Abstract
We investigate effective equations governing the volume expansion of spatially averaged portions of inhomogeneous cosmologies in spacetimes filled with an arbitrary fluid. This work is a follow-up to previous studies focused on irrotational dust models (Paper I) and irrotational perfect fluids (Paper II) in flow-orthogonal foliations of spacetime. It complements them by considering arbitrary foliations, arbitrary lapse and shift, and by allowing for a tilted fluid flow with vorticity. As for the first studies, the propagation of the spatial averaging domain is chosen to follow the congruence of the fluid, which avoids unphysical dependencies in the averaged system that is obtained. We present two different averaging schemes and corresponding systems of averaged evolution equations providing generalizations of Papers I and II. The first one retains the averaging operator used in several other generalizations found in the literature. We extensively discuss relations to these formalisms and pinpoint limitations, in particular regarding rest mass conservation on the averaging domain. The alternative averaging scheme that we subsequently introduce follows the spirit of Papers I and II and focuses on the fluid flow and the associated \(1+3\) threading congruence, used jointly with the \(3+1\) foliation that builds the surfaces of averaging. This results in compact averaged equations with a minimal number of cosmological backreaction terms. We highlight that this system becomes especially transparent when applied to a natural class of foliations which have constant fluid proper time slices.
Similar content being viewed by others
Notes
Greek letters are assigned to spacetime indices, they run in \(\{ 0, 1, 2, 3 \}\), and Latin letters refer to space indices, running in \(\{ 1, 2, 3 \}\). The signature of the metric is taken as \((- + + +)\), and units are such that \(c = 1\). The coordinate system \(x^\mu \) is associated to the coordinate vector basis \(\{ \varvec{\partial }_{x^\mu } \} := \{ \varvec{\partial }_t, \varvec{\partial }_{x^i} \}\) and its dual exact basis \(\{ {{\mathbf {d}}} x^\mu \} := \{ {\mathbf{d}} t, {{\mathbf {d}}} x^i \}\). Unless otherwise specified, components of tensorial objects should be understood as expressed in these bases, with arguments \((t, x^i)\).
The shear, vorticity and acceleration of the fluid, as seen in the normal frames, can be derived from the projections onto the three-surfaces of the proper shear \({\varvec{\sigma }}\), proper vorticity \({\varvec{\omega }}\) and proper acceleration \({\varvec{a}}\), respectively. For instance, the second would read \(h^\alpha _{ \mu } h^\beta _{ \nu } \, \omega _{\alpha \beta }= h^\alpha _{ \mu } h^\beta _{ \nu } \, b^\delta _{ \alpha } b^\xi _{ \beta } \nabla _{[ \delta } u_{\xi ]}\), which differs from \(h^\alpha _{ \mu } h^\beta _{ \nu } \nabla _{[ \alpha } u_{\beta ]}\) when \({\varvec{a}}\) is not null.
For massive particles, the rest mass density \(\varrho \) is defined naturally and the associated definition of \({\varvec{u}}\) as the barycentric velocity ensures the conservation of the mass current \(\varrho \, {\varvec{u}}\). For massless particles such as a pure photon gas, a conserved number density current can be constructed instead in general, defining a 4-velocity \({\varvec{u}}\) as its direction and \(\varrho \), now a particle number density, as its norm; similar definitions from other conserved currents (such as the baryon current density) are possible, e.g., when nuclear reactions occur and the actual rest mass is not locally conserved [44, sec. 3.1], [42, p. 72]. Any of these definitions may be used in applications as appropriate for the specific matter model at hand, although we will, for simplicity, refer to \(\varrho \) as the ‘rest mass density’ and its volume integral as the ‘rest mass’ in what follows. Note that for a general fluid model, a definition of \({\varvec{u}}\) based on the energy frame may not always be compatible with such constructions based on conserved currents.
The proper time is not uniquely defined a priori, but it is fully determined by the choice of an initial Cauchy surface to build one of its level sets [48, p.74]. Another proper time function \(\tau '\), taking the constant value \(\tau '_{{\mathbf {i}}}\) on another initial hypersurface \(\varGamma '\), would differ from \(\tau \) by a function \(\varphi \) constant along the fluid flow lines, \(\tau '(t,X^i) = \tau (t,X^i) + \varphi (X^i)\). This relation follows by writing
$$\begin{aligned} \tau ' := \tau '_{{\mathbf {i}}} + \int _{t_{\varGamma '}(X^i)}^{t} \frac{N({\hat{t}}, X^i )}{\gamma ({\hat{t}}, X^i )} \; {\mathrm{d}}{\hat{t}} \, , \end{aligned}$$with \(\varGamma '\) parametrized by \(t = t_{\varGamma '}(X^i)\), yielding
$$\begin{aligned} \varphi (X^i) = \tau '_{{\mathbf {i}}} - \tau _{{\mathbf {i}}} + \int _{t_{\varGamma '}(X^i)}^{t_\varGamma (X^i)} \frac{N({\hat{t}}, X^i )}{\gamma ({\hat{t}}, X^i )} \; {\mathrm{d}}{\hat{t}} \, . \end{aligned}$$The expressions defining \(\tau \) and \(\tau '\) are here given in terms of comoving coordinates. They could alternatively be written covariantly, by setting the value of \(\tau - \tau _{{\mathbf {i}}}\) (resp. \(\tau ' - \tau '_{{\mathbf {i}}}\)) at a given spacetime event as the total length of the unique fluid flow line joining this event to the hypersurface \(\varGamma \) (resp. \(\varGamma '\)). The properties of both proper times and their relation through \(\varphi \) of course hold in this description.
For the same reason, the operators \({\mathrm{d}}/{\mathrm{d}}\tau \) and \(\partial _t |_{x^i}\) do not commute either with \(\int _{{\mathcal {D}}_{{\varvec{x}}}} \cdot \; {\mathrm{d}}^3x\).
In contrast to the operator \({\mathrm{d}}/{\mathrm{d}}t\), the operator \({\mathrm{d}}/{\mathrm{d}}\tau \) does not commute in general with \(\int _{{\mathcal {D}}_{{\varvec{X}}}} \cdot \; {\mathrm{d}}^3X\), since \({\mathrm{d}}/{\mathrm{d}}\tau = (\gamma / N) \; {\mathrm{d}}/{\mathrm{d}}t\) depends on the spatial coordinates.
One could in the same way parametrize the Friedmann model by a different time coordinate while staying within the homogeneous foliation, and similarly get rescaled terms and an arbitrarily altered acceleration term (see, e.g., the system of equations (20) in Paper II [19] or the system of equations (40) in [66]). The usual form of the Friedmann equations removes this freedom by choosing the proper time as the most natural time parameter in this situation. As, additionally, the spatial coordinates generally used in this framework are comoving with the fluid content, this picture corresponds to what we termed in this work a Lagrangian description.
Alternatively, we can derive the integrability condition directly from the Einstein equations. For this we can derive the local evolution equations for the square of the trace-free part of the extrinsic curvature and for the scalar 3-curvature using (3.2) and (3.4). Averaging these equations and combining them, we also obtain the integrability condition above (cf. [15, appendix]).
This is not obvious at first glance, due to a different choice of the scalars that have been averaged, i.e. in contrast to our case the averaged quantities in [74] do not involve the factor \(N^2\). Hence, the averaged equations do not appear identical to those obtained in the present work. To see that they are equivalent, the use of the corresponding local equations is necessary. The notations also differ (mostly because the description adopted in [74] is explicitly 4-covariant); one should take care in particular of the fact that in [74] the notation \(\partial _t\) is used for the coordinate-time covariant derivative along \({\varvec{n}}\) (i.e. \(N n^\mu \nabla _\mu \) in the notations of the present work), rather than for the coordinate-time partial derivative \(\partial _t |_{x^i}\).
Accordingly, and in contrast to a statement of [53], the resulting averaged system of equations, as expressed in \(3+1\) form, is not identical to that of Brown et al. [12] for a non-vanishing shift, as in this latter study the domain is transported along \(\varvec{\partial }_t\). This becomes true if a vanishing shift is chosen, due to the proportionality of \({\varvec{n}}\) and \(\varvec{\partial }_t\) in this case. As correctly stated, however, the averaged system of equations in \(3+1\) form of [53] becomes identical to that of Paper II for an irrotational perfect fluid if the fluid rest frames are used to generate the spatial hypersurfaces. This is indeed expected as in this case \({\varvec{n}} = {\varvec{u}}\), hence the domain has the same (fluid-comoving) evolution as in Paper II.
This system of averaged equations is given in a background-independent scheme as a preliminary step in [60, 87]. However, the emphasis is subsequently put on linear perturbation theory around a Friedmannian background, on which the main conclusions are based. Accordingly, no or negligible contributions from backreaction are found in this setting, which is expected due to the nonlinear and background-free nature of backreaction. We emphasize that mixing background-dependent applications with a background-free framework may imply strong restrictions, e.g. the small backreaction found by Russ et al. [80] in second-order perturbation theory with a Friedmannian background must in reality vanish due to the geometric constraints imposed (see the comments in Paper I [18], Sect. 3.4.).
The averaged equations are in general defined for arbitrary domains. If an assumption is adopted that distinct fluid congruences coincide or “average out” on the boundary, the arbitrariness of the domain choice has to be given up.
The averaged energy conservation equation and the integrability condition (see Sect. 3.4.2 above) are not always considered. The \(3+1\) approach of Beltrán Jiménez et al. [8] differs from the one used here in that it does neither include lapse nor shift, while Tanaka and Futamase [87] consider a nontrivial lapse along with a vanishing shift. In the approach of Räsänen [74], the formulation is only given in explicitly 4-covariant terms; also in this case can a \(3+1\) formulation be readily deduced, for comparison with the above averaged equations, upon making a coordinate choice including the appropriate time t.
Such additional differences with the results of [65] arise from a definition of the effective scale factor in this latter study that makes its evolution different from that of the cubic root of the domain’s volume. Since the aim of an averaging framework is to investigate the regional dynamics of comoving domains lying within spatial hypersurfaces, it seems to us to be more appropriate to define the scale factor from the volume of these domains. The reader may refer to [88] for a comparison of the different averaged energy constraints obtained for different choices of \(a_{\mathcal {D}}\), and for an analysis of the backreaction effects obtained for each such choice in a Friedmann-Lemaître-Robertson-Walker (FLRW) model perturbed up to second order. Note, however, that in these studies the domain also follows the congruence of the coordinate frames along \(\varvec{\partial }_t\), implying the drawbacks already highlighted in Sect. 3.5.5.
As \(\sqrt{h (t,x^k )} \,{\mathrm{d}}^3 x\), the fluid-orthogonal volume 3-form \(\sqrt{b (t,x^k )}\, {\mathrm{d}}^3 x\) is also invariant under a change of spatial coordinates, as can be checked either directly or by rewriting it as \(\gamma (t,x^k ) \sqrt{h(t,x^k )} \,{\mathrm{d}}^3 x\), \(\gamma = -n^\mu u_\mu \) being a 4-scalar. It reads in particular \(\sqrt{b (t,x^i )} \,{\mathrm{d}}^3 x = \sqrt{b (t, f^i(t,{\varvec{X}}))} \, J(t,X^i) \,{\mathrm{d}}^3 X\) in comoving spatial coordinates \(X^i\), with \(b (t, f^i(t,{\varvec{X}})) \, J(t,X^i)^2\) being the determinant of the spatial components of the fluid rest frame projector \({\mathbf {b}}\) in the comoving coordinate system \((t,X^i)\).
The difference to the averager used in Sect. 3 can be made explicit by introducing l as the counterpart of \(\ell \):
$$\begin{aligned} \frac{1}{a_{\mathcal {D}}} \frac{{\mathrm{d}} a_{\mathcal {D}}}{{\mathrm{d}} t} = \left\langle {\frac{1}{l} \frac{{\mathrm{d}} l}{{\mathrm{d}} t}} \right\rangle _{{\mathcal {D}}} \;\; \text {with} \;\; \frac{1}{l} \frac{{\mathrm{d}} l}{{\mathrm{d}}t} := \frac{1}{3} \left( \frac{N}{\gamma } \varTheta - \frac{1}{\gamma } \frac{{\mathrm{d}} \gamma }{{\mathrm{d}}t} \right) \, = \frac{1}{\ell } \frac{{\mathrm{d}} \ell }{{\mathrm{d}}t} - \frac{1}{3} \frac{1}{\gamma } \frac{{\mathrm{d}} \gamma }{{\mathrm{d}}t} \; . \end{aligned}$$We thus have \(l^3 = \ell ^3 / \gamma \), i.e. l is an isotropically averaged length (cubic root of a volume) associated with the volume contraction of \(\ell ^3\) by the Lorentz factor \(\gamma \): lengths are contracted by \(\gamma \) in one spatial direction and are not affected in the other orthogonal two directions, implying a factor \(\gamma ^{1/3}\) for the isotropically averaged length contraction. In comoving spatial coordinates, one can see from the first relation in (2.39) that \(\ell \) may be chosen as \(\ell \propto b^{1/6}\) from the determinant b, so that, accordingly, l may be chosen as \(l \propto h^{1/6}\). The continuity equation (2.24) also shows that one may define \(\ell \) as \(\ell \propto \varrho ^{-1/3}\), in any spatial coordinates.
However, \({\mathscr {R}}\) can indeed arise as the 3-Ricci scalar associated to a \({\varvec{u}}\)-orthogonal spatial ‘Riemann-like’ tensor which can be built from the \({\varvec{u}}\)-orthogonal spatial covariant derivative operator (defined for tensors fully orthogonal to \({\varvec{u}}\) as the projection through \({\mathbf {b}}\) on every component of their covariant 4-derivative) as well as from its spacetime embedding [47, 70, 76]. For non-vanishing vorticity, this Riemann-like tensor does not possess all the symmetry properties of a true Riemann tensor, and the way of defining such a spatial curvature tensor is not unique. Despite of this, \({\mathscr {R}}\) may be seen as the scalar part of local 3-curvature associated with this tensor in the \({\varvec{u}}\)-orthogonal subspace of the tangent space at each spacetime point. Boersma and Dray introduce so-called parametric manifolds to define this quantity as the curvature of the parametric submanifold [9]. Alternatively, we may see it simply as a definition through equation (4.16).
In the standard cosmological model it is assumed that the cosmological constant \({\Lambda }\) models Dark Energy; the averaged equations show that we then also have to account for Dark Energy backreaction\({\tilde{{\mathcal {L}}}}^b_{\mathcal {D}}\) in cases where \(N \ne \gamma \) and \({\Lambda }\ne 0\), cf. (4.19). Note that a change of time parameter within a given foliation, \(t \mapsto T(t)\), changes the value of this term at each time according to an affine transformation as the lapse factor in \({\tilde{\varLambda }}^b_{\mathcal {D}}\) gets rescaled.
In the paper introducing the morphon field [28], the possibility of phantom energies has been discussed too, which in this effective picture does not violate energy conditions. We have omitted this possible parametrization here.
The language of a given effective scalar field theory can be freely specified. We may think of other effective scalar field theories, e.g. non-minimally coupled, especially if we set the scalar field analogy within an extrinsic averaging formalism, where another dictionary could be a better choice. In this line, the analogy—here set up for fluid-intrinsic averaging—could have interesting implications for the relation of different effective scalar field theories to different foliation choices. By construction, the scalar field obtained here for any given domain \({\mathcal {D}}\) obeys the evolution equations of a homogeneous scalar field, being built from pure functions of t. One may, however, define it first (following the above procedure) as a pure function of the time t of a given foliation, and then consider this field in another foliation choice, where it will in general be inhomogeneous. In this way, the scalar field would acquire a nonvanishing spatial gradient and would so allow for a comparison with phenomenological inhomogeneous scalar fields employed in standard perturbation theory.
More precisely, the notation \({\mathbf {b}}_{|{\Sigma }}\) denotes the restriction of the type-(0,2) tensor \({\mathbf {b}}\) to \(T{\Sigma } \otimes T{\Sigma }\) where \(T{\Sigma }\) is the tangent space to the hypersurface; i.e., it corresponds to the type-(0,2) tensor on the submanifold \({\Sigma }\) obtained from the evaluation of \({\mathbf {b}}\) on vectors of \(T{\Sigma }\) only. We will use the same notation in Appendix D for 3-forms, meaning their restriction to \(T{\Sigma } \otimes T{\Sigma } \otimes T{\Sigma }\) in this case.
The integration constant \({\mathcal {W}}_{\mathcal {D}}^E(t_{{\mathbf {i}}}) + {\mathcal {Q}}_{\mathcal {D}}^E(t_{{\mathbf {i}}})\) is fully determined by the—arbitrary—choice of the constant \(k_{{{\mathcal {D}}_{{\mathbf {i}}}}}\) and by the initial conditions on \(H_{\mathcal {D}}^b\) and \(\langle \varrho \rangle ^b_{\mathcal {D}}\)via the averaged energy constraint equation (5.5) at \(t = {t_{{\mathbf {i}}}}\). With the choice \(k_{{{\mathcal {D}}_{{\mathbf {i}}}}} = 0\), the initial data of a reference Einstein-de Sitter FLRW model would, for instance, correspond to \({\mathcal {W}}_{\mathcal {D}}^E(t_{{\mathbf {i}}}) + {\mathcal {Q}}_{\mathcal {D}}^E(t_{{\mathbf {i}}})=0\).
In Newtonian cosmology we have to abandon the background-free character of general relativity. In order to obtain unique solutions, we have to introduce a background in terms of a linear reference velocity field, \(V_i = H_{ij} \, x_j\), with homogeneous expansion, shear and vorticity, \(H_{ij} := (\varTheta _H (t)/3) \, \delta _{ij} + \Sigma _{ij}(t) + \Omega _{ij}(t)\), with \(\Sigma _{ij}\) symmetric and traceless and \(\Omega _{ij}\) antisymmetric. Deviations thereof are to be bound to a 3-torus topology. As a result of the integrability of the Newtonian variables on flat space sections, the kinematical backreaction (which does not depend on the background variables) can be written in terms of full divergences of deviation vector fields. Hence, this backreaction has to vanish on the boundary-free 3-torus, see [27] and the recent discussion [24].
References
Alcubierre, M.: Introduction to \(3+1\) Numerical Relativity. Oxford University Press, Oxford (2008)
Alles, A., Buchert, T., Al Roumi, F., Wiegand, A.: Lagrangian theory of structure formation in relativistic cosmology III: gravitoelectric perturbation and solution schemes at any order. Phys. Rev. D 92, 023512 (2015). https://doi.org/10.1103/PhysRevD.92.023512. arXiv:1503.02566
Al Roumi, F., Buchert, T., Wiegand, A.: Lagrangian theory of structure formation in relativistic cosmology IV: Lagrangian approach to gravitational waves. Phys. Rev. D 96, 123538 (2017). https://doi.org/10.1103/PhysRevD.96.123538. arXiv:1711.01597
Arnowitt, R., Deser, S., Misner, C.: Republication of: the dynamics of general relativity. Gen. Relativ. Gravit. 40, 1997 (2008). https://doi.org/10.1007/s10714-008-0661-1. arXiv:gr-qc/0405109
Asada, H.: Lagrangian description of fluid flow with pressure in relativistic cosmology. Phys. Rev. D 62, 127301 (2000). https://doi.org/10.1103/PhysRevD.62.127301. arXiv:astro-ph/0010349
Asada, H., Kasai, M.: Lagrangian description of fluid flow with vorticity in relativistic cosmology. Phys. Rev. D 59, 123515 (1999). https://doi.org/10.1103/PhysRevD.59.123515. arXiv:astro-ph/9904005
Barbosa, R.M., Chirinos Isidro, E.G., Zimdahl, W., Piattella, O.F.: Cosmic bulk viscosity through backreaction. Gen. Relativ. Gravit. 48, 51 (2016). https://doi.org/10.1007/s10714-016-2043-4. arXiv:1512.07835
Beltrán Jiménez, J., de la Cruz-Dombriz, Á., Dunsby, P.K.S., Sáez-Gómez, D.: Backreaction mechanism in multifluid and extended cosmologies. J. Cosmol. Astropart. Phys. JCAP1405, 031 (2014). https://doi.org/10.1088/1475-7516/2014/05/031. arXiv:1312.5680
Boersma, S., Dray, T.: Slicing, threading and parametric manifolds. Gen. Relativ. Gravit. 27, 319 (1995). https://doi.org/10.1007/BF02109128. arXiv:gr-qc/9407020
Bolejko, K.: Volume averaging in the quasispherical Szekeres model. Gen. Relativ. Gravit. 41, 1585 (2009). https://doi.org/10.1007/s10714-008-0727-0. arXiv:0808.0376
Bolejko, K.: Cosmological backreaction within the Szekeres model and emergence of spatial curvature. J. Cosmol. Astropart. Phys. JCAP1706, 025 (2017). https://doi.org/10.1088/1475-7516/2017/06/025. arXiv:1704.02810
Brown, I.A., Behrend, J., Malik, K.A.: Gauges and cosmological backreaction. J. Cosmol. Astropart. Phys. JCAP0911, 027 (2009). https://doi.org/10.1088/1475-7516/2009/11/027. arXiv:0903.3264
Brown, I.A., Latta, J., Coley, A.: Backreaction: Gauge and frame dependences. Phys. Rev. D 87, 043518 (2013). https://doi.org/10.1103/PhysRevD.87.043518. arXiv:1211.0802
Brown, I.A., Coley, A.A., Herman, D.L., Latta, J.: Physical effects of consistent cosmological averaging. Phys. Rev. D 88, 083523 (2013). https://doi.org/10.1103/PhysRevD.88.083523. arXiv:1308.5072
Brunswic, L., Buchert, T.: Gauss–Bonnet–Chern approach to the averaged Universe. arXiv:2002.08336
Buchert, T.: Lagrangian perturbation approach to the formation of large-scale structure. In: Bonometto, S., Primack, J., Provenzale, A. (eds.) International School of Physics Enrico Fermi, Course CXXXII: Dark Matter in the Universe, Varenna 1995, pp. 543–564. IOP Press, Amsterdam (1996). arXiv:astro-ph/9509005
Buchert, T.: On average properties of inhomogeneous cosmologies, In: Eriguchi, Y., et al. (eds.) 9th JGRG Meeting, Hiroshima 1999, J.G.R.G., vol. 9, p. 306 (2000) arXiv:gr-qc/0001056
Buchert, T.: [Paper I] On average properties of inhomogeneous fluids in general relativity: dust cosmologies. Gen. Relativ. Gravit. 32, 105 (2000). https://doi.org/10.1023/A:1001800617177. arXiv:gr-qc/9906015
Buchert, T.: [Paper II] On average properties of inhomogeneous fluids in general relativity: perfect fluid cosmologies. Gen. Relativ. Gravit. 33, 1381 (2001). https://doi.org/10.1023/A:1012061725841. arXiv:gr-qc/0102049
Buchert, T.: A cosmic equation of state for the inhomogeneous universe: can a global far-from-equilibrium state explain dark energy? Class. Quantum Gravity 22, L113 (2005). https://doi.org/10.1088/0264-9381/22/19/L01. arXiv:gr-qc/0507028
Buchert, T.: On globally static and stationary cosmologies with or without a cosmological constant and the dark energy problem. Class. Quantum Gravity 23, 817 (2006). https://doi.org/10.1088/0264-9381/23/3/017. arXiv:gr-qc/0509124
Buchert, T.: Dark energy from structure—a status report. Gen. Relativ. Gravit. 40, 467 (2008). https://doi.org/10.1007/s10714-007-0554-8. arXiv:0707.2153
Buchert, T.: Toward physical cosmology: focus on inhomogeneous geometry and its non-perturbative effects. Class. Quantum Gravity 28, 164007 (2011). https://doi.org/10.1088/0264-9381/28/16/164007. arXiv:1103.2016
Buchert, T.: On backreaction in Newtonian cosmology. Mon. Not. R. Astron. Soc. Lett. 473, L46 (2018). https://doi.org/10.1093/mnrasl/slx160. arXiv:1704.00703
Buchert, T., Carfora, M.: Cosmological parameters are dressed. Phys. Rev. Lett. 90, 031101 (2003). https://doi.org/10.1103/PhysRevLett.90.031101. arXiv:gr-qc/0210045
Buchert, T., Carfora, M., Ellis, G.F.R., Kolb, E.W., MacCallum, M.A.H., Ostrowski, J.J., Räsänen, S., Roukema, B.F., Andersson, L., Coley, A.A., Wiltshire, D.L.: Is there proof that backreaction of inhomogeneities is irrelevant in cosmology? Class. Quantum Gravity 32, 215021 (2015). https://doi.org/10.1088/0264-9381/32/21/215021. arXiv:1505.07800
Buchert, T., Ehlers, J.: Averaging inhomogeneous Newtonian cosmologies. Astron. Astrophys. 320, 1 (1997). arXiv:astro-ph/9510056
Buchert, T., Larena, J., Alimi, J.-M.: Correspondence between kinematical backreaction and scalar field cosmologies—the ‘morphon field’. Class. Quantum Gravity 23, 6379 (2006). https://doi.org/10.1088/0264-9381/23/22/018. arXiv:gr-qc/0606020
Buchert, T., Mädler, T.: Editorial note to: On the Newtonian limit of Einstein’s theory of gravitation (by Jürgen Ehlers). Gen. Relativ. Gravit. 51, 162 (2019). https://doi.org/10.1007/s10714-019-2623-1. arXiv:1910.12106
Buchert, T., Mourier, P., Roy, X.: On cosmological backreaction and its dependence on space-time foliation. Class. Quantum Gravity 35, 24LT02 (2018). https://doi.org/10.1088/1361-6382/aaebce. arXiv:1805.10455
Buchert, T., Nayet, C., Wiegand, A.: Lagrangian theory of structure formation in relativistic cosmology II: average properties of a generic evolution model. Phys. Rev. D 87, 123503 (2013). https://doi.org/10.1103/PhysRevD.87.123503. arXiv:1303.6193
Buchert, T., Obadia, N.: Effective inhomogeneous inflation: curvature inhomogeneities of the Einstein vacuum. Class. Quantum Gravity 28, 162002 (2011). https://doi.org/10.1088/0264-9381/28/16/162002. arXiv:1010.4512
Buchert, T., Ostermann, M.: Lagrangian theory of structure formation in relativistic cosmology I: Lagrangian framework and definition of a nonperturbative approximation. Phys. Rev. D 86, 023520 (2012). https://doi.org/10.1103/PhysRevD.86.023520. arXiv:1203.6263
Buchert, T., Räsänen, S.: Backreaction in late-time cosmology. Ann. Rev. Nucl. Part. Sci. 62, 57 (2012). https://doi.org/10.1146/annurev.nucl.012809.104435. arXiv:1112.5335
Chirinos Isidro, E.G., Barbosa, R.M., Piattella, O.F., Zimdahl, W.: Averaged Lemaître–Tolman–Bondi dynamics. Class. Quantum Gravity 34, 035001 (2016). https://doi.org/10.1088/1361-6382/34/3/035001. arXiv:1608.00452
Clarkson, C., Ellis, G.F.R., Larena, J., Umeh, O.: Does the growth of structure affect our dynamical models of the Universe? The averaging, backreaction, and fitting problems in cosmology. Rep. Prog. Phys. 74, 112901 (2011). https://doi.org/10.1088/0034-4885/74/11/112901. arXiv:1109.2314
Clifton, T., Sussman, R.A.: Cosmological backreaction in spherical and plane symmetric dust-filled space-times. Class. Quantum Gravity 36, 205004 (2019). https://doi.org/10.1088/1361-6382/ab3a14. arXiv:1904.02557
Coley, A.A., Wiltshire, D.L.: What is general relativity? Phys. Scripta 92, 053001 (2017). https://doi.org/10.1088/1402-4896/aa6857. arXiv:1612.09309
Daverio, D., Dirian, Y., Mitsou, E.: General relativistic cosmological N-body simulations I: time integration. J. Cosmol. Astropart. Phys. JCAP1910, 065 (2019). https://doi.org/10.1088/1475-7516/2019/10/065. arxiv:1904.07841
Delgado Gaspar, I., Hidalgo, J.C., Sussman, R.A.: Non-comoving baryons and cold dark matter in cosmic voids. EPJC 79, 106 (2019). https://doi.org/10.1140/epjc/s10052-019-6606-x. arXiv:1811.03634
Ehlers, J.: Akad. Wiss. Lit. (Mainz); Abh. Math.-Nat. Kl. No. 11, 793 (1961); translation: Contributions to the relativistic mechanics of continuous media. Gen. Relativ. Gravit. 25, 1225 (1993). https://doi.org/10.1007/BF00759031
Ehlers, J.: Survey of general relativity theory. In: Israel, W. (ed.) Relativity, Astrophysics and Cosmology: Proceedings of the Summer School held 14–26 August, 1972 at the Banff Centre, Banff, Alberta, Dordrecht/Boston, vol. 38., D. Reidel Publishing Co., Astrophysics and Space Science Library (1973). https://doi.org/10.1007/978-94-010-2639-0_1
Ehlers, J., Buchert, T.: Newtonian cosmology in Lagrangian formulation: foundations and perturbation theory. Gen. Relativ. Gravit. 29, 733 (1997). https://doi.org/10.1023/A:1018885922682. arXiv:astro-ph/9609036
Ellis, G.F.R.: Republication of: Relativistic cosmology (originally published by G.F.R. Ellis in: Proceedings of the International School of Physics: General Relativity and Cosmology, 1971). Gen. Relativ. Gravit. 41, 581 (2009). https://doi.org/10.1007/s10714-009-0760-7
Ellis, G.F.R.: Relativistic cosmology: its nature, aims and problems. In: Bertotti, B., de Felice, F., Pascolini, A. (eds.) General Relativity and Gravitation, pp. 215–288. D. Reidel Publishing Co., Dordrecht (1984)
Ellis, G.F.R.: Inhomogeneity effects in cosmology. Class. Quantum Gravity 28, 164001 (2011). https://doi.org/10.1088/0264-9381/28/16/164001. arXiv:1103.2335
Ellis, G.F.R., Bruni, M., Hwang, J.: Density–gradient–vorticity relation in perfect-fluid Robertson–Walker perturbations. Phys. Rev. D 42, 1035 (1990). https://doi.org/10.1103/PhysRevD.42.1035
Ellis, G.F.R., Maartens, R., MacCallum, M.A.H.: Relativistic Cosmology. Cambridge University Press, Cambridge (2012). https://doi.org/10.1017/CBO9781139014403
Ellis, G.F.R., Stoeger, W.: The ‘fitting problem’ in cosmology. Class. Quantum Gravity 4, 1697 (1987). https://doi.org/10.1088/0264-9381/4/6/025
Fanizza, G., Gasperini, M., Marozzi, G., Veneziano, G.: Generalized covariant prescriptions for averaging cosmological observables. J. Cosmol. Astropart. Phys. JCAP1402, 017 (2020). https://doi.org/10.1088/1475-7516/2020/02/017. arXiv:1911.09469
Friedrich, H.: Evolution equations for gravitating ideal fluid bodies in general relativity. Phys. Rev. D 57, 2317 (1998). https://doi.org/10.1103/PhysRevD.57.2317
Gasperini, M., Marozzi, G., Veneziano, G.: Gauge invariant averages for the cosmological backreaction. J. Cosmol. Astropart. Phys. JCAP0903, 011 (2009). https://doi.org/10.1088/1475-7516/2009/03/011. arXiv:0901.1303
Gasperini, M., Marozzi, G., Veneziano, G.: A covariant and gauge invariant formulation of the cosmological “backreaction”. J. Cosmol. Astropart. Phys. JCAP1002, 009 (2010). https://doi.org/10.1088/1475-7516/2010/02/009. arXiv:0912.3244
Gasperini, M., Marozzi, G., Nugier, F., Veneziano, G.: Light-cone averaging in cosmology: formalism and applications. J. Cosmol. Astropart. Phys. JCAP1107, 008 (2011). https://doi.org/10.1088/1475-7516/2011/07/008. arXiv:1104.1167
Gourgoulhon, E.: \(3+1\) Formalism in General Relativity. Bases of numerical relativity. Lecture Notes in Physics, vol. 846. Springer, Berlin (2012)
Hawking, S.W., Ellis, G.F.R.: The Large Scale Structure of Spacetime. Cambridge Monographs on Mathematical Physics, pp. 88–95. Cambridge University Press, Cambridge (1973). https://doi.org/10.1017/CBO9780511524646
Heinesen, A., Mourier, P., Buchert, T.: On the covariance of scalar averaging and backreaction in relativistic inhomogeneous cosmology. Class. Quantum Gravity 36, 075001 (2019). https://doi.org/10.1088/1361-6382/ab0618. arXiv:1811.01374
Heinesen, A., Mourier, P.: (in preparation)
Jantzen, R.T., Carini, P., Bini, D.: The many faces of gravitoelectromagnetism. Ann. Phys. 215, 1 (1992). https://doi.org/10.1016/0003-4916(92)90297-Y. arXiv:gr-qc/0106043
Kasai, M., Asada, H., Futamase, T.: Toward a no-go theorem for an accelerating universe through a nonlinear backreaction. Progr. Theor. Phys. 115, 827 (2006). https://doi.org/10.1143/PTP.115.827. arXiv:astro-ph/0602506
King, A.R., Ellis, G.F.R.: Tilted homogeneous cosmological models. Commun. Math. Phys. 31, 209 (1973). https://doi.org/10.1007/BF01646266
Kolb, E.W.: Backreaction of inhomogeneities can mimic dark energy. Class. Quantum Gravity 28, 164009 (2011). https://doi.org/10.1088/0264-9381/28/16/164009
Korzyński, M.: Nonlinear effects of general relativity from multiscale structure. Class. Quantum Gravity 32, 215013 (2015). https://doi.org/10.1088/0264-9381/32/21/215013. arXiv:1412.3865
Krasiński, A.: Inhomogeneous Cosmological Models. Cambridge University Press, Cambridge (1997)
Larena, J.: Spatially averaged cosmology in an arbitrary coordinate system. Phys. Rev. D 79, 084006 (2009). https://doi.org/10.1103/PhysRevD.79.084006. arXiv:0902.3159
Li, Y.-Z., Mourier, P., Buchert, T., Wiltshire, D.L.: Lagrangian theory of structure formation in relativistic cosmology V: irrotational fluids. Phys. Rev. D 98, 043507 (2018). https://doi.org/10.1103/PhysRevD.98.043507. arXiv:1806.05467
Macpherson, H.J., Price, D.J., Lasky, P.D.: Einstein’s universe: cosmological structure formation in numerical relativity. Phys. Rev. D 99, 063522 (2019). https://doi.org/10.1103/PhysRevD.99.063522. arxiv:1807.01711
Magni, S.: Backreaction and the covariant formalism of general relativity. Master Thesis (2012). arXiv:1202.0430
Marozzi, G.: The cosmological backreaction: gauge (in)dependence, observers and scalars. J. Cosmol. Astropart. Phys. JCAP1101, 012 (2011). https://doi.org/10.1088/1475-7516/2011/01/012. arXiv:1011.4921
Massa, E.: Space tensors in general relativity III: the structural equations. Gen. Relativ. Gravit. 5, 715 (1974). https://doi.org/10.1007/BF00761928
Misner, C.W., Thorne, K., Wheeler, J.A.: Gravitation. Freedman, San Francisco (1973). https://doi.org/10.1002/asna.19752960110
Olson, D.W., Sachs, R.K.: The production of vorticity in an expanding, self-gravitating fluid. Astrophys. J. 185, 91 (1973). https://doi.org/10.1086/152399
Räsänen, S.: Backreaction: directions of progress. Class. Quantum Gravity 28, 164008 (2011). https://doi.org/10.1088/0264-9381/28/16/164008. arXiv:1102.0408
Räsänen, S.: Light propagation in statistically homogeneous and isotropic universes with general matter content. J. Cosmol. Astropart. Phys. JCAP1003, 018 (2010). https://doi.org/10.1088/1475-7516/2010/03/018. arXiv:0912.3370
Räsänen, S.: Light propagation and the average expansion rate in near-FLRW universes. Phys. Rev. D 85, 083528 (2012). https://doi.org/10.1103/PhysRevD.85.083528. arXiv:1107.1176
Roy, X.: On the \(1+3\) formalism in general relativity (2014). arXiv:1405.6319
Roy, X., Buchert, T.: Chaplygin gas and effective description of inhomogeneous universe models in general relativity. Class. Quantum Gravity 27, 175013 (2010). https://doi.org/10.1088/0264-9381/27/17/175013. arXiv:0909.4155
Roy, X., Buchert, T.: Relativistic cosmological perturbation scheme on a general background: scalar perturbations for irrotational dust. Class. Quantum Gravity 29, 115004 (2012). https://doi.org/10.1088/0264-9381/29/11/115004. arXiv:1202.5766
Roy, X., Buchert, T., Carloni, S., Obadia, N.: Global gravitational instability of FLRW backgrounds—interpreting the dark sectors. Class. Quantum Gravity 28, 165004 (2011). https://doi.org/10.1088/0264-9381/28/16/165004. arXiv:1103.1146
Russ, H., Soffel, M.H., Kasai, M., Börner, G.: Age of the universe: influence of the inhomogeneities on the global expansion factor. Phys. Rev. D 56, 2044 (1997). https://doi.org/10.1103/PhysRevD.56.2044. arXiv:astro-ph/9612218
Smarr, L., York, J.W.: Kinematical conditions in the construction of spacetime. Phys. Rev. D 17, 2529 (1978). https://doi.org/10.1103/PhysRevD.17.2529
Smirnov, J.: Gauge-invariant average of Einstein equations for finite volumes (2014). arXiv:1410.6480
Sussman, R.A.: Back-reaction and effective acceleration in generic LTB dust models. Class. Quantum Gravity 28, 235002 (2011). https://doi.org/10.1088/0264-9381/28/23/235002. arXiv:1102.2663
Sussman, R.A.: Weighed scalar averaging in LTB dust models, part I: statistical fluctuations and gravitational entropy. Class. Quantum Gravity 30, 065015 (2013). https://doi.org/10.1088/0264-9381/30/6/065015. arXiv:1209.1962
Sussman, R.A., Delgado Gaspar, I., Hidalgo, J.C.: Coarse-grained description of cosmic structure from Szekeres models. J. Cosmol. Astropart. Phys. JCAP1603, 012 (2016). https://doi.org/10.1088/1475-7516/2016/03/012. arXiv:1507.02306
Sussman, R.A., Hidalgo, J.C., Delgado Gaspar, I., Germàn, G.: Nonspherical Szekeres models in the language of cosmological perturbations. Phys. Rev. D 95, 064033 (2018). https://doi.org/10.1103/PhysRevD.95.064033. arXiv:1701.00819
Tanaka, H., Futamase, T.: A phantom does not result from a backreaction. Progr. Theor. Phys. 117, 183 (2007). https://doi.org/10.1143/PTP.117.183. arXiv:astro-ph/0612151
Umeh, O., Larena, J., Clarkson, C.: The Hubble rate in averaged cosmology. J. Cosmol. Astropart. Phys. JCAP1103, 029 (2011). https://doi.org/10.1088/1475-7516/2011/03/029. arXiv:1011.3959
Vigneron, Q., Buchert, T.: Dark matter from backreaction? Collapse models on galaxy cluster scales. Class. Quantum Gravity 36, 175006 (2019). https://doi.org/10.1088/1361-6382/ab32d1. arXiv:1902.08441
Wald, R.M.: General Relativity, pp. 218–220. University of Chicago Press, Chicago (1984)
Wiltshire, D.L.: Cosmic clocks, cosmic variance and cosmic averages. New J. Phys. 9, 377 (2007). https://doi.org/10.1088/1367-2630/9/10/377. arXiv:gr-qc/0702082
Wiltshire, D.L.: What is dust?—Physical foundations of the averaging problem in cosmology. Class. Quantum Gravity 28, 164006 (2011). https://doi.org/10.1088/0264-9381/28/16/164006. arXiv:1106.1693
Wiltshire, D.L.: Cosmic structure, averaging and dark energy, In: Novello, M., Perez Bergliaffa, S.E. (eds.) Cosmology and gravitation: XVth Brazilian School of Cosmology and Gravitation. Cambridge Scientific Publishers, Cambridge, ISBN 1908106395, 9781908106391, pp. 203–244 (2014). arXiv:1311.3787
Acknowledgements
This work is part of a project that has received funding from the European Research Council (ERC) under the European Union’s Horizon 2020 research and innovation programme (grant agreement ERC advanced grant 740021–ARTHUS, PI: T. Buchert). PM was supported by a ‘spécifique Normalien’ Ph.D. grant for most of this work, and is grateful to the Erwin Schrödinger Institute’s 2017 Summer school ‘Between Geometry and Relativity’ in Vienna for providing insights that were of use for this work. XR acknowledges support from the Claude Leon Foundation when this work started in 2012 at the Cosmology and Gravity group of the University of Cape Town. XR thanks Chris Clarkson and Obinna Umeh for fruitful discussions and also acknowledges hospitality at the Centre de Recherche Astrophysique de Lyon, and thanks TB for his invitations. We wish to thank Fosca Al Roumi for valuable comments on an earlier manuscript, Léo Brunswic, Mauro Carfora, Asta Heinesen, Étienne Jaupart, Jan Ostrowski, Roberto Sussman, Nezihe Uzun, Quentin Vigneron, and David Wiltshire for many valuable discussions, and Julien Larena, Syksy Räsänen for useful comments on the final manuscript.
Author information
Authors and Affiliations
Corresponding author
Additional information
Publisher's Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Work supported by ERC advanced Grant 740021–ARTHUS.
Appendices
A \(3+1\) evolution equations along the congruence of the fluid
The evolution equations for \(h_{ij}\) and \({\mathcal {K}}^i_{ j}\) along the congruence of the fluid are obtained from expressions (3.1) and (3.2), by relating the derivative \(\partial _t |_{x^i}\) to \({\mathrm{d}}/{\mathrm{d}}t\) with the help of (2.25). They read:
Comoving coordinate system. In the comoving picture, as described in Sect. 2.4, we have \({\varvec{N}} = N {\varvec{v}}\), or equivalently \({\varvec{V}} = {\varvec{0}}\). Equations (A.1) and (A.2) hence read:
For these equations, focusing on the local foliation-related variables \({\mathbf {h}}\), \({\mathcal {K}}^i_{\ j}\) rather than fluid variables, the additional assumption of a fluid proper time foliation and of the choice \(t=\tau \) (implying \(N=\gamma \)), corresponding to the Lagrangian picture, does not affect the comoving picture equations (A.3)–(A.4) above, except for the possibility of replacing each occurrence of N by \(\gamma \). We shall accordingly not rewrite the equations for this case.
The local equations of evolution along the fluid flow for arbitrary slices and coordinates (A.1)–(A.2) allow for an alternative derivation of the coordinate-time derivative of the extrinsic volume, Eq. (3.17):
by starting over from (3.14) and expanding its integrand as
The trace of (A.1) can then be used together with (3.8) to obtain:
This expression then allows to catch up with the end of the derivation given in Sect. 3.2.2, so that a similar use of relations (3.16) and (2.10) again gives the evolution of the volume (A.5).
B Extrinsic averaging operator in fluid-intrinsic variables
The system of equations for extrinsic averages on \({\mathcal {D}}\) derived in Sect. 3 (Theorem 1) is mostly expressed in terms of geometric variables of the \({\varvec{n}}\)-orthogonal hypersurfaces, such as their extrinsic curvature. We here present an alternative formulation of the same equations focusing instead on the intrinsic, rest-frame kinematic and dynamical quantities of the fluid (see Sect. 2.2.2) which do not depend on the foliation choice.
We can first rewrite the volume expansion rate (3.19) and the commutation rule (3.22) in terms of the intrinsic local expansion rate of the fluid \(\varTheta \) by re-expressing the three-divergence of \(N {\varvec{v}}\) as
where we have employed (2.6) for the last equality. Noticing that \({\mathcal {K}}= - \nabla _\mu n^\mu \) and making use of expression (2.7), we get:
where we have defined a tilted and scaled expansion rate\({\tilde{\varTheta }}^T\) out of the scaled rate \({\tilde{\varTheta }}= (N/\gamma ) \, \varTheta \). The factor \(N / \gamma \) in \({\tilde{\varTheta }}^T\) above adjusts the local clock rates between the proper time of the fluid and the coordinate time. This can also be seen upon writing:
where we have used the relation (2.26) between \(\mathrm {d}/\mathrm {d}t\) and \(\mathrm {d}/\mathrm {d}\tau \). The additional tilt term \(- \gamma ^{-1} \, \mathrm {d}\gamma / \mathrm {d}t\) can be understood as the effect of the evolving mutual tilt (Lorentz boost) between the hypersurfaces in which \({\mathcal {D}}\) is embedded, and the fluid flow. This affects the local volume measure so that the evolution of the volume is not only due to the fluid’s intrinsic expansion.
The above rewriting (B.2) allows us to recast the volume and scale factor evolution rates, and the commutation rule, respectively, into the following expressions:
We notice that, even for the general configuration we are investigating (see Fig. 1), the commutation rule, as well as the domain volume expansion rate, can be cast into a simple form with respect to the fluid quantities, although this extrinsic averaging framework requires the explicit additional contribution from the evolving tilt.
The use of the Raychaudhuri equation (4.14) and the energy constraint (4.16) (instead of the scalar parts of the extrinsic \(3+1\) Einstein equations (3.2), (3.4)), together with the above alternative form of the commutation rule, allows for a rewriting of the evolution equations for the extrinsic effective scale factor \(a_{\mathcal {D}}\). This yields the following equivalent formulation of Theorem 1, in terms of rescaled fluid-intrinsic kinematic and dynamical variables, \({\tilde{\sigma }}^2 = (N^2/\gamma ^2) \, \sigma ^2\), \({\tilde{\omega }}^2 = (N^2/\gamma ^2) \, \omega ^2\), \(\tilde{{\mathscr {R}}}= (N^2/\gamma ^2) \, {\mathscr {R}}\), \({\tilde{\epsilon }}= (N^2/\gamma ^2) \, \epsilon \), \({\tilde{p}} = (N^2/\gamma ^2) \, p\), and \(\tilde{{\mathcal {A}}} = (N^2/\gamma ^2) \, {\mathcal {A}}\) (with \({\mathcal {A}}= \nabla _\mu a^\mu \)), as well as \({\tilde{{\Lambda }}} := (N^2/\gamma ^2) \,{\Lambda }\):
Corollary 1a
(extrinsically averaged evolution equations in fluid variables) The averaged evolution equations for the extrinsic effective scale factor \(a_{\mathcal {D}}\) can also be written under the following form:
with alternative, ‘tilted’ kinematical and dynamical backreactions, reading respectively:
We recall that, as in Theorem 1a, the left-hand sides in the above equations should be seen as derivatives with respect to the chosen parameter t, and be interpreted according to the physical meaning of the latter. In particular, the term \(3 \, a_{\mathcal {D}}^{-1} \, \mathrm {d}^2 a_{\mathcal {D}}/ \mathrm {d}t^2\) in equation (B.7) corresponds in a Lagrangian picture to the proper time scale factor acceleration, but not in general.
Under this form, only two backreaction terms appear, \({\tilde{{\mathcal {Q}}}}^{\mathrm{T}}_{\mathcal {D}}\) and \({\tilde{{\mathcal {P}}}}^{\mathrm{T}}_{\mathcal {D}}\), as the tilt only contributes under these combinations. These backreaction terms will not in general be directly related to the terms \({\mathcal {Q}}_{\mathcal {D}}\) and \({\mathcal {P}}_{\mathcal {D}}\) appearing in Theorem 1a, as they do not collect the same local terms in their expressions. They do coincide, however, for a fluid-orthogonal foliation assuming an irrotational fluid (with in this case \({\tilde{{\mathcal {Q}}}}^{\mathrm{T}}_{\mathcal {D}}= {\mathcal {Q}}_{\mathcal {D}}\) and \({\tilde{{\mathcal {P}}}}^{\mathrm{T}}_{\mathcal {D}}= {\mathcal {P}}_{\mathcal {D}}\), while \({\mathcal {T}}_{\mathcal {D}}= 0\)).
Note that there is no explicit non-perfect fluid contribution to these effective evolution equations, although the non-perfect fluid components of the energy-momentum tensor do have an influence on the dynamics via the local (and average, see below) evolution of \(\epsilon \) and p.
As before, this set of equations goes along with an integrability condition, and must be complemented by the evolution equation for the averaged energy density and pressure.
Corollary 1b
(integrability and energy balance conditions to Corollary 1a) The integrability condition corresponding to Corollary 1a reads:
while the associated averaged conservation equation for the scaled energy density \({\tilde{\epsilon }}\) and pressure \({\tilde{p}}\) becomes:
(From the commutation rule, the expression \(\langle {{\tilde{\varTheta }}^T}\rangle _{\mathcal {D}}\left\langle {{\tilde{p}}} \right\rangle _{{\mathcal {D}}} - \langle {{\tilde{\varTheta }}^T {\tilde{p}}}\rangle _{\mathcal {D}}\) can also be written as \(\left\langle {{\mathrm{d}} {\tilde{p}} / \mathrm {d}t} \right\rangle _{{\mathcal {D}}} - \mathrm {d}\left\langle {{\tilde{p}}} \right\rangle _{{\mathcal {D}}}/\mathrm {d}t\).)
C Summarized literature comparison
We present in Table 1 a comparative overview of the various formalisms used in the existing generalization proposals of the system of averaged scalar equations of Papers I and II to general foliations, discussed in Sects. 3.5.4–3.5.6, including the extrinsic averaging formalism of Sect. 3 of this work.
In this table we express all notations in terms of those used in this work to make comparisons easier. In particular, when considering the 4-scalar expressions of [53, 74, 82], we define the lapse N as \((- \partial _\mu T \, \partial ^\mu T)^{-1/2}\), where T is the scalar function which labels the hypersurfaces. This quantity (noted \({\Gamma }\) in [74]) satisfies \(n_\mu = - N \partial _\mu T\), thus playing an analogous role to the \(3+1\) lapse, and it indeed coincides with the usual lapse if the 4-scalar formalism is split into a \(3+1\) description with the natural choice of T as the time coordinate. Since the domain propagation varies between these papers, in terms of the present notations the averaging domain should be noted \({\mathcal {E}}_{{\varvec{\partial }}_t}\) or \({\mathcal {E}}_{{\varvec{n}}}\), and \({\mathcal {D}}\) for the present paper. Rather than indicating the corresponding domain for each average and backreaction term, we remove any domain-labelling subscript for these objects in the table for notational ease.
Additional specificities of some of the papers compared here:
-
Tanaka and Futamase [87]: The commutation rule obtained in the corresponding framework is not explicitly provided or derived. It thus remains implicit as a necessary intermediate step for the derivation of the evolution equations for the effective scale factor from averages of the local dynamical equations. We note, moreover, that the main aim of the paper, as for [60], is to set up a link to a background-dependent perturbation approach.
-
Räsänen [74]: The velocity field \({\varvec{u}}\) that is introduced in addition to \({\varvec{n}}\) is fully general and is not related either to \({\varvec{n}}\) nor to the content of the model universe (it could be chosen to be the 4-velocity of some fluid as in the present approach, but this would be a restriction of generality). It is supposed to represent the 4-velocity field of the observers. In the application paper [75], this field is restricted to be everywhere very close to \({\varvec{n}}\) (and so has a small vorticity), whereas \({\varvec{n}}\) is assumed to be chosen such that it builds hypersurfaces of statistical homogeneity and isotropy. These restrictions are already both suggested in the original paper [74] but the equations are kept general.
-
Beltrán Jiménez et al. [8]: The main objective of using a general \(T_{\mu \nu }\) in this paper is to account for theories beyond General Relativity whose differences are transferred into \(T_{\mu \nu }\) as effective terms. Note also that this paper features an additional average equation giving the evolution of the average squared rate of shear \(\partial _t \left\langle {\sigma ^2} \right\rangle \), as well as the corresponding local equation; these equations are absent from the other papers quoted in this Appendix, including the present work (the reason being that the resulting system is still not closed by adding this equation; work about looking deeper into the hierarchy of equations will be published in [15], where the next level of the hierarchy is explored through a topological approach).
-
Smirnov [82]: Not only the choice of hypersurfaces (or of \({\varvec{n}}\)) and the choice of the time that parametrizes them obey specific criteria in this paper, but this is also the case for the averaging domain, although this is not explicit in the averaged equations and it could as well be any domain propagated along the chosen \({\varvec{n}}\). Indeed, the domain is there chosen as a ‘sphere’ in some \({\varvec{n}}\)-comoving coordinates \(Z^i\) on the hypersurfaces, as defined by \(H_{i j} Z^i Z^j \le r_0\) for some \(r_0 > 0\) and with \(H_{i j}\) the components of the spatial metric in these coordinates. This choice was a response to the series of papers of Gasperini et al. and Marozzi [52, 53, 69] to show how one may determine the boundary of the domain via a scalar function (here in the sense that the \(Z^i\) are fixed a priori without any link to the actual spacetime coordinates choices).
-
In the present work, we also introduce, in Sect. 4, a different (intrinsic) averaging formalism that measures scalar quantities and volume in the local rest frames of the fluid, even if they are integrated over a domain lying in the not necessarily fluid-orthogonal hypersurfaces. We then obtain the corresponding commutation rule and averaged dynamics under rather simple forms in terms of the intrinsic kinematic and dynamical quantities of the fluid —for instance, the effective Hubble parameter, still defined as 1/3 of the volume expansion rate, can be simply expressed as the average of \((N/\gamma ) \, (\varTheta /3)\)— and only two backreactions, kinematical and dynamical, distinct in general from the terms \({\mathcal {Q}}_{\mathcal {D}}\), \({\mathcal {P}}_{\mathcal {D}}\) of Sect. 3. This formalism and this system of equations clearly differ from the literature compared in this Appendix (including our Sect. 3, although it otherwise follows the same setup), due to the different volume and averaging operator definition.
We also give in the present work, in Appendix B, a re-expression of the averaged equations arising from the extrinsic averaging operator of Sect. 3, in terms of averages of fluid-intrinsic kinematic and dynamical variables. This is obtained at the expense of the appearance of additional contributions from the evolution of the tilt factor.
D Remarks on volume 3-forms and manifestly covariant rewritings
Let us consider the two different volume 3-forms obtained as the respective Hodge duals \(\star {{{{\underline{\varvec{n}}}}}}\) , \(\star {{{{\underline{\varvec{u}}}}}}\) to the 1-forms \({{{{\underline{\varvec{n}}}}}} = n_\mu \, {\mathbf {d}}x^\mu \) (metric-dual to the normal vector to the slices \({\varvec{n}}\)) and \({{{{\underline{\varvec{u}}}}}} = u_\mu \, {\mathbf {d}}x^\mu \) (metric-dual to the fluid 4-velocity vector \({\varvec{u}}\)). We can decompose these 3-forms as follows in the exact basis \(({\mathbf {d}}x^\mu ) = ({\mathbf {d}}t,{\mathbf {d}}x^i)\), associated with the reference coordinates \((t,x^i)\):
where \(\varepsilon _{\mu \nu \rho \sigma }\) denote the components of the four-dimensional Levi-Civita tensor, while \(\epsilon _{ijk}\) is simply the Levi-Civita symbol with 3 (spatial) indices. We have used above the components expressions (2.1) and (2.14) of \({\varvec{n}}\) and \({\varvec{u}}\), respectively, and we have related the modulus g of the 4-metric determinant to the spatial determinants h and bvia\(h = g/N^2\) (see, e.g., [55, sec. 5.2.3], or [53, sec. 3]) and \(b = \gamma ^2 h = (\gamma /N)^2 g\) (from Eq. (3.23)). Note that in the expression for \(\star {{{{\underline{\varvec{n}}}}}}\), \({\mathbf {d}}x^i + N^i {\mathbf {d}}t\) is the hypersurface-projected element \(h^i_{\ \mu }\, {\mathbf {d}}x^\mu \), related to the rewriting of the four-dimensional line element (2.5) in all generality as \(\mathrm {d}s^2 = - N^2 \mathrm {d}t^2 + h_{ij} (\mathrm {d}x^i + N^i \mathrm {d}t) (\mathrm {d}x^j + N^j \mathrm {d}t)\).
The restrictions of these 3-forms to the tangent spaces to the spatial hypersurfaces thus read \(( \star {{{{\underline{\varvec{n}}}}}} )_{|{\Sigma }} = \sqrt{h} \, \mathrm {d}^3 x\) and \(( \star {{{{\underline{\varvec{u}}}}}} )_{|{\Sigma }} = \sqrt{b} \, \mathrm {d}^3 x\), respectively, since the integration element on the slices \(\mathrm {d}^3 x\) previously introduced corresponds to \((1 / 6) \, \epsilon _{ijk} \, ({\mathbf {d}}x^i \wedge {\mathbf {d}}x^j \wedge {\mathbf {d}}x^k )_{|{\Sigma }}\). We thus get, for any scalar \(\psi \), using here again the label \({}^h\) for the extrinsic volume and averages for a clearer distinction:
the \(\psi = 1\) case giving the volumes \({\mathcal {V}}_{\mathcal {D}}^h\) and \({\mathcal {V}}_{\mathcal {D}}^b\). We thus recover under this form the extrinsic and intrinsic operators of Sects. 3 and 4 as natural averages (and volumes) definitions based, respectively, on the hypersurface normal, or on the fluid 4-velocity.
The similar roles played by \({\varvec{n}}\) in the extrinsic averaging formalism and by \({\varvec{u}}\) in the intrinsic formalism can alternatively be highlighted by rewriting the corresponding volumes and averages under a manifestly 4-covariant form using a 4-scalar window function [53, 57]. The framework of Gasperini et al. [53] indeed gives a manifestly covariant form of the extrinsic volume and of the associated averaging operator as (for any scalar \(\psi \)):
respectively. Here, \({\mathfrak {W}}_{\mathcal {D}}^h\) is the window function selecting the averaging domain:
where \({\mathcal {H}}\) is the Heaviside step function. The scalar functions A, B respectively define the foliation and the four-dimensional tube spanned by the domain, with the parameters \(A_0\), \(r_0\) selecting respectively a specific slice as \(\{ x^\alpha \vert \; A(x^\alpha ) = A_0 \}\) and the domain boundaries from the condition \(B(x^\alpha ) \le r_0\). (In [53], a hypersurface-orthogonal propagation is then set for the averaging domain; this is done by requiring \(n^\mu \partial _\mu B = 0\).) As pointed out and analyzed in more detail in [57], the fluid volume (4.5) and the intrinsic averager (4.6) can be written under the same form by simply replacing \({\varvec{n}}\) by \({\varvec{u}}\) in the window function, i.e., by replacing the window function \({\mathfrak {W}}_{\mathcal {D}}^h\) by a window function \({\mathfrak {W}}_{\mathcal {D}}^b\) in (D.4), with
With the addition of the constraint \(u^\mu \partial _\mu B = 0\), which defines a comoving domain propagation, this yields the same intrinsic averaging framework as that considered in our Sect. 4 above.
E Remark on the threading lapse
We here add a technical remark. In the fluid-intrinsic approach we can borrow one element from the \(1+3\) formalism to foliate spacetime, the so-called spacetime threading, although spatial volume averaging only makes sense on hypersurfaces. We recall that in the \(1+3\) decomposition, and in comoving spatial coordinates, the four-dimensional line element (2.33) reads (see, e.g., [59, sec. 10]):
with \({\mathcal {M}}\) the threading lapse and \(\underline{\varvec{{\mathcal {M}}}}\) the threading shift, which relate to the lapse and to the Eulerian velocity dual 1-form \({{{{\underline{\varvec{v}}}}}}\) as follows:
In the Lagrangian description we have:
Note that in the intrinsic form of the general averaged equations (see Sect. 4.2), we only deal with occurrences of \(N / \gamma ={\mathcal {M}}\).
F Erratum
We wish to point out a small mistake in Paper II [19] (repeated in the appendix of [22] after equations (A23) and (A28) therein). For this we recall the spatial components of the 4-acceleration, \(a_i = N_{|| i} / N\), its 4-divergence \({\mathcal {A}} := \nabla _{\mu }a^{\mu } = a^i_{|| i} + a^i a_i\), and the expression of the latter in terms of the lapse N or the injection energy per fluid element h (related to the relativistic enthalpy),
which are correctly written (for such a fluid-orthogonal framework). However, the first equality in equation (10a) of Paper II (and of its review in [22]) is incorrect, \({{{\mathcal {A}}}} \ne \left( N^{|| i} /N \right) _{|| i}\), due to an omission of the \(a^i a_i\) contribution to \({\mathcal {A}}\).
There is also an imprecise statement: in Paper II, in footnote 3, it is stated that for scalars the operator || amounts to a partial derivative. This statement is only true for spatial components (for a scalar, \({ }_{||i} = \partial _i\), but \({ }_{||0} {} = N^i \partial _i \ne \partial _t\); \({ }_{||0}\) was identically zero for scalars in Paper II due to the vanishing shift).
Rights and permissions
About this article
Cite this article
Buchert, T., Mourier, P. & Roy, X. On average properties of inhomogeneous fluids in general relativity III: general fluid cosmologies. Gen Relativ Gravit 52, 27 (2020). https://doi.org/10.1007/s10714-020-02670-6
Received:
Accepted:
Published:
DOI: https://doi.org/10.1007/s10714-020-02670-6