Downloaded from https://iranpaper.

ir
https://www.tarjomano.com/order

Eur. Phys. J. Plus (2021) 136:4

https://doi.org/10.1140/epjp/s13360-020-01007-1

Regular Article

Complete set of GW polarization modes in

higher-derivative f (R, R, T ) theories of gravity

Mojtaba Haghshenas, Tahereh Azizia

Department of Theoretical Physics, Faculty of Basic Sciences, University of Mazandaran, Babolsar
47416-95447, Iran

Received: 24 October 2020 / Accepted: 8 December 2020

© Società Italiana di Fisica and Springer-Verlag GmbH Germany, part of Springer Nature 2021

Abstract In this paper, we investigate the gravitational waves in a particular class of alter-
native theories of gravity which is known as higher-derivative f (R, R, T ) gravity theory.
In low curvature regime and by using Newman–Penrose analysis for the linear order of the
field equations, we show that in addition to the usual plus and cross polarization modes which
appear in general relativity, all other modes can exist too. Actually, according to our analysis,
in such a theory by choosing the proper values from the parameter space of the model, a
complete set of gravitational wave polarization modes can be produced in this concept.

1 Introduction

The detection of gravitational waves (GWs) by the LIGO Scientific Collaboration and Virgo
Collaboration has opened a new window to study the gravitational physics and astrophysics
[1,2]. In particular, the observation of the polarization state of GWs is an important issue to test
the general theory of relativity (GR) and alternative theories of gravity. In GR the gravitational
waves exhibit two tensor polarization states, the so-called plus and cross modes. In contrast,
in the context of the general metric theories of gravity, there are six possible polarization
modes for GWs which correspond to the six independent Newman–Penrose (NP) quantities
Ψ2 , Ψ3 , Ψ4 and Φ22 . The resulting modes are two tensor polarization modes (the cross and
plus modes), two vector modes the so-called vector-x and vector-y polarizations and two
scalar polarization modes that include the longitudinal polarization and transverse scalar
polarization mode [3]. Therefore, the detection of the polarization states of GWs can be
considered as a new tool to probe the nature of gravity. It is worth pointing out that, until
now, the Ligo and Virgo observatories have detected only the pure tensor polarization modes
that correspond to the usual GR modes. Nevertheless, with the advent of future space-borne
detectors, there will exist a better chance to access an unprecedented signal sensitivity, hence
it is worthwhile to explore GWs in alternative theories of gravity [4]. The discovery of the
accelerating expansion of our universe has motivated a lot of research on the modification
of GR as a possible approach to describe the current cosmic speed-up [5,6]. One way to
modify gravity is to consider a general function of the Ricci scalar in the Einstein–Hilbert
action which is dubbed as f (R) gravity. It is shown that this modified theory can explain

a e-mail: [email protected] (corresponding author)

0123456789().: V,-vol 123

 Downloaded from https://iranpaper.ir
https://www.tarjomano.com/order

4 Page 2 of 17 Eur. Phys. J. Plus (2021) 136:4

the accelerating phase of the universe without the need to introduce any exotic matter such
as dark energy [7]. Therefore, an extensive amount of research is carried out to explore
the gravitational and cosmological implication of f (R) gravity (see for instance [8] and
references therein for some reviews on this subject).
A further generalization of f (R) gravity that includes a coupling between matter and
geometry has been proposed by Harko et al. in which the Lagrangian density is defined
by an arbitrary function f (R, T ) where T is the trace of the stress-energy tensor [9]. The
presence of T in the Lagrangian may be induced by exotic imperfect fluids or quantum
effects (conformal anomaly). Different implications of this model have been discussed in the
literature for example, the coupling of the matter and geometry leads to the presence of an
extra force perpendicular to the four velocity. As a result, the motion of a massive particle is
not along the geodesic path. Moreover, the accelerating phase in f (R, T ) gravity theory can
be deduced not only from the geometrical contribution but also from the matter content of
the universe. In a recent work, Houndjo et al. [10] extended manifestly the f (R, T ) model
to include the higher derivative term R, where  denotes to the d’Alembertian operator. In
fact, this term is appeared in studying some dynamical features of string theory [11] and can
be considered as a potentially important contribution to the original idea of f (R, T ) gravity.
Some gravitational and cosmological results of this model are described in [12,13].
In the literature, there are a lot of works that have been done to investigate the polarization
states of GWs in alternative theories of gravity. Some of them are based on linearizing the
theory consisting of metric perturbations around Minkowski background [14–26] and the
other works are based on NP formalism according to the Petrov classification [26–33].
In this paper, we study the polarization state of GWs in the context of f (R, R, T ) gravity
model. Using the Newman–Penrose formalism, we obtain all possible polarization modes
in some particular models of this scenario. The paper is organized as follows. In Sect. 2,
we introduce the higher-derivative f (R, R, T ) gravitational model. In Sect. 3, we briefly
review the Newman–Penrose formalism for classifying the polarizations of null gravitational
waves. In Sects. 4 and 5, we investigate the polarization states of GWs in two particular
models of higher-derivative f (R, R, T ) gravity. Finally, the results and conclusions are
discussed in Sect. 6.

2 Higher-derivative f (R, R, T ) theory

We start with the following action


1 √
S= d4 x −g[ f (R, R, T ) + L m ]. (1)
2κ 2

where R is the Ricci scalar, T is the trace of the stress-energy tensor and  = g μν ∇μ ∇ν
denotes to the d’Alembertian operator. Varying the action with respect to the metric leads to
the equations of motion for the f (R, R, T ) gravity in the following form [10]

1 
f R Rμν − gμν f + (gμν  − ∇μ ∇ν ) f R + 2 f R (∇(μ ∇ν) R − Rμν ) − Rμν  + gμν 2
2 
− ∇μ ∇ν − ∇μ R∇ν + 2g αβ ∇α Rμν ∇β f R = κ 2 Tμν − f T (Tμν + Θμν ), (2)

123
 Downloaded from https://iranpaper.ir
https://www.tarjomano.com/order

Eur. Phys. J. Plus (2021) 136:4 Page 3 of 17 4

where subscripts R, R and T denote to the derivative of the corresponding quantities with
respect to R, R and T , respectively and
δTαβ
Θμν = g αβ , (3)
δg μν
in which the corresponding stress-energy tensor is defined as
√
2 δ( −gL m )
Tμν = − √ . (4)
−g δg μν
Taking the trace of Eq. (2), we obtain the following equations of motion for the Ricci scalar
as
 
f R − 2 f + 2 f R + 2 f R R − R + 32 − ∇μ R∇ μ + 2g αβ ∇α R f R
= κ 2 T − f T (T + Θ). (5)
In the following sections, we will consider some general forms of the function f (R, R, T )
and investigate the polarization modes for several cases.

3 Brief overview of the Newman–Penrose analysis

Newman and Penrose in 1962 started a study in order to investigate gravitational radiation by
using spin coefficients [34]. They proposed four null-tetrad, namely (lμ , n μ , m μ , m̄ μ ) which
two of them lμ and n μ are real null vectors and the two other m μ and m̄ μ are complex null
vectors. According to their suggestion, at any point, these four null-tetrads will be related to
the Cartesian tetrads (tˆ, x̂, ŷ, ẑ) as follows
1
l = √ (tˆ + ẑ),
2
1
n = √ (tˆ − ẑ),
2
(6)
1
m = √ (x̂ + i ŷ),
2
1
m̄ = √ (x̂ − i ŷ).
2
On the other hand, these null-tetrads will be related to each other as
lμ l μ = m μ m μ = m̄ μ m̄ μ = n μ n μ = 0,
lμ n μ = −m μ m̄ μ = 1, (7)
μ μ μ μ
lμ m = lμ m̄ = n μ m = n μ m̄ = 0.
It is important to notice that in principle, any tensor in Cartesian basis can be converted to
the null-tetrads basis as [34]
Tabc... = Tμνλ... a μ bν cλ ... . (8)
where the sets (a, b, c, . . .) and (μ, ν, λ, . . .) run over the (l, n, m, m̄) and (t, x, y, z), respec-
tively. In their analysis, they showed that the Riemann tensor can be broken down into the
irreducible parts in spinor basis which correspond to ten components of Weyl tensor Ψ ’s,

123
 Downloaded from https://iranpaper.ir
https://www.tarjomano.com/order

4 Page 4 of 17 Eur. Phys. J. Plus (2021) 136:4

nine components of the traceless Ricci tensor Φ’s and the Ricci scalar Λ. According to this
approach, the Riemann tensor in spinor terms decomposes as follows
Rabcd = Ψ ABC D A B  C  D  + Ψ̄ A B  C  D  AB C D
+ Φ ABC  D  A B  C D + ΦC D A B  AB C  D  (9)
− 2Λ AB C D A (C  D  )B  − 2Λ A B  C  D A(C D)B

where AB is the metric-like acting Levi–Civita tensor. In above expression, the five nonzero
components of the Weyl spinor are defined as

Ψ0 = Ψ0000 = Ψ ABC D o A o B oC o D = Cabcd l a m b l c m d ,

Ψ1 = Ψ0001 = Ψ ABC D o A o B oC ι D = Cabcd l a n b l c m d ,
Ψ2 = Ψ0011 = Ψ ABC D o A o B ιC ι D = Cabcd l a m b m̄ c n d , (10)
Ψ3 = Ψ0111 = Ψ ABC D o ι ι ι = Cabcd l n m̄ n ,
A B C D a b c d

Ψ4 = Ψ1111 = Ψ ABC D ι A ι B ιC ι D = Cabcd m̄ a n b m̄ c n d .

where Cabcd is the Weyl tensor and the pair of spinors (o A , ι A ) are the spin-frame basis which
is normalized by the following conditions

AB o o = 0,
A B
AB o ι = 0.
A B
(11)
and the components of the Ricci spinor are
  1
Φ00 = Φ AB A B  o A o B ō A ō B = − Rab l a l b ,
2
 B 1
Φ01 = Φ AB A B  o o ō ῑ = − Rab l a m b ,
A B A
2
 B 1
Φ02 = Φ AB A B  o o ῑ ῑ = − Rab m a m b ,
A B A
2
 B 1
Φ10 = Φ AB A B  o ι ō ō = − Rab l a m̄ b ,
A B A
2
  1
Φ11 = Φ AB A B  o A ι B ō A ῑ B = − Rab (l a n b + m a m̄ b ), (12)
4
  1
Φ12 = Φ AB A B  o A ι B ῑ A ῑ B = − Rab n a m b ,
2
  1
Φ20 = Φ AB A B  ι A ι B ō A ō B = − Rab m̄ a m̄ b ,
2
  1
Φ21 = Φ AB A B  ι A ι B ō A ῑ B = − Rab n a m̄ b ,
2
  1
Φ22 = Φ AB A B  ι A ι B ῑ A ῑ B = − Rab n a n b .
2
Finally, the scalar curvature in the spinor formalism Λ is defined as
1
Λ= R. (13)
24
It is worth noting that in the spinor formalism, the Ricci tensor Rab can be divided into a
trace-free part consisting of the Ricci spinor and the scalar curvature as follows
Rab = R A A B B  = −2Φ AB A B  + 6Λ AB A B  . (14)

123
 Downloaded from https://iranpaper.ir
https://www.tarjomano.com/order

Eur. Phys. J. Plus (2021) 136:4 Page 5 of 17 4

In 1973 Eardley et al. introduced an approach in order to study gravitational waves polar-
ization using NP quantities [3]. They claimed that in metric theories of gravity, near the
gravitational waves and in the low curvature regime, the Riemann tensor components will
reduce to six components which correspond to six independent polarization modes. This
idea was based on the exploration of the electric components of the Riemann tensor Ri0 j0
in null tetrads basis. Using the electric components of the Riemann tensor and NP notation,
the amplitudes of the six polarization modes are given by
1 1
Ψ2 = − Rz0z0 = − Rnlnl , (15)
6 6
1 1
Ψ3 = (−Rx0z0 + i R y0z0 ) = − Rnln m̄ , (16)
2 2
Ψ4 = R y0y0 − Rx0x0 + 2i Rx0y0 = −Rn m̄n m̄ , (17)
Φ22 = −(Rx0x0 − R y0y0 ) = −Rnmn m̄. (18)
From Eqs. (16) and (17), it is clear that Ψ4 and Ψ3 are complex and each one is related to
the two independent polarization modes, namely tensor modes +, × and vector-x, y modes,
respectively. The other two functions Ψ2 and Φ22 are real scalar functions which represent
the longitudinal scalar mode and a breathing scalar mode, respectively. In addition, there
are also some useful relationships which we use them frequently in order to handle the NP
quantities in the rest of this article as
Rnl = Rnlnl ,
R = −2Rnl = −2Rnlnl ,
Rnn = 2Rnmn m̄ , (19)
Rnm = Rnlnm ,
Rn m̄ = Rnln m̄ .
These polarization modes (Ψ2 , Ψ3 , Ψ4 , Φ22 ) in general form are observer dependent but
there are certain invariant statements about them that are true for all standard observers if
they are true for any one. Hence, they were classified gravitational waves to five different
classes from the perspective of polarization modes, the so-called E(2) classification which
is defined as follows [3]
– Class I I6 : Ψ2  = 0. All standard observers measure the same nonzero amplitude in the
Ψ2 mode. But the presence or the absence of all other modes is observer-dependent;
– Class I I I5 : Ψ2 = 0  = Ψ3 . All standard observers measure the absence of Ψ2 and the
presence of Ψ3 . But the presence or absence of Ψ4 and Φ22 is observer-dependent;
– Class N3 : Ψ2 = Ψ3 = 0, Ψ4  = 0, Φ22  = 0. Presence or absence of all modes is
observer-independent;
– Class N2 : Ψ2 = Ψ3 = Φ22 = 0; Ψ4  = 0. Observer-independent;
– Class O1 : Ψ2 = Ψ3 = Ψ4 = 0; Φ22  = 0. Observer-independent.

4 GWs in the case f (R, R, T ) = R − α R−β + γ R + 2λT

In principle, one can consider many general forms for the function f (R, R, T ). In this
regard, we start with assuming a well-known form for this function as f (R, R, T ) =
R − α R −β + γ R + 2λT , where α, β, γ and λ are constants. Note that in this case, the
curvature scalar is minimally coupled to its higher-order derivative term and the trace of the

123
 Downloaded from https://iranpaper.ir
https://www.tarjomano.com/order

4 Page 6 of 17 Eur. Phys. J. Plus (2021) 136:4

stress-energy tensor. As well, the derivatives of the f function with respect to R and R are
obtained as
f R = 1 + αβ R −β−1 (20)
and
f R = γ , (21)
respectively. By applying Eqs. (20) and (21) to the field Eq. (2) in low curvature regime, we
get the field equations in the linear order in Ricci tensor and curvature scalar as follows
1 1 1
Rμν − 2γ Rμν − gμν R + αgμν R −β − γ gμν R + αβgμν R −β−1
2 2 2 (22)
− αβ∇μ ∇ν R −β−1 + 2γ ∇μ ∇ν R = κ 2 Tμν − f T (Tμν + Θμν ).
Obviously, the field equations for this case have a complicated form and it is a difficult task
to handle them. However, considering the fact that we are working in low curvature regime,
can help us to simplify the issue. In order to simplification, we can specify some values for
β and then by applying some approximation, we will be able to solve the field equations.

4.1 Case λ = 0, β = −2

By considering this assumption, the field Eq. (22) in the vacuum take the following form
1 1
Rμν − 2γ Rμν − gμν R + (4α − γ )gμν R + 2(γ − α)∂μ ∂ν R = 0. (23)
2 2
Taking the trace of the above equation leads to
2(3α − γ )R − R = 0. (24)
In order to solve the trace Eq. (24), one can consider two special cases. The simple one is to
assume γ = 3α. In this case, it will be easily calculated that R = 0 and so, by substituting
this result into the field Eq. (23), we obtain
1
Rμν − Rμν = 0. (25)
2γ
Solving this differential equation leads to the following expression for the Ricci tensor
1
Rμν = Aμν exp (iqα x α ), qα q α = − . (26)
2γ
Due to the fact that Rμν  = 0 everywhere and according to the equations that already we have
mentioned in Sect. 3 (Eqs. (15)–(18)), we will get the following results
Rlklk = R = 0 , Rlml m̃  = 0 , Rlklm  = 0 , Rlkl m̃  = 0. (27)
Therefore, from Eqs. (15)–(18), the Newman–Penrose quantities will be obtained as
Ψ2 = 0 , Ψ3  = 0 , Ψ4  = 0 , Φ22  = 0. (28)
Consequently, there are five polarization modes in this case (γ = 3α), which two of them are
vector modes according to the fact that Ψ3  = 0, two are tensor modes + and × corresponding
to Ψ4  = 0 and a breathing scalar mode that corresponds to Φ22  = 0. Notice that Ψ3 and Ψ4
are complex, so each of them represents two polarization modes. It is worth mentioning that

123
 Downloaded from https://iranpaper.ir
https://www.tarjomano.com/order

Eur. Phys. J. Plus (2021) 136:4 Page 7 of 17 4

the classification for this case is (I I I5 ). Another interesting case that should be considered
is γ  = 3α. In this case, the solution of the trace Eq. (24) is given by
1
R = R0 exp (ikα1 x α ), kα1 k1α = − . (29)
2(3α − γ )
On the other hand, Eq. (23) can be rewritten as follows
1 1
Rμν − Rμν = Sμν , (30)
2γ 2γ
where
γ − 2α
Sμν = gμν R + 2(γ − α)∂μ ∂ν R. (31)
4(3α − γ )
In general, the solution of (30) is in the following form [24]
Rμν = Aμν exp i(k1 z − ωt) + Bμν exp i(k2 z − ωt) + c.c.. (32)
Notice that, everywhere in this article, we assume that the gravitational wave travels in the
direction of z-axis. Now, by substituting Eq. (32) in (30), one can calculate Aμν . Therefore,
by a little calculation, we can obtain
 
(3α − γ ) γ − 2α
Aμν = R0 2(γ − α)kμ kν −
1 1
gμν (33)
3α 4(3α − γ )
and
1 1
k1 = ω2 − , k2 = ω2 − . (34)
2(3α − γ ) 2γ
Note that Bμν can also be calculated in the same manner as Aμν . Finally, by substituting (33)
and (34) in (32), the nonzero components of the Ricci tensor can be obtained as follows
 
R0 (3α − γ ) (γ − 2α)
Rtt = 2(γ − α)ω2 + exp i(k1 z − ωt)
3α 4(3α − γ )
+ Btt exp i(k2 z − ωt) + c.c., (35)
R0 (3α − γ )  1 (γ − 2α) 
Rzz = 2(γ − α)[ω2 − ]+ exp i(k1 z − ωt)
3α 2(3α − γ ) 4(3α − γ )
+ Bzz exp i(k2 z − ωt) + c.c., (36)
R0 (3α − γ ) 1
Rt z = 2(γ − α)ω ω2 − exp i(k1 z − ωt)
3α 2(3α − γ )
+ Bt z exp i(k2 z − ωt) + c.c., (37)
and the other nonzero component will be
Ri j = Bi j exp i(k2 z − ωt) + c.c. i, j = x, y, z. (38)
Since all Ricci components are nonzero in this case, all component of the Riemann tensor in
the tetrad basis are non-null too so that
Rlklk = R  = 0 , Rlml m̃  = 0 , Rlklm  = 0 , Rlkl m̃  = 0. (39)
Therefore, the Newman–Penrose quantities satisfy
Ψ2  = 0 , Ψ3  = 0 , Ψ4  = 0 , Φ22  = 0. (40)

123
 Downloaded from https://iranpaper.ir
https://www.tarjomano.com/order

4 Page 8 of 17 Eur. Phys. J. Plus (2021) 136:4

As a result, all six possible polarization modes exist in this case and the corresponded class
is (I I6 ) which is the most general class for the metric theories of gravity.

4.2 Case λ = 0, β < −2

In this subsection, we consider β < −2 and as before, according to the fact that we are
working in the weak field regime, we can carry out some approximations. Since R  1, one
can assume that R −β−1  R and consequently, the field Eq. (22) could be simplified to the
following form
1 3
Rμν − 2γ Rμν − gμν R + γ ∂μ ∂ν R = 0. (41)
2 2
Thus, the trace of the field equations satisfy
1
R + R = 0, (42)
2γ
which its solution is obtained as
1
R = R0 exp (ikα1 x α ) , kα1 k1α = − . (43)
2γ
Now we can rewrite the field equations in the same form as (30) and in a similar approach to
the previous subsection, we obtain the nonzero components of the Ricci tensor as follows
1 1
Rμν − Rμν = Hμν , (44)
2γ 2γ
where
3 1
Hμν = γ ∂μ ∂ν R − gμν R. (45)
2 2
Again the solution of Eq. (44) is given by

Rμν = Aμν exp i(k1 z − ωt) + Bμν exp i(k2 z − ωt) + c.c., (46)

where Aμν satisfies

3γ gμν
Aμν = R0 + kμ1 kν1 (47)
4 3γ
and

1
k1 = k2 = ω2 − . (48)
2γ

As a result, the non-zero components of the Ricci tensor can be obtained as

1
Rtt = R0 (3γ ω2 − 1) exp i(k1 z − ωt) + Btt exp i(k2 z − ωt) + c.c. , (49)
4
1
Rzz = R0 (6γ ω2 − 1) exp i(k1 z − ωt) + Bzz exp i(k2 z − ωt) + c.c. , (50)
8
3γ 1
Rzt = R0 (ω ω2 − ) exp i(k1 z − ωt) + Bzt exp i(k2 z − ωt) + c.c. (51)
4 2γ

123
 Downloaded from https://iranpaper.ir
https://www.tarjomano.com/order

Eur. Phys. J. Plus (2021) 136:4 Page 9 of 17 4

and
Ri j = Bi j exp i(k2 z − ωt) + c.c., i, j = x, y, z. (52)
Therefore, the components of the Riemann tensor in the tetrad basis are
Rlklk = R  = 0 , Rlml m̃  = 0 , Rlklm  = 0 , Rlkl m̃  = 0. (53)
It is straightforward to obtain the Newman–Penros quantities. So we get
Ψ2  = 0 , Ψ3  = 0 , Ψ4  = 0 , Φ22  = 0. (54)
As the same result, there are all six polarization modes and according to the Newman–Penrose
analysis the corresponding class of GWs is (I I6 ).

4.3 Case λ = 0, β > 1

As we have mentioned previously, our framework is in the low curvature regime (R  1),
so we assume R −β−1  R. Therefore, considering the first-order terms in R and Rμν , Eq.
(22) reads
1
Rμν − 2γ Rμν + αgμν R −β + 3αβ∂μ ∂ν R −β−1 = 0. (55)
2
Taking the trace of the above equation and ignoring the terms which containing R in com-
parison with the term that including R −β and R −β−1 , we get
2 −β
R −β−1 + R = 0. (56)
3β
This is a familiar form for the differential equation, so we will follow the approach explained
in [35] exactly. Hence, by defining a new variable as Ω ≡ R −β−1 , Eq. (56) takes the following
form
β
2 1+β
Ω + Ω = 0, (57)
3β
The above equation can be rewritten in a general form as follows
∂U
Ω − = 0, (58)
∂Ω
where U is the potential and by using Eqs. (57) and (58), it will be described as
2(1 + β) 2β+1
U (Ω) = − Ω β+1 . (59)
3β(2β + 1)
Remember the assumption which the GW is traveling in the direction of the z-axis, so
Ω = ddzΩ2 , consequently the solution of (57) reads (see for more details [24])
2

 
(β+1) 2(β+1)
Ω(z) = iξ(z − z 0 ) + Ω0 2 , (60)

where
 1
1 2
ξ= . (61)
3β(2β + 1)(β + 1)

123
 Downloaded from https://iranpaper.ir
https://www.tarjomano.com/order

4 Page 10 of 17 Eur. Phys. J. Plus (2021) 136:4

It should be noted that in Equ. (60), Ω0 is the value of Ω in z 0 . Since the field Ω is Lorentz-
invariant, we can apply Lorentz transformation on the static solution to obtain the time
dependent expression. The result is

 
(β+1) 2(β+1)
(z − z 0 ) − vt
Ω(z, t) = iξ √ + Ω0 2 . (62)
1 − v2

So the Ricci scalar can be found easily as

 −2
(z − z 0 ) − vt − 21
R(z, t) = iξ √ + R0 . (63)
1 − v2

Now to obtain the components of the Ricci tensor, we rearrange (55) in the following form

1 1 1
Rμν − Rμν = αgμν R −β + 3αβ∂μ ∂ν R −β−1 . (64)
2γ 2γ 2

Note that, as one can easily distinguish, there is an important difference between this case
(λ = 0, β < −2) and the previous cases that make Eq. (64) more complicated than the
former cases. Indeed, unlike the previous cases, Eq. (63) implies that the Ricci scalar has a
non-oscillatory solution so, it seems that it is impossible to handle the equation of motion
(64) in order to obtain an exact solution for the Ricci tensor. Nevertheless, one can use again
the static solution of curvature scalar to obtain a time independent solution for the Ricci
tensor and then by applying the Lorentz transformation, find the time dependent solution. To
get more clarity, we will do calculations for the components of the Ricci tensor separately.
According to Eq. (64) and by substituting the static form of the Ricci scalar (63), three general
types of equations for nonzero components of (Rμν ) can be found as follows

1 1 1 −β
Rzz − Rzz = α R + 3αβ∂z ∂z R −β−1 , (65)
2γ 2γ 2
1 1 1 −β
Rμν − Rμν = ( α R ) ∀ μ = ν = z , (66)
2γ 2γ 2
1
Rμν − Rμν = 0 ∀ μ  = ν. (67)
2γ

We start with the third Eq. (67) which has a clear solution as

1 z − vt 1 z − vt
Rμν = c1 exp √ √ + c2 exp − √ √ ∀ μ  = ν, (68)
2γ 1 − v 2 2γ 1 − v 2

where c1 and c2 are constants of the integral. Substituting static form of the Ricci scalar in
Eq. (66) one finds

1 α
Rμν − Rμν = (iξ z + b)2β ∀ μ = ν  = z. (69)
2γ 4γ

123
 Downloaded from https://iranpaper.ir
https://www.tarjomano.com/order

Eur. Phys. J. Plus (2021) 136:4 Page 11 of 17 4

1/2
Where for convenience, in the above equation we have put b = −iξ z 0 + R0 which is a
constant so the solution reads
 √z−vt α exp ( √1 V )(iξ V + b)2β
1 z − vt 1−v 2 2γ
Rμν = exp − √ √ √ dV
2γ 1 − v 2 1 4 2γ
 √z−vt
1 z − vt 1−v 2
+ exp √ √
2γ 1 − v 2 1
α exp (− √12γ W )(iξ W + b)2β 1 z − vt
− √ dW + c3 exp √ √
4 2γ 2γ 1 − v 2
1 z − vt
+ c4 exp − √ √ ∀ μ = ν  = z. (70)
2γ 1 − v 2
Note that the terms which contain integrals are actually the expanded forms of the “Ei”
functions (Exponential Integral function). Now with a similar approach, the first Eq. (65) has
the following solution
  √z−vt
1 z − vt 2 z − vt 1−v 2
Rzz = exp − √ √ exp √
2γ 1 − v 2 γ 1−v 2 1

3α exp (− √12γ V ) (iξ V + b)2β

− √ dV
4 2γ
 √z−vt 3α exp ( √1 W )(iξ W + b)2β 
1−v 2 2γ 1 z − vt
+ √ dW + c5 exp √ √
1 4 2γ 2γ 1 − v 2
1 z − vt
+ c6 exp − √ √ . (71)
2γ 1 − v 2
Then by using Ricci tensor components, the NP quantities satisfy
Ψ2  = 0 , Ψ3  = 0 , Ψ4  = 0 , Φ22  = 0. (72)
As a result, we see all six polarization modes and (I I6 ) class of gravitational wave is appeared
again.

4.4 Case λ  = 0

Since we are interested in working in the case of the absence of matter fields, we can consider
the cosmological constant as a source (see [30] for more details) so, with this assumption
(L m = −Λ) the energy-momentum tensor becomes
Tμν = −Λgμν . (73)
Taking the trace of the above relation, one gets
T = −4Λ . (74)
Now by using Eq. (3), the quantity Θμν is obtained as follows
δTαβ δ
Θμν = g αβ = gαβ μν (−Λgαβ ) = Λgμν . (75)
δg μν δg
Assuming a general form for f (R, R, T ) function as ( f (R, R, T ) = R + α R 2 + βR +
2λT ) and by virtue of the useful derivatives, f R = 1 + 2α R, f R = β and f T = 2λ, the

123
 Downloaded from https://iranpaper.ir
https://www.tarjomano.com/order

4 Page 12 of 17 Eur. Phys. J. Plus (2021) 136:4

equations of motion for the Ricci tensor, take the following form in the first order in R and
Rμν
1 3
Rμν − 2βRμν = gμν R − β∂μ ∂ν R + Λgμν , (76)
2 2
where = (4λ − k 2 ) is a constant. Taking the trace of the field Eq. (76) leads to
1 2
R − R= Λ, (77)
2ξ ξ
where ξ = − β4 . Now to solve the above equation, at first we consider the static solution and
then since the Ricci scalar is Lorentz invariant, by applying the Lorentz transformations, we
obtain the time dependent solution. In this respect, we rewrite Eq. (77) as follows
d2 R 1 2
− R= Λ, (78)
dz 2 2ξ ξ
which its solution is
z z
R(z) = c1 exp √ + c2 exp − √ − 4 Λ, (79)
2ξ 2ξ
where c1 and c2 are constants of the integration. Then by applying the Lorentz transformation,
we have the following result for time dependent solution
1 z − vt 1 z − vt
R(z, t) = c1 exp √ √ + c2 exp − √ √ − 4 Λ. (80)
2ξ 1 − v 2 2ξ 1 − v 2
Note that like the previous subsection, we have the same situation in which the solution of
the Ricci scalar is a non-periodic function. So, as we have mentioned earlier, the approach
that we have used in Sect. (4.1) cannot be applied directly in this case. Therefore in order to
obtain a dynamical solution to the Ricci tensor, one can use exactly the method which has
been employed in Sect. (4.3). By substituting Eq. (79) in (75) we’ll get the following three
general forms of equations as
1 1 1 3
Rzz − Rzz = − R + β∂z ∂z R − Λ , (81)
2β 2β 2 2
1 1 1
Rμν − Rμν = R+ Λ ∀ μ = ν  = z, (82)
2β 2β 2
1
Rμν − Rμν = 0 ∀ μ  = ν. (83)
2β
Equation (83) has a simple form which, after applying the Lorentz transformation, has the
following solution
1 z − vt 1 z − vt
Rμν = c1 exp √ √ + c2 exp − √ √ ∀ μ  = ν. (84)
2β 1 − v 2 2β 1 − v 2
For the second Eq. (82), by putting the expression of the Ricci scalar from (79) and with a
bit of manipulation, we get
1 z − vt 1 z − vt 1 1 z − vt
Rμν = c3 sin √ √ + c4 cos √ √ + c3 exp √ √
2 2ξ 1 − v 2 2 2ξ 1 − v 2 10 2ξ 1 − v 2
1 1 z − vt
+ c4 exp − √ √ − Λ ∀ μ = ν  = z. (85)
10 2ξ 1 − v 2

123
 Downloaded from https://iranpaper.ir
https://www.tarjomano.com/order

Eur. Phys. J. Plus (2021) 136:4 Page 13 of 17 4

Finally, the last nonzero component of the Ricci tensor is obtained from Eq. (81) as follows
1 z − vt 1 z − vt 7 1 z − vt
Rzz = c5 sin √ √ + c6 cos √ √ + c5 exp √ √
2 2ξ 1 − v 2 2 2ξ 1 − v 2 10 2ξ 1 − v 2
7 1 z − vt
+ c6 exp − √ √ − Λ. (86)
10 2ξ 1 − v 2
Consequently, since all components of the Ricci tensor are nonzero, the NP quantities are
Ψ2  = 0 , Ψ3  = 0 , Ψ4  = 0 , Φ22  = 0. (87)
As a result, all six polarization modes will be appeared and the gravitational wave classifi-
cation is (I I6 ).

5 GWs in the case f (R, R, T ) = R + α R2 + β RR

In this section, we investigate the question that how presence of the term which includes RR
in action will effect on the equations of motion and consequently, the number of polarization
modes. In this respect, we assume f (R, R, T ) = R + α R 2 + β RR, so the function
derivatives will be
f R = 1 + 2α R + βR (88)
and
f R = β R. (89)
Now we begin with the trace of the field Eq. (5) in order to obtain the equations of motion for
the Ricci scalar. Considering the first-order terms in R, we have the following relationship
for the Ricci scalar
1
R − R = 0, (90)
6α
which has a solution that reads
1
R = R0 exp(iqα x α ) , qα q α = . (91)
6α
Once again, by using the field Eq. (2) and the fact that R  1, we have the following
dynamical equations for the Ricci tensor
1
Rμν = gμν R − 2αgμν R + 2α∂μ ∂ν R. (92)
2
Substituting the expression that we have already obtained for the Ricci scalar i.e. Eq. (91) in
the above relationship, the nonzero components of Ricci tensor reads
5
Rtt = 2αk 2 − R, (93)
6

1
Rt z = −2αk k 2 − R, (94)
6α
5
Rzz = + 2αk 2 R (95)
6

123
 Downloaded from https://iranpaper.ir
https://www.tarjomano.com/order

4 Page 14 of 17 Eur. Phys. J. Plus (2021) 136:4

and also Rx x = R yy = 0. Therefore for NP quantities, we have

Ψ2  = 0 , Ψ3 = 0 , Ψ4  = 0 , Φ22  = 0. (96)

So there are four polarization modes and classification for this case is (I I6 ). It is important to
note that according to this analysis, in such higher-derivative models, the presence of RR
term in action will decrease the number of polarization modes from six to four modes.

6 Conclusion

This paper is concentrated on investigating the number of GWs polarization modes by using
NP analysis in higher derivative theory of gravity. We considered two particular classes of
higher derivative theory. At first, in Sect. 4 we assumed that f (R, R, T ) = R − α R −β +
γ R + 2λT which in such theories by choosing the different values for the coefficients, we
found the number of polarization modes for several cases. In this specific class of alternative
theory, we revealed that by choosing β = −2 and γ = 3α the NP values will be Ψ2 = 0
which for null gravitational waves means that the massive longitudinal mode does not exist,
therefore five polarization modes will appear. Otherwise (γ  = 3α), in this case not only
Ψ3  = 0, Ψ4  = 0 and Φ22  = 0, but also the NP quantity Ψ2 is nonzero. So all six independent
polarization modes will exist in this case.
In the second step, in Sect. 5 we considered the term that including RR in order to
study how the presence of such terms affects the number of polarization modes. By these
assumptions, we found that Ψ3 = 0 which means unlike the first case, for null GWs the
vector modes x and y do not exist. Therefore, the number of independent polarization modes
will be decreased from six to four polarizations modes.
As a conclusion, by assuming the general form of f function to be f (R, R, T ) =
R − α R −β + γ R δ R + 2λT , the results can be summarized in Table 1. Furthermore, we
determined the E(2) class of Lorentz group for each case which is shown in the table. Notice
that we assume the GR and R 2 theory of gravity as a particular class of a higher-derivative
model. In this perspective, as one can easily distinguish, by choosing the proper values for
the coefficient of f function, a complete set of GWs polarization modes can be produced.
For instance, in the GR case, according to only nonzero NP quantity Ψ4  = 0, we expect
only two cross and plus tensor modes for null GWs, but for higher-derivative case with
(λ = δ = 0, β > 1) one can expects all six independent polarization modes to be revealed.
It is worth noticing to compare our results to some similar works that recently have been
done in the framework of higher-order gravity in view of a clear classification of GWs in these
theories. In a recent work by Capozziello et al. the GWs in higher-order gravity with a different
approach has been studied [24]. They have  considered a generic higher-order gravitational
p √
Lagrangian density L g = (R + a0 R 2 + k=1 ak Rk R) −g where p is the order of
theory and then they perturbed the metric gμν with respect to the flat Minkowski spacetime
ημν in order to obtain the linearized equation in the perturbed metric h μν . By solving the
linearized equations, they have calculated six polarization states for the gravitational waves.
Consequently, they have concluded that there are all six possible polarization states in any
order of the theory and this result is completely general for any order. In fact, the results of
[24] fits very well with our outputs which already have been mentioned in previous sections.
However, by NP analysis approach in the special case f (R, R, T ) = R + α R 2 + β RR
which is corresponds to order p = 1, we have shown that the number of polarization states
reduces to four modes instead of six modes.

123
 Eur. Phys. J. Plus

Table 1 GW Polarization modes in higher-derivative theory of gravity

Coefficients Functional form NP quantities Polarization

(2021) 136:4
Downloaded from https://iranpaper.ir

modes (Classi-
fication)

λ=α=γ =0 R (GR) Ψ2 = Ψ3 = Φ22 = 0, Ψ4  = 0 +, × (N2 )

λ = γ = 0, α  = 0, β = −2 R − α R 2 (R 2 Gravity) Ψ3 = Φ22 = 0, Ψ4  = 0, Ψ2  = 0 +, ×, l (I I6 )
λ = 0, γ = δ = 1, β = −2 R − α R 2 + β RR Ψ3 = 0, Ψ2  = 0, Ψ4  = 0, Φ22  = 0 +, ×, l, b (I I6 )
λ = δ = 0, γ = 3α, β = −2 R − α R 2 + 3αR Ψ2 = 0, Ψ3  = 0, Ψ4  = 0, Φ22  = 0 +, ×, b, x, y (I I I5 )
λ = δ = 0, γ  = 3α, β = −2 R − α R 2 + γ R Ψ2  = 0, Ψ3  = 0, Ψ4  = 0, Φ22  = 0 +, ×, l, b, x, y (I I6 )
λ = δ = 0, β < −2 R − α R −β + γ R Ψ2  = 0, Ψ3  = 0, Ψ4  = 0, Φ22  = 0 +, ×, l, b, x, y (I I6 )
λ = δ = 0, β > 1 R − α R −β + γ R Ψ2  = 0, Ψ3  = 0, Ψ4  = 0, Φ22  = 0 +, ×, l, b, x, y (I I6 )
λ  = 0, δ = 0, β = −2 R − α R 2 + γ R + 2λT Ψ2  = 0, Ψ3  = 0, Ψ4  = 0, Φ22  = 0 +, ×, l, b, x, y (I I6 )
https://www.tarjomano.com/order

Page 15 of 17
4

123
 Downloaded from https://iranpaper.ir
https://www.tarjomano.com/order

4 Page 16 of 17 Eur. Phys. J. Plus (2021) 136:4

In another article by the same authors, GWs in higher-order teleparallel gravity has been
investigated [25]. They studied the teleparallel equivalent of higher-order Lagrangians as
L R = −R + a0 R 2 + a1 RR by means of the boundary term B = 2∇μ T μ . According to
their results, there are two transverse and longitudinal polarization modes in addition to the
two standard + and × modes. These outcomes have a suitable compatibility with our results
by the NP analysis approach in the absence of teleparallel terms.
Another work that worth comparison is a recent study that have considered the GWs in
higher-order local and non-local gravity [26]. The authors investigated
 gravitational waves

in the higher-order local gravity with an action including f R, R, 2 R, ..., n R =
n 
R + i=1 αi Ri R and in non-local gravity with an action contains f R, −1 R, −2 R, ...,
 n
−n R = R + i=1 αi R−i R. They have shown that the number of polarization modes
in both local and non-local models are similar however, the physical meaning behind them
are different. They have obtained two breathing and longitudinal scalar modes in addition
to standard GR tensor modes which are matching to our results that have been obtained in
Sect. 5 as well.
Hence, the higher-derivative theories of gravity could be a powerful and interesting subject
to further investigation in the context of the gravitational wave attributes.

References

1. B. Abbott et al., LIGO scientific collaboration and virgo collaboration. Phys. Rev. Lett. 116, 061102
(2016)
2. B.P. Abbott et al., Phys. Rev. Lett. 119(14), 141101 (2017)
3. D.M. Eardley, D.L. Lee, A.P. Lightman, Phys. Rev. D 8, 3308 (1973)
4. P. Amaro-Seoane et al., GW Notes 6, 4–110 (2013)
5. S. Nojiri, S.D. Odintsov, V.K. Oikonomou, Phys. Rep. 692, 1 (2017)
6. T. Clifton, P.G. Ferreira, A. Padilla, C. Skordis, Phys. Rep. 513, 1–189 (2012)
7. S. Nojiri, S.D. Odintsov, Phys. Rep. 505, 59 (2011)
8. A. De Felice, S. Tsujikawa, Living Rev. Relativ. 13, 3 (2010)
9. T. Harko, F.S.N. Lobo, S. Nojiri, S.D. Odintsov, Phys. Rev. D 84, 024020 (2011)
10. M.J.S. Houndjo, M.E. Rodrigues, N.S. Mazhari, D. Momeni, Myrzakulov, Int. J. Mod. Phys. D 26,
1750024 (2017)
11. A. Hindawi, B.A. Ovrut, D. Waldram, Phys. Rev. D 53, 5597 (1996). arXiv:hep-th/9509147
12. Z. Yousaf, M. Sharif, M. Ilyas, M.Z. Bhatti, Int. J. Geom. Meth. Mod. Phys. 15, 1850146 (2018)
13. M. Ilyas, Z. Yousaf, M.Z. Bhatti, Mod. Phys. Lett. A 34, 1950082 (2019)
14. S. Capozziello, C. Corda, Int. J. Mod. Phys. D 15, 1119–1150 (2006)
15. C. Corda, JCAP 0704, 009 (2007)
16. S. Capozziello, C. Corda, M.F. De Laurentis, Phys. Lett. B 669, 255–259 (2008)
17. C. Corda, Eur. Phys. J. C 65, 257–267 (2010)
18. L. Yang, C.-C. Lee, C.-Q. Geng, JCAP 1108, 029 (2011)
19. C.P.L. Berry, J.R. Gair, Phys. Rev. D 83, 104022 (2011)
20. K. Bamba, S. Capozziello, M. De Laurentis, S. Nojiri, D. Sez-Gmez, Phys. Lett. B 727, 194–198 (2013)
21. S. Capozziello, A. Stabile, Astrophys. Space Sci. 358(2), 27 (2015)
22. S. Hou, Y. Gong, Y. Liu, Eur. Phys. J. C 78, 378 (2018)
23. Rong-jia Yang, Phys. Lett. B 784, 212–216 (2018)
24. S. Capozziello, M. Capriolo, L. Caso, Int. J. Geom. Methods Mod. Phys. 16, 1950047 (2019)
25. S. Capozziello, M. Capriolo, L. Caso, Class. Quantum Gravity 37, 235013 (2020)
26. S. Capozziello, M. Capriolo, S. Nojiri, Phys. Lett. B 810, 135821 (2020)
27. M.E.S. Alves, O.D. Miranda, J.C.N. de Araujo, Phys. Lett. B 679, 401–406 (2009)
28. M.E.S. Alves, O.D. Miranda, J.C.N. de Araujo, Class. Quantum Gravity 27, 145010 (2010)
29. M. Alves, P. Moraes, J. de Araujo, M. Malheiro, Phys. Rev. D 94(2), 024032 (2016). [17]
30. H.Rizwana Kausar, L. Philippoz, P. Jetzer, Phys. Rev. D 93(12), 124071 (2016)

123
 Downloaded from https://iranpaper.ir
https://www.tarjomano.com/order

Eur. Phys. J. Plus (2021) 136:4 Page 17 of 17 4

31. M. Sharif, A. Siddiqa, Astrophys. Space Sci. 362, 226 (2017)

32. H.R. Kausar, Int. J. Mod. Phys. D 26(5), 1741010 (2017)
33. O. Bertolami, C. Gomes, F.S.N. Lobo, Eur. Phys. J. C 78, 303 (2018)
34. E. Newman, R. Penrose, J. Math Phys. 3, 566 (1962)
35. R. Rajaraman, Solitons and Instantons: An Introduction to Solitons and Instantons in Quantum Field
Theory (Elsevier Science Publishers, Netherlands, 1982)

123