Beruflich Dokumente
Kultur Dokumente
moving objects
Dalimil Mazac
February 5, 2010
Abstract
We review a mathematical construction providing an interesting view-
point for understanding the change of shape of moving objects. In general,
Lorentz contraction of relativistically moving objects is not directly ob-
served since light rays coming from dierent parts of given object which
arrive at observing point at given instant come from dierent times. To
account for this combined eect, the use is made of the correspondence
between the restricted Lorentz group and the Mobius group, where the
connection between celestial sphere and complex plane is given by the
stereographic projection. As an application, it is shown this implies that
a moving sphere will keep its apparent shape independently of velocity.
1 Introduction
It is a well-known consequence of special relativity that a moving body is con-
tracted by the factor
1
= 1 v 2 /c2
in the direction of its motion. Taking the sphere as an example, we can see
that from the point of view of an inertial observer, with respect to whom the
sphere moves uniformly at velocity v, it will be attened to an axially symmetric
ellipsoid with the ratio of semi-axes precisely .
Is the shape of a moving body, as seen by an observer at given instant, going
to be given solely by the length contraction? The answer is no since, thanks to
the nite value of speed of light, light rays coming from dierent parts of the
body which enter the observers eye at the same time leave the body at dierent
times and hence reect its position at distinct instants. As an illustration, let
us imagine we are standing nearby a straight bit of a railway line, along which
a train is coming at a speed comparable to that of light (possibly imaginable in
far future in Japan). As its rear end is further than the front, a light ray takes
more time to travel from the rear than from the front, and if we require that
both the light rays considered arrive at our eye at the same time, the rear end
light ray had to be emitted earlier. As a result, we will observe an apparent
extension of the train (besides other nontrivial eects).
Thus, the resulting transformation of an objects shape is a combination of
length contraction and the mentioned optical eect. The following shows that
this transformation is always given by a Mobius transformation of C projected
1
onto the celestial sphere through the stereographic map, and hence that the two
eects exactly cancel in the case of a moving sphere.
2 Initial insights
The question we aim to answer is: Given an object and an observer moving
uniformly wit respect to it, what is the shape of the object from observers
perspective at given instant?
Let us analyze the situation in reference frame S, which is the rest frame of
the observed object, with origin generally not coinciding with the object itself
(coordinates in S are unprimed). The observer moves at a uniform velocity
in S, and his reference frame will be denoted S (with primed coordinates).
Let us choose a particular time of observation, that is a particular observers
proper time, and take the point in spacetime, where the relevant rays enter the
observers eye, as the common origin of S and S .1 So the coordinates in S and
S are related by a homogeneous Lorentz transformation.
The events seen by the observer at t = 0 are precisely the points in past
spacetime with vanishing Minkowski norm, that is (t, x, y, z) such thatc = 1 in
this article.
x x = t2 x2 y 2 z 2 = 0.
Note that the 4-vectors of events that happened in the same direction from the
origin are multiples of each other.
It is important to realize that properties of events x , y
the ray of light emitted at x arrives at the space origin at time t = 0
x and y happened in the same space direction from the space origin
are independent of the choice of an inertial reference frame. The rst is the result
of invariance of the interval under Lorentz transformations and the second holds
by their linearity. Let us then call the set of events x for which x x = 0 and
such that all of them happened in the same space direction from the origin
a given ray.2 The transition from S to S then amounts to a transformation
between their rays. This is well-dened precisely by the above properties of
events on a light cone since two events of the same ray in S will map to the
same ray in S under Lorentz transformation.
Our task is to nd the right transformation between space directions of rays
when going from S to S , for which we exploit a remarkable correspondence (a
near isomorphism) of groups SO(1, 3) and SL(2, C), and hence also the group
of Mobius transformations.3 In the following, we shall see that if we identify the
1 Thechoice of S and S thereby depends on the selected time of observation.
mathematical terms, we impose the equivalence relation on the past light cone such
2 In
2
space direction of each ray with a point on the unit sphere S 2 , and perform the
stereographic projection of this onto C , the Lorentz transformation between
the rays of S and S translates as a Mobius transformation of C .
Further we note that the stereographic projection maps spherical circles on
S 2 to circles and lines in the complex plane (bijectively), and nally, that every
Mobius transformation sends circles and lines in C to circles and lines again.
Putting all the pieces together, we will be able to conclude that every object
with circular bounding shape4 in S, such as sphere, will be observed to have the
same shape in the S frame. The mathematical derivation of given arguments
is given in the following parts.
3
we can get the square of the Minkowski norm as the determinant
det(X) = t2 x2 y 2 z 2 .
we can see it preserves determinant of X and hence the norm of the 4-vector that
X represents. Thus every matrix A SL(2, C) represents a Lorentz transforma-
tion of spacetime. Moreover, we can verify that the resulting Lorentz transfor-
mation always preserves the positive direction of time and its determinant is 1.
What we got is a homomorphism : SL(2, C) SO+ (1, 3), which is also sur-
jective (this is harder to see). The kernel of this homomorphism is just {1, 1},
and consequently SO+ (1, 3) is isomorphic to SL(2, C)/{1, 1} = P SL(2, C).
The reasoning behind the proof of isomorphism between P SL(2, C) and the
Mobius group is notably clearer. Every Mobius transformation of the extended
complex plane
az + b
z 7 , ac bd = 0 (3)
cz + d
can be obtained from the matrix
( )
a b
, ac bd = 1.
c d
4
Figure 1: Stereographic projection
the sphere (hence it is going to touch it precisely at C), the circular shape of its
intersection with our elliptic cone will be preserved. Some angle hunting then
tells us that if this plane makes the angle with the cone axis, the complex
plane will make the angle = with it. It follows that the elliptic cone
intersects the complex plane in a circle too, in an extreme case in a circle with
an innite radius - a line.
To nish this section, let us show that Mobius transformations take circles
7 the ordinary cone
5
Figure 2: Circles under the stereographic projection
6
5 Construction of the argument
Every ray, as dened above, corresponds to a point that an observer at x =
x = 0 sees in the sky, and so can be specied by two real numbers u, v. Let
us represent a ray by the S-coordinates of a particular event lying on it.8
u2 v 2 1
2u
x =
.
(4)
2v
u +v 1
2 2
This choice will be justied later, meanwhile we can at least verify that x x =
0. Now, let us use the representation of 4-vectors by 2 2 hermitian matrices,
as we did in the proof of isomorphism between SO+ (1, 3) a P SL(2, C). For the
sake of the argument, the signs of t and y were changed, which does not present
any further problems
t ( )
x t + z x + iy
x = y
7 X = .
x iy t z
z
And so
u2 v 2 1 ( )
2u zz z
x =
7 X=2 , (5)
2v z 1
u2 + v 2 1
where z = u + iv (z is not the z-coordinate of space) with star denoting complex
conjugation. The equivalence relation which dissects the past light cone into the
rays, is thus translated as an equivalence relation of 2 2 hermitian matrices
with zero determinant, where A is equivalent to B, if A is a nonzero multiple of
B. The set of these equivalence classes is denoted S, with little ambiguity with
the S reference frame.
Any equivalence class from S can be represented by its member matrix
( ) ( )
zz z 0 0
, or ,
z 1 0 0
7
point, as if, instead of g, we used its image (g) M and act by it on B(x).
This action isomorphism can be written as10
B(gx) = (g)B(x),
which is equivalent to saying that the right hand half of the following diagram
commutes
/ P SL(2, C) /M
SO+ (1, 3) (6)
rays /S B / C {}
2
stereographic projection
with the original 1 in the z-coordinate disappearing thanks to the shift of the
spheres centre. It remains to show that the point on the sphere which lies in the
same direction as (7), will be mapped to u + iv by the stereographic projection,
perhaps multiplied by a constant factor.
Rewrite the position vector (7) in cylindrical coordinates, and express the
image complex number in the polar form
(, z, ) = (2 u2 + v 2 , u2 + v 2 , arctan(v/u)) 7 rei
matrices.
11 This nally explains the motivation behind the choice (4).
8
north pole, A the point of intersection of the ray SX and our circle, and nally
X = (2u, 0).
Pythagoras theorem tells us that |SX| = u2 + 1, and so |AX| = u2 . Points
AX X form an isosceles triangle, which is also congruent to SAO. It follows
that the stereographic projection maps A precisely to X , yielding the nal map
of the ray to the complex plane in the form
u2 v 2 1
2u
7 2(u + iv)
2v
u +v 1
2 2
6 Summary
Diagram (6) provides the connection between the mathematical construction
and the physical situation analyzed. It tells us: If you want to nd out where
a given point in sky seen by an observer at the origin of S will be shifted if we
go to the origin of a dierent frame S , which relates to S through a Lorentz
transformation, you only need to take the right Mobius transformation and act
with it on the point in C obtained from the sky point by the stereographic
projection, and nally map this back onto the sky. Lorentz transformations of
the sky are thus precisely Mobius transformations projected onto the celestial
sphere, and their analysis reduces to the investigation of the action of Mobius
group.
The most striking property of the Mobius transformations is the invariance
of circles and lines, which holds even after the back-projection on the sphere. A
sphere stationary in S will clearly have an observed circular boundary in this
frame, and this is unchanged when we transform to S , showing a moving sphere
always still looks spherical. Moving objects in general do not appear length-
contracted but rather rotated12 , an eect also called the Terrell rotation, bearing
the name of a physicist who rst derived this eect in 1959, simultaneously with
and independently of Roger Penrose. See the original articles [1] and [2] for
reference.
Finally, we mention that the mathematical correspondence expressed in the
diagram (6) has proved to be very fruitful in more advanced areas of theoretical
physics, especially giving rise to Penroses twistor theory, which is an approach
to quantum gravity.
References
[1] James Terrell, Invisibility of the Lorentz Contraction,
Phys. Rev. 116 1041-1045 (1959).
[2] Roger Penrose, The Apparent Shape of a Relativistically Moving Sphere,
Proc. Cambridge Phil. Soc. 55 137-139 (1959).
12 The angle of apparent rotation is given by sin = v/c at the instant when an object
moving at velocity v is passing perpendicularly nearby the observer. The equivalence of this
to the formula for the aberration of light from distant stars is not coincidental.