Arxiv:1702.02552V2 [Gr-Qc]

arXiv:1702.02552v2 [gr-qc] 13 Oct 2017 n hi iedrvtvsaegvna niiilspace- initial an at given hypersurface are derivatives like time their and oriaetasomto,tm vlto falteten general the of all freedom of components evolution metric the dy- time of are transformation, Because (EFE) coordinate equations field equations. Einstein’s namical the of ponents osnticuescn iedrvtv ¨ derivative time which second equation), include Friedmann the not of the one does by (i.e., determined constraints simply is four universe the of lution Friedmann- the consider evo- physical we Lemaˆıtre-Robertson-Walker the when cosmology, of (FLRW) physically example, otherwise set For are special that possible. at all a among satisfied allowed only are are uni- born, states they a is when if (spacetime) Therefore, time verse hypersurface. all spacelike at guar- initial hold an identities they initial Bianchi on that contracted constraints antee the four and give just conditions, EFE of components nld eodtm eiaie faymti component, metric any and of derivatives time second include a mgn nvrewt aiu ausof values various we with though universe given, a are imagine curvature can and density parameter energy once Hubble The factor. 1 rvt ihfe nta odtos ouint h cosm the to solution a conditions: initial free with Gravity pta rm1t ,wt h incneto aea e.[1]. are ref. indices as Latin same convention and as sign 3 denoted the to is with derivative 0 3, partial to from 1 run from indices spatial Greek usual, As eiaieas derivative and ti elkonfc htntalo h e com- ten the of all not that fact known well a is It ∂ 0 2 ~ g r e qa ounity. to equal set are 0 µ ontapa tali F.Teeoete0 the Therefore EFE. in all at appear not do uigiflto,adCBBmd oaiainwudb obser be would relativity. polarization general B-mode anisot by CMB predictions metric degrees and of increased inflation, cosm modes the the during non-oscillatory of and new larger Because regions, predicts much Λ theory argument. scales small anthropic on in varies the only by Λ formed but are inflation, galaxies un Then are by Λ Λ constant nonzero cosmological with the variations. for metric Implications against solved. geodesic again remain th unstable the paths replacing is geodesic by w derived spacetime is condition equations Minkowski field unimodular flat gravitational the the a to and there However, addition meanings, six. in most constraints at three are freedom princip of action th the degrees con in extend physical variations cosmological to metric the on proposed solve constraints to was four motivation it equ a work field with Einstein’s previous conditions, the the of In components ten conditions. the of four because ∇ nsadr eea eaiiyteuies antb start be cannot universe the relativity general standard In µ .INTRODUCTION I. 1 rasmcln h udmna constants fundamental The semicolon. a or h 0 The . eerhCne o h al nvre colo cec,T Science, of School Universe, Early the for Center Research g µν antb eemndee if even determined be cannot eateto srnm,Sho fSine h University The Science, of School Astronomy, of Department µ opnnso F onot do EFE of components ∂ µ racma n covariant a and comma, a or -- og,Bny-u oy 1-03 aa and Japan 113-0033, Tokyo Bunkyo-ku, Hongo, 7-3-1 H etbeb M -oepolarization B-mode CMB by testable -- og,Bny-u oy 1-03 Japan 113-0033, Tokyo Bunkyo-ku, Hongo, 7-3-1 ˙ = a/a a sdetermined is ftescale the of Dtd etme 2 2018) 12, September (Dated: H . oooiTotani Tomonori g µν A µ c anmtvto fti okwstecsooia con- cosmological the was work this of motivation main il httegaiyter hudb bet describe to conditions. able initial prin- free be guiding from should starting a theory evolution gravity time adopting the I), that Paper ciple (hereafter [2] study time pos- describe physically to states. any able initial from sible be starting would spacetime theory of evolution a disap- such would and of equations constraint pear, DOFs four physical the six Then by gravity. into equations constrained field it variations gravitational Rather, derive metric of to redundancy. natural set a be such full expect may disfavor to the may unreasonable deriving not nature in is that it redundancy and of EFE, sort 10-component a re- is transformation variations there a coordinate metric request to to about lated have condition not action do coor- stationary we under and of invariant the transformation, because already However, dinate six is transformation. action most coordinate Einstein-Hilbert at from are four gravity metric the of ten of (DOFs) the degrees origi- physical freedom all constraints the of while four independently, variations such components considering of from existence nates the EFE, rive 2 3 nifiieia oriaetasomto of transformation coordinate infinitesimal An n a osdrta iia ruetmyas pl for apply also may argument similar a that consider may pa EFE. One Bian a than contracted by rather law the integration, conservation to tial energy-momentum leads the this and about identities condition action ary rtsmti aitosof variations metric erates uhater fgaiywspooe nteprevious the in proposed was gravity of theory a Such de- to principle action the of viewpoint the From e hc sraoal eas h gravity’s the because reasonable is which le, e osrit ihnwoe eurn that requiring ones new with constraints ree h ntblt rbe ste naturally then is problem instability The tn rbe.Ti a oeb setting by done was This problem. stant hne;teter ovre noEFE into converges theory the changed; oygnrtdb unu fluctuation quantum by generated ropy etopolm bu hster;the theory; this about problems two re tos(F)aecntanso initial on constraints are (EFE) ations r nrdcdwtotcerphysical clear without introduced ere dwt rirr nta conditions, initial arbitrary with ed tprubtos eeanwstof set new a Here perturbations. st rvt hoyt lo reinitial free allow to theory gravity e ffedmi ercdnmc,the dynamics, metric in freedom of lgclcntn rbe ssolved is problem constant ological e ieetyfo h standard the from differently ved hntepeetHbl horizon. Hubble present the than eUiest fTokyo of University he lgclcntn problem constant ological fTokyo of δg µν = − ζ µ ; ν 2 hsipisthat implies This . − ζ ν ; µ x n h station- the and , µ → x µ + ζ 3 µ The gen- chi r- 2 stant (Λ) problem (see e.g. [3–6] for reviews). The key posed by Paper I. One is that the three constraints of is how to set four constraints on metric variations δgµν δg0i = 0 were introduced in a rather ad hoc way without to extract six gravity DOFs. The unimodular condition, clear physical background. The other is that, as will be µν δ(√ g)=(1/2)√ gg δgµν = 0, was chosen as the first shown in II, the flat Minkowski spacetime becomes un- one.−4 This condition− has been studied for a long time stable against§ perturbations. The purpose of this work is starting from Einstein [7–19], and it is interesting con- to present a new version of the gravitational field equa- cerning the Λ problem because a term Λg(x) gµν appears tions ( III), which are similar but modified from those in the field equations as a Lagrange multiplier. How- in Paper§ I, by replacing the three constraints with those ever, introducing only this constraint does not solve the having a more solid physical basis: geodesic paths are Λ problem, because the contracted Bianchi identities re- kept geodesic against metric variations. It will be found quire Λg to be a universal integration constant, and we that the instability problem naturally disappears in this do not know how to set its initial value. In Paper I, the new version. Then implications for cosmology will be three more constraints were assumed to be δg0i = 0, sim- discussed in IV, in particular primordial metric fluctua- ply to make the 0i components of EFE ineffective. This tions produced§ by quantum effect during inflation. Since violates general covariance, and for a consistent theory it this theory has more DOFs of metric dynamics than stan- was assumed that any spacetime is initially created as a dard general relativity, new modes of metric anisotropy spacelike hypersurface, and the conditions δg0i = 0 hold would be generated, which may be tested by future ob- in preferred reference frames, i.e., synchronous coordi- servations such as B-mode polarization of the cosmic mi- nate systems starting from this hypersurface. crowave background radiation (CMB). The motivation of just allowing free initial conditions may not be strong enough to consider an alternative theory of gravity. However, the proposed theory gives a II. STABILITY OF THE FIELD EQUATIONS IN solution both for the smallness and coincidence prob- PAPER I lems of Λ. A universe is assumed to start with highly inhomogeneous conditions, and hence Λg(x) is not a uni- The gravitational field equations proposed in Paper I versal constant. However, once a portion of the uni- are verse starts inflation [20–25], anisotropy rapidly disap- µν µν µν µν pears and Λ (x) converges into a cosmological constant G κT =Λg(x) g + Ξ , (1) g − term Λ. Hence the universe can be described by EFE+Λ where Gµν = Rµν (1/2) Rgµν is the Einstein tensor, within the present-day Hubble horizon, but the final to- κ = 8πG, G the Newton’s− gravitational constant, and tal value of Λ Λ +Λvac can be positive or negative and g T µν the energy-momentum tensor of matter. The term changes on scales≡ much larger than the Hubble horizon, including a scalar field Λg(x) comes from the unimodu- where Λvac is the contribution from vacuum energy den- lar condition on δg , and the tensor Ξµν comes from the sity of all relevant fields in the particle physics theory. µν Then the anthropic argument for the cosmological con- three conditions of δg0i = 0. In this theory, it is assumed that any spacetime realized in nature is finite toward past stant [26, 27] applies; galaxy formation is possible only in timelike directions, and the initial spacelike hypersur- in the regions of ρΛ . ρ , like concentration of human M face at the birth of the universe can be defined at least as populations to coastal| | areas on Earth regarding Λ as al- a classical theory. Then we can define a synchronous co- titude, where ρΛ and ρ are energy densities of Λ and M ordinate system starting from this hypersurface, in which matter, respectively. The probability distribution of Λ is g00 = 1 and g0i = 0 throughout the spacetime. In this expected to be flat per unit Λ because Λ Λg and | | ≪ | | reference frame, it is assumed that δg0i do not repre- Λvac . Combined with galaxy formation efficiency as a sent gravity DOFs, and hence the 0i components of EFE |function| of Λ, the probability of finding a small Λ value simply become ineffective by Ξµν , which takes the form as observed [28–35] is not extremely small (typically 5– 00 ∼ of Ξ = Ξij = 0 with three nonzero components of 10 % [36]). Moreover, since the theory becomes EFE+Λ 0i i0 in the present Hubble horizon, it passes all the classical Ξ = Ξ . Since the initial hypersurface of constant time is fixed, the synchronous condition allows only transfor- tests supporting general relativity. ′i i j However, there are two problems in the theory pro- mations within spatial coordinates [x = f (x )], and the form of Ξµν is unchanged by these. The term Ξµν can be converted into any coordinate systems if we define the ordinary tensor transformation law, and then electromagnetic dynamics, because there are similarities between eqs. (1) become covariant. The theory is derived from mathematical structures of EFE and the Maxwell equations. the same Lagrangians as general relativity and hence the However, there is considerable difference in the nature of gravity stationary action condition does not depend on choice and electromagnetic forces as well, and here we simply assume of reference frames. However, the general principle of that this principle is valid only for gravity. 4 Note that we request only the variational condition of δ(√ g)= relativity is violated because the constraints on δgµν to 0, though the original unimodular condition was √ g = 1.− This extract the gravity DOFs have a preferred frame. Since variational condition is sufficient to generate the− cosmological the preferred frame can physically be specified by the ini- constant term in aribtrary coordinate systems with √ g = 1. − 6 tial spacelike hypersurface, this is a consistent theory to 3 determine evolution of the gravitational fields and space- term, and see also Paper I for some theoretical moti- time. vations to introduce this condition in the action princi- However, there is a problem in these equations, which ple. Then we should seek for new three constraints on becomes apparent when we consider linear perturbations δgµν on a more solid physical basis. Here geodesics may from the flat Minkowski spacetime. In the synchronous be relevant, because the effective number of components µν frame, we consider perturbations of hij ,Λg(x), and Ξ , of the geodesic equations is indeed three. The princi- where gµν = ηµν + hµν and ηµν is the metric of the ple of equivalence tells us that gravity can be erased in Minkowski spacetime as usual. The contracted Bianchi any local inertial frames, and hence it is reasonable to 5 µν µν identities of eqs. (1) lead to ∂µΛg g + µΞ = 0, and expect that a geodesic should remain geodesic against the ν = 0 and i components are: ∇ metric variations related to the physical DOFs of gravity. It is difficult to find covariant constraints on δg 0i µν ∂0Λg + ∂iΞ = 0 (2) keeping any geodesics, and here the same assumption as 0i ∂iΛg + ∂0Ξ =0 . (3) Paper I is adopted that any spacetime realized in nature − starts from a well-defined spacelike hypersurface. The Therefore we find paths of fixed spatial coordinates in the synchronous ref- 2 erence frame starting from the initial hypersurface are ∂0 Λg + ∆Λg =0 , (4) geodesics, and they are natural “backbone” of spacetime where ∆ is the three-dimensional Laplacian. This is when we consider evolution of spacetime to the timelike clearly unstable, because a mode of wavenumber k will direction. Therefore it is assumed that metric variations grow exponentially. The field Λ (x) is a scalar and hence should keep these geodesics in the action principle to de- g rive the gravitational field equations. this result does not depend on choice of coordinate sys- µ µ tems. When we consider a plane wave propagating into Let u = dx /dτ be the four-velocity field of these 3 backbone geodesics xµ(τ), where τ is the proper time the x direction, the mode (h11 +h22) is prohibited in the α µ ordinary EFE because the 00 and 33 components give along the geodesics. Therefore u αu = 0, and in the synchronous frame τ = x0 and uµ =∇ (1, 0, 0, 0). The con- α µ (h11 + h22) 33 = (h11 + h22) 00 =0 . (5) dition of keeping geodesics is given by δ(u αu )=0 , , ∇ against δgµν . Here, it should be noted that the backbone i 0 However, in this theory because of the freedom about paths x (x ) are kept unchanged by δgµν , and accord- µ Λg(x) we find ingly u changes by δgµν because a four-velocity includes 2 µ ν proper time that depends on metric, as dτ = gµν dx dx . (h11 + h22),00 + (h11 + h22),33 =0 , (6) Metric and four velocity fields after variations are de- µ µ µ noted asg ˜µν gµν + δgµν andu ˜ u + δu , respec- which also grows exponentially. Therefore, even if the ≡ µ ≡ theory predicts that inflation produces a spatially flat tively, and then variations of u become µν universe obeying EFE+Λ with negligible Ξ (Paper I), dxµ δ(dτ) a small perturbation will rapidly grow after inflation and δuµ = δ = uµ (7) dτ − dτ the Minkowski spacetime is not stable. This is the prob- 1 ρ σ µ µ lem that should be solved in the new field equations pre- = δgρσu u u u . (8) sented below. −2 ∝ µ Becauseu ˜µu˜ = 1 holds after variation, we find

III. A NEW APPROACH TO DERIVE THE α µ α µ u˜ ˜ α(˜uµu˜ )=2˜uµu˜ ˜ αu˜ =0 , (9) GRAVITATIONAL FIELD EQUATIONS ∇ ∇

where ˜ α is a covariant derivative usingg ˜µν . Note that A. Variations keeping geodesics α ∇µ u˜ ˜ αu˜ is not necessarily zero, if the path is no longer geodesic∇ after variation. Therefore, The constraints of δg0i = 0 in Paper I were introduced α ˜ µ α µ α µ in a rather ad hoc way simply to make the 0i compo- u˜µu˜ αu˜ = δ(uµu αu )= uµδ(u αu )=0 (10), ∇ ∇ ∇ nents of EFE ineffective, and this may be the cause of i.e., variations of the geodesic equations are perpendic- the difficulty described in II. In this work the unimod- µ ular condition is kept as one§ of the four constraints be- ular to u . This means that the condition of keeping cause it has a connection to the cosmological constant geodesics consists of three constraints rather than four, which is not surprising because there are only three in- dependent components in the geodesic equations. Variations of the geodesic equations can be written µ µ 5 The evolution of matter is determined by the least action prin- with δu and δΓρσ as ciple for the matter action in a fixed spacetime geometry, and α µ α µ α µ hence the matter equations of motion are not changed from the δ(u αu )= δu αu + u α(δu ) µν ∇ ∇ µ ∇ standard theory and the energy conservation µT = 0 holds. α β ∇ + u u δΓαβ , (11) 4

µ where Γρσ are the Christoffel symbols. Using eq. (8) and variation about δg00 using the constraint (14) as the geodesic equations, the first term on the right hand 0i 0i 4 side vanishes and we find 2 G κT √ gδg0i d x (17) Z − − 1 i 4 α µ α ρ σ µ α β µ 0 0 δ(u αu )= u u u u αδgρσ + u u δΓ (12). =2 ∂ (ξ √ g) δg i d x (18) ∇ −2 ∇ αβ Z − i 4 = 2 ξ √ g ∂0δg0i d x (19) Because this is perpendicular to uµ and the first term − Z − µ on the right hand side is proportional to u , in the syn- i 4 = ξ √ g ∂iδg00 d x (20) chronous frame these constraints are equivalent to re- − Z − questing that the spatial components of the second term i ∂i(ξ √ g) 4 are zero, i.e., = − √ gδg00 d x , (21) Z √ g − − α β i ij 2u u δΓαβ = g (2∂0δg0j ∂j δg00)=0 . (13) where the boundary terms are assumed to vanish as − usual. Now the stationary action condition about δg00 i Therefore the three constraints can simply be expressed and δgij , and the definition of ξ (eq. 16) can be com- as bined into ten-component new field equations as follows: µν µν µν µν G κT =Λg g + Ξ (22) − 2∂0δg0i ∂iδg00 =0 . (14) − where i ˜i 00 ∂i(ξ √ g) ∂iξ i α i Ξ = − = = ∂iξ Γ ξ ,(23) − √ g −√ g − − iα − − B. The new field equations i ˜i 0i ∂0(ξ √ g) ∂0ξ i α i Ξ = − = = ∂0ξ +Γ0 ξ , (24) √ g √ g α Here a new set of field equations is derived from the ij − − action principle by adopting the constraints (eq. 14) in Ξ =0 . (25) the synchronous coordinate system, instead of δg0i = 0 Here, the unimodular condition is also adopted to gener- µν i i in Paper I. An important feature of the constraints (eq. ate the term Λg(x) g , and ξ˜ ξ √ g. Note that now 14) is that they include derivatives of metric, in contrast Ξµν is different from that in Paper≡ I; it− is similar but Ξ00 to another constraint of the unimodular condition. In is nonzero and related to Ξ0i through ξi in this new ver- derivation of local differential field equations generally, sion. Since ξi is a time integration of the 0i component of variation of the action δS is expressed with δgµν (but EFE, these are integro-differential equations about time, without their derivatives using partial integrations), and i.e., nonlocal. then δS = 0 is requested for a local metric variation If we define a tensor Ξµν by the ordinary transforma- that is nonzero only at an infinitesimal region. A con- tion law, the field equations (22) would become a co- straint without derivatives can easily be incorporated in variant form holding in any coordinate systems. How- such a local metric variation, but it is not simple when a ever, the above expression of Ξµν by ξi (eqs. 23–25) is constraint includes derivatives, leading to nonlocal field valid in synchronous coordinate systems, and there is a equations. freedom of coordinate transformation within the spatial ′i i j Variation of the action by δgµν in general relativity is coordinates, x = f (x ), even if the initial spacelike hypersurface is fixed. Therefore if Ξµν can be defined as a tensor, eqs. (23)–(25) should be common in all these 1 µν µν 4 δS = (G κT ) √ gδgµν d x , (15) synchronous systems under an appropriate transforma- 2κ Z − − tion law of ξi, which can be shown as follows. The transformation law of Ξµν by the spatial coordinate transfor- and define ξi(x) in a synchronous coordinate system as 0 mation should be integrations of the 0i components over time (t x ), as ′ ≡ Ξ 00 = Ξ00 (26) ′i t ′0 ∂x 0 i 0i 0i ′ i Ξ i = Ξ j (27) ξ √ g G κT √ g dt + C , (16) ∂xj − ≡ Zts − − ′i ′j ′ij ∂x ∂x kl Ξ = k l Ξ =0 . (28) where integration is over a path of constant spatial co- ∂x ∂x ordinates (i.e., a backbone geodesic) starting from the In a synchronous frame, the Christoffel symbol in eq. i j α ij initial hypersurface on which time is ts, and C (x ) = (24) is Γ0α = (1/2) g gij,0, which is invariant under spa- i j 0i ξ (ts, x )√ g are integration constants. Then the 0i and tial coordinate transformations. Then Ξ obeys the ten- i0 metric variations− in eq. (15) can be converted into a sor transformation law if ξi are transformed as spatial 5 components of a four-vector ξµ whose 0-th component that any spacetime realized in nature is born as a space- ′0 0 is ξ = ξ = 0. Using this four vector, eq. (23) be- like hypersurface on which Λg = 0, by physics beyond 00 µ comes Ξ = µξ , which is a scalar and hence con- the level of a classical theory. sistent with the−∇ transformation law of Ξ00 (eq. 26). Now the theory can determine spacetime evolution Therefore Ξµν is consistently expressed by eqs. (23)– without ambiguity from given initial conditions about µν (25) in all the synchronous coordinate systems starting gij , ∂0gij , and T . The principle of free initial con- µν i from the initial hypersurface. Finally, if we define Ξ ditions is satisfied because of the freedom about ∂iξ˜ and i in non-synchronous coordinate systems as those trans- ∂0ξ˜ in the 00 and 0i components, respectively. The ini- i formed from the synchronous systems by the ordinary tial value of ∂iξ˜ is determined by the 00 component with µν i tensor law, a tensor Ξ is defined consistently in any Λg = 0, and evolution of ξ˜ is determined by the 0i com- 2 coordinate systems. ponents. The metric evolution is determined by ∂0 gij of the ij components, and then Λg evolution is determined by the time derivative of the 00 component. Therefore 2 C. Ambiguity in spacetime determination from the ij components to determine ∂0 gij are the same as initial conditions and its removal EFE on the initial hypersurface, and deviation from EFE emerges by evolution of Λg from zero, which is related to The role of a gravitational theory should be to de- nonlocal integration of physical quantities over the past termine time evolution of spacetime, i.e., gij in a syn- backbone paths starting from the initial hypersurface. µν i chronous frame, when gij , ∂0gij , and T on the initial Note that ξ affect the metric evolution only through the i i hypersurface are given. However, the equations (22) are form of ∂iξ˜ , and hence any ξ on the initial hypersurface 2 i not yet sufficient. The second time derivatives ∂0 gij ap- give the same spacetime solution if ∂iξ˜ is the same. pear only in the ij components, but these also include Λg(x). Though Λg can be erased using the 00 compo- i nent, ∂iξ˜ then appears in the ij components, which must D. Covariant derivation by Lagrange multipliers be determined independently. One may consider that the i µν three constraints from the 0i components determine ∂iξ˜ , The fact that Ξ becomes a tensor using the four- i but the 0i components determine only ∂0ξ˜ , with no con- vector field ξµ implies that a more covariant derivation straints on ξ˜i. If we consider ξi as new fields of physical of the field equations (22) in general coordinate systems DOFs, such an ambiguity may not be a problem. How- may be possible, and indeed it is, using the concept of ever, our motivation is just to remove the constraints on Lagrange multipliers as shown below. An important dif- initial conditions in EFE with increased DOFs of metric ference from the case of the unimodular condition is that dynamics, rather than to complicate the gravity theory the constraints to keep geodesics include metric deriva- by introducing new physical fields. If there is another tives, and hence derivatives of multipliers appear in the i constraint on the initial conditions of Λg and/or ∂iξ˜ , the field equations by partial integrations, which is related spacetime evolution would be unambiguously determined to the nonlocal nature of the constraints as already dis- without new physical fields. cussed. If the gravity plus matter action S is stationary This ambiguity arises because the field equations in- against δgµν under the constraints of keeping geodesics, ˜i α µ clude both ξ and their time derivatives. This is a conse- δ(u αu ) = 0, a modified action S˜ S + SC should quence of the nonlocal nature of the constraints to keep be stationary∇ with a Lagrange multiplier≡ four-vector ξµ, geodesics, which lead to the field equations including in- where tegrations of the 0i components of EFE over the past (ξ˜i). 1 α µ 4 The nonlocality is somewhat concordant with the basic SC = ξµ u αu √ g d x (29) assumption of the theory that there are preferred refer- κ Z ∇ − ence frames defined by the initial spacelike hypersurface, and a factor of 1/κ is introduced to make the resultant which is also a global property of the spacetime. Then field equations equivalent to eqs. (22). Because of the it would be natural to relate the initial conditions about µ α µ ˜i orthogonality between u and δ(u αu ), a condition of ξ and Λg to the initial spacelike hypersurface. A nat- ξ uµ = 0 can be set. Variation of S∇ by δg is then ural initial condition is that dynamical equations about µ C µν 2 ˜i ∂0 gij are not affected by ξ or Λg on the initial hypersur- 1 α µ 4 δSC = ξµ δ(u αu )√ g d x (30) face. This is achieved by requesting a covariant condition κ Z ∇ − Λg = 0 on the initial hypersurface, because the effect of 1 α β µ 4 ξ˜i is propagated to the ij components through Λ in the = ξµu u δΓ √ g d x , (31) g κ Z αβ − 00 component.6 Therefore, in this theory it is assumed

surface. However, in this option fluctuation of Λg produced by 6 Another possible constraint to remove the ambiguity is setting quantum effect during inflation would be too large to be consis- 00 µ Ξ = µξ = 0, rather than Λg = 0 on the initial hyper- tent with the presently observed universe (see IV B). −∇ § 6 where the contribution from the first term on the right tiplied by a factor of 2) become µ hand side of eq. (12) vanishes by ξµu = 0, and we do not 3 00: (h11 + h22),33 =2Λg 2 ξ 3 (37) have to consider variation of √ g because of the geodesic − , − equations for unvaried quantities. Since variations of the 11 : h22 + h33,00 = 2Λg (38) µ − Christoffel symbols δΓαβ are a tensor, δSC should be a 22 : h11 + h33,00 = 2Λg (39) scalar. We find − 33: (h11 + h22),00 = 2Λg (40) − 1 α β µρ 12 : h12 = 0 (41) δSC = ξµu u g (δgρβ,α + δgαρ,β δgαβ,ρ) − 2κ Z − 13 : h13,00 = 0 (42) n − µρ 4 + δg (gρβ,α + gαρ,β gαβ,ρ) √ g d x 23 : h23,00 = 0 (43) − o − − 1 01 : h13,03 = 2 ξ,0 (44) 1 ρ α β ρ α β − − = α(ξ u u )δgρβ + β(ξ u u )δgαρ 2 −2κ Z h∇ ∇ 02 : h23,03 = 2 ξ,0 (45) ρ α β − − 3 ρ(ξ u u )δgαβ 03: (h11 + h22),03 = 2 ξ 0 , (46) −∇ − , µ 4 µν + (terms including Γρσ) √ g d x , (32) where η ∂µ∂ν and the superscripts in eqs. (22) i − have been≡ converted into subscripts by the ordinary ten- 00 0j where partial integrations with vanishing boundaries sor laws, and note that Ξ00 = Ξ and Ξ0i = gij Ξ 0i ∼ have been used to get the second line, and the terms in- Ξ up to the first order perturbations. It can be − cluding Christoffel symbols appear by converting partial seen that the two modes of gravitational waves (h12 and µ derivatives into covariant ones and ∂α(√ g)=Γ √ g. h11 h22) in EFE are unchanged. − µα − − Since δSC is a scalar, the terms including Christoffel sym- Since free initial conditions are allowed in this the- bols should vanish, which of course can be checked by a ory, we consider plane wave perturbations hij = 3 direct calculation. Therefore, Aij exp(ikx ) for all the ij components on the initial hypersurface. For simplicity, here particular solutions with 7 1 µν 4 initial conditions of Λg = 0 and hij,0 = 0 are considered . δSC = Ξ δgµν √ g d x (33) −2κ Z − Then because ∂0Λg =0,Λg is zero throughout the spacetime, making the ij components the same as EFE. The i 3 where difference from EFE appears as ξ,0 and ξ,3 only in the 0µ components. There is no acceleration for h13 and h23, but µν µ ν ρ ν µ ρ ρ µ ν Ξ ρ(ξ u u )+ ρ(ξ u u ) ρ(ξ u u ) (34). ≡ ∇ ∇ − ∇ the difference from EFE is that their first time derivatives can be nonzero. However, we have assumed hij,0 = 0 ini- Then the field equations become eq. (22) after adding tially for the particular solutions considered here, and the unimodular condition. It is straightforward to verify µν hence they are kept constant with time. Then these two by component calculations that this tensor Ξ is exactly modes are not different from those possible in EFE, and the same as eqs. (23)–(25) in a synchronous coordinate they can be erased by a linear infinitesimal coordinate system. ′µ µ µ ′ transformation x = x + ζ and hij = hij ζi,j ζj,i, ν − − where ζµ ηµν ζ . For example, h13 disappears by a spatial coordinate≡ transformation keeping the synchronous −1 3 E. Linear perturbation analysis condition, ζ1 = A13(ik) exp(ikx ). Now give a look at the mode h11 + h22, which is also To examine the stability of flat spacetime in this the- force free and hence A11 + A22 becomes constant by the i assumed initial condition of h 0 = 0. It should be noted ory, consider linear perturbations about hij ,Λg and ξ ij, from the Minkowski spacetime without matter in a syn- that this mode is prohibited in EFE because of the 00 chronous coordinate. The contracted Bianchi identities component, (h11 + h22),33 = 0, in contrast to h13 and µν µν h23. But in this theory A11 + A22 can be nonzero by µ(Λg g + Ξ ) = 0 for the 0 and i components are: 3 ∇ the freedom of ξ,3. This mode cannot be erased by co- i i ordinate transformations, and it may have observational ∂0Λg + ∂0( ∂iξ )+ ∂i(∂0ξ )= ∂0Λg =0 , (35) − 2 i consequences that are different from EFE. Finally, the ∂iΛg + ∂0 ξ =0 . (36) − mode h33 feels a constant acceleration as 3 Therefore at the level of linear perturbation, the scalar 2h33,00 = 2ξ 3 = (h11 + h22),33 . (47) − , field Λg(x) is always a constant along the time coordinate. This is in sharp contrast to the theory proposed in

Paper I in which Λg exponentially grows, and hence the 7 instability problem has disappeared. Though this theory assumes Λg = 0 on the initial hypersurface when a spacetime was born, nonzero fluctuations of Λg may be Next we consider plane wave perturbations along the produced in later nonlinear cosmological evolution. However, 3 1 2 x direction, and hence derivatives about x and x dis- as argued in IV B, these conditions are expected to hold for § appear. The 10 components of the field equations (mul- fluctuations generated by inflation. 7

2 µν Though it seems to evolve in time as h33 = (Bt + and the contracted Bianchi identities µ(δΛg g + 3 µν ∇ A33) exp(ikx ) where A33 is constant and B = (A11 + Ξ ) = 0 become 2 − A22)k /4, this depends on the choice of a coordinate system. In fact, this mode is transformed into ν = 0 : ∂0δΛg = 0 (54) 1 0i 0i ν = i : ∂ δΛ + ∂0Ξ +5HΞ =0 . (55) ′ 1 3 2 i g h00 = 2B exp(ikx ) (48) −a − k2 ′ ′ Therefore δΛ becomes constant about time in the h = h = 0 (49) g 03 33 linear theory also in the FLRW metric. From the second equation, Ξ0i should exponentially decay if by a coordinate transformation of 0i 2 0i 5HΞ & ∂iδΛg/a , and hence 5HΞ is limited be- | | | 2 | | | 1 3 low ∂iδΛg/a . Consider a fluctuation whose comov- 0 | | ζ =2Bt 2 exp(ikx ) (50) ing wavenumber is ki H at the beginning of infla- k ∼ 1 tion, at which a is normalized to unity. Then we find 2 3 0i 2 00 0i µν ζ3 = (Bt + A33) exp(ikx ) . (51) Ξ . δΛg /a and Ξ Ξ , meaning that Ξ is ik | | | | | | ∼ | |µν 00 unimportant compared with δΛg g (i.e., Ξ δΛg 0i | | ≪ | | This is no longer a synchronous coordinate system, but and Ξ δΛg /a) when a 1. Therefore fluctua- | | ≪ | | ≫ the mode becomes a static gravitational field. The ini- tions at the beginning of inflation rapidly disappear and ′ tial fluctuation A33 disappears in h00, and B is related the field equations converge into EFE with a nonzero to the mode (h11 + h22), and hence the mode h33 is cosmological constant Λ on scales much smaller than the essentially the same as (h11 + h22). This implies that comoving scale of initial inhomogeneity before inflation 2 −1 the growing mode h33 t is an artifact by the con- (k ). The symmetry of general covariance is thus spon- ∝ i strained choice of synchronous coordinate systems, which taneously restored. However, the fluctuations of Λg on −1 becomes a static perturbed Minkowski spacetime in non- the comoving scale ki remain after inflation, with the synchronous frames. finalvalue of Λ = Λg +Λvac in the presently observed universe. Then the cosmological constant problem is solved by the anthropic argument, as explained in I (see Paper § IV. IMPLICATIONS FOR COSMOLOGY I for more discussions).

A. Solving the cosmological constant problem B. Primordial metric anisotropy generated by inflation Implications for the cosmological constant problem are not much changed from those obtained in Paper I. It is In IV A it was shown that the observable universe reasonable to assume that the universe started from a should§ be described by EFE+Λ after inflation at the level highly inhomogeneous state without the four constraints of a classical theory. It is the standard paradigm that the of EFE on initial conditions. In the theory Λg = 0 is large scale structures observed in the present universe are assumed on the initial hypersurface, but the linear per- generated by quantum fluctuations of the inflaton field, turbation theory cannot be adopted in such a state and and the two modes of primordial gravitational waves (h12 hence Λg(x) would soon evolve to nonzero values. Then and h11 h22) are also generated by quantum fluctua- in some regions of the spacetime inflation begins by the tions of metric,− which may be observed as B-mode polar- vacuum energy density of an inflaton field with a Hub- ization of CMB (see [37] for a review). These quantum ble parameter value of H. The inflating regions become fluctuations are assumed to become classical fluctuations isotropic and homogeneous, and hence anisotropic quan- when the wavelength of a mode becomes larger than the i tities such as ξ should become unimportant, and the Hubble horizon during inflation, though there are some contracted Bianchi identities ensure that Λg(x) converges fundamental theoretical issues about such a process. to a universal constant. This should be examined by nu- In the proposed theory, the four constraints on initial merical studies in future work, but can be verified at conditions in EFE have been removed, and DOFs of met- the linear theory level in the flat FLRW metric as fol- ric dynamics are increased. As shown in III E, the two lows. Consider linear perturbations about hij , δΛg, and gravitational wave modes are not changed in§ the new the- i ξ from the FLRW metric in a synchronous gauge, ig- ory. The newly increased modes do not obey the wave noring matter perturbations for simplicity. Here hij are 2 equation, but quantum fluctuations should exist for any defined by gij = a (δij hij ) so that they become the − − dynamical DOF. Therefore if the standard paradigm of same as those in II and IIIE when a = 1. The fluctu- quantum generation of density and metric fluctuations ation δΛ (x) is discriminated§ § from the nonzero constant g is correct, all the six modes of hij fluctuations are ex- Λg of the background metric. In this case pected to appear. The amplitude of primordial gravita- 2 3 2 2 tional waves is predicted to be ∆h k hk GH , 00 i 2 ∼ h| | i ∼ Ξ = ∂iξ (52) where hk is the power spectrum of metric fluctua- 0i − i i h| | i Ξ = ∂0ξ +3Hξ (53) tions of a comoving wavenumber k. This can be derived 8 by the uncertainty principle about time (H−1) and ki- V. CONCLUDING REMARKS netic energy within the Hubble volume (h˙ 2/G H−3) × using h˙ Hh. The same argument can be applied to the In this work a new set of gravitational field equations ∼ new modes, and hence they are expected to have similar was derived from the action principle, in which the La- amplitudes to those of the two gravitational wave modes. grangians are the same as standard general relativity but An important difference is, however, that the new modes four constraints are imposed on δgµν to extract six DOFs are not oscillatory. Matter fluctuations and gravitational of gravity. Then the four constraints on initial condi- waves are predicted to be Gaussian, because the ground tions in EFE are removed. This is motivated by a prin- state wave function of the harmonic oscillator is Gaus- ciple that the gravity theory should be able to describe sian. This implies that the non-oscillatory new modes spacetime evolution starting from free initial conditions would have strong deviation from Gaussian. on the initial spacelike hypersurface. Such field equa- The theory assumes Λg = 0 on the initial spacelike tions were originally proposed in Paper I, but here a hypersurface when the universe was born as a classical new version was derived by replacing the three of the spacetime. It is an interesting question how to treat this four constraints with more physically motivated ones, re- for fluctuations classicalized when they become super- questing that geodesics remain geodesic against varia- horizon during inflation. Ultimately the theory of quan- tions. Another constraint of the unimodular condition tum gravity would be required to answer this question, is unchanged. As a result, the theory becomes nonlocal but it is reasonable to assume δΛg = 0 as the initial with integro-differential field equations. The constraints condition at classicalization also for such metric pertur- of keeping geodesics were introduced independently of bations. It should be noted that the initial hypersurface the motivation of free initial conditions, but they natu- to determine the field equations may also be perturbed rally lead to the field equations similar to those of Paper from the background isotropic FLRW metric, which may I and the principle of free initial conditions is satisfied. be different for different wavelength modes. Then metric Furthermore, the instability problem in the version of perturbations in various wavelengths cannot be treated Paper I naturally disappears. These results lend some in a single synchronous coordinate system, but the syn- credence to the proposed theory as a candidate of the chronous coordinates in which Ξµν takes the simple form true gravity theory. of eqs. (23)–(25) may be different for different wavelength The field equations (22) are given in a covariant form, modes. Systematic formulations to treat such various which are derived from the covariant condition of the modes in a single coordinate system are interesting for least action, but the general principle of relativity is vi- future work, but here evolution of a single wavelength olated because the tensor Ξµν takes a simple form in mode is considered for simplicity. preferred coordinate systems defined by the initial space- Consider a mode of metric perturbation in its syn- like hypersurface when a classical spacetime is born. At chronous coordinate system. Because δΛg = 0 through- the cost of this, the theory allows free initial conditions out the spacetime within the linear theory for the FLRW about the metric, their first time derivatives, and matter metric, evolution of hij is determined by the ij com- distribution. Moreover, the cosmological constant prob- ponents of linearized EFE. Classicalization occurs when lem is solved in this theory. One may think that the the mode becomes superhorizon, and hence we can ig- theory is similar to the Einstein-Aether (EA) theory in nore spatial derivative terms in later evolution. In this which a timelike vector field is introduced and hence there case the equations for the metric perturbations become is a preferred reference frame [38, 39]. However, in the hïj = 3Hh˙ ij , and hence all components of hij are present theory the preferred frame is defined by a syn- quickly− frozen and become constant with time. After in- chronous coordinate system with respect to the initial flation, such frozen fluctuations gradually enter the Hub- hypersurface, and uµ is not a physical dynamical field. ble horizon from shorter wavelength modes. Then the The vector field ξµ is Lagrange multipliers rather than a new modes of metric anisotropy should affect the obser- dynamical field including kinetic terms, unlike the vec- vational signal of CMB B-mode polarization with am- tor field in the EA theory. The concept of the proposed plitudes similar to the ordinary two gravitational wave theory is to increase DOFs of metric dynamics by re- modes, on scales larger than the horizon at the recom- moving the four constraints on initial conditions in EFE, bination. Especially, the mode (h11 + h22) cannot be rather than introducing new physical fields. The pro- erased by gauge transformations as discussed in III E. posed theory does not include any new adjustable pa- The gravitational wave modes start to oscillate after§ they rameters, while the EA theory includes several. become subhorizon, and are damped on the time scale of The theory is indistinguishable from EFE+Λ after in- H−1 by the frictional term in the equations of motion. flation as a classical theory, and hence this theory passes Though the new modes are not oscillatory, their evolu- all observational/experimental tests supporting general tion would also change after entering the horizon. Quan- relativity and the standard cosmological model with Λ titative evolution must be calculated by a careful treat- and cold dark matter. However, it has a particular pre- ment of perturbation equations in an expanding universe diction about the primordial metric anisotropy generated with coupling to matter fluctuations, which is beyond the by quantum fluctuations during inflation. Because of the scope of this paper. increased DOFs of metric dynamics, new modes of fluc- 9 tuation would be generated including h11 + h22 in plane that gravitational fields should be quantized in the same wave solutions to the x3 direction, which is prohibited in way as other fields, which may not be true. Another standard general relativity. The two gravitational wave possibility is then that primordial metric anisotropy (in- modes predicted by EFE are unchanged, and amplitudes cluding ordinary gravitational waves) is not generated at of the new modes are similar to those. However, the new all by quantum metric dynamics during inflation. The modes are non-oscillatory and highly non-Gaussian, un- solution to the cosmological constant problem by this like the ordinary gravitational waves. These predictions theory is still valid even in such a case, though the the- may be tested by future observations seeking for CMB ory would become indistinguishable from EFE+Λ within B-mode polarization. A fundamental assumption here is the present Hubble horizon.

[1] L. D. Landau and E. M. Lifshitz, “The Classical Theory mological constant,” Int. J. Mod. Phys. D 10, 49 (2001) of Fields”, Elsevier (1975). [hep-th/9911102]. [2] T. Totani, “Initial Conditions of Inhomogeneous Uni- [17] L. Smolin, “The Quantization of unimodular gravity and verse and the Cosmological Constant Problem,” JCAP the cosmological constant problems,” Phys. Rev. D 80, 1606, no. 06, 003 (2016) [arXiv:1512.04411 [gr-qc]]. 084003 (2009) [arXiv:0904.4841 [hep-th]]. [3] J. Frieman, M. Turner and D. Huterer, “Dark Energy [18] G. F. R. Ellis, H. van Elst, J. Murugan and J. P. Uzan, and the Accelerating Universe,” Ann. Rev. Astron. As- “On the Trace-Free Einstein Equations as a Viable Al- trophys. 46, 385 (2008) [arXiv:0803.0982 [astro-ph]]. ternative to General Relativity,” Class. Quant. Grav. 28, [4] R. R. Caldwell and M. Kamionkowski, “The Physics of 225007 (2011) [arXiv:1008.1196 [gr-qc]]. Cosmic Acceleration,” Ann. Rev. Nucl. Part. Sci. 59, 397 [19] G. F. R. Ellis, “The Trace-Free Einstein Equations (2009) [arXiv:0903.0866 [astro-ph.CO]]. and inflation,” Gen. Rel. Grav. 46, 1619 (2014) [5] J. Martin, “Everything You Always Wanted To Know [arXiv:1306.3021 [gr-qc]]. About The Cosmological Constant Problem (But Were [20] A. A. Starobinsky, “Spectrum of relict gravitational radi- Afraid To Ask),” Comptes Rendus Physique 13, 566 ation and the early state of the universe,” JETP Lett. 30, (2012) [arXiv:1205.3365 [astro-ph.CO]]. 682 (1979) [Pisma Zh. Eksp. Teor. Fiz. 30, 719 (1979)]. [6] D. H. Weinberg, M. J. Mortonson, D. J. Eisenstein, C. Hi- [21] D. Kazanas, “Dynamics of the Universe and Spontaneous rata, A. G. Riess and E. Rozo, “Observational Probes Symmetry Breaking,” Astrophys. J. 241, L59 (1980). of Cosmic Acceleration,” Phys. Rept. 530, 87 (2013) [22] A. H. Guth, “The Inflationary Universe: A Possible Solu- [arXiv:1201.2434 [astro-ph.CO]]. tion to the Horizon and Flatness Problems,” Phys. Rev. [7] A. Einstein, “Spielen Gravitationsfelder im Aufbau der D 23, 347 (1981). materiellen Elementarteilchen eine wesentliche Rolle?”, [23] K. Sato, “First Order Phase Transition of a Vacuum and Sitzungsber. Preuss. Akad. Wiss. Berlin (Math. Phys.) Expansion of the Universe,” Mon. Not. Roy. Astron. Soc. 349–56 (1919); A. Einstein, The Principle of Relativ- 195, 467 (1981). ity (London: Methuen) (1923) (Engl. transl.); A. Ein- [24] A. D. Linde, “A New Inflationary Universe Scenario: A stein, The Principle of Relativity (Mineola, NY: Dover) Possible Solution of the Horizon, Flatness, Homogene- (Lorentz, H.A. et al., ed.), reprinted (1952), paper X. ity, Isotropy and Primordial Monopole Problems,” Phys. [8] J. L. Anderson and D. Finkelstein, “Cosmological con- Lett. B 108, 389 (1982). stant and fundamental length,” Am. J. Phys. 39, 901 [25] A. Albrecht and P. J. Steinhardt, “Cosmology for Grand (1971). Unified Theories with Radiatively Induced Symmetry [9] J. J. van der Bij, H. van Dam and Y. J. Ng, “The Ex- Breaking,” Phys. Rev. Lett. 48, 1220 (1982). change of Massless Spin Two Particles,” Physica 116A, [26] J. D. Barrow, and F. J. Tipler, The Anthropic Cosmo- 307 (1982). logical Principle. Oxford Univ. Press, New York (1986) [10] W. Buchmuller and N. Dragon, “Einstein Gravity From [27] S. Weinberg, “Anthropic Bound on the Cosmological Restricted Coordinate Invariance,” Phys. Lett. B 207, Constant,” Phys. Rev. Lett. 59, 2607 (1987). 292 (1988). [28] M. Fukugita, F. Takahara, K. Yamashita and Y. Yoshii, [11] W. G. Unruh, “A Unimodular Theory of Canonical “Test For The Cosmological Constant With The Number Quantum Gravity,” Phys. Rev. D 40, 1048 (1989). Count Of Faint Galaxies,” Astrophys. J. 361, L1 (1990) [12] M. Henneaux and C. Teitelboim, “The Cosmological [29] G. Efstathiou, W. J. Sutherland and S. J. Maddox, “The Constant and General Covariance,” Phys. Lett. B 222, cosmological constant and cold dark matter,” Nature 195 (1989). 348, 705 (1990). [13] Y. J. Ng and H. van Dam, “Possible solution to the cos- [30] Y. Yoshii, “Detection and selection effects in observations mological constant problem,” Phys. Rev. Lett. 65, 1972 of faint galaxies,” Astrophys. J. 403, 552 (1993) (1990). [31] L. M. Krauss and M. S. Turner, “The Cosmological con- [14] Y. J. Ng and H. van Dam, “Unimodular Theory of Grav- stant is back,” Gen. Rel. Grav. 27, 1137 (1995) ity and the Cosmological Constant,” J. Math. Phys. 32, [32] J. P. Ostriker and P. J. Steinhardt, “The Observational 1337 (1991). case for a low density universe with a nonzero cosmolog- [15] D. R. Finkelstein, A. A. Galiautdinov and J. E. Baugh, ical constant,” Nature 377, 600 (1995). “Unimodular relativity and cosmological constant,” J. [33] A. G. Riess et al. [Supernova Search Team Collabora- Math. Phys. 42, 340 (2001) [gr-qc/0009099]. tion], “Observational evidence from supernovae for an [16] Y. J. Ng and H. van Dam, “A Small but nonzero cos- accelerating universe and a cosmological constant,” As- 10

tron. J. 116, 1009 (1998) Roy. Astron. Soc. 464, 1563 (2017) [arXiv:1607.00180 [34] S. Perlmutter et al. [Supernova Cosmology Project Col- [astro-ph.CO]]. laboration], “Measurements of Omega and Lambda from [37] M. Kamionkowski and E. D. Kovetz, “The Quest for 42 high redshift supernovae,” Astrophys. J. 517, 565 B Modes from Inﬂationary Gravitational Waves,” (1999) Ann. Rev. Astron. Astrophys. 54, 227 (2016) [35] D. N. Spergel et al. [WMAP Collaboration], “First year [arXiv:1510.06042 [astro-ph.CO]]. Wilkinson Microwave Anisotropy Probe (WMAP) obser- [38] T. Jacobson, “Einstein-aether gravity: A Status report,” vations: Determination of cosmological parameters,” As- PoS QG-PH, 020 (2007) [arXiv:0801.1547 [gr-qc]]. trophys. J. Suppl. 148, 175 (2003) [39] T. Clifton, P. G. Ferreira, A. Padilla and C. Skordis, [36] T. Sudoh, T. Totani, R. Makiya and M. Nagashima, “Modiﬁed Gravity and Cosmology,” Phys. Rept. 513, 1 “Testing anthropic reasoning for the cosmological con- (2012) [arXiv:1106.2476 [astro-ph.CO]]. stant with a realistic galaxy formation model,” Mon. Not.