Professional Documents
Culture Documents
where the Lagrangian L is a function of the scalar, Φ and it’s spacetime derivatives ∂µ Φ,
L = L (Φ(x), ∂µ Φ(x)) .
For the free scalar field, the Lagrangian can be taken to be,
1 1
L = − ∂µ Φ ∂ µ Φ − m2 Φ2 . (1)
2 2
This form is dictated by Lorentz/Poincare invariance and the funny factors of − 12 are there so that
the kinetic terms come with the right sign and right factor, e.g. 12 Φ̇2 . m is invariant quantity with
the dimensions of mass (or energy). Upon quantizing the system, the parameter, m will turn out
to be the mass of the scalar field quanta/particles.
Check:
• Lorentz invariance is rather obvious because most terms in the action is Lorentz invariant,
d4 x, m2 , Φ2 . Even the kinetic term, ∂µ Φ ∂ µ Φ, is because the Lorentz index µ is contracted,
viz:
∂µ Φ(x) → ∂µ0 Φ0 (x0 ) = Λµ ν ∂ν Φ(x),
1
=⇒ ∂µ Φ(x) ∂ µ Φ(x) → ∂µ0 Φ0 (x0 )∂ 0µ Φ0 (x0 ) = Λµ ν ∂ν Φ(x) Λµ α ∂ α Φ(x)
= (Λµ ν ) (Λµ α ) ∂ν Φ(x) ∂ α Φ(x)
= δαν ∂ν Φ(x) ∂ α Φ(x)
= ∂ν Φ(x) ∂ ν Φ(x).
remains invariant.
• Translation invariance is also obvious because the action integral being defined over all space
and time i.e ranges of integration being (−∞, ∞), is independent of the origin of coordinates
and there is no explicit dependence on the coordinates, x.
or ˆ
δI = d4 x (∂ρ ω µ ν ) M µ νρ .
This form is consistent with our expectation that in the special case when a and ω are constants
these expressions must vanish as the action is invariant under the global/constant changes. The
2
companion terms i.e. T µ ν or M µ νρ which are some functions of scalar field and it’s derivatives are
the conserved quantities. This can be inferred from the above expressions by a simple integration
by parts and abandoning the total derivative term. For example for the translation invariance:
ˆ
δI = d4 x (∂µ aν ) Tµ ν
ˆ ˆ
= d x ∂µ (a T ν ) − d4 x (∂µ Tµ ν ) aν
4 ν µ
ˆ
= − d4 x (∂µ Tµ ν ) aν .
Now when we go to the global case, i.e. aν becomes a constant, it can be dragged out of the
integral and we have the expression,
ˆ
ν
δI = −a d4 x (∂µ Tµ ν ) .
Since the change in the action for this global aν is zero, the integrand must vanish. The vanishing
of the integrand is nothing but the continuity equation,
∂µ Tµ ν = 0,
in other words, T µ ν is a conserved current. In the following we use the Noether procedure to
extract the conserved quantities, M µνρ and Tµ µν
ν explicitly in terms of Φ and it’s derivatives, ∂µ Φ.
3
The derivatives transform like,
∂xν
∂µ → ∂µ0 = ∂ν
∂x0µ
= δµν − ∂µ aν ∂ν
= ∂µ − ∂µ aν ∂ν . (3)
Φ0 (x0 ) = Φ(x),
while,
∂xν
∂µ0 Φ0 (x0 ) = ∂ν Φ(x)
∂x0µ
= ∂µ Φ(x) − ∂µ aν ∂ν Φ(x). (4)
4
where we have identified the conserved current,
∂L
Tµ ν ≡ δνµ L − ∂ν Φ(x) .
∂ (∂µ Φ(x)
Raising the ν-index we arrive at the expression of the canonical stress-energy-momentum tensor,
∂L
Tµν = η µν L − ∂ ν Φ(x),
∂ (∂µ Φ(x)
= η µν L − (∂ µ Φ) (∂ν Φ) . (5)
• One can go ahead and check that charges corresponding to the current is nothing but the
four-momentum, P µ , i.e. ˆ
P = d3 x T0ν .
ν
• This tensor is symmetric between the indices, µ, ν. This is only true for scalar fields. For
the Maxwell field, we will see that the corresponding stress tensor will not be symmetric.
• Note that the stress-energy-momentum is non-unique to some extent. One can always add
a term like, ∂λ B λµν where B is a tensor that has the following antisymmetric properties,
B λµν = −B µλν .
5
The new quantity1 ,
T µν = Tµν + ∂λ B λµν
is also conserved,
∂µ T µν = ∂µ Tµν + ∂µ ∂λ B λµν = 0.
For the Maxwell field, one can exploit this ambiguity to define a stress tensor which is
symmetric in the indices, µ and ν,
T µν = T νµ .
Before we do that we need to first obtain the expression for the charges conserved as a result
of Lorentz invariance.
Λµ ν → Λµ ν (x)
|J| = 1 + ω ρ ρ + xρ ∂σ ω σ ρ
= 1 + xρ ∂ σ ω σ ρ . (6)
6
The corresponding change in the action integral is,
ˆ
0 0
d4 x0 L Φ0 (x0 ), ∂µ0 Φ0 (x0 )
I [Φ (x )] =
ˆ
= d4 x (1 + xρ ∂σ ω σ ρ ) L (Φ(x), ∂µ Φ(x) + ωµ ν ∂ν Φ(x) + (∂µ ωρ ν ) xρ ∂ν Φ(x))
ˆ
∂L(x)
= d x (1 + x ∂σ ω ρ ) L (Φ(x), ∂µ Φ(x)) + ωµ ν ∂ν Φ(x)
4 ρ σ
∂(∂µ Φ(x))
ν ρ ∂L(x) 2
+ (∂µ ωρ ) x ∂ν Φ(x) + O(ω )
∂(∂µ Φ(x))
ˆ ˆ
4 4 ρ σ ν ρ ∂L
= d x L (Φ(x), ∂µ Φ(x)) + d x x ∂σ ω ρ L + ∂µ ωρ x ∂ν Φ + . . . + O(ω 2 ).
∂(∂µ Φ)
where in the last line steps we have only bothered to show the terms which contain derivatives
of ω because the non-derivative terms integrate out to zero. Thus the first order variation in the
action,
ˆ
0 0 4 ρ σ ν ρ ∂L
δI ≡ I [Φ (x )] − I [Φ(x)] = d x x ∂σ ω ρ L + ∂µ ωρ x ∂ν Φ
∂(∂µ Φ)
ˆ
4 ρ µν ∂L ν
= d x ∂µ ωνρ x η L − ∂ Φ
∂(∂µ Φ)
ˆ
= d4 x ∂ρ ωµν xν Tρµ
ˆ
= d4 x ∂ρ ωµν (xν Tρµ − xµ Tρν ) .
This object is antisymmetric in the first pair of indices, i.e. µ, ν while it is symmetric in the
exchange of the third index, ρ with any one of the first pair of indices i.e. µ or ν .
´ ´
One can check that the charges i.e., Lµν ≡ d3 x M µν0 = d3 x (xµ T0ν − xν T0µ ) are simply the
components of the angular momentum and the boost generators.
7
Again, the action is symmetric under translations,
x → x0 = x + a,
Aµ (x) → A0µ (x0 ) = Aµ (x).
since the integration range is all over spacetime and there is no explicit dependence on x (only
implicit dependence on x via the fields). As a result there exists a conserved current, Tµν , which
can be extracted thru the Noether procedure,
∂L
Tµν = η µν L − ∂ ν Aρ ,
∂ (∂µ Aρ )
1
= − η µν Fρσ F ρσ + F µρ ∂ ν Aρ . (9)
4
Evidently this is not symmetric under µ ↔ ν.
What about the angular momentum and boost generators? Here we need to remind ourselves
that the Maxwell field carries spin-1 hence the angular momentum must get a contribution from
internal spin in addition to orbital term and hence would look different from the scalar example
(8). One can go thru the similar steps, and obtain,
ˆ
1
δI = d4 x ∂µ ωνρ [− (xν F µα F ρ α − xρ F µα F ν α ) + (η µν xρ L − η µρ xν L)]
2
ˆ
1
= d4 x ∂µ ωνρ [xρ (η µν L + F µα F ν α ) − ν ↔ ρ ] .
2
8
So this last line helps us identify a conserved current, M ρνµ which are the Lorentz generators
∂µ M ρνµ = 0,
So we see that the Lorentz generators for the Maxwell field are can be expressed as,
M µνρ ≡ xµ T νρ − xν T µρ ,
T µν ≡ η µν L + F µα F ν α
1
= − η µν Fρσ F ρσ + F µα F ν α . (10)
4
This tensor is evidently different from the canonical stress tensor for the Maxwell field (9). In the
following exercise you will prove that this symmetric tensor, T µν is a “new and improved” stress
tensor for Maxwell fields i.e. it is conserved and it gives us right expression fro energy density and
momentum density of EM field.
Homework:
2. Check that indeed components of these coincide with energy density and momentum density
of the Maxwell field, i.e. , T 00 = 12 (E2 + B2 ) and T 0i = E × B.
2
This is a a result of scale invariance of the Maxwell action. A scale transformation is given by scaling all fields
and coordinates by their mass (energy) dimensions x0 → λx, Aµ → λ−1 Aµ . The scalar action is not invariant under
this scaling symmetry due to the “mass term” 21 m2 φ2 which does not scale properly because m is a parameter and
not a field.