Prof.

David Hilditch
André Cordeiro (TA) Relatividade
Christoph Gärtlein (TA) Week 3 — Solutions 2023/2024

*: problems we will discuss in class

Solution 1: Michelson-Morley Experiment With a Real Wind

a) We know:

l
(u − v) · tA→B = l ⇔ tA→B = (1)
u−v

On the way back:

l
(u + v) · tB→A = l ⇔ tB→A = (2)
u+v

l l 2l
t1 = tA→B + tB→A = + = v2
(3)
u−v u−v


u 1− u2

Using t0 = 2l/u:

t0
t1 = (4)
1 − (v/u)2

1
Prof. David Hilditch
André Cordeiro (TA) Relatividade
Christoph Gärtlein (TA) Week 3 — Solutions 2023/2024

b) No, because ther would be still one trip with speed (u − v) and another with (u + v).
What changes is that now :

l l
tA→B = , tB→A = (5)
u+v u−v
But t1 = tA→B + tB→A is the same.
c) The plane still travels at speed u but needs to counter the wind speed so that it doesn’t
change its direction.

√
up = u2 − v 2 (6)

Therefore, a component of the airplane speed towards north is v to counter

√ the wind
speed. The component towards east (and west on the way back) is up = u2 − v 2 (see
Pythagoras). Thus:
l
tA→B = tB→A = √ (7)
u2 − v 2
2l t0
t2 = 2 · tA→B = √ =q (8)
2
u −v 2 2
1− v u2

d) No, the difference would be the speed component towards north, which now must
√ be to-
wards south. The speed component towards east/west would be the same up = u2 − v 2 .
Therefore:

t0
t2 = q (9)
v2
1− u2

2
Prof. David Hilditch
André Cordeiro (TA) Relatividade
Christoph Gärtlein (TA) Week 3 — Solutions 2023/2024

e) The essential difference is the existence of a medium, the so called "ether", and so the
speed is not invariant.
r
t2 v2
= 1 − 2 ̸= 1 (10)
t1 u

3
Prof. David Hilditch
André Cordeiro (TA) Relatividade
Christoph Gärtlein (TA) Week 3 — Solutions 2023/2024

Solution 2: A moving, slanting rod

a) Rest frame:

In the frame S:

∆x = γ(∆x′ + v∆t′ ) (11)

∆y = ∆y ′ (12)
∆t = γ(∆t′ + cv2 ∆x′ ) (13)

4
Prof. David Hilditch
André Cordeiro (TA) Relatividade
Christoph Gärtlein (TA) Week 3 — Solutions 2023/2024

The measurement in S is done with t constant. so ∆t = 0.

v
∆t′ = −2
∆x′ (14)
c
⇒ ∆y = ∆y ′ = L′ sin θ′ (15)
1 1
⇒ ∆x = γ(∆x′ + v∆t′ ) = ∆x′ = L′ cos θ′ (16)
γ γ

The total length of the rod measured in S is

q
L = L′2 sin2 θ′ + L′2 cos2 θ′ /γ 2 (17)
q
⇔ = L′2 sin2 θ′ + L′2 cos2 θ′ (1 − β 2 ) (18)
p
= L′2 − L′2 cos2 θ′ β 2 (19)
p
⇔ = L′ 1 − cos2 θ′ β 2 (20)

b)

∆y L′ sin θ′ γ
tan θ = = ′ ′
= γ tan θ′ (21)
∆x L cos θ
⇔ θ = arctan(γ tan θ′ ) (22)

c) Use Mathematica e.g.: L ≈ 0.938, θ ≈ 32.2o , and γ ≈ 1.091.

5
Prof. David Hilditch
André Cordeiro (TA) Relatividade
Christoph Gärtlein (TA) Week 3 — Solutions 2023/2024

Solution 3: Lorentz Transformations in 3D

We have:

⃗x = x∥ ⃗n + x⊥ ⃗e⊥ (23)
⃗v = v⃗n , ⃗n · ⃗n = 1 (24)
⃗n · ⃗e⊥ = 0 (25)

a) Recalling the Lorentz transformations

′
 v 
t = γ t − 2 x , ⃗v = v⃗eR (26)
c
x′ = γ(x − vt) (27)
y′ = y (28)
z′ = z (29)

Therefore, the transformation occurs in the direction of the velocities, so x → x∥ and

y → x⊥ .
The Lorentz transformations for (ct, x∥ , x⊥ ) are:

 v 
t′ = γ t − 2 x∥ (30)
c
x′∥ = γ(x∥ − vt) (31)
x′⊥ = x⊥ (32)

⃗x · ⃗v
b) Now we can use that x∥ = , so
v

 
′ v ⃗x · ⃗v
t = γ t− 2 (33)
c v
 
⃗x · ⃗v
⇔ t′ = γ t− 2 (34)
c

c)

x⃗′ = x′∥ ⃗n + x′⊥ ⃗e⊥ (35)

 
= γ(x∥ − vt) ⃗n + x⊥⃗e⊥ (36)

6
Prof. David Hilditch
André Cordeiro (TA) Relatividade
Christoph Gärtlein (TA) Week 3 — Solutions 2023/2024

We can use that x⊥⃗e⊥ = ⃗x − x∥ ⃗n = ⃗x − (⃗n · ⃗x)⃗n

So

x⃗′ = γ(⃗n · ⃗x − vt) ⃗n + ⃗x − (⃗n · ⃗x)⃗n

 
(37)
 
= (γ − 1)(⃗n · ⃗x − γvt) ⃗n + ⃗x (38)
 
= ⃗x + (γ − 1)(⃗n · ⃗x − γvt) ⃗n (39)

7
Prof. David Hilditch
André Cordeiro (TA) Relatividade
Christoph Gärtlein (TA) Week 3 — Solutions 2023/2024

Solution 4: Relativistic Velocity Addition in 3D

First, decompose any vector into components perpendicular and parallel to the velocity,

⃗v
⃗x = x∥ ⃗e∥ + ⃗x⊥ , where ⃗e∥ = , and ⃗x⊥ · ⃗e∥ = 0 . (40)
|⃗v |
Note that although this does not uniquely define ⃗x⊥ , since any rotation about ⃗v leaves the
last condition fulfilled, this does not affect this proof. This is because boosts do not affect
the perpendicular components of position vectors.
Following the hint, consider the interval between two different space-time coordinates of the
object (think of this as two events in its space-time trajectory). Label this interval (∆t, ∆⃗x)
in S and (∆t′ , ∆⃗x′ ) in S ′ . These are related by the Lorentz boost


′
 v 
∆t
 = γ ∆t − 2 ∆x∥ ,

 c 
∆x′∥ = γ −v ∆t + ∆x∥ , (41)


 ′
∆⃗x⊥ = ∆⃗x⊥ .
Expressing the object’s velocities as distance covered per unit of time, and performing the
same decomposition into perpendicular and parallel components,

∆⃗x ∆x∥ ∆⃗x⊥

⃗u = =⇒ u∥ = and ⃗u⊥ = , (42)
∆t ∆t ∆t
with a similar result for S ′ .
Replacing this into the Lorentz boost,

  
′
∆t

 = γ 1 − cv2 u∥ ∆t ,
 
∆x∥ ′
= γ −v + u∥ ∆t , (43)


 ′
∆⃗x⊥ = ∆⃗x⊥ .
Finally, dividing the last two equations by the first one,

′
u∥ − v
 u = v ,
 ∥


1 − 2 u∥

c (44)
′ ⃗u⊥
 ⃗
u = v ,
 ⊥

  
γ 1 − 2 u∥

c
2 2 −1/2
where γ = γ(v) = (1 − v /c ) , as we expected. Notice how the perpendicular components
are affected differently by the boost.

8
Prof. David Hilditch
André Cordeiro (TA) Relatividade
Christoph Gärtlein (TA) Week 3 — Solutions 2023/2024

Solution 5: In the Rear-view Mirror

The velocities of the incident and reflected light rays in the laboratory frame are given by
⃗ui = (c cos θi , c sin θi ) , (45)
⃗ur = (−c cos θr , c sin θr ) , (46)
with similar expressions for the light rays in the mirror frame.

a) The transformation law for the longitudinal component of the velocities yields the follow-
ing transformation for cos θi ,
cos θi − (−v/c) cos θi + β
cos θi′ = = . (47)
1 − (−v/c) cos θi 1 + β cos θi
The incident and reflected angles are the same in the frame of the mirror, therefore
cos θi + β
cos θr′ = . (48)
1 + β cos θi
This deviates from the classical result cos θr′ = cos θi , yielding a larger value for the cosine
of the reflection angle. Since cos x is a decreasing function, this means that reflected
angles appear smaller than in the classical case.
However, images appear with the same width as in the classical case! Although the angles
are smaller due to relativistic effects, there is still one ray of light emitted along the line of
sight of the observer in the frame of the mirror (although this light ray would be emitted
at a much wider angle in the frame of the source)1 .
b) Applying the transformation law to the reflected light ray we find
− cos θr + β cos θr − β
− cos θr′ = ⇐⇒ cos θr′ = . (49)
1 + β(− cos θr ) 1 − β cos θr
Inverting this requires flipping a sign,
cos θr′ + β
cos θr = . (50)
1 + β cos θr′
Putting these results together yields the intended result
(1 + β 2 ) cos θi + 2β
cos θr = . (51)
1 + β 2 + 2β cos θi
Plotting this with the help of some software like Mathematica or the online graphing
calculator Desmos reveals that the cosine of the reflected angle becomes even larger. For
the reasons stated above, this still does not impact the size of the image.
1
This is a good example of what can go wrong when one mixes relativity with geometric optics, to the
point that a previous version of these solutions claimed the opposite effect!

9
Prof. David Hilditch
André Cordeiro (TA) Relatividade
Christoph Gärtlein (TA) Week 3 — Solutions 2023/2024

Solution 6: Fitting a log in a barn

a) Drawing the reference frames in the same spacetime diagram we get:

b) In the frame S with (ct, x), the log P Q fits completely inside the barn F R at t = 0.
In the frame S ′ with (ct′ , x′ ), the log is always too long to fit inside the barn at any time
t′ = const.

c) As we learned in the lecture, the principle of simultaneous events is observer dependent.

Therefore, these two results are not in contradiction with each other.

10
Prof. David Hilditch
André Cordeiro (TA) Relatividade
Christoph Gärtlein (TA) Week 3 — Solutions 2023/2024

Solution 7: Time Ordering and Causality

a) Consider the Lorentz boosts from S to S ′ and vice-versa,

c∆t′ = γ c∆t − v ∆x c∆t = γ c∆t′ + v ∆x′

     
c and c . (52)
∆x′ = γ − v c∆t + ∆x ∆x = γ v c∆t′ + ∆x′
   
c c

We will take the case where the time interval is positive in both reference frames. This
implies
c∆t′ v
c∆t > 0 ⇐⇒ ′
>− , (53)
∆x c
c∆t v
c∆t′ > 0 ⇐⇒ >+ . (54)
∆x c

Imposing that this condition must true for all relative velocities v/c ∈ [−1, 1], we find
c∆t′
> +1 , (55)
∆x′
c∆t
> +1 , (56)
∆x
accounting for the different possible signs of ∆x. This therefore implies

(c∆t′ )2 > (∆x′ )2 , and (c∆t)2 > (∆x)2 . (57)

This condition is equivalent to requiring that the velocity of an object travelling between
the two events must be smaller than the speed of light c (in both reference frames, of
course) for their time ordering to be preserved across all reference frames. This region of
spacetime is known as the light cone.
Because time ordering between events is required to establish a causal relation, all physical
theories that comply with Special Relativity must only predict correlations between events
within the light cone. This is the source of the common rule of thumb “information cannot
propagate faster than light”.
b) Showing that the conditions are equivalent is a straightforward exercise,

(c∆t′ )2 − (∆x′ )2 = γ 2 (c∆t − β ∆x)2 − γ 2 (−β c∆t + ∆x)2 = (58)

= γ 2 1 − β 2 (c∆t)2 − γ 2 1 − β 2 (∆x)2 =



(59)
= (c∆t)2 − (∆x)2 , (60)

with the cross terms cancelling. This quantity is related to the invariant space-time
interval, and when it is positive both conditions in (57) are verified.

11
Prof. David Hilditch
André Cordeiro (TA) Relatividade
Christoph Gärtlein (TA) Week 3 — Solutions 2023/2024

c) The laser spot moves with a linear velocity of vspot = 2 × 3.84 × 108 m/s ≃ 2.6 c. Although
this is above the speed of light, causality is not broken since no two points in the lunar
surface can affect each other using this supra-luminal laser spot — if for example two
astronauts tried to use this scheme to communicate “faster than light”, they would need
to coordinate with whoever controls the laser, and that communication would be limited
by the speed of light.

12