Non-coordinates basis in General Relativity and Cartan’s structure equations

Wytler Cordeiro dos Santos

Universidade de Brası́lia, CEP 70910-900, DF, Brasil

Abstract
The basic and fundamental aspects of General Relativity are in general analysed in mathematical level of coordinate
arXiv:1711.09503v1 [gr-qc] 27 Nov 2017

basis or holonomic frame by several authors in the literature. However, for many purposes it is more convenient to
use a general basis, often called in four dimensions, a tetrad or vierbein, very useful in a local frame with orthonormal
basis or pseudo-orthonormal basis. This text presents an introduction to non-coordinate basis and the two Cartan’s
structure equations that are mathematical implements in Riemannian geometry that facilitate the calculation of cur-
vature tensors. The purpose of this text is to approach the language and the notation of tetrad field or vierbein with
conceptual and calculational details.
Keywords: Cartan Equations, Orthonormal Basis, Null Complex Tetrad

1. Introduction

The foundations for a precise and mathematical formulation of General Relativity is obtained with basic ideas
of differential geometry, i.e.: basic properties of manifolds and tensor fields. However in a level of basic courses of
General Relativity, such as of references [1, 2, 3, 4, 5, 6], we have worked just only with coordinate basis or world
indices. In some advanced topics of General Relativity it is necessary to work with objects defined on manifold as
torsion and curvature in other perspective, i.e.: in a non-coordinate basis or local indices. In a non-coordinate basis,
it arises the structure equations of Cartan. The Cartan formalism works with two equations that approach torsion and
curvature in the differential forms language.
The discussion of this text begins with vectors and tensors on manifolds, perhaps it is util to revisite some dis-
cussions and approaches in differential manifolds in references [7, 8, 9, 10, 11]. Coming after the differential forms,
exterior product an exterior derivatives, definition of torsion and curvature tensors and metric. Here we focus on
vectors and tensors on Riemannian manifold with Lorentzian signature of spacetime metric: (-+++), and then we
introduce the local frame with orthonormal basis and pseudo-orthornormal basis. The pseudo-orthonormal basis is
commonly known by complex null tetrad that is very useful to Newman-Penrose formalism [12, 13, 14]. The struc-
ture equations of Cartan are presented as a compact and efficient method for calculating the components of curvature
tensor with respect to a non-coordinate basis, and then as examples are presented the Schwarzschild and Brinkmann
spacetimes.

2. Riemannian Geometry

In a nutshell, a manifold M is a topological space which locally resembles to Rn , where a homeomorphism enable
us to obtain in each point in a manifold a set of numbers called the local coordinate. The inner product between two
vectors in a tangent space of a manifold is defined with aid of a metric tensor, in this case the manifold is named
Riemannian manifold.

Email address: wytler@fis.unb.br (Wytler Cordeiro dos Santos)

December 15, 2021

2.1. Tangent vectors and differential forms
In general a vector can not be considered as an arrow connecting two points of the manifold. For a consistent
generalization of the concept of vector in Rn , we can identifies vector on manifold M with tangent vector. So, a
tangent vector V at p ∈ M is a linear operator which assigns to each differentiable function f on M a real number
V( f ). This operator satisfies the axioms:
1. V( f + g) = V( f ) + V(g)
2. V( f g) = gV( f ) + f V(g)
3. V(c f ) = cV( f ), c is a constant.

A tangent vector is just a directional derivative along a curve γ(t) through p ∈ M

!    
dxµ  dxµ ∂ f 
   
∂f  
 ∂f 
 
 



 = µ


 

 = µ
 .

 (1)
∂t γ t=t

0
∂x γ(t ) dt t=t

0

0
dt ∂x  γ(t )
0

The directional derivative of a function f is obtained by applying the differential operator V to f , where

dxµ 

∂ 
V = Vµ µ , where V µ =



 (2)
∂x dt 
t=t 0

that is ! 

∂f 
 ∂f
= V µ µ = V[ f ].



 (3)
∂t 

γ t=t
0
∂x
∂
Thus every tangent vector at a point p can be expressed as a linear combination of the coordinate derivatives µ . The
∂x
directional derivatives along the coordinate lines at p form a basis of an n-dimensional
( vector
) space whose elements
∂
are the tangent vector at p. This space is called the tangent space T p and the basis is called a coordinate basis
∂xµ
or holonomic( frame.
) Thus every tangent vector at a point p can be expressed as a linear combination of the coordinate
∂
derivatives . Any tangent vectors to M at p of the n-dimensional vector space T p are referred as contravariant
∂xµ
vector.
The commutator, [U, V], of two vector fields U and V is defined by

[U, V]( f ) = U(V( f )) − V(U( f )),

∂ ∂
for a given coordinates basis U = U κ κ
and V = V λ λ we have that
∂x ∂x
! !
∂ ∂ ∂ ∂
[U, V] = Uκ κ Vλ λ − Vλ λ Uκ κ
∂x ! ∂x ∂x ∂x !
λ κ
κ ∂V ∂ κ λ ∂2 λ ∂U ∂ λ κ ∂2
= U + U V − V − V U
∂xκ ! ∂xλ ∂xκ!∂xλ ∂xλ ∂xκ ∂xλ ∂xκ
λ κ
∂V ∂ ∂U ∂
= Uκ − Vλ
∂xκ ∂xλ λ
! ∂x ∂x
κ
λ λ
∂V ∂U ∂ ∂
= Uκ κ − Vκ κ = Wλ λ ,
∂x ∂x ∂xλ ∂x

∂V λ κ ∂U
λ
where W λ = U κ − V . Thus, the commutator defines a new vector
∂xκ ∂xκ
[U, V] = W. (4)
2
The commutators satisfy the Jacobi identity

[U, [V, W]] + [V, [W, U]] + [W, [U, V]] = 0 (5)

for arbitrary U, V and W.

A one-form (1-form) or covariant vector or dual vector ω at p maps a tangent vector V into a real number, the
contraction, denoted by hω, Vi, and this mapping is linear:

hω, aU + bVi = ahω, Ui + bhω, Vi (6)

holds for all a, b ∈ R and U, V ∈ T p . Linear combinations of 1-forms ω, τ are defined by the rule

haω + bτ, Vi = ahωVi + bhτ, Vi. (7)

An arbitrary 1-form ω is written as a linear combination,

ω = ωµ dxµ , (8)
( )
∂
where {dxµ } is a dual basis to the basis of tangent vectors, being called cotangent basis. The 1-forms form a
∂xµ
basis of n-dimensional dual vector space called the dual or cotangent space T ∗p with basis {dxµ } with conditions
* +
µ ∂ ∂xµ
dx , ν = ν = δµ ν . (9)
∂x ∂x

The contraction hω, Vi is the inner product between a contravariant vector V and a covariant or dual vector ω. For
ω ∈ T ∗p and V ∈ T p we have that inner product is
* + * +
∂ ∂
hω, Vi = ωµ dxµ , V ν ν = ωµ V ν dxµ , ν = ωµ V ν δµ ν = ωµ V µ . (10)
∂x ∂x

In terms of the dual basis, the differential d f on an arbitrary function f is given by

∂f µ
df = dx . (11)
∂xµ
2.2. Tensors
A tensor of type (q, r) at a point p is a multilinear object which maps q elements of T p and r elements of T ∗p to a
q
real number. Tr (M) at p denotes the set of type (q, r) tensors at p ∈ M, by tensor product

Trq (M) p = T p ⊗ · · · ⊗ T p ⊗ T ∗p ⊗ · · · ⊗ T ∗p . (12)

| {z } | {z }
q factors r factors
( )
∂
In particular, T01 = T p and T10 = T ∗p . An element of Trq (M) is written in terms of the basis and {dxν } as
∂xµ

∂ ∂
T = T µ1···µq ν1···νr ⊗ · · · ⊗ µq ⊗ dxν1 ⊗ · · · ⊗ dxνr , (13)
∂x µ1 ∂x
where all indices run from 1 to n. The coefficients T µ1···µq ν1···νr (
with contravariant
) indices µ1 · · · µq and covariant
∂
indices ν1 · · · νr are the components of T with respect to the basis and {dxν }.
∂xµ

3
2.3. Differential forms, exterior product and exterior derivatives
A differential form of order r, or r-form, is a totally antisymmetric tensor of type (0, r). If ρ and σ are r-form and
s-form respectively, one can define a (r + s)-form ρ ∧ σ from them, where ∧ is the skew-symmetrized tensor product
⊗. The exterior product or wedge product ρ ∧ σ is the tensor of type (0, r + s) [15]. For example, if

(i) ρ = rµ dxµ and σ = sν dxν

ρ and σ are 1-form, then

ρ ∧ σ = rµ sν (dxµ ⊗ dxν − dxν ⊗ dxµ ) = rµ sν dxµ ∧ dxν = −σ ∧ ρ,

ρ ∧ σ is a 2-form.
Now if
(ii) ρ = rλ dxλ , σ = sµ dxµ and τ = tν dxν
then

ρ∧σ∧τ = rλ sµ tν (dxλ ⊗dxµ ⊗dxν +dxν ⊗dxλ ⊗dxµ +dxµ ⊗dxν ⊗dxλ −dxλ ⊗dxν ⊗dxµ −dxν ⊗dxµ ⊗dxλ −dxµ ⊗dxλ ⊗dxν )

or
ρ ∧ σ ∧ τ = rλ sµ tν dxλ ∧ dxµ ∧ dxν ,
ρ ∧ σ ∧ τ is a 3-form.
For the skew-symmetrized tensor product ∧, it is verified that the exterior product satisfies the fllowing axioms:
1. is linear in each dxµ
2. vanish if any two factors coincide
3. changes sign if any two factors are interchanged.

If we denote the vector space of r-forms at point p ∈ M by Ωr (M) and an element ρ ∈ Ωr (M) is expaned as

ρ = rµ1,µ2,··· ,µr dxµ1 ∧ dxµ2 ∧ · · · ∧ dxµr ,

1 ≤ µ1 < µ2 < · · · < µr ≤ n, with r ≤ n.

The above axiom 2 implies that these exterior products vanish for r > n. !
n
We have already stated that rµ1,µ2···µr are totally antisymmetric, then there are r-combinations of choices of
r
!
n
set (µ1, µ2, · · · , µr), thus the dimensional of the vector space Ωr (M) is . For example, a manifold which has
r
dim(M) = 3 with dx, dy and dz. The 2-forms: dx ∧ dy, dx ∧ dz and dy ∧ dz are oriented area elements. dx ∧ dy ∧ dz
is oriented volume element. We have:
1 ∗ 0 1
1. {dx,
! dy, dz} ∈ Ω (M) = T (M) where Ω (M) is vector space of smooth functions. Ω (M) has dimension
3
= 3.
1
!
3
2. {dx ∧ dy, dx ∧ dz, dy ∧ dz} ∈ Ω2 (M) where Ω2 (M) has dimension = 3.
2
!
3 3 3
3. {dx ∧ dy ∧ dz} ∈ Ω (M) where dim(Ω (M)) = = 1.
3

4
 Let ρ ∈ Ωr (M), σ ∈ Ω s (M) and τ ∈ Ωt (M), the exterior product is associative:

(ρ ∧ σ) ∧ τ = ρ ∧ (σ ∧ τ). (14)

However, the commutative law is slighty changed:

ρ ∧ σ = (−1)rsσ ∧ ρ. (15)

The definition of an r-form is extended to include ordinary function f on manifold M. Functions on M are called
0-form; 1-form is a covariant vector, 2-form is an antisymmetric covariant tensor of rank 2 and so on.
If ρ is an r-form given by
ρ = rµ1,µ2,··· ,µr dxµ1 ∧ dxµ2 ∧ · · · ∧ dxµr ,
then the exterior derivative of ρ is written dρ and is defined by
∂rµ1,µ2,··· ,µr ν
dρ = dx ∧ dxµ1 ∧ dxµ2 ∧ · · · ∧ dxµr , (16)
∂xν
The exterior derivative d is a map Ωr (M) → Ωr+1 (M). For example: if M = R3 and
1. 0-form: f = f (x, y, z) then the action of exterior derivative is
∂f ∂f ∂f
df = dx + dy + dz, (17)
∂x ∂y ∂z
it is identified with gradient;
2. 1-form: ρ = ρ x (x, y, z) dx + ρy (x, y, z) dy + ρz (x, y, z) dz then the action of exterior derivative is
! ! !
∂ρy ∂ρ x ∂ρz ∂ρy ∂ρ x ∂ρz
dρ = − dx ∧ dy + − dy ∧ dz + − dz ∧ dx (18)
∂x ∂y ∂y ∂z ∂z ∂x
it is identified with rotational;
3. 2-form: σ = σ xy (x, y, z) dx ∧ dy + σyz (x, y, z) dy ∧ dz + σzx (x, y, z) dz ∧ dx then the action of exterior derivative is
!
∂σyz ∂σzx ∂σ xy
dσ = + + dx ∧ dy ∧ dz (19)
∂x ∂y ∂z
it is identified with divergence.

The second exterior derivative of an r-form, it is verified that d 2 ρ = 0. If ρ is an r-form given by

ρ = rµ1,µ2,··· ,µr dxµ1 ∧ dxµ2 ∧ · · · ∧ dxµr ,

then the d2 ρ is
∂2 rµ1,µ2,··· ,µr ν
d2 ρ = dx ∧ dxλ ∧ dxµ1 ∧ dxµ2 ∧ · · · ∧ dxµr ,
∂xλ ∂xν
∂2 rµ1,µ2,··· ,µr
since is symmetric with respect ν and λ while dxν ∧ dxλ is antisymmetric. Another important property is
∂xλ ∂xν
d(ρ ∧ σ) = dρ ∧ σ + (−1)r (ρ ∧ dσ) (20)

where ρ is an r-form.
For example, the electromagnetic potential
 
 φ 
A 
(Aµ ) =  x  ,
 Ay 
Az
5
can be assigned by 1-form A = Aν dxν . It remains for us to describe the electromagnetic tensor F = d A as a 2-form,
!
∂Aν µ ν 1 ∂Aν ∂Aν
dA = dx ∧ dx = + dxµ ∧ dxν
∂xµ 2 ∂xµ ∂xµ !
1 ∂Aν µ ∂Aµ
= µ
dx ∧ dxν + ν dxν ∧ dxµ
2 ∂x ! ∂x
1 ∂Aν ∂Aµ
= − ν dx ∧ dxν
µ
2 ∂xµ ∂x
1
= Fµν dx ∧ dxν ,
µ
2
1
where F = Fµν dxµ ∧ dxν . The components of Fµν is given by:
2
 
 0 −E x −Ey −Ez 

 E 0 Bz −By 
(Fµν ) =  x .
 Ey −Bz 0 B x 
Ez By −B x 0

The action of exterior derivative on F results in

1 ∂Fµν λ
dF = dx ∧ dxµ ∧ dxν ,
2 ∂xλ
where dF = d2 A = 0, then
1 
∂λ Fµν dxλ ∧ dxµ ∧ dxν = ∂λ Fµν dxλ ∧ dxµ ∧ dxν + ∂µ Fνλ dxµ ∧ dxν ∧ dxλ + ∂ν Fλµ dxν ∧ dxλ ∧ dxµ = 0,
3
we can interchange dxµ ∧ dxν ∧ dxλ and dxν ∧ dxλ ∧ dxµ of above equation,

dxµ ∧ dxν ∧ dxλ = dxλ ∧ dxµ ∧ dxν

and
dxν ∧ dxλ ∧ dxµ = dxλ ∧ dxµ ∧ dxν ,
such as  
∂λ Fµν + ∂µ Fνλ + ∂ν Fλµ dxλ ∧ dxµ ∧ dxν = 0,
which is known as the Bianchi identity, that follows from this, the two homogeneous Maxwell’s equations,

∂B
∇·B=0 and + ∇ × E = 0.
∂t
2.4. The covariant derivative
The exterior derivative is a limited generalization acting only on forms. So, it is necessary to introduce covariant
derivative ∇ X in the direction of the vector X at p on manifold M, that maps an arbitrary tensor into a tensor of same
type. The symbol ∇ is called affine connection. Let X, Y and Z vector fields and f a scalar function. The covariant
derivative satisfies the following conditions:

∇ X (Y + Z) = ∇ X Y + ∇ X Z; (21)

∇(X+Y) Z = ∇ X Z + ∇Y Z; (22)
∇ f X Y = f ∇ X Y; (23)
∇ X ( f Y) = X[ f ]Y + f ∇ X Y. (24)

6
where from (3) we have that
∂f
∇ X f = X[ f ] = X µ .
∂x µ
∂ ∂
The covariant derivative of the basis vector Eν = ν in the direction of basis vector Eµ = can be expanded
∂x ∂x µ
in terms of basis vectors:
∇ Eµ Eν = Γρ µν Eρ (25)
where D E
Γρ µν = dx ρ , ∇ Eµ Eν (26)
is called connection coefficients.
The covariant derivative ∇ X Y, where X = X µ Eµ and Y = Y ν Eν is given by vector

∇ X Y = ∇(X µ Eµ ) (Y ν Eν )

with conditions (23) and (24) we have

 
∇ X Y = X µ ∇ Eµ (Y ν Eν ) = X µ Eµ [Y ν ]Eν + Y ν ∇ Eµ Eν

with (25) we have !

µ ∂Y ν
∇X Y = X Eν + Y ν Γρ µν Eρ
∂xµ
thus, this covariant derivative is given by the vector
 
∇ X Y = X µ ∂µ Y ρ + Y ν Γρ µν Eρ . (27)

By definition, the affine connection ∇ maps two vectors X and Y to a new vector given by of the right hand side of
(27), whose ρth component is  
X µ ∂µ Y ρ + Y ν Γρ µν ≡ X µ ∇µ Y ρ .
If we compute the covariant derivative ∇ Eµ Y, we have:
   
∇ Eµ Y = Eµ [Y ν ]Eν + Y ν ∇ Eµ Eν = ∂µ Y ρ + Y ν Γρ µν Eρ = ∇µ Y ρ Eρ ,

where ∇µ Y ρ is the ρth component of vector ∇ Eµ Y.

The covariant derivative of 1-form ω = ων dxν is obtained from contraction of ω with a tangent vector Y,

hω, Yi = f.

If we put Y = Eν and ω = dxρ in above equation, it becomes

hdxρ , Eν i = δρ ν ,

then the covariant derivative of above equation with respect to Eµ is

∇ Eµ hdxρ , Eν i = 0
D E D E
∇ Eµ dxρ , Eν + dxρ , ∇ Eµ Eν = 0
we can use (26) D E
∇ Eµ dxρ , Eν = −Γρ µν ,
and as consequence of above equation we can identify

∇ Eµ dxρ = −Γρ µσ dxσ . (28)

7
Furthermore, if we calculate derivative of 1-form ω = ων dxν with respect to Eµ , we have that,

∇ Eµ ω = Eµ [ων ]dxν + ων ∇ Eµ dxν

ν ν σ
= ∂µ ων dx − ων Γ µσ
 dx
ν σ
= ∂ ω −ω Γ dx
 µ σ  ν µσ
σ
= ∇µ ωσ dx ,

where we can identify the σth component of 1-form ∇ Eµ ω as

∇µ ωσ = ∂µ ωσ − ων Γν µσ . (29)

We require that the Leibnitz rule be true for any tensor products,
 
∇ Eµ (T1 ⊗ T2 ) = ∇ Eµ T1 ⊗ T2 + T1 ⊗ ∇ Eµ T2 . (30)

If we have a tensor of type (1,1) T = T µ ν Eν ⊗ dxµ , under the action of covariant derivative and Leibnitz rule we have
that
 
∇ Eρ T = ∇ Eρ T µ ν Eν ⊗ dxµ
 
= Eρ [T µ ν ]Eν ⊗ dxµ + T µ ν ∇ Eρ Eν ⊗ dxµ + T µ ν Eν ⊗ ∇ Eρ dxµ
 
= ∂ρ T µ ν Eν ⊗ dxµ + T µ ν Γσ ρν Eσ ⊗ dxµ + T µ ν Eν ⊗ −Γµ ρσ dxσ
 
= ∂ρ T µ ν + Γν ρσ T µ σ − Γσ ρµ T σ ν Eν ⊗ dxµ ,

where we have that:

∇ρ T µ ν = ∂ρ T µ ν + Γν ρσ T µ σ − Γσ ρµ T σ ν (31)

2.5. The torsion and curvature

Intrinsic objects defined on manifolds that measure the bends on manifolds are torsion tensor and curvature (Rie-
mann) tensor. Let X, Y and Z be vectors of tangent space T p , then the torsion tensor and curvature tensor are defined
by
T(X, Y) ≡ ∇ X Y − ∇Y X − [X, Y] torsion tensor (32)
R(X, Y)Z ≡ ∇ X ∇Y Z − ∇Y ∇ X Z − ∇[X,Y] Z curvature tensor. (33)
The torsion tensor gives an intrinsic characterization of how tangent spaces twist about a curve when they are parallel
transported. The curvature or Riemann ( tensor describes
) how the tangent spaces roll along the curve.
∂
With respect to coordinate basis Eµ = and the dual basis {dx µ }, the components of torsion tensor is given
∂x µ
by    
T λ µν = hdxλ , T(Eµ , Eν )i = hdxλ , ∇ Eµ Eν − ∇ Eν Eµ − [Eµ , Eν ] i = hdxλ , Γκ µν Eκ − Γκ νµ Eκ − 0 i
where it follows that
T λ µν = Γλ µν − Γλ νµ . (34)
And the components of curvature tensor with respect to coordinate basis is given by

Rκ λµν = hdx κ , R(Eµ , Eν )Eλ i, (35)

where we have

 
R(Eµ , Eν )Eλ = ∇ Eµ ∇ Eν Eλ − ∇ Eν ∇ Eµ Eλ − ∇[Eµ ,Eν ] Eλ
   
= ∇ Eµ Γ ρ νλ Eρ − ∇ Eν Γ ρ µλ Eρ − 0,

we recall (24) where ∇ X ( f Y) = X[ f ]Y + f ∇ X Y, such as,

  h i
R(Eµ , Eν , Eλ ) = Eµ Γ ρ νλ Eρ + Γ ρ νλ ∇ Eµ Eρ − Eν Γ ρ µλ Eρ − Γ ρ µλ ∇ Eν Eρ
8
 ρ ρ σ ρ ρ σ
=  µ Γ νλ Eρ + Γ νλ Γ µρ Eσ − ∂ν Γ µλ Eρ − Γ µλ Γ νρ Eσ
∂
ρ ρ σ ρ σ ρ
= ∂µ Γ νλ − ∂ν Γ µλ + Γ νλ Γ µσ − Γ µλ Γ νσ Eρ ,
it follows that
Rκ λµν = ∂µ Γ κ νλ − ∂ν Γ κ µλ + Γ σ νλ Γκ µσ − Γ σ µλ Γκ νσ . (36)
We readily find
T λ µν = −T λ νµ and Rκ λµν = −Rκ λνµ . (37)

2.6. Metric tensors

We have defined the contraction hω, Vi as the inner product between a contravariant vector V and a covariant or
dual vector ω. For ω ∈ T ∗p and V ∈ T p we have that inner product is
hω, Vi = ωµ V µ .
Now, we can define a linear map g p (U, V) as the inner product between two contravariant vectors U, V ∈ T p . The
Riemannian metric g p is a type (0,2) tensor field on manifold M which satisfies
(i) g p (U, V) = g p (V, U) ∈ R;
(ii) g p (U, U) ≥ 0 assigns a magnitude of vector U ∈ T p at point p. g p (U, U) = 0 is true only when U = 0.

g p is called pseudo-Riemannian metric when the metric is no longer positive definite. Since g p is a type (0,2) tensor
field on M, it is expaned in terms of dxµ ⊗ dxν as
g p = gµν (p) dxµ ⊗ dxν . (38)
( )
∂
The components of g p with respect a coordinate basis Eµ = µ are
∂x
gµν (p) = g p (Eµ , Eν ) = g p (Eν , Eµ ), (39)
where we can see that (gµν ) is a symmetric matrix (we can omit p in metric tensor). The matrix (gµν ) has an inverse
denoted by (gµν ) where we have
gµν gνλ = gλν gνµ = δµ λ . (40)
The map g(U, ) : T p M → R by V 7→ g(U, V) is identified with 1-form. Thus, if U µ are components of a
contravariant vector U ∈ T p , then Uµ are the components of a uniquely associated covariant vector, where
Uµ = gµν U ν , (41)
such as
g(U, V) = g(U µ Eµ , V ν Eν ) = U µ V ν g(Eµ , Eν ) = gµν U µ V ν = Uν V ν = U µ Vµ . (42)
Due to isomorphism between tangent space T p and cotangent sapce T ∗p we have
U µ = gµν Uν . (43)
In Special Relativity, two separated events with coordinates (t, x, y, z) and (t+dt, x+dx, y+dy, z+dz) in a particular
inertial frame have the square of the infinitesimal interval between them given by
ds2 = −dt2 + dx2 + dy2 + dz2 , (44)
which is known as the line element of Minkowski spacetime and is invariant under any Lorentz transformation [1, 2,
3, 4, 5, 6]. We can rewrite this line element in matrix form
  
−1 0 0 0  dt 
   0 1 0 0 dx
ds2 = dt dx dy dz     ,
 0 0 1 0 dy
0 0 0 1 dz
9
or
ds2 = ηµν dxµ dxν . (45)
It is common to regard (ηµν ) as η = diag(−1, 1, 1, 1) whose signature of Minkowski spacetime metric is (− + ++). The
Minkowski spacetime is denoted by (R4 , η).
Any metric g in four-dimensional differentiable manifold M with signatures (− + ++) is called Lorentzian metric
(a pseudo-Riemannian metric tensor). General Relativity is based on the concept of spacetime, which is a four-
dimensional differentiable manifold M, free of torsion with Lorentzian metric, denoted by (M, g) or just V4 . There are
others Theories of Gravitation such as Einstein-Cartan(-Sciama-Kibble) Theory of Gravity concepted in a Riemann-
Cartan spacetime with non-vanishing torsion and curvature. Riemann-Cartan spacetimes are denoted by U4 [16].
In accordance with the concept of line element of Minkowski spacetime (45), if we take an infinitesimal displace-
∂
ment dx µ µ ∈ T p to M and plug into g, we have
∂x
! !
µ ∂ ν ∂ µ ν ∂ ∂
2
ds = g dx , dx = dx dx g , = gµν dx µ dx ν (46)
∂x µ ∂x ν ∂x µ ∂x ν
that represents the length of the infinitesimal arc determined by the coordinate displacement x µ → x µ + dx µ .
In spacetime (M, g) where the metric is Lorentzian, the vectors are divided into three classes:
(i) g(U, U) > 0, U is spacelike,
(ii) g(U, U) < 0, U is timelike,
(iii) g(U, U) = 0, U is lightlike or null.
Now that a manifold M can be endowed with a metric g, we can put some restrictions on the possible form of
connections Γλ µν . We demand that the metric gµν be covariantly constant,
∇λ gµν = ∂λ gµν − Γκ λµ gκν − Γκ λν gκµ = 0. (47)
Then, the affine connection ∇ is said to be a metric connection. In a coordinate basis, with (25) and (39), we obtain
that    
Γλµν = g Eλ , ∇ Eµ Eν = g Eλ , Γκ µν Eκ = Γκ µν gλκ . (48)
In General Relativity the spacetime is V4 where the manifold is free of torsion, then with equation (47) we can obtain
the Christoffel relations
1 
Γλµν = ∂ν gλµ + ∂µ gλν − ∂λ gµν (49)
2
for the coordinate components of the connection.
From equation (47) we have that
∂λ gµν − Γνλµ − Γµλν = 0, (50)
where we can rewrite above equation to 1-form language
∂gµν λ
dx − Γνλµ dxλ − Γµλν dxλ = 0
∂xλ
or
d(gµν ) − Γµλν dxλ − Γνλµ dxλ = 0.
Introducing the connection 1-forms
Γµν ≡ Γµλν dxλ , (51)
we have that
d(gµν ) − Γµν − Γνµ = 0. (52)
If we have constant metric or a rigid frame, ∂λ gµν = 0, then
Γµν = −Γνµ . (53)
We will see in the next sections that the orthonormal and pseudo-orthonormal basis are rigid frames with the connec-
tion 1-forms obeying the above propertie.
10
3. Non-coordinate basis

In General Relativity there is a possibility that to set up a system of locally inertial coordinates, valid in a suffi-
ciently small region of spacetime. Strictly speaking, this must be an infinitesimal region surrounding at a point p ∈ M
with coordinates xµ . We can denote these local coordinates by yα . We can denote the transformation matrix, which
relates the two sets of coordinates - coordinate basis and locally inertial coordinates - by
∂xµ ∂yα
eα µ (p) = and the inverse ωα µ (p) = (54)
∂yα ∂xµ
If we set up a locally inertial frame of reference at each point of spacetime, in such a way that the directions of their
axes vary smoothly from one point to another, then we obtain a set of four vector fields e0 µ , · · · , e3 µ which specify, at
each point, the directions of these axes.
For instance, a tangent vector at a point p ∈ M can be expressed as a linear combination of a coordinate( ) basis or
∂
holonomic frame in accordance with equation (2). In coordinate basis, T p is spanned by {Eµ } = and T ∗p by
∂xµ
{dx µ }. Now, we can express a new basis of vectors in a system of locally inertial coordinates (local Lorentz spacetime)
by linear combinations
êα = eα µ Eµ . (55)
for a new basis of T p and a new dual basis of T ∗p ,

θ̃β = ωβ ν dxν . (56)

From (55) we can see that new basis {êα } is the frame of basis vectors which is obtained by a rotation of the basis {Eµ }
preserving the orietation, where {eα µ } ∈ GL(4, R). The new basis {êα } is called non-coordinate basis of the tangent
space T p .
∂f
The directional derivative in the general spacetime coordinate Eµ [ f ] = µ implies that êα [ f ] also is a directional
∂x
derivative, from (55) it follows that
∂f
êα [ f ] = eα µ µ , (57)
∂x
or
êα [ f ] = ∂α f, (58)
the directional derivative in the local Lorentz spacetime.
Since {θ̃β } is a new basis of the dual space T ∗p of the tangent space T p in local Lorentz coordinates, we require that

hêα , θ̃β i = δα β ,

where it follows
heα µ Eµ , ωβ ν dxν i = eα µ ωβ ν hEµ , dxν i = eα µ ωβ µ = δα β , (59)
α µ α µ µ µ
where ω µ is the inverse of eα . Also, it is valid that ω ν eα = δ ν . The matrix eα is called the tetrad or vierbein
field. The vierbein field, eα µ , has two kinds of indices: µ = 1, 2, 3, 4, labels the general spacetime coordinate and
α = 1, 2, 3, 4 labels the local spacetime. We use κ, λ, µ, ν, ρ, σ, τ, φ, etc to denote the coordinate basis {Eµ } and {dxν },
while α, β, γ, δ, ǫ, etc, to denote the non-coordinate basis {êα } and {θ̃β }.
From expression (55) where êα = eα µ Eµ , we can obtain a reverse expression by using ωα ν ,

ωα ν êα = ωα ν eα µ Eµ = δµ ν Eµ

that it gives
Eµ = ωα µ êα . (60)
α α µ
In the same way, the reverse expression for (56), where θ̃ = ω µ dx , results that

dxµ = eα µ θ̃α . (61)

11
 ∂f
We can verify that the differential of an arbitrary scalar function f given by d f = µ dxµ in coordinate basis is
∂x
given in non-coordinate basis by substituting dxµ by use of the equation (61),

∂f µ α
df = eα θ̃ ,
∂xµ
we can use (57) and (58) where it results in
d f = θ̃α ∂α f. (62)
Let a tensor field of type (1,1) given by T = T µ ν Eµ ⊗ dxν expressed in non-coordinate basis as

T = T µ ν (ωα µ êα ) ⊗ (eβ ν θ̃β ) = (T µ ν ωα µ eβ ν ) êα ⊗ θ̃β

where we have
T = T α β êα ⊗ θ̃β ,
with a way to transit tensors from coordinate basis to non-coordinate basis by

T α β = T µ ν ωα µ eβ ν . (63)

The metric tensor g defined by (38) in terms of non-coordinate basis is

g = gµν dxµ ⊗ dxν = gαβ ωα µ ωβ ν dxµ ⊗ dxν = gαβ (ωα µ dxµ ) ⊗ (ωβ ν dxν )
g = gαβ θ̃α ⊗ θ̃β . (64)

The components of g with respect a coordinate basis {êα } are

g(êα , êβ ) = gαβ (65)

where we have

g(êα , êβ ) = g(eα µ Eµ , eβ ν Eν ) = eα µ eβ ν g(Eµ , Eν ) = eα µ eβ ν gµν , (66)

that follows that

eα µ eβ ν gµν = gαβ . (67)
In the inverse way it results that
gµν = ωα µ ωβ ν gαβ . (68)
The inner product is preserved by vierbein eα µ that maps the tangent space to local frame. Let U and V two vectors
in tangent space, the inner product in coordinate basis is given by (42),

g(U, V) = gµν U µ V ν .

The above inner product in local frame is obtained by substituting (68) into above equation,

g(U, V) = ωα µ ωβ ν gαβ U µ V ν = gαβ (ωα µ U µ )(ωβ ν V ν ) = gαβ U α U β .

From (55) we can see that the non-coordinate basis {êα } is the frame of basis vectors which is obtained by a rotation
of the basis {Eµ } preserving the orietation and inner product.
The action of covariant derivative in (59) results in

(∇ν eα µ ) ωβ µ + eα µ ∇ν ωβ µ = 0 (69)

that it follows
ωβ µ ∇ν eα µ = −eα µ ∇ν ωβ µ . (70)

12
To lower the coordinate indice of the vierbein field, eα µ , we ca use

gµν eα µ = eαν

and when we use (68), gµν = ωα µ ωβ ν gαβ , on above equation, it becomes

eαν = ωβ µ ωγ ν gβγ eα µ = (ωβ µ eα µ )(gβγ ωγ ν ) = δβ α ωβν

from which we find

eαµ = ωαµ . (71)
Now, we can rewrite (68) such as
gµν = ωα µ ωαν , (72)
we can see that the vierbein fields can be interpreted as ‘square root’ of metric.
From the identification of vectors with directional derivatives we can conclude that the result of the successive
application of two vectors to a function depends on the order in which operators are applied. Let êα and êβ two vectors
of a non-coordinate basis {êα },
∂ ∂
êα = eα µ µ and êβ = eβ ν ν ,
∂x ∂x
then the commutation operation applied to a function f follows
!
∂ ∂ ∂ ∂
[êα , êβ ] f = eα µ µ eβ ν ν − eβ ν ν eα µ µ f
∂x ∂x ∂x ∂x !
ν 2 µ 2
µ ∂e β ∂ f µ ν ∂ f ν ∂eα ∂ f ν µ ∂ f
= eα + eα eβ − eβ − eβ eα
∂xµ ∂xν ∂xµ ∂x!ν ∂xν ∂xµ ∂xν ∂xµ
ν µ
∂e β ∂ ∂e α ∂
= eα µ µ ν − eβ ν ν f
∂x ∂x ∂x ∂xµ !
ν
∂eβ ∂ ∂eα ν ∂
= eα µ µ ν − eβ µ µ ν f
∂x ∂x ∂x
! ∂x
ν ν
µ ∂eβ µ ∂eα ∂
= eα − eβ f.
∂xµ ∂xµ ∂xν
It is util to introduce the commutator coefficients
!
∂eβ ν µ ∂eα
ν
eα µ − e β = Dγ αβ eγ ν , (73)
∂xµ ∂xµ
where the commutator of êα and êβ becomes
h i ∂
êα , êβ = Dγ αβ eγ ν ν ,
∂x
or simply h i
êα , êβ = Dγ αβ êγ , (74)
similar to Lie algebra. Let us now isolate the commutator coefficients from (73) by using ωδ ν
 
Dγ αβ eγ ν ωδ ν = ωδ ν eα µ ∂µ eβ ν − eβ µ ∂µ eα ν
 
Dγ αβ δγ δ = ωδ ν eα µ ∂µ eβ ν − eβ µ ∂µ eα ν
 
Dδ αβ = ωδ ν eα µ ∂µ eβ ν − eβ µ ∂µ eα ν . (75)

Due to the commutator from this equation, it is important to notice that

Dδ αβ = −Dδ βα , (76)

where the commutator coefficients have antisymmetry in the index pair (αβ).
13
3.1. Conection coefficients
In the coordinate basis the connection coefficients are given by (26). Now, let {êα } be the non-coordinate basis and
{θ̃β } the dual basis, thus the connection coefficients with respect to the non-coordinates is obtained in the same way
that it is obtained in coordinate basis (25). Thus ∇êβ êγ = Γα βγ êα where it get

Γα βγ = hθ̃α , ∇êβ êγ i, (77)

where we can use equations (55) e (56), such that the connection coefficients equation follows

Γα βγ = hωα µ dxµ , ∇(eβ ν Eν ) eγ ρ Eρ i. (78)

Let us use the properties (23) of covariant derivative, ∇ f X Y = f ∇ X Y, where we can obtain from equation (78) the
following result
Γα βγ = hωα µ dxµ , eβ ν ∇(Eν ) eγ ρ Eρ i, (79)
and now we can use the other propertie of covariant derivative (24), ∇ X ( f Y) = X[ f ]Y + f ∇ X Y, to rewrite the equation
(79) as
Γα βγ = hωα µ dxµ , eβ ν (Eν [eγ ρ ]Eρ + eγ ρ ∇(Eν ) Eρ )i. (80)
From identity (25), where ∇(Eν ) Eρ = Γσ νρ Eσ , we can calculate the scalar product

Γα βγ = ωα µ eβ ν (∂ν [eγ ρ ]δµ ρ + eγ ρ Γσ νρ δµ σ ), (81)

that results in
µ
Γα βγ = ωα µ eβ ν (∂ν eγ µ + eγ ρ Γνρ ), (82)
or
Γα βγ = ωα µ eβ ν ∇ν eγ µ . (83)
In addition, it is important to observe that the νth component of covariant derivative of vierbein field eγ µ acts only on
indice µ of coordinate basis {Eµ }.
We have seen in equation (28) that ∇ Eµ dxρ = −Γρ µσ dxσ , similarly we have

Γγ βα θ̃α = −∇(e˜β ) θ̃γ , (84)

where the connection coefficients in term of (55) e (56) reduce to,

Γα βγ = −heγ , ∇(eβ ) θα i
= −heγ µ Eµ , ∇(eβ ν Eν ) (ωα ρ dxρ )i
= −heγ µ Eµ , eβ ν ∇ Eν (ωα ρ dxρ )i
= −heγ µ Eµ , eβ ν (Eν [ωα ρ ]dxρ + ωα ρ ∇ Eν dxρ )i
= −heγ µ Eµ , eβ ν (∂ν ωα ρ dxρ − ωα ρ Γρνσ dxσ )i
= −heγ µ Eµ , eβ ν (∂ν ωα ρ − ωα σ Γσνρ )dxρ i
= −eγ µ eβ ν ∇ν ωα ρ δµ ρ , (85)

such that
Γα βγ = −eγ µ eβ ν ∇ν ωα µ . (86)
It is important to note that from (83),
Γα βγ = eβ ν ωα µ ∇ν eγ µ ,
| {z }
where the underbraced term can be replaced with aid of the identity (70), where we have that ωα µ ∇ν eγ µ = −eγ µ ∇ν ωα µ ,
and consequently the connection coefficients are given by (86).

14
4. Torsion tensor in non-coordinate basis and the first Cartan equation
The first structure equation of Cartan can be obtained from definition of torsion tensor (32),
T(X, Y) = ∇ X Y − ∇Y X − [X, Y],
where the components of torsion tensor in non-coordinate basis follow that
T α βγ = hθ̃α , T (êβ , êγ )i,
such that
T α βγ = hθ̃α , ∇êβ êγ − ∇êγ êβ − [êβ , êγ ]i.
With aid of
∇êβ êγ = Γδ βγ êδ
and from (74) where
[êβ , êγ ] = Dδ βγ êδ ,
the torsion tensor becomes
T α βγ = hθ̃α , Γδ βγ êδ − Γδ γβ êδ − Dδ βγ êδ i
from which we find
T α βγ = Γα βγ − Γα γβ − Dα βγ . (87)
It is straightforward to notice that the difference between torsion tensor in coordinate basis of equation (34) and torsion
tensor in non-coordinate basis of equation (87) is the presence of commutator coefficient, Dα βγ .
We can obtain an expression that make a transition of torsion tensor, T µ νρ , from coordinate basis to torsion tensor
α
T βγ in non-coordinate basis. From equation (87) we have
T α βγ = Γα βγ − Γα γβ − Dα βγ  
= −ωα µ eγ ν ∇ν eβ µ + ωα µ eβ ν ∇ν eγ µ − ωα ν eβ µ ∂µ eγ ν − eγ µ ∂µ eβ ν
     
= −ωα µ eγ ν ∂ν eβ µ + eβ ρ Γµ νρ + ωα µ eβ ν ∂ν eγ µ + eγ ρ Γµ νρ − ωα ν eβ µ ∂µ eγ ν − eγ µ ∂µ eβ ν
= −ωα µ eγ ν eβρ Γµ νρ + ωα µ e β ν eγ ρ Γµ νρ
= ωα µ eβ ν eγ ρ Γµ νρ − Γµ ρν ,
we have from (34) that T µ νρ = Γµ νρ − Γµ ρν which it follows
T α βγ = ωα µ eβ ν eγ ρ T µ νρ . (88)
If a manifold is free of torsion we have from (87) that
Γα βγ − Γα γβ = Dα βγ (89)
α α α
or 2Γ [βγ] =D βγ . As a consequence, in coordinate basis we have that D βγ = 0 such that
α α
Γ βγ =Γ γβ . (90)
If the torsion tensor vanishes and from requirement that the metric tensor must be covariantly constant, ∇α gβγ = 0,
we have that the cycle permutation of (α, β, γ) yields
∇α gβγ = ∂α gβγ − Γγαβ − Γβαγ = 0
∇β gαγ = ∂β gαγ − Γγβα − Γαβγ = 0
∇γ gαβ = ∂γ gαβ − Γβγα − Γαγβ = 0
Let us compute ∇α gβγ + ∇β gαγ − ∇γ gαβ , to obtain an expression for the connection coefficients
∂α gβγ + ∂β gαγ − ∂γ gαβ − Γγαβ − Γβαγ − Γγβα − Γαβγ + Γβγα + Γαγβ = 0.
We require that the torsion tensor is vanish, then from equation (89), Γαγβ − Γαβγ = Dαγβ , we have
∂α gβγ + ∂β gαγ − ∂γ gαβ − Γγαβ − Γγαβ + Γγαβ − Γγβα + Dβγα + Dαγβ = 0,
with Dαγβ = −Dαβγ and Dβγα = −Dβαγ , the above equation results in
1 
Γγαβ = ∂α gβγ + ∂β gαγ − ∂γ gαβ + Dγαβ − Dαβγ − Dβαγ . (91)
2
15
4.1. The first structure equation of Cartan
Let us introduce the torsion 2-form
1 α β
Tα = T βγ θ̃ ∧ θ̃γ . (92)
2
From equation (87) where we have
T α βγ = Γαβγ − Γαγβ − Dα βγ ,
we can calculate the torsion two-form (92) as it follows
1h α β i
Tα = Γ βγ θ̃ ∧ θ̃γ − Γα γβ θ̃β ∧ θ̃γ − Dα βγ θ̃β ∧ θ̃γ
2
1h α β i
= Γ βγ θ̃ ∧ θ̃γ + Γα γβ θ̃γ ∧ θ̃β − Dα βγ θ̃β ∧ θ̃γ
2
1h α β i
= 2Γ βγ θ̃ ∧ θ̃γ − Dα βγ θ̃β ∧ θ̃γ . (93)
2
We can compute the second term in above equation by using (75) that follows
1 α β 1 α  µ 
D βγ θ̃ ∧ θ̃γ = ω ν eβ ∂µ eγ ν − eγ µ ∂µ eβ ν θ̃β ∧ θ̃γ
2 2
= ωα ν eβ µ ∂µ eγ ν θ̃β ∧ θ̃γ . (94)
We can notice that
∂µ (eγ ν ωα ν ) = ∂µ (δγ α ) = 0,
where it results in
ωα ν ∂µ eγ ν = −eγ ν ∂µ ωα ν ,
hence, the equation (94) with above identity becomes
1 α β    
D βγ θ̃ ∧ θ̃γ = − ∂µ ωα ν eγ ν eβ µ θ̃β ∧ θ̃γ = − ∂µ ωα ν (eβ µ θ̃β ) ∧ (eγ ν θ̃γ ).
2
With aid of expression (61) it follows that
1 α β  
D βγ θ̃ ∧ θ̃γ = − ∂µ ωα ν (dxµ ) ∧ (dxν ). (95)
2
Now, if we compute an exterior derivative of expression (56) we have that
 
dθ̃α = ∂µ ωα ν dxµ ∧ dxν . (96)
Hence we have that the equation (95) becomes
1 α β
D βγ θ̃ ∧ θ̃γ = −dθ̃α , (97)
2
where it can be returned in expression (93), that results in
1h α β i  
Tα = 2Γ βγ θ̃ ∧ θ̃γ − Dα βγ θ̃β ∧ θ̃γ = Γα βγ θ̃β ∧ θ̃γ + dθ̃α . (98)
2
We have introduced the conection 1-form in equation (51) in coordinate basis, similarly we have the conection 1-form
in non-coordinate basis
Γα γ ≡ Γα βγ θ̃β . (99)
With this conection 1-form in equation of torsion 2-form, we arrive in
T α = dθ̃α + Γα β ∧ θ̃β first structure equation of Cartan . (100)
If the manifold is free of torsion then it follows that
dθ̃α = −Γα β ∧ θ̃β . (101)
For non-coordinate basis, the connection coefficients can be computed from (101).
16
5. Curvature tensor in non-coordinate basis and the second Cartan equation

Let us examine the curvature tensor given by (33)

R(X, Y)Z = ∇ X ∇Y Z − ∇Y ∇ X Z − ∇[X,Y] Z.

From this definition we can obtain the components of curvature tensor in non-coordinate basis with X = êγ , Y = êδ
and Z = êβ , such as
R(êγ , êδ )êβ = ∇êγ (∇δ êβ ) − ∇êδ (∇γ êβ ) − ∇[êγ ,êδ ] êβ ,
with ∇êβ êγ = Γα βγ êα and (74) it results in

R(êγ , êδ )êβ = ∇êγ (Γα δβ êα ) − ∇êδ (Γα γβ êα ) − ∇(Dǫ γδ êǫ ) êβ .

We can use covariant derivative properties (23) and (24),

∇ f X Y = f ∇ X Y,

for the third term and

∇ X ( f Y) = X[ f ]Y + f ∇ X Y
for the first and second terms. Thus

R(êγ , êδ )êβ = êγ [Γα δβ ]êα + Γα δβ ∇êγ êα − êδ [Γα γβ ]êα − Γα γβ ∇êδ êα − Dǫ γδ ∇êǫ êβ ,

that results in
R(êγ , êδ )êβ = êγ [Γα δβ ]êα − êδ [Γα γβ ]êα + Γα δβ Γǫ γα êǫ − Γα γβ Γǫ δα êǫ − Dǫ γδ Γα ǫβ êα ,
or n o
R(êγ , êδ )êβ = êγ [Γα δβ ] − êδ [Γα γβ ] + Γǫ δβ Γα γǫ − Γǫ γβ Γαδǫ − Dǫ γδ Γα ǫβ êα .
From equation (35) we have that
R(êγ , êδ )êβ = Rα βγδ êα ,
it follows that the components of curvature tensor with respect to non-coordinate basis is given by

Rα βγδ = êγ [Γα δβ ] − êδ [Γα γβ ] + Γǫ δβ Γα γǫ − Γǫ γβ Γαδǫ − Dǫ γδ Γα ǫβ

with (58) where we have êγ [Γα δβ ] = ∂γ Γα δβ

Rα βγδ = ∂γ Γα δβ − ∂δ Γα γβ + Γǫ δβ Γα γǫ − Γǫ γβ Γα δǫ − Dǫ γδ Γα ǫβ . (102)

Furthermore, similar to torsion tensor, we have to notice that the principal difference between curvature tensor in
coordinate basis of equation (36) and curvature tensor in non-coordinate basis of equation (102) is the presence of
commutator coefficient, Dǫ γδ .

5.1. The second structure equation of Cartan

Similarly as we have worked with the torsion 2-form, we can introduce the curvature 2-form
1 α
Θα β = R βγδ θ̃γ ∧ θ̃δ . (103)
2
Let us substitute Rα βγδ from equation (102) into curvature two-form

1h α γ i
Θα β = ∂γ Γ δβ θ̃ ∧ θ̃δ − ∂δ Γα γβ θ̃γ ∧ θ̃δ + Γǫ δβ Γα γǫ θ̃γ ∧ θ̃δ − Γǫ γβ Γα δǫ θ̃γ ∧ θ̃δ − Dǫ γδ Γα ǫβ θ̃γ ∧ θ̃δ ,
2

17
with θ̃γ ∧ θ̃δ = −θ̃δ ∧ θ̃γ and from equation (89) when the manifold is free of torsion, where Γα βγ − Γα γβ = Dα βγ , we
have
1h α γ i
Θα β = ∂γ Γ δβ θ̃ ∧ θ̃δ − ∂δ Γα γβ (−θ̃δ ∧ θ̃γ ) + Γǫ δβ Γα γǫ θ̃γ ∧ θ̃δ − Γǫ γβ Γα δǫ (−θ̃δ ∧ θ̃γ ) − (Γǫ γδ − Γǫ δγ )Γα ǫβ θ̃γ ∧ θ̃δ ,
2
that it follows
1h i
Θα β = 2 ∂γ Γα δβ θ̃γ ∧ θ̃δ + 2 Γǫ δβ Γα γǫ θ̃γ ∧ θ̃δ − Γα ǫβ Γǫ γδ θ̃γ ∧ θ̃δ + Γα ǫβ Γǫ δγ (−θ̃δ ∧ θ̃γ ) ,
2
or

Θα β = ∂γ Γα δβ θ̃γ ∧ θ̃δ + Γǫ δβ Γα γǫ θ̃γ ∧ θ̃δ − Γα ǫβ Γǫ γδ θ̃γ ∧ θ̃δ

= ∂γ Γα δβ θ̃γ ∧ θ̃δ + (Γα γǫ θ̃γ ) ∧ (Γǫ δβ θ̃δ ) − Γα ǫβ (Γǫ γδ θ̃γ ) ∧ θ̃δ .

From definition from conection 1-form from (99), we have

Θα β = ∂γ Γα δβ θ̃γ ∧ θ̃δ + Γα ǫβ θ̃δ ∧ Γǫ δ +Γα ǫ ∧ Γǫ β . (104)

| {z }

The underbraced terms can be simplifyied by use of exterior derivative rule (20). If f is a 0-form and ξ is one-form
we have that
d( f ∧ ξ) = (d f ) ∧ ξ + f ∧ dξ, (105)
then
dΓα β = d(Γα γβ θ̃γ ) = (∂δ Γα γβ )θ̃δ ∧ θ̃γ + Γα γβ dθ̃γ .
In a manifold free of torsion we have from first Cartan equation (101), dθ̃γ = −Γγ β ∧ θ̃β , where it can be inserted into
above equation that follows

dΓα β = (∂δ Γα γβ )θ̃δ ∧ θ̃γ + Γα γβ (−Γγ δ ∧ θ̃δ ) = (∂γ Γα δβ )θ̃γ ∧ θ̃δ + Γα γβ θ̃δ ∧ Γγ δ .

Now we can change the underbraced terms from equation (104) by above result, that it results in the following
curvature two-form

Θα β = dΓα β + Γα γ ∧ Γγ β second structure equation of Cartan (106)

The second structure equation of Cartan is an efficient method for calculating the components of curvature tensor
(102) with respect to a non-coordinate basis. The second Cartan’s equation gives an algororithm for the calculation of
curvature from the conection coefficients Γα β .
We have seen that the second exterior derivative of an r-form is zero, d 2 Γα β = 0, thus in this sense we can write
the Bianchi identity for the curvature 2-form,

dΘα β − Θα γ ∧ Γγ β + Γα γ ∧ Θγ β = 0.

with aid of exterior derivative property (20).

6. Local Lorentz transformations

In an m-dimensional Riemannian manifold, the metric tensor gµν has m(m+1)

2 degrees of freedom, in 4-dimensional
spacetime it has 10 degrees of freedom. While the vierbein field eα µ has m2 degrees of freedom, in 4-dimensional
spacetime it has 16 degrees of freedom. There are many non-coordinate bases which yield the same metric g, for
example, the the two main non-coordinate bases are (i) orthonormal basis and (ii) pseudo-orthonormal basis. Each of
them is related to the other by local rotation,

θ̃′α (p) = Λα β (p)θ̃β (p) (107)

18
at each point p. From equation (56), where we have,

θ̃α = ωα µ dxµ ,

and where it yields

θ̃′α (p) = ω′α µ dxµ = Λα β (p)θ̃β (p)
or
ω′α µ dxµ = Λα β ωβ µ dxµ
where it results in
ω′α µ = Λα β ωβ µ . (108)
Under the local transformation, Λα β , the indices α, β · · · rotated while µ, ν · · · , from coordinate basis (world indices)
are not affected.
The basis vector of non-coordinate basis in a referential frame O is given by

êα = eα µ Eµ (109)

while
ê′α = e′α µ Eµ (110)
is given in the referential frame O′ . From expression (108) where ω′α µ = Λα β ωβ µ and from identity (59) where
eα µ ωβ µ = δα β , we can obtain

e′α µ ω′β µ = δα β
e′α µ Λβ γ ωγ µ = δα β (111)

it can be multiplied by (Λ−1 )δ β , such that it follows

(Λ−1 )δ β Λβ γ e′α µ ωγ µ = (Λ−1 )δ β δα β

δδ γ ωγ µ e′α µ = (Λ−1 )δ α
ωδ µ e′α µ = (Λ−1 )δ α (112)

and multiplying by eδ ν we have that

eδ ν ωδ µ e′α µ = eδ ν (Λ−1 )δ α
δν µ e′α µ = eδ ν (Λ−1 )δ α

that results in
e′α µ = eβ µ (Λ−1 )β α . (113)
Now, we can return to equation (110), where we have

ê′α = e′α µ Eµ = eβ µ (Λ−1 )β α Eµ .

In local frame O, from (109), where êα = eα µ Eµ and from (60) where Eµ = ωγ µ êγ , the above equation yields

ê′α = eβ µ (Λ−1 )β α ωγ µ êγ

= eβ µ ωγ µ êγ (Λ−1 )β α êγ
= δβ γ (Λ−1 )β α êγ

so that
ê′α = (Λ−1 )β α êβ . (114)
Let a tensor field of type (1,1) given by T = T α β êα ⊗ θ̃β in local frame O. In any rotated local frame O′ it is given
by
β
T ′ = T ′α β ê′α ⊗ θ˜′ .
19
It is possible to obtain a rule how this tensor transforms by use of equation (107) where θ̃′α = Λα β θ̃β and the equation
(114) where ê′α = (Λ−1 )β α êβ , with respective inverses

θ̃α = (Λ−1 )α β θ̃′β and êα = Λβ α ê′β . (115)

Thus the tensor T = T α β êα ⊗ θ̃β in a local frame O is related to the other by local rotation by

T = T α β êhα ⊗ θ̃β i h i
= T α β Λγ α ê′γ ⊗ (Λ−1 )β δ θ̃′δ
h i
= Λγ α T α β (Λ−1 )β δ ê′γ ⊗ θ̃′δ

from which we find the transformation rule,

T ′α β = Λα γ T γ δ (Λ−1 )δ β . (116)

This transformation rule for tensor field of type (1,1) can be written in tensor form,

T ′ = ΛTΛ−1 . (117)

The upper non-coordinate indices are rotated by Λ while lower non-coordinate indices are rotated by Λ−1 . However
not all objects with indices are the components of a tensor, an important example is provided by the transformation
properties of the connection coefficients Γα βγ determined by (77). The connection coefficients under the rotation is

Γ′ α βγ = hθ̃′α , ∇ê′β ê′γ i = hΛα δ θ̃δ , ∇[(Λ−1 )ǫ β êǫ ] (Λ−1 )ζ γ êζ i,

withe properties (23), (24) and (58) it follows that

Γ′ α βγ = Λα δ hθ̃δ , (Λ−1 )ǫ β ∇êǫ (Λ−1 )ζ γ êζ i

h i
= Λα δ (Λ−1 )ǫ β hθ̃δ , ∂ǫ (Λ−1 )ζ γ êζ + (Λ−1 )ζ γ Γη ǫζ êη i
h i
= Λα δ (Λ−1 )ǫ β ∂ǫ (Λ−1 )ζ γ δδ ζ + (Λ−1 )ζ γ Γη ǫζ δδ η
= Λα δ (Λ−1 )ǫ β ∂ǫ (Λ−1 )δ γ + Λα δ (Λ−1 )ǫ β (Λ−1 )ζ γ Γδ ǫζ ,

Because of the first term on the right-hand side of above equation, the connection coefficients do not transform as the
components of a tensor. We can see how to transform the connection 1-form Γα γ by calculate Γ′ α βγ θ̃′β ,

Γ′ α βγ θ̃′β = Λα δ (Λ−1 )ǫ β θ̃′β ∂ǫ (Λ−1 )δ γ + Λα δ (Λ−1 )ζ γ Γδ ǫζ (Λ−1 )ǫ β θ̃′β ,

with transformation (Λ−1 )α β θ̃′β = θ̃α from (115) it follows that

Γ′ α γ = Λα δ θ̃ǫ ∂ǫ (Λ−1 )δ γ + Λα δ (Λ−1 )ζ γ Γδ ǫζ θ̃ǫ ,

we can use (62), where θ̃ǫ ∂ǫ (Λ−1 )δ γ = d(Λ−1 )δ γ , to reduce the above equation to

Γ′ α γ = Λα δ d(Λ−1 )δ γ + Λα δ Γδ ζ (Λ−1 )ζ γ . (118)

Also, the above transformation rule of the connection 1-form Γα β , can be obtained by rotate the torsion 2-form from
first Cartan equation (100),
T ′α = dθ̃′α + Γ′α β ∧ θ̃′β .
On the left-hand side of the above equation we have T ′α = Λα γ T γ , while on the right-hand side we can apply (107),

 Λα γ T γ = d(Λα γ θ̃γ ) + Γ′α β ∧ (Λβ γ θ̃γ )

Λα γ dθ̃γ + Γγ β ∧ θ̃β = dΛα γ ∧ θ̃γ + Λα γ dθ̃γ + Γ′α β ∧ Λβ γ θ̃γ
Λα γ Γγ β ∧ θ̃β = dΛα γ ∧ θ̃γ + Γ′α β ∧ Λβ γ θ̃γ .
20
We can isolate Γ′α β to obtain
Γ′α β Λβ γ ∧ θ̃γ = Λα β Γβ γ ∧ θ̃γ − dΛα γ ∧ θ̃γ
where it reduces to
Γ′α β Λβ γ = Λα β Γβ γ − dΛα γ ,
and multiplying both sides by Λ−1 from the right, we have
Γ′α β Λβ γ (Λ−1 )γ δ = Λα β Γβ γ (Λ−1 )γ δ − (dΛα γ )(Λ−1 )γ δ .
We can use
Γ′α β Λβ γ (Λ−1 )γ δ = Γ′α β δβ δ = Γ′α δ
and
h i
d Λα γ )(Λ−1 )γ δ = (dΛα γ )(Λ−1 )γ δ + Λα γ d(Λ−1 )γ δ ,
dδα δ = (dΛα γ )(Λ−1 )γ δ + Λα γ d(Λ−1 )γ δ
where it follows that
(dΛα γ )(Λ−1 )γ δ = −Λα γ d(Λ−1 )γ δ . (119)
′α
Thus we can express Γ β as
Γ′α β = Λα γ Γγ δ (Λ−1 )δ β + Λα γ d(Λ−1 )γ β . (120)
Thus, the above result is the same one obtained in equation (118).
It is easy to notice from expression (116), that the curvature two-form transforms as
Θ′α β = Λα γ Θγ δ (Λ−1 )δ β , (121)
under a local frame transformation Λ.

7. The non-coordinate orthonormal basis

General Relativity is based on the concept of spacetime, which is a 4-dimensional differentiable manifold M
with Lorentzian metric,
( ) by (M, g) where the components of the metric tensor g of spacetime with respect a
denoted
∂
coordinate basis Eµ = µ are given by (39),
∂x
g(Eµ , Eν ) = gµν .
However, for many purposes it is more convenient to use a non-coordinate orthonormal basis, or orthonormal tetrad
or Lorentz frame {êα }, where we have:
g(êα , êβ ) = ηαβ , (122)
where (ηαβ ) = diag(−1, 1, 1, 1) are the components of metric tensor of Minkowski reference system. The relationship
between coordinate basis with non-coordinate orthonormal basis is given by (66), where
g(êα , êβ ) = eα µ eβ ν gµν ,
that follows that
ηαβ = eα µ eβ ν gµν . (123)
In the inverse way it results that
gµν = ωα µ ωβ ν ηαβ .
The metric tensor given by (64) is
g = ηαβ θ̃α ⊗ θ̃β
= −(θ̃0 ⊗ θ̃0 ) + (θ̃1 ⊗ θ̃1 ) + (θ̃2 ⊗ θ̃2 ) + (θ̃3 ⊗ θ̃3 ). (124)
The metric tensor in non-coordinate orthonormal basis is a system of locally inertial coordinates or local Lorentz
spacetime.
21
7.1. The Ricci rotation coefficients
In the non-coordinate orthonormal basis the Minkowskian metric tensor is constant and consequently this local
reference system is a rigid frame with ∂γ ηαβ = 0, then we can review equation (91) where we have,
1 
Γγαβ = Dγαβ − Dαβγ − Dβαγ . (125)
2
we can interchange γ ↔ β into above equation such as
1 
Γβαγ = Dβαγ − Dαγβ − Dγαβ
2
1 
= − Dγαβ + Dαγβ − Dβαγ ,
2
where we can use equation (76) for the second term in right hand side of above equation, Dαγβ = −Dαβγ , that it
follows,
1 
Γβαγ = − Dγαβ − Dαβγ − Dβαγ .
2
Now we can compare above equation with (125) and obtain that
Γγαβ = −Γβαγ . (126)
The connection coefficients in non-coordinate orthonormal basis also referred as the Ricci rotation coefficients [5,
8, 12]. The antisymmetry of the Ricci rotation coefficients in the index pair (βγ) yields n2 (n−1) 2 , in 4-dimensional
spacetime V 4 , we have 24 components Γγαβ .
The matrix of Ricci rotation coefficients (Γαγβ ) is 4 × 4 × 4 matrix and we can see this in 4 layers
 
 0 Γ001 Γ002 Γ003 

−Γ 0 Γ102 Γ103 
Γα0β =  001 
−Γ002 −Γ102 0 Γ203 
−Γ003 −Γ103 −Γ203 0
 
 0 Γ011 Γ012 Γ013 

−Γ 0 Γ112 Γ113 
Γα1β =  011 
−Γ012 −Γ112 0 Γ213 
−Γ013 −Γ113 −Γ213 0
 
 0 Γ021 Γ022 Γ023 

−Γ 0 Γ122 Γ123 
Γα2β =  021 
−Γ022 −Γ122 0 Γ223 
−Γ023 −Γ123 −Γ223 0
 
 0 Γ031 Γ032 Γ033 

−Γ 0 Γ132 Γ133 
Γα3β =  031 
−Γ032 −Γ132 0 Γ233 
−Γ033 −Γ133 −Γ233 0
We can see the 24 terms, 6 in each matrix layer.
In a coordinate basis the Ricci rotation coefficients are substituted by the Christoffel symbols Γµ νρ that are sym-
metric in the index pair (νρ), Γµ νρ = Γµ ρν . It yields n2 (n+1) 4
2 terms, where in in 4-dimensional manifold V , there are 40
α
components Γ βγ . Thus, we have advantage in use of non-coordinate orthonormal basis for calculating curvature. The
Cartan’s method for the calculation of curvature is more compact and efficient in many applications. The calculation
of the connection 1-forms (51) are obtained from (126),
Γαγβ θ̃γ = −Γβγα θ̃γ
Γαβ = −Γβα , (127)
with the first Cartan equation determine Γαβ uniquely. In a non-coordinate orthonormal basis in 4-dimensional mani-
fold, at most six independent connection 1-forms survive.
22
7.2. The first equation of Cartan
In a non-coordinate orthonormal basis the first Cartan equation yields a system of four equations




 −dθ̃0 = Γ0 1 ∧ θ̃1 + Γ0 2 ∧ θ̃2 + Γ0 3 ∧ θ̃3


−dθ̃1 = Γ1 0 ∧ θ̃0 + Γ1 2 ∧ θ̃2 + Γ1 3 ∧ θ̃3


 (128)




 −dθ̃2 = Γ2 0 ∧ θ̃0 + Γ2 1 ∧ θ̃1 + Γ2 3 ∧ θ̃3

−dθ̃3 = Γ3 0 ∧ θ̃0 + Γ3 1 ∧ θ̃1 + Γ3 2 ∧ θ̃2 .


There are 12 Ricci rotation coefficients in above system, but with the antisymmetry of the connection 1-forms (127),
Γβα = −Γαβ the Ricci rotation coefficients are reduced to 6. Let α = 0 (timelike) and β = i = 1, 2, 3 (spacelike), where
we have Γ0i = −Γi0 , thus,
Γ0 i = η00 Γ0i = (−1)Γ0i or Γ0 i = Γi0 . (129)
Moreover, we have that
Γi 0 = ηi j Γ j0 = δi j Γ j0 , (130)
for i = 1, it yields
Γ1 0 = Γ10 , (131)
0
with respect to (129) where Γ 1 = Γ10 we have that

Γ1 0 = Γ0 1 . (132)

The next step, let α = 1 e β = 2 (both spacelike), and from (127), we have Γ12 = −Γ21 ,

Γ1 2 = η1k Γk2 = δ11 Γ12 = Γ12 . (133)

For Γ21 we have

Γ2 1 = η2k Γk1 = δ22 Γ21 = Γ21 = −Γ12 , (134)
comparing the two above expression we have that

Γ1 2 = −Γ2 1 . (135)

With these results we can write the system of equations (128) as





 −dθ̃0 = Γ0 1 ∧ θ̃1 + Γ0 2 ∧ θ̃2 + Γ0 3 ∧ θ̃3


−dθ̃1 = Γ0 1 ∧ θ̃0 + Γ1 2 ∧ θ̃2 + Γ1 3 ∧ θ̃3


 (136)




 −dθ̃2 = Γ0 2 ∧ θ̃0 − Γ1 2 ∧ θ̃1 + Γ2 3 ∧ θ̃3

−dθ̃3 = Γ0 3 ∧ θ̃0 − Γ1 3 ∧ θ̃1 − Γ2 3 ∧ θ̃2


with six independent connection 1-forms: Γ0 1 , Γ0 2 , Γ0 3 , Γ1 2 , Γ1 3 e Γ2 3 to be determined.

7.3. The second equation of Cartan

For the calculation of curvature we use the second Cartan equation from (106),

Θαβ = dΓαβ + Γαγ ∧ Γγ β

where the curvature 2-forms Θαβ are antisymmetric in the index pair αβ. From (127) we have that

Θαβ = −dΓβα − Γγα ∧ Γγ β = −dΓβα − Γγ α ∧ Γγβ = −dΓβα + Γγβ ∧ Γγ α = −dΓβα − Γβγ ∧ Γγ α ,

it follows that
Θαβ = −Θβα . (137)

23
The antisymmetry of the curvature 2-forms, Θαβ , yields a system of six equations




Θ01 = dΓ01 + Γ02 ∧ Γ2 1 + Γ03 ∧ Γ3 1

= dΓ02 + Γ01 ∧ Γ1 2 + Γ03 ∧ Γ3 2




Θ02


Θ03

 = dΓ03 + Γ01 ∧ Γ1 3 + Γ02 ∧ Γ2 3




Θ12 = dΓ12 + Γ10 ∧ Γ0 2 + Γ13 ∧ Γ3 2






Θ13 = dΓ13 + Γ10 ∧ Γ0 3 + Γ12 ∧ Γ2 3

= dΓ23 + Γ20 ∧ Γ0 3 + Γ21 ∧ Γ1 3

Θ23

We have seen that there are six independent connection 1-forms: Γ0 1 , Γ0 2 , Γ0 3 , Γ1 2 , Γ1 3 e Γ2 3 to be determined. With
conditions (132), Γ1 0 = Γ0 1 = −Γ01 , and (135), Γ1 2 = −Γ2 1 = Γ12 , the above system of equations from second Cartan
equation becomes 



 Θ01 = dΓ01 − Γ02 ∧ Γ12 − Γ03 ∧ Γ13





 Θ02 = dΓ02 + Γ01 ∧ Γ12 − Γ03 ∧ Γ23


Θ03 = dΓ03 + Γ01 ∧ Γ13 + Γ02 ∧ Γ23


 (138)



 Θ12 = dΓ12 + Γ01 ∧ Γ02 − Γ13 ∧ Γ23






 Θ13 = dΓ13 + Γ01 ∧ Γ03 + Γ12 ∧ Γ23


Θ23 = dΓ23 + Γ02 ∧ Γ03 − Γ12 ∧ Γ13 .

Now let us see an example of Cartan’s formalism applied to General Relativity.

7.4. Example of Schwarzschild spacetime

When the non-null components of metric tensor g are only diagonal components, it is practical to work with the
non-coordinate orthonormal basis, as the case may be Schwarzschild spacetime. In coordinate basis, the metric of the
Schwarzschild spacetime is given by
!  
2 2M 2  1  2 2 2 2
ds = − 1 − dt +   dr + r (dθ + sin θdφ), (139)
r 1 − 2M
r

where the parameters run over the range r > 2M, 0 ≤ θ < π and 0 ≤ φ < 2π. The metric of the Schwarzschild
spacetime is already in diagonal form, thus it is easy to rewrite it in non-coordinate orthonormal basis in accordance
with (124)
ds2 = ηαβ θ̃α ⊗ θ̃β = −θ̃0 ⊗ θ̃0 + θ̃1 ⊗ θ̃1 + θ̃2 ⊗ θ̃2 + θ̃3 ⊗ θ̃3 .
From (56) we have that

θ̃0 = ω0 µ dxµ , θ̃1 = ω1 µ dxµ , θ̃2 = ω2 µ dxµ and θ̃3 = ω3 µ dxµ . (140)

and from (68) where gµν = ωα µ ωβ ν ηαβ , it follows that

gµν = ω0 µ ω0 ν η00 + ω1 µ ω1 ν η11 + ω2 µ ω2 ν η22 + ω3 µ ω3 ν η33

= −ω0 µ ω0 ν + ω1 µ ω1 ν + ω2 µ ω2 ν + ω3 µ ω3 ν ,

the diagonal terms of gµν

gtt = (ω0 t )2 , grr = (ω1 r )2 , gθθ = (ω2 θ )2 and gφφ = (ω3 φ )2

where the vierbein fields are interpreted as square root of metric and it results in
r
0 2M 1
ω t = 1− , ω1 r = q , ω2 θ = r and ω3 φ = r sin θ. (141)
r 2M
1− r

24
Thus the 1-form terms from (140) become
r
0 2M dr
θ̃ = dt 1 − , θ̃1 = q , θ̃2 = rdθ and θ̃3 = r sin θdφ. (142)
r 2M
1− r

The inverse terms are

r
θ̃0 1 2M θ̃2 θ̃3
dt = q , dr = θ̃ 1− , dθ = and dφ = . (143)
2M r r r sin θ
1− r

With (142) and (143) we can calculate 2-form d θ̃α ,

• dθ̃0 results in
M/r2 M/r2
dθ̃0 = q (dr ∧ dt) = q θ̃1 ∧ θ̃0 ,
2M 2M
1− r 1− r

• dθ̃1 results in
−M/r2
dθ̃1 =  3/2 (dr ∧ dr) = 0,
1 − 2M
r

• dθ̃2 results in q
2M
1− r
2
dθ̃ = dr ∧ dθ = θ̃1 ∧ θ̃2 ,
r
• dθ̃3 results in q
2M
1− r cot θ 2
3
dθ̃ = sin θdr ∧ dφ + r cos θdθ ∧ dφ = θ̃1 ∧ θ̃3 + θ̃ ∧ θ̃3 .
r r
With these results we can solve the first structure equation of Cartan as the system of equations (136),

M/r2
−q θ̃1 ∧ θ̃0 = Γ0 1 ∧ θ̃1 ,
2M
1− r
q 0 = Γ0 1 ∧ θ̃0 + Γ1 2 ∧ θ̃2 + Γ1 3 ∧ θ̃3 ,
2M
1− r
− θ̃1 ∧ θ̃2 = −Γ1 2 ∧ θ̃1 ,
q r
2M
1− r cot θ 2
− θ̃1 ∧ θ̃3 − θ̃ ∧ θ̃3 = −Γ1 3 ∧ θ̃1 − Γ2 3 ∧ θ̃2 .
r r
Thus, we have four independent connection 1-forms
M
Γ0 1 = q θ̃0 = −Γ01 ,
2M
r2 1− r
q
1 − 2M
r
Γ1 2 = − θ̃2 = Γ12 ,
q r
2M
1− r
Γ1 3 = − θ̃3 = Γ13 ,
r
cot θ 3
Γ2 3 = − θ̃ = Γ23 ,
r
25
with Γ0 2 = Γ0 3 = 0. Now we can calculate dΓα β with objective to solve the second structure equation of Cartan. The
calculation of 2-form dΓ0 1 follows that
 
 2

 2M M · 2M/r  M
dΓ0 1 = − q 0
dθ̃0

−    dr ∧ θ̃ + q
 r3 1 − 2M 2r2 1 − 2M 3/2  r2 1 − r 2M
r r
 r
M  M  2M 1 M M
= − q    + 2 1 − θ̃ ∧ θ̃0 + q · q θ̃1 ∧ θ̃0
r 1− r 2M r
r3 1 − 2M r r2 1 − 2Mr r2 1 − 2M
r
 
 M  M  2
M 

= − 3 +2 +    θ̃1 ∧ θ̃0
r r − 2M 4
r 1− r 2M 

2M 0
= θ̃ ∧ θ̃1
r3
or
2M 0
dΓ01 = − θ̃ ∧ θ̃1 (144)
r3
The calculation of 2-form dΓ1 2 follows that
q  q
 1 − 2M  1 − 2M
 r  r
dΓ1 2 = −d   ∧ θ̃2 − dθ̃2

 r  r
  q q
 r 2M 2M
2M 2 2M/r 2
 1 − r 1− r
 
 dr ∧ θ̃2 −
=  1 − r − q · θ̃1 ∧ θ̃2
 r  r r
2r 1 − 2M r
 q  
 1 − 2M  r 2M 
2M 1 − r  1

 r M 
=  − q 
 1 − −  θ̃ ∧ θ̃2
 r2 3 2M 
 r r2 
r 1− r
M 1
= − 3 θ̃ ∧ θ̃2
r
or
M 1
dΓ12 = − θ̃ ∧ θ̃2 . (145)
r3
The calculation of 2-form dΓ1 3 follows that
q
2M

 1 − r 2M 
M  1 1− r
1 3
dΓ 3 =  − 
 θ̃ ∧ θ̃ − dθ̃3
r2 r3 r
q q 
2M  2M

 1 − 2M

M  1 1−  1 −
r  r cot θ 2


r
=  2
− 3
 θ̃ ∧ θ̃3 − 
 θ̃1 ∧ θ̃3 + θ̃ ∧ θ̃3  ,
r r r  r r 

where it reduces to q
2M
M 1− r
dΓ13 = − 3 θ̃1 ∧ θ̃3 − cot θ θ̃2 ∧ θ̃3 . (146)
r r2
The calculation of 2-form dΓ2 3 follows that
 cot θ  cot θ 3
dΓ2 3 = −d ∧ θ̃3 − dθ̃ ,
r r

26
where it results is
1 2
dΓ23 = − θ̃ ∧ θ̃3 . (147)
r2
The curvature 2-forms are founf from the second structure Cartan equation (138).
2M
Θ01 = dΓ01 − Γ02 ∧ Γ12 − Γ03 ∧ Γ13 = − 3 θ̃0 ∧ θ̃1 ,
r
M 0
Θ02 = dΓ02 + Γ01 ∧ Γ12 − Γ03 ∧ Γ23 = 3 θ̃ ∧ θ̃2 ,
r
M
Θ03 = dΓ03 + Γ01 ∧ Γ13 + Γ02 ∧ Γ23 = 3 θ̃0 ∧ θ̃3 ,
r
M
Θ12 = dΓ12 + Γ01 ∧ Γ02 − Γ13 ∧ Γ23 = − 3 θ̃1 ∧ θ̃2 ,
r
M 1
Θ13 = dΓ13 + Γ01 ∧ Γ03 + Γ12 ∧ Γ23 = − 3 θ̃ ∧ θ̃3 ,
r
2M
Θ23 = dΓ23 + Γ02 ∧ Γ03 − Γ12 ∧ Γ13 = 3 θ̃2 ∧ θ̃3 .
r
We have introduced the curvature two-form given by equation (103),
1
Θαβ = Rαβγδ θ̃γ ∧ θ̃δ .
2
Thus we can obtain the Riemann curvature tensor in non-coordinate orthonormal basis. For
2M 0
Θ01 = − θ̃ ∧ θ̃1 ,
r3
we have
1 1 2M
R0101 θ̃0 ∧ θ̃1 + R0110 θ̃1 ∧ θ̃0 = − 3 θ̃0 ∧ θ̃1 ,
2 2 r
that results in
2M
R0101 = − .
r3
In the same way we have for others components of Riemann tensor,
M
R0202 = ,
r3
M
R0303 = ,
r3
M
R1212 = − 3,
r
M
R1313 = − 3,
r
2M
R2323 = .
r3
Now we can compute the components of Ricci tensor

Rβδ = Rγ βγδ = ηαγ Rαβγδ .

The components of above Riemann tensor show that we have readily Rβδ = 0 with β , δ, however it is possible to
compute R00 , R11 , R22 and R33 , but they are null as it follows
2M M M
R00 = R1 010 + R2 020 + R3 030 = − + 3 + 3 = 0,
r3 r r
2M M M
R11 = R0 101 + R2 121 + R3 131 = + − 3 − 3 = 0,
r3 r r
27
 M M 2M
R22 = R0 202 + R1 212 + R3 232 = − − + 3 = 0,
r3 r3 r
M M 2M
R33 = R0 303 + R1 313 + R2 323 = − 3 − 3 + 3 = 0.
r r r
Thus Rαβ = 0 and consequently we have that the Einstein tensor Gαβ = Rαβ − 21 gαβ R = 0, that is a vacuum solution.
In coordinate basis or world indices we can obtain that

Rµν = ωα µ ωβ ν Rαβ (148)

where clearly Rµν = 0. However we can se that the non null components of Riemann tensor in coordinate basis are
obtained from
Rµνρσ = ωα µ ωβ ν ωγ ρ ωδ σ Rαβρσ , (149)
where with aid of (141). It follows that
 2  2 2M ! 2M
Rtrtr = ω0 t ω1 r ω0 t ω1 r R0101 = ω0 t ω1 r − 3 = − 3 ,
r r
 2  2  M  ! !
2M 2 M M 2M
Rtθtθ = ω0 t ω2 θ ω0 t ω2 θ R0202 = ω0 t ω2 θ = 1 − r = 1 − ,
r3 r r3 r r
   ! !
0 3 0 3
 2 
0 2 2 M 2M  2 2  M M 2M
Rtφtφ = ω t ω φ ω t ω φ R0303 = ω t ω φ = 1− r sin θ 3 = 1− sin2 θ,
r3 r r r r
  !−1
1 2 1 2
 2  2  −M 
1 2
 r2  M M 2M
Rrθrθ = ω r ω θ ω r ω θ R1212 = ω r ω θ = − 
 
 3 = − 1 − ,
r3 1 − 2M r
r r r
 2 2  !−1
  
3 2 −M  r sin θ  M M 2M
 2 
1 3 1 3 1
Rrφrφ = ω r ω φ ω r ω φ R1313 = ω r ω φ = −   3 = − 1− sin2 θ,
r3 1 − 2M r
r r r
 2  2 2M !   2M
Rθφθφ = ω2 θ ω3 φ ω2 θ ω3 φ R2323 = ω2 θ ω3 φ 3
= r4 sin2 θ 3 = 2Mr sin2 θ.
r r
After some practices, we can see that the non-coordinate orthonormal basis to advance in an easy method to
calculate the Riemann tensor of spacetimes that have the diagonal shape in the metric tensor g in coordinate basis.

8. Newman-Penrose null tetrad - pseudo-orthonormal basis

A pseudo-orthonormal basis or null tetrad formalism is due to Newman and Penrose [14]. It has proved very useful
in the construction of exact solutions on General Relativity. This formalism is adapted to treatment of the propagation
of radiation in spacetime as will be seen at the end of this section with the example of Brinkmann metric.
We can start with metric of Minkowski spacetime in orthonormal basis where the line element is

ds2 = −dt2 + dx2 + dy2 + dz2 .

An alternative coordinate system for Minkowski spacetime can be through the introduction of advanced and retarded
null coordinates, u and v with the following coordinate transformations
1 1
u = √ (t − x) and v = √ (t + x) , (150)
2 2
so that the line element of Minkowski spacetime can be

ds2 = −2dudv + dy2 + dz2 .

28
With these null coordinates, we can see that if u is constant, we have du = 0, then dx = dt or dx
dt = 1, the speed of light.
The null coordinate u may be thought of as light ray at the speed of light in direction +x, while the null coordinate v
may be thought of as light ray at the speed of light in direction −x. These null coordinates, v and u are orthogonal to
the yz-plane. In the similar way, it is convenient to parametrise the yz-plane in terms of the complex coordinate
1 1
ζ = √ (y + iz) , ζ̄ = √ (y − iz) , (151)
2 2

thus we can see that in terms of these coordinate transformations the line element ds2 of Minkowski spacetime is

ds2 = −2dudv + 2dζdζ̄. (152)

Now we can express the 1-form terms of ds2 as

θ̃0 = dv, θ̃1 = du, θ̃2 = dζ, θ̃3 = d ζ̄,

where we have the metric tensor g in pseudo-orthonormal coordinate basis given by

g = −2θ̃0 ⊗ θ̃1 + 2θ̃2 ⊗ θ̃3 (153)

that in matrix form is  

0 θ̃0 

 0 −1 0
  −1 0 0

0 θ̃1 
g = θ̃0 θ̃1 θ̃2 θ̃3   
 0 0 0 1 θ̃2 
 
0 0 1 0 θ̃3
or
g = γαβ θ̃α ⊗ θ̃β (154)
since  
 0 −1 0 0

−1 0 0 0
(γαβ ) =   (155)
 0 0 0 1
0 0 1 0
γαβ is defined the components of metric tensor with respect to the non-coordinate pseudo-orthonormal basis.
Again, from a tangent space T p spanned by {Eµ } in coordinate basis, we can obtain the non-coordinate pseudo-
orthonormal basis by a rotation. From equation (55) we have

êα = eα µ Eµ ,

Thus we have a tetrad,





 ê0 = e0 µ Eµ = kµ Eµ = k,


ê1

 = e1 µ Eµ = lµ Eµ = l,
 (156)




 ê2 = e2 µ Eµ = mµ Eµ = m,


ê3 = e3 µ Eµ = m̄µ Eµ = m̄.

The above set of complex null vectors {êα } = {k, l, m, m̄} is called a Newman-Penrose null tetrad. This is usually
abbreviated to NP null tetrad [13]. The reason for it designation is because the condition (67), where we have,

eα µ eβ ν gµν = γαβ .

It is straightforward to notice that the diagonal terms of metric tensor γαβ from (155) are zero. For Example

e0 µ e0 ν gµν = γ00 = 0,
29
where
kµ kν gµν = kµ kµ = 0.
And the same way the other three contractions are null,

lµ lµ = 0, mµ mµ = 0 and m̄µ m̄µ = 0

The non-null contractions are given by

e0 µ e1 ν gµν = γ01 ,
that results in kµ lµ = −1 and
e2 µ e3 ν gµν = γ23 ,
that results mµ m̄µ = 1.
The relationships (150) and (151), in flat spacetime implies that,
∂ ∂ ∂ ∂
k= , l= , m= and m̄ = , (157)
∂v ∂u ∂ζ ∂ζ̄
where we see that m and m̄ are complex conjugates. In coordinate basis we can see from (150) and (151) that,
       
1  1  0  0 
1 1 1 −1 1 0 1  0 
kµ = √   , lµ = √   , mµ = √   , and m̄µ = √   , (158)
2 0 2  0  2 1 2  1 
0 0 i −i

From equation (59)

hêα , θ̃β i = eα µ ωβ µ = δα β ,
we can identify that
hê0 , θ̃0 i = e0 µ ω0 µ = kµ ω0 µ = 1,
where from condition kµ lµ = −1, the above equation results for ω0 µ = −lµ . In the same way we have

e1 µ ω1 µ = lµ ω1 µ = 1,

that from kµ lµ = −1, it results in ω1 µ = −kµ . For the others terms we have

e2 µ ω2 µ = mµ ω2 µ = 1 ⇒ ω2 µ = m̄µ ,

and
e3 µ ω3 µ = m̄µ ω3 µ = 1 ⇒ ω3 µ = mµ .
So, as θ̃α = ωα µ dxµ , we have that the differential 1-forms, the basis for cotangent space in pseudo-orthonormal
coordinate, are 



 θ̃0 = ω0 µ dxµ = −lµ dxµ


θ̃1 = ω1 µ dxµ = −kµ dxµ


 (159)



 θ̃2 = ω2 µ dxµ = m̄µ dxµ


θ̃3 = ω3 µ dxµ = mµ dxµ


In coordinate basis, from (158) we have

       
−1 −1 0  0 
1  1 
  1 −1 1 0 1  0 
kµ = √   , lµ = √   , mµ = √   , and m̄µ = √   , (160)
2  0  2  0  2 1 2  1 
0 0 i −i

where we have the only nonzero contractions kµ lµ = −1 and mµ m̄µ = 1.

30
 Now we can see the way to write the components of metric tensor in coordinate basis from the terms lµ , kµ , mµ
and m̄µ . Recall equation (68),
gµν = ωα µ ωβ ν γαβ ,
it results in

gµν = ω0 µ ω1 ν γ01 + ω1 µ ω0 ν γ10 + ω2 µ ω3 ν γ23 + ω3 µ ω2 ν γ32

= (−lµ )(−kν )(−1) + (−kµ )(−lν )(−1) + (m̄µ )(mν )(1) + (mµ )(m̄ν )(1)
= −(lµ kν + lν kµ ) + (m̄µ mν + m̄ν mµ ),

where we can simplify the symmetrization of index pairs by round brackets

gµν = −2k(µ lν) + 2m(µ m̄ν) . (161)

8.1. Ricci rotation coefficients

We have seen that in orthonormal basis there are 24 Ricci rotation coefficients. But in pseudo-orthonormal basis,
because m and m̄ are complex conjugates, the number of Ricci rotation coefficients reduces to 16 coefficients. The
Ricci rotation coefficients are given by
Γαβγ = g(êα , ∇êβ êγ ).
In terms of NP null tetrad we have for Γ100 the following calculation

Γ100 = g(ê1 , ∇ê0 ê0 ) = g(lµ Eµ , ∇(kν Eν ) kρ Eρ ) = lµ kν g(Eµ , ∇ν kρ Eρ ) = lµ kν ∇ν kρ g(Eµ , Eρ ) = lµ kν ∇ν kρ gµρ ,

it follows that
Γ100 = lµ kν ∇ν kµ .
Notice that Γ100 = Γ100 because lµ and kµ are real numbers. Similarly to orthonormal basis, the pseudo-orthonormal
basis is a rigid frame with ∂δ γαβ = 0, so the Ricci rotation coefficients are given by (125) and results in (126),
Γγαβ = −Γβαγ , such as Γ100 = −Γ001 = −kµ kν ∇ν lµ .
The calculation for Γ200 is

Γ200 = g(ê2 , ∇ê0 ê0 ) = g(mµ Eµ , ∇(kν Eν ) kρ Eρ ) = mµ kν ∇ν kµ .

and for Γ300 = m̄µ kν ∇ν kµ , where we ca see that

Γ300 = mµ kν ∇ν kµ = Γ200 .

The calculation for Γ101 is

Γ101 = g(ê1 , ∇ê0 ê1 ) = g(lµ Eµ , ∇(kν Eν ) lρ Eρ ) = lµ kν ∇ν lµ .

where we observe that Γ101 = Γ101 .

We have Ricci rotation coefficient Γ102 = lµ kν ∇ν mµ and Γ103 = lµ kν ∇ν m̄µ , where we can see Γ103 = Γ102 . With
these calculations we can recognize the 24 Ricci rotation coefficient Γαβγ as

Γ100 = lµ kν ∇ν kµ Γ200 = mµ kν ∇ν kµ Γ300 = m̄µ kν ∇ν kµ Γ201 = mµ kν ∇ν lµ

Γ301 = m̄µ kν ∇ν lµ Γ302 = m̄µ kν ∇ν mµ Γ110 = lµ lν ∇ν kµ Γ210 = mµ lν ∇ν kµ
Γ310 = m̄µ lν ∇ν kµ Γ211 = mµ lν ∇ν lµ Γ311 = m̄µ lν ∇ν lµ Γ312 = m̄µ lν ∇ν mµ
(162)
Γ120 = lµ mν ∇ν kµ Γ220 = mµ mν ∇ν kµ Γ320 = m̄µ mν ∇ν kµ Γ221 = mµ mν ∇ν lµ
Γ321 = m̄µ mν ∇ν lµ Γ322 = m̄µ mν ∇ν mµ Γ130 = lµ m̄ν ∇ν kµ Γ230 = mµ m̄ν ∇ν kµ
Γ330 = m̄µ m̄ν ∇ν kµ Γ231 = mµ m̄ν ∇ν lµ Γ331 = m̄µ m̄ν ∇ν lµ Γ332 = m̄µ m̄ν ∇ν mµ

Newman and Penrose classify 12 independent complex linear combination of the above Ricci rotation coefficients
and called spin coefficients as follows

−κ = Γ200 = Γ̄300 = mµ kν ∇ν kµ ,
31
 −ρ = Γ320 = Γ̄230 = m̄µ mν ∇ν kµ ,

−σ = Γ220 = Γ̄330 = mµ mν ∇ν kµ ,

−τ = Γ210 = Γ̄310 = mµ lν ∇ν kµ ,

ν = Γ311 = Γ̄211 = m̄µ lν ∇ν lµ ,

µ = Γ321 = Γ̄231 = m̄µ mν ∇ν lµ ,

λ = Γ331 = Γ̄211 = m̄µ m̄ν ∇ν lµ ,

π = Γ301 = Γ̄201 = m̄µ kν ∇ν lµ ,

1 1µ ν 
−ǫ = (Γ100 − Γ302 ) = l k ∇ν kµ − m̄µ kν ∇ν mµ ,
2 2
1 1µ ν 
−β = (Γ120 − Γ322 ) = l m ∇ν kµ − m̄µ mν ∇ν mµ ,
2 2
1 1 µν 
γ = (Γ011 − Γ213 ) = k l ∇ν lµ − mµ lν ∇ν m̄µ ,
2 2
1 1 µ ν 
α = (Γ031 − Γ233 ) = k m̄ ∇ν lµ − mµ m̄ν ∇ν m̄µ . (163)
2 2
From expression (52), where we have to connection 1-forms in pseudo-orthonormal coordinate that

d(γαβ ) − Γαβ − Γβα = 0,

and with a constant metric, a rigid frame, d(γαβ ) = 0, it results to connection 1-forms that

Γαβ = −Γβα .

In this frame we have

Γα β = γαδ Γδβ , (164)
and we have that only non-null components of metric tensor from (155) are γ01 = γ10 = −1 and γ23 = γ32 = 1 where
we can use above equation (164) to find the connection 1-forms Γα β from Γδβ . The calculation for Γ0 0 follows

Γ0 0 = γ00 Γ00 + γ01 Γ10 + γ02 Γ20 + γ03 Γ30 ,

which is simplified in
Γ0 0 = −Γ10 .
The calculations for the others connection 1-forms are displayed below

Γ0 1 = γ00 Γ01 + γ01 Γ11 + γ02 Γ21 + γ03 Γ31 = 0;

Γ0 2 = γ00 Γ02 + γ01 Γ12 + γ02 Γ22 + γ03 Γ32 = −Γ12 ;
Γ0 3 = γ00 Γ03 + γ01 Γ13 + γ02 Γ23 + γ03 Γ33 = −Γ13 ;
Γ1 0 = γ10 Γ00 + γ11 Γ10 + γ12 Γ20 + γ13 Γ30 = 0;
Γ1 1 = γ10 Γ01 + γ11 Γ11 + γ12 Γ21 + γ13 Γ31 = −Γ01 ;
Γ1 2 = γ10 Γ02 + γ11 Γ12 + γ12 Γ22 + γ13 Γ32 = −Γ02 ;
Γ1 3 = γ10 Γ03 + γ11 Γ13 + γ12 Γ23 + γ13 Γ33 = −Γ03 ;
Γ2 0 = γ20 Γ00 + γ21 Γ10 + γ22 Γ20 + γ23 Γ30 = −Γ03 ;
Γ2 1 = γ20 Γ01 + γ21 Γ11 + γ22 Γ21 + γ23 Γ31 = −Γ13 ;

32
 Γ2 2 = γ20 Γ02 + γ21 Γ12 + γ22 Γ22 + γ23 Γ32 = −Γ23 ;
Γ2 3 = γ20 Γ03 + γ21 Γ13 + γ22 Γ23 + γ23 Γ33 = 0;
Γ3 0 = γ30 Γ00 + γ31 Γ10 + γ32 Γ20 + γ33 Γ30 = −Γ02 ;
Γ3 1 = γ30 Γ01 + γ31 Γ11 + γ32 Γ21 + γ33 Γ31 = −Γ12 ;
Γ3 2 = γ30 Γ02 + γ31 Γ12 + γ32 Γ22 + γ33 Γ32 = 0;
Γ3 3 = γ30 Γ03 + γ31 Γ13 + γ32 Γ23 + γ33 Γ33 = Γ23 . (165)

The above terms can be rearranged in a matrix form

 
−Γ01 0 −Γ12 −Γ13 

α  0 −Γ01 −Γ02 −Γ03 
(Γ β ) =  . (166)
−Γ03 −Γ13 −Γ23 0 
−Γ02 −Γ12 0 Γ23
Thus with these connection 1-forms we can see the first structure equation of Cartan (101),

dθ̃α = −Γα β ∧ θ̃β ,

in pseudo-orthonormal coordinate basis that results the four below equations





 dθ̃0 = Γ01 ∧ θ̃0 + Γ12 ∧ θ̃2 + Γ13 ∧ θ̃3


dθ̃1 = Γ01 ∧ θ̃1 + Γ02 ∧ θ̃2 + Γ03 ∧ θ̃3


 (167)



 dθ̃2 = Γ03 ∧ θ̃0 + Γ13 ∧ θ̃1 + Γ23 ∧ θ̃2


dθ̃3 = Γ02 ∧ θ̃0 + Γ12 ∧ θ̃1 − Γ23 ∧ θ̃3


The dual basis for complex null tetrad from (159),

θ̃0 = −lµ dxµ , θ̃1 = −kµ dxµ , θ̃2 = m̄µ dxµ and θ̃3 = mµ dxµ

has for each component the following complex conjugate

θ̃0 = −lµ dxµ = θ̃0 , θ̃1 = −kµ dxµ = θ̃1 , ¯ µ dxµ = mµ dxµ = θ̃3
θ̃2 = m̄ and θ̃3 = m̄µ dxµ = θ̃2 .

The complex conjugate for first equation of system (167), results in

d(θ̃0 ) = Γ01 ∧ θ̃0 + Γ12 ∧ θ̃2 + Γ13 ∧ θ̃3

where it reduces to

dθ̃0 = Γ01 ∧ θ̃0 + Γ12 ∧ θ̃3 + Γ13 ∧ θ̃2

dθ̃0 = Γ01 ∧ θ̃0 + Γ12 ∧ θ̃2 + Γ13 ∧ θ̃3

and it yields
Γ01 = Γ01 , Γ12 = Γ13 .
The complex conjugate for second equation of system (167), results in

d(θ̃1 ) = Γ01 ∧ θ̃1 + Γ02 ∧ θ̃2 + Γ03 ∧ θ̃3

where it reduces to

dθ̃1 = Γ01 ∧ θ̃1 + Γ02 ∧ θ̃3 + Γ03 ∧ θ̃2

dθ̃1 = Γ01 ∧ θ̃1 + Γ02 ∧ θ̃2 + Γ03 ∧ θ̃3

and it yields Γ02 = Γ03 . The complex conjugate for third equation of system (167), results in dθ̃2 = dθ̃3 , that is

Γ03 ∧ θ̃0 + Γ13 ∧ θ̃1 + Γ23 ∧ θ̃2 = Γ02 ∧ θ̃0 + Γ12 ∧ θ̃1 − Γ23 ∧ θ̃3 ,
33
and it results that Γ23 = Γ32 . We verify that exchanging the indices 2 and 3 implies complex conjugation. Thus we
can write the system of equations (167) with three equations,




 dθ̃0 = Γ01 ∧ θ̃0 + Γ12 ∧ θ̃2 + Γ12 ∧ θ̃3

 1

 dθ̃ = Γ01 ∧ θ̃1 + Γ02 ∧ θ̃2 + Γ02 ∧ θ̃3 , (168)


dθ̃2 = Γ02 ∧ θ̃0 + Γ12 ∧ θ̃1 + Γ23 ∧ θ̃2


where the fourth equation of system (167) is obtained from complex conjugation of third equation of system (168).
We can see that in pseudo-orthonormal basis - complex null tetrad - the number of independent connection 1-forms
Γαβ reduces from six to four.

8.2. The second equation of Cartan in pseudo-orthonormal basis

We have seen that in pseudo-orthonormal basis there are four connection 1-forms Γ01 , Γ02 , Γ12 and e Γ23 , with
complex conjugate Γ02 = Γ03 , Γ12 = Γ13 and Γ23 = Γ32 . We can observe that from second equation of Cartan (106),

Θαβ = dΓαβ + Γαγ ∧ Γγ β ,

it can be solved reducing from six to four the number of independent curvature 2-forms Θ01 , Θ02 , Θ13 e Θ32 .
Let us calculate the curvature 2-form Θ02

Θ02 = dΓ02 + Γ0γ ∧ Γγ 2

= dΓ02 + Γ01 ∧ Γ1 2 + Γ02 ∧ Γ2 2 + Γ03 ∧ Γ3 2 ,

where we can see the values of Γ1 2 , Γ2 2 and Γ3 2 on matrix (166), then it results in

Θ02 = dΓ02 + Γ01 ∧ (−Γ02 ) + Γ02 ∧ (−Γ23 ) + Γ03 ∧ (0),

being simplified in
Θ02 = dΓ02 + Γ02 ∧ (Γ01 + Γ23 ). (169)
The calculation for curvature 2-form Θ13 is

Θ13 = dΓ13 + Γ1γ ∧ Γγ 3

= dΓ13 + Γ10 ∧ Γ0 3 + Γ12 ∧ Γ2 3 + Γ13 ∧ Γ3 3
= dΓ13 + Γ10 ∧ (−Γ13 ) + Γ12 ∧ (0) + Γ13 ∧ Γ23
= dΓ13 + Γ13 ∧ (Γ10 + Γ23 )

where we obtain
Θ13 = dΓ13 − Γ13 ∧ (Γ01 + Γ32 ) (170)
The calculation for curvature 2-form Θ01 is

Θ01 = dΓ01 + Γ01 ∧ Γ1 1 + Γ02 ∧ Γ2 1 + Γ03 ∧ Γ3 1

= dΓ01 + Γ01 ∧ (−Γ01 ) + Γ02 ∧ (−Γ13 ) + Γ03 ∧ (−Γ12 )
= dΓ01 − Γ02 ∧ Γ13 − Γ03 ∧ Γ12 ,

and finally for Θ32 , we have

Θ32 = dΓ32 + Γ30 ∧ Γ0 2 + Γ31 ∧ Γ1 2 + Γ32 ∧ Γ2 2

= dΓ32 + Γ30 ∧ (−Γ12 ) + Γ31 ∧ (−Γ02 ) + Γ32 ∧ Γ23
= dΓ32 + Γ03 ∧ Γ12 − Γ02 ∧ Γ13 ,

here we can sum the above equations

Θ01 + Θ32 = d(Γ01 + Γ32 ) + 2Γ13 ∧ Γ02 . (171)

34
Thus we can rearrange the three complex equations (169), (170) and (171) in a system of 2-forms equations of the
second equation of Cartan, 



 Θ02 = dΓ02 + Γ02 ∧ (Γ01 + Γ23 )



 Θ13 = dΓ13 − Γ13 ∧ (Γ01 + Γ32 ) . (172)



Θ01 + Θ32 = d(Γ01 + Γ32 ) + 2Γ13 ∧ Γ02

The above system of 2-forms curvature equations has proved very useful in Newman-Penrose formalism [12, 17],
where exact solutions of Einstein equations are obtained.

8.3. Example of Brinkmann spacetime

An example most important which represents gravitational waves is the Brinkmann spacetime. This spacetime
can be used to explain the non-expanding waves, known as plane-fronted waves with parallel rays, or simply pp-
waves. The pp-waves may represent gravitational waves, electromagnetic waves, massless radiation associated with
Weyl fermions, some other forms of matter moving at the speed of light, or any combination of these. The metric for
pp-waves can be written in the form
ds2 = −2du dv − 2H du2 + 2dζ dζ̄, (173)
where H = H(ζ, ζ̄, u).
It is straightforward to notice that the shape of the Brinkmann metric is almost the same shape of the pseudo-
orthonormal metric of Minkowski spacetime (152)

ds2 = −2dudv + 2dζdζ̄,

where the aditional term −2H du2 is responsible for curvature of spacetime. When H = 0, the metric (173) reduces
to Minkowski spacetime. We can intuit that the the easiest way to work with Brinkmann metric is with complex null
tetrad. In non-coordinate basis pseudo-orthonormal the metric tensor is given by (153),

g = −2θ̃0 ⊗ θ̃1 + 2θ̃2 ⊗ θ̃3 ,

where we can immediately identify θ̃2 = dζ e θ̃3 = dζ̄. While the other sector of metric can be can be identified by

− 2θ̃0 ⊗ θ̃1 = −2du dv − 2H du2 ,

with θ̃α = ωα µ dxµ , we have that

du ⊗ dv + H du ⊗ du = θ̃0 ⊗ θ̃1 = (ω0 µ dxµ ) ⊗ (ω1 ν dxν )

= (ω0 u du + ω0 v dv) ⊗ (ω1 u du + ω1 v dv)
= ω0 u ω1 u du ⊗ du + ω0 u ω1 v du ⊗ dv + ω0 v ω1 u dv ⊗ du
+ω0 v ω1 v dv ⊗ dv,

where it yields the following equations

ω0 u ω1 u = H;
ω u ω v + ω0 v ω1 u
0 1
= 1;
ω0 v ω1 v = 0.

The solution for the above system of equations is displayed in a matrix,

 0
ω0 v ω0 ζ ω0 ζ̄  H
  
ω u 1 0 0
ω1 1   
ω1 v ω1 ζ ω ζ̄   1 0 0 0
(ω µ ) =  2 u
α  =  , (174)
ω u ω2 v ω2 ζ ω2 ζ̄   0 0 1 0
ω3 u ω3 v ω3 ζ ω3 ζ̄ 0 0 0 1

35
thus we obtain the dual basis as 



 θ̃0 = Hdu + dv


θ̃1

 = du
 . (175)



 θ̃2 = dζ


θ̃3
 = dζ̄
The matrix of vierbein is obtained from inversion of matrix (ωα µ ),
 u 
e0 e0 v e0 ζ e0 ζ̄  0 1 0 0

 u   
e e1 v e1 ζ e1 ζ̄  1 −H 0 0
(eα µ ) =  1 u v ζ ζ̄  = 
  , (176)
e2 e2 e2 e2  0 0 1 0
e3 u e3 v e3 ζ e3 ζ̄ 0 0 0 1

Then the tetrad of vectors of the pseudo-orthonormal basis is





 ê0 = k = ∂v


ê1 = l = ∂u − H∂v


 , (177)



 ê2 = m = ∂ζ



ê3 = m̄ = ∂ζ̄

the above tetrad can be expressed as 




 k = kα ∂x∂α

 l = lα ∂x∂α



 , (178)



 m = mα ∂x∂α


 m̄ = m̄α ∂α

∂x
∂
where ∂xα = (∂u , ∂v , ∂ζ , ∂ζ̄ ) and
       
0  1  0 0
1 −H  0 0
kα =   , lα =   , mα =   , m̄α =   . (179)
0  0  1 0
0 0 0 1

The exterior derivatives of 1-forms of (175) result that the only non-null is dθ̃0
∂H ∂H ∂H
dθ̃0 = dζ ∧ du + dζ̄ ∧ du + du ∧ du,
∂ζ ∂ζ̄ ∂u
∂H
here we denote ∂ζ = H,ζ and the above equation reduces in

dθ̃0 = H,ζ dζ ∧ du + H,ζ̄ dζ̄ ∧ du,

or
dθ̃0 = H,ζ θ̃2 ∧ θ̃1 + H,ζ̄ θ̃3 ∧ θ̃1 . (180)
Let us put this result in the first equation of Cartan (168),




 dθ̃0 = Γ01 ∧ θ̃0 + Γ12 ∧ θ̃2 + Γ12 ∧ θ̃3

 1

 dθ̃ = Γ01 ∧ θ̃1 + Γ02 ∧ θ̃2 + Γ02 ∧ θ̃3 ,


dθ̃2 = Γ02 ∧ θ̃0 + Γ12 ∧ θ̃1 + Γ23 ∧ θ̃2


where we obtain 



 −H,ζ θ̃1 ∧ θ̃2 − H,ζ̄ θ̃1 ∧ θ̃3 = −Γ01 ∧ θ̃0 + Γ12 ∧ θ̃2 + Γ13 ∧ θ̃3



 0 = Γ01 ∧ θ̃1 + Γ02 ∧ θ̃2 + Γ02 ∧ θ̃3 ,


0 = Γ02 ∧ θ̃0 + Γ12 ∧ θ̃1 + Γ23 ∧ θ̃2


36
the solution of this system results in

Γ12 = −H,ζ θ̃1 and Γ13 = −H,ζ̄ θ̃1 . (181)

The only independent non-null Ricci rotation coefficient is Γ12 = Γ13 .

We can solve the second equation of Cartan for Brinkmann spacetime with (172)




 Θ02 = dΓ02 + Γ02 ∧ (Γ01 + Γ23 )



 Θ13 = dΓ13 − Γ13 ∧ (Γ01 + Γ32 ) .



Θ01 + Θ32 = d(Γ01 + Γ32 ) + 2Γ13 ∧ Γ02

Here we can see that just the second equation of the above system has the nonzero 1-form Γ13 . Then the only nonzero
curvature 2-form is
∂H,ζ̄ ∂H,ζ̄ ∂H,ζ̄
Θ13 = dΓ13 = d(−H,ζ̄ θ̃1 ) = − du ∧ θ̃1 − dζ ∧ θ̃1 − dζ̄ ∧ θ̃1
∂u ∂ζ ∂ζ̄
= −H,ζ ζ̄ θ̃2 ∧ θ̃1 − H,ζ̄ ζ̄ θ̃3 ∧ θ̃1 (182)

From definition of curvature 2-form (103), we have

1 α
Θα β = R βγδ θ̃γ ∧ θ̃δ ,
2
such that

Θ13 = R1301 θ̃0 ∧ θ̃1 + R1302 θ̃0 ∧ θ̃2 + R1303 θ̃0 ∧ θ̃3 + R1312 θ̃1 ∧ θ̃2
+R1313 θ̃1 ∧ θ̃3 + R1323 θ̃2 ∧ θ̃3 ,

by comparing the above result with (182), we have the following components of Riemann tensor

R1312 = H,ζ ζ̄ , and R1313 = H,ζ̄ ζ̄ .

The components of the Ricci tensor can be computed via Rαβ = γδǫ Rǫαδβ . We use the symmetry of Riemann tensor
where R3121 = H,ζ ζ̄ . The complex conjugate of this component is

R̄3121 = R2131 = H,ζ ζ̄ .

There is only one nonzero component of Ricci tensor in pseudo-orthonormal basis

R11 = γ23 R3121 + γ32 R2131 = 2H,ζ ζ̄ .

The Ricci tensor is given by

Ric = Rαβ θ̃α ⊗ θ̃β ,
we recall from (56) where θ̃α = ωα µ dxµ , to rewrite the tensor Ricci in coordinate basis as

Ric = Rαβ (ωα µ dxµ ) ⊗ (ωβ ν dxν ),

we have only R11 , 0, then it results in

Ric = R11 ω1 µ ω1 ν dxµ ⊗ dxν ,
and with ω1 µ = −kµ from (159) the Ricci tensor in cordinate basis is given by

Ric = 2H,ζ ζ̄ kµ kν dxµ ⊗ dxν ,

where we can identify the components of Ricci tensor in coordinate basis as

Rµν = 2H,ζ ζ̄ kµ kν . (183)

37
If we use (160) we have that the matrix of Ricci tensor in coordinate basis is
 
 1 −1 0 0
−1 1 0 0
(Rµν ) = H,ζ ζ̄   .
 0 0 0 0
0 0 0 0
because gµν kµ kν = kµ kµ = 0, we have that the scalar curvature R = gµν Rµν is zero. Then, according to Einstein’ General
Relativity, the curvature os spacetime is related to the distribution of matter, where specifically the components of the
Ricci tensor are directly related to the local energy-momentum tensor T µν by Einstein’s field equations. We have from
(183) that the energy-momentum tensor T µν of Brinkmann spacetime is given by
T µν = ρ kµ kν . (184)
The above tensor is the energy-momentum tensor of null dust, where ρ denotes the energy density. The energy-
momentum tensor (184) has a similar structure as that of dust (pressure-free perfect fluid) this explains the notation
null dust. Null dust or pure radiation or incoherent radiation, represents a simple matter source in General Relativity
and describes the flux of massless particles, for example photons, with a fixed propagation direction given by the null
vector field kµ .

9. Conclusion
The purpose of this manuscript was to provide a self-contained review of manipulate some objects such as con-
nection coefficients and curvature tensors of a spacetime (M, g) in non-coordinate basis with concepts of vierbein
field. In a traditional manner the authors that introduce the General Relativity [1, 2, 3, 4, 5, 6], have taught to calculate
curvature tensors directly from metric tensor gµν in coordinate basis. However for an advanced way, some times it is
necessary to calculate the connection coefficients and curvature tensors with aid of Cartan’s structure equations which
deal with vierbein fields eα µ , the ’square root’ of the metric field tensor. So, we have seen that the non-coordinate
bases, orthonormal and pseudo-orthonormal (complex null tetrad), are useful for pedagogical study of conection co-
efficients and curvature in language of differential forms and also they are useful to manipulating and pratice the
Cartan’s structure equations. This manuscript has the introduction of complex null tetrad, the pseudo-orthonormal
basis, that is useful to the formalism of Newman-Penrose applied to General Relativity. The Newman-Penrose for-
malism adopts a tetrad in which the two axes u (advanced) and v (retarded) along the direction of propagation are
chosen to be lightlike, while the two complex axes ζ and ζ̄ are transverse to direction of propagation. It provides a
particular powerful way to deal with fields that propagates at the speed of light as we have seen with the example of
Brinkmann spacetime [12, 13, 14, 17].

References
[1] R. D’Inverno, Introducing Einstein’s Relativity, Oxford, UK: Clarendon (1992);
[2] L.D. Landau, & E.M. Lifschitz, Classical Theory of Fields, Oxford: Pergamon Press (1984);
[3] M.P. Hobson, G.P. Efstathiou, A.N. Lasenby, General Relativity, an introduction for physicists, Cambridge University Press (2006);
[4] B. F. Schutz, A First Course in General Relativity, Cambridge University Press (1985);
[5] R.M. Wald, General Relativity, The University of Chicago Press (1984);
[6] S. Weinberg, Gravitation and Cosmology: Principles and Applications of the General Theory of Relativity, John Wiley & Sons, Inc. (1972);
[7] S.W. Hawking, G.F.R. Ellis, The Large Scale Structure of Space-time, Cambridge University Press (1973);
[8] M. Nakahara, Geometry, Topology and Physics, IOP Publishing Ltd, (1990);
[9] L.P. Einsenhart, Riemannian Geometry, Princeton University Press, (1949);
[10] W.M. Boothby, An Introduction to Differentiable Manifolds and Riemannian Geometry, Academic Press an imprint of Elsevier Science,
second edition, (2003);
[11] C. Nash & S. Sen, Topology and Geometry for Physicists, Academic Press Inc (1983);
[12] H. Stephani, D. Kramer, M. A.H. MacCallum, C. Hoenselaers, E. Herlt, Exact solutions of Einstein’s field equations, Cambridge University
Press (2003);
[13] J. Stewart, Advanced General Relativity, Cambridge University Press (1991);
[14] E.T. Newman & R. Penrose, An approach to gravitational radiation by a method of spin coefficients, JMP 3, 566, (1962);
[15] H, Flanders, Forms with Applications to the Physical Sciences, Dover Publications, Inc. New York, (1989);
[16] M. Blagojević, F.W. Hehl - editors - Gauge Theories of Gravitation, a reader with commentaries, Imperial College Press (2013);
[17] J.B. Griffiths & J. Podolský, Exact Space-Times in Einstein’s General Relativity, Cambridge University Press (2009).

38