Electrodynamic Radiation Reaction and General

Relativity
arXiv:0803.3921v1 [gr-qc] 27 Mar 2008

Carlos Kozameh1 , Ezra T. Newman2 , Raul Ortega3 Gilberto Silva-Ortigoza4

1
FaMaF, Univ. of Cordoba,
Cordoba, Argentina
[email protected]
2
Dept of Physics and Astronomy,
Univ. of Pittsburgh,
Pittsburgh, PA 15260, USA
[email protected]
3
FACEN, Univ. of Catamrca,
Catamarca, Argentina
[email protected]
4
Facultad de Ciencias Fı́sico Matemáticas
de la Universidad Autónoma de Puebla,
Apartado Postal 1152, 72001,
Puebla, Pue., México
[email protected]
March 27, 2008

Abstract
We argue that the well-known problem of the instabilities associated
with the self-forces (radiation reaction forces) in classical electrodynamics
are possibly stabilized by the introduction of
gravitational forces via general relativity.

1 Introduction
The problems and difficulties associated with the motion of charged particles
interacting with both an external electromagnetic field and its own self-field
have not been resolved even after over a century of investigation[1, 2, 3, 4, 5].
The problems arise in many different contexts: the difficulties in giving appro-
priate initial conditions, infinite self-energy problems, model building, Lorentz
invariance difficulties and perhaps the most serious, the instabilities (or pre-
acceleration) in the solutions to the equations of motion.

1
 The best known equations of motion, coming from a point structureless
particle, are the Abraham-Lorentz equations and the relativist generalization
the Abraham-Lorentz-Dirac equations. They are given, respectively, by[3]

−
→ −
→ → → 2q 2 −
− →
m v̇ = q E + q B × −
v + 3 v̈ , (1)
3c
and[2]

2q 2 a 1
mv̇ a = qF ab vb + (v̈ + 2 v a v̇ b v̇b ), (2)
3c3 c
−
→ −
→
where F ab (or E , B ) are an external field and are derived by a variety of means,
but always with severe approximations.
It is generally acknowledged that there are fundamental difficulties with this
issue. And there seems to be a variety of different reasons, explanations and
suggested remedies. They range from: quantum theory is the resolution to the
approximations leading to these equations are wrong or even that there is no
real problem. The author, J. D. Jackson, summarizes the situation in his well
known graduate text[1] as:
“The difficulties presented by this problem touch one of the most funda-
mental aspects of physics, the nature of the elementary particle. Although
partial solutions, workable within limited areas, can be given, the basic prob-
lem remains unsolved. One might hope that the transition from classical to
quantum-mechanical treatments would remove the difficulties. While there is
still hope that this may eventually occur, the present quantum-mechanical dis-
cussions are beset with even more elaborate troubles than the classical ones. It
is one of the triumphs of comparatively recent years (˜1948 - 1950) that the
concepts of Lorentz covariance and gauge invariance were exploited sufficiently
cleverly to circumvent these difficulties in quantum electrodynamics and so al-
low the calculation of very small radiative effects to extremely high precision,
in full agreement with experiment. From a fundamental point of view, however,
the difficulties remain.”
The purpose of this note is to describe a totally new point of view towards
this problem. This new view is completely classical, with no reliance on quan-
tum theory. It however does rely heavily on the Einstein-Maxwell equations of
general relativity.
The basic situation that we address is to first consider an arbitrary com-
pact gravitating-electromagnetic system which is taken to be the particle whose
motion we want to describe. The system is given by local mass and charge
densities and currents. There are no external fields acting on it. It is an iso-
lated system with arbitrary internal degrees of fredom. The program is to solve
the Einstein-Maxwell equations in the future null asymptotic region and, from
the asymptotic field (the asymptotic Weyl and Maxwell tensors), determine a
center of mass and center of charge and their laws of motion. Aside from condi-
tions that the total charge Q be non-vanishing and the important requirement
that the centers of mass and charge should coincide, there is no further model
building.

2
 Since our detailed calculations, which were done in the language of the spin-
coefficient formalism, are long and complicated and have appeared elsewhere[6],
we will just summarize the ideas and results. Basically the calculations are done
2nd order in deviations from Reissner-Nordstrom. In the spherical harmonic
expansions, with frequent use of Clebsch-Gordon products, only terms up to
the l = 2 harmonics are kept.

2 The Complex Center of charge: Its Identifica-

tion
The basic starting idea in this work is essentially simple. It is in the generaliza-
tions and implementations where difficulties arise.
Starting in Minkowski space in a given Lorentzian frame with spatial origin,
−
→
the electric dipole moment D E is calculated from an integral over the (localized)
−
→
charge distribution. If there is a shift, R , in the origin, the dipole transforms
as
−→
∗ −
→ −
→
DE = DE − QR. (3)
−
→
If D E is time dependent, we obtain the center of charge world-line by taking
−→∗ −
→ − →
DE = 0, i.e., from R = D E /Q. It is this idea that we want to generalize and
extend to gravitational fields.
First, however, we want to discuss other dipole issues in flat-space. Starting
on the time-like world-line at the spatial origin, we construct the family of future
directed light-cones, C0 , and investigate behavior of the Maxwell field in the limit
as null infinity is approached, i.e., Penrose’s I+ . Using the null tetrad formalism
−
→ → −
and where the Maxwell pair ( E , B or Fab ) is replaced by the complex vector
−
→ −
→
E + i B , or more accurately by their tetrad components, (φ0 , φ1 , φ2 ), with[7]
φ0 = Fab la mb (4)
1
φ1 = Fab (la nb + ma mb )
2
φ2 = Fab ma nb .
The asymptotic (peeling) behavior of these fields for a compact source is given
by
φ00
φ0 = + O(r−4 ) (5)
r3
φ0
φ1 = 21 + O(r−3 )
r
φ02
φ2B = + O(r−2 ).
r
The vector field la is the tangent field to the null geodesic generators of the
null cones C0 . At I+ , nb is the tangent field to the null generators of I+ while

3
(ma , mb ) are (the complex conjugate pair) tangent to the two surface, S 2 , the
intersection of C0 with I+ .
The r independent quantities (φ00 , φ01 , φ02 , ...) are functions ‘living’ on I+ ,
i.e., functions of the retarded time, u, (the light-cone cuts of I+ ) and (ζ, ζ),
the complex stereographic coordinates labeling the, S 2 , generators of I+ . The
components of their spherical harmonic decomposition,

φ00 = φ00i Y1i1 + φ00ij Y2ij

1
+ ..., (6)
φ01 = Q + φ01i Y1i0 + φ01ij Y2ij
0
+ ..., (7)
φ02 = 0 −1 0 −1
φ2i Y1i + φ2ij Y2ij + ..., (8)

are the asymptotically defined multipole moments and their time derivatives.
For example, the l = 0 harmonic component of φ01 is proportional to the total
source charge. For us the important quantity is φ00 i , the l = 1 component of
φ00 : φ00 i is proportional to the (asymptotically defined ) complex dipole moment,
~C = D
D ~ E + iD~ M , where D
~ M is the magnetic dipole moment.
The problem now is: how does the D ~ C transform under an origin shift to
an arbitrary world-line? With an origin shift there will be new light-cones and
a new null vector field, l∗a , obtained from the old one, la , by a null rotation at
I+ . This can be expressed explicitly by[9, 10]

L L
l∗ = l +m + m + O(r−2 ) (9)
r r
m∗ = m + O(r−1 )
n∗ = n

where L = L(u, ζ, ζ) is a stereographic angle field given on I+ (still to be

described) that determines the new null geodesic field, l∗ .
The transformation law for φ00 i (given only approximately for small origin
shifts[8]) is[6]
0 0
φ∗0
0 i = (φ0 − 2Lφ1 + ...)i (10)
If we are given a Minkowski space world-line, xa = ξ a (s), for the apex of the
new light cones, then L = L(u, ζ, ζ) is given in the parametric form

L(u, ζ, ζ) = ξ a (s)ma (ζ, ζ), (11)

a
u = ξ (s)la (ζ, ζ),

with
√
2 ζ +ζ ζ − ζ −1 + ζζ
la (ζ, ζ) = (1, , −i , ), (12)
2 1 + ζζ 1 + ζζ 1 + ζζ
√
1 2 2 2
ma (ζ, ζ) = (0, Y1i (ζ, ζ)) = (0, 1 − ζ , −i(1 + ζ ), 2ζ), (13)
2P
By the appropriate choice of ξ a (s), from Eq.(10), with the use of Eq.(11),
a a
one can force the real part of φ∗0
0 i to vanish, thereby making x = ξ (s) the

4
center of charge. If however we generalized the choice of L(u, ζ, ζ) and allowed
it to be defined parametrically by

a
L(u, ζ, ζ) = ξC (τ )ma (ζ, ζ), (14)
a
u= ξC (τ )la (ζ, ζ),

where z a = ξCa
is a complex analytic world in complex Minkowski space, then by
∗0
setting φ0 i = 0, in Eq.(10) the complex world-line is determined. This complex
curve (which is purely formal) defines the complex center of charge. Using this
”curve” as the origin both the elecric and magnetic dipoles vanish.
Theorem[11] - copy from home

3 The Complex Center of Mass

For asymptotically flat Einstein-Maxwell space-times the situation is totally
analogous: the shear-free null geodesics originating from light-cones from world-
lines (real or complex) are replaced by (regular) asymptotically shear-free null
geodesic congruences generated by a complex world-line[11], in the space of the
complex Poincare translation subgroup of the BMS group. The Maxwell asymp-
totic dipole transforms exactly as in the flat space case, i.e., as in Eq.(10) with
however a slight change in the parametric description of the function L(u, ζ, ζ) :

L = ξ i (τ )Y1i1 (ζ, ζ) − 6ξ ij (τ )Y2ij

1
(ζ, ζ), (15)
1 1
u = √ ξ 0 (τ ) − ξ i (τ )Y1i0 (ζ, ζ) + ξ ij (τ )Y2ij
0
(ζ, ζ) + ..., (16)
2 2
where the extra terms come from the existence of a non-vanishing Bondi shear,
given up to l = 2 terms, by

σ = 24ξ ij (u)Y2ij
2
+ ... (17)
Turning the gravitational behavior, the relevant (for us) tetrad components
of the Weyl tensor[10, 6]

ψ1 = −Cabcd la mb lc md
ψ2 = −Cabcd ma nb lc md

have the asymptotic form (the peeling theorem)

ψ10 (u, ζ, ζ)
ψ1 = + O(r−5 )
r4
ψ 0 (u, ζ, ζ)
ψ2 = 2 3 + O(r−4 )
r

5
 The leading terms have the harmonic expansion:

ψ20 = Υ + ψ2i
0
Y1i0 + ψ2ij
0 0
Y2ij + ... (18)
ψ10 = 0
ψ1i Y1i1 + 0
ψ1ij 1
Y2ij + ... (19)

The mass aspect, defined by

Ψ = ψ20 + ð2 σ + σ(σ)· ,

is real and has the expansion

Ψ = Ψ = Ψ0 + Ψi Y1i0 + Ψij Y2ij

0
+ ...

The Bondi mass and linear momentum (four-momentum) is obtained from the
l = (0, 1) harmonic components of Ψ by:
√
2 2G
Ψ0 = − 2 M (20)
c
6G
Ψi = − 3 P i . (21)
c
The complex gravitational dipole moment (roughly mass dipole plus i times an-
gular momentum) is identified as being proportional to the l = 1 harmonic of
ψ10 , i.e., as ψ1i
0 0
. (Though many authors add further terms to ψ1i for this identi-
fication, to our order of approximation they all agree with our identification[6].)
0
The transformation (to second order) of ψ1i to an arbitrary (complex) world-
line, analogous to Eq.(10), using Eq.(15), is[6]
∗0
ψ1i = (ψ10 − 3Lψ20 + ...)i . (22)
∗0 i
Setting ψ1i = 0, thereby defining the complex center or mass, ξ (u), yields after
a lengthy calculation,
√
6 2G 1
0
ψ1i = − 2 M [ξ i (w) + i ǫkji v k ξ j ] + Gi , (23)
c 2
where Gi is a known non-linear function of quadrupole terms.
At this point we make our only assumption on the physical system being
considered. We saw that we could determine a complex center of charge or a
complex gravitational center of mass by setting either ϕ∗0 ∗0
0i or ψ1i to zero. We
now assume, for the rest of this work, that the two complex world-lines coincide.
Aside from taking Q 6= 0, there are no other conditions on the internal structure
of our source (particle).
We now turn to the dynamics, which are contained in the asymptotic Bianchi
identities. They can be written:

6
 0
(ψ10 )· = −ðΨ + ðσ(σ)· + 3σð(σ)· + ð3 σ + 2kφ01 φ2 (24)
0
·
Ψ =σσ + · ·
kφ02 φ2 (25)
−4
k = 2Gc (26)

From these two equations, (24) and (25), we extract the equations of motion
with the radiation reaction term. Rather than going thru the details (long with
rather unattractive calculations) we will describe what we did in words and
then give
√ the results. First we point out that we change the Bondi u-variable
to w = 2uc−1 , w being the retarded time. Derivatives with respect to w are
denoted by prime, (′ ).
We first extract from Eq.(24) its l = 1 part and then decompose it into its
real and imaginary parts. This yields two results: the imaginary part determines
the dynamics of the total angular momentum, i.e., the conservation of angular
momentum. Other than remarking that we identify S i = M cξIi as the intrinsic
spin (with ξIi the imaginary part of ξ i ), this is not our interest here and will
not be discussed any further. The real l = 1 part can be solved for the linear
momentum P i that was sitting in the l = 1 part of Ψ :

2Q2 k ′
P k = M vR
k
v + W k.
− (27)
3c3 R
This is a major result that come from our identification of the complex centers
of mass and charge. First of all we see kinematic expressions for the Bondi
3-momentum, the mv term and then the radiation reaction contribution to
the momentum. The W contains further kinematic terms involving spin and
quadrupole interactions that are known but not displayed here[6].
Extracting the l = (0, 1) harmonics from Eq.(25) yields the Bondi mass and
momentum loss equations:

G 2Q2 i ′ i ′
M′ =− (Qij ′′′ ij ′′′ ij ′′′
Mass QMass + QSpin QSpin ) −
ij ′′′
(v v + vIi ′ vIi ′ ) (28)
5c 7 3c5 R R
1
− (Dij ′′′ DE ij ′′′
+ DM ij ′′′ ij ′′′
DM )
180c7 E
2G lj ′′′ ij ′′′ lj ′′′ ij ′′′ Q2 l ′ i ′
Pk′ = Fk ≡ (Q Q − Q Q )ǫ ilk − (v v − vR l′ i′
vI )ǫilk
15c6 Spin Mass Mass Spin
3c4 I R
(29)
Q 1
+ (v j ′ Djk ′′′ + vIj ′ DM jk ′′′
)+ (Dlj ′′′ DM ij ′′′
− DMlj ′′′ ij ′′′
DE )ǫilk
15c5 R E 540c6 E
with the mass and spin quadrupoles related to the ξ ij by

ij G
ξ ij = (ξR + iξIij ) = √ (Qij′′ ij′′
Mass + iQSpin ). (30)
12 2c4

7
The mass loss equation is thus exactly the usual quadrupole energy loss plus
the classical dipole and quadrupole electromagnetic energy loss.
It is however Eq.(29) that is of most interest to us. By substituting the
kinematic expression for the momentum, Eq.(27) into Eq.(29) we obtain our
generalized Abraham-Lorentz equations of motion:

k′ k 2Q2 k ′′
M vR + vR M′ − v + Rk = F k . (31)
3c3 R
Note that though it is similar to the Abraham-Lorentz equations there are
many differences that are hidden in the known but complicated expressions for
M ′ , Rk and F k . The F k is the Bondi recoil (or rocket) force due to the momen-
tum loss, while Rk can be considered to be a gravitational radiation reaction
force depending on internal degrees of freedom spin and quadrupole moments.
M ′ has exactly the classical dipole energy loss terms plus two additional terms.
Though it is very hard to directly see if the solutions to Eq.(31) are well behaved,
in the conclusion we will discuss this issue in more general terms.

4 Conclusions
We have considered the situation of a gravitating - electromagnetic source of
compact support viewed from future null infinity. The only restriction made on
the distributions is that the total charge is non-vanishing and that the complex
asymptotic electromagnetic dipole moment be proportional the complex gravita-
tional dipole moment so that the complex centers of charge and mass coincide.
Though it is not clear how severe this condition is, it certainly is a serious re-
striction. It has been shown that for this situation the gyromagnetic ration, the
ratio of the spin-angular momentum to the magnetic moment is that of Dirac’s,
namely g = 2. We showed that in a manner completely analogous to the flat
space Maxwell case, one could determine the transformation laws for the two
dipole moments and thereby go to the center of mass/charge, determining a
unique complex world-line. Then, using the Bianchi Identities, which play the
role of dynamical equations, we were able to give kinematic significance to the
Bondi linear momentum, in the sense that we had
−
→ 2Q2 −
→
P = M−
→
v − 3 v ′ + ... (32)
3c
From the Bondi momentum loss equation it immediately followed that we
had a generalized version of the Abraham-Lorentz equations of motion for an
isolated massive charged particle. It should be emphasized that the quadrupole
quantity, ξ ij (u), is arbitrary and in most cases it is taken as non-vanishing in a
finite interval so that the gravitational radiation exists also in a finite interval.
If however the motion, from Eq.(31), is unstable, the particle acceleration will
be unbounded and there will be an infinite energy loss via the electromagnetic
dipole radiation. This would be a physically unacceptable situation, indicating
that something is seriously wrong with the Einstein-Maxwell equations.

8
 The question then is does the general relativity (gravitational) contribu-
tions to the equations of motion stabilize the equations. Though we do not see
any immediate prospects for a direct proof, we make a few comments. Looking
at Eq.(31), we see that the term M ′ v has the same form as in the Abraham-
Lorentz equation but now is more negative because of the extra radiation terms
and has the correct sign to try to stabilize the motion. Whether or not it does
stabilize is an open question. If not, perhaps higher order terms that have been
left out in our approximations might succeed. And finally, it is known that the
vacuum Einstein equations are stable in the neighborhood of Minkowski space.
If the same were true of the Einstein-Maxwell equations with compact sources,
that would constitute a proof that our physical situation was indeed stable and
the run-away behavior was prevented by the inclusion of classical general rela-
tivity. Unfortunately, this is a difficult question and, to our understanding, the
answer is unknown. It would be surprising if it turned out that the asymptotic
Einstein-Maxwell equations were unstable.

5 Acknowledgements
G.S.O. acknowledges the financial support from CONACYT and Sistema Na-
cional de Investigadores (SNI-México). C.K. thanks CONICET and SECY-
TUNC for support.

6 References
References
[1] J. D. Jackson, Classical Electrodynamics, 2 nd edition, Wiley &Sons, Hobo-
ken, NJ, (1975).
[2] W. Thirring, Principles of Quantum Electrodynamics (p.25), Academic
Press, NY and London, (1958).
[3] L. Landau, E. M. Lifschitz, Classical Theory of Fields, Addison-Wesley,
Reading, Mass., (1962).

[4] Spohn, Herbert Dynamics of Charged Particles and Their Radiation Field,
Cambridge University Press, Cambridge, (2004).
[5] Fritz Rohrlich, Classical Charged Particles, World Scientific, London, 3rd
edition, (2007).
[6] C. Kozameh, E.T. Newman and Gilberto Silva-Ortigoza, ”On Extracting
Physical Content from Asymptotically Flat Space-Time Metrics” submitted
Class.Quantum.Grav., (2008).
[7] Ezra T. Newman and Roger Penrose, J.Math.Phys, 3, 566–768, (1962).

9
 [8] The description of the finite transformation is considerably more compli-
cated.
[9] B. Aronson and E. T. Newman, J. Math. Phys. 13, 1847-51, (1972).
[10] E.T. Newman, K.P.Tod, in General Relativity and Gravitation, Vol 2, Alan
Held, (ed.), Plenum (1980).
[11] Kozameh and E.T. Newman, Class. Quantum Grav. 22, 4659-4665, (2005)

10