This action might not be possible to undo. Are you sure you want to continue?

SPIN-09/11

INTRODUCTION TO THE

THEORY OF BLACK HOLES

∗

Gerard ’t Hooft

Institute for Theoretical Physics

Utrecht University

and

Spinoza Institute

Postbox 80.195

3508 TD Utrecht, the Netherlands

e-mail: g.thooft@uu.nl

internet: http://www.phys.uu.nl/~thooft/

Version June 9, 2009

∗

Lectures presented at Utrecht University, 2009.

1

Contents

1 Introduction 2

2 The Metric of Space and Time 4

3 Curved coordinates 5

4 A short introduction to General Relativity 6

5 Gravity 9

6 The Schwarzschild Solution 10

7 The Chandrasekhar Limit 13

8 Gravitational Collapse 14

9 The Reissner-Nordstr¨om Solution 18

10 Horizons 20

11 The Kerr and Kerr-Newman Solution 22

12 Penrose diagrams 23

13 Trapped Surfaces 25

14 The four laws of black hole dynamics 29

15 Rindler space-time 31

16 Euclidean gravity 32

17 The Unruh eﬀect 34

18 Hawking radiation 38

19 The implication of black holes for a quantum theory of gravity 40

20 The Aechelburg-Sexl metric 44

1

21 History 47

1. Introduction

According to Newton’s theory of gravity, the escape velocity v from a distance r from

the center of gravity of a heavy object with mass m, is described by

1

2

v

2

=

Gm

r

. (1.1)

What happens if a body with a large mass m is compressed so much that the escape

velocity from its surface would exceed that of light, or, v > c ? Are there bodies with a

mass m and radius R such that

2Gm

Rc

2

≥ 1 ? (1.2)

This question was asked as early as 1783 by John Mitchell. The situation was investigated

further by Pierre Simon de Laplace in 1796. Do rays of light fall back towards the surface

of such an object? One would expect that even light cannot escape to inﬁnity. Later, it

was suspected that, due to the wave nature of light, it might be able to escape anyway.

Now, we know that such simple considerations are misleading. To understand what

happens with such extremely heavy objects, one has to consider Einstein’s theory of

relativity, both Special Relativity and General Relativity, the theory that describes the

gravitational ﬁeld when velocities are generated comparable to that of light.

Soon after Albert Einstein formulated this beautiful theory, it was realized that his

equations have solutions in closed form. One naturally ﬁrst tries to ﬁnd solutions with

maximal symmetry, being the radially symmetric case. Much later, also more general

solutions, having less symmetry, were discovered. These solutions, however, showed some

features that, at ﬁrst, were diﬃcult to comprehend. There appeared to be singularities

that could not possibly be accepted as physical realities, until it was realized that at least

some of these singularities were due only to appearances. Upon closer examination, it

was discovered what their true physical nature is. It turned out that, at least in principle,

a space traveller could go all the way in such a “thing” but never return. Indeed, also

light would not emerge out of the central region of these solutions. It was John Archibald

Wheeler who dubbed these strange objects “black holes”.

Einstein was not pleased. Like many at ﬁrst, he believed that these peculiar features

were due to bad, or at least incomplete, physical understanding. Surely, he thought,

those crazy black holes would go away. Today, however, his equations are much better

understood. We not only accept the existence of black holes, we also understand how

they can actually form under various circumstances. Theory allows us to calculate the

2

behavior of material particles, ﬁelds or other substances near or inside a black hole. What

is more, astronomers have now identiﬁed numerous objects in the heavens that completely

match the detailed descriptions theoreticians have derived. These objects cannot be

interpreted as anything else but black holes. The “astronomical black holes” exhibit no

clash whatsoever with other physical laws. Indeed, they have become rich sources of

knowledge about physical phenomena under extreme conditions. General Relativity itself

can also now be examined up to great accuracies.

Astronomers found that black holes can only form from normal stellar objects if these

represent a minimal amount of mass, being several times the mass of the Sun. For low

mass black holes, no credible formation process is known, and indeed no indications have

been found that black holes much lighter than this “Chandrasekhar limit” exist anywhere

in the Universe.

Does this mean that much lighter black holes cannot exist? It is here that one could

wonder about all those fundamental assumptions that underly the theory of quantum

mechanics, which is the basic framework on which all atomic and sub-atomic processes

known appear to be based. Quantum mechanics relies on the assumption that every

physically allowed conﬁguration must be included as taking part in a quantum process.

Failure to take these into account would necessarily lead to inconsistent results. Mini

black holes are certainly physically allowed, even if we do not know how they can be

formed in practice. They can be formed in principle. Therefore, theoretical physicists

have sought for ways to describe these, and in particular they attempted to include them

in the general picture of the quantum mechanical interactions that occur in the sub-atomic

world.

This turned out not to be easy at all. A remarkable piece of insight was obtained

by Stephen Hawking, who did an elementary mental exercise: how should one describe

relativistic quantized ﬁelds in the vicinity of a black hole? His conclusion was astonishing.

He found that the distinction between particles and antiparticles goes awry. Diﬀerent ob-

servers will observe particles in diﬀerent ways. The only way one could reconcile this with

common sense was to accept the conclusion that black holes actually do emit particles, as

soon as their Compton wavelengths approach the dimensions of the black hole itself. This

so-called “Hawking radiation” would be a property that all black holes have in common,

though for the astronomical black holes it would be far too weak to be observed directly.

The radiation is purely thermal. The Hawking temperature of a black hole is such that

the Wien wave length corresponds to the radius of the black hole itself.

We assume basic knowledge of Special Relativity, assuming c = 1 for our unit system

nearly everywhere, and in particular in the last parts of these notes also Quantum Mechan-

ics and a basic understanding at an elementary level of Relativistic Quantum Field Theory

are assumed. It was my intention not to assume that students have detailed knowledge of

General Relativity, and most of these lectures should be understandable without knowing

too much General Relativity. However, when it comes to discussing curved coordinates,

Section 3, I do need all basic ingredients of that theory, so it is strongly advised to famil-

iarize oneself with its basic concepts. The student is advised to consult my lecture notes

“Introduction to General Relativity”, http://www.phys.uu.nl/ thooft/lectures/genrel.pdf

3

whenever something appears to become incomprehensible. Of course, there are numerous

other texts on General Relativity; note that there are all sorts of variations in notation

used.

2. The Metric of Space and Time

Points in three-dimensional space are denoted by a triplet of coordinates, x = (x, y, z) ,

which we write as (x

1

, x

2

, x

3

) , and the time at which an event takes place is indicated

by a fourth coordinate t = x

0

/c , where c is the speed of light. The theory of Special

Relativity is based on the assumption that all laws of Nature are invariant under a special

set of transformations of space and time:

¸

¸

¸

t

x

y

z

=

¸

¸

¸

a

0

0

a

0

1

a

0

2

a

0

3

a

1

0

a

1

1

a

1

2

a

1

3

a

2

0

a

2

1

a

2

2

a

2

3

a

3

0

a

3

1

a

3

2

a

3

3

¸

¸

¸

t

x

y

z

,

or x

µ

=

¸

ν=0,···,3

a

µ

ν

x

ν

, or x

= Ax , (2.1)

provided that the matrix A is such that a special quantity remains invariant:

−c

2

t

2

+ x

2

+ y

2

+ z

2

= −c

2

t

2

+ x

2

+y

2

+z

2

; (2.2)

which we also write as:

¸

µ,ν=0,···,3

g

µν

x

µ

x

ν

is invariant, g

µν

=

¸

¸

¸

−1 0 0 0

0 1 0 0

0 0 1 0

0 0 0 1

. (2.3)

A matrix A with this property is called a Lorentz transformation. The invariance is

Lorentz invariance. Usually, we also demand that

a

0

0

> 0 , det(A) = +1 , (2.4)

in which case we speak of special Lorentz transformations. The special Lorentz transfor-

mations form a group called SO(3, 1) .

In what follows, summation convention will be used: in every term of an equation

where an index such as the index ν in Eqs. (2.1) and (2.3) occurs exactly once as a

superscript and once as a subscript, this index will be summed over the values 0, , 3 ,

so that the summation sign,

¸

, does not have to be indicated explicitly anymore: x

µ

=

a

µ

ν

x

ν

and s

2

= g

µν

x

µ

x

ν

. In the latter expression, summation convention has been

implied twice.

More general linear transformations will turn out to be useful as well, but then (2.2)

will not be invariant. In that case, we simply have to replace g

µν

by an other quantity,

as follows:

g

µν

= (A

−1

)

α

µ

(A

−1

)

β

ν

g

αβ

, (2.5)

4

so that the expression

s

2

= g

µν

x

µ

x

ν

= g

µν

x

µ

x

ν

(2.6)

remains obviously valid. Thus, Nature is invariant under general linear transformations

provided that we use the transformation rule (2.5) for the tensor g

µν

. This tensor will

then be more general than (2.3). It is called the metric tensor. The quantity s deﬁned

by Eq. (2.6) is assumed to be positive (when the vector is spacelike), i times a positive

number (when the vector is timelike), or zero (when x

µ

is lightlike). It is then called the

invariant length of a Lorentz vector x

µ

.

In the general coordinate frame, one has to distinguish co-vectors x

µ

from contra-

vectors x

µ

. they are related by

x

µ

= g

µν

x

ν

; x

µ

= g

µν

x

ν

, (2.7)

where g

µν

is the inverse of the metric tensor matrix g

µν

. Usually, they are denoted by

the same symbol; in a vector or tensor, replacing a subscript index by a superscript index

means that, tacitly, it is multiplied by the metric tensor or its inverse, as in Eqs. (2.7).

3. Curved coordinates

The coordinates used in the previous section are such that they can be used directly to

measure, or deﬁne, distances and time spans. We will call them Cartesian coordinates.

Now consider just any coordinate frame, that is, the original coordinates (t, x, y, z) are

completely arbitrary, in general mutually independent, diﬀerentiable functions of four

quantities u = ¦u

µ

, µ = 0, , 3¦ . Being diﬀerentiable here means that every point is

surrounded by a small region where these functions are to a good approximation linear.

There, the formalism described in the previous section applies. More precisely, at a

given point x in space and time, consider points x + dx , separated from x by only an

inﬁnitesimal distance dx . Then we deﬁne ds by

ds

2

= g

µν

dx

µ

dx

ν

= g

µν

(u) du

µ

du

ν

. (3.1)

The prime was written to remind us that g

µν

in the u coordinates is a diﬀerent function

than in the x coordinates, but in later sections this will be obvious and we omit the

prime. Under a coordinate transformation, g

µν

transforms as Eq. (2.5), but now these

coeﬃcients are also coodinate dependent:

g

µν

(u) =

∂x

α

∂u

µ

∂x

β

∂u

ν

g

αβ

(x) . (3.2)

In the original, Cartesian coordinates, a particle on which no force acts, will go along a

straight line, which we can describe as

dx

µ

(τ)

dτ

= v

µ

= constant; v

µ

v

µ

= −1 ;

d

2

x

µ

(τ)

dτ

2

= 0 , (3.3)

5

y

x

x ′

y ′

Figure 1: A transition from one coordinate frame ¦x, y¦ to an other, curved

coordinate frame ¦x

, y

¦ .

where τ is the eigen time of the particle. In terms of curved coordinates u

µ

(x) , this no

longer holds. Suppose that x

µ

are arbitrary diﬀerentiable functions of coordinates u

λ

.

Then

dx

µ

dτ

=

∂x

µ

∂u

λ

du

λ

dt

;

d

2

x

µ

dτ

2

=

∂

2

x

µ

∂u

λ

∂u

κ

du

κ

dτ

du

λ

dτ

+

∂x

µ

∂u

λ

d

2

u

λ

dτ

2

. (3.4)

Therefore, eq. (3.3) is then replaced by an equation of the form

g

µν

(u)

du

µ

dτ

du

ν

dτ

= −1 ;

d

2

u

µ

(τ)

dτ

2

+ Γ

µ

κλ

(u)

du

κ

dτ

du

λ

dτ

= 0 , (3.5)

where the function Γ

µ

κλ

(u) is given by

Γ

µ

κλ

(u) =

∂u

µ

∂x

α

∂

2

x

α

∂u

κ

∂u

λ

. (3.6)

Here, it was used that partial derivatives are invertible:

∂u

µ

∂x

α

∂x

α

∂u

κ

= δ

α

κ

. (3.7)

Γ

µ

κλ

is called the connection ﬁeld. Note that it is symmetric under interchange of its two

subscript indices:

Γ

µ

κλ

= Γ

µ

λκ

. (3.8)

4. A short introduction to General Relativity

• A scalar function φ(x) of some arbitrary curved set of coordinates x

µ

, ia a function

that keeps the same values upon any coordinate transformation. Thus, a coordinate

transformation x

µ

→ u

λ

implies that φ(x) = φ

(u(x)) , where φ

(u) is the same

scalar function, but written in terms of the new coordinates u

λ

. Usually, we will

omit the prime.

6

• A co-vector is any vectorial function A

α

(x) of the curved coordinates x

µ

that,

upon a curved coordinate transformation, transforms just as the gradient of a scalar

function φ(x) . Thus, upon a coordinate transformation, this vectorial function

transforms as

A

α

(x) =

∂u

λ

∂x

α

A

λ

(u) . (4.1)

.

• A contra-vector B

µ

(x) transforms with the inverse of that matrix, or

B

µ

(x) =

∂x

µ

∂u

λ

B

λ

(u) . (4.2)

This ensures that the product A

α

(x)B

α

(x) transforms as a scalar:

A

α

(x)B

α

(x) =

∂u

λ

∂x

α

∂x

α

∂u

κ

A

λ

(u)B

κ

(u) = A

λ

(u)B

λ

(u) , (4.3)

where Eq. (3.7) was used.

• A tensor A

β

1

β

2

···

α

1

α

2

···

(x) is a function that transforms just as the product of covectors

A

1

α

1

, A

2

α

2

, , and covectors B

β

1

1

, B

β

2

2

, . Superscript indices always refer to the

contravector transformation rule and subscript indices to the covector transforma-

tion rule.

The gradient of a vector or tensor, in general, does not transform as a vector or tensor.

To obtain a quantity that does transform as a true tensor, one must replace the gradient

∂/∂x

µ

by the so-called the covariant derivative D

µ

, which for covectors is deﬁned as

D

µ

A

λ

(x) =

∂A

λ

(x)

∂x

µ

−Γ

ν

µλ

(x)A

ν

(x) , (4.4)

for contravectors:

D

µ

B

κ

(x) =

∂B

κ

(x)

∂x

µ

+ Γ

κ

µν

(x)B

ν

(x) , (4.5)

and for tensors:

D

µ

A

β

1

β

2

···

α

1

α

2

···

(x) =

∂

∂x

µ

A

β

1

β

2

···

α

1

α

2

···

(x) − Γ

ν

µα

1

(x)A

β

1

β

2

···

να

2

···

(x) −Γ

ν

µα

2

(x)A

β

1

β

2

···

α

1

ν···

(x) −

+ Γ

β

1

µν

(x)A

νβ

2

···

α

1

α

2

···

(x) + . (4.6)

In these expressions, Γ

ν

µλ

is the connection ﬁeld that we introduced in Eq. (3.6); there,

however, we assumed a ﬂat coordinate frame to exist. Now, this might not be so. In that

case, we use the metric tensor g

µν

(x) to deﬁne Γ

ν

µλ

. It goes as follows. If we had a ﬂat

7

coordinate frame, the metric tensor g

µν

would be constant, so that its gradient vanishes.

Suppose that we demand the covariant derivative of g

µν

to vanish as well. We have

D

µ

g

αβ

=

∂

∂x

µ

g

αβ

−Γ

ν

µα

g

νβ

−Γ

ν

µβ

g

αν

. (4.7)

Lowering indices using the metric tensor, this can be written as

D

µ

g

αβ

=

∂

∂x

µ

g

αβ

−Γ

βµα

−Γ

αµβ

. (4.8)

Taking his covariant derivative to vanish, and using the fact that Γ is symmetric in its

last two indices, we derive

Γ

µ

κλ

=

1

2

g

µα

(∂

κ

g

αλ

+ ∂

λ

g

ακ

−∂

α

g

κλ

) , (4.9)

where g

µα

is the inverse of g

µν

, that is, g

νµ

g

µα

= δ

α

ν

, and ∂

κ

stands short for the partial

derivative: ∂

κ

= ∂/∂u

κ

.

Eq. (4.9) will now be used as a deﬁnition of the connection ﬁeld Γ. Note that it is

always symmetric in its two subscript indices:

Γ

µ

κλ

= Γ

ν

λκ

. (4.10)

This deﬁnition implies that D

µ

g

αβ

= 0 automatically, as an easy calculation shows, and

that the covariant derivatives of all vectors and tensors again transform as vectors and

tensors.

It is important to note that the connection ﬁeld Γ

α

αβ

itself does not transform as a

tensor; indeed, it is designed to ﬁx quantities that aren’t tensors back into forms that are.

However, there does exist a quantity that is constructed out of the connection ﬁeld that

does transform as a tensor. This is the so-called Riemann curvature. This object will be

used to describe to what extent space-time deviates from being ﬂat. It is a tensor with

four indices, deﬁned as follows:

R

µ

καβ

= ∂

α

Γ

µ

κβ

−∂

β

Γ

µ

κα

+ Γ

µ

ασ

Γ

σ

κβ

−Γ

µ

βσ

Γ

σ

κα

; (4.11)

in the last two terms, the index σ is summed over, as dictated by the summation con-

vention. In the lecture course on general Relativity, the following statement is derived:

If V is a simply connected region in space-time, then the Riemann curvature

R

µ

καβ

= 0 everywhere in V , if and only if a ﬂat coordinate frame exists in V ,

that is, a coordinate frame in terms of which g

µν

(x) = g

0

µν

everywhere in V .

The Ricci curvature is a two-index tensor deﬁned by contracting the Riemann curvature:

R

κα

= R

µ

κµα

. (4.12)

The Ricci scalar R is deﬁned by contracting this once again, but because there are only

two subscript indices, this contraction must go with the inverse metric tensor:

R = g

µν

R

µν

. (4.13)

8

With some eﬀort, one can derive that the Riemann tensor obeys the following (partial)

diﬀerential equations, called Bianchi identity:

D

α

R

µ

κβγ

+ D

β

R

µ

κγα

+ D

γ

R

µ

καβ

= 0 . (4.14)

From that, we derive that the Ricci tensor obeys

g

µν

D

µ

R

να

−

1

2

D

α

R = 0 . (4.15)

5. Gravity

Consider a coordinate frame ¦x

µ

¦ where g

µν

is time independent: ∂

0

g

µν

= 0 , and a

particle that, at one instant, is at rest in this coordinate frame: dx

µ

/dτ = (1, 0, 0, 0) .

Then, according to Eq. (3.5), it will undergo an acceleration

d

2

x

i

dτ

2

= −Γ

i

00

=

1

2

g

ij

∂

j

g

00

. (5.1)

Since this acceleration is independent of the particle’s mass, this is a perfect description of

a gravitational force. In that case, −

1

2

g

00

can serve as an expression for the gravitational

potential (note that, usually, g

00

is negative). This is how the use of curved coordinates

can serve as a description of gravity – in particular there must be curvature in the time

dependence.

From here it is a small step to think of a space-time where the metric g

µν

(x) can

be any diﬀerentiable function of the coordinates x . Coordinates x in terms of which

g

µν

is completely constant do not have to exist. The gravitational ﬁeld of the Earth, for

instance, can be modelled by choosing g

00

(x) to take the shape of the Earth’s gravitational

potential. We then use Eqs. (3.5) and (4.9) to describe the motion of objects in free fall.

This is the subject of the discipline called General Relativity.

Of course, no coordinate frame exists in which all objects on or near the Earth move

in straight lines, and therefore we expect the Riemann curvature not to vanish. Indeed,

we need to have equations that determine the connection ﬁeld surrounding a heavy object

like the Earth such that it describes the gravitational ﬁeld correctly. In addition, we wish

these equations to be invariant under Lorentz transformations. This is achieved if the

equations can be written entirely in terms of vectors and tensors, i.e. all terms in the

equations must transform as such under coordinate transformations. The gravitational

equivalence principle requires that they transform as such under all (diﬀerentiable) curved

coordinate transformations.

Clearly, the mass density, or equivalently, energy density (x, t) must play the role as

a source. However, it is the 00 component of a tensor T

µν

(x) , the mass-energy-momentum

distribution of matter. So, this tensor must act as the source of the gravitational ﬁeld.

Einstein managed to ﬁgure out the correct equations that determine how this matter

distribution produces a gravitational ﬁeld. T

µν

(x) is deﬁned such that in ﬂat space-time

(with c = 1 ), T

0

0

= −T

00

= (x) is the energy distribution, T

i0

= T

0i

is the matter

9

ﬂow, which equals the momentum density, and T

ij

is the tension; for a gas or liquid with

pressure p , the tension is T

ij

= −p δ

ij

. The continuity equation in ﬂat, local coordinates

is

∂

i

T

iµ

−∂

0

T

0µ

= 0 , µ = 0, 1, 2, 3. (5.2)

Under general coordinate transformations, T

µν

transforms as a tensor, just as g

µν

does,

see Eq. (3.2).

In curved coordinates, or in a gravitational ﬁeld, the energy-momentum tensor does

not obey the continuity equation (5.2), but instead:

g

µν

D

µ

T

να

= 0 , (5.3)

So, the partial derivative ∂

µ

has been replaced by the covariant derivative. This means

that there is an extra term containing the connection ﬁeld Γ

λ

αβ

. This is the gravitational

ﬁeld, which adds or removes energy and momentum to matter.

Einstein’s ﬁeld equation now reads:

R

µν

−

1

2

Rg

µν

= −8πGT

µν

, (5.4)

where G is Newton’s constant. The second term in this equation is crucial. In his ﬁrst

attempts to write an equation, Einstein did not have this term, but then he hit upon

inconsistencies: there were more equations than unknowns, and they were, in general,

conﬂicting. Now we know the importance of the equation for energy-momentum conser-

vation (5.3), written more compactly as D

µ

T

µ

ν

= 0 . It matches precisely the Bianchi

identity (4.15) for the Ricci tensor, because that can also be written as

g

µν

D

µ

(R

να

−

1

2

Rg

να

) = 0 . (5.5)

6. The Schwarzschild Solution

When Einstein found his equation, Eq. (5.4), end of 1915, he quickly derived approximate

solutions, in order to see its consequences for observations, so that Eddington could set

up his expedition to check the deﬂection of star light by the gravitational ﬁeld of the sun.

Einstein, however, did not expect that the equation could be solved exactly. It was Karl

Schwarzschild, in 1916, who discovered that an exact, quite non-trivial solution can be

found. We will here skip the details of its derivation, which is straightforward, though

somewhat elaborate, and we will see more of that later. Schwarzschild’s description of

the metric g

µν

(x) that solves Einstein’s equations is most easily expressed in the modern

notation:

ds

2

= g

µν

dx

µ

dx

ν

= −(1 −2M/r) dt

2

+

dr

2

1 −2M/r

+ r

2

dΩ

2

, (6.1)

dΩ

2

= dθ

2

+ sin

2

θ dϕ

2

, (6.2)

10

where Newton’s constant G has been absorbed in the deﬁnition of the mass parameter

1

:

M = Gm. The advantage of this notation is that one can read oﬀ easily what the

metric looks like if we make a coordinate transformation: just remember that dx

µ

is an

inﬁnitesimal displacement of a point in space and time. Notice from the dependence on

dx

0

2

, that indeed, −

1

2

g

00

is the gravitational potential −M/r , apart from the constant

1.

Like other researchers in the early days, Schwarzschild himself was very puzzled by

the singularity at r = 2M . He decided to replace the coordinate r by a “better” radial

coordinate, let’s call it ˜ r , deﬁned as ˜ r = (r

3

−(2M)

3

)

1/3

. The reason for this substitution

was that Schwarzschild used simpliﬁed equations that only hold if the space-time-volume

element, det(g

µν

) = −1 , and the shift he used simply subtracts an amount (2M)

3

from

the space-time volume enclosed by r . Now, the singularity occurs at the “origin”, ˜ r = 0 .

Schwarzschild died only months after his paper was published. His solution is now famous,

but the substitution r → ˜ r (in the paper, the notation is diﬀerent) was unnecessary. The

apparent singularity at r = 2M is easier to describe when it is kept right where it is,

though indeed, we can use any coordinate frame we like to describe this metric. We

emphasize that, whether or not the singularity is moved to the origin, only depends on

the coordinate frame used, and has no physical signiﬁcance whatsoever.

2

One elegant coordinate substitution is the replacement of r and t by the Kruskal-

Szekeres coordinates x and y , which are deﬁned by the following two equations:

xy =

r

2M

−1

e

r/(2M)

, (6.3)

x/y = e

t/(2M)

. (6.4)

The angular coordinates θ and ϕ are kept the same. By taking the log of Eq. (6.3) and

(6.4), and partially diﬀerentiating with respect to x and y , we read oﬀ:

dx

x

+

dy

y

=

dr

r −2M

+

dr

2M

=

dr

2M(1 −2M/r)

, (6.5)

dx

x

−

dy

y

=

dt

2M

. (6.6)

The Schwarzschild metric is now given by

ds

2

= 16M

2

1 −

2M

r

dx dy

xy

+ r

2

dΩ

2

=

32M

3

r

e

−r/(2M)

dx dy + r

2

dΩ

2

. (6.7)

Notice now that, in the last expression, the zero and the pole at r = 2M have

cancelled out. The function r(x, y) can be obtained by inverting the algebraic expression

1

throughout these notes, we will denote the total mass of an object by m, and use the symbol M

for Gm.

2

There are some heated discussions of this on weblogs of amateur physicists who did not grab this

point!

11

0 2M

t

r

h

o

r

i

z

o

n

y

x

xy = −1

f

u

t

u

r

e

h

o

r

i

z

o

n

p

a

s

t

h

o

r

i

z

o

n

xy = −1

ΙΙ

Ι

0

b) a)

Figure 2: a ) The black hole in the Schwarzschild coordinates r, t . The horizon

is at r = 2M . b ) Kruskal-Szekeres coordinates; here, the coordinates of the

horizon are at x = 0 and at y = 0 . The orientation of the local lightcones is

indicated. Thin red lines are the time = Constant limes in the physical part

of space-time.

(6.3) and is regular in the entire region xy > −1 . In particular, nothing special seems to

happen on the two lines x = 0 and y = 0 . Apparently, there is no physical singularity

or curvature singularity at r → 2M . We do notice that the line x = 0, θ and ϕ both

constant, is lightlike, since two neighboring points on that line obey dx = dθ = dϕ = 0 ,

and this implies that ds = 0 , regardless the value of dy . similarly, the line y = 0 is

lightlike. Indeed, we can also read oﬀ from the original expression (6.1) that if r = 2M ,

the lines with constant θ and ϕ are lightlike, as ds = 0 regardless the value of dt .

The line y = 0 is called the future horizon and the line x = 0 is the past horizon (see

Section 10).

An other important thing to observe is that Eq. (6.4) attaches a real value for the

time t when x and y both have the same sign, such as is the case in the region marked

I in Fig. 2 b , but if xy < 0 , as in region II , the coordinate t gets an imaginary part.

This means that region II is not part of our universe. Actually, t does not serve as a

time coordinate there, but as a space coordinate, since there, dt

2

enters with a positive

sign in the metric (6.1). r is then the time coordinate.

Even if we restrict ourselves to the regions where t is real, we ﬁnd that, in general,

every point (r, t) in the physical region of space-time is mapped onto two points in the

(x, y) plane: the points (x, y) and (−x, −y) are mapped onto the same point (r, r) . This

leads to the picture of a black hole being a wormhole connecting our universe to another

universe, or perhaps another region of the space-time of our universe. However, there are

no timelike or light like paths connecting these two universes. If this is a wormhole at all,

it is a purely spacelike one.

12

7. The Chandrasekhar Limit

Consider Einstein’s equation (5.4), and some spherically symmetric, stationary distribu-

tion of matter. Let p(r) be the r dependent pressure, and (r) the r dependent local

mass density. An equation of state for the material relates p to . In terms of an aux-

iliary variable m(r) , roughly to be interpreted as the gravitational mass enclosed within

a sphere with radius r , and putting c = 1 , one can derive the following equations from

General Relativity:

dp

dr

= −G

( + p)(m + 4πp r

3

)

r

2

(1 −2Gm/r)

, (7.1)

dm

dr

= 4π r

2

. (7.2)

These equations are known as the Tolman-Oppenheimer-Volkoﬀ equations. The last of

these, Eq. (7.2) seems to be easy to interpret. The ﬁrst however, Eq. (7.1), seems to

imply that not only energy but also pressure causes gravitational attraction. This is a

peculiar consequence of the trace part of Einstein’s equation (5.4). In many cases, such

as the calculation of the gravitational forces between stars and planets, the pressure term

cancels out precisely. This is when there is a boundary where the pressure vanishes.

The resulting space-time metric is calculated to take the form

ds

2

= −A(r)dt

2

+ B(r)dr

2

+ r

2

(dθ

2

+ sin

2

θ dϕ

2

) , (7.3)

where

B(r) =

1

1 −2Gm(r)/r

; (7.4)

and

log(A(r)B(r)) = −8πG

∞

r

(p + ρ) r dr

1 −2Gm(r)/r

. (7.5)

Note that the Tolman-Oppenheimer-Volkoﬀ equations (7.1) and (7.2) are exact, as soon

as spherical symmetry and time independence are assumed.

There is no stable solution of the Tolman-Oppenheimer-Volkoﬀ equations if, anywhere

along the radius r , the enclosed gravitational mass M(r) = Gm(r) exceeds the value

r/2 . If the equation of state allows the pressure to be high while the density is small,

then large amounts of total mass will still show stable solutions, but if we have a liquid

that is cool enough to show a ﬁxed energy density

0

even when the pressure is low,

the enclosed mass would be approximately

4

3

π

0

r

3

, so with suﬃciently large quantities of

mass you can always exceed that limit. Thus, at suﬃciently low temperatures, no stable,

non-singular solution can exist if the baryonic mass N

B

exceeds some critical value.

Integrating inwards, one ﬁnds that there will be values of r where M(r)/r exceeds the

critical value 1/2 so that A(r) and B(r) develop singularities.

13

Substituting some realistic equation of state at suﬃciently low temperature, one de-

rives that the smallest amount of total mass needed to make a black hole is then a little

more than one solar mass. The Chandrasekhar limit refers to the largest amount of

mass one can make of a substance where only electron pressure resists the gravitational

attraction. This limit is about 1.44 solar masses.

One must ask what happens when larger quantities of mass are concentrated in a small

enough volume. If no stable solution exists, this must mean that the system collapses

under its own weight. What will happen to it?

8. Gravitational Collapse

An extreme case is matter of the form where the pressure p vanishes everywhere. This

is called dust. When at rest, in a local Lorentz frame, dust has only an energy density

T

00

= − while all other components of T

µν

vanish. In any other coordinate frame, the

energy-momentum tensor takes the form

T

µν

dust

= −(x)v

µ

v

ν

, (8.1)

where v

µ

is the local velocity dx

µ

/dτ of the dust grains.

In that case (and if we insist on spherical symmetry, so that the total angular mo-

mentum vanishes), gravitational implosion can never be avoided. It is instructive to show

some simple exact solutions.

Consider as initial state a large sphere of matter contracting at a certain speed v . We

could take v to be anything, but for simplicity we here choose it to be the velocity of

light. Thus, at t →−∞ we take for the energy density T

0

0

(and for simplicity G = 1 ),

T

0

0

(r, t) →

0

δ(r + t)/r

2

, (8.2)

where the factor r

−2

was inserted to ensure conservation of energy at inﬁnity:

E =

∞

0

dr 4πr

2

T

0

0

(r, t) = 4π

0

. (8.3)

Thus, at t →−∞, matter is assumed to be conﬁned into a thin, dense shell with radius

r →[t[ .

With this initial condition, and the equation of state p = 0 , it is not so diﬃcult simply

to guess the exact solution: we assume the metric to be stationary both before and after

the passage of the dust shell, but while the dust shell passes there is a jump proportional

to a theta step function. Both inside the dust shell and outside, spherical symmetry

demands that the only admissible solution will then be the Schwarzschild metric with

mass parameter M , however, M outside is diﬀerent from M inside. With our initial

condition (8.2), we have to choose M inside to be zero, but, for future use, we will also

consider more general solutions, with diﬀerent values for M

in

and M

out

. We will have to

verify afterwards that the conﬁguration obtained is indeed a correct solution, but we can

14

already observe that spherical symmetry would not have left us any alternative. What

has to be done now is to carefully formulate the matching conditions of the two regions

at the location of the contracting dust shell.

space

s

i

n

g

u

l

a

r

i

t

y

h

o

r

i

z

o

n

t

i

m

e

M = 0

M

1

M

2

M

3

b) a)

M = 0

M

1

M

2

M

3

r

t

com

Figure 3: Several shells of matter (shaded lines) implode to form a black hole,

whose mass M increases. In a) , time is neither the Schwarzschild time nor

the t

com

coordinate, but it indicates the causal order of events. The dotted

line is the location of the horizon. In b) , the coordinates t

com

and r are

used. The dotted line here is the apparent Schwarzschild horizon r = 2M .

Here, 0 < M

1

< M

2

< M

3

.

The contracting dust shell follows a lightlike geodesic in the radial direction, given by

ds

2

= 0 , or

dr

dt

= −

A

B

=

2M

r

−1 , (8.4)

so that

dt

dr

=

−r

r −2M

→ t(r) = −r −2M log(r −2M) . (8.5)

This is a reason to use a modiﬁed coordinate frame both in the inside region and the out-

side region. Inside, we use the Schwarzschild metric with coordinates (t

in

, r, θ, ϕ) , and

outside we use (t

out

, r, θ, ϕ) , but in both regions we make the transition to coordinates

15

(t

com

, r, θ, ϕ) , where

t

com

= t

in

+r + 2M

in

log(r −2M

in

) ,

t

com

= t

out

+ r + 2M

out

log(r −2M

out

) . (8.6)

Remember that for our original problem, M

in

= 0 , so that, according to the initial

condition (8.2), the dust shell moves at the orbit t

com

= 0 . We call t

com

the co-moving

time. t

com

= C

st

is a geodesic in both regions. The matching condition will now be that

at the points t

com

= 0 the two regions are stitched together. The coordinates r, θ and

ϕ will be the same for both regions (otherwise the metric g

µν

would show inadmissible

discontinuities

3

). At t

com

< 0 we have M = M

in

and at t

com

> 0 we have M = M

out

.

In terms of the new coordinates, the metric is

ds

2

= −A

dt −

B

A

dr

2

+Bdr

2

+r

2

dΩ

2

= −Adt

2

+ 2 dt dr + r

2

dΩ

2

=

−1 +

µ(t)

r

dt

2

+ 2 dt dr + r

2

dΩ

2

, (8.7)

where

µ(t) = 2M(t) , M(t) = θ(t)M

out

+ θ(−t)M

in

, (8.8)

and we dropped the subscript “com” for the time coordinate t .

In fact, any monotonously rising function µ(t) will be a solution to Einstein’s equa-

tions where dust ﬂows inwards with the speed of light. To check the solution now, let us

evaluate the Ricci curvature for the metric (8.7) in these coordinates:

g

00

= −1 +

µ

r

, g

10

= g

01

= 1 , g

11

= 0 ,

g

00

= 0 , g

10

= g

01

= 1 , g

11

= 1 −

µ

r

; (8.9)

deﬁning ˙ µ = dµ/dt , we ﬁnd

4

for the Christoﬀel symbols, Γ

αµν

=

1

2

(∂

µ

g

αν

+∂

ν

g

αµ

−∂

a

g

µν

) ,

Γ

000

=

˙ µ

2r

, Γ

100

=

µ

2r

2

, Γ

010

= Γ

001

= −

µ

2r

2

;

Γ

122

= −r , Γ

133

= −r sin

2

θ ,

Γ

121

= Γ

221

= r , Γ

313

= Γ

331

= r sin

2

θ ,

Γ

233

= −r

2

sin θ cos θ , Γ

323

= Γ

332

= r

2

sin θ cos θ ; (8.10)

3

In a space-time where the Ricci tensor is allowed to have a Dirac delta distribution, there must always

be a coordinate frame such that: the second derivatives of the metric g

µν

may have delta peaks, but

the ﬁrst derivatives have at most discontinuities in the form of step functions, while the metric itself is

continuous. If then a coordinate transformation is applied with a discontinuous ﬁrst derivative, such as

x → (a + b θ(y))y , with a > 0 and a + b > 0 , the metric g

xx

may show a discontinuity; compare the

discontinuity in g

00

as a function of t

com

in Eq. (8.9).

4

Remember that indices are raised and lowered by multiplying these ﬁelds with the metric tensor g

µν

or its inverse, g

µν

.

16

Γ

0

00

=

µ

2r

2

, Γ

1

10

= Γ

1

01

= −

µ

2r

2

,

Γ

1

00

=

˙ µ

2r

+

µ

2r

2

−

µ

2

2r

3

,

Γ

0

22

= −r , Γ

0

33

= −r sin

2

θ , Γ

1

22

= µ −r , Γ

1

33

= (µ −r) sin

2

θ ,

Γ

2

12

= Γ

2

21

=

1

r

, Γ

3

13

= Γ

3

31

= cot θ ;

√

−g = r

2

sin θ , log

√

−g = 2 log r + log θ . (8.11)

Inserting these in the equation for the Ricci curvature,

R

µν

= −(log

√

−g)

,µ,ν

+ Γ

α

µν,α

−Γ

β

αµ

Γ

α

βν

+ Γ

α

µν

(log

√

−g)

α

, (8.12)

we ﬁnd that

R

00

=

˙ µ

r

2

, (8.13)

while all other components of R

µν

in these coordinates vanish. It follows also for the

trace

5

that R = 0 , because g

00

= 0 . Hence

GT

00

=

−1

8π

˙ µ

r

2

=

−

˙

M

4πr

2

, (8.14)

while all other components, notably also T

11

, vanish. To see that this is indeed the

energy momentum tensor of our dust shell, we note that, in our co-moving coordinates,

the 4-velocity is v

µ

= (0, −Λ, 0, 0) , where Λ tends to inﬁnity (our “dust” goes with the

speed of light). From Eq. (8.9), we derive that v

µ

= (−Λ, 0, 0, 0) , and we have agreement

with Eq. (8.14) if

G =

˙

M

4πr

2

Λ

2

. (8.15)

We have

v

µ

= −Λ∂

µ

t

com

, (8.16)

so that, in the original Schwarzschild coordinates, where t

com

is replaced by t

in

or t

out

,

according to Eq. (8.6),

v

µ

= Λ

−1,

−1

1 −2M/r

, 0, 0

; v

µ

= Λ

1

1 −2M/r

, −1, 0, 0

; (8.17)

here M is the local mass parameter. From the second expression in Eq. (8.17), we see

that indeed the shell is moving inwards, with the local speed of light. The situation is

5

The fact that R = 0 ensues from our choice to have the dust move with the speed of light. Of course,

this is a limiting case, where all of the energy of the dust is kinetic, and the rest mass is negligible. The

Ricci scalar R refers to this fact, that the dust has negligible rest mass.

17

sketched in Figures 3 a) and b) . One readily ﬁnds that, if M

in

is taken to be zero,

M

out

is indeed the total energy E of our initial dust shell, deﬁned while it was still at

r →∞, in Eq. (8.3).

With a bit more work, this exercise can be repeated for dust going slower than the

speed of light. What we see is that, just behind the ﬁrst dust shell, the Schwarzschild

metric emerges. Subsequent shells of dust go straight through the horizon, generating

Schwarzschild metrics with larger mass parameters M . Taking M a continuous function

of t

com

leads to a description of less singular, spherically symmetric clouds of dust,

coalescing to form a black hole.

At this stage it is very important to observe that this description of “gravitational

collapse” allows for small perturbations to be added to it. For instance, one might assume

a tiny amount of angular momentum or other violations of spherical symmetry in the

initial state. This is what we mean when we say that the solution is ‘robust’. The

horizon at r = 2M might wobble a bit, but it cannot be removed by small perturbations

only. This is because the horizon is not a true singularity but rather an artefact of the

coordinates chosen. The singularity at r = 0 on the other hand, is very sensitive to

small perturbations, but it does not play a role in the physically observable properties of

the black hole; it is well hidden way behind the horizon (an observation called Cosmic

Censorship).

9. The Reissner-Nordstr¨om Solution

The Maxwell equations in curved space-time, when written in terms of the antisymmetric,

covariant tensor F

µν

(x) , are easy to ﬁnd by replacing partial derivatives by covariant

derivatives:

• The homogeneous Maxwell equation remains the same:

∂

α

F

βγ

+∂

β

F

γα

+ ∂

γ

F

αβ

= 0 , (9.1)

because the contributions of the connection ﬁelds cancel out due to the complete

antisymmetry under permutations of α, β , and γ . Hence, we still have a vector

potential ﬁeld A

µ

obeying

F

µν

= ∂

µ

A

ν

−∂

ν

A

µ

. (9.2)

• The inhomogeneous Maxwell equation is now

D

µ

F

µ

ν

= g

αβ

D

α

F

βν

= −J

ν

, (9.3)

where J

ν

(x) is the electro magnetic charge and current distribution. This can be

rewritten as

∂

µ

(

√

−g F

µν

) = −

√

−g J

ν

, (9.4)

18

where the quantity g is the determinant of the metric:

g = det

µ,ν

(g

µν

) , (9.5)

so that we have the conservation law

∂

µ

(

√

−g J

µ

) = 0 , (9.6)

because the Maxwell ﬁeld F

µν

is antisymmetric in its two indices.

• The energy momentum distribution of the Maxwell ﬁeld is

T

µν

= −F

µα

F

α

ν

+ (

1

4

F

αβ

F

αβ

−J

α

A

α

)g

µν

. (9.7)

Spherical symmetry can still be used as a starting point for the construction of a solution

of the combined Einstein-Maxwell equations for the ﬁelds surrounding a “planet” with

electric charge Q and mass m. Just as Eq. (7.3) we choose

ds

2

= −Adt

2

+Bdr

2

+ r

2

(dθ

2

+ sin

2

θ dϕ

2

) , (9.8)

but now also a static electric ﬁeld, deﬁned by E

i

(x) = F

0i

= −F

i0

:

E

r

= E(r) ; E

θ

= E

ϕ

= 0 ;

B = 0 . (9.9)

Let us assume that the source J

µ

of this ﬁeld is inside the planet and we are only

interested in the solution outside the planet. So there we have

J

µ

= 0 . (9.10)

Since g = −ABr

4

sin

2

θ , and F

0r

= −

1

AB

E(r) , the inhomogeneous Maxwell law (9.4)

implies

∂

r

E(r)r

2

√

AB

= 0 , (9.11)

and consequently,

E(r) =

Q

√

AB

4πr

2

, (9.12)

where Q is an integration constant, to be identiﬁed with electric charge since at r →∞

both A and B tend to 1 .

The homogeneous parts of Maxwell’s law are automatically obeyed because there is a

ﬁeld A

0

(potential ﬁeld) with

E

r

= −∂

r

A

0

. (9.13)

19

The ﬁeld (9.12) contributes to T

µν

:

T

00

= −E

2

/2B = −AQ

2

/32π

2

r

4

; (9.14)

T

11

= E

2

/2A = BQ

2

/32π

2

r

4

; (9.15)

T

22

= −E

2

r

2

/2AB = −Q

2

/32π

2

r

2

, (9.16)

T

33

= T

22

sin

2

θ = −Q

2

sin

2

θ /32π

2

r

2

. (9.17)

We ﬁnd

T

µ

µ

= g

µν

T

µν

= 0 ; R = 0 , (9.18)

a general property of the free Maxwell ﬁeld. In this case we have (putting G = 1 )

R

µν

= −8π T

µν

. (9.19)

Herewith the Einstein equation (5.4) lead to the following solution:

A = 1 −

2M

r

+

Q

2

4πr

2

; B = 1/A. (9.20)

This is the Reissner-Nordstr¨om solution (1916, 1918).

If we choose Q

2

/4π < M

2

there are two “horizons”, the roots of the equation A = 0 :

r = r

±

= M ±

M

2

−Q

2

/4π . (9.21)

Again these singularities are artifacts of our coordinate choice and can be removed by

generalizations of the Kruskal-Szekeres coordinates.

We have not shown the complete derivations of these solutions. In principle, the

information given in these notes should suﬃce to derive them, but if further details are

needed we refer to the various more elaborate texts in General Relativity. In these lecture

notes we concentrate on the physical properties of the various metrics that were found.

10. Horizons

Consider the metric (6.1), and a light ray going radially inward. The equation for such a

light ray is

ds

2

= 0 ; dΩ = 0 , or

dt

dr

=

1

1 −2M/r

. (10.1)

The solution of this equation is

t = t

0

±(r + 2M log(r −2M)) , (10.2)

where t

0

is an integration constant, and we choose the minus sign, so that, at r very

close to 2M ,

r(t) →2M + e

(t

0

−t)/2M−1

. (10.3)

20

Similarly, a ray going radially outward is given by

r(t) →2M + e

(t−t

1

)/2M−1

. (10.4)

Note that neither of these light rays ever pass through the barrier r = 2M . Rays going

at an angle rather than radially also will not pass through this point. This is why this

point is called a horizon. In general, a horizon forms when the coeﬃcient A in the metric

(9.8) tends to zero suﬃciently fast.

In terms of the Kruskal coordinates, Eqs. (6.7), the inwards and outwards light rays

are easier to ﬁnd: the in-rays are at x = C

st

, and the out-rays at y = C

st

, where C

st

is any positive constant. In these coordinates, however, we see something new. The light

rays do actually cross the horizon. Beyond the horizon, they hit upon the singularity at

xy = −1 , which is the point r = 0 .

In solutions that are more general than the Schwarzschild solution, the horizon is

deﬁned to be the boundary line between two regions of space-time. Region I is the region

deﬁned by all space-time points x from which a geodesic can start, heading towards the

future direction, that reaches the boundary at r = ∞. Region II is the collection of

points that have no such geodesics attached to them. This means that particles, or indeed

astronauts, cannot reach inﬁnity from these points, regardless the value and direction of

their initial velocity. The boundary line between I and II , the horizon, is a surface

formed by lightlike geodesics pointing radially outwards.

There is also such a surface of lightlike geodesics pointing inwards. For the Schwarz-

schild solution, both horizons are at r = 2GM , but we see in the Kruskal frame that

actually, the two horizons do not coincide.

In the Reissner-Nordstr¨om solution, we see that the function A(r) has two zeros. The

largest one, at

r = M +

M

2

−Q

2

/4π , (10.5)

coincides with the Schwarzschild horizon in the limit Q →0 . It has the same properties

as the Schwarzschild horizon.

The second horizon, at

r = M −

M

2

−Q

2

/4π , (10.6)

goes to the singularity r = 0 in the limit Q → 0 . But at ﬁnite Q it is also a lightlike

surface.

Whatever happens within the horizon might be called physically irrelevant, since in-

formation concerning the interior region cannot be sent out using light rays. However,

later we will see that quantum eﬀects do depend on details of the horizon region, and to

understand these, one may have to pass beyond the horizon.

The true, physical singularity that occurs at r = 0 , is far hidden from observation;

this singularity may be compared with singularities in physical equations when some

parameters such as the time parameter become complex. Kepler’s elliptical orbits, for

instance, show delicate singularities at points in complex time, but the planets in the

solar system do not seem to be bothered about that.

21

11. The Kerr and Kerr-Newman Solution

A fast rotating planet has a gravitational ﬁeld that is no longer spherically symmetric but

is only symmetric under rotations around the z -axis. We here just give the solution:

ds

2

= −dt

2

+ (r

2

+ a

2

) sin

2

θdϕ

2

+

2Mr(dt −a sin

2

θdϕ)

2

r

2

+ a

2

cos

2

θ

+(r

2

+a

2

cos

2

θ)

dθ

2

+

dr

2

r

2

−2Mr + a

2

. (11.1)

This solution was found by R. Kerr in 1963. To prove that this is indeed a solution of

Einstein’s equations requires patience but is not diﬃcult. For a derivation using more ele-

mentary principles more powerful techniques and machinery of mathematical physics are

needed. The free parameter a in this solution can be identiﬁed with angular momentum:

J = a M . (11.2)

c) The Newman et al solution

For sake of completeness we also mention that rotating planets can also be electrically

charged. The solution for that case was found by Newman et al in 1965. The metric is:

ds

2

= −

∆

Y

(dt −a sin

2

θdϕ)

2

+

sin

2

θ

Y

(adt −(r

2

+ a

2

)dϕ)

2

+

Y

∆

dr

2

+ Y dθ

2

, (11.3)

where

Y = r

2

+ a

2

cos

2

θ , (11.4)

∆ = r

2

−2Mr + Q

2

/4π + a

2

. (11.5)

The vector potential is

A

0

= −

Qr

4πY

; A

3

=

Qra sin

2

θ

4πY

. (11.6)

Eq. (11.2) here also describes the total angular momentum in the solution. The Kerr-

Newman solution is the most general stationary solution for a black hole with electric

charge, if no matter is present.

As with the Reissner-Nordstr¨om solution, the zeros of the function ∆(r) ar not true

singularities but rather coordinate singularities. The only genuine singularity in the cur-

vature of space and time occurs where Y (r, θ) = 0 , but this occurs only when both r

and θ are zero: the singularity lies along the equator at r = 0 .

Exercise: show that when Q = 0 , Eqs. (11.1) and (11.3) coincide.

Exercise: ﬁnd the non-rotating magnetic monopole solution by postulating a radial

magnetic ﬁeld.

22

Exercise: derive the gravitational ﬁeld for a non-relativistic source by linearizing

Einstein’s equation (5.4), and use that to derive Eq. (11.2).

Exercise for the advanced student: describe geodesics in the Kerr solution.

12. Penrose diagrams

h

o

r

i

z

o

n

h

o

r

i

z

o

n

x

P

y

P

r = 0 singularity

r = 0 singularity

III

II

IV I

∞

+

∞

−

∞

Figure 4: The Penrose diagram for the Schwarzschild solution

∞ ∞

∞

+

∞

−

∞

∞

−

∞

+

r

=

0

h

o

r

i

z

o

n

r = 0

r

=

0

II

I

∞

+

∞

−

∞

c b a

Figure 5: a ) Penrose diagram for the Minkowski vacuum with Cartesian

transverse coordinates, b ) Minkowski space in polar coordinates; c ) Penrose

diagram for a black hole formed by matter (darker color represents matter

falling in).

It is of interest to ﬁnd coordinate systems that are such that they cover all of space-

time that is continuously connected to the region that one has studied before, preferably

avoiding any coordinate-induced singularities. This is not always possible, but we can

try to choose the best possible coordinates. A good example is the Kruskal-Szekeres

23

r

=

r

+

r

=

r

+

r

=

r +

r

=

r +

r

=

0

s

i

n

g

u

l

a

r

i

t

y

r

=

0

s

i

n

g

u

l

a

r

i

t

y

r

=

r

−

r

=

r

−

r

=

r −

r

=

r −

r

=

r

+

r

=

r

+

r

=

r +

r

=

r +

r

=

0

s

i

n

g

u

l

a

r

i

t

y

r

=

0

s

i

n

g

u

l

a

r

i

t

y

r

=

r

−

r

=

r

−

r

=

r −

r

=

r −

VII

VI V

III

II

IV I

∞

+

∞

−

∞

r

=

r

+

r

=

r

+

r

=

r +

r

=

r +

r

i

n

g

s

i

n

g

u

l

a

r

i

t

y

r

i

n

g

s

i

n

g

u

l

a

r

i

t

y

r

=

r

−

r

=

r

−

r

=

r −

r

=

r −

r

=

r

+

r

=

r

+

r

=

r +

r

=

r +

r

i

n

g

s

i

n

g

u

l

a

r

i

t

y

r

i

n

g

s

i

n

g

u

l

a

r

i

t

y

r

=

r

−

r

=

r

−

r

=

r −

r

=

r −

VII

VI’ VI V’ V

III

II

IV I

∞

+

∞

−

∞

b a

Figure 6: Penrose diagram for the Reissner-Nordstr¨om ( a ) and the Kerr black

hole ( b ). The singularity at r = 0 is a natural boundary in case of Reissner-

Nordstr¨om, but it is a ring singularity in case of Kerr- and Kerr-Newman,

through which one can continue to asymptotically ﬂat universes V

and V I

,

containing a negative mass Kerr Newman black hole.

coordinate system, (6.3) — (6.4), for the Schwarzschild black hole. At every point in the

xy frame of these coordinates, light rays are constrained to form angles of 45

◦

or less

with the vertical (vertical meaning the line dx +dy = 0 ). Or, light rays themselves form

trajectories of the form dx ≥ 0, dy ≤ 0 , where one of the equal signs is reached as soon

as dθ = dϕ = 0 .

Now these are not the only coordinates with this property. If the x coordinate is

replaces by any monotonously increasing, diﬀerentiable function x

P

of x , and y by any

monotonously increasing diﬀerentiable function y

P

of y , we still have the same property.

This freedom we can use to obtain one other desirable feature: map the point x = ∞

to x

P

= 1 and the same for y and y

P

. Furthermore, we can assure that the r = 0

singularity is mapped onto a straight line, here the line x

P

−y

P

= 1 . In the Schwarzschild

case, this feature is reached if we choose

x = tan(x

P

π/2) , y = tan(y

P

π/2) . (12.1)

24

Space-time is then sketched in Fig. 4.

A Penrose diagram now is a representation of two of the space-time coordinates in

such a way that the local light cones always show that light rays go with a maximal

velocity +1 to the right or -1 to the left, so that the fastest way to transmit information

is by rays that are tilted by 45

◦

to the left or to the right, such as is the case in Figure

4. The other two coordinates, θ and ϕ usually deﬁne a two-sphere. Characteristic

boundaries are represented as much as possible by straight lines, which is usually possible

and has the advantage that the entire space-time can be represented in a ﬁnite patch of

the coordinates.

The diagram of Fig. 4 shows four regions of space-time, separated by horizons. Region

I is the region that can be reached from inﬁnity and from which one can also escape to

inﬁnity. Region II is the domain behind the horizon that can be reached by test objects

falling in, but from where no escape back to inﬁnity is possible. The r = 0 singularity

lies in the future of any test object there. Region III is a domain that cannot be reached

from inﬁnity, but escape to inﬁnity is allowed. Finally, region IV is only connected to

the physical spacetime I by spacelike geodesics.

The Penrose diagram of ﬂat Minkowski space-time is shown in Figure 5 a . Figure 5 b

describes a black hole formed by matter. We see that at negative times it corresponds to

that of Minkowski space-time. At early times, the point r = 0 in 3-space forms a timelike

geodesic; at later times it becomes spacelike

13. Trapped Surfaces

A black hole is characterized by the presence of a region in space-time from which no

trajectories can be found that escape to inﬁnity while keeping a velocity smaller than

that of light. This implies the presence of trapped surfaces there. We start from the

following deﬁnitions.

Consider a two-dimensional, closed, convex, spacelike surface S in a curved space-

time. Let A be the surface area of S (calculated using the induced metric on S . Deﬁne

a time coordinate t such that t = 0 on that surface. Suppose that our surface at t = 0

divides 3-space into two regions: an outer region V

1

and an inner region V

2

. A small

instant later, at time t = ε , 3-space is divided in three regions:

– an outer region V

1

that is spacelike separated from S ,

– an inner region V

2

that is also spacelike separated from S , and

– a region V

3

between V

1

and V

2

that can be reached by timelike geodesics from S .

Its boundary can be reached with lightlike geodesics from S .

Let S

1

be the boundary between V

1

and V

3

and S

2

be the boundary between V

2

and

V

3

. See Fig. 7. The surfaces S

1

and S

2

have areas A

1

and A

2

. Now, we deﬁne the

25

S

VV

3

V

2

V

1

S

2

S

1

Figure 7: A surface S at t = 0 as described in the text. A little later, at

t = ε , signals moving inwards and outwards divide 3-space into the regions

V

1

, V

2

and V

3

.

expansion rates θ

1

, θ

2

of these two surfaces as follows:

θ

1

=

dA

1

dε

, θ

2

=

dA

2

dε

. (13.1)

Under non-exotic circumstances, such as in a ﬂat space-time, certainly the outer surface

expansion rate is positive: θ

1

> 0 . The inner one is usually negative. However, inside a

black hole, we can have a trapped surface. S is called trapped iﬀ both expansion rates

are negative or zero:

θ

1

≤ 0 and θ

2

≤ 0 . (13.2)

A surface is marginally trapped if the equal sign in Eq. (13.2) holds. For a pure Schwarz-

schild black hole, the surface r = 2M is marginally trapped. This is because all light-like

geodesics leaving this surface have r = 2M , so that its area, which in the local induced

metric is 4π(2M)

2

, does not increase with time.

What happens in the presence of matter, when the solutions of Einstein’s equations

look a lot more complicated? In that case, we can still deﬁne trapped surfaces, and they

obey a number of important theorems. One of the most important theorems is:

If, in all locally regular coordinate frames, the matter distribution in a space-

time obeys the constraint that the energy density is non-negative anywhere,

or, in our notation,

T

00

≤ 0 (13.3)

in all coordinate frames, then

- a trapped surface stays trapped forever, and

- the area of the largest trapped surface can only stay constant or increase.

The importance of this theorem is that it shows that black holes cannot disappear once

they have been formed.

6

Indeed, other theorems show the inevitability of singularities

6

Of course, we have not yet considered quantum mechanical eﬀects. These can indeed cause black

holes to shrink, and presumably disappear altogether, see Section 18.

26

forming inside trapped surfaces after some time, such as the r = 0 singularity of the

Schwarzschild black hole.

We will not give the general proofs, but instead consider the much simpler spherically

symmetric case. Consider the most general spherically symmetric metric,

ds

2

= −A(r, t)dt

2

+ B(r, t)dr

2

+ C(r, t)drdt +r

2

dΩ

2

. (13.4)

It is the direct generalization of the vacuum solution (6.1). Actually, the cross term

C(r, t)drdt can easily be removed by a proper redeﬁnition of the time coordinate t , so

there is no loss of generality if we put

C(r, t) = 0 (13.5)

(the r coordinate is ﬁxed by demanding the angular dependence as in Eq. (13.4)).

It is now convenient to choose light cone coordinates x and y , which are deﬁned by

demanding that the lines x =constant and y =constant are in fact light rays ds = 0 .

This implies that the coeﬃcients for dx

2

and dy

2

must vanish. We then get the direct

generalization of the Kruskal metric (6.7), which in the presence of matter reads

ds

2

= 2A(x, y)dxdy + r

2

(x, y)dΩ

2

. (13.6)

From this metric, we calculate how it relates to the matter distribution T

µν

. Here follows

the calculation, which one might decide to skip at ﬁrst reading, but we list it to enable

the reader to check. The connection ﬁelds are easily calculated:

Γ

x

xx

=

∂

x

A

A

, Γ

x

θθ

=

−r ∂

y

r

A

, Γ

x

ϕϕ

=

−r ∂

y

r sin

2

θ

A

,

Γ

y

yy

=

∂

y

A

A

, Γ

y

θθ

=

−r ∂

x

r

A

, Γ

y

ϕϕ

=

−r ∂

x

r sin

2

θ

A

,

Γ

θ

xθ

= Γ

ϕ

xϕ

=

∂

x

r

r

, Γ

θ

yθ

= Γ

ϕ

yϕ

=

∂

y

r

r

,

Γ

θ

ϕϕ

= −cos θ sin θ , Γ

ϕ

θϕ

= cot θ , (13.7)

and all others are zero, except the ones obtained from the above by interchanging the two

subscript indices.

From these, the Ricci tensor can be derived, and one obtains

R

xx

=

2 ∂

x

A∂

x

r

Ar

−

2 ∂

2

x

r

r

, R

yy

=

2 ∂

y

A∂

y

r

Ar

−

2 ∂

2

y

r

r

,

R

xy

=

∂

x

A∂

y

A

A

2

−

∂

x

∂

y

A

A

−

2 ∂

x

∂

y

r

r

,

R

θθ

= 1 −

2 ∂

x

r ∂

y

r + 2 r ∂

x

∂

y

r

A

, R

ϕϕ

= sin

2

θ R

θθ

. (13.8)

This, we plug into Einstein’s equation, Eq. (5.4), where we, temporarily, ignore the factor

8πG:

T

xx

=

2 ∂

2

x

r

r

−

2/, ∂

x

A∂

x

r

Ar

=

A

r

∂

x

∂

x

r

A

, T

yy

=

A

r

∂

y

∂

y

r

A

, (13.9)

27

T

xy

=

1

r

2

(A −2 ∂

x

r∂

y

r −2 r ∂

x

∂

y

r) =

A −∂

x

∂

y

(r

2

)

r

2

, (13.10)

T

θθ

= −

r ∂

x

∂

y

log A + 2 ∂

x

∂

y

r

A

, T

ϕϕ

= sin

2

θ T

θθ

. (13.11)

Now, keeping y zero or very small, we can regard the coordinate x as our time

coordinate. In particular, we will use the ﬁrst of Eq. (13.9). We claim that the positive

energy condition will also require

T

xx

≤ 0 . (13.12)

Proof: Consider some given values for T

xx

, T

xy

, and T

yy

. Now go to a coordinate frame

¦τ, ¦ , where τ serves as the new time variable, and

x = λ( + τ) , y =

1

λ

( −τ) , (13.13)

where the parameter λ is chosen suﬃciently large. Since

∂

∂τ

= λ

∂

∂x

−

1

λ

∂

∂y

, (13.14)

we have

T

ττ

= λ

2

T

xx

−2T

xy

+λ

−2

T

yy

. (13.15)

Demanding this to be negative or zero for all λ implies Ineq. (13.12).

b

a

y

x

a′

Figure 8: In the lightcone coordinates ¦x, y¦ , we may have a point (x

0

, y

0

)

where ∂

x

r = 0, ∂

y

r > 0 , so that this point represents a trapped surface. If no

matter is present, the line a , corresponding to y = y

0

, then forms a series of

trapped surfaces. But if matter falls in (green arrow), then beyond that point

on the same line a

, ∂

x

r < 0 , so that a new line b emerges at y = y

1

> y

0

,

where ∂

x

r = 0 . This is a new trapped surface with a larger area.

Now suppose that, at some positive value of A, we have a marginally trapped surface,

so that, along a line y = y

0

= constant, we have ∂

x

r = 0 . Since y runs in the

negative time direction, we have ∂

y

r ≥ 0 . If there is no matter around, then according

28

to Eq. (13.9), r will keep the same value; the marginally trapped surface is then also a

horizon. However, if there is matter around, obeying Ineq. (13.12), w the quantity ∂

x

r

will be negative some time later. This means that the line y = y

0

is now well within

this trapped surface. We can now go to a slightly larger value of y to ﬁnd a marginally

trapped surface. Since ∂r/∂y > 0 (the surface is also trapped at the inner side), this

surface has a larger r value, hence a larger area. Thus, as soon as matter falls in, the

marginally trapped surface is replaced by a larger one. We can therefore conclude that

the area of the horizon increases when matter falls in. See Fig. 8.

We used spherical symmetry for this simple argument, but it can be generalized to

the non-symmetric case. All one needs to know is that all matter that falls through the

horizon, has a positive energy density in any locally regular coordinate frame. In that

case, the total area of the horizon can only increase.

14. The four laws of black hole dynamics

Consider the most general black hole solution, the Kerr-Newman solution (11.3). The

horizon occurs where ∆ = 0 , because at that point the lightlike geodesics going in and

going out coincide:

∆(r) = 0 ,

dr

dt

= 0 ,

dθ

dt

= 0 ,

dϕ

dt

=

a

r

2

+ a

2

. (14.1)

Deﬁning the roots of ∆ to be

r

±

= M ±

M

2

−a

2

−Q

2

/4π , (14.2)

we ﬁnd that the horizon is at r = r

+

, and its area is

Σ =

π

0

dθ

2π

0

dϕ sin θ (r

2

+

+ a

2

) = 4π(r

2

+

+a

2

) (14.3)

(where the dependence of the function Y (r, θ) dropped out).

The free parameters of this solution are M, Q and a . For reasons to become clear

shortly, we now wish to express these in terms of the three mutually independent param-

eters Q, J , and Σ. We have the following equations,

r

2

+

+ a

2

= Σ/4π , r

+

r

−

−a

2

= Q

2

/4π , 2aM = a(r

+

+ r

−

) = 2J . (14.4)

They allow us to eliminate r

+

, r

−

and a the following way:

4J

2

a

2

−

Σ

4π

+ a

2

Q

2

+ Σ

4π

2

= 0 ;

a

2

=

J

2

Σ

4πJ

2

+

1

16π

(Q

2

+ Σ)

2

. (14.5)

29

This gives us the dependence of M , the total mass/energy, on the independent parameters

Q, /J and Σ:

M

2

(Q, J, Σ) =

1

16π

Q

4

Σ

+ 2Q

2

+ Σ

+

4πJ

2

Σ

. (14.6)

How does this change upon small variations of our free parameters? We write

dM = τ dΣ + ΩdJ + φdQ . (14.7)

These derivative functions are now derived to be

τ =

r

+

−r

−

16π(r

2

+

+ a

2

)

, Ω =

a

r

2

+

+ a

2

, φ =

Qr

+

4π(r

2

+

+ a

2

)

. (14.8)

these now have the following interpretation. Ω is an angular velocity. For all systems

with angular momentum J , the increase in energy upon an increase of J is the angular

velocity. Indeed, this is the angular velocity that any object acquires when it goes through

the horizon, see Eq. (14.1).

Similarly, φ is the electrostatic potential for a test charge crossing at the horizon.

This is seen as follows. The vector potential (11.6) holds in the coordinates (t, r, θ, ϕ) .

If we want the vector potential for a test particle that rotates with angular velocity

Ω = a/(r

2

+

+ a

2

) , we have to transform to the co-rotating coordinates (t, r, θ, ˜ ϕ) , with

˜ ϕ = ϕ −Ωt . The vector ﬁeld transforms as follows:

˜

A

µ

d˜ x

µ

= A

µ

dx

µ

, (14.9)

and since d˜ ϕ = dϕ −Ωdt , we have at r = r

+

, where Y = r

2

+

+ a

2

cos

2

θ ,

A

0

dt + A

3

dϕ =

˜

A

0

dt +

˜

A

3

(dϕ −Ωdt) →

˜

A

3

= A

3

;

˜

A

0

= A

0

+ ΩA

3

= −

Qr

+

4πY

1 −

a

2

sin

2

θ

r

2

+

+ a

2

=

=

−Qr

+

4π(r

2

+

+ a

2

)

. (14.10)

This is the vector potential felt by the test charge with angular velocity Ω.

Σ is the area of the horizon, but what is τ ? In all respects, this equation resembles

the entropy equation in statistical mechanics. It was found by Bardeen, Carter and

Hawking, and they noticed this similarity. In that case, τ acts as a temperature. It

could not be the real temperature of a black hole, as was thought at ﬁrst, because the

black hole temperature is zero: nothing can come out, so also no thermal radiation. But

the similarity with the entropy law went further. Due to the trapped surface theorems,

we also know that the area of a horizon cannot decrease. This it has in common with

entropy. thus, the second law of thermodynamics has an analogy in black holes:

the second law of black hole physics states that the total area of all horizons cannot

decrease, just like the total entropy in thermodynamics.

30

The ﬁrst law of black hole physics is equation (14.7). It states that the increase of

mass of a black hole is the sum of all kinds of energy that is added to it. The amount

dU = τdΣ is then interpreted as heat energy.

The quantity

τ =

κ

8π

=

M

2

−a

2

−Q

2

/4π

8π(r

2

+

+a

2

)

, (14.11)

cannot normally go to zero, and it takes the same value all across the horizon, just like

the temperature for an object in equilibrium. κ is sometimes referred to as the “surface

gravity” at the horizon. Very near the horizon, we again replace the angular coordinate

ϕ by ˜ ϕ = ϕ −

a

r

2

+

+a

2

t , so that, at constant θ and ˜ ϕ, the metric (11.3) approaches

ds

2

→Y

−(r −r

+

)(r

+

−r

−

)

(r

2

+

+ a

2

)

2

dt

2

+

dr

2

(r −r

+

)(r

+

−r

−

)

. (14.12)

The ratio between the time component and the space component is therefore the square

of

(r −r

+

)(r −r

−

)

r

2

+

+ a

2

. (14.13)

Diﬀerentiating with respect to r gives something that could be called the gravitational

ﬁeld at the horizon, which is κ in Eq. (14.11). We return to this topic in Section 18.

Thus, the zeroth law of black hole dynamics states that the “temperature” τ , or the

“surface gravity”, is constant on the horizon.

The third law would be that τ cannot be zero. This, indeed, is a delicate limit. It

occurs when r

+

and r

−

coincide, or

a

2

+Q

2

/4π →M

2

. (14.14)

This is called the extreme limit of the Kerr-Newman black hole. It is dubious whether

this limit can be reached in practice, but this has not been proven. The extreme limit of

black holes has many special properties, and plays an important role in string theories.

15. Rindler space-time

Consider ordinary Minkowski space-time, described by the coordinates (t, x, y, z) , where

the metric is deﬁned as

ds

2

= −dt

2

+ dx

2

+ dy

2

+ dz

2

. (15.1)

It is instructive to transform towards the curved coordinates (τ, , ˜ x) , where

e

τ

= z + t , e

−τ

= z −t , ˜ x = (x, y) . (15.2)

31

A substitution

τ →τ + λ , (15.3)

where λ is a constant, would leave z

2

−t

2

invariant, and hence corresponds to a Lorentz

transformation in Minkowski space-time. We can be sure that Nature’s laws will not

change, and so, in this curved coordinate frame, the laws of nature are invariant under

translations of the new time variable τ . The metric in the new coordinates is

ds

2

= −

2

dτ

2

+ d

2

+ d˜ x

2

. (15.4)

In this coordinate frame, therefore, an observer experiences a gravitational potential pro-

portional to . Actually, at any position =

0

, this observer would be tempted to

redeﬁne time as

˜

t = τ/

0

, so that the gravitational potential would feel as V = /

0

,

with gradient 1/

0

. Therefore, the actual gravitational ﬁeld strength felt by the observer

is inversely proportional to the distance from the origin.

This space-time is called Rindler space-time, and it is very instructive for the study

of gravitational ﬁelds, since all physical phenomena observed in this world can be derived

from what they are in Minkowski space-time without any gravitational ﬁeld.

In fact, any small region very close to the horizon of a non-extremal black hole can de

compared with Rindler space-time. Near θ ≈ π/2 , replace the Schwarzschild coordinates

as follows,

t/4M = τ , 8M(r −2M) =

2

, ˜ x = (2Mθ, 2Mϕ) , (15.5)

then, close to r ≈ 2M , the metric (6.1) is

ds

2

≈ −

r −2M

2M

dt

2

+

2M dr

2

r −2M

+ d˜ x

2

=

= −

2

dτ

2

+ d

2

+ d˜ x

2

. (15.6)

In fact, the ﬂat Minkowski coordinates z and t , (15.2), are closely related to the Kruskal-

Szekeres coordinates (6.3)—(6.6).

Thus, we can ﬁnd out about quantum phenomena near a horizon by studying them

ﬁrst in ﬂat Minkowski space-time, then in Rindler space-time, and then in the black hole.

16. Euclidean gravity

Mathematical functions in space-time coordinates can often be extended to complex values

of these coordinates. There, they continue to obey the same equations. In particular, it

seems to be interesting to replace the time coordinate t by an imaginary time: t = i

˜

t .

In Euclidean space, the metric then becomes

ds

2

= +d

˜

t

2

+ dx

2

+ dy

2

+ dz

2

. (16.1)

32

The invariance group is then not the Poincar´e group with the Lorentz group as its homo-

geneous part, but it now has the orthogonal group SO(4) . Thus, Lorentz transformations

are replaced by ordinary rotations:

z

= z cos γ +

˜

t sin γ ,

˜

t

= −z sin γ +

˜

t cos γ ,

(16.2)

under which the metric (16.1) is invariant.

In Rindler spacetime, one can also extend to imaginary values of the Rindler time τ :

τ = i˜ τ ,

t = i

˜

t ,

z = cos ˜ τ ,

˜

t = sin ˜ τ .

(16.3)

This means that, in Euclidean space, the transition towards Rindler spacetime is nothing

more than a transition to cylindrical coordinates. Rindler time translations are simply

rotations in Euclidean space.

In the absence of matter, Einstein’s equations in ordinary spacetime remain unchanged

when we go to Euclidean spacetime. So we can take Schwarzschild’s solution and extend

it to Euclidean times:

ds

2

=

1 −

2M

r

d

˜

t

2

+

1

1 −2M/r

dr

2

+ r

2

dΩ

2

. (16.4)

To see what happens at the horizon, we do the same substitution in the Kruskal-Szekeres

coordinates. We ﬁnd that Eqs. (6.3)—(6.7) turn into

[x[ = [y[ , xy is real , (16.5)

and, writing [x[ = [y[ = ,

x = y

∗

= e

i

˜

t/(4M)

= ξ + iη , ξ = cos(

˜

t

4M

) , η = sin(

˜

t

4M

) . (16.6)

2

= ξ

2

+ η

2

=

r

2M

−1

e

r/(2M)

, (16.7)

ds

2

=

32M

3

r

e

−r/(2M)

(dξ

2

+ dη

2

) + r

2

dΩ

2

. (16.8)

We see that the solution is rotationally invariant in (ξ, η) space, and the metric is regular

at the origin. Time translations, also at very large values of r , are now rotations, and

these are periodic. So, after a time translation over one period,

T = 8πM , (16.9)

points in space and time return to their original positions. At large distances, this space-

time is not excactly ﬂat Euclidean spacetime, because points in Euclidean spacetime that

are separated by one period T in Euclidean time have to be identiﬁed as being the same

point. If they were not the same point, singularities would arise at the origin of (ξ, η)

space. This observation will be very important later on (Section 18).

The region of Euclidean space-time that we described, where the metric is positive

everywhere, only refers to the region r ≥ 2M , so, one cannot go through the horizon

here. The situation is sketched in Figure 9. We see that, asymptotically, this spacetime

is not described by ordinary ﬂat Euclidean spacetime, usually denoted as R

4

, but by a

cylinder, R

3

⊗S

1

, where S

1

stands for the circle.

33

Figure 9: The Schwarzschild black hole in Euclidean gravity. Asymptotically,

this spacetime is a cylinder.

17. The Unruh eﬀect

For the following sections some basic knowledge of quantum ﬁeld theory is required. We

consider a quantized scalar ﬁeld Φ(t, x) in Minkowski space-time, and we shall investigate

what it looks like in Rindler space-time. for the time being, we shall not need to include

interactions, so we are talking of a free ﬁeld. Its local commutation rules are

[Φ(t, x), Φ(t, x

)] = 0 , [Φ(t, x),

˙

Φ(t, x

)] = iδ

3

(x −x

) . (17.1)

Assuming this ﬁeld to obey the Klein-Gordon equation,

(

∂

2

−∂

2

0

−m

2

)Φ = 0 , (17.2)

one ﬁnds the Fourier mode expansion

Φ(t, x) =

d

3

k

2k

0

(

k)(2π)

3

a(

k)e

i

k·x−ik

0

t

+ a

†

(

k)e

−i

k·x+ik

0

t

, (17.3)

˙

Φ(t, x) =

−ik

0

d

3

k

2k

0

(

k)(2π)

3

a(

k)e

i

k·x−ik

0

t

−a

†

(

k)e

−i

k·x+ik

0

t

, (17.4)

where k

0

(

k) =

k

2

+ m

2

(always with the positive sign), and the operators a(

k) and

a

†

(

**k) obey the commutation rules for operators that respectively annihilate and create a
**

particle:

[a(

k), a(

k

)] = 0 , [a(

k), a

†

(

k

)] = δ

3

(

k −

k

) , (17.5)

In the Rindler space coordinates (15.2), the Klein-Gordon equation (17.2) reads

(∂

)

2

−∂

2

τ

+

2

(

˜

∂

2

−m

2

)

Φ = 0 . (17.6)

Solutions periodic in τ are

Φ

ω,

˜

k

(τ, , ˜ x) = K

ω,

1

2

µe

τ

,

1

2

µe

−τ

e

i

˜

k·˜ x

= K

ω,

1

2

µ,

1

2

µ

e

i

˜

k·˜ x−iωτ

, (17.7)

34

where µ

2

=

˜

k

2

+ m

2

and

K(ω, α, β) =

∞

0

ds

s

s

iω

e

−isα+iβ/s

. (17.8)

We used the fact that the function K obeys

K(ω, σα, β/σ) = σ

−iω

K(ω, α, β) ; (17.9)

it can be expressed in terms of the familiar Bessel and Hankel functions. Eq. (17.7)

is readily obtained by taking one of the plane wave solutions in Minkowski space-time,

k

3

= 0 , k

0

= µ, rewriting it in terms of the Rindler coordinates and τ , and then

Fourier transforming it with respect to τ . It is not diﬃcult to verify directly (using

partial integration in s ) that the partial diﬀerential equation (17.6) is obeyed.

We now normalize the Fourier components of a ﬁeld Φ(τ, , ˜ x) with respect to τ as

follows:

Φ(τ, , ˜ x) = A(τ, , ˜ x) + A

†

(τ, , ˜ x) , (17.10)

A(τ, , ˜ x) =

∞

−∞

dω

d

2

˜

k

2(2π)

4

K(ω,

1

2

µ,

1

2

µ) e

i

˜

k·˜ x−iωτ

a

2

(

˜

k, ω) , (17.11)

so that the operator a

2

is identiﬁed as

a

2

(

˜

k, ω) =

∞

−∞

dk

3

√

2πk

0

a(

k) e

iω ln

k

3

+k

0

µ

, (17.12)

where k

0

(

k) =

k

3

2

+ µ

2

. The inverse of this Fourier transform is

a(

k) =

∞

−∞

dω

√

2π k

0

a

2

(

k, ω) e

−iω ln

k

3

+k

0

µ

(17.13)

(remember that k

0

is a function of k

3

, and ∂k

0

/∂k

3

= k

3

/k

0

). Plugging Eq. (17.13) into

Eq. (17.3) gives us Eq. (17.11), if the variable s in (17.8) is identiﬁed with

k

0

−k

3

µ

=

µ

k

0

+k

3

.

From the commutation rules (17.5), we derive similar commutation rules for a

2

:

[a

2

(

˜

k, ω), a

†

2

(

˜

k

, ω

)] = δ

2

(

˜

k −

˜

k

)δ(ω −ω

) . (17.14)

However, before interpreting these as annihilation and creation operators, we must be

aware of the fact that the integration in Eq. (17.11) also goes over negative values for

ω . There, the operator a(

˜

k, ω) annihilates a negative amount of energy, so it really is a

creation operator. Therefore, we must rearrange the positive and negative ω contributions

when they are added in Eq. (17.10). To do this, it is convenient to note some properties

of the functions K .

First, one has

K

∗

(ω, α, β) = K(−ω, −α, −β) . (17.15)

35

Next, let α > 0 and β > 0 . In the deﬁnition (17.8), the integrand is bounded in the

region Im(s) ≤ 0 . Therefore, one may rotate the integration contour as follows:

s →s e

−iφ

, 0 ≤ φ ≤ π . (17.16)

Taking the case φ = π , we ﬁnd that

K(−ω, α, β) =

∞

0

ds

s

s

−iω

e

−πω

e

isα−iβ/s

= e

−πω

K

∗

(ω, α, β) if α > 0 , β > 0 ,(17.17)

and similarly one has

K(−ω, αβ) = e

+πω

K

∗

(ω, α, β) if α < 0 , β < 0 . (17.18)

This allows us to collect the positive and negative ω contributions in Eqs. (17.10) and

(17.11) as follows. In the region > 0 , we have

Φ(τ, , ˜ x) =

∞

0

dω e

−iωτ

d

2

˜

k e

i

˜

k·˜ x

2(2π)

4

K(ω,

1

2

µ,

1

2

µ)

a

2

(

˜

k, ω) + e

−πω

a

†

2

(−

˜

k, −ω)

+ H.c. (17.19)

In the opposite quadrant of Rindler space, where < 0 , we have

Φ(τ, , ˜ x) =

∞

0

dω e

−iωτ

d

2

˜

k e

i

˜

k·˜ x

2(2π)

4

K(ω,

1

2

µ,

1

2

µ)

a

2

(

˜

k, ω) + e

+πω

a

†

2

(−

˜

k, −ω)

+ H.c. (17.20)

At this point, it is opportune to deﬁne the new creation and annihilation operators

a

I

, a

†

I

, a

II

and a

†

II

, applying the following Bogolyubov transformation, when ω > 0 :

¸

¸

¸

a

I

(

˜

k, ω)

a

†

I

(−

˜

k, ω)

a

II

(

˜

k, ω)

a

†

II

(−

˜

k, ω)

=

1

√

1 −e

−2πω

¸

¸

¸

1 0 0 e

−πω

0 1 e

−πω

0

0 e

−πω

1 0

e

−πω

0 0 1

¸

¸

¸

a

2

(

˜

k, ω)

a

†

2

(−

˜

k, ω)

a

2

(

˜

k, −ω)

a

†

2

(−

˜

k, −ω)

(17.21)

Inverting this, we see that we have

¸

¸

¸

a

2

(

˜

k, ω)

a

†

2

(−

˜

k, ω)

a

2

(

˜

k, −ω)

a

†

2

(−

˜

k, −ω)

=

1

√

1 −e

−2πω

¸

¸

¸

1 0 0 −e

−πω

0 1 −e

−πω

0

0 −e

−πω

1 0

−e

−πω

0 0 1

¸

¸

¸

a

I

(

˜

k, ω)

a

†

I

(−

˜

k, ω)

a

II

(

˜

k, ω)

a

†

II

(−

˜

k, ω)

(17.22)

We then see that, at > 0 , the ﬁeld Φ depends only on a

I

and a

†

I

, while at < 0 ,

the ﬁeld Φ depends only on a

II

and a

†

II

. The normalization has again been chosen such

that

[a

I

(

˜

k, ω), a

†

I

(

˜

k

, ω

)] = [a

II

(

˜

k, ω), a

†

II

(

˜

k

, ω

)] = δ(ω −ω

)δ

2

(

˜

k −

˜

k

) ,

[a

I

, a

II

] = [a

I

, a

†

ii

] = 0 . (17.23)

36

The importance of this is the following. The operators a

I

and a

II

are deﬁned such that

they only annihilate objects with positive energy, and their hermitean conjugates only

create positive energies. In the quadrant > 0 , only the combination created by a

†

I

can

be detected using the ﬁeld F(τ, , ˜ x) , and in the quadrant < 0 only the other operators,

a

II

act. It is important to realize this, because if we had not paid attention to this, we

could have kept the original operators, deﬁning a

2

(

˜

k, ω) and a

†

2

(

˜

k, −ω) as annihilation

operators at ω > 0 , without the apparent need for a Bogolyubov transformation.

Just because the ﬁelds depend on time as e

−iωτ

, the Rindler space Hamiltonian, that

is, the operator that generates a boost in the Rindler time parameter τ , is

H =

∞

−∞

dω ω

d

2

˜

k a

†

2

(

˜

k, ω)a

2

(

˜

k, ω)

=

∞

0

dω ω

d

2

˜

k

a

†

I

(

˜

k, ω)a

I

(

˜

k, ω) −a

†

II

(

˜

k, ω)a

II

(

˜

k, ω)

= H

I

R

−H

II

R

. (17.24)

One may also verify that, if H

M

(x) is the Hamiltonian density in Minkowski space-time

at time t = 0 , then

H

I

R

=

>0

d

3

x H

M

(x) , H

II

R

=

<0

dx[[H

M

(x) . (17.25)

Consequently, all observables constructed out of the ﬁelds φ in the quadrant I where

> 0 commute with H

II

R

and the observables in quadrant II, < 0 commute with

H

I

R

.

The vacuum state as experienced by an observer in Rindler space is the ground state

of H

I

R

, and since [H

I

R

, H

II

R

] = 0 , we can also have the ground state of H

II

R

. We refer to

this state as [0, 0`

R

. It is called the Boulware vacuum.

However, the Boulware vacuum is not at all the lowest energy state in Minkowski

space-time. Let is take that state, [Ω`

M

, which is deﬁned as

a(

k)[Ω`

M

= 0 , a

2

(

˜

k, ω)[Ω`

M

= 0 . (17.26)

This describes empty space-time as experienced by an observer who is stationary in

Minkowski space, or freely falling in the gravitational ﬁeld of Rindler space-time. What

does [Ω`

M

look like to the observer who is at a ﬁxed position (, ˜ x) , with > 0 , in

Rindler space? For this observer, the operator a

I

describes the particles. We have

a

I

(

˜

k, ω)[Ω`

M

= e

−πω

a

†

II

(−

˜

k, ω)[Ω`

M

,

a

II

(

˜

k, ω)[Ω`

M

= e

−πω

a

†

I

(−

˜

k, ω)[Ω`

M

. (17.27)

These equations are easy to solve. We ﬁnd

[Ω`

M

=

¸

˜

k,ω

√

1 −e

−2πω

∞

¸

n=0

[n`

I

[n`

II

e

−πωn

, (17.28)

37

where the square root is added for normalization. Note that

H

R

[Ω`

M

= (H

I

R

−H

II

R

)[Ω`

M

= 0 , (17.29)

which conﬁrms that the Minkowski vacuum is Lorentz invariant; remember that H

R

is

the generator of Lorentz boosts.

Consider any observable O in the positive sector of Rindler space-time. It must

commute with H

II

R

, and therefore

O([ψ`

I

[ψ

`

II

) = [λ

`

II

(O[ψ`

I

) . (17.30)

Let us concentrate on only one sector of

˜

k and ω . There, the expectation value of such

an operator is

M

'Ω[O[Ω`

M

= (1 −e

−2πω

)

¸

n

1

,n

2

II

'n

1

[

I

'n

1

[O[n

2

`

I

[n

2

`

II

e

−πω(n

1

+n

2

)

=

¸

n≥0

I

'n[O[n`

I

e

−2πnω

(1 −e

−2πω

) = Tr (O

Ω

) , (17.31)

where

Ω

is the density matrix C e

−βH

I

corresponding to a thermal state ate the tem-

perature

T = 1/(k β) = 1/(2πk) , (17.32)

where k is Boltzmann’s constant: β = 1/(kT) . Note that this temperature is dimension-

less. This is because the time unit, τ , in the Rindler coordinates (15.2), is dimensionless.

In Section 15, we saw that, at the distance =

0

from the Rindler horizon, the strength

of the gravitational ﬁeld is

1

0

, and furthermore that time has to be rescaled by a factor

0

. Therefore, we conclude that an observer who is being accelerated by a gravita-

tional ﬁeld with strength g in relativistic units, experiences radiation with a temperature

T = g/(2πk) . This is the Unruh eﬀect.

18. Hawking radiation

The region of space-time in the vicinity of the horizon of a black hole, approximately

takes the form of Rindler space, that is, the Schwarzschild time coordinate t relates to

Rindler time there as in Eq. (15.5): t = 4Mτ . Therefore, the temperature experienced

there is given by

kT

H

= 1/8πM = 1/8πGm

BH

. (18.1)

This is the Hawking temperature of a black hole.

The value for this temperature could have been derived more intuitively as follows.

The free energy F of any thermal quantum system is computed as

e

−βF

= Tr (e

−βH

) , (18.2)

38

where β = 1/kT , and k is the Boltzmann constant. From this expression for F , one can

compute the thermal average of any operator O of a system. Assume a small disturbance:

e

−βF(J)

= Tr (e

−β(H+JO)

) , (18.3)

then we have, keeping β ﬁxed:

'O`

T

=

¸

E

'E[O[E`e

−βE

¸

E

e

−βE

=

−

∂

β∂J

Tr e

−βH−βJO

[

J=0

Tr e

−βH

=

−

∂

β∂J

log(Tr (e

−βH−βJO

))[

J=0

=

∂F

∂J

[

J=0

. (18.4)

The operator e

−βH

happens to be the evolution operator e

−iHP

for a time period P =

−iβ , and taking the trace means that the evolution operator is connected to itself after

this period in imaginary time. So, this essentially means that quantum mechanics over

a space-time that is periodic in imaginary time is equivalent to working out thermal

expectation values of operators at a temperature T equal to

P = β = /(kT) , (18.5)

where we re-inserted the constant .

The Unruh temperature g/(2πk) is thus connected to the fact that it refers to a

Rindler space that has periodicity 2π/g , and the Hawking temperature 1/(8πkM) follows

from the periodicity (16.9) derived for the Kruskal spacetime in Section 16.

We can now also understand why the temperature, or equivalently, the surface gravity

(14.11), cannot depend on the position along the horizon: if a solution is periodic with

period P at one spot, it cannot have any diﬀerent periodicity elsewhere, since the space-

time must still have the same analytic form after any number of periods in this particular

Euclidean direction.

For the general Kerr-Newman solution, the metric near the horizon approaches

Eq. (14.12). Writing

r −r

+

=

2

, dr = 2 d , (18.6)

we ﬁnd

ds

2

→

4Y

r

+

−r

−

−κ

2

2

dτ

2

+ d

2

,

κ =

r

+

−r

−

2(r

2

+

+ a

2

)

. (18.7)

So, rather than the “surface gravity”, we should view κ as the parameter that scales the

time variable in Rindler space-time at the horizon. Therefore, the Hawking temperature

of a Kerr Newman black hole is

kT = κ/2π =

r

+

−r

−

2(r

2

+

+ a

2

)

=

M

2

−a

2

−Q

2

/4π

2M

2

−Q

2

/4π + 2M

M

2

−a

2

−Q

2

/4π

. (18.8)

39

Clearly then, the parameter that looked like a temperature when we phrased the “four

laws of black hole dynamics”, really is a temperature! It is actually 4 times the parameter

τ that was introduced in section 14. Scaling everything else there accordingly, we ﬁnd

that the actual entropy S , as it occurs in the equation dU = TdS+ is 4 times smaller:

The temperature of a black hole equals kT = 4κ, where κ is the “surface

gravity”. The entropy of a black hole equals

S =

1

4

kΣ , (18.9)

where Σ is the area of the horizon.

Note that we kept Boltzmann’s constant k in our descriptions of temperature and entropy.

Units for the temperature could be chosen such that it is one.

The fact that there are particles with a certain temperature near the horizon of a black

hole, means that some of these thermally excited particles can escape to inﬁnity, and be

observed there. Indeed, with its temperature T = 1/(8πkM) , there will be radiation

emerging from the horizon. The intensity of the radiation will be proportional with T

4

close to the horizon, and the total energy loss per unit of time due to this radiation will

be approximately

U = C

1

ΣT

4

= C

2

M

2

M

−4

= C

2

M

−2

, (18.10)

where C

1

and C

2

are constants depending not only on the geometric details of the black

hole (what will be its apparent surface area as seen from inﬁnity?), but also, weakly, on

temperature, because the number of particle types participating in the radiation depends

on whether the temperature is suﬃciently high to excite particles with given rest masses.

Let us nevertheless take Eq. (18.10) as a rough approximation. Assuming conservation

of total mass/energy (General Relativity would be inconsistent if we did not), we must

conclude that, if left by itself, a black hole should loose mass:

dM

dt

≈ −C

2

M

−2

→

dM

3

dt

≈ −3C

2

;

M(t) ≈ (3C

2

)

1

3

(t

0

−t)

1

3

. (18.11)

Clearly, at some moment t = t

0

the black hole must disappear altogether. What exactly

happens at that moment, however, cannot be understood without a more complete under-

standing of quantum gravity than we possess today. We do expect this to be quite a bang,

because the total mass-energy emitted in the last second, turns out to be formidable, once

we put our conventional units back in.

19. The implication of black holes for a quantum theory of grav-

ity

In thermodynamics, the entropy S of a system with no other adjustable parameters obeys

TdS = dU , (19.1)

40

where U is the energy stored as heat. The quantity F , deﬁned as

F = U −TS , (19.2)

is called the Helmholtz free energy, and it obeys

dF = −SdT , or S = −

∂F

∂T

. (19.3)

In statistical physics, the free energy F has been identiﬁed by the equation

e

−βF

=

¸

E

e

−βE

= Tr (e

−βH

) , (19.4)

where β = 1/(kT) , while k is Boltzmann’s constant and H is the quantum Hamiltonian.

The sum is over all quantum states [E` . The quantity e

−βE

is the Boltzmann factor

describing the probability of any state [E` to occur when there is thermal equilibrium.

The total energy U is then given by

U =

¸

E

E e

−βE

¸

E

e

−βE

=

Tr (H e

−βH

)

Tr (e

−βH

)

=

−

∂

∂β

e

−βF(β)

e

−βF

=

∂

∂β

(βF) . (19.5)

Eq. (19.2) can be written as

S = kβ(U −F) , (19.6)

and we derive

S/k =

Tr (βHe

−βH

)

Tr (e

−βH

)

+ log Tr (e

−βH

) = log Tr (e

βH−βH

) , (19.7)

where U has been written as an average: U = 'H` .

In these expressions, it was assumed that our system is a grand canonical ensemble. We

also can consider micro canonical ensembles, which may be a collection of many systems

but always in such a way that the total energy U is kept ﬁxed. In that case, the sum is

only over all states with the same energy E = U . Then, the exponent in Eq. (19.7) is 1,

and the entropy is then seen to be

S = k log(Tr (1)) , (19.8)

or, the entropy is nothing but the logarithm of the total number of states over which

we sum. This is a quite general result: In a quantum system, the entropy is k times

the logarithm of the total number of quantum states that can describe the system we are

looking at.

For a black hole, this is a fundamental feature. Since here, we conclude from Eq. (18.9)

that the total number of “black hole microstates” is given by

= C e

Σ/4

. (19.9)

41

C is an unknown constant. This is because entropy is always deﬁned apart from

an unknown additive constant. In Eq. (19.9), this is a multiplicative constant, which is

unknown.

Much research is going into identifying these quantum states. Can we write a Schr¨od-

inger equation for black holes? This question is compounded by the fact that the pure

Minkowski vacuum state, when transformed into Rindler coordinates, emerges as a density

matrix, e

−βH

, which is a mixture of quantum states. Does a collapsing system smoothly

transmute from a pure quantum state into a mixed state? This we do not believe. To

understand this situation better, we must study the collapsing system in diﬀerent coordi-

nate frames, and include the consideration that the metric of space and time itself must

be subject to quantum oscillations. This is beyond the scope of this lecture course.

The application of thermodynamics to black holes could be criticized for the following

reason: let us try to calculate the ‘speciﬁc heat’ of a black hole. What is dU/dT , or, the

amount of heat needed to raise the temperature by one degree? The temperature is given

by Eq. (18.1), so

U = 1/(8πkT) . (19.10)

Therefore,

dU

dT

= −1/(8πkT

2

) < 0 , (19.11)

so the temperature goes down when heat is added. This means that the black hole is

fundamentally unstable thermally.

But there is another way to derive that the number of black hole microstates is the

exponent of

1

4

the area Σ of the horizon. Consider a quantum mechanical description

of the process of capturing something. The cross section for capture in a Schwarzschild

black hole can roughly be estimated to be

7

σ(

k) = πr

2

+

= 4πM

2

, (19.12)

where

k is the momentum of the ingoing particle. Now we also know the probability W

for emitting a particle, which is given by the thermal probability:

Wdt =

σ(

k)v

V

e

−β

H

E

dt , (19.13)

where β

H

is the inverse Hawking temperature:

β

H

= 1/kT

H

= 8πM . (19.14)

and V is the volume of the space where the particle is released.

7

The actual value will be considerably larger, and momentum dependent, because the orbits cannot

be straight lines, but in the present argument only the order of magnitude is of signiﬁcance.

42

Now we assume that the process is also coverned by a Schr¨odinger equation. This

means that there exist quantum mechanical transition amplitudes,

T

in

=

BH

'M + GE[ [M`

BH

[E`

in

, (19.15)

and T

out

=

BH

'M['E[ [M + GE`

BH

, (19.16)

where the states [M`

BH

represent black hole states with mass M/G, and the states [E`

are states of surrounding particles with total energy E , conﬁned to a volume V . In

terms of these amplitudes, using the so-called Fermi Golden Rule, the cross section σ

and the emission probabilities W can be written as

σ = [T

in

[

2

(M +GE)/v , (19.17)

W = [T

out

[

2

(M)

1

V

, (19.18)

where (M) stands for the presumed density of quantum levels of a black hole with

mass M . The factor v

−1

in Eq. (19.17) is a kinematical factor, and the factor V

−1

in

Eq. (19.18) arises from the normalization of the wave functions.

Time reversal invariance would relate T

in

to T

out

. To be precise, all we need is CPT

invariance, since a parity transformation P and a charge conjugation C have no eﬀect on

our calculation of σ . Dividing the expressions (19.17) and (19.18), and using Eq, (19.13),

one ﬁnds:

(M + GE)

(M)

= e

βE

= e

8πM

. (19.19)

This is easy to integrate:

d log (M)

dM

= 8πM/G , (19.20)

(M) = C e

4πM

2

/G

= e

S/k

. (19.21)

Thus, we found a direct expression for the density of quantum levels, which was now

deﬁned to be the logarithm of an entropy. It coincides with the thermodynamic expression

(19.9).

Clearly, this analysis suggests that black holes obey a Schr¨ odinger equation describing

the evolution of internal quantum states, and we can estimate rather precisely the dimen-

sionality of this internal Hilbert space. It is as if there is one Boolean degree of freedom

per unit of area A

0

of the horizon:

= 2

A/A

0

, A

0

= 4Glog 2 . (19.22)

But how can we understand the details of this Schr¨odinger equation? Curiously, the

answer to this question does not appear to follow from any of the ﬁrst principles that

have been discussed so-far. To the contrary, there seems to be a contradiction. According

to Hawking’s derivation of the radiation process, any black hole, regardless its past, ends

43

up as a thermodynamically mixed state. Would this also hold for a black hole that started

out as a collapsing star in a quantum mechanically pure state? Can pure states evolve

into mixed states? Not according to conventional quantum mechanics.

From a physical point of view, the distinction between pure states and mixed states

for macroscopic objects is pointless. Black holes should be regarded as being macroscopic.

So, it is very likely that what we perceive as a mixed state is actually a pure state whose

details we were unable to resolve. However, if that is true, the derivation given by Hawking

is wanting. We should search for a more precise analysis.

In fact, approximations and simpliﬁcations were made in Hawking’s derivation. In

particular, in and outgoing particles were assumed not to interact with one another.

Usually, this is a reasonable assumption. However, in this case, it is easy to observe that

particles from the collapse entering the black hole at early times and Hawking particles

leaving the black hole at late times, meet each other very close to the horizon. The

center-of-mass energy that this encounter represents diverges exponentially with the time

lapse, so it can easily surpass the mass-energy of the entire universe. Such passings cannot

go without mutual interactions; they would merge to form gigantic black holes, but long

before that happens, our analysis has become invalid. This is where our procedures should

be repaired. This, however, is diﬃcult and research is in progress.

In the mean time, there have been other developments, notably in string theory. Ac-

cording to string theory, D-brane conﬁgurations form soliton-like conﬁgurations that play

the role of black holes. For these black holes, the microstates can be counted, provided

that they are not too far separated from the extreme limit. The counting appears to

conﬁrm the result (19.21).

In a uniﬁed theory of all particles and forces, the primary building blocks are the

heaviest and most compact forms of matter. We see that such forms of matter are black

holes. There cannot be “other” primary forms of matter, since all massive objects must

be surrounded by gravitational ﬁelds, i.e., they are black holes. The properties of these

black holes, in turn, must be determined by ﬁeld theories describing particles at their

horizons. So, the question of unifying all forces and matter forms ends up in a logical

spiral. This makes the problem interesting and challenging from a theoretical point of

view. A more precise and coherent theoretical approach might lead to further insights.

20. The Aechelburg-Sexl metric

We wish to ﬁnd the space-time metric surrounding a particle that goes almost with the

speed of light towards the positive z -direction. Consider the schwarzschild metric in the

case of a very light mass m:

ds

2

= dx

2

+

2µ

r

(dt

2

+ dr

2

) , (20.1)

where µ = Gm, and dx

2

is the ﬂat metric d˜ x

2

−dt

2

. This, we rewrite as

ds

2

= dx

2

+

2µ

r

(u dx)

2

+

2µ

r

dr

2

, r =

x

2

+ (u x)

2

, (20.2)

44

where

u = (1, 0, 0, 0) ; u

2

= g

µν

u

µ

u

ν

= −1 . (20.3)

In these expressions, we neglected all eﬀects that are of higher order in the particle’s mass

µ, since µ is chosen to be small.

Written this way, we can now easily give this particle a Lorentz boost. In the boosted

frame we can take

mu

µ

= p

µ

→(p, 0, 0, p) , Gp =

µv

1 −v

2

/c

2

µ . (20.4)

In the limit µ →0 , p ﬁxed, one has r →[x u[ .

It will turn out to be useful to compare this metric with the ﬂat space-time metric in

two coordinate frames y

µ

±

, deﬁned as

y

µ

(±)

= x

µ

±2µu

µ

log r . (20.5)

We have

dy

2

(±)

= dx

2

±

4µ

r

(u dx)dr −4µ

2

dr

2

r

2

; (20.6)

ds

2

−dy

2

(±)

=

2µ

r

d

r ∓(u x)

2

+ 4µ

2

(d log r)

2

. (20.7)

Now consider the limit (20.4). We keep p ﬁxed but let µ tend to zero. We now claim

that when (p x) > 0 , the metric ds

2

approaches the ﬂat metric dy

2

(+)

, whereas when

(p x) < 0 , we have ds

2

→dy

2

(−)

, and at the plane deﬁned by (p x) = 0 these two ﬂat

space-times are glued together according to

y

µ

(+)

= y

µ

(−)

+ 4p

µ

log [˜ x[ , (20.8)

where ˜ x = (0, x, y, 0) are the transverse part of the coordinates y

µ

.

This is seen as follows. First, we note that the last term of Eq. (20.7) can be ignored.

Next, given a small positive number λ, we divide space-time in three regions:

A) (u x) > λ ;

B) (u x) < −λ ,

C) [(u x)[ ≤ λ .

(20.9)

In region (A) , we use

r −(u x) =

x

2

r + (u x)

, (20.10)

which is therefore bounded by

x

2

λ

. Thus, the ﬁrst term in Eq. (20.7) for y

(+)

is bounded

by

µ

λ

2

times a coordinate dependent function (note that r ≥ [˜ x[ ). Similarly, in region

(B) , Eq. (20.7) for y

(−)

will tend to zero as µ/λ

2

. In the region (C) , we have that r

45

and (u x) are both bounded by terms that are ﬁnite or proportional to λ. So, in (C) ,

both equations (20.7) are bounded by functions of the form µ or µλ

2

. Choosing λ such

that, as µ →0 , both µλ

2

→0 and µ/λ

2

→0 , allows us to conclude that

A) ds

2

→dy

2

(+)

if (p x) ≥ 0 ,

B) ds

2

→dy

2

(−)

if (p x) ≤ 0 ,

C) y

(+)

= y

(−)

+ 4µu

µ

log r at (p x) ≈ 0 ,

(20.11)

which is equivalent to Eq. (20.8). This deﬁnes the Aechelburg-Sexl metric.

Deﬁning x

±

= z ± t , one ﬁnds for a source particle moving with the speed of light

to the positive z -direction that two ﬂat space-times, one with coordinates (x

±

(+)

, ˜ x

(+)

)

and one with coordinates (x

±

(−)

, ˜ x

(−)

) are connected together at the point x

−

(+)

= x

−

(−)

,

in such a way that ˜ x

(+)

= ˜ x

(−)

and

x

+

(+)

−x

+

(−)

= 4Gp

+

log [˜ x[ = 8Gp log [˜ x[ . (20.12)

The r.h.s. of this equation happens to be a Green function,

δx

+

= −pf(˜ x) ,

˜

∂

2

f(˜ x) = −16πGδ

2

(˜ x) . (20.13)

This result can be generalized to describe a light particle falling into the horizon of

a black hole. For the Schwarzschild observer, its energy is taken to be so small that its

gravitational ﬁeld appears to be negligible, and the black hole mass will hardly be aﬀected

by the energy added to it. However, in Kruskal coordinate space, see Eqs. (6.3)–(6.7), the

energy is seen to grow exponentially as Schwarzschild time t progresses. Let us therefore

choose the Kruskal coordinate frame such that the particle came in at large negative time

t . This means that in Eq. (6.3), x ≈ 0 , or, the particle moves in along the past horizon.

In view of the result derived above, one can guess in which way the particle that goes in

will deform the metric: we cut Kruskal space in halves across the x -axis, and glue the

pieces together, again after a shift, deﬁned by

y

(+)

= y

(−)

−16πGp

x

F(y

(−)

, θ, ϕ) , (20.14)

where (±) now refers to the regions x > 0 and x < 0 . This corresponds to a metric

with a delta-distributed Riemann curvature on the plane x = 0 . The function F is yet

to be determined.

By demanding that the Ricci curvature must still vanish at the seam, one can compute

the equations for F . It is then found that F has to obey

−

˜

∂

2

F + F =

˜

δ

2

(Ω) , (20.15)

where

˜

∂

2

is the spherical Laplacian ( + 1) and

˜

δ

2

(Ω) the Dirac delta function on the

sphere (θ, ϕ) . The quantity p

x

is the momentum of the particle falling in, with respect

to the Kruskal coordinate frame. This equation, which clearly shows a strong similarity

with the case derived earlier for Rindler space, can be solved in an integral form. It turns

out not to depend on y

(−)

itself. At small angular distances θ , one gets

F(θ) →(1/2π) log(1/θ) . (20.16)

46

It is important to note that this shift in the Kruskal y coordinate aﬀects the Hawking

radiation. It does not aﬀect its thermal nature, not the temperature itself, but it will

aﬀect the microstates. This may be an important starting point for further investigations

of the quantum structure of a black hole.

21. History

A brief history of black holes in General Relativity:

8

• 1915: Einstein formulates the general theory of relativity.

• 1916: Karl Schwarzschild publishes his exact spherically symmetric and static so-

lution, showing a singularity at r = 2M .

• 1924: Eddington introduces coordinates that are well behaved at r = 2M .

• 1930: Using general relativity, Subrahmanyan Chandrasekhar calculates that a

non-rotating body of electron-degenerate matter above 1.44 solar masses (the Chan-

drasekhar limit) would collapse.

• 1933: LeMaˆıtre realizes the signiﬁcance of Eddington’s result: r = 2M is a ﬁcti-

tious singularity.

• 1958: David Finkelstein introduces the concept of the event horizon by presenting

Eddington-Finkelstein coordinates, which enabled him to show that ”The Schwarz-

schild surface R = 2M is not a singularity, but that it acts as a perfect unidirectional

membrane: causal inﬂuences can cross it in only one direction”. All theories up to

this point, including Finkelstein’s, covered only non-rotating black holes.

• 1960: Kruskal and Szekeres obtain the maximal extension of the Schwarzschild

solution.

• 1960: Penrose introduces global methods in the study of General Relativity.

• 1963: Roy Kerr ﬁnds a generalization of the Schwarzschild metric and interprets it

as the ﬁeld of a “spinning particle”.

• 1967: John Wheeler uses the words “black hole” in a public lecture. Unoﬃcially,

the phrase has been used earlier by others.

Black hole uniqueness theorems make people believe that black holes cannot form,

because time reversal invariance of Nature’s laws would then imply that only per-

fectly symmetric initial states could collapse gravitationally. Roger Penrose saw the

ﬂaw of that argument: there may be perturbations in the black hole metric in the

form of multipole components, but they all die out or radiate away exponentially.

8

I made use here of notes made by A. Ashtekar

47

• late 1960’s - early 1970’s: Bekenstein, Bardeen, Carter, Penrose and Hawking

explore the structure and properties of black holes. Bekenstein proposes that black

holes should carry entropy, proportional to the horizon area. Bardeen, Carter and

Hawking prove the ﬁrst theorems on black hole mechanics.

• 1974: Hawking discovers black hole evaporation. Quantum ﬁelds on a black hole

background space-time radiate thermal (i.e. black body) spectrum of particles, with

a temperature of kT = κ/2π .

• 1982: Bunting and Mazur independently derive a generalized uniqueness theorem:

any isolated, time-independent black hole in general relativity is described by the

Kerr metric. hence the equilibrium state of every (uncharged) black hole is fully

described by only two parameters: mass and angular momentum (represented by

M and J ).

• 1995: Strominger, Vafa, Maldacena and others discover how to describe the black

hole microstates in terms of D-branes in string theory. The description is particu-

larly detailed at or near the extreme limit, and usually the black hole is considered

in more than 4 dimensional space-time.

48

Contents

1 Introduction 2 The Metric of Space and Time 3 Curved coordinates 4 A short introduction to General Relativity 5 Gravity 6 The Schwarzschild Solution 7 The Chandrasekhar Limit 8 Gravitational Collapse 9 The Reissner-Nordstr¨m Solution o 10 Horizons 11 The Kerr and Kerr-Newman Solution 12 Penrose diagrams 13 Trapped Surfaces 14 The four laws of black hole dynamics 15 Rindler space-time 16 Euclidean gravity 17 The Unruh eﬀect 18 Hawking radiation 19 The implication of black holes for a quantum theory of gravity 20 The Aechelburg-Sexl metric 2 4 5 6 9 10 13 14 18 20 22 23 25 29 31 32 34 38 40 44

1

21 History

47

1.

Introduction

According to Newton’s theory of gravity, the escape velocity v from a distance r from the center of gravity of a heavy object with mass m , is described by

1 2 v 2

=

Gm . r

(1.1)

What happens if a body with a large mass m is compressed so much that the escape velocity from its surface would exceed that of light, or, v > c ? Are there bodies with a mass m and radius R such that 2G m ≥1? R c2 (1.2)

This question was asked as early as 1783 by John Mitchell. The situation was investigated further by Pierre Simon de Laplace in 1796. Do rays of light fall back towards the surface of such an object? One would expect that even light cannot escape to inﬁnity. Later, it was suspected that, due to the wave nature of light, it might be able to escape anyway. Now, we know that such simple considerations are misleading. To understand what happens with such extremely heavy objects, one has to consider Einstein’s theory of relativity, both Special Relativity and General Relativity, the theory that describes the gravitational ﬁeld when velocities are generated comparable to that of light. Soon after Albert Einstein formulated this beautiful theory, it was realized that his equations have solutions in closed form. One naturally ﬁrst tries to ﬁnd solutions with maximal symmetry, being the radially symmetric case. Much later, also more general solutions, having less symmetry, were discovered. These solutions, however, showed some features that, at ﬁrst, were diﬃcult to comprehend. There appeared to be singularities that could not possibly be accepted as physical realities, until it was realized that at least some of these singularities were due only to appearances. Upon closer examination, it was discovered what their true physical nature is. It turned out that, at least in principle, a space traveller could go all the way in such a “thing” but never return. Indeed, also light would not emerge out of the central region of these solutions. It was John Archibald Wheeler who dubbed these strange objects “black holes”. Einstein was not pleased. Like many at ﬁrst, he believed that these peculiar features were due to bad, or at least incomplete, physical understanding. Surely, he thought, those crazy black holes would go away. Today, however, his equations are much better understood. We not only accept the existence of black holes, we also understand how they can actually form under various circumstances. Theory allows us to calculate the 2

behavior of material particles, ﬁelds or other substances near or inside a black hole. What is more, astronomers have now identiﬁed numerous objects in the heavens that completely match the detailed descriptions theoreticians have derived. These objects cannot be interpreted as anything else but black holes. The “astronomical black holes” exhibit no clash whatsoever with other physical laws. Indeed, they have become rich sources of knowledge about physical phenomena under extreme conditions. General Relativity itself can also now be examined up to great accuracies. Astronomers found that black holes can only form from normal stellar objects if these represent a minimal amount of mass, being several times the mass of the Sun. For low mass black holes, no credible formation process is known, and indeed no indications have been found that black holes much lighter than this “Chandrasekhar limit” exist anywhere in the Universe. Does this mean that much lighter black holes cannot exist? It is here that one could wonder about all those fundamental assumptions that underly the theory of quantum mechanics, which is the basic framework on which all atomic and sub-atomic processes known appear to be based. Quantum mechanics relies on the assumption that every physically allowed conﬁguration must be included as taking part in a quantum process. Failure to take these into account would necessarily lead to inconsistent results. Mini black holes are certainly physically allowed, even if we do not know how they can be formed in practice. They can be formed in principle. Therefore, theoretical physicists have sought for ways to describe these, and in particular they attempted to include them in the general picture of the quantum mechanical interactions that occur in the sub-atomic world. This turned out not to be easy at all. A remarkable piece of insight was obtained by Stephen Hawking, who did an elementary mental exercise: how should one describe relativistic quantized ﬁelds in the vicinity of a black hole? His conclusion was astonishing. He found that the distinction between particles and antiparticles goes awry. Diﬀerent observers will observe particles in diﬀerent ways. The only way one could reconcile this with common sense was to accept the conclusion that black holes actually do emit particles, as soon as their Compton wavelengths approach the dimensions of the black hole itself. This so-called “Hawking radiation” would be a property that all black holes have in common, though for the astronomical black holes it would be far too weak to be observed directly. The radiation is purely thermal. The Hawking temperature of a black hole is such that the Wien wave length corresponds to the radius of the black hole itself. We assume basic knowledge of Special Relativity, assuming c = 1 for our unit system nearly everywhere, and in particular in the last parts of these notes also Quantum Mechanics and a basic understanding at an elementary level of Relativistic Quantum Field Theory are assumed. It was my intention not to assume that students have detailed knowledge of General Relativity, and most of these lectures should be understandable without knowing too much General Relativity. However, when it comes to discussing curved coordinates, Section 3, I do need all basic ingredients of that theory, so it is strongly advised to familiarize oneself with its basic concepts. The student is advised to consult my lecture notes “Introduction to General Relativity”, http://www.phys.uu.nl/ thooft/lectures/genrel.pdf 3

···. does not have to be indicated explicitly anymore: xµ = aµν xν and s 2 = gµν xµ xν . · · · .1) provided that the matrix A is such that a special quantity remains invariant: −c2 t 2 + x 2 + y 2 + z 2 = −c2 t2 + x2 + y 2 + z 2 . which we write as (x1 . gµν µ.5) . x2 .2) will not be invariant. so that the summation sign.whenever something appears to become incomprehensible. 1) . summation convention has been implied twice. In the latter expression.3) occurs exactly once as a superscript and once as a subscript. The Metric of Space and Time Points in three-dimensional space are denoted by a triplet of coordinates. which we also write as: gµν xµ xν is invariant. as follows: gµν = (A−1 )αµ (A−1 )βν gαβ . and the time at which an event takes place is indicated by a fourth coordinate t = x0 /c . this index will be summed over the values 0. 1 2 3 z a3 a3 a3 a3 z 0 1 2 3 or xµ = ν=0. there are numerous other texts on General Relativity. z) .ν=0. In what follows.···. In that case.3) A matrix A with this property is called a Lorentz transformation. 3 . note that there are all sorts of variations in notation used. The invariance is Lorentz invariance. where c is the speed of light. 4 (2. More general linear transformations will turn out to be useful as well.2) 0 0 . . we simply have to replace gµν by an other quantity. The special Lorentz transformations form a group called SO(3. x = (x. The theory of Special Relativity is based on the assumption that all laws of Nature are invariant under a special set of transformations of space and time: 0 t a 0 a0 a0 a0 t 1 2 3 x a1 a1 a1 a1 x 0 1 2 3 = 2 y a 0 a2 a2 a2 y . but then (2. 2.1) and (2. summation convention will be used: in every term of an equation where an index such as the index ν in Eqs. (2. (2. 0 1 −1 0 = 0 0 0 1 0 0 0 0 1 0 (2. y. (2. x3 ) . 0 det(A) = +1 .3 (2. Of course.3 aµν xν . Usually. or x = Ax . we also demand that a0 > 0 .4) in which case we speak of special Lorentz transformations.

3) .5). in a vector or tensor. dτ v µ v µ = −1 . the original coordinates (t. diﬀerentiable functions of four quantities u = {uµ . which we can describe as dxµ (τ ) = v µ = constant. Thus. Under a coordinate transformation. z) are completely arbitrary.6) is assumed to be positive (when the vector is spacelike). (2. Curved coordinates The coordinates used in the previous section are such that they can be used directly to measure. (2. at a given point x in space and time.5) for the tensor gµν . Nature is invariant under general linear transformations provided that we use the transformation rule (2. Usually. We will call them Cartesian coordinates. It is then called the invariant length of a Lorentz vector xµ . This tensor will then be more general than (2.3). There. it is multiplied by the metric tensor or its inverse. replacing a subscript index by a superscript index means that. Cartesian coordinates. ∂uµ ∂uν (3. dτ 2 (3. but now these coeﬃcients are also coodinate dependent: gµν (u) = ∂xα ∂xβ gαβ (x) . distances and time spans.7). More precisely. one has to distinguish co-vectors xµ from contravectors xµ . µ = 0. i times a positive number (when the vector is timelike). 3.2) In the original. (2. In the general coordinate frame. xµ = g µν xν . (3. separated from x by only an inﬁnitesimal distance dx .so that the expression s2 = gµν xµ xν = gµν xµ xν (2.7) where g µν is the inverse of the metric tensor matrix gµν . consider points x + dx . tacitly. the formalism described in the previous section applies. or deﬁne. they are related by xµ = gµν xν . Being diﬀerentiable here means that every point is surrounded by a small region where these functions are to a good approximation linear. as in Eqs. (2. The quantity s deﬁned by Eq. Now consider just any coordinate frame. in general mutually independent. It is called the metric tensor. a particle on which no force acts. they are denoted by the same symbol. y. 5 d2 xµ (τ ) =0. Then we deﬁne ds by ds2 = gµν dxµ dxν = gµν (u) duµ duν .1) The prime was written to remind us that gµν in the u coordinates is a diﬀerent function than in the x coordinates. · · · . or zero (when xµ is lightlike). 3} . that is. x. will go along a straight line. gµν transforms as Eq. but in later sections this will be obvious and we omit the prime.6) remains obviously valid.

dτ ∂uλ dt dτ 2 ∂uλ ∂uκ dτ dτ ∂u dτ 2 Therefore.8) 4. A short introduction to General Relativity • A scalar function φ(x) of some arbitrary curved set of coordinates xµ . ∂xα ∂uκ (3. Suppose that xµ are arbitrary diﬀerentiable functions of coordinates uλ .4) (3. In terms of curved coordinates uµ (x) . ia a function that keeps the same values upon any coordinate transformation. a coordinate transformation xµ → uλ implies that φ(x) = φ (u(x)) . where τ is the eigen time of the particle. 6 .7) Γµ is called the connection ﬁeld. Note that it is symmetric under interchange of its two κλ subscript indices: Γµ = Γµ . where φ (u) is the same scalar function. Then dxµ ∂xµ duλ d 2 xµ ∂ 2 xµ duκ duλ ∂xµ d2 uλ = . ∂xα ∂uκ ∂uλ Here. but written in terms of the new coordinates uλ . = + λ . y } . Thus.3) is then replaced by an equation of the form duµ duν d2 uµ (τ ) duκ duλ = −1 . y} to an other. κλ dτ dτ dτ 2 dτ dτ where the function Γµ (u) is given by κλ gµν (u) ∂uµ ∂ 2 xα . (3. κλ λκ (3. Usually. this no longer holds.6) (3.y′ y x′ x Figure 1: A transition from one coordinate frame {x. + Γµ (u) =0.5) (3. eq. curved coordinate frame {x . it was used that partial derivatives are invertible: Γµ (u) = κλ ∂uµ ∂xα α = δκ . we will omit the prime.

∂xα ∂uκ (4. Γν is the connection ﬁeld that we introduced in Eq. transforms just as the gradient of a scalar function φ(x) . (3.1) This ensures that the product Aα (x)B α (x) transforms as a scalar: Aα (x)B α (x) = where Eq.7) was used. this vectorial function transforms as ∂uλ Aα (x) = Aλ (u) .• A co-vector is any vectorial function Aα (x) of the curved coordinates xµ that. Thus. there. To obtain a quantity that does transform as a true tensor. µλ ∂xµ ∂B κ (x) + Γκ (x)B ν (x) . µλ however. B2 2 . or B µ (x) = ∂xµ λ B (u) . ∂uλ (4. (3. µν ∂xµ (4. In that case.3) The gradient of a vector or tensor.2) (4. · · · .6) µν α In these expressions. • A contra-vector B µ (x) transforms with the inverse of that matrix. 2 • A tensor Aα1β12β······ (x) is a function that transforms just as the product of covectors α β β A1 1 . A2 2 . one must replace the gradient ∂/∂xµ by the so-called the covariant derivative Dµ . Now. (4. It goes as follows. ∂uλ ∂xα Aλ (u)B κ (u) = Aλ (u)B λ (u) . which for covectors is deﬁned as Dµ Aλ (x) = for contravectors: Dµ B κ (x) = and for tensors: 2 Dµ Aα1β12β······ (x) = α ∂Aλ (x) − Γν (x)Aν (x) . upon a curved coordinate transformation. we use the metric tensor gµν (x) to deﬁne Γν . and covectors B1 1 .6). upon a coordinate transformation. in general.4) (4.5) ∂ β A β1 β2 ··· (x) − Γν 1 (x)Aναβ1···2 ··· (x) − Γν 2 (x)Aα1β1 β2 ··· (x) − · · · µα µα ν··· 2 ∂xµ α1 α2 ··· ··· + Γβ1 (x)Aα1νβ22··· (x) + · · · . Superscript indices always refer to the α α contravector transformation rule and subscript indices to the covector transformation rule. · · · . does not transform as a vector or tensor. ∂xα . we assumed a ﬂat coordinate frame to exist. If we had a ﬂat µλ 7 . this might not be so.

that is. the index σ is summed over. It is important to note that the connection ﬁeld Γα itself does not transform as a αβ tensor.10) This deﬁnition implies that Dµ gαβ = 0 automatically.13) . It is a tensor with four indices. This is the so-called Riemann curvature. κλ 2 (4. a coordinate frame in terms of which gµν (x) = gµν everywhere in V . deﬁned as follows: Rµ = ∂α Γµ − ∂β Γµ + Γµ Γσ − Γµ Γσ . and using the fact that Γ is symmetric in its last two indices. there does exist a quantity that is constructed out of the connection ﬁeld that does transform as a tensor. this contraction must go with the inverse metric tensor: R = g µν Rµν . ∂xµ (4. as an easy calculation shows. Note that it is always symmetric in its two subscript indices: Γµ = Γν . λκ κλ (4. we derive Γµ = 1 g µα (∂κ gαλ + ∂λ gακ − ∂α gκλ ) . κα ασ κβ καβ κβ βσ κα (4. Eq. as dictated by the summation convention. so that its gradient vanishes. and that the covariant derivatives of all vectors and tensors again transform as vectors and tensors. However. the following statement is derived: If V is a simply connected region in space-time.12) The Ricci scalar R is deﬁned by contracting this once again. (4.9) α where g µα is the inverse of gµν .8) Taking his covariant derivative to vanish. In the lecture course on general Relativity. but because there are only two subscript indices. it is designed to ﬁx quantities that aren’t tensors back into forms that are. if and only if a ﬂat coordinate frame exists in V .7) Lowering indices using the metric tensor. µα µβ ∂xµ ∂ gαβ − Γβµα − Γαµβ . Suppose that we demand the covariant derivative of gµν to vanish as well. The Ricci curvature is a two-index tensor deﬁned by contracting the Riemann curvature: Rκα = Rµ . 8 (4. and ∂κ stands short for the partial κ derivative: ∂κ = ∂/∂u . the metric tensor gµν would be constant. indeed. then the Riemann curvature µ Rκαβ = 0 everywhere in V . 0 that is. this can be written as Dµ gαβ = (4. We have Dµ gαβ = ∂ gαβ − Γν gνβ − Γν gαν .9) will now be used as a deﬁnition of the connection ﬁeld Γ . gνµ g µα = δν .coordinate frame. This object will be used to describe to what extent space-time deviates from being ﬂat. κµα (4.11) in the last two terms.

So. This is the subject of the discipline called General Relativity. Tµν (x) is deﬁned such that in ﬂat space-time 0 (with c = 1 ).5). From here it is a small step to think of a space-time where the metric gµν (x) can be any diﬀerentiable function of the coordinates x . Then. This is achieved if the equations can be written entirely in terms of vectors and tensors.With some eﬀort. called Bianchi identity: Dα Rµ + Dβ Rµ + Dγ Rµ = 0 . it is the 00 component of a tensor Tµν (x) . (3. This is how the use of curved coordinates can serve as a description of gravity – in particular there must be curvature in the time dependence.5) and (4. 0. the mass-energy-momentum distribution of matter. all terms in the equations must transform as such under coordinate transformations. Einstein managed to ﬁgure out the correct equations that determine how this matter distribution produces a gravitational ﬁeld. it will undergo an acceleration d2 xi 1 = −Γi = 2 g ij ∂j g00 . we derive that the Ricci tensor obeys g µν Dµ Rνα − 1 Dα R = 0 . t) must play the role as a source. Of course. the mass density. κγα κβγ καβ From that.e. can be modelled by choosing g00 (x) to take the shape of the Earth’s gravitational potential. However. 2 (4. (3. at one instant. T0 = −T00 = (x) is the energy distribution. The gravitational equivalence principle requires that they transform as such under all (diﬀerentiable) curved coordinate transformations. 00 2 dτ (5. and a particle that. 0. 0) . g00 is negative). we need to have equations that determine the connection ﬁeld surrounding a heavy object like the Earth such that it describes the gravitational ﬁeld correctly. In addition. for instance.15) (4. Indeed. energy density (x. usually. The gravitational ﬁeld of the Earth. Coordinates x in terms of which gµν is completely constant do not have to exist. i. according to Eq. In that case. Ti0 = T0i is the matter 9 . Clearly.14) 5. one can derive that the Riemann tensor obeys the following (partial) diﬀerential equations. or equivalently. this tensor must act as the source of the gravitational ﬁeld. no coordinate frame exists in which all objects on or near the Earth move in straight lines. Gravity Consider a coordinate frame {xµ } where gµν is time independent: ∂0 gµν = 0 . we wish these equations to be invariant under Lorentz transformations. this is a perfect description of a gravitational force. and therefore we expect the Riemann curvature not to vanish. We then use Eqs.9) to describe the motion of objects in free fall. is at rest in this coordinate frame: dxµ /dτ = (1.1) Since this acceleration is independent of the particle’s mass. − 1 g00 can serve as an expression for the gravitational 2 potential (note that.

In his ﬁrst attempts to write an equation.3) So. see Eq. and they were.4).3). the partial derivative ∂µ has been replaced by the covariant derivative. in 1916.2) . We will here skip the details of its derivation. 2. Tµν transforms as a tensor. local coordinates is ∂i Tiµ − ∂0 T0µ = 0 . (5. so that Eddington could set up his expedition to check the deﬂection of star light by the gravitational ﬁeld of the sun. end of 1915. for a gas or liquid with pressure p .4) where G is Newton’s constant. and we will see more of that later. Now we know the importance of the equation for energy-momentum conserµ vation (5. written more compactly as Dµ Tν = 0 . just as gµν does. which equals the momentum density.15) for the Ricci tensor. This means that there is an extra term containing the connection ﬁeld Γλ . Eq. but then he hit upon inconsistencies: there were more equations than unknowns. Einstein’s ﬁeld equation now reads: 1 Rµν − 2 R gµν = −8πG Tµν . quite non-trivial solution can be found.5) 6. 10 dr2 + r2 dΩ2 . 1. who discovered that an exact.ﬂow. Einstein did not have this term. 2 (5. Einstein. 1 − 2M/r (6. in general. It was Karl Schwarzschild. because that can also be written as g µν Dµ (Rνα − 1 Rgνα ) = 0 .1) (6. though somewhat elaborate. the energy-momentum tensor does not obey the continuity equation (5. but instead: g µν Dµ Tνα = 0 . The Schwarzschild Solution When Einstein found his equation. Schwarzschild’s description of the metric gµν (x) that solves Einstein’s equations is most easily expressed in the modern notation: ds2 = gµν dxµ dxν = − (1 − 2M/r) dt2 + dΩ2 = dθ2 + sin2 θ dϕ2 . in order to see its consequences for observations.2) Under general coordinate transformations. however. (3. the tension is Tij = −p δij . and Tij is the tension. (5. conﬂicting.2). µ = 0. or in a gravitational ﬁeld.2). he quickly derived approximate solutions. The continuity equation in ﬂat. which adds or removes energy and momentum to matter. (5. The second term in this equation is crucial. 3. It matches precisely the Bianchi identity (4. (5. In curved coordinates. did not expect that the equation could be solved exactly. which is straightforward. This is the gravitational αβ ﬁeld.

whether or not the singularity is moved to the origin. deﬁned as r = (r3 −(2M )3 )1/3 . Notice from the dependence on 2 dx0 . we will denote the total mass of an object by m . Like other researchers in the early days. in the last expression. r = 0 . (6.3) (6. that indeed. He decided to replace the coordinate r by a “better” radial coordinate. Schwarzschild himself was very puzzled by the singularity at r = 2M . 2M t/(2M ) x/y = e . apart from the constant 2 1. and the shift he used simply subtracts an amount (2M )3 from the space-time volume enclosed by r . ˜ Schwarzschild died only months after his paper was published. We emphasize that.7) (6. r Notice now that. 2 There are some heated discussions of this on weblogs of amateur physicists who did not grab this point! 1 11 . − 1 g00 is the gravitational potential −M/r .4) The angular coordinates θ and ϕ are kept the same. The function r(x. y) can be obtained by inverting the algebraic expression throughout these notes. though indeed. His solution is now famous. Now.6) 32M 3 −r/(2M ) e dx dy + r2 dΩ2 . and has no physical signiﬁcance whatsoever. we can use any coordinate frame we like to describe this metric. The advantage of this notation is that one can read oﬀ easily what the metric looks like if we make a coordinate transformation: just remember that dxµ is an inﬁnitesimal displacement of a point in space and time. The reason for this substitution ˜ ˜ was that Schwarzschild used simpliﬁed equations that only hold if the space-time-volume element. and use the symbol M for Gm . only depends on the coordinate frame used. The ˜ apparent singularity at r = 2M is easier to describe when it is kept right where it is. the singularity occurs at the “origin”. By taking the log of Eq. and partially diﬀerentiating with respect to x and y .4). the zero and the pole at r = 2M have cancelled out. but the substitution r → r (in the paper. x y 2M The Schwarzschild metric is now given by ds2 = 16M 2 1 − = 2M r dx dy + r2 dΩ2 xy (6. which are deﬁned by the following two equations: r − 1 er/(2M ) . let’s call it r .5) (6.3) and (6. det(gµν ) = −1 .where Newton’s constant G has been absorbed in the deﬁnition of the mass parameter1 : M = Gm . xy = (6. x y r − 2M 2M 2M (1 − 2M/r) dx dy dt − = .2 One elegant coordinate substitution is the replacement of r and t by the KruskalSzekeres coordinates x and y . we read oﬀ: dx dy dr dr dr + = + = . the notation is diﬀerent) was unnecessary.

the lines with constant θ and ϕ are lightlike. −y) are mapped onto the same point (r. y) and (−x. However. the coordinates of the horizon are at x = 0 and at y = 0 . y) plane: the points (x. This means that region II is not part of our universe. every point (r. We do notice that the line x = 0.1) that if r = 2M . 2 b . there are no timelike or light like paths connecting these two universes. If this is a wormhole at all.1). it is a purely spacelike one. since two neighboring points on that line obey dx = dθ = dϕ = 0 . t does not serve as a time coordinate there. but as a space coordinate. nothing special seems to happen on the two lines x = 0 and y = 0 . such as is the case in the region marked I in Fig. t . similarly. we can also read oﬀ from the original expression (6. or perhaps another region of the space-time of our universe.4) attaches a real value for the time t when x and y both have the same sign.3) and is regular in the entire region x y > −1 . Indeed. The horizon is at r = 2M . Actually. in general. The orientation of the local lightcones is indicated. is lightlike. t) in the physical region of space-time is mapped onto two points in the (x. there is no physical singularity or curvature singularity at r → 2M . Apparently. and this implies that ds = 0 . the coordinate t gets an imaginary part. the line y = 0 is lightlike. dt2 enters with a positive sign in the metric (6. but if x y < 0 . θ and ϕ both constant. here. In particular. we ﬁnd that. since there. This leads to the picture of a black hole being a wormhole connecting our universe to another universe. An other important thing to observe is that Eq. 12 fu st pa riz ho on . (6.t horizon xy = −1 ΙΙ 0 Ι x tu re h 0 2M r or iz on xy = −1 y a) b) Figure 2: a ) The black hole in the Schwarzschild coordinates r. r) . b ) Kruskal-Szekeres coordinates. (6. r is then the time coordinate. Thin red lines are the time = Constant limes in the physical part of space-time. as ds = 0 regardless the value of dt . as in region II . The line y = 0 is called the future horizon and the line x = 0 is the past horizon (see Section 10). regardless the value of dy . Even if we restrict ourselves to the regions where t is real.

so with suﬃciently large quantities of mass you can always exceed that limit.1) and (7. The ﬁrst however. 13 .2) seems to be easy to interpret. the pressure term cancels out precisely. dr r2 (1 − 2Gm/r) dm = 4π r2 .5) Note that the Tolman-Oppenheimer-Volkoﬀ equations (7. where B(r) = and ∞ (7. Thus. The Chandrasekhar Limit Consider Einstein’s equation (5. seems to imply that not only energy but also pressure causes gravitational attraction. one ﬁnds that there will be values of r where M (r)/r exceeds the critical value 1/2 so that A(r) and B(r) develop singularities. stationary distribution of matter.3) 1 . anywhere along the radius r . Eq. dr (7. and (r) the r dependent local mass density. 1 − 2Gm(r)/r (7. non-singular solution can exist if the baryonic mass NB exceeds some critical value. Integrating inwards. (7. then large amounts of total mass will still show stable solutions. Let p(r) be the r dependent pressure. 1 − 2Gm(r)/r (7. but if we have a liquid that is cool enough to show a ﬁxed energy density 0 even when the pressure is low. An equation of state for the material relates p to .2) are exact. 4 the enclosed mass would be approximately 3 π 0 r3 .2) These equations are known as the Tolman-Oppenheimer-Volkoﬀ equations.1). Eq. There is no stable solution of the Tolman-Oppenheimer-Volkoﬀ equations if.4) log(A(r)B(r)) = −8πG r (p + ρ) r dr . The resulting space-time metric is calculated to take the form ds2 = −A(r)dt2 + B(r)dr2 + r2 (dθ2 + sin2 θ dϕ2 ) . This is when there is a boundary where the pressure vanishes. roughly to be interpreted as the gravitational mass enclosed within a sphere with radius r . (7.1) (7. the enclosed gravitational mass M (r) = Gm(r) exceeds the value r/2 . If the equation of state allows the pressure to be high while the density is small.7. The last of these. as soon as spherical symmetry and time independence are assumed.4). In terms of an auxiliary variable m(r) . This is a peculiar consequence of the trace part of Einstein’s equation (5. and putting c = 1 . In many cases. and some spherically symmetric. such as the calculation of the gravitational forces between stars and planets. at suﬃciently low temperatures.4). no stable. one can derive the following equations from General Relativity: dp ( + p)(m + 4πp r3 ) = −G .

The Chandrasekhar limit refers to the largest amount of mass one can make of a substance where only electron pressure resists the gravitational attraction. spherical symmetry demands that the only admissible solution will then be the Schwarzschild metric with mass parameter M . When at rest. We will have to verify afterwards that the conﬁguration obtained is indeed a correct solution. It is instructive to show some simple exact solutions. (8. If no stable solution exists.1) where v µ is the local velocity dxµ /dτ of the dust grains. In that case (and if we insist on spherical symmetry. With our initial condition (8. With this initial condition. in a local Lorentz frame. dust has only an energy density T00 = − while all other components of Tµν vanish. (8. t) = 4π 0 . at t → −∞ . the energy-momentum tensor takes the form µν Tdust = − (x)v µ v ν . This is called dust.3) Thus. In any other coordinate frame. This limit is about 1. and the equation of state p = 0 . however. matter is assumed to be conﬁned into a thin. this must mean that the system collapses under its own weight. Gravitational Collapse An extreme case is matter of the form where the pressure p vanishes everywhere. it is not so diﬃcult simply to guess the exact solution: we assume the metric to be stationary both before and after the passage of the dust shell. at t → −∞ we take for the energy density T 00 (and for simplicity G = 1 ). so that the total angular momentum vanishes). What will happen to it? 8. but for simplicity we here choose it to be the velocity of light. (8. we have to choose M inside to be zero. Both inside the dust shell and outside. T 00 (r. t) → 0 δ(r + t)/r2 . One must ask what happens when larger quantities of mass are concentrated in a small enough volume. Consider as initial state a large sphere of matter contracting at a certain speed v . gravitational implosion can never be avoided.44 solar masses. Thus. but while the dust shell passes there is a jump proportional to a theta step function.2) where the factor r−2 was inserted to ensure conservation of energy at inﬁnity: ∞ E= 0 dr 4πr2 T 00 (r. but we can 14 . dense shell with radius r → |t| .Substituting some realistic equation of state at suﬃciently low temperature.2). but. we will also consider more general solutions. one derives that the smallest amount of total mass needed to make a black hole is then a little more than one solar mass. for future use. with diﬀerent values for M in and M out . We could take v to be anything. M outside is diﬀerent from M inside.

The dotted line is the location of the horizon. The dotted line here is the apparent Schwarzschild horizon r = 2M .5) This is a reason to use a modiﬁed coordinate frame both in the inside region and the outside region. In b) . What has to be done now is to carefully formulate the matching conditions of the two regions at the location of the contracting dust shell. and outside we use (t out . or 2 dr = − dt so that dt −r = dr r − 2M → 2M A = −1 .already observe that spherical symmetry would not have left us any alternative. θ. but in both regions we make the transition to coordinates 15 . B r (8. ϕ) . we use the Schwarzschild metric with coordinates (t in . whose mass M increases. but it indicates the causal order of events. The contracting dust shell follows a lightlike geodesic in the radial direction. ϕ) . given by ds = 0 . singularity horizon M3 time tcom M3 M2 M2 M1 M1 space M=0 M=0 r a) b) Figure 3: Several shells of matter (shaded lines) implode to form a black hole. time is neither the Schwarzschild time nor the t com coordinate. Here.4) t(r) = −r − 2M log(r − 2M ) . the coordinates t com and r are used. θ. (8. r. r. 0 < M1 < M2 < M3 . Inside. In a) .

(8. Γ323 = Γ332 = r2 sin θ cos θ . θ.7) (8. g10 = g01 = 1 . g11 = 0 . In terms of the new coordinates. the metric gxx may show a discontinuity.7) in these coordinates: µ g00 = −1 + . t com = C st is a geodesic in both regions. 4 Remember that indices are raised and lowered by multiplying these ﬁelds with the metric tensor gµν or its inverse. The matching condition will now be that at the points t com = 0 the two regions are stitched together. such as x → (a + b θ(y))y .(t com . Γ233 = −r2 sin θ cos θ . r where µ(t) = 2M (t) . with a > 0 and a + b > 0 . 2r2 = −r sin2 θ . the metric is ds2 = −A dt − B dr A 2 + Bdr2 + r2 dΩ2 = −Adt2 + 2 dt dr + r2 dΩ2 µ(t) = −1 + dt2 + 2 dt dr + r2 dΩ2 . In a space-time where the Ricci tensor is allowed to have a Dirac delta distribution. let us evaluate the Ricci curvature for the metric (8. r µ 00 g = 0 . but the ﬁrst derivatives have at most discontinuities in the form of step functions. (8. 2r2 = −r .2). In fact. there must always be a coordinate frame such that: the second derivatives of the metric gµν may have delta peaks. any monotonously rising function µ(t) will be a solution to Einstein’s equations where dust ﬂows inwards with the speed of light.8) and we dropped the subscript “com” for the time coordinate t .9) r deﬁning µ = dµ/dt . r. we ﬁnd4 for the Christoﬀel symbols. so that. 2r Γ100 = Γ122 µ . the dust shell moves at the orbit t com = 0 . (8. 3 Γ313 = Γ331 = r sin2 θ . (8.6) Remember that for our original problem. Γ010 = Γ001 = − Γ133 µ . ˙ 2 Γ000 = µ ˙ . ϕ) .10) Γ121 = Γ221 = r . compare the discontinuity in g00 as a function of t com in Eq. 16 . g 10 = g 01 = 1 . t com = t out + r + 2M out log(r − 2M out ) . We call t com the co-moving time. g µν . g 11 = 1 − . To check the solution now. according to the initial condition (8. while the metric itself is continuous. θ and ϕ will be the same for both regions (otherwise the metric gµν would show inadmissible discontinuities3 ). M in = 0 . The coordinates r. If then a coordinate transformation is applied with a discontinuous ﬁrst derivative. where t com = t in + r + 2M in log(r − 2M in ) . (8. At t com < 0 we have M = M in and at t com > 0 we have M = M out .9). Γαµν = 1 (∂µ gαν +∂ν gαµ −∂a gµν ) . M (t) = θ(t)M out + θ(−t)M in .

where all of the energy of the dust is kinetic. It follows also for the trace5 that R = 0 . we see that indeed the shell is moving inwards.11) Inserting these in the equation for the Ricci curvature.14) while all other components. log −g = 2 log r + log θ .6).9). 5 17 . 0. 0) . Γ313 = Γ331 = cot θ . −Λ. 0. (8. where t com is replaced by t in or t out .13) while all other components of Rµν in these coordinates vanish. we derive that vµ = (−Λ. µ . 0 1 − 2M/r .16) ˙ M . r √ √ 2 −g = r sin θ .Γ022 = −r . To see that this is indeed the energy momentum tensor of our dust shell. The Ricci scalar R refers to this fact.17) here M is the local mass parameter. 8π r2 4πr2 (8. (8. in the original Schwarzschild coordinates. vanish. with the local speed of light. (8. From Eq. the 4-velocity is v µ = (0. according to Eq. 1 Γ212 = Γ221 = .15) so that. 0. vµ = Λ −1. in our co-moving coordinates. r2 (8. Of course. Hence G T00 = ˙ ˙ −M −1 µ = . 0 1 − 2M/r .ν + Γα − Γβ Γα + Γα (log −g) α . 2r2 µ ˙ µ µ2 Γ100 = + 2− 3 .µ.14) if G = We have vµ = −Λ∂µ t com . 0. −1 . Γ000 = Γ110 = Γ101 = − µ . Γ 33 = (µ − r) sin2 θ . 0.17). (8. we note that. and the rest mass is negligible. (8. µν. 4πr2 Λ2 (8. Γ 22 = µ − r . vµ = Λ 1 . notably also T11 . −1. that the dust has negligible rest mass. where Λ tends to inﬁnity (our “dust” goes with the speed of light). 2r2 (8. this is a limiting case. and we have agreement with Eq.α αµ βν µν we ﬁnd that R00 = µ ˙ . (8. 2r 2r 2r 0 2 1 1 Γ 33 = −r sin θ . The situation is The fact that R = 0 ensues from our choice to have the dust move with the speed of light.12) (8. √ √ Rµν = −(log −g). because g 00 = 0 . 0) . From the second expression in Eq.

sketched in Figures 3 a) and b) . 9. With a bit more work. but it does not play a role in the physically observable properties of the black hole. β . the Schwarzschild metric emerges. spherically symmetric clouds of dust. This can be rewritten as √ √ (9. when written in terms of the antisymmetric. The horizon at r = 2M might wobble a bit. What we see is that. This is because the horizon is not a true singularity but rather an artefact of the coordinates chosen. and γ . (8. deﬁned while it was still at r → ∞ . (9. we still have a vector potential ﬁeld Aµ obeying Fµν = ∂µ Aν − ∂ν Aµ .1) because the contributions of the connection ﬁelds cancel out due to the complete antisymmetry under permutations of α . it is well hidden way behind the horizon (an observation called Cosmic Censorship).4) ∂µ ( −g F µν ) = − −g J ν . this exercise can be repeated for dust going slower than the speed of light. For instance. Taking M a continuous function of t com leads to a description of less singular.3). This is what we mean when we say that the solution is ‘robust’. but it cannot be removed by small perturbations only. is very sensitive to small perturbations. are easy to ﬁnd by replacing partial derivatives by covariant derivatives: • The homogeneous Maxwell equation remains the same: ∂α Fβγ + ∂β Fγα + ∂γ Fαβ = 0 . Subsequent shells of dust go straight through the horizon. one might assume a tiny amount of angular momentum or other violations of spherical symmetry in the initial state. in Eq. coalescing to form a black hole. One readily ﬁnds that. if M in is taken to be zero. covariant tensor Fµν (x) . • The inhomogeneous Maxwell equation is now µ Dµ F ν = g αβ Dα Fβν = −Jν . At this stage it is very important to observe that this description of “gravitational collapse” allows for small perturbations to be added to it. The singularity at r = 0 on the other hand. (9. just behind the ﬁrst dust shell. M out is indeed the total energy E of our initial dust shell.2) (9. Hence. 18 . The Reissner-Nordstr¨m Solution o The Maxwell equations in curved space-time. generating Schwarzschild metrics with larger mass parameters M .3) where Jν (x) is the electro magnetic charge and current distribution.

3) we choose ds2 = −Adt2 + Bdr2 + r2 (dθ2 + sin2 θ dϕ2 ) . B = 0.13) 19 . (9. (7.8) Let us assume that the source J µ of this ﬁeld is inside the planet and we are only interested in the solution outside the planet.where the quantity g is the determinant of the metric: g = det(gµν ) . the inhomogeneous Maxwell law (9. So there we have Jµ = 0 . but now also a static electric ﬁeld.4) implies ∂r and consequently. because the Maxwell ﬁeld Fµν is antisymmetric in its two indices. (9. Just as Eq. 4πr2 (9. (9.7) Spherical symmetry can still be used as a starting point for the construction of a solution of the combined Einstein-Maxwell equations for the ﬁelds surrounding a “planet” with electric charge Q and mass m .11) √ Q AB E(r) = .5) so that we have the conservation law √ ∂µ ( −g J µ ) = 0 . The homogeneous parts of Maxwell’s law are automatically obeyed because there is a ﬁeld A0 (potential ﬁeld) with Er = −∂r A0 . • The energy momentum distribution of the Maxwell ﬁeld is Tµν = −Fµα Fν α + ( 1 Fαβ F αβ − J α Aα )gµν . to be identiﬁed with electric charge since at r → ∞ both A and B tend to 1 . and F 0r = − AB E(r) . µ. Eθ = Eϕ = 0 .10) 1 Since g = −ABr4 sin2 θ . deﬁned by Ei (x) = F0i = −Fi0 : Er = E(r) .ν (9. (9.12) where Q is an integration constant. 4 (9. E(r)r2 √ AB = 0.6) (9.9) (9.

3) .1) The solution of this equation is t = t0 ± (r + 2M log(r − 2M )) . (9.14) (9.16) (9. but if further details are needed we refer to the various more elaborate texts in General Relativity. (9. T33 = T22 sin θ = −Q sin θ /32π r . Herewith the Einstein equation (5.19) This is the Reissner-Nordstr¨m solution (1916. and we choose the minus sign. or dt 1 = . (10.20) (9. We ﬁnd µ Tµ = g µν Tµν = 0 .12) contributes to Tµν : T00 = − E 2 /2B = −AQ2 /32π 2 r4 . T11 = E 2 /2A = BQ2 /32π 2 r4 . the roots of the equation A = 0 : r = r± = M ± M 2 − Q2 /4π .18) a general property of the free Maxwell ﬁeld. 1918). dr 1 − 2M/r (10. r(t) → 2M + e(t0 −t)/2M −1 . the information given in these notes should suﬃce to derive them. In this case we have (putting G = 1 ) Rµν = −8π Tµν . 2 2 2 2 2 2 2 2 2 2 (9. 20 (10. T22 = −E r /2AB = −Q /32π r . In these lecture notes we concentrate on the physical properties of the various metrics that were found. The equation for such a light ray is ds2 = 0 . dΩ = 0 . and a light ray going radially inward. Horizons Consider the metric (6.The ﬁeld (9. at r very close to 2M . r 4πr2 B = 1/A .21) Again these singularities are artifacts of our coordinate choice and can be removed by generalizations of the Kruskal-Szekeres coordinates.17) R = 0. (9. In principle.15) (9.2) where t0 is an integration constant. 10. so that.4) lead to the following solution: A = 1− 2M Q2 + .1). o If we choose Q2 /4π < M 2 there are two “horizons”. We have not shown the complete derivations of these solutions.

physical singularity that occurs at r = 0 . In solutions that are more general than the Schwarzschild solution. The second horizon. the two horizons do not coincide. (10. at r=M+ M 2 − Q2 /4π . that reaches the boundary at r = ∞ . Whatever happens within the horizon might be called physically irrelevant.7). For the Schwarzschild solution. for instance. is a surface formed by lightlike geodesics pointing radially outwards. The true. In these coordinates. which is the point r = 0 . and to understand these. is far hidden from observation. a horizon forms when the coeﬃcient A in the metric (9. the horizon is deﬁned to be the boundary line between two regions of space-time. Eqs. The boundary line between I and II . However.5) coincides with the Schwarzschild horizon in the limit Q → 0 . (6. (10. however. show delicate singularities at points in complex time. Kepler’s elliptical orbits. The o largest one. or indeed astronauts. later we will see that quantum eﬀects do depend on details of the horizon region. one may have to pass beyond the horizon. This means that particles. 21 .4) Note that neither of these light rays ever pass through the barrier r = 2M . at r=M− M 2 − Q2 /4π . the horizon. This is why this point is called a horizon. There is also such a surface of lightlike geodesics pointing inwards. this singularity may be compared with singularities in physical equations when some parameters such as the time parameter become complex. cannot reach inﬁnity from these points.Similarly. but the planets in the solar system do not seem to be bothered about that. The light rays do actually cross the horizon. but we see in the Kruskal frame that actually. In terms of the Kruskal coordinates.8) tends to zero suﬃciently fast. the inwards and outwards light rays are easier to ﬁnd: the in-rays are at x = C st . where C st is any positive constant. Rays going at an angle rather than radially also will not pass through this point. In the Reissner-Nordstr¨m solution. regardless the value and direction of their initial velocity. both horizons are at r = 2GM .6) goes to the singularity r = 0 in the limit Q → 0 . Region II is the collection of points that have no such geodesics attached to them. since information concerning the interior region cannot be sent out using light rays. In general. and the out-rays at y = C st . we see something new. a ray going radially outward is given by r(t) → 2M + e(t−t1 )/2M −1 . Region I is the region deﬁned by all space-time points x from which a geodesic can start. they hit upon the singularity at xy = −1 . But at ﬁnite Q it is also a lightlike surface. heading towards the future direction. (10. It has the same properties as the Schwarzschild horizon. we see that the function A(r) has two zeros. Beyond the horizon.

The metric is: ds2 = − where Y = r2 + a2 cos2 θ .3) coincide. The only genuine singularity in the curvature of space and time occurs where Y (r.6) ∆ sin2 θ Y (dt − a sin2 θdϕ)2 + (adt − (r2 + a2 )dϕ)2 + dr2 + Y dθ2 . but this occurs only when both r and θ are zero: the singularity lies along the equator at r = 0 . (11. Eqs. To prove that this is indeed a solution of Einstein’s equations requires patience but is not diﬃcult. The KerrNewman solution is the most general stationary solution for a black hole with electric charge.3) Eq. r − 2M r + a2 (11. 4πY Qra sin2 θ A3 = .5) ∆ = r2 − 2M r + Q2 /4π + a2 . (11. The free parameter a in this solution can be identiﬁed with angular momentum: J = aM .1) and (11.4) (11. The Kerr and Kerr-Newman Solution A fast rotating planet has a gravitational ﬁeld that is no longer spherically symmetric but is only symmetric under rotations around the z -axis. As with the Reissner-Nordstr¨m solution. Y Y ∆ (11. The solution for that case was found by Newman et al in 1965.1) This solution was found by R. We here just give the solution: ds2 = −dt2 + (r2 + a2 ) sin2 θdϕ2 + 2M r(dt − a sin2 θdϕ)2 r2 + a2 cos2 θ dr2 +(r2 + a2 cos2 θ) dθ2 + 2 . For a derivation using more elementary principles more powerful techniques and machinery of mathematical physics are needed.11. (11. (11. Kerr in 1963. The vector potential is Qr A0 = − . 4πY (11.2) here also describes the total angular momentum in the solution. Exercise: ﬁnd the non-rotating magnetic monopole solution by postulating a radial magnetic ﬁeld. the zeros of the function ∆(r) ar not true o singularities but rather coordinate singularities. if no matter is present. θ) = 0 . 22 .2) c) The Newman et al solution For sake of completeness we also mention that rotating planets can also be electrically charged. Exercise: show that when Q = 0 .

(11. Penrose diagrams r = 0 singularity ∞+ xP ho riz on II IV III I ho r = 0 singularity ∞ Figure 4: The Penrose diagram for the Schwarzschild solution on riz yP ∞− r=0 ∞+ ∞+ ∞+ ho riz on II I ∞ ∞ r=0 ∞ ∞ r=0 ∞− ∞− a b ∞− c Figure 5: a ) Penrose diagram for the Minkowski vacuum with Cartesian transverse coordinates. 12. This is not always possible. It is of interest to ﬁnd coordinate systems that are such that they cover all of spacetime that is continuously connected to the region that one has studied before. and use that to derive Eq. b ) Minkowski space in polar coordinates.Exercise: derive the gravitational ﬁeld for a non-relativistic source by linearizing Einstein’s equation (5. preferably avoiding any coordinate-induced singularities. c ) Penrose diagram for a black hole formed by matter (darker color represents matter falling in). Exercise for the advanced student: describe geodesics in the Kerr solution. A good example is the Kruskal-Szekeres 23 .4). but we can try to choose the best possible coordinates.2).

light rays are constrained to form angles of 45◦ or less with the vertical (vertical meaning the line dx + dy = 0 ). for the Schwarzschild black hole.3) — (6.and Kerr-Newman. dy ≤ 0 . At every point in the x y frame of these coordinates. 24 (12. we can assure that the r = 0 singularity is mapped onto a straight line. coordinate system. y = tan(yP π/2) . where one of the equal signs is reached as soon as dθ = dϕ = 0 . containing a negative mass Kerr Newman black hole. this feature is reached if we choose x = tan(xP π/2) . light rays themselves form trajectories of the form dx ≥ 0. If the x coordinate is replaces by any monotonously increasing. Or. we still have the same property.1) ring singularity r= r+ r= r+ ∞− r= − III ∞− − r= r r − − r= r= r− r= r− r r r− r= r− . and y by any monotonously increasing diﬀerentiable function yP of y . The singularity at r = 0 is a natural boundary in case of ReissnerNordstr¨m.r + r + r + r= r= VII r = 0 singularity r = 0 singularity ring singularity VII ring singularity r r= V VI V’ V r= VI r − r= r= r+ − r= r + r r= r= r − − r r= IV I + ∞ r= r + + r r + r= r= r= r+ r= r− II r= r+ r= r+ r= r− III r= r− VI’ r= r− ∞+ r= r+ II ∞+ ∞ r= r+ r = 0 singularity IV I r = 0 singularity ring singularity r= r= a b Figure 6: Penrose diagram for the Reissner-Nordstr¨m ( a ) and the Kerr black o hole ( b ).4). diﬀerentiable function xP of x . but it is a ring singularity in case of Kerr. here the line xP −yP = 1 . In the Schwarzschild case. (6. Now these are not the only coordinates with this property. o through which one can continue to asymptotically ﬂat universes V and V I . This freedom we can use to obtain one other desirable feature: map the point x = ∞ to xP = 1 and the same for y and yP . Furthermore.

Region I is the region that can be reached from inﬁnity and from which one can also escape to inﬁnity. Consider a two-dimensional. Suppose that our surface at t = 0 divides 3-space into two regions: an outer region V1 and an inner region V2 . Now. such as is the case in Figure 4. Let S1 be the boundary between V1 and V3 and S2 be the boundary between V2 and V3 . convex. At early times. but escape to inﬁnity is allowed. region IV is only connected to the physical spacetime I by spacelike geodesics. θ and ϕ usually deﬁne a two-sphere. Trapped Surfaces A black hole is characterized by the presence of a region in space-time from which no trajectories can be found that escape to inﬁnity while keeping a velocity smaller than that of light. 4 shows four regions of space-time. The surfaces S1 and S2 have areas A1 and A2 . Finally. Let A be the surface area of S (calculated using the induced metric on S . We see that at negative times it corresponds to that of Minkowski space-time. 3-space is divided in three regions: – an outer region V1 that is spacelike separated from S . Region III is a domain that cannot be reached from inﬁnity. separated by horizons. which is usually possible and has the advantage that the entire space-time can be represented in a ﬁnite patch of the coordinates. the point r = 0 in 3-space forms a timelike geodesic. 7. Figure 5 b describes a black hole formed by matter. Deﬁne a time coordinate t such that t = 0 on that surface. but from where no escape back to inﬁnity is possible. See Fig. A Penrose diagram now is a representation of two of the space-time coordinates in such a way that the local light cones always show that light rays go with a maximal velocity +1 to the right or -1 to the left. at time t = ε . at later times it becomes spacelike 13. Region II is the domain behind the horizon that can be reached by test objects falling in. so that the fastest way to transmit information is by rays that are tilted by 45◦ to the left or to the right. closed. The other two coordinates. A small instant later. This implies the presence of trapped surfaces there. The diagram of Fig. Characteristic boundaries are represented as much as possible by straight lines. we deﬁne the 25 . – an inner region V2 that is also spacelike separated from S . Its boundary can be reached with lightlike geodesics from S . The r = 0 singularity lies in the future of any test object there. and – a region V3 between V1 and V2 that can be reached by timelike geodesics from S . 4. The Penrose diagram of ﬂat Minkowski space-time is shown in Figure 5 a . We start from the following deﬁnitions.Space-time is then sketched in Fig. spacelike surface S in a curved spacetime.

1) dε dε Under non-exotic circumstances. and . we can have a trapped surface. The importance of this theorem is that it shows that black holes cannot disappear once they have been formed. (13. One of the most important theorems is: If. S is called trapped iﬀ both expansion rates are negative or zero: θ1 = θ1 ≤ 0 and θ2 ≤ 0 . inside a black hole. and presumably disappear altogether.2) A surface is marginally trapped if the equal sign in Eq. V2 and V3 .a trapped surface stays trapped forever. A little later. θ2 of these two surfaces as follows: dA1 dA2 . so that its area. This is because all light-like geodesics leaving this surface have r = 2M . the matter distribution in a spacetime obeys the constraint that the energy density is non-negative anywhere. However. other theorems show the inevitability of singularities Of course. in our notation. and they obey a number of important theorems. These can indeed cause black holes to shrink.the area of the largest trapped surface can only stay constant or increase. at t = ε . (13. T00 ≤ 0 in all coordinate frames. certainly the outer surface expansion rate is positive: θ1 > 0 . What happens in the presence of matter. S2 S1 expansion rates θ1 . does not increase with time.3) 26 .S V2 V3 V1 Figure 7: A surface S at t = 0 as described in the text. θ2 = . which in the local induced metric is 4π(2M )2 . signals moving inwards and outwards divide 3-space into the regions V1 . For a pure Schwarzschild black hole.6 Indeed. (13. then . 6 (13. when the solutions of Einstein’s equations look a lot more complicated? In that case. the surface r = 2M is marginally trapped. we can still deﬁne trapped surfaces. see Section 18. The inner one is usually negative. or.2) holds. in all locally regular coordinate frames. such as in a ﬂat space-time. we have not yet considered quantum mechanical eﬀects.

7) and all others are zero. (13. and one obtains Rxx 2 2 ∂x A ∂x r 2 ∂x r = − Ar r 2 2 ∂y A ∂ y r 2 ∂y r . Consider the most general spherically symmetric metric. The connection ﬁelds are easily calculated: Γx xx Γy yy ∂x A = . which are deﬁned by demanding that the lines x = constant and y = constant are in fact light rays ds = 0 . (5. such as the r = 0 singularity of the Schwarzschild black hole.4)). the cross term C(r. y)dxdy + r2 (x. . Here follows the calculation. = A2 A r Rxy Rθθ = 1 − 2 ∂ x r ∂y r + 2 r ∂ x ∂ y r . (13. A ∂y r Γθ = Γϕ = . θϕ Γx ϕϕ (13. A ∂y A = . t)drdt can easily be removed by a proper redeﬁnition of the time coordinate t .6) From this metric. except the ones obtained from the above by interchanging the two subscript indices. A −r ∂x r sin2 θ y Γ ϕϕ = . t) = 0 (13.4). t)dt2 + B(r.4) It is the direct generalization of the vacuum solution (6. we plug into Einstein’s equation. so there is no loss of generality if we put C(r. the Ricci tensor can be derived. ds2 = −A(r.9) . This implies that the coeﬃcients for dx2 and dy 2 must vanish. (13. t)drdt + r2 dΩ2 . but we list it to enable the reader to check. ignore the factor 8πG : Txx = 2 2 ∂x r 2/. temporarily. Rϕϕ = sin2 θ Rθθ .5) (the r coordinate is ﬁxed by demanding the angular dependence as in Eq. Tyy = A ∂y r ∂y r A . Ryy = − . (13. which one might decide to skip at ﬁrst reading. . Eq. ∂x A ∂x r A − = ∂x r Ar r ∂x r A 27 . t)dr2 + C(r. It is now convenient to choose light cone coordinates x and y . A −r ∂y r = A −r ∂x r Γy = θθ A ∂x r Γθ = Γϕ = xθ xϕ r θ Γ ϕϕ = − cos θ sin θ Γx θθ . Actually. which in the presence of matter reads ds2 = 2A(x. we calculate how it relates to the matter distribution Tµν . −r ∂y r sin2 θ = . but instead consider the much simpler spherically symmetric case. From these.forming inside trapped surfaces after some time.8) A This.7). (13.1). We then get the direct generalization of the Kruskal metric (6. where we. We will not give the general proofs. Ar r ∂x A ∂y A ∂x ∂y A 2 ∂x ∂y r − − . yθ yϕ r Γϕ = cot θ . y)dΩ2 . .

along a line y = y0 = constant. we can regard the coordinate x as our time coordinate. We claim that the positive energy condition will also require Txx ≤ 0 . where ∂x r = 0 . y0 ) where ∂x r = 0. y} . we have a marginally trapped surface.12) Proof: Consider some given values for Txx . Txy . we have ∂y r ≥ 0 . we will use the ﬁrst of Eq. and x = λ( + τ ) . we may have a point (x0 . } . Since y runs in the negative time direction. ∂x r < 0 .9). then forms a series of trapped surfaces. so that a new line b emerges at y = y1 > y0 .10) (13. If no matter is present. In particular. (13. the line a . and Tyy .Txy = A − ∂x ∂y (r2 ) 1 (A − 2 ∂x r∂y r − 2 r ∂x ∂y r) = .13) where the parameter λ is chosen suﬃciently large. But if matter falls in (green arrow). then beyond that point on the same line a . (13.14) x a′ b a y Figure 8: In the lightcone coordinates {x. Tϕϕ = sin2 θ Tθθ .15) (13. A (13. where τ serves as the new time variable. so that this point represents a trapped surface. then according 28 . y= 1 ( − τ) . corresponding to y = y0 . Since ∂ ∂ 1 ∂ =λ − . Now go to a coordinate frame {τ. ∂y r > 0 . we have ∂x r = 0 . Demanding this to be negative or zero for all λ implies Ineq. This is a new trapped surface with a larger area. Now suppose that. (13. so that. (13. r2 r2 r ∂x ∂y log A + 2 ∂x ∂y r Tθθ = − . If there is no matter around.11) Now.12). λ (13. ∂τ ∂x λ ∂y we have Tτ τ = λ2 Txx − 2Txy + λ−2 Tyy . at some positive value of A . keeping y zero or very small.

For reasons to become clear shortly. θ) dropped out). However. dt dϕ a = 2 . and its area is π 2π Σ= 0 dθ 0 2 2 dϕ sin θ (r+ + a2 ) = 4π(r+ + a2 ) (14. but it can be generalized to the non-symmetric case. (14. r− and a the following way: 4J 2 Σ a − 4π 2 +a 2 Q2 + Σ 4π 2 = 0. the marginally trapped surface is replaced by a larger one. as soon as matter falls in. obeying Ineq. Q and a . 2aM = a(r+ + r− ) = 2J .9). All one needs to know is that all matter that falls through the horizon. Thus. In that case.to Eq. we now wish to express these in terms of the three mutually independent parameters Q. We used spherical symmetry for this simple argument. See Fig. 8. if there is matter around. (13. (14. dt dθ =0. because at that point the lightlike geodesics going in and going out coincide: ∆(r) = 0 . the total area of the horizon can only increase.12). dt r + a2 (14. this surface has a larger r value. the marginally trapped surface is then also a horizon. (13. and Σ . (14. has a positive energy density in any locally regular coordinate frame.3) (where the dependence of the function Y (r. 2 r+ + a2 = Σ/4π .1) Deﬁning the roots of ∆ to be r± = M ± M 2 − a2 − Q2 /4π . We can therefore conclude that the area of the horizon increases when matter falls in. the Kerr-Newman solution (11.3). We can now go to a slightly larger value of y to ﬁnd a marginally trapped surface.4) They allow us to eliminate r+ . We have the following equations. dr =0. The four laws of black hole dynamics Consider the most general black hole solution.5) . a2 = J 2Σ 1 4πJ 2 + 16π (Q2 + Σ)2 29 . hence a larger area. 14.2) we ﬁnd that the horizon is at r = r+ . J . r will keep the same value. w the quantity ∂x r will be negative some time later. r+ r− − a2 = Q2 /4π . This means that the line y = y0 is now well within this trapped surface. The free parameters of this solution are M. Since ∂r/∂y > 0 (the surface is also trapped at the inner side). The horizon occurs where ∆ = 0 .

If we want the vector potential for a test particle that rotates with angular velocity 2 Ω = a/(r+ + a2 ) . Σ is the area of the horizon. so also no thermal radiation. ϕ) . just like the total entropy in thermodynamics. see Eq. τ acts as a temperature. This is seen as follows. The vector potential (11.6) How does this change upon small variations of our free parameters? We write dM = τ dΣ + Ω dJ + φ dQ . with ˜ ϕ = ϕ − Ωt . 2 4π(r+ + a2 ) (14.7) a . It was found by Bardeen. But the similarity with the entropy law went further. we have to transform to the co-rotating coordinates (t. 2 16π(r+ + a2 ) Ω= 2 r+ (14. this equation resembles the entropy equation in statistical mechanics. The vector ﬁeld transforms as follows: ˜ ˜ x Aµ d˜µ = Aµ dxµ .8) these now have the following interpretation. θ. (14. where Y = r+ + a2 cos2 θ . we have at r = r+ . 30 .1). but what is τ ? In all respects. For all systems with angular momentum J . φ is the electrostatic potential for a test charge crossing at the horizon. the second law of thermodynamics has an analogy in black holes: the second law of black hole physics states that the total area of all horizons cannot decrease. 16π Σ Σ (14. Σ) = 1 Q4 4πJ 2 + 2Q2 + Σ + . r. Indeed. This it has in common with entropy.9) ˜ ˜ A0 dt + A3 dϕ = A0 dt + A3 (dϕ − Ωdt) → ˜ A3 = A3 . Ω is an angular velocity.10) This is the vector potential felt by the test charge with angular velocity Ω . thus. 2 4π(r+ + a2 ) = (14. the increase in energy upon an increase of J is the angular velocity. on the independent parameters Q. Due to the trapped surface theorems. In that case. θ. we also know that the area of a horizon cannot decrease. Carter and Hawking. ˜ (14. as was thought at ﬁrst. Similarly. Qr+ a2 sin2 θ ˜ A0 = A0 + ΩA3 = − 1− 2 4πY r+ + a2 −Qr+ = .6) holds in the coordinates (t. r. + a2 φ= Q r+ . /J and Σ : M 2 (Q. These derivative functions are now derived to be τ= r+ − r− . J. this is the angular velocity that any object acquires when it goes through the horizon. because the black hole temperature is zero: nothing can come out.This gives us the dependence of M . 2 and since dϕ = dϕ − Ωdt . and they noticed this similarity. It could not be the real temperature of a black hole. the total mass/energy. ϕ) .

Thus. so that.7). and plays an important role in string theories. e−τ = z − t . ˜ (15. This.1) . Rindler space-time Consider ordinary Minkowski space-time.The ﬁrst law of black hole physics is equation (14. and it takes the same value all across the horizon. Very near the horizon. It is instructive to transform towards the curved coordinates (τ.13) Diﬀerentiating with respect to r gives something that could be called the gravitational ﬁeld at the horizon. is constant on the horizon. 2 r+ + a2 (14. The quantity τ= κ = 8π M 2 − a2 − Q2 /4π . . indeed. It states that the increase of mass of a black hole is the sum of all kinds of energy that is added to it. where ˜ eτ = z + t . (14.12) The ratio between the time component and the space component is therefore the square of (r − r+ )(r − r− ) . κ is sometimes referred to as the “surface gravity” at the horizon.14) This is called the extreme limit of the Kerr-Newman black hole. where the metric is deﬁned as ds2 = −dt2 + dx2 + dy 2 + dz 2 . described by the coordinates (t. x) .3) approaches ˜ ˜ + ds2 → Y dr2 −(r − r+ )(r+ − r− ) 2 dt + 2 (r+ + a2 )2 (r − r+ )(r+ − r− ) . or a2 + Q2 /4π → M 2 . x. y) . (14. The amount dU = τ dΣ is then interpreted as heat energy. 2 8π(r+ + a2 ) (14. which is κ in Eq. y.11) cannot normally go to zero. at constant θ and ϕ . just like the temperature for an object in equilibrium. The extreme limit of black holes has many special properties. z) . we again replace the angular coordinate a ϕ by ϕ = ϕ − r2 +a2 t . but this has not been proven.2) (15. It occurs when r+ and r− coincide. The third law would be that τ cannot be zero. or the “surface gravity”. is a delicate limit. It is dubious whether this limit can be reached in practice. 31 x = (x. 15.11). (14. We return to this topic in Section 18. the zeroth law of black hole dynamics states that the “temperature” τ . the metric (11.

6) In fact. 32 (16. close to r ≈ 2M .6). 8M (r − 2M ) = 2 . In Euclidean space.A substitution τ →τ +λ .1) is ds2 ≈ − r − 2M 2 2M dr2 dt + + d˜2 = x 2M r − 2M = − 2 dτ 2 + d 2 + d˜2 . with gradient 1/ 0 . it ˜ seems to be interesting to replace the time coordinate t by an imaginary time: t = it .1) . at any position = 0 .4) In this coordinate frame. We can be sure that Nature’s laws will not change. we can ﬁnd out about quantum phenomena near a horizon by studying them ﬁrst in ﬂat Minkowski space-time.3)—(6. the laws of nature are invariant under translations of the new time variable τ . In particular. and then in the black hole. the actual gravitational ﬁeld strength felt by the observer is inversely proportional to the distance from the origin. x = (2M θ. Euclidean gravity Mathematical functions in space-time coordinates can often be extended to complex values of these coordinates. and hence corresponds to a Lorentz transformation in Minkowski space-time. (15. and it is very instructive for the study of gravitational ﬁelds. ˜ (15. There. 16. in this curved coordinate frame. any small region very close to the horizon of a non-extremal black hole can de compared with Rindler space-time. (15. are closely related to the KruskalSzekeres coordinates (6.3) where λ is a constant. this observer would be tempted to ˜ redeﬁne time as t = τ / 0 . so that the gravitational potential would feel as V = / 0 . 2M ϕ) . therefore. Actually.2). an observer experiences a gravitational potential proportional to . Near θ ≈ π/2 . Therefore. the ﬂat Minkowski coordinates z and t . t/4M = τ . and so. x (15. since all physical phenomena observed in this world can be derived from what they are in Minkowski space-time without any gravitational ﬁeld. the metric then becomes ˜ ds2 = +dt 2 + dx2 + dy 2 + dz 2 . the metric (6.5) then. In fact. replace the Schwarzschild coordinates as follows. x (15. they continue to obey the same equations. Thus. would leave z 2 − t2 invariant. then in Rindler space-time. The metric in the new coordinates is ds2 = − 2 dτ 2 + d 2 + d˜2 . This space-time is called Rindler space-time.

in Euclidean space.2) ˜ ˜ t = −z sin γ + t cos γ . (16.6) r 2 = ξ 2 + η2 = − 1 er/(2M ) . asymptotically. Time translations. So. writing |x| = |y| = . usually denoted as R4 .7) turn into |x| = |y| . In the absence of matter. If they were not the same point. x y is real . We see that. so. η = sin( 4M ) . (6.9) points in space and time return to their original positions. this spacetime is not excactly ﬂat Euclidean spacetime.3)—(6. this spacetime is not described by ordinary ﬂat Euclidean spacetime.1) is invariant. Thus. The situation is sketched in Figure 9. (16. ˜ (16. but by a cylinder. singularities would arise at the origin of (ξ. under which the metric (16. η) space. only refers to the region r ≥ 2M . and. also at very large values of r . after a time translation over one period. where the metric is positive everywhere. ˜ T = 8πM . we do the same substitution in the Kruskal-Szekeres coordinates. t t ξ = cos( 4M ) . Rindler time translations are simply rotations in Euclidean space.8) r We see that the solution is rotationally invariant in (ξ. So we can take Schwarzschild’s solution and extend it to Euclidean times: 2M ˜2 1 ds2 = 1 − dt + dr2 + r2 dΩ2 . τ z = cos τ . R3 ⊗ S1 . because points in Euclidean spacetime that are separated by one period T in Euclidean time have to be identiﬁed as being the same point. At large distances.7) 2M 32M 3 −r/(2M ) 2 ds2 = e (dξ + dη 2 ) + r2 dΩ2 . (16. This observation will be very important later on (Section 18). 33 . (16. Einstein’s equations in ordinary spacetime remain unchanged when we go to Euclidean spacetime. one can also extend to imaginary values of the Rindler time τ : τ = i˜ . and these are periodic.5) x = y ∗ = eit/(4M ) = ξ + iη . η) space. but it now has the orthogonal group SO(4) . The region of Euclidean space-time that we described.3) ˜. ˜ ˜ (16. ˜ = sin τ . (16. where S1 stands for the circle. and the metric is regular at the origin. We ﬁnd that Eqs.The invariance group is then not the Poincar´ group with the Lorentz group as its homoe geneous part. In Rindler spacetime. one cannot go through the horizon here.4) r 1 − 2M/r To see what happens at the horizon. Lorentz transformations are replaced by ordinary rotations: ˜ z = z cos γ + t sin γ . are now rotations. (16. the transition towards Rindler spacetime is nothing more than a transition to cylindrical coordinates. t = it t ˜ This means that.

5) a(k)eik·x−ik t − a† (k)e−ik·x+ik 0 0t a(k)eik·x−ik t + a† (k)e−ik·x+ik 0 0t . for the time being. Its local commutation rules are [Φ(t. [a(k). x )] = 0 .2) one ﬁnds the Fourier mode expansion Φ(t.2). we shall not need to include interactions. .7) 34 . We consider a quantized scalar ﬁeld Φ(t. x )] = iδ 3 (x − x ) . 2 µ eτ . (17.3) . (17. (17. x) = d3 k 2k 0 (k)(2π)3 ˙ Φ(t. the Klein-Gordon equation (17. x) = K ω. ˜ 2 ˜ ( ∂ 2 − m2 ) Φ = 0 .2) reads ( ∂ )2 − ∂τ2 + Solutions periodic in τ are x 1 1 Φω.Figure 9: The Schwarzschild black hole in Euclidean gravity. Asymptotically.1) Assuming this ﬁeld to obey the Klein-Gordon equation. this spacetime is a cylinder. ˙ [Φ(t.6) (17. x). 1 µ e−τ eik·˜ = K ω. 17. x) = −ik 0 d3 k 2k 0 (k)(2π)3 where k 0 (k) = k 2 + m2 (always with the positive sign). a(k )] = 0 . Φ(t.k (τ. (17. a† (k )] = δ 3 (k − k ) . (17. 2 µ ˜ ˜ 2 2 ˜ x eik·˜−iωτ . x). 2 (∂ 2 − ∂0 − m2 )Φ = 0 . x) in Minkowski space-time. Φ(t. and we shall investigate what it looks like in Rindler space-time. and the operators a(k) and a† (k) obey the commutation rules for operators that respectively annihilate and create a particle: [a(k).4) In the Rindler space coordinates (15. so we are talking of a free ﬁeld. (17. 1 µ . The Unruh eﬀect For the following sections some basic knowledge of quantum ﬁeld theory is required.

ω )] = δ 2 (k − k )δ(ω − ω ) . ω) = where k 0 (k) = ∞ −∞ (17.8) We used the fact that the function K obeys K(ω. . if the variable s in (17. 2 µ ) eik·˜−iωτ a2 (k. . ω) e −iω ln k3 +k0 µ (17. β) = K(−ω. so it really is a creation operator.˜ where µ2 = k 2 + m2 and ∞ K(ω. (17.11) iω ln dk 3 √ a(k) e 0 2πk k3 +k0 µ . ω) annihilates a negative amount of energy. . rewriting it in terms of the Rindler coordinates and τ .11) also goes over negative values for ˜ ω . −β) . k 3 = 0 .14) However.7) is readily obtained by taking one of the plane wave solutions in Minkowski space-time. x) = ˜ dω K(ω. Eq. 2 4 2(2π) −∞ so that the operator a2 is identiﬁed as ˜ a2 (k. −α. . σα.6) is obeyed. x) with respect to τ as ˜ follows: Φ(τ. It is not diﬃcult to verify directly (using partial integration in s ) that the partial diﬀerential equation (17. Plugging Eq. x) = A(τ. the operator a(k. The inverse of this Fourier transform is ∞ a(k) = −∞ √ dω 2π k 0 a2 (k. Therefore. α. and then Fourier transforming it with respect to τ . (17. a† (k . β/σ) = σ −iω K(ω. (17.9) it can be expressed in terms of the familiar Bessel and Hankel functions.3) gives us Eq. First. (17.13) into 0 3 µ Eq. and ∂k 0 /∂k 3 = k 3 /k 0 ). α. s (17. 35 (17. We now normalize the Fourier components of a ﬁeld Φ(τ. we must rearrange the positive and negative ω contributions when they are added in Eq. we must be aware of the fact that the integration in Eq. 2 (17. (17. (17.13) (remember that k 0 is a function of k 3 . β) . β) = 0 ds iω −isα+iβ/s s e .12) k 3 2 + µ2 . one has K ∗ (ω. x) + A† (τ. α.8) is identiﬁed with k −k = k0 +k3 . There. µ From the commutation rules (17. x) .11). (17.5). ω) . 1 µ . k 0 = µ . ˜ ˜ ˜ ∞ ˜ d2 k ˜x 1 ˜ A(τ. To do this.10) (17. before interpreting these as annihilation and creation operators. it is convenient to note some properties of the functions K . we derive similar commutation rules for a2 : ˜ ˜ ˜ ˜ [a2 (k. (17. . ω).15) .10).

we see that we have ˜ ˜ aI (k.18) and similarly one has K(−ω. the ﬁeld Φ depends only on aI and a† .c. at > 0 . The normalization has again been chosen such II that ˜ ˜ ˜ ˜ ˜ ˜ [aI (k. where ∞ (17. a† ] = 0 . ω) e−πω 0 0 1 a† (−k. α. ω). ω) 1 − e−2πω 0 ˜ ˜ a† (−k. −ω) 2 II (17. This allows us to collect the positive and negative ω contributions in Eqs. Taking the case φ = π . ω) 1 0 0 e−πω a2 (k.20) At this point.Next. ω) + e−πω a† (−k. −ω) −e 1 0 aII (k. ω) ˜ ˜ a† (−k. ω) 0 1 e−πω 0 a† (−k. we have ˜ Φ(τ. ω) a2 (k.8). 1 µ ) 2 ˜ ˜ × a2 (k. 1 µ ) 2 ˜ ˜ × a2 (k. ω). ω) 0 a2 (k. it is opportune to deﬁne the new creation and annihilation operators aI . ω) = √ 1 I 2 −πω ˜ ˜ aII (k. ω) 1 0 0 −e−πω ˜ ˜ a† (−k. ω )] = [aII (k. applying the following Bogolyubov transformation. 2 (17. ω )] = δ(ω − ω )δ 2 (k − k ) . we ﬁnd that ∞ 0≤φ≤π . aII ] = [aI . α.22) We then see that. I II [aI . β) if α < 0 .19) < 0 . a† . β) = 0 ds −iω −πω isα−iβ/s s e e = e−πω K ∗ (ω.11) as follows. ω) −e−πω 0 0 1 a† (−k.21) e 1 0 1 − e−2πω ˜ ˜ a† (−k. one may rotate the integration contour as follows: s → s e−iφ . a† (k .23) . 2 µ . −ω) (17. we have ∞ Φ(τ. (17.(17. the integrand is bounded in the region Im(s) ≤ 0 . β > 0 .c. I the ﬁeld Φ depends only on aII and a† . x) = ˜ 0 dω e −iωτ ˜ x d2 k eik·˜ 2(2π)4 1 K(ω. x) = ˜ 0 dω e −iωτ ˜ x d2 k eik·˜ ˜ 2(2π)4 1 K(ω. β < 0 .10) and (17. In the region > 0 . a† (k . α. β) if α > 0 . ii 36 (17. −ω) + H. while at < 0 .17) s (17. ω) I 2 = √ 1 −πω ˜ ˜ a2 (k. . aII and a† . (17. In the deﬁnition (17. Therefore. −ω) + H. −ω) 2 II Inverting this. αβ) = e+πω K ∗ (ω. 2 In the opposite quadrant of Rindler space. when ω > 0 : I II ˜ ˜ aI (k. ω) + e+πω a† (−k. 2 µ .16) K(−ω. let α > 0 and β > 0 . ω) 0 1 −e−πω 0 a† (−k. .

0 R . the operator that generates a boost in the Rindler time parameter τ . in ˜ Rindler space? For this observer. What does |Ω M look like to the observer who is at a ﬁxed position ( . all observables constructed out of the ﬁelds φ in the quadrant I where II > 0 commute with HR and the observables in quadrant II. ω)|Ω M M ˜ = e−πω a† (−k. . We refer to this state as |0. ω) and a† (k. HR ] = 0 . I (17.The importance of this is the following. >0 II HR = dx| |HM (x) . ω) = HR − HR (17. Let is take that state. ω)|Ω M .25) Consequently. if HM (x) is the Hamiltonian density in Minkowski space-time at time t = 0 . ω)|Ω M =0. and their hermitean conjugates only create positive energies. We have ˜ aI (k. x) . we ˜ ˜ could have kept the original operators. deﬁning a2 (k. with > 0 .26) This describes empty space-time as experienced by an observer who is stationary in Minkowski space. and in the quadrant < 0 only the other operators.27) These equations are easy to solve. which is deﬁned as a(k)|Ω M =0. It is important to realize this. |Ω M . The vacuum state as experienced by an observer in Rindler space is the ground state I I II II of HR . (17. It is called the Boulware vacuum. and since [HR . x) . that is. ˜ aII act.28) 37 . without the apparent need for a Bogolyubov transformation. However. The operators aI and aII are deﬁned such that they only annihilate objects with positive energy. ω) − a† (k. (17. −ω) as annihilation 2 operators at ω > 0 . only the combination created by a† can I be detected using the ﬁeld F (τ. then I HR = d3 x HM (x) . < 0 commute with I HR . ω) 2 I II ˜ ˜ ˜ ˜ ˜ d2 k a† (k. or freely falling in the gravitational ﬁeld of Rindler space-time. ˜ a2 (k. the operator aI describes the particles. ω)aI (k.ω √ ∞ 1− e−2πω n=0 |n I |n II e−πωn . II ˜ = e−πω a† (−k. ω)|Ω ˜ aII (k.24) . we can also have the ground state of HR . the Rindler space Hamiltonian. <0 (17. In the quadrant > 0 . is ∞ H = −∞ ∞ dω ω dω ω 0 ˜ ˜ ˜ d2 k a† (k. ω)|Ω M . I II = One may also verify that. because if we had not paid attention to this. the Boulware vacuum is not at all the lowest energy state in Minkowski space-time. We ﬁnd |Ω M = ˜ k. ω)a2 (k. Just because the ﬁelds depend on time as e−iωτ . ω)aII (k.

and therefore O(|ψ I |ψ II ) = |λ II (O|ψ I ) .2). It must II commute with HR . (17. at the distance = 0 from the Rindler horizon. Consider any observable O in the positive sector of Rindler space-time. the expectation value of such an operator is M Ω|O|Ω M = (1 − e−2πω ) n1 . Note that this temperature is dimensionless. This is the Unruh eﬀect.31) where Ω is the density matrix C e−βHI corresponding to a thermal state ate the temperature T = 1/(k β) = 1/(2πk) . (17. τ . The free energy F of any thermal quantum system is computed as e−βF = Tr (e−βH ) . Hawking radiation The region of space-time in the vicinity of the horizon of a black hole.5): t = 4M τ .where the square root is added for normalization. that is.30) ˜ Let us concentrate on only one sector of k and ω . experiences radiation with a temperature T = g/(2πk) . the temperature experienced there is given by kTH = 1/8πM = 1/8πGmBH . Note that HR |Ω M I II = (HR − HR )|Ω M =0.32) where k is Boltzmann’s constant: β = 1/(kT ) . In Section 15. (15. in the Rindler coordinates (15. we saw that. is dimensionless. 18. the strength of the gravitational ﬁeld is 10 . the Schwarzschild time coordinate t relates to Rindler time there as in Eq.n2 II n1 |I n1 |O|n2 I |n2 −πω(n1 +n2 ) II e = n≥0 I n|O|n I e−2πnω (1 − e−2πω ) = Tr (O Ω) .1) . This is because the time unit. The value for this temperature could have been derived more intuitively as follows. 38 (18. and furthermore that time has to be rescaled by a factor 0 . remember that HR is the generator of Lorentz boosts. (17. approximately takes the form of Rindler space. This is the Hawking temperature of a black hole.2) (18. There. (17. Therefore. Therefore.29) which conﬁrms that the Minkowski vacuum is Lorentz invariant. we conclude that an observer who is being accelerated by a gravitational ﬁeld with strength g in relativistic units.

and taking the trace means that the evolution operator is connected to itself after this period in imaginary time.6) . cannot depend on the position along the horizon: if a solution is periodic with period P at one spot. So.5) The Unruh temperature g/(2πk) is thus connected to the fact that it refers to a Rindler space that has periodicity 2π/g .11). it cannot have any diﬀerent periodicity elsewhere. where we re-inserted the constant . rather than the “surface gravity”. (18.8) . since the spacetime must still have the same analytic form after any number of periods in this particular Euclidean direction. We can now also understand why the temperature. Assume a small disturbance: e−βF (J) = Tr (e−β(H+JO) ) . then we have. (18. 2 2(r+ + a2 ) 2M 2 − Q2 /4π + 2M M 2 − a2 − Q2 /4π 39 (18. dr = 2 d . Therefore. or equivalently.4) The operator e−βH happens to be the evolution operator e−iHP for a time period P = −iβ . (18. one can compute the thermal average of any operator O of a system.where β = 1/kT . 2 2(r+ + a2 ) 2 2 . From this expression for F . β∂J ∂J E (18. the Hawking temperature of a Kerr Newman black hole is kT = κ/2π = M 2 − a2 − Q2 /4π r+ − r− = . keeping β ﬁxed: O T (18. we should view κ as the parameter that scales the time variable in Rindler space-time at the horizon. and k is the Boltzmann constant. Writing r − r+ = we ﬁnd ds2 → 4Y − κ2 2 dτ 2 + d r+ − r− r+ − r− κ= .9) derived for the Kruskal spacetime in Section 16. (14. For the general Kerr-Newman solution. and the Hawking temperature 1/(8πkM ) follows from the periodicity (16. the metric near the horizon approaches Eq.7) So.3) = ∂ − β∂J Tr e−βH−βJO |J=0 E|O|E e−βE = = −βE Tr e−βH Ee ∂ ∂F − log(Tr (e−βH−βJO ))|J=0 = |J=0 . the surface gravity (14.12). this essentially means that quantum mechanics over a space-time that is periodic in imaginary time is equivalent to working out thermal expectation values of operators at a temperature T equal to P = β = /(kT ) .

Let us nevertheless take Eq. We do expect this to be quite a bang. and be observed there. The fact that there are particles with a certain temperature near the horizon of a black hole. The implication of black holes for a quantum theory of gravity In thermodynamics. because the total mass-energy emitted in the last second. the parameter that looked like a temperature when we phrased the “four laws of black hole dynamics”. The intensity of the radiation will be proportional with T 4 close to the horizon. dt dt M (t) ≈ (3C2 ) 3 (t0 − t) 3 .9) where Σ is the area of the horizon. turns out to be formidable. however. weakly. and the total energy loss per unit of time due to this radiation will be approximately U = C1 Σ T 4 = C2 M 2 M −4 = C2 M −2 . once we put our conventional units back in. 19. Scaling everything else there accordingly. The entropy of a black hole equals 1 S = 4 kΣ . we ﬁnd that the actual entropy S . (18. where κ is the “surface gravity”. the entropy S of a system with no other adjustable parameters obeys T dS = dU . with its temperature T = 1/(8πkM ) . really is a temperature! It is actually 4 times the parameter τ that was introduced in section 14.11) Clearly. What exactly happens at that moment. Indeed. (18. if left by itself. as it occurs in the equation dU = T dS +· · · is 4 times smaller: The temperature of a black hole equals kT = 4κ . we must conclude that. because the number of particle types participating in the radiation depends on whether the temperature is suﬃciently high to excite particles with given rest masses.1) . (18. Units for the temperature could be chosen such that it is one. but also. Note that we kept Boltzmann’s constant k in our descriptions of temperature and entropy. cannot be understood without a more complete understanding of quantum gravity than we possess today. Assuming conservation of total mass/energy (General Relativity would be inconsistent if we did not).Clearly then.10) as a rough approximation. a black hole should loose mass: dM dM 3 ≈ −C2 M −2 → ≈ −3C2 .10) where C1 and C2 are constants depending not only on the geometric details of the black hole (what will be its apparent surface area as seen from inﬁnity?). means that some of these thermally excited particles can escape to inﬁnity. on temperature. 1 1 (18. there will be radiation emerging from the horizon. 40 (19. at some moment t = t0 the black hole must disappear altogether.

the entropy is nothing but the logarithm of the total number of states over which we sum. In these expressions. ∂T (19. 41 (19. or S=− ∂F . this is a fundamental feature. which may be a collection of many systems but always in such a way that the total energy U is kept ﬁxed. Then. the sum is only over all states with the same energy E = U . Since here. For a black hole. while k is Boltzmann’s constant and H is the quantum Hamiltonian.5) Eq. In that case. The quantity F . and we derive S/k = Tr (βHe−βH ) + log Tr (e−βH ) = log Tr (e βH Tr (e−βH ) −βH (19.3) (19.8) or.2) In statistical physics. The sum is over all quantum states |E . (19. the entropy is k times the logarithm of the total number of quantum states that can describe the system we are looking at.7) is 1. the exponent in Eq.where U is the energy stored as heat.9) that the total number of “black hole microstates” is given by = C · eΣ/4 . The total energy U is then given by U = E ∂ − ∂β e−βF (β) E e−βE Tr (H e−βH ) ∂ = = = (βF ) . (19. (18. −βE Tr (e−βH ) e−βF ∂β Ee (19. (19.2) can be written as S = kβ(U − F ) .6) ). it was assumed that our system is a grand canonical ensemble. and the entropy is then seen to be S = k log(Tr (1)) . is called the Helmholtz free energy. deﬁned as F = U − TS .7) where U has been written as an average: U = H .4) where β = 1/(kT ) . and it obeys dF = −SdT . This is a quite general result: In a quantum system.9) . (19. the free energy F has been identiﬁed by the equation e−βF = E e−βE = Tr (e−βH ) . The quantity e−βE is the Boltzmann factor describing the probability of any state |E to occur when there is thermal equilibrium. we conclude from Eq. We also can consider micro canonical ensembles. (19.

which is given by the thermal probability: W dt = σ(k)v −βH E e dt . (18. emerges as a density matrix. and momentum dependent. In Eq. V (19. (19.12) where k is the momentum of the ingoing particle. The application of thermodynamics to black holes could be criticized for the following reason: let us try to calculate the ‘speciﬁc heat’ of a black hole. we must study the collapsing system in diﬀerent coordinate frames. dT (19. The actual value will be considerably larger. Can we write a Schr¨do inger equation for black holes? This question is compounded by the fact that the pure Minkowski vacuum state. because the orbits cannot be straight lines.14) 42 . The cross section for capture in a Schwarzschild black hole can roughly be estimated to be7 2 σ(k) = πr+ = 4πM 2 .9). which is unknown. To understand this situation better.1). What is dU/dT . This is because entropy is always deﬁned apart from an unknown additive constant. Therefore. when transformed into Rindler coordinates. Consider a quantum mechanical description 4 of the process of capturing something.11) (19. and V is the volume of the space where the particle is released. dU = −1/(8πkT 2 ) < 0 . this is a multiplicative constant. which is a mixture of quantum states. This means that the black hole is fundamentally unstable thermally. (19. or. e−βH . and include the consideration that the metric of space and time itself must be subject to quantum oscillations. Now we also know the probability W for emitting a particle. but in the present argument only the order of magnitude is of signiﬁcance. 7 (19. so U = 1/(8πkT ) . This is beyond the scope of this lecture course. Much research is going into identifying these quantum states. But there is another way to derive that the number of black hole microstates is the exponent of 1 × the area Σ of the horizon.10) so the temperature goes down when heat is added.C is an unknown constant. the amount of heat needed to raise the temperature by one degree? The temperature is given by Eq.13) where βH is the inverse Hawking temperature: βH = 1/kTH = 8πM . Does a collapsing system smoothly transmute from a pure quantum state into a mixed state? This we do not believe.

13). (19. Dividing the expressions (19.Now we assume that the process is also coverned by a Schr¨dinger equation.22) But how can we understand the details of this Schr¨dinger equation? Curiously. It is as if there is one Boolean degree of freedom per unit of area A0 of the horizon: = 2A/A0 . which was now deﬁned to be the logarithm of an entropy. (19. dM (M ) = C e4πM 2 /G (19. any black hole. It coincides with the thermodynamic expression (19. The factor v −1 in Eq. this analysis suggests that black holes obey a Schr¨dinger equation describing o the evolution of internal quantum states. According to Hawking’s derivation of the radiation process.17) (19.17) is a kinematical factor. (19. (19. where the states |M BH represent black hole states with mass M/G . (M ) This is easy to integrate: d log (M ) = 8πM/G .9). and we can estimate rather precisely the dimensionality of this internal Hilbert space.16) M | E| |M + GE .21) = eS/k . Clearly. To the contrary. since a parity transformation P and a charge conjugation C have no eﬀect on our calculation of σ .15) (19. one ﬁnds: (M + GE) = eβE = e8πM . the o answer to this question does not appear to follow from any of the ﬁrst principles that have been discussed so-far. using the so-called Fermi Golden Rule.17) and (19. and the factor V −1 in Eq. Time reversal invariance would relate Tin to Tout . To be precise. In terms of these amplitudes. there seems to be a contradiction.18) arises from the normalization of the wave functions. conﬁned to a volume V . A0 = 4G log 2 . This o means that there exist quantum mechanical transition amplitudes. and using Eq. Tin = and Tout = BH BH M + GE| |M BH |E in BH . Thus. (19.18) where (M ) stands for the presumed density of quantum levels of a black hole with mass M . regardless its past.20) (19. we found a direct expression for the density of quantum levels. V (19. the cross section σ and the emission probabilities W can be written as σ = |Tin |2 (M + GE)/v . 1 W = |Tout |2 (M ) .18). all we need is CP T invariance. ends 43 .19) (19. and the states |E are states of surrounding particles with total energy E .

1) where µ = Gm . However. meet each other very close to the horizon. the primary building blocks are the heaviest and most compact forms of matter.e. i. We see that such forms of matter are black holes. According to string theory. they are black holes.up as a thermodynamically mixed state. in this case. A more precise and coherent theoretical approach might lead to further insights. We should search for a more precise analysis. we rewrite as x ds2 = dx2 + 2µ 2 2µ (u · dx)2 + dr . provided that they are not too far separated from the extreme limit. this is a reasonable assumption. since all massive objects must be surrounded by gravitational ﬁelds. This. is diﬃcult and research is in progress. The center-of-mass energy that this encounter represents diverges exponentially with the time lapse. D -brane conﬁgurations form soliton-like conﬁgurations that play the role of black holes. the question of unifying all forces and matter forms ends up in a logical spiral. In fact. For these black holes. So. Consider the schwarzschild metric in the case of a very light mass m : ds2 = dx2 + 2µ 2 (dt + dr2 ) . there have been other developments.21). the distinction between pure states and mixed states for macroscopic objects is pointless. in turn. Black holes should be regarded as being macroscopic. In particular. (20. must be determined by ﬁeld theories describing particles at their horizons. however. Usually. The properties of these black holes.. approximations and simpliﬁcations were made in Hawking’s derivation. the derivation given by Hawking is wanting. There cannot be “other” primary forms of matter. it is easy to observe that particles from the collapse entering the black hole at early times and Hawking particles leaving the black hole at late times. This. The counting appears to conﬁrm the result (19. in and outgoing particles were assumed not to interact with one another. So. and dx2 is the ﬂat metric d˜ 2 − dt2 . the microstates can be counted. notably in string theory. they would merge to form gigantic black holes. r (20. This makes the problem interesting and challenging from a theoretical point of view. but long before that happens. it is very likely that what we perceive as a mixed state is actually a pure state whose details we were unable to resolve. In the mean time. In a uniﬁed theory of all particles and forces.2) . if that is true. From a physical point of view. 20. This is where our procedures should be repaired. The Aechelburg-Sexl metric We wish to ﬁnd the space-time metric surrounding a particle that goes almost with the speed of light towards the positive z -direction. r r 44 r= x2 + (u · x)2 . Such passings cannot go without mutual interactions. so it can easily surpass the mass-energy of the entire universe. our analysis has become invalid. However. Would this also hold for a black hole that started out as a collapsing star in a quantum mechanically pure state? Can pure states evolve into mixed states? Not according to conventional quantum mechanics.

(20. Next.7) can be ignored. the ﬁrst term in Eq. (20. deﬁned as µ y(±) = xµ ± 2µ uµ log r . we have ds2 → dy(−) . p) . since µ is chosen to be small. and at the plane deﬁned by (p · x) = 0 these two ﬂat space-times are glued together according to µ µ y(+) = y(−) + 4pµ log |˜| . 0. u2 = gµν uµ uν = −1 .6) (20. p ﬁxed. one has r → |x · u| . We now claim 2 that when (p · x) > 0 . we use r − (u · x) = 2 (u · x) > λ . we can now easily give this particle a Lorentz boost.4). y. the metric ds2 approaches the ﬂat metric dy(+) . we neglected all eﬀects that are of higher order in the particle’s mass µ . Now consider the limit (20. 0) .4) 1 − v 2 /c2 In the limit µ → 0 . Eq.5) We have 2 dy(±) = dx2 ± 2 ds2 − dy(±) = 4µ dr2 (u · dx)dr − 4µ2 2 . whereas when 2 (p · x) < 0 . 0) are the transverse part of the coordinates y µ . It will turn out to be useful to compare this metric with the ﬂat space-time metric in µ two coordinate frames y± . (20. in region 2 (B) . Thus. First. (20.where u = (1. 0. we divide space-time in three regions: A) B) C) In region (A) . |(u · x)| ≤ λ .10) which is therefore bounded by x . 0.9) x2 . x. r r 2 (20. In the region (C) . Gp = µ.3) In these expressions.7) 2µ d r r (u · x) + 4µ2 (d log r)2 . Similarly. 0. (u · x) < −λ .7) for y(+) is bounded λ µ x by λ2 times a coordinate dependent function (note that r ≥ |˜| ).8) where x = (0. we have that r 45 . x (20. given a small positive number λ . (20. Written this way. ˜ This is seen as follows. (20. r + (u · x) (20. We keep p ﬁxed but let µ tend to zero. (20. we note that the last term of Eq.7) for y(−) will tend to zero as µ/λ . In the boosted frame we can take µv muµ = pµ → (p.

So. with respect to the Kruskal coordinate frame. one gets F (θ) → (1/2π) log(1/θ) . which clearly shows a strong similarity with the case derived earlier for Rindler space. both µλ2 → 0 and µ/λ2 → 0 .14) where (±) now refers to the regions x > 0 and x < 0 . θ. x ≈ 0 . allows us to conclude that 2 A) ds2 → dy(+) 2 B) ds2 → dy(−) C) y(+) = y(−) + 4µ uµ log r if (p · x) ≥ 0 . one can guess in which way the particle that goes in will deform the metric: we cut Kruskal space in halves across the x -axis. Let us therefore choose the Kruskal coordinate frame such that the particle came in at large negative time t . This means that in Eq.s. ϕ) . of this equation happens to be a Green function. (20. and glue the pieces together. At small angular distances θ . For the Schwarzschild observer.16) . x ˜ x ∂ 2 f (˜) = −16πGδ 2 (˜) . its energy is taken to be so small that its gravitational ﬁeld appears to be negligible.8). The function F is yet to be determined.h. one can compute the equations for F . x (20. in (C) . (20. It turns out not to depend on y(−) itself. In view of the result derived above.11) which is equivalent to Eq. This equation. in Kruskal coordinate space. However. one ﬁnds for a source particle moving with the speed of light to the positive z -direction that two ﬂat space-times. (20.7) are bounded by functions of the form µ or µλ2 . (20. This corresponds to a metric with a delta-distributed Riemann curvature on the plane x = 0 . and the black hole mass will hardly be aﬀected by the energy added to it. at (p · x) ≈ 0 . see Eqs. or. one with coordinates (x± . both equations (20.7). the particle moves in along the past horizon. can be solved in an integral form. It is then found that F has to obey ˜ ˜ −∂ 2 F + F = δ 2 (Ω) . The quantity px is the momentum of the particle falling in. Choosing λ such that. again after a shift. Deﬁning x± = z ± t . (6.and (u · x) are both bounded by terms that are ﬁnite or proportional to λ . 46 (20. if (p · x) ≤ 0 . x(−) ) are connected together at the point x(+) = x− . as µ → 0 . δx+ = −pf (˜) . ϕ) .13) (20.15) ˜ ˜ where ∂ 2 is the spherical Laplacian ( + 1) and δ 2 (Ω) the Dirac delta function on the sphere (θ. x x (+) (−) The r.3). (6.12) This result can be generalized to describe a light particle falling into the horizon of a black hole. By demanding that the Ricci curvature must still vanish at the seam. the energy is seen to grow exponentially as Schwarzschild time t progresses. deﬁned by y(+) = y(−) − 16πG px F (y(−) . ˜ (−) in such a way that x(+) = x(−) and ˜ ˜ x+ − x+ = 4Gp+ log |˜| = 8Gp log |˜| . This deﬁnes the Aechelburg-Sexl metric. x(+) ) (+) ˜ ± − and one with coordinates (x(−) .3)–(6.

All theories up to this point. 21. Roger Penrose saw the ﬂaw of that argument: there may be perturbations in the black hole metric in the form of multipole components. • 1916: Karl Schwarzschild publishes his exact spherically symmetric and static solution. covered only non-rotating black holes. • 1924: Eddington introduces coordinates that are well behaved at r = 2M .44 solar masses (the Chandrasekhar limit) would collapse. • 1930: Using general relativity. • 1958: David Finkelstein introduces the concept of the event horizon by presenting Eddington-Finkelstein coordinates. the phrase has been used earlier by others. because time reversal invariance of Nature’s laws would then imply that only perfectly symmetric initial states could collapse gravitationally. Ashtekar 47 . but that it acts as a perfect unidirectional membrane: causal inﬂuences can cross it in only one direction”. • 1960: Penrose introduces global methods in the study of General Relativity. 8 I made use here of notes made by A. not the temperature itself. • 1933: LeMaˆ ıtre realizes the signiﬁcance of Eddington’s result: r = 2M is a ﬁctitious singularity. This may be an important starting point for further investigations of the quantum structure of a black hole. but they all die out or radiate away exponentially. but it will aﬀect the microstates. • 1963: Roy Kerr ﬁnds a generalization of the Schwarzschild metric and interprets it as the ﬁeld of a “spinning particle”. • 1967: John Wheeler uses the words “black hole” in a public lecture. showing a singularity at r = 2M . Unoﬃcially. Subrahmanyan Chandrasekhar calculates that a non-rotating body of electron-degenerate matter above 1. Black hole uniqueness theorems make people believe that black holes cannot form. including Finkelstein’s. • 1960: Kruskal and Szekeres obtain the maximal extension of the Schwarzschild solution. History A brief history of black holes in General Relativity:8 • 1915: Einstein formulates the general theory of relativity. which enabled him to show that ”The Schwarzschild surface R = 2M is not a singularity.It is important to note that this shift in the Kruskal y coordinate aﬀects the Hawking radiation. It does not aﬀect its thermal nature.

Bekenstein proposes that black holes should carry entropy. and usually the black hole is considered in more than 4 dimensional space-time. hence the equilibrium state of every (uncharged) black hole is fully described by only two parameters: mass and angular momentum (represented by M and J ). Bardeen. black body) spectrum of particles. 48 .• late 1960’s .e. with a temperature of kT = κ/2π . Vafa. Quantum ﬁelds on a black hole background space-time radiate thermal (i. Maldacena and others discover how to describe the black hole microstates in terms of D -branes in string theory. time-independent black hole in general relativity is described by the Kerr metric. Carter and Hawking prove the ﬁrst theorems on black hole mechanics. proportional to the horizon area. • 1974: Hawking discovers black hole evaporation. • 1982: Bunting and Mazur independently derive a generalized uniqueness theorem: any isolated. Bardeen. Penrose and Hawking explore the structure and properties of black holes. • 1995: Strominger.early 1970’s: Bekenstein. The description is particularly detailed at or near the extreme limit. Carter.

- Neutrino Less Double Beta Decay
- Neutrino Mass and New Physics
- Standard Model
- Fermion Masses in the Standard Model
- [Michael Dine] Supersymmetry and String Theory Be(BookFi)
- Neutrinos
- Cp-violing phases in type ii seesaw framework
- WIMP Dark Matter as Radiative Neutrino Mass
- Group Theory
- Informal Lectures on Lattice Gauge Theory
- Introduction to Super String Theory by Peskin
- What is Entropy

- Problem 2
- vectoranalysisan015954mbp
- Tens Or
- Tensor
- Tensor
- Class 06 Handout
- Beckman Loffler Report
- Electrodynamics of Massless Charged Particles
- bq2
- m473
- Errata - A Most Incomprehensible Thing
- Some Possible Features of General Expressions for Lovelock Tensors and for the Coefficients of Lovelock Lagrangians Up to the 15th Order in Curvature (and Beyond)
- Summary Sheet I. Tensor Analysis, Geodesics, Covariant Derivatives
- Cft July November 2013
- Tensor Calculus
- Introduction to General Relativity - G. T.hooft
- Stress Energy Tensor - Txy
- Heavens Gr
- Einstein - The Meaning of Relativity
- Chapter 8, Astrophysics
- Solution 1
- Automated delineation of white matter fiber tracts with a multiple region-of-interest approach.
- Tensor Analysis and Curvilinear Coordinates
- Ur Tensors
- Gr Lectures
- Albert Einstein -The Meaning of Relativity
- the Meaning of Relativity
- Decomposing a Tensor
- Brief Review of Tensors
- Proof Epsilon Delta
- BH_lecturenotes[1]

Are you sure?

This action might not be possible to undo. Are you sure you want to continue?

We've moved you to where you read on your other device.

Get the full title to continue

Get the full title to continue reading from where you left off, or restart the preview.

scribd