i

Geometry in Physics
Contents
1 Exterior Calculus page 1
1.1 Exterior Algebra 1
1.2 Dierential forms in R
n
7
1.3 Metric 29
1.4 Gauge theory 39
1.5 Summary and outlook 48
2 Manifolds 49
2.1 Basic structures 49
2.2 Tangent space 53
2.3 Summary and outlook 65
3 Lie groups 66
3.1 Generalities 66
3.2 Lie group actions 67
3.3 Lie algebras 69
3.4 Lie algebra actions 72
3.5 From Lie algebras to Lie groups 74
ii
1
Exterior Calculus
Dierential geometry and topology are about mathematics of objects that are, in a sense,
smooth. These can be objects admitting an intuitive or visual understanding curves, surfaces,
and the like or much more abstract objects such as high dimensional groups, bundle spaces,
etc. While dierential geometry and topology, respectively, are overlapping elds, the perspective
at which they look at geometric structures are dierent: dierential topology puts an emphasis
on global properties. Pictorially speaking, it operates in a world made of amorphous or jelly-like
objects whose properties do not change upon continuous deformation. Questions asked include:
is an object compact (or innitely extended), does it have holes, how is it embedded into a host
space, etc?
In contrast, dierential geometry puts more weight on the actual look of geometric struc-
tures. (In dierential geometry a coee mug and a donut are not equivalent objects, as they
would be in dierential topology.) It often cares a about distances, local curvature, the area
of surfaces, etc. Dierential geometry heavily relies on the fact that any smooth object, looks
locally at (you just have to get close enough.) Mathematically speaking, smooth objects can
be locally modelled in terms of suitably constructed linear spaces, and this is why the con-
cepts of linear algebra are of paramount importance in this discipline. However, at some point
one will want to explore how these at approximations change upon variation of the reference
point. Variations belong to the department of calculus, and this simple observations shows that
dierential geometry will to a large extend be about the synthesis of linear algebra and calculus.
In the next section, we will begin by introducing the necessary foundations of linear algebra,
notably tensor algebra. Building on these structures, we will then advance to introduce elements
of calculus.
1.1 Exterior Algebra
In this section, we introduce the elements of (multi)linear algebra relevant to this course.
Throughout this chapter, V will be an Rvector space of nite dimension n.
1.1.1 Dual Basis
Let V

be the dual vector space of V , i.e. the space of all linear mappings from V to the real
numbers:
V

= : V R, linear
1
2 Exterior Calculus
differential topology:
compactness?
holes?
embedding in outer space?
differential geometry:
geometric structure?
curvature?
distances?
areas?
Figure 1.1 The arena of dierential topology and geometry. The magnied area illustrates that a
smooth geometric structure looks locally at. Further discussion, see text
Let e
i
a basis of V , and e
i
be the associated basis of V

. The latter is dened by i, j =

1, . . . , N : e
i
(e
j
) =
i
j
. A change of basis e
i
e

i
is dened by a linear transformation
A GL(n). Specically, with
e

i
= (A
1
)
j
i
e
j
, (1.1)
the components v
i
of a vector v = v
i
e
i
transform as v
i
= A
i
j
v
j
. We denote this transforma-
tion behaviour, i.e. transformation under the matrix A
i
j
representing A, as contravariant
transformation. Accordingly, the components v
i
are denoted contravariant components. By
(general) conventions, contravariant components carry their indices upstairs, as superscripts.
Similarly, the components of linear transformations are written as A
i
j
, with the contravariant
indices (upstairs) to the left of the covariant indices (downstairs). For the systematics of this
convention, see below.
The dual basis transforms by the transpose of the inverse of A, i.e. e
i
= A
i
j
e
j
, implying
that the components w
i
of a general element w w
i
e
i
V

transform as w

i
= (A
1
)
j
i
w
j
.
Transformation under the matrix (A
1
)
j
i
is denoted as covariant transformation. Covariant
components carry their indices downstairs, as subscripts.
INFO Recall, that for a general vector space, V , there is no canonical mapping V V

to its
dual space. Such mappings require additional mathematical structure. This additional information
may either lie in the choice of a basis |e
i
in V . As discussed above, this xes a basis |e
i
in V .
A canonical (and basis-independent) mapping also exists if V comes with a scalar product, , ) :
1.1 Exterior Algebra 3
V V R, (v, v

) v, v

). For then we may assign to v V a dual vector v

dened by the
condition w V : v

(w) = v, w). If the coordinate representation of the scalar product reads

v, v

) = v
i
g
ij
v
j
, the dual vector has components v

i
= v
j
g
ji
.
Conceptually, contravariant (covariant) transformation are the ways by which elements of
vector spaces (their dual spaces) transform under the representation of a linear transformation
in the vector space.
1.1.2 Tensors
Tensors (latin: tendo I span) are the most general objects of multilinear algebra. Loosely
speaking, tensors generalize the concept of matrices (or linear maps), to maps that are multi-
linear. To dene what is meant by this, we need to recapitulate that the tensor product V V

of two vector spaces V and V

is the set of all (formal) combinations v v

subject to the
following rules:
v (v

+w

) v v

+v w

(v +w) v

v v

+w v

c(v w) (cv) w v (cw).

Here c R, v, w V and v

, w

. We have written , because the above relations dene

what is meant by addition and multiplication by scalars in V V

; with these denitions V V

becomes a vector space, often called the tensor space V V

. In an obvious manner, the

denition can be generalized to tensor products of higher order, V V

. . . .
We now consider the specic tensor product
T
q
p
(V ) = (
q
V ) (
p
V

), (1.2)
where we dened the shorthand notation
q
V V V
. .
q
. Its elements are called tensors
of degree (q, p). Now, a dual vector is something that maps vectors into the reals. Conversely,
we may think of a vector as something that maps dual vectors (or linear forms) into the reals.
By extension, we may think of a tensor T
q
p
as an object mapping q linear forms and p
vectors into the reals:
: (
q
V

) (
p
V ) R,
(v

1
, . . . v

q
, v
1
, . . . , v
p
) (v

1
, . . . v

q
, v
1
, . . . , v
p
).
By construction, these maps are multilinear, i.e. they are linear in each argument, (. . . , v +
w, . . . ) = (. . . , v, . . . ) +(. . . , w, . . . ) and (. . . , cv, . . . ) = c(. . . , v, . . . ).
The tensors form a linear space by themselves: with ,

T
q
p
(V ), and X (
q
V

)(
p
V ),
we dene (+

)(X) = (X) +(X

) through the sum of images, and (cX) = c(X). Given

a basis e
i
of V , the vectors
e
i
1
e
i
q
e
j
1
e
j
p
, i
1
, . . . , j
p
= 1, . . . , N,
4 Exterior Calculus
form a natural basis of tensor space. A tensor T
q
p
(V ) can then be expanded as
=
N

i
1
,...,j
p
=1

i
1
,...,i
p
j
1
,...,j
p
e
i
1
e
i
q
e
j
1
e
j
p
. (1.3)
A few examples:
T
1
0
is the space of vectors, and
T
0
1
the space of linear forms.
T
1
1
is the space of linear maps, or matrices. (Think about this point!) Notice that the
contravariant indices generally appear to the left of the covariant indices; we have used this
convention before when we wrote A
i
j
.
T
0
2
is the space of bilinear forms.
T
0
N
contains the determinants as special elements (see below.)
Generally, a tensor of valence (q, p) is characterized by the q contravariant and the p covariant
indices of the constituting vectors/co-vectors. It may thus be characterized as a mixed tensor (if
q, p ,= 0) that is contravariant of degree q and covariant of degree p. In the physics literature,
tensors are often identied with their components,
i
1
,...,i
p
j
1
,...,j
p
which are then rather
implicitly characterized by their transformation behavior.
The list above may illustrate, that tensor space is suciently general to encompass practically
all relevant objects of (multi)linear algebra.
1.1.3 Alternating forms
In our applications below, we will not always have to work in full tensor space. However, there is
one subspace of T
0
p
(V ), the so-called space of alternating forms, that will play a very important
role throughout:
Let
p
V

T
0
p
(V ) be the set of plinear real valued alternating forms:

p
V

= :
p
V R, multilinear & alternating. (1.4)
Here, alternating means that (. . . , v
i
, . . . , v
j
, . . . ) = (. . . , v
j
, . . . , v
i
, . . . ).
A few remarks on this denition:
The sum of two alternating forms is again an alternating form, i.e.
p
is a (real) vector space
(a subspace of T
0
p
(V ).)

p
V

is the pth completely antisymmetric tensor power of V

,
p
V

= (
p
1
V

)
asym.
.

1
V = V

and
0
V R.

p>n
V = 0.
dim
p
V

=
_
n
p
_
.
Elements of
p
V

are called forms (of degree p).

1.1 Exterior Algebra 5
1.1.4 The wedge product
Importantly, alternating forms can be multiplied with each other, to yield new alternating forms.
Given a pform and a qform, we dene this so-called wedge product (exterior product) by
:
p
V

q
V

p+q
V

,
(, ) ,
( )(v
1
, . . . , v
p+q
)
1
p!q!

PS
p+q
sgn P (v
P1
, . . . , v
Pp
)(v
P(P+1)
, . . . , v
P(p+q)
).
For example, for p = q = 1, ( )(v, w) = (v)(w) (w)(v). For p = 0 and q = 1,
(v) = (v), etc. Important properties of the wedge product include (
p
V

q
V

,
r
V

, c R):
bilinearity, i.e. (
1
+
2
) =
1
+
2
and (c) = c( ).
associativity, i.e. ( ) = ( ) .
graded commutativity, = ()
pq
.
INFO A (real) algebra is an R-vector space W with a product operation
W W W,
u, v u v,
subject to the following conditions (u, v, w W, c R):
(u +v) w = u w +v w,
u (v +w) = u v +u w,
c(v w) = (cv) w +v (cw).
The direct sum of vector spaces
V

p=0

p
V

(1.5)
together with the wedge product denes a real algebra, the so-called an exterior algebra or
Grassmann algebra. We have dimV

n
p=0
_
n
p
_
= 2
n
. A basis of
p
V

is given by all
forms of the type
e
i
1
e
i
p
, 1 i
1
< < i
p
n.
To see this, notice that i) these forms are clearly alternating, i.e. (for xed p) they belong
to
p
V , they are b) linearly independent, and c) there are 2
n
of them. These three criteria
guarantee the basis-property. Any pform can be represented in the above basis as
=

i
1
<<i
p

i
1
,...,i
p
e
i
1
e
i
p
, (1.6)
6 Exterior Calculus
where the coecients
i
1
,...,i
p
R obtain as
i
1
,...,i
p
= (e
i
1
, . . . , e
i
p
). Alternatively (yet
equivalantly), may be represented by the unrestricted sum
=
1
p!

i
1
,...,i
p

i
1
,...,i
p
e
i
1
e
i
p
,
where
i
1
,...,i
p
is totally antisymmetric in its indices. By construction, the components
of a pform transform as a covariant tensor of degree p. Indeed, it is straightforward
to verify that under the basis transformation
i
1
,...,i
p

i
1
,...,i
p
, where

i
1
,...,i
p
=
(A
1T
)
j
1
i
1
. . . (A
1T
)
j
p
i
p

j
1
,...,j
p
.
1.1.5 Inner derivative
For any v V there is a mapping i
v
:
p
V

p1
V

lowering the tensor degree by one. It

is dened by
(i
v
)(v
1
, . . . , v
p1
) (v, v
1
, . . . , v
p1
).
The components of i
v
obtain by contraction of the components of :
(i
v
)
i
1
,...,i
p1
= v
i

i,i
1
,...,i
p1
.
Properties of the inner derivative:
i
v
is a linear mapping, i
v
( +

) = i
v
+i
v

.
It is also linear in its parametric argument, i
v+w
= i
v
+i
w
.
i
v
obeys the (graded) Leibnitz rule:
i
v
( ) = (i
v
) + ()
p
(i
v
),
p
V

. (1.7)
It is for this rule that we call i
v
a derivative.
The inner derivative is antisymmetric in the sense that i
v
i
w
= i
w
i
v
, in particular,
(i
v
)
2
= 0.
1.1.6 Pullback
Given a form
p
V

and a linear map F : W V we may dene a form (F

)
p
W by
the pullback operation,
F

:
p
V

p
W

,
F

,
(F

)(w
1
, . . . , w
p
) (Fw
1
, . . . , Fw
p
). (1.8)
To obtain a componentrepresentation of the pullback, we consider bases e
i
and f
i
of
V and W, respectively. The map F is dened by a component matrix (mind the horizontal
positioning of indices, F
j
i
= (F
T
)
i
j
.)
Ff
i
= F
j
i
e
j
.
1.2 Dierential forms in R
n
7
It is then straightforward to verify that the action on the dual basis is given by
F

e
i
= F
i
j
f
j
.
(Cf. with the component representation of basis changes, (1.1).) The action of F

on general
forms may be obtained by using the rules
F

is linear.
F

( ) = (F

) (F

),
(F G)

= G

.
The rst two properties state that F

is a (Grassmann)algebra homomorphism, i.e. a linear map

between algebras, compatible with the product operation.
1.1.7 Orientation
Given a basis e
i
of a vector space V , we may dene a topdimensional form e
1
e
2

e
n

n
V

. Take another basis e

i
. We say that the two bases are oriented in the same
way if (e

1
, e

2
, . . . , e

n
) > 0. One veries that (a) orientation denes an equivalence relation
and (b) that there are only two equivalence classes. A vector space together with a choice of
orientation is called an oriented vector space.
EXERCISE Prove the statements above. To this end, show that (e

1
, e

2
, . . . , e

n
) = det(A
1
),
where A is the matrix mediating the basis change, i.e. e

i
= (A
1T
)
j
i
e
j
. Reduce the identication
of equivalence classes to a classication of the sign of determinants.
1.2 Dierential forms in R
n
In this section, we generalize the concepts of forms to dierential forms, that is forms
x
continuously depending on a parameter x. Throughout this section, we assume x U, where
U is an open subset of R
n
. This will later be generalized to forms dened in dierent spaces.
1.2.1 Tangent vectors
Let U R
n
be open in R
n
. For practical purposes, we will often think of U as parameterized
by coordinates. A system of coordinates is dened by a dieomorphism (an bijective map such
that the map and its inverse are dierentiable),
x : K U,
(x
1
, . . . , x
n
) x(x
1
, . . . , x
n
), (1.9)
where the coordinate domain K R
n
is an open subset by itself (cf. Fig. 1.2.)
1
If no confusion
is possible, we do not explicitly discriminate between elements x U and their representation
in terms of an ncomponent coordinate vector.
1
At this stage, the distinction K vs. U both open subsets of R
n
may appear tautological. However, the usefulness of the
distinction between these two sets will become evident in chapter 2.
8 Exterior Calculus
Figure 1.2 An open subset of R
n
and its coordinate representation
EXAMPLE For a chosen basis in R
n
, each element x U may be canonically associated to an
ncomponent coordinate vector (x
1
, . . . , x
n
). We then need not distinguish between K and U and
x : U U is the identity mapping.
EXAMPLE let U = R
2
|negative x axis. With K =]0, r[]0, 2[ a system of polar coordinates
is dened by the dieomorphism K U, (r, ) (r cos , r sin ) = x(r, ).
The tangent space of U at x is dened by
T
x
U = x R
n
= (x, )[ R
n
. (1.10)
Through the denition a(x, ) + b(x, ) (x, a + b), a, b R,
T
x
U becomes a real ndimensional vector space. Note that each
x carries its own copy of R
n
; For x ,= y, T
x
U and T
y
U are to
be considered as independent vector spaces. Also, the space R
n
attached to x U should not be confused with the host space R
n
U in which U lives.
EXAMPLE Consider a curve : [0, 1] U. Monitoring its velocity, d
t
(t) , we obtain entries
(, ) T

U of tangent space. Even if U is small, the velocities may be arbitrarily large, i.e. the
space of velocities R
n
is really quite dierent from the base U (not to mention that the latter doesnt
carry a vector space structure.)
Taking the union of all tangent spaces, we obtain the socalled tangent bundle of U,
TU
_
xU
T
x
U = (x, )[x U, R
n
= U R
n
. (1.11)
Notice that TU R
2n
is open in R
2n
. It is therefore clear that we may contemplate dierentiable
mapping to and from the tangent bundle. Throughout, we will assume all mappings to be
suciently smooth to permit all derivative operations we will want to consider.
PHYSICS (M) Assume that U denes the set of n generalized coordinates of a mechanical system.
The Lagrangian function L(x, x) takes coordinates and generalized velocities as its arguments. Now,
we have just seen that (x, x) T
x
U, i.e.
1.2 Dierential forms in R
n
9
Figure 1.3 Visualizaton of a coordinate frame
The Lagrangian is a function on the tangent bundle TU of the
coordinate manifold U.
It is important not to get confused about the following point: for a given curve x = x(t), the
generalized derivatives x(t) are dependent variables and L(x, x) is determined by the coordinate
curves x(t). However, the function L as such is dened without reference to specic curves. The way
it is dened, it is a function of the 2n variables (x, ) T
x
U.
1.2.2 Vector elds and frames
A smooth mapping
v : U TU
x (x, (x)) v(x) (1.12)
is called a vector eld on U.
Consider a set of n vector elds b
i
. We call this set a frame (or nframe), if (x
U : b
i
(x) linear independent.) The n vectors b
i
(x) thus form a basis of T
x
U, smoothly
depending on x.
For example, a set of coordinates x
1
, . . . , x
n
parameterizing U induces a frame often denoted
(/x
1
, . . . , /x
n
) or (
1
, . . . ,
n
) for brevity. It is dened by the n vector elds

x
i

_
x,
x(x
1
, . . . , x
n
)
x
i
_
, i = 1, . . . , n.
Notice that the notation hints at a connection (vectors derivatives of functions), a point to
which we will return below.
Frames provide the natural language in which to formulate the concept of moving coordinate
systems pervasive in physical applications.
10 Exterior Calculus
EXAMPLE Let U R
2
be parameterized by cartesian coordinates x = (x
1
, x
2
). Then
1
=
(x, (1, 0)) and
2
= (x, (0, 1)), i.e.
1,2
are vector elds locally tangent to the axes of a cartesian
coordinate system.
Now consider the same set parameterized by polar coordinates, x = (r cos , r sin ). Then

r
= (x, (cos , sin )) and

= (x, r sin , r cos ) are tangent to the axes of a polar coordinate

system. (To obtain a proper orthonormal frame, we would need to normalize these vector elds (see
section 1.3.)
Every vector v(x) T
x
U may be decomposed with respect to a frame as
v(x) =

i
v
i
(x)b
i
(x),
and every vector eld v admits the global decomposition v =

i
v
i
b
i
where v
i
: U R are
smooth functions. Expansions of this type usually appear in connection with coordinate systems
(frames). For example, the curve (, ) may be represented as
(, ) =

i

i

x
i

,
where
i
=
i
(t) is the coordinate representation of the curve.
The connection between two frames b
i
and b

i
is given by
b

i
(x) =

i
(A
1T
)
j
i
(x)b
j
(x),
where A(x) GL(n) is a groupvalued smooth function frames transform covariantly. (Notice
that A
1
(x) is the inverse of the matrix A(x), and not the inverse of the function x A.)
Specically, the transformation between the coordinate frames of two systems x
i
and y
i

follows from
x
y
i
=

j
x
x
j
x
j
y
i
.
This means,

y
i
=
x
j
y
i

x
j
,
implying that (A
1
)
j
i
=
x
j
y
i
.
1.2.3 The tangent mapping
Let U R
n
and v R
m
be open subsets equipped with coordinates (x
1
, . . . , x
n
) and
(y
1
, . . . , y
m
), respectively. Further, let
F : U V
x y = F(x),
(x
1
, . . . , x
n
) (F
1
(x
1
, . . . , x
n
), . . . , F
m
(x
1
, . . . , x
n
)) (1.13)
1.2 Dierential forms in R
n
11
Figure 1.4 On the denition of the tangent mapping
be a smooth map. The map F gives rise to an induced map T
x
F : T
x
U T
F(x)
V of tangent
spaces which may be dened as follows: Let : [0, 1] U be a curve such that v = (x, ) =
((t), (t)). We then dene the tangent map through
T
x
Fv = (F((t)), d
t
(F )(t)) T
F(x)
V.
To make this a proper denition, we need to show that the r.h.s. is independent of the particular
choice of . To this end, recall that = d
t

i
x
i
= v
i
x
i
, where
i
are the coordinates of
the curve . Similarly, d
t
(F ) = d
t
(F )
j
y
j
. However, by the chain rule, d
t
(F )
j
=
F
j
x
i
d
t

i
=
F
j
x
i
v
i
. We thus have
v = v
i

x
i
, (T
x
F)v =
F
j
x
i
v
i

y
j
, (1.14)
which is manifestly independent of the representing curve. For xed x, the tangent map is a
linear map T
x
F : T
x
U T
F(x)
V . This is why we do not put its argument, v, in brackets, i.e.
T
x
Fv instead of T
x
F(v).
Given two maps U
F
V
G
W, we have T
x
(G F) = T
F(x)
G T
x
F. In the literature, the
map TF is alternatively denoted as F

.
1.2.4 Dierential forms
A dierential form of degree p (or pform, for short) is a map that assigns to every x U a
covariant tensor of degree p,
x

p
T
x
U. Given p vector elds, v
1
, . . . , v
p
,
x
(v
1
(x), . . . , v
p
(x))
is a real number which we require to depend smoothly on x. We denote by
p
U the set of all
pforms on U. (Not to be confused with the set of pforms of a single vector space discussed
above.) With the obvious linear structure,
(a +b)
x
a
x
+b
x
,
12 Exterior Calculus

p
U becomes an innite dimensional real vector space. (Notice that
0
U is the set of real
valued functions on U.) We nally dene
U
n

p=0

p
U,
the exterior algebra of forms. The wedge product of forms and the inner derivative with
respect to a vector eld are dened pointwise. They have the same algebraic properties as their
ancestors dened in the previous section and will be denoted by the same symbols.
INFO Dierential forms in physics why? In Physics, we tend to associate everything that comes
with a sense of magnitude and direction with a vector. From a computational point of view, this
identication is (mostly) o.k., conceptually, however, it may obscure the true identity of physical
objects. Many of our accustomed vector elds arent vector elds at all. And if they are not, they
are usually dierential forms in disguise.
Let us illustrate this point on examples: consider the concept of force, i.e. one of the most
elementary paradigms of a vector F in physics teaching. However, force isnt a vector at all! What
is more, the non-vectorial nature of force is not rooted in some abstract mathematical thinking but
rather in considerations close to experimentation! For force is measured by measuring the amount of
work, W, needed to move a test charge along a small curve segment. Now, locally, a curve can be
identied with its tangent vector , and work is a scalar. This means that force is something that
assigns to a vector ( ) a scalar (W). In other words, force is a one form. In physics, we often write
this as W = F t, where t is a reference time interval. This notation hints at the existence of a
scalar product. Indeed, a scalar product is required to identify the true force, a linear form in dual
space, with a vector. However, as long as we think about forces as dierential forms, no reference to
scalar products is needed.
Another example is current (density). In physics, we identify this with a vector eld, much like force
above. However, this isnt how currents are measured. Current densities are determined by xing a
small surface area in space, and counting the number of charges that pass through that area per unit
time. Now, a small surface may be represented as the parallelogram spanned by two vectors v, w,
i.e. current density is something that assigns to two vectors a number (of charges). In other words,
current density is a two-form. (It truly is a two form (rather than a general covariant 2-tensor)
for the area of the surface is given by the anti-symmetric combination v w. Think about this
point, or see below.) In electrodynamics we allow for time varying currents. It then becomes useful
to upgrade current to a three form, whose components determine the number of charges associated
to a space-time box spanned by a spatial area and a certain extension in time.
The simplest non-trivial forms are the dierential one-forms. Locally, a one-form maps tangent
vectors into the reals:
x
: T
x
U R. Thus,
x
is an element of the dual space of tangent
space, the so-called cotangent space, (T
x
U)

. For a given basis (b

1
(x), . . . , b
n
(x)) of T
x
M,
2
the cotangent space is spanned by (b
1
(x), . . . , b
n
(x)). The induced set of 1forms (b
1
, . . . , b
n
)
2
Depending on the context, we sometimes indicate the x-dependence of mathematical objects, F, by a bracket notation,

F(x)

,
and sometimes by subscripts

F

x
.
1.2 Dierential forms in R
n
13
denes a frame of the cotangent bundle, i.e. of the unication (TU)

x
(T
x
U)

. The
connection between a cotangent frame b
i
and another one b
i
is given by
b
i
(x) =
n

j=1

i
j
(x)b
j
(x),
where
1
(x) is the matrix (function) generating the change of frames b
i
b

i
.
For a given coordinate system x = x(x
1
, . . . , x
n
) consider the coordinate frame
i
. Its dual
frame will be denoted by dx
i
. The action of the dual frame is dened by
dx
i
_

x
j
_
=
i
j
. (1.15)
Under a change of coordinate systems, x
i
y
i
,
dx
i
dy
i
=

j
y
i
x
j
dx
j
, (1.16)
i.e. in this case,
i
j
=
y
i
x
j
. (Notice the mnemonic character of this formula.)
By analogy to (1.6), we expand a general pform as
=

i
1
,...,i
p

i
1
,...,i
p
b
i
1
b
i
p
, (1.17)
where
i
1
,...,i
p
(x) is a continuous function antisymmetric in its indices. Specically, for the
coordinate forms introduced in (1.15), this expansion assumes the form
=

i
1
,...,i
p

i
1
,...,i
p
dx
i
1
dx
i
p
. (1.18)
PHYSICS (M) Above, we had argued that the Lagrangian of a mechanical system should be
interpreted as a function L : TU R on the tangent bundle. But where is Hamiltonian mechanics
dened? The Hamiltonian is a function H = (q, p) of coordinates and momenta. The latter are dened
through the relation p
i
=
L(q, q)
q
i
. The positioning of the indices suggests, that the components
op
i
do not transform as the components of a vector. Indeed, consider a coordinate transformation
q
i
= q
i
(q). We have q
i
=
q
i
q
j
q
j
, which shows that velocities transform covariantly with the matrix
A
i
j
=
q
i
q
j
. The momenta, however, transform as
p

i
=
L(q(q

), q(q

, q

))
q
i
=
L(q(q

), q(q

, q

))
q
j
q
j
q
i
=
q
j
q
i
p
j
,
where we noted that q
j
=
q
j
q
i
q
i
implies
q
j
q
i
=
q
j
q
i
. This is covariant transformation behavior
under the matrix (A
1
)
i
j
=
q
j
q
i
. We are thus led to the conclusions that
The momenta p
i
of a mechanical system are the components of a covariant rank 1
tensor. Accordingly, the Hamiltonian H(q, p) is a function on the cotangent bundle TU

.
14 Exterior Calculus
We can push this interpretation a little bit further: recall (cf. info block on p ??) that a vector space
V can be canonically identied with its dual, V

, i V comes with a scalar product. This is relevant

to our present discussion, for
the Lagrangian of a mechanical systems dened a scalar product on tangent space, TU.
This follows from the fact that the kinetic energy T in L = T U usually has the structure T =
1
2
q
i
F
ij
(q) q
j
of a bilinear form in velocities. This gives p
i
= F
ij
q
j
, (where the q
j
on the r.h.s. needs
to be expressed as a function of q and p).
1.2.5 Pullback of dierential forms
Let F : U V be a smooth map as dened in (1.13). We may then dene a pullback operation
F

:
p
V

p
U

.
For
p
V

it is dened by
(F

)
x
(v
1
, . . . , v
p
) =
F(x)
(F

v
1
, . . . , F

v
p
).
For xed x, this operation reduces to the vector space pullback operation dened in section
1.1.6. In the specic case p = 0, is a function and F

= F. Straightforward generalization
of the identities discussed in section 1.1.6 obtains
F

is linear,
F

( ) = (F

) (F

),
(G F)

= F

.
Let y
j
and x
i
be coordinate systems of V and U, resp. (Notice that TU and TV need not
have the same dimensionality!). Then
(F

dy
j
)
_

x
i
_
= dy
j
_
F

_

x
i
__
= dy
j
_
F
l
x
i

y
l
_
=
F
j
x
i
,
or
(F

dy
j
) =
F
j
x
i
dx
i
. (1.19)
EXAMPLE In the particular case dimU = dimV = n,
3
and degree() = n (a topdimensional
form), can be written as
= gdy
1
dy
n
,
where g : V R is a smooth function. The pullback of is then given by
F

= (g F) det
_
F
x
_
dx
1
dx
n
,
3
By dimU we refer to the dimension of the embedding vector space R
n
U.
1.2 Dierential forms in R
n
15
where F/x is shorthand for the matrix |F
j
/x
i
. The appearance of a determinant signals that
topdimensional dierential forms have something to do with integration and volume.
1.2.6 Exterior derivative
The exterior derivative
d :
p
U
p+1
U,
d,
is a mapping that increases the degree of forms by one. It is one of the most important operations
in the calculus of dierential forms. To mention but one of its applications, the exterior derivative
encompasses the operations of vector analysis, div, grad, curl, and will be the key to understand
these operators in a coherent fashion.
To start with, consider the particular case of 0forms, i.e. smooth functions f : U R. The
1form df is dened as
df =

i
f
x
i
dx
i
. (1.20)
A number of remarks on this denition:
Eq. (1.20) is consistent with our earlier notation dx
j
for the duals of

x
f
. Indeed, for f = x
j
a coordinate function, dx
j
=
x
j
x
i
dx
i
= dx
j
.
Eq. (1.20) actually is a denition (is independent of the choice of coordinates.) For a dierent
coordinate system, y
i
,
df =
f
y
j
dy
j
=
f
x
i
x
i
y
j
y
j
x
k
dx
k
=
f
x
i
dx
i
,
where we used the chain rule and the transformation behavior of the dx
j
s Eq. (1.16).
The evaluation of df on a vector eld v = v
i
x
i
obtains
dfv =
f
x
i
v
i
,
i.e. the directional derivative of f in the direction identied by the vector components v
i
.
Specically, for v
i
= d
t

i
, dened by the velocity of a curve, dfd
t
=
f
x
i
d
t

i
= d
t
(f ).
The operation d dened by (1.20) obeys the Leibnitz rule,
d(fg) = dfg +gdf,
and is linear
d(af +bg) = adf +bdg, a, b = const. .
To generalize Eq. (1.20) to forms of arbitrary degree p, we need the following
Theorem: There is precisely one map d : U U with the following properties:
16 Exterior Calculus
d obeys the (graded) Leibnitz rule
d( ) = (d) + ()
p
(d), (1.21)
where p is the degree of .
d is linear
d(a +b) = ad +bd, a, b = const.
d is nilpotent, d
2
= 0.
d(
p
U)
p+1
U.
d(
0
U) is given by the denition above.
To prove this statement, we represent an arbitrary form
p
U as an unrestricted sum,
=
1
p!

i
1
,...,i
p

i
1
,...,i
p
dx
i
1
dx
i
p
.
Leibnitz rule and nilpotency then imply
d =
1
p!

i
1
,...,i
p
(d
i
1
,...,i
p
)dx
i
1
dx
i
p
, (1.22)
i.e. we have found a unique representation. However, it still needs to be shown that this repre-
sentation meets all the criteria listed above.
As for 1.), let
=
1
p!

i
1
,...,i
p

i
1
,...,i
p
dx
i
1
dx
i
p
, =
1
q!

i
1
,...,i
q

j
1
,...,j
q
dx
j
1
dx
j
q
.
Then,
d( ) =
1
p!q!

i
1
,...,i
p
j
1
,...,j
p
d(
i
1
,...,i
p

j
1
,...,j
q
)dx
i
1
dx
i
p
dx
j
1
dx
j
q
=
=
1
p!q!

i
1
,...,i
p
j
1
,...,j
p
_
(d
i
1
,...,i
p
)
j
1
,...,j
p
) +
i
1
,...,i
p
d
j
1
,...,j
q

dx
i
1
dx
i
p
dx
j
1
dx
j
q
=
= (d) + ()
p
d,
where we used the chain rule, and the relation ddx
i
1
dx
i
p
= ()
p
dx
i
1
dx
i
p
d.
Property 3. follows straightforwardly from the symmetry/antisymmetry of second derivatives

2
x
i
x
j

i
1
,...,i
p
/dierentials dx
i
dx
j
.
Importantly, pullback and exterior derivative commute,
F

d = d F

, (1.23)
for any smooth function F. (The proof amounts to a straightforward application of the chain
rule.)
1.2 Dierential forms in R
n
17
INFO Let us briey address the connection between exterior dierentiation and vector calculus
alluded to in the beginning of the section. The discussion below relies on an identication of co-
and contravariant components that makes sense only in cartesian coordinates. It is, thus, limited in
scope and merely meant to hint at a number of connections whose coordinate invariant (geometric)
meaning will be discussed somewhat further down.
Consider a oneform =
i
dx
i
, i.e. an object characterized by d (covariant) components. Its
exterior derivative can be written as
d =

i
x
j
dx
j
dx
i
=
1
2
_

i
x
j


j
x
i
_
dx
j
dx
i
.
Considering the case n = 3 and interpreting the coecients
i
, as components of a vector eld v
i
,
4
we are led to identify the three coecients of the form d as the components of the curl of |
i
.
Similarly, identifying the three components of the twoform =
ijk

i
dx
j
dx
k
(
ijk
is the fully
antisymmetric tensor) with a vector eld,
i
v
i
, we nd d =
_

x
i

i
_
dx
1
dx
2
dx
3
, i.e. the
threeform d is dened by the divergence v =
i
v
i
.
Finally, for a function , we have d =
i
dx
i
, i.e. (the above naive identication understood) a
vector eld whose components are dened by the gradient = |
i
. (For a coordinate invariant
representation of the vector dierential operators, we refer to section xx.)
Notice that in the present context relations such as v = 0 or f = 0 all follow from
the nilpotency of the exterior derivative, d
2
= 0.
PHYSICS (E) The dierential forms of electrodynamics. One of the objectives of the present
course is to formulate electrodynamics in a geometrically oriented manner. Why this geometric
approach? If one is primarily interested in applied electrodynamics (i.e. the solution of Maxwells
equations in a concrete setting), the geometric formulation isnt of that much value: calculations are
generally formulated in a specic coordinate system, and once one is at this stage, its all down to the
solution of dierential equations. The strengths of the geometric approach are more of conceptual
nature. Specically, it will enable us to
understand the framework of physical foundations needed to formulate electrodynamics. Do we
need a metric (the notion of distances) to formulate electrodynamics? How does the structure of
electrodynamics emerge from the condition of relativistic invariance? These questions and others
are best addressed in a geometry oriented framework.
formulate the equations of electrodynamics in a very concise, and easy to memorize manner. This
compact formulation is not only of aesthetic value, rather, it connects to the third and perhaps
most important aspect:
understand electrodynamics as a representative of a family of theories known as gauge theories.
The interpretation of electrodynamics as a gauge theory has paved the way to one of the most
rewarding and important developments in modern theoretical physics, the understanding of funda-
mental forces electromagnetism, weak and strong interactions, and gravity as part of one unifying
scheme.
We will here formulate the geometric view of electrodynamics in a bottom up approach. That is,
we will introduce the basic objects of the theory, elds, currents, etc. rst on a purely formal level.
The actual meaning of the denitions will then gradually get disclosed as we go along.
4
The naive identication
i
v
i
does not make much sense, unless we are working in cartesian coordinates. Thats why the
expressions derived here hold only in cartesian frames.
18 Exterior Calculus
Electrodynamics is formulated in (4 = 3 + 1)dimensional space, three space dimensions and one
time dimension. For the time being, we identify this space with R
4
. Unless stated otherwise, cartesian
coordinates (x
0
, x
1
, x
2
, x
3
, x
4
) will be understood, where x
0
= ct measures time, and c is the speed
of light.
Let us begin by introducing the sources of the theory: we dene the current 3-form, j
3
R
4
,
j =
1
3!

dx

dx

dx

, (1.24)
where

is the fully antisymmetric tensor in four dimensions.

5
We may rewrite this as
j =j
0
dx
1
dx
2
dx
3

j
1
dx
2
dx
3
dx
0

j
2
dx
3
dx
1
dx
0

j
3
dx
1
dx
2
dx
0
.
We tentatively identify j
1,2,3
with the vectorial components of the
current density familiar from electrodynamics. This means that, say, j
1
will be the number of charges owing through surface elements in the
(23)-plane during a given time interval. Comparing with our heuristic
discussion in the beginning of the chapter, we have extended the deni-
tion by a dynamical component, i.e. j
1
actually measures the number
of charges associated with a space-time box spanned by a spatial area
in the (23) plane and a stretch in the 0-direction (time.) To be more pre-
cise, take a triplet of vectors (x
2
e
2
, x
3
e
3
, x
0
e
0
), anchored at a space time point (x
0
, x
1
, x
2
, x
3
).
The value obtained by evaluating the current form on these arguments, j(x
2
e
2
, e
3
e
3
, cte
0
),
then is the number of charges passing through the area x
2
x
3
at (x
1
, x
2
, x
3
) during time interval
[t, t + t], where t = x
0
/c (see the gure, where the solid arrows represent the time line of
particles passing through the shaded spatial area in time t.) Similarly, j
0
will be interpreted as (c
times) the number of charges in a spatial box spanned by the (1, 2, 3)coordinates. Thus, j
0
= c,
where is the physical charge density.
On physical grounds, we need to impose a continuity equation,

=
t
+

3
i=1

x
i
j
i
=

= 0.
6
This condition is equivalent to
the vanishing of the exterior derivative,
dj = 0. (1.25)
To check this, compute
dj =
1
3!

dx

dx

dx

dx

=
=
1
3!

dx
0
dx
1
dx
2
dx
3
=

dx
0
dx
1
dx
2
dx
3
,
where the second equality follows from the antisymmetry of the wedge product and the third from
the properties of the antisymmetric tensor (see below),

= 3!

.
5
I.e.

vanishes if any two of its indices are identical. Otherwise, it equals the sign of the permutation (0, 1, 2, 3)
(, , , ).
6
Following standard conventions, indices i, j, k, . . . are spatial and run from 1 to 3, while , , are space-time indices
running from zero to three.
1.2 Dierential forms in R
n
19
Figure 1.5 Cartoon on the measurement prescriptions for electric elds (left) and magnetic elds
(right). Discussion, see text.
Next in our list of denitions are the electric and magnetic elds. The electric eld bears similarity
to a force. At least, it is measured like a force is, that is by displacement of a test particle along (small)
stretches in space and recording the corresponding work. Much like a force, the electric eld is something
that converts a vector (innitesimal curve segment) into a number (work). We thus describe the eld
in terms of a one-form
E = E
i
dx
i
, (1.26)
where E
i
= E
i
(x, t) are the time dependent coecients of the eld strength. In cartesian coordinates
(and assuming the standard scalar product), these can be identied with the components of the electric
eld vector. The magnetic eld, in contrast, is not measured like a force. Rather, one measures the
magnetic ux threading spatial areas (e.g. by measuring the torque exerted on current loops.) In analogy
to our discussion on page ?? we thus dene the magnetic eld as a two-form
B = B
1
dx
2
dx
3
+B
2
dx
3
dx
1
+B
3
dx
1
dx
2
. (1.27)
Attentive readers may wonder how this expression ts into our previous discussion of coordinate repre-
sentations of dierential forms: as it is written, B is not of the canonical form B = B
ij
dx
i
dx
j
. On
a similar note, one may ask how the coecients B
i
above relate to the familiar magnetic eld vector
B
v
of electrodynamics (the subscript v serves to distinguish B
v
from the dierential form.) The rst
thing to notice is that B
v
is not a conventional vector, rather it is a pseudovector, or axial vector. A
pseudovector is a vector that transforms conventionally under spatial rotation. However, under parity
non-conserving operations (reection, for example), it transforms like a vector, plus it changes sign (see
Fig. 1.6.) Usually, pseudovectors obtain by taking the cross product of conventional vectors. Familiar
examples include angular momentum, l, (l = r p) and the magnetic eld, B
v
.
To understand how this relates to the two-form introduced above, let us rewrite the latter as B =
1
2
B
i

ijk
dx
i
dx
k
, where B
i
transforms contravariantly (like a vector.) Comparison with B shows that,
e.g., B
1
= B
1

123
=

B
1
. I.e. in a given basis, B
1
= B
1
and we may identify these with the components
of the eld (pseudo)vector B
v
. However, under a basis change, B
i
A
i
j
B
j
transforms like a vector,
while (check it!)
ijk

ijk
det(A). This means B
i
A
i
j
B
j
det(A). For rotations, det A = 1 and
|B
i
transforms like a vector. However, for reections det A = 1, we pick up an additional sign
change. Thus, |B
i
transforms like a pseudovector, and may be identied with the magnetic eld
strength (pseudo)vector B
v
. The distinction between pseudo- and conventional vectors can be avoided,
if we identify B with what it actually is, a dierential two-form.
20 Exterior Calculus
Figure 1.6 On the axial nature of the magnetic eld. Spatial reection at a plane leads to reection
of the eld vector plus a sign change. In contrast, ordinary vectors (such as the current density
generating the eld) just reect.
We combine the components of the electric eld, E, and the magnetic eld, B, into the socalled
eld strength tensor,
F E dx
0
+B F

dx

dx

. (1.28)
Comparison with the component representations in (1.26) and (1.27) shows that
|F

=
_
_
_
0 E
1
E
2
E
3
E
1
0 B
3
B
2
E
2
B
3
0 B
1
E
3
B
2
B
1
0
_
_
_. (1.29)
It is straightforward to verify that the homogeneous Maxwell equations, commonly written as
(CGS units)
B = 0,
1
c

t
B +E = 0,
are equivalent to the closedness of the eld strength tensor,
dF = 0. (1.30)
We next turn to the discussion of the two partner elds of E and B, the electric D-eld and the
magnetic H-eld, respectively. In vacuum, these are usually identied with E and B, i.e. E = D
and B = H, resp. However, in general, the elds are dierent and this shows quite explicitly in
the present formalism. In principle, D and H can be introduced with reference to a measurement
prescription, much like we did above with E and B. However, this discussion would lead us too far
astray and we here introduce these elds pragmatically. That is, we require that D and H satisfy a
dierential equation whose component-representation equals the inhomogeneous Maxwell equations.
To this end, we dene the two form D by
D = D
1
dx
2
dx
3
+D
2
dx
3
dx
1
+D
3
dx
1
dx
2
, (1.31)
similar in structure to the magnetic (!) form B. The eld H is dened as a one-form,
H = H
i
dx
i
. (1.32)
1.2 Dierential forms in R
n
21
The components |D
i
and |H
i
appearing in these denitions are two be identied with the vector
components in the standard theory where, again, the appearance of covariant indices indicates that
the vectorial interpretation becomes problematic in non-metric environments. We now dene the
dierential two-form
G H dt +D = G

dx

dx

, (1.33)
with component representation
|G

=
_
_
_
0 H
1
H
2
H
3
H
1
0 D
3
D
2
H
2
D
3
0 D
1
H
3
D
2
D
1
0
_
_
_. (1.34)
With these denitions it is straightforward to verify that the inhomogeneous Maxwell equations
assume the form
dG = j. (1.35)
What the present discussion does not tell us is how the two main players in the theory, the covariant
tensors F and G are connected to each other. To establish this connection, we need additional
structure, viz. a metric, and this is a point to which we will return below.
To conclude this section let us briey address the coupling between electromagnetic elds and matter.
This coupling is mediated by the Lorentz force acting on charges q, which, in standard notation, assumes
the form F = q(E + v B). Here, v is the velocity vector, and all other quantities are vectorial, too.
Translating to forms, this reads
F = q(E i
v
B), (1.36)
where F is the force one-form.
INFO The general fully antisymmetric tensor or Levi-Civita symbol or -tensor is a mixed tensor
dened by

1
,...,
n

1
,...,
n
=
_
_
_
+1 , (
1
, . . . ,
n
) an even permutation of (
1
, . . . ,
n
),
1 , (
1
, . . . ,
n
) an odd permutation of (
1
, . . . ,
n
),
0 , else.
(1.37)
In the particular case (
1
, . . . ,
n
) = (1, . . . , n) one abbreviates the notation to
1,...,n

1
,...,
n

1
,...,
n
. Similarly,

1
,...,
n
1,...,n

1
,...,
n
. Important identities fullled by the tensor include

1
,...,
n

1
,...,
n
= n! and
,
2
,...,
n

,
2
,...,
n
= (n 1)!

.
1.2.7 Poincare Lemma
Forms which are annihilated by the exterior derivative, d = 0, are called closed. For example,
every nform dened on U R
n
is closed. Also, forms = d that can be written as exterior
derivatives of another form so called exact forms are closed. One may wonder whether
every closed form is exact, d = 0
?
= d. The answer to this question is negative; in general
closedness does not imply exactness.
22 Exterior Calculus
EXAMPLE On U = R
2
|0 consider the form
=
xdy ydx
x
2
+y
2
.
It is straightforward to verify that d = 0. Nonetheless, is not exact: one may verify that =
d arctan(y/x), everywhere where the dening function exists. However the function arctan(y/x) is
illdened on the positive xaxis, i.e. can not be represented as the exterior derivative of a smooth
function on the entire domain of denition of . (A more direct way to see this is to notice that on
R
2
|0, = d, where is the polar angle. The latter jumps at the positive xaxis by 2, i.e. it
is not well dened on all of U.)
The conditions under which a closed form is exact are stated by the
Theorem (Poincare Lemma): On a starshaped
7
open subset U R
n
a form
p
U is
exact if and only if it is closed.
The Lemma is proven by explicit construction. We here restrict ourselves to the case of
oneforms,
1
U and assume that the reference point establishing the starshapedness of
U is at the origin. Let the oneform =
i
dx
i
be closed. Then, the zeroform (function)
f(x) =
_
1
0
dt

i
(tx)x
i
satises the equation df = , i.e. we have shown that is exact.
(Notice that the existence of the integration path tx, t [0, 1] relies on the starshapedness of
the domain.) Indeed,
df =
_
1
0
dt
_
t

i
(y)
y
j

y=tx
dx
j
x
i
+
i
(tx)dx
i
_
=
=
_
1
0
dt
_
td
t

i
(tx)dx
i
+
i
(tx)dx
i
+t
_

i
y
j


j
y
i
_
. .
0 (d=0)
dx
j
dx
i
_
=
=
i
(x)dx
i
,
where in the last line we have integrated by parts. (Notice that
_
1
0
dt
i
(tx)x
i
=
_
1
0
dt
i
(tx)d
t
(tx) =
_
dx
i

i
coincides with the standard vectoranalysis denition of the lineintegral of the irro-
tational eld
i
along the straight line from the origin to x. The proof of the Lemma for
forms of higher degree p > 1 is similar.
INFO The above example, and the proof of the Lemma suggest a connection (exactness geom-
etry). This connection is the subject of cohomology theory.
1.2.8 Integration of forms
Orientation of open subsets U R
n
We generalize the concept of orientation introduced in section 1.1.7 to open subsets U R
n
.
Consider a nowhere vanishing form
n
U, i.e. a form such that for any frame (b
1
, . . . , b
n
),
7
A subset U R
n
is starshaped if there is a point x U such that any other y U is connected to x by a straight line.
1.2 Dierential forms in R
n
23
x U :
x
(b
1
(x), . . . , b
n
(x)) ,= 0. We call the frame (b
1
, . . . , b
n
) oriented if
x U,
x
(b
1
(x), . . . , b
n
(x)) > 0. (1.38)
A set of coordinates (x
1
, . . . , x
n
) is called oriented if the associated frames (/
x
1, . . . , /
x
n)
are oriented.
Conversely, an orientation of U may be introduced by declaring
= dx
1
dx
n
to be an orienting form. Finally, two forms
1
and
2
dene the same orientation i they dier
by a positive function
1
= f
2
.
Integration of nforms
Let K U be a suciently (in the sense that all regular integrals we are going to consider exist)
regular subset. Let (x
1
, . . . , x
n
) be an oriented coordinate system on U. An arbitrary nform

n
U may then be written as = f dx
1
dx
n
, where f is a smooth function given by
f(x) =
x
_

x
1
, . . . ,

x
n
_
.
The integral of the nform is then dened as
_
K

_
K
f(x) dx
1
. . . dx
n
, = f(x) dx
1
dx
n
, (1.39)
where the notation
_
K
(. . . )dx
1
. . . dx
n
(no wedges between dierentials) refers to the integral
of standard calculus over the domain of coordinates spanning K.
Under a change of coordinates, (x
1
, . . . , x
n
) (y
1
, . . . , y
n
), the integral changes as
_
K,x
sgn det
_
x
i
y
j
_ _
K,y
,
where
_
K,x
is shorthand for the evaluation of the integral in the coordinate representation x.
The sign factor sgn det(x
i
/y
j
) has to be read as
sgn det
_
x
i
y
j
_
=

det
_
x
i
y
j
_

det
_
x
i
y
j
_ ,
where the modulus of the determinant in the numerator comes from the variable change in the
standard integral and the determinant in the denominator reects is due to = f(x)dx
1

dx
n
= f(x(y)) det(x
i
/y
j
) dy
1
dy
n
. This results tells us that (a) the integral is
invariant under an orientation preserving change of coordinates (the denition of the integral
canonical) while (b) it changes sign under a change of orientation.
Let F : V U, y F(y) x be a dieomorphism, i.e. a bijective map such that F and
F
1
are smooth. We also assume that F is orientation preserving, i.e. that dy
1
dy
n

n
V
24 Exterior Calculus
Figure 1.7 On the denition of oriented p-dimensional surfaces embedded in R
n
and F

(dx
1
dx
n
)
n
V dene the same orientation. This is equivalent to the condition
(why?) det(dx
i
/dy
j
) > 0. We then have
_
F
1
(K)
F

=
_
K
. (1.40)
The proof follows from F

(f(x) dx
1
dx
n
) = det(dx
i
/dy
j
)f(x(y)) dy
1
dy
n
, the
denition of the integral Eq. (1.39), and the standard calculus rules for variable changes.
Integration of pforms
A pdimensional oriented surface T R
n
is dened by a smooth parameter representation
Q : K T, (
1
, . . . ,
p
) Q(
1
, . . . ,
p
) where K V R
p
, and V is an oriented open
subset of R
p
, i.e. a subset equipped with an oriented coordinate system.
The integral of a pform over the surface T is dened by
_
T
=
_
K
Q

. (1.41)
On the rhs we have an integral over a pform in pdimensional space to which we may apply
the denition (1.39). If p = n, we may choose a parameter representation such that V = U,
K = T and Q =inclusion map. With this choice, Eq. (1.41) reduces to the ndimensional case
discussed above. Also, a change of parameters is given by an orientation preserving map in
parameter space V . To this parameter change we may apply our analysis above (identication
p = n understood). This shows that the denition (1.41) is parameter independent.
Examples:
For p = 1, T is a curve in R
n
. The integral of a one form =
i
dx
i
along T is dened by
_
T
=
_
[0,1]
Q

=
_
[0,1]

i
(Q())
Q
i

d, (1.42)
where we assumed a parameter representation Q : [0, 1] T, Q(). The last expres-
sion conforms with the standard calculus denition of line integrals provided we identify the
components of the form
i
with a vector eld.
1.2 Dierential forms in R
n
25
For p = 2 and n = 3 we have a surface in threedimensional space. Assuming a parameter
representation Q : [a, b] [c, d] T, (
1
,
2
) Q(
1
,
2
), and a form = v
1
dx
2
dx
3
+
v
2
dx
3
dx
1
+v
3
dx
1
dx
2
=
1
2

ijk
v
i
dx
j
dx
k
, we get
_
T
=
_
[a,b][c,d]
Q

= (1.43)
=
1
2
_
[a,b][c,d]

ijk
v
i
(Q(
1
,
2
))
_
Q
j

1
Q
k

2

Q
j

2
Q
k

1
_
d
1
d
2
=
=
_
[a,b][c,d]

ijk
v
i
(Q(
1
,
2
))
Q
j

1
Q
k

2
d
1
d
2
. (1.44)
In the standard notation of calculus, this would read
_
d
1
d
2
v (

1Q

2Q) =
_
dS n v,
where dS n =

1Q

2Q is surface element normal vector eld.

Cells and chains
A pcell of in R
p
is a triple = (D, Q, Or) consisting of (cf. Fig. 1.8)
a convex polyhedron D R
p
,
a dierentiable mapping (parameterization) Q : D K R
n
, and
an orientation (denoted Or) of R
p
.
Using the language of cells, our previous denition of the integral of a pform assumes the form
_

=
_
D
Q

,
where we have written
_

(instead of
_
Q(D)
for the integral over the cell. The pcell diering
from in the choice of orientation is denoted . If no confusion is possible, we will designate
cells = (D, Q, Or) just by reference to their basepolyhedron D, or to their image Q(D).
(E.g. what we mean when we speak of the 1cell [a, b] is (i) the interval [a, b] plus (ii) the
image Q([a, b]) with (iii) some choice of orientation.)
EXAMPLE Let be the unitcircle, S
1
in twodimensional space. We may represent by a 1cell
in R
2
whose base polyhedron is the interval I = [0, 2] of (oriented) R
1
and the map Q : [0, 2]
R
2
, t (cos t, sin t) The integral of the 1form = x
1
dx
2

1
R
2
over then evaluates to
_

=
_
I
Q

=
_
I
cos t d(sin t) =
_
2
0
cos
2
t dt = .
Now, let be the unitdisk, D
2
in two dimensional space. We describe by
Q : D [0, 1] [0, 2] D
2
,
(r, ) (r cos , r sin ).
The integral of the oneform d(x
1
dx
2
) = dx
1
dx
2
over is given by
_
D
2
dx
1
dx
2
=
_
D
Q

(dx
1
dx
2
) =
_
det
_
d(x
1
, x
2
)
d(r, )
_
. .
r
dr d =
_
1
0
dr
_
2
0
r = .
We thus observe
_
S
1
x
1
dx
2
=
_
D
2
d(x
1
x
2
), a manifestation of Stokes theorem ...
26 Exterior Calculus
Figure 1.8 On the concept of oriented cells and their boundaries. Discussion, see text
As a generalization of a single cell, we introduce chains. A pchain is a formal sum
c = m
1

1
+ +m
r

r
,
where
i
are pcells and the multiplicities m
i
Z integer valued coecients. Introducing the
natural identications
m
1

1
+m
2

2
= m
2

2
+m
1

1
,
0 = 0,
c + 0 = c,
m
1
+m
2
= (m
1
+m
2
),
(m
1

1
+m
2

2
) + (m

1
+m

2
) = m
1

1
+m
2

2
+m

1
+m

2
,
the space of pchains, C
p
, becomes a linear space.
8
Let be a pcell. Its boundary is a (p 1)chain in which may be dened as follows:
Consider the p 1 dimensional faces D
i
of the polyhedron D underlying . The mappings
Q
i
: D
i
R
n
are the restrictions of the parent map Q : D R
n
to the faces D
i
. The faces
D
i
inherit their orientation from that of D. To see this, let (e
1
, . . . , e
p
) be a positively oriented
frame of D. At a point x
i
D
i
consider a vector n normal and outwardly directed (w.r.t. the
bulk of D.) A frame (f
1
, . . . , f
p1
) is positively oriented, if (n, f
1
, f
p1
) is oriented in the same
way as (e
1
, . . . , e
p
). We thus dene
9
=

i
,
where
i
= (D
i
, Q

D
i
, Or
i
) and Or
i
is the induced orientation. The collection of faces, D
i
,
denes the boundary of the base polyhedron D:
D =

i
D
i
.
8
Strictly speaking, the integervaluedness of the coecients of elementary cells implies that C
p
is an abelian group (rather
than the real vector space we would have obtained were the coecients arbitrary.)
9
These denitions work for (p > 1)cells. To include the case p = 1 into our denition, we agree that a 0chain is a collection
of points with multilplicities. The boundary of a 1cell dened in terms of a line segment

AB from a point A to B is
B A.
1.2 Dierential forms in R
n
27
The boundary of a pchain =

i
m
i

i
is dened as =

i
m
i

i
. We may, thus,
interpret as a linear operator, the boundary operator mapping p chains onto p 1 chains:
: C
p
C
p1
. (1.45)
Let
p
U be a pform and =

m
i

i
C
p
be a pchain. The integral of over is
dened as
_

i
m
i
_

i
. (1.46)
1.2.9 Stokes theorem
Theorem (Stokes): Let C
p+1
be an arbitrary (p + 1)chain and
p
U be an arbitrary
pform on U. Then,
_

=
_

d. (1.47)
Examples
Before proving this theorem, we go through a number of examples. (For simplicity, we assume
that = (D, Q, Or) is an elementary cell, where D = [0, 1]
p+1
is a (k + 1)dimensional unit
cube and Q(D) U lies in an open subset U R
n
. We assume cartesian coordinates on
U.)
p = 0, n = 1: Consider the case where Q : D Q(D) is just the inclusion mapping embedding
the line segment D = [0, 1] into R. The boundary D = 1 0 is a zerochain containing the
two terminating points 1 and 0. Then,
_
[0,1]
= (1) (0) and
_
[0,1]
d =
_
[0,1]

x
dx,
where is any 0form (function), i.e. Stokes theorem reduces to the well known integral
formula of onedimensional calculus.
p = 0 arbitrary n: The cell = ([0, 1], Q, Or) denes a curve = Q([0, 1]) in R
n
. Stokes
theorem assumes the form
(Q(1)) (Q(0)) =
_

d =
_
[1,0]
d( Q) =
_
1
0

x
i

Q(s)

Q
i
(s)ds,
i.e. it relates the line integral of a gradient eld
i
to the value of the potential at the
boundary points.
p = 1, n = 3 : The cell = ([0, 1] [0, 1], Q, Or) represents a smooth surface embedded into
R
3
. With =
i
dx
i
, the integral
_

=
_

i
dx
i
=
_
[0,1]
2

i
(Q())
Q
i

d
28 Exterior Calculus
reduces to the line integral over the boundary ([0, 1]
2
) = [0, 1] 0 +1 [0, 1] [0, 1]
1 0 [0, 1] of the square [0, 1]
2
(cf. Eq. (1.42)). With d
i
dx
i
=

i
x
j
dx
j
dx
i
, we
have (cf. Eq. (1.43))
_

d =
_

i
x
j
dx
j
dx
i
=
_
[0,1]
2

kji
( Q)
j
x
j

klm
Q
l

1
Q
m
d
2
d
1
d
2
.
Note that (in the conventional notation of vector calculus)
kji
i
x
j
= ( )
k
, i.e. we
rediscover the formula
_

ds =
_

dS (),
(known in calculus as Stokes law.)
p = 2, n = 3 : The cell = ([0, 1]
3
, Q, Or) denes a three dimensional volume Q([0, 1]
3
) in
threedimensional space. Its boundary is a smooth surface. Let
1
2

ij
dx
i
dx
j
be a
twoform. Its exterior derivative is given by d =
1
2

kij

ij
x
k
dx
k
dx
i
dx
j
and the l.h.s. of
Stokes theorem assumes the form
_

=
1
2
_

ij
dx
i
dx
j
=
_
([0,1]
3
)
( Q)
ij
Q
i

1
Q
j

2
d
1
d
2
.
The r.h.s. is given by
_

d =
1
2
_

kij

ij
x
k
dx
k
dx
i
dx
j
=
1
2
_
[0,1]
3

kij
( Q)
ij
x
k
det
_
Q

_
d
1
d
2
d
3
.
Identifying the three independent components of with a vector according to
ij
=
ijk
v
k
,
we have
1
2

kij

ij
x
k
= v and Stokes theorem reduces to Gau law of calculus,
_

dS v =
_

dV ( v).
Proof of Stokes theorem
Thanks to Eq. (1.46) it suces to prove Stokes theorem for individual cells.
To start with, let us assume that the cell = ([0, 1]
p+1
, Q, Or) has a unit(p+1)cube as its
base. (Later on, we will relax that assumption.) Consider [0, 1]
p+1
to be dissected into N 1
small cubes r
i
of volume 1/N 1. Then, =

i
where
i
= (r
i
, Q, Or) and =

i
.
Since
_

d =

i
_

i
d and
_

=
_

i
, it is sucient to prove Stokes theorem for the
small microcells.
Without loss of generality, we consider the cornercube r
1
([0, ]
p+1
, Q, Or), where =
N

1
p+1
. For (notational) simplicity, we set p = 1; the proof for general p is absolutely analogous.
By denition,
_
r
1
d =
_
[0,]
2
Q

d =
_
[0,]
2
dQ

.
1.3 Metric 29
Assuming that the 1form is given by =
i
dx
i
, we have
dQ

= dQ

(
i
dx
i
) = d(( Q)
i
d(x
i
Q
. .
Q
i
)) = d( Q)
i
dQ
i
=
=
_
( Q)
i

1
Q
i

2

( Q)
i

2
Q
i

1
_
d
1
d
2
.
We thus have
_
[0,]
2
dQ

=
_

0
d
1
d
2
_
( Q)
i

1
Q
i

2

( Q)
i

2
Q
i

1
_
=
=
2
_
( Q)
i

1
Q
i

2

( Q)
i

2
Q
i

1
_
(0, 0) +O(
3
).
We want to relate this expression to
_
r
1
=
_
[0,]
2
Q

=
_
[0,]
2
( Q)
i
dQ
i
. Considering
the rst of the four stretches contributing to the boundary [0, ]
2
= [0, ] 0+[0, ]
[0, ] 0 [0, ] we have
_

0
_
( Q)
i
Q
i

1
_
(
1
, 0) d
1

_
( Q)
i
Q
i

1
_
(0, 0) +O(
2
).
Evaluating the three other contributions in the same manner and adding up, we obtain
_
r
1
=
__
( Q)
i
Q
i

1
_
(0, 0) +
_
( Q)
i
Q
i

2
_
(0, )

_
( Q)
i
Q
i

1
_
(0, )
_
( Q)
i
Q
i

2
_
(0, 0)
_
+O(
2
) =
=
2
_

2
_
( Q)
i
Q
i

1
_

1
_
( Q)
i
Q
i

2
__
(0, 0) +O(
3
) =
=
2
_
( Q)
i

1
Q
i

2

( Q)
i

2
Q
i

1
_
(0, 0) +O(
3
),
i.e. the same expression as above. This proves Stokes theorem for an individual microcell and,
thus, (see the argument given above) for a dcube.
INFO The proof of the general case proceeds as follows: A dsimplex is the volume spanned by
d + 1 linearly independent points (a line in one dimension, a triangle in two dimensions, a tetrad in
three dimensions, etc.) One may show that (a) a dcube may be dieomorphically mapped onto a
dsimplex. This implies that Stokes theorem holds for cells whose underlying base polyhedron is a
simplex. Finally, one may show that (b) any polyhedron may be decomposed into simplices, i.e. is
a chain whose elementary cells are simplices. As Stokes theorem trivially carries over from cells to
chains one has, thus, proven it for the general case.
1.3 Metric
In our so far discussion, notions like length or distances concepts of paramount importance
in elementary geometry where not an issue. Yet, the moment we want to say something about
30 Exterior Calculus
the actual shape of geometric structures, means to measure length have to be introduced. In
this section, we will introduce the necessary mathematical background and various follow up
concepts built on it. Specically, we will reformulate the standard operations of vector analysis
in a manner not tied to specic coordinate systems. For example, we will learn to recognize the
allinfamous expression of the Laplace operator in spherical coordinates as a special case of a
structure that not dicult to conceptualize.
1.3.1 Reminder: Metric on vector spaces
As before, V is an ndimensional R-vector space. A (pseudo)metric on V is a bilinear form
g : V V R,
(v, w) g(v, w) v, ,
which is symmetric, g(v, w) = g(w, v) and nondegenerate: w V, g(v, w) = 0 v = 0.
Occasionally, we will use the notation (V, g) to designate the pair (vector space,its metric).
Due to its linearity, the full information on g is stored in its value on the basis vectors e
i
,
g
ij
g(e
i
, e
j
).
(Indeed, g may be represented as g = g
ij
e
i
e
j
T
0
2
(V ).) The matrix g
ij
is symmetric
and has nonvanishing determinant, det(g
ij
) ,= 0. Under a change of basis, e

i
= (A
1
)
j
i
e
j
it
transforms covariantly, g

ii
= (A
1
)
j
i
(A
1
)
j

i
g
i

j
, or
g

ii
= (A
1T
gA
1
)
ii
.
Being a symmetric bilinear form, the metric may be diagonalized, i.e. a basis e
i
exists, wherein
g
ij
= g( e
i
, e
j
)
ij
. Introducing orthonormalized basis vectors by
i
= e
i
/(g( e
i
, e
i
)
1/2
), the
matrix representing g assumes the form
g = diag(1, . . . , 1
. .
r
, 1, . . . , 1
. .
nr
), (1.48)
where we have ordered the basis according to the signature of the eigenvalues. The set of trans-
formations A leaving the metric forminvariant, A
1T
A
1
= denes the (pseudo)orthogonal
group, O(r, nr). The dierence, 2rn between the number of positive and negative eigenval-
ues is called the signature of the metric. (According to the theorem of Sylvester), the signature
is invariant under changes of basis. A metric with signature n is called positive denite.
Finally, a metric may be dened by choosing any basis and declaring it to be orthonormal,
g(
i
,
j
)
ij
.
1.3.2 Induced metric on dual space
Let
i
be an orthonormal basis of V and
i
be the corresponding dual basis. We dene a
metric

g
on V

by requiring

g
(
i
,
j
)
ij
. Here,
ij
is the inverse of the matrix
ij
.
10
10
The distinction has only notational signicance: {
ij
} is self inverse, i.e.
ij
=
ij
.
1.3 Metric 31
Now, let e
i
be an arbitrary oriented basis and e
i
its dual. The matrix elements of the metric
and the dual metric are dened as, respectively, g
ij
g(e
i
, e
j
) and g
ij

g
(e
i
, e
j
). (Mind the
position of the indices!) By construction, one is inverse to the other,
g
ik
g
kj
=
j
i
. (1.49)
Canonical isomorphism V V

While V and V

are isomorphic to each other (by virtue of the mapping e

i
e
i
) the isomorphy
between the two spaces is, in general, not canonical; it relies on the choice of the basis e
i
.
However, for a vector space with a metric g, there is a basisinvariant connection to dual
space: to each vector v, we may assign a dual vector v

by requiring that w V : v

(w)
!
=
g(v, w). We thus have a canonical mapping:
J : V V

,
v v

g(v, . ). (1.50)
In physics, this mapping is called raising or lowering of indices: For a basis of vectors e
i
, we
have
J(e
i
) = g
ij
e
j
For a vector v = v
i
e
i
, the components of the corresponding dual vector J(v) v
i
e
i
obtain as
v
i
= g
ij
v
j
, i.e. again by lowering indices.
Volume form
If
i
is an oriented orthonormal basis, the nform

1

n
(1.51)
is called a volume form. Its value (v
1
, . . . , v
n
) is the volume of the parallel epiped spanned
by the vectors v
1
, . . . , v
n
.
As a second important application of the general basis, we derive a general representation
of the volume form. Let the mapping
i
e
i
be given by
i
= (A
1
)
j
i
e
j
. Then, we have
the component representation of the metric

ij
= A
i
k
g
kl
(A
T
)
j
l
. (1.52)
Taking the determinant of this relation, using that det A > 0 (orientation!), and noting that
detg
ij
/ det
ij
= [ detg
ij
[, we obtain the relation
det A = [g[
1/2
, (1.53)
where we introduced the notation
g detg
ij
= (detg
ij
)
1
.
32 Exterior Calculus
At the same time, we know that the representation of the volume form in the new basis reads
as = det Ae
1
e
n
, or
= [g[
1/2
e
1
e
n
. (1.54)
1.3.3 Hodge star
We begin this section with the observation that the two vector spaces
p
V

and
np
V

have
the same dimensionality
dim
p
V

= dim
np
V

=
_
n
p
_
.
By virtue of the metric, a canonical isomorphism between the two may be constructed. This
mapping, the socalled Hodge star is constructed as follows: Starting from the elementary scalar
product of 1forms, g
ij
=

g
(e
i
, e
j
) e
i
, e
j
, a scalar product of pforms may be dened by
e
i
1
e
i
p
, e
j
1
,
e
j
p
dete
i
k
, e
j
l
) = detg
i
k
j
l
.
Since every form
p
V

may be obtained by linear combination of basisforms e

i
1
,
e
i
p
this formula indeed denes a scalar product on all of
p
V

. Indeed, noting that for any matrix

X = X
ij
, det X =
1,...,n
i
1
,...,i
n
X
i
1
1
. . . X
i
n
n
, i.e.
detg
i
k
j
l
=
i
1
,...,i
p
k
1
,...,k
p
g
k
1
j
1
. . . g
k
p
j
p
,
we obtain the explicit representation
, =
1
p!

i
1
,...,i
p

i
1
,...,i
p
. (1.55)
(Notice that the r.h.s. of this expression may be equivalently written as , =
1
p!

i
1
,...,i
p

i
1
,...,i
p
.
The Hodge star is now dened as follows
:
p
V

np
V

,
,

p
V

: ,
!
= (). (1.56)
To see that this relation uniquely denes an (n p)form, we identify the components
()
i
p+1
,...,i
n
of the target form . To this end, we consider the particular case, = e
1
e
p
.
Separate evaluation of the two sides of the denition obtains
e
1
e
p
, =

p!

i
1
,...,i
p
e
1
e
p
[e
i
1
e
i
p
=
=

p!

i
1
,...,i
p
j
1
,...,j
p
g
1j
1
. . . g
pj
p

i
1
,...,i
p
= g
1i
1
. . . g
pi
p

i
1
,...,i
p
=
1,...,p
e
1
e
p
() =
1
(n p)!
()
i
p+1
,...,i
n
e
1
e
p
e
i
p+1
e
i
n
=
()
p+1,...,n
[g[
1/2
,
1.3 Metric 33
Comparing these results we arrive at the equation
()
p+1,...,n
= [g[
1/2

1,...,p
=
[g[
1/2
p!

i
1
,...,i
p
,p+1,...,n

i
1
,...,i
p
.
It is clear from the construction that this result holds for arbitrary index congurations, i.e. we
have obtained the coordinate representation of the Hodge star
()
i
p+1
,...,i
n
=
[g[
1/2
p!

i
1
,...,i
n

i
1
,...,i
p
. (1.57)
In words: the coecients of the Hodged form obtain by (i) raising the p indices of the coecients
of the original form, (ii) contracting these coecients with the rst p indices of the tensor,
and (iii) multiplying all that by
_
[g[/p!.
We derive two important properties of the star: The star operation is compatible with the
scalar product,
,
p
V

: , = sgn g , , (1.58)
and it is selfinvolutary in the sense that

p
V

: = sgn g ()
p(np)
. (1.59)
INFO The proof of Eqs. (1.58) and (1.59): The rst relation is proven by brute force computation:
, ) =
1
(n p)!
()
i
p+1
,...,i
n
()
i
p+1
,...,i
n
=
=
[g[
(n p)!(p!)
2

i
1
,...,i
n

i
1
,...,i
p

l
1
,...,l
p

j
1
,...,j
n
g
l
1
j
1
. . . g
l
p
j
p
g
i
p+1
j
p+1
. . . g
i
n
j
n
. .
g
1

l
1
...l
p
i
p+1
...i
n
=
=
sgn g
(p!)
2

l
1
,...,l
p
i
1
,...,i
p

i
1
,...,i
p

l
1
,...,l
p
=
sgn g
p!

i
1
,...,i
p

i
1
,...,i
p
= sgn g , ).
To prove the second relation, we consider two forms
p
V

and
np
V

. Then,
, ) = = ()
p(np)
= ()
p(np)
, ) =
= ()
p(np)
sgn g , ) = ()
p(np)
sgn g, ).
Holding for every this relation implies (1.59).
1.3.4 Isometries
As a nal metric concept of linear algebra we introduce the notion of isometries. Let V and V

be two vector spaces with metrics g and g

, respectively. An isometry F : V V

is a linear
mapping that conforms with the metric:
v, w V : g(v, w) = g

(Fv, Fw). (1.60)

EXAMPLE (i) The set of isometries of R
4
with Minkowski metric = diag(1, 1, 1, 1) is the
Lorentz group SO(3, 1). (ii) The canonical mapping J : V V

is an isometry of (V, g) and (V

g
).
34 Exterior Calculus
1.3.5 Metric structures on open subsets of R
n
Let U R
n
be open in R
n
. A metric on U is a collection of vector space metrics
g
x
: T
x
U T
x
U R,
smoothly depending on x. To characterize the metric we may choose a frame b
i
. This obtains
a matrixvalued function
g
ij
(x) g
x
(b
i
(x), b
j
(x)).
The change of one frame to another, b

i
= (A
1
)
j
i
b
j
transforms the metric as g
ij
(x) g

ij
(x) =
(A
1T
)
k
i
(x)g
kl
(x)(A
1
)
l
j
(x). Similarly to our discussion before, we dene an induced metric
g
x
on cotangent space by requiring that g
x
(b
i
x
, b
j
x
) g
ij
(x) be inverse to the matrix g
ij
(x).
EXAMPLE Let U = R
3
(negative xaxis). Expressed in the cartesian frame

x
i
, the Euclidean
metric assumes the form g
ij
=
ij
. Now consider the coordinate frame corresponding to polar
coordinates, (

r
,

). Transformation formulae such as

r
=
x
1
r

x
1
+
x
2
r

x
2
+
x
3
r

x
3
= sin cos

x
1
+ sin sin

x
2
+ cos

x
3
obtain the matrix
A
1
=
_
_
sin cos r sin sin r cos cos
sin sin r sin cos r cos sin
cos 0 r sin
_
_
Expressed in the polar frame, the metric then assumes the form
g = A
1T
1A
1
= |g
ij
=
_
_
1
r
2
sin
2

r
2
_
_
.
The dual frames corresponding to cartesian and polar coordinates on R
3
(negative xaxis) are given
by, respectively, (dx
1
, dx
2
, dx
3
) and (dr, d, d). The dual metric in polar coordinates reads
|g
ij
=
_
_
1
r
2
sin
2

r
2
_
_
.
Also notice that
_
[g[ = r
2
sin , i.e. the volume form
= dx
1
dx
2
dx
3
= r
2
sin dr d d
transforms in the manner familiar from standard calculus.
Now, let (U R
n
, g) and (U

R
m
, g

) be two metric spaces and F : U U

a smooth
mapping. F is an isometry, if it conforms with the metric, i.e. fullls the condition
x U, v, w T
x
U : g
x
(v, w)
!
= g

F(x)
(F

v, F

w).
In other words, for all x U, the mapping T
x
F must be an isometry between the two spaces
(T
x
U, g
x
) and (T
F(x)
U

, g

F(x)
). Chosing coordinates x
i
and y
j
on U and U

, respectively,
1.3 Metric 35
dening g
ij
= g(

x
i
,

x
j
) and g

ij
= g

(

y
i
,

y
j
), and using Eq. (1.14), using Eq. (1.14), one
obtains the condition
g
ij
!
=
F
l
x
i
F
k
x
j
g

lk
. (1.61)
A space (U, g) is called at if an isometry (U, g) (V, ) onto a subset V R
n
with metric
exists. Otherwise, it is called curved.
11
Turning to the other operations introduced in section 1.3.2, the volume form and the Hodge
star are dened locally: ()
x
=
x
, and =
_
[g[b
1
b
n
. It is natural to extend the
scalar product introduced in section 1.3.3 by integration of the local scalar products against
the volume form:
, :
p
U
p
U R,
(, ) ,
_
U

x
,
x
.
Comparison with Eq. (1.56) then obtains the important relation
, =
_
U
. (1.62)
Given two p-forms on a metric space, this denes a natural way to produce a number. On p xx
below, we will discuss applications of this prescription in physics.
1.3.6 Holonomic and orthonormal frames
In this section we focus on dual frames. A dual frame may have two distinguished properties:
An orthonormal frame is one wherein
g
ij
=
ij
.
It is always possible to nd an orthonormal frame; just subject the symmetric matrix g
ij
to a
GramSchmidt orthonormalization procedure.
A holonomic frame is one whose basis forms
i
are exact, i.e. a frame for which n functions
x
i
exist such that
i
= dx
i
. The linear independence of the
i
implies that the functions x
i
form
a system of coordinates. Open subsets of R
n
may always be parameterized by global systems of
coordinates, i.e. holonomic frames exist. In a holonomic frame,
g = dx
i
g
ij
dx
j
, g

=

x
i
g
ij

x
j
.
It turns out, however, that it is not always possible to nd frames that are both orthonormal
and holonomic. Rather,
11
The denition of curved subsets of R
n
makes mathematical sense but doesnt seem to be a very natural context. The
intuitive meaning of curvature will become more transparent once we have introduced dierentiable manifolds.
36 Exterior Calculus
the neccessary and sucient condition ensuring the existence of orthonormal and
holonomic frames is that the space (U, g) must be at.
To see this, consider the holonomic representation g = dx
i
g
ij
dx
j
and ask for a transformation
onto new coordinates x
i
to y
j
such that g = dy
i

ij
dy
j
be an orthonormal representa-
tion. Substituting dx
i
=
dx
i
y
j
dy
j
into the dening equation of the metric, we obtain g =
x
i
y
l
x
j
y
k
g
ij
dy
l
dy
k
!
= dy
l

lk
dy
k
. Comparison with (1.61) we see that the coordinate transforma-
tion must be an isometry.
EXAMPLE We may consider a piece of the unit sphere as parameterized by, say, the angular values
0 < < 45 deg and 0 < < 90 deg. The metric on the sphere obtains by projection of the Euclidean
metric of R
3
onto the submanifold r = 1, g
ij
= diag(sin
2
, 1). (Focusing on the coordinate space
we may, thus, think of our patch of the sphere as an open subset of R
2
(the coordinate space)
endowed with a nonEuclidean metric.) We conclude that (sin d, d) is an orthonormal frame.
Since d(sin d) = cos d d ,= 0 it is, however, not holonomic.
1.3.7 Laplacian
Coderivative
As before, U R
n
is an open oriented subset of R
n
with metric g. In the linear algebra of metric
spaces, taking the adjoint of an operator A is an important operation: Av, w = v, A

w.
Presently, the exterior derivative, d, is our most important linear operator. It is, thus, natural
to ask for an operator, , that is adjoint to d in the sense that

p1
U,
p
U : d,
!
= f, +. . . ,
where the ellipses stand for boundary terms
_
U
(. . . ) (which vanish if U is boundaryless or ,
have compact support inside U.) Clearly, must be an operator that decreases the degree of
forms by one. An explicit formula for may be obtained by noting that
d, =
_
U
d = ()
p1
_
U
d
(1.59)
=
= ()
p
sgn g()
(p1)(np+1)
_
U
d = sgn g()
(p+1)n+1
, d ,
where in the second equality we integrated by parts (ignoring surface terms). This leads to the
identication of the coderivative, a dierential operator that lowers the degree of forms by one:
:
p
U
p1
U,
sgn g()
n(p+1)+1
d .
Two more remarks on the coderivative:
It squares to zero, (d)(d) dd = 0.
1.3 Metric 37
Applied to a oneform
i
dx
i
, it obtains (exercise)
=
1
[g[
1/2

x
i
([g[
1/2

i
). (1.63)
Laplacian and the operations of vector analysis
We next combine exterior derivative and coderivative to dene a second order dierential oper-
ator,
:
p
U
p
U,
(d +d). (1.64)
By construction, this operator is self adjoint,
, = , .
If the metric is positive denite it is called the the Laplacian. If r = n 1 (cf. Eq. (1.48)) it is
called the dAlambert operator or wave operator (and usually denoted by .)
Using Eq. (1.63), it is straightforward to verify that the action of on 0forms (functions)
f
0
U is given by
f =
1
[g[
1/2

x
i
_
[g[
1/2
g
ij

x
j
f
_
. (1.65)
EXAMPLE Substitution of the spherical metric discussed in the example on p 34, it is straightforward
to verify that the Laplacian in three dimensional spherical coordinates assumes the form
=
1
r
2

r
r
2

r
+
1
r
2
sin

sin

+
1
r
2
sin
2

2
.
We are now, at last, in a position to establish contact with the operations of vector analysis.
Let f
0
U be a function. Its gradient is the vector eld
grad f J
1
df =
_
g
ij

x
j
f
_

x
i
. (1.66)
Let v = v
i
x
i
be a vector eld dened on an open subset U R
3
. Its divergence is dened as
div v Jv =
1
[g[
1/2

x
i
[g[
1/2
v
i
. (1.67)
Finally, let v = v
i
x
i
be a vector eld dened on an open subset U R
3
of threedimensional
space. Its curl is dened as
curl v J
1
dJv =
1
[g[
1/2

ijk
_

x
i
v
j
_

x
k
. (1.68)
38 Exterior Calculus
PHYSICS (E) Let us get back to our discussion of Maxwell theory. On p 19 we had introduced
two distinct two-forms, F and G, containing the electromagnetic elds as coecients. However, the
connection between these two objects was left open. On the other hand, a connection of some sort
must exist, for we know that in vacuum the electric eld E (entering F) and the displacement eld D
(entering G) are not distinct. (And analogously for B and H.) Indeed, there exists a relation between
F and G, and it is provided by the metric.
To see this, chose an orthonormal frame wherein the Minkovski metric assumes the form
=
_
_
_
1
1
1
1
_
_
_. (1.69)
Assuming vacuum, E = B and B = H, and using Eq. (1.57), it is then straightforward to verify that
the 2 form F and the (4 2)-form G are related through
G = F. (1.70)
How does this equation behave under a coordinate transformation of Minkovski space? Under a
general transformation, the components of F transform as those of a second rank covariant tensor.
The equation dF = 0 is invariant under such transformations. However, the equation G = F
involves the Hodge star, i.e. an operation depending on a metric. It is straightforward to verify (do it)
remains form-invariant only under isometric coordinate transformations. Restricting ourselves to linear
coordinate transformations, this identies the invariance group of Maxwell theory as the Poincare
group, i.e. the group of linear isometries of Minkowski space. The subgroup of the Poincare group
stabilizing at least one point (i.e. discarding translations of space) is the Lorentz group.
The Maxwell equations now assume the form d F = j and dF = 0, resp. In the traditional
formulation of electromagnetism, the solution of these equations is facilitated by the introduction of
a vector potential. Let us formulate the ensuing equations in the language of dierential forms:
On open subsets of R
4
, the closed form F may be generated from a potential one-form, A,
F = dA, (1.71)
whereupon the inhomogeneous Maxwell equations assume the form
d dA = j. (1.72)
Now, rather than working with the second order dierential operator dd, it would be nicer to express
Maxwell theory in terms of the self adjoint Laplacian, . To this end, we act on the inhomogeneous
Maxwell equation with a Hodge star to obtain d dA = dA = j. We now require A to obey
the Lorentz gauge condition A = 0.
12
A potential obeying the Lorentz condition can always be
found by applying a gauge transformation A A

A+df, where f is a 0-form (a function). The

12
In a component notation, A = A

dx

and
A = d A = d

1
3!

dx

dx

dx

=
=
1
3!

dx

dx

dx

dx

,
which is the familiar expression for the Lorentz gauge.
1.4 Gauge theory 39
condition A

= 0 then translates to df = f
!
= A, i.e. a linear dierential equation that can be
solved. In the Lorentz gauge, the inhomogeneous Maxwell equations assume the form
A = j, (1.73)
where = (d +d).
So far, we have achieved little more than a reformulation of known equations. However, as we are
going to discuss next, the metric structures introduced above enable us to interpret Maxwell theory
from an entirely new perspective: it will turn out that the equations of electromagnetism can be
derived entirely on the basis of geometric reasoning, i.e. without reference to physical input!
Much like Newtons equations are equations of motions for point particles, the Maxwell equations
(1.73) can be interpreted as equations of motions for a eld, A. It is then natural to ask whether these
equations, too, can be obtained from a Lagrangian variational principle. What we need to formulate a
variational principle is an action functional S[A], whose variation S[A]/A = 0 obtains Eq. (1.73)
as its Euler-Lagrange equation.
At rst sight, one may feel at a loss as to how to construct a suitable action. It turns out, however,
that geometric principles almost uniquely determine the form of the action functional: lets postulate
that our action be as simple as possible, i.e. a low order polynomial in the degrees of freedom of the
theory, the potential, A. Now, to construct an action, we need something to integrate over, that is
4-forms. Now, our so-fare development of the theory is based on the dierential forms, A, F, and j.
Out of these, we can construct the 4-forms F F, F F and j A. The rst of these is exact,
F F = dA dA = d(A dA) and hence vanishes under integration. Thus, a natural candidate of
an action reads
S[A] =
_
(c
1
F F +c
2
j A) ,
where c
i
are constants.
13
Let us now see what we get upon variation of the action. Substituting A A+a into the action
we obtain
S[A+a] =
_
(c
1
(dA+da) (dA+da) +c
2
j (A+a)) .
Expanding to rst order in a and using the symmetry of the scalar product
_
=
_
, we
arrive at
S[A+a] S[A] =
_
a (2c
1
d dAc
2
j) .
Stationarity of the integral is equivalent to the condition 2c
1
d dAc
2
j = 0. Comparison with Eq.
(1.72) shows that this condition is equivalent to the Maxwell equation, provided we set c
1
= c
2
/2.
Summarizing, we have seen that the structure of the Maxwell equations which entails the entire
body of electromagnetic phenomena follows largely from purely geometric reasoning!
1.4 Gauge theory
In this section, we will apply the body of mathematical structures introduced above to one of the
most important paradigmes of modern physics, gauge theory. Gauge principles are of enormously
13
In principle, one might allow c
i
= c
i
(A) to be functions of A. However, this would be at odds with our principle of maximal
simplicity.
40 Exterior Calculus
general validity, and it stands to reason that this is due to their geometric origin. This chapter
aims to introduce the basic ideas behind gauge theory, within the framework of the mathematical
theory developed thus far. In fact, we will soon see that a more complete coverage of gauge
theories, notably the discussion of non-abelian gauge theory, requires the introduction of more
mathematical structure: (Lie) group theory, the theory of dierential manifolds, and bundle
theory. Our present discussion will be heuristic in that we touch these concepts, without any
ambition of mathematical rigor. In a sense, the discussion of this section is meant to motivate
the mathematical contents of the chapters to follow.
1.4.1 Field matter coupling in classical and quantum mechanics (reminder)
Gauge theory is about the coupling of matter to so-called gauge elds. According to the modern
views of physics, the latter mediate forces (electromagnetic, strong, weak, and gravitational),
i.e. what we are really up to is a description of matter and its interactions. The most basic
paradigm of gauge theory is the coupling of (charged) matter to the electromagnetic eld.
We here recapitulate the traditional description of eld/matter coupling, both in classical and
quantum mechanics. (Readers familiar with the coupling of classical and quantum point particles
to the electromagnetic eld may skip this section.)
Consider the Lagrangian of a charged point particle coupled to the electromagnetic eld.
Representing the latter by a four-potential with components A

= (, A
i
) the corresponding
Lagrangian function is given by (we set the particle charge to unity)
14
L =
m
2
x
i
x
i
+ x
i
A
i
.
Exercise: consider the Euler-Lagrange equations (d
t

x
i L
x
i )L = 0 to verify that you obtain
the Newton equation of a particle subject to the Lorentz force, m x = E + v B, where
E =
t
A and B = A.
The canonical momentum is then given by p
i
=
x
i L = mx
i
+A
i
, which implies the Hamilton
function
H =
1
2m
(p A)
i
(p A)
i
+.
We may now quantize the theory to obtain the Schrodinger equation ( = 1 throughout)
_
i
t

1
2m
(iA)
i
(iA)
i

_
(x, t) = 0. (1.74)
What happens to this equation under a gauge transformation, A A+,
t
? Sub-
stitution of the transformed elds into (1.74) obtains a changed Schrodinger equation. However,
the gauge dependent contributions get removed if the gauge the wave function as (x, t)
14
In this section, we work in a non-relativistic setting. We assume a standard Euclidean metric and do not pay attention to the
co- or contravariance of indices.
1.4 Gauge theory 41
(x, t)e
i(x,t)
. We thus conclude that a gauge transformation in quantum mechanics is
dened by
A A+,

t
,
e
if
. (1.75)
1.4.2 Gauge theory: general setup
The take home message of the previous section is that gauge transformations act on the C-
valued functions of quantum mechanics through the multiplication by phases. Formally, this
denes a U(1)-action in C. Let us now anticipate a little to say that the states relevant to the
more complex gauge theories of physics will take values in higher dimensional vector spaces C
n
.
Natural extensions of the (gauge) group action will then be U(n) actions or SU(n) actions. We
thus anticipate that a minimal arena of gauge theory will involve
a domain of space time formally an open subset U R
d
, where d = 4 corresponds to
(3 + 1)dimensional space time.
15
A bundle of vector spaces B

xU
V
x
, where V
x
V and V is an rdimensional real
or complex vector space.
16
A transformation group G (gauge group) acting in the spaces V
x
. Often, this will be a norm-
preserving group, i.e. G = U(n) or SU(n) for complex vector spaces and G = O(n) or SO(n)
for real vector spaces.
A (matter) eld, i.e. a map : U B, x (x). (This is the generalization of a wave
function.)
Some dynamical input (the generalization of a Hamiltonian.) At rst sight, the choice the
dynamical model appears to be solely determined by the physics of the system at hand.
However, we will see momentarily that important (physical!) features of the system follow
entirely on the basis of geometric considerations. Notably, we will be forced to introduced a
structure known in mathematics as a connection, and in physics as a gauge eld.
15
In condensed matter physics, one is often interested in cases d < 4.
16
This is our second example of a vector bundle. (The tangent bundle TU was the rst.) For the general theory of bundle
spaces, see chapter ...
42 Exterior Calculus
Let us try to demystify the last point in the
list above. Physical models generally involve
the comparison of states at nearby points. For
example the derivative operation in a quantum
Hamiltonian,
x
(x, t) compares wave function
amplitudes at two innitesimally close points. In
other words, we will want to take derivatives of
states. Now, it is clear how to take the derivative
of a real scalar eld V = R: just form the
quotients
lim
0
1

(
()

(0)
), (1.76)
where is a curve in U locally tangent to the direction in which we want to dierentiate. How-
ever, things start to get problematic when
x
V
x
and V
x
V is a higher dimensional vector
space. The problem now is that
()
V
()
and
(0)
V
(0
live in dierent vector spaces.
But how do we compare vectors of dierent spaces? Certainly, the naive formula (1.76) wont
work anymore. For the concrete evaluation of the expression above requires the introduction of
components, i.e. the choice of bases of the two spaces. The change of the basis in only one of
the spaces would change the outcome of the derivative (cf. the gure above) which shows that
(1.76) is a meaningless expression.
We wish to postulate the freedom to independently choose a basis at dierent points in
space times an integral part of the theory. (For otherwise, we would need to come up with
some principle that synchronizes bases uniformly in space and time. This would amount to an
instantaneous action at the distance a concept generally deemed as problematic.) Still, we
need some extra structure that will enable us to compare elds at dierent points. The idea is
to introduce a principle that determines when two elds
x
and
x
are to be identied. This
principle must be gauge invariant in that identical elds remain identical after two independent
changes of bases at x and x

. A change of basis at x is mediated by an element of the gauge

group g
x
G. Here, g
x
is to be interpreted as a linear transformation g
x
: V
x
V
x
acting in
the eld space at x. The components of the eld in the new representation will be denoted by
g
x

x
.
In mathematics, the principle establishing a gauge covariant relation between elds at dierent
points is called a connection. The idea of a connection can be introduced in dierent ways. We
here start by dening an operation called parallel transport. Parallel transport will assign to
each
x
V
x
and each curve connecting x and x

an element []
x
V
x
which we interpret
as the result of transporting the eld
x
along to the space V
x
. In view of the isomorphy
V
x
V V
x
, we may think of [] G as an element of the gauge group. Importantly,
parallel transport is dened so as to commute with gauge transformations, which is to say that
the operation of parallel transport must not depend on the bases used to represent the spaces
V
x
and V
x
, resp.
In formulas the condition of gauge covariance is expressed as follows: subject
x
to a gauge
transformation to obtain g
x

x
. Parallel translation will yield []g
x

x
. This has to be equal to
1.4 Gauge theory 43
the result g
x
[]
x
obtained if we rst parallel transport and only then gauge transform. We
are thus lead to the condition
[] = g
x
[]g
1
x
. (1.77)
As usual, conditions of this type are easiest
to interpret for curve segments of innitesimal
length. For such curves, [] id. will be close
to the group identity, and [] id. will be
approximately and element of the Lie algebra,
g.
17
Innitesimal parallel transport will thus be a
prescription assigning to an innitesimally short
segment (represented by a tangent vector) an
element close to the group identity (represented
by a Lie algebra element). In other words,
Innitesimal parallel transport is described by a gvalued one-form, A, on U.
Let us now derive more concrete expressions for the parallel transportation of elds. To this end,
let (t) = ((t))(0) denote the elds obtained by parallel translation along a curve (t). We
then have
(t+) = ((t+))(0) = ((t+))(((t)))
1
((t))(0) = ((t+))(((t)))
1
(t).
Taylor expansion to rst order obtains
(t +) = (t) A(d
t
(t))(t) +O(
2
), (1.78)
where A(d
t
((t)) = d

=0
((t +))(((t)))
1
g is the Lie algebra element obtained by
evaluating the one form A on the tangent vector d
t
(t). Taking the limit 0, we obtain a
dierential equation for parallel transport
d
t
(t) = A(d
t
(t))(t). (1.79)
Having expressed parallel transport in terms of a gvalued one-form, the question arises what
conditions gauge invariance imply on this form. Comparing with (1.77) and denoting the gauge
transformed connection form by A

, we obtain (all equalities up to rst order in )

(t +) [id. A

(d
t
)]

(t) =
= g(t +)(t +) = g(t +) [id. A(d
t
)] (t) =
= g(t +) [id. A(d
t
)] g
1
(t)

(t) =
_
id. +((d
t
g(t))g
1
(t) g(t)A(d
t

t
))g
1
(t)

(t).
17
Referring for a more substantial discussion to chapter xx below, we note that the Lie algebra of a Lie group G is the space
of all tangent vectors d
t

t=0
g(t) where g(t) is a smooth curve in g with g(0) = id.
44 Exterior Calculus
where g(t) is shorthand for g((t)). Comparing terms, and using that (d
t
g(t))g
1
=
g(t)d
t
g
1
(t), we arrive at the identication
A

= gAg
1
+gdg
1
, (1.80)
where gdg
1
is the gvalued one form dened by (gdg
1
)(v) = g((0))d
t
[
t=0
g
1
((t)) where
d
t
(t) = v.
Notice what happens in the case G = U(1) and g = iR relevant to conventional quantum
mechanics. In this case, writing g = e
i
, where is a real valued function on U, the dierential
form gdg
1
= id collapses to a real valued form. Also g
1
Ag = A, on account of the
commutativity of the group. This leads to the transformation law A

= A id reminiscent
of the transformation behavior of the electromagnetic potential. (The extra i appearing in this
relation is a matter of convention.) This suggests a tentative identication
connection form (mathematics) = gauge potential (physics).
Eq. (1.79) describes the innitesimal variant of parallel transport. Mathematically, this is a
system of ordinary linear dierential equations with time dependent coecients. Equations of this
type can be solved in terms of so-called path ordered exponentials: Let us dene the generalized
exponential series
[
t
] P exp
_

t
A
_

j=1
()
j
_
t
0
dt
1
_
t
1
0
dt
2
. . .
_
t
j1
0
dt
j
A( (t
1
))A( (t
2
)) . . . A(
t
j
),
(1.81)
where
t
is a shorthand for the extension of a curve = (s)[s [0, 1] up to the parameter
value s = t.
The series is constructed so as to solve the dierential equation
d
t
P exp
_

t
A
_
= A( (t))P exp
_

_
(t)
A
_
,
with initial condition P exp
_

0
A
_
= id.. Consequently
(t) = P exp
_

t
A
_
(0)
describes the parallel transport of (0) along curve segments of nite length.
INFO In the abelian case G = U(1) relevant to electrodynamics, the (matrices representing the)
elements A( ) at dierent times commute. In this case,
_
t
0
dt
1
_
t
1
0
dt
2
. . .
_
t
j1
0
dt
j
A( (t
1
))A( (t
2
)) . . . A(
t
j
) =
1
j!
__

t
0
dt A( )
_
j
and [
t
] = exp
_

t
0
dt A( )
_
collapses to an ordinary exponential. In components, this may be
written as
[] = e

t
0
ds A

((s))

(s)
.
1.4 Gauge theory 45
1.4.3 Field strength
Our discussion above shows that a connection naturally brings about an object, A, behaving
similar to a generalized potential. This being so, one may wonder whether the eld strength
corresponding to the potential carries geometric meaning, too. As we are going to show next,
the answer is armative.
Consider a connection as represented by its connection oneform, A. The ensuing parallel
transporters [] generally depend on the curve, i.e. parallel transport along two curves con-
necting two points x and x

will not, in general give identical results. Equivalently, [] may

dier from unity, even if is closed. To understand the consequences, let us consider the case
of the abelian group G = U(1) relevant to quantum electrodynamics. In this, case, (see info
section above), parallel transport around a closed loop in space-time can be written as
[] = e

A
= e

S()
dA
= e

S()
F
,
where S() may be any surface surrounded by . This shows that the existence of a non-trivial
parallel transporter around a closed loop is equivalent to the presence of a non-vanishing eld
strength form. (Readers familiar with quantum mechanics may interpret this phenomenon as
a manifestation of the Aharonov-Bohm eect: a non-vanishing Aharonov-Bohm phase along a
closed loop (in space) is indicative of a magnetic eld penetrating the loop.)
How does the concepts of a eld strength generalize to the non-abelian case? As a result of
a somewhat tedious calculation one nds that the non-abelian generalization of F is given by
F = dA+A A. (1.82)
The gvalued components of the two-form F are given by
F

+ [A

, A

],
where [X, Y ] = XY Y X is the matrix commutator. Under a gauge transformation A
gAg
1
+gdg
1
. Substituting this into (1.82), we readily obtain
F

= gFg
1
. (1.83)
INFO To prove Eq. (1.82), we consider an innitesimal curve of length . The area bounded by
the curve will then be of O(
2
). We wish to identify contributions to the path ordered exponential of
this order. A glance at the abelian expression exp(
_
S()
F) = 1 + F O(
2
) +O(
3
) shows that
this is sucient to identify the generalization of F.
We thus expand
[] = id. +
_
1
0
dt
1
A( (t
1
)) +
_
1
0
dt
1
_
t
1
0
dt
2
A( (t
1
))A( (t
2
)) +O(
3
).
The term of rst order in A is readily identied as
_

A =
_
S()
dA, which is of O(
2
). Turning to
the second term, we represent the product of matrices
A( (t
1
))A( (t
2
)) =
1
2
([A( (t
1
)), A( (t
2
))]
+
+ [A( (t
1
)), A( (t
2
))]

)
46 Exterior Calculus
as a sum of a symmetrized and an anti-symmetrized contribution. Here, [A, B]

= AB BA. The
symmetric contribution evaluates to
_
1
0
dt
1
_
t
1
0
dt
2
[A( (t
1
)), A( (t
2
))]
+
=
__
1
0
dt A(t)
_
2
= O(
4
),
and can be discarded. Turning to the antisymmetric contribution, we obtain
_
1
0
dt
1
_
t
1
0
dt
2
[A( (t
1
)), A( (t
2
))]
+
=
1
2
A

_
1
0
dt
1
_
t
1
0
dt
2
(

(t
1
)

(t
2
)

(t
1
)

(t
2
)) +O(
3
) =
=
1
2
A

_
1
0
dt
1
(

(t
1
)

(t
1
)

(t
1
)

(t
1
)) +O(
3
) =
=
1
2
A

(d

) +O(
3
) =
=
1
2
A

_
S()
(d

) +O(
3
) =
=
_
S()
A A+O(
3
).
In the crucial rst equality, we noted that for an innitesimal curve, A( (t)) = (A

)
(t)

(t)
(A

)
(0)

(t) A

(t), i.e. the coecients of the potential form can be pulled out of the integral.
Combining terms, we arrive at
[] = id. +
_
S()
(dA+A A) +O(
3
).
Comparing with the abelian expression, we obtain (1.82) for the non-abelian generalization of the
eld strength form.
The discussion above illustrates the appearance of maps carrying representations dierent
from the fundamental group representation of G in V : Let us assume that we are interested
in the variation of a smooth map : U X. Here, X = B corresponds to a V -valued function.
However, we may also choose to consider forms X =
p
U, or just ordinary functions X = R.
These maps generally carry a representation of the group G whose specics depend on the target
space and on the denition of the map. For X = V , this representation will be the fundamental
representation considered above,
x
g
x

x
, where g
x
= g
x,ij
is the matrix representing
g
x
U. For X =
2
U, we may encounter other representations. For example, for = F, the
eld strength form, F
x
gF
x
g
1
transforms according to the adjoint representation, cf. Eq.
(1.83). Finally, for X = R, does not transform under G, transformation behavior which we
may formally assign to the singlet representation.
It is straightforward to generalize the notion of parallel transport to objects transforming
according to arbitrary group representations. For example, for an object transforming according
to the adjoint representation, the analog of (1.78) reads
(t +) = (t) [A(d
t
(t))(t) (t)A(d
t
(t))] +O(
2
),
1.4 Gauge theory 47
which immediately leads to
d
t
(t) = [A(d
t
), (t)]. (1.84)
The generalization to objects transforming under yet dierent representations of G should be
straightforward.
1.4.4 Exterior covariant derivative
With the notion of parallel transport in place, we are now in a position to dene a meaningful
derivative operation. The idea simply is to measure variations of objects dened in U in terms
of deviations from the parallel transported objects. Consider, thus, a curve (t), as before. The
derivative of a function along is described by the dierential quotient
D

(t)
( (t)) lim
0
1

(t+)

(t)
+A( (t))
(t)

.
Here, both
(t+)
and the parallel transport of
(t)
, i.e.
(t)
+A( (t))
(t)
are considered
to be elements of V
(t+)
, and we assumed to transform under the fundamental representation
of G. This expression denes the so-called (exterior) covariant derivative of along . The
general covariant derivative (prior to reference to a direction of dierentiation) is dened as
D d +A , (1.85)
where in the case of a V
x
-valued function (zero-form), the wedge product reduces to the con-
ventional product between the matrix A and the vector . In components:
(D)
i
d
i
+A
i
j

j
.
The wedge product becomes important, once we generalize to the covariant derivative of dier-
ential forms. For example, for a two-form transforming under the adjoint representation, the
covariant derivative reads (cf. Eq. (1.84)),
D = d A + A. (1.86)
Let us discuss the most important properties of the covariant derivative:
By design, the covariant derivative is compatible with gauge transformations: with

= g,
A

= gAg
1
+gdg
1
,
we have
D

= g(D),
where D

d +A

. This is checked by direct substitution of the denitions.

The covariant derivative obeys the Leibniz rule
D( ) = D + ()
q
D.
48 Exterior Calculus
Unlike with d
2
= 0, the covariant derivative is not nilpotent. Rather, one may check by
straightforward substitution that
D
2
= F,
where F is the eld strength form (1.82).
Finally, let us consider the covariant derivative of F itself. The eld strength transforms
under the adjoint representation, (1.83), which means that the covariant derivative DF =
dF +AF F A is well dened. Substituting the denition (1.82), we readily obtain the
Bianchi identity,
DF = 0, (1.87)
which generalizes the homogeneous Maxwell equations to general gauge theories.
The covariant derivative plays a very important role in physics. Important applications include all
areas of gauge theory, and general relativity. In the latter context, the role of the connection
is assumed by the so-called Riemannian connection associated to the curvature of space time.
We will return to this point in chapter xx below.
1.5 Summary and outlook
In this chapter, we have introduced a minimal framework of mathematical operations relevant
to dierential geometry: we discussed the (exterior) multilinear algebra, and its generalization
to alternating and locally linear maps on the tangent bundle of open subsets of R
n
, i.e. the
apparatus of dierential forms. We learned how to dierentiate and integrate dierential forms,
thus generalizing the basic operations of vector analysis. Finally, we introduced the concept of
a metric as an important means to characterize geometric structures.
The mathematical framework introduced above is powerful enough to describe various appli-
cations of physics in a unied and ecient way. We have seen how to formulate the foundations
of classical mechanics and electrodynamics in a coordinate invariant way. Here, the notion of
coordinate invariance implies three major advantages: (i) the underlying structure of the theory
becomes maximally visible, i.e. formulas arent cluttered with indices, etc., (ii) the change(ability)
between dierent coordinate systems is exposed in transparent terms, and (iii) formulas relating
to specic coordinate systems (think of the formula (1.65)) are formulated so as to expose the
underlying conceptual structure. (You cant say this about the standard formula of the Laplacian
in spherical coordinates.)
However, in our discussion of the nal example gauge theory we were clearly pushing
limits, and several important limitations of our so far theory became evident. What we need,
at least, is an extension from geometry on open subsets of R
n
to more general geometric
structures. Second, we should like to give the notion of bundles i.e. mathematical constructs
where some mathematical structure is locally attached to each point of a base structure a
more precise denition. And thirdly, the ubiquitous appearance of continuous groups in physical
applications calls for a geometry oriented discussion of group structures. In the following chapters
we will discuss these concepts in turn.
2
Manifolds
In this chapter, we will learn how to describe the geometry of structures that cannot be identied
with open subsets of R
n
. Objects of this type are pervasive both in mathematics, and in physical
applications. In fact, it is the lack of an identication with a single coordinate domain in R
n
that makes a geometric structure interesting. Prominent examples of such manifolds include
spheres, tori, the celebrated Moebius strip, continuous groups, and many more.
2.1 Basic structures
2.1.1 Dierentiable manifolds
Spaces M which locally (yet not necessarily globally) look like open subsets of R
n
are called
manifolds. The precise meaning of the notion look like is provided by the following denition:
A chart of a manifold M is a pair (U, ), where U M is an open (!) subset of M and
: U (U) R
n
,
x (x)
is a homeomorphism ( is invertible and both and
1
are continuous) of U onto an image
(U) R
n
which is open in R
n
. Notice that the denition above requires the existence of
a certain amount of mathematical structure: We rely on the existence of open subsets and
continous mappings. This means that M must, at least, be a topological space. More precisely,
M must be a Hausdor space
1
whose topology is generated by a countable basis. Loosely
speaking, the Hausdorness of M means that it is a topological space (the notion of openness
and continuity exists) on which we may meaningfully identify distinct points.
The chart assigns to any point x U M a set of ncoordinates
i
(x), the coordinates of
x with respect to the chart . If we are working with a denite chart we will, to avoid excessive
notation, often use the alternative designation x
i
(x), or just x
i
. Given the notion of charts,
we are able to dene topological manifolds. A topological manifold M is a Hausdor space
with countable basis such that every point of M lies in a coordinate neighbourhood, i.e. in the
domain of denition U of a chart. A collection of charts (U
r
,
r
) such that

r
U
r
= M covers
M is called an atlas of M.
1
A Hausdor space is a topological space for which any two distinct points possess disjoint neighbourhoods. (Exercise: look up
the denitions of topological spaces, bases of topologies, and neighbourhoods.)
49
50 Manifolds
Figure 2.1 On the denition of topological manifolds. Discussion, see text.
INFO Minimal manifolds as dened above are called topological manifolds. However, most man-
ifolds that are encountered in (physical) practice may be embedded into some suciently high
dimensional R
n
. (Do not confuse the notions embedding and identifying. I.e. we may think of the
twosphere as a subset of R
3
, it is, however, not possible to identify it with a subspace of R
2
.) In
such cases, the manifolds inherits its topology from the standard topology of R
n
, and we need not
worry about topological subtleties.
Consider now two charts (U
1
,
1
) and (U
2
,
2
) with non-empty intersection U
1
U
2
. Each
x U
1
U
2
then possesses two coordinate representations x
1

1
(x) and x
2

2
(x). These
coordinates are related to each other by the map
2

1
1
, i.e.

2

1
1
:
1
(U
1
U
2
)
2
(U
1
U
2
),
x
1
x
2
=
2

1
1
(x
1
),
or, in components, x
i
2
=
i
2
(
1
1
(x
1
1
, . . . , x
n
1
)). The coordinate transformation
2

1
1
denes
a homeomorphism between the open subsets
1
(U
1
U
2
) and
2
(U
1
U
2
). If, in addition, all
coordinate transformations of a given atlas are C

, the atlas is called a C

atlas. (In practice,

we will exclusively deal with C

systems.)
EXAMPLE The most elementary example of a manifold is an open subset U R
n
. It may be
covered by a oneatlas chart containing just (U, id
U
).
EXAMPLE Consider the twosphere S
2
R
3
, i.e. the set of all points x R
3
fullling the
condition (Euclidean metric in R
3
) (x
1
)
2
+ (x
2
)
2
+ (x
3
)
2
= 1. We cover S
2
by two chartsdomains,
2.1 Basic structures 51
U
1
|x S
2
[x
3
> 1 (S
2
south pole) and U
2
|x S
2
[x
3
< 1 (S
2
north pole). The two
coordinate mappings (aka stereographic projections of the sphere)
1
are dened by

1
(x
1
, x
2
, x
3
) =
1
1 +x
3
(x
1
, x
2
)
1
(U
1
) R
2
, x
3
> 1

2
(x
1
, x
2
, x
3
) =
1
1 x
3
(x
1
, x
2
)
2
(U
2
) R
2
, x
3
< 1.
If the union of two atlases of M, (U
i
,
i
) and (V
i
,
i
) is again an atlas, the two parent
atlases are called compatible. Compatibility of atlases denes an equivalence relation. Individual
equivalence classes of this relation are called dierentiable structures. I.e. a dierentiable
structure on M contains a maximum set of mutually compatible atlases. A manifold M equipped
with a dierentiable structure is called a dierentiable manifold. (Throughout we will refer to
dierentiable manifolds just as manifolds.)
EXAMPLE Let M = R be equipped with the standard topology of R and a dierentiable structure
be dened by the onechart atlas |(R,
1
), where
1
(x) = x. Another dierentiable structure is
dened by |(R,
2
), where
2
(x) = x
3
. These two atlases indeed belong to dierent dierentiable
structures. For,
1

1
2
: x x
1/3
is not dierentiable at x = 0.
2.1.2 Dierentiable mappings
A function f : M R is called a dierentiable function (at x M) if for any chart U x,
the function f
1
: (U) R
n
R is dierentiable in the ordinary sense of calculus, i.e.
f(x
1
, . . . , x
n
) has to be a dierentiable function at (x) (cf. Fig. 2.2, top.) It does, in fact,
suces to verify dierentiability for just one chart of Ms dierentiable structure. For with any
other chart, , f
1
= (f
1
)(
1
) and dierentiability follows from the dierentiability
of the two constituent maps. The algebra of dierentiable functions of M is called C

(M).
More generally, we will want to consider maps
F : M M

between dierentiable manifolds M and M

of dimensions n and n

, resp. (cf. Fig. 2.2, center

part.) Let, x M, U x the domain of a chart and U

F(x) be a chart of M

containing
xs image. The function F is dierentiable at x if

F
1
: (U) R
n

(F(U)) R
n

is dierentiable in the sense of ordinary calculus (i.e. F

i
(x
1
, . . . , x
n
)
i
(F(x
1
, . . . , x
n
)),
i = 1, . . . , n

are dierentiable at (x).) The set of all smooth mappings F : M M

will be
designated by (

(M, M

).
The map F is a dieomorphism if it is invertible and both F and F
1
are dierentiable.
(What this means is that for any two chart domains,

F
1
dieomorphically maps (U)
onto

(F(U)). If a dieomorphism F : M M

exists, the two manifolds M and M

are
dieomorphic. In this case, of course, dimM = dimM

.
INFO The denitions above provide the link to the mathematical apparatus developed in the previous
chapter. By virtue of charts, maps between manifolds may be locally reduced to maps between open
subsets of R
n
(viz. the maps expressed in terms of local coordinates.) It may happen, though, that
52 Manifolds
Figure 2.2 On the denition of continuous maps of manifolds into the reals, or between manifolds.
Discussion, see text.
a map meaningfully dened for a local coordinate neighbourhood dees extension to the entire atlas
covering M. Examples will be encountered below.
2.1.3 Submanifolds
A subset N of an ndimensional manifold M is called a qdimensional submanifold of M if
for each point x
0
N there is a chart (U, ) of M such that x
0
U and for all x U N,
(x) = (x
1
, . . . , x
q
, a
q+1
, . . . , a
n
),
with
q+1
, . . . , a
n
xed. Dening

U = U N and :

U R
q
, (x) = (x
1
, . . . , x
q
), we obtain
a chart (

U, ) of the qdimensional manifold N. A (compatible) collection of such charts denes

a dierentiable structure of N.
EXAMPLE An open subset U M is an ndimensional submanifold of M. We may think of a
collection of isolated points in M as a zerodimensional manifold.
EXAMPLE Let f
i
: M R, i = 1, . . . , p be a family of p functions on M. The set of solutions of
the equations f
1
(x) = = f
p
(x) = const. denes a pdimensional submanifold on M if the map
M R
p
, x (f
1
(x), . . . , f
p
(x)) has rank p.
2.2 Tangent space 53
2.2 Tangent space
In this section, we will generalize the notion of tangent space as introduced in the previous
chapter to the tangent space of a manifold. To make geometric sense, this denition will have
to be independent of the chosen atlas of the manifold.
2.2.1 Tangent vectors
The notation

x
i
introduced in section 1.2.1 to designate coordinate vector elds suggests an
interpretation of vector elds as dierential operators. (Indeed, the components v
i
of v = v
i
x
i
where obtained by dierentiating the coordinate functions in the direction of v.) This view
turns out to be very convenient when working on manifolds. In the following, we will give the
derivative-interpretation of vectors the status of a precise denition.
Consider a curve : [a, a] M such that (0) = x. To dene a vector v
x
tangent to the
manifold at x M we take directional derivatives of functions f : M R at x in the direction
identied by . I.e. the action of the tangent vector v
x
identied by the curve on a function f
is dened by
v
x
(f) d
t

t=0
f((t)).
Of course there are other curves

such that

(t
0
) = (0) and d
t
f((t))

t=0
= d
t
f(

(t))

t=t
0
.
All these curves are tangent to each other at x and will generate the same tangent vector
action. It is thus appropriate to identify a tangent vectors v
x
at x as equivalence classes of
curves tangent to each other at x.
For a given chart (U, ), the components v
i
x
of the vector v
x
are obtained by letting v
x
act
on the coordinate functions:
v
i
x
v
x
(x
i
) = d
t

i
(t)

t=0
,
where
i

i
. According to the chain rule, the action of the vector on a general function
is then given by
v
x
(f) = v
i
x

f
x
i
, (2.1)
where

f = f
1
: (U) R
n
R is a real valued function of n variables and
x
i

f its
partial derivative. In a notation emphasizing v
x
s action as a dierentiable operator we have
v
x
= v
i
x

x
i
.
Notice the analogy to our earlier denition in section 1.2.1; the action of a vector is dened by
taking directional derivatives in the direction identied by its components.
The denition above is coordinate independent and conforms with our earlier denition of
tangent vectors. At the same time, however, it appears to be somewhat unnatural to link the
denition of vector (dierential operators) to a set of curves.
2
Indeed, there exists an alternative
2
Cf. the introduction of partial derivatives in ordinary calculus: while the action of a partial derivative operation on a function
is dened in terms of a curve (i.e. the curve identifying the direction of the derivative), in practice one mostly applies partial
derivatives without explicit reference to curves.
54 Manifolds
denition of vectors which does not build on the notion of curves: A tangent vector v
x
at
x M is a derivation, i.e. a linear map from the space of smooth functions
3
dened on some
open neighbourhood of x into the real numbers fullling the conditions
v
x
(af +bg) = av
x
f +bv
x
g, a, b R, f, g functions, linearity,
v
x
(fg) = fv
x
g + (v
x
f)g, Leibnitz rule.
To see that this denition is equivalent to the one given above, let y be a point innitesimally
close to x. We may then Taylor expand
f(y) = f(x) + ((y) (x))
i


f
x
i

x
.
Again dening the components v
i
x
of the vector (in the chart ) as
v
i
x
= v
x

i
,
and taking the limit y x we nd that the action of the vector on f is given by (2.1).
We may nally relate the denitions of tangent vectors given above to (the mathematical
formulation of ) a denition pervasive in the physics literature. Let (, U) and (

, U

) be two
charts such that x U U

. The action of a tangent vector v

x
on a function then aords the
two representations
v
x
f = v
i
x

f
x
i

(x)
= v
i
x

x
i

(x)
,
where

f

= f
1
=

f (
1
). Using the abbreviated notation (

1
)(x) = x

(x) we
have

f
x
i
=


f
x
j
x
j
x
i
,
we obtain the transformation law of vector components
v
i
x
=
x
i
x
j
v
j
x
. (2.2)
This leads us to yet another possibility to dene tangent vectors: A tangent vector v
x
is
described by a triple (U, , V ), where V R
n
is an ncomponent object (containing the
components of v
x
in the chart .) The triples (U, , V ) and (U

, V

) describe the same

tangent vector i the components are related by V
i
=
x
i
x
j
V
j
.
We have, thus introduced three dierent (yet equivalent) ways of dening tangent vectors on
manifolds:
Tangent vectors as equivalence classes of curves, as
derivative operations acting on (germs of) functions, and
a denition in terms of (contravariant) transformation behaviour of components.
3
To be precise, vectors map socalled germs of functions into the reals. A germ of a function is obtained by identifying functions
for which a neighbourhood around x exists in which the reference functions coincide.
2.2 Tangent space 55
2.2.2 Tangent space
Dening a linear structure in the obvious manner, (av
x
+ bw
x
)(f) = av
x
(f) + bw
x
(f), the
set of all tangent vectors at x becomes a linear space, the tangent space T
x
M. The union

xM
T
x
M TM denes the tangent bundle of the manifold. Notice that
The tangent bundle, TM of an n-dimensional manifold, M, is a 2n-dimensional
manifold by itself.
For, in a chart domain of M with coordinates x
i
, the elements of TM may be identied
in terms of coordinates x
1
, . . . , x
n
, v
1
, . . . , v
n
, where the vectorial components of T
x
M are
parameterized of v = v
i
x
i
. In a similar manner, we may introduce the cotangent bundle
as the space TM

xM
T

x
M, where T

x
M is the dual space of T
x
M. Again, TM

is a
2n-dimensional manifold. Below, we will see that it comes with a very interesting mathematical
structure.
A vector eld on the domain of a chart, (U, ) is a smooth mapping
v : U TU,
x v
i
x

x
i
T
x
U.
A vector eld on the entire manifold is obtained by extending this mapping from a single chart
to an entire atlas and requiring the obvious compatibility relation, x U U

v
x
= v

x
. (In
coordinates, this condition reads as (2.2).)
The set of all smooth vector elds on M is denoted by vect (M). For v vect (M) and
f C

M, the action of the vector eld on the function obtains another function, v(f), dened
by
(v(f))(x) = v
x
(f).
A frame on (a subset of) M is a set (b
1
, . . . , b
n
) of n vector elds linearly independent at
each point of their denition. A coordinate system (U, ) denes a local frame (

x
1
, . . . ,

x
n
).
However, in general no frame extensible to all of M exists. If such a frame exists, the manifold
is called parallelizable.
EXAMPLE Open subsets of R
n
, Lie groups (see next chapter), and certain spheres S
1
, S
3
, S
7
are
examples of parallelizable manifolds. Nonparallelizable are all other spheres, the Moebius strip and
many others more.
2.2.3 Tangent mapping
For a smooth map F : M M

between two dierentiable manifolds, the tangent mapping

may be dened by straightforward extension of our denition in section 1.2.3. For x M, we
56 Manifolds
dene
TF
x
: T
x
M T
F(x)
M

,
v
x
TF
x
(v
x
),
[TF
x
(v
x
)]f v
x
(f F).
For two coordinate systems (U, ), and (U

) covering x U and the image point F(x) U

,
respectively, the components (TF
x
(v
x
))
i
of the vector (TF
x
)(v
x
) obtain as
(TF
x
(v
x
))
i
=


F
i
x
j
v
j
,
where

F =

F
1
. The tangent mapping of the composition of two maps G F evaluates
to
T(G F)
x
= TG
F(x)
TF
x
.
2.2.4 Dierential Forms
Dierential forms are dened by straightforward generalization of our earlier denition of dier-
ential forms on open subsets of R
n
: A pform on a dierentiable manifold maps x M to

x

p
(T
x
M)

. The xdependence is required to be smooth, i.e. for v

1
, . . . , v
p
vect (M),

x
(v
1
(x), . . . , v
p
(x)) is a smooth function of x. The vector space of pforms is denoted by
p
M
and the algebra of forms of general degree by
M
n

p=0

p
M.
The mathematics of dierential forms on manifolds completely parallels that of forms on open
subsets of R
n
. Specically,
The wedge product of dierential forms and the inner product of a vector eld and a form
are dened as in section 1.2.4.
A (dual) nframe is a set of n linearly independent 1forms. On a coordinate domain, the
coordinate forms (dx
1
, . . . , dx
n
) locally dene a frame, dual to the coordinate vector elds
(

x
1
, . . . ,

x
n
). In the domain of overlap of two charts, a pform aords the two alternative
coordinate representations
=
1
p!

i
1
,...,i
p
dx
i
1
dx
i
p
,
=
1
p!

i
1
,...,i
p
dx
i
1
dx
i
p
,

i
1
,...,i
p
=
j
1
,...,j
p
x
j
1
x
i
1
. . .
x
j
p
x
i
p
.
2.2 Tangent space 57
For a given chart, the exterior derivative of a pform is dened as in Eq. (1.22). The
coordinate invariance of that denition pertains to manifolds, i.e. the denition of the exterior
derivative does not depend on the chosen chart; given an atlas, d may be dened on the
entire manifold.
The pullback of a dierential form under a smooth mapping between manifolds is dened as
before. Again, pullback and exterior derivative commute.
The one mathematical concept whose generalization from open subsets of R
n
to manifolds
requires some thought is Poincares lemma. As a generalization of our earlier denition of star
shaped subsets of R
n
, we dene the notion of contractible manifolds: A manifold M is called
contractible contractible manifold if the identity mapping M M, x x may be continuously
deformed to a constant map, M M, x x
0
, x
0
M xed. In other words, there has to
exist a family of continuous mappings,
F : [0, 1] M M, (t, x) F(x, t),
such that F(x, 1) = x and F(x, 0) = x
0
. For xed x, F(x, t) denes a curve starting at x
0
and
ending at x. (Exercise: show that R
n
is contractible, while R
n
0 is not.)
Poincares lemma now states that on a contractible manifold a pform is exact if and only
if it is closed.
2.2.5 Lie derivative
A vector eld implies the notion of transport on a manifold. The idea is to trace the behavior
of mathematical objects functions, forms, vectors, etc. as one ows along the directions
specied by the reference eld. In this section, we dene and explore the properties of the ensuing
derivative operation, the Lie derivative.
The ow of a vector eld
Given a vector eld, v, we may attribute to each point x M a curve whose tangent vector at
x equals v
x
. The union of all these curves denes the ow of the vector eld.
More precisely, we wish to introduce a oneparameter group of dieomorphisms,
: V U,
(x, )

(x),
where R parameterizes the oneparameter group for xed x M. Think of (x, ) as a
curve parameterized by . We parameterize this curve such that
0
(x) = x. For each domain
of a chart, U, and x U, we further require that V = supp() x R = x interval
(x, 0), i.e. for each x the parameter interval of the oneparameter group is nite. is a group
of dieomorphisms in the sense that

+
(x) =

(x)).
58 Manifolds
Each map denes a vector eld v vect(M),
v : M TM,
x (x, d

=0

(x)),
i.e. x is mapped onto the tangent vector of the curve

(x) at = 0. Conversely, each vector

eld v denes a oneparameter group of dieomorphisms, the ow of a vector eld. The ow
graphically, the trajectories traced out by a swarm of particles whose velocities at x(t) equal
v
x(t)
is dened by (see the gure above)
x M : d

(x)
!
= v

(x)
.
Here, we interpret

: (

(M) (

(M), f

(f) as a map between functions on M,

where (

(f))(x) f(

(x)). The equation above then reads as (d

)(f) = d

f(

) =
f
x
i
d

=
f
x
i
v
i
. With a local decomposition v = v
i
x
i
this translates to the set of rst order
ordinary dierential equations,
i = 1, . . . , n : d

(x)
!
= v
i

(x)
.
Together with the initial condition
i
0
(x) = x
i
, we obtain a uniquely solvable problem (over at
least a nite parameter interval of .)
EXAMPLE Let M = R
n
and v
x
= Ax, where A GL(n). The ow of this vector eld is
: M R M, (x, )

(x) = exp(A)x.
Lie derivative of forms
Given a vector eld and its ux one may ask how a dierential form
p
M changes as one
moves along its ux lines. The answer to this question is provided by the socalled Lie derivative.
The Lie derivative compares
x

p
(T
x
M)

with the pullback of

(x)

p
(T

(x)
)

under

(where we interpret

: U U as a dieomorphism of an open neighbourhood

U x.) The rate of change of along the ux lines is described by the dierential quotient
lim
1

(x)
)
x
), the Lie derivative of in the direction of v.
Formally, the Lie derivative is a degreeconserving map,
L
v
:
p
M
p
M,
L
v
,
L
v
= d

=0
(

(x)
). (2.3)
Properties of the Lie derivative (all immediate consequences of the denition):
L
v
is linear, L
v
(w +w

) = L
v
w +L
v
w

, and
obeys the Leibniz rule, L
v
( ) = (L
v
) + L
v
.
It commutes with the exterior derivative dL
v
= L
v
d and
L
v
is linear in v.
2.2 Tangent space 59
for f L
0
M a function, L
v
f = dfv reduces to the directional derivative, L
v
f = v(f). This
can also be written as L
v
f = d

=0
f

.
In practice, the computation of Lie derivatives by the formula (2.3) is cumbersome: one rst
has to compute the ow of the eld v, then its pullback map, and nally dierentiate w.r.t.
time. Fortunately there exists an alternative prescription due to Cartan that is drastically more
simple:
L
v
= i
v
d +d i
v
. (2.4)
In words: to compute the Lie derivative, L
v
, of arbitrary forms, one simply has to apply an
exterior derivative followed by insertion of v (plus the reverse sequence of operations.) Now,
this looks like a manageable operation. Let us sketch the proof of Eq. (2.4). One rst checks
that the operations on both the left and the right hand side of the equation are derivations.
Consequently, it is sucient to prove the equality for zero forms (functions) and one-forms, dg.
(In combination with the Leibniz rule, the expandability of an arbitrary form into wedge products
of these building blocks then implies the general identity.)
For functions (cf. the list of properties above), we have L
v
f = df(v) = i
v
df = (i
v
d +di
v
)f,
where we used that i
v
f = 0 by denition. For one-forms, dg, we obtain
L
v
dg = d(L
v
g) = d(i
v
dg) = (di
v
+i
v
d)dg.
This proves Eq. (2.4).
Lie derivative of vector elds
A slight variation of the above denitions leads to a derivative operation on vector elds: Let
v be a vector eld and its ux. Take another vector eld w. We may then compare w
x
with
the image of w

(x)
under the tangent map (T

(x)
: T

(x)
M T
x
M. This leads to
the denition of the Lie derivative of vector elds,
L
v
: vect(M) vect(M),
w L
v
w,
(L
v
w)
x
= d

=0
(T

(x)
(w

(x)
). (2.5)
The components of the vector eld L
v
w = (L
v
w)
i
x
i
may be evaluated as
(L
v
w)
i
= d

=0
(T

(x)
(w

(x)
)(x
i
) = d

=0

x
k
w
k

(x)
=
_
d

=0

x
k
_
w
k
+d

=0
w
i

(x)
=
v
i
x
k
w
k
+
w
i
x
k
v
k
.
We thus arrive at the identication
L
v
w =
_
w
i
x
k
v
k

v
i
x
k
w
k
_

x
i
. (2.6)
60 Manifolds
Eq. (2.6) implies an alternative interpretation of the Lie derivative: Application of v to a function
f (

(M) produces another function v(f) (

(M). To this function we may apply the

vector eld w to produce the function w(v(f)) (wv)(f). Evidently, the formal combination
wv acts as a derivative operator on the function f. It is, however, not a derivation. To see this,
we choose a coordinate representation and obtain
(wv)(f) = w
i

x
i
v
j

x
j
f = w
i
v
j
x
i

x
j
f +w
i
v
j

2
x
i
x
j
f.
The presence of second order derivatives signals that wv is not a linear derivative operation.
However, consider now the skew symmetric combination
vw wv [v, w].
Application of this combination to f obtains:
(vw wv) =
_
w
i
x
k
v
k

v
i
x
k
w
k
_

x
i
f.
The second order derivatives have canceled out, which signals that [v, w] is a vector eld; the
space of vector elds admits a product operation to be explored in more detail below. Second,
our result above implies the important identication
[v, w] = L
v
w. (2.7)
EXAMPLE Consider the vector eld v = x
1

x
2 x
2

x
1. The ow of this eld is given by the
(linear) map:

(x) = O

x, where the matrix

O

=
_
cos sin
sin cos
_
.
Now, consider the constant vector eld w =
x
1. With w
i
x
=
i,1
, we obtain
L
v
w = (L
v
w)
i

x
i
= d

=0
(O

x)
i
x
j
w
j

x
i
= d

=0
(O

x)
i
x
1

x
i
=

x
2
.
Notice that L
v
w ,= 0, even at the origin where
t
0 = 0 is stationary.
PHYSICS (M) Consider the cotangent bundle, TM

of a manifold M of generalized coordinates

|q
i
. The bundle TM

is parameterized by coordinates |q
i
, p
i
where the canonical momenta cor-
respond to a Hamiltonian H = H(q, p).
Now, consider the so-called symplectic two-form =

i
dq
i
dp
i
. A the existence of a two-
form on the vector spaces T
(q,p)
TM

. (Dont be afraid of the accumulation of tangent/co-tangent

structures! Just think of TM

as a manifold with coordinates |(q

i
, p
i
) and T
(q,p)
TM

its tangent
space at (q, p).) enables us to switch between T
(q,p)
TM

and its dual space (T

(q,p)
TM

, i.e.
the space of one-forms on T
(q,p)
TM

. We apply this correspondence to the particular one-form

dH T(TM

: dene a vector eld X

H
T(TM

) by the condition
(X
H
, ) dH(). (2.8)
In words: substitution of a second vector, Y T(TM

), into ((X
H
, Y ) gives the same as evaluating
dH(Y ). The vector eld X
H
is called the Hamiltonian vector eld of H.
2.2 Tangent space 61
The Hamiltonian vector eld represents a very useful concept in the description of mechanical
systems. We rst note that the ow of the Hamiltonian vector eld represents describes the
mechanical trajectories of the system. We may check this by direct evaluation of (2.8) on the vectors

q
i
and

p
i
. Decomposing X
H
= (Q
H
, P
H
) into a coordinate and a momentum sector, we obtain
(X
H
,

q
i
) = dq
j
dp
j
(X
H
,

q
i
) = (P
H
)
i
!
= dH(

q
i
) =
H
q
i
.
In a similar manner, we get (Q
H
)
i
=
H
p
i
. Thus, the components of the Hamiltonian ow
(
Q
,
P
) obey the equations
d
t
(
Q
)
i
= (Q
H
)
i
=
H
p
i
,
d
t
(
P
)
i
= (P
H
)
i
=
H
q
i
, (2.9)
which we recognize as Hamiltons equations of motion.
Many fundamental statements of classical mechanics can be concisely expressed in the language of
Hamiltonian ows and the symplectic two form. For example, the time evolution of a function in phase
space is described by f
t
(x) f(
t
(x)), where we introduced the shorthand notation x = (q, p),
and assumed the absence of explicit time dependence in H for simplicity. In incremental form, this
assumes the form d
t
f
t
(x) = df(d
t

t
(x)) = df(X
H
) = X
H
(f) = L
X
H
f. For example, the statement
of the conservation of energy assumes the form d
t
H
t
(x) = d
H
(X
H
) = (X
H
, X
H
) = 0.
Phase space ow maps regions in phase space onto oth-
ers and this concept is very powerful in the description of
mechanical motion. By way of example, consider a subset
A TM

. We interpret A as the set of initial conditions of

a large number, N, of point particles, where we assume for
simplicity, that these particles populate A at constant den-
sity N/vol(A). (Notice that we havent dened yet, what the
volume of A is (cf. the gure.) The ow will transport the
volume A to another one,

(A). One may now ask, in which

way the density of particles changes in the process: will the evaporate to ll all of phase space? Or
may the distribution shrink down to a very small volume? The answer to these questions is given
by Liouvilles theorem, stating that the density of phase space points remains constant. In other
words, the volume of A remains constant under phase space evolution.
4
To properly formulate Liouvilles theorem, we rst need to clarify what is meant by the volume
of A. This is easy enough, because phase space comes with a canonical volume form. Much like a
metric induces a volume form, the symplectic two-form does so two: dene

1
n!

. .
n
= S

i
dq
i
dp
i
,
where S is an inessential sign factor. The volume of A is then dened as
vol(A) =
_
A
.
4
Notice that nothing is said about the shape of A. If the dynamics is suciently wild (chaotic) an initially regular A may
transform into a ragged object, whose laments cover all of phase space. In this case, particles do get scattered over phase
space. Nonetheless they stay conned in a structure of constant volume.
62 Manifolds
Liouvilles theorem states that
_
A
=
_

(A)
.
Using Eq. 1.40, this is equivalent to
_
A
=
_
A

, or to the condition

(x)
=
x
. Comparing
with the denition (2.3), we may reformulate this as a vanishing Lie derivative, L
X
H
= 0. Finally,
using the Leibniz property (i.e. the fact that the Lie-derivative separately acts on the -factors
constituting , we conclude that
Liouvilles theorem is equivalent to a vanishing of the Lie derivative L
X
H
= 0 of the
symplectic form in the direction of the Hamiltonian ow.
The latter statement is proven by straightforward application of Eq. (2.4): L
X
H
= i
X
H
d+di
X
H
=
0 + 0 = 0, where the rst 0 follows from the d = 0 and the second from di
X
H

(2.8)
= ddH = 0.
The discussion above heavily relied on the existence of the symplectic form, . In general, a manifold
equipped with a two-form that is skew symmetric and non-degenerate is called a symplectic manifold.
Many of the niceties that came with the existence of a scalar product also apply in the symplectic case
(think of the existence of a canonical mapping between tangent and cotangent space, or the existence
of a volume form.) However, the signicance of the symplectic form to the formulation of classical
mechanics goes much beyond that, as we exemplied above.
What remains to be show is that phase space actually is a symplectic manifold: our introduction of
above was ad hoc and tied to a specic system of coordinates. To see, why any cotangent bundle
is symplectic, consider the projection : T(TM

)
2.2.6 Orientation
As with the open subsets of R
n
discussed above, an orientation on a manifold may be introduced
by dening a nowhere vanishing nform . However, not for every manifold can such nforms
be dened, i.e. not every manifold is orientable (the Moebius strip being a prominent example
of a nonorientable manifold.)
Given an orientation (provided by a no-where vanishing
n-form), a chart (U, ) is called positively oriented, if the
corresponding coordinate frame obeys
x
(

x
1
, . . . ,

x
n
) >
0, or, equivalently, = fdx
1
dx
n
with a positive
function f. An atlas containing oriented charts is called ori-
ented. Orientation of an atlas is equivalent to the statement
that for any two sets of overlapping coordinate systems
x
i
and y
i
, det(x
i
/y
j
) > 0. (Exercise: show that for
the Moebius strip no orientable atlas exists.)
2.2 Tangent space 63
Figure 2.3 On the denition of manifolds with boundaries
2.2.7 Manifolds with boundaries
An ndimensional manifold is an object locally looking like an open subset of R
n
. Replacing R
n
by the half space
H
n
x = (x
1
, . . . , x
n
R
n
[x
n
0,
we obtain what is called a manifold with boundary. The boundary of M, M, is the set of all
points mapping onto the boundary of H, H
n
= x = (x
1
, . . . , x
n
R
n
[x
n
= 0,
M =
_
r

1
r
(
r
(M) H
n
),
where the index r runs over all charts of an atlas. (One may show that this denition is inde-
pendent of the chosen atlas.)
EXAMPLE Show that the unit ball B
n
= |x = (x
1
, . . . , x
n
) R
n
[(x
1
)
2
+ + (x
n
)
2
1 in
ndimensions is a manifold whose boundary is the unit sphere S
n1
.
With the above denitions, M is (a) a manifold of dimensionality n1 which (b) is bound-
aryless, M = . As with the boundary of cells discussed earlier, the boundary M of a
manifold inherits an orientation from the bulk, M. To see this, let x M be a boundary point
and v = v
i
x
i
T
x
M be a tangent vector. If v
n
= 0, v T
x
(M) is tangent to the boundary
(manifold). If v
n
< 0, v is called a outward normal vector. (Notice, however, that normal
does not imply orthogonality; we are not using a metric yet.)
With any outward normal vector n, the (n 1)form i
n

n1
M then denes an
orientation of the boundary manifold.
64 Manifolds
2.2.8 Integration
Partition of unity
An atlas (U

, ) of a manifold M is called locally nite, if every x M possesses a neigh-

bourhood such that the number of charts with U

U ,= is nite. Locally nite atlases always

exist.
A partition of unity subordinate to a locally nite covering of M is a family of mappings h

with the following properties:

(i) x M, h

(x) 0,
(ii) supp(h

) U

,
(iii) x M,

(x) = 1.
(Due to the local niteness of the covering, the sum in 3. contains only a nite number of
terms.)
EXAMPLE Let |B

[ I be a countable set of unit balls centered at points x

R
n
covering
R
n
. Dene the functions
f

_
exp((1 [x x

[
2
)
1
) , [x x

[ 1,
0 , else
Then the functions
h

(x)
f

(x)

(x)
dene a partition of unity in R
n
.
Integration
Let M be a manifold (with or without boundary), (U

, ) a locally nite atlas, h

a
partition of unity and
n
M an nform. For an integration domain U U

contained in a
single chart domain, the integral over is dened as
_
U
=
_
(U)

1
,
where the second integral is evaluated according to our earlier denition of integrals over open
subsets of R
n
. If U is not contained in a single chart, we dene
_
U
=

_
UU

.
One may show that the denition does not depend on the reference partition of unity.
Finally, Stokes theorem assumes the form

p1
M :
_
M
d =
_
M
.
(For a manifold without boundary, the l.h.s. vanishes.)
2.3 Summary and outlook 65
Metric
A metric on a manifold is a nondegenerate symmetric bilinear form g
x
on each tangent space
T
x
M which depends smoothly on x (i.e. for two vector elds v
1
, v
2
the function g
x
(v
1
(x), v
2
(x))
depends smoothly on x.) In section 1.3.5 we introduced metric structures on open subsets of
U R
n
. Most operations relating to the metric, the canonical isomorphism T
x
U
J
(T
x
U)

,
the Hodge star, the coderivative, the denition of a volume form, etc. where dened locally.
Thanks to the local equivalence of manifolds to open subsets of R
n
, these operations carry over
to manifolds without any changes. (To globally dene a volume form, the manifold must be
orientable.)
There are but a few global aspects where the dierence between a manifold and an open
subset of R
n
may play a role. While, for example, it is always possible to dene a metric of
any given signature on an open subset of R
n
, the situation on manifolds is more complex, i.e.
a global metric of predesignated signature need not exist.
2.3 Summary and outlook
In this section, we introduced the concept of manifolds to describe geometric structures that
cannot be globally identied with open subsets of R
n
. Conceptually, all we had to do to achieve
this generalization was to patch up the local description of a manifold provided by charts and
the ensuing dierentiable structures to a coherent global description. By construction, the
transition from one chart to another is mediated by dierentiable functions between subsets of
R
n
, i.e. objects we know how to handle. In the next section, we will introduce a very important
family of dierentiable manifolds, viz. manifolds carrying a group structure.
3
Lie groups
In this chapter we will introduce Lie groups, a class of manifolds of paramount importance in
both physics and mathematics. Loosely speaking, a Lie group is a manifold carrying a group
structure. Or, changing the perspective, a group that is at the same time a manifold. The
linkage of concepts from group theory and the theory of dierential manifolds generates a rich
mathematical structure. At the same time, Lie groups are the natural mathematical objects
to describe symmetries in physics, notably in quantum physics. In section In section 1.4 above,
we got a rst impression of the importance of Lie groups in quantum theory, when we saw how
these objects implement the concept of gauge transformations. In this section, however, the
focus will be on the mathematical theory of Lie groups.
3.1 Generalities
A (nite dimensional, real) Lie group, G, is a dierentiable manifold carrying a group structure.
One requires that the group multiplication, GG G, (g, g

) gg

and the group inversion,

G G, g g
1
be dierentiable maps.
Notice that there are lots of mathematical features that may be attributed to a Lie group
(manifold):
manifold: dimensionality, compactness, conectedness, etc.
group: abelian, simplicity, nilpotency, etc.
The joint group/manifold structure entails two immediate further denitions: a Lie subgroup
H G is a subgroup of G which is also a submanifold. We denote by e the unit element
of G and by G
e
the connected component of G. The set G
e
is a Lie subgroup of the same
dimensionality as G (think why!)
A few elementary (yet important) examples of Lie groups:
There are but two dierent connected onedimensional Lie groups: the real numbers R with
its additive group structure is a simply connected noncompact abelian Lie group. The unit
circle z C[[z[ = 1 with complex multiplication as group operation is a compact abelian
nonsimply connected Lie group (designated by U(1) or SO(2) depending on whether one
identies C with R
2
or not.)
The general linear group, GL(n), i.e. the set of all real (nn)matrices with nonvanishing
determinants is a Lie group. Embedded into R
n
it contains two connected components,
GL
+
(n) and GL

(n), the set of matrices of positive and negative determinant, respectively

66
3.2 Lie group actions 67
(the former containing the group identity, 1
n
.) GL(n) is noncompact, nonconnected, and
nonabelian.
The orthogonal group, O(n) is the group of all real orthogonal matrices, i.e. O(n) =
O GL(n)[O
T
O = 1
n
. It is the maximal compact subgroup of GL(n). It is of dimension
n(n 1)/2 and nonconnected. Similarly, the special orthogonal group, SO(n) = O
O(n)[ det O = 1 is the maximal compact subgroup of GL
+
(n). It is connected yet not simply
connected. (Think of SO(2).)
The special unitary group, SU(n) is the group of all complex valued (n n)matrices, U,
obeying the conditions U

U = 1
n
and det U = 1. Alternatively, one may think of SU(n) as
a real manifold viz. as a real subgroup of Gl(2n). It is of dimension n
2
1, compact, and
simply connected.
By way of example, consider SU(2). We are going to show that SU(2) is isomorphic to
the real manifold S
3
, the three sphere. To see this, write an element U SU(2) (in complex
representation) as
U =
_
a b
c d
_
.
The conditions U

U = 1
2
and det U = 1 translate to a a +c c = 1, b

b +d

d = 1, a

b +c

d = 0,
and ad cb = 1. Dening a = x
1
+ ix
2
and c = x
3
+ ix
4
, x
1
, . . . , x
4
R, these conditions
are resolved by
U =
_
x
1
+ix
2
x
3
+ix
4
x
3
+ix
4
x
1
ix
2
_
,
4

i=1
(x
i
)
2
= 1.
This representation establishes an dieomorphism between SU(2) and the three sphere, S
3
=
(x
1
, x
2
, x
3
, x
4
) R
4
[

4
i=1
(x
i
)
2
= 1.
3.2 Lie group actions
3.2.1 Generalities
Let G be a Lie group and M an arbitrary manifold. A (left) action
1
of G on M is a dierentiable
mapping,
: GM M,
(g, x) (g, x)
g
(x),
assigning to each group element a smooth map
g
: M M. The composition of these maps
must be compatible with the group structure multiplication in the sense that

gg
=
g

g
,
e
= id
M
.
1
Sometimes group actions are also called group representations. However, we prefer to reserve that terminology for the linear
group actions to be dened below.
68 Lie groups
EXAMPLE Let G = SO(3) be the three dimensional rotation group and M = S
2
the twosphere.
The group G acts on M by rotating the unitvectors constituting S
2
. More generally, the isometries
of a Riemannian manifold dene a group acting on the manifold.
A right action of G on M is dened in the same manner, only that the compatibility relation
reads

gg
=
g

g
.
Notation: Instead of
g
we will occasionally write (g) or just g. For brevity, the left/right action
of g on x M is often designated as gx/xg
If is a left action, then g
g
1 denes a right action. For a xed x M, we dene the
map,
bit
x
: G M,
g bit
x
g
g
1x. (3.1)
The orbit of x is the image of bit
x
,
orbit(x) bit
x
(G).
If orbit(x) = M, the action of the group is called transitive. (Exercise: why is it sucient to
prove transitivity for an arbitrary reference point?) For a transitive group action, two arbitrary
points, x, y M are connected by a group transformation, y =
g
x. An action is called faithful
if there are no actions other than
e
acting as the identity transform: (x M :
g
x = x)
g = e. It is called free i bit
x
is injective for all x M. (A free action is faithful.) The isotropy
group of an element x M, is dened as
I(x) g G[
g
x = x.
The action is free i the isotropy group of all x M contains only the unit element, I(x) = e.
EXAMPLE The action of the rotation group SO(3) on the twosphere S
2
is transitive and faithful,
but not free. The isotropy group is SO(2).
3.2.2 Action of a Lie group on itself
A Lie group acts on itself, M = G, in a number of dierent and important ways. The action by
left translation is dened by
L
g
: G G,
h gh, (3.2)
i.e. g acts by left multiplication. This representation is transitive and free. Second, it acts on
itself by the inner automorphism,
aut
g
: G G,
h ghg
1
. (3.3)
3.3 Lie algebras 69
In general, this representation is neither transitive nor faithful. The stability group, I(e) = G.
Finally, the right translation is a right action dened by
R
g
: G G,
h hg. (3.4)
Right and left translation commute, and we have aut
g
= L
g
R
g
1.
3.2.3 Linear representations
An action is called a linear representation or just representation if the manifold it acts upon
is a vector space, M = V , and if all dieomorphisms
g
are linear. Put dierently, a linear
representation is a group homeomorphism G GL(V ). Linear representations are not transitive
(think of the zero vector.) They are called irreducible representations if V does not possess

g
invariant subspaces other than itself (and space spanned by the zero vector.)
EXAMPLE The fundamental representation of SO(3) on R
3
is irreducible.
A group action is called an ane representation if it acts on an ane space and if all
g
are ane maps.
2
Two more remarks on representations,
Given a linear (or ane) representation, , a general left action may be constructed as

g

F
g
F
1
, where F : V V is some dieomorphism;

g
need no longer be linear.
Conversely, given a left action
g
it is not always straightforward to tell whether
g
is a linear
representation

g
disguised by some dieomorphism,
g
= F
g
F
1
.
Depending on the dimensionality of the vector space V , one speaks of a nite or an innite
dimensional representation. For example, given an open subset of U R
n
the Lie group
GL(n) acts on the frame bundle of U the innite dimensional vector space formed by all
frames by left multiplication. This is an innite dimensional representation of GL(n).
3.3 Lie algebras
3.3.1 Denition
Recall that a (nite or innite) algebra is a vector space V equipped with a product operation,
V V V . A Lie algebra is an algebra whose product (the bracket notation is standard for
Lie algebras)
[ , ] : V V V
(v, w) [v, w]
satises certain additional properties:
[ , ] is bilinear,
2
A map is called ane if it is the sum of a linear map and a constant map.
70 Lie groups
skewsymmetric: v, w V, [v, w] = [w, v], and satises the
Jacobi identity, u, v, w V, [u, [v, w]] + [w, [u, v]] + [v, [w, u]] = 0.
EXAMPLE In section 2.2.5 we have introduced the Lie derivative Eq. (2.5) as a derivative operation
on the innite dimensional space of vector elds on a manifold, vect(M). Alternatively, we may think
of the Lie derivative as a product,
[ , ] : vect(M) vect(M) vect(M),
(v, w) L
v
w = [v, w] (3.5)
assigning to two vector elds a new one, [v, w]. The above product operation is called the Lie
bracket of vector elds. Both, skew symmetry and the Jacobi identity are immediate consequences
of Eq. (2.6). We thus conclude that vect(M) is an innite dimensional Lie algebra, the Lie algebra
of vector elds on a manifold.
3.3.2 Lie algebra of a Lie group
Let A vect(G) be a vector eld on a Lie group G. A is a left invariant vector eld if it is
invariant under all left translations,
g, h G : (TL
g
)
h
A
h
= A
gh
.
EXAMPLE The left invariant vector elds on the abelian Lie group R
n
are the constant vector
elds.
Due to the linearity of TL
g
, linear combinations of left invariant vector elds are again left
invariant, i.e. the set of left invariant vector elds forms a linear space, here denoted by g.
However, as we are going to show below, g carries much more mathematical structure than
just linearity. To see this, a bit of preparatory work is required: Let M be a manifold and
F : M M a smooth map. A vector eld v on M is called invariant under the map F, if x
M : TF
x
v
x
= v
F(x)
. Recalling that TF
x
v
x
(f) = v
x
(f F), and that v
F(x)
(f) = (v(f) F)(x)
this condition may be rewritten as
f (

(M) : v(f F) = v(f) F.

Now, consider two vector elds A, B g. Applying the above invariance criterion to F = L
g
and considering the two cases v = A, f = B(

f), and v = B, f = A(

f), where

f (

(G), we
obtain
(AB)(

f) L
g
= A(B(

f)) L
g
= A(B(

f) L
g
) = A(B(

f L
g
)) = (AB)(

f L
g
),
(BA)(

f) L
g
= B(A(

f)) L
g
= B(A(

f) L
g
) = B(A(

f L
g
)) = (BA)(

f L
g
).
Subtraction of these formulas gives
(AB BA)(

f) L
g
= (AB AB)(

f L
g
),
which shows that
A, B g [A, B] g,
3.3 Lie algebras 71
i.e. that the space of left invariant vector elds, g, forms a Lie subalgebra of the space of vector
elds. g is called the Lie algebra of G.
The left action of a Lie group on itself is transitive. Specically, each element g may be
reached by left multiplication of the unit element, g = L
g
e = ge. As we shall see, this implies
an isomorphism of the tangent space T
e
G onto the Lie algebra of the group. Indeed, the left
invariance criterion implies that
A
g
= (TL
g
)
e
A
e
, (3.6)
i.e. the value of the vector eld at arbitrary g is determined by its value in the tangent space at
unity, T
e
G. As a corollary we conclude that
The dimension of the Lie algebra, dim(g) = dim(G), and
On g there exists a global frame, B
i
, i = 1, . . . , dim(g), where (B
i
)
g
= (TL
g
)
e
E
i
and
E
i
is a basis of T
e
G; Lie group manifolds are parallelizable.
EXAMPLE By way of example, let us consider the Lie group GL(n). (Many of the structures
discussed below instantly carry over to other classical matrix groups.) GL(n) R
n
2
is open in R
n
2
and can be represented in terms of a global set of coordinates. The standard coordinates are, of
course, x
ij
(g), where x
ij
is the ith row and jth column of the matrix representing g. A tangent
vector at e can be represented as (remember: summation convention)
A
e
= a
ij

x
ij
.
In order to compute the corresponding invariant vector eld A
g
, we rst note that the L
g
acts by
left matrixmultiplication,
x
ij
(L
g
h) = x
ik
(g)x
kj
(h).
Using Eq. (1.14), we then obtain
((TL
g
)
e
A
e
)
ij
=
x
ij
(L
g
h)
x
kl
(h)

h=e
a
kl
= x
ik
(g)a
kl
,
Or A
g
= gA
e
, where g (A
e
) is identied with the n n matrices |x
ij
(g) (|a
ij
) and matrix
multiplication is implied. Another representation reads
A
g
= x(g)
ik
a
kj

x
ij
x
ik
a
kj

x
ij
.
This latter expression may be used to calculate the Lie bracket of two left invariant vector elds,
[A, B] = x
ij
a
kj

x
ij
x
lm
b
mn

x
ln
(a b) = x
ik
(ab ba)
kj

x
ij
,
where |b
ij
are the matrix indices identifying B and ab is shorthand for standard matrix multiplication.
This result (a) makes the left invariance of [A, B] manifest, and (b) shows that the commutator [A, B]
simply obtains by taking the matrix commutator of the coordinate matrices [a, b], (more formally, the
identication A a = |a
ij
is a homomorphism of the Lie algebras (left invariant vector elds, Lie
bracket) and (ordinary matrices, matrix commutator).
When working with (left invariant) vector elds on Lie groups, it is often convenient to employ
the equivalence classes of curves denition of vector elds. For a given A T
e
G, there are
72 Lie groups
many curves
A
(t) tangent to A at t = 0:
A
(0) = e, d
t

t=0

A
(t) = A. Presently, however,
it will be convenient to consider a distinguished curve, viz. g
A,t

A,t
(e), where
t
(e) is the
ow of the left invariant vector eld A
g
corresponding to A. We note that the vector eld A
evaluated at g
A,s
, A
g
A,s
, aords two alternative representations. On the one hand, A
g
A,s
=
T(L
g
A,s
)
e
A
e
= d
t

t=0
g
A,s
g
A,t
and on the other hand, A
g
A,s
= d
s
g
A,s
= d
t

t=0
g
A,s+t
. This
implies that
g
A,s
g
A,t
= g
A,s+t
,
i.e. g
A,t
is a one parameter subgroup of G. Later on, we shall see that these subgroups play
a decisive role in establishing the connection between the Lie algebra and the global structure
of the group. Presently, we only note that the left invariant vector eld A
g
may be represented
as (cf. Eq. (3.6))
A
g
= d
t

t=0
gg
A,t
.
Put dierently, the ow of the left invariant vector eld
3
,
A,t
acts as

A,t
: G G,
g gg
A,t
,
for
A,0
= id
G
and d
t

t=0

A,t
(g) = A
g
, as required. This latter representation may be used
to compute the Lie derivative of two left invariant vector elds A and B, L
A
B. With B
g
=
d
s

s=0
gg
B,s
, we have
(L
A
B)
e
= d
t

t=0
(T
t
)

A,t
(e)
(B

A,t
(e)
) = d
2
s,t

s,t=0
(T
t
)
g
A,t
g
A,t
g
B,s
=
= d
2
s,t

s,t=0
g
A,t
g
B,s
g
A,t
.
Using that g
t
g
t
= g
tt
= g
0
= e, i.e. that g
t
= g
1
t
, we conclude that the Lie derivative of
two left invariant vector elds is given by
(L
A
B)
e
= d
2
s,t

s,t=0
g
A,t
g
B,s
g
1
A,t
. (3.7)
3.4 Lie algebra actions
3.4.1 From the action of a Lie group to that of its algebra
Let F
a
: M M, a R be a one parameter family of dieomorphisms, smoothly depending
on the parameter a. Assume that F
0
= id
M
. For a innitesimal, we may write,
F
a
(x) x +a

a

a=0
F
a
(x) +O(a
2
).
3
Notice that g
A,t
was constructed from the ow through the origin whilst we here dene the global ow.
3.4 Lie algebra actions 73
This shows that, asymptotically for small a, F
a
may be identied with a vector eld (whose
components are given by
a

a=0
x
i
(F
a
(x)). Two more things we know are that (a) the innites-
imal variant of Lie group elements (the elements of the tangent space at unity) constitute the
Lie algebra, and (b) Lie group actions map Lie group elements into di(M). Summarizing
G
repr.
di(M),
innit. innit.
g
?
vect(M).
The diagram suggests that there should exist an innitesimal variant of Lie group representa-
tions mapping Lie algebra elements onto vector elds. Also, one may expect that this mapping
is a Lie algebra homomorphism, i.e. is compatible with the Lie bracket.
Identifying g = T
e
G, the representation of the Lie algebra, i.e. an assignment g A v
vect(M) may be constructed as follows: consider x M. The group G acts on x as x
g
(x).
We may think of

(x) : G M, g
g
(x) as a smooth map from G to M. Specically, the
unit element e maps onto x. Thus, the tangent mapping T

(x) maps g into T

x
M. For a given
Lie group representation we thus dene the induced representation of the Lie algebra as
:g vect(M),
A
A
,
(
A
)
x
= (T

(x))
e
(A).
INFO It is an instructive exercise to show that is a Lie algebra (anti) homomorphism. Temporarily
denoting
A


A, what we need to check is that

[A, B] = [

A,

B]. Again, it will be convenient to
work in the curve representation of vector elds. With A = d
t

t=0
g
A,t
and B = d
s

s=0
g
B,s
, we have

A
x
= d
t

t=0
g
A,t
x, where we denote the group action on M by
g
(x) gx. Similarly, the ow of
the vector eld

A is given by
A,t
(x) = g
A,t
x. We may now evaluate the Lie bracket of the image
vector elds as
[

A,

B]
x
= (L
A

B)(x) = d
t

t=0
(T
A,t
)

A,t
(x)

B

A,t
(x)
= d
2
s,t

s,t=0
g
A,t
g
B,s
g
A,t
(x) =
= d
2
s,t

s,t=0
g
A,t
g
B,s
g
A,t
x.
Comparison with (3.7) shows that [

A,

B]
x
=

L
A
B
x
=

[A, B]
x
, i.e. g vect(M) is a Lie algebra
anti (the sign) homomorphism.
3.4.2 Linear representations
Let : G GL(V ) di(V ) be a linear representation of a Lie group. To understand
what vector elds describe the Lie algebra, let A T
e
G be a Lie algebra element and g
A,t
be a representing curve. The representation
g
(v) M
g
v maps group elements g onto linear
transformations M
g
GL(V ). In a given basis, M
g
is represented by an (nn)matrix M
ij
g
.
74 Lie groups
Specically, the generating curve g
A,t
M
g
A,t
is represented by a matrix valued curve. We
thus nd
(
A
)
v
= d
t

t=0
M
g
A,t
v.
The vector eld (
A
)
v
depends linearly on v. Dening X
A
d
t

t=0
M
g
A,t
gl(V ) (X
A
is an
element of the Lie algebra, gl(V ) of the group GL(V )), we may write have (
A
)
v
X
A
v.
Adjoint representation
In section 3.2.2 we have seen that a Lie group acts on itself by the inner automorphism, aut,
where aut
g
(h) = ghg
1
. Associated to this action we have a linear representation of G on g,
the adjoint representation, Ad, of the group on its Lie algebra:
Ad : Gg g,
(g, A) Ad
g
(A) (Taut
g
)
e
A,
where we identify A T
e
G as an element of the tangent vector space. Since aut
g
(e) = e, the
image of A, (Taut
g
)
e
A T
e
G is again in the Lie algebra. With g
A,t
a curve representing A,
we have Ad
g
(A) = d
t

t=0
gg
A,t
g
1
.
The corresponding linear representation of the Lie algebra is denoted the adjoint represen-
tation of the Lie algebra,

Ad ad. The adjoint representation is a representation of the Lie
algebra on itself. According to our previous discussion, we have
ad
A
(B) = d
2
t,s

t,s=0
g
A,t
g
B,s
g
1
A,t
= [A, B].
The result
ad
A
(B) = [A, B] (3.8)
plays a pivotal role in the representation theory of Lie groups. Let T
a
be a basis of the Lie
algebra. The expansion coecients f
abc
of [T
a
, T
b
],
[T
a
, T
b
] f
abc
T
c
(3.9)
are called the structure constants of the Lie algebra. Two Lie algebras are isomorphic, if they
share the same structure constants.
3.5 From Lie algebras to Lie groups
Above, we have seen how plenty of structure information is encoded in the Lie algebra g. In
this nal section, we will show that this information actually suces to recover the structure
of the whole group G, at least in some vicinity of the unit element. The section contains the
mathematics behind the physicists strategy to generate a global transformation (an element
of the Lie group or one of its actions) out of innitesimal transformations, or transformation
generators (an element of the Lie algebra or one of its actions.)
3.5 From Lie algebras to Lie groups 75
3.5.1 The exponential mapping
In section 3.3.2 we have introduced the ow g
A,t
of a left invariant vector eld A through the
origin. We now dene the map
exp : T
e
G G,
A exp(A) g
A,1
. (3.10)
Let us try to understand the background of the denotation exp. We rst note that for s R,
g
sA,t
= g
A,st
. Indeed, g
sA,t
solves the dierential equation d
t
g
sA,t
= sAg
sA,t
with initial
condition g
sA,0
= e. However, by the chain rule, d
t
g
A,st
= sd
st
g
A,st
= sAg
A,st
(with initial
condition g
A,s0
= e.) Thus, g
A,st
and g
sA,t
solve the same rst order initial value problem which
implies their equality. Using the homogeneity relation g
sA,t
= g
A,st
, we nd that
exp(sA) exp(tA) = g
sA,1
g
tA,1
= g
A,s
g
A,t
= g
A,s+t
= g
(s+t)A,1
= exp((s +t)A),
i.e. the function exp satises the fundamental relation of exponential functions which explains
its name. The denotation exp hints at another important point. Dening monomials of Lie
algebra elements A
n
in the obvious manner, i.e. through the nfold application of the vector A,
we may tentatively try the power series representation
exp(A) =

n=0
1
n!
A
n
. (3.11)
To see that the r.h.s. of this equation indeed does the job, we use that exp(tA) = g
t,A
must
satisfy the dierential equation d
t
exp(tA) = exp(tA)A = A
exp(tA)
. It is straightforward to
verify that the r.h.s. of Eq. (3.11) solves this dierential equation and, therefore, appears to
faithfully represent the exponential function.
4
INFO We are using the cautious attribute appears to because the interpretation of the r.h.s. of
the power series representation is not entirely obvious. A priori, monomials A
n
neither lie in the Lie
algebra, nor in the group, i.e. the actual meaning of the series requires interpretation. In cases, where
G GL(n) is (subset of) the matrix group GL(n) no diculties arise: certainly, A
n
is a matrix
and det exp(A) = exp(ln det exp(A)) = exp tr ln exp(A) = exp tr(A) ,= 0 is nonvanishing, i.e.
exp(A) GL(n) as required. For the general interpretation of the power series interpretation, we
refer to the literature.
3.5.2 Normal coordinates
The exponential map provides the key to extrapolating from local structures (Lie algebra) to
global ones (Lie group). Let us quote a few relevant facts:
In general, the exponential map is neither injective, nor surjective. However, in some open
neighbourhood of the origin, exp denes a dieomorphism. This feature may be used to dene
4
Equivalently, one might have argued that the fundamental relation exp(x) exp(y) = exp(x + y) implies the power series
representation.
76 Lie groups
a specic set of local coordinates, the socalled normal coordinates: Assume that a basis of
T
e
G has been chosen. The normal coordinates of exp(tA) are then dened by exp(tA)
i
tA
i
,
where A
i
are the components of A T
e
G in the chosen basis. (For a proof of the faithfulness
of this representation, see the info block below.)
There is one and only one simply connected simply connected Lie group G with Lie(G) = g.
For every other connected Lie group H with Lie(H) = g, there is a group homomorphism
G H whose kernel is a discrete subgroup of H. (Example: g = R, G = R, H = S
1
, with
kernel Z (2).) G is called the universal covering group of H.
INFO We wish to prove that, locally, exp denes a dieomorphism. To this end, let g
i
(g) be an
arbitrary coordinate system and
i
(g) be the normal coordinates. The (local) faithfulness of the latter
is proven, once we have shown that the Jacobi matrix
g
i

j
is nonsingular at the origin. Without loss
of generality, we assume that the basis spanning T
e
G is the basis of coordinate vectors

g
i
of the
reference system.
Thus, let
i
tA
i
be the normal coordinates of some group element g. Now consider the specic
tangent vector B

t

t=0
g T
e
G. Its normal coordinates are given by B
i
n
=

t

i
= A
i
. By
denition, the components of the group element g in the original system of normal coordinates will
be given by g
i
(g) = exp
i
(tA). The components of the vector B are given by B
i
= d
t

t=0
exp
i
(tA) =
d
t

t=0
g
i
1,tA
= d
t

t=0
g
i
t,A
= A
i
. Thus, B
i
= B
i
n
coincide. At the same time, by denition, B
i
=
g
i

j
B
j
n
, implying that
g
i

j
= id has maximal rank, at least at the origin. By continuity, the coordinate
transformation will be nonsingular in at least an open neighbourhood of g = e.
EXAMPLE The group SU(2) as the universal covering group of SO(3). The groups SU(2)
and SO(3) have isomorphic Lie algebras. The three dimensional algebra so(3) consists of all three
dimensional antisymmetric real matrices. It may be conveniently spanned by the three matrices
T
1
=
_
_
0 0 0
0 0 1
0 1 0
_
_
, T
2
=
_
_
0 0 1
0 0 0
1 0 0
_
_
, T
3
=
_
_
0 1 0
1 0 0
0 0 0
_
_
,
generating rotations around the 1, 2 and 3 axis, respectively. The structure constants in this basis,
[T
i
, T
j
] =
ijk
T
k
coincide with the fully antisymmetric tensor
ijk
. In contrast, the Lie algebra
su(2) of SU(2) consists of all antiHermitean traceless two dimensional complex matrices. It is three
dimensional and may be spanned by the matrices
i

i
2

i
, where

1
=
_
0 1
1 0
_
,
2
=
_
0 i
i 0
_
,
3
=
_
1 0
0 1
_
are the familiar Pauli matrices. As with so(3), we have [
i
,
j
] =
ijk

k
, i.e. the two algebras are
isomorphic to each other.
Above, we have seen that the group manifold SU(2) S
3
is isomorphic to the threesphere, i.e.
it is simply connected. In contrast, the manifold SO(3) is connected yet not simply connected.
To see this, we perform a gedanken experiment: consider a long twodimensional exible strip
embedded in three dimensional space. Let the (unit)length of the strip be parameterized by [0, 1]
and let v() be the vector pointing in the narrow direction of the strip. Assuming the width of
the strip to be uniform, the mapping v(0) v() is mediated by an SO(3) transformation O().
Further, O() denes a curve in SO(3) Assuming that v(0) | v(1), this curve is closed. As we
will see, however, it cannot be contracted to a trivial (constant) curve if v(1) obtains from v(0) by
3.5 From Lie algebras to Lie groups 77
a 2 rotation (in which case our strip looks like a Moebius strip.) However, it can be contracted, if
v(1) was obtained from v(0) by a 4 rotation.
To explicate the connection SU(2) SO(3), we introduce the auxiliary function
f : R
3
su(2),
v = v
i
e
i
v
i

i
.
For an arbitrary matrix U SU(2), we have Uf(v)U
1
su(2) (Uf(v)U
1
is antiHermitean and
traceless.) Further, the map f is trivially bijective, i.e. it has an inverse. We may thus dene an action
of SU(2) in R
3
as

U
v f
1
(Uf(v)U
1
).
Due to the linearity of f,
U
actually is a linear representation. Furthermore
U
: R
3
R
3
SO(3),
i.e. : SU(2) SO(3), U
U
denes a map between the two groups SU(2) and SO(3). (To see
that
U
SO(3), we compute the norm
U
v. First note that for A su(2), [f
1
(A)[
2
= 4 det(A).
However, det(Uf(v)U
1
) = det(f(v)), from where follows the normpreserving of
U
. One may
also check that
U
preserves the orientation, i.e.
U
is a norm and orientation preserving map,

U
SO(3).)
It can be checked that the mapping : SU(2) SO(3) is surjective. However, it is not injective.
To see this, we need to identify a group element g SU(2) such that v : g
1
f(v)g = f(v), or,
equivalently, h SU(2) : g
1
hg = h. The two group elements satisfying this requirement are
g = e and g = e. We have thus found hat the universal covering group of SO(3) is SU(2) and that
the discrete kernel of the group homomorphism SU(2) SO(3) is |e, e.