This action might not be possible to undo. Are you sure you want to continue?

**An Introduction to Riemannian Geometry
**

Sigmundur Gudmundsson

(Lund University)

(version 1.270 - 4 June 2010)

The latest version of this document can be found at

http://www.matematik.lu.se/matematiklu/personal/sigma/

1

Preface

These lecture notes grew out of an M.Sc. course on diﬀerential

geometry which I gave at the University of Leeds 1992. Their main

purpose is to introduce the beautiful theory of Riemannian Geometry

a still very active area of mathematical research. This is a subject with

no lack of interesting examples. They are indeed the key to a good

understanding of it and will therefore play a major role throughout

this work. Of special interest are the classical Lie groups allowing

concrete calculations of many of the abstract notions on the menu.

The study of Riemannian geometry is rather meaningless without

some basic knowledge on Gaussian geometry that i.e. the geometry of

curves and surfaces in 3-dimensional space. For this I recommend the

excellent textbook: M. P. do Carmo, Diﬀerential geometry of curves

and surfaces, Prentice Hall (1976).

These lecture notes are written for students with a good under-

standing of linear algebra, real analysis of several variables, the classical

theory of ordinary diﬀerential equations and some topology. The most

important results stated in the text are also proved there. Other are

left to the reader as exercises, which follow at the end of each chapter.

This format is aimed at students willing to put hard work into the

course.

For further reading I recommend the very interesting textbook: M.

P. do Carmo, Riemannian Geometry, Birkh¨auser (1992).

I am very grateful to my many enthusiastic students who through-

out the years have contributed to the text by ﬁnding numerous typing

errors and giving many useful comments on the presentation.

Norra N¨obbel¨ov, 17 February 2008

Sigmundur Gudmundsson

Contents

Chapter 1. Introduction 5

Chapter 2. Diﬀerentiable Manifolds 7

Chapter 3. The Tangent Space 21

Chapter 4. The Tangent Bundle 35

Chapter 5. Riemannian Manifolds 47

Chapter 6. The Levi-Civita Connection 59

Chapter 7. Geodesics 69

Chapter 8. The Riemann Curvature Tensor 83

Chapter 9. Curvature and Local Geometry 95

3

CHAPTER 1

Introduction

On the 10th of June 1854 Georg Friedrich Bernhard Riemann (1826-

1866) gave his famous ”Habilitationsvortrag” in the Colloquium of the

Philosophical Faculty at G¨ottingen. His talk ”

¨

Uber die Hypothesen,

welche der Geometrie zu Grunde liegen” is often said to be the most

important in the history of diﬀerential geometry. Johann Carl Friedrich

Gauss (1777-1855), at the age of 76, was in the audience and is said to

have been very impressed by his former student.

Riemann’s revolutionary ideas generalized the geometry of surfaces

which had been studied earlier by Gauss, Bolyai and Lobachevsky.

Later this lead to an exact deﬁnition of the modern concept of an

abstract Riemannian manifold.

5

CHAPTER 2

Diﬀerentiable Manifolds

In this chapter we introduce the important notion of a diﬀeren-

tiable manifold. This generalizes curves and surfaces in R

3

studied in

classical diﬀerential geometry. Our manifolds are modelled on the stan-

dard diﬀerentiable structure on the vector spaces R

m

via compatible

local charts. We give many examples, study their submanifolds and

diﬀerentiable maps between manifolds.

For a natural number m let R

m

be the m-dimensional real vector

space equipped with the topology induced by the standard Euclidean

metric d on R

m

given by

d(x, y) =

_

(x

1

−y

1

)

2

+ . . . + (x

m

−y

m

)

2

.

For positive natural numbers n, r and an open subset U of R

m

we shall

by C

r

(U, R

n

) denote the r-times continuously diﬀerentiable maps

from U to R

n

. By smooth maps U → R

n

we mean the elements

of

C

∞

(U, R

n

) =

∞

r=1

C

r

(U, R

n

).

The set of real analytic maps from U to R

n

will be denoted by

C

ω

(U, R

n

). For the theory of real analytic maps we recommend the

book: S. G. Krantz and H. R. Parks, A Primer of Real Analytic Func-

tions, Birkh¨auser (1992).

Deﬁnition 2.1. Let (M, T ) be a topological Hausdorﬀ space with

a countable basis. Then M is said to be a topological manifold if

there exists a natural number m and for each point p ∈ M an open

neighbourhood U of p and a continuous map x : U → R

m

which is a

homeomorphism onto its image x(U) which is an open subset of R

m

.

The pair (U, x) is called a (local) chart (or local coordinates) on M.

The natural number m is called the dimension of M. To denote that

the dimension of M is m we write M

m

.

According to Deﬁnition 2.1 a topological manifold M is locally

homeomorphic to the standard R

m

for some natural number m. We

7

8 2. DIFFERENTIABLE MANIFOLDS

shall now deﬁne a diﬀerentiable structure on M via local charts and

turn M into a diﬀerentiable manifold.

Deﬁnition 2.2. Let M be a topological manifold. Then a C

r

-atlas

on M is a collection

/ = ¦(U

α

, x

α

)[ α ∈ I¦

of local charts on M such that / covers the whole of M i.e.

M =

_

α

U

α

and for all α, β ∈ I the corresponding transition maps

x

β

◦ x

−1

α

[

x

α

(U

α

∩U

β

)

: x

α

(U

α

∩ U

β

) →R

m

are r-times continuously diﬀerentiable.

A chart (U, x) on M is said to be compatible with a C

r

-atlas / on

M if the union /∪¦(U, x)¦ is a C

r

-atlas on M. A C

r

-atlas

ˆ

/ is said to

be maximal if it contains all the charts that are compatible with it. A

maximal atlas

ˆ

/ on M is also called a C

r

-structure on M. The pair

(M,

ˆ

/) is said to be a C

r

-manifold, or a diﬀerentiable manifold of

class C

r

, if M is a topological manifold and

ˆ

/ is a C

r

-structure on M.

A diﬀerentiable manifold is said to be smooth if its transition maps

are C

∞

and real analytic if they are C

ω

.

It should be noted that a given C

r

-atlas /on a topological manifold

M determines a unique C

r

-structure

ˆ

/ on M containing /. It simply

consists of all charts compatible with /.

Example 2.3. For the standard topological space (R

m

, T ) we have

the trivial C

ω

-atlas

/ = ¦(R

m

, x)[ x : p →p¦

inducing the standard C

ω

-structure

ˆ

/ on R

m

.

Example 2.4. Let S

m

denote the unit sphere in R

m+1

i.e.

S

m

= ¦p ∈ R

m+1

[ p

2

1

+ + p

2

m+1

= 1¦

equipped with the subset topology induced by the standard T on R

m+1

.

Let N be the north pole N = (1, 0) ∈ RR

m

and S be the south pole

S = (−1, 0) on S

m

, respectively. Put U

N

= S

m

¸ ¦N¦, U

S

= S

m

¸ ¦S¦

and deﬁne x

N

: U

N

→R

m

, x

S

: U

S

→R

m

by

x

N

: (p

1

, . . . , p

m+1

) →

1

1 −p

1

(p

2

, . . . , p

m+1

),

x

S

: (p

1

, . . . , p

m+1

) →

1

1 + p

1

(p

2

, . . . , p

m+1

).

2. DIFFERENTIABLE MANIFOLDS 9

Then the transition maps

x

S

◦ x

−1

N

, x

N

◦ x

−1

S

: R

m

¸ ¦0¦ →R

m

¸ ¦0¦

are given by

p →

p

[p[

2

so / = ¦(U

N

, x

N

), (U

S

, x

S

)¦ is a C

ω

-atlas on S

m

. The C

ω

-manifold

(S

m

,

ˆ

/) is called the standard m-dimensional sphere.

Another interesting example of a diﬀerentiable manifold is the m-

dimensional real projective space RP

m

.

Example 2.5. On the set R

m+1

¸ ¦0¦ we deﬁne the equivalence

relation ≡ by:

p ≡ q if and only if there exists a λ ∈ R

∗

such that p = λq.

Let RP

m

be the quotient space (R

m+1

¸ ¦0¦)/ ≡ and

π : R

m+1

¸ ¦0¦ →RP

m

be the natural projection mapping a point p ∈ R

m+1

¸ ¦0¦ to the

equivalence class [p] ∈ RP

m

i.e. the line

[p] = ¦λp ∈ R

m+1

[ λ ∈ R

∗

¦

through the origin generated by p. Equip RP

m

with the quotient topol-

ogy induced by π and T on R

m+1

. For k ∈ ¦1, . . . , m + 1¦ deﬁne the

open subset

U

k

= ¦[p] ∈ RP

m

[ p

k

,= 0¦

of RP

m

and the charts x

k

: U

k

→R

m

by

x

k

: [p] →(

p

1

p

k

, . . . ,

p

k−1

p

k

, 1,

p

k+1

p

k

, . . . ,

p

m+1

p

k

).

If [p] ≡ [q] then p = λq for some λ ∈ R

∗

so p

l

/p

k

= q

l

/q

k

for all l. This

means that the map x

k

is well deﬁned for all k. The corresponding

transition maps

x

k

◦ x

−1

l

[

x

l

(U

l

∩U

k

)

: x

l

(U

l

∩ U

k

) →R

m

are given by

(

p

1

p

l

, . . . ,

p

l−1

p

l

, 1,

p

l+1

p

l

, . . . ,

p

m+1

p

l

) →(

p

1

p

k

, . . . ,

p

k−1

p

k

, 1,

p

k+1

p

k

, . . . ,

p

m+1

p

k

)

so the collection

/ = ¦(U

k

, x

k

)[ k = 1, . . . , m + 1¦

is a C

ω

-atlas on RP

m

. The diﬀerentiable manifold (RP

m

,

ˆ

/) is called

the m-dimensional real projective space.

10 2. DIFFERENTIABLE MANIFOLDS

Example 2.6. Let

ˆ

C be the extended complex plane given by

ˆ

C = C ∪ ¦∞¦

and put C

∗

= C ¸ ¦0¦, U

0

= C and U

∞

=

ˆ

C ¸ ¦0¦. Then deﬁne the

local coordinates x

0

: U

0

→ C and x

∞

: U

∞

→ C on

ˆ

C by x

0

: z → z

and x

∞

: w →1/w, respectively. The corresponding transition maps

x

∞

◦ x

−1

0

, x

0

◦ x

−1

∞

: C

∗

→C

∗

are both given by z →1/z so / = ¦(U

0

, x

0

), (U

∞

, x

∞

)¦ is a C

ω

-atlas on

ˆ

C. The real analytic manifold (

ˆ

C,

ˆ

/) is called the Riemann sphere.

For the product of two diﬀerentiable manifolds we have the following

interesting result.

Proposition 2.7. Let (M

1

,

ˆ

/

1

) and (M

2

,

ˆ

/

2

) be two diﬀerentiable

manifolds of class C

r

. Let M = M

1

M

2

be the product space with the

product topology. Then there exists an atlas / on M making (M,

ˆ

/)

into a diﬀerentiable manifold of class C

r

and the dimension of M sat-

isﬁes

dimM = dimM

1

+ dimM

2

.

Proof. See Exercise 2.1.

The concept of a submanifold of a given diﬀerentiable manifold

will play an important role as we go along and we shall be especially

interested in the connection between the geometry of a submanifold

and that of its ambient space.

Deﬁnition 2.8. Let m, n ∈ N be natural numbers such that 1 ≤

m ≤ n and (N

n

,

ˆ

B) be a C

r

-manifold. A subset M of N is said to

be a submanifold of N if for each point p ∈ M there exists a chart

(U

p

, x

p

) ∈

ˆ

B such that p ∈ U

p

and x

p

: U

p

→R

m

R

n−m

satisﬁes

x

p

(U

p

∩ M) = x

p

(U

p

) ∩ (R

m

¦0¦).

The natural number (n −m) is called the codimension of M in N.

Proposition 2.9. Let m, n ∈ N be natural numbers such that 1 ≤

m ≤ n and (N

n

,

ˆ

B) be a C

r

-manifold. Let M be a submanifold of N

equipped with the subset topology and π : R

m

R

n−m

→ R

m

be the

natural projection onto the ﬁrst factor. Then

/ = ¦(U

p

∩ M, (π ◦ x

p

)[

U

p

∩M

)[ p ∈ M¦

is a C

r

-atlas for M. In particular, the pair (M,

ˆ

/) is an m-dimensional

C

r

-manifold. The diﬀerentiable structure

ˆ

/ on the submanifold M of

N is called the induced structure of

ˆ

B.

2. DIFFERENTIABLE MANIFOLDS 11

Proof. See Exercise 2.2.

Our next step is to prove the implicit function theorem which is a

useful tool for constructing submanifolds of R

n

. For this we use the

classical inverse function theorem stated below. Note that if

F : U →R

m

is a diﬀerentiable map deﬁned on an open subset U of R

n

then its

diﬀerential dF

p

: R

n

→ R

m

at a point p ∈ U is a linear map given by

the mn matrix

dF

p

=

_

_

∂F

1

/∂x

1

(p) . . . ∂F

1

/∂x

n

(p)

.

.

.

.

.

.

∂F

m

/∂x

1

(p) . . . ∂F

m

/∂x

n

(p)

_

_

.

If γ : R → U is a curve in U such that γ(0) = p and ˙ γ(0) = v ∈ R

n

then the composition F ◦ γ : R → R

m

is a curve in R

m

and according

to the chain rule we have

dF

p

v =

d

ds

(F ◦ γ(s))[

s=0

,

which is the tangent vector of the curve F ◦ γ at F(p) ∈ R

m

.

Hence the diﬀerential dF

p

can be seen as a linear map that

maps tangent vectors at p ∈ U to tangent vectors at the image

F(p) ∈ R

m

. This will later be generalized to the manifold

setting.

Fact 2.10 (The Inverse Function Theorem). Let U be an open sub-

set of R

n

and F : U →R

n

be a C

r

-map. If p ∈ U and the diﬀerential

dF

p

: R

n

→R

n

of F at p is invertible then there exist open neighbourhoods U

p

around p

and U

q

around q = F(p) such that

ˆ

F = F[

U

p

: U

p

→U

q

is bijective and

the inverse (

ˆ

F)

−1

: U

q

→ U

p

is a C

r

-map. The diﬀerential (d

ˆ

F

−1

)

q

of

ˆ

F

−1

at q satisﬁes

(d

ˆ

F

−1

)

q

= (dF

p

)

−1

i.e. it is the inverse of the diﬀerential dF

p

of F at p.

Before stating the implicit function theorem we remind the reader

of the deﬁnition of the following notions.

12 2. DIFFERENTIABLE MANIFOLDS

Deﬁnition 2.11. Let m, n be positive natural numbers, U be an

open subset of R

n

and F : U → R

m

be a C

r

-map. A point p ∈ U is

said to be critical for F if the diﬀerential

dF

p

: R

n

→R

m

is not of full rank, and regular if it is not critical. A point q ∈ F(U)

is said to be a regular value of F if every point of the pre-image

F

−1

(¦q¦) of q is regular and a critical value otherwise.

Note that if n ≥ m then p ∈ U is a regular point of

F = (F

1

, . . . , F

m

) : U →R

m

if and only if the gradients ∇F

1

, . . . , ∇F

m

of the coordinate functions

F

1

, . . . , F

m

: U → R are linearly independent at p, or equivalently, the

diﬀerential dF

p

of F at p satisﬁes the following condition

det(dF

p

(dF

p

)

t

) ,= 0.

Theorem 2.12 (The Implicit Function Theorem). Let m, n be nat-

ural numbers such that m < n and F : U → R

m

be a C

r

-map from

an open subset U of R

n

. If q ∈ F(U) is a regular value of F then the

pre-image F

−1

(¦q¦) of q is an (n −m)-dimensional submanifold of R

n

of class C

r

.

Proof. Let p be an element of F

−1

(¦q¦) and K

p

be the kernel of

the diﬀerential dF

p

i.e. the (n −m)-dimensional subspace of R

n

given

by K

p

= ¦v ∈ R

n

[ dF

p

v = 0¦. Let π

p

: R

n

→ R

n−m

be a linear map

such that π

p

[

K

p

: K

p

→ R

n−m

is bijective, π

p

[

K

⊥

p

= 0 and deﬁne the

map G

p

: U →R

m

R

n−m

by

G

p

: x →(F(x), π

p

(x)).

Then the diﬀerential (dG

p

)

p

: R

n

→ R

n

of G

p

, with respect to the

decompositions R

n

= K

⊥

p

⊕K

p

and R

n

= R

m

⊕R

n−m

, is given by

(dG

p

)

p

=

_

dF

p

[

K

⊥

p

0

0 π

p

_

,

hence bijective. It now follows from the inverse function theorem that

there exist open neighbourhoods V

p

around p and W

p

around G

p

(p)

such that

ˆ

G

p

= G

p

[

V

p

: V

p

→W

p

is bijective, the inverse

ˆ

G

−1

p

: W

p

→V

p

is C

r

, d(

ˆ

G

−1

p

)

G

p

(p)

= (dG

p

)

−1

p

and d(

ˆ

G

−1

p

)

y

is bijective for all y ∈ W

p

.

Now put

˜

U

p

= F

−1

(¦q¦) ∩ V

p

then

˜

U

p

=

ˆ

G

−1

p

((¦q¦ R

n−m

) ∩ W

p

)

2. DIFFERENTIABLE MANIFOLDS 13

so if π : R

m

R

n−m

→R

n−m

is the natural projection onto the second

factor, then the map

˜ x

p

= π ◦ G

p

[

˜

U

p

:

˜

U

p

→(¦q¦ R

n−m

) ∩ W

p

→R

n−m

is a chart on the open neighbourhood

˜

U

p

of p. The point q ∈ F(U) is

a regular value so the set

B = ¦(

˜

U

p

, ˜ x

p

)[ p ∈ F

−1

(¦q¦)¦

is a C

r

-atlas for F

−1

(¦q¦).

Employing the implicit function theorem we yield the following in-

teresting examples of the m-dimensional sphere S

m

and its tangent

bundle TS

m

as diﬀerentiable submanifolds of R

m+1

and R

2m+2

, respec-

tively.

Example 2.13. Let F : R

m+1

→R be the C

ω

-map given by

F : (p

1

, . . . , p

m+1

) →

m+1

i=1

p

2

i

.

The diﬀerential dF

p

of F at p is given by dF

p

= 2p, so

dF

p

(dF

p

)

t

= 4[p[

2

∈ R.

This means that 1 ∈ R is a regular value of F so the ﬁbre

S

m

= ¦p ∈ R

m+1

[ [p[

2

= 1¦ = F

−1

(¦1¦)

of F is an m-dimensional submanifold of R

m+1

. It is the standard

m-dimensional sphere introduced in Example 2.4.

Example 2.14. Let F : R

m+1

R

m+1

→R

2

be the C

ω

-map deﬁned

by F : (p, v) → (([p[

2

− 1)/2, ¸p, v)). The diﬀerential dF

(p,v)

of F at

(p, v) satisﬁes

dF

(p,v)

=

_

p 0

v p

_

.

Hence

det(dF (dF)

t

) = [p[

2

([p[

2

+[v[

2

) = 1 +[v[

2

> 0

on F

−1

(¦0¦). This means that

F

−1

(¦0¦) = ¦(p, v) ∈ R

m+1

R

m+1

[ [p[

2

= 1 and ¸p, v) = 0¦

which we denote by TS

m

is a 2m-dimensional submanifold of R

2m+2

.

We shall later see that TS

m

is what is called the tangent bundle of the

m-dimensional sphere.

14 2. DIFFERENTIABLE MANIFOLDS

We shall now apply the implicit function theorem to construct the

important orthogonal group O(m) as a submanifold of the set of the

real vector space of mm matrices R

m×m

.

Example 2.15. Let Sym(R

m

) be the linear subspace of R

m×m

con-

sisting of all symmetric mm-matrices

Sym(R

m

) = ¦A ∈ R

m×m

[ A

t

= A¦.

Then it is easily seen that the dimension of Sym(R

m

) is m(m + 1)/2.

Let F : R

m×m

→Sym(R

m

) be the map deﬁned by

F : A →A

t

A.

If A : R →R

m×m

is a curve in R

m×m

then

d

ds

(F ◦ A(s)) =

˙

A(s)

t

A(s) + A(s)

t

˙

A(s),

so the diﬀerential dF

A

of F at A ∈ R

m×m

satisﬁes

dF

A

: X →X

t

A + A

t

X.

This means that for an arbitrary element A of

O(m) = F

−1

(¦e¦) = ¦A ∈ R

m×m

[ A

t

A = e¦

and Y ∈ Sym(R

m

) we have dF

A

(AY/2) = Y . Hence the diﬀerential

dF

A

is surjective, so the identity matrix e ∈ Sym(R

m

) is a regular value

of F. Following the implicit function theorem O(m) is a submanifold

of R

m×m

of dimension m(m− 1)/2. The set O(m) is the well known

orthogonal group.

The concept of a diﬀerentiable map U → R

n

, deﬁned on an open

subset of R

m

, can be generalized to mappings between manifolds. We

shall see that the most important properties of these objects in the

classical case are also valid in the manifold setting.

Deﬁnition 2.16. Let (M

m

,

ˆ

/

1

) and (N

n

,

ˆ

/

2

) be two C

r

-manifolds.

A map φ : M → N is said to be diﬀerentiable of class C

r

if for all

charts (U, x) ∈

ˆ

/

1

and (V, y) ∈

ˆ

/

2

the map

y ◦ φ ◦ x

−1

[

x(U∩φ

−1

(V ))

: x(U ∩ φ

−1

(V )) ⊂ R

m

→R

n

is of class C

r

. A diﬀerentiable map γ : I → M deﬁned on an open

interval of R is called a diﬀerentiable curve in M. A diﬀerentiable

map f : M →R with values in R is called a diﬀerentiable function on

M. The set of smooth functions deﬁned on M is denoted by C

∞

(M).

It is an easy exercise, using Deﬁnition 2.16, to prove the follow-

ing result concerning the composition of diﬀerentiable maps between

manifolds.

2. DIFFERENTIABLE MANIFOLDS 15

Proposition 2.17. Let (M

1

,

ˆ

/

1

), (M

2

,

ˆ

/

2

), (M

3

,

ˆ

/

3

) be C

r

-mani-

folds and φ : (M

1

,

ˆ

/

1

) →(M

2

,

ˆ

/

2

), ψ : (M

2

,

ˆ

/

2

) →(M

3

,

ˆ

/

3

) be diﬀer-

entiable maps of class C

r

. Then the composition ψ ◦ φ : (M

1

,

ˆ

/

1

) →

(M

3

,

ˆ

/

3

) is a diﬀerentiable map of class C

r

.

Proof. See Exercise 2.5.

Deﬁnition 2.18. Two manifolds (M

1

,

ˆ

/

1

) and (M

2

,

ˆ

/

2

) of class

C

r

are said to be diﬀeomorphic if there exists a bijective C

r

-map

φ : M

1

→ M

2

, such that the inverse φ

−1

: M

2

→ M

1

is of class C

r

. In

that case the map φ is said to be a diﬀeomorphism between (M

1

,

ˆ

/

1

)

and (M

2

,

ˆ

/

2

).

It can be shown that the 2-dimensional sphere S

2

in R

3

and the

Riemann sphere, introduced earlier, are diﬀeomorphic, see Exercise

2.7.

Deﬁnition 2.19. Two C

r

-structures

ˆ

/

1

and

ˆ

/

2

on the same topo-

logical manifold M are said to be diﬀerent if the identity map id

M

:

(M,

ˆ

/

1

) →(M,

ˆ

/

2

) is not a diﬀeomorphism.

It can be seen that even the real line R carries diﬀerent diﬀerentiable

structures, see Exercise 2.6.

Deep Result 2.20. Let (M

m

1

,

ˆ

/

1

), (M

m

2

,

ˆ

/

2

) be two diﬀerentiable

manifolds of class C

r

and of equal dimensions. If M

1

and M

2

are home-

omorphic as topological spaces and m ≤ 3 then (M

1

,

ˆ

/

1

) and (M

2

,

ˆ

/

2

)

are diﬀeomorphic.

The following remarkable result was proved by John Milnor in his

famous paper: Diﬀerentiable structures on spheres, Amer. J. Math.

81 (1959), 962-972.

Deep Result 2.21. The 7-dimensional sphere S

7

has exactly 28

diﬀerent diﬀerentiable structures.

The next very useful proposition generalizes a classical result from

the real analysis of several variables.

Proposition 2.22. Let (N

1

,

ˆ

/

1

) and (N

2

,

ˆ

/

2

) be two diﬀerentiable

manifolds of class C

r

and M

1

, M

2

be submanifolds of N

1

and N

2

, re-

spectively. If φ : N

1

→ N

2

is a diﬀerentiable map of class C

r

such

that φ(M

1

) is contained in M

2

, then the restriction φ[

M

1

: M

1

→M

2

is

diﬀerentiable of class C

r

.

Proof. See Exercise 2.8.

16 2. DIFFERENTIABLE MANIFOLDS

Example 2.23. The result of Proposition 2.22 can be used to show

that the following maps are all smooth.

(i) φ

1

: S

2

⊂ R

3

→S

3

⊂ R

4

, φ

1

: (x, y, z) →(x, y, z, 0),

(ii) φ

2

: S

3

⊂ C

2

→S

2

⊂ CR, φ

2

: (z

1

, z

2

) →(2z

1

¯ z

2

, [z

1

[

2

−[z

2

[

2

),

(iii) φ

3

: R

1

→S

1

⊂ C, φ

3

: t →e

it

,

(iv) φ

4

: R

m+1

¸ ¦0¦ →S

m

, φ

4

: x →x/[x[,

(v) φ

5

: R

m+1

¸ ¦0¦ →RP

m

, φ

5

: x →[x],

(vi) φ

6

: S

m

→RP

m

, φ

6

: x →[x].

In diﬀerential geometry we are especially interested in diﬀerentiable

manifolds carrying a group structure compatible with their diﬀeren-

tiable structure. Such manifolds are named after the famous math-

ematician Sophus Lie (1842-1899) and will play an important role

throughout this work.

Deﬁnition 2.24. A Lie group is a smooth manifold G with a

group structure such that the map ρ : GG →G with

ρ : (p, q) →p q

−1

is smooth. For an element p in G the left translation by p is the map

L

p

: G →G deﬁned by L

p

: q →p q.

Note that the standard diﬀerentiable R

m

equipped with the usual

addition + forms an abelian Lie group (R

m

, +) with ρ : R

m

R

m

→R

m

given by

ρ : (p, q) →p −q.

Corollary 2.25. Let G be a Lie group and p be an element of G.

Then the left translation L

p

: G →G is a smooth diﬀeomorphism.

Proof. See Exercise 2.10

Proposition 2.26. Let (G, ) be a Lie group and K be a submani-

fold of G which is a subgroup. Then (K, ) is a Lie group.

Proof. The statement is a direct consequence of Deﬁnition 2.24

and Proposition 2.22.

The set of non-zero complex numbers C

∗

together with the standard

multiplication forms a Lie group (C

∗

, ). The unit circle (S

1

, ) is an

interesting compact Lie subgroup of (C

∗

, ). Another subgroup is the

set of the non-zero real numbers (R

∗

, ) containing the positive real

numbers (R

+

, ) and the 0-dimensional sphere (S

0

, ) as subgroups.

Example 2.27. Let H be the set of quaternions deﬁned by

H = ¦z +wj[ z, w ∈ C¦

equipped with the addition +, multiplication and conjugation ¯

2. DIFFERENTIABLE MANIFOLDS 17

(i) (z + wj) = ¯ z −wj,

(ii) (z

1

+w

1

j) + (z

2

+ w

2

j) = (z

1

+ z

2

) + (w

1

+w

2

)j,

(iii) (z

1

+w

1

j) (z

2

+ w

2

j) = (z

1

z

2

−w

1

¯ w

2

) + (z

1

w

2

+w

1

¯ z

2

)j

extending the standard operations on R and C as subsets of H. Then

it is easily seen that the non-zero quaternions (H

∗

, ) form a Lie group.

On H we deﬁne a scalar product

HH →H, (p, q) →p ¯ q

and a real valued norm given by [p[

2

= p ¯ p. Then the 3-dimensional

unit sphere S

3

in H

∼

= R

4

with the restricted multiplication forms a

compact Lie subgroup (S

3

, ) of (H

∗

, ). They are both non-abelian.

We shall now introduce some of the classical real and complex

matrix Lie groups. As a reference on this topic we recommend the

wonderful book: A. W. Knapp, Lie Groups Beyond an Introduction,

Birkh¨auser (2002).

Example 2.28. Let Nil

3

be the subset of R

3×3

given by

Nil

3

= ¦

_

_

1 x z

0 1 y

0 0 1

_

_

∈ R

3×3

[ x, y, z ∈ R¦.

Then Nil

3

has a natural diﬀerentiable structure determined by the

global coordinate φ : Nil

3

→R

3

with

φ :

_

_

1 x z

0 1 y

0 0 1

_

_

→(x, y, z).

It is easily seen that if ∗ is the standard matrix multiplication, then

(Nil

3

, ∗) is a Lie group.

Example 2.29. Let Sol

3

be the subset of R

3×3

given by

Sol

3

= ¦

_

_

e

z

0 x

0 e

−z

y

0 0 1

_

_

∈ R

3×3

[ x, y, z ∈ R¦.

Then Sol

3

has a natural diﬀerentiable structure determined by the

global coordinate φ : Sol

3

→R

3

with

φ :

_

_

e

z

0 x

0 e

−z

y

0 0 1

_

_

→(x, y, z).

It is easily seen that if ∗ is the standard matrix multiplication, then

(Sol

3

, ∗) is a Lie group.

18 2. DIFFERENTIABLE MANIFOLDS

Example 2.30. The set of invertible real mm matrices

GL

m

(R) = ¦A ∈ R

m×m

[ det A ,= 0¦

equipped with the standard matrix multiplication has the structure of

a Lie group. It is called the real general linear group and its neutral

element e is the identity matrix. The subset GL

m

(R) of R

m×m

is open

so dimGL

m

(R) = m

2

.

As a subgroup of GL

m

(R) we have the real special linear group

SL

m

(R) given by

SL

m

(R) = ¦A ∈ R

m×m

[ det A = 1¦.

We will show in Example 3.13 that the dimension of the submanifold

SL

m

(R) of R

m×m

is m

2

−1.

Another subgroup of GL

m

(R) is the orthogonal group

O(m) = ¦A ∈ R

m×m

[ A

t

A = e¦.

As we have already seen in Example 2.15 the dimension of O(m) is

m(m−1)/2.

As a subgroup of O(m) and SL

m

(R) we have the special orthog-

onal group SO(m) which is deﬁned as

SO(m) = O(m) ∩ SL

m

(R).

It can be shown that O(m) is diﬀeomorphic to SO(m) O(1), see

Exercise 2.9. Note that O(1) = ¦±1¦ so O(m) can be seen as two

copies of SO(m). This means that

dimSO(m) = dimO(m) = m(m−1)/2.

Example 2.31. The set of invertible complex mm matrices

GL

m

(C) = ¦A ∈ C

m×m

[ det A ,= 0¦

equipped with the standard matrix multiplication has the structure of

a Lie group. It is called the complex general linear group and its

neutral element e is the identity matrix. The subset GL

m

(C) of C

m×m

is open so dim(GL

m

(C)) = 2m

2

.

As a subgroup of GL

m

(C) we have the complex special linear

group SL

m

(C) given by

SL

m

(C) = ¦A ∈ C

m×m

[ det A = 1¦.

The dimension of the submanifold SL

m

(C) of C

m×m

is 2(m

2

−1).

Another subgroup of GL

m

(C) is the unitary group U(m) given

by

U(m) = ¦A ∈ C

m×m

[

¯

A

t

A = e¦.

2. DIFFERENTIABLE MANIFOLDS 19

Calculations similar to those for the orthogonal group show that the

dimension of U(m) is m

2

.

As a subgroup of U(m) and SL

m

(C) we have the special unitary

group SU(m) which is deﬁned as

SU(m) = U(m) ∩ SL

m

(C).

It can be shown that U(1) is diﬀeomorphic to the circle S

1

and that

U(m) is diﬀeomorphic to SU(m) U(1), see Exercise 2.9. This means

that dimSU(m) = m

2

−1.

For the rest of this manuscript we shall assume, when not

stating otherwise, that our manifolds and maps are smooth

i.e. in the C

∞

-category.

20 2. DIFFERENTIABLE MANIFOLDS

Exercises

Exercise 2.1. Find a proof for Proposition 2.7.

Exercise 2.2. Find a proof for Proposition 2.9.

Exercise 2.3. Let S

1

be the unit circle in the complex plane C

given by S

1

= ¦z ∈ C[ [z[

2

= 1¦. Use the maps x : C ¸ ¦i¦ → C and

y : C ¸ ¦−i¦ →C with

x : z →

i + z

1 + iz

, y : z →

1 + iz

i + z

to show that S

1

is a 1-dimensional submanifold of C

∼

= R

2

.

Exercise 2.4. Use the implicit function theorem to show that the

m-dimensional torus

T

m

= ¦z ∈ C

m

[ [z

1

[ = = [z

m

[ = 1¦

is a diﬀerentiable submanifold of C

m

∼

= R

2m

.

Exercise 2.5. Find a proof of Proposition 2.17.

Exercise 2.6. Equip the real line R with the standard topology

and for each odd integer k ∈ Z

+

let

ˆ

/

k

be the C

ω

-structure deﬁned on

R by the atlas

/

k

= ¦(R, x

k

)[ x

k

: p →p

k

¦.

Prove that the diﬀerentiable structures

ˆ

/

k

are all diﬀerent but that the

diﬀerentiable manifolds (R,

ˆ

/

k

) are all diﬀeomorphic.

Exercise 2.7. Prove that the 2-dimensional sphere S

2

as a diﬀer-

entiable submanifold of the standard R

3

and the Riemann sphere

ˆ

C are

diﬀeomorphic.

Exercise 2.8. Find a proof of Proposition 2.22.

Exercise 2.9. Let the spheres S

1

, S

3

and the Lie groups SO(n),

O(n), SU(n), U(n) be equipped with their standard diﬀerentiable

structures introduced above. Use Proposition 2.22 to prove the fol-

lowing diﬀeomorphisms

S

1

∼

= SO(2), S

3

∼

= SU(2),

SO(n) O(1)

∼

= O(n), SU(n) U(1)

∼

= U(n).

Exercise 2.10. Find a proof of Corollary 2.25.

Exercise 2.11. Let (G, ∗) and (H, ) be two Lie groups. Prove that

the product manifold GH has the structure of a Lie group.

CHAPTER 3

The Tangent Space

In this chapter we introduce the notion of the tangent space T

p

M of

a diﬀerentiable manifold M at a point p ∈ M. This is a vector space of

the same dimension as M. We start by studying the standard R

m

and

show how a tangent vector v at a point p ∈ R

m

can be interpreted as

a ﬁrst order linear diﬀerential operator, annihilating constants, when

acting on real valued functions locally deﬁned around p.

Let R

m

be the m-dimensional real vector space with the standard

diﬀerentiable structure. If p is a point in R

m

and γ : I → R

m

is a

C

1

-curve such that γ(0) = p then the tangent vector

˙ γ(0) = lim

t→0

γ(t) −γ(0)

t

of γ at p is an element of R

m

. Conversely, for an arbitrary element v

of R

m

we can easily ﬁnd a curve γ : I → R

m

such that γ(0) = p and

˙ γ(0) = v. One example is given by

γ : t →p + t v.

This shows that the tangent space, i.e. the space of tangent vectors,

at the point p ∈ R

m

can be identiﬁed with R

m

.

We shall now describe how ﬁrst order diﬀerential operators annihi-

lating constants can be interpreted as tangent vectors. For a point p

in R

m

we denote by ε(p) the set of diﬀerentiable real-valued functions

deﬁned locally around p. Then it is well known from multi-variable

analysis that if v ∈ R

m

and f ∈ ε(p) then the directional derivative

∂

v

f of f at p in the direction of v is given by

∂

v

f = lim

t→0

f(p + tv) −f(p)

t

.

Furthermore the operator ∂ has the following properties:

∂

v

(λ f + µ g) = λ ∂

v

f + µ ∂

v

g,

∂

v

(f g) = ∂

v

f g(p) + f(p) ∂

v

g,

∂

(λ·v+µ·w)

f = λ ∂

v

f + µ ∂

w

f

for all λ, µ ∈ R, v, w ∈ R

m

and f, g ∈ ε(p).

21

22 3. THE TANGENT SPACE

Deﬁnition 3.1. For a point p in R

m

let T

p

R

m

be the set of ﬁrst

order linear diﬀerential operators at p annihilating constants i.e. the

set of mappings α : ε(p) →R such that

(i) α(λ f + µ g) = λ α(f) +µ α(g),

(ii) α(f g) = α(f) g(p) + f(p) α(g)

for all λ, µ ∈ R and f, g ∈ ε(p).

The set of diﬀential operators T

p

R

m

carries the structure of a real

vector space. This is given by the addition + and the multiplication

by real numbers satisfying

(α + β)(f) = α(f) + β(f),

(λ α)(f) = λ α(f)

for all α, β ∈ T

p

R

m

, f ∈ ε(p) and λ ∈ R.

The above mentioned properties of the operator ∂ show that we

have a well deﬁned linear map Φ : R

m

→T

p

R

m

given by

Φ : v →∂

v

.

Theorem 3.2. For a point p in R

m

the linear map Φ : R

m

→T

p

R

m

deﬁned by Φ : v →∂

v

is a vector space isomorphism.

Proof. Let v, w ∈ R

m

such that v ,= w. Choose an element u ∈

R

m

such that ¸u, v) , = ¸u, w) and deﬁne f : R

m

→ R by f(x) = ¸u, x).

Then ∂

v

f = ¸u, v) , = ¸u, w) = ∂

w

f so ∂

v

,= ∂

w

. This proves that the

map Φ is injective.

Let α be an arbitrary element of T

p

R

m

. For k = 1, . . . , m let ˆ x

k

:

R

m

→R be the map given by

ˆ x

k

: (x

1

, . . . , x

m

) →x

k

and put v

k

= α(ˆ x

k

). For the constant function 1 : (x

1

, . . . , x

m

) →1 we

have

α(1) = α(1 1) = 1 α(1) + 1 α(1) = 2 α(1),

so α(1) = 0. By the linearity of α it follows that α(c) = 0 for any

constant c ∈ R. Let f ∈ ε(p) and following Lemma 3.3 locally write

f(x) = f(p) +

m

k=1

(ˆ x

k

(x) −p

k

) ψ

k

(x),

where ψ

k

∈ ε(p) with

ψ

k

(p) =

∂f

∂x

k

(p).

3. THE TANGENT SPACE 23

We can now apply the diﬀerential operator α ∈ T

p

R

m

and yield

α(f) = α(f(p) +

m

k=1

(ˆ x

k

−p

k

) ψ

k

)

= α(f(p)) +

m

k=1

α(ˆ x

k

−p

k

) ψ

k

(p) +

m

k=1

(ˆ x

k

(p) −p

k

) α(ψ

k

)

=

m

k=1

v

k

∂f

∂x

k

(p)

= ∂

v

f,

where v = (v

1

, . . . , v

m

) ∈ R

m

. This means that Φ(v) = ∂

v

= α so the

map Φ : R

m

→ T

p

R

m

is surjective and hence a vector space isomor-

phism.

Lemma 3.3. Let p be a point in R

m

and f : U →R be a function

deﬁned on an open ball around p. Then for each k = 1, 2, . . . , m there

exist functions ψ

k

: U →R such that

f(x) = f(p) +

m

k=1

(x

k

−p

k

) ψ

k

(x) and ψ

k

(p) =

∂f

∂x

k

(p)

for all x ∈ U.

Proof. It follows from the fundamental theorem of calculus that

f(x) −f(p) =

_

1

0

∂

∂t

(f(p + t(x −p)))dt

=

m

k=1

(x

k

−p

k

)

_

1

0

∂f

∂x

k

(p + t(x −p))dt.

The statement then immediately follows by setting

ψ

k

(x) =

_

1

0

∂f

∂x

k

(p + t(x −p))dt.

Let p be a point in R

m

, v ∈ R

m

be a tangent vector at p and

f : U → R be a C

1

-function deﬁned on an open subset U of R

m

containing p. Let γ : I → U be a curve such that γ(0) = p and

˙ γ(0) = v. The identiﬁcation given by Theorem 3.2 tells us that v acts

on f by

v(f) =

d

dt

(f ◦ γ(t))[

t=0

= df

p

( ˙ γ(0)) = ¸gradf

p

, ˙ γ(0)) = ¸gradf

p

, v).

24 3. THE TANGENT SPACE

Note that the real number v(f) is independent of the choice of the

curve γ as long as γ(0) = p and ˙ γ(0) = v. As a direct consequence of

Theorem 3.2 we have the following useful result.

Corollary 3.4. Let p be a point in R

m

and ¦e

k

[ k = 1, . . . , m¦ be

a basis for R

m

. Then the set ¦∂

e

k

[ k = 1, . . . , m¦ is a basis for the

tangent space T

p

R

m

at p.

We shall now use the ideas presented above to generalize to the

manifold setting. Let M be a diﬀerentiable manifold and for a point

p ∈ M let ε(p) denote the set of diﬀerentiable functions deﬁned on an

open neighborhood of p.

Deﬁnition 3.5. Let M be a diﬀerentiable manifold and p be a

point on M. A tangent vector X

p

at p is a map X

p

: ε(p) →R such

that

(i) X

p

(λ f + µ g) = λ X

p

(f) + µ X

p

(g),

(ii) X

p

(f g) = X

p

(f) g(p) +f(p) X

p

(g)

for all λ, µ ∈ R and f, g ∈ ε(p). The set of tangent vectors at p is called

the tangent space at p and denoted by T

p

M.

The tangent space T

p

M of M at p has the structure of a real vector

space. The addition + and the multiplication by real numbers are

simply given by

(X

p

+Y

p

)(f) = X

p

(f) + Y

p

(f),

(λ X

p

)(f) = λ X

p

(f)

for all X

p

, Y

p

∈ T

p

M, f ∈ ε(p) and λ ∈ R.

Deﬁnition 3.6. Let φ : M → N be a diﬀerentiable map between

manifolds. Then the diﬀerential dφ

p

of φ at a point p in M is the

map dφ

p

: T

p

M →T

φ(p)

N such that for all X

p

∈ T

p

M and f ∈ ε(φ(p))

(dφ

p

(X

p

))(f) = X

p

(f ◦ φ).

We shall now give some motivations for the above deﬁnitions and

hopefully convince the reader that they are not only abstract nonsense.

Proposition 3.7. Let φ : M →N and ψ : N →

¯

N be diﬀerentiable

maps between manifolds, then for each p ∈ M we have

(i) the map dφ

p

: T

p

M →T

φ(p)

N is linear,

(ii) if id

M

: M →M is the identity map, then d(id

M

)

p

= id

T

p

M

,

(iii) d(ψ ◦ φ)

p

= dψ

φ(p)

◦ dφ

p

.

3. THE TANGENT SPACE 25

Proof. The only non-trivial statement is the relation (iii) which

is called the chain rule. If X

p

∈ T

p

M and f ∈ ε(ψ ◦ φ(p)), then

(dψ

φ(p)

◦ dφ

p

)(X

p

)(f) = (dψ

φ(p)

(dφ

p

(X

p

)))(f)

= (dφ

p

(X

p

))(f ◦ ψ)

= X

p

(f ◦ ψ ◦ φ)

= d(ψ ◦ φ)

p

(X

p

)(f).

**Corollary 3.8. Let φ : M → N be a diﬀeomorphism with inverse
**

ψ = φ

−1

: N → M. Then the diﬀerential dφ

p

: T

p

M → T

φ(p)

N at p is

bijective and (dφ

p

)

−1

= dψ

φ(p)

.

Proof. The statement is a direct consequence of the following re-

lations

dψ

φ(p)

◦ dφ

p

= d(ψ ◦ φ)

p

= d(id

M

)

p

= id

T

p

M

,

dφ

p

◦ dψ

φ(p)

= d(φ ◦ ψ)

φ(p)

= d(id

N

)

φ(p)

= id

T

φ(p)

N

.

**We are now ready to prove the following interesting theorem. This
**

is of course a direct generalization of the corresponding result in the

classical theory for surfaces in R

3

.

Theorem 3.9. Let M

m

be an m-dimensional diﬀerentable manifold

and p be a point in M. Then the tangent space T

p

M at p is an m-

dimensional real vector space.

Proof. Let (U, x) be a chart on M. Then the linear map dx

p

:

T

p

M → T

x(p)

R

m

is a vector space isomorphism. The statement now

follows from Theorem 3.2 and Corollary 3.8.

Let M be an m-dimensional manifold and (U, x) be a local chart

around p ∈ M. Then the diﬀerential dx

p

: T

p

M → T

x(p)

R

m

is a

bijective linear map so for a given element X

p

∈ T

p

M there exists

a tangent vector v in R

m

such that dx

p

(X

p

) = v. The image x(U)

is an open subset of R

m

containing x(p) so we can ﬁnd a curve c :

(−ǫ, ǫ) → x(U) with c(0) = x(p) and ˙ c(0) = v. Then the composition

γ = x

−1

◦ c : (−ǫ, ǫ) → U is a curve in M through p since γ(0) = p.

The element d(x

−1

)

x(p)

(v) of the tangent space T

p

M denoted by ˙ γ(0)

is called the tangent to the curve γ at p. It follows from the relation

˙ γ(0) = d(x

−1

)

x(p)

(v) = X

p

that the tangent space T

p

M can be thought of as the set of all tangents

to curves through the point p.

26 3. THE TANGENT SPACE

If f : U → R is a C

1

-function deﬁned locally on U then it follows

from Deﬁnition 3.6 that

X

p

(f) = (dx

p

(X

p

))(f ◦ x

−1

)

=

d

dt

(f ◦ x

−1

◦ c(t))[

t=0

=

d

dt

(f ◦ γ(t))[

t=0

It should be noted that the value X

p

(f) is independent of the choice

of the chart (U, x) around p and the curve c : I → x(U) as long as

γ(0) = p and ˙ γ(0) = X

p

. This leads to the following construction.

Proposition 3.10. Let M

m

be a diﬀerentiable manifold, (U, x) be

a local chart on M and ¦e

k

[ k = 1, . . . , m¦ be the canonical basis for

R

m

. For an arbitrary point p in U we deﬁne (

∂

∂x

k

)

p

in T

p

M by

_

∂

∂x

k

_

p

: f →

∂f

∂x

k

(p) = ∂

e

k

(f ◦ x

−1

)(x(p)).

Then the set

¦(

∂

∂x

k

)

p

[ k = 1, 2, . . . , m¦

is a basis for the tangent space T

p

M of M at p.

Proof. The local chart x : U →x(U) is a diﬀeomorphism and the

diﬀerential (dx

−1

)

x(p)

: T

x(p)

R

m

→ T

p

M of the inverse x

−1

: x(U) → U

satisﬁes

(dx

−1

)

x(p)

(∂

e

k

)(f) = ∂

e

k

(f ◦ x

−1

)(x(p))

=

_

∂

∂x

k

_

p

(f)

for all f ∈ ε(p). The statement is then a direct consequence of Corollary

3.4.

We shall now determine the tangent spaces of some of the explicit

diﬀerentiable manifolds introduced in Chapter 2. We start with the

sphere.

Example 3.11. Let γ : (−ǫ, ǫ) → S

m

be a curve into the m-

dimensional unit sphere in R

m+1

with γ(0) = p and ˙ γ(0) = v. The

curve satisﬁes

¸γ(t), γ(t)) = 1

and diﬀerentiation yields

¸ ˙ γ(t), γ(t)) +¸γ(t), ˙ γ(t)) = 0.

3. THE TANGENT SPACE 27

This means that ¸v, p) = 0 so every tangent vector v ∈ T

p

S

m

must be

orthogonal to p. On the other hand if v ,= 0 satisﬁes ¸v, p) = 0 then

γ : R →S

m

with

γ : t →cos(t[v[) p + sin(t[v[) v/[v[

is a curve into S

m

with γ(0) = p and ˙ γ(0) = v. This shows that the

tangent space T

p

S

m

is actually given by

T

p

S

m

= ¦v ∈ R

m+1

[ ¸p, v) = 0¦.

In order to determine the tangent spaces of the classical Lie groups

we need the diﬀerentiable exponential map Exp : C

m×m

→ C

m×m

for

matrices given by the following converging power series

Exp : X →

∞

k=0

X

k

k!

.

For this map we have the following well-known result.

Proposition 3.12. Let Exp : C

m×m

→ C

m×m

be the exponential

map for matrices. If X, Y ∈ C

m×m

, then

(i) det(Exp(X)) = exp(traceX),

(ii) Exp(

¯

X

t

) = Exp(X)

t

, and

(iii) Exp(X +Y ) = Exp(X)Exp(Y ) whenever XY = Y X.

Proof. See Exercise 3.2

The real general linear group GL

m

(R) is an open subset of R

m×m

so its tangent space T

p

GL

m

(R) at any point p is simply R

m×m

. The

tangent space T

e

SL

m

(R) of the special linear group SL

m

(R) at the

neutral element e can be determined as follows.

Example 3.13. If X is a matrix in R

m×m

with traceX = 0 then

deﬁne a curve A : R →R

m×m

by

A : s →Exp(sX).

Then A(0) = e,

˙

A(0) = X and

det(A(s)) = det(Exp(sX)) = exp(trace(sX)) = exp(0) = 1.

This shows that A is a curve into the special linear group SL

m

(R) and

that X is an element of the tangent space T

e

SL

m

(R) of SL

m

(R) at the

neutral element e. Hence the linear space

¦X ∈ R

m×m

[ traceX = 0¦

of dimension m

2

−1 is contained in the tangent space T

e

SL

m

(R).

28 3. THE TANGENT SPACE

The curve given by s → Exp(se) = exp(s)e is not contained in

SL

m

(R) so the dimension of T

e

SL

m

(R) is at most m

2

−1. This shows

that

T

e

SL

m

(R) = ¦X ∈ R

m×m

[ traceX = 0¦.

Example 3.14. Let A : (−ǫ, ǫ) → O(m) be a curve into the or-

thogonal group O(m) such that A(0) = e. Then A(s)

t

A(s) = e for all

s ∈ (−ǫ, ǫ) and diﬀerentiation yields

¦

˙

A(s)

t

A(s) + A(s)

t

˙

A(s)¦[

s=0

= 0

or equivalently

˙

A(0)

t

+

˙

A(0) = 0. This means that each tangent vector

of O(m) at e is a skew-symmetric matrix.

On the other hand, for an arbitrary real skew-symmetric matrix X

deﬁne the curve B : R →R

m×m

by B : s →Exp(sX). Then

B(s)

t

B(s) = Exp(sX)

t

Exp(sX)

= Exp(sX

t

)Exp(sX)

= Exp(s(X

t

+X))

= Exp(0)

= e.

This shows that B is a curve on the orthogonal group, B(0) = e and

˙

B(0) = X so X is an element of T

e

O(m). Hence

T

e

O(m) = ¦X ∈ R

m×m

[ X

t

+ X = 0¦.

The dimension of T

e

O(m) is therefore m(m − 1)/2. This conﬁrms

our calculations of the dimension of O(m) in Example 2.15 since we

know that dim(O(m)) = dim(T

e

O(m)). The orthogonal group O(m) is

diﬀeomorphic to SO(m) ¦±1¦ so dim(SO(m)) = dim(O(m)) hence

T

e

SO(m) = T

e

O(m) = ¦X ∈ R

m×m

[ X

t

+ X = 0¦.

We have proved the following result.

Proposition 3.15. Let e be the neutral element of the classical

real Lie groups GL

m

(R), SL

m

(R), O(m), SO(m). Then their tangent

spaces at e are given by

T

e

GL

m

(R) = R

m×m

T

e

SL

m

(R) = ¦X ∈ R

m×m

[ traceX = 0¦

T

e

O(m) = ¦X ∈ R

m×m

[ X

t

+ X = 0¦

T

e

SO(m) = T

e

O(m) ∩ T

e

SL

m

(R) = T

e

O(m)

3. THE TANGENT SPACE 29

For the classical complex Lie groups similar methods can be used

to prove the following.

Proposition 3.16. Let e be the neutral element of the classical

complex Lie groups GL

m

(C), SL

m

(C), U(m), and SU(m). Then their

tangent spaces at e are given by

T

e

GL

m

(C) = C

m×m

T

e

SL

m

(C) = ¦Z ∈ C

m×m

[ traceZ = 0¦

T

e

U(m) = ¦Z ∈ C

m×m

[

¯

Z

t

+ Z = 0¦

T

e

SU(m) = T

e

U(m) ∩ T

e

SL

m

(C).

Proof. See Exercise 3.4

The rest of this chapter is devoted to the introduction of special

types of diﬀerentiable maps, the so called immersions, embeddings and

submersions. If γ

M

: (−ǫ, ǫ) →M is a curve on M such that γ

M

(0) = p

then a diﬀerentiable map φ : M →N takes γ

M

to a curve

γ

N

= φ ◦ γ

M

: (−ǫ, ǫ) →N

on N with γ

N

(0) = φ(p). The interpretation of the tangent spaces

given above shows that the diﬀerential dφ

p

: T

p

M → T

φ(p)

N of φ at p

maps the tangent ˙ γ

M

(0) at p to the tangent ˙ γ

N

(0) at φ(p) i. e.

dφ

p

( ˙ γ

M

(0)) = ˙ γ

N

(0).

Deﬁnition 3.17. A diﬀerentiable map φ : M → N between man-

ifolds is said to be an immersion if for each p ∈ M the diﬀerential

dφ

p

: T

p

M → T

φ(p)

N is injective. An embedding is an immersion

φ : M →N which is a homeomorphism onto its image φ(M).

For positive integers m, n with m < n we have the inclusion map

φ : R

m+1

→R

n+1

given by

φ : (x

1

, . . . , x

m+1

) →(x

1

, . . . , x

m+1

, 0, . . . , 0).

The diﬀerential dφ

x

at x is injective since dφ

x

(v) = (v, 0). The map

φ is obviously a homeomorphism onto its image φ(R

m+1

) hence an

embedding. It is easily seen that even the restriction φ[

S

m : S

m

→ S

n

of φ to the m-dimensional unit sphere S

m

in R

m+1

is an embedding.

Deﬁnition 3.18. Let M be an m-dimensional diﬀerentiable man-

ifold and U be an open subset of R

m

. An immersion φ : U → M is

called a local parametrization of M.

If M is a diﬀerentiable manifold and (U, x) a chart on M then the

inverse x

−1

: x(U) →U of x is a parametrization of the subset U of M.

30 3. THE TANGENT SPACE

Example 3.19. Let S

1

be the unit circle in the complex plane C.

For a non-zero integer k ∈ Z deﬁne φ

k

: S

1

→C by φ

k

: z →z

k

. For a

point w ∈ S

1

let γ

w

: R → S

1

be the curve with γ

w

: t → we

it

. Then

γ

w

(0) = w and ˙ γ

w

(0) = iw. For the diﬀerential of φ

k

we have

(dφ

k

)

w

( ˙ γ

w

(0)) =

d

dt

(φ

k

◦ γ

w

(t))[

t=0

=

d

dt

(w

k

e

ikt

)[

t=0

= kiw

k

.

This shows that the diﬀerential (dφ

k

)

w

: T

w

S

1

∼

= R → T

w

kC

∼

= R

2

is

injective, so the map φ

k

is an immersion. It is easily seen that φ

k

is an

embedding if and only if k = ±1.

Example 3.20. Let q ∈ S

3

be a quaternion of unit length and

φ

q

: S

1

→ S

3

be the map deﬁned by φ

q

: z → qz. For w ∈ S

1

let

γ

w

: R → S

1

be the curve given by γ

w

(t) = we

it

. Then γ

w

(0) = w,

˙ γ

w

(0) = iw and φ

q

(γ

w

(t)) = qwe

it

. By diﬀerentiating we yield

dφ

q

( ˙ γ

w

(0)) =

d

dt

(φ

q

(γ

w

(t)))[

t=0

=

d

dt

(qwe

it

)[

t=0

= qiw.

Then [dφ

q

( ˙ γ

w

(0))[ = [qwi[ = [q[[w[ = 1 implies that the diﬀerential dφ

q

is injective. It is easily checked that the immersion φ

q

is an embedding.

In the next example we construct an interesting embedding of the

real projective space RP

m

into the vector space Sym(R

m+1

) of the real

symmetric (m+ 1) (m + 1) matrices.

Example 3.21. Let m be a positive integer and S

m

be the m-

dimensional unit sphere in R

m+1

. For a point p ∈ S

m

let

L

p

= ¦λp ∈ R

m+1

[ λ ∈ R¦

be the line through the origin generated by p and ρ

p

: R

m+1

→R

m+1

be

the reﬂection about the line L

p

. Then ρ

p

is an element of End(R

m+1

)

i.e. the set of linear endomorphisms of R

m+1

which can be identiﬁed

with R

(m+1)×(m+1)

. It is easily checked that the reﬂection about the

line L

p

is given by

ρ

p

: q →2¸q, p)p −q.

It then follows from the equations

ρ

p

(q) = 2¸q, p)p −q = 2p¸p, q) −q = (2pp

t

−e)q

that the matrix in R

(m+1)×(m+1)

corresponding to ρ

p

is just

(2pp

t

−e).

We shall now show that the map φ : S

m

→R

(m+1)×(m+1)

given by

φ : p →ρ

p

3. THE TANGENT SPACE 31

is an immersion. Let p be an arbitrary point on S

m

and α, β : I →S

m

be two curves meeting at p, that is α(0) = p = β(0), with a = ˙ α(0)

and b =

˙

β(0). For γ ∈ ¦α, β¦ we have

φ ◦ γ : t →(q →2¸q, γ(t))γ(t) −q)

so

(dφ)

p

( ˙ γ(0)) =

d

dt

(φ ◦ γ(t))[

t=0

= (q →2¸q, ˙ γ(0))γ(0) + 2¸q, γ(0)) ˙ γ(0)).

This means that

dφ

p

(a) = (q →2¸q, a)p + 2¸q, p)a)

and

dφ

p

(b) = (q →2¸q, b)p + 2¸q, p)b).

If a ,= b then dφ

p

(a) ,= dφ

p

(b) so the diﬀerential dφ

p

is injective, hence

the map φ : S

m

→R

(m+1)×(m+1)

is an immersion.

If the points p, q ∈ S

m

are linearly independent, then the lines

L

p

and L

q

are diﬀerent. But these are just the eigenspaces of ρ

p

and

ρ

q

with the eigenvalue +1, respectively. This shows that the linear

endomorphisms ρ

p

, ρ

q

of R

m+1

are diﬀerent in this case.

On the other hand, if p and q are parallel then p = ±q hence

ρ

p

= ρ

q

. This means that the image φ(S

m

) can be identiﬁed with the

quotient space S

m

/ ≡ where ≡ is the equivalence relation deﬁned by

x ≡ y if and only if x = ±y.

This of course is the real projective space RP

m

so the map φ induces

an embedding ψ : RP

m

→Sym(R

m+1

) with

ψ : [p] →ρ

p

.

For each p ∈ S

m

the reﬂection ρ

p

: R

m+1

→ R

m+1

about the line L

p

satisﬁes

ρ

p

ρ

t

p

= e.

This shows that the image ψ(RP

m

) = φ(S

m

) is not only contained in

the linear space Sym(R

m+1

) but also in the orthogonal group O(m+1)

which we know is a submanifold of R

(m+1)×(m+1)

The following result was proved by Hassler Whitney in his famous

paper, Diﬀerentiable Manifolds, Ann. of Math. 37 (1936), 645-680.

Deep Result 3.22. For 1 ≤ r ≤ ∞ let M be an m-dimensional

C

r

-manifold. Then there exists a C

r

-embedding φ : M → R

2m+1

into

the (2m + 1)-dimensional real vector space R

2m+1

.

32 3. THE TANGENT SPACE

The classical inverse function theorem generalizes to the manifold

setting as follows.

Theorem 3.23 (The Inverse Function Theorem). Let φ : M →N

be a diﬀerentiable map between manifolds with dimM = dimN. If

p is a point in M such that the diﬀerential dφ

p

: T

p

M → T

φ(p)

N at

p is bijective then there exist open neighborhoods U

p

around p and U

q

around q = φ(p) such that ψ = φ[

U

p

: U

p

→ U

q

is bijective and the

inverse ψ

−1

: U

q

→U

p

is diﬀerentiable.

Proof. See Exercise 3.8

We shall now generalize the classical implicit function theorem to

manifolds. For this we need the following deﬁnition.

Deﬁnition 3.24. Let m, n be positive natural numbers and φ :

N

n

→M

m

be a diﬀerentiable map between manifolds. A point p ∈ N

is said to be critical for φ if the diﬀerential

dφ

p

: T

p

N →T

φ(p)

M

is not of full rank, and regular if it is not critical. A point q ∈ φ(N) is

said to be a regular value of φ if every point of the pre-image φ

−1

(¦q¦)

of ¦q¦ is regular and a critical value otherwise.

Theorem 3.25 (The Implicit Function Theorem). Let φ : N

n

→

M

m

be a diﬀerentiable map between manifolds such that n > m. If

q ∈ φ(N) is a regular value, then the pre-image φ

−1

(¦q¦) of q is an

(n−m)-dimensional submanifold of N

n

. The tangent space T

p

φ

−1

(¦q¦)

of φ

−1

(¦q¦) at p is the kernel of the diﬀerential dφ

p

i.e. T

p

φ

−1

(¦q¦) =

Ker dφ

p

.

Proof. Let (V

q

, ψ

q

) be a chart on M with q ∈ V

q

and ψ

q

(q) = 0.

For a point p ∈ φ

−1

(¦q¦) we choose a chart (U

p

, ψ

p

) on N such that

p ∈ U

p

, ψ

p

(p) = 0 and φ(U

p

) ⊂ V

q

. The diﬀerential of the map

ˆ

φ = ψ

q

◦ φ ◦ ψ

−1

p

[

ψ

p

(U

p

)

: ψ

p

(U

p

) →R

m

at the point 0 is given by

d

ˆ

φ

0

= (dψ

q

)

q

◦ dφ

p

◦ (dψ

−1

p

)

0

: T

0

R

n

→T

0

R

m

.

The pairs (U

p

, ψ

p

) and (V

q

, ψ

q

) are charts so the diﬀerentials (dψ

q

)

q

and

(dψ

−1

p

)

0

are bijective. This means that the diﬀerential d

ˆ

φ

0

is surjective

since dφ

p

is. It then follows from the implicit function theorem 2.12

that ψ

p

(φ

−1

(¦q¦)∩U

p

) is an (n−m)-dimensional submanifold of ψ

p

(U

p

).

Hence φ

−1

(¦q¦)∩U

p

is an (n−m)-dimensional submanifold of U

p

. This

is true for each point p ∈ φ

−1

(¦q¦) so we have proven that φ

−1

(¦q¦) is

an (n −m)-dimensional submanifold of N

n

.

3. THE TANGENT SPACE 33

Let γ : (−ǫ, ǫ) →φ

−1

(¦q¦) be a curve, such that γ(0) = p. Then

(dφ)

p

( ˙ γ(0)) =

d

dt

(φ ◦ γ(t))[

t=0

=

dq

dt

[

t=0

= 0.

This implies that T

p

φ

−1

(¦q¦) is contained in and has the same dimen-

sion as the kernel of dφ

p

, so T

p

φ

−1

(¦q¦) = Ker dφ

p

.

Deﬁnition 3.26. For positive integers m, n with n ≥ m a map

φ : N

n

→ M

m

between two manifolds is said to be a submersion if

for each p ∈ N the diﬀerential dφ

p

: T

p

N →T

φ(p)

M is surjective.

If m, n ∈ N such that n ≥ m then we have the projection map

π : R

n

→ R

m

given by π : (x

1

, . . . , x

n

) → (x

1

, . . . , x

m

). Its diﬀerential

dπ

x

at a point x is surjective since

dπ

x

(v

1

, . . . , v

n

) = (v

1

, . . . , v

m

).

This means that the projection is a submersion. An important sub-

mersion between spheres is given by the following.

Example 3.27. Let S

3

and S

2

be the unit spheres in C

2

and C

R

∼

= R

3

, respectively. The Hopf map φ : S

3

→S

2

is given by

φ : (x, y) →(2x¯ y, [x[

2

−[y[

2

).

The map φ and its diﬀerential dφ

p

: T

p

S

3

→ T

φ(p)

S

2

are surjective for

each p ∈ S

3

. This implies that each point q ∈ S

2

is a regular value and

the ﬁbres of φ are 1-dimensional submanifolds of S

3

. They are actually

the great circles given by

φ

−1

(¦(2x¯ y, [x[

2

−[y[

2

)¦) = ¦e

iθ

(x, y)[ θ ∈ R¦.

This means that the 3-dimensional sphere S

3

is a disjoint union of great

circles

S

3

=

_

q∈S

2

φ

−1

(¦q¦).

34 3. THE TANGENT SPACE

Exercises

Exercise 3.1. Let p be an arbitrary point on the unit sphere S

2n+1

of C

n+1

∼

= R

2n+2

. Determine the tangent space T

p

S

2n+1

and show that

it contains an n-dimensional complex subspace of C

n+1

.

Exercise 3.2. Find a proof for Proposition 3.12.

Exercise 3.3. Prove that the matrices

X

1

=

_

1 0

0 −1

_

, X

2

=

_

0 −1

1 0

_

, X

3

=

_

0 1

1 0

_

form a basis for the tangent space T

e

SL

2

(R) of the real special linear

group SL

2

(R) at e. For each k = 1, 2, 3 ﬁnd an explicit formula for the

curve γ

k

: R →SL

2

(R) given by

γ

k

: s →Exp(sX

k

).

Exercise 3.4. Find a proof for Proposition 3.16.

Exercise 3.5. Prove that the matrices

Z

1

=

_

0 −1

1 0

_

, Z

2

=

_

0 i

i 0

_

, Z

3

=

_

i 0

0 −i

_

form a basis for the tangent space T

e

SU(2) of the special unitary group

SU(2) at e. For each k = 1, 2, 3 ﬁnd an explicit formula for the curve

γ

k

: R →SU(2) given by

γ

k

: s →Exp(sZ

k

).

Exercise 3.6. For each k ∈ N

0

deﬁne φ

k

: C → C and ψ

k

: C

∗

→

C by φ

k

, ψ

k

: z → z

k

. For which k ∈ N

0

are φ

k

, ψ

k

immersions,

submersions or embeddings.

Exercise 3.7. Prove that the map φ : R

m

→C

m

given by

φ : (x

1

, . . . , x

m

) →(e

ix

1

, . . . , e

ix

m

)

is a parametrization of the m-dimensional torus T

m

in C

m

.

Exercise 3.8. Find a proof for Theorem 3.23.

Exercise 3.9. Prove that the Hopf-map φ : S

3

→ S

2

with φ :

(x, y) →(2x¯ y, [x[

2

−[y[

2

) is a submersion.

CHAPTER 4

The Tangent Bundle

In this chapter we introduce the tangent bundle TM of a diﬀer-

entiable manifold M. Intuitively, this is the object that we get by

glueing at each point p of M the corresponding tangent space T

p

M.

The diﬀerentiable structure on M induces a diﬀerentiable structure on

the tangent bundle TM turning it into a diﬀerentiable manifold.

We have already seen that for a point p ∈ R

m

the tangent space

T

p

R

m

can be identiﬁed with the m-dimensional vector space R

m

. This

means that if we at each point p ∈ R

m

glue T

p

R

m

to R

m

we yield the

so called tangent bundle of R

m

TR

m

= ¦(p, v)[ p ∈ R

m

, v ∈ T

p

R

m

¦.

For this we have the natural projection π : TR

m

→R

m

deﬁned by

π : (p, v) →p

and for a point p ∈ M the ﬁbre π

−1

(¦p¦) over p is precisely the tangent

space T

p

R

m

at p.

Classically, a vector ﬁeld X on R

m

is a smooth map X : R

m

→R

m

but we would like to view it as a map X : R

m

→TR

m

into the tangent

bundle and with abuse of notation write

X : p →(p, X(p)).

Following Proposition 3.10 two vector ﬁelds X, Y : R

m

→TR

m

can be

written as

X =

m

k=1

a

k

∂

∂x

k

and Y =

m

k=1

b

k

∂

∂x

k

,

where a

k

, b

k

: R

m

→ R are smooth functions deﬁned on R

m

. If f :

R

m

→R is another such function the commutator [X, Y ] acts on f as

follows:

[X, Y ](f) = X(Y (f)) −Y (X(f))

=

m

k,l=1

_

a

k

∂

∂x

k

(b

l

∂

∂x

l

) −b

k

∂

∂x

k

(a

l

∂

∂x

l

)

_

(f)

35

36 4. THE TANGENT BUNDLE

=

m

k,l=1

_

a

k

∂b

l

∂x

k

∂

∂x

l

+ a

k

b

l

∂

2

∂x

k

∂x

l

−b

k

∂a

l

∂x

k

∂

∂x

l

−b

k

a

l

∂

2

∂x

k

∂x

l

_

(f)

=

m

k,l=1

_

a

k

∂b

l

∂x

k

−b

k

∂a

l

∂x

k

__

∂

∂x

l

_

(f).

This shows that the commutator [X, Y ] is a smooth vector ﬁeld on R

m

.

We shall now generalize to the manifold setting. This leads us ﬁrst

to the following notion of a topological vector bundle.

Deﬁnition 4.1. Let E and M be topological manifolds and π :

E →M be a continuous surjective map. The triple (E, M, π) is called

an n-dimensional topological vector bundle over M if

(i) for each p ∈ M the ﬁbre E

p

= π

−1

(¦p¦) is an n-dimensional

vector space,

(ii) for each p ∈ M there exists a bundle chart (π

−1

(U), ψ) con-

sisting of the pre-image π

−1

(U) of an open neighbourhood U

of p and a homeomorphism ψ : π

−1

(U) → U R

n

such that

for all q ∈ U the map ψ

q

= ψ[

E

q

: E

q

→ ¦q¦ R

n

is a vector

space isomorphism.

A continuous map σ : M → E is called a section of the bundle

(E, M, π) if π ◦ σ(p) = p for each p ∈ M.

Deﬁnition 4.2. A topological vector bundle (E, M, π) over M, of

dimension n, is said to be trivial if there exists a global bundle chart

ψ : E →M R

n

.

If n is a positive integer and M is a topological manifold then we

have the n-dimensional vector bundle (M R

n

, M, π) where

π : M R

n

→M

is the projection map π : (p, v) →p. The identity map ψ : M R

n

→

M R

n

is a global bundle chart so the bundle is trivial.

Deﬁnition 4.3. Let (E, M, π) be an n-dimensional topological vec-

tor bundle over M. A collection

B = ¦(π

−1

(U

α

), ψ

α

)[ α ∈ I¦

of bundle charts is called a bundle atlas for (E, M, π) if M = ∪

α

U

α

and for α, β ∈ I there exists a map A

α,β

: U

α

∩ U

β

→ GL

n

(R) such

4. THE TANGENT BUNDLE 37

that the corresponding continuous map

ψ

β

◦ ψ

−1

α

[

(U

α

∩U

β

)×R

n : (U

α

∩ U

β

) R

n

→(U

α

∩ U

β

) R

n

is given by

(p, v) →(p, (A

α,β

(p))(v)).

The elements of ¦A

α,β

[ α, β ∈ I¦ are called the transition maps of

the bundle atlas B.

Deﬁnition 4.4. Let E and M be diﬀerentiable manifolds and π :

E →M be a diﬀerentiable map such that (E, M, π) is an n-dimensional

topological vector bundle. A bundle atlas B for (E, M, π) is said to be

diﬀerentiable if the corresponding transition maps are diﬀerentiable. A

diﬀerentiable vector bundle is a topological vector bundle together

with a maximal diﬀerentiable bundle atlas. By C

∞

(E) we denote the

set of all smooth sections of (E, M, π).

From now on we shall assume, when not stating otherwise,

that all our vector bundles are smooth.

Deﬁnition 4.5. Let (E, M, π) be a vector bundle over a manifold

M. Then we deﬁne the operations + and on the set C

∞

(E) of smooth

sections of (E, M, π) by

(i) (v + w)

p

= v

p

+ w

p

,

(ii) (f v)

p

= f(p) v

p

for all v, w ∈ C

∞

(E) and f ∈ C

∞

(M). If U is an open subset of M

then a set ¦v

1

, . . . , v

n

¦ of smooth sections v

1

, . . . , v

n

: U → E on U is

called a local frame for E if for each p ∈ U the set ¦(v

1

)

p

, . . . , (v

n

)

p

¦

is a basis for the vector space E

p

.

With the above deﬁned operations C

∞

(E) becomes a module over

C

∞

(M) and in particular a vector space over the real numbers as the

constant functions in C

∞

(M).

Example 4.6. Let M

m

be a diﬀerentiable manifold with maximal

atlas

ˆ

/. Deﬁne the set TM by

TM = ¦(p, v)[ p ∈ M, v ∈ T

p

M¦

and let π : TM →M be the projection map satisfying

π : (p, v) →p.

Then the ﬁbre π

−1

(¦p¦) over a point p ∈ M is the m-dimensional tan-

gent space T

p

M. The triple (TM, M, π) is called the tangent bundle

of M.

38 4. THE TANGENT BUNDLE

We shall now equip TM with the structure of a smooth manifold.

For every local chart x : U → R

m

from the maximal atlas

ˆ

/ of M we

deﬁne a local chart

x

∗

: π

−1

(U) →R

m

R

m

on the tangent bundle TM by the formula

x

∗

: (p,

m

k=1

v

k

_

∂

∂x

k

_

p

) →(x(p), (v

1

, . . . , v

m

)).

Proposition 3.10 shows that the map x

∗

is well-deﬁned. The collection

¦(x

∗

)

−1

(W) ⊂ TM[ (U, x) ∈

ˆ

/ and W ⊂ x(U) R

m

open¦

is a basis for a topology T

TM

on TM and (π

−1

(U), x

∗

) is a chart on the

2m-dimensional topological manifold (TM, T

TM

).

If (U, x) and (V, y) are two charts in

ˆ

/ such that p ∈ U ∩ V then

the transition map

(y

∗

) ◦ (x

∗

)

−1

: x

∗

(π

−1

(U ∩ V )) →R

m

R

m

is given by

(a, b) →(y ◦ x

−1

(a),

m

k=1

∂y

1

∂x

k

(x

−1

(a))b

k

, . . . ,

m

k=1

∂y

m

∂x

k

(x

−1

(a))b

k

).

Since we are assuming that y ◦ x

−1

is diﬀerentiable it follows that

(y

∗

) ◦ (x

∗

)

−1

is also diﬀerentiable. This means that

/

∗

= ¦(π

−1

(U), x

∗

)[ (U, x) ∈

ˆ

/¦

is a C

r

-atlas on TM so (TM,

´

/

∗

) is a diﬀerentiable manifold. It is

trivial that the surjective projection map π : TM →M is diﬀerentiable.

We shall now describe how the tangent bundle (TM, M, π) can be

given the structure of a diﬀerentiable vector bundle. For each point

p ∈ M the ﬁbre π

−1

(¦p¦) is the tangent space T

p

M and hence an m-

dimensional vector space. For a local chart x : U →R

m

in the maximal

atlas

ˆ

/ of M we deﬁne ¯ x : π

−1

(U) →U R

m

by

¯ x : (p,

m

k=1

v

k

_

∂

∂x

k

_

p

) →(p, (v

1

, . . . , v

m

)).

The restriction ¯ x

p

= ¯ x[

T

p

M

: T

p

M → ¦p¦ R

m

to the tangent space

T

p

M is given by

¯ x

p

:

m

k=1

v

k

_

∂

∂x

k

_

p

→(v

1

, . . . , v

m

),

4. THE TANGENT BUNDLE 39

so it is a vector space isomorphism. This implies that the map

¯ x : π

−1

(U) →U R

m

is a local bundle chart. It is not diﬃcult to see that

B = ¦(π

−1

(U), ¯ x)[ (U, x) ∈

ˆ

/¦

is a bundle atlas turning (TM, M, π) into an m-dimensional topological

vector bundle, see Exercise 4.1. It immediately follows from above that

(TM, M, π) together with the maximal bundle atlas

ˆ

B deﬁned by B is

a diﬀerentiable vector bundle.

Deﬁnition 4.7. Let M be a diﬀerentiable manifold, then a section

X : M →TM of the tangent bundle is called a vector ﬁeld. The set

of smooth vector ﬁelds X : M →TM is denoted by C

∞

(TM).

Example 4.8. We have seen earlier that the 3-sphere S

3

in H

∼

= C

2

carries a group structure given by

(z, w) (α, β) = (zα −w

¯

β, zβ + w¯ α).

This makes (S

3

, ) into a Lie group with neutral element e = (1, 0).

Put v

1

= (i, 0), v

2

= (0, 1) and v

3

= (0, i) and for k = 1, 2, 3 deﬁne the

curves γ

k

: R →S

3

with

γ

k

: t →cos t (1, 0) + sin t v

k

.

Then γ

k

(0) = e and ˙ γ

k

(0) = v

k

so v

1

, v

2

, v

3

are elements of the tangent

space T

e

S

3

. They are linearily independent so they generate T

e

S

3

.

The group structure on S

3

can be used to extend vectors in T

e

S

3

to

vector ﬁelds on S

3

as follows. For p ∈ S

3

let L

p

: S

3

→ S

3

be the left

translation on S

3

by p given by L

p

: q → p q. Then deﬁne the vector

ﬁelds X

1

, X

2

, X

3

∈ C

∞

(TS

3

) by

(X

k

)

p

= (dL

p

)

e

(v

k

) =

d

dt

(L

p

(γ

k

(t)))[

t=0

.

It is left as an exercise for the reader to show that at a point p =

(z, w) ∈ S

3

the values of X

k

at p is given by

(X

1

)

p

= (z, w) (i, 0) = (iz, −iw),

(X

2

)

p

= (z, w) (0, 1) = (−w, z),

(X

3

)

p

= (z, w) (0, i) = (iw, iz).

Our next aim is to introduce the Lie bracket on the set of vector

ﬁelds C

∞

(TM) on M.

40 4. THE TANGENT BUNDLE

Deﬁnition 4.9. Let M be a smooth manifold. For two vector ﬁelds

X, Y ∈ C

∞

(TM) we deﬁne the Lie bracket [X, Y ]

p

: C

∞

(M) →R of

X and Y at p ∈ M by

[X, Y ]

p

(f) = X

p

(Y (f)) −Y

p

(X(f)).

The next result shows that the Lie bracket [X, Y ]

p

actually is an

element of the tangent space T

p

M.

Proposition 4.10. Let M be a smooth manifold, X, Y ∈ C

∞

(TM)

be vector ﬁelds on M, f, g ∈ C

∞

(M, R) and λ, µ ∈ R. Then

(i) [X, Y ]

p

(λ f + µ g) = λ [X, Y ]

p

(f) + µ [X, Y ]

p

(g),

(ii) [X, Y ]

p

(f g) = [X, Y ]

p

(f) g(p) + f(p) [X, Y ]

p

(g).

Proof.

[X, Y ]

p

(λf +µg)

= X

p

(Y (λf + µg)) −Y

p

(X(λf + µg))

= λX

p

(Y (f)) +µX

p

(Y (g)) −λY

p

(X(f)) −µY

p

(X(g))

= λ[X, Y ]

p

(f) + µ[X, Y ]

p

(g).

[X, Y ]

p

(f g)

= X

p

(Y (f g)) −Y

p

(X(f g))

= X

p

(f Y (g) + g Y (f)) −Y

p

(f X(g) + g X(f))

= X

p

(f)Y

p

(g) + f(p)X

p

(Y (g)) +X

p

(g)Y

p

(f) + g(p)X

p

(Y (f))

−Y

p

(f)X

p

(g) −f(p)Y

p

(X(g)) −Y

p

(g)X

p

(f) −g(p)Y

p

(X(f))

= f(p)¦X

p

(Y (g)) −Y

p

(X(g))¦ + g(p)¦X

p

(Y (f)) −Y

p

(X(f))¦

= f(p)[X, Y ]

p

(g) + g(p)[X, Y ]

p

(f).

**Proposition 4.10 implies that if X, Y are vector ﬁelds on M then
**

the map [X, Y ] : M → TM given by [X, Y ] : p → [X, Y ]

p

is a section

of the tangent bundle. In Proposition 4.12 we shall prove that this

section is smooth. For this we need the following technical lemma.

Lemma 4.11. Let M

m

be a smooth manifold and X : M → TM

be a section of TM. Then the following conditions are equivalent

(i) the section X is smooth,

(ii) if (U, x) is a chart on M then the functions a

1

, . . . , a

m

: U →R

given by

X[

U

=

m

k=1

a

k

∂

∂x

k

,

are smooth,

4. THE TANGENT BUNDLE 41

(iii) if f : V →R deﬁned on an open subset V of M is smooth, then

the function X(f) : V →R with X(f)(p) = X

p

(f) is smooth.

Proof. (i) ⇒ (ii): The functions a

k

= π

m+k

◦ x

∗

◦ X[

U

: U →

TM → x(U) R

m

→ R are restrictions of compositions of smooth

maps so therefore smooth.

(ii) ⇒ (iii): Let (U, x) be a chart on M such that U is contained

in V . By assumption the map

X(f[

U

) =

m

i=1

a

i

∂f

∂x

i

is smooth. This is true for each such chart (U, x) so the function X(f)

is smooth.

(iii) ⇒(i): Note that the smoothness of the section X is equivalent

to x

∗

◦ X[

U

: U → R

2m

being smooth for all charts (U, x) on M. On

the other hand, this is equivalent to x

∗

k

= π

k

◦ x

∗

◦ X[

U

: U →R being

smooth for all k = 1, 2, . . . , 2m and all charts (U, x) on M. It is trivial

that the coordinates x

∗

k

= x

k

for k = 1, . . . , m are smooth. But x

∗

m+k

=

a

k

= X(x

k

) for k = 1, . . . , m hence also smooth by assumption.

Proposition 4.12. Let M be a manifold and X, Y ∈ C

∞

(TM) be

vector ﬁelds on M. Then the section [X, Y ] : M →TM of the tangent

bundle given by [X, Y ] : p →[X, Y ]

p

is smooth.

Proof. Let f : M → R be an arbitrary smooth function on M

then [X, Y ](f) = X(Y (f)) − Y (X(f)) is smooth so it follows from

Lemma 4.11 that the section [X, Y ] is smooth.

For later use we prove the following useful result.

Lemma 4.13. Let M be a smooth manifold and [, ] be the Lie

bracket on the tangent bundle TM. Then

(i) [X, f Y ] = X(f) Y + f [X, Y ],

(ii) [f X, Y ] = f [X, Y ] −Y (f) X

for all X, Y ∈ C

∞

(TM) and f ∈ C

∞

(M).

Proof. If g ∈ C

∞

(M), then

[X, f Y ](g) = X(f Y (g)) −f Y (X(g))

= X(f) Y (g) + f X(Y (g)) −f Y (X(g))

= (X(f) Y +f [X, Y ])(g)

This proves the ﬁrst statement and the second follows from the skew-

symmetry of the Lie bracket.

42 4. THE TANGENT BUNDLE

Deﬁnition 4.14. A real vector space (V, +, ) equipped with an

operation [, ] : V V → V is said to be a real Lie algebra if the

following relations hold

(i) [λX + µY, Z] = λ[X, Z] + µ[Y, Z],

(ii) [X, Y ] = −[Y, X],

(iii) [X, [Y, Z]] + [Z, [X, Y ]] + [Y, [Z, X]] = 0

for all X, Y, Z ∈ V and λ, µ ∈ R. The equation (iii) is called the Jacobi

identity.

Theorem 4.15. Let M be a smooth manifold. The vector space

C

∞

(TM) of smooth vector ﬁelds on M equipped with the Lie bracket

[, ] : C

∞

(TM) C

∞

(TM) →C

∞

(TM) is a real Lie algebra.

Proof. See exercise 4.4.

If φ : M → N is a surjective map between diﬀerentiable manifolds

then two vector ﬁelds X ∈ C

∞

(TM),

¯

X ∈ C

∞

(TN) are said to be

φ-related if dφ

p

(X) =

¯

X

φ(p)

for all p ∈ M. In that case we write

¯

X = dφ(X).

Proposition 4.16. Let φ : M →N be a map between diﬀerentiable

manifolds, X, Y ∈ C

∞

(TM),

¯

X,

¯

Y ∈ C

∞

(TN) such that

¯

X = dφ(X)

and

¯

Y = dφ(Y ). Then

[

¯

X,

¯

Y ] = dφ([X, Y ]).

Proof. Let f : N →R be a smooth function, then

[

¯

X,

¯

Y ](f) = dφ(X)(dφ(Y )(f)) −dφ(Y )(dφ(X)(f))

= X(dφ(Y )(f) ◦ φ) −Y (dφ(X)(f) ◦ φ)

= X(Y (f ◦ φ)) −Y (X(f ◦ φ))

= [X, Y ](f ◦ φ)

= dφ([X, Y ])(f).

**Proposition 4.17. Let φ : M → N be a smooth bijective map
**

between diﬀerentiable manifolds. If X, Y ∈ C

∞

(TM) are vector ﬁelds

on M, then

(i) dφ(X) ∈ C

∞

(TN),

(ii) the map dφ : C

∞

(TM) →C

∞

(TN) is a Lie algebra homomor-

phism i.e. [dφ(X), dφ(Y )] = dφ([X, Y ]).

Proof. The fact that the map φ is bijective implies that dφ(X)

is a section of the tangent bundle. That dφ(X) ∈ C

∞

(TN) follows

4. THE TANGENT BUNDLE 43

directly from the fact that

dφ(X)(f)(φ(p)) = X(f ◦ φ)(p).

The last statement is a direct consequence of Proposition 4.16.

Deﬁnition 4.18. Let M be a smooth manifold. Two vector ﬁelds

X, Y ∈ C

∞

(TM) are said to commute if [X, Y ] = 0.

Let (U, x) be local coordinates on a manifold M and let

¦

∂

∂x

k

[ k = 1, 2, . . . , m¦

be the induced local frame for the tangent bundle. For k = 1, 2, . . . , m

the vector ﬁeld ∂/∂x

k

is x-related to the constant coordinate vector

ﬁeld e

k

in R

m

. This implies that

dx([

∂

∂x

k

,

∂

∂x

l

]) = [e

k

, e

l

] = 0.

Hence the local frame ﬁelds commute.

Deﬁnition 4.19. Let G be a Lie group with neutral element e. For

p ∈ G let L

p

: G →G be the left translation by p with L

p

: q →pq. A

vector ﬁeld X ∈ C

∞

(TG) is said to be left invariant if dL

p

(X) = X

for all p ∈ G, or equivalently, X

pq

= (dL

p

)

q

(X

q

) for all p, q ∈ G. The

set of all left invariant vector ﬁelds on G is called the Lie algebra of

G and is denoted by g.

The Lie algebras of the classical Lie groups introduced earlier are

denoted by gl

m

(R), sl

m

(R), o(m), so(m), gl

m

(C), sl

m

(C), u(m) and

su(m), respectively.

Proposition 4.20. Let G be a Lie group and g be the Lie algebra

of G. Then g is a Lie subalgebra of C

∞

(TG) i.e. if X, Y ∈ g then

[X, Y ] ∈ g,

Proof. If p ∈ G then

dL

p

([X, Y ]) = [dL

p

(X), dL

p

(Y )] = [X, Y ]

for all X, Y ∈ g. This proves that the Lie bracket [X, Y ] of two left

invariant vector ﬁelds X, Y is left invariant and thereby that g is a Lie

subalgebra of C

∞

(TG).

Note that if X is a left invariant vector ﬁeld on G then

X

p

= (dL

p

)

e

(X

e

)

44 4. THE TANGENT BUNDLE

so the value X

p

of X at p ∈ G is completely determined by the value

X

e

of X at e. This means that the map ∗ : T

e

G →g given by

∗ : X →(X

∗

: p →(dL

p

)

e

(X))

is a vector space isomorphism and that we can deﬁne a Lie bracket

[, ] : T

e

GT

e

G →T

e

G on the tangent space T

e

G by

[X, Y ] = [X

∗

, Y

∗

]

e

.

Proposition 4.21. Let G be one of the classical Lie groups and

T

e

G be the tangent space of G at the neutral element e. Then the Lie

bracket on T

e

G

[, ] : T

e

GT

e

G →T

e

G

is given by

[X

e

, Y

e

] = X

e

Y

e

−Y

e

X

e

where is the usual matrix multiplication.

Proof. We shall prove the result for the case when G is the real

general linear group GL

m

(R). For the other real classical Lie groups

the result follows from the fact that they are all subgroups of GL

m

(R).

The same proof can be used for the complex cases.

Let X, Y ∈ gl

m

(R) be left invariant vector ﬁelds on GL

m

(R), f :

U → R be a function deﬁned locally around the identity element e ∈

GL

m

(R) and p be an arbitrary point in U. Then the derivative X

p

(f)

is given by

X

p

(f) =

d

ds

(f(p Exp(sX

e

)))[

s=0

= df

p

(p X

e

) = df

p

(X

p

).

The real general linear group GL

m

(R) is an open subset of R

m×m

so

we can use well-known rules from calculus and the second derivative

Y

e

(X(f)) is obtained as follows:

Y

e

(X(f)) =

d

dt

(X

Exp(tY

e

)

(f))[

t=0

=

d

dt

(df

Exp(tY

e

)

(Exp(tY

e

) X

e

))[

t=0

= d

2

f

e

(Y

e

, X

e

) + df

e

(Y

e

X

e

).

The Hessian d

2

f

e

of f is symmetric, hence

[X, Y ]

e

(f) = X

e

(Y (f)) −Y

e

(X(f)) = df

e

(X

e

Y

e

−Y

e

X

e

).

**Theorem 4.22. Let G be a Lie group. Then the tangent bundle
**

TG of G is trivial.

4. THE TANGENT BUNDLE 45

Proof. Let ¦X

1

, . . . , X

m

¦ be a basis for T

e

G. Then the map ψ :

TG →GR

m

given by

ψ : (p,

m

k=1

v

k

(X

∗

k

)

p

) →(p, (v

1

, . . . , v

m

))

is a global bundle chart so the tangent bundle TG is trivial.

46 4. THE TANGENT BUNDLE

Exercises

Exercise 4.1. Let (M

m

,

ˆ

/) be a smooth manifold and (U, x), (V, y)

be two charts in

ˆ

/ such that U ∩ V ,= ∅. Let

f = y ◦ x

−1

: x(U ∩ V ) →R

m

be the corresponding transition map. Show that the local frames

¦

∂

∂x

i

[ i = 1, . . . , m¦ and ¦

∂

∂y

j

[ j = 1, . . . , m¦

for TM on U ∩ V are related by

∂

∂x

i

=

m

j=1

∂(f

j

◦ x)

∂x

i

∂

∂y

j

.

Exercise 4.2. Let m be a positive integer an SO(m) be the corre-

sponding special orthogonal group.

(i) Find a basis for the tangent space T

e

SO(m),

(ii) construct a non-vanishing vector ﬁeld Z ∈ C

∞

(TSO(m)),

(iii) determine all smooth vector ﬁelds on SO(2).

The Hairy Ball Theorem. Let m be a positive integer. Then there

does not exist a continuous non-vanishing vector ﬁeld X ∈ C

0

(TS

2m

)

on the even dimensional sphere S

2m

.

Exercise 4.3. Let m be a positive integer. Use the Hairy Ball The-

orem to prove that the tangent bundles TS

2m

of the even-dimensional

spheres S

2m

are not trivial. Construct a non-vanishing vector ﬁeld

X ∈ C

∞

(TS

2m+1

) on the odd-dimensional sphere S

2m+1

.

Exercise 4.4. Find a proof for Theorem 4.15.

CHAPTER 5

Riemannian Manifolds

In this chapter we introduce the notion of a Riemannian manifold

(M, g). The metric g provides us with an inner product on each tangent

space and can be used to measure the length of curves in the manifold.

It deﬁnes a distance function and turns the manifold into a metric space

in a natural way. A Riemannian metric on a diﬀerentiable manifold is

an important example of what is called a tensor ﬁeld.

Let M be a smooth manifold, C

∞

(M) denote the commutative ring

of smooth functions on M and C

∞

(TM) be the set of smooth vector

ﬁelds on M forming a module over C

∞

(M). Deﬁne

C

∞

0

(TM) = C

∞

(M)

and for each positive integer r ∈ Z

+

let

C

∞

r

(TM) = C

∞

(TM) ⊗ ⊗C

∞

(TM)

be the r-fold tensor product of C

∞

(TM) over C

∞

(M).

Deﬁnition 5.1. Let M be a diﬀerentiable manifold. A smooth

tensor ﬁeld A on M of type (r, s) is a map A : C

∞

r

(TM) →C

∞

s

(TM)

which is multi-linear over C

∞

(M) i.e. satisfying

A(X

1

⊗ ⊗X

k−1

⊗(f Y +g Z) ⊗X

k+1

⊗ ⊗X

r

)

= f A(X

1

⊗ ⊗X

k−1

⊗Y ⊗X

k+1

⊗ ⊗X

r

)

+g A(X

1

⊗ ⊗X

k−1

⊗Z ⊗X

k+1

⊗ ⊗X

r

)

for all X

1

, . . . , X

r

, Y, Z ∈ C

∞

(TM), f, g ∈ C

∞

(M) and k = 1, . . . , r.

For the rest of this work we shall for A(X

1

⊗ ⊗X

r

) use the notation

A(X

1

, . . . , X

r

).

The next general result provides us with the most important prop-

erty of tensor ﬁelds. It shows that the value of A(X

1

, . . . , X

r

) at a point

p ∈ M only depends on the values of the vector ﬁelds X

1

, . . . , X

r

at p

and is independent of their values away from p.

Proposition 5.2. Let A : C

∞

r

(TM) → C

∞

s

(TM) be a tensor ﬁeld

of type (r, s) and p ∈ M. Let X

1

, . . . , X

r

and Y

1

, . . . , Y

r

be smooth

47

48 5. RIEMANNIAN MANIFOLDS

vector ﬁelds on M such that (X

k

)

p

= (Y

k

)

p

for each k = 1, . . . , r. Then

A(X

1

, . . . , X

r

)(p) = A(Y

1

, . . . , Y

r

)(p).

Proof. We shall prove the statement for r = 1, the rest follows by

induction. Put X = X

1

and Y = Y

1

and let (U, x) be local coordinates

on M. Choose a function f ∈ C

∞

(M) such that f(p) = 1,

support(f) = ¦p ∈ M[ f(p) ,= 0¦

is contained in U and deﬁne the vector ﬁelds v

1

, . . . , v

m

∈ C

∞

(TM) on

M by

(v

k

)

q

=

_

f(q) (

∂

∂x

k

)

q

if q ∈ U

0 if q / ∈ U

Then there exist functions ρ

k

, σ

k

∈ C

∞

(M) such that

f X =

m

k=1

ρ

k

v

k

and f Y =

m

k=1

σ

k

v

k

.

This implies that

A(X)(p) = f(p)A(X)(p) = A(f X)(p) =

m

k=1

ρ

k

(p)A(v

k

)(p)

and similarily

A(Y )(p) =

m

k=1

σ

k

(p)A(v

k

)(p).

The fact that X

p

= Y

p

shows that ρ

k

(p) = σ

k

(p) for all k. As a direct

consequence we see that

A(X)(p) = A(Y )(p).

**For a tensor A we shall by A
**

p

denote the multi-linear restriction of

A to the r-fold tensor product

T

p

M ⊗ ⊗T

p

M

of the vector space T

p

M over R given by

A

p

: ((X

1

)

p

, . . . , (X

r

)

p

) →A(X

1

, . . . , X

r

)(p).

Deﬁnition 5.3. Let M be a smooth manifold. A Riemannian

metric g on M is a tensor ﬁeld

g : C

∞

2

(TM) →C

∞

0

(TM)

such that for each p ∈ M the restriction

g

p

= g[

T

p

M⊗T

p

M

: T

p

M ⊗T

p

M →R

5. RIEMANNIAN MANIFOLDS 49

with

g

p

: (X

p

, Y

p

) →g(X, Y )(p)

is an inner product on the tangent space T

p

M. The pair (M, g) is

called a Riemannian manifold. The study of Riemannian manifolds

is called Riemannian Geometry. Geometric properties of (M, g) which

only depend on the metric g are called intrinsic or metric properties.

The standard inner product on the vector space R

m

given by

¸X, Y )

R

m = X

t

Y =

m

k=1

X

k

Y

k

deﬁnes a Riemannian metric on R

m

. The Riemannian manifold

E

m

= (R

m

, ¸, )

R

m)

is called the m-dimensional Euclidean space.

Deﬁnition 5.4. Let γ : I → M be a C

1

-curve in M. Then the

length L(γ) of γ is deﬁned by

L(γ) =

_

I

_

g( ˙ γ(t), ˙ γ(t))dt.

By multiplying the Euclidean metric on subsets of R

m

by a factor

we obtain important examples of Riemannian manifolds.

Example 5.5. For a positive integer m equip the real vector space

R

m

with the Riemannian metric g given by

g

x

(X, Y ) =

4

(1 +[x[

2

R

m

)

2

¸X, Y )

R

m.

The Riemannian manifold Σ

m

= (R

m

, g) is called the m-dimensional

punctured round sphere. Let γ : R

+

→ Σ

m

be the curve with

γ : t → (t, 0, . . . , 0). Then the length L(γ) of γ can be determined as

follows

L(γ) = 2

_

∞

0

_

¸ ˙ γ, ˙ γ)

1 +[γ[

2

dt = 2

_

∞

0

dt

1 +t

2

= 2[arctan(t)]

∞

0

= π.

Example 5.6. Let B

m

1

(0) be the open unit ball in R

m

given by

B

m

1

(0) = ¦x ∈ R

m

[ [x[

R

m < 1¦.

By the m-dimensional hyperbolic ball we mean B

m

1

(0) equipped with

the Riemannian metric

g

x

(X, Y ) =

4

(1 −[x[

2

R

m

)

2

¸X, Y )

R

m.

50 5. RIEMANNIAN MANIFOLDS

Let γ : (0, 1) →B

m

1

(0) be a curve given by γ : t →(t, 0, . . . , 0). Then

L(γ) = 2

_

1

0

_

¸ ˙ γ, ˙ γ)

1 −[γ[

2

dt = 2

_

1

0

dt

1 −t

2

= [log(

1 + t

1 −t

)]

1

0

= ∞

As we shall now see a Riemannian manifold (M, g) has the structure

of a metric space (M, d) in a natural way.

Proposition 5.7. Let (M, g) be a Riemannian manifold. For two

points p, q ∈ M let C

pq

denote the set of C

1

-curves γ : [0, 1] →M such

that γ(0) = p and γ(1) = q and deﬁne the function d : M M → R

+

0

by

d(p, q) = inf¦L(γ)[ γ ∈ C

pq

¦.

Then (M, d) is a metric space i.e. for all p, q, r ∈ M we have

(i) d(p, q) ≥ 0,

(ii) d(p, q) = 0 if and only if p = q,

(iii) d(p, q) = d(q, p),

(iv) d(p, q) ≤ d(p, r) +d(r, q).

The topology on M induced by the metric d is identical to the one M

carries as a topological manifold (M, T ), see Deﬁnition 2.1.

Proof. See for example: Peter Petersen, Riemannian Geometry,

Graduate Texts in Mathematics 171, Springer (1998).

A Riemannian metric on a diﬀerentiable manifold induces a Rie-

mannian metric on any of its submanifolds as follows.

Deﬁnition 5.8. Let (N, h) be a Riemannian manifold and M be

a submanifold of N. Then the smooth tensor ﬁeld g : C

∞

2

(TM) →

C

∞

0

(M) given by

g(X, Y ) : p →h

p

(X

p

, Y

p

).

is a Riemannian metric on M called the induced metric on M in

(N, h).

The Euclidean metric ¸, ) on R

n

induces Riemannian metrics on the

following submanifolds.

(i) the m-dimensional sphere S

m

⊂ R

m+1

,

(ii) the tangent bundle TS

m

⊂ R

n

, where n = 2m + 2,

(iii) the m-dimensional torus T

m

⊂ R

n

, with n = 2m

(iv) the m-dimensional real projective space

RP

m

⊂ Sym(R

m+1

) ⊂ R

n

,

where n = (m + 2)(m+ 1)/2.

5. RIEMANNIAN MANIFOLDS 51

The vector space C

m×m

, of complex m m matrices, carries a

natural Euclidean metric g given by

g(Z, W) = Re(trace(

¯

Z

t

W)).

This induces metrics on the submanifolds of C

m×m

such as R

m×m

and

the classical Lie groups GL

m

(R), SL

m

(R), O(m), SO(m), GL

m

(C),

SL

m

(C), U(m) and SU(m).

Our next important step is to prove that every diﬀerentiable mani-

fold M can be equipped with a Riemannian metric g. For this we need

the following fact from topology.

Fact 5.9. Every locally compact Hausdorﬀ space with countable

basis is paracompact.

Corollary 5.10. Let (M, T ) be a topological manifold. Let the

collection (U

α

)

α∈I

be an open covering of M such that for each α ∈ I

the pair (U

α

, ψ

α

) is a chart on M. Then there exists

(i) a locally ﬁnite open reﬁnement (W

β

)

β∈J

such that for all β ∈ J,

W

β

is an open neighbourhood for a chart (W

β

, ψ

β

), and

(ii) a partition of unity (f

β

)

β∈J

such that support(f

β

) ⊂ W

β

.

Theorem 5.11. Let (M

m

,

ˆ

/) be a diﬀerentiable manifold. Then

there exists a Riemannian metric g on M.

Proof. For each point p ∈ M let (U

p

, φ

p

) ∈

ˆ

/ be a chart such that

p ∈ U

p

. Then (U

p

)

p∈M

is an open covering as in Corollary 5.10. Let

(W

β

)

β∈J

be a locally ﬁnite open reﬁnement, (W

β

, x

β

) be charts on M

and (f

β

)

β∈J

be a partition of unity such that support(f

β

) is contained

in W

β

. Let ¸, )

R

m be the Euclidean metric on R

m

. Then for β ∈ J

deﬁne g

β

: C

∞

2

(TM) →C

∞

0

(TM) by

g

β

(

∂

∂x

β

k

,

∂

∂x

β

l

)(p) =

_

f

β

(p) ¸e

k

, e

l

)

R

m if p ∈ W

β

0 if p / ∈ W

β

Note that at each point only ﬁnitely many of g

β

are non-zero. This

means that the well-deﬁned tensor g : C

∞

2

(TM) →C

∞

0

(TM) given by

g =

β∈J

g

β

is a Riemannian metric on M.

Deﬁnition 5.12. Let (M, g) and (N, h) be Riemannian manifolds.

A map φ : (M, g) → (N, h) is said to be conformal if there exists a

function λ : M →R such that

e

λ(p)

g

p

(X

p

, Y

p

) = h

φ(p)

(dφ

p

(X

p

), dφ

p

(Y

p

)),

52 5. RIEMANNIAN MANIFOLDS

for all X, Y ∈ C

∞

(TM) and p ∈ M. The function e

λ

is called the

conformal factor of φ. A conformal map with λ ≡ 0 is said to be

isometric. An isometric diﬀeomorphism is called an isometry.

Example 5.13. On the standard unit sphere S

m

we have an action

O(m + 1) S

m

→S

m

of the orthogonal group O(m + 1) given by

(A, x) →A x

where is the standard matrix multiplication. The following shows

that the O(m + 1)-action on S

m

is isometric

¸AX, AY ) = X

t

A

t

AY = X

t

Y = ¸X, Y ).

Example 5.14. Equip the orthogonal group O(m) as a submani-

fold of R

m×m

with the induced metric given by

¸X, Y ) = trace(X

t

Y ).

For x ∈ O(m) the left translation L

x

: O(m) →O(m) by x is given by

L

x

: y →xy. The tangent space T

y

O(m) of O(m) at y is

T

y

O(m) = ¦yX[ X

t

+ X = 0¦

and the diﬀerential (dL

x

)

y

: T

y

O(m) →T

xy

O(m) of L

x

is given by

(dL

x

)

y

: yX →xyX.

We then have

¸(dL

x

)

y

(yX), (dL

x

)

y

(yY ))

xy

= trace((xyX)

t

xyY )

= trace(X

t

y

t

x

t

xyY )

= trace(yX)

t

(yY ).

= ¸yX, yY )

y

.

This shows that the left translation L

x

: O(m) →O(m) is an isometry

for each x ∈ O(m).

Deﬁnition 5.15. Let G be a Lie group. A Riemannian metric g

on G is said to be left invariant if for each x ∈ G the left translation

L

x

: G →G is an isometry.

As for the orthogonal group O(m) an inner product on the tangent

space at the neutral element of any Lie group can be transported via

the left translations to obtain a left invariant Riemannian metric on

the group.

Proposition 5.16. Let G be a Lie group and ¸, )

e

be an inner

product on the tangent space T

e

G at the neutral element e. Then for

each x ∈ G the bilinear map g

x

(, ) : T

x

GT

x

G →R with

g

x

(X

x

, Y

x

) = ¸dL

x

−1(X

x

), dL

x

−1(Y

x

))

e

5. RIEMANNIAN MANIFOLDS 53

is an inner product on the tangent space T

x

G. The smooth tensor ﬁeld

g : C

∞

2

(TG) →C

∞

0

(G) given by

g : (X, Y ) →(g(X, Y ) : x →g

x

(X

x

, Y

x

))

is a left invariant Riemannian metric on G.

Proof. See Exercise 5.5.

We shall now equip the real projective space RP

m

with a Riemann-

ian metric.

Example 5.17. Let S

m

be the unit sphere in R

m+1

and Sym(R

m+1

)

be the linear space of symmetric real (m+1)(m+1) matrices equipped

with the metric g given by

g(A, B) =

1

8

trace(A

t

B).

As in Example 3.21 we deﬁne a map φ : S

m

→Sym(R

m+1

) by

φ : p →(ρ

p

: q →2¸q, p)p −q).

Let α, β : R → S

n

be two curves such that α(0) = p = β(0) and put

a = ˙ α(0), b =

˙

β(0). Then for γ ∈ ¦α, β¦ we have

dφ

p

( ˙ γ(0)) = (q →2¸q, ˙ γ(0))p + 2¸q, p) ˙ γ(0)).

If B is an orthonormal basis for R

m+1

, then

g(dφ

p

(a), dφ

p

(b)) = trace(dφ

p

(a)

t

dφ

p

(b))/8

=

q∈B

¸¸q, a)p +¸q, p)a, ¸q, b)p +¸q, p)b)/2

=

q∈B

¦¸p, p)¸a, q)¸q, b) +¸a, b)¸p, q)¸p, q)¦/2

= ¦¸a, b) +¸a, b)¦/2

= ¸a, b)

This proves that the immersion φ is isometric. In Example 3.21 we have

seen that the image φ(S

m

) can be identiﬁed with the real projective

space RP

m

. This inherits the induced metric from R

(m+1)×(m+1)

and

the map φ : S

m

→ RP

m

is what is called an isometric double cover of

RP

m

.

Long before John Nash became famous in Hollywood he proved

the next remarkable result in his paper The embedding problem for

Riemannian manifolds, Ann. of Math. 63 (1956), 20-63. It implies

that every Riemannian manifold can be realized as a submanifold of a

Euclidean space. The original proof of Nash was later simpliﬁed, see

54 5. RIEMANNIAN MANIFOLDS

for example Matthias Gunther, On the perturbation problem associated

to isometric embeddings of Riemannian manifolds, Annals of Global

Analysis and Geometry 7 (1989), 69-77.

Deep Result 5.18. For 3 ≤ r ≤ ∞ let (M, g) be a Riemannian

C

r

-manifold. Then there exists an isometric C

r

-embedding of (M, g)

into a Euclidean space R

n

.

We shall now see that parametrizations can be very useful tools for

studying the intrinsic geometry of a Riemannian manifold (M, g). Let

p be a point of M and

ˆ

ψ : U → M be a local parametrization of M

with q ∈ U and

ˆ

ψ(q) = p. The diﬀerential d

ˆ

ψ

q

: T

q

R

m

→ T

p

M is

bijective so there exist neighbourhoods U

q

of q and U

p

of p such that

the restriction ψ =

ˆ

ψ[

U

q

: U

q

→ U

p

is a diﬀeomorphism. On U

q

we

have the canonical frame ¦e

1

, . . . , e

m

¦ for TU

q

so ¦dψ(e

1

), . . . , dψ(e

m

)¦

is a local frame for TM over U

p

. We then deﬁne the pull-back metric

˜ g = ψ

∗

g on U

q

by

˜ g(e

k

, e

l

) = g(dψ(e

k

), dψ(e

l

)).

Then ψ : (U

q

, ˜ g) → (U

p

, g) is an isometry so the intrinsic geometry of

(U

q

, ˜ g) and that of (U

p

, g) are exactly the same.

Example 5.19. Let G be one of the classical Lie groups and e

be the neutral element of G. Let ¦X

1

, . . . , X

m

¦ be a basis for the Lie

algebra g of G. For x ∈ G deﬁne ψ

x

: R

m

→G by

ψ

x

: (t

1

, . . . , t

m

) →L

x

(

m

k=1

Exp(t

k

X

k

))

where L

x

: G →G is the left-translation given by L

x

(y) = xy. Then

(dψ

x

)

0

(e

k

) = X

k

(x)

for all k. This means that the diﬀerential (dψ

x

)

0

: T

0

R

m

→ T

x

G is an

isomorphism so there exist open neighbourhoods U

0

of 0 and U

x

of x

such that the restriction of ψ to U

0

is bijective onto its image U

x

and

hence a local parametrization of G around x.

We shall now study the normal bundle of a submanifold of a given

Riemannian manifold. This is an important example of the notion of

a vector bundle over a manifold.

Deﬁnition 5.20. Let (N, h) be a Riemannian manifold and M be

a submanifold of N. For a point p ∈ M we deﬁne the normal space

N

p

M of M at p by

N

p

M = ¦v ∈ T

p

N[ h

p

(v, w) = 0 for all w ∈ T

p

M¦.

5. RIEMANNIAN MANIFOLDS 55

For all p ∈ M we have the orthogonal decomposition

T

p

N = T

p

M ⊕N

p

M.

The normal bundle of M in N is deﬁned by

NM = ¦(p, v)[ p ∈ M, v ∈ N

p

M¦.

Example 5.21. Let S

m

be the unit sphere in R

m+1

equipped with

its standard Euclidean metric ¸, ). If p ∈ S

m

then the tangent space

T

p

S

m

of S

m

at p is

T

p

S

m

= ¦v ∈ R

m+1

[ ¸v, p) = 0¦

so the normal space N

p

S

m

of S

m

at p satisﬁes

N

p

S

m

= ¦λp ∈ R

m+1

[ λ ∈ R¦.

This shows that the normal bundle NS

m

of S

m

in R

m+1

is given by

NS

m

= ¦(p, λp) ∈ R

2m+2

[ p ∈ S

m

, λ ∈ R¦.

Theorem 5.22. Let (N

n

, h) be a Riemannian manifold and M

m

be a smooth submanifold of N. Then the normal bundle (NM, M, π)

is a smooth (n −m)-dimensional vector bundle over M.

Proof. See Exercise 5.7.

We shall now determine the normal bundle NO(m) of the orthog-

onal group O(m) as a submanifold of R

m×m

.

Example 5.23. The orthogonal group O(m) is a subset of the

linear space R

m×m

equipped with the Riemannian metric

¸X, Y ) = trace(X

t

Y )

inducing a left invariant metric on O(m). We have already seen that

the tangent space T

e

O(m) of O(m) at the neutral element e is

T

e

O(m) = ¦X ∈ R

m×m

[ X

t

+ X = 0¦

and that the tangent bundle TO(m) of O(m) is given by

TO(m) = ¦(x, xX)[ x ∈ O(m), X ∈ T

e

O(m)¦.

The space R

m×m

of real mm matrices has a linear decomposition

R

m×m

= Sym(R

m

) ⊕T

e

O(m)

and every element X ∈ R

m×m

can be decomposed X = X

⊤

+ X

⊥

in

its symmetric and skew-symmetric parts given by

X

⊤

= (X −X

t

)/2 and X

⊥

= (X +X

t

)/2.

If X ∈ T

e

O(m) and Y ∈ Sym(R

m

) then

¸X, Y ) = trace(X

t

Y )

56 5. RIEMANNIAN MANIFOLDS

= trace(Y

t

X)

= trace(XY

t

)

= trace(−X

t

Y )

= −¸X, Y ).

This means that the normal bundle NO(m) of O(m) in R

m×m

is given

by

NO(m) = ¦(x, xY )[ x ∈ O(m), Y ∈ Sym(R

m

)¦.

A given Riemannian metric g on M can be used to construct a

family of natural metrics on the tangent bundle TM of M. The best

known such examples are the Sasaki and Cheeger-Gromoll metrics. For

a detailed survey on the geometry of tangent bundles equipped with

these metrics we recommend the paper S. Gudmundsson, E. Kappos,

On the geometry of tangent bundles, Expo. Math. 20 (2002), 1-41.

5. RIEMANNIAN MANIFOLDS 57

Exercises

Exercise 5.1. Let m be a positive integer and φ : R

m

→C

m

be the

standard parametrization of the m-dimensional torus T

m

in C

m

given

by φ : (x

1

, . . . , x

m

) → (e

ix

1

, . . . , e

ix

m

). Prove that φ is an isometric

parametrization.

Exercise 5.2. Let m be a positive integer and

π

m

: (S

m

−¦(1, 0, . . . , 0)¦, ¸, )

R

m+1) →(R

m

,

4

(1 +[x[

2

)

2

¸, )

R

m)

be the stereographic projection given by

π

m

: (x

0

, . . . , x

m

) →

1

1 −x

0

(x

1

, . . . , x

m

).

Prove that π

m

is an isometry.

Exercise 5.3. Let B

2

1

(0) be the open unit disk in the complex plane

equipped with the hyperbolic metric

g(X, Y ) =

4

(1 −[z[

2

)

2

¸X, Y )

R

2.

Equip the upper half plane ¦z ∈ C[ Im(z) > 0¦ with the Riemannian

metric

g(X, Y ) =

1

Im(z)

2

¸X, Y )

R

2

and prove that the holomorphic function

π : B

2

1

(0) →¦z ∈ C[ Im(z) > 0¦

given by

π : z →

i +z

1 + iz

is an isometry.

Exercise 5.4. Equip the unitary group U(m) as a submanifold of

C

m×m

with the induced metric given by

¸Z, W) = Re(trace(

¯

Z

t

W)).

Show that for each z ∈ U(m) the left translation L

z

: U(m) → U(m)

given by L

z

: w →zw is an isometry.

Exercise 5.5. Find a proof for Proposition 5.16.

58 5. RIEMANNIAN MANIFOLDS

Exercise 5.6. Let m be a positive integer and GL

m

(R) be the

corresponding real general linear group. Let g, h be two Riemannian

metrics on GL

m

(R) deﬁned by

g

x

(xZ, xW) = trace((xZ)

t

xW), h

x

(xZ, xW) = trace(Z

t

W).

Further let ˆ g,

ˆ

h be the induced metrics on the special linear group

SL

m

(R) as a subset of GL

m

(R).

(i) Which of the metrics g, h, ˆ g,

ˆ

h are left-invariant?

(ii) Determine the normal space N

e

SL

m

(R) of SL

m

(R) in GL

m

(R)

with respect to g

(iii) Determine the normal bundle NSL

m

(R) of SL

m

(R) in GL

m

(R)

with respect to h.

Exercise 5.7. Find a proof for Theorem 5.22.

CHAPTER 6

The Levi-Civita Connection

In this chapter we introduce the Levi-Civita connection ∇ of a

Riemannian manifold (M, g). This is the most important example of

the general notion of a connection in a smooth vector bundle. We

deduce an explicit formula for the Levi-Civita connection for Lie groups

equipped with left-invariant metrics. We also give an example of a

connection in the normal bundle of a submanifold of a Riemannian

manifold and study its properties.

On the m-dimensional real vector space R

m

we have the well-known

diﬀerential operator

∂ : C

∞

(TR

m

) C

∞

(TR

m

) →C

∞

(TR

m

)

mapping a pair of vector ﬁelds X, Y on R

m

to the directional deriv-

ative ∂

X

Y of Y in the direction of X given by

(∂

X

Y )(x) = lim

t→0

Y (x + tX(x)) −Y (x)

t

.

The most fundamental properties of the operator ∂ are expressed by

the following. If λ, µ ∈ R, f, g ∈ C

∞

(R

m

) and X, Y, Z ∈ C

∞

(TR

m

)

then

(i) ∂

X

(λ Y + µ Z) = λ ∂

X

Y + µ ∂

X

Z,

(ii) ∂

X

(f Y ) = X(f) Y + f ∂

X

Y ,

(iii) ∂

(f X +g Y )

Z = f ∂

X

Z + g ∂

Y

Z.

The next result shows that the diﬀerential operator ∂ is compatible

with both the standard diﬀerentiable structure on R

m

and its Euclidean

metric.

Proposition 6.1. Let the real vector space R

m

be equipped with

the standard Euclidean metric ¸, ) and X, Y, Z ∈ C

∞

(TR

m

) be smooth

vector ﬁelds on R

m

. Then

(iv) ∂

X

Y −∂

Y

X = [X, Y ],

(v) X(¸Y, Z)) = ¸∂

X

Y , Z) +¸Y, ∂

X

Z).

59

60 6. THE LEVI-CIVITA CONNECTION

We shall now generalize the diﬀerential operator ∂ on the Euclidean

space R

m

to the so called Levi-Civita connection ∇ on a Riemannian

manifold (M, g). First we introduce the concept of a connection in a

smooth vector bundle.

Deﬁnition 6.2. Let (E, M, π) be a smooth vector bundle over M.

A connection on (E, M, π) is a map

ˆ

∇ : C

∞

(TM)C

∞

(E) →C

∞

(E)

such that

(i)

ˆ

∇

X

(λ v + µ w) = λ

ˆ

∇

X

v + µ

ˆ

∇

X

w,

(ii)

ˆ

∇

X

(f v) = X(f) v + f

ˆ

∇

X

v,

(iii)

ˆ

∇

(f X +g Y )

v = f

ˆ

∇

X

v + g

ˆ

∇

Y

v.

for all λ, µ ∈ R, X, Y ∈ C

∞

(TM), v, w ∈ C

∞

(E) and f, g ∈ C

∞

(M).

A section v ∈ C

∞

(E) of the vector bundle E is said to be parallel

with respect to the connection

ˆ

∇ if

ˆ

∇

X

v = 0

for all vector ﬁelds X ∈ C

∞

(TM).

Deﬁnition 6.3. Let M be a smooth manifold and

ˆ

∇ be a connec-

tion on the tangent bundle (TM, M, π). Then we deﬁne the torsion

T : C

∞

2

(TM) →C

∞

1

(TM) of

ˆ

∇ by

T(X, Y ) =

ˆ

∇

X

Y −

ˆ

∇

Y

X −[X, Y ],

where [, ] is the Lie bracket on C

∞

(TM). The connection

ˆ

∇ is said to

be torsion-free if its torsion T vanishes i.e.

[X, Y ] =

ˆ

∇

X

Y −

ˆ

∇

Y

X

for all X, Y ∈ C

∞

(TM).

Deﬁnition 6.4. Let (M, g) be a Riemannian manifold. Then a

connection

ˆ

∇ on the tangent bundle (TM, M, π) is said to be metric

or compatible with the Riemannian metric g if

X(g(Y, Z)) = g(

ˆ

∇

X

Y , Z) + g(Y,

ˆ

∇

X

Z)

for all X, Y, Z ∈ C

∞

(TM).

Let (M, g) be a Riemannian manifold and ∇ be a metric and

torsion-free connection on its tangent bundle (TM, M, π). Then it is

easily seen that the following equations hold

g(∇

X

Y , Z) = X(g(Y, Z)) −g(Y, ∇

X

Z),

g(∇

X

Y , Z) = g([X, Y ], Z) + g(∇

Y

X, Z)

6. THE LEVI-CIVITA CONNECTION 61

= g([X, Y ], Z) + Y (g(X, Z)) −g(X, ∇

Y

Z),

0 = −Z(g(X, Y )) + g(∇

Z

X, Y ) + g(X, ∇

Z

Y )

= −Z(g(X, Y )) + g(∇

X

Z + [Z, X], Y ) + g(X, ∇

Y

Z −[Y, Z]).

By adding these relations we yield

2 g(∇

X

Y , Z) = ¦X(g(Y, Z)) + Y (g(Z, X)) −Z(g(X, Y ))

+g(Z, [X, Y ]) + g(Y, [Z, X]) −g(X, [Y, Z])¦.

If ¦E

1

, . . . , E

m

¦ is a local orthonormal frame for the tangent bundle

then

∇

X

Y =

m

k=1

g(∇

X

Y , E

i

)E

i

.

As a direct consequence there exists at most one metric and torsion

free connection on the tangent bundle.

Deﬁnition 6.5. Let (M, g) be a Riemannian manifold then the

map ∇ : C

∞

(TM) C

∞

(TM) →C

∞

(TM) given by

2 g(∇

X

Y , Z) = ¦X(g(Y, Z)) + Y (g(X, Z)) −Z(g(X, Y ))

+g([Z, X], Y ) + g([Z, Y ], X) + g(Z, [X, Y ])¦.

is called the Levi-Civita connection on M.

It should be noted that the Levi-Civita connection is an intrinsic

object on (M, g) only depending on the diﬀerentiable structure of the

manifold and its Riemannian metric.

Proposition 6.6. Let (M, g) be a Riemannian manifold. Then the

Levi-Civita connection ∇ is a connection on the tangent bundle TM of

M.

Proof. It follows from Deﬁnition 3.5, Theorem 4.15 and the fact

that g is a tensor ﬁeld that

g(∇

X

(λ Y

1

+ µ Y

2

), Z) = λ g(∇

X

Y

1

, Z) + µ g(∇

X

Y

2

, Z)

and

g(∇

Y

1

+ Y

2

X, Z) = g(∇

Y

1

X, Z) + g(∇

Y

2

X, Z)

for all λ, µ ∈ R and X, Y

1

, Y

2

, Z ∈ C

∞

(TM). Furthermore we have for

all f ∈ C

∞

(M)

2 g(∇

X

fY , Z)

= ¦X(f g(Y, Z)) + f Y (g(X, Z)) −Z(f g(X, Y ))

+f g([Z, X], Y ) + g([Z, f Y ], X) + g(Z, [X, f Y ])¦

62 6. THE LEVI-CIVITA CONNECTION

= ¦X(f) g(Y, Z) + f X(g(Y, Z)) + f Y (g(X, Z))

−Z(f) g(X, Y ) −f Z(g(X, Y )) +f g([Z, X], Y )

+g(Z(f) Y +f [Z, Y ], X) + g(Z, X(f) Y +f [X, Y ])¦

= 2 ¦X(f) g(Y, Z) +f g(∇

X

Y , Z)¦

= 2 g(X(f) Y + f ∇

X

Y , Z)

and

2 g(∇

f X

Y , Z)

= ¦f X(g(Y, Z)) +Y (f g(X, Z)) −Z(f g(X, Y ))

+g([Z, f X], Y ) + f g([Z, Y ], X) + g(Z, [f X, Y ])¦

= ¦f X(g(Y, Z)) +Y (f) g(X, Z) + f Y (g(X, Z))

−Z(f) g(X, Y ) −f Z(g(X, Y ))

+g(Z(f) X, Y ) +f g([Z, X], Y )

+f g([Z, Y ], X) + f g(Z, [X, Y ]) −g(Z, Y (f) X)¦

= 2 f g(∇

X

Y , Z).

This proves that ∇ is a connection on the tangent bundle (TM, M, π).

**The next result is called the Fundamental theorem of Riemannian
**

geometry.

Theorem 6.7. Let (M, g) be a Riemannian manifold. Then the

Levi-Civita connection is a unique metric and torsion free connection

on the tangent bundle (TM, M, π).

Proof. The diﬀerence g(∇

X

Y , Z) − g(∇

Y

X, Z) equals twice the

skew-symmetric part (w.r.t the pair (X, Y )) of the right hand side of

the equation in Deﬁnition 6.5. This is the same as

=

1

2

¦g(Z, [X, Y ]) −g(Z, [Y, X])¦ = g(Z, [X, Y ]).

This proves that the Levi-Civita connection is torsion-free.

The sum g(∇

X

Y , Z) + g(∇

X

Z, Y ) equals twice the symmetric part

(w.r.t the pair (Y, Z)) on the right hand side of Deﬁnition 6.5. This is

exactly

=

1

2

¦X(g(Y, Z)) + X(g(Z, Y ))¦ = X(g(Y, Z)).

This shows that the Levi-Civita connection is compatible with the Rie-

mannian metric g on M.

6. THE LEVI-CIVITA CONNECTION 63

A vector ﬁeld X ∈ C

∞

(TM) on (M, g) induces the ﬁrst order

covariant derivative

∇

X

: C

∞

(TM) →C

∞

(TM)

in the direction of X by

∇

X

: Y →∇

X

Y .

Deﬁnition 6.8. Let G be a Lie group. For a left invariant vector

ﬁeld Z ∈ g we deﬁne the map ad(Z) : g →g by

ad(Z) : X →[Z, X].

Proposition 6.9. Let (G, g) be a Lie group equipped with a left

invariant metric. Then the Levi-Civita connection ∇ satisﬁes

g(∇

X

Y , Z) =

1

2

¦g([X, Y ], Z) + g(ad(Z)(X), Y ) + g(X, ad(Z)(Y ))¦

for all X, Y, Z ∈ g. In particular, if for all Z ∈ g the map ad(Z) is

skew symmetric with respect to g then

∇

X

Y =

1

2

[X, Y ].

Proof. See Exercise 6.2.

Proposition 6.10. Let G be one of the classical compact Lie groups

O(m), SO(m), U(m) or SU(m) equipped with the left-invariant metric

g(Z, W) = Re (trace (

¯

Z

t

W)).

Then for each X ∈ g the operator ad(X) : g →g is skew symmetric.

Proof. See Exercise 6.3.

Example 6.11. Let (M, g) be a Riemannian manifold with Levi-

Civita connection ∇. Further let (U, x) be local coordinates on M and

put X

i

= ∂/∂x

i

∈ C

∞

(TU). Then ¦X

1

, . . . , X

m

¦ is a local frame of

TM on U. For (U, x) we deﬁne the Christoﬀel symbols Γ

k

ij

: U →R

of the connection ∇ with respect to (U, x) by

m

k=1

Γ

k

ij

X

k

= ∇

X

i

X

j

.

On the subset x(U) of R

m

we deﬁne the metric ˜ g by

˜ g(e

i

, e

j

) = g

ij

= g(X

i

, X

j

).

The diﬀerential dx is bijective so Proposition 4.17 implies that

dx([X

i

, X

j

]) = [dx(X

i

), dx(X

j

)] = [e

i

, e

j

] = 0

64 6. THE LEVI-CIVITA CONNECTION

and hence [X

i

, X

j

] = 0. From the deﬁnition of the Levi-Civita connec-

tion we now yield

m

k=1

Γ

k

ij

g

kl

= ¸

m

k=1

Γ

k

ij

X

k

, X

l

)

= ¸∇

X

i

X

j

, X

l

)

=

1

2

¦X

i

¸X

j

, X

l

) + X

j

¸X

l

, X

i

) −X

l

¸X

i

, X

j

)¦

=

1

2

¦

∂g

jl

∂x

i

+

∂g

li

∂x

j

−

∂g

ij

∂x

l

¦.

If g

kl

= (g

−1

)

kl

then

Γ

k

ij

=

1

2

m

l=1

g

kl

¦

∂g

jl

∂x

i

+

∂g

li

∂x

j

−

∂g

ij

∂x

l

¦.

Deﬁnition 6.12. Let N be a smooth manifold, M be a submanifold

of N and

˜

X ∈ C

∞

(TM) be a vector ﬁeld on M. Let U be an open

subset of N such that U ∩ M ,= ∅. A local extension of

˜

X to U is

a vector ﬁeld X ∈ C

∞

(TU) such that

˜

X

p

= X

p

for all p ∈ U ∩ M. If

U = N then X is called a global extension.

Fact 6.13. Let (N, h) be a Riemannian manifold and M be a sub-

manifold of N. Then vector ﬁelds

˜

X ∈ C

∞

(TM) and

˜

Y ∈ C

∞

(NM)

have global extensions X, Y ∈ C

∞

(TN).

Let (N, h) be a Riemannian manifold and M be a submanifold

equipped with the induced metric g. Let Z ∈ C

∞

(TN) be a vector

ﬁeld on N and

˜

Z = Z[

M

: M →TN be the restriction of Z to M. Note

that

˜

Z is not necessarily an element of C

∞

(TM) i.e. a vector ﬁeld on

the submanifold M. For each p ∈ M the tangent vector

˜

Z

p

∈ T

p

N can

be decomposed

˜

Z

p

=

˜

Z

⊤

p

+

˜

Z

⊥

p

in a unique way into its tangential part (

˜

Z

p

)

⊤

∈ T

p

M and its normal

part (

˜

Z

p

)

⊥

∈ N

p

M. For this we write

˜

Z =

˜

Z

⊤

+

˜

Z

⊥

.

Let

˜

X,

˜

Y ∈ C

∞

(TM) be vector ﬁelds on M and X, Y ∈ C

∞

(TN)

be their extensions to N. If p ∈ M then (∇

X

Y )

p

only depends on the

value X

p

=

˜

X

p

and the value of Y along some curve γ : (−ǫ, ǫ) → N

such that γ(0) = p and ˙ γ(0) = X

p

=

˜

X

p

. For this see Remark 7.3. Since

X

p

∈ T

p

M we may choose the curve γ such that the image γ((−ǫ, ǫ))

is contained in M. Then

˜

Y

γ(t)

= Y

γ(t)

for t ∈ (−ǫ, ǫ). This means

that (∇

X

Y )

p

only depends on

˜

X

p

and the value of

˜

Y along γ, hence

6. THE LEVI-CIVITA CONNECTION 65

independent of the way

˜

X and

˜

Y are extended. This shows that the

following maps

˜

∇ and B are well deﬁned.

Deﬁnition 6.14. Let (N, h) be a Riemannian manifold and M be

a submanifold equipped with the induced metric g. Then we deﬁne

two operators

˜

∇ : C

∞

(TM) C

∞

(TM) →C

∞

(TM)

and

B : C

∞

(TM) C

∞

(TM) →C

∞

(NM)

by

˜

∇

˜

X

˜

Y = (∇

X

Y )

⊤

and B(

˜

X,

˜

Y ) = (∇

X

Y )

⊥

,

where X, Y ∈ C

∞

(TN) are any extensions of

˜

X,

˜

Y .

The operator B is called the second fundamental form of M in

(N, h). It is symmetric and hence tensorial in both its arguments, see

Exercise 6.7.

Theorem 6.15. Let (N, h) be a Riemannian manifold and M be a

submanifold of N with the induced metric g. Then the operator

˜

∇ is

the Levi-Civita connection of the submanifold (M, g).

Proof. See Exercise 6.8.

The Levi-Civita connection on (N, h) induces a metric connection

¯

∇ on the normal bundle NM of M in N as follows.

Proposition 6.16. Let (N, h) be a Riemannian manifold and M

be a submanifold with the induced metric g. Let X, Y ∈ C

∞

(TN) be

vector ﬁelds extending

˜

X ∈ C

∞

(TM) and

˜

Y ∈ C

∞

(NM). Then the

map

¯

∇ : C

∞

(TM) C

∞

(NM) →C

∞

(NM) given by

¯

∇

˜

X

˜

Y = (∇

X

Y )

⊥

is a well-deﬁned connection on the normal bundle NM satisfying

˜

X(h(

˜

Y ,

˜

Z)) = h(

¯

∇

˜

X

˜

Y ,

˜

Z) + h(

˜

Y ,

¯

∇

˜

X

˜

Z)

for all

˜

X ∈ C

∞

(TM) and

˜

Y ,

˜

Z ∈ C

∞

(NM).

Proof. See Exercise 6.9.

66 6. THE LEVI-CIVITA CONNECTION

Exercises

Exercise 6.1. Let M be a smooth manifold and

ˆ

∇ be a con-

nection on the tangent bundle (TM, M, π). Prove that the torsion

T : C

∞

2

(TM) →C

∞

1

(TM) of

ˆ

∇ is a tensor ﬁeld of type (2, 1).

Exercise 6.2. Find a proof for Proposition 6.9.

Exercise 6.3. Find a proof for Proposition 6.10.

Exercise 6.4. Let G be the 3-dimensional subgroup of GL

3

(R)

given by

G = ¦

_

_

e

z

0 x

0 e

z

y

0 0 1

_

_

[ p = (x, y, z) ∈ R

3

¦.

Let X, Y, Z ∈ g be left-invariant vector ﬁelds on G such that

X

e

=

∂

∂x

[

p=0

, Y

e

=

∂

∂y

[

p=0

, Z

e

=

∂

∂z

[

p=0

.

Show that

[X, Y ] = 0, [Z, X] = X, [Z, Y ] = Y.

Let g be a left-invariant Riemannian metric on G such that ¦X, Y, Z¦ is

an orthonormal basis for the Lie algebra g. Calculate the vector ﬁelds

∇

X

Y , ∇

Y

X, ∇

X

Z, ∇

Z

X, ∇

Y

Z, ∇

Z

Y .

Exercise 6.5. Let SO(m) be the special orthogonal group equipped

with the metric

¸X, Y ) =

1

2

trace(X

t

Y ).

Prove that ¸, ) is left-invariant and that for any left-invariant vector

ﬁelds X, Y ∈ so(m) we have

∇

X

Y =

1

2

[X, Y ].

Let A, B, C be elements of the Lie algebra so(3) with

A

e

=

_

_

0 −1 0

1 0 0

0 0 0

_

_

, B

e

=

_

_

0 0 −1

0 0 0

1 0 0

_

_

, C

e

=

_

_

0 0 0

0 0 −1

0 1 0

_

_

.

Prove that ¦A, B, C¦ is an orthonormal basis for so(3) and calculate

∇

A

B, ∇

B

C, ∇

C

A.

6. THE LEVI-CIVITA CONNECTION 67

Exercise 6.6. Let SL

2

(R) be the real special linear group equipped

with the metric

¸X, Y )

p

=

1

2

trace((p

−1

X)

t

(p

−1

Y )).

Let A, B, C be elements of the Lie algebra sl

2

(R) with

A

e

=

_

0 −1

1 0

_

, B

e

=

_

0 1

1 0

_

, C

e

=

_

1 0

0 −1

_

.

Prove that ¦A, B, C¦ is an orthonormal basis for sl

2

(R) and calculate

∇

A

B, ∇

B

C, ∇

C

A.

Exercise 6.7. Let (N, h) be a Riemannian manifold with Levi-

Civita connection ∇ and (M, g) be a submanifold with the induced

metric. Prove that the second fundamental form B of M in N is

symmetric and tensorial in both its arguments.

Exercise 6.8. Find a proof for Theorem 6.15.

Exercise 6.9. Find a proof for Proposition 6.16.

CHAPTER 7

Geodesics

In this chapter we introduce the notion of a geodesic on a Riemann-

ian manifold (M, g). This is a solution to a second order non-linear

system of ordinary diﬀerential equations. We show that geodesics are

solutions to two diﬀerent variational problems. They are critical points

to the so called energy functional and furthermore locally shortest paths

between their endpoints.

Deﬁnition 7.1. Let M be a smooth manifold and (TM, M, π) be

its tangent bundle. A vector ﬁeld X along a curve γ : I → M is

a curve X : I → TM such that π ◦ X = γ. By C

∞

γ

(TM) we denote

the set of all smooth vector ﬁelds along γ. For X, Y ∈ C

∞

γ

(TM) and

f ∈ C

∞

(I) we deﬁne the operations + and by

(i) (X +Y )(t) = X(t) + Y (t),

(ii) (f X)(t) = f(t) X(t).

This turns (C

∞

γ

(TM), +, ) into a module over C

∞

(I) and a real vector

space over the constant functions in particular. For a given smooth

curve γ : I → M in M the smooth vector ﬁeld X : I → TM with

X : t →(γ(t), ˙ γ(t)) is called the tangent ﬁeld along γ.

The next result gives a rule for diﬀerentiating a vector ﬁeld along a

given curve and shows how this is related to the Levi-Civita connection.

Proposition 7.2. Let (M, g) be a smooth Riemannian manifold

and γ : I →M be a curve in M. Then there exists a unique operator

D

dt

: C

∞

γ

(TM) →C

∞

γ

(TM)

such that for all λ, µ ∈ R and f ∈ C

∞

(I),

(i) D(λ X + µ Y )/dt = λ (DX/dt) + µ (DY/dt),

(ii) D(f Y )/dt = df/dt Y + f (DY /dt), and

(iii) for each t

0

∈ I there exists an open subinterval J

0

of I such that

t

0

∈ J

0

and if X ∈ C

∞

(TM) is a vector ﬁeld with X

γ(t)

= Y (t)

for all t ∈ J

0

then

_

DY

dt

_

(t

0

) = (∇

˙ γ

X)

γ(t

0

)

.

69

70 7. GEODESICS

Proof. Let us ﬁrst prove the uniqueness, so for the moment we

assume that such an operator exists. For a point t

0

∈ I choose a chart

(U, x) on M and open interval J

0

such that t

0

∈ J

0

, γ(J

0

) ⊂ U and put

X

i

= ∂/∂x

i

∈ C

∞

(TU). Then a vector ﬁeld Y along γ can be written

in the form

Y (t) =

m

k=1

α

k

(t)

_

X

k

_

γ(t)

for some functions α

k

∈ C

∞

(J

0

). The second condition means that

(1)

_

DY

dt

_

(t) =

m

k=1

α

k

(t)

_

DX

k

dt

_

γ(t)

+

m

k=1

˙ α

k

(t)

_

X

k

_

γ(t)

.

Let x ◦ γ(t) = (γ

1

(t), . . . , γ

m

(t)) then

˙ γ(t) =

m

k=1

˙ γ

k

(t)

_

X

k

_

γ(t)

and the third condition for D/dt imply that

(2)

_

DX

j

dt

_

γ(t)

= (∇

˙ γ

X

j

)

γ(t)

=

m

k=1

˙ γ

k

(t)(∇

X

k

X

j

)

γ(t)

.

Together equations (1) and (2) give

(3)

_

DY

dt

_

(t) =

m

k=1

¦ ˙ α

k

(t) +

m

i,j=1

Γ

k

ij

(γ(t)) ˙ γ

i

(t)α

j

(t)¦

_

X

k

_

γ(t)

.

This shows that the operator D/dt is uniquely determined.

It is easily seen that if we use equation (3) for deﬁning an operator

D/dt then it satisﬁes the necessary conditions of Proposition 7.2. This

proves the existence of the operator D/dt.

Remark 7.3. It follows from the fact that the Levi-Civita connec-

tion is tensorial in its ﬁrst argument i.e.

∇

f Z

X = f ∇

Z

X

and the equation

_

DY

dt

_

(t

0

) = (∇

˙ γ

X)

γ(t

0

)

in Proposition 7.2 that the value (∇

Z

X)

p

of ∇

Z

X at p only depends

on the value of Z

p

of Z at p and the values of Y along some curve γ

satisfying γ(0) = p and ˙ γ(0) = Z

p

. This allows us to use the notation

∇

˙ γ

Y for DY/dt.

7. GEODESICS 71

The Levi-Civita connection can now be used to deﬁne a parallel

vector ﬁeld and a geodesic on a manifold as solutions to ordinary dif-

ferential equations

Deﬁnition 7.4. Let (M, g) be a Riemannian manifold and γ : I →

M be a C

1

-curve. A vector ﬁeld X along γ is said to be parallel if

∇

˙ γ

X = 0.

A C

2

-curve γ : I →M is said to be a geodesic if its tangent ﬁeld ˙ γ is

parallel along γ i.e.

∇

˙ γ

˙ γ = 0.

The next result shows that for given initial values at a point p ∈ M

we get a parallel vector ﬁeld globally deﬁned along any curve through

that point.

Theorem 7.5. Let (M, g) be a Riemannian manifold and I = (a, b)

be an open interval on the real line R. Further let γ : I → M be a

smooth curve, t

0

∈ I and X

0

∈ T

γ(t

0

)

M. Then there exists a unique

parallel vector ﬁeld Y along γ such that X

0

= Y (t

0

).

Proof. Without loss of generality we may assume that the image

of γ lies in a chart (U, x). We put X

i

= ∂/∂x

i

so on the interval I the

tangent ﬁeld ˙ γ is represented in our local coordinates by

˙ γ(t) =

m

i=1

ρ

i

(t)

_

X

i

_

γ(t)

with some functions ρ

i

∈ C

∞

(I). Similarly let Y be a vector ﬁeld along

γ represented by

Y (t) =

m

j=1

σ

j

(t)

_

X

j

_

γ(t)

.

Then

_

∇

˙ γ

Y

_

(t) =

m

j=1

¦ ˙ σ

j

(t)

_

X

j

_

γ(t)

+ σ

j

(t)

_

∇

˙ γ

X

j

_

γ(t)

¦

=

m

k=1

¦ ˙ σ

k

(t) +

m

i,j=1

σ

j

(t)ρ

i

(t)Γ

k

ij

(γ(t))¦

_

X

k

_

γ(t)

.

This implies that the vector ﬁeld Y is parallel i.e. ∇

˙ γ

Y ≡ 0 if and

only if the following ﬁrst order linear system of ordinary diﬀerential

72 7. GEODESICS

equations is satisﬁed:

˙ σ

k

(t) +

m

i,j=1

σ

j

(t)ρ

i

(t)Γ

k

ij

(γ(t)) = 0

for all k = 1, . . . , m. It follows from Fact 7.6 that to each initial value

σ(t

0

) = (v

1

, . . . , v

m

) ∈ R

m

with

Y

0

=

m

k=1

v

k

_

X

k

_

γ(t

0

)

there exists a unique solution σ = (σ

1

, . . . , σ

m

) to the above system.

This gives us the unique parallel vector ﬁeld Y

Y (t) =

m

k=1

σ

k

(t)

_

X

k

_

γ(t)

along I.

The following result is the well-known theorem of Picard-Lindel¨of.

Fact 7.6. Let f : U →R

n

be a continuous map deﬁned on an open

subset U of R R

n

and L ∈ R

+

such that

[f(t, x) −f(t, y)[ ≤ L [x −y[

for all (t, x), (t, y) ∈ U. If (t

0

, x

0

) ∈ U then there exists a unique local

solution x : I →R

n

to the following initial value problem

x

′

(t) = f(t, x(t)), x(t

0

) = x

0

.

Lemma 7.7. Let (M, g) be a Riemannian manifold, γ : I → M

be a smooth curve and X, Y be parallel vector ﬁelds along γ. Then the

function g(X, Y ) : I → R given by t → g

γ(t)

(X

γ(t)

, Y

γ(t)

) is constant.

In particular, if γ is a geodesic then g( ˙ γ, ˙ γ) is constant along γ.

Proof. Using the fact that the Levi-Civita connection is metric

we obtain

d

dt

(g(X, Y )) = g(∇

˙ γ

X, Y ) + g(X, ∇

˙ γ

Y ) = 0.

This proves that the function g(X, Y ) is constant along γ.

The following result on parallel vector ﬁelds is a useful tool in Rie-

mannian geometry. It will be employed in Chapter 9.

Proposition 7.8. Let (M, g) be a Riemannian manifold, p ∈ M

and ¦v

1

, . . . , v

m

¦ be an orthonormal basis for the tangent space T

p

M.

Let γ : I →M be a smooth curve such that γ(0) = p and X

1

, . . . , X

m

be

parallel vector ﬁelds along γ such that X

k

(0) = v

k

for k = 1, 2, . . . , m.

7. GEODESICS 73

Then the set ¦X

1

(t), . . . , X

m

(t)¦ is a orthonormal basis for the tangent

space T

γ(t)

M for all t ∈ I.

Proof. This is a direct consequence of Lemma 7.7.

For the important geodesic equation we have the following local

existence result.

Theorem 7.9. Let (M, g) be a Riemannian manifold. If p ∈ M

and v ∈ T

p

M then there exists an open interval I = (−ǫ, ǫ) and a

unique geodesic γ : I →M such that γ(0) = p and ˙ γ(0) = v.

Proof. Let (U, x) be a chart on M such that p ∈ U and put

X

i

= ∂/∂x

i

. For an open interval J and a C

2

-curve γ : J →U we put

γ

i

= x

i

◦ γ : J →R. The curve x ◦ γ : J →R

m

is C

2

so we have

(dx)

γ(t)

( ˙ γ(t)) =

m

i=1

˙ γ

i

(t)e

i

giving

˙ γ(t) =

m

i=1

˙ γ

i

(t)

_

X

i

_

γ(t)

.

By diﬀerentiation we then obtain

∇

˙ γ

˙ γ =

m

j=1

∇

˙ γ

_

˙ γ

j

(t)

_

X

j

_

γ(t)

_

=

m

j=1

¦¨ γ

j

(t)

_

X

j

_

γ(t)

+

m

i=1

˙ γ

j

(t) ˙ γ

i

(t)

_

∇

X

i

X

j

_

γ(t)

¦

=

m

k=1

¦¨ γ

k

(t) +

m

i,j=1

˙ γ

j

(t) ˙ γ

i

(t)Γ

k

ij

◦ γ(t)¦

_

X

k

_

γ(t)

.

Hence the curve γ is a geodesic if and only if

¨ γ

k

(t) +

m

i,j=1

˙ γ

j

(t) ˙ γ

i

(t)Γ

k

ij

(γ(t)) = 0

for all k = 1, . . . , m. It follows from Fact 7.10 that for initial values

q

0

= x(p) and w

0

= (dx)

p

(v) there exists an open interval (−ǫ, ǫ) and

a unique solution (γ

1

, . . . , γ

m

) satisfying the initial conditions

(γ

1

(0), . . . , γ

m

(0)) = q

0

and ( ˙ γ

1

(0), . . . , ˙ γ

m

(0)) = w

0

.

**The following result is a second order consequence of the well-known
**

theorem of Picard-Lindel¨of.

74 7. GEODESICS

Fact 7.10. Let f : U → R

n

be a continuous map deﬁned on an

open subset U of R R

n

and L ∈ R

+

such that

[f(t, x) −f(t, y)[ ≤ L [x −y[

for all (t, x), (t, y) ∈ U. If (t

0

, x

0

) ∈ U and x

1

∈ R

n

then there exists a

unique local solution x : I →R

n

to the following initial value problem

x

′′

(t) = f(t, x(t)), x(t

0

) = x

0

, x

′

(t

0

) = x

1

.

The Levi-Civita connection ∇ on a given Riemannian manifold

(M, g) is an inner object completely determined by the metric g. Hence

the same applies for the condition

∇

˙ γ

˙ γ = 0

for a given curve γ : I →M. This means that the image of a geodesic

under a local isometry is again a geodesic.

Let E

m

= (R

m

, ¸, )

R

m) be the Euclidean space. For the trivial chart

id

R

m : R

m

→ R

m

the metric is given by g

ij

= δ

ij

, so Γ

k

ij

= 0 for all

i, j, k = 1, . . . , m. This means that γ : I → R

m

is a geodesic if and

only if ¨ γ(t) = 0 or equivalently γ(t) = t a+b for some a, b ∈ R

m

. This

proves that the geodesics are the straight lines.

Deﬁnition 7.11. A geodesic γ : I → (M, g) in a Riemannian

manifold is said to be maximal if it cannot be extended to a geodesic

deﬁned on an interval J strictly containing I. The manifold (M, g)

is said to be complete if for each point (p, v) ∈ TM there exists a

geodesic γ : R →M deﬁned on the whole of R such that γ(0) = p and

˙ γ(0) = v.

Proposition 7.12. Let (N, h) be a Riemannian manifold with Levi-

Civita connection ∇ and M be a submanifold equipped with the induced

metric g. A curve γ : I →M is a geodesic in M if and only if

(∇

˙ γ

˙ γ)

⊤

= 0.

Proof. Following Theorem 6.15 the Levi-Civita connection

˜

∇ on

(M, g) satisﬁes

˜

∇

˙ γ

˙ γ = (∇

˙ γ

˙ γ)

⊤

.

Example 7.13. Let E

m+1

be the (m + 1)-dimensional Euclidean

space and S

m

be the unit sphere in E

m+1

with the induced metric. At

a point p ∈ S

m

the normal space N

p

S

m

of S

m

in E

m+1

is simply the

line generated by p. If γ : I →S

m

is a curve on the sphere, then

˜

∇

˙ γ

˙ γ = ¨ γ

⊤

= ¨ γ − ¨ γ

⊥

= ¨ γ −¸¨ γ, γ)γ.

7. GEODESICS 75

This shows that γ is a geodesic on the sphere S

m

if and only if

(4) ¨ γ = ¸¨ γ, γ)γ.

For a point (p, v) ∈ TS

m

deﬁne the curve γ = γ

(p,v)

: R →S

m

by

γ : t →

_

p if v = 0

cos([v[t) p + sin([v[t) v/[v[ if v ,= 0.

Then one easily checks that γ(0) = p, ˙ γ(0) = v and that γ satisﬁes the

geodesic equation (4). This shows that the non-constant geodesics on

S

m

are precisely the great circles and the sphere is complete.

Example 7.14. Let Sym(R

m+1

) be equipped with the metric

¸A, B) =

1

8

trace(A

t

B).

Then we know that the map φ : S

m

→Sym(R

m+1

) with

φ : p →(2pp

t

−e)

is an isometric immersion and that the image φ(S

m

) is isometric to

the m-dimensional real projective space RP

m

. This means that the

geodesics on RP

m

are exactly the images of geodesics on S

m

. This

shows that the real projective spaces are complete.

Deﬁnition 7.15. Let (M, g) be a Riemannian manifold and γ :

I →M be a C

r

-curve on M. A variation of γ is a C

r

-map

Φ : (−ǫ, ǫ) I →M

such that for all s ∈ I, Φ

0

(s) = Φ(0, s) = γ(s). If the interval is

compact i.e. of the form I = [a, b], then the variation Φ is called

proper if for all t ∈ (−ǫ, ǫ), Φ

t

(a) = γ(a) and Φ

t

(b) = γ(b).

Deﬁnition 7.16. Let (M, g) be a Riemannian manifold and γ :

I → M be a C

2

-curve on M. For every compact interval [a, b] ⊂ I we

deﬁne the energy functional E

[a,b]

by

E

[a,b]

(γ) =

1

2

_

b

a

g( ˙ γ(t), ˙ γ(t))dt.

A C

2

-curve γ : I → M is called a critical point for the energy

functional if every proper variation Φ of γ[

[a,b]

satisﬁes

d

dt

(E

[a,b]

(Φ

t

))[

t=0

= 0.

We shall now prove that geodesics can be characterized as the crit-

ical points of the energy functional.

76 7. GEODESICS

Theorem 7.17. A C

2

-curve γ : I = [a, b] → M is a critical point

for the energy functional if and only if it is a geodesic.

Proof. For a C

2

-map Φ : (−ǫ, ǫ) I → M, Φ : (t, s) → Φ(t, s)

we deﬁne the vector ﬁelds X = dΦ(∂/∂s) and Y = dΦ(∂/∂t) along Φ.

The following shows that the vector ﬁelds X and Y commute.

∇

X

Y −∇

Y

X = [X, Y ]

= [dΦ(∂/∂s), dΦ(∂/∂t)]

= dΦ([∂/∂s, ∂/∂t])

= 0,

since [∂/∂s, ∂/∂t] = 0. We now assume that Φ is a proper variation of

γ. Then

d

dt

(E

[a,b]

(Φ

t

)) =

1

2

d

dt

(

_

b

a

g(X, X)ds)

=

1

2

_

b

a

d

dt

(g(X, X))ds

=

_

b

a

g(∇

Y

X, X)ds

=

_

b

a

g(∇

X

Y , X)ds

=

_

b

a

(

d

ds

(g(Y, X)) −g(Y, ∇

X

X))ds

= [g(Y, X)]

b

a

−

_

b

a

g(Y, ∇

X

X)ds.

The variation is proper, so Y (a) = Y (b) = 0. Furthermore X(0, s) =

∂Φ/∂s(0, s) = ˙ γ(s), so

d

dt

(E

[a,b]

(Φ

t

))[

t=0

= −

_

b

a

g(Y (0, s), (∇

˙ γ

˙ γ)(s))ds.

The last integral vanishes for every proper variation Φ of γ if and only

if ∇

˙ γ

˙ γ = 0.

A geodesic γ : I → (M, g) is a special case of what is called a

harmonic map φ : (M, g) → (N, h) between Riemannian manifolds.

Other examples are conformal immersions ψ : (M

2

, g) →(N, h) which

parametrize the so called minimal surfaces in (N, h). For a reference on

harmonic maps see H. Urakawa, Calculus of Variations and Harmonic

Maps, Translations of Mathematical Monographs 132, AMS(1993).

7. GEODESICS 77

Let (M

m

, g) be an m-dimensional Riemannian manifold, p ∈ M

and

S

m−1

p

= ¦v ∈ T

p

M[ g

p

(v, v) = 1¦

be the unit sphere in the tangent space T

p

M at p. Then every point

w ∈ T

p

M ¸ ¦0¦ can be written as w = r

w

v

w

, where r

w

= [w[ and

v

w

= w/[w[ ∈ S

m−1

p

. For v ∈ S

m−1

p

let γ

v

: (−α

v

, β

v

) → M be

the maximal geodesic such that α

v

, β

v

∈ R

+

∪ ¦∞¦, γ

v

(0) = p and

˙ γ

v

(0) = v. It can be shown that the real number

ǫ

p

= inf¦α

v

, β

v

[ v ∈ S

m−1

p

¦

is positive so the open ball

B

m

ǫ

p

(0) = ¦v ∈ T

p

M[ g

p

(v, v) < ǫ

2

p

¦

is non-empty. The exponential map exp

p

: B

m

ǫ

p

(0) → M at p is

deﬁned by

exp

p

: w →

_

p if w = 0

γ

v

w

(r

w

) if w ,= 0.

Note that for v ∈ S

m−1

p

the line segment λ

v

: (−ǫ

p

, ǫ

p

) → T

p

M

with λ

v

: t → t v is mapped onto the geodesic γ

v

i.e. locally we have

γ

v

= exp

p

◦λ

v

. One can prove that the map exp

p

is smooth and it

follows from its deﬁnition that the diﬀerential

d(exp

p

)

0

: T

p

M →T

p

M

is the identity map for the tangent space T

p

M. Then the inverse

mapping theorem tells us that there exists an r

p

∈ R

+

such that if

U

p

= B

m

r

p

(0) and V

p

= exp

p

(U

p

) then exp

p

[

U

p

: U

p

→V

p

is a diﬀeomor-

phism parametrizing the open subset V

p

of M.

The next result shows that the geodesics are locally the shortest

paths between their endpoints.

Theorem 7.18. Let (M, g) be a Riemannian manifold. Then the

geodesics are locally the shortest path between their end points.

Proof. Let p ∈ M, U = B

m

r

(0) in T

p

M and V = exp

p

(U) be such

that the restriction

φ = exp

p

[

U

: U →V

of the exponential map at p is a diﬀeomorphism. We deﬁne a metric ˜ g

on U such that for each X, Y ∈ C

∞

(TU) we have

˜ g(X, Y ) = g(dφ(X), dφ(Y )).

This turns φ : (U, ˜ g) → (V, g) into an isometry. It then follows from

the construction of the exponential map, that the geodesics in (U, ˜ g)

78 7. GEODESICS

through the point 0 = φ

−1

(p) are exactly the lines λ

v

: t →t v where

v ∈ T

p

M.

Now let q ∈ B

m

r

(0) ¸ ¦0¦ and λ

q

: [0, 1] → B

m

r

(0) be the curve

λ

q

: t → t q. Further let σ : [0, 1] → U be any curve such that

σ(0) = 0 and σ(1) = q. Along the curve σ we deﬁne the vector ﬁeld X

with X : t → σ(t) and the tangent ﬁeld ˙ σ : t → ˙ σ(t) to σ. Then the

radial component ˙ σ

rad

of ˙ σ is the orthogonal projection of ˙ σ onto the

line generated by X i.e.

˙ σ

rad

: t →

˜ g( ˙ σ(t), X(t))

˜ g(X(t), X(t))

X(t).

Then it is easily checked that

[ ˙ σ

rad

(t)[ =

[˜ g( ˙ σ(t), X(t))[

[X(t)[

and

d

dt

[X(t)[ =

d

dt

_

˜ g(X(t), X(t)) =

˜ g( ˙ σ(t), X(t))

[X(t)[

.

Combining these two relations we yield

[ ˙ σ

rad

(t)[ ≥

d

dt

[X(t)[.

This means that

L(σ) =

_

1

0

[ ˙ σ(t)[dt

≥

_

1

0

[ ˙ σ

rad

(t)[dt

≥

_

1

0

d

dt

[X(t)[dt

= [X(1)[ −[X(0)[

= [q[

= L(λ

q

).

This proves that in fact γ is the shortest path connecting p and q.

Deﬁnition 7.19. Let (N, h) be a Riemannian manifold and M be

a submanifold with the induced metric g. Then the mean curvature

vector ﬁeld of M in N is the smooth section H : M → NM of the

normal bundle NM given by

H =

1

m

trace B =

1

m

m

k=1

B(X

k

, X

k

).

7. GEODESICS 79

Here B is the second fundamental form of M in N and ¦X

1

, . . . , X

m

¦

is any local orthonormal frame for the tangent bundle TM of M. The

submanifold M is said to be minimal in N if H ≡ 0 and totally

geodesic in N if B ≡ 0.

Proposition 7.20. Let (N, h) be a Riemannian manifold and M be

a submanifold equipped with the induced metric g. Then the following

conditions are equivalent:

(i) M is totally geodesic in N

(ii) if γ : I → M is a curve, then the following conditions are

equivalent

(a) γ : I →M is a geodesic in M,

(b) γ : I →M is a geodesic in N.

Proof. The result is a direct consequence of the following decom-

position formula

∇

˙ γ

˙ γ = (∇

˙ γ

˙ γ)

⊤

+ (∇

˙ γ

˙ γ)

⊥

=

˜

∇

˙ γ

˙ γ + B( ˙ γ, ˙ γ).

**Proposition 7.21. Let (N, h) be a Riemannian manifold and M be
**

a complete submanifold of N. For a point (p, v) of the tangent bundle

TM let γ

(p,v)

: I → N be the maximal geodesic in N with γ(0) = p

and ˙ γ(0) = v. Then M is totally geodesic in (N, h) if and only if

γ

(p,v)

(I) ⊂ M for all (p, v) ∈ TM.

Proof. See Exercise 7.3.

Corollary 7.22. Let (N, h) be a Riemannian manifold, p ∈ N

and V be an m-dimensional linear subspace of the tangent space T

p

N

of N at p. Then there exists (locally) at most one totally geodesic

submanifold M of (N, h) such that T

p

M = V .

Proof. See Exercise 7.4.

Proposition 7.23. Let (N, h) be a Riemannian manifold and M be

a submanifold of N which is the ﬁxpoint set of an isometry φ : N →N.

Then M is totally geodesic in N.

Proof. Let p ∈ M, v ∈ T

p

M and γ : I → N be the maximal

geodesic with γ(0) = p and ˙ γ(0) = v. The map φ : N → N is an

isometry so φ ◦ γ : I → N is a geodesic. The uniqueness result of

Theorem 7.9, φ(γ(0)) = γ(0) and dφ( ˙ γ(0)) = ˙ γ(0) then imply that

φ(γ) = γ. Hence the image of the geodesic γ : I → N is contained in

M, so following Proposition 7.21 the submanifold M is totally geodesic

in N.

80 7. GEODESICS

Corollary 7.24. Let n < m be positive integers. Then the n-

dimensional sphere

S

n

= ¦(x, 0) ∈ R

n+1

R

m−n

[ [x[

2

= 1¦

is a totally geodesic submanifold of

S

m

= ¦(x, y) ∈ R

n+1

R

m−n

[ [x[

2

+[y[

2

= 1¦.

Proof. The statement is a direct consequence of the fact that S

n

is the ﬁxpoint set of the isometry φ : S

m

→ S

m

of S

m

with (x, y) →

(x, −y).

Corollary 7.25. Let H

m

be the m-dimensional hyperbolic space

modelled on the upper half space R

+

R

m−1

equipped with the Rie-

mannian metric

g(X, Y ) =

1

x

2

1

¸X, Y )

R

m,

where x = (x

1

, . . . , x

m

) ∈ H

m

. Then the n-dimensional hyperbolic

space

H

n

= ¦(x, 0) ∈ H

m

[ x ∈ R

n

¦

is totally geodesic in H

m

.

Proof. See Exercise 7.6.

7. GEODESICS 81

Exercises

Exercise 7.1. The result of Exercise 5.3 shows that the two di-

mensional hyperbolic disc H

2

introduced in Example 5.6 is isometric

to the upper half plane M = (¦(x, y) ∈ R

2

[ y ∈ R

+

¦ equipped with the

Riemannian metric

g(X, Y ) =

1

y

2

¸X, Y )

R

2.

Use your local library to ﬁnd all geodesics in (M, g).

Exercise 7.2. Let n be a positive integer and O(n) be the orthog-

onal group equipped with the standard left-invariant metric

g(A, B) = trace(A

t

B).

Prove that a C

2

-curve γ : (−ǫ, ǫ) →O(n) is a geodesic if and only if

γ

t

¨ γ = ¨ γ

t

γ.

Exercise 7.3. Find a proof for Proposition 7.21.

Exercise 7.4. Find a proof for Corollary 7.22.

Exercise 7.5. For the real parameter θ ∈ (0, π/2) deﬁne the 2-

dimensional torus T

2

θ

by

T

2

θ

= ¦(cos θe

iα

, sin θe

iβ

) ∈ S

3

[ α, β ∈ R¦.

Determine for which θ ∈ (0, π/2) the torus T

2

θ

is a minimal submanifold

of the 3-dimensional sphere

S

3

= ¦(z

1

, z

2

) ∈ C

2

[ [z

1

[

2

+[z

2

[

2

= 1¦.

Exercise 7.6. Find a proof for Corollary 7.25.

Exercise 7.7. Determine the totally geodesic submanifolds of the

m-dimensional real projective space RP

m

.

Exercise 7.8. Let the orthogonal group O(n) be equipped with

the left-invariant metric

g(A, B) = trace(A

t

B)

and let K be a Lie subgroup of O(n). Prove that K is totally geodesic

in O(n).

CHAPTER 8

The Riemann Curvature Tensor

In this chapter we introduce the Riemann curvature tensor and

the notion of sectional curvature of a Riemannian manifold. These

generalize the Gaussian curvature playing a central role in classical

diﬀerential geometry.

We prove that the Euclidean spaces, the standard spheres and the

hyperbolic spaces all have constant sectional curvature. We determine

the Riemannian curvature tensor for manifolds of constant sectional

curvature and also for an important class of Lie groups. We then derive

the important Gauss equation comparing the sectional curvatures of a

submanifold and that of its ambient space.

Let (M, g) be a Riemannian manifold and ∇ be its Levi-Civita

connection. Then to each vector ﬁeld X ∈ C

∞

(TM) we have the ﬁrst

order covariant derivative

∇

X

: C

∞

(TM) →C

∞

(TM)

in the direction of X satisfying

∇

X

: Z →∇

X

Z.

We shall now generalize this and introduce the covariant derivative of

tensor ﬁelds of type (r, 0) or (r, 1).

As motivation, let us assume that A is a tensor ﬁeld of type (2, 1).

If we diﬀerentiate A(Y, Z) in the direction of X applying the naive

”product rule”

∇

X

(A(Y, Z)) = (∇

X

A)(Y, Z) + A(∇

X

Y ) + A(∇

X

Z)

we get

(∇

X

A)(Y, Z) = ∇

X

(A(Y, Z)) −A(∇

X

Y ) −A(∇

X

Z),

where ∇

X

A is the ”covariant derivative” of the tensor ﬁeld A in the

direction of X. This naive idea turns out to be very useful and leads

to the following formal deﬁnition.

Deﬁnition 8.1. Let (M, g) be a Riemannian manifold with Levi-

Civita connection ∇. For a tensor ﬁeld A : C

∞

r

(TM) → C

∞

0

(TM) of

83

84 8. THE RIEMANN CURVATURE TENSOR

type (r, 0) we deﬁne its covariant derivative

∇A : C

∞

r+1

(TM) →C

∞

0

(TM)

by

∇A : (X, X

1

, . . . , X

r

) →(∇

X

A)(X

1

, . . . , X

r

) =

X(A(X

1

, . . . , X

r

)) −

r

k=1

A(X

1

, . . . , X

k−1

, ∇

X

X

k

, X

k+1

, . . . , X

r

).

A tensor ﬁeld A of type (r, 0) is said to be parallel if ∇A ≡ 0.

The following result can be seen as, yet another, compatibility of

the Levi-Civita connection ∇ of (M, g) with the Riemannian metric g.

Proposition 8.2. Let (M, g) be a Riemannian manifold. Then the

metric g is a parallel tensor ﬁeld of type (2, 0).

Proof. See Exercise 8.1.

Let (M, g) be a Riemannian manifold. Then its Levi-Civita con-

nection ∇ is tensorial in its ﬁrst argument i.e. if X, Y ∈ C

∞

(TM) and

f, g ∈ C

∞

(M) then

∇

(fX + gY )

Z = f∇

X

Z + g∇

Y

Z.

This means that a vector ﬁeld Z ∈ C

∞

(TM) on M induces a natural

tensor ﬁeld Z : C

∞

1

(TM) →C

∞

1

(TM) of type (1, 1) given by

Z : X →∇

X

Z.

Deﬁnition 8.3. Let (M, g) be a Riemannian manifold with Levi-

Civita connection ∇. For a tensor ﬁeld B : C

∞

r

(TM) → C

∞

1

(TM) of

type (r, 1) we deﬁne its covariant derivative

∇B : C

∞

r+1

(TM) →C

∞

1

(TM)

by

∇B : (X, X

1

, . . . , X

r

) →(∇

X

B)(X

1

, . . . , X

r

) =

∇

X

(B(X

1

, . . . , X

r

)) −

r

k=1

B(X

1

, . . . , X

k−1

, ∇

X

X

k

, X

k+1

, . . . , X

r

).

A tensor ﬁeld B of type (r, 1) is said to be parallel if ∇B ≡ 0.

Deﬁnition 8.4. Let X, Y ∈ C

∞

(TM) be two vector ﬁelds on the

Riemannian manifold (M, g) with Levi-Civita connection ∇. Then the

second order covariant derivative

∇

2

X, Y

: C

∞

(TM) →C

∞

(TM)

8. THE RIEMANN CURVATURE TENSOR 85

is deﬁned by

∇

2

X, Y

: Z →(∇

X

Z)(Y ),

where Z is the natural (1, 1)-tensor ﬁeld induced by Z ∈ C

∞

(TM).

As a direct consequence of Deﬁnitions 8.3 and 8.4 we see that if

X, Y, Z ∈ C

∞

(TM) then the second order covariant derivative ∇

2

X, Y

satisﬁes

∇

2

X, Y

Z = ∇

X

(Z(Y )) −Z(∇

X

Y ) = ∇

X

∇

Y

Z −∇

∇

X

Y

Z.

Deﬁnition 8.5. Let (M, g) be a Riemannian manifold with Levi-

Civita connection ∇. Then we deﬁne its Riemann curvature operator

R : C

∞

(TM) C

∞

(TM) C

∞

(TM) →C

∞

(TM)

as twice the skew-symmetric part of the second covariant derivative ∇

2

i.e.

R(X, Y )Z = ∇

2

X, Y

Z −∇

2

Y, X

Z.

The next remarkable result shows that the curvature operator is a

tensor ﬁeld.

Theorem 8.6. Let (M, g) be a Riemannian manifold with Levi-

Civita connection ∇. Then the curvature R : C

∞

3

(TM) →C

∞

1

(TM)

R(X, Y )Z = ∇

X

∇

Y

Z −∇

Y

∇

X

Z −∇

[X, Y ]

Z

is a tensor ﬁeld on M of type (3, 1).

Proof. See Exercise 8.2.

The reader should note that the Riemann curvature tensor R is

an intrinsic object since it only depends on the intrinsic Levi-Civita

connection ∇. The following result shows that the curvature tensor

has many nice properties of symmetry.

Proposition 8.7. Let (M, g) be a Riemannian manifold. For vec-

tor ﬁelds X, Y, Z, W ∈ C

∞

(TM) on M we then have

(i) R(X, Y )Z = −R(Y, X)Z,

(ii) g(R(X, Y )Z, W) = −g(R(X, Y )W, Z),

(iii) R(X, Y )Z + R(Z, X)Y + R(Y, Z)X = 0,

(iv) g(R(X, Y )Z, W) = g(R(Z, W)X, Y ),

(v) 6 R(X, Y )Z = R(X, Y + Z)(Y + Z) −R(X, Y −Z)(Y −Z)

+ R(X +Z, Y )(X +Z) −R(X −Z, Y )(X −Z).

Proof. See Exercise 8.3.

86 8. THE RIEMANN CURVATURE TENSOR

Part (iii) of Proposition 8.7 is the so called ﬁrst Bianchi identity.

The second Bianchi identity is a similar result concerning the covariant

derivative ∇R of the curvature tensor. This will not be treated here.

Let (M, g) be a Riemannian manifold and p ∈ M. Then a section

V at p is a 2-dimensional subspace of the tangent space T

p

M. The set

G

2

(T

p

M) = ¦V [ V is a section of T

p

M¦

of sections is called the Grassmannian of 2-planes at p.

Lemma 8.8. Let (M, g) be a Riemannian manifold, p ∈ M and

X, Y, Z, W ∈ T

p

M be tangent vectors at p such that the two sections

span

R

¦X, Y ¦ and span

R

¦Z, W¦ are identical. Then

g(R(X, Y )Y, X)

[X[

2

[Y [

2

−g(X, Y )

2

=

g(R(Z, W)W, Z)

[Z[

2

[W[

2

−g(Z, W)

2

.

Proof. See Exercise 8.4.

This leads to the following natural deﬁnition of the sectional cur-

vature.

Deﬁnition 8.9. Let (M, g) be a Riemannian manifold and p ∈ M

Then the function K

p

: G

2

(T

p

M) →R given by

K

p

: span

R

¦X, Y ¦ →

g(R(X, Y )Y, X)

[X[

2

[Y [

2

−g(X, Y )

2

is called the sectional curvature at p. We often write K(X, Y ) for

K(span

R

¦X, Y ¦).

Deﬁnition 8.10. Let (M, g) be a Riemannian manifold, p ∈ M

and K

p

: G

2

(T

p

M) → R be the sectional curvature at p. Then we

deﬁne the functions δ, ∆ : M →R by

δ : p → min

V ∈G

2

(T

p

M)

K

p

(V ) and ∆ : p → max

V ∈G

2

(T

p

M)

K

p

(V ).

The Riemannian manifold (M, g) is said to be

(i) of positive curvature if δ(p) ≥ 0 for all p,

(ii) of strictly positive curvature if δ(p) > 0 for all p,

(iii) of negative curvature if ∆(p) ≤ 0 for all p,

(iv) of strictly negative curvature if ∆(p) < 0 for all p,

(v) of constant curvature if δ = ∆ is constant,

(vi) ﬂat if δ ≡ ∆ ≡ 0.

The next result shows how the curvature tensor can be expressed

in terms of local coordinates.

8. THE RIEMANN CURVATURE TENSOR 87

Proposition 8.11. Let (M, g) be a Riemannian manifold and let

(U, x) be local coordinates on M. For i, j, k, l = 1, . . . , m put

X

i

=

∂

∂x

i

, g

ij

= g(X

i

, X

j

) and R

ijkl

= g(R(X

i

, X

j

)X

k

, X

l

).

Then

R

ijkl

=

m

s=1

g

sl

_

∂Γ

s

jk

∂x

i

−

∂Γ

s

ik

∂x

j

+

m

r=1

¦Γ

r

jk

Γ

s

ir

−Γ

r

ik

Γ

s

jr

¦

_

,

where Γ

k

ij

are the Christoﬀel symbols of the Levi-Civita connection ∇

of (M, g) with respect to (U, x).

Proof. Using the fact that [X

i

, X

j

] = 0 we obtain

R(X

i

, X

j

)X

k

= ∇

X

i

∇

X

j

X

k

−∇

X

j

∇

X

i

X

k

=

m

s=1

¦∇

X

i

(Γ

s

jk

X

s

) −∇

X

j

(Γ

s

ik

X

s

)¦

=

m

s=1

_

∂Γ

s

jk

∂x

i

X

s

+

m

r=1

Γ

s

jk

Γ

r

is

X

r

−

∂Γ

s

ik

∂x

j

X

s

−

m

r=1

Γ

s

ik

Γ

r

js

X

r

_

=

m

s=1

_

∂Γ

s

jk

∂x

i

−

∂Γ

s

ik

∂x

j

+

m

r=1

¦Γ

r

jk

Γ

s

ir

−Γ

r

ik

Γ

s

jr

¦

_

X

s

.

**For the m-dimensional vector space R
**

m

equipped with the Eu-

clidean metric ¸, )

R

m the set ¦∂/∂x

1

, . . . , ∂/∂x

m

¦ is a global frame for

the tangent bundle TR

m

. In this situation we have g

ij

= δ

ij

, so Γ

k

ij

≡ 0

by Example 6.11. This implies that R ≡ 0 so E

m

is ﬂat.

Example 8.12. The standard sphere S

m

has constant sectional

curvature +1 (see Exercises 8.7 and 8.8) and the hyperbolic space H

m

has constant sectional curvature −1 (see Exercise 8.9).

Our next aim is a formula for the curvature tensor for manifolds of

constant sectional curvature. This we present in Corollary 8.16. First

we need some preparations.

Lemma 8.13. Let (M, g) be a Riemannian manifold, p ∈ M and

Y ∈ T

p

M. Then the map

˜

Y : T

p

M →T

p

M given by

˜

Y : X →R(X, Y )Y

is a symmetric endomorphism of the tangent space T

p

M.

88 8. THE RIEMANN CURVATURE TENSOR

Proof. For Z ∈ T

p

M we have

g(

˜

Y (X), Z) = g(R(X, Y )Y, Z) = g(R(Y, Z)X, Y )

= g(R(Z, Y )Y, X) = g(X,

˜

Y (Z)).

**For a tangent vector Y ∈ T
**

p

M with [Y [ = 1 let ^(Y ) be the normal

space to Y

^(Y ) = ¦X ∈ T

p

M[ g(X, Y ) = 0¦.

The fact that

˜

Y (Y ) = 0 and Lemma 8.13 ensure the existence of an

orthonormal basis of eigenvectors X

1

, . . . , X

m−1

for the restriction of

the symmetric endomorphism

˜

Y to ^(Y ). Without loss of generality,

we can assume that the corresponding eigenvalues satisfy

λ

1

(p) ≤ ≤ λ

m−1

(p).

If X ∈ ^(Y ), [X[ = 1 and

˜

Y (X) = λX then

K

p

(X, Y ) = g(R(X, Y )Y, X) = g(

˜

Y (X), X) = λ.

This means that the eigenvalues satisfy

δ(p) ≤ λ

1

(p) ≤ ≤ λ

m−1

(p) ≤ ∆(p).

Deﬁnition 8.14. Let (M, g) be a Riemannian manifold. Then

deﬁne the smooth tensor ﬁeld R

1

: C

∞

3

(TM) → C

∞

1

(TM) of type

(3, 1) by

R

1

(X, Y )Z = g(Y, Z)X −g(X, Z)Y.

Proposition 8.15. Let (M, g) be a smooth Riemannian manifold

and X, Y, Z be vector ﬁelds on M. Then

(i) [R(X, Y )Y −

δ+∆

2

R

1

(X, Y )Y [ ≤

1

2

(∆−δ)[X[[Y [

2

(ii) [R(X, Y )Z −

δ+∆

2

R

1

(X, Y )Z[ ≤

2

3

(∆−δ)[X[[Y [[Z[

Proof. Without loss of generality we can assume that [X[ = [Y [ =

[Z[ = 1. If X = X

⊥

+ X

⊤

with X

⊥

⊥ Y and X

⊤

is a multiple of Y

then R(X, Y )Z = R(X

⊥

, Y )Z and [X

⊥

[ ≤ [X[ so we can also assume

that X ⊥ Y . Then R

1

(X, Y )Y = ¸Y, Y )X −¸X, Y )Y = X.

The ﬁrst statement follows from the fact that the symmetric endo-

morphism of T

p

M with

X →¦R(X, Y )Y −

∆ +δ

2

X¦

restricted to ^(Y ) has eigenvalues in the interval [

δ−∆

2

,

∆−δ

2

].

8. THE RIEMANN CURVATURE TENSOR 89

It is easily checked that the operator R

1

satisﬁes the conditions of

Proposition 8.7 and hence D = R−

∆+δ

2

R

1

as well. This implies that

6 D(X, Y )Z = D(X, Y + Z)(Y + Z) −D(X, Y −Z)(Y −Z)

+ D(X + Z, Y )(X + Z) −D(X −Z, Y )(X −Z).

The second statement then follows from

6[D(X, Y )Z[ ≤

1

2

(∆−δ)¦[X[([Y +Z[

2

+[Y −Z[

2

)

+[Y [([X + Z[

2

+[X −Z[

2

)¦

=

1

2

(∆−δ)¦2[X[([Y [

2

+[Z[

2

) + 2[Y [([X[

2

+[Z[

2

)¦

= 4(∆−δ).

**As a direct consequence of Proposition 8.15 we have the following
**

useful result.

Corollary 8.16. Let (M, g) be a Riemannian manifold of constant

curvature κ. Then the curvature tensor R is given by

R(X, Y )Z = κ (g(Y, Z)X −g(X, Z)Y ).

Proof. The result follows directly from κ = δ = ∆.

Proposition 8.17. Let (G, g) be a Lie group equipped with a left-

invariant metric such that for all X ∈ g the endomorphism

ad(X) : g →g

is skew-symmetric with respect to g. Then for any left-invariant vector

ﬁelds X, Y, Z ∈ g the curvature tensor R is given by

R(X, Y )Z = −

1

4

[[X, Y ], Z].

Proof. See Exercise 8.6.

We shall now prove the important Gauss equation comparing the

curvature tensors of a submanifold and its ambient space in terms of

the second fundamental form.

Theorem 8.18. Let (N, h) be a Riemannian manifold and M be a

submanifold of N equipped with the induced metric g. Let X, Y, Z, W ∈

C

∞

(TN) be vector ﬁelds extending

˜

X,

˜

Y ,

˜

Z,

˜

W ∈ C

∞

(TM). Then

g(

˜

R(

˜

X,

˜

Y )

˜

Z,

˜

W) −h(R(X, Y )Z, W)

= h(B(

˜

Y ,

˜

Z), B(

˜

X,

˜

W)) −h(B(

˜

X,

˜

Z), B(

˜

Y ,

˜

W)).

90 8. THE RIEMANN CURVATURE TENSOR

Proof. Using the deﬁnitions of the curvature tensors R,

˜

R, the

Levi-Civita connection

˜

∇ and the second fundamental form of

˜

M in

M we obtain

g(

˜

R(

˜

X,

˜

Y )

˜

Z,

˜

W)

= g(

˜

∇

˜

X

˜

∇

˜

Y

˜

Z −

˜

∇

˜

Y

˜

∇

˜

X

˜

Z −

˜

∇

[

˜

X,

˜

Y ]

˜

Z,

˜

W)

= h((∇

X

(∇

Y

Z −B(Y, Z)))

⊤

−h(∇

Y

(∇

X

Z −B(X, Z)))

⊤

, W)

−h((∇

[X, Y ]

Z −B([X, Y ], Z))

⊤

, W)

= h(∇

X

∇

Y

Z −∇

Y

∇

X

Z −∇

[X, Y ]

Z, W)

−h(∇

X

(B(Y, Z)), W) +∇

Y

(B(X, Z)), W)

= h(R(X, Y )Z, W)

h((B(Y, Z)), ∇

X

W) −h(B(X, Z)), ∇

Y

W)

= h(R(X, Y )Z, W)

+h(B(Y, Z), B(X, W)) −h(B(X, Z), B(Y, W)).

**We shall now apply the Gauss equation to the classical situation of
**

a regular surface Σ as a submanifold of the 3-dimensional Euclidean

space R

3

. Let ¦

˜

X,

˜

Y ¦ be a local orthonormal frame for the tangent

bundle TΣ around a point p ∈ Σ and

˜

N be the local Gauss map with

˜

N =

˜

X

˜

Y . If X, Y, N are local extensions of

˜

X,

˜

Y ,

˜

N then the second

fundamental form B of Σ in R

3

satisﬁes

B(

˜

X,

˜

Y ) = < ∂

X

Y , N > N

= − < Y, ∂

X

N > N

= − < Y, dN(X) > N

= < Y, S

p

(X) > N,

where S

p

: T

p

Σ → T

p

Σ is the shape operator at p. If we now apply

the fact that R

3

is ﬂat, the Gauss equation tells us that the sectional

curvature K(

˜

X,

˜

Y ) of Σ at p satisﬁes

K(

˜

X,

˜

Y ) = <

˜

R(

˜

X,

˜

Y )

˜

Y ,

˜

X >

= < B(

˜

Y ,

˜

Y ), B(

˜

X,

˜

X) > − < B(

˜

X,

˜

Y ), B(

˜

Y ,

˜

X) >

= det S

p

.

In other word, the sectional curvature K(

˜

X,

˜

Y ) is the determinant of

the shape operator S

p

at p i.e. the classical Gaussian curvature.

8. THE RIEMANN CURVATURE TENSOR 91

As a direct consequence of the Gauss equation we have the following

useful result, see for example Exercises 8.8 and 8.9.

Corollary 8.19. Let (N, h) be a Riemannian manifold and (M, g)

be a totally geodesic submanifold of N. Let X, Y, Z, W ∈ C

∞

(TN) be

vector ﬁelds extending

˜

X,

˜

Y ,

˜

Z,

˜

W ∈ C

∞

(TM). Then

g(

˜

R(

˜

X,

˜

Y )

˜

Z,

˜

W) = h(R(X, Y )Z, W).

We conclude this chapter by deﬁning the Ricci and scalar curvatures

of a Riemannian manifold. These are obtained by taking traces over the

curvature tensor and play an important role in Riemannian geometry.

Deﬁnition 8.20. Let (M, g) be a Riemannian manifold, then

(i) the Ricci operator r : C

∞

1

(TM) →C

∞

1

(M) is deﬁned by

r(X) =

m

i=1

R(X, e

i

)e

i

,

(ii) the Ricci curvature Ric : C

∞

2

(TM) →C

∞

0

(TM) by

Ric(X, Y ) =

m

i=1

g(R(X, e

i

)e

i

, Y ), and

(iii) the scalar curvature s ∈ C

∞

(M) by

s =

m

j=1

Ric(e

j

, e

j

) =

m

j=1

m

i=1

g(R(e

i

, e

j

)e

j

, e

i

).

Here ¦e

1

, . . . , e

m

¦ is any local orthonormal frame for the tangent bun-

dle.

Corollary 8.21. Let (M, g) be a Riemannian manifold of constant

sectional curvature κ. Then the following holds

s = m (m−1) κ.

Proof. Let ¦e

1

, . . . , e

m

¦ be an orthonormal basis, then Corollary

8.16 implies that

Ric(e

j

, e

j

) =

m

i=1

g(R(e

j

, e

i

)e

i

, e

j

)

=

m

i=1

g(κ(g(e

i

, e

i

)e

j

−g(e

j

, e

i

)e

i

), e

j

)

= κ(

m

i=1

g(e

i

, e

i

)g(e

j

, e

j

) −

m

i=1

g(e

i

, e

j

)g(e

i

, e

j

))

92 8. THE RIEMANN CURVATURE TENSOR

= κ(

m

i=1

1 −

m

i=1

δ

ij

) = (m−1) κ.

To obtain the formula for the scalar curvature s we only have to mul-

tiply the constant Ricci curvature Ric(e

j

, e

j

) by m.

For further reading on diﬀerent notions of curvature we recommend

the interesting book, Wolfgang K¨ uhnel, Diﬀerential Geometry: Curves

- Surfaces - Manifolds, AMS (2002).

8. THE RIEMANN CURVATURE TENSOR 93

Exercises

Exercise 8.1. Let (M, g) be a Riemannian manifold. Prove that

the tensor ﬁeld g of type (2, 0) is parallel with respect to the Levi-Civita

connection.

Exercise 8.2. Let (M, g) be a Riemannian manifold. Prove that

the curvature R is a tensor ﬁeld of type (3, 1).

Exercise 8.3. Find a proof for Proposition 8.7.

Exercise 8.4. Find a proof for Lemma 8.8.

Exercise 8.5. Let R

m

and C

m

be equipped with their standard

Euclidean metric g given by

g(z, w) = Re

m

k=1

z

k

¯ w

k

and let

T

m

= ¦z ∈ C

m

[ [z

1

[ = ... = [z

m

[ = 1¦

be the m-dimensional torus in C

m

with the induced metric. Find an

isometric immersion φ : R

m

→T

m

, determine all geodesics on T

m

and

prove that the torus is ﬂat.

Exercise 8.6. Find a proof for Proposition 8.17.

Exercise 8.7. Let the Lie group S

3

∼

= SU(2) be equipped with the

metric

g(Z, W) =

1

2

Re(trace(

¯

Z

t

W)).

(i) Find an orthonormal basis for T

e

SU(2).

(ii) Prove that (SU(2), g) has constant sectional curvature +1.

Exercise 8.8. Let S

m

be the unit sphere in R

m+1

equipped with

the standard Euclidean metric ¸, )

R

m+1. Use the results of Corollaries

7.24, 8.19 and Exercise 8.7 to prove that (S

m

, ¸, )

R

m+1) has constant

sectional curvature +1.

Exercise 8.9. Let H

m

be the m-dimensional hyperbolic space mod-

elled on the upper half space R

+

R

m−1

equipped with the Riemannian

metric

g(X, Y ) =

1

x

2

1

¸X, Y )

R

m,

94 8. THE RIEMANN CURVATURE TENSOR

where x = (x

1

, . . . , x

m

) ∈ H

m

. For k = 1, . . . , m let the vector ﬁelds

X

k

∈ C

∞

(TH

m

) be given by

(X

k

)

x

= x

1

∂

∂x

k

and deﬁne the operation ∗ on H

m

by

(α, x) ∗ (β, y) = (α β, α y + x).

Prove that

(i) (H

m

, ∗) is a Lie group,

(ii) the vector ﬁelds X

1

, . . . , X

m

are left-invariant,

(iii) [X

k

, X

l

] = 0 and [X

1

, X

k

] = X

k

for k, l = 2, ..., m,

(iv) the metric g is left-invariant,

(v) (H

m

, g) has constant curvature −1.

Compare with Exercises 6.4 and 7.1.

CHAPTER 9

Curvature and Local Geometry

This chapter is devoted to the study of the local geometry of Rie-

mannian manifolds and how this is controlled by the curvature tensor.

For this we introduce the notion of a Jacobi ﬁeld which is a useful

tool in diﬀerential geometry. With this at hand we yield a funda-

mental comparison result describing the curvature dependence of local

distances.

Let (M, g) be a Riemannian manifold. By a smooth 1-parameter

family of geodesics we mean a C

∞

-map

Φ : (−ǫ, ǫ) I →M

such that the curve γ

t

: I →M given by γ

t

: s →Φ(t, s) is a geodesic for

all t ∈ (−ǫ, ǫ). The variable t ∈ (−ǫ, ǫ) is called the family parameter

of Φ.

Proposition 9.1. Let (M, g) be a Riemannian manifold and Φ :

(−ǫ, ǫ) I → M be a 1-parameter family of geodesics. Then for each

t ∈ (−ǫ, ǫ) the vector ﬁeld J

t

: I →C

∞

(TM) along γ

t

given by

J

t

(s) =

∂Φ

∂t

(t, s)

satisﬁes the second order ordinary diﬀerential equation

∇

˙ γ

t

∇

˙ γ

t

J

t

+R(J

t

, ˙ γ

t

) ˙ γ

t

= 0.

Proof. Along Φ we put X(t, s) = ∂Φ/∂s and J(t, s) = ∂Φ/∂t.

The fact that [∂/∂t, ∂/∂s] = 0 implies that

[J, X] = [dΦ(∂/∂t), dΦ(∂/∂s)] = dΦ([∂/∂t, ∂/∂s]) = 0.

Since Φ is a family of geodesics we have ∇

X

X = 0 and the deﬁnition

of the curvature tensor then gives

R(J, X)X = ∇

J

∇

X

X −∇

X

∇

J

X −∇

[J, X]

X

= −∇

X

∇

J

X

= −∇

X

∇

X

J.

95

96 9. CURVATURE AND LOCAL GEOMETRY

Hence for each t ∈ (−ǫ, ǫ) we have

∇

˙ γ

t

∇

˙ γ

t

J

t

+R(J

t

, ˙ γ

t

) ˙ γ

t

= 0.

**Deﬁnition 9.2. Let (M, g) be a Riemannian manifold, γ : I →M
**

be a geodesic and X = ˙ γ. A C

2

vector ﬁeld J along γ is called a

Jacobi ﬁeld if

(5) ∇

X

∇

X

J +R(J, X)X = 0

along γ. We denote the space of all Jacobi ﬁelds along γ by ¸

γ

(TM).

We shall now give an example of a 1-parameter family of geodesics

in the (m + 1)-dimensional Euclidean space E

m+1

.

Example 9.3. Let c, n : R → E

m+1

be smooth curves such that

the image n(R) of n is contained in the unit sphere S

m

. If we deﬁne a

map Φ : R R →E

m+1

by

Φ : (t, s) →c(t) +s n(t)

then for each t ∈ R the curve γ

t

: s → Φ(t, s) is a straight line and

hence a geodesic in E

m+1

. By diﬀerentiating with respect to the family

parameter t we yield the Jacobi ﬁeld J ∈ ¸

γ

0

(TE

m+1

) along γ

0

with

J(s) =

d

dt

Φ(t, s)[

t=0

= ˙ c(0) +s ˙ n(0).

The Jacobi equation (5) on a Riemannian manifold is linear in J.

This means that the space of Jacobi ﬁelds ¸

γ

(TM) along γ is a vector

space. We are now interested in determining the dimension of this

space

Proposition 9.4. Let γ : I → M be a geodesic, 0 ∈ I, p = γ(0)

and X = ˙ γ along γ. If v, w ∈ T

p

M are two tangent vectors at p then

there exists a unique Jacobi ﬁeld J along γ, such that J

p

= v and

(∇

X

J)

p

= w.

Proof. Let ¦X

1

, . . . , X

m

¦ be an orthonormal frame of parallel vec-

tor ﬁelds along γ, see Proposition 7.8. If J is a vector ﬁeld along γ,

then

J =

m

i=1

a

i

X

i

where a

i

= g(J, X

i

) are smooth functions on I. The vector ﬁelds

X

1

, . . . , X

m

are parallel so

∇

X

J =

m

i=1

˙ a

i

X

i

and ∇

X

∇

X

J =

m

i=1

¨ a

i

X

i

.

9. CURVATURE AND LOCAL GEOMETRY 97

For the curvature tensor we have

R(X

i

, X)X =

m

k=1

b

k

i

X

k

,

where b

k

i

= g(R(X

i

, X)X, X

k

) are smooth functions on I depending on

the geometry of (M, g). This means that R(J, X)X is given by

R(J, X)X =

m

i,k=1

a

i

b

k

i

X

k

.

and that J is a Jacobi ﬁeld if and only if

m

i=1

(¨ a

i

+

m

k=1

a

k

b

i

k

)X

i

= 0.

This is equivalent to the second order system

¨ a

i

+

m

k=1

a

k

b

i

k

= 0 for all i = 1, 2, . . . , m

of linear ordinary diﬀerential equations in a = (a

1

, . . . , a

m

). A global

solution will always exist and is uniquely determined by a(0) and ˙ a(0).

This implies that J exists globally and is uniquely determined by the

initial conditions

J(0) = v and (∇

X

J)(0) = w.

**Corollary 9.5. Let (M, g) be an m-dimensional Riemannian man-
**

ifold and γ : I →M be a geodesic in M. Then the vector space ¸

γ

(TM)

of all Jacobi ﬁelds along γ has the dimension 2m.

The following Lemma shows that when proving results about Ja-

cobi ﬁelds along a geodesic γ we can always assume, without loss of

generality, that [ ˙ γ[ = 1.

Lemma 9.6. Let (M, g) be a Riemannian manifold, γ : I →M be

a geodesic and J be a Jacobi ﬁeld along γ. If λ ∈ R

∗

and σ : λI → I

is given by σ : t →t/λ, then γ ◦ σ : λI →M is a geodesic and J ◦ σ is

a Jacobi ﬁeld along γ ◦ σ.

Proof. See Exercise 9.1.

Next we determine the Jacobi ﬁelds which are tangential to a given

geodesic.

98 9. CURVATURE AND LOCAL GEOMETRY

Proposition 9.7. Let (M, g) be a Riemannian manifold, γ : I →

M be a geodesic with [ ˙ γ[ = 1 and J be a Jacobi ﬁeld along γ. Let J

⊤

be the tangential part of J given by

J

⊤

= g(J, ˙ γ) ˙ γ and J

⊥

= J −J

⊤

be its normal part. Then J

⊤

and J

⊥

are Jacobi ﬁelds along γ and there

exist a, b ∈ R such that J

⊤

(s) = (as + b) ˙ γ(s) for all s ∈ I.

Proof. We now have

∇

˙ γ

∇

˙ γ

J

⊤

+R(J

⊤

, ˙ γ) ˙ γ = ∇

˙ γ

∇

˙ γ

(g(J, ˙ γ) ˙ γ) + R(g(J, ˙ γ) ˙ γ, ˙ γ) ˙ γ

= g(∇

˙ γ

∇

˙ γ

J, ˙ γ) ˙ γ

= −g(R(J, ˙ γ) ˙ γ, ˙ γ) ˙ γ

= 0.

This shows that the tangential part J

⊤

of J is a Jacobi ﬁeld. The

fact that ¸

γ

(TM) is a vector space implies that the normal part J

⊥

=

J −J

⊤

of J also is a Jacobi ﬁeld.

By diﬀerentiating g(J, ˙ γ) twice along γ we obtain

d

2

ds

2

(g(J, ˙ γ)) = g(∇

˙ γ

∇

˙ γ

J, ˙ γ) = −g(R(J, ˙ γ) ˙ γ, ˙ γ) = 0

so g(J, ˙ γ(s)) = (as + b) for some a, b ∈ R.

Corollary 9.8. Let (M, g) be a Riemannian manifold, γ : I → M

be a geodesic and J be a Jacobi ﬁeld along γ. If

g(J(t

0

), ˙ γ(t

0

)) = 0 and g((∇

˙ γ

J)(t

0

), ˙ γ(t

0

)) = 0

for some t

0

∈ I, then g(J(t), ˙ γ(t)) = 0 for all t ∈ I.

Proof. This is a direct consequence of the fact that the function

g(J, ˙ γ) satisﬁes the second order ordinary diﬀerential equation

¨

f = 0

and the initial conditions f(0) = 0 and

˙

f(0) = 0.

Our next aim is to show that if the Riemannian manifold (M, g)

has constant sectional curvature then we can solve the Jacobi equation

∇

X

∇

X

J +R(J, X)X = 0

along any given geodesic γ : I →M. For this we introduce the follow-

ing notation. For a real number κ ∈ R we deﬁne the c

κ

, s

κ

: R → R

by

c

κ

(s) =

_

¸

_

¸

_

cosh(

_

[κ[s) if κ < 0,

1 if κ = 0,

cos(

√

κs) if κ > 0.

9. CURVATURE AND LOCAL GEOMETRY 99

and

s

κ

(s) =

_

¸

_

¸

_

sinh(

_

[κ[s)/

_

[κ[ if κ < 0,

s if κ = 0,

sin(

√

κs)/

√

κ if κ > 0.

It is a well known fact that the unique solution to the initial value

problem

¨

f + κ f = 0, f(0) = a and

˙

f(0) = b

is the function f : R →R satisfying f(s) = ac

κ

(s) + bs

κ

(s).

Example 9.9. Let C be the complex plane with the standard Eu-

clidean metric ¸, )

R

2 of constant sectional curvature κ = 0. The ro-

tations about the origin produce a 1-parameter family of geodesics

Φ

t

: s →se

it

. Along the geodesic γ

0

: s →s we get the Jacobi ﬁeld

J

0

(s) =

∂Φ

t

∂t

(0, s) = is

with [J

0

(s)[ = [s[ = [s

κ

(s)[.

Example 9.10. Let S

2

be the unit sphere in the standard Eu-

clidean 3-space C R with the induced metric of constant sectional

curvature κ = +1. Rotations about the R-axis produce a 1-parameter

family of geodesics Φ

t

: s → (sin(s)e

it

, cos(s)). Along the geodesic

γ

0

: s →(sin(s), cos(s)) we get the Jacobi ﬁeld

J

0

(s) =

∂Φ

t

∂t

(0, s) = (isin(s), 0)

with [J

0

(s)[

2

= sin

2

(s) = [s

κ

(s)[

2

.

Example 9.11. Let B

2

1

(0) be the open unit disk in the complex

plane with the hyperbolic metric

g(X, Y ) =

4

(1 −[z[

2

)

2

¸, )

R

2

of constant sectional curvature κ = −1. Rotations about the origin

produce a 1-parameter family of geodesics Φ

t

: s →tanh(s/2)e

it

. Along

the geodesic γ

0

: s →tanh(s/2) we get the Jacobi ﬁeld

J

0

(s) =

∂Φ

t

∂t

(0, s) = i tanh(s/2)

with

[J

0

(s)[

2

=

4 tanh

2

(s/2)

(1 −tanh

2

(s/2))

2

= sinh

2

(s) = [s

κ

(s)[

2

.

100 9. CURVATURE AND LOCAL GEOMETRY

Let (M, g) be a Riemannian manifold of constant sectional curva-

ture κ and γ : I → M be a geodesic with [X[ = 1 where X = ˙ γ.

Further let P

1

, P

2

, . . . , P

m−1

be parallel vector ﬁelds along γ such that

g(P

i

, P

j

) = δ

ij

and g(P

i

, X) = 0. Any vector ﬁeld J along γ may now

be written as

J(s) =

m−1

i=1

f

i

(s)P

i

(s) +f

m

(s)X(s).

This means that J is a Jacobi ﬁeld if and only if

m−1

i=1

¨

f

i

(s)P

i

(s) +

¨

f

m

(s)X(s) = ∇

X

∇

X

J

= −R(J, X)X

= −R(J

⊥

, X)X

= −κ(g(X, X)J

⊥

−g(J

⊥

, X)X)

= −κJ

⊥

= −κ

m−1

i=1

f

i

(s)P

i

(s).

This is equivalent to the following system of ordinary diﬀerential equa-

tions

(6)

¨

f

m

(s) = 0 and

¨

f

i

(s) + κf

i

(s) = 0 for all i = 1, 2, . . . , m−1.

It is clear that for the initial values

J(s

0

) =

m−1

i=1

v

i

P

i

(s

0

) + v

m

X(s

0

),

(∇

X

J)(s

0

) =

m−1

i=1

w

i

P

i

(s

0

) + w

m

X(s

0

)

or equivalently

f

i

(s

0

) = v

i

and

˙

f

i

(s

0

) = w

i

for all i = 1, 2, . . . , m

we have a unique and explicit solution to the system (6) on the whole

of I.

In the next example we give a complete description of the Jacobi

ﬁelds along a geodesic on the 2-dimensional sphere.

Example 9.12. Let S

2

be the unit sphere in the standard Eu-

clidean 3-space C R with the induced metric of constant curvature

κ = +1 and γ : R →S

2

be the geodesic given by γ : s →(e

is

, 0). Then

9. CURVATURE AND LOCAL GEOMETRY 101

˙ γ(s) = (ie

is

, 0) so it follows from Proposition (9.7) that all Jacobi ﬁelds

tangential to γ are given by

J

T

(a,b)

(s) = (as +b)(ie

is

, 0) for some a, b ∈ R.

The vector ﬁeld P : R → TS

2

given by s → ((e

is

, 0), (0, 1)) satisﬁes

¸P, ˙ γ) = 0 and [P[ = 1. The sphere S

2

is 2-dimensional and ˙ γ is parallel

along γ so P must be parallel. This implies that all the Jacobi ﬁelds

orthogonal to ˙ γ are given by

J

N

(a,b)

(s) = (0, a cos s + b sin s) for some a, b ∈ R.

In more general situations, where we do not have constant curvature

the exponential map can be used to produce Jacobi ﬁelds as follows. Let

(M, g) be a complete Riemannian manifold, p ∈ M and v, w ∈ T

p

M.

Then s →s(v+tw) deﬁnes a 1-parameter family of lines in the tangent

space T

p

M which all pass through the origin 0 ∈ T

p

M. Remember that

the exponential map

(exp)

p

[

B

m

ε

p

(0)

: B

m

ε

p

(0)

→exp(B

m

ε

p

(0)

)

maps lines in T

p

M through the origin onto geodesics on M. Hence the

map

Φ

t

: s →(exp)

p

(s(v + tw))

is a 1-parameter family of geodesics through p ∈ M, as long as s(v+tw)

is an element of B

m

ε

p

(0)

. This means that

J(s) =

∂Φ

t

∂t

(0, s)

is a Jacobi ﬁeld along the geodesic γ : s → Φ

0

(s) with γ(0) = p and

˙ γ(0) = v. It is easily veriﬁed that J satisﬁes the initial conditions

J(0) = 0 and (∇

X

J)(0) = w.

The following technical result is needed for the proof of the main

Theorem 9.14 at the end of this chapter.

Lemma 9.13. Let (M, g) be a Riemannian manifold with sectional

curvature uniformly bounded above by ∆ and γ : [0, α] → M be a

geodesic on M with [X[ = 1 where X = ˙ γ. Further let J : [0, α] →TM

be a Jacobi ﬁeld along γ such that g(J, X) = 0 and [J[ , = 0 on (0, α).

Then

(i) d

2

([J[)/ds

2

+ ∆ [J[ ≥ 0,

(ii) if f : [0, α] →R is a C

2

-function such that

(a)

¨

f + ∆ f = 0 and f > 0 on (0, α),

(b) f(0) = [J(0)[, and

(c)

˙

f(0) = [∇

X

J(0)[,

102 9. CURVATURE AND LOCAL GEOMETRY

then f(s) ≤ [J(s)[ on (0, α),

(iii) if J(0) = 0, then [∇

X

J(0)[ s

∆

(s) ≤ [J(s)[ for all s ∈ (0, α).

Proof. (i) Using the facts that [X[ = 1 and ¸X, J) = 0 we obtain

d

2

ds

2

([J[) =

d

2

ds

2

_

g(J, J) =

d

ds

(

g(∇

X

J, J)

[J[

)

=

g(∇

X

∇

X

J, J)

[J[

+

[∇

X

J[

2

[J[

2

−g(∇

X

J, J)

2

[J[

3

≥

g(∇

X

∇

X

J, J)

[J[

= −

g(R(J, X)X, J)

[J[

≥ −∆ [J[.

(ii) Deﬁne the function h : [0, α) →R by

h(s) =

_

|J(s)|

f(s)

if s ∈ (0, α),

lim

s→0

|J(s)|

f(s)

= 1 if s = 0.

Then

˙

h(s) =

1

f

2

(s)

(

d

ds

([J(s)[)f(s) −[J(s)[

˙

f(s))

=

1

f

2

(s)

_

s

0

(

d

2

dt

2

([J(t)[)f(t) −[J(t)[

¨

f(t))dt

=

1

f

2

(s)

_

s

0

f(t)(

d

2

dt

2

([J(t)[) + ∆ [J(t)[)dt

≥ 0.

This implies that

˙

h(s) ≥ 0 so f(s) ≤ [J(s)[ for all s ∈ (0, α).

(iii) The function f(s) = [(∇

X

J)(0)[ s

∆

(s) satisﬁes the diﬀerential

equation

¨

f(s) + ∆f(s) = 0

and the initial conditions f(0) = [J(0)[ = 0,

˙

f(0) = [(∇

X

J)(0)[ so it

follows from (ii) that [(∇

X

J)(0)[ s

∆

(s) = f(s) ≤ [J(s)[.

Let (M, g) be a Riemannian manifold of sectional curvature which is

uniformly bounded above, i.e. there exists a ∆ ∈ R such that K

p

(V ) ≤

∆ for all V ∈ G

2

(T

p

M) and p ∈ M. Let (M

∆

, g

∆

) be another Rie-

mannian manifold which is complete and of constant sectional curva-

ture K ≡ ∆. Let p ∈ M, p

∆

∈ M

∆

and identify T

p

M

∼

= R

m

∼

= T

p

∆

M

∆

.

9. CURVATURE AND LOCAL GEOMETRY 103

Let U be an open neighbourhood of R

m

around 0 such that the

exponential maps (exp)

p

and (exp)

p

∆

are diﬀeomorphisms from U onto

their images (exp)

p

(U) and (exp)

p

∆

(U), respectively. Let (r, p, q) be

a geodesic triangle i.e. a triangle with sides which are shortest paths

between their endpoints. Furthermore let c : [a, b] → M be the side

connecting r and q and v : [a, b] → T

p

M be the curve deﬁned by

c(t) = (exp)

p

(v(t)). Put c

∆

(t) = (exp)

p

∆

(v(t)) for t ∈ [a, b] and then

it directly follows that c(a) = r and c(b) = q. Finally put r

∆

= c

∆

(a)

and q

∆

= c

∆

(b).

Theorem 9.14. For the above situation the following inequality for

the distance function d is satisﬁed

d(q

∆

, r

∆

) ≤ d(q, r).

Proof. Deﬁne a 1-parameter family s → s v(t) of straight lines

in T

p

M through p. Then Φ

t

: s → (exp)

p

(s v(t)) and Φ

∆

t

: s →

(exp)

p

∆

(s v(t)) are 1-parameter families of geodesics through p ∈ M,

and p

∆

∈ M

∆

, respectively. Hence J

t

= ∂Φ

t

/∂t and J

∆

t

= ∂Φ

∆

t

/∂t are

Jacobi ﬁelds satisfying the initial conditions J

t

(0) = J

∆

t

(0) = 0 and

(∇

X

J

t

)(0) = (∇

X

J

∆

t

)(0) = ˙ v(t). Using Lemma 9.13 we now obtain

[ ˙ c

∆

(t)[ = [J

∆

t

(1)[

= [(∇

X

J

∆

t

)(0)[ s

∆

(1)

= [(∇

X

J

t

)(0)[ s

∆

(1) ≤ [J

t

(1)[

= [ ˙ c(t)[

The curve c is the shortest path between r and q so we have

d(r

∆

, q

∆

) ≤ L(c

∆

) ≤ L(c) = d(r, q).

**We now add the assumption that the sectional curvature of the
**

manifold (M, g) is uniformly bounded below i.e. there exists a δ ∈ R

such that δ ≤ K

p

(V ) for all V ∈ G

2

(T

p

M) and p ∈ M. Let (M

δ

, g

δ

)

be a complete Riemannian manifold of constant sectional curvature δ.

Let p ∈ M and p

δ

∈ M

δ

and identify T

p

M

∼

= R

m

∼

= T

p

δ

M

δ

. Then a

similar construction as above gives two pairs of points q, r ∈ M and

q

δ

, r

δ

∈ M

δ

and shows that

d(q, r) ≤ d(q

δ

, r

δ

).

Combining these two results we obtain locally

d(q

∆

, r

∆

) ≤ d(q, r) ≤ d(q

δ

, r

δ

).

104 9. CURVATURE AND LOCAL GEOMETRY

Exercises

Exercise 9.1. Find a proof for Lemma 9.6.

Exercise 9.2. Let (M, g) be a Riemannian manifold and γ : I →M

be a geodesic such that X = ˙ γ ,= 0. Further let J be a non-vanishing

Jacobi ﬁeld along γ with g(X, J) = 0. Prove that if g(J, J) is constant

along γ then (M, g) does not have strictly negative curvature.

1

Preface

These lecture notes grew out of an M.Sc. course on diﬀerential geometry which I gave at the University of Leeds 1992. Their main purpose is to introduce the beautiful theory of Riemannian Geometry a still very active area of mathematical research. This is a subject with no lack of interesting examples. They are indeed the key to a good understanding of it and will therefore play a major role throughout this work. Of special interest are the classical Lie groups allowing concrete calculations of many of the abstract notions on the menu. The study of Riemannian geometry is rather meaningless without some basic knowledge on Gaussian geometry that i.e. the geometry of curves and surfaces in 3-dimensional space. For this I recommend the excellent textbook: M. P. do Carmo, Diﬀerential geometry of curves and surfaces, Prentice Hall (1976). These lecture notes are written for students with a good understanding of linear algebra, real analysis of several variables, the classical theory of ordinary diﬀerential equations and some topology. The most important results stated in the text are also proved there. Other are left to the reader as exercises, which follow at the end of each chapter. This format is aimed at students willing to put hard work into the course. For further reading I recommend the very interesting textbook: M. P. do Carmo, Riemannian Geometry, Birkh¨user (1992). a I am very grateful to my many enthusiastic students who throughout the years have contributed to the text by ﬁnding numerous typing errors and giving many useful comments on the presentation. Norra N¨bbel¨v, 17 February 2008 o o Sigmundur Gudmundsson

.

The Tangent Space Chapter 4. The Riemann Curvature Tensor Chapter 9. The Tangent Bundle Chapter 5. Curvature and Local Geometry 5 7 21 35 47 59 69 83 95 3 . Riemannian Manifolds Chapter 6. Introduction Chapter 2. Geodesics Chapter 8. The Levi-Civita Connection Chapter 7.Contents Chapter 1. Diﬀerentiable Manifolds Chapter 3.

.

CHAPTER 1 Introduction On the 10th of June 1854 Georg Friedrich Bernhard Riemann (18261866) gave his famous ”Habilitationsvortrag” in the Colloquium of the ¨ Philosophical Faculty at G¨ttingen. Riemann’s revolutionary ideas generalized the geometry of surfaces which had been studied earlier by Gauss. o welche der Geometrie zu Grunde liegen” is often said to be the most important in the history of diﬀerential geometry. at the age of 76. His talk ”Uber die Hypothesen. was in the audience and is said to have been very impressed by his former student. Bolyai and Lobachevsky. 5 . Later this lead to an exact deﬁnition of the modern concept of an abstract Riemannian manifold. Johann Carl Friedrich Gauss (1777-1855).

.

This generalizes curves and surfaces in R3 studied in classical diﬀerential geometry. T ) be a topological Hausdorﬀ space with a countable basis. The natural number m is called the dimension of M . R. R ) = r=1 ∞ n C r (U. Rn ). Birkh¨user (1992). For a natural number m let Rm be the m-dimensional real vector space equipped with the topology induced by the standard Euclidean metric d on Rm given by d(x.1. Let (M. Parks. Our manifolds are modelled on the standard diﬀerentiable structure on the vector spaces Rm via compatible local charts. The set of real analytic maps from U to Rn will be denoted by C ω (U. a Deﬁnition 2. According to Deﬁnition 2. We 7 . y) = For positive natural numbers n. G. To denote that the dimension of M is m we write M m .1 a topological manifold M is locally homeomorphic to the standard Rm for some natural number m. By smooth maps U → Rn we mean the elements of ∞ (x1 − y1 )2 + . Rn ) denote the r-times continuously diﬀerentiable maps from U to Rn . r and an open subset U of Rm we shall by C r (U. . The pair (U. We give many examples. Krantz and H. Then M is said to be a topological manifold if there exists a natural number m and for each point p ∈ M an open neighbourhood U of p and a continuous map x : U → Rm which is a homeomorphism onto its image x(U ) which is an open subset of Rm . + (xm − ym )2 . C (U.CHAPTER 2 Diﬀerentiable Manifolds In this chapter we introduce the important notion of a diﬀerentiable manifold. For the theory of real analytic maps we recommend the book: S. Rn ). . x) is called a (local) chart (or local coordinates) on M . study their submanifolds and diﬀerentiable maps between manifolds. A Primer of Real Analytic Functions.

xβ ◦ x−1 |xα (Uα ∩Uβ ) : xα (Uα ∩ Uβ ) → Rm α Example 2. . 0) ∈ R × Rm and S be the south pole S = (−1. 1 − p1 1 xS : (p1 . A C r -atlas A is said to be maximal if it contains all the charts that are compatible with it. T ) we have the trivial C ω -atlas A = {(Rm . . 0) on S m . . . Example 2.8 2. x)} is a C r -atlas on M . or a diﬀerentiable manifold of (M. Put UN = S m \ {N }.4. Let M be a topological manifold. DIFFERENTIABLE MANIFOLDS shall now deﬁne a diﬀerentiable structure on M via local charts and turn M into a diﬀerentiable manifold. . A diﬀerentiable manifold is said to be smooth if its transition maps are C ∞ and real analytic if they are C ω . pm+1 ). . . 1 + p1 . pm+1 ) → (p2 . . pm+1 ). Let N be the north pole N = (1. respectively. A ˆ maximal atlas A on M is also called a C r -structure on M . pm+1 ) → (p2 . Let S m denote the unit sphere in Rm+1 i.e.3. . . It simply consists of all charts compatible with A. M= Uα α A = {(Uα . . It should be noted that a given C r -atlas A on a topological manifold ˆ M determines a unique C r -structure A on M containing A. x) on M is said to be compatible with a C r -atlas A on ˆ M if the union A ∪ {(U. if M is a topological manifold and A is a C r -structure on M . . The pair ˆ is said to be a C r -manifold. . β ∈ I the corresponding transition maps are r-times continuously diﬀerentiable. For the standard topological space (Rm . US = S m \ {S} and deﬁne xN : UN → Rm . xS : US → Rm by 1 xN : (p1 . A chart (U. A) ˆ class C r . . 2 S m = {p ∈ Rm+1 | p1 + · · · + p2 = 1} m+1 equipped with the subset topology induced by the standard T on Rm+1 . . Then a C r -atlas on M is a collection of local charts on M such that A covers the whole of M i. . xα )| α ∈ I} and for all α. x)| x : p → p} ˆ inducing the standard C ω -structure A on Rm . Deﬁnition 2.e.2.

.. A) is called the m-dimensional real projective space.. The diﬀerentiable manifold (RP m .. pk pk pk pk If [p] ≡ [q] then p = λq for some λ ∈ R∗ so pl /pk = ql /qk for all l.. . The corresponding transition maps are given by p1 pl−1 pl+1 pm+1 p1 pk−1 pk+1 pm+1 ( .. . m + 1} xk ◦ x−1 |xl (Ul ∩Uk ) : xl (Ul ∩ Uk ) → Rm l [p] = {λp ∈ Rm+1 | λ ∈ R∗ } . .. ) → ( ... Example 2. This means that the map xk is well deﬁned for all k. . . DIFFERENTIABLE MANIFOLDS 9 Then the transition maps are given by xS ◦ x−1 .. . be the natural projection mapping a point p ∈ Rm+1 \ {0} to the equivalence class [p] ∈ RP m i. ) pl pl pl pl pk pk pk pk so the collection ˆ is a C ω -atlas on RP m ..2. .. . . . xN ◦ x−1 : Rm \ {0} → Rm \ {0} N S p→ p |p|2 so A = {(UN .. xN ). xk )| k = 1. . A = {(Uk . .e. . 1. . ..5. m + 1} deﬁne the open subset Uk = {[p] ∈ RP m | pk = 0} of RP m and the charts xk : Uk → Rm by p1 pk−1 pk+1 pm+1 xk : [p] → ( . ). On the set Rm+1 \ {0} we deﬁne the equivalence relation ≡ by: Let RP m be the quotient space (Rm+1 \ {0})/ ≡ and π : Rm+1 \ {0} → RP m p ≡ q if and only if there exists a λ ∈ R∗ such that p = λq. . . Another interesting example of a diﬀerentiable manifold is the mdimensional real projective space RP m .. xS )} is a C ω -atlas on S m . . For k ∈ {1. 1.. 1. The C ω -manifold ˆ (S m .... A) is called the standard m-dimensional sphere. Equip RP m with the quotient topology induced by π and T on Rm+1 . the line through the origin generated by p.. (US .

Then ˆ is a C r -atlas for M . Proof. U0 = C and U∞ = C \ {0}. A) r into a diﬀerentiable manifold of class C and the dimension of M satisﬁes dim M = dim M1 + dim M2 . respectively. B) be a C r -manifold. n ∈ N be natural numbers such that 1 ≤ ˆ m ≤ n and (N n . Deﬁnition 2. DIFFERENTIABLE MANIFOLDS ˆ and put C∗ = C \ {0}. A subset M of N is said to be a submanifold of N if for each point p ∈ M there exists a chart ˆ (Up . (U∞ . Proposition 2.9. Let M be a submanifold of N equipped with the subset topology and π : Rm × Rn−m → Rm be the natural projection onto the ﬁrst factor.8. Let C be the extended complex plane given by ˆ C = C ∪ {∞} . Then deﬁne the ˆ local coordinates x0 : U0 → C and x∞ : U∞ → C on C by x0 : z → z and x∞ : w → 1/w.7. x∞ ◦ x−1 . x0 ◦ x−1 : C∗ → C∗ 0 ∞ ˆ Example 2. (π ◦ xp )|Up ∩M )| p ∈ M } xp (Up ∩ M ) = xp (Up ) ∩ (Rm × {0}).6. Let m. A) is an m-dimensional ˆ C r -manifold. Let m. In particular. x∞ )} is a C ω -atlas on ˆ ˆ ˆ C. n ∈ N be natural numbers such that 1 ≤ ˆ m ≤ n and (N n . The real analytic manifold (C. The corresponding transition maps are both given by z → 1/z so A = {(U0 . A = {(Up ∩ M.10 2. See Exercise 2. Let (M1 . Let M = M1 × M2 be the product space with the ˆ product topology. The concept of a submanifold of a given diﬀerentiable manifold will play an important role as we go along and we shall be especially interested in the connection between the geometry of a submanifold and that of its ambient space. A1 ) and (M2 . A) is called the Riemann sphere. B) be a C r -manifold. For the product of two diﬀerentiable manifolds we have the following interesting result. x0 ). Then there exists an atlas A on M making (M. ˆ ˆ Proposition 2. xp ) ∈ B such that p ∈ Up and xp : Up → Rm × Rn−m satisﬁes The natural number (n − m) is called the codimension of M in N . the pair (M.1. The diﬀerentiable structure A on the submanifold M of ˆ N is called the induced structure of B. A2 ) be two diﬀerentiable manifolds of class C r .

2. This will later be generalized to the manifold setting. The diﬀerential (dF −1 )q of ˆ F −1 at q satisﬁes ˆ (dF −1 )q = (dFp )−1 i. .e. ∂Fm /∂x1 (p) . . Let U be an open subset of Rn and F : U → Rn be a C r -map. ds which is the tangent vector of the curve F ◦ γ at F (p) ∈ Rm . . Before stating the implicit function theorem we remind the reader of the deﬁnition of the following notions. If p ∈ U and the diﬀerential dFp : Rn → Rn of F at p is invertible then there exist open neighbourhoods Up around p ˆ and Uq around q = F (p) such that F = F |Up : Up → Uq is bijective and ˆ ˆ the inverse (F )−1 : Uq → Up is a C r -map. DIFFERENTIABLE MANIFOLDS 11 Proof. . Hence the diﬀerential dFp can be seen as a linear map that maps tangent vectors at p ∈ U to tangent vectors at the image F (p) ∈ Rm .10 (The Inverse Function Theorem). . dFp = . Our next step is to prove the implicit function theorem which is a useful tool for constructing submanifolds of Rn . . it is the inverse of the diﬀerential dFp of F at p. . . See Exercise 2. Fact 2. For this we use the classical inverse function theorem stated below. ∂F1 /∂xn (p) . . .2. Note that if is a diﬀerentiable map deﬁned on an open subset U of Rn then its diﬀerential dFp : Rn → Rm at a point p ∈ U is a linear map given by the m × n matrix ∂F1 /∂x1 (p) . ∂Fm /∂xn (p) F : U → Rm If γ : R → U is a curve in U such that γ(0) = p and γ(0) = v ∈ Rn ˙ m m then the composition F ◦ γ : R → R is a curve in R and according to the chain rule we have d dFp · v = (F ◦ γ(s))|s=0 .

with respect to the ⊥ decompositions Rn = Kp ⊕ Kp and Rn = Rm ⊕ Rn−m . n be natural numbers such that m < n and F : U → Rm be a C r -map from an open subset U of Rn . . hence bijective. Fm : U → R are linearly independent at p. It now follows from the inverse function theorem that there exist open neighbourhoods Vp around p and Wp around Gp (p) ˆ ˆ −1 such that Gp = Gp |Vp : Vp → Wp is bijective. A point p ∈ U is said to be critical for F if the diﬀerential dFp : Rn → Rm is not of full rank. the diﬀerential dFp of F at p satisﬁes the following condition det(dFp · (dFp )t ) = 0. .11. . Note that if n ≥ m then p ∈ U is a regular point of F = (F1 . n be positive natural numbers. . Fm ) : U → Rm if and only if the gradients ∇F1 . DIFFERENTIABLE MANIFOLDS Deﬁnition 2. If q ∈ F (U ) is a regular value of F then the pre-image F −1 ({q}) of q is an (n − m)-dimensional submanifold of Rn of class C r . πp |Kp = 0 and deﬁne the ⊥ m n−m map Gp : U → R × R by Then the diﬀerential (dGp )p : Rn → Rn of Gp . . 0 πp Gp : x → (F (x). p ˜ Now put Up = F −1 ({q}) ∩ Vp then ˜ ˆp Up = G−1 (({q} × Rn−m ) ∩ Wp ) . Let m.12 2. πp (x)). Theorem 2. . . .12 (The Implicit Function Theorem). and regular if it is not critical. Let p be an element of F −1 ({q}) and Kp be the kernel of the diﬀerential dFp i. d(G−1 )Gp (p) = (dGp )−1 and d(G−1 )y is bijective for all y ∈ Wp . Let m. . Proof. Let πp : Rn → Rn−m be a linear map such that πp |Kp : Kp → Rn−m is bijective. is given by (dGp )p = dFp |Kp 0 ⊥ .e. ∇Fm of the coordinate functions F1 . . A point q ∈ F (U ) is said to be a regular value of F if every point of the pre-image F −1 ({q}) of q is regular and a critical value otherwise. the (n − m)-dimensional subspace of Rn given by Kp = {v ∈ Rn | dFp · v = 0}. . or equivalently. . U be an open subset of Rn and F : U → Rm be a C r -map. the inverse Gp : Wp → Vp ˆp ˆp is C r .

v) of F at (p.2. then the map ˜ xp = π ◦ Gp |Up : Up → ({q} × Rn−m ) ∩ Wp → Rn−m ˜ ˜ ˜ is a chart on the open neighbourhood Up of p. Let F : Rm+1 → R be the C ω -map given by m+1 F : (p1 .4. p. F −1 ({0}) = {(p. . so dFp · (dFp )t = 4|p|2 ∈ R. v) satisﬁes p 0 . v) → ((|p|2 − 1)/2. respectively. The diﬀerential dF(p. DIFFERENTIABLE MANIFOLDS 13 so if π : Rm × Rn−m → Rn−m is the natural projection onto the second factor. It is the standard m-dimensional sphere introduced in Example 2. Let F : Rm+1 ×Rm+1 → R2 be the C ω -map deﬁned by F : (p. This means that det(dF · (dF )t ) = |p|2 (|p|2 + |v|2 ) = 1 + |v|2 > 0 p. pm+1 ) → p2 . Example 2.v) = v p Hence on F −1 ({0}).13. This means that 1 ∈ R is a regular value of F so the ﬁbre of F is an m-dimensional submanifold of Rm+1 . Example 2. v) ∈ Rm+1 × Rm+1 | |p|2 = 1 and . ˜ ˜ B = {(Up . . . v ). . We shall later see that T S m is what is called the tangent bundle of the m-dimensional sphere.14. xp )| p ∈ F −1 ({q})} Employing the implicit function theorem we yield the following interesting examples of the m-dimensional sphere S m and its tangent bundle T S m as diﬀerentiable submanifolds of Rm+1 and R2m+2 . The point q ∈ F (U ) is a regular value so the set is a C r -atlas for F −1 ({q}). v = 0} S m = {p ∈ Rm+1 | |p|2 = 1} = F −1 ({1}) which we denote by T S m is a 2m-dimensional submanifold of R2m+2 . dF(p. i i=1 The diﬀerential dFp of F at p is given by dFp = 2p.

y ◦ φ ◦ x−1 |x(U ∩φ−1 (V )) : x(U ∩ φ−1 (V )) ⊂ Rm → Rn O(m) = F −1 ({e}) = {A ∈ Rm×m | At A = e} . The set O(m) is the well known orthogonal group.16. Let Sym(Rm ) be the linear subspace of Rm×m consisting of all symmetric m × m-matrices Then it is easily seen that the dimension of Sym(Rm ) is m(m + 1)/2. A map φ : M → N is said to be diﬀerentiable of class C r if for all ˆ ˆ charts (U.16. so the identity matrix e ∈ Sym(Rm ) is a regular value of F . ds so the diﬀerential dFA of F at A ∈ Rm×m satisﬁes This means that for an arbitrary element A of dFA : X → X t A + At X. F : A → At A. to prove the following result concerning the composition of diﬀerentiable maps between manifolds. deﬁned on an open subset of Rm . The set of smooth functions deﬁned on M is denoted by C ∞ (M ). A2 ) be two C r -manifolds. using Deﬁnition 2. x) ∈ A1 and (V. DIFFERENTIABLE MANIFOLDS Example 2. ˆ ˆ Deﬁnition 2. The concept of a diﬀerentiable map U → Rn . Following the implicit function theorem O(m) is a submanifold of Rm×m of dimension m(m − 1)/2. It is an easy exercise. A1 ) and (N n . and Y ∈ Sym(Rm ) we have dFA (AY /2) = Y . We shall see that the most important properties of these objects in the classical case are also valid in the manifold setting.15. We shall now apply the implicit function theorem to construct the important orthogonal group O(m) as a submanifold of the set of the real vector space of m × m matrices Rm×m . Hence the diﬀerential dFA is surjective. Let (M m . A diﬀerentiable map γ : I → M deﬁned on an open interval of R is called a diﬀerentiable curve in M . Sym(Rm ) = {A ∈ Rm×m | At = A}. can be generalized to mappings between manifolds.14 2. A diﬀerentiable map f : M → R with values in R is called a diﬀerentiable function on M . y) ∈ A2 the map is of class C r . Let F : Rm×m → Sym(Rm ) be the map deﬁned by If A : R → Rm×m is a curve in Rm×m then d ˙ ˙ (F ◦ A(s)) = A(s)t A(s) + A(s)t A(s).

and (M2 . A2 ) be two diﬀerentiable r manifolds of class C and M1 . A1 ). A1 ) → ˆ (M3 . Two manifolds (M1 . The next very useful proposition generalizes a classical result from the real analysis of several variables. The following remarkable result was proved by John Milnor in his famous paper: Diﬀerentiable structures on spheres. A2 ). A3 ) be diﬀerˆ ˆ ˆ folds and φ : (M1 . A1 ) and (M2 . A3 ) is a diﬀerentiable map of class C r . m ˆ m ˆ Deep Result 2. see Exercise 2. Let (N1 . 81 (1959). See Exercise 2. It can be seen that even the real line R carries diﬀerent diﬀerentiable structures.22. A1 ) and (M2 . Then the composition ψ ◦ φ : (M1 . If M1 and M2 are homeˆ ˆ omorphic as topological spaces and m ≤ 3 then (M1 .6. ˆ ˆ Deﬁnition 2. J. A1 ). Math. If φ : N1 → N2 is a diﬀerentiable map of class C r such that φ(M1 ) is contained in M2 . introduced earlier.19.20. DIFFERENTIABLE MANIFOLDS 15 ˆ ˆ ˆ Proposition 2. A r It can be shown that the 2-dimensional sphere S 2 in R3 and the Riemann sphere. then the restriction φ|M1 : M1 → M2 is diﬀerentiable of class C r . Proof. The 7-dimensional sphere S 7 has exactly 28 diﬀerent diﬀerentiable structures.8. A3 ) be C r -maniˆ1 ) → (M2 . Amer. In ˆ that case the map φ is said to be a diﬀeomorphism between (M1 . Proof. See Exercise 2. A ˆ entiable maps of class C r . M2 be submanifolds of N1 and N2 . Deep Result 2. Two C r -structures A1 and A2 on the same topological manifold M are said to be diﬀerent if the identity map idM : ˆ ˆ (M. A2 ) is not a diﬀeomorphism. are diﬀeomorphic. Let (M1 . A2 ) → (M3 . respectively. A2 ) of class C are said to be diﬀeomorphic if there exists a bijective C r -map φ : M1 → M2 . A2 ) are diﬀeomorphic. ψ : (M2 .17.5. A1 ) ˆ2 ). ˆ ˆ Proposition 2. (M2 . .7. A2 ). see Exercise 2.2. A1 ) and (N2 . A2 ) be two diﬀerentiable r manifolds of class C and of equal dimensions. (M2 . ˆ ˆ Deﬁnition 2. (M3 . A1 ) → (M. such that the inverse φ−1 : M2 → M1 is of class C r .21.18. Let (M1 . 962-972.

·). Let H be the set of quaternions deﬁned by H = {z + wj| z. z. ·). (ii) φ2 : S 3 ⊂ C2 → S 2 ⊂ C×R. (v) φ5 : Rm+1 \ {0} → RP m . Then (K. multiplication · and conjugation ¯ . φ5 : x → [x].24. ·) containing the positive real numbers (R+ .27. ·) is a Lie group. ·) is an interesting compact Lie subgroup of (C∗ .23. q) → p · q −1 Note that the standard diﬀerentiable Rm equipped with the usual addition + forms an abelian Lie group (Rm . y.25.10 Proposition 2. A Lie group is a smooth manifold G with a group structure · such that the map ρ : G × G → G with is smooth. φ3 : t → eit . Example 2. The unit circle (S 1 . φ2 : (z1 .16 2. Another subgroup is the set of the non-zero real numbers (R∗ . |z1 |2 −|z2 |2 ). DIFFERENTIABLE MANIFOLDS In diﬀerential geometry we are especially interested in diﬀerentiable manifolds carrying a group structure compatible with their diﬀerentiable structure. ·) and the 0-dimensional sphere (S 0 . Let (G. Proof. ·) be a Lie group and K be a submanifold of G which is a subgroup. Deﬁnition 2. φ6 : x → [x]. φ1 : (x. ¯ (iii) φ3 : R1 → S 1 ⊂ C. The result of Proposition 2. +) with ρ : Rm ×Rm → Rm given by ρ : (p.26. φ4 : x → x/|x|. Such manifolds are named after the famous mathematician Sophus Lie (1842-1899) and will play an important role throughout this work. Corollary 2. w ∈ C} equipped with the addition +.22. Then the left translation Lp : G → G is a smooth diﬀeomorphism. The set of non-zero complex numbers C∗ together with the standard multiplication · forms a Lie group (C∗ . y. Proof. 0).24 and Proposition 2. z2 ) → (2z1 z2 . Example 2.22 can be used to show that the following maps are all smooth. The statement is a direct consequence of Deﬁnition 2. (i) φ1 : S 2 ⊂ R3 → S 3 ⊂ R4 . Let G be a Lie group and p be an element of G. For an element p in G the left translation by p is the map Lp : G → G deﬁned by Lp : q → p · q. ρ : (p. (vi) φ6 : S m → RP m . ·) as subgroups. See Exercise 2. q) → p − q. z) → (x. (iv) φ4 : Rm+1 \ {0} → S m .

·) of (H∗ . Lie Groups Beyond an Introduction. On H we deﬁne a scalar product and a real valued norm given by |p|2 = p · p. 0 0 1 H × H → H. z). then (Sol3 . z ∈ R}. Then it is easily seen that the non-zero quaternions (H∗ . Knapp. Let N il3 be the subset of R3×3 given by 1 x z N il3 = {0 1 y ∈ R3×3 | x. 0 0 1 Then N il3 has a natural diﬀerentiable structure determined by the global coordinate φ : N il3 → R3 with 1 x z φ : 0 1 y → (x. a Example 2. y. z). W. ·) form a Lie group. y. (iii) (z1 + w1 j) · (z2 + w2 j) = (z1 z2 − w1 w2 ) + (z1 w2 + w1 z2 )j ¯ ¯ extending the standard operations on R and C as subsets of H.2. y. DIFFERENTIABLE MANIFOLDS 17 (i) (z + wj) = z − wj. ∗) is a Lie group.29. y. Then Sol3 has a natural diﬀerentiable structure determined by the global coordinate φ : Sol3 → R3 with z e 0 x φ : 0 e−z y → (x. They are both non-abelian. Example 2. 0 0 1 It is easily seen that if ∗ is the standard matrix multiplication. Birkh¨user (2002). Then the 3-dimensional ¯ unit sphere S 3 in H ∼ R4 with the restricted multiplication forms a = compact Lie subgroup (S 3 .28. (p. Let Sol3 be the subset of R3×3 given by z e 0 x Sol3 = { 0 e−z y ∈ R3×3 | x. then (N il3 . 0 0 1 . ¯ (ii) (z1 + w1 j) + (z2 + w2 j) = (z1 + z2 ) + (w1 + w2 )j. ∗) is a Lie group. z ∈ R}. We shall now introduce some of the classical real and complex matrix Lie groups. As a reference on this topic we recommend the wonderful book: A. q) → p · q ¯ It is easily seen that if ∗ is the standard matrix multiplication. ·).

It is called the complex general linear group and its neutral element e is the identity matrix. DIFFERENTIABLE MANIFOLDS Example 2. The subset GLm (R) of Rm×m is open so dim GLm (R) = m2 .30. It is called the real general linear group and its neutral element e is the identity matrix.18 2. The set of invertible complex m × m matrices GLm (C) = {A ∈ Cm×m | det A = 0} equipped with the standard matrix multiplication has the structure of a Lie group. As a subgroup of GLm (C) we have the complex special linear group SLm (C) given by The dimension of the submanifold SLm (C) of Cm×m is 2(m2 − 1). The set of invertible real m × m matrices GLm (R) = {A ∈ Rm×m | det A = 0} equipped with the standard matrix multiplication has the structure of a Lie group. Example 2. SLm (C) = {A ∈ Cm×m | det A = 1}. Note that O(1) = {±1} so O(m) can be seen as two copies of SO(m). As a subgroup of O(m) and SLm (R) we have the special orthogonal group SO(m) which is deﬁned as It can be shown that O(m) is diﬀeomorphic to SO(m) × O(1). The subset GLm (C) of Cm×m is open so dim(GLm (C)) = 2m2 . As we have already seen in Example 2. As a subgroup of GLm (R) we have the real special linear group SLm (R) given by SLm (R) = {A ∈ Rm×m | det A = 1}.15 the dimension of O(m) is m(m − 1)/2. We will show in Example 3.31.13 that the dimension of the submanifold SLm (R) of Rm×m is m2 − 1. see Exercise 2.9. Another subgroup of GLm (R) is the orthogonal group O(m) = {A ∈ Rm×m | At A = e}. Another subgroup of GLm (C) is the unitary group U(m) given by ¯ U(m) = {A ∈ Cm×m | At A = e}. This means that dim SO(m) = dim O(m) = m(m − 1)/2. . SO(m) = O(m) ∩ SLm (R).

DIFFERENTIABLE MANIFOLDS 19 Calculations similar to those for the orthogonal group show that the dimension of U(m) is m2 .e. As a subgroup of U(m) and SLm (C) we have the special unitary group SU(m) which is deﬁned as It can be shown that U(1) is diﬀeomorphic to the circle S 1 and that U(m) is diﬀeomorphic to SU(m) × U(1).2. . SU(m) = U(m) ∩ SLm (C).9. in the C ∞ -category. when not stating otherwise. This means that dim SU(m) = m2 − 1. see Exercise 2. For the rest of this manuscript we shall assume. that our manifolds and maps are smooth i.

Exercise 2. S 3 and the Lie groups SO(n).8. Let (G. Exercise 2.7. DIFFERENTIABLE MANIFOLDS Exercises Exercise 2.7.2. Let the spheres O(n). Find a proof for Proposition 2. ·) be two Lie groups.10. Exercise 2. U(n) be equipped structures introduced above. SU(n). with their standard diﬀerentiable Proposition 2. Find a proof of Proposition 2.9. = Exercise 2. y:z→ 1 + iz i+z 1 to show that S is a 1-dimensional submanifold of C ∼ R2 . Equip the real line R with the standard topology ˆ and for each odd integer k ∈ Z+ let Ak be the C ω -structure deﬁned on R by the atlas Ak = {(R. = Exercise 2. Use lowing diﬀeomorphisms S 1 ∼ SO(2). Find a proof of Proposition 2. . diﬀerentiable manifolds (R.25. = SO(n) × O(1) ∼ O(n).17. ∗) and (H. Exercise 2. Let S 1 be the unit circle in the complex plane C given by S 1 = {z ∈ C| |z|2 = 1}.3. A Exercise 2.20 2.1. SU(n) × U(1) ∼ U(n). Prove that the product manifold G × H has the structure of a Lie group. Find a proof of Corollary 2. = = Exercise 2. ˆ Prove that the diﬀerentiable structures Ak are all diﬀerent but that the ˆk ) are all diﬀeomorphic.22 to prove the folS 3 ∼ SU(2). Use the maps x : C \ {i} → C and y : C \ {−i} → C with i+z 1 + iz x:z→ . = S 1 .4. Use the implicit function theorem to show that the m-dimensional torus T m = {z ∈ Cm | |z1 | = · · · = |zm | = 1} is a diﬀerentiable submanifold of Cm ∼ R2m . Exercise 2. xk )| xk : p → pk }.6.9.11.5. Exercise 2. Find a proof for Proposition 2.22. Prove that the 2-dimensional sphere S 2 as a diﬀerˆ entiable submanifold of the standard R3 and the Riemann sphere C are diﬀeomorphic.

v. This is a vector space of the same dimension as M . Then it is well known from multi-variable analysis that if v ∈ Rm and f ∈ ε(p) then the directional derivative ∂v f of f at p in the direction of v is given by f (p + tv) − f (p) ∂v f = lim . for an arbitrary element v of Rm we can easily ﬁnd a curve γ : I → Rm such that γ(0) = p and γ(0) = v. t→0 t Furthermore the operator ∂ has the following properties: ∂v (λ · f + µ · g) = λ · ∂v f + µ · ∂v g. at the point p ∈ Rm can be identiﬁed with Rm . One example is given by ˙ This shows that the tangent space. g ∈ ε(p). for all λ. Let Rm be the m-dimensional real vector space with the standard diﬀerentiable structure. We shall now describe how ﬁrst order diﬀerential operators annihilating constants can be interpreted as tangent vectors. Conversely. annihilating constants. ∂v (f · g) = ∂v f · g(p) + f (p) · ∂v g. the space of tangent vectors. w ∈ Rm and f. µ ∈ R.CHAPTER 3 The Tangent Space In this chapter we introduce the notion of the tangent space Tp M of a diﬀerentiable manifold M at a point p ∈ M . For a point p in Rm we denote by ε(p) the set of diﬀerentiable real-valued functions deﬁned locally around p. ∂(λ·v+µ·w) f = λ · ∂v f + µ · ∂w f 21 γ : t → p + t · v. i. when acting on real valued functions locally deﬁned around p. .e. We start by studying the standard Rm and show how a tangent vector v at a point p ∈ Rm can be interpreted as a ﬁrst order linear diﬀerential operator. If p is a point in Rm and γ : I → Rm is a C 1 -curve such that γ(0) = p then the tangent vector γ(t) − γ(0) γ(0) = lim ˙ t→0 t m of γ at p is an element of R .

v = u. Let v. Let α be an arbitrary element of Tp Rm . Theorem 3. This proves that the map Φ is injective. . (λ · α)(f ) = λ · α(f ) for all α. Proof. f ∈ ε(p) and λ ∈ R. Then ∂v f = u. xm ) → 1 we x have α(1) = α(1 · 1) = 1 · α(1) + 1 · α(1) = 2 · α(1). For k = 1. . . ∂xk where ψk ∈ ε(p) with ψk (p) = . v = u. . THE TANGENT SPACE Deﬁnition 3. . . m let xk : ˆ Rm → R be the map given by m xk : (x1 . . w and deﬁne f : Rm → R by f (x) = u. The above mentioned properties of the operator ∂ show that we have a well deﬁned linear map Φ : Rm → Tp Rm given by Φ : v → ∂v . This is given by the addition + and the multiplication · by real numbers satisfying (α + β)(f ) = α(f ) + β(f ). Choose an element u ∈ R such that u.22 3. The set of diﬀential operators Tp Rm carries the structure of a real vector space. w = ∂w f so ∂v = ∂w . Let f ∈ ε(p) and following Lemma 3. . β ∈ Tp Rm .e. w ∈ Rm such that v = w. . . For a point p in Rm the linear map Φ : Rm → Tp Rm deﬁned by Φ : v → ∂v is a vector space isomorphism. the set of mappings α : ε(p) → R such that (i) α(λ · f + µ · g) = λ · α(f ) + µ · α(g). . g ∈ ε(p). For a point p in Rm let Tp Rm be the set of ﬁrst order linear diﬀerential operators at p annihilating constants i. µ ∈ R and f. For the constant function 1 : (x1 . .1. xm ) → xk ˆ and put vk = α(ˆk ).3 locally write m f (x) = f (p) + k=1 (ˆk (x) − pk ) · ψk (x). (ii) α(f · g) = α(f ) · g(p) + f (p) · α(g) for all λ. x . so α(1) = 0.2. x ∂f (p). By the linearity of α it follows that α(c) = 0 for any constant c ∈ R.

where v = (v1 . Then for each k = 1.3. . This means that Φ(v) = ∂v = α so the map Φ : Rm → Tp Rm is surjective and hence a vector space isomorphism. Lemma 3. THE TANGENT SPACE 23 We can now apply the diﬀerential operator α ∈ Tp Rm and yield m α(f ) = α(f (p) + k=1 m (ˆk − pk ) · ψk ) x m = α(f (p)) + k=1 m α(ˆk − pk ) · ψk (p) + x k=1 (ˆk (p) − pk ) · α(ψk ) x = k=1 vk ∂f (p) ∂xk = ∂v f. . γ(0) = gradfp . Let γ : I → U be a curve such that γ(0) = p and γ(0) = v. It follows from the fundamental theorem of calculus that 1 ∂ f (x) − f (p) = (f (p + t(x − p)))dt 0 ∂t m 1 = k=1 (xk − pk ) · 1 0 ∂f (p + t(x − p))dt.3. . . . ˙ ˙ dt . Proof. m there exist functions ψk : U → R such that m f (x) = f (p) + k=1 (xk − pk ) · ψk (x) and ψk (p) = ∂f (p) ∂xk for all x ∈ U . ∂xk Let p be a point in Rm . . ∂xk The statement then immediately follows by setting ψk (x) = 0 ∂f (p + t(x − p))dt. Let p be a point in Rm and f : U → R be a function deﬁned on an open ball around p. 2.2 tells us that v acts ˙ on f by v(f ) = d (f ◦ γ(t))|t=0 = dfp (γ(0)) = gradfp . . . v ∈ Rm be a tangent vector at p and f : U → R be a C 1 -function deﬁned on an open subset U of Rm containing p. The identiﬁcation given by Theorem 3. v . vm ) ∈ Rm .

m} be a basis for Rm . g ∈ ε(p). µ ∈ R and f. Let φ : M → N and ψ : N → N be diﬀerentiable maps between manifolds. The tangent space Tp M of M at p has the structure of a real vector space. f ∈ ε(p) and λ ∈ R. As a direct consequence of ˙ Theorem 3. . Let M be a diﬀerentiable manifold and for a point p ∈ M let ε(p) denote the set of diﬀerentiable functions deﬁned on an open neighborhood of p. . Then the diﬀerential dφp of φ at a point p in M is the map dφp : Tp M → Tφ(p) N such that for all Xp ∈ Tp M and f ∈ ε(φ(p)) (dφp (Xp ))(f ) = Xp (f ◦ φ). . ¯ Proposition 3. (ii) Xp (f · g) = Xp (f ) · g(p) + f (p) · Xp (g) for all λ. Let φ : M → N be a diﬀerentiable map between manifolds. The addition + and the multiplication · by real numbers are simply given by (Xp + Yp )(f ) = Xp (f ) + Yp (f ). . then d(idM )p = idTp M . . (iii) d(ψ ◦ φ)p = dψφ(p) ◦ dφp . . . . Let M be a diﬀerentiable manifold and p be a point on M .4. We shall now use the ideas presented above to generalize to the manifold setting. then for each p ∈ M we have (i) the map dφp : Tp M → Tφ(p) N is linear. (λ · Xp )(f ) = λ · Xp (f ) for all Xp . .6. m} is a basis for the tangent space Tp Rm at p. (ii) if idM : M → M is the identity map.5.2 we have the following useful result.24 3. Deﬁnition 3. Corollary 3.7. Then the set {∂ek | k = 1. We shall now give some motivations for the above deﬁnitions and hopefully convince the reader that they are not only abstract nonsense. Let p be a point in Rm and {ek | k = 1. The set of tangent vectors at p is called the tangent space at p and denoted by Tp M . Yp ∈ Tp M . THE TANGENT SPACE Note that the real number v(f ) is independent of the choice of the curve γ as long as γ(0) = p and γ(0) = v. A tangent vector Xp at p is a map Xp : ε(p) → R such that (i) Xp (λ · f + µ · g) = λ · Xp (f ) + µ · Xp (g). Deﬁnition 3.

The element d(x−1 )x(p) (v) of the tangent space Tp M denoted by γ(0) ˙ is called the tangent to the curve γ at p. Proof. THE TANGENT SPACE 25 Proof. Then the composition ˙ −1 γ = x ◦ c : (−ǫ. Let (U. ǫ) → U is a curve in M through p since γ(0) = p. The statement now follows from Theorem 3. If Xp ∈ Tp M and f ∈ ε(ψ ◦ φ(p)). The statement is a direct consequence of the following relations dψφ(p) ◦ dφp = d(ψ ◦ φ)p = d(idM )p = idTp M .9. The image x(U ) is an open subset of Rm containing x(p) so we can ﬁnd a curve c : (−ǫ. dφp ◦ dψφ(p) = d(φ ◦ ψ)φ(p) = d(idN )φ(p) = idTφ(p) N . Theorem 3. Corollary 3. Then the linear map dxp : Tp M → Tx(p) Rm is a vector space isomorphism. . We are now ready to prove the following interesting theorem. x) be a chart on M . Then the tangent space Tp M at p is an mdimensional real vector space. This is of course a direct generalization of the corresponding result in the classical theory for surfaces in R3 .3.8. Let M m be an m-dimensional diﬀerentable manifold and p be a point in M . Then the diﬀerential dφp : Tp M → Tφ(p) N at p is bijective and (dφp )−1 = dψφ(p) .2 and Corollary 3.8. x) be a local chart around p ∈ M . Let M be an m-dimensional manifold and (U. then (dψφ(p) ◦ dφp )(Xp )(f ) = = = = (dψφ(p) (dφp (Xp )))(f ) (dφp (Xp ))(f ◦ ψ) Xp (f ◦ ψ ◦ φ) d(ψ ◦ φ)p (Xp )(f ). Proof. It follows from the relation γ(0) = d(x−1 )x(p) (v) = Xp ˙ that the tangent space Tp M can be thought of as the set of all tangents to curves through the point p. Let φ : M → N be a diﬀeomorphism with inverse ψ = φ−1 : N → M . The only non-trivial statement is the relation (iii) which is called the chain rule. Then the diﬀerential dxp : Tp M → Tx(p) Rm is a bijective linear map so for a given element Xp ∈ Tp M there exists a tangent vector v in Rm such that dxp (Xp ) = v. ǫ) → x(U ) with c(0) = x(p) and c(0) = v.

We start with the sphere. x) be a local chart on M and {ek | k = 1.4. . x) around p and the curve c : I → x(U ) as long as γ(0) = p and γ(0) = Xp . ǫ) → S m be a curve into the mdimensional unit sphere in Rm+1 with γ(0) = p and γ(0) = v. .11. 2. γ(t) = 0. Let γ : (−ǫ. Example 3. . . We shall now determine the tangent spaces of some of the explicit diﬀerentiable manifolds introduced in Chapter 2. . ˙ ˙ . Proof. m} be the canonical basis for ∂ Rm . The ˙ curve satisﬁes γ(t). γ(t) = 1 and diﬀerentiation yields γ(t).26 3.6 that Xp (f ) = (dxp (Xp ))(f ◦ x−1 ) d = (f ◦ x−1 ◦ c(t))|t=0 dt d (f ◦ γ(t))|t=0 = dt It should be noted that the value Xp (f ) is independent of the choice of the chart (U. For an arbitrary point p in U we deﬁne ( ∂xk )p in Tp M by ∂ ∂xk Then the set p :f → {( ∂f (p) = ∂ek (f ◦ x−1 )(x(p)). ∂xk ∂ )p | k = 1.10. Let M m be a diﬀerentiable manifold. m} ∂xk is a basis for the tangent space Tp M of M at p. . This leads to the following construction. . ˙ Proposition 3. The local chart x : U → x(U ) is a diﬀeomorphism and the diﬀerential (dx−1 )x(p) : Tx(p) Rm → Tp M of the inverse x−1 : x(U ) → U satisﬁes (dx−1 )x(p) (∂ek )(f ) = ∂ek (f ◦ x−1 )(x(p)) ∂ (f ) = ∂xk p for all f ∈ ε(p). THE TANGENT SPACE If f : U → R is a C 1 -function deﬁned locally on U then it follows from Deﬁnition 3. The statement is then a direct consequence of Corollary 3. (U. γ(t) + γ(t). .

k! For this map we have the following well-known result. In order to determine the tangent spaces of the classical Lie groups we need the diﬀerentiable exponential map Exp : Cm×m → Cm×m for matrices given by the following converging power series ∞ γ : t → cos(t|v|) · p + sin(t|v|) · v/|v| Exp : X → k=0 Xk . THE TANGENT SPACE 27 This means that v.13. Proof. Example 3. If X.3. p = 0 so every tangent vector v ∈ Tp S m must be orthogonal to p. If X is a matrix in Rm×m with traceX = 0 then deﬁne a curve A : R → Rm×m by ˙ Then A(0) = e. t ¯ (ii) Exp(X t ) = Exp(X) . This shows that A is a curve into the special linear group SLm (R) and that X is an element of the tangent space Te SLm (R) of SLm (R) at the neutral element e. Y ∈ Cm×m . then (i) det(Exp(X)) = exp(traceX).2 The real general linear group GLm (R) is an open subset of Rm×m so its tangent space Tp GLm (R) at any point p is simply Rm×m . p = 0 then γ : R → S m with is a curve into S m with γ(0) = p and γ(0) = v. The tangent space Te SLm (R) of the special linear group SLm (R) at the neutral element e can be determined as follows. This shows that the ˙ m tangent space Tp S is actually given by Tp S m = {v ∈ Rm+1 | p. v = 0}. Hence the linear space of dimension m2 − 1 is contained in the tangent space Te SLm (R). det(A(s)) = det(Exp(sX)) = exp(trace(sX)) = exp(0) = 1. A(0) = X and A : s → Exp(sX).12. Proposition 3. Let Exp : Cm×m → Cm×m be the exponential map for matrices. See Exercise 3. {X ∈ Rm×m | traceX = 0} . and (iii) Exp(X + Y ) = Exp(X)Exp(Y ) whenever XY = Y X. On the other hand if v = 0 satisﬁes v.

Let A : (−ǫ. ˙ ˙ {A(s)t A(s) + A(s)t A(s)}|s=0 = 0 This shows that B is a curve on the orthogonal group. The orthogonal group O(m) is diﬀeomorphic to SO(m) × {±1} so dim(SO(m)) = dim(O(m)) hence Te SO(m) = Te O(m) = {X ∈ Rm×m | X t + X = 0}. ǫ) → O(m) be a curve into the orthogonal group O(m) such that A(0) = e. This shows that Te SLm (R) = {X ∈ Rm×m | traceX = 0}. Let e be the neutral element of the classical real Lie groups GLm (R). O(m). SLm (R). SO(m). Then their tangent spaces at e are given by Te GLm (R) Te SLm (R) Te O(m) Te SO(m) = = = = Rm×m {X ∈ Rm×m | traceX = 0} {X ∈ Rm×m | X t + X = 0} Te O(m) ∩ Te SLm (R) = Te O(m) .28 3. ǫ) and diﬀerentiation yields ˙ ˙ or equivalently A(0)t + A(0) = 0. Then B(s)t B(s) = = = = = Exp(sX)t Exp(sX) Exp(sX t )Exp(sX) Exp(s(X t + X)) Exp(0) e. Example 3.14. for an arbitrary real skew-symmetric matrix X deﬁne the curve B : R → Rm×m by B : s → Exp(sX). Then A(s)t A(s) = e for all s ∈ (−ǫ. THE TANGENT SPACE The curve given by s → Exp(se) = exp(s)e is not contained in SLm (R) so the dimension of Te SLm (R) is at most m2 − 1. B(0) = e and ˙ B(0) = X so X is an element of Te O(m). This means that each tangent vector of O(m) at e is a skew-symmetric matrix. Hence Te O(m) = {X ∈ Rm×m | X t + X = 0}. This conﬁrms our calculations of the dimension of O(m) in Example 2.15. The dimension of Te O(m) is therefore m(m − 1)/2. Proposition 3.15 since we know that dim(O(m)) = dim(Te O(m)). We have proved the following result. On the other hand.

embeddings and submersions. . x) a chart on M then the inverse x−1 : x(U ) → U of x is a parametrization of the subset U of M . xm+1 . . If γM : (−ǫ. . A diﬀerentiable map φ : M → N between manifolds is said to be an immersion if for each p ∈ M the diﬀerential dφp : Tp M → Tφ(p) N is injective. . . e.3. If M is a diﬀerentiable manifold and (U. n with m < n we have the inclusion map φ : Rm+1 → Rn+1 given by The diﬀerential dφx at x is injective since dφx (v) = (v. . An embedding is an immersion φ : M → N which is a homeomorphism onto its image φ(M ). THE TANGENT SPACE 29 For the classical complex Lie groups similar methods can be used to prove the following.17. SLm (C). γN = φ ◦ γM : (−ǫ. U(m). . Proof. . It is easily seen that even the restriction φ|S m : S m → S n of φ to the m-dimensional unit sphere S m in Rm+1 is an embedding. . An immersion φ : U → M is called a local parametrization of M . . Then their tangent spaces at e are given by Te GLm (C) Te SLm (C) Te U(m) Te SU(m) = = = = Cm×m {Z ∈ Cm×m | traceZ = 0} ¯ {Z ∈ Cm×m | Z t + Z = 0} Te U(m) ∩ Te SLm (C). the so called immersions.16. Let M be an m-dimensional diﬀerentiable manifold and U be an open subset of Rm .18. . 0). ˙ ˙ Deﬁnition 3. See Exercise 3. Proposition 3. The map φ is obviously a homeomorphism onto its image φ(Rm+1 ) hence an embedding. For positive integers m. φ : (x1 . Deﬁnition 3.4 The rest of this chapter is devoted to the introduction of special types of diﬀerentiable maps. ǫ) → N . The interpretation of the tangent spaces given above shows that the diﬀerential dφp : Tp M → Tφ(p) N of φ at p maps the tangent γM (0) at p to the tangent γN (0) at φ(p) i. ˙ ˙ dφp (γM (0)) = γN (0). . ǫ) → M is a curve on M such that γM (0) = p then a diﬀerentiable map φ : M → N takes γM to a curve on N with γN (0) = φ(p). and SU(m). xm+1 ) → (x1 . Let e be the neutral element of the classical complex Lie groups GLm (C). 0. 0).

For the diﬀerential of φk we have ˙ d d (φk ◦ γw (t))|t=0 = (wk eikt )|t=0 = kiwk . p p − q = 2p p. Let q ∈ S 3 be a quaternion of unit length and φq : S 1 → S 3 be the map deﬁned by φq : z → qz. dt dt Then |dφq (γw (0))| = |qwi| = |q||w| = 1 implies that the diﬀerential dφq ˙ is injective. By diﬀerentiating we yield ˙ d d (φq (γw (t)))|t=0 = (qweit )|t=0 = qiw. γw (0) = iw and φq (γw (t)) = qweit . q − q = (2ppt − e)q (2ppt − e). φ : p → ρp Lp = {λp ∈ Rm+1 | λ ∈ R} that the matrix in R(m+1)×(m+1) corresponding to ρp is just We shall now show that the map φ : S m → R(m+1)×(m+1) given by . It is easily seen that φk is an embedding if and only if k = ±1. For a point p ∈ S m let be the line through the origin generated by p and ρp : Rm+1 → Rm+1 be the reﬂection about the line Lp . so the map φk is an immersion.30 3. Then γw (0) = w. Let S 1 be the unit circle in the complex plane C.20. It is easily checked that the reﬂection about the line Lp is given by ρp : q → 2 q.21. It is easily checked that the immersion φq is an embedding. Let m be a positive integer and S m be the mdimensional unit sphere in Rm+1 . For w ∈ S 1 let γw : R → S 1 be the curve given by γw (t) = weit . dt dt This shows that the diﬀerential (dφk )w : Tw S 1 ∼ R → Twk C ∼ R2 is = = injective. It then follows from the equations ρp (q) = 2 q. THE TANGENT SPACE Example 3. p p − q. For a non-zero integer k ∈ Z deﬁne φk : S 1 → C by φk : z → z k . dφq (γw (0)) = ˙ In the next example we construct an interesting embedding of the real projective space RP m into the vector space Sym(Rm+1 ) of the real symmetric (m + 1) × (m + 1) matrices. the set of linear endomorphisms of Rm+1 which can be identiﬁed with R(m+1)×(m+1) . Then ρp is an element of End(Rm+1 ) i. For a point w ∈ S 1 let γw : R → S 1 be the curve with γw : t → weit . Then γw (0) = w and γw (0) = iw. (dφk )w (γw (0)) = ˙ Example 3.19. Example 3.e.

x ≡ y if and only if x = ±y. ρq of Rm+1 are diﬀerent in this case. γ(t) γ(t) − q) (dφ)p (γ(0)) = ˙ This means that and d (φ ◦ γ(t))|t=0 dt = (q → 2 q. with a = α(0) ˙ ˙ and b = β(0). For γ ∈ {α. that is α(0) = p = β(0). On the other hand. Ann.3. But these are just the eigenspaces of ρp and ρq with the eigenvalue +1. respectively. Diﬀerentiable Manifolds. Let p be an arbitrary point on S m and α. β : I → S m be two curves meeting at p. ˙ ˙ dφp (a) = (q → 2 q. of Math. If a = b then dφp (a) = dφp (b) so the diﬀerential dφp is injective. p a) dφp (b) = (q → 2 q. hence the map φ : S m → R(m+1)×(m+1) is an immersion. β} we have so φ ◦ γ : t → (q → 2 q. THE TANGENT SPACE 31 is an immersion. For 1 ≤ r ≤ ∞ let M be an m-dimensional C r -manifold.22. Then there exists a C r -embedding φ : M → R2m+1 into the (2m + 1)-dimensional real vector space R2m+1 . . This shows that the image ψ(RP m ) = φ(S m ) is not only contained in the linear space Sym(Rm+1 ) but also in the orthogonal group O(m + 1) which we know is a submanifold of R(m+1)×(m+1) The following result was proved by Hassler Whitney in his famous paper. Deep Result 3. γ(0) γ(0) + 2 q. γ(0) γ(0)). b p + 2 q. a p + 2 q. 645-680. This means that the image φ(S m ) can be identiﬁed with the quotient space S m / ≡ where ≡ is the equivalence relation deﬁned by This of course is the real projective space RP m so the map φ induces an embedding ψ : RP m → Sym(Rm+1 ) with For each p ∈ S m the reﬂection ρp : Rm+1 → Rm+1 about the line Lp satisﬁes ρp · ρt = e. This shows that the linear endomorphisms ρp . p ψ : [p] → ρp . p b). If the points p. if p and q are parallel then p = ±q hence ρp = ρq . q ∈ S m are linearly independent. then the lines Lp and Lq are diﬀerent. 37 (1936).

A point q ∈ φ(N ) is said to be a regular value of φ if every point of the pre-image φ−1 ({q}) of {q} is regular and a critical value otherwise. This is true for each point p ∈ φ−1 ({q}) so we have proven that φ−1 ({q}) is an (n − m)-dimensional submanifold of N n . ψq ) are charts so the diﬀerentials (dψq )q and −1 ˆ (dψp )0 are bijective. If p is a point in M such that the diﬀerential dφp : Tp M → Tφ(p) N at p is bijective then there exist open neighborhoods Up around p and Uq around q = φ(p) such that ψ = φ|Up : Up → Uq is bijective and the inverse ψ −1 : Uq → Up is diﬀerentiable. This means that the diﬀerential dφ0 is surjective since dφp is. See Exercise 3. A point p ∈ N is said to be critical for φ if the diﬀerential n is not of full rank. It then follows from the implicit function theorem 2. For a point p ∈ φ−1 ({q}) we choose a chart (Up .24. Let φ : N n → M be a diﬀerentiable map between manifolds such that n > m.e. Deﬁnition 3. Tp φ−1 ({q}) = Ker dφp . ψq ) be a chart on M with q ∈ Vq and ψq (q) = 0. If q ∈ φ(N ) is a regular value.8 We shall now generalize the classical implicit function theorem to manifolds. The tangent space Tp φ−1 ({q}) of φ−1 ({q}) at p is the kernel of the diﬀerential dφp i. ψp (p) = 0 and φ(Up ) ⊂ Vq . Proof.12 that ψp (φ−1 ({q})∩Up ) is an (n−m)-dimensional submanifold of ψp (Up ). . at the point 0 is given by −1 ˆ dφ0 = (dψq )q ◦ dφp ◦ (dψp )0 : T0 Rn → T0 Rm .23 (The Inverse Function Theorem). then the pre-image φ−1 ({q}) of q is an (n−m)-dimensional submanifold of N n . ψp ) on N such that p ∈ Up .25 (The Implicit Function Theorem). Let (Vq . Hence φ−1 ({q})∩Up is an (n−m)-dimensional submanifold of Up .32 3. n be positive natural numbers and φ : N → M m be a diﬀerentiable map between manifolds. Let φ : M → N be a diﬀerentiable map between manifolds with dim M = dim N . and regular if it is not critical. Let m. Proof. THE TANGENT SPACE The classical inverse function theorem generalizes to the manifold setting as follows. The diﬀerential of the map ˆ φ = ψq ◦ φ ◦ ψ −1 |ψ (U ) : ψp (Up ) → Rm p p p The pairs (Up . For this we need the following deﬁnition. Theorem 3. m dφp : Tp N → Tφ(p) M Theorem 3. ψp ) and (Vq .

Example 3. . ˙ dt dt −1 This implies that Tp φ ({q}) is contained in and has the same dimension as the kernel of dφp . . q∈S 2 φ : (x. For positive integers m. . y)| θ ∈ R}. vn ) = (v1 . Then dq d (dφ)p (γ(0)) = (φ ◦ γ(t))|t=0 = |t=0 = 0. . . . vm ). The Hopf map φ : S 3 → S 2 is given by = The map φ and its diﬀerential dφp : Tp S 3 → Tφ(p) S 2 are surjective for each p ∈ S 3 . Its diﬀerential dπx at a point x is surjective since dπx (v1 . y . . . ǫ) → φ−1 ({q}) be a curve. . respectively. Deﬁnition 3. . so Tp φ−1 ({q}) = Ker dφp . |x|2 − |y|2 )}) = {eiθ (x. This means that the projection is a submersion. such that γ(0) = p. . Let S 3 and S 2 be the unit spheres in C2 and C × R ∼ R3 .3.26. If m. y) → (2x¯. . xm ). y φ−1 ({(2x¯. n with n ≥ m a map φ : N n → M m between two manifolds is said to be a submersion if for each p ∈ N the diﬀerential dφp : Tp N → Tφ(p) M is surjective. THE TANGENT SPACE 33 Let γ : (−ǫ.27. |x|2 − |y|2 ). . xn ) → (x1 . They are actually the great circles given by This means that the 3-dimensional sphere S 3 is a disjoint union of great circles S3 = φ−1 ({q}). n ∈ N such that n ≥ m then we have the projection map π : Rn → Rm given by π : (x1 . . An important submersion between spheres is given by the following. . . This implies that each point q ∈ S 2 is a regular value and the ﬁbres of φ are 1-dimensional submanifolds of S 3 .

.2. . . 3 ﬁnd an explicit formula for the curve γk : R → SL2 (R) given by γk : s → Exp(sXk ).3. Exercise 3.23. . . For each k = 1. . Find a proof for Proposition 3. y) → (2x¯. |x|2 − |y|2 ) is a submersion. For each k = 1. Find a proof for Theorem 3. Let p be an arbitrary point on the unit sphere S 2n+1 of Cn+1 ∼ R2n+2 .34 3. Z2 = Z1 = i 0 1 0 i 0 0 −i Z3 = form a basis for the tangent space Te SU(2) of the special unitary group SU(2) at e. Exercise 3. . y φ : (x1 . . Prove that the Hopf-map φ : S 3 → S 2 with φ : (x. Exercise 3. submersions or embeddings.4. Exercise 3. . Exercise 3. Find a proof for Proposition 3. Exercise 3. 2. For which k ∈ N0 are φk . .1. Prove that the matrices 0 i 0 −1 .16. 3 ﬁnd an explicit formula for the curve γk : R → SU(2) given by Exercise 3.8. ψk : z → z k . . Prove that the map φ : Rm → Cm given by is a parametrization of the m-dimensional torus T m in Cm .7. Exercise 3.6. THE TANGENT SPACE Exercises Exercise 3. For each k ∈ N0 deﬁne φk : C → C and ψk : C∗ → C by φk . eixm ) γk : s → Exp(sZk ). 2. xm ) → (eix1 . ψk immersions. Prove that the matrices 0 −1 1 0 . X2 = X1 = 1 0 0 −1 0 1 1 0 X3 = form a basis for the tangent space Te SL2 (R) of the real special linear group SL2 (R) at e.5.12.9. Determine the tangent space Tp S 2n+1 and show that = it contains an n-dimensional complex subspace of Cn+1 .

Y : Rm → T Rm can be written as m m ∂ ∂ X= ak and Y = . = k. v)| p ∈ Rm . v ∈ Tp Rm }. Y ] acts on f as follows: [X. this is the object that we get by glueing at each point p of M the corresponding tangent space Tp M .10 two vector ﬁelds X.l=1 ak ∂ ∂ ∂ ∂ (bl ) − bk (al ) (f ) ∂xk ∂xl ∂xk ∂xl 35 . X(p)). a vector ﬁeld X on Rm is a smooth map X : Rm → Rm but we would like to view it as a map X : Rm → T Rm into the tangent bundle and with abuse of notation write Following Proposition 3. Intuitively.CHAPTER 4 The Tangent Bundle In this chapter we introduce the tangent bundle T M of a diﬀerentiable manifold M . Y ](f ) = X(Y (f )) − Y (X(f )) m T Rm = {(p. bk ∂xk ∂xk k=1 k=1 where ak . bk : Rm → R are smooth functions deﬁned on Rm . v) → p X : p → (p. π : (p. We have already seen that for a point p ∈ Rm the tangent space Tp Rm can be identiﬁed with the m-dimensional vector space Rm . This means that if we at each point p ∈ Rm glue Tp Rm to Rm we yield the so called tangent bundle of Rm For this we have the natural projection π : T Rm → Rm deﬁned by and for a point p ∈ M the ﬁbre π −1 ({p}) over p is precisely the tangent space Tp Rm at p. The diﬀerentiable structure on M induces a diﬀerentiable structure on the tangent bundle T M turning it into a diﬀerentiable manifold. If f : Rm → R is another such function the commutator [X. Classically.

M. of dimension n. M. A continuous map σ : M → E is called a section of the bundle (E. We shall now generalize to the manifold setting. π) over M . Let (E. A topological vector bundle (E. π) where is the projection map π : (p. is said to be trivial if there exists a global bundle chart ψ : E → M × Rn . ∂xk ∂xk ∂xl This shows that the commutator [X. M. The identity map ψ : M × Rn → M × Rn is a global bundle chart so the bundle is trivial. A collection of bundle charts is called a bundle atlas for (E. Deﬁnition 4. π) is called an n-dimensional topological vector bundle over M if (i) for each p ∈ M the ﬁbre Ep = π −1 ({p}) is an n-dimensional vector space. This leads us ﬁrst to the following notion of a topological vector bundle.36 4.l=1 ∂2 ∂al ∂ − bk al (f ) ∂xk ∂xl ∂xk ∂xl ∂bl ∂al ∂ ak − bk (f ). (ii) for each p ∈ M there exists a bundle chart (π −1 (U ). π) be an n-dimensional topological vector bundle over M . ψα )| α ∈ I} π : M × Rn → M .1. Deﬁnition 4. Let E and M be topological manifolds and π : E → M be a continuous surjective map. M.2. M. Deﬁnition 4. β ∈ I there exists a map Aα. M. v) → p. The triple (E. THE TANGENT BUNDLE m = k.3. π) if M = ∪α Uα and for α. ψ) consisting of the pre-image π −1 (U ) of an open neighbourhood U of p and a homeomorphism ψ : π −1 (U ) → U × Rn such that for all q ∈ U the map ψq = ψ|Eq : Eq → {q} × Rn is a vector space isomorphism. If n is a positive integer and M is a topological manifold then we have the n-dimensional vector bundle (M × Rn . π) if π ◦ σ(p) = p for each p ∈ M . Y ] is a smooth vector ﬁeld on Rm .β : Uα ∩ Uβ → GLn (R) such B = {(π −1 (Uα ).l=1 ak ∂bl ∂ ∂2 + ak b l ∂xk ∂xl ∂xk ∂xl −bk m = k.

M.β (p))(v)). Deﬁne the set T M by T M = {(p. Deﬁnition 4. v)| p ∈ M. Let M m be a diﬀerentiable manifold with maximal ˆ atlas A. when not stating otherwise. . v ∈ Tp M } and let π : T M → M be the projection map satisfying Then the ﬁbre π −1 ({p}) over a point p ∈ M is the m-dimensional tangent space Tp M . A bundle atlas B for (E. π : (p. Example 4. . The triple (T M. π) be a vector bundle over a manifold M . . . π) is called the tangent bundle of M .β | α. By C ∞ (E) we denote the set of all smooth sections of (E. . Let E and M be diﬀerentiable manifolds and π : E → M be a diﬀerentiable map such that (E. (ii) (f · v)p = f (p) · vp for all v. Deﬁnition 4. M. . Let (E. A diﬀerentiable vector bundle is a topological vector bundle together with a maximal diﬀerentiable bundle atlas. Then we deﬁne the operations + and · on the set C ∞ (E) of smooth sections of (E. v) → (p. With the above deﬁned operations C ∞ (E) becomes a module over C (M ) and in particular a vector space over the real numbers as the constant functions in C ∞ (M ). π) is said to be diﬀerentiable if the corresponding transition maps are diﬀerentiable. . that all our vector bundles are smooth. If U is an open subset of M then a set {v1 . v) → p. ∞ (p. . THE TANGENT BUNDLE 37 that the corresponding continuous map −1 ψβ ◦ ψα |(Uα ∩Uβ )×Rn : (Uα ∩ Uβ ) × Rn → (Uα ∩ Uβ ) × Rn is given by The elements of {Aα. . M. M.6. vn : U → E on U is called a local frame for E if for each p ∈ U the set {(v1 )p . π) is an n-dimensional topological vector bundle. π) by (i) (v + w)p = vp + wp . β ∈ I} are called the transition maps of the bundle atlas B. vn } of smooth sections v1 . . M.4. π). From now on we shall assume.5. . . (vn )p } is a basis for the vector space Ep . .4. (Aα. w ∈ C ∞ (E) and f ∈ C ∞ (M ). M.

The restriction xp = x|Tp M : Tp M → {p} × Rm to the tangent space ¯ ¯ Tp M is given by m xp : ¯ k=1 vk ∂ ∂xk p → (v1 . . A∗ ) is a diﬀerentiable manifold. We shall now describe how the tangent bundle (T M. . (x (a))bk ). vm )). .10 shows that the map x∗ is well-deﬁned. . . ¯ k=1 vk ∂ ∂xk p ) → (p. y) are two charts in A such that p ∈ U ∩ V then the transition map is given by (y ∗ ) ◦ (x∗ )−1 : x∗ (π −1 (U ∩ V )) → Rm × Rm m Proposition 3. . M. THE TANGENT BUNDLE We shall now equip T M with the structure of a smooth manifold. . x) and (V. k=1 p is a basis for a topology TT M on T M and (π −1 (U ). . . . b) → (y ◦ x−1 (a). . vm )). x∗ )| (U. . ˆ If (U. ∂xk ∂xk k=1 m Since we are assuming that y ◦ x−1 is diﬀerentiable it follows that (y ∗ ) ◦ (x∗ )−1 is also diﬀerentiable. This means that ˆ A∗ = {(π −1 (U ). TT M ). . . vm ). x) ∈ A} is a C r -atlas on T M so (T M. x∗ ) is a chart on the 2m-dimensional topological manifold (T M. . x) ∈ A and W ⊂ x(U ) × Rm open} (a.38 4. . It is trivial that the surjective projection map π : T M → M is diﬀerentiable. (v1 . The collection ˆ {(x∗ )−1 (W ) ⊂ T M | (U. . k=1 ∂y1 −1 ∂ym −1 (x (a))bk . For a local chart x : U → Rm in the maximal ˆ atlas A of M we deﬁne x : π −1 (U ) → U × Rm by ¯ m x : (p. ˆ For every local chart x : U → Rm from the maximal atlas A of M we deﬁne a local chart on the tangent bundle T M by the formula m x∗ : π −1 (U ) → Rm × Rm vk ∂ ∂xk ) → (x(p). (v1 . x∗ : (p. For each point p ∈ M the ﬁbre π −1 ({p}) is the tangent space Tp M and hence an mdimensional vector space. π) can be given the structure of a diﬀerentiable vector bundle.

π) together with the maximal bundle atlas B deﬁned by B is a diﬀerentiable vector bundle. (X3 )p = (z. The group structure on S 3 can be used to extend vectors in Te S 3 to vector ﬁelds on S 3 as follows. v2 . . Then deﬁne the vector ﬁelds X1 . i) = (iw. 0) = (iz. It is not diﬃcult to see that ˆ B = {(π −1 (U ). M. v2 = (0. iz). Let M be a diﬀerentiable manifold. Example 4. ¯ This makes (S 3 . (X2 )p = (z. This implies that the map x : π −1 (U ) → U × Rm ¯ is a local bundle chart. 3 deﬁne the curves γk : R → S 3 with γk : t → cos t · (1. Deﬁnition 4. 0). i) and for k = 1. We have seen earlier that the 3-sphere S 3 in H ∼ C2 = carries a group structure · given by ¯ (z. 0). 0) + sin t · vk . The set of smooth vector ﬁelds X : M → T M is denoted by C ∞ (T M ). ·) into a Lie group with neutral element e = (1. x)| (U. w) · (α. X3 ∈ C ∞ (T S 3 ) by d (Lp (γk (t)))|t=0 . w) · (i. Then γk (0) = e and γk (0) = vk so v1 . w) ∈ S 3 the values of Xk at p is given by (Xk )p = (dLp )e (vk ) = (X1 )p = (z. β) = (zα − wβ. M.1.8. then a section X : M → T M of the tangent bundle is called a vector ﬁeld. 2. THE TANGENT BUNDLE 39 so it is a vector space isomorphism. They are linearily independent so they generate Te S 3 . π) into an m-dimensional topological vector bundle. −iw). X2 . zβ + wα). It immediately follows from above that ˆ (T M. see Exercise 4. x) ∈ A} ¯ is a bundle atlas turning (T M.7. dt It is left as an exercise for the reader to show that at a point p = (z. Put v1 = (i. v3 are elements of the tangent ˙ space Te S 3 .4. Our next aim is to introduce the Lie bracket on the set of vector ﬁelds C ∞ (T M ) on M . w) · (0. For p ∈ S 3 let Lp : S 3 → S 3 be the left translation on S 3 by p given by Lp : q → p · q. 1) and v3 = (0. 1) = (−w. z). w) · (0.

Lemma 4.10. X.10 implies that if X. Proposition 4. Y ]p (f · g) = [X.12 we shall prove that this section is smooth. f. For two vector ﬁelds X. Y ]p (f ). Y ]p (f ) · g(p) + f (p) · [X. In Proposition 4. Y ] : p → [X. [X. For this we need the following technical lemma. . Y ]p (g). Y ]p (g). Y ]p (g). Y ]p (λf + µg) = Xp (Y (λf + µg)) − Yp (X(λf + µg)) = λXp (Y (f )) + µXp (Y (g)) − λYp (X(f )) − µYp (X(g)) = λ[X. R) and λ. Y ]p actually is an element of the tangent space Tp M . Proposition 4. Y ]p (λ · f + µ · g) = λ · [X. Y ]p (f ) + µ[X. (ii) if (U. am : U → R given by m ∂ X|U = ak . Y ∈ C ∞ (T M ) we deﬁne the Lie bracket [X. . = = = = = [X. . THE TANGENT BUNDLE Deﬁnition 4. Y ]p is a section of the tangent bundle. . Proof. Then the following conditions are equivalent (i) the section X is smooth. Y ] : M → T M given by [X. Y ∈ C ∞ (T M ) be vector ﬁelds on M . Y ]p : C ∞ (M ) → R of X and Y at p ∈ M by The next result shows that the Lie bracket [X. Let M be a smooth manifold. (ii) [X. Y ]p (f ) + µ · [X. Let M be a smooth manifold. Y are vector ﬁelds on M then the map [X. [X. Y ]p (f ) = Xp (Y (f )) − Yp (X(f )). g ∈ C ∞ (M.11.9.40 4. . Y ]p (g) + g(p)[X. x) is a chart on M then the functions a1 . µ ∈ R. Let M m be a smooth manifold and X : M → T M be a section of T M . Y ]p (f · g) Xp (Y (f · g)) − Yp (X(f · g)) Xp (f · Y (g) + g · Y (f )) − Yp (f · X(g) + g · X(f )) Xp (f )Yp (g) + f (p)Xp (Y (g)) + Xp (g)Yp (f ) + g(p)Xp (Y (f )) −Yp (f )Xp (g) − f (p)Yp (X(g)) − Yp (g)Xp (f ) − g(p)Yp (X(f )) f (p){Xp (Y (g)) − Yp (X(g))} + g(p){Xp (Y (f )) − Yp (X(f ))} f (p)[X. Then (i) [X. ∂xk k=1 are smooth.

4. THE TANGENT BUNDLE

41

(iii) if f : V → R deﬁned on an open subset V of M is smooth, then the function X(f ) : V → R with X(f )(p) = Xp (f ) is smooth. Proof. (i) ⇒ (ii): The functions ak = πm+k ◦ x∗ ◦ X|U : U → T M → x(U ) × Rm → R are restrictions of compositions of smooth maps so therefore smooth. (ii) ⇒ (iii): Let (U, x) be a chart on M such that U is contained in V . By assumption the map

m

X(f |U ) =

ai

i=1

∂f ∂xi

is smooth. This is true for each such chart (U, x) so the function X(f ) is smooth. (iii) ⇒ (i): Note that the smoothness of the section X is equivalent to x∗ ◦ X|U : U → R2m being smooth for all charts (U, x) on M . On the other hand, this is equivalent to x∗ = πk ◦ x∗ ◦ X|U : U → R being k smooth for all k = 1, 2, . . . , 2m and all charts (U, x) on M . It is trivial that the coordinates x∗ = xk for k = 1, . . . , m are smooth. But x∗ k m+k = ak = X(xk ) for k = 1, . . . , m hence also smooth by assumption. Proposition 4.12. Let M be a manifold and X, Y ∈ C ∞ (T M ) be vector ﬁelds on M . Then the section [X, Y ] : M → T M of the tangent bundle given by [X, Y ] : p → [X, Y ]p is smooth. Proof. Let f : M → R be an arbitrary smooth function on M then [X, Y ](f ) = X(Y (f )) − Y (X(f )) is smooth so it follows from Lemma 4.11 that the section [X, Y ] is smooth. For later use we prove the following useful result. Lemma 4.13. Let M be a smooth manifold and [, ] be the Lie bracket on the tangent bundle T M . Then (i) [X, f · Y ] = X(f ) · Y + f · [X, Y ], (ii) [f · X, Y ] = f · [X, Y ] − Y (f ) · X for all X, Y ∈ C ∞ (T M ) and f ∈ C ∞ (M ). Proof. If g ∈ C ∞ (M ), then [X, f · Y ](g) = X(f · Y (g)) − f · Y (X(g)) = X(f ) · Y (g) + f · X(Y (g)) − f · Y (X(g)) = (X(f ) · Y + f · [X, Y ])(g) This proves the ﬁrst statement and the second follows from the skewsymmetry of the Lie bracket.

42

4. THE TANGENT BUNDLE

Deﬁnition 4.14. A real vector space (V, +, ·) equipped with an operation [, ] : V × V → V is said to be a real Lie algebra if the following relations hold (i) [λX + µY, Z] = λ[X, Z] + µ[Y, Z], (ii) [X, Y ] = −[Y, X], (iii) [X, [Y, Z]] + [Z, [X, Y ]] + [Y, [Z, X]] = 0 for all X, Y, Z ∈ V and λ, µ ∈ R. The equation (iii) is called the Jacobi identity. Theorem 4.15. Let M be a smooth manifold. The vector space C (T M ) of smooth vector ﬁelds on M equipped with the Lie bracket [, ] : C ∞ (T M ) × C ∞ (T M ) → C ∞ (T M ) is a real Lie algebra.

∞

Proof. See exercise 4.4. If φ : M → N is a surjective map between diﬀerentiable manifolds ¯ then two vector ﬁelds X ∈ C ∞ (T M ), X ∈ C ∞ (T N ) are said to be ¯ φ(p) for all p ∈ M . In that case we write φ-related if dφp (X) = X ¯ X = dφ(X). Proposition 4.16. Let φ : M → N be a map between diﬀerentiable ¯ ¯ ¯ manifolds, X, Y ∈ C ∞ (T M ), X, Y ∈ C ∞ (T N ) such that X = dφ(X) ¯ = dφ(Y ). Then and Y ¯ ¯ [X, Y ] = dφ([X, Y ]). Proof. Let f : N → R be a smooth function, then ¯ ¯ [X, Y ](f ) = dφ(X)(dφ(Y )(f )) − dφ(Y )(dφ(X)(f )) = X(dφ(Y )(f ) ◦ φ) − Y (dφ(X)(f ) ◦ φ) = X(Y (f ◦ φ)) − Y (X(f ◦ φ)) = [X, Y ](f ◦ φ) = dφ([X, Y ])(f ).

Proposition 4.17. Let φ : M → N be a smooth bijective map between diﬀerentiable manifolds. If X, Y ∈ C ∞ (T M ) are vector ﬁelds on M , then (i) dφ(X) ∈ C ∞ (T N ), (ii) the map dφ : C ∞ (T M ) → C ∞ (T N ) is a Lie algebra homomorphism i.e. [dφ(X), dφ(Y )] = dφ([X, Y ]). Proof. The fact that the map φ is bijective implies that dφ(X) is a section of the tangent bundle. That dφ(X) ∈ C ∞ (T N ) follows

4. THE TANGENT BUNDLE

43

directly from the fact that dφ(X)(f )(φ(p)) = X(f ◦ φ)(p). The last statement is a direct consequence of Proposition 4.16. Deﬁnition 4.18. Let M be a smooth manifold. Two vector ﬁelds X, Y ∈ C ∞ (T M ) are said to commute if [X, Y ] = 0. Let (U, x) be local coordinates on a manifold M and let { ∂ | k = 1, 2, . . . , m} ∂xk

be the induced local frame for the tangent bundle. For k = 1, 2, . . . , m the vector ﬁeld ∂/∂xk is x-related to the constant coordinate vector ﬁeld ek in Rm . This implies that dx([ ∂ ∂ , ]) = [ek , el ] = 0. ∂xk ∂xl

Hence the local frame ﬁelds commute. Deﬁnition 4.19. Let G be a Lie group with neutral element e. For p ∈ G let Lp : G → G be the left translation by p with Lp : q → pq. A vector ﬁeld X ∈ C ∞ (T G) is said to be left invariant if dLp (X) = X for all p ∈ G, or equivalently, Xpq = (dLp )q (Xq ) for all p, q ∈ G. The set of all left invariant vector ﬁelds on G is called the Lie algebra of G and is denoted by g. The Lie algebras of the classical Lie groups introduced earlier are denoted by glm (R), slm (R), o(m), so(m), glm (C), slm (C), u(m) and su(m), respectively. Proposition 4.20. Let G be a Lie group and g be the Lie algebra of G. Then g is a Lie subalgebra of C ∞ (T G) i.e. if X, Y ∈ g then [X, Y ] ∈ g, Proof. If p ∈ G then dLp ([X, Y ]) = [dLp (X), dLp (Y )] = [X, Y ] for all X, Y ∈ g. This proves that the Lie bracket [X, Y ] of two left invariant vector ﬁelds X, Y is left invariant and thereby that g is a Lie subalgebra of C ∞ (T G). Note that if X is a left invariant vector ﬁeld on G then Xp = (dLp )e (Xe )

Let G be a Lie group. Then the tangent bundle T G of G is trivial. Let X. Y ]e (f ) = Xe (Y (f )) − Ye (X(f )) = dfe (Xe · Ye − Ye · Xe ). Y ∗ ]e .21.44 4. Then the derivative Xp (f ) is given by d Xp (f ) = (f (p · Exp(sXe )))|s=0 = dfp (p · Xe ) = dfp (Xp ).22. Ye ] = Xe · Ye − Ye · Xe where · is the usual matrix multiplication. Theorem 4. ] : Te G × Te G → Te G is given by [Xe . Y ] = [X ∗ . Xe ) + dfe (Ye · Xe ). For the other real classical Lie groups the result follows from the fact that they are all subgroups of GLm (R). hence [X. Y ∈ glm (R) be left invariant vector ﬁelds on GLm (R). ds The real general linear group GLm (R) is an open subset of Rm×m so we can use well-known rules from calculus and the second derivative Ye (X(f )) is obtained as follows: Ye (X(f )) = d (XExp(tYe ) (f ))|t=0 dt d = (dfExp(tYe ) (Exp(tYe ) · Xe ))|t=0 dt 2 = d fe (Ye . Proposition 4. We shall prove the result for the case when G is the real general linear group GLm (R). Proof. THE TANGENT BUNDLE so the value Xp of X at p ∈ G is completely determined by the value Xe of X at e. This means that the map ∗ : Te G → g given by is a vector space isomorphism and that we can deﬁne a Lie bracket [. ∗ : X → (X ∗ : p → (dLp )e (X)) The Hessian d2 fe of f is symmetric. ] : Te G × Te G → Te G on the tangent space Te G by [X. . The same proof can be used for the complex cases. Then the Lie bracket on Te G [. Let G be one of the classical Lie groups and Te G be the tangent space of G at the neutral element e. f : U → R be a function deﬁned locally around the identity element e ∈ GLm (R) and p be an arbitrary point in U .

vm )) is a global bundle chart so the tangent bundle T G is trivial. . . Xm } be a basis for Te G. . (v1 . . Then the map ψ : T G → G × Rm given by m ψ : (p.4. . Let {X1 . . k=1 ∗ vk · (Xk )p ) → (p. THE TANGENT BUNDLE 45 Proof. . . .

Let m be a positive integer an SO(m) be the corresponding special orthogonal group. . (i) Find a basis for the tangent space Te SO(m). . (V. Use the Hairy Ball Theorem to prove that the tangent bundles T S 2m of the even-dimensional spheres S 2m are not trivial. Let (M m . . ∂xi ∂yj Exercise 4. Let m be a positive integer. Exercise 4. m} ∂xi ∂yj for T M on U ∩ V are related by ∂ = ∂xi m j=1 f = y ◦ x−1 : x(U ∩ V ) → Rm ∂(fj ◦ x) ∂ · .4. THE TANGENT BUNDLE Exercises ˆ Exercise 4. Exercise 4. m} and { | j = 1.15. . . (iii) determine all smooth vector ﬁelds on SO(2). y) ˆ be two charts in A such that U ∩ V = ∅. x). . Construct a non-vanishing vector ﬁeld X ∈ C ∞ (T S 2m+1 ) on the odd-dimensional sphere S 2m+1 . Show that the local frames ∂ ∂ { | i = 1.46 4. . . The Hairy Ball Theorem. A) be a smooth manifold and (U. Find a proof for Theorem 4.3. . (ii) construct a non-vanishing vector ﬁeld Z ∈ C ∞ (T SO(m)).2. Let be the corresponding transition map. Then there does not exist a continuous non-vanishing vector ﬁeld X ∈ C 0 (T S 2m ) on the even dimensional sphere S 2m .1. Let m be a positive integer.

Let M be a smooth manifold. . g ∈ C ∞ (M ) and k = 1. . . . . f. . . The next general result provides us with the most important property of tensor ﬁelds. . s) is a map A : Cr (T M ) → Cs (T M ) ∞ which is multi-linear over C (M ) i. s) and p ∈ M . . . .2. . Deﬁnition 5. g). Let X1 . For the rest of this work we shall for A(X1 ⊗ · · · ⊗ Xr ) use the notation A(X1 . It shows that the value of A(X1 . . ∞ ∞ Proposition 5. . C ∞ (M ) denote the commutative ring of smooth functions on M and C ∞ (T M ) be the set of smooth vector ﬁelds on M forming a module over C ∞ (M ). satisfying A(X1 ⊗ · · · ⊗ Xk−1 ⊗ (f · Y + g · Z) ⊗ Xk+1 ⊗ · · · ⊗ Xr ) = f · A(X1 ⊗ · · · ⊗ Xk−1 ⊗ Y ⊗ Xk+1 ⊗ · · · ⊗ Xr ) +g · A(X1 ⊗ · · · ⊗ Xk−1 ⊗ Z ⊗ Xk+1 ⊗ · · · ⊗ Xr ) ∞ Cr (T M ) = C ∞ (T M ) ⊗ · · · ⊗ C ∞ (T M ) for all X1 . Xr ) at a point p ∈ M only depends on the values of the vector ﬁelds X1 . Xr at p and is independent of their values away from p. r. . It deﬁnes a distance function and turns the manifold into a metric space in a natural way. . Let M be a diﬀerentiable manifold. Z ∈ C ∞ (T M ).CHAPTER 5 Riemannian Manifolds In this chapter we introduce the notion of a Riemannian manifold (M. . .e. Xr and Y1 . . Xr . . . Yr be smooth 47 . Let A : Cr (T M ) → Cs (T M ) be a tensor ﬁeld of type (r. . A smooth ∞ ∞ tensor ﬁeld A on M of type (r. A Riemannian metric on a diﬀerentiable manifold is an important example of what is called a tensor ﬁeld. Y. . . The metric g provides us with an inner product on each tangent space and can be used to measure the length of curves in the manifold. . . Xr ). Deﬁne ∞ C0 (T M ) = C ∞ (M ) and for each positive integer r ∈ Z+ let be the r-fold tensor product of C ∞ (T M ) over C ∞ (M ). .1. .

A Riemannian metric g on M is a tensor ﬁeld such that for each p ∈ M the restriction ∞ ∞ g : C2 (T M ) → C0 (T M ) gp = g|Tp M ⊗Tp M : Tp M ⊗ Tp M → R . . . We shall prove the statement for r = 1. . x) be local coordinates on M . . Put X = X1 and Y = Y1 and let (U. RIEMANNIAN MANIFOLDS vector ﬁelds on M such that (Xk )p = (Yk )p for each k = 1. . Yr )(p). . . . . . Choose a function f ∈ C ∞ (M ) such that f (p) = 1. k=1 m A(X)(p) = f (p)A(X)(p) = A(f · X)(p) = and similarily A(Y )(p) = k=1 m ρk (p)A(vk )(p) k=1 σk (p)A(vk )(p). σk ∈ C (M ) such that m m support(f ) = {p ∈ M | f (p) = 0} f ·X = This implies that ρk v k k=1 and f · Y = σk vk . (Xr )p ) → A(X1 . .48 5. As a direct consequence we see that A(X)(p) = A(Y )(p). Proof. . . . Xr )(p). Let M be a smooth manifold. . . Xr )(p) = A(Y1 . Deﬁnition 5. . . . the rest follows by induction. is contained in U and deﬁne the vector ﬁelds v1 . For a tensor A we shall by Ap denote the multi-linear restriction of A to the r-fold tensor product of the vector space Tp M over R given by Tp M ⊗ · · · ⊗ Tp M Ap : ((X1 )p . .3. vm ∈ C ∞ (T M ) on M by ∂ f (q) · ( ∂xk )q if q ∈ U (vk )q = 0 if q ∈ U / ∞ Then there exist functions ρk . . r. . . The fact that Xp = Yp shows that ρk (p) = σk (p) for all k. . Then A(X1 .

The standard inner product on the vector space Rm given by m X. . g) is called a Riemannian manifold. Y ) = (1 + |x|2 m )2 R The Riemannian manifold Σm = (Rm . Then the length L(γ) of γ can be determined as follows ∞ L(γ) = 2 0 m Example 5. 0. Then the length L(γ) of γ is deﬁned by L(γ) = I g(γ(t). Geometric properties of (M. ˙ ˙ By multiplying the Euclidean metric on subsets of Rm by a factor we obtain important examples of Riemannian manifolds. γ ˙ ˙ dt = 2 1 + |γ|2 ∞ 0 dt = 2[arctan(t)]∞ = π. Y )(p) is an inner product on the tangent space Tp M . RIEMANNIAN MANIFOLDS 49 with gp : (Xp . Y ) = (1 − |x|2 m )2 R . Example 5.4. Y Rm . Deﬁnition 5. . Rm ) is called the m-dimensional Euclidean space. g) is called the m-dimensional punctured round sphere. The pair (M. gx (X. . gx (X. γ(t))dt. Y Rm . 0). Let γ : I → M be a C 1 -curve in M . . Y Rm = Xt · Y = m Xk Yk k=1 deﬁnes a Riemannian metric on R . 0 2 1+t m By the m-dimensional hyperbolic ball we mean B1 (0) equipped with the Riemannian metric 4 X. For a positive integer m equip the real vector space Rm with the Riemannian metric g given by 4 X. Yp ) → g(X. The study of Riemannian manifolds is called Riemannian Geometry. . The Riemannian manifold E m = (Rm . γ. Let γ : R+ → Σm be the curve with γ : t → (t. g) which only depend on the metric g are called intrinsic or metric properties. Let B1 (0) be the open unit ball in Rm given by m B1 (0) = {x ∈ Rm | |x|Rm < 1}.5.5.6.

T ). 0. The topology on M induced by the metric d is identical to the one M carries as a topological manifold (M. q ∈ M let Cpq denote the set of C 1 -curves γ : [0. r ∈ M we have (i) d(p. . with n = 2m (iv) the m-dimensional real projective space RP m ⊂ Sym(Rm+1 ) ⊂ Rn . q) = 0 if and only if p = q.8. see Deﬁnition 2.50 5.e. q) ≤ d(p. for all p. d) is a metric space i. where n = (m + 2)(m + 1)/2. g) has the structure of a metric space (M. Then 1 L(γ) = 2 0 γ. (ii) the tangent bundle T S m ⊂ Rn .1. d) in a natural way. Riemannian Geometry. Proof. The Euclidean metric . Then the smooth tensor ﬁeld g : C2 (T M ) → ∞ C0 (M ) given by g(X. Let (M. Graduate Texts in Mathematics 171. 1] → M such that γ(0) = p and γ(1) = q and deﬁne the function d : M × M → R+ 0 by d(p. h). See for example: Peter Petersen. . (ii) d(p. is a Riemannian metric on M called the induced metric on M in (N. Then (M. Yp ). p). q. Springer (1998). Proposition 5. where n = 2m + 2. γ ˙ ˙ dt = 2 1 − |γ|2 1 0 1+t 1 dt = [log( )] = ∞ 2 1−t 1−t 0 As we shall now see a Riemannian manifold (M. (iii) d(p. . q) ≥ 0. . For two points p. (iv) d(p. RIEMANNIAN MANIFOLDS m Let γ : (0. Deﬁnition 5. 1) → B1 (0) be a curve given by γ : t → (t. (iii) the m-dimensional torus T m ⊂ Rn . r) + d(r. g) be a Riemannian manifold. on Rn induces Riemannian metrics on the following submanifolds.7. 0). h) be a Riemannian manifold and M be ∞ a submanifold of N . Y ) : p → hp (Xp . Let (N. . q) = inf{L(γ)| γ ∈ Cpq }. A Riemannian metric on a diﬀerentiable manifold induces a Riemannian metric on any of its submanifolds as follows. q) = d(q. q). (i) the m-dimensional sphere S m ⊂ Rm+1 .

SLm (R). Deﬁnition 5. g) → (N. g) and (N. Wβ is an open neighbourhood for a chart (Wβ . O(m). h) be Riemannian manifolds. carries a natural Euclidean metric g given by ¯ g(Z. Fact 5. Let the collection (Uα )α∈I be an open covering of M such that for each α ∈ I the pair (Uα . For this we need the following fact from topology. of complex m × m matrices. ψβ ). Then there exists (i) a locally ﬁnite open reﬁnement (Wβ )β∈J such that for all β ∈ J. GLm (C). SO(m). SLm (C). U(m) and SU(m). xβ ) be charts on M and (fβ )β∈J be a partition of unity such that support(fβ ) is contained in Wβ . Then there exists a Riemannian metric g on M . Every locally compact Hausdorﬀ space with countable basis is paracompact.9. and (ii) a partition of unity (fβ )β∈J such that support(fβ ) ⊂ Wβ . ˆ Proof. Corollary 5. For each point p ∈ M let (Up . Yp ) = hφ(p) (dφp (Xp ). T ) be a topological manifold. This ∞ ∞ means that the well-deﬁned tensor g : C2 (T M ) → C0 (T M ) given by g= β∈J gβ is a Riemannian metric on M . A) be a diﬀerentiable manifold.11. h) is said to be conformal if there exists a function λ : M → R such that eλ(p) gp (Xp . A map φ : (M. ˆ Theorem 5. W ) = Re(trace(Z t · W )). (Wβ . Our next important step is to prove that every diﬀerentiable manifold M can be equipped with a Riemannian metric g. el 0 Rm if p ∈ Wβ if p ∈ Wβ / Note that at each point only ﬁnitely many of gβ are non-zero. Let (M m . Let (M. dφp (Yp )). φp ) ∈ A be a chart such that p ∈ Up .10. . ∂ ∂xβ l )(p) = fβ (p) · ek . Then for β ∈ J ∞ ∞ deﬁne gβ : C2 (T M ) → C0 (T M ) by gβ ( ∂ ∂xβ k .12. Rm be the Euclidean metric on Rm . Let (M.10. The vector space Cm×m .5. RIEMANNIAN MANIFOLDS 51 This induces metrics on the submanifolds of Cm×m such as Rm×m and the classical Lie groups GLm (R). ψα ) is a chart on M . Let (Wβ )β∈J be a locally ﬁnite open reﬁnement. Then (Up )p∈M is an open covering as in Corollary 5. Let .

An isometric diﬀeomorphism is called an isometry. Example 5. yY y . For x ∈ O(m) the left translation Lx : O(m) → O(m) by x is given by Lx : y → xy. Yx ) = dLx−1 (Xx ). Y . Let G be a Lie group and .16. Deﬁnition 5. e be an inner product on the tangent space Te G at the neutral element e. (dLx )y (yY ) = trace((xyX)t xyY ) = trace(X t y t xt xyY ) = trace(yX)t (yY ). Y = trace(X t Y ). On the standard unit sphere S m we have an action O(m + 1) × S m → S m of the orthogonal group O(m + 1) given by where · is the standard matrix multiplication. The tangent space Ty O(m) of O(m) at y is and the diﬀerential (dLx )y : Ty O(m) → Txy O(m) of Lx is given by We then have (dLx )y : yX → xyX. As for the orthogonal group O(m) an inner product on the tangent space at the neutral element of any Lie group can be transported via the left translations to obtain a left invariant Riemannian metric on the group. Y ∈ C ∞ (T M ) and p ∈ M . Let G be a Lie group. xy (A. dLx−1 (Yx ) e . AY = X t At AY = X t Y = X.14. x) → A · x Ty O(m) = {yX| X t + X = 0} (dLx )y (yX). A conformal map with λ ≡ 0 is said to be isometric. A Riemannian metric g on G is said to be left invariant if for each x ∈ G the left translation Lx : G → G is an isometry. Then for each x ∈ G the bilinear map gx (. The function eλ is called the conformal factor of φ. = yX. Example 5.13. This shows that the left translation Lx : O(m) → O(m) is an isometry for each x ∈ O(m). Equip the orthogonal group O(m) as a submanifold of Rm×m with the induced metric given by X.52 5. RIEMANNIAN MANIFOLDS for all X. The following shows that the O(m + 1)-action on S m is isometric AX. ) : Tx G × Tx G → R with gx (Xx .15. Proposition 5.

It implies that every Riemannian manifold can be realized as a submanifold of a Euclidean space.21 we deﬁne a map φ : S m → Sym(Rm+1 ) by Let α. See Exercise 5.5. The smooth tensor ﬁeld ∞ ∞ g : C2 (T G) → C0 (G) given by is a left invariant Riemannian metric on G. q q. Then for γ ∈ {α. b This proves that the immersion φ is isometric. In Example 3. q }/2 = { a. a p + q.17. q p. p p − q). p a.5. We shall now equip the real projective space RP m with a Riemannian metric. b + a. Yx )) g(dφp (a). of Math. Y ) → (g(X. q. b = β(0). see . Y ) : x → gx (Xx . b p + q. β : R → S n be two curves such that α(0) = p = β(0) and put ˙ a = α(0). β} we have ˙ If B is an orthonormal basis for Rm+1 . Long before John Nash became famous in Hollywood he proved the next remarkable result in his paper The embedding problem for Riemannian manifolds. p a. Let S m be the unit sphere in Rm+1 and Sym(Rm+1 ) be the linear space of symmetric real (m+1)×(m+1) matrices equipped with the metric g given by 1 g(A. B) = trace(At B). ˙ ˙ ˙ φ : p → (ρp : q → 2 q. b p. Proof. g : (X. p b /2 = q∈B { p. 20-63. b }/2 = a. γ(0) p + 2 q. 8 As in Example 3. Ann. The original proof of Nash was later simpliﬁed. Example 5. b + a. p γ(0)). then = dφp (γ(0)) = (q → 2 q. This inherits the induced metric from R(m+1)×(m+1) and the map φ : S m → RP m is what is called an isometric double cover of RP m .21 we have seen that the image φ(S m ) can be identiﬁed with the real projective space RP m . RIEMANNIAN MANIFOLDS 53 is an inner product on the tangent space Tx G. 63 (1956). dφp (b)) = trace(dφp (a)t · dφp (b))/8 q∈B q.

Let ˆ p be a point of M and ψ : U → M be a local parametrization of M ˆ ˆ with q ∈ U and ψ(q) = p. . tm ) → Lx ( Exp(tk Xk )) k=1 where Lx : G → G is the left-translation given by Lx (y) = xy. Deﬁnition 5. dψ(em )} is a local frame for T M over Up . For a point p ∈ M we deﬁne the normal space Np M of M at p by Np M = {v ∈ Tp N | hp (v. Then (dψx )0 (ek ) = Xk (x) for all k. Let G be one of the classical Lie groups and e be the neutral element of G. dψ(el )).54 5. The diﬀerential dψq : Tq Rm → Tp M is bijective so there exist neighbourhoods Uq of q and Up of p such that ˆ the restriction ψ = ψ|Uq : Uq → Up is a diﬀeomorphism. Xm } be a basis for the Lie algebra g of G. Let (N. g) is an isometry so the intrinsic geometry of ˜ (Uq . . Then there exists an isometric C r -embedding of (M. . ˜ Example 5. . . . . We shall now study the normal bundle of a submanifold of a given Riemannian manifold. For 3 ≤ r ≤ ∞ let (M. This means that the diﬀerential (dψx )0 : T0 Rm → Tx G is an isomorphism so there exist open neighbourhoods U0 of 0 and Ux of x such that the restriction of ψ to U0 is bijective onto its image Ux and hence a local parametrization of G around x. On the perturbation problem associated to isometric embeddings of Riemannian manifolds. ˜ Then ψ : (Uq . r We shall now see that parametrizations can be very useful tools for studying the intrinsic geometry of a Riemannian manifold (M. g) be a Riemannian C -manifold. We then deﬁne the pull-back metric g = ψ ∗ g on Uq by ˜ g (ek . Let {X1 . . Annals of Global Analysis and Geometry 7 (1989). This is an important example of the notion of a vector bundle over a manifold. . g) into a Euclidean space Rn . g ) → (Up . em } for T Uq so {dψ(e1 ). . . g ) and that of (Up . .19.18. On Uq we have the canonical frame {e1 . g). h) be a Riemannian manifold and M be a submanifold of N . w) = 0 for all w ∈ Tp M }. . 69-77. el ) = g(dψ(ek ). . .20. RIEMANNIAN MANIFOLDS for example Matthias Gunther. . For x ∈ G deﬁne ψx : Rm → G by m ψx : (t1 . Deep Result 5. . g) are exactly the same.

. Y = trace(X t Y ) inducing a left invariant metric on O(m). xX)| x ∈ O(m). N M = {(p. Y X ⊤ = (X − X t )/2 and X ⊥ = (X + X t )/2. h) be a Riemannian manifold and M m be a smooth submanifold of N . We have already seen that the tangent space Te O(m) of O(m) at the neutral element e is and that the tangent bundle T O(m) of O(m) is given by The space Rm×m of real m × m matrices has a linear decomposition and every element X ∈ Rm×m can be decomposed X = X ⊤ + X ⊥ in its symmetric and skew-symmetric parts given by If X ∈ Te O(m) and Y ∈ Sym(Rm ) then X. Example 5. Let S m be the unit sphere in Rm+1 equipped with its standard Euclidean metric .22.7. M. λp) ∈ R2m+2 | p ∈ S m . Proof. See Exercise 5.21. = trace(X t Y ) Rm×m = Sym(Rm ) ⊕ Te O(m) T O(m) = {(x. We shall now determine the normal bundle N O(m) of the orthogonal group O(m) as a submanifold of Rm×m . Te O(m) = {X ∈ Rm×m | X t + X = 0} N S m = {(p. p = 0} Np S m = {λp ∈ Rm+1 | λ ∈ R}. Let (N n . π) is a smooth (n − m)-dimensional vector bundle over M . The orthogonal group O(m) is a subset of the linear space Rm×m equipped with the Riemannian metric X. . X ∈ Te O(m)}. v)| p ∈ M. If p ∈ S m then the tangent space Tp S m of S m at p is so the normal space Np S m of S m at p satisﬁes Tp S m = {v ∈ Rm+1 | v. This shows that the normal bundle N S m of S m in Rm+1 is given by Theorem 5. λ ∈ R}. Then the normal bundle (N M. RIEMANNIAN MANIFOLDS 55 For all p ∈ M we have the orthogonal decomposition The normal bundle of M in N is deﬁned by Example 5.23. v ∈ Np M }. Tp N = Tp M ⊕ Np M.5.

Expo. RIEMANNIAN MANIFOLDS = = = = This means that the normal bundle N O(m) of O(m) in Rm×m is given by N O(m) = {(x. Y . The best known such examples are the Sasaki and Cheeger-Gromoll metrics. xY )| x ∈ O(m). Y ∈ Sym(Rm )}. On the geometry of tangent bundles.56 5. 1-41. Kappos. Gudmundsson. . 20 (2002). Math. For a detailed survey on the geometry of tangent bundles equipped with these metrics we recommend the paper S. A given Riemannian metric g on M can be used to construct a family of natural metrics on the tangent bundle T M of M . trace(Y t X) trace(XY t ) trace(−X t Y ) − X. E.

Exercise 5. Exercise 5. . .2. Y ) = Im(z)2 and prove that the holomorphic function 2 π : B1 (0) → {z ∈ C| Im(z) > 0} 4 X. . RIEMANNIAN MANIFOLDS 57 Exercises Exercise 5. xm ) → (eix1 . Let m be a positive integer and φ : Rm → Cm be the standard parametrization of the m-dimensional torus T m in Cm given by φ : (x1 .1. 0)}. eixm ). xm ).5. Let B1 (0) be the open unit disk in the complex plane equipped with the hyperbolic metric g(X.3. . Prove that φ is an isometric parametrization. xm ) → Prove that πm is an isometry. . . given by π:z→ is an isometry.5. 0. . . . (1 + |x|2 )2 Rm ) 1 (x1 .4. . . . Equip the unitary group U(m) as a submanifold of C with the induced metric given by m×m i+z 1 + iz ¯ Z. be the stereographic projection given by πm : (x0 . . . Show that for each z ∈ U(m) the left translation Lz : U(m) → U(m) given by Lz : w → zw is an isometry. Y ) = Equip the upper half plane {z ∈ C| Im(z) > 0} with the Riemannian metric 1 X. Let m be a positive integer and πm : (S m − {(1. Y R2 g(X. W = Re(trace(Z t W )). . Find a proof for Proposition 5. . . 4 . . . Rm+1 ) → (Rm . . 1 − x0 2 Exercise 5. Exercise 5. .16. Y (1 − |z|2 )2 R2 . .

Exercise 5. h are left-invariant? (ii) Determine the normal space Ne SLm (R) of SLm (R) in GLm (R) with respect to g (iii) Determine the normal bundle N SLm (R) of SLm (R) in GLm (R) with respect to h.22.6. h be the induced metrics on the special linear group ˆ ˆ SLm (R) as a subset of GLm (R). ˆ ˆ (i) Which of the metrics g. xW ) = trace((xZ)t xW ). h. Further let g . g . Let m be a positive integer and GLm (R) be the corresponding real general linear group. RIEMANNIAN MANIFOLDS Exercise 5.58 5. . Let g. h be two Riemannian metrics on GLm (R) deﬁned by gx (xZ.7. xW ) = trace(Z t W ). Find a proof for Theorem 5. hx (xZ.

1. This is the most important example of the general notion of a connection in a smooth vector bundle. g ∈ C ∞ (Rm ) and X. . (ii) ∂X(f · Y ) = X(f ) · Y + f · ∂XY . ∂XZ . Proposition 6. Z + Y. If λ. and X. Y on Rm to the directional derivative ∂XY of Y in the direction of X given by (∂XY )(x) = lim t→0 Y (x + tX(x)) − Y (x) . Z ) = ∂XY . (v) X( Y. 59 The most fundamental properties of the operator ∂ are expressed by the following. On the m-dimensional real vector space Rm we have the well-known diﬀerential operator ∂ : C ∞ (T Rm ) × C ∞ (T Rm ) → C ∞ (T Rm ) mapping a pair of vector ﬁelds X. t The next result shows that the diﬀerential operator ∂ is compatible with both the standard diﬀerentiable structure on Rm and its Euclidean metric. We also give an example of a connection in the normal bundle of a submanifold of a Riemannian manifold and study its properties. Y. (iii) ∂(f · X + g · Y )Z = f · ∂XZ + g · ∂Y Z. Y. g).CHAPTER 6 The Levi-Civita Connection In this chapter we introduce the Levi-Civita connection ∇ of a Riemannian manifold (M. f. Y ]. µ ∈ R. Let the real vector space Rm be equipped with the standard Euclidean metric . Z ∈ C ∞ (T Rm ) then (i) ∂X(λ · Y + µ · Z) = λ · ∂XY + µ · ∂XZ. Then (iv) ∂XY − ∂Y X = [X. Z ∈ C ∞ (T Rm ) be smooth vector ﬁelds on Rm . We deduce an explicit formula for the Levi-Civita connection for Lie groups equipped with left-invariant metrics.

ˆ ˆ [X. Z) + g(Y. Z) = X(g(Y. A section v ∈ C ∞ (E) of the vector bundle E is said to be parallel ˆ with respect to the connection ∇ if ˆ ∇ v=0 X ∞ for all vector ﬁelds X ∈ C (T M ). Y ]. Then it is easily seen that the following equations hold g(∇ Y . π) is said to be metric or compatible with the Riemannian metric g if ˆ ˆ X(g(Y.60 6. ˆ (ii) ∇ X X ˆ ˆ ˆ (iii) ∇ · X + g · Y )v = f · ∇ v + g · ∇ v. First we introduce the concept of a connection in a smooth vector bundle. Let (E. Let (M. µ ∈ R.e. M. g ∈ C ∞ (M ). Y. π). Deﬁnition 6. Z) X Y . Y ] = ∇ Y − ∇ X X Y ∞ for all X. M. M. Z) + g(∇ X. π). Let (M. (f X Y for all λ. M. ∇ Z) X X ∞ for all X. π) is a map ∇ : C ∞ (T M )×C ∞ (E) → C ∞ (E) such that ˆ ˆ ˆ (i) ∇ (λ · v + µ · w) = λ · ∇ v + µ · ∇ w. Deﬁnition 6.3. ] is the Lie bracket on C (T M ). Then we deﬁne the torsion ∞ ∞ ˆ T : C2 (T M ) → C1 (T M ) of ∇ by ˆ ˆ T (X. ˆ Deﬁnition 6. ∇ Z). Z)) − g(Y.4. g) be a Riemannian manifold and ∇ be a metric and torsion-free connection on its tangent bundle (T M. X X g(∇ Y . Then a ˆ connection ∇ on the tangent bundle (T M. Z ∈ C (T M ). THE LEVI-CIVITA CONNECTION We shall now generalize the diﬀerential operator ∂ on the Euclidean space Rm to the so called Levi-Civita connection ∇ on a Riemannian manifold (M. ˆ A connection on (E. v. Y ]. Y ∈ C (T M ). M. Y ) = ∇ Y − ∇ X − [X. g). X Y ∞ ˆ where [. Let M be a smooth manifold and ∇ be a connection on the tangent bundle (T M. π) be a smooth vector bundle over M . X X X ˆ (f · v) = X(f ) · v + f · ∇ v.2. g) be a Riemannian manifold. Z) = g([X. w ∈ C ∞ (E) and f. Y ∈ C ∞ (T M ). The connection ∇ is said to be torsion-free if its torsion T vanishes i. Z)) = g(∇ Y . X.

Y ) + g([Z. Y1 . [Y. . X) + g(Z.15 and the fact that g is a tensor ﬁeld that and g(∇ (λ · Y1 + µ · Y2 ). [X. X k=1 As a direct consequence there exists at most one metric and torsion free connection on the tangent bundle. Z)) + f · Y (g(X.5. Y ) + g(X. Deﬁnition 6. Y 0 = −Z(g(X. ∇ Z − [Y. Z) X X X g(∇ + Y X. Let (M. Z)) + Y (g(Z. Furthermore we have for all f ∈ C ∞ (M ) = {X(f · g(Y. Y )) + g(∇ Z + [Z. Z) + g(∇ X.5. Proof. Y ) + g(X. X]) − g(X. f · Y ]. is called the Levi-Civita connection on M . . Z) = λ · g(∇ Y1 . Let (M. THE LEVI-CIVITA CONNECTION 61 = g([X.6. Y2 . Y )) X If {E1 . X) + g(Z. Z])}. [X. Z]). X]. Z)) − g(X. Z)) − Z(f · g(X. [Z. Z) + Y (g(X. Ei )Ei . It follows from Deﬁnition 3. Y )) +f · g([Z. X)) − Z(g(X. ∇ Y ) Z Z By adding these relations we yield = −Z(g(X. X Y 2 · g(∇ Y . X]. Z)) + Y (g(X. Z) = {X(g(Y. Z) + µ · g(∇ Y2 .6. Z) X . Then the Levi-Civita connection ∇ is a connection on the tangent bundle T M of M. f · Y ])} 2 · g(∇ f Y . Z) Y1 Y1 Y2 2 ∞ for all λ. Y ]. g) only depending on the diﬀerentiable structure of the manifold and its Riemannian metric. It should be noted that the Levi-Civita connection is an intrinsic object on (M. [X. µ ∈ R and X. . g) be a Riemannian manifold. Y ])}. ∇ Z). Em } is a local orthonormal frame for the tangent bundle then m +g(Z. Y )) X +g([Z. X]. Y ]. Theorem 4. Proposition 6. Y ]) + g(Y. g) be a Riemannian manifold then the map ∇ : C ∞ (T M ) × C ∞ (T M ) → C ∞ (T M ) given by 2 · g(∇ Y . Z) = {X(g(Y. . Y ) + g([Z. Z) = g(∇ X. Z ∈ C (T M ). Y )) + g(∇ X. Z)) − Z(g(X. ∇ Y = X g(∇ Y .

X])} = g(Z. [X. Y ))} = X(g(Y. Z) − g(∇ X. Z)) on the right hand side of Deﬁnition 6. This is exactly 1 = {X(g(Y. X This proves that ∇ is a connection on the tangent bundle (T M.r. Z)) − Z(f · g(X. The next result is called the Fundamental theorem of Riemannian geometry. Z)} X = 2 · g(X(f ) · Y + f · ∇ Y .5.5. Z)) + f · Y (g(X. Z)). Z)) + X(g(Z. M. Y )) +g([Z. THE LEVI-CIVITA CONNECTION = {X(f ) · g(Y.7. π). [f · X. Z) + g(∇ Z. [X. Z). f · X]. X) + g(Z. Y ) +f · g([Z. The sum g(∇ Y . Y ])} = {f · X(g(Y. Theorem 6. Z) + f · Y (g(X. Y ) +g(Z(f ) · Y + f · [Z. 2 This shows that the Levi-Civita connection is compatible with the Riemannian metric g on M . Proof. Y ) − f · Z(g(X. Z)) + Y (f ) · g(X. Z) f = {f · X(g(Y. [X.62 6. Y ]). Y ) equals twice the symmetric part X X (w. Z) + f · g(∇ Y . Y )) of the right hand side of the equation in Deﬁnition 6. Z) X and 2 · g(∇ · XY . This is the same as 1 = {g(Z. X]. Y )) + f · g([Z. The diﬀerence g(∇ Y . Y ]) − g(Z. Let (M. Then the Levi-Civita connection is a unique metric and torsion free connection on the tangent bundle (T M. Y ) + f · g([Z. Y ) + f · g([Z. M. Y ])} = 2 · {X(f ) · g(Y. X) + f · g(Z. Y ]) − g(Z. X) + g(Z.t the pair (Y. . g) be a Riemannian manifold. 2 This proves that the Levi-Civita connection is torsion-free. Z)) −Z(f ) · g(X. Z) equals twice the X Y skew-symmetric part (w.t the pair (X. Z)) + Y (f · g(X. X(f ) · Y + f · [X. [Y. Y )) +g(Z(f ) · X. Y ) − f · Z(g(X. Y ]. Z)) −Z(f ) · g(X. Z) + f · X(g(Y. X]. Y (f ) · X)} = 2 · f · g(∇ Y .r. Y ]. Y ]. π).

See Exercise 6. Y ]. x) be local coordinates on M and put Xi = ∂/∂xi ∈ C ∞ (T U ). Proposition 6. Let G be one of the classical compact Lie groups O(m). Xj ]) = [dx(Xi ). Let (M. Then for each X ∈ g the operator ad(X) : g → g is skew symmetric. Xm } is a local frame of T M on U .3. Y ) + g(X. Y ].2.17 implies that dx([Xi .11. Then the Levi-Civita connection ∇ satisﬁes 1 g(∇ Y . ad(Z)(Y ))} X 2 for all X. Proof. if for all Z ∈ g the map ad(Z) is skew symmetric with respect to g then 1 ∇ Y = [X. Further let (U. In particular. ˜ The diﬀerential dx is bijective so Proposition 4. Y. x) by m k=1 m Γk Xk = ∇ Xj . Z) = {g([X. THE LEVI-CIVITA CONNECTION 63 A vector ﬁeld X ∈ C ∞ (T M ) on (M. For (U.6. g) be a Lie group equipped with a left invariant metric. . . g) be a Riemannian manifold with LeviCivita connection ∇. ij Xi On the subset x(U ) of R we deﬁne the metric g by ˜ g (ei . X X Deﬁnition 6. SO(m). g) induces the ﬁrst order covariant derivative ∇ : C ∞ (T M ) → C ∞ (T M ) X in the direction of X by ∇ :Y →∇ Y. X 2 Proof. Let (G. Proposition 6.10.8. X]. U(m) or SU(m) equipped with the left-invariant metric ¯ g(Z. Z ∈ g. dx(Xj )] = [ei . Then {X1 . For a left invariant vector ﬁeld Z ∈ g we deﬁne the map ad(Z) : g → g by ad(Z) : X → [Z. . W ) = Re (trace (Z t W )). Xj ). x) we deﬁne the Christoﬀel symbols Γk : U → R ij of the connection ∇ with respect to (U. . ej ) = gij = g(Xi . ej ] = 0 .9. Example 6. Z) + g(ad(Z)(X). See Exercise 6. Let G be a Lie group.

Xj } 2 1 ∂gjl ∂gli ∂gij = { + − }. Xl + Xj Xl . Let N be a smooth manifold. Xj ] = 0. If U = N then X is called a global extension. Note ˜ that Z is not necessarily an element of C ∞ (T M ) i. If p ∈ M then (∇ Y )p only depends on the X ˜ value Xp = Xp and the value of Y along some curve γ : (−ǫ. ǫ). Fact 6. Xi − Xl Xi . Xl ij ∇ Xj . Y ∈ C ∞ (T N ) be their extensions to N . h) be a Riemannian manifold and M be a sub˜ ˜ manifold of N . This means ˜ ˜ that (∇ Y )p only depends on Xp and the value of Y along γ. A local extension of X to U is ∞ ˜ a vector ﬁeld X ∈ C (T U ) such that Xp = Xp for all p ∈ U ∩ M . THE LEVI-CIVITA CONNECTION and hence [Xi . a vector ﬁeld on ˜ the submanifold M . For each p ∈ M the tangent vector Zp ∈ Tp N can be decomposed ˜ ˜⊤ ˜⊥ Zp = Zp + Zp ˜ in a unique way into its tangential part (Zp )⊤ ∈ Tp M and its normal ˜ ˜ ˜ ˜ part (Zp )⊥ ∈ Np M . ǫ) → N ˜ such that γ(0) = p and γ(0) = Xp = Xp . 2 ∂xi ∂xj ∂xl = If g kl = (g −1 )kl then Γk ij 1 = 2 m l=1 g kl { ∂gjl ∂gli ∂gij + − }. ∂xi ∂xj ∂xl Deﬁnition 6. Xl Xi 1 = {Xi Xj . Since ˙ Xp ∈ Tp M we may choose the curve γ such that the image γ((−ǫ.3. Let (N.64 6. Then vector ﬁelds X ∈ C ∞ (T M ) and Y ∈ C ∞ (N M ) ∞ have global extensions X. Then Yγ(t) = Yγ(t) for t ∈ (−ǫ.13. For this we write Z = Z ⊤ + Z ⊥ . For this see Remark 7. ǫ)) ˜ is contained in M . Y ∈ C ∞ (T M ) be vector ﬁelds on M and X.e. Y ∈ C (T N ). From the deﬁnition of the Levi-Civita connection we now yield m m Γk gkl = ij k=1 k=1 Γk Xk . hence X . Let (N.12. Let Z ∈ C ∞ (T N ) be a vector ˜ ﬁeld on N and Z = Z|M : M → T N be the restriction of Z to M . Let U be an open ˜ subset of N such that U ∩ M = ∅. h) be a Riemannian manifold and M be a submanifold equipped with the induced metric g. ˜ ˜ Let X. M be a submanifold ˜ of N and X ∈ C ∞ (T M ) be a vector ﬁeld on M .

X X X ∞ ˜ Y. h) induces a metric connection ¯ ∇ on the normal bundle N M of M in N as follows. Then the ¯ : C ∞ (T M ) × C ∞ (N M ) → C ∞ (N M ) given by map ∇ ¯ ˜ ∇˜Y = (∇ Y )⊥ X X is a well-deﬁned connection on the normal bundle N M satisfying ˜ ˜ ˜ ¯ ˜ ˜ ˜ ¯ ˜ X(h(Y . Y ∈ C ∞ (T N ) be ˜ ˜ vector ﬁelds extending X ∈ C ∞ (T M ) and Y ∈ C ∞ (N M ). THE LEVI-CIVITA CONNECTION 65 ˜ ˜ independent of the way X and Y are extended. Let (N. The operator B is called the second fundamental form of M in (N. Z ∈ C ∞ (N M ). h). see Exercise 6. h) be a Riemannian manifold and M be a submanifold with the induced metric g. Then the operator ∇ is the Levi-Civita connection of the submanifold (M. g). . The Levi-Civita connection on (N. ∇˜Z) X X ˜ ∈ C ∞ (T M ) and Y .16. Let X.6. Then we deﬁne two operators ˜ ∇ : C ∞ (T M ) × C ∞ (T M ) → C ∞ (T M ) and by B : C ∞ (T M ) × C ∞ (T M ) → C ∞ (N M ) ˜ ˜ ˜ ˜ ∇˜Y = (∇ Y )⊤ and B(X. h) be a Riemannian manifold and M be a ˜ submanifold of N with the induced metric g.14. Z) + h(Y . It is symmetric and hence tensorial in both its arguments. ˜ ˜ for all X Proof. This shows that the ˜ following maps ∇ and B are well deﬁned.9. Proof. Y ∈ C (T N ) are any extensions of X. See Exercise 6. Let (N. h) be a Riemannian manifold and M be a submanifold equipped with the induced metric g.15. Deﬁnition 6. Theorem 6. Z)) = h(∇˜Y . ˜ where X.8. Proposition 6.7. Let (N. See Exercise 6. Y ) = (∇ Y )⊥ .

C 0 1 Ae = 0 Prove that {A. Ye = |p=0 . Y ] = Y. Exercise 6. Let G given by z e G = { 0 0 Xe = Show that [X. 0 1 0 1 0 0 0 0 ∇ B. ∇ Z.10. ∂x ∂y ∂z Let A. 1).4. ∇ C. 2 Prove that . ∇ X. Let M be a smooth manifold and ∇ be a connection on the tangent bundle (T M. Find a proof for Proposition 6. Y ∈ so(m) we have 1 ∇ Y = [X. Let g be a left-invariant Riemannian metric on G such that {X. Y = trace(X t Y ). X Y X Z Y Z Exercise 6. Y ] = 0. Be = 0 0 0 . Ce = 0 0 −1 . X] = X.1. X 2 be elements of the Lie algebra so(3) with 0 0 0 0 0 −1 −1 0 0 0 . Calculate the vector ﬁelds ∇ Y . [Z.2. 0 1 Let X. M.9. Z ∈ g be left-invariant vector ﬁelds on G such that ∂ ∂ ∂ |p=0 . ∇ X. ∇ A. A B C be the 3-dimensional subgroup of GL3 (R) 0 x ez y | p = (x. is left-invariant and that for any left-invariant vector ﬁelds X. Let SO(m) be the special orthogonal group equipped with the metric 1 X.3. Prove that the torsion ∞ ∞ ˆ T : C2 (T M ) → C1 (T M ) of ∇ is a tensor ﬁeld of type (2. B. z) ∈ R3 }. [Z. Exercise 6. THE LEVI-CIVITA CONNECTION Exercises ˆ Exercise 6. Y. B. Z} is an orthonormal basis for the Lie algebra g. Y ]. Ze = |p=0 . Y. π). y. Exercise 6. ∇ Z.5. Find a proof for Proposition 6. C} is an orthonormal basis for so(3) and calculate . ∇ Y .66 6.

∇ C.8. B. Exercise 6.16. Prove that the second fundamental form B of M in N is symmetric and tensorial in both its arguments. 2 Let A. Find a proof for Theorem 6. C be elements of the Lie algebra sl2 (R) with Ae = 0 −1 1 0 . Y p = trace((p−1 X)t (p−1 Y )). Let SL2 (R) be the real special linear group equipped with the metric 1 X. C} is an orthonormal basis for sl2 (R) and calculate . Prove that {A. g) be a submanifold with the induced metric. Find a proof for Proposition 6. A B C Exercise 6. ∇ B. THE LEVI-CIVITA CONNECTION 67 Exercise 6.9. Let (N. Be = 0 1 1 0 .6. Exercise 6.15. B. h) be a Riemannian manifold with LeviCivita connection ∇ and (M.7. ∇ A.6. Ce = 1 0 0 −1 .

.

˙ The next result gives a rule for diﬀerentiating a vector ﬁeld along a given curve and shows how this is related to the Levi-Civita connection. Proposition 7. ·) into a module over C ∞ (I) and a real vector space over the constant functions in particular. Y ∈ Cγ (T M ) and f ∈ C ∞ (I) we deﬁne the operations + and · by (i) (X + Y )(t) = X(t) + Y (t). g). M. Deﬁnition 7. Let M be a smooth manifold and (T M. g) be a smooth Riemannian manifold and γ : I → M be a curve in M . Then there exists a unique operator D ∞ ∞ : Cγ (T M ) → Cγ (T M ) dt such that for all λ. π) be its tangent bundle. µ ∈ R and f ∈ C ∞ (I). This is a solution to a second order non-linear system of ordinary diﬀerential equations.2. A vector ﬁeld X along a curve γ : I → M is ∞ a curve X : I → T M such that π ◦ X = γ. and (iii) for each t0 ∈ I there exists an open subinterval J0 of I such that t0 ∈ J0 and if X ∈ C ∞ (T M ) is a vector ﬁeld with Xγ(t) = Y (t) for all t ∈ J0 then DY (t0 ) = (∇˙ X)γ(t0 ) .CHAPTER 7 Geodesics In this chapter we introduce the notion of a geodesic on a Riemannian manifold (M. Let (M. For a given smooth curve γ : I → M in M the smooth vector ﬁeld X : I → T M with X : t → (γ(t). (ii) (f · X)(t) = f (t) · X(t). For X. +. (ii) D(f · Y )/dt = df /dt · Y + f · (DY /dt). γ dt 69 . By Cγ (T M ) we denote ∞ the set of all smooth vector ﬁelds along γ. (i) D(λ · X + µ · Y )/dt = λ · (DX/dt) + µ · (DY /dt). γ(t)) is called the tangent ﬁeld along γ. ∞ This turns (Cγ (T M ). We show that geodesics are solutions to two diﬀerent variational problems.1. They are critical points to the so called energy functional and furthermore locally shortest paths between their endpoints.

The second condition means that (1) DY (t) = dt DXk αk (t) dt k=1 m m m γ(t) + k=1 αk (t) Xk ˙ γ(t) . It is easily seen that if we use equation (3) for deﬁning an operator D/dt then it satisﬁes the necessary conditions of Proposition 7.2. ∇ · ZX = f · ∇ X f Z and the equation DY (t0 ) = (∇˙ X)γ(t0 ) γ dt in Proposition 7.3. For a point t0 ∈ I choose a chart (U. This proves the existence of the operator D/dt. This shows that the operator D/dt is uniquely determined. Let us ﬁrst prove the uniqueness.e. . GEODESICS Proof. γ(J0 ) ⊂ U and put Xi = ∂/∂xi ∈ C ∞ (T U ). . γm (t)) then γ(t) = ˙ k=1 γk (t) Xk ˙ γ(t) and the third condition for D/dt imply that (2) DXj dt m γ(t) = (∇˙ Xj )γ(t) = γ k=1 γk (t)(∇ Xj )γ(t) . ˙ X k Together equations (1) and (2) give (3) DY (t) = dt m m k=1 {αk (t) + ˙ Γk (γ(t))γi (t)αj (t)} Xk ˙ ij i. ˙ . Then a vector ﬁeld Y along γ can be written in the form m Y (t) = k=1 αk (t) Xk γ(t) for some functions αk ∈ C ∞ (J0 ). This allows us to use the notation ˙ ∇γ Y for DY /dt. Let x ◦ γ(t) = (γ1 (t).2 that the value (∇ X)p of ∇ X at p only depends Z Z on the value of Zp of Z at p and the values of Y along some curve γ satisfying γ(0) = p and γ(0) = Zp . .70 7. so for the moment we assume that such an operator exists. x) on M and open interval J0 such that t0 ∈ J0 . . It follows from the fact that the Levi-Civita connection is tensorial in its ﬁrst argument i. Remark 7.j=1 γ(t) .

k=1 This implies that the vector ﬁeld Y is parallel i. Then there exists a unique parallel vector ﬁeld Y along γ such that X0 = Y (t0 ). γ A C 2 -curve γ : I → M is said to be a geodesic if its tangent ﬁeld γ is ˙ parallel along γ i. Then m ∇˙ Y (t) = γ = j=1 m {σj (t) Xj ˙ {σk (t) + ˙ γ(t) m + σj (t) ∇˙ Xj γ γ(t) } γ(t) k σj (t)ρi (t)Γij (γ(t))} Xk i.e. Let (M. ∇˙ Y ≡ 0 if and γ only if the following ﬁrst order linear system of ordinary diﬀerential .e. Without loss of generality we may assume that the image of γ lies in a chart (U. Further let γ : I → M be a smooth curve. GEODESICS 71 The Levi-Civita connection can now be used to deﬁne a parallel vector ﬁeld and a geodesic on a manifold as solutions to ordinary differential equations Deﬁnition 7. t0 ∈ I and X0 ∈ Tγ(t0 ) M . A vector ﬁeld X along γ is said to be parallel if ∇˙ X = 0.j=1 . Proof. Theorem 7. g) be a Riemannian manifold and I = (a.7.4. b) be an open interval on the real line R. γ˙ The next result shows that for given initial values at a point p ∈ M we get a parallel vector ﬁeld globally deﬁned along any curve through that point.5. g) be a Riemannian manifold and γ : I → M be a C 1 -curve. Similarly let Y be a vector ﬁeld along γ represented by m Y (t) = j=1 σj (t) Xj γ(t) . Let (M. ∇˙ γ = 0. x). We put Xi = ∂/∂xi so on the interval I the tangent ﬁeld γ is represented in our local coordinates by ˙ m γ(t) = ˙ i=1 ρi (t) Xi γ(t) with some functions ρi ∈ C ∞ (I).

If (t0 . . It follows from Fact 7. In particular. . Let (M. m. Let f : U → Rn be a continuous map deﬁned on an open subset U of R × Rn and L ∈ R+ such that for all (t. γ : I → M be a smooth curve and X. 2. Then the function g(X. . Proposition 7.6. Using the fact that the Levi-Civita connection is metric we obtain d (g(X. . Let (M.j=1 σj (t)ρi (t)Γk (γ(t)) = 0 ij for all k = 1. Y be parallel vector ﬁelds along γ.72 7. .8. γ) is constant along γ. y) ∈ U . |f (t. if γ is a geodesic then g(γ. . σm ) to the above system. p ∈ M and {v1 . x(t0 ) = x0 . . GEODESICS equations is satisﬁed: m σk (t) + ˙ i. . Lemma 7. . . vm } be an orthonormal basis for the tangent space Tp M . ∇˙ Y ) = 0. . Y ) is constant along γ. ˙ ˙ Proof. Y )) = g(∇˙ X. x) − f (t. x). . Xm be parallel vector ﬁelds along γ such that Xk (0) = vk for k = 1. x(t)). m. Y ) : I → R given by t → gγ(t) (Xγ(t) . The following result on parallel vector ﬁelds is a useful tool in Riemannian geometry. . g) be a Riemannian manifold. The following result is the well-known theorem of Picard-Lindel¨f. . . g) be a Riemannian manifold. Y ) + g(X. (t. Yγ(t) ) is constant. y)| ≤ L · |x − y| . . . γ γ dt This proves that the function g(X. x0 ) ∈ U then there exists a unique local solution x : I → Rn to the following initial value problem x′ (t) = f (t. . . It will be employed in Chapter 9. .7. o Fact 7. . . . This gives us the unique parallel vector ﬁeld Y m Y (t) = k=1 σk (t) Xk γ(t) along I.6 that to each initial value σ(t0 ) = (v1 . vm ) ∈ Rm with m Y0 = k=1 vk Xk γ(t0 ) there exists a unique solution σ = (σ1 . . Let γ : I → M be a smooth curve such that γ(0) = p and X1 .

x) be a chart on M such that p ∈ U and put Xi = ∂/∂xi .10 that for initial values q0 = x(p) and w0 = (dx)p (v) there exists an open interval (−ǫ. γm ) satisfying the initial conditions (γ1 (0). . If p ∈ M and v ∈ Tp M then there exists an open interval I = (−ǫ.9. Let (U. γm (0)) = q0 and (γ1 (0). This is a direct consequence of Lemma 7. . . For the important geodesic equation we have the following local existence result. . Proof. . . . ˙ Proof. g) be a Riemannian manifold. GEODESICS 73 Then the set {X1 (t). o .7. . ǫ) and a unique geodesic γ : I → M such that γ(0) = p and γ(0) = v.j=1 γj (t)γi (t)Γk ◦ γ(t)} Xk ˙ ˙ ij γ(t) . . . . For an open interval J and a C 2 -curve γ : J → U we put γi = xi ◦ γ : J → R.7. By diﬀerentiation we then obtain m ∇˙ γ = γ˙ = j=1 m ∇˙ γj (t) Xj γ ˙ {¨j (t) Xj γ {¨k (t) + γ m γ(t) m + γ(t) i=1 m j=1 m γj (t)γi (t) ∇ Xj ˙ ˙ Xi γ(t) } = k=1 i. m. Theorem 7. . It follows from Fact 7.j=1 γj (t)γi (t)Γk (γ(t)) = 0 ˙ ˙ ij for all k = 1. ˙ ˙ The following result is a second order consequence of the well-known theorem of Picard-Lindel¨f. . . . . The curve x ◦ γ : J → Rm is C 2 so we have m (dx)γ(t) (γ(t)) = ˙ i=1 γi (t)ei ˙ giving γ(t) = ˙ m γi (t) Xi ˙ i=1 γ(t) . . ǫ) and a unique solution (γ1 . Xm (t)} is a orthonormal basis for the tangent space Tγ(t) M for all t ∈ I. . . Let (M. . γm (0)) = w0 . Hence the curve γ is a geodesic if and only if γk (t) + ¨ i.

Following Theorem 6.11. . Let (N. x) − f (t. Let E m = (Rm .13. v) ∈ T M there exists a geodesic γ : R → M deﬁned on the whole of R such that γ(0) = p and γ(0) = v. Rm ) be the Euclidean space. y)| ≤ L · |x − y| . x(t0 ) = x0 . GEODESICS Fact 7. so Γk = 0 for all ij i. h) be a Riemannian manifold with LeviCivita connection ∇ and M be a submanifold equipped with the induced metric g. Let f : U → Rn be a continuous map deﬁned on an open subset U of R × Rn and L ∈ R+ such that for all (t. At a point p ∈ S m the normal space Np S m of S m in E m+1 is simply the line generated by p. (t. g) satisﬁes ˜ ˙ ∇˙ γ = (∇˙ γ)⊤ . . .12. g) is an inner object completely determined by the metric g. m. γ˙ ˜ Proof. γ γ. j. k = 1. A geodesic γ : I → (M. Hence the same applies for the condition ∇˙ γ = 0 γ˙ for a given curve γ : I → M . b ∈ Rm . The manifold (M. g) in a Riemannian manifold is said to be maximal if it cannot be extended to a geodesic deﬁned on an interval J strictly containing I. The Levi-Civita connection ∇ on a given Riemannian manifold (M. γ γ˙ Example 7. then ˜ ˙ ¨ ∇˙ γ = γ ⊤ = γ − γ ⊥ = γ − γ . For the trivial chart idRm : Rm → Rm the metric is given by gij = δij . x′ (t0 ) = x1 . This ¨ proves that the geodesics are the straight lines. A curve γ : I → M is a geodesic in M if and only if (∇˙ γ)⊤ = 0. x). ¨ ¨ ¨ ¨ γ |f (t. ˙ Proposition 7.74 7. This means that the image of a geodesic under a local isometry is again a geodesic. If (t0 . . g) is said to be complete if for each point (p. Deﬁnition 7.10. y) ∈ U . x0 ) ∈ U and x1 ∈ Rn then there exists a unique local solution x : I → Rn to the following initial value problem x′′ (t) = f (t. Let E m+1 be the (m + 1)-dimensional Euclidean space and S m be the unit sphere in E m+1 with the induced metric. This means that γ : I → Rm is a geodesic if and only if γ (t) = 0 or equivalently γ(t) = t · a + b for some a.15 the Levi-Civita connection ∇ on (M. x(t)). If γ : I → S m is a curve on the sphere. .

GEODESICS 75 This shows that γ is a geodesic on the sphere S m if and only if (4) γ = γ . This shows that the real projective spaces are complete. Let (M. Let (M. ǫ) × I → M g(γ(t). g) be a Riemannian manifold and γ : I → M be a C 2 -curve on M . s) = γ(s). . Let Sym(Rm+1 ) be equipped with the metric 1 trace(At B). Φ0 (s) = Φ(0.15. b]. For every compact interval [a.16.b] (Φt ))|t=0 = 0.b] (γ) = 1 2 b φ : p → (2ppt − e) Φ : (−ǫ. g) be a Riemannian manifold and γ : I → M be a C r -curve on M . Example 7.b] by E[a. Then one easily checks that γ(0) = p. This means that the geodesics on RP m are exactly the images of geodesics on S m . A variation of γ is a C r -map such that for all s ∈ I.14.7. Deﬁnition 7. ¨ ¨ For a point (p. This shows that the non-constant geodesics on S m are precisely the great circles and the sphere is complete. Φt (a) = γ(a) and Φt (b) = γ(b). B = is an isometric immersion and that the image φ(S m ) is isometric to the m-dimensional real projective space RP m . ˙ ˙ a A C 2 -curve γ : I → M is called a critical point for the energy functional if every proper variation Φ of γ|[a.b] satisﬁes d (E[a. b] ⊂ I we deﬁne the energy functional E[a. then the variation Φ is called proper if for all t ∈ (−ǫ. γ γ. γ(t))dt. 8 Then we know that the map φ : S m → Sym(Rm+1 ) with A. Deﬁnition 7. dt We shall now prove that geodesics can be characterized as the critical points of the energy functional.e. γ(0) = v and that γ satisﬁes the ˙ geodesic equation (4). v) ∈ T S m deﬁne the curve γ = γ(p.v) : R → S m by γ:t→ p cos(|v|t) · p + sin(|v|t) · v/|v| if v = 0 if v = 0. ǫ). of the form I = [a. If the interval is compact i.

h). . Φ : (t. AMS(1993). Furthermore X(0. g) → (N. A C 2 -curve γ : I = [a. dΦ(∂/∂t)] dΦ([∂/∂s. g) → (N. ∇ X))ds X ds b a = [g(Y. h) between Riemannian manifolds.b] (Φt ))|t=0 = − g(Y (0. ∂/∂t] = 0. X)ds X b = a = a ( d (g(Y. Translations of Mathematical Monographs 132. ∇ Y −∇ X = X Y = = = [X. so Y (a) = Y (b) = 0. X)]b − a g(Y. For a reference on harmonic maps see H. The following shows that the vector ﬁelds X and Y commute. GEODESICS Theorem 7. (∇˙ γ)(s))ds. s) → Φ(t. b] → M is a critical point for the energy functional if and only if it is a geodesic. γ˙ A geodesic γ : I → (M. γ˙ dt a The last integral vanishes for every proper variation Φ of γ if and only if ∇˙ γ = 0. Proof. h) which parametrize the so called minimal surfaces in (N. We now assume that Φ is a proper variation of γ. Urakawa. s). X)ds) dt 2 dt a 1 b d = (g(X. s) = ∂Φ/∂s(0. g) is a special case of what is called a harmonic map φ : (M. ǫ) × I → M .b] (Φt )) = ( g(X. ∇ X)ds. X The variation is proper. X)ds Y g(∇ Y . since [∂/∂s. X))ds 2 a dt b = a b g(∇ X. For a C 2 -map Φ : (−ǫ. s) we deﬁne the vector ﬁelds X = dΦ(∂/∂s) and Y = dΦ(∂/∂t) along Φ. ∂/∂t]) 0. Other examples are conformal immersions ψ : (M 2 . Then b d 1d (E[a. Y ] [dΦ(∂/∂s). so ˙ b d (E[a.76 7. X)) − g(Y. Calculus of Variations and Harmonic Maps.17. s) = γ(s).

7. βv | v ∈ Sp } m is non-empty. v) = 1} be the unit sphere in the tangent space Tp M at p. It then follows from ˜ the construction of the exponential map. that the geodesics in (U. Let (M. Let p ∈ M . βv ) → M be the maximal geodesic such that αv . It can be shown that the real number ˙ is positive so the open ball m−1 ǫp = inf{αv . Theorem 7. locally we have γv = expp ◦λv . One can prove that the map expp is smooth and it follows from its deﬁnition that the diﬀerential m Bǫp (0) = {v ∈ Tp M | gp (v. where rw = |w| and m−1 m−1 vw = w/|w| ∈ Sp . γv (0) = p and γv (0) = v. g ) → (V. g) into an isometry. βv ∈ R+ ∪ {∞}. The exponential map expp : Bǫp (0) → M at p is deﬁned by p if w = 0 expp : w → γvw (rw ) if w = 0. Y ∈ C ∞ (T U ) we have d(expp )0 : Tp M → Tp M g (X. g ) ˜ .e. The next result shows that the geodesics are locally the shortest paths between their endpoints. ˜ This turns φ : (U. Then the geodesics are locally the shortest path between their end points. U = Br (0) in Tp M and V = expp (U ) be such that the restriction φ = expp |U : U → V ˜ of the exponential map at p is a diﬀeomorphism. v) < ǫ2 } p is the identity map for the tangent space Tp M . For v ∈ Sp let γv : (−αv . m−1 Note that for v ∈ Sp the line segment λv : (−ǫp . Then every point w ∈ Tp M \ {0} can be written as w = rw · vw .18. Then the inverse mapping theorem tells us that there exists an rp ∈ R+ such that if m Up = Brp (0) and Vp = expp (Up ) then expp |Up : Up → Vp is a diﬀeomorphism parametrizing the open subset Vp of M . Y ) = g(dφ(X). m Proof. GEODESICS 77 Let (M m . We deﬁne a metric g on U such that for each X. p ∈ M and m−1 Sp = {v ∈ Tp M | gp (v. g) be a Riemannian manifold. ǫp ) → Tp M with λv : t → t · v is mapped onto the geodesic γv i. dφ(Y )). g) be an m-dimensional Riemannian manifold.

Xk ). Then the ˙ ˙ radial component σrad of σ is the orthogonal projection of σ onto the ˙ ˙ ˙ line generated by X i. h) be a Riemannian manifold and M be a submanifold with the induced metric g.e. g (X(t). X(t)) = ˜ |X(t)| = . X(t)) ˜ d g (σ(t). Then the mean curvature vector ﬁeld of M in N is the smooth section H : M → N M of the normal bundle N M given by 1 1 trace B = H= m m m B(Xk . Further let σ : [0. 1] → Br (0) be the curve λq : t → t · q. Deﬁnition 7. X(t)) ˜ ˙ d g (X(t). Let (N. 1] → U be any curve such that σ(0) = 0 and σ(1) = q. σrad : t → ˙ Then it is easily checked that |σrad (t)| = ˙ and |˜(σ(t). Along the curve σ we deﬁne the vector ﬁeld X with X : t → σ(t) and the tangent ﬁeld σ : t → σ(t) to σ.19. dt dt |X(t)| Combining these two relations we yield d |σrad (t)| ≥ |X(t)|. k=1 . GEODESICS through the point 0 = φ−1 (p) are exactly the lines λv : t → t · v where v ∈ Tp M . X(t))| g ˙ |X(t)| g (σ(t).78 7. ˙ dt This means that 1 L(σ) = 0 1 |σ(t)|dt ˙ |σrad (t)|dt ˙ 1 ≥ ≥ 0 d |X(t)|dt 0 dt = |X(1)| − |X(0)| = |q| = L(λq ). This proves that in fact γ is the shortest path connecting p and q. X(t)) ˜ ˙ X(t). m m Now let q ∈ Br (0) \ {0} and λq : [0.

Corollary 7. GEODESICS 79 Here B is the second fundamental form of M in N and {X1 .4.3. Proof. Let (N. so following Proposition 7.v) : I → N be the maximal geodesic in N with γ(0) = p and γ(0) = v. (b) γ : I → M is a geodesic in N . Then M is totally geodesic in (N. The map φ : N → N is an ˙ isometry so φ ◦ γ : I → N is a geodesic. Then M is totally geodesic in N .21 the submanifold M is totally geodesic in N . Proof. Let (N. The uniqueness result of Theorem 7. Hence the image of the geodesic γ : I → N is contained in M . The submanifold M is said to be minimal in N if H ≡ 0 and totally geodesic in N if B ≡ 0. h) be a Riemannian manifold and M be a submanifold equipped with the induced metric g. Proof. Let (N. Proposition 7. Then there exists (locally) at most one totally geodesic submanifold M of (N. .20. v ∈ Tp M and γ : I → N be the maximal geodesic with γ(0) = p and γ(0) = v. Xm } is any local orthonormal frame for the tangent bundle T M of M . See Exercise 7.23. φ(γ(0)) = γ(0) and dφ(γ(0)) = γ(0) then imply that ˙ ˙ φ(γ) = γ.v) (I) ⊂ M for all (p. See Exercise 7. h) be a Riemannian manifold and M be a complete submanifold of N . . v) of the tangent bundle T M let γ(p. γ). Let (N. h) be a Riemannian manifold. For a point (p.7. .9. then the following conditions are equivalent (a) γ : I → M is a geodesic in M . v) ∈ T M . Let p ∈ M . The result is a direct consequence of the following decomposition formula ˜ ˙ ∇˙ γ = (∇˙ γ)⊤ + (∇˙ γ)⊥ = ∇˙ γ + B(γ. Then the following conditions are equivalent: (i) M is totally geodesic in N (ii) if γ : I → M is a curve. h) if and only if ˙ γ(p. ˙ ˙ γ˙ γ˙ γ˙ γ Proposition 7.22. Proposition 7. Proof.21. p ∈ N and V be an m-dimensional linear subspace of the tangent space Tp N of N at p. h) be a Riemannian manifold and M be a submanifold of N which is the ﬁxpoint set of an isometry φ : N → N . . h) such that Tp M = V . .

Y Rm . 0) ∈ H m | x ∈ Rn } is totally geodesic in H m . . Proof. y) → (x. Then the n-dimensional hyperbolic space H n = {(x.80 7. The statement is a direct consequence of the fact that S n is the ﬁxpoint set of the isometry φ : S m → S m of S m with (x. Proof.6. 0) ∈ Rn+1 × Rm−n | |x|2 = 1} Corollary 7. xm ) ∈ H . S m = {(x. . See Exercise 7.25. −y).24. . Then the ndimensional sphere is a totally geodesic submanifold of S n = {(x. Let n < m be positive integers. GEODESICS Corollary 7. Let H m be the m-dimensional hyperbolic space modelled on the upper half space R+ × Rm−1 equipped with the Riemannian metric 1 g(X. Y ) = 2 X. . . x1 m where x = (x1 . y) ∈ Rn+1 × Rm−n | |x|2 + |y|2 = 1}.

Prove that a C 2 -curve γ : (−ǫ. Let n be a positive integer and O(n) be the orthogonal group equipped with the standard left-invariant metric g(A. Determine the totally geodesic submanifolds of the m-dimensional real projective space RP m .4.5. B) = trace(At B). For the real parameter θ ∈ (0. β ∈ R}.3. Exercise 7. Exercise 7. B) = trace(At B) and let K be a Lie subgroup of O(n).8. Exercise 7. Tθ2 = {(cos θeiα . z2 ) ∈ C2 | |z1 |2 + |z2 |2 = 1}.6.21.7. Find a proof for Corollary 7.2.22. Find a proof for Corollary 7. y) ∈ R2 | y ∈ R+ } equipped with the Riemannian metric 1 g(X. sin θeiβ ) ∈ S 3 | α. Exercise 7. ¨ ¨ Exercise 7. Let the orthogonal group O(n) be equipped with the left-invariant metric g(A. The result of Exercise 5. Y R2 .6 is isometric to the upper half plane M = ({(x. y Use your local library to ﬁnd all geodesics in (M. . Find a proof for Proposition 7. GEODESICS 81 Exercises Exercise 7.7. π/2) deﬁne the 2dimensional torus Tθ2 by Determine for which θ ∈ (0.1. Y ) = 2 X. g).3 shows that the two dimensional hyperbolic disc H 2 introduced in Example 5. Prove that K is totally geodesic in O(n).25. Exercise 7. Exercise 7. π/2) the torus Tθ2 is a minimal submanifold of the 3-dimensional sphere S 3 = {(z1 . ǫ) → O(n) is a geodesic if and only if γ t · γ = γ t · γ.

.

We determine the Riemannian curvature tensor for manifolds of constant sectional curvature and also for an important class of Lie groups. Z) in the direction of X applying the naive ”product rule” we get ∇ (A(Y. X X We shall now generalize this and introduce the covariant derivative of tensor ﬁelds of type (r. Then to each vector ﬁeld X ∈ C ∞ (T M ) we have the ﬁrst order covariant derivative ∇ : C ∞ (T M ) → C ∞ (T M ) X in the direction of X satisfying ∇ : Z → ∇ Z. Z)) − A(∇ Y ) − A(∇ Z). Let (M. We then derive the important Gauss equation comparing the sectional curvatures of a submanifold and that of its ambient space. Let (M. 0) or (r. the standard spheres and the hyperbolic spaces all have constant sectional curvature. This naive idea turns out to be very useful and leads to the following formal deﬁnition. We prove that the Euclidean spaces.1.CHAPTER 8 The Riemann Curvature Tensor In this chapter we introduce the Riemann curvature tensor and the notion of sectional curvature of a Riemannian manifold. Deﬁnition 8. These generalize the Gaussian curvature playing a central role in classical diﬀerential geometry. g) be a Riemannian manifold with Levi∞ ∞ Civita connection ∇. 1). X X X X where ∇ A is the ”covariant derivative” of the tensor ﬁeld A in the X direction of X. Z) = ∇ (A(Y. Z)) = (∇ A)(Y. 1). let us assume that A is a tensor ﬁeld of type (2. g) be a Riemannian manifold and ∇ be its Levi-Civita connection. For a tensor ﬁeld A : Cr (T M ) → C0 (T M ) of 83 . As motivation. Z) + A(∇ Y ) + A(∇ Z) X X X X (∇ A)(Y. If we diﬀerentiate A(Y.

. See Exercise 8. Proposition 8.1. . X1 .e. Xr ) → (∇ B)(X1 . g) be a Riemannian manifold with Levi∞ ∞ Civita connection ∇. 1) is said to be parallel if ∇B ≡ 0. Let (M. . . Xr ) → (∇ A)(X1 . . Let (M. . . . (f X Y This means that a vector ﬁeld Z ∈ C ∞ (T M ) on M induces a natural ∞ ∞ tensor ﬁeld Z : C1 (T M ) → C1 (T M ) of type (1. Then its Levi-Civita connection ∇ is tensorial in its ﬁrst argument i. .84 8. Xk−1 . . . compatibility of the Levi-Civita connection ∇ of (M. X Deﬁnition 8. . ∇ Xk . . Xr ) = X r ∇ (B(X1 . Then the metric g is a parallel tensor ﬁeld of type (2. . . yet another. . Then the second order covariant derivative ∇2 Y : C ∞ (T M ) → C ∞ (T M ) X. Y ∈ C ∞ (T M ) and f. . . . 0) we deﬁne its covariant derivative by ∞ ∞ ∇A : Cr+1 (T M ) → C0 (T M ) ∇A : (X. 0) is said to be parallel if ∇A ≡ 0. g) be a Riemannian manifold. . Xk+1 . Xk−1 . . Xr )) − k=1 A(X1 . ∇ Xk . g) with Levi-Civita connection ∇.2. . . . 1) we deﬁne its covariant derivative by ∞ ∞ ∇B : Cr+1 (T M ) → C1 (T M ) ∇B : (X. A tensor ﬁeld B of type (r. Xr ). X A tensor ﬁeld A of type (r. . if X. Xr ). g) be a Riemannian manifold. . Xr )) − X k=1 B(X1 . 1) given by Z : X → ∇ Z. Let X. . For a tensor ﬁeld B : Cr (T M ) → C1 (T M ) of type (r. . . g) with the Riemannian metric g. THE RIEMANN CURVATURE TENSOR type (r. . . 0). Let (M. . . . X Deﬁnition 8.4. Proof. .3. . Xr ) = X r X(A(X1 . The following result can be seen as. g ∈ C ∞ (M ) then ∇ X + gY )Z = f ∇ Z + g∇ Z. Y ∈ C ∞ (T M ) be two vector ﬁelds on the Riemannian manifold (M. . . X1 . . Xk+1 . . .

See Exercise 8. satisﬁes ∇2 Y Z = ∇ (Z(Y )) − Z(∇ Y ) = ∇ ∇ Z − ∇ Y Z. Z. Y + Z)(Y + Z) − R(X.3. Y )Z = ∇2 Y Z − ∇2 XZ. The following result shows that the curvature tensor has many nice properties of symmetry. Y )Z = −R(Y.8. Y.6. Y )Z.4 we see that if X. Let (M. X)Y + R(Y. (ii) g(R(X. W )X. X. W ∈ C ∞ (T M ) on M we then have (i) R(X. For vector ﬁelds X. (v) 6 · R(X.2. Y ). X)Z. W ) = g(R(Z. Z ∈ C ∞ (T M ) then the second order covariant derivative ∇2 Y X. Y )Z = ∇ ∇ Z − ∇ ∇ Z − ∇ Y ]Z X Y Y X [X. g) be a Riemannian manifold. Proof. Theorem 8. (iv) g(R(X. Y − Z)(Y − Z) + R(X + Z. Y )(X − Z). Let (M. See Exercise 8. g) be a Riemannian manifold with Levi∞ ∞ Civita connection ∇. The next remarkable result shows that the curvature operator is a tensor ﬁeld. R(X. is a tensor ﬁeld on M of type (3. (iii) R(X. Y. THE RIEMANN CURVATURE TENSOR 85 is deﬁned by where Z is the natural (1. The reader should note that the Riemann curvature tensor R is an intrinsic object since it only depends on the intrinsic Levi-Civita connection ∇. Proof. Y )Z + R(Z. Y )W. g) be a Riemannian manifold with LeviCivita connection ∇. W ) = −g(R(X. Let (M. As a direct consequence of Deﬁnitions 8. R : C ∞ (T M ) × C ∞ (T M ) × C ∞ (T M ) → C ∞ (T M ) ∇2 Y : Z → (∇ Z)(Y ). Y )Z. Z). X Deﬁnition 8. Y )(X + Z) − R(X − Z. Y.e.3 and 8. Then we deﬁne its Riemann curvature operator as twice the skew-symmetric part of the second covariant derivative ∇2 i. X. Z)X = 0. X . 1). Proposition 8. Y )Z = R(X. 1)-tensor ﬁeld induced by Z ∈ C ∞ (T M ).5. X X X Y ∇ X. Then the curvature R : C3 (T M ) → C1 (T M ) R(X.7.

W ∈ Tp M be tangent vectors at p such that the two sections spanR {X. Z) = . Y )Y. The Riemannian manifold (M. Y } and spanR {Z. g) be a Riemannian manifold and p ∈ M Then the function Kp : G2 (Tp M ) → R given by Kp : spanR {X. g) be a Riemannian manifold. Y }). We often write K(X. Y )2 is called the sectional curvature at p. Then we deﬁne the functions δ. Deﬁnition 8. See Exercise 8. W } are identical.10. Let (M. This will not be treated here. p ∈ M and X. THE RIEMANN CURVATURE TENSOR Part (iii) of Proposition 8. Z. The second Bianchi identity is a similar result concerning the covariant derivative ∇R of the curvature tensor. Deﬁnition 8.7 is the so called ﬁrst Bianchi identity. Let (M. X) |X|2 |Y |2 − g(X. W )W. X) g(R(Z. The set of sections is called the Grassmannian of 2-planes at p. p ∈ M and Kp : G2 (Tp M ) → R be the sectional curvature at p. g) is said to be (i) of positive curvature if δ(p) ≥ 0 for all p. Y ) for K(spanR {X. g) be a Riemannian manifold and p ∈ M . W )2 G2 (Tp M ) = {V | V is a section of Tp M } Proof. (iii) of negative curvature if ∆(p) ≤ 0 for all p. (iv) of strictly negative curvature if ∆(p) < 0 for all p.9.8.4. Y )Y. (ii) of strictly positive curvature if δ(p) > 0 for all p. Let (M. Y } → g(R(X. Y. Then a section V at p is a 2-dimensional subspace of the tangent space Tp M . (v) of constant curvature if δ = ∆ is constant. ∆ : M → R by δ:p→ V ∈G2 (Tp M ) min Kp (V ) and ∆ : p → V ∈G2 (Tp M ) max Kp (V ). This leads to the following natural deﬁnition of the sectional curvature. 2 |Y |2 − g(X. Lemma 8. g) be a Riemannian manifold. Y )2 |X| |Z|2 |W |2 − g(Z. Let (M.86 8. (vi) ﬂat if δ ≡ ∆ ≡ 0. The next result shows how the curvature tensor can be expressed in terms of local coordinates. Then g(R(X. .

gij = g(Xi . Example 8. This we present in Corollary 8. Xj )Xk . Xj )Xk = ∇ ∇ Xk − ∇ ∇ Xk Xi Xj Xj Xi m = s=1 m {∇ (Γs · Xs ) − ∇ (Γs · Xs )} Xi jk Xj ik s ∂Γjk · Xs + ∂xi m = s=1 m Γs Γr Xr jk is r=1 m ∂Γs ik · Xs − − ∂xj m r Γs Γjs Xr ik r=1 = s=1 ∂Γs jk ∂xi − ∂Γs ik + ∂xj r=1 {Γr Γs − Γr Γs } Xs .16. p ∈ M and ˜ Y ∈ Tp M . First we need some preparations. jk ir ik jr k where Γij are the Christoﬀel symbols of the Levi-Civita connection ∇ of (M. x) be local coordinates on M . l = 1. Using the fact that [Xi .11. .13. .12. j. g) with respect to (U. ∂/∂xm } is a global frame for the tangent bundle T Rm . The standard sphere S m has constant sectional curvature +1 (see Exercises 8. For i. Xl ). . Lemma 8.9). Let (M. THE RIEMANN CURVATURE TENSOR 87 Proposition 8. . Rm the set {∂/∂x1 .8.7 and 8. ∂xi s ∂Γjk ∂Γs ik − + ∂xi ∂xj m Rijkl = s=1 gsl r=1 {Γr · Γs − Γr · Γs } .11. . Xj ] = 0 we obtain R(Xi . so Γk ≡ 0 ij by Example 6. k. . Let (M. . Proof. Xj ) and Rijkl = g(R(Xi . . This implies that R ≡ 0 so E m is ﬂat. Then the map Y : Tp M → Tp M given by ˜ Y : X → R(X. jk ir ik jr For the m-dimensional vector space Rm equipped with the Euclidean metric . m put Xi = Then m ∂ .8) and the hyperbolic space H m has constant sectional curvature −1 (see Exercise 8. In this situation we have gij = δij . g) be a Riemannian manifold. x). Our next aim is a formula for the curvature tensor for manifolds of constant sectional curvature. Y )Y is a symmetric endomorphism of the tangent space Tp M . g) be a Riemannian manifold and let (U. .

Y )Y. Y X − X. Z) = g(R(X. Let (M. Y ) ˜ = g(R(Z. Z be vector ﬁelds on M . Y ) = 0}. Y. Z)Y. Y )Z| ≤ 3 (∆ − δ)|X||Y ||Z| 2 Proof. |X| = 1 and Y (X) = λX then This means that the eigenvalues satisfy δ(p) ≤ λ1 (p) ≤ · · · ≤ λm−1 (p) ≤ ∆(p). X) = λ. The ﬁrst statement follows from the fact that the symmetric endomorphism of Tp M with X → {R(X. Z)X. Y )Y. . Y )Y. Without loss of generality we can assume that |X| = |Y | = |Z| = 1. ˜ The fact that Y (Y ) = 0 and Lemma 8. ˜ Kp (X. g) be a smooth Riemannian manifold and X. Z)X − g(X. Y (Z)). Y ) = g(R(X. Z) = g(R(Y. 2 2 . Then 1 (i) |R(X. For Z ∈ Tp M we have ˜ g(Y (X). ∆−δ ]. Without loss of generality. Then ∞ ∞ deﬁne the smooth tensor ﬁeld R1 : C3 (T M ) → C1 (T M ) of type (3. Y )Z = g(Y. Proposition 8. Y Y = X. Deﬁnition 8. .88 8. 1) by R1 (X. X) = g(X. . . Y )Y = Y. X) = g(Y (X). we can assume that the corresponding eigenvalues satisfy ˜ If X ∈ N (Y ). Then R1 (X. For a tangent vector Y ∈ Tp M with |Y | = 1 let N (Y ) be the normal space to Y N (Y ) = {X ∈ Tp M | g(X. Y )Y − δ+∆ R1 (X.15. THE RIEMANN CURVATURE TENSOR Proof. Let (M. Y )Z − δ+∆ R1 (X. Y )Z = R(X ⊥ . Xm−1 for the restriction of ˜ the symmetric endomorphism Y to N (Y ). Y )Y − ∆+δ · X} 2 λ1 (p) ≤ · · · ≤ λm−1 (p). If X = X ⊥ + X ⊤ with X ⊥ ⊥ Y and X ⊤ is a multiple of Y then R(X. restricted to N (Y ) has eigenvalues in the interval [ δ−∆ . Y )Z and |X ⊥ | ≤ |X| so we can also assume that X ⊥ Y .14.13 ensure the existence of an orthonormal basis of eigenvectors X1 . g) be a Riemannian manifold. Y )Y | ≤ 2 (∆ − δ)|X||Y |2 2 2 (ii) |R(X.

Let (G. Y. W ) ˜ ˜ ˜ ˜ ˜ ˜ ˜ ˜ = h(B(Y . B(X.18. ad(X) : g → g . Theorem 8. Proof. Z]. Y )Z = − [[X. B(Y .6. Y − Z)(Y − Z) + D(X + Z. Let (N. We shall now prove the important Gauss equation comparing the curvature tensors of a submanifold and its ambient space in terms of the second fundamental form. Then for any left-invariant vector ﬁelds X. Z). W )). Proposition 8. 4 Proof. Y )Z = D(X. g) be a Riemannian manifold of constant curvature κ. See Exercise 8. Y )Z| ≤ 2 +|Y |(|X + Z|2 + |X − Z|2 )} 1 (∆ − δ){2|X|(|Y |2 + |Z|2 ) + 2|Y |(|X|2 + |Z|2 )} = 2 = 4(∆ − δ). W ∈ C ∞ (T M ). Z)Y ). g) be a Lie group equipped with a leftinvariant metric such that for all X ∈ g the endomorphism is skew-symmetric with respect to g. This implies that 2 6 · D(X. The second statement then follows from 1 (∆ − δ){|X|(|Y + Z|2 + |Y − Z|2 ) 6|D(X. W ) − h(R(X. Z).16. Then the curvature tensor R is given by R(X.15 we have the following useful result. The result follows directly from κ = δ = ∆. h) be a Riemannian manifold and M be a submanifold of N equipped with the induced metric g. Y )Z = κ · (g(Y. As a direct consequence of Proposition 8.7 and hence D = R − ∆+δ · R1 as well. Z ∈ g the curvature tensor R is given by 1 R(X. Y )(X − Z). Then ˜ ˜ ˜ ˜ ˜ g(R(X. Y )Z. W ∈ ˜ ˜ ˜ ˜ C ∞ (T N ) be vector ﬁelds extending X. Z)X − g(X. Let (M. Z.17. Y )Z. Y ]. Y + Z)(Y + Z) − D(X. Let X. Corollary 8.8. Y . THE RIEMANN CURVATURE TENSOR 89 It is easily checked that the operator R1 satisﬁes the conditions of Proposition 8. Y. W )) − h(B(X. Y )(X + Z) − D(X − Z. Z.

X) > = det Sp . = h((∇ (∇ Z − B(Y. Sp (X) > N. If we now apply the fact that R3 is ﬂat. W ) + ∇ (B(X. Y ). W ) [X. W )) − h(B(X. W ) X Y Y X [ −h((∇ Y ]Z − B([X. We shall now apply the Gauss equation to the classical situation of a regular surface Σ as a submanifold of the 3-dimensional Euclidean ˜ ˜ space R3 . Y )Z. If X. R. Y ]. Y ) of Σ at p satisﬁes ˜ ˜ ˜ ˜ ˜ ˜ ˜ K(X. Y . X) > − < B(X. dN (X) > N = < Y. B(Y . Z)). the sectional curvature K(X. Y ) = < R(X. X > ˜ ˜ ˜ ˜ ˜ ˜ ˜ ˜ = < B(Y .e. Y )Z. Let {X. ∇ W ) X Y = h(R(X. N > N = − < Y. Z)))⊤ − h(∇ (∇ Z − B(X. Z)).90 8. where Sp : Tp Σ → Tp Σ is the shape operator at p. the ˜ ˜ Levi-Civita connection ∇ and the second fundamental form of M in M we obtain ˜ ˜ ˜ ˜ ˜ g(R(X. B(Y. Y ]Z. Y ) = < ∂XY . Z)))⊤ . Z). B(X. N are local extensions of X. W ) ˜ ˜ ˜ ˜ ˜ ˜ ˜ ˜ ˜ ˜ ˜ = g(∇˜∇˜ Z − ∇˜ ∇˜Z − ∇X. W ) h((B(Y. Z). ˜ ˜ In other word. . Y. Y )Z. W )). Z))⊤ . ∂XN > N = − < Y. the classical Gaussian curvature. Using the deﬁnitions of the curvature tensors R. Z)). Y ) is the determinant of the shape operator Sp at p i. W ) X Y Y X −h(∇ (B(Y. W ) X Y Y X [X. W ) +h(B(Y. ∇ W ) − h(B(X. Z)). Y )Y . Y ). the Gauss equation tells us that the sectional ˜ ˜ curvature K(X. N then the second 3 fundamental form B of Σ in R satisﬁes ˜ ˜ B(X. W ) X Y = h(R(X. Y } be a local orthonormal frame for the tangent ˜ bundle T Σ around a point p ∈ Σ and N be the local Gauss map with ˜ ˜ ˜ ˜ ˜ ˜ N = X × Y . THE RIEMANN CURVATURE TENSOR ˜ Proof. = h(∇ ∇ Z − ∇ ∇ Z − ∇ Y ]Z. B(X.

ej ) − g(ei . ei )ei . then ∞ ∞ (i) the Ricci operator r : C1 (T M ) → C1 (M ) is deﬁned by m r(X) = i=1 R(X. ei )ei . g) be a Riemannian manifold of constant sectional curvature κ. Proof. W ∈ C ∞ (T N ) be ˜ ˜ ˜ ˜ vector ﬁelds extending X. Then the following holds s = m · (m − 1) · κ.8 and 8. We conclude this chapter by deﬁning the Ricci and scalar curvatures of a Riemannian manifold. ej ) m m = i=1 = κ( i=1 g(ei . ei ). W ) = h(R(X. Deﬁnition 8. .21. .16 implies that m Ric(ej . Y .8. h) be a Riemannian manifold and (M. W ). Then ˜ ˜ ˜ ˜ ˜ g(R(X. ej )g(ei . Corollary 8. g) be a Riemannian manifold. ej )ej . ej ) g(κ(g(ei . ei )ej − g(ej . em } is any local orthonormal frame for the tangent bundle. see for example Exercises 8. Here {e1 . .19. ei )ei . Let {e1 . Let (M. ei )ei ).20. ej )) i=1 . Let X. . Let (N. Y. . ∞ ∞ (ii) the Ricci curvature Ric : C2 (T M ) → C0 (T M ) by m Ric(X. Z. Y )Z. Y ) = i=1 g(R(X. . Y ). W ∈ C ∞ (T M ). . then Corollary 8. and (iii) the scalar curvature s ∈ C ∞ (M ) by m m m s= j=1 Ric(ej . g) be a totally geodesic submanifold of N .9. Z. em } be an orthonormal basis. ej ) = i=1 m g(R(ej . ej ) = j=1 i=1 g(R(ei . These are obtained by taking traces over the curvature tensor and play an important role in Riemannian geometry. . THE RIEMANN CURVATURE TENSOR 91 As a direct consequence of the Gauss equation we have the following useful result. ei )g(ej . Corollary 8. Let (M. Y )Z.

92

8. THE RIEMANN CURVATURE TENSOR

m

m

= κ(

i=1

1−

i=1

δij ) = (m − 1) · κ.

To obtain the formula for the scalar curvature s we only have to multiply the constant Ricci curvature Ric(ej , ej ) by m. For further reading on diﬀerent notions of curvature we recommend the interesting book, Wolfgang K¨hnel, Diﬀerential Geometry: Curves u - Surfaces - Manifolds, AMS (2002).

8. THE RIEMANN CURVATURE TENSOR

93

Exercises

Exercise 8.1. Let (M, g) be a Riemannian manifold. Prove that the tensor ﬁeld g of type (2, 0) is parallel with respect to the Levi-Civita connection. Exercise 8.2. Let (M, g) be a Riemannian manifold. Prove that the curvature R is a tensor ﬁeld of type (3, 1). Exercise 8.3. Find a proof for Proposition 8.7. Exercise 8.4. Find a proof for Lemma 8.8. Exercise 8.5. Let Rm and Cm be equipped with their standard Euclidean metric g given by

m

g(z, w) = Re

k=1

zk wk ¯

and let be the m-dimensional torus in Cm with the induced metric. Find an isometric immersion φ : Rm → T m , determine all geodesics on T m and prove that the torus is ﬂat. Exercise 8.6. Find a proof for Proposition 8.17. Exercise 8.7. Let the Lie group S 3 ∼ SU(2) be equipped with the = metric 1 ¯ g(Z, W ) = Re(trace(Z t W )). 2 (i) Find an orthonormal basis for Te SU(2). (ii) Prove that (SU(2), g) has constant sectional curvature +1. Exercise 8.8. Let S m be the unit sphere in Rm+1 equipped with the standard Euclidean metric , Rm+1 . Use the results of Corollaries 7.24, 8.19 and Exercise 8.7 to prove that (S m , , Rm+1 ) has constant sectional curvature +1. Exercise 8.9. Let H m be the m-dimensional hyperbolic space modelled on the upper half space R+ ×Rm−1 equipped with the Riemannian metric 1 g(X, Y ) = 2 X, Y Rm , x1 T m = {z ∈ Cm | |z1 | = ... = |zm | = 1}

94

8. THE RIEMANN CURVATURE TENSOR

where x = (x1 , . . . , xm ) ∈ H m . For k = 1, . . . , m let the vector ﬁelds Xk ∈ C ∞ (T H m ) be given by ∂ (Xk )x = x1 · ∂xk m and deﬁne the operation ∗ on H by Prove that (i) (H m , ∗) is a Lie group, (ii) the vector ﬁelds X1 , . . . , Xm are left-invariant, (iii) [Xk , Xl ] = 0 and [X1 , Xk ] = Xk for k, l = 2, ..., m, (iv) the metric g is left-invariant, (v) (H m , g) has constant curvature −1. Compare with Exercises 6.4 and 7.1. (α, x) ∗ (β, y) = (α · β, α · y + x).

CHAPTER 9 Curvature and Local Geometry This chapter is devoted to the study of the local geometry of Riemannian manifolds and how this is controlled by the curvature tensor. ǫ) the vector ﬁeld Jt : I → C ∞ (T M ) along γt given by Jt (s) = Φ : (−ǫ. The variable t ∈ (−ǫ. Proposition 9. By a smooth 1-parameter family of geodesics we mean a C ∞ -map such that the curve γt : I → M given by γt : s → Φ(t. s) is a geodesic for all t ∈ (−ǫ. Let (M. ǫ). = −∇ ∇ X X J = −∇ ∇ J. X)X = ∇ ∇ X − ∇ ∇ X − ∇ X]X J X X J [J.1. With this at hand we yield a fundamental comparison result describing the curvature dependence of local distances. X] = [dΦ(∂/∂t). γt )γt = 0. Along Φ we put X(t. For this we introduce the notion of a Jacobi ﬁeld which is a useful tool in diﬀerential geometry. s) ∂t satisﬁes the second order ordinary diﬀerential equation ∇˙ ∇˙ Jt + R(Jt . g) be a Riemannian manifold and Φ : (−ǫ. ∂/∂s] = 0 implies that [J. X X 95 . Then for each t ∈ (−ǫ. ǫ) × I → M be a 1-parameter family of geodesics. The fact that [∂/∂t. Let (M. s) = ∂Φ/∂t. ǫ) × I → M ∂Φ (t. ∂/∂s]) = 0. ǫ) is called the family parameter of Φ. dΦ(∂/∂s)] = dΦ([∂/∂t. g) be a Riemannian manifold. ˙ ˙ γt γt Proof. s) = ∂Φ/∂s and J(t. Since Φ is a family of geodesics we have ∇ X = 0 and the deﬁnition X of the curvature tensor then gives R(J.

If J is a vector ﬁeld along γ. s)|t=0 = c(0) + s · n(0). . Let c. If we deﬁne a map Φ : R × R → E m+1 by then for each t ∈ R the curve γt : s → Φ(t. . such that Jp = v and (∇ J)p = w. .3. Xi ) are smooth functions on I. Let {X1 . γt )γt = 0. 0 ∈ I.8. w ∈ Tp M are two tangent vectors at p then ˙ there exists a unique Jacobi ﬁeld J along γ.96 9. We denote the space of all Jacobi ﬁelds along γ by Jγ (T M ). This means that the space of Jacobi ﬁelds Jγ (T M ) along γ is a vector space.4. A C 2 vector ﬁeld J along γ is called a ˙ Jacobi ﬁeld if ∇ ∇ J + R(J.2. Xm are parallel so m m ∇ J= X i=1 ai Xi and ∇ ∇ J = ˙ X X ai Xi . γt γt Deﬁnition 9. ǫ) we have ˙ ˙ ∇˙ ∇˙ Jt + R(Jt . Let γ : I → M be a geodesic. n : R → E m+1 be smooth curves such that the image n(R) of n is contained in the unit sphere S m . . γ : I → M be a geodesic and X = γ. CURVATURE AND LOCAL GEOMETRY Hence for each t ∈ (−ǫ. . ¨ i=1 . ˙ ˙ dt The Jacobi equation (5) on a Riemannian manifold is linear in J. g) be a Riemannian manifold. Example 9. . X Proof. s) is a straight line and hence a geodesic in E m+1 . Let (M. We are now interested in determining the dimension of this space Proposition 9. Xm } be an orthonormal frame of parallel vector ﬁelds along γ. . If v. see Proposition 7. . By diﬀerentiating with respect to the family parameter t we yield the Jacobi ﬁeld J ∈ Jγ0 (T E m+1 ) along γ0 with d J(s) = Φ(t. X)X = 0 X X along γ. p = γ(0) and X = γ along γ. (5) We shall now give an example of a 1-parameter family of geodesics in the (m + 1)-dimensional Euclidean space E m+1 . s) → c(t) + s · n(t) ai Xi where ai = g(J. The vector ﬁelds X1 . then m J= i=1 Φ : (t.

X)X.1.5. Let (M.k=1 ai bk Xk . The following Lemma shows that when proving results about Jacobi ﬁelds along a geodesic γ we can always assume. X)X = i. . If λ ∈ R∗ and σ : λI → I is given by σ : t → t/λ. Next we determine the Jacobi ﬁelds which are tangential to a given geodesic. . See Exercise 9. X)X = k=1 bk Xk . i and that J is a Jacobi ﬁeld if and only if m m (¨i + a i=1 k=1 ak bi )Xi = 0. Proof. X)X is given by m R(J. . ˙ Lemma 9. . CURVATURE AND LOCAL GEOMETRY 97 For the curvature tensor we have m R(Xi . m k of linear ordinary diﬀerential equations in a = (a1 . k This is equivalent to the second order system m ai + ¨ k=1 ak bi = 0 for all i = 1. that |γ| = 1. . X Corollary 9. Let (M.9. . . g) be an m-dimensional Riemannian manifold and γ : I → M be a geodesic in M . . i where bk = g(R(Xi . Xk ) are smooth functions on I depending on i the geometry of (M. g). This means that R(J. then γ ◦ σ : λI → M is a geodesic and J ◦ σ is a Jacobi ﬁeld along γ ◦ σ. g) be a Riemannian manifold. Then the vector space Jγ (T M ) of all Jacobi ﬁelds along γ has the dimension 2m. γ : I → M be a geodesic and J be a Jacobi ﬁeld along γ. without loss of generality. ˙ This implies that J exists globally and is uniquely determined by the initial conditions J(0) = v and (∇ J)(0) = w.6. A global solution will always exist and is uniquely determined by a(0) and a(0). am ). 2. .

γ : I → M be a geodesic and J be a Jacobi ﬁeld along γ. ˙ Corollary 9. cos(√κs) if κ > 0. γ(t0 )) = 0 ˙ ˙ γ for some t0 ∈ I. ˙ Proof. γ) satisﬁes the second order ordinary diﬀerential equation f = 0 ˙ ˙(0) = 0. The fact that Jγ (T M ) is a vector space implies that the normal part J ⊥ = J − J ⊤ of J also is a Jacobi ﬁeld.8. γ)γ) + R(g(J.98 9. γ(t0 )) = 0 and g((∇˙ J)(t0 ). For a real number κ ∈ R we deﬁne the cκ . γ)γ. We now have ∇˙ ∇˙ J ⊤ + R(J ⊤ . and the initial conditions f (0) = 0 and f Our next aim is to show that if the Riemannian manifold (M. g) be a Riemannian manifold. For this we introduce the following notation. cκ (s) = 1 if κ = 0. γ) twice along γ we obtain ˙ d2 (g(J. ˙ Proof. γ)) = g(∇˙ ∇˙ J. then g(J(t). Then J ⊤ and J ⊥ are Jacobi ﬁelds along γ and there exist a. γ : I → M be a geodesic with |γ| = 1 and J be a Jacobi ﬁeld along γ. γ(t)) = 0 for all t ∈ I. This shows that the tangential part J ⊤ of J is a Jacobi ﬁeld. g) be a Riemannian manifold. γ)γ γ γ ˙ ˙ ˙ ˙ ˙ ˙ = −g(R(J. g) has constant sectional curvature then we can solve the Jacobi equation ∇ ∇ J + R(J. γ)γ = ∇˙ ∇˙ (g(J. By diﬀerentiating g(J. γ)γ = 0. γ)γ.7. b ∈ R. γ)γ and J ⊥ = J − J ⊤ ˙ ˙ . CURVATURE AND LOCAL GEOMETRY Proposition 9. γ) = −g(R(J. Let (M. This is a direct consequence of the fact that the function ¨ g(J. γ) = 0 ˙ ˙ ˙ ˙ γ γ ˙ ds2 so g(J. X)X = 0 X X along any given geodesic γ : I → M . sκ : R → R by cosh( |κ|s) if κ < 0. γ)γ. Let J ⊤ ˙ be the tangential part of J given by be its normal part. γ)γ ˙ ˙ ˙ ˙ ˙ ˙ ˙ ˙ γ γ γ γ = g(∇˙ ∇˙ J. J ⊤ = g(J. Let (M. γ(s)) = (as + b) for some a. If g(J(t0 ). b ∈ R such that J ⊤ (s) = (as + b)γ(s) for all s ∈ I.

10. Along the geodesic γ0 : s → s we get the Jacobi ﬁeld J0 (s) = with |J0 (s)| = |s| = |sκ (s)|. Example 9. (1 − tanh2 (s/2))2 . Along the geodesic γ0 : s → (sin(s). sin(√κs)/√κ if κ > 0. cos(s)) we get the Jacobi ﬁeld J0 (s) = ∂Φt (0. s) = (isin(s). It is a well known fact that the unique solution to the initial value problem ¨ f + κ · f = 0. Let S 2 be the unit sphere in the standard Euclidean 3-space C × R with the induced metric of constant sectional curvature κ = +1. Let C be the complex plane with the standard Euclidean metric . Rotations about the R-axis produce a 1-parameter family of geodesics Φt : s → (sin(s)eit . s) = i · tanh(s/2) ∂t 4 .9.9. Let B1 (0) be the open unit disk in the complex plane with the hyperbolic metric g(X. s) = is ∂t with |J0 (s)|2 = sin2 (s) = |sκ (s)|2 . sκ (s) = s if κ = 0. Rotations about the origin produce a 1-parameter family of geodesics Φt : s → tanh(s/2)eit . f (0) = a and f˙(0) = b sinh( |κ|s)/ |κ| if κ < 0. R2 of constant sectional curvature κ = 0. (1 − |z|2 )2 R2 4 · tanh2 (s/2) = sinh2 (s) = |sκ (s)|2 . CURVATURE AND LOCAL GEOMETRY 99 and is the function f : R → R satisfying f (s) = acκ (s) + bsκ (s).11. The rotations about the origin produce a 1-parameter family of geodesics Φt : s → seit . 0) ∂t ∂Φt (0. 2 Example 9. Along the geodesic γ0 : s → tanh(s/2) we get the Jacobi ﬁeld J0 (s) = with |J0 (s)|2 = ∂Φt (0. cos(s)). Y ) = of constant sectional curvature κ = −1. Example 9.

0). i=1 This is equivalent to the following system of ordinary diﬀerential equations ¨ ¨ (6) fm (s) = 0 and fi (s) + κfi (s) = 0 for all i = 1. Pj ) = δij and g(Pi . This means that J is a Jacobi ﬁeld if and only if m−1 i=1 ¨ ¨ fi (s)Pi (s) + fm (s)X(s) = ∇ ∇ J X X = −R(J. . 2. 2. m − 1. It is clear that for the initial values m−1 J(s0 ) = i=1 m−1 vi Pi (s0 ) + vm X(s0 ). ˙ Further let P1 . . . P2 . In the next example we give a complete description of the Jacobi ﬁelds along a geodesic on the 2-dimensional sphere. Let S 2 be the unit sphere in the standard Euclidean 3-space C × R with the induced metric of constant curvature κ = +1 and γ : R → S 2 be the geodesic given by γ : s → (eis . X)X = −κJ ⊥ = −κ = −κ(g(X. . m we have a unique and explicit solution to the system (6) on the whole of I. . .100 9. . . X)J ⊥ − g(J ⊥ . wi Pi (s0 ) + wm X(s0 ) i=1 (∇ J)(s0 ) = X or equivalently fi (s0 ) = vi and f˙i (s0 ) = wi for all i = 1. . Example 9. . Pm−1 be parallel vector ﬁelds along γ such that g(Pi . X)X) m−1 fi (s)Pi (s). X)X = −R(J ⊥ . Then . . .12. X) = 0. g) be a Riemannian manifold of constant sectional curvature κ and γ : I → M be a geodesic with |X| = 1 where X = γ. CURVATURE AND LOCAL GEOMETRY Let (M. Any vector ﬁeld J along γ may now be written as m−1 J(s) = i=1 fi (s)Pi (s) + fm (s)X(s).

0). g) be a Riemannian manifold with sectional curvature uniformly bounded above by ∆ and γ : [0. as long as s(v+tw) m is an element of Bεp (0) . a cos s + b sin s) for some a. (0. b ∈ R. The sphere S 2 is 2-dimensional and γ is parallel ˙ ˙ along γ so P must be parallel. g) be a complete Riemannian manifold. p ∈ M and v.7) that all Jacobi ﬁelds ˙ tangential to γ are given by The vector ﬁeld P : R → T S 2 given by s → ((eis . This implies that all the Jacobi ﬁelds orthogonal to γ are given by ˙ N J(a. Hence the map Φt : s → (exp)p (s(v + tw)) is a 1-parameter family of geodesics through p ∈ M . α] → R is a C 2 -function such that ¨ (a) f + ∆ · f = 0 and f > 0 on (0. α). (ii) if f : [0. b ∈ R. w ∈ Tp M . α).b) (s) = (0. This means that ∂Φt (0. α] → M be a geodesic on M with |X| = 1 where X = γ. γ = 0 and |P | = 1. Let (M. X) = 0 and |J| = 0 on (0. Let (M. (b) f (0) = |J(0)|. CURVATURE AND LOCAL GEOMETRY 101 γ(s) = (ieis . X The following technical result is needed for the proof of the main Theorem 9. Remember that the exponential map m m m (exp)p |Bεp (0) : Bεp (0) → exp(Bεp (0) ) maps lines in Tp M through the origin onto geodesics on M . T J(a. Then (i) d2 (|J|)/ds2 + ∆ · |J| ≥ 0. 1)) satisﬁes P. Then s → s(v +tw) deﬁnes a 1-parameter family of lines in the tangent space Tp M which all pass through the origin 0 ∈ Tp M .13.14 at the end of this chapter.b) (s) = (as + b)(ieis . 0) for some a. X . It is easily veriﬁed that J satisﬁes the initial conditions ˙ J(s) = J(0) = 0 and (∇ J)(0) = w.9. and (c) f˙(0) = |∇ J(0)|. where we do not have constant curvature the exponential map can be used to produce Jacobi ﬁelds as follows. α] → T M ˙ be a Jacobi ﬁeld along γ such that g(J. 0) so it follows from Proposition (9. Lemma 9. s) ∂t is a Jacobi ﬁeld along the geodesic γ : s → Φ0 (s) with γ(0) = p and γ(0) = v. Further let J : [0. In more general situations.

α) → R by h(s) = Then d (|J(s)|)f (s) − |J(s)|f˙(s)) ds s d2 1 ¨ ( 2 (|J(t)|)f (t) − |J(t)|f (t))dt = f 2 (s) 0 dt s 1 d2 = f (t)( 2 (|J(t)|) + ∆ · |J(t)|)dt f 2 (s) 0 dt ≥ 0. f˙(0) = |(∇ J)(0)| so it X follows from (ii) that |(∇ J)(0)| · s∆ (s) = f (s) ≤ |J(s)|. Let p ∈ M . X Proof. J) d2 d2 X g(J. J) = − |J| ≥ −∆ · |J|.e. Let (M∆ . g∆ ) be another Riemannian manifold which is complete and of constant sectional curvature K ≡ ∆. X)X. there exists a ∆ ∈ R such that Kp (V ) ≤ ∆ for all V ∈ G2 (Tp M ) and p ∈ M . J)2 X X X + X = 3 |J| |J| g(∇ ∇ J. J = 0 we obtain d g(∇ J. i. ˙ This implies that h(s) ≥ 0 so f (s) ≤ |J(s)| for all s ∈ (0. α). (i) Using the facts that |X| = 1 and X. J) |∇ J|2 |J|2 − g(∇ J. CURVATURE AND LOCAL GEOMETRY then f (s) ≤ |J(s)| on (0. (iii) The function f (s) = |(∇ J)(0)| · s∆ (s) satisﬁes the diﬀerential X equation ¨ f (s) + ∆f (s) = 0 and the initial conditions f (0) = |J(0)| = 0. = = lims→0 |J(s)| f (s) if s ∈ (0. α). = 1 if s = 0. (iii) if J(0) = 0. g) be a Riemannian manifold of sectional curvature which is uniformly bounded above. α). J) = ( (|J|) = ) ds2 ds2 ds |J| g(∇ ∇ J. then |∇ J(0)| · s∆ (s) ≤ |J(s)| for all s ∈ (0. . p∆ ∈ M∆ and identify Tp M ∼ Rm ∼ Tp∆ M∆ .102 9. α). |J(s)| f (s) (ii) Deﬁne the function h : [0. J) X X ≥ |J| g(R(J. X ˙ h(s) = 1 f 2 (s) ( Let (M.

9. Proof. respectively. Hence Jt = ∂Φt /∂t and Jt∆ = ∂Φ∆ /∂t are t Jacobi ﬁelds satisfying the initial conditions Jt (0) = Jt∆ (0) = 0 and (∇ J t )(0) = (∇ Jt∆ )(0) = v(t). Let (Mδ . p. Furthermore let c : [a. r∆ ) ≤ d(q. b] → M be the side connecting r and q and v : [a. there exists a δ ∈ R such that δ ≤ Kp (V ) for all V ∈ G2 (Tp M ) and p ∈ M . rδ ). = |c(t)| ˙ .14.e. a triangle with sides which are shortest paths between their endpoints. g) is uniformly bounded below i. We now add the assumption that the sectional curvature of the manifold (M. gδ ) be a complete Riemannian manifold of constant sectional curvature δ.13 we now obtain ˙ X X |c∆ (t)| = |Jt∆ (1)| ˙ = |(∇ Jt∆ )(0)| · s∆ (1) X = |(∇ Jt )(0)| · s∆ (1) ≤ |Jt (1)| X The curve c is the shortest path between r and q so we have d(r∆ . For the above situation the following inequality for the distance function d is satisﬁed d(q∆ . Then Φt : s → (exp)p (s · v(t)) and Φ∆ : s → t (exp)p∆ (s · v(t)) are 1-parameter families of geodesics through p ∈ M . r). r) ≤ d(qδ . Put c∆ (t) = (exp)p∆ (v(t)) for t ∈ [a. Then a = = similar construction as above gives two pairs of points q. b] and then it directly follows that c(a) = r and c(b) = q. and p∆ ∈ M∆ . r) ≤ d(qδ . respectively. Let (r. d(q. CURVATURE AND LOCAL GEOMETRY 103 Let U be an open neighbourhood of Rm around 0 such that the exponential maps (exp)p and (exp)p∆ are diﬀeomorphisms from U onto their images (exp)p (U) and (exp)p∆ (U ). r∆ ) ≤ d(q. q∆ ) ≤ L(c∆ ) ≤ L(c) = d(r. Finally put r∆ = c∆ (a) and q∆ = c∆ (b). b] → Tp M be the curve deﬁned by c(t) = (exp)p (v(t)). Let p ∈ M and pδ ∈ Mδ and identify Tp M ∼ Rm ∼ Tpδ Mδ . Using Lemma 9. Deﬁne a 1-parameter family s → s · v(t) of straight lines in Tp M through p. Theorem 9. q). rδ ∈ Mδ and shows that Combining these two results we obtain locally d(q∆ . rδ ).e. r ∈ M and qδ . q) be a geodesic triangle i.

Further let J be a non-vanishing ˙ Jacobi ﬁeld along γ with g(X.2. J) = 0.6. Prove that if g(J. Find a proof for Lemma 9. J) is constant along γ then (M. .1. Exercise 9. g) be a Riemannian manifold and γ : I → M be a geodesic such that X = γ = 0. Let (M.104 9. g) does not have strictly negative curvature. CURVATURE AND LOCAL GEOMETRY Exercises Exercise 9.

Sign up to vote on this title

UsefulNot useful- Eat
- Fibre Bundles
- BelbrunoGideaTopputo-SIAM
- lero08
- Renteln P.-solution Manual for Manifolds, Tensors, And Forms-CUP (2013)
- Nonholonomic Ricci Flows
- Manual
- MIT Calculus Chapter 13
- APS
- 13.6 Parametric Surfaces and Their Area-Part2
- Chapter2.pdf
- Geometry of Homoclinic Connections in a Planar System
- Domains of Effective Stability near L5 in the R3BP
- Gauss Bonet Theorem
- 0401 Correa
- Winitzki - Heidelberg Lectures on Advanced General Relativity 2007
- Berger GR 2006
- Journal Entry #1
- Day4Exercises[1]
- exam3fall
- 01_04 (1)
- HW1
- Calculus I Notes, Section 3-8
- Graph-Based Learning via Auto-Grouped Sparse Regularization and Kernelized Extension.pdf
- A Liveliness Analysis of a Distributed Constrained Coordination Strategy for Multi-Agent Linear Systems
- Math 114 PSet 2
- Juliette Kennedy and Saharon Shelah- More on regular reduced products
- Lecture 01
- THE EXTENICS NORM APPLIED TO A TWO-DIMENSIONAL ROBOTIC WORKSPACES
- welfarism
- Riemann

Are you sure?

This action might not be possible to undo. Are you sure you want to continue?

We've moved you to where you read on your other device.

Get the full title to continue

Get the full title to continue reading from where you left off, or restart the preview.