THE ENERGY-MOMENTUM PSEUDOTENSOR Tµν of matter satisfies the (covariant) divergenceless equation Tµν ;ν = 0 (3) We know it is not a conservation law, because it cannot be written as an ordinary divergence. In a locally inertial frame (LIF): eq. (3) becomes ∂T µν = 0, (4) ∂xν this means that T µν can be written as: ∂ (5) T µν = α η µνα , ∂x where η µνα is antisymmetric in ν and α; INDEED ∂ 2η µνα = 0, ∂xν ∂xα because the derivative operator is symmetric in ν and α We want to find the expression of η µνα : write Einstein eqs. 4 8πG c 1 Rµν − g µν R . Gµν = 4 T µν −→ T µν = (6) c 8πG 2 In a LIF Rµν is: 2 2 2 2 ∂ g 1 ∂ g ∂ g ∂ g γβ αδ γδ αβ . Rµν = gµα gνβ gγδ + − − 2 ∂xα ∂xδ ∂xγ ∂xβ ∂xα ∂xβ ∂xγ ∂xδ By replacing in eq. (6), T µν becomes 4 c 1 ∂ ∂ µν αβ µα νβ Tµν = (−g) g g − g g (7) 16πG (−g) ∂xβ ∂xα The part within { } is antisymmetric in ν and α, symmetric in µ and ν, and it is the quantity η µνα we were looking for. ∂ µνα η , ∂xα Tµν = η µνα = 4 c 1 ∂ (−g) gµν gαβ − gµαgνβ β 16πG (−g) ∂x Since in a LIF gµν,α = 0 we can extract write this equation as (−g)T µν ∂ζ µνα = , ∂xα 1 (−g) and (8) where ζ µνα = (−g)η µνα c4 ∂ µν αβ µα νβ . (9) (−g) g g − g g = 16πG ∂xβ EQ. (8) has been derived in a locally inertial frame. µνα In any other frame (−g)Tµν will not equate ∂ζ∂xα , therefore, in a generic frame ∂ζ µνα − (−g)Tµν 6= 0. α ∂x We shall call this difference (−g)t µν (−g)tµν , i.e. ∂ζ µνα µν = − (−g)T . ∂xα The quantities tµν are symmetric, because T µν and ∂ζ µνα are symmetric in µ and ν. It follows that ∂xα (−g) (T µν ∂ζ µνα +t )= ∂xα , µν −→ ∂ µν µν [(−g) (T + t )] = 0, ∂xν this is THE CONSERVATION LAW OF THE TOTAL ENERGY AND MOMENTUM OF MATTER + GRAVITATIONAL FIELD VALID IN ANY REFERENCE FRAME. (−g)t If we express stein’s eqs. G µν µν ∂ζ µνα = − (−g)Tµν . α ∂x T µν in terms of gµν and by using Ein- 8πG = 4 T µν c −→ T µν c4 µν 1 µν R − g R . = 8πG 2 and eq. (9) it is possible to show that tµν can be written as follows t µν = + + + c4 δ µα νβ µν αβ σ δ σ δ σ 2Γ αβ Γ δσ − Γ ασ Γ βδ − Γ αδ Γ βσ g g − g g 16πG g µα g βδ (Γν ασ Γσ βδ + Γν βδ Γσ ασ − Γν δσ Γσ αβ − Γν αβ Γσ δσ ) g να g βδ (Γµασ Γσ βδ + Γµβδ Γσ ασ − Γµδσ Γσ αβ − Γµαβ Γσ δσ ) αβ δσ µ ν µ ν g g (Γ αδ Γ βσ − Γ αβ Γ δσ ) This is the stress-energy pesudotensor of the gravitazional field. tµν it is not a tensor because : 1) it is the ordinary derivative, (not the covariant one) of a tensor 2) it is a combination of the Γ’s that are not tensors. However, as the Γ’s, it behaves as a tensor under a linear coordinate transformation. Let us consider an emitting source and the associated 3-dimensional coordinate frame O (x, y, z). Be an observer located at P = (x1, y1, z1) at a distance √ 2 r = x1 + y12 + z12 from the origin. The observer wants to detect the wave coming along r the direction identified by the versor n = |r| . y y’ P x’ n x z z’ Consider a second frame O’ (x0, y 0 , z 0 ), with origin coincident with O, and having the x0-axis aligned with n. Assuming that the wave traveling along x0 direction is linearly polarized and has only one polarization, the corresponding metric tensor will be g µ0 ν 0 (t) −1 = 0 0 0 (x) (y) (z) 0 0 0 , 1 0 0 0 TT 0 [1 + h+ (t, x )] 0 TT 0 0 0 [1 − h+ (t, x )] The observer wants to measure the energy which flows per unit time across the unit surface orthogonal to x0, 0 i.e. t0x , therefore he needs to compute the Christoffel symbols i.e. the derivatives of hTµ0Tν 0 . The metric perx0 turbation has the form hT T (t, x0) = const · f (t − 0 x c ), the only derivatives which matter are those with respect to time and x0 ∂hTT const ˙ ≡ ḣTT = f, ∂t x0 1 TT const const 0 1 const ˙ ∂hTT TT 0 f = − ≡ h = − f + f ∼ − ḣ , ∂x0 x02 x0 c x0 c where we have retained only the dominant 1/x0 term. Thus, the non-vanishing Christoffel symbols are: Γ0y0y0 = −Γ0z0 z0 = 0 0 Γx y0y0 = −Γx z0 z0 = 1 2 1 2c 0 0 Γy 0y0 = −Γz 0z0 = ḣT+T 0 1 TT ḣ 2 + 0 Γy y0x0 = −Γz z0 x0 = − ḣT+T 1 TT ḣ . 2c + By substituting Christoffel’s symbols in tµν , we find t0x ≡ Thus 2 TT 0 c dh (t, x ) dEGW = dx0 dS 16πG dt 3 TT 0 dEGW c dh (t, x ) = dtdS 16πG dt 2 2 . . In general, if both polarization are present g µ0 ν 0 0 t0x = (t) −1 = 0 0 0 2 (x) (y) (z) 0 0 0 , 1 0 0 0 0 TT TT 0 [1 + h+ (t, x )] h× (t, x ) 0 0 TT TT 0 h× (t, x ) [1 − h+ (t, x )] TT 2 dh+ c 16πG dt TT 2 dh × + dt 2 = TT 2 X dhjk c 32πG jk dt . This is the energy per unit time which flows across a unit surface orthogonal to the direction x0. However, the direction x0 is arbitrary; if the observer il located in a different position and computes the energy flux he receives, he will find formally the same but with hTjkT referreed to the TT-gauge associated with the new direction. Therefore, if we consider a generic direction ~r = r~n t0r = 2 TT X dhjk (t, r) 2 c 32πG jk dt . (10) Since in GR the energy of the gravitational field cannot be defined locally, to find the GW-flux we need to average over several wavelenghts, i.e. 3 TT 2 + X dhjk dEGW c = ct0r = . dtdS 32πG jk dt We shall now express the energy flux directly in terms of the quadrupole moment. * Since h̄TTµ0 = 0, µ = 0, 3 r 2G d2 TT ) (t − h̄ik (t, r) = 4 · 2 QTT ik c r dt c by direct substitution we find 2 + dEGW c3 *X dhTT jk = dtdS 32πG jk dt * " ... G r !#2 + TT X Qjk t − = 8πc5 r2 jk c * " ... r !#2+ G X Pjkmn Qmn t − = . 8πc5 r2 jk c From this formula we can compute the gravitational GW luminosity LGW = dEdt LGW = Z Z dE dEGW GW 2 dS = r dΩ dtdS dtdS * G 1 Z X = dΩ 2c5 4π jk ... Pjkmn Qmn r !!2+ t− . c Let us compute this integral. By using the properties of Pmnjk we find ... = = ... ... ... Pmnjk Qjk Pmnrs Qrs = PjkrsQjk Qrs ... ... 1 δjr δks − njnrδks − nknsδjr + 2 njnknrns QjkQrs ... ... ... ... ... ... QjkQjk − 2nkQkrQrsns + 21 njnknrnsQjk Qrs . The integrals of the n’s over the solid angle are: 1 Z 1 dΩninj = δij 4π 3 1 Z 1 dΩninj nr ns = (δij δrs + δir δjs + δis δjr ) 4π 15 so that 1 R 4π ... ... ... ... dΩ Qjk Qjk − 2nkQkrQrsns + = ... ... 2 Q Q 5 jk jk ... ... 1 nn nnQ Q 2 j k r s jk rs and, finally, the power emitted in gravitational waves by an evolving source is LGW dEGW G *X ... r !+ r ! ... Q t − Qjk t − = = 5 . dt 5c jk jk c c where Qij = qij − 1 δ ij qkk 3 We shall now compute the GW-luminosity of a binary system We shall use the formula: LGW G *X ... ... + Q Qjk . = 5 5c jk jk where µ 1 Qij = qij − δij δ kl qkl = l20 2 2 ωK t sin 2ω ωK t 0 cos 2ω ωK t −cos cos 2ω ωK t 0 sin 2ω 0 0 0 and µ Qij = l20 8 ωK 3 2 ... X jk ... ... ωK t sin 2ω −cos ωK t cos 2ω 0 2 Q Qjk = 32 µ l40 ωK t 0 − cos 2ω ωK t 0 sin 2ω −sin 0 0 ωK 6 jk ωK = v u u u t GM l30 M3 = 32 µ G l50 2 3 By direct substitution we find dEGW 32 G4 µ2M3 LGW ≡ = dt 5 c5 l50 and if m1 = m2 = m and consequently µ = 2m 64G4m5 LGW = 5c5l50 For the binary pulsar PSR 1913+16 LGW = 0.7 · 1031 erg/s m 2 and M = Since we know how much energy is radiated in GW, we can compute the consequent variation of the orbital period The total energy of the system is 1 Eorb = µωK 2 l20 − U, 2 therefore where Eorb U=− µM Gµ Gm1m2 =− l0 l0 µM 1 Gµ =− 2 l0 and µM 1 dl0 1 dl0 dEorb 1 Gµ = = −Eorb dt 2 l0 l0 dt l0 dt Since ωK = lnGM−3lnl0 → → 2lnω ωK 2 = GMl−3 0 ωK 1 dω 3 1 dl0 =− ωK dt 2 l0 dt and since ωK 1 dω 1 dT 1 dl0 2 1 dT =− → = ωK dt T dt l0 dt 3 T dt and dEorb 2 Eorb dT dT 3 T dEorb =− → =− dt 3 T dt dt 2 Eorb dt The energy lost in GWs must be compensated by a variation of orbital energy (adiabatic approximation dEorb + LGW = 0 dt → dEorb = −LGW dt we find dT 3 T LGW = dt 2 Eorb dT 3 T = LGW dt 2 Eorb For PSR1913+16 T = 27907 s, Eorb ∼ −1.4 · 1048 ergs, LGW ∼ 0.7 · 1031 erg/s with these data we find dT ∼ −2.2 10−13 dt By refining the calculations, using the equations of motion appropriate for an eccentric orbit with ' 0.617 one finds dT = −2.4 · 10−12 dt The observed value is dT = −2.3 · 10−12 (±0.22 · 10−12) dt FIRST INDIRECT EVIDENCE OF THE EXISTENCE OF GRAVITATIONAL WAVES For PSRJ0737-3039 T = 8640 s, Eorb ∼ −2.55 · 1048 ergs, LGW ∼ 2.24 · 1032 erg/s dT ∼ −1.2 10−12 dt ORBITAL EVOLUTION ωK 1 dω 1 dT =− ωK dt T dt dT 3 T = LGW dt 2 Eorb moreover, LGW dEGW 32 G4 µ2M3 = ≡ dt 5 c5 l50 Eorb = − µM 1 Gµ 2 l0 therefore ωK 1 dω 96 G3 µM2 = . ωK dt 5 c5 l40 When integrated it gives ωK (t) = 3/8 ωK in tcoal [tcoal − t] in 4 5 c l0 256 G3 µM2 5 where ωK in = ωK (t = 0) 3/8 tcoal = (11) lin 0 = l0 (t = 0) (12) and The orbital frequency, and consequently the frequency of the emitted wave ν GW = ωK /π, change accordingly: ν GW (t) = 3/8 ν in GW tcoal 3/8 [tcoal − t] (13) the frequency increases with time Since ω K = is s GM , l30 using eq. (11) the orbital separation l0(t) = lin 0 1/4 tcoal [tcoal − t]1/4 the orbital distance decreases with time. (14) WAVEFORM: AMPLITUDE AND PHASE We have seen that if a binary system moves on a circular orbit and we look, for example, in the direction orthogonal to the orbital plane, the wave we would detect is hTT ij µMG2 TT 4µ =− Akl rl0c4 and ωK = s GM l30 ATT ij ωK t sin 2ω ωK t 0 cos 2ω sin 2ω ω cos ω = t −cos 2ω t 0 K K 0 0 0 = π ν GW . We shall model the waveform emitted in the inspiralling as follows: 1) an instantaneous amplitude µMG2 ωK 2/3(t) µMG2 4µ 4µ = · 1/3 1/3 h0(t) = rl0(t)c4 rc4 G M 4π 2/3 G5/3 M5/3 2/3 = ν GW (t) c4 r where M5/3 = µ M2/3 → M = µ3/5 M2/5 = chirp mass Since the frequency increases in time the amplitude increases too. 2) Since the wave frequency is changing in time, the ωK t, will be replaced by phase appearing in ATT ij , i.e. 2ω Φ(t) = Z t 2πνν GW (t) dt + Φ in , where Φ in = Φ(t = 0) The phase Φ(t) = Z t 2πνν GW (t) dt + Φ in , Since ν GW (t) = and Φ in = Φ(t = 0) 3/8 ν in t coal GW [tcoal − t]3/8 3/8 ν in tcoal = 53/8 where 1 8π 5/8 3 c GM then 1 ν GW (t) = 8π 3 5/8 c GM 5 tcoal − t 3/8 and the integrated phase will be 5/8 c3 + Φin Φ(t) = 2 5GM (tcoal − t) If we know the phase we can measure the chirp mass Thus, the signal emitted during the inspiralling will be hTT ij 4π 2/3 G5/3 M5/3 2/3 TT = ν (t)A GW ij c4 r where ATT ij = sin Φ(t) 0 cos Φ (t) sin Φ (t) −cos cos Φ(t) 0 0 0 0 LIGO[40 Hz − 1 − 2 kHz] LISA[10−4 − 10−1] Hz VIRGO[10 Hz − 1 − 2 kHz] 1e-21 [Hz -1/2 ] 1e-20 Sh 1/2 GEO 1e-22 VIRGO LIGO 1e-23 10 100 1000 10000 log νGW (Hz) Let us consider 3 binary system a) m1 = m2 = 1.4 M b) m1 = m2 = 10 M c) m1 = m2 = 106 M Let us first calculate what is the orbital distance between the two bodies on the innermost stable circular orbit (ISCO) and the corresponding emission frequency lI0SCO ∼ 6GM , c2 ωK = v u u u t v u u u u t GM 1 GM ISCO ν ν → = = π GW GW l30 π (lISCO )3 0 ν GW = 1570.4 Hz a) l0 ISCO = 24, 8 km b) l0ISCO = 177, 2 km ν GW = 219.8 Hz ν GW = 2.2·10−3 Hz c) l0ISCO = 17.720.415, 3 km a) and b) are interesting for LIGO and VIRGO, c) will be detected by LISA Let us consider LIGO and VIRGO, and let us compute the time a given signal spends in the detector bandwidth before coalescence. From 3/8 ν in GW tcoal ν GW (t) = [tcoal − t]3/8 we get 8/3 in ν GW t = tcoal 1 − . ν GW (t) Putting : ν in GW = lowest frequency detectable by the antenna, and ISCO ν max we find GW = ν GW (LIGO) a) [40 − 1570.4 Hz] ∆t = 24.86 s b) [10 − 219.8 Hz] ∆t = 0.93 s (VIRGO) [10 − 1570.4 kHz] ∆t = 16.7 m [10 − 219.8 kHz] ∆t = 37.82 s VIRGO catches the signal for a longer time! WHAT ABOUT LISA? LISA [10−4 − 10−1] Hz 1e-20 Detection threshold LISA 1-yr observation 1e-21 1e-22 1e-23 1e-24 0.0001 0.001 0.01 ν [ Hz ] 0.1 1 Let us consider 2 BH-BH binary systems a) m1 = m2 = 102 M b) m1 = m2 = 106 M Orbital distance between the two bodies on the innermost stable circular orbit (ISCO) and the corresponding emission frequency lI0SCO ∼ 6GM , c2 ωK = v u u u t v u u u u t GM 1 GM ISCO ν ν = π → = GW GW l30 π (lISCO )3 0 a) lISCO = 1772 km 0 = 17.720.415, 3 km b) lISCO 0 ν GW = 21.98 Hz ν GW = 2.2 10−3 Hz Time a given signal spends in the detector bandwidth before coalescence. a) m1 = m2 = 102 M b) m1 = m2 = 106 M t = tcoal 1 − ν in ν GW (t) 8/3 LISA a) [10−4 − 10−1 Hz] ∆t = 556.885 years b) [10−4 − 2.2 · 10−3 Hz] ∆t = 0, 12 years = 43 d 18 h 43 m 24 s
© Copyright 2024 Paperzz