Note 2
Note 2
Equivalence principle
General relativity is based on Einstein’s equivalence principle. The principle postulates local
equivalence of inertial and gravitational forces. Because of that principle general relativity appears
as a geometric theory: massive bodies distort the geometry of space-time in their vicinity which
influences the motion of other bodies. This is different to the electromagnetic forces where charged
bodies do not alter the space-time geometry but rather create the electromagnetic field around
them which then affects the motion of other charged bodies.
Inertial forces
Inertial forces are apparent forces that act on all bodies in non-inertial frames of reference such
as a rotating frame (centrifugal and Coriolis forces) or a uniformly accelerating frame (elevator
force). Inertial forces do not arise from any physical interaction but rather from the non-linear
motion of the frame itself.
In an inertial frame a free body moves without acceleration, and its (non-relativistic) equation
of motion in Cartesian coordinates is
~r¨ = 0, (1)
.
where ~r = {x, y, z} are the three spatial coordinates of the body and the dot above a letter denotes
time derivative. This is an equation for a linear motion: in inertial frames free bodies move without
acceleration along straight lines.
In a non-inertial uniformly accelerating frame that moves with acceleration ~a relative to an
inertial frame the corresponding equation of motion for the free body is
1
note2 [September 8, 2020]
The masses in the left- and right-hand sides of this equation are often referred to separately
as inertial and gravitational masses.
Many experiments attempted to observe the difference between gravitational and inertial
masses, including Galileo’s measurements of acceleration of balls of different composition rolling
down inclined planes, and Newton’s measurements of the period of pendulums with different
mass but identical length. All experiments reported no observed difference, with the most precise
experiment to day (arXiv:0712.0607) having the accuracy of one part in 1013 .
There are theories of gravitation where the equality of gravitational and inertial masses is just
a coinsidence but in general relativity this is a fundamental postulate.
The principle is customarily illustrated by two spacecrafts, one on Earth, the other accelerating
with the Earth’s gravitational acceleration g in the outer space. The observers in these spacecrafts
can not determine by doing local experiments inside spacecrafts whether their craft is accelerating
or at rest in a gravitational field.
However, unlike inertial forces gravitational forces vanish at large distances from the sources
of gravitation and therefore non-local experiments can well distinguish between inertial and grav-
itational forces.
mv 2
Z
2
S= − mc − mφ dt . (7)
2
This can be viewed—just as we did for a free body—as the non-relativistic limit of
Z s Z
2φ 2 2 2
S = −mc 1 + 2 c dt − dr = −mc ds . (8)
c
Therefore one can conclude that in a (weak) gravitational potential φ the metric of the space
becomes (approximately)
2φ
ds2 = 1 + 2 c2 dt2 − dr2 . (9)
c
This is clearly a metric of a non-Euclidean space as, apparently, time runs differently at different
depths in a gravitational potential.
1 Locally means within a limited space, where variations of the fields can be neglected.
2
note2 [September 8, 2020]
However, if there exist a coordinate transformation that globally turns the metric into (pseudo)-
Euclidean, the space is still called flat. In our example such transformation is
x = r cos θ ,
(11)
y = r sin θ .
If such transformation does not exist, the space is called curved or Riemann space. The
geometry of a curved space is called non-Euclidean geometry or Riemann geometry.
Exercises
1. A pendulum is suspended from the roof of a car moving in a line (straight curve) with
constant acceleration a. Find the angle the pendulum makes with the vertical. Explain
what is happening from the viewpoint of an inertial observer outside of the car and a non-
inertial observer in the car.
2. A bucket of water slides freely under gravity down a slope of a fixed angle α to the horizontal.
What is the angle of inclination of the surface of water relative to the base of the bucket?
3. Newtonian gravitation
2 The accepted coordinates must be differentiable (to allow development of differential geometry) and the
3
note2 [September 8, 2020]
(a) Argue that the equation of motion of a test body with coordinate ~r in Newton’s theory
of gravitation,
X GN Mk ~r − ~rk
~r¨ = − , (12)
|~r − ~rk |2 |~r − ~rk |
k
where Mk and ~rk are the masses and coordinates of gravitating bodies, can be written
as
~r¨ = −∇φ
~ , (13)
where φ is the gravitational potential,
X G N Mk
φ(~r) = − . (14)
|~r − ~rk |
k
(c) Argue that the equation of motion (12) and the equation for the gravitational poten-
tial (15) are invariant under Galilean transformation.
4. Gravitation as geometry of space
(a) Argue that the non-relativistic motion of a test body with mass m in a Newtonian
gravitational potential φ can be described by the Lagrangian
1
L = mv 2 − mφ − mc2 , (17)
2
where v is the velocity of the body.
(b) Argue that this Lagrangian can be obtained Ras the weak-field non-relativistic limit
(v c and φ c2 ) from the action S = −mc ds where the infinitesimal interval ds
is given as
2φ 2 2
2
ds = 1 + 2 c dt − d~r2 . (18)
c
(c) Argue that this metric describes a curved (non-Euclidean) space and that the proper
time runs differently at different levels in the gravitational potential.
5. Explain the “twins paradox” of special relativity from the viewpoints of i) the inertial twin
and ii) the non-inertial (travelling) twin.
6. What is the path of a free particle in a uniform gravitational field?
7. Argue that equivalence principle implies bending of light in gravitational fields.
8. Argue that equivalence principle implies gravitational redshift.
9. Construct an experiment to find out whether one is in an inertial frame.
10. Are the following coordinate transformations (x, y) → (x0 , y 0 ) allowed in general relativity?
(a) x0 = x, y 0 = 1
p
(b) x0 = x2 + y 2 , y 0 = arctan(y/x)
p
(c) x0 = 1/ x2 + y 2 , y 0 = arctan(y/x)
(d) x0 = ln(x), y 0 = y
3 Hint: ∇2 r1 = −4πδ(~
r).