Physics 214 Solution Set 2 Winter 2024

1. [Jackson, problem 11.10]

(a) For the Lorentz boost and rotation matrices K and S show that

(ǫ̂·S)3 = −ǫ̂·S , (1)

(ǫ̂′ ·K)3 = ǫ̂′ ·K , (2)

where ǫ̂ and ǫ̂′ are any real unit 3-vectors.

We are given
     
0 0 0 0 0 0 0 0 0 0 0 0
0 0 0 0 0 0 0 1 0 0 −1 0
S1 = 
0 0
, S2 =  , S3 =  ,
0 −1 0 0 0 0 0 1 0 0
0 0 1 0 0 −1 0 0 0 0 0 0

     
0 1 0 0 0 0 1 0 0 0 0 1
1 0 0 0 0 0 0 0 0 0 0 0
K1 = 
0
, K2 =  , K3 =  .
0 0 0 1 0 0 0 0 0 0 0
0 0 0 0 0 0 0 0 1 0 0 0

To prove eq. (1), we evaluate the matrix ǫ̂·S explicitly,

 
0 0 0 0
0 0 −ǫ3 ǫ2 
ǫ̂·S = 
0
,
ǫ3 0 −ǫ1 
0 −ǫ2 ǫ1 0

and then compute (ǫ̂·S)3 via matrix multiplication. Indeed,

 
0 0 0 0
 0 2
−ǫ2 − ǫ32
ǫ1 ǫ2 ǫ1 ǫ3 
(ǫ̂·S)2 = 
0
,
ǫ1 ǫ2 −ǫ21 − ǫ23 ǫ2 ǫ3 
0 ǫ1 ǫ3 ǫ2 ǫ3 −ǫ21 − ǫ22

and  
0 0 0 0
0 0 −ǫ3 ǫ2 
(ǫ̂·S)3 = (ǫ̂·S)2 ǫ̂·S = −(ǫ21 + ǫ22 + ǫ23 ) 
0
 = −ǫ̂·S ,
ǫ3 0 −ǫ1 
0 −ǫ2 ǫ1 0
after using the fact that ǫ̂ is a real unit 3-vector, which implies that ǫ21 + ǫ22 + ǫ23 = 1.

1
 To prove eq. (2), we evaluate the matrix ǫ̂·K explicitly,
 
0 ǫ′1 ǫ′2 ǫ′3
ǫ′1 0 0 0
ǫ̂ ′ ·K = 
ǫ2
, (3)
′
0 0 0
ǫ′3 0 0 0
and then compute (ǫ̂ ′ ·K)3 via matrix multiplication. Indeed,
 ′2 
ǫ1 + ǫ′22 + ǫ′32 0 0 0
 0 ǫ′2
ǫ′ ′
1 ǫ2 ǫ′1 ǫ′3 
(ǫ̂ ′ ·K)2 = 

1 ,
0 ′ ′
ǫ1 ǫ2 ǫ′22 ǫ′2 ǫ′3 
′ ′
0 ǫ1 ǫ3 ǫ′2 ǫ′3 ǫ′32
and  
0 ǫ′1 ǫ′2 ǫ′3
ǫ ′
0 0 0
(ǫ̂ ′ ·K)3 = (ǫ̂ ′ ·K)2 ǫ̂ ′ ·K = (ǫ′12 + ǫ′22 + ǫ′32 )  1
ǫ′2
 = ǫ̂ ′ ·K ,
0 0 0
ǫ′3 0 0 0
after using the fact that ǫ̂ ′ is a real unit 3-vector.

ALTERNATIVE SOLUTION:
The following alternative solution to part (a) is noteworthy. First, observe that the first row
and column of S1 , S2 and S3 are all zeros. Hence we can simply focus on the remaining 3 × 3
block. That is, we write the Si in block matrix form,
 
0 0Tk
(Si )jk =  , (4)
0j −ǫijk
where 0T is a row vector of three zeros, 0 is a column vector of three zeros, and


+1 , if (ijk) is an even permutation of (123) ,
ǫijk = −1 , if (ijk) is an odd permutation of (123) ,


0, otherwise ,
is the three-dimensional Levi-Civita tensor. After excluding the first row and column, jk labels
the three remaining rows and columns of the Si .
Thus, we can compute (ǫ̂·S)3 by pretending that the first row and column do not exist. More
explicitly,1
(ǫ̂·S)3jk = (ǫ̂·S)jℓ (ǫ̂·S)ℓm (ǫ̂·S)mk = ǫi (Si )jℓ ǫp (Sp )ℓm ǫq (Sq )mk
= −ǫi ǫp ǫq ǫijℓ ǫpℓm ǫqmk = ǫi ǫp ǫq ǫijℓ ǫpmℓ ǫqmk
= ǫi ǫp ǫq (δip δjm − δim δjp )ǫqmk = ǫq ǫqjk − ǫm ǫj ǫq ǫqmk , (5)
1
In eq. (5), we employ the Einstein summation convention. In this derivation, we make use of the antisymmetry
properties of the Levi-Civita tensor and employ the identity ǫijℓ ǫpmℓ = δip δjm − δim δjp .

2
after noting that ǫi ǫi = ǫ̂·ǫ̂ = 1 since ǫ̂ is an arbitrary real unit vector. We now observe
that ǫm ǫj ǫq ǫqmk = 0 since ǫm ǫq is symmetric under the interchange of m and q whereas ǫqmk is
antisymmetric under the same interchange of indices. Thus, eq. (5) yields

(ǫ̂·S)3jk = ǫq ǫqjk = −ǫq (Sq )jk = −(ǫ̂·S)jk ,

which establishes eq. (1).

To establish eq. (2), we rewrite ǫ̂ ′ ·K given by eq. (3) in block matrix form [analogous to the
form of the Si in eq. (4)],  
′
0 ǫk
(ǫ̂ ′ ·K)jk =  , (6)
′
ǫj 0jk
where 0jk stands for the matrix elements of the 3 × 3 zero matrix. In particular, j labels the row
and k labels the column. Then,
      
′ ′ ′ ′ ′ ′
0 ǫℓ 0 ǫi 0 ǫk ǫ̂ ·ǫ̂ 0i 0 ǫk
(ǫ̂ ′ ·K)3jk =    =  
′ ′ ′ T ′ ′ ′
ǫj 0jℓ ǫℓ 0ℓi ǫi 0ik 0j ǫj ǫi ǫi 0ik
      
1 0i 0 ǫ′k 0 ǫ′k 0 ǫ′k
=  = =  = (ǫ̂ ′ ·K)jk ,
0Tj ǫ′j ǫ′i ǫ′i 0ik ǫ′j ǫ̂ ′ ·ǫ̂ ′ 0jk ǫ′j 0jk

after using the fact that ǫ̂′ is is real unit vector. Once again, eq. (2) is established.

(b) Use the result of part (a) to show that:

exp −ζ β̂·K = I − β̂·K sinh ζ + (β̂·K)2 [cosh ζ − 1] ,

where I is the 4 × 4 identity matrix.

We employ the series expansion for the exponential (which defines the matrix exponential),
∞
X

(−ζ)n
exp −ζ β̂·K = (β̂·K)n . (7)
n=0
n!

In part (a), we established the following result: (β̂·K)3 = β̂·K. Hence, it follows that

(β̂·K)2n = (β̂·K)2 , (β̂·K)2n+1 = β̂·K , for n = 1, 2, 3, . . . .

Thus, we can rewrite the series given in eq. (7) as

X ζn X ζn
exp −ζ β̂·K = I − β̂·K + (β̂·K)2 , (8)
n! n even n!
n odd
n≥1 n≥2

3
after using the fact that (β̂·K)0 = I is the 4 × 4 identity matrix. Using,
∞
X X∞
ζ 2n+1 ζ 2n
= sinh ζ , = cosh ζ ,
n=0
(2n + 1)! n=0
(2n)!

and noting that the last summation in eq. (8) starts at n = 2, we end up with


exp −ζ β̂·K = I − β̂·K sinh ζ + (β̂·K)2 [cosh ζ − 1] , (9)
which is the desired result.

REMARKS:
To understand the significance of eq. (9), let us write it explicitly in matrix form. It is
convenient to use the block matrix form of eq. (6), where j labels the row and k labels the
column,
     
T T
1 0k 0 β̂ k 1 0k
I = , (β̂·K)jk =  , (β̂·K)2jk =  .
0j δjk β̂ j 0jk 0j β̂ j β̂ k
(10)
Then, eq. (9) yields
 

cosh ζ − β̂ k sinh ζ
exp −ζ β̂·K =  .
−β̂ j sinh ζ δjk + β̂ j β̂ k (cosh ζ − 1)

In class, we identified ζ = tanh−1 β as the rapidity, which satisfies

1
γ=p = cosh ζ , βγ = sinh ζ .
1 − β2
~ = β β̂ = (β1 , β2 , β3 ), it follows that
Hence, after writing β
 

γ −γβk 
exp −ζ β̂·K = 

, (11)
βj βk 
−γβj δjk + (γ − 1) 2
β
which we recognize as the boost matrix defined in eq. (11.98) of Jackson.

AN ALTERNATIVE METHOD FOR COMPUTING exp −ζ β̂·K :
Using eq. (3),
 
0 −ζβ1 /β −ζβ2 /β −ζβ3 /β
−ζβ1 /β 0 0 0 
M ≡ −ζ β̂·K = 
−ζβ2 /β
, (12)
0 0 0 
−ζβ3 /β 0 0 0

4
In order to compute f (M) = exp M, we shall employ the following formula of matrix algebra.
Denote the distinct eigenvalues of the n × n matrix M by λi and define the following polynomial,2
p(x) = (x − λ1 )(x − λ2 ) · · · (x − λm ) . (13)
Then, M is diagonalizable if and only if p(M) = 0n , where 0n is the n × n zero matrix. In this
case, any function of M is given by 3
 
m
X m
Y M − λj I n 
f (M) = 
f (λi )  , (14)
i=1 j=1
λi − λj 
j6=i

where I n is the n × n identify matrix and m is the number of distinct eigenvalues.4

We first compute the eigenvalues of M, which are roots of the characteristic polynomial,
   
−ζβ1 /β 0 0 −ζβ1 /β −λ 0
ζβ1 ζβ2
det(M − λI 4 ) = λ4 + det −ζβ2 /β −λ 0 − det −ζβ2 /β 0 0
β β
−ζβ3 /β 0 −λ −ζβ3 /β 0 −λ
 
−ζβ1 /β −λ 0
ζβ3
− det −ζβ2 /β 0 −λ = λ2 (λ2 − ζ 2 ) , (15)
β
−ζβ3 /β 0 0
after using β 2 = β12 + β22 + β32 . Thus, the three distinct eigenvalues of M are λi = 0, ζ, −ζ.
We can check that M is diagonalizable by evaluating:
p(M) = M(M − ζI 4 )(M + ζI 4 )
  
0 −ζβ1 /β −ζβ2 /β −ζβ3 /β −ζ −ζβ1 /β −ζβ2 /β −ζβ3 /β
−ζβ1 /β 0 0  
0  −ζβ1 /β −ζ 0 0 
=−ζβ2 /β

0 0 0  −ζβ2 /β 0 −ζ 0 
−ζβ3 /β 0 0 0 −ζβ3 /β 0 0 −ζ
 
ζ −ζβ1 /β −ζβ2 /β ζβ3 /β
−ζβ1 /β ζ 0 0 
×
−ζβ2 /β

0 ζ 0 
−ζβ3 /β 0 0 ζ
  
1 β1 /β β2 /β β3 /β 1 −β1 /β −β2 /β β3 /β
β1 /β β12 /β 2 β1 β2 /β 2 2 
β1 β3 /β  −β1 /β 1 0 0
= ζ3 
β2 /β
 = 04 .
β1 β2 /β 2 β22 /β 2 β2 β3 /β 2 −β2 /β 0 1 0
β3 /β β1 β3 /β 2 β2 β3 /β 2 β32 /β 2 −β3 /β 0 0 1
(16)
2
Note that m ≤ n since it is possible that the characteristic polynomial possesses one or more multiple roots.
3
For example, see eqs. (7.36) and (7.3.11) of Carl D. Meyer, Matrix Analysis and Applied Linear Algebra (SIAM,
Philadelphia, PA, 2000) or Chapter V, Section 2.2 of F.R. Gantmacher, Theory of Matrices–Volume I (Chelsea
Publishing Company, New York, NY, 1959).
4
If the n × n matrix M is not diagonalizable then p(M ) 6= 0n , in which case the formula for f (M ) is more
complicated than the one given in eq. (14).

5
 We now apply eq. (14) to f (M) = exp M. It then follows that
1 ζ 1 −ζ 1
exp M = − (M − ζI 4 )(M + ζI 4 ) + e M(M + ζI 4 ) + e M(M − ζI 4 )
ζ2 2ζ 2 2ζ 2
   
sinh ζ cosh ζ − 1
= I4 + M+ 2
M2 . (17)
ζ ζ

Inserting M = −ζ β̂·K, we recover the result of eq. (9).

2. [Jackson, problem 11.13] An infinitely long straight wire of negligible cross-sectional area is
at rest and has a uniform linear charge density q0 in the inertial frame K ′ . The frame K ′ (and
the wire) move with velocity ~
v parallel to the direction of the wire with respect to the laboratory
frame K.
(a) Write down the electric and magnetic fields in cylindrical coordinates in the rest frame of
the wire. Using the Lorentz transformation properties of the fields, find the components of the
electric and magnetic fields in the laboratory.

In the rest frame of the wire (i.e. frame K ′ ), choose the z-axis to point along the wire. Then, to
compute the electric field, we draw a cylinder of length L and radius r ′ , whose symmetry axis
coincides with the z-axis. Applying Gauss’ law in gaussian units,
I
~ ′ · n̂ da = 4πQ ,
E (18)
S

where Q is the total charge enclosed inside the cylinder. In cylindrical coordinates (r ′ , φ′ , z ′ ),5 the
symmetry of the problem implies that E ~ ′ (r ′ , φ′ , z ′ ) = E ′ (r ′ )r̂ ′ , where E ′ (r ′ ) depends only on the
radial distance from the symmetry axis. Choosing the surface S to be the surface of the cylinder,
we have n̂ = r̂ ′ , and so eq. (18) reduces to

2πr ′ LE ′ (r ′ ) = 4πQ .

Defining the linear charge density (i.e. charge per unit length) by q0 = Q/L, we conclude that6

~ ′ (r ′ ) = 2q0 r̂ ′ .
E (19)
r′
~ ′ = 0.
Since there are no moving charges in the rest frame of the wire, it follows that B
The transformation laws for the electric and magnetic field between reference frames K and
5
We denote the radial coordinate of cylindrical coordinates in frame K ′ to be r′ rather than the more traditional
′
ρ , since we reserve the letter ρ for charge density.
6
The direction of the unit vectors r̂, φ̂ and ẑ are the same in frames K and K ′ , so no extra primed-superscript
is required on these quantities.

6
K ′ are given by7
h i γ2 ~ ~ ~ ′
~ ~ ′ ~ ~
E =γ E −β×B − ′
β(β· E ) ,
γ+1
h i 2
~ =γ B
B ~′+β ~ ′ − γ β(
~×E ~ β·
~ B~ ′) .
γ+1

For this problem, β~ = β ẑ. Using the results of part (a), and noting that r = r ′ (since the radial
direction is perpendicular to the direction of the velocity of frame K ′ with respect to K), it follows
that
~ = 2γq0 r̂ ,
E B~ = 2γβq0 φ̂ , (20)
r r
where we have used ẑ· r̂ = 0 and ẑ × r̂ = φ̂.

(b) What are the charge and current densities associated with the wire in its rest frame? In
the laboratory?

In reference frame K ′ there are no moving charges, so that J~ ′ = 0. The corresponding charge
density is
q0
ρ′ (r ′ ) = δ(r ′ ) . (21)
2πr ′
To check this, let us integrate over a cylinder of length L and arbitrary nonzero radius, whose
symmetry axis coincides with the z-axis. Then,
Z Z Z
ρ (r ) dV = ρ (r ) r dr dφ dz = q0 dr ′ δ(r ′ )dz ′ = q0 L = Q .
′ ′ ′ ′ ′ ′ ′

Since J µ = (cρ ; J~) is a four-vector, the relevant transformation law between frames K and K ′
are:
~ J~ ′ ) ,
cρ = γ(cρ′ + β· (22)
γ − 1 ~ ~′ ~ ~ ′.
J~ = J~ ′ + (β· J )β + γ βcρ (23)
β2

Plugging in J~ ′ = 0 and the result of eq. (21), and noting that β

~ = β ẑ and r ′ = r, it follows that8

γq0 γβc q0
ρ(r) = δ(r) , J~ = ẑ δ(r) = ρ(r)v ẑ = ρ(r)~
v, (24)
2πr 2πr
after using v ≡ β c.
7
Eq. (11.149) of Jackson provides the equations to transform the fields from reference frame K to reference
frame K ′ . To transform the fields from K ′ to K, simply change the sign of β.~
8
We can interpret q ≡ γq0 as the linear charge density as observed in reference frame K. This is not unexpected
due to the phenomenon of length contraction.

7
 (c) From the laboratory charge and current densities, calculate directly the electric and mag-
netic fields in the laboratory. Compare with the results of part (a).

This is an electrostatics and magnetostatics problem, so we can use Gauss’ law to compute E ~
~ ~
and Ampère’s law to compute B. The computation of E is identical to the one given in part (a)
with q0 replaced by γq0 . Hence, it immediately follows from eq. (19) that

~ 2γq0
E(r) = r̂ ,
r
in agreement with eq. (20). Ampère’s law in gaussian units is
I
~ ·d~ 4πI
B ℓ= ,
C c

where I is the current enclosed in the loop C. With J~ given by eq. (24),
Z Z
I= ~
J · n̂ da = ρ(r)v r dr dφ = γq0 v ,
A

after noting that n̂ = ẑ points along the direction of the current flow and da = rdrdφ is the
infinitesimal area element perpendicular to the current flow. Using the symmetry of the problem,
B~ = B(r)φ̂. Thus, evaluating Ampère’s law with a contour C given by a circle centered at r = 0
that lies in a plane perpendicular to the current flow, d~ ~ and
ℓ = rdφφ
4πI 4πγq0 v
2πrB(r) = = ,
c c
which yields
~ 2γβq0 v
B(r) = φ̂ ,
r
after using v = βc, in agreement with eq. (20).

3. [Jackson, problem 11.15] In a certain reference frame, a static uniform electric field E0 is
parallel to the x-axis, and a static uniform magnetic field B0 = 2E0 lies in the x–y plane, making
an angle θ with respect to the x-axis. Determine the relative velocity of a reference frame in
which the electric and magnetic fields are parallel. What are the fields in this frame for θ ≪ 1
and θ → 21 π?

In frame K, we have
~ = E0 x̂ ,
E ~ = Bx x̂ + By ŷ ,
B (25)
with
~ ·B
E ~ = |E|
~ |B|
~ cos θ = E0 B0 cos θ = 2E02 cos θ , (26)
~ = E0 and |B|
after writing |E| ~ = B0 = 2E0 . It follows that

Bx = 2E0 cos θ , By = 2E0 sin θ . (27)

8
 The electric and magnetic fields are parallel in a reference frame K ′ which is moving at a
velocity ~
v ≡ cβ~ with respect to reference frame K. That is, the fields in K ′ satisfy,

~′×B
E ~ ′ = 0. (28)

The electric and magnetic fields in frame K ′ are related to the corresponding fields in frame K
by eq. (11.149) of Jackson,
2
~ ′ = γ(E
E ~ +β ~ − γ β(
~ × B) ~ β·
~ E)
~ , (29)
γ+1
2
~ ′ = γ(B
B ~ −β ~ − γ β(
~ × E) ~ β·
~ B)
~ . (30)
γ+1
These relations can be rewritten in the following form,
~ k′ = E
E ~k , B~ k′ = B
~k , (31)
   
~ ′ ~ ~ ~
E⊥ = γ E⊥ + β × B⊥ , ~ ′ ~ ~ ~
B⊥ = γ B⊥ − β × E⊥ . (32)

~ and fields with a ⊥ subscript are

In eqs. (31) and (32), fields with a k subscript are parallel to β
~ For example,
perpendicular to β.
~×E
β ~ E
~ k = 0 and β· ~⊥ = 0 ,

which implies that

~ ~ ~ ~ ~ ~ ~ ~ ~
~ k = (β· E)β
E ~ − (β· E)β = β × (E × β) .
~⊥ = E
and E
β2 β2 β2

The form of eqs. (31) and (32) suggests that the relative velocity ~
v should point in the z-
direction. That is,
β~ = β ẑ ,
in which case E ~ k = Ez ẑ and B~ k = Bz ẑ. Since Ez = Bz = 0, it follows from eq. (31) that
Ez′ = Bz′ = 0. Using eq. (32), the transverse fields are given by

Ex′ = γ(Ex − βBy ) = γE0 (1 − 2β sin θ) , Ey′ = γ(Ey + βBx ) = 2βγE0 cos θ , (33)
Bx′ = γ(Bx + βEy ) = 2γE0 cos θ , By′ = γ(Ey + βBx ) = γE0 (2 sin θ − β) , (34)

after using eqs. (25)–(27). Moreover, eq. (28) implies that Ex′ By′ − Ey′ Bx′ = (E ~′ ×B
~ ′ )z = 0 .
Inserting the results for the primed fields in this last equation, it then follows that

γ 2 E02 (1 − 2β sin θ)(2 sin θ − β) − 4βγ 2 E02 cos2 θ = 0 .

Multiplying out the factors above and writing cos2 θ = 1 − sin2 θ, the above equation simplifies to

2β 2 sin θ − 5β + 2 sin θ = 0 . (35)

9
This is a quadratic equation in β which is easily solved. The larger of the two roots is greater
than 1, which we reject since 0 ≤ β ≤ 1 (i.e., 0 ≤ v ≤ c). The smaller of the two roots is
non-negative and less than 1. Thus, we conclude that
p
v 5 − 25 − 16 sin2 θ
β= = . (36)
c 4 sin θ
The two limiting cases are easily analyzed. In the case of θ ≪ 1, we can work to first order
in θ. From eq. (36) we find that β ≃ 25 θ. Since θ ≪ 1 it follows that β ≪ 1, in which case

γ = (1 − β 2 )−1/2 ≃ 1 + O(β 2 ) .

Since we are working to first order in θ, we also must work to first order in β. In particular we
can neglect terms such as βθ. Hence, in this limiting case, eqs. (33) and (34) yield
~ ′ = 1B
E ~ ′ = E0 (x̂ + 2β ŷ) , for β ≃ 52 θ ≪ 1 , (37)
2

where we have neglected terms that are second √ order (or higher) in β. Finally, in the limit of
θ → 21 π, eq. (36) yields β = 21 . Then γ = 2/ 3, and eqs. (33) and (34) yield
√
E~ ′ = 0, ~ ′ = 3E0 ŷ ,
B for θ = 12 π . (38)

REMARK 1:
Recall that in class, we showed that the quantity F µν Feµν = 12 ǫµναβ F µν F αβ = −4E ~ ·B~ is a
Lorentz invariant. This means that if E ~ and B~ are perpendicular in one frame, then they must
1
be perpendicular in all frames. Thus, if θ = 2 π in frame K and θ = 0 in frame K ′ , then it must
be true that either the electric field or the magnetic field (or both) vanish in frame K ′ , since the
only vector that is both perpendicular and parallel to a given fixed nonzero vector is the zero
vector. This is indeed the case here, as can be seen in eq. (38).

REMARK 2:
It is easy to show that eq. (36) implies that 0 ≤ β ≤ 21 . If we multiply the numerator and
p
denominator of eq. (36) by 5 + 25 − 16 sin2 θ, we obtain,
4 sin θ
β= p .
5+ 25 − 16 sin2 θ
Since the polar angle lies in the range 0 ≤ θ ≤ π or equivalently 0 ≤ sin θ ≤ 1, it follows
immediately that β ≥ 0 (where β = 0 corresponds to sin θ = 0 as expected). Finally, it is easy to
verify that
4 sin θ 1
p ≤ . (39)
2
5 + 25 − 16 sin θ 2
Since the denominator on the left hand side above is positive, we can rewrite eq. (39) as
 p 
4 sin θ ≤ 21 5 + 25 − 16 sin2 θ . (40)

10
This inequality is manifestly true for sin θ = 0. For sin θ > 0, eq. (40) can be rearranged into the
following form p
8 sin θ − 5 ≤ 25 − 16 sin2 θ . (41)
Squaring both sides and simplifying the resulting expression then yields sin θ (sin θ − 1) ≤ 0.
Dividing both sides of the equation by sin θ (which is assumed positive) yields 0 ≤ sin θ ≤ 1,
which is valid for all polar angles θ. Hence, eq. (39) is established. The inequality becomes an
equality if sin θ = 1, in which case β = 21 .

REMARK 3: Non-uniqueness of the solution

In our analysis above, we found one solution to the problem. However, it is easy to see that
there are an infinite number of solutions. That is, there are an infinite number of Lorentz boost
matrices such that
~ µ α Λ(β)
F ′µν = Λ(β) ~ ν β F αβ , (42)
where F αβ is the electromagnetic field strength tensor made up of the E ~ and B~ fields given in
eqs. (25) and (27), F ′µν
is the electromagnetic field strength tensor made up of the E ~ ′ and B~′
fields such that E~ ×B
′ ~ = 0, and Λ(β)
′ ~ is the Lorentz boost matrix in the direction of β ~ given
in eq. (11). We have already found one such boost matrix, namely Λ(β ẑ), where β is given by
eq. (36). This boost matrix produces the E ~ ′ and B
~ ′ fields given in eqs. (33) and (34). Since E~′
and B~ ′ are parallel in the primed reference frame, we can write

~ ′ = E ′ n̂ ,
E ~ ′ = B ′ n̂ ,
B (43)

~ ′ and B
where n̂ is the common direction of E ~ ′ . Using eq. (33), one obtains an explicit form for
n̂ that is given by,

(1 − 2β sin θ)x̂ + 2β cos θ ŷ ~ + β cos θ B

(sin θ − 2β)E ~
n̂ = p = p , (44)
1 − 4β sin θ + 4β 2 E0 sin θ 1 − 4β sin θ + 4β 2

where β is given by eq. (36). We used eqs. (25)–(27) to obtain the final expression above.
If one applies the following Lorentz transformation to reference frame K,

Λ = Λ(β ′n̂)Λ(β ẑ) , (45)

~ ′ and B
then in the resulting reference frame K ′′ the E ~ ′ fields are also parallel, for any choice of β ′ .
This result follows from eq. (31), which states that the components of the electric and magnetic
field that are parallel to the boost direction are unaffected by the Lorentz transformation. Having
found the reference frame K ′ after applying Λ(β ẑ) where E ~ ′ and B
~ ′ are parallel and point in the
n̂ direction, one can perform an arbitrary boost in the direction parallel to n̂ without modifying
E~ ′ and B~ ′ further.
One can evaluate the right hand side of eq. (45) explicitly. Here, I will make use of Pawel
Klimas, Lecture Notes on Classical Electrodynamics, which has been posted to the Physics 214
webpage. Using eqs. (1.73) and (1.78) of Klimas’ notes,

~ ′′ ) ,
Λ(β ′n̂)Λ(β ẑ) = OΛ(β (46)

11
where O is a Lorentz transformation corresponding to a pure rotation9 and
 ′   
~ ′′ 1 β n̂ γββ ′
β = + 1+ n̂· ẑ β ẑ , (47)
1 + ββ ′ n̂· ẑ γ γ+1
where γ ≡ (1 − β 2 )−1/2 . In light of eq. (44), it follows that n̂· ẑ = 0, and eq. (47) simplifies to10
~ ′′ = β ′ (1 − β 2 )1/2 n̂ + β ẑ .
β (48)
Note that the parallel electric and magnetic field remain parallel if one transforms the reference
frame by a pure rotation. Thus, we can neglect the pure rotation O in eq. (45) to conclude that
starting from reference frame K, the application of the boost Λ(β ~ ′′ ) to produce reference frame
~ ′′ and B
K ′′ yields E ~ ′′ fields that are parallel.
To summarize, the complete answer to the problem posed by Jackson (although
probably not
′ 2 1/2
what Jackson meant to ask) is that any boost of the form Λ β (1 − β ) n̂ + β ẑ , where β and
n̂ are fixed by eqs. (36) and (44), respectively, will yield a reference frame K ′′ such that the E ~ ′′
and B ~ fields are parallel, for any choice of the parameter β ′ , where 0 ≤ β ′ ≤ 1.
′′

ALTERNATIVE SOLUTION:

Plugging the results for the electric and magnetic fields in reference frame K ′ given by eqs. (29)
and (30) into eq. (28), one can work out the following expressions. First,
 
~ +β
(E ~ × B)
~ × (B ~ −β~ × E)
~ =E ~ ×B ~ −β~ E 2 + B 2 − β·(
~ E ~ × B)
~ + E( ~ E)
~ β· ~ + B( ~ B)
~ β· ~ , (49)

~ and B ≡ |B|.
where E ≡ |E| ~ Second,

(E ~ × B)
~ +β ~ = −β
~ ×β ~×E ~ β·
~ − β( ~ B) ~,
~ + β 2B (50)
~ × (B
β ~ × E)
~ −β ~×B
~ =β ~ β·
~ − β( ~ E)
~ + β 2E
~. (51)
Hence, we obtain,
 
~ ′ ~ ′ 2~ ~ ~ 2 2 2 ~ ~ ~
 γ3  ~ ~ 2 ~ ~ 2

E × B = γ E × B − β γ [E + B − β·(E × B) − (β· E) + (β· B)
γ+1
   

2 ~ ~ ~
 γβ 2 γ3
~ ~ ~
+γ E(β· E) + B(β· B) 1 − − ~ ~ ~ ~ ~ ~ ~ ~
(β· E)β × B − (β· B)β × E .
γ+1 γ+1
(52)
We can simplify the above expression by using γ 2 = (1 − β 2 )−1 , which yields β 2 = (γ 2 − 1)/γ 2 .
Hence,
γβ 2 γ−1 1
1− =1− = . (53)
γ+1 γ γ
9
The rotation O is called the Wigner rotation. As explained below eq. (48), the parallel electric and magnetic
fields remain parallel under a pure rotation, and thus we will not require an explicit expression for the Wigner
rotation in this problem.
10 ~ ′′ |, then β ′′ 2 = β ′ 2 (1 − β 2 ) + β 2 . One can then check that 0 ≤ β 2 , β ′ 2 ≤ 1 implies that
If we define β ′′ ≡ |β
0 ≤ β ′′ 2 ≤ 1, as required by special relativity.

12
We then end up with,
3
 
  2 γ
~ ×B
E ~ = γ E×B+γ
′ ~ ′
E(2~
~ β·
~ E)+
~ B(
~ β·
~ B)
~ −γ hβ−
~ ~ E)
(β· ~ β×
~ B−(
~ β·
~ B)
~ β×
~ E
~ , (54)
γ+1

where
 
  γ  2  2
2 2 ~ · B)
h ≡ E + B − k E B − (E ~ 2 2 2
− 2 ~ · B)
k 1 E + k 2 (E ~ ~ · B)
+ k 1 (E ~ + k2 B 2
. (55)
γ+1

The only way to satisfy E~′ ×B~ ′ = 0 is if the right hand side of eq. (54) is proportional to
~ × B.
E ~ 11
~ ×B
One way of ensuring that the right hand side of eq. (54) is proportional to E ~ is to take
~ to be parallel to E
β ~ × B.
~ That is, there exists a nonzero constant k such that

~ = kE
β ~ ×B
~. (56)

~ E
Note that eq. (56) implies that β· ~ = β·
~ B~ = 0. Hence, eq. (54) simplifies to,
 2

~ ×B
′ ~ =γ β
′ 2 ~ 1 + β 2 2
E −E −B . (57)
k

It then follows that

~′×B
~′=0 1 + β2
E =⇒ = E2 + B2 . (58)
k
Using eqs. (25) and (27), E 2 + B 2 = 5E02 and
2
~ = kE
β ~ = 2E0 k sin θ β
~ ×B ~. (59)
β
Thus, one can identify,
β
k= . (60)
2E02 sin θ
Plugging this result into eq. (58) yields,

2 sin θ(1 + β 2 ) = 5β , (61)

which reproduces the result previously obtained in eq. (35).

As a check of our calculation, let us verify explicitly that E~ ′ is parallel to B
~ ′ . Inserting
eq. (56) into eqs. (29) and (30) yields,


E~ ′ = γE ~ 1 − kB 2 + γk B(
~ E~ · B)
~ , (62)


B~ ′ = γB ~ 1 − kE 2 + γk E(
~ E~ · B)
~ . (63)
11
P
Recall that if {~
vi } is a set of linearly independent vectors, then the only solution to i ci ~
vi = 0 is ci = 0 for
all i.

13
In light of eqs. (25)–(27) and eq. (60),
 
~ ′ = γE0 (1 − 2β sin θ)x̂ + 2β cos θ ŷ ,
E (64)
 
~ ′ = γE0 2 cos θ x̂ + (2 sin θ − β)ŷ .
B (65)

We can now check that

~′×B
E ~ ′ = [2 sin θ(1 + β 2 ) − 5β]ẑ = 0 , (66)
after employing eq. (61), which completes the check of the calculation.
The two limiting cases are now easily analyzed. In the case of θ ≪ 1, we can work to first
order in θ. As noted below eq. (36), β ≃ 25 θ and γ = (1 − β 2 )−1/2 ≃ 1 + O(β 2 ). Since we are
working to first order in θ, we also must work to first order in β. In particular we can neglect
terms such as βθ. Hence, in this limiting case, eqs. (64) and (65) yield

~ ′ = 1B
E ~ ′ = E0 (x̂ + 2β ŷ) , for β ≃ 52 θ ≪ 1 , (67)
2

where we have neglected terms that are second √ order (or higher) in β. Finally, in the limit of
θ → 21 π, eq. (36) yields β = 21 . Then γ = 2/ 3, and eqs. (64) and (65) yield
√
~ ′ = 0,
E ~′=
B 3E0 ŷ , for θ = 12 π . (68)

Thus, we have reproduced the results of eqs. (37) and (38).

REMARK 4:
Another strategy to find all possible boosts that result in parallel electric and magnetic fields
is to start with eqs. (29) and (30) and impose the condition E~′×B ~ ′ = 0 to determine the most
general form for the boost. We again denote the boost parameter by β. ~
~ B
Since E, ~ and E ~ ×B ~ are three linearly independent vectors, β ~ can be written in the
following form,
~ = k1 E
β ~ + k2 B
~ + kE
~ ×B ~, (69)
where the constants k1 , k2 and k are to be determined. It then follows that,
~ E
β· ~ = E 2 k 1 + (E
~ · B)k
~ 2, ~ B
β· ~ = (E
~ · B)k
~ 1 + B 2 k2 ,
   
~×E
β ~ = −k2 E ~ ×B ~ − k (E~ · B)
~ E~ − E 2B~ , ~×B
β ~ = k1 E
~ ×B ~ + k (E~ · B)
~ B~ − B2E
~ .
 
~ E
β·( ~ × B)
~ = k|E ~ × B| ~ 2 = k E 2 B 2 − (E
~ · B)
~ 2 ,

~ where
and β ≡ |β|,
 
~ · B)
β = k12 E 2 + 2k1 k2 (E ~ + k22 B 2 + k 2 E 2 B 2 − (E
~ · B)
~ 2 . (70)

This last equation is needed to obtain an expression for γ ≡ (1 − β 2 )−1/2 . Plugging the above
results into eq. (54) yields,

~′×B
E ~ ′ = c1 E
~ + c2 B
~ + c3 E
~ ×B
~, (71)

14
in reference frame K ′′ , where
  3  
c1 = −γ 2 k1 h + γ k1 E 2 + k2 (E ~ + γ kk1 E 2 B 2 − (E
~ · B) ~ · B)
~ 2 , (72)
γ+1
  3  
c2 = −γ 2 k2 h + γ k1 (E ~ + k2 B 2 + γ kk2 E 2 B 2 − (E
~ · B) ~ · B)
~ 2 , (73)
γ+1
3  
γ ~ · B)
~ + k22 B 2 .
c3 = γ 2 (1 − kh) − k12 E 2 + 2k1 k2 (E (74)
γ+1

and h is given by eq. (55), which we rewrite below for the reader’s convenience,
 
 2 2  γ  2  
2 2
2 2 ~
h ≡ E + B − k E B − (E · B) − ~ 2 2 ~ ~ ~ ~
k1 E + k2 (E · B) + k1 (E · B) + k2 B . (75)
γ+1

To find solutions {k1 , k2 , k} to the equation E ~′×B ~ ′ = 0, we set c1 = c2 = c3 = 0. This

yields three nonlinear equations for the three unknowns, k1 , k2 and k. The one solution obtained
previously with β ~ = β0 ẑ corresponds to k1 = k2 = 0 and kh = 1, where k is given by eq. (60).
Here, we write β0 to distinguish this special case from the general case under consideration. In
this special case, c1 = c2 = 0 automatically and c3 = 0 yields kh = 1 which implies that
 
~ · B)
k(E 2 + B 2 ) − k 2 E 2 B 2 − (E ~ 2 = 1. (76)

~ · B)
Using eqs. (25)–(27), E 2 + B 2 = 5E02 and E 2 B 2 − (E ~ 2 = 4E04 sin2 θ. Hence,

4E04 k 2 sin2 θ − 5E02 k + 1 = 0 . (77)

Using eq. (60) to eliminate k (replacing β with β0 as noted above), eq. (56) is equivalent to

5β0
β02 − +1 = 0, (78)
2 sin θ
~ = β0 ẑ, as expected.
which yields eq. (61) for the special case of β
More generally, one can verify that eq. (48) provides a family of solutions to eqs. (72)–(74)
with c1 = c2 = c3 = 0. In light of eqs. (44) and (48), we can identify,

β ′(sin θ − 2β0 )
k1 = p , (79)
γ0 E0 sin θ 1 − 4β0 sin θ + 4β02
β ′ β0 cos θ
k2 = p , (80)
γ0 E0 sin θ 1 − 4β0 sin θ + 4β02
β0
k= 2
, (81)
2E0 sin θ

where β0 is given by eq. (78), γ0 ≡ (1−β02 )−1/2 and β ′ is an arbitrary number such that 0 ≤ β ′ ≤ 1.
I have checked using Mathematica that after plugging in eqs. (79)–(81) into eqs. (72)–(75) along

15
with E 2 = E02 , B 2 = 4E02 , and E~ ·B~ = 2E02 cos θ, the end result is,
  
2γCE0 2(β02 + 1) sin θ − 5β0 1 − 21 β0 sin θ − C 2 cos2 θ
c1 = − γ+ , (82)
(1 + γ) sin θ 1 − β02 − C 2 (1 − 4β0 sin θ + 4β02 )
  
γCE0 cos θ 2(β02 + 1) sin θ − 5β0 1 − C 2 (1 − 2β0 sin θ)
c2 = γ+ , (83)
(1 + γ) sin θ 1 − β02 − C 2 (1 − 4β0 sin θ + 4β02
  
γ 2 2(β02 + 1) sin θ − 5β0 1 + γ − γC 2 (1 − 2β0 sin θ)
c3 = , (84)
2(1 + γ) sin θ
where
β′
C≡ p . (85)
γ0 1 − 4β0 sin θ + 4β02
Indeed, if β0 satisfies eq. (78) then we find that c1 = c2 = c3 = 0. Thus, I have verified that a
boost to the frame with boost parameter given by eq. (48) yields E ~′×B ~ ′ = 0. I believe that
{k1 , k2 , k} given by eqs. (79)–(81) provides all possible solutions, but I do not have a proof of this
statement.

4. [Jackson, problem 11.18] The electric and magnetic fields of a particle of charge q moving
in a straight line with speed v = βc, given by eq. (11.52) of Jackson, become more and more
concentrated as β → 1, as indicated in Fig. 11.9 on p. 561 of Jackson. Choose axes so that
the charge moves along the z axis in the positive direction, passing the origin at t = 0. Let the
spatial coordinates of the observation point be (x, y, z) and define the transverse vector ~
r ⊥ , with
components x and y. Consider the fields and the source in the limit of β = 1.
(a) Show that the fields can be written as

~ = 2q ~
E
r⊥
δ(ct − z) , ~ = 2q v̂ × ~
B
r⊥
δ(ct − z) , (86)
2 2
r⊥ r⊥
where v̂ is a unit vector in the direction of the particle’s velocity.

We begin with eq. (11.154) on p. 560 of Jackson,

qR~
~=
E , (87)
R3 γ 2 (1 − β 2 sin2 ψ)3/2
where ψ is the angle between the vectors ~ v and R. ~ I have modified Jackson’s notation by
~
employing the symbol R for the vector that points from the charge q to the observation point ~
r=
12
(x, y, z) in reference frame K. Eq. (87) was also derived in class along with the corresponding
result for the magnetic field,
q(~
v × R)~
~ =
B . (88)
cR3 γ 2 (1 − β 2 sin2 ψ)3/2
12
Jackson denotes the vector that points from the charge q to the observation point (x, y, z) by ~r. However,
I prefer to employ ~
r to represent the vector that points from the origin of reference frame K to the observation
point, as shown in Fig. 1.

16
 x

(x, y, z)

~
r ~
R
ψ
vt
~
z
q
Figure 1: A charge q moving at constant velocity ~
v in the z-direction as seen from reference frame
K. The angle ψ is defined so that v̂· R̂ = cos ψ.

The reference frame K is exhibited in Fig. 1. It is evident from this figure that
~ =~
R r −~
vt . (89)

The velocity vector is taken to lie along the z-direction. That is, ~
v = v ẑ.
It is convenient to introduce the notation where

r ⊥ = xx̂ + y ŷ ,
~ r k = z~
~ z, (90)

so that ~ v = 0 and ~
r ⊥ ·~ ~ into components parallel
v = 0. Likewise, we can resolve the vector R
r k ×~
and perpendicular to the velocity vector,
~ =R
R ~k + R
~⊥ ,

where
~ k ≡ Rk ẑ = (z − vt)ẑ ,
R ~⊥ = ~
R r⊥ . (91)
~ ⊥ | ≡ R⊥ = R sin ψ. It follows that
after making use of eq. (89). In particular, note that |R

R3 (1 − β 2 sin2 ψ)3/2 = (R2 − β 2 R2 sin2 ψ)3/2 = (R⊥

2
+ Rk2 − β 2 R⊥
2 3/2
)

= [Rk2 + R⊥
2
(1 − β 2 )]3/2 = (Rk2 + R⊥
2
/γ 2 )3/2 . (92)

Note that in obtaining eq. (92) we used R2 = R⊥ 2

+ Rk2 and γ ≡ (1 − β 2 )−1/2 . Moreover, since
~⊥ = ~
R r⊥ [cf. eq. (91)], we may replace R⊥ with r⊥ ≡ |~ r ⊥ | = (x2 + y 2)1/2 in the above formulae.
Eqs. (87), (91) and (92) then yield
 
~= γq ~
r ⊥ + (z − vt)ẑ
E . (93)
(γ 2 Rk2 + r⊥
2 3/2
)

Likewise, eqs.(88), (91) and (92) yield

~ = γq(~
v×~ r⊥)
B 2 2 2 3/2
. (94)
c(γ Rk + r⊥ )

17
 Consider the limit of β → 1. In this limit, γ → ∞, and we see that
(
γ 0, if Rk 6= 0 ,
lim 2 2 2 3/2
=
γ→∞ (γ R + r ) ∞, if Rk = 0 .
k ⊥

This implies that

γ
lim = kδ(Rk ) , (95)
γ→∞ (γ 2 R2 2 3/2
+ r⊥ )
k

for some constant k. Note that in light of eq. (91),

lim Rk = z − ct ,
γ→∞

since γ → ∞ in the limit of v → c. To determine k, we integrate eq. (95) from −∞ to ∞, since Rk

can be any real number (either positive, negative or zero) depending on the value of the time t.
Thus, employing the substitution u = γRk ,
Z ∞ Z ∞ ∞
γ dRk du u 2
k= 2 2 2 3/2
= 2 2 3/2
= 2 2 2 1/2
= 2 .
−∞ (γ Rk + r⊥ ) −∞ (u + r⊥ ) r⊥ (u + r⊥ ) −∞ r⊥

Hence, we conclude that

γ 2
lim = 2 δ(z − ct) .
γ→∞ (γ 2 R2 2 3/2
+ r⊥ ) r⊥
k

Inserting this result back into eqs. (114) and (94) yields

~ = 2q ~
E
r⊥
δ(z − ct) , ~ = 2q v̂ × ~
B
r⊥
δ(z − ct) . (96)
2 2
r⊥ r⊥
~ above, we noted that in the limit of v → c, the z-component of the electric field
In obtaining E
is proportional to (z − ct)δ(z − ct) = 0 due to a property of the delta function. In obtaining B ~
above, we noted that limv→c ~ v /c = v̂. Finally, since the delta function is an even function of its
argument, we can write δ(z − ct) = δ(ct − z), and eq. (86) is verified.

(b) Show by substitution into the Maxwell equations that these fields are consistent with the
4-vector source density
J α = qcv α δ (2) (~
r ⊥ )δ(ct − z) ,
where the 4-vector v α = (1 ; v̂).

The four-vector current is given by J µ = (cρ ; J~). Hence, using the Maxwell equations in gaussian
units,
0
~ E
∇· ~ = 4πρ = 4πJ .
c
Hence, using eq. (96) and noting that Ez = 0, it follows that
   
0 c ~ ~ c ~ ~ ∂Ez qc ~ ~
r⊥
J = ∇· E = ∇⊥ · E + = δ(z − ct)∇⊥ · 2
, (97)
4π 4π ∂z 2π r⊥

18
where
~ ⊥ ≡ x̂ ∂/∂x + ŷ ∂/∂y .
∇ (98)
For ~
r ⊥ ≡ x x̂ + y ŷ 6= 0, an elementary computation yields
     
~ ⊥· ~
r ⊥ ∂ x ∂ y y 2 − x2 x2 − y 2
∇ 2
= + = + = 0. (99)
r⊥ ∂x x2 + y 2 ∂y x2 + y 2 (x2 + y 2 )2 (x2 + y 2 )2
To determine the behavior at ~
r ⊥ = 0, we consider the two-dimensional analogue of the divergence
theorem, Z   I Z 2π
~ ⊥· ~r ⊥ ~
r ⊥
dxdy ∇ 2
= r⊥ dφ 2 · r̂ ⊥ = dφ = 2π , (100)
A r⊥ C r⊥ 0

where A is a circular disk and C is the circular boundary of the disk. Note that r̂ ⊥ = ~
r ⊥ /r⊥ is
the outward normal to the circular boundary.
Eqs. (99) and (100) imply that
 
~ ~
r⊥
∇⊥ · 2
= 2πδ (2) (~
r⊥) , (101)
r⊥

where δ (2) (~
r ⊥ ) is a two-dimensional delta function. Inserting this result into eq. (97), we end up
with
J 0 = qcδ (2) (~
r ⊥ )δ(z − ct) . (102)
Next, we employ the Maxwell equation,
~
∇ ~ − 1 ∂ E = 4π J~ ,
~ ×B (103)
c ∂t c
to evaluate J~. First, we compute
r ⊥ = ẑ × (xx̂ + y ŷ) = xŷ − y x̂ ,
v̂ × ~ (104)

where we have used the fact that ~v points in the z direction. It then follows that
 
x̂ ŷ ẑ
   
v̂ × ~
r ⊥  ∂ ∂ ∂
~ ×
∇ 2
δ(z − ct) = det 
 ∂x ∂y

∂z  . (105)
r⊥  −y 
x
δ(z − ct) δ(z − ct) 0
x2 + y 2 x2 + y 2
Evaluating the determinant and making use of eqs. (90), (98) and (101) yields,
      
~ v̂ × ~r⊥ xx̂ + y ŷ ′ ∂ x ∂ y
∇× 2
δ(z − ct) = − 2 δ (z − ct) + + δ(z − ct)
r⊥ x + y2 ∂x x2 + y 2 ∂y x2 + y 2
 
~
r⊥ ′ ~ ~
r⊥
= − 2 δ (z − ct) + ẑ ∇⊥ · 2
δ(z − ct)
r⊥ r⊥

~
r⊥ ′
=− 2
δ (z − ct) + 2π ẑ δ (2) (~
r ⊥ )δ(z − ct) . (106)
r⊥

19
The prime refers to differentiation with respect to z. Finally, we compute
   
∂ ~ r⊥ ∂ ~ r⊥ c~
r⊥
2
δ(z − ct) = −c 2
δ(z − ct) = − 2 δ ′ (z − ct) . (107)
∂t r⊥ ∂z r⊥ r⊥

Inserting eq. (96) into eq. (103) and using eqs. (106) and (107), we obtain
   
qc v̂ × ~
r ⊥ q ∂ ~
r ⊥
J~ = ~ ×
∇ 2
δ(z − ct) − 2
δ(z − ct)
2π r⊥ 2π ∂t r⊥
 
qc ~
r⊥ ′ (2) ~
r⊥ ′
= − 2 δ (z − ct) + 2π ẑ δ (~ r ⊥ )δ(z − ct) + 2 δ (z − ct)
2π r⊥ r⊥

= qcv̂ δ (2) (~
r ⊥ )δ(z − ct) , (108)

after using the fact that v̂ = ẑ. Combining eqs. (102) and (108), we can write

J α = qcv α δ (2) (~
r ⊥ )δ(z − ct) ,

where the four-vector v α = (1 ; v̂).

(c) Show that the fields of part (a) are derivable from either of the following 4-vector potentials:

A0 = Az = −2qδ(ct − z) ln(λr⊥ ) , ~⊥ = 0 ,
A (109)

or
A0 = Az = 0 , ~ ⊥ = −2qΘ(ct − z)∇
A ~ ⊥ ln(λr⊥ ) , (110)
where λ is an irrelevant parameter setting the scale of the logarithm. Show that the two potentials
differ by a gauge transformation and find the corresponding gauge function χ.

~ Given the four-vector potential, the electromagnetic

The four-vector potential is Aµ = (Φ ; A).
fields are determined by
~
E ~ 0 − 1 ∂A ,
~ = −∇A ~ =∇
B ~ ×A
~.
c ∂t
Inserting the scalar and vector potentials given in eq. (109),
h i
~ δ(ct − z) ln(λr⊥ ) + 2q ẑ ln(λr⊥ ) ∂ δ(ct − z)
~ = 2q ∇
E
c ∂t
 h i  
∂ ∂ ∂ 1 ∂
= 2qδ(z − ct) x̂ + ŷ 1
2
ln(x2 + y 2 ) + ln λ + 2q ẑ ln(λr⊥ ) + δ(ct − z)
∂x ∂y ∂z c ∂t
~
r⊥
= 2q 2
δ(z − ct) ,
r⊥

20
 2
after using ~
r⊥ = xx̂ + y ŷ and r⊥ = x2 + y 2 . In particular, note that
 
∂ 1∂
+ f (ct − z) = 0 ,
∂z c ∂t
for any function of ct − z. Using eq. (109) to compute the magnetic field,
 
x̂ ŷ ẑ
 
∂ ∂ ∂ 
~ =∇
B ~ ×A ~ = det  
 ∂x ∂y ∂z 
 
0 0 −2q ln(λr⊥ )δ(ct − z)
 
∂ ∂
= −2qδ(ct − z) x̂ ln(λr⊥ ) − ŷ ln(λr⊥ )
∂y ∂x
 
∂ 1 2 2
 ∂ 1 2 2

= −2qδ(ct − z) x̂ ln(x + y ) + ln λ − ŷ ln(x + y ) + ln λ
∂y 2 ∂x 2

2q v̂ × ~r⊥
=− 2
(y x̂ − xŷ)δ(ct − z) = 2q 2
δ(ct − z) ,
r⊥ r⊥
after employing eq. (104).
Repeating these calculations using eq. (110),
~⊥  
1 ∂ A ∂ ∂ 1  ~
r⊥
~ =−
E = 2qδ(ct − z) x̂ + ŷ ln(x2
+ y 2
) + ln λ = 2q 2 δ(ct − z) ,
2
c ∂t ∂x ∂y r⊥
d
after using the relation between the delta function and the step function, δ(x) = Θ(x). In the
dx
computation of the magnetic field, we require the following result:
 
∂ ∂ 1  xx̂ + y ŷ
~ ⊥ ln(λr⊥ ) = x̂
∇ + ŷ ln(x2 + y 2 ) + ln λ = 2 .
2
∂x ∂y x + y2
Hence, it follows that
 
x̂ ŷ ẑ
 
 ∂ ∂ ∂
~ =∇
B ~ ×A ~ = −2q det   ∂x ∂y

∂z 
 
 x y 
Θ(ct − z) 2 Θ(ct − z) 0
x + y2 x2 + y 2
    
y x̂ − xŷ ∂ y ∂ x
= 2 δ(ct − z) + ẑ Θ(ct − z) −
x + y2 ∂x x2 + y 2 ∂y x2 + y 2

v̂ × ~r⊥
= 2q 2
δ(ct − z) ,
r⊥

21
after employing eq. (104) and noting that
   
∂ y ∂ x 2xy 2xy
2 2
− 2 2
=− 2 2 2
+ 2 = 0.
∂x x + y ∂y x + y (x + y ) (x + y 2 )2

Finally, we demonstrate that eqs. (109) and (110) differ by a gauge transformation. Under a
gauge transformation (using gaussian units),

~→A
~′ = A
~ + ∇χ
~ , 1 ∂χ
A A0 → A′ 0 = A0 − .
c ∂t
Denoting Aµ by eq. (109) and A′ µ by eq. (110), it follows that
∂χ
= −2qcδ(ct − z) ln(λr⊥ ) ,
∂t
~ ⊥ χ = −2qΘ(ct − z)∇
∇ ~ ⊥ ln(λr⊥ ) ,

∂χ
= 2qδ(ct − z) ln(λr⊥ ) .
∂z
The solution to these equations can be determined by inspection,

χ(~
x, t) = −2qΘ(ct − z) ln(λr⊥ ) ,

up to an overall additive constant.

5. [Jackson, problem 11.19] A particle of mass M and 4-momentum P decays into two particles
of mass m1 and m2 .
(a) Use the conservation of energy and momentum in the form p2 = P − p1 , and use the
invariance of scalar products of 4-vectors to show that the total energy of the first particle in the
rest frame of the decaying particle is13
M 2 + m21 − m22
E1 = ,
2M
and that E2 is obtained by interchanging m1 and m2 .

In the rest frame of the decaying particle (with 3-momentum P ~ , relativistic energy E and
mass M), we can explicitly write out the momentum four-vectors,7

P = (M ; ~0) , p1 = (E1 ; p
~1 ) , p1 = (E2 ; p
~2 ) . (111)

In particular, P ~ = ~0 (the zero-vector), since by definition the decaying particle in its rest frame
~ = γM~
is at rest, i.e. its velocity is zero so that P v = ~0. Likewise, E = γM, but in the rest frame
γ = 1 so that E = M as indicated in eq. (111).
13
Note that on p. 566, Jackson indicates that in particle kinematics, the symbols p and m stand for cp and mc2 ,
respectively. We follow this convention in this problem.

22
 To obtain an expression for E1 , we start from the conservation of the 4-momentum,

p2 = P − p1 .

Squaring this equation, and using the fact that squares of the corresponding 4-momenta are
Lorentz-invariant, P 2 = M 2 and p2i = mi (for i = 1, 2), it follows that

m22 = M 2 + m21 − 2P · p1 . (112)

The dot product of the two 4-momenta is a Lorentz invariant and can be evaluated in any
reference frame. In particular, using the rest-frame forms for the 4-momenta given in eq. (111),
it immediately follows that P · p1 = ME1 . Inserting this result in eq. (112) yields

m2 = M 2 + m21 − 2ME1 .

Solving for E1 then yields

M 2 + m21 − m22
E1 = . (113)
2M
To obtain E2 one can repeat the derivation above starting from p1 = P − p2 . Squaring this
equation and using P · p2 = ME2 , we see that the resulting derivation is identical to the previous
one with m1 and m2 interchanged. Thus,
M 2 + m22 − m21
E2 = . (114)
2M
Alternatively, we can obtain E2 from energy conservation, P 0 = p01 + p02 . That is,

E1 + E2 = M . (115)

Inserting eq. (113) into eq. (115) and solving for E2 , we again obtain eq. (114).

REMARK:
Although Jackson does not ask you to compute the magnitude of the 3-momentum of the
decaying particles, this can be easily done. By momentum conservation, P ~ =p ~2 = ~0, from
~1 + p
which it follows that p~1 = −~p2 . Hence, we can re-express the rest-frame four vectors p1 and p2
given in eq. (111) as p1 = (E1 ; p
~) and p2 = (E2 ; −~
p), where

~≡p
p ~1 = −~
p2 . (116)

Using the formula for relativistic energy, E12 = |~

p |2 +m21 , and the formula for E1 given in eq. (113),
we can solve for |~
p |,
q p
(M 2 + m21 − m22 )2 − 4M 2 m21
p | = E12 − m21 =
|~ .
2M
In the literature, this formula is often written as

λ1/2 (M 2 , m21 , m22 )

|~
p| = ,
2M
23
where the function λ(x, y, z) is defined by

λ(x, y, z) ≡ x2 + y 2 + z 2 − 2xy − 2xz − 2yz .

This function is ubiquitous in relativistic kinematics and is known as the triangle function.14

(b) Show that the kinetic energy Ti of the ith particle in the same frame is:
 
mi ∆M
Ti = ∆M 1 − − , (117)
M 2M
where ∆M ≡ M − m1 − m2 is the mass excess or Q-value of the process.

The kinetic energy Ti of the ith particle is defined as (cf. footnote 7):

Ti = Ei − mi ,

That is, we subtract the rest energy from the relativistic energy of the particle. Using eq. (113),
it immediately follows that:
M 2 + m21 − m22 (M − m1 )2 − m22 (M − m1 − m2 )(M − m1 + m2 )
T1 = − m1 = = .
2M 2M 2M
If we define ∆M ≡ M − m1 − m2 , then simple algebra yields
M − m1 + m2 m1 ∆M
=1− − .
2M M 2M
Hence, it follows that  
m1 ∆M
T1 = ∆M 1− − . (118)
M 2M
The calculation of T2 proceeds similarly by interchanging m1 and m2 . Since ∆M is invariant
under this interchange, it follows that:
 
m2 ∆M
T2 = ∆M 1 − − . (119)
M 2M
Thus, we have confirmed eq. (117). Note that
 
∆M + m1 + m2
T1 + T2 = ∆M 2 − = ∆M , (120)
M
after using the definition of ∆M given below eq. (117).
14
The√function λ is√called the triangle function since 41 [−λ(x, y, z)]1/2 is the area of a triangle with sides of
√
lengths x, y and z. However, in the application to the decay of one particle into two particles, we have
λ(M 2 , m21 , m22 ) ≥ 0 if the decay is kinematically allowed (i.e. consistent with energy conservation), in which case
M ≥ m1 + m2 and the corresponding triangle of lengths M , m1 and m2 does not exist. The triangle function can
be expressed in many equivalent ways. For further details, see E. Byckling and K. Kajantie, Particle Kinematics
(John Wiley & Sons Ltd., London, UK, 1973). This book is considered to be the bible for relativistic kinematics,
although it is sadly long out of print.

24
 (c) The charged pi-meson (M = 139.6 MeV) decays into a mu-meson15 (m1 = 105.7 MeV)
and a neutrino (m2 = 0). Calculate the kinetic energies of the mu-meson and the neutrino in
the pi-meson’s rest frame. The unique kinetic energy of the muon is the signature of a two-body
decay. It entered importantly in the discovery of the pi-meson in photographic emulsions by
Powell and coworkers in 1947.

For charged pi-meson decay into a muon and a neutrino, the sum of the muon and neutrino kinetic
energies [cf. eq. (120)] is T1 + T2 = ∆M = 139.6 − 105.7 = 33.9 MeV. Using eqs. (118) and (119),
 
105.7 33.9
T1 = 1 − − 33.9 MeV = (0.1214)(33.9 MeV) = 4.1 MeV ,
139.6 2(139.6)
 
33.9
T2 = 1 − 33.9 MeV = (0.8786)(33.9 MeV) = 29.8 MeV .
2(139.6)

6. [Jackson, problem 11.27, part (a)] A charge density ρ′ of zero total charge, but with a dipole
moment p ~ ′ , exists in reference frame K ′ . There is no current density in K ′ . The frame K ′ moves
with velocity ~ ~ in the frame K. Find the charge and current densities ρ and J~ in the frame
v = βc
K and show that there is a magnetic dipole moment m ~ = 12 (~ ~ correct to first order in β.
p ′ × β),
What is the electric dipole moment p ~ in K to the same order in β?16

In frame K ′ , we have J~ ′ = 0 and the charge density ρ′ (~ x ′ ) is assumed to be time-independent.

Furthermore, the total charge in frame K ′ is zero, i.e.
Z
x ′ ) d3x′ = 0 ,
ρ′ (~ (121)

whereas the electric dipole moment in frame K ′ , denoted by p ~ ′ , is assumed to be nonzero. By

definition, the electric dipole moment is given by
Z
′
~ = ~
p x ′ ρ′ (~
x ′ ) d3 x′ . (122)

Frame K ′ moves with velocity ~ ~ with respect to frame K. We shall employ eqs. (22) and
v = βc
(23), which we repeat here:
~ J~ ′ ) ,
cρ = γ(cρ′ + β· (123)
γ − 1 ~ ~′ ~ ~ ′.
J~ = J~ ′ + (β· J )β + γ βcρ (124)
β2

Setting J~ ′ = 0, it follows that

x ′) ,
x) = γρ′ (~
ρ(~ J~(~ ~ ′ (~
x) = γ βcρ ~ x) .
x ′ ) = βcρ(~ (125)
15
The nomenclature mu-meson is no longer standard. The currently accepted name for this particle is the muon.
16
Jackson denotes the electric dipole moment vector in frame K ′ by p ~, which is misleading notation. Thus, I
have denoted the electric dipole moment in frame K ′ by p
~ ′ and the electric dipole moment in frame K by p~.

25
Hence, the electric dipole moment in frame K is given by
Z
~= ~
p x ρ(~x) d3 x , (126)

and the magnetic dipole moment is defined (in gaussian units) in frame K as
Z Z Z 
1 ~ 3 1 ~ 3 1 3 ~,
m~ = x × J (~
~ x) d x = x × β ρ(~
~ x) d x = x ρ(~
~ x) d x × β
2c 2 2
after employing eq. (125) in the second step above. In light of eq. (126), it follows that

~ = 12 p
m ~.
~×β (127)

In order to relate p~ to p
~ ′ , we need to compare ρ(~ x ′ ). Eq. (11.19) of Jackson gives
x) and ρ′ (~
the four-vector x′ µ in terms of xµ . The inverse relation is obtained by changing the sign of β, ~
which yields
~ x ′) , γ−1 ~ ′ ~ ~ ′0 .
x0 = γ(x′0 + β·~ x=~
~ x′ + (β·~ x )β + γ βx (128)
β2
Working to first order in β, we can set γ = (1 − β 2 )−1/2 ≃ 1, and
~ x′ ,
x0 ≃ x′0 + β·~ x≃~
~ ~ ′.
x ′ + βx (129)
0

In the same approximation, d3 x′ ≃ d3 x and ρ(~ x ′ ). Hence, using eq. (121),

x) ≃ ρ′ (~
Z Z Z
~= ~
p x ρ(~ 3
x) d x ≃ (~ ′ ~ ′ ′
x + βx0 )ρ (~ ′ 3 ′
x )d x = ~ x ′ ρ′ (~
x ′ ) d3 x′ = p
~′ . (130)

to first order in β. Using eqs. (127) and (130), we conclude that to first order in β,

~ = 21 p
m ~.
~′ × β

ADDENDUM:

Although Jackson only asks for the result to first order in β, it is not too hard to derive
the exact result. By assumption, ρ′ (~ x ′ ) is a time-independent charge density in frame K ′ which
satisfies eq. (121) [i.e. the total charge vanishes]. We are asked to compute p ~ and m ~ in frame K
at a fixed time x0 . Using eq. (128), it follows that
x0 ~ ′
x′0 = − β·~
x .
γ
Inserting this result back into eq. (128) then yields
 
′ γ−1 ~ x ′ )β
~ + βx
~ 0.
x=~
~ x + − γ (β·~
β2

We can simplify this expression by noting that γ = (1 − β 2 )−1/2 implies that β 2 γ 2 = γ 2 − 1.

26
Substituting for β 2 above,
γ−1 γ
2
−γ = − . (131)
β γ+1
Hence,
γ ~ x ′ )β
~ + βx
~ 0.
~ ~′−
x=x (β·~ (132)
γ+1
We can evaluate the Jacobian of the transformation, eq. (132),17
∂xi γ
′
= δij − βi βj , (133)
∂xj γ+1
which we evaluate at fixed x0 . To compute the determinant of the Jacobian, we use the following
general result which is easily proved (see the Appendix following this addendum):
a| 2 .
det(δij − ai aj ) = 1 − |~
Hence,    
∂xi γβ 2 1 γ2 − 1 1
det =1− = γ+1− = .
∂x′j γ+1 γ+1 γ γ
It follows that  
3 ∂xi d3 x′
d x = det d3 x′ = , (134)
∂x′j γ
which is just the well-known length contraction in special relativity. Using eqs. (125) and (134),
it follows that the charge dq located inside an infinitesimal volume element is
dq = ρ(~ x ′ ) d3x′ ,
x) d3 x = ρ′ (~ (135)
since the factors of γ cancel out. Eq. (135) is expected since the electric charge is a Lorentz scalar,
which must be independent of the reference frame used to evaluate it.
We now can compute p ~ in frame K. Using eqs. (132) and (135),
Z Z Z  
′ ′ γ ~ x )β
′ ~ + βx
~ 0 ρ′ (~
~= ~
p x ρ(~ 3
x) d x = ~ ′
x ρ (~ 3 ′
x )d x = x −
~ (β·~ x ′ ) d3 x′ . (136)
γ+1
Using eq. (122), it follows that
γ ~ ′ ~
p ~′ −
~=p
(β·~p )β (137)
γ+1
where we have used eq. (121) [i.e., the total charge in frame K ′ vanishes] to eliminate the last
term in eq. (136). Indeed, if we work to first order in β, then the second term on the right hand
side of eq. (137) can be dropped and we recover the result of eq. (130).
To obtain m,
~ we start with the exact result obtained in eq. (127). Inserting eq. (137) for p~
~ ~
and using β × β = 0, it follows that 18

~ = 1p
m ~′ × β~, (138)
2

which we now see is exact to all orders in β.

17
In eq. (133), we do not distinguish between lowered and raised indices, as all involved quantities are three-
dimensional.
18
In the literature, one often finds the result of eq. (138) quoted without the factor of 21 . For a discussion of this
discrepancy, see V. Hnizdo, Magnetic dipole moment of a moving electric dipole, American Journal of Physics 80,
645–647 (2012).

27
REMARK:

One can derive eq. (137) more directly as follows. Using eqs. (125) and (128),
Z Z  
′ γ−1 ~ ′ ~ ~
~= ~
p x ρ(~ 3
x) d x = γ x +
~ x )β + γ βx0 ρ′ (~
(β·~ ′
x ′ ) d3 x
β2
Z  
′ γ−1 ~ ′ ~ ~
=γ x +
~ 2
(β·~x )β + γ βx0 ρ′ (~
′
x ′ ) δ(x0 − ct)d4 x .
β
R
In the last step, I inserted the integral δ(x0 − ct)dx0 = 1 which does not change the result. This
is useful, since d4 x = d4 x′ = d3 x′ dx′0 . Using eq. (128) to express x0 in terms of x′0 ,
Z    
′ γ −1 ~ ′ ~ ~ ′ ′ ′ ′ ~ x −
′ ct
~=
p x +
~ 2
(β·~ x )β + γ βx 0 ρ (~ x ) δ x0 + β·~ d3 x′ dx′0 ,
β γ
~ x ′ − ct/γ). Integrating over x′ yields
after writing γδ(x0 − ct) = δ([x0 − ct]/γ) = δ(x′0 + β·~ 0
Z     Z
γ−1 ~ x ′ )β
~ ρ′ (~ ~ ρ′ (~
p~= x′ +
~ − γ (β·~ x ′ ) d3 x′ + ctβ x ′ ) d3 x′ .
β2
R ′ ′ ′ 3 ′
Using eqs. (121) and (131) along with p ~′ = ~ x ρ (~ x ) d x , we end up with
γ ~ ′ ~
p ~′ −
~=p (β·~
p )β ,
γ+1
which reproduces eq. (137). Of course, the two derivations are equivalent.

APPENDIX: Proof of a determinantal formula

Theorem: Let A be an n × n matrix, whose matrix elements are given by:

Aij = δij − ai aj . (139)
Then,
n
X
2 2
det A = 1 − |~
a| , where |~
a| = a2i . (140)
i=1

Proof: The determinant of A is defined as

det A = ǫi1 i2 ···in A1i1 A2i2 · · · Anin ,
where there is an implicit sum over repeated indices. Plugging in eq. (139),
det A = ǫi1 i2 ···iN (δ1i1 − a1 ai1 )(δ2i2 − a2 ai2 ) · · · (δnin − an ain ) .
Expanding the product above, and using the Kronecker deltas to perform the sums, we obtain
det A = ǫ123···n − a1 ai1 ǫi1 23···n − a2 ai2 ǫ1i2 3···n − · · · − an ain ǫ123···in
= ǫ123···n (1 − a21 − a22 − · · · − a2n ) , (141)

28
after performing the final set of summations and using the fact that ǫi1 i2 ···in vanishes unless all
of its indices are distinct. Note the absence of any terms in eq. (141) that are quartic (or higher
order) in the ai , since all such terms will be symmetric under the interchange of two indices that
are summed against two corresponding indices of the Levi-Civita tensor. For example,

ǫi1 i2 ···iN a1 ai1 a2 ai2 = 0 ,

since ai1 ai2 is symmetric under the interchange of i1 and i2 whereas ǫi1 i2 ···iN is antisymmetric
under this interchange of indices.
Since ǫ123···n = 1, eq. (141) yields
n
X
det A = 1 − a2i ,
i=1

which completes the proof.

29