Given a real vector space V , an alternating k

Given a real vector space V , an alternating k-form on V is a map ω : V k → R
which is multilinear and alternating, i.e.
ω(. . . , v + w, . . .) = ω(. . . , v, . . .) + ω(. . . , w, . . .);
ω(. . . , cv, . . .) = cω(. . . , v, . . .);
ω(. . . , v, w, . . .) = −ω(. . . , w, v, . . .).
Note, that the alternating property implies that for σ ∈ Sk ,
ω(vσ(1) , . . . , vσ(k) ) = sign(σ)ω(v1 , . . . , vk ).
The set of alternating k-forms on V is a real vector space Ak (V ) with
(ω + η)(v1 , . . . , vk ) = ω(v1 , . . . , vk ) + η(v1 , . . . , vk );
(cω)(v1 , . . . , vk ) = cω(v1 , . . . , vk ).
An example of an alternating form is the determinant, det ∈ An (Rn ).
Given ω ∈ Ak (V ) and η ∈ Aj (V ) the wedge product ω ∧ η ∈ Ak+j (V ) is
defined by
X
ω ∧ η(v1 , . . . , vk+j ) =
sign(σ)ω(vσ(1) , . . . , vσ(j) )η(vσ(j+1) , . . . , vσ(j+k) ).
σ∈Sk+j
Corollary: The following properties hold for the wedge product.
ω ∧ η = (−1)jk η ∧ ω;
(ω ∧ η) ∧ ψ = ω ∧ (η ∧ ψ);
(ω + α) ∧ η = ω ∧ η + α ∧ η;
ω ∧ (η + β) = ω ∧ η + ω ∧ β;
(cω) ∧ η = ω ∧ (cη) = c(ω ∧ η).
Theorem: Let {e1 , . . . , en } be a basis for V . Let {e1 , . . . , en } be the dual basis
for A1 (V ), i.e. linear maps V → R such that ei (ej ) = δji . Then {eI1 ∧ . . . ∧ eIk :
1 ≤ I1 < · · · < Ik ≤ n} is a basis for Ak (V ).
Proof: ?
A multi-index is a sequence I1 , . . . , Ik ∈ N of the form 1 ≤ I1 < · · · < Ik ≤ n.
The set of such multi-indices is denoted I(n, k). Note that #I(n, k) = (nk ). For
I ∈ I(n, k), ω I denotes ω I1 ∧ . . . ∧ ω Ik , so the basis above can be written more
compactly as {eI : I ∈ I(n, k)}.
Corollary: dim Ak (V ) = (nk ) where n = dim V .
Given a linear map f : W → V , the pullback f ∗ ω ∈ Ak (W ) is defined by
f ∗ ω(w1 , . . . , wn ) = ω(f (w1 ), . . . , f (wn )).
1
Corollary: The following properties hold for the pullback.
f ∗ (ω + α) = f ∗ ω + f ∗ α;
f ∗ (cω) = cf ∗ ω;
f ∗ (ω ∧ η) = f ∗ ω ∧ f ∗ η.
Given a manifold M , a differential k-form on M is a map ω : M → ∪p∈M Ak (Tp M )
with ω(p) ∈ Ak (Tp M ) such that for any smooth vector fields v1 , . . . , vn , the map
M → R, p 7→ ω(p)(v1 (p), . . . , vn (p)) is smooth. The set of differential k-forms
on M is denoted Ωk (M ) and is endowed with a real vector space structure and a
wedge product inherited from the Ak (Tp M )’s. Also, Ω0 (M ) is the ring of scalar
fields over which Ωk (M ) is a module. Scalar multiplication is both senses and
wedge products are all compatible.
The exterior derivative d : Ω0 (M ) → Ω1 (M ) is defined by dg(p) = Dp g.
Theorem: Let U be an open subset of Rn and let xi ∈ Ω0 (U ) be the standard
coordinate projections, i.e. P
xi (p) = pi . Then there exists unique {ωI ∈ Ω0 (U ) :
I ∈ I(n, k)} such that ω = I∈I(n,k) ωI dxI .
Proof: The xi are restrictions of the linear maps ei which are dual to the
standard basis for Rn , hence dxi = ei , so this follows from the basis theorem
for Ak (V ) with V =P
Rn .
n
Corollary: dg = i=1 ∂i gdxi .
P
The exterior derivative d : Ωk (U ) → Ωk+1 (U ) is defined by dω = I∈I(n,k) dωI ∧
dxI .
Given a smooth map f : N → M , the pullback f ∗ ω ∈ Ωk (N ) is defined by
∗
f ω(p) = (Dp f )∗ ω(p).
Theorem: For ω ∈ Ωk (M ) there exists a unique dω ∈ Ωk+1 (M ), the exterior
derivative, such that for any parametrization f : U → M , df ∗ ω = f ∗ dω.
Proof: ?
Note, the three definitions of the exterior derivative are all compatible.
Corollary: The following properties hold for the exterior derivative and the
pullback.
f ∗ (ω + η) = f ∗ ω + f ∗ η;
f ∗ (cω) = cf ∗ ω;
f ∗ (ω ∧ η) = f ∗ ω ∧ f ∗ η;
d(ω + η) = dω + dη;
d(cω) = cdω;
d(ω ∧ η) = dω ∧ η + (−1)k ω ∧ dη;
df ∗ ω = f ∗ dω;
ddω = 0.
Pn
Proof: On U , ddg = d j=1 ∂j gdxj = i,j=1 ∂i ∂j gdxi ∧ dxj . Equality of
mixed partials and antisymmetry of wedge product
of 1-forms therefore implies
P
that ddg = −ddg. So, ddg = 0. So, ddω = I∈I(n,k) ddωI ∧ dxI = 0. On M ,
f ∗ ddω = ddf ∗ ω = 0, so ddω = 0 (why?).
Pn
2