Tensor Formulation of Relativity

ics
am
yn
od
tr
ec
El
Shobhit Mahajan
&
shobhit.mahajan@gmail.com
r y
eo
Th
tic
ne
ag
om
tr
ec
El

Version 1.0
Last modified on February 22, 2020
 Shobhit Mahajan Levi Civita Tensor & Tensor Formulation of Special Relativity

0.1 Levi-Civita tensor

ics
were cyclic, anticyclic or any two or more were equal. This allowed us to simplify the cross product of
a vector as

am
1
(A × B)i = ijk (Aj Bk − Ak Bj ) = ijk Aj Bk
2

yn
We now generalise this to 3 + 1 dimensions which we use in Special Relativity. We define the Levi
Cevita tensor as a rank 4 tensor which is totally antisymmetric. This immediately tells us
that if any two of the indices are equal, the tensor is zero. We define it as follows:

od
tr
αβγδ = +1 if αβγδ are even permutations of 0, 1, 2, 3
αβγδ = −1 if αβγδ are odd permutations of 0, 1, 2, 3

ec
αβγδ = 0 otherwise

El
and the covariant form,

αβγδ = −1 if αβγδ
&
are even permutations of 0, 1, 2, 3
y
αβγδ = 1 if αβγδ are odd permutations of 0, 1, 2, 3
r
αβγδ = 0 otherwise
eo

Now what this means in operation is the following. Start always with 0123 which we
know that 0123 = 1 and keep switching indices till you get the combination you are looking
Th

for. Every time you interchange an index, you pick up a − sign.

1032 = −0132 = 0123 = 1

1320 = −1302 = 1032 = −0132 = 0123 = +1

1203 = −1023 = 0123 = +1

3201 = −3021 = 0321 = −0231 = 0213 = −0123 = −1

and so on.
ec

We also define generalised Kronecker delta tensors as follows:

 µ 
µν
 δ α δν α 
δ αβ =  µ
  (1)
δ β δν β 
and

1
 Shobhit Mahajan Levi Civita Tensor & Tensor Formulation of Special Relativity

 µ 
 δ α δν α δλα 
 
δ µνλ αβγ =  δ µ β δ ν β δ λ β 
 (2)
 δµγ δν γ δλγ 

ics
and so on.

am
With these definitions, we can prove the following identities.

yn
αβγδ αβγδ = −4!

od
αβγδ µβγδ = −3!δ α µ
αβγδ µνγδ = −2!δ αβ µν

tr
αβγδ µνρδ = −δ αβγ µνρ (3)

ec
Basically use the summation convention and sum over the required indices keeping in mind the rule
for picking up the minus sign.

El
As an example let us prove the first identity:
&
αβγδ αβγδ = −4!
y
Take α = 0. Then we have , using the definitions of αβγδ and αβγδ ,
r
eo

α = 0, β = 1, γ = 2, δ = 3 0123 0123 = −1
Th

α = 0, β = 1, γ = 3, δ = 2 0132 0132 = −1
α = 0, β = 2, γ = 3, δ = 1 0231 0231 = −1
tic

α = 0, β = 2, γ = 1, δ = 3 0213 0213 = −1
α = 0, β = 3, γ = 2, δ = 1 0321 0321 = −1
ne

α = 0, β = 3, γ = 1, δ = 2 0312 0312 = −1
ag

Thus with α = 0, we have the sum as −6. But α can take 4 values, 0, 1, 2, 3. Thus the expression

αβγδ αβγδ = −24 = −4!

Similarly, though with significantly more algebra, one can verify the other identities.
tr
ec

0.2 Tensor formulation of Special Relativity

We now apply the generalised concepts of tensors to the case of Special Relativity which is what we
need for this course. Recall that in special relativity, the transformations that we are interested in are
the Lorentz transformations.

From the definition of the invariant line element, ds2 and the metric, we can easily see that the
metric in this case, called the Minkowski metric is given by

2
 Shobhit Mahajan Levi Civita Tensor & Tensor Formulation of Special Relativity

ηµν = diag(1, −1, −1, −1)

ics
 
1 0 0 0
 0 −1 0 0 
η=

am

 0 0 −1 0 
0 0 0 −1

yn
For the case of transformations, recall that the contravariant vector transforms as

od
∂x0µ ν
A0µ = A (4)
∂xν

tr
In the case of relativity, the transformation matrices are the Lorentz matrices, that is

ec
R ∂x0µ

El
ν
= Λµ ν Relativity (5)
∂x
Similarly for the covariant vectors, we have &
R
y
∂xν
A0µ = Aν (6)
r
∂x0µ
eo

R
or
Th

∂xν
= Λµ ν Relativity (7)
∂x0µ
Contravariant vectors in Special relativity are called Four Vectors. Frequently, in this
tic

course when we speak of four vectors we shall be referring to contravariant vectors which transform as
Eq(4) with the transformation matrices being the Lorentz transformations, Eq(5).
ne

Once we have the transformation matrices for contravariant and covariant vectors, we can easily see
ag

what the transformation matrices for tensors would be in relativity.

Contravariant tensors:

R
tr

∂x0µ1 ∂x0µp ν1 ν2 ,···νp

∂xν1 ∂xνp
El

or in relativity

R T 0µ1 µ2 ,···µp = Λµ1 ν1 · · · Λµp νp T ν1 ν2 ,···νp Relativity (9)

Covariant tensors:

3
 Shobhit Mahajan Levi Civita Tensor & Tensor Formulation of Special Relativity

R Tµ0 1 µ2 ,···µp =
∂xν1
∂x0µ1
· · ·
∂xνp
Tν ν ,···ν
∂x0µp 1 2 p
(10)

ics
or in relativity

am
Tµ0 1 µ2 ,···µp = Λµ1 ν1 · · · Λµp νp Tν1 ν2 ,···νp Relativity (11)

yn
and for a mixed tensor

R 
∂x0µ1 ∂x0µm

∂xβ1 ∂xβn


od
0µ1 µ2 ,···µm
T ν1 ν2 ···νn = · · · · · · T α1 α2 ···αm β1 β2 ,···βn (12)
∂xα1 ∂xαm ∂x0ν1 ∂x0νn

tr
or in relativity

ec
T 0µ1 µ2 ,···µm ν1 ν2 ···νn = Λµ1 α1 · · · Λµm αm Λν1 β1 · · · Λνn βn T α1 α2 ···αm β1 β2 ,···βn

El
Relativity (13)

Raising and lowering of indices is easy once we have the metric. We can easily relate the contravariant
&
and covariant tensors of any rank. For instance, for vectors, which in this case are called four vectors,
we have
y
Aµ = ηµν Aν
r
eo

Since the RHS has a repeated index, we need to sum over it. Thus
Th

A0 = η0ν Aν = η00 A0 = A0
Ai = ηiν Aν = ηij Aj = −δij Aj = −Ai
tic

We know that 4-vectors, which recall are the contravariant vectors transform as the usual coordinates
under coordinate transformations. Thus we can write the 4-vector as

R
ne

4 Vector Aµ = (A0 , Ai ) = (A0 , A) (14)

where the spatial part is the usual 3-vector.

From the definition of covariant vectors and the Minkowski metric, we can see that

R
tr

R
Covariant Vector Aµ = (A0 , Ai ) = (A0 , −Ai ) = (A0 , −A) (15)
ec

This is extremely important to remember- the covariant vector’s spatial part is

the negative of the usual 3-vector.

Thus if we take a scalar product, Aµ Aµ , we get

R Aµ Aµ = ηµν Aµ Aν = η00 A0 A0 + ηµi Aµ Ai = (A0 )2 − δji Aj Ai = (A0 )2 − (A)2 (16)

4
 Shobhit Mahajan Levi Civita Tensor & Tensor Formulation of Special Relativity

and

R Bµ Aµ = ηµν Aµ B ν = η00 A0 B 0 + ηµi Aµ B i = A0 B 0 − δji Aj B i = A0 B 0 − A · B (17)

ics
NOTE: The summation is always adding up the elements. The reason we get a minus
sign in the spatial components is because the metric has a negative sign. Alternatively,

am
think about it as follows: The covariant vector has a spatial part which is negative of the
spatial part of the contravariant vector which is the ordinary three vector.

yn
What about rank 2 vectors? We can easily relate their components also for contra and covariant
indices.

od
tr
Tµν = ηµα ηνβ T αβ

ec
Thus

El
T00 = η0α η0β T αβ = T 00
T0i = η0α ηiβ T αβ = η00 ηij T 0j = −δij T 0j
&
Ti0 = ηiα η0β T αβ = ηij T j0 = −δij T j0
Tii = ηiα ηiβ T αβ = ηik ηil T kl = δik δjl T kl
r y
Note that if we choose T µν = η µν we can easily see that
eo

η00 = η 00 = 1, ηij = η ij = −δij

Thus the components of the contravariant and covariant Minkowski metrics are identical.
tic
ne
ag
om
tr
ec
El