Title

Radouane Gannouji,1, ∗ Jonas Pinheiro da Silva,2, † and Hermano Velten3, ‡

1
Instituto de Fı́sica, Pontificia Universidad Católica de Valparaı́so
Av. Brasil 2950, Valparaı́so, Chile
2
PPGCosmo, CCE, Universidade Federal do Espı́rito Santo (UFES)
Av. Fernando Ferrari, 540, CEP 29.075-910, Vitória, ES, Brasil.
3
Departamento de Fı́sica, Universidade Federal de Ouro Preto (UFOP),
Campus Universitário Morro do Cruzeiro, 35.402-136, Ouro Preto, Brazil
(Dated: January 14, 2025)
...

I. INTRODUCTION

We consider the following action

√
Z
1
S= d4 x −gf (R, T ) + Sm + Sr (1)
2κ2
where R is the curvature scalar, T is the trace of the energy-momentum tensor and Sm , Sr represent respectively the
non-relativistic and the relativistic contributions. Of course the variation should not assume the equations of motion
and therefore we should consider T to be generic, but it is more natural to consider that only the non-relativistic part
couples non-minimally. Following this principle, we have

δT = δ(g µν Tµν
(m)
) = g µν δTµν
(m) (m) µν
+ Tµν δg (2)

Also we assume that the non-relativistic energy-momentum tensor depends only on the metric
(m)
(m) δTµν
δTµν = δg αβ (3)
δg αβ
which implies
(m)
δTµν 
(m) (m)

δT = g µν δg αβ
+ T (m) µν
µν δg ≡ θ αβ + Tαβ δg αβ (4)
δg αβ
where we have defined
(m)
(m) δTµν
θαβ = g µν (5)
δg αβ
Performing the variation of (1) wrt the metric gives
1 
(m) (m)

f,R Rµν − f gµν + gµν □f,R − ∇µν f,R + f,T θµν + Tµν = κ2 Tµν (6)
2
Taking the covariant derivative of this equation and using the Bianchi identity as well as

∇ν □f,R − □∇ν f,R = Rµν ∇µ f,R (7)

we find
h  i 1
κ2 ∇µ Tµν = ∇µ f,T θµν
(m) (m)
+ Tµν − f,T ∇ν T (m) (8)
2

∗ [email protected]
† [email protected]
‡ [email protected]
 2

Consistently with our previous assumption, only the non-relativistic part should not be conserved, namely
∇µ Tµν
(r)
=0 (9)
h  i 1
∇µ κ2 Tµν(m) (m)
− f,T θµν (m)
+ Tµν = − f,T ∇ν T (m) (10)
2
(m)
In order to find the Friedmann equations we need to find the expression of θµν . We have different approaches reaching
(m) (m)
different expressions. See the discussion in the appendix. We will assume θµν = −2Tµν .

II. EQUATIONS OF MOTION

We consider the FLRW metric ds2 = −dt2 + a(t)2 dx2 , the equations are
Rf,R − f
3H 2 f,R = κ2 (ρm + ρr ) − T f,T + − 3H f˙,R (11)
2
 4 
−2Ḣf,R = κ2 ρm + ρr − T f,T − H f˙,R + f¨R (12)
3
and the equations of conservation are
ρ̇r + 4Hρr = 0 (13)
f,T
(κ2 + f,T )(ρ̇m + 3Hρm ) = −ρm f˙,T − ρ̇m (14)
2
We will study the case f (R, T ) = R + f (T ) so the equations reduce to
f
3H 2 = κ2 (ρm + ρr ) − T f,T − (15)
2
2
 4 
−2Ḣ = κ ρm + ρr − T f,T (16)
3
We define the variables
κ2 ρm κ2 ρr T f,T f
Ωm = , Ω r = , x1 = − , x2 = − , (17)
3H 2 3H 2 3H 2 6H 2
which transforms the Friedmann equations into
Ω m = 1 − Ωr − x 1 − x 2 (18)
Ḣ
−2 = 3Ωm + 4Ωr + 3x1 = 3 + Ωr − 3x2 (19)
H2
and the conservation equations into
ρ̇r + 4Hρr = 0 (20)
2
κ + f,T
ρ̇m + 3H ρm = 0 (21)
κ2 + 32 f,T + T f,T T
We see that if ρr has the standard evolution, ρm grows in a Universe with an effective Hubble function
κ2 + f,T
Hef f = H (22)
κ2 + 32 f,T + T f,T T
The autonomous system is
dΩr  
= Ωr Ωr − 3x2 − 1 (23)
d ln a
dx1  
= x1 3 + Ωr − 3x2 − 3h(1 + m) (24)
d ln a
dx2   3
= x2 3 + Ωr − 3x2 − x1 h (25)
d ln a 2
 3

with the normalized effective Hubble function

κ2 + f,T
h≡ (26)
κ2 + 32 f,T + T f,T T

and
T f,T T
m= (27)
f,T

which are function of

x1 T f,T
r≡ =− (28)
2x2 f
Finally, we have

2Ḣ Ωr
weff = −1 − = − x2 (29)
3H 2 3
The critical points are

Label x1 x2 Ωr Ωm weff Nature

P1 0 0 0 1 0 matter
P2 0 0 1 0 1/3 radiation
P3 0 1 0 0 −1 de-Sitter
P4 2x2 (1 + m) x2 0 1 − (3 + 2m)x2 −x2 -

The P4 point is defined for specific models. For a given model, we can obtain m(r) and this point needs to be
solution of the equation m(r) = −r − 1.
In the monomial case f (T ) = αT n , the system is simpler because in that case x1 = 2nx2 , m = n − 1, r = −1 so we
need a new variable instead of r, we will use x2 /Ωm , so we have

Ωm + 2nx2 1 − x 2 − Ωr
h= = (30)
Ωm + n(2n + 1)x2 1 + (n − 1)(2n + 1)x2 − Ωr

which reduces the system to

dΩr  
= Ωr Ωr − 3x2 − 1 (31)
d ln a
dx2  
= x2 3 + Ωr − 3x2 − 3nx2 h (32)
d ln a
 4

Appendix A: Harko et al. approach

In [1], the authors considered the definition of energy momentum tensor

√
2 δ −gLm δLm
Tµν = −√ = gµν Lm − 2 µν (A1)
−g δg µν δg

From which we can obtain

δTαβ δ 2 Lm
θµν = g αβ µν
= −2Tµν + gµν Lm − 2g αβ µν αβ (A2)
δg δg δg

They assume that the matter is derived from Lm = p where p is the pressure. It seems that they assume that the
last term is zero, which I am unconvinced. But assuming this hypothesis, they get

θµν = −2Tµν + gµν p (A3)

and therefore for non-relativistic matter, we get

(m) (m)
θµν = −2Tµν (A4)

Appendix B: Lm = p approach

Let us assume Harko et al. approach but with more details. In order to derive the energy-momentum tensor, we
need to follow the approach of [2]. We consider the action as

√
Z
Smat = d4 x −g p(µ, s) (B1)

The pressure depends on 2 parameters, the relativistic enthalpy per unit of baryon density, µ or chemical potential
and the entropy per baryons s = S/N with N the number of baryons. The relativistic enthalpy is defined as ρ + p
which gives
ρ+p
µ= (B2)
n
where ρ is the energy density and p the pressure. The total energy E is related to the internal energy U as E = U +N m,
which implies that dU = dE because the number of baryons N is conserved. Therefore the first law of thermodynamics
can be written

dU = dE = T dS − pdV (B3)

Using that V = N/n, E = ρV = ρN/n and S = N s, we obtain from the first law
ρ p
dρ − dn = T nds + dn (B4)
n n
From eq. (B2) we obtain

dρ = −dp + ndµ + µdn (B5)

which gives for the first law

dp = ndµ − T nds (B6)

which gives the pressure as a function of (µ, s). That expression is important because we get

∂p ρ+p
=n= (B7)
∂µ µ
 5

In order to obtain the energy-momentum tensor, [2] considers that the chemical potential depends on various potentials
that we vary. Most exactly, it is assumed that the 4-velocity is
1 
uµ = ∂µ ϕ + α∂µ β + θ∂µ s (B8)
µ
where (α, β) are associated to rotational movement. Because the 4-velocity is normalized uµ uµ = −1, we obtain
  
µ2 = −g µν ∂µ ϕ + α∂µ β + θ∂µ s ∂ν ϕ + α∂ν β + θ∂ν s (B9)

from which we obtain

δµ µ
µν
= − uµ uν (B10)
δg 2
The entropy is independent of the metric but the chemical potential is dependent. Therefore we get, if we do the
variation wrt the metric
√ h 1
Z
∂p δµ i
δSm = d4 x −g − gµν pδg µν + µν
δg µν (B11)
2 ∂µ δg
√ h
Z
1 i
=− d4 x −g gµν p + nµuµ uν δg µν (B12)
2
which implies
2 δSm
Tµν = − √ = nµuµ uν + pgµν = (ρ + p)uµ uν + pgµν (B13)
−g δg µν
In this formalism, we have

δ 2 Lm δ  
µν αβ
= µν nµuα uβ (B14)
δg δg δg
Of course, the velocity depends on the metric because of the normalization condition, we have from g µν uµ uν = −1
1
uµ δuµ = − uµ uν δg µν (B15)
2
from which we get
1
δuµ = uα uβ uµ δg αβ + Aµ , with uµ Aµ = 0 (B16)
2
We will assume Aµ = 0, I don’t know why. So we get

δ 2 Lm δ   δµ δn δuα δuβ
= nµuα uβ = nuα uβ µν + µuα uβ µν + µnuβ µν + µnuα µν (B17)
δg µν δg αβ δg µν δg δg δg δg
µ δn
= − nuα uβ uµ uν + µuα uβ µν + µnuα uβ uµ uν (B18)
2 δg
µ δn
= nuα uβ uµ uν + µuα uβ µν (B19)
2 δg
Finally, we need to calculate the last term. For that we know that the conserved flux of particles is jµ = nuµ . I don’t
know why but I will assume that it is independent of the metric, so δjµ = 0 which implies uµ δn = −nδuµ from which
we get

δn = nuµ δuµ (B20)

and from the normalization condition uµ uµ = −1, we have

1
uµ δuµ = − uµ uν δg µν (B21)
2
 6

which gives from (B20)

n
δn = − uµ uν δg µν (B22)
2
Going back to (B17), we obtain

δ 2 Lm
=0 (B23)
δg µν δg αβ
From which we can conclude
δ 2 Lm
θµν = −2Tµν + gµν Lm − 2g αβ (B24)
δg µν δg αβ
= −2Tµν + gµν p (B25)

which means
(m) (m)
θµν = −2Tµν (B26)

Appendix C: Lm = ρ approach

We can use an other approach with the density. For simplicity, we will not consider entropy. The action takes the
form
√ h
Z i
Sm = − d4 x −g ρ(n) + j µ ∂µ ϕ (C1)

where ρ is the total energy density, n = N/V is the particle number density and the particle current j µ = nuµ as we
defined before. We will consider that we have 3 fields, the metric gµν , the current j µ and the field ϕ. Variation with
respect to ϕ gives the conservation of the baryon number
√ 
∂µ −gj µ = 0 ⇔ ∇µ j µ = 0 (C2)

Because of the normalization condition of the 4-velocity, we have n2 = −gµν j µ j ν implying

1 n
δn = − jµ δj µ + uµ uν δg µν (C3)
n 2
The variation wrt j µ gives

ρ′ (n)
∂µ ϕ = jµ (C4)
n
And finally, the variation wrt the metric is
δSm 1 √   1√
= gµν −g ρ(n) + j α
∂α ϕ − −gnρ′ (n)uµ uν (C5)
δg µν 2 2
1 √   1√
= gµν −g ρ(n) − nρ′ (n) − −gnρ′ (n)uµ uν (C6)
2 2
From eq.(B4) we have

∂ρ ρ+p
= (C7)
∂n n
which gives
δSm 1 √ 1√
= − gµν −gp − −g(ρ + p)uµ uν (C8)
δg µν 2 2
 7

which gives

2 δSm
Tµν = − √ = (ρ + p)uµ uν + pgµν (C9)
−g δg µν

In this approach, we have

 
Lm = − ρ(n) + j µ ∂µ ϕ (C10)

which gives us

δLm 1
= − (ρ + p)uα uβ (C11)
δg αβ 2

from which we get

δ 2 Lm 1hn ′ ′
i
= − (ρ (n) + p (n)) + ρ + p uµ uν uα uβ (C12)
δg µν δg αβ 2 2
1 1
h  dp  i
=− 1+ + 1 (ρ + p)uµ uν uα uβ (C13)
2 2 dρ
1 dp 
=− 3+ (ρ + p)uµ uν uα uβ (C14)
4 dρ

implying

δ 2 Lm 1 dp 
g αβ = 3 + (ρ + p)uµ uν (C15)
δg µν δg αβ 4 dρ

and finally

1 dp 
θµν = −2Tµν + gµν Lm − 3+ (ρ + p)uµ uν (C16)
2 dρ

with on-shell
   
Lm = − ρ(n) + j µ ∂µ ϕ = − ρ(n) − nρ′ (n) = p (C17)

and the final expression is

1 dp 
θµν = −2Tµν + gµν p − 3+ (ρ + p)uµ uν (C18)
2 dρ

which implies

(m) (m) 3 7 (m)

θµν = −2Tµν − ρuµ uν = − Tµν (C19)
2 2

Appendix D: Tµν approach

(m)
We could also try to do the calculations directly, from Tµν = ρuµ uν . We know that
(m)
(m) δTµν
θαβ = g µν (D1)
δg αβ

and the density as well as the velocity are function of the baryon number density, n. As we have seen
ρ+p ρ
ρ′ (n) = = (D2)
n n
 8

and
1
δuµ = uα uβ uµ δg αβ (D3)
2
so we have
(m)
(m) δTµν 
′ δn   δ ln n 
θαβ = g µν = g µν
ρ (n) + ρu u
α β u u
µ ν = −ρ + u u
α β (D4)
δg αβ δg αβ δg αβ

We have seen two different results. If we assume that the only field is the metric, we obtain (B22)

δ ln n 1
αβ
= − uα uβ (D5)
δg 2

but if we assume that the fields are metric and current j µ , we have (C3)

δ ln n 1
= uα uβ (D6)
δg αβ 2
So depending on the approach, we obtain

(m) 1 (m) (m) 3 (m)

θµν = − Tµν , or θµν = − Tµν (D7)
2 2
Finally, if we do not assume only matter, so we start with Tµν = (ρ + p)uµ uν + pgµν , we obtain

δTµν h
′ ′ ′
i δn
= (ρ (n) + p (n))uµ uν + p (n)gµν + (ρ + p)uµ uν uα uβ − pgµα gνβ (D8)
δg αβ δg αβ
which gives
 dp  δ ln n
θµν = −(ρ + p)uµ uν − pgµν + 3 − 1 (ρ + p) µν (D9)
dρ δg
which gives for non-relativistic matter

(m) δ ln n
θµν = −ρuµ uν − ρ (D10)
δg µν

which is similar to (D4) and at least in that case we can assume matter previous to the derivation.
In conclusion, these models need to have a special prescription on the matter sector which doesn’t seem to be
physical. Why should we use Lm = p in particular.

[1] T. Harko, F. S. N. Lobo, S. Nojiri, and S. D. Odintsov, f (R, T ) gravity, Phys. Rev. D 84, 024020 (2011), arXiv:1104.2669
[gr-qc].
[2] B. F. Schutz, Perfect Fluids in General Relativity: Velocity Potentials and a Variational Principle, Phys. Rev. D 2, 2762
(1970).