0% found this document useful (0 votes)
13 views

457661lecture 6

Uploaded by

Angelo Oppio
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
0% found this document useful (0 votes)
13 views

457661lecture 6

Uploaded by

Angelo Oppio
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 5

Spin Geometry: Lecture 6

Shuhan Jiang

By definition, every covariant derivative ∇ on E defines a linear map Ω0 (E) → Ω1 (E). It


can be extended to a linear map d∇ : Ωp (E) → Ωp+1 (E) by setting

d∇ (ω ⊗ s) = dω ⊗ s + (−1)p ω ∧ ∇(s).

d∇ is referred to as an exterior covariant derivative on E. By construction,

d∇ |Ω0 (E) = ∇.

Exercise 2.11. Show that d∇ can be locally written as

d∇ = dxµ ∧ ∇µ .

Example 2.9. The exterior derivative d is an exterior covariant derivative on R.

d∇ should not be confused as a covariant derivative on the vector bundle ΛT ∗ M ⊗ E.


Proposition 2.5. A covariant derivative ∇ on E induces covariant derivatives

1. on the endomorphism bundle End(E), which we denote by ∇ with a slight abuse of


notation, and is defined by:

∇(H)(s) := ∇(H(s)) − H(∇(s))

for any H ∈ Γ(End(E)) and s ∈ Γ(E);

2. on the dual bundle E ∨ , which we again denote by ∇ with a slight abuse of notation,
and is defined by:

∇(α)(s) := d(α(s)) − α(∇(s))

for any α ∈ Γ(E ∨ ) and s ∈ Γ(E);

Covariant derivatives ∇i on vector bundles Ei over the same base manifold, i = 1, 2, induces
a covariant derivative ∇ on the tensor product bundle E1 ⊗ E2 by

∇(s1 ⊗ s2 ) = ∇1 (s1 ) ⊗ s2 + s1 ⊗ ∇2 (s2 ).

The proof is left as an exercise:

1
Exercise 2.12. Check that

1. the above connections are well-defined;


2. the induced connections on End(E) and E ⊗ E ∨ are equivalent under the natural iso-
morphism End(E) ∼ = E ⊗ E ∨.

Show that a connection ∇ on E also induces connections on T (E), Λ(E), and Sym(E).
Proposition 2.6. There exists a 2-form F with values in End(E) such that F = d∇ ◦ d∇ .
F is called the curvature of ∇. F satisfies the so-called second Bianchi identity
d∇ F = 0.

Proof. Let’s first show that F = d∇ ◦ d∇ is C ∞ (M )-linear:


F (f s) = d∇ (df ⊗ s + f ∇s) = d2 f ⊗ s − df ⊗ ∇s + df ⊗ ∇s + f d∇ (∇s) = f (F s).
The second Bianchi identity can also be proved by the following one line computation:
d∇ (F )(s) = d∇ (F (s)) − F (d∇ s) = d∇ (d∇ ◦ d∇ (s)) − (d∇ ◦ d∇ )(d∇ s) = 0.

Locally, we have
F (s) = (d + A)2 (s) = d2 (s) + d(A(s)) + A(d(s)) + (A ∧ A)(s) = (dA + A ∧ A)(s).
Therefore,
F = dA + A ∧ A. (2.7)
Component-wise, we have
∂ ∂ ∂Aν ∂Aµ
Fµν = F ( µ
, ν) = µ
− + [Aµ , Aν ]. (2.8)
∂x ∂x ∂x ∂xν
Remark 2.2. The formulas (2.7) and (2.8) probably need some explanation. In general, let
A = p!1 Aµ1 ...µp dxµ1 ∧ · · · ∧ dxµp and B = q!1 Bν1 ...νq dxν1 ∧ · · · ∧ dxνq be two matrix-valued p-form
and q-form, where the matrix valued coefficients Aµ1 ,...,µp are skew symmetric with respect to
the indices. The wedge product between A and B is defined as
1
A ∧ B := Aµ1 ...µp Bν1 ...νq dxµ1 ∧ · · · ∧ dxµp ∧ dxν1 ∧ · · · ∧ dxνq
p!q!
X X
= Aµ1 ...µp Bν1 ...νq dxµ1 ∧ · · · ∧ dxµp ∧ dxν1 ∧ · · · ∧ dxνq
µ1 <···<µp ν1 <···<νq
X X
= (−1)|σ| Aσ(ρ1 )...σ(ρp ) Bσ(ρp+1 )...σ(ρp+q ) dxρ1 ∧ · · · ∧ dxρp+q
ρ1 <···<ρp+q (p,q)-shuffles σ
1 X
= (−1)|σ| Aσ(ρ1 )...σ(ρp ) Bσ(ρp+1 )...σ(ρp+q ) dxρ1 ∧ · · · ∧ dxρp+q .
(p + q)!
(p,q)-shuffles σ

2
where Aµ1 ,...,µp Bν1 ,...,νq is the matrix product of Aµ1 ,...,µp and Bν1 ,...,νq . Component-wise, we
have X
(A ∧ B)ρ1 ...ρp+q = (−1)|σ| Aσ(ρ1 )...σ(ρp ) Bσ(ρp+1 )...σ(ρp+q ) .
(p,q)-shuffles σ

In particular, letting A be a matrix valued 1-form, we have


X
(A ∧ A)µν = (−1)|σ| Aσ(µ) Bσ(ν) = Aµ Aν − Aν Aµ = [Aµ , Aν ].
(1,1)-shuffles σ

Replacing the matrix product with the Lie product in the above definition, we obtain the wedge
product of Lie algebra valued forms, which is usually denoted as

[A ∧ B], or A[∧]B, or simply [A, B].

By definition,
[A, B] = A ∧ B − (−1)pq B ∧ A.
For matrix valued 1-forms, we have

[A, A] = 2A ∧ A.

Thus, the curvature F can also be expressed as


1
F = dA + [A, A].
2
Remark 2.3. Another way to derive the formula (2.8) is by noticing that
1
d2∇ = dxµ ∧ dxν ∇µ ∇ν = dxµ ∧ dxν [∇µ , ∇ν ].
2
Hence,
∂ ∂ ∂Aν ∂Aµ
Fµν = [∇µ , ∇ν ] = [ µ
+ Aµ , ν + Aν ] = µ
− + [Aµ , Aν ].
∂x ∂x ∂x ∂xν
Example 2.10. Let E = T M . Let ∇ = d + A be an affine connection of M . Under the
identification End(T M ) ∼
= T ∗ M ⊗ T M , we can write

A = Γρµν dxµ ⊗ dxν ⊗ .
∂xρ
If ∇ is the Levi-Civita connection of M , then Γρµν are the Christoffel symbols, and the cur-
vature 2-form F of ∇ is exactly the Riemannian curvature tensor of M .

Example 2.11. The curvature F of the covariant derivative on O(−1) defined by (2.6) is

X
n
dck ∧ dck X n
ck cl dcl ∧ dck
F = dA + A ∧ A = ∂A = − .
k=1
1 + |c|2 k,l=1
(1 + |c|2 )2

3
In particular, for n = 1, we have only one coordinate function which we denote by c.
1 2i
F = dc ∧ dc = rdr ∧ dϕ.
(1 + |c| )
2 2 (1 + r2 )2
Since dimR CP1 = 2, we can integrate the 2-form 2πi F
over CP1 . The result is:
Z Z Z 2π Z ∞ ∞
F 1 ∞ 1 1 1
= dr dϕ r = d(r 2
) = − = 1.
CP1 2πi π 0 0 (1 + r2 )2 0 (1 + r2 )2 1 + r2 0

A connection with vanishing curvature is called a flat connection.


Exercise 2.13. Show that the connection defined in Example 2.5 is a flat connection.
Hint 2.10. Use Exercise 2.7.

4 Bochner Laplacians
Let (E, ⟨·, ·⟩) be a metric vector bundle over a compact Riemannian manifold (M, g). Let ∇
be a metric connection on E. Let ∇∗ denote the formal adjoint of ∇, i.e., ∇∗ is a R-linear
map ∇∗ : Γ(T ∗ M ⊗ E) → Γ(E) such that
Z Z
⟨s, ∇(t)⟩ volg = ⟨∇∗ (s), t⟩ volg
M M

for all s ∈ Ω1 (E) and t ∈ Ω0 (E).


Definition 2.9. The Bochner Laplacian on E is defines as
∆B = ∇∗ ∇.

Let’s consider the vector bundle ΛT ∗ M and the Levi-Civita connection ∇ on ΛT ∗ M . Let
∇∗ be the formal adjoint of ∇ with respect to the bundle metric on ΛT ∗ M induced by the
Riemannian metric g on M .
Proposition 2.7. Let ω be a differential k-form on M . We have
(∆B − ∆H )(ω) = Ric(ω), (2.9)
where ∆H = dd∗ + d∗ d is the Hodge Laplacian of M . Ric ∈ End(Λk T ∗ M ) is the so-called
Weitzenböck operator, which is defined by
X
n X
k
Ric(ω)(X1 , · · · , Xk ) = (R(ei , Xj )ω)(X1 , · · · , ei , · · · , Xk ), (2.10)
i=1 j=1

where R is the Riemannian curvature tensor and {ei }ni=1 is a local orthonormal frame.1

P·)n of the vector field X, we have Ric(X )(Y ) =


1 ♯ ♯
Pn Applying (2.10) to the
Pn metric dual X = g(X,
i=1 g(R(ei , Y )X, ei ) = − i=1 g(R(ei , Y )ei , X) = Ric(X, Y ), where Ric(·, ·) is

i=1 (R(ei , Y )X )(ei ) =
the Ricci curvature tensor of g, hence the notation.

4
More generally, let’s consider the tensor product bundle ΛT ∗ M ⊗ E, where E is a metric
vector bundle bundle, equipped with the connection

∇ = ∇LC ⊗ 1 + 1 ⊗ ∇E ,

where ∇LC is the Levi-Civita connection of M and ∇E is a metric connection on E. ΛT ∗ M ⊗


E has an induced bundle metric, with respect to which ∇ is a metric connection.

Proposition 2.8. Let ω be a E-valued k-form on M . We have

(∆B − ∆H )(ω) = Ric(ω) + FE (ω), (2.11)

where ∆H := d∇E d∗∇E + d∗∇E d∇E and FE ∈ Ωk (End(E)) is defined by

n X
X k
FE (ω)(X1 , · · · , Xk ) = FE (ei , Xj )(ω(X1 , · · · , ei , · · · , Xk )),
i=1 j=1

where FE is the curvature of ∇E .

The proofs are postponed to the next section.

You might also like