Electrovacuum spacetime near an extreme horizon

Carmen Li; James Lucietti

doi:10.4310/ATMP.2019.v23.n7.a5

Li, Carmen;Lucietti, James

2018-09-21 00:00:00

We determine all inﬁnitesimal transverse deformations of extreme horizons in Einstein- Maxwell theory that preserve axisymmetry. In particular, we show that the general static transverse deformation of the AdS ×S near-horizon geometry is a two-parameter family, which contains the known extreme charged, accelerating, static black hole solution held in equilibrium by an external electric or magnetic ﬁeld (Ernst solution) and a special case of the extreme Kerr-Newman-Melvin solution. More generally, we ﬁnd a three- parameter family of deformations of the extreme Kerr-Newman horizon, which contains the extreme Kerr-Newman-Melvin solution and a rotating generalisation of the Ernst so- lution. We also consider vacuum gravity with a cosmological constant and prove unique- ness of axisymmetric transverse deformations of the extreme Kerr-AdS horizon. Finally, we completely classify transverse deformations of extreme horizons in three-dimensional Einstein-Maxwell theory with a negative cosmological constant. k.k.li@lancaster.ac.uk j.lucietti@ed.ac.uk arXiv:1809.08164v1 [gr-qc] 21 Sep 2018 Contents 1 Introduction and main results 2 2 Near an extreme horizon in Einstein-Maxwell theory 5 3 Moduli spaces of electrovacuum deformations 8 3.1 AdS × S . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9 3.2 Kerr-Newman horizon . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14 4 Uniqueness of solutions with a cosmological constant 20 4.1 Static horizons . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20 4.2 Kerr-AdS horizon . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21 5 All three-dimensional solutions 24 A Transverse deformations of known solutions 25 A.1 Gaussian null coordinates for axisymmetric extreme black holes . . . . . . . . 25 A.2 Majumdar-Papapetrou solution . . . . . . . . . . . . . . . . . . . . . . . . . . 27 A.3 Kerr-Newman-AdS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28 A.4 Kerr-Newman-Melvin . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 29 A.5 Ernst solution and rotating generalisation . . . . . . . . . . . . . . . . . . . . 30 A.5.1 Ernst solution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30 A.5.2 Rotating generalisation . . . . . . . . . . . . . . . . . . . . . . . . . . . 31 1 Introduction and main results Extreme horizons in General Relativity in four and higher dimensions possess a number of remarkable rigidity properties. Most notably, it has been established that for stationary and axially symmetric spacetimes containing an extreme (Killing) horizon, the near-horizon geome- try possesses an enhanced isometry group SO(2, 1), in a wide range of Einstein-Maxwell-scalar type theories (which include various D = 4, 5 supergravity theories) and Einstein-Yang-Mills theory [2–5]. Furthermore, a number of explicit classiﬁcations of near-horizon geometries have been derived assuming axial or supersymmetry (mostly in D = 4, 5), see [1] for a review. In addition, general restrictions on the topology of extreme horizons have been established [4,6]. These are all steps towards the general black hole classiﬁcation program, which is a major open problem in higher dimensional General Relativity. Near-horizon classiﬁcations are only possible due to the following special property of ex- treme horizons. The Einstein equations restricted to an extreme horizon reduce to a set of geometric equations purely for data intrinsic to the horizon; the extrinsic data on the horizon decouples if and only if the horizon is extreme. Indeed, this is why one can deﬁne a precise notion of the near-horizon geometry for a spacetime containing an extreme horizon. The Ein- stein equations thus reduce to a problem of Riemannian geometry on a spatial section of the horizon which is blind to the exterior spacetime. This great simpliﬁcation is responsible for the above rigidity and classiﬁcation results. However, it also highlights a major diﬃculty: given a 2 near-horizon geometry how do you reconstruct the possible corresponding extreme black hole solutions (should they exist)? Nevertheless, the rigidity of near-horizon geometries provides the key boundary conditions required to establish uniqueness and classiﬁcation theorems for extreme black holes. This has been accomplished for D = 4 Einstein-Maxwell black holes [7–10], D = 5 stationary and biaxisymmetric vacuum black holes [9], supersymmetric black holes to D = 4, N = 2 supergravity [11], and most recently for supersymmetric and biaxisymmetric black holes in D = 5 minimal supergravity [12]. These works all combine global constraints on the spacetime arising from asymptotic ﬂatness and the assumed special symmetry (axial or supersymmetry), together with the near-horizon classiﬁcation. Despite these advances, it is desirable to develop a ‘quasi-local’ approach which does not rely on global spacetime assumptions or special symmetry structures. It is well known that a given near-horizon geometry may not arise as the near-horizon limit of any black hole, or may arise as the near-horizon limit of distinct black hole solutions. It is therefore of interest to study the more general inverse problem: what are the possible extreme black holes with a given near-horizon geometry? In a previous paper we initiated a systematic study of this question in vacuum gravity by introducing the notion of inﬁnitesimal transverse deformations of an extreme horizon [17]. These deformations are solutions to the linearised Einstein equations in the background near- horizon geometry (itself a solution due to extremality). This revealed that the moduli space of such deformations, for horizons with compact cross-sections, is ﬁnite dimensional – there cannot be ‘too many’ solutions. Furthermore, by classifying axially symmetric solutions, we established uniqueness of transverse deformations of the extreme Kerr horizon, thereby ex- tending the known rigidity result for the near-horizon geometry [13,16], without invoking any global assumptions on the spacetime. The purpose of this paper is to extend our study of transverse deformations of extreme horizons to Einstein-Maxwell theories. In fact, following the vacuum theory [17], it has been already shown that the moduli space of transverse deformations to extreme horizons with compact cross-sections is ﬁnite dimensional in a large class of Einstein-Maxwell-scalar theo- ries [18, 19]. In this paper we will go further and explicitly determine all axisymmetric defor- mations in D = 4 Einstein-Maxwell theory. The results (and analysis) are more complicated than in the vacuum theory. First, we recall that near-horizon geometries in this theory are known to be unique: the general static solution is AdS × S and the general axisymmetric solution is given by the extreme Kerr-Newman horizon [13,16]. Our main results are contained in the following theorems: Theorem 1 The moduli space of smooth, static and axisymmetric, transverse deformations of the AdS ×S near-horizon geometry in Einstein-Maxwell theory is 2-dimensional. It contains the extreme Reissner-Nordstro¨m solution, the extreme Ernst solutions and a special case of the extreme Kerr-Newman-Melvin solution. Theorem 2 The moduli space of smooth, axisymmetric, transverse deformations of the ex- treme Kerr-Newman horizon in Einstein-Maxwell theory is 3-dimensional. It contains the extreme Kerr-Newman solution, the extreme Kerr-Newman-Melvin solution and the extreme rotating Ernst solution. 3 Remarks. 1. The extreme Kerr-Newman-Melvin solution is a 3-parameter family of regular station- ary and axisymmetric solutions to Einstein-Maxwell theory, constructed by immersing a Kerr-Newman solution in an external magnetic ﬁeld, see e.g. [27–30]. It occupies a 2-dimensional region of the 3-dimensional moduli space of transverse deformations. It contains a special case in which the near-horizon geometry and deformation are static. 2. The extreme Ernst solution is a 2-parameter family of regular static and axisymmetric so- lutions to Einstein-Maxwell theory, given by balancing a charged C-metric in an external electric or magnetic ﬁeld, see e.g. [32]. In the case of vanishing acceleration parameter it reduces to the Reissner-Nordstr¨om solution. The Ernst solution has a near-horizon geometry isometric to AdS × S [31] (as it must by the above mentioned uniqueness theorem). We ﬁnd that although the dimensionality matches, this solution does not ﬁll all parts of the moduli space of static deformations. 3. A rotating generalisation of the Ernst solution can be constructed from a charged rotating C-metric held in equilibrium by an external magnetic ﬁeld [33]. It has a near-horizon geometry isometric to that of the Kerr-Newman black hole [33] (again this is guaranteed by the near-horizon uniqueness theorem). It occupies a 3-dimensional region of the 3- dimensional moduli space. 4. We also found a 1-parameter family of axisymmetric deformations of AdS × S that do not preserve staticity. The above results show that the deformations are more general than those corresponding to the Reissner-Nordstr¨om and Kerr-Newman solutions. Hence, we ﬁnd that the ‘local’ no- hair theorem established in vacuum gravity [17] does not persist in Einstein-Maxwell theory, although the physical interpretation of the extra parameters is essentially clear: background electromagnetic ﬁeld and acceleration. Indeed, since we do not impose any global assumptions such as asymptotic ﬂatness, regular solutions corresponding to black holes in external elec- tromagnetic ﬁelds are naturally captured in our classiﬁcation. Having said this, as noted in the remarks above, not all regions of the moduli space are occupied by the known solutions, so it would be interesting to determine if these extend to other solutions or if there is some obstruction to extending these linearised solutions to higher order. We also consider the case of a pure cosmological constant. Uniqueness of axisymmetric near-horizon geometries has been previously established [16]. Our main result is: Theorem 3 Any smooth, axisymmetric, marginally trapped, transverse deformation of the extreme Kerr-AdS horizon corresponds to an extreme Kerr-AdS black hole. This is interesting because black hole uniqueness theorems for solutions with a cosmological constant have not been established, and indeed may be violated. The violations of uniqueness are expected for solutions with a single Killing ﬁeld (a combination of the stationary and axial symmetries) [20,21], so our result is perhaps not surprising. To this end, it would be interesting to extend our analysis to deformations which do not preserve the axial symmetry. On the other hand, for axisymmetric solutions it is possible that the no-hair theorem for Kerr-AdS is indeed true; our results are consistent with this and may be taken as weak evidence for this possibility. 4 It is worth recalling that gravitational perturbations of the near-horizon geometry of the extreme Kerr(-AdS) black hole and of AdS × S have been previously considered [22–25]. These works consider dynamical perturbations. In contrast, our transverse deformations are perturbations which are non–dynamical since by construction they preserve the structure of the extreme Killing horizon. Naturally, it would be interesting to extend our results to Einstein-Maxwell solutions with a cosmological constant. Preliminary work on this suggests the analysis and results are (even) more complicated in this case. Due to the lack of solution generating techniques, black hole solutions in external electromagnetic ﬁelds with a cosmological constant are not known. Nev- ertheless, the classiﬁcation of near-horizon geometries of extreme horizons has been also solved in this case [16]. Hence investigating their transverse deformations may provide some evi- dence for the existence (or nonexistence) of putative dS/AdS black hole solutions in external electromagnetic ﬁelds. Finally, as a simpler application of our general formalism, we consider three dimensional Einstein-Maxwell theory with a negative cosmological constant. This theory admits charged generalisations of the (extreme) BTZ black hole [34–37]. In this case we show that one can easily determine all transverse deformations of the possible extreme horizons in the theory. This paper is organised as follows. In Section 2 we derive the linearised Einstein-Maxwell equations for a general transverse deformation of an extreme horizon in D-dimensional space- time. In Section 3 we determine the general axisymmetric deformations of AdS × S and more generally of the extreme Kerr-Newman horizon. In Section 4 we prove uniqueness of deformations of static extreme horizons with a cosmological constant and we prove uniqueness of axisymmetric deformations of the extreme Kerr-AdS horizon. In Section 5 we determine all deformations of extreme horizons in three-dimensional Einstein-Maxwell theory with a neg- ative cosmological constant. In the Appendix we give a general method for computing the transverse deformation arising from an axisymmetric extreme black hole solution and apply it to the known examples. 2 Near an extreme horizon in Einstein-Maxwell theory Let (M, g) be a D-dimensional spacetime containing a smooth degenerate Killing horizon H of a complete Killing ﬁeld n possessing a cross-section S (a (D − 2)-dimensional spacelike submanifold transverse to n). In the neighbourhood of such a horizon we introduce Gaussian null coordinates (GNC) and the associated near-horizon geometry, see e.g. [1] for more details. The spacetime metric in these coordinates takes the form a 1 2 a b r F (r, x)dv + γ (r, x)dx dx , (1) g = 2dv dr + rh (r, x)dx + a ab where the horizon is at r = 0, the vector ﬁeld ℓ = ∂ is transverse and geodesic, the normal Killing ﬁeld n = ∂ , and (x ) are coordinates on a cross-section S. Degeneracy of the horizon is equivalent to g = O(r ). From this form of the metric, it is easy to see there exists a well vv deﬁned notion of a near-horizon geometry. a a For any ε > 0, consider the diﬀeomorphism φ : (v, r, x ) → (v/ε, εr, x ) and deﬁne the 1-parameter family of metrics ∗ a 1 2 a b g(ε) ≡ φ g = 2dv dr + rh (εr, x)dx + r F (εr, x)dv + γ (εr, x)dx dx . (2) a ab 5 (0) The near-horizon geometry is deﬁned as the ε → 0 limit of g(ε), which we denote by g . Smoothness of the metric guarantees it exists and it is given by (0) (0) (0) a 1 2 (0) a b g = 2dv dr + rh (x)dx + r F (x)dv + γ (x)dx dx , (3) a ab (0) (0) (0) where F = F| , h (x) = h | , γ (x) = γ | , are a function, 1-form and Riemannian r=0 a r=0 ab r=0 ab (0) metric on S. In particular, γ (x) is the induced metric on S. ab The transverse deformation of an extreme horizon, introduced in [17], is deﬁned as the ﬁrst (1) d variation of g(ε) at ε = 0, i.e. g ≡ g(ε)| . This is equivalent to a linear perturbation of ε=0 dε the near-horizon geometry and is explicitly given by (1) (1) 3 (1) 2 2 (1) a a b g = r F (x)dv + 2r h (x)dvdx + rγ (x)dx dx , (4) a ab (1) (1) where γ = ∂ γ | etc. The quantity γ (x) is (twice) the extrinsic curvature of S with r ab r=0 ab ab respect to the null normal ℓ. As for any linearised perturbation, diﬀeomorphism invariance (1) (1) (0) implies the gauge freedom g → g −L g , where ξ is a vector ﬁeld. The most general such (0) ξ which preserves the form of the deformation (and is not an isometry of g ) is [17] 2 (0) a a 1 1 1 ξ = f∂ + r h D f ∂ − rD f ∂ , (5) v r a 2 4 2 (0) where D is the metric connection of γ and f(x) is a smooth function on S. Thus f param- ab eterises a supertranslation on the null surface H. In terms of this the gauge transformation rules for the ﬁrst order data are (1) (1) (0) γ → γ + D D f − h D f (6) a b b) ab ab (a (0) (0) (0) (1) (1) 1 (0) 1 b 1 (0) b 1 (0) b 1 b h → h − F D f − (D h )D f − h h D f + (D h )D f + h D D f a a b a a a a a 2 4 b 4 b 2 4 b (1) (1) 1 a (0) (0) (0) F → F + (D f) D F − h F . We emphasise that the Gaussian null chart is ﬁxed by a choice of cross-section S and coordinates on S. The above gauge freedom emerges as we are eﬀectively requiring the metric to be of Gaussian null form only to ﬁrst order in the aﬃne parameter r. Now suppose (M, g) is solution to the D-dimensional Einstein-Maxwell-Λ equations σ ρσ R = 2F F − g F F + Λg (7) νµ σµ νµ ρσ νµ (D − 2) d ∗ F = 0 , dF = 0 (8) where F is the Maxwell 2-form. In Gaussian null coordinates, a smooth Maxwell ﬁeld takes the general form a a 1 a b F = Ψ(r, x)dv ∧ dr + rW (r, x)dv ∧ dx + Z (r, x)dr ∧ dx + B (r, x)dx ∧ dx . (9) a a ab Deﬁne the 1-parameter family of Maxwell ﬁelds F(ε) ≡ φ F where φ is the diﬀeomorphism deﬁned above. The near-horizon limit of the Maxwell ﬁeld is the ε → 0 limit of F(ε). By smoothness of F its near-horizon limit exists and is given by (0) (0) (0) (0) a a b B (x)dx ∧ dx , (10) F = Ψ (x)dv ∧ dr + rW (x)dv ∧ dx + ab In general the F term does not admit a near horizon limit; however since on a Killing horizon 0 = va μ ν ab R K K | = 2γ F F | , smoothness implies we must have F = rW for some smooth W (r, x). μν N va vb r=0 va a a 6 (0) (0) (0) where Ψ = Ψ| , W = W | , B = B | are a function, 1-form and 2-form on S. r=0 a r=0 ab r=0 ab The near-horizon Bianchi identity reduces to (0) (0) (0) W = D Ψ , dB = 0. (11) (0) The 2-form B is the Maxwell ﬁeld induced on S. ab The 1-parameter family (g(ε),F(ε)) must also satisfy the Einstein-Maxwell equations. In (0) (0) particular, the near-horizon limit (g ,F ) satisﬁes the Einstein-Maxwell equations, resulting (0) (0) (0) (0) (0) in geometric equations for the horizon data (F , h , γ , Ψ , B ) intrinsic to S. The near- ab ab horizon Einstein equations are (0) (0) (0) (0) (0)c (0) (0) (0) (0) (0)2 (0)2 1 2 1 R = h h − D h + Λγ + 2B B + γ Ψ − γ B (12) (a a ac ab b b) ab b ab ab 2 D−2 D−2 (0) 1 (0)2 1 a (0) D−3 (0)2 1 (0)2 F = h − D h + Λ − 2 Ψ − B (13) 2 2 D−2 D−2 and the near-horizon Maxwell equation is (0) (0) (0) b (0)b D − h Ψ + D − h B = 0 . (14) a ba The classiﬁcation of solutions to these equations has been extensively studied [1]. In particular, for D = 4 all static solutions and all axially symmetric solutions can be determined [5,7,13–16]. Now, we deﬁne the transverse deformation of the Maxwell ﬁeld as the ﬁrst variation of the dF(ε) (1) Maxwell ﬁeld F ≡ | . In GNC this reads ε=0 dε (1) (1) (1) 2 (1) a (1) a a b F = rΨ (x)dv ∧ dr + r W (x)dv ∧ dx + Z (x)dr ∧ dx + rB (x)dx ∧ dx . (15) a a 2 ab (1) (1) where Ψ = ∂ Ψ| etc are deﬁned as above, except Z = Z | instead. The linearised r r=0 a r=0 (1) Bianchi identity dF = 0 reduces to (1) 1 (1) (1) (1) W = dΨ , B = dZ . (16) The diﬀeomorphism invariance implies the linearised Maxwell ﬁeld transforms simultaneously (1) (1) (0) with the linearised metric as F → F −L F , where ξ is the vector ﬁeld (5). This implies the ﬁrst order data transforms as (1) (1) (0) (0) (0) a Ψ → Ψ + D Ψ − 2h Ψ D f (0) (1) (1) 1 (0) b Z → Z + Ψ D f − B D f . (17) a a ab d νµ The linearised Maxwell equation is (g (ε)∇ (ε)F (ε))| = 0. µ νρ ε=0 dε After some lengthy calculations we ﬁnd that the linearised Einstein-Maxwell equations (1) (1) reduce to a pair of linear PDEs for ﬁrst order data (γ , Z ), deﬁned on the background ab (0) (0) (0) (0) horizon (S, γ , h , Ψ , B ), given by ab ab (1) (0) (1) (0) (1) (1) (1) 1 (1) 3 (0)c 3 (1) (0)c c 1 (0)2 0 = Δ γ + D D γ + h D γ − h D γ − h D γ + h D γ − h γ L (a b) c b) (a ab ab (a b)c (a b)c ab 2 2 2 2 (0) (1) (0) (1) (0) (0) (0) 1 (0) (1) (0)c c (0) (1)cd 2 (0)e (1)cd + h h γ + D h γ − D h γ + 2B B γ − γ B B γ (a a b ac bd ab c de 2 b)c (a b)c D−2 (1) (0) (1) (0) (1) (0)c (0)2 2 (0)2 (1) (0) 4 (0) (1) −2Ψ γ + γ Ψ γ + 4Ψ D Z − γ Ψ D · Z + 4D Z B (a c ab ab ab D−2 b) D−2 (a b) (0) (1) (0) (1) (0) (0) (0) (1) (0) 4 (0)cd (0) (1)c (1)c (0)c + γ D Z B − 4Ψ h Z − 4Z h B + 4Z D B − 4Z B h [c (a ab d] (a b) (a b)c b)c (a b)c D−2 (0) (0) (1) (0) 4 (1) c (0) 4 (0)cd + γ Z D Ψ + γ B Z h , (18) ab c ab [c d] D−2 D−2 7 (1) (1) (0)c (1) (0) (0) 1 2 (1)cd where Δ γ = − D γ +R γ −R γ is the Lichnerowicz operator of (S, γ ), and ab ab (a b)c acbd ab (0) (1) (0) 2 (1) (1)b (0)b (0)b (1) (0) (1) (0)bc (1) 0 = D Z − R Z + h D Z − 3h D Z − h D · Z + 4B B Z a b a a a c ab b ab (0) (0) (0)2 (1) (0) (1) (0)2 (1) (1)b (1) (0) (1)bc −4Ψ Z − 2F Z + 2h Z + 2Z D h − Z D · h − B D γ [b c a a a a ab a] (1) (0) (1) (1) (0) b (1)bc (0) (1)bc (0) (0)b (0)b (0)c +Ψ D γ − γ D B − γ B h − γ W − γ B h ab ca ab c ab ab c (1)d (0) (1)d (0) (0) (0)bc (0)bc b (1) (0) (1) (0) (0) (1) ˆ ˆ −γ C B − γ C B + B D γ − Ψ D γ + h Ψ γ , (19) bc da ab cd ab 2 (1)c (1) (1) (1) (1) 1 (0)cd (1) (1) where C = γ (D γ +D γ −D γ ). The rest of the ﬁrst order data (F , h , Ψ ) a b d a ab 2 db ad ab is then determined algebraically by (1) (1) (1) (1) 1 (0)b 1 b 1 (1) 1 (0) (1) (0)b (0) (1) h = h γ − D γ + D γ − h γ + 2 B Z + Ψ Z (20) a ab ab a a b a 2 2 2 4 (1) (1) (1) (0)a (1) 1 a (1) 1 (0)a (0)b 1 (0)(a b) F = h h − D h − h h γ + h D γ (21) a a ab ab 3 3 6 (1) 1 (a (0)b) 1 (0) (1) 1 (0)a (1) (0) (1) + D h γ − F γ − h D γ − h γ ab a 6 6 12 4 D−3 (0) (1) 2 D−4 (0) (0)a (1) 2 D−4 (0)a (1) − Ψ Ψ − Ψ h Z + W Z a a 3 D−2 3 D−2 3 D−2 (1) (1) (0) (0) (0) 2 (0)[ab] 2 (0)cd (1)ab − B B + 2Z h + γ B B γ [ab] [a b] [ac] [bd] 3(D−2) 3(D−2) (1) (0)a (1) a (1) (1) (0) Ψ = −h Z + D Z − γ Ψ . (22) a a Equations (18), (20), (21) correspond to the Einstein equations, whereas (19) and (22) corre- spond to the Maxwell equations (although simpliﬁed by combining them). Notice that (18) is automatically traceless, so the number of independent PDEs given by (18) and (19) is the same (1) (1) as the number of degrees of freedom (γ , Z ), once the gauge freedom (6), (17) is accounted ab for (recall this is parameterised by one function f). Crucially, (18) and (19) are linear elliptic (1) (1) (once gauge ﬁxed) PDEs for the ﬁrst order data (γ , Z ) deﬁned on the background of the ab (0) (0) (0) (0) horizon geometry (S, γ , h , Ψ , B ). By application of standard results for Fredholm op- ab ab erators we deduce: The moduli space of transverse deformations of a near-horizon geometry of an extremal horizon with compact cross-sections in D-dimensional Einstein-Maxwell-Λ theory is ﬁnite dimensional. This result was established for vacuum gravity in [17] and subsequently generalised to a large class of Einstein-Maxwell-dilaton theories in [19]. 3 Moduli spaces of electrovacuum deformations In D = 4 Einstein-Maxwell theory, an essentially complete understanding of the possible near- horizon geometry of extreme horizons with compact cross-sections has been achieved. In this section we will set the cosmological constant Λ = 0. The general static near-horizon geometry is AdS ×S [7], whereas the general axisymmetric near-horizon geometry is that of the extreme Kerr-Newman black hole [13, 16]. We will determine the complete moduli space of transverse deformations which preserve the axisymmetry of these near-horizon geometries. 8 2 3.1 AdS × S This near-horizon geometry of course arises as the near-horizon limit of the extreme Reissner- Nordstr¨om solution. It can be written as 2 2 r dx (0) 2 2 2 2 g = − dv + 2dvdr + r + (1 − x )dφ (23) 2 2 r 1 − x (0) F = − dv ∧ dr + q dx ∧ dφ , 2 2 2 where the electric q and magnetic q charges satisfy q + q = r > 0. The metric in the e m e m + round brackets is just the unit S written in polar coordinates, so −1 < x < 1 and φ is 2π periodic and x = ±1 are the usual coordinate singularities corresponding to the poles of S . We will consider axisymmetric transverse deformations to this near-horizon geometry, i.e., we assume the ﬁrst order data is invariant under the axial Killing ﬁeld ∂ . The gauge freedom for such deformations must be generated by an axisymmetric function f(x) [17]. The gauge transformations (6) and (17) are (1) (1) ′ ′′ γ → γ − f (x) + f (x) xx xx 1 − x (1) (1) γ → γ xφ xφ (1) (1) 2 ′ γ → γ − x(1 − x )f (x) φφ φφ (1) (1) ′ Z → Z − f (x) x x 2r q (x − 1) (1) (1) m Z → Z − f (x) . (24) φ φ 2r (1) Thus γ is gauge invariant. Using the above transformations we may also deﬁne the following xφ gauge invariant variables (1) 2 (1) Q = q Z + q (1 − x )Z 1 e m (1) (1) 2 2 2 (1) 2 2 Q = x (1 − x ) γ + x(1 − x )γ + (2x − 1)γ xx φφ φφ (1) (1) 2 (1) Q = 2q xZ − 2q x(1 − x )Z + γ . (25) 3 m e φ x φφ It is convenient to express the linearised Einstein-Maxwell equations in terms of these gauge invariant quantities. It is important to note that Q , Q , Q are all globally deﬁned functions 1 2 3 on S which vanish at the poles x = ±1. This is easy to see by writing the Q in terms of the 2 2 globally deﬁned vector ﬁelds m = ∂ and X = (1 − x )∂ which both vanish at the poles . It φ x (1) can also be shown that γ is globally deﬁned and vanishes at the poles [17]. xφ The linearised Maxwell equations (19) and Einstein equations (18) reduce to 4 ODEs (two from the Maxwell equations and two from the Einstein equations, due to the latter being 2 2 (1) (1) 2 (1) For example Q = x γ (X, X) + xX[γ (m, m)] + (2x − 1)γ (m, m). 9 traceless). Writing these in terms of the gauge invariant quantities we ﬁnd h i (1) 3 2 2 2 ′′ 0 = 2q x (1 − x ) 1 − x γ − 1 − x Q + 2Q (26) m 1 xφ 1 2 2 ′′ ′ 2 2 2 ′ 2 − q x(1 − x ) (−xQ + 2Q ) + 2(1 − x )(2x − 1)Q + x(1 − x )Q + 2(3x − 1)Q e 3 2 3 3 2 h i h i ′′ ′ (1) (1) (1) 2 2 2 2 0 = 1 − x (1 − x )γ + 2x (1 − x )γ + 2(1 − x )γ (27) xφ xφ xφ 2 ′ + 4(1 − x )Q + 8xQ h i (1) 3 2 2 2 ′′ 0 = 2q x (1 − x ) (1 − x )γ − (1 − x )Q + 2Q (28) e 1 xφ 1 2 2 ′′ ′ 2 2 2 ′ 2 + q x(1 − x ) (−xQ + 2Q ) + 2(1 − x )(2x − 1)Q + x(1 − x )Q + 2(3x − 1)Q m 3 2 3 3 2 2 ′ 2 2 2 ′ 2 2 0 = −x(1 − x )Q + 2(1 − 3x )Q − 2x(1 − x ) Q + 2(1 − x )(1 − 3x )Q . (29) 2 3 2 3 Note that equations (26) and (28) imply that the coeﬃcients of q and q in each must vanish e m 2 2 separately (since q + q 6= 0) and are thus equivalent to e m h i (1) 2 2 ′′ 0 = 1 − x γ − 1 − x Q + 2Q (30) xφ 1 2 2 ′′ ′ 2 2 0 = x(1 − x ) (−xQ + 2Q ) + 2(1 − x )(2x − 1)Q 3 3 2 ′ 2 +x(1 − x )Q + 2(3x − 1)Q . (31) (1) Thus the pairs of variables (γ , Q ) and (Q , Q ) decouple, each satisfying a pair of ODEs. 1 2 3 xφ (1) Let us ﬁrst consider the equations for (γ , Q ) which are (30) and (27). We can obtain an xφ h i (1) expression for (1 − x ) γ in terms of Q and its derivatives from (30), and substitute this xφ (1) expression into (27) to solve for γ in terms of Q and its derivatives only xφ 2 2 ′′′ 2 ′ (1 − x ) Q + 2(1 − x )Q + 4xQ (1) 1 1 1 γ = . (32) xφ −2 (1 − x ) Substituting this back into the ﬁrst equation (30) gives a fourth order equation for Q (4) 1 2 ′′′ ′′ 0 = − (1 − x )Q + 2xQ − 2Q , (33) 1 1 1 which has the general solution C 3 C 2 2 1 2 Q = x + x −2x + (1 − x ) log[1 − x] − (1 − x ) log[1 + x] + C + C x , (34) 1 3 4 6 8 where the C ’s are constants. Smoothness of Q and the endpoints x = ±1 requires C = 0. i 1 2 Then (32) gives 4 2 2 C (2x − 3x + 3) + 12C x + 6C (1 + x ) (1) 1 3 4 γ = , (35) xφ −6(1 − x ) which is not smooth at the endpoints in general. Imposing smoothness at x = ±1 then yields the constraints C = 0 and C = −6C . Thus by relabelling the constant C → K , the general 3 1 4 4 1 (1) solutions for Q and γ are simply xφ Q = K x(1 − x ) (36) 1 1 (1) γ = 2K (1 − x ) . (37) xφ 10 Now let us consider the equations for (Q , Q ) which are (31) and (29). It is easy to see 2 3 that the Q terms can be eliminated by adding (29) to (31), resulting in 2 ′′ 0 = (1 − x )Q + 2Q , (38) whose general solution which is smooth at x = ±1 is Q = K (1 − x ) , (39) 3 2 where K is a constant. Substituting this into (29) gives the following equation for Q : 2 2 2 2 ′ 2 3 0 = 2(1 − 3x )Q − x(1 − x )Q + 2K (1 − x ) , (40) 2 2 which has the general solution 2 2 2 2 2 Q = −K (1 − x ) + K x (1 − x ) (41) 2 2 3 where K is another constant. (1) To summarise, the general smooth solution for γ , Q , Q , Q is given by (37, 36, 41, 39). 1 2 3 xφ Using (25), we may now invert this general solution for the gauge invariant variables, to obtain (1) (1) the general solution for individual metric components γ and Maxwell ﬁeld components Z . ab Because of the gauge freedom, one of them will be a free smooth function. Regularity at the poles implies that without loss of generality, we can write (1) γ = (1 − x ) (m + xg(x)) , (42) φφ where m is a constant and g(x) is some smooth function. Then the gauge transformation for (1) γ is simply g(x) → g(x) − f (x). We then ﬁnd φφ 2 2 2 ′ 3 4 m − K + x (K + K ) − x (1 − x )g (x) + x g(x) − K x 2 2 3 3 (1) γ = xx 2 2 x (1 − x ) q (m − K ) + 2K q x + q xg(x) e 2 1 m e (1) Z = 2 2 2(q + q )x e m q (K − m) + 2K q x − q xg(x) (1) m 2 1 e m Z = (1 − x ) . (43) 2 2 2(q + q )x e m The solution (43) must be smooth for all −1 < x < 1. Therefore, to avoid the pole at x = 0 we must set K = m. Then, the general solution is K + xg(x) (1) ′ γ = + K − g (x) xx (1 − x ) (1) γ = 2K (1 − x ) xφ (1) γ = (1 − x ) (K + xg(x)) φφ 2K q x + q g(x) 1 m e (1) Z = 2 2 2(q + q ) e m 2K q x − q g(x) (1) 1 e m Z = (1 − x ) . (44) 2 2 2(q + q ) e m 11 2 Observe that near the poles x → 1 we have (1) (1) (1) (1) 2 −2 2 2 γ = γ (1 − x ) + O(1) , γ = O(1 − x ) , γ = O(1 − x ) , (45) xx φφ xφ φφ (1) so this family of deformations γ is indeed smooth tensor ﬁeld on S [17]. In summary, we have ab found a three parameter K , K , K family of smooth axisymmetric transverse deformations of 1 2 3 AdS × S . Since these linear deformations are deﬁned only up to an overall scale, there are only two independent physical deformations. Let us now consider the conditions for the deformation to correspond to a marginally trapped surface (MTS); see [17] for a full discussion of extreme MTS. The mean transverse expansion, which is a gauge invariant, is given by Z Z Z 1 1 (1) 2 (1) 2 2 γ = 2πr γ dx = 2π 2K + K (1 − x ) + (x − 1)g(x) dx 2 3 S −1 −1 = 8π K + , (46) (1) so the MTS condition γ > 0 is satisﬁed if K + > 0 . (47) For the extreme Reissner-Nordstr¨om black hole solution it is easy to check that K = K = 0, K = 2r . (48) 1 3 2 + Thus, any solution with K = K = 0 which obeys the MTS condition is equivalent to a 1 3 positive multiple of the extreme Reissner-Nordstr¨om solution. Now for static spacetimes, the normal Killing ﬁeld n is hypersurface orthogonal i.e. n∧dn = 0 everywhere. To ﬁrst order in Gaussian null coordinates this is equivalent to [26] (1) (0) (1) (1) (1) (0) dh = h ∧ h and dF = 2F h . (49) Therefore, we will call a transverse deformation static if it obeys these equations. For the case at hand these simplify to (1) (1) dh = 0 and dF = 0 . (50) With the general solution (44), we ﬁnd that (20) and (21) imply K x − g(x) 2K x(1 − x ) 3K + K 3 1 2 3 (1) (1) dx + dφ , F = . (51) h = 2 2 4 2r r 3r + + + Therefore our deformation is static if and only if K = 0. We deduce that there is a 2-parameter family (K , K ) of static transverse deformations. What do these correspond to? 2 3 It is useful to note that in order to compute the parameters K , K that specify a transverse 2 3 deformation of a known static solution, it is suﬃcient to compute Q which only depends on (1) γ . In the Appendix we give a general method for doing this. ab There are, of course, several known static and axisymmetric solutions to Einstein-Maxwell theory with extreme horizons. The most obvious is the Majumdar-Papapetrou multi-black 12 2 hole solution. The near-horizon geometry of each horizon is AdS × S . In the Appendix we compute the corresponding ﬁrst order deformation and we ﬁnd that it is indistinguishable from the Reissner-Nordstr¨om solution, i.e., K > 0 and K = 0. Thus, perhaps surprisingly, multi- 2 3 black holes are not visible at ﬁrst order in the GNC; in other words, the extrinsic curvature of the horizon is unchanged by the presence of another black hole. Another well known solution is the Reissner-Nordstr¨om solution in an external electric or magnetic ﬁeld (a Melvin universe). However, there are no static solutions which are smooth in this family. Both the electric Reissner-Nordstr¨om in the electric Melvin background and the magnetic Reissner-Nordstr¨om in the magnetic Melvin background suﬀer from conical singu- larities. The conical singularities may be avoided for e.g. an electric Reissner-Nordstr¨om in a magnetic Melvin background (or vice-versa), however this solution is no longer static and is a special case of the general Kerr-Newman-Melvin solution (discussed in the next section). So what is the interpretation of the parameter K ? In fact there is another smooth static and axisymmetric spacetime in Einstein-Maxwell theory with a extreme horizon : the extreme 2 2 Ernst solution is a two-parameter family (e, b) with e > 0 and e b < 4 corresponding to a static accelerating black hole with electric (or magnetic) charge e held in equilibrium by a uniform electric (or magnetic) ﬁeld b (see Appendix) . Its near-horizon geometry is AdS × S [31], as must be the case from the near-horizon uniqueness theorem, with 2 2 2 e(4 + b e ) r = . (52) 2 2 4(4 − b e ) Computing the ﬁrst order transverse deformation and the corresponding gauge invariant vari- ables (see Appendix), we ﬁnd it takes the form of our general solution with 2 2 3 2 3 2 2 3 2e(4 + b e ) 12b e (4 + b e ) K = 0, K = , K = . (53) 1 2 3 2 2 3 2 2 4 (4 − b e ) (4 − b e ) Thus the deformation in this case has both K and K nonvanishing. For b → 0 this reduces to 2 3 the data for the Reissner-Nordstr¨om black hole, as it should. The scale invariant combination 3 2 2 K + 4 + b e = ≥ 1 , (54) 2 2 K 4 − b e where the inequality follows from the fact the RHS is a monotonic function in the domain 2 2 e b < 4. This shows that not all marginally trapped deformations with K 6= 0 correspond to an Ernst solution. It would be interesting to determine if there are any black hole solutions which occupy the remaining part of the moduli space. What about deformations with K = 0? In fact, the extreme Kerr-Melvin solution admits a special case in which the near-horizon geometry is static (see Appendix). The deformation corresponding to this has K = K = 0 and K > 0. Therefore, although the extreme Kerr- 1 2 3 Melvin solution is stationary and never static, in this special case it gives rise to a static deformation of AdS × S with K = 0. For this solution staticity is broken at second order 2 2 in the GNC expansion. Conversely, any marginally trapped, static deformation with K = 0 corresponds to such a Kerr-Melvin solution. We thank Gary Gibbons for pointing out this solution. 13 Therefore, we have established Theorem 1 stated in the Introduction. It is worth empha- sising that above we also found a one-parameter family (K ) of deformations which do not preserve staticity. We will discuss their interpretation in the next section where we analyse deformations of the more general extreme Kerr-Newman horizon. 3.2 Kerr-Newman horizon The horizon data in this case is [16] (see also Appendix) 2 2 2 2 2 2 ρ dx (1 − x ) a + r dφ + + (0) γ = + (55) 2 2 1 − x ρ 2 2 2 2ar (1 − x ) a + r 2a x (0) h = − dx + dφ 2 4 ρ ρ + + 4 4 2 2 2 4 a x + a r (6x − 4) + r + + (0) F = − , 2 2 2 a q x − 2aq r x − q r e m + e (0) + Ψ = , 2 2 2 2 2 a + r a q x + 2aq r x − q r m e + m + + (0) B = − dx ∧ dφ , 2 2 2 2 2 2 2 2 2 2 where ρ = r + a x and r = a + q + q . We will assume a > 0 and q + q > 0. The + + + m e m e horizon metric is deﬁned for −1 < x < 1, φ is 2π periodic, and extends to a smooth metric on S where the endpoints x = ±1 correspond to the ﬁxed points of the axial symmetry. We will assume the ﬁrst order deformation is also invariant under the axial Killing ﬁeld ∂ , in which case the gauge transformation function f must also be axisymmetric [17]. The gauge transformation rules (6) and (17) are 2 2 2 ′ x a (2x − 1) + r f (x) (1) (1) ′′ γ → γ + f (x) − (56) xx xx (1 − x ) ρ 2 2 2 ′ ar (1 − x ) a + r f (x) (1) (1) + γ → γ − xφ xφ 2 2 2 ′ x (1 − x ) a + r f (x) (1) (1) γ → γ − φφ φφ 2 2 2 ′ q r − a x + 2aq rx f (x) e m (1) (1) Z → Z − x x 2ρ 2 2 2 2 2 2 ′ (1 − x ) a + r a q x + 2aq r x − q r f (x) m e + m (1) (1) + + Z → Z − . φ φ 2ρ It is straightforward to check the following variables are invariant under the above gauge 14 transformation: 2 2 2 2 2 2 (1) Q = (a + r )(1 − x ) −2aq r x + q (r − a x ) Z 1 e + m + + x (1) 2 2 2 2 2 2 +(r + a x ) 2aq r x + q (r − a x ) Z m + e + + φ (1) 2 2 3 2 2 2 (1) 2 2 2 2 3 Q = (a + r ) x (1 − x ) γ + x(1 − x )(r + a x ) γ + xx + φφ (1) 2 2 2 2 2 2 2 2 2 −(r + a x ) r (1 − 2x ) + a x (5x − 6) γ + + φφ (1) 2 2 2 2 2 2 2 2 2 2 2 2 (1) Q = −(q + q )(r + a x ) γ + 2ar (a + r )(1 − x ) 2aq r x + q (r − a x ) Z 3 + m + e e m + xφ + + x (1) 2 2 2 2 2 2 +2ar (r + a x ) 2aq r x + q (a x − r ) Z + e + m + + φ (1) (1) 2 2 2 2 2 2 Q = (a + r ) x γ − ar (r + a x )γ . (57) 4 + + xφ + φφ Note that these do not reduce to the variables used in the static case (25) in the static limit 4 static a → 0. However, they are closely related: in particular, as a → 0 we have Q → r Q , Q → 1 2 + 1 6 static r Q . Nevertheless, as in the static case one can show these Q are also globally deﬁned + 2 and vanish at the poles of S . In terms of these gauge invariant variables, the Maxwell equations (19) are equivalent to (by taking linear combinations of the x and φ components) 2 2 2 2 2 2 2 2 ′′ 2 2 2 2 2 2 2 2 ′ 0 = (a + r ) (1 − x )(r + a x ) Q − 2a (a + r ) x(1 − x )(r + a x )Q + + 1 + + 1 2 2 2 4 4 2 2 2 2 2 2 2 2 2 2 2 ′ +2(a + r ) −r + 3a x + a r (1 + x ) Q + (a + r ) (1 − x )(r + a x )Q + + + + + 3 2 2 2 2 2 4 4 2 2 2 2 −2a (a + r ) x(1 − x )Q + 2(r − a )(r + a x ) Q (58) 3 4 + + + 2 2 2 2 2 2 2 2 4 ′′ 2 2 2 2 2 2 2 3 2 2 2 ′ 0 = (a − r )x (1 − x ) (r + a x ) Q + 2(a − r )x(1 − x ) (r + a x ) (2a x − r )Q + + 4 + + + 4 2 2 2 2 2 2 2 4 4 2 4 4 2 2 4 +2(a − r )(1 − x )(r + a x ) r − 2r x + 3a x + 2a r x Q + + + + + 2 2 2 3 2 2 2 2 2 2 ′′ 2 2 2 2 4 2 2 2 2 2 ′ −(a + r ) x (1 − x ) (r + a x ) Q + 2a (a + r ) x (1 − x ) (r + a x )Q + + 3 + + 3 2 2 2 3 2 4 4 2 2 2 2 2 2 −2(a + r ) x (1 − x )(−r + 3a x + a r + a r x )Q + + + + 2 2 2 2 2 2 2 ′ 2 2 2 2 2 2 2 +ar (a − r )x(1 − x )(r + a x ) Q + 2ar (a − r )(3x − 1)(r + a x ) Q + + 2 + + 2 + + 2 2 2 2 2 3 2 2 2 2 2 ′ 4 2 2 2 2 4 2 2 +4a r (a + r ) x (1 − x ) (r + a x )Q − 8a r (a + r ) x (1 − x ) Q . (59) + + + 1 + + The xφ component of the Einstein equation (18) gives 2 2 2 2 2 2 3 ′′ 2 2 2 2 2 2 2 2 2 4 ′ 0 = x (1 − x ) (r + a x ) Q − 2x(1 − x )(r + a x ) (r − a x + 2a x )Q + 4 + + 4 2 2 2 2 2 4 2 4 4 4 2 2 2 2 2 ′ −2(r + a x ) 2a r x (1 + x ) + (r + a x )(3x − 1) Q + ar x(1 − x )(r + a x )Q 4 + + + + + 2 2 2 2 2 2 2 2 3 2 2 2 2 ′ 2ar (3x − 1)(r + a x )Q + 4(a + r ) x (1 − x )(r + a x )Q + 2 + + + 1 2 2 2 4 2 2 2 2 2 2 3 2 +8(a + r ) x (r + a x )Q − 4(a + r ) x (1 − x )Q , (60) 1 3 + + + 2 2 and the xφ component plus (a + r ) times the xx component of the Einstein equation gives 2 2 2 2 2 3 ′′ 2 2 2 2 3 ′ 0 = (1 − x ) (r + a x ) Q − 2x(1 − x )(r + a x ) Q + 4 + 4 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 ′ −2(r + a x ) r (3x − 1) + 2a − a x (1 − x ) Q + 2(a + r )(1 − x ) (r + a x )Q + + + + 3 2 2 2 2 2 2 2 2 2 2 2 2 ′ −4a (a + r )x(1 − x ) Q + 4(a + r ) x(1 − x )(r + a x )Q + + + 1 2 2 2 4 4 2 2 2 2 2 2 2 +8(a + r ) x (r + a x ) + a r + a r x (2x − 1) Q . (61) + + + + The remaining components of the Einstein equation are automatically satisﬁed due to the fact it is traceless. 15 2 2 2 The Maxwell equation (59) can be further simpliﬁed by subtracting from it (a − r )(r + + + 2 2 a x ) times the Einstein equation (60) resulting in 2 2 2 2 2 2 2 2 2 ′′ 2 2 2 2 2 2 2 2 2 ′ 0 = −(a + r ) (1 − x ) (r + a x ) Q + 2a (a + r ) x(1 − x ) (r + a x )Q + + 3 + + 3 2 2 2 2 4 2 2 2 2 2 2 2 4 4 2 2 2 2 3 ′ −2(a + r ) (1 − x )(r + a x − a r + 3a r x )Q + 2(a − r )(1 − x )(r + a x ) Q + + + + + + 4 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 2 4 4 2 ′ +4(a − r )(a + r ) x(r + a x ) Q + 4(a + r ) (1 − x )(r + a x )(r − a x )Q + + + + + + 1 2 2 2 4 2 6 6 4 2 4 2 4 2 −8(a + r ) x(a r − r + a x + a r − 2a r x )Q . (62) + + + + + Therefore, the Einstein-Maxwell equations are equivalent to the four equations (58), (60), (61), (62); note only (60) contains Q terms, so let us focus on the remaining three equations ﬁrst. We can rearrange (58) and write 2 2 2 ′ 2 2 2 2 2 2 2 2 2 ′′ (r + a x )Q − 2a x Q = −(a + r ) (1 − x )(r + a x ) Q + 3 + + 1 2 2 2 (a + r ) (1 − x ) 2 2 2 2 2 2 2 2 ′ +2a (a + r ) x(1 − x )(r + a x )Q + + 1 2 2 2 4 4 2 2 2 2 2 2 −2(a + r ) (−r + 3a x + a r + a r x )Q + + + + 4 4 2 2 2 2 +2(a − r )(r + a x ) Q , (63) + + and express (62) as 2 2 2 2 2 2 2 2 2 2 2 ′ 2 0 = −(a + r ) (1 − x ) (1 − x )(r + a x ) (r + a x )Q − 2a x Q + + + 3 2 2 2 2 2 ′ 2 2 2 2 2 −2a x(1 − x ) (r + a x )Q − 2a x Q + 2(r + a x ) Q + ... , (64) 3 3 + 3 + where ... denote the terms not involving Q . Thus, using (63) we can substitute for (r + 2 2 ′ 2 a x )Q − 2a x Q in (64) to ﬁnd an expression for Q in terms of Q and Q terms. It turns 3 3 1 4 out that the Q terms all cancel and we are left with simply 2 2 2 2 2 2 ′′′ 2 2 2 2 ′ 2 2 2(1 − x )Q = (1 − x ) (r + a x )Q + 2(1 − x )(r + a x )Q + 4(a + r )x Q . (65) 3 1 + 1 + 1 + Now, using (65) we can substitute for all the Q terms in (58) to get (4) 2 2 2 2 2 2 2 2 2 3 2 2 2 0 = 4(a − r )(1 − x )(r + a x )Q − (a + r )(1 − x ) (r + a x )Q + + + + 1 2 2 2 2 2 2 2 ′′′ 2 2 2 2 2 2 2 ′′ +2(a + r )x(1 − x ) (r + a x )Q − 4(a + r )(1 − x ) (r + a x )Q + + 1 + + 1 2 2 2 2 2 2 ′ 2 2 2 2 2 −4(a + r )x(1 − x )(r + a x )Q − 8(a + r )(r + a x )Q , (66) + + 1 + + 2 2 2 2 which allows us to solve for Q in terms of Q (note r − a = q + q > 0 by assumption). 4 1 + e m We may obtain another equation for Q and Q only by taking the linear combination 1 4 2 2 2 −2(1 − x ) (58) +(a + r ) (61) to eliminate the Q terms, resulting in 2 2 2 2 2 2 2 2 ′′ 0 = −2(a + r )(1 − x ) (r + a x ) Q + + 1 2 2 2 2 2 2 2 2 2 2 ′ +4(a + r )x(1 − x )(r + a x )(r + 2a − a x )Q + + + 1 2 2 2 2 2 2 2 2 2 4 2 2 2 2 2 3 ′′ +4(a + r ) (r + r x − 3a x + 5a x )Q + (1 − x ) (r + a x ) Q + + + + 4 2 2 2 2 3 ′ 2 2 2 2 3 −2x(1 − x )(r + a x ) Q − 2(1 − x )(r + a x ) Q . (67) + 4 + 16 Eliminating Q from the equations (66) and (67) then gives us a remarkably simple sixth order equation for Q : (6) (5) (4) 2 2 2 2 0 = (1 − x ) Q − 12x(1 − x )Q − 6(1 − 5x )Q . (68) 1 1 1 The general solution to (68) which is regular at x = ±1 is simply the polynomial 3 2 ˜ ˜ Q = A x + A x + A x + A (69) 1 1 2 3 4 ˜ ˜ where A , A , A , A are constants (we have discarded the log terms in the most general solution 1 2 3 4 to the diﬀerential equation (68) since they are not regular at the poles). We can use (66) to determine the general solution to Q . Regularity at the poles forces ˜ ˜ A = −A and A = −A , thus giving 3 1 4 2 2 2 2 2(a + r )(A + A x)(1 − x ) 2 1 Q = −(A + A x)(1 − x ) , Q = − , (70) 1 2 1 4 2 2 r + a x and from (65) we deduce 2 2 2 Q = 2(A r − A a x)(1 − x ) . (71) 3 1 2 Finally, from (60) we ﬁnd 2 2 3 2 2 2 2 2 2 8A a(a + r )x (1 − x ) 2A r (a + r )(1 − x ) 1 2 + + + 2 2 2 Q = − + A x (1 − x ) , (72) 2 3 r a where A is an integration constant. In order to able to take the a → 0 static limit to compare with our analysis of the static solutions, without loss of generality we will deﬁne the constant A = A /a. We have therefore found the general solution for the gauge invariant variables 2 2 Q , Q , Q , Q parameterised in terms of constants A , A , A . 1 2 3 4 1 2 3 Now we can invert the equation (57) to get the general solution for the individual compo- (1) (1) (1) nents of γ and Z . Let us choose γ to be the arbitrary function corresponding to the ab φφ gauge freedom. We may write (1 − x ) (m + xh(x)) (1) γ = (73) φφ 2 2 2 2 (r + a x ) where m is some constant and h(x) is some smooth function. Under a gauge transformation 2 2 3 ′ 2 (56) we ﬁnd h(x) → h(x) − (a + r ) f (x)/ρ , so the function h(x) may be thought of as + + parameterising the gauge freedom. The solution for Q implies 2 2 2 2 2 amr (1 − x ) − 2A a(a + r )(1 − x ) 2A (1 − x ) 1 + 2 (1) + γ = − + xφ 2 2 2 2 2 2 2 2 2 2 2 (a + r )(r + a x ) (a + r ) x(r + a x ) + + + + ar (1 − x )h(x) + . (74) 2 2 2 2 2 2 (a + r ) (r + a x ) + + To avoid the pole at x = 0 we must set 2 2 2A (a + r ) m = . (75) 17 In summary, the general solution is a three parameter family parameterised by constants A , A , A , and a smooth function h(x) corresponding to the gauge freedom, where 1 2 3 2 2 2 2 2A (r − 2a + 3a x ) 8A ax A 1 + 3 (1) γ = + + xx 2 2 2 2 2 2 2 2 2 3 r (a + r ) r (a + r ) (1 − x ) (a + r ) + + + + + 2 2 2 2 ′ 2 2 2 2 (1 − x )(r + a x )h (x) − x(r − 3a + 4a x )h(x) + + 2 2 3 2 (a + r ) (1 − x ) 2 2 ar h(x) − 2A (a + r ) + 1 (1) + 2 γ = (1 − x ) xφ 2 2 2 2 2 2 (a + r ) (r + a x ) + + 2 2 2A (a + r ) + r xh(x) (1) 2 + γ = (1 − x ) φφ 2 2 2 2 r (r + a x ) 2 2 2 3 2 2 3 q (2ar + ar x − a x ) + q (3a r x − r x) a(aq x − q r ) e m + e m + (1) + + + Z = A + A 1 2 2 2 2 4 2 2 2 2 2 2 4 2 2 r (r − a )(a + r ) (r + a x ) r (r − a )(r + a x ) + + + + + + + 2 2 2 2aq r x + q (r − a x ) m + e + h(x) 2 2 3 2 2 2 2(a + r ) (r + a x ) + + 2 2 2 3 2 2 3 q (2ar + ar x − a x ) − q (3a r x − r x) m e + (1) + + + 2 Z = −A (1 − x ) 4 2 4 2 2 2 r (r − a )(r + a x ) + + 2 2 2 2aq r x − q (r − a x ) a(aq x + q r ) m e + e + m 2 2 −A (1 − x ) + (1 − x )h(x) . 2 2 2 2 2 2 2 2 2 2 2 2 2 r (r − a )(r + a x ) 2(a + r ) (r + a x ) + + + + + (76) It can be checked that near the poles x → 1 we have (1) (1) (1) (1) 2 −2 2 2 γ = γ (1 − x ) + O(1) γ = O(1 − x ) γ = O(1 − x ) (77) xx φφ xφ φφ (1) so the ﬁrst order quantity γ is a indeed smooth tensor ﬁeld on S [17]. In the static limit a → ab 4 3 0 our solution reduces to the general static solution with K = −A /r , K = 2A /r , K = 1 1 2 2 3 + + A /r . We will now impose that the deformation is such that S is a MTS. The appropriate gauge (1) invariant quantity is the mean expansion Γγ where Γ > 0 is the unique (up to scale) (0) −1 (0) (0) ˜ ˜ function such that that h = Γ h − d log Γ and D · h = 0 [17]. For the extreme Kerr- 2 2 2 Newman horizon one can set Γ = ρ /(r + a ) [16] and we ﬁnd + + 8π (1) 2 2 Γγ = (A + 6A r (a + r )) . (78) 3 2 + 2 2 3 3(a + r ) S + Therefore, the MTS condition for our ﬁrst order deformation requires 2 2 A + 6A r (a + r ) > 0 . (79) 3 2 + The extreme Kerr-Newman black hole solution gives rise to a transverse deformation (see Appendix). It is easily checked this takes the form of our solution with 2 2 2 2 2 2 2 2 A = 0, A = (q + q )(a + r ), A = 12a r (a + r ) , (80) 1 2 3 + e m + + 18 which clearly satisﬁes our MTS condition. It is worth noting that in order to extract the parameters A , A , A of a transverse deformation for a given known solution, it is suﬃcient 1 2 3 (1) to compute Q which in fact only depends on γ . We give a general recipe for doing this in ab the Appendix. Another family of stationary and axisymmetric solutions with an extreme horizon is given by the Kerr-Newman-Melvin solution, i.e. a Kerr-Newman black hole in an external elec- tric/magnetic ﬁeld. This is a rather complicated solution so it is helpful to consider a few instructive special cases. The extreme Kerr-Melvin solution is a 2-parameter family parameterised by (a˜, b), cor- responding to the rotation parameter a˜ and the external magnetic ﬁeld b (see Appendix). It’s near-horizon geometry is isometric to that of an extreme Kerr-Newman black hole with parameters 2 2 2 2 2 2 2 2 2 2 r = a˜ (1 + a˜ b ) , a = a˜ (1 − a˜ b ) . (81) A computation reveals that the ﬁrst order transverse deformation then takes our general form (see Appendix) with 7 4 4 3 A = A = 0, A = 48a˜ (1 + a˜ b ) , (82) 1 2 3 which clearly satisﬁes our MTS condition. Notice this is of a diﬀerent form to the deformation corresponding to the Kerr-Newman black hole (which has A 6= 0); thus at ﬁrst order it is possible to distinguish an external magnetic ﬁeld from intrinsic charge. Another interesting special case is the extreme Reissner-Nordstr¨om-Melvin solution which is also a 2-parameter family parameterised by (q˜, b), corresponding to the electric charge pa- rameter q˜ and external magnetic ﬁeld b (see Appendix). Its near-horizon geometry is again given by that of an extreme Kerr-Newman black hole with parameters 2 2 1 2 2 2 2 4 2 r = q˜ (1 + q˜ b ) , a = q˜ b , (83) and the corresponding ﬁrst order deformation is given by 3 2 2 4 4 2 2 A = 0, A = q˜ (4 − b q˜ )(b q˜ + 24b q˜ + 16), 1 2 3 2 8 2 2 4 4 2 2 2 A = b q˜ (b q˜ + 12)(b q˜ + 24b q˜ + 16) . (84) It may be veriﬁed that this obeys the MTS condition. The general extreme Kerr-Newman-Melvin is a much more complicated solution parame- terised by three-parameters (a˜, q˜, b). It’s near-horizon geometry has parameters [28–30] 1 2 2 2 1 2 2 2 r˜ (3a˜ + r˜ )b , a = a˜ − q˜r˜ b − a˜(a˜ + 3r˜ )b (85) r = r˜ + a˜q˜b + + + + + + + 4 4 2 2 where r˜ = a˜ + q˜ . We have veriﬁed that the corresponding deformation also has A = 0 + 1 and generically A , A 6= 0; indeed, this case interpolates between Kerr-Newman, Kerr-Melvin 2 3 and Reissner-Nordstr¨om-Melvin described above. It is worth noting that although the extreme Kerr-Newman-Melvin solution is a stationary and axisymmetric solution, it includes a two parameter special case for which the near-horizon geometry is static and hence isometric to the AdS × S solution. As noted above, in the static horizon limit a → 0 we have A → 2 1 −r K , so we deduce that this special case of Kerr-Newman-Melvin solutions can only lead to deformations of AdS × S with K = 0, i.e. they preserve staticity to ﬁrst order in GNC even 2 1 though the full solution is not static. 19 All known solutions discussed so far have A = 0. So what about the interpretation of the parameter A ? In fact, as in the static case, a more general class of spacetimes containing a smooth extremal horizon maybe be constructed. By applying a Harrison transformation to a rotating, dyonic C-metric one can construct a regular solution which corresponds to a rotating, dyonic, accelerating black hole held in equilibrium by a uniform magnetic ﬁeld [33]. This is a rotating version of the Ernst solution, so we will simply refer to it as the rotating Ernst solution. It has an extremal limit where the surface gravity of the horizon is zero. The resulting extreme solution is a four parameter family parameterised by (a˜, q˜, p˜, A, b), subject to a constraint coming from the removal of conical singularities at x = ±1. Here, (a˜, q˜, p˜) are the rotation parameter, electric and magnetic charges, A is the acceleration parameter, and b is the external magnetic ﬁeld. For simplicity we will only consider the special case a˜ = 0 given in [33]; for q˜ 6= 0 even this case is rotating due to the external magnetic ﬁeld and hence suﬃciently general for our pur- 2 2 poses. As shown in the Appendix, the regular solution can be parameterised by z = q˜ + p˜ and (A, b) subject to certain inequalities. Its near-horizon geometry is isometric to that of the Kerr-Newman black hole, as it must be by the near-horizon uniqueness theorems, with parameters (r , a) given by (187). The corresponding ﬁrst order deformation takes our general form with A , A , A given by (189); in particular note that generically all three constants 1 2 3 A , A , A are nonvanishing. For A = 0, this solution reduces to Reissner-Nordstr¨om-Melvin, 1 2 3 in which case A = 0 as found above. This provides an interpretation for the parameter A . 1 1 However, as in the static case, presumably the rotating Ernst solution does not occupy all parts of the moduli space, i.e. although this gives a three parameter family of deformations not all values of A , A , A may be realised. It would be interesting if there are new solutions 1 2 3 which ﬁll out the rest of the moduli space, or if there is some obstruction to promoting our linearised solutions to higher order in the GNC expansion in these regions of moduli space. We have now established Theorem 2 stated in the Introduction. It is worth noting that this general four parameter family of extreme rotating Ernst solutions discussed above will contain a three parameter family of solutions with a static near-horizon geometry. Presumably the corresponding deformations will give the non-static K 6= 0 deformations of the AdS × S 1 2 near-horizon geometry found in the previous section. 4 Uniqueness of solutions with a cosmological constant Here we will consider the vacuum case with a cosmological constant Λ. The possible static near-horizon geometries with compact cross-sections S are given by (0) (0) (0) (0) F = Λ, h = 0, R = Λγ , (86) ab ab 2 2 corresponding to dS ×S or AdS ×H if Λ > 0 and Λ < 0 respectively [14]. The most general 2 2 axisymmetric near-horizon geometry is given by that of Kerr-(A)dS [15]. We will determine the moduli space of transverse deformations of these near-horizon geometries. 4.1 Static horizons (1) (1) In this case, the gauge freedom (6) reduces to γ → γ + D D f and may be used to ﬁx a a b ab ab (1) gauge in which γ is a constant. This can always be done since it involves solving Poisson’s 20 (0) equation on a compact manifold S. For the sake of generality, we will assume that (S, γ ) is ab an n = D − 2 dimensional maximally symmetric space, so (0) (0) (0) (0) (0) R = K(γ γ − γ γ ), (87) abcd ac bd ad bc where Λ = K(n− 1). For D = 4, 5, so S is 2 or 3 dimensional, this is the only possibility since S is an Einstein space by equation (86). The linearised Einstein equation (18) reduces to (1) (1) − D γ˜ = −2nKγ˜ , (88) ab ab (1) (1) (0) 1 (1) where γ˜ = γ − γ γ is the traceless part of the deformation. ab ab ab n 2 First assume Λ > 0, so S is locally isometric to the round metric on S . Then, since −D (1) is a positive deﬁnite operator on a compact manifold, we must have γ˜ ≡ 0. Thus, the only ab (1) (0) n (1) deformations of dS ×S are given by γ = γ γ . These correspond to the Narai solution ab ab (this is the extreme limit of Schwarzschild de-Sitter in which the black hole and cosmological horizons are coincident). Now consider Λ < 0, so S is locally hyperbolic space H . The above argument no longer (1) works and we have to work a little harder. The Einstein equation states that γ˜ is an ab eigentensor of −D with eigenvalue 2n|K|. First consider n = 2, in which case S Σ where Σ is a Riemann surface of genus g ≥ 2. It is well known that in general n = 2 symmetric tensor harmonics are all derived from scalar harmonics on S. A basis for traceless symmetric tensor harmonics is given by (0) (0)c D D Y − γ D Y, D ǫ D Y , (89) a b (a c ab b) (0) (0) where ǫ is the volume form of (S, γ ) and Y are the scalar harmonics obeying −D Y = λY . ab ab It can be checked that (89) are eigentensors of −D both with eigenvalue λ−2nK. Comparing (1) to the Einstein equation we deduce that γ˜ must be a linear combination of the λ = 0 ab harmonics. However, the only λ = 0 scalar harmonics on a compact manifold are Y =constant, (1) so in fact we must have γ˜ ≡ 0. Thus, we also deduce that the only deformations of AdS ×Σ 2 g ab (1) (0) 1 (1) are given by γ = γ γ . These correspond to the extreme Schwarzschild-AdS -hyperbolic ab ab black hole. This argument fails for n > 2, since then one can have non-scalar derived tensor harmonics. Hence, the moduli space of deformations may be more complicated in this case, and need not correspond to the Schwarzschild-AdS-hyperbolic black hole. 4.2 Kerr-AdS horizon The extreme Kerr-AdS horizon data with Λ = −3g is given by [15] (see also Appendix) 2 2 2 2 ρ (r + a )(1 − x )Δ (0) + 2 + 2 γ = dx + dφ (90) 2 2 (1 − x )Δ ρ Ξ 2 2 2 2 2ar (r + a )(1 − x )Δ 2a x + x (0) h = dφ − dx , (91) 2 2 Ξρ ρ + + 21 2 2 2 2 2 2 2 2 2 −1 where ρ = r +a x , Δ = 1−a g x and Ξ = 1−a g and the parameters obey 0 < r < g x + + + and 1 + 3g r a = r . (92) 1 − g r The coordinate ranges are −1 < x < 1 and φ is 2π-periodic and the endpoints x = ±1 are coordinate singularities corresponding to the ﬁxed points of the axial symmetry (the poles of S ). We will consider axisymmetric deformations, in which case as noted earlier the function f parameterising the allowed gauge transformation must also be axisymmetric. It is convenient to introduce gauge invariant variables. It is straightforward to check that (1) (1) 2 2 2 4 2 4 2 2 2 2 2 X = (r + a )x[−r + a g x + a (−1 + g r (−1 + 2x ))]γ (x) + ar ρ Ξγ (93) + + + xφ + φφ is invariant under our gauge transformations (6). Further, X is smooth and vanishes at the poles x = ±1. Note that for g = 0 this is proportional to the gauge invariant variable (also (1) called X) used in the pure vacuum case [17]. One can then use this to eliminate γ in favour φφ of X and doing this for the xφ component of the linearised Einstein equation (18) gives 2x 2 ′ 2 4 2 2 2 (1 − x )X + [−4r + a g x (−1 + 5x ) 5Δ ρ 2 2 2 2 2 +a (−3 − x + 4g r (−1 + 2x )]X + Y = 0 , (94) where Y is another gauge invariant variable which is a complicated linear combination of ′ ′′ ′ (1) (1) (1) (1) γ , γ , γ , γ (its explicit form is unilluminating). Similarly, the φφ component of the xx xφ xφ xφ linearised Einstein equation reduces to an equation of the form ′′ ′ a X + a X + a X + Z = 0 (95) 1 2 3 where a , a , a are complicated functions of x and Z is another gauge invariant variable which 1 2 3 ′ ′′ ′ (1) (1) (1) (1) (1) is a linear combination of γ , γ , γ , γ , γ . Because the linearised Einstein equation is xx xx xφ xφ xφ automatically traceless the ﬁnal component is redundant. In the pure vacuum case g = 0 one can check that Z ∝ Y so these two variables are not independent; one may then eliminate Y to get a second order ODE for X as found in [17]. However, for g 6= 0 the variables Y and Z are independent and we must proceed diﬀerently, as follows. (1) (1) For g 6= 0 one can invert the deﬁnitions of Y and Z to solve algebraically for γ , γ xx xx ′ ′′ ′ (1) (1) (1) (1) (1) in terms of Y, Z, γ , γ , γ . Then imposing the ‘integrability condition’ γ = dγ /dx xx xx xφ xφ xφ results in a gauge invariant equation of the form ′ ′ b Y + c Z + b Y + c Z = 0 , (96) 1 1 2 2 where b , b , c , c are unsightly functions of x. 1 2 1 2 Using (94) and (95) to eliminate Y and Z in (96) ﬁnally gives the following third order ODE for X, ′′′ ′ α X + α X + α X = 0 (97) 1 2 3 22 where 3 3 2 4 4 2 2 2 2 α = 1 − x 1 − 3g r x − g r x + 1 (98) + + 2 2 2 4 4 2 2 2 2 4 4 4 α = −4 1 − 3g r 1 − x 1 − 3g r x − g r x + 1 3g r x + + + + 2 2 4 2 2 +g r x + 6x − 1 + 2x + 1 (99) 2 2 8 8 6 6 6 6 4 2 α = 8x 1 − 3g r 9g r x + 3g r 2x + 9x − 4x + 1 + + + 4 4 2 4 2 2 2 4 2 2 +g r x x + 18x − 1 + g r 3x − 2x − 1 − x − 2 . (100) + + Remarkably, one can ﬁnd the general solution to this equation in terms of elementary functions. The general solution which is smooth at x → ±1 is simply 2 2 2 2 2 A(1 − x )[4 − x (1 + 3g r ) ] X = − (101) 4 4 2 2 2 2 4[1 − 3g r x − g r (1 + x )] + + where A is an integration constant. The other functions Y and Z are then also determined by (1) (1) (94) and (95). Using X, Y, Z, one can then solve for the deformation γ , γ algebraically in xx φφ (1) terms of γ (x) (an arbitrary function reﬂecting the gauge freedom) and the constant A (i.e. xφ there are no further integration constants). Using this general solution one can then compute the gauge invariant mean expansion (1) Γγ . In this case we may take [15] Γ = (102) Ξ(r + a ) and we ﬁnd 2 7 πA (1 − g r ) (1) Γγ = (103) 2 6 2 2 3 2 2 2 2 2g r (1 + g r ) (1 − 3g r ) 1 + 3g r S + + + + so the MTS condition is simply A > 0. Now, using the ﬁrst order data for the extreme Kerr-AdS black hole (see Appendix), it is straightforward to verify that X takes precisely the above form (as it must) with 2 7 2 2 2 32g r (1 + g r ) 1 + 3g r + + + A = (104) 2 2 2 2 3 (1 − 3g r ) (1 − g r ) + + which indeed obeys the MTS condition. Since such linear deformations are only determined up to scale, we deduce that the general solution must be gauge equivalent to the ﬁrst order data of the extreme Kerr-AdS black hole. This establishes Theorem 3. Therefore, the uniqueness theorem established for the vacuum extreme Kerr horizon [17], persists with a cosmological constant. We emphasise that this result does not invoke any global assumption on the spacetime and hence is valid for both asymptotically AdS and locally AdS spacetimes. 23 5 All three-dimensional solutions −2 Three-dimensional Einstein-Maxwell theory with Λ = −2ℓ < 0 admits black hole solu- tions [34–37]. It is easy to completely classify near-horizon geometries with compact cross- sections S = S [1]. Being one-dimensional, S has no curvature and all tensors are scalars, (0) so we can introduce a periodic coordinate x ∼ x + 2πR so that γ = 1. Furthermore, the (0) Maxwell ﬁeld induced on S must vanish so B = 0. There are two classes of near-horizon solutions: (i) AdS × S ; (ii) locally AdS [1]. 2 3 1 (0) −2 (0) (0) −1 For AdS × S the horizon data is F = −2ℓ , h = 0, Ψ = ±ℓ . The linearised Einstein equation (18) is automatically satisﬁed, whereas the Maxwell equation (19) reduces to 2 (1) (1) d Z dγ 1 (0) − Ψ = 0 . (105) dx dx Integrating we ﬁnd the general solution (1) 1 (0) (1) Z (x) = a + Ψ γ − b , (106) (1) (1) where a, b are integration constants and periodicity of Z ﬁxes b = γ . Thus the 2πR S (1) deformation is parameterised by a constant a and an arbitrary function γ , reﬂecting the (1) (1) ′′ gauge freedom (6) which reduces to γ → γ + f . The remaining ﬁrst order data is (1) (0) (1) (1) 1 (0) 2 (1) 1 (0) h = 2Ψ Z , F = (Ψ ) b , Ψ = − Ψ b . (107) 3 2 The MTS condition is simply b > 0. The static charged extreme BTZ solution corresponds to (1) (1) a = 0 and γ a positive constant (so b = γ > 0). The rotating generalisation [35–37] also has 1 (1) the AdS ×S near-horizon geometry [1] and presumably corresponds to the a 6= 0, γ = b > 0 deformations. (0) (0) The locally AdS near-horizon geometry is given by the vacuum solution F = 0, h = (0) ±2/ℓ, Ψ = 0. The linearised Einstein equation (18) is again automatically satisﬁed, whereas the Maxwell equation (19) now reduces to 2 (1) (1) d Z dZ (0) (0)2 (1) − 3h + 2h Z = 0 . (108) dx dx (1) dZ Multiplying by the ﬁrst and third terms become total derivatives, and integrating this dx over S the boundary terms vanish (by periodicity), leaving (1) dZ dx = 0 . (109) dx (1) (1) This implies Z is a constant. Substituting back into (108) we deduce that Z = 0. There- (1) fore, the general deformation in this case is given by an arbitrary function γ and the remaining ﬁrst order data is (1) 1 (0) (1) (1) (1) h = h γ , F = 0, Ψ = 0 . (110) In fact this is the general vacuum deformation [17]. Therefore we ﬁnd that at ﬁrst order there are no electrovacuum deformations which are not vacuum. While charged black holes with an 24 1 AdS × S near-horizon geometry are known (as discussed above), we are not aware of any charged black holes with a locally AdS near-horizon geometry (see discussion in [1]). Indeed, such solutions may not exist and our result supports this possibility. In the vacuum case, the full nonlinear solution to the Einstein equation is known for arbi- (1) trary γ and is diﬀeomorphic to the extreme BTZ black hole (the diﬀeo is large, in the sense that the asymptotic Virasoro charges change) [38]. It would be interesting to ﬁnd the full non linear solution in the Einstein-Maxwell case. Acknowledgements JL would like to thank Gary Gibbons for useful comments. CL was supported by the Pol- ish National Science Centre grant No. 2015/17/B/ST2/02871. JL is supported by STFC [ST/L000458/1]. A Transverse deformations of known solutions In this Appendix we compute the transverse deformations corresponding to the known extreme black holes solutions. First, we present a general analysis for stationary and axisymmetric solutions. Then we apply it to various examples. A.1 Gaussian null coordinates for axisymmetric extreme black holes Consider a stationary and axisymmetric spacetime of the form BdR 2 2 2 2 2 ds = −R Adt + + W dx + X(dφ + Rωdt) (111) where the metric components are functions of (R, x). The surface R = 0 is a smooth extremal horizon with normal n = ∂ if 2 2 B − (c − c˜R) A = O(R ) , ω = + O(R) , (112) where b, c, c˜ are constants (assume c > 0). These conditions can be written as 2 2 B (x) = c A (x), B (x) = c A (x) − 2cc˜A (x), ω (x) = (113) 0 0 1 1 0 0 where B (x) = ∂ B| etc. Indeed, in terms of the new coordinates (V, ϕ) deﬁned by n R=0 t = V + + c˜log R, φ = ϕ + b log R (114) we have B − (c − c˜R) A 2 2 2 2 ds = A −R dV + 2(c − c˜R)dV dR + dR dR + W dx + X dϕ + RωdV + (b − (c − c˜R)ω) (115) 25 and hence the above conditions imply g = O(1) near R = 0 so that the spacetime metric Rµ is smooth and non-degenerate at R = 0. In particular R = 0 is an extremal Killing horizon. Its near-horizon geometry can be extracted by scaling (V, R) → (V/ǫ, ǫR) and letting ǫ → 0, giving 2 2 2 2 2 ds = A [−R dV + 2c dV dR] + W dx + X (dϕ + Rω dV ) , (116) 0 0 0 0 NH which takes the familiar form of a circle ﬁbration over AdS . We now ﬁnd Gaussian null coordinates for extremal horizons of the above form. The Killing ﬁelds are n = ∂ and m = ∂ . We will assume the existence of an axisymmetric cross-section S, t φ i.e. such that m is tangent to S. Then, we need to ﬁnd null geodesics γ(r) such that γ˙ · n = 1 and γ˙ · m = 0, where r is an aﬃne parameter synchronised so that r = 0 at the horizon. These give 1 ω t = − , φ = . (117) R A RA The null condition is then −1 2 2 ˙2 − A + BR + R Wx˙ = 0 , (118) which together with the geodesic equation for x and the initial conditions R(0, y) = 0, x(0, y) = y, x˙(0, y) = 0 (119) uniquely determines R(r, y) and x(r, y). We can develop the Taylor series in r for the solution. Using the null constraint we ﬁnd r (c˜A (y) − cA (y))r 0 1 2 2 R(r, y) = 1 − + O(r ) , x(r, y) = y + O(r ) . (120) 2 2 cA (y) 2c A (y) 0 0 Integrating for t, φ then gives c A (y) t = v + + c˜log r + f(y) + O(r), r ω A (y) c˜ω 0 1 0 φ = ψ + g(y) + cω log r + ω (y) − − + O(r ) , (121) 0 1 A (y) 2A (y) 2c 0 0 where v, ψ and f(y), g(y) are constant along the geodesics. The above deﬁnes a new chart (v, r, y, ψ) near the horizon. The Killing ﬁelds in this chart are n = ∂ , m = ∂ . In order for this to deﬁne a Gaussian null chart we need to impose v ψ g = g = 0. The latter condition is just ∂ · m = 0 which we have already imposed. The ry rψ r former is ˙ ˙ BRR BRR y y ∂ · ∂ = −R Att + + x˙x W = t + + O(r) , (122) r y y y y 2 2 R R where t = ∂ t etc, which vanishes at r = 0 if and only if y y cA f(y) = − c˜log A . (123) 2A Since ∂ is geodesic this is suﬃcient to guarantee it vanishes for r > 0. Hence, we have GNC. We have not fully ﬁxed the coordinates on the horizon yet. Indeed, g = X(φ + Rωt ) , (124) yψ y y 26 and requiring this to vanish on the horizon implies cω A ′ 0 g (y) = − . (125) Our coordinate change is now fully ﬁxed. We deduce the remaining components of the metric are R B 2 2 2 2 2 g = − R At + x W + X(φ + Rωt ) = W + O(r ), (126) yy y y y y 2 2 2 g = X, g = −R A + R Xω , g = RXω (127) ψψ vv vψ g = −R At + RXω(φ + ωRt ) . (128) yv y y y From this we can extract the horizon data ′ 2 ω X A −A + ω X 0 0 0 0 (0) 2 2 (0) 0 (0) 0 γ = W dy + X dψ , h = dψ − dy, F = . (129) 0 0 cA A c A 0 0 (1) The ﬁrst order data γ = ∂ γ | is r ab r=0 ab 1 1 X ω (1) (1) 0 (1) 1 γ = W , γ = X , γ = . (130) 1 1 yy ψψ yψ cA (y) cA (y) A 0 0 0 (1) Observe that for a static solution ω = 0 and hence γ = 0. yψ To apply the method to rotating black holes, we need to take account of the horizon rotating. Thus we let t = t and φ = φ + Ω t, so BdR 2 2 2 2 2 ˜ ˜ + W dx + X(dφ + Ωdt) (131) ds = −R Adt + where Ω ≡ −Ω + Rω. Then the Killing ﬁeld null on the horizon is n = ∂ + Ω ∂ . H ˜ H ˜ We now apply the above to work out the ﬁrst order data for several important examples. A.2 Majumdar-Papapetrou solution The Majumdar-Papapetrou solution is determined by an arbitrary harmonic function H on 3 3 R . For a black hole at the origin of R the solution is 2 −2 2 2 2 2 2 ℓ ds = −H dt + H (dR + R dΩ ), H = + h R Y (θ) (132) ℓ ℓ ℓ=0 where we assume H is axisymmetric (so the spacetime is). Thus the metric takes the above general form (111) with A = , B = (Q + RH) , (133) ˜ 2 (Q + RH) ℓ 2 2 where H = h R Y (θ). Hence A = 1/Q and c = Q , c˜ = −2Qh and therefore ℓ ℓ 0 0 ℓ=0 (0) 2 2 (1) 2 γ = Q dΩ , γ = 2Qh dΩ . (134) 2 2 Observe that the ﬁrst order data in this case depends only the monopole term. Therefore to ﬁrst order in GNC the multi-centred black hole solution is indistinguishable from a single black hole solution. 27 A.3 Kerr-Newman-AdS The Kerr-Newman-AdS solution, with Λ = −3g ≤ 0, in standard Boyer-Lindquist coordinates is given by 2 2 2 2 Δ a(1 − x ) ρ dr ρ 2 2 2 ds = − dt − dφ + dr + dx (135) 2 2 ρ Ξ Δ (1 − x )Δ r x 2 2 2 Δ (1 − x ) r + a + adt − dφ , ρ Ξ 1 a(1 − x ) 2 2 2 F = − q (r − a x ) + 2q rax dt ∧ dr + dr ∧ dφ (136) e m ρ Ξ 2 2 1 (r + a ) 2 2 2 + q (r − a x ) − 2q rax adt ∧ dx + dx ∧ dφ , m e ρ Ξ where 2 2 2 2 2 2 2 2 2 ρ = r + a x , Ξ = 1 − g a , Δ = 1 − g a x (137) 2 2 2 2 2 Δ = (r + a ) 1 + g r − 2mr + z . (138) The parameters m, a encode the mass and rotation, whereas q and q are electric and magnetic e m 2 2 2 charges respectively and z = q + q . The horizon is located at the biggest root r > 0 of e m Δ = 0. In the extreme limit the parameters obey 2 4 2 2 2 2 2 g r + a (1 + g r ) + (r + z ) + + + m = , (139) 2r 2 2 4 2 2 2 2 z = 3g r + 1 + g a r − a (140) + + and 2 2 2 2 2 Δ = (r − r ) [1 + g (r + 2rr + a + 3r )] . (141) r + + 2 2 2 The Kerr-Newman case in given by setting g = 0 in the above, in which case r = a + z . 2 2 2 2 2 The Kerr-AdS case is given by setting z = 0, in which case a = (1 + 3g r )/(1 − g r ) and + + 2 2 g r < 1. Writing the metric in our general form (111) and setting R = r − r we can read oﬀ the horizon data. One ﬁnds the constants are aΞ 2ar Ξ Ω = , ω = , (142) H 0 2 2 2 2 2 r + a (r + a ) + + 2 2 2 2 2 2 2r −a g + 4g r + 1 r + a c = , c˜ = − , (143) 2 2 2 2 2 2 2 2 1 + g a + 6g r (a g + 6g r + 1) and the horizon data is 2 2 2 2 2 a g + 6g r + 1 ρ + + A = (144) (a + r ) 2 2 2 2 2 2 ρ dx (1 − x ) a + r Δ dφ + + (0) γ = + , (145) 2 2 (1 − x ) Δ Ξ ρ 28 2 2 2 2 where ρ = r + a x . The ﬁrst order data is + + 2 2 2r a + r (1) γ = (146) xx (1 − x ) Δ ρ 3 2 2 2 2 2 2a x (1 − x ) a + r g r + 1 (1) + + γ = (147) xφ ΞΔ ρ 2 2 2 2 2 2 2r (1 − x ) a + r Δ a (2x − 1) + r (1) + x + + γ = . (148) φφ 2 6 Ξ ρ The near-horizon Maxwell ﬁeld and deformation can be computed using the coordinate change to GNC (120), (121). We ﬁnd 2 2 2 a q x − 2aq r x − q r e m + e (0) Ψ = (149) 2 2 2 2 2 a + r a q x + 2aq r x − q r m e + m (0) + + B = − (150) xφ Ξρ and 2 2 2 a[q (r − a x ) − 2aq r x] m e + (1) + Z = − (151) Ξρ 2 2 2 2 2 2 a (1 − x ) a + r a q x − 2aq r x − q r e m + e (1) + + Z = . (152) Ξρ A.4 Kerr-Newman-Melvin The Kerr-Newman-Melvin spacetime may be constructed by applying a Harrison transforma- tion to the Kerr-Newman metric, see eg. [27]. In particular, the extreme limit has been studied in detail in recent years [28–30]. The extreme solution is a three parameter family which de- pends on (a˜, q˜, b), where a˜, q˜ are the rotation and charge parameter of the seed Kerr-Newman solution and b parameterises the external magnetic ﬁeld. The extreme solution can be written as 2 2 2 2 Δ dr dx A|Λ| (1 − x ) 2 2 2 2 2 ds = |Λ| ρ − dt + + + (dφ − ωˆdt) (153) 2 2 A Δ 1 − x ρ where 2 2 2 2 2 2 2 2 2 2 Δ = (r − m) , ρ = r + a˜ x , A = (r + a˜ ) − Δa˜ (1 − x ) , (154) 2 2 the parameter m = a˜ + q˜ > 0 and Λ, ωˆ are functions that depend on the magnetic ﬁeld parameter b. The general solution is rather complicated. For simplicity we will consider two special cases. The Reissner-Nordstr¨om-Melvin solution is given by a˜ = 0 and arises as a special case of the rotating Ernst solution we give in the next section. If one includes magnetic charge in the seed Kerr-Newman, one gets conical singularities on the poles of horizon. Removing these again gives a three parameter family of solutions. We will not consider this solution. 29 The extreme Kerr-Melvin solution is given by m = a˜ and the functions 2 2 2 b A(1 − x ) a˜ 1 1 2 2 2 2 Λ = 1 + − ia˜ x 3 − x + (1 − x ) (155) 4 2 2 2 ρ ρ 2 4 a˜ ρ b 4 4 2 2 2 2 ωˆ = (1 − b a˜ ) − Δ + −8a˜rx (3 − x ) − 6a˜r(1 − x ) (156) 2 2 r + a˜ A 16 3 2 3 4 2 2a˜ (1 − x ) 4a˜ x 2 2 3 2 2 2 2 2 2 + [r(r + a˜ ) + 2a˜ ] + (r + a˜ )(3 − x ) − 4a˜ (1 − x ) A A where b is the magnetic ﬁeld parameter. Setting R = r − a˜ we may extract the constants 4 4 4 4 1 − a˜ b 1 − a˜ b Ω = , ω = , c = 2a˜ , c˜ = −2a˜ (157) H 0 2a˜ 2a˜ and the horizon geometry is that of extreme Kerr-Newman (see previous section with g = 0) with 2 2 2 2 2 2 2 2 2 2 r = a˜ (1 + a˜ b ) , a = a˜ (1 − a˜ b ) (158) 2 2 2 r + a x A = . (159) 4a˜ The ﬁrst order data is 3 2 2 2 2 2 2 4a˜ [(1 + a˜ b ) − 2a˜ b x ] (1) γ = (160) xx 2 2 2 2 (1 − x ) (r + a x ) 7 2 2 4 4 16a˜ x (1 − x ) (1 + a˜ b ) (1) γ = (161) φφ 2 2 2 (r + a x ) 5 2 4 4 4a˜ x (1 − x ) (1 − a˜ b ) (1) γ = . (162) xφ 2 2 (r + a x ) 2 2 Observe that for b = 0 this reduces to the extreme Kerr data. For a˜ b = 1 the near-horizon geometry is static so this gives a deformation of AdS × S ; in fact it is a static deformation, although the full solution is not static. A.5 Ernst solution and rotating generalisation A.5.1 Ernst solution The Ernst solution represents a static charged accelerating black hole held in equilibrium by an external ﬁeld, see e.g. [32]. This is given by " # 2 2 2 2 2 2 1 dr r dx˜ Pr (1 − x˜ )dφ 2 2 2 ds = D −Qdt + + + (163) 2 2 2 (1 + Arx˜) Q P (1 − x˜ ) D where in the extreme limit −2 2 2 2 2 Q = r (r − e) (1 − A r ), P = (1 + Aex˜) , (164) 2 2 2 b r (1 − x˜ )P D = (1 + ebx˜) + . (165) 4(1 + Arx˜) 30 The parameter e represents the electric (or magnetic) charge, b the external electric (or mag- netic) ﬁeld and A the acceleration. The black hole horizon is at r = e and there is an acceleration horizon at r = 1/|A|. We require the acceleration horizon to be outside the black 2 2 hole so e A < 1 and e < r < 1/|A|. The conical singularities at x˜ = ±1 can be simultaneously removed if A = (166) 2 2 1 + e b 2 2 and the period of φ is chosen appropriately (see below). The condition e A < 1 thus implies 2 2 e b < 4. For b = 0 this solution reduces to the extreme Reissner-Nordstr¨om solution. Setting R = r − e we may use the above general formulas to extract the horizon data and the ﬁrst order deformation. We ﬁnd the horizon data 2 2 2 2 2 2 2 2 2 4 (4 − b e ) e (4 + b e ) 2e(4 + b e ) A = , c = , c˜ = − (167) 2 2 2 2 2 2 4 16e (4 − b e ) (4 − b e ) dx (0) 2 2 2 γ = r + (1 − x )dφ , (168) 1 − x where 2 2 2 e(4 + b e ) r = , (169) 2 2 4(4 − b e ) and to reveal the round metric on the horizon we have changed coordinates to (x, φ) x − Ae 16 −1 x˜ = , φ = Z φ, Z = , (170) 2 2 2 1 − Aex (4 + b e ) where φ is 2π periodic. We then ﬁnd that the ﬁrst order data is 2 2 2 2 512e(1 − x )[4 − 4bex + b e (−1 + 2x )] (1) γ = (171) φφ 2 2 4 2 2 (4 − b e ) (4 + b e ) 2 2 3 2 2 2 2e(4 + b e ) [4 − 4bex − b e (−3 + 2x )] (1) γ = (172) xx 2 2 4 2 (4 − b e ) (1 − x ) (1) γ = 0 . (173) xφ (1) Notice that γ = 0 is due to the fact the solution is static. For b = 0 this reduces to xφ (1) 2 γ = 2e dΩ , corresponding to the Reissner-Nordstr¨om solution. A.5.2 Rotating generalisation A magnetised accelerating Kerr-Newman solution can be obtained by magnetising the accel- erating Kerr-Newman via the Ehlers-Harrison transformation. The seed extremal accelerating Kerr-Newman metric can be written as [32] h i 1 G(r) ρ 2 2 2 ds = − dt + a˜ 1 − x˜ dφ + dr ρ G(r) (1 + rx˜A) h i H(x˜) ρ 2 2 2 + r + a˜ dφ + adt + dx˜ , (174) ρ H(x˜) 31 where 2 2 2 2 ρ = r + a˜ x˜ , 2 2 2 G(r) := 1 − A r (r − m) , H(x˜) := 1 − x˜ (1 + Ax˜m) , 2 2 2 and m = a˜ + q˜ + p˜ > 0. The black hole horizon is located at r = m and there is also an acceleration horizon at r = r = . Thus for the acceleration horizon to be outside the black |A| 2 2 hole horizon, we must have r > m and so A m < 1. The coordinate ranges are m < r < r , A A −1 < x˜ < 1. For A 6= 0 this metric has conical singularities at x˜ = ±1 which cannot be simultaneously removed. For A = 0 the solution reduces to the extremal Kerr-Newman black hole. For simplicity we will set a˜ = 0; for q˜ 6= 0 this still leads to a rotating solution due to the presence of the external magnetic ﬁeld. The magnetised solution is given in [33] and takes the form 2 2 2 |Λ| G(r) r r 2 2 2 c2 ds = − dt + dr + dx˜ r G(r) H(x˜) (1 + rx˜A) r H(x˜) + dφ − ωˆdt (175) |Λ| (1 + rx˜A) where r H(x˜) 1 2 2 2 2 Λ = 1 + bx˜(p˜− iq˜) + b + (p˜ + q˜ )x˜ (176) 4 2 (1 + Arx˜) 3 2 2 2 2 2 2 2 2q˜b q˜b [(r − 2mr)(1 + Arx˜ + x˜ ) + x˜ (p˜ + q˜ )(1 − A r )] ωˆ = − + (177) r 2r(1 + Arx˜) and the parameter b encodes the background magnetic ﬁeld (so b = 0 reduces to the above seed). We ﬁnd it convenient to introduce the parameterisation q˜ = z sin α and p˜ = z cos α, so 2 2 that m = z and hence A z < 1. In general the solution has conical singularities on the axis of symmetry x˜ = ±1. For b 6= 0, simultaneous removal of these can be achieved by setting 2 2 4 4 A(16 + 24b z + b z ) cos α = (178) 2 2 2 2 4b(1 + A z )(4 + b z ) and ﬁxing the period to be 2 2 2π 16(1 + A z ) Δφ = , Z = . (179) 2 2 4 4 Z 16 + 24b z + b z This gives a 3-parameter family of solutions parameterised by (z, A, b). The parameter ranges can be obtained as follows. From the expression for cos α, we have 2 2 2 2 2 4 16b 2 [(4 + b z ) − 16b A z ] − A 2 2 2 (4+b z ) sin α = . (180) 2 2 2 2 16b (1 + A z ) We ﬁnd that this is simpler than solving for A as was done in [33]. 32 2 2 The condition A z < 1 mentioned above implies 2 2 2 2 2 4 2 2 2 2 2 2 2 2 (4 + b z ) − 16b A z > (4 + b z ) − 16b z = (4 − b z ) ≥ 0 , (181) and hence sin α ≥ 0 implies 16b A ≤ . (182) 2 2 2 (4 + b z ) For A = 0 the solution reduces to Reissner-Norstrom-Melvin with parameters q˜, b (note cos α = 2 2 0 in this case so z = ±q˜). For A = ±4b/(4 + b z ) we have q˜ = 0 (note cos α = ±1), which corresponds to the static Ernst solution. Setting R = r − m we may write this solution in our general form (111) to extract the near-horizon data. We ﬁnd the constants 2 2 bq(4 + b z ) 4 1 2 Ω = , ω = bq b + (183) H 0 2z z 2 2 z 1 + A z c = , c˜ = − , (184) 2 2 2 2 2 1 − A z (1 − A z ) thus establishing regularity of the event horizon. The horizon data can be written as 2 2 2 2 2 2 2 r + a x (1 − x )(r + a ) + + (0) 2 2 γ = dx + dφ (185) 2 2 2 2 2 1 − x (r + a x ) 2 2 2 (1 − A z ) 2 2 2 A = (r + a x ) , (186) where we have deﬁned the constants 2 2 2 2 2 2 4 4 2 2 2 2 2 z [(4 + b z ) − 16A b z ] z [16b − A (4 + b z ) ] 2 2 r = , a = , (187) 2 2 2 2 2 2 2 2 2 2 16(1 − A z ) (1 + A z ) 16(1 − A z ) (1 + A z ) and changed coordinates to x˜ + Az x = , φ = Zφ . (188) 1 + Azx˜ This reveals the horizon geometry is isometric to that of the extreme Kerr-Newman (as it must be!), with parameters r , a as given above. Note that the parameter ranges ensure that r > 0 and a ≥ 0. Note that a static near-horizon geometry a = 0 corresponds to the static Ernst solution. The ﬁrst order deformation takes our general form with 7 2 2 4 4 2 2 2 2 2 2 Ar z (4 − b z ) (b z + 24b z + 16) (A (b z + 4) − 16b ) A = (189) 2 2 5 2 2 2 2 256a(1 − A z ) (A z + 1) (b z + 4) 4 2 2 4 4 2 2 2 2 2 4 4 4 z (4 − b z ) (b z + 24b z + 16) (b (8z − 16A z ) + b z + 16) A = 2 2 5 2 2 2 2 256r (1 − A z ) (A z + 1) (b z + 4) 8 4 4 2 2 2 6 6 4 4 2 2 2 4 4 2 2 3z (b z + 24b z + 16) (−A (b z + 44b z + 48b z + 64) + 2b (b z + 16b z + 48)) A = 2 2 6 2 2 2 2 2048(1 − A z ) (A z + 1) (b z + 4) 33 The MTS condition 6 2 2 4 4 3 3z (16 + 24b z + b z ) 2 2 A + 6A r (r + a ) = > 0 (190) 3 2 + 2 2 5 2 2 3 2048(1 − A z ) (1 + A z ) is satisﬁed. For A = 0 the above gives the ﬁrst order data for the Reissner-Nordstr¨om-Melvin solution given in the main text in equation (84). References [1] H. K. Kunduri and J. Lucietti, Living Rev. Rel. 16 (2013) 8 doi:10.12942/lrr-2013-8 [arXiv:1306.2517 [hep-th]]. [2] H. K. Kunduri, J. Lucietti and H. S. Reall, Class. Quant. Grav. 24 (2007) 4169 doi:10.1088/0264-9381/24/16/012 [arXiv:0705.4214 [hep-th]]. [3] P. Figueras, H. K. Kunduri, J. Lucietti and M. Rangamani, Phys. Rev. D 78 (2008) 044042 doi:10.1103/PhysRevD.78.044042 [arXiv:0803.2998 [hep-th]]. [4] J. Lucietti, Class. Quant. Grav. 29 (2012) 235014 doi:10.1088/0264-9381/29/23/235014 [arXiv:1209.4042 [gr-qc]]. [5] C. Li and J. Lucietti, Class. Quant. Grav. 30 (2013) 095017 doi:10.1088/0264- 9381/30/9/095017 [arXiv:1302.4616 [hep-th]]. [6] M. Khuri, E. Woolgar and W. Wylie, arXiv:1804.01220 [hep-th]. [7] P. T. Chrusciel and P. Tod, Commun. Math. Phys. 271 (2007) 577 doi:10.1007/s00220- 007-0191-9 [gr-qc/0512043]. [8] A. J. Amsel, G. T. Horowitz, D. Marolf and M. M. Roberts, Phys. Rev. D 81 (2010) 024033 doi:10.1103/PhysRevD.81.024033 [arXiv:0906.2367 [gr-qc]]. [9] P. Figueras and J. Lucietti, Class. Quant. Grav. 27 (2010) 095001 doi:10.1088/0264- 9381/27/9/095001 [arXiv:0906.5565 [hep-th]]. [10] P. T. Chrusciel and L. Nguyen, Annales Henri Poincare 11 (2010) 585 doi:10.1007/s00023- 010-0038-3 [arXiv:1002.1737 [gr-qc]]. [11] P. T. Chrusciel, H. S. Reall and P. Tod, Class. Quant. Grav. 23 (2006) 2519 doi:10.1088/0264-9381/23/7/018 [gr-qc/0512116]. [12] V. Breunho¨lder and J. Lucietti, arXiv:1712.07092 [hep-th]. [13] J. Lewandowski and T. Pawlowski, Class. Quant. Grav. 20 (2003) 587 doi:10.1088/0264- 9381/20/4/303 [gr-qc/0208032]. [14] P. T. Chrusciel, H. S. Reall and P. Tod, Class. Quant. Grav. 23 (2006) 549 doi:10.1088/0264-9381/23/2/018 [gr-qc/0512041]. 34 [15] H. K. Kunduri and J. Lucietti, J. Math. Phys. 50 (2009) 082502 doi:10.1063/1.3190480 [arXiv:0806.2051 [hep-th]]. [16] H. K. Kunduri and J. Lucietti, Class. Quant. Grav. 26 (2009) 055019 doi:10.1088/0264- 9381/26/5/055019 [arXiv:0812.1576 [hep-th]]. [17] C. Li and J. Lucietti, Class. Quant. Grav. 33 (2016) no.7, 075015 doi:10.1088/0264- 9381/33/7/075015 [arXiv:1509.03469 [gr-qc]]. [18] M. Dunajski, J. Gutowski and W. Sabra, Class. Quant. Grav. 34 (2017) no.4, 045009 doi:10.1088/1361-6382/aa5992 [arXiv:1610.08953 [hep-th]]. [19] A. Fontanella and J. B. Gutowski, J. Phys. A 50 (2017) no.21, 215202 doi:10.1088/1751- 8121/aa6cbf [arXiv:1610.09949 [hep-th]]. [20] H. K. Kunduri, J. Lucietti and H. S. Reall, Phys. Rev. D 74 (2006) 084021 doi:10.1103/PhysRevD.74.084021 [hep-th/0606076]. [21] O. J. C. Dias, J. E. Santos and B. Way, JHEP 1512 (2015) 171 doi:10.1007/JHEP12(2015)171 [arXiv:1505.04793 [hep-th]]. [22] A. J. Amsel, G. T. Horowitz, D. Marolf and M. M. Roberts, JHEP 0909 (2009) 044 doi:10.1088/1126-6708/2009/09/044 [arXiv:0906.2376 [hep-th]]. [23] O. J. C. Dias, H. S. Reall and J. E. Santos, JHEP 0908 (2009) 101 doi:10.1088/1126- 6708/2009/08/101 [arXiv:0906.2380 [hep-th]]. [24] O. J. C. Dias, J. E. Santos and M. Stein, JHEP 1210 (2012) 182 doi:10.1007/JHEP10(2012)182 [arXiv:1208.3322 [hep-th]]. [25] A. P. Porfyriadis, JHEP 1807 (2018) 064 doi:10.1007/JHEP07(2018)064 [arXiv:1805.12409 [hep-th]]. [26] P. Figueras, J. Lucietti and T. Wiseman, Class. Quant. Grav. 28 (2011) 215018 doi:10.1088/0264-9381/28/21/215018 [arXiv:1104.4489 [hep-th]]. [27] G. W. Gibbons, A. H. Mujtaba and C. N. Pope, Class. Quant. Grav. 30 (2013) no.12, 125008 doi:10.1088/0264-9381/30/12/125008 [arXiv:1301.3927 [gr-qc]]. [28] I. Booth, M. Hunt, A. Palomo-Lozano and H. K. Kunduri, Class. Quant. Grav. 32 (2015) no.23, 235025 doi:10.1088/0264-9381/32/23/235025 [arXiv:1502.07388 [gr-qc]]. [29] F. Hejda and J. Bicak, arXiv:1510.00301 [gr-qc]. [30] J. Bicak and F. Hejda, Phys. Rev. D 92 (2015) no.10, 104006 doi:10.1103/PhysRevD.92.104006 [arXiv:1510.01911 [gr-qc]]. [31] F. Dowker, J. P. Gauntlett, S. B. Giddings and G. T. Horowitz, Phys. Rev. D 50 (1994) 2662 doi:10.1103/PhysRevD.50.2662 [hep-th/9312172]. [32] J. B. Griﬃths and J. Podolsky, doi:10.1017/CBO9780511635397 35 [33] M. Astorino, Phys. Lett. B 760 (2016) 393 doi:10.1016/j.physletb.2016.07.019 [arXiv:1605.06131 [hep-th]]. [34] M. Banados, C. Teitelboim and J. Zanelli, Phys. Rev. Lett. 69 (1992) 1849 doi:10.1103/PhysRevLett.69.1849 [hep-th/9204099]. [35] C. Martinez, C. Teitelboim and J. Zanelli, Phys. Rev. D 61 (2000) 104013 doi:10.1103/PhysRevD.61.104013 [hep-th/9912259]. [36] G. Clement, Phys. Rev. D 49 (1994) 5131. doi:10.1103/PhysRevD.49.5131 [37] G. Clement, Phys. Lett. B 367 (1996) 70 doi:10.1016/0370-2693(95)01464-0 [gr-qc/9510025]. [38] C. Li and J. Lucietti, Phys. Lett. B 738 (2014) 48 doi:10.1016/j.physletb.2014.09.012 [arXiv:1312.2626 [hep-th]].

http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png

General Relativity and Quantum Cosmology arXiv (Cornell University)

http://www.deepdyve.com/lp/arxiv-cornell-university/electrovacuum-spacetime-near-an-extreme-horizon-Mihovq8ro7

Electrovacuum spacetime near an extreme horizon

Loading next page...

References

References for this paper are not available at this time. We will be adding them shortly, thank you for your patience.

ISSN: 1095-0761
eISSN: ARCH-3332
DOI: 10.4310/ATMP.2019.v23.n7.a5
Publisher site: See Article on Publisher Site

Abstract

Journal

General Relativity and Quantum Cosmology – arXiv (Cornell University)

Published: Sep 21, 2018

Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Electrovacuum spacetime near an extreme horizon

Electrovacuum spacetime near an extreme horizon

Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Electrovacuum spacetime near an extreme horizon

Electrovacuum spacetime near an extreme horizon

References

Abstract

Journal

Recommended Articles

There are no references for this article.

Our policy towards the use of cookies