Ben Sauerwine

Classical Electrodynamics Homework 11

v v
1) Suppose in a general coordinate system K , E and B fields are given. Find v
the velocity,
v with respect to K , of a new coordinate system K ' such that E '
and B ' are parallel in K ' .

I notice that when taking the Lorentz transform, the components of electric and
magnetic field in the direction of the boost are not affected butv the
v components in the
orthogonal directions are. Thus, my boost should be in the B × E direction (and if the
v v
fields start with B × E = 0 , there is no need to transform at all.)

Writing the electromagnetic fields as a Lorentz tensor, I have:

⎡ E ⎤
⎢0 − 0 0 ⎥
c
⎢ v v ⎥
⎢ ⎛ v B ⋅ E v ⎞⎥ ⎡ tv ⎤
⎜ B− r E ⎟ ⎢ E ⎥
⎢E v ⎜ ⎟⎥ ⎢ ⎥
2
B ⋅⎜ E
⎢ 0 0 v v ⎟⎥ E
⎢ ⎥
⎜ B − Bv⋅ E E ⎟⎥ ⎢ v B ⋅ E v ⎥
⎢ c v v v v
⎢ ⎜ E2 ⎟⎥ B − r E
μν ⎢ ⎝ v v ⎠⎥ ⎢ E 2v ⎥
F =
⎢0 −
B ⋅ E ⎥ v B⋅E v ⎥
⎢ v
⎥⎢ B − v 2 E ⎥
0 0
⎢ E
⎢ v v ⎥⎢ E ⎥
⎢ ⎛ v B⋅E v ⎞ ⎥ ⎢ v v ⎥
⎜ B− r E ⎟ v v ⎢ B× E ⎥
⎢ v ⎜ ⎟ B⋅E ⎥ v v
⎥ ⎢⎣ B × E ⎥⎦
2
⎢ 0 − B ⋅⎜ vE v ⎟ 0
⎜ B − Bv⋅ E E ⎟
⎢ v v E ⎥
⎢ ⎜ 2 ⎟ ⎥
⎣ ⎝ E ⎠ ⎦
Above, I have defined the dimensions v inv the vector multiplying the tensor. Now
taking the boost to be solely in the B × E direction, I have the Lorentz transform:
⎡ γ 0 0 − βγ ⎤
⎢ 0 1 0 0 ⎥⎥
Λμν = ⎢
⎢ 0 0 1 0 ⎥
⎢ ⎥
⎣− βγ 0 0 γ ⎦
v
The above corresponds to a boost β . Now applying this, I get (using Mathematica to
take the dot products)
⎡ 0 F 01 + βF 13 βF 23 0 ⎤
⎢ 13 ⎥
− F − βF
01 13
0 0 βF + F ⎥
01
Λμ α Λν β F αβ = γ ⎢
⎢ − βF 23 0 0 − F 23 ⎥
⎢ ⎥
⎣⎢ 0 − βF 01 − F 13 F 23 0 ⎦⎥
Now certainly all that’s left to do is find the value of velocity β to ensure that in this
 v v
frame B × E = 0 . In terms of the result of the boosted matrix, this constraint will be:
(F 23 )(− βF 23 ) − (βF 01 + F 13 )(− F 01 − βF 13 ) = 0
where I have ignored the constant factor imparted on the electric field from the
definition of the F tensor.

β=
( ) − (F ) + (F )
− F 01
2 13 2 23 2
± ((F ) + (F ) − (F ) )
01 2 13 2 23 2
2
( ) (F )
− 4 F 01
2 13 2

v v 2 F 01 F 13
in the B × E direction.

I may try to rewrite this to take on a more aesthetically pleasing form

⎡ v v ⎛ v v2 ⎞
2 ⎤
⎢ v 2 (B ⋅ E)2 ⎜v v (B⋅vE) v v 2 ⎟ ⎥
v2 B − v v v2 − v2 v2 v v 2 ⎥
(B⋅ E) ⎜ E (B⋅ E) ⎟ 4 E B − (B⋅ E) ⎥
2
1⎛ c ⎞ E ⎢ 2 B
β = ⎜ v v ⎟⎢− 2 − E2 E2
v v 2 + v2 ± ⎜ 2 + v v 2 − v2 ⎟ − 2 v v
2 ⎝ B ⋅ E ⎠⎢ c ⎛ v B ⋅ E v ⎞ E ⎜ c ⎛ v B⋅ E v ⎞ E ⎟ c ⎛ v B⋅ E v ⎞2 ⎥
⎢ ⎜⎜ B − v 2 E ⎟⎟ ⎜ ⎜⎜ B − v 2 E ⎟⎟ ⎟ ⎜⎜ B − v 2 E ⎟⎟ ⎥
⎢ ⎝ E ⎠ ⎝ ⎝ E ⎠ ⎠ ⎝ E ⎠ ⎥
⎣ ⎦
Here, one must take the positive root—while this is not a pretty solution, it is valid.
Simpler forms exist, but this one works just as well.

2) Derive the general transformation of electric and magnetic fields under the
Lorentz transformation, given in eq (11.149) in Jackson.

For a boost in one direction, I have:

⎡ γ − βγ 0 0⎤
⎢− βγ γ 0 0⎥⎥
μ
Λν = ⎢
⎢ 0 0 1 0⎥
⎢ ⎥
⎣ 0 0 0 1⎦
which will act on a general EM field tensor
⎡ Ex Ey E ⎤
⎢ 0 − − − z⎥
⎢ c c c ⎥
E
⎢ x 0 − Bz By ⎥
F =μν ⎢ c ⎥
⎢ Ey ⎥
⎢ Bz 0 − Bx ⎥
⎢ c ⎥
⎢ Ez − B Bx 0 ⎥
⎣⎢ c ⎦⎥
y

My generalized boost tensor is identical to Jackson’s equation (11.19) and (11.98).

Jackson’s formulas
v v v v γ2 v v v
E ' = γ (E + β × B ) − β (β ⋅ E )
γ +1
v v v v γ2 v v v
B' = γ (B − β × E ) − β (β ⋅ B )
γ +1
will be given by
F μν = Λμ α Λν β F αβ if I take the boost in the appropriate direction. Using the basis
⎡ t ⎤
⎢ β ⎥
⎢ || ⎥ , indicating time, the direction parallel to boost and two directions
⎢ β (1) + ⎥
⎢ (2 ) ⎥
⎣β + ⎦
perpendicular to boost, I may rewrite
⎡ E|| E (1) + E (2 ) + ⎤
⎢ 0 − − − ⎥
⎢ E c c c ⎥
⎢ || 0 − B (2 ) + B (1) + ⎥
F μν = ⎢⎢ c(1) ⎥ in the basis above.
E + ⎥
⎢ B (2 ) + 0 − B|| ⎥
⎢ (c2 ) ⎥
⎢ E + − B (1)
⎢⎣ c + B|| 0 ⎥⎥
⎦
μν μ ν αβ
Taking F = Λ α Λ β F with Mathematica now gives
γ 2 (1 − β 2 ) = 1
(1) (2 )
⎡ E|| E⊥ (2 ) E⊥ (1) ⎤
⎢ 0 − −γ + γβB⊥ −γ − γβB⊥ ⎥
⎢ c c c
(1) (2 ) ⎥
⎢ E ||
0
(2 )
− γB⊥ + γβ
E⊥ (1)
γB⊥ + γβ
E⊥ ⎥
⎢ c ⎥
Λμ α Λν β F αβ = ⎢ (1)
c
(1)
c
⎥
⎢γ E ⊥ − γβB⊥ (2 ) γB⊥ (2 ) − γβ E ⊥ 0 − B|| ⎥
⎢ c c ⎥
⎢ E ⊥ (2 ) (1) (1) E⊥
(2 ) ⎥
⎢γ + γβB⊥ − γB⊥ − γβ B|| 0 ⎥
⎣ c c ⎦
Now it ris clear to see from the form of this matrix that the terms
v v v v v v r
E ' ⊥ = γE ⊥ + γβ × B B' ⊥ = γB⊥ − γβ × E
v r v v
E '|| = E|| ⋅ βˆ B'|| = B|| ⋅ βˆ
Of course, to get the beautiful form of Jackson, one must take note that:
γ2 2 ⎛ γ2 ⎛ 1 ⎞⎞
γE i − β Ei = ⎜⎜ γ − ⎜⎜1 − 2 ⎟⎟ ⎟ E i = E i
γ +1 ⎝ γ + 1 ⎝ γ ⎠ ⎟⎠
which
v v allows v us to remove the dependence on the perpendicular components, e.g.
E = E|| + E ⊥

v v v
v2
3) Show that the Lagrangian given by L = −mc 2 1 − 2
+ eA ⋅ v − eA0 gives,
c
through Euler-Lagrange equations, the equation of motion for a particle
moving in the electromagnetic fields. What happens to this Lagrangian if we
perform a gauge transformation? Discuss this consequence.
 v
Note that here, A μ = (φ , A) . This is indeed a properly formulated Lagrangian with the
first term being that of a free particle, the last term being just the Coulomb potential, and
the center interaction term corresponding to the magnetic force.
v v
∂p v v v v v v v v ∂A
Locally, I expect = qv × B + qE . Note that B = ∇ × A E = −∇ ⋅ φ − .
∂t ∂t

∂L d ⎛ ∂L ⎞
From the Euler-Lagrange equations, I expect = ⎜ ⎟ (note the total derivative
∂qi dt ⎜⎝ ∂q& i ⎟⎠
versus partial). Applying these derivatives locally, I have then
v v v
v2
L = −mc 1 − 2 + eA ⋅ v − eA0
2

∂L v v v v v v v v v v v
v = ∇(eA ⋅ v ) − e∇φ = e(v × (∇ × A) + (v ⋅ ∇ )A) − e∇φ
∂q
v v v
d ∂L d v ∂ v dA ∂p ∂A v v v
v = (mγv + eA) = m γv + e + e(v ⋅ ∇ )A
v
= +e
dt ∂q& dt ∂t dt ∂t ∂t
∂L d ∂L
v= v
∂q dt ∂q&
v v
v v v v v v v ∂p ∂A v v v
e(v × (∇ × A) + (v ⋅ ∇ )A) − e∇φ = +e + e(v ⋅ ∇ )A
∂t ∂t
v v
v v v ⎛ v ∂A ⎞ ∂p
e(v × (∇ × A)) + e⎜⎜ − ∇φ − ⎟⎟ =
⎝ ∂t ⎠ ∂t
v ∂p v
v v
ev × B + eE =
∂t

∂f v v v
Under a gauge transformation, φ → φ − A → A + ∇f
∂t
 v v v ∂f ⎞
+ e(A + ∇f ) ⋅ v − e⎜ A0 + ⎟
v2 v ⎛
L = − mc 2 1 −
c 2
⎝ ∂t ⎠
∂L v v v v⎛ ∂f ⎞
v = ∇(e(A + ∇f ) ⋅ v ) − e∇⎜ φ − ⎟
v
∂q ⎝ ∂t ⎠
⎛v v v
v v v v v v
( ) v v v
= e(v × (∇ × A) + (v ⋅ ∇ )A) + e⎜ v × ∇×∇f + (v ⋅ ∇ )∇f
⎜
⎞
⎟ − e∇φ + e∇ ∂f
⎟
v v
∂t
⎝ →0 ⎠
d ∂L d v ∂ v d v v
v = (mγv + eA) = m γv + e (A + ∇f )
v
dt ∂q& dt ∂t dt
v v
∂p ∂A v v v ∂ v v v v
= +e + e(v ⋅ ∇ )A + e ∇f + e(v ⋅ ∇ )∇f
∂t ∂t ∂t
∂L d ∂L
v= v
∂q dt ∂q&
(obvious cancellations )
v v
v v v v ∂p ∂A
e(v × (∇ × A)) − e∇φ = +e
∂t ∂t

and now I see I am on the same track as in the original derivation of the equations of
motion, so despite the fact that the Lagrangian itself isn’t gauge invariant (see the first
v v ∂f
line above—canceling the e(A + ∇f ) ⋅ v vs e
v
certainly makes no sense), I still get
∂t
the same equations of motion, so this isn’t a big problem.