See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/330661987

Gravitoelectromagnetism: Formulation

Article · January 2019

CITATIONS READS

0 106

1 author:

Suvankar Paul
Indian Institute of Technology Kanpur
15 PUBLICATIONS   25 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Gravitational lensing images and shadows as a probe of strong gravity View project

All content following this page was uploaded by Suvankar Paul on 21 February 2019.

The user has requested enhancement of the downloaded file.

 Gravitoelectromagnetism: Formulation
Suvankar Paul ∗

It is well known that there lies a very close analogy between Newton’s gravitational law and Coulomb’s
law of electric field. This analogy leads us to investigate further into the similarities between these two
theories. One natural question in this regard arises in the possibility of existence of Gravitomagnetism.
In classical electrodynamics, we know that a moving electric charge or an electric current produces a
magnetic field around it. Therefore, we ask if an analogous mass current can produce the corresponding
gravito-magnetic fields. Many authors tried to formulate a theory of gravito-electromagnetism, but it was
Einstein [1] and then Thirring [2] who were successful in giving a formal theory of gravito-electrodynamics
within the framework of General relativity. In fact, Lens and Thirring in [3],[4],[5], (English version of
these three articles can be found in [6]) explicitly showed how a rotating mass produces a gravitomagnetic
field around it and affects the motion of planetary orbits. In this article I present a very short and brief
review of the formalism behind gravitoelectromagnetic effects. There are many articles regarding different
aspects of gravitomagnetic effects that can be found in the literature, for example [7],[8] etc.

1 Linearized Gravity Approach

In general relativistic framework, we can write Einstein’s equations in such a simplified way that it can
directly connect to the analogy of gravity with Maxwell’s equations in classical electrodynamics. It is
possible only if we consider the linearized weak field approximation of Einstein’s equation. In its actual
form, Einstein’s field equation in GR is

1 8πG
Gµν = Rµν − Rgµν = 4 Tµν (1)
2 c
where Gµν is called the Einstein tensor, Rµν is the Ricci tensor, R is the Ricci or curvature scalar, Tµν
is known as the stress-energy tensor, gµν is the metric tensor of the resulting spacetime, G and c are
the universal gravitational constant and speed of light in vacuum respectively. Here, µ, ν etc run for
(0, 1, 2, 3).
In absence of any gravitational field, the spacetime becomes flat represented by the usual Minkowski
metric [ηµν = diagonal(−1, 1, 1, 1)], i.e., gµν = ηµν . If a gravity source, whose gravitational field is very
weak, and which is moving with non-relativistic velocities, is introduced in the otherwise flat background,
the effective metric of the resulting spacetime can be approximated as,

gµν (x) = ηµν + hµν (x) (2)

where ‘x’ is a short hand notation for the coordinates (ct(= x0 ), x1 , x2 , x3 ), and |hµν | ≪ 1 is the small
perturbation over the flat background due to the source. Again, let us define,
∗ svnkr@iitk.ac.in

1
 1
h̄µν = hµν − hηµν (3)
2
where h = hµµ = η µν hνµ . With this new definition, the metric becomes,
 
h
gµν = 1+ ηµν + h̄µν (4)
2
Putting the above metric in the field equation (1), expanding it in powers of h̄µν , keeping only the linear
order terms and using the Lorentz guage condition, ∂ν h̄µν = 0, we obtain

16πG
h̄µν = − Tµν (5)
c4
where  ≡ ∂ µ ∂µ is known as the D’Alembertian operator. Now, before proceeding further in this
description, it is important to recall the laws of classical electromagnetism in its covariant form to
compare the above equation with the corresponding equation in electrodynamics.
In classical electrodynamics, the 4-vector potential is defined as
   
φ i φ
Aµ = ,A , or, Aµ = − , Ai (6)
c c
where φ is the scalar potential, Ai ’s are the components of the 3-vector potential and i runs for (1, 2, 3).
The electromagnetic field tensor is defined as

Fµν = ∂µ Aν − ∂ν Aµ , or, F µν = ∂ µ Aν − ∂ ν Aµ (7)

Maxwell’s equations, in terms of F µν , are therefore written as

∂ν F µν = µ0 J µ =⇒ ∂ν (∂ µ Aν − ∂ ν Aµ ) = µ0 J µ (8)

where J µ = cρ, j i is the 4-vector associated with charge (ρ) and current (j i ) densities, i.e. source of the

EM field, and µ0 is the free space permeability. Using the Lorentz guage condition, ∂µ Aµ = 0 (exactly
similar to the condition ∂ν h̄µν = 0 in derivation of Eq.(5)), the above equation reduces to

∂ν ∂ ν Aµ = −µ0 J µ =⇒ Aµ = −µ0 J µ

or, Aµ = −µ0 Jµ (9)

Comparing Eqs.(5) and (9), we see that the two equations are very similar in form indicating the fact
that there indeed exists electromagnetic-like analogous terms in linearized gravity also. The retarded
solution of Eq.(9) is given by

Z J µ t − |−

→ →
− →
−
µ x − x′ |/c, x′
Aµ ct, − → 0
d3 x′


x = →′
− (10)
4π →
−
|x − x |
As defined earlier, J µ = cρ, j i , or, Jµ = −cρ, j i . Writing the temporal and spatial parts of the above

expression separately, we obtain

2
 Z −c ρ t − |−

→ →′
− →′ 
−

µ0 x − x |/c, x
A0 ct, −
→
x = →
− d3 x′
4π |−
→x − x′ |
Z ρ t − |−

→ →′
− →′ 
−
→
−
φ ct, x


x − x |/c, x
µ0 c
=⇒ − =− →
− d3 x′
c 4π →
−
|x − x |′

Z ρ t−|x −−
→
− →′ →
−

1 x |/c, x′
=⇒ φ ct, −→ d3 x′


x = →′
− (11)
4πǫ0 →
−
|x − x |

and
Z j i t − |−

→ →
− →
−

µ0 x − x′ |/c, x′
Ai ct, −
→
x = →
− d3 x′ (12)
4π |−
→
x − x′ |
Equations (11) and (12) describe the familiar expressions of scalar and vector potentials of the corre-
sponding electromagnetic fields produced by a given charge and current distributions (source terms).
These two equations give us information about how an electromagnetic field is produced by a source
and the Lorentz force law will then provide us information about how a test charge or current will be
affected by this field. Therefore, equations (11) and (12) along with the Lorentz force law describe the
entire classical electromagnetism completely.
We shall now return to the problem of gravitational field under consideration. From Eqs.(5) and (9),
it is found that h̄µν , in case of gravity, plays the role of Aµ , and Tµν acts as the source of gravitational
field just like Jµ acts as the source of electromagnetic field. Therefore, following Eq.(10), we will have a
similar type of retarded solution for h̄µν , given by

Z Tµν t − |−

→ →
− −
→

4G x − x′ |/c, x′
h̄µν ct, −
→
x = 4 →′
− d3 x′ (13)
c →
−
|x − x |
To proceed further in this analysis, we first need to specify the nature of the source of gravitational field,
i.e., Tµν . Let us consider the source to be a perfect fluid so that the corresponding stress-energy tensor
becomes
 
P
Tµν = ρ + 2 uµ uν + P gµν (14)
c
where ρ is rest mass density, P is isotropic pressure, and uµ = cdt/dτ, dxi /dτ is the component of the

four-velocity vector of the fluid (τ is the proper time). Using the well known relation, dt = γdτ ,q
the form
v2
of uµ or uµ can be written as, uµ = γ(c, v i ), or, uµ = γ(−c, v i ), where v i = dxi /dt and γ(v) = 1/ 1 − c2 .
We should remember that, in the weak field approximation, |v i | ≪ c as well as |P | ≪ 1. Therefore, if
we expand γ(v) in powers of v/c, we find, in the weak field limit, γ(v) = 1 + 21 (v 2 /c2 ) + O(c−4 ) ≈ 1,
neglecting terms involving v 2 /c2 and all higher orders.
Considering different components of Tµν from Eq.(14) separately, we obtain

3
  
P
T00 = ρ + 2 (u0 )2 + P (η00 + h00 )
c
P
≈ ρc2 + 2 c2 − P (as |P | |h00 | ≪ 1)
c
=⇒ T00 ≈ ρc2 (15)

 
P
T0i = ρ+ u0 ui + P (η0i + h0i )
c2
P
≈ −ρcvi − cvi − P h0i
c2 v 
 i
=⇒ T0i ≈ −ρcvi = −cji (as |P |   ≪ 1, |P ||h0i | ≪ 1) (16)
c

 
P
Tij =
ρ + 2 ui uj + P (ηij + hij )
c
vi vj
≈ ρvi vj + P 2 + P (δij + hij )
c v v 
 i j
=⇒ Tij ≈ ρvi vj + P δij (as |P |  2  ≪ 1, |P ||hij | ≪ 1) (17)
c

With the above forms of Tµν under the present approximation, the expressions of h̄µν from Eq.(13) takes
the form

Z ρ t − |−

→ →
− →
−
x − x′ |/c, x′ 4φ ct, −
→


4G x
h̄00 ct, −
→
x = 2 →
−
3 ′
d x =− (18)
c |−
→
x − x′ | c2

Z ji t − |−

→ →
− →
−
x − x′ |/c, x′ 2Ai ct, −
→

→
−
4G 3 ′ x
h̄0i ct, x = − 3 →
− d x =− (19)
c |−
→
x − x′ | c2

Z (ρvi vj + P δij ) t − |−

→ →′
− →′ 
−

4G x − x |/c, x
h̄ij ct, →
−
x = 4 →
− d3 x′ ≈ 0, (20)
c |−
→
x − x′ |
 v v    
 i j P 
as  4  ≪ 1, and  4  ≪ 1
 
c c

where ji = ρvi is the mass current density, φ is the Newtonian gravitational potential and Ai is the
gravito-magnetic vector potential arising due to the mass current ji . Their expressions are respectively
given by

Z ρ t − |−

→ →
− →
−
x − x′ |/c, x′
φ ct, −
→ d3 x′ ,


x = −G →
− (21)
|−
→
x − x′ |

4
 Z ji t − |−

→ →
− −′ 
→

2G x − x′ |/c, x
Ai ct, −
→
x = →
− d3 x′ , (22)
c |−
→
x − x′ |
Z j i t − |−

→ →
− −′ 
→
2G x − x′ |/c, x
→
−
or, Ai ct, x =


d3 x′ (23)
→
−
c |−
→
x − x′ |

It is important to note that there is an intrinsic negative sign in the definition of Newtonian gravita-
tional potential (φ) as in Eq.(21), wheres no negative sign appears in the definition of electromagnetic
scalar potential (φ), as in Eq.(11). This is due to the fact that gravity is always attractive for normal
matter and on the other hand, electromagnetic force can be attractive as well as repulsive. As a result,
Newtonian gravitational potential for a mass M at a distance r is written as , φg (r) = − GM
r , but scalar
1 Q
potential for an electric charge Q at the same distance r is written as, φe (r) = 4πǫ r . Hence, all the
expressions involving φ below will differ by a negative sign from the corresponding relations in classical
electrodynamics. Moreover, from equations (21) and (22), it is seen that Ai is smaller in magnitude than
φ by a factor of ∼ 1/c.
Using the above definitions of gravito-potentials, we can easily define the corresponding gravitoelectric
and gravitomagnetic fields, as1
   
−
→ →
− 1 ∂ 1−→ 1−→ −
→ →
−
E g = ∇φ − A and B g = ∇ ×
A (24)
c ∂t 2 2
→
− ∂ →
− i →
− →
−
where ∇ ≡ ∂x i and A ≡ A . Once again, there is a flip of sign in front of ∇φ in the definition of E g ,

as compared to the standard definition of electric field in classical electrodynamics. Now, the Lorentz
guage condition, ∂ν h̄µν = 0, for µ = 0, becomes
   i
00 0i 1 ∂ 4φ ∂ 2A
∂0 h̄ + ∂i h̄ = 0 =⇒ − 2 + i =0
c ∂t c ∂x c2
 
1 ∂φ −→ 1−→
− +∇· A =0 (25)
c ∂t 2
→
−
Therefore, taking the divergence of E g and using the condition (25), we get
  
− −
→ → 2 1 ∂ → − 1−→
∇ · Eg = ∇ φ − ∇. A = φ (26)
c ∂t 2
Again, the 00-component of Eq.(5) gives

16πG 4φ 16πG
h00 = − T00 =⇒ (− 2 ) = − 4 (ρc2 ) =⇒ φ = 4πGρ (27)
c4 c c
Combining Eq.(26) and Eq.(27) yields

− −
→ →
∇ · E g = 4πGρ (28)
−
→
Similarly, taking divergence of B g gives

− −
→ →
∇ · Bg = 0 (29)
1 In this equation and all the following equations involving −
→ −
→
A, there is an extra 12 factor in front of A, which is absent
in standard classical electrodynamics. This is due to the fact that linearized gravity approximation involves a spin-2 field,
wheres standard electrodynamics deals with a spin-1 field of photons. For this reason, gravitomagnetic charge is twice the
gravitoelectric charge.

5
 →
− →
−
Again, taking the curls of both E g and B g from Eq.(24), we get
   →
−
− −
→ → 1 ∂ − → 1−→ → −
− → 1 ∂Bg
∇ × Eg = − ∇× A =⇒ ∇ × E g = − (30)
c ∂t 2 c ∂t
       
→ −
− → →
− →
− 1−→ → −
− → 1−→ 2 1−→
∇ × Bg = ∇ × ∇ × A =∇ ∇· A −∇ A
2 2 2
Now, from 0i-component of Eq.(5)
→
−
16πG 2A 16πG →
−
h0i = − 4
T 0i =⇒ (− 2
) = − 4 (−c j )
c c c
2
   
2 1−→ 4πG −
→ 1 ∂ 1−→
=⇒ ∇ A =− j + 2 2 A
2 c c ∂t 2
Therefore,
  
− −
→ → 4πG −
→ 1 ∂ → − 1 ∂ 1− →
∇ × Bg = j + ∇φ − A
c c ∂t c ∂t 2
→
−
−
→ − → 4πG −→ 1 ∂Eg
=⇒ ∇ × B g = j + (31)
c c ∂t
Equations (28), (29), (30) and (31) together constitute the Maxwell’s equations for the gravitoelectric field
→
− →
−
( E g ) and gravitomagnetic field ( B g ). Finally, the gravito-electromagnetic Lorentz force is determined
by (section 4.4 of [9])

−
→ →
− 1 −
→ → 
−
F g = −m E g − 4m v × Bg (32)
c
Again, let us consider the components h̄µν and find out the effective metric under the assumed
approximation at play here. We have
! !
h̄00 h̄0i −4φ/c2 −2Ai /c2
h̄µν = = (33)
h̄i0 0 −2Ai /c2 0
We know

1 1 1
h̄µν = hµν − hηµν =⇒ η νµ h̄µν = η νµ hµν − hη νµ ηµν =⇒ η νµ h̄µν = hνν − hδ νν
2 2 2
=⇒ η νµ h̄µν = h − 2h = −h =⇒ h = −η νµ h̄µν = −η 00 h̄00 − δ ij h̄ij
4φ
=⇒ h = h̄00 = − (34)
c2
Therefore,
!
−2φ/c2 −2Ai /c2
hµν = (35)
−2Ai /c2 −2φ/c2 δij
So the metric components of the background spacetime and the corresponding line element become
   
2φ 2Ai 2φ
g00 = − 1 + 2 , g0i = − 2 , gij = 1 − 2 δij , (36)
c c c
   
2φ 4Ai 2φ
ds2 = −c2 1 + 2 dt2 − 2 cdtdxi + 1 − 2 δij dxi dxj (37)
c c c

6
 It is important to remember that, in case of purely Newtonian gravity, only the 00-component of
the Einstein equation contributes resulting in the so called Poisson’s equation, ∇2 φ = 4πGρ, and this
φ comes from the 00-component of the above metric. All other components of the metric are equal to
the respective components of the flat Minkowski metric. So, in the Newtonian limit, all the off-diagonal
components of the metric are zero. Therefore, the gravito-magnetic vector potential Ai is found to be
the first Post-Newtonian correction over the Newtonian gravity.
Apart from all the above mentioned similarities between linearized gravity and classical electrody-
namics, there exits another important analogy worth mentioning here. In Eq.(32), it seems that there
is an arbitrariness in the definition of gravitomagnetic Lorentz force to make an one to one correspon-
dence with the standard Lorentz force definition in electrodynamics. But it can be shown that the same
→
−
expression of F g as in Eq.(32) can be obtained from the geodesic equation of a test particle moving in
this linearized gravity background. Mathematically, the geodesic equation is given as

d2 xµ ν
µ dx dx
λ
+ Γ νλ =0 (38)
dτ 2 dτ dτ
Simplifying the above equation in accordance with the approximation under consideration and neglecting
the relevant terms, we can arrive at the form of Lorentz force given in equation (32).

2 Curved Spacetime Approach

In the above section, we have introduced the concept of gravito-electromagnetic effects in a perturba-
tive method by linearizing Einstein’s theory of general relativity. This method is valid for very weak
gravitational sources moving with non-relativistic velocities in an effective flat background spacetime.
But there exists another approach by which the theory of gravito-magnetism can be applied for strong
gravity backgrounds also in a general curved spacetime. This methodology involves introduction of a new
coordinate system called Fermi Normal Coordinates (FNC). There are numerous excellent reviews and
discussions on FNC in the literature, for example see [10],[11],[12],[13], etc (in case of timelike geodesics),
and [14],[15] (in case of null geodesics). The essential idea behind FNC lies in the fact that, in this coor-
dinate system, the spacetime becomes locally flat along the entire geodesic of a test particle moving in
this spacetime. Since the metric becomes exactly flat on the geodesic locally, the signature of curvature
of the global curved background arises in the higher order derivatives of the metric. If xα = (cτ, xi )
represents FNC, then the metric can be expanded in the following forms

g00 = −1 − R0i0j xi xj + ... (39)

2
g0i = − R0jik xj xk + ... (40)
3
1
gij = δij − Rikjl xk xl + ... (41)
3
where, τ is the proper time along the timelike geodesic of the test particle and the components of
Riemann tensor mentioned above are taken to be the projections on the orthonormal tetrad frame (λαµ )
associated with the FNCs. Here, indices denoted by α, β, ... etc represent local Fermi normal coordinates,
and indices µ, ν, ... etc indicate global coordinates. Therefore, we can write

Rαβγδ = λαµ λβν λγρ λδσ Rµνρσ (42)

7
 Now the metric in Fermi normal coordinates looks, by construction, to be a perturbation over the
Minkowski metric, similar to the previous case. Therefore, we can easily extend all the analysis of the
previous section in this approach either. More specifically, if we compare equations (39) and (40) with
Eq.(36), we find the gravito-electromagnetic potentials as

c2
φ τ, −
→
x = R0i0j (τ ) xi xj + ... (43)
2

c2
Ai τ, −
→
x = R0jik (τ ) xj xk + ... (44)
3
→ −
− → − →
With these definitions of φ and Ai , we can define all the other quantities, like E g , B g , F g etc in terms of
the components of Riemann tensor and can complete the entire formulation of Gravito-electromagnetic
effects in a curved spacetime.

References
[1] A. Einstein Phys. Z. 14 (1913) 1261.

[2] H. Thirring Phys. Z. 19 (1918) 204.

[3] H. Thirring Phys. Z. 19 (1918) 33.

[4] J. Lens and H. Thirring Phys. Z. 19 (1918) 156.

[5] H. Thirring Phys. Z. 22 (1921) 29.

[6] B. Mashhoon, F. W. Hehl, and D. S. Theiss, “On the gravitational effects of rotating masses: The
thirring-lense papers,” General Relativity and Gravitation 16 no. 8, (Aug, 1984) 711–750.

[7] M. L. Ruggiero and A. Tartaglia, “Gravitomagnetic effects,” Nuovo Cim. B117 (2002) 743–768,
arXiv:gr-qc/0207065 [gr-qc].

[8] B. Mashhoon, “Gravitoelectromagnetism: A Brief review,” arXiv:gr-qc/0311030 [gr-qc].

[9] R. M. Wald, General Relativity. University of Chicago Press, 2010.

[10] F. K. Manasse and C. W. Misner, “Fermi normal coordinates and some basic concepts in
differential geometry,” J. Math. Phys. 4 (1963) 735–745.

[11] M. Ishii, M. Shibata, and Y. Mino, “Black hole tidal problem in the Fermi normal coordinates,”
Phys. Rev. D71 (2005) 044017, arXiv:gr-qc/0501084 [gr-qc].

[12] E. Poisson, A relativist’s Toolkit, The Mathematics of Black-Hole Mechanics. Cambridge

University Press, 2004.

[13] C. W. Misner, K. S. Thorne, and J. A. Wheeler, Gravitation. W. H. Freeman and Company, 1973.

[14] M. Blau, D. Frank, and S. Weiss, “Fermi coordinates and Penrose limits,”
Class. Quant. Grav. 23 (2006) 3993–4010, arXiv:hep-th/0603109 [hep-th].

[15] P. Banerjee, S. Paul, and T. Sarkar, “Curvature couplings of massless fermions in analog gravity,”
Phys. Lett. B789 (2019) 160–166.

View publication stats