Scribd
Upload
8 views

lecture17

The document discusses Riemannian manifolds, defining them as m-dimensional C k-differentiable manifolds with a positive definite symmetric (2, 0) tensor field. It explains concepts such as charts, atlases, differentiable structures, and critical points in the context of smooth maps. Additionally, it provides examples including spheres and special linear groups as manifolds.

Uploaded by

roger.chemoul86
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
8 views

lecture17

The document discusses Riemannian manifolds, defining them as m-dimensional C k-differentiable manifolds with a positive definite symmetric (2, 0) tensor field. It explains concepts such as charts, atlases, differentiable structures, and critical points in the context of smooth maps. Additionally, it provides examples including spheres and special linear groups as manifolds.

Uploaded by

roger.chemoul86
Copyright
© © All Rights Reserved
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 2

DIFFERENTIAL GEOMETRY

MATH 136

Unit 17: Riemannian Manifolds

17.1. A locally Euclidean space M of dimension m is a subset of some Rn such


that every x ∈ M has a neighborhood U , that is homeomorphic to an open subset
R = ϕ(U ) of Rm . The pair (U, ϕ) is called a chart producing a coordinate system
on U : there is a parametrization r(ϕ(x)) = x, which is a regular map from R ⊂ Rm
to Rn , meaning that dr has rank m everywhere. A C k atlas S on a locally Euclidean
space M is a collection F = {Ui , ϕi }i∈I of charts such that i∈I Ui = M , and that all
ϕij = ϕi ◦ ϕ−1
j are in C k (ϕi (Uj ∩ Ui ), Rn ). An atlas is called maximal, if (U, ϕ) is a
chart such that ϕ ◦ ϕ−1
i and ϕi ◦ ϕ−1 are C k for all i ∈ I, then (U, ϕ) ∈ F. Two atlases
F, G are called equivalent if their union F ∪ G is an atlas. Given an atlas F, the
union of all atlases equivalent to F is called a differentiable structure generated
by A. 1

Figure 1. A m-manifold M ⊂ Rn is shown with part of an atlas


F. Each patch Ui is a regularly parametrized by r : Ri → Ui with
Ri = ϕi (Ui ) ⊂ Rm . The map r has maximal rank m everywhere on Ri .

17.2. A m-dimensional C k -differentiable manifold is a pair (M, F), where M is a


m-dimensional locally Euclidean space and where F is a differentiable C k structure on
M . What this means is that near every point we are in a parametrized region r(U )
1The concept can be difficult: F is not unique in general. On S 7 , there are 28 different smooth
structures. The smooth Poincare conjecture claims that S 4 has a unique differentiable structure.
Differential Geometry

and so can use what we have done in this course like forming ru , rv , define fundamental
forms etc. Instead of C k , we usually just say smooth. 2
17.3. If E = Rm is the space of column vectors of dimension m, its dual E ∗ is
defined as the space of all linear maps f : E → R, It is the space of row vectors. If
{e1 , . . . , em } is a basis of E,
P then {e1 , . . . , em } denotes a basis of E ∗ . Every element
P in
E can be written as v = i v ei , every element in E ∗ can be written as v = i vi ei .
i

For p, q ≥ 0, the linear space Tqp of all multi-linear maps (E ∗ )p × E q → R is called


the space of tensors of type (p, q). Column vectors are (1, 0)-tensors in T01 = E,
while row vectors are (0, 1)-tensors in T10 = E ∗ , bilinear maps are (0, 2) tensors in
T20 . A tensor field of type (p, q) on a m-manifold M is a smooth assignment of a
(p, q) tensor to every point. Such a map is also called a section of the tensor bundle,
generalizing that a vector field is a section of the tangent bundle T M . For a (0, 2)
tensor field g for example smooth means that for any vector fields X, Y , the function
x → g(x)(X(x), Y (x)) is smooth. If f : M → Rk is a smooth map, then df is a
(0, 1) tensor field. This is also called a 1-form. A vector field means a (1, 0) tensor
field. The first fundamental form g is by definition a (2, 0) tensor field, a bilinear form
attached to every point. A Riemannian manifold (M, g) is a smooth manifold M
with a positive definite symmetric (2, 0) tensor field g.
17.4. Let M be a m-manifold and f : M → Rk be smooth. A point x ∈ M is called a
critical point and f (x) a critical value, if the rank of df (x) is not m. Non-critical
points are called regular points.
Theorem 1. If M is a m-manifold and f : M → Rk is smooth and y is a regular
value, then Mf = f −1 (y) is a manifold of dimension m − k.
Proof. If x ∈ f −1 (y) is given, the Jacobean map df (x) has rank k and the kernel
H = ker(df ) of df (x) is (m − k)-dimensional and H ⊥ is k dimensional. Take a chart
(U, ϕ) in M which contains x. Define
g = f ◦ ϕ−1 : ϕ(U ) ⊂ Rm → Rk .
The projection L : Rm = H ⊕ H ⊥ → Rm−k , (h, h′ ) 7→ h′ onto the orthogonal comple-
ment which is non-singular on H. The map F : ϕ(U ) → Rn ×Rm−k as F (z) = (g(z), Lz)
has derivative dF (u) = (dg(u), Lu) and which is nonsingular. By the inverse func-
tion theorem, a neighborhood ϕ(U ) of ϕ(x) is mapped by the diffeomorphism F onto
a neighborhood F (ϕ(U )) of F (ϕ(x)). We get so a chart Ux = ϕ−1 ◦ F −1 (F (ϕ(U )) ∩
{ψ(y)} × Rm−k ) on f −1 (y) which is mapped by ϕx = F ◦ ϕ into a (m − k)-dimensional
space. Doing the same construction at any point x ∈ M produces an atlas for f −1 (y)
and verifies that f −1 (y) is a manifold. □
17.5. Examples: a) The d − sphere is the set M = S d = {x ∈ Rd+1 | x21 + x22 +
· · · + x2d+1 = 1} ⊂ Rd+1 . Take two points A = (0, . . . , 0, 1) and B = (0, . . . , 0, −1). The
standard differentiable structure F on S d is generated by F = {(S d \{A}, ϕA ), (S d \
{B}, ϕB )}, where ϕA are the stereographic projections from A to {xd+1 = 0}. b)
The set SL(n, R) of n × n matrices of determinant 1 is a manifold.
Oliver Knill, [email protected], Math 136, Fall, 2024
2A theorem of Whitney assures that any smooth compact m-manifold M (defined more abstractly
using fancy-schmancy paracompact Hausdorff spaces) is part of Rn with n = 2m + 1.

You might also like