Time-Varying G in Shift-Symmetric Scalar-Tensor Theories With A Vainshtein Mechanism

Download as pdf or txt
Download as pdf or txt
You are on page 1of 4

Time-varying G in shift-symmetric scalar-tensor theories with a Vainshtein mechanism

Eugeny Babichev,1, 2 Cdric Deayet,2, 3 and Gilles Esposito-Far`se3 e e


Laboratoire de Physique Thorique dOrsay, Btiment 210, e a Universit Paris-Sud 11, F-91405 Orsay Cedex, France e 2 AstroParticule & Cosmologie, UMR 7164-CNRS, Universit Denis Diderot-Paris 7, CEA, Observatoire de Paris, e 10 rue Alice Domon et Lonie Duquet, F-75205 Paris Cedex 13, France e 3 GRCO, Institut dAstrophysique de Paris, UMR 7095-CNRS, Universit Pierre et Marie Curie-Paris 6, 98bis boulevard Arago, F-75014 Paris, France e (Dated: July 7, 2011) We show that the current bounds on the time variation of the Newton constant G can put severe constraints on many interesting scalar-tensor theories which possess a shift symmetry and a nonminimal matter-scalar coupling. This includes in particular Galileon-like models with a Vainshtein screening mechanism. We underline that this mechanism, if ecient to hide the eects of the scalar eld at short distance and in the static approximation, can in general not alter the cosmological time evolution of the scalar eld. This results in a locally measured time variation of G which is too large when the matter-scalar coupling is of order one.
PACS numbers: 04.50.Kd, 98.80.-k, 04.80.-y
1

arXiv:1107.1569v1 [gr-qc] 8 Jul 2011

Many theories in which gravity is modied with respect to general relativity (GR) contain, in addition to the metric, a scalar eld which is coupled directly to matter. Such scalar-tensor theories appear naturally in low energy limits of string theory and are also obtained from phenomenological brane-world constructions (such as the DGP model [1]). Some are also of current interest as able to produce an interesting cosmology via a large distance modication of gravity. In such theories, in contrast to GR, matter not only interacts via the helicity-2 graviton, but also via the exchange of the scalar eld. In general, one faces the following dilemma: Either this eld is coupled to matter with gravitational strength, as required to produce order one deviations from GR, but then the theory cannot pass local tests of gravity, or the coupling is very small, but then there are no signicant eects of the scalar. A canonical example is the Brans-Dicke theory [2, 3] and its extensions [4] whose parameters are tightly constrained by the local tests of gravity and observations of binary pulsars (see for instance [5]). A way out of this dilemma is provided by the Vainshtein mechanism, rst proposed in the context of massive gravity [6, 7] (a proof was recently provided in [8]). Indeed, close to localized bodies, this allows to screen eects which lead to large deviations from GR at large distances. This mechanism was also shown to be present in the DGP brane model [1] as well as its decoupling limit [9]. It was later generalized and shown to apply to large class of scalar-tensor models, called in [10]k-mouage gravity models, with a non-linear kinetic self-interaction of a scalar eld providing a self-screening of the scalar force ` la Vainshtein (hence the name k-mouage). This a class contains in particular the Galileon model [11], and its covariantized versions [1215]. Many applications of the Galileon model and its extensions to the late-time acceleration, including minimally coupled [1619] as well as non-minimally coupled models [2022], have been con-

sidered, while various constraints coming from cosmology as well as from local observations have been studied [17, 19, 2325]. In this paper, we point out that in spite of the fact that the Vainshtein screening indeed allows to pass most of the constraints coming from local observations by cutting o the spatial variation of the scalar eld near massive bodies, the tests on the constancy of the Newton constant may easily rule out many models. Indeed, we show that in many shift-symmetric models, the evolution with time of the scalar eld is (approximately) the same everywhere and it follows its cosmological behavior. If the scalar is directly coupled to matter, this induces a variation of the Newton constant G, which is tightly constrained by a number of observations (see, e.g., the review [26]). The most stringent bounds come from binary-pulsar data [27] and above all Lunar Laser Ranging experiments [28], the latter giving |G/G| < 1.3 1012 yr1 , or in terms of the Hubble value today, H0 , |G/G| < 0.02H0 . As we will see below, the time variation of the scalar eld is generically of order of the Hubble scale H0 (unless it is in the cosmological screening regime with a tiny energy scale M H0 ). This, whenever the direct coupling of the scalar eld to matter is of order of one, induces a too large variation of Newtons constant, |G/G| H0 . We consider the following general action, S=
2 MP 2

g d4 x g (R + Ls + LNL ) + Sm [ , m ] ,

(1) where R is the Ricci scalar of the metric g , Ls = 2 () is the standard kinetic term of a scalar eld (normalized to be dimensionless), LNL describes some generic nonlinear self-interaction of , and the matter elds (collectively denoted as m ) are minimally coupled to the physical metric g = A2 ()g . Because of this coupling, gravity is modied at large distances through

2 a scalar exchange, while GR is supposed to be restored at small distances thanks the Vainshtein screening eect made possible by the self-interaction LNL . This screening occurs for rather generic nonlinear interaction terms LNL [10] and we do not specify a precise from of it. We however assume that LNL is shift symmetric, i.e., it does not change under the transformation + const. For example, we can choose LNL to be in the Galileon or kessence families. For dimensional reasons LNL must contain a mass scale M . We will assume that this is the only additional scale entering our action. In general, this scale is xed by phenomenological requirements, e.g. to get present day acceleration of the Universe from Galileons, M should be of order of the Hubble scale H0 . Note that by a conformal transformation, action (1) can always be rewritten in a form where matter is only minimally coupled to one metric g , going to the so called Jordan frame. Here we will work rather in the Einstein frame (1), with the understanding that our results would apply to any theory whose action can be put in the form (1) by a suitable eld redenition. We will also not consider eects that can arise, even in theories with minimal coupling to gravity in the Einstein frame, due to the non-linear kinetic coupling of the graviton to the scalar eld (called kinetic braiding in [16]). The variation of Eq. (1) with respect to the metric g gives the (modied) Einstein equations,
2 (st) (NL) (m) MP G = T + T + T , (st) (NL) (m)

(2)

where T , T and T are respectively the energymomentum tensors for the standard scalar kinetic term, its nonlinear term, and the matter contribution. The equation of motion for the scalar eld is
2 ( + JNL ) = ()MP T (m) ,

(3)

Three dierent cosmological regimes can be identied. In a regime where the cosmological energy density of the scalar eld, , is subdominant compared to the matter energy density, m , m , the scalar eld equation (4) is decoupled from the metric equation (2). Then 2 from the Einstein equations it follows that m = 3MP H 2 , 2 thus the r.h.s. of (4) is H . If the scalar eld is away from the cosmological Vainshtein regime (i.e., when the nonlinear term in the l.h.s. of (4) is negligible), then from Eq. (4) one can see that a particular solution to (4) is |cosm | H. Notice that the general solution also contains a homogeneous decaying solution C0 exp 3 dtH with an arbitrary constant C0 , however unless this constant (or, equivalently, the initial condition) is ne tuned, the time variation of remains of order of H. In the second regime matter is again dominant, m , but the scalar eld is in the cosmological Vainshtein regime. Then Eq. (4) contains, in addition to H, also the nonlinear scale M . Therefore the solution to (4) generically contains a combination of scales H and M . When this scale is small with respect to H0 , the time evolution of cosm may thus be suppressed, |cosm | H0 . This is the cosmological analog of the original Vainshtein mechanism. The third regime is realized in the case when the scalar eld is dominant, m , and in particular when the late-time acceleration of the Universe is driven by the scalar eld. In this case, both the metric and the scalar eld equations depend only on one dimensionful parameter, M , which is of order of H0 . We thus conclude that the typical value of the present variation of the scalar eld is the Hubble scale, |cosm | H0 . Let us consider now the local eects caused by the conformal coupling of the scalar eld. For a slow time evolution, cosm (t) can be written as the linear approximation, cosm (t) = cosm (t0 ) + cosm (t0 ) t. (5)

where () d ln (A) /d, the nonlinear current JNL is obtained by variation of LNL with respect to the gradi ent of the scalar eld, JNL LNL /, , and T (m) is the trace of the matter energy-momentum tensor in the Einstein frame. Note that the eld equations (3) can always be written as the divergence of a current, because of shift symmetry. We also rescale the Planck mass so that A() = 1 at present. We stress that the trace of the matter energy-momentum tensor dened in the Einstein (m) frame, T , diers by the factor A4 () from the trace of the (conserved) Jordan-frame energy momentum tensor. However, as we will see below, the time variation of is small (of order of the Hubble time or less), so that the change of with time can be neglected in the r.h.s. of (3), giving only small corrections. For instance, if A() = e , the approximation A4 () 1 is valid for || 1, i.e., for |t| H 1 . The homogeneous cosmological evolution of the scalar eld, cosm (t) can be easily read from (3), 2 0 cosm + 3H cosm 0 JNL = ()MP T (m) .

Note that (5) imposes the boundary value of the eld far from localized sources. The solution to the full equation of motion (3) at any point of space-time (including the regions close to massive bodies, in particular, inside the Vainshtein radius) depends on both time and space coordinates. The key observation here is that thanks to the shift symmetry of the equation of motion, the PDE (3) allows separation of variables in the following way, (t, r) = (r) + cosm (t0 )t + cosm (t0 ), (6)

(4)

where r is the distance to the source. It is not dicult to see that the above ansatz (which has also been used in other contexts [29]) passes through the full equation of motion (3), giving an ordinary dierential equation of the second-order on (r), with possible remnants from the time-dependence in a form of a constant, cosm (t0 ). The last two terms in the above ansatz give the boundary condition for the PDE imposed by the cosmological evolution (5), provided we choose the radial dependent part

3 of (5), (r), to vanish at innity. Now, the ODE on (r) is of the second order, and supplied with two boundary conditions, (r = ) = 0 and (r = 0) = 0 (the last one comes from the regularity at the origin), which is in general sucient to nd a unique solution. Provided that this solution is non singular (which is in some cases a strong mathematical requirement), and assuming that (5) is a good approximation for the time-dependent cosmological evolution of , our ansatz gives a solution for all times to the eld equation (3) for a spherical source centered at r = 0. The key point of this paper is that the time derivative of (t, r) is found to be independent of r, i.e., it is set by the cosmological evolution even inside the regions where the Vainshtein screening operates. It is worth mentioning that our ansatz (6) and boundary conditions are in fact selecting a particular class of solutions. Indeed, if formulated in terms of a Cauchy problem, they implies (t = 0, r) = (r) and (t = 0, r) = cosm (t0 ). In principle, there is no reason not to choose some radially dependent initial velocity, (r) = C(r). In contrast to (6), such solutions, however, are not stationary, and they should relax to the stationary one, assuming that the latter is stable. A numerical check of this relaxation, as required by the nature of the eld equations, goes however far beyond the scope of this paper. The time-dependence of leads to a variation of the eective Newton constant with time. This can be seen by making the conformal transformation to the Jordan (physical) frame, with the metric g . Generically, there are two eects, which enter the nal result for the evolution of Newtons constant: the exchange of helicity-0 modes and the rescaling of the coordinates via the conformal transformation of the metric. It should be noted that in standard scalar-tensor theories (without screening mechanisms), these eects are of the same order, so that they even can compensate each others (as in Barkers theory [30]) giving no change of G. In our case, however, the exchange of is screened by the Vainshtein mechanism, so that only one eect the stretching of coordinates is important. As a result, the eective Planck mass in the action gains a dependence on cosm (t), MP = A1 (cosm ) MP . Thus, the observed Newton constant evolves with time as G/G 2cosm (t). (7) 1). In order for a theory to explain the accelerated expansion of the universe at present days, and pass the constraints on the variation of G, one should assume < 0.01. It is interesting to note that a similar constraint on the matter-scalar coupling constant was obtained for the covariant Galileon theory from a combined analysis of supernovae, baryonic acoustic oscillations and cosmic microwave background [22]. Let us now briey discuss the case of non shiftsymmetric theories. There is a class of such theories which can be put in the form (1) by suitable eld redenitions, in which case our conclusions apply. When this is not the case, the situation must be carefully reanalyzed. Indeed, the fact that the ansatz (6) leads to a mere ODE to solve for (r) is a direct consequence of the shift symmetry and it might be that some screening of the time variation of G occurs when this symmetry is lost. For example, we may introduce a mass term in the action, m2 2 , which explicitly breaks the shift symmetry. If the mass is big enough (say, much bigger than the present Hubble scale, H0 ), then the cosmological evolution of is suppressed, because the scalar follows the minimum of the eective potential, min H H/m2 . However, such theories do not possess either an interesting self-accelerating scenario driven by the kinetic term. We also stress that even in the shift-symmetric case, it might be that the ODE obeyed by (r) does not possess regular solutions, or that it leads to a stationary solution (6) which happens to be unstable as a solution of the PDE (3). If so, it may open a way out of our conclusions, necessitating the use of a more general ansatz than (6) to solve the eld equations, which could in turn result in a Vainshtein suppression of the time variation of G in the solar system. It also interesting to mention another possibility of avoiding any signicant time evolution of G, namely, to violate our assumption of a conformal matter-scalar coupling. In particular, in the relativistic MOND theory, called TeVeS [31], where the physical metric is related to the Einstein one in a disformal way, the time variation of the Newton constant is strictly zero [32]. This is a consequence of the dierent scalings of time and space coordinates with A(cosm ) when one imposes that the physical metric g tends to the Minkowski metric at spatial innity. This absence of time variation of G also applies to other theories with disformal coupling, in particular to the improved relativistic MOND [33], where the k-mouage screening has been used to pass solar-system and binary-pulsar constraints. Let us nally underline that the DGP brane model, although equivalent to a scalar-tensor theory of the Galileon type in UV (in particular in the decoupling limit [9]), is not fully described by such a theory at cosmological scales (IR limit). Therefore our analysis does not apply to the DGP model, and this explains why Ref. [34] did not nd any local time variation of G. To conclude, we have shown that a generic scalartensor theory with conformal coupling of a scalar eld

As we have seen before, depending on the regime, |cosm | H (in the matter domination regime) or |cosm | H (when the scalar eld is dominant). There fore, presently one has |G/G| 2 H0 if the scalar eld is subdominant and away from the cosmological screen ing, and |G/G| H0 when the scalar eld is dominant, in particular, when it drives the late-time acceleration of the universe. This applies in particular for a constant , i.e., a conformal coupling g = e2 g . The observational constraints from Lunar Laser Rang ing give |G/G| < 0.02H0 which is enough to rule out theories of the kind considered here with a scalar coupling to matter of the order of the gravitational one (i.e.,

4 to matter, and with a shift symmetry of the scalar Lagrangian, is tightly constrained by the bounds on time variation of the Newton constant. The models which fall into this category are not only standard Brans-Dicke-like theories, but also those featuring a Vainshtein screening mechanism due to the kinetic self-coupling, in particular non-minimally coupled Galileon models. We argued that the local time evolution of the scalar eld is set by its cosmological evolution and not screened by the Vainshtein eect, which is only able to suppress the fth force due to the exchange of helicity-0 degree of freedom. The time derivative of the scalar eld is of order of the Hubble scale, unless the whole universe is in the screening regime (cosmological Vainshtein eect with M H0 ). This induces a time evolution of Newtons constant of the same order. This result applies for both the matter domination and the scalar eld domination cosmological regimes. It also does not depend on a particular form of the non-linear self-interaction term, provided it is shiftsymmetric. The key point is that the time dependence in (6), which is crucial in (7), is the same irrespectively of the precise structure of LNL . The evolution of the Newton constant is, however, tightly constrained by observa tions, |G/G| 0.02H0, therefore the conformal coupling on such theories is constrained too. For example, if the non-minimal coupling is of the form exp (2), then || should be less than 102 .
Acknowledgments

We thank Jrme Martin for discussions. The work eo of C.D. and G.E.-F. was in part supported by the ANR grant THALES.

[1] G. R. Dvali, G. Gabadadze, and M. Porrati, Phys. Lett. B 485, 208 (2000) [arXiv:hep-th/0005016]. [2] P. Jordan, Nature (London) 164, 637 (1949); Schwerkraft und Weltall (Vieweg, Braunschweig, 1955); Z. Phys. 157, 112 (1959); M. Fierz, Helv. Phys. Acta 29, 128 (1956). [3] C. Brans and R. H. Dicke, Phys. Rev. 124, 925 (1961). [4] T. Damour and G. Esposito-Far`se, Class. Quant. Grav. e 9, 2093 (1992). [5] G. Esposito-Far`se, Fundam. Theor. Phys. 162, 461 e (2011) [arXiv:0905.2575 [gr-qc]]. [6] A. I. Vainshtein, Phys. Lett. B 39, 393 (1972). [7] C. Deayet, G. R. Dvali, G. Gabadadze, and A. I. Vainshtein, Phys. Rev. D 65, 044026 (2002) [hep-th/0106001]. [8] E. Babichev, C. Deayet, and R. Ziour, Phys. Rev. Lett. 103, 201102 (2009) [arXiv:0907.4103 [gr-qc]]; Phys. Rev. D 82, 104008 (2010) [arXiv:1007.4506 [gr-qc]]; JHEP 0905, 098 (2009) [arXiv:0901.0393 [hep-th]]. [9] M. A. Luty, M. Porrati, and R. Rattazzi, JHEP 0309, 029 (2003) [arXiv:hep-th/0303116 [hep-th]]. [10] E. Babichev, C. Deayet, and R. Ziour, Int. J. Mod. Phys. D 18, 2147 (2009) [arXiv:0905.2943 [hep-th]]. [11] A. Nicolis, R. Rattazzi, and E. Trincherini, Phys. Rev. D 79, 064036 (2009) [arXiv:0811.2197 [hep-th]]. [12] G. W. Horndeski, Int. J. Theor. Phys. 10, 363 (1974). [13] C. Deayet, G. Esposito-Far`se, and A. Vikman, Phys. e Rev. D 79, 084003 (2009) [arXiv:0901.1314 [hep-th]]. [14] C. Deayet, S. Deser, and G. Esposito-Far`se, Phys. Rev. e D 80, 064015 (2009) [arXiv:0906.1967 [gr-qc]]; Phys. Rev. D 82, 061501 (2010) [arXiv:1007.5278 [gr-qc]]. [15] C. Deayet, X. Gao, D. A. Steer, and G. Zahariade, arXiv:1103.3260 [hep-th]. [16] C. Deayet, O. Pujolas, I. Sawicki, and A. Vikman, JCAP 1010, 026 (2010) [arXiv:1008.0048 [hep-th]]. [17] S. Nesseris, A. De Felice, and S. Tsujikawa, Phys. Rev. D 82, 124054 (2010) [arXiv:1010.0407 [astro-ph.CO]]. [18] A. De Felice and S. Tsujikawa, Phys. Rev. Lett. 105, 111301 (2010) [arXiv:1007.2700 [astro-ph.CO]]. [19] A. De Felice, R. Kase, and S. Tsujikawa, Phys. Rev. D 83, 043515 (2011) [arXiv:1011.6132 [astro-ph.CO]]. [20] N. Chow and J. Khoury, Phys. Rev. D 80, 024037 (2009) [arXiv:0905.1325 [hep-th]]. [21] F. P. Silva and K. Koyama, Phys. Rev. D 80, 121301

[22] [23] [24] [25] [26] [27]

[28] [29] [30] [31]

[32] [33] [34]

(2009) [arXiv:0909.4538 [astro-ph.CO]]; A. De Felice and S. Tsujikawa, arXiv:1008.4236 [hep-th]; R. Gannouji and M. Sami, Phys. Rev. D 82, 024011 (2010) [arXiv:1004.2808 [gr-qc]]; A. De Felice, S. Mukohyama, and S. Tsujikawa, Phys. Rev. D 82, 023524 (2010) [arXiv:1006.0281 [astro-ph.CO]]; A. De Felice and S. Tsujikawa, JCAP 1007, 024 (2010) [arXiv:1005.0868 [astroph.CO]]. A. Ali, R. Gannouji, and M. Sami, Phys. Rev. D 82, 103015 (2010) [arXiv:1008.1588 [astro-ph.CO]]. P. Brax, C. Burrage, and A.-C. Davis, [arXiv:1106.1573 [hep-ph]]; C. Burrage and D. Seery, JCAP 1008, 011 (2010) [arXiv:1005.1927 [astro-ph.CO]]. T. Kobayashi, Phys. Rev. D 81, 103533 (2010) [arXiv:1003.3281 [astro-ph.CO]]. T. Kobayashi, H. Tashiro, and D. Suzuki, Phys. Rev. D 81, 063513 (2010) [arXiv:0912.4641 [astro-ph.CO]]. C. M. Will, Living Rev. Rel. 9, 3 (2005) [gr-qc/0510072]. A. T. Deller, J. P. W. Verbiest, S. J. Tingay, and M. Bailes, Astrophys. J. 685, L67 (2008) [arXiv:0808.1594 [astro-ph]]; K. Lazaridis, N. Wex et al., Mon. Not. R. Astron. Soc. 400, 805 (2009) [arXiv:0908.0285 [astro-ph.GA]]. J. G. Williams, S. G. Turyshev, and D. H. Boggs, Phys. Rev. Lett. 93, 261101 (2004) [arXiv:gr-qc/0411113]. E. Babichev, Phys. Rev. D 83, 024008 (2011) [arXiv:1009.2921 [hep-th]]. B. M. Barker, Astrophys. J. 219, 5 (1978). J. D. Bekenstein, Phys. Rev. D 48, 3641 (1993) [arXiv:grqc/9211017]; J. D. Bekenstein and R. H. Sanders, Astrophys. J. 429, 480 (1994) [arXiv:astro-ph/9311062]; R. H. Sanders, Astrophys. J. 480, 492 (1997) [arXiv:astro-ph/9612099]; J. D. Bekenstein, Phys. Rev. D 70, 083509 (2004) [Erratum-ibid. D 71, 069901 (2005)] [arXiv:astro-ph/0403694]. J. D. Bekenstein and E. Sagi, Phys. Rev. D 77, 103512 (2008) [arXiv:0802.1526 [astro-ph]]. E. Babichev, C. Deayet, and G. Esposito-Far`se, e arXiv:1106.2538 [gr-qc]. A. Lue and G. Starkman, Phys. Rev. D67, 064002 (2003) [astro-ph/0212083].

You might also like