Arxiv:1201.1448V1 [Gr-Qc]

Home , Perfect fluid

arXiv:1201.1448v1 [gr-qc] 6 Jan 2012 field ihtetoeuvln arnindensities Lagrangian fluid equivalent scalar perfect a two a describing the with of with associated not possibility is the fluid short which namely effective fluids this field, the perfect In of by property not satisfied purposes. a is discuss formal correspondence we purely note this for because even still careful perfect one be one, formal to a needs as the duality regarding correspon- fluid while field/perfect convenient However, scalar a purposes. formal only for but dence identification, an not nainr etrain noesd ftedaiyby duality the of side cor- computing one this of on in possibility interest perturbations recent the inflationary which the by one, of motivated light perfect respondence the a in regard- as relevant following duality against is the caveat field a in fluid/scalar equiva- to shown the not leads ing as are property “fluid”, This 3], field [2, section. scalar fluid a perfect for lent a for equivalent are and re eiaie ftesaa ed hl h efc fluid perfect the tensor while stress-energy field, scalar first the field involves of below) scalar derivatives is (2) order the eq. defined—it by consequence, be (given a tensor to stress-energy As average par- point-like continuum. an of a need made already not not not does is does and it and ticles average; fundamental A an more interactions. from is their derive microscopic of instead, using and field, theory particles scalar fluid kinetic the a of via laws models fluid obtained the be and particles, can as- constituent quantities a its microscopic with systems; averaging sociated physical by different field obtained very scalar is are a fluid that fluid clear perfect is a it and view, of point conceptual the nov eiaie xlctybtcnb ecie using density described energy be the can only but explicitly derivatives involve edi eea eaiiycnb ersne saperfect scalar a as coupled (see, represented minimally be fluid can universe. a Relativity late General the that in the recognized field in in widely inflation quintessence is of par- of It are contexts and universe and the early decades in many important for ticularly cosmology and gravity clrfilshv enpeeti h ieaueon literature the in present been have fields Scalar L u 2 a ec,tefli ecito fasaa edis field scalar a of description fluid the Hence, . = h orsodnebtenasaa edada ffcieper effective an and field scalar a between correspondence The e.g. − ett neetv efc udwt qaino tt deter state dens of Lagrangian the equation because with complete fluid not perfect is effective correspondence an to lent qiaetfrapretfli,aenteuvln o minim a for equivalent not are fluid, L perfect a for equivalent ASnmes 44.b 42.y 04.20.-q 04.20.Fy, 04.40.-b, numbers: PACS ρ 1 e.[]fradtie icsin.From discussion). detailed a for [1] Ref. , hs w arnindniis which densities, Lagrangian two These . ti ieyakolde ht o omlproe,aminim a purposes, formal for that, acknowledged widely is It →L ←→ INTRODUCTION T 2 ab mut oecagn cnncl clrfil ihaphant a with field scalar “canonical” a exchanging to amounts ( = ρ hsc eatetadSA eerhCutr ihpsUni Bishop’s Cluster, Research STAR and Department Physics pressure , P 60CleeSre,Sebok,Q´bc aaaJM1Z7 J1M Québec, Canada Sherbrooke, Street, College 2600 + ρ ) u P a u n four-velocity and , b + g P ab L osnot does 1 aei Faraoni Valerio = P cietesaa edeetv ud ti elknown well per- is a It for [2]. equivalent fluid are fect fluid. densities effective Lagrangian these field that scalar the scribe hog potential a through ldsaa field scalar pled is metric spacetime the of perfect signature barotropic a [5]. be- and in equivalence discussed field the was scalar fluid has and k-essence [7]) [8], a galaxies in tween in a discussed matter as recently dark suggested been to been alternative has (which possible Ricci spacetime the to of explicitly curvature coupling fluid perfect a [6]). for densities also see 5], ([4, side other the using eaiiyadteisei odtriewehrboth whether determine to is densities issue Lagrangian the and Relativity assuming hc sotie (when obtained is which odnequation Gordon rmtecvratcnevto equation conservation covariant the from h tensor The udsrs-nrytensor stress-energy fluid h nrydniyadpesr eaiet comov- a to four-velocity relative with pressure observer and ing density energy The ∇ is h clrfil nrymmnu tensor energy-momentum field scalar the c iial ope clrfield scalar coupled minimally A e smmc efc uduigamnmlycou- minimally a using fluid perfect a mimick us Let the particular, in [9]; Ref. of notations the adopt We Lagrangian two these between non-equivalence The φ IIKN EFC LI IHA WITH FLUID PERFECT A MIMICKING IIAL OPE CLRFIELD SCALAR COUPLED MINIMALLY satmlk etrfil 1.Tefli four-velocity fluid The [1]. field vector timelike a is ∗ T lyculdsaa ed h exchange The field. scalar coupled ally ab ities ie ytesaa oeta.This potential. scalar the by mined [ ∇ φ T lyculdsaa edi equiva- is field scalar coupled ally = ] c ab φ L [ ∇ 1 φ ∇ c = sue h omo neetv perfect effective an of form the assumes ] φ φ a P nacre pctm self-interacting spacetime curved a in φ u ,with 0, 6= ∇ V a msaa field. scalar om and L ( b = 1 φ φ φ .W suesadr General standard assume We ). ∇ = versity, p − − L c T 2 P |∇ φ 2 1 dV ab dφ = ∇ osntvns identically) vanish not does g c and ab a φ − ( = (1) 0 = φ u ∇ ∇ ρ a hc are which , u c L c u P etfluid fect a φ φ a 2 | − ∇ + φ r ie by given are = in( sign = c ++. + ρ by h Klein- the obeys φ − ) − u ρ ∇ a g V orcl de- correctly u b T b ab ∇ ab + c . for 0 = g P φ ∇ ρ ab c (3) (2) φ = ). if 2

a b ab Tab[φ]u u and P = Tab[φ]h /3, respectively, where yields hab gab + uaub is the Riemannian 3-metric in the 3- ≡ c 4 c dV space orthogonal to u (this 3 + 1 decomposition makes d x √ g ( δφ) cφ + δφ =0 . (14) − − ∇ ∇ dφ sense when cφ is timelike). One easily obtains ∇ Z c c Using the identity ( δφ) cφ = (δφ cφ) δφφ and 1 c c ∇ ∇ ∇ ∇ − ρ = φ cφ V sign ( φ cφ) , (4) discarding the boundary term, one obtains 2 ∇ ∇ − ∇ ∇ dV φ + =0 . (15) 1 1 c c P = 1+ sign ( φ cφ) φ cφ dφ 3 − 2 ∇ ∇ ∇ ∇ Eq. (15) is not the Klein-Gordon equation because of c [4 + sign ( φ cφ)] V (φ) . (5) the incorrect sign of the potential derivative term. The − ∇ ∇ } difference between eq. (15) and the Klein-Gordon equa- If we restrict ourselves to the case in which cφ is time- tion (1) disappears, of course, if V = constant. In this c ∇ like, φ cφ< 0, we have case the potential term gabV (φ) in the stress-energy ∇ ∇ − tensor Tab[φ] can be attributed entirely to a cosmological 1 c constant, i.e., to gravity instead of matter. If this con- ρ = φ cφ + V (φ) , (6) −2∇ ∇ stant vanishes, V 0, then eqs. (6) and (7) yield ρ = P and eq. (15) coincides≡ with the Klein-Gordon equation. 1 c P = φ cφ V (φ) , (7) However, there is still something wrong: the scalar field −2 ∇ ∇ − sourcing gravity and appearing in the total action and 4 R Stotal = Sgravity +Smatter = d x √ g +S[φ] (16) − 2κ (P + ρ) uaub + Pgab (8) Z (where κ 8πG) will give the incorrect field equations ≡ 1 c = aφ bφ gab φ cφ Vgab (9) 1 ∇ ∇ − 2 ∇ ∇ − Rab gabR = κTab[φ] (17) − 2 −

Tab[φ] (10) instead of the Einstein equations ≡ c 1 in addition to ucu = 1. The last equation shows Rab gabR = κTab[φ] . (18) not only that a minimally− coupled scalar field can be − 2 given a perfect fluid description, but also that any per- When V 0, the correct Lagrangian density would be ≡ fect fluid with barotropic equation of state P = P (ρ) can = P = ρ instead of 2 = ρ. (Then 3 2 = ρ be mimicked by a scalar field with appropriate poten- Lcan only describe a perfectL fluid− with stiffL ≡ equation −L of tial V (φ). Roughly speaking, prescribing the equation of state P = ρ.) state P = P (ρ) corresponds to assigning a suitable po- The conclusion is that, for a perfect fluid, 1 = P is tential, but the correspondence between equation of state the Lagrangian density reproducing the correctL equations and scalar field potential is not one-to-one [10]. of motion, while 2 = ρ (or 3 = ρ) provides incorrect The Klein-Gordon Lagrangian density for the scalar equations of motion.L − L field is the well known [9] For completeness, we conclude this section by going back to eqs. (4) and (5) and considering the case of a 1 c spacelike cφ, although this choice obviously does not KG = φ cφ V (φ) , (11) ∇ c L −2 ∇ ∇ − correspond to any physical fluid. If φ cφ > 0 it is √ c ∇ ∇ which coincides with = P and the variation of the ac- ua = aφ/ φ cφ and, using eqs. (3)-(7), L1 ∇ ∇ ∇ tion S = d4x√ g P with respect to φ reproduces c KG u uc = 1 , (19) the Klein-Gordon equation− (1), as is also well known [9]. Let us tryR to adopt instead the other candidate La- 1 c grangian density ρ = φ cφ V (φ) , (20) 2∇ ∇ −

1 c 2 = ρ = φ cφ V (φ) ; (12) 1 c 5 P = φ cφ V (φ) . (21) L − 2 ∇ ∇ − −6 ∇ ∇ − 3 then, the variational principle Taking again = P yields the incorrect field equation L1 dV δS δ d4x √ g = 0 (13) φ 5 =0 , (22) 2 ≡ − L2 − dφ Z 3 while taking = = ρ yields again one obtains the incorrect energy-momentum tensor L ±L2 ∓ dV φ + =0 . (23) ab a b 1 ab c ab S = ∂ φ ∂ φ + η ∂ φ∂cφ V η , (28) dφ (2) − 2 − Hence, the Lagrangians , all give incorrect field 1 2 which does not reproduce T ab[φ] and the Klein-Gordon equations for a scalar fieldL with±L spacelike gradient. c equation unless V 0. Finally, we consider a null φ. In this case, the pur- ≡ pose of considering a scalar field∇ representation of a per- To reiterate the argument, one can consider an- fect fluid would be the modelling of the only perfect fluid other situation in which the Noether symmetry ap- with null four-velocity that makes sense physically, i.e.,a proach is applicable: that of a spatially homo- geneous and isotropic Friedmann-Lemaˆıtre-Robertson- null dust with associated stress-energy tensor Tab = uaub c Walker universe with line element with ucu = 0, describing a distribution of coherent mass- less φ-waves [11, 12]. In this case, it must be V = 0 c ds2 = dt2 + a2(t) dx2 + dy2 + dz2 (29) and and , which are both proportional to φ cφ, L1 L2 ∇ ∇ − vanish identically and the usual Lagrangian density (11) cannot describe the null dust. In fact, a minimally cou- in comoving coordinates (for simplicity, we consider here pled scalar field cannot work as a model of conformally only the spatially flat metric). The minimally coupled invariant fluid such as a null dust. To model such a fluid, Klein-Gordon field in this spacetime depends only on the the physics of the scalar field would need to be confor- comoving time, φ = φ(t), and its energy density and mally invariant, which can only be achieved by coupling pressure are conformally the scalar to the Ricci curvature of spacetime R via the introduction of the term Rφ2/12 in the ac- 1 ˙2 − ρ(t) = φ + V (φ) , (30) tion SKG [13–15]. Moreover, the scalar φ must be either 2 4 free (V 0), or have a quartic self-coupling V = λφ 1 [14]. In≡ general, mimicking a null fluid or an imperfect P (t) = φ˙2 V (φ) . (31) 2 − fluid with a scalar field requires that the latter be coupled non-minimally to the curvature [1, 16]. By adopting the Lagrangian density

˙ THE NOETHER APPROACH 1 a, φ, φ = P, (32) L In flat spacetime there is an independent line of ap- a suitable point-like Lagrangian is proach to the Lagrangian description of a perfect fluid. ˙2 The Noether theorem applied to the translational Killing 3 3 φ 3 L1 = 1√ g = 1a = a V = a P. (33) fields of the Poincarégroup for a field φ described by L − L 2 − ! the Lagrangian density [φ, ∂aφ, ηab] (where ηab is the L Minkowski metric) yields the (independent) canonical The Euler-Lagrange equation energy-momentum tensor [9]

d ∂L1 ∂L1 ab ∂ b ab = 0 (34) S = L ∂ φ η . (24) dt ∂φ˙ − ∂φ ∂ (∂aφ) − L This tensor is conserved and coincides with the canonical then yields the correct Klein-Gordon equation Tab[φ] of eq. (2) up to a constant [9]. By adopting ¨ ˙ dV 1 ab φ +3Hφ + =0 . (35) η ∂aφ∂bφ V (φ) (25) dφ L1 ≡−2 − one obtains By contrast, the second point-like Lagrangian 1 ab a b ab c ab ab ˙2 S(1) = ∂ φ ∂ φ + η ∂ φ∂cφ + V η = T [φ] . 3 3 φ 3 − 2 − L2 = 2√ g = a 2 = a + V = a ρ (36) (26) L − L − 2 ! − b As is well known, the conservation equation Tab[φ] = ∇ 0 reproduces the Klein-Gordon equation. By contrast, yields the incorrect equation of motion using dV 1 ab φ¨ +3Hφ˙ =0 . (37) 2 η ∂aφ∂bφ + V (φ) (27) − dφ L ≡−2 4

CONCLUSIONS

∗ The duality between a minimally coupled scalar field [email protected] and a perfect fluid is widely acknowledged, but it is not a [1] M.S. Madsen, Class. Quantum Grav. 5, 627 (1988);see complete one. Lagrangian densities which are equivalent also M. Madsen, Astrophys. Sp. Sci. 113, 205 (1985). for a perfect fluid cease to be equivalent for an effective [2] R.L. Seliger and G.B. Whitham, Proc. R. Soc. (Lon- don) A305, 1 (1968); B. Schutz 1970 Phys. Rev. D, 2 fluid constructed out of a scalar field. Specifically, the 2762 (1970); J.D. Brown, Class. Quantum Grav. 10, 1579 change = P to = ρ, which does not have conse- L1 L2 − (1993). quences for the equations of motion of a real fluid, does [3] S.W. Hawking and G.F.R. Ellis, The Large Scale Struc- change the Klein-Gordon equation of motion of a scalar. ture of Space-Time (Cambridge University Press, Cam- Could this change point to the possibility that a scalar bridge, 1973). field exist in nature which satisfies eq. (15) instead of [4] L. Boubekeur, P. Creminelli, J. Norena, and F. Vernizzi, 0808 eq. (1)? In the cosmological literature, such a field (sat- J. Cosmol. Astropart. Phys. , 028 (2008). [5] A. Diez-Tejedor and A. Feinstein, Int. J. Mod. Phys. D isfying eq. (37) instead of (35)) is known as a phantom 14, 1561 (2005); F. Arroja and M. Sasaki, Phys. Rev. D field [17–19]. This hypothetical scalar field would cause 81, 107301 (2010). superacceleration of the universe (i.e., Hubble parameter [6] S. Unnikrishnan and L. Sriramkumar, Phys. Rev. D 81, H a/a˙ increasing with time, H˙ > 0) and leads to a Big 103511 (2010). Rip≡ singularity at a finite time in the future [18, 19]. [7] O. Bertolami, C.G. Böhmer, T. Harko, and F.S.N. Lobo, Phys. Rev. D 75, 104016 (2007); O. Bertolami and J. Phantom fields have been the subject of much at- Páramos, Phys. Rev. D 77, 084018 (2008); O. Berto- tention in cosmology, due to repeated reports from as- lami, F.S.N. Lobo, and J. Páramos, Phys. Rev. D 78, tronomers that the effective equation of state parameter 064036 (2008); O. Bertolami, J. Páramos, T. Harko, w P/ρ of the cosmic fluid lies in a range which does and F.S.N. Lobo, in The Problems of Modern Cosmol- not≡ exclude, or even favours, w< 1: this is a signature ogy, P.M. Lavrov ed. (Tomsk State Pedagogical Univer- of a phantom scalar field which causes− superacceleration. sity, 2009) [arXiv:0811.2876]; V. Faraoni, Phys. Rev. D 76 There is consensus that phantom scalar fields are unsta- , 127501 (2007); O. Bertolami and J. Páramos, Class. Quantum Grav. 25, 245017 (2008); T.P. Sotiriou, Phys. ble, classically and, even more, quantum mechanically Lett. B 664, 225 (2008); O. Bertolami and M.C. Se- [18, 20], therefore a phantom field is extremely unlikely. queira, Phys. Rev. D 79, 104010 (2009); O. Bertolami However, a phantom can appear in low-energy effective and J. Páramos, arXiv:0906.4757; O. Bertolami and M. actions which are eventually modified by higher order Carvalho Sequeira, arXiv:0910.3876; T.P. Sotiriou and corrections. In the cosmological literature, a duality be- V. Faraoni, Class. Quantum Grav. 25, 205002 (2008); D. tween a canonical scalar field and a phantom one is is Puetzfeld and Yuri N. Obukhov, Phys. Rev. D 78, 121501 obtained by changing the sign of the kinetic energy den- (2008). [8] V. Faraoni, Phys. Rev. D 80, 124040 (2009). sity term in eqs. (30) and (31). Our discussion puts a new [9] R.M. Wald, General Relativity (Chicago University twist on this duality, in that a phantom scalar can be ob- Press, Chicago, 1984). tained from a canonical one by the exchange 1 2. [10] V. Faraoni, Am. J. Phys. 69, 372 (2001). L ←→ L It is presently unknown whether this exchange has a more [11] H. Stephani, D. Kramer, M. MacCallum, C. Hoenselaers, fundamental meaning. and E. Hertl, Exact Solutions of Einstein’s Field Equa- tions, 2nd edition (Cambridge University Press, Cam- We have mentioned that a null fluid could be con- 4 bridge, 2003). structed using a scalar field with potential V (φ) = λφ [12] J. Bicak and K.V. Kuchar, Phys. Rev. D 56, 4878 (1997). conformally coupled to the curvature and with lighlike [13] N.A. Chernikov and E.A. Tagirov, Ann. Inst. H. gradient. In general, a non-minimally coupled scalar field PoincaréA 9, 109 (1968). corresponds to an imperfect fluid whose stress-energy [14] C.G. Callan Jr., S. Coleman and R. Jackiw, Ann. Phys. tensor contains terms which can be interpreted as heat (NY) 59, 42 (1970). currents and anisotropic stresses [1]. Since a Lagrangian [15] F.G. Friedlander, The Wave Equation on a Curved Spacetime (Cambridge University Press, Cambridge, description of imperfect fluids has not yet been devel- 1975). oped, we cannot comment on this aspect of the non- [16] V. Faraoni, Int. J. Theor. Phys. 40, 2259 (2001). minimally coupled scalar field/imperfect fluid duality. It [17] R.R. Caldwell, Phys. Lett. B 545, 23 (2002). is plausible, however, that the obstruction to a perfect [18] R.R. Caldwell, M. Kamionkowski, and N.N. Weinberg, duality found for perfect fluids will carry over to (ef- Phys. Rev. Lett. 91, 071301 (2003). fective) imperfect fluids constructed with non-minimally [19] S.M. Carroll, M. Hoffman, and M. Trodden, Phys. Rev. 68 coupled scalar fields, if they are found to admit a La- D , 023509 (2003). [20] S. Nojiri and S.D. Odintsov, Phys. Lett. B 562, 147 grangian description. (2003); V. Faraoni, Phys. Rev. D 69, 123520 (2004); J.M. The author thanks a referee for constructive remarks Cline, S. Jeon, and G.D. Moore, Phys. Rev. D 70, 043543 and the Natural Sciences and Engineering Research (2004); S.D.H. Hsu, A. Jenkins, and M.B. Wise, Phys. Council of Canada for financial support. Lett. B 597, 270 (2004); R.V. Buniy and S.D.H. Hsu, 5

Phys. Lett. B 632, 543 (2006).