EINSTEIN-HILBERT ACTIONS WITH TORSION

Nenad O. Vesić a and Dragoljub D. Dimitrijević b

arXiv:1906.05672v1 [math.DG] 13 Jun 2019

Abstract
In this paper, we studied the full Einstein-Hilbert actions with respect to non-symmetric
metrics and the corresponding torsion. The first concrete result in this paper are the general
formulae


for pressure and density with respect to the Madsen’s article the equation (3.1),
in [10] . Based on these results, we obtained the expression of energy-momentum tensor
with respect to non-symmetric metrics. We started the generalization of the Bianchi type-I
model of cosmology with respect to the corresponding non-symmetric metrics.

Key words: Einstein-Hilbert action, energy-momentum tensor, non-symmetric metric ten-

sor, torsion, cosmological model

2010 Math. Subj. Classification: 53B50, 58Z05, 46G05

1 Introduction
Cosmology aims to explain the origin and evolution of the Universe, the underlying physical
processes, and to obtain a deeper understanding of the laws of physics [1, 2]. We have only one
universe to study, and we cannot make experiments with it, only observations.
Cosmology is based on the Einstein’s theory of general relativity, i.e. theoretical studies take
place in the context of gravitational theories based on Einstein’s theory of general relativity.
Spacetime and hence the evolution of the Universe is determined by the matter present via the
Einstein’s equations for gravitational field.
There are three ideas underlying Einstein’s theory of general relativity. The first is that
spacetime may be described as a curved, four-dimensional pseudo-Riemannian manifold. The
laws of physics must be expressed in a form that is valid independently of any coordinate system
used to label points in spacetime. The second essential idea underlying the general relativity
is that at every spacetime point there exist locally inertial reference frames, corresponding to
locally flat coordinates (carried by freely falling observers), in which the physics of general rela-
tivity is locally indistinguishable from that of special relativity. This is Einstein’s famous strong
equivalence principle and it makes general relativity an extension of special relativity to a curved
spacetime. The third key idea is that mass curves spacetime in a manner described by the tensor
field equations of Einstein.
a,b
Faculty of Science and Mathematics, University of Niš
a
Department of Mathematics, Serbian Ministry of Education, Science and Technological Development, Grant
No. 174012
b
Department of Physics, Serbian Ministry of Education, Science and Technological Development, Grant No.
174020, ICTP – SEENET-MTP NT-03 project "Cosmology-Classical and Quantum Challenges"

1
 Einstein’s theory of general relativity defines equations for gravity. It is a system of non-linear
partial differential

equations of up to second order for the components gij of the spacetime metric
tensor ĝ see [3] . They determine the structure of spacetime in a covariant and coordinate-
independent way. This statement agrees with the first idea that the laws of physics must be
expressed in a form that is valid independently of any coordinate system.
Very important fact is that the use of geometry provides important additional insights by
which much information can be gained from Einstein’s equations in a systematic way. Spacetime
itself is equipped with a pseudo-Riemannian structure and encodes gravity (gravitational field)
in a geometrical way. Consequently, geometry (or more precisely differential geometry) provides
means to understand the structure of the spacetime itself.

1.1 Motivation
Note that Einstein’s gravity is (appropriately) described by the pseudo-Riemann geometry
which is torsion-free. The spacetime metric represents the gravitational field. The connections
are given by the Christoffel symbol compatible with the metric structure.
Allowing spacetime to have non-zero torsion, which arise naturally in generalized gauge the-
ories of gravity and in string theory, we can analyze cosmological models with a such geometry.
We will consider an action of the generalized Einstein-Hilbert-like form constructed for non-zero
torsion case.
In this paper we investigate cosmological aspects of spacetime with torsion and discuss mod-
ified expressions obtained involving torsion in general relativity. We will not discuss extended
Einstein’s general relativity which includes spin, i.e. Einstein-Cartan-Kibble-Sciama theory of
gravity [4,5]. We will discuss a model with (dominant) cosmological perfect fluid described in the
usual way. This perfect cosmological fluid naturally arises as a consequence of non-zero torsion.
This paper is composed of the introduction and the following sections:

- In the second section, we will recall the necessary results from differential geometry.

- In the third section, we will generalize the energy-momentum tensor. This part will start with
the expression of the energy-momentum tensor from the Madsen’s article [10]. After that, we
will consider 4-dimensional spacetimes equipped with different non-symmetric metrics.

- In the fourth section, we will apply the obtained general results to study the Friedmann
spacetime the Bianchi type-I spacetime with torsion. Precisely, we will consider the energy-
momentum tensor with respect to special non-symmetric metrics.

2 Necessary observations from differential geometry

Different Riemannian and generalized Riemannian spaces have been studied by a lot of au-
thors. Some of them are L. P. Eisenhart [8,9], M. Blau [6], M. S. Madsen [10], V. N. Ponomarev,
A. O. Barvinsky, Y. N. Obukhov [15], N. S. Sinyukov [17], J. Mikeš and his research team [11],
Lj. S. Velimirović, S. M. Minčić, M. S. Stanković [18, 19] and many others. S. M. Minčić [12–14]
obtained curvature tensors for non-symmetric affine connection spaces (the affine connection
spaces with torsion). Because the generalized Riemannian spaces are special non-symmetric
affine connection spaces, these results will be useful in this article.

2
2.1 Riemannian and generalized Riemannian spaces
Let us present definitions and observations necessary for further research in this paper.

Definition 2.1. [8,9] An N -dimensional manifold MN equipped with the non-symmetric metric
g
tensor ĝ with the components gij = gij (x0 , . . . , xN −1 ) is the generalized Riemannian space GRN .

Because the tensor ĝ is non-symmetric, the symmetric and anti-symmetric part of the com-
ponents gij are

1
1

gij = gij + gji and gij = gij − gji . (2.1)
2 ∨ 2
It evidently holds the equality gij = gij + gij .

∨

We guess that the matrix gij N ×N is non-singular, i.e. g = det gij N ×N 6= 0. For this

−1
reason, the contravariant metric is the inverse matrix g ij = gij N ×N .
g
The generalized Christoffel symbols of the first kind of the space GRN are
1

Γi.jk = ∂k gjα − ∂α gjk + ∂j gαk . (2.2)
2
g
The affine connection coefficients of the space GRN are the generalized Christoffel symbols
of the second kind
1

Γijk = g iα Γα.jk = giα ∂k gjα − ∂α gjk + ∂j gαk , (2.3)
2
One may easily check that it holds Γijk 6= Γikj . For this reason, the symmetric and anti-
symmetric parts of the affine connection coefficient Γijk are

1 i
1

Γijk = Γjk + Γikj and Γijk = Γijk − Γikj . (2.4)
2 ∨ 2
g
The differences T ijk = Γijk − Γikj are called the components of the torsion tensor of the space
g
GRN .

Remark 2.1. The affine connection ∇ of an affine connection space is the bilinear transfor-
mation of the set of differentiable vector spaces on a manifold M. With respect to the affine
connections with or without torsion, the corresponding covariant derivatives are defined.
i ...i i ...i i ...i i ...i
In the case of the geometrical objects Xj11...jpq and Yj11...jqp , the commutator [Xj11...jpq , Yj11...jqp ]
vanishes.
In [15], the anti-symmetric part Γijk is called the torsion tensor. Generally, if ∇ is the affine
∨
g
connection of the space GRN the torsion tensor is defined as

ˆ Ŷ + [X̂, Ŷ ],
T̂ (X̂, Ŷ ) = ∇Ŷ X̂ − ∇ X̂

3
for the commutator [X̂, Ŷ ] = X̂ Ŷ − Ŷ X̂.
g
Coordinately, it has the form T ijk = Γijk − Γikj and we will use this definition for components
of torsion.

It is easy to prove that it is satisfied the following equations

1

Γijk = giα ∂k gjα − ∂α gjk + ∂j gαk , (2.5)
2
i 1 iα

Γjk = g ∂k gjα − ∂α gjk + ∂j gαk . (2.6)
∨ 2 ∨ ∨ ∨

Moreover, the following expressions also hold

1

Γi.jk = giα Γαjk = ∂k gji − ∂i gjk + ∂j gik . (2.6’)
∨ ∨ 2 ∨ ∨ ∨

Similarly as in the case of the symmetric and anti-symmetric part of metric tensor, we get
Γijk = Γijk + Γijk . Moreover, with respect to the definition of the generalized Riemannian space
∨
g
GRN (see [8, 9]), it holds

1  
Γαiα = ∂i g and Γαiα = 0. (2.7)
2g ∨

The affine connection space equipped with the affine connection which affine connection
g
coefficients are Γijk is the Riemannian space RN and it is called the associated space of the space
g
GRN .
g
With respect to the affine connection of the associated space RN , one kind of covariant
derivative is defined as (see [8, 9, 11, 17])
g
∇k Xji = ∂k Xji + Γiαk Xjα − Γαjk Xαi , (2.8)

for a geometrical object Xji of the type (1, 1).

g
There is one identity of Ricci type with regard to the covariant derivative ∇. The components
of the corresponding curvature tensor are
g
Rijmn = ∂n Γijm − ∂m Γijn + Γαjm Γiαn − Γαjn Γiαm . (2.9)
g
The components of the Ricci-curvature tensor and scalar curvature for the space RN are

g g g g
Rij = Rαijα = ∂α Γαij − ∂j Γαiα + Γβij Γαβα − Γβiα Γαjβ and R = Rαβ gαβ . (2.10)

g g
With respect to the affine connection ∇e of the generalized Riemannian space GRN , four kinds
of the covariant derivative are defined [12–14]

4
 g
e k X i = ∂k X i + Γi X α − Γα X i ,
∇ (2.11)
j j αk j jk α
0
g
e k Xji = ∂k Xj + Γi Xjα − Γα Xαi ,
∇ (2.12)
kα kj
1
g
e k X i = ∂k X i + Γi X α − Γα X i ,
∇ (2.13)
j j αk j kj α
2
g
e k Xji = ∂k Xji + Γi Xjα − Γα Xαi .
∇ (2.14)
kα jk
3

Based on these covariant derivatives, four curvature tensors, eight derived curvature tensors
and fifteen curvature pseudotensors are obtained [12–14].
In this paper, we will deal with the lagrangian obtained with respect to the curvature and
derived curvature tensors. The components of the curvature tensors and the derived curvature
g
tensors for the space GRN are elements of the family [20]

g g g g g g g g g g g g
ei
R i i ′ i α i ′ α i α i
(2.15)
jmn = Rjmn + u∇n T jm + u ∇m T jn + v T jm T αn + v T jn T αm + w T mn T αj ,

g ĝ
for real coefficients u, u′ , v, v ′ , w and the components T ijk of the torsion tensor T .
With respect to the equations (2.7, 2.15), one obtains that the families of Ricci-curvature
g
tensors and the scalar curvature of the space GRN are

g g g g g g g g g g
R e = R − (v ′ + w)g γδ T α T β .
eij = Rij + u∇α T α − (v ′ + w)T α T β and R (2.16)
ij iβ jα γβ δα

Six of the curvature tensors in the family (2.15) are linearly independent, for example

g g
ei i 1 g gi 1 g gi 1 gα gi 1 gα gi
R jmn = Rjmn + ∇n T jm − ∇m T jn + T jm T αn − T T , (2.17)
0 2 2 4 4 jn αm
g g g g g g g g
eijmn = Rijmn − 1 ∇n T ijm + 1 ∇m T ijn + 1 T αjm T iαn − 1 gα gi
R T T , (2.18)
1 2 2 4 4 jn αm
g g g g g g g g
ei i 1 i 1 i 1 α i 1 gα gi 1 gα gi
R jmn = Rjmn + ∇n T jm + ∇m T jn − T jm T αn + T T − T T , (2.19)
2 2 2 4 4 jn αm 2 mn αj
g g
ei i 1 g gi 1 g gi 1 gα gi 1 gα gi 1 gα gi
R jmn = Rjmn + ∇n T jm + ∇m T jn − T jm T αn + T T + T T , (2.20)
3 2 2 4 4 jn αm 2 mn αj
g g g g g g
1 1
eijmn = Rijmn − T αjm T iαn + T αjn T iαm ,
R (2.21)
4 4 4
g g g g
ei i 1 α i 1 gα gi
R jmn = Rjmn + T jm T αn + T jn T αm . (2.22)
5 4 4
g
The components of the corresponding Ricci-curvature tensors of the space GRN are

5
 g g g g g g g g g g g g
eij = Rij + 1 ∇α T α + 1 T β T α
R eij = Rij − 1 ∇α T α + 1 T β T α ,
R
0 2 ij 4 iα jβ 1 2 ij 4 iα jβ
g g g g g g g g g g g g
eij = Rij + 1 ∇α T α + 1 T β T α , R
R ij
eij = Rij + 1 ∇α T α − 3 T β T α ,
ij
(2.23)
2 2 4 iα jβ 2 4 iα jβ
3
g g g g g g
eij = Rij − 1 T β T α ,
R eij = Rij .
R
4 4 iα jβ 5
g g g
eij , R
Three of them, for instance R eij , R
eij , are linearly independent.
0 1 5
g
The corresponding scalar curvatures of the space GRN are
g g g g g g g g
e = R + 1 gγδ T βγα T α
R e = R + 1 gγδ T βγα T α ,
R
0 4 δβ 1 4 δβ
g g g g g g g g
e =R+
R 1 γδ β α e = R − 3 gγδ T βγα T α , (2.24)
2 4 g T γα T δβ , R
3 4 δβ
g g g g g g
e = R − 1 gγδ T βγα T α , R
R e = R.
4 4 δβ 5
g g g
e, R
Two of the scalar curvatures, e.g. R e = R, are linearly independent.
0 5
g
In our research about physics, we will stay focused on the generalized Riemannian space GR4
equipped with the metric tensor
 
s0 (t) n0 (t) n1 (t) n2 (t)
 −n0 (t) s1 (t) n3 (t) n4 (t) 
ĝ = 
 −n1 (t) −n3 (t)
, (2.25)
s2 (t) n5 (t) 
−n2 (t) −n4 (t) −n5 (t) s3 (t)
for differentiable functions s1 (t), s2 (t), s3 (t), n0 (t), . . . , n5 (t), depending on the time coordinate t.
The variables in the space GR4 will be x0 = t and the variables x1 , x2 , x3 are the space variables.

2.2 Variations
We will recall the infinitesimal deformations as in [18,19] in this part of the paper. A necessary
rule will be obtained in here.
g g
A transformation f : GRN → GRN defined as x ≡ (xi ) → x ≡ (xi ), for

xi = xi + εz i (xj ), i, j = 0, . . . , N − 1, (2.26)
g
where ε is an infinitesimal, is the infinitesimal deformation of the space GRN determined by the
vector field z = z(xi ).
A local coordinate system in which the point x is endowed with coordinates xi and the point x
with the coordinates xi will be denoted by (s). In another coordinate system (s′ ), corresponding
′ ′
to the point x = (xi ) new coordinates xi = xi , i.e. as new coordinates xi of the point x = (xi )
′
we choose old coordinates of the point x = (xi ). In other words, the equalities x = (xi ) = (xi )
are satisfied at the system (s′ ).

6
 A considered geometric object A with respect to the system (s) at the point x = xi will be
denoted as A(s, x).
The point x is said to be deformed point of the point x, if the equation (2.26) holds. Geo-
metrical object A(s, x) is the deformed object A(s, x) with respect to the deformation (2.26), if
it holds

A(s′ , x) = A(s, x), (2.27)

for the coordinate systems s and s′ .
The limit

A(s, x) − A(s, x)
δA = lim (2.28)
ε→0 ε
is the (first) variation of the geometric object A(s, x). We may notice A(s, x) = A(s, x) + εδA.
Let
Z
 
I[f ] = F t, f (t), f ′ (t) dt, (2.29)

be a functional for f , where f ′ (t) = df /dt. If f is varied by adding to it a function δf , and the
integrand F (x, f + δf, f ′ + δf ′ ) is expanded in powers of δf , then the change in the value of I
to first order in δf is
Z  
∂F d ∂F
δI[f ] = − δqdt. (2.30)
∂f dt ∂f ′
The function δF/δf (t) is the functional derivative of F with respect to f at the point t. This
functional derivative may be computed as

δF ∂F d ∂F
= − . (2.31)
δf (t) ∂f dt ∂f ′
As we may conclude, if values of a function F do not change when we change values of a
function f (t), with respect to t, then the functional derivative of F with respect to f vanishes.
In other words, if the function F is not expressed as a composition of the function f (t) and some
other function g, then the functional derivative δF/δf (x) is equal 0.
With respect to the equation (2.31), we may define the functional derivative of a tensor X̂
i ...i
with components Xj00...jpq with respect to a tensor Ŷ with components Ynm0 ...n
0 ...mu .
v
i ...i
First of all, notice that the component Xj00...jpq of the tensor X̂ is expressed as
p
X q
X
i ...i (u)
Xj00...jpq = d(u) · X + ∂(v) X(v) , (2.32)
u=0 v=0

for the orthogonal base (du ) ≡ (d/dxu ), u = 0, . . . , p, of the space of functionals and the base
(∂v ) = (∂/∂xv ), v = 0, . . . , q, of the space of positions.
i ...i i ...ip
If we treat the components Xj00...jpq and Ynm0 ...n0 ...mu as functions X 0
v
m0 ...mu
j0 ...jq (ε) and Yn0 ...nv (ε) of
an infinitesimal deformation, we may conclude that the function F from the equation (2.29) is

  
F ε, Ynm0 ...n
0 ...mu
v
(ε), d Ynm0 ...n
0 ...mu
v
(ε) /dε = F ε, Ynm0 ...n
0 ...mu
v
(ε), δYnm0 ...n
0 ...mu
v
. (2.33)

7
 i ...i
After comparing the equations (2.31), (2.33), we will conclude that if the components Xj00...jpq (ε)
i ...i
and Ynm0 ...n
0 ...mu (ε) as well as the components X 0
v
m0 ...mu
j0 ...jq (ε) and δYn0 ...nv are functionally independent,
p

i ...i
the variational derivative of the component Xj00...jpq with respect to the component Ynm0 ...n
0 ...mu is
v

equal zero. Moreover, if the tensor X̂ is not expressed as a composition of the tensor Ŷ and
arbitrary tensor Ẑ, the functional derivative of the tensor X̂ with respect to the tensor Ŷ is equal
zero.

3 Theoretical considerations
In this section, we will theoretically consider the Einstein-Hilbert action with torsion and the
corresponding energy-momentum tensor.

3.1 Four-dimensional spacetime

g
Let GR4 be the generalized Riemannian space equipped with the metric tensor (2.25). The
symmetric and anti-symmetric parts of this tensor are

   
s0 (t) 0 0 0 0 n0 (t) n1 (t) n2 (t)
 0 s1 (t) 0 0   n3 (t) n4 (t) 
ĝ =   , ĝ =  −n0 (t) 0 . (3.1)
 0 0 s2 (t) 0  ∨
 −n1 (t) −n3 (t) 0 n5 (t) 
0 0 0 s3 (t) −n2 (t) −n4 (t) −n5 (t) 0


The contravariant metric tensor ĝ −1 with the components gij is

−1 
s0 (t) 0 0 0

−1 
 0 s1 (t) 0 0 
ĝ −1 = 
−1 . (3.2)
 0 0 s2 (t) 0 

−1
0 0 0 s3 (t)
g
The generalized Christoffel symbols of the first and the second kind of the space GR4 are

1  1
−1  
Γi.jk = ∂k gji − ∂i gjk + ∂j gik and Γijk = si (t) ∂k gji − ∂i gjk + ∂j gik . (3.3)
2 2
The covariant anti-symmetric Christoffel symbols are

Γ0.12 = − 12 n′3 (t), Γ0.21 = 1 ′

n (t),
2 3
Γ1.02 = 1 ′
n (t),
2 3
Γ1.20 = − 21 n′3 (t), Γ2.01 = − 21 n′3 (t), Γ2.10 = 1 ′
n (t),
2 3
∨ ∨ ∨ ∨ ∨ ∨
Γ0.13 = − 12 n′4 (t),
∨
Γ0.31 =
∨
1 ′
n (t),
2 4
Γ1.03 =
∨
1 ′
n (t),
2 4
Γ1.30 = − 21 n′4 (t),
∨
Γ3.01 = − 21 n′4 (t),
∨
Γ3.10 =
∨
1 ′
n (t),
2 4 (3.4)
Γ0.23 = − 12 n′5 (t), Γ0.32 = 1 ′
n (t),
2 5
Γ2.03 = 1 ′
n (t),
2 5
Γ2.30 = − 21 n′5 (t), Γ3.02 = − 21 n′5 (t), Γ3.20 = 1 ′
n (t),
2 5
∨ ∨ ∨ ∨ ∨ ∨

and Γi.jk = 0 in all other cases.

∨
The corresponding anti-symmetric parts of the generalized Christoffel symbols Γijk are

8
 1
−1 1
−1
Γ012 = g 0α Γα.12 = − n′3 (t) s0 (t) , Γ021 = n′3 (t) s0 (t) ,
∨ ∨ 2 ∨ 2
1
−1 1
−1
Γ013 = g 0α Γα.13 = − n′4 (t) s0 (t) , Γ031 = n′4 (t) s0 (t) ,
∨ ∨ 2 ∨ 2
0 0α 1
−1 0 1 ′
−1
Γ23 = g Γα.23 = − n′5 (t) s0 (t) , Γ32 = n5 (t) s0 (t) ,
∨ ∨ 2 ∨ 2
1
−1 1
−1
Γ120 = g 1α Γα.20 = − n′3 (t) s1 (t) , Γ102 = n′3 (t) s1 (t) ,
∨ ∨ 2 ∨ 2
1
−1 1
−1
Γ130 = g 1α Γα.30 = − n′4 (t) s1 (t) , Γ103 = n′4 (t) s1 (t) ,
∨ ∨ 2 ∨ 2
2 2α 1
−1 2 1 ′
−1
Γ01 = g Γα.01 = − n′3 (t) s2 (t) , Γ10 = n3 (t) s2 (t) ,
∨ ∨ 2 ∨ 2
1
−1 1
−1
Γ30 = g 2α Γα.30 = − n′5 (t)
2
s2 (t) , Γ203 = n′5 (t) s2 (t) ,
∨ ∨ 2 ∨ 2
1
−1 1
−1
Γ301 = g 3α Γα.01 = − n′4 (t) s3 (t) , Γ310 = n′4 (t) s3 (t) ,
∨ ∨ 2 ∨ 2
1
−1 1
−1
Γ302 = g 3α Γα.02 = − n′5 (t) s3 (t) , Γ320 = n′5 (t) s3 (t) ,
∨ ∨ 2 ∨ 2
and Γijk = 0 in all other cases.
∨
Because Γγiδ Γδjγ = Γγjδ Γδiγ , it is enough to obtain the following geometrical objects for further
∨ ∨ ∨ ∨
calculations:

1  ′
2
−1 
(3.5)

2
2
Γγ0δ Γδ0γ = −

−1
−1
−1
−1
−1
n3 (t) s1 (t) s2 (t) + n′4 (t) s1 (t) s3 (t) + n′5 (t) s2 (t) s3 (t) ,
∨ ∨ 2
1  
(3.6)

2
2
Γγ1δ Γδ1γ = −

−1
−1
−1
−1
n′3 (t) s0 (t) s2 (t) + n′4 (t) s0 (t) s3 (t) ,
∨ ∨ 2
1
Γγ1δ Γδ2γ = − n′4 (t)n′5 (t) s0 (t) (3.7)

−1
−1
s3 (t) ,
∨ ∨ 2
1
Γγ1δ Γδ3γ = n′3 (t)n′5 (t) s0 (t) (3.8)

−1
−1
s2 (t) ,
∨ ∨ 2
1  
(3.9)

2
2
Γγ2δ Γδ2γ = −

−1
−1
−1
−1
n′3 (t) s0 (t) s1 (t) + n′5 (t) s0 (t) s3 (t) ,
∨ ∨ 2
1
Γγ2δ Γδ3γ = − n′3 (t)n′4 (t) s0 (t) (3.10)

−1
−1
s1 (t) ,
∨ ∨ 2
1  
(3.11)

2
2
Γγ3δ Γδ3γ = −

−1
−1
−1
−1
n′4 (t) s0 (t) s1 (t) + n′5 (t) s0 (t) s2 (t) ,
∨ ∨ 2

and Γγiδ Γδjγ = 0 in all other cases for i ≤ j.

∨ ∨
g g
e, R and
With respect to the tensor ĝ −1 , the above mentioned linearly independent scalars R
0
the metric determinant g = s0 (t)s1 (t)s2 (t)s3 (t), we obtain that the scalar curvature of the space
g

GR4 with torsion is see the equations (2.24) and (2.25)
g g 



e = R − 3 · g−1 s3 (t) n′ (t) 2 + s2 (t) n′ (t) 2 + s1 (t) n′ (t) 2 .
R (3.12)
3 4 5
2

9
 The scalar curvature (3.12) can be used to define the full Einstein-Hilbert action.
p g
Note that the Lagrangian density in [6] is |g|R. Here, we examine the Lagrangian density
p g
in the case of the space with torsion to be |g|Re.
0
We are aimed to express a term LM describing a dominant cosmological fluid appearing in
the model as the function of torsion. The full Lagrangian density of the model takes the form
 g 
p 1 1 γδ g β g α
L = |g| R + g T γα T δβ . (3.13)
2κ 4
Remark 3.1. The same effect would be achieved if we get the metric
 
s0 (t) 2κn0 (t) 2κn1 (t) 2κn2 (t)
 −2κn0 (t) s 1 (t) 2κn3 (t) 2κn4 (t) 
ĥ = 
 −2κn1 (t) −2κn3 (t)
.

s2 (t) 2κn5 (t)
−2κn2 (t) −2κn4 (t) −2κn5 (t) s3 (t)
g g g g
and the Lagrangian L = R + 41 gγδ T αγβ T βδγ , for LM = 41 gγδ T αγβ T βδγ .
Precisely, we guess that the symmetric and anti-symmetric part of the metric g will satisfy
the equality gij = gij + 2κgij in our paper. It is done for the simplifying of the expression through
∨
the computation process.

Remark 3.2. We could easily put 2κ = 1. In this way, some researchers may find wrong
conclusions. To avoid that, we will analyze the general case with the constant κ.

Lemma 3.1. In the space GR4 equipped with the metrics

 
g11 g12 g13 g14
∨ ∨ ∨
 
 −g12 g22 g23 g24 
ĝ =  ∨ ∨
 −g13 −g23 g33 g34
∨ ,

 ∨ ∨ ∨ 
−g14 −g24 −g34 g44
∨ ∨ ∨

g g
1


the tensor T̂ whose components are T i.jk = 2 ∂k gji − ∂i gjk + ∂j gik is not a function of neither
∨ ∨ ∨
gpq nor ∂r gpq .

Proof. The following equalities are satisfied:

g g g g g g
T i.jk = −T i.kj , T i.jk = −T j.ik , T i.jk = T k.ji .
g
That means the torsion tensor T i.jk may be expressed as
g
T i.jk = tijk + τijk ,
for a geometrical object tijk anti-symmetric in any pair of indices i, j, k and τijk = τkji . For this
reason, we get

10
 1 g g
1

τijk = T i.jk + T k.ji = g✚
∂k✚ ❍ i gjk ❍
ji −∂ ✟✟❍ ✟✟ ❍ ✟✟ = 0. (3.14)
❍❍ +∂ ❍
✟❍
j gik +∂
❍i gjk −∂ g
✟ ∨ ✟✟❍
k ji +∂
❍j gki
2 2 ✚ ∨ ∨ ✟
❍ ❍∨ ❍ ∨ ✟
❍ ❍∨

g
In this way, we proved that the torsion tensor T i.jk is totally anti-symmetric.
g
Assume that the covariant torsion tensor T̂ with the components T i.jk is a function of some
g
∂r gpq . Because T i.jk is anti-symmetric in any pair of the indices i, j, k but the geometrical object
∂r gpq is symmetric in the indices p and q, these indices should be mute (dummy) in the functional
g
correspondence between the objects ∂r gpq and T i.jk .
That means that it is satisfied the equation
g
(i)(k)
T i.jk = ∂j g(i)(k) qik + θi.jk ,
for some geometrical object θi.jk which is not a function of ∂j gik . Because the symmetric metric
ĝ in this lemma is diagonal, the first summand in the last equation is not equal zero if and only
g
if i = k. Because the component T i.ji = 0, i.e. it is the trivial function, its functional derivative
by any function is equal 0.

3.2 Variations of the action

The variation of the action with the Lagrangian density (3.13) is

Z    g 
4 1 g 1 γδ g β g α p p 1 1 γδ g β g α
δS = d x R + g T γα T δβ δ |g| + |g|δ R + g T γα T δβ . (3.15)
2κ 4 2κ 4
p p R p g
Because δ |g| = − 12 |g|gαβ δgαβ , d4 x |g|gαβ δRαβ = 0 (from standard calculations) and
g g
δR = Rαβ δgαβ , the equation (3.15) transforms to
Z  g 
1 4 1 1 γδ g ζ g ǫ p
δS = − d x R + g T γǫ T δζ |g|gαβ δgαβ
2 2κ 4
 g g 
(3.16)
Z p g
Z p δ gǫζ T γǫδ T δζγ
1 1
+ d4 x |g|Rαβ δgαβ + d4 x |g| δgαβ ,
2κ 4 δgαβ
i.e.

Z  
1 4 1 g 1 g g p
δS = − d x R + gγδ T ζγǫ T ǫδζ |g|gαβ δgαβ
2 2κ 4
 
ǫζ ργ σδ
g g (3.16’)
Z
1p g 1
Z p δ g g g T ρ.ǫδ T σ.ζγ
+ d4 x |g|Rαβ δgαβ + d4 x |g| δgαβ .
2κ 4 δgαβ
Based on the equations (2.6, 2.6’), we conclude that the anti-symmetric part Γi.jk is anti-
∨
symmetric by indices i and j as well as j and k, but it is symmetric by indices i and k.

11
 g g g g
Let us define the scalar object F (gij ) = g ǫζ g ργ gσδ T ρ.ǫδ T σ.ζγ . Because T i.jk = −T i.kj =
g g g g g
−T j.ik = T j.ki = −T k.ji = T k.ij , gαβ δgαβ = −gαβ δgαβ , gǫζ = gζǫ , δT i.jk /δgpq = 0 and gǫζ = gζǫ
we obtain

 
δ F (gij ) g g g g g g g g
= 2g ργ gσδ T ρ.αδ T σ.βγ + 2gǫζ gσδ T α.ǫδ T β.σζ + 2gǫζ g ργ T α.ρǫ T β.ζγ = 6T γαδ T δβγ . (3.17)
δgαβ

Hence, the equation (3.16’) is equivalent to

Z  
4
p 1 g 1 γδ g ζ g ǫ 1 g 2 gγ gδ
δS = d x |g| − Rgαβ − g T γǫ T δζ gαβ + Rαβ + T αδ T βγ δgαβ . (3.16’’)
4κ 8 2κ 3

Variation of the action is zero, δS = 0, if and only if

g
 
1g 1 γδ g ζ g ǫ 4 gγ gδ
Rij − Rgij = κ g T γǫ T δζ gij − T iδ T jγ . (3.18)
2 4 3
In this way we obtained the set of generalized Einstein equations.
g
Because there are two linearly independent scalar curvatures of the space GR4 , for example
g g g
e and R
R e = R, the generalized Einstein equations (3.18) reduce to the Einstein’s equations
0 5
g 1g
Rij − Rgij = 0, (3.19)
2
g
p p g
e ≡ |g|R (LM ).
in the case of the lagrangian L = |g|R
5
The equations (3.18, 3.19) are the linearly independent dynamical equations for metric field
with torsion. Any other equation of motion for models with torsion may be expressed as the
corresponding linear combination of the equations (3.18, 3.19).
Components of the corresponding energy-momentum tensor for the cosmological fluid are
written on the right side of (3.18)

1 γδ g ζ g ǫ
g 4g g
g T γǫ T δζ gij − T γiδ T δjγ .
T ij = (3.20)
4 3
With respect to the expressions (3.5—3.11), we get

g g 
2
2
2 
T α0β T β0α = −2g−1 s0 (t) n′3 (t) s3 (t) + n′4 (t) s2 (t) + n′5 (t) s1 (t) , (3.21)
g g 
2
2 
T α1β T β1α = −2g−1 s1 (t) n′3 (t) s3 (t) + n′4 (t) s2 (t) , (3.22)
g g
T α1β T β2α = −2g−1 s1 (t)s2 (t)n′4 (t)n′5 (t), (3.23)
g g
T α1β T β3α = 2g−1 s1 (t)s3 (t)n′3 (t)n′5 (t), (3.24)
g g
T α2β T β1α = −2g−1 s1 (t)s2 (t)n′4 (t)n′5 (t), (3.25)

12
 g g 
2
2 
T α2β T β2α = −2g−1 s2 (t) n′3 (t) s3 (t) + n′5 (t) s1 (t) , (3.26)
g g
T α2β T β3α = −2g−1 s2 (t)s3 (t)n′3 (t)n′4 (t), (3.27)
g g
T α3β T β1α = 2g−1 s1 (t)s3 (t)n′3 (t)n′5 (t), (3.28)
g g
T α3β T β2α = −2g−1 s2 (t)s3 (t)n′3 (t)n′4 (t), (3.29)
g g 
2
2 
T α3β T β3α = −2g−1 s3 (t) n′4 (t) s2 (t) + n′5 (t) s1 (t) , (3.30)

g g g g
but T α0β T βiα = T αiβ T β0α = 0 for i ∈ {1, 2, 3}.
Because the matrix ĝ −1 is diagonal, the following equation holds
g g 
2
2
2 
gγδ T ζγǫ T ǫδζ = −6g −1 n′3 (t) s3 (t) + n′4 (t) s2 (t) + n′5 (t) s1 (t) . (3.31)

3.3 General formulae

The components of the energy-momentum tensor for a non-ideal (cosmological) fluid are [10]
g g g g
T ij = (p + ρ)ui uj + pgij + ∆ij , (3.32)
g g
ˆ
for the components ui of a velocity û, the pressure p and the density ρ, as well as the tensor ∆
whose components ∆ij are symmetric in the indices i and j. In the case of an ideal fluid, (3.32)
reduces to
g g g g
T ij = (p + ρ)ui uj + pgij . (3.33)
From the M. S. Madsen’s article [10], we may read that the energy-momentum tensor for
non-ideal fluid is expressed more explicitly as
g g g g g

T ij = ρui uj + qi uj + q j ui − phij + πij , (3.34)
g g
for the velocity components ui such that uα uα = 1, the energy density ρ = T αβ uα uβ , the
g g g g
geometrical objects hij = gij − ui uj , q i = T αβ uα hβi , Πij = phij + πij = −T αβ hαi hβj , and the
g
pressure p = 31 Παα .
From the expression of the objects Πij and hij , we get
g


Πij = −T αβ δiα − uα ui δjβ − uβ uj ,
which leads to
g g g g
Πij = −T ij + T iβ uβ uj + T αj uα ui − T αβ uα uβ ui uj . (3.35)
If we multiply the last equation with g ij , we will obtain the following results
g g g g
Παα = −T αα + T αβ uβ uα + T αβ uα uβ − T αβ uα uβ uγ uγ ,

13
i.e.
g g
Παα = −T αα + T αβ uα uβ . (3.36)
g
Hence, the pressure p of the fluid is

g 1g
1 g g

p = − T αβ g αβ − uα uβ = − T αα − ρ . (3.37)
3 3
In the comoving reference frame ui = δi0 , the last equation reduces to

g 1g 1g
p = − T αα + T 00 . (3.37’)
3 3
The following lemma holds.
g g
Lemma 3.2. The pressure p, the density ρ and the components of the energy-momentum tensor
for a non-ideal fluid satisfy the equation (3.37). In the comoving reference frame, the pressure
reduces to (3.37’).
g
The energy density ρ is expressed as
g g
ρ = T αβ uα uβ . (3.38)

In the comoving reference frame the energy density reduces to

g g
ρ = T 00 . (3.38’)
Having in mind that the usual equation of state for a fluid is given by
g g g
p = ω · ρ,
g g
where ω is the state parameter, then ω can be expressed as

 −1
g 1 gα g β γ 1
ω = − T α T βγ u u + . (3.39)
3 3
In the comoving reference frame the last expression reduces to

g 1 g g
−1 1
ω = − T αα T 00 + . (3.40)
3 3

With respect to the previous computations and the last Lemma, we obtain that the next
theorem holds:
g g
Theorem 3.1. The components T ij of the energy-momentum tensor and the components T ijk of
the torsion tensor satisfy the equation
 
g (3.16 ′′ ) 1 g ζ g ǫ γδ 16 γ δ
T ij = T T g gij − δi δj . (3.41)
4 γǫ δζ 3

14
 The trace of the energy-momentum tensor is
g 
2
2
2 
T αα = 2g−1 n′3 (t) s3 (t) + n′4 (t) s2 (t) + n′5 (t) s1 (t) . (3.42)
The energy-density is

g 3 
2
2
2  4g g
ρ = − g−1 n′3 (t) s3 (t) + n′4 (t) s2 (t) + n′5 (t) s1 (t) − T γαδ T δβγ uα uβ . (3.43)
2 3
The pressure is

g 7 
2
2
2  4g g
p = − g−1 n′3 (t) s3 (t) + n′4 (t) s2 (t) + n′5 (t) s1 (t) − T γαδ T δβγ uα uβ . (3.44)
6 9
ĝ
The components of the 1-form q are

g g g
q i = T αi uα − ui T αβ uα uβ . (3.45)
g
The state parameter ω is

   4g g 
g 1 −1 ′

2 ′

2 ′

2 γ δ α β
ω= g n3 (t) s3 (t) + n4 (t) s2 (t) + n5 (t) s1 (t) + T αδ T βγ u u
3 9
    (3.46)
3 −1 ′

2 ′

2 ′

2 4 γ g δ α β −1
g
· g n3 (t) s3 (t) + n4 (t) s2 (t) + n5 (t) s1 (t) + T αδ T βγ u u .
2 3
The components of the energy-momentum tensor, the torsion-tensor, and the metric tensor
satisfy the equation

g 3 
2
2
2  4g g
T ij = − g−1 n′3 (t) s3 (t) + n′4 (t) s2 (t) + n′5 (t) s1 (t) gij − T αiβ T βjα, (3.47)
2 3
g g
for the above obtained T αiβ T βjα .
g g g g
Corollary 3.1. In the comoving reference frame, ui = δi0 , the expressions for ρ, p, q i and ω,
reduce to

g 1 −1
 ′
2
2
2 
p=− g 21 + 8s0 (t) n3 (t) s3 (t) + n′4 (t) s2 (t) + n′5 (t)s1 (t) , (3.48)
18
g g 1
 ′
2
2
2 
ρ = T 00 = − g−1 9 − 16s0 (t) n3 (t) s3 (t) + n′4 (t) s2 (t) + n′5 (t) s1 (t) , (3.49)
6
(
1 −1

 ′
2
2
2 
g −6g 9 − 16s0 (t) n3 (t) s3 (t) + n′4 (t) s2 (t) + n′5 (t) s1 (t) , i = 0,
qi = (3.50)
0 i > 0,
g 1

−1
ω = 21 + 8s0 (t) 9 − 16s0 (t) . (3.51)
3

15
4 Some special cases
In this section, we examine Friedmann-like and Bianchi type-I-like models with torsion.

4.1 Friedmann spacetime with torsion

Let be s0 (t) ≡ −1, s1 (t) = s2 (t) = s3 (t) = s(t), n3 (t) = n4 (t) = n5 (t) = n(t). In this case, it
is satisfied the following equations

g
2
−2
T 00 = 9.5 n′ (t) s(t) , (4.1)
g
2
−1
T 11 = 4 n′ (t) s(t) , (4.2)
g g
2
−1 g g
2
−1
T 12 = T 21 = 3 n′ (t) s(t) , T 13 = T 31 = −2 n′ (t) s(t) , (4.3)
g
2
−1 g g
2
−1
T 22 = 4 n′ (t) s(t) , T 23 = T 32 = 2 n′ (t) s(t) , (4.4)
g
2
−1
T 33 = 4 n′ (t) s(t) , (4.5)
g g
and T i0 = T 0i = 0 for i ∈ {1, 2, 3}.
In the comoving reference frame, we get

g (3.48 ) 39 ′
2
2 g (3.49) 25

2
2 g (3.51) 13
p = n (t) s(t) , ρ = 2 n′ (t) s(t) , ω = ≈ 0.173. (4.6)
18 75

Let be s0 (t) ≡ −1, s1 (t) = s2 (t) = s3 (t) = s(t), n3 (t) = n4 (t) = n(t), n5 (t) = 0. In this case,
we get:

g 17 ′
2
−2
T 00 = n (t) s(t) , (4.7)
3
g 1 ′
2
−1
T 11 = n (t) s(t) , (4.8)
3
g 1
2
−1 g g
2
−1
T 22 = n′ (t) s(t) , T 23 = T 32 = 2 n′ (t) s(t) , (4.9)
3
g 23 ′
2
−1
T 33 =− n (t) s(t) , (4.10)
12
g g
and T ij = T ji = 0 in all other cases for i, j ∈ {0, 1, 2, 3}.
In the comoving reference frame, we also obtain

g (3.48) 13 ′
2
−2 g (3.49) 25

2
−2 g (3.51) 13
p = n (t) s(t) , ρ = 3 n′ (t) s(t) , ω = ≈ 0.173. (4.11)
9 75

16
4.2 Bianchi type-I spacetime with torsion
The Bianchi type-I cosmological model is the main subject for our observations in this sub-
section. This model is characterized by the first square form
2 2 2
ds2 = −dt2 + s1 (t) dx1 + s2 (t) dx2 + s3 (t) dx3 . (4.12)
All above obtained results may be treated as generalizations of the Bianchi type-I spatetime
model with torsion. In this section, we will generalize the Bianchi type-I spatetime model with
a bidiagonal metrics.
As we may see from the results above, the components n0 (t), n1 (t), n2 (t) do not affect the
action. For this reason, we will pay attention to the following metric
 
−1 0 0 0
 0 s1 (t) c(t) 0 
b̂ = 
 0 −c(t) s2 (t)
. (4.13)
0 
0 0 0 s3 (t)

The symmetric and anti-symmetric part of b̂ are

   
−1 0 0 0 0 0 0 0
 0 s1 (t) 0 0   0 0 c(t) 0 
b̂ = 
 0
 and b̂ = 
 
. (4.14)
0 s2 (t) 0 ∨ 0 −c(t) 0 0 
0 0 0 s3 (t) 0 0 0 0

The determinant of the matrix b̂ is f = −s1 (t)s2 (t)s3 (t).

We obtain the following values of energy-momentum tensor components in this case:

b 25
−1 ′
2
T 00 = s1 (t)s2 (t) c (t) , (4.15)
6
b 7
−1 ′
2
T 11 = − s2 (t) c (t) , (4.16)
6
b 7
−1 ′
2
T 22 = − s1 (t) c (t) , (4.17)
6
b 3
−1
2
T 33 = s1 (t)s2 (t) s3 (t) c′ (t) , (4.18)
2
b
and T ij = 0 in all other cases.
In the comoving reference frame, we get the following values

b (3.48) 13  −1
2 b (3.49) 25

 −1
2 b (3.51) 13
p = s1 (t)s2 (t) c′ (t) , ρ = 6 s1 (t)s2 (t) c′ (t) , ω = . (4.19)
18 75

4.2.1 The Friedmann-Lemaitre-Robertson-Walker model

In this part of the paper, we are aimed to consider the Friedmann-Lemaitre-Robertson-Walker
cosmological model as a special case of the Bianchi type-I model.

17
 The metric (4.13) reduces to the metric for the FLRW model in the case of s1 (t) = s2 (t) =
s3 (t) = s(t). Because our previous results for pressure and density are obtained for general
non-symmetric metric, we conclude that the results for FLRW model are the special cases of the
results obtained for the Bianchi type-I model. In this case, i.e. in the case of the metric
 
−1 0 0 0
 0 s(t) c(t) 0 
fˆ = 
 0 −c(t) s(t) 0 
 (4.20)
0 0 0 s(t)
we have

2
2
2
f 25 c′ (t) f f 7 c′ (t) f 3 c′ (t)
T 00 = ·
2 , T 11 = T 22 = − · , T 33 = · ,
6 s(t) 6 s(t) 2 s(t)

2
2 (4.21)
f 13 c′ (t) f 25 c′ (t) f 13
p= ·
2 , ρ = ·
2 , ω = .
18 s(t) 6 s(t) 75

5 Conclusion
After recalled the necessaries from differential geometry (Section 2), we geometrically studied
and generalized the concept of the energy-momentum tensor.
In the Section 3 of this paper, we obtained the general formulae of the energy-momentum
tensor, the pressure and the density with respect to the non-symmetric metrics and the torsion-
tensor. After that, we analyzed the energy-momentum tensors obtained with respect to the
special non-symmetric metrics.
In the Section 4, we applied the general formulae to generalize the Bianchi type-I spacetime
model. We studied the special case of the non-symmetric metrics with non-zero components
placed on both diagonals of the non-symmetric metric matrices.
It is worth mentioning that although the state parameters for Friedmann and Bianchi type-
I spacetime model with torsion have the same numerical value, it is obvious that those two
cosmological fluids have different dynamics, i.e. their pressures and densities are different.
Based on the results presented in this paper, in the future

we will examine how many gener-
alized Riemannian spaces in the Eisenhart’s sense see [8,9] does a lagrangian LM 6= 0 generate.
We will try to give the physical interpretations of these results. Moreover, we will use different
concepts
of the generalized Riemannian space the Eisenhart’s model used in this paper
is one
of them to expand the Shapiro’s model of the cosmology with torsion presented in [16] .

References
[1] V. Mukhanov, Physical Foundations of Cosmology, Cambridge University Press, Cam-
bridge, UK (2005).

[2] S. Weinberg, Cosmology, Oxford University Press, Oxford, UK (2008).

[3] M. Bojowald, Canonical Gravity and Applications: Cosmology, Black Holes, and Quantum
Gravity, Cambridge University Press, Cambridge (2011).

18
 [4] T. W. B. Kibble, Lorentz Invariance and the Gravitational Field, J. Math. Phys. 2, 212
(1961), https://doi.org/10.1063/1.1703702.

[5] D. W. Sciama, The Physical Structure of General Relativity, Rev. Mod. Phys. 36, 463
(1964), 463–469.

[6] M. Blau, Lecture Notes on General Relativity.

[7] L. P. Eisenhart, Non-Riemannian Geometry, New York, 1927.

[8] L. P. Eisenhart, Generalized Riemannian Spaces I, Proc. Nat. Acad. Sci. USA, 37 (1951),
311–315.

[9] L. P. Eisenhart, Generalized Riemannian spaces II, Proc. Nat. Acad. Sci. USA, 38 (1952),
505–508.

[10] M. S. Madsen, Scalar fields in curved spacetimes, Class. Quantum Grav. 5 (1988), 627–639.

[11] J. Mikeš, E. Stepanova, A. Vanžurova, et al., Differential geometry of special map-

pings, Olomouc: Palacky University, 2015.

[12] S. M. Minčić, Curvature tensors of the space of non-symmetric affine connexion, obtained
from the curvature pseudotensors, Matematički Vesnik, 13 (28) (1976), 421–435.

[13] S. M. Minčić, Independent curvature tensors and pseudotensors of spaces with non-
symmetric affine connexion, Coll. Math. Soc. János Bolayai, 31. Dif. geom., Budapest (Hun-
gary), (1979), 445–460.

[14] S. M. Minčić, On Ricci Type Identities in Manifolds With Non-Symmetric Affine Connec-
tion, PUBLICATIONS DE L’INSTITUT MATHÉMATIQUE Nouvelle série, tome 94 (108)
(2013), 205–217.

[15] V. N. Ponomarev, A. O. Barvinsky, Y. N. Obukhov, Gauge Approach and Quan-

tization Methods in Gravity Theory, Nuclear Safety Institute of the Russian Academy of
Sciences, Nauka, Moscow, 2017.

[16] I. L. Shapiro, Physical Aspects of the Space-Time Torsion, Physics Reports, Volume 357,
Issue 2, January 2002, 113–213.

[17] N. S. Sinyukov, Geodesic mappings of Riemannian spaces, (in Russian), "Nauka", Moscow,
1979.

[18] Lj. S. Velimirović, S. M. Minčić, M. S. Stanković, Infinitesimal Deformations of

Basic Tensor in Generalized Riemannian Space, Filomat 21:2 (2007), 235–242.

[19] Lj. S. Velimirović, S. M. Minčić, M. S. Stanković, Infinitesimal Deformations of

Curvature Tensors at Non-Symmetric Affine Connection Space, Matematički Vesnik, 54
(2002), 219–226.

[20] M. Lj. Zlatanović, New projective tensors for equitorsion geodesic mappings, Applied
Mathematics Letters 25 (2012), No. 5, 890–897.

19