Gravitational waves in Kasner spacetimes and Rindler wedges in Regge-Wheeler
gauge: Unruh effect
Yuuki Sugiyama∗
Department of Physics, Graduate School of Science, Hiroshima University, Higashi-Hiroshima 739-8526, Japan and
Department of Physics, Kyushu University, 744 Motooka, Nishi-Ku, Fukuoka 819-0395, Japan
Kazuhiro Yamamoto†
Department of Physics, Kyushu University, 744 Motooka, Nishi-Ku, Fukuoka 819-0395, Japan
Tsutomu Kobayashi‡
Department of Physics, Rikkyo University, Toshima, Tokyo 171-8501, Japan
Abstract
We derive the solutions of gravitational waves in the future (F) expanding and the past (P) shrink-ing Kanser spacetimes as well as in the left (L) and right (R) Rindler wedges in the Regge-Wheeler gauge. The solutions for all metric components are obtained in an analytic form in each region. We identify the master variables, which are equivalent to massless scalar fields, to describe the gravita-tional degrees of freedom for the odd parity and even parity modes under the transformation in the two-dimensional plane symmetric space. Then, the master variables are quantized, and we develop the quantum field theory of the gravitational waves in the F, P, L, and R regions. We demonstrate that the mode functions of the quantized gravitational waves in the left and right Rindler wedges are obtained by an analytic continuation of the left-moving and right-moving wave modes in Kasner spacetime. On the basis of these analyses, we discuss the Unruh effect of quantized gravitational waves for an observer in a uniformly accelerated motion in Minkowski spacetime.
I. INTRODUCTION
Gravitational waves (GWs), predicted in general relativity, have been directly detected by the advanced LIGO in 2015 for the first time. The first event GW150914 is thought to come from a merger of a binary black holes [1]. Gravitational waves are useful not only for testing general relativity itself but also for exploring the black hole physics. The detection of gravitational waves from a coalescence of neutron star binary has enhanced their importance as astrophysical tools. Electromagnetic counterparts of gravitational wave events, in particular like neutron star merger events, have been explored at various wavelengths, i.e., optical waves, infrared waves, X-rays, andγ-rays (see e.g., [2]). The role of gravitational waves in cosmology is also important for exploring the primordial universe. The primordial gravitational waves can be generated in an inflationary era from vacuum fluctuations, which might be detected in polarization anisotropies in the cosmic microwave background as a smoking gun of inflation, although they have not been detected [3].
The standard model of inflation yields a homogeneous and isotropic universe satisfying the cosmological principle. Observations show that the universe on very large scales looks being consistent with the cosmological principle. However, some authors claim that the cosmic microwave background radiation shows anomalous features on large scales such as hemispherical power asymmetry and dipole modulation [4, 5].
Models of inflation that may explain the large scale anomalies have been proposed. For example, in Refs. [6, 7], the authors proposed anisotropic inflation during which a U(1) gauge field plays an important role in producing anisotropic expansion. They also investigated the cosmological perturbations in the anisotropic inflation models. Various anisotropic inflation models have been proposed so far, and cosmological perturbations are found to be useful for characterizing the features of each anisotropic inflation model in terms of, e.g., the matter power spectrum and non-gaussian features (see Refs. [8, 9]).
Many authors have investigated the behavior of gravitational waves in a universe breaking the spatial homogeneity or the spatial isotropy [10–13]. Cho and Speliotopoulos investigated the propagation of gravitational waves in Bianchi type-I (B-I) universe [14]. The authors of [15, 16] investigated the primordial gravitational waves from a pre-inflationaly
era in a B-I universe. Gravitational waves in a universe breaking the cosmological principle are not trivial. For example, scalar, vector, and tensor modes do not decouple. The evolution of metric perturbations is not easy to be solved in an analytic manner, and therefore numerical approaches or an approximation of small anisotropy are often taken.
In order to explore gravitational waves in an anisotropic universe or in a non-inertial space, we focus on gravitational waves in the future expanding and the past shrinking Kanser spacetimes as well as those in the left and right Rindler wedges. Kasner spacetime is a special case of the B-I universe, which is one of the simplest model of an anisotropic universe [17]. The Rindler metric is a patch of Minkowski spacetime described by the coordinates of uniformly accelerating observers. Minkowski spactime is covered by the future (F) expanding and the past (P) shrinking Kanser spacetimes and the left (L) and right (R) Rinder wedges (See Fig. 1). One of the aims of the present paper is to investigate the quantum aspects of gravitational waves in Kasner spacetimes and Rindler wedges including the Unruh effect.
The Hawking radiation is the well-known prediction of quantum field theory in a black hole spacetime [18]. The Unruh effect predicts that a uniformly accelerating observer in Minkowski spacetime sees the vacuum state for an inertial observer as a thermally excited state with the temperaturea/2π, whereais the acceleration [19]. The Unruh effect is an analogy of the Hawking radiation through the equivalence principle, which is described on the basis of the quantum field theory in Rindler wedges. The Unruh effect is explained by the fact that the Minkowski vacuum state is expressed as an entangled state between the left and right Rindler wedges when it is constructed on the Rindler vacuum. This is well known for the case of a scalar field and a Dirac field [20–22]. In the present paper, we investigate the vacuum structure of gravitational waves in Minkowski spactime, by finding the explicit expression of the solution of the tensor modes in the F, P, R, L regions. This investigation is a generalization of the works on a scalar field and a Dirac field to the gravitational waves. In the present paper, the Regge-Wheeler gauge is extensively used to a plane symmetric spacetime, which is often used in a spherically symmetric spacetime [23]. Our result is useful to describe the Unruh effect of gravitational waves.
The present paper is organised as follows: In Sec. II, we first derive the action for the master variables of the gravitational waves in the F, P, R, L regions using the Regge-Wheeler gauge for the odd modes and the even modes, respectively [23]. The resultant action for the master variables is found to be equivalent to that of a massless scalar field, which will be quantized in Sec. III. Then, we discuss the Bogoliubov transformation for two different sets of the solutions, and we show that the solutions in Kasner spacetimes (F and P regions) are analytically continued to the left and right Rindler wedges (L and R regions). In Sec. IV, we demonstrate the analytic continuations of the metric components of the F and P regions to those of the L and R regions. Then, we demonstrate that the Minkowski vacuum state is described as an entangled state between the two modes described in the whole region of the Minkowski spacetime, which leads to a description of the Unruh effect of the gravitational waves. In Sec. V, we demonstrate the calculation of the expectation values of the energy density of the vacuum fluctuations of the gravitational waves associated with the Rindler vacuum state. Section V is devoted to summary and conclusions. In Appendix A, we demonstrate the derivation of the action for the master variables of the gravitational waves in the F, P, L, R regions. In Appendix B, we briefly review the positive mode frequency function of the Minkowski vacuum state in the F region. In Appendix C, we summarize the results of Sec. IV, i.e., the analytic continuation property of the metric perturbation of the gravitational waves in the F, P R, L regions in the Regge-Wheeler gauge. In Appendix D, we present some details of the calculation in Sec. V. Throughout the present paper, we use the natural units c=~=kB = 1.
II. CLASSICAL SOLUTIONS OF GWS IN RINDLER AND KASNER SPACETIME A. GWs in future expanding Kasner spacetime (F region)
We start with driving the solution of GWs in the future expanding Kasner spacetime (the F region) by using the Regge-Wheeler gauge. The line element of the F region is given by
ds2=e2aη(−dη2+dζ2) +dy2+dz2≡¯gFµνdxµdxν, (1) where −∞< η <∞,−∞< ζ <∞, andais a constant. The Kasner coordinates of the F region are related to the global coordinates of Minkowski spacetime as
t=1 ae
aηcoshaζ, x= 1 ae
aηsinhaζ, (2)
as shown in Fig. 1. The metric perturbationshF
µν are defined by
τ
x
R
F
L
t
P
τ
∼ξ
∼ξ
η
ζ
η
ζ
∼ ∼FIG. 1. Four regions of Minkowski spacetime and corresponding coordinates.
Following [23], we write
hF0a =∂ah0F+ab∂bχF, (4)
hF1a =∂ahF1 +ab∂bψF, (5)
hFab=hFδab+∂a∂bh3F+c(a∂b)∂cγF, (6)
where a, b, and c stand for 2 or 3, and ab is the completely antisymmetric tensor: 22 =33 = 0, 23= 1 = −32.
Therefore, the metric perturbations are written using the 10 functions, (hF00, hF01, hF11, hF0, hF1, χF, ψF, hF, hF3, γF).
With respect to the parity transformation on they-zplane, these metric perturbations can be classified into the odd modes described by (χF, ψF, γF) and the even modes described by (hF
00, hF01, hF11, h0F, hF1, hF, hF3).
Let us consider the gauge transformation generated by the infinitesimal transformation generator Kµ =Kµodd+K even µ (7) with Kµodd≡ 0 0 ab∂bλ , Kµeven≡ δη δζ ∂aδx , (8) whereKodd
µ andKµevenare the generator of the gauge transformation for the odd and even modes, respectively. Using this gauge freedom, we take the Regge-Wheeler gauge in whichhF
0 =hF1 =hF3 =γF = 0. The metric perturbations
are then described by the 6 functions (hF
00, hF01, hF11, hF, χF, ψF): hFµν = hF 00 hF01 ∂zχF −∂yχF hF 01 hF11 ∂zψF −∂yψF ∂zχF ∂zψF hF 0 −∂yχF −∂yψF 0 hF . (9)
in terms of the two decoupled master variables asS(2)=SgF(o)+SgF(e), SgF(o)= 1 32πG Z dηdζdydz (∂ηφF)2−(∂ζφF)2−e2aη(∂⊥φF)2, (10) SgF(e)= 1 32πG Z dηdζdydz(∂ηhF)2−(∂ζhF)2−e2aη(∂⊥hF)2, (11) whereSF g(o)andS F
g(e)are the actions for the odd and even modes, respectively, and we defined (∂⊥φ
F)2= (∂
yφF)2+ (∂zφF)2 (and similarly for (∂⊥hF)2). HereφF itself is not a metric perturbation variable, but it is related toψF and
χF in a nontrivial way. One may notice thatφF and hF are subject to the same action as that of a massless scalar field living in the Kasner spacetime (1). Once the solutions for φF and hF are obtained, all the other variables are determined in terms of these two master variables through the constraint equations, as demonstrated below. This is as it should be, because we have only two dynamical degrees of freedom in general relativity. These actions are obtained by substituting the metric perturbations to the Einstein-Hilbert action, expanding it to quadratic order, and performing integration of parts. The detailed derivation is presented in Appendix A.
Let us then proceed to solving the equations of motion. We obtain the following equation of motion for the odd mode,
−∂η2+∂ζ2+e2aη∂⊥2
φF = 0, (12)
where∂2
⊥=∂y2+∂z2. The odd-mode metric perturbations,ψF andχF, are given in terms ofφF through the constraint equations,
∂2⊥ψF =∂ηφF, ∂⊥2χF =∂ζφF. (13) (See Appendix A 1 for the derivation.) WhenφF is written in a mode-expanded form
φF =ϕF,o(η)ei(kxζ+k⊥·x⊥), (14)
the coefficientϕF,o(η) obeys
[∂η2+k2(η)]ϕF,o(η) = 0, (15)
where we definedk2(η) =k2
x+e2aηκ2 withκ2=|k⊥|2. Then, using Eq. (13), we have
ψF =−∂ηϕ F,o κ2 e i(kxζ+k⊥·x⊥), (16) χF =−ikxϕ F,o κ2 e i(kxζ+k⊥·x⊥). (17)
For the even mode,hF, we have the same equation of motion as Eq. (12). The other even-mode metric perturbations are given in terms ofhF through
∂⊥2hF00=e2aη∂⊥2hF+ 2a∂ηhF −2∂η2h
F, (18)
∂⊥2hF01=−2∂ζ(∂ηhF−ahF), (19) ∂⊥2hF11=−e2aη∂⊥2hF+ 2a∂ηhF−2∂2ζhF. (20) (See Appendix A 2 for the derivation.) Then, we writehF in a mode-expanded form
hF =ϕF,e(η)ei(kxζ+k⊥·x⊥), (21)
andϕF,e(η) obeys the same equation as Eq. (15),
[∂2η+k2(η)]ϕF,e(η) = 0. (22)
From Eqs. (18), (19), and (20), we have hF00= e2aηϕF,e−2a κ2∂ηϕ F,e+ 2 κ2∂ 2 ηϕF,e ei(kxζ+k⊥·x⊥), (23) hF01= 2ikx κ2 (∂ηϕ F,e−aϕF,e)ei(kxζ+k⊥·x⊥), (24) hF11= −k 2(η) κ2 ϕ F,e−k 2 x κ2ϕ F,e−2a κ2∂ηϕ F,e ei(kxζ+k⊥·x⊥). (25)
Finally, summarizing the above results, the metric perturbations in the F region in the Regge-Wheeler gauge are given by hFµν = (ehF,µνo+ehF,µνe)ei(kxζ+k⊥·x⊥) (26) with ehF,µνo= 0 0 kxkz κ2 ϕ F,o −kxky κ2 ϕ F,o ∗ 0 −ikz κ2∂ηϕ F,o iky κ2∂ηϕ F,o ∗ ∗ 0 0 ∗ ∗ ∗ 0 , (27) e hF,µνe = e2aηϕF,e−2a κ2∂ηϕ F,e+ 2 κ2∂ 2 ηϕ F,e 2ikx κ2 (∂ηϕ F,e −aϕF,e) 0 0 ∗ −k 2(η) κ2 ϕ F,e−kx2 κ2ϕ F,e−2a κ2∂ηϕ F,e 0 0 ∗ ∗ ϕF,e 0 ∗ ∗ ∗ ϕF,e . (28)
We can thus determine all the metric components simply by solving a massless Klein-Gordon equation in the Kasner spacetime.
B. GWs in past shriking Kasner spacetime (P region)
Let us then derive the solution of GWs in the past shriking Kasner spacetime (the P region) in the Regge-Wheeler gauge. The analysis presented in this and subsequent subsections follows closely, and hence has some overlaps with, the one in the previous subsection. The line element of the P-region is written as
ds2=e−2aηe(−d e
η2+deζ2) +dy2+dz2≡g¯µνP dxµdxν, (29) where the coordinates in the P region is related to those in the global Minkowski coordinates as
t=−1
ae
−aηecoshaeζ, x=1 ae
−aeηsinhaeζ, (30)
as shown in Fig. 1. The coordinates in the P region is obtained by the analytic continuation of the coordinates in the F region asζ=−ζeandη=−ηe−πi/a(TABLE I). We can write the metric perturbations in the P region in the Regge-Wheeler gauge asgP µν= ¯gµνP +hPµν with hPµν = hP 00 hP01 ∂zχP −∂yχP hP01 hP11 ∂zψP −∂yψP ∂zχP ∂zψP hP 0 −∂yχP −∂yψP 0 hP . (31) F→R τ =ζ− π 2ai, ξ=η+ π 2ai F→L τ˜=−ζ− π 2ai, ˜ ξ=η+ π 2ai P→R τ =−ζ˜− π 2ai, ξ=−η˜− π 2ai P→L τ˜= ˜ζ− π 2ai, ˜ ξ=−η˜− π 2ai
TABLE I. Analytic continuations and the relations among the coordinates in the F, P, R, and L regions. A more detailed discussion is given in Sec. IV.
The master variable for the odd modes, φP, can be found in a similar way to those in the F region, giving the action and equation of motion
SgP(o) = 1 32πG Z d˜ηdζdydz˜ h(∂η˜φP)2−(∂ζ˜φP)2−e−2aη˜(∂⊥φP)2 i , (32) h −∂2η˜+∂ 2 ˜ ζ +e −2aη˜ ∂⊥2iφP = 0. (33)
The master variable of the even mode,hP, has the same action and hence obeys the same equation of motion as Eqs. (32) and (33). The metric perturbations are given by solving
∂2⊥ψP =∂η˜φP, ∂⊥2χP =∂ζ˜φP, (34) and ∂2⊥hP00=e−2aη˜∂⊥2hP−2a∂η˜hP−2∂2η˜h P, (35) ∂2⊥hP01=−2∂ζ˜(∂˜ηhP +ahP), (36) ∂⊥2hP11=−e−2aη˜∂⊥2hP−2a∂ηhP−2∂ζ2˜h P. (37)
WhenφP andhP are written in the forms
φP =ϕP,o(η)ee i(kxeζ+k⊥·x⊥), hP =ϕP,e(
e
η)ei(kxeζ+k⊥·x⊥), (38)
we have the metric perturbations in the P region in the Regge-Wheeler gauge,
hPµν = (ehP,µνo+ehP,µνe)ei(kxζe+k⊥·x⊥), (39) with e hP,µνo= 0 0 kxkz κ2 ϕ P,o −kxky κ2 ϕ P,o ∗ 0 −ikz κ2∂ηϕ˜ P,o iky κ2∂ηϕ˜ P,o ∗ ∗ 0 0 ∗ ∗ ∗ 0 , (40) ehP,µνe= e−2aη˜ϕP,e+2a κ2∂η˜ϕ P,e+ 2 κ2∂ 2 ˜ ηϕ P,e 2ikx κ2 (∂˜ηϕ P,e+aϕP,e) 0 0 ∗ − ˜ k2( e η) κ2 ϕ P,e −k 2 x κ2ϕ P,e+2a κ2∂eηϕ P,e 0 0 ∗ ∗ ϕP,e 0 ∗ ∗ ∗ ϕP,e , (41) whereek2(η) = (ke x2+κ2e−2aηe).
C. GWs in right Rindler wedge (R region)
Next, we derive the solution of GWs in the right Rindler wedge (the R region). The line element in the R region is given by
ds2=e2aξ(−dτ2+dξ2) +dy2+dz2≡g¯µνRdxµdxν, (42) whereais a constant which is interpreted as a uniform acceleration. The coordinates in the R region is related to the coordinates of Minkowski spacetime as
t= 1 ae
aξsinhaτ , x= 1 ae
as shown in Fig. 1. The metric perturbations in the R region can be written as gRµν= ¯gµνR +hRµν (44) with hRµν = hR00 hR01 ∂zχR −∂yχR hR 01 hR11 ∂zψR −∂yψR ∂zχR ∂zψR hR 0 −∂yχR −∂yψR 0 hR . (45)
The action for the odd parity master variable in the R region is given by SgR(o)= 1 32πG Z dτ dξdydz (∂τφR)2−(∂ξφR)2−e2aξ(∂φR)2 , (46)
which leads to the equation of motion
−∂τ2+∂ξ2+e2aξ∂⊥2φR= 0. (47) The metric perturbationsψR andχR are related toφR as
∂⊥2ψR=∂τφR, ∂2⊥χR=∂ξφR. (48) WritingφR in the form,
φR=ϕR,o(ξ)e−iωτ+ik⊥·x⊥, (49)
one sees thatϕR,o obeys
[∂ξ2+K2(ξ)]ϕR,o= 0 (50)
with K2(ξ) = ω2−κ2e2aξ. Thus, also in the right Rindler wedge the problem reduces to analyzing a system of a massless scalar field.
For the even mode, we obtain the same action as that of the odd mode, SgR(e)= 1 32πG Z dτ dξdydz (∂τhR)2−(∂ξhR)2−e2aξ(∂hR)2 , (51)
and hence the equation of motion is also the same as that ofφR. UsinghR one can write the other metric functions as ∂⊥2hR00=e 2aξ ∂⊥2hR+ 2a∂ξhR−2∂τ2h R , (52) ∂⊥2hR01=−2∂τ(∂ξhR−ahR), (53) ∂⊥2hR11=−e 2aξ ∂⊥2hR+ 2a∂ξhR−2∂ξ2h R . (54)
WritinghR in the form,
hR=ϕR,e(ξ)e−iωτ+ik⊥·x⊥, (55)
the coefficientϕR,e obeys
[∂2ξ+K2(ξ)]ϕR,e= 0, (56)
whereK was already defined above. The solutions of Eqs. (52), (53), (54) are obtained as hR00= e2aξϕR,e−2a κ2∂ξϕ R,e−2ω 2 κ2ϕ R,e e−iωτ+ik⊥·x⊥, (57) hR01=−2iω κ2 (∂ξϕ R,e−aϕR,e)e−iωτ+ik⊥·x⊥, (58) hR11= −e2aξϕR,e−2a κ2∂ξϕ R,e+ 2 κ2∂ 2 ξϕ R,e e−iωτ+ik⊥·x⊥. (59)
We finally have the metric perturbations in the R region in the Regge-Wheeler gauge hµνR = (ehR,µνo+ehµνR,e)e−iωτ+ik⊥·x⊥ (60) with e hR,µνo= 0 0 −ikz κ2∂ξϕ R,o iky κ2∂ξϕ R,o ∗ 0 −ωkz κ2 ϕ R,o ωky κ2 ϕ R,o ∗ ∗ 0 0 ∗ ∗ ∗ 0 , (61) ehR,µνe= −K 2(ξ) κ2 ϕ R,e− 1 κ2 ω 2ϕR,e+ 2a∂ξϕR,e −2iω κ2 (∂ξϕ R,e−aϕR,e) 0 0 ∗ −e2aξϕR,e− 2 κ2 a∂ξϕ R,e−∂2 ξϕ R,e 0 0 ∗ ∗ ϕR,e 0 ∗ ∗ ∗ ϕR,e . (62)
D. GWs in left Rindler wedge (L region)
Similarly, we can obtain the action for the L region. The line element in the L region is given by ds2=e2aξe(−d
e
τ2+deξ2) +dy2+dz2≡g¯µνL dxµdxν, (63) which is related to the Minkowski coordinates through the relation
t= 1 ae aeξsinha e τ , x=−1 ae aeξcosha e τ , (64)
as shown in Fig. 1. In a similar way to the case for the R region, the metric perturbation is defined bygL
µν = ¯gµνL +hLµν with hLµν = hL 00 hL01 ∂zχL −∂yχL hL01 hL11 ∂zψL −∂yψL ∂zχL ∂zψL hL 0 −∂yχL −∂yψL 0 hL . (65)
We obtain the action for the master variable for the odd modes as SgL(o)= 1 32πG Z d˜τ dξdydz˜ h(∂τ˜φL)2−(∂ξ˜φL)2−e2a ˜ ξ(∂φL)2i, (66)
which leads to equation of motion
h
−∂τ2˜+∂ξ2˜+e2aξ˜∂2⊥iφL= 0. (67) The action and equation of motion for the even mode master variable hL are the same as Eqs. (66) and (67), respectively. The metric components in the L region are related tohL by
∂⊥2ψL=∂ηφ˜ L, ∂2⊥χL=∂ξ˜φ L, (68) and ∂2⊥hL00=e2aξ˜∂⊥2hL+ 2a∂ξ˜hL−2∂˜τ2h L, (69) ∂2⊥hL01=−2∂τ˜(∂ξ˜hL−ahL), (70) ∂2⊥hL11=−e2aξ˜∂⊥2hL+ 2a∂ξ˜hL−2∂ξ2˜hL. (71)
Similarly, whenφL andhL are written in the forms
φL=ϕL,o( ˜ξ)e−iωτ˜−ik⊥·x⊥, hL=ϕL,e( ˜ξ)e−iωτ˜−ik⊥·x⊥, (72)
respectively, we have the metric perturbations in the L region in the Regge-Wheeler gauge
hLµν= (ehL,µνo+ehµνL,e)e−iωτ˜−ik⊥·x⊥ (73) with e hL,µνo= 0 0 −ikz κ2∂ξ˜ϕ L,o iky κ2∂ξ˜ϕ L,o ∗ 0 −ωkz κ2 ϕ L,o ωky κ2 ϕ L,o ∗ ∗ 0 0 ∗ ∗ ∗ 0 , (74) e hL,µνe= −K˜ 2( ˜ξ) κ2 ϕ L,e− 1 κ2 ω2ϕL,e+ 2a∂ξ˜ϕ L,e −2iω κ2 (∂ξ˜ϕ L,e−aϕL,e) 0 0 ∗ −e2aξ˜ϕL,e− 2 κ2 a∂ξ˜ϕ L,e−∂2 ˜ ξϕ L,e 0 0 ∗ ∗ ϕL,e 0 ∗ ∗ ∗ ϕL,e , (75) where ˜K2( ˜ξ) = (ω2−κ2e2aξ˜).
III. QUANTIZATION OF THE GRAVITATIONAL WAVES
Having thus obtained the classical solution for all the components of the metric perturbations, let us move to consider the quantization of the gravitational waves. As we have shown in the previous section, the master variables are essentially regarded as two decoupled massless scalar fields. Therefore, we closely follow the quantization procedure of massless scalar fields living on the Kasner/Rindler metric [21], and the consequences of the quantization of the master variables are very similar to those in the case of a massless scalar field.
A. Quantization in Kasner spactimes (F and P regions)
We perform quantization of the gravitaional waves described by the actions in the previous section. Here we consider the odd and even modes in the F region (10) and (11). For the later convenience, we introduce the canonically normalized variablesϕF(o)=φF/√16πGandϕF(e)=hF/√16πG, so that the corresponding actions reduce to
SgF(λ)=1 2
Z
dηdζdydzh(∂ηϕF(λ))2−(∂ζϕF(λ))2−e2aη(∂⊥ϕF(λ))
2i, (76)
whereλ= o,e. The canonical momenta are defined as
ΠF(λ)(η, ζ, y, z)≡ δS F g(λ) δ∂ηϕF(λ) =∂ηϕ F (λ)(η, ζ, y, z), (77)
with which the commutation relations are imposed as
[ ˆϕF(λ)(η, ζ, y, z),ΠˆF(λ)(η, ζ0, y0, z0)] =iδ(ζ−ζ0)δ(x⊥−x0⊥), others = 0, (78)
wherex⊥= (y, z).
The field operator satisfies the Heisenberg equation of motions
−∂η2+∂ζ2+e2aη∂⊥2 ˆ
whose solution can be written as ˆ ϕF(λ)(η, ζ, y, z) = Z ∞ −∞ dkx Z ∞ −∞ dk⊥ h vkF,λ xk⊥(η, ζ, y, z)ˆa F,λ kxk⊥+ h.c. i , (80) wherek⊥= (ky, kz), ˆa F,λ kxk⊥(ˆa F,λ†
kxk⊥) is the annihilation (creation) operators satisfying andv
F,λ
kx,k⊥(η, ζ, y, z) is a function
written in the form ofvkF,λ
x,k⊥ =f(η)e
ikxζeik⊥·x⊥ with mode functionsf satisfying the equation of motion
∂η2f(η) +kx2f(η) +e2aη|k⊥|2f(η) = 0, (81)
and the normalization condition (∂ηf)f∗−f(∂ηf∗) =i/(2π)3.
Let us write a solution of the equation of motion (81) using the Bessel function of the first kind as [20], vF,λk xk⊥(η, ζ, y, z) = −i 2πp4asinh(π|kx|/a) J−i|kx|/a κeaη a eikxζeik⊥·x⊥ (82)
with κ=|k⊥|. This function behaves as vkF,λxk⊥(η, ζ, y, z)∝e
−i|kx|ηeikxζeik⊥·x⊥ in the η → −∞ limit. This means
that this function is a positive frequency mode function near the horizon. Furthermore, we note that the mode functions vF,λk
xk⊥(η, ζ, y, z) with positive kxrepresent right-moving wave modes in theζ direction, whereas the modes
with negative kx represent left-moving wave modes. One can see this from the behavior of the mode functions near the future horizon [21].
One can choose a different solution of Eq. (81) written in terms of the Hankel function, with which the solution is written as uF,λk xk⊥(η, ζ, y, z) = −i 4π√2ae π|kx|/2aH(2) i|kx|/a κeaη a eikxζeik⊥·x⊥. (83)
With this choice, the field operator can be expressed as ˆ ϕF(λ)(η, ζ, y, z) = Z ∞ −∞ dkx Z ∞ −∞ dk⊥ h uF,λk xk⊥(η, ζ, y, z)ˆb F,λ kxk⊥+ h.c. i , (84) where ˆbF,λk xk⊥(ˆb F,λ†
kxk⊥) is the annihilation (creation) operators satisfying [ˆb
F,λ kxk⊥, ˆbF,λ† k0 xk0⊥ ] = δ(kx−kx0)δ(k⊥−k0⊥). The mode functionsuF,λk xk⊥(η, ζ, y, z) behaves asu F,λ kxk⊥(η, ζ, y, z)∝e −iκeaη/a
eikxζeik⊥·x⊥in theη→ ∞limit, and therefore
this is the natural choice as the positive frequency mode at late times. As we will show in this section as well as in the appendix B,uF,λk
xk⊥(η, ζ, y, z) is the mode function associated with the Minkowski vacuum state.
Similarly, we can write the same quantized field in the P region (see Fig. 1) in two different ways, ˆ ϕP(λ)(˜η,ζ, y, z) =˜ Z ∞ −∞ dkx Z ∞ −∞ dk⊥ h uP,λk xk⊥(˜η, ˜ ζ, y, z)ˆbP,λk xk⊥+h.c i , (85) ˆ ϕP(λ)(˜η,ζ, y, z) =˜ Z ∞ −∞ dkx Z ∞ −∞ dk⊥ h vkP,λ xk⊥(˜η, ˜ ζ, y, z)ˆaP,λk xk⊥+ h.c. i , (86) where ˆaP,λk xk⊥(ˆa P,λ† kxk⊥) and ˆb P,λ kxk⊥(ˆb P,λ†
kxk⊥) are the annihilation (creation) operators, and the mode functions are
uP,λk xk⊥(˜η, ˜ ζ, y, z) = ie ikxζ˜ 4π√2ae −π|kx|/2aH(1) i|kx|/a κe−aη˜ a eik⊥·x⊥, (87) vP,λk xk⊥(˜η, ˜ ζ, y, z) = ie ikxζ˜ 2πp4asinh(π|kx|/a) Ji|kx|/a κe−aη˜ a eik⊥·x⊥. (88) The modesvkP,λ
xk⊥(η, ζ, y, z) with positivekxrepresent right-moving wave modes and those with negativekx represent
left-moving wave modes as is discussed in Ref. [21]. As mentioned above, the mode function uFk(P),λ
xk⊥ (η, ζ, y, z) that is written in terms of the Hankel function in the
F(P) region corresponds to the positive frequency mode function for the Minkowski vacuum (see Appendix B). This leads us to define the Minkowski vacuum with the annihilation operator ˆbλ
kxk⊥ by
ˆ bFk(K),λ
for any (kx,k⊥) andλ. On the other hand,vkFx(kP⊥),λ(η, ζ, y, z) is the Kasner mode function, with which we can define
the Kasner vacuum state by
ˆ aFk(P),λ
xk⊥ |0iK = 0 (90)
for any (kx,k⊥) andλ.
The Bogoliubov transformation between the two sets of the mode functions u(kF,P)λ
xk⊥ and v (F,P)λ
kxk⊥ can be found
straightforwardly as follows. Using some mathematical formula for the Hankel and Bessel functions, we arrive at
vkF,λ x,k⊥(η, ζ, y, z) = −i 4πp4asinh(π|kx|/a) eπ|kx|/aeikxζH(2) i|kx|/a κeaη a + e−ikxζH(2) i|kx|/a κeaη a ∗ eik⊥·x⊥, (91) vkP,λ x,k⊥(˜η, ˜ ζ, y, z) = i 4πp4asinh(π|kx|/a) ( eikxζ˜H(1) i|kx|/a κe−aη˜ a +e−π|kx|/a e−ikxζ˜H(1) i|kx|/a κe−aη˜ a ∗) eik⊥·x⊥. (92) From these equations we can read off the Bogoliubov coefficients as
vkF(P),λ x,k⊥ (η, ζ, y, z) =αkxu F(P),λ kx,k⊥ (η, ζ, y, z) +β−kx h uF−(kP),λ x,−k⊥(η, ζ, y, z) i∗ , (93) with αkx = 1 √ 1−e−2π|kx|/a , β−kx =− e−π|kx|/a √ 1−e−2π|kx|/a . (94)
It then follows that the annihilation and creation operators associated with the Minkowski and Kasner modes are related as ˆ bFk(P),λ x,k⊥ = 1−e−2π|kx|/a −1/2 ˆ aFk(P),λ x,k⊥ −e −π|kx|/aaˆF(P),λ† −kx,−k⊥ . (95)
Note that kx is momentum in the direction of ζ ( ˜ζ) in the F-region (P-region), which takes both positive and negative values. The Minkowski vacuum state can be described using the states associated with the Kasner vacuum as |0iM ∝exp 1 2 Z ∞ −∞ dkx Z ∞ −∞ dk⊥e−π|kx|/aakFx(,Pk)⊥,λ†a F(P),λ† −kx,−k⊥ |0iK, (96)
where the Kanser vacuum is defined by Eq. (90) for any (kx,k⊥) andλ. The above expression can be rewritten as
|0iM = ∞ Y kx=0 ∞ Y k⊥=−∞ p 1−e−2πkx/a ∞ X n=0 e−πkxn/a|n, λ, kx,k ⊥iK⊗ |n, λ,−kx,−k⊥iK, (97) where |n, λ, kx,k⊥iK = (n!)−1/2 a F(P),λ† kx,k⊥ n
|0iK is then-th excited state from the Kasner vacuum state |0iK char-acterized by kx,k⊥ and λ. This expression shows that the Minkowski vacuum is expressed as an entangled state
between the positive and negative momentum modes.
The expectation value of the number operator constructed from ˆaFk(P),λ
x,k⊥ and ˆa
F(P),λ†
kx,k⊥ is given by a thermal
distri-bution with the temperatureT =a/2π,
Nkx ≡Mh0|aˆ F(P),λ† kx,k⊥ ˆa F(P),λ kx,k⊥ |0iM = 1 e2π|kx|/a−1δ (3)(0), (98)
which is responsible for the entanglement of the Minkowski vacuum state (97), where the divergent factor δ(3)(0) accounts for an infinite spatial volume.
B. Quantization in Rindler spacetime (R and L regions)
Quantization of the gravitational waves in the right and left Rindler wedges described by the action (46) and (51) can be done in a similar way. We expand the canonically normalized master variables as
ˆ ϕR(λ)(τ, ξ, y, z) = Z ∞ 0 dω Z ∞ −∞ dk⊥[vω,R,λk⊥(τ, ξ, y, z)ˆa R,λ ω,k⊥+ h.c.], (99) ˆ ϕL(λ)(˜τ ,ξ, y, z) =˜ Z ∞ 0 dω Z ∞ −∞ dk⊥[vω,L,λk⊥(˜τ , ˜ ξ, y, z)ˆaL,λω,k ⊥+ h.c.], (100)
where we introduced the creation and annihilation operators satisfying and the mode functions are given by
vω,R,λk ⊥(τ, ξ, y, z) = r sinhπω/a 4π4a e −iωτK iω/a κeaξ a eik⊥·x⊥, (101) vω,L,λk ⊥(˜τ , ˜ ξ, y, z) = r sinhπω/a 4π4a e −iωτ˜K iω/a κeaξ˜ a ! e−ik⊥·x⊥. (102)
The right (left) Rindler vacuum state is defined by ˆ
aRω,(kL⊥),λ|0iR(L)= 0 (103)
for anyλ, ω, k⊥.
In the next section, we demonstrate that the analytic continuations of the modes from the F and P regions to the R and L regions reproduce the solutions in the latter regions from those in the former (and vice versa). The analytic continuations yield identities between the mode functions and the metric perturbations in the different regions. The Bogoliubov transformation and the description of the Minkowski vacuum in the F and P regions are generalized to the entire spacetime including the R and L regions, leading to the description of the Unruh effect of metric perturbations.
IV. ANALYTIC CONTINUATION FOR METRIC PERTURBATION COMPONENTS A. Analytic continuation of mode functions from Kasner to Rindler
We first consider the analytic continuation of the master variables [20, 21]. The right and left Rindler wedges are described by the line elements (42) and (63), respectively, and their coordinates are related to those of the Minkowski spacetime by Eqs. (43) and (64) (see Fig. 1). The coordinates of the four regions, i.e., the F, P, R, and L regions, are related to each other by the analytic continuations, as summarized in TABLE I. By inspecting the explicit form of the mode functions (91), (92), (101), and (102), we find that they are analytically continued as
vF,λ−ω,k ⊥(η, ζ, y, z)= ( vω,R,λk ⊥(τ, ξ, y, z) (R region) 0 (L region) (forω=−kx>0), (104) vF,λω,−k ⊥(η, ζ, y, z)= ( 0 (R region) vL,λω,k⊥(˜τ ,ξ, y, z)˜ (L region) (forω=kx>0), (105) vP,λω,k ⊥(˜η, ˜ ζ, y, z) = ( vR,λω,k ⊥(τ, ξ, y, z) (R region) 0 (R region) (forω=kx>0), (106) v−P,λω,−k ⊥(˜η, ˜ ζ, y, z) = ( 0 (R region) vω,L,λk ⊥(˜τ , ˜ ξ, y, z) (L region) (forω=−kx>0), (107) where we used the formulasK−ν(z) =Kν(z), Kν(z) =−(πi/2)e−νπi/2H
(2)
ν (e−πi/2z), andKν(z) = (πi/2)eνπi/2H
(1)
ν (eπi/2z). See Ref. [21] for a more detailed derivation. Thus, it can be seen that the left-moving (right-moving) modes in the F (P) region are equivalent to the Rindler modes in the R region, whereas the right-moving (left-moving) modes in the F (P) region are equivalent to the Rindler modes in the L region.
We can then identify the creation and annihilation operators in the different regions as ˆ aI,λω,k ⊥ ≡ˆa F,λ −ω,k⊥ = ˆa P,λ ω,k⊥= ˆa R,λ ω,k⊥, (108) ˆ aII,λω,k ⊥ ≡ˆa F,λ ω,−k⊥ = ˆa P,λ −ω,−k⊥= ˆa L,λ ω,k⊥, (109)
which satisfyˆaI(II)ω,k,λ
⊥ ,ˆa I(II),λ0† ω0,k0 ⊥ =δλ,λ0δ(ω−ω0)δ(k⊥−k⊥0 ) andaˆI(II)ω,k,λ ⊥ ,ˆa I(II),λ0 ω0,k0 ⊥ =aˆI(II)ω,k,λ† ⊥ ,aˆ (I(II),λ0† ω0,k0 ⊥ = 0.
B. Analytic continuation of metric perturbations from F region to R and L regions
Next, let us derive the metric tensors in the Rindler wedges by means of the analytic continuation. A brief summary of the results of this section is presented in Appendix C.
The analytic continuation of the metric tensor consists of the analytic continuation of the mode functions, demon-strated in the previous subsection, and the transformation of the metric tensor under the coordinate transformation. This latter step is unique to the case of gravitational waves. For example, under the transformation of the coordinates from the F region to the R region, the perturbations of the metric tensor transform as
hFµν(x0) = ∂x ρ ∂x0µ ∂xσ ∂x0νh R ρσ(x), (110) wherex={τ, ξ, y, z} andx0 ={η, ζ, y, z}.
The odd parity metric perturbations in the F region can be written as ˜ hF,µνo(η, kx,k⊥) = 1 κ2O F µν(η, kx,k⊥)ϕF,o, (111) where OµνF (η, kx,k⊥)≡ 0 0 kxkz −kxky ∗ 0 −ikz∂η iky∂η ∗ ∗ 0 0 ∗ ∗ ∗ 0 , (112)
andϕF,o is defined by Eq. (C2). Similarly, the even parity metric perturbations in the F region can be written as ˜ hF,µνe(η, kx,k⊥) = 1 κ2E F µν(η, kx,k⊥)ϕF,e, (113) where EµνF (η, kx,k⊥)≡ κ2e2aη−2a∂η+ 2∂2 η −2ikxa+ 2ikx∂η 0 0 ∗ −(k2(η) +k2 x)−2a∂η 0 0 ∗ ∗ κ2 0 ∗ ∗ ∗ κ2 , (114)
andϕF,e is defined by Eq. (C2).
The metric perturbations in the R region may be expressed in a similar way as ˜ hR,µνo(ω, ξ,k⊥) = 1 κ2O R µν(ω, ξ,k⊥)ϕR,o (115) with ORµν(ω, ξ,k⊥)≡ 0 0 −ikz∂ξ iky∂ξ ∗ 0 −ωkz ωky ∗ ∗ 0 0 ∗ ∗ ∗ 0 (116)
for the odd modes, and ˜ hR,µνe(ω, ξ,k⊥) = 1 κ2E R µν(ω, ξ,k⊥)ϕR,e, (117) with EµνR(ω, ξ,k⊥)≡
−(K2(ξ) +ω2)−2a∂ξ 2iωa−2iω∂ξ 0 0
∗ −κ2e2aξ−2a∂ξ+ 2∂2 ξ 0 0 ∗ ∗ κ2 0 ∗ ∗ ∗ κ2 , (118)
for the even modes, whereϕR,oandϕR,e are defined by Eq. (C6).
Now, let us show thatOµνF andEµνF are related, respectively, toORµν andEµνR in the way inferred from a coordinate transformation. The coordinate transformation we consider here is given by the analytic continuation from the F region to the R region (TABLE I),
τ=ζ−iπ
2a, ξ=η+ iπ 2a. One can directly check that
OµνF (η,−ω,k⊥) = ∂xρ ∂x0µ ∂xσ ∂x0νO R ρσ(ω, ξ,k⊥), EµνF (η,−ω,k⊥) = ∂xρ ∂x0µ ∂xσ ∂x0νE R ρσ(ω, ξ,k⊥). (119)
Using this and Eq. (104), we see, for positiveω, that
ehF,µνo(η,−ω,k⊥)e−iωζe+ik⊥·x⊥= 1 κ2O F µν(η,−ω,k⊥)vF,−ωok⊥ = 1 κ2 ∂xρ ∂x0µ ∂xσ ∂x0νO R ρσ(ω, ξ,k⊥)vωR,ko⊥ = ∂xρ ∂x0µ ∂xσ ∂x0νehR,ρσo(ω, ξ,k⊥)e−iωτe+ik⊥·x⊥, (120) ehF,µνe(η,−ω,k⊥)e−iωζe+ik⊥·x⊥ = 1 κ2E F µν(η,−ω,k⊥)vF,−ωek⊥ = 1 κ2 ∂xρ ∂x0µ ∂xσ ∂x0νE R ρσ(ω, ξ,k⊥)vR,ωke⊥= ∂xρ ∂x0µ ∂xσ ∂x0νehR,ρσe(ω, ξ,k⊥)e−iωτe+ik⊥·x⊥. (121)
This result shows that, by the analytic continuation and the transformation law of the metric under the coordinate transformation, all the metric components in the F region can be reproduced from the R region, and vice versa.
In a similar way, we obtain the metric tensor in the L region by using analytic continuation from the F region. To this end we write the odd parity metric perturbations in the L region as
˜ hL,µνo(ω,ξ,˜k⊥) = 1 κ2O L µν(ω,ξ,˜k⊥)ϕL,o, (122) where we defined OLµν(ω,ξ,˜ k⊥)≡ 0 0 −ikz∂ξ˜ iky∂ξ˜ ∗ 0 −ωkz ωky ∗ ∗ 0 0 ∗ ∗ ∗ 0 (123)
andϕL,ois defined by Eq. (C8). Similarly, we write the even parity metric perturbations in the L region as ˜ hL,µνe(ω,ξ,˜k⊥) = 1 κ2E L µν(ω,ξ,˜k⊥)ϕL,e, (124) where we defined EµνL (ω,ξ,˜ k⊥)≡ −( ˜K2( ˜ξ) +ω2)−2a∂ ˜ ξ 2iωa−2iω∂ξ˜ 0 0 ∗ −κ2e2aξ˜−2a∂ξ˜+ 2∂ 2 ˜ ξ 0 0 ∗ ∗ κ2 0 ∗ ∗ ∗ κ2 , (125)
andϕL,e is defined by (C8). The analytic continuation from the F region to the L region reads (see TABLE I) ˜ τ=−ζ− iπ 2a, ˜ ξ=η+ iπ 2a, using which we can show that the following relations hold:
OµνF (η, ω,−k⊥) = ∂xρ ∂x0µ ∂xσ ∂x0νO L ρσ(ω,ξ,˜ k⊥), EµνF (η, ω,−k⊥) = ∂xρ ∂x0µ ∂xσ ∂x0νE L ρσ(ω,ξ,˜k⊥). (126)
Therefore, by combining the above result and the relation between the mode functions (105) we obtain, for positive ω, e hF,µνo(η, ω,−k⊥)eiωζe−ik⊥·x⊥ = 1 κ2O F µν(η, ω,−k⊥)vωF,−ok⊥ = 1 κ2 ∂xρ ∂x0µ ∂xσ ∂x0νO L ρσ(ω,ξ,˜k⊥)vωL,ko⊥ = ∂xρ ∂x0µ ∂xσ ∂x0νehL,ρσo(ω,ξ,˜ k⊥)e−iωτ˜e−ik⊥·x⊥, (127) e hF,µνe(η, ω,−k⊥)eiωζe−ik⊥·x⊥ = 1 κ2E F µν(η, ω,−k⊥)vωF,−ek⊥ = 1 κ2 ∂xρ ∂x0µ ∂xσ ∂x0νE L ρσ(ω,ξ,˜ k⊥)vωL,ke⊥ = ∂xρ ∂x0µ ∂xσ ∂x0νehL,ρσe(ω,ξ,˜ k⊥)e−iωτ˜e−ik⊥·x⊥, (128)
showing that all the metric components in the L region can be reproduced from the F region by the analytic contin-uation and the transformation law of the metric under the coordinate transformation.
C. Analytic continuation of metric perturbations from P region to R region and L region
Next let us consider the analytic continuation from the P region to the R region and the L region. In the P region, we introduce the following decomposition,
˜ hP,µνo(˜η, kx,k⊥) = 1 κ2O P µν(η, kx,k⊥)ϕP,o, (129) with OPµν(˜η, kx,k⊥)≡ 0 0 kxkz −kxky ∗ 0 −ikz∂η˜ iky∂η˜ ∗ ∗ 0 0 ∗ ∗ ∗ 0 (130)
andϕP,o defined by Eq. (C4) for the odd modes, and
˜ hP,µνe(˜η, kx,k⊥) = 1 κ2E P µν(˜η, kx,k⊥)ϕF,e, (131) with EµνP (˜η, kx,k⊥)≡ κ2e−2aη˜+ 2a∂ ˜ η+ 2∂η2˜ 2ikxa+ 2ikx∂η˜ 0 0 ∗ −(˜k2(˜η) +k2 x) + 2a∂η˜ 0 0 ∗ ∗ κ2 0 ∗ ∗ ∗ κ2 (132)
and ϕP,e defined by Eq. (C4) for the even modes. In a similar way to the case of the F region, using the analytic continuation from the P region to the R region,
τ=−ζ˜−iπ
2a, ξ=−η˜− iπ 2a, we obtain the following relations,
OPµν(˜η, ω,k⊥) = ∂xρ ∂x0µ ∂xσ ∂x0νO R ρσ(ω, ξ,k⊥), E P µν(˜η, ω,k⊥) = ∂xρ ∂x0µ ∂xσ ∂x0νE R ρσ(ω, ξ,k⊥), (133)
From this and Eq. (106), we obtain, for positiveω, ehP,µνo(˜η, ω,k⊥)eiω ˜ ζeik⊥·x⊥= 1 κ2O P µν(˜η, ω,k⊥)vωP,ko⊥ = 1 κ2 ∂xρ ∂x0µ ∂xσ ∂x0νO R ρσ(ω,ξ,˜k⊥)vωR,ko⊥ = ∂xρ ∂x0µ ∂xσ ∂x0νehR,ρσo(ω,ξ,˜k⊥)e−iωτe+ik⊥·x⊥, (134) ehP,µνe(˜η, ω,k⊥)eiω ˜ ζeik⊥·x⊥ = 1 κ2E P µν(˜η, ω,k⊥)vωP,ke⊥ = 1 κ2 ∂xρ ∂x0µ ∂xσ ∂x0νE R ρσ(ω, ξ,k⊥)vR,ωke⊥= ∂xρ ∂x0µ ∂xσ ∂x0νehR,ρσe(ω,ξ,˜k⊥)e−iωτe+ik⊥·x⊥. (135)
In a similar way, we see that the following relations hold with the help of the analytic continuation from the P region to the L region,
OPµν(˜η,−ω,−k⊥) = ∂xρ ∂x0µ ∂xσ ∂x0νO L ρσ(ω,ξ,˜ k⊥), EµνP (˜η,−ω,−k⊥) = ∂xρ ∂x0µ ∂xσ ∂x0νE L ρσ(ω,ξ,˜k⊥), (136)
where the analytic continuation here is given by ˜ τ= ˜ζ−iπ 2a, ˜ ξ=−η˜− iπ 2a. Then, from the above result and Eq. (107) we obtain, for positiveω,
e hP,µνo(˜η,−ω,−k⊥)e−iω ˜ ζe−ik⊥·x⊥ = 1 κ2O P µν(˜η,−ω,−k⊥)vP,−ωo−k⊥ = 1 κ2 ∂xρ ∂x0µ ∂xσ ∂x0νO L ρσ(ω,ξ,˜k⊥)vωL,ko⊥ = ∂xρ ∂x0µ ∂xσ ∂x0νehL,ρσo(ω,ξ,˜ k⊥)e−iωτ˜e−ik⊥·x⊥, (137) e hP,µνe(˜η,−ω,−k⊥)e−iω ˜ ζe−ik⊥·x⊥ = 1 κ2E P µν(˜η,−ω,−k⊥)vP,−ωe−k⊥ = 1 κ2 ∂xρ ∂x0µ ∂xσ ∂x0νE L ρσ(ω,ξ,˜ k⊥)vωL,ke⊥ = ∂xρ ∂x0µ ∂xσ ∂x0νehL,ρσe(ω,ξ,˜ k⊥)e−iωτ˜e−ik⊥·x⊥. (138)
Thus, we have shown that the mode functions of the metric perturbations in the R region and the L region are given by the analytic continuations from the F (P) region. The results are summarized as follows:
e hF,λµν (η,−ω,k⊥)e−iωζe+ik⊥·x⊥ = ∂xρ ∂x0µ ∂xσ ∂x0νehR,λρσ (ω, ξ,k⊥)e−iωτe+ik⊥·x⊥ =ehP,λµν (eη, ω,k⊥)e+iωeζe+ik⊥·x⊥, (139) e hF,λµν (η, ω,−k⊥)e+iωζe−ik⊥·x⊥ = ∂xρ ∂x0µ ∂xσ ∂x0νehL,λρσ (ω,ξ,ek⊥)e−iωτee−ik⊥·x⊥=ehP,λ µν (eη,−ω,−k⊥)e −iωeζe−ik⊥·x⊥. (140)
These results clearly demonstrate the relation between the modes of gravitational waves in each region: the left-moving (right-left-moving) wave modes in the F(P) region are equivalent to the Rindler modes in the R region, whereas the right-moving (left-moving) wave modes in the F(P) region are equivalent to the Rindler modes in the L region.
D. Description of the Minkowski vacuum state and the Unruh effect
As a result of the analytic continuations from the F (P) region to the R and L regions [Eqs. (139) and (140)], and the equivalence of the operators [Eqs. (108) and (109)], the Minkowski vacuum is described as an entanglement state between the states defined in the left and right Rindler wedges and the F and P regions in a unified way,
|0iM = ∞ Y ω=0 ∞ Y k⊥=−∞ p 1−e−2πω/a ∞ X n=0 e−πωn/a|n, λ, ω,k⊥iI⊗ |n, λ, ω,k⊥iII, (141) where we defined |n, λ, ω,k⊥iI(II) = (n!)−1/2 a I(II),λ† ω,k⊥ n
|0iI(II), and the Rindler vacuum state is defined in each
region by
aIω,(IIk),λ
for any (ω,k⊥). The expectation value of the number operator in the Rindler wedges reads NI(II)≡Mh0|ˆa I(II),λ† ω,k⊥ aˆ I(II),λ ω,k⊥ |0iM = 1 e2πω/a−1δ (3)(0), (143)
which shows the thermal distribution with the temperature determined by the acceleration, T =a/2π. This can be understood as the Unruh effect of the gravitational waves.
V. CONCLUSION
In the present paper, we have derived the solutions of gravitational waves in the future expanding Kasner spactime (the F region) and the past shrinking Kasner spacetime (the P region) as well as the left (L) and the right (R) Rindler wedges in an analytic form. In the derivation, we have performed the metric tensor decomposition in spacetime endowed with two-dimensional plane symmetry and used (an analog of) the Regge-Wheeler gauge. In this formulation, the odd parity and even parity modes defined with respect to the parity transformation in the two-dimensional plane are decoupled. We have introduced the two master variables associated with the gravitational wave degrees of freedom and derived the quadratic actions for them, each of which is equivalent to the action of a massless scalar field in the corresponding background. The master variables are quantized in each region. The mode functions corresponding to the Minkowski vacuum state in the F (P) region are identified and the Bogoliubov transformation between the mode functions of the Minkowski vacuum state and those of the Kasner vacuum state are presented. From the relation it can be seen that the Minkowski vacuum state for the quantized gravitational waves is described as an entangled state constructed on the basis of left-moving and right-moving wave modes associated with the Kasner vacuum state. We have also demonstrated that the metric components of the quantized gravitational waves associated with the Kasner vacuum state in the F and P regions are analytically continued to those in the left and right Rindler wedges. This analytic continuation from the F (P) region to the R(L) region allows us to see that the Minkowski vacuum state is described as an entangle state between the left and right Rindler states. Our result gives a description of the Unruh effect for gravitational waves in the Rindler wedges.
The description will be useful for investigating the Unruh effect on the vacuum fluctuations of gravitational waves in a frame of uniform acceleration or equivalently in a frame of uniform gravitation by the equivalence principle. A thorough investigation is left for a further study. However, here we evaluate the energy density of gravitational waves to discuss the difference between the Minkowski vacuum and the Rindler vacuum. We calculate the following effective energy-momentum tensor,
TµνGW(x) =− 1
8πGh
(2)G
µνi, (144)
whereh(2)Gµνiis the second-order part of the Einstein tensor andh· · · istands for the temporal and spatial average.
Using the expression (99) in the Rindler wedge, we obtain
Mh0|Tˆτ τGW|0iM = e−2aξ 4π2 Z ∞ −∞ dω ω 3 e2πω/a−1, (145) Rh0|Tˆτ τGW|0iR= e−2aξ 4π2 Z ∞ 0 dωω3, (146)
The details of the calculation are presented in Appendix D. Equation (145) shows that the energy density associated with the Minkowski vacuum state obeys the Planck distribution with the temperature T = a/2π. The regularized energy momentum tensor is obtained by subtracting the expectation values associated with the Minkowski vacuum. Then,h0M|Tˆτ τGW|0Mireduces to zero after regularization, while (R)h0|Tˆτ τGW|0i(R)reads
Rh0|Tˆτ τGW|0i regularized R =Rh0|Tˆτ τGW|0iR−Mh0|Tˆτ τGW|0iM =−e −2aξ 2π2 Z ∞ 0 dω ω 3 e2πω/a−1 =− e−2aξ 2π2 a4 240. (147)
In the Schwarzschild spacetime with a massM, whose metric is given by ds2=−(1−2GM /r)dt2+ dr
2
(1−2GM /r)+r
the expectation value of the energy momentum tensor for a massless scalar field with respect to the Boulware vacuum state|Biis given in ther→2GM limit by [24]
hB|Tˆµν|Biren∼ − 1 2π2(1−2GM/r)2 Z ∞ 0 dω ω 3 exp (2πω/κ)−1diag −1,1 3, 1 3, 1 3 , (149)
whereκ= 1/4GM. This yields an expression for the energy density similar to Eq. (147),
hB|Tˆ00|Biren∼ − 1 2π2(1−2GM/r) Z ∞ 0 dω ω 3 exp (2πω/κ)−1 =− 1 2π2(1−2GM/r) κ4 240. (150)
One can see a pathological behavior that the renormalized expectation value of the energy density diverges at the Schwarzschild horizon when the Boulware vacuum state is adopted outside the horizon of the Schwarzschild spacetime. This is quite similar to the result of Eq. (147) that the energy density diverges at the Rindler horizon when the Rindler vacuum state is adopted in the Rindler wedge. Thus, the structure of the Rindler vacuum state and the Minkowski vacuum state on the Minkowski spacetime is analogous to that of the Boulware vacuum state and the Hartle-Hawking state on the Schwarzschild spacetime [24]. The expectation values of the other components of the energy momentum tensor will be obtained in a similar way,
Rh0|TˆξξGW|0i regularized R =Rh0|TˆξξGW|0iR−Mh0|TˆξξGW|0iM =−e −2aξ 6π2 Z ∞ 0 dωω 3+ 4ωa2 e2πω/a−1 =− e−2aξ 6π2 41a4 240 (151) Rh0|TˆyyGW|0i regularized R =Rh0|TˆzzGW|0i regularized R =Rh0|TˆyyGW|0iR−Mh0|TˆyyGW|0iM =−e −4aξ 6π2 Z ∞ 0 dω ω 3+ωa2 e2πω/a−1 =− e−4aξ 6π2 11a4 240 (152)
The result do not satisfy the trace-free property, hTµ
µi = 0, which would be expected by the equation of motion averaged within a certain region of spacetime. This violation of the trace-free property of the energy momentum tensor could be related to the fact that the gravitational waves violate the conformal invariance and that the gravitational waves become ambiguous for the modes with the wavelength longer than the characteristic scale of the spacetime. The characteristic scale of the Rindler space is 1/a, and the modes of the gravitational waves with the wavelength longer than the scale 1/a contributes to the propertyhTµ
µi 6= 0. If we take the range of the integration with respect toω in Eqs. (151) and (152) to beω much larger thana, then the trace-free property is obtained.
ACKNOWLEDGMENTS
We thank T. Inagaki, M. Taniguchi, Y. Matsuo, H. Sakamoto, H. Shimoji, D. Sakuma, K. Ueda, T. Tanaka, A. Higuchi, Y. Nambu, J. Soda, S. Kanno, A. Matsumura, and H. Suzuki for useful discussions related to the topic in the present paper. This work was partially supported by MEXT/JSPS KAKENHI Grant Numbers 17K05444 (KY), 17H06359 (KY), JP20H04745 (TK), and JP20K03936 (TK).
Appendix A: Derivation of the master variables and their quadratic Lagrangians
In this appendix, we derive the quadratic Lagrangians for the master variables for odd and even parity perturbations. Although we focus on the F region, the derivation can be extended straightforwardly to the P, R, and L regions.
1. Odd parity perturbations
To simplify the notation we omit the labelF from the variables. Around the future expanding Kasner metric we expand the Einstein-Hilbert Lagrangian,√−gR/2, to second order in metric perturbations. Performing integration by parts, we obtain 8πGLodd =1 4e −2aηh(∂ ⊥ψ)˙ 2−2∂⊥χ0∂⊥ψ˙+ (∂⊥χ0)2 i −1 4(∂ 2 ⊥ψ)2+ 1 4(∂ 2 ⊥χ)2, (A1)
where a dot and a prime denote differentiation with respect to η and ξ, respectively. By introducing the auxiliary fieldφ, the above Lagrangian can be written equivalently as
8πGLodd= 1 4 h −e2aη(∂⊥φ)2+ 2∂⊥φ(∂⊥ψ˙−∂⊥χ0) i −1 4(∂ 2 ⊥ψ) 2+1 4(∂ 2 ⊥χ) 2. (A2)
Indeed, from the Euler-Lagrange equation forφone has
φ=e−2aη( ˙ψ−χ0), (A3) using which one can removeφfrom Eq. (A2) to reproduce the original Lagrangian (A1).
Now, the Euler-Lagrange equations forψandχderived from Eq. (A2) read, respectively,
∂⊥2ψ= ˙φ, ∂⊥2χ=φ0. (A4) Substituting these constraint equations back to Eq. (A2), we arrive at the quadratic Lagrangian for the dynamical variableφ, 8πGLodd =1 4 h ( ˙φ)2−(φ0)2−e2aη(∂⊥φ)2 i . (A5)
Thus, the odd parity sector is found to be governed by the single master variableφ. The metric perturbations in the Regge-Wheeler gauge,ψ andχ, are determined in terms ofφthrough the constraint equations (A4).
2. Even parity perturbations
We also expand the Einstein-Hilbert Lagrangian to second order in ever parity perturbations to get 8πGLeven=−1 4 ˙ h2+1 2e −2aηh 11¨h+ae−2aηh01h0−e−2aηh01h˙0+ 1 4e −2aη(∂ ⊥h01)2 +1 4∂⊥h11∂⊥h+ 1 4(h 0)2−a 2e −2aηh 11h˙ +h00 1 4e −2aη∂2 ⊥h11+ 1 4∂ 2 ⊥h+ 1 2e −2aηh00−a 2e −2aηh˙ , (A6)
where we performed integration by parts and omitted total derivatives. From the Euler-Lagrange equations for h00,
h01, andh11, we obtain the following constraint equations:
∂⊥2h11=−e2aη∂⊥2h+ 2ah˙ −2h00, (A7)
∂⊥2h01=−2( ˙h0−ah0), (A8)
∂⊥2h00=e2aη∂2⊥h+ 2ah˙ −2¨h. (A9)
These equations can be used to eliminateh11,h01, andh11from the Lagrangian. After integration by parts, we end
up with the quadratic Lagrangian for the single master variableh, 8πGLeven=1 4 h ˙ h2−(h0)2−e2aη(∂⊥h)2 i , (A10)
which has essentially the same form as Eq. (A5). The other components of the even parity metric perturbations are determined through the constraint equations (A7)–(A9).
Appendix B: Positive frequency mode function in the F-region
Here we briefly review the positive frequency mode function associated with the Minkowski vacuum state in the F region [26]. The relation between the coordinates in the future expanding Kasner spacetime (T, χ) and Minkowski spacetime (t, x) is given by
Here, T andχ are related to the coordinates used in the main text, (η, ζ), as T =eaη/a and χ =aζ. The integral representation for the Hankel function is given by
e−ircoshK+issinhK= 1 2i Z ∞ −∞ dpe−iKp r+s r−s ip/2 eπp/2Hip(2) r2−s21/2, (B2)
where Im(r±s)<0. By transforming the variables ass=κx, r=κt, K= sinh−1(q/κ), we obtain
e−iωkt+iqx= 1 2i Z ∞ −∞ dpe−iKp t+x t−x ip/2 eπp/2Hip(2)κ t2−x21/2 , (B3) whereωκ=κcoshK= p
q2+κ2. In terms of the coordinates (T, χ), the right-hand side is written as
e−iωκt+iqx= 1
2i Z ∞
−∞
dpe−iKpeipχeπp/2Hip(2)(κT). (B4)
This equation shows that Hip(2)(κT) is indeed the positive frequency mode function associated with the Minkowski vacuum state. It is easy to see that the complex conjugate gives the negative frequency mode function via the relation
h
eπp/2Hip(2)(κT)i
∗
=e−πp/2Hip(1)(κT). (B5)
Appendix C: Quantized GWs: Summary of results
We present a summary of the results of the quantized gravitational waves in the F, P, R, and L regions. The quantized metric perturbation in the F-region can be written by introducingω=|kx|as
ˆ hFµν(η, ζ, y, z) =√16πG X λ=o,e Z ∞ −∞ dkx Z ∞ −∞ d2k⊥ h e hF,λµν (η, kx,k⊥)eikxζ+ik⊥·x⊥ˆaF,λkx,k⊥+ h.c. i =√16πG X λ=o,e Z ∞ 0 dω Z ∞ −∞ d2k⊥ h e hF,λµν (η,−ω,k⊥)e−iωζ+ik⊥·x⊥ˆaI,λω,k⊥+eh F,λ µν (η, ω,−k⊥)eiωζ−ik⊥·x⊥aˆII,λω,k⊥+ h.c. i , (C1) where the Fourier componentsehF,µνo andehF,µνe are given by Eqs. (27) and (28) with
ϕF,λ= −i 2πp4asinh(π|kx|/a)J−i|kx|/a κeaη a , (C2)
We have a similar expression in the P-region ˆ hPµν(˜η,ζ, y, z)˜ =√16πG X λ=o,e Z ∞ −∞ dkx Z ∞ −∞ d2k⊥ h e hP,λµν (˜η, kx,k⊥)eikx ˜ ζ+ik⊥·x⊥aP,λ kx,k⊥+ h.c. i =√16πG X λ=o,e Z ∞ 0 dω Z ∞ −∞ d2k⊥ h e hP,λµν ( ˜η,ω,k⊥)eiω ˜ ζ+ik⊥·x⊥ˆaI,λ ω,k⊥+eh P,λ µν (˜η,−ω,−k⊥)e−iω ˜ ζ−ik⊥·x⊥ˆaII,λ ω,k⊥+ h.c. i . (C3) Here the metric perturbationsehP,µνo andehP,µνe are given by Eqs. (40) and (41), respectively, with
ϕP,λ= i 2πp4asinh(π|kx|/a) Ji|kx|/a κe−aη˜ a , (C4)
The quantized tensor mode in the right Rindler wedge is obtained as ˆ hRµν(τ, ξ, y, z) =√16πG X λ=o,e Z ∞ 0 dω Z ∞ −∞ d2k⊥ h e hR,λµν (ω, ξ,k⊥)e−iωτ+ik⊥·x⊥ˆaI,λω,k⊥+ h.c. i , (C5)
where the metric perturbationsehR,µνo and ˜hR,µνe are given by Eqs. (61) and (62), respectively, with
ϕR,λ= r sinhπω/a 4π4a e −iωτK iω/a κeaξ a . (C6)
The quantized tensor mode in the left Rindler wedge is obtained as ˆ hLµν(eτ ,ξ, y, z) =˜ √ 16πG X λ=o,e Z ∞ 0 dω Z ∞ −∞ d2k⊥ h ehL,λµν (ω,ξ,˜k⊥)e−iωτ˜−ik⊥·x⊥ˆaII,λω,k⊥+ h.c. i , (C7)
where the metric perturbationsehL,µνo and ˜hL,µνe are given by Eqs. (74) and (75), respectively, with
ϕL,λ= r sinhπω/a 4π4a e −iω˜τK iω/a κeaξ˜ a ! . (C8)
Appendix D: Energy density of GWs
In this appendix, we derive the vacuum expectation value of the energy density of GWs [Eqs. (145) and (146)]. The energy-momentum tensor of GWs is given by the second-order part of the Einstein tensor as
TµνGW=− 1
8πGh
(2)Gµνi, (D1)
whereh· · · idenotes spatial and temporal average. For example, the time-time component of this energy-momentum tensor gives the energy density of GWs. Let us now focus on the odd modes in the R region. From a direct computation with the use of the linear equation of motion (47), we find
(2)G
τ τ =− 1 2(∂τφ
R)2+ (total derivative), (D2)
where the total derivative will be omitted upon averaging. We obtain the similar results of other components. Thus, it can be seen that the energy density can be expressed simply in terms of the master variable. It then follows immediately that Tτ τGW= 1 16πGh(∂τφ R)2i=h(∂ τϕR(o)) 2i, (D3)
where one should recall thatφR=√16πGϕR
(o).
Let us move to the calculation of the vacuum expectation value of the energy density with respect to the Minkowski vacuum,
Mh0|Tˆτ τGW|0iM =Mh0|(∂τϕˆR(o)) 2
|0iM. (D4)
Substituting Eq. (99) into Eq. (D4), we obtain
Mh0|Tˆτ τGW|0iM = lim x0→x Z ∞ 0 dω Z dk⊥ " ∂τvω,R,ko⊥(x)∂τ0vω,R,ko∗ ⊥(x 0) 1−e−2πω/a + ∂τvω,R,ko∗⊥(x)∂τ0vω,R,ko ⊥(x 0) e2πω/a−1 # = lim x0→x Z ∞ 0 dωω2 Z dk⊥ " vω,R,ko ⊥(x)v R,o∗ ω,k⊥(x 0) 1−e−2πω/a + vR,ω,ko∗ ⊥(x)v R,o ω,k⊥(x 0) e2πω/a−1 # , (D5)
where we used the following formulas, Mh0|ˆaR,ω,ko⊥ˆa R,o† ω0,k0 ⊥|0iM = 1 1−e−2πω/aδ(ω−ω 0)δ(k ⊥−k0⊥), (D6) Mh0|ˆaω,R,ko†⊥ˆaR,ω0o,k0 ⊥|0iM = e−2πω/a 1−e−2πω/aδ(ω−ω 0)δ(k ⊥−k0⊥). (D7)
Note here that for a technical reason we first consider two separate spacetime points and then take the coincident limit. Thek⊥-integration can be done following Ref. [21]. First, we have
Z dk⊥vR,ω,ko⊥(x)v R,o∗ ω,k⊥(x 0) = sinh(πω/a) 2π3a e −iω(τ−τ0)Z ∞ 0 dκκKiω/a(ακ)Kiω/a(βκ)J0(γκ), (D8)
where α=eaξ/a, β=eaξ0/a, γ =|x
⊥−x0⊥|, and we used K−ν(z) =Kν(z) and R2π
0 dϕe
iκγcosϕ = 2πJ
0(κγ). The
integral in the right-hand side can be performed by using the formula [25, 26], Z ∞ 0 dκκν+1Kµ(ακ)Kµ(βκ)Jν(γκ) = 1 2 r π 2 γν (αβ)ν+1Γ(ν+µ+ 1)Γ(ν−µ+ 1) Θ 2−1−ν/2−1/4 B−µ−ν−1/12/2(Θ), (D9) where Θ = (α2+β2+γ2)/2αβ andBµ−−ν−1/12/2(ζ) (Re(µ±ν)>−1, Re(ν)>−1) is the associated Legendre function,
Biω/a−1/2−1/2(ζ) = √1 2π 1 iω/a 1 (ζ2−1)1/4 ζ+pζ2−1iω/a−ζ+p ζ2−1−iω/a . (D10)
Explicitly, in our case we have Θ = (ea(ξ−ξ0)+e−a(ξ−ξ0)+a2e−a(ξ+ξ0)|x⊥−x0⊥|2)/2 and
Z ∞ 0 dκκKiω/a κeaξ a Kiω/a κeaξ0 a ! J0(γκ) = πa 2e−a(ξ+ξ0) 4isinh(πω/a) 1 √ Θ2−1 Θ +pΘ2−1iω/a−Θ +pΘ2−1−iω/a . (D11)
In the coincidence limit,x0→x(Θ→1), this reduces to
lim x→x0 Z ∞ 0 dκκKiω/a κeaξ a Kiω/a κeaξ0 a ! J0(γκ) = πaωe−2aξ 2 sinh(πω/a), (D12) where we used lim Θ→1 1 √ Θ2−1 Θ +pΘ2−1iω/a−Θ +pΘ2−1−iω/a = 2iω a . (D13)
Therefore, we finally obtain the following result:
lim x0→x Z ∞ 0 dωω2 Z dk⊥ " vR,ω,ko ⊥(x)v R,o∗ ω,k⊥(x 0) 1−e−2πω/a + vω,R,ko∗ ⊥(x)v R,o ω,k⊥(x 0) e2πω/a−1 # = e −2aξ 4π2 Z ∞ 0 dω ω3 1−e−2πω/a + ω3 e2πω/a−1 = e −2aξ 4π2 Z ∞ −∞ dω ω3 e2πω/a−1 , (D14)
where in evaluating the first term we changed the integration variable asω→ −ω.
In the case of the Rindler vacuum state, it is easy to see that the vacuum expectation value is given by
Rh0|Tˆτ τGW|0iR= lim x0→x Z ∞ 0 dω Z dk⊥∂τvR,ω,ko⊥(x)∂τ0v R,o∗ ω,k⊥(x 0) =e −2aξ 4π2 Z ∞ 0 dωω3. (D15)
Then, we obtain the regularized energy density (147). Repeating similar computations, we obtain (151) and (152), where we use the following results
lim x0→x ∂ ∂ξ ∂ ∂ξ0 Z dk⊥vR,ω,ko⊥(x)v R,o∗ ω,k⊥(x 0) = e−2aξ 4π2 (4a2+ω2) 3 , lim x0→x ∂ ∂y ∂ ∂y0 Z dk⊥vR,ω,ko⊥(x)v R,o∗ ω,k⊥(x 0) = e−4aξ 4π2 (a2+ω2) 3 , lim x0→x ∂ ∂z ∂ ∂z0 Z dk⊥vR,ω,ko⊥(x)v R,o∗ ω,k⊥(x 0) = e−4aξ 4π2 (a2+ω2) 3 , lim x0→x ∂ ∂xµ ∂ ∂xν0 Z dk⊥vR,ω,ko⊥(x)v R,o∗ ω,k⊥(x 0) = 0 (forµ6=ν). (D16)
[1] B. P. Abbott et al. (LIGO Scientific Collaboration and Virgo Collaboration) Phys. Rev. Lett.116, 061102 “Observation of Gravitational Waves from a Binary Black Hole Merger”(2016).
[2] B. P. Abbott et al. (LIGO Scientific Collaboration and Virgo Collaboration) Phys. Rev. Lett. Phys. Rev. Lett.119, 161101 “GW170817: Observation of Gravitational Waves from a Binary Neutron Star Inspiral”(2017).
[3] Y. Akrami et al. (Planck Collaboration) Astron. and Astrophys.641, A10 “Planck 2018 results. X. Constraints on infla-tion”(2020).
[4] D. J. Schwarz, C. J. Copi, D. Huterer, and G. D. Starkman, Class. Quant. Grav.33, 184001, 1510.07929 “CMB Anomalies after Planck”(2016).
[5] P. Fosalba, E. Gaztanaga, arXiv:2011.00910
[6] M. Watanabe, S. Kanno, J. Soda, Prog. Theor. Phys. 123 , 1041-1068 “The Nature of Primordial Fluctuations from Anisotropic Inflation”(2010).
[7] J. Soda, “Statistical anisotropy from anisotropic inflation” Classical and Quantum Gravity, 29, 083001 (2012).
[8] A. Dey, E. D. Kovetz, and S. Paban, , J. Cosmol. Astropart. Phys.06, 025 “Power spectrum and non-Gaussianities in anisotropic inflation”(2014).
[9] A. Maleknejad, M.M. Sheikh-Jabbari, J. Soda, “Gauge Fields and Inflation” Phys. Rept.528161 (2013).
[10] K. Tomita, M. Den, “Gauge-invariant perturbations in anisotropic homogeneous cosmological models” Phys. Rev. D.34
3570 (1986)
[11] T. S. Pereira, C. Pitrou, J-P. Uzan, “Theory of cosmological perturbations in an anisotropic universe” JCAP0709:006 (2007)
[12] C. Pitrou, T. S. Pereira, J-P. Uzan, “Predictions from an anisotropic inflationary era” JCAP 0804:004 (2008)
[13] A. E. Gumrukcuoglu, C. R. Contaldi, M. Peloso, “Inflationary perturbations in anisotropic backgrounds and their imprint on the CMB” JCAP0711:005,2007
[14] H. T. Cho and A. D. Speliotopoulos, “Gravitational waves in Binchi type-I universe: The classical theory” Phys. Rev. D
525445 (1995).
[15] Y. Furuya, Y. Niiyama and Y. Sendouda, JCAP 01 009 “Probing pre-inflationary anisotropy with directional variations in the gravitational wave background”(2017).
[16] Y. Furuya, Y. Niiyama and Y. Sendouda, Class.Quant.Grav.36085007 “A quantisation procedure in the presence of an initial Kasner singularity: primordial gravitational waves from triaxially anisotropic pre-inflation”(2019).
[17] E. Kasner, American Journal of Mathematics43, 217 “Geometrical Theorems on Einstein’s Cosmological Equations”(1921). [18] S.W. Hawking, Comm. Math. Phys.43199 “Particle creation by black holes”(1975).
[19] Unruh, W. G., Phys. Rev. D14, 870-892 “Notes on black-hole evaporation”(1976).
[20] L. C.B.Crispino, A. Higuchi, G. E.A.Matsas, Rev. Mod. Phys.80, 787 “The Unruh effect and its applications”(2008). [21] A. Higuchi, S. Iso, K. Ueda, K. Yamamoto, Phys. Rev. D96083531 “Entanglement of the Vacuum between Left, Right,
Future, and Past: The Origin of Entanglement-Induced Quantum Radiation”(2017). [22] K. Ueda, A. Higuchi, Y. Nan, Ar Rohim, K. Yamamoto, in preparation
[23] U.H. Gerlach, U.K. Sengupta, Phys. Rev. D19, 2268 “Gauge-invariant perturbations on most general spherically symmetric space-times”(1979).
[24] D. W. Sciama, P. Candelas, D. Deutsch, Advances in Phys.,30327 “Quantum field theory, horizons and thermodynamics” (1981).
[25] I. S. Gradshteyn. I. M. Ryzhik, A. Jeffrey, “Table of Integrals, Series, and Products, 5th edition” (Academic Press, 1994). [26] W. Magnus, F. Oberhettinger, R. P. Soni, “Formulas and Theorems for the Special Functions of Mathematical Physics”
