Beruflich Dokumente
Kultur Dokumente
: U
R
d
(same d for all s) such that
(i) M =
A
U
;
(ii) any two charts are compatible: for every , the change of local coordinates
is a smooth (C
(U
) R
d
.
(iii) the collection of charts
(U
) and
(U
) of R
d
.
In practice, one only needs to worry about the rst two conditions in the above deni-
tion. Given a collection of compatible charts covering M, i.e. satisfying (i) and (ii), there
is a unique way to extend it to a maximal collection to satisfy (iii). I leave this last claim
without proof but refer to Warner, p. 6.
Denition. A topological space equipped with a C
structure.
Remarks . 1. Some variations of the denition of the dierentiable structure are possible.
Much of the material in these lectures could be adapted to C
k
rather than C
dierentiable
manifolds, for any integer k > 0.
2
On the other hand, replacing R
d
with the complex coordinate space C
n
and smooth
maps with holomorphic (complex analytic) maps between domains in C, leads of an im-
portant special class of complex manifoldsbut that is another story.
2
if k = 0 then the denition of dierentiable structure has no content
differential geometry 3
By a manifold I will always mean in these lectures a smooth (real) manifold, unless
explicitly stated otherwise.
2. Here is an example of what can happen if one omits a Hausdor property. Consider
the following line with a double point
M = (, 0) 0
, 0
(0, ),
with two charts being the obvious identity maps (the induced topology is assumed)
1
: U
1
= (, 0) 0
(0, ) R,
2
: U
2
= (, 0) 0
(0, ) R, so
1
(0
) =
1
(0
and 0
cannot be separated).
Omitting the 2nd countable property would allow, e.g. an uncountable collection (dis-
joint union) of lines
.
0<<1
R
,
each line R
equipped with the usual topology and charts being the identity maps R
R.
3
Examples. The Euclidean R
d
is made into a manifold using the identity chart. The
complex coordinate space C
n
becomes a 2n-dimensional manifold via the chart C
n
R
2n
replacing every complex coordinate z
j
by a pair of real coordinates Re z
j
, Imz
j
.
The sphere S
n
= x R
n+1
:
n
i=0
x
2
i
= 1 is made into a smooth manifold of
dimension n, by means of the two stereographic projections onto R
n
= x R
n+1
:
x
0
= 0, from the North and the South poles (1, 0, . . . , 0). The corresponding change of
coordinates is given by (x
1
, . . . , x
n
) (x
1
/[x[
2
, . . . , x
n
/[x[
2
).
The real projective space RP
n
is the set of all lines in R
n+1
passing through 0. Elements
of RP
n
are denoted by x
0
: x
1
: . . . : x
n
, where not all x
i
are zero. Charts can be given by
i
(x
0
: x
1
: . . . : x
n
) = (x
0
/x
i
, . . .
i . . . , x
n
/x
i
) R
n
with changes of coordinates given by
j
1
i
: (y
1
, . . . , y
n
) y
1
: . . . : (1 in ith place) : . . . : y
n
RP
n
_
y
1
y
j
, . . . ,
y
i1
y
j
,
1
y
j
,
y
i
y
j
, . . . ,
y
j1
y
j
,
y
j+1
y
j
. . .
y
n
y
j
_
,
smooth functions on their domains of denition (i.e. for y
j
,= 0). Thus RP
n
is a smooth
n-dimensional manifold.
Denition. Let M, N be smooth manifolds. A continuous map f : M N is called
smooth (C
) if for each p M, for some (hence for every) charts and , of M and
N respectively, with p in the domain of and f(p) in the domain of , the composition
f
1
(which is a map between open sets in R
n
, R
k
, where n = dimM, k = dimN)
is smooth on its domain of denition.
Exercise: write out the domain of denition for f
1
.
Two manifolds M and N are called dieomorphic is there exists a smooth bijective
map M N having smooth inverse. Informally, dieomorphic manifolds can be thought
of as the same.
3
For a more interesting example on a non 2nd countable manifold see Example Sheet Q1.10.
4 alexei kovalev
1.2 Matrix Lie groups
Consider the general linear group GL(n, R) consisting of all the n n real matrices A
satisfying det A ,= 0. The function A det A is continuous and GL(n, R) is the inverse
image of the open set R 0, so it is an open subset in the n
2
-dimensional linear space of
all the n n real matrices. Thus GL(n, R) is a manifold of dimension n
2
. Note that the
result of multiplication or taking the inverse depends smoothly on the matrix entries.
Similarly, GL(n, C) is a manifold of dimension 2n
2
(over R).
Denition. A group G is called a Lie group if it is a smooth manifold and the map
(, ) GG
1
G is smooth.
Let A be an n n complex matrix. The norm given by [A[ = nmax
ij
[a
ij
[ has a useful
property that [AB[ [A[[B[ for any A, B. The exponential map on the matrices is dened
by
exp(A) = I + A + A
2
/2! + . . . + A
n
/n! + . . . .
The series converge absolutely and uniformly on any set [A[ , by the Weierstrass
M-test. It follows that e.g. exp(A
t
) = (exp(A))
t
and exp(C
1
AC) = C
1
exp(A)C,
for any invertible matrix C. Furthermore, the term-by-term dierentiated series also con-
verge uniformly and so exp(A) is C
i
= x
i
(x) the coordinates of this vector are
transformed according to the familiar chain rule, applied to x
i
=
_
x
i
x
j
_
p
a
j
, (1.4)
where x
i
, x
i
are the local coordinates on U, U
respectively.
4
All the tangent vectors at a
given point p form the tangent space denoted T
p
M.
Remarks. It is easy to check that T
p
M is naturally a vector space.
The transformation law (1.4) is the dening property of a tangent vector.
Notation. A choice local coordinates x
i
on a neighbourhood U M denes a linear
isomorphism T
p
M R
n
. A basis of T
p
M corresponding to the standard basis of R
n
via
this isomorphism is a usually denoted by (
x
i
)
p
. The expression of a tangent vector in
local coordinates a(U, ) = (a
1
, . . . , a
n
) then becomes a
i
(
x
i
)
p
.
4
Here and below the convention is used that if the same letter appears as an upper and lower index
then the summation is performed over the range of this index. E.g. the summation in j = 1, . . . , n in this
instance.
6 alexei kovalev
As can be seen from (1.4), the standard basis vectors of T
p
M given by the local coor-
dinates x
i
and x
i
are related by
_
x
i
_
p
=
_
x
j
x
i
_
p
_
x
j
_
p
. (1.4
)
This is what one would expect in view of the chain rule from the calculus. The formula (1.4
)
tells us that every tangent vector a
i
(
x
i
)
p
gives a well-dened rst-order derivation
a
i
(
x
i
)
p
: f C
(M) a
i
f
x
i
(p) R. (1.5)
In fact the following converse statement is true although I shall not prove it here
5
. Given
p M, every linear map a : C
x
j
. Writing X
(i)
= X
(i)
j
(x)
x
j
, we obtain
b
j
= a
i
X
(i)
j
(x), which shows that
1
is smooth. The matrix X(x) = (X
(i)
j
(x)) expresses a
change of basis of T
p
M, from (
x
j
)
p
to X
(j)
(p), and is smooth in x, so the inverse matrix
C(x) is smooth in x too. Therefore a
i
= b
j
C
j
i
(x) veries that is smooth.
Remark . The hypothesis of Theorem 1.9 is rather restrictive. In general, a manifold need
not admit any non-vanishing smooth (or even continuous) vector elds at all (as we shall
see, this is the case for any even-dimensional sphere S
2n
) and the tangent bundle TM will
not be a product M R
n
.
Denition. The dierential of a smooth map F : M N at a point p M is a linear
map
(dF)
p
: T
p
M T
F(p)
N
given in any chart by (dF)
p
: (
x
i
)
p
(
y
j
x
i
)(p) (
y
j
)
F(p)
. Here x
i
are local coordinates
on M, y
j
on N, and for j = 1, . . . , dimN, y
j
= F
j
(x
1
, . . . , x
n
) (n = dimM) give the
expression F
1
for F in these local coordinates.
It follows, by direct calculations in local coordinates, that the dierential is independent
of the choice of charts and that the chain rule d(F
2
F
1
)
p
= (dF
2
)
g(p)
(dF
1
)
p
holds.
Every (smooth) vector eld, say X on M, denes a linear dierential operator of rst
order X : C
(M) C
F(p)
f =
x
i
p
(f F) (x
i
on M and y
j
on N) yields a coordinate-free relation
_
((dF)X)f
_
F = X(f F), (1.10)
Let X, Y be vector elds regarded as dierential operators on C
: g G (dL
g
)
e
T
g
G TG. (1.11)
differential geometry 9
Then X
1
(h), . . . , X
m
(h)
dene m = dimG vector elds whose values at each h G give a basis of T
h
G.
Hence, in view of Theorem 1.9, we obtain
Theorem 1.13. The tangent bundle TG of any Lie group G is isomorphic to the product
GR
dimG
.
The smoothness of X
(h) = X
(gh), i.e.
(dL
g
)X
= X
L
g
. (1.14)
and we shall call any vector eld satisfying (1.14) left-invariant.
It follows from Proposition 1.12 that the vector space l(G) of all the left-invariant vector
elds on G form a nite-dimensional subspace (of dimension m) in the space V (G) of all
vector elds. In fact more is true.
Theorem 1.15. The space of all the left-invariant vector elds on G is a nite-dimensional
Lie algebra, hence a Lie subalgebra of V (G).
Proof. The theorem may be restated as saying that for every pair X
, X
of left-invariant
vector elds, [X
, X
, X
]f L
g
= [X
, X
](f L
g
) = X
(f L
g
) X
(f L
g
)
= X
_
(dL
g
)X
f L
g
_
X
_
(dL
g
)X
f L
g
_
= (dL
g
)X
((dL
g
)X
f) L
g
(dL
g
)X
((dL
g
)X
f) L
g
= [(dL
g
)X
, (dL
g
)X
]f L
g
and using the left-invariant property (1.14) of X
and X
L
g
, X
L
g
]f L
g
= ([X
, X
] L
g
)f L
g
.
Thus (dL
g
)[X
, X
]f L
g
= ([X
, X
] L
g
)f L
g
, for each g G, f C
, X
, X
] = X
l(G) is an
isomorphism of the Lie algebras, where the Lie bracket on g is as dened in Theorem 1.7.
We shall prove Theorem 1.16 in the next section.
10 alexei kovalev
1.4 Submanifolds
Suppose that M is a manifold, N M, and N is itself a manifold, denote by : N M
the inclusion map.
Denition. N is said to be an embedded submanifold
8
of M if
(i) the map is smooth;
(ii) the dierential (d)
p
at any point of p N is an injective linear map;
(iii) is a homeomorphism onto its image, i.e. a D N is open in the topology of
manifold N if and only if D is open in the topology induced on N from M (i.e. the open
subsets in N are precisely the intersections with N of the open subsets in M).
Remark. Often the manifold N is not given as a subset of M but can be identied with a
subset of M by means of an injective map : N M. In this situation, the conditions in
the above denition make sense for (N) (regarded as a manifold dieomorphic to N). If
these conditions hold for (N) then one says that the map embeds N in M and writes
: N M.
Example. A basic example of embedded submanifold is a (parameterized) curve or surface
in R
3
. Then the condition (i) means that the parameterization is smooth and (ii) means
that the parameterization is regular (recall introductory remarks on p.1). The condition
(iii) eliminates e.g. the irrational twist ow t R (t, t) R
2
/Z
2
= S
1
S
1
, RQ,
on the torus.
A surface or curve in R
3
(or more generally in R
n
) is often dened by an equation or
a system of equations, i.e. as the zero locus of a smooth map on R
3
(respectively on R
n
).
E.g. x
2
+y
2
1 = 0 (a circle) or (x
2
+y
2
+b
2
+z
2
a
2
)
2
4b
2
(x
2
+y
2
) = 0, (a torus with
radii b > a > 0). However a smooth (even polynomial) map may in general have bad
points in its zero locus (Example sheet 1, Q.6, 8 and 9). When does a system of equations
on a manifold dene a submanifold?
Denition. A value q N of a smooth map f between manifolds M and N is called a
regular value if for any p M such that f(p) = q the dierential of f at p is surjective,
(df)
p
(T
p
M) = T
q
N.
Theorem 1.17. Let f : M N be a smooth map between manifolds and q N a regular
value of f. The inverse image of a regular value P = f
1
(q) = p M : f(p) = q
(if it is non-empty) is an embedded submanifold of M, of dimension dimM dimN.
We shall need the following result from advanced calculus.
Inverse Mapping Theorem. Suppose that f : U R
n
R
n
is a smooth map dened
on an open set U, 0 U and f(0) = 0. Then f has an smooth inverse g, dened on some
neighbourhood of 0 with g(0) = 0, if and only if (df)
0
is an invertible linear map of R
n
.
Note that the Inverse Mapping Theorem, as stated above, is a local result, valid only if
one restricts attention to a suitably chosen neighbourhood of a point. The statement will
in general no longer hold with neighbourhoods replaced by manifolds. (Consider e.g. the
map of R to the unit circle S
1
C given by f(x) = e
ix
.)
8
In these notes I will sometimes write submanifold meaning embedded submanifold.
differential geometry 11
Proof of Theorem 1.17. Firstly, P is Hausdor and second countable because M is so.
Let p be an arbitrary point of P. We may assume without loss of generality that
there are local coordinates x
i
, y
j
, i = 1, . . . , n, j = 1, . . . , k = dimN, n + k = dimM,
dened in a neighbourhood of p in M and local coordinates dened in a neighbourhood
of f(p) in N such that x
i
(p) = y
j
(p) = 0 and f is expressed in these local coordinates
as a map f = (f
1
, . . . , f
k
) on a neighbourhood of 0 in R
n
with values in R
k
, f(0) = 0,
det(f
i
/y
j
)(0, 0) ,= 0.
Then
det
_
1 0
(f
i
/x
i
) (f
i
/y
j
)
_
(0, 0) ,= 0
and so, by the Inverse Function Theorem, the x
i
, f
j
form a valid set of new local coordinates
on a (perhaps smaller) neighbourhood of p in M. The local equation for the intersection of
P with that neighbourhood takes in the new coordinates a simple form f
j
= 0, j = 1, . . . , k.
Furthermore the rst projection (x
i
, f
j
) (x
i
) restricts to a homeomorphism from a
neighbourhood of p in P onto a neighbourhood of zero in R
n
. Dene this rst projection
to be a coordinate chart on P with x
i
the local coordinates. Then the family of all such
charts covers P and it remains to verify that any two charts dened in this way are in fact
compatible.
So let x
i
, f
j
, x
i
, f
j
be two sets of local coordinates near p as above. Then, by the
construction, for every p in P, we have that x
i
= x
i
(x, f), f
j
= f
j
(x, f) with f
j
(x, 0) = 0
identically in x. Therefore, (f
i
/x
j
)(0, 0) = 0. Then
det
_
x
i
/x
i
x
i
/f
j
f
j
/x
i
f
j
/f
j
_
(0, 0) = det
_
x
i
/x
i
x
i
/f
j
0 f
j
/f
j
_
(0, 0) ,= 0,
so we must have det(x
i
/x
j
)(0) ,= 0 for the n n Jacobian matrix, and the change from
x
i
s to x
i
s is a dieomorphism near 0.
Remarks. 1. It is not true that every submanifold of M is obtainable as the inverse image
of a regular value for some smooth map on M. One counterexample is RP
1
RP
2
(check it!).
2. Sometimes in the literature one encounters a statement a subset P is (or is not) a
submanifold of M. Every subset P of a manifold M has a topology induced from M. It
turns out that there is at most one smooth structure on the topological space P such that
P is an embedded submanifold of M, but I shall not prove it here.
Theorem 1.18 (Whitney embedding theorem). Every smooth n-dimensional manifold can
be embedded in R
2n
(i.e. is dieomorphic to a submanifold of R
2n
).
We shall assume the Whitney embedding theorem without proof here (a proof of em-
bedding in R
2n+1
is found in e.g. Guillemin and Pollack, Ch.1 9).
It is worth to remark that the possibility of embedding any manifold in some R
N
, with
N possibly very large, does not particularly simplify the study of manifolds in practical
terms (but is relatively easier to prove). The essence of the Whitney embedding theorem is
12 alexei kovalev
the minimum possible dimension of the ambient Euclidean space as way of measuring the
topological complexity of the manifold. The result is sharp, in that the dimension of the
ambient Euclidean space could not in general be lowered (as can be checked by considering
e.g. the Klein bottle).
We can now give, as promised, a proof of Theorem 1.16
Theorem 1.16. Suppose that G GL(n, R) is a subgroup and an embedded submani-
fold of GL(n, R), and smooth structure on G is dened by the log-charts. Then the map
g X
, X
] = X
[,]
for a matrix Lie group where
the LHS is the Lie bracket of left-invariant vector elds and the RHS is dened using
Theorem 1.7. Note rst that for G = GL(nR), all the calculations can be done on an open
subset of R
n
2
= Matr(n, R) with coordinates x
i
j
, i.j = 1. . . . , n. The map L
g
, and hence
also dL
g
, is the usual left multiplication by a xed matrix g = (x
i
j
). Respectively, the
left-invariant vector elds are X
(g) = x
i
k
k
j
x
i
j
and [X
, X
] = X
[,]
is an easy calculation.
Thus the theorem holds for GL(n, R).
Now consider a general case G GL(n, R). Denote, as before, the inclusion map by .
Any left-invariant vector eld X
, X
l(GL(n, R))
with , g, where g is understood as the image of log-chart for G near I. We have
[X
, X
] = X
[,]
, by the above calculations on GL(n, R), and the Lie bracket of matrices
[, ] g by Theorem 1.7. Therefore, X
[,]
restricts to give a well-dened left-invariant
vector eld on G, and [X
, X
] = X
[,]
holds for any X
, X
l(G) as claimed.
Remark. Notice that the dierential d considered in the above proof identies g with a
linear subspace of n n matrices, by considering G as a hypersurface in GL(n, R) R
n
2
.
In the example of G = O(n), for any path A(t) or orthogonal matrices with A(0) = I,
we calculate 0 =
d
dt
t=0
A(t)A
(t) =
A(0)A
(0) + A(0)
A
(0) =
A(0) +
A
(0), i.e.
A(0) is
skew-symmetric. Thus the log chart at the identity actually maps onto a neighbourhood
of zero in the tangent space T
I
G which explains why the log-chart construction in 1.2
worked.
1.5 Exterior algebra of dierential forms. De Rham cohomology
See the reference card on multilinear algebra
The dierential forms
Consider a smooth manifold M of dimension n. The dual space to the tangent space T
p
M,
p M, is called the cotangent space to M at p, denoted T
p
M. Suppose that a chart and
hence the local coordinates are given on a neighbourhood of p. The dual basis to (
x
i
)
p
is
differential geometry 13
traditionally denoted by (dx
i
)
p
, i = 1, . . . , n. (Sometimes I may drop the subscript p from
the notation.) Thus an arbitrary element of T
p
M is expressed as
i
a
i
(dx
i
)
p
for some
a
i
R.
Recall from (1.4
i
, induces
a change of basis of tangent space T
p
M from
x
i
to
x
i
. By linear algebra, there is a
corresponding change of the dual basis of T
p
M, from dx
i
to dx
i
given by the transposed
matrix. Thus the transformation law is
dx
j
=
x
j
x
i
dx
i
. (1.19)
Like (1.4
) the eq.(1.19) is a priori merely a notation which resembles familiar results from
the calculus. See however further justication in the remark on the next page.
A disjoint union of all the cotangent spaces T
M = .
pM
T
p
M of a given manifold
M is called the cotangent bundle of M. The cotangent bundle can be given a smooth
structure making it into a manifold of dimension 2 dimM by an argument very similar
to one for the tangent bundle (but with a change of notation, replacing any occurrence
of (1.4
) with (1.19)).
A smooth eld of linear functionals is called a (smooth) dierential 1-form (or just
1-form). More precisely, a dierential 1-form is a map : M T
M such that
p
T
p
M
near every p M and is expressed in any local coordinates x = (x
1
, . . . , x
n
) by =
i
a
i
(x)dx
i
where a
i
(x) are some smooth functions of x.
Remark . The 1-forms are of course the dual objects to the vector elds. In particular,
a
i
(x(p)) is obtained as the value of
p
on the tangent vector (
x
i
)
p
. Consequently, is
smooth on M if and only if (X) is a smooth function for every vector eld X on M.
Notice that the latter condition does not use local coordinates.
One can similarly consider a space
r
T
p
M of alternating multilinear functions on T
p
M
. . . T
p
M(r factors), for any r = 0, 1, 2, . . . n, and proceed to dene the r-th exterior power
r
T
M = M R and
0
(M) = C
M is isomorphic to M R.
Proof (gist). That (a)(c) is proved similarly to the proof of Theorem 1.9.
14 alexei kovalev
Using linear algebra, we nd that the transformation of the dierential forms of top
degree under a change of coordinates is given by
dx
1
. . . dx
n
= det
_
x
j
x
i
_
dx
1
. . . dx
n
.
Now (a)(b) is easy to see.
To obtain, (b)(a) we assume the following.
Theorem 1.21 (Partition of unity). For any open cover M
A
U
, there exists a
countable collection of functions
i
C
for some =
i
(i.e. depending on i);
(ii) the collection is locally nite: each x M has a neighbourhood W
x
such that
i
(x) ,= 0
on W
x
for only nitely many i; and
(iii)
i
0 on M for all i and
i
i
(x) = 1 for all x M.
The collection
i
satisfying the above is called a partition of unity subordinate to U
.
Choose a partition of unity
i
subordinate to the given family of coordinate neigh-
bourhoods covering M. For each i, choose local coordinates x
()
i
valid on the support of
i
.
Dene
i
= dx
()
1
. . .dx
()
n
in these local coordinates, then
is a well-dened (smooth)
n-form on all of M (extended by zero outside the coordinate domain) and =
p
M. In local coordinates
x
i
dened near p we have df(x) =
f
x
i
(x)dx
i
, thus df is a well-dened dierential 1-form,
whose coecients are those of the gradient of f.
Remark . Observe that any local coordinate x
i
on an open domain U M is a smooth
function on U. Then the formal symbols dx
i
actually make sense as the dierentials of
these smooth functions (which justies the previously introduced notation, cf.(1.19)).
Theorem 1.22 (exterior dierentiation). There exists unique linear operator
d :
k
(M)
k+1
(M), k 0, such that
(i) if f
0
(M) then df coincides with the dierential of a smooth function f;
(ii) d( ) = d + (1)
deg
d for any two dierential forms ,;
(iii) dd = 0 for every dierential form .
differential geometry 15
Proof (gist). On an open set U R
n
, or in the local coordinates on a coordinate domain
on a manifold, application of conditions (ii), then (iii) and (i), yields
d(fdx
i
1
. . . dx
i
r
) = df dx
i
1
. . . dx
i
r
=
f
x
i
dx
i
dx
i
1
. . . dx
i
r
, (1.23)
for any smooth function f. Extend this to arbitrary dierential forms by linearity. The
conditions (i),(ii),(iii) then follow by direct calculation, in particular the last of these holds
by independence of the order of dierentiation in second partial derivatives. This proves
the uniqueness, i.e. that if d exists then it must be expressed by (1.23) in local coordinates.
Observe another important consequence of (1.23): the operator d is necessarily local,
which means that the value (d)
p
at a point p is determined by the values of dierential
form on a neighbourhood of p.
To establish the existence of d one now needs to show that the dening formula (1.23)
is consistent, i.e. the result of calculation does not depend on the system of local coordi-
nates in which it is performed. So let d
(fdx
i
1
. . . dx
i
r
) = d
f dx
i
1
. . . dx
i
r
+
r
j=1
(1)
j
fdx
i
1
. . . d
(dx
i
j
) . . . dx
i
r
.
But d
f = df and d
(dx
k
) = d
(d
x
k
) = 0, by (i) and (iii) and because we know that
the dierential of a smooth function (0-form) is independent of the coordinates. Hence the
right-hand side of the above equality becomes
f
x
j
dx
j
dx
i
1
. . .dx
i
r
= d(fx
i
1
. . .dx
i
r
),
and so the exterior dierentiation is well-dened.
De Rham cohomology
A dierential form is said to be closed when d = 0 and exact when = d for some
dierential form . Thus exact forms are necessarily closed (but the converse is not in
general true, e.g. example sheet 2, Q4).
Denition. The quotient space
H
k
dR
(M) =
closed k-forms on M
exact k-forms on M
is called the k-th de Rham cohomology group of the manifold M.
Any smooth map between manifolds, say f : M N, induces a pull-back map
between the cotangent spaces f
: T
f(p)
N T
p
M, which is a linear map dened, for any
dierential form on N, by
(f
)
p
(v
1
, . . . , v
r
) =
f(p)
((df)
p
v
1
, . . . , (df)
p
v
r
),
16 alexei kovalev
using the dierential of f. The chain rule for dierentials of smooth maps immediately
gives
(f g)
= g
. (1.24)
It is also straightforward to check that f
() = (f
)(f
)
and f
(d) = d(f
), hence f
preserves the
subspaces of closed and exact dierential forms. Therefore, every smooth map f : M N
induces a linear map on the de Rham cohomology
f
: H
r
(N) H
r
(M)
A consequence of the chain rule (1.24) is that if f is a dieomorphism then f
is a linear
isomorphism. Thus the de Rham cohomology is a dieomorphism invariant, i.e. dieo-
morphic manifolds have isomorphic de Rham cohomology.
9
Poincare lemma. H
k
(D) = 0 for any k > 0, where D denotes the open unit ball in R
n
.
The proof goes by working out a way to invert the exterior derivative. More precisely,
one constructs linear maps h
k
:
k
(U)
k1
(U) such that
h
k+1
d + d h
k
= id
k
(U)
.
Remark. In the degree 0, one has H
0
(M) = R for any connected manifold.
Basic integration on manifolds
Throughout this subsection M is an oriented n-dimensional manifold. Let
n
(M)
and, as before, denote supp = p M :
p
,= 0 (the support of ). Suppose that the
closure of supp is compact.
Consider rst the special case when the closure of supp is contained in the domain
of just one coordinate chart, (U, ) say. If f(x)dx
1
. . . dx
n
is the local expression
for then the integral
_
(U
)
f(x)dx
1
. . . dx
n
makes sense as in the multivariate calculus.
The value of this integral is independent of the choice of local coordinates, provided only
that the change is orientation-preserving, i.e. the Jacobian is positive. This is because the
local expression for changes precisely as required by the change of variables formula for
integrals of functions of n variables (which involves the absolute value of the Jacobian).
Thus
_
M
=
_
U
is well dened when supported in just one coordinate chart.
Now let be any n-form with compact support. Consider an oriented system of charts
(U
.
Denition. In the above situation, the integral of over M is given by
_
M
=
_
U
.
9
In fact, more is true. It can be shown, using topology, that the de Rham cohomology depends only on
the topological space underlying a smooth manifold and that homeomorphic manifolds have isomorphic de
Rham cohomology. The converse in not true: there are manifolds with isomorphic de Rham cohomology
but e.g. with dierent fundamental groups.
differential geometry 17
Note that the sum in the right-hand side is nite. It can be checked that
_
M
does
not depend on the choice of a partition of unity on M, and is therefore well-dened.
Stokes Theorem (for manifolds without boundary)
10
. Suppose that
n1
(M)
has a compact support. Then
_
M
d = 0.
Let
x
0
and straightforward calculus shows that the norm of the linear map dened by the Jacobi
matrix
_
d([x[X(x/[x[))
x
0
is bounded independent of x
0
,= 0. Hence there is
0
> 0, such
that (df)
x
0
is a linear isomorphism of R
n+1
onto itself for any x
0
,= 0 and any [[ <
0
. But
then, by the Inverse Mapping Theorem (page 10), for any x ,= 0, f maps some open ball
B(x,
x
) of radius
x
> 0 about x dieomorphically onto its image.
Furthermore, it can be checked, by inspection of the proof of the Inverse Mapping The-
orem, that (1)
x
can be taken to be continuous in x and (2)
x
can be chosen independent
of if [[ <
0
. We shall assume these two latter claims without proof. Consequently,
x
can be taken to depend only on [x[ (as S
n
([x[) is compact).
10
Manifolds with boundary are not considered in these lectures.
18 alexei kovalev
Taking a smaller
0
> 0 if necessary, we ensure that f is one-to-one if [[ <
0
. For
we have [f(x)[ =
1 +
2
[x[ and so it suces to check that that f is one-to-one on each
S
n
([x[). But two points on S
n
([x[) far away from each other cannot be mapped to one
because [f(x) x[ < [x[ and two distinct points at a distance less than say
1
2
|x|
cannot
be mapped to one because f restricts to a dieomorphism (hence a bijection) on a
|x|
-ball
about each point.
A similar reasoning shows that f is surjective (onto) if in suciently small. Indeed,
f(B(x,
x
)) is an open set homeomorphic to a ball and the boundary of f(B(x,
x
)) is
within small distance (1 +
x
)[x[ from the boundary of B(x,
x
). Therefore, x must be
inside the boundary of f(B(x,
x
)) and thus in the image of f. In all of the above, small
0
can be chosen independent of x because
x
depends only on [x[ and f is homogeneous
of degree 1, f(x) = f(x) for each positive .
p.t.o.
differential geometry 19
Step 2. Now, as f is a dieomorphism f maps the embedded submanifold S
n
(1)
dieomorphically onto the embedded submanifold f(S
n
(1)) = S
n
(
1 +
2
). Consider a
dierential n-form on R
n+1
=
n
i=0
(1)
i
x
i
dx
0
. . . (omit dx
i
) . . . dx
n
,
We have
_
f(S
n
(1))
=
_
S
n
(1)
f
_
S
n
=
_
r
1
d
ds
__
S
n
(s)
_
ds
=
_
r
1
__
S
n
d
ds
_
(s)
_
_
ds
applying, on each coordinate patch, a theorem on dierentiation of an integral depending
on a parameter s, from calculus
=
_
1|x|r
d
replacing, again on each coordinate patch, a repeated integration with an (n+1)-dimensional
integral
=
_
1|x|r
(n + 1)dx
0
. . . dx
n
= c
n+1
(r
n+1
1),
where c
n+1
is n + 1 times the volume of (n + 1)-dimensional ball (the value of c
n+1
does
not matter here). Put r =
1 +
2
and then the right-hand side is not a polynomial in
if n + 1 is odd (i.e. when n is even). A contradiction.
Some page references
20 alexei kovalev
to Warner, and GuilleminPollack, and GallotHulinLafontaine.
N.B. The material in these books is sometimes covered dierently from the Lectures and
may contain additional topics, thus the references are not quite one-to-one.
smooth manifolds [W] 1.21.6
tangent and cotangent bundles [W] 1.25
exponential map on a matrix Lie group [W] 3.35
left invariant vector elds [W] 3.63.7
submanifolds [W] 1.271.31, 1.38
dierential forms [GP] 153165,174178
de Rham cohomology [GP] pp.178182
Poincare Lemma [W] 4.18
partition of unity [GP] p.52 or [W] 1.81.11
integration and Stokes Theorem [GP] 165168, 183185
non-existence of vector elds without zeros on S
2n
(Milnors proof) [GHL] 1.41
Part III: Dierential geometry (Michaelmas 2004)
Alexei Kovalev (A.G.Kovalev@dpmms.cam.ac.uk)
2 Vector bundles.
Denition. Let B be a smooth manifold. A manifold E together with a smooth submer-
sion
1
: E B, onto B, is called a vector bundle of rank k over B if the following
holds:
(i) there is a k-dimensional vector space V , called typical bre of E, such that for any
point p B the bre E
p
=
1
(p) of over p is a vector space isomorphic to V ;
(ii) any point p B has a neighbourhood U, such that there is a dieomorphism
1
(U)
U
U V
pr
1
U U
and the diagram commutes, which means that every bre E
p
is mapped to p V .
U
is called a local trivialization of E over U and U is a trivializing neighbour-
hood for E.
(iii)
U
[
E
p
: E
p
V is an isomorphism of vector spaces.
Some more terminology: B is called the base and E the total space of this vector bundle.
: E B is said to be a real or complex vector bundle corresponding to the typical bre
being a real or complex vector space. Of course, the linear isomorphisms etc. are understood
to be over the respective eld R or C. In what follows vector bundles are taken to be real
vector bundles unless stated otherwise.
Denition. Any smooth map s : B E such that s = id
B
is called a section of E.
If s is only dened over a neighbourhood in B it is called a local section.
Examples. 0. A trivial, or product, bundle E = B V with the rst projection.
Sections of this bundle are just the smooth maps C
(B; V ).
1. The tangent bundle TM of a smooth manifold M has already been discussed in
Chapter 1. It is a real vector bundle of rank n = dimM which in general is not trivial.
2
The sections of TM are the vector elds. In a similar way, the cotangent bundle T
M and,
1
A smooth map is called a submersion if its dierential is surjective at each point.
2
Theorems 1.9 and 1.26 in Chapter 1 imply that TS
2n
cannot be trivial.
20
differential geometry 21
more generally, the bundle of dierential p-forms
p
T
n
p
with sections being the dierential 1-forms, respectively p-forms. Exercise: verify that
the vector bundles
p
T
, U
a composite map
(b, v) = (b,
(b)v),
(b, v) (U
) R
k
. For every xed b the above composition is a linear isomorphism
of R
k
depending smoothly on b. The maps
: U
= id
R
k,
= id
R
k,
= id
R
k .
(2.1)
The left-hand side is dened on the intersection U
covering the base and such that all the corresponding transition
functions
, for
all , , where k is the rank of E. Then this latter system (U
) of local trivializations
over U
1
take values in the orthogonal group (respectively, unitary group) then
is called an orthogonal trivialization (resp. unitary trivialization).
Principal bundles.
Let G be a Lie group. A smooth free right action of G on a manifold P is a smooth
map P G P, (p, h) ph, such that (1) for any p P, ph = p if and only if h is the
identity element of G; and (2) (ph
1
)h
2
= p(h
1
h
2
) for any p P, any h
1
, h
2
G. (It follows
that for each h G, P h P is a dieomorphism.)
Denition . A (smooth) principal G-bundle P over B is a smooth submersion
: P B onto a manifold B, together with a smooth right free action P G P, such
that the set of orbits of G in P is identied with B (as a set), P/G = B, and also for any
b B there exists a neighbourhood U B of b and a dieomorphism
U
:
1
(U) UG
such that pr
1
U
= [
1
(U)
, i.e. the following diagram is commutative
1
(U)
U
U G
pr
1
U U
(2.2)
and
U
commutes with the action of G, i.e. for each h G,
U
(ph) = (b, gh), where
(b, g) =
U
(p), (p) = b U.
A local section of the principal bundle P is a smooth map s : U P dened on a
neighbourhood U B and such that s = id
U
.
differential geometry 23
For a pair of overlapping trivializing neighbourhoods U
, U
one has
(b, g) = (b,
(b, g)),
where for each b U
, the
, we must have
(b, g)h =
(b, 1
G
) G. It is sensible to slightly simplify the notation and write g
(b)g
for this left multiplication. We nd that, just like vector bundles, the principal G-bundles
have transition functions
: U
, corresponding
to an open cover of B and satisfying (2.1), via the following Steenrod construction.
For each trivializing neighbourhood U
B for E consider U
G. Dene an equivalence
relation between elements (b, h) U
G, (b
, h
) U
G, so that (b, h) (b
, h
) precisely
if b
= b and h
/ (2.3)
the disjoint union of all U
1
: (w, wz)
1
(U
1
) (1 : z, w
1 +[z[
2
) U
1
C,
2
: (w, w)
1
(U
2
) ( : 1, w
1 +[[
2
) U
2
C.
Calculating the inverse, we nd
1
1
(1 : z, w) =
1 +[z[
2
,
wz
1 +[z[
2
and so
2
1
1
(1 : z, w) =
2
1 +[z[
2
,
wz
1 +[z[
2
=
2
[[w
[[
2
+ 1
,
[[w
[[
2
+ 1
: 1,
[[
1 : z,
z
[z[
w
2,1
takes values in the unitary group U(1) = S
1
= z C : [z[ = 1, a subgroup of
GL(1, C) = C 0 (it is for this reason the square root factor was useful in the local
trivialization). Theorem 2.4 now ensures that Hopf bundle E is a well-dened vector
bundle, moreover a vector bundle with a U(1)-structure. Hence there is a invariantly
dened notion of length of any vector in any bre of E. The length of (wz
1
, wz
2
) may
be calculated in the local trivializations
1
or
2
by taking the mudulus of the second
component of
i
(in C). For each i = 1, 2, this coincides with the familiar Hermitian
length
[wz
1
[
2
+[wz
2
[
2
of (wz
1
, wz
2
) in C
2
.
We can now use
2,1
to construct the principal S
1
-bundle (i.e. U(1)-bundle) P CP
1
associated to Hopf vector bundle E, cf. Theorem 2.4. If U(1) is identied with a unit
circle S
1
C , any bre of P may be considered as the unit circle in the respective
bre of E. Thus P is identied as the space of all vectors in E of length 1, so P =
(w
1
, w
2
) C
2
[ w
1
w
1
+w
2
w
2
=1 is the 3-dimensional sphere and the bundle projection is
: (w
1
, w
2
) S
3
w
1
: w
2
S
2
,
1
(p)
= S
1
,
where we used the dieomorphism S
2
= CP
1
for the target space. (Examples 1, Q3(ii).)
This principal S
1
-bundle S
3
over S
2
is also called Hopf bundle. It is certainly not trivial,
as S
3
is not dieomorphic to S
2
S
1
. (The latter claim is not dicult to verify, e.g.
by showing that the de Rham cohomology H
1
(S
3
) is trivial, whereas H
1
(S
2
S
1
) is not.
Cf. Examples 2 Q5.)
differential geometry 25
Pulling back vector bundles and principal bundles
Let P be a principal bundle over a base manifold B and E an associated vector bundle
over B. Consider a smooth map f : M B.
The pull-back of a vector (respectively, principal) bundle is a bundle f
E (f
P)
over M such that there is a commutative diagram (vertical arrows are the bundle projec-
tions)
f
E
F
E
M
f
B,
(2.5)
such that the restriction of F to each bre (f
E)
p
over p M is an isomorphism onto a
bre E
f(p)
.
A very basic special case of the above is when f maps M to a point in B; then the pull-
back f
E (and f
E
(resp. f
E
= E X,
with the projection (e, x) E X ((e), x) B X.
The construction may be extended to a general vector bundle, by working in local
trivialization. Then one has to ensure that the pull-back must be well-dened independent
of the choice of local trivialization. To this end, let
be a system of transition
functions for E. Dene
f
f
and f
E and f
P as required by (2.5).
26 alexei kovalev
2.1 Bundle morphisms and automorphisms.
Let (E, B, ) and (E
, B
a smooth map.
Denition. A smooth map F : E E
f(p)
for any p B,
so that
E
F
E
B
f
B
is a commutative diagram,
F = f .
More explicitly, suppose that the local trivializations :
1
(U) U V ,
1
(U
) U
F[
1
(U)
1
is expressed as
F
U
: (b, v) (f(b), h(b)v), (2.6)
for some smooth h : U L(V, V
, E
is a bundle morphism E E
.
Examples. 1. If : M N is a smooth map between manifolds M,N then its dierential
d : TM TN is a morphism of tangent bundles.
2. Recall the pull-back of a given vector bundle (E, B, ) via a smooth map f : M B.
For every local trivialization E[
U
of E, the corresponding local trivialization of the pull-
back bundle is given by f
E[
f
1
(U)
(f
1
(U))V , where V is the typical bre of E (hence
also of f
E E which
restricts to a linear isomorphism between any pair of bres (f
E)
p
and E
f(p)
, p M. (This
isomorphism becomes just the identity map of the typical bre V in the indicated local
trivializations.) It follows that F is a bundle morphism covering the given map f : M B.
3. Important special case of bundle morphisms occurs when f is a dieomorphism of B
onto B
. A morphism F : E E
(B, G) C
over U
, is replaced by
. We have
(e) = h
((e))
= h
h
1
.
Now let
1
p
, . . . ,
m
p
be linearly independent covectors in T
p
E, p
1
(U), and dene
S
p
:= v T
p
E[
i
p
(v) = 0, i = 1, . . . , m.
We can write, using local coordinates on U,
i
p
= f
i
k
dx
k
+ g
i
j
da
j
, i = 1, . . . , m, f
i
k
, g
i
j
R. (2.7)
and any tangent vector in T
p
E as v = B
k
(
x
k
)
p
+ C
i
(
a
i
)
p
, B
k
, C
i
R. The
j
p
cannot all
vanish on a vertical vector, i.e. on a vector having B
k
= 0 for all k. That is,
if g
i
j
C
j
= 0 for all i = 1, . . . , m then C
i
= 0 for all i = 1, . . . , m.
differential geometry 29
Therefore the m m matrix g = (g
i
j
) must be invertible. Denote the inverse matrix by
c = g
1
, c = (c
i
l
). Replace
i
by
i
= c
i
l
l
= da
i
+e
i
k
dx
k
, this does not change the space S
p
.
The above arrangement can be made for every p
1
(U), with f
i
k
(p) and g
i
j
(p)
in (2.7) becoming functions of p. Call a map p S
p
a eld of horizontal subspaces if
the functions f
i
k
(p) and g
i
j
(p) are smooth. To summarize,
Proposition 2.8. Let S = S
p
, p E, be an arbitrary smooth eld of horizontal subspaces
in TE. Let x
k
, a
j
be local coordinates on
1
(U) arising, as above, from some local trivial-
ization of E over a coordinate neighbourhood U. Then S
p
is expressed as S
p
=
m
j=1
Ker
j
p
,
where
j
= da
j
+ e
j
k
(x, a)dx
k
, (2.9)
for some smooth functions e
j
k
(x, a). These e
j
k
(x, a) are uniquely determined by a local
trivialization.
Denition. A eld of horizontal subspaces S
p
T
p
E is called a connection on E if in
every local trivialization it can be written as S
p
= Ker (
1
p
, . . . ,
m
p
) as in (2.9), such that
the functions e
i
k
(p) = e
i
k
(x, a) are linear in the bre variables,
e
i
k
(x, a) =
i
jk
(x)a
j
. (2.10a)
and so
i
p
= da
i
+
i
jk
(x)a
j
dx
k
, (2.10b)
where
i
jk
: U B R are smooth functions called the coecients of connection S
p
in a given local trivialization.
As we shall see below, the linearity condition in a
i
ensures that the horizontal sections
(which are to become the analogues of constant vector-functions) form a linear subspace of
the vector space of all sections of E. (A very reasonable thing to ask for.)
I will sometimes use an abbreviated notation
i
p
= da
i
+ A
i
j
a
j
, where A
i
j
=
i
jk
dx
k
,
So Proposition 2.8 identies a connection with a system of matrices A = (A
i
j
) of dierential
1-forms, assigned to trivializing neighbourhoods U B.
The transformation law for connections.
Now consider another coordinate patch U
X, U
= (x
k
:
1
(U
) U
R
m
with x
k
, a
i
the coordinates on U
R
m
.
Notation: throughout this subsection, the apostrophe
will refer to the local trivialization
of E over U
are written as
i
i
and from U
to U as
i
i
, thus the matrix (
i
i
) is inverse to (
i
i
).
Likewise (x
k
/x
k
/x
k
).
Recall that on U
U we have
x
k
= x
k
(x), a
i
=
i
i
(x)a
i
. (2.11)
Then
dx
k
=
x
k
x
k
dx
k
, da
i
= d
i
i
a
i
+
i
i
da
i
=
i
i
x
k
a
i
dx
k
+
i
i
da
i
,
so
= da
i
+
i
k
a
j
dx
k
= d
i
i
a
i
+
i
i
da
i
+
x
k
x
k
i
k
a
j
dx
k
,
and
i
i
i
= da
i
+ (
i
i
j
x
k
+
i
i
x
k
x
k
i
k
j
j
)a
j
dx
k
.
But then
i
i
i
=
i
and we nd, by comparing with (2.10b), that
i
jk
=
i
k
i
i
j
j
x
k
x
k
+
i
i
j
x
k
(2.12a)
and, using A
i
j
=
i
jk
dx
k
,
A
i
j
=
i
i
A
i
j
j
j
+
i
i
d
i
j
, (2.12b)
Writing A
and A
= A
1
+ d
1
= A
1
(d)
1
. (2.12c)
The above calculations prove.
Theorem 2.13. Any system of functions
i
jk
, i, j = 1, . . . , m, k = 1, . . . , n attached to the
local trivializations and satisfying the transformation law (2.12) denes on E a connection
A, whose coecients are
i
jk
.
Remark. Suppose that we x a local trivialization and a connection A on E and regard
as a bundle automorphism of E, Aut E. With his shift of view, the formula (2.12c)
expresses the action of the group Aut E on the space of connections on E. (Cf. the remark
on bundle automorphisms and linear algebra, page 25.)
Before considering the third view on connections we need a rigorous and systematic
way to consider vectors and matrices of dierential forms.
differential geometry 31
The endomorphism bundle End E. Dierential forms with values in vector
bundles.
Let (E, B, ) be a vector bundle with typical bre V and transition functions
. If G
is a linear map V V , or endomorphism of V , G End V , in a trivialization labelled
by , then the same endomorphism in trivialization will be given by
(recall
that
=
1
).
This may be understood in the sense that the structure group of V acts linearly on
End V . Exploiting, once again, the idea of Theorem 2.4 and the accompanying remarks
one can construct from E a new vector bundle End E, with the same structure group as
for E and with a typical bre End V . This is called the endomorphism bundle of a
vector bundle E and denoted End E.
One can further extend the above construction and dene over B the vector bundle
whose bres are linear maps T
b
B E
b
, b B or, more generally, the antisymmetric
multilinear maps T
b
B . . . T
b
B E
b
on r-tuples of tangent vectors. (So the typical
bre of the corresponding bundle is the tensor product
r
(R
n
)
1
B
(End E).
Covariant derivatives
Denition . A covariant derivative on a vector bundle E is a R-linear operator
E
: (E) (T
E
(fs) = df s + f
E
s (2.14)
for any s (E) and function f C
(B).
32 alexei kovalev
Here I used () to denote the space of sections of a vector bundle. Thus (E) =
0
B
(E)
and (T
B E) =
1
B
(E).
Example. Consider a connection A and put
E
= d
A
dened in a local trivialization by
d
A
s = ds + As, s (E).
More explicitly, one can write s = (s
1
, . . . , s
m
) with the help of a local trivialization, where
s
j
are smooth functions on the trivializing neighbourhood, and then
d
A
(s
1
, . . . , s
m
) =
(
s
1
x
k
+
1
jk
s
j
)dx
k
, . . . , (
s
m
x
k
+
m
jk
s
j
)dx
k
.
The operator d
A
is well-dened as making a transition to another trivialization we have
s = s
and A = A
1
(d)
1
, which yields the correct transformation law for
d
A
s = ds + As = d(s
) + (A
1
(d)
1
)s
= (ds
+ A
) = (d
A
s)
.
Theorem 2.15. Any covariant derivative
E
arises as d
A
from some connection A.
Proof (gist). Firstly, any covariant derivative
E
is a local operation, which means that
is s
1
, s
2
are two sections which are equal over an open neighbourhood U of b B then
(
E
s
1
)[
b
= (
E
s
2
)[
b
. Indeed, let U
0
be a smaller neighbourhood of b with the closure
U
0
U and consider a cut-o function C
(B), so that 0 1, [
U
0
= 1,
[
B\U
= 0. Then 0 = d((s
1
s
2
)) = (s
1
s
2
) d+
E
(s
1
s
2
), whence s
1
(b) = s
2
(b).
So, it suces to consider
E
in some local trivialization of E.
The proof now simply produces the coecients
i
jk
of the desired A in an arbitrary
local trivialization of E, over U say. Any local section of E dened over U may be written,
with respect to the local trivialization, as a vector valued function U R
m
(m being
the rank of E). Let e
j
, j = 1, . . . m denote the sections corresponding in this way to the
constant vector-valued functions on U equal to the j-th standard basis vector of R
m
. Then
the coecients of the connection A are (uniquely) determined by the formula
i
jk
=
(
E
e
j
)(
x
k
)
i
, (2.16)
where
E
(e
j
) is a vector of dierential 1-forms which takes the argument a vector eld. We
used a local coordinate vector eld
x
k
(which is well-dened provided that a trivializing
neighbourhood U for E is also a coordinate neighbourhood for B) and obtained a local
section of E, expressed as a smooth map U V . Then
i
jk
C
E
s =
E
(s
j
e
j
) = (ds
i
+ s
j
i
jk
dx
k
)e
i
= d
A
s
where as usual A = (A
i
j
) = (
i
jk
dx
k
), so we recover the d
A
dened above. It remains to ver-
ify that
i
jk
s actually transform according to (2.12) in any change of local trivialization, so
we get a well-dened connection. The latter calculation is straightforward (and practivally
equivalent to verifying that d
A
is well-dened independent of local trivialization.)
differential geometry 33
The denition of covariant derivative further extends to dierential forms with values
in E by requiring the Leibniz rule, as follows
d
A
( ) = (d
A
) + (1)
q
(d),
q
B
(E),
r
(B),
with above extended in an obvious way to multiply vector-valued dierential forms and
usual dierential forms. It is straightforward to verify, considering local bases of sections
e
j
and dierential forms dx
k
, that in any local trivialization one has d
A
= d + A ,
where
q
B
(E).
Remark (Parallel sections). A section s of E is called parallel, or covariant constant,
if d
A
s = 0. In a local trivialization over coordinate neighbourhood U the section s is
expressed as s = (s
1
(x), . . . , s
m
(x)), x U and the graph of s is respectively =
(x
k
, s
j
(x)) U R
m
: x B, a submanifold of U R
m
. The tangent spaces to are
spanned by
x
k
+
s
j
x
k
a
j
, for k = 1, . . . , n. We nd that the 1-forms
i
s(x)
= da
i
+
i
jk
dx
k
s
j
,
i = 1, . . . , m, vanish precisely on the tangent vectors to .
This is just the horizontality condition for a tangent vector to E and we see that
a section s is covariant constant if and only if any tangent vector to the graph of s is
horizontal. Another form of the same statement: s : B E denes an embedding of B
in E as the graph of s and the tangent space to the submanifold s(B) at p E is the
horizontal subspace at p (relative to A) if and only if d
A
s = 0.
To sum up, a connection on a vector bundle E can be given in three equivalent
ways:
(1) as a (smooth) eld of horizontal subspaces in TE depending linearly on the bre
coordinates, as in (2.10a);
or
(2) as a system of matrix-valued 1-forms A
i
j
(a system of smooth functions
i
jk
) assigned to
every local trivialization of E and satisfying the transformation rule (2.12) on the overlaps;
or
(3) as a covariant derivative
E
on the sections of E and, more generally, on the dierential
forms with values in E.
2.3 Curvature.
Let A be a connection on vector bundle E and consider the repeated covariant dierentia-
tion of an arbitrary section (or r-form) s
r
B
(E) (assume r = 0 though). Calculation in
a local trivialization gives
d
A
d
A
s = d(ds+As)+A(ds+As) = (dA)sA(ds)+A(ds)+A(As) = (dA+AA)s.
Thus d
A
d
A
is a linear algebraic operator, i.e. unlike the dierential operator but d
A
, the
d
A
d
A
commutes with the multiplication by smooth functions.
d
A
d
A
(fs) = fd
A
d
A
s, for any f C
(B). (2.17)
34 alexei kovalev
Notice that the formula (2.17) does not make explicit reference to any local trivialization.
We nd that (d
A
d
A
s) at any point b B is determined by the value s(b) at that point. It
follows that the operator d
A
d
A
is a multiplication by an endomorphism-valued dierential
2-form. (This 2-form can be recovered explicitly in coordinates similarly to (2.16), using
a basis e
i
say of local sections and a basis of dierential 1-forms dx
k
in local coordinates
on B.)
Denition. The form
F(A) = dA + A A
2
(B; End E).
is called the curvature form of a connection A.
Denition. A connection A is said to be at is its curvature form vanishes F(A) = 0.
Example. Consider a trivial bundle B R
m
, so the space of sections is just the vector-
functions C
(B; R
m
). Then exterior derivative applied to each component of a vector-
function is a well-dened linear operator satisfying Leibniz rule (2.14). The corresponding
connection is called trivial, or product connection. It is clearly a at connection.
The converse is only true with an additional topological condition that the base B is
simply-connected; then any at connection on E induces a (global) trivialization E
= B
R
m
(Examples 3, Q5) and will be a product connection with respect to this trivialisation.
Bianchi identity.
Covariant derivative on a vector bundle E with respect to a connection A can be extended,
in a natural way, to any section of B of End E by requiring the following formula to hold,
(d
A
B)s = d
A
(Bs) B(d
A
s),
for every section s of E. Notice that this is just a suitable form of Leibniz rule.
The denition further extends to dierential forms with values End E, by setting for
every
p
(B; End E) and
q
(B; E),
(d
A
) = d
A
( ) (1)
p
(d
A
).
(Write =
k
B
k
k
and =
j
s
j
j
.) It follows that a suitable Leibniz rule also holds
when
1
p
(B; End E),
2
q
(B; End E) to give d
A
(
1
2
) = (d
A
1
)
2
+(1)
p
(d
A
2
). In the special case of trivial vector bundle, E = BR and with d
A
= d the exterior
dierentiation, the above formuli recover the familiar results for the usual dierential forms.
In particular, any
2
(B; End E) in a local trivialization becomes a matrix of 2-forms
and its covariant derivative is a matrix of 3-forms given by
d
A
= d + A A.
Now, for any section s of E, we can write d
A
(d
A
d
A
)s = (d
A
d
A
)d
A
s, i.e. d
A
(F(A)s) =
F(A)d
A
s and comparing with the Leibniz rule above we obtain.
Proposition 2.18 (Bianchi identity). Every connection A satises d
A
F(A) = 0.
differential geometry 35
2.4 Orthogonal and unitary connections
Recall that an orthogonal structure on a (real) vector bundle E dened as a family
of
local trivializations (covering the base) of this bundle so that all the transition functions
between these take values in the orthogonal group O(m), m = rk E. These trivializations
the
corresponding local trivializations. Then on each restriction E[
W
we may consider a
trivial product connection d
()
dened using
. Now consider
a partition of unity
i
subordinate to W
. The expressions
i
s,
i
d
(i)
s make sense over all
of B as we may extend by zero away from W
i
. Now dene
E
s :=
i=1
d
(i)
(
i
s) =
i=1
i
d
(i)
s, (2.22)
where for the second equality we used Leibniz rule for d
(i)
and the property
i=1
i
= 1
(so
i=1
d
i
= 0). The
E
dened by (2.22) is manifestly linear in s and Leibniz rule for
E
holds because it does for each d
(i)
.
Part III: Dierential geometry (Michaelmas 2003, 2004)
Alexei Kovalev (A.G.Kovalev@dpmms.cam.ac.uk)
3 Riemannian geometry
3.1 Riemannian metrics and the LeviCivita connection
Let M be a smooth manifold.
Denition. A bilinear symmetric positive-denite form
g
p
: T
p
M T
p
M R
dened for every p M and smoothly depending on p is called a Riemannian metric
on M.
Positive-denite means that g
p
(v, v) > 0 for every v ,= 0, v T
p
M. Smoothly depending
on p means that for every pair X
p
,Y
p
of C
-smooth function of p M.
Alternatively, consider a coordinate neighbourhood on M containing p and let x
i
,
i = 1, . . . , dimM be the local coordinates. Then any two tangent vectors u, v T
p
M
may be written as u = u
i
(
x
i
)
p
, v = v
i
(
x
i
)
p
and g
p
(u, v) = g
ij
(p)u
i
v
j
, where the functions
g
ij
(p) = g(
_
x
i
_
p
,
_
x
j
_
p
) express the coecients of the metric g in local coordinates. One
often uses the following notation for a metric in local coordinates
g = g
ij
dx
i
dx
j
.
The bilinear form (metric) g will be smooth if and only if the local coecients g
ij
= g
ij
(x)
are smooth functions of local coordinates x
i
on each coordinate neighbourhood.
Example 3.1. Recall (from Chapter 1) that any smooth regularly parameterized surface S
in R
3
,
r : (u, v) U R
2
r(u, v) R
3
.
is a 2-dimensional manifold (more precisely, we assume here that S satises all the dening
conditions of an embedded submanifold). The rst fundamental form
1
Edu
2
+ 2Fdudv +
Gdv
2
is a Riemannian metric on S.
The following formulae are proved in multivariate calculus.
A curve on S may be given as (t) = r(u(t), v(t)), a t b. The length of is then
computed as
_
b
a
[ (t)[dt =
_
b
a
E u
2
+ 2F u v + G v
2
dt.
The area of S is
__
U
EGF
2
dudv.
1
E = (r
u
, r
u
), F = (r
u
, r
v
), G = (r
v
, r
v
) using the Euclidean inner product
36
differential geometry 37
Theorem 3.2. Any smooth manifold M can be given a Riemannian metric.
Proof. Indeed, M may be embedded in R
m
by Whitney theorem. Then the restriction
(more precisely, a pull-back) of the Euclidean metric of R
m
to M denes a Riemannian
metric on M. (Examples 3, Q1).
Remark . A metric, being a bilinear form on the tangent spaces, can be pulled back via a
smooth map, f say, in just the same way as a dierential form. But a pull-back f
g of a
metric g will be a well-dened metric only if f has an injective dierential.
Remark . As a Riemannian metric on M is an inner product on the vector bundle TM,
Theorem 3.2 is also a consequence of Q7 of Examples 2.
Denition. A connection on a manifold M is a connection on its tangent bundle TM.
Recall that a choice of local coordinates x on M determines a choice of local trivial-
ization of TM (using the basis vector elds
x
i
on coordinate patches). The transition
function for two trivializations of TM is given by the Jacobi matrices of the correspond-
ing change of coordinates (
i
i
) = (
x
i
x
i
).
Let
i
jk
be the coecients (Christoel symbols) of a connection on M in local coordi-
nates x
i
. For any other choice x
i
i
jk
=
i
x
i
x
i
x
j
x
j
x
k
x
k
+
x
i
x
i
2
x
i
x
j
x
k
(3.3)
One can see from the above formula that if
i
jk
are the coecients of a connection on M
then
i
kj
also are the coecients of some well-dened connection on M (in general, this
would be a dierent connection).
The dierence T
i
jk
=
i
jk
i
kj
is called the torsion of a connection (
i
jk
). The trans-
formation law for T
i
jk
is T
i
jk
= T
i
x
i
x
i
x
j
x
j
x
k
x
k
, thus the torsion of a connection is a well-
dened antisymmetric bilinear map sending a pair of vector elds X, Y to a vector eld
T(X, Y ) = T
i
jk
X
j
Y
k
on M.
Denition. A connection on M is symmetric if its torsion vanishes, i.e. if
i
jk
=
i
kj
.
Notation: given a connection (covariant derivative) D :
0
M
(TM)
1
M
(TM) and a
smooth vector eld X on M, we write D
X
for the composition of D and contraction of
1-forms (in
1
M
(TM)) with X. Thus D
X
:
0
M
(TM)
0
M
(TM) is a linear dierential
operator acting on vector elds on M. In local coordinates, it is expressed as (D
X
Y )
i
=
X
j
j
Y
i
+
i
jk
Y
j
X
k
.
Here is a way to dene a symmetric connection independent of the local coordinates.
Proposition 3.4. A connection D is symmetric if and only if D
X
Y D
Y
X = [X, Y ].
The proof is an (easy) straightforward computation.
38 alexei kovalev
Theorem 3.5. On any Riemannian manifold (M, g) there exists a unique connection D
such that
(1) d(g(X, Y ))(Z) = g(D
Z
X, Y ) + g(X, D
Z
Y ) for any vector elds X, Y, Z on M; and
(2) the connection D is symmetric.
D is called the LeviCivita connection of the metric g.
The condition (1) in the above theorem is sometimes written more neatly as
dg(X, Y ) = g(DX, Y ) + g(X, DY ).
Proof. Uniqueness. The conditions (1) and (2) determine the coecients of LeviCivita in
local coordinates as follows. A coordinate vector eld
x
i
with constant coecients has
covariant derivative D
x
i
=
p
ik
x
p
dx
k
. The condition (1) with X =
x
i
, Y =
x
j
, Z =
x
k
gives
x
k
g
ij
=
p
ik
g
pj
+
p
jk
g
ip
. (3.6a)
Cycling i, j, k in the above formula, one can write two more relations
x
j
g
ki
=
p
kj
g
pi
+
p
ij
g
kp
, (3.6b)
x
i
g
jk
=
p
ji
g
pk
+
p
ki
g
jp
. (3.6c)
Let (g
iq
) denote the inverse matrix to (g
iq
), so
p
jk
g
qp
g
iq
=
i
jk
. Adding the rst two
equations of (3.6) and subtracting the third, dividing by 2, and multiplying both sides of
the resulting equation by (g
iq
), one obtains the formula
i
jk
=
1
2
g
iq
_
g
qj
x
k
+
g
kq
x
j
g
jk
x
q
_
(3.7)
(also taking account of the symmetry condition (2)). Thus if LeviCivita connection exists
then its coecients in local coordinates are expressed in terms of the metric by (3.7).
Existence. In view of the above calculations it suces to make sure that the for-
mula (3.7) indeed gives a well-dened connection on M. This can be done by verifying that
the
i
jk
s transform in the right way, i.e. as in (3.3), under a change of local coordinates.
The transformation law for g
ij
is g
i
j
=
x
i
x
i
g
ij
x
j
x
j
l
Y
i
+
i
jk
Y
j
Z
k
)
i
for any vector elds Z = Z
l
l
, Y = Y
i
i
on U. Let Y = Z = . Then
at any point p = (t) we have Z
l
l
Y
i
= x
l
x
i
x
l
= x
i
by the chain rule. It follows that the
40 alexei kovalev
equation (3.9) is equivalent to (3.11) if the latter if restricted to the points of the curve .
It can also be seen, by inspection of the above construction, that D
is independent of
the choice of extension of (t) to a vector eld on U.
We have d( , )
g
= (D
, )
g
+ ( , D
)
g
from the dening properties of the Levi
Civita. Hence d[ [
2
g
= 0, by (3.11). But, by the construction, [ [
g
is a function of
x
1
only, so d[ [
2
g
= x
1
(t)
d
dx
1
[ [
2
g
. As x
1
(t) ,= 0 on U by earlier assumption, we must have
(d/dx
1
)[ [
2
g
= 0 and [ [
g
= const as we had to prove.
Examples. 1. On R
n
with the Euclidean metric
(dx
i
)
2
we have
i
ik
= 0, so the Levi
Civita is just the exterior derivative D = d. The geodesics
x
i
= 0 are straight lines
p
(t, a) = p + at parameterized with constant velocity.
2. Consider the sphere S
n
with the round metric (i.e. the restriction of the Euclidean
metric to S
n
R
n+1
). Then p S
n
and a T
p
S
n
may be regarded as the vectors
in R
n+1
. Suppose a ,= 0, then the orthogonal reection L in the 2-dimensional subspace
P = spanp, p+a is an isometry of S
n
. Now L preserves the metric and p and a, the data
which determines the geodesic
p
(, a). As
p
(, a) is moreover uniquely determined it must
be contained in the xed point set of L. But the xed point set in a curve, the great circle
P S
n
. We nd that great circles, parameterized with velocity of constant lengthand
only theseare the geodesics on S
n
.
Observe that for any geodesic
p
(t, a) and any real constant the path
p
(t, a) is also
a geodesic and
p
(t, a) =
p
(t, a).
By application of a general result in the theory of ordinary dierential equations, a
geodesic
p
(t, a) must depend smoothly on its initial conditions p, a. Furthermore, there
exist
1
> 0 and
2
> 0 independent of a and such that if [a[ <
1
then
p
(t, a) exists for
all 2
2
< t < 2
2
. It follows that
p
(1, a) is dened whenever [a[ < =
1
2
.
Denition. The exponential map at a point p of a Riemannian manifold (M, g) is
exp
p
: a N T
p
M (1; p, a) M.
Proposition 3.12. (d exp
p
)
0
= id(T
p
M)
Proof. We use the canonical identication a T
p
M
d
dt
(ta)[
t=0
to dene (d exp
p
)
0
as a
linear map on T
p
M (rather than on T
0
(T
p
M)).
Let [a[ < , so
p
(t, a) =
p
(1, ta) is dened for 0 t 1. Then we have
(d exp
p
)
0
a =
d
dt
exp
p
(ta)[
t=0
=
d
dt
p
(1, ta)[
t=0
=
d
dt
p
(t, a)[
t=0
= (0, a) = a.
Corollary 3.13. The exponential map exp
m
denes a dieomorphism from a neighbour-
hood of zero in T
m
M to a neighbourhood of m in M.
differential geometry 41
Proof. Apply the Inverse Mapping Theorem (page 8 of these notes).
Corollary 3.13 means that the exponential map denes near every point p of a Rieman-
nian manifold a system of local coordinatescalled normal (or geodesic) coordinates
at p. It is not dicult to see that the geodesics
p
(t, a) are given in these coordinates by
rays emanating from the origin.
It also makes sense to speak of geodesic polar coordinates at p M dened by the
polar coordinates on T
p
M via a dieomorphism
f : (r, v) ]0, [ S
n1
exp
p
(r v) M. (3.14)
Here ]0, [S
n1
is regarded as a subset in T
p
M
= R
n
via the inner product g(p). If
0 < r < then the image
r
= f(r S
n1
T
p
M) of the metric sphere of radius r
is well-dened on M and is called a geodesic sphere about p. (So
r
is an embedded
submanifold of M.) The following remarkable result asserts that the geodesic spheres are
orthogonal to their radii.
Gauss Lemma. The geodesic (t; p, a) is orthogonal to
r
. Thus the metric g in geodesic
polar coordinates has local expression g = dr
2
+h(r, v), where for any 0 < r < , h(r, v) is
the metric induced by g on
r
(by restriction).
Proof. Let X be an arbitrary smooth vector eld on the unit sphere S
n1
T
p
M. Use
polar coordinates to make sense of X as a vector eld (independent of r) on the punctured
unit ball B 0 T
p
M. Dene a vector eld
X(r, v) = rX(v) on B 0. The map
exp
p
induces a vector eld Y (f(r, v)) = (d exp
p
)
rv
X(r, v) on the punctured geodesic ball
B
p = exp
p
(B 0) in M.
We shall be done if we show that Y is everywhere orthogonal to the radial vector
eld
r
. Note that, by construction, any geodesic from p is given in normal coordinates by
(t; p, a) = at, so (t; p, a)/[a[ =
r
. Here [a[ means the norm in the inner product g
p
on
the vector space T
p
M. By application of Corollary 3.13, the family (t; p, a), where [a[ = 1,
denes a smooth vector eld on B
X D
r
_
= (df)
d
dr
X = (df)(
X/r) = Y/r,
with the help of Proposition 3.4. Therefore, we nd
d
dr
g(Y, ) = g(D
Y, ) + g(Y, D
) = g(D
Y, ) = g(D
Y
+
1
r
Y, ) =
1
r
g(Y, ).
as 2g(D , ) = d g( , ) = 0 by Proposition 3.11. Thus
d
dr
G = G/r, where G = g(Y, ).
Hence G is linear in r and
d
dr
G independent of r. But lim
r0
d
dr
G = lim
r0
g(X,
r
) = 0, as
(d exp
p
)
0
is an isometry by Proposition 3.12, and so g(Y, ) = 0 and the result follows.
42 alexei kovalev
3.3 Curvature of a Riemannian manifold
Let g be a metric on a manifold M. The (full) Riemann curvature R = R(g) of g is, by
denition, the curvature of the LeviCivita connection of g. Thus R
2
M
(End(TM)),
locally a matrix of dierential 2-forms R =
1
2
(R
i
j,kl
dx
l
dx
k
), i, j, k, l = 1 . . . n = dimM.
The coecients (R
i
j,kl
) form the Riemann curvature tensor of (M, g). Given two vec-
tor elds X, Y , one can form an endomorphism R(X, Y ) End(TM); its matrix in
local coordinates is R(X, Y )
i
j
= R
i
j,kl
X
k
Y
l
(as usual X = X
k
k
, Y = Y
l
l
). Denote
R
kl
= R(
k
,
l
) End(TM).
Recall that in local coordinates a connection (covariant derivative) may be written as
d + A, with A =
i
jk
dx
k
= A
k
dx
k
. We write D
k
= D
x
k
=
x
k
+ A
k
. The denition of the
curvature form of a connection (Chapter 2, p. 31) yields an expression in local coordinates
R
i
j,kl
=
_
D
l
D
k
x
j
D
k
D
l
x
j
_
i
, or R
kl
= [D
k
, D
l
], (3.15)
considering the coecient at dx
l
dx
k
. Now D
X
= X
k
D
k
and so we have [D
X
, D
Y
] =
[X
k
D
k
, X
l
D
l
] = X
k
(
k
Y
l
)D
l
X
k
Y
l
D
k
D
l
+ Y
k
(
k
X
l
)D
l
+ X
k
Y
l
D
l
D
k
= X
k
Y
l
R
kl
[X, Y ]
l
D
l
. We have thus proved
Lemma 3.16. R(X, Y ) = D
[X,Y ]
[D
X
, D
Y
].
One also can combine (3.15) with (3.7) and thus obtain an explicit local expression for
R
i
j,kl
in terms of the coecients of the metric g and their rst and second derivatives.
It is convenient to consider R
ij,kl
= g
iq
R
q
j,kl
, which denes a map on 4-tuples of vector
elds (X, Y, Z, T) g(R(X, Y )Z, T).
Proposition 3.17.
(i) R
ij,lk
= R
ij,kl
= R
ji,kl
;
(ii) R
i
j,kl
+ R
i
k,lj
+ R
i
l,jk
= 0;
(iii) R
ij,kl
= R
kl,ij
.
Proof. (i) The rst equality is clear. For the second equality, one has, from the denition
of the LeviCivita connection,
g
kl
x
i
= g(D
i
x
k
,
x
l
) + g(
x
k
, D
i
x
l
), and further
2
g
kl
x
j
x
i
= g(D
j
D
i
x
k
,
x
l
) + g(D
i
x
k
, D
j
x
l
) + g(D
j
x
k
, D
i
x
l
) + g(
x
k
, D
j
D
i
x
l
).
The right-hand side of the above expression is symmetric in i, j as
2
g
kl
x
i
x
j
=
2
g
kl
x
j
x
i
. The
anti-symmetric part of the right-hand side (which has to be zero) equals R
ij,kl
+ R
ji,kl
.
(ii) Firstly, (D
k
x
j
)
i
=
i
jk
= (D
j
x
k
)
i
, by the symmetric property of the Levi-Civita.
The claim now follows by straightforward computation using (3.15).
(iii) Multiplying (ii) by g
iq
gives R
ij,kl
+ R
ik,lj
+ R
il,jk
= 0. Similarly, R
jk,li
+ R
jl,ik
+
R
ji,kl
= 0, R
kl,ij
+ R
ki,jl
+ R
kj,li
= 0, and R
li,jk
+ R
lj,ki
+ R
lk,ij
= 0. Adding up the four
identities and making cancellations using (i) (the octahedron trick) gives the required
result.
differential geometry 43
Corollary 3.18. The Riemann curvature tensor (R
ij,kl
)
p
denes, at any point p M a
symmetric bilinear form on
2
T
p
M.
There are natural ways to extract simpler quantities (i.e. with less components) from
the Riemann curvature tensor.
Denition. The Ricci curvature of a metric g at a point p M, Ric
p
= Ric(g)
p
, is the
trace of the endomorphism v R
p
(x, v)y of T
p
M depending on a pair of tangent vectors
x, y T
p
M.
Thus in local coordinates Ric(p) is expressed as a matrix Ric = (Ric
ij
), Ric
ij
=
q
R
q
i,jq
.
That is, the Ricci curvature at p is a bilinear form on T
p
M. A consequence of Proposi-
tion 3.17(iii) is that this bilinear form is symmetric, Ric
ij
= Ric
ji
.
Denition. The scalar curvature of a metric g at a point p M, s = scal(g)
p
is a
smooth function on M obtained by taking the trace of the bilinear form Ric
ij
with respect
to the metric g.
If local coordinates are chosen so that g
ij
(p) =
ij
, then the latter denition means
that s(p) =
i
Ric
ii
(p) =
i,j
R
ij,ij
(p). For a general g
ij
, the formula may be written
as s =
i
g
ij
Ric
ij
, where g
ij
is the induced inner product on the cotangent space with
respect to the dual basis, algebraically (g
ij
) is the inverse matrix of (g
ij
)).
3.3.1 Some examples
(1) It makes sense to consider the condition
Ric = g (3.19)
for some constant R, as both the metric and its Ricci curvature are symmetric bilinear
forms on the tangent spaces to M. When the condition (3.19) is satised, the Riemannian
manifold (M, g) is called Einstein manifold. In particular, if (3.19) holds with = 0 then
M is said to be Ricci-at.
(2) Recall that if is a surface in R
3
(smooth, regularly parameterized by (u, v) in an
open set in R
2
) then there is a metric induced on , expressed as the rst fundamental
form Edu
2
+ 2Fdudv + Gdv
2
. The second fundamental form Ldu
2
+ 2Mdudv + Ndv
2
is
dened by taking the inner products L = (r
uu
, n), M = (r
uv
, n), N = (r
vv
, n) with the
unit normal vector to , n = r
u
r
v
/[r
u
r
v
[ (the subscripts u and v denote respective
partial derivatives). The quantity
K =
LN M
2
EGF
2
is called the gaussian curvature of . A celebrated theorema egregium, proved by Gauss,
asserts that K is determined by the coecients of rst fundamental form, i.e. by the metric
on (and so K is independent of the choice of an isometric embedding of in R
3
).
44 alexei kovalev
Taking up a general view on as a 2-dimensional Riemannian manifold, one can check
that 2(EG F
2
)
1
R
12,12
= s, the scalar curvature of . From the results of the next
section, we shall see (among other things) that the scalar curvature of a surface is twice
its gaussian curvature s = 2K.
3.4 Riemannian submanifolds
When a manifold M
n
is an embedded submanifold of a Riemannian manifold, say V
n+r
,
the Riemannian metric g
V
on V induces, by restriction, a Riemannian metric g
M
on M.
What is the relation between the LeviCivita connection
D of g
V
and the LeviCivita
connection D of g
M
?
To see this relation, it is convenient to consider the vector bundle E =
(TV ) over M,
where : M V is the embedding map. (Informally, E is just the restriction of TV to M
if the latter is regarded as a subset of V .)
In the next proposition, we write x
k
for local coordinates on M, y
k
=
y
x
k
, where
X
)[
M
=
X
, where in the left-hand side we use the connection on E. In
particular, the right-hand side is independent of the choices of extensions
X and .
Thus the connection
on E makes natural sense from all three points of view. Note
that we did not require any metric to dene this induced connection.
Each bre E
x
of E contains T
x
M as a subspace. Using now the metric on M we obtain
a direct sum decomposition
E
x
= T
x
M (T
x
M)
. (3.21)
The disjoint union of the orthogonal complements .
xM
(T
x
M)
X
Y )
x
= (
X
Y )
x
+ (h(X, Y ))
x
, according to (3.21), where h(X, Y ) is some section
of N
M/V
. It turns out that is a well-dened covariant derivative (connection) on M and
h is a bilinear map T
x
M T
x
M (T
x
M)
D
X
Y = D
X
Y + II(X, Y ),
where D is the LeviCivita connection of the induced metric on M, and II is a symmetric
bilinear map called the second fundamental form of M in V .
Theorem 3.23 (Weingarten formula). For any vector eld X on M and section of the
normal bundle N
M/V
,
D
X
= S
X +
X
,
where for any , S
is a connection on N
M/V
. Furthermore, the shape operator is symmetric with respect
to the induced Riemannian metric M,
g
M
(S
X, Y ) = g
M
(X, S
Y ) = g
V
(II(X, Y ), ),
for any vector eld Y on M.
By direct application of the above, we can compute the Riemann curvature R = (R
ij,kl
)
of M in terms of the curvature of the ambient manifold and the second fundamental form.
Theorem 3.24 (Gauss).
R(X, Y, Z, T) =
R(X, Y, Z, T) + g
V
(II(X, Z), II(Y, T)) g
V
(II(X, T), II(Y, Z)).
Corollary 3.25. The curvature of a submanifold M of a at manifold is determined by
the second fundamental form of M.
When M is a surface in the Euclidean R
3
, this is equivalent to theorema egregium
discussed in the previous section.
3.5 LaplaceBeltrami operator
Throughout this section M is a connected oriented Riemannian manifold of dimension n.
Let g denote a metric on M and let the orientation be given by a nowhere-zero n-form .
Starting from the vector elds
x
1
, . . .
x
n
at a point x in a coordinate neighbourhood U,
we can apply GramSchmidt process with x as a parameter. Thus we obtain a new system
of (smooth) vector elds e
1
, . . . , e
n
which give an orthonormal basis of tangent vectors on a
perhaps smaller neighbourhood of p (still denote this neighbourhood by U). Let
1
, . . .
n
on U be the dual 1-forms to e
1
, . . . , e
n
, in the sense that
j
(e
i
) =
ij
at any point in U.
Then
j
give at every point p of U a basis of T
p
M, the dual basic to e
j
.
The metric on M induces, for every p = 0, . . . , n an inner product on the bundle
p
T
M
by making
i
1
(x) . . .
i
p
(x) : 1 i
1
< . . . < i
p
n an orthonormal basis of
p
T
x
M.
46 alexei kovalev
If
j
is another system of local 1-forms, on another coordinate neighbourhood U
say,
and
j
are orthonormal at every point in U
then
1
. . .
n
= det()
1
. . .
n
on U
U,
for some orthogonal matrix (depending on x U
(M). Then
g
is called the volume form of M.
In (positively oriented) local coordinates,
g
=
_
det g
ij
dx
1
. . . dx
n
.
Denition. The Hodge star on M is a linear operator on the dierential forms
:
p
T
x
M
np
T
x
M,
such that for any two p-forms ,
p
T
x
M one has = , )
g
g
(x), where
g
is
the volume form on M.
It follows that if
i
is an orthonormal basis of a cotangent space T
x
M then necessarily
(
1
. . .
p
) =
p+1
. . .
n
. In particular, 1 =
g
and
g
= 1. By permutations
of indices and by linearity, the Hodge star is then uniquely determined for any dierential
form on M. Further, it follows that = (1)
p(np)
on the p-forms.
Using the Hodge star we construct a dierential operator
:
p
(M)
p1
(M)
putting = (1)
n(p+1)+1
d if p ,= 0 and = 0 on
0
(M) = C
(M).
Denition. The LaplaceBeltrami operator, or Laplacian, on M is a linear dieren-
tial operator :
p
(M)
p
(M) given by
= d + d.
Straightforward computation shows that when M is the Euclidean R
n
the denition
gives f =
2
f
(x
1
)
2
. . .
2
f
(x
n
)
2
for any smooth function f. For a general metric
g = (g
ij
), the local expression becomes
g
f =
1
det g
x
j
_
_
det g g
ij
f
x
i
_
.
Proposition 3.26. The operator is the adjoint
2
of d in the sense that
_
M
d, )
g
g
=
_
M
, )
g
g
,
for every compactly supported
p1
(M),
p
(M).
2
It is more correct to say that is the formal adjoint of d for reasons that have to do with the Analysis.
differential geometry 47
Using the inner product on the spaces
p
T
p
M, p M, we can dene an inner product on
p
(M), called the L
2
inner product, by putting , )
L
2 =
_
M
, )
g
g
. The inner product
makes each
p
(M) into a normed space, with L
2
-norm dened by || = (, )
L
2)
1/2
. In
particular, = 0 if and only if || = 0.
Thus Proposition 3.26 says that d, )
L
2 = , )
L
2 and, consequently, , )
L
2 =
, )
L
2. It follows immediately that the LaplaceBeltrami operator is self-adjoint.
A dierential form
p
(M) is called harmonic if = 0.
Corollary 3.27. Every harmonic dierential form on a compact manifold is closed and
co-closed: = 0 if and only if both d = 0 and = 0.
Proof. Integration by parts, 0 = d + d, )
L
2 = , )
L
2 +d, d)
L
2.
It is also easily checked that = on any
p
(M). Therefore the Hodge star of any
harmonic form is again harmonic.
Hodge Decomposition Theorem. Let M be a compact oriented Riemannian manifold.
For every 0 p dimM, the space H
p
of harmonic p-forms is nite-dimensional. Fur-
thermore, there are L
2
-orthogonal direct sum decompositions
p
(M) =
p
(M) H
p
= d
p
(M) d
p
(M) H
p
= d
p1
(M)
p+1
(M) H
p
(where we formally put
1
(M) = 0).
Remark: the compactness condition on M cannot be removed.
3
Short summary of the proof. We need to introduce the concept of a weak solution of
= . (3.28)
A weak solution of (3.28) is by denition, a linear functional l :
p
(M) R which is
(i) bounded, [l()[ C||, for some C > 0 independent of , and
(ii) satises l() = , )
L
2.
Any solution of (3.28) denes a weak solution by putting l
() = , )
L
2.
The proof of Hodge Decomposition Theorem requires some results from Functional
Analysis.
Regularity Theorem. Any weak solution l of (3.28) is of the form l() = , )
L
2, for
some
p
(M) (and hence denes a solution of (3.28)).
Compactness Theorem. Assume that a sequence
n
p
(M) satises |
n
| < C and
|
n
| < C, for some C independent of n. Then
n
contains a Cauchy subsequence.
3
The reason is that certain results in Analysis fail on non-compact sets, but this is another story.
48 alexei kovalev
We shall assume the above two theorems (and the HahnBanach theorem below)
without proof.
Compactness Theorem implies at once that H
p
must be nite-dimensional (for,
otherwise, there would exist an innite orthonormal sequence of forms). As H
p
is nite-
dimensional, we can write an L
2
-orthogonal decomposition
p
(M) = H
p
(H
p
)
.
It is easy to see that
p
(M) (H
p
)
p
(M) R of (3.28).
Dene l rst on a subspace
p
(M), by putting l() = , )
L
2. It is not hard to
check that l is well-dened. Further, (ii) is automatically satised (on this subspace); we
claim that (i) holds too. To verify the latter claim, we show that l is bounded below on
p
(M) using, once again, the Compactness Theorem.
In order to extend l to all of
p
(M), we appeal to
HahnBanach Theorem. Suppose that L is a normed vector space, and L
0
a subspace
of L, and l : L
0
R a linear functional satisfying l(x
0
) < |x
0
|, for all x
0
L
0
. Then l
extends to a linear functional on L with l(x) < |x| for all x L.
Thus we obtain a weak solution of (3.28) and deduce that
p
(M) =
p
(M) H
p
as desired. The two other versions of the L
2
-orthogonal decomposition of
p
(M) follow
readily by application of Proposition 3.26.
Corollary 3.29. Every de Rham cohomology class a H
r
(M) of a compact oriented Rie-
mannian manifold M is represented by a unique harmonic dierential r-form
r
(M),
[] = a. Thus H
r
= H
r
(M).
Proof. Uniqueness. If
1
,
2
are harmonic p-forms and
1
2
= d then |d|
2
=
d,
1
2
)
L
2 = , (
1
2
))
L
2 = 0.
Existence. If is such that d = 0 then || = 0. Hence any closed p-form must
be in d
p1
(M) H
p
.
Corollary 3.29 is a surprising result: an analytical object (harmonic forms) turns out to
be equivalent to a topological object (de Rham cohomology) via some dierential geometry.
Here is a way to see why such a result can be true.
A de Rham cohomology class, a H
r
(M) say, can be represented by many dierential
forms; consider the (innite-dimensional) ane space
B
a
=
r
(M) [ d = 0, [] = a H
r
(M)
=
r
(M)[ = + d, for some
r1
(M).
When does a closed form have the smallest L
2
-norm amongst all the closed forms in a
given de Rham cohomology class B
a
?
differential geometry 49
Such a form must be a critical point of the function F( +d) = | +d)|
2
on B
a
, so
the partial derivatives of F in any direction should vanish. That is, we must have
0 =
d
dt
t=0
+ t d, + t d)
L
2 = 2, d)
L
2.
Integrating by parts, we nd that
_
M
, )
g
= 0 must hold for every
r1
(M). This
forces = 0, and so the extremal points of F are precisely the harmonic forms .
Page references for this chapter
all to GallotHulinLafontaine except
to Warner
Riemannian metrics pp.5253,
LeviCivita connection pp.6971,
curvature of a Riemannian manifold pp.107108,111112,155156
geodesics, Gauss Lemma pp.8085,8990,
Riemannian submanifolds pp.217220,
the Laplacian and the Hodge Decomposition Theorem [W], pp. 140141,220226.