Lecture 18

M. Siddikov

[email protected]

November 6, 2023
Outline Today’s plan:
Dipole emission (electric and magnetic).
. − Various problems
. − Multipole emission
Emission of a pointlike charge in motion
−Discussed at the end of block 2 (Lienard-
Wiechert potentials, Larmor’s formula).
Please see Lectures 10, 11 for details.

Special Relativity
. − Recap of main concepts

Certamen 2
−Tomorrow !!!
Summary of last lecture
Block 2: EM waves are emitted if a charge moves with acceleration, a = r̈ ̸= 0

Last lecture:
. Microscopically, may have partial cancellations of emis-
sions of individual charges.

In some systems (e.g. rotating sphere) no emission at

all, if locally ρ, j don’t depend on time

−Discussion in terms of “individual” emissions NOT very effective. Need to

analyze the system as a whole.
ˆ ˆ
1 ρ (r ′ , tr ) µ0 j (r ′ , tr )
ϕ (r , tr ) = dV ′ , A = dV ′
4πε0 |r − r ′ | 4π |r − r ′ |
′ ′
|r − r | |r | n · r
tr = t − ≈t− + ,
c | {z c} c
t0

∗
Emission terms correspond to expansion of |r − r ′ | in numerator (tr ).
Denominator expansion gives contirbutions which are negligible at large
distances r ≫ r ′ .
Summary of last lecture (dipole emission)
Magnetic dipole emission: Compare: electric dipole:

⃗ ⃗¨   n 1
⃗ = − ∂ A = µ0⃗n × M (t0 ) +O 1
E
B≈−
c
× Ȧ = −
4πε0 c 3 r
n × P̈ (t0 ) ,
∂t 4πc r r2
1 h i
E ≈ n × n × Ȧ = n × B
B = rotA ≈ n × E c
Both E and B are transverse to direc-
2 tion of propagation ⃗n
µ0  M̈ (t0 ) 1 1
S= 2 3
sin2 θ n S≈ E ×B = |B|2 n
| 16π
{z c } r2 µ0 µ0
const
2
˛

2 1 1 2
 µ 
0
2 |B| = P̈ (t0 ) sin2 θ
I = r2 dΩ S · n = 3
M̈ (t0 ) 4πε0 c 3 r2
6πc
| {z } | {z }
const
const

Spatial distribution of flux density is sim- Total emission (power per unit of
ilar, but polarization of (E , B) is different time):
.  electric dipole: Eθ , Bφ  µ 
0
2
 magnetic dipole: Bθ , Eφ I = P̈ (t0 )
6πc
| {z }
const

where θ is angle between n and P̈ (t0 )

Control problem

Electric dipole: Magnetic dipole emission:

µ0 1 2
S≈ P̈ (t0 ) sin2 θ n 2
16π 2 c 2 r 2 M̈ (t0 )
µ0
S= 2 4
sin2 θ n
16π c r2

Assume that all charges in the system move with small nonrelativistic velocities
(|ṙ i | ≪ c). Try to estimate which contribution is more important, electric or
magnetic dipole emission?
X
ρ(r , t) = qi δ 3 (r − r i (t))
i
X
j (r , t) = qi ṙ i (t) δ 3 (r − r i (t))
i
Multipole emission
From: Jackson, “Classical Electrodynamics”, Chapter 9
−We obtained results for electric and magnetic dipole emission making direct expansion
∗
This method becomes quite tedious when we go to higher orders (quadrupole,
octupole ...)

−Now we’ll consider very special case, when time dependence is periodic,

ρ(r , t) = ρ(r , ω)e −iωt , J(r , t) = J(r , ω)e −iωt ,

and will get exact results for multipole emission

∗
For any smooth function of time t, can use Fourier expansion:
ˆ ˆ
dω dω
ρ(r , t) = ρ(r , ω) e −iωt J(r , t) = J(r , ω) e −iωt (1)
2π 2π
ρ̇+divj = 0 ⇒ ρ(r , ω) = −iω −1 divJ(r , ω)

and in view of linearity of Maxwell’s equations, can find the total field as a linear
superposition of each ω
Multipole emission for periodic dependence

ρ(r , t) = ρ(r , ω)e −iωt , J(r , t) = J(r , ω)e −iωt ,

|r − r ′ | |r | n · r ′
tr = t − ≈t− + ,
c | {z c} | {z
c }
t0 δt
X (−iω δt)N
e −iωtr = e −iω(t0 +δt) = e −iωt0 e −iωδt = e −iωt0
N!
N

ˆ ˆ
1 ρ (r ′ , tr ) e −iωt+ikr X (−ik)N
N
dV ′ dV ′ ρ r ′ , ω n · r ′


ϕ (r , tr ) = ≈
4πε0 |r − r ′ | 4πε0 r N!
N
ˆ ˆ
µ0 ′
j (r , tr ) µ0 e −iωt+ikr X
(−ik)N
N
dV ′ dV ′ J r ′ , ω n · r ′


A (r , tr ) = ′
≈
4π |r − r | 4πr N!
N

−Formal expansion parameter: kr ′ ∼ 2π |r ′ |/λ ≪ 1, so the series converges rapidly

−Different terms correspond to dipole, quadrupole, ... contributions
−Get outgoing (spherically divergent) waves ∼ e −iω(t−r /c) /r multiplied by the n-
dependent amplitude (orientation w.r.t. corresponding multipole moment matters)
Multipole emission
ˆ ˆ
1 ρ (r ′ , tr ) e −iωt+ikr X (−ik)N
N
dV ′ dV ′ ρ r ′ , ω n · r ′


ϕ (r , tr ) = ≈
4πε0 |r − r ′ | 4πε0 r N!
N

ˆ ˆ
µ0 j (r ′ , tr ) µ0 e −iωt+ikr X (−ik)N
N
dV ′ dV ′ J r ′ , ω n · r ′


A (r , tr ) = ≈
4π |r − r ′ | 4πr N!
N

So far we considered the so-called far/emission zone, where ⟨|r ′ |⟩ ≪ λ ≪ r

−we disregarded r ′ /r from expansion of denominator compared to r ′ /λ from expansion
of numerator
Now we are going to consider other regions:
a) near (static) zone,
⟨|r ′ |⟩ ≪ r ≪ λ (2)
b) intermediate (induction) zone,

⟨|r ′ |⟩ ≪ r ∼ λ (3)

 In the next slides we’ll analyze in detail those regions and try to get analytic results
for potentials
Near (static) zone ⟨|r ′ |⟩ ≪ r ≪ λ
Can assume that

|r − r ′ | |r |
tr = t − ≈t− , (4)
c | {z c}
t0

′
since “extra” term ∼ (n · r )/c in expansion of tr gives terms which behave as

∼ (ω/c)N ∼ 1/λN ,

whereas expansion of denominators gives terms which behave as ∼ 1/r N ≫ 1/λN

⇒ all points contribute at the same retarded time t0
⇒can just use static solution from Lecture 4,
ˆ
1 X 4π Yℓm (θ, φ)
ℓ
dV ′ r ′ Yℓm (θ′ , φ′ )ρ r ′ , t0


ϕ(r , t) = ℓ+1
4πε0 2ℓ + 1 r
ℓm

ˆ
µ0 X 4π Yℓm (θ, φ)
ℓ
dV ′ r ′ Yℓm (θ′ , φ′ )J r ′ , t0


A(r , t) =
4π 2ℓ + 1 r ℓ+1
ℓm

−fields are “quasistationary” (differ from stationary case only by t → t0 ).

Intermediate (induction) zone ⟨|r ′ |⟩ ≪ r ∼ λ
Can’t assume neither kr ≫ 1 nor kr ≪ 1
Can decompose
′
1 e ik|r −r | X
jℓ (kr ′ )hℓ (kr )Yℓm
∗
(θ′ , φ′ )Yℓm (θ, φ)
(1)
G = ′
= ik (1)
4π |r − r |
ℓm

Proof: (Similar to what we did for static case in Lecture 3)

The function G is the solution of the equation

∆ + k2 G = δ r − r ′

−for r ̸= r ′ rhs is zero, homogeneous Helmholtz

−expect G = G (r − r ′ ) and thus

∆r ′ + k 2 G = ∆r + k 2 G = δ r − r ′

(subindex r ′ reminds that ∆ acts on r ′ -coordinates)

⇒Can solve in spherical coordinates using separation of variables, seeking solution as

G = Rℓm (r )Rℓm (r ′ )Yℓm

∗
(θ′ , φ′ )Yℓm (θ, φ)
(1) (2)

−Solutions for radial parts for r ≫ r ′ should satisfy border conditions:

a) limr →∞ ∼ e ikr b) limr ′ →0 < ∞ ⇒This gives (1).
Intermediate (induction) zone ⟨|r ′ |⟩ ≪ r ∼ λ

′
1 e ik|r −r | X
jℓ (kr ′ )hℓ (kr )Yℓm
∗
(θ′ , φ′ )Yℓm (θ, φ)
(1)
G = ′
= ik (1)
4π |r − r |
ℓm

so
X ˆ
ϕ(r , t) = ε−1 dV ′ jℓ (kr ′ )Yℓm
∗
(θ′ , φ′ )ρ r ′ e −iωt
(1)

0 ik hℓ (k r )Yℓm (θ, φ)
ℓm
X ˆ
dV ′ jℓ (kr ′ )Yℓm
∗
(θ′ , φ′ )J r ′ e −iωt
(1)

A(r , t) = µ0 ik hℓ (k r )Yℓm (θ, φ)
ℓm

(1)
⇒Can recover all previous results via Taylor expansion of jℓ , hℓ at small or large kr
−Can reconstruct E , B using standard formulas

B = rotA,
∂A
E = −∇ϕ −
∂t
Dipole emission vs spherical harmonics
Electric dipole:
1
B≈− n × P̈ (t0 ) , E = n × cB
4πε0 c 3 r
Magnetic dipole emission:

µ0 n × M̈ (t0 ) 1
E = , B =− n×E
4πc r c
Polarization:
 electric dipole: Eθ , Bφ
 magnetic dipole: Bθ , Eφ

Assume that a system of particle oscillates with frequency ω ̸= 0, and

P̈, M̈ ∼ ẑ (do not change direction). Demonstrate that fields of electric and
magnetic dipoles correspond to spherical harmonics TM10 and TE10 (assuming
P̈, M̈ point in direction of axis ẑ).
Dipole emission vs spherical harmonics
Electric dipole: Lecture 13:
−Monochromatic solutions of the wave
1
B≈− n × P̈ (t0 ) , equation, TE mode∗ :
4πε0 c 3 r
1 h i mω Yℓm (θ, ϕ)
Eθ = r R(r )
E ≈ n × n × Ȧ = n × B ℓ(ℓ + 1) sin θ
c
Magnetic dipole emission: iω
Eϕ = r R(r ) ∂θ Yℓm (θ, ϕ)
ℓ(ℓ + 1)
∂A µ0 n × M̈ (t0 )
E =− =
∂t 4πc r Er = 0, Br = R(r )Yℓm (θ, ϕ)
B = rotA ≈ n × E 1 ∂
Bθ = − (rEφ )
iωr ∂r
Polarization:
1 ∂
 electric dipole: Eθ , Bφ Bϕ = (rEθ )
iωr ∂r
 magnetic dipole: Bθ , Eφ
1
R(r ) = (αjℓ (kr ) + βnℓ (kr ))
r
1  (1) (1) (2) (2)

= αℓ hℓ (kr ) + αℓ hℓ (kr )
r
∗
For TM mode, just replace fields
1
E → B, B→− E
c2
Dipole emission vs spherical harmonics
In spherical coordinates n ≡ r̂ , ẑ = r̂ cos θ − θ̂ sin θ, so
Electric dipole:

sin θ φ̂ sin θ θ̂
B≈ P̈ (t0 ) , E ≈ n × B = − P̈ (t0 )
4πε0 c 3 r 4πε0 c 3 r
Magnetic dipole emission:

µ0 sin θφ̂ M̈ (t0 ) µ0 sin θθ̂ M̈ (t0 )

E =− , B ≈n×E =
4πc r 4πc r
For monochromatic waves

P̈(t0 )
∼ exp (iωt0 ) = exp (iωt − ikr ) ,
M̈(t0 )

On the other hand, at large distances

1  (1) (1) (2) (2)

 1  (1) (2)

R(r ) = αℓ hℓ (kr ) + αℓ hℓ (kr ) ≈ 2 αℓ e ikr + αℓ e −ikr
r r

(1) (2)
so conclude that should choose αℓ = 0, αℓ ̸= 0
At large distances R(r ) ∼ O 1/r 2 , and since we omit such correction, we don’t have radial

components Er , Br for dipole emission.

Dipole emission vs spherical harmonics
Magnetic dipole emission: Our result is independent of ϕ, dependence
on θ from sin θ, so can see that only har-
µ0 sin θϕ̂ M̈ (t0 ) monics with ℓ = 1, m = 0 contributes:
E =−
4πc r

µ0 sin θθ̂ M̈ (t0 ) Eθ = Bϕ = 0,

B ≈n×E =
 
1
4πc r Br ∼ R(r ) ∼ O ,
r2
Spherical, TE mode:
mω Yℓm (θ, ϕ) however
Eθ = r R(r )
ℓ(ℓ + 1) sin θ

iω 1 (2) −ikr
Eϕ = r R(r ) ∂θ Yℓm (θ, ϕ) r R(r ) ≈ α e
ℓ(ℓ + 1) r 1

Er = 0, Br = R(r )Yℓm (θ, ϕ) so we get nonzero contributions for Eϕ , Bθ , and

1 ∂
Bθ = − (rEφ ) µ0 M̈ (t0 )
iωr ∂r (2)
α1 ∼
1 ∂ 2πc
Bϕ = (rEθ )
iωr ∂r
Dipole emission vs spherical harmonics

Summary:
−Magnetic dipole field corresponds to spherical harmonics TE10 , with amplitude
(2)
α1 ∼ M̈ (t0 )

−The analysis of the electric dipole field is literally the same, just need to permute
E , B and P̈, M̈. Final result: electric dipole corresponds to spherical harmonics
(2)
TM10 , with amplitude α1 ∼ P̈ (t0 ) .
Control question

In all our evaluations we always assumed that dipole moment (electric dipole P̈for
definiteness) is always collinear to axis ẑ.

Assume now that P̈ is NOT collinear to ẑ (though still points in the same direction).
What will be the electromagnetic field in terms of spherical harmonics TMℓm (θ, φ) in
this case?

−Might be useful in some problems ...

Dipole emission vs spherical harmonics
−We considered in this problem that P̈, M̈ ∼ ẑ, that’s why we got only m = 0
−If dipole moment P̈ is NOT collinear to ẑ, then we can do the following:
∗
Introduce ẑ ′ in direction of P̈ and write out the dipole fields using spherical angles
θ′ , φ′ w.r.t. that axis ẑ ′
∗
Recall rotation of spherical harmonics from group theory
ℓ
X
Dm, m′ (α, β, γ) Y ℓ, m′ θ′ , φ′
(ℓ)

Y ℓ, m (θ, φ) = (1)
m′ =−ℓ

∗ (ℓ)
The matrices Dm, m′ (α, β, γ) are called Wigner’s D-matrices:
∗∗
depend on (Euler’s) angles α, β, γ which characterize orientation of ẑ ′ w.r.t. ẑ
∗∗
Tabulated, well-known. See any book on group theory (Sections about Represen-
tations of rotation group) or Wikipedia. Some results for ℓ = 1:
s  
(1) (1 − m)! m sin β sin β
Dm, 0 = P1 (cos β) = − √ , cos β, √
(1 + m)! 2 2

where β is angle between ẑ and ẑ ′

⇒If in right-hand side of (1) had only TM10 , can see that in left-hand side will get
also the harmonics TM1,±1 (ℓ = 1, m = ±1).
Revision of Special relativity

Now we are going to start the review of It is expected that You’ve seen
Special Relativity some basics in undergraduate
courses, so I will focus on the aspects
related to electromagnetism

Please try to recall everything

You remember about Special Rela-
tivity (I need your replies to choose
proper learning pace)
Control question 1

What motivated development of Special Relativity?

What was wrong with “old good” formulas from Classical mechanics?
Control question 2

Are there any experimental evidence which shows that SR is not just a trick to “solve”
some problems of electromagnetism?
Control question 3

Which concepts/main results of Special Relativity differ from those of Newtonian

Mechanics & Galileo’s transformations ?
Galileo’s relativity
Classical mechanics: −Time is absolute
All inertial reference frames are
equvalent t′ = t

−Transformation of coordinates

x′ = x − V t

⇒Transformation of distances between

points:
∆x ′ = ∆x
−Transformation of velocities:

v = v′ + V

−Transformation of forces (recall F = m a):

F = F′
Electromagnetism vs Galileo’s relativity
Maxwell’s equations are incompatible The force between two charges which
with Galileo’s principle: move with the same constant velocity in
lab frame
EM wave equation is covariant
(=doesn’t change its form) in any in- F = q (E + v × B) ,
ertial frame  Results for electric and magnetic fields
1 ∂2E in moving frame:
− ∆E = 0, q 1 − v 2 /c 2 R̂
c 2 ∂t 2 E =
4πε0 1 − v 2 sin2 θ/c 2 3/2 R 2


1 ∂2B
− ∆B = 0,
c 2 ∂t 2
Same velocity c in any frame, in con- q 1 − v 2 /c 2 v × R̂
B=
tradiction with Galileo’s 4πε0 1 − v 2 sin2 θ/c 2 3/2 R 2

v 1 = v 2 + V frame
where θ is angle between R̂ and v , and R =
Note that similar equations appear for r − v t (instant, not retarded separation).
acoustic waves in mechanics of flu-  In the rest frame we get
ids and elastic medium, but they were q1 q2 R̂
found in the medium rest frame, so F = F Coulomb =
4πε0 R 2
there the problem does not exist !!!
⇒different force between electric charges
in different frames, contradicts Galileo’s
Attempts to reconcile EM with Galileo’s relativity

Early attempts (all failed!):

 Modify equations of Maxwell (add more terms to restore covariance)
− Numerous checks, no deviations were detected.
 ether (ficticious invisible substance which we do not feel)
− EM field⇒strain of ether (Maxwell’s strain tensor⇒”strain of ether”)
− ether breaks explicitly the equivalence of different frames
− motion of ether⇒”ether wind”
− Attempts to measure light speed difference in different direction-zero results
Attempts to reconcile EM with Galileo’s relativity
Attempts to measure light speed difference:
 Major challenge: speed c ≈ 3 × 108 m/s ≫ speeds of macroscopic objects
Main idea:
−Use interference of light rays which travel via different paths, in different mediums. If
the motion of the medium could affect the velocity of light, then interference patters
should shift as a function of velocity
Typical experiments of that period:
. Fizeau experiment (1851, repeated in different modifi-
cations).
−Tried to detect if the moving water can drag ether and
in this way affect the velocity of light.
−Galileo’s theory could not explain the results,
vlight ̸= (c/n)±vwater . At the same time, vlight ̸= c/n.

−In SR the effect is interpreted via modification of electric and magnetic polarizabilities
of material in moving frame (D ̸= εε0 E , B ̸= µµ0 H), so vlight ̸= c/n (can use
vlight = c/n only in rest frame of water). In any other frame should use Einstein’s
velocity addition formula.
−Shift of intereference picture ∼ O(v /c)
Michelson-Morley interferometer (1887)
. Tried to detect if the relative motion of the observer
(interferometer) w.r.t. ether affects the speed of light
−Assume that L1 and L2 are the lengths of the arms of
the interferometer, the interferometer moves in hor-
izontal direction with velocity vx with respecct to
ether. Then the corresponding phase shift according
to Galileo’s theory is given by
 
L1 L1 2 L1
∆φ = ω + − =
c − vx c + vx c
 
ωL1 1 1
= + −2
c 1 − vx /c 1 + vx /c
2 2
 
2L1 vx /c
∼ O vx2 /c 2


=
λ 1 − vx2 /c 2

−Expected: Should have periodic oscillations of the phase shift due to rotation of the
Earth (change of orientation of v )
−Experiment: null result, no shift (within errors of experiment). There is no ether
drag, speed of light c = const
Electromagnetism vs Galileo’s relativity
Theoretical developments:
−We know that Cartesian components of E , B in vacuum should satisfy

1 ∂2ψ
− ∆ψ = 0, (1)
c 2 ∂t 2
−Assume that a plane EM wave moves just in 1 dimension in direction x̂.

1 ∂2ψ ∂2ψ
− = 0, (1′ )
c 2 ∂t 2 ∂x 2
Note that as of now we don’t know how E , B transform from one frame to another
−However, if we assume that E , B tranform linearly through each other,

Eα → Kαβ Eβ + Lαβ Bβ , Bα → Mαβ Eβ + Nαβ Nβ ,

where K , L, M, N are some coordinate-independent matrices, then, taking into account

that all components of E , B should satisfy the same equation (1’), can conclude that
in new frame E , B also must satisfy (1’)

Demonstrate explicitly that (1’) is not covariant under Galileo’s transformations, i.e.
if it has this form in one reference frame, it will get additional terms in another frame
Electromagnetism vs Galileo’s relativity

1 ∂2ψ ∂2ψ
2 2
− = 0, (1′ )
c ∂t ∂x 2
−Time is absolute, transformation of time coordinates from one frame to another:

t ′ = t, x′ = x − V t

∂ ∂x ′ ∂ ∂
= =
∂x ∂x ∂x ′ ∂x ′
∂ ∂t ′ ∂ ∂x ′ ∂ ∂ ∂
= ′
+ = ′ −v ′
∂t ∂t ∂t ∂t ∂x ′ ∂t ∂x
2
∂2 ∂2

∂ ∂ ∂ ∂ ∂
= − v = +v 2 ′2 − 2v ′ ′
∂t 2 ∂t ′ ∂x ′ ∂t ′2 ∂x ∂t ∂x

1 ∂2ψ ∂ 2 ψ v 2 ∂ψ 2v ∂ ∂
′2
− + − 2 ′ ′ψ = 0
2
c ∂t ∂x ′2 c 2 ∂x ′2 c ∂t ∂x
−due to additional terms (in red) the covariance is broken, wave equation is NOT
covariant w.r.t. Galileo’s transformation.
Electromagnetism vs Galileo’s relativity
Theoretical developments
(Lorentz, 1899-1904):

Assume that a plane EM wave moves just in 1 dimension in direction x̂. Let’s try to
find (mathematically) a linear transformation (x, t) ↔ (x ′ , t ′ ) which keeps the wave
equation

1 ∂2ψ ∂2ψ
− = 0, (1)
c 2 ∂t 2 ∂x 2
covariant (i.e. has the same form in all reference frames)

Observation: The light moves with the same velocity in all reference frames, so its
front should satisfy
x 2 − c 2 t 2 = (x ′ )2 − c 2 (t ′ )2 = 0 (2)
Let’s introduce a variable x0 = ic t. In this case the equation (2) turns into

x02 + x 2 = (x0′ )2 + (x ′ )2 = 0 (3)

⃗ = (x0 , x) in 4D space. Recall that
-equation for the norm (absolute value) of vector X
only orthogonal transformations keep the value of vector invariant.
Lorentz transformations
x02 + x 2 = (x0′ )2 + (x ′ )2 = 0 (3)
⃗ = (x0 , x) in 4D space. Recall that only
-equation for the absolute value of vector X
orthogonal transformations keep the value of vector invariant.

x ′ = cos α x + sin α ict, ict ′ = − sin α x + cos α ict (4)

From (4): x ′ is real, so sin α must have only imaginary part. Recall that

cos(iη) = cosh η, sin(iη) = i sinh η

so we may rewrite (4, 5) in a form

x ′ = cosh η x − sinh η ct, ct ′ = − sinh η x + cosh η ct (5)

Compare with Galileo’s formulas (v ≪ c):

v
sinh η ≈ , cosh η ≈ 1
c
But cosh2 η − sinh2 η ≡ 1, so need to generalize
v /c 1
sinh η = p , cosh η = p
1 − v 2 /c 2 1 − v 2 /c 2
Lorentz transformations (3+1)

Generalization to 3+1 dimensional case:

x02 + x 2 = (x0′ )2 + (x ′ )2 = const

⃗ = (x0 , x) should be linear,
We expect that transformation of components of vector X
akin to rotations in ordinary 3D space:

X µ → Λµν X ν
where Λµν are some matrices. If we had ordinary 4D Euclidean space, we would conclude
that matrices Λ are orthogonal,
ΛT · Λ = 1
However, now we should be careful because x0 = ict , so some “rotation angles” are
purely imaginary. We’ll need some additional mathematical concept to be able to
formulate carefully conditions on Λµν .