Академический Документы
Профессиональный Документы
Культура Документы
Schwarzschild Solution
Problem Set #5: 5.3, 5.4, 5.5 (Due Monday Dec. 2nd)
Then the Forbenius’s Theorem implies that the integral curves of these vector
fields are constraint to sub-manifolds. In the case of the vector fields V (1) ,
V (2) and V (3) the submanifolds will be two-spheres that would foliate (almost)
64
CHAPTER 5. SCHWARZSCHILD SOLUTION 65
all of the manifold into two-spheres. For example, R3 can be foliated with
concentric two-spheres centered in the origin. Then the Killing vector fields
represent rotations around x, y and z axis respectively. These rotations move
points around but the point remain at the same sphere at a fixed distance
from origin. Does this foliation cover all of the R3 ? Could you think of other
manifolds that can be foliated with two-spheres?
The foliation based on symmetries of the manifold can be used to put a
coordinate system. If a manifold is n-dimentional and it is foliated with m-
dimantional submanifolds then we can use coordinates ui (where i = 1...m)
to move around on a given submanifolds and coordinates v I (where i = 1...n)
to move between submanifolds. If the submanifolds are maximally symmetric
(e.g. two spheres) then one can show that
ds2 = gµν dxµ dxν = gIJ (v)dv I dv J + f (v)γij (u)dui duj . (5.6)
All it says is that the metric of submanifolds γij (u) is the same on different
submanifolds (since gIJ (v) and f (v) are not functions of u) and that the
cross-terms dv I dui can always be eliminated by redefining coordinates such
that ∂/∂v I is orthogonal to the submanifolds. For a spherically symmetric
four dimensional manifold we get
ds2 = gaa (a, b)da2 + gab (a, b) (dadb + dbda) + gbb (a, b)db2 + r 2 (a, b)dΩ2 (5.7)
where
dΩ2 = dθ2 + sin2 θdφ2 . (5.8)
Our remaining task is to change the coordinates so that (5.7) takes the
form of Schwarzschild metric (5.1). The first step is to change coordinates
from a, b to a, r by inverting r(a, b), i.e.
ds2 = gaa (a, r)da2 + gar (a, r) (dadr + drda) + grr (a, r)dr 2 + r 2 dΩ2 . (5.9)
or
'! "2 ! "! " ! "2 (
∂t ∂t ∂t ∂t
ds2 = m(t, r) da2 + (dtdr + drdt) + dr 2 +n(t, r)dr 2 +r 2 dΩ2 .
∂a ∂a ∂r ∂r
(5.13)
By matching the coefficients of (5.9) and (5.13) we get three equations
! "2
∂t
m = gaa
∂a
! "2
∂t
n+m = grr
∂r
! "! "
∂t ∂t
m = gar (5.14)
∂a ∂r
that can always be solved for three unknown functions t(a, r), m(a, r) and
n(a, r). Therefore we can always put a spherically symmetric metric into the
form of (5.10). But since the signature of the metric is Lorentzian either m
or n must be negative and thus without loss of generality we can assume that
m is negative. Then (5.10) can be rewritten in terms of exponentials as
ds2 = −e2α(t,r) dt2 + e2β(t,r) dr 2 + r 2 dΩ2 . (5.15)
The final step is to determine α and β by solving the vacuum Einstein
equation. If we label (t, r, θ, φ) as (0, 1, 2, 3) then non-vanishing Christoffel
symbols,
Γ000 = ∂0 α
Γ001 = Γ010 = ∂1 α
Γ011 = e2(β−α) ∂0 β
Γ111 = ∂1 β
Γ101 = Γ110 = ∂0 β
Γ100 = e2(α−β) ∂1 α
Γ122 = −re−2β
1
Γ221 = Γ212 =
r
Γ133 = −re−2β sin2 θ
1
Γ331 = Γ313 =
r
Γ233 = − sin θ cos θ
cos θ
Γ332 = Γ323 = . (5.16)
sin θ
CHAPTER 5. SCHWARZSCHILD SOLUTION 67
R0101 = e2(β−α) ∂02 β + (∂0 β)2 − ∂0 α∂0 β + ∂1 α∂1 β − ∂12 α − (∂1 α)2
# $ # $
R0202 = −re−2β ∂1 α
R0303 = −re−2β sin2 θ∂1 α
R0212 = −re−2α ∂0 β
R0313 = −re−2α sin2 θ∂0 β
R1212 = −re−2β ∂1 β
R1313 = −re−2β sin2 θ∂1 β
R2323 1 − e−2β sin2 θ∂1 β
# $
= (5.17)
The requirement that our metric solves the vacuum Einstein equation is
equivalent to
Rµν = 0 (5.19)
for all µ and ν. In particular
and
Since all of the metric component are independent of time this proves that the
spherically symmetric solution of vacuum Einstein equation must possesses
a time-like Killing vector. Such metrics are called stationary. If in addition
the time-like Killing vector is orthogonal to a family of space-like hyper-
surfaces, then the metric is call static.
To find α(r) and β(r) we note that
2
R00 = 0 , R11 = 0 ⇒ e2(α−β) R00 +R11 = 0 ⇒ (∂1 α + ∂1 β) ⇒ α = −β+const.
r
(5.25)
and by once again redefining time coordinates
But since
µ
R22 = 0 ⇒ e2α (2r∂1 α + 1) = 1 ⇒ ∂1 re2α = 1 ⇒ e2α = 1 +
# $
r
(5.27)
we arrive at our final expression
) µ* 2 ) µ *−1 2
ds2 = − 1 + dt + 1 + dr + r 2 dΩ2 . (5.28)
r r
g00 = −1 − 2Φ (5.31)
grr = 1 − 2Φ (5.32)
with potential
GM
Φ=− (5.33)
r
if we set
µ = −2GM. (5.34)
CHAPTER 5. SCHWARZSCHILD SOLUTION 69
GM
Γ100 = (r − 2GM)
r3
−GM
Γ111 =
r (r − 2GM)
−GM
Γ001 = Γ010 =
r (r − 2GM)
1
Γ212 = Γ221 =
r
Γ122 = − (r − 2GM)
1
Γ313 = Γ331 =
r
Γ133 = − (r − 2GM) sin2 θ
Γ233 = − sin θ cos θ
cos θ
Γ323 = Γ332 = (5.40)
sin θ
leading to the following geodesic equations
d2 t 2GM dr dt
2
+ =0 (5.41)
dλ r (r − 2GM) dλ dλ
"2 "2 '! "2 ! "2 (
d2 r GM
! !
dt GM dr dθ 2 dφ
+ (r − 2GM) − −(r − 2GM) + sin θ =0
dλ2 r 3 dλ r (r − 2GM) dλ dλ dλ
(5.42)
"2
d2 θ 2 dθ dθ
!
dφ
+ − sin θ cos θ =0 (5.43)
dλ2 r dλ dλ dλ
d2 φ 2 dθ dr cos θ dθ dφ
2
+ +2 = 0. (5.44)
dλ r dλ dλ sin θ dλ dλ
To solve these equation we use the symmetry of the problem manifested
by the four Killing vectors (three rotations and one time translation). The
existence of Killing vectors imply that
dxµ
Kµ = const. (5.45)
dt
is conserved along geodesics of the three components of angular momenta
(one for magnitude and two for direction) and one component for the energy.
CHAPTER 5. SCHWARZSCHILD SOLUTION 71
where
1 ϵGM L2 L2 GM
V (r) = ϵ − + 2− . (5.56)
2 r 2r r3
The first two terms in the potential are the same as in Newtonian mechan-
ics, but the last term is a new term due to the effects of general relativity.
Note that in the limit of small masses and large distances the correction is
negligible.
To find circular orbits we can differentiate the potential and set it to zero.
In Newtonian gravity
dV (r) 1 # L2
= 4 ϵGMr 2 − L2 r = 0 ⇒ r =
$
(5.57)
dr r ϵGM
and in general relativity
√
dV (r) 1 # L2 ± L4 − 12ϵG2 M 2 L2
= 4 ϵGMr 2 − L2 r + 3GML2 = 0 ⇒ r =
$
.
dr r 2ϵGM
(5.58)
In the limit of large angular momentum the solutions are given by
+ 2
L
stable orbit
r = ϵGM (5.59)
3GM unstable orbit
r = 6GM (5.60)
when √
L= 12ϵGM (5.61)
which is the lowest bound of the radii of a stable orbits in Schwarzschild
metric. Evidently, the possible range of closed stable orbits
The situation with non-circular orbits is much more subtle since one needs
to solve an equation for φ(λ) which is usually done using power series ap-
proximation. The solutions are not ellipses as in Newtonian mechanics and
would experience precession. To find the angular solutions we make use of
(5.50) or
! "−2
dφ r4
= 2. (5.64)
dλ L
to rewrite (5.55) as
! "2
dr 1 2GM 3 E2 4
+ 2 r4 − r + r 2
− 2GM r = r (5.65)
dφ L rL2 L2
where we have chosen to work with proper time by setting ϵ = 1. Then in a
new variable
L2
x= (5.66)
GMr
the equation is
! "2
dx L2 2G2 M 2 3 2E 2 L2
+ 2 − 2x + x2 − x = 2 2. (5.67)
dφ GM L2 GM
By differentiating it with respect to φ we get
d2 x 3G2 M 2 2
− 1 + x = x. (5.68)
dφ2 L2
dx
since dφ ̸= 0 for non-circular orbits. In Newtonian limit the right hand side
would be zero and the solution would be given by
x0 = 1 + e cos φ (5.69)
where ,
b2
e= 1− (5.70)
a2
is the eccentricity where a and b are the semi-major and semi-minor axes.
To study the leading correction due to effects of general relativity we expand
x ≈ x0 + x1 (5.71)
to obtain
d2 x1 3G2 M 2 3G2 M 2
!! " "
2 1 1 2
+x1 ≈ (1 + e cos φ) = 1 + e2 + 2e cos φ + e cos 2φ .
dφ2 L2 L2 2 2
(5.72)
CHAPTER 5. SCHWARZSCHILD SOLUTION 74
5.4 Black-Holes
Consider a stationary observer with respect to Schwarzschild coordinates
with four-velocity
'! "−1/2 (
2GM
U= 1− , 0, 0, 0 . (5.80)
r
Thus the photons moving away from the origin (or larger values of Newtonian
potential) are red-shifted (or loose energy), and the photons moving towards
the origin (or smaller values of Newtonian potential) are blue-shifted (or gain
energy).
Consider the behavior of light cones in Schwarzschild coordinates for con-
stant φ and θ = 0, i.e.
! " ! "−1
2 2GM 2 2GM
ds = 0 = − 1 − dt + 1 − dr 2 (5.84)
r r
Then ! "2 ! "−2
dt 2GM
= 1− (5.85)
dr r
or ! "−1
dt 2GM
=± 1− . (5.86)
dr r
In the limit r → 2GM the two solutions close up as
dt
lim = ∞. (5.87)
r→2GM dr
This means that it would take an infinite coordinate time to reach the surface
r = 2GM for an arbitrary observer, but it does not mean that the proper
time to reach the surface is infinite. The main problem is that the coordinate
time runs too fast near r = 2GM. To fix the problem we can introduce the
so called tortoise coordinates
) r *
r ∗ = r + 2GM ln −1 (5.88)
2GM
in which (5.86) has a simple solution
t = ±r ∗ + const. (5.89)
CHAPTER 5. SCHWARZSCHILD SOLUTION 76
r = const (5.96)
Alternatively one could have chosen to follow the past directed curves by
defining
u = t − r∗ (5.97)
so that the metric is
! "
2 2GM
ds = − 1 − du2 + (dudr + drdu) + r 2 dΩ2 . (5.98)
r
CHAPTER 5. SCHWARZSCHILD SOLUTION 77
In these coordinates
du
=0 (5.99)
dr
and ! "−1
du 2GM
= −2 1 − (5.100)
dr r
describe respectively the outgoing and in-falling null trajectories.
To describe both the future directed, past directed and space-like geodesics
all on the same chart we define
) r ! "
′
*1/2 r+t
v = −1 exp (5.101)
2GM 4GM
) r ! "
*1/2 r−t
u′ = − −1 exp . (5.102)
2GM 4GM
In the new coordinates the metric
16G3 M 3 ) r *
ds2 = − exp − (dv ′ du′ + du′ dv ′ ) + r 2 dΩ2 (5.103)
r 2GM
does not have any coordinates singularities at the horizon for any geodesic
passing through r = 2GM. We can now switch from two null coordinates v ′
and u′ back having only one time coordinate
! "
1 ′ ′
) r *1/2 ) r * t
T = (v + u ) = −1 exp sinh (5.104)
2 2GM 4GM 4GM
and one more spatial coordinate
! "
1 ) r *1/2 ) r * t
R = (v − u) = −1 exp cosh (5.105)
2 2GM 4GM 4GM
32G3 M 3 ) r *#
ds2 = −dT 2 + dR2 + r 2 dΩ2
$
exp − (5.106)
r 2GM
where ) r * ) r *
T 2 − R2 = 1 − exp . (5.107)
2GM 2GM
There are a number of useful properties of Kruskal coordinates:
• Null rays are at ±45% angles
T = ±R + const. (5.108)
CHAPTER 5. SCHWARZSCHILD SOLUTION 78