Bounds for axially symmetric linear perturbations for the extreme Kerr black hole Sergio Dain and Ivan Gentile de Austria 5 Facultad de Matema´tica, Astronom´ıa y F´ısica, FaMAF, 1 0 Universidad Nacional de C´ordoba, 2 Instituto de F´ısica Enrique Gaviola, IFEG, CONICET, n a Ciudad Universitaria, (5000) C´ordoba, Argentina. J 5 1 January 19, 2015 ] c q Abstract - r g We obtain remarkably simple integral bounds for axially symmet- [ ric linear perturbations for the extreme Kerr black hole in terms of 1 conserved energies. From theseestimates we deducepointwisebounds v 6 for the perturbations outside the horizon. 4 8 3 1 Introduction 0 . 1 0 Inthisarticlewecontinuetheworkinitiatedin[4]destinedtostudythelinear 5 stability of extreme Kerr black hole under axially symmetric gravitational 1 : perturbations using conserved energies. For a general introduction to the v subject and a list of relevant references we refer to [4]. The main result of i X thatarticleisthatthereexistsapositivedefiniteandconservedenergyforthe r a axially symmetric gravitational perturbations. In the present article, using this energy, we prove the existence of integral bounds for the first and second derivatives of the perturbation. In particular, these bounds imply pointwise estimates for the perturbation outside the black hole horizon. This result is presented in theorem 1.1 and corollary 1.2. In the following we introduce the basic definitions and notation needed to formulate the theorem and then we discuss the meaning and scope of these estimates. Axially symmetric perturbations are characterized by two functions σ 1 and ω which represent the linear perturbation of the norm and the twist 1 of the axial Killing vector. The coordinates system (t,ρ,z) is fixed by the 1 maximal-isothermal gauge condition. Partial derivatives with respect to the space coordinates (ρ,z) is denoted by ∂ and ∂ respectively and partial ρ z derivative with respect to t is denotes with a dot. We use the following notation to abbreviate the products of gradients in the spatial coordinates (ρ,z) for functions f and g ∂f∂g = ∂ f∂ g +∂ f∂ g, |∂f|2 = (∂ f)2 +(∂ f)2. (1) ρ ρ z z ρ z The 2-dimensional Laplacian ∆ is defined by ∆f = ∂2f +∂2f, (2) ρ z and the operators (3)∆ and (7)∆ are defined by ∂ f ∂ f (3)∆f = ∆f + ρ , (7)∆f = ∆f +5 ρ . (3) ρ ρ The operators (3)∆ and (7)∆ correspond to the flat Laplace operator in 3- dimensions and 7-dimensions respectively written in cylindrical coordinates and acting on axially symmetric functions. Thedomainforthespacecoordinates(ρ,z)isthehalfplaneR2 definedby + 0 ≤ ρ < ∞, −∞ < z < ∞. The axis of symmetry is give by ρ = 0. In these coordinates, the horizon is located at the origin r = 0, where r = ρ2 +z2. We follow the same notation and conventions used in [4] and we refper to that article for further details. The linear equations for axially symmetric gravitational perturbations for the extreme Kerr black hole in the maximal-isothermal gauge were obtained in [4]. In appendix A we briefly review the set of equations needed in the proofoftheorem1.1. Thebackgroundquantities aredenotedwithasubindex 0, and the first order perturbation with a subindex 1. The square norm of the axial Killing vector of the background extreme Kerr metric is denoted by η and the background function σ is defined by 0 0 η eσ0 = 0. (4) ρ2 The other relevant background quantities are the twist ω and the function 0 q . In appendix B we review the behaviour of these explicit functions. 0 It is useful to define the following rescaling of ω 1 ω 1 ω¯ = . (5) 1 η2 0 The extreme Kerr solution depends on only one parameter m which repre- 0 sents the total mass of the black hole. The first order linearization of the 2 total mass of the spacetime m vanished. The second order expansion of 1 the total mass m provides a positive definite and conserved quantity for 2 the perturbation (see [4]) that is given explicitly in equation (75). Taking time derivatives of the linear equations we get an infinity number of con- served quantities that have the same form as m but in terms of the time 2 derivatives of the corresponding quantities. For the result presented bellow, we will make use of m¯ which is obtained taking one time derivative of the 2 equations. The explicit expression for m¯ is given in equation (77). The 2 conserved quantities m and m¯ depend only on the initial conditions for the 2 2 perturbation. We will assume in the following that the perturbations satisfy the fall off and boundary conditions discussed in detail in [4]. Physically, these condi- tions imply that the system is isolated (and hence it has finite total energy m ) and that that the perturbations do not change the angular momentum 2 of the background (i.e. ω vanished at the axis). 1 Theorem 1.1. Axially symmetric linear gravitational perturbations (σ , ω ) 1 1 for the extreme Kerr black hole satisfy the following bound 1 σ2 η2|∂ω¯ |2 +|∂η |2ω¯2 +|∂σ |2 + 1 ρdρdz ≤ Cm , (6) ZR2 (cid:18)2 0 1 0 1 1 r2(cid:19) 2 + (3)∆σ 2 +η2 (7)∆ω¯ 2 e−2(q0+σ0)ρdρdz ≤ C(m¯ +m ), (7) 1 0 1 2 2 ZR2 (cid:16) (cid:17) + (cid:0) (cid:1) (cid:0) (cid:1) where C is a positive constant that depends only on the mass m of the 0 background extreme Kerr black hole. The conserved quantity m involves first spatial derivatives of σ and ω , 2 1 1 the quantity m¯ involves also second spatial derivatives of σ and ω . Note 2 1 1 however, that these terms appears in a rather complicated way and hence it is by no means obvious that m and m¯ satisfy the bound (6) and (7). 2 2 Besides σ and ω , gravitational perturbations involve other quantities 1 1 (the shift vector β , the metric function q , the second fundamental form 1 1 χAB,see[4]forthedetails). Theseotherfunctionsare,inprinciple, calculated 1 in terms σ and ω using the coupled system of equations. It is remarkable 1 1 that the estimates (6) and(7) canbe written purely interms of thegeometric functions σ and ω (which precisely encode the dynamical degree of freedom 1 1 of the system) without involving the other functions. The functions (σ ,ω ) satisfy the linear evolution equations (71)–(72). 1 1 These equations have the well known structure of a wave map coupled with a non-trivial background metric. Recently, a model problem for ananalogous wave map (but without the coupling) was studied in [6]. Remarkably enough 3 theestimates proved intheorem1.1makeuse onlyonthewave mapstructure of the equations but they hold for the complete coupled system. Also, these estimates are robust in the sense that they make use of energies that are also available for the non-linear equations. Sincetheestimates (6) and(7) essentially control uptosecond derivatives of the functions (σ ,ω ), using Sobolev embeddings we can obtain pointwise 1 1 bounds. This is, of course, one of the main motivations to obtain these kind of estimates. Note however, that the different terms are multiplied by the background functions. Particularly relevant is the factor e−2(σ0+q0) that appears in (7). This explicit function is positive, goes to 1 at infinity but vanished like r2 at the origin, where the black hole horizon is located. That means that the estimate (7) degenerate at the origin (but not at infinity) and we can not expect to control pointwise the functions (σ ,ω ) at the 1 1 origin using only this estimate. In the following corollary we prove pointwise bounds outside the horizon. Let δ > 0 an arbitrary, small, number. We define the following two domains Ω = (ρ,z) ∈ R2,such that 0 < δ ≤ r (8) δ + Γ = (cid:8)(ρ,z) ∈ R2,such that 0 < δ ≤ ρ(cid:9) (9) δ + (cid:8) (cid:9) We have the following result. Corollary 1.2. Under the same assumptions of theorem 1.1, the following pointwise bounds hold sup|σ | ≤ C (m¯ +m ), (10) 1 δ 2 2 Ωδ sup|ω¯ | ≤ C (m¯ +m ), (11) 1 δ 2 2 Γδ where the constant C depends on m and δ. δ 0 The bounds (10) and (11) are not intended to be sharp, they aremeant as example of possible pointwise bounds that can be deduced from (6) and (7). It is certainly conceivable that sharped weighted bounds can be proved using the estimates (6) and (7). But it is also clear that no pointwise bound at the horizon can be proved using these estimates, because the factor e−2(σ0+q0) vanishes there. The situation strongly resemble the problem studied in [5]. In that article the wave equation on the extreme Reissner-Nordstr¨om black hole was analyzed using conserved energies. In order to prove a pointwise bound at the horizon it was not enough with the first two energies. An extra energy which involves “integration in time” was needed. It is a relevant open question whether the same strategy can be applied to the present case, which is certainly much more complicated. 4 2 Proof of Theorem 1.1 In this section we prove theorem 1.1. We begin with the estimate (6), which represents the most important part of the theorem. In the integrand on the left hand side of (6) the terms involve up to first derivatives of σ and ω . 1 1 The integral is bounded only with the energy m , the higher order energy 2 m¯ is not needed for this estimate. Moreover, to prove the bound (6) we will 2 make use only of the last three terms in the energy density ε given by (76). 2 Note that in ε appears the same terms as in the integrand on the left hand 2 side of (6). However they appear arranged in different form (i.e. there are many cross products) and it not obvious how to deduce the bound (6). The proof of (6) can be divided in two parts. The first part consists in in- tegral estimates, this is the subtle part of the proof. The second part consists on pointwise estimates. In the arguments, we make repeatedly use of differ- ent forms of the standard Cauchy inequality, for readability we summarize them bellow. Let a ···a be arbitrary real numbers, then we have 1 n 1 a2 +a2 ≥ (a +a )2, (12) 1 2 2 1 2 and, in general, 1 a2 +a2 +... +a2 ≥ (a +a +... +a )2. (13) 1 2 n n 1 2 n Let λ > 0, then a2 a a ≤ λa2 + 2. (14) 1 2 1 4λ Inthefollowing two lemmas weprove therelevant integral estimates. The relevance of lemma 2.1 in the proof of the estimate (6) is clear: in this lemma the bound for the fourth term in (6) is proved. This integral bound is the key to prove the bounds for the second and the third term in (6). We will see in the following, that in order to prove these bounds we will need the integral estimate proved in lemma 2.2 with v = ω¯ . 1 Lemma 2.1. Consider the mass m given by (75) and (76). Then, the 2 following inequality holds σ2 m ≥ 1 ρdρdz. (15) 2 ZR2 r2 + Proof. The mass m is the second variation of the total ADM mass (see 2 [4]). The first three terms in (76) correspond to the dynamical part of the 5 mass (these terms vanished for stationary solutions), the last three terms correspond to the stationary part of the mass. These terms are precisely the secondvariationofthemassfunctionalextensively studiedinconnectionwith the mass angular momentum inequality (see [2], [3] and reference therein). In a recent article [7], an important estimate has been proved for the second variation of this functional in terms of the distance function in the hyperbolic plane. From lemma 2.3 in [7] we deduce the following inequality m ≥ 2 |∂d((η ,ω ),(η ,ω ))|2ρdρdz, (16) 2 1 1 0 0 ZR2 + where d is the distance function in the hyperbolic plane between the two points (η ,ω ) and (η ,ω ), where η = ρ2eσ1 (see, for example, [2] for the 1 1 0 0 1 explicit expression of d). Toobtainthedesired lower boundfortheright handside oftheinequality (16) we first use the following weighted Poincare inequality proved in [1] (equation (31) in [1] with δ = −1/2) d2 2 |∂d|2ρdρdz ≥ ρdρdz, (17) ZR2 ZR2 r2 + + and then we use the following bound for the distance function d proved in [2] (see equation (138) in that reference) |d| ≥ |σ |. (18) 1 Lemma 2.2. Let η andω bethe normand the twist functionforthe extreme 0 0 Kerr black hole, and let v be an arbitrary smooth function with compact support outside the axis. Then, the following inequality holds |∂ω |2v2ρdρdz ≤ 3 |∂η |2v2ρdρdz + η2|∂v|2ρdρdz. (19) ZR2 0 ZR2 0 ZR2 0 + + + Proof. We will use that the background function η and ω satisfy equation 0 0 (81). Let v is an arbitrary function with compact support outside of the axis. We multiply (81) by η−2δv2 (where δ is an arbitrary number) and integrate, 0 we obtain |∂ω |2 η−2δv2(3)∆(lnη )ρdρdz = − 0 η−2δv2ρdρdz. (20) ZR2+ 0 0 ZR2+ η02 0 6 Integrating by parts the left hand side of (20) we obtain the following useful identity |∂ω |2η−2δ−2v2ρdρdz = −2δ η−2δ−2|∂η |2v2ρdρdz+ ZR2 0 0 ZR2 0 0 + + +2 η−2δ−1v∂v∂η ρdρdz (21) 0 0 ZR2 + We take δ = −1 in (21), we obtain |∂ω |2v2ρdρdz = 2 |∂η |2v2ρdρdz +2 η v∂v∂η ρdρdz, (22) 0 0 0 0 ZR2 ZR2 ZR2 + + + ≤ 3 |∂η |2v2ρdρdz + η2|∂v|2ρdρdz, (23) 0 0 ZR2 ZR2 + + where, to obtain line (23) we have used in the second term of the right hand side of(22) theinequality (14) witha = η ∂ω¯ , a = ∂η ω¯ andλ = 1/2. 1 0 1 2 0 1 We prove now the pointwise bounds in terms of the energy density ε . In 2 the following we denote by C a generic positive constant that depends only on the background parameter m . 0 We begin with the first term in the integrand in (6). From the explicit expression for ε given in (76), keeping only the fifth and sixth terms, we 2 obtain ε 2 ≥ ∂ ω η−1 −η−1σ ∂ω 2 + η−1σ ∂ω −ω η−2∂η 2 (24) ρ 1 0 0 1 0 0 1 0 1 0 0 (cid:0) (cid:0) (cid:1) (cid:1) (cid:0) (cid:1) = (η ∂ω¯ +ω¯ ∂η −η−1σ ∂ω )2 +(η−1σ ∂ω −ω¯ ∂η )2 (25) 0 1 1 0 0 1 0 0 1 0 1 0 where in (25) we have used the definition of ω¯ given in (5). We use the 1 Cauchy inequality (12) in (25) to finally obtain ε 1 2 ≥ η2(∂ω¯ )2. (26) ρ 2 0 1 From the second term in (6) we take ε given in (76) and keep only the 2 last term, we obtain 2 ε ∂ω 2 0 ≥ σ −∂η ω¯ , (27) 1 0 1 ρ (cid:18) η (cid:19) 0 |∂ω |2 ∂ω = 0 σ2 +|∂η |2ω¯2 −2σ ω¯ 0∂η , (28) η2 1 0 1 1 1 η 0 0 0 |∂ω |2 1 ≥ − 0 σ2 + |∂η |2ω¯2, (29) η2 1 2 0 1 0 C 1 ≥ − σ2 + |∂η |2ω¯2, (30) r2 1 2 0 1 7 where in the inequality (29) we have used the Cauchy inequality (14) with λ = 1 and in line (30) we have used the bound (83) for the background quantities. We have obtained ε C 1 2 + σ2 ≥ |∂η |2ω¯2. (31) ρ r2 1 2 0 1 We integrate the bounds (26) and (31), and use the integral bound (15) to obtain 1 σ2 η2|∂ω¯ |2 +|∂η |2ω¯2 + 1 ρdρdz ≤ Cm . (32) ZR2 (cid:18)2 0 1 0 1 r2(cid:19) 2 + To prove (6) it only remains to bound the term |∂σ |2. For that term, we 1 use the fourth term in (76) to obtain ε 2 ≥ ∂σ +ω η−2∂ω 2 (33) ρ 1 1 0 0 (cid:0) (cid:1) = (∂σ +ω¯ ∂ω )2, (34) 1 1 0 = |∂σ |2 +ω¯2|∂ω |2 +2ω¯ ∂σ ∂ω , (35) 1 1 0 1 1 0 1 ≥ |∂σ |2 −|∂ω |2ω¯2, (36) 2 1 0 1 where in line (34) we have just used the definition of ω¯ and in line (36) we 1 1 have used inequality (14) with λ = . Then, we have obtained 4 ε 1 2 +|∂ω |2ω¯2 ≥ |∂σ |2. (37) ρ 0 1 2 1 We integrate the pointwise estimate (37), to handle second term on the left hand side of (37) we use the integral bound (15) v = ω¯ and the bound (32). 1 Hence we have obtained the desired estimate (6). We turn to the bound (7) which involves second derivatives of the func- tions σ and ω¯ and hence we need the higher order mass m¯ . 1 1 2 We begin with the term with σ . We use the evolution equation (71) to 1 8 obtain 2 2 (3)∆σ 2 = e2(σ0+q0)p˙+ σ |∂ω |2 −∂ω ∂ω (38) 1 (cid:18) η2 1 0 1 0 (cid:19) (cid:0) (cid:1) 0 (cid:0) (cid:1) 2 2 = e2(σ0+q0)p˙+ σ |∂ω |2 −2η ω¯ ∂η ∂ω +η2∂ω ∂ω¯ (cid:18) η2 1 0 0 1 0 0 0 0 1 (cid:19) 0 (cid:0) (cid:1) (39) 16 64 ≤ 4e4(σ0+q0)p˙2 + σ2|∂ω |4 + (∂η ∂ω )2ω¯2 +16(∂ω ∂ω¯ )2 η4 1 0 η2 0 0 1 0 1 0 0 (40) |∂ω |2σ2 64 ≤ 4e4(σ0+q0)p˙2 +C 0 1 + |∂η |2|∂ω |2ω¯2+16|∂ω¯ |2|∂ω |2, η2 r2 η2 0 0 1 1 0 0 0 (41) where in line (39) we have used the definition of ω¯ (5), in line (40) we have 1 used the inequality (13) and line (41) follows from the bound (83) and the Cauchy-Schwartz inequality. Multiplying by e−2(σ0+q0) the inequality (41) we obtain |∂ω |2 σ2 e−2(σ0+q0) (3)∆σ 2 ≤ 4e2(σ0+q0)p˙2 +e−2(σ0+q0) 0 C 1 +64|∂η |2ω¯2+16|∂ω¯ |2η2 1 η2 (cid:18) r2 0 1 1 0(cid:19) (cid:0) (cid:1) 0 (42) ε¯ σ2 ≤ 2 2 +C 1 +|∂η |2ω¯2 +|∂ω¯ |2η2 (43) ρ (cid:18)r2 0 1 1 0(cid:19) where in line (43) we have used (78) and the bound (85). Integrating (43) and using the previous bounds we finally obtain e−2(σ0+q0) (3)∆σ 2ρdρdz ≤ C(m¯ +m ). (44) 1 2 2 ZR2 + (cid:0) (cid:1) To estimate the second derivatives of ω¯ we proceed in a similar way. We 1 will make use of the evolution equation (72). First, it is useful to write this equation in terms of ω¯ instead of ω . To do that we first obtain the following 1 1 relation 4 ω η2(7)∆ω¯ = (3)∆ω − ∂ ω −4∂ω ∂σ −2ω (3)∆σ +8 1∂ σ +4ω |∂σ |2, 0 1 1 ρ ρ 1 1 0 1 0 ρ ρ 0 1 0 (45) where we have used the definition of ω¯ given in (5), the expression of η in 1 0 terms of σ given in (4), the definitions of the operators (3)∆ and (7)∆ given 0 9 in (3) and the identity (82). Using the evolution equation (72) and equation (45) we obtain η2(7)∆ω¯ = e2(σ0+q0d˙−2η2ω¯ (3)∆σ −2η2∂σ ∂ω¯ +2∂ω ∂σ . (46) 0 1 0 1 0 0 0 1 0 1 To obtain the estimate, we take the square of each side of equation (46) and use the Cauchy inequality (13) to obtain η4 (7)∆ω¯ 2 ≤ 4e4(σ0+q0)d˙2 +16η4 (3)∆σ 2ω¯2 +16η4(∂σ ∂ω¯ )2 +16(∂ω ∂σ )2 0 1 0 0 1 0 0 1 0 1 (cid:0) (cid:1) (cid:0) (cid:1) (47) ≤ 4e4(σ0+q0)d˙2 +16|∂ω |2 |∂ω |2ω¯2+|∂σ |2 +16η4|∂σ |2|∂ω¯ |2 0 0 1 1 0 0 1 (cid:0) (cid:1) (48) where in line (48) we have used the Cauchy-Schwartz inequality andequation (79) to substitute the factor (3)∆σ . We multiply by e−2(σ0+q0)η−2 each side 0 0 of inequality (48) η2e−2(σ0+q0) (7)∆ω¯ 2 ≤ 4e2(σ0+q0)d˙2+16e−2(σ0+q0)|∂ω0|2 |∂ω |2ω¯2 +|∂σ |2 0 1 η2 η2 0 1 1 (cid:0) (cid:1) 0 0 (cid:0) (cid:1) +16e−2(σ0+q0)|∂σ |2η2|∂ω¯ |2 (49) 0 0 1 Then, we bound the first term on the right hand side of inequality (49) with the energy density ε¯ (78), for the other terms we use the inequalities (85) 2 and (86) to bound the background functions by a constant C. We obtain ε¯ η2e−2(σ0+q0) (7)∆ω¯ 2 ≤ 2 2 +C |∂ω |2ω¯2 +|∂σ |2 +η2|∂ω¯ |2 . (50) 0 1 ρ 0 1 1 0 1 (cid:0) (cid:1) (cid:0) (cid:1) Integrating (50) and using (6) we finally have η2e−2(σ0+q0) (7)∆ω¯ 2ρdρdz ≤ C(m¯ +m ). (51) 0 1 2 2 ZR2 + (cid:0) (cid:1) 3 Proof of Corollary 1.2 In the proof of the corollary 1.2 we essentially use an appropriated variant of the Sobolev embedding and standard cut off functions arguments. Let χ : R → R be a smooth cut off function such that χ ∈ C∞(R), 0 ≤ χ ≤ 1, χ(r) = 1for0 ≤ r ≤ 1, χ(r) = 0for2 ≤ r. Defineχ (r) = χ(r/δ). δ Consider the following function σ¯ = (1−χ )σ (52) 1 δ 1 10