0% found this document useful (0 votes)
15 views5 pages

Topics Summary Sheet Please

The document discusses various concepts related to black holes, including energy conditions, Lorentzian geometry, and hypersurfaces. It outlines the Einstein equations, initial value problems, and properties of causal vectors and their implications in the context of spacetime. Additionally, it covers the definitions and properties of null hypersurfaces, Killing horizons, and the behavior of geodesics in relation to black hole mechanics.
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
15 views5 pages

Topics Summary Sheet Please

The document discusses various concepts related to black holes, including energy conditions, Lorentzian geometry, and hypersurfaces. It outlines the Einstein equations, initial value problems, and properties of causal vectors and their implications in the context of spacetime. Additionally, it covers the definitions and properties of null hypersurfaces, Killing horizons, and the behavior of geodesics in relation to black hole mechanics.
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 5

Black holes

GR & DG background • Dominant energy condition: T (X, Y ) ≥ 0 for all f.d. II fundamental form: g(∇[X,Y ] Z, N ) = −K([X, Y ], Z). Then,
product rule for 1-forms: if α ∈ Ω1 (M ) then causal vectors X, Y . Energy-momentum current: simply evalutate the LHS of the statement expanding the
dα(X, Y ) = X(α(Y )) − Y (α(X)) − α([X, Y ]) Ja := −Tab X b along integral curves of a f.d. causal definition of Riemann.
Example of leibniz: vector X. (Energy-momentum flow of matter does not
∇k K l ij = ∂k K l ij + Γlmk K mij − Γm l m l exceed c). Initial Value problem
ik K mj − Γjk K im
Einstein-equation: Rab − 1
Rgab
= 8πGTab • Weak energy condition: T (X, X) ≥ 0 for all f.d. causal Definition (Energy density /matter momentum
2
Koszul formula: 2g(∇X Y, Z) = X(g(Y, Z)) + Y (g(Z, X)) − vectors X. Only requires positivity in energy density density): If Tab is the energy-momentum tensor of matter
Z(g(X, Y )) + g([X, Y ], Z) + g([Z, X], Y ) − g([Y, Z], X). ρ := T (X, X) measured by observers with tangent X and N a unit normal to a spacelike hypersurface, then
ρ := Tab N a N b is the matter energy density and ji := −N a Tai
• Null energy condition: T (X, X) ≥ 0 for all f.d. null
Lorentzian Geometry vectors X. By Einstein equations we get Rab X a X b ≥ 0
matter momentum density for an observer with velocity N .
Prop(Hamiltonian and Einstein constraint): Let (M, g)
Lorentzian vector space for all null X.
be a spacetime satisfying the Einstein equation and S ⊂ M a
Lemma 2.1: if v, w ∈ V and g(v, w) = 0 then • Strong energy condition: Ric(X, X) ≥ 0 for all f.d. spacelike hypersurface. Then
• If v is timelike =⇒ w is spacelike
causal vectors X. By Einstein equations this equivalent R′ − K ij Kij + K 2 = 16πρ , Dj Kji − Di K = −8πji , which are
to (Tab − 21 gab T )X a X b ≥ 0 where T := g ab Tab if the Hamiltonian and momentum constraint respectively.
• If v is null =⇒ w is null collinear to v or spacelike M 4 (gravity is attractive). Definition(Vaccum intial data set):A vaccum intialn data
• If v is spacelike =⇒ w is any set is a triple (S, h, K) where (S, h) is a riemannian manifold
Hypersurfaces and K is a symmetric tensor field on S, that satisfy the
Cauchy-Schwarz and triangle: For all causal vectors v, w ∈ V Definition(Hypersurfaces): Let (M, g) be a Lorentzian Hamiltonian and momentum constraint (ρ = j = 0).
• |g(v, w)| ≥ |v||w| manifold and S ⊂ M a hypersurface. If N ∈ Tp M is timelike, Remark:A time symmetric intial data set is when K = 0 it
spacelike, null for all p ∈ S we call S a spacelike,timelike or follows that R′ = 0 (Hamiltonian).
• |v + w| ≥ |v| + |w| null hypersurface respectively(independent of choice of N ). Theorem(Choquet-Bruhat):Let (S̃, h̃, K̃) be a smooth,
Definition(Time-cone): The timecone of a timelike vector (Space/Time)like Hypersurface vector decomposition: (n − 1)-dim vacuum initial data set. Then there exist an
v ∈ V is the set C(v) := {w timelike|g(v, w) < 0}. The causal For any p ∈ S ⊂ M , we can decompose X ∈ Tp M as n-dimensional vacuum spacetime (M, g) and a smooth
cone of a timelike vector v ∈ V is defined similarly: X = X∥ + X⊥ , where X∥ ∈ Tp S and X⊥ = ±g(X, N )N . embedding ϕ : S̃ → M such that h̃ = ϕ∗ h and K̃ = ϕ∗ K,
C̄(v) := {w causal|g(v, w) < 0}. Projector onto Tp S: ha a a
b = δb ∓ N Nb ; (first sign: timelike
where h and K are induced metric and extrinsic curvature of
lemma 2.2: Timelike vectors v, w ∈ V are in the same hypersurface) the hypersurface S := ϕ(S̃) ⊂ M .
timecone iff g(v, w) < 0. Projected (induced) metric/first fundamental form:
hab = gab ∓ Na Nb / h(X, Y ) = g(X, Y ) , X, Y ∈ Tp S. Null hypersurfaces
Lorentzian manifolds If S is spacelike/timelike h is a Euclidean/Lorentzian inner A normal to a null hypersurface H is also tangent to H!
Definition (f.d/p.d): A causal vector Xp ∈ Tp M in (M, g) is: product. Theorem(null geodesics in H): Let H ⊂ M be a null
L-C of (S,h):For X, Y ∈ X(M ) then DX Y := (∇X Y )∥ . hypersurface and N ∈ X(H) be a normal vector. Then the
• Future-directed (f.d) if Xp ∈ C̄(Tp ), that is Extrinsic curvature/second fundamental form:
g(Xp , Tp ) < 0 integral curves of N are null geodesics in H/∇N N = αN (these
K(X, Y ) = −g(N, ∇X Y ), X, Y ∈ X(S) are null generators of H).
• Past-directed (p.d) if Xp ∈ C̄(−Tp ), that is The extrinsic curvature is symmetric on S and it measures the Notation:X̄ ∈ Tp H/N just means X mod N .
g(Xp , Tp ) > 0 failure of the covariant derivative along vectors tangent to S First Fundamental form: for p ∈ H and X, Y ∈ Tp H we
∇X Y = DX Y ∓ K(X, Y )N =⇒ K(X, Y ) = g(∇X N, Y ) define hp : Tp H/N × Tp H/N → R by h(X̄, Ȳ ) = g(X, Y ) this is
where T ∈ X(M ) that defines the time orientation. lemma(another way to calculate K via h): for all X, Y ∈ X(S),
Definition (James): For L-C connection: positive definite and defines a metric on T H/N .
K(X, Y ) = 12 (LN h)(X, Y ) Definition(Null shape operator):null shape operator
• ∇a eb = Γcba ec . Prop(Gauss equation): Let R denote the riemann tensor b = bN of H wrt N is, for each p ∈ H, the linear map
for (M, g) and R′ the Riemann tensor of (S, h). Then for all b : Tp H/N → Tp H/N , b(X̄) = ∇X N
• Γijk = 1 il
2
g (∂j glk + ∂k glj − ∂l gjk ) W, X, Y, Z ∈ X(M ), g(W, R(X, Y )Z) = b is self adjoint wrt to h, i.e. h(b(X̄), Ȳ ) = h(X̄, b(Ȳ )) for all
Covariant derivative along curve: h(W, R′ (X, Y )Z) ∓ K(Y, Z)K(X, W ) ± K(X, Z)K(Y, W ) (to X̄, Ȳ ∈ Tp H/N
derive consider R terms and decompose ∇Y into DY and N ) Proof: Notice g(X, N ) = 0 and evaluate 0 = Y (g(X, N )) from
• DY
dt
= (∇γ̇ Y )γ = (Ẏ k + Γkij Y i γ̇ j )∂k |γ in components we have, Rijkl = Rijkl ′ ∓ Kik Kjl ± Kjk Kil metric compatibility. Then use definition of b and
Definition: Y is called parallel along γ if ∇γ̇ Y = 0. Prop (Scalar R ): R ∓ 2Rab N N b = R′ ± Kij K ij ∓ K 2
′ a [X, Y ] ∈ X(H), so normal to N and evaluate the lhs.
Definition: γ is geodesic iff ∇γ̇ γ̇ = 0. where K = hij Kij . (to derive consider R definition then Definition(Null Second fundamental form): null second
In coordinates: decompose g ab via the projector and find wrt R then use fundamental form B = BN of H with respect to N is, for each
• γ̈ k (t) + Γkij (γ(t))γ̇ i (t)γ̇ j (t) = 0. Gauss for the RHS ). p ∈ H, the bilinear form B : Tp H/N × Tp H/N → R,
Prop(Codazzi-eq): For all X, Y, Z ∈ X(S), B(X̄, Ȳ ) = h(b(X̄), Ȳ ) = g(∇X N, Y ) (where B is symmetric).
Definition (Curvature, James): Rdcab ed = R(ea , eb )ec . g(N, R(X, Y )Z) = (DY K)(X, Z) − (DX K)(Y, Z)/N a Raijk = In an adapted basis {eA }A=1··· ,n−2 for Tp H/N ,
Definition(Ricci, James): Rab = g ca Rabcd / Dk Kij − Dj Kik bAB = hAC BCB .
Ric(X, Y ) = (R(ed , Y )X)d Proof: Take g(N, ∇X ∇Y Z) = −K(X, DY Z) − X[K(Y, Z)] Definition(covariant derivative on b): Let γ be an
ENERGY CONDITIONS: from ∇X Y = DX Y ∓ K(X, Y )N twice. Since, by definition of integral curve of N (null generator on H) given a vector along
γ (Y (s) ∈ Tγ(s) H) we define the derivative of b along γ as Definition/prop(Killing vector): K ∈ X(M ) is killing if α : (−ε, ε) × I → M , (s, t) 7→ γs (t) is smooth one-to-one with
b′ (X̄) := b(X̄)′ − b(X̄ ′ ) where Ȳ ′ := ∇γ̇ Y . LK g = 0 =⇒ ∇µ Kν + ∇ν Kµ = 0. smooth inverse.
Prop(Riccati equation): Let γ be affinely parametrised null Definition(Killing Horizon): A null hypersurface is H ⊂ M Thus, a variation α parametrises a 2-dimensional surface
generator of H. b satisfies Riccati equation along γ is a Killing horizon if there is a Killing vector of (M, g) that is Σ := im(α) ⊂ M . A vector field along α is a smooth map
b′ + b2 + R = 0 where R is the linear map normal to H. where on H, ∇K K = κK where κ ∈ C ∞ (H) is Y : (−ε, ε) × I → T M such that Y (s, t) ∈ Tγs (t) M . Define,
˙ γ(s).
˙ called the surface gravity of the killing horizon. α̇ = ∂α , α′ = ∂α , which are vector fields along α that are
R(X̄) := R(X, γ(s)) ∂t ∂s
Prop:Any Killing horizon H is totally geodesic (To prove one tangent to Σ. Under suitable definition of covariant
Proof: First note that from the de nition of the Riemann
can use the identity derivatives, we have [γ̇, γ ′ ] = γ∗ [∂t , ∂s ] = 0, so ∇α̇ α′ = ∇α′ α̇.
tensor R(N, X)N ∈ X(H) for all X ∈ X(H) and R(N, X)N
g(∇X K, Y ) = 21 (LK g)(X, Y ) + 21 (dK ♭ )(X, Y ), then compute The tangent to the original curve is γ̇ = α̇|s=0 , while a vector
depends only on X̄ so we can indeed define a linear map
R(X̄) = R(X, N )N . Then, rescale N so that ∇N N = 0. Then, dK ♭ (X, Y ) show its zero on H and that thus B is zero on H). field along γ given by V = α′ |s=0 is called a variation vector
∇N ∇X N = R(N, X)N + ∇[N,X] N . Now, consider a null Prop(Ricci on killing):H ⊂ M is a killing horizon then for field.
generator γ̇(s) = Nγ(s) .Then, evaluate b′ (X̄) and get the
all X ∈ X(H), X(κ) = −Ric(K, X) on H. A special case is if these are geodesics Ddtα̇ = 0 in which case α
An easier way to calculate surface gravity: is called a geodesic variation.
result.
Definition(null expansion): The null expansion of a d(g(K, K)) = −2κK ♭ on H. Rate of convergence of geodesics/Jacobi equation: If γ
2
hypersurface H wrt N is θ = trb = hAB BAB . A cross-section Theorem(zeroth law of black hole mechanics):The is geodesic, Ddt2J = R(γ̇, J)γ̇ is the Jacobi equation.
of H is a codimension-1-manifold Σ ⊂ H which is transverse surface gravity κ of a killing horizon H in a spacetime (M, g) Proposition (VERY IMPORTANT): If α is a geodesic
to H (Np ∈ / Tp Σ for all p ∈ Σ). θ = divΣ N , θ measures the that obeys the Einstein equation with DEC is a constant on variation, the deviation vector α′ is a Jacobi field along each
expansion of the null generators. every connected component of H. Proof: Use DEC to get that geodesic in the family (proof is very easy, simply use the lhs of
Prop(Raychaudhuri equation): Let γ be an affine V a = Tba K b |H is causal on H. On the other hand, by Einstein Jacobi equation and symmetry of connection for variation).

parametrised null geodesic generator of a null hypersurface equation and by Ric(K, K) = K(κ) = 0, Interpretation: Dα is understood as the relative ’velocity’ of
dt
1
H.Then along γ, θ̇ + trσ 2 + n−2 θ2 + Ric(γ̇, γ̇) = 0 where 0 = Ric(K, K) = 8πT (K, K) = 8πg(K, V ), so V is on H. 2 ′
Thus, V is null and V = f K. By Einstein equation then nearby geodesics, while Ddtα 2 as relative ’acceleration,’ so the
1
σ := n−2
θI is called the trace-free part of b called the shear rhs of Jacobi eqn is sometimes understood as ”tidal force.”
Ric(X, K) = 8πT (X, K) = 8πg(X, V ) = 8πf g(X, K) = 0, so
(to prove take the trace of the riccati equation above/ also X(κ) = 0, so statement. Definition: The first and second variation of arclength are
note R0̄00̄0 = g 0̄0 R000̄0 = 0 in a basis for Tp M by d d2
L′ (0) := ds L[γs ]|s=0 , L′′ (0) = ds 2 L[γs ]|s=0 , where
{ea }a=0,0̄,1,···n−2 s.t. e0̄ = L, e0 = N where L is null Curves and congruences Rbp
L[γs ] = a −g(γ˙s , γ˙s )dt is the arclength of each curve.
g(N, L) = 1 and {eA }A=1,...n−2 are a basis of span⊥ (N, L). Let V beR a variation vector field.
Local Lorentz geometry P A first variation is
Then, trR = n−2 A
L′ (0) = I g(V, ∇γ̇ (γ̇/|γ̇|))dt + ki=1 g(V, [γ̇/|γ̇|]i ), where []i is
P
A=1 g(eA , R(eA , N )N ) = R 0A0 =
a 0̄
R 0a0 + R 00̄0 = Ric(N, N ). Definition(exponential map): exponential map the jump in the unit velocity over breaks. A second variation
2
Prop(simple ver of Hawkings Area thm):Let (M, g) be a expp : Op ⊆ Tp M → M is the function expp (v) = γv (1), where is L′′ (0) = I g(V, Ddt2V − R(γ̇, V )γ̇)dt + ki=1 g(V, [∇γ̇ V ]i ),
R P
spacetime satisfying the NEC and H ⊂ M a null hypersurface. Op is the set of tangent vector such that the unique geodesic where [∇γ̇ V ]i is the jump in ∇γ̇ V at breaks. Notice, this
If the null expansion of H is θ = θ0 < 0 at a point on a f.d. γv throuh p with γ̇(0) = v is defined on at least [0, 1]. vanishes if V is Jacobi. This makes sense as a deviation vector
null generator γ, then θ → −∞ along γ within affine Op is star-shaped domain in Tp M containing origin that is if of a geodesic variation is always Jacobi, so the nearby curves
(n−2)
parameter s ≤ |θ | . In particular, if the null generators of H v ∈ Op then sv ∈ Op for all s ∈ [0, 1] (γsv (t) = γv (st)). in the variation are also geodesic and hence extremise
0 prop(normal nbhd):For every p ∈ M there exists a arclength. These are proven later in ”example questions” in
are future complete then H has θ ≥ 0 everywhere.
star-shaped neighbourhood 0 ∈ Õp in Tp M such that this piecewise setting.
Killing horizons expp : Õp → Up is a diffeomorphism onto a neighbourhood Definition: R The index
2
form is
p ∈ Up in M . Up is called a normal neighbourhood. I(V, W ) = I g(V, DdtW 2 − R(γ̇, W )γ̇)dt +
Pk
Definition(Totatlly geodesic): A null hypersurface H is i=1 g(V, [∇γ̇ W ]i ).
Prop(maximal curves in normal nbhd): Let p ∈ M and This form is symmetric and I(V, V ) = L′′ (0). The proof is by
totally geodesic if BN = 0 on H.(Note this independent of Up be a normal neighbourhood of a Lorentzian manifold. If
scalling N ). differentiating g(V, ∇γ̇ W ) and integrating along I, and
there is a timelike curve from p to q in Up , then the radial rearranging.
Prop(properties of such hypersurfaces): Let H ⊂ M be a geodesic from p to q is the unique longest timelike curve in Up
totally geodesic null hypersurface with a null normal N. from p to q (up to a reparametrisation). Geodesic congruences
• ∇X Y ∈ X(H) for all X, Y ∈ X(H), that is, L-C An open set U ⊆ M is convex if it is a normal neighbourhood Definition: A congruence of curves in an open set O ⊂ M is
connection ∇ of (M, g) restricts to an affine connection of each points. Importantly, it can be show each p ∈ M is a family of curves such that through each point in O passes
on H (geodesics in M intially tangent to H remain in conatained in a convex neighbourhood U. exactly one curve in the family. We say a congruence is
H). Proposition(in a convex nbhd timelike geodesics are smooth if the vector field U on O defined by the tangent
maximal): Let U be a convex neighbourhood in a lorentzian vectors to the curves is smooth, so U ∈ X(O). A geodesic
• There exists ω ∈ Ω1 (H) such that for all X ∈ X(H), manifold. If there is a timelike curve γ from p to q in U , then congruence is a congruence for which every curve is a geodesic.
∇X N = ω(X)N on H. the unique geodesic from p to q in U , denoted pq, is timelike Timelike congruences: Parametrise by proper time and
Proof: B(X̄, Ȳ ) = −g(∇X Y, N ). This vanishes if and L(pq) ≥ L(γ) for all causal curves γ from p to q with consider hab = gab + Ua Ub , which has the property hab U a = 0.
equality iff γ is pq (up to reparametrisation). Thus, for each p ∈ O, the linear map ha a a
∇X Y ∈ X(H). Further, by definition B(X̄, Ȳ ) = g(∇X N, Y ), b = δb + U Ub projects
which vanishes for all Y ∈ X(H) iff ∇X N is normal to H for Tp M to span(Up )⊥ . Since U is timelike, h defines a positive
all X on hypersurface. This is the case iff ∇X N = ω(X)N for Variation of arclegth definite inner product on span(Up )⊥ referring to h as spatial
all X The shape operator bN = 0, so the null expansion θ = 0 Definition. A variation of curve γ is a 1-parameter family of metric. More generally, any tensor vanishing under single
and shear σ = 0 on H which implies Ric(N, N ) = 0 on H. curves γs for |s| < ε such that γ0 = γ and the map contraction with U a or Ua we refer to as spatial tensor. If U is
normal to a hypersurface S then h is the induced metric on S, and q, because otherwise there would be a conjugate point Causal structure
however in general there is no hypersurface orthogonal to U . between p and q. Now consider the congruence of all timelike
Futures and past
ANALOGY TO NULL HYPERSURFACES: geodesics through p. As mentioned above, Jacobi fields along γ
b(X) = ∇X U and B(X, Y ) = g(X, b(Y )). B(X, U ) = b(U ) = 0, that vanish at p are equivalent to variational fields for Definition(causality): Let (M, g) be a spacetime and
since ∇U U = 0 because it is tangent to geodesics. variation of γ in the congruence. Therefore, b(J) = ∇γ̇ J, p, q ∈ M
In a local basis ba a c
which in our basis is J˙i = bij J j . Then, the previous hold, iff • if there exists a f.d. timlike curve in M from p to q we
b = ∇b U and Bab = gac bb = ∇b Ua .
Expansion, shear and twist of a congruence is given by write p << q. We define the chronological (or
1 Ȧij = bik Akj , so b = ȦA−1 . In particular, we deduce,
θ = hab Bab , σab = B(ab) − n−1 θhab and ωab = B[ab] . Thus, d d
timelike) future of p as
θ = trȦA−1 = (det A)−1 dt det A, so θ = dt log | det A|. Now,
Bab = n−1 1
θhab + σab + ωab . Notice σ and ω are purely I + (p) = {q ∈ M |p << q}
since A satisfies the linear ODE before there exists a smooth
spatial. If U is normal, then B|S coincides with K, so B is solution everywhere along γ with the stated initial conditions, for a subset S ⊂ M , we define its chronological future as
symmetric, so there is no twist. d
so in particular dt det A is smooth along γ. Hence, θ → −∞ as
[
I + (S) = I + (p)
Frobenius’ theorem: U a is hypersurface orthogonal iff t → t1 iff det A → 0+ as t → t1 which is equivalent to saying A p∈S
U[a ∇b Uc] = 0. For a timelike geodesic congruence this is is non-invertible at q.
equivalent to the vanishing of twist. • if there exists a f.d. causal curve in M from p to q we
Proposition: Let (M, g) be an SEC spacetime and γ a
Riccati for congruence: ∇U b + b2 + R = 0 for a spatial timelike geodesic through p and suppose its expansion
write p < q. We define the causal future of p as
tensor R(X) = R(X, U )U . Proof by definition starting from θ(r) = θ0 < 0 where r is a point on γ to the future of p. Then, J + (p) = {q ∈ M |p ≤ q}
the left most term and applying definition of b, curvature and within proper time t ≤ (n − 1)/|θ0 | from r along γ there exists
symmetry where p ≤ q if p = p or p < q. For a subset S ⊂ M , we
a point q conjugate to p. define its causal future as
Raychaudhuri-Landau: If X is a geodesic congruence. Proposition (IMPORTANT): The necessary and sufficient [
Then, along each geodesic in the congruence: J + (S) = J + (p)
1 condition that a timelike curve γ from p to q is a local
θ̇ = − n−1 θ2 − σ ab σab + ω ab ωab − Rab γ̇ a γ̇ b . Proof: trace the maximum of the proper time over all variations of γ is that γ
p∈S
Riccati. is a geodesic with no congruence points between p and q. • The chronological and causal past of a point
Proposition: Let U be tangent to a hypersurface orthogonal (proof undoable, Im not even writing it) p, I − (p) and J − (p) resp., and of a set S, I − (S) and
timelike geodesic congruence and suppose SEC. If the Proposition: Let (M, g) be a spacetime satisfying SEC. J − (S) resp., are defined as above except the causality
expansion θ takes a negative value θ0 < 0 at any point on a Suppose there is a spacelike hypersurface S with K = K0 < 0 relations << and ≤ are reversed.
geodesic in the congruence then θ → −∞ along that geodesic at some q ∈ S. Then, within proper time t ≤ (n − 1)/|K0 | I + (S)/J + (S) is the set of points that be reached by f.d.
within proper time t ≤ (n − 1)/|θ0 |. there exists a point p conjugate to Σ along the geodesic γ timlike/causal curves starting from S. The Causal relations
Proof: Twist is zero, cause U is normal. The shear is spatial so through q orthogonal to S. << and ≤ are transitive!
σ 2 ≥ 0. Apply SEC to get dθ dt
1
≤ − n−1 θ2 . Integrating and Proposition: Let γ be a timelike curve between p and q ∈ S Definition(Topology): A manifold M is a topological space
rearranging gives the result using p = γ(0) and θ(p) = θ0 . where S is a spacelike hypersurface. The necessary and i.e with a collection of open subsets (including the empty set ∅
Conjugate points sufficient condition that γ is a local maximum of the proper and M ) and closed subsets (the complement in M of an open
time over all variations of γ with endpoints on S and p is that set). Definition (open and closed
Conjugate points along causal geodesics are of fundamental
γ is an unbroken geodesic orthogonal to S with no point subset/interior/closure/boundary):
significance for two reasons. Firstly, they correspond to points
conjugate to S between S and p. • A subset S ⊆ M is open if for all p ∈ S there is
where geodesic congruences containing the geodesic are
Definition: Let Σ be a codimension-2 spacelike hypersurface. neighbourhood U of p such that U ⊂ M (a
singular, that is, points where all nearby geodesics meet.
A point p is called conjugate to Σ if there exists a Jacobi field neighbourhood of p is a set that contains an open set
Secondly, they characterise the failure of a geodesic to be a
J¯ along γ that is non-zero on Σ, but vanishes at p. For each that contains p).
local maximum of arclength between two points, that is, that
r ∈ Σ, (Tr Σ)⊥ is a 2-dimensional Lorentzian vector space and
there exist longer nearby curves. The interplay of these two
hence there are 2 independent future-directed null vectors. • A subset S ⊆ M is closed if it contains all of its limit
phenomena are crucial for establishing the singularity points
These are called the ”outgoing” and ”ingoing” null geodesics
theorems and global properties of black holes. • Given a subset S ⊂ M the interior of S, denoted
from Σ and form a pair of null hypersurfaces (rather than
Definition: Two points p, q on a geodesic γ are conjugate if int(S), is the set of p ∈ S such that there is
congruence) Proposition: Let (M, g) be a spacetime
there is a non-zero Jacobi field on γ vanishing at p and q. neighbourhood of p ∈ S(hence its open)
satisfying NEC and Σ be an orientable codimension-2 spacelike
Proposition: Let γ be a timelike geodesic through p and
consider the congruence of all such geodesics. Then, a point q
surface. Suppose, null expansion of the outgoing or ingoing • The closure of S, denoted by S, is the union of S with
geodesics from Σ is θ(r) = θ0 < 0 for some r ∈ Σ. Then, its limit points.
on γ to the future of p is conjugate to q iff along γ, θ → −∞
within affine parameter s ≤ (n − 2)/|θ0 | there exists a point
at q. • The (topological) boundary of S is Ṡ = S\int(S) and
conjugate to Σ along the null geodesic through r.
Proof: Choose proper time, e0 = γ̇ and {ei } orthonormal note that S = Ṡ ∪ S
Proposition: Let Σ be a codimension-2 spacelike
frame of span(e0 )⊥ . Consider Jacobi field J ∈ span(e0 )⊥ , so S is open iff S =int(S) and S is closed iff S = S. prop: If U
hypersurface and γ a causal curve from Σ to p. The necessary
in this basis Jacobi equation becomes J¨i = Ri 00j J j . Now, is a convex set in M , then for each p ∈ U
and sufficient condition for the NONexistence of variation of γ
γ(0) = p and J(0) = 0. The solution to the ODE must linearly to a timelike curve from Σ to p is that γ is a null geodesic • I + (p, U ) is open in U (and hence M ) and is the set of
depend on the initial conditions, so J i (t) = Aij (t)J˙j (0). Thus, orthogonal to Σ with no point conjugate to Σ between Σ and p. points in U that are along f.d. timelike geodesics in U
Aij (0) = 0 and Ȧij (0) = δji and Äij = Ri 00k Akj . Now, a point that start at p
q = γ(t1 ) is conjugate to p = γ(0) iff J(t1 ) = 0 for some initial • J + (p, U ) is the closure in U of I + (p, U ) and J˙+ (p, U ) is
data J˙i (0) ̸= 0. By earlier, this only occurs if Aij (t1 ) is the set of points in U that are along f.d. null geodesics
non-invertible. But A(t) is invertible all along γ, between p in U that start at p.
Prop(I ± (p) is open): Let p << q in M . Then there exist lemma:An achronal boundary I˙+ (S) is achronal and edgeless. there exists a unique (up to diffeomorphism) n-dimensional
open neighbourhoods Ṽ of p and Ũ of q such that p′ << q ′ for prop:If S is achronal, S\edge(S) is C 0 −hypersurface in M . spacetime (M, g), called the maximal Cauchy development of
all p′ ∈ Ṽ , q ′ ∈ Ũ in general though J + (p/S) is not closed Prop:An achronal boundary I˙+ (S) (non empty) is a closed, (S̃, h̃, K̃), such that (i) (M, g) satisfies the vacuum Einstein
consider Minkowski minus a point. achronal, C 0 -hypersurface in M . prop:Let S ⊂ M be closed. equations; (ii) there is a smooth embedding φ : S̃ → M such
prop(null geodesics/causul curves):If q ∈ J + (p)\I + (p) Then every p ∈ J˙+ (S)\S is on a null geodesic contained in that φ∗ h = h̃ and φ∗ K = K̃, where h and K are the induced
then any causal curve from p to q is a null geodesic. In general J˙+ (S) that either is past inextendible in M or has a past metric and extrinsic curvature of the hypersurface
I + (S) = I + (I + (S)) = I + (J + (S)) = J + (I + (S)) also endpoint S. S := φ(S̃) ⊂ M ; (iii) (M, g) is globally hyperbolic with Cauchy
generically I + (S) ̸= J + (S) surface S; and (iv) any other spacetime satisfying (i)–(iii) is
prop(set relations): For all subsets S ⊂ M :
Global hyperbolicity isometric to a subset of (M, g).
definition(strongly causal)A spacetime is strongly causal at
• int(J + (S))= I + (S)
p ∈ M if for any neighbourhood V of p there exists a Singularity theorems
• J + (S) ⊂ I + (S) neighbourhood U ⊂ V of p such that no causal curve in M
meets U at more than one point. A spacetime is strongly Thm Hawking:Let (M, g) be a globally hyperbolic spacetime
• J˙+ (S) = I˙+ (S)
causal if it is for all p. satisfying the Strong Energy Condition (SEC). Suppose there
To prove: Definition(globally hyperbolic): A spacetime (M, g) is is a smooth spacelike Cauchy surface S on which the trace of
• By definition, I + (S) ⊂ J + (S), and since I + (S) is open, globally hyperbolic if : the extrinsic curvature satisfies tr K ≤ K0 < 0 for some
it follows that I + (S) ⊂ int(J + (S)). This follows because constant K0 . Then any future-directed timelike curve γ
• It is strongly causal n−1
for any p ∈ I + (S) ⊂ J + (S), there is a neighbourhood V emanating from S has length L[γ] ≤ |K , and consequently
0|
of p such that V ⊂ I + (S) ⊂ J + (S), and by definition, • J + (p) ∩ J − (q) are compact for all p, q ∈ M (internal (M, g) is future timelike geodesically incomplete.
int(J + (S)) is the set of points p ∈ J + (S) such that compactness) Defintion(trapped-surfaces):A trapped surface is a
there is a neighbourhood V of p where V ⊂ J + (S). prop(properties of g.h. spacetime):Let M be a globally compact, codimension-2, spacelike, orientable, submanifold
Now for the converse. If p ∈ int(J + (S)), there is a hyperbolic spacetime.Then Σ ⊂ M for which the null expansion of both families of null
convex neighbourhood U of p such that geodesics orthogonal to Σ are strictly negative on Σ (recall the
• J ± (p) is closed for all p ∈ M
I − (p, U ) ∩ J + (S) ̸= ∅. Hence, there is a q ∈ J + (S) such outgoing and ingoing null geodesics from Σ form a pair of null
that q ∈ I − (p, U ), which implies q ≪ p, and therefore • J ± (A) is closed for all compact A ⊂ M hypersurfaces that meet at Σ)
p ∈ I + (J + (S)) = I + (S). Thm(Penrose):Let (M, g) be a spacetime that satisfies the
• J + (A) ∩ J − (B) is compact for all compact A, B ⊂ M
Hence, we have shown the reverse inclusion: Null Energy Condition (NEC) with a noncompact Cauchy
int(J + (S)) ⊂ I + (S). prop:In a g.h. spacetime (M, g), if K ⊂ M is a compact surface S. If M contains a trapped surface Σ, then there is at
subset, then every p ∈ J˙+ (K) is on a past-directed null least one future-inextendible null geodesic orthogonal to Σ
• ii) We will prove J + (p) I + (p).
⊂ Recall, by definition, geodesic from p to K that lies in J˙+ (K). with affine length no greater than n−2 , where θ0 < 0 is the
p ∈ J + (p), and it is easy to see that p ∈ I + (p) (just propLet (M, g) be a globally hyperbolic spacetime and Σ a
|θ0 |
consider a sequence of points on a timelike curve maximum value of the null expansion for both families of null
compact orientable,co-dimension 2 submanifold of M .Then geodesics orthogonal to Σ. In particular, (M, g) is future null
through p that converge to p). Now let q ∈ J + (p) where every p ∈ J˙+ (Σ) is on a f.d. null geodesic orthagonal to Σ that
p ≤ q, and let γ be a future-directed causal curve from p geodesically incomplete.
lies in J˙+ (Σ) has no points conjugate to Σ between Σ and p.
to q. Let U be a convex neighbourhood of q, and let prop:Let M be a globally hyperbolic spacetime. For p, q ∈ M
q − ∈ γ ⊂ J − (q, U ). Hence, with p << q there is a maximal future directed timelike Asymptotic flatness
q ∈ J + (q − , U ) = I + (q − , U ) (null/causal). But geodesic from p to q.
q − ∈ J + (p), and hence Definition(Cauchy surface): A Cauchy surface for a Asymptotically flat initial data
I + (q − , U ) ⊂ I + (J + (p)) = I + (p), which implies that spacetime M is an achronal set S ⊂ M which every
q ∈ I + (p), as required. Definition(assymptotically flat end):(n − 1)-dimensional
inextendible causal curve in M . initial data set (S, h, K) is an asymptotically flat end if:
• By definition, I + (S) ⊂ J + (S) and hence theorem(Geroch):A spacetime M is globally hyperbolic iff it
I + (S) ⊂ J + (S) = J + (S), since all limit points of has a cauchy surface. • S is diffeomorphic to Rn−1 \B where B is a closed ball
I + (S) are also limit points of J + (S) and J + (S) is Prop(homeos to CS):If M is a Cauchy surface S, then i)M in Rn−1 centred at the origin.
closed. Similarly, part (b) implies is homeomorphic to R × S; ii) any Cauchy surface is
J + (S) ⊂ I + (S) = I + (S), and therefore homeomorphic to S • If (xi ) denote the pull-back to S of a Cartesian
J + (S) = I + (S). Hence Definition Let S be an chronal set in a spacetime M . the coordinate system on Rn−1 , then as r → ∞, the
J˙+ (S) = J + (S) \ int J + (S) = I + (S) \ I + (S) = I˙+ (S), future domain of dependence of S is D+ (S) = {p ∈ induced metric and extrinsic curvature

where the second equality follows from J + (S) = I + (S) M |every past inextendible causal curve from p meets S} The hij = δij + O(r−τ ) , Kij = O(r−τ −1 )

and part (a), and the final equality since I + (S) is open. past domain dependence is defined by replacing past where r = xi xi for some fall off rate τ > 0 and their
inextendible with future inextendible and derivatives (∂k )p hij = O(r−τ −p ) and
achronal boundries D(S) = D+ (S) ∪ D− (S). (Any signal must travel along a (∂k )p Kij = O(r−τ −p−1 ) for 0 ≤ p ≤ 2.
definition(achronal A subset S ⊂ M is achronal if no two causal curve, any point D+ (S) only receives signals from S)
points in S can be joined by a timelike curve. prop(fact of cauchy surfaces) Let S be an achronal subset (S, h, K) is asymptotically flat with N ends if there is a
defintion(edge):Let S ⊂ M be achronal. Then p ∈ S is an of a spacetime M . Then S is a cauchy surface for M iff compact set K such that S\K is diffeomorphic to the union of
edge point of S if every open neighbourhood U of p contains a M = D(S). N asymptomatically flat ends (the idea being you change
timelike curve γ from I − (p, U ) to I + (p, U ) that does not meet Thm(Choquet-Bruhat & Geroch): Let (S̃, h̃, K̃) be a coordinates such that if you were to tend to r → p you change
S. The set of edge points of S edge(S). smooth (n − 1)-dimensional vacuum initial data set. Then coods such that r∗ → ∞).
Rb
Asymptotically flat spacetimes ∂s [g(α′ , ∇α̇ (α̇/|α̇|))]dt + ki=1 ∂s [g(α′ , [α̇/|α̇|]i )] for the first
P
null infinity, there are coordinates (u, Ω, θ, ϕ) in a a
Definition(conformal boundary): A spacetime (M, g) neighbourhood of I + ⊂ M such that term a ∂s [g(α′ , ∇α̇ (α̇/|α̇|))]dt = ab g(α′ , ∇α′ ∇α̇ (α̇/|α̇|))dt =
Rb R
Rb
admits a conformal boundary (or conformal infinity) if there ḡ|Ω=0 = 2udΩ + dθ2 + sin2 θdϕ2 g(α′ , ∇α̇ ∇α′ (α̇/|α̇|) + R(α′ , α̇)α̇/|α̇|)dt|s=0 =
exists a spacetime (M̄ , ḡ) such that: Rab
where ∂u is tangent to affine null generators of I + . a g(V, ∇γ̇ ∇γ̇ V + R(V, γ̇)γ̇)dt for the second term
• M ⊂ M̄ can be extended to a manifold with a boundary ∂s [g(α′ , [α̇/|α̇|]i )] = g(∇α′ α′ , [α̇/|α̇|]i ) + g(α′ , ∇α′ [α̇/|α̇|]i )
∂M within M̄ , that is, M ∪ ∂M ⊂ M̄ . evaluating at s = 0 ,
Black holes in GR ∂s [g(α′ , [α̇/|α̇|]i )]|s=0 = g(V, ∇α′ [α̇/|α̇|]i |s=0 ) where the first
• There exists a smooth function Ω on M̄ such that Definition(black hole): Let (M, g) be a spacetime that is term has vanished since γ has no breaks but also
i) Ω > 0 and ḡ = Ω2 g on M (ḡ is conformal to g), asymptotically flat at null infinity. The blackhole is the region ∇α′ (α̇/|α̇|)|s=0 = ∇α̇ V which implies ∇α′ [α̇/|α̇|]i |s=0 =
ii) Ω = 0 on ∂M , and iii) dΩ ̸= 0 on ∂M . B := M \(M ∩ J − (I + )) and the future event horizon is its (∇α′ [α̇/|α̇|])(t,s)=(t+ ,0) − (∇α′ [α̇/|α̇|])(t,s)=(t− ,0) = [∇γ̇ V ]i .
The function Ω is called a defining function and the level set boundary H+ := Ḃ = M ∩ J˙− (I + ). The whitehole i i

Ω = 0 defines a smooth hypersurface in M̄ . W := M \(M ∩ J + (I − )) and the past event horizon is


Example: consider minkowski space in terms of retarded time H− := Ẇ = M ∩ J˙+ (I − ).
Prop(Penrose 1968):The event horizon H+ is a null
(u = t − r) this reads as g = −du2 + 2dudr + r2 dΩ2 . Now let
hypersurface in (M, g) and its null generators have no future
x = r−1 > 0 to obtain g = −du2 + 2x−1 dudx + x−2 dΩ2 . We
endpoints.
can choose Ω = x so ḡ = −x2 du2 + 2dudx + dΩ2 . This extends
Definition(strongly asymptotically predictable): A
the metric smoothly to x ≤ 0 and x = 0 is conformal boundary
spacetime (M, g) that is asymptotically flat at null infinity is
(r → ∞) this corresponds to null hypersurface I + = R × S 2
strongly asymptotically predictable if there exist an open set
where u ∈ R and ∂u is tangent to affine null generators. Prop
V̄ ⊂ M̄ with M ∩ J − (I + ) ⊂ V̄ such that (V̄ , ḡ) is globally
(formulas for conformally related spacetimes): Rab of
hyperbolic.
(M, g) is related to the R¯ab of (M, ḡ) by
Thm(Black holes cannot bifurcate): Let (M, g) be a
Rab = R̄ab + (n − 2)Ω−1 ∇¯a ∇¯b Ω + g¯ cd
ab ḡ (Ω
−1 ∇¯c ∇¯d Ω − (n −
strongly asymptotically predictable spacetime and S1 , S2
1)Ω−2 (dΩ)c (dΩ)d )
¯ is L-C of ḡ. Cauchy surfaces of V̄ with S2 ⊂ I + (S1 ). If B1 is a blackhole
where ∇
at time S1 , then J + (B1 ) ∩ S2 is nonempty and contained in a
The ricci
 scalar is given by
4(n−1) n−2 ¯ n−2
 connected component of B ∩ S2 .
R = Ω2 R̄ − Ω 2 ∆(Ω− 2 ) where ∆ ¯ := g¯ab ∇
¯ a ∇¯b
n−2
is the Laplacian on (M̄ , ḡ). Example question and answers/help for
L-C of ∇¯ :
¯ X Y = ∇X Y + X(log Ω)Y + Y (log Ω)X − g(X, Y )(∇ log Ω)
calculations

df ♯ surface normal Let f ∈ C ∞ (M ) be a smooth function
for all X, Y ∈ X(M ) where (∇f )a := g ab (df )b (to derive
and consider the hyper-surface defined as the level set
consider the Koszul formula) Prop(defining
S = {p ∈ M |f (p) = c} one can show that N a = g ab (df )b is
function):Suppose (M, g) admits a conformal boundary and
normal to S (if proof needed consider X(f ) = 0).
Ric(g) = 0 on V ∩ M where V is a neighbourhood of ∂M in
surface gravity computation For Killing horizons an easy
M̄ . Then:
way to compute surface gravity is d(g(K, K)) = −2κK ♭ on H,
• ∂M is a null hypersurface in (M̄ , ḡ). To prove contract killings equation with K a to get
• One can choose a defining function Ω such that K a ∇a Kb + K a ∇b Ka = κKb + 12 ∇b (K a Ka ) = 0 on H.
∇¯a ∇¯b Ω = 0 on ∂M . first and second variation of arc length: let
γ : [a, b] → M be a piecewise smooth curve with breaks
Definition(assymptotically flat at null infty): A a < t1 < · · · < tk < b and α(s, t) = γs (t) a piecewise smooth
4-dimensional spacetime (M, g) is asymptotically flat at null variation of γ with same breaks.
Rb p
infinity if: d
α̇, α̇)dt = − ab 2|1α̇| ∂s [g(α̇, α̇)]dt =
R
ds
L[γ s ] = a ∂ s −g(
• It admits a conformal boundary − ab g( |α̇ , ∇α′ α̇)dt = − ab g(α̇/|α̇|, ∇α̇ α′ )dt =
R R
α̇|
• Ric(g) = 0 on V ∩ M where V is an neighbourhood of
Rb
−( a ∂t [g(α̇/|α̇|, α′ )] + g(∇α̇ (α̇/|α̇|), α′ )) for first term split
∂M in M̄ . the integral into the piecewise smooth segments where t0 = a
t−
• ∂M = I + ∪ I − (disjoint union) where each I ± and tk+1 = b, so ab ∂t [g(α̇/|α̇|, α)] = k+1 ′ i+1
R P
i=0 [g(α̇/|α̇|, α )] + =
diffeomorphic to R × S 2 and are called future (past) ti

null infinity. Pk t−
[g(α̇/|α̇|, α′ )]t− + ′ i+1 ′
i=1 [g(α̇/|α̇|, α )]t+ − [g(α̇/|α̇|, α )]t+ =
1 k
• J − (M ) ∩ I + = ∅. Pk i
− i g(α′ , [α̇/|α̇|]
i) where α′ vanishes at the end points.
• The affinely parameterised null generators of I ± are d
= ab g(∇α̇ (α̇/|α̇|), α′ )dt + ki=1 g(α′ , [α̇/|α̇|]i )
R P
ds
L[γs ]
complete. evaluating at s = 0 and recalling γ̇ = α̇|s=0 and V = α′ |s=0 .
Proposition (near I ± spacetime approaches 2nd variation: differentiate the 1st variation to get
d2
Minkowski): For a spacetime that is asymptotically flat at ds2
L[γs ]|s=0 =

You might also like