Lecture 21
M. Siddikov
[email protected]
November 20, 2023
Outline
Discussion of exam 1
−See separate file “Soluciones...pdf” in
Aula.
Revision of Special Relativity
−Electrodynamics in tensorial notations.
Electromagnetic tensor and Maxwell’s
equations
−Energy-momentum tensor for electro-
magnetic fields
Certamenes
Certamen 3 (+Examen final):
−28/11
Summary of last lecture
Action for a particle in external EM field:
r
v2
Z
S = S0 + Sint , S0 = −mc 2 dt 1−
c2
Z Z
dxµ
Sint = −q dxµ Aµ (x) = q dτ (u · A) , uµ ≡
dτ
du µ dp µ
Equation of motion : m ≡ = q F µν uν (1)
dτ dτ
0 Ex /c Ey /c Ez /c
−Ex /c 0 −Bz By
Fµν ≡ ∂µ Aν − ∂ν Aµ =
−Ey /c
(2)
Bz 0 −Bx
−Ez /c −By Bx 0
−Use ηµν , η µν to raise/lower indices (affects signs in front of electric field)
Fµν is gauge-invariant: (1) does not
Fµν ≡ ∂µ Aν − ∂ν Aµ (1)
change under
Fµν is called electromagnetic field tensor
Aµ → Aµ + ∂µ f (x),
Antisymmetric, rank 2⇒6 independent
components where f (x) is arbitrary function. Obviously,
this translates in components into familiar
0 Ex /c Ey /c Ez /c 1
−Ex /c 0 −Bz By ϕ→ϕ− ∂ f,
Fµν = c t
−Ey /c Bz 0 −Bx
A → A + ∇f
−Ez /c −By Bx 0
Contravariant components:
Relation to vector and scalar potentials:
F µν = η µα Fαβ η βν =
Aµ = (ϕ/c, A) ,
0 −Ex /c −Ey /c −Ez /c
Aµ = ηµν Aν = (ϕ/c, −A) Ex /c 0 −Bz By
=
(fix constants from E = −∇ϕ − ∂A/∂t, Ey /c Bz 0 −Bx
B = ∇ × A) Ez /c −By Bx 0
Transformation of Fµν under boosts
The derivative and vector potential transform as
∂ ∂ ∂x α ∂ ∂
→ ′µ
≡ = Λµα α Aµ → Aµ′ ≡ Λµβ′ Aβ (1)
∂x µ ∂x ∂x ′µ ∂x α ∂x
From (1) we can see that transformation of Fµν is given by
Fµν ≡ (∂µ Aν − ∂ν Aµ ) → Λµα Λνβ ∂α Aβ − Λνα Λµβ ∂α Aβ
Swap (rename) dummy indices α ↔ β:
Λµα Λνβ ∂α Aβ − Λνα Λµβ ∂α Aβ = Λµα Λνβ (∂α Aβ − ∂β Aα ) = Λµα Λνβ Fαβ
⇒Thus Fµν transforms as
Fµν → Λµα Λνβ Fαβ
For contravariant tensor get
F µν = η µα η νβ Fαβ
µα
and we know that η doesn’t change under Lorentz boosts
so
F µν → η µα η νβ Λαγ Λβδ Fγδ = η µα η νβ Λαγ Λβδ ηγσ ηδτ F στ = Λµσ Λντ F στ
F µν → Λµα Λνβ F αβ (1)
Fαβ → Λαµ Λβν Fµν
Explicit expressions for boost in direction x̂ :
Ex′ = Ex , Bx′ = Bx
Ey′ = γ (Ey − v Bz ) = γ (E + v × B)y ,
Ez′ = γ (Ez + v By ) = γ (E + v × B)z ,
′
v 1
By = γ By + 2 Ez = γ B − 2 v × E ,
c c y
v 1
Bz′ = γ Bz − 2 Ey = γ B − 2 v × E .
c c z
γ v /c γ 0 0
vγ
γ 0 0 Ey′ ≡ c F02
′
= Λ0α Λ2β Fαβ
Λµν (v x̂) =
c ,
0 0 1 0 only (αβ) = (0, 2) and (αβ) = (1, 2)
0 0 0 1 contribute
1
γ(v ) = p Λ00 Λ22 F02 + Λ01 Λ22 F12 =
1 − v 2 /c 2
= γ · 1 · Ey + γ v · 1 · (−Bz )
Ex′ ≡ c F01
′
= Λ0α Λ1β Fαβ
⇒ Ey′ = γ (Ey − v Bz ).
only (αβ) = (0, 1) and (αβ) = (1, 0)
contribute; take into account F10 = −F01 : Ez′ ≡ c F03
′
= Λ0α Λ3β Fαβ
Λ00 Λ11 − Λ01 Λ10 = γ 2 − γ 2 v 2 /c 2
only (αβ) = (0, 3) and (αβ) = (1, 3)
2
1 − v /c 2 contribute
= =1
1 − v 2 /c 2 Λ00 Λ33 F03 + Λ01 Λ33 F13 =
⇒ Ex′ = Ex . = γ · 1 · Ez + γv · 1 · (By )
Bx′ ≡ F23
′
= Λ2α Λ3β Fαβ ⇒ Ez′ = γ (Ez + v By ).
−Similarly we obtain resuts for By ′ , Bz ′
only (αβ)=(2, 3) gives nonzero result⇒
Bx′ = Bx
Technical comment (only for rank 2 tensors!)
.
F µν → Λµα Λνβ F αβ , (1)
Fµν → Λµα Λνβ Fαβ (2)
Can write (1,2) in matrix notations:
µν
F µν = Λ · F · ΛT
Do not confuse Λµα and Λµα ;
... due to Λµν Λαν = δαµ expect
Λµα (v ) = Λµα (−v )
.Mathematica: We have simple product of
3 matrices
In agreement with our earlier result:
Ex′ = Ex , Bx′ = Bx ,
Ey′ = γ (Ey − v Bz ) , Ez′ = γ (Ez + v By ) ,
v v
By′ = γ By + 2 Ez , Bz′ = γ Bz − 2 Ey .
c c
Boosts in arbitrary directions
Boosts in direction of x̂:
Ex′ = Ex , Bx′ = Bx ,
Ey′ = γ (Ey − v Bz ) , Ez′ = γ (Ez + v By ) ,
v v
By′ = γ By + 2 Ez , Bz′ = γ Bz − 2 Ey .
c c
Can rewrite this as
E||′ = E|| , B||′ = B|| ,
1
E⊥ ′ = γ (E⊥ + v × B ⊥ ) , B⊥ ′ = γ B⊥ − 2 v × E ⊥ , (1)
c
where E ⊥ , B ⊥ are parts of the field ⊥ to vector v (components y , z in our case)
Can generalize this result for boosts in arbitary direction: decompose fields as
E = E || + E ⊥ , B = B || + B ⊥ ,
(E · v ) v (E · v ) v
E || = (E · v̂ ) v̂ = , E ⊥ = E − E || = E −
v2 v2
and similar for B || , B ⊥ . Take into account that E || , B || do not change under boost in
direction of v , whereas E ⊥ , B ⊥ change according to (1).
Boosts in arbitrary directions
Final result:
γ 2 /c 2
E ′ = γ (E + v × B) − v (v · E )
γ+1
γ 2 /c 2
v ×E
B′ = γ B − 2
− v (v · B),
c γ+1
(Jackson, 11.149)
−Alternatively, could get this result writing Λ(v ) in components for arbitrary boosts
µν
F µν = Λ · F · ΛT
vi vj Λi0
Λ00 = γ, Λi j = δij + (γ − 1) , Λi 0 = −γvi , Λ0i = ,
v2 c2
Control question
In Leture 6 (p. 4) we introduced vectors Q, Q ∗ defined as
i i √
Q=B+ E, Q∗ = B − E, c −1 = ε 0 µ0
c c
and demonstrated that Maxwell’s equations might be rewritten in such a way that they
include either only Q (and its derivatives) or only Q ∗ , but not both.
r
i ∂Q µ0
∇×Q + = µ0 j , ∇·Q =i ρ,
c ∂t ε0
i ∂Q ∗
r
µ0
∇ × Q∗ − = µ0 j , ∇ · Q ∗ = −i ρ,
c ∂t ε0
−Show that the fields Q, Q ∗ do NOT mix under Lorentz boosts, and find explicit
transformation law.
−Explain this result from the point of view of Lorentz group properties which we
discussed in the previous lectures.
γ 2 /c 2
E ′ = γ (E + v × B) − v (v · E )
γ+1
γ 2 /c 2
v ×E
B′ = γ B − − v (v · B),
c2 γ+1
⇓
γ 2 /c 2
i ′ v ×Q
Q ′= B ′ + E =γ Q +i − v (v · Q)
c c γ+1
v × Q∗ γ 2 /c 2
′ i
(Q ∗ ) = B ′ − E ′ = γ Q ∗ − i − v (v · Q ∗ )
c c γ+1
Note that under rotations fields E , B, Q, Q ∗ all transform in a similar way, as
Ea → Rab Eb , Ba → Rab Bb , Qa → Rab Qb
where Rab is rotation matrix
γ 2 /c 2
i ′ v ×Q
Q ′= B ′ + E =γ Q +i − v (v · Q) (1)
c c γ+1
v × Q∗ γ 2 /c 2
′ i
(Q ∗ ) = B ′ − E ′ = γ Q ∗ − i − v (v · Q ∗ ) (2)
c c γ+1
Transformation of E , B, Q, Q ∗ under rotations
Qa → Rab Qb , Qa∗ → Rab Qb∗ ,
⇓
Group-theoretic point of view: Fields (E , B) belong to (≈transform as elements of)
some reducible representation of Lorentz group. Vectors Q, Q ∗ correspond to vectors
which form 2 irreducible sub-representations of smaller dimension.
−Earlier result: Lorentz algebra is ∼ so(3)L ⊕ so(3)R .
∗
If we drop in (1,2) last O(v 2 /c 2 ) terms (assuming v ≪ c), can see that trans-
formation of the fields Q, Q ∗ coincides with infinitesimal rotations of vectors by
elements of ∼ so(3)L , so(3)R
∗
In some advanced books You may find statement that “electromagnetic fields
(E , B) transform/belong to representation (1, 0) + (0, 1)”. This simply refers to
such decomposition into irreducible Q, Q ∗ which don’t mix under Lorentz boosts.
Note that components of Gαβ are related to
Demonstrate that the quantities
Fαβ by substitution E /c → B, B → −E /c
I1 = Fµν F µν ≡ Fµν Fαβ η µα η νβ
Gµν G µν = −I1
and
1 1
I2 = ϵαβµν F αβ F µν ≡ Fµν Fαβ ϵαβµν I2 = Fµν G µν = ϵαβµν F αβ F µν =
2 2
are invariants. Evaluate their values in = 2 ϵ0ijk F 0i F jk = 4 E · B/c
terms of electric and magnetic fields Take into account that in the penultimate
line jk ̸= 0, i and there is permutation kj
Earlier we have seen that contraction de- Note that
creases rank by 2;
no “free” (uncontracted) indices⇒scalars 1
I2 = ϵαβµν F αβ F µν =
2
I1 = 2F0i F 0i + Fij F ij = 2 B 2 − E 2 /c 2
1
= ϵαβµν ∂ α Aβ − ∂ β Aα F µν =
2
”Dual” tensor:
= ϵαβµν ∂ α Aβ F µν = ∂ α ϵαβµν Aβ F µν
1 − ϵαβµν Aβ ϵαβµν ∂ α F µν
Gµν ≡ ϵµναβ F αβ =
2
0 −Bx −By −Bz But
Bx 0 −Ez /c Ey /c ϵαβµν ∂ α F µν = ϵαβµν ∂ α (∂ µ Aν − ∂ ν Aµ ) = 0
=
By Ez /c 0 −Ex /c
Bz −Ey /c Ex /c 0 ⇒ I2 = ∂ α Jα , Jα = ϵαβµν Aβ F µν
Demonstrate that there are only two independent invariants which we may construct
using Fµν , i.e. any other scalar built using Fµν might be represented as a function of
I1 and I2 .
Some examples:
... Det(F ) = (E · B)2 /c 2 = I22 /4
... If use notation F · F = Fµα F αν , then can show with Mathematica that
Tr (F · F ) ≡ I1
I12 I2
Tr (F · F · F · F ) = − 2
2 4
I13 3 I1 I22
Tr (F · F · F · F · F · F ) = − ,
4 16
...
| · {z
Tr( F ... · F} ) = 0
odd number
If You have any ideas how we could prove this statement in general, please explain
