Gravitational radiation and the Bondi mass
National Center for Theoretical Sciences, Mathematics Division March 16th
, 2007
Wen-ling Huang Department of Mathematics University of Hamburg, Germany
Structuring
1. Gravitational waves
2. Bondi’s radiating space-time
3. Positive mass theorem at null infinity (a) Schoen-Yau’s method
Structuring
1. Gravitational waves
2. Bondi’s radiating space-time
3. Positive mass theorem at null infinity (a) Schoen-Yau’s method
Gravitational waves (GW)
• predicted by Einstein’s general relativity
• time dependent solutions of Einstein’s field equations which radiate or transport energy
• created by accelerated masses • radiate with the light speed
Examples for GW
• Binary pulsars
• Collapsing black holes • Double star
Detection of GW
Gravitational waves encode the history of the universe.
However, they are very weak and have not been detected yet.
How weak are they ?
For black holes that weigh about 10 times as much as the sun and which are a billion light-years away, the wave strength is about 10−21 when they arrive at the Earth.
Therefore the waves produce tides in the earth’s ocean by 10−21· 107 = 10−14 meter, or 10 times the diameter of an atom’s nucleus. There exist several GW detectors on the world.
GW Detectors
• LIGO – USA
• VIRGO – France + Italy
• Geo 600 – Germany and GB • TAMAX – Japan
Detection of GW (2)
Although they have not been detected yet, the existence of grav-itational waves has been proved indirectly from observations of the pulsar PSR 1913+16. This rapidly rotating binary system should emit gravitational radiation, hence lose energy and rotate faster. The observed relative change in period agrees remarkably with the theoretical value.
GW and energy
A fundamental conjecture is that gravitational waves can not carry away more energy than they have initially in an isolated gravitational system.
Structuring
1. Gravitational waves
2. Bondi’s radiating space-time
3. Positive mass theorem at null infinity (a) Schoen-Yau’s method
Bondi’s radiating space-time (1) Historical notes
• Late 1950s-early 1960s:
Pirani, Bondi, Robinson, Trautman and others • Bondi, Van der Burg, and Metzner:
– vacuum solutions of Einstein’s field equations outside an isolated (i.e. spatially bounded) axisymmetric system
– definition of mass along outgoing null hypersurfaces in the limit r → ∞
– the total mass, measured at null infinity, is non-increasing with respect to the retarded time
Ref.: H. Bondi, M. van der Burg, A. Metzner, Gravitational waves in general relativity VII. Waves from axi-symmetric isolated systems, Proc. Roy. Soc. London A 269(1962), 21-52.
Bondi’s radiating space-time (2) Historical notes
• Sachs 1962:
Generalization of the work of Bondi, Van der Burg, Metzner to asymptotically flat space-times
Ref.: R. Sachs, Gravitational waves in general relativity VIII. Waves in asymptotically flat space-time, Proc. Roy. Soc. London, A 270(1962), 103-126.
Bondi’s radiating space-time (3) Historical notes
Interpretation of Bondi mass as the total mass of the isolated physical system measured after the loss due to the gravita-tional radiation up to that time
Bondi’s radiating space-time (4) Retarded time
Minkowski space-time:
• photon emitted when t = t0 takes a certain amount of time to
reach an observer located at distance r ≥ 0 from the source, so the observer notices it when t = t1,
t1 = t0 + r
c = t0 + r (where speed of light c := 1) • the time t0 = t1 − r is defined as retarded time.
• set u = t − r:
the hypersurface defined by u = t − r = k, k a constant, is the future directed null cone with vertex r = 0, t = k.
Bondi’s radiating space-time (5) Bondi’s radiating space-time
Vacuum space-time (Tij = 0) with Bondi’s radiating metric
gBondi =
+ r2e2γU2cosh 2δ + r2e−2γW2cosh 2δ + 2r2U W sinh 2δ − V r e
2β
du2 −2e2βdu dr − 2r2e2γU cosh 2δ + W sinh 2δdu dθ
−2r2e−2γW cosh 2δ + U sinh 2δsin θ du dϕ
+r2e2γ cosh 2δ dθ2 + e−2γ cosh 2δ sin2θ dϕ2 + 2 sinh 2δ sin θ dθ dϕ
• β, γ, δ, U, V , W : functions of (u, r, θ, ϕ) which are smooth for 0 < r0 ≤ r, 0 ≤ θ ≤ π, 0 ≤ ϕ ≤ 2π
• u: retarded time, r: radius function, θ, ϕ: spherical coordinates.
Bondi’s radiating space-time (6) Examples
(i) Minkowski space-time:
gMink = −dt2 + dr2 + r2 dθ2 + sin2 θdϕ2.
Using the retarded time u = t − r, the metric can be written as
Bondi’s radiating space-time (7) Examples
(ii) Schwarzschild space-time:
gSchw = − 1 − 2m r dt
2 + dr2
1 − 2mr + r
2 dθ2 + sin2 θdϕ2,
where m is the mass.
Using the retarded time u = t−r −2m ln r−2m , the metric can be written as gSchw = − 1 − 2m r du 2 − 2dudr + r2 dθ2 + sin2 θdϕ2.
Bondi’s radiating space-time (8) Examples
(iii) Kerr space-time: gKerr = − 1 − 2mr Σ dt2 − 4mar sin 2 θ Σ dtdφ + Σ 4dr 2 + Σdθ2 + r2 + a2 + 2mra2 sin2 θ Σ2 sin2 θdφ2 where Σ ≡ r2 + a2 cos2 θ, 4 ≡ r2 − 2mr + a2,
m is the mass, ma is the angular momentum as measured from infinity.
Retarded time u ? It is an interesting problem to obtain the Kerr solution in the form of a Bondi metric.
Bondi’s radiating space-time (9)
The outgoing radiation conditions imply, as r → ∞, γ = c(u, θ, ϕ) r + O(r −3), δ = d(u, θ, ϕ) r + O(r −3), β = −c 2 + d2 4r2 + O(r −4), U = −l(u, θ, ϕ) r2 + O(r −3), W = −¯l(u, θ, ϕ) r2 + O(r −3), V = −r + 2M(u, θ, ϕ) + O(r−1),
Bondi’s radiating space-time (10)
Bondi’s radiating metric under the outgoing radiation conditions
gBondi = −1− 2M r + O(r −2)du2 −21 − c 2 + d2 4r2 + O(r −4)dudr +2l + 2cl + 2dl r + O(r −2)dudθ +2¯l− 2c¯l− 2dl r + O(r −2)sin θdudϕ +r21 + 2c r + O(r −2)dθ2 +r21− 2c r + O(r −2)sin2 θdϕ2 +r24d r + O(r −2)sin θdθdϕ2 ,
Bondi’s radiating space-time (11) We have two regularity assumptions
Condition A
Each of the six functions β, γ, δ, U , V , W together with its derivatives up to the second orders are equal at ϕ = 0 and 2π. Condition B
For all u and ϕ,
Bondi’s radiating space-time (12) At null infinity:
The Bondi energy-momentum of u0-slice: mν(u0) = 1 4π I S2 M(u0, θ, ϕ)nνdS for ν = 0, 1, 2, 3, where
n0 = 1, n1 = sin θ cos ϕ, n2 = sin θ sin ϕ, n3 = cos θ. m0: “Bondi energy” or “Bondi mass”
mi: “Bondi momentum”
• Minkowski space-time: mν(u0) = 0.
Bondi’s radiating space-time (13) We define M(u, θ, ϕ) = M (u, θ, ϕ) − 1 2 l,2 + l cot θ + ¯l,3 csc θ = M (u, θ, ϕ) − 1 2 − 2c(u, θ, ϕ) + c,22(u, θ, ϕ) − csc2 θ c,33(u, θ, ϕ) + 2 csc θ d,23(u, θ, ϕ)
+3 cot θ c,2(u, θ, ϕ) + 2 cot θ csc θ d,3(u, θ, ϕ). Its u-derivative is
Bondi’s radiating space-time (14) Under condition A and B, we have
I S2 l,2 + l cot θ + ¯l,3 csc θ nνdS = 0. This implies mν(u0) = 1 4π I S2 M(u0, θ, ϕ) nν dS = 1 4π I S2 M(u0, θ, ϕ) nν dS. ν = 0, 1, 2, 3.
Bondi’s radiating space-time (15) Bondi mass loss formula
d du mν = − 1 4π I S2 (c,0)2 + (d,0)2nνdS, ν = 0, 1, 2, 3. When ν = 0, this is the famous Bondi mass loss formula
d du m0 = − 1 4π I S2 (c,0)2 + (d,0)2dS ≤ 0.
The Bondi mass is non-increasing in u, i.e., more and more energy is radiated away.
Bondi’s radiating space-time (16)
Generalized Bondi energy-momentum loss formula Applying H¨older’s inequality and Cauchy-Schwarz’ inequality, we obtain d du m0 − s X 1≤i≤3 m2i ≤ 0.
Bondi’s radiating space-time (17) Positive Mass Conjecture at Null Infinity
In Bondi’s vacuum radiating space-times, the Bondi mass must be nonnegative, i.e., a finite gravitational gravitational system cannot radiate away more energy than it has initially.
Structuring
1. Gravitational waves
2. Bondi’s radiating space-time
3. Positive mass theorem at null infinity (a) Schoen-Yau’s method
Positive mass theorem at null infinity (1) The Positive mass theorem near infinity
1982, Schoen-Yau:
Solving the Jang’s equation–prescribing the mean curvatures.
1982-, Israel-Nester, Horowitz-Perry, Ashtekar-Horowitz, Renla-Tod, Ludvigsen-Vickers, etc.:
Witten’s method–the Dirac operator.
2006, W.-l. Huang, S.T. Yau, and X. Zhang: Detailed proof.
Ref.: W.-l. Huang, S.T. Yau, and X. Zhang, Positivity of the Bondi mass in Bondi’s radiating spacetimes, Rend. Lincei. Mat. Appl. 17 (2006), 335-349.
Positive mass theorem at null infinity (2) Asymptotically null initial data set
In Minkowski space-time, the spacelike hypersurface t = p1 + r2
has the hyperbolic metric ˘g and the nontrivial second form ˘h, ˘ g = dr 2 1 + r2 + r 2 dθ2 + sin2 θdϕ2, ˘h = dr2 1 + r2 + r 2 dθ2 + sin2 θdϕ2
in polar coordinates (r, θ, ϕ) where 0 < r < ∞, 0 ≤ θ < π, 0 ≤ ϕ < 2π.
Positive mass theorem at null infinity (3) Asymptotically null initial data set
Denote the associated orthonormal frame {˘ei} and coframe {˘ei} by ˘ e1 = p1 + r2 ∂ ∂r, e˘ 1 = √ dr 1 + r2, ˘ e2 = 1 r ∂ ∂θ, e˘ 2 = rdθ, ˘ e3 = 1 r sin θ ∂ ∂ϕ, e˘ 3 = r sin θdϕ. ˘
Positive mass theorem at null infinity (4) Asymptotically null initial data set
An initial data set (M3, g, p) (p is not necessarily symmetric) is asymptotically null of order τ if, outside a compact subset,
• M is diffeomorphic to R3\BR,
• the metric g and the 2-tensor p are
g(˘ei, e˘j) = ˘g(˘ei, e˘j) + aij, p(˘ei, e˘j) = ˘p(˘ei, e˘j) + bij where aij and bij satisfy
aij = O(r−τ), bij = O(r−τ), ˘
∇kaij = O(r−τ), ∇˘ kbij = O(r−τ),
˘
Positive mass theorem at null infinity (5)
In Bondi’s radiating vacuum space-time, the spacelike hypersur-face given by u = p1 + r2 − r + (c 2 + d2) u=0 12r3 + a3(θ, ϕ) r4 + o(r −4)
is asymptotically null of order 1.
(Since Bondi’s radiating metric is complicated, we have calculated the induced metric g and the second fundamental form h using Mathematica 5.0.)
Structuring
1. Gravitational waves
2. Bondi’s radiating space-time
3. Positive mass theorem at null infinity (a) Schoen-Yau’s method
Schoen-Yau’s method (1) Jang’s equation: gij − f ifj 1 + |∇f|2 f,ij p1 + |∇f|2 − hij = 0.
If Jang’s equation has a solution f which has the asymptotic expansion
f = p1 + r2 + p(θ, ϕ) ln r + o(1)
for r sufficiently large, then p(θ, ϕ) and M(0, θ, ϕ) must be constant.
Schoen-Yau’s method (2) Proof. For r sufficiently large,
J(f ) ≈ ln r r3 4S2 p(θ, ϕ) + p(θ, ϕ) − 2M(0, θ, ϕ) r3 , where 4S2 = ∂2 ∂θ2 + cot θ ∂ ∂θ + csc 2 θ ∂2 ∂ϕ2. J(f ) = 0 implies 4S2 p(θ, ϕ) = 0, p(θ, ϕ) − 2M(0, θ, ϕ) = 0.
There is no nonconstant harmonic function on S2 ⇒ p(θ, ϕ) and M(0, θ, ϕ) = p(θ,ϕ)2 are constant.
Schoen-Yau’s method (3)
Theorem 1. Let L3,1, g˜ be a vacuum Bondi’s radiating space-time with Bondi-metric g. Suppose that Condition A˜ and Condition B hold.
If there exists u0 such that M(u0, θ, ϕ) is constant, then
m0(u) ≥ s X 1≤i≤3 m2i(u) for all u ≤ u0.
Schoen-Yau’s method (4) Idea of the proof:
• Let M(u0, θ, ϕ) = p2 be constant for some u0.
• Without loss of generality, u0 = 0.
• On the hypersurface u = p1 + r2 − r + (c 2 + d2) u=0 12r3 + a3(θ, ϕ) r4 + o(r −4),
Jang’s equation has a solution
f = p1 + r2 + p ln r + q,
Schoen-Yau’s method (5) Idea of the proof:
• ADM total energy E(¯g) = p.
• Scalar curvature ¯R ≥ 2 |Y |2g¯ − 2 divg¯Y .
• ¯g can be transformed conformally to a metric ˆg with ˆ
R = 0 and E(¯g) ≥ E(ˆg).
• The positive mass theorem at spatial infinity implies p = E(¯g) ≥ E(ˆg) ≥ 0.
• Bondi mass on the u = 0 slice: m0(0) = p
Schoen-Yau’s method (6) Idea of the proof:
• From the Bondi mass-loss formula,
m0(u) ≥ m0(u0) ≥ 0 ∀u ≤ u0.
• From the generalized Bondi energy-momentum-loss formula, m0(u) ≥
s
X
1≤i≤3
Structuring
1. Gravitational waves
2. Bondi’s radiating space-time
3. Positive mass theorem at null infinity (a) Schoen-Yau’s method
Witten’s method (1)
Let (X, g, h) be an asymptotically null spacelike hypersurface. • total energy: E = ˘∇ja1j − ˘∇1trg˘(a) − a11 − δ11trg˘(a), Eν(X) = 1 16π rlim→∞ I Sr E nν r dS • total linear momentum:
P = b11 − δ11trg˘(b), Pν(X) = 1 8π rlim→∞ I Sr P nν r dS Sr: sphere of radius r in R3, ν = 0, 1, 2, 3.
Witten’s method (2)
Positive mass theorem (X. Zhang, 2002):
Let (X, gij, pij) be a 3-dimensional asymptotically null initial
data set of order τ = 3. Denote
µ := 1 2 R + (p i i) 2 − pijpij, ϕj := ∇ipji − ∇jpii, σj := 2∇i pij − pji.
If the initial data set satisfies the dominant energy condition
µ ≥ maxnqXϕ2j, qX (ϕj + σj)2 o , then E0(X) − P0(X) ≥ s X i=1,2,3 Ei(X) − Pi(X)2 . If equality holds, then
Witten’s method (3)
Remark: The proof of the theorem can still go through if the order τ > 32 and the Eν − Pν are finite for ν = 0, 1, 2, 3.
Witten’s method (4)
Theorem 2. Let L3,1, g˜ be a vacuum Bondi’s radiating space-time with Bondi-metric g. Suppose that Condition A˜ and Condition B hold.
If there exists u0 such that c u=u0 = d u=u0 = 0, then m0(u) ≥ s X 1≤i≤3 m2i(u) for all u ≤ u0.
Witten’s method (5) Proof.
• Without loss of generality, assume u0 = 0.
• Choose an asymptotically null spacelike hypersurface X with u = p1 + r2 − r + (c
2 + d2)
u=0
12r3 + O(r
−4).
• E−2P = −2r33(l,2+l cot θ+¯l,3 csc θ)u=0+
4M (0,θ,ϕ) r3 +O(r−4). • Eν(X) − Pν(X) = mν(0). • E0(X) − P0(X) ≥ q P3 i=1 Ei(X) − Pi(X) 2 . • m0(u) ≥ q P3 i=1 m2i(u)
Open questions
1. What does it mean in physics that M (u0, θ, ϕ) = constant ? 2. What is the physical meaning of the condition
c(u0, θ, ϕ) = 0 = d(u0, θ, ϕ) ? Does it preclude gravitational waves ?