4 Introduction to General Relativity

In this chapter we discuss Einstein’s General Theory of Relativity, a theory to describe gravity.

4.1 The Equivalence Principle

According to the Newton’s second law of motion, a force applied on an object is proportional to its
acceleration
F~ = mi~a ,
where the constant of proportionality mi is called the inertial mass. On the other hand, the gravitational
force that acts on an object is proportional to the gravitational acceleration

F~g = mg ~g ,

where the mg is called the gravitational mass. There is no reason to think that inertial mass and the
gravitational mass are the same – mathematically, they are just constant of proportionality. Since the very
first experiment by Eotvos, several ever precise experiments have shown that the mi /mg ratio is equal to
unity to very high accuracy. That the inertial and gravitation mass are equal is called the weak equivalence
principle of Newton.
There are important consequences of this principle: Firstly, it is impossible to distinguish between
gravity and an accelerated frame. For example, let there be an observer inside a closed elevator on the
surface of the earth. If he observes the motion of the particle inside the elevator he can determine the
gravitational acceleration g. In fact, all his experiments would finally led him to the conclusion that
he is in the gravitational field of the earth. Let the the elevator now be in a spaceship accelerating
with an acceleration that is equal to g. If he repeats the experiments, he will reach the same identical
conclusions. So, he will not be able to distinguish between the gravitational field of the earth and an
accelerating spaceship. Therefore, the weak equivalence principle is also stated as: The motion of freely
particles are same in a gravitational field and a uniformly accelerated frame in sufficiently small enough
region of spacetime. We have added “sufficiently small enough region of spacetime” because gravitational
inhomogeneities in a large enough spacetime are detectable.
Above thought experiment shows that you can trade gravity with uniform acceleration. A more powerful
way to look at is this: you can turn o↵ or completely cancel gravity in a coordinate system that is falling
freely in a purely gravitational field. Such cancellation works only in a very small region of space. For
example, the earth is a freely falling reference frame towards the sun. Thats why we do not feel the gravity
of the sun – it is completely canceled out. But small inhomogeneities in the field is sufficient to cause tides
in the ocean. So, freely falling frames are locally equivalent to an inertial frame. Why a freely falling frame
in a gravitational field is equivalent to an inertial frame? Imagine laboratory floating in deep space and
free from gravity – this is inertial, meaning not acellerated. Consider the same laboratory falling freely
towards earth. If you are trapped inside, there is no experiment to distinguish between the two cases.
This fact, that locally inertial frame can be chosen in arbitrary gravitational field is stated by the strong
equivalence priciple: In an arbitrary gravitational field, it is possible to chose a locally inertial coordinate
system such that in a sufficiently small region of spacetime, the laws of nature have the same form as in
an unaccelerated Cartesian coordinate system in the absence of gravity. The usefulness of the principle is
that, even you are in a gravitational field, locally you can construct a inertial frame where special relativity
holds. In the above statement, the word Cartesian mean that in this coordinate, the Newton’s law is of the
type f ↵ = md2 x↵ /d⌧ 2 .
It is easy to show that freely falling frame cancels gravity. Consider a particle of mass m above the
earth moving under the influence of a force F~ . In an Cartesian coordinate system the equation of motion
of the particle is given by
d2 x
m 2 = m~g + F~ .
dt
Now, lets do a coordinate transformation x0 = x 12 gt2 which is a freely falling coordinate system. In
principle, one can think of transformation of time also, but for non-relativistic motions near the earth

23
surface, time is same for the particle and an observer who is at rest on the surface of the earth. If you
substitute this then you get that the equation of motion as

d2 x0
m = F~ ,
dt2
where gravity is completely removed. This example shows that both the original observer with coordinates
x and his freely falling friend with coordinate x0 agrees on the mechanics of particle. The only di↵erence is
that the x0 friend do not feel any gravity.
To summarize
• You can not di↵erentiate between uniform gravity and uniform acceleration
• You can not distinguish an inertial frame from a frame that is falling under the influence of gravity
only (also called freely falling under gravity)

4.2 Light and gravity: curvature of spacetime

An important outcome of the EP is that gravity curves the spacetime. This can be understood by a
thought experiment of how light responds to a gravitational field. Naively, you can expect that the light
being massless it is una↵ected by gravity. But consider this experiment where you are inside freely falling
accelerating elevator towards the earth, and your friend is standing at the surface of the earth. You shine
a laser from one of the side wall to the other, see figure 10. Both of you trace the path of the light ray.
Let us assume that the elevator is transparent so that both of you agree where light starts and where it
hits. Your observation: you are in an inertial frame, and according to the strong EP the laws of light will
be same as in an inertial frame in the absence of gravity. That means you observe that the light travels in
a straight line. Your friend: by the time the light ray travels from one wall to the other, the elevator has
moved a bit. So according to your friend the light is bend (remember that both agree where light hits the
wall) towards the earth. Here are more explanations.
There is an equivalent experiment where the elevator is accelerated upwards, see figure . If someone
shines a light from outside to the opposite wall, then to someone inside the light will appear to bend. But
according to EP this situation is equivalent to the elevator in a uniform gravitational field. So gravity
should also bend light.

Figure 10: Light bends in uniformly accelerated frame. According to the EP, light will also bend in a
gravitational filed.

The light bending is actually the consequence of the curvature of the spacetime. Recall that light follows
paths that correspond to the least distances – these are called geodesics. For example in a flat surface it
is a straight line, and on the surface of a sphere it is the arc. Therefore bending of light means a geodesic
that is not a straight line – hence it must be a curved spacetime.

24
4.3 Gravitational redshift
The aforementioned light-bending experiment is hard to perform. However, there is an e↵ect called grav-
itational redshift, yet another consequence of EP, that has been measured. Gravitational redshift refer to
the fact that, when light ray travels from a lower gravitational potential to a higher gravitational potential,
the frequencies are redshifted. Consider a light source E at the surface of the earth that is emitting light
waves at constant frequency, and there is a receiver R at height h. According to equivalence principle,
this setup equivalent to both the emitter E and the receiver R accelerated with constant acceleration g by
rockets with distance h apart.
At time t0 the trailing craft emits a photon of frequency ⌫E . The frequency reaches the receiver at time
t = h/c, where c is the speed of light. By the time the light ray travels from E to R, the receiver R has
acquired a velocity v = gt = gh/c. Hence from classical Doppler e↵ect the received frequency ⌫R is
v gh ⌫R ⌫E gh
⌫R = ⌫E (1 ) = ⌫E (1 ), = . (62)
v c2 ⌫E c2
This ⌫R < ⌫E meaning redshift of the received light.
Actually, the gravitational redshift is the cause of gravitational time dilation – two di↵erent clocks at
di↵erent height runs di↵erently. This e↵ect is accounted for in GPS system. You can read more here, or
here.

4.4 Free particle in gravity

Equivalence principle tells us that if we jump from a non-inertial frame to a freely falling one, we remove
gravity. In the following, we will do the reverse.
Consider a particle moving freely in nothing but gravitational forces. According to the equivalence
principle there is a freely falling coordinate system ⇠ ↵ where the motion is same as in an inertial frame,
i.e., the motion is a straight line in spacetime
d2 ⇠ ↵
= 0,
d⌧ 2
where d⌧ 2 = ⌘ d⇠ d⇠ is the proper time in the freely falling frame. Suppose we chose an arbitrary
Cartesian coordinate system xµ which can be anything (rest at earth, accelerated, rotating). In terms of
the arbitrary coordinate system, we can write the proper time as
@⇠ ↵ µ @⇠
d⌧ 2 = ⌘↵ dx dx⌫ = gµ⌫ dxµ dx⌫ ,
@xµ @x⌫
where we have a defined metric
@⇠ ↵ @⇠
gµ⌫ ⌘ ⌘↵ . (63)
@xµ @x⌫
Note that if the new coordinate is non-inertial, ie., accelerated, then the new metric gµ⌫ is di↵erent from
the Minkowski metric ⌘µ⌫ . We can now express the motion in the arbitrary coordinate system xµ as
✓ ◆
d @⇠ ↵ dxµ
= 0,
d⌧ @xµ d⌧
@⇠ ↵ d2 xµ @ 2 ⇠ ↵ dxµ dx⌫
or, + = 0
@xµ d⌧ 2 @xµ @x⌫ d⌧ d⌧
✓ ◆
dx @x @ 2 ⇠ ↵ dxµ dx⌫
or, 2
+ ↵ µ ⌫
= 0.
d⌧ @⇠ @x @x d⌧ d⌧
↵
@⇠ @x
To obtain the last line we have multiplied by @x /@⇠ ↵ and have made use of the product rule @x µ @⇠ ↵ = µ .

In the above equation, the object within the bracket is called the affine connection or Christo↵el symbol

@x @ 2 ⇠ ↵
µ⌫ ⌘ . (64)
@⇠ ↵ @xµ @x⌫

25
Therefore, the equation of motion looks as

d2 x dxµ dx⌫
+ µ⌫ =0 . (65)
d⌧ 2 d⌧ d⌧

In general, it is the equation of motion of an object that is accelerated. According to the EP, acceleration
is equivalent to gravity. Therefore, in some way, the affine connections must describe gravity. As we will
see later, affine connections have geometric origin and associated to curvature of space. This serves as a
hint that a theory of gravity is geometric theory and curvature of spacetime is associated to gravity.
For the motion of a massless particle, for example photon or neutrino, the derivation is same as above
with only exception that the proper time ⌧ is replaced by some arbitrary parameter that changes along
its worldline. Then, from the perspective of freely falling coordinate the equation of motion is

d2 ⇠ ↵
= 0,
d 2
When seen from the arbitrary coordinate system xµ , it becomes

d2 x dxµ dx⌫
+ µ⌫ =0 . (66)
d 2 d d

4.5 Relation between µ⌫ and gµ⌫

Note that in the definition of the affine connections (64), there is reference to the old coordinate system ⇠
to which the particle has no connection to. The affine connection can be expressed in terms of the metric
tensor gµ⌫ and its derivatives as
✓ ◆
g↵ @g⌫↵ @gµ↵ @g⌫µ
µ⌫ = + . (67)
2 @xµ @x⌫ @x↵

The right hand side is also called the Christo↵el symbol. Often, the affine connection and Christo↵el symbol
are interchangeably used. The µ⌫ is not a tensor.
Let see what we just did. There was a particle in motion under nothing but gravitation. The equivalence
principle helped us construct a locally inertial frame ⇠ ↵ in which the motion is that of a free particle in a
straight line. We then switched to an arbitrary coordinate system, and found that the particle is accelerated.
The acceleration is determined by the affine connection µ⌫ . Therefore, the affine connection µ⌫ , and hence
the metric gµ⌫ must determine the gravitation in the new system xµ .

4.6 Geodesic equation

The shortest path between two points in a flat surface is a straight line. This is easy to prove using functional
minimization. Consider two infinitesimally separated points on the plane (x, y) and (x + dx, y + dy). Then
the infinitesimal distance is p
dx2 + dy 2 .
The distance between any two points A and B is simply the integration of the above
Z p Z p
dy
dx2 + dy 2 = 1 + (y 0 )2 dx , y0 = .
dx

26
One can consider infinitely many paths connecting A and B. By keeping the end points fixed, if the
coordinates are changed by infinitesimal amount, then the shortest path is the one for which
Z p
0 = 1 + (y 0 )2 dx .
Z ✓ ◆
@ p
= dx 1 + (y 0 )2 y0
@y 0
Z ✓ ◆
@ p 0 )2
d
= dx 1 + (y y
@y 0 dx
Z  ✓ ◆ ✓ ◆
d @ p 0 )2 y
d @ p 0 )2
= dx 1 + (y 1 + (y y
dx @y 0 dx @y 0
Since the end points are fixed, the first terms vanishes, and only the second term inside the integration
will survive. As the integral is zero for any arbitrary change y, the quantity that multiplies it must also
vanish, i.e., ✓ ◆
d @ p 0 )2
1 + (y = 0,
dx @y 0
solving this equation leads to the equation of straight line

y = mx + c ,

where m, c are constants. One can proceed in the same way to find equation of shortest paths in any
geometry.
Let us apply the above machinery to find the equation of shortest path of the freely falling particle in the
xµ coordinate system. You can skip the derivation and go to last page for comments. Let the particle falls
1
from a point A to another point B. An infinitesimal proper distance is d⌧ = gµ⌫ dxµ dx⌫ ) 2 . Therefore,
the length of the proper distance from A to B is
Z 1
⌧= gµ⌫ dxµ dx⌫ 2 .

Shortest path is the one for which the variation on the paths xµ ! xµ + xµ , keeping the end points A, B
fixed, corresponds to
Z ✓ ◆ 12
0= ⌧ = gµ⌫ dxµ dx⌫ ,
Z ✓ ◆ 1 
1 2
= gµ⌫ dxµ dx⌫ (gµ⌫ )dxµ dx⌫ gµ⌫ (dxµ )dx⌫ gµ⌫ dxµ (dx⌫ ) .
2

We can write the variation (gµ⌫ ) in terms of partial derivative as

@(gµ⌫ ) ↵
(gµ⌫ ) = x .
@x↵
For the variation of the di↵erential, it is well know from the properties of functional derivative that
variation of di↵erential (dxµ ) can be written as di↵erential of variation d( xµ )1 . We will use these to write
✓ ◆ 12
µ ⌫
the next line. Another thing we note is that gµ⌫ dx dx is simply 1/d⌧ . We divide by one more
factor of d⌧ and multiply d⌧ . The two factors of 1/d⌧ is taken inside the square bracket to form d/d⌧ . So
with all these we get
Z 
1 @gµ⌫ ↵ dxµ dx⌫ d xµ dx⌫ dxµ d x⌫
0 = x + g µ⌫ + g µ⌫ d⌧ . (68)
2 @x↵ d⌧ d⌧ d⌧ d⌧ d⌧ d⌧
1 In fact, derivative of a variation is the same if variation of the derivative

27
The last two terms can be integrated by parts. For example
Z  Z  ✓ ◆
1 dxµ d( x⌫ ) 1 d dxµ ⌫ dgµ⌫ dxµ ⌫ d2 xµ ⌫
gµ⌫ = gµ⌫ x x gµ⌫ x .
2 d⌧ d⌧ 2 d⌧ d⌧ d⌧ d⌧ d⌧ 2

The first term being the surface term it vanishes, since the end points of the path are held fixed. Therefore
we can write
Z  Z 
1 dxµ d( x⌫ ) 1 dgµ⌫ dxµ ⌫ d2 xµ ⌫
gµ⌫ = x + gµ⌫ x .
2 d⌧ d⌧ 2 d⌧ d⌧ d⌧ 2
Z 
1 @gµ⌫ dx dxµ ⌫ d2 xµ ⌫
= x + gµ⌫ x x⌫ d⌧ ,
2 @x d⌧ d⌧ d⌧ 2

where in the second line we have used chain rule for the derivative of gµ⌫ . There is a similar result for the
second terms of (68). Now collecting all the terms, and rearranging the dummy indices we get
Z 
d2 xµ 1 @gµ⌫ @gµ @g⌫ dxµ dx⌫
0= gµ⌫ 2
+ ⌫
x⌫ d⌧ ,
d⌧ 2 @x @x @xµ d⌧ d⌧

As the variation x⌫ are arbitrary, the integrand must vanish identicallyi.e.,

d2 xµ 1 @gµ⌫ @gµ @g⌫ dxµ dx⌫

gµ⌫ 2
+ = 0.
d⌧ 2 @x @x⌫ @xµ d⌧ d⌧
Multiplying this equation by g ⇢⌫ , and rearrnging the indices we get

d2 x⌫ ⌫ dxµ dx
+ µ =0 (69)
d⌧ 2 d⌧ d⌧

where the ⌫µ are the affine connections defined in (67). In the xµ coordinate system the solution of the
above equation gives a path for which the proper distance (time) is shortest – these kinds of paths are
called geodesics. But this is the same equation that we have obtained in the previous subsection – for a
particle that is moving freely in purely gravitational field. So we can conclude that if a particle moves
freely under the influence of noting but gravitational field, they will follow the geodesics, or ’straight line’
in that geometry. You can see that the solutions of this equation will not give straight line. What it means
is that geodesics are curved – or the spacetime is curved.
From geometrical point of view ⌫µ are non-zero in a non-Eucledian space or a curved space. For
example the surface of a sphere ⌫µ are non-zero. Therefore the curvature of spacetime is associated with
the ⌫µ . One can therefore say that gravity manifests itself as the the curvature of the spacetime – free
particles only follow geodesics in those curved spacetime. This is the GTR way of looking at gravity. In
other words, you can describe gravity through the curvature of spacetime.
For massless particles like photon and neutroino, the geodesic equation can be expressed in terms of
parameter
d2 x⌫ dxµ dx
2
+ ⌫µ =0 (70)
d d d
This is called a null geodesic.

28