Sei sulla pagina 1di 8

The de Rham Cohomology and a Variational Principle:

Why electromagnetism is mathematically trivial


Quinton Westrich
Abstract
We derive the Maxwell equations for electromagnetism in the language of differential
forms. First, we define a few objects, such as differential forms and the de Rham
cohomology vector spaces. Then we show how the Poincaré Lemma applied to the de
Rham cohomology on Rn relates to two of the Maxwell equations and the existence
of a potential field. Finally, we apply a variational principle to obtain the last two
Maxwell equations. Charge conservation is a direct result of these latter equations.

Contents
1 Exterior Algebra 1

2 Differential Forms 3

3 The de Rham Cohomology 5

4 Maxwell’s Equations for Electromagnetism 6

5 A Variational Principle 7

6 Conclusion 8

1 Exterior Algebra
Convention 1.1. All vector spaces are finite dimensional over R.

Definition 1.2. A map


k
Y
ϕ: Vi → R,
i=1

where the Vi are vector spaces, is called k-linear if it is linear in each argument, i.e., if

v 7→ ϕ(v1 , . . . , vi−1 , v, vi+1 , . . . , vk )

is linear for each choice of v1 , . . . , vi−1 , vi+1 , . . . , vk .

1
Remark 1.3. The set of all k-linear maps from V k to R forms a vector space and is denoted

T k (V ) = {ϕ : V k → R : ϕ is k-linear}.

So T 1 (V ) = V ∗ and we set T 0 (V ) = R.

Definition 1.4. An element ϕ ∈ T k (V ) is called alternating if

ϕ(v1 , . . . , vi , . . . , vj , . . . , vk ) = −ϕ(v1 , . . . , vj , . . . , vi , . . . , vk ).

Definition 1.5. The vector space Λk V of all k-linear alternating forms with real
values is
Λk V = {ϕ ∈ T k (V ) : ϕ is alternating}.

Definition 1.6. The wedge product, or exterior product, of a k-form and a `-form is a
(k + `)-form:

∧ : Λk V × Λ` V → Λk+` V
∧ : (ϕ, ψ) 7→ ϕ ∧ ψ
1 X
(ϕ ∧ ψ)(v1 , . . . , vk+` ) = sgn (σ)ϕ(vσ(1) , . . . , vσ(k) )ψ(vσ(k+1) , . . . , vσ(k+`) )
k!`! σ∈S
n

Remarks 1.7.

1. If β 1 , . . . , β n is a basis for V ∗ , then a basis of Λk V is given by

β i1 ∧ β i 2 ∧ · · · ∧ β ik , 1 ≤ i1 < i2 < · · · < ik ≤ n,

and therefore  
k n
dim Λ V = .
k
Thus, any p-form can be written in this basis as
X
ϕ= ϕi1 ···ik β i1 ∧ · · · ∧ β ik , ϕi1 ···ik ∈ R,
i1 <i2 <···<ik

or
1 X
ϕ= ϕi1 ···ik β i1 ∧ · · · ∧ β ik , ϕi1 ···ik ∈ R.
k! i ,...,i
1 k

2. dim Λn V = 1 so any n-form can be written as

ϕ = kβ 1 ∧ · · · ∧ β n .

An orientation of V is the choice of one of the two equivalence classes of Λn V r {0},


i.e., some non-zero n-form ϕ modulo a positive factor.

2
2 Differential Forms
Convention 2.1. All manifolds and maps are smooth, unless qualified otherwise.
Definition 2.2. A section of a vector bundle π : E → B is a continuous function s : B → E
such that π ◦ s = idB . If M is a manifold, a section of T M is called a vector field on M .
Definition 2.3. If f : M → R, then the differential of f is the section of T ∗ M defined by

df (p)(X) = Xp (f ), for Xp ∈ Tp M.

Remark 2.4. If p ∈ U , where U is an open subset of a manifold M , and x : U → Rn is a


coordinate chart for M , then a basis for Tp (U ) is given by

∂ ∂
,..., n .
∂x1 p ∂x p

A vector field for T U is given by



X(p) = .
∂x1 p
In this case, for any f : M → R, we have

∂f
df (p)(X) = .
∂x1 p

In particular,
∂x1

1
dx (p)(X) = = 1.
∂x1 p
In general,
!
∂xi


dxi (p) = = δij ,
∂xj p ∂xj p
so that dx1p , . . . , dxnp is a basis for Tp∗ U . In particular, we have
n
X ∂f i
df = dx .
i=1
∂xi

Definition 2.5. Let k ∈ {0, 1, . . . , n} and M be a smooth manifold. A differential form


of degree k, or k-form for short, is a mapping

ω : M → Λk (T M ), ϕ : p 7→ ϕp ∈ Λk (Tp M ).

Remark 2.6. If p ∈ U , where U is an open subset of a manifold M , and x : U → Rn is a


coordinate chart for M , dx1p , . . . , dxnp , is a basis for Λ1 (Tp U ). Thus, a basis for Λk (Tp U ) is

dxi1 ∧ dxi2 ∧ · · · ∧ dxik , 1 ≤ i1 < i2 < · · · < ik ≤ n.

3
Definition 2.7. Let ω give an orientation of M , g be a metric on M , and dxi be an oriented
orthonormal basis of T ∗ M = Λ1 (T M ). The Hodge star is the linear isomorphism
∗ : Λk (T M ) → Λn−k (T M )
1
∗(dxi1 ∧ · · · ∧ dxik ) = εi ···i η i1 i1 · · · η ik ik dxik+1 ∧ · · · ∧ dxin .
(n − k)! 1 n
Definition 2.8. The exterior derivative d is the linear map
d : Λk (T M ) → Λk+1 (T M )
1 X
dω = dωi1 ···ik ∧ dxi1 ∧ · · · ∧ dxik
k! i ,...,i
1 k

1 X ∂
= i
ωi1 ···ik dxi ∧ dxi1 ∧ · · · ∧ dxik .
k! i,i ,...,i ∂x
1 k

Remarks 2.9. 1. Although the Hodge star and the exterior derivative were both defined
in terms of a coordinate system, they are in fact both independent of the choice of
coordinates.
2. Let f : R3 → R. Then
∂f ∂f ∂f
df =
dx + dy + dz.
∂x ∂y ∂z
The 1-form df can be identified with the vector
 
∂f ∂f ∂f
∇f = , , .
∂x ∂y ∂z

3. Let ω = ω1 dx + ω2 dy + ω3 dz be a 1-form. Then


     
∂ω3 ∂ω2 ∂ω3 ∂ω1 ∂ω2 ∂ω1
dω = − dy ∧ dz + − dx ∧ dz + − dx ∧ dy.
∂y ∂z ∂x ∂z ∂x ∂y
Identifying
dy ∧ dz ↔ x̂, dx ∧ dz ↔ ŷ, dx ∧ dy ↔ ẑ,
we obtain
dω = ∇ × (ω1 , ω2 , ω3 ).

4. Let ω = ω3 dx ∧ dy − ω2 dx ∧ dz + ω1 dy ∧ dz. Then


dω = ∇ · (ω1 , ω2 , ω3 ) dx ∧ dy ∧ dz.
Theorem 2.10 (Stoke’s Theorem). If M is an oriented n-manifold with boundary ∂M and
ω is an (n − 1)-form on M with compact support, then
Z Z
dω = ω.
M ∂M

4
3 The de Rham Cohomology
Definition 3.1. The de Rham complex on a manifold M is the graded algebra
n
M

Ω (M ) = Λk (T M )
k=0

together with the operator d. It gives rise to the exact sequence


d d
· · · −→ Λk−1 (T M ) −→ Λk (T M ) −→ Λk+1 (T M ) −→ · · · .

Definition 3.2. Forms ω which satisfy dω = 0 are called closed. For each k, we have the
vector space

Z k (M ) = {ω ∈ Λk (M ) : ω is closed} = ker{d : Λk (T M ) → Λk+1 (T M )}.

If there exists a (k − 1)-form ϕ such that ω = dϕ, then ω is called exact. The vector space

B k (M ) = {ω ∈ Λk (M ) : ω is exact} = im {d : Λk−1 (T M ) → Λk (T M )}.

Further, we set B 0 (M ) = {0}.

Theorem 3.3. d2 = 0.

Proof. Note that d2 = (d : Λk (T M ) → Λk+1 (T M )) ◦ (d : Λk−1 (T M ) → Λk (T M )). Now,

1 X ∂2
ddω = ωi ···i dxj ∧ dxi ∧ dxi1 ∧ · · · ∧ dxip = 0
k!(k + 1)! i,j,i ,...,i ∂xj ∂xi 1 k
1 k

since the partial derivatives commute whereas dxi ∧ dxj = −dxj ∧ dxi .

Corollary 3.4. Every exact form is closed.

Corollary 3.5. B k (M ) is a vector subspace of Z k (M ).

Definition 3.6. The quotient vector space

ker{d : Λk (T M ) → Λk+1 (T M )}
H k (M ) = Z k (M )/B k (M ) =
im {d : Λk−1 (T M ) → Λk (T M )}
is called the k-dimensional de Rham cohomology vector space of M .

Theorem 3.7 (Poincaré Lemma). If M is smoothly contractible to a point p0 ∈ M , then


every closed form ω on M is exact.

k n {0} for k > 0
Corollary 3.8. H (R ) =
R for k = 0.

5
4 Maxwell’s Equations for Electromagnetism
Maxwell’s equations are
∇·E = ρ ∇·B = 0
∂E ∂B
∇×B− =J ∇×E− =0
∂t ∂t
where E is the electric field and B is the magnetic field, ρ is the charge density, and J is the
electric current density. (We use Heaviside units and set c = 1.)
To describe electromagnetism in terms of differential forms, we set our spacetime to
be Minkowski space: R4 with Cartesian coordinates x0 , x1 , x2 , x3 as an oriented coordinate
system and metric
 
∂ ∂
g , = ηµν , η = diag(1, −1, −1, −1).
∂xµ ∂xν
Definition 4.1. The electromagnetic field 2-form (tensor) is
F = −E1 dx0 ∧ dx1 − E2 dx0 ∧ dx2 − E3 dx0 ∧ dx3 + B1 dx2 ∧ dx3 − B2 dx1 ∧ dx3 + B3 dx1 ∧ dx2
1
= Fµν dxµ ∧ dxν .
2!
∂B
(
∇×E− =0
Theorem 4.2. dF = 0 ⇐⇒ ∂t
∇·B=0
Corollary 4.3. F is exact.
Remark 4.4. Gauge freedom occurs since F = dAλ for any Aλ = A0 + dλ, where
λ ∈ Λ0 (R4 ). The map A 7→ A + dλ is called a gauge transformation.
∂E
(
∇×B− =0
Theorem 4.5. d∗F = 0 ⇐⇒ ∂t
∇·E=0
Definition 4.6. The current density 4-vector j µ has components (ρ, J 1 , J 2 , J 3 ). The
current density 3-form is
j = ρ dx1 ∧ dx2 ∧ dx3 − J 1 dx0 ∧ dx2 ∧ dx3 + J 2 dx0 ∧ dx1 ∧ dx3 − J 3 dx0 ∧ dx1 ∧ dx2
1
= εµνλρ j µ dxν ∧ dxλ ∧ dxρ .
3!
(It is natural to define the current 3-form since integrating over a volume now gives the total
charge contained in that volume.)
∂E
(
∇×B− =J
Theorem 4.7. d∗F = j ⇐⇒ ∂t
∇·E=ρ
Corollary 4.8 (The Continuity Equation). dj = ∂µ j µ dx0 ∧ dx1 ∧ dx2 ∧ dx3 = 0

6
5 A Variational Principle
Theorem 5.1. An extremal of the action integral
Z  
1
S[A] = − F ∧ ∗F − j ∧ A
2
obeys the differential equation
d ∗ F = j.
Moreoever, this action integral is gauge invariant provided dj = 0 (i.e. provided the current
is conserved).
Definition 5.2. Let M be an oriented manifold with metric g of signature r − s. The
coderivative δω of a differential form ω is given by the map
δ : Λk (T M ) → Λk−1 (T M ) given by δ = (−1)nk+n+1+s ∗ d ∗ .
Lemma 5.3. Let ϕ ∈ Λk (T M ), and M be an oriented n-manifold with metric of signature
r − s. Then
Z Z
ϕ ∧ ∗ψ = ψ ∧ ∗ϕ
Z Z
dϕ ∧ ∗ψ = ϕ ∧ ∗δψ

∗ ∗ ϕ = (−1)k(n−1)+s ϕ
d(ϕ ∧ ψ) = dϕ ∧ ψ + (−1)k ϕ ∧ dψ.
Proof of Theorem 5.1. The standard first variation calculation is
Z  
1 1
S[A + a] − S[A] = − d(A + a) ∧ ∗d(A + a) − j ∧ (A + a) + dA ∧ ∗dA + j ∧ A
2 2
Z  
1 1
= − (dA + da) ∧ (∗dA + ∗da) − j ∧ A − j ∧ a + dA ∧ ∗dA + j ∧ A
2 2
Z  
1 1
= − (dA ∧ ∗dA + da ∧ ∗dA + dA ∧ ∗da + da ∧ ∗da) − j ∧ a + dA ∧ ∗dA
2 2
Z  
1
= − (da ∧ ∗dA + dA ∧ ∗da + da ∧ ∗da) − j ∧ a
2
Z  
1
= − da ∧ ∗dA − j ∧ a − da ∧ ∗da
|2 {z }
nonlinear in a
Z
−a ∧ ∗δdA + a ∧ j + O(a2 )

=
Z
= a ∧ (− ∗ δdA + j) + O(a2 )
Z
= a ∧ (−d ∗ F + j) + O(a2 ).

7
Since a was arbitrary and the linear term in the expansion is required to vanish, we have the
desired result
d ∗ F = j.
To show gauge invariance, we compute
Z  
1 1
S[A + dλ] − S[A] = − d(A + dλ) ∧ ∗d(A + dλ) − j ∧ (A + dλ) + dA ∧ ∗dA + j ∧ A
2 2
Z  
1 1
= − dA ∧ ∗dA − j ∧ A − j ∧ dλ + dA ∧ ∗dA + j ∧ A
2 2
Z
= (−j ∧ dλ)
Z
= d(j ∧ λ) − (dj) ∧ λ
Z Z
= j∧λ− (dj) ∧ λ.
∂M M

Modulo boundary contributions and provided dj = 0, we have that S[A + dλ] = S[A] for an
arbitrary gauge transformation (0-form) λ. This gives the desired gauge invariance.

6 Conclusion
The Maxwell electromagnetic field equations can be written compactly as

dF = 0
d ∗ F = j.

If spacetime is contractible, then dF = 0 is equivalent to F being exact, i.e. the existence of


a “potential” 1-form A such that F = dA. Finally, the functional S[A] obtains an extremum
for A if d ∗ F = j, which then implies dj = 0, i.e. charge conservation.

References
[1] Bott and Tu. Differential Forms in Algebraic Topology. Springer-Verlag, 1982.

[2] Bredon. Topology and Geometry. Springer-Verlag, 1993.

[3] Göckeler and Schücker. Differential Geometry, Gauge Theories, and Gravity. Cambridge
University Press, 1987.

Potrebbero piacerti anche