This paper was converted on www.awesomepapers.org from LaTeX by an anonymous user.
Want to know more? Visit the Converter page.

Semiclassical Rotating AdS Black Holes with Quantum Hair in Holography

Ryusei Hamaki to69647@gmail.com College of Systems Engineering and Science, Shibaura Institute of Technology, Saitama 330-8570, JAPAN    Kengo Maeda maeda302@sic.shibaura-it.ac.jp Faculty of Engineering, Shibaura Institute of Technology, Saitama 330-8570, JAPAN
Abstract

In the context of the AdS/CFT duality, we study semiclassical stationary rotating AdS black holes with non-trivial quantum hair in three and five dimensions. We construct these solutions by perturbing the BTZ black hole and the five-dimensional Myers-Perry AdS black hole according to holographic semiclassical equations. In the three-dimensional case, the vacuum expectation value of the stress-energy tensor diverges as 1/λn(n=1,2)\sim 1/\lambda^{n}~{}(n=1,2) along a radial null geodesic as the affine parameter λ\lambda approaches zero at the Cauchy horizon, depending on the type of perturbation. In the five-dimensional case, most hairy solutions exhibit strong divergences, either in the stress-energy tensor or in the parallelly propagated Riemann components, along the radial null geodesic crossing the Cauchy horizon. Nevertheless, there exists a specific class of semiclassical solutions that retain a C0C^{0}-regular Cauchy horizon, where perturbations remain bounded. For extremal black holes, the vacuum expectation value of the stress-energy tensor diverges along a radial null geodesic transverse to the event horizon in both three and five dimensions, even though all components of the perturbed metric vanish in this limit.

I Introduction

Charged and rotating black holes have garnered significant attention in both the gravity and high-energy physics communities. A key topic of interest is the stability of the Cauchy horizon, which lies deep inside the event horizon. According to the strong cosmic censorship (SCC) conjecture, proposed by Penrose [1], the evolution of spacetime should be uniquely determined by regular initial data under physically reasonable conditions. In this framework, the presence of a Cauchy horizon is generally considered unphysical, as it would indicate a breakdown of determinism in general relativity. In classical general relativity, it is well established that the Cauchy horizon is generically unstable against perturbations [2, 3]. However, the resulting singularity is weak in the sense that tidal distortions remain finite for an observer crossing the Cauchy horizon [4]. Consequently, many C0C^{0} extensions of the metric are possible beyond the Cauchy horizon, suggesting a potential violation of the SCC in classical theory.

Thus, it is natural to ask whether quantum effects can uphold the SCC conjecture. One tractable approach is to study quantum field theory in a fixed curved spacetime background. A quantized free scalar field has been extensively analyzed in various charged and rotating black hole backgrounds [5, 6, 7, 8, 9]. Recent work, for instance, rigorously demonstrated that the vacuum expectation value of the stress-energy tensor diverges along a radial null geodesic as Cλ2\sim C\lambda^{-2}, where λ\lambda is the affine parameter, near the Cauchy horizon (λ=0\lambda=0) of a four-dimensional Reissner-Nordström-de Sitter black hole [9]. In contrast, for the near-extremal rotating BTZ black hole, a three-dimensional vacuum solution with a negative cosmological constant [10], the vacuum expectation value remains finite, as shown in [8], due to the vanishing coefficient C=0C=0 (see [9] for detailed analysis). These findings indicate that the coefficient CC, which determines the formation of a strong curvature singularity at the Cauchy horizon, depends on the specific background spacetime. Therefore, it remains unclear whether quantum field theory in a fixed curved background can consistently support the SCC conjecture. An alternative scenario supporting the SCC conjecture could involve significant quantum modifications to the background geometry itself. As observed within a classical theory [11], one might expect that the backreaction destroys the Cauchy horizon of charged or rotating black holes.

Another intriguing feature of charged or rotating black holes is the existence of extremal states with zero temperature. Motivated by the AdS/CFT duality [12], recent studies have explored static or stationary perturbations of extremal Reissner-Nordström-AdS black holes and rotating AdS black holes near their event horizons [13, 14, 15, 16]. As one approaches the horizon from the outside, the energy density or a curvature component in the parallelly propagated frame along a causal geodesic crossing the horizon diverges, despite the perturbations converging to zero and all scalar curvature invariants remaining finite. Such singularities are absent in non-extremal black holes with finite temperature. However, the curvature at the horizon grows indefinitely as the temperature decreases [15]. This behavior suggests a potential violation of the weak cosmic censorship (WCC), which posits that a distant observer near future null infinity cannot witness the formation of a naked singularity. Therefore, it becomes important to investigate whether quantum effects can prevent extremal black holes from developing curvature singularities at their event horizons.

In this paper, we investigate semiclassical rotating AdS black holes in both non-extremal and extremal cases. Given the absence of a complete theory of quantum gravity, the most tractable approach is the semiclassical procedure, where gravity is treated classically, while matter fields are treated quantum mechanically through the semiclassical Einstein equations. The source term in these equations is provided by the vacuum expectation value of the stress-energy tensor. Previous studies have been limited to lower-dimensional semiclassical black hole solutions sourced by a quantized, free, massless scalar field [17], or to four-dimensional spherically symmetric solutions under the s-wave approximation of a massless, minimally coupled scalar field [18]. To extend these works, we explore semiclassical solutions for strongly coupled quantum fields with a gravity dual via the AdS/CFT duality.

Initially proposed in Ref.[19], the semiclassical equations on a dd-dimensional boundary spacetime can be reformulated as mixed boundary conditions on a (d+1d+1)-dimensional asymptotically AdS bulk spacetime by promoting the boundary metric to a dynamical field. Following this procedure, an analytical semiclassical (static) AdS black hole solution was obtained by introducing even-parity perturbations to the BTZ solution [20]. Extending this analysis to rotating cases, we construct three- and five-dimensional semiclassical rotating AdS black holes by perturbing the rotating BTZ solution and the five-dimensional Myers-Perry AdS solution [21], respectively.

In the three-dimensional case, the non-extremal solution is obtained analytically by adding odd-parity perturbation to the static hairy solution [20] and applying a Lorentz boost. For the rotating solution, the vacuum stress-energy tensor diverges as λn(n=1,2)\sim\lambda^{-n}~{}(n=1,2) along a radial null geodesic with affine parameter λ\lambda crossing the Cauchy horizon (λ=0\lambda=0), depending on the type of perturbation. In the five-dimensional case, we derive three coupled second-order differential equations with six independent solutions. Interestingly, while the stress-energy tensor or the parallelly propagated Riemann components along the radial null geodesic strongly diverges as λ2\sim\lambda^{-2} near the Cauchy horizon for most parameter choices, it is possible to fine-tune the parameters to construct a specific hairy solution with a C0C^{0}-regular Cauchy horizon.

In the extremal case, the semiclassical extremal solution in three-dimensions is obtained analytically. The perturbation falls off as a power law, (rr0)γ\sim(r-r_{0})^{\gamma}, near the extremal horizon, r=r0r=r_{0}, with a very slow decay rate, i. e. , 0<γ<10<\gamma<1. This slow decay indicates the formation of a curvature singularity on the horizon, similar to the case of classical perturbations in extremal black holes [13, 14, 15, 16]. This singular behavior also leads to the divergence of the vacuum expectation value of the stress-energy tensor along a radial null geodesic transverse to the horizon. In the five-dimensional case, there are six independent mode solutions near the extremal horizon. We show that the power-law index γ\gamma satisfies γ<2\gamma<2 for all parameters of the background Myers-Perry AdS solution, similar to previous semiclassical models [18]. This suggests that the formation of a curvature singularity on the extremal horizon is a generic feature of semiclassical extremal rotating black holes.

This paper is organized as follows: In section II, we briefly review the holographic set up for constructing the semiclassical solutions. In section III and IV, we construct three and five-dimensional semiclassical rotating AdS black hole solutions. Section V is devoted to summary and discussions.

II Preliminaries: Holographic setting

We start with d+1(d=3, 5)d+1~{}(d=3,\,5)-dimensional asymptotically AdS bulk spacetime with metric

dsd+12\displaystyle ds_{d+1}^{2} =GMNdXMdXN\displaystyle=G_{MN}dX^{M}dX^{N}
=Ω2(z)dz2+gμν(z,x)dxμdxν=Ω2(z)(dz2+dsd2),\displaystyle=\Omega^{-2}(z)dz^{2}+g_{\mu\nu}(z,x)dx^{\mu}dx^{\nu}=\Omega^{-2}(z)(dz^{2}+ds_{d}^{2}),
dsd2\displaystyle ds_{d}^{2} :=g~μν(z,x)dxμdxν,Ω(z):=Lsinz,\displaystyle:=\tilde{g}_{\mu\nu}(z,x)dx^{\mu}dx^{\nu},\qquad\Omega(z):=\frac{\ell}{L}\sin\frac{z}{\ell}, (1)

where XM=(z,xμ)X^{M}=(z,\,x^{\mu}) and LL (\ell) is the bulk (boundary) AdS length and the AdS boundary is located at z=0z=0 and z=πz=\pi. The bulk action consists of the d+1d+1-dimensional Einstein Hilbert action SbulkS_{\text{bulk}}, Gibbons-Hawking term SGHS_{\text{GH}}, and the counter term Sct(d)S_{\text{ct}}^{(d)} given by

Sbulk=SEH+SGH+Sct(d)\displaystyle S_{\text{bulk}}=S_{\text{EH}}+S_{\text{GH}}+S_{\text{ct}}^{(d)}
=dd+1XG16πGd+1(R(G)+d(d1)L2)+ddxg8πGd+1K+Sct(d),\displaystyle=\int\frac{d^{d+1}X\sqrt{-G}}{16\pi G_{d+1}}\left(R(G)+\frac{d(d-1)}{L^{2}}\right)+\int\frac{d^{d}x\sqrt{-g}}{8\pi G_{d+1}}K+S_{\text{ct}}^{(d)}, (2)

where GdG_{d} is the dd-dimensional gravitational constant and Sct(d)S^{(d)}_{\text{ct}} is given by

Sct(3)=d3xg16πG4(4L+LR(g)),\displaystyle S^{(3)}_{\text{ct}}=-\int\frac{d^{3}x\sqrt{-g}}{16\pi G_{4}}\left(\frac{4}{L}+LR(g)\right), (3)
Sct(5)=d5xg16πG6(8L+L3R(g)+L39{Rμν(g)Rμν(g)516R2(g)}).\displaystyle S^{(5)}_{\text{ct}}=-\int\frac{d^{5}x\sqrt{-g}}{16\pi G_{6}}\left(\frac{8}{L}+\frac{L}{3}R(g)+\frac{L^{3}}{9}\left\{R_{\mu\nu}(g)R^{\mu\nu}(g)-\frac{5}{16}R^{2}(g)\right\}\right). (4)

We consider dd-dimensional spacetime with the conformal boundary metric 𝒢μν:=limz0g~μν(z,x)\mathcal{G}_{\mu\nu}:=\lim_{z\to 0}\tilde{g}_{\mu\nu}(z,x) satisfying the semiclassical Einstein equations [20]. According to the AdS/CFT dictionary [22, 23], the vacuum expectation value of the boundary stress-energy tensor is given by the variation of the bulk action (II) with respect to the conformal boundary metric 𝒢μν\mathcal{G}_{\mu\nu} as 111In detail derivation, see Ref. [20].

𝒯μν=limz018πGd+1L[L2(d2)Ωd2{K~K~μνg~μν2(K~αβK~αβ+K~2)}\displaystyle\left\langle~{}\mathcal{T}_{\mu\nu}~{}\right\rangle=\lim_{z\to 0}\frac{1}{8\pi G_{d+1}L}\Biggl{[\dfrac{L^{2}}{(d-2)\Omega^{d-2}}\Bigl{\{}\tilde{K}\tilde{K}_{\mu\nu}-\dfrac{\tilde{g}_{\mu\nu}}{2}(\tilde{K}^{\alpha\beta}\tilde{K}_{\alpha\beta}+\tilde{K}^{2}})\Bigr{\}}
Lg~νρ(LΩd2z+1)K~μρδμρK~Ωd1\displaystyle-L\tilde{g}_{\nu\rho}\left(\frac{L\Omega}{d-2}\frac{\partial}{\partial z}+1\right)\frac{{\tilde{K}_{\mu}}^{\rho}-{\delta_{\mu}}^{\rho}\tilde{K}}{\Omega^{d-1}}
+(d1)g~μν(d2)Ωd{L2Ω2(ΩΩ)+1d22(1LΩ)2}]+τ(d)μν,\displaystyle+\frac{(d-1)\tilde{g}_{\mu\nu}}{(d-2)\Omega^{d}}\Bigl{\{}L^{2}\Omega^{2}\left(\frac{\Omega^{\prime}}{\Omega}\right)^{\prime}+1-\frac{d-2}{2}(1-L\Omega^{\prime})^{2}\Bigr{\}}\Biggr{]}+\tau^{(d)}_{\mu\nu}, (5)

where a prime denotes differentiation with respect to zz and the extrinsic curvature K~μν\tilde{K}_{\mu\nu} of the conformal metric g~μν\tilde{g}_{\mu\nu} is defined by

K~μν=12zg~μν,\displaystyle\tilde{K}_{\mu\nu}=-\frac{1}{2}\partial_{z}\tilde{g}_{\mu\nu}, (6)

and

τμν(3)\displaystyle\tau^{(3)}_{\mu\nu} =0,\displaystyle=0,
τμν(5)\displaystyle\tau^{(5)}_{\mu\nu} =L372πG6Ω{(D~258R~)R~μν38D~μD~νR~+2R~αμβνR~αβ\displaystyle=\frac{L^{3}}{72\pi G_{6}\Omega}\Biggl{\{}\left(\tilde{D}^{2}-\frac{5}{8}\tilde{R}\right)\tilde{R}_{\mu\nu}-\frac{3}{8}\tilde{D}_{\mu}\tilde{D}_{\nu}\tilde{R}+2\tilde{R}_{\alpha\mu\beta\nu}\tilde{R}^{\alpha\beta}
g~μν2(R~αβR~αβ+14D~2R~516R~2)}.\displaystyle-\frac{\tilde{g}_{\mu\nu}}{2}\left(\tilde{R}_{\alpha\beta}\tilde{R}^{\alpha\beta}+\frac{1}{4}\tilde{D}^{2}\tilde{R}-\frac{5}{16}\tilde{R}^{2}\right)\Biggr{\}}. (7)

Here, R~μν\tilde{R}_{\mu\nu} is the Ricci tensor of the conformal metric g~μν\tilde{g}_{\mu\nu}.

The bulk geometry (II) is the solution of the d+1d+1-dimensional bulk Einstein equations with a negative cosmological constant Λd+1=d(d1)/(2L2)\Lambda_{d+1}=-d(d-1)/(2L^{2}) if the dd-dimensional metric g~μν=g~μν(x)\tilde{g}_{\mu\nu}=\tilde{g}_{\mu\nu}(x) satisfies the vacuum Einstein equations with a negative cosmological constant Λd=(d1)(d2)/(22)\Lambda_{d}=-(d-1)(d-2)/(2\ell^{2}). So, if the metric g~μν\tilde{g}_{\mu\nu} corresponds to that of the BTZ black hole [10] or the 55-dimensional Myers-Perry AdS black hole [21], the bulk Einstein equations are automatically satisfied. Furthermore, the vacuum expectation value of the stress-energy tensor (II) vanishes identically in both d=3, 5d=3,\,5 cases, due to the absence of the Weyl anomaly in odd dimensions. This implies that these bulk geometries are solutions to the semiclassical Einstein equations 222Here, for simplicity, we ignored higher curvature corrections in d=5d=5 case. on the boundary:

μν2𝒢μν+Λd𝒢μν=8πGd𝒯μν,\displaystyle\mathcal{R}_{\mu\nu}-\frac{\mathcal{R}}{2}\mathcal{G}_{\mu\nu}+\Lambda_{d}\,\mathcal{G}_{\mu\nu}=8\pi G_{d}\left\langle~{}\mathcal{T}_{\mu\nu}~{}\right\rangle, (8)

where μν\mathcal{R}_{\mu\nu} is the Ricci tensor of the metric 𝒢μν\mathcal{G}_{\mu\nu}.

We aim to find semiclassical solutions of Eqs. (8) with non-zero expectation values, i. e. , 𝒯μν0\left\langle~{}\mathcal{T}_{\mu\nu}~{}\right\rangle\neq 0, by perturbing the background solution. To achieve this, we expand the metric g~μν\tilde{g}_{\mu\nu} as

g~μν=𝒢¯μν(x)+ϵhμν(z,x)+,\displaystyle\tilde{g}_{\mu\nu}=\overline{\mathcal{G}}_{\mu\nu}(x)+\epsilon h_{\mu\nu}(z,\,x)+\cdots, (9)

where 𝒢¯μν(x)\overline{\mathcal{G}}_{\mu\nu}(x) is the background metric of the BTZ black hole or the 55-dimensional Myers-Perry AdS black hole, and ϵ\epsilon is an infinitesimally small parameter. Under the metric ansatz hμν=ξ(z)Hμν(x)h_{\mu\nu}=\xi(z)H_{\mu\nu}(x), the perturbed bulk Einstein equations can be decomposed with a separation constant m2m^{2} as

ξ′′+m2ξ(d1)ΩΩξ=0,\displaystyle\xi^{\prime\prime}+m^{2}\xi-\frac{(d-1)\Omega^{\prime}}{\Omega}\xi^{\prime}=0, (10)
𝒟¯2Hμν+2¯μανβHαβ=m2Hμν,\displaystyle\bar{\mathcal{D}}^{2}H_{\mu\nu}+2\overline{\mathcal{R}}_{\mu\alpha\nu\beta}H^{\alpha\beta}=m^{2}H_{\mu\nu}, (11)

where 𝒟¯μ\overline{\mathcal{D}}_{\mu} and ¯μναβ\overline{\mathcal{R}}_{\mu\nu\alpha\beta} are the covariant derivative and the Riemann tensor with respect to the background metric 𝒢¯μν\overline{\mathcal{G}}_{\mu\nu}, respectively, and we impose the transverse and traceless condition

hμμ=hμν𝒢¯μν=0,𝒟¯νhμν=0.\displaystyle{h_{\mu}}^{\mu}=h_{\mu\nu}\overline{\mathcal{G}}^{\mu\nu}=0\,,\quad\bar{\mathcal{D}}^{\nu}h_{\mu\nu}=0\,. (12)

The AdS boundary of the bulk spacetime (II) consists of two portions: z=0z=0, where the conformal boundary metric 𝒢μν\mathcal{G}_{\mu\nu} satisfying the semiclassical Einstein equations, and z=πz=\pi. According to the procedure [20], we shall impose the Dirichlet boundary condition at z=πz=\pi. As shown in Ref. [24], the solution of Eq. (10) is expressed by the hypergeometric function as

ξ(y)=C(1y)d2F(12p,12+p, 1+d2;1y),y:=1cosz2,\displaystyle\xi(y)=C(1-y)^{\frac{d}{2}}F\left(\frac{1}{2}-p,\,\frac{1}{2}+p,\,1+\frac{d}{2};1-y\right),\qquad y:=\frac{1-\cos\frac{z}{\ell}}{2}, (13)

where CC is a constant and pp is defined by

p=(d1)24+m^2\displaystyle p=\sqrt{\frac{(d-1)^{2}}{4}+\hat{m}^{2}} (14)

with m^2:=m22\hat{m}^{2}:=m^{2}\ell^{2}. As shown in the perturbations of the maximally symmetric AdS spacetimes [20, 24], the perturbed semiclassical Eqs. (8) reduce to an algebraic equation characterized by the dimensionless constant γd\gamma_{d}, defined by

γd:=GdLπGd+1(L)d2.\displaystyle\gamma_{d}:=\frac{G_{d}L}{\pi G_{d+1}}\left(\frac{L}{\ell}\right)^{d-2}. (15)

For d=3d=3, the first order perturbation of Eqs. (8) using Eqs. (II), (II), and (13) results in the algebraic equation [20]

γ3=tanπpπp,\displaystyle\gamma_{3}=\frac{\tan\pi p}{\pi p}, (16)

which allows a hairly solution in the range

γ3>1,1<m^2<34.\displaystyle\gamma_{3}>1,\qquad-1<\hat{m}^{2}<-\frac{3}{4}. (17)

For d=5d=5, the perturbed semiclassical Eqs. (8) in the Myers-Perry AdS black hole background reduce to the algebraic equation (77), which is identical to that derived in the perturbation of the 55-dimensional maximally symmetric AdS solution [24]. The solution was found for a negative mass satisfying the inequality

154<m^2<74\displaystyle-\frac{15}{4}<\hat{m}^{2}<-\frac{7}{4} (18)

when γ5\gamma_{5} is larger than a critical value [24]. In the following sections, we shall focus on the solutions of (11) under the inequalities, (17) and (18), ensuring that the algebraic equations (16) and (77) are also satisfied.

III 33-dimensional rotating AdS black holes

In this section, we investigate 33-dimensional rotating AdS black holes by solving Eqs. (11) for a negative mass satisfying the inequality (17) in the background of the rotating BTZ solution [10]. In the non-extremal case, an analytic solution is constructed by adding an odd-parity perturbation to the static solution [20] and by applying a Lorentz boost to the perturbed solution. In the extremal case, we also obtain an analytic semiclassical solution in which HμνH_{\mu\nu} in Eqs. (11) converges to zero near the horizon. Nevertheless, we demonstrate that the expectation value of the stress-energy tensor diverges along a null geodesic crossing the event horizon.

III.1 The non-extremal solution

We start with the metric ansatz for the non-extremal AdS black hole solution,

ds32=𝒢μνdxμdxν=(𝒢¯μν+ϵHμν)dxμdxν\displaystyle ds_{3}^{2}=\mathcal{G}_{\mu\nu}dx^{\mu}dx^{\nu}=(\overline{\mathcal{G}}_{\mu\nu}+\epsilon H_{\mu\nu})dx^{\mu}dx^{\nu}
=rh2g(u)2(1+sinh2βu)u(1+ϵK(u))dt2+24u2g(u)(1+ϵU(u))du2\displaystyle=-\frac{r_{h}^{2}g(u)}{\ell^{2}(1+\sinh^{2}\beta\,u)u}(1+\epsilon K(u))dt^{2}+\frac{\ell^{2}}{4u^{2}g(u)}(1+\epsilon U(u))du^{2}
+rh2(1+sinh2βu)u(1+ϵZ(u))[(coshβsinhβu(1+sinh2βu)+ϵΩ(u))dt+dφ]2,\displaystyle+\frac{r_{h}^{2}(1+\sinh^{2}\beta\,u)}{u}(1+\epsilon Z(u))\left[\left(-\frac{\cosh\beta\sinh\beta\,u}{\ell(1+\sinh^{2}\beta\,u)}+\epsilon\Omega(u)\right)dt+d\varphi\right]^{2},
g(u)=1u,\displaystyle g(u)=1-u, (19)

where 0u0\leq u\leq\infty, 0φ2π0\leq\varphi\leq 2\pi, β(0<β<)\beta~{}(0<\beta<\infty) is a boost parameter, and the event horizon (the AdS boundary) is located at u=1u=1 (u=0u=0). It can be verified that the metric with ϵ=0\epsilon=0 reduces to standard form of the rotating BTZ solution [10] through the coordinate transformation

u=rh2r2rh2sinh2β,\displaystyle u=\frac{r_{h}^{2}}{r^{2}-r_{h}^{2}\sinh^{2}\beta}, (20)

where the radii of the event horizon and the Cauchy horizon correspond to r=rhcoshβr=r_{h}\cosh\beta, and r=rhsinhβr=r_{h}\sinh\beta (u=u=\infty), respectively.

Under the Lorentz boost,

(t/φ)=(coshβsinhβsinhβcoshβ)(t^/φ^),\displaystyle\begin{pmatrix}t/\ell\\ \varphi\end{pmatrix}=\begin{pmatrix}\cosh\beta&\sinh\beta\\ \sinh\beta&\cosh\beta\end{pmatrix}\begin{pmatrix}\hat{t}/\ell\\ \hat{\varphi}\end{pmatrix}, (21)

the metric (III.1) reduces to

ds32=rh2g2u(1+ϵT(u))dt^2+24u2g(1+ϵU(u))du2+rh2u(1+ϵR(u))dφ^22ϵrh2uS(u)dt^dφ^,\displaystyle ds_{3}^{2}=-\frac{r_{h}^{2}g}{\ell^{2}u}(1+\epsilon T(u))d\hat{t}^{2}+\frac{\ell^{2}}{4u^{2}g}(1+\epsilon U(u))du^{2}+\frac{r_{h}^{2}}{u}(1+\epsilon R(u))d\hat{\varphi}^{2}-\frac{2\epsilon r_{h}^{2}}{u}S(u)d\hat{t}d\hat{\varphi}, (22)

where

K=gR+T+(TgR)cosh(2β)2sinh(2β)S2+u(cosh(2β)1),\displaystyle K=\frac{gR+T+(T-gR)\cosh(2\beta)-2\ell\sinh(2\beta)S}{2+u(\cosh(2\beta)-1)},
Z=R+gT+(RgT)cosh(2β)+2sinh(2β)S2+u(cosh(2β)1),\displaystyle Z=\frac{R+gT+(R-gT)\cosh(2\beta)+2\ell\sinh(2\beta)S}{2+u(\cosh(2\beta)-1)},
Ω=2S{u+(2u)cosh(2β)}+g(RT)sinh(2β){2+u(cosh(2β)1)}2.\displaystyle\Omega=-2\frac{\ell S\{u+(2-u)\cosh(2\beta)\}+g(R-T)\sinh(2\beta)}{\ell\{2+u(\cosh(2\beta)-1)\}^{2}}. (23)

In this metric form (22), the perturbation for SS is decoupled from the other variables, (U,T,R)(U,\,T,\,R), since they correspond to odd and even-parity mode perturbations under φ^φ^\hat{\varphi}\leftrightarrow-\hat{\varphi} transformation.

For the even-parity mode, one obtains a master equation of UU [25], and RR, TT are determined by Eq. (12) as

R=(T+U),\displaystyle R=-(T+U), (24)
T=(23u)U2(1u)U,\displaystyle T=-\left(2-\frac{3}{u}\right)U-2(1-u)U^{\prime}, (25)
gU′′2uU+8m^24u2U=0,\displaystyle gU^{\prime\prime}-\frac{2}{u}U^{\prime}+\frac{8-\hat{m}^{2}}{4u^{2}}U=0, (26)

where a prime denotes differentiation with respect to uu. For the odd-parity mode, the equation of SS becomes

gS′′m^24u2S=0.\displaystyle gS^{\prime\prime}-\frac{\hat{m}^{2}}{4u^{2}}S=0. (27)

By imposing regularity on the event horizon, u=1u=1, the solutions are analytically given by

U=c1ζ(u),S=c2guζ(u),\displaystyle U=c_{1}\zeta(u),\qquad S=\frac{c_{2}g}{\ell u}\zeta(u),
ζ(u):=u3p2F(1p2,3p2,2;1u),\displaystyle\zeta(u):=u^{\frac{3-p}{2}}F\left(\frac{1-p}{2},\frac{3-p}{2},2;1-u\right), (28)

where c1c_{1}, c2c_{2} are arbitrary constants, FF is the hypergeometric function, and pp is the parameter defined by p=1+m^2p=\sqrt{1+\hat{m}^{2}} (14). So, by Eqs. (III.1), (24), and (25), the general rotating AdS black hole solution with quantum hair is described by the two parameters, c1c_{1} and c2c_{2}, as follows:

K=c1{u(2u)+(u26u+6)cosh(2β)}2c2(1u)sinh(2β)u(2u+ucosh(2β))ζ2c1g{u+(2u)cosh(2β)}2u+ucosh(2β)ζ,\displaystyle K=\frac{c_{1}\{u(2-u)+(u^{2}-6u+6)\cosh(2\beta)\}-2c_{2}(1-u)\sinh(2\beta)}{u(2-u+u\cosh(2\beta))}\zeta-\frac{2c_{1}g\{u+(2-u)\cosh(2\beta)\}}{2-u+u\cosh(2\beta)}\zeta^{\prime},
L=2c1{u(2u)+(u23u+3)cosh(2β)}c2(1u)sinh(2β)u(2u+ucosh(2β))ζ+2c1g{u+(2u)cosh(2β)}2u+ucosh(2β)ζ,\displaystyle L=-2\frac{c_{1}\{u(2-u)+(u^{2}-3u+3)\cosh(2\beta)\}-c_{2}(1-u)\sinh(2\beta)}{u(2-u+u\cosh(2\beta))}\zeta+\frac{2c_{1}g\{u+(2-u)\cosh(2\beta)\}}{2-u+u\cosh(2\beta)}\zeta^{\prime},
Ω=2gc2{u+(2u)cosh(2β)}3c1(2u)sinh(2β)u(2u+ucosh(2β))2ζ8c1g2sinh(2β)(2u+ucosh(2β))2ζ.\displaystyle\Omega=-2g\frac{c_{2}\{u+(2-u)\cosh(2\beta)\}-3c_{1}(2-u)\sinh(2\beta)}{\ell u(2-u+u\cosh(2\beta))^{2}}\zeta-\frac{8c_{1}g^{2}\sinh(2\beta)}{\ell(2-u+u\cosh(2\beta))^{2}}\zeta^{\prime}. (29)

Expanding ζ\zeta (III.1) near the null infinity, u=0u=0, one obtains

U=O(u3p2),R,T,S=O(u1p2).\displaystyle U=O(u^{\frac{3-p}{2}}),\qquad R,\,T,\,S=O(u^{\frac{1-p}{2}}). (30)

This guaranties the asymptotic boundary condition that KK, ZZ, and Ω\Omega decay to zero at u=0u=0 for 0<p<1/20<p<1/2 (17).

Now, consider the behavior of the vacuum expectation value of the perturbed stress-energy tensor δTμν\delta\left\langle~{}T_{\mu\nu}~{}\right\rangle near the Cauchy horizon, u=u=\infty. Near the AdS boundary z=0z=0, the metric hμνh_{\mu\nu} in (9) can be expanded as

hμν(z,x)=hμν(0)(x)+(zL)2hμν(2)(x)+(zL)3hμν(3)(x)+,\displaystyle h_{\mu\nu}(z,\,x)=h^{(0)}_{\mu\nu}(x)+\left(\frac{z}{L}\right)^{2}h^{(2)}_{\mu\nu}(x)+\left(\frac{z}{L}\right)^{3}h^{(3)}_{\mu\nu}(x)+\cdots, (31)

and the coefficient hμν(3)(x)h^{(3)}_{\mu\nu}(x) corresponds to δTμν\delta\left\langle~{}T_{\mu\nu}~{}\right\rangle [20]. Consequently, under the metric ansatz, hμν=ξ(z)Hμν(x)h_{\mu\nu}=\xi(z)H_{\mu\nu}(x), we obtain the perturbed stress-energy tensor in the coordinate system (22) as

δTμνdxμdxν3ϵ16πG4Lδ(ds32)\displaystyle\delta\left\langle~{}T_{\mu\nu}~{}\right\rangle dx^{\mu}\otimes dx^{\nu}\sim\frac{3\epsilon}{16\pi G_{4}L}\delta(ds_{3}^{2})
=3ϵ16πG4L(rh2g2uT(u)dt^2+24u2gU(u)du2+rh2uR(u)dφ^22rh2uS(u)dt^dφ^).\displaystyle=\frac{3\epsilon}{16\pi G_{4}L}\left(-\frac{r_{h}^{2}g}{\ell^{2}u}T(u)d\hat{t}^{2}+\frac{\ell^{2}}{4u^{2}g}U(u)du^{2}+\frac{r_{h}^{2}}{u}R(u)d\hat{\varphi}^{2}-\frac{2r_{h}^{2}}{u}S(u)d\hat{t}d\hat{\varphi}\right). (32)

Let us denote the tangent vector along an affine parameterized null geodesic crossing the Cauchy horizon as VμV^{\mu}. Decomposing the tangent null vector VμV^{\mu} into the background V¯μ\overline{V}^{\mu} and the first order deviation δVμ\delta V^{\mu}, i. e. , Vμ=V¯μ+δVμV^{\mu}=\overline{V}^{\mu}+\delta V^{\mu}, one obtains

δ(TμνVμVν)=2Tμν¯δVμV¯ν+δTμνV¯μV¯ν=δTμνV¯μV¯ν,\displaystyle\delta(\left\langle~{}T_{\mu\nu}~{}\right\rangle V^{\mu}V^{\nu})=2\overline{\left\langle~{}T_{\mu\nu}~{}\right\rangle}\delta V^{\mu}\overline{V}^{\nu}+\delta\left\langle~{}T_{\mu\nu}~{}\right\rangle\overline{V}^{\mu}\overline{V}^{\nu}=\delta\left\langle~{}T_{\mu\nu}~{}\right\rangle\overline{V}^{\mu}\overline{V}^{\nu}, (33)

where we used the fact the background stress-energy tensor Tμν¯\overline{\left\langle~{}T_{\mu\nu}~{}\right\rangle} is zero in the second equality.

The background null vector V¯μ\overline{V}^{\mu} in the metric (22) is expressed by two parameters {\cal E} and 𝒫{\cal P},

V¯μ=u(2rh(2g𝒫2)uu+gt^+𝒫φ^)μ.\displaystyle\overline{V}^{\mu}=u\left(\frac{2r_{h}}{\ell}\sqrt{({\cal E}^{2}-g{\cal P}^{2})u}\frac{\partial}{\partial u}+\frac{\ell{\cal E}}{g}\frac{\partial}{\partial\hat{t}}+{\cal P}\frac{\partial}{\partial\hat{\varphi}}\right)^{\mu}. (34)

Contracting δTμν\delta\left\langle~{}T_{\mu\nu}~{}\right\rangle in Eq. (III.1) with V¯μ\overline{V}^{\mu}, one obtains

δ(TμνVμVν)3ϵ16πG4Lrh2ug[𝒫2g(RU)2𝒫S+2(UT)].\displaystyle\delta(\left\langle~{}T_{\mu\nu}~{}\right\rangle V^{\mu}V^{\nu})\sim\frac{3\epsilon}{16\pi G_{4}L}\frac{r_{h}^{2}u}{g}\left[{\cal P}^{2}g(R-U)-2\ell{\cal E}{\cal P}S+{\cal E}^{2}(U-T)\right]. (35)

Expanding ζ\zeta in Eq. (III.1) as a series in 1/u1/u, the asymptotic behavior of ζ\zeta near the CH (u=u=\infty) is given by

ζαu,α:=4cos(πp2)π(1p2).\displaystyle\zeta\simeq\alpha u,\qquad\alpha:=\frac{4\cos\left(\frac{\pi p}{2}\right)}{\pi(1-p^{2})}. (36)

Inserting this into Eqs. (24), (25), and (III.1), one obtains the leading behaviors of UU, TT, RR, and SS near the CH as

URc1αu,T=O(1),Sc2αu,\displaystyle U\simeq-R\simeq c_{1}\alpha u,\quad T=O(1),\quad S\simeq-\frac{c_{2}\alpha}{\ell}u, (37)

and, by Eqs. (35), the expectation value of the stress-energy tensor diverges as

δ(TμνVμVν)3ϵαrh28πG4L(c1𝒫2u2c2𝒫u)c1(λ2)+c2(λ1).\displaystyle\delta(\left\langle~{}T_{\mu\nu}~{}\right\rangle V^{\mu}V^{\nu})\sim-\frac{3\epsilon\alpha r_{h}^{2}}{8\pi G_{4}L}(c_{1}{\cal P}^{2}u^{2}-c_{2}{\cal E}{\cal P}u)\simeq c_{1}(\lambda^{-2})+c_{2}(\lambda^{-1}). (38)

Here, λ\lambda is the affine parameter along the null geodesic, with λ=0\lambda=0 corresponding to the location of the CH, u=u=\infty. For the even-parity solution with c10,c2=0c_{1}\neq 0,\,c_{2}=0, the expectation value of the stress-energy tensor strongly diverges, supporting the SCC conjecture. In contrast, for the odd-parity solution with c1=0,c20c_{1}=0,\,c_{2}\neq 0, the divergence is weaker in the sense that the distance η\eta between the two neighboring null geodesics remains finite at the Cauchy horizon. This is because integrating δ(TμνVμVν)\delta(\left\langle~{}T_{\mu\nu}~{}\right\rangle V^{\mu}V^{\nu}) twice with respect to λ\lambda results in a finite value at the CH [4]. Thus, a C0C^{0} extension of the metric beyond the CH is possible.

III.2 The extremal solution

The extremal rotating black hole corresponds to the limit β\beta\to\infty in the Lorentz boost (21). So, instead of the metric ansatz (III.1), we consider the following metric ansatz for the extremal rotating solution:

ds2=r022u(1u)2(1+ϵT(u))dt2+2(1+ϵU(u))4u2(1u)2du2\displaystyle ds^{2}=-\frac{r_{0}^{2}}{\ell^{2}u}(1-u)^{2}(1+\epsilon T(u))dt^{2}+\frac{\ell^{2}(1+\epsilon U(u))}{4u^{2}(1-u)^{2}}du^{2}
+r02u(1+ϵR(u)){(u+ϵS(u))dt+dφ}2,\displaystyle+\frac{r_{0}^{2}}{u}(1+\epsilon R(u))\left\{\left(-\frac{u}{\ell}+\epsilon S(u)\right)dt+d\varphi\right\}^{2}, (39)

where ϵ=0\epsilon=0 corresponds to the extremal BTZ black hole 333The coodinate transformation u=r02/r2u=r_{0}^{2}/r^{2} produces the standard form of the extremal BTZ black hole metric with horizon radius r0r_{0} [10].. Eqs. (11) and (12) lead to

R=(T+U),\displaystyle R=-(T+U), (40)
S=1u2u{2(1u)uU(3u)U+2uT},\displaystyle S=\frac{1-u}{2\ell u}\left\{2(1-u)uU^{\prime}-(3-u)U+2uT\right\}, (41)
(1u)2U′′2u(1u2)U+8m^24u2U=0,\displaystyle(1-u)^{2}U^{\prime\prime}-\frac{2}{u}(1-u^{2})U^{\prime}+\frac{8-\hat{m}^{2}}{4u^{2}}U=0, (42)
(1u)2T′′m^24u2T+2(1u2)uU+m^242u2U=0.\displaystyle(1-u)^{2}T^{\prime\prime}-\frac{\hat{m}^{2}}{4u^{2}}T+\frac{2(1-u^{2})}{u}U^{\prime}+\frac{\hat{m}^{2}-4}{2u^{2}}\,U=0. (43)

The general solution of Eq. (42) is expressed by the parameter p=1+m^2p=\sqrt{1+\hat{m}^{2}} (14) as

U=a1(1uu)3+p2+a2(1uu)3p2.\displaystyle U=a_{1}\left(\frac{1-u}{u}\right)^{-\frac{3+p}{2}}+a_{2}\left(\frac{1-u}{u}\right)^{-\frac{3-p}{2}}. (44)

Each mode function in (44) diverges at the extremal horizon, u=1u=1 under the inequality (17), 0<p<1/20<p<1/2, and therefore, a1=a2=0a_{1}=a_{2}=0, i. e., U=0U=0. Substituting U=0U=0 into Eqs. (40), (41), and (43), one obtains

T=b1u1p2(1u)1+p2+b2u1+p2(1u)1p2,\displaystyle T=b_{1}u^{\frac{1-p}{2}}(1-u)^{\frac{1+p}{2}}+b_{2}u^{\frac{1+p}{2}}(1-u)^{\frac{1-p}{2}},
R=T,S=(1u)T.\displaystyle R=-T,\qquad S=\frac{(1-u)T}{\ell}. (45)

In the range 0<p<1/20<p<1/2, the perturbation goes to zero towards the AdS boundary, u=0u=0, which is the same as the non-extremal case (III.1). It also converges to zero at the event horizon, u=1u=1. Nevertheless, we can show that the stress-energy tensor along null geodesics crossing the event horizon (33) diverges. Let us consider a null geodesic on the background BTZ black hole with tangent vector V¯μ\overline{V}^{\mu},

V¯μ=u(22r0(1u)2t+uu+u2r0(1u)2φ)μ,\displaystyle\overline{V}^{\mu}={\cal E}u\left(\frac{\ell^{2}}{2r_{0}(1-u)^{2}}\frac{\partial}{\partial t}+\sqrt{u}\frac{\partial}{\partial u}+\frac{\ell u}{2r_{0}(1-u)^{2}}\frac{\partial}{\partial\varphi}\right)^{\mu}, (46)

where {\cal E} is a positive constant. By Eqs. (III.2) and (46), we obtain

δ(TμνVμVν)=δTμνV¯μV¯ν3ϵ16πG4LHμνV¯μV¯ν\displaystyle\delta(\left\langle~{}T_{\mu\nu}~{}\right\rangle V^{\mu}V^{\nu})=\delta\left\langle~{}T_{\mu\nu}~{}\right\rangle\overline{V}^{\mu}\overline{V}^{\nu}\sim\frac{3\epsilon}{16\pi G_{4}L}H_{\mu\nu}\overline{V}^{\mu}\overline{V}^{\nu}
ϵ322uT(u)64πG4L(1u)2=ϵO((1u)3+p2)=ϵO((λ)3+p2),\displaystyle\sim-\epsilon\frac{3{\cal E}^{2}\ell^{2}uT(u)}{64\pi G_{4}L(1-u)^{2}}=\epsilon\,O\left((1-u)^{-\frac{3+p}{2}}\right)=\epsilon\,O\left((-\lambda)^{-\frac{3+p}{2}}\right), (47)

where λ\lambda is the affine parameter of the null geodesic with λ(1u)\lambda\sim-(1-u). This singular behavior is similar to the semiclassical result in the extremal Reissner-Nordström black hole [18].

IV 55-dimensional rotating AdS black holes

In this section we construct 55-dimensional semiclassical rotating AdS black hole solutions by solving Eqs. (11) and (12). In general, it is difficult to solve these equations in the background of rotating AdS black holes because the perturbations are partial differential equations. One notable exception is the odd-dimensional Myers-Perry (MP) AdS black hole [21] with equal angular momenta, which is known to be cohomogeneity-one solution of the vacuum Einstein equations with a negative cosmological constant. In this case, the equations become coupled ordinary differential equations, making it possible to analyze the singular behavior of the Cauchy horizon both analytically and numerically. Hereafter, we construct the solutions of Eqs. (11) and (12) for the 55-dimensional MP AdS black hole with equal angular momenta in both non-extremal and extremal cases.

IV.1 The non-extremal solution

We consider the following metric ansatz for perturbation of the 55-dimensional MP AdS black hole with equal angular momenta as

ds2=r^02ηuh(1+ϵT(u))dt2+24u2η(1+ϵU(u))du2+r^0224u(1+ϵR(u))(dθ2+sin2θdφ2)\displaystyle ds^{2}=-\frac{\hat{r}_{0}^{2}\eta}{uh}(1+\epsilon T(u))dt^{2}+\frac{\ell^{2}}{4u^{2}\eta}(1+\epsilon U(u))du^{2}+\frac{\hat{r}_{0}^{2}\ell^{2}}{4u}(1+\epsilon R(u))(d\theta^{2}+\sin^{2}\theta d\varphi^{2})
+r^022hu(1+ϵZ(u))(dψ+cosθ2dφ(Ω+ϵS(u))dt)2,\displaystyle+\frac{\hat{r}_{0}^{2}\ell^{2}h}{u}(1+\epsilon Z(u))\left(d\psi+\frac{\cos\theta}{2}d\varphi-(\Omega+\epsilon S(u))dt\right)^{2},
η=(1u)(1κ2u)(ur^02+u+κ2u+1),\displaystyle\eta=(1-u)(1-\kappa^{2}u)\left(\frac{u}{\hat{r}_{0}^{2}}+u+\kappa^{2}u+1\right),
h=1+κ2{1+(1+κ2)r^02}u2,Ω=M2ar^04h4u2,\displaystyle h=1+\kappa^{2}\{1+(1+\kappa^{2})\hat{r}_{0}^{2}\}u^{2},\qquad\Omega=\frac{M^{2}a}{\hat{r}_{0}^{4}h\ell^{4}}u^{2},
M=r^0(1+κ2)(1+r^02)(1+κ2r^02),\displaystyle M=\ell\hat{r}_{0}\sqrt{(1+\kappa^{2})(1+\hat{r}_{0}^{2})(1+\kappa^{2}\hat{r}_{0}^{2})},
a=κr^01+(1+κ2)r^02(1+κ2)(1+r^02)(1+κ2r^02),\displaystyle a=\ell\kappa\hat{r}_{0}\sqrt{\frac{1+(1+\kappa^{2})\hat{r}_{0}^{2}}{(1+\kappa^{2})(1+\hat{r}_{0}^{2})(1+\kappa^{2}\hat{r}_{0}^{2})}}, (48)

where each surface on t=t=const. and u=u=const. is a homogenously squashed S3S^{3}, which consists of S1S^{1} fibers over S2S^{2} (the agular coordinate ranges are 0ψ2π0\leq\psi\leq 2\pi, 0θπ0\leq\theta\leq\pi, and 0φ2π0\leq\varphi\leq 2\pi), and r^0(=r0/)\hat{r}_{0}~{}(=r_{0}/\ell) is the horizon radius on S2S^{2} normalized by \ell. Here, the radial coordinate uu is defined such that the event horizon and the Cauchy horizon are located at u=1u=1 and u=1/κ2u=1/\kappa^{2} in the range 0<κ10<\kappa\leq 1. Note that the ϵ=0\epsilon=0 background metric reduces to the standard form of the 55-dimensional MP AdS black hole [21] with equal angular momenta under the coordinate transformation u=r02/r2u=r_{0}^{2}/r^{2}.

Under the metric ansatz, Eqs.  (11) and (12) are decomposed into the three coupled second order ordinary differential equations for UU, TT, and ZZ

U′′+(2ηη5u)U+(h2h2+hr^02uηhηhη2h2Ω2η)T+[hr^02uη+1u2η(2η2uη)]Z\displaystyle U^{\prime\prime}+\left(\frac{2\eta^{\prime}}{\eta}-\frac{5}{u}\right)U^{\prime}+\left(\frac{h^{\prime 2}}{h^{2}}+\frac{h}{\hat{r}_{0}^{2}u\eta}-\frac{h^{\prime}\eta^{\prime}}{h\eta}-\frac{\ell^{2}h^{2}\Omega^{\prime 2}}{\eta}\right)T+\left[-\frac{h}{\hat{r}_{0}^{2}u\eta}+\frac{1}{u^{2}\eta}(2\eta-2-u\eta^{\prime})\right]Z
+[8u2h2h2m^24u2η+(hh4u)ηη+2h2Ω2η]U=0,\displaystyle+\left[\frac{8}{u^{2}}-\frac{h^{\prime 2}}{h^{2}}-\frac{\hat{m}^{2}}{4u^{2}\eta}+\left(\frac{h^{\prime}}{h}-\frac{4}{u}\right)\frac{\eta^{\prime}}{\eta}+\frac{\ell^{2}h^{2}\Omega^{\prime 2}}{\eta}\right]U=0, (49)
T′′+(hh1u)T+(4u+hh2ηη)UhhZ+[hr^02uη+1u2η(22η+uη)]Z\displaystyle T^{\prime\prime}+\left(\frac{h^{\prime}}{h}-\frac{1}{u}\right)T^{\prime}+\left(\frac{4}{u}+\frac{h^{\prime}}{h}-\frac{2\eta^{\prime}}{\eta}\right)U^{\prime}-\frac{h^{\prime}}{h}Z^{\prime}+\left[\frac{h}{\hat{r}_{0}^{2}u\eta}+\frac{1}{u^{2}\eta}(2-2\eta+u\eta^{\prime})\right]Z
+[h2h2+1η(m^24u2hr^02u+hηh+2h2Ω2)]T\displaystyle+\left[-\frac{h^{\prime 2}}{h^{2}}+\frac{1}{\eta}\left(-\frac{\hat{m}^{2}}{4u^{2}}-\frac{h}{\hat{r}_{0}^{2}u}+\frac{h^{\prime}\eta^{\prime}}{h}+\ell^{2}h^{2}\Omega^{\prime 2}\right)\right]T
+[10u2+2h2h2+1η(hr^02u2hηh+4+m^2+10uη2u222h2Ω2)]U=0,\displaystyle+\left[-\frac{10}{u^{2}}+\frac{2h^{\prime 2}}{h^{2}}+\frac{1}{\eta}\left(\frac{h}{\hat{r}_{0}^{2}u}-\frac{2h^{\prime}\eta^{\prime}}{h}+\frac{4+\hat{m}^{2}+10u\eta^{\prime}}{2u^{2}}-2\ell^{2}h^{2}\Omega^{\prime 2}\right)\right]U=0, (50)
Z′′+(ηη+hh1u)Z+(ηηhh2u)U+(ηηhh)T+(h2h2hηhη2h2Ω2η)T\displaystyle Z^{\prime\prime}+\left(\frac{\eta^{\prime}}{\eta}+\frac{h^{\prime}}{h}-\frac{1}{u}\right)Z^{\prime}+\left(\frac{\eta^{\prime}}{\eta}-\frac{h^{\prime}}{h}-\frac{2}{u}\right)U^{\prime}+\left(\frac{\eta^{\prime}}{\eta}-\frac{h^{\prime}}{h}\right)T^{\prime}+\left(\frac{h^{\prime 2}}{h^{2}}-\frac{h^{\prime}\eta^{\prime}}{h\eta}-\frac{\ell^{2}h^{2}\Omega^{\prime 2}}{\eta}\right)T
+[8u2h2h2+1η(2hr^02u+hηh8uη+8+m^22u2+2h2Ω2)]U\displaystyle+\left[\frac{8}{u^{2}}-\frac{h^{\prime 2}}{h^{2}}+\frac{1}{\eta}\left(-\frac{2h}{\hat{r}_{0}^{2}u}+\frac{h^{\prime}\eta^{\prime}}{h}-\frac{8u\eta^{\prime}+8+\hat{m}^{2}}{2u^{2}}+\ell^{2}h^{2}\Omega^{\prime 2}\right)\right]U
[3hr^02uη+1u2η(2+m^242η+uη)]Z=0,\displaystyle-\left[\frac{3h}{\hat{r}_{0}^{2}u\eta}+\frac{1}{u^{2}\eta}\left(2+\frac{\hat{m}^{2}}{4}-2\eta+u\eta^{\prime}\right)\right]Z=0, (51)

and SS and RR are determined by

S=5ηuη22uh2ΩU+hηηh22h3ΩT+ηh22h3ΩZηU2h2Ω,\displaystyle S=\frac{5\eta-u\eta^{\prime}}{2\ell^{2}uh^{2}\Omega^{\prime}}U+\frac{h\eta^{\prime}-\eta h^{\prime}}{2\ell^{2}h^{3}\Omega^{\prime}}T+\frac{\eta h^{\prime}}{2\ell^{2}h^{3}\Omega^{\prime}}Z-\frac{\eta U^{\prime}}{\ell^{2}h^{2}\Omega^{\prime}}, (52)
R=U+T+Z2,\displaystyle R=-\frac{U+T+Z}{2}, (53)

where a prime denotes differentiation with respect to uu. Let us consider the asymptotic solution near the AdS boundary, u=0u=0. Introducing the new variables, P=uUP=uU^{\prime}, Q=uZQ=uZ^{\prime}, R=uTR=uT^{\prime}, and 𝑽=(U,Z,T,P,Q,R)\bm{V}=(U,Z,T,P,Q,R), Eqs. (IV.1), (IV.1), and (IV.1) are transformed into the six coupled first order differential equations near u=0u=0,

ud𝑽duA𝑽,A:=(000100000010000001(8m^24)00600m^224m^2402208m^220m^24402).\displaystyle u\frac{d{\bm{V}}}{du}\simeq A{\bm{V}},\qquad A:=\begin{pmatrix}0&0&0&1&0&0\\ 0&0&0&0&1&0\\ 0&0&0&0&0&1\\ -\left(8-\frac{\hat{m}^{2}}{4}\right)&0&0&6&0&0\\ \frac{\hat{m}^{2}}{2}-4&\frac{\hat{m}^{2}}{4}&0&2&2&0\\ 8-\frac{\hat{m}^{2}}{2}&0&\frac{\hat{m}^{2}}{4}&-4&0&2\end{pmatrix}. (54)

Substituting 𝑽=uζ𝑲{\bm{V}}=u^{\zeta}{\bm{K}} into (54), we arrive at the eigenvalue equation |AζI|=0|A-\zeta I|=0, where II is the identity matrix and ζ\zeta represents the eigenvalue. The solutions are given by

ζ=3±p2,𝑲1±=(8p2±8p+12,2pp2±8p+12,±2p±6,4p±2,pp±2, 1),\displaystyle\zeta=3\pm\frac{p}{2},\qquad{\bm{K}}_{1\pm}=\left(-\frac{8}{p^{2}\pm 8p+12},\,\mp\frac{2p}{p^{2}\pm 8p+12},\,\pm\frac{2}{p\pm 6},\,\mp\frac{4}{p\pm 2},\,-\frac{p}{p\pm 2},\,1\right),
ζ=1±p2,𝑲2±=(0, 0, 1, 0, 0, 1±p2),𝑲3±=(0, 1, 0, 0, 1±p2, 0).\displaystyle\zeta=1\pm\frac{p}{2},\qquad{\bm{K}}_{2\pm}=\left(0,\,0,\,1,\,0,\,0,\,1\pm\frac{p}{2}\right),\qquad{\bm{K}}_{3\pm}=\left(0,\,1,\,0,\,0,\,1\pm\frac{p}{2},\,0\right). (55)

By the inequality (18), the range of pp is 1/2<p<3/21/2<p<3/2. Within this range, all the variables, UU, ZZ, TT, and RR converge to zero for all eigenvalues ζ\zeta considered above. Additionally, we can show that SS in Eq. (52) also converges to zero as follows.

Among all modes, 𝑲2{\bm{K}}_{2-} and 𝑲3{\bm{K}}_{3-} exhibit the slowest decay at infinity. For the 𝑲2{\bm{K}}_{2-} mode, using Eqs. (IV.1), (IV.1), and (IV.1), UU, ZZ, TT can be expanded as a series in uu as

Uu2p2(1+p)r^02,Tu1p2,Z=O(u2p2).\displaystyle U\simeq-\frac{u^{2-\frac{p}{2}}}{(1+p)\hat{r}_{0}^{2}},\qquad T\simeq u^{1-\frac{p}{2}},\qquad Z=O\left(u^{2-\frac{p}{2}}\right). (56)

Substituting Eq. (56) into Eq. (52) and using Ω=O(u2)\Omega=O(u^{2}) and h=O(u)h^{\prime}=O(u), we find that S=O(u1p/2)S=O(u^{1-p/2}), which converges to zero in the range 1/2<p<3/21/2<p<3/2. Similarly, for the 𝑲3{\bm{K}}_{3-} mode, using Eqs. (IV.1), (IV.1), and (IV.1), one finds

Z=O(u1p2),U=T=O(u3p2),\displaystyle Z=O\left(u^{1-\frac{p}{2}}\right),\qquad U=T=O\left(u^{3-\frac{p}{2}}\right), (57)

leading to the same asymptotic behavior, S=O(u1p/2)S=O(u^{1-p/2}).

One can also derive the asymptotic solutions of Eqs. (IV.1), (IV.1), and (IV.1) near the event horizon (Cauchy horizon) by transforming them into the coupled first order differential equations near u=1(u=1/κ2)u_{\ast}=1~{}(u_{\ast}=1/\kappa^{2}). Introducing variables PP, QQ, and RR as P=(uu)UP=(u_{\ast}-u)U^{\prime}, Q=(uu)ZQ=(u_{\ast}-u)Z^{\prime}, R=(uu)TR=(u_{\ast}-u)T^{\prime}, and setting 𝑽{\bm{V}} as 𝑽=(U,Z,T,P,Q,R):=(uu)ζ𝑲\bm{V}=(U,Z,T,P,Q,R):=(u_{\ast}-u)^{\zeta}{\bm{K}}, the equations become

(uu)d𝑽duA𝑽,A:=(000100000010000001000100000101000201).\displaystyle(u_{\ast}-u)\frac{d{\bm{V}}}{du}\simeq A{\bm{V}},\qquad A:=\begin{pmatrix}0&0&0&1&0&0\\ 0&0&0&0&1&0\\ 0&0&0&0&0&1\\ 0&0&0&1&0&0\\ 0&0&0&1&0&1\\ 0&0&0&-2&0&-1\end{pmatrix}. (58)

The solutions of the eigenvalue equation |A+ζI|=0|A+\zeta I|=0 are ζ=±1\zeta=\pm 1 and 0, where the eigenvectors of ζ=±1\zeta=\pm 1 are given by

ζ=1:𝑲=(1, 0,1, 1, 0,1),ζ=1:𝑲+=(0,1,1,0,1,1).\displaystyle\zeta=-1:\,\,\,{\bm{K}}_{-}=(1,\,0,\,-1,\,1,\,0,\,-1),\qquad\zeta=1:\,\,\,{\bm{K}}_{+}=(0,1,-1,0,-1,1). (59)

The zero mode solutions consist of two independent regular solutions and one logarithmic divergent mode solution, Zln(uu)Z\sim\ln(u_{\ast}-u), as derived from Eq. (IV.1).

By imposing regularity at the event horizon, and expanding UU, TT, and ZZ as

U=u0u1(1u)+u2(1u)2+,\displaystyle U=u_{0}-u_{1}(1-u)+u_{2}(1-u)^{2}+\cdots,
T=t0t1(1u)+t2(1u)2+,\displaystyle T=t_{0}-t_{1}(1-u)+t_{2}(1-u)^{2}+\cdots,
Z=z0z1(1u)+z2(1u)2+,\displaystyle Z=z_{0}-z_{1}(1-u)+z_{2}(1-u)^{2}+\cdots, (60)

Eqs. (IV.1)-(IV.1) lead to the condition

u0=t0,\displaystyle u_{0}=t_{0}, (61)

and the regular solutions are uniquely determined by the three free parameters, u0u_{0}, z0z_{0}, and z1z_{1}. Since we consider linear perturbations of the MP AdS black hole, hereafter, we set u0=1u_{0}=1 without loss of generality. Therefore, there are two independent regular solutions at the event horizon: a zero-mode solution parametrized by z0z_{0} and the ζ=1\zeta=1 convergent solution with the eigenvector 𝑲+{\bm{K}}_{+} in Eq. (59) parametrized by z1z_{1}.

Let us consider the behavior of the vacuum expectation value of the stress-energy tensor along an achronal null geodesic crossing the Cauchy horizon, u=1/κ2u_{\ast}=1/\kappa^{2}. By introducing new coordinates, (v,ψ^)(v,\hat{\psi})

v=th2r^0uη1+ϵU1+ϵT𝑑u,\displaystyle v=t-\int\frac{\ell\sqrt{h}}{2\hat{r}_{0}\sqrt{u}\,\eta}\sqrt{\frac{1+\epsilon U}{1+\epsilon T}}du,
ψ^=ψh2r^0uη(Ω+ϵS)1+ϵU1+ϵT𝑑u,\displaystyle\hat{\psi}=\psi-\int\frac{\ell\sqrt{h}}{2\hat{r}_{0}\sqrt{u}\,\eta}(\Omega+\epsilon S)\sqrt{\frac{1+\epsilon U}{1+\epsilon T}}du, (62)

the metric (IV.1) takes the form

ds2=r^02ηhu(1+ϵT)dv2uuh(1+ϵU)(1+ϵT)dudv\displaystyle ds^{2}=-\frac{\hat{r}_{0}^{2}\eta}{hu}(1+\epsilon T)dv^{2}-\frac{\ell}{u\sqrt{uh}}\sqrt{(1+\epsilon U)(1+\epsilon T)}\,dudv
+r^022hu(1+ϵZ)(dψ^+cosθ2dϕ(Ω+ϵS)dv)2\displaystyle+\frac{\hat{r}_{0}^{2}\ell^{2}h}{u}(1+\epsilon Z)\left(d\hat{\psi}+\frac{\cos\theta}{2}d\phi-(\Omega+\epsilon S)dv\right)^{2}
+r^0224u(1+ϵR)(dθ2+sin2θdϕ2).\displaystyle+\frac{\hat{r}_{0}^{2}\ell^{2}}{4u}(1+\epsilon R)(d\theta^{2}+\sin^{2}\theta d\phi^{2}). (63)
Refer to caption
Figure 1: The functions UU (blue, solid), TT (green, dashes), ZZ (red, dotted), and SS (brown, dotdashed) are plotted in the range u(0,1)u\in(0,1) for the parameters, κ=0.5\kappa=0.5, r^0=3\hat{r}_{0}=3, m^2=3\hat{m}^{2}=-3, and (z0,z1)=(0,1)(z_{0},\,z_{1})=(0,1).

Now, u\partial_{u} is the tangent null vector along the achronal null geodesic and the affine parametrized tangent vector VμV^{\mu} is given by

𝑽=𝑽¯+δ𝑽=u32h(1+ϵU)(1+ϵT)u\displaystyle{\bm{V}}=\overline{{\bm{V}}}+\delta{\bm{V}}=\frac{{\cal E}u^{\frac{3}{2}}\sqrt{h}}{\ell\sqrt{(1+\epsilon U)(1+\epsilon T)}}\frac{\partial}{\partial u}
=uuh[h2r^0uηt+u+h2r^0uηΩψ]+O(ϵ),\displaystyle=\frac{u\sqrt{uh}\,{\cal E}}{\ell}\left[\frac{\ell\sqrt{h}}{2\hat{r}_{0}\sqrt{u}\eta}\frac{\partial}{\partial t}+\frac{\partial}{\partial u}+\frac{\ell\sqrt{h}}{2\hat{r}_{0}\sqrt{u}\eta}\Omega\frac{\partial}{\partial\psi}\right]+O(\epsilon), (64)

where {\cal E} is a positive constant. The solution of (58) includes the growing mode with the eigenvalue ζ=1\zeta=-1 near the Cauchy horizon. The eigenvector 𝑲{\bm{K}}_{-} in Eq. (59) causes the strong divergence of the stress-energy tensor,

δ(TμνVμVν)ϵHμνV¯μV¯ν=ϵ2uh(UT)4η=ϵC1λ2,\displaystyle\delta(\left\langle~{}T_{\mu\nu}~{}\right\rangle V^{\mu}V^{\nu})\sim\epsilon H_{\mu\nu}\overline{V}^{\mu}\overline{V}^{\nu}=\epsilon\frac{{\cal E}^{2}uh(U-T)}{4\eta}=\epsilon\frac{C_{1}}{\lambda^{2}}, (65)

where C1C_{1} approaches a constant near the Cauchy horizon, and λ(1/κ2u)\lambda~{}(\sim 1/\kappa^{2}-u) is the affine parameter, which is defined to be zero at the Cauchy horizon.

Fig. 1 shows the behavior of the functions UU (blue, solid), TT (green, dashed), ZZ (red, dotted), and SS (brown, dotdashed) in the range u(0,1)u\in(0,1) for the parameter values, κ=0.5\kappa=0.5, r^0=3\hat{r}_{0}=3, m^2=3\hat{m}^{2}=-3, and (z0,z1)=(0,1)(z_{0},\,z_{1})=(0,1): The function UU decays as Uu3/2U\sim u^{3/2}, while the other functions TT, ZZ, and SS decays as u1/2\sim u^{1/2}, as shown in Eqs. (56) and (57).

Inside the event horizon, the perturbation generally grows toward the Cauchy horizon for most of the parameter space (z0,z1)(z_{0},z_{1}). As indicated in the linear analysis (58) and (59), UTU-T and ZZ diverge near the Cauchy horizon as

UTC11/κ2u,ZC2ln(1/κ2u),\displaystyle U-T\sim\frac{C_{1}}{1/\kappa^{2}-u},\qquad Z\sim C_{2}\ln(1/\kappa^{2}-u), (66)

as shown in Figs. 3 and 3 for the same parameters with Fig. 1. The former divergence leads to a strong divergence of the null energy (65). As demonstrated in a quantum field theory on the Reissner-Nordström de Sitter background [9], one expects that the coefficient C1C_{1} always takes a nonzero value in higher-dimensional spacetimes, except in the three-dimensional case analyzed in the previous section.

In contrast, for the five-dimensional semiclassical solution, there exists a class of parameters for which the coefficient C1C_{1} vanishes. Fig. 5 shows the curve z1=z1(z0)z_{1}=z_{1}(z_{0}) in the parameter space (z0,z1)(z_{0},z_{1}) where C1=0C_{1}=0, for the same background parameters as Fig. 1. One might then consider that an observer traveling along the achronal null geodesic would not experience any divergence in the null energy (65), allowing the null geodesic to be extended beyond the Cauchy horizon. However, in our case, backreaction effects from quantum corrections are incorporated through the semiclassical equations (8). As a result, a strong curvature singularity still forms at the Cauchy horizon, even along this curve, due to the presence of the logarithmic mode (66) near the Cauchy horizon, as discussed below.

Refer to caption
Figure 2: The plot of UTU-T for the parameters κ=0.5\kappa=0.5, r^0=3\hat{r}_{0}=3, m^2=3\hat{m}^{2}=-3, and (z0,z1)=(0,1)(z_{0},\,z_{1})=(0,1). The best-fit curve (dashed curve) is UT61.5(1/κ2u)124.6U-T\simeq 61.5(1/\kappa^{2}-u)^{-1}-24.6.
Refer to caption
Figure 3: The plot of ZZ for the parameters κ=0.5\kappa=0.5, r^0=3\hat{r}_{0}=3, m^2=3\hat{m}^{2}=-3, and (z0,z1)=(0,1)(z_{0},\,z_{1})=(0,1). The best-fit curve (dashed curve) corresponds to Z0.15ln(1/κ2u)+6.76Z\simeq-0.15\ln(1/\kappa^{2}-u)+6.76.
Refer to caption
Figure 4: The curve z1=z1(z0)z_{1}=z_{1}(z_{0}), along which the coefficient C1C_{1} vanishes, is plotted for the parameters κ=0.5\kappa=0.5 and r^0=3\hat{r}_{0}=3.
Refer to caption
Figure 5: The coefficient C2C_{2} is plotted along the curve z1=z1(z0)z_{1}=z_{1}(z_{0}) for the same parameters κ=0.5\kappa=0.5 and r^0=3\hat{r}_{0}=3.

Let us calculate a curvature component in the frame that is parallelly propagated along the null geodesic with tangent vector (IV.1). A unit spacelike vector 𝑬θ{\bm{E}}_{\theta} defined by

𝑬θ=2ur^01+ϵRθ\displaystyle{\bm{E}}_{\theta}=\frac{2\sqrt{u}}{\hat{r}_{0}\ell\sqrt{1+\epsilon\,R}}\frac{\partial}{\partial\theta} (67)

is invariant along the null geodesic orbit, i. e. , VνDνEθμ=0V^{\nu}D_{\nu}E_{\theta}^{\mu}=0, and orthogonal to 𝑽{\bm{V}}. Up to O(ϵ\epsilon), we find that the following Riemann component strongly diverges as

δ(μανβEθμVαEθνVβ)ϵC22(1+r^02)(1+κ2)42κ8(1κ2u)21λ2\displaystyle\delta(\mathcal{R}_{\mu\alpha\nu\beta}{E}_{\theta}^{\mu}V^{\alpha}{E}_{\theta}^{\nu}V^{\beta})\sim\epsilon\frac{C_{2}{\cal E}^{2}(1+\hat{r}_{0}^{2})(1+\kappa^{2})}{4\ell^{2}\kappa^{8}\left(\frac{1}{\kappa^{2}}-u\right)^{2}}\sim\frac{1}{\lambda^{2}} (68)

in the presence of the logarithmic mode ZC2ln(1/κ2u)Z\sim C_{2}\ln(1/\kappa^{2}-u). This is a parallelly propagated (p. p.) curvature singularity [26] along the null geodesic transverse to the Cauchy horizon. The shear tensor σμν\sigma_{\mu\nu} of the null geodesic congruence, contracted with the unit spacelike vector 𝑬θ{\bm{E}}_{\theta}, obeys the following equation along the null geodesic,

d(σμνEθμEθν)dλδ(CμανβEθμVαEθνVβ)δ(μανβEθμVαEθνVβ)1λ2\displaystyle\frac{d(\sigma_{\mu\nu}E_{\theta}^{\mu}E_{\theta}^{\nu})}{d\lambda}\sim\delta(C_{\mu\alpha\nu\beta}{E}_{\theta}^{\mu}V^{\alpha}{E}_{\theta}^{\nu}V^{\beta})\sim\delta(\mathcal{R}_{\mu\alpha\nu\beta}{E}_{\theta}^{\mu}V^{\alpha}{E}_{\theta}^{\nu}V^{\beta})\sim\frac{1}{\lambda^{2}} (69)

along the curve z1=z1(z0)z_{1}=z_{1}(z_{0}) with C1=0C_{1}=0. Here, we have used the fact that μνVμVνTμνVμVν\mathcal{R}_{\mu\nu}V^{\mu}V^{\nu}\sim\left\langle~{}T_{\mu\nu}~{}\right\rangle V^{\mu}V^{\nu} is subdominant compared to the divergence of the Riemann tensor component by Eqs. (8). Thus, the shear tensor itself diverges as 1/λ1/\lambda. Since the integral of the shear tensor also diverges, the p. p. singularity corresponds to a strong curvature singularity, characterized by an infinite stretching or crushing of the null geodesic congruence.

Fig. 5 illustrates that the coefficient C2C_{2} is generically non-zero along the curve z1=z1(z0)z_{1}=z_{1}(z_{0}), except at a particular point z0=z06.81z_{0}=z_{0}^{\ast}\simeq 6.81 for (κ,r^0)=(0.5, 3)(\kappa,\hat{r}_{0})=(0.5,\,3). Contrary to the naive expectation that quantum fluctuations would universally destroy the Cauchy horizon for any semiclassical solutions, we find that a C0C^{0}-regular Cauchy horizon exists at a specific point where the perturbation remains finite. This occurs due to the fact that the solutions of (58) include only two growing modes, which can be eliminated by a suitable adjustment of the two free parameters (z0,z1)(z_{0},z_{1}) on the regular event horizon.

IV.2 The extremal solution

In this subsection, we investigate semiclassical κ=1\kappa=1 extremal solutions of Eqs. (IV.1), (IV.1), and (IV.1). Near the extremal horizon, u=1u=1, defining PP, QQ, RR, and 𝑽{\bm{V}} as P=(1u)UP=(1-u)U^{\prime}, Q=(1u)ZQ=(1-u)Z^{\prime}, R=(1u)TR=(1-u)T^{\prime}, and 𝑽=(U,Z,T,P,Q,R){\bm{V}}=(U,\,Z,\,T,\,P,\,Q,\,R), respectively, 𝑽{\bm{V}} obeys

(1u)d𝑽duA𝑽,A:=(000100000010000001r^02(m^216)84+12r^022+4r^021+3r^022r^021+3r^02300r^02(m^2+8)+42+6r^0224+(32+m^2)r^024+12r^022+4r^021+3r^02212r^02(8m^2)+42+6r^022+2r^021+3r^02(m^28)r^024+12r^02401).\displaystyle(1-u)\frac{d{\bm{V}}}{du}\simeq A{\bm{V}},\qquad A:=\begin{pmatrix}0&0&0&1&0&0\\ 0&0&0&0&1&0\\ 0&0&0&0&0&1\\ \frac{\hat{r}_{0}^{2}(\hat{m}^{2}-16)-8}{4+12\hat{r}_{0}^{2}}&\frac{2+4\hat{r}_{0}^{2}}{1+3\hat{r}_{0}^{2}}&\frac{2\hat{r}_{0}^{2}}{1+3\hat{r}_{0}^{2}}&3&0&0\\ \frac{\hat{r}_{0}^{2}(\hat{m}^{2}+8)+4}{2+6\hat{r}_{0}^{2}}&\frac{24+(32+\hat{m}^{2})\hat{r}_{0}^{2}}{4+12\hat{r}_{0}^{2}}&\frac{2+4\hat{r}_{0}^{2}}{1+3\hat{r}_{0}^{2}}&2&1&2\\ \frac{\hat{r}_{0}^{2}(8-\hat{m}^{2})+4}{2+6\hat{r}_{0}^{2}}&-2+\frac{2\hat{r}_{0}^{2}}{1+3\hat{r}_{0}^{2}}&\frac{(\hat{m}^{2}-8)\hat{r}_{0}^{2}}{4+12\hat{r}_{0}^{2}}&-4&0&-1\end{pmatrix}. (70)

Substituting 𝑽=(1u)ζ𝑲{\bm{V}}=(1-u)^{\zeta}{\bm{K}} into Eq. (70), we arrive at the eigenvalue equation |A+ζI|=0|A+\zeta I|=0, where ζ\zeta represents the eigenvalue. The solutions to this equation are determined by

k(z)=0,\displaystyle k(z)=0,
k(z):=64(1+3r^02)3z316(1+3r^02)2{35r^02(653ρ)}z2\displaystyle k(z):=64(1+3\hat{r}_{0}^{2})^{3}z^{3}-16(1+3\hat{r}_{0}^{2})^{2}\{35-\hat{r}_{0}^{2}(65-3\rho)\}z^{2}
+4(1+3r^02){259+r^02(115438ρ)+3r^04(377(22ρ)ρ)}z\displaystyle+4(1+3\hat{r}_{0}^{2})\{259+\hat{r}_{0}^{2}(1154-38\rho)+3\hat{r}_{0}^{4}(377-(22-\rho)\rho)\}z
{15+r^02(21+ρ)}{15+r^02(18+r^02(27ρ))(5+ρ)},\displaystyle-\{15+\hat{r}_{0}^{2}(21+\rho)\}\{15+\hat{r}_{0}^{2}(18+\hat{r}_{0}^{2}(27-\rho))(5+\rho)\}, (71)

where zz and ρ\rho are defined as z:=(ζ+1/2)2z:=(\zeta+1/2)^{2} and ρ=m^2>0\rho=-\hat{m}^{2}>0, respectively. Here, k(z)k(z) is a cubic polynomial whose coefficients depend on the parameters r^0\hat{r}_{0} and ρ\rho.

We demonstrate that there exist at least two independent, convergent solutions with positive ζ\zeta, and every solution of k(z)=0k(z)=0 satisfies ζ<2\zeta<2 near the extremal horizon. This ensures that the solutions remain bounded near the extremal horizon, consistent with the expected physical behavior in this regime. The result is summarized in the following proposition:

Proposition
There exist at least two independent, convergent solutions with positive ζ\zeta to the coupled first-order differential equations (70) near the extremal horizon. Furthermore, all solutions satisfy ζ<2\zeta<2 within the mass range specified by (18).
Proof)
By Lemma 1 in Appendix B, the function k(z)k(z) has a maximum value kmaxk_{\text{max}} at z=zz=z_{-} and a minimum value kmink_{\text{min}} at z=z+z=z_{+}, where z±z_{\pm} satisfy 1/4<z<z+<25/41/4<z_{-}<z_{+}<25/4 in the range (18). From Lemma 2 in Appendix B, we have kmax>0k_{\text{max}}>0 and kmin<0k_{\text{min}}<0 within this range. Furthermore, it is straightforward to verify that k(25/4)>0k(25/4)>0 in the same range (18). By the intermediate value theorem, k(z)k(z) has at least two distinct roots, corresponding to convergent solutions with positive ζ\zeta, and all solutions satisfy ζ<2\zeta<2.   \Box

For the mode solutions, the asymptotic AdS boundary condition S(0)=0S(0)=0 is satisfied, as shown in the previous section. Consequently, the above proposition implies the existence of at least two extremal solutions that converge to the extremal Myers-Perry (MP) AdS black hole at the extremal horizon.

Although the extremal hairy black hole solution approaches the extremal MP AdS black hole at the horizon, the stress-energy tensor Tμν\left\langle~{}T_{\mu\nu}~{}\right\rangle near the event horizon diverges along the null geodesic with the tangent vector given in Eq.(IV.1), similar to the 33-dimensional case discussed in Sec.III.2. Contracting Tμν\left\langle~{}T_{\mu\nu}~{}\right\rangle with the null vector (IV.1) and using the eigenvector 𝑲{\bm{K}} that satisfies |A+ζI|=0|A+\zeta I|=0, we find

δ(TμνVμVν)ϵγ5w516πG522uh(UT)4η(1u)ζ2.\displaystyle\delta(\left\langle~{}T_{\mu\nu}~{}\right\rangle V^{\mu}V^{\nu})\sim\frac{\epsilon\gamma_{5}w_{5}}{16\pi G_{5}\ell^{2}}\frac{{\cal E}^{2}uh(U-T)}{4\eta}\sim(1-u)^{\zeta-2}. (72)

Thus, all semiclassical extremal solutions must exhibit a singularity at the horizon.

V Summary and discussions

We investigated odd-dimensional holographic semiclassical rotating AdS black holes within the framework of the AdS/CFT duality, considering a strongly coupled quantum field. Within the regime of linear perturbations, the backreaction of the quantum field on the geometry is incorporated via the semiclassical Eqs. (8) on the conformal AdS boundary. The separation constant m2m^{2} in Eqs. (11) is constrained within the ranges given in Eqs. (17) and (18), as shown in Refs. [20, 24]. Thus, the only remaining task for constructing the semiclassical solutions is to solve the massive Kaluza-Klein graviton equations (11) with a negative mass m2m^{2}.

In the 33-dimensional case, we analytically derived both non-extremal and extremal semiclassical black hole solutions. One key feature is that the vacuum expectation value of the stress-energy tensor Tμν\left\langle~{}T_{\mu\nu}~{}\right\rangle always diverges near the Cauchy horizon, independent of the parameters of the background BTZ solution. This divergence contrasts with the regular behavior of Tμν\left\langle~{}T_{\mu\nu}~{}\right\rangle on the Cauchy horizon for a quantum scalar field in a fixed near-extremal BTZ black hole background [8]. As pointed out in [8], this finite behavior arises due to a special property of the background BTZ geometry. On the other hand, in our semiclassical model, the vacuum expectation value of the stress-energy tensor is computed on the modified BTZ geometry via the semiclassical equations. Since this special property is absent in the modified geometry, we infer that the vacuum expectation value generically diverges on the Cauchy horizon in the semiclassical theory. Another key feature in the non-extremal case is that the degree of divergence depends on the type of perturbation, corresponding to even- or odd-parity modes in the static frame of the background BTZ solution. For even-parity perturbations, the null energy along a null geodesic with tangent vector VμV^{\mu} exhibits a strong divergence as TμνVμVν1/λ2\left\langle~{}T_{\mu\nu}~{}\right\rangle V^{\mu}V^{\nu}\sim 1/\lambda^{2} with respect to the affine parameter λ\lambda (λ=0\lambda=0 at the Cauchy horizon). Conversely, for odd-parity perturbations, the divergence is milder, scaling as TμνVμVν1/λ\left\langle~{}T_{\mu\nu}~{}\right\rangle V^{\mu}V^{\nu}\sim 1/\lambda. These findings indicate that a C0C^{0} extension of the metric beyond the Cauchy horizon is feasible in the absence of even-parity perturbations, although generic perturbation supports the SCC conjecture.

In the 5-dimensional case, semiclassical non-extremal rotating AdS black holes are characterized by two parameters at the regular event horizon. Although TμνVμVν\left\langle~{}T_{\mu\nu}~{}\right\rangle V^{\mu}V^{\nu} typically diverges strongly as C/λ2\sim C/\lambda^{2} near the Cauchy horizon for most parameter choices, there exists a class of semiclassical solutions in which the coefficient CC vanishes, independent of the background MP black hole parameters. Even in such cases, we demonstrated that a parallely propagated Riemann tensor component along the null geodesic still generically diverges strongly as C/λ2\sim C/\lambda^{2}. This appears to be a peculiar phenomenon observed only in the semiclassical model, which incorporates the backreaction on the geometry and it supports the SCC conjecture.

In contrast, there exist semiclassical solutions with a C0C^{0}-regular Cauchy horizon where perturbations remain finite. This is somewhat surprising, as one might expect quantum effects to universally induce strong divergences through backreaction, independent of the background geometry. Unlike in the 3-dimensional case, only two of the six independent mode solutions contribute to the singular behavior at the Cauchy horizon. By carefully tuning the two free parameters at the event horizon, a semiclassical solution with a C0C^{0}-regular Cauchy horizon can be constructed. In scenarios with a fixed curved background, such strong divergence has been observed in free quantum fields on fixed 4-dimensional Reissner-Nordström-dS black hole backgrounds [9] and in strongly coupled field theories on fixed 5-dimensional Myers-Perry AdS black hole spacetimes [27]. In both cases, the null energy density exhibits a strong divergence near the Cauchy horizon in the absence of backreaction effects. However, when backreaction is incorporated, it has the potential to weaken the divergence, suggesting that quantum effects could moderate the singularity under semiclassical perturbations.

For the extremal case, we examine semiclassical solutions under linear perturbations. Notably, we obtained analytic 3-dimensional semiclassical black hole solutions that converge to the extremal BTZ solution at the event horizon. Since the decay rate is very slow, the null energy TμνVμVν\left\langle~{}T_{\mu\nu}~{}\right\rangle V^{\mu}V^{\nu} diverges along a null geodesic with tangent vector VμV^{\mu} transverse to the event horizon. In the 5-dimensional hairy AdS black hole case, there exist at least two mode solutions that vanish at the extremal horizon. We proved that the decay rate for the mode solutions is very slow at the extremal horizon, and consequently TμνVμVν\left\langle~{}T_{\mu\nu}~{}\right\rangle V^{\mu}V^{\nu} always diverges along a null geodesic crossing the event horizon for any extremal semiclassical solution. These results indicate that such singularities at the event horizon are a generic feature of semiclassical rotating extremal solutions, similar to the findings for classical perturbations [13, 14, 15, 16]. Whether this generic singular behavior extends to semiclassical charged extremal black hole solutions, as suggested in the s-wave approximation [18], remains an open question.

Finally, the thermodynamic stability of these semiclassical rotating black hole solutions warrants further investigation. In the non-rotating case, 3- and 5-dimensional semiclassical hairy black hole solutions are thermodynamically more stable than the BTZ or hyperbolic AdS black hole solutions with a vanishing stress-energy tensor Tμν=0\left\langle~{}T_{\mu\nu}~{}\right\rangle=0 [25, 24]. Furthermore, extending semiclassical solutions beyond the linear perturbation regime will be crucial for a deeper understanding of the backreaction effects. A detailed analysis of the rotating case will be pursued in future work.

Acknowledgments

We wish to thank Takashi Okamura and Akihiro Ishibashi for useful discussions. This work is supported in part by JSPS KAKENHI Grant No. 20K03975 (K.M.) and also supported by MEXT KAKENHI Grant-in-Aid for Transformative Research Areas A Extreme Universe No.21H05186 (K.M.).

Appendix A d=5d=5 Semiclassical equations

Under the transverse and traceless condition (12), the first order perturbation of the following curvature tensors and the derivatives are given by

δμν=ϵ(12𝒟¯2Hμν+42Hμν+¯μανβHαβ)=ϵ(m22+42)Hμν,\displaystyle\delta\mathcal{R}_{\mu\nu}=-\epsilon\left(\frac{1}{2}\overline{\mathcal{D}}^{2}H_{\mu\nu}+\frac{4}{\ell^{2}}H_{\mu\nu}+\overline{\mathcal{R}}_{\mu\alpha\nu\beta}H^{\alpha\beta}\right)=-\epsilon\left(\frac{m^{2}}{2}+\frac{4}{\ell^{2}}\right)H_{\mu\nu},
δ(𝒟2μν)=ϵ2m2D2¯Hμν,δ(μν)=20ϵ2(42+m22)Hμν,\displaystyle\delta(\mathcal{D}^{2}\mathcal{R}_{\mu\nu})=-\frac{\epsilon}{2}m^{2}\overline{D^{2}}H_{\mu\nu},\qquad\delta(\mathcal{R}\mathcal{R}_{\mu\nu})=\frac{20\epsilon}{\ell^{2}}\left(\frac{4}{\ell^{2}}+\frac{m^{2}}{2}\right)H_{\mu\nu},
δ(αμβναβ)=4ϵ2(m22+42)Hμνϵ2m2¯μανβHαβ,\displaystyle\delta(\mathcal{R}_{\alpha\mu\beta\nu}\mathcal{R}^{\alpha\beta})=\frac{4\epsilon}{\ell^{2}}\left(\frac{m^{2}}{2}+\frac{4}{\ell^{2}}\right)H_{\mu\nu}-\frac{\epsilon}{2}m^{2}\overline{\mathcal{R}}_{\mu\alpha\nu\beta}H^{\alpha\beta},
δ=δ(𝒟μ𝒟ν)=δ(μνμν)=0,\displaystyle\delta\mathcal{R}=\delta(\mathcal{D}_{\mu}\mathcal{D}_{\nu}\mathcal{R})=\delta(\mathcal{R}_{\mu\nu}\mathcal{R}^{\mu\nu})=0, (73)

where we used Eq. (11) in the second equality in the first line. Substituting this into the perturbation of τμν(5)\tau_{\mu\nu}^{(5)} in (II), one obtains the same tensor [24]

δτμν(5)=ϵL472πG6sinz(m42+9m242924)Hμν.\displaystyle\delta\tau^{(5)}_{\mu\nu}=-\frac{\epsilon L^{4}}{72\pi\ell G_{6}\sin\frac{z}{\ell}}\left(\frac{m^{4}}{2}+\frac{9m^{2}}{4\ell^{2}}-\frac{9}{2\ell^{4}}\right)H_{\mu\nu}. (74)

Inserting Eq. (74) into (II) and expanding (13) near z=0z=0, one obtains the perturbed stress-energy tensor [24]

δTμν=ϵ16πG5γ5w5(m^2)2Hμν,\displaystyle\delta\left\langle~{}T_{\mu\nu}~{}\right\rangle=-\frac{\epsilon}{16\pi G_{5}}\frac{\gamma_{5}w_{5}(\hat{m}^{2})}{\ell^{2}}H_{\mu\nu},
w5(m^2)=π9m^2(3+m^2)4+m^2tan(π4+m^2).\displaystyle w_{5}(\hat{m}^{2})=\frac{\pi}{9}\frac{\hat{m}^{2}(3+\hat{m}^{2})\sqrt{4+\hat{m}^{2}}}{\tan(\pi\sqrt{4+\hat{m}^{2}})}. (75)

By Eqs. (A), the perturbation of Eqs. (8) becomes

ϵm22Hμν=ϵ16πG5γ5w5(m^2)2Hμν,\displaystyle-\epsilon\frac{m^{2}}{2}H_{\mu\nu}=-\frac{\epsilon}{16\pi G_{5}}\frac{\gamma_{5}w_{5}(\hat{m}^{2})}{\ell^{2}}H_{\mu\nu}, (76)

which is equivalent to the algebraic equation [24]

γ5=m^2w5(m^2).\displaystyle\gamma_{5}=\frac{\hat{m}^{2}}{w_{5}(\hat{m}^{2})}. (77)

Appendix B Lemmas

Lemma 1
The function k(z)k(z) in (IV.2) takes the maximum and the minimum values at z=zz=z_{-} and z=z+z=z_{+}, respectively. z±z_{\pm} satisfies the inequality,

14<z<z+<254.\displaystyle\frac{1}{4}<z_{-}<z_{+}<\frac{25}{4}. (78)

proof)
By k(z)=0k^{\prime}(z)=0, z±z_{\pm} is given by

z±=136[653ρ+40+3ρ1+3r^02±1228+r^02(686ρ)+r^04(5212ρ)1+3r^02]=:p1±12q136(1+3r^02),\displaystyle z_{\pm}=\frac{1}{36}\left[65-3\rho+\frac{40+3\rho}{1+3\hat{r}_{0}^{2}}\pm\frac{12\sqrt{28+\hat{r}_{0}^{2}(68-6\rho)+\hat{r}_{0}^{4}(52-12\rho)}}{1+3\hat{r}_{0}^{2}}\right]=:\frac{p_{1}\pm 12\sqrt{q_{1}}}{36(1+3\hat{r}_{0}^{2})}, (79)

where q1>0q_{1}>0 in the range (18) for an arbitrary value of r^0\hat{r}_{0}. By

z14=32+r^02(563ρ)4q112(1+3r^02)=:p24q112(1+3r^02),\displaystyle z_{-}-\frac{1}{4}=\frac{32+\hat{r}_{0}^{2}(56-3\rho)-4\sqrt{q_{1}}}{12(1+3\hat{r}_{0}^{2})}=:\frac{p_{2}-4\sqrt{q_{1}}}{12(1+3\hat{r}_{0}^{2})}, (80)

and p2>0p_{2}>0 in the range (18), the inequality

p2216q1=576+96r^02(26ρ)+9r^04{(ρ8)2+192}>0\displaystyle p_{2}^{2}-16q_{1}=576+96\hat{r}_{0}^{2}(26-\rho)+9\hat{r}_{0}^{4}\left\{(\rho-8)^{2}+192\right\}>0 (81)

shows z1>1/4z_{1}>1/4. Similarly, one finds

254z+=40+r^02(160+3ρ)4q112(1+3r^02)=p34q112(1+3r^02),\displaystyle\frac{25}{4}-z_{+}=\frac{40+\hat{r}_{0}^{2}(160+3\rho)-4\sqrt{q_{1}}}{12(1+3\hat{r}_{0}^{2})}=\frac{p_{3}-4\sqrt{q_{1}}}{12(1+3\hat{r}_{0}^{2})}, (82)

where p3>0p_{3}>0. The inequality (81) combined with

p3216q1=1152+48r^02(244+7ρ)+9r^04{2752+ρ(128+ρ)}>0\displaystyle p_{3}^{2}-16q_{1}=1152+48\hat{r}_{0}^{2}(244+7\rho)+9\hat{r}_{0}^{4}\{2752+\rho(128+\rho)\}>0 (83)

proves the above lemma 1.

Lemma 2
The maximum and minimum values of k(z)k(z) are positive and negative, respectively.
proof)
The maximum and minimum values of k(z)k(z) are given by

k(z±)=25627[2(1+r^02){2r^04(359ρ)+r^02(59ρ)20}{r^04(266ρ)+r^02(343ρ)+14}q1].\displaystyle k(z_{\pm})=\frac{256}{27}\left[2(1+\hat{r}_{0}^{2})\{2\hat{r}_{0}^{4}(35-9\rho)+\hat{r}_{0}^{2}(5-9\rho)-20\}\mp\{\hat{r}_{0}^{4}(26-6\rho)+\hat{r}_{0}^{2}(34-3\rho)+14\}\sqrt{q_{1}}\right]. (84)

The minimum value k(z+)k(z_{+}) is maximized at

ρ0=3+2r^0211+2r^02>3\displaystyle\rho_{0}=3+\frac{2}{\hat{r}_{0}^{2}}-\frac{1}{1+2\hat{r}_{0}^{2}}>3 (85)

with respect to ρ\rho. When ρ015/4\rho_{0}\leq 15/4,

k(z+)|ρ=ρ0=512(1+r^02)(2+3r^02)<0.\displaystyle k(z_{+})\Bigr{|_{\rho=\rho_{0}}}=-512(1+\hat{r}_{0}^{2})(2+3\hat{r}_{0}^{2})<0. (86)

When ρ0>15/4\rho_{0}>15/4,

k(z+)|ρ=154\displaystyle k(z_{+})\Bigr{|_{\rho=\frac{15}{4}}} =3227[20(16+39r^02+21r^042r^06)+714(8+13r^02+2r^04)32]\displaystyle=-\frac{32}{27}\left[20(16+39\hat{r}_{0}^{2}+21\hat{r}_{0}^{4}-2\hat{r}_{0}^{6})+7\sqrt{14}(8+13\hat{r}_{0}^{2}+2\hat{r}_{0}^{4})^{\frac{3}{2}}\right]
<3227[20(16+39r^02+21r^04)+(142840)r^06]<0,\displaystyle<-\frac{32}{27}\left[20(16+39\hat{r}_{0}^{2}+21\hat{r}_{0}^{4})+(14\sqrt{28}-40)\hat{r}_{0}^{6}\right]<0, (87)

showing k(z+)<0k(z_{+})<0 in the range (18). Here, we used 714(8+13r^02+2r^04)32>1428r^067\sqrt{14}(8+13\hat{r}_{0}^{2}+2\hat{r}_{0}^{4})^{\frac{3}{2}}>14\sqrt{28}\,\hat{r}_{0}^{6} in the first inequality in the second line.

On the other hand, the maximum value k(z)k(z_{-}) is a monotonously decreasing function of ρ\rho, and hence,

k(z)|ρ=154=64027(1+r^02)(16+23r^022r^04)+2242714(8+13r^02+2r^04)8+13r^02+2r^04\displaystyle k(z_{-})\Bigr{|_{\rho=\frac{15}{4}}}=-\frac{640}{27}(1+\hat{r}_{0}^{2})(16+23\hat{r}_{0}^{2}-2\hat{r}_{0}^{4})+\frac{224}{27}\sqrt{14}(8+13\hat{r}_{0}^{2}+2\hat{r}_{0}^{4})\sqrt{8+13\hat{r}_{0}^{2}+2\hat{r}_{0}^{4}}
>64027(1+r^02)(16+23r^022r^04)+2242728(8+13r^02+2r^04)(2+r^02)\displaystyle>-\frac{640}{27}(1+\hat{r}_{0}^{2})(16+23\hat{r}_{0}^{2}-2\hat{r}_{0}^{4})+\frac{224}{27}\sqrt{28}(8+13\hat{r}_{0}^{2}+2\hat{r}_{0}^{4})(2+\hat{r}_{0}^{2})
=6427[16(7710)+(2387390)r^02+(1197210)r^04+(20+147)r^06]>0,\displaystyle=\frac{64}{27}\left[16(7\sqrt{7}-10)+(238\sqrt{7}-390)\hat{r}_{0}^{2}+(119\sqrt{7}-210)\hat{r}_{0}^{4}+(20+14\sqrt{7})\hat{r}_{0}^{6}\right]>0,\qquad\Box (88)

where we used 8+13r^02+2r^04>8+8r^02+2r^048+13\hat{r}_{0}^{2}+2\hat{r}_{0}^{4}>8+8\hat{r}_{0}^{2}+2\hat{r}_{0}^{4} in the second inequality.

References

  • [1] R. Penrose, Singularities of spacetime, in Theoretical principles in astrophysics and relativity (W. R. N.R. Liebowitz and P.O.Vandervoort, eds.), pp. 217?243. Chicago University Press, 1978.
  • [2] M. Simpson and R. Penrose, “Internal instability in a Reissner-Nordstrom black hole,” Int. J. Theor. Phys. 7 (1973), 183-197 doi:10.1007/BF00792069
  • [3] P. R. Brady and C. M. Chambers, “Nonlinear instability of Kerr type Cauchy horizons,” Phys. Rev. D 51 (1995), 4177-4186 doi:10.1103/PhysRevD.51.4177 [arXiv:gr-qc/9501025 [gr-qc]].
  • [4] A. Ori, “Inner structure of a charged black hole: An exact mass-inflation solution,” Phys. Rev. Lett. 67 (1991), 789-792 doi:10.1103/PhysRevLett.67.789.
  • [5] W. A. Hiscock, “Stress Energy Tensor Near a Charged, Rotating, Evaporating Black Hole,” Phys. Rev. D 15 (1977), 3054-3057 doi:10.1103/PhysRevD.15.3054.
  • [6] W. A. Hiscock, “QUANTUM MECHANICAL INSTABILITY OF THE KERR-NEWMAN BLACK HOLE INTERIOR,” Phys. Rev. D 21 (1980), 2057-2063 doi:10.1103/PhysRevD.21.2057
  • [7] A. R. Steif, “The Quantum stress tensor in the three-dimensional black hole,” Phys. Rev. D 49 (1994), 585-589 doi:10.1103/PhysRevD.49.R585 [arXiv:gr-qc/9308032 [gr-qc]].
  • [8] O. J. C. Dias, H. S. Reall and J. E. Santos, “The BTZ black hole violates strong cosmic censorship,” JHEP 12 (2019), 097 doi:10.1007/JHEP12(2019)097 [arXiv:1906.08265 [hep-th]].
  • [9] S. Hollands, R. M. Wald and J. Zahn, “Quantum instability of the Cauchy horizon in Reissner–Nordström–deSitter spacetime,” Class. Quant. Grav. 37 (2020) no.11, 115009 doi:10.1088/1361-6382/ab8052 [arXiv:1912.06047 [gr-qc]].
  • [10] M. Banados, C. Teitelboim and J. Zanelli, “The Black hole in three-dimensional space-time,” Phys. Rev. Lett. 69 (1992), 1849-1851 doi:10.1103/PhysRevLett.69.1849 [arXiv:hep-th/9204099 [hep-th]].
  • [11] S. A. Hartnoll, G. T. Horowitz, J. Kruthoff and J. E. Santos, “Gravitational duals to the grand canonical ensemble abhor Cauchy horizons,” JHEP 10 (2020), 102 doi:10.1007/JHEP10(2020)102 [arXiv:2006.10056 [hep-th]].
  • [12] J. M. Maldacena, “The Large N limit of superconformal field theories and supergravity,” Adv. Theor. Math. Phys. 2 (1998), 231-252 doi:10.4310/ATMP.1998.v2.n2.a1 [arXiv:hep-th/9711200 [hep-th]].
  • [13] K. Maeda, T. Okamura and J. i. Koga, “Inhomogeneous charged black hole solutions in asymptotically anti-de Sitter spacetime,” Phys. Rev. D 85 (2012), 066003 doi:10.1103/PhysRevD.85.066003 [arXiv:1107.3677 [gr-qc]].
  • [14] N. Iizuka, A. Ishibashi and K. Maeda, “Flows of extremal attractor black holes,” JHEP 09 (2022), 093 doi:10.1007/JHEP09(2022)093 [arXiv:2206.04845 [hep-th]].
  • [15] G. T. Horowitz, M. Kolanowski and J. E. Santos, “Almost all extremal black holes in AdS are singular,” JHEP 01 (2023), 162 doi:10.1007/JHEP01(2023)162 [arXiv:2210.02473 [hep-th]].
  • [16] G. T. Horowitz and J. E. Santos, “Smooth extremal horizons are the exception, not the rule,” [arXiv:2411.07295 [hep-th]].
  • [17] M. Casals, A. Fabbri, C. Martínez and J. Zanelli, “Quantum Backreaction on Three-Dimensional Black Holes and Naked Singularities,” Phys. Rev. Lett. 118 (2017) no.13, 131102 doi:10.1103/PhysRevLett.118.131102 [arXiv:1608.05366 [gr-qc]].
  • [18] J. Arrechea, C. Barceló, R. Carballo-Rubio and L. J. Garay, “Reissner–Nordström geometry counterpart in semiclassical gravity,” Class. Quant. Grav. 38 (2021) no.11, 115014 doi:10.1088/1361-6382/abf628 [arXiv:2102.03544 [gr-qc]].
  • [19] G. Compere and D. Marolf, “Setting the boundary free in AdS/CFT,” Class. Quant. Grav. 25 (2008), 195014 doi:10.1088/0264-9381/25/19/195014 [arXiv:0805.1902 [hep-th]].
  • [20] A. Ishibashi, K. Maeda and T. Okamura, “Semiclassical Einstein equations from holography and boundary dynamics,” JHEP 05 (2023), 212 doi:10.1007/JHEP05(2023)212 [arXiv:2301.12170 [hep-th]].
  • [21] S. W. Hawking, C. J. Hunter and M. Taylor, “Rotation and the AdS / CFT correspondence,” Phys. Rev. D 59 (1999), 064005 doi:10.1103/PhysRevD.59.064005 [arXiv:hep-th/9811056 [hep-th]].
  • [22] S. de Haro, S. N. Solodukhin and K. Skenderis, “Holographic reconstruction of space-time and renormalization in the AdS / CFT correspondence,” Commun. Math. Phys. 217, 595-622 (2001) doi:10.1007/s002200100381 [arXiv:hep-th/0002230 [hep-th]].
  • [23] V. Balasubramanian and P. Kraus, “A Stress tensor for Anti-de Sitter gravity,” Commun. Math. Phys. 208 (1999), 413-428 doi:10.1007/s002200050764 [arXiv:hep-th/9902121 [hep-th]].
  • [24] A. Ishibashi, K. Maeda and T. Okamura, “The higher dimensional instabilities of AdS in holographic semiclassical gravity,” [arXiv:2412.18764 [hep-th]].
  • [25] A. Ishibashi, K. Maeda and T. Okamura, JHEP 02 (2024), 146 doi:10.1007/JHEP02(2024)146 [arXiv:2312.10311 [hep-th]].
  • [26] S. W. Hawking and G. F. R. Ellis, The large scale structure of spacetime, Cambridge University Press, Cambridge, 1973.
  • [27] A. Ishibashi, K. Maeda and E. Mefford, Phys. Rev. D 96 (2017) no.2, 024005 doi:10.1103/PhysRevD.96.024005 [arXiv:1703.09743 [hep-th]].