Scribd
Upload
26 views2 pages

KNILL Lecture19

The document discusses the curvature tensor in differential geometry, detailing the Riemann curvature tensor and its properties, including the sectional curvature and Ricci curvature. It also introduces pseudo Riemannian manifolds and the conditions under which a manifold can support a metric of a specific signature. Theorems related to the curvature and metrics are presented, emphasizing their significance in understanding geometric structures and their applications in relativity.

Uploaded by

roger.chemoul86
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
26 views2 pages

KNILL Lecture19

The document discusses the curvature tensor in differential geometry, detailing the Riemann curvature tensor and its properties, including the sectional curvature and Ricci curvature. It also introduces pseudo Riemannian manifolds and the conditions under which a manifold can support a metric of a specific signature. Theorems related to the curvature and metrics are presented, emphasizing their significance in understanding geometric structures and their applications in relativity.

Uploaded by

roger.chemoul86
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/ 2

DIFFERENTIAL GEOMETRY

MATH 136

Unit 19: Curvature Tensor

19.1. While Γijk = 12 [ ∂u∂ i gjk + ∂u∂ j gki − ∂u∂ k gij ] is not a tensor, the Riemann curvature
s ∂ s ∂ s X
r s
X
Rikj = Γ − Γ + Γ Γ − Γrik Γsrj
∂uk ij ∂uj ik r
ij rk
r

is a (1, 3) tensor. Think of it as a matrix Ris describing a linear transformation when


rotating around a small square in the k, j plane. We can also look at at the (0, 4)-tensor
s
P
Rmikj = s gms Rijk .
intuition, let us rewrite this without coordinates. Let X = i X i ei
P
19.2. To gainP more
and Y = i
i Y ei be vector fields (1,0)
P tensor fields. 1 Reflected in the notation
i
ei = ∂P ui is that a vector field X = i X ei also defines a linear map on functions
Xf = i x fui = df X, the directional derivative. 2 Since every vector field X also
i

is a linear map, one can look at the commutator [X, Y P ] = XY − Y X,Pwhich is by


Leibniz again a vector field. Proof: in coordinates X = j X j ej , Y j = j Y j ej , this
Lie bracket is [X, Y ]i = j X j ∂j Y i − Y j ∂j X i .
P

19.3. The covariant derivative ∇X Y is a new vector field. Axiomatically it is deter-


mined by Leibniz ∇X (f Y ) = (Xf )Y + f ∇X Y , metric compatibility g(∇X Y, Z) +
g(Y, ∇X Z) = Xg(Y, Z), and being torsion free ∇X Y − ∇Y X = [X, Y ]. The fun-
damental theorem of Riemannian P k geometry assures that there exactly one such
derivative: and this is ∇ei ej = k Γij ek determining the Riemann curvature tensor R.
19.4. The curvature tensor now also can be written P as rR(X, Y )Z = ∇X ∇Y Z −
∇Y ∇XZ − ∇[X,Y ] Z meaning g(es , R(ek , ej )ei ) = r gsr Rikj er = Rsikj . Intuitively,
R(ek , ej ) tells what happens if one parallel transports along a small rectangular loop
spanned by ek , ej . A linear transformation Asi results from looping in the k, j plane.
19.5. For linearly independent vectors u, v, the sectional curvature
g(R(u, v)v, u)
K(u, v) =
g(u, u)g(v, v) − g(u, v)2
is probably the most intuitive approach to curvature as it does not depend on the
coordinate system. It only depends on the plane defined by the tangent vectors u, v.
If M is two dimensional, it agrees with the Gauss curvature. But this is not obvious
as it reestablishes the Theorema egregium!
1Most of the literature uses capital letters for vector fields.
P i P i
i X ei rather than i v ei .
2The directional derivative is written as ∇f · v in multi-variable calculus.
Differential Geometry

Theorem 1. The sectional curvatures determine the Riemann curvature tensor.


P j
19.6. The Ricci curvature R is a contraction of the curvature tensor Rik = j Rijk .
The scalar curvature is then the contraction of the Ricci curvature S = j,k g jk Rjk .
P

In two dimensions, it is twice the Gauss curvature. The Einstein tensor G is defined
as R − Sg/2 . A metric is called an Einstein metric if R = λg for some λ. Define
R
the Hilbert functional
R P S(g) = S dVg and the inner product on (0, 2) tensors as
M g
⟨a, b⟩g = M i,j a(ei , ej )b(ei , ej ) dV . Under which conditions is Hilbert functional
extremal? 3
d
Theorem 2. dt
S(g + th) = ⟨Sg/2 − R), h⟩g .
Theorem 3. Every 2-manifold is an Einstein manifold: Sg/2 − R = 0.
R
Proof. The reason is that K = S/2 and that the Hilbert functional S(g) = 2 M KdV =
4πχ(M ) does not depend on the metric by the global Gauss-Bonnet theorem. □
We see that in the 2 dimensional case the Ricci tensor R is K times the Riemannian
metric tensor g. Again, this is not obvious as it re-establishes the Theorema egregium.
19.7. In order to prepare for relativity, we also need to generalize Riemannian man-
ifolds. A metric tensor on a linear space E is a symmetric (0, 2) tensor which is
non-degenerate that is g(u, v) = 0, ∀v ∈ E ⇒ u = 0. A metric tensor field
g is a tensor field g ∈ T20 (M ) such that g(x) is a metric tensor in T20 (Tx M ). This
means that for any vector fields X, Y the function x → g(x)(X(x), Y (x)) is smooth.
A pseudo Riemannian manifold is a smooth manifold with a metric tensor field
g on M . A pseudo Riemannian manifold (M, g) is a Riemannian manifold, if g is
positive definite, meaning
p g(x)(v, v) ≥ 0 for all v. The length of a vector v ∈ Tp M is
defined as ||v|| = |g(p)(u, u)|, where g(p)(u, v) = ij gij (p)ui v j . 4 A vector of length
P

zero is called null. Vectors u for which ij gij ui uj < 0 are time like, vectors u with
P
P i j
Rb
ij gij u u > 0 space like. The length of the curve is defined by a ||ẋ(t)|| dt.

19.8. Does every manifold allow a pseudo Riemannian manifold of a certain signature?
Theorem 4. On any Riemannian manifold there exists a Riemannian metric g.
Proof. There is a tensor field g ∈ T20 (M ) which is symmetric, non-degenerate and
positive definite: let {Ui , ϕi } be an atlas for M and let pi be a partition of unity,
subordinate to the cover Ui . Let q be a Riemannian metric on Rn . For example
∗
[q]
P = Diag(1, 1, 1 . . . , 1). Let qi = ϕi q be the pull back metrics on Ui . Define g(p) =
P definite because for p ∈ M and u in the
i gi (p)qi (p). This is smooth and positive
tangent space Tp M , we have g(p)(u, u) = i gi qi (u, u) > 0. □
19.9. It is not always possible to build on a given manifold a metric of a given signature.
For example, on the sphere M = S 2 , there exists no Lorentzian metric, that is a
metric of signature (−1, 1). The reason is that one can not comb a 2-sphere.
Oliver Knill, [email protected], Math 136, Fall, 2024
3A proof can be found on pages 312-320 in Kuehnel.
4Note the appearance of an absolute value.

You might also like