Chapter 12 Dynamics of Relativistic Particles & EM Fields

12.1 Lagrangian & Hamiltonian for a Relativistic Charged Particle

dp u in External EM Fields
= e [E
× B] 
dt c d U e p
 = F   U  & 
U =  c ,  u=
dE d mc m
= e uE
dt
It is useful to consider the formulation of the dynamics from the viewpoint of
Lagrangian and Hamiltonian mechanics.

The principle of least action: the motion of a mechanical system is such that in
going from one configuration at one time to another configuration at another
time, the action is an extremum.
t2 d L L
A =
t L [qi t , q i t  , t ] d t   A=0   = 0 Euler-Lagrange
1
d t  q i  qi eqn of motion
wish to extend the formalism to relativistic motion in a manner consistent with
SR and leading for charged particles in external fields to the right eqns.

A. Elementary Approach to a Relativistic Lagrangian

From the 1st postulate of SR the action integral must be a Lorentz scalar
because the eqns of motion are determined by the extremum condition.
 t2 2
A=
t L d t =
 L d   L is Lorentz invariant & A are invariant
1 1

The Lagrangian for a free particle can be a function of its velocity & mass, but
not of its position. The only Lorentz invariant of the velocity available is UU=c2

  L free  c 2  L free =
m c2

=m c 2 1
 u2
c 2


If a particle stays at rest initially and after in the frame, the integral over proper
d
dt
 m u =0

time will be larger than if it moves with a nonzero velocity along its path. So a
straight world line gives the maximum integral over proper time.

This motion at constant velocity is the solution of the free-particle eqn of motion.

for a relativistic charged particle in external EM fields

interaction L  L nonrel =e  e   e
int int
 L int = U A =e 
uA
part c c


2
u e d 
 L =m c2 1  2
 e 
uA  Lorentz force law =
u
c c dt t
canonical P   L =  m u
e A  P = p
e A p =  m u
j j j
momentum  uj c c
c Pe A

Hamiltonian H  Pu  L = c P e A
m c
e  2 2 4
u=
P  e A /c
m
2 2
c
2
 total energy W = H  W  e 2  c P e A2 = m 2 c 4
2 2 2 E W e  e
 p = m c p =  , p=  , P A
c c c
the total energy W/c acts as the time component of a canonically conjugate 4-
momentum P of which P is the space part.
the eqns of motion are invariant under a gauge transformation of the potentials.

Since the Lagrangian involves the potentials, it is not invariant. But the change
in the Lagrangian is of such a form (a total time derivative) that it does not alter
the action integral or the eqns of motion. [Problem 12.2]

B. Manifestly Covariant Treatment of the Relativistic Lagrangian

2
mc
L free =

U U

 A=m c
U 1
 U 
d

dU
 2 = U  U  = c2
the eqn of constraint: U  
U =0
d
can be incorporated by the Lagrange multiplier method, but we try another way.

 d x  d x
 d x d x
s2

U 

U d =
d d

d = g   d x d x
 
 A=m c

s1
g 
ds ds
ds *

 g 
d x d x 
ds ds
d s=c d   2 = c2
U
 d d x / d s d2 x
*  mc =0  m =0 free particle motion


2
ds d x d x  d
ds ds
For a charged particle in an external field

[  d x d x  d x
]
s2
e
A=
s1
m c g 
ds ds

c
A  x
ds
ds


d 
L
d s  d x / d s 
 L

d 2 x  e d A e d x   

 =0
ds ds c

ds
 =m c g
L 
 d x d x 

e
A
d x

 m 2

  A =0
d c d c d
d2 x  e     d x e  & d A d x   
 m 2
=  A   A  = F U= =  A
d c d c m d d


2 2
L e 1 e e

P = =m U

A 
  = P U  L
H P 
 = 

P 
A  c   A
 d x  / d s c m c c

dx H  1  e 
=
= P  A  2
d  P m c e
P 
  0W
2 2
 A = m c  H
d P H  e e c
= = P   A   A
d  x m c c
12.2 Motion in a Uniform, Static Magnetic Field
dp e eB ce
= v ×B B= = B
E=0 dv
 dt c  = v × B mc E
B= const dE v = const dt gyration frequency
=0 
dt  = const precession
A circular motion  to
& a uniform translation
to B

 e1  i e 2 
v  e 3 & counterclock rotation (for positive charge)
i  B t
 v t = a  B e
a : gyration radius
helix radius = a  a e B = c p 
i  t
 x t = X0
a e i e1
e 2 
v  t e 3 % 
B
v
1
pitch angle  = tan
a B
This form is convenient for the determination of particle momenta.

For particles with charge the same in magnitude as the electronic charge
p  MeV /c= 3.00 ×104 B a gauss-cm = 300 B a Tesla-m 
12.3 Motion in Combined, Uniform, Static E & B Fields
u
Let E  B E '  = 0, E '  = u E
× B= 0
c
E×B 
choose u =c

2
B E
B2 B '  = 0, B '  = = 1 2 B EB
u B
d p' v' e
 Lorentz force eqn = e E '
× B ' = v ' × B in K '
dt' c c u
In K' the only field acting is a static B' pointing in the same direction as B, but
weaker by a factor -1.
 The drift velocity has physical meaning only if it is less than c, ie, if |E|<|B|.

If |E|>|B|, E is so strong that the particle is continually accelerated in the

direction of E and its average energy continues to increase with time

For E ! B

choose u ' = c
E×B 
E ” =0, E ”  =
E
u ' 
= 1

u'
B2
E 2
E

E2 B ”  = 0, B ”  = u ' B × E= 0

c
In K” the particle is acted on by a purely E” which causes hyperbolic motion with
ever-increasing velocity.

If a beam of particles having a spread in velocities is normally incident on a

region containing uniform crossed E & B, only those particles with velocities
equal to cE/B will travel without deflection.
Suitable entrance and exit slits will allow only a very narrow band of velocities
around cE/B to be transmitted.
Combined with momentum selectors, like a deflecting magnet, the E
velocity
selectors can extract a very pure & monoenergetic beam of particles of a definite
mass from a mixed beam with different masses and momenta
commonly used
in high-energy accelerators.
 If E has a component parallel to B, the behavior of the particle cannot be
understood in such simple term
EB and E 2  B2 are the only 2 Lorentzinvariants

E B  EB= 0  " a Lorentz frame where E = 0 if B! E

B= 0 if E ! B
If E

0, E & B will exist simultaneously in all Lorentz frames. Consequently
motion in combined fields must be considered.
12.4 Particle Drifts in Nonuniform, Static B Fields
Often the variations are gentle enough that a perturbation solution to the motion
is an adequate approximation.

consider a gradient  to the direction of B  nB = 0

expansion about the origin of
  B  x =
e
mc [
B x   0 1
# B0
nx
B0 ] coordinates where  = 0
B

# : coordinate in the direction n

Since the direction of
is unchanged, the motion
to
remains a uniform
translation. We then consider only modifications in the transverse motion.
v  = v0
v 1
d v
dt
= v  ×  B x 

d v1
dt [
 v1
v0 # B0
nx 0
B0 ] × 0
d v0
dt
= v0 ×  0

v 0 = 0 ×x 0  X
&  X : center of gyration X = 0 here 
 0 × v0
% x 0  X = 2
0
# B0

d v1
dt [
 v1  v 0 # B0
nx 0
B0 ]
 0 × x0 ×  0 
v G $ v1 % =
B0
 0 ×$ x 0  nx 0 %

gradient drift velocity

 the rectangular components of x0 oscillate sinusoidally with peak amplitude a
and a phase difference of 90°, so only the component of x0
to n contributes

a2 a2 # B 0 vG a coordinate
 $ x 0  nx 0 %= n  vG =  0 ×n  = 2
B ×  
B
2 2 B0 B a 2B indep.

The particle tends to spiral around a field line, but the field line curves off to the
side. This is equivalent to a centrifugal acceleration of magnitude v
2/R.
 The acceleration can be viewed as arising from an effective electric field
2 m R × B0 v 2
v
m  vC = c v 2 R × B0 e B0
E eff = 2
R  e
 2
R B
2  vC = $  B =
e R 0
 B R R B0 mc
curvature drift velocity
The sign in the eqn is for positive charges and is independent of the sign of v
.
For negative particles the opposite sign arises from B.

A straightforward derivation comes from solving the Lorentz force eqn directly.

with origin at the center of curvature, B has only  &-component, B&= B0(R/').
 ' &
 0
¨ 2 ' &=   R v2 = const
'2 &=
Lorentz R '
z̈ =  B '  z =  B ln
v0 (  B x
v0 '= R
x
forece  ' R
eqn R v2 v 2
¨ ' & = B
2
' z  ẍ
2B
3 2
 x(   B v0
' R R
2 2
v v0 v
 $ x%( 2
 v ) B R  $ z % ( v 0
 B $ x %(  $
 R
B
B B R
 B R
If J = 0   × B =0  = & v  =  B a transverse velocity
B R2 of gyration
2 v 2
v 2 R × B 172 T K 
 v D = vG
vC =  v D cm / s= for nonrelativistic
2 B R RB R  m B gauss charged particle
For a toroidal tube with a strong field, the plasmas inside will drift out to the
walls in a short time. Hotter the plasmas, greater the drift rate.

One way to prevent this 1st-order drift in toroidal geometries is to twist the torus
into a 8.

The particles make many circuits around the closed path, so they feel no net
curvature or gradient of the field, and no net drift, at least to 1st order in 1/R.
12.5 Adiabatic Invariance of Flux Through Orbit of Particle
we now consider motion parallel to the lines of force.

for slowly varying fields a powerful tool is the concept of adiabatic invariants.

If qi and pi are the generalized canonical coordinates & momenta, and for each
coordinate which is periodic, the action integral is defined by

Ji *p dqi i
over a complete cycle of qi

For a given mechanical system the action integrals are constants.

if a change is slow compared to the periods of motion and is not related to the
periods (ie, adiabatic change), the action integrals are invariant.

One system can be changed into another system with an adiabatic change, but
the values of the action integrals have the same values in both systems.

For a charged particle in a uniform, static B, the transverse motion is periodic.

e e
J= * *
P d + =  m v  d +

c
* Ad + P=p

c
A

e e
= * 2
 m B a d ,

c
* 2
Ad + =2 -  m  B a

c

Bn d a
S
v  d +
 2 e
 J = -  m B a = B - a2 n is antiparallel to B
c
В-a2 is the flux through the particle's orbit.
If the particle moves through regions where B varies slowly, the adiabatic
invariance of J means that the flux linked by the particle's orbit remains constant.

If
increases, a will decrease so that В-a2 remains unchanged.

]
B a2 e B a2
2
are adiabatic invariants .= magnetic moment
p /B
2c of the current loop
.

If B= const  v = const  energy = const  . is itself an adiabatic invariant

v = v  er
v e z 2 2 2 2 2
 v 
v  = v 0 = v 0
v  0 = const
v0 = v z = 0= v  0 e r
v 0 e z
2 2
v v0 B z
= invariant  v 2 = v20  v2 0 B = Be r axial magnetic induction
B B0 B0

2 2
e e 2
v0 2 2
v0
 z̈ =  B' 
' & ' &  z B  
 z B ' &a  B 0 =
mc 2mc 2 B0 B 0
 To 1st order in small quantities the constancy of flux linking the orbit follows
directly from the eqns of motion.

The adiabatic invariance of the flux linking an orbit is useful in particle motions
in all types of spatially varying magnetic fields.
12.6 Lowest Order Relativistic Corrections to the Lagrangian for
Interacting Charge Particles: The Darwin Lagrangian
When the finite velocity of propagation of EM fields is taken into account, this is
possible that the Lagrangian is a function of the instantaneous velocities and
coordinates of all the particles, since the values of the potentials at one particle
due to the other particles depend on their state of motion at “retarded” times.

consider a conventional Lagrangian of the interaction of two or more charged

particles with each other, and it is possible only at nonrelativistic velocities.
NR q 1 q2
L int
= =q 1 12
r
to generalize beyond the static limit, we must determine both 12 and A12.

In general there are relativistic corrections to 12 & A12. But in the Coulomb
gauge, the scalar potential is given correctly to all orders in v/c by the Coulomb
potential. Thus, all that needs to be considered is the vector potential A12.
 3
J  x '  d x '
J t x ' = q2 v 2  x '  x 2 
q2
'
v2x '  x 2 

J t  x  =  ×  ×
4 -x  x '
4- x '  x 23
J x= q 2 v 2  x  x2 
q 2 v2 q2 d x'
3
v 2 x '  x 2 
 A12 
cr


4 - c x '  x
'
x '  x 2 3
/ y = x '  x2

=
q2 v 2
cr

4-c
q2
r

v 2y d3 y
y
3
y  r
=
q2 v 2
c [ r
 r
v 2r
2r ] [
=
q2
2cr
v2

v2r
r
2
r
]
 L int =e 

e
c
uA
q1 q 2
r [ 1

1
2c 2 
important in a quantum mechanics of relativistic corrections in 2-electron atom.
v 1v 2

v 1r v 2r 
r 2 ] Darwin 1920

Breit interaction:
interaction replace the velocity vectors with their corresponding quantum-
mechanical operators.

For a system of particles, correct to order 1/c2 inclusive,

L Darwin =
1
2
0 2
mi v 1

i [ v 2i
4 c2 ] 
1
2
0 '
q1 q2
ri j [ 1
v iv j
v i1r  v j1r 
2 c2 ] 0 ' : no self-
energy

the Darwin Lagrangian has uses usually in the purely classical domain.
12.7 Lagrangian for the EM Field
The Lagrangian approach to continuous fields closely parallels the techniques
used for discrete point particles. The finite number of coordinates are replaced by
an infinite number of degrees of freedom. Each point in space-time corresponds
to a finite number of values of the discrete index. The generalized coordinate is
replaced by a continuous field. The generalized velocity is replaced by the 4-
vector gradient
i  x , k L= 0 L i  qi , q i  
2 &k ,   &k  d 3 x
qi  &k x   2 : Lagrangian density
d L 2 2
   =
q i   & k  x d t  q i   &k 
 &k
1
 Action A= 6 3
2 d x d t=
c

2 d4 x
The Lorentz-invariant nature of the action is preserved provided the Lagrangian
density is a Lorentz scalar since the 4d volume element is invariant.
expect the free-field Lagrangian to be quadratic in the velocities, a scalar under
proper Lorentz transformations, and the interaction involves the source densities
1  1 
g. 3 g4 5 1
 2 = F  F  J  A =  3 A5   5 A3  . A4   4 A.  J  A
16 - c 16 - c
2 g. 3 g4 5 3 5 5 3 .4 . 4 4 . 35 1
  
= [         F
          F ]= F 
 A 16 - 4-
 2 1 4-

= J   F   = J  # inhomogeneous Maxwell eqns
A c c
the definition of the field strength tensor F in terms of the 4-vector potential A3
was chosen so that the homogeneous equations were satisfied automatically.
1
 7   =   8  . 4 F . 4 =   8  . 4 . A4 = 8  . 4   . A4 = 0
2
The conservation of the source current density can be obtained from (#)
4-
0    F   =  J    J  = 0
c
12.8 Proca Lagrangian; Photon Mass Effects
The conventional Maxwell eqns and the Lagrangian are based on the hypothesis
that the photon has zero mass.

the Proca Lagrangian: add a “mass” term into the Lagrangian

1  .2 1 m c reciprocal Compton
2 Proca = F  F
A A  J  A .=
16 - 8- c : wavelength of the photon
4-
   F  
.2 A = J ,  7   = 0
c
In the Proca eqns the potentials as well as the fields enter, thus the potentials
acquire real physical significance through the mass term.
4-
Lorentz gauge  4 A4 = 0   F  
.2 A =9 A
.2 A = J
c
2 2 4- 0
  A  . A = J  static limit  A =     x  for a rest point charge
c
e. r
   x = q spherically symmetric Yukawa form
r
the exponential factor alters the character of the earth's B sufficiently to permit
us to set quite stringent limits on the photon mass.
 absence of sources  9 A
. 2 A =0  2 = c 2 k 2
.2 c2 square of energy / :
2 2 2 2
consider some resonant system (cavity/lumped circuit)  = 0
. c 0 = c k

measure the difference between  and 0 in a circuit for a given photon mass.

However, lumped circuits are incapable of setting any limit on the photon mass.

for a solid conducting sphere of radius a at the center of a hollow conducting

shell of inner radius b held at zero potential, the capacitance is increased by
.2 a 2 b = .2 c 2 = 1
= C= for . b ) 1    = O . 2 d 2  0 =
3 0 2 0
2
0 L C
very insensitive in practice to a possible photon mass.

for .=0 the  modes of a transmission line are degenerate modes, with
propagation at a phase velocity=c. The situation does not alter if .0. The only
difference is that the transverse behavior of the fields is governed by (;t2-.2)<=0
instead of the Laplace eqn.

TBA
12.9 Effective "Photon" Mass in Superconductivity; London
Penetration Depth
effect the expulsion of B from the interior of a superconductor as it
Meissner effect:
transits from the normal state (T>Tc) to the superconducting state (T<Tc).

If B is applied after the material is superconducting, it penetrates a very small

distance called the London penetration depth (~10 nm).

Being a perfect conductor, a superconductor is perfectly diamagnetic.

Q Q2 Q
J = Q nQ v = nQ P  nQ A P = mQ v
A
mQ mQ c c
The superconducting state is a coherent Q2
 J = nQ A
state of the charge carriers with P =0 mQ c
4-  4 - n Q Q2
9
2 2 2 2
A= J   A  0 A  . A = 0 . =
c mQ c 2
no current flows across the interface between normal and superconducting
media, so the normal component of A vanishes.

For  t A =0
planar symmetry  A  e±. x  3 L = .1 =
 mQ c 2
4 - nQ Q 2
  
: Q 4 - n Q a 30 e2
 the effective photon mass m , eff =  m  , eff c2 = ~ few eVs
3L c e mQ / m e a 0
the charge carriers in low-temperature superconductors are pairs of electrons
loosely bound by a 2nd-order interaction through lattice phonons
Q =2 e n eff 22 3 6 .2 =8 - r 0 n Q
 , nQ = = O 10 cm   3 L = O 4 × 10 cm 
mQ = 2 m e 2 r 0 = electron radius

n eff E F : the Fermi energy

2
BCS theory  n Q T = 0= = E F N 0
2 3 N 0 : density of state at
the Fermi surface
in high-temperature superconductors penetration depths are found to be an
order of magnitude smaller than in conventional superconductors.

Measurements of 3L(T) can be done by incorporating the superconductor into a

resonant circuit and studying the shift in resonant frequency with change in
temperature. 2 3 2-3
8- L L
Z s (i Gaussian units=i Z 0 SI units
c 3 3
the impedance is inductive, corresponding to an inductance per unit area,L=.03L.
12.10 Canonical and Symmetric Stress Tensors; Conservation Laws
A. Generalization of the Hamiltonian: Canonical Stress Tensor
L dH L
pi = i
 H = pi q i  L  = 0 if =0
 q dt t

Hamiltonian density: H =
H d 3
x

H should
Since the energy of a particle is the time component of a 4-vector,
transform in the same way. Since the invariant 4-volume element is d4x=d3xdx0, H
transform as the time-time component of a 2nd-rank tensor.
2
 H = t &k  2 vs H = pi q i  L
 t & k
The inferred Lorentz transformation properties of H suggest that the covariant
generalization of the Hamiltonian density is the canonical stress tensor:
2
T  &k  g   2
  &k
1 .4 E 2  B2
For the free EM field Lagrangian 2 em = F. 4 F =
16 - 8-

 2 em  3  1   3 
 T  3
 A  g 2 em
= F 3  A  g 2 em
  A 4-
 8 - T 00 =E 2
B2 
2  E
 4 - T 0 i =E × Bi
 Ai E E = 0 &  × B = 0 E
4 - T i 0 =E × B
 ×  Bi  0  E i 
suppose that the fields are localized in some finite region of space
1

T 00 d 3 x =
8-

E 2
B2  d 3 x = E field total energy of the fields

1

T
0i 3
d x=
4-

i 3
 E × B d x = c P field
i
linear momentum of the fields

The differential conservation statement:   T =0

Proof:  T   = 
 
[ 2
  &k


] 
 &k   2 =  &k 
 2
  &k

2
  &k

  &k  2


2  2
=  &k
   &k   2 =   2 &k ,   &k    2 =0
 &k   &k
The conservation law or continuity eqn yields the conservation of total energy
and momentum upon integration over all of 3-space at fixed time
d d
0=
  T   d 3 x = 0
T 0  d3 x

i T i  d3 x 
dt
E field = 0,
dt
P field = 0
B. Symmetric Stress Tensor
i
Deficiencies (1) T00 & T0 differ from the usual expressions for E and P densities.
(2) lack of symmetry
(3) it involves the potentials explicitly, and so is not gauge invariant
(4) its trace is not zero, as required for zero-mass photons.
1
the angular momentum of the field L field =
4-c

x × E × B d 3 x


its covariant generalization
3rd-rank tensor M = T   x  T   x
 M    = 0  conservation of the total angular momentum of the field
       
 0 = x  T
T  x  T T =T T
 conservation of angular momentum requires T   = T  
1 1 1 1
T   = F  3  A3  g  2 em = [ F 3 F3 
g   F . 4 F . 4 ] F  3 3 A
4- 4- 4 4-
 1  3  1 3  1 3   3 3
 T D  F 3 A = F 3 A = F  3 A
A  3 F  3 F =0
4- 4- 4-
1
=
4-
3 F 3  A     T D  = 0,
T 
D
d3 x = 0

 symmetric stress tensor > = T

  
T D =
1
4- [ F

3 F

3 1
4
g

F. 4 F
.4
]
 1
>00 =  E 2
B2   u : energy density
8-
0i 1 i
 > = E × Bi  c g : momentum density
4-

>i j =
4-
1
[ E i E j
Bi B j 
1
2 ]
i j  E 2
B2    T iMj : Maxwell stress tnesor

 
> =
[ u
c g T
cg
M
i j
] [
, >  =
u
c g T
1
c g
M
i j
] [
, > 
 =
u
cg
c g
T M
i j
] [
, > =
u cg
c g T iMj ]
0 2
0=  > =  t u
S S = c g : Poynting vector
   >  = 0  c
i i j M
0=  > =  t g  T i j 6.121 both for source-free
      
 M => x > x   M =0 angular momentum conservation
 The conservation of M 00 i is a statement on the center of mass motion
. Conservation Laws for EM Fields Interacting with Charged Particles
 1 1
4 -  >  =  .  F . 4 F 4  
   F . 4 F . 4 = F 4   . F . 4
F . 4  . F 4 
F . 4  F . 4
4 2
 1 4 1 . 4 . 4  .4 1 . 1
  >
F J4= F . 4  F
 F
 F   F. 4 = J4
c 8- 4- c
1
= F . 4  . F 4 
 4 F .  = 0 . F 4 
  F . 4
 4 F  . = 0
8-
1 1 conservation of energy
 t u
S= JE
1
  >  = F 4 J4  c c & momentum for EM
c 1 fields with sources
t gi   j T iMj =' E i   J × Bi 
c J =c ' , J 
 1 4 1
 f  F J4 =  JE , c ' E
J × B Lorentz force density
c c
d dp u dE
For the sources are discrete

f d x=
3

dt

P particle
dt
=e [E

c
× B] ,
dt
= e uE

d


  >
f  d x =
 3

dt
 
 P field
P particle = 0 conservation of 4-momentum for
the system of particles and fields
A more equitable treatment of a combined system of particles and fields
2 = 2 free-field
2 free-particle
2 interaction
12.11 Solution of the Wave Eqn in Covariant Form; Invariant Green
Functions
4- 4-
 F   = J  9 A     A =9 A = J   x   A =0
c c
The solution can be accomplished by finding a Green function for the eqn
9x D  x , x ' = 4   x  x ' =  x 0  x ' 0    x  x ' 
In the absence of boundary surfaces, the Green function can depend only on the
4-vector difference
z = x  x '  D  x , x ' = D  x  x ' = D  z   9z D  z = 4   z 
1
 D  z =
2 -

4
 
D k e
i kz 4
d k kz = k 0 z 0  kz

   k = 1
D 4   z  =
1


ei kz d 4 k
 k
k 2 -4
 i k 0 z 0
1 ei kz 4 1 @
e
 D  z =
2 -

4 kk
d k =
2 -4

e
i kz
@ 2
k ?
0
2
d k0 d3 k ?=k

The k0 integrand has two simple poles at k0=±?.

Green functions that differ in their behavior are obtained by choosing different
contours of integration relative to the poles.
 @
, z 0  sin ? z0 , z 0 
 Dr  z =
2 -

3
?
e i kz 3
d k=
2- 2
R

sin ? R sin ? z  d ?
0
0 R=z

@
, z 0  ,  z0 
=
8- 2
R

@
[e
i ?  z 0  R
e
i ?  z 0
R
] d ?=
4-R
[ z0  R  z0
R]

,  x0  x ' 0  retarded causal  Green function

 Dr  x  x ' =   x0  x ' 0  R  the source time x ' 0 is always earlier
4-R
than the observation time x 0
  ei  R / c
D = Fourier transform of D r  x  x '  with respect to x 0 vs chapter 6
4-R
,  x ' 0  x0 advanced Green function
Da  x  x '  =   x 0  x ' 0
R
4-R choosing the contour a
These Green functions can be put in covariant form

 [ x  x ' 2 ]=  [ x 0  x ' 0 2 x  x '2 ]=  [ x 0  x ' 0  R  x 0  x ' 0

R]
1
= [  x 0  x ' 0  R
  x 0  x ' 0
R]
2R
1
Dr  x  x '  = ,  x 0  x ' 0   [ x  x ' 2 ]
 2- explicitly invariant expression
1
Da  x  x ' = ,  x ' 0  x 0   [ x  x ' 2 ]
2-
The , functions, apparently noninvariant, are actually invariant under proper
Lorentz transformations when constrained by the delta functions.

the retarded (advanced) Green function is different from zero only on the
forward (backward) light cone of the source point.
4-

A  x  = 
Ain  x 

c

D  x  x '  J  x '  d
r
4
x' * 1
The solution of the wave eqn
4-

A  x  = 
Aout  x 

c

D  x  x '  J  x '  d
a
4
x' * 2
 In the limit x0A-B, the integral in (*1) vanishes, assuming the sources are
localized in space and time, the retarded nature of the Green function.


Ain : the incident or incoming potential specified at x 0   @

Aout : the asymptotic outgoing potential specified at x 0 
@

The radiation fields: difference between the outgoing and the incoming fields.
4-

Arad  x = 
Aout  x  
Ain  x =
c

D  x  x '  J  x '  d 4 x ' D  z = Dr  z  Da  z 

For a charged particle ' x , t =e  [x  r t ] r is the position in frame K

J x , t = e v t   [ x  r t] v = r is the velocity
can be written as a 4-vector current in manifestly covariant form
J  x = e c
U   
4 
[ x  r  ] d r =[c t , r t ] in the inertial frame K
U =  c ,  v