COSMOLOGICAL EVOLUTION WITH NEGATIVE

ENERGY DENSITIES
A. A. SAHARIAN1,2∗, R. M. AVAGYAN1,2 , E. R. BEZERRA DE MELLO3 ,
V. KH. KOTANJYAN1,2 , T. A. PETROSYAN1,2 , H. G. BABUJYAN1
arXiv:2205.01619v3 [gr-qc] 8 Aug 2022

1
Department of Physics, Yerevan State University,
1 Alex Manogian Street, 0025 Yerevan, Armenia
2
Institute of Applied Problems of Physics NAS RA,
25 Hr. Nersessian Street, 0014 Yerevan, Armenia
3
Departamento de Fı́sica, Universidade Federal da Paraı́ba,
58059-900, Caixa Postal 5008, João Pessoa, PB, Brazil

August 10, 2022

Abstract
For general number of spatial dimensions we investigate the cosmological dynamics driven
by a cosmological constant and by a source with barotropic equation of state. It is assumed
that for both those sources the energy density can be either positive or negative. Exact
solutions of the cosmological equations are provided for flat models. For models with curved
space and with zero cosmological constant the general solutions are expressed in terms of the
hypergeometric function. The qualitative evolution is described for all values of the equation
of state parameter. We specify the values of that parameter and the combinations of the
signs for the cosmological constant and matter energy density for which the cosmological
dynamics is nonsingular. An example is considered with positive cosmological constant and
negative matter energy density induced by the polarization of the hyperbolic vacuum.

Keywords: cosmological evolution, cosmological constant, negative energy density

1 Introduction
The investigation of cosmological dynamics is carried out mainly within the framework of homo-
geneous and isotropic models described by Friedmann-Robertson-Walker (FRW) line element.
In particular, the models containing a positive cosmological constant in addition to the matter
and radiation sources of the expansion have been actively studied. This theoretical activity
is motivated by the observational evidence [1, 2] for accelerated expansion of the universe in
recent epoch driven by a source (dark energy) with properties close to a positive cosmological
constant. The cosmological model with a positive cosmological constant and cold dark matter
(CDM) in addition to the usual matter (ΛCDM model) is in good agreement with observational
data on the large scale structure and dynamics of the universe. Recently a problem appeared
∗
E-mail: saharian@[Link]

1
that is related to the value of the Hubble parameter H0 at present determined by two differ-
ent ways. The first one is based on direct low redshift observations [3]-[6] and gives the value
H0 ≈ 73 km/s/Mpc. The second way combines the Planck data on temperature anisotropies
of the cosmic microwave background radiation [7] with the ΛCDM model and gives the result
H0 ≈ 67 km/s/Mpc. A number of models have been discussed in the literature to address this
problem, also called Hubble tension (for a review see [8]). In particular, they include the models
with negative cosmological constant (see [9]-[13]). The maximally symmetric solution of the
Einstein field equations with a negative cosmological constant as the only source of the grav-
itation is given by anti-de Sitter (AdS) spacetime. This geometry appears as a ground state
in string theories and in supergravity. It plays an important role in braneworld models with
large extra dimensions and in holographic duality models relating two theories living in different
numbers of spatial dimensions. An example of the latter is the AdS/CFT correspondence (see,
for example, [14]) establishing the duality between supergravity and string theories on the AdS
bulk and conformal field theory on the AdS boundary.
Another example for a gravitational source with the negative energy density, that can play
an important role in the expansion of the early universe, is provided by the vacuum polarization.
The vacuum expectation value of the energy-momentum tensor for quantum fields may break
the energy conditions of the singularity theorems in general relativity (see, e.g., [15]). This
can serve as the key for solving the singularity problems in the cosmological dynamics. Here
we consider the cosmological dynamics for both cases of positive and negative energy densities.
Various combinations of cosmological constant and of a source with barotropic equation of
state will be studied. Having in mind possible applications in higher-dimensional models, in
particular, motivated by string theories, the discussion is presented for a general number of
spatial dimensions. The qualitative evolution in cosmological models with scalar fields having
negative potentials has been considered in [16]-[19]. Various cases of exact solutions to Friedmann
equations in general number of spatial dimensions were discussed in [20] by using Chebyshev’s
theorem. Cosmological solutions in (3+1)-dimensional spacetime with a single positive and
negative energy component in a flat universe and for a negative energy component in a curved
universe have been described in [21].
The present paper is organized as follows. In the next section we present the cosmological
equations and some qualitative features. The solutions for flat model with a cosmological con-
stant and barotropic matter are given in section 3. They serve as past or future attractors for
models with curved space and include various special cases previously considered in the litera-
ture. In section 4 we discuss models with curved space. First, the general solutions are presented
in terms of the hypergeometric function for models with zero cosmological constant. Various
special cases where the time-dependence of the scale factor is expressed in terms of elementary
functions were discussed in the literature. Then we describe the qualitative evolution in models
with curved space driven by a cosmological constant and barotropic matter source.

2 Cosmological equations
We consider (D + 1)-dimensional background spacetime described by the FRW line element

dr 2
 
2 2 2 2 2 2
ds = N (t)dt − a (t) + r dΩD−1 , (1)
1 − kr 2

where dΩ2D−1 is the line element on a unit sphere S D−1 and k = 0, ±1. The choices N (t) = 1
and N (t) = a(t) correspond to the synchronous and conformal time coordinates, respectively.
Depending on the equation of state the first or the second choice of the time coordinate is

2
convenient to present the cosmological solutions in simpler form. Assuming that the dynamics
is governed by General Relativity in (D + 1)-dimensional spacetime, the set of cosmological
equations takes the form
  !
d ȧ ȧ ȧ Ṅ k 8πGD 2
+ D − + (D − 1)N 2 2 = N (ε − p) ,
dt a a a N a D−1
 2
ȧ N 2k 16πGD
+ 2 = N 2 ε, (2)
a a D(D − 1)
where the dot stands for the derivative with respect to t, GD is the gravitational constant in
(D+1)-dimensional spacetime, ε is the energy density and p is the pressure for the sources driving
the cosmological evolution. The latter two quantities obey the equation ε̇ + D(ȧ/a)(ε + p) = 0
which is obtained from the covariant conservation equation for the energy-momentum tensor.
This relation can also be obtained from (2). For the second derivative of the scale factor we get
 
ä Ṅ ȧ 8πGD 2 D−2
− =− N p+ ε . (3)
a Na D−1 D
From this relation it follows that the accelerated expansion in terms of the synchronous time
coordinate (N (t) = 1) is obtained under the condition p < (2 − D)ε/D. The latter condition is
satisfied by the positive cosmological constant Λ with the energy density εΛ = Λ/(8πGD ) and
pressure pΛ = −εΛ .
In the discussion below we assume that the matter source contains two parts with ε = εΛ +εm
and p = pΛ + pm . Here, the part with the equation of state pΛ = −εΛ corresponds to the
cosmological constant Λ with the constant energy density εΛ and the second contribution has
an equation of state pm = wεm with w =const. The condition p < (2 − D)ε/D for the second
source is reduced to w < wc ≡ 2/D − 1 for εm > 0 and to w > wc for εm < 0. From the covariant
conservation equation we get
εm = εm0 (a/a0 )−α , (4)
with the notation
α = D(1 + w). (5)
We will assume that the cosmological constant Λ and the constant εm0 = εm |a=a0 can be either
positive or negative. Note that from the second equation in (2) it follows that one needs to have
the condition 16πGD ε ≥ D(D − 1)ka−2 and the total energy density ε should be nonnegative
in models with k = 0, 1.
Let us consider the qualitative features of the evolution in terms of the synchronous time.
Taking N (t) = 1, the second equation in (2) is rewritten as
k 2Λ 16πGD εm0
H2 + = + , (6)
a2 D(D − 1) D(D − 1) (a/a0 )α
where H = ȧ/a is the Hubble function. From here it follows that for w > −1 and for a positive
cosmological constant the late time evolution (large values of the scale factor) is dominated by
the first term in the right hand side. In this case the de Sitter solution a(t) ∝ eHΛ t , with
s
2|Λ|
HΛ = (7)
D(D − 1)

(here we consider the case Λ > 0, the notation HΛ for Λ < 0 is used below), is the future
attractor for the general solution. For a negative cosmological constant, Λ < 0, and for w > −1,

3
from (6) we see that with increasing a at some moment t = tm the Hubble function becomes
zero. The corresponding value for the scale factor a = am is determined from (6) putting H = 0.
At that moment from the first equation (2) we get
w − wc 1+w
Ḣt=tm = −Dk + Λ. (8)
2a2m D−1

For k = 0, 1 and w > wc the right-hand side is negative and for t > tm one obtains H < 0
and the initial expansion is followed by the contraction. The same is the case for k = −1 and
−1 < w < wc . For α > 2 and εm0 > 0, the early expansion, corresponding to small values of
the scale factor, is dominated by the matter source and the solutions with flat space serve as
attractors for models with k = ±1.

3 Cosmological solutions in flat model

Simple exact solutions of the cosmological equations can be found in the case of flat model,
k = 0. In the absence of the matter source the equation (6) has solutions only for Λ ≥ 0. For
positive cosmological constant the de Sitter solution, a(t) ∝ e±HΛ t , is obtained. To see the
influence of the matter source, first we consider the case of positive cosmological constant and
positive matter density, corresponding to εm0 , εΛ > 0. In the synchronous time coordinate, for
the Hubble function we get

εm0 1/α
r  a α a
m m
H = ±HΛ 1 + , ≡ , (9)
a a0 εΛ

with α defined by (5). The integration of this equation leads to the following expressions for the
Hubble function and the scale factor:

H(t) = ±HΛ coth (β |t|) , a(t) = am sinh2/α (β |t|) , (10)

where s
1 D|Λ|
β = |α|HΛ = |1 + w| . (11)
2 2(D − 1)
For w > −1 the solution (10) for the scale factor coincides with that found in [20]. In that
case and for expansion models one has 0 < t < ∞ with the upper sign in the expression for
the Hubble function. At late times, βt ≫ 1, one has an approximately de Sitter expansion
with a(t) ∝ eHΛ t . Near the singularity point t = 0 we obtain a(t) ∝ |t|2/α . The case w < −1
corresponds to the phantom phase (for the effective phantom phase generated by different types
of sources see [22]). In this case α < 0 and for the expansion models we have −∞ < t < 0. The
point t = 0 corresponds to the Big Rip singularity. The universe starts with de Sitter expansion
a(t) ∝ eHΛ t , β |t| ≫ 1, in the infinite past and ends the evolution at Big Rip singularity at t = 0
with the behavior a(t) ∝ |t|−2/|α| . In figure 1 we have plotted the ratio a/am versus HΛ t for
D = 3. The full and dashed curves correspond to the values w = 0 (dust matter), w = −2/3 and
w = −3/2 (phantom matter). Note that under certain conditions (see [23]) the energy density
for the axion field scales as εaxion ∼ 1/a3 and the corresponding dynamics is described by the
curve with w = 0 in figure 1 (the cosmological dynamics with the axion field and holographic
dark energy has been recently discussed in [24]). For expanding models we have 0 < t < +∞
for sources with w > −1 and −∞ < t < 0 for w < −1. The singular point t = 0 corresponds to
the Big Bang in the first case and to the Big Rip in the second case. For w > wc and w < −1

4
 4

w=0 w=-3/2

a/am
2

w=-2/3
1

0
-3 -2 -1 0 1 2 3
HΛ t

Figure 1: The time dependence of the scale factor in the model with εm0 , εΛ > 0 for D = 3 and
w = 0, −2/3, −3/2.

one has ȧ|t=0 = ∞ and for −1 < w < wc we get ȧ|t=0 = 0. We see that for εm0 , εΛ > 0 all the
flat models contain singularities.
Next we consider the case εΛ < 0 < εm0 . For the Hubble function we find
q
H = ±HΛ (am /a)α − 1. (12)

The time dependences for the Hubble function and for the scale factor read

H(t) = ±HΛ tan (β|t|) , a(t) = am cos2/α (β|t|) , (13)

with −π/2β < t < π/2β. For w > −1 this solution coincides with that presented in [20]. The
authors of [20] emphasize that the solution (13) gives rise to a periodic universe. However, it
should be noted that, though the function a(t) in (13) is periodic with the period tL = π/β,
the periods are separated by singular points |t| = π(l + 1/2)/β, l = 0, 1, 2, . . ., and the evolution
pieces separated by those points present the copies of the same universe with a finite lifetime
tL (for discussion of various types of singularities in the cosmological context see, for example,
[25, 26, 27]). The dependence of the scale factor on the synchronous time coordinate, described
by (13), is depicted in figure 2 for D = 3 and w = 0, −2/3, −3/2. In models with w > −1
the expansion phase with −π/2β < t < 0 is followed by the contraction one for 0 < t < π/2β.
The maximal value of the scale factor is determined by (9). For sources with w < −1 the same
relation determines the minimal value of the scale factor. Similar to the previous case, the flat
models contain singularities for all values of the parameters.
Now we turn to the case εm0 < 0 < εΛ . The Hubble function is expressed as
q
H = ±HΛ 1 − (am /a)α , (14)

where am is the minimal (maximal) value of the scale factor for w > −1 (w < −1). The time
dependence is given by the formulas

H = ±HΛ tanh (β |t|) , a = am cosh2/α (β |t|) , (15)

5
 4

w=-3/2
3

a/am
2

1
w=-2/3
w=0
0
-3 -2 -1 0 1 2 3
HΛ t

Figure 2: The same as in figure 1 for the model mith εΛ < 0 < εm0 .

with −∞ < t < +∞. The time dependence of the scale factor given by (15) is plotted in figure
3 for the values of the parameters D = 3 and w = 1/3, 0, −2/3, −3/2. The models in this case
have no singularities. The value a = am determines the minimum/maximum value of the scale
factor. Note that flat cosmological models with εm0 , εΛ < 0 are not allowed by the equation (6).
The corresponding models with curved space will be discussed in the next section.
5

w=0
4
w=1/3

3
a/am

2 w=-2/3

1
w=-3/2

0
-3 -2 -1 0 1 2 3
HΛ t

Figure 3: The scale factor versus the time coordinate in the model with εm0 < 0 < εΛ for D = 3
and w = 1/3, 0, −2/3, −3/2.

In [28, 29] it has been shown that in de Sitter spacetime the vacuum expectation value of
the energy-momentum tensor for a conformally coupled massless scalar field in the hyperbolic
vacuum has the form
 
D E
k CD 1 1
Ti = εΛ diag (1, 1, · · · , 1) + D+1 diag 1, − , · · · , − , (16)
a D D

6
where the first term in the right-hand side corresponds to a cosmological constant and the
constant CD is negative. The second term can be identified with the source we have considered
above having the equation of state pm = εm /D. Hence, for this source one gets w = 1/D
and εm0 < 0. The corresponding cosmological solution is given by (15) with α = D + 1 and
β = (D + 1) HΛ /2. The time dependence of the scale factor for D = 3 is presented in figure 3
by the curve with w = 1/3. The corresponding models are nonsingular.

4 Cosmological dynamics in models with curved space

Passing to the models with curved space, first let us recall the well-known solutions in the
absence of matter sources. The models with k = 1 are allowed only in the case Λ > 0 and the
corresponding solution for the scale factor is given by a = cosh (HΛ t) /HΛ . For k = −1 and
Λ > 0 the solution reads a = sinh (HΛ |t|) /HΛ . For k = −1 and negative cosmological constant
we have the solution a = sin (HΛ |t|) /HΛ . Note that in models with k = −1 and HΛ |t| ≪ 1 the
evolution is approximated by linear scale factor a(t) = |t|. The latter describes a flat spacetime
and corresponds to the Milne universe.
Another special case corresponds to the absence of cosmological constant. From the equation
(6) we get
dy p
= ± γ0 y γ − k, (17)
d(t/a0 )
with the notations
a 16πGD εm0 a20
y= , γ0 = , γ = D(wc − w). (18)
a0 D(D − 1)
Separating the variables, the integrals in (17) can be expressed in terms of the incomplete beta
function Bz (u, v). Presenting the latter through the hypergeometric function F (a, b; c; z) (see,
for example, [30]), for the models with k = −1 we find

γ0 y γ
 
a 1 1
t= √ γ F , 1; 1 + , . (19)
γ0 y + 1 2 γ γ0 y γ + 1

In a similar way, for the models with k = 1 the integration gives

 1/2  
2a0 1 1 1 1 3 1
t= 1− F , + ; ;1 − . (20)
1/γ
γγ0 γ0 y γ 2 2 γ 2 γ0 y γ

The various special cases of these general formulas have been considered in the literature. In
particular, the examples when for general number of spatial dimension the solutions are expressed
in terms of elementary functions have been discussed in [20].
Now we turn to the general case of models with curved space in the presence of a cosmological
constant and barotropic matter. The equation (6) is rewritten as

dx p
= ±HΛ sΛ x2 + bxγ − k, (21)
dt
where
x = HΛ a, sΛ = sgn(Λ), (22)
and
16πGD εm0 a20
b= (HΛ a0 )γ . (23)
D(D − 1)

7
Simple solutions are found for the special√ case of the source with w = wc . For Λ > 0 and
γ0 − k > 0 the solution has
√ the form x = γ0 − k sinh (HΛ |t|). In the case Λ > 0 and γ0 − k < 0,
the solution reads x = k − γ0 cosh (HΛ√t). For Λ < 0 one needs to have γ0 − k > 0 and the
corresponding solution is given by x = γ0 − k sin (HΛ t), 0 < t < π/HΛ . For γ0 = 0 the first
two solutions are reduced to the de Sitter solutions.
We will denote by x = xm > 0 the value of the function x(t) at its possible extremum,
dx/dt|x=xm = 0. The extrema are zeros of the expression under the square root in (21). Taking
the corresponding value of the time coordinate as t = 0 and expanding near the extremum we
get  
a(t) D wc − w
≈1+ sΛ (1 + w) + k (HΛ t)2 , (24)
am 4 x2m
where am = xm /HΛ . The nature of the extremum (minimum or maximum) is determined
by the sign of the expression in the square brackets. Note that for the extremum we have
bx−α 2
m = k/xm −sΛ . In the definition of the constant b we have taken a0 = a(t0 ) and εm0 = εm (t0 )
for a fixed time t = t0 . Taking t0 = tm , where tm corresponds to the extremal value xm ,
x(tm ) = xm , from (6) we get the following relation

sΛ k
x2m = , (25)
1 + ε(m) /εΛ

where ε(m) = ε(tm ) is the matter energy density at the extremum point. Note that, assuming
the presence of the extremum x = xm , the equation (21) is written as

ε(m) γ
r
dy
= ±HΛ sΛ (y 2 − 1) + (y − 1), (26)
dt |εΛ |

with y = x/xm = a/am .

Let us consider different combinations of the signs for the energy densities. For εm0 , εΛ > 0
and w > wc , the early dynamics, corresponding to small values of x, is dominated by the source
with the energy density εm and the expansion law is close to the one for the flat model. At
late times, corresponding to x ≫ 1, the expansion is dominated by the cosmological constant
and, again, the curvature term is subdominant. The solution corresponding to the flat model is
the future attractor for models with curved space. The dependence of the scale factor on time
coordinate is qualitatively similar to that depicted in figure 1 for w = 0.
For εm0 , εΛ > 0, w < wc , and k = −1, the early dynamics (x ≪ 1) for expanding models
is dominated by the curvature term and a(t) ≈ t, t → 0. As it has been mentioned above,
the spacetime with k = −1 and a(t) = t is flat and corresponds to the Milne universe. The
matter energy density behaves as εm ∝ x−α and for w > −1 it diverges at t = 0 like εm ∼ t−α .
In the model with εm0 , εΛ > 0, w < wc , and k = 1 the scale factor has a minimal value that
corresponds to the zero x = xm of the expression in the right-hand side of (21). At this point
the Hubble function becomes zero. The time-dependence of the scale factor near the minimum,
HΛ t ≪ 1, is given by (24) with sΛ = 1 and am = amin . At late times of the expansion, x ≫ 1,
the curvature term in (21) can be ignored and the cosmological dynamics is well approximated
by the solutions for flat model (see the graphs with w = −2/3, −3/2 in figure 1). We conclude
that the models with εm0 , εΛ > 0, w < wc and k = 1 are nonsingular.
Let us turn to the models with εΛ < 0 < εm0 . For w > wc the maximum allowed value for x
is determined by the zero x = xm of the right-hand side in (21). The asymptotic behavior near
the maximum is described by (24) with sΛ = 1 and am = amax . For x ≪ 1, in the right-hand
side of (21) we can omit the curvature term and x2 . The scale factor is approximated by the
solution for the flat model and near the Big Bang, corresponding to t = −t1 , t1 > 0, one has

8
a(t) ∝ (t + t1 )2/α . The model has finite lifetime 2t1 and the corresponding time-dependence of
the scale factor is qualitatively similar to that for the flat model presented by the graph with
w = 0 in figure 2.
For εΛ < 0 < εm0 and −1 < w < wc the function x(t) has a maximal allowed value x = xm
determined by the zero of the right-hand side in (21). Taking x(0) = xm , near the maximum
point we have the approximation (24) with sΛ = 1 and am = amax . For k = −1, the models
start the expansion at t = −t1 with the scale factor a(t) ≈ t + t1 and the behavior of the scale
factor is close to the one for the Milne universe. The expansion is stopped at t = 0 and for t > 0
the model enters the contraction phase. The latter is ended at t = t1 with a(t) ≈ t1 − t. Hence,
the k = −1 models have lifetime 2t1 and the Milne universe is the past and future attractor
for the corresponding dynamics. Note that, though the first derivative of the scale factor is
finite at the points t = ±t1 (|ȧ|t=±t1 = 1), the matter energy density diverges at those points
as εm ∼ 1/|t ± t1 |D(1+w) . The models with k = 1 start the expansion from the finite value
of the scale factor amin at t = −tmin . At that point ȧ(−tmin ) = 0. At t = 0 the scale factor
takes its maximal value amax = xm /HΛ and then it enters into the contraction phase. Near the
maximum we have the approximation (24). The evolution is ended at t = tmin with a = amin
and ȧ(tmin ) = 0. Hence, in this case we have nonsingular evolution for −tmin ≤ t ≤ tmin . Joining
the evolutionary pieces with duration 2tmin , we obtain a model with periodically oscillating scale
factor in the limits amin ≤ a ≤ amax for −∞ < t < +∞.
In models with εΛ < 0 < εm0 , w < −1, and for large values of x the expansion law is close to
the one for the flat model and the corresponding behavior is qualitatively close to the one given
by the curve with w = −3/2 in figure 2. For small values of x and for models with k = −1 the
expansion/contraction law is approximated by a(t) ≈ |t|. At t = 0 the matter energy density
vanishes as εm ∼ |t|D|1+w| . In models with k = 1 the scale factor has a minimum value a = amin
determined by the zero of the right-hand side in (21) and the evolution for all values of x ≥ xmin
is qualitatively similar to that described by the curve with w = −3/2 in figure 2. The expansion
models have Big Rip singularity.
Now let us consider models with the energy densities in the range εm0 < 0 < εΛ . For w > wc
the scale factor has a minimal value a = amin which is determined by the zero of the right-
hand side in (21). Taking t = 0 for the corresponding value of the time coordinate, near the
minimum one has the approximation (24) with am = amin and xm = xmin . For w > wc and for
large values of x the evolution is approximated by de Sitter spacetime with the Hubble constant
HΛ . The behavior of the scale factor is qualitatively similar to that depicted in figure 3 by
the curves with w = 0, 1/3 and the corresponding models have no singularities. An example
with positive cosmological constant, negative matter energy density and the equation of state
parameter w = 1/D > wc is provided by (16). In the range −1 < w < wc and for large values of
x the evolution is again dominated by the cosmological constant with de Sitter spacetime being
the past or future attractor. In the same range for w and for k = −1 one gets the approximate
solution a(t) ≈ |t| for HΛ |t| ≪ 1, corresponding to the Milne universe. The matter energy
density diverges at t = 0. For −1 < w < wc and k = 1 the scale factor has a minimal value
determined by the zero of the right-hand side of (21). Near that minimum the scale factor is
approximated by (24) and the model is nonsingular. In the range w < −1 the scale factor has the
maximal value amax which is given by the zero of the right-hand side in (21). For models with
k = −1 the expansion starts at t = −t1 with the asymptotic a(t) ≈ t + t1 (curvature dominated
expansion) and ends at t = 0 with the asymptotic given by (24). The expansion phase is followed
by the contraction for 0 < t < t1 with a(t) ≈ t1 − t near t = t1 . For models with k = 1 the scale
factor varies between two nonzero values 0 < amin ≤ a ≤ amax < ∞. The corresponding models
are nonsingular and can be extended for t ∈ (−∞, +∞). The qualitative dynamics is similar to
that we have described above for the case k = 1, εΛ < 0 < εm0 , −1 < w < wc .

9
 Finally, for εm0 , εΛ < 0, in accordance with (6), the models with k = 0 and k = 1 are not
allowed. Let us consider the features of the cosmological dynamics in this case for k = −1. For
w > wc , from the condition for the positivity of the expression under the square root in (21),
we can see that the model is allowed under the constraint
w − wc α/2
 
2
|b| < . (27)
D (w − wc ) w + 1
This condition restricts the allowed values for the negative energy density εm0 . In the range
determined by (27), the right-hand side of (21) has two zeros and they determine the minimal
and maximal values for the scale factor, 0 < amin ≤ a(t) ≤ amax . At those points ȧ = 0 and
H = 0. Near the extrema the scale factor is approximated by (24) with sΛ = −1 and k = −1.
From (24) it follows that r
1 w − wc
amin < < amax . (28)
HΛ 1+w
For εm0 , εΛ < 0 and w < wc the right hand side of (21) has a single zero that determines the
maximal value of the scale factor amax = a(0). Near the maximum the scale factor behaves
like (24) with sΛ = −1 and k = −1. For small values of x the dynamics is dominated by the
curvature term with the Milne universe as the asymptotic. The expansion starts at t = −t1 with
a(t) ≈ t + t1 and stops at t = 0 with the maximal value of the scale factor. The evolution for
0 < t < t1 corresponds to the contraction phase with the future attractor a(t) ≈ t1 − t. At the
points t = ±t1 the matter energy density vanishes for w < −1 and diverges for −1 < w < wc .

5 Conclusion
We have considered the dynamics of (D + 1)-dimensional FRW cosmological models driven by
the cosmological constant and the matter source with barotropic equation of state assuming that
the energy densities for those sources can be either positive or negative. Exact solutions are
provided for models with flat space which include various special cases previously considered in
the literature. In particular, it has been demonstrated that nonsingular solutions are obtained
only for negative energy density of the matter, regardless the sign of the cosmological constant.
The corresponding scale factor is given by (15). Another classes of exact solutions, expressed in
terms of the hypergeometric function (see (19) and (20)), are obtained for models with curved
space in the absence of cosmological constant. A number of special cases of those solutions,
when they are expressed in terms of elementary function, have been discussed in the literature
(see, for example, [20]). The qualitative evaluation for models with curved spaces and with a
cosmological constant and matter source has been described in the second part of section 4 for all
the values of the equation of state parameter w and for all combinations of the signs of the energy
densities. Depending on the values of w one can have Big Bang or Big Rip type singularities. We
have also specified nonsingular models with curved space. For k = 1, nonsingular modelas are
obtained for the following combinations of conditions: (i) (ε0m > 0, εΛ > 0, w < wc ), (ii) (ε0m >
0, εΛ < 0, −1 < w < wc ), (iii) (ε0m < 0, εΛ > 0). In models (ii ) and (ε0m < 0, εΛ > 0, w > −1)
the evolution of the scale factor, as a function of time coordinate, is periodically oscillatory
in the limits amin ≤ a(t) ≤ amax . In the remaining cases, the qualitative evolution of k = 1
nonsingular models is similar to that depicted in figure 3 for w = −2/3, 0, 1/3. For models with
negative curvature space there exists at least one point on the time axis where the scale factor
becomes zero. Near those points the evolution is dominated by the matter source for w > wc
and by the curvature term for w < wc . In the second case the scale factor is approximated by a
linear expansion/contraction as a function of the time coordinate. At the point with zero scale
factor the matter energy density diverges for −1 < w < wc and vanishes for w < −1.

10
 We have seen that the negative energy densities for both the cosmological constant and mat-
ter source enlarge the possible scenarios of cosmological dynamics. Bearing in mind applications
in higher-dimensional models, it would be interesting to generalize the corresponding results for
models with extra compact dimensions. The compactification leads to additional contributions
to the vacuum expectation value of the energy-momentum tensor. In general, the effective pres-
sures along compact dimensions differ and for massless conformally coupled fields the topological
contributions are equivalent to barotropic perfect fluid with anisotropic pressures. In particular,
the coefficients w in the respective equations of state may have different signs. In the corre-
sponding anisotropic cosmological models one can have an expansion for a part of dimensions
and a contraction for the remaining ones. The analysis of different cosmological scenarios can
be done in a way similar to that we have described above. We can also use the methods of
qualitative analysis of dynamical systems to classify qualitatively different cosmological models.
The corresponding results for a toroidal compactification will be presented elsewhere.

Acknowledgments
A.A.S., R.M.A.,T.A.P. were supported by Grants No. 20RF-059, No. 21AG-1C047 and No.
20AA-1C005 of the Science Committee of the Ministry of Education, Science, Culture and
Sport RA. E.R.B.d.M. is partially supported by CNPQ under Grant No. 301.783/2019-3.

References
[1] A. G. Riess, A. V. Filippenko, P. Challis et al., Astron. J., 116, 1009, 1998.

[2] S. Perlmutter, G. Aldering, G. Goldhaber et al., Astrophys. J., 517, 565, 1999.

[3] W. L. Freedman, B. F. Madore, D. Hatt et al., Astrophys. J., 882, 34, 2019.

[4] K. C. Wong, S. H. Suyu, G. C.-F. Chen et al., Mon. Not. Roy. Astron. Soc., 498, 1420,
2020.

[5] D. W. Pesce, J. A. Braatz, M. J. Reid, Astrophys. J. Lett., 891, L1, 2020.

[6] A. G. Riess, W. Yuan, L. M. Macri et al., A comprehensive measurement of the local value
of the Hubble constant with 1 km/s/Mpc uncertainty from the Hubble Space Telescope and
the SH0ES team, arXiv:2112.04510.

[7] Planck Collaboration, et al., Planck 2018 results. VI. Cosmological parameters, Astron.
Astrophys. 641, A6, 2020.

[8] W. L. Freedman, Astrophys. J., 919, 16, 2021.

[9] L. Visinelli, S. Vagnozzi, U. Danielsson, Symmetry, 11, 1035, 2019.

[10] O. Akarsu, J. D. Barrow, L. A. Escamilla, J. A. Vazquez, Phys. Rev. D 101, 063528, 2020.

[11] K. Dutta, A. Roy, Ruchika et al., Gen. Rel. Grav. 52, 15, 2020.

[12] R. Calderón, R. Gannouji, B. L’Huillier, D. Polarski, Phys. Rev. D 103, 023526, 2021.

[13] A. A. Sen, S. A. Adil, S. Sen, Do cosmological observations allow a negative Λ?,

arXiv:2112.10641.

11
[14] H. Năstase, Introduction to AdS/CFT Correspondence (Cambridge University Press, Cam-
bridge, England, 2015).

[15] N. D. Birrell, P. C. W. Davies, Quantum Fields in Curved Space (Cambridge University

Press, Cambridge, 1982).

[16] A. A. Saharian, Astrophysics, 40, 335, 1997.

[17] A. A. Saharian, V. L. Sargsyan, Astrophysics, 42, 350, 1999.

[18] A. A. Saharian, Astrophysics, 45, 380, 2002.

[19] G. Felder, A. Frolov, L. Kofman, A. Linde, Phys. Rev. D, 66, 023507, 2002.

[20] S. Chen, G. W. Gibbons, Y. Lic, Y. Yang, JCAP 12(2014)035.

[21] R. J. Nemiroff, R. Joshia, B. R. Patla, JCAP 06(2015)006.

[22] E. Elizalde, S. Nojiri, S. D. Odintsov, P. Wang, Phys. Rev. D, 71, 103504, 2005.

[23] S. D. Odintsov, V. K. Oikonomou, Phys. Rev. D, 101, 044009, 2020.

[24] A. A. Saharian, A. V. Timoshkin, Int. J. Geom. Meth. Mod. Phys., 18, 2150174, 2021.

[25] S. Nojiri, S. D. Odintsov, S. Tsujikawa, Phys. Rev. D, 71, 063004, 2005.

[26] K. Bamba, S. Capozziello, S, Nojiri, S. D. Odintsov, Astrophys. Space Sci., 342, 155, 2012.

[27] A. Yu. Kamenshchik, Class. Quantum Grav., 30, 173001, 2013.

[28] A. A. Saharian, T. A. Petrosyan, Phys. Rev. D, 104, 065017, 2021.

[29] A. A. Saharian, T. A. Petrosyan, V. S. Torosyan, Ann. Phys., 437, 168728, 2022.

[30] F.W. Olver et al., NIST Handbook of Mathematical Functions (Cambridge University
Press, USA, 2010).

12