Gravitoelectromagnetism: A Brief Review
Gravitoelectromagnetism: A Brief Review
Gravitoelectromagnetism: A Brief Review
Gravitoelectromagnetism:
A Brief Review
Bahram Mashhoon
Department of Physics and Astronomy
University of Missouri-Columbia
Columbia, Missouri 65211, USA
1.1 Introduction
The analogy between gravitation and electromagnetism has a long history. The
similarity between Newton’s law of gravitation and Coulomb’s law of electricity
naturally led to a gravitoelectric description of Newtonian gravitation. Moreover,
on the basis of advances in electrodynamics in the second half of the nineteenth
century, Holzmüller [1] and Tisserand [2] postulated that the gravitational force
exerted by the Sun on the planets of the solar system had an additional “magnetic”
component. This extra force led to the precession of the planetary orbits; therefore,
it could be adjusted in order to account for the excess perihelion precession of
Mercury. Decades later, however, Einstein’s general relativity provided a beautiful
explanation of the excess motion of Mercury’s perihelion in terms of a relativistic
gravitoelectric correction to the Newtonian gravitational potential of the Sun [3].
Furthermore, general relativity, which is a field theory of gravitation, contains a
gravitomagnetic field due to mass current [4]. Indeed, to bring together Newtonian
gravitation and Lorentz invariance in a consistent field-theoretic framework, the
introduction of a gravitomagnetic field is unavoidable.
1
2 CHAPTER 1. GRAVITOELECTROMAGNETISM: A BRIEF REVIEW
According to general relativity, the proper rotation of the Sun produces a gravit-
omagnetic field and the influence of this field on planetary orbits was first considered
by de Sitter [5] and later in a more general form by Lense and Thirring [4]. The
gravitomagnetic contribution to the excess motion of Mercury’s perihelion turns out
to be much smaller and in the opposite sense compared to the main gravitoelectric
motion; in fact, it turns out that the Lense-Thirring precession of planetary orbits
is too small to be measurable at present. On the other hand, evidence for the grav-
itomagnetic field of the Earth has been offered by Ciufolini by studying the motion
of laser-ranged satellites LAGEOS and LAGEOS II [6]. The precise measurement
of this field via superconducting gyroscopes in a drag-free satellite in polar orbit
about the Earth is one of the aims of NASA’s GP-B [7].
Within the framework of general relativity, gravitoelectromagnetism (“GEM”)
has been discussed by a number of authors [3, 8]; a more extensive list of references is
provided in [9]. The purpose of this review is to present the two principal approaches
to GEM and briefly describe some of their consequences.
1 8πG
Rµν − gµν R = 4 Tµν (1.1)
2 c
take the form
16πG
2h̄µν = − Tµν (1.2)
c4
after imposing the transverse gauge condition h̄µν,ν = 0. The general solution of
(1.2) is a superposition of a particular solution together with the general solution of
the wave equation; however, we are only interested in the special retarded solution
of (1.2) given by
4G Tµν (ct − |x − x′ |, x′ ) 3 ′
Z
h̄µν = 4 d x. (1.3)
c |x − x′ |
Let us define the matter density ρ and matter current j = ρv via T 00 = ρc2 and
T 0i = cj i , respectively. Moreover, it is useful to define the GEM potentials Φ and A
in terms of ρ and j as h̄00 = 4Φ/c2 and h̄0i = −2Ai /c2 , respectively. Assuming that
the source consists of a finite distribution of slowly moving matter with |v| << c,
Tij ∼ ρvi vj + pδij , where p is the pressure, and (1.3) imply that h̄ij = O(c−4 ).
1.2. LINEAR PERTURBATION APPROACH TO GEM 3
All terms of O(c−4 ) will be neglected in this analysis. Under these conditions, the
spacetime metric has the GEM form
2 2 Φ 2 4 Φ
ds = −c 1 − 2 2 dt − (A · dx)dt + 1 + 2 2 δij dxi dxj . (1.4)
c c c
In the Newtonian limit, Φ reduces to the Newtonian gravitational potential, while
A = O(c−1 ). If the source distribution is confined around the origin of spatial
coordinates, then far from the source
GM GJ×x
Φ∼ , A∼ , (1.5)
r c r3
where r = |x| and M and J are the mass and angular momentum of the source,
respectively. Moreover, the transverse gauge condition reduces to
1 ∂Φ 1
+∇· A = 0. (1.6)
c ∂t 2
We define the GEM fields via
1 ∂ 1
E = −∇Φ − A , B = ∇ × A, (1.7)
c ∂t 2
in direct analogy with electromagnetism. It follows from these definitions that the
GEM fields have dimensions of acceleration and
1 ∂ 1 1
∇×E=− B , ∇· B = 0. (1.8)
c ∂t 2 2
Furthermore, (1.2) implies that
1 1 ∂ 4πG
∇ · E = 4πGρ, ∇× B = E+ j. (1.9)
2 c ∂t c
The GEM field equations (1.8) and (1.9) contain the continuity equation ∇ · j +
∂ρ/∂t = 0, as expected.
For a complete GEM theory, we need an analogue of the Lorentz force law. The
Lagrangian for the motion of a test particle of mass m, L = −mcds/dt, can be
written to linear order in Φ and A as
1
v2 2 v2
2 2m
L = −mc 1 − 2 + mγ 1 + 2 Φ − γv · A, (1.10)
c c c
where γ is the Lorentz factor. The equation of motion, dp/dt = F, where p = γmv
is the kinetic momentum, takes a simple familiar form if ∂A/∂t = 0 and F is
expressed to lowest order in v/c, Φ and A; then,
v
F = −mE − 2m × B. (1.11)
c
4 CHAPTER 1. GRAVITOELECTROMAGNETISM: A BRIEF REVIEW
1∂ 1 ′ 1
Φ′ = Φ − Ψ, A = A + ∇Ψ, (1.13)
c ∂t 2 2
7 X
4πGt00 = − E 2 + A(i,j) A(i,j) , (1.14)
2
i,j
There is some similarity between these and the corresponding relations in classical
electrodynamics. In particular, the GEM Poynting vector is given by
c
S =− E × B. (1.17)
2πG
For instance, gravitational energy circulates around a stationary source of mass m
and angular momentum J = Jẑ with a flow velocity
J
vg = k sin θ φ̂ (1.18)
Mr
in the same sense as the rotation of the mass. Here we employ spherical polar coordi-
nates and k = 4/7. The flow given by (1.18) is divergence-free and the corresponding
circulation is independent of the radial distance r and is given by 2πk(J/M ) sin2 θ.
1.3. GRAVITATIONAL LARMOR THEOREM 5
dλµ(α)
= φαβ λµ(β) , (1.19)
dτ
where φαβ (τ ) is the antisymmetric acceleration tensor of the observer. In analogy
with the Faraday tensor, φαβ consists of an“electric” part φ0i = ai and a “magnetic”
6 CHAPTER 1. GRAVITOELECTROMAGNETISM: A BRIEF REVIEW
part φij = ǫijk Ωk . Here a and Ω are spacetime scalars that represent respectively
the translational acceleration, ai = aµ λµ(i) , and the rotational frequency of the local
spatial frame with respect to the local nonrotating (i.e. Fermi-Walker transported)
frame. Consider now a geodesic system of coordinates X µ established along the
worldline of the fiducial observer. At any event τ along the worldline, the straight
spacelike geodesic lines orthogonal to the worldline span a hyperplane that is Eu-
clidean space. Let xµ be the coordinates of a point on this hyperplane; then,
The Minkowski metric ηµν dxµ dxν with respect to the new coordinates takes the
form gµν dX µ dX ν , where
These geodesic coordinates are admissible if g00 < 0; a detailed discussion of the
nature of the boundary of the admissible region is given in [12].
A comparison of the metric given by (1.21) and (1.22) with (1.4) reveals that
they are Larmor equivalent at the linear order once
1
Φ = −aL · X, A = − ΩL × X, (1.23)
2
and we neglect spatial curvature. The corresponding GEM fields to lowest order are
E = −∇Φ = aL and B = ∇ × A = −ΩL , as expected from the traditional Larmor
theorem with qE = −m and qB = −2m.
The gravitational Larmor theorem [13] is essentially Einstein’s principle of equiv-
alence formulated within the GEM framework. Einstein’s heuristic principle of
equivalence traditionally refers to the Einstein “elevator” and its translational ac-
celeration in connection with the gravitoelectric field of the source. However, it
follows from the gravitational Larmor theorem that a rotation of the elevator is
generally necessary as well in order to take due account of the gravitomagnetic field
of the source.
In classical electrodynamics, a charged spinning test particle has a magnetic
dipole moment µ = qS/(2mc), where m, q and S are respectively the mass, charge
and the spin of the particle. In an external magnetic field B, the test dipole has an
interaction energy −µ · B and precesses due to a torque µ × B. In a similar way, a
test gyroscope of spin S with q → qB = −2m has a gravitomagnetic dipole moment
µg = −S/c and precesses in the exterior field of a rotating source of mass M and
spin J with the frequency [14]
GJ
ΩP = [3(Ĵ · r̂)r̂ − Ĵ], (1.24)
c2 r 3
1.4. SPACETIME CURVATURE APPROACH TO GEM 7
where ΩP = B/c and the gravitomagnetic field B is given by the curl of the vector
potential in (1.5) that corresponds to a source with gravitomagnetic dipole moment
J/c. The related interaction energy is S · ΩP . A major aim of the GP-B is to
measure (1.24) for gyros in a polar orbit about the Earth.
It follows from (1.24) that 2π/ΩP is a characteristic timescale for the gravitomag-
netic field. More generally, gravitomagnetic effects reveal an interesting temporal
structure around a rotating mass; this can be further illustrated by the phenomena
associated with the gravitomagnetic clock effect [15] and the gravitomagnetic time
delay [16].
A more exact long-term post-Schwarzschild analysis of the orbital motion of
an ideal test gyroscope in the field of a rotating source reveals that besides the
gravitoelectric geodetic (i.e. de Sitter-Fokker) precession of the gyro axis there is
a complex gravitomagnetic component involving precessional as well as nutational
motions—the latter is known as relativistic nutation [17]. The net gravitomagnetic
spin motion reduces in the post-Newtonian approximation to equation (1.24).
relative velocity of the test particle) V = dX/dT is in general arbitrary (|V| < 1
at X = 0). To linear order in velocity, one can show that (1.35) takes the Lorentz
form
d2 X
m 2 = qE E + qB V × B, (1.36)
dT
where qE = −m and qB = −2m as before.
The stress-energy tensor in the new approach can be constructed essentially from
the Faraday tensor (1.33) as in Maxwell’s theory, i.e.
αβ 1 α βγ 1 αβ γδ
GT = F γ F − g Fγδ F , (1.37)
4π 4
This tensor as well as the Faraday tensor (1.33) vanishes along the worldline of the
reference observer in the Fermi system; indeed, this is an immediate consequence of
the inertial character of this system along the reference trajectory and a realization
of Einstein’s principle of equivalence. Thus this treatment depends on our choice of
a reference observer and the corresponding Fermi coordinate system.
To obtain a coordinate-independent measure of the stress-energy content of the
gravitational field, we invoke the notion that the physical measurement of such a
quantity requires an averaging process [21]. Starting from an event (T, 0) on the
reference worldline, we average the tensor given in (1.38) over a small sphere of
radius ǫL, where 0 < ǫ << 1 and L is an invariant length scale that is characteristic
of the source of the gravitational field under consideration. For instance, L could be
GM/c2 or, in the absence of such a scale, the Planck length. The quadratic nature
of (1.38) in the spatial coordinates implies that the averaging involves
where k = 1/5 or 1/3 depending on whether the averaging involves the volume or
the surface of the sphere, respectively. In either case, the constant k can be absorbed
in the definition of L. Thus
kǫ2 L2
hTαβ i = T̄µνρσ λµ(α) λν(β) λρ(0) λσ(0) , (1.40)
4πG
where T̄µνρσ (x) is the Bel tensor given by
1 1
T̄µνρσ (x) = (Rµξρζ Rν ξ ρζ + Rµξσζ Rν ξ ρζ ) − gµν Rαβργ Rαβσ γ . (1.41)
2 4
10 CHAPTER 1. GRAVITOELECTROMAGNETISM: A BRIEF REVIEW
This tensor bears a certain similarity with the Maxwell stress-energy tensor and
on this basis was first defined by Bel [22] for Einstein spaces in 1958. The Bel
superenergy tensor is symmetric and trace-free in its first pair of indices and only
symmetric in the second pair of indices. In a Ricci-flat spacetime, the Riemann
tensor reduces to the Weyl conformal tensor Cµνρσ and the Bel tensor reduces to
the completely symmetric and trace-free Bel-Robinson tensor Tµνρσ given by
1 1
Tµνρσ = (Cµξρζ Cν ξσζ + Cµξσζ Cν ξρ ζ ) − gµν gρσ Cαβγδ C αβγδ . (1.42)
2 16
An invariant average GEM stress-energy tensor of the gravitational field can
thus be defined up to a constant positive multiplicative factor by [23]
L2
T̄(α)(β) = T̄µνρσ λµ(α) λν(β) λρ(0) λσ(0) , (1.43)
G
∗
where T̄(α)(β) is symmetric and traceless. In the Ricci-flat case, T̄(α)(β) → T(α)(β) ,
∗
where T(α)(β) is the gravitational stress-energy tensor. This designation refers to the
fact that in a Ricci-flat spacetime, the spatial components of the curvature tensor
in (1.27), which were essentially ignored in our GEM analysis, are indeed basically
∗
given by its electric components; therefore, T(α)(β) involves all of the components
of the spacetime curvature. The stress-energy tensors T̄µν and Tµν ∗ have properties
respect to an inertial frame; therefore, the spin appears to rotate with respect to a
rotating observer. To this motion of spin corresponds, according to quantum me-
chanics, a Hamiltonian H = −γΩ · S. The general formula for the transformation
of energy turns out to be E ′ = γ(E − ~ΩM), where Ω is the frequency of rotation
of the observer and ~M is the component of the total angular momentum along
the axis of rotation; that is, M = 0, ±1, ±2, . . . for a scalar or a vector particle,
while M ∓ 21 = 0, ±1, ±2, . . . for a Dirac particle. This formula relates the energy
of a quantum system measured by a rotating observer E ′ to measurements per-
formed in a global inertial frame and can be written in the JWKB approximation
as E ′ = γ(E − Ω · J), where J = r × p + S is the total angular momentum. Thus,
E ′ = γ(E − v · p) − γΩ · S, so that in the absence of intrinsic spin we recover the
classical expression for the energy of a particle as measured in the rotating frame
with v = Ω × r. The spin-rotation coupling therefore involves an energy shift given
by the Hamiltonian H = −γΩ · S [27].
Observational evidence for such an energy shift in the case of fermions has been
provided in certain high-precision experiments by way of a small frequency offset due
to the coupling between the nuclear spin of mercury and the rotation of the Earth
[28, 29]; moreover, a direct approach using neutron or atom interferometry has
been proposed in [30]. For photons, helicity-rotation coupling has been confirmed
to rather high accuracy using rotating GPS receivers [31]; moreover, experimental
evidence exists for such a coupling in the microwave and optical regimes in terms of
the frequency shift of polarized radiation [30]. The modifications of Doppler effect
and aberration due to the coupling of photon spin with the rotation of the source
and/or receiver have been the subject of recent studies [32, 33].
Let us now turn to the coupling of spin with gravitomagnetic fields; the spin-
gravity coupling is naturally related to the spin-rotation coupling by way of Ein-
stein’s principle of equivalence. That is, starting from the spin-rotation Hamiltonian,
the transformation Ω → −ΩP leads, according to the gravitational Larmor theorem,
to the spin-gravity Hamiltonian.
It follows from these ideas that in Earth-based experiments, to every Hamilto-
nian we must add the spin-rotation-gravity interaction Hamiltonian δH ∼ = −Ω⊕ ·
S + ΩP · S, where Ω⊕ and ΩP refer to the rotation frequency of the Earth and
the corresponding gravitomagnetic precession frequency, respectively. Thus in the
approximation under consideration here a particle with intrinsic spin behaves es-
sentially like an ideal gyroscope. The energy difference corresponding to a spin-1/2
particle polarized vertically up and down relative to the surface of the Earth can be
estimated from ~Ω⊕ ∼ = 10−19 eV and ~ΩP ∼= 10−29 eV. The measurement of the latter
term is beyond present capabilities by several orders of magnitude. In this connec-
tion, however, we note that near Jupiter ~ΩP ∼ = 10−27 eV, and therefore it is likely
that with further improvements in magnetometer design, the spin-gravitomagnetic
coupling could become measurable in a satellite in orbit near the surface of Jupiter
in the foreseeable future [34]. It is important to recognize that such a relativistic
12 CHAPTER 1. GRAVITOELECTROMAGNETISM: A BRIEF REVIEW
quantum gravitational effect, like all other gravitational effects, is subject to the
whole mass-energy content of the universe. It follows that our treatment has been
based on certain cosmological assumptions regarding the distribution of angular
momentum in the universe; specifically, we have assumed that on the largest scales
there is no preferred sense of rotation. Moreover, in δH the spin-gravity coupling
term has a gradient. Therefore, there exists a gravitomagnetic Stern-Gerlach force
−∇(ΩP · S) on a spinning particle that is independent of the its mass and hence
violates the universality of the gravitational acceleration. The weight of a body
thus depends on its spin, but the effect is too small to be directly measurable in the
foreseeable future. It is interesting to note that the Stern-Gerlach force has an exact
analogue in the classical Mathisson-Papapetrou spin-curvature force. The results of
this section are in agreement with the consequences of Dirac-type wave equations
in the gravitational field of a rotating mass [35].
References
[2] F. Tisserand, Compt. Rend. 75, 760 (1872); 110, 313 (1890).
[4] H. Thirring, Phys. Z. 19, 33 (1918); 22, 29 (1921); J. Lense and H. Thirring,
Phys. Z. 19, 156 (1918); B. Mashhoon, F.W. Hehl and D.S. Theiss, Gen.
Rel. Grav. 16, 711 (1984).
[5] W. de Sitter, Mon. Not. Roy. Astron. Soc. 76, 699 (1916).
[7] C.W.F. Everitt et al., in Near Zero: Festschrift for William M. Fairbank,
edited by C.W.F. Everitt (Freeman, San Francisco, 1986).
[8] A. Matte, Canadian J. Math. 5, 1 (1953); L. Bel, Compt. Rend. 247, 1094
(1958); P. Teyssandier, Phys. Rev. D 16, 946 (1977); 18, 1037 (1978);
V. Braginsky, C. Caves and K.S. Thorne, Phys. Rev. D 15, 2047 (1977);
R. Jantzen, P. Carini and D. Bini, Ann. Phys. (NY) 215, 1 (1992); I.
Ciufolini and J.A. Wheeler, Gravitation and Inertia (Princeton University
Press, Princeton, 1995); M.A.G. Bonilla and J.M.M. Senovilla, Phys. Rev.
Lett. 11, 783 (1997); R. Maartens and B.A. Bassett, Class. Quantum Grav.
15, 705 (1998); S.J. Clark and R.W. Tucker, Class. Quantum Grav. 17,
4125 (2000); J.M.M. Senovilla, Mod. Phys. Lett. A 15, 159 (2000); Class.
Quantum Grav. 17, 2799 (2000); M.L. Ruggiero and A. Tartaglia, Nuovo
Cimento B 117, 743 (2002); L. Iorio and D.M. Lucchesi, Class. Quantum
Grav. 20, 2477 (2003).
13
14 REFERENCES
[15] J.M. Cohen and B. Mashhoon, Phys. Lett. A 181, 353 (1993); B. Mashhoon,
F. Gronwald and H.I.M. Lichtenegger, in Testing Relativistic Gravity in
Space, edited by C. Lämmerzahl, C.W.F. Everitt and F.W. Hehl (Springer,
Berlin, 2000); B. Mashhoon, F. Gronwald and D.S. Theiss, Ann. Physik 8,
135 (1999); W.B. Bonnor and B.R. Steadman, Class. Quantum Grav. 16,
1853 (1999); B. Mashhoon and N.O. Santos, Ann. Physik 9, 49 (2000); A.
Tartaglia, Gen. Rel. Grav. 32, 1745 (2000) ; B. Mashhoon, L. Iorio and H.
Lichtenegger, Phys. Lett. A 292, 49 (2001) ; L. Iorio, H.I.M. Lichtenegger
and B. Mashhoon, Class. Quantum Grav. 19, 39 (2002) ; D. Bini, R.T.
Jantzen and B. Mashhoon, Class. Quantum Grav. 19, 17 (2002).
[16] I. Ciufolini and F. Ricci, Class. Quantum Grav. 19, 3863, 3875 (2002); I.
Ciufolini, S. Kopeikin, B. Mashhoon and F. Ricci, Phys. Lett. A 308, 101
(2003).
[20] C. Chicone and B. Mashhoon, Class. Quantum Grav. 19, 4231 (2002).
[23] B. Mashhoon, J.C. McClune and H. Quevedo, Phys. Lett. A 231, 47 (1997);
Class. Quantum Grav. 16, 1137 (1999); 17, 533 (2000).
[25] B. Mashhoon and D.S. Theiss, Phys. Rev. Lett. 49, 1542 (1982).
REFERENCES 15
[26] B. Mashhoon, H.J. Paik and C.M. Will, Phys. Rev. D 39, 2825 (1989).
[28] B.J. Venema et al., Phys. Rev. Lett. 68, 135 (1992).
[33] J.D. Anderson and B. Mashhoon, Phys. Lett. A 315, 199 (2003).
[35] C.G. de Oliveira and J. Tiomno, Nuovo Cimento 24, 672 (1962); F.W. Hehl
and W.T. Ni, Phys. Rev. D 42, 2045 (1990); Y.Q. Cai and G. Papini, Phys.
Rev. Lett. 66, 1259 (1991); I.D. Soares and J. Tiomno, Phys. Rev. D 54,
2808 (1996); G. Papini, Phys. Rev. D 65, 077901 (2002); L.H. Ryder and
B. Mashhoon, in: Proc. Ninth Marcel Grossmann Meeting (Rome, 2000),
edited by V.G. Gurzadyan, R.T. Jantzen and R. Ruffini (World Scientific,
Singapore, 2002), pp. 486-497.