Home Search Collections Journals About Contact us My IOPscience

Radiation reaction as a non-conservative force

This content has been downloaded from IOPscience. Please scroll down to see the full text.

2016 Eur. J. Phys. 37 055202

(http://iopscience.iop.org/0143-0807/37/5/055202)

View the table of contents for this issue, or go to the journal homepage for more

Download details:

IP Address: 89.202.245.164
This content was downloaded on 07/07/2016 at 11:47

Please note that terms and conditions apply.

 European Journal of Physics
Eur. J. Phys. 37 (2016) 055202 (14pp) doi:10.1088/0143-0807/37/5/055202

Radiation reaction as a non-conservative

force
Sandeep Aashish and Asrarul Haque1
Department of Physics, BITS Pilani Hyderabad Campus, Hyderabad-500078, AP, India

E-mail: sandeepaashish@gmail.com and ahaque@hyderabad.bits-pilani.ac.in

Received 15 February 2016, revised 20 May 2016

Accepted for publication 9 June 2016
Published 6 July 2016

Abstract
We study a system of a nite size charged particle interacting with a radiation
eld by exploiting Hamiltons principle for a non-conservative system recently
introduced by Galley [1]. This formulation leads to the equation of motion of
the charged particle that turns out to be the same as that obtained by Jackson
[2]. We show that the radiation reaction stems from the non-conservative part
of the effective action for a charged particle. We notice that a charge inter-
acting with a radiation eld modeled as a heat bath affords a way to justify that
the radiation reaction is a non-conservative force. The topic is suitable for
graduate courses on advanced electrodynamics and classical theory of elds.

Keywords: radiation reaction, Hamiltons principle for non-conservative

systems, AbrahamLorentz equation

1. Introduction

An accelerating charge radiates and therefore loses energy [24]. Larmor power accounts for
the rate of loss of energy for a non-relativistic accelerated point charge. Loss of energy could
be conceived to be caused by a damping force operating on the charge produced by the
radiation eld. This damping force is known as the radiation reaction [2].
An accelerating charge is thus a non-conservative system which experiences the radiation
reactiona non-conservative force. However, an accelerating charge together with its own
eld as a system conserves energy and is therefore conservative.
Although the radiation reaction has been studied extensively in the past [510] with
several notable attempts to rectify the issues of causality violation and runaway solutions [11
17], it continues to be the subject of several recent studies [1821].

1
Author to whom any correspondence should be addressed.

0143-0807/16/055202+14$33.00 2016 IOP Publishing Ltd Printed in the UK 1

Eur. J. Phys. 37 (2016) 055202 S Aashish and A Haque

We perform a systematic study of the radiation reaction as a non-conservative force using

Hamiltons principle for non-conservative systems to obtain the equation of motion of a
charge and establish its non-conservative nature.
The radiation reaction, studied in an ad hoc manner in the past in light of Hamiltons
principle [2224] for the construction of the relevant Lagrangian, could be attributed
essentially to the prescribed Lagrangian in Galleys formulation of Hamiltons principle.
Polonyi [25] has studied the radiation reaction by working out the effective dynamics of a
single point charge having recast Schwingers closed time path formalism. Polonyis effective
action approach to the radiation reaction appears similar to that of the derived effective action
of Galleys formulation; however, his method does not have an immediate scope to construct
a prescribed Lagrangian like Galleys formulation.
Galley [1] has formulated Hamiltons principle compatible with initial data which leads
to the EulerLagrange equations of motion for both conservative and non-conservative
systems. In the present work, we obtain the equation of motion for an accelerating charge by
exploiting Galleys formulation of Hamiltons principle for a non-conservative system. The
equation of motion so obtained for a rigid, spherically symmetric charge instantaneously at
rest turns out to be the same as that obtained by Jackson [2]. The radiation reaction is shown
to stem from the non-conservative part of the effective action for a charged particle. The
AbrahamLorentz equation [26] is derived using the effective non-conservative Lagrangian
for a point charge. We notice, on the basis of a correspondence between a charge interacting
with radiation eld and that of a particle interacting with innite bath oscillators, that the
radiation reaction could be realized as a non-conservative force.
In section 2, we briey review Hamiltons principle for a non-conservative system due to
Galley. In section 3, we observe distinctly why the usual derivation of the EulerLagrange
equation from the effective action provides an incomplete equation of motion for an accel-
erating charge. In sub-sections 3.1 and 3.2, we derive the correct equation of motion for a
nite size charge and a point charge respectively using Galleys formulation of Hamiltons
principle for a non-conservative system.

2. Preliminaries

We understand that Hamiltons principle is formulated as a boundary value problem in time to

account for conservative systems. In this section, we shall briey discuss the formulation of
Hamiltons principle for a non-conservative system due to Galley [1].

2.1. Hamiltons action principle for a non-conservative system

Galleys formulation of Hamiltons principle, which is compatible with initial value pro-
blems, is endowed with a systematic algorithm as to how to obtain the EulerLagrange
equation for a non-conservative system. The essential ingredients required to obtain the
EulerLagrange equation for a non-conservative system are:
1. a dynamical system and
2. coordinates qi and velocities qi .
Consider a dynamical system with coordinates qi and velocities qi . The action, a func-
tional of the dynamical variables qi, is dened by

2
Eur. J. Phys. 37 (2016) 055202 S Aashish and A Haque

tf
S [qi] = t i
dtL (qi , qi) . (1)

Under an arbitrary variation of the paths, qi (t ) qi (t ) + hi (t ), with the end points held
xed, i.e., h (ti ) = 0 = h (tf ), Hamiltons principle states that for the actual paths the rst-
order variation of S vanishes
dS d tf
=
d = 0 de t i
L (qi (t ) , qi (t )) dt
e = 0
= 0,

which in turn leads to the EulerLagrange equation

d L L
- = 0.
dt qi (t ) qi (t )

Thus, Hamiltons principle is formulated not as an initial value problem but as a boundary
value problem in time since its formulation requires knowledge of the initial and nal values
of qi(t). However, the solution to the derived equation of motion requires initial data, i.e., the
values of qi(t) and qi (t ) at the initial time. It is this disparity between the two that lies at the
heart of the incompatibility of Hamiltons principle with dissipative (time-asymmetric)
systems. To this end, we consider the following example. Consider the action for a harmonic
oscillator
tf 1 1
S= t i
dt mx 2 - kx 2
2 2
( 2)

which leads to the equation of motion

d2
m 2 + k x (t ) = 0. (3)
dt

We can rewrite the action as

tf 1 2 1 2 1 d tf 1 1
t i
mx (t ) -
2 2
kx (t ) dt =
i
t
m
2 dt
(xx ) - mxx - kx 2 (t ) dt
2 2
tf 1 d2
=-
ti 2
x (t ) m 2 + k x (t ) dt.
dt
(4)

1 d
The total derivative term 2 m dt (xx) vanishes as the end points are held xed. Note that the
integrand in (4) is time-symmetric. Now, we consider the equation of motion for a damped
harmonic oscillator
d2 d
m 2 + b + k x (t ) = 0.
dt dt
In order to derive the EulerLagrange equation for the damped harmonic oscillator using
Hamiltons principle, getting inspired from (4), one might be tempted to dene the action as
follows
tf 1 d2 d
Sd [x (t )] = - t i 2
x (t ) m 2 + b + k x (t ) dt .
dt dt

However, under the arbitrary variation of the path x (t ) x (t ) + h (t ) with the endpoints
held xed, h (ti ) = 0 = h (tf ), the vanishing of the rst-order variation of Sd

3
Eur. J. Phys. 37 (2016) 055202 S Aashish and A Haque

dSd 1 tf d

d e= 0
=-
2 t i
dt (mx + kx ) h + b (hx ) = 0
dt
(5)

leads to the incorrect equation of motion

d2
m 2 + k x (t ) = 0
dt
for the damped harmonic oscillator because the velocity term in (5) appears as a total
derivative and does not contribute.
Galley has developed a consistent formulation of Hamiltons principle that is compatible
with initial value problems. The formulation of Hamiltons principle for a generic system is
accomplished through the following steps.
Double the dynamical variables [2729] qi: qi (qi1, qi2 ). This is done to facilitate the
nonconservativity.
Dene action, a functional of the coordinates qi1 and qi2, as
tf
S [qi1, qi2] = ti dt L(qi1, i2 , qi1, i2) (6)

where the new Lagrangian reads

L(qi1, i2 , qi1, i2) = L (qi1, qi1) - L (qi2 , qi2) + K (qi1, i2 , qi1, i2 , t ) . (7)
K depends on both variables and encodes the important aspects of the formulation.
When
K = 0, two paths qi1 and qi2 are decoupled and L describes a
conservative system,
0, the two paths qi1 and qi2 get coupled with each other and L
describes a nonconservative system.
It is thus the non-zero K that determines the non-conservative nature of a dynamical
system.
Subject an arbitrary variation of the paths: qi1, i2 (t ) qi1, i2 (t , 0) + hi1, i2 (t ) with the
following pair of boundary conditions
i. hi1 (ti ) = hi2 (ti ) = 0 at the initial time as well as
ii. qi1 (tf , ) = qi2 (tf , ) and qi1 (tf , ) = qi2 (tf , ) (equality conditions) at the nal
timeso that Hamiltons principle for the corresponding change in the action becomes
dS [qi1, qi2]
= 0,
d = 0
qi1= qi2 = qi

where qi1 = qi2 = qi is called the physical limit. Hamiltons principle leads to the
following equation of motion
L d L d
- = - K (qi1, i2 , qi1, i2 , t ) .
qi dt qi qi2 dt qi2
qi1= qi2 = qi

It is noteworthy that the equality conditions imply that the data qi1, i2 (tf ) and qi1, i2 (tf )
are not held xed (see gure 1).

4
Eur. J. Phys. 37 (2016) 055202 S Aashish and A Haque

Figure 1. The solid lines represent the paths qi1 (t ) and qi2 (t ) and the dashed lines stand
for their innitesimal ( 1) departures. At the initial time ti both variables are xed,
i.e., qi1 (ti ) = qi2 (ti ) = qi0 and at the nal time tf their values coincide, i.e.,
qi1 (tf , ) = qi2 (tf , ), but have an arbitrary variation.

Change the variables qi1, i2 (t ) (for conveniences sake only) to

qi1 + qi2
qi + and qi - qi1 - qi2 (8)
2
such that qi - 0 and qi + qi for qi1 = qi2 (physical limit (PL)). Hamiltons principle in
terms of these new variables
dS [qi ] d S [q ]
i
= =0 (9)
de e = 0 d
i - t ) PL
q (
qi -= 0, qi += qi

leads to the EulerLagrange equation for a generic system as

dpi L L K
= = + (10)
dt qi - PL qi qi - PL
where the conjugate momenta pi are given by
L L K
pi = = + . (11)
qi - PL qi qi - PL

K is zero for conservative systems whereas open systems, which can exchange energy
with the environment, possess non-zero K. A non-zero K gives rise to the non-conservative
force as we shall see in section 3. Non-zero K could be either
prescribed when the underlying variables that cause non-conservative (e.g. dissipative)
forces are neither given nor modeled or
derived when all the degrees of freedom of a closed system are given or modeled.
Non-zero derived K could be obtained from the effective action (such as the Fokker
action) that describes some interacting system, for instance, a system of a charge interacting
with the radiation eld. It is wort noting that the interim variables (qi) need not be associated
with any physical meaning until after the physical limit is applied. The novelty of the
formulation lies in its generality in the sense that

it is compatible with both conservative and non-conservative systems and

5
Eur. J. Phys. 37 (2016) 055202 S Aashish and A Haque

in this formulation it is possible to have a consistent Lagrangian consisting of distinct

conservative as well as non-conservative part with the freedom to have either prescribed
or derived K for the non conservative part of Lagrangian.

3. Effective action for an accelerating charge

Consider a system of a nite size charged particle of charge e coupled with its own

electromagnetic eld that is subjected to an external potential V (x ). Such a system of a
charged particle plus electromagnetic eld is described by the Lagrangian [30]
L = L 0 + L em + L int (12)
where
1 2
L0 = mxi - V (x ) (13)
2
1 1
L em = d3x 2 0 (-f - A )2 - 2m0 ( A )2
1 1
= d3x 2 0 (if + A i )2 - 2m0 (iAj iAj - iAj j Ai ) (14)

L int = d3x (-rf + Ji Ai ); (15)

and Ji are charge and current densities of the charged particle respectively; f and Ai are the
potentials associated to the self eld of the charged particle. The action associated with the
Lagrangian (12) is given by

S [xi , f , Ai ] = dtL (xi , xi , f, f , if, Ai , A i , j Ai )

1
= dt 2 mxi2 - V (x )

1 1
+ dtd 3x 0 (i f + A i )2 -
2 2m 0
(i Aj i Aj - i Aj j Ai )

+ dtd 3x (-rf + Ji Ai ). (16)

Variations of the action (16) with respect to potentials f and Ai in the Lorentz gauge
f
. A + 0 m0 =0
t
lead to the following equations of motion for f and Ai
1
f = - r (17)
0
Ai = - m 0 Ji. (18)

6
Eur. J. Phys. 37 (2016) 055202 S Aashish and A Haque

The appropriate solutions to equations (17) and (18) for radiation reaction can be given by
1
f (x , t ) =
4p 0 d3xdt Gret (x - x , t - t ) r (x , t ) (19)

m0
Ai (x , t ) =
4p
d3x dt Gret (x - x , t - t ) Ji (x , t ) (20)

where the function Gret (x - x , t - t ) stands for the retarded Green function and has
x -x
nonzero support at t - t = c
. We obtain the effective action by substituting (19) and
(20) in the action (16) as
1 1 1
Seff = dt 2 mxi2 + 2 4p 0 d3x d3xdt {m0 0 Ji (x , t ) Ji (x , t ) Gret (x - x , t - t )

- r (x , t ) r (x , t ) Gret (x - x , t - t )}] .
(21)

The products Ji (x , t ) Ji (x , t ) and r (x , t ) r (x , t ) are symmetric under the exchange of
variables xi xi and t t and couple only to the symmetric part of the retarded Green
function i.e.

G (x - x , t - t ) + Gret (x - x , t - t )
G+ ret
2

Gret (x - x , t - t ) + Gadv (x - x , t - t )
= (22)
2

since Gret (x - x , t - t ) = Gadv (x - x , t - t ) which has nonzero support at
x -x
t - t = c
. The effective action thus becomes
1 1 1
Seff = dt 2 mxi2 + 2 4p 0 d3x d3xdt {m0 0 Ji (x , t ) Ji (x , t ) G+

- r (x , t ) r (x , t ) G+}] . (23)
The equation of motion for the nite size charged particle (rigid) could be obtained by the
variation of an effective action (23) with respect to xi. We have the EulerLagrange equation
as
1 1 G+

mx + d3x d3x dt r (x , t ) 2
4p 0 c t
J (x , t ) + G+r (x , t ) = 0

(24)

where we have used for a rigid object Ji (x , t ) = r (x , t ) xi (t ). Equation (24) would exhibit
non-causal behavior because the presence of Gadv in G+ would render the evolution of
equation of motion acausal. The second term of (24)
1 1 G+

d3x d3x dt r (x , t ) 2
4p 0 c t
J (x , t ) + G+r (x , t )

R
subject to small c , rigid, spherically symmetric and instantaneously at rest charge distribution,
after performing a Taylor series expansion [2], takes the following form

1 1 2 1 + ( - 1)n n + 1
2 4p 0
3
n=0

n!c n + 2 t n + 1
x (t ) d3x d3xr (x ) Rn-1r (x )

7
Eur. J. Phys. 37 (2016) 055202 S Aashish and A Haque

where R = x - x . Equation (24) now becomes

1 1 2 1 + ( - 1)n n + 1
mx +
2 4p 0
3
n=0

n!c n + 2 t n + 1
x (t ) d3x d3xr (x ) Rn-1r (x ) = 0 (25)

which in fact does not represent the correct equation of motion for an accelerating charge.
Owing to the merely temporally symmetric contribution to (25), the second term in (25)

contains terms proportional to only an even order time derivative like x (t ), x (t ) etc and

hence does not incorporate a radiation reaction term which is proportional to x (t ). It is the
retarded Green function Gret = G+ + G- that can provide the correct equation of motion

1 2 ( - 1) n n + 1
mx +
4p 0
3 n!c n+ 2 t n+ 1 x (t ) d3x d3xr (x ) Rn- 1r (x ) = 0. (26)
n=0
In the following section, we shall apply Hamiltons principle for a non-conservative system
due to Galley for an accelerating charge in order to obtain its correct equation of motion.

3.1. Equation of motion for an accelerated charge using Hamiltons principle for a non-
conservative system

The Lagrangian that describes a system of an accelerating charge plus its electromagnetic
eld reads

1 1
d3x 2 ((if)2 + A i + 2i fA i ) -
2
L = mxi2 - V (x ) + 0
((j Ai )2 - j Ai i Aj )
2 2m 0
- rf + Ji Ai ]. (27)
We begin with doubling the variables as follows
xi xi1, xi2
Ai Ai1 , Ai2
f f1, f 2 . (28)
Next, we subject the change of variables as given below
Ai1 + Ai2 f + f2 x + xi 1
Ai + = f+ = 1 xi + = i 1
2 2 2
Ai - = Ai1 - Ai2 f- = f1 - f 2 xi - = xi1 - xi2 (29)
so that the action becomes
S [xi + , xi -, f+, f-, Ai + , Ai - ]
= dt L

= dt mxi +xi - - V- + d3x ( 0 i f+i f- + 0 A i + A i -
1 1
-
j Ai - j Ai + + 0 A i - i f+ + 0 A i + f- + j Ai + i Aj -
m0 2m 0
1
- j Ai - i Aj + - r+f- - r-f+ + (Ji +Ai - + Ji -Ai + ) (30)
2m 0

where V- V (x1) - V (x2 ). Equations of motion for Ai and f that follow from the
variation of action (30) are given by

8
Eur. J. Phys. 37 (2016) 055202 S Aashish and A Haque

Ai = - m 0 Ji (31)
1
f = - r (32)
0
where we have identied
r1 + r 2 1 1 1
r+ = = r x+ + x- + r x+ - x- (33)
2 2 2 2

1 1
r- = r1 - r2 = r x+ + x- - r x+ - x- (34)
2 2
and
Ji1 + Ji2 1
Ji + = = xi +r+ + xi -r-
2 4
Ji - = Ji1 - Ji2 = xi +r- + xi -r+. (35)
Now the solutions to (31) and (32) in the Lorentz gauge are given by
m
4p
Ai + (x , t ) = 0 d3x dt Gret (x - x , t - t ) J+(x , t ) (36)

1
f+(x , t ) =
4p 0 d3xdt Gret (x - x , t - t ) r+(x , t ) (37)

up to solutions of the homogeneous equations, which for the radiation reaction could be
appropriately set to zero and
m
4p
Ai - (x , t ) = 0 d3x dt Gadv (x - x , t - t ) Ji -(x , t ) (38)

1
f-(x , t ) =
4p 0 d3x dt Gadv (x - x , t - t ) r-(x , t ). (39)

We can obtain the effective action by substituting (36), (37), (38) and (39) in the action (30).
The effective action turns out to be

Seff = dt [mxi+xi- - V-
m
+ 0 d3xJi -(x , t ) d3x dt Gret (x - x , t - t ) Ji +(x , t )
4p
1
-
4p 0 d3xr-(x, t ) d3x dt Gret (x - x , t - t ) r+(x , t )]. (40)

The effective Lagrangian could be read from the effective action (40) as
L = mxi +xi - - V-
m0
+ d3x d3x dt 4p Ji-(x, t ) Gret (x - x , t - t ) Ji+(x , t )
1
- r-(x , t ) Gret (x - x , t - t ) r+(x , t ) . (41)
4p 0
Thus, the effective Lagrangian consists of conservative as well as non-conservative parts.
The non-conservative part K of is given by

9
Eur. J. Phys. 37 (2016) 055202 S Aashish and A Haque

m0
K= d3x d3x dt 4p Ji-(x, t ) Gret (x - x , t - t ) Ji+(x , t )
1
- r-(x , t ) Gret (x - x , t - t ) r+(x , t ) (42)
4p 0
while the conservative part of (in the physical limit) reads
1 2
L=mxi - V (x ) . (43)
2
The EulerLagrange equation for xi(t)
d L L K d K
+ = + (44)
dt xi xi xi - dt xi - PL
follows from the variation of action Seff (40) which in turn yields
1 1
mxi + d3x d3xdt r (x , t ) 4p 0 c 2 t [Gret Ji (x , t )]

+ i [Gret r (x , t )]] = Fiext . (45)
where
K m0
=
xi - PL 4p d3x d3x dt r (x , t ) Gret (x - x , t - t ) Ji (x , t ) (46)

K 1
=-
xi - PL 4p 0 d3x d3x dt r (x , t ) i [Gret (x - x , t - t ) r (x , t )] (47)

V
- Fiext . (48)
xi
Moreover, for a rigid, spherically symmetric charged particle instantaneously at rest, equation
of motion (45) takes the form
1 2 ( - 1)n n + 1
mx +
4 p 0 n = 0 3 n! c n + 2 t n + 1
x (t ) d3x d3x r (x ) R n - 1r (x ) = Fext . (49)

This is in fact the same equation as obtained by Jackson [2]. We observe that the radiation
reaction
m e2
FRad = 0 x
6pc
comes from the expansion of the second term for n = 1 in the LHS of equation (49). Thus the
radiation reaction stems from K. This is important to note that it is now possible to obtain
distinctly the non-conservative force from the non-conservative part K of the effective
Lagrangian . Moreover, it is now feasible to predict almost effortlessly a prescribed
Lagrangian
1 2 ( - 1) n n + 1
L = mx+ x- + x- Fext -
4p 0 n = 0 3 n!c n + 2
x-
t n + 1
x+In (50)

10
Eur. J. Phys. 37 (2016) 055202 S Aashish and A Haque

that would result the same equation of motion as (49), where

In d3x d3x r (x ) R n - 1r (x ) . (51)

We note that the prescribed Lagrangian (50) is seemingly different from the derived one (41);
however, both of them lead to the same equation of motion in the physical limit. We have an
advantage of this formalism that we can have the freedom to choose K (the last term in (50)).
2 (-1)n n
For example, K = 4p
1
n = 0 +
x- n x+In is an equally acceptable choice.
0 3 n!c n 2 t
3.2. Equation of motion for a point charge from the effective Lagrangian

We can obtain the effective Lagrangian for an accelerating point charge from in (41), using
the following form of Greens function
1 R
Gret (x - x , t - t ) = d t - t + (52)
R c
and performing integration over the variable t , as follows
m0 1
L = mzi +zi - - V- +
4p d3x d3xJi-(x , z, t ) R Ji+(x , z, tr )
1 1
-
4p 0 d3x d3xr-(x , z, t ) R r+(x , z, tr ) (53)

where tr = t - R c and V- is given by

1 1
V- = V z + + z- - V z + - z- . (54)
2 2
If the velocity of the charge is small compared to that of light, then the change in charge and
R
current densities could be considered negligible [31] over the time interval c . Hence, in the
point charge limit (R 0), we can neglect the higher order terms in the expansion of r+(tr )
and Ji +(tr ) about t
R
Ji +(tr ) Ji +(t ) - Ji +(t ) (55)
c
R R2 R3
r+(tr ) r+(t ) -
r+(t ) + 2 r+(t ) - 3 r+ (t ) . (56)
c 2c 6c
Using the above expansions, the effective Lagrangian can now be expressed as
L = mzi +zi - - V-
m 1 1 1
+ d3x d3x 0 Ji -(x , z , t ) Ji +(x , z , t ) -
4p R 4p 0 2c 2 -
r (x , z , t ) Rr+(x , z , t )

m 1 1
- 0 Ji -(x , z , t ) Ji +(x , z , t ) + r (x , z , t ) R 2r+ (x , z , t )
4p c 4p 0 6c 3 -
1 1 1 1
- r (x , z , t ) r+(x , z , t ) + r (x , z , t ) r+(x , z , t )].
4p 0 - R 4p 0 c -
(57)

11
Eur. J. Phys. 37 (2016) 055202 S Aashish and A Haque

The equation of motion

d L L
= (58)
dt zi - PL zi - PL
of the point charge gives
m0 1 i R
mzi +
4p d3x d3xr (x - z , t ) R t Ji (x , t ) + 2R-1 r (x - z , t )
1 1 2 R2
-
4p 0 c 3 d3x d3x r (x - z , t ) 2 Ji (x , t ) + i r (x - z , t ) = Fiext
t 6
(59)

where
L m0 1 1
= mzi +
zi - PL d3x d3xr (x - z , t ) 4p R Ji (x , z , t ) - c Ji (x , z , t )
L V 1 1
=-
zi - PL xi
- d3x d3x
4p 0 2c 2
r (x - z , t ) i Rr (x - z , t )

1 1 1 1
- r (x - z , t ) i R 2r (x - z , t ) + r (x - z , t ) i r (x - z , t )
4p 0 6c 3 4p 0 R
1 1
- i r (x - z , t ) r (x - z , t )
4p 0 c
1

0 = d3x d3x r (x - z , t ) i r (x - z , t )
R


0 = i d3x d3x r (x - z , t ) r (x - z , t ) .

For a spherically symmetric charge distribution (see pp 7523 of [2])

r i R n + 1 2
J (x , t ) + (x - z , t ) = zi r (x - z , t ) . (60)
t (n + 1)(n + 2) R n - 1 3

By making use of (60) and substituting r (x - z ) = ed (x - z ) and Ji (x , t ) = ezi d (x - z )

in (59), followed by integration over x and x , we obtain the equation of motion of a point
charge as
m e2 m e2
mz + lim 0 z - 0 z = Fext . (61)
R 0 6pR 6p c
m0 e 2
The term which is divergent in the limit R 0 possesses the dimension of mass and
6pR
couples with the bare mass m to give the renormalized mass M,
m0 e2
M = m + lim (62)
R 0 6p R

so that the equation of motion can now be expressed as the famous AbrahamLorentz
equation,

M (z - tz ) = Fext (63)
m e2
where t = 6p0Mc . We note that the AbrahamLorentz equation can also be obtained directly

from (49) by substituting r (x ) = ed (x ) and taking the limit R 0 . Moreover, the details of

12
Eur. J. Phys. 37 (2016) 055202 S Aashish and A Haque

the current section are important in their own right as they enable us to obtain the relevant
effective Lagrangian which describes the effective dynamics of a point charge.
We know that the AbrahamLorentz equation suffers from the problems of causality-
violation and runaway solutions. It has been shown by Rohrlich [16, 32] that if the external
force varies slowly over the size of a charged particle (of the order ct ) such that
d
t Fext Fext (t ) (64)
dt
then the (nite) charge distribution can be approximated as a point charge and the effect of
radiation reaction on the motion of charge is negligible. This allows us to write

Mz Fext (65)
or

Fext
z @ . (66)
M

The equation of motion obtained by substituting for z in (63) is free of the above mentioned
problemscausality-violation and runaway solutionsand is given by

Mz = F + tF .
ext ext (67)
A Lagrangian may now be prescribed for an accelerating point charge in light of equation (67)
as follows

L = Mz+ z- + z- Fext - tz- Fext (68)
with the non-conservative part of the Lagrangian given by

K = z- Fext - tz- Fext . (69)
The prescribed Lagrangian (68) produces the same equation of motion as (67).

4. Conclusion and discussion

We have shown that Hamiltons principle for a non-conservative system such as a nite size
charged particle interacting with a radiation eld furnishes the same equation of motion as
that obtained by Jackson [2]. Moreover, in the limit of a point charge, we obtain the Abra-
hamLorentz equation. We nd that the radiation reaction is derivable from the non-con-
servative part of the effective Lagrangian. A systematic study of the radiation reaction, based
on the interaction of a nite size charged particle with a radiation eld modeled as a heat bath,
has been made in the past [13, 33]. This model affords a way to justify that an accelerating
charge interacting with a radiation eld is a non-conservative system. Since if, for a particle
interacting with a heat bath modeled as a large number of independent harmonic oscillators,
the number of the bath oscillators is about 20 or larger, the Poincar recurrence time (the
amount of time after which a system returns to a state very close to the initial state) turns out
to be innite [34, 35], which renders the dynamics of the interacting particle non-
conservative.

References

[1] Galley C R 2013 Classical mechanics of non-conservative systems Phys. Rev. Lett. 110 174301
[2] Jackson J D 1999 Classical Electrodynamics 3rd edn (New York: Wiley) ch 16

13
Eur. J. Phys. 37 (2016) 055202 S Aashish and A Haque

[3] Barut A O 1980 Electrodynamics and Classical Theory of Fields and Particles (New York:
Dover) ch 5
[4] Schwinger J, DeRaad L L Jr., Milton K, Tsai W and Norton J 1998 Classical Electrodynamics
(Boulder, CO: Westview) ch 32
[5] Boyer T H 1972 Mass renormalization and radiation damping for a charged particle in uniform
circular motion Am. J. Phys. 40 1843
[6] Lenive H, Moniz E J and Sharp D H 1977 Motion of extended charges in classical
electrodynamics Am. J. Phys. 45 75
[7] Grifths D J and Szeto E W 1978 Dumbbell model for the classical radiation reaction Am. J. Phys.
46 244
[8] Templin J D 1998 An approximate method for the direct calculation of radiation reaction Am. J.
Phys. 66 403
[9] Templin J D 1999 Radiation reaction and runaway solutions in acoustics Am. J. Phys. 67 407
[10] Rohrlich F 2000 The self-force and radiation reaction Am. J. Phys. 68 1109
[11] Cook R J 1984 Radiation reaction revisited Am. J. Phys. 52 894
[12] Valentini A 1988 Resolution of causality violation in the classical radiation reaction Phys. Rev.
Lett. 61 17
[13] Ford G W and OConnell R F 1991 Radiation reaction in electrodynamics and the elimination of
runaway solutions Phys. Lett. A 157 4 5
[14] Ford G W and OConnell R F 1993 Relativistic form of radiation reaction Phys. Lett. A 174 182
[15] Spohn H 2000 The critical manifold of the LorentzDirac equation Europhys. Lett. 50 287
[16] Rohrlich F 2008 Dynamics of a charged particle Phys. Rev. E 77 046609
[17] Grifths D J, Proctor T C and Schroeter D F 2010 AbrahamLorentz versus LandauLifshitz Am.
J. Phys. 78 391
[18] Haque A 2014 A simple derivation of Lorentz self-force Eur. J. Phys. 35 5
[19] Aashish S and Haque A 2015 Average Lorentz self-force from electric eld lines Eur. J. Phys. 36
055012
[20] Steane A M 2015 Reduced-order AbrahamLorentzDirac equation and the consistency of
classical electromagnetism Am. J. Phys. 83 256
[21] Steane A M 2015 Tracking the radiation reaction energy when charged bodies accelerate Am. J.
Phys. 83 703
[22] Bateman H 1931 On dissipative systems and related variational principles Phys. Rev. 38 815
[23] Barone P M V B and Mendes A C R 2007 Lagrangian description of the radiation damping Phys.
Lett. A 364 438
[24] Bender C M and Gianfreda M 2015 PT-symmetric interpretation of the electromagnetic self-force
J. Phys. A: Math. Theor. 48 34FT01
[25] Polonyi J 2014 Effective dynamics of a classical point charge Ann. Phys. 342 239
[26] Grifths D J 2012 Introduction to Electrodynamics 4th edn (Harlow: Pearson) sections 11.2.2
and 11.2.3
[27] Staruszkiewicz A 1970 An example of a consistent relativistic mechanics of point particles Ann.
Phys. (NY) 480 362
[28] Jaranowski P and Schfer G 1997 Radiative 3.5 post-Newtonian ADM Hamiltonian for many-
body point-mass systems Phys. Rev. D 55 4712
[29] Knigsdrffer C, Faye G and Schfer G 2003 Binary black-hole dynamics at the third-and-a-half
post-Newtonian order in the ADM formalism Phys. Rev. D 68 044004
[30] Zangwill A 2013 Modern Electrodynamics 1st edn (Cambridge: Cambridge University Press)
section 24.3
[31] Landau L D and Lifshitz E M 1975 The Classical Theory of Fields (Course of Theoretical Physics
Series vol 2) 4th edn (Amsterdam: Elsevier) section 65,75
[32] Rohrlich F 2002 Dynamics of classical quasi-point charge Phys. Lett. A 303 30710
[33] OConnell R F 2012 Radiation reaction: general approach and applications, especially to
electrodynamics Contemp. Phys. 53 4
[34] Weiss U 2008 Quantum Dissipative Systems 3rd edn (Singapore: World Scientic)
[35] Griener W 2003 Classical Mechanics: Systems of Particles and Hamiltonian Dynamics (New
York: Springer)

14