A Summary of the Black Hole Perturbation Theory Steven Hochman Introduction Many frameworks for doing perturbation theory The two most popular ones Direct examination of the Einstein equations -> Zerilli-Regge-Wheeler equations for Schwarzschild. Newman-Penrose formalism -> Bardeen-Press equation for the Schwarzschild type, and the Teukolsky equation for Kerr type black holes. The Metric In spherical polar coordinates the flat space Minkowski metric can be written as ds2 = −dt2 + dr2 + r2 dΩ2 where r2 dΩ2 = r2 dθ2 + r2 sin2 θdφ2 Schwarzschild The Schwarzschild metric is a vacuum solution ! 2M 2 ds = − 1 − r " 2 dr 2 2 $ dt2 + # + r dΩ 1 − 2M r The coordinates above fail at R = 2M Killing Vectors Killing vectors tell us something about the physical nature of the spacetime. Invariance under time translations leads to conservation of energy Invariance under rotations leads to conservation of the three components of angular momentum. Angular momentum as a three-vector: one component the magnitude and two components the direction. Killing Vectors of Schwarzchild Two Killing vectors: conservation of the direction of angular momentum -> we can choose pi = 2 for plane Energy conservation is shown in the timelike Killing vector µ µ K = (∂t ) = (1, 0, 0, 0) Magnitude of the angular momentum conserved by the final spacelike Killing vector R = (∂φ ) = (0, 0, 0, 1) µ µ Geodesics in Schwarzschild The geodesic equation can be written after some simplification as 1 2 ! dr dλ "2 + V (r) = ε, The potential is 1 GM L2 GM L3 V (r) = ! − ! + 2− 2 r 2r r3 The Event Horizon and the Tortoise Null cones close up ! dt 2GM =± 1− dr r "−1 Replace t with coordinate that moves more slowly t = ±r + constant ∗ where ! r r = r + 2GM ln 2GM ∗ " −1 More Tortoise ds = 2 ! 2GM 1− r " (−dt + dr ) + r dΩ R = 2GM -> - infinity Transmission Reflection 2 ∗2 2 2 Kruskal Coordinates 32G M −r/2GM 2 2 2 2 ds = − e (−dT + dR ) + r dΩ r 2 3 3 Null cones T = ±R + constant Unlike the tortoise the event horizon is not infinitely far away, and is defined by T = ±R Vishveshwara Kerr ! 2GM r 2 ds = − 1 − ρ2 " 2 2GM ar sin θ ρ2 2 2 dt − (dtdφ + dφdt) + dr ρ2 ∆ 2 sin θ 2 2 2 2 2 2 2 2 +ρ dθ + [(r + a ) − a ∆ sin θ]dφ ρ2 where ∆(r) = r − 2GM r + a 2 2 and ρ (r, θ) = r + a cos θ 2 Angular momentum 2 2 2 Einstein Field Equation Rµν 1 − Rgµν = 8πGTµν 2 Can also be written as Rµν = 8πG(Tµν 1 − T gµν ) 2 Perturbations Rµν = 0 For a perturbation ! gµν = gµν + hµν Inserting this in Rµν + δRµν = 0 But δRµν = 0 Schwarzschild Perturbations Regge and Wheeler - Spherical Harmonics Stability? Gauge invariance? Physical Continuity? “Ring down” Zerilli - Falling particle Tensor Harmonics Separate the solution into a product of four factors, each a function of a single coordinate. This separation is best achieved by generalizing the method of spherical harmonics already established for vectors, scalars, and spinors. Parity Scalar functions have even parity. Two kinds of vectors, each of different parity: One the gradient of a the spherical harmonic and has even parity. The pseudogradient of the spherical harmonic, and has odd parity. There are three kinds of tensors. One is given by the double gradient of the spherical harmonic and has even parity. Another is a constant times the metric of the sphere, also with even parity. The last is obtained by taking the double pseudogradient; it has odd parity. 2 ∂ ∂ its even(electric or polar) and odd(magnetic +G(t, r)( dθ −(cos )] into or axial) parity parts θ)( sin1 θ )( ∂φ 2 )] Sym Sym r2 [K(t, r) sin2 θ Odd/Magnetic/Axial parity = YLMSym × ∂2 1 ∂ ∂ +G(t, r)[( 0 0 −h0 (t, r)( sin θ )( ∂φ ) h0 (t, r)(sin θ)( ∂θ ) ∂φ2 ) ∂ ∂ θ)( ) +(sin θ)( θ)(cos 0 ∂θ 0 −h (t, r)( 1 )( ∂ ) h (t, r)(sin ) 1 1 sin θ ∂φ ∂θ 2 2 1 ∂ 1 1 ∂ ∂ Symof Sym Due to the spherical symmetry the background equations (**)sinand do(cos θ)( ∂θ h2 (t, r)[( sinmetric, )( ) 2 h2 (t, r)[( )( (**)) + ) θ ∂θ∂φ θ ∂φ∂φ 2 not mix terms belonging to different L and parity. L θon∂ the −(cos θ)( sin12M −ofsin )] )( is∂ )]the projection Black Hole Perturbation Theory 15 ∂θ∂θ θ ∂φ 2 Even and Odd A Summary of the ∂ z − axis. To apply quantum language can say that∂ L,M Sym Sym to a classical Sym problem,−hwe 2 (t, r)[(sin θ)( ∂θ∂φ ) − (cos θ)( ∂φ ) andharmonics the parityfrom are above constants of able the motion. Theperturbation existence of hstill another constant Using the tensor we are to split the µν M Even/Electric/Polar Parity = YL × fromand theodd(magnetic circumstance that the background into its even(electricfollows or polar) axial) parity parts metric is independent of the cotime, or ∂ M (1 −consider 2M/r)Ha0 (t, r) H r) h0 (t, r)(ω∂θ h0 (t, r T = ct. On can perturbation of1a(t,definite frequency, =) kc, Odd/Magnetic/Axial parity = Ythis L ×account we r) h (t, r)( ∂ ) −1 Sym h (t, r)(sin − 2M/r) h1 (t, r 1 ∂ the perturbation ∂have so that of hµν(1will aH time of the 2 (t, dependence 1 A Summary of the Black 0Hole Perturbation Theory 15 ∂θ 0 every −h0component (t, r)( )( ) θ)( ) 0 sin θ ∂φ ∂θ (iωt) (−ikT ) 2 the = e . proceed to determine completely form of the r[K(t, r) r G(t, r)[( 1We ∂thereforeSym ∂ Sym 0form e 0 −h1 (t, r)( sin θ )(∂φ ) h1 (t, r)(sin θ)( ∂θ ) 2 ∂ Using the tensorharmonics fromsolution above we are able to split L theand perturbation µν frequency. individual of parity, M∂ 2values, hand Ther)( general +G(t, −(cos θ)( 1 specified ∂2 1 1 ∂ 2 )] Sym dθ Sym h (t, r)[( )( ) h (t, r)[( )( ) + (cos θ)( ) 2 2 sin θ or∂θ∂φ 2parity parts sin θ ∂φ∂φ ∂θ into its even(electric or polar) and will odd(magnetic axial) of solution be a superposition these individual with coefficients 2solutions Sym Sym Sym adjusted r2 [K(t, ∂ 1 ∂ )] − sin θ )] M −(cos θ)( sin2 θ )( ∂φ ∂θ∂θ is no need to work Odd/Magnetic/Axialparity L × to = fit Ythe appropriate boundary conditions and initial values. There +G(t, r ∂2 ∂ Sym Sym Sym −h (t, r)[(sin θ)( ) − (cos θ)( ) 2 ∂ as all will lead ∂ radial equation. ∂θ∂φ ∂φ In this case we will an0 (t, arbitrary tor)(sin the same 0 0with−h r)( sin1 θ )(M∂φ ) h0 (t, θ)( ∂θ ) +(sin θ)(c M 1 ∂ advantage that φ will completely ∂ 0 from the calculations. Even/Electric/Polar Parity = × the M 1= with disappear L 0take −h (t,0Yr)( )(the ) spherical symmetry h1 (t, r)(sinofθ)( )background metric, equations (**) and (**) do sin θ ∂φ ∂θ Due to the The odd Sym waves contain three unknown functions: 2 2 1 ∂ amount 1 of work 1remains. ∂ ∂ odd ∂ ∂ three unknown A2 (t, considerable The waves contain Sym h r)[( )( ) h (t, r)[( )( ) + (cos θ)( ) (1 − 2M/r)H (t, r) H (t, r) h (t, r)( ) h (t, r)( ) 2 0 not 0 0 M is terms21belonging L and ∂θparity. of L on the sin θmix ∂θ∂φ sinto θ different ∂φ∂φ ∂θ ∂φ the projection 2 ∂ 1 ∂ ∂ ∂ functions of r: (h , h , h ). The even waves contain seven unknown functions: −1 1 2 −(cos )] ∂θ ) to a classical )((1 0−)]2M/r) Sym θ)( H2 (t,−r)sinhθ1language (t, r)( h1 (t, r)( z sin − 2axis. problem, we can say that L,M ∂θ∂θ ∂φ θ ∂φ To apply quantum of gauge 2 (H , H , H , G, K, h , h ). The calculations can be greatly simplified by the use ∂ ∂ 2 2 0 1 2 0 1 another constant Sym Sym r[K(t, r) r G(t, r)[(∂ /∂θ∂φ) Sym Sym Sym −h (t, r)[(sin θ)( ) − (cos θ)( ) and the parity are constants of the motion. The existence of still 2 ∂θ∂φ ∂φ transformations. ∂2 1 ∂ +G(t, r)( −(cos θ)( )( )] follows from the circumstance that the background metric is independent of the cotime M 2 )] dθ sin θ ∂φ Even/Electric/Polar Parity = Y × L Sym T = ct. On thisSym Sym r2 [K(t, r)ofsin θ account we can consider a perturbation a2definite frequency, ω = kc ∂ ∂ 2 (1 − 2M/r)H (t, r) H (t, r) h (t, r)( ) h (t, r)( ) 5.4. Gauge Transformations 0 1 0 0 ∂ dependence of the ∂θ ∂φ so that every component of the perturbation h will have a time +G(t, r)[( ) 2 µν ∂ ∂φ ∂ −1 ∂ Sym (1 − 2M/r) H (t, r) h (t, r)( ) h (t, r)( (iωt) (−ikT ) 2 1 1 ∂θ ∂φ form e gauge, = e which . We therefore proceed determine completely +(sin θ)(cos θ)( ) Whatthe The Regge-Wheeler we are about to 2walk to through, is unique. weform of the 2 ∂θ Sym Sym r[K(t, r) r G(t, r)[(∂ /∂θ∂φ) frequency. solution of specified parity, L and M(**) values, and aresymmetry trying toindividual do the is create gauge invariant quantities and work in a do fixed gauge. Actually,The genera Due to the spherical of background metric, equations (**) and ∂2 1 ∂ +G(t, r)(of )] −(cos θ)( sin θ solutions )( ∂φ )] dθ2 these solution will beparity. athe superposition individual coefficients with these two tasks are one and same. As long as one works in athe uniquely fixed gauge, adjusted not mix terms belonging to different L and M is the projection of L on 2 2 Sym quantities Sym rin[K(t, r) sinthat θ one There to one fit the appropriate boundary conditions and initial values. is no need to work is Sym dealing with are gauge invariant, the sense can translate To applythe 2 z − axis. quantum language to a classical problem, we can say that L,M ∂ +G(t, r)[( ) equation. with an arbitrary Mwithout as all will the same radial In this case we wil ∂φ2 into any gauge one wants changing the physical problem. In the following and theparity are them constants of the motion. The existence oflead stillto another constant ∂ +(sin θ)(cos θ)( disappear ) takethe M background = 0 with themetric advantage that φ will from the calculations follows from the circumstance that is independent ofcompletely the cotime,∂θ A considerable work remains. The Due toTthe spherical symmetry of can the consider background metric,amount equations (**) and (**) do ct. On this account we a seven perturbation of a of definite frequency, ω =odd kc, waves contain three unknown The =even waves contain unknown functions: of r:hM(his h1 , hprojection The of even waves not mix belonging to different Lperturbation and parity. L on the contain seven unknown functions 0 , the 2 ). so terms that every component of thefunctions µν will have a time dependence of the (H0to , Hproceed H2 , G,toK,determine h0 , h1 ). The calculations be of greatly (−ikT ) z − axis. apply language a1 , classical problem, we can say that L,M formTo e(iωt) = equantum . We therefore completely thecan form the simplified by the use of gauge transformations. and the parity are constants of the motion. of and still frequency. another constant individual solution of specified parity, LThe andexistence M values, The general followssolution from thewill circumstance that the background metric is independent of the cotime, be a superposition of these individual solutions with coefficients adjusted 5.4.a Gauge Transformations T = ct. On this account we can consider perturbation of a definite frequency, ω = kc, Gauge Transformations A Summary of the Black Hole Perturbation Theory 16 we will use calculations in the Regge-Wheeler gauge. Any result can be expressed in a gauge of invariant manner substitutingTheory the Regge-Wheeler gauge quantities in terms A Summary the Black Holeby Perturbation 16 of The Regge-Wheeler a is unique fixed gauge a general gauge. (See Gleiser gauge 1996 for proof) of in thethe Black Hole the Perturbation Theory 16 we will A useSummary calculations Regge-Wheeler gauge. Any result can beviewed expressed in a Different waves can represent same physical phenomena in different systems ofmanner coordinates. Consider an coordinate gauge invariant by substituting theinfinitesimal Regge-Wheeler gauge transformation: quantities in terms of The quantities are gauge invariant wegauge. will use in the Regge-Wheeler gauge. Any result can be expressed in a a general (Seecalculations Gleiser !α α 1996 α for proof) α α x = x + ξ (ξ ! x ). (76) in terms of gauge waves invariant manner bythe substituting the phenomena Regge-Wheeler gauge quantities Different can represent same physical viewed in different Any result can be expressed in a gauge invariant α aofgeneral gauge.displacements (See Gleiser for coordinate proof) infinitesimal ξ 1996 transform like a vector. In the new frame we shall systemsThe coordinates. Consider an infinitesimal transformation: manner by substituting the Regge-Wheeler gauge have: Different can represent the αsame αphysical phenomena viewed in different !α α waves α x =x + (ξ ! x ).gauge (76) quantities inξ terms of a general systems of coordinates. infinitesimal coordinate transformation: ! gµν + h!µν = gµνConsider + ξµ ;ν +ξνan ;µ +h (77) µν . α The infinitesimal displacements ξ transform like a vector. In the new frame we shall !α α α α α Consider x = x + ξ (ξ ! x ).the Schwarzschild (76) Now h is defined as the difference between the perturbed metric and µν have: metric written in spherical coordinates. According to this definition, the difference in α ! ! The infinitesimal displacements ξ like a vector. In the new g + h = g + ξ ; +ξ ; +h . (77) frame we shall µν theµvalue ν ν µ µνtransform µν µν have the new frame will have: new difference old Now hµν is defined ashthe between the perturbed metric and the Schwarzschild(78) µν = hµν + ξµ ;ν +ξν . ! metric written in spherical this definition, the difference in gµν +coordinates. h!µν = gµν +According ξµ ;ν +ξν ;to (77) µ +hµν . This result can be interpreted by saying that the infinitesimal changes in the coordinates the new frame will have the value of the hhµν ,isundergo aasgauge transformation quitethe similar to the metric well known gauge Now defined the difference between perturbed and the Schwarzschild µν new old hµν = hfor + ξµelectromagnetic ;ν +ξν . (78) transformation field. We use this to simplify the description µν the metric written in spherical coordinates. According to this definition, the difference in of the hµν , undergo a gauge transformation quite similar to the well known gauge The gauge for transformation can be field. performed on this any toindividual partial wave. transformation the electromagnetic We use simplify the description Obviously no real simplification result unless the resulting wave still belongs to of the perturbation and make it will unique. the original eigenvalues. This requirement the possible for ξ α .partial This vector The gauge transformation can be limits performed on anychoices individual wave. turns out to no be areal spherical harmonic of the same L and as thewave partial underto Obviously simplification will result unless theparity resulting stillwave belongs consideration. Such a gauge transformation allowsthe uspossible to impose additional the original eigenvalues. This requirement limits choices for ξ α . simplifying This vector conditions thea perturbation hµν . We havesame therefore turns out on to be spherical harmonic of the L andchosen parity to as eliminate the partialthose waveterms under which contain the derivatives the highest order with to the angles. simplifying The final consideration. Such a gauge of transformation allows us respect to impose additional α radial equations then simplify. Moreover, the desired gauge transformation can then conditions on the perturbation hµν . We have therefore chosen to eliminateξ those terms be foundcontain by the the use derivatives of finite operations only, without arbitrary and boundary which of the highest order with respectconstants to the angles. The final conditions. radial equations then simplify. Moreover, the desired gauge transformation ξ α can then α The gauge thatoperations simplifies the odd wave must have the be found by thevector use ofξ finite only,general without arbitrary constants andform boundary Regge-Wheeler Gauge conditions. The gauge vector odd ξ 0 = 0; ξα1 = 0; that ξ µ = Λ(T, simplifies r); "µν (∂/∂xν )YLM (θ, φ),the (µ, ν = general 2, 3) (79) The gauge vector ξ that simplifies the general odd wave must have the form wave has the form according to the foregoing arguments. Moreover, the radial function Λ can be adjusted ξ 0 = 0; ξ 1 = 0; ξ µ = Λ(T, r); "µν (∂/∂xν )YLM (θ, φ), (µ, ν = 2, 3) (79) to annul the radial factor h2 (T, r). The final canonical form for an odd wave of angular momentum L and according to the foregoing arguments. Moreover, thetotal radial function Λ can be adjusted The final canonical form for an odd wave L, projection M= 0 is then to annul the radial factor h2 (T, r). is M = 0 L and The final canonical form for an odd wave of total 0 angular 0 0momentum h0 (r) projection M = 0 is then 0 0 0 h (r) 1 (−ikT ) hodd = e (sin θ)(∂/∂θ)P (cos θ) × L µν 00 00 h00 (r) 00 Sym Sym 0 01 (r) 0 0 0 h odd (−ikT ) A Summary ofhthe Black Hole Perturbation Theory 17 = e (sin θ)(∂/∂θ)P (cos θ) × L µνthe Black Hole Perturbation Theory A Summary of 17 even 0 the 0 0general 0 The gauge vector that simplifies 0 The gauge transformation that simplifies even waves is Sym Sym 0 The gauge transformation that simplifies even waves is wave has the form M M ξ0 = (θ, φ); ξ1 = (θ, φ); (80) 0 M0 (T, r)YL M 1 M1 (T, r)YL M ξ = M0 (T, r)YL (θ, φ); ξ = M1 (T, r)YL (θ, φ); (80) M 2 M ξ2 = (θ, φ); ξ 3 = (θ, φ).(81) 2 M (T, r)(∂/∂θ)YL M 3 M (T, r)(1/ sin θ)(∂/∂φ)Y 2 L M ξ = M (T, r)(∂/∂θ)YL (θ, φ); ξ = M (T, r)(1/ sin θ)(∂/∂φ)YL (θ, φ).(81) We adjust the factors M0 , M1 , and M to annul the factors G, h0 , and h1 in the We adjust the factors M0 , M1 , and M to annul the factors G, h0 , and h1 in the even perturbations, thereby obtaining the even waves in the formwave L, M = 0 The final canonical form for ancanonical even even perturbations, thereby obtaining the even waves in the canonical form H0 (1 − 2M/r) is H1 0 0 H (1 − 2M/r) H 0 0 0 1 −1 Sym H (1 − 2M/r) 0 0 2 even (−ikT ) −1 Sym H (1 − 2M/r) 0 0 hµνeven= e (−ikTP)L (cos θ) × 2 2 hµν = e PL (cos θ) × 0 0 r K 0 2 0 0 r K 2 02 0 0 0 r K sin θ 0 0 0 r2 K sin2 θ There are therefore two unknown functions of r in the odd case (h0 , h1 ) and four unknown The Choice of Gauge There are now only two unknown functions for the odd case and four for the even case This helps tremendously with the differential equations But even perturbations increase with distance and remain in unchanging magnitude for odd 1/r? We can choose another gauge (Radiation) Solutions Even/Electric/Polar Odd/Magnetic/Axial L = 0,1,2.... Static k=0 Solutions for L values There is no L = 0 odd/magnetic perturbation L = 0, L = 1 even and L = 1 odd: the changes from perturbations in mass, velocity, and angular momentum, have exact solutions. L >=2 describe the radiation, no exact solutions. Odd/Magnetic Solutions A Summary of the Black Hole Perturbation Theory equations For odd waves there are three non-trivial condition. Thus we have a system of two first-order linear equations. The tw 18 Can be expressed as a wave equation known as the order equations can be expressed, after a series of substitutions, as a simple secon Regge-Wheeler Equation condition. Thus we have known a systemasofthe twoRegge-Wheeler first-order linear equation equations. for Theodd twoperturbatio firstSchrödinger equation A Summary of the Black Hole Perturbation Theory A Summary of the Black Hole Theory 18 order equations canPerturbation be expressed, after a series of substitutions, as a simple second-order 2 d Ψodd equation known the Regge-Wheeler equation 2 first-order condition.Schrödinger Thus we have a system of twoas linear equations. The for twoodd first-perturbations: dr∗2 + k (r)Ψodd = 0. order equations can be expressed, d2 Ψodd after2 a series of substitutions, as a simple second-order + Regge-Wheeler k (r)Ψodd = 0.equation for odd perturbations: Schrödinger equation as ∗2 the This can known alsodrbe written in the time-domain as In time domain 2 Ψoddalso 2be written Thisd can as d2 Ψodd d2 Ψtime-domain + k (r)Ψ = 0.in the odd odd dr∗2 − + V (r)Ψ odd ∗2 d2 Ψodd dt2 d2 Ψodd dr − +asV (r)Ψodd = 0, This can also be written in the time-domain dr∗2 dt2 with d2 Ψodd with with dr∗2 d2 Ψodd − + V (r)Ψodd = 0, dt2 2 L = (82) (83) (83) 2 3 V (r) = [−L(L + 2 1)/r + 3 6M/r ](1 − 2M/r). V (r) = [−L(L + 1)/r + 6M/r ](1 − 2M/r). with L = = 0, (82) 3 (r) =is[−L(L + 1)/r + 6M/r ](1 − 2M/r). ForLL==0 V0there there is odd no odd parity type harmonic. For no parity type harmonic. LL == 1 parity perturbation represents the addition of a angular m0 a, where m0 a The 1 perturbation represents the addition of amomentum angular momentum For L = 0 thereThe is no odd type harmonic. the perspective this can themomentum conserved the The perturbation Lfrom = 1 perturbation represents the addition ofseen a angular mangular 0 a, wheremomentum from theZerilli Zerilli perspective thisbecan beasseen as the conserved angular of momentum 0 no from the Zerilli this can be seen as the conserved angular momentum of the fallingperspective fallingparticle. particle. falling particle.Static solutions of the odd type exist for L "= 0. For the odd equations with k = 0, Static solutions of the odd exist L "= 0. For 1 addition of angular Statichsolutions of the odd type exist formomentum L "= 0.type For the oddfor equations with k =the 0, odd equations with 1 must vanish. h1 must vanish. h must vanish. with dr dt 1 perturbation represents the addition of a angular momentum m0 a, wh For L = 0 The thereLis=no odd parity type harmonic. with 2 3 from perspective this can seen asofthe conserved angular momentum = [−L(L + 1)/r + 6M/r ](1 2M/r). The L the =V (r) 1 Zerilli perturbation represents the−be addition a angular momentum m0 a, whereof 2 3 V (r) = [−L(L + 1)/r + 6M/r ](1 − 2M/r). falling particle. from the Zerilli perspective this can be seen as the conserved angular momentum of the For L = 0 there is no odd parity type harmonic. For there is no solutions odd parity of type harmonic. Static odd type exist for L "= 0. mFor the odd equations with k = falling The LL==10particle. perturbation represents thethe addition of a angular momentum 0 a, where The = 1solutions perturbation the addition of aangular momentum m0the a, where hL1perspective must vanish. from the Zerilli this of canrepresents be seen the conserved of Static the odd as type exist for Langular "= 0. momentum For the odd equations with k = 0, the Zerilli perspective this can be seen as the conserved angular momentum of the fallingfrom particle. h1 must vanish. fallingsolutions particle.of the odd type exist for L "= 0. For the odd equations with k = 0, Static 5.7. Even/Electric Static solutions of the oddSolutions type exist for L "= 0. For the odd equations with k = 0, h1 must vanish. Even/Electric Solutions h5.7. vanish. 1 must Even/Electric Solutions For even waves,are the ten Einstein field equations give seven nontrivial conditio For even waves there seven non-trivial 5.7. Even/Electric Solutions one algebraic relation of two of theequations unknown functions; first-order differen For even waves, the ten Einstein field give seven three nontrivial conditions: equations: One algebraic relation, three first5.7. Even/Electric Solutions equations; and three three second-order equations. This leaves six equati even the ten Einstein equations givedifferential seven functions; nontrivial conditions: onewaves, algebraic relation offield two ofsecond-order the unknown three first-order differential orderForequations, and equations. For even waves, the ten Einstein field equations give seven nontrivial conditions: one algebraic relation of two of the unknown functions; three equations. first-order differential for theand three remaining unknowns. Three of the first-order equations sufficient equations; three second-order differential This leaves six are equations one algebraic relation of two of the unknown functions; three first-order differential equations; and three second-order differential equations. This leaves six equations determine a solution provided the divergence conditions on the source are satisfi for the three remaining unknowns. Three of the first-order equations are term sufficient to equations; and three second-order equations. Thisare leaves six the equations for the three remaining unknowns. Three differential of equation the first-order equations sufficient to Can be expressed as a wave known as Afteraremaining some substitutions and simplifications we can arrive thesatisfied. second-or determine solution provided the divergence conditions on the source term for the three unknowns. Three ofother the on first-order equations are sufficient to atare determine a solution provided the divergence conditions the source term are satisfied. Zerilli determine Equation Schrödinger equation known the Zerilli for arrive even perturbations: a solution provided divergence on equation theatsource term are satisfied. Aftersubstitutions some substitutions and otherasconditions simplifications we at the second-order After some and otherthe simplifications we can arrive the can second-order After some substitutions and otheras simplifications weperturbations: can arrive at theperturbations: second-order 2known Schrödinger equation known asdthe Zerilli equation for even Schrödinger equation the Zerilli equation for even Ψ even 2 equation for even perturbations: Schrödinger equation known as the+Zerilli k (r)Ψeven = 0. ( 2 ∗2 d Ψeven d22Ψ dr even 2 + k (r)Ψ (84) even = 20. d Ψ ∗2 even (84) 2 + k (r)Ψeven = 0. dr + k (r)Ψ 0. the time-domain as (84) ∗2be even = in dr This can also written ∗2 dr This can also be written in the time-domain as In time domain 2 Thiscancan also be2 written in thedtime-domain as This also be written in2 Ψ the time-domain d Ψevenas even 2 d Ψeven d Ψeven − = 0, + V (r)Ψeven = 0, ( 2− 2 2+ V (85) ∗2(r)Ψ 2 even dt2 d Ψ d Ψ dr ∗2 2 even even dΨ d+ Ψeven dr dt− even V (r)Ψ = 0, (85) − +even V (r)Ψ (85) ∗2 2 even = 0, dr dt ∗2 2 with dr dt with with ! " # # !3 ! "$ "# # ! $ "$ $ (L−1)L(L+1)(L+2) 3M 3 2M 1 72M 12M 3M with (L−1)L(L+1)(L+2) 2M 1 72M 12M " # # ! "$ $ V (r) = 1 − !rV (r) − (L − 1)(L + 2) 1 − + , 3 − r3 (L 1)(L +(L−1)L(L+1)(L+2) + , λ2 = r 511 − r −3M r 22) 1 − 2 5 2 72Mrr3 12M with λ r r r V (r) = 1 − − (L − 1)(L + 2) 1 − + , ! 2M " # # ! "$ $ 2 r λ2 r5 r3 3 r 2M 1 72M 12M 3Mr and V (r) = 1 − r − r3 (L − 1)(L + 2) 1 − r + (L−1)L(L+1)(L+2) , 2 5 λ r r2 and and 6M . λ = L(L + 1) − 2 + r and 6M . + 1) − 2 + 6M . λ = L(L + 1) − λ 2=+L(L r r L = 0 addition of mass L = 1 shift of the cm λ = L(L + 1) − 2 + 6M . r the solution is time independent. From the Zerilli perspective, this addition m0 γ0 is the mass-energy of the falling particle. For L = 1, the nonzero hµν can be removed by a gauge transformation which can be interpreted by a distant observer as a shift of the origin of the coordinate system. When the perturbation is at large r, the shift looks like a transformation to the centerof-momentum system where the particles orbit each other with distances from the center of momentum which are in inverse proportion to their relativistic masses. In the static case where k = 0, one of the unknown functions vanishes, namely H1 . For L = 0 the solution is trivial; the difference between the Schwarzschild metric mass and the new mass with addition δm. For L = 1, we find a solution that corresponds to a displacement of the center of attraction by the amount δz. Solutions for L>=2 Radiation not solve 5.8. Solutions for L ≥Can 2 and Gravitational Waves the equations explicitly For L ≥ 2, we can not solve the equations explicitly. Asymptotically for large r in the radiation gauge, the perturbation hµν in the metric is at the sum of two transverse traceless Asymptotically large r the perturbation tensor harmonics listed above as d and f. Asymptotically for large r the radial factors sum or two traces tensor harmonics. have the form M (m) hM 0L (ω, r) ≈ −rAL (ω)eiωr is the ∗ (86) Using a Green's function formed from high and frequency-limit solutions, we obtain amplitudes for M (e) KLM (ω, r) ≈ AL ingoing (ω)eiωr . the r=2M and outgoing (87) r=infinity radiation for radiation, a particle falling radially The amount of the escaping which is of more physical interest, is determined into the black M (m) M (e) completely by the amplitude coefficients AL and AL . hole. ∗ Using a Green’s function formed from the high frequency-limit solutions of the homogeneous equations, we obtain amplitudes for the ingoing (at r = 2M ) and amplitude peaks attheapproximately 3/16piM outgoing (at r = ∞) The radiation for a particle falling radially into black hole. The amplitude, as a function of frequency ω, increases like a power of ω. Amplitude total energy Integrating this, the lawestimated peaks at approximately ω = 3/16πM , and decreases exponentially for high frequencies. radiated is Integrating this, the estimated total energy radiated is (1/625)(m2o /M ) times a factor of order 1. To determine theTo distribution of the energy in time rather than frequency, must determine distribution inwe time use Fourier form the Fourier integrals for even and odd waves, and construct the stress-energy tensor from these time-dependent fields. static for As VishveshwaraNo indicates, there seemperturbations to be no static perturbations for L L>=2 ≥ 2 on Stability The Schwarzschild metric background gives an equilibrium state. If the metric is perturbed, however, will it remain stable? The collapsed Schwarzschild metric must be proven to be stable against small perturbations. A problem with coordinates chosen by Regge-Wheeler prevented from judging whether any divergence shown by the perturbations at the surface was real or due to the coordinate singularity at r=2M. Using new Kruskal coordinates, Vishveshwara was able to determine background metric finite at the surface and the divergence of the perturbations with imaginary frequency time dependence violate the small perturbation assumption. Thus perturbations with imaginary frequencies are physically unacceptable and the metric is indeed stable. on a spherically symmetric background metric into its normal modes using the tensor that the background metric is indeed finite at the surface and spherical harmonics discussed above. A problem with the coordinates chosen by perturbations with imaginary frequency time dependence violate them, however, prevented them from judging whether any divergence shown by the perturbations withtoimaginary frequencies are p perturbations atassumption. the surface (rThus = 2M ) was real or due the coordinate singularity andthe thenew metric is indeed stable. at r = 2M . Using Kruskal coordinates, Vishveshwara was able to determine that the background metric is indeed finite at the surface and the divergence of the perturbations with imaginary frequency time dependence violate the small perturbation 5.10. Quasi-Normal Modes assumption. Thus perturbations with imaginary frequencies are physically unacceptable The Transmission-Reflection Perspective and the metric is5.11. indeed stable. ! Newman-Penrose Formalism x if x ≥ 0; The second 5.10. popular method solving perturbation Quasi-Normal Modes for |x| = −x if x < 0. equations is the Newman-Penrose (NP) 5.11. The Transmission-Reflection Perspective formalism. ! x if x ≥ 0; 6. Newman-Penrose Formalism |x| =a is The NP formalism notation −x if x < 0. for writing various quantities and equations that appear insolving relativity. The second popular method for perturbation equations is 6. Newman-Penrose Formalism It starts by considering a complex tetrad (NP) formalism. The NPnull formalism is a notation for writing such that equations that appear in relativity. It isstarts by considering The second popular method for solving perturbation equations the Newman-Penrose − → − − → → − → ( l , n , m, m) suchisthat, (NP) formalism. The NP formalism a notation for writing various quantities and equations that appear in relativity. It starts by− considering a complex null tetrad − → − − → → → − → − l · n = 1 = − m · m. − → → ( l ,→ n,− m, m) such that, − →A− − → → is for the directional derivatives along l ·→ nnotation = 1 = −− m introduced · m. (88) tet A notation is introduced for the D directional vectors: = lµ ∂µ , derivatives ∆ =along nµ ∂µtetrad , δ = mµ ∂µ , A Summary of the Black Hole Perturbation Theory 21 µ = l ∂α, ∆κ, =λ, n µ, ∂µ ,ν, π, (used δσ,=τmare ∂µheavily , notation δ ∗ = for min ∂the The projections ofD And the tensor µ . spin (89) µ ,Weyl β, γ, &, ρ, a coeffic The projections of the Weyl tensor (used heavily in NP formalism in place of Gµν NP formalism in place of G.. and R..) then become and Rµν ) then become And α, β, γ, &, κ, λ, µ, ν, π, ρ, σ, τ are a notation for the spin coefficients of the null tetrad. µ µ µ Ψ0 = −Cµνρσ lµ mν lρ mσ Ψ1 = −Cµνρσ lµ nν lρ mσ Ψ2 = −Cµνρσ lµ mν mρ nσ Ψ3 = −Cµνρσ lµ nν mρ nσ Ψ4 = −Cµνρσ nµ mν nρ mσ . 7. Kerr Perturbations and Rµν ) then become σ µ ν ρ σ Ψ0 = −Cµνρσ lµ mν lρofmthe Ψ1 Hole = −C l m Ψ2 = −Cµνρσ lµ mν mρ n21σ A Summary Black Perturbation µνρσ l n Theory ν ρ σ µ ν ρ σ Ψ3 = −CµνρσThe lµ nprojections m n of the Weyl tensor (used heavily Ψ = −C n m nof m 4 µνρσ in NP formalism in place Gµν . Kerr Perturbations and Rµν ) then become 7. Kerr Perturbations Ψ0 = −Cµνρσ lµ mν lρ mσ Ψ1 = −Cµνρσ lµ nν lρ mσ Ψ2 = −Cµνρσ lµ mν mρ nσ Due to the complexity of Ψ3 = −Cµνρσ lµ nνthe mρ nσ Kerr metric, it Ψ4 = −Cµνρσ nµ mν nρ mσ . becomes use the Einstein The Kerr spinning blackdifficult hole metricto is non-diagonal, so it hasequations cross-terms between Kerrget Perturbations directly7. to a solvable perturbation equation. spatial and time coordinates. Due to the complexity of the Kerr metric, it becomes Theequations Kerr spinning black hole metric is non-diagonal, so it has cross-terms between difficult to useTo the obtain Einstein directly to get a solvable perturbation equation. To the perturbation equation for rotating spatial and time coordinates. Due to the complexity of the Kerr metric, it becomes obtain the perturbation equation for black directly holes,Newman-Penrose the Newmanblack holes, used the difficult toTeukolsky use therotating Einstein equations toTeukolsky get a solvable used perturbation equation. To formalism. over much laborious calculation obtainSkipping the perturbation equation for rotating blackwe holes, Teukolsky usedTeukolsky the NewmanPenrose formalism. Skipping over much laborious calculation arrive at the Penrose Skipping over much laborious calculation we arrive at the Teukolsky we arrive at formalism. the Teukolsky equation. equation. ! equation. " ! " ! 2∂ 2 Ψ " 2 2 (r2 +a2 )2 ∂ !2 Ψ ∂ 2 Ψ" ∂ 2 Ψ a4M 2 (r2 +a2 )24M ar ar ∂ 2 Ψ1 a 1 ∂ Ψ 2 − a sin θ + + − − a sin θ + − 2 2 2 + 2 2 ∆ ∂t ∆ ∂t∂ψ ∆ ∆ θ ∂φ!sin θ ∂φ"2 $ $ ∆ # !sin $ " # $ ∂t −s ∂ #∆#s+1∂t∂ψ a(r−M ) ∂Ψ 1 ∂ ∂Ψ i cos θ ∂Ψ cos θ+ ∂Ψ −∆1 ∂r∂ ∆ ∂r ∂Ψ − sin θ ∂θ sin θa(r−M −)2s i ∆ −s ∂ s+1 ∂Ψ 2θ ∂θ ∂φ sin −∆ ∂r ∆ ∂r − sin!θM∂θ sin θ − 2s + " 2θ 2 2 ∂θ ∆ ∂φ sin (r " −a ) ∂Ψ 2 ! −2s − r − ia cos θ + (s cot θ − s)Ψ = 0, 2 2 ∆ ∂t M (r −a ) ∂Ψ 2 −2s − r − ia cos θ ∂t + (s cot θ − s)Ψ = 0, ∆ where where once again where once again ∆(r) = r2 − 2GM r + a2 . (90) −1 2 If s = 2 to 2, Ψ =Ψ 0 and if s =achieve −2, Ψ = ρ−4 Ψangular . While not possible separation in (90) 4 where ρ = (r−ia cos θ) ∆(r) = r − 2GM r + a . This equation more involveddomain than the Zerilli While it is the time domain, inis considerably the frequency it equation. is not possible to achieve −1 time domain, in the frequency domain −4 angular separation in the separable. If s = 2, Ψ =Ψ 0 and if sit=is −2, Ψ = ρ Ψ4 where ρ = (r−ia cos θ) . separable. If you assume This equation is considerably more−iωtinvolved than the Zerilli equation. While it is imφ Ψ=e e S(θ, ω)R(r, ω), (91) not possible to achieve angular separation in the time domain, in the frequency domain M 2 Connections When a=0 in the Teukolsky equation you are then left with the Bardeen-Press equation for Schwarzchild black holes. The Bardeen-Press equation contains in its real and imaginary parts the Zerilli and the Regge-Wheeler equations respectively. References [1] B.F. Whiting, Class Notes from General Relativity I (Transcribed by various students - usually Shawn Mitryk, Gainesville, 2008). [2] S.M. Carroll, Spacetime and Geometry (Addison-Wesley, San Fransisco, 2004). [3] J.B. Hartle, Gravity (Addison-Wesley, San Fransisco, 2003). [4] B.F Schutz, A First Course in General Relativity (Cambridge University Press, Cambridge, 1985). [5] B.F Schutz, Geometrical Methods of Mathematical Physics (Cambridge University Press, Cambridge, 1980). [6] S. Chandrasekhar, The Mathematical Theory of Black Holes (Oxford University Press, New York, 1992). [7] S. Chandrasekhar, Selected Papers, Volume 6: The Mathematical Theory of Black Holes and of Colliding Plane Waves (University of Chicago Press, Chicago, 1991). [8] D.J. Griffiths, Introduction to Electrodynamics, Third Edition (Prentice-Hall, Upper Saddle River, 1999). [9] D.J. Griffiths, Introduction to Quantum Mechanics, Second Edition (Prentice-Hall, Upper Saddle River, 2005). [10] T. Regge and J.A. Wheeler, Stability of a Schwarzschild Singularity Phys. Rev. 108, 1063 (1957). [11] F.J. Zerilli, Gravitational Field of a Particle Fal ling in a Schwarzschild Geometry Analyzed in Tensor Harmonics Phys. Rev. D 2, 2141 (1970). [12] F.J. Zerilli, Effective Potential for Even-Parity Regge-Wheeler Gravitational Perturbation Equations Phys. Rev. Lett 24, 737 (1970). [13] C.V. Vishveshwara, Stability of the Schwarzschild Metric Phys. Rev. D 1, 2870 (1970). [14] S. Chandrasekhar, On the Equations Governing the Perturbations of the Schwarzschild Black Hole Proc. R. Soc. 343, 289 (1975). [15] S. Chandrasekhar, and S. Detweiler, The Quasi-Normal Modes of the Schwarzschild Black Hole Proc R. Soc. 344, 441 (1975). [16] S. Chandrasekhar, On One-Dimensional Potential Barriers Having Equal Reflection and Transmission Coefficients Proc. R. Soc. 369, 425 (1980). [17] S. Chandrasekhar, and S. Detweiler, On the Equations Governing the Axisymmetric Perturbation of the Kerr Black Hole Proc R. Soc. 345, 145 (1975). [18] J.M. Bardeen, and W.H. Press, Radiation Fields in the Schwarzschild Background J. Math. Phys. 14, 7 (1972). [19] J.M. Stewart M. Walker, Perturbations of Space-Times in General Relativity Proc. R. Soc. 341, 49 (1974). [20] S.A. Teukolsky, Perturbations of a Rotating Black Hole Astrophys J. 185, 635 (1973).