# GENERAL RELATIVITY &

COSMOLOGY

for Undergraduates

Professor John W. Norbury

Physics Department

University of Wisconsin-Milwaukee

P.O. Box 413

Milwaukee, WI 53201

1997

Contents

1 NEWTONIAN COSMOLOGY 5

1.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5

1.2 Equation of State . . . . . . . . . . . . . . . . . . . . . . . . . 5

1.2.1 Matter . . . . . . . . . . . . . . . . . . . . . . . . . . . 6

1.2.2 Radiation . . . . . . . . . . . . . . . . . . . . . . . . . 6

1.3 Velocity and Acceleration Equations . . . . . . . . . . . . . . 7

1.4 Cosmological Constant . . . . . . . . . . . . . . . . . . . . . . 9

1.4.1 Einstein Static Universe . . . . . . . . . . . . . . . . . 11

2 APPLICATIONS 13

2.1 Conservation laws . . . . . . . . . . . . . . . . . . . . . . . . 13

2.2 Age of the Universe . . . . . . . . . . . . . . . . . . . . . . . 14

2.3 Inﬂation . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15

2.4 Quantum Cosmology . . . . . . . . . . . . . . . . . . . . . . . 16

2.4.1 Derivation of the Schr¨ odinger equation . . . . . . . . . 16

2.4.2 Wheeler-DeWitt equation . . . . . . . . . . . . . . . . 17

2.5 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 18

2.6 Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 19

2.7 Answers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20

2.8 Solutions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 21

3 TENSORS 23

3.1 Contravariant and Covariant Vectors . . . . . . . . . . . . . . 23

3.2 Higher Rank Tensors . . . . . . . . . . . . . . . . . . . . . . . 26

3.3 Review of Cartesian Tensors . . . . . . . . . . . . . . . . . . . 27

3.4 Metric Tensor . . . . . . . . . . . . . . . . . . . . . . . . . . . 28

3.4.1 Special Relativity . . . . . . . . . . . . . . . . . . . . . 30

3.5 Christoﬀel Symbols . . . . . . . . . . . . . . . . . . . . . . . . 31

1

2 CONTENTS

3.6 Christoﬀel Symbols and Metric Tensor . . . . . . . . . . . . . 36

3.7 Riemann Curvature Tensor . . . . . . . . . . . . . . . . . . . 38

3.8 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 39

3.9 Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 40

3.10 Answers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 41

3.11 Solutions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 42

4 ENERGY-MOMENTUM TENSOR 45

4.1 Euler-Lagrange and Hamilton’s Equations . . . . . . . . . . . 45

4.2 Classical Field Theory . . . . . . . . . . . . . . . . . . . . . . 47

4.2.1 Classical Klein-Gordon Field . . . . . . . . . . . . . . 48

4.3 Principle of Least Action . . . . . . . . . . . . . . . . . . . . 49

4.4 Energy-Momentum Tensor for Perfect Fluid . . . . . . . . . . 49

4.5 Continuity Equation . . . . . . . . . . . . . . . . . . . . . . . 51

4.6 Interacting Scalar Field . . . . . . . . . . . . . . . . . . . . . 51

4.7 Cosmology with the Scalar Field . . . . . . . . . . . . . . . . 53

4.7.1 Alternative derivation . . . . . . . . . . . . . . . . . . 55

4.7.2 Limiting solutions . . . . . . . . . . . . . . . . . . . . 56

4.7.3 Exactly Solvable Model of Inﬂation . . . . . . . . . . . 59

4.7.4 Variable Cosmological Constant . . . . . . . . . . . . . 61

4.7.5 Cosmological constant and Scalar Fields . . . . . . . . 63

4.7.6 Clariﬁcation . . . . . . . . . . . . . . . . . . . . . . . . 64

4.7.7 Generic Inﬂation and Slow-Roll Approximation . . . . 65

4.7.8 Chaotic Inﬂation in Slow-Roll Approximation . . . . . 67

4.7.9 Density Fluctuations . . . . . . . . . . . . . . . . . . . 72

4.7.10 Equation of State for Variable Cosmological Constant 73

4.7.11 Quantization . . . . . . . . . . . . . . . . . . . . . . . 77

4.8 Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80

5 EINSTEIN FIELD EQUATIONS 83

5.1 Preview of Riemannian Geometry . . . . . . . . . . . . . . . . 84

5.1.1 Polar Coordinate . . . . . . . . . . . . . . . . . . . . . 84

5.1.2 Volumes and Change of Coordinates . . . . . . . . . . 85

5.1.3 Diﬀerential Geometry . . . . . . . . . . . . . . . . . . 88

5.1.4 1-dimesional Curve . . . . . . . . . . . . . . . . . . . . 89

5.1.5 2-dimensional Surface . . . . . . . . . . . . . . . . . . 92

5.1.6 3-dimensional Hypersurface . . . . . . . . . . . . . . . 96

5.2 Friedmann-Robertson-Walker Metric . . . . . . . . . . . . . . 99

5.2.1 Christoﬀel Symbols . . . . . . . . . . . . . . . . . . . . 101

CONTENTS 3

5.2.2 Ricci Tensor . . . . . . . . . . . . . . . . . . . . . . . . 102

5.2.3 Riemann Scalar and Einstein Tensor . . . . . . . . . . 103

5.2.4 Energy-Momentum Tensor . . . . . . . . . . . . . . . 104

5.2.5 Friedmann Equations . . . . . . . . . . . . . . . . . . 104

5.3 Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105

6 Einstein Field Equations 107

7 Weak Field Limit 109

8 Lagrangian Methods 111

4 CONTENTS

Chapter 1

NEWTONIAN

COSMOLOGY

1.1 Introduction

Many of the modern ideas in cosmology can be explained without the need

to discuss General Relativity. The present chapter represents an attempt to

do this based entirely on Newtonian mechanics. The equations describing

the velocity (called the Friedmann equation) and acceleration of the universe

are derived from Newtonian mechanics and also the cosmological constant

is introduced within a Newtonian framework. The equations of state are

also derived in a very simple way. Applications such as conservation laws,

the age of the universe and the inﬂation, radiation and matter dominated

epochs are discussed.

1.2 Equation of State

In what follows the equation of state for non-relativistic matter and radiation

will be needed. In particular an expression for the rate of change of density,

˙ ρ, will be needed in terms of the density ρ and pressure p. (The deﬁnition

˙ x ≡

dx

dt

, where t is time, is being used.) The ﬁrst law of thermodynamics is

dU +dW = dQ (1.1)

where U is the internal energy, W is the work and Q is the heat transfer.

Ignoring any heat transfer and writing dW = Fdr = pdV where F is the

5

6 CHAPTER 1. NEWTONIAN COSMOLOGY

force, r is the distance, p is the pressure and V is the volume, then

dU = −pdV. (1.2)

Assuming that ρ is a relativistic energy density means that the energy is

expressed as

U = ρV (1.3)

from which it follows that

˙

U = ˙ ρV +ρ

˙

V = −p

˙

V (1.4)

where the term on the far right hand side results from equation (1.2). Writing

V ∝ r

3

implies that

˙

V

V

= 3

˙ r

r

. Thus

˙ ρ = −3(ρ +p)

˙ r

r

(1.5)

1.2.1 Matter

Writing the density of matter as

ρ =

M

4

3

πr

3

(1.6)

it follows that

˙ ρ ≡

dρ

dr

˙ r = −3ρ

˙ r

r

(1.7)

so that by comparing to equation (1.5), it follows that the equation of state

for matter is

p = 0. (1.8)

This is the same as obtained from the ideal gas law for zero temperature.

Recall that in this derivation we have not introduced any kinetic energy, so

we are talking about zero temperature.

1.2.2 Radiation

The equation of state for radiation can be derived by considering radiation

modes in a cavity based on analogy with a violin string [12]. For a standing

wave on a string ﬁxed at both ends

L =

nλ

2

(1.9)

1.3. VELOCITY AND ACCELERATION EQUATIONS 7

where L is the length of the string, λ is the wavelength and n is a positive

integer (n = 1, 2, 3.....). Radiation travels at the velocity of light, so that

c = fλ = f

2L

n

(1.10)

where f is the frequency. Thus substituting f =

n

2L

c into Planck’s formula

U = ¯ hω = hf, where h is Planck’s constant, gives

U =

nhc

2

1

L

∝ V

−1/3

. (1.11)

Using equation (1.2) the pressure becomes

p ≡ −

dU

dV

=

1

3

U

V

. (1.12)

Using ρ = U/V , the radiation equation of state is

p =

1

3

ρ. (1.13)

It is customary to combine the equations of state into the form

p =

γ

3

ρ (1.14)

where γ ≡ 1 for radiation and γ ≡ 0 for matter. These equations of state

are needed in order to discuss the radiation and matter dominated epochs

which occur in the evolution of the Universe.

1.3 Velocity and Acceleration Equations

The Friedmann equation, which speciﬁes the speed of recession, is obtained

by writing the total energy E as the sum of kinetic plus potential energy

terms (and using M =

4

3

πr

3

ρ )

E = T +V =

1

2

m˙ r

2

−G

Mm

r

=

1

2

mr

2

(H

2

−

8πG

3

ρ) (1.15)

where the Hubble constant H ≡

˙ r

r

, m is the mass of a test particle in the

potential energy ﬁeld enclosed by a gas of dust of mass M, r is the distance

from the center of the dust to the test particle and G is Newton’s constant.

8 CHAPTER 1. NEWTONIAN COSMOLOGY

Recall that the escape velocity is just v

escape

=

2GM

r

=

8πG

3

ρr

2

, so that

the above equation can also be written

˙ r

2

= v

2

escape

−k

13 −2 (1.16)

with k

≡ −

2E

m

. The constant k

**can either be negative, zero or positive
**

corresponding to the total energy E being positive, zero or negative. For

a particle in motion near the Earth this would correspond to the particle

escaping (unbound), orbiting (critical case) or returning (bound) to Earth

because the speed ˙ r would be greater, equal to or smaller than the escape

speed v

escape

. Later this will be analagous to an open, ﬂat or closed universe.

Equation (1.15) is re-arranged as

H

2

=

8πG

3

ρ +

2E

mr

2

.13 −3 (1.17)

Deﬁning k ≡ −

2E

ms

2

and writing the distance in terms of the scale factor R

and a constant length s as r(t) ≡ R(t)s, it follows that

˙ r

r

=

˙

R

R

and

¨ r

r

=

¨

R

R

,

giving the Friedmann equation

H

2

≡ (

˙

R

R

)

2

=

8πG

3

ρ −

k

R

2

(1.18)

which speciﬁes the speed of recession. The scale factor is introduced because

in General Relativity it is space itself which expands [19]. Even though this

equation is derived for matter, it is also true for radiation. (In fact it is also

true for vacuum, with Λ ≡ 8πGρ

vac

, where Λ is the cosmological constant

and ρ

vac

is the vacuum energy density which just replaces the ordinary den-

sity. This is discussed later.) Exactly the same equation is obtained from

the general relativistic Einstein ﬁeld equations [13]. According to Guth [10],

k can be rescaled so that instead of being negative, zero or positive it takes

on the values −1, 0 or +1. From a Newtonian point of view this corresponds

to unbound, critical or bound trajectories as mentioned above. From a geo-

metric, general relativistic point of view this corresponds to an open, ﬂat or

closed universe.

In elementary mechanics the speed v of a ball dropped from a height r

is evaluated from the conservation of energy equation as v =

√

2gr, where

g is the acceleration due to gravity. The derivation shown above is exactly

analagous to such a calculation. Similarly the acceleration a of the ball is

calculated as a = g from Newton’s equation F = m¨ r, where F is the force

1.4. COSMOLOGICAL CONSTANT 9

and the acceleration is ¨ r ≡

d

2

r

dt

2

. The acceleration for the universe is obtained

from Newton’s equation

−G

Mm

r

2

= m¨ r.13 −5 (1.19)

Again using M =

4

3

πr

3

ρ and

¨ r

r

=

¨

R

R

gives the acceleration equation

F

mr

≡

¨ r

r

≡

¨

R

R

= −

4πG

3

ρ. (1.20)

However because M =

4

3

πr

3

ρ was used, it is clear that this acceleration

equation holds only for matter. In our example of the falling ball instead of

the acceleration being obtained from Newton’s Law, it can also be obtained

by taking the time derivative of the energy equation to give a =

dv

dt

= v

dv

dr

=

(

√

2gr)(

√

2g

1

2

√

r

) = g. Similarly, for the general case one can take the time

derivative of equation (1.18) (valid for matter and radiation)

d

dt

˙

R

2

= 2

˙

R

¨

R =

8πG

3

d

dt

(ρR

2

). (1.21)

Upon using equation (1.5) the acceleration equation is obtained as

¨

R

R

= −

4πG

3

(ρ + 3p) = −

4πG

3

(1 +γ)ρ (1.22)

which reduces to equation (1.20) for the matter equation of state (γ = 0).

Exactly the same equation is obtained from the Einstein ﬁeld equations [13].

1.4 Cosmological Constant

In both Newtonian and relativistic cosmology the universe is unstable to

gravitational collapse. Both Newton and Einstein believed that the Universe

is static. In order to obtain this Einstein introduced a repulsive gravitational

force, called the cosmological constant, and Newton could have done exactly

the same thing, had he believed the universe to be ﬁnite.

In order to obtain a possibly zero acceleration, a positive term (conven-

tionally taken as

Λ

3

) is added to the acceleration equation (1.22) as

¨

R

R

= −

4πG

3

(ρ + 3p) +

Λ

3

(1.23)

10 CHAPTER 1. NEWTONIAN COSMOLOGY

which, with the proper choice of Λ can give the required zero acceleration

for a static universe. Again exactly the same equation is obtained from the

Einstein ﬁeld equations [13]. What has been done here is entirely equivalent

to just adding a repulsive gravitational force in Newton’s Law. The question

now is how this repulsive force enters the energy equation (1.18). Identifying

the force from

¨ r

r

=

¨

R

R

≡

F

repulsive

mr

≡

Λ

3

(1.24)

and using

F

repulsive

=

Λ

3

mr ≡ −

dV

dr

(1.25)

gives the potential energy as

V

repulsive

= −

1

2

Λ

3

mr

2

(1.26)

which is just a repulsive simple harmonic oscillator. Substituting this into

the conservation of energy equation

E = T +V =

1

2

m˙ r

2

−G

Mm

r

−

1

2

Λ

3

mr

2

=

1

2

mr

2

(H

2

−

8πG

3

ρ −

Λ

3

) (1.27)

gives

H

2

≡ (

˙

R

R

)

2

=

8πG

3

ρ −

k

R

2

+

Λ

3

. (1.28)

Equations (1.28) and (1.23) constitute the fundamental equations of motion

that are used in all discussions of Friedmann models of the Universe. Exactly

the same equations are obtained from the Einstein ﬁeld equations [13].

Let us comment on the repulsive harmonic oscillator obtained above.

Recall one of the standard problems often assigned in mechanics courses.

The problem is to imagine that a hole has been drilled from one side of the

Earth, through the center and to the other side. One is to show that if a

ball is dropped into the hole, it will execute harmonic motion. The solution

is obtained by noting that whereas gravity is an inverse square law for point

masses M and m separated by a distance r as given by F = G

Mm

r

2

, yet if one

of the masses is a continous mass distribution represented by a density then

F = G

4

3

πρmr. The force rises linearly as the distance is increased because

the amount of matter enclosed keeps increasing. Thus the gravitational force

for a continuous mass distribution rises like Hooke’s law and thus oscillatory

solutions are encountered. This sheds light on our repulsive oscillator found

1.4. COSMOLOGICAL CONSTANT 11

above. In this case we want the gravity to be repulsive, but the cosmological

constant acts just like the uniform matter distribution.

Finally authors often write the cosmological constant in terms of a vac-

uum energy density as Λ ≡ 8πGρ

vac

so that the velocity and acceleration

equations become

H

2

≡ (

˙

R

R

)

2

=

8πG

3

ρ −

k

R

2

+

Λ

3

=

8πG

3

(ρ +ρ

vac

) −

k

R

2

(1.29)

and

¨

R

R

= −

4πG

3

(1 +γ)ρ +

Λ

3

= −

4πG

3

(1 +γ)ρ +

8πG

3

ρ

vac

. (1.30)

1.4.1 Einstein Static Universe

Although we have noted that the cosmological constant provides repulsion,

it is interesting to calculate its exact value for a static universe [14, 15]. The

Einstein static universe requires R = R

0

= constant and thus

˙

R =

¨

R = 0.

The case

¨

R = 0 will be examined ﬁrst. From equation (1.23) this requires

that

Λ = 4πG(ρ + 3p) = 4πG(1 +γ)ρ. (1.31)

If there is no cosmological constant (Λ = 0) then either ρ = 0 which is an

empty universe, or p = −

1

3

ρ which requires negative pressure. Both of these

alternatives were unacceptable to Einstein and therefore he concluded that

a cosmological constant was present, i.e. Λ = 0. From equation (1.31) this

implies

ρ =

Λ

4πG(1 +γ)

(1.32)

and because ρ is positive this requires a positive Λ. Substituting equa-

tion (1.32) into equation (1.28) it follows that

Λ =

3(1 +γ)

3 +γ

[(

˙

R

R

0

)

2

+

k

R

2

0

]. (1.33)

Now imposing

˙

R = 0 and assuming a matter equation of state (γ = 0)

implies Λ =

k

R

2

0

. However the requirement that Λ be positive forces k = +1

giving

Λ =

1

R

2

0

= constant. (1.34)

12 CHAPTER 1. NEWTONIAN COSMOLOGY

Thus the cosmological constant is not any old value but rather simply the

inverse of the scale factor squared, where the scale factor has a ﬁxed value

in this static model.

Chapter 2

APPLICATIONS

2.1 Conservation laws

Just as the Maxwell equations imply the conservation of charge, so too do

our velocity and acceleration equations imply conservation of energy. The

energy-momentum conservation equation is derived by setting the covariant

derivative of the energy momentum tensor equal to zero. The same result is

achieved by taking the time derivative of equation (1.29). The result is

˙ ρ + 3(ρ +p)

˙

R

R

= 0. (2.1)

This is identical to equation (1.5) illustrating the intersting connection be-

tweeen thermodynamics and General Relativity that has been discussed re-

cently [16]. The point is that we used thermodynamics to derive our velocity

and acceleration equations and it is no surprise that the thermodynamic for-

mula drops out again at the end. However, the velocity and acceleration

equations can be obtained directly from the Einstein ﬁeld equations. Thus

the Einstein equations imply this thermodynamic relationship in the above

equation.

The above equation can also be written as

d

dt

(ρR

3

) +p

dR

3

dt

= 0 (2.2)

and from equation (1.14), 3(ρ +p) = (3 +γ)ρ, it follows that

d

dt

(ρR

3+γ

) = 0. (2.3)

13

14 CHAPTER 2. APPLICATIONS

Integrating this we obtain

ρ =

c

R

3+γ

(2.4)

where c is a constant. This shows that the density falls as

1

R

3

for matter and

1

R

4

for radiation as expected.

Later we shall use these equations in a diﬀerent form as follows. From

equation (2.1),

ρ

+ 3(ρ +p)

1

R

= 0 (2.5)

where primes denote derivatives with respect to R, i.e. x

≡ dx/dR. Alter-

natively

d

dR

(ρR

3

) + 3pR

2

= 0 (2.6)

so that

1

R

3+γ

d

dR

(ρR

3+γ

) = 0 (2.7)

which is consistent with equation (2.4)

2.2 Age of the Universe

Recent measurements made with the Hubble space telescope [17] have de-

termined that the age of the universe is younger than globular clusters. A

possible resolution to this paradox involves the cosmological constant [18].

We illustrate this as follows.

Writing equation (1.28) as

˙

R

2

=

8πG

3

(ρ +ρ

vac

)R

2

−k (2.8)

the present day value of k is

k =

8πG

3

(ρ

0

+ρ

0vac

)R

2

0

−H

2

0

R

2

0

(2.9)

with H

2

≡ (

˙

R

R

)

2

. Present day values of quantities have been denoted with a

subscript 0. Substituting equation (2.9) into equation (2.8) yields

˙

R

2

=

8πG

3

(ρR

2

−ρ

0

R

2

0

+ρ

vac

R

2

−ρ

0vac

R

2

0

) −H

2

0

R

2

0

. (2.10)

2.3. INFLATION 15

Integrating gives the expansion age

T

0

=

R

0

0

dR

˙

R

=

R

0

0

dR

8πG

3

(ρR

2

−ρ

0

R

2

0

+ρ

vac

R

2

−ρ

0vac

R

2

0

) −H

2

0

R

2

0

.

(2.11)

For the cosmological constant ρ

vac

= ρ

0vac

and because R

2

< R

2

0

then a

non zero cosmological constant will give an age larger than would have been

obtained were it not present. Our aim here is simply to show that the

inclusion of a cosmological constant gives an age which is larger than if no

constant were present.

2.3 Inﬂation

In this section only a ﬂat k = 0 universe will be discussed. Results for

an open or closed universe can easily be obtained and are discussed in the

references [13].

Currently the universe is in a matter dominated phase whereby the dom-

inant contribution to the energy density is due to matter. However the early

universe was radiation dominated and the very early universe was vacuum

dominated. Setting k = 0, there will only be one term on the right hand

side of equation (1.29) depending on what is dominating the universe. For a

matter (γ = 0) or radiation (γ = 1) dominated universe the right hand side

will be of the form

1

R

3+γ

(ignoring vacuum energy), whereas for a vacuum

dominated universe the right hand side will be a constant. The solution

to the Friedmann equation for a radiation dominated universe will thus be

R ∝ t

1

2

, while for the matter dominated case it will be R ∝ t

2

3

. One can see

that these results give negative acceleration, corresponding to a decelerating

expanding universe.

Inﬂation [19] occurs when the vacuum energy contribution dominates the

ordinary density and curvature terms in equation (1.29). Assuming these

are negligible and substituting Λ = constant, results in R ∝ exp(t). The

acceleration is positive, corresponding to an accelerating expanding universe

called an inﬂationary universe.

16 CHAPTER 2. APPLICATIONS

2.4 Quantum Cosmology

2.4.1 Derivation of the Schr¨ odinger equation

The Wheeler-DeWitt equation will be derived in analogy with the 1 dimen-

sional Schr¨ odinger equation, which we derive herein for completeness. The

Lagrangian L for a single particle moving in a potential V is

L = T −V (2.12)

where T =

1

2

m˙ x

2

is the kinetic energy, V is the potential energy. The action

is S =

Ldt and varying the action according to δS = 0 results in the

Euler-Lagrange equation (equation of motion)

d

dt

(

∂L

∂ ˙ x

) −

∂L

∂x

= 0 (2.13)

or just

˙

P =

∂L

∂x

(2.14)

where

P ≡

∂L

∂ ˙ x

. (2.15)

(Note P is the momentum but p is the pressure.) The Hamiltonian H is

deﬁned as

H(P, x) ≡ P ˙ x −L( ˙ x, x). (2.16)

For many situations of physical interest, such as a single particle moving in

a harmonic oscillator potential V =

1

2

kx

2

, the Hamiltonian becomes

H = T +V =

P

2

2m

+V = E (2.17)

where E is the total energy. Quantization is achieved by the operator re-

placements P →

ˆ

P = −i

∂

∂x

and E →

ˆ

E = i

∂

∂t

where we are leaving oﬀ

factors of ¯ h and we are considering the 1-dimensional equation only. The

Schr¨ odinger equation is obtained by writing the Hamiltonian as an operator

ˆ

H acting on a wave function Ψ as in

ˆ

HΨ =

ˆ

EΨ (2.18)

and making the above operator replacements to obtain

(−

1

2m

∂

2

∂x

2

+V )Ψ = i

∂

∂t

Ψ (2.19)

which is the usual form of the 1-dimensional Schr¨odinger equation written

in conﬁguration space.

2.4. QUANTUM COSMOLOGY 17

2.4.2 Wheeler-DeWitt equation

The discussion of the Wheeler-DeWitt equation in the minisuperspace ap-

proximation [20, 21, 11, 22] is usually restricted to closed (k = +1) and

empty (ρ = 0) universes. Atkatz [11] presented a very nice discussion for

closed and empty universes. Herein we consider closed, open and ﬂat and

non-empty universes. It is important to consider the possible presence of

matter and radiation as they might otherwise change the conclusions. Thus

presented below is a derivation of the Wheeler-DeWitt equation in the min-

isuperspace approximation which also includes matter and radiation and

arbitrary values of k.

The Lagrangian is

L = −κR

3

[(

˙

R

R

)

2

−

k

R

2

+

8πG

3

(ρ +ρ

vac

)] (2.20)

with κ ≡

3π

4G

. The momentum conjugate to R is

P ≡

∂L

∂

˙

R

= −κ2R

˙

R. (2.21)

Substituting L and P into the Euler-Lagrange equation,

˙

P −

∂L

∂R

= 0, equa-

tion (1.29) is recovered. (Note the calculation of

∂L

∂R

is simpliﬁed by using

the conservation equation (2.5) with equation (1.14), namely ρ

+ ρ

vac

=

−(3 +γ)ρ/R). The Hamiltonian H ≡ P

˙

R −L is

H(

˙

R, R) = −κR

3

[(

˙

R

R

)

2

+

k

R

2

−

8πG

3

(ρ +ρ

vac

)] ≡ 0 (2.22)

which has been written in terms of

˙

R to show explicitly that the Hamiltonian

is identically zero and is not equal to the total energy as before. (Compare

equation (1.29)). In terms of the conjugate momentum

H(P, R) = −κR

3

[

P

2

4κ

2

R

4

+

k

R

2

−

8πG

3

(ρ +ρ

vac

)] = 0 (2.23)

which, of course is also equal to zero. Making the replacement P → −i

∂

∂R

and imposing HΨ = 0 results in the Wheeler-DeWitt equation in the min-

isuperspace approximation for arbitrary k and with matter or radiation (ρ

term) included gives

{−

∂

2

∂R

2

+

9π

2

4G

2

[(kR

2

−

8πG

3

(ρ +ρ

vac

)R

4

]}Ψ = 0. (2.24)

18 CHAPTER 2. APPLICATIONS

Using equation (2.4) the Wheeler-DeWitt equation becomes

{−

∂

2

∂R

2

+

9π

2

4G

2

[kR

2

−

Λ

3

R

4

−

8πG

3

cR

1−γ

]}Ψ = 0. (2.25)

This just looks like the zero energy Schr¨odinger equation [21] with a potential

given by

V (R) = kR

2

−

Λ

3

R

4

−

8πG

3

cR

1−γ

. (2.26)

For the empty Universe case of no matter or radiation (c = 0) the po-

tential V (R) is plotted in Figure 1 for the cases k = +1, 0, −1 respectively

corresponding to closed [21], open and ﬂat universes. It can be seen that only

the closed universe case provides a potential barrier through which tunnel-

ing can occur. This provides a clear illustration of the idea that only closed

universes can arise through quantum tunneling [22]. If radiation (γ = 1 and

c = 0) is included then only a negative constant will be added to the poten-

tial (because the term R

1−γ

will be constant for γ = 1) and these conclusions

about tunneling will not change. The shapes in Figure 1 will be identical

except that the whole graph will be shifted downwards by a constant with

the inclusion of radiation. (For matter (γ = 0 and c = 0) a term growing

like R will be included in the potential which will only be important for very

small R and so the conclusions again will not be changed.) To summarize,

only closed universes can arise from quantum tunneling even if matter or

radiation are present.

2.5 Summary

2.6. PROBLEMS 19

2.6 Problems

2.1

20 CHAPTER 2. APPLICATIONS

2.7 Answers

2.1

2.8. SOLUTIONS 21

2.8 Solutions

2.1

2.2

22 CHAPTER 2. APPLICATIONS

Chapter 3

TENSORS

3.1 Contravariant and Covariant Vectors

Let us imagine that an ’ordinary’ 2-dimensional vector has components (x, y)

or (x

1

, x

2

) (read as x superscript 2 not x squared) in a certain coordinate

system and components (x, y) or (x

1

, x

2

) when that coordinate system is ro-

tated by angle θ (but with the vector remaining ﬁxed). Then the components

are related by [1]

x

y

=

cos θ sin θ

sin θ cos θ

x

y

(3.1)

Notice that we are using superscipts (x

i

) for the components of our or-

dinary vectors (instead of the usual subscripts used in freshman physics),

which henceforth we are going to name contravariant vectors. We empha-

size that these are just the ordinary vectors one comes across in freshman

physics.

Expanding the matrix equation we have

x = xcos θ +y sin θ (3.2)

y = −xsin θ +y cos θ

from which it follows that

∂x

∂x

= cos θ

∂x

∂y

= sin θ (3.3)

23

24 CHAPTER 3. TENSORS

∂y

∂x

= −sin θ

∂y

∂y

= cos θ

so that

x =

∂x

∂x

x +

∂x

∂y

y (3.4)

y =

∂y

∂x

x +

∂y

∂y

y

which can be written compactly as

x

i

=

∂x

i

∂x

j

x

j

(3.5)

where we will always be using the Einstein summation convention for doubly

repeated indices. (i.e. x

i

y

i

≡

¸

i

x

i

y

i

)

Instead of deﬁning an ordinary (contravariant) vector as a little arrow

pointing in some direction, we shall instead deﬁne it as an object whose com-

ponents transform according to equation(3.5). This is just a fancy version

of equation(3.1), which is another way to deﬁne a vector as what happens

to the components upon rotation (instead of the deﬁnition of a vector as a

little arrow). Notice that we could have written down a diferential version

of (3.5) just from what we know about calculus. Using the inﬁnitessimal dx

i

(instead of x

i

) it follows immediately that

dx

i

=

∂x

i

∂x

j

dx

j

(3.6)

which is identical to (3.5) and therefore we must say that dx

i

forms an

ordinary or contravariant vector (or an inﬁnitessimally tiny arrow).

While we are on the subject of calculus and inﬁnitessimals let’s think

about

∂

∂x

i

which is kind of like the ’inverse’ of dx

i

. From calculus if f =

f(x, y) and x = x(x, y) and y = y(x, y) (which is what (3.3) is saying) then

∂f

∂x

=

∂f

∂x

∂x

∂x

+

∂f

∂y

∂y

∂x

(3.7)

∂f

∂y

=

∂f

∂x

∂x

∂y

+

∂f

∂y

∂y

∂y

or simply

∂f

∂x

i

=

∂f

∂x

j

∂x

j

∂x

i

. (3.8)

3.1. CONTRAVARIANT AND COVARIANT VECTORS 25

Let’s ’remove’ f and just write

∂

∂x

i

=

∂x

j

∂x

i

∂

∂x

j

. (3.9)

which we see is similar to (3.5), and so we might expect that ∂/∂x

i

are

the ’components’ of a ’non-ordinary’ vector. Notice that the index is in the

denominator, so instead of writing ∂/∂x

i

let’s just always write it as x

i

for

shorthand. Or equivalently deﬁne

x

i

≡

∂

∂x

i

(3.10)

Thus

x

i

=

∂x

j

∂x

i

x

j

. (3.11)

So now let’s deﬁne a contravariant vector A

µ

as anything whose components

transform as (compare (3.5))

A

µ

≡

∂x

µ

∂x

ν

A

ν

(3.12)

and a covariant vector A

µ

(often also called a one-form, or dual vector or

covector)

A

µ

=

∂x

ν

∂x

µ A

ν

(3.13)

In calculus we have two fundamental objects dx

i

and the dual vector ∂/∂x

i

.

If we try to form the dual dual vector ∂/∂(∂/∂x

i

) we get back dx

i

[2]. A

set of points in a smooth space is called a manifold and where dx

i

forms a

space, ∂/∂x

i

forms the corresponding ’dual’ space [2]. The dual of the dual

space is just the original space dx

i

. Contravariant and covariant vectors are

the dual of each other. Other examples of dual spaces are row and column

matrices (x y) and

x

y

**and the kets < a| and bras |a > used in quantum
**

mechanics [3].

Before proceeding let’s emphasize again that our deﬁnitions of contravari-

ant and covariant vectors in (3.13) and (3.13) are nothing more than fancy

versions of (3.1).

26 CHAPTER 3. TENSORS

3.2 Higher Rank Tensors

Notice that our vector components A

µ

have one index, whereas a scalar

(e.g. t = time or T = temperature) has zero indices. Thus scalars are called

tensors of rank zero and vectors are called tensors of rank one. We are

familiar with matrices which have two indices A

ij

. A contravariant tensor of

rank two is of the form A

µν

, rank three A

µνγ

etc. A mixed tensor, e.g. A

µ

ν

,

is partly covariant and partly contravariant.

In order for an object to be called a tensor it must satisfy the tensor

transformation rules, examples of which are (3.13) and (3.13) and

T

µν

=

∂x

µ

∂x

α

∂x

ν

∂x

β

T

αβ

.

(3.14)

T

µ

ν

=

∂x

µ

∂x

α

∂x

β

∂x

ν

T

α

β

. (3.15)

T

µν

ρ

=

∂x

µ

∂x

α

∂x

ν

∂x

β

∂x

γ

∂x

ρ

T

αβ

γ

. (3.16)

Thus even though a matrix has two indices A

ij

, it may not necessarily be

a second rank tensor unless it satisﬁes the above tensor tranformation rules

as well. However all second rank tensors can be written as matrices.

Higher rank tensors can be constructed from lower rank ones by forming

what is called the outer product or tensor product [14] as follows. For instance

T

α

β

≡ A

α

B

β

(3.17)

or

T

αβ

γδ

≡ A

α

γ

B

β

δ

. (3.18)

The tensor product is often written simply as

T = A⊗B (3.19)

(do Problem 3.1) (NNN Next time discuss wedge product - easy - just

introduce antisymmetry).

We can also construct lower rank tensors from higher rank ones by a

process called contraction, which sets a covariant and contravariant index

equal, and because of the Einstein summation convention equal or repeated

3.3. REVIEW OF CARTESIAN TENSORS 27

indices are summed over. Thus contraction represents setting two indices

equal and summing. For example

T

αβ

γβ

≡ T

α

γ

(3.20)

Thus contraction over a pair of indices reduces the rank of a tensor by two

[14].

The inner product [14] of two tensors is deﬁned by forming the outer

product and then contracting over a pair of indices as

T

α

β

≡ A

α

γ

B

γ

β

. (3.21)

Clearly the inner product of two vectors (rank one tensors) produces a scalar

(rank zero tensor) as

A

µ

B

µ

= constant ≡ A.B (3.22)

and it can be shown that A.B as deﬁned here is a scalar (do Problem 3.2).

A scalar is a tensor of rank zero with the very special transformation law of

invariance

c = c. (3.23)

It is easily shown, for example, that A

µ

B

µ

is no good as a deﬁnition of inner

product for vectors because it is not invariant under transformations and

therefore is not a scalar.

3.3 Review of Cartesian Tensors

Let us review the scalar product that we used in freshman physics. We wrote

vectors as A = A

i

ˆ e

i

and deﬁned the scalar product as

A.B ≡ ABcos θ (3.24)

where A and B are the magnitudes of the vectors A and B and θ is the

angle between them. Thus

A.B = A

i

ˆ e

i

.B

j

ˆ e

j

= (ˆ e

i

.ˆ e

j

)A

i

B

j

≡ g

ij

A

i

B

j

(3.25)

28 CHAPTER 3. TENSORS

where the metric tensor g

ij

is deﬁned as the dot product of the basis vectors.

A Cartesian basis is deﬁned as one in which g

ij

≡ δ

ij

(obtained from

ˆ e

i

.ˆ e

j

= |ˆ e

i

||ˆ e

j

| cos θ = cos θ = δ

ij

). That is, the basis vectors are of unit

length and perpendicular to each other in which case

A.B = A

i

B

i

= A

x

B

x

+A

y

B

y

+.... (3.26)

where the sum (+...) extends to however many dimensions are being consid-

ered and

A.A ≡ A

2

= A

i

A

i

(3.27)

which is just Pythagoras’ theorem, A.A ≡ A

2

= A

i

A

i

= A

2

x

+A

2

y

+.......

Notice that the usual results we learned about in freshman physics, equa-

tions (3.26) and (3.27), result entirely from requiring g

ij

= δ

ij

=

1 0

0 1

**in matrix notation.
**

We could easily have deﬁned a non-Cartesian space, for example, g

ij

=

1 1

0 1

**in which case Pythagoras’ theorem would change to
**

A.A ≡ A

2

= A

i

A

i

= A

2

x

+A

2

y

+A

x

A

y

. (3.28)

Thus it is the metric tensor g

ij

≡ ˆ e

i

.ˆ e

j

given by the scalar product of the

unit vectors which (almost) completely deﬁnes the vector space that we are

considering. Now let’s return to vectors and one-forms (i.e. contravariant

and covariant vectors).

3.4 Metric Tensor

We have already seen (in Problem 3.2) that the inner product deﬁned by

A.B ≡ A

µ

B

µ

transforms as a scalar. (The choice A

µ

B

µ

won’t do because

it is not a scalar). However based on the previous section, we would expect

that A.B can also be written in terms of a metric tensor. The most natural

way to do this is

A.B ≡ A

µ

B

µ

= g

µν

A

ν

B

µ

(3.29)

assuming g

µν

is a tensor.

3.4. METRIC TENSOR 29

In fact deﬁning A.B ≡ A

µ

B

µ

≡ g

µν

A

ν

B

µ

makes perfect sense because it

also transforms as a scalar (i.e. is invariant). (do Problem 3.3) Thus either

of the two right hand sides of (3.29) will do equally well as the deﬁnition of

the scalar product, and thus we deduce that

A

µ

= g

µν

A

ν

(3.30)

so that the metric tensor has the eﬀect of lowering indices. Similarly it can

raise indices

A

µ

= g

µν

A

ν

(3.31)

How is vector A written in terms of basis vectors ? Based on our expe-

rience with Cartesian vectors let’s deﬁne our basis vectors such that

A.B ≡ A

µ

B

µ

= g

µν

A

ν

B

µ

≡ (e

µ

.e

ν

)A

ν

B

µ

(3.32)

which imples that vectors can be written in terms of components and basis

vectors as

A = A

µ

e

µ

= A

µ

e

µ

(3.33)

Thus the basis vectors of a covariant vector (one-form) transform as con-

travariant vectors. Contravariant components have basis vectors that trans-

form as one-froms [5] (pg. 63-64).

The above results illuminate our ﬂat (Cartesian) space results where

g

µν

≡ δ

µν

, so that (3.31) becomes A

µ

= A

µ

showing that in ﬂat space there

is no distinction between covariant and contravariant vectors. Because of

this it also follows that A = A

µ

ˆ e

µ

and A.B = A

µ

B

µ

which were our ﬂat

space results.

Two more points to note are the symmetry

g

µν

= g

νµ

(3.34)

30 CHAPTER 3. TENSORS

and the inverse deﬁned by

g

µα

g

αν

= δ

ν

µ

= g

ν

µ

(3.35)

so that g

ν

µ

is the Kronecker delta. This follows by getting back what we start

with as in A

µ

= g

µν

A

ν

= g

µν

g

να

A

α

≡ δ

α

µ

A

α

.

3.4.1 Special Relativity

Whereas the 3-dimensional Cartesian space is completely characterized by

g

µν

= δ

µν

or

g

µν

=

¸

¸

1 0 0

0 1 0

0 0 1

¸

(3.36)

Obviously for unit matrices there is no distinction between δ

ν

µ

and δ

µν

. The

4-dimensional spacetime of special relativity is speciﬁed by

η

µν

=

¸

¸

¸

¸

1 0 0 0

0 −1 0 0

0 0 −1 0

0 0 0 −1

¸

(3.37)

If a contravariant vector is speciﬁed by

A

µ

= (A

0

, A

i

) = (A

0

, A) (3.38)

it follows that the covariant vector is A

µ

= η

µν

A

ν

or

A

µ

= (A

0

, A

i

) = (A

0

, −A) (3.39)

Note that A

0

= A

0

.

Exercise: Prove equation (3.39) using (3.38) and (3.37).

Thus, for example, the energy momentum four vector p

µ

= (E, p) gives

p

2

= E

2

− p

2

. Of course p

2

is the invariant we identify as m

2

so that

E

2

= p

2

+m

2

.

Because of equation (3.38) we must have

∂

µ

≡

∂

∂x

µ

= (

∂

∂x

0

, ) = (

∂

∂t

, ) (3.40)

implying that

∂

µ

≡

∂

∂x

µ

= (

∂

∂x

0

, −) = (

∂

∂t

, −) (3.41)

3.5. CHRISTOFFEL SYMBOLS 31

Note that ∂

0

= ∂

0

=

∂

∂t

(with c ≡ 1). We deﬁne

P

2

≡ ∂

µ

∂

µ

= ∂

0

∂

0

+∂

i

∂

i

= ∂

0

∂

0

−∂

i

∂

i

=

∂

2

∂x

0

2

−

2

=

∂

2

∂t

2

−

2

(3.42)

(Note that some authors [30] instead deﬁne P

2

≡

2

−

∂

2

∂t

2

).

Let us now brieﬂy discuss the fourvelocity u

µ

and proper time. We shall

write out c explicitly here.

Using dx

µ

≡ (cdt, dx) the invariant interval is

ds

2

≡ dx

µ

dx

µ

= c

2

dt

2

−dx

2

. (3.43)

The proper time τ is deﬁned via

ds ≡ cdτ =

cdt

γ

(3.44)

which is consistent with the time dilation eﬀect as the proper time is the

time measured in an observer’s rest frame. The fourvelocity is deﬁned as

u

µ

≡

dx

µ

dτ

≡ (γc, γv) (3.45)

such that the fourmomentum is

p

µ

≡ mu

µ

= (

E

c

, p) (3.46)

where m is the rest mass.

Exercise: Check that (mu

µ

)

2

= m

2

c

2

. (This must be true so that E

2

=

(pc)

2

+ (mc

2

)

2

).

3.5 Christoﬀel Symbols

Some good references for this section are [7, 14, 8]. In electrodynamics in

ﬂat spacetime we encounter

E = −

φ (3.47)

and

B =

×A (3.48)

32 CHAPTER 3. TENSORS

where E and Bare the electric and magnetic ﬁelds and φ and Aare the scalar

and vectors potentials.

is the gradient operator deﬁned (in 3 dimensions)

as

≡

ˆ

i∂/∂x +

ˆ

j∂/∂y +

ˆ

k∂/∂z

= ˆ e

1

∂/∂x

1

+ ˆ e

2

∂/∂x

2

+ ˆ e

3

∂/∂x

3

. (3.49)

Clearly then φ and A are functions of x, y, z, i.e. φ = φ(x, y, z) and

A = A(x, y, z). Therefore φ is called a scalar ﬁeld and A is called a vector

ﬁeld. E and B are also vector ﬁelds because their values are a function of

position also. (The electric ﬁeld of a point charge gets smaller when you

move away.) Because the left hand sides are vectors, (3.47) and (3.48) imply

that the derivatives

φ and

×A also transform as vectors. What about

the derivative of tensors in our general curved spacetime ? Do they also

transform as tensors ?

Consider a vector ﬁeld A

µ

(x

ν

) as a function of contravariant coordinates.

Let us introduce a shorthand for the derivative as

A

µ,ν

≡

∂A

µ

∂x

ν

(3.50)

We want to know whether the derivative A

µ,ν

is a tensor. That is does A

µ,ν

transform according to A

µ,ν

=

∂x

α

∂x

µ

∂x

β

∂x

ν A

α,β

. ? To ﬁnd out, let’s evaluate the

derivative explicitly

A

µ,ν

≡

∂A

µ

∂x

ν

=

∂

∂x

ν

(

∂x

α

∂x

µ

A

α

)

=

∂x

α

∂x

µ

∂A

α

∂x

ν

+

∂

2

x

α

∂x

ν

∂x

µ

A

α

(3.51)

but A

α

is a function of x

ν

not x

ν

, i.e. A

α

= A

α

(x

ν

) = A

α

(x

ν

) . Therefore

we must insert

∂Aα

∂x

ν =

∂Aα

∂x

γ

∂x

γ

∂x

ν so that

A

µ,ν

≡

∂A

µ

∂x

ν

=

∂x

α

∂x

µ

∂x

γ

∂x

ν

∂A

α

∂x

γ

+

∂

2

x

α

∂x

ν

∂x

µ

A

α

=

∂x

α

∂x

µ

∂x

γ

∂x

ν

A

α,γ

+

∂

2

x

α

∂x

ν

∂x

µ

A

α

(3.52)

We see therefore that the tensor transformation law for A

µ,ν

is spoiled by

the second term. Thus A

µ,ν

is not a tensor [8, 7, 14].

3.5. CHRISTOFFEL SYMBOLS 33

To see why this problem occurs we should look at the deﬁnition of the

derivative [8],

A

µ,ν

≡

∂A

µ

∂x

ν

= lim

dx→0

A

µ

(x +dx) −A

µ

(x)

dx

ν

(3.53)

or more properly [7, 14] as lim

dx

γ

→0

Aµ(x

γ

+dx

γ

)−Aµ(x

γ

)

dx

ν

.

The problem however with (3.53) is that the numerator is not a vector

because A

µ

(x +dx) and A

µ

(x) are located at diﬀerent points. The diﬀernce

between two vectors is only a vector if they are located at the same point.

The diﬀerence betweeen two vectors located at separate points is not a vector

because the transformations laws (3.12) and (3.13) depend on position. In

freshman physics when we represent two vectors A and B as little arrows,

the diﬀerence A − B is not even deﬁned (i.e. is not a vector) if A and B

are at diﬀerent points. We ﬁrst instruct the freshman student to slide one

of the vectors to the other one and only then we can visualize the diﬀerence

between them. This sliding is achieved by moving one of the vectors parallel

to itself (called parallel transport), which is easy to do in ﬂat space. Thus

to compare two vectors (i.e. compute A−B) we must ﬁrst put them at the

same spacetime point.

Thus in order to calculate A

µ

(x +dx) −A

µ

(x) we must ﬁrst deﬁne what

is meant by parallel transport in a general curved space. When we parallel

transport a vector in ﬂat space its components don’t change when we move

it around, but they do change in curved space. Imagine standing on the

curved surface of the Earth, say in Paris, holding a giant arrow (let’s call

this vector A) vertically upward. If you walk from Paris to Moscow and keep

the arrow pointed upward at all times (in other words transport the vector

parallel to itself), then an astronaut viewing the arrow from a stationary

position in space will notice that the arrow points in diﬀerent directions in

Moscow compared to Paris, even though according to you, you have par-

allel transported the vector and it still points vertically upward from the

Earth. Thus the astronaut sees the arrow pointing in a diﬀerent direction

and concludes that it is not the same vector. (It can’t be because it points

diﬀerently; it’s orientation has changed.) Thus parallel transport produces a

diﬀerent vector. Vector A has changed into a diﬀerent vector C.

To ﬁx this situation, the astronaut communicates with you by radio and

views your arrow through her spacecraft window. She makes a little mark

on her window to line up with your arrow in Paris. She then draws a whole

series of parallel lines on her window and as you walk from Paris to Moscow

she keeps instructing you to keep your arrow parallel to the lines on her

34 CHAPTER 3. TENSORS

window. When you get to Moscow, she is satisﬁed that you haven’t rotated

your arrow compared to the markings on her window. If a vector is parallel

transported from an ’absolute’ point of view (the astronaut’s window), then

it must still be the same vector A, except now moved to a diﬀerent point

(Moscow).

Let’s denote δA

µ

as the change produced in vector A

µ

(x

α

) located at x

α

by an inﬁnitessimal parallel transport by a distance dx

α

. We expect δA

µ

to

be directly proportional to dx

α

.

δA

µ

∝ dx

α

(3.54)

We also expect δA

µ

to be directly proportional to A

µ

; the bigger our arrow,

the more noticeable its change will be. Thus

δA

µ

∝ A

ν

dx

α

(3.55)

The only sensible constant of proportionality will have to have covariant µ

and α indices and a contravariant ν index as

δA

µ

≡ Γ

ν

µα

A

ν

dx

α

(3.56)

where Γ

ν

µα

are called Christoﬀel symbols or coeﬃcients of aﬃne connection

or simply connection coeﬃcients . As Narlikar [7] points out, whereas the

metric tensor tells us how to deﬁne distance betweeen neighboring points, the

connection coeﬃcients tell us how to deﬁne parallelism betweeen neighboring

points.

Equation (3.56) deﬁnes parallel transport. δA

µ

is the change produced

in vector A

µ

by an inﬁnitessimal transport by a distance dx

α

to produce a

new vector C

µ

≡ A

µ

+δA

µ

. To obtain parallel transport for a contravariant

vector B

µ

note that a scalar deﬁned as A

µ

B

µ

cannot change under parallel

transport. Thus [8]

δ(A

µ

B

µ

) = 0 (3.57)

from which it follows that (do Problem 3.4)

δA

µ

≡ −Γ

µ

να

A

ν

dx

α

. (3.58)

We shall also assume [8] symmetry under exchange of lower indices,

Γ

α

µν

= Γ

α

νµ

. (3.59)

(We would have a truly crazy space if this wasn’t true [8]. Think about it !)

3.5. CHRISTOFFEL SYMBOLS 35

Continuing with our consideration of A

µ

(x

α

) parallel transported an in-

ﬁnitessimal distance dx

α

, the new vector C

µ

will be

C

µ

= A

µ

+δA

µ

. (3.60)

whereas the old vector A

µ

(x

α

) at the new position x

α

+dx

α

will be A

µ

(x

α

+

dx

α

) . The diﬀerence betweeen them is

dA

µ

= A

µ

(x

α

+dx

α

) −[A

µ

(x

α

) +δA

µ

] (3.61)

which by construction is a vector. Thus we are led to a new deﬁnition of

derivative (which is a tensor [8])

A

µ;ν

≡

dA

µ

dx

ν

= lim

dx→0

A

µ

(x +dx) −[A

µ

(x) +δA

µ

]

dx

ν

(3.62)

Using (3.53) in (3.61) we have dA

µ

=

∂Aµ

∂x

ν

dx

ν

−δA

µ

=

∂Aµ

∂x

ν

dx

ν

−Γ

µα

A

dx

α

and (3.62) becomes A

µ;ν

≡

dAµ

dx

ν

=

∂Aµ

∂x

ν

− Γ

µν

A

(because

dx

α

dx

ν

= δ

α

ν

) which

we shall henceforth write as

A

µ;ν

≡ A

µ,ν

−Γ

µν

A

(3.63)

where A

µ,ν

≡

∂Aµ

∂x

ν

. The derivative A

µ;ν

is often called the covariant deriva-

tive (with the word covariant not meaning the same as before) and one can

easily verify that A

µ;ν

is a second rank tensor (which will be done later in

Problem 3.5). From (3.58)

A

µ

;ν

≡ A

µ

,ν

+ Γ

µ

ν

A

(3.64)

For tensors of higher rank the results are, for example, [14, 8]

A

µν

;λ

≡ A

µν

,λ

+ Γ

µ

λ

A

ν

+ Γ

ν

λ

A

µ

36 CHAPTER 3. TENSORS

(3.65)

and

A

µν;λ

≡ A

µν,λ

−Γ

µλ

A

ν

−Γ

νλ

A

µ

(3.66)

and

A

µ

ν;λ

≡ A

µ

ν,λ

+ Γ

µ

λ

A

ν

−Γ

νλ

A

µ

(3.67)

and

A

µν

αβ;λ

≡ A

µν

αβ,λ

+ Γ

µ

λ

A

ν

αβ

+ Γ

ν

λ

A

µ

αβ

−Γ

αλ

A

µν

β

−Γ

βλ

A

µν

α

. (3.68)

3.6 Christoﬀel Symbols and Metric Tensor

We shall now derive an important formula which gives the Christoﬀel symbol

in terms of the metric tensor and its derivatives [8, 14, 7]. The formula is

Γ

α

βγ

=

1

2

g

α

(g

β,γ

+g

γ,β

−g

βγ,

).

(3.69)

Another result we wish to prove is that

Γ

µ

= (ln

√

−g)

,µ

=

1

2

[ln(−g)]

,µ

(3.70)

where

g ≡ determinant|g

µν

|. (3.71)

Note that g = |g

µν

|. Let us now prove these results.

Proof of Equation (3.69). The process of covariant diﬀerentiation should

never change the length of a vector. To ensure this means that the covariant

derivative of the metric tensor should always be identically zero,

g

µν;λ

≡ 0. (3.72)

Applying (3.66)

g

µν;λ

≡ g

µν,λ

−Γ

µλ

g

ν

−Γ

νλ

g

µ

≡ 0 (3.73)

3.6. CHRISTOFFEL SYMBOLS AND METRIC TENSOR 37

Thus

g

µν,λ

= Γ

µλ

g

ν

+ Γ

νλ

g

µ

(3.74)

and permuting the µνλ indices cyclically gives

g

λµ,ν

= Γ

λν

g

µ

+ Γ

µν

g

λ

(3.75)

and

g

νλ,µ

= Γ

νµ

g

λ

+ Γ

λµ

g

ν

(3.76)

Now add (3.75) and (3.76) and subtract (3.74) gives [8]

g

λµ,ν

+g

νλ,µ

−g

µν,λ

= 2Γ

µν

g

λ

(3.77)

because of the symmetries of (3.59) and (3.34). Multiplying (3.77) by g

λα

and using (3.34) and (3.35) (to give g

λ

g

λα

= g

λ

g

λα

= δ

α

) yields

Γ

α

µν

=

1

2

g

λα

(g

λµ,ν

+g

νλ,µ

−g

µν,λ

). (3.78)

which gives (3.69). (do Problems 3.5 and 3.6).

Proof of equation (3.70) [14] (Appendix II) Using g

α

g

β,α

= g

α

g

βα,

(obtained using the symmetry of the metric tensor and swapping the names

of indices) and contracting over αν, equation (3.69) becomes (ﬁrst and last

terms cancel)

Γ

α

β,α

=

1

2

g

α

(g

β,α

+g

α,β

−g

βα,

).

=

1

2

g

α

g

α,β

(3.79)

Deﬁning g as the determinant |g

µν

| and using (3.35) it follows that

∂g

∂g

µν

= gg

µν

(3.80)

a result which can be easily checked. (do Problem 3.7) Thus (3.79) be-

comes

Γ

α

β,α

=

1

2g

∂g

∂g

λα

∂g

λα

∂x

β

=

1

2g

∂g

∂x

β

=

1

2

∂ ln g

∂x

β

(3.81)

which is (3.70), where in (3.70) we write ln(−g) instead of ln g because g is

always negative.

38 CHAPTER 3. TENSORS

3.7 Riemann Curvature Tensor

The Riemann curvature tensor is one of the most important tensors in gen-

eral relativity. If it is zero then it means that the space is ﬂat. If it is

non-zero then we have a curved space. This tensor is most easily derived

by considering the order of double diﬀerentiation on tensors [28, 2, 9, 7, 8].

Firstly we write in general

A

µ

,αβ

≡

∂

2

A

µ

∂x

α

∂x

β

(3.82)

and also when we write A

µ

;αβ

we again mean second derivative. Many authors

instead write A

µ

,αβ

≡ A

µ

,α,β

or A

µ

;αβ

≡ A

µ

;α;β

. We shall use either notation.

In general it turns out that even though A

µ

,αβ

= A

µ

,βα

, however in general

it is true that A

µ

;αβ

= A

µ

;βα

. Let us examine this in more detail. Firstly

consider the second derivative of a scalar φ. A scalar does not change under

parallel transport therefore φ

;µ

= φ

,µ

. From (3.63) we have (φ

;µ

is a tensor,

not a scalar)

φ

;µ;ν

= φ

,µ;ν

= φ

,µ,ν

−Γ

µν

φ

,

(3.83)

but because Γ

µν

= Γ

νµ

it follows that φ

;µν

= φ

;νµ

meaning that the order

of diﬀerentiation does not matter for a scalar. Consider now a vector. Let’s

diﬀerentiate equation (3.64). Note that A

µ

;ν

is a second rank tensor, so we

use (3.67) as follows

A

µ

;ν;λ

= A

µ

;ν,λ

+ Γ

µ

λ

A

;ν

−Γ

νλ

A

µ

;

=

∂

∂x

λ

(A

µ

;ν

) + Γ

µ

λ

A

;ν

−Γ

νλ

A

µ

;

= A

µ

,ν,λ

+ Γ

µ

ν,λ

A

+ Γ

µ

ν

A

,λ

+ Γ

µ

λ

A

;ν

−Γ

νλ

A

µ

;

(3.84)

Now interchange the order of diﬀerentiation (just swap the ν and λ indices)

A

µ

;λ;ν

= A

µ

,λ,ν

+ Γ

µ

λ,ν

A

+ Γ

µ

λ

A

,ν

+ Γ

µ

ν

A

;λ

−Γ

λν

A

µ

;

(3.85)

Subtracting we have

A

µ

;ν;λ

−A

µ

;λ;ν

= A

(Γ

µ

ν,λ

−Γ

µ

λ,ν

+ Γ

µ

λθ

Γ

θ

ν

−Γ

µ

νθ

Γ

θ

λ

)

≡ A

R

µ

λν

(3.86)

with the famous Riemann curvature tensor deﬁned as

3.8. SUMMARY 39

R

α

βγδ

≡ −Γ

α

βγ,δ

+ Γ

α

βδ,γ

+ Γ

α

γ

Γ

βδ

−Γ

α

δ

Γ

βγ

(3.87)

Exercise: Check that equations (3.86) and (3.87) are consistent.

The Riemann tensor tells us everything essential about the curvature of

a space. For a Cartesian spcae the Riemann tensor is zero.

The Riemann tensor has the following useful symmetry properties [9]

R

α

βγδ

= −R

α

βδγ

(3.88)

R

α

βγδ

+R

α

γδβ

+R

α

δβγ

= 0 (3.89)

and

R

αβγδ

= −R

βαγδ

(3.90)

All other symmetry properties of the Riemann tensor may be obtained from

these. For example

R

αβγδ

= R

γδαβ

(3.91)

Finally we introduce the Ricci tensor [9] by contracting on a pair of indices

R

αβ

≡ R

αβ

(3.92)

which has the property

R

αβ

= R

βα

(3.93)

(It will turn out later that R

αβ

= 0 for empty space [9] ). Note that the

contraction of the Riemann tensor is unique up to a sign, i.e. we could have

deﬁned R

αβ

or R

αβ

or R

αβ

as the Ricci tensor and we would have the

same result except that maybe a sign diﬀernce would appear. Thus diﬀerent

books may have this sign diﬀerence.

However all authors agree on the deﬁnition of the Riemann scalar (ob-

tained by contracting R

α

β

)

R ≡ R

α

α

≡ g

αβ

R

αβ

(3.94)

Finally the Einstein tensor is deﬁned as

G

µν

≡ R

µν

−

1

2

Rg

µν

(3.95)

After discussing the stress-energy tensor in the next chapter, we shall put

all of this tensor machinery to use in our discussion of general relativity

following.

3.8 Summary

40 CHAPTER 3. TENSORS

3.9 Problems

3.1 If A

µ

and B

ν

are tensors, show that the tensor product (outer product)

deﬁned by T

µ

ν

≡ A

µ

B

ν

is also a tensor.

3.2 Show that the inner product A.B ≡ A

µ

B

µ

is invariant under transfor-

mations, i.e. show that it satisﬁes the tensor transformation law of a scalar

(thus it is often called the scalar product).

3.3 Show that the inner product deﬁned by A.B ≡ g

µν

A

µ

B

ν

is also a scalar

(invariant under transformations), where g

µν

is assumed to be a tensor.

3.4 Prove equation (3.58).

3.5 Derive the transformation rule for Γ

α

βγ

. Is Γ

α

βγ

a tensor ?

3.6 Show that A

µ;ν

is a second rank tensor.

3.7 Check that

∂g

∂gµν

= gg

µν

. (Equation (3.80)).

3.10. ANSWERS 41

3.10 Answers

no answers; only solutions

42 CHAPTER 3. TENSORS

3.11 Solutions

3.1

To prove that T

µ

ν

is a tensor we must show that it satisﬁes the

tensor transformation law T

µ

ν

=

∂x

µ

∂x

α

∂x

β

∂x

ν T

α

β

.

Proof T

µ

ν

= A

µ

B

ν

=

∂x

µ

∂x

α

A

α ∂x

β

∂x

ν B

β

=

∂x

µ

∂x

α

∂x

β

∂x

ν A

α

B

β

=

∂x

µ

∂x

α

∂x

β

∂x

ν T

α

β

QED.

3.2

First let’s recall that if f = f(θ, α) and θ = θ(x, y) and α =

α(x, y) then

∂f

∂θ

=

∂f

∂x

∂x

∂θ

+

∂f

∂y

∂y

∂θ

=

∂f

∂x

i

∂x

i

∂θ

.

Now

A.B = A

µ

B

µ

=

∂x

µ

∂x

α

A

α ∂x

β

∂x

µ B

β

=

∂x

µ

∂x

α

∂x

β

∂x

µ A

α

B

β

=

∂x

β

∂x

α

A

α

B

β

by the chain rule

= δ

β

α

A

α

B

β

= A

α

B

α

= A.B

3.11. SOLUTIONS 43

3.3

A.B ≡ g

µν

A

µ

B

ν

=

∂x

α

∂x

µ

∂x

β

∂x

ν g

αβ

∂x

µ

∂x

γ

∂x

ν

∂x

δ

A

γ

B

δ

=

∂x

α

∂x

µ

∂x

β

∂x

ν

∂x

µ

∂x

γ

∂x

ν

∂x

δ

g

αβ

A

γ

B

δ

=

∂x

α

∂x

γ

∂x

β

∂x

δ

g

αβ

A

γ

B

δ

= δ

α

γ

δ

β

δ

g

αβ

A

γ

B

δ

= g

αβ

A

α

B

β

= A

α

B

α

= A.B

3.4

3.5

3.6

3.7

44 CHAPTER 3. TENSORS

Chapter 4

ENERGY-MOMENTUM

TENSOR

It is important to emphasize that our discussion in this chapter is based

entirely on Special Relativity.

4.1 Euler-Lagrange and Hamilton’s Equations

Newton’s second law of motion is

F =

dp

dt

(4.1)

or in component form (for each component F

i

)

F

i

=

dp

i

dt

(4.2)

where p

i

= m˙ q

i

(with q

i

being the generalized position coordinate) so that

dp

i

dt

= ˙ m˙ q

i

+ m¨ q

i

. If ˙ m = 0 then F

i

= m¨ q

i

= ma

i

. For conservative forces

F = −V where V is the scalar potential. Rewriting Newton’s law we have

−

dV

dq

i

=

d

dt

(m˙ q

i

) (4.3)

Let us deﬁne the Lagrangian L(q

i

, ˙ q

i

) ≡ T −V where T is the kinetic energy.

In freshman physics T = T( ˙ q

i

) =

1

2

m˙ q

2

i

and V = V (q

i

) such as the harmonic

oscillator V (q

i

) =

1

2

kq

2

i

. That is in freshman physics T is a function only

of velocity ˙ q

i

and V is a function only of position q

i

. Thus L(q

i

, ˙ q

i

) =

45

46 CHAPTER 4. ENERGY-MOMENTUM TENSOR

T( ˙ q

i

) − V (q

i

). It follows that

∂L

∂q

i

= −

dV

dq

i

and

∂L

∂ ˙ q

i

=

dT

d ˙ q

i

= m˙ q

i

= p

i

. Thus

Newton’s law is

F

i

=

dp

i

dt

∂L

∂q

i

=

d

dt

(

∂L

∂ ˙ q

i

) (4.4)

with the canonical momentum [1] deﬁned as

p

i

≡

∂L

∂ ˙ q

i

(4.5)

The second equation of (4.4) is known as the Euler-Lagrange equations of

motion and serves as an alternative formulation of mechanics [1]. It is usually

written

d

dt

(

∂L

∂ ˙ q

i

) −

∂L

∂q

i

= 0 (4.6)

or just

˙ p

i

=

∂L

∂q

i

(4.7)

We have obtained the Euler-Lagrange equations using simple arguments. A

more rigorous derivation is based on the calculus of variations [1] as discussed

in Section 7.3.

We now introduce the Hamiltonian H deﬁned as a function of p and q as

H(p

i

, q

i

) ≡ p

i

˙ q

i

−L(q

i

, ˙ q

i

) (4.8)

For the simple case T =

1

2

m˙ q

2

i

and V = V ( ˙ q

i

) we have p

i

∂L

∂ ˙ q

i

= m˙ q

i

so that

T =

p

2

i

2m

and p

i

˙ q

i

=

p

2

i

m

so that H(p

i

, q

i

) =

p

2

i

2m

+ V (q

i

) = T + V which is the

total energy. Hamilton’s equations of motion immediately follow from (4.8)

as

∂H

∂p

i

= ˙ q

i

(4.9)

because L = L(p

i

) and

∂H

∂q

i

= −

∂L

∂q

i

so that from (4.4)

−

∂H

∂q

i

= ˙ p

i

. (4.10)

4.2. CLASSICAL FIELD THEORY 47

4.2 Classical Field Theory

Scalar ﬁelds are important in cosmology as they are thought to drive inﬂa-

tion. Such a ﬁeld is called an inﬂaton, an example of which may be the Higgs

boson. Thus the ﬁeld φ considered below can be thoguht of as an inﬂaton,

a Higgs boson or any other scalar boson.

In both special and general relativity we always seek covariant equations

in which space and time are given equal status. The Euler-Lagrange equa-

tions (4.6) are clearly not covariant because special emphasis is placed on

time via the ˙ q

i

and

d

dt

(

∂L

∂ ˙ q

i

) terms.

Let us replace the q

i

by a ﬁeld φ ≡ φ(x) where x ≡ (t, x). The generalized

coordiante q has been replaced by the ﬁeld variable φ and the discrete index

i has been replaced by a continuously varying index x. In the next section

we shall show how to derive the Euler-Lagrange equations from the action

deﬁned as

S ≡

Ldt (4.11)

which again is clearly not covariant. A covariant form of the action would

involve a Lagrangian density L via

S ≡

Ld

4

x =

Ld

3

xdt (4.12)

with L ≡

Ld

3

x. The term −

∂L

∂q

i

in equation (4.6) gets replaced by the

covariant term −

∂L

∂φ(x)

. Any time derivative

d

dt

should be replaced with

∂

µ

≡

∂

∂x

µ

which contains space as well as time derivatives. Thus one can

guess that the covariant generalization of the point particle Euler-Lagrange

equations (4.6) is

∂

µ

∂L

∂(∂

µ

φ)

−

∂L

∂φ

= 0 (4.13)

which is the covariant Euler-Lagrange equation for scalar ﬁelds. This will

be derived rigorously in the next section.

In analogy with the canonical momentum in equation (4.5) we deﬁne the

covariant momentum density

Π

µ

≡

∂L

∂(∂

µ

φ)

(4.14)

so that the Euler-Lagrange equations become

∂

µ

Π

µ

=

∂L

∂φ

(4.15)

48 CHAPTER 4. ENERGY-MOMENTUM TENSOR

The canonical momentum is deﬁned as

Π ≡ Π

0

=

∂L

∂

˙

φ

(4.16)

The energy momentum tensor is (analagous to (4.8))

T

µν

≡ Π

µ

∂

ν

φ −g

µν

L (4.17)

with the Hamiltonian density

H ≡

Hd

3

x

H ≡ T

00

= Π

˙

φ −L (4.18)

4.2.1 Classical Klein-Gordon Field

In order to illustrate the foregoing theory we shall use the example of the

classical, massive Klein-Gordon ﬁeld deﬁned with the Lagrangian density

(HL units ??)

L

KG

=

1

2

(∂

µ

φ∂

µ

φ −m

2

φ

2

)

=

1

2

[

˙

φ

2

−(φ)

2

−m

2

φ

2

] (4.19)

The covariant momentum density is more easily evaluated by re-writing

L

KG

=

1

2

g

µν

(∂

µ

φ∂

ν

φ − m

2

φ

2

). Thus Π

µ

=

∂L

∂(∂µφ)

=

1

2

g

µν

(δ

α

µ

∂

ν

φ + ∂

µ

φδ

α

ν

)

=

1

2

(δ

α

µ

∂

µ

φ + ∂

ν

φδ

α

ν

) =

1

2

(∂

α

φ + ∂

α

φ) = ∂

α

φ. Thus for the Klein-Gordon

ﬁeld we have

Π

α

= ∂

α

φ (4.20)

giving the canonical momentum Π = Π

0

= ∂

0

φ = ∂

0

φ =

˙

φ,

Π =

˙

φ (4.21)

Evaluating

∂L

∂φ

= −m

2

φ, the Euler-Lagrange equations give the ﬁeld equation

as ∂

µ

∂

µ

φ +m

2

φ or

(P

2

+m

2

)φ = 0

¨

φ −

2

φ +m

2

φ = 0 (4.22)

4.3. PRINCIPLE OF LEAST ACTION 49

which is the Klein-Gordon equation for a free, massive scalar ﬁeld. In mo-

mentum space p

2

= −P

2

, thus

(p

2

−m

2

)φ = 0 (4.23)

(Note that some authors [30] deﬁne P

2

≡

2

−

∂

2

∂t

2

diﬀerent from (3.42), so

that they write the Klein-Gordon equation as (P

2

−m

2

)φ = 0 or (p

2

+m

2

)φ =

0.)

The energy momentum tensor is

T

µν

≡ Π

µ

∂

ν

φ −g

µν

L

= ∂

µ

φ∂

ν

φ −g

µν

L

= ∂

µ

φ∂

ν

φ −

1

2

g

µν

(∂

α

φ∂

α

φ −m

2

φ

2

). (4.24)

Therefore the Hamiltonian density is H ≡ T

00

=

˙

φ

2

−

1

2

(∂

α

φ∂

α

φ − m

2

φ

2

)

which becomes [31]

H =

1

2

˙

φ

2

+

1

2

(φ)

2

+

1

2

m

2

φ

2

=

1

2

[Π

2

+ (φ)

2

+m

2

φ

2

] (4.25)

where we have relied upon the results of Section 3.4.1.

4.3 Principle of Least Action

derive EL eqns properly for q and φ (do later). Leave out for now.

4.4 Energy-Momentum Tensor for Perfect Fluid

The best references for this section are [9](Pg. 124-125), [7], and [32](Pg.

155). The book by D’Inverno [32] also has a nice discussion of the Navier-

Stokes equation and its relation to the material of this section. Other ref-

erences are [8](Pg. 83), [15](Pg. 330), [33](Pg. 259), [34](Pg. 38), and

[2].

These references show that the energy-momentum tensor for a perfect

ﬂuid is

50 CHAPTER 4. ENERGY-MOMENTUM TENSOR

T

µν

= (ρ +p)u

µ

u

ν

−pη

µν

(4.26)

where ρ is the energy density and p is the pressure. We shall now work this

out for several speciﬁc cases [9]. Fig. 2.5 of Narlikar’s book [7] is particularly

helpful.

Motionless dust represents a collection of particles at rest. Thus u

µ

=

(c, 0), so that T

00

= ρ. The equation of state for dust is p = 0 so that

T

ii

= 0 = T

0i

= T

ij

. Thus

T

µν

=

¸

¸

¸

¸

ρ 0 0 0

0 0 0 0

0 0 0 0

0 0 0 0

¸

(4.27)

Motionless ﬂuid representes a collection of particles all moving randomly

(such that they exert a pressure) but the whole collection is at rest, such

as a gas of particles at non-zero temperature, but conﬁned in a motionless

container. In this case u

µ

= (c, 0) again, but now p = 0. Thus again T

00

= ρ

but now T

ii

= p and T

ij

= 0 so that

T

µν

=

¸

¸

¸

¸

ρ 0 0 0

0 p 0 0

0 0 p 0

0 0 0 p

¸

(4.28)

Motionless radiation is characterized by the equation of state p =

1

3

ρ.

Again the radiation is conﬁned to a container not in motion so that u

µ

=

(γc, 0). (The

1

3

just comes from randomizing the pressure in 3 dimensions

[7].) Thus

T

µν

=

4

3

ρu

µ

u

ν

−

1

3

ρη

µν

=

¸

¸

¸

¸

ρ 0 0 0

0

1

3

ρ 0 0

0 0

1

3

ρ 0

0 0 0

1

3

ρ

¸

(4.29)

Thus the general case is the motionless ﬂuid energy-momentum tensor

in equation (4.28). The special cases of motionless dust or motionless radi-

ation are obtained with the respective substitutions of p = 0 or p =

1

3

ρ in

equation (4.28).

4.5. CONTINUITY EQUATION 51

4.5 Continuity Equation

In classical electrodynamics the fourcurrent density is j

µ

≡ (cρ, j) and the

covariant conservation law is ∂

µ

j

µ

= 0 which results in the equation of

continuity

∂ρ

∂t

+ .j = 0. This can also be obtained from the Maxwell

equations by taking the divergence of Amp`ere’s law. (do Problems 4.1

and 4.2) Thus the four Maxwell equations are entirely equivalent to only

three Maxwell equations plus the equation of continuity.

We had a similar situation in Chapters 1 and 2 where we found that the

velocity and acceleration equations imply the conservation equation. Thus

the two velocity and acceleration equations are entirely equivalent to only

the velocity equation plus the conservation law.

In analogy with electrodynamics the conservation law for the energy-

momentum tensor is

T

µν

;ν

= 0 (4.30)

In the next chapter we shall show how equation (2.1) can be derived from

this.

4.6 Interacting Scalar Field

We represent the interaction of a scalar ﬁeld with a scalar potential V (φ).

Recall our elementary results for L = T −V =

1

2

m˙ q

2

i

−V (q

i

) for the coordi-

nates ˙ q

i

. These discrete coordinates ˙ q

i

have now been replaced by continuous

ﬁeld variables φ(x) where φ has replaced the generalized coordinate q and

the discrete index i has been replaced by a continuous index x. Thus V (q

i

)

naturally gets replaced with V (φ) where φ ≡ φ(x).

Thus for an interacting scalar ﬁeld we simply tack on −V (φ) to the free

Klein-Gordon Lagrangian of equation (4.19) to give

L =

1

2

(∂

µ

φ∂

µ

φ −m

2

φ

2

) −V (φ)

≡ L

O

+L

I

(4.31)

where L

O

≡ L

KG

and L

I

≡ −V (φ). Actually the Lagrangian of (4.31)

refers to a minimally coupled scalar ﬁeld as opposed to conformally coupled

[21] (Pg. 276). It is important to emphasize that V (φ) does not contain

derivative terms such as ∂

µ

φ. Thus the covariant momentum density and

canonical momentum remain the same as equations (4.20) and (4.21) for the

52 CHAPTER 4. ENERGY-MOMENTUM TENSOR

free particle case namely Π

α

= ∂

α

φ and π =

˙

φ. Solving the Euler-Lagrange

equations now gives

(P

2

+m

2

)φ +V

= 0

¨

φ −

2

φ +m

2

φ +V

= 0 (4.32)

with

V

≡

dV

dφ

(4.33)

The energy-momentum tensor is the same as for the free particle case,

equation (4.24), except for the addition of g

µν

V (φ) as in

T

µν

= ∂

µ

φ∂

ν

φ −g

µν

[

1

2

(∂

α

φ∂

α

φ −m

2

φ

2

) −V (φ)] (4.34)

yielding the Hamiltonian density the same as for the free particle case, equa-

tion (4.25), except for the addition of V (φ) as in

H ≡ T

00

=

1

2

˙

φ

2

+

1

2

(φ)

2

+

1

2

m

2

φ

2

+V (φ)

=

1

2

[Π

2

+ (φ)

2

+m

2

φ

2

] +V (φ). (4.35)

The purely spatial components are T

ii

= ∂

i

φ∂

i

φ−g

ii

[

1

2

(∂

α

φ∂

α

φ−m

2

φ

2

)−

V (φ)] and with g

ii

= −1 (i.e. assume Special Relativity NNN) we obtain

T

ii

=

1

2

˙

φ

2

+

1

2

(φ)

2

−

1

2

m

2

φ

2

−V (φ) (4.36)

Note that even though T

ii

has repeated indices let us not assume

¸

i

is

implied in this case. That is T

ii

refers to T

ii

= T

11

= T

22

= T

33

and not

T

ii

= T

11

+T

22

+T

33

. Some authors (e.g Serot and Walecka [34]) do assume

the latter convention and therefore will disagree with our results by

1

3

.

Let us assume that the eﬀects of the scalar ﬁeld are averaged so as

to behave like a perfect (motionless) ﬂuid. In that case, comparing equa-

tion (4.28), we make the identiﬁcation [13, 34]

E ≡ ρ ≡< T

00

> (4.37)

and

p ≡< T

ii

> (4.38)

4.7. COSMOLOGY WITH THE SCALAR FIELD 53

where E ≡ ρ is the energy density and p is the pressure. (Note that because

Serot and Walecka do assume the Einstein summation convention for T

ii

,

they actually write p ≡

1

3

< T

ii

>.) Making these identiﬁcations we have

ρ =

1

2

˙

φ

2

+

1

2

(φ)

2

+

1

2

m

2

φ

2

+V (φ) (4.39)

and

p =

1

2

˙

φ

2

+

1

2

(φ)

2

−

1

2

m

2

φ

2

−V (φ) (4.40)

Let us also assume that the scalar ﬁeld is massless and that φ = φ(t)

only, i.e. φ = φ(x), so that spatial derivatives disappear. (See Pg. 276-277

of Kolb and Turner [21] and Pg. 138 of Islam [13]). Therefore we ﬁnally

obtain [13, 21].

ρ =

1

2

˙

φ

2

+V (φ)

(4.41)

and

p =

1

2

˙

φ

2

−V (φ)

(4.42)

4.7 Cosmology with the Scalar Field

We have ﬁnished with our discussion of the energy-momentum tensor and

therefore we should now move onto the next chapter. However, with the

tools at hand (energy-momentum tensor and Friedmann equations) we can

discuss the relevance of the scalar ﬁeld to cosmology without needing the

formalism of General Relativity. Therefore before proceeding to the next

chapter we shall make a brief digression and discuss the evolution of the

scalar ﬁeld.

If one is considering cosmological evolution driven by a scalar ﬁeld, one

can simply substitute the above expressions for ρ and p into the Friedmann

54 CHAPTER 4. ENERGY-MOMENTUM TENSOR

and acceleration equations (1.29) and (1.30) to obtain the time evolution of

the scale factor as in

H

2

≡ (

˙

R

R

)

2

=

8πG

3

[

1

2

˙

φ

2

+

1

2

(φ)

2

+

1

2

m

2

φ

2

+V (φ)] −

k

R

2

+

Λ

3

(4.43)

and

¨

R

R

= −

4πG

3

[

1

2

˙

φ

2

+

1

2

(φ)

2

−

1

2

m

2

φ

2

−V (φ)] +

Λ

3

(4.44)

The equation for the time evolution of the scalar ﬁeld is obtained either by

taking the time derivative of equation (4.43) or more simply by substituting

the expression for ρ and p in equations (4.41) and (4.42) into the conservation

equation (2.1) to give

¨

φ + 3H[

˙

φ +

(φ)

2

˙

φ

] +m

2

φ +V

= 0. (4.45)

Note that this is a new Klein-Gordon equation quite diﬀerent to equation (4.32).

The diﬀerence occurs because we have now incorporated gravity via the

Friedmann and conservation equation. We shall derive this equation again

in Chapter 7.

Again assuming the ﬁeld is massless and ignoring spatial derivatives we

have

¨

φ + 3H

˙

φ +V

= 0

(4.46)

Notice that this is the equation for a damped harmonic oscillator (V =

1

2

kx

2

and

dV

dx

≡ V

= kx with F = −V

) as

m¨ x +d ˙ x +kx = 0 (4.47)

Kolb and Turner [21] actually also include a particle creation term due to

the decay of the scalar ﬁeld, which will cause reheating, and instead write

¨

φ + 3H

˙

φ + Γ

˙

φ +V

= 0 (4.48)

4.7. COSMOLOGY WITH THE SCALAR FIELD 55

4.7.1 Alternative derivation

We can derive the equation of motion (4.46) for the scalar ﬁeld in a quicker

manner [29] (Pg. 73), but this derivation only seems to work if we set m = 0

and φ = 0 at the beginning. (Exercise: ﬁnd out what goes wrong if m = 0

and φ = 0.)

Consider a Lagrangian for φ which already has the scale factor built into

it as

L = R

3

[

1

2

(∂

µ

φ∂

µ

φ −m

2

φ

2

) −V (φ)] (4.49)

The R

3

factor comes from

√

−g = R

3

for a Robertson-Walker metric. This

will be discussed in Chapter 7. Notice that it is the same factor which

sits outside the Friedmann Lagrangian in equation (2.20). The equation of

motion is (do Problem 4.3)

¨

φ −

2

φ + 3H

˙

φ +m

2

φ +V

= 0 (4.50)

which is diﬀerent to (4.45). (NNNN why ???) However if m = 0 and φ = 0

it is the same as (4.46).

Let’s only consider

L = R

3

[

1

2

˙

φ

2

−V (φ)] (4.51)

which results from setting m = 0 and φ = 0 in (4.49). The equation of

motion is

¨

φ + 3H

˙

φ +V

= 0 (4.52)

Notice how quickly we obtained this result rather than the long procedure

to get (4.46). We didn’t even use the energy-momentum tensor. Also realize

that because φ = 0 the above Lagrangian formalism is really no diﬀerent to

our old fashioned formalism where we had q

i

(t). Here we have only φ = φ(t)

(not φ = φ(x)), and so we only have i = 1, i.e. q

i

≡ φ.

Identifying the Lagrangian as [29] L = R

3

(T −V ) we immediately write

down the total energy density ρ = T + V =

1

2

˙

φ

2

+ V (φ). Taking the time

derivative ˙ ρ =

˙

φ

¨

φ + V

˙

φ = −3H

˙

φ

2

from (4.46) and substituting into the

conservation equation (2.1), ˙ ρ = −3H(ρ + p) we obtain the pressure as

p =

1

2

˙

φ

2

− V (φ). Thus our energy density and pressure derived here agree

with our results above (4.39) and (4.40). Notice that the pressure is nothing

more than p =

L

R

3

. [29].

56 CHAPTER 4. ENERGY-MOMENTUM TENSOR

4.7.2 Limiting solutions

Assuming that k = Λ = 0 the Friedmann equation becomes

H

2

≡ (

˙

R

R

)

2

=

8πG

3

(

1

2

˙

φ +V ) (4.53)

This equation together with equation (4.46) form a set of coupled equations

where solutions give φ(t) and R(t). We solve the coupled equations in the

standard way by ﬁrst eliminating one variable, then solving one equation,

then substituting the solution back into the other equation to solve for the

other variable. Let’s write equation (4.46) purely in terms of φ by eliminating

R which appears in the form H =

˙

R

R

. We eliminate R by substituting H

from (4.53) into (4.46) to give

¨

φ +

12πG(

˙

φ

2

+ 2V )

˙

φ+V’=0

¨

φ

2

+ 2

¨

φV

−12πG(

˙

φ

2

+ 2V )

˙

φ

2

+V

2

= 0 (4.54)

Notice that this is a non-linear diﬀerential equation for φ, which is diﬃcult

to solve in general. In this section we shall study the solutions for certain

limiting cases. Once φ(t) is obtained from (4.54) it is put back into (4.53)

to get R(t).

Potential Energy=0

Setting V = 0 we then have ρ =

1

2

˙

φ

2

= p. Thus our equation of state is

p = ρ (4.55)

or γ = 3.

With V = V

= 0 we have

¨

φ

2

+

√

12πG

˙

φ

2

= 0 (4.56)

which has the solution (do problem 4.4)

φ(t) = φ

o

+

1

√

12πG

ln[1 +

√

12πG

˙

φ(t −t

o

)] (4.57)

4.7. COSMOLOGY WITH THE SCALAR FIELD 57

(Note that the solution is equation (9.18) of [29] is wrong.) Upon substitut-

ing this solution back into the Friedmann equation (4.53) and solving the

diﬀerential equation we obtain (do problem 4.5)

R(t) = R

o

[1 +

√

12πG

˙

φ

o

(t −t

o

)]

1/3

. (4.58)

This result may be understood from another point of view. Writing the

Friedmann equations as

H

2

≡ (

˙

R

R

)

2

=

8πG

3

ρ (4.59)

and

ρ =

α

R

m

(4.60)

then the solution is always

R∝t

2/m

(4.61)

which always gives

ρ ∝

1

t

2

. (4.62)

If

ρ = constant (4.63)

(corresponding to m = 0) then the solution is

R∝e

t

(4.64)

(do problem 4.6). Note that for m < 2, one obtains power law inﬂation.

For ordinary matter (m = 3), or radiation (m = 4) we have R ∝ t

2/3

and

R ∝ t

1/2

respectively. Returning to the scalar ﬁeld solution (4.57) the density

is ρ =

1

2

˙

φ

2

for V = 0. Thus

˙

φ(t) =

˙

φ

o

1 +

√

12πG

˙

φ

o

(t −t

o

)

(4.65)

combined with (

R

Ro

)

3

= 1 +

√

12πG

˙

φ

o

(t −t

o

) from (4.58) yields

˙

φ(t) =

˙

φ

o

R

3

o

R

3

(4.66)

to give the density

58 CHAPTER 4. ENERGY-MOMENTUM TENSOR

ρ =

1

2

˙

φ

2

o

R

6

o

R

6

(4.67)

corresponding to m = 6 and thus R ∝ t

1/3

in agreement with (4.58). Note

also that this density ρ ∝

1

R

6

also gives ρ ∝

1

t

2

.

Thus for a scalar ﬁeld with V = 0, we have p = ρ (γ = 3) and ρ ∝

1

R

6

.

Contrast this with matter for which p = 0(γ = 0) and ρ ∝

1

R

3

or radiation

for which p =

1

3

ρ (γ = 1/3) and ρ ∝

1

R

4

.

However equation (4.67) may not be interpreted as a decaying Cosmo-

logical Constant because p = ρ (see later).

Kinetic Energy=0

Here we take

˙

φ =

¨

φ = 0, so that ρ = V and p = −V giving

p = −ρ (4.68)

or γ = −3 which is a negative pressure equation of state. Our equation of

motion for the scalar ﬁeld (4.54) becomes

V

= 0 (4.69)

meaning that

V = V

o

(4.70)

which is constant. Substituting the solution into the Friedmann equation

(4.53) gives

H

2

= (

˙

R

R

)

2

=

8πG

3

V

o

(4.71)

which acts as a Cosmological Constant and which has the solution (do prob-

lem 4.7)

R(t) = R

o

e

8πG

3

Vo(t−to)

(4.72)

which is an inﬂationary solution, valid for any V .

Warning

We have found that if k = Λ = 0 and if ρ ∝

1

R

m

then R∝t

2

for any

value of m. All of this is correct. To check this we might substitute into the

Friedmann equation as

H

2

= (

˙

R

R

)

2

∝

1

t

2

(4.73)

4.7. COSMOLOGY WITH THE SCALAR FIELD 59

and say

˙

R

R

∝

1

t

giving

1

R

dR

dt

dt ∝

dt

t

which yields ln R ∝ lnt and thus

R ∝ t

2/m

. The result R ∝t is wrong because we have left out an important

constant.

Actually if

˙

R

R

=

c

t

then lnR = c ln t = ln t

c

giving R ∝ t

c

instead of R∝t.

Let’s keep our constants then. Write ρ =

d

2

R

m

then R = (

md

2

)

2/m

t

2/m

and ρ =

d

2

(

md

2

)

2

t

2

=

(2/m)

2

t

2

. Substituting into the Friedmann equation gives

(

˙

R

R

)

2

=

(2/m)

2

t

2

or

˙

R

R

=

(2/m)

t

with the above constant C =

2

m

yielding

R ∝t

2/m

in agreement with the correct result above.

The lesson is be careful of constants when doing back-of-the-envelope

calculations.

4.7.3 Exactly Solvable Model of Inﬂation

Because (4.54) is a diﬃcult non-linear equation, exactly solvable models are

very rare. We shall examine the model of Barrow [35] which can be solved

exactly and leads to power law inﬂation. The advantage of an exactly solv-

able model is that one can develop ones physical intuition better. Barrow’s

model [35] is brieﬂy introduced by Islam [13].

Any scalar ﬁeld model is speciﬁed by writing down the potential V (φ).

Barrow’s potential is

V (φ) ≡ βe

−λφ

(4.74)

where β and λ are constants to be determined. Barrow [35] claims that a

particular solution to (4.54) is (which was presumably guessed at, rather

then solving the diﬀerential equation)

φ(t) =

√

2Alnt (4.75)

where

√

2A is just some constant. We check this claim by substituting (4.74)

and (4.75) into (4.54). From this we ﬁnd (do problem 4.9) that

λ =

2

A

(4.76)

and

β = −A (4.77)

or

β = A(24πGA−1) (4.78)

60 CHAPTER 4. ENERGY-MOMENTUM TENSOR

Note that Barrow is wrong when he writes λA =

√

2. Also he uses units

with 8πG = 1, so that the second solution (4.78), he writes correctly as

β = A(3A−1). Also Barrow doesn’t use the ﬁrst solution (4.77) for reasons

we shall see shortly.

Having solved for φ(t) we now substitute into (4.53) to solve for R(t).

(Recall φ(t) and R(t) are the solutions we seek to our coupled equations

(4.46) and (4.53).) Substituting V =

Λ

t

2

and

˙

φ =

√

2A

t

(see solution to

problem 4.9) we have

H

2

≡ (

˙

R

R

)

2

=

8πG

3

(

1

2

2A

t

2

+

β

t

2

) =

8πG

3

(A+β)

1

t

2

(4.79)

giving an equivalent density

ρ =

A+β

t

2

(4.80)

Clearly we see why we reject the ﬁrst solution (4.77) with β = −A. It would

give zero density. Using the second solution (4.78) with β = A(24πGA−1)

yields

ρ =

24πGA

2

t

2

. (4.81)

Solving the Friedmann equation (4.79) gives

R∝t

8πGA

(4.82)

where D is some constant. Setting 8πG ≡ 1 we have

R∝t

A

(4.83)

in agreement with Barrow’s solution. Power law inﬂation results for

A > 1. (4.84)

Inverting the solution (4.83) we have t

2

= C

R

2/A

where C

is some constant.

Substituting into (4.81) we have

ρ ∝

1

R

2/A

(4.85)

which corresponds to a Weak decaying Cosmological Constant. (See sections

4.7.4 and 4.7.5) For the inﬂationary result A > 1 we have

2

A

≡ m < 2 which

corresponds to the quantum tunneling solution!!

Note of course that (4.85) can also be obtained via ρ =

1

3

˙

φ

2

+V. We have

V=

β

t

2

∝

1

R

2/A

and

˙

φ =

√

2A

t

giving

˙

φ

2

∝

1

R

2/A

.

4.7. COSMOLOGY WITH THE SCALAR FIELD 61

4.7.4 Variable Cosmological Constant

In this section we address the question as to when the density can be inter-

preted as a Cosmological Constant. Recall the Friedmann equations

H

2

≡ (

˙

R

R

)

2

=

8πG

3

ρ −

k

R

2

+

Λ

3

(4.86)

and

¨

R

R

= −

4πG

3

(ρ + 3p) +

Λ

3

= −

4πG

3

ρ(1 +γ) +

Λ

3

(4.87)

for p =

γ

3

ρ. Suppose ρ = k = 0, then we have

H

2

= (

˙

R

R

)

2

=

Λ

3

(4.88)

and

¨

R

R

=

Λ

3

(4.89)

where two things have happened. Firstly the velocity and acceleration equa-

tions both have the same right hand side. Secondly the acceleration is posi-

tive. What sort of density would give the same result. Again for k = 0

H

2

= (

˙

R

R

)

2

=

8πG

3

ρ (4.90)

and

¨

R

R

=

8πG

3

ρ (4.91)

only for γ = −3 or p = −ρ. From our conservation equation, ˙ ρ = −3H(ρ+p),

this can only happen for ρ=constant. Thus constant density with equation

of state p = −ρ acts identically to a Cosmological Constant. In addition

the solution is automatically are of exponential inﬂation, R ∝ e

Ht

. (Ex-

ercise:verify this.) Let us deﬁne a Strong Cosmological Constant as one in

which the velocity and acceleration equations both have the same right hand

side, (which automatically implies that the acceleration is positive). Such a

Strong Cosmological Constant must be a true constant.

62 CHAPTER 4. ENERGY-MOMENTUM TENSOR

On the other hand we can imagine densities that still give a positive

acceleration (i.e. inﬂation) but do not normally give the velocity and accel-

eration with the same right hand side. Examining (4.54) indicates that the

acceleration is guarenteed to be positive if γ < −1 giving p < −

1

3

ρ. (Recall

that the exponential inﬂation above required γ = −3, which is consistent

with the inequality.) Thus negative pressure gives inﬂation. (Although not

all negative pressure gives inﬂation, e.g. p = −

1

4

ρ.) The inﬂation due to

γ < −1 will not be exponential inﬂation, but something weaker like per-

haps power law inﬂation. Let us deﬁne a Weak Cosmological Constant as

one which arises from negative pressure (actually p < −

1

3

ρ) to give a positive

acceleration (inﬂation) only. The velocity and acceleration equations need

not have the same right hand side.

Recall that ordinary matter and radiation, or any positive pressure equa-

tion of state, neccesarily leads to negative acceleration (with Λ = 0). Thus

positive pressure leads to attractive gravity. However positive acceleration

implies a repulsive gravity or antigravity. Thus negative pressure (actually

p < −

1

3

ρ) leads to antigravity. This is why we wish to use the term weak Cos-

mological Constant (even though right hand sides are not the same) because

it is consonant with antigravity.

Let us summarize. We consider ρ alone with Λ ≡ 0. For

¨

R > 0 ⇒γ < −1 and ρ =anything (e.g. ρ = ρ(R) or ρ =constant)

¨

R > 0 and ρ =constant ⇒γ = −3

For γ < −1, ρ behaves as a weak Cosmological Constant and for γ = −3,

ρ behaves as a Strong Cosmological Constant.

Finally let us emphasize that it is perfectly legitimate to consider a Weak

Cosmological Constant as a real Cosmological Constant. Einstein’s original

motivation in introducing Λ was to obtain a static universe. Thus all he

wanted was a antigravity term; i.e. all he wanted was a weak Cosmological

Constant. It “accidentally” happened that the right hand sides turned out

to be equal, giving a strong Λ.

Interpreting ρ as a Cosmological Constant leads us to expect that a

weak Cosmological Constant can vary. This follows from ρ=anything above.

i.e. ρ=constant or ρ = ρ(R) giving a variable function. (But a strong

Cosmological Constant cannot vary).

4.7. COSMOLOGY WITH THE SCALAR FIELD 63

4.7.5 Cosmological constant and Scalar Fields

Refer back to the density and pressure of the scalar ﬁeld in equations (4.41))

and (4.42). We had

ρ =

1

2

˙

φ

2

+V (φ)

and

ρ =

1

2

˙

φ

2

−V (φ).

For the case where PE = V = 0 we have p = ρ. The pressure is positive

and therefore equation (4.67), ρ ∝

1

R

6

, cannot be interpreted as a (variable)

Cosmological Constant.

For the case where KE =

1

2

˙

φ

2

= 0 we have p = −ρ meaning that ρ can

be interpreted as a Strong Cosmological Constant. (We found V

= 0 and

thus ρ = V = constant = V

o

and p = −V

o

.)

These results are true in general (assuming m = ∇φ = 0) as we have not

yet speciﬁed V (φ).

Let us now consider the Barrow model V (φ) = βe

−λφ

. we found that

φ(t) =

2

λ

ln t and β =

2

λ

2

(

6

λ

2

− 1) for 8πG ≡ 1. Introducing A ≡

2

λ

we can

re-write as β = A(3A − 1) and φ(t) =

√

2Aln t and V (φ) = A(3A − 1)e

φ

2

A

.

Substituting we obtain

ρ

Barrow

=

3A

2

t

2

(4.92)

and

ρ

Barrow

=

3A

2

t

2

(

2A

3

−1) = (

2A

3

−1)ρ (4.93)

the general equation of state is p =

γ

3

ρ giving the Barrow equation of state

γ

Barrow

=

2

A

−3 (4.94)

in equation (4.84) we concluded that power law inﬂation results for A > 1.

Substituting this into (4.94) implies

γ

Barrow

< −3 (4.95)

which we expect because power law inﬂation implies

¨

R > 0. thus for A > 1

the Barrow pressure is negative with γ < −1 and thus ρ

Barrow

=

3A

2

t

2

cor-

responds to a Weak Cosmological Constant. Furthermore this Cosmological

Constant is variable and decays with time. in equation (4.85) we wrote this

as ρ

Barrow

∝

1

R

2/A

.

64 CHAPTER 4. ENERGY-MOMENTUM TENSOR

4.7.6 Clariﬁcation

We wish to clarify the distinction between a Strong Cosmological Constant,

a Weak Cosmological Constant and a Varying Cosmological Constant. A

Strong Cosmological Constant occurs when the Friedmann equation are of

the form

H

2

≡ (

˙

R

R

)

2

=

Λ

3

(4.96)

and

−qH

2

≡

¨

R

R

=

Λ

3

(4.97)

and comparing with

(

˙

R

R

)

2

=

8πG

3

ρ (4.98)

¨

R

R

= −

8πG

3

(ρ + 3p) (4.99)

leads us to conclude that the equation of state for a Strong Cosmological

Constant is p = −ρ, where we have identiﬁed Λ ≡ 8πGρ

v

. The conservation

equation

˙ ρ = −3H(ρ +p) (4.100)

leads us to conclude that Λ ≡ 8πGρ

v

=constant for a Strong Cosmological

Constant.

A Variable Cosmological Constant Λ(R), can also be considered but the

equation cannot be like (4.96) and (4.97) with no ρ term. As we shall see be-

low a Varying Cosmological Constant automatically involves matter creation

(existence of a ρ term). Thus Λ(R) cannot be written down in the Friedmann

equation without also writing ρ. Thus for a Varying Cosmological Constant

(let k = 0) we have

H

2

≡ (

˙

R

R

)

2

=

8πG

3

ρ +

Λ(R)

3

(4.101)

and

−qH

2

≡

¨

R

R

= −

4πG

3

(1 +γ)ρ +

Λ

3

(4.102)

One can easily show that the conservation equation becomes

˙ ρ + ˙ ρ

v

=-3H(ρ+p) =−H(3+γ)ρ

4.7. COSMOLOGY WITH THE SCALAR FIELD 65

(4.103)

(do Problem 4.10). Other ways of writing this are (see equations 2.1-2.7

).

d(ρR

3

)

dt

+p

dR

3

dt

+R

3

˙ ρ

v

= 0 (4.104)

d(ρR

3+γ

)

dt

+R

3+γ

˙ ρ

v

= 0 (4.105)

ρ

+ρ

v

= −

3

R

(ρ +p) (4.106)

d(ρR

3

)

dR

+ 3pR

2

+R

3

ρ

v

= 0 (4.107)

1

R

3+γ

d(ρR

3+γ

)

dR

= −ρ

v

(4.108)

The last equation show clearly that if ρ

v

= 0 (i.e. ρ

v

= constant) then

matter or radiation is created or destroyed. This is why we cannot write

(4.96) and (4.97) if Λ is allowed to vary. We must include particle creation

terms as in (??) and (4.102).

For a Variable Cosmological Constant the same term

Λ(R)

3

appears in

both right hand sides of the Friedmann equation (4.101) and (4.102). How-

ever we don’t know its equation of state. Also γ can be anything and thus

˙ ρ

v

can lead to creation of matter, radiation or anything else. Further Λ(R)

can vary.

Both a Strong and Weak Cosmological Constant have well deﬁned equa-

tion of state, both with (diﬀerent) negative pressure. A Weak Cosmological

Constant can vary but the right hand sides of the Friedmann equations diﬀer.

Thus we see that we have 3 diﬀerent types of Cosmological Constant

namely, Strong, Weak, and Variable, each with both similar and dissimiliar

properties. These are summarize in the Table.

4.7.7 Generic Inﬂation and Slow-Roll Approximation

We shall ﬁrst discuss features that are common to many models of inﬂation

based on scalar ﬁelds [21]. These models typically have a large region of the

potential where the potential is ﬂat (slow-roll region) and then a harmonic

66 CHAPTER 4. ENERGY-MOMENTUM TENSOR

Weak Strong Variable

Right Hand Sides

of Velocity and Diﬀerent same Same (but must include ρ)

Acceleration Eqns.

Eqn. State p < −

1

3

ρ p = −ρ Unknown

Λ = 8πGρ

v

Variable Constant Variable

Table 4.1: Properties of 3 types of Cosmological Constants

region where the potential looks similar to an oscillatior (rapid oscillation

region). Such a potential is shown in Fig 4.1.

Let us ﬁrst analyze this situation from a simple physics point of view. If

Fig 4.1 were actually a plot of V (x) versus x the equation of motion would

be

m¨ x +b ˙ x +V

= 0 (4.109)

which, for V =

1

2

kx

2

or V

**= kx would represent a damped harmonic oscil-
**

lator, which indeed is the case in the rapid oscillation region. In the slow

roll regime we have V ≈constant or V = V

o

− Cx where C is small, giving

V

**= −C. The physical situation would represent a ball rolling down a hill
**

with friction into a valley [21].

In the slow roll regime friction dominates and the ball moves at terminal

velocity [21], ˙ x=constant and ˙ x doesn’t change much so that ¨ x = 0. This is

the slow roll regime we have

b ˙ x +V

= 0 (4.110)

and for V = V

o

−Cx we have

b ˙ x≈C (4.111)

When the ball ﬁnally reaches the valley we are back to the full equation

(4.109) where V

**≡ kx so that the ball experiences rapid, damped oscilla-
**

tions.

Let’s return to the scalar ﬁeld equations which are (for Λ = k = 0)

H

2

≡ (

˙

R

R

)

2

=

8πG

3

(

1

2

˙

φ +V ) (4.112)

4.7. COSMOLOGY WITH THE SCALAR FIELD 67

and

¨

φ + 3H

˙

φ +V

= 0 (4.113)

where the friction term 3H is due to the expansion of the universe [21]. Based

on analogy with the rolling ball, the slow roll approximation (

˙

φ ≈constant,

¨

φ ≈ 0) is

3H

˙

φ +V

≈ 0 (4.114)

In addition, the slow roll region is characterized by a small kinetic energy

1

2

˙

φ

2

<< V , so that

H

2

≈

8πG

3

V

o

≈ constant (4.115)

For the rolling ball we only had one equation, but here we have two equations

(4.114) and (4.115) involved in the slow roll approximations. The slow roll

equation can also be obtained directly from equation (4.54) where we set

˙

φ

2

<< 2V

o

and

¨

φ ≈ 0 to give

24πGV

o

˙

φ +V

= 0 (4.116)

which is consistent with combining (4.114) and (4.115). Normally we would

have the formula for V(φ) and solve (4.116) or (4.54) ﬁrst for φ(t) and then

substitute our answer into the Friedmann equation to obtain R(t). However

we have not yet speciﬁed V (φ). We have speciﬁed that we are in a slow roll

regime thus making (4.115) valid which we can solve directly for R(t) giving

an exponential inﬂationary solution

R(t)≈R

o

e

t

8πG

3

V

O

(4.117)

Thus, no matter what V (φ) is, the slow regime always give approximately

exponential inﬂation.

To obtain φ(t) we must specify V (φ) even in the slow-roll regime.

4.7.8 Chaotic Inﬂation in Slow-Roll Approximation

The Chaotic inﬂation model is deﬁned by the potential

V (φ) =

1

2

m

2

φ

2

(4.118)

which is exactly analogous to the harmonic oscillator potential V (x) =

1

2

kx

2

.

68 CHAPTER 4. ENERGY-MOMENTUM TENSOR

Using V

o

=

1

2

m

2

φ

2

o

the slow roll solution (4.117) becomes

R(t) = R

o

e

t

4π

3

√

Gmφ

= R

o

e

t

4π

3

m

Mp

φ

(4.119)

where G ≡

1

M

2

p

is in agreement with equation (9.23) of Madsen [29]. Having

speciﬁed V (φ) in (4.118) we can solve (4.116), which

˙

φ = −

1

√

24πG

V

√

V

o

(4.120)

or

˙

φ

φ

= −

m

√

12πG

1

φ

o

(4.121)

This gives

φ(t) = φ

o

e

−

m

φo

√

12πG

t

= φ

o

e

−

mMp

φo

√

12π

t

= φ

o

−

mM

p

√

12π

t (4.122)

is the slow roll region in agreement with equation (9.25) of Madsen [29]. For

m << M

p

or for short times we see that φ(t) will be approximately constant,

or slowing rolling. We see that φ(t) is a decaying exponential in time. Thus

there will be a “half-life” or “lifetime” associated with slow roll which we

deﬁne as

τ =

φ

o

√

12π

mM

p

(4.123)

When t = τ we see that φ = φ

o

1

e

. That is the amplitude is reduced by the

factor

1

e

We expect that the slow roll approximation will be valid for t < τ.

Our solution of φ(t) in (4.122) came from solving (4.116). We can also

obtain φ(t) slightly diﬀerently. Let’s not assume V = V

o

=constant, but only

that

˙

φ

2

<< 2V and

¨

φ ≈ 0. Then instead of (4.116) we have

√

24πGV

˙

φ +V

= 0 (4.124)

which, for chaotic inﬂation becomes

√

12πGmφ

˙

φ +m

2

φ = 0 (4.125)

4.7. COSMOLOGY WITH THE SCALAR FIELD 69

or

˙

φ = −

m

√

12πG

(4.126)

which has the solution

φ(t) = φ

o

−

m

√

12πG

t

= φ

o

−

mM

p

√

12π

t (4.127)

in agreement with (4.122) for short time.

We can further investigate the validity of the slow roll approximation by

evaluating the potential as a function of time and checking that it is constant

for short times. We do this by substituting our solution for φ(t) back into

the potential. We get

V =

1

2

m

2

φ

2

=

1

2

m

2

φ

2

o

e

−

mMp

φo

√

3π

t

≈

1

2

m

2

φ

o

(φ

o

−

mM

p

√

3π

t) (4.128)

Thus we see that for m << M

p

or for short times the potential is indeed

constant.

For short times (or for m << M

p

) we have veriﬁed that φ and V are

approximately constant. This means that

˙

φ ≈ 0 and ρ ≈ V

o

which give

exponential inﬂation. (Also p ≈ −V

o

, so that p = −ρ).

In order to solve the horizon, ﬂatness and monopole problems, most

models require a high degree of inﬂation typically amounting to about 60

e-folds [21]. Given R(t) = R

o

e

Ht

, the number of e-folds after time t is

N = ln(

R(t)

R

o

) = Ht. (4.129)

(Actually a better formula is given in equation (8.26) of the book by Kolb

and Turner [21] (pg.278)). After one lifetime τ, the number of e-folds is

N = Hτ =

4π

3

m

M

p

φ

o

φ

o

√

12π

mM

p

= 4π

φ

2

o

M

2

p

(4.130)

where H =

4π

3

m

Mp

φ

o

taken from (4.119) and τ is from (4.123). Thus the

requirement N ≥ 60 yields

N ≥ 60 ⇒ φ

o

≥

√

5M

p

. (4.131)

70 CHAPTER 4. ENERGY-MOMENTUM TENSOR

Notice how the ﬂatness problem is solved in inﬂation. We have H

2

≡ (

˙

R

R

)

2

=

8πG

3

ρ −

k

R

2

≈

8πG

3

V

o

−

k

R

2

. During inﬂation (slow roll) V

o

stays constant but

by the end of inﬂation R = R

o

e

60

or

1

R

2

=

1

R

2

o

e

−120

. The term

k

R

2

has

dropped by e

−120

, whereas V

o

has remained constant. Thus the term

k

R

2

is

entirely negligible. Inﬂation does not give k = 0, but rather gives

k

R

2

≈ 0

which is equivalent to k = 0. This is an improtant distinction. The universe

can have k = 0 or k = +1 or k = −1. No matter what the value of k, it gets

diluted by inﬂation and is equivalent to a universe with k = 0. thus within

our horizon the universe is ﬂat. Quantum cosmology predict that a universe

which arises via tunnelling must have k = +1. This is perfectly OK with

inﬂation which simply dilutes the curvature.

Quantum tunnelling requires k = +1. Inﬂation actually says nothing

about the value of k. It simply predicts that

k

R

2

≈ 0 at the end of inﬂation.

On Earth, the reason many people believe the Earth is ﬂat is because

we cannot see beyond the horizon. Up to the horizon it looks ﬂat. If we

could see beyond the horizon we would see the curvature. Similarly for our

universe. According to inﬂation the size of the universe is much larger then

the distance to the horizon ( i.e. as far as we can see) the universe looks

ﬂat becuase

k

R

2

is negligible. If we could see beyond the horizon we would

see the curvature. And quantum tunnelling predicts that what we would see

would be a universe of positive curvature.

Cosmological Constant associated with Chaotic Inﬂation

Let us now calculate the density as a function of R. We solve (4.119) for

t = t(R) as

t =

3

4π

M

p

mφ

o

ln(

R

R

o

). (4.132)

Substituting into (4.122) we have

φ(R) = φ

o

(

R

o

R

)

M

2

p

4πφ

2

o

(4.133)

and with V (φ) =

1

2

m

2

φ

2

we have

V (R) =

1

2

m

2

φ

2

o

(

R

o

R

)

M

2

p

2πφ

2

o

. (4.134)

Also from (4.122)

˙

φ(t) = −

mM

p

√

12π

e

−

mMp

φo

√

12π

t (4.135)

4.7. COSMOLOGY WITH THE SCALAR FIELD 71

so that

˙

φ(R) = −

mM

p

√

12π

(

R

o

R

)

M

2

p

4πφ

2

o

. (4.136)

Finally evaluating ρ =

1

2

˙

φ

2

+V we have

ρ(R) =

1

2

m

2

(

M

2

p

12π

+φ

2

o

)(

R

o

R

)

M

2

p

2πφ

2

o

. (4.137)

Thus

ρ(R) ∝

1

R

m

where m ≡

M

2

p

2πφ

2

o

. (4.138)

Recall previously that an inﬂationary solution requires m < 2, yielding

φ

o

>

M

p

√

4π

= 0.3 M

p

(4.139)

which is entirely consistent with (4.131)!

What is the equation of state? using p =

1

2

˙

φ

2

−V we obtain

p =

1

2

m

2

(

M

2

p

12π

−φ

2

o

)(

R

o

R

)

M

2

p

2πφ

2

o

. (4.140)

The question is, is the pressure negative? We ﬁnd that for inﬂation to occur

we need φ

o

>

Mp

√

4π

. Write this as

φ

o

=

M

p

√

4π

with > 1 (4.141)

Thus the density becomes

ρ(R) =

m

2

M

2

p

8π

(

1

3

+

2

)(

R

o

R

)

m

(4.142)

and

p(R) =

m

2

M

2

p

8π

(

1

3

−

2

)(

R

o

R

)

m

(4.143)

Deﬁning κ ≡

m

2

M

2

p

8π

(

Ro

R

)

m

, we write ρ = (

1

3

+

2

)κ and p = (

1

3

−

2

)κ, giving

p =

1−3

2

1+3

2

ρ. The requirement > 1 yields

p < −

1

2

ρ (4.144)

72 CHAPTER 4. ENERGY-MOMENTUM TENSOR

which means negative pressure! Writing p =

γ

3

ρ gives

γ = −

3

2

. (4.145)

These results are in agreement with our previous constraints that in order

to have positive

¨

R (inﬂation) we needed p <

1

3

ρ or γ < −1.

Our chaotic inﬂation model is the slow roll approximation gives negative

pressure (but not p = −ρ) and corresponds to a Weak Decaying Cosmological

Constant!

4.7.9 Density Fluctuations

An important result that we shall use without proof is that ﬂuctuations of

the scalar ﬁeld are given approximately by

δφ ≈

H

2π

(4.146)

this result is discussed by Linde [36] (Pg.17,50), Kolb and Turner [21](Pg.284),

Collins, Martin and Squires [37] (Pg.410) and by Dolgov, Sazhin and Zel-

dovich [38].

Using ρ =

1

2

˙

φ

2

+V(φ) we have

dρ

dφ

= V

(φ) or

δρ = V

δφ≈V

H

2π

(4.147)

which, upon assuming

˙

φ = 0, gives

δρ

ρ

=

2

3π

GV

2

V

=

2

3π

V

M

p

√

V

(4.148)

where we have used G ≡

1

M

2

p

. For the chaotic inﬂation model, V =

1

2

m

2

φ

2

this yields

δρ

ρ

=

4

3π

m

M

p

. (4.149)

This is an intensely important formula often written as

δρ

ρ

≈m

√

G =

m

M

p

(4.150)

4.7. COSMOLOGY WITH THE SCALAR FIELD 73

the density ﬂuctuations observed by COBE are

δρ

ρ

≈ 10

−5

yielding m =

10

−5

M

p

.

The above formula is not very useful for the Barrow model where a well

deﬁned inﬂation means m ≡ m

inflation

is not present. In that case the

formula is written more usefully as [?, ?, ?]

δρ

ρ

=

ρ

inflation

ρ

p

(4.151)

4.7.10 Equation of State for Variable Cosmological Constant

In this section we wish to demonstrate that Variable Cosmological Constant

models have negative pressure [?, ?, ?].

Firstly if one assumes

p ≡

γ

3

ρ (4.152)

then the conservation law follows as

1

R

3+γ

d

dR

(ρR

3+γ

) = −ρ

v

. (4.153)

Let’s assume that

ρ

v

≡

α

R

m

= ρ

vo

(

R

o

R

)

m

. (4.154)

Integrating the conservation law we have

ρ =

A

R

3+γ

+χρ

v

(4.155)

where

χ ≡

m

3 +γ −m

(4.156)

and A is a constant give by

A = (ρ

o

−χρ

vo

)R

3+γ

o

. (4.157)

The pressure is

p =

γ

3

A

R

3+γ

+

γ

3

m

3 +γ −m

ρ

v

(4.158)

This looks like bad news. Assuming that ρ

v

dominate over the ﬁrst term

at some stage of evolution, it looks like the pressure only get negative for

m> 3 + γ. However, there are two things to keep in mind. Firstly, the

74 CHAPTER 4. ENERGY-MOMENTUM TENSOR

pressure p is not the pressure of radiation or matter or vacuum because

ρ =

A

R

3+γ

+χρ, and p =

γ

3

ρ. The pressure that we would want to be negative

would be the vacuum pressure p

v

, which we shall work out below. Secondly,

the key point is not as much having p negative but rather having

¨

R positive.

The equation

−qH

2

≡

¨

R

R

= −

4πG

3

(ρ + 3p) +

8πG

3

ρ

v

(4.159)

can still give postive

¨

R even if p is not negative, because the ρ

v

term has to

be considered. The Friedmann equation is

H

2

≡ (

˙

R

R

)

2

=

8πG

3

(ρ +ρ

v

) −

k

R

2

(4.160)

Let us evaluate the right hand sides of these two equations (4.159) and

(4.160).

From (4.155) we have

H

2

≡ (

˙

R

R

)

2

=

8πG

3

[

A

R

3+γ

+ (χ + 1)ρ

v

] −

k

R

2

(4.161)

where

χ + 1 =

3 +γ

3 +γ −m

. (4.162)

Also assuming p=

γ

3

ρ we have

−qH

2

≡

¨

R

R

= −

4πG

3

(1 +γ)(

A

R

3+γ

+χρ

v

) +

8πG

3

ρ

v

(4.163)

= −

4πG

3

[1 +γ)

A

R

3+γ

+

(3 +γ)(m−2)

3 +γ −m

ρ

v

]. (4.164)

Deﬁne

θ ≡

(3 +γ)(m−2)

3 +γ −m

(4.165)

we see that θ is always negative for m < 2! Thus if the vacuum term domi-

nates equation (4.164) then

¨

R will be positive for m < 2. This agrees with

our previous consideration that if ρ =

α

R

m

dominates the velocity equation

then m < 2 leads to inﬂation.

Note however that θ can be negative for other values of m as shown in

Table 4.2.

4.7. COSMOLOGY WITH THE SCALAR FIELD 75

m θ(γ = 0) θ(γ = 1)

0.5 -1.8 -1.7

1 -1.5 -1.3

1.5 -1.0 -0.8

2 0 0

2.5 3.0 1.33

3 ∞ 4

4 -6 ∞

5 -4.5 -12

6 -4 -8

Table 4.2: θ as a function of γ and m

Having established that a decaying Cosmological Constant can lead to

negative pressure, let us now work out the vacuum equation of state for a

decaying Cosmological Constant. Looking at (4.161) let us deﬁne

˜ ρ ≡

A

R

3+γ

(4.166)

and

˜ ρ

v

≡ (χ + 1)ρ

v

=

3 +γ

3 +γ −m

ρ

v

(4.167)

so that

ρ +ρ

v

= ˜ ρ + ˜ ρ

v

(4.168)

giving

H

2

≡ (

˙

R

R

)

2

=

8πG

3

(ρ +ρ

v

) −

k

R

2

=

8πG

3

(˜ ρ + ˜ ρ

v

) −

k

R

2

(4.169)

and from (4.164) we have

−qH

2

≡

¨

R

R

= −

4πG

3

[(1 +γ)ρ −2ρ

v

] (4.170)

= −

4πG

3

[(1 +γ)˜ ρ + (m−2)˜ ρ

v

] (4.171)

which we would like to write as

−qH

2

≡

¨

R

R

= −

4πG

3

(˜ ρ + 3˜ p) −

4πG

3

(˜ ρ

v

+ 3˜ p

v

). (4.172)

This is achieved if we make the following deﬁntions

76 CHAPTER 4. ENERGY-MOMENTUM TENSOR

˜ p ≡

γ

3

˜ ρ

(4.173)

and

˜ p

v

≡

m−3

3

˜ ρ

v

≡

γv

3

˜ ρ

v

,

(4.174)

which is our vacuum equation of state for a decaying Cosmological Constant.

We see that for m< 3 we have ˜ p

v

and γ

v

negative. For a m < 2 we have

γ

v

< −1 which we saw previously is the condition for inﬂation assuming

vacuum domination of the density and pressure.

It is also satisfying to note that the equation of state for the non-vacuum

component (equation 4.173) is identical to the equation of state for a perfect

ﬂuid that we encountered for models without a Cosmological Constant.

Alternative Derivation

Our deﬁnition of ˜ ρ and ˜ ρ

v

above are not unique. We present alternative

deﬁnitions below which will give the same vacuum equation of state but

diﬀerent for the non-vacuum component.

In equation (4.161) the density is

ρ +ρ

v

=

A

R

3+γ

+ (χ + 1)ρ

v

= (ρ

o

−χρ

vo

)(

R

o

R

)

3+γ

+ (χ + 1)ρ

vo

(

R

o

R

)

m

= ρ

o

(

R

o

R

)

3+γ

+ρ

vo

[(χ + 1)(

R

o

R

)

m

−χ(

R

o

R

)

3+γ

] (4.175)

We previoulsy deﬁned ˜ ρ ≡

A

R

3+γ

and ˜ ρ

v

≡ (χ + 1)ρ

v

. However we might

alternatively deﬁne

¯ ρ ≡ ρ

o

(

R

o

R

)

3+γ

(4.176)

and

¯ ρ

v

≡ ρ

vo

[(χ + 1)(

R

o

R

)

m

−χ(

R

o

R

3+γ

]

= (χ + 1)ρ

v

−χ

ρ

vo

ρ

o

¯ ρ (4.177)

4.7. COSMOLOGY WITH THE SCALAR FIELD 77

still obtaining

ρ +ρ

v

= ˜ ρ + ˜ ρ

v

= ¯ ρ + ¯ ρ

v

. (4.178)

Assuming p =

γ

3

ρ we have (compare to equation 4.171)

−qH

2

≡

¨

R

R

= −

4πG

3

[(1 +γ)ρ −2ρ

v

]

= −

4πG

3

[(1 +γ −m

ρ

vo

ρ

o

)¯ ρ + (m−2)¯ ρ

v

] (4.179)

(where we have used the second expression in 4.177), which we would like

to write as

−qH

2

≡

¨

R

R

= −

4πG

3

(¯ ρ + 3¯ p) −

4πG

3

(¯ ρ

v

+ 3¯ p

v

). (4.180)

This is achieved if we make the following deﬁnitions

¯ p ≡

γ−m

ρvo

ρo

3

¯ ρ ≡

Γ

3

¯ ρ

(4.181)

and

¯ p

v

≡

m−3

3

¯ ρ

v

=

γv

3

¯ ρ

v

(4.182)

4.7.11 Quantization

All of our proceeding work with the scalar ﬁeld was at the classical level. In

this section we wish to consider quantum eﬀects.

In section (2.4.2) we dervied the wheeler-DeWitt equation in minisuper-

space approximation. We began with the Lagrangian in equation (2.20)

L = −κR

3

[(

˙

R

R

)

2

−

k

R

2

+

8πG

3

(ρ +ρ

v

)] (4.183)

and identiﬁed the conjugate momentum p ≡

∂L

∂

˙

R

= −2κ

˙

RR and derived the

Wheeler-DeWitt equation, after quantizing with p →i

∂

∂R

, as

{−

∂

2

∂R

2

+ 4κ

2

[kR

2

−

8πG

3

(ρ +ρ

v

)R

4

]}ψ = 0. (4.184)

78 CHAPTER 4. ENERGY-MOMENTUM TENSOR

Notice that our quantization ’didn’t do anything to the density.”

In the work that we have done in the present chapter we have made

an eﬀort to write the scalar ﬁeld as a function of R, i.e. φ = φ(R) and

using ρ =

1

2

˙

φ

2

+ V (φ) we have written as an eﬀective density ρ(R) for

the scalar ﬁeld. Our intention has been to simply insert this ρ(R) into

the Wheeler-DeWitt equation (4.184). In our work on inﬂation we found

that for ρ ∝

1

R

m

dominating the Friedmann equation then inﬂation occurs

for m < 2. If this density also dominates ρ + ρ

v

in the Wheeler-DeWitt

equation, then a tunnelling potential will only be present for m < 2. Thus

inﬂation and quantum tunnelling require the same condition. This leads us

to the hypothesis that inﬂation and quantum tunnelling are identical! Or in

other words, inﬂation is simply a classical description of quantum tunnelling.

We call this hypothesis Quantum Inﬂation.

Quantum inﬂation is easy to validate for ordinary densities, either ρ or

ρ

v

, that behave like ρ ∝

1

R

m

. With our discussion of the scalar ﬁeld we have

written ρ

φ

∝

1

R

m

so it would seem that the idea of quantum inﬂation also

works for scalar ﬁelds.

In our quantization procedure we ”didn’t do anything to the density.” In

terms of scalar ﬁelds then this quantization procedure is diﬀerent to what

other people do with quantization. The usual procedure [?, ?, ?, 20, 21] is

to quantize φ and R seperately and arrive at a Wheeler- DeWitt equation

in terms of both of these variables. Let us now study this procedure.

We begin with the Lagrangian

L = −κR

3

[(

˙

R

R

)

2

−

k

R

2

] + 2π

2

R

3

[

1

2

˙

φ

2

−V (φ)] (4.185)

where κ ≡

3π

4G

.

From this one can deduce that

(

˙

R

R

)

2

=

8πG

3

(

1

2

˙

φ

2

+V ) −

k

R

2

(4.186)

and also

¨

φ + 3H

˙

φ +V

= 0 (4.187)

provided one uses

¨

R

R

= −

4πG

3

(ρ + 3p) with ρ =

1

2

˙

φ

2

+ V and p =

1

2

˙

φ

2

− V .

(NNN see FE)

The canonical momenta are

Π

R

≡

∂L

∂

˙

R

= −2κR

˙

R (4.188)

4.7. COSMOLOGY WITH THE SCALAR FIELD 79

and

Π

φ

≡

∂L

∂

˙

φ

= 2π

2

R

3

˙

φ. (4.189)

The Hamiltonian (H = p

i

˙ q

i

−L) becomes (using κ =

3π

4G

)

H = Π

R

˙

R + Π

φ

˙

φ −L (4.190)

= −κR

3

[(

˙

R

R

)

2

+

k

R

2

−

8πG

3

(

1

2

˙

φ

2

+V )] ≡ 0 (4.191)

where the result H = 0 is obtained by comparing the expression for the

Hamiltonian to the Friedmann equation (4.186). This Hamiltonian is exactly

analogous to the Hamiltonian we had in equation (2.22) where we had ρ

instead of

1

2

˙

φ

2

+V .

Writing H in terms of the conjugate momenta we have

H = −κR

3

[

Π

2

R

4κ

2

R

4

+

d

R

2

−

8πG

3

(

Π

2

φ

8π

4

R

6

+V )] = 0 (4.192)

which, of course is also equal to zero. This Hamiltonian is inherited to

equation (11.34), Pg.462 of the book by Kolb and Turner [21]. This equation

is re-arranged as

Π

2

R

−

3

4πG

2

R

2

Π

2

φ

+

9π

2

4G

2

(kR

2

−

8πG

3

V R

4

) = 0. (4.193)

In order to compare to our signal Wheeler-DeWitt equation let’s replace Π

φ

with Π

φ

= 2π

2

R

3

˙

φ, which results in

Π

2

R

+

9π

2

4G

2

[kR

2

−

8πG

3

(

1

2

˙

φ

2

+V )R

4

] = 0 (4.194)

which is exactly analogous to our original Wheeler-DeWitt equation (2.24)

where we had ρ instead of

1

2

˙

φ

2

+V .

Equation (4.193) is quantized by making the replacements

Π

R

→−i

∂

∂R

(4.195)

and

Π

φ

→−i

∂

∂φ

(4.196)

and setting Hψ = 0 to give

80 CHAPTER 4. ENERGY-MOMENTUM TENSOR

[−

∂

2

∂R

2

+

3

4πG

1

R

2

∂

2

∂φ

2

+

9π

2

4G

2

(kR

2

−

8πG

3

VR

4

)]ψ = 0

(4.197)

which is the Wheeler-DeWitt equation in minisuperspace approximation for

a quantized scalar ﬁeld φ. This is identical to equation (10.1.11), Pg.270 of

the book by Linde [36].

We identify the potential as

U(R, φ) =

9π

2

4G

2

(kR

2

−

8πG

3

V R

4

)

(4.198)

which is identical to equation (11.35), Pg.463 of the book by Kolb and Turner

[21]. In equation (10.1.25), Pg277 of the book by Linde [36], he gives an

expression for V (R, φ) =

3π

G

V (a), so that our result does agree with Linde.

We can see that the above method of quantizing the scalar ﬁeld φ directly

is still consistent with our idea of Quantum Inﬂation. Recall that ρ(R)

and V (R) in terms of φ(R) obvioulsy ρ, V and

˙

φ

2

must have the same R

dependence. Thus if ρ ∝

1

R

m

then also V ∝

1

R

m

in the same way. Thus our

potential U(R, φ) will alsways exhibit a tunnelling shape for m < 2. Thus

Quantum Inﬂation still works for U(R, φ) when φ is quantized seperately.

4.8 Problems

4.1 Show that ∂

µ

j

µ

= 0 is the equation of continuity

∂ρ

∂t

+ ∇· j = 0 where

j

µ

≡ (ρ, j).

4.2 Show that the above equation of continuity also results from taking the

divergence of Amp` er´ es law.

4.3 Show that the Lagrangian in equation (4.49) yields the equation of mo-

tion (4.50).

4.4 A) Solve the diﬀerential equation (4.56). B) Check your answer by

substituting your solution back into the equation.

4.5 Now put your solution from problem 4.4 into the Friedmann equation

(4.53) and solve for R(t).

4.8. PROBLEMS 81

4.6 A) If (

˙

R

R

)

2

∝

1

R

m

, show that R ∝ t

2/m

. B) If H

2

≡ (

˙

R

R

)

2

= constant

show that R ∝ e

Ht

.

4.7 If (

˙

R

R

)

2

=

8πG

3

V

o

, show that R = R

o

e

8πG

3

Vo(t−to)

.

4.8 Show that inﬂationary solutions are characterized by

¨

R > 0 and non-

inﬂationary expansions by

¨

R < 0

4.9 Barrow’s model is V (φ) ≡ βe

−λφ

. Check that φ(t) =

√

2Aln t is a

soluttion and evaluate the constants β and λ in terms of A.

4.10 Prove equations (4.103)-(4.108).

82 CHAPTER 4. ENERGY-MOMENTUM TENSOR

Chapter 5

EINSTEIN FIELD

EQUATIONS

The Einstein’s ﬁeld equations are

G

µν

= 8πGT

µν

+ Λg

µν

(5.1)

which are a set of 16 coupled equations which will give g

µν

(buried inside

G

µν

) gives T

µν

. Actually there are only 10 independent equations because

of the symmetry g

µν

= g

νµ

.

In principle our job is easy. Just write down T

µν

and solve for g

µν

which

speciﬁes the metric. Then we can calculate the paths of light rays, the orbits

of planets, etc. In practice the solution of the Einstein ﬁeld equations are

exceedingly diﬃcult and only a few exact solutions are known.

Two excellent reference for this section are Chapter 7 of the book by

Lawden [?] and Chapter 2 of the book by Islam [13].

In actual practice, the way one usually solves the Einstein’s equations

is to specify a metric in general terms which contains unknown coeﬃcients.

This metric is substituted into the Einstein equations and one solves for the

unknown coeﬃcients.

Thus we need to learn how to derive the metric for the spaces un-

der consideration. Let us learn how to derive the Friedmann-Robertson-

Walker (FRW) metric which is the metric appropriate to a homogeneous

and isotropic universe but where size can change with time.

83

84 CHAPTER 5. EINSTEIN FIELD EQUATIONS

5.1 Preview of Riemannian Geometry

5.1.1 Polar Coordinate

A general N-dimensional Riemannian space, denoted by R

N

, is one in which

the distance ds between two neighboring points can be written (Pg. 88 [?])

ds

2

= g

µν

dx

µ

dx

ν

. (5.2)

If coordinates can be found such that

ds

2

= dx

µ

dx

ν

(5.3)

over the whole space then the space is said to be Euclidean and is denoted

by E

N

. Clearly E

N

is a special case of R

N

.

Below we shall restrict our discussion to the spatial part of the metric

denoted as

d

2

= h

ij

dx

i

dx

j

. (5.4)

We shall very often have (eg. FRW and special relativity metrics)

ds

2

= c

2

dt

2

−d

2

(5.5)

so that

h

ij

= −g

ij

. (5.6)

Consider the two dimensional space where

d

2

= dr

2

+r

2

dθ

2

. (5.7)

Here h

11

= 1, h

12

= h

21

= 0, h

22

= r

2

. The space looks like R

2

but actually it

is E

2

because we can ﬁnd coordinate such that (5.3) is true. These coordinate

are the two-dimensional plane polar coordinates

x = r cos θ (5.8)

y = r sin θ (5.9)

in which

d

2

= dx

2

+dy

2

, (5.10)

with g

11

= g

22

=1 and g

12

= g

21

= 0 or g

µν

= δ

µν

.

5.1. PREVIEW OF RIEMANNIAN GEOMETRY 85

Recall that for 3-dimensional spherical polar coordinates

x = r sin θ cos φ (5.11)

y = r sin θ sin φ (5.12)

z = r cos θ (5.13)

and the increments of length d

r

, d

θ

, d

φ

is the e

r

, e

θ

and e

φ

directions

respectively are

d

r

= dr (5.14)

d

θ

= rdθ (5.15)

d

φ

= r sin θdφ. (5.16)

Thus the surfaces of a sphere is an example of a space which is R

2

and

cannot be reduced to E

2

. On the surface of the sphere the distance between

two points is

d

2

= d

2

θ

+d

2

φ

(5.17)

= r

2

dθ

2

+r

2

sin

2

θdφ

2

(5.18)

where h

11

= r

2

, h

12

= h

21

= 0, h

22

= r

2

sin

2

θ. For this surface it is not

possible to ﬁnd x, y such that ds

2

= dx

2

+ dy

2

and therefore the surface of

a sphere is not E

∈

but rather a genuine R

∈

space.

5.1.2 Volumes and Change of Coordinates

The measure is the volume element in some set of coordinates. There are 3

ways to calculate the measure.

1) Inﬁnitesimal Length Method is one in which one identiﬁes the inﬁn-

tesimal increments of length and simply multiplies them together to get the

volume element. In Cartesian Coordinates we have d

x

= dx, d

y

= dy, and

d

z

= dz to give

dV = d

x

d

y

d

z

= dxdydz. (5.19)

In 2-d plane polar coordinates d

r

= dr and d

θ

= rdθ to give

dV = d

r

d

θ

= rdrdθ (5.20)

(actually this ”volume” is an area). In 3-D spherical polar coordinates d

r

=

dr, d

θ

= dθ and d

φ

= r sin dφ to give

dV = d

r

d

θ

d

φ

= r

2

sin θdrdθdφ. (5.21)

86 CHAPTER 5. EINSTEIN FIELD EQUATIONS

2) Jacobian Method is the one usually mentioned in introductory calculus

books ([?] Pg.746) for changing variables. Suppose x = x(u, v) and y =

y(u, v) then

f(x, y)dxdy =

f[x(u, v), y(u, v)] | J(u, v) | dudv (5.22)

where | J(u, v) | is the modulus of the Jacobian deﬁned as

J(u, v) ≡

∂x

∂u

∂x

∂v

∂y

∂u

∂y

∂v

(5.23)

For 3-dimensions with x = x(u, v, w), y = y(u, v, w) and z = z(u, v, w) we

have

J(u, v, w) ≡

∂x

∂u

∂x

∂v

∂x

∂w

∂y

∂u

∂y

∂v

∂y

∂w

∂z

∂u

∂z

∂v

∂z

∂w

(5.24)

For Cartesian coordinates obviously J(x, y, z) = 1. For plane polar coordi-

nates (x = r cos θ, y = r sin θ)

J(r, θ) =

cos θ −r sin θ

sin θ r cos θ

= r (5.25)

and for spherical polar coordinates (x = r sin θ cos φ, y = r sin θ cos φ, z =

r cos θ)

J(r, θ, φ) =

**sin θ cos φ r cos θ cos φ −r sin θ sin φ
**

sin θ sin φ r cos θ sin φ r sin θ cos φ

cos θ −r sin θ 0

= r

2

sin θ (5.26)

The volume element in 2-d is

dV =| J(u, v) | dudv (5.27)

and in 3-d it is

dV =| J(u, v, w) | dudvdw (5.28)

which then reproduce equations (5.19), (5.139) and (5.140) for Cartesian,

plane polar and spherical polar coordinates.

3) Metric Tensor Method is what we prefer in general relativity. Here

([21], Pg.33)

5.1. PREVIEW OF RIEMANNIAN GEOMETRY 87

dV =

√

hdudvdw

(5.29)

where h is the determinant of the spatial metric tensor. Thus

| J |=

√

h

(5.30)

For a plane polar coordinates ds

2

= dr

2

+r

2

dθ

2

≡ g

ij

dx

i

dx

j

so that

h =

1 0

0 r

2

= r

2

(5.31)

giving

√

h = r so that dV =

√

hdrdθ in agreement with (5.139). For spherical

polar coordinates d

2

= dr

2

+r

2

dθ

2

+r

2

sin

2

θdφ

2

≡ h

ij

dx

i

dx

j

giving

h =

1 0 0

0 r

2

0

0 0 r

2

sin

2

θ

= r

4

sin

2

θ (5.32)

giving

√

h = r

2

sin θ so that dV =

√

hdrdθdφ = r

2

sin θdrdθdφ in agreement

with (5.140).

Thus deﬁne

d

2

x ≡ dudv

d

3

x ≡ dudvdw

d

4

x ≡ dudvdwdt (5.33)

so that the measure is

dV =

√

hd

2

x or

√

hd

3

x

(5.34)

depending on the number of dimensions. It is important to remember that

d

3

x or d

4

x om this notation is not d

1

d

2

d

3

or d

1

d

2

d

3

d

4

but simply only

the coordinates. For example in spherical polar coordinates

d

3

x ≡ drdθdφ (5.35)

or with time

d

4

x ≡ drdθdφdt. (5.36)

The measure is volume is obtained with dV =

√

hd

3

x or

√

−gd

4

x (because

for 4-d we use g

µν

and h

ij

= −g

ij

and h = −g). Thus in general

88 CHAPTER 5. EINSTEIN FIELD EQUATIONS

dV = d

3

x

or

dV = d

4

x

in contrast to conventions used, say in undergraduate physics books.

5.1.3 Diﬀerential Geometry

A Good reference for diﬀerential geometry is the book by Lipschutz [?].

Introductory material is discussed by Purcell and Varberg [?] (Pg.625-634,

285-290).

Previously we wrote the deﬁnition of a circle as x

2

+ y

2

= r

2

which

could be written generally as y = y(x). The same equation can be expressed

parametrically in terms of the parameter θ as x = r cos θ and y = r sin θ or

generally as x = x(θ) and y = y(θ). For many curves the form y = y(x)

can be clumsy and nowadays mathematicians always prefer the parametric

representation. (See [?] Pg.570). Another way to write our equation for the

circle (radius=1) is

x = cos θˆ e

1

+ sin θˆ e

2

(5.37)

where ˆ e

1

and ˆ e

2

are basis vectors in E

2

.

Thus a general curve (Chapter 3 of [?]) is expressed as

x = x(t) (5.38)

where t is the parameter. If the basis is chosen to be E

2

then x = x(t) is

equivalent to two scalar equations x

1

= x

1

(t) and x

2

= x

2

(t). Thus a curve

can be speciﬁed in any number of dimensions. For our circle above we have

x

1

= x

1

(θ) = cos θ and x

2

= x

2

= sin θ.

A general surface (Chapter 8 of [?]) is expressed as

x = x(u, v). (5.39)

If, for example, the basis in E

3

then

x(u, v) = x

1

(u, v)ˆ e

1

+x

2

(u, v)ˆ e

2

+x

3

(u, v)ˆ e

3

. (5.40)

Good references of the next 3 sections are the books by Kolb and Turner

([21], Pg.31-35) and Landau ([?], Pg.177-181) and Ohanian and Ruﬃni ([8],

Pg546-552) and Chow ([28], Pg.197-200) and the article by Kung [?].

5.1. PREVIEW OF RIEMANNIAN GEOMETRY 89

5.1.4 1-dimesional Curve

Let us ﬁrst consider the circle, often called the one sphere denoted by S

1

.

Recall that for a circle the radius R, the proper way to express it is in terms

of 1-dimensional parameter θ as

x(θ) = R(cos θˆ e

1

+ sin θˆ e

2

). (5.41)

However in introductory books one always introduces a ﬁcticous extra di-

mension and embeds the 1-d curve in a 2-d Euclidean space via

x

2

+y

2

= R

2

(5.42)

which we recognize as the equation for a circle. Remember though this

equation is really overkill. It is a 2-d equation for a 1-d curve! The 1-

parameter equation (5.41) is much better. We can also write

x

2

1

+x

2

2

= R

2

. (5.43)

The element of length in the 2-d Euclidean space is

d

2

= dx

2

1

+dx

2

2

. (5.44)

In an ordinary 2-d Euclidean space x and y (or x

1

and x

2

) are free to vary

independently and this is how the whole 2-d space get covered. Equation

(5.44) is true in general. However the reason that (5.42) or (5.43) describes

a circle is because it constrains the value of y in terms of x. This constraint

(5.42) picks our only those points in E

2

which give the circle.

Equation (5.44) covers all of E

2

. We can constrain it for the circle by

reducing the two parameters x

1

and x

2

to only one parameter. Thus we will

have d for the circle. We do this using the 2-d constraint (5.43) and writing

y =

√

R

2

−x

2

and dy =

−x

√

R

2

−x

2

dx so that

dy

2

=

x

2

R

2

−x

2

dx

2

. (5.45)

Note that dy

2

≡ (dy)

2

and dy

2

= d(y

2

). Thus (5.44) becomes

d

2

= dx

2

+

x

2

R

2

−x

2

dx

2

=

R

2

R

2

−x

2

dx

2

. (5.46)

90 CHAPTER 5. EINSTEIN FIELD EQUATIONS

This can also be written in terms of the dimensionless coordinate

r ≡

x

R

(5.47)

to give

d

2

= R

2

dr

2

1 −r

2

(5.48)

where R is the radius of the space (the circle).

Another convenient coordinate system for the circle use the angle θ from

plane polar coordinates speciﬁed via

x = Rcos θ

y = Rsin θ. (5.49)

Identifying the increments of length d

R

and d

θ

in the ˆ e

R

and ˆ e

θ

directions

as

d

R

= dR

d

θ

= Rdθ (5.50)

then

d

2

= d

2

R

+d

2

θ

= dR

2

+R

2

dθ

2

(5.51)

which gives the distance d in the 2-d space. To restrict ourselves to the rim

of the circle (curved 1-d space) we ﬁx d

R

= dR = 0 and get

d

2

= R

2

dθ

2

(5.52)

which makes it obvious that the space is the one sphere (circle) of radius R.

Using simple trigonometry one can show that (5.52) is the same as (5.46).

(Do Problem 5.1)

Using d

2

≡ h

ij

dx

i

dx

j

we evidently have

h

ij

= (R

2

) (5.53)

which is a 1-dimensional ”motion”. The determinant is obviously h = R

2

giving

√

h = R. This allows us to calculate the volume (we are calling the

length a geological volume) as

V =

√

hd

1

x =

2π

0

Rdθ = 2πR (5.54)

5.1. PREVIEW OF RIEMANNIAN GEOMETRY 91

The 1-d curve that we described above is the circle or one sphere denoted

S

1

. However there are three 1-d spaces which are homogeneous and isotropic.

There are i) the ﬂat x line (R

1

), ii) the positively curved one sphere (S

1

)

derived above and iii) the negatively curved hyperbolic curve (H

1

). [21]

The formulas for a space of constant negative curvature [21] can be ob-

tained with the replacement

R →iR (5.55)

to yield

d

2

=

−R

2

−R

2

−x

2

dx

2

=

R

2

R

2

+x

2

dx

2

(5.56)

or using r ≡

x

R

d

2

= R

2

dr

2

1 +r

2

. (5.57)

These results are also obtained by embedding in Minkowski space ( do Prob-

lem 5.2).

The line element for a space of zero curvature is obviously just

d

2

= dx

2

(5.58)

or using r ≡

x

R

d

2

= R

2

dr

2

. (5.59)

These formulas are obtained from S

1

or H

1

by letting R →∞.

We can collect our results for R

1

, S

1

and H

1

into a single formula

d

2

=

R

2

R

2

−kx

2

dx

2

(5.60)

or using r ≡

x

R

d

2

= R

2 dr

2

1−kR

2

(5.61)

where k = 0, +1, −1 for ﬂat, closed and open curves respectively. (i.e. for

R

1

, S

1

and H

1

respectively)

92 CHAPTER 5. EINSTEIN FIELD EQUATIONS

5.1.5 2-dimensional Surface

Kolb and Turner [21] analyze this problem very nicely. Other references are

listed following equation (5.40).

A surface is represented by two parameters u, v and expressed as x =

x(u, v) as mentioned previously. However we shall introduce a surface ﬁcti-

cious coordinate (three parameters) and embedding the surface in E

3

. Thus

with 3 parameters the equation for the two sphere is

x

2

+y

2

+z

2

+R

2

(5.62)

which we recognize as the equation for a sphere used in introductory books.

However this equation is overkill. It is a 3-d equation for a 2-d surface. We

can also write

x

2

1

+x

2

2

+x

2

3

= R

2

. (5.63)

Any 3-d Euclidean space E

3

has length element

d

2

= dx

2

1

+dx

2

2

+dx

2

3

(5.64)

which under normal circumstances would map out the whole 3-d volume.

However (5.63) restricts x

3

according to

x

2

3

= R

2

−x

2

1

−x

2

2

. (5.65)

Writing

dx

3

=

∂x

3

∂x

1

dx

1

+

∂x

3

∂x

2

dx

2

(5.66)

and with

∂x

3

∂x

1

=

−x

1

√

R

2

−x

2

1

−x

2

2

we have

dx

3

= −

x

1

dx

1

+x

2

dx

2

R

2

−x

2

1

−x

2

2

(5.67)

or

dx

2

3

=

(x

1

dx

1

+x

2

dx

2

)

2

R

2

−x

2

1

−x

2

2

(5.68)

to give ([21], Pg.32)

d

2

= dx

2

1

+dx

2

2

+

x

1

dx

1

+x

2

dx

2

)

2

R

2

−x

2

1

−x

2

2

(5.69)

5.1. PREVIEW OF RIEMANNIAN GEOMETRY 93

which is re-written as

d

2

=

1

R

2

−x

2

1

−x

2

2

[(R

2

−x

2

2

)dx

2

1

+(R

2

−x

2

1

)dx

2

2

+x

1

x

2

dx

1

dx

2

+x

2

x

1

dx

2

dx

1

].

(5.70)

Let us introduce plane polar coordinates in the x

3

plane as

x

1

= r

cos θ x

2

= r

sin θ. (5.71)

These coordinates are shown very clearly in Fig 2.1, Pg.32 of Kolb and

Turner [21]. Thus

θ : 0 → 2π

r

: 0 → R. (5.72)

Using

dx

i

=

∂x

i

∂x

1

dr

+

∂x

i

∂θ

dθ (5.73)

then (5.69) and (5.70) become

d

2

=

R

2

R

2

−r

2

dr

2

+r

2

dθ

2

. (5.74)

This can also be written in terms of the dimensionless coordinate

r ≡

r

R

(5.75)

to give

d

2

= R

2

[

dr

2

1 −r

2

+r

2

dθ

2

] (5.76)

where

r : 0 → 1. (5.77)

Another convenient coordinate system for the two sphere uses angles θ

and φ from spherical polar coordinates speciﬁed via

x = Rsin θ cos φ (5.78)

y = Rsin θ sin φ (5.79)

z = Rcos θ (5.80)

and substituting into (5.64) directly yields

d

2

= R

2

(dθ

2

+ sin

2

θdφ

2

). (5.81)

94 CHAPTER 5. EINSTEIN FIELD EQUATIONS

Equation (5.81) is alternatively obtained by identifying the increments of

length d

R

, d

θ

, d

φ

in the ˆ e

R

, ˆ e

θ

and ˆ e

φ

directions as

d

R

= dR (5.82)

d

θ

= Rdθ (5.83)

d

φ

= Rsin θdφ (5.84)

then

d

2

= d

2

R

+d

2

θ

+d

2

φ

= dR

2

+R

2

(dθ

2

+ sin

2

θdφ

2

) (5.85)

gives the distance ds in the 3-d space. To restrict ourselves to the surface of

the sphere (curved 2-d space) we ﬁnd d

R

= dR = 0 and get

d

2

= R

2

(dθ

2

+ sin

2

θdφ

2

) (5.86)

in agreement with (5.81).

Using d

2

≡ h

ij

dx

i

dx

j

we evidently have for S

2

h

ij

=

R

2

0

0 R

2

sin

2

θ

(5.87)

The determinant is obviously

h = R

4

sin

2

θ (5.88)

giving

√

h = R

2

sin θ. The volume (we are calling the surface area a gener-

alized volume) is

V =

√

hd

2

x =

√

hdθdφ +R

2

π

0

sin theta

2π

0

dφ = 4πR

2

(5.89)

Actually there are three 2-d spaces which are homogenous and isotropic.

There are 1) the ﬂat x − y plane (R

2

), ii) the positively curved two sphere

(S

2

) and iii) the negatively curved two hyperbola (H

2

).

As before we can obtain the formula for H

2

with the replacement

R →iR (5.90)

to yield

d

2

= R

2

[

dr

2

1 +r

2

+r

2

dθ

2

]. (5.91)

5.1. PREVIEW OF RIEMANNIAN GEOMETRY 95

This result is also obtained by embedding in Minkowski space. The metric

corresponding to (5.81) is ([21], Pg.34, equation 2.17)

d

2

= R

2

(dθ

2

+ sin

2

θdφ

2

). (5.92)

We can collect our results for R

2

, S

2

and H

2

into a single formula

d

2

= R

2

(

dr

2

1 −kr

2

+r

2

dθ

2

) (5.93)

where k=0, +1, -1 for ﬂat (R

2

), closed (S

2

) and open (H

2

) surfaces respec-

tively.

The volume can be alternatively calculated using d

2

= h

ij

dx

i

dx

j

in

(5.93) we have

h

ij

=

R

2

1−kr

2

0

0 R

2

r

2

(5.94)

giving the determinant

h =

R

2

r

2

1 −kr

2

(5.95)

or

√

h =

R

2

r

√

1−kr

2

. The volume is

V =

√

hd

2

x =

√

h = R

2

0

0

rdr

√

1 −kr

2

2π

0

dθ (5.96)

= 2πR

2

0

0

rdr

√

1 −kr

2

. (5.97)

The limits of integration

0

0

dr can be clearly seen from Fig 2.1 of Kolb and

Turner [21], Pg.32. What this really means is

0

0

dr ≡ 2

R

0

dr (5.98)

where r = 0 at θ = 0 and r = R at θ =

π

2

and r = 0 again at θ = π.

(See bottom Pg.179 and top pg.180 of the book by Lawden [?] for more

explanation)

The integral and its limits are more clearly done with the substitution

√

kr ≡ sin χ (5.99)

96 CHAPTER 5. EINSTEIN FIELD EQUATIONS

when

χ : 0 → π. (5.100)

Thus

π

0

dχ is equivalent to

0

0

dr = 2

R

0

dr. Thus the volume in equation

(5.97) becomes

V = 2πR

2

π

0

1

√

k

1

√

k

sin χdχ (5.101)

giving

V =

4πR

2

k

. (5.102)

Thus for k = +1 we have V = 4πR

2

as before. For k = 0 we have V = ∞

and for k = −1 we need to do the integral again. We would ﬁnd V = ∞ for

k = −1.

5.1.6 3-dimensional Hypersurface

Proceeding upwards in our number of dimensions we might inquire a ”curved

volume”. But the curvature can really only be imagined with respect to

embedding in a 4-dimensional Euclidean space E

4

. We call the 4-d space as

a hypersurface.

Our 4-d Euclidean space (into which we will embed the hypersurface)

has length element

d

2

= dx

2

1

+dx

2

2

+dx

2

3

+dx

2

4

(5.103)

(it is d

2

and not ds

2

).

A hypersurface is represented by three parameters u, v, w and is ex-

pressed as x = x(u, v, w). Introduce an extra ﬁcticious coordinate for the

three sphere S

3

as

x

2

+y

2

+z

2

+w

2

= R

2

x

2

1

+x

2

2

+x

2

3

+x

2

4

= R

2

(5.104)

which restricts x

4

as

x

2

4

= R

2

−x

2

1

−x

2

2

−x

2

3

. (5.105)

Writing

dx

4

=

∂x

4

∂x

1

dx

1

+

∂x

4

∂x

2

dx

2

+

∂x

4

∂x

3

dx

3

(5.106)

5.1. PREVIEW OF RIEMANNIAN GEOMETRY 97

and with

∂x

4

∂x

1

=

−x

1

√

R

2

−x

2

1

−x

2

2

−x

2

3

etc. we have

dx

4

= −

x

1

dx

1

+x

2

dx

2

+x

3

dx

3

R

2

−x

2

1

−x

2

2

−x

2

3

(5.107)

or

dx

2

4

=

(x

1

dx

1

+x

2

dx

2

+x

3

dx

3

)

2

R

2

−x

2

1

−x

2

2

−x

2

3

(5.108)

to give ([21], Pg.34)

d

2

= dx

2

1

+dx

2

2

+dx

2

3

+

(x

1

dx

1

+x

2

dx

2

+x

3

dx

3

)

2

R

2

−x

2

1

−x

2

2

−x

2

3

. (5.109)

Let us introduce spherical polar coordinates in the x

4

hyperplane as

x

1

= r

**sin θ cos φ (5.110)
**

x

2

= r

**sin θ sin φ (5.111)
**

x

3

= r

cos θ (5.112)

where

θ : 0 → π (5.113)

φ : 0 → 2π (5.114)

r

: 0 → R. (5.115)

Using

dx

i

=

∂x

i

∂r

dr

. +

∂x

i

∂θ

dθ +

∂x

i

∂φ

dφ (5.116)

then (5.109) becomes

d

2

=

R

2

R

2

−r

2

dr

2

+r

2

dθ

2

+r

2

sin

2

θdφ

2

(5.117)

Introducing the dimensionless coordinate

r ≡

r

R

(5.118)

gives

d

2

= R

2

[

dr

2

1 −r

2

+r

2

dθ

2

+r

2

sin

2

θdφ

2

] (5.119)

98 CHAPTER 5. EINSTEIN FIELD EQUATIONS

where

r : 0 → 1. (5.120)

Another convenient coordinate system for the three sphere uses angles

χ, θ, φ from 4-dimensional hyperspherical polar coordinates speciﬁed via

x = Rsin χsin θ cos φ (5.121)

y = Rsin χsin θ sin φ (5.122)

z = Rsin χcos θ (5.123)

w = Rcos χ. (5.124)

Substituting into (5.119) directly yields

d

2

= R

2

[dχ

2

+ sin

2

χ(dθ

2

+ sin

2

θdφ

2

)]. (5.125)

Using d

2

= h

ij

dx

i

dx

j

we have

h

ij

=

¸

¸

R

2

0 0

0 R

2

sin

2

χ 0

0 0 R

2

sin

2

χsin

2

θ

¸

(5.126)

The determinant is

h = R

6

sin

4

χsin

2

θ (5.127)

giving

√

h = R

3

sin

2

χsin θ. The volume is

V =

√

hd

3

x =

√

hdχdθdφ

= R

3

π

0

sin

2

χdχ

π

0

sin θdθ

2π

0

dφ (5.128)

where the limits

π

0

dχ are the same as in the previous section. Thus

V = 4πR

3

π

0

sin

2

χdχ = 4πR

3

[

χ

2

−

sin 2χ

4

]

π

0

(5.129)

giving

V = 2π

2

R

3

(5.130)

for the volume of our hypersphere. Compare this to the volume of a Eu-

clidean sphere

4

3

πR

3

.

For a ﬂat, open and closed hyperspheres the metric is

d

2

= R

2

[

dr

2

1 −kr

2

+r

2

dθ

2

+r

2

sin

2

θdφ

2

] (5.131)

5.2. FRIEDMANN-ROBERTSON-WALKER METRIC 99

The volume can be calculated alternatively. Using d = h

ij

dx

i

dx

j

in

(5.131) we have

h

ij

=

¸

¸

R

2

1−kr

2

0 0

0 R

2

r

2

0

0 0 R

2

r

2

sin

2

θ

¸

(5.132)

giving the determinant h =

R

6

r

4

sin

2

θ

1−kr

2

. The volume is

V =

√

hd

3

x =

√

hdrdθdφ

= R

3

0

0

r

2

dr

√

1 −kr

2

π

0

sin θdθ

2π

0

dφ

= 4πR

3

0

0

r

2

dr

√

1 −kr

2

. (5.133)

The limits of integration are the same as discussed in equation (5.98).

Using the substitution

√

kr ≡ sin χ

(5.134)

with

χ

: 0 → π (5.135)

becuase χ : 0 → π. Thus the volume is

V = 4πR

3

π

0

1

k

3/2

sin

2

χdχ (5.136)

giving

V =

2π

2

R

3

k

3/2

. (5.137)

For k = +1 this agrees with our result before.

5.2 Friedmann-Robertson-Walker Metric

The metric of Special Relativity is

ds

2

= c

2

dt −(dx

2

dy

2

dz

2

). (5.138)

Clearly the spatial part is a 3-d Euclidean ﬂat space. We have seen that the

spatial metric for a homogeneous, isotropic curved space with a size R(t)

that can change in time is

ds

2

= R

2

(t)[

dr

2

1 −kr

2

+r

2

(dθ

2

+ sin

2

θdφ

2

)]. (5.139)

100 CHAPTER 5. EINSTEIN FIELD EQUATIONS

Replacing the spatial part of the special relativity metric with (5.139) we

have the Friedmann-Robertson-Walker (FRW) metric [13]

ds

2

= c

2

dt

2

−R

2

(t)[

dr

2

1 −kr

2

+r

2

(dθ

2

+ sin

2

θdφ

2

)] (5.140)

where R(t) is called the scale factor and the constant k can be 0, ±1 depend-

ing on the curvature. This can also be derived with the use of Killing vectors

[13].

Writing ds

2

≡ g

µν

dx

µ

dx

ν

and identifying

x

0

= ct (5.141)

x

1

= r (5.142)

x

2

= θ (5.143)

x

3

= φ (5.144)

we have

g

00

= 1 (5.145)

g

11

=

−R

2

1 −kr

2

(5.146)

g

22

= −R

2

r

2

(5.147)

g

33

= −R

2

r

2

sin

2

θ (5.148)

Deﬁning the determinant

g ≡ detg

µν

= g

00

g

11

g

22

g

33

(5.149)

= −

R

6

r

4

sin

2

θ

1 −kr

2

(5.150)

(Note that this is not g = detg

µν

.) Thus

√

−g =

R

3

r

2

sin θ

√

1 −kr

2

. (5.151)

If g

µν

is represented by a matrix [g

µν

], then we found previously that

g

µν

is just the inverse of this metric namely [g

µν

]

−1

. For a diagonal matrix

(which we have for the FRW metric) each matrix element is simply given by

g

µν

=

1

gµν

. Thus it’s easy to get

g

00

= 1 (5.152)

5.2. FRIEDMANN-ROBERTSON-WALKER METRIC 101

g

11

= −

(1 −kr

2

)

R

2

(5.153)

g

22

=

−1

R

2

r

2

(5.154)

g

33

=

−1

R

2

r

2

sin

2

θ

. (5.155)

5.2.1 Christoﬀel Symbols

We now calculate the Christoﬀel symbols using equation (3.69). Fortunately

we need not calculate all of them. We can use the symmetry Γ

α

βγ

= Γ

α

γβ

to

shorten the job. We have

Γ

α

βγ

≡

1

2

g

α

(g

β,γ

+g

αγ,β

−g

βγ,

) = Γ

α

γβ

=

1

2

g

αα

(g

αβ,γ

+g

αγ,β

−g

βγ,α

) (5.156)

which follows because g

α

= 0 unless = α. (g

µν

is a diagonal matrix for

the FRW metric.) The only non-zero Christoﬀel symbols are the following:

Γ

0

11

=

1

2

g

00

(g

01,1

+g

01,1

−g

11,0

) = −

1

2

g

11,0

because g

01

= 0 and g

00

= 1. This becomes (let’s now set c ≡ 1)

Γ

0

11

= −

1

2

g

11,0

= −

1

2

∂

∂t

(

−R

2

1 −kr

2

)

=

1

2

1

1 −kr

2

∂R

2

∂t

=

2R

˙

R

2(1 −kr

2

)

=

R

˙

R

1 −kr

2

(5.157)

because r = r(t) and R = R(t). Proceeding

Γ

0

22

= −

1

2

g

22,0

= −

1

2

∂

∂t

(−R

2

r

2

) = r

2

R

˙

R (5.158)

Γ

0

33

= r

2

sin

2

θR

˙

R (5.159)

Γ

1

11

=

kr

1 −kr

2

(5.160)

Γ

1

22

= −r(1 −kr

2

) (5.161)

Γ

1

33

= −r(1 −kr

2

) sin

2

θ (5.162)

102 CHAPTER 5. EINSTEIN FIELD EQUATIONS

Γ

2

12

= Γ

3

13

=

1

r

(5.163)

Γ

2

33

= −sin θ cos θ (5.164)

Γ

3

23

= cot θ (5.165)

Γ

1

01

= Γ

2

02

= Γ

3

03

=

˙

R

R

(5.166)

(do Problems 5.2 and 5.3)

5.2.2 Ricci Tensor

Using equation (??) we can now calculate the Ricci tensor. For the FRW

metric it turns out that R

µν

= 0 for µ = ν, so that the non-zero components

are R

00

, R

11

, R

22

, R

33

. Proceeding we have

R

00

=

1

√

−g

(Γ

00

√

−g), −(ln

√

−g)

,00

−Γ

0θ

Γ

θ

0

but Γ

00

= 0 giving

R

00

= −(ln

√

−g)

,00

−Γ

0

0θ

Γ

θ

00

−Γ

1

0θ

Γ

θ

01

−Γ

2

0θ

Γ

θ

02

−Γ

3

0θ

Γ

θ

03

when we have performed the sum over . The term Γ

θ

0θ

= 0. In the last

three terms we have Γ

α

0θ

where α = 1, 2, 3. Now Γ

α

0θ

= 0 for θ = α, so that

we must have θ = 1, 2, 3 in the third, forth and ﬁfth terms respectively. Also

the second term contains Γ

0

0θ

which is always 0. Thus

R

00

= −(ln

√

−g)

,00

−Γ

1

10

Γ

1

01

−Γ

2

02

Γ

2

02

−Γ

3

03

Γ

3

03

= −(ln

√

−g)

,00

−(Γ

1

01

)

2

−(Γ

2

02

)

2

−(Γ

3

03

)

2

= −(ln

√

−g)

,00

−3(

˙

R

R

)

2

Now

(

√

−g)

,0

∂

√

−g

∂x

=

∂

√

−g

∂t

=

r

2

sin θ

√

1 −kr

2

∂R

3

∂t

=

r

2

sin θ

√

1 −kr

2

3R

2

˙

R

and

(ln

√

−g)

,µ

≡

∂ ln

√

−g

∂x

µ

=

∂ ln

√

−g

∂

√

−g

∂

√

−g

∂x

µ

=

1

√

−g

∂

√

−g

∂x

µ

5.2. FRIEDMANN-ROBERTSON-WALKER METRIC 103

so that

(ln

√

−g)

,0

=

1

−

√

−g

∂

√

−g

∂x

0

=

√

1 −kr

2

R

3

r

2

sin θ

R

2

sin θ

√

1 −kr

2

3R

2

˙

R = 3

˙

R

R

giving

(ln

√

−g)

,00

= 3

∂

∂t

(

˙

R

R

) = 3

R

¨

R −

˙

R

2

R

2

= 3

¨

R

R

−3(

˙

R

R

)

2

.

We ﬁnally have

R

00

= −3

˙

R

R

. (5.167)

One can similaraly show that

R

11

=

R

¨

R −2

˙

R

2

+ 2k

1 −kr

2

(5.168)

R

22

= r

2

(R

¨

R + 2

˙

R

2

+ 2k) (5.169)

R

33

= r

2

sin

2

θ(R

˙

R + 2

˙

R

2

+ 2k) (5.170)

(do Problem 5.4)

5.2.3 Riemann Scalar and Einstein Tensor

We now calculate the Ricci scalar R≡ RR

α

α

≡ g

αβ

R

αβ

. The only non-zero

contributions are

R = g

00

R

00

+g

11

R

11

+g

22

R

22

+g

33

R

33

(5.171)

= −6[

¨

R

R

+ (

˙

R

R

)

2

+

k

R

2

] (5.172)

(do Problem 5.5). Finally we calculate the Einstein tensor G

µν

≡ R

µν

−

1

2

Rg

µν

. The only non-zero component are for µ = ν. We obtain

G

00

= 3[(

˙

R

R

)

2

+

k

R

2

] (5.173)

G

11

=

−1

1 −kr

2

(2

¨

RR +

˙

R

2

+k) (5.174)

G

22

= −r

2

(2

¨

RR +

˙

R

2

+k) (5.175)

G

33

= −r

2

sin

2

θ(2

¨

RR +

˙

R

2

+k) (5.176)

(do Problem 5.6).

104 CHAPTER 5. EINSTEIN FIELD EQUATIONS

5.2.4 Energy-Momentum Tensor

For a perfect ﬂuid the energy momentum tensor is given in equation (4.26)

as

T

µν

= (ρ +p)u

µν

−pη

µν

(5.177)

The tensor for T

µν

is written is (4.28) for the metric of Special Relativity.

For an arbitrary metric in General Relativity we have

T

µν

= (ρ +p)u

µν

−pg

µν

(5.178)

where we shall use g

µν

from our FRW model. For a motionless ﬂuid recall

that u

µ

= (c, 0) or U

µ

= (c, −0) = (c, 0) = (1, 0) for c ≡ 1. Thus

T

00

= ρ +p −p = ρ (5.179)

and

T

ii

= −pg

ii

(5.180)

because u

i

= 0. Upon substitution of the FRW values for the metric given

is equations (5.145)-(5.148) we have

T

µν

=

¸

¸

¸

¸

ρ 0 0 0

0 p

R

2

1−kr

2

0 0

0 0 pR

2

r

2

0

0 0 0 pR

2

r

2

sin

2

θ

¸

(5.181)

5.2.5 Friedmann Equations

Finally we substitute our results into the Einstein ﬁeld equations G

µν

=

8πGT

µν

+ Λg

µν

.

The µν = 00 component is

3[(

˙

R

R

)

2

+

k

R

2

] = 8πGρ + Λ

giving

H

2

≡ (

˙

R

R

)

2

=

8πG

3

ρ −

k

R

2

+

Λ

3

. (5.182)

The µν = 11 component is

−1

1 −kr

2

(2

¨

RR +

˙

R

2

+k) = 8πGp

R

2

1 −kr

2

+ Λ

−R

2

1 −kr

2

5.3. PROBLEMS 105

giving

2

¨

R

R

+ (

˙

R

R

)

2

+

k

R

2

= −8πGp + Λ.

But we now use our previous result (5.182) to give

2

¨

R

R

+

8πG

3

ρ +

Λ

3

= −8πGp + Λ

to ﬁnally give

¨

R

R

=

4πG

3

(ρ + 3p) +

Λ

3

(5.183)

(do Problem 5.7).

5.3 Problems

5.1 For the FRW metric show that Γ

1

33

−(1 −kr

2

)r sin

2

θ and Γ = cot θ.

5.2 Show that, for example, Γ

1

22

= Γ

2

23

= 0 for the FRW metric.

5.3 Show that R

22

= r

2

(R

¨

R + 2 cot R

2

+ 2k) for the FRW metric.

5.4 Show that the Ricci scalar is R= −6[

¨

R

R

+(

˙

R

R

)

2

+

k

R

2

] for the FRW metric.

5.5 Calculate G

µν

for the FRW metric.

5.6 Show that the µν = 22 and µν = 33 components of the Einstein’s

equations for the FRW metric yield the same equation (5.63) as the µν = 11

component.

106 CHAPTER 5. EINSTEIN FIELD EQUATIONS

Chapter 6

Einstein Field Equations

G

µν

= kT

µν

go through history e.g. he ﬁrst tried R

µν

= kT

µν

etc

107

108 CHAPTER 6. EINSTEIN FIELD EQUATIONS

Chapter 7

Weak Field Limit

derivation of G

µν

= kT

µν

from equiv princ.

109

110 CHAPTER 7. WEAK FIELD LIMIT

Chapter 8

Lagrangian Methods

Lagrangians for G

µν

etc.

(NNNN have assumed special relativity g

00

= +1) (NNN φ term seems

to disagree with Kolb and Turner Pg. 276 eqn 8.20).

111

112 CHAPTER 8. LAGRANGIAN METHODS

Bibliography

[1] J.B. Marion, Classical Dynamics of Particles and Systems, 3rd ed.,

(Harcourt, Brace, Jovanovich College Publishers, New York, 1988).

QA845 .M38

[2] J.Foster and J.D Nightingale, A Short Course in General Relativity,

2nd ed., (Springer-Verlag, 1995). QC173.6 .F67

[3] S. Gasiorowicz, Quantum Physics, (Wiley, New York, 1996).

[4] H.A. Atwater, Introduction to General Relativity, (Pergamon, New

York, 1974).

[5] B.F. Schutz, A First Course in General Relativity, (Cambridge Univer-

sity Press, New York, 1990). QC173.6.S38

[6] B.F. Schutz, Geometrical methods of mathematical physics, (Cambridge

University Press, New York, 1980). QC20.7.D52

[7] J.V. Narlikar, Introduction to cosmology, 2nd ed., (Cambridge Univer-

sity Press, New York, 1993). QB981.N3

[8] H. Ohanian and R. Ruﬃni, A Gravitation and Spacetime, 2nd ed.,

(W.W. Norton and Company, New York, 1994). QC178 .O35

[9] J.L. Martin, General Relativity, (Ellis Horwood Limited, Chichester,

England, 1988).

[10] A. Guth and P. Steinhardt, The Inﬂationary Universe, in The New

Physics, edited by P. Davies, (Cambridge University Press, New York,

1989).

[11] D. Atkatz, American J. Phys. 62, 619 (1994).

113

114 BIBLIOGRAPHY

[12] R. Kubo, Statistical Mechanics, (North-Holland, Amsterdam, 1967).

[13] J.N. Islam, An Introduction to Mathematical Cosmology, (Cambridge

University Press, New York, 1992).

[14] H.A. Atwater, Introduction to General Relativity, (Pergamon, New

York, 1974).

[15] R. Adler, M. Bazin, M. Schiﬀer, Introduction to General Relativity,

(McGraw-Hill, New York, 1975).

[16] T. Jacobson, Phys. Rev. Lett. 75, 1260 (1995)

[17] W. Freedman et al, Nature D 371, 757 (1994).

[18] L.M. Krauss and M.S. Turner, The cosmological constant is back, Gen-

eral Relativity and Gravitation, 27, 1137 (1995).

[19] A. Guth, Phys. Rev. D 23, 347 (1981).

[20] J. Hartle and S. Hawking, Phys. Rev. D 28, 2960 (1983).

[21] E.W. Kolb and M.S. Turner, The Early Universe, (Addison-Wesley,

1990).

[22] D. Atkatz and H. Pagels, Phys. Rev. D 25, 2065 (1982).

[23] F.W. Byron and Fuller, Mathematics of Classical and Quantum Physics,

vols. 1 and 2,

(Addison-Wesley, Reading, Masachusetts, 1969). QC20.B9

[24] G.B. Arfken and H.J. Weber, Mathematical Methods for Physicists, 4th

ed., (Academic Press, San Diego, 1995). QA37.2.A74

[25] H.C. Ohanian, Classical Electrodynamics, (Allyn and Bacon, Boston,

1988). QC631.O43

[26] J.D. Jackson, Classical Electrodynamics, (Wiley, New York, 1975).

QC631.J3

[27] J.B. Marion, Classical Electromagnetic Radiation, (Academic Press,

New York, 1965). QC631.M37

[28] T. L. Chow, General Relativity and Cosmology, (Wuerz Publishing Ltd.,

Winnipeg, Canada, 1994).

BIBLIOGRAPHY 115

[29] M.S. Madsen, The Dynamic Cosmos, (Chapman and Hall, New York,

1995).

[30] H. Muirhead, The physics of elementary particles, (Pergamon Press,

New York, 1965). QC721.M94

[31] M. Leon, Particle physics: an introduction, (Academic Press, New York,

1973). QC793.2.L46

[32] R. D’Inverno, Introducing Einstein’s Relativity, (Clarendon Press, Ox-

ford, 1993).

[33] P.J.E. Peebles, Principles of Physical Cosmology, (Princeton University

Press, Princeton, New Jersey, 1993).

[34] B.D. Serot and J. D. Walecka, The Relativistic Nuclear ManyBody Prob-

lem, Advances in Nuclear Physics, 16, 1 (1986). edited by J.W. Negele

and E. Vogt (Plenum Press, New York, 1986).

[35] J. D. Barrow, Cosmic No-Hair Theorem and Inﬂation, Physics

Lett. B, 187, 112, (1987)

[36] A. D. Linde, Inﬂation and Quantum Cosmology, (Academic Press, New

York, 1990)

[37] P. D. B. Collins, A. D. Martin and E. J. Squires, Particle Physics and

Cosmology, (Wiley, New YOrdk, 1987)

[38] A. D. Dolgov, M. V. Sazhin and Y. B. Zeldovich, Basis of Modern

Cosmology (Editions Fronti` eres, B.P.33, 91192 Gif-Sur-Yvette Cedex,

France, 1990)

Contents

1 NEWTONIAN COSMOLOGY 1.1 Introduction . . . . . . . . . . . . . . 1.2 Equation of State . . . . . . . . . . . 1.2.1 Matter . . . . . . . . . . . . . 1.2.2 Radiation . . . . . . . . . . . 1.3 Velocity and Acceleration Equations 1.4 Cosmological Constant . . . . . . . . 1.4.1 Einstein Static Universe . . . 5 5 5 6 6 7 9 11 13 13 14 15 16 16 17 18 19 20 21 23 23 26 27 28 30 31

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

. . . . . . .

2 APPLICATIONS 2.1 Conservation laws . . . . . . . . . . . . . . . 2.2 Age of the Universe . . . . . . . . . . . . . . 2.3 Inﬂation . . . . . . . . . . . . . . . . . . . . . 2.4 Quantum Cosmology . . . . . . . . . . . . . . 2.4.1 Derivation of the Schr¨dinger equation o 2.4.2 Wheeler-DeWitt equation . . . . . . . 2.5 Summary . . . . . . . . . . . . . . . . . . . . 2.6 Problems . . . . . . . . . . . . . . . . . . . . 2.7 Answers . . . . . . . . . . . . . . . . . . . . . 2.8 Solutions . . . . . . . . . . . . . . . . . . . . 3 TENSORS 3.1 Contravariant and Covariant Vectors 3.2 Higher Rank Tensors . . . . . . . . . 3.3 Review of Cartesian Tensors . . . . . 3.4 Metric Tensor . . . . . . . . . . . . . 3.4.1 Special Relativity . . . . . . . 3.5 Christoﬀel Symbols . . . . . . . . . . 1

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

. . . . . .

2 3.6 3.7 3.8 3.9 3.10 3.11 Christoﬀel Symbols and Metric Tensor Riemann Curvature Tensor . . . . . . Summary . . . . . . . . . . . . . . . . Problems . . . . . . . . . . . . . . . . Answers . . . . . . . . . . . . . . . . . Solutions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

CONTENTS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 36 38 39 40 41 42

4 ENERGY-MOMENTUM TENSOR 45 4.1 Euler-Lagrange and Hamilton’s Equations . . . . . . . . . . . 45 4.2 Classical Field Theory . . . . . . . . . . . . . . . . . . . . . . 47 4.2.1 Classical Klein-Gordon Field . . . . . . . . . . . . . . 48 4.3 Principle of Least Action . . . . . . . . . . . . . . . . . . . . 49 4.4 Energy-Momentum Tensor for Perfect Fluid . . . . . . . . . . 49 4.5 Continuity Equation . . . . . . . . . . . . . . . . . . . . . . . 51 4.6 Interacting Scalar Field . . . . . . . . . . . . . . . . . . . . . 51 4.7 Cosmology with the Scalar Field . . . . . . . . . . . . . . . . 53 4.7.1 Alternative derivation . . . . . . . . . . . . . . . . . . 55 4.7.2 Limiting solutions . . . . . . . . . . . . . . . . . . . . 56 4.7.3 Exactly Solvable Model of Inﬂation . . . . . . . . . . . 59 4.7.4 Variable Cosmological Constant . . . . . . . . . . . . . 61 4.7.5 Cosmological constant and Scalar Fields . . . . . . . . 63 4.7.6 Clariﬁcation . . . . . . . . . . . . . . . . . . . . . . . . 64 4.7.7 Generic Inﬂation and Slow-Roll Approximation . . . . 65 4.7.8 Chaotic Inﬂation in Slow-Roll Approximation . . . . . 67 4.7.9 Density Fluctuations . . . . . . . . . . . . . . . . . . . 72 4.7.10 Equation of State for Variable Cosmological Constant 73 4.7.11 Quantization . . . . . . . . . . . . . . . . . . . . . . . 77 4.8 Problems . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80 5 EINSTEIN FIELD EQUATIONS 5.1 Preview of Riemannian Geometry . . . . . . 5.1.1 Polar Coordinate . . . . . . . . . . . 5.1.2 Volumes and Change of Coordinates 5.1.3 Diﬀerential Geometry . . . . . . . . 5.1.4 1-dimesional Curve . . . . . . . . . . 5.1.5 2-dimensional Surface . . . . . . . . 5.1.6 3-dimensional Hypersurface . . . . . 5.2 Friedmann-Robertson-Walker Metric . . . . 5.2.1 Christoﬀel Symbols . . . . . . . . . . 83 84 84 85 88 89 92 96 99 101

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . . . .

. . . . . . .5 Friedmann Equations . . .2 Ricci Tensor . . . . .2. . . . . . .2. . . . . . . . . . . . . . . . . . . . . . . . . . . . Problems . . . .
3 102 103 104 104 105 107 109 111
5. . . . . . . . .CONTENTS 5. . . . . . . .3
6 Einstein Field Equations 7 Weak Field Limit 8 Lagrangian Methods
. .2. . . . . .4 Energy-Momentum Tensor . 5. . . .2. . . . . . . . . . . . . . . . . 5. . .3 Riemann Scalar and Einstein Tensor 5. .

4
CONTENTS
.

The equations describing the velocity (called the Friedmann equation) and acceleration of the universe are derived from Newtonian mechanics and also the cosmological constant is introduced within a Newtonian framework. is being used.2
Equation of State
In what follows the equation of state for non-relativistic matter and radiation will be needed. The present chapter represents an attempt to do this based entirely on Newtonian mechanics. In particular an expression for the rate of change of density.1 Introduction
Many of the modern ideas in cosmology can be explained without the need to discuss General Relativity.
1.1)
where U is the internal energy.Chapter 1
NEWTONIAN COSMOLOGY
1. the age of the universe and the inﬂation. The equations of state are also derived in a very simple way. Ignoring any heat transfer and writing dW = F dr = pdV where F is the 5
. (The deﬁnition ˙ x ≡ dx . ρ. where t is time.) The ﬁrst law of thermodynamics is ˙ dt dU + dW = dQ (1. Applications such as conservation laws. will be needed in terms of the density ρ and pressure p. W is the work and Q is the heat transfer. radiation and matter dominated epochs are discussed.

9)
. Writing ˙ V ˙ V ∝ r3 implies that V = 3 r . p is the pressure and V is the volume. Thus r ρ = −3(ρ + p) ˙ r ˙ r (1.5)
1. r is the distance.5).7) dr r so that by comparing to equation (1. it follows that the equation of state for matter is p = 0.2).
1.2. so we are talking about zero temperature. (1.2)
Assuming that ρ is a relativistic energy density means that the energy is expressed as U = ρV (1. For a standing wave on a string ﬁxed at both ends L= nλ 2 (1. then dU = −pdV.6
CHAPTER 1.2.8) ρ≡ ˙ This is the same as obtained from the ideal gas law for zero temperature. Recall that in this derivation we have not introduced any kinetic energy.3) from which it follows that ˙ ˙ ˙ U = ρV + ρV = −pV ˙ (1. (1.1
Matter
Writing the density of matter as ρ= it follows that M
4 3 3 πr
(1.4)
where the term on the far right hand side results from equation (1.6)
dρ r ˙ r = −3ρ ˙ (1.2
Radiation
The equation of state for radiation can be derived by considering radiation modes in a cavity based on analogy with a violin string [12]. NEWTONIAN COSMOLOGY
force.

is obtained by writing the total energy E as the sum of kinetic plus potential energy terms (and using M = 4 πr3 ρ ) 3 1 Mm 8πG 1 E = T + V = mr2 − G ˙ = mr2 (H 2 − ρ) 2 r 2 3 (1. 2 L
(1.15)
˙ where the Hubble constant H ≡ r . the radiation equation of state is 1 p = ρ. These equations of state are needed in order to discuss the radiation and matter dominated epochs which occur in the evolution of the Universe.
. 3. 3 It is customary to combine the equations of state into the form p= γ ρ 3 (1.2) the pressure becomes p≡− dU 1U = . so that c = fλ = f 2L n (1. Radiation travels at the velocity of light.14) (1. VELOCITY AND ACCELERATION EQUATIONS
7
where L is the length of the string.
1. dV 3V (1. λ is the wavelength and n is a positive integer (n = 1.. where h is Planck’s constant. 2. which speciﬁes the speed of recession.). m is the mass of a test particle in the r potential energy ﬁeld enclosed by a gas of dust of mass M . Thus substituting f = 2L c into Planck’s formula U = hω = hf .10)
n where f is the frequency.3..13)
where γ ≡ 1 for radiation and γ ≡ 0 for matter..12)
Using ρ = U/V .11)
Using equation (1..1.3
Velocity and Acceleration Equations
The Friedmann equation. r is the distance from the center of the dust to the test particle and G is Newton’s constant. gives ¯
U=
nhc 1 ∝ V −1/3 .

This is discussed later. where g is the acceleration due to gravity. For a particle in motion near the Earth this would correspond to the particle escaping (unbound). Later this will be analagous to an open.
(1. The constant k can either be negative. general relativistic point of view this corresponds to an open.8
CHAPTER 1. zero or negative. Even though this equation is derived for matter. The scale factor is introduced because in General Relativity it is space itself which expands [19]. k can be rescaled so that instead of being negative. zero or positive it takes on the values −1. (In fact it is also true for vacuum. it follows that giving the Friedmann equation ˙ R 8πG k H 2 ≡ ( )2 = ρ− 2 R 3 R
=
and
=
¨ R R.16)
with k ≡ − 2E . From a geometric.) Exactly the same equation is obtained from the general relativistic Einstein ﬁeld equations [13]. where Λ is the cosmological constant and ρvac is the vacuum energy density which just replaces the ordinary density. orbiting (critical case) or returning (bound) to Earth because the speed r would be greater.15) is re-arranged as H2 = 8πG 2E .18)
which speciﬁes the speed of recession.17)
2E Deﬁning k ≡ − ms2 and writing the distance in terms of the scale factor R
and a constant length s as r(t) ≡ R(t)s.
so that
(1. ﬂat or closed universe. NEWTONIAN COSMOLOGY
2GM r
Recall that the escape velocity is just vescape = the above equation can also be written
2 r2 = vescape − k 13 − 2 ˙
=
8πG 2 3 ρr . ﬂat or closed universe. According to Guth [10]. The derivation shown above is exactly analagous to such a calculation.13 − 3 ρ+ 3 mr2
r ˙ r ˙ R R r ¨ r
(1. where F is the force r
. critical or bound trajectories as mentioned above. Equation (1. equal to or smaller than the escape ˙ speed vescape . From a Newtonian point of view this corresponds to unbound. In elementary mechanics the speed v of a ball dropped from a height r √ is evaluated from the conservation of energy equation as v = 2gr. with Λ ≡ 8πGρvac . 0 or +1. Similarly the acceleration a of the ball is calculated as a = g from Newton’s equation F = m¨. it is also true for radiation. zero or positive m corresponding to the total energy E being positive.

Exactly the same equation is obtained from the Einstein ﬁeld equations [13].
1. dt2
9
The acceleration for the universe is obtained Mm = m¨.19)
gives the acceleration equation (1.4
Cosmological Constant
In both Newtonian and relativistic cosmology the universe is unstable to gravitational collapse. R = 2RR = dt 3 dt Upon using equation (1. In order to obtain a possibly zero acceleration. COSMOLOGICAL CONSTANT and the acceleration is r ≡ ¨ from Newton’s equation
d2 r .20) for the matter equation of state (γ = 0). and Newton could have done exactly the same thing. a positive term (conventionally taken as Λ ) is added to the acceleration equation (1.1.22) (1. Both Newton and Einstein believed that the Universe is static. it can also be obtained by taking the time derivative of the energy equation to give a = dv = v dv = dt dr √ √ 1 ( 2gr)( 2g 2√r ) = g. called the cosmological constant.22) as 3 ¨ R 4πG Λ =− (ρ + 3p) + R 3 3 (1. In order to obtain this Einstein introduced a repulsive gravitational force. for the general case one can take the time derivative of equation (1.13 − 5 r r2 =
¨ R R
−G Again using M = 4 πr3 ρ and 3
r ¨ r
(1. it is clear that this acceleration 3 equation holds only for matter. Similarly.5) the acceleration equation is obtained as ¨ R 4πG 4πG =− (ρ + 3p) = − (1 + γ)ρ R 3 3 (1.4.21)
which reduces to equation (1. had he believed the universe to be ﬁnite.20)
¨ F r ¨ R 4πG ≡ ≡ =− ρ.23)
. mr r R 3
However because M = 4 πr3 ρ was used.18) (valid for matter and radiation) d ˙2 ˙ ¨ 8πG d (ρR2 ). In our example of the falling ball instead of the acceleration being obtained from Newton’s Law.

The problem is to imagine that a hole has been drilled from one side of the Earth.24) r R mr 3 and using Λ dV Frepulsive = mr ≡ − (1.28)
Equations (1. One is to show that if a ball is dropped into the hole. Substituting this into the conservation of energy equation 1 M m 1 Λ 2 1 2 2 8πG Λ E = T + V = mr2 − G ˙ − mr = mr (H − ρ − ) (1. Let us comment on the repulsive harmonic oscillator obtained above. Thus the gravitational force for a continuous mass distribution rises like Hooke’s law and thus oscillatory solutions are encountered. Exactly the same equations are obtained from the Einstein ﬁeld equations [13].27) 2 r 23 2 3 3 gives ˙ R 8πG Λ k H 2 ≡ ( )2 = ρ− 2 + . Recall one of the standard problems often assigned in mechanics courses. What has been done here is entirely equivalent to just adding a repulsive gravitational force in Newton’s Law. yet if one r of the masses is a continous mass distribution represented by a density then F = G 4 πρmr.10
CHAPTER 1. Identifying the force from ¨ Frepulsive r ¨ R Λ = ≡ ≡ (1. The force rises linearly as the distance is increased because 3 the amount of matter enclosed keeps increasing. The solution is obtained by noting that whereas gravity is an inverse square law for point m masses M and m separated by a distance r as given by F = G M2 . NEWTONIAN COSMOLOGY
which. with the proper choice of Λ can give the required zero acceleration for a static universe.26)
which is just a repulsive simple harmonic oscillator. it will execute harmonic motion. This sheds light on our repulsive oscillator found
.28) and (1.25) 3 dr gives the potential energy as Vrepulsive = − 1Λ 2 mr 23 (1. through the center and to the other side.23) constitute the fundamental equations of motion that are used in all discussions of Friedmann models of the Universe.18). The question now is how this repulsive force enters the energy equation (1. R 3 R 3 (1. Again exactly the same equation is obtained from the Einstein ﬁeld equations [13].

From equation (1. Λ = 0.32) into equation (1.29)
1. 15]. In this case we want the gravity to be repulsive.34) R0
. (1.4. Substituting equation (1.31) this implies Λ ρ= (1. ¨ = 0 will be examined ﬁrst.23) this requires The case R that Λ = 4πG(ρ + 3p) = 4πG(1 + γ)ρ. COSMOLOGICAL CONSTANT
11
above.1
Einstein Static Universe
Although we have noted that the cosmological constant provides repulsion.32) 4πG(1 + γ) and because ρ is positive this requires a positive Λ. From equation (1. or p = − 1 ρ which requires negative pressure.1. 3+γ R0 R0 (1. i.33)
˙ Now imposing R = 0 and assuming a matter equation of state (γ = 0) k implies Λ = R2 . (1. Both of these 3 alternatives were unacceptable to Einstein and therefore he concluded that a cosmological constant was present. However the requirement that Λ be positive forces k = +1 0 giving 1 Λ = 2 = constant. but the cosmological constant acts just like the uniform matter distribution. R 3 3 3 3 (1.e.30)
(1. Finally authors often write the cosmological constant in terms of a vacuum energy density as Λ ≡ 8πGρvac so that the velocity and acceleration equations become ˙ R 8πG Λ k k 8πG H 2 ≡ ( )2 = ρ− 2 + = (ρ + ρvac ) − 2 R 3 R 3 3 R and ¨ R 4πG Λ 4πG 8πG =− (1 + γ)ρ + = − (1 + γ)ρ + ρvac . The ˙ ¨ Einstein static universe requires R = R0 = constant and thus R = R = 0. it is interesting to calculate its exact value for a static universe [14.4.28) it follows that Λ= ˙ 3(1 + γ) R 2 k [( ) + 2 ].31) If there is no cosmological constant (Λ = 0) then either ρ = 0 which is an empty universe.

. NEWTONIAN COSMOLOGY
Thus the cosmological constant is not any old value but rather simply the inverse of the scale factor squared. where the scale factor has a ﬁxed value in this static model.12
CHAPTER 1.

The above equation can also be written as d dR3 (ρR3 ) + p =0 dt dt and from equation (1. dt 13 (2.2)
.29). Thus the Einstein equations imply this thermodynamic relationship in the above equation. it follows that d (ρR3+γ ) = 0.5) illustrating the intersting connection betweeen thermodynamics and General Relativity that has been discussed recently [16]. the velocity and acceleration equations can be obtained directly from the Einstein ﬁeld equations. 3(ρ + p) = (3 + γ)ρ.1)
This is identical to equation (1. The result is ρ + 3(ρ + p) ˙ ˙ R = 0.3) (2. The same result is achieved by taking the time derivative of equation (1. so too do our velocity and acceleration equations imply conservation of energy. The point is that we used thermodynamics to derive our velocity and acceleration equations and it is no surprise that the thermodynamic formula drops out again at the end. R (2. However.14).Chapter 2
APPLICATIONS
2. The energy-momentum conservation equation is derived by setting the covariant derivative of the energy momentum tensor equal to zero.1 Conservation laws
Just as the Maxwell equations imply the conservation of charge.

2
Age of the Universe
Recent measurements made with the Hubble space telescope [17] have determined that the age of the universe is younger than globular clusters. Substituting equation (2.8) yields
8πG 2 2 2 2 ˙ R2 = (ρR2 − ρ0 R0 + ρvac R2 − ρ0vac R0 ) − H0 R0 .28) as 8πG ˙ R2 = (ρ + ρvac )R2 − k 3 the present day value of k is k=
˙
(2.5) R
where primes denote derivatives with respect to R. 1 ρ + 3(ρ + p) = 0 (2.9)
R with H 2 ≡ ( R )2 . Alternatively d (2. Writing equation (1.8)
8πG 2 2 2 (ρ0 + ρ0vac )R0 − H0 R0 3
(2.1).6) (ρR3 ) + 3pR2 = 0 dR so that 1 R3+γ d (ρR3+γ ) = 0 dR (2.10)
. Present day values of quantities have been denoted with a subscript 0. APPLICATIONS
ρ=
c R3+γ
(2. i.7)
which is consistent with equation (2.4)
2. R4 Later we shall use these equations in a diﬀerent form as follows. x ≡ dx/dR. 3
(2. We illustrate this as follows.4)
1 where c is a constant.9) into equation (2.14 Integrating this we obtain
CHAPTER 2.e. A possible resolution to this paradox involves the cosmological constant [18]. This shows that the density falls as R3 for matter and 1 for radiation as expected. From equation (2.

INFLATION Integrating gives the expansion age
15
R0
.3
Inﬂation
In this section only a ﬂat k = 0 universe will be discussed. Inﬂation [19] occurs when the vacuum energy contribution dominates the ordinary density and curvature terms in equation (1. Results for an open or closed universe can easily be obtained and are discussed in the references [13]. Currently the universe is in a matter dominated phase whereby the dominant contribution to the energy density is due to matter.29). corresponding to an accelerating expanding universe called an inﬂationary universe. whereas for a vacuum dominated universe the right hand side will be a constant.2. Setting k = 0. 2 2 2 2 − ρ0 R0 + ρvac R2 − ρ0vac R0 ) − H0 R0 (2. Our aim here is simply to show that the inclusion of a cosmological constant gives an age which is larger than if no constant were present. The solution to the Friedmann equation for a radiation dominated universe will thus be 1 2 R ∝ t 2 . For a matter (γ = 0) or radiation (γ = 1) dominated universe the right hand side 1 will be of the form R3+γ (ignoring vacuum energy).29) depending on what is dominating the universe. The acceleration is positive.3. while for the matter dominated case it will be R ∝ t 3 . results in R ∝ exp(t). corresponding to a decelerating expanding universe. One can see that these results give negative acceleration.
0 0 8πG 2 3 (ρR
T0 =
dR = ˙ R
R0
dR
2. However the early universe was radiation dominated and the very early universe was vacuum dominated. there will only be one term on the right hand side of equation (1.
. Assuming these are negligible and substituting Λ = constant.11) 2 For the cosmological constant ρvac = ρ0vac and because R2 < R0 then a non zero cosmological constant will give an age larger than would have been obtained were it not present.

The ¯ Schr¨dinger equation is obtained by writing the Hamiltonian as an operator o ˆ H acting on a wave function Ψ as in ˆ ˆ HΨ = EΨ (2.12) where T = 1 mx2 is the kinetic energy. ˙ ˙ (2. which we derive herein for completeness.) The Hamiltonian H is deﬁned as H(P. Quantization is achieved by the operator re∂ ∂ ˆ ˆ placements P → P = −i ∂x and E → E = i ∂t where we are leaving oﬀ factors of h and we are considering the 1-dimensional equation only.13)
∂L . (−
. V is the potential energy. the Hamiltonian becomes 2
P2 +V =E (2. APPLICATIONS
2. The o Lagrangian L for a single particle moving in a potential V is L=T −V (2.4
2.14) (2.16
CHAPTER 2.17) 2m where E is the total energy.18) H=T +V = and making the above operator replacements to obtain 1 ∂2 ∂ + V )Ψ = i Ψ (2. (2.15) ∂x ˙ (Note P is the momentum but p is the pressure. The action ˙ 2 is S = Ldt and varying the action according to δS = 0 results in the Euler-Lagrange equation (equation of motion) d ∂L ∂L ( )− =0 dt ∂ x ˙ ∂x or just ∂L ˙ P = ∂x where P ≡ (2.4.19) 2 2m ∂x ∂t which is the usual form of the 1-dimensional Schr¨dinger equation written o in conﬁguration space.16) For many situations of physical interest.1
Quantum Cosmology
Derivation of the Schr¨dinger equation o
The Wheeler-DeWitt equation will be derived in analogy with the 1 dimensional Schr¨dinger equation. x). such as a single particle moving in a harmonic oscillator potential V = 1 kx2 . x) ≡ P x − L(x.

Making the replacement P → −i ∂R and imposing HΨ = 0 results in the Wheeler-DeWitt equation in the minisuperspace approximation for arbitrary k and with matter or radiation (ρ term) included gives
{−
∂2 9π 2 8πG + [(kR2 − (ρ + ρvac )R4 ]}Ψ = 0. R) = −κR3 [( )2 + 2 − (ρ + ρvac )] ≡ 0 R R 3
(2. QUANTUM COSMOLOGY
17
2.29) is recovered.21)
∂L ˙ Substituting L and P into the Euler-Lagrange equation. The Hamiltonian H ≡ P R − L is
˙ R k 8πG ˙ H(R. In terms of the conjugate momentum H(P. open and ﬂat and non-empty universes. (Note the calculation of ∂R is simpliﬁed by using the conservation equation (2. of course is also equal to zero. 2 2 ∂R 4G 3
(2. The Lagrangian is ˙ R k 8πG L = −κR3 [( )2 − 2 + (ρ + ρvac )] R R 3 with κ ≡
3π 4G . equa∂L tion (1.24)
.2. It is important to consider the possible presence of matter and radiation as they might otherwise change the conclusions.23)
∂ which. ˙ ∂R (2. 21.
(2.2
Wheeler-DeWitt equation
The discussion of the Wheeler-DeWitt equation in the minisuperspace approximation [20. Thus presented below is a derivation of the Wheeler-DeWitt equation in the minisuperspace approximation which also includes matter and radiation and arbitrary values of k.4. R) = −κR3 [ P2 k 8πG + 2− (ρ + ρvac )] = 0 2 R4 4κ R 3 (2. namely ρ + ρvac = ˙ −(3 + γ)ρ/R).22)
˙ which has been written in terms of R to show explicitly that the Hamiltonian is identically zero and is not equal to the total energy as before.5) with equation (1.20)
The momentum conjugate to R is P ≡ ∂L ˙ = −κ2RR. P − ∂R = 0. Herein we consider closed. (Compare equation (1.29)).14). 22] is usually restricted to closed (k = +1) and empty (ρ = 0) universes.4. Atkatz [11] presented a very nice discussion for closed and empty universes. 11.

The shapes in Figure 1 will be identical except that the whole graph will be shifted downwards by a constant with the inclusion of radiation. (For matter (γ = 0 and c = 0) a term growing like R will be included in the potential which will only be important for very small R and so the conclusions again will not be changed.18
CHAPTER 2.
2.25)
This just looks like the zero energy Schr¨dinger equation [21] with a potential o given by Λ 8πG 1−γ (2. open and ﬂat universes. 0. It can be seen that only the closed universe case provides a potential barrier through which tunneling can occur.5
Summary
. −1 respectively corresponding to closed [21].26) V (R) = kR2 − R4 − cR . If radiation (γ = 1 and c = 0) is included then only a negative constant will be added to the potential (because the term R1−γ will be constant for γ = 1) and these conclusions about tunneling will not change.) To summarize. only closed universes can arise from quantum tunneling even if matter or radiation are present.4) the Wheeler-DeWitt equation becomes {− ∂2 9π 2 Λ 8πG 1−γ + [kR2 − R4 − cR ]}Ψ = 0. APPLICATIONS
Using equation (2. This provides a clear illustration of the idea that only closed universes can arise through quantum tunneling [22]. 2 2 ∂R 4G 3 3 (2. 3 3 For the empty Universe case of no matter or radiation (c = 0) the potential V (R) is plotted in Figure 1 for the cases k = +1.

6. PROBLEMS
19
2.6
2.1
Problems
.2.

20
CHAPTER 2.1
Answers
. APPLICATIONS
2.7
2.

8
2.1
Solutions
2. SOLUTIONS
21
2.2.8.2
.

22
CHAPTER 2. APPLICATIONS
.

x2 ) (read as x superscript 2 not x squared) in a certain coordinate system and components (x. y) or (x1 . Expanding the matrix equation we have
x = x cos θ + y sin θ y = −x sin θ + y cos θ from which it follows that ∂x = cos θ ∂x 23 ∂x = sin θ ∂y
(3.Chapter 3
TENSORS
3.1 Contravariant and Covariant Vectors
Let us imagine that an ’ordinary’ 2-dimensional vector has components (x. which henceforth we are going to name contravariant vectors.2)
(3.3)
.1)
Notice that we are using superscipts (xi ) for the components of our ordinary vectors (instead of the usual subscripts used in freshman physics). We emphasize that these are just the ordinary vectors one comes across in freshman physics. x2 ) when that coordinate system is rotated by angle θ (but with the vector remaining ﬁxed). y) or (x1 . Then the components are related by [1] x y = cos θ sin θ sin θ cos θ x y (3.

xi yi ≡ i xi yi ) Instead of deﬁning an ordinary (contravariant) vector as a little arrow pointing in some direction.7)
.8)
(3. From calculus if f = f (x.4)
which can be written compactly as xi = ∂xi j x ∂xj (3. which is another way to deﬁne a vector as what happens to the components upon rotation (instead of the deﬁnition of a vector as a little arrow). ∂xi ∂xj ∂xi (3.1).5).3) is saying) then ∂f ∂f ∂x ∂f ∂y = + ∂x ∂x ∂x ∂y ∂x ∂f ∂f ∂x ∂f ∂y = + ∂y ∂x ∂y ∂y ∂y or simply ∂f ∂xj ∂f = . This is just a fancy version of equation(3. While we are on the subject of calculus and inﬁnitessimals let’s think ∂ about ∂xi which is kind of like the ’inverse’ of dxi .5) just from what we know about calculus. Using the inﬁnitessimal dxi (instead of xi ) it follows immediately that dxi = ∂xi j dx ∂xj (3.6)
which is identical to (3.5) and therefore we must say that dxi forms an ordinary or contravariant vector (or an inﬁnitessimally tiny arrow). we shall instead deﬁne it as an object whose components transform according to equation(3.5)
where we will always be using the Einstein summation convention for doubly repeated indices. (i. y) and y = y(x. TENSORS ∂y = cos θ ∂y
∂x ∂x x+ y ∂x ∂y ∂y ∂y x+ y y= ∂x ∂y
(3.e. Notice that we could have written down a diferential version of (3. y) and x = x(x. y) (which is what (3.24 ∂y = − sin θ ∂x so that x=
CHAPTER 3.

3. and so we might expect that ∂/∂xi are the ’components’ of a ’non-ordinary’ vector. ∂xi ∂xi ∂xj
25
(3.11) ∂xi So now let’s deﬁne a contravariant vector Aµ as anything whose components transform as (compare (3.13) and (3. dxi
.1. Or equivalently deﬁne xi ≡ Thus ∂ ∂xi (3. Contravariant and covariant vectors are the dual of each other. ∂/∂xi forms the corresponding ’dual’ space [2].13) In calculus we have two fundamental objects and the dual vector ∂/∂xi .5).10)
∂xj xj . so instead of writing ∂/∂xi let’s just always write it as xi for shorthand. Before proceeding let’s emphasize again that our deﬁnitions of contravariant and covariant vectors in (3. CONTRAVARIANT AND COVARIANT VECTORS Let’s ’remove’ f and just write ∂ ∂xj ∂ = .12) and a covariant vector Aµ (often also called a one-form.9)
which we see is similar to (3.1). The dual of the dual space is just the original space dxi . or dual vector or covector)
∂xν ∂xµ Aν
Aµ =
(3. If we try to form the dual dual vector ∂/∂(∂/∂xi ) we get back dxi [2]. Other examples of dual spaces are row and column x matrices (x y) and and the kets < a| and bras |a > used in quantum y mechanics [3]. Notice that the index is in the denominator. A set of points in a smooth space is called a manifold and where dxi forms a space. (3.5)) xi =
µ ∂xµ ν ∂xν A
A ≡
(3.13) are nothing more than fancy versions of (3.

Higher rank tensors can be constructed from lower rank ones by forming what is called the outer product or tensor product [14] as follows. TENSORS
3. γ
(3.18)
The tensor product is often written simply as T =A⊗B (3. t = time or T = temperature) has zero indices. ∂xα ∂xν β
(3.g. rank three Aµνγ etc. ∂xα ∂xβ T
(3. We are familiar with matrices which have two indices Aij . However all second rank tensors can be written as matrices. (3.just introduce antisymmetry). which sets a covariant and contravariant index equal.16) ∂xα ∂xβ ∂xρ γ Thus even though a matrix has two indices Aij .13) and (3.easy .13) and
T
µν
=
∂xµ ∂xν αβ . examples of which are (3. whereas a scalar (e. For instance
α Tβ ≡ Aα Bβ
(3.15)
∂xµ ∂xν ∂xγ αβ T .g. In order for an object to be called a tensor it must satisfy the tensor transformation rules. A contravariant tensor of rank two is of the form Aµν . Aµ .1) (NNN Next time discuss wedge product . it may not necessarily be a second rank tensor unless it satisﬁes the above tensor tranformation rules as well. A mixed tensor.14) Tν = Tρ =
µν µ
∂xµ ∂xβ α T . and because of the Einstein summation convention equal or repeated
.19)
(do Problem 3.17)
or
αβ β Tγδ ≡ Aα Bδ .26
CHAPTER 3. e. ν is partly covariant and partly contravariant.2
Higher Rank Tensors
Notice that our vector components Aµ have one index. Thus scalars are called tensors of rank zero and vectors are called tensors of rank one. We can also construct lower rank tensors from higher rank ones by a process called contraction.

3.20)
Thus contraction over a pair of indices reduces the rank of a tensor by two [14]. A scalar is a tensor of rank zero with the very special transformation law of invariance c = c.21)
Clearly the inner product of two vectors (rank one tensors) produces a scalar (rank zero tensor) as Aµ Bµ = constant ≡ A. Thus contraction represents setting two indices equal and summing.3
Review of Cartesian Tensors
Let us review the scalar product that we used in freshman physics. For example
αβ α Tγβ ≡ Tγ
(3. REVIEW OF CARTESIAN TENSORS
27
indices are summed over.24)
where A and B are the magnitudes of the vectors A and B and θ is the angle between them.22)
and it can be shown that A.3. (3.2). Thus
A.B ≡ AB cos θ (3.23) It is easily shown. We wrote vectors as A = Ai ei and deﬁned the scalar product as ˆ A.ˆj )Ai Bj e e ≡ gij Ai Bj (3.B as deﬁned here is a scalar (do Problem 3.3. γ
(3.Bj ej ˆ ˆ = (ˆi . The inner product [14] of two tensors is deﬁned by forming the outer product and then contracting over a pair of indices as
γ α Tβ ≡ Aα Bβ .25)
.B (3. for example.B = Ai ei . that Aµ B µ is no good as a deﬁnition of inner product for vectors because it is not invariant under transformations and therefore is not a scalar.

A ≡ A2 = Ai Ai = A2 + A2 + . (The choice Aµ B µ won’t do because it is not a scalar). The most natural way to do this is A. That is.. contravariant and covariant vectors).ˆj = |ˆi ||ˆj | cos θ = cos θ = δij ).. x y (3.B can also be written in terms of a metric tensor.. x y Notice that the usual results we learned about in freshman physics..B = Ai Bi = Ax Bx + Ay By + . we would expect that A.B ≡ Aµ B µ = gµν Aν B µ assuming gµν is a tensor. result entirely from requiring gij = δij = 0 1 in matrix notation..A ≡ A2 = Ai Ai (3. A Cartesian basis is deﬁned as one in which gij ≡ δij (obtained from ei .. equa1 0 tions (3.27).26)
where the sum (+.27) which is just Pythagoras’ theorem. (3.B ≡ Aµ B µ transforms as a scalar.
3.A ≡ A2 = Ai Ai = A2 + A2 + Ax Ay .) extends to however many dimensions are being considered and A.. TENSORS
where the metric tensor gij is deﬁned as the dot product of the basis vectors. gij = 1 1 in which case Pythagoras’ theorem would change to 0 1 A. However based on the previous section.26) and (3.. Now let’s return to vectors and one-forms (i...29)
.28
CHAPTER 3.2) that the inner product deﬁned by A.28)
ˆ e Thus it is the metric tensor gij ≡ ei ..ˆj given by the scalar product of the unit vectors which (almost) completely deﬁnes the vector space that we are considering. (3. A.4
Metric Tensor
We have already seen (in Problem 3. for example. We could easily have deﬁned a non-Cartesian space.e. the basis vectors are of unit ˆ e e e length and perpendicular to each other in which case A.

63-64).32)
which imples that vectors can be written in terms of components and basis vectors as A = Aµ eµ = Aµ eµ (3. Two more points to note are the symmetry gµν = gνµ (3.3. Because of this it also follows that A = Aµ eµ and A.B ≡ Aµ B µ = gµν Aν B µ ≡ (eµ . METRIC TENSOR
29
In fact deﬁning A.34)
. (do Problem 3.33)
Thus the basis vectors of a covariant vector (one-form) transform as contravariant vectors.e. Similarly it can raise indices Aµ = g µν Aν (3.4. is invariant).B ≡ Aµ B µ ≡ gµν Aν B µ makes perfect sense because it also transforms as a scalar (i.eν )Aν B µ (3. Contravariant components have basis vectors that transform as one-froms [5] (pg. so that (3.B = Aµ Bµ which were our ﬂat ˆ space results.31) How is vector A written in terms of basis vectors ? Based on our experience with Cartesian vectors let’s deﬁne our basis vectors such that A.30) so that the metric tensor has the eﬀect of lowering indices.31) becomes Aµ = Aµ showing that in ﬂat space there is no distinction between covariant and contravariant vectors. and thus we deduce that Aµ = gµν Aν (3.29) will do equally well as the deﬁnition of the scalar product.3) Thus either of the two right hand sides of (3. The above results illuminate our ﬂat (Cartesian) space results where gµν ≡ δµν .

This follows by getting back what we start α with as in Aµ = gµν Aν = gµν g να Aα ≡ δµ Aα .4.38)
Note that A0 = A0 .38) we must have ∂µ ≡ implying that ∂µ ≡ ∂ ∂ ∂ =( . the energy momentum four vector pµ = (E.38) and (3.− ) ∂xµ ∂x0 ∂t (3.36) 0 0 1
ν Obviously for unit matrices there is no distinction between δµ and δµν .39) (3.41) ∂ ∂ = ( 0. −A) (3. TENSORS
ν ν gµα g αν = δµ = gµ
(3.30 and the inverse deﬁned by
CHAPTER 3. Ai ) = (A0 . Exercise: Prove equation (3.37)
If a contravariant vector is speciﬁed by Aµ = (A0 . Ai ) = (A0 .40)
.35)
ν so that gµ is the Kronecker delta.1
Special Relativity
Whereas the 3-dimensional Cartesian space is completely characterized by gµν = δµν or 1 0 0 gµν = 0 1 0 (3. Thus. for example.39) using (3.
3.− ) = ( . p) gives 2 = E 2 − p2 . ∂t ) (3. Because of equation (3. Of course p2 is the invariant we identify as m2 so that p E 2 = p2 + m2 .37). ∂xµ ∂x )=( ∂ . The 4-dimensional spacetime of special relativity is speciﬁed by
ηµν =
1 0 0 0 0 −1 0 0 0 0 −1 0 0 0 0 −1
(3. A) it follows that the covariant vector is Aµ = ηµν Aν or Aµ = (A0 .

5. In electrodynamics in ﬂat spacetime we encounter E=− φ (3.
3. (This must be true so that E 2 = (pc)2 + (mc2 )2 ). The proper time τ is deﬁned via ds ≡ cdτ = cdt γ
(3. 8].42)
∂ (Note that some authors [30] instead deﬁne P2 ≡ 2 − ∂t2 ).46) dxµ ≡ (γc. The fourvelocity is deﬁned as uµ ≡ such that the fourmomentum is pµ ≡ muµ = ( E . We deﬁne
P2 ≡ ∂µ ∂ µ = ∂0 ∂ 0 + ∂i ∂ i = ∂0 ∂0 − ∂i ∂i ∂2 ∂2 = − 2= 2− 2 ∂t ∂x0 2
2
(3.44)
which is consistent with the time dilation eﬀect as the proper time is the time measured in an observer’s rest frame.45)
where m is the rest mass. 14.47) and B= ×A (3.48)
.5
Christoﬀel Symbols
Some good references for this section are [7. γv) dτ (3. We shall Let us now brieﬂy discuss the fourvelocity u write out c explicitly here.3. µ and proper time. p) c (3. Using dxµ ≡ (cdt.43)
(3. dx) the invariant interval is
ds2 ≡ dxµ dxµ = c2 dt2 − dx2 . CHRISTOFFEL SYMBOLS Note that ∂ 0 = ∂0 =
∂ ∂t
31
(with c ≡ 1). Exercise: Check that (muµ )2 = m2 c2 .

7. z) and A = A(x.50)
We want to know whether the derivative Aµ.ν ≡ ∂Aµ ∂xα ∂xγ ∂Aα ∂ 2 xα ν = ∂xµ ∂xν ∂xγ + ∂xν ∂xµ Aα ∂x ∂xα ∂xγ ∂ 2 xα = µ ∂xν Aα. y.ν is spoiled by the second term.
. Thus Aµ. E and B are also vector ﬁelds because their values are a function of position also.ν α β transform according to Aµ. let’s evaluate the ∂x ∂x derivative explicitly Aµ.e. is the gradient operator deﬁned (in 3 dimensions) as ˆ ≡ˆ i∂/∂x + ˆ j∂/∂y + k∂/∂z = e1 ∂/∂x1 + e2 ∂/∂x2 + e3 ∂/∂x3 .ν = ∂xµ ∂xν Aα.e.48) imply that the derivatives φ and × A also transform as vectors. i. ? To ﬁnd out.β . z. (The electric ﬁeld of a point charge gets smaller when you move away. φ = φ(x.52)
We see therefore that the tensor transformation law for Aµ. i. y. Let us introduce a shorthand for the derivative as Aµ.ν ≡ ∂Aµ ∂ ∂xα = ( Aα ) ∂xν ∂xν ∂xµ ∂ 2 xα ∂xα ∂Aα + ν µ Aα = ∂xµ ∂xν ∂x ∂x
(3.47) and (3.32
CHAPTER 3.) Because the left hand sides are vectors.51)
but Aα is a function of xν not xν . Aα = Aα (xν ) = Aα (xν ) .ν ≡ ∂Aµ ∂xν (3. TENSORS
where E and B are the electric and magnetic ﬁelds and φ and A are the scalar and vectors potentials. ˆ ˆ ˆ (3.49)
Clearly then φ and A are functions of x. Therefore φ is called a scalar ﬁeld and A is called a vector ﬁeld.ν is not a tensor [8. Therefore γ α α we must insert ∂Aν = ∂Aγ ∂xν so that ∂x ∂x ∂x Aµ. y. 14]. (3. That is does Aµ. What about the derivative of tensors in our general curved spacetime ? Do they also transform as tensors ? Consider a vector ﬁeld Aµ (xν ) as a function of contravariant coordinates.γ + ∂xν ∂xµ Aα ∂x
(3.ν is a tensor. z).

the astronaut communicates with you by radio and views your arrow through her spacecraft window. 14] as limdxγ →0 µ . Imagine standing on the curved surface of the Earth. If you walk from Paris to Moscow and keep the arrow pointed upward at all times (in other words transport the vector parallel to itself).53) is that the numerator is not a vector because Aµ (x + dx) and Aµ (x) are located at diﬀerent points. Thus to compare two vectors (i.53)
µ or more properly [7. then an astronaut viewing the arrow from a stationary position in space will notice that the arrow points in diﬀerent directions in Moscow compared to Paris. dxν The problem however with (3. The diﬀerence betweeen two vectors located at separate points is not a vector because the transformations laws (3. which is easy to do in ﬂat space. To ﬁx this situation. In freshman physics when we represent two vectors A and B as little arrows. Aµ. compute A − B) we must ﬁrst put them at the same spacetime point. the diﬀerence A − B is not even deﬁned (i. She then draws a whole series of parallel lines on her window and as you walk from Paris to Moscow she keeps instructing you to keep your arrow parallel to the lines on her
.e. Vector A has changed into a diﬀerent vector C.) Thus parallel transport produces a diﬀerent vector.12) and (3. it’s orientation has changed.3.ν ≡ Aµ (x + dx) − Aµ (x) ∂Aµ = lim ν dx→0 ∂x dxν
A (xγ +dxγ )−A (xγ )
(3. even though according to you. The diﬀernce between two vectors is only a vector if they are located at the same point.5. but they do change in curved space. you have parallel transported the vector and it still points vertically upward from the Earth. Thus in order to calculate Aµ (x + dx) − Aµ (x) we must ﬁrst deﬁne what is meant by parallel transport in a general curved space. is not a vector) if A and B are at diﬀerent points. She makes a little mark on her window to line up with your arrow in Paris. Thus the astronaut sees the arrow pointing in a diﬀerent direction and concludes that it is not the same vector. When we parallel transport a vector in ﬂat space its components don’t change when we move it around. CHRISTOFFEL SYMBOLS
33
To see why this problem occurs we should look at the deﬁnition of the derivative [8]. We ﬁrst instruct the freshman student to slide one of the vectors to the other one and only then we can visualize the diﬀerence between them. holding a giant arrow (let’s call this vector A) vertically upward.13) depend on position. This sliding is achieved by moving one of the vectors parallel to itself (called parallel transport). say in Paris.e. (It can’t be because it points diﬀerently.

δAµ ∝ dxα (3. then it must still be the same vector A.34
CHAPTER 3. Γα = Γα .57) from which it follows that (do Problem 3. Thus δAµ ∝ Aν dxα (3.4) δAµ ≡ −Γµ Aν dxα . To obtain parallel transport for a contravariant vector B µ note that a scalar deﬁned as Aµ B µ cannot change under parallel transport. the more noticeable its change will be. whereas the metric tensor tells us how to deﬁne distance betweeen neighboring points. να We shall also assume [8] symmetry under exchange of lower indices. When you get to Moscow. δAµ is the change produced in vector Aµ by an inﬁnitessimal transport by a distance dxα to produce a new vector Cµ ≡ Aµ + δAµ . except now moved to a diﬀerent point (Moscow). she is satisﬁed that you haven’t rotated your arrow compared to the markings on her window. Think about it !)
. We expect δAµ to be directly proportional to dxα . TENSORS
window.59) (3. Equation (3. µν νµ (3. Let’s denote δAµ as the change produced in vector Aµ (xα ) located at xα by an inﬁnitessimal parallel transport by a distance dxα . Thus [8] δ(Aµ B µ ) = 0 (3.56)
where Γν are called Christoﬀel symbols or coeﬃcients of aﬃne connection µα or simply connection coeﬃcients . If a vector is parallel transported from an ’absolute’ point of view (the astronaut’s window). the bigger our arrow.56) deﬁnes parallel transport.58)
(We would have a truly crazy space if this wasn’t true [8]. the connection coeﬃcients tell us how to deﬁne parallelism betweeen neighboring points.54)
We also expect δAµ to be directly proportional to Aµ .55)
The only sensible constant of proportionality will have to have covariant µ and α indices and a contravariant ν index as δAµ ≡ Γν Aν dxα µα (3. As Narlikar [7] points out.

62) becomes Aµ.ν ≡ ∂xν . [14.ν is a second rank tensor (which will be done later in Problem 3.ν is often called the covariant derivative (with the word covariant not meaning the same as before) and one can easily verify that Aµ.64) For tensors of higher rank the results are. The derivative Aµ.3. From (3.ν ≡ Aµ (x + dx) − [Aµ (x) + δAµ ] dAµ = lim ν dx→0 dx dxν
∂A ∂A
(3. (3.53) in (3.ν ≡ dxν = ∂xν − Γµν A (because dx ν = δν ) which dx we shall henceforth write as
Aµ.62)
µ µ Using (3. The diﬀerence betweeen them is dAµ = Aµ (xα + dxα ) − [Aµ (xα ) + δAµ ] (3.61)
which by construction is a vector.5). the new vector Cµ will be Cµ = Aµ + δAµ .61) we have dAµ = ∂xν dxν − δAµ = ∂xν dxν − Γµα A dxα α dAµ ∂Aµ α and (3. for example.λ λ
ν
+ Γν Aµ λ
.ν − Γµν A (3.58)
∂A
Aµ ≡ Aµ + Γµ A .λ . Thus we are led to a new deﬁnition of derivative (which is a tensor [8]) Aµ.ν ν (3.63)
µ where Aµ.5.60)
whereas the old vector Aµ (xα ) at the new position xα + dxα will be Aµ (xα + dxα ) .ν ≡ Aµ. CHRISTOFFEL SYMBOLS
35
Continuing with our consideration of Aµ (xα ) parallel transported an inﬁnitessimal distance dxα .ν . 8]
Aµν ≡ Aµν + Γµ A .

Let us now prove these results.65)
and Aµν.6
Christoﬀel Symbols and Metric Tensor
We shall now derive an important formula which gives the Christoﬀel symbol in terms of the metric tensor and its derivatives [8. Applying (3.µ = 1 [ln(−g)]. ).69).67)
(3. (3. λ α αβ.66) gµν.λ λ and
ν Aµν ≡ Aµν + Γµ Aαβ + Γν Aµ − Γαλ Aµν − Γβλ Aµν .λ ≡ gµν. gµν.72)
− Γνλ gµ ≡ 0
(3.β
− gβγ.λ λ αβ β ν
− Γνλ Aµ
(3.36
CHAPTER 3.γ
+g
γ.70) where g ≡ determinant|gµν |. To ensure this means that the covariant derivative of the metric tensor should always be identically zero.69)
Another result we wish to prove is that √ Γµ = (ln −g).73)
.µ 2 (3.68)
3.λ − Γµλ A and Aµ ≡ Aµ + Γµ Aν − Γνλ Aµ ν. The process of covariant diﬀerentiation should never change the length of a vector.λ ν. 14. (3.λ ≡ 0.λ − Γµλ g
ν
(3. 7].71) Note that g = |g µν |. TENSORS (3.λ ≡ Aµν.λ αβ.66) (3. Proof of Equation (3. The formula is
Γα = 1 g α (g βγ 2
β.

34).78) µν 2 which gives (3.35) it follows that ∂g = gg µν ∂gµν
(3.74)
and permuting the µνλ indices cyclically gives
µ
(3. (3.34) and (3.µ − gµν.76) and subtract (3.77)
because of the symmetries of (3.ν + gνλ.β
− gβα.λ = Γµλ g gλµ.59) and (3. CHRISTOFFEL SYMBOLS AND METRIC TENSOR Thus gµν.6.35) (to give gλ g λα = g λ g λα = δ α ) yields 1 Γα = g λα (gλµ.74) gives [8] gλµ.µ = Γνµ g
λ ν
37
+ Γνλ gµ + Γµν gλ + Γλµ gν
(3.69) becomes (ﬁrst and last terms cancel) 1 Γα = g α (g β. Multiplying (3.70) [14] (Appendix II) Using g α g β.λ ).7) Thus (3.ν = Γλν g and gνλ.
.79) becomes 1 ∂g ∂gλα Γα = β.6).α = g α gβα.5 and 3.α
+g
α.75) (3.80)
a result which can be easily checked.69).81) 2 ∂xβ which is (3.3.76)
Now add (3. ). Proof of equation (3.77) by g λα and using (3. (3.λ = 2Γµν gλ (3.α 2
β. (do Problem 3. (obtained using the symmetry of the metric tensor and swapping the names of indices) and contracting over αν.75) and (3.79)
1 = g α g α.α 2g ∂gλα ∂xβ 1 ∂g = 2g ∂xβ 1 ∂ ln g = (3.70) we write ln(−g) instead of ln g because g is always negative.β 2 Deﬁning g as the determinant |gµν | and using (3. where in (3. (do Problems 3.µ − gµν. equation (3.ν + gνλ.70).

αβ ∂ 2 Aµ ∂xα ∂xβ (3.µν = φ. Firstly .αβ .µ = φ. λ λ ∂x µ µ µ = A.λ .85)
+ Γµ Γθ − Γµ Γθ ) λθ ν νθ λ
µ ≡ A Rλ ν
(3. 2.λ.ν.ν λ λ Subtracting we have Aµ − Aµ = A (Γµ . From (3.ν.64).α. 8].λ .38
CHAPTER 3.ν = φ.ν . however in general . A scalar does not change under parallel transport therefore φ.ν
(3.λ A + Γµ A.λ + Γν .ν − Γµν φ.86)
with the famous Riemann curvature tensor deﬁned as
.β .µ is a tensor.µ.αβ .ν ν λ
.ν − Γνλ Aµ . We shall use either notation.λ − Γµ .µ.7
Riemann Curvature Tensor
The Riemann curvature tensor is one of the most important tensors in general relativity. . not a scalar) φ.αβ .63) we have (φ.83) but because Γµν = Γνµ it follows that φ.α. Firstly we write in general Aµ ≡ . This tensor is most easily derived by considering the order of double diﬀerentiation on tensors [28.ν use (3.ν A + Γµ A.βα consider the second derivative of a scalar φ.ν − Γνλ Aµ . If it is non-zero then we have a curved space.λ λ ∂ (Aµ ) + Γµ A.ν = φ.µ. =
(3. Let us examine this in more detail.αβ .λ.βα it is true that Aµ = Aµ . . .84)
Now interchange the order of diﬀerentiation (just swap the ν and λ indices) Aµ = Aµ + Γµ . (3. 7.β In general it turns out that even though Aµ = Aµ .λ − Γλν Aµ ν .82)
and also when we write Aµ we again mean second derivative.ν.ν . 9. so we .λ.νµ meaning that the order of diﬀerentiation does not matter for a scalar.ν − Γνλ Aµ ν .λ + Γλ A. Many authors .ν.ν + Γµ A. Note that Aµ is a second rank tensor. Let’s diﬀerentiate equation (3. Consider now a vector. TENSORS
3. If it is zero then it means that the space is ﬂat.αβ instead write Aµ ≡ Aµ or Aµ ≡ Aµ .67) as follows Aµ = Aµ + Γµ A.µ .

92)
(3.93) (It will turn out later that Rαβ = 0 for empty space [9] ).8. For example Rαβγδ = Rγδαβ (3.87) Exercise: Check that equations (3.86) and (3.95) Gµν ≡ Rµν − Rgµν 2 After discussing the stress-energy tensor in the next chapter. However all authors agree on the deﬁnition of the Riemann scalar (obα tained by contracting Rβ )
α R ≡ Rα ≡ g αβ Rαβ β
(3. Thus diﬀerent books may have this sign diﬀerence.89) (3. SUMMARY
α Rβγδ ≡ −Γα + Γα + Γαγ Γβδ − Γαδ Γβγ βγ.88) =0 (3. For a Cartesian spcae the Riemann tensor is zero. The Riemann tensor tells us everything essential about the curvature of a space. we could have deﬁned R αβ or Rα β or Rαβ as the Ricci tensor and we would have the same result except that maybe a sign diﬀernce would appear. we shall put all of this tensor machinery to use in our discussion of general relativity following.δ βδ.3.90)
+
α Rγδβ
+
α Rδβγ
and Rαβγδ = −Rβαγδ All other symmetry properties of the Riemann tensor may be obtained from these. The Riemann tensor has the following useful symmetry properties [9]
α α Rβγδ = −Rβδγ α Rβγδ
(3.
3. i. Note that the contraction of the Riemann tensor is unique up to a sign.γ
39
(3.91) Finally we introduce the Ricci tensor [9] by contracting on a pair of indices Rαβ ≡ Rα which has the property Rαβ = Rβα (3.94)
Finally the Einstein tensor is deﬁned as 1 (3.87) are consistent.e.8
Summary
.

6 Show that Aµ.ν is a second rank tensor. 3.9
Problems
3. (Equation (3. 3. 3.5 Derive the transformation rule for Γα .80)). Is Γα a tensor ? βγ βγ 3.e.7 Check that
∂g ∂gµν
= gg µν .3 Show that the inner product deﬁned by A. where gµν is assumed to be a tensor.
. i.B ≡ Aµ Bµ is invariant under transformations.4 Prove equation (3.2 Show that the inner product A. TENSORS
3. 3. show that the tensor product (outer product) µ deﬁned by Tν ≡ Aµ Bν is also a tensor.1 If Aµ and Bν are tensors.58). show that it satisﬁes the tensor transformation law of a scalar (thus it is often called the scalar product).40
CHAPTER 3. 3.B ≡ gµν Aµ B ν is also a scalar (invariant under transformations).

ANSWERS
41
3. only solutions
.10.3.10
Answers
no answers.

42
CHAPTER 3. y) then ∂f = ∂f ∂x + ∂f ∂y = ∂xi ∂x .1
Solutions
µ To prove that Tν is a tensor we must show that it satisﬁes the β µ ∂xµ α tensor transformation law T ν = ∂xα ∂xν Tβ . ∂θ ∂x ∂θ ∂y ∂θ ∂θ Now A. TENSORS
3.2 First let’s recall that if f = f (θ. ∂x
Proof T ν = A B ν = = =
∂xµ ∂xβ α ∂xα ∂xν A Bβ ∂xµ ∂xβ α ∂xα ∂xν Tβ
µ
µ
∂xµ α ∂xβ ∂xα A ∂xν Bβ
QED.11
3.B = A B µ = = = =
∂xµ ∂xβ α ∂xα ∂xµ A Bβ ∂xβ α ∂xα A Bβ β δα Aα Bβ µ ∂xµ α ∂xβ ∂xα A ∂xµ Bβ
by the chain rule
= Aα Bα = A.
3.B
. y) and α = i ∂f α(x. α) and θ = θ(x.

3.3 A. SOLUTIONS 3.7
.B ≡ g µν A B = = = = =
µ ν ∂xµ ∂xν γ δ ∂xα ∂xβ ∂xµ ∂xν gαβ ∂xγ ∂xδ A B ∂xα ∂xβ ∂xµ ∂xν γ δ ∂xµ ∂xν ∂xγ ∂xδ gαβ A B ∂xα ∂xβ γ δ ∂xγ ∂xδ gαβ A B α β δγ δδ gαβ Aγ B δ gαβ Aα B β
43
= Aα Bα = A.5
3.6
3.11.B
3.4
3.

TENSORS
.44
CHAPTER 3.

For conservative forces ˙ ˙ q ˙ q dt F = − V where V is the scalar potential. That is in freshman physics T is a function only 2 of velocity qi and V is a function only of position qi . qi ) ≡ T − V where T is the kinetic energy.
4.Chapter 4
ENERGY-MOMENTUM TENSOR
It is important to emphasize that our discussion in this chapter is based entirely on Special Relativity.2)
˙ where pi = mqi (with qi being the generalized position coordinate) so that dpi = mqi + m¨i . Rewriting Newton’s law we have − dV d = (mqi ) ˙ dqi dt (4.1
Euler-Lagrange and Hamilton’s Equations
dp dt
Newton’s second law of motion is F= (4. Thus L(qi .3)
˙ Let us deﬁne the Lagrangian L(qi . qi ) = ˙ ˙ 45
. If m = 0 then Fi = m¨i = mai . In freshman physics T = T (qi ) = 1 mqi and V = V (qi ) such as the harmonic ˙ ˙2 2 2 oscillator V (qi ) = 1 kqi .1)
or in component form (for each component Fi ) Fi = dpi dt (4.

A more rigorous derivation is based on the calculus of variations [1] as discussed in Section 7. ˙ ∂qi
(4. qi ) ≡ pi qi − L(qi .8)
∂L For the simple case T = 1 mqi and V = V (qi ) we have pi ∂ qi = mqi so that ˙2 ˙ ˙ 2 ˙
i i i ˙ T = 2m and pi qi = m so that H(pi .9) ∂pi
p2
p2
p2
because L = L(pi ) and
∂H ∂qi
∂L = − ∂qi so that from (4. It follows that ˙ Newton’s law is
=
dT dqi ˙
= mqi = pi .5)
The second equation of (4.7)
We have obtained the Euler-Lagrange equations using simple arguments.4) is known as the Euler-Lagrange equations of motion and serves as an alternative formulation of mechanics [1]. Thus ˙
dpi dt ∂L d ∂L = ( ) ∂qi dt ∂ qi ˙ Fi = with the canonical momentum [1] deﬁned as pi ≡ ∂L ∂ qi ˙
(4. qi ) ˙ ˙ (4.4)
(4. Hamilton’s equations of motion immediately follow from (4.46
CHAPTER 4.10)
. It is usually written d ∂L ∂L )− =0 (4. ENERGY-MOMENTUM TENSOR
∂L ∂qi dV = − dqi and ∂L ∂ qi ˙
T (qi ) − V (qi ).6) ( dt ∂ qi ˙ ∂qi or just pi = ˙ ∂L ∂qi (4. We now introduce the Hamiltonian H deﬁned as a function of p and q as H(pi . qi ) = 2m + V (qi ) = T + V which is the total energy.4)
−
∂H = pi .8) as ∂H = qi ˙ (4.3.

This will be derived rigorously in the next section. an example of which may be the Higgs boson.2
Classical Field Theory
Scalar ﬁelds are important in cosmology as they are thought to drive inﬂation. Such a ﬁeld is called an inﬂaton. The Euler-Lagrange equations (4. Any time derivative dt should be replaced with ∂ ∂µ ≡ ∂xµ which contains space as well as time derivatives. ˙ ˙ Let us replace the qi by a ﬁeld φ ≡ φ(x) where x ≡ (t. a Higgs boson or any other scalar boson. In the next section we shall show how to derive the Euler-Lagrange equations from the action deﬁned as S ≡ Ldt (4. The generalized coordiante q has been replaced by the ﬁeld variable φ and the discrete index i has been replaced by a continuously varying index x.2. Thus the ﬁeld φ considered below can be thoguht of as an inﬂaton.6) are clearly not covariant because special emphasis is placed on d ∂L time via the qi and dt ( ∂ qi ) terms. In both special and general relativity we always seek covariant equations in which space and time are given equal status.4.12)
∂L with L ≡ Ld3 x. A covariant form of the action would involve a Lagrangian density L via S≡ Ld4 x = Ld3 xdt (4. In analogy with the canonical momentum in equation (4.15)
. x). Thus one can guess that the covariant generalization of the point particle Euler-Lagrange equations (4. The term − ∂qi in equation (4. CLASSICAL FIELD THEORY
47
4.13) ∂(∂µ φ) ∂φ
which is the covariant Euler-Lagrange equation for scalar ﬁelds.11) which again is clearly not covariant.6) is ∂L ∂L ∂µ − =0 (4.14)
so that the Euler-Lagrange equations become ∂µ Πµ = (4.6) gets replaced by the ∂L d covariant term − ∂φ(x) .5) we deﬁne the covariant momentum density Πµ ≡ ∂L ∂(∂µ φ) ∂L ∂φ (4.

8)) Tµν ≡ Πµ ∂ν φ − gµν L with the Hamiltonian density H≡ Hd3 x (4.1
Classical Klein-Gordon Field
In order to illustrate the foregoing theory we shall use the example of the classical. the Euler-Lagrange equations give the ﬁeld equation ∂φ as ∂µ ∂ µ φ + m2 φ or (P2 + m2 )φ = 0 ¨ φ−
2
φ + m2 φ = 0
(4.16)
The energy momentum tensor is (analagous to (4.20)
˙ giving the canonical momentum Π = Π0 = ∂ 0 φ = ∂0 φ = φ. massive Klein-Gordon ﬁeld deﬁned with the Lagrangian density (HL units ??) 1 LKG = (∂µ φ∂ µ φ − m2 φ2 ) 2 1 ˙ = [φ2 − ( φ)2 − m2 φ2 ] 2
(4.48
CHAPTER 4.22)
.17)
˙ H ≡ T00 = Πφ − L
4. ˙ Π=φ (4. ENERGY-MOMENTUM TENSOR
The canonical momentum is deﬁned as Π ≡ Π0 = ∂L ˙ ∂φ (4.18) (4.2.19)
The covariant momentum density is more easily evaluated by re-writing ∂L α α LKG = 1 g µν (∂µ φ∂ν φ − m2 φ2 ). Thus Πµ = ∂(∂µ φ) = 1 g µν (δµ ∂ν φ + ∂µ φδν ) 2 2
α α = 1 (δµ ∂ µ φ + ∂ ν φδν ) = 1 (∂ α φ + ∂ α φ) = ∂ α φ. Thus for the Klein-Gordon 2 2 ﬁeld we have Πα = ∂ α φ (4.21)
Evaluating ∂L = −m2 φ.

3. Other references are [8](Pg. In momentum space p2 = −P2 . [7]. massive scalar ﬁeld. [15](Pg. These references show that the energy-momentum tensor for a perfect ﬂuid is
.4
Energy-Momentum Tensor for Perfect Fluid
The best references for this section are [9](Pg.25)
4. [33](Pg.1.23)
∂ (Note that some authors [30] deﬁne P2 ≡ 2 − ∂t2 diﬀerent from (3. so that they write the Klein-Gordon equation as (P2 −m2 )φ = 0 or (p2 +m2 )φ = 0.4.24)
˙ Therefore the Hamiltonian density is H ≡ T00 = φ2 − 1 (∂α φ∂ α φ − m2 φ2 ) 2 which becomes [31] 1˙ 1 1 H = φ2 + ( φ)2 + m2 φ2 2 2 2 1 2 2 = [Π + ( φ) + m2 φ2 ] 2 where we have relied upon the results of Section 3. 330). 38).
(4.
4.3
Principle of Least Action
derive EL eqns properly for q and φ (do later). Leave out for now. [34](Pg. and [2]. 83).4. and [32](Pg. The book by D’Inverno [32] also has a nice discussion of the NavierStokes equation and its relation to the material of this section.) The energy momentum tensor is
Tµν ≡ Πµ ∂ν φ − gµν L = ∂µ φ∂ν φ − gµν L 1 = ∂µ φ∂ν φ − gµν (∂α φ∂ α φ − m2 φ2 ). 155).42). 2 (4. thus (p2 − m2 )φ = 0
2
(4. PRINCIPLE OF LEAST ACTION
49
which is the Klein-Gordon equation for a free. 124-125). 259).

Fig.
. so that T 00 = ρ. The special cases of motionless dust or motionless radiation are obtained with the respective substitutions of p = 0 or p = 1 ρ in 3 equation (4. 0).26)
where ρ is the energy density and p is the pressure. 3 Again the radiation is conﬁned to a container not in motion so that uµ = (γc.50
CHAPTER 4. We shall now work this out for several speciﬁc cases [9]. Thus uµ = (c. In this case uµ = (c. but conﬁned in a motionless container. Thus again T 00 = ρ but now T ii = p and T ij = 0 so that
T µν =
ρ 0 0 0
0 p 0 0
0 0 p 0
0 0 0 p
(4.) Thus 4 1 T µν = ρuµ uν − ρη µν 3 3 ρ 0 0 0 0 1ρ 0 0 3 = 0 0 1ρ 0 3 0 0 0 1ρ 3
(4.28). 2.28)
Motionless radiation is characterized by the equation of state p = 1 ρ. The equation of state for dust is p = 0 so that T ii = 0 = T 0i = T ij . (The 1 just comes from randomizing the pressure in 3 dimensions 3 [7]. Thus
T µν =
ρ 0 0 0
0 0 0 0
0 0 0 0
0 0 0 0
(4.5 of Narlikar’s book [7] is particularly helpful.29)
Thus the general case is the motionless ﬂuid energy-momentum tensor in equation (4. but now p = 0. 0).27)
Motionless ﬂuid representes a collection of particles all moving randomly (such that they exert a pressure) but the whole collection is at rest. ENERGY-MOMENTUM TENSOR T µν = (ρ + p)uµ uν − pη µν (4.28). Motionless dust represents a collection of particles at rest. 0) again. such as a gas of particles at non-zero temperature.

Thus for an interacting scalar ﬁeld we simply tack on −V (φ) to the free Klein-Gordon Lagrangian of equation (4. Thus V (qi ) naturally gets replaced with V (φ) where φ ≡ φ(x). In analogy with electrodynamics the conservation law for the energymomentum tensor is µν T.5. 276).31)
where LO ≡ LKG and LI ≡ −V (φ).6
Interacting Scalar Field
We represent the interaction of a scalar ﬁeld with a scalar potential V (φ). These discrete coordinates qi have now been replaced by continuous ˙ ˙ ﬁeld variables φ(x) where φ has replaced the generalized coordinate q and the discrete index i has been replaced by a continuous index x. Actually the Lagrangian of (4.2) Thus the four Maxwell equations are entirely equivalent to only three Maxwell equations plus the equation of continuity.19) to give 1 L = (∂µ φ∂ µ φ − m2 φ2 ) − V (φ) 2 ≡ LO + LI
(4.j = 0. j) and the covariant conservation law is ∂µ j µ = 0 which results in the equation of continuity ∂ρ + . We had a similar situation in Chapters 1 and 2 where we found that the velocity and acceleration equations imply the conservation equation.4.1 e and 4.20) and (4. (do Problems 4.1) can be derived from this. Thus the covariant momentum density and canonical momentum remain the same as equations (4. This can also be obtained from the Maxwell ∂t equations by taking the divergence of Amp`re’s law.21) for the
. CONTINUITY EQUATION
51
4.31) refers to a minimally coupled scalar ﬁeld as opposed to conformally coupled [21] (Pg. It is important to emphasize that V (φ) does not contain derivative terms such as ∂µ φ. Recall our elementary results for L = T − V = 1 mqi − V (qi ) for the coordi˙2 2 nates qi .30) In the next chapter we shall show how equation (2.5
Continuity Equation
In classical electrodynamics the fourcurrent density is j µ ≡ (cρ.
4. Thus the two velocity and acceleration equations are entirely equivalent to only the velocity equation plus the conservation law.ν = 0 (4.

36)
Note that even though Tii has repeated indices let us not assume i is implied in this case. equation (4.e. 34] E ≡ ρ ≡< T00 > and p ≡< Tii > (4. Some authors (e.34)
yielding the Hamiltonian density the same as for the free particle case. equation (4. except for the addition of V (φ) as in 1˙ 1 1 H ≡ T00 = φ2 + ( φ)2 + m2 φ2 + V (φ) 2 2 2 1 2 = [Π + ( φ)2 + m2 φ2 ] + V (φ). assume Special Relativity NNN) we obtain 1˙ 1 1 Tii = φ2 + ( φ)2 − m2 φ2 − V (φ) 2 2 2 (4. we make the identiﬁcation [13.25).35)
The purely spatial components are Tii = ∂i φ∂i φ−gii [ 1 (∂α φ∂ α φ−m2 φ2 )− 2 V (φ)] and with gii = −1 (i.37)
. except for the addition of gµν V (φ) as in 1 Tµν = ∂µ φ∂ν φ − gµν [ (∂α φ∂ α φ − m2 φ2 ) − V (φ)] 2 (4.33)
The energy-momentum tensor is the same as for the free particle case. That is Tii refers to Tii = T11 = T22 = T33 and not Tii = T11 + T22 + T33 . comparing equation (4.52
CHAPTER 4. 3 Let us assume that the eﬀects of the scalar ﬁeld are averaged so as to behave like a perfect (motionless) ﬂuid.24).38) (4. In that case.32)
(4.g Serot and Walecka [34]) do assume the latter convention and therefore will disagree with our results by 1 . ENERGY-MOMENTUM TENSOR
˙ free particle case namely Πα = ∂ α φ and π = φ.28). Solving the Euler-Lagrange equations now gives (P2 + m2 )φ + V = 0 ¨ φ− with V ≡
2
φ + m2 φ + V = 0 dV dφ
(4. 2
(4.

φ = φ(x). 138 of Islam [13]).
˙ ρ = 1 φ2 + V (φ) 2 (4. (See Pg. COSMOLOGY WITH THE SCALAR FIELD
53
where E ≡ ρ is the energy density and p is the pressure.e.4.39)
(4.7.7
Cosmology with the Scalar Field
We have ﬁnished with our discussion of the energy-momentum tensor and therefore we should now move onto the next chapter. Therefore we ﬁnally obtain [13. i. with the tools at hand (energy-momentum tensor and Friedmann equations) we can discuss the relevance of the scalar ﬁeld to cosmology without needing the formalism of General Relativity.) Making these identiﬁcations we have 3 1˙ 1 1 ρ = φ2 + ( φ)2 + m2 φ2 + V (φ) 2 2 2 and 1˙ 1 1 p = φ2 + ( φ)2 − m2 φ2 − V (φ) 2 2 2 (4. they actually write p ≡ 1 < Tii >. However.41) and
˙ p = 1 φ2 − V (φ) 2 (4. Therefore before proceeding to the next chapter we shall make a brief digression and discuss the evolution of the scalar ﬁeld. 21]. 276-277 of Kolb and Turner [21] and Pg.42)
4.40)
Let us also assume that the scalar ﬁeld is massless and that φ = φ(t) only. (Note that because Serot and Walecka do assume the Einstein summation convention for Tii . one can simply substitute the above expressions for ρ and p into the Friedmann
. If one is considering cosmological evolution driven by a scalar ﬁeld. so that spatial derivatives disappear.

43) or more simply by substituting the expression for ρ and p in equations (4.1) to give
2 ¨ ˙ ( φ) ] + m2 φ + V = 0. ENERGY-MOMENTUM TENSOR
and acceleration equations (1. The diﬀerence occurs because we have now incorporated gravity via the Friedmann and conservation equation.44) (4. Again assuming the ﬁeld is massless and ignoring spatial derivatives we have
¨ ˙ φ + 3H φ + V = 0 (4.29) and (1.46) Notice that this is the equation for a damped harmonic oscillator (V = 1 kx2 2 and dV ≡ V = kx with F = −V ) as dx m¨ + dx + kx = 0 x ˙ (4. We shall derive this equation again in Chapter 7. which will cause reheating. and instead write ¨ ˙ ˙ φ + 3H φ + Γφ + V = 0 (4.54
CHAPTER 4.47)
Kolb and Turner [21] actually also include a particle creation term due to the decay of the scalar ﬁeld.48)
.43)
The equation for the time evolution of the scalar ﬁeld is obtained either by taking the time derivative of equation (4.30) to obtain the time evolution of the scale factor as in ˙ R 8πG 1 ˙ 2 1 1 k Λ H 2 ≡ ( )2 = [ φ + ( φ)2 + m2 φ2 + V (φ)] − 2 + R 3 2 2 2 R 3 and ¨ R 1 Λ 4πG 1 ˙ 2 1 =− [ φ + ( φ)2 − m2 φ2 − V (φ)] + R 3 2 2 2 3 (4. φ + 3H[φ + ˙ φ
(4.45)
Note that this is a new Klein-Gordon equation quite diﬀerent to equation (4.42) into the conservation equation (2.41) and (4.32).

(NNNN why ???) However if m = 0 and it is the same as (4.51)
which results from setting m = 0 and φ = 0 in (4.) Consider a Lagrangian for φ which already has the scale factor built into it as 1 L = R3 [ (∂µ φ∂ µ φ − m2 φ2 ) − V (φ)] (4. Identifying the Lagrangian as [29] L = R3 (T − V ) we immediately write ˙ down the total energy density ρ = T + V = 1 φ2 + V (φ).46) for the scalar ﬁeld in a quicker manner [29] (Pg.20).1). This will be discussed in Chapter 7. 73).40). We didn’t even use the energy-momentum tensor.46) and substituting into the ˙ ˙ ˙¨ derivative ρ = φφ + V φ = −3H φ ˙ conservation equation (2. and so we only have i = 1.1
Alternative derivation
We can derive the equation of motion (4. COSMOLOGY WITH THE SCALAR FIELD
55
4. [29].49) 2 √ The R3 factor comes from −g = R3 for a Robertson-Walker metric. but this derivation only seems to work if we set m = 0 and φ = 0 at the beginning.3) ¨ φ−
2
˙ φ + 3H φ + m2 φ + V = 0
(4. qi ≡ φ. ρ = −3H(ρ + p) we obtain the pressure as ˙ ˙ p = 1 φ2 − V (φ). Thus our energy density and pressure derived here agree 2 with our results above (4. The equation of motion is (do Problem 4.50) φ=0
which is diﬀerent to (4.39) and (4. The equation of motion is ¨ ˙ φ + 3H φ + V = 0 (4. Taking the time 2 2 from (4. i.7.
.46). (Exercise: ﬁnd out what goes wrong if m = 0 and φ = 0. Notice that it is the same factor which sits outside the Friedmann Lagrangian in equation (2.e. Here we have only φ = φ(t) (not φ = φ(x)).49).45).4. Let’s only consider 1˙ L = R3 [ φ2 − V (φ)] 2
(4.7.52) Notice how quickly we obtained this result rather than the long procedure to get (4. Also realize that because φ = 0 the above Lagrangian formalism is really no diﬀerent to our old fashioned formalism where we had qi (t).46). Notice that the pressure is nothing L more than p = R3 .

56

CHAPTER 4. ENERGY-MOMENTUM TENSOR

4.7.2

Limiting solutions

Assuming that k = Λ = 0 the Friedmann equation becomes ˙ R 8πG 1 ˙ H 2 ≡ ( )2 = ( φ+V) R 3 2 (4.53)

This equation together with equation (4.46) form a set of coupled equations where solutions give φ(t) and R(t). We solve the coupled equations in the standard way by ﬁrst eliminating one variable, then solving one equation, then substituting the solution back into the other equation to solve for the other variable. Let’s write equation (4.46) purely in terms of φ by eliminating ˙ R R which appears in the form H = R . We eliminate R by substituting H from (4.53) into (4.46) to give

¨ φ+

˙ ˙ 12πG(φ2 + 2V )φ+V’=0

2

¨ ¨ ˙ ˙ φ2 + 2φV − 12πG(φ2 + 2V )φ2 + V

=0

(4.54)

Notice that this is a non-linear diﬀerential equation for φ, which is diﬃcult to solve in general. In this section we shall study the solutions for certain limiting cases. Once φ(t) is obtained from (4.54) it is put back into (4.53) to get R(t). Potential Energy=0 ˙ Setting V = 0 we then have ρ = 1 φ2 = p. Thus our equation of state is 2 p=ρ or γ = 3. With V = V = 0 we have ¨ φ2 + √ ˙ 12πGφ2 = 0 (4.56) (4.55)

which has the solution (do problem 4.4) φ(t) = φo + √ √ 1 ˙ ln[1 + 12πGφ(t − to )] 12πG (4.57)

4.7. COSMOLOGY WITH THE SCALAR FIELD

57

(Note that the solution is equation (9.18) of [29] is wrong.) Upon substituting this solution back into the Friedmann equation (4.53) and solving the diﬀerential equation we obtain (do problem 4.5) √ ˙ R(t) = Ro [1 + 12πGφo (t − to )]1/3 . (4.58) This result may be understood from another point of view. Writing the Friedmann equations as ˙ R 8πG H 2 ≡ ( )2 = ρ R 3 and ρ= then the solution is always R∝t2/m which always gives ρ∝ If ρ = constant (corresponding to m = 0) then the solution is R∝et (4.64) (4.63) 1 . t2 (4.62) (4.61) α Rm (4.59)

(4.60)

(do problem 4.6). Note that for m < 2, one obtains power law inﬂation. For ordinary matter (m = 3), or radiation (m = 4) we have R ∝ t2/3 and R ∝ t1/2 respectively. Returning to the scalar ﬁeld solution (4.57) the density ˙ is ρ = 1 φ2 for V = 0. Thus 2 ˙ φ(t) =

R combined with ( Ro )3 = 1 +

1+

√

˙ φo ˙ 12πGφo (t − to )

(4.65)

√

˙ 12πGφo (t − to ) from (4.58) yields ˙ 3 φo Ro ˙ φ(t) = R3 (4.66)

to give the density

58

**CHAPTER 4. ENERGY-MOMENTUM TENSOR ρ=
**

˙ 6 1 φ2 Ro o 2 R6

(4.67) corresponding to m = 6 and thus R ∝ t1/3 in agreement with (4.58). Note 1 also that this density ρ ∝ R6 also gives ρ ∝ t1 . 2 1 Thus for a scalar ﬁeld with V = 0, we have p = ρ (γ = 3) and ρ ∝ R6 . 1 Contrast this with matter for which p = 0(γ = 0) and ρ ∝ R3 or radiation 1 for which p = 1 ρ (γ = 1/3) and ρ ∝ R4 . 3 However equation (4.67) may not be interpreted as a decaying Cosmological Constant because p = ρ (see later). Kinetic Energy=0 ˙ ¨ Here we take φ = φ = 0, so that ρ = V and p = −V giving p = −ρ (4.68)

or γ = −3 which is a negative pressure equation of state. Our equation of motion for the scalar ﬁeld (4.54) becomes V =0 meaning that V = Vo (4.70) which is constant. Substituting the solution into the Friedmann equation (4.53) gives ˙ R 8πG H 2 = ( )2 = (4.71) Vo R 3 which acts as a Cosmological Constant and which has the solution (do problem 4.7) R(t) = Ro e

8πG Vo (t−to ) 3

(4.69)

(4.72)

which is an inﬂationary solution, valid for any V . Warning We have found that if k = Λ = 0 and if ρ ∝ R1 then R∝t2 for any m value of m. All of this is correct. To check this we might substitute into the Friedmann equation as ˙ R 1 H 2 = ( )2 ∝ 2 (4.73) R t

54) is a diﬃcult non-linear equation. rather then solving the diﬀerential equation) √ φ(t) = 2Alnt (4. The result R ∝t is wrong because we have left out an important constant.4.75) into (4. Substituting into the Friedmann t2 ˙ (2/m) R with the above constant C R = t
equation gives
4.54) is (which was presumably guessed at. R ∝t The lesson is be careful of constants when doing back-of-the-envelope calculations. ˙ R Actually if R = c then ln R = c ln t = ln tc giving R ∝ tc instead of R∝t. We check this claim by substituting (4.54).74) where β and λ are constants to be determined. We shall examine the model of Barrow [35] which can be solved exactly and leads to power law inﬂation. COSMOLOGY WITH THE SCALAR FIELD
˙
59
1 and say R ∝ 1 giving R dR dt ∝ dt which yields ln R ∝ lnt and thus R t dt t R ∝ t2/m . Barrow [35] claims that a particular solution to (4. Barrow’s model [35] is brieﬂy introduced by Islam [13].77) 2 A (4. t d2 Let’s keep our constants then. From this we ﬁnd (do problem 4. Any scalar ﬁeld model is speciﬁed by writing down the potential V (φ).76)
. The advantage of an exactly solvable model is that one can develop ones physical intuition better.75) √ where 2A is just some constant.7. Barrow’s potential is V (φ) ≡ βe−λφ (4. Write ρ = Rm then R = ( md )2/m t2/m 2
and ρ =
˙
d2 ( md )2 t2 2 (2/m)2
=
2 R or = m yielding ( R )2 = t2 2/m in agreement with the correct result above.78) (4.3
Exactly Solvable Model of Inﬂation
Because (4.9) that λ= and β = −A or β = A(24πGA − 1) (4.74) and (4.7.
(2/m)2 . exactly solvable models are very rare.

79)
(4.84) (4.81) t2 Solving the Friedmann equation (4.79) gives R∝t8πGA where D is some constant. (4. Setting 8πG ≡ 1 we have R∝tA in agreement with Barrow’s solution.78). Also Barrow doesn’t use the ﬁrst solution (4.77) with β = −A.60
CHAPTER 4.82)
Inverting the solution (4. so that the second solution (4.81) we have ρ∝ 1 R2/A (4. (Recall φ(t) and R(t) are the solutions we seek to our coupled equations √ Λ ˙ (4.80)
Clearly we see why we reject the ﬁrst solution (4. Using the second solution (4. (See sections 2 4.46) and (4.9) we have ˙ R 8πG 1 2A β 8πG 1 H 2 ≡ ( )2 = ( 2 + 2) = (A + β) 2 R 3 2 t t 3 t giving an equivalent density ρ= A+β t2 (4.85) can also be obtained via ρ = 1 φ2 +V.83) (4. Substituting into (4.4 and 4. Also he uses units with 8πG = 1.77) for reasons we shall see shortly.) Substituting V = t2 and φ = 2A (see solution to t problem 4. ENERGY-MOMENTUM TENSOR
√ Note that Barrow is wrong when he writes λA = 2. Having solved for φ(t) we now substitute into (4. (4.85)
which corresponds to a Weak decaying Cosmological Constant. 2
t R t R
. Power law inﬂation results for A > 1.83) we have t2 = C R2/A where C is some constant. It would give zero density.78) with β = A(24πGA − 1) yields 24πGA2 ρ= .53).7.53) to solve for R(t).5) For the inﬂationary result A > 1 we have A ≡ m < 2 which corresponds to the quantum tunneling solution!! ˙ Note of course that (4.7. We have 3 √ 1 1 ˙ ˙ V= β ∝ 2/A and φ = 2A giving φ2 ∝ 2/A . he writes correctly as β = A(3A − 1).

) Let us deﬁne a Strong Cosmological Constant as one in which the velocity and acceleration equations both have the same right hand side. Firstly the velocity and acceleration equations both have the same right hand side.88)
(4. ρ = −3H(ρ+p). ˙ this can only happen for ρ=constant.87)
for p = γ ρ. Recall the Friedmann equations ˙ R 8πG Λ k H 2 ≡ ( )2 = ρ− 2 + R 3 R 3 and ¨ R R = − 4πG Λ (ρ + 3p) + 3 3 4πG Λ = − ρ(1 + γ) + 3 3 (4.91)
only for γ = −3 or p = −ρ. Thus constant density with equation of state p = −ρ acts identically to a Cosmological Constant.4.89)
where two things have happened.4
Variable Cosmological Constant
In this section we address the question as to when the density can be interpreted as a Cosmological Constant.7. then we have 3 ˙ R Λ H 2 = ( )2 = R 3 and ¨ R Λ = R 3 (4. In addition the solution is automatically are of exponential inﬂation. Suppose ρ = k = 0. Secondly the acceleration is positive. COSMOLOGY WITH THE SCALAR FIELD
61
4. What sort of density would give the same result. Again for k = 0 ˙ R 8πG H 2 = ( )2 = ρ R 3 and ¨ R 8πG = ρ R 3 (4.
.90)
(4. R ∝ eHt . From our conservation equation. (Exercise:verify this.7. (which automatically implies that the acceleration is positive). Such a Strong Cosmological Constant must be a true constant.86)
(4.

62

CHAPTER 4. ENERGY-MOMENTUM TENSOR

On the other hand we can imagine densities that still give a positive acceleration (i.e. inﬂation) but do not normally give the velocity and acceleration with the same right hand side. Examining (4.54) indicates that the acceleration is guarenteed to be positive if γ < −1 giving p < − 1 ρ. (Recall 3 that the exponential inﬂation above required γ = −3, which is consistent with the inequality.) Thus negative pressure gives inﬂation. (Although not all negative pressure gives inﬂation, e.g. p = − 1 ρ.) The inﬂation due to 4 γ < −1 will not be exponential inﬂation, but something weaker like perhaps power law inﬂation. Let us deﬁne a Weak Cosmological Constant as one which arises from negative pressure (actually p < − 1 ρ) to give a positive 3 acceleration (inﬂation) only. The velocity and acceleration equations need not have the same right hand side. Recall that ordinary matter and radiation, or any positive pressure equation of state, neccesarily leads to negative acceleration (with Λ = 0). Thus positive pressure leads to attractive gravity. However positive acceleration implies a repulsive gravity or antigravity. Thus negative pressure (actually p < − 1 ρ) leads to antigravity. This is why we wish to use the term weak Cos3 mological Constant (even though right hand sides are not the same) because it is consonant with antigravity. Let us summarize. We consider ρ alone with Λ ≡ 0. For ¨ R > 0 ⇒ γ < −1 and ρ =anything (e.g. ρ = ρ(R) or ρ =constant) ¨ R > 0 and ρ =constant ⇒ γ = −3 For γ < −1, ρ behaves as a weak Cosmological Constant and for γ = −3, ρ behaves as a Strong Cosmological Constant. Finally let us emphasize that it is perfectly legitimate to consider a Weak Cosmological Constant as a real Cosmological Constant. Einstein’s original motivation in introducing Λ was to obtain a static universe. Thus all he wanted was a antigravity term; i.e. all he wanted was a weak Cosmological Constant. It “accidentally” happened that the right hand sides turned out to be equal, giving a strong Λ. Interpreting ρ as a Cosmological Constant leads us to expect that a weak Cosmological Constant can vary. This follows from ρ=anything above. i.e. ρ=constant or ρ = ρ(R) giving a variable function. (But a strong Cosmological Constant cannot vary).

4.7. COSMOLOGY WITH THE SCALAR FIELD

63

4.7.5

Cosmological constant and Scalar Fields

Refer back to the density and pressure of the scalar ﬁeld in equations (4.41)) and (4.42). We had 1˙ ρ = φ2 + V (φ) 2 and 1˙ ρ = φ2 − V (φ). 2 For the case where P E = V = 0 we have p = ρ. The pressure is positive 1 and therefore equation (4.67), ρ ∝ R6 , cannot be interpreted as a (variable) Cosmological Constant. ˙ For the case where KE = 1 φ2 = 0 we have p = −ρ meaning that ρ can 2 be interpreted as a Strong Cosmological Constant. (We found V = 0 and thus ρ = V = constant = Vo and p = −Vo .) These results are true in general (assuming m = φ = 0) as we have not yet speciﬁed V (φ). Let us now consider the Barrow model V (φ) = βe−λφ . we found that 2 2 6 2 φ(t) = λ ln t and β = λ2 ( λ2 − 1) for 8πG ≡ 1. Introducing A ≡ λ we can √ 2 re-write as β = A(3A − 1) and φ(t) = 2A ln t and V (φ) = A(3A − 1)eφ A . Substituting we obtain 3A2 ρBarrow = 2 (4.92) t and 3A2 2A 2A ρBarrow = 2 ( − 1) = ( − 1)ρ (4.93) t 3 3 the general equation of state is p = γ ρ giving the Barrow equation of state 3 2 −3 (4.94) A in equation (4.84) we concluded that power law inﬂation results for A > 1. Substituting this into (4.94) implies γBarrow = γBarrow < −3 (4.95)

¨ which we expect because power law inﬂation implies R > 0. thus for A > 1 2 the Barrow pressure is negative with γ < −1 and thus ρBarrow = 3A cort2 responds to a Weak Cosmological Constant. Furthermore this Cosmological Constant is variable and decays with time. in equation (4.85) we wrote this 1 as ρBarrow ∝ R2/A .

64

CHAPTER 4. ENERGY-MOMENTUM TENSOR

4.7.6

Clariﬁcation

We wish to clarify the distinction between a Strong Cosmological Constant, a Weak Cosmological Constant and a Varying Cosmological Constant. A Strong Cosmological Constant occurs when the Friedmann equation are of the form ˙ R Λ H 2 ≡ ( )2 = (4.96) R 3 and ¨ R Λ −qH 2 ≡ = (4.97) R 3 and comparing with ˙ R 8πG ( )2 = ρ (4.98) R 3 ¨ R 8πG =− (ρ + 3p) (4.99) R 3 leads us to conclude that the equation of state for a Strong Cosmological Constant is p = −ρ, where we have identiﬁed Λ ≡ 8πGρv . The conservation equation ρ = −3H(ρ + p) ˙ (4.100) leads us to conclude that Λ ≡ 8πGρv =constant for a Strong Cosmological Constant. A Variable Cosmological Constant Λ(R), can also be considered but the equation cannot be like (4.96) and (4.97) with no ρ term. As we shall see below a Varying Cosmological Constant automatically involves matter creation (existence of a ρ term). Thus Λ(R) cannot be written down in the Friedmann equation without also writing ρ. Thus for a Varying Cosmological Constant (let k = 0) we have ˙ R 8πG Λ(R) H 2 ≡ ( )2 = ρ+ R 3 3 and ¨ R 4πG Λ =− (1 + γ)ρ + R 3 3 One can easily show that the conservation equation becomes −qH 2 ≡ ρ + ρv =-3H(ρ+p) =−H(3+γ)ρ ˙ ˙ (4.101)

(4.102)

Further Λ(R) ˙ can vary. ρv = constant) then matter or radiation is created or destroyed. Weak.7.7
Generic Inﬂation and Slow-Roll Approximation
We shall ﬁrst discuss features that are common to many models of inﬂation based on scalar ﬁelds [21]. For a Variable Cosmological Constant the same term Λ(R) appears in 3 both right hand sides of the Friedmann equation (4. Strong.101) and (4.
4. Other ways of writing this are (see equations 2. This is why we cannot write (4.e.105) (4.7 ).102). d(ρR3 ) dR3 ˙ (4.96) and (4.1-2.7.108)
The last equation show clearly that if ρv = 0 (i. COSMOLOGY WITH THE SCALAR FIELD
65 (4. A Weak Cosmological Constant can vary but the right hand sides of the Friedmann equations diﬀer. radiation or anything else.106) (4.102). These models typically have a large region of the potential where the potential is ﬂat (slow-roll region) and then a harmonic
.107) (4. each with both similar and dissimiliar properties. These are summarize in the Table.104) +p + R3 ρv = 0 dt dt d(ρR3+γ ) ˙ + R3+γ ρv = 0 dt 3 ρ + ρv = − (ρ + p) R d(ρR3 ) + 3pR2 + R3 ρv = 0 dR 1 R3+γ d(ρR3+γ ) = −ρv dR (4.97) if Λ is allowed to vary. Thus we see that we have 3 diﬀerent types of Cosmological Constant namely.4.103)
(do Problem 4.10). However we don’t know its equation of state. Also γ can be anything and thus ρv can lead to creation of matter. and Variable. Both a Strong and Weak Cosmological Constant have well deﬁned equation of state. both with (diﬀerent) negative pressure. We must include particle creation terms as in (??) and (4.

Eqn. for V = 1 kx2 or V = kx would represent a damped harmonic oscil2 lator. which indeed is the case in the rapid oscillation region. Such a potential is shown in Fig 4. The physical situation would represent a ball rolling down a hill with friction into a valley [21]. Let us ﬁrst analyze this situation from a simple physics point of view. giving V = −C.1. ENERGY-MOMENTUM TENSOR Weak Strong same Variable Same (but must include ρ)
Right Hand Sides of Velocity and Acceleration Eqns. State
Diﬀerent
p < −1ρ 3
p = −ρ
Unknown
Λ = 8πGρv
Variable
Constant
Variable
Table 4.112) (4. x=constant and x doesn’t change much so that x = 0. If Fig 4. In the slow roll regime friction dominates and the ball moves at terminal velocity [21].66
CHAPTER 4. This is ˙ ˙ ¨ the slow roll regime we have bx + V = 0 ˙ and for V = Vo − Cx we have bx≈C ˙ (4. Let’s return to the scalar ﬁeld equations which are (for Λ = k = 0) ˙ R 8πG 1 ˙ H 2 ≡ ( )2 = ( φ+V) R 3 2 (4.111) When the ball ﬁnally reaches the valley we are back to the full equation (4.109) which.1 were actually a plot of V (x) versus x the equation of motion would be m¨ + bx + V = 0 x ˙ (4.1: Properties of 3 types of Cosmological Constants region where the potential looks similar to an oscillatior (rapid oscillation region). damped oscillations. In the slow roll regime we have V ≈constant or V = Vo − Cx where C is small.109) where V ≡ kx so that the ball experiences rapid.110)
.

115) valid which we can solve directly for R(t) giving an exponential inﬂationary solution R(t)≈Ro et
8πG VO 3
(4. no matter what V (φ) is. the slow roll region is characterized by a small kinetic energy 1 ˙2 2 φ << V . COSMOLOGY WITH THE SCALAR FIELD and ¨ ˙ φ + 3H φ + V = 0
67
(4.54) ﬁrst for φ(t) and then substitute our answer into the Friedmann equation to obtain R(t). but here we have two equations (4. We have speciﬁed that we are in a slow roll regime thus making (4.8
Chaotic Inﬂation in Slow-Roll Approximation
The Chaotic inﬂation model is deﬁned by the potential 1 V (φ) = m2 φ2 2 (4.118)
which is exactly analogous to the harmonic oscillator potential V (x) = 1 kx2 . The slow roll equation can also be obtained directly from equation (4. 2
. Normally we would have the formula for V(φ) and solve (4. so that 8πG H2 ≈ Vo ≈ constant (4.7.114) and (4. To obtain φ(t) we must specify V (φ) even in the slow-roll regime.
4.115). the slow roll approximation (φ ≈constant.116) or (4.115) 3 For the rolling ball we only had one equation.113)
where the friction term 3H is due to the expansion of the universe [21].114) In addition.4.7.115) involved in the slow roll approximations.117)
Thus. Based ˙ on analogy with the rolling ball. However we have not yet speciﬁed V (φ).54) where we set ˙ ¨ φ2 << 2Vo and φ ≈ 0 to give ˙ 24πGVo φ + V = 0 (4. ¨ φ ≈ 0) is ˙ 3H φ + V ≈ 0 (4. the slow regime always give approximately exponential inﬂation.114) and (4.116)
which is consistent with combining (4.

which
1 V ˙ √ φ = −√ 24πG Vo or ˙ φ 1 m = −√ φ φo 12πG
(4.122)
is the slow roll region in agreement with equation (9.119)
1 where G ≡ M 2 is in agreement with equation (9. Having p speciﬁed V (φ) in (4.117) becomes o 2 R(t) = Ro et = Ro e
t
4π 3
√
Gmφ
4π m φ 3 Mp
(4. That is the amplitude is reduced by the e factor 1 We expect that the slow roll approximation will be valid for t < τ .122) came from solving (4.125)
.120)
(4.116).23) of Madsen [29]. or slowing rolling. For m << Mp or for short times we see that φ(t) will be approximately constant. but only ˙ ¨ that φ2 << 2V and φ ≈ 0. ENERGY-MOMENTUM TENSOR Using Vo = 1 m2 φ2 the slow roll solution (4.116). We see that φ(t) is a decaying exponential in time.68
CHAPTER 4. e Our solution of φ(t) in (4. Then instead of (4. Let’s not assume V = Vo =constant.123) mMp When t = τ we see that φ = φo 1 .121)
This gives φ(t) = φo e
−
φo m √ t 12πG
= φo e
−
φo
mMp √ t 12π
mMp = φo − √ t 12π
(4.124) which. Thus there will be a “half-life” or “lifetime” associated with slow roll which we deﬁne as √ φo 12π τ= (4. for chaotic inﬂation becomes √ ˙ 12πGmφφ + m2 φ = 0
(4.116) we have √ ˙ 24πGV φ + V = 0 (4.25) of Madsen [29]. We can also obtain φ(t) slightly diﬀerently.118) we can solve (4.

126)
which has the solution
(4. Thus the 3 requirement N ≥ 60 yields √ N ≥ 60 ⇒ φo ≥ 5Mp .130) 3 Mp mMp Mp
m where H = 4π Mp φo taken from (4. the number of e-folds is √ 4π m φo 12π φ2 N = Hτ = φo = 4π o2 (4.131)
. Ro (4. This means that φ ≈ 0 and ρ ≈ Vo which give exponential inﬂation.123).4. We can further investigate the validity of the slow roll approximation by evaluating the potential as a function of time and checking that it is constant for short times. Given R(t) = Ro eHt . the number of e-folds after time t is N = ln( R(t) ) = Ht. In order to solve the horizon. After one lifetime τ .127)
in agreement with (4. ﬂatness and monopole problems.122) for short time.278)).7. We get V = ≈ 1 2 2 1 2 2 − φmM3π t √p m φ = m φo e o 2 2 1 2 mMp m φo (φo − √ t) 2 3π
(4. For short times (or for m << Mp ) we have veriﬁed that φ and V are ˙ approximately constant. (4. so that p = −ρ). We do this by substituting our solution for φ(t) back into the potential. most models require a high degree of inﬂation typically amounting to about 60 e-folds [21]. (Also p ≈ −Vo .128)
Thus we see that for m << Mp or for short times the potential is indeed constant.26) of the book by Kolb and Turner [21] (pg.129)
(Actually a better formula is given in equation (8.119) and τ is from (4. COSMOLOGY WITH THE SCALAR FIELD or m ˙ φ = −√ 12πG φ(t) = φo − √ m t 12πG mMp = φo − √ t 12π
69
(4.

We solve (4. If we could see beyond the horizon we would see the curvature. And quantum tunnelling predicts that what we would see would be a universe of positive curvature. According to inﬂation the size of the universe is much larger then the distance to the horizon ( i. thus within our horizon the universe is ﬂat. but rather gives R2 ≈ 0 which is equivalent to k = 0. the reason many people believe the Earth is ﬂat is because we cannot see beyond the horizon. whereas Vo has remained constant. Quantum cosmology predict that a universe which arises via tunnelling must have k = +1.132) t= 4π mφo Ro Substituting into (4. ENERGY-MOMENTUM TENSOR
˙
R Notice how the ﬂatness problem is solved in inﬂation. Quantum tunnelling requires k = +1.
Cosmological Constant associated with Chaotic Inﬂation Let us now calculate the density as a function of R. The universe can have k = 0 or k = +1 or k = −1.135)
. Inﬂation does not give k = 0. No matter what the value of k. It simply predicts that R2 ≈ 0 at the end of inﬂation. Up to the horizon it looks ﬂat. Similarly for our universe. Thus the term R2 is k entirely negligible. it gets diluted by inﬂation and is equivalent to a universe with k = 0.122) we have
p Ro o φ(R) = φo ( ) 4πφ2 R M2
(4. o 2 R M2
(4.122) mMp − φ √12π ˙ φ(t) = − √ t e o 12π
mMp
(4. On Earth. During inﬂation (slow roll) Vo stays constant but 1 1 k by the end of inﬂation R = Ro e60 or R2 = R2 e−120 .133)
and with V (φ) = 1 m2 φ2 we have 2
p 1 Ro o V (R) = m2 φ2 ( ) 2πφ2 .70
CHAPTER 4. This is perfectly OK with inﬂation which simply dilutes the curvature.119) for t = t(R) as 3 Mp R ln( ). Inﬂation actually says nothing k about the value of k. The term R2 has
o
k dropped by e−120 . We have H 2 ≡ ( R )2 = 8πG k 8πG k 3 ρ − R2 ≈ 3 Vo − R2 . (4. as far as we can see) the universe looks k ﬂat becuase R2 is negligible.134)
Also from (4.e. If we could see beyond the horizon we would see the curvature. This is an improtant distinction.

144)
.136)
(4. yielding Mp φo > √ = 0. Rm 2πφ2 o
(4.142)
2
)(
(4.139)
(4. COSMOLOGY WITH THE SCALAR FIELD so that
71
mMp Ro 4πφ2 ˙ φ(R) = − √ ( ) o.
Deﬁning κ ≡ 8π p ( Ro )m .143)
2 )κ.138)
Recall previously that an inﬂationary solution requires m < 2.131)! ˙ What is the equation of state? using p = 1 φ2 − V we obtain 2
Mp 2 Mp Ro 1 o − φ2 )( ) 2πφ2 . Write this as Mp φo = √ with 4π Thus the density becomes ρ(R) = and p(R) =
m2 M 2 2 m2 Mp 1 ( + 8π 3 2 m2 Mp 1 ( − 8π 3 2
>1
(4. The requirement > 1 yields 1+3 1 p<− ρ 2
2 )κ
giving
(4.4. is the pressure negative? We ﬁnd that for inﬂation to occur Mp we need φo > √4π . ρ(R) = m2 ( o 2 12π R 2
2 Mp
(4. we write ρ = ( 1 + R 3 2 p = 1−3 2 ρ. p = m2 ( o 2 12π R 2
(4.140)
The question is.3 Mp 4π which is entirely consistent with (4.7.141)
)(
Ro m ) R Ro m ) R and p = ( 1 − 3
(4.137)
Thus ρ(R) ∝
2 Mp 1 where m ≡ . 12π R ˙ Finally evaluating ρ = 1 φ2 + V we have 2
Mp 2 Mp Ro 1 o + φ2 )( ) 2πφ2 .

410) and by Dolgov.148)
For the chaotic inﬂation model. 2 (4. upon assuming φ = 0. Martin and Squires [37] (Pg.146) this result is discussed by Linde [36] (Pg.145)
These results are in agreement with our previous constraints that in order ¨ to have positive R (inﬂation) we needed p < 1 ρ or γ < −1. Collins. Sazhin and Zeldovich [38]. gives δρ = ρ where we have used G ≡ this yields
1 2. 3π Mp (4. Mp
H 2π
(4.284). Kolb and Turner [21](Pg.7. V = 1 m2 φ2 2 δρ = ρ 4 m .72
CHAPTER 4.147)
2 GV 3π V
2
=
V 2 √ 3π Mp V
(4.50). ENERGY-MOMENTUM TENSOR
which means negative pressure! Writing p = γ ρ gives 3 3 γ=− . 3 Our chaotic inﬂation model is the slow roll approximation gives negative pressure (but not p = −ρ) and corresponds to a Weak Decaying Cosmological Constant!
4.149)
This is an intensely important formula often written as √ δρ m ≈m G = ρ Mp (4.17.9
Density Fluctuations
An important result that we shall use without proof is that ﬂuctuations of the scalar ﬁeld are given approximately by δφ ≈
H 2π
(4. dρ ˙ Using ρ = 1 φ2 +V(φ) we have dφ = V (φ) or 2 δρ = V δφ≈V ˙ which.150)
.

there are two things to keep in mind. However.
(4. COSMOLOGY WITH THE SCALAR FIELD
73
the density ﬂuctuations observed by COBE are δρ ≈ 10−5 yielding m = ρ 10−5 Mp . In that case the formula is written more usefully as [?.10
Equation of State for Variable Cosmological Constant
In this section we wish to demonstrate that Variable Cosmological Constant models have negative pressure [?. The above formula is not very useful for the Barrow model where a well deﬁned inﬂation means m ≡ minf lation is not present.158)
This looks like bad news. ?.151)
4.155)
(4.4.154)
(4.7. it looks like the pressure only get negative for m> 3 + γ. Firstly. dR (4. ?. ?] ρinf lation δρ = ρ ρp (4.157)
The pressure is p= γ A m γ + ρv 3+γ 3R 33+γ−m (4. ?].153)
α Ro = ρvo ( )m .156)
and A is a constant give by
3+γ A = (ρo − χρvo )Ro . Assuming that ρv dominate over the ﬁrst term at some stage of evolution. the
. Firstly if one assumes γ p≡ ρ (4.7. m R R Integrating the conservation law we have ρ= where χ≡ A + χρv R3+γ m 3+γ−m
(4.152) 3 then the conservation law follows as 1 R3+γ Let’s assume that ρv ≡ d (ρR3+γ ) = −ρv .

Secondly.161)
(4. 3+γ−m (4.159) R 3 3 ¨ can still give postive R even if p is not negative.162)
Deﬁne
(4. ¨ the key point is not as much having p negative but rather having R positive. This agrees with α our previous consideration that if ρ = Rm dominates the velocity equation then m < 2 leads to inﬂation. (4. Note however that θ can be negative for other values of m as shown in Table 4. From (4.
.160).159) and (4. and p = γ ρ. The pressure that we would want to be negative 3 would be the vacuum pressure pv .164) 3 R 3+γ−m θ≡ (3 + γ)(m − 2) 3+γ−m 3+γ . which we shall work out below.160)
Let us evaluate the right hand sides of these two equations (4.164) then R will be positive for m < 2. The equation ¨ 4πG 8πG R −qH 2 ≡ =− (ρ + 3p) + ρv (4.2.155) we have ˙ R 8πG A k H 2 ≡ ( )2 = [ 3+γ + (χ + 1)ρv ] − 2 R 3 R R where χ+1= Also assuming p= γ ρ we have 3 −qH 2 ≡ ¨ R R = − 8πG 4πG A (4.74
CHAPTER 4. The Friedmann equation is ˙ R 8πG k H 2 ≡ ( )2 = (ρ + ρv ) − 2 R 3 R (4. ENERGY-MOMENTUM TENSOR
pressure p is not the pressure of radiation or matter or vacuum because A ρ = R3+γ + χρ.163) (1 + γ)( 3+γ + χρv ) + ρv 3 R 3 4πG (3 + γ)(m − 2) A = − [1 + γ) 3+γ + ρv ]. because the ρv term has to be considered.165)
we see that θ is always negative for m < 2! Thus if the vacuum term domi¨ nates equation (4.

5 3 4 5 6 θ(γ = 0) -1.170) (4.172)
This is achieved if we make the following deﬁntions
. ρ R 3 3 (4.5 -1.169) (4.8 -1.33 4 ∞ -12 -8 Table 4.2: θ as a function of γ and m
75
Having established that a decaying Cosmological Constant can lead to negative pressure.171) (4.166)
(4.7.5 1 1.164) we have −qH 2 ≡ ¨ R R = − 4πG [(1 + γ)ρ − 2ρv ] 3 4πG = − [(1 + γ)˜ + (m − 2)˜v ] ρ ρ 3 (4. COSMOLOGY WITH THE SCALAR FIELD m 0.3 -0.4. Looking at (4.5 -4 θ(γ = 1) -1.7 -1.161) let us deﬁne ρ≡ ˜ and ρv ≡ (χ + 1)ρv = ˜ so that ρ + ρv = ρ + ρv ˜ ˜ giving ˙ R 8πG k 8πG k H 2 ≡ ( )2 = (ρ + ρv ) − 2 = (˜ + ρv ) − 2 ρ ˜ R 3 R 3 R and from (4.168) A R3+γ 3+γ ρv 3+γ−m (4.5 2 2.0 ∞ -6 -4.8 0 1. let us now work out the vacuum equation of state for a decaying Cosmological Constant.0 0 3.167)
which we would like to write as −qH 2 ≡ ¨ R 4πG 4πG p =− (˜ + 3˜) − ρ p (˜v + 3˜v ).

161) the density is ρ + ρv = A + (χ + 1)ρv R3+γ Ro Ro = (ρo − χρvo )( )3+γ + (χ + 1)ρvo ( )m R R Ro 3+γ Ro m Ro = ρo ( ) + ρvo [(χ + 1)( ) − χ( )3+γ ] R R R
A R3+γ
(4.76
CHAPTER 4.175)
We previoulsy deﬁned ρ ≡ ˜ alternatively deﬁne
and ρv ≡ (χ + 1)ρv .173)
and pv ≡ ˜
m−3 ˜ 3 ρv
≡
γv ˜ 3 ρv
. It is also satisfying to note that the equation of state for the non-vacuum component (equation 4. We present alternative ˜ ˜ deﬁnitions below which will give the same vacuum equation of state but diﬀerent for the non-vacuum component.173) is identical to the equation of state for a perfect ﬂuid that we encountered for models without a Cosmological Constant. (4.176)
ρ ≡ ρo ( ¯ and
ρv ≡ ρvo [(χ + 1)( ¯
Ro m Ro 3+γ ] ) − χ( R R ρvo = (χ + 1)ρv − χ ρ ¯ ρo
(4.177)
. However we might ˜ Ro 3+γ ) R (4.174)
which is our vacuum equation of state for a decaying Cosmological Constant. For a m < 2 we have ˜ γv < −1 which we saw previously is the condition for inﬂation assuming vacuum domination of the density and pressure. Alternative Derivation Our deﬁnition of ρ and ρv above are not unique. In equation (4. ENERGY-MOMENTUM TENSOR p ≡ γρ ˜ 3˜ (4. We see that for m< 3 we have pv and γv negative.

177).171) ¨ R R = − 4πG [(1 + γ)ρ − 2ρv ] 3 4πG ρvo = − )¯ + (m − 2)¯v ] ρ ρ [(1 + γ − m 3 ρo
−qH 2 ≡
(4. We began with the Lagrangian in equation (2. In section (2.20) ˙ R k 8πG L = −κR3 [( )2 − 2 + (ρ + ρv )] R R 3 (4. ρ R 3 3 (4.181)
and pv ≡ ¯
m−3 ¯ 3 ρv
=
γv ¯ 3 ρv
(4.4. 2 ∂R 3
(4.2) we dervied the wheeler-DeWitt equation in minisuperspace approximation.178)
we have (compare to equation 4. ˜ ˜ ¯ ¯ Assuming p =
γ 3ρ
77
(4.182)
4.179)
(where we have used the second expression in 4. COSMOLOGY WITH THE SCALAR FIELD still obtaining ρ + ρv = ρ + ρv = ρ + ρv .11
Quantization
All of our proceeding work with the scalar ﬁeld was at the classical level.180)
This is achieved if we make the following deﬁnitions p≡ ¯
γ−m ρvo ρo 3
ρ ≡ Γρ ¯ 3¯ (4.183)
∂L ˙ and identiﬁed the conjugate momentum p ≡ ∂ R = −2κRR and derived the ˙ ∂ Wheeler-DeWitt equation. as
{−
∂2 8πG + 4κ2 [kR2 − (ρ + ρv )R4 ]}ψ = 0. after quantizing with p → i ∂R . which we would like to write as −qH 2 ≡ ¨ R 4πG 4πG p =− (¯ + 3¯) − ρ p (¯v + 3¯v ).7.184)
.4. In this section we wish to consider quantum eﬀects.7.

184).185)
˙ R 8πG 1 ˙ 2 k ( )2 = ( φ +V)− 2 R 3 2 R and also ¨ ˙ φ + 3H φ + V = 0
(4.188)
. then a tunnelling potential will only be present for m < 2. ?. 20. We call this hypothesis Quantum Inﬂation. In our work on inﬂation we found that for ρ ∝ R1 dominating the Friedmann equation then inﬂation occurs m for m < 2.” In terms of scalar ﬁelds then this quantization procedure is diﬀerent to what other people do with quantization. either ρ or ρv . With our discussion of the scalar ﬁeld we have m written ρφ ∝ R1 so it would seem that the idea of quantum inﬂation also m works for scalar ﬁelds. Thus inﬂation and quantum tunnelling require the same condition. that behave like ρ ∝ R1 .186)
(4. We begin with the Lagrangian ˙ R k 1˙ L = −κR3 [( )2 − 2 ] + 2π 2 R3 [ φ2 − V (φ)] R R 2
3π where κ ≡ 4G . Let us now study this procedure. 3 2 2 (NNN see FE) The canonical momenta are
ΠR ≡
∂L ˙ = −2κRR ˙ ∂R
(4. i.187)
¨ R ˙ ˙ provided one uses R = − 4πG (ρ + 3p) with ρ = 1 φ2 + V and p = 1 φ2 − V . From this one can deduce that
(4. inﬂation is simply a classical description of quantum tunnelling. Quantum inﬂation is easy to validate for ordinary densities. This leads us to the hypothesis that inﬂation and quantum tunnelling are identical! Or in other words. If this density also dominates ρ + ρv in the Wheeler-DeWitt equation.DeWitt equation in terms of both of these variables.e. The usual procedure [?.78
CHAPTER 4. Our intention has been to simply insert this ρ(R) into the Wheeler-DeWitt equation (4. In our quantization procedure we ”didn’t do anything to the density.” In the work that we have done in the present chapter we have made an eﬀort to write the scalar ﬁeld as a function of R. ?. φ = φ(R) and ˙ using ρ = 1 φ2 + V (φ) we have written as an eﬀective density ρ(R) for 2 the scalar ﬁeld. 21] is to quantize φ and R seperately and arrive at a Wheeler. ENERGY-MOMENTUM TENSOR
Notice that our quantization ’didn’t do anything to the density.

˙ ∂φ
3π 4G )
(4. This Hamiltonian is inherited to equation (11.194)
which is exactly analogous to our original Wheeler-DeWitt equation (2.186).193)
In order to compare to our signal Wheeler-DeWitt equation let’s replace Πφ ˙ with Πφ = 2π 2 R3 φ.191)
where the result H = 0 is obtained by comparing the expression for the Hamiltonian to the Friedmann equation (4.190) (4.7. 2 Writing H in terms of the conjugate momenta we have H = −κR3 [ Π2 d 8πG Π2 φ R + 2− + V )] = 0 ( 4κ2 R4 R 3 8π 4 R6 (4.34).462 of the book by Kolb and Turner [21]. This Hamiltonian is exactly analogous to the Hamiltonian we had in equation (2.24) ˙ where we had ρ instead of 1 φ2 + V . COSMOLOGY WITH THE SCALAR FIELD and Πφ ≡
79
∂L ˙ = 2π 2 R3 φ.4. This equation is re-arranged as Π2 − R 3 2 2 9π 2 8πG Πφ + (kR2 − V R4 ) = 0. 2 2 4πG R 4G 3 (4. of course is also equal to zero.192)
which.195)
(4. Pg.196)
.193) is quantized by making the replacements ΠR → −i and Πφ → −i and setting Hψ = 0 to give ∂ ∂R ∂ ∂φ (4.22) where we had ρ ˙ instead of 1 φ2 + V . which results in Π2 + R 9π 2 8πG 1 ˙ 2 [kR2 − ( φ + V )R4 ] = 0 2 4G 3 2 (4. 2 Equation (4.189)
The Hamiltonian (H = pi qi − L) becomes (using κ = ˙
˙ ˙ H = ΠR R + Πφ φ − L ˙ R k 8πG 1 ˙ 2 = −κR3 [( )2 + 2 − ( φ + V )] ≡ 0 R R 3 2
(4.

Pg277 of the book by Linde [36].197)
which is the Wheeler-DeWitt equation in minisuperspace approximation for a quantized scalar ﬁeld φ.5 Now put your solution from problem 4.25).1. ee 4. Thus Quantum Inﬂation still works for U (R. Recall that ρ(R) ˙ and V (R) in terms of φ(R) obvioulsy ρ. In equation (10.56). Pg.4 A) Solve the diﬀerential equation (4. Thus our potential U (R. This is identical to equation (10. V and φ2 must have the same R 1 1 dependence.50). We identify the potential as U (R. φ) will alsways exhibit a tunnelling shape for m < 2. φ) when φ is quantized seperately. so that our result does agree with Linde.49) yields the equation of motion (4.35). 4.53) and solve for R(t).1.
+
· j = 0 where
4. Pg. he gives an expression for V (R.463 of the book by Kolb and Turner [21]. φ) =
9π 2 (kR2 4G2
−
8πG 3 V
R4 ) (4.8
Problems
∂ρ ∂t
4.4 into the Friedmann equation (4. φ) = 3π V (a).80
CHAPTER 4.2 Show that the above equation of continuity also results from taking the divergence of Amp`r´s law. 4. G We can see that the above method of quantizing the scalar ﬁeld φ directly is still consistent with our idea of Quantum Inﬂation.
.3 Show that the Lagrangian in equation (4.198)
which is identical to equation (11.270 of the book by Linde [36]. ENERGY-MOMENTUM TENSOR
∂ [− ∂R2 +
2
3 1 ∂2 4πG R2 ∂φ2
+
9π 2 (kR2 4G2
−
8πG 4 3 VR )]ψ
=0 (4. j).1 Show that ∂µ j µ = 0 is the equation of continuity j µ ≡ (ρ. B) Check your answer by substituting your solution back into the equation. Thus if ρ ∝ Rm then also V ∝ Rm in the same way.
4.11).

B) If H 2 ≡ ( R )2 = constant m R show that R ∝ eHt .
show that R = Ro e
8πG Vo (t−to ) 3
.8 Show that inﬂationary solutions are characterized by R > 0 and non¨ inﬂationary expansions by R < 0 4. √ 2A ln t is a
. 4. R 4.8. PROBLEMS
˙ ˙
81
R 4. Check that φ(t) = soluttion and evaluate the constants β and λ in terms of A.
¨ 4. show that R ∝ t2/m .108).4.6 A) If ( R )2 ∝ R1 .10 Prove equations (4.9 Barrow’s model is V (φ) ≡ βe−λφ .103)-(4.7 If ( R )2 = ˙ 8πG 3 Vo .

82
CHAPTER 4. ENERGY-MOMENTUM TENSOR
.

83
. In actual practice. Actually there are only 10 independent equations because of the symmetry gµν = gνµ . Let us learn how to derive the Friedmann-RobertsonWalker (FRW) metric which is the metric appropriate to a homogeneous and isotropic universe but where size can change with time. the orbits of planets. etc. Thus we need to learn how to derive the metric for the spaces under consideration.Chapter 5
EINSTEIN FIELD EQUATIONS
The Einstein’s ﬁeld equations are Gµν = 8πGTµν + Λgµν (5. Just write down Tµν and solve for gµν which speciﬁes the metric. In practice the solution of the Einstein ﬁeld equations are exceedingly diﬃcult and only a few exact solutions are known. Two excellent reference for this section are Chapter 7 of the book by Lawden [?] and Chapter 2 of the book by Islam [13]. In principle our job is easy. Then we can calculate the paths of light rays. This metric is substituted into the Einstein equations and one solves for the unknown coeﬃcients. the way one usually solves the Einstein’s equations is to specify a metric in general terms which contains unknown coeﬃcients.1)
which are a set of 16 coupled equations which will give gµν (buried inside Gµν ) gives Tµν .

(5. Below we shall restrict our discussion to the spatial part of the metric denoted as d 2 = hij dxi dxj . The space looks like R2 but actually it is E 2 because we can ﬁnd coordinate such that (5. (5.2)
over the whole space then the space is said to be Euclidean and is denoted by E N .1
Preview of Riemannian Geometry
Polar Coordinate
A general N-dimensional Riemannian space. denoted by RN .7)
Here h11 = 1. FRW and special relativity metrics) ds2 = c2 dt2 − d so that hij = −gij . is one in which the distance ds between two neighboring points can be written (Pg. h22 = r2 .84
CHAPTER 5. If coordinates can be found such that ds2 = dxµ dxν (5.6)
= dr2 + r2 dθ2 .
.1
5. These coordinate are the two-dimensional plane polar coordinates x = r cos θ y = r sin θ in which d
2
(5.3) is true. Consider the two dimensional space where d
2 2
(5.8) (5. 88 [?]) ds2 = gµν dxµ dxν . h12 = h21 = 0. Clearly E N is a special case of RN .4) We shall very often have (eg. EINSTEIN FIELD EQUATIONS
5.5)
(5.3) (5.
(5.9)
= dx2 + dy 2 .10)
with g11 = g22 =1 and g12 = g21 = 0 or gµν = δµν .1.

eθ and eφ directions (5.17) (5.19) In 2-d plane polar coordinates d
r
= dr and d
θ
θ
= rdθ to give (5. and d z = dz to give dV = d x d y d z = dxdydz.21)
.1. PREVIEW OF RIEMANNIAN GEOMETRY Recall that for 3-dimensional spherical polar coordinates x = r sin θ cos φ y = r sin θ sin φ z = r cos θ and the increments of length d r .1. On the surface of the sphere the distance between two points is d
2
= d
2
2 θ
+d
2
2 φ 2 2 2
(5. h22 = r2 sin2 θ. y such that ds2 = dx2 + dy 2 and therefore the surface of a sphere is not E∈ but rather a genuine R∈ space.2
Volumes and Change of Coordinates
The measure is the volume element in some set of coordinates.12) (5. d respectively are d d d
r θ φ φ
85
(5.13)
is the er .11) (5. For this surface it is not possible to ﬁnd x. 1) Inﬁnitesimal Length Method is one in which one identiﬁes the inﬁntesimal increments of length and simply multiplies them together to get the volume element. d θ . There are 3 ways to calculate the measure. In 3-D spherical polar coordinates d dr. d θ = dθ and d φ = r sin dφ to give dV = d r d θ d
φ
=
= r2 sin θdrdθdφ.
5.20)
r
dV = d r d
= rdrdθ
(actually this ”volume” is an area). (5.18)
= r dθ + r sin θdφ
where h11 = r2 . h12 = h21 = 0.15) (5.5.
(5.14) (5. In Cartesian Coordinates we have d x = dx. d y = dy.16)
= dr = rdθ = r sin θdφ.
Thus the surfaces of a sphere is an example of a space which is R2 and cannot be reduced to E 2 .

v) | dudv (5. v.33) = r2 sin θ (5. w) ≡
∂y ∂u ∂z ∂u
(5. z = r cos θ) J(r.28) which then reproduce equations (5. 3) Metric Tensor Method is what we prefer in general relativity.26)
The volume element in 2-d is (5.746) for changing variables. θ. v) ≡
∂x ∂u ∂y ∂u ∂x ∂v
(5. z) = 1. v. w) | dudvdw (5. y)dxdy = f [x(u. w) we have
∂x ∂u ∂x ∂v ∂y ∂v ∂z ∂v ∂x ∂w ∂y ∂w ∂z ∂w
J(u. w) and z = z(u. y. v. y(u. y = y(u. y = r sin θ) J(r. EINSTEIN FIELD EQUATIONS
2) Jacobian Method is the one usually mentioned in introductory calculus books ([?] Pg. Suppose x = x(u. y = r sin θ cos φ.86
CHAPTER 5. Here ([21].19).139) and (5. For plane polar coordinates (x = r cos θ. plane polar and spherical polar coordinates. v) then f (x. w).140) for Cartesian. v) and y = y(u.27)
. Pg. θ) = cos θ −r sin θ sin θ r cos θ =r (5. v).24)
For Cartesian coordinates obviously J(x.22)
where | J(u. v) | is the modulus of the Jacobian deﬁned as J(u. v. v) | dudv and in 3-d it is dV =| J(u. φ) = sin θ cos φ r cos θ cos φ −r sin θ sin φ sin θ sin φ r cos θ sin φ r sin θ cos φ cos θ −r sin θ 0 dV =| J(u.25)
and for spherical polar coordinates (x = r sin θ cos φ. v)] | J(u. (5.23)
∂y ∂v
For 3-dimensions with x = x(u. v.

140).34) depending on the number of dimensions.31)
√ √ giving h = r so that dV = hdrdθ in agreement with (5.139). PREVIEW OF RIEMANNIAN GEOMETRY dV = √ hdudvdw
87
(5. Thus deﬁne h= d2 x ≡ dudv d3 x ≡ dudvdw d4 x ≡ dudvdwdt so that the measure is dV = √ hd2 x or √ hd3 x (5.1. For spherical polar coordinates d 2 = dr2 + r2 dθ2 + r2 sin2 θdφ2 ≡ hij dxi dxj giving 1 0 0 0 r2 0 = r4 sin2 θ (5. For example in spherical polar coordinates d3 x ≡ drdθdφ or with time d4 x ≡ drdθdφdt. The measure is volume is obtained with dV = hd3 x or −gd4 x (because for 4-d we use gµν and hij = −gij and h = −g).35) (5. Thus in general √ √ (5.33)
.30) For a plane polar coordinates ds2 = dr2 + r2 dθ2 ≡ gij dxi dxj so that h= 1 0 0 r2 = r2 (5.32) 2 sin2 θ 0 0 r √ √ giving h = r2 sin θ so that dV = hdrdθdφ = r2 sin θdrdθdφ in agreement with (5. It is important to remember that d3 x or d4 x om this notation is not d 1 d 2 d 3 or d 1 d 2 d 3 d 4 but simply only the coordinates.36) (5.5.29) where h is the determinant of the spatial metric tensor. Thus √ | J |= h (5.

285-290).40) (5. For many curves the form y = y(x) can be clumsy and nowadays mathematicians always prefer the parametric representation. Previously we wrote the deﬁnition of a circle as x2 + y 2 = r2 which could be written generally as y = y(x).31-35) and Landau ([?]. say in undergraduate physics books. v)ˆ2 + x3 (u.197-200) and the article by Kung [?]. v). Pg. for example.570). If the basis is chosen to be E 2 then x = x(t) is equivalent to two scalar equations x1 = x1 (t) and x2 = x2 (t). A general surface (Chapter 8 of [?]) is expressed as x = x(u.177-181) and Ohanian and Ruﬃni ([8]. Pg.1. Another way to write our equation for the circle (radius=1) is x = cos θˆ1 + sin θˆ2 e e (5. Pg.
.38)
where t is the parameter. Pg546-552) and Chow ([28].39)
Good references of the next 3 sections are the books by Kolb and Turner ([21]. EINSTEIN FIELD EQUATIONS dV = d3 x
or dV = d4 x in contrast to conventions used.88
CHAPTER 5. The same equation can be expressed parametrically in terms of the parameter θ as x = r cos θ and y = r sin θ or generally as x = x(θ) and y = y(θ). e e e (5. Introductory material is discussed by Purcell and Varberg [?] (Pg. v)ˆ3 .625-634. If. (See [?] Pg. the basis in E 3 then x(u.
5.37) ˆ where e1 and e2 are basis vectors in E 2 . For our circle above we have x1 = x1 (θ) = cos θ and x2 = x2 = sin θ. Thus a curve can be speciﬁed in any number of dimensions. v) = x1 (u. v)ˆ1 + x2 (u.3
Diﬀerential Geometry
A Good reference for diﬀerential geometry is the book by Lipschutz [?]. ˆ Thus a general curve (Chapter 3 of [?]) is expressed as x = x(t) (5.

R2 − x2
.42) picks our only those points in E 2 which give the circle.45)
Note that dy 2 ≡ (dy)2 and dy 2 = d(y 2 ). PREVIEW OF RIEMANNIAN GEOMETRY
89
5.1.42)
which we recognize as the equation for a circle. Recall that for a circle the radius R.44)
In an ordinary 2-d Euclidean space x and y (or x1 and x2 ) are free to vary independently and this is how the whole 2-d space get covered. often called the one sphere denoted by S 1 .46)
R2 dx2 .44) covers all of E 2 . R2 − x2 (5.42) or (5. However the reason that (5.5.43) describes a circle is because it constrains the value of y in terms of x. We do this using the 2-d constraint (5.41)
However in introductory books one always introduces a ﬁcticous extra dimension and embeds the 1-d curve in a 2-d Euclidean space via x2 + y 2 = R2 (5. the proper way to express it is in terms of 1-dimensional parameter θ as x(θ) = R(cos θˆ1 + sin θˆ2 ). We can also write x2 + x2 = R2 .43) and writing d −x y = R2 − x2 and dy = √R2 −x2 dx so that dy 2 = x2 dx2 . e e (5. 1 2
(5. Thus we will have√ for the circle.44) is true in general.44) becomes d
2
= dx2 + =
x2 dx2 R2 − x2 (5. This constraint (5. Thus (5. Equation (5. 1 2 The element of length in the 2-d Euclidean space is d
2
(5. We can constrain it for the circle by reducing the two parameters x1 and x2 to only one parameter.1.4
1-dimesional Curve
Let us ﬁrst consider the circle. Equation (5. It is a 2-d equation for a 1-d curve! The 1parameter equation (5. Remember though this equation is really overkill.43)
= dx2 + dx2 .41) is much better.

To restrict ourselves to the rim of the circle (curved 1-d space) we ﬁx d R = dR = 0 and get d
2
= R2 dθ2
(5.52)
which makes it obvious that the space is the one sphere (circle) of radius R.46).50)
= d
= dR + R dθ2
2 R+ 2
d
2 θ 2
(5. Another convenient coordinate system for the circle use the angle θ from plane polar coordinates speciﬁed via = R2 x = R cos θ y = R sin θ. This allows us to calculate the volume (we are calling the length a geological volume) as V = √ hd1 x =
0 2π
Rdθ = 2πR
(5. Identifying the increments of length d as d d then d
2 R θ R
(5.49)
θ
and d
in the eR and eθ directions ˆ ˆ
= dR = Rdθ (5. (Do Problem 5. Using simple trigonometry one can show that (5.54)
.90
CHAPTER 5.53)
which is a 1-dimensional ”motion”.47)
dr2 (5. The determinant is obviously h = R2 √ giving h = R.52) is the same as (5.51)
which gives the distance d in the 2-d space.1) Using d 2 ≡ hij dxi dxj we evidently have hij = (R2 ) (5. EINSTEIN FIELD EQUATIONS
This can also be written in terms of the dimensionless coordinate x r≡ R to give d
2
(5.48) 1 − r2 where R is the radius of the space (the circle).

(5.57)
These results are also obtained by embedding in Minkowski space ( do Problem 5. S 1 and H 1 respectively)
.2). PREVIEW OF RIEMANNIAN GEOMETRY
91
S1. S 1 and H 1 into a single formula d or using r ≡
x R 2
=
R2 dx2 R2 − kx2
(5. closed and open curves respectively.e. The line element for a space of zero curvature is obviously just d or using r ≡
x R 2 2
= dx2
(5.
The 1-d curve that we described above is the circle or one sphere denoted However there are three 1-d spaces which are homogeneous and isotropic. We can collect our results for R1 .1. +1.55) to yield d
2
= =
−R2 dx2 −R2 − x2 R2 dx2 R2 + x2 dr2 .59)
These formulas are obtained from S 1 or H 1 by letting R → ∞. −1 for ﬂat. 1 + r2
(5.56)
or using r ≡
x R
d
2
= R2
(5.58)
d
= R2 dr2 . for R1 .61) where k = 0. ii) the positively curved one sphere (S 1 ) derived above and iii) the negatively curved hyperbolic curve (H 1 ). [21] The formulas for a space of constant negative curvature [21] can be obtained with the replacement R → iR (5.60)
d
2
dr = R2 1−kR2
2
(5.5. (i. There are i) the ﬂat x line (R1 ).

62)
which we recognize as the equation for a sphere used in introductory books. v) as mentioned previously.63) restricts x3 according to x2 = R2 − x2 − x2 .68)
= dx2 + dx2 + 1 2
x1 dx1 + x2 dx2 )2 R2 − x2 − x2 1 2
(5. 3 1 2 Writing dx3 = and with
∂x3 ∂x1
(5. Thus with 3 parameters the equation for the two sphere is x2 + y 2 + z 2 + R2 (5.69)
.65)
∂x3 ∂x3 dx1 + dx2 ∂x1 ∂x2
(5. However we shall introduce a surface ﬁcticious coordinate (three parameters) and embedding the surface in E 3 .67)
dx3 = − or dx2 = 3 to give ([21].40). Other references are listed following equation (5. It is a 3-d equation for a 2-d surface. A surface is represented by two parameters u. We can also write x2 + x2 + x2 = R2 .1. (5.66)
=√
−x1 R2 −x2 −x2 1 2
we have x1 dx1 + x2 dx2 R2 − x2 − x2 1 2 (5. However this equation is overkill.32) d
2
(x1 dx1 + x2 dx2 )2 R2 − x2 − x2 1 2
(5. EINSTEIN FIELD EQUATIONS
5.63) 1 2 3 Any 3-d Euclidean space E 3 has length element d
2
= dx2 + dx2 + dx2 1 2 3
(5.64)
which under normal circumstances would map out the whole 3-d volume.92
CHAPTER 5. Pg. However (5. v and expressed as x = x(u.5
2-dimensional Surface
Kolb and Turner [21] analyze this problem very nicely.

75)
= R2 [
dr2 + r2 dθ2 ] 1 − r2
(5.79) (5.70) become d
2
: :
0 → 2π 0 → R.70) Let us introduce plane polar coordinates in the x3 plane as = x1 = r cos θ x2 = r sin θ.
(5. (5. −r2 r R
(5.73)
=
R2
(5.77) Another convenient coordinate system for the two sphere uses angles θ and φ from spherical polar coordinates speciﬁed via x = R sin θ cos φ y = R sin θ sin φ z = R cos θ and substituting into (5.80)
= R2 (dθ2 + sin2 θdφ2 ).69) and (5.1. Thus θ r Using dxi = then (5.1. (5.76)
(5.32 of Kolb and Turner [21].72)
∂xi ∂xi dr + dθ ∂x1 ∂θ R2 dr 2 + r 2 dθ2 . PREVIEW OF RIEMANNIAN GEOMETRY which is re-written as d
2
93
1 [(R2 −x2 )dx2 +(R2 −x2 )dx2 +x1 x2 dx1 dx2 +x2 x1 dx2 dx1 ].5.71)
These coordinates are shown very clearly in Fig 2. (5. Pg.81)
.74)
This can also be written in terms of the dimensionless coordinate r≡ to give d where r : 0 → 1.78) (5. 2 1 1 2 R2 − x2 − x2 1 2 (5.64) directly yields d
2 2
(5.

As before we can obtain the formula for H 2 with the replacement R → iR to yield d
2
(5. To restrict ourselves to the surface of the sphere (curved 2-d space) we ﬁnd d R = dR = 0 and get d
2
= R2 (dθ2 + sin2 θdφ2 )
(5.90)
= R2 [
dr2 + r2 dθ2 ].82) (5. ii) the positively curved two sphere (S 2 ) and iii) the negatively curved two hyperbola (H 2 ).87)
sin theta
0
dφ = 4πR2
(5. Using d 2 ≡ hij dxi dxj we evidently have for S 2 hij = The determinant is obviously h = R4 sin2 θ (5.86)
in agreement with (5.81) is alternatively obtained by identifying the increments of length d R .89)
Actually there are three 2-d spaces which are homogenous and isotropic.88) √ giving h = R2 sin θ. d φ in the eR . 1 + r2
(5. d θ . The volume (we are calling the surface area a generalized volume) is V = √ hd2 x = √ hdθdφ + R2
0 π 2π
R2 0 0 R2 sin2 θ
(5. EINSTEIN FIELD EQUATIONS
Equation (5.85)
gives the distance ds in the 3-d space.83) (5.84)
= d
= dR + R (dθ2 + sin2 θdφ2 )
2 R+ 2
d
2 θ 2
+d
2 φ
(5. eθ and eφ directions as ˆ ˆ ˆ d d d then d
2 R θ φ
= dR = Rdθ = R sin θdφ
(5.91)
.94
CHAPTER 5.81). There are 1) the ﬂat x − y plane (R2 ).

98)
where r = 0 at θ = 0 and r = R at θ = π and r = 0 again at θ = π.
(5.5.180 of the book by Lawden [?] for more explanation) The integral and its limits are more clearly done with the substitution √ kr ≡ sin χ (5. 2 (See bottom Pg. closed (S 2 ) and open (H 2 ) surfaces respectively.179 and top pg. The metric corresponding to (5.93)
where k=0. S 2 and H 2 into a single formula d
2
= R2 (
dr2 + r2 dθ2 ) 1 − kr2
(5.96) (5.94)
R2 r 2 1 − kr2
(5.95)
The volume is √ hd2 x =
0 0
V
=
√
h = R2
0
0
√
rdr 1 − kr2
2π
dθ
0
(5. +1.92)
We can collect our results for R2 .97)
= 2πR2
√
rdr .34. equation 2. The volume can be alternatively calculated using d 2 = hij dxi dxj in (5. What this really means is 0 0
dr ≡ 2
R
dr
0
(5. Pg.32.93) we have hij = giving the determinant h= or √ h=
√R r . Pg.81) is ([21].17) d
2
= R2 (dθ2 + sin2 θdφ2 ). -1 for ﬂat (R2 ).99)
.1. PREVIEW OF RIEMANNIAN GEOMETRY
95
This result is also obtained by embedding in Minkowski space. 1−kr2
2
R2 1−kr2
0 R2 r 2
0
(5.1 of Kolb and Turner [21]. 1 − kr2
0 The limits of integration 0 dr can be clearly seen from Fig 2.

5.100) Thus the volume in equation (5. w and is expressed as x = x(u.96 when
CHAPTER 5.6
3-dimensional Hypersurface
Proceeding upwards in our number of dimensions we might inquire a ”curved volume”. But the curvature can really only be imagined with respect to embedding in a 4-dimensional Euclidean space E 4 . EINSTEIN FIELD EQUATIONS
χ : 0 → π.1.105) (5.
π Thus 0 dχ is equivalent to (5.102)
Thus for k = +1 we have V = 4πR2 as before. We would ﬁnd V = ∞ for k = −1. v. We call the 4-d space as a hypersurface. For k = 0 we have V = ∞ and for k = −1 we need to do the integral again.106) (5.
V = 2πR2 giving
1 1 √ √ sin χdχ k k 4πR2 . Introduce an extra ﬁcticious coordinate for the three sphere S 3 as x2 + y 2 + z 2 + w2 = R2 x2 + x2 + x2 + x2 = R2 1 2 3 4 which restricts x4 as x2 = R2 − x2 − x2 − x2 . v. A hypersurface is represented by three parameters u. k
V =
(5. Our 4-d Euclidean space (into which we will embed the hypersurface) has length element d 2 = dx2 + dx2 + dx2 + dx2 (5.104)
. 4 1 2 3 Writing dx4 = ∂x4 ∂x4 ∂x4 dx1 + dx2 + dx3 ∂x1 ∂x2 ∂x3 (5.101)
=2
π 0
R 0 dr. w).97) becomes 0 0 dr
(5.103) 1 2 3 4 (it is d 2 and not ds2 ).

PREVIEW OF RIEMANNIAN GEOMETRY and with
∂x4 ∂x1
97
=√
−x1 R2 −x2 −x2 −x2 1 2 3
etc. R2 − x2 − x2 − x2 1 2 3
(5.118)
= R2 [
dr2 + r2 dθ2 + r2 sin2 θdφ2 ] 1 − r2
(5.113) (5. we have x1 dx1 + x2 dx2 + x3 dx3 R2 − x2 − x2 − x2 1 2 3 (5.112)
: : :
0 → π 0 → 2π 0 → R.109) becomes d
2
(5.111) (5.117)
Introducing the dimensionless coordinate r≡ gives d
2
r R
(5.114) (5.5.108)
= dx2 + dx2 + dx2 + 1 2 3
(x1 dx1 + x2 dx2 + x3 dx3 )2 .115)
∂xi ∂xi ∂xi dr .
(5. + dθ + dφ ∂r ∂θ ∂φ
(5. Pg.107)
dx4 = − or dx2 = 4 to give ([21].34) d
2
(x1 dx1 + x2 dx2 + x3 dx3 )2 R2 − x2 − x2 − x2 1 2 3
(5.109)
Let us introduce spherical polar coordinates in the x4 hyperplane as x1 = r sin θ cos φ x2 = r sin θ sin φ x3 = r cos θ where θ φ r Using dxi = then (5.116)
=
R2 dr 2 + r 2 dθ2 + r 2 sin2 θdφ2 R2 − r 2
(5.110) (5.1.119)
.

129)
(5.122) (5. θ. V = The volume is √ 3 √ hd x = hdχdθdφ
π 0
(5.124)
= R2 [dχ2 + sin2 χ(dθ2 + sin2 θdφ2 )].119) directly yields d Using d
2 2
(5.121) (5.98 where
CHAPTER 5.126)
The determinant is giving √ h = R6 sin4 χ sin2 θ h= R3 sin2 χ sin θ.130)
for the volume of our hypersphere. φ from 4-dimensional hyperspherical polar coordinates speciﬁed via x = R sin χ sin θ cos φ y = R sin χ sin θ sin φ z = R sin χ cos θ w = R cos χ. Substituting into (5. Thus
π 0
V = 4πR3 giving
sin2 χdχ = 4πR3 [ V = 2π 2 R3
χ sin 2χ π − ] 2 4 0
(5. Compare this to the volume of a Euclidean sphere 4 πR3 .
(5.128)
dχ are the same as in the previous section.120)
Another convenient coordinate system for the three sphere uses angles χ.123) (5.
(5. 3 For a ﬂat.125)
= hij dxi dxj we have R2 0 0 2 sin2 χ 0 hij = 0 R 0 0 R2 sin2 χ sin2 θ
(5. EINSTEIN FIELD EQUATIONS
r : 0 → 1. open and closed hyperspheres the metric is d
2
= R2 [
dr2 + r2 dθ2 + r2 sin2 θdφ2 ] 1 − kr2
(5.127)
= R3 where the limits
π 0
sin2 χdχ
0
π
2π
sin θdθ
0
dφ
(5.131)
.

isotropic curved space with a size R(t) that can change in time is ds2 = R2 (t)[ dr2 + r2 (dθ2 + sin2 θdφ2 )].2
Friedmann-Robertson-Walker Metric
ds2 = c2 dt − (dx2 dy 2 dz 2 ).131) we have
hij =
R2 1−kr2
0 0
6 4
0 0 2 r2 0 R 2 r 2 sin2 θ 0 R
2
(5. We have seen that the spatial metric for a homogeneous.137)
5. (5. Using the substitution √ kr ≡ sin χ (5. k 3/2 For k = +1 this agrees with our result before.134) with χ : 0 → π becuase χ : 0 → π.5.2.132)
r sin giving the determinant h = R 1−kr2 θ .98).139)
.136)
giving 2π 2 R3 . 1 − kr2 0
√
2π
dφ
0
(5. Using d = hij dxi dxj in (5.133)
The limits of integration are the same as discussed in equation (5. Thus the volume is V = 4πR3
0 π
(5.135)
1 k 3/2
sin2 χdχ
(5. FRIEDMANN-ROBERTSON-WALKER METRIC
99
The volume can be calculated alternatively. V = (5. 1 − kr2 (5.138)
The metric of Special Relativity is
Clearly the spatial part is a 3-d Euclidean ﬂat space. The volume is √ 3 √ V = hd x = hdrdθdφ
= R3
0
0
= 4πR3
π r2 dr sin θdθ 1 − kr2 0 0 r2 dr √ .

148) (5.151)
If gµν is represented by a matrix [gµν ].146) (5.152)
.) Thus √ R3 r2 sin θ −g = √ . then we found previously that is just the inverse of this metric namely [gµν ]−1 .100
CHAPTER 5. ±1 depending on the curvature. Thus it’s easy to get g µν g 00 = 1 (5. For a diagonal matrix (which we have for the FRW metric) each matrix element is simply given by 1 g µν = gµν . 1 − kr2 (5.143) (5.145) (5.140)
where R(t) is called the scale factor and the constant k can be 0.141) (5.147) (5.144)
g22 = −R2 r2 g33 = −R2 r2 sin2 θ Deﬁning the determinant g ≡ detgµν = g00 g11 g22 g33 = − R6 r4 sin2 θ 1 − kr2
(5.139) we have the Friedmann-Robertson-Walker (FRW) metric [13] ds2 = c2 dt2 − R2 (t)[ dr2 + r2 (dθ2 + sin2 θdφ2 )] 1 − kr2 (5. Writing ds2 ≡ gµν dxµ dxν and identifying x0 = ct x1 = r x2 = θ x3 = φ we have g00 = 1 g11 = −R2 1 − kr2 (5. This can also be derived with the use of Killing vectors [13].142) (5. EINSTEIN FIELD EQUATIONS
Replacing the spatial part of the special relativity metric with (5.149) (5.150)
(Note that this is not g = detg µν .

0 ) = − g11.α ) 2
(5.γ + gαγ.157)
Γ1 = −r(1 − kr2 ) 22 Γ1 = −r(1 − kr2 ) sin2 θ 33
.153) (5. Fortunately we need not calculate all of them.2.1
Christoﬀel Symbols
We now calculate the Christoﬀel symbols using equation (3.162)
(5. (g µν is a diagonal matrix for the FRW metric. ) = Γα γβ 2 1 αα g (gαβ.0 = − ( 11 2 2 ∂t 1 − kr2 ˙ ˙ 1 ∂R2 1 2RR RR = = = 2 ∂t 2) 2 1 − kr 2(1 − kr 1 − kr2 because r = r(t) and R = R(t). R r sin θ g 11 = −
101 (5.γ + gαγ.β − gβγ.0 = − (−R2 r2 ) = r2 RR 22 2 2 ∂t ˙ Γ0 = r2 sin2 θRR 33 Γ1 = 11 kr 1 − kr2 (5. We have Γα βγ ≡ = 1 α g (g β.155)
5.2.161) (5.159) (5.) The only non-zero Christoﬀel symbols are the following: 1 1 Γ0 = g 00 (g01.β − gβγ.154) (5.1 + g01. We can use the symmetry Γα = Γα to βγ γβ shorten the job. FRIEDMANN-ROBERTSON-WALKER METRIC (1 − kr2 ) R2 −1 g 22 = 2 2 R r −1 g 33 = 2 2 2 .156)
which follows because g α = 0 unless = α.5.158) (5.0 11 2 2 because g01 = 0 and g 00 = 1.160) (5. This becomes (let’s now set c ≡ 1) 1 1 ∂ −R2 ) Γ0 = − g11. Proceeding 1 1∂ ˙ Γ0 = − g22.1 − g11.69).

Also the second term contains Γ0 which is always 0.0 = 3R2 R =√ =√ ∂x ∂t 1 − kr2 ∂t 1 − kr2 and √ √ √ √ √ ∂ ln −g ∂ ln −g ∂ −g 1 ∂ −g √ (ln −g).00 − (Γ01 ) − (Γ02 ) − (Γ03 ) ˙ √ R = −(ln −g).00 − Γ0 Γθ − Γ1 Γθ − Γ2 Γθ − Γ3 Γθ 0θ 00 0θ 01 0θ 02 0θ 03 when we have performed the sum over . Thus 0θ √ R00 = −(ln −g). 2. The term Γθ = 0.163) (5. R33 . so that 0θ 0θ we must have θ = 1.166)
Γ2 = − sin θ cos θ 33 Γ3 = cot θ 23 Γ1 = Γ2 = Γ3 = 01 02 03 (do Problems 5. −(ln −g). so that the non-zero components are R00 . Proceeding we have √ √ 1 R00 = √ (Γ00 −g).2 and 5. EINSTEIN FIELD EQUATIONS Γ2 = Γ3 = 12 13 1 r (5.3)
5. R22 .00 − Γ1 Γ1 − Γ2 Γ2 − Γ3 Γ3 10 01 02 02 03 03 √ 1 2 2 2 3 2 = −(ln −g). 2.2.µ ≡ = =√ ∂xµ ∂ −g ∂xµ −g ∂xµ
. In the last 0θ three terms we have Γα where α = 1. Now Γα = 0 for θ = α.102
CHAPTER 5. R11 .164) (5. For the FRW metric it turns out that Rµν = 0 for µ = ν.2
Ricci Tensor
Using equation (??) we can now calculate the Ricci tensor. 3.165) ˙ R R (5. forth and ﬁfth terms respectively.00 − 3( )2 R Now √ √ √ ∂ −g ∂ −g r2 sin θ ∂R3 r2 sin θ ˙ ( −g).00 − Γ0θ Γθ 0 −g but Γ00 = 0 giving √ R00 = −(ln −g). 3 in the third.

168) (5. FRIEDMANN-ROBERTSON-WALKER METRIC so that
103
√ √ ˙ 1 − kr2 R2 sin θ 1 ∂ −g R ˙ √ (ln −g).2.5).3
Riemann Scalar and Einstein Tensor
We now calculate the Ricci scalar R≡ RRα ≡ g αβ Rαβ . 2 ∂t R R R R We ﬁnally have ˙ R R00 = −3 .0 = √ = 3 2 3R2 R = 3 − −g ∂x0 R r sin θ 1 − kr2 R √
giving ¨ ˙ ¨ ˙ ˙ √ ∂ R R RR − R2 R (ln −g).4)
5. We obtain ˙ R k G00 = 3[( )2 + 2 ] R R −1 ¨ ˙ G11 = (2RR + R2 + k) 1 − kr2 ¨ ˙ G22 = −r2 (2RR + R2 + k) ¨ G33 = −r sin θ(2RR + R + k)
2 2
(5.169) (5.175) (5.170)
¨ ˙ R22 = r2 (RR + 2R2 + 2k) ˙ ˙ R33 = r2 sin2 θ(RR + 2R2 + 2k) (do Problem 5.172)
(do Problem 5. The only non-zero component are for µ = ν.2.173) (5. The only non-zero α contributions are R = g 00 R00 + g 11 R11 + g 22 R22 + g 33 R33 ¨ ˙ R k R = −6[ + ( )2 + 2 ] R R R (5.167)
(5.
.176)
˙2
(do Problem 5. R One can similaraly show that R11 = ¨ ˙ RR − 2R2 + 2k 1 − kr2
(5.171) (5.5. Finally we calculate the Einstein tensor Gµν ≡ Rµν − 1 2 Rgµν .00 = 3 ( ) = 3 = 3 − 3( )2 .174) (5.6).

0) = (1. EINSTEIN FIELD EQUATIONS
5.5
Friedmann Equations
Finally we substitute our results into the Einstein ﬁeld equations Gµν = 8πGTµν + Λgµν .28) for the metric of Special Relativity.181)
5.179)
Tµν =
ρ 0 R2 0 p 1−kr2 0 0 0 0
0 0 0 0 0 pR2 r2 0 pR2 r2 sin2 θ
(5.182)
. −0) = (c.4
Energy-Momentum Tensor
For a perfect ﬂuid the energy momentum tensor is given in equation (4. 0) for c ≡ 1.2.180) because ui = 0.178)
where we shall use gµν from our FRW model.145)-(5.26) as Tµν = (ρ + p)uµν − pηµν (5. For an arbitrary metric in General Relativity we have Tµν = (ρ + p)uµν − pgµν (5. 0) or Uµ = (c.2. For a motionless ﬂuid recall that uµ = (c. Upon substitution of the FRW values for the metric given is equations (5.177) The tensor for T µν is written is (4. The µν = 00 component is ˙ R k 3[( )2 + 2 ] = 8πGρ + Λ R R giving ˙ R 8πG Λ k H 2 ≡ ( )2 = ρ− 2 + .148) we have
(5. R 3 R 3 The µν = 11 component is R2 −R2 −1 ¨ ˙ (2RR + R2 + k) = 8πGp +Λ 1 − kr2 1 − kr2 1 − kr2 (5.104
CHAPTER 5. Thus T00 = ρ + p − p = ρ and Tii = −pgii (5.

PROBLEMS giving 2 ¨ ˙ R k R + ( )2 + 2 = −8πGp + Λ. (5. for example.
k R 5.4 Show that the Ricci scalar is R= −6[ R +( R )2 + R2 ] for the FRW metric. 22 23 ¨ 5.6 Show that the µν = 22 and µν = 33 components of the Einstein’s equations for the FRW metric yield the same equation (5. 5.183) ¨ R 8πG Λ + ρ + = −8πGp + Λ R 3 3
5.1 For the FRW metric show that Γ1 − (1 − kr2 )r sin2 θ and Γ = cot θ.5 Calculate Gµν for the FRW metric.7). R ¨ ˙
5.63) as the µν = 11 component. Γ1 = Γ2 = 0 for the FRW metric.182) to give 2 to ﬁnally give ¨ R 4πG Λ = (ρ + 3p) + R 3 3 (do Problem 5.3 Show that R22 = r2 (RR + 2 cot R2 + 2k) for the FRW metric.2 Show that.3.5. R R R
105
But we now use our previous result (5.3
Problems
5.
. 33 5.

106
CHAPTER 5. EINSTEIN FIELD EQUATIONS
.

he ﬁrst tried Rµν = kT µν etc
107
.g.Chapter 6
Einstein Field Equations
Gµν = kT µν go through history e.

EINSTEIN FIELD EQUATIONS
.108
CHAPTER 6.

Chapter 7
Weak Field Limit
derivation of Gµν = kT µν from equiv princ.
109
.

WEAK FIELD LIMIT
.110
CHAPTER 7.

20). φ term seems
111
. 276 eqn 8.Chapter 8
Lagrangian Methods
Lagrangians for Gµν etc. (NNNN have assumed special relativity g00 = +1) (NNN to disagree with Kolb and Turner Pg.

112
CHAPTER 8. LAGRANGIAN METHODS
.

New York.Bibliography
[1] J.. QC178 . A Gravitation and Spacetime. 1989).A. A First Course in General Relativity.. American J.. Classical Dynamics of Particles and Systems. QC173.W. Davies. 2nd ed. (Cambridge University Press. Atkatz. [5] B. (Wiley. 1980).D Nightingale.O35 [9] J. 1993). New York. New York.Foster and J. Gasiorowicz.. Steinhardt. Ruﬃni. Quantum Physics. QA845 . 1988).D52 [7] J. 62.N3 [8] H. 1988).F67 [3] S.L. (Ellis Horwood Limited. New York. 113
. New York. (W. 1994). [10] A. Narlikar. (Cambridge University Press. QC173.B. Schutz. Brace. (Cambridge University Press. Introduction to cosmology. 1996). Martin. (Harcourt. Ohanian and R. [4] H. 1995). 2nd ed. Geometrical methods of mathematical physics.S38 [6] B. edited by P. Schutz.F. New York. 2nd ed. 1974). New York. in The New Physics. General Relativity. Chichester. QB981.6.F. 1990).7. A Short Course in General Relativity. Phys. (Springer-Verlag. Guth and P. 619 (1994). Jovanovich College Publishers.V. Norton and Company. Introduction to General Relativity.6 . (Cambridge University Press. Marion.M38 [2] J. New York. The Inﬂationary Universe. (Pergamon. England. [11] D. QC20. Atwater. 3rd ed.

1990). Amsterdam. QA37. Phys. 1995). Arfken and H.M.M37 [28] T. New York. QC631. D 25. Chow. (Addison-Wesley. Lett. Kubo. Ohanian. Reading. Phys. Guth. (Addison-Wesley. Marion. 75. 1975). 757 (1994). D 28. Mathematical Methods for Physicists. 2960 (1983). Statistical Mechanics. [16] T.D. (Wiley. Mathematics of Classical and Quantum Physics. 4th ed. Jacobson. Rev. Phys. 1975).C. (Allyn and Bacon. New York. (Pergamon. Phys. New York. 347 (1981). Rev. Kolb and M. L. 1260 (1995) [17] W. 1967). 1974). Schiﬀer. Pagels. Byron and Fuller. Winnipeg.J.S. Introduction to General Relativity.W. The cosmological constant is back.
.. Turner. M. 1988). [19] A.. (Academic Press. 1965). (Wuerz Publishing Ltd. D 23.114
BIBLIOGRAPHY
[12] R. Adler. 1 and 2. New York. Turner.A74 [25] H. [21] E. QC631. Masachusetts. Krauss and M. vols.O43 [26] J. 1994).B9 [24] G. Weber.S. (Academic Press.J3 [27] J.B. Classical Electrodynamics.B. Hawking. Jackson. General Relativity and Cosmology. [13] J. Boston. Introduction to General Relativity. 2065 (1982). Rev. (Cambridge University Press. [20] J. [18] L. Atkatz and H.A. Canada. 1137 (1995). Hartle and S.2. QC20. 27. Bazin. [15] R. (McGraw-Hill. New York. The Early Universe. 1992). [23] F. Classical Electromagnetic Radiation. Nature D 371. [22] D. (North-Holland. Rev. QC631. Islam. Atwater. General Relativity and Gravitation. An Introduction to Mathematical Cosmology. Classical Electrodynamics. San Diego. [14] H.N. Freedman et al.W. 1969). M.

1993). New York. (Chapman and Hall. 1995). Basis of Modern Cosmology (Editions Fronti`res. [33] P. e France. The physics of elementary particles. QC721. B. 91192 Gif-Sur-Yvette Cedex. Walecka. D’Inverno. 1965). D. Cosmic No-Hair Theorem and Inﬂation. 1990)
. B. New York. Sazhin and Y. Negele and E. Zeldovich.E.L46 [32] R. [35] J.M94 [31] M.S. B. D. New Jersey.D. (Academic Press. J. 1973).W. Princeton. Peebles. Dolgov. Collins. [34] B. 1993). Vogt (Plenum Press. 112. New YOrdk. New York. Linde. New York. Introducing Einstein’s Relativity.J. New York. Serot and J. Principles of Physical Cosmology. 1 (1986). The Relativistic Nuclear ManyBody Problem. B. D. QC793. Martin and E. 1990) [37] P. A. The Dynamic Cosmos. Leon. 1987) [38] A.P. Madsen. V. D. [30] H. Particle Physics and Cosmology. 187. Barrow. (Clarendon Press. D. Muirhead. Oxford. edited by J. Inﬂation and Quantum Cosmology.2. Particle physics: an introduction.BIBLIOGRAPHY
115
[29] M.33. Advances in Nuclear Physics. Squires. (Princeton University Press. (1987) [36] A. (Pergamon Press. 1986). D. (Wiley. 16. (Academic Press. Physics Lett. M.