betaFeedback

WorksPhysics and Astronomy

The Spectral Anatomy of Complex-Time Singularities

Abstract

The properties of the bulk spacetime in a holographic theory may be probed by the analytic structure of the boundary correlator G(t,x)G(t,x). In this letter, we analyse the singularities of G(t,x)G(t,x) by deriving them from Fourier space. We show that the quasinormal modes are characterised by two distinct momentum scaling regimes. The first regime is defined by a finite k/ωγk/ω^γ for some characteristic exponent γγ which is extracted from the singularity, and probes G(t,0)G(t,0), reproducing the bouncing geodesic analysis. The second regime is defined by a finite k/ωk/ω, and probes the complexified bulk away from the singularity, which allows us to study the singularities of G(t,x≠0)G(t,x\neq 0). We extract coefficients and exponents of the singularities for general spacetime dimensions, and allow for more exotic interiors. The heavy reliance of our computation on quasinormal modes opens the doors to a similar analysis being done on spacetimes that are accessible only numerically.

Introduction

Note: During the final stages of this work, a preprint [1] appeared with substantial overlap with our results. The present work was developed independently.

In the large-N limit of holographic field theories, the boundary two-point functions G(t,x)G(t,x) encode information about the bulk geometry [2, 3, 4]. Recently, this has attracted a lot of attention in the context of the imprint of the black hole singularity on the boundary. The relevant computation involves a geodesic computation of the analytically extended bulk spacetime, known in this context as the ‘bouncing geodesic’ [5, 6], which leads to a branch point of G(t,0)G(t,0) in complex time [7, 8, 9] – the bouncing singularity. These studies often require a study of complexified geodesics and OPE [7, 8, 10, 11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28], and become especially difficult when x≠0x\ne0. Another perspective is offered by the Fourier space perspective, by studying G(ω,k)G(\omega,k). There, the correlator is determined by the quasinormal modes (QNMs) [29], which allows for derivation of the singularities G(t,k)G(t,k) via large-ω\omega transasymptotics [9, 23, 30, 31, 32, 33, 34, 35, 36, 37, 38, 39, 40, 41, 42, 43, 44, 45, 46]. However, fixed-xx and fixed-kk correlators have different analytic structure in complex time, obscuring the precise connection between the QNM and the geodesic approaches [9].

In this work, we address this gap by studying the large-kk behaviour of QNMs and performing the subsequent Fourier transform. We are specifically treating a dd-dimensional boundary (d≥3d\ge3) dual to the planar anti-de Sitter-Schwarzschild black hole (AdS-SC), and a probe scalar with scaling dimension Δ=d/2+ν\Delta=d/2+\nu. Many of the results of this paper are, however, applicable to more general geometries, such as black holes with a Kasner interior [39, 47, 48, 49, 50]. We perform a WKB analysis in large ω\omega to derive the asymptotic QNMs at various momenta kk. We derive several distinct momentum regimes, the classification of which is heavily dependent on the near-singularity behaviour of the bulk equations of motion.

More specifically, the singularity singles out an exponent γ\gamma with 0<γ<1/20<\gamma<1/2, and the classification of large-nn QNMs is the following:1

  • The small-kk regime k≪nγk\ll n^\gamma is the one that has been studied extensively in WKB studies of QNMs [51, 52, 53], with the first correction given in [39]:

ωn∼πZ[n+ν−12+ilog⁡22π+α0k2n2γ].(1)\omega_n\sim\frac{\pi}{\mathcal{Z}}\left[n+\frac{\nu-1}{2}+\mathrm{i}\frac{\log2}{2\pi}+\alpha_0\frac{k^2}{n^{2\gamma}}\right]. \tag*{(1)}
  • The inner-kk regime k∼nγk \sim n^\gamma is characterised by a non-trivial function FF that must be computed from near-singularity physics

ωn∼πZ[n+ν−12+F(knγ)].(2)\omega_n \sim\frac{\pi}{\mathcal{Z}}\left[n+\frac{\nu-1}{2}+F\left(\frac{k}{n^\gamma}\right)\right]. \tag*{(2)}
  • The intermediate-kk regime n≫k≫nγn \gg k \gg n^\gamma corresponds in the bulk to the region which is close to the singularity in terms of the AdS radius, but far from singularity in terms of the wavelength of the scalar solution

ωn∼πZn[1+α1(kn)2−2d].(3)\omega_n \sim\frac{\pi}{\mathcal{Z}}n\left[1+\alpha_1\left(\frac{k}{n}\right)^{2-\frac{2}{d}}\right]. \tag*{(3)}
  • The outer-kk regime k∼nk \sim n probes the complexified bulk away from the singularity, and is governed by functions H(k/n)H(k/n) and Hˉ(k/n)\bar{H}(k/n) that can be expressed in closed form. Explicitly,

ωn∼nH(k/n)+ν−12Hˉ(k/n).(4)\omega_n \sim n H(k/n)+\frac{\nu-1}{2}\bar{H}(k/n). \tag*{(4)}
  • The near-null regime k≫nk \gg n probes the near-boundary region and was previously derived in [54, 55]

ωn∼k+α2n(kn)−d−2d+2.(5)\omega_n \sim k+\alpha_2 n\left(\frac{k}{n}\right)^{-\frac{d-2}{d+2}}. \tag*{(5)}

The definition and value of the optical distance Z\mathcal{Z} and explicit expressions for α0\alpha_0, α1\alpha_1, and α2\alpha_2 are provided below. The small-kk and inner regimes probe the singularity (see dashed green line on Figure 1), the outer regime probes the far-from-singularity bulk (see green line on Figure 1), and the near-null regime probes the near boundary region. The intermediate regime is especially interesting, as it probes the region which is close to the singularity in terms of the AdS radius, but far from it in terms of the characteristic wavelength of the solution ∼1/ω\sim1/\omega (see red region in Figure 1).

Figure 1. Schematic representation of the distinct WKB regimes determining physics in the complex-zz plane. Here the boundary is located at z=0z=0, horizon at z=1z=1 and curvature singularity at z=∞z=\infty. The green lines track the anti-Stokes lines in the outer-kk regime k∼ωk\sim\omega (solid green) and inner-kk regime k∼ωγk\sim\omega^\gamma (dashed green). Slightly away from the singularity the red area denotes the intermediate regime ω≫k≫ωγ\omega\gg k\gg\omega^\gamma. The solid red line tracks the movement of the turning point z∗z_* for the WKB in the direction of growing kk.

To derive G(t,x)G(t,x) from QNMs, one would require the full momentum dependence, not just the relevant scaling regimes. However, we show that the results presented above are sufficient to study the near singularity behaviour of the retarded correlator

G(t→tc,x)∼Fc(t−tc)r,(6)G(t\to t_c,x)\sim\frac{F_c}{(t-t_c)^r}, \tag*{(6)}

specifically, to compute the position of the singularity tct_c, coefficient FcF_c, and the power rr, all of which are a function of xx. When x=0x=0, those can be derived from the inner scaling regime (2), with the exponent γ\gamma playing the central role. This is coherent with the bouncing geodesic story, where the expression (6) descends from the geodesic bounce. When x≠0x\ne0, the data in Eq. (6) can be derived from the outer regime (4), probing instead the far-from-singularity region of spacetime. The results presented here agree with and extend the results currently present in literature. Importantly, the analysis presented here may be extended to more interesting bulk geometries as well as geometries that are only accessible numerically. This is due to the fact that QNMs are easily computable on a numerical background, in great contrast to complex geodesics.

In Section II we outline the WKB computation, which we then apply to the inner, outer and intermediate scaling regimes. In Section III we perform the Fourier computation and connection to both bouncing and bulk-cone geodesics. In Section IV we discuss the application of the computation to more general geometries and potential future directions.

WKB computation

Throughout, we are working with ω\omega in the fourth complex quadrant (Im⁡ω<0\operatorname{Im}\omega<0 and Re⁡ω>0\operatorname{Re}\omega>0), a real positive kk, and a complexified bulk radial coordinate so that z=0z=0 is the boundary, z=1z=1 the horizon, and z=∞z=\infty the singularity (see Figure 1). That is encoded in the emblackening factor f(z)=−z2gtt=1/z2gzzf(z)=-z^2g_{tt}=1/z^2g_{zz} (f(z)=1−zdf(z)=1-z^d in the AdS-SC). It is convenient to introduce

μ=kω,(7)\mu=\frac{k}{\omega}, \tag*{(7)}

and consider the physics away from the lightcone ∣μ∣<1|\mu|<1.

We are interested in solving the bulk wave equation of a minimally coupled scalar in the regime of large ω\omega. The WKB solution approximates solutions well when the characteristic wavelength of solution is much smaller than the characteristic scale of the coefficients of the equation of motion. This is precisely applicable to large-ω\omega limit of the bulk wave equation (see e.g. [30, 32, 35, 39, 51, 52, 53, 56] for the technical details). We introduce

p(z)=1−μ2f(z)f(z).(8)p(z)=\frac{\sqrt{1-\mu^2 f(z)}}{f(z)}. \tag*{(8)}

The solution is valid away from the zeroes and singularities of p(z)p(z). We set the ingoing boundary condition at the horizon, meaning that, in the near-horizon matching region

ψhorizon(z)∼eiω∫zp(z′) dz′.(9)\psi_{\mathrm{horizon}}(z)\sim e^{i\omega\int^z p(z')\,\mathrm{d}z'}. \tag*{(9)}

The WKB approximation breaks down near the boundary, where an explicit near-boundary solution must be constructed, relating the boundary fall-offs to the WKB solution in the respective matching region. We introduce a function Ω(ω,k)\Omega(\omega,k) that parametrises the specific superposition of the two near-boundary solutions, so that, in the matching region,

ψbdry(z)∼ie2πiΩ+iω∫zp(z′) dz′+e−iω∫zp(z′) dz′.(10)\psi_{\mathrm{bdry}}(z)\sim ie^{2\pi i\Omega+i\omega\int^z p(z')\,\mathrm{d}z'}+e^{-i\omega\int^z p(z')\,\mathrm{d}z'}. \tag*{(10)}

The boundary and the horizon WKB solution may be matched to one another by matching the coefficients in front of the oscillatory exponentials. The exponentials are oscillatory only when

Im⁡(ω∫zp(z′) dz′)=0,(11)\operatorname{Im}\left(\omega\int^{z}p(z')\,\mathrm{d}z'\right)=0, \tag*{(11)}

which defines anti-Stokes lines in the complex zz plane. For generic values of ω\omega and kk, there is always a trivial solution as there is an anti-Stokes line connecting the horizon and boundary directly, giving Ω→−i∞\Omega\to-i\infty. The non-trivial solutions relevant for our study come when the anti-Stokes lines emanating from the boundary and horizon meet at a turning point or the singularity, yielding a non-trivial Ω(ω,k)\Omega(\omega,k). The boundary retarded correlator may be expressed as

G(ω,k)≈πd2Γ(−ν)4νΓ(d2+ν)(ω2−k2)νcos⁡π(Ω(ω,k)−ν2)cos⁡π(Ω(ω,k)+ν2),(12)G(\omega,k)\approx\frac{\pi^{\frac{d}{2}}\Gamma(-\nu)}{4^{\nu}\Gamma\left(\frac{d}{2}+\nu\right)}\left(\omega^{2}-k^{2}\right)^{\nu}\frac{\cos\pi\left(\Omega(\omega,k)-\frac{\nu}{2}\right)}{\cos\pi\left(\Omega(\omega,k)+\frac{\nu}{2}\right)}, \tag*{(12)}

where the normalisation is chosen so that G(x)=1/x2ΔG(x)=1/x^{2\Delta} in the vacuum. The WKB approximation holds well when ω→∞\omega\to\infty in the direction of the complex plane where Im⁡Ω≈0\operatorname{Im}\Omega\approx0 and Im⁡ω<0\operatorname{Im}\omega<0. The quasinormal modes are the solutions of

Ω(ω,k)+ν−12+n=0.(13)\Omega(\omega,k)+\frac{\nu-1}{2}+n=0. \tag*{(13)}

Outer regime (k∼ωk\sim\omega)

At finite μ\mu, i.e. when k∼ωk\sim\omega, the non-trivial solutions to the WKB problem are the ones where there is an anti-Stokes line emanating from the boundary, hitting a turning point z∗z_\ast, and connecting to the horizon by an anti-Stokes line that is rotated for −2π/3-2\pi/3 (see solid green curves in Figure 1). The relevant turning point is the solution to

z∗d=1−1μ2,(14)z_\ast^{d}=1-\frac{1}{\mu^{2}}, \tag*{(14)}

with Re⁡z∗>0\operatorname{Re}z_\ast>0 and Im⁡z∗>0\operatorname{Im}z_\ast>0 (in case of more such solutions, the relevant one is the one with the smallest imaginary part). The problem is solved by the standard Airy solution around the turning point. The result of the computation is

Ω=−Ψ(μ)ω+O(ω−1),(15)\Omega=-\Psi(\mu)\omega+\mathcal{O}(\omega^{-1}), \tag*{(15)}

where

Ψ(μ)≡1π∫0z∗p(z′) dz′,(16)\Psi(\mu)\equiv\frac{1}{\pi}\int_{0}^{z_\ast}p(z')\,\mathrm{d}z', \tag*{(16)}

the contour of integration being taken as suggested by the red line in Figure 1. This integral may be expressed in closed form

Ψ(μ)=z∗1−μ2dπB(1d,32)2F1(1,1d,12+1d;1−1μ2).(17)\Psi(\mu)=\frac{z_\ast\sqrt{1-\mu^{2}}}{d\pi}B\left(\frac{1}{d},\frac{3}{2}\right){}_2F_1\left(1,\frac{1}{d},\frac{1}{2}+\frac{1}{d};1-\frac{1}{\mu^{2}}\right). \tag*{(17)}

Using Eq. (13), the QNMs may be expressed by introducing H(x)H(x) such that

H(x)Ψ(xH(x))=1,Hˉ(x)=H(x)−xH′(x).(18)H(x)\Psi\left(\frac{x}{H(x)}\right)=1,\qquad\bar{H}(x)=H(x)-xH'(x). \tag*{(18)}

The limit of μ→1\mu\to1 yields an expansion

Ψ(μ→1)∼(1−μ2)12+1deiπdB(1d,32)dπ,(19)\Psi(\mu\to1)\sim(1-\mu^{2})^{\frac{1}{2}+\frac{1}{d}}e^{i\frac{\pi}{d}}\frac{B\left(\frac{1}{d},\frac{3}{2}\right)}{d\pi}, \tag*{(19)}

giving (see Eq. (5))

α2=12[e−iπddπB(1d,32)]2d2+d.(20)\alpha_{2}=\frac{1}{2}\left[e^{-i\frac{\pi}{d}}\frac{d\pi}{B\left(\frac{1}{d},\frac{3}{2}\right)}\right]^{\frac{2d}{2+d}}. \tag*{(20)}

The limit of μ→0\mu\to0 gives

Ψ(μ→0)∼1π[Z+μ2−2deiπdB(12,1d)2(1−d)],(21)\Psi(\mu\to0)\sim\frac{1}{\pi}\left[\mathcal{Z}+\mu^{2-\frac{2}{d}}e^{i\frac{\pi}{d}}\frac{B\left(\frac{1}{2},\frac{1}{d}\right)}{2(1-d)}\right], \tag*{(21)}

where we have introduced the optical distance (see e.g. [39])

Z≡πΨ(0)=θeiθsin⁡θ(22)\mathcal{Z}\equiv\pi\Psi(0)=\theta\frac{e^{i\theta}}{\sin\theta} \tag*{(22)}

where θ=π/d\theta=\pi/d. It is related to the inverse temperature β\beta through

Im⁡Z=β4.(23)\operatorname{Im}\mathcal{Z}=\frac{\beta}{4}. \tag*{(23)}

This limit will be elaborated on in Section II C.

Inner regime (k∼ωγk\sim\omega^\gamma)

When μ=0\mu=0, we get p(z)≈1/f(z)p(z)\approx1/f(z), and z∗z_\ast reaches the singularity and the computation does not involve a turning point anymore. There is an anti-Stokes line emerging from the boundary to the singularity with arg⁡(z)∼θ\arg(z)\sim\theta and another from the horizon to the singularity with arg⁡(z)∼θ−π/(d−1)\arg(z)\sim\theta-\pi/(d-1), see green dashed lines in Figure 1. These lines enter the near-singularity regime where WKB breaks down. We must therefore match the near-singularity solution to the two anti-Stokes lines. To be more concrete, we introduce a new variable yy

y≡ω(∫0zdz′f(z′)−Z)(24)y\equiv\omega\left(\int_{0}^{z}\frac{\mathrm{d}z'}{f(z')}-\mathcal{Z}\right) \tag*{(24)}

so that, near the singularity y∼ωz1−d/(d−1)y\sim\omega z^{1-d}/(d-1). The near-singularity equation reduces to

(d2dy2+1+14y2+λ2y2(1−γ))ψ(y)=0,(25)\left(\frac{\mathrm{d}^{2}}{\mathrm{d}y^{2}}+1+\frac{1}{4y^{2}}+\frac{\lambda^{2}}{y^{2(1-\gamma)}}\right)\psi(y)=0, \tag*{(25)}

where

λ=Akωγ.(26)\lambda=\sqrt{A}\frac{k}{\omega^\gamma}. \tag*{(26)}

In the AdS-SC case, we have

A=(d−1)d1−d,γ=d−22(d−1).(27)A=(d-1)^{\frac{d}{1-d}},\qquad\gamma=\frac{d-2}{2(d-1)}. \tag*{(27)}

This calculation, however, is applicable to more exotic singularities, such as Kasner. We therefore leave AA and γ\gamma arbitrary, so long as 0<γ<1/20<\gamma<1/2. In the matching regions, we connect to the ingoing boundary conditions from the horizon

ψsing(y)∝eiy,(28)\psi_{\mathrm{sing}}(y)\propto e^{iy}, \tag*{(28)}

and to the boundary via

ψsing(ye−iπ)∝ie2πiΦ(λ)−iy+eiy(29)\psi_{\mathrm{sing}}(ye^{-i\pi})\propto ie^{2\pi i\Phi(\lambda)-iy}+e^{iy} \tag*{(29)}

where the e−iπe^{-i\pi} continuation corresponds precisely to matching the solution between the horizon and boundary anti-Stokes line. This introduces Φ(λ)\Phi(\lambda) as a function encoding the mapping between the matching coefficients. While there is no explicit form for Φ(λ)\Phi(\lambda), it is easily computable by numerical integration. By matching the solutions on the anti-Stokes lines, we get

Ω=−Zπω+Φ(λ).(30)\Omega=-\frac{\mathcal{Z}}{\pi}\omega+\Phi(\lambda). \tag*{(30)}

Solving for the quantisation condition (13) yields

F(x)=Φ(ZγπγAx).(31)F(x)=\Phi\left(\frac{\mathcal{Z}^{\gamma}}{\pi^{\gamma}}\sqrt{Ax}\right). \tag*{(31)}

The small-λ\lambda regime of Φ(λ)\Phi(\lambda) has been derived in [39]

Φ(λ→0)∼ilog⁡22π+λ2γie−iπγB(12,12−γ)4Γ(1−γ)2+…,(32)\Phi(\lambda\to0)\sim i\frac{\log2}{2\pi}+\lambda^{2\gamma}\frac{ie^{-i\pi\gamma}B\left(\frac{1}{2},\frac{1}{2}-\gamma\right)}{4\Gamma(1-\gamma)^2}+\ldots, \tag*{(32)}

giving, in Eq. (1),

α0=A(Zπ)2γie−iπγB(12,12−γ)4Γ(1−γ)2.(33)\alpha_0=A\left(\frac{\mathcal{Z}}{\pi}\right)^{2\gamma}\frac{ie^{-i\pi\gamma}B\left(\frac{1}{2},\frac{1}{2}-\gamma\right)}{4\Gamma(1-\gamma)^2}. \tag*{(33)}

Intermediate regime (ω≫k≫ωγ\omega\gg k\gg\omega^\gamma)

As mentioned before, the small-kk regime (1) and the near-null regime (5) correspond to the near-singularity and near-boundary WKB matching computations respectively. The intermediate regime is more interesting and subtle, and may be derived in two distinct, but ultimately equivalent, ways.

The first way is to expand Φ(λ)\Phi(\lambda) for large λ\lambda. This involves performing a WKB analysis on Eq. (25). The anti-Stokes lines at y→∞±y\to\infty^{\pm} are matched via a turning point at

y∗=(λe−iπ2)11−γ,(34)y_*=\left(\lambda e^{-i\frac{\pi}{2}}\right)^{\frac{1}{1-\gamma}}, \tag*{(34)}

with a standard Airy computation rotating for −2π/3-2\pi/3 from the boundary line to the horizon line. This gives

Φ(λ→∞)∼λ11−γeiπ21−2γ1−γB(12,1−2γ2(1−γ))2π.(35)\Phi(\lambda\to\infty)\sim\lambda^{\frac{1}{1-\gamma}}\frac{e^{i\frac{\pi}{2}\frac{1-2\gamma}{1-\gamma}}B\left(\frac{1}{2},\frac{1-2\gamma}{2(1-\gamma)}\right)}{2\pi}. \tag*{(35)}

On the other hand, we can expand Ψ(μ)\Psi(\mu) for small μ\mu, giving

Ψ(μ→0)∼1π[Z+μ2−2deiπ2B(12,1d)2(1−d)].(36)\Psi(\mu\to0)\sim\frac{1}{\pi}\left[\mathcal{Z}+\mu^{2-\frac{2}{d}}e^{i\frac{\pi}{2}}\frac{B\left(\frac{1}{2},\frac{1}{d}\right)}{2(1-d)}\right]. \tag*{(36)}

Taking Eq. (27) into account, these two expressions contain precisely the same data, and an overlap identity holds:

λ≫1≫μ:Zπ−Ψ(μ)=Φ(λ)ω.(37)\lambda\gg1\gg\mu:\qquad\frac{\mathcal{Z}}{\pi}-\Psi(\mu)=\frac{\Phi(\lambda)}{\omega}. \tag*{(37)}

This gives, in Eq. (3),

α1=−eiπ2B(12,1d)2π(1−d)(πZ)d2−1.(38)\alpha_1=-e^{i\frac{\pi}{2}}\frac{B\left(\frac{1}{2},\frac{1}{d}\right)}{2\pi(1-d)}\left(\frac{\pi}{\mathcal{Z}}\right)^{\frac{d}{2}-1}. \tag*{(38)}

The fact that the expansions in two distinct regimes have an overlap is expected, as there is no intermediate scale that could interfere with the smooth transition. Nevertheless, the fact that the two computations give the same result serves as an important validity check, and it is pertinent to understand it better. The outer regime holds whenever the turning point z∗z_* lies many WKB wavelengths away from the singularity. On the other hand, the inner regime holds whenever the matching computation is done in the near-singularity region. The overlapping region is

1≪∣z∗∣d−1≪∣ω∣,(39)1\ll|z_*|^{d-1}\ll|\omega|, \tag*{(39)}

which singles out a specific region of the complexified bulk which is close to the singularity in terms of AdS radius, but far away from it in terms of the WKB wavelength (see red region in Figure 1).

Fourier integrals and complex time singularities

To study the singularities of G(t,x)G(t,x), it is convenient to expand the cosine of Eq. (12) into a geometric series

G(ω,k)∝∑nGn(ω,k)(40)G(\omega,k)\propto\sum_n G_n(\omega,k) \tag*{(40)}

with

Gn(ω,k)∝(ω2−k2)νe−2πin(Ω(ω,k)+ν−12).(41)G_n(\omega,k)\propto(\omega^2-k^2)^\nu e^{-2\pi in\left(\Omega(\omega,k)+\frac{\nu-1}{2}\right)}. \tag*{(41)}

where the proportionality holds up to the Ω\Omega-independent n=0n=0 term, which will not affect the outcome of our computation.2 For now, we discard the universal proportionality factors and recover them for the final results. The crucial observation is the fact that a singularity of the type (6) emerges from integrals of the type

∫(ωr−1+⋯ )eiω(tc−t)+⋯ dω∼Γ(r)(it−itc)r,(42)\int\left(\omega^{r-1}+\cdots\right)e^{i\omega(t_c-t)+\cdots}\,\mathrm{d}\omega\sim\frac{\Gamma(r)}{(it-it_c)^r}, \tag*{(42)}

where some convergent contour must be chosen. Non-equivalent choices of the contour for the same values of complex tc−tt_c-t correspond to picking the sheet when going around the branch point in complex time. Crucially, the singularity’s leading coefficient, exponent, and location do not depend on the subleading terms denoted by the ellipsis [31], nor on the lower bound of integration. We are therefore justified in using the large-ω\omega WKB analysis of Section II to obtain singularity results. Secondly, since Ω∼ω\Omega\sim\omega, every term in the sum of (40) is expected to result in a singularity tnt_n. Importantly, two terms with a different exponent cannot contribute to each other’s leading singularity behaviour. Therefore it is justified to treat the terms in the sum independently.

x=0

We consider the integral

G(t,0)∝∫0∞e−iωt dω∫0∞kd−2 dk G(ω,k).(43)G(t,0) \propto\int_0^\infty e^{-i\omega t}\,\mathrm{d}\omega\int_0^\infty k^{d-2}\,\mathrm{d}k\,G(\omega,k). \tag*{(43)}

We may do a substitution k=k(λ)k=k(\lambda) (see Eq. (26)). For large ω\omega we get

kd−2 dk G(ω,k)∼ω2ν+γ(d−1)A1−d2×∑nλd−2e−2πin(Ω+ν−12) dλ.(44)k^{d-2}\,\mathrm{d}k\,G(\omega,k) \sim\omega^{2\nu+\gamma(d-1)} A^{\frac{1-d}{2}} \times\sum_n \lambda^{d-2}e^{-2\pi i n\left(\Omega+\frac{\nu-1}{2}\right)}\,\mathrm{d}\lambda. \tag*{(44)}

If we pick arg⁡ω=−θ\arg\omega=-\theta, we may plug in the WKB result of Eq. (30). The leading large-ω\omega behaviour factorises into the ω\omega-dependent and the λ\lambda-dependent factors, meaning we can integrate over them separately.3 The x=0x=0 computation therefore maps onto the inner regime, which is exactly what we would expect from the bouncing geodesic picture. We define

M(n)≡∫0eiγθ∞λd−2 dλ e−2πin(Φ(λ)+ν−12).(45)M(n)\equiv\int_0^{e^{i\gamma\theta}\infty}\lambda^{d-2}\,\mathrm{d}\lambda\,e^{-2\pi i n\left(\Phi(\lambda)+\frac{\nu-1}{2}\right)}. \tag*{(45)}

Due to (35), the integral is not convergent for positive nn, which is a consequence of the geometric series expansion not commuting with the momentum integral. We must therefore evaluate M(n)M(n) in some convergent region and then analytically continue.4 This may be justified by a careful resummation. Note that the integral (45) is evaluated strictly for the complex argument that emerges in the QNM expression [2], and is therefore directly accessible from quasinormal modes.

Defining tn=2nZt_n=2n\mathcal{Z} and using Eq. (42), the ω\omega integral yields

G(t→tn,0)∼CA1−d2M(n)Γ(p)eiπ1−p2(t−tn(0))p,(46)G(t\to t_n,0)\sim C\frac{A^{\frac{1-d}{2}}M(n)\Gamma(p)e^{i\pi\frac{1-p}{2}}}{(t-t_n(0))^p}, \tag*{(46)}

where p=2ν+1+(d−1)γp=2\nu+1+(d-1)\gamma, and

C=22−d−2νπΓ(d−12)Γ(1+ν)Γ(d2+ν).(47)C=\frac{2^{2-d-2\nu}\sqrt{\pi}}{\Gamma\left(\frac{d-1}{2}\right)\Gamma(1+\nu)\Gamma\left(\frac{d}{2}+\nu\right)}. \tag*{(47)}

We emphasise again that the normalisation is chosen so that G(x)=1/x2ΔG(x)=1/x^{2\Delta} in the vacuum. The location of the singularities tnt_n recovers the standard bouncing geodesic result at n=1n=1 [8], including the correct exponents, and extends the results to other values of nn. These should be understood as nn-bounces off the curvature singularity, see e.g. [9, 30, 31].

nonzero x

We now consider x≠0x\ne0. Similarly as for the small momentum, we Fourier transform each individual term Gn(ω,k)G_n(\omega,k) in the sum (40), which, after substitution k=μωk=\mu\omega, lead to

Gn(t,x)∼x3−d2∫0∞e−iωt dω Hn(ω),(48)G_n(t,x)\sim x^{\frac{3-d}{2}}\int_0^\infty e^{-i\omega t}\,\mathrm{d}\omega\,\mathcal{H}_n(\omega), \tag*{(48)}

where

Hn(ω)=ωd+12+2ν∫0∞dμ μd−12Jd−32(xωμ)(1−μ2)νe2πinΨ(μ).(49)\mathcal{H}_n(\omega)=\omega^{\frac{d+1}{2}+2\nu}\int_0^\infty\mathrm{d}\mu\,\mu^{\frac{d-1}{2}}J_{\frac{d-3}{2}}(x\omega\mu)(1-\mu^2)^\nu e^{2\pi i n\Psi(\mu)}. \tag*{(49)}

By arguments similar to before, the approximation here is justified by the fact that it isolates the leading large-ω\omega behaviour. Expanding the Bessel function in the large argument, we get

Hn(ω)∼ωd2+2ν2πx∑±e∓iπ4(d−2)××∫dμ μd−22(1−μ2)νeiω(±xμ+2πnΨ(μ)).(50)\begin{aligned} \mathcal{H}_n(\omega)\sim\frac{\omega^{\frac{d}{2}+2\nu}}{\sqrt{2\pi x}}\sum_{\pm}e^{\mp i\frac{\pi}{4}(d-2)}\times\\ &\times\int\mathrm{d}\mu\,\mu^{\frac{d-2}{2}}(1-\mu^2)^\nu e^{i\omega(\pm x\mu+2\pi n\Psi(\mu))}. \tag*{(50)} \end{aligned}

This integral can now be treated using the saddle point approximation; the saddle points μ±\mu_\pm are located at

±x+2πnΨ′(μ±)=0,(51)\pm x+2\pi n\Psi'(\mu_\pm)=0, \tag*{(51)}

giving an explicit expression

x=±2ndeiπ/dB(12,1d)[μ±1−μ±2]d−2d.(52)x=\pm\frac{2n}{d}e^{i\pi/d}B\left(\frac{1}{2},\frac{1}{d}\right)\left[\frac{\mu_\pm}{\sqrt{1-\mu_\pm^2}}\right]^{\frac{d-2}{d}}. \tag*{(52)}

The complex time singularities then appear at

tn+(x)=xμ++2πnΨ(μ+),tn−(x)=tn+(−x).(53)t_n^+(x)=x\mu_++2\pi n\Psi(\mu_+),\qquad t_n^-(x)=t_n^+(-x). \tag*{(53)}

where xx is implicitly expressed by μ±\mu_{\pm}. Evaluating Eq. (50) on the saddles, we get

Hn(ω)∼x−1/2∑±Fn±(x)ωq−1eiωtn±(x),(54)\mathcal{H}_{n}(\omega) \sim x^{-1/2}\sum_{\pm}F_{n}^{\pm}(x)\omega^{q-1}e^{i\omega t_{n}^{\pm}(x)}, \tag*{(54)}
Fn±(x)=e∓iπ4(d−2)μ±d−22(1−μ±2)ν2in(d−2)d2eiπ2dB(12,1d)μ±−d2(1−μ±2)1d−32,(55)F_{n}^{\pm}(x)=\frac{e^{\mp\frac{i\pi}{4}(d-2)}\mu_{\pm}^{\frac{d-2}{2}}(1-\mu_{\pm}^{2})^{\nu}}{\sqrt{\frac{2in(d-2)}{d^{2}}}e^{\frac{i\pi}{2d}}B\left(\frac{1}{2},\frac{1}{d}\right)\mu_{\pm}^{-\frac{d}{2}}(1-\mu_{\pm}^{2})^{\frac{1}{d}-\frac{3}{2}}}, \tag*{(55)}

where we assumed both saddles are crossed by the deformed contours. Finally, plugging this into Eq. (48), and decomposing we get the final result as

G(t→tn±(x),x)∼Dx1−d2Fn±(x)eiπ(n(1−ν)−q2)Γ(q)(t−tn±(x))q,(56)G(t\to t_{n}^{\pm}(x),x)\sim D\frac{x^{1-\frac{d}{2}}F_{n}^{\pm}(x)e^{i\pi\left(n(1-\nu)-\frac{q}{2}\right)}\Gamma(q)}{\left(t-t_{n}^{\pm}(x)\right)^{q}}, \tag*{(56)}

where q=2ν+d+12q=2\nu+\frac{d+1}{2} and

D=iπ212−d−2νΓ(1+ν)Γ(d2+ν).(57)D=i\frac{\sqrt{\pi}2^{\frac{1}{2}-d-2\nu}}{\Gamma(1+\nu)\Gamma\left(\frac{d}{2}+\nu\right)}. \tag*{(57)}

First, note that the power qq of the singularity differs from the case x=0x=0 by q−p=1/2q-p=1/2. This result precisely matches one of the complex bulk-cone singularities, which are expected to describe the large-xx regime, cf. Eq. (4.27) of [15]. As a further comparison, taking Im⁡(x)=0\operatorname{Im}(x)=0, one reconstructs the location of the complex bulk-cone singularity [15] and continuously connects it to the bouncing singularity, simply using the saddle point result (53). Concretely, taking (53), defining t∗=tn+t_{*}=t_{n}^{+} and focusing on d=4d=4, one finds these two regimes to be connected by a junction point at x∗=B(12,14)/2≈2.6221x_{*}=B\left(\frac{1}{2},\frac{1}{4}\right)/2\approx2.6221 where the behaviour sharply, yet continuously, changes (see Figure 2)

Figure 2. Behaviour of the saddle time t∗t_{*} (59) and (58) as a function of x∈Rx\in\mathbb{R} in d=4d=4. For x<x∗x<x_{*} the time t∗t_{*} smoothly connects the standard x=0x=0 bouncing singularity and the junction point. For x>x∗x>x_{*} the time t∗t_{*} continuously transits to the (complex) bulk-cone singularity of [15]. For x→0x\to0, Im⁡(t∗)\operatorname{Im}(t_{*}) obeys the expected x3x^{3} behaviour.

0≤x≤x∗:t∗(x)=β2+iβ2[1−Ix4x∗4(34,12)],(58)0\leq x\leq x_{*}:\quad t_{*}(x)=\frac{\beta}{2}+i\frac{\beta}{2}\left[1-I_{\frac{x^{4}}{x_{*}^{4}}}\left(\frac{3}{4},\frac{1}{2}\right)\right], \tag*{(58)}
x≥x∗:t∗(x)=β2±B(12,14)4∫1x2x∗2y dyy2−1,(59)x\geq x_{*}:\quad t_{*}(x)=\frac{\beta}{2}\pm\frac{B\left(\frac{1}{2},\frac{1}{4}\right)}{4}\int_{1}^{\frac{x^{2}}{x_{*}^{2}}}\frac{\sqrt{y}\,dy}{\sqrt{y^{2}-1}}, \tag*{(59)}

where Iz(a,b)I_{z}(a,b) is the regularised incomplete Beta function. Interestingly, the existence of such junction point is special to d=4d=4.

Finally, let us examine the small-xx limit of these results. Expanding tn±t_{n}^{\pm} for x→0x\to0, we find

tn±(x)−2nZ≃d−22(d−1)(d2nB(12,1d))dd−2e∓iπd−2x2(d−1)d−2,(60)t_{n}^{\pm}(x)-2n\mathcal{Z}\simeq\frac{d-2}{2(d-1)}\left(\frac{d}{2nB\left(\frac{1}{2},\frac{1}{d}\right)}\right)^{\frac{d}{d-2}}e^{\mp\frac{i\pi}{d-2}}x^{\frac{2(d-1)}{d-2}}, \tag*{(60)}

and in particular, in d=4d=4,

tn±≃2nZ∓4i3n2B(12,14)x3,(61)t_{n}^{\pm}\simeq2n\mathcal{Z}\mp\frac{4i}{3n^{2}B\left(\frac{1}{2},\frac{1}{4}\right)}x^{3}, \tag*{(61)}

which is precisely the bouncing geodesic prediction [15]. While this analysis has been done for real xx under the assumption that appropriate contours of the ω\omega-integrals

exist, the natural connection to WKB emerges when considering a complex xx as defined through μ\mu. Specifically, a choice of μ\mu for which Im⁡[Ψ(μ)/μ]=0\operatorname{Im}[\Psi(\mu)/\mu]=0, there is an anti-Stokes line relating that value of xx directly to a turning point in the bulk z∗z_{*}. The x→0x\to0 limit then drives the bulk point to the curvature singularity.

Discussion and generalisations

In this work, we have focused on the planar AdS-Schwarzschild geometry. The computations presented here are, however, applicable to a broader context. In the most narrow sense, they are applicable to any bulk geometry which exhibits the same anti-Stokes topology as the cases studied here. On general grounds, this is expected for all settings where only one pair of QNM branches is present. This was exploited in [39], where a relevant scalar deformation flows toward a Kasner interior. For example, in the case d=3d=3 the power γ\gamma is given by the Kasner exponent ptp_{t}, and the optical distance takes the form

γ=1+pt2(1−pt),Z=∫0(1+iϵ)∞gzzgtt dz.(62)\gamma=\frac{1+p_{t}}{2(1-p_{t})},\qquad\mathcal{Z}=\int_{0}^{(1+i\epsilon)\infty}\sqrt{\frac{g_{zz}}{g_{tt}}}\,dz. \tag*{(62)}

Our result (46) then precisely recovers the exponent pp for the boundary dual to such a spacetime, as was derived in [20, 26].

More general settings will involve more QNM branches – examples include charged black holes [13, 33], Gauss-Bonnet black holes [57, 58], etc. Nevertheless, our analysis is agnostic of possible other QNM branches as long as there is a branch associated with the anti-Stokes geometry described here. A more interesting and subtle question is about the connection between the exponent γ\gamma and the presence or absence of a real bouncing geodesic. Our analysis breaks down if γ≥1/2\gamma\ge1/2, and it would be interesting to understand what happens to QNM scaling as well as the presence of the complex-time singularity in that case. This would be interesting to understand in charged black holes with Kasner interior [49, 50] or regular black holes [44]. Another very interesting avenue would be to apply the same analysis for the mixmaster interior [48, 59, 60, 61]. Importantly, because the analysis relies only on QNMs, the singularity structure of G(t,x)G(t,x) is in principle accessible by numerical means, even for bulk geometries which are not known in closed form (see e.g. [62, 63]). This approach avoids trying to reach the singularity by means of crude numerical analytic continuation. Lastly, the bouncing singularity in complex time is expected to resolve away from the strict large-NN regime [21, 22, 23, 24, 25, 31, 44, 46, 64]. It would be interesting to better understand this resolution in the Fourier picture, and translate the results to the x≠0x \ne0 case.

Acknowledgments

S.V. is funded by the Marie Skłodowska-Curie Actions programme GA-101177446 and the Slovenian Research and Innovation Agency (ARIS), contract number 5110-18/2025-5. M.V. is funded by the NWA ORC programme Emergence at All Scales. In the process of creating this paper, LLMs were used exclusively as a resource and for error diagnosis. Calculations, plots, and text were all done by the authors and the authors alone.

References

References

  1. [1]H. F. Jia and M. Rangamani, (2026), arXiv:2611.13261 [hep-th].arxiv.org/abs/2611.13261
  2. [2]J. M. Maldacena, Adv. Theor. Math. Phys. 2, 231 (1998), arXiv:hep-th/9711200.DOI
  3. [3]S. S. Gubser, I. R. Klebanov, and A. M. Polyakov, Phys. Lett. B 428, 105 (1998), arXiv:hep-th/9802109.DOI
  4. [4]E. Witten, Adv. Theor. Math. Phys. 2, 253 (1998), arXiv:hep-th/9802150.arxiv.org/abs/hep-th/9802150
  5. [5]L. Fidkowski, V. Hubeny, M. Kleban, and S. Shenker, JHEP 02, 014 (2004), arXiv:hep-th/0306170.DOI
  6. [6]G. Festuccia and H. Liu, JHEP 04, 044 (2006), arXiv:hep-th/0506202.
  7. [7]E. Parisini, K. Skenderis, and B. Withers, JHEP 05, 296 (2024), arXiv:2312.03820 [hep-th].arxiv.org/abs/2312.03820
  8. [8]N. Čeplak, H. Liu, A. Parnachev, and S. Valach, JHEP 10, 105 (2024), arXiv:2404.17286 [hep-th].arxiv.org/abs/2404.17286
  9. [9]S. Grozdanov, S. Valach, and M. Vrbica, JHEP 08, 022 (2026), arXiv:2603.15598 [hep-th].arxiv.org/abs/2603.15598
  10. [10]I. Burić, I. Gusev, and A. Parnachev, JHEP 05, 059 (2026), arXiv:2508.08373 [hep-th].arxiv.org/abs/2508.08373
  11. [11]S. Valach, (2025), arXiv:2508.17139 [hep-th].arxiv.org/abs/2508.17139
  12. [12]J. Barrat, D. N. Bozkurt, et al., JHEP 05, 180 (2026), arXiv:2510.20894 [hep-th].arxiv.org/abs/2510.20894
  13. [13]N. Čeplak, H. Liu, A. Parnachev, and S. Valach, JHEP 05, 001 (2026), arXiv:2511.09638 [hep-th].arxiv.org/abs/2511.09638
  14. [14]Y. Jia and M. Kulaxizi, (2026), arXiv:2602.06558 [hep-th].arxiv.org/abs/2602.06558
  15. [15]I. J. Araya, C. Esper, Y. Jia, M. Kulaxizi, and A. Parnachev, (2026), arXiv:2602.12893 [hep-th].arxiv.org/abs/2602.12893
  16. [16]I. Burić, C.-M. Chang, et al., (2026), arXiv:2607.05258 [hep-th].arxiv.org/abs/2607.05258
  17. [17]I. Burić, I. Gusev, and A. Parnachev, (2026), arXiv:2607.24499 [hep-th].arxiv.org/abs/2607.24499
  18. [18]P. Arnaudo, C. Iossa, R. Karlsson, and B. Withers, (2026), arXiv:2607.24909 [hep-th].arxiv.org/abs/2607.24909
  19. [19]J. Barrat, D. N. Bozkurt, E. Marchetto, A. Miscioscia, and E. Pomoni, (2026), arXiv:2607.24919 [hep-th].arxiv.org/abs/2607.24919
  20. [20]N. Čeplak and S. Valach, (2026), arXiv:2609.23776 [hep-th].arxiv.org/abs/2609.23776
  21. [21]G. Batra, A. Levine, and S. H. Shenker, (2026), arXiv:2609.32071 [hep-th].arxiv.org/abs/2609.32071
  22. [22]Z. Komargodski, (2026), arXiv:2609.31309 [hep-th].arxiv.org/abs/2609.31309
  23. [23]I. Burić, I. Gusev, and A. Parnachev, (2026), arXiv:2610.12311 [hep-th].arxiv.org/abs/2610.12311
  24. [24]G. Batra, (2026), arXiv:2610.22478 [hep-th].arxiv.org/abs/2610.22478
  25. [25]M. Dodelson, H. W. Lin, H. Xie, and J. Yeh, (2026), arXiv:2610.31700 [hep-th].arxiv.org/abs/2610.31700
  26. [26]H. Oh, (2026), arXiv:2610.31699 [hep-th].arxiv.org/abs/2610.31699
  27. [27]H. Oh, (2026), arXiv:2611.10998 [hep-th].arxiv.org/abs/2611.10998
  28. [28]J. Chakravarty, (2026), arXiv:2611.15933 [hep-th].arxiv.org/abs/2611.15933
  29. [29]M. Dodelson, C. Iossa, R. Karlsson, and A. Zhiboedov, JHEP 01, 036 (2024), arXiv:2304.12339 [hep-th].arxiv.org/abs/2304.12339
  30. [30]N. Afkhami-Jeddi, S. Caron-Huot, et al., Phys. Rev. D 114, 025005 (2026), arXiv:2510.21673 [hep-th].
  31. [31]M. Dodelson, C. Iossa, and R. Karlsson, JHEP 07, 161 (2026), arXiv:2511.09616 [hep-th].arxiv.org/abs/2511.09616
  32. [32]H. F. Jia and M. Rangamani, (2025), arXiv:2512.15114 [hep-th].arxiv.org/abs/2512.15114
  33. [33]S. Ali Ahmad, A. Almheiri, and S. Lin, (2026), arXiv:2601.02354 [hep-th].arxiv.org/abs/2601.02354
  34. [34]S. Giombi, Y.-Z. Li, and J. Shan, JHEP 07, 170 (2026), arXiv:2603.11012 [hep-th].arxiv.org/abs/2603.11012
  35. [35]H. F. Jia and M. Rangamani, (2026), arXiv:2604.10803 [hep-th].arxiv.org/abs/2604.10803
  36. [36]P. Arnaudo and B. Withers, (2026), arXiv:2605.16489 [gr-qc].arxiv.org/abs/2605.16489
  37. [37]S. Grozdanov, V. Movrin, and S. Valach, (2026), arXiv:2606.11297 [hep-th].arxiv.org/abs/2606.11297
  38. [38]S. Grozdanov, V. Movrin, and S. Valach, (2026), arXiv:2608.13643 [hep-th].arxiv.org/abs/2608.13643
  39. [39]S. A. Hartnoll and A. Zhiboedov, (2026), arXiv:2609.07514 [hep-th].arxiv.org/abs/2609.07514
  40. [40]C. Coviello, A. Gupta, R. A. Hennigar, K. Shi, and A. Svesko, (2026), arXiv:2609.26724 [hep-th].arxiv.org/abs/2609.26724
  41. [41]A. P. Saha and P. Shivanna, (2026), arXiv:2609.29243 [gr-qc].arxiv.org/abs/2609.29243
  42. [42]Y. Shi, C. Ding, Y. Zhou, C. Liu, W. Deng, and S. Long, (2026), arXiv:2610.16973 [hep-th].arxiv.org/abs/2610.16973
  43. [43]J. Chakravarty and Y. Zigdon, (2026), arXiv:2610.30134 [hep-th].arxiv.org/abs/2610.30134
  44. [44]H.-Y. Chen, Y. Hikida, and Y. Koga, (2026), arXiv:2610.30915 [hep-th].arxiv.org/abs/2610.30915
  45. [45]P. Arnaudo and B. Withers, (2026), arXiv:2611.03791 [gr-qc].arxiv.org/abs/2611.03791
  46. [46]M. Liu and S.-M. Ruan, (2026), arXiv:2611.03347 [hep-th].arxiv.org/abs/2611.03347
  47. [47]E. Kasner, Am. J. Math. 43, 217 (1921).
  48. [48]V. A. Belinsky, I. M. Khalatnikov, and E. M. Lifshitz, Adv. Phys. 19, 525 (1970).
  49. [49]S. A. Sarkar, G. T. Horowitz, J. Kruthoff, and J. E. Santos, JHEP 10, 102 (2020), arXiv:2006.10056 [hep-th].
  50. [50]A. Frenkel, S. A. Hartnoll, J. Kruthoff, and Z. D. Shi, JHEP 08, 003 (2020), arXiv:2004.01192 [hep-th].DOI
  51. [51]J. Natario and R. Schiappa, Adv. Theor. Math. Phys. 8, 1001 (2004), arXiv:hep-th/0411267.arxiv.org/abs/hep-th/0411267
  52. [52]V. Cardoso, J. Natario, and R. Schiappa, J. Math. Phys. 45, 4698 (2004), arXiv:hep-th/0403132.DOI
  53. [53]L. Motl and A. Neitzke, Adv. Theor. Math. Phys. 7, 307 (2003), arXiv:hep-th/0301173.
  54. [54]G. Festuccia and H. Liu, Adv. Sci. Lett. 2, 221 (2009), arXiv:0811.1033 [gr-qc].DOI
  55. [55]J. F. Fuini, C. F. Uhleman, and L. G. Yaffe, JHEP 12, 042 (2016), arXiv:1610.03491 [hep-th].DOI
  56. [56]M. A. Stephanov and Y. Yin, JHEP 02, 017 (2012), arXiv:1111.5303 [hep-ph].DOI
  57. [57]N. Čepłak, H. Liu, A. Parnachev, and S. Valach, “Stringy splitting of black hole singularities,” (to appear).
  58. [58]S. Grozdanov and A. O. Starinets, JHEP 04, 080 (2019), arXiv:1812.09288 [hep-th].DOI
  59. [59]C. W. Misner, Phys. Rev. Lett. 22, 1071 (1969).
  60. [60]M. De Clerck, S. A. Hartnoll, and J. E. Santos, JHEP 07, 202 (2024), arXiv:2312.11622 [hep-th].
  61. [61]E. Cáceres, Á. J. Murcia, A. K. Patra, and J. F. Pedraza, JHEP 08, 164 (2026), arXiv:2603.06560 [hep-th].arxiv.org/abs/2603.06560
  62. [62]R. A. Konoplya and A. Zhidenko, JHEP 09, 139 (2017), arXiv:1705.07732 [hep-th].DOI
  63. [63]A. Jansen, Eur. Phys. J. Plus 132, 546 (2017), arXiv:1709.09178 [gr-qc].arxiv.org/abs/1709.09178
  64. [64]A. Almheiri, M. Dodelson, and Z. Komargodski, (2026), arXiv:2610.31610 [hep-th].arxiv.org/abs/2610.31610

Paper details

vlnc:W_q5k5cwsm
Version
v1 · latest
Published
Fresh — added to the Hub within the last 30 days
Added
License
CC BY 4.0
Usage

views · 0

downloads · 0

A view counts once per browser per day, a download each time the PDF is requested.

  • Physics and Astronomy

LLMs

View statement
Looking for related papers…
Contents