recast in the form of eq. (9), by first identifying the model functions III. Slow-roll parameters and computational (Φ), (Φ), (Φ) and σ(Φ). This is under the implicit algorithm AassumptionB thatV the model under consideration includes only one dynamical scalar field Φ. As we explained pre- The action functional (9) can be identified as the Ein- viously, we may use (8) to compute the invariant quan- stein frame regarding theg ˆµν metric. Then the equa- tity (Φ) and invert that relation to obtain Φ( ). IΦ IΦ tions of motion coincide in both formulations of gravity, By using Φ( Φ) we can calculate the invariant poten- I end although in the Palatini formalism the Levi-Civita con- tial ( Φ) and then solve ǫ( Φ ) = 1 to obtain the IV I I nection is derived on-shell from its constraint equation field value at the end of inflation. The field value ∗ IΦ δ ˆ = 0. The invariant quantity Φ assumes the role is obtained by integrating (15) and assuming that the (Γ)S I of the inflaton field driving inflation, governed by its number of e-folds lies somewhere in the allowed region potential ( Φ). Assuming then the usual slow-roll of N (50 60) e-folds. Finally, the inflationary ob- IV I conditions, we can rewrite the potential slow-roll pa- servables≃ are− readily obtained from eqs. (12)-(15) using rameters (PSRPs) as [32–34] the field value Φ∗ . In the followingI sections, we apply this procedure 2 1 d ln in the study of the inflationary predictions for scalar- ǫ = IV , (10) 2 d tensor theories in the metric and Palatini formulations. IΦ 1 d2 η = IV . (11) d 2 IV IΦ IV. When do identical metric and Palatini actions yield (almost) the same observables? At this point, we assumed that the integral in eq. (8) is 2 solvable and the relation of Φ(Φ) invertible , so that I Comparing eqs. (7) and (8) we can see that the differ- we can obtain a relation of Φ( Φ). Then, after a direct I ence between the metric and Palatini formulation arise from a different definition of the invariant field value. Therefore, given the action in eq. (1) (i.e. a set of func- 2 The problem is still solvable also when IΦ(Φ) is not invertible. tions , and ), the metric and Palatini formulations In that case Φ is used as a new variable and the chain rule is usuallyA generateB V different invariant actions and there- applied in the computation of the derivatives. fore different predictions. However, it might happen 4

that the two formulations produce the same slow-roll where 0 is a constant of integration that does not carry parameters when the invariant potential and the invari- any physicalA meaning and can be used to conveniently ant field value possess certain properties. The slow-roll set the zero-value of the invariant field according to the parameters are independent of the overall normalization problem at hand. Imposing eq. (17) we obtain of the invariant potential, therefore it is enough to as- n sume that invariant potential in the two formulations, (Φ) (Φ) . ln A V 2 (25) as functions of the corresponding invariant field values, 0 ∝ (Φ) are proportional to each other: A A Therefore, the metric and Palatini formulations produce g Γ . (16) the same slow-roll parameters when IV ∝ IV Unfortunately, we cannot provide a general criterion 2 (Φ) (Φ) ( ′(Φ)) , (26) that is more explicit than eq. (16), because eq. (8) con- A B ∝ A (Φ) n tains an integral over Φ and the corresponding solving (Φ) (Φ)2 ln A . (27) technique is strongly dependent on the actual definition V ∝ A 0 A of and . On the other hand, we can provide a couple ofA explicitB examples: one relatively simple (A) and one From the two equations, we can immediately see that more complicated (B). the following class of non-minimal Coleman-Weinberg models where3

(Φ) = ξΦ2 , (28) A. Example: Power law invariant potential A (Φ) = 1 , (29) B Given eqs. (7) and (8), the simplest way to satisfy Φ n (Φ) = β ln Φ4 , (30) eq. (16) is by requiring V Φ 0 n Φ , (17) IV ∝ I satisfies the conditions (26) and (27), and therefore gen- g Γ erates slow-roll parameters that cannot discriminate be- Φ Φ , (18) I ∝ I tween metric and Palatini gravity. These results are in where n is some nonzero power. The class of models agreement with the findings of [49] and the strong cou- (17) is well-known (e.g. [1]) and goes under the name pling limit of [50]. of monomial inflation. Using eqs. (10), (11) and (17) Moreover, eqs. (26) and (27) can also be used to back- we see that the corresponding slow-roll parameters are engineer models. For instance, choosing n = 1 and a natural inflation potential n2 ǫ = , (19) 2 2 Φ IΦ (Φ) = M 4 1 cos , (31) V − Φ n(n 1) 0 η = −2 . (20) IΦ the addition of the following non-trivial non-minimal We can appreciate that the case n = 2 is even more spe- coupling to gravity and non-canonical kinetic function cial, since it accidentally implies also that ǫ = η. With z a couple of straightforward computations, we can eas- (Φ) = , (32) A W (z) ily verify that the tensor-to-scalar ratio and the scalar r spectral index are: 2 Φ 3/2 sin Φ W (z) (Φ) = 0 , (33) 16n 3/2 2 r = . (21) B 4z (W (z)+1) n +4N where W (z) is the Lambert W -function and n z =1 cos Φ , would generate 2( + 2) Φ0 ns =1 , (22) − − n +4N Φ , (34) Furthermore, the combination of eqs. (8) and (18) IV ∝ I 1 implies ǫ = 2 , (35) 2 Φ 2 I (Φ) ′(Φ) η =0 , (36) B A , (23) (Φ) ∝ (Φ) A A therefore

3 ′(Φ) (Φ) Without loss of generality we assume Φ > 0 in Section IV and Γ,g(Φ) dΦ A = ln A , (24) IΦ ∝ (Φ) V. Z 0 A A

regardless of the adopted gravity formulation. There- Using (44), the scalar spectral index and the tensor-to- fore, for n = 1, the scalar spectral index and the tensor- scalar ratio can be expressed by: to-scalar ratio have the following values: 1 4N ′ 4N ′ − ns =0.9701, r =0.0796, for N = 50, nW e 1+ n + ln 4N ′W e (37) ns =1 , ns =0.9751, r =0.0664, for N = 60, 1 2 − 4N ′ N ′ ln 4N ′W − which are out of the 2σ Planck boundaries [1], but still e (46) allowed at 3σ. Therefore, if the universe happened to be described by a non-minimal scalar ﬁeld with model n2W 4N ′ functions (31), (32), (33) in action (1), the slow-roll e r = . (47) parameters would not be able to distinguish whether 1 2 N N W 4N ′ − the underlying theory is metric or Palatini in character. 2 ′ ln 4 ′ e An example to illustrate this possibility can be real- B. Example: Logarithmic invariant potential ized by the model functions

(Φ) = exp 1 acosh(2Φ) 1 Φ 4Φ2 1 , (48) A more complicated way to satisfy eq. (16) is the A 2√6 − √6 − following choice: (Φ) = exp 1 acosh(2Φ) 1 Φp4Φ2 1 , (49) n 2√6 √6 ln( m) , (38) B − − Φ 2 IV ∝ I (Φ) = (Φ)2 ln Φ2 . p (50) g ( Γ)l , (39) V A IΦ ∝ IΦ where Φ is a subscript while l,m,n are some powers. In this case, by integrating eq. (8) one obtains the in- The class of models (38) can be interpreted as a cosmo- variant field in the two formalisms as logical constant subject to quantum corrections. Be- Γ =Φ , (51) ing somehow new (only the case n = 1 is well-known Φ Ig (e.g. [1])), such a class deserves a deeper investigation =Φ2 . (52) IΦ than the previous example. Despite the nonlinear relation between the invariant fields in the two formalisms While the invariant potentials differ by a constant fac- in eq. (39), the expressions of the slow-roll parameters tor, coming from eq. (38) are 2 Γ Γ 2 =4 ln Φ , (53) n IV I ǫ = , (40) 2 2 2 g g 2 Φ ln Φ = ln Φ , (54) I I IV I n n 1 ln Φ η = − − I , (41) as expected, the slow-roll parameters coincide: 2 2 Φ ln Φ 4 I I ǫ = , (55) where the power m canceled out because of properties 2 2 2 ln Φ of the logarithm. Inflation ends when IΦ I 1 2 2 ln Φ end n n − η = − I (56) Φ = W , (42) 2 2 I √2 √2 Φ ln Φ I I while the number of e-folds turns out to be: and therefore yield the same ns and r (as functions of 2 2 ∗ 1 Φ Φ IΦ Φ) in both metric and Palatini formulations (cf. eqs. N = I + I ln Φ . (43) I end (38), (55) and (56) for n = 2). In this case the scalar n − 4 2 I Φ h iI spectral index and the tensor-to-scalar ratio take the Therefore the value of the invariant field at the horizon following values: crossing is: ns =0.9699, r =0.0556, for N = 50, 1 (57) 4N ′ − ∗ = 4N W , (44) ns =0.9752, r =0.0450, for N = 60, IΦ ′ e s which are within the 2σ Planck boundaries [1]. where we have defined: The model (48)-(50) looks rather contrived, but it 2 n2 n − employs a parametrization where the calculational logic N ′ = nN W − 8 √2 is easy to see. However, hidden somewhere in the infi- 2 1 (45) nite possibilities of reparametrizations there might exist n2 n − n n − + W ln W . a physically better motivated form of the same model, 4 √ √ √ 2 2 2 ! but where the calculations become harder to deal with. 6

It is not easy to guess what a nicer parametrization plings in eq. (28) satisfy the following condition: could be, but as an extra illustration let us just perform a simple scalar field redefinition ξg ξΓ = , (64) 1+6ξg 1+ Φ¯ 2 Φ= . (58) 4Φ¯ where ξg,Γ are respectively the non-minimal coupling under the metric and the Palatini formulation [49, 50]. The model functions (48)-(50) transform under eq. (58) A more general discussion about the generation of ex- into actly equivalent invariant potentials is presented in the next section. 1 √6(Φ¯ 4 1) − ¯(Φ)¯ = Φ¯ 2√6 e− 48Φ¯ 2 , (59) A 1 2 2 √6(Φ¯ 4 1) (1 Φ¯ ) − ¯ ¯ ¯ 2√6 48Φ¯ 2 V. When do different metric and Palatini actions (Φ) = Φ − ¯ 4 e− , (60) B 16Φ yield the same observables? 1 √6(Φ¯ 4 1) 2 2 2 − (1+Φ¯ ) ¯(Φ)¯ = Φ¯ √6 e− 24Φ¯ 2 ln , (61) V 16Φ¯ 2 As described in [33], equivalent inflationary theo- 4 and contrary to the previous form the functions ¯(Φ)¯ ries are described by one invariant function : ( Φ). IV I and ¯(Φ)¯ do not coincide any more. A direct integrationA However, inflationary models can be produced by us- B ing three generating functions: (Φ), (Φ) and (Φ). of eq. (8) yields the expressions of the invariant field now A B V as Therefore, the a priori knowledge of ( Φ) allows us to derive only one constraint that (Φ),IV I(Φ) and (Φ) 1+ Φ¯ 2 have to satisfy, leaving two functionsA outB of the threeV Γ = , (62) IΦ¯ 4Φ¯ completely undetermined. Generally, we can express 1+ Φ¯ 4 1 the invariant field Φ as the inverse function of the in- g 5 I ¯ = + , (63) variant potential in eq. (7) IΦ 16Φ¯ 2 8 1 where the last term had to be added as an integra- (Φ) − 1 , tion constant to maintain an explicit equivalence. By Φ = V 2 − (Φ) (65) I (Φ) ≡ IV construction, we get the same invariant potentials (53), A ! (54), and PSRPs (55), (56). Note that if we had omit- where the superscript “ 1” stands for inverse function. ted the constant of integration in (63), the proportion- Using eq. (8) we can write− ality of invariant fields (39) would still hold, but the proportionality of invariant potentials (38) would not be completely obvious at first sight. Nevertheless, a 2 direct calculation of the derivatives in the inflationary 1 (Φ) 3 ′(Φ) − (Φ) = dΦ B + δj Γ A , (66) parameters (12), (13) would yield the same result with IV s (Φ) 2 (Φ) Z A A or without the integration constant. As a final comment, let us stress that in all the exam- where the parameter δj Γ indicates the adopted grav- ples of this section the invariant PSRPs, and therefore ity formulation. Therefore, given the invariant function r n N ( Φ), eq. (66) is the constraint that (Φ), (Φ) and the predictions for , s and , coincide in the met- IV I A B ric and Palatini cases, since the invariant potentials are proportional to each other and the overall factor cancels out. However, the amplitude of the scalar power spec- 4 trum (14) depends explicitly on the invariant potential, As already discussed in [33], the full gravitation equivalence I and thus this observable will be sensitive to the differ- needs to take into account also the invariant m, that describes the couplings to matter. Therefore, ensuring only the same in- ence in the actual normalization of the invariant po- variant potential, the (p)reheating mechanism might still affect tentials. The normalization can be crucial in satisfying the value of the observables and break the equivalence [49, 82]. 9 the observational constraints, currently As 2.1 10− On the other hand, we can see from (6) that, by adjusting ac- [1]. The metric and Palatini models will yield≃ the× same cordingly the function σ, we can easily obtain theories where I phenomenology also in this respect if a strict equiva- m is the same under both gravity formulations, restoring the equivalence of observables also when (p)reheating is considered. lence between the invariant potentials holds, not just a 5 I−1 The computation of V is quite delicate. In many cases proportionality (16). Starting from exactly the same in- IV (IΦ) is not a bijective (i.e. invertible) function, therefore I−1 variant actions, this is never the case. However, for the V can be consistently identified only after a proper defini- I I examples considered before, a change in the normalization of the domain of V ( Φ). For more details about this tion of the model functions of the metric and Palatini topic see [83], and also [30, 84]. This is related to the possible occurrence of a singularity or exceptional point in the theory action will have the final effect of generating exactly the which could have implications for cosmology or in the presence same invariant potentials. For instance, for what con- of black holes (see e.g. [85, 86]). Nevertheless, here we restrict cerns Example A, the invariant potential under metric our attention to slow-roll inflation and smooth functions, ex- and Palatini are the same when the non-minimal cou- pecting no singularities to arise. 7

4 (Φ) must satisfy in order to create equivalent inflation- η − , (72) V 2 ≈ √ α Φ ary theories also among different gravity formulations. 3αe 3 I This means that (apart from pathological cases) we can while the number of e-folds is randomly choose two functions among (Φ), (Φ) and A B 3α 2 (Φ). If the third one satisfies eq. (66), then the cor- √ α ∗ V N e 3 IΦ . (73) rect ( Φ) is always generated. However, the solution ≈ 4 IV I of the constraint (66) is strongly dependent on the ini- Therefore, at the leading order in N we get that tial choice of model functions, invariant potential and gravity formulation (metric or Palatini). Nevertheless, 2 ns 1 , (74) until the constraint (66) is satisfied, the same invariant ≈ − N potential ( Φ) and therefore the same inflationary 12α IV I r . (75) observables (eqs. (12)-(15)) are generated, regardless of ≈ N 2 initial model functions and gravity formulation. Planck data sets log10 α< 1.3 at 95% CL [1]. When (Φ) and (Φ) are given, it is always possible We can invert eq. (70) to obtain to solve eq.A (66) andV obtain the corresponding value for the non-canonical kinetic function 3α 1 (Φ) 1 . − (Φ) = ln 1 2 V 2 (76) 1 2 2 IV − 2 − M s (Φ) d − (Φ) 3 ′(Φ) r A ! (Φ) = (Φ) IV δ A , B A dΦ − 2 j Γ (Φ) ( A ) Therefore, the constraint in eq. (66) becomes (67) where δj Γ reflects the adopted gravity formulation (see 2 (Φ) 3 ′(Φ) eq. (5)). dΦ B + δj Γ A s (Φ) 2 (Φ) Instead, if (Φ) and (Φ) are fixed, the constraint Z A A A B can be formally solved as 3α 1 (Φ) . = ln 1 2 V 2 (77) 2 1 − 2 − M s (Φ) (Φ) = (Φ) − (Φ) (68) r A ! V A IV IV 1 Let us consider now some case by case examples and where − (Φ) is given in eq. (66). However, in this see how the initial choice of the model functions affects IV case, since the integral of an elementary function is not the solving strategy of eq. (77). automatically elementary, choosing (Φ) and (Φ) as elementary functions of Φ does not alwaysA ensureB that (Φ) is elementary as well. 1. A and V are fixed V Finally, when (Φ) and (Φ) are chosen, the con- B V straint eq. (66) becomes the following differential equa- Taking for instance the following natural inflation po- tion: tential and non-minimal coupling to gravity 2 1 2 2 3 ′(Φ) d − (Φ) (Φ) = 1+ ξΦ (78) (Φ) δj Γ A IV + (Φ) = 0 , A A 2 (Φ) − dΦ B Φ ( A ) (Φ) = M 4 1 cos (79) (69) V − Φ0 to be solved in order to determine (Φ). we obtain the invariant potential in eq. (70) if Next, we present an example in orderA to better illus-

trate the different issues arising in each configuration. Φ Φ 2 (Φ) cos 4ξΦ0Φ sin (Φ) = 3α A 2Φ0 − 2Φ0 Γ 4 (Φ)Φ2 √ Φ B 0 (Φ) 2 sin 2Φ A A − 0 (80) A. Example: Generalized Starobinsky invariant potential in the Palatini case, with (Φ) given in eq. (78) and A ξ2Φ2 Let us consider the following invariant potential: (Φ)g = (Φ)Γ 6 (81) B B − (Φ) A 2 2 4 √ α Φ ( Φ)= M 1 e− 3 I , (70) in the metric case. IV I − which generalizes the Starobinsky potential [87–89]. 2. A and B are fixed The model is well-known too, therefore we just summarize briefly the main features. At the leading order in the invariant field value the slow-roll parameters are We consider now a non-minimally coupled scalar field with a canonical kinetic term 4 ǫ , (71) 2 2 2 (Φ) = Φ , (82) ≈ 2√ α Φ 3αe 3 I A 3α 8

(Φ) = 1 . (83) VI. Summary and conclusions B Solving eq. (66), we can see that we reproduce the invariant potential in eq. (70) if the potential is In the present paper we studied the slow-roll parameters and inflationary observables in the framework of 4M 4 Φ 2 scalar-tensor theories of gravity in the metric and Pala- (Φ) = 1 0 Φ4 (84) V Γ 9α2 − Φ tini formulations. The model functions (Φ), (Φ), (Φ) allow us to construct quantities, whichA are invari-B in the Palatini case and antV under a conformal transformation of the metric and behave as scalar functions under the scalar field redefini- 4+α 2 4 √ α tion. Using this frame invariant approach we expressed 4M Φ0 4 (Φ)g = 1 Φ (85) the slow-roll parameters ǫ, η, as well as the inflationary V 9α2  − Φ  observable quantities ns, r, As, and explained in de-   tail how to compute them in the case of different model in the metric case, where Φ0 is an integration constant. functions. Next, in the main part of the paper, we clarified what conditions must be met for the metric and Palatini for- 3. B and V are fixed malisms to give the same observable quantities. Due to the fact that most of the observable quantities are in- In this last example we consider a non-canonically dependent of the overall normalization factor, we con- normalized scalar and a quartic potential cluded that it is sufficient for the invariant potentials in 6(α 1)Φ2 both formulations to be proportional to each other in or- (Φ) = − , (86) der to obtain equal predictions for r, n and N (but not B (Φ) s A As) in both formulations. We illustrated this general (Φ) = M 4 Φ4 , (87) V statement by two specific examples. After that, start- with α> 1. We need to determine (Φ) by solving the ing from the same invariant potential, we showed how A 1 by fixing two out of the three model functions we can differential equation in eq. (69), where − , (Φ) and (Φ) are given respectively in eqs. (76),I (V86)B and (87). straightforwardly obtain the third. We demonstrated TheV specific choice in eq. (86) allows us to solve such the different possibilities by considering as an example differential equation in both the metric and the Pala- an invariant potential of the Starobinsky form. One tini cases. With a convenient choice of the integration then sees how seemingly different models of inflation constants, the corresponding solution is can give the same values of the observed parameters. A deeper case-by-case study may unveil other con- 2 (Φ)g =1+Φ (88) figurations where the same model functions, but with A different values of the free parameters, share the same in the metric case and invariant potential and therefore give the same values α 1 for observables. The framework described here pro- 2 2√ α− (Φ) =Φ 1+ Φ− (89) A Γ vides a tool that enables to easily check different models against observations, as well as to reconstruct variations in the Palatini case. The special value of α = 1 re- of models with a given phenomenology. quires an additional comment. In this case the potential If the next generation satellites (LITEBIRD [78], (70) becomes exactly the Starobinsky one and the non- PIXIE [79], PICO [80]) will be launched and after data canonical kinetic term (86) becomes identically zero in will be collected, the available parameter space will be both the metric and the Palatini formulations. For the even more constrained, leaving us with a reduced set of first case this not a problem because it coincides with allowed invariant potentials and more indications about the formulation of the Starobinsky model via the aux- which gravity formulation satisfies additional criteria iliary field in the Jordan frame. On the other hand, as like elegance, simplicity or minimality. discussed in Section II, in the Palatini formulation the invariant field Φ is not dynamical and the problem does not have a solution.I However, it is still possible to re- Acknowledgments produce the potential (70) from (87) in the Palatini formulation by relaxing the condition (86). For instance, choosing This work was supported by the Estonian Research Council grants MOBJD381, MOBTT5, MOBTT86, αΦ2 PRG356 and by the EU through the European Re- (Φ) = , (90) B Γ (Φ) gional Development Fund CoE program TK133 “The A Dark Side of the Universe.” A. Kozak was a beneficiary we would get of the ”International scholarship exchange of PhD can- didates and academic staff” programme (PROM) or- 2 √ 2 (Φ) =Φ 1+Φ− 3 . (91) A Γ ganized by Polish National Agency for Academic Ex- 9

change (NAWA). The research of A. Lykkas is co- and Lifelong Learning” in the context of the project ﬁnanced by Greece and the European Union (Euro- “Strengthening Human Resources Research Potential pean Social Fund - ESF) through the Operational Pro- via Doctorate Research” (MIS-5000432), implemented gramme “Human Resources Development, Education by the State Scholarships Foundation (IKY).

A. Appendix

The most general action for a class of Palatini scalar-tensor theories of gravity featuring non-metricity vectors entering the action functional in a linear way can be written as follows [37, 38]:

1 1 = d4x √ g (Φ)R(g, Γ) (Φ)( Φ)2 (Φ) (Φ)Qµ Φ (Φ)Q¯µ Φ + [e2σ(Φ)g ,χ ]. S − 2A − 2B ∇ −V −C1 ∇µ −C2 ∇µ Sm µν m Z (A1) The action contains three independent variables: metric tensor, affine connection, and scalar field. It also features six arbitrary functions of the scalar field: , , 1, 2, , σ , providing, together with the dynamical variables, the so-called “frame” for the action (A1). The{A vectorsB C QC µ Vand}Q¯µ are defined as

Qµ = gµν gαβ Γg = gµν gαβQ , (A2a) ∇ν αβ ναβ Q¯µ = gµν gαβ Γg = gµν gαβQ . (A2b) − ∇α νβ − ανβ The Γ is deﬁned with respect to the independent connection, therefore the covariant derivative of the metric will not vanish∇ in general. In the Palatini approach, the metric tensor is fundamentally independent of the connection. When we use the Weyl (or conformal) transformation of the metric, the connection remains unchanged. We might use this freedom and postulate additional transformations of the connection preserving the light cones. We introduce the following transformation formulae for the dynamical variables entering the action functional:

2¯γ1(Φ)¯ gµν =e g¯µν , (A3a) Γα = Γ¯α +2δα ∂ γ¯ (Φ)¯ g¯ g¯αβ∂ γ¯ (Φ)¯ , (A3b) µν µν (µ ν) 2 − µν β 3 Φ= f¯(Φ)¯ . (A3c)

The transformations are governed by three smooth functions of the scalar field (γ1,γ2,γ3), and are accompanied by a redefinition of the scalar field. The transformations (A3a))-(A3c)) are invertible

2γ1(Φ) g¯µν =e gµν , (A4a) Γ¯α =Γα +2δα ∂ γ (Φ) g gαβ∂ γ (Φ), (A4b) µν µν (µ ν) 2 − µν β 3 Φ=¯ f(Φ), (A4c)

and the relations between the gamma functions and the diﬀeomorphism of the scalar ﬁeld are given by:

γ¯ = γ f, (A5a) i − i ◦ 1 f¯ = f − . (A5b)

The action (A1) turns out to be form-invariant under the action of transformations (A3a)-(A3c), which means that solutions to the ﬁeld equations obtained in one frame are mapped into corresponding solutions in another frame, assuming that the six functions of the scalar ﬁeld , , , , , α change in the following way: {A B C1 C2 V } ¯ ¯(Φ)=e¯ 2¯γ1(Φ) (f¯(Φ))¯ , (A6a) A A 2¯γ (Φ)¯ 2 ¯(Φ)=e¯ 1 (f¯(Φ))(¯ f¯′(Φ))¯ B "B

+ f¯′(Φ)¯ (f¯(Φ))(8¯¯ γ′ (Φ)¯ 10¯γ′ (Φ)+2¯¯ γ′ (Φ))¯ (f¯(Φ))(2¯¯ γ′ (Φ)¯ 7¯γ′ (Φ)+5¯¯ γ′ (Φ))¯ C1 1 − 2 3 −C2 1 − 2 3 2 2 +3 4 (f¯(Φ))¯¯ γ′ (Φ)¯¯ γ′ (Φ)¯ (f¯(Φ))¯ γ¯′ (Φ)¯ (f¯(Φ))¯ γ¯′ (Φ)¯ A 2 3 − A 2 − A 3 10

d (f¯(Φ))¯ γ ¯ γ ¯ f¯ ¯ γ ¯ γ ¯ γ ¯ , + A ¯ (¯2′ (Φ)+¯3′ (Φ)) 2 ( (Φ))¯1′ (Φ)(¯2′ (Φ)+¯3′ (Φ)) (A6b) fΦ − A # 2¯γ (Φ)¯ 3 1 ¯ (Φ)=e¯ 1 f¯′(Φ)¯ (f¯(Φ))¯ (f¯(Φ))¯ γ¯′ (Φ)¯ + γ¯′ (Φ)¯ , (A6c) C1 C1 − A 2 2 2 3 h i 2¯γ (Φ)¯ ¯ (Φ)=e¯ 1 f¯′(Φ)¯ (f¯(Φ))¯ (f¯(Φ))¯ 3¯γ′ (Φ)¯ γ¯′ (Φ)¯ , (A6d) C2 C2 − A 2 − 3 ¯ ¯(Φ)=e¯ 4¯γ1(Φ)h (f¯(Φ))¯ , i (A6e) V V σ¯(Φ)¯ = σ(f¯(Φ))+¯ γ ¯1(Φ)¯ . (A6f)

It is always possible to choose the functions (γ2,γ3) in such a way that the functions 1 and 2 vanish. Indeed, one must take C C

2 1(Φ) 2(Φ) γ¯′ (Φ) = − C −C , (A7a) 2 6 (Φ) A 2 1(Φ) + 2(Φ) γ¯′ (Φ) = − C C . (A7b) 3 2 (Φ) A Such a choice will transform the action (A1) to the following one:

1 1 = d4x √ g (Φ)R(g, Γ)¯ ¯(Φ)( Φ)2 (Φ) + [e2σ(Φ)g ,χ ], (A8) S − 2A − 2B ∇ −V Sm µν m Z n o where

2 2 ′(Φ) 2(Φ) 4 1(Φ) 11 (Φ) 4 (Φ) 16 (Φ) (Φ) ¯(Φ) = (Φ) + A C − C + C2 − C1 − C1 C2 , (A9) B B (Φ) 6 (Φ) A A which justiﬁes the choice of the initial action (1) without the functions. Ci

[1] Y. Akrami et al. (Planck), “Planck 2018 re- arXiv:1108.5857 [gr-qc]. sults. X. Constraints on inflation,” (2018), [8] Takeshi Chiba and Masahide Yamaguchi, “Conformal- arXiv:1807.06211 [astro-ph.CO]. Frame (In)dependence of Cosmological Observations [2] Alexei A. Starobinsky, “A New Type of Isotropic in Scalar-Tensor Theory,” JCAP 1310, 040 (2013), Cosmological Models Without Singularity,” arXiv:1308.1142 [gr-qc]. Phys. Lett. 91B, 99–102 (1980). [9] Alexander Yu. Kamenshchik and Chris- [3] Fedor L. Bezrukov and Mikhail Shaposhnikov, tian F. Steinwachs, “Question of quantum “The Standard Model Higgs boson as the equivalence between Jordan frame and Ein- inflaton,” Phys. Lett. B659, 703–706 (2008), stein frame,” Phys. Rev. D 91, 084033 (2015), arXiv:0710.3755 [hep-th]. arXiv:1408.5769 [gr-qc]. [4] Fedor L. Bezrukov, Amaury Magnin, and Mikhail [10] Guillem Domènech and Misao Sasaki, “Con- Shaposhnikov, “Standard Model Higgs boson mass formal Frame Dependence of Inflation,” from inflation,” Phys. Lett. B675, 88–92 (2009), JCAP 1504, 022 (2015), arXiv:1501.07699 [gr-qc]. arXiv:0812.4950 [hep-ph]. [11] Mario Herrero-Valea, “Anomalies, equiv- [5] F. Bezrukov and M. Shaposhnikov, “Standard Model alence and renormalization of cosmologi- Higgs boson mass from inflation: Two loop analysis,” cal frames,” Phys. Rev. D93, 105038 (2016), JHEP 07, 089 (2009), arXiv:0904.1537 [hep-ph]. arXiv:1602.06962 [hep-th]. [6] F. Bezrukov, A. Magnin, M. Shaposhnikov, [12] Daniel Burns, Sotirios Karamitsos, and Apos- and S. Sibiryakov, “Higgs inflation: consis- tolos Pilaftsis, “Frame-Covariant Formula- tency and generalisations,” JHEP 01, 016 (2011), tion of Inflation in Scalar-Curvature The- arXiv:1008.5157 [hep-ph]. ories,” Nucl. Phys. B907, 785–819 (2016), [7] Israel Quiros, Ricardo Garcia-Salcedo, Jose Edgar arXiv:1603.03730 [hep-ph]. Madriz Aguilar, and Tonatiuh Matos, “The con- [13] D. J. Brooker, S. D. Odintsov, and R. P. formal transformation’s controversy: what are Woodard, “Precision predictions for the pri- we missing?” Gen. Rel. Grav. 45, 489–518 (2013), mordial power spectra from f(R) models of 11

inflation,” Nucl. Phys. B911, 318–337 (2016), [29] Laur Järv, Piret Kuusk, Margus Saal, and Ott Vil- arXiv:1606.05879 [gr-qc]. son, “Invariant quantities in the scalar-tensor theories [14] Krishnakanta Bhattacharya and Bibhas Ranjan Ma- of gravitation,” Physical Review D 91, 024041 (2015), jhi, “Fresh look at the scalar-tensor theory of grav- arXiv:1411.1947 [gr-qc]. ity in Jordan and Einstein frames from undis- [30] Laur Järv, Piret Kuusk, Margus Saal, and Ott cussed standpoints,” Phys. Rev. D95, 064026 (2017), Vilson, “Transformation properties and gen- arXiv:1702.07166 [gr-qc]. eral relativity regime in scalar–tensor theories,” [15] Sachin Pandey and Narayan Banerjee, “Equiv- Classical and Quantum Gravity 32, 235013 (2015), alence of Jordan and Einstein frames at the arXiv:1504.02686 [gr-qc]. quantum level,” Eur. Phys. J. Plus 132, 107 (2017), [31] Laur Jarv, Piret Kuusk, Margus Saal, and Ott arXiv:1610.00584 [gr-qc]. Vilson, “The formalism of invariants in scalar-tensor [16] Sebastian Bahamonde, Sergei D. Odintsov, V. K. and multiscalar-tensor theories of gravitation,” in Oikonomou, and Petr V. Tretyakov, “Decelera- 14th Marcel Grossmann Meeting on Recent Developments in Theoretical and Experimental General Relativity, Astrophysics, and Relativistic Field Theories (MG14) Rome, Italy, July 12-18, 2015 tion versus acceleration universe in different frames (2015) arXiv:1512.09166 [gr-qc]. of F (R) gravity,” Phys. Lett. B766, 225–230 (2017), [32] Piret Kuusk, Mihkel Rünkla, Margus Saal, and Ott arXiv:1701.02381 [gr-qc]. Vilson, “Invariant slow-roll parameters in scalar–tensor [17] Sotirios Karamitsos and Apostolos Pilaftsis, theories,” Class. Quant. Grav. 33, 195008 (2016), “Frame Covariant Nonminimal Multifield In- arXiv:1605.07033 [gr-qc]. flation,” Nucl. Phys. B927, 219–254 (2017), [33] Laur Järv, Kristjan Kannike, Luca Marzola, An- arXiv:1706.07011 [hep-ph]. tonio Racioppi, Martti Raidal, Mihkel Rünkla, [18] Michael S. Ruf and Christian F. Steinwachs, Margus Saal, and Hardi Veermäe, “Frame- “Quantum equivalence of f(R) gravity and scalar- Independent Classification of Single-Field Infla- tensor theories,” Phys. Rev. D97, 044050 (2018), tionary Models,” Phys. Rev. Lett. 118, 151302 (2017), arXiv:1711.07486 [gr-qc]. arXiv:1612.06863 [hep-ph]. [19] Sotirios Karamitsos and Apostolos Pilaftsis, “On the [34] Alexandros Karam, Thomas Pappas, and Cosmological Frame Problem,” Proceedings, 17th Hel- Kyriakos Tamvakis, “Frame-dependence of lenic School and Workshops on Elementary Particle higher-order inflationary observables in scalar- Physics and Gravity (CORFU2017): Corfu, Greece, tensor theories,” Phys. Rev. D96, 064036 (2017), September 2-28, 2017, PoS CORFU2017, 036 (2018), arXiv:1707.00984 [gr-qc]. arXiv:1801.07151 [hep-th]. [35] Piret Kuusk, Laur Järv, and Ott Vilson, “Invari- [20] Krishnakanta Bhattacharya, Ashmita Das, and ant quantities in the multiscalar-tensor theories of Bibhas Ranjan Majhi, “Noether and Abbott- gravitation,” Int. J. Mod. Phys. A31, 1641003 (2016), Deser-Tekin conserved quantities in scalar-tensor arXiv:1509.02903 [gr-qc]. theory of gravity both in Jordan and Ein- [36] Alexandros Karam, Angelos Lykkas, and stein frames,” Phys. Rev. D97, 124013 (2018), Kyriakos Tamvakis, “Frame-invariant ap- arXiv:1803.03771 [gr-qc]. proach to higher-dimensional scalar-tensor [21] Israel Quiros and Roberto De Arcia, “On local gravity,” Phys. Rev. D97, 124036 (2018), scale invariance and the questionable theoretical ba- arXiv:1803.04960 [gr-qc]. sis of the conformal transformations’ issue,” (2018), [37] Aleksander Kozak and Andrzej Borowiec, arXiv:1811.02458 [gr-qc]. “Palatini frames in scalar–tensor theories [22] Kevin Falls and Mario Herrero-Valea, “Frame of gravity,” Eur. Phys. J. C79, 335 (2019), (In)equivalence in Quantum Field Theory and arXiv:1808.05598 [hep-th]. Cosmology,” (2018), arXiv:1812.08187 [hep-th]. [38] Andrzej Borowiec and Aleksander Kozak, “New class [23] Saikat Chakraborty, Sanchari Pal, and Alberto Saa, of hybrid metric-Palatini scalar-tensor theories of grav- “Dynamical equivalence of f(R) gravity in Jordan ity,” (2020), arXiv:2003.02741 [gr-qc]. and Einstein frames,” Phys. Rev. D99, 024020 (2019), [39] Manuel Hohmann, “Scalar-torsion theories of grav- arXiv:1812.01694 [gr-qc]. ity III: analogue of scalar-tensor gravity and con- [24] Israel Quiros, “Selected topics in scalar-tensor theories formal invariants,” Phys. Rev. D98, 064004 (2018), and beyond,” (2019), 10.1142/S021827181930012X, arXiv:1801.06531 [gr-qc]. arXiv:1901.08690 [gr-qc]. [40] Attilio Palatini, “Deduzione invariantiva delle [25] Khamphee Karwan and Phongpichit Channuie, equazioni gravitazionali dal principio di hamilton,” “Generalized Conformal Transformation and Infla- Rendiconti del Circolo Matematico di Palermo (1884-1940) 43, 203–212 (1919). tionary Attractors,” Phys. Rev. D 100, 023514 (2019), [41] M. Ferraris, M. Francaviglia, and C. Reina, arXiv:1811.03006 [gr-qc]. “Variational formulation of general rel- [26] Debottam Nandi, “Note on stability in conformally ativity from 1915 to 1925 “palatini’s connected frames,” Phys. Rev. D 99, 103532 (2019), method” discovered by einstein in 1925,” arXiv:1904.00153 [gr-qc]. General Relativity and Gravitation 14, 243–254 (1982). [27] Yohei Ema, “Dynamical Emergence of Scalaron [42] Q. Exirifard and M. M. Sheikh-Jabbari, “Love- in Higgs Inflation,” JCAP 09, 027 (2019), lock gravity at the crossroads of Palatini and met- arXiv:1907.00993 [hep-ph]. ric formulations,” Phys. Lett. B661, 158–161 (2008), [28] G.G.L. Nashed, W. El Hanafy, S.D. Odintsov, and arXiv:0705.1879 [hep-th]. V.K. Oikonomou, “Thermodynamical correspondence [43] Florian Bauer and Durmus A. Demir, “Inflation of f(R) gravity in Jordan and Einstein frames,” (2019), with Non-Minimal Coupling: Metric versus Pala- arXiv:1912.03897 [gr-qc]. tini Formulations,” Phys. Lett. B665, 222–226 (2008), 12

arXiv:0803.2664 [hep-ph]. no massless scalar,” Phys. Rev. D 99, 124018 (2019), [44] Florian Bauer, “Filtering out the cosmological arXiv:1902.07876 [hep-th]. constant in the Palatini formalism of modified [61] Ryusuke Jinno, Mio Kubota, Kin-ya Oda, and gravity,” Gen. Rel. Grav. 43, 1733–1757 (2011), Seong Chan Park, “Higgs inflation in metric arXiv:1007.2546 [gr-qc]. and Palatini formalisms: Required suppression of [45] Nicola Tamanini and Carlo R. Contaldi, “In- higher dimensional operators,” JCAP 03, 063 (2020), flationary Perturbations in Palatini Gener- arXiv:1904.05699 [hep-ph]. alised Gravity,” Phys. Rev. D83, 044018 (2011), [62] Katsuki Aoki and Keigo Shimada, “Scalar-metric- arXiv:1010.0689 [gr-qc]. affine theories: Can we get ghost-free theories [46] Florian Bauer and Durmus A. Demir, from symmetry?” Phys. Rev. D 100, 044037 (2019), “Higgs-Palatini Inflation and Unitar- arXiv:1904.10175 [hep-th]. ity,” Phys. Lett. B698, 425–429 (2011), [63] Massimo Giovannini, “Post-inflationary phases arXiv:1012.2900 [hep-ph]. stiffer than radiation and Palatini formula- [47] Gonzalo J. Olmo, “Palatini Approach to tion,” Class. Quant. Grav. 36, 235017 (2019), Modified Gravity: f(R) Theories and Be- arXiv:1905.06182 [gr-qc]. yond,” Int. J. Mod. Phys. D20, 413–462 (2011), [64] Tommi Tenkanen and Luca Visinelli, “Axion dark mat- arXiv:1101.3864 [gr-qc]. ter from Higgs inflation with an intermediate H∗,” [48] Syksy Rasanen and Pyry Wahlman, “Higgs inflation JCAP 08, 033 (2019), arXiv:1906.11837 [astro-ph.CO]. with loop corrections in the Palatini formulation,” [65] Nilay Bostan, “Non-minimally coupled quartic JCAP 11, 047 (2017), arXiv:1709.07853 [astro-ph.CO]. inflation with Coleman-Weinberg one-loop cor- [49] Antonio Racioppi, “Coleman-Weinberg rections in the Palatini formulation,” (2019), linear inflation: metric vs. Palatini arXiv:1907.13235 [gr-qc]. formulation,” JCAP 1712, 041 (2017), [66] Nilay Bostan, “Quadratic, Higgs and hilltop arXiv:1710.04853 [astro-ph.CO]. potentials in the Palatini gravity,” (2019), [50] Antonio Racioppi, “New universal attrac- arXiv:1908.09674 [astro-ph.CO]. tor in nonminimally coupled gravity: Lin- [67] Tommi Tenkanen, “Trans-Planckian ear inflation,” Phys. Rev. D97, 123514 (2018), censorship, inflation, and dark mat- arXiv:1801.08810 [astro-ph.CO]. ter,” Phys. Rev. D 101, 063517 (2020), [51] Laur Järv, Antonio Racioppi, and Tommi arXiv:1910.00521 [astro-ph.CO]. Tenkanen, “Palatini side of inflationary at- [68] Ioannis D. Gialamas and A.B. Lahanas, tractors,” Phys. Rev. D 97, 083513 (2018), “Reheating in R2 Palatini inflationary arXiv:1712.08471 [gr-qc]. models,” Phys. Rev. D 101, 084007 (2020), [52] Flavio Bombacigno and Giovanni Montani, “Big arXiv:1911.11513 [gr-qc]. bounce cosmology for Palatini R2 gravity with [69] Antonio Racioppi, “Non-Minimal (Self-)Running In- a Nieh–Yan term,” Eur. Phys. J. C 79, 405 (2019), flation: Metric vs. Palatini Formulation,” (2019), arXiv:1809.07563 [gr-qc]. arXiv:1912.10038 [hep-ph]. [53] Syksy Rasanen and Eemeli Tomberg, “Planck scale [70] I. Antoniadis, A. Karam, A. Lykkas, and K. Tam- black hole dark matter from Higgs inflation,” vakis, “Palatini inflation in models with an R2 term,” JCAP 01, 038 (2019), arXiv:1810.12608 [astro-ph.CO]. JCAP 1811, 028 (2018), arXiv:1810.10418 [gr-qc]. [54] Syksy Rasanen, “Higgs inflation in the Pala- [71] I. Antoniadis, A. Karam, A. Lykkas, T. Pappas, and tini formulation with kinetic terms for the K. Tamvakis, “Rescuing Quartic and Natural Inflation metric,” (2018), 10.21105/astro.1811.09514, in the Palatini Formalism,” JCAP 1903, 005 (2019), arXiv:1811.09514 [gr-qc]. arXiv:1812.00847 [gr-qc]. [55] Juan P. Beltrán Almeida, Nicolás Bernal, Javier Rubio, [72] Ignatios Antoniadis, Alexandros Karam, Angelos and Tommi Tenkanen, “Hidden Inflaton Dark Matter,” Lykkas, Thomas Pappas, and Kyriakos Tam- JCAP 03, 012 (2019), arXiv:1811.09640 [hep-ph]. vakis, “Single-field inflation in models with an R2 [56] Keigo Shimada, Katsuki Aoki, and Kei- term,” in 19th Hellenic School and Workshops on ichi Maeda, “Metric-affine Gravity and In- Elementary Particle Physics and Gravity (2019) flation,” Phys. Rev. D 99, 104020 (2019), arXiv:1912.12757 [gr-qc]. arXiv:1812.03420 [gr-qc]. [73] Tommi Tenkanen, “Tracing the high energy [57] Tomo Takahashi and Tommi Tenkanen, “Towards theory of gravity: an introduction to Pala- distinguishing variants of non-minimal inflation,” tini inflation,” Gen. Rel. Grav. 52, 33 (2020), JCAP 04, 035 (2019), arXiv:1812.08492 [astro-ph.CO]. arXiv:2001.10135 [astro-ph.CO]. [58] Ryusuke Jinno, Kunio Kaneta, Kin-ya [74] Tommi Tenkanen and Eemeli Tomberg, “Ini- Oda, and Seong Chan Park, “Hillclimb- tial conditions for plateau inflation: a case ing inflation in metric and Palatini formu- study,” (2020), 10.1088/1475-7516/2020/04/050, lations,” Phys. Lett. B 791, 396–402 (2019), arXiv:2002.02420 [astro-ph.CO]. arXiv:1812.11077 [gr-qc]. [75] Amy Lloyd-Stubbs and John McDonald, “Sub- [59] Tommi Tenkanen, “Minimal Higgs in- Planckian φ2 Inflation in the Palatini Formula- flation with an R2 term in Palatini tion of Gravity with an R2 term,” (2020), gravity,” Phys. Rev. D 99, 063528 (2019), arXiv:2002.08324 [hep-ph]. arXiv:1901.01794 [astro-ph.CO]. [76] Ignatios Antoniadis, Angelos Lykkas, and Kyriakos [60] Ariel Edery and Yu Nakayama, “Palatini formula- Tamvakis, “Constant-roll in the Palatini-R2 models,” tion of pure R2 gravity yields Einstein gravity with JCAP 04, 033 (2020), arXiv:2002.12681 [gr-qc]. 13

[77] D.M. Ghilencea, “Palatini quadratic gravity: sponta- arXiv:1903.03707 [hep-th]. neous breaking of gauged scale symmetry and infla- [84] Laur Jarv, Piret Kuusk, and Margus Saal, tion,” (2020), arXiv:2003.08516 [hep-th]. “Potential dominated scalar-tensor cosmolo- [78] T. Matsumura et al., “Litebird: Mis- gies in the general relativity limit: phase sion overview and focal plane layout,” space view,” Phys. Rev. D 81, 104007 (2010), Journal of Low Temperature Physics 184, 824–831 (2016). arXiv:1003.1686 [gr-qc]. [79] A Kogut, D.J Fixsen, D.T Chuss, J Dotson, E Dwek, [85] F. Briscese, E. Elizalde, S. Nojiri, and S.D. M Halpern, G.F Hinshaw, S.M Meyer, S.H Moseley, Odintsov, “Phantom scalar dark energy as modi- M.D Seiffert, D.N Spergel, and E.J Wollack, “The pri- fied gravity: Understanding the origin of the Big mordial inflation explorer (PIXIE): a nulling polarime- Rip singularity,” Phys. Lett. B 646, 105–111 (2007), ter for cosmic microwave background observations,” arXiv:hep-th/0612220. Journal of Cosmology and Astroparticle Physics 2011, 025–025[86] Sebastian (2011). Bahamonde, S.D. Odintsov, V.K. [80] Brian M. Sutin et al., “PICO - the probe of inflation Oikonomou, and Matthew Wright, “Correspon- and cosmic origins,” Proceedings, SPIE Astronom- dence of F (R) Gravity Singularities in Jordan and ical Telescopes + Instrumentation 2018: Modeling, Einstein Frames,” Annals Phys. 373, 96–114 (2016), Systems Engineering, and Project Management arXiv:1603.05113 [gr-qc]. for Astronomy VIII: Austin, USA, June 10-15, 2018, [87] Sergio Ferrara, Renata Kallosh, Andrei Linde, and Proc. SPIE Int. Soc. Opt. Eng. 10698, 106984F (2018), Massimo Porrati, “Minimal Supergravity Mod- arXiv:1808.01368 [astro-ph.IM]. els of Inflation,” Phys. Rev. D88, 085038 (2013), [81] Eanna E. Flanagan, “The Conformal arXiv:1307.7696 [hep-th]. frame freedom in theories of gravita- [88] Renata Kallosh, Andrei Linde, and Diederik tion,” Class. Quant. Grav. 21, 3817 (2004), Roest, “Superconformal Inflationary α-Attractors,” arXiv:gr-qc/0403063 [gr-qc]. JHEP 11, 198 (2013), arXiv:1311.0472 [hep-th]. [82] Javier Rubio and Eemeli S. Tomberg, “Preheating [89] Mario Galante, Renata Kallosh, Andrei Linde, in Palatini Higgs inflation,” JCAP 04, 021 (2019), and Diederik Roest, “Unity of Cosmological Infla- arXiv:1902.10148 [hep-ph]. tion Attractors,” Phys. Rev. Lett. 114, 141302 (2015), [83] Sotirios Karamitsos, “Beyond the Poles in At- arXiv:1412.3797 [hep-th]. tractor Models of Inflation,” JCAP 09, 022 (2019),