Lorentz tensors

Gerardo Urrutia S anchez

Facultad de Ciencias, Universidad Nacional

Autonoma de Mexico, Mexico D.F. 04510, Mexico
Abstract
In this note I intend exhibit a general idea of Lorentz 4-tensors. First dening the dierentials and
gradient in the minkowsky space which later used for construct the 4-vectors laws of transformation.
Step by step development the idea of tensor, this by increasing the rank of the Lorentz 4-vectors
to reach a generalization of Lorentz 4-tensors.
1
I. INTRODUCTION
In this note reviews the elementary mathematics of Lorentz tensors with development
algebraic simple. The principal reason of this study is generally understand the vectors idea
in physics. The physical laws are constructed with this objets because they do not depend of
the coordinates i.e. the laws of physics are the same in all reference inertial systems. Is this
hard to imagine a physical problem which does not involve some sort of continuous space.
It might be physical three-dimentional space, four-dimentional spacetime, phase space for a
proble in classical quantum mechanics, the space of all termodynamic equilibrium states, or
some still more abstract space. All these spaces have dierent geometrical properties, but
they all share something in common, something which has to do with their being continuous
spaces rather than, say, lattices of discrete points. We will have a rapprochement with the
4-tensor whose foundation is in the dierential forms in a continuous space.
II. DIFFERENTIAL AND GRADIENT IN THE MINKOWSKY SPACE
The usual conguration in Special Relativity with Lorentz transformations is
x

=(x vt), (1)

t

=(t vx/c
2
), (2)
y

=y, (3)
z

=z, (4)
with 1/
_
1v
2
/c
2
. For express the Ecs. (1)-(4) in the compact form, you may change
the label in the coordinates: x
0
=ct, x
1
=x

, x
2
=y

, x
3
=z

, x
0
=ct, x
1
=x, x
2
=y, x
3
=z.
Then rewrite the Ecs. (1)-(4)
x
0
= x
0
x
1
+ 0 + 0, (5)
x
1
=x
0
+ x
1
+ 0 + 0, (6)
x
2
= 0 + 0 + x
2
+ 0, (7)
x
3
= 0 + 0 + 0 + x
3
, (8)
2
Im introduced the factor with value v/c. The primed coordinates are lineal and
homogeneous functions of the not-primed coordinates. This observation is very important
for subsequent denitios of 4-vectors and tensors.
The matrix form of the Ecs. (5)-(8)
_
_
_
_
_
_
_
x
0
x
1
x
2
x
3
_
_
_
_
_
_
_
=
_
_
_
_
_
_
_
0 0
0 0
0 0 1 0
0 0 0 1
_
_
_
_
_
_
_
_
_
_
_
_
_
_
x
0
x
1
x
2
x
3
_
_
_
_
_
_
_
. (9)
With compact notation rewrite the Ecs. (5)-(8)
x

=
3

=0

, (10)
with x

=(x
0
, x
1
, x
2
, x
3
); x

=(x
0
, x
1
, x
2
, x
3
) and

=
_
_
_
_
_
_
_
0 0
0 0
0 0 1 0
0 0 0 1
_
_
_
_
_
_
_
, (11)
(11) is the lineal transformation (Lorentz) represented by a matrix. The index represents
the components in rows and the index represents the components in the columns. In
Special Relativity the greek indices take values , , , ... = 0, 1, 2, 3 and the latin indices
take values i, j, k, l... = 1, 2, 3. Can also be written 1+3 form x

=(x
0
, x
i
).
The Eq. (10) with the Einstein notation is
x

. (12)
The Einstein convention over repeated indices dene; when a index twice repeats then
understood a sum, that is to say:
P

=
3

=0
P

= P
0
Q
0
+ P
1
Q
1
+ P
2
Q
2
+ P
3
Q
3
(13)
The index is named dumb because it can be replaced by another i.e. P

= P

=
P

. Is important to emphasize that a index should not appear more than twice. For
3
example the expression A

not be written as A

. This expression can be interpreted

in many ways, turns out to be an ambiguity. The free index is one that does not repeat two
times. In the expression A

the index is free y and is dumb. For any equation the free
indices on both sides are the same and must retain their sense covariant and contravariant.
For example,

= J

its a correct expression, but the equations G

= T

and

= g

are not corect.

In Relativity theory the index notation is as shown in the following expression (put

in
the index but not in the tensor functions)
x

. (14)
With this we can use several times the same greek letter to denote dierent indices. For
example, the Ec. (14) is written by
x

. (15)
Now, The coordinates x
0
, x
1
, x
2
y x
3
are independent, then
x

=
_
_
_
1 si =
0 si =
(16)
This expression is similar to the Kronecker delta , dened as

=
_
_
_
1 si =
0 si =
(17)
therefore
x

, (18)
Developing the sums explicitly, these identities are correct:

and

= 4. The
Kronecker delta serves to convert a free index on a dummy index. For example, T

the
index is free. Applying T

in

then is converted in a dumb index:

T

= T

. (19)
When deriving the Ec. (12) with respect x

we obtain
x

(20)
4
We used the Ecs. (18) and (19). The identity
x

(21)
is characteristic for the Lorentz transformations. The index in the matrix

corresponds
to coordinate x

and the index to coordinate x

.
We have the ingredients to dene the transformation rules of two 4-vectores:
The 4-dierential and 4-gradient. Taking the dierential of the Ec. (12) and remembering
that

is a matrix with constant elements, gives a transformation law for the 4-dierential
dx

dx

(22)
The 4-dierential have a freee super index or contravariant. The 4-diferential is a function
of the coordinates in both reference frames: dx

= f

(x) and dx

= g

(x

). The Ec. (22)

shows that the 4-diferential dx

in the primate system is lineal and homogeneous function

of the 4-dierential dx

in the not primate system.

Now consider a Lorentz 4-scalar (x) with transformation rule

(x

) = (x). (23)
For all values of x

and x the 4-scalar is invariant. Note the absence of the indices in the
4-scalar. Also note that for simplicity hemos escrito (x

) = (x

) y (x

) = (x) to specify the

arguments of the 4-scalar in both systems. In general, adopt this convention in the 4-tensors
study. Taking the partial derivative with respect x

of the Ec. (23) and applying the chain

rule for obtain the expression

=
x

(24)
Of the Ec. (21) we know that x

/x

. But does it mean x

/x

? Multiply the
Ecs. (22) y (24)
dx

=
_

_
dx

(25)
This product of the 4-dierential and the 4-gradient lack of indices, then could be a Lorentz
4-scalar. For the Eq. (25) be a law of transformation of a Lorentz 4-scalar we needed that
the quantity within the square brackets meets

. (26)
5
Inserting the Ec. (26) in (25) veried the scalar character
dx

dx

= dx

(27)
The object x

/x

in the Ec. (26) is the inverse matrix of

. To denote x

/x

with

then the Ec. (26) take a form

. (28)
Other frequently notations for the inverse Lorentz transformations is: (
1
)

or (

)
1
.
Therefore the Ec. (28) is also expressed as

)
1
=

and

(
1
)

. notice that,
the Ec. (28) with primated indices can be written as

. (29)
The Ec. (24) can be written as

(30)
As

= then we obtain the transformation law of the 4-gradient operator.

. (31)
Note; the 4-gradient have a fre index in covariant form. The 4-gradient operator may
be considered a function on both sides in the two reference frames: /x

= g

(x

) and
/x

= f

(x). Then the Ec. (31) is interpreted as /x

in the primate system. Is a lineal

and homogeneous function of the 4-gradient /x

in the not primated system.

The inverse of the 4-dierential and the 4-gradient we obtain by multiplying the Ecs. (22)
y (31). Multiply the Ec. (22) by

and we obtain the inverse transformation of the 4-

dierential.

dx

dx

dx

= dx

. (32)
In other words, the inverse of dx

dx

is
dx

dx

. (33)
In analogy, multiplying the Ec. (31) by

=

x

, (34)
6
we obtain the inverse transformation of 4-gradient.

. (35)
Now we introduce a new notation for the 4-gradient

(36)
With this notation, the Ec. (35) is more elegant.

. (37)
III. LORENTZ 4-VECTORS
The 4-dierential dx

is a example of covariant Lorentz 4-vector. In general we speak;

that a set of 4 components T

(x

) dened in primated system; form the components of a

covariant 4-vector of Lorentz. If this components is related lineal and homogeneous with
respect of the components T

(x) dened in the noprimated inercial system, with the next

law of transformation
T

(x

)=

(x). (38)
A covariant Lorantz 4-vector also is appointed as covariant Lorentz 4-tensor of rank 1. The
rank (order) of 4-tensor is equal a number of free indices. The inverse transformation of
Ec. (38) is
T

(x)=

(x

), (39)
obtained by a usual procedure

= T

. (40)
The 4-gradient is example of a covariant 4-vector. In general a set of 4 components T

(x)
dened in the noprimated inertial reference system, form the components of a covariant
Lorentz 4-vector, if this coordinates is related lineal and homogeneous with respect of the
components T

(x) dened in a not-primate inertial frame system in the next transformation

law
T

(x

)=

(x) (41)
7
The inverse transformation of 4-vector obeys the next law of transformation
T

(x)=

(x

), (42)
and we obtain with the same procedure that became the Ec. (38) to Ec. (39). The product
T

is a Lorentz 4-escalar.
T

= T

. (43)
The associated transformations at contravariant Lorentz 4-vectors form a group. we put
in the indices to change the notation. For example; considered the Lorentz transformation
x

between double-quote coordinates and the quote coordinates. T

(x

) and
T

(x

) are the components of 4-vector in the double-quote and quote frames. Therefore
T

(x

) =

(x

). (44)
Now consider the Lorentz transformation x

between quote and not-quote coordi-

nates. T

(x

) y T

(x) are the components of 4-vector in the quote and not-quote reference
frames. Therefore
T

(x

) =

(x). (45)
replacing the Ec. (45) in Ec. (44) obtain the law of transformation of contravariant Lorentz
4-vector associated at the double-quote and not-quoted systems.
T

(x

) =

(x) =
x

(x) =
x

(x) =

(x) (46)
This means that the product of Lorentz transformations of two contravariant 4-vectors
is also a Lorentz transformations of contravariant 4-vector. This result, also is vaild for the
covariant Lorentz 4-vectors case and more general for the Lorentz 4-tensors case.
IV. LORENTZ 4-TENSORS
Construct the product T

and obtain the transformation law. In efect, multiplicand

the laws of transformation T

and T

, we obtain the transformation

T

. (47)
8
The product T

have a two free super-index libres, i.e., is a double-contravariant function.

Is clear that T

= f

(x

)
and
T

= g

(x). Of the Ec. (47) continues that the product T

in the quote system is

a bilineal and homogeneous function of the product T

in the not-quote system.

The product T

is a example of double-contravariant Lorentz 4-tensor, which we dene:

a set of 4
2
components T

(x) dened in the quoted system, form the double contravariant

Lorentz 4-tensor (or covariant of rank 2). If this components are relacionated with bilineal
and homogeneous with respect at T

(x) components dened in a reference frame not-quote

with the nex law of transformation
T

(x

)=

(x). (48)
Observe that, if T

and T

are contravariant 4-vectors then T

is a double-contravariant
Lorentz 4-tensor. But not every double-contravariant Lorentz 4-tensor can be written as
the product of two contravariant 4-vectors. The transformation rule of double-contravariant
Lorentz 4-tensor is
T

(x

)=

(x). (49)
and inverse transformation
T

(x)=

(x

). (50)
Analogamente, we obtain the transformation rule of a Lorentz second order mixed 4-tensor
(with one covariant and contravariant index)
T

(x)=

(x

). (51)
the inverse transformation
T

(x)=

(x

). (52)
With 4-vectors and 4-tensors transformation laws of decond order, we build the 4-tensors
laws of transformation with higer rank or order. Is simple. Need to properly multiply the
9
4-vectors and4-tensors laws of transformation of second order.
The previous results can be generalized for Lorentz 4-tensor of higth rank (order) r. A
set of 4
r=p+q
components T

2
...p

2
...q
(x

).
A set of 4
r=p+q
components T

2
...p

2
...q
(x

) dened in the quote inertial form the components

of Lorentz 4-tensor of rank r = p+q (with p times contravariant and q times contravariant),
if this components are multilineal and homogeneous related with respect to the components
T

2
...p

2
...q
(x) dened in the inertial reference not-quote frame, in accordance with the next
transformation law
T

2
...p

2
...q
(x

)=

2
...
p
p

2
...
q
q
T

2
...p

2
...q
(x). (53)
Is evident that the laws of transformation of 4-vectors in Ecs. (38),(41) and of the order
two 4-tensors in Ecs. (48), (50) y (51) are particular cases of the Ec. (53). The inverse
transformation of Ec. (53) is dened by
T

2
...p

2
...q
(x)=

2
...
p
p

2
...
q
q
T

2
...p

2
...q
(x

). (54)

Electronic address: geursan@ciencias.unam.mx

1
Shutz B, Geometrical methods of mathematical physics, 3 ed Cambridge university press, UK,
1999
2
Grits D, introduction to electrodynamics, 3 ed, Prentice hall, USA, 1999.
10