LECTURE 1

1.1. Brief review of notions from Topology and Analysis

Did I use to know all these things? Was the summer so long?

We start by recalling some notions from Analysis and Topology, which will be
used throughout the course.

Some Topology
Definition 1.1.1. A topology on a set X is a set T of subsets of X, whose
elements are called open subsets, satisfying the axioms:
[
∅, X ∈ T , U, V ∈ T ⇒ U ∩ V ∈ T , (Ui ∈ T , ∀i ∈ I) ⇒ Ui ∈ T .
i∈I

The pair (X, T ) is called a topological space.

Definition 1.1.2. Let Y be a subset of a topological space (X, T ). The induced
topology on Y , also called the subspace topology, is given by:
T |Y := {U ∩ Y : U ∈ T }.
Definition 1.1.3. Let (X, TX ) and (Y, TY ) be topological spaces. A function
f : X −→ Y
−1
is called continuous if f (U ) is open in X for every open set U in Y .
The function f is called a homeomorphism if it is continuous, bijective and
its inverse f −1 : Y → X is also continuous.
Definition 1.1.4. A topological space (X, T ) is called Hausdorff if every two
distinct points can be separated by two open sets; in other words, for every x, y ∈ X,
with x 6= y, there exist U, V ∈ T such that x ∈ U , y ∈ V and U ∩ V = ∅.
Definition 1.1.5. A topological space (X, T ) is called second countable if there
exists a countable family of open sets {Ui }i∈N with the following
S property: for every
open set U there exists an index set I ⊂ N such that U = i∈I Ui . In other words,
the topology has a countable basis.
Next, we recall the definition of the Euclidean topology on
Rn = {x = (x1 , . . . , xn ) : x1 , . . . , xn ∈ R}.
9
10 IOAN MĂRCUT
, , MANIFOLDS

Definition 1.1.6. (1) The open ball in Rn of radius ε > 0 and center x0 ∈ Rn
is the set
Bε (x0 ) = {x ∈ Rn : |x − x0 | < ε},
where | · | is the usual Euclidean length:
p
|x − y| := (x1 − y 1 )2 + . . . + (xn − y n )2 .
(2) A subset U ⊂ Rn is called open if for every x0 ∈ U there exists a positive
number ε > 0 such that Bε (x0 ) ⊂ U . The collection of all such sets forms the
so-called Euclidean topology on Rn .

Smooth functions
Let U ⊂ Rn be an open set. A function f : U → Rm is called differentiable at a
point x ∈ U , if there exists a linear map
dx f : Rn −→ Rm
such that:
|f (x + h) − f (x) − dx f (h)|
lim = 0.
|h|→0 |h|
The map dx f is called the differential of f at x. If dx f exists, then it is also
uniquely determined, since it can be described as the directional derivative of f :
f (x + εv) − f (x)
dx f (v) = lim , v ∈ Rn .
ε→0 ε
Denote the components of f by f = (f 1 , . . . , f m ). The matrix of dx f in the standard
bases of Rn and Rm is given by the Jacobian matrix:
∂f 1 ∂f 1 ∂f 1
 
∂x12 ∂x22 . . . ∂x n
 ∂f ∂f ∂f 2 
dx f =  ∂x1
 ∂x2 . . . ∂x n
.

 ... ... ... ... 
∂f m ∂f m m

∂x1 ∂x2 . . . ∂f
∂xn
Recall also the chain rule for the differential:
Proposition 1.1.7 (The chain rule). Consider two maps f : U → Rm , g : V → U ,
where U ⊂ Rl and V ⊂ Rn are open sets. If g is differentiable at x ∈ V and f
is differentiable at g(x) ∈ U then f ◦ g is differentiable at x, and its differential is
given by
dx (f ◦ g) = (dg(x) f ) ◦ dx g : Rn −→ Rm .
In terms of partial derivatives the chain rule takes the form:
l
∂(f ◦ g) X ∂f ∂g j
i
(x) = j
(g(x)) i (x), for all 1 ≤ i ≤ n.
∂x j=1
∂y ∂x

The first, and most general class of maps are continuous maps; these are called
also functions of class zero, or just C 0 -functions.
Denote the space of all linear maps from Rn to Rm by Lin(Rn , Rm ). Writing
the linear maps in the standard bases, this space can be identified with the space of
n×m-matrices, hence with Rn×m . Using this identification, we endow Lin(Rn , Rm )
with the Euclidean topology on Rn×m .
 LECTURE 1. 11

Definition 1.1.8. Let U ⊂ Rn be an open set. A map f : U → Rm is said to

be a C 1 -function if f is differentiable at every x ∈ U , and its differential is a
continuous function from U to Lin(Rn , Rm ):

df : U → Lin(Rn , Rm ), x 7→ dx f ∈ Lin(Rn , Rm ).

Functions of class C k are defined inductively. We say that f : U → Rm is

a C -function, for k ≥ 2, if it is differentiable and its differential is a C k−1 -
k

function.

Let us recall some standard properties of C k -functions:

Proposition 1.1.9. Let U ⊂ Rn be an open set, and let

f = (f 1 , . . . , f m ) : U −→ Rm

be a function.
(1) If f is a C k -function then it is also a C k−1 -function.
(2) The function f is of class C k if and only if its partial derivatives of order ≤ k:

∂lf i ∂f i
  
∂ ∂
= . . . . . . : U → R,
∂xj1 ∂xj2 . . . ∂xjl ∂xj1 ∂xj2 ∂xjl

for 1 ≤ i ≤ m, 0 ≤ l ≤ k, 1 ≤ j1 , . . . , jl ≤ n
exist and are continuous on U .
(3) If f is of class C k then its partial derivatives “commute”, meaning that:

∂lf i ∂lf i
= ,
∂xj1 ∂xj2 . . . ∂xjl ∂xjσ(1) ∂xjσ(2) . . . ∂xjσ(l)
for 1 ≤ i ≤ m, 1 ≤ l ≤ k, 1 ≤ j1 , . . . , jl ≤ n, and every permutation σ of the
set {1, . . . , l}.

We will mostly work with the following type of maps:

Definition 1.1.10. A smooth function, also called a C ∞ -function, is a function

f : U → Rm , where U is an open subset of Rn , which is of class C k for all k ≥ 0.
Equivalently, f is smooth if and only if its partial derivatives of all orders exist and
are continuous.

The following lemma, which will be repeatedly used throughout this course,
illustrates the richness of the class of smooth functions.

Lemma 1.1.11. Fix a point a ∈ Rm , and positive numbers 0 <  < δ. There exists
a smooth map χ : Rm → R such that
• 0 ≤ χ(x) ≤ 1, for all x ∈ Rm ,
• χ(x) = 1, for all x ∈ B (a),
• χ(x) = 0, for all x ∈
/ Bδ (a).

The functions constructed in the lemma are usually called bump-functions

(think about their graph in Rm+1 ). In Exercise 1.2 we show how to construct such
functions and how to prove the lemma.
12 IOAN MĂRCUT
, , MANIFOLDS

Analytic functions and Borel’s Lemma

For completeness, we recall also the class of analytic functions. Let f : U → Rm
be a smooth function defined on an open set U ⊂ Rn . The formal Taylor series
of f at a point a ∈ U is given by:
X 1 ∂ i1 +...+in f
Pfa (T1 , . . . , Tn ) := (a)T1i1 . . . Tnin .
i1 ! . . . in ! (∂x )i1 . . . (∂xn )in
1
i1 ,...,in ≥0

This expression belongs to the vector space of formal power series in the Ti ’s with
coefficients in Rm , denoted:
 
 X 
Rm [[T1 , . . . , Tn ]] := ci1 ,...,in T1i1 . . . Tnin : ci1 ,...,in ∈ Rm .
 
i1 ,...,in ≥0

Definition 1.1.12. The function f is called analytic around a if there exists  > 0
such that for all x ∈ B (a), the Taylor series evaluated on T = x − a converges
absolutely to f (x):
X 1 ∂ i1 +...+in f
f (x) = (a)(x1 − a1 )i1 . . . (xn − an )in .
i1 ! . . . in ! (∂x )i1 . . . (∂xn )in
1
i1 ,...,in ≥0

Finally, f is called analytic on U , or of class C ω , if it is analytic around every

point in U .
In this course we will develop the theory of smooth manifolds starting from
the class of smooth functions. There are theories of manifolds which are devel-
oped based on other classes of functions. There are two main reasons why smooth
functions are usually preferred: first, smooth functions are closed under several
operations (derivatives of smooth functions are smooth, solutions to ODE’s with
smooth coefficients are smooth etc.), and second, the class of smooth function is
very rich and very flexible (especially when compared to analytic functions; see
Exercise 1.3).
Also to show the contrast with analytic functions, we state the following (for a
proof, solve Exercises 1.4 and 1.5):
Theorem ♣ 1.1.13 (Borel’s Lemma). For any formal power series
X
Q(T1 , . . . , Tn ) := ci1 ,...,in T1i1 . . . Tnin ∈ R[[T1 , . . . , Tn ]],
i1 ,...,in ≥0

there exists a smooth function f : Rn → R which has Q as its Taylor series at 0.

1.2. Manifolds
Topological manifolds
Definition 1.2.1. An m-dimensional topological manifold is a Hausdorff, sec-
ond countable topological space, for which every point has an open neighborhood
homeomorphic to an open set in Rm .
The last condition means that a topological manifold looks locally like Rm .
The following terminology is used:
 LECTURE 1. 13

Definition 1.2.2. Let M be a topological space. An m-dimensional chart (U, ϕ)

on M consists of a homeomorphism ϕ : U → U e from an open set U ⊂ M to an
e ⊂ Rm . The components of ϕ are denoted by
open set U
ϕ = (x1ϕ , . . . , xm
ϕ ), xiϕ : U −→ R, for 1 ≤ i ≤ m,
and are called the local coordinates on M corresponding to the chart (U, ϕ). The
inverse map ϕ−1 : U e → U is called a local parameterization of M .

Non-mathematically, an atlas is a book with charts. Here is an atlas of Earth:

Definition 1.2.3. Let M be a topological space. An m-dimensional topological

atlas on M (or C 0 -atlas on M ) consists of a collection of m-dimensional charts
A = {(Uα , ϕα )}α∈I covering M , i.e. M = ∪α∈I Uα .
With this terminology, note that a Hausdorff, second countable topological
space is an m-dimensional topological manifold if and only if it admits an m-
dimensional topological atlas.

Smooth manifolds
Let us recall:
Definition 1.2.4. A map f : U → V between open sets U, V ⊂ Rm is called a
diffeomorphism if f is bijective and both f and f −1 are smooth.
Note that (as in the case of homeomorphisms) the condition that f −1 be smooth
is not automatically satisfied. The standard example of a smooth bijection which
is not a diffeomorphism is
f : R −→ R, f (t) = t3 .
In a local chart a topological manifold is described as an open piece of Rm . To
develop analysis on manifolds, one needs to introduce derivatives and integration of
functions, notions which, locally in charts, should coincide with those from multi-
variable calculus. However, a function that is differentiable in one chart might fail
to be differentiable in different chart! To circumvent this, on a smooth manifold
one only works with mutually compatible charts.
Definition 1.2.5. Let M be a topological space. Let (U, ϕ) and (V, ψ) be two m-
dimensional charts on M . The map
ϕ ◦ ψ −1 : ψ(U ∩ V ) −→ ϕ(U ∩ V )
14 IOAN MĂRCUT
, , MANIFOLDS

is a called the change of coordinates map or the transition map between the
two charts. The two charts are said to be compatible if the transition map is a
diffeomorphism.
Definition 1.2.6. An m-dimensional topological atlas A on an m-dimensional
topological manifold M is said to be an m-dimensional smooth atlas (or C ∞ -
atlas) if every two charts in A are compatible.
Here are some simple examples of C ∞ -atlases:
Example 1.2.7. (1) On Rm there is a smooth atlas with only one chart
A = {(Rm , idRm )}.
Another smooth atlas is the collection of all diffeomorphisms between open
subsets of Rm :
B = { (U, ϕ) : U, V ⊂ Rm are open and ϕ : U → V is a diffeomorphism}.
In fact, B consists of all charts compatible with (Rm , idRm ).
(2) Consider the m-dimensional sphere with the induced topology from Rm+1 :
S m = (x0 , x1 , . . . , xm ) : (x0 )2 + (x1 )2 + . . . + (xm )2 = 1 ⊂ Rm+1 .


We construct an atlas on S m with only two charts. The “south pole” and the
“north pole” of S m are the points
s := (0, 0, . . . , 0, −1) ∈ S m , resp. n := (0, 0, . . . , 0, 1) ∈ S m .
The stereographic projection through the north pole is the map
πn : S m \{n} −→ Rm
which sends a point p ∈ S m , p 6= n, to the point q = πn (p) which is the
intersection of the plane xm = 0 and the line through n and p.
n
Sm

q
xm = 0
p

Explicitly, for p = (x0 , . . . , xm ), the point q must satisfy q = (y, 0) = (1−t)n+tp

1
for some t ∈ R. The last coordinate gives 1 − t + txm = 0, hence t = 1−x m,

and so,
1
πn (x0 , . . . , xm ) = (x0 , . . . , xm−1 ).
1 − xm
Similarly, we have the stereographic projection through the south pole:
1
πs : S m \{s} → Rm , πs (x0 , . . . , xm ) = (x0 , . . . , xm−1 ).
1 + xm
We claim that the following is a smooth atlas on S m :
A = {(S m \{n}, πn ), (S m \{s}, πs )}.
 LECTURE 1. 15

First we show that the stereographic projections are homeomorphisms. Note

that the formula defining πn makes sense on the open set U = Rm+1 \{xm = 1},
and that it defines a continuous map U → Rm . Thus πn is the restriction to
U ∩ S m of a continuous map and, using the definition of the induced topology,
this implies that πn is continuous. The same argument implies that πs is
continuous. Finally, let us determine the inverse of πn . We have that p =
πn−1 (y) is the point at the intersection of S m and the line through q = (y, 0)
and n, and which is not n. So p = t(y, 0) + (1 − t)n, with t ∈ R. Writing
|p|2 = 1, we obtain: t2 |y|2 + (1 − t)2 = 1; or t2 (|y|2 + 1) = 2t. The solution
2
t = 0 corresponds to n; thus t = 1+|y| 2 . Using this, we obtain that the inverse

of πn is given by:
πn−1 : Rm −→ S m \{n},
2y 1 2y 2 2y m |y|2 − 1
 
−1 1 2 m
πn (y , y , . . . , y ) = , ,..., 2 , .
|y|2 + 1 |y|2 + 1 |y| + 1 |y|2 + 1
All functions appearing in this expression are continuous, and this proves that
πn is a homeomorphism. A similar calculation shows that also πs is a homeo-
morphism.
Finally, we need to check that the charts are C ∞ -compatible. Since πn (s) =
πs (n) = 0, the transition map is defined as follows:
πs ◦ πn−1 : Rm \{0} −→ Rm \{0},
and using the formulas above, we obtain that this map is given by:
|y|2 − 1
 
2y 1 2y y
πs ◦ πn−1 (y) = πs 2
, 2
= |y|2 −1 2
= 2.
|y| + 1 |y| + 1 1 + 2 |y| + 1 |y|
|y| +1

This map is clearly smooth. Note that the map satisfies (πs ◦ πn−1 )2 = idRm \{0} .
This shows that it equals its own inverse, hence it is a diffeomorphism.
We introduce a relation on atlases:
Definition 1.2.8. Let M be an m-dimensional topological manifold. Consider the
following relation on m-dimensional C ∞ -atlases on M :
def.
A1 ∼ A2 ⇐⇒ A1 ∪ A2 is a C ∞ − atlas.
As expected:
Proposition 1.2.9. The relation ∼ is an equivalence relation on the set of m-
dimensional C ∞ -atlases of M .
Proof. Reflexivity and symmetry of the relation ∼ are obvious. We will check that
transitivity holds. Consider three m-dimensional C ∞ -atlases on M : A1 , A2 and
A3 such that A1 ∼ A2 and A2 ∼ A3 . To check that A1 ∼ A3 we need to show
that each pair of charts (U1 , ϕ1 ) ∈ A1 and (U3 , ϕ3 ) ∈ A3 are compatible, i.e. that
the map
(1) ϕ1 ◦ ϕ−1
3 : ϕ3 (U1 ∩ U3 ) −→ ϕ1 (U1 ∩ U3 )

is a diffeomorphism. This map is clearly a bijection, with inverse

(2) ϕ3 ◦ ϕ−1
1 : ϕ1 (U1 ∩ U3 ) −→ ϕ3 (U1 ∩ U3 ).
16 IOAN MĂRCUT
, , MANIFOLDS

So, it suffices to prove that the maps (1) and (2) are smooth. Let p ∈ U1 ∩ U3 .
Since A2 is an atlas, there exists a chart (U2 , ϕ2 ) ∈ A2 such that p ∈ U2 . Since
A1 ∼ A2 and A2 ∼ A3 is follows that the following maps are diffeomorphisms:
ϕ1 ◦ ϕ−1
2 : ϕ2 (U1 ∩ U2 ) −→ ϕ1 (U1 ∩ U2 ),

ϕ2 ◦ ϕ−1
3 : ϕ3 (U2 ∩ U3 ) −→ ϕ2 (U2 ∩ U3 ).
In particular, the restriction of their composition is a diffeomorphism:
ϕ1 ◦ ϕ−1
3 : ϕ3 (U1 ∩ U2 ∩ U3 ) −→ ϕ1 (U1 ∩ U2 ∩ U3 ).
Hence, the map (1) and its inverse (2) are smooth when restricted to the open
neighborhoods ϕ3 (U1 ∩ U2 ∩ U3 ) and ϕ1 (U1 ∩ U2 ∩ U3 ), respectively, of ϕ3 (p) and
ϕ1 (p), respectively. Since p was chosen arbitrary in U1 ∩ U3 , it follows that (1) and
(2) are smooth everywhere. This finishes the proof. 
Next, we define:
Definition 1.2.10. An m-dimensional C ∞ -atlas A on the topological manifold M
is said to be maximal if
A ∼ B =⇒ B ⊂ A
∞
for any m-dimensional C -atlas B on M .
We have that:
Proposition 1.2.11. Any equivalence class of smooth atlases has a unique maximal
representative. The maximal C ∞ -atlas equivalent to the C ∞ -atlas A is given by:
Amax = {(U, ϕ) : A ∪ {(U, ϕ)} is a C ∞ -atlas} .
Proof. Let A be an m-dimensional C ∞ -atlas on M . It is clear that there exists
at most one maximal smooth atlas which is equivalent to A, because if there were
two, A1max and A2max , then by transitivity of the relation ∼ we would have that
A1max ∼ A2max , and since both atlases are maximal, this implies that
A1max ⊂ A2max ⊂ A1max ;
thus A1max = A2max .
To prove existence of a maximal C ∞ -atlas equivalent to A, let Amax be the set
constructed in the statement. We check that Amax is indeed a C ∞ -atlas. First,
since A ⊂ Amax , it follows that the open sets in Amax cover M . Second, we check
that each pair of charts (U, ϕ) and (V, ψ) in Amax are compatible. Clearly A ∼
A ∪ {(U, ϕ)} and A ∼ A ∪ {(V, ψ)}, therefore, by transitivity of ∼, A ∪ {(U, ϕ)} ∼
A ∪ {(V, ψ)}, hence A ∪ {(U, ϕ)} ∪ {(V, ψ)} is a C ∞ -atlas. This implies that the
charts (U, ϕ) and (V, ψ) are compatible, and proves that Amax is indeed a C ∞ -atlas.
Finally, we prove that Amax is maximal. Note first that A ∼ Amax . Let B be
a C ∞ -atlas such that B ∼ Amax . Then, by transitivity of ∼, B ∼ A, which implies
that every chart in B is compatible with every chart in A; hence A ∪ {(U, ϕ)} is an
atlas for every (U, ϕ) ∈ B. The definition of Amax implies now that B ⊂ Amax . 
Definition 1.2.12. An m-dimensional differentiable structure (or smooth
structure) on a topological manifold M is an equivalence class of m-dimensional
C ∞ -atlases on M . Equivalently, by Proposition 1.2.11, a differentiable structure is
the same as a maximal atlas on M .
We are ready to give the general definition of a smooth manifold:
 LECTURE 1. 17

Definition 1.2.13. An m-dimensional smooth manifold is a topological manifold

endowed with an m-dimensional differentiable structure.
Example 1.2.14. The topological space Rm and S m are smooth manifolds when
endowed with the differentiable structure corresponding to the C ∞ -atlases con-
structed in Example 1.2.7. Note that B Example 1.2.7 (1) is the maximal atlas on
Rm . The obtained differentiable structures on Rm and S m are called the standard
differentiable structures. A differentiable structure which is not equivalent to the
standard one is called an exotic differentiable structure. These do exist in certain
dimensions; see Section 2.3 for more details.
Remark 1.2.15. The axiom of being second countable insures that manifolds are
not “too big”. Let us mention that there are examples of Hausdorff topological
spaces, endowed with a smooth atlas, but which are not second countable. An
easy example is the disjoint union of an uncountable collection of manifolds of the
same dimension; e.g. an uncountable collection of points with the discrete topology
is not second countable, and has a 0-dimensional atlas. There are also connected
examples, but these are more difficult to describe; a famous one is the Long Line
(see e.g. [17, 18]).

1.3. Exercises
Exercise 1.1. (a) Show that Rn is Hausdorff and second countable.
(b) Show that if X is a Hausdorff topological space and Y ⊂ X is a subset, then
the induced topology on Y is Hausdorff.
(c) Show that if X is a second countable topological space and Y ⊂ X is a subset,
then the induced topology on Y is second countable.
Exercise 1.2. (a) Prove that the following function is smooth but is not analytic

0, x ≤ 0;
f : R −→ R, f (x) =
e−1/x , x > 0.
(b) Show that g : R → R, g(x) := f (x)f (1 − x) is a smooth function which is
positive on (0, 1) and zero elsewhere. Show also that the function h : R → R,
Rx
g(y)dy
h(x) := R01
0
g(y)dy
is a smooth and it satisfies: h(x) = 0 for x < 0, 0 < h(x) < 1 for 0 < x < 1,
and h(x) = 1 for x > 1.
(c) Use (a) and (b) to prove Lemma 1.1.11.
Exercise 1.3. Prove that there are no analytic functions χ : Rm → R with the
properties from Lemma 1.1.11.
The following exercise is used to prove Borel’s Lemma:
Exercise 1.4. Let Cbk (Rn ) denote the set of C k -function f : Rn → R which have
all partial derivatives up to order k bounded. On this space we define the so-called
C k -norm:
∂ i1 +...+in f
 
n
kf kk := sup (x) : x ∈ R , i1 + . . . + i n ≤ k .
(∂x1 )i1 . . . (∂xn )in
Let Cb∞ (Rn ) denote the set of smooth maps with all partial derivatives bounded.
18 IOAN MĂRCUT
, , MANIFOLDS

(a) Prove that (Cbk (Rn ), k · kk ) is a Banach space, i.e. prove that: every sequence
of functions in Cbk (Rn ) which is Cauchy with respect to the norm k · kk is in
fact convergent to an element of Cbk (Rn ).
(b) Prove that Cb∞ (Rn ) endowed with the family of norms {k · kk }k≥0 is a Fréchet
space, i.e. prove that: every sequence of functions in Cb∞ (Rn ) which is Cauchy
with respect to all the norms {k · kk }k≥0 is in fact convergent with respect to
all the norms {k · kk }k≥0 to an element in Cb∞ (Rn ).
In the next exercise you are asked to prove Borel’s Lemma; note that it uses
the previous exercise.
Exercise 1.5. Consider a formal power series
X
Q(T1 , . . . , Tn ) := ci1 ,...,in T1i1 . . . Tnin ∈ R[[T1 , . . . , Tn ]].
i1 ,...,in ≥0
n
Let χ : R → R be a smooth function such that χ(x) = 1, for x ∈ B1 (0), and
χ(x) = 0 for x ∈
/ B2 (0) (by Lemma 1.1.11, such functions exist).
For k ≥ 0, define the following function depending on a positive number k > 0:
 
X x
fk (x) := χ · ci1 ,...,in (x1 )i1 . . . (xn )in .
k
i1 +...+in =k

(a) Prove that fk ∈ Cb∞ (Rn ).

(b) For k ≥ 1, prove that one can choose k > 0 such that
kfk kk−1 < 2−k .
(c) Let k > 0 be as in (b). Show that the series:
X
fk
k≥0

converges to a smooth function f ∈ Cb∞ (Rn ).

(d) Prove that the Taylor series of f at 0 is Q.
Exercise 1.6. Consider the map
f : R −→ R, f (t) = t3 .
(a) Show that f is a smooth homeomorphism of R.
(b) Show that, there are no open neighborhoods U and V of 0 in R such that
f |U : U → V is a diffeomorphism.
(c) On the topological manifold R consider the atlases
A1 = {(R, idR )} and A2 = {(R, f )}.
Is A1 ∼ A2 ?
Exercise 1.7. On the unit sphere S m ⊂ Rm+1 consider the following collection of
2(m + 1) maps:
B = {(Ui , ϕi ) : 0 ≤ i ≤ m,  = ±1},
Ui = {(x0 , . . . , xm ) ∈ S m :  · xi > 0},
ϕ (x0 , . . . , xm ) = (x0 , . . . , xbi , . . . , xm ),
i
where the circumflex b indicates that the underlying term has been deleted.
(a) Prove that the pairs (Ui , ϕi ) are indeed charts, and prove that B is a smooth
atlas on S m .
 LECTURE 1. 19

(b) Prove that the atlas A from Example 1.2.7 (2) and the atlas B define the same
differentiable structure on S m .