Topology Summary

This is a summary of the results discussed in lectures. These notes will be gradually growing
as the time passes by. Most of the time proofs will not be given. You are supposed to be able
to understand them from your notes. Parts in red are things that were not mentioned in the
corresponding lecture but may be used in the future lectures or in exercise sheets. They will be
discussed when needed but it would be very useful if you could think about them before that.
Lecture 1-1,2
Definition. Sets A and B have the same cardinality if there is an invertible function f : A 7 B.
We write Card(A) = Card(B).
Remark. In more simple but less precise terms Card(A) = Card(B) if there is a one-to-one
correspondence between the elements of A and B.
Definition. A set A is countable if it has the same cardinality as the set of natural numbers
N = {1, 2, ..., n, ...}.
We proved the following statement:
Theorem 1. The sets [0, 1] and (0, 1] have the same cardinality.
The proof makes use of a simple observation that A = {1, 2, ...} and B = {2, 3, ...} have the
same cardinality.
Definition. If X is a set and T is a family of subsets of X then T is a topology on X if it
satises the following properties:
1. if U1 , U2 T then U1 U2 T ,
2. if U : with U T for each then

iI

Ui T , where is a set of indices.

3. , X T
A topological space is a set X together with a topology T . We call the sets in T open sets.
In the future we often say that X is a topological space without mentioning the relevant
topology T ; it would always be clear from the context which topology is the relevant one. One
sometimes says that X is equipped with topology T .
Remark. Any nite intersection of open sets is open (prove this). You cannot do innite inter
sections. For example in R the sets Un = (1/n, 1/n), 1 n < , are all open but n Un = {0}
which is not open.
Definition. A set A in a topological space X is closed if X \ A (i.e., the complement of A) is
open.

Examples.
1. Suppose X is a set. Let T be the power set of X; i.e., T contains all subsets of X;
i.e., every set is open. Trivially it satises the axioms for a topology. We call this the
discrete topology.
2. Let X be a metric space. Let T be the collection of metric-open sets in X; that is the
topology-open sets are exactly the metric-open sets. The axioms are satised (indeed that
is why we chose our axioms for a topological space).
Remark: The discrete metric gives the discrete topology.
3. Suppose X is a set. Let T = {, X}. It is easy to check that this is a topology. We call
this the indiscrete topology.
4. Suppose X is an innite set. A set U is open if either U = or X \ U only contains nitely
many points. Check that if T is a collection of such sets then it is a topology. We call this
the co-nite topology.
Sometimes we can have two dierent metrics on the same set. These can sometimes give
dierent topologies e.g., R with the normal metric and R with the discrete metric. But sometimes dierent metrics give the same topology: e.g., R2 with Euclidean metric and R2 with l1
(Manhattan metric).
Definition. Suppose that d1 and d2 are two metrics on the same space X. We say they are
equivalent if every d1 -open ball Bd1 (x, r) contains a d2 -open ball Bd2 (x, s) and vice versa.
Lecture 2-1,2
Definition. Suppose X is a topological space. We say X is Hausdor if for every pair of distinct
points x, y X we can nd disjoint open sets U and V with x U and y V .
Exercise. Every metric space is Hausdor.
Definition. Suppose that X and Y are topological spaces and f : X 7 Y is a function. We say
f is continuous if for every open set V in Y we have f 1 (V ) is open in X.
Remarks.
1. f 1 (V ) is dened to be {x X : f (x) V }
2. This is not saying the image of an open set is open: it is saying the pre-image of an open
set is open.
3. The pre-image is better behaved than the image: e.g. f 1 (V1 V2 ) = f 1 (V1 ) f 1 (V2 )
and similarly for unions. This would not be true for images.
Exercise. Suppose X and Y are topological spaces and f : X 7 Y . f is continuous if and only if
for every closed set V Y the set f 1 (V ) is closed.

Exercise. Suppose that X and Y are metric spaces and f : X 7 Y is a function. We have two
denitions of continuity: one using the metric and one using the corresponding topology (i.e.,
one using , and one using the pre-image of open sets). Show that these two denitions are the
same: i.e., f is continuous in the metric sense if and only if it is continuous in the topology sense.
Theorem 2. Suppose X, Y and Z are topological spaces and f : X 7 Y and g : Y 7 Z are
(both) continuous. Then g f : X 7 Z is continuous.
Definition. Suppose X and Y are topological spaces. A function f : X Y is a homeomorphism
if
1. f is a bijection and
2. f is continuous and
3. f 1 is also continuous.
Remark. We really do need all three of these conditions.
Definition. If f : X Y is a homeomorphism then the topological spaces X and Y are
homeomorphic.
Examples. Suppose f : [0, 1] R is continuous. We know f is bounded. But this is not true on
(0, 1). So [0, 1] and (0, 1) are not homeomorphic.
Examples. Are R and R2 homeomorphic? Suppose f : R R2 is a homeomorphism. Then f
maps R \ {0} to R2 \ {f (0)}. The rst looks like two pieces: the second like one piece. This
suggests they are not homeomorphic (and this will be formalized later).
Examples. Are [0, 1) and S 1 = {z C : |z| = 1} homeomorphic? Consider the function
f : [0, 1) S 1 by f (x) = e2ix . Is this a homeomorphism? It is a bijection; it is continuous.
BUT the inverse is not continuous! It is not a homeomorphism.
Exercise. Show that homeomorphism is an equivalence relation.
Remark. We did this in the lecture.
Definition. A topological property is one which is preserved under homeomorphism. That is if
X and Y are homeomorphic then X has the property if and only if Y has the property.
Essentially we are only interested in topological properties.
Theorem 3. Being Hausdor is a topological property.
Exercise. Prove this theorem.
Remark. In fact you should prove a somewhat stronger statement: if f : X Y is a continuous
injection and Y is Hausdor then X is Hausdor.
Examples. For metric spaces we have a notion of boundedness: that is a metric space is bounded if
there is some real number M such that d(x, y) M for all x, y. Boundedness is not a topological
property. For example, (0, 1) and (1, ) are homeomorphic but one is bounded and one is not.

Sequences in Topological Spaces

Definition. Suppose that X is a topological space and (xn )
n=1 is a sequence of points in X.
We say that xn converges to y if for any U open in X with y U there is N such that for all
n > N : xn U .
3

Lecture 3-1,2

Sequences in Topological Spaces

Warning: sequences are not as well behaved in topological spaces as in metric spaces. Here is
once again the denition of a converging sequence.
Definition. Suppose that X is a topological space and (xn )
n=1 is a sequence of points in X. We
say that xn converges to x if
(U open in X with x U ) N n > N : xn U.
Examples. 1. If X has discrete topology and xn converges to x. Take U = {x}. Then there is a
N such that for all n > N we have xn {x}, i.e., xn = x. The only sequences which converge are
those which are eventually x: i.e., there is an N when xn reaches x and stays there for evermore.
2. What if X has indiscrete topology. Then there is only one open set U containing x. It is
trivially true that there is an N such that xn U for all n > N ; indeed N = 1 will do. Every
sequence converges to every single point (all at the same time)!
Theorem 4. Suppose X is a Hausdor topological space and (xn )
n=1 is a sequence and xn x
and xn y then x = y.
Informally this tells us that convergent sequences have unique limits.
Theorem 5. Suppose X and Y are topological spaces and f : X Y is continuous and (xn )
n=1
is a sequence of points in X with xn x. Then f (xn ) f (x).
Corollary 6. Convergence is a topological property.
Remark. Completeness is not a topological property.
Definition. Suppose A is a subset of a topological space X. A point x X is a limit point of A
if every open set U x meets A at a point other than x itself.
Theorem 7. Suppose A is a subset of a topological space X. Then A is closed if and only if A
contains all of its limit points.

Subspaces
Definition. Suppose (X, T ) is a topological space and S is a subset of X. Then the
subspace topology on S is the family
T S = {U S : U T },
that is the open sets in S are the intersection of open sets U in X with S.
Definition. If S is a subspace of a topological space X the inclusion map is the function i : S X
given by i(x) = x.
Lemma 8. The inclusion map i : S X is continuous.
Lemma 9. Suppose X and Y are topological spaces, that S is a subspace of X and that f : Y S
is a function. Then f is continuous and if and only if i f is continuous.
4

Lecture 4-1,2

Paths and Path Connectedness.

Definition. Suppose X is a topological space. Then a path in X is a continuous function
: [0, 1] X. It starts at (0) and nishes at (1).
Definition. A topological space X is path connected if for any two points x, y X there is a
path starting at x and nishing at y (abbreviated to a path from x to y).
Theorem 10. Suppose X and Y are topological spaces and X is path connected and f : X Y
is continuous. Then f (X) is path connected.
Corollary 11. Path connectedness is a topological property.
Remark. Deleting a point idea. E.g. (0, 1] and (0, 1) are not homeomorphic.
Definition (Concatenation of paths.). Suppose that X is a topological space and that , : [0, 1]
X are paths, and that (1) = (0) (i.e. starts where nishes. Then the join path : [0, 1]
X is given by
{
(2t)
t 1/2
(t) =
(2t 1) t 1/2.
[The two denitions of (1/2) agree so that is ok.]
Lemma 12 (The Gluing Lemma.). Suppose that X and Y are topological spaces and A, B
are closed subsets of X with A B = X. Further suppose that fA : A Y and fB : B Y
are continuous and fA and fB agree on A B. Then there is a (unique) continuous function
f : X Y which agrees with fA on A and fB on B.
Proof. 1. The only (and the obvious) way to dene f : X Y is to set f (x) = fA (x) if x A \ B,
f (x) = fB (x) if x B \ A, and f (x) = fA (x) = fB (x) if x A B.
2. Note that for any set U Y one has f 1 (U ) = fA1 (U ) fB1 (U ). If U is closed, then fA1 (U )
and fB1 (U ) are closed subsets in A and B, respectively because of the continuity of fA and fB .
It remains to note that closed subsets of (A, T A ) and of (B, T B ) are closed in (X, T X ). Hence
f 1 (U ) is closed in (X, T X ), which means that f is continuous.
Remark 1. f is continuous i for any closed U Y the pre-image fA1 (U ) is closed (see ex. 1 in
Exercise sheet 4).
Corollary. Join paths are continuous (i.e., join paths are paths).
Exercise. Explain how this statement follows from Lemma 12.
Theorem 13. Suppose X is a topological space and let be the relation on X dened by x y
if there is an xy path. The is an equivalence relation.
Definition. Each equivalence class in the relation is called a path component.
Exercise. Show that the number of path components is a topological property.
5

Lecture 5-1,2.

Compactness
Definition. Suppose that (X, T ) is a topological space and A is a subset of X. Then a family
of open subsets C T is a cover of A if for any x A there is a U C such that x U .
Compactness: one of the most important properties in topology.
Definition. Suppose that X is a topological space and A is a subset of X. Then A is compact
if every open cover of A has a nite sub-cover.
Examples: 1) [0, 1] is a compact set; 2) any closed subset of [0, 1] is compact; 3) any closed
bounded subset of Rn is compact.
Definition. A topological space X is sequentially compact if every sequence (xn )
n=1 of points
in X has a convergent subsequence (to something in X).
Theorem 14. I. Suppose X is a compact metric space. Then X is sequentially compact.
II. If a metric space X is sequentially compact then it is compact.
The detailed proof of this theorem was given in the lecture. Here we summarize the three
steps leading to the proof of part II:
1. Suppose X is sequentially compact and > 0. Then there exist nitely many points
x1 , x2 , . . . , xn such that every point of X is within -distance of one of the xi (i.e., the balls
B(xi , ) cover X).
2. Suppose that C is a cover of X. If it does not have a nite sub-cover then for any > 0
there is at least one ball B(y, ) which cannot be covered by a nite subcover of C.
3. Consider now a sequence n 0 as n and let B(yj , j ) be a sequence of balls such that
each of them cannot be covered by a nite number of open sets from C. Due to sequential
compactness, we can choose a subsequence yjk y as k . Since there is a U C s.t.
y U , there is a ball B(y, ) U . But B(yjk , jk ) B(y, ) whenever (yjk , y) < 31 and
jk < 13 which contradicts the assumption that all B(yjk , jk ) cannot be covered by a nite
number of subsets from C.
Theorem 15 (Heine-Borel Lemma). The set [0, 1] is compact (in R).
Lemma 16. If X is a Hausdor space then x X is a closed set.
Theorem 17. Suppose X is a Hausdor topological space and that A is a compact subset of X.
Then A is closed.
Theorem 18. If A is a closed subset of a compact set then A is compact.
Theorem 19. Suppose that X is a metric space and that A X is compact. Then A is bounded.

Lecture 6-1,2.

Properties of Functions on Compact Spaces

Theorem 20. Suppose X and Y are topological spaces and X is compact and f : X Y is
continuous. Then f (X), the image of X, is compact. In particular compactness is a topological
property.
Corollary 21. Suppose X is a compact topological space and that f : X R is continuous.
Then f is bounded.
Lemma 22. Suppose that A R is not closed. Then it is not compact.
Remark. If X is a compact topological space and f : X R is continuous. Then f (X) is compact:
in particular f (X) is closed and bounded. So for example it attains its least upper bound.
Corollary 23. A set A R is compact if and only if it is closed and bounded.
Theorem 24. Suppose that X, Y are topological spaces, X is compact and Y is Hausdor.
Further suppose that f : X Y is a continuous bijection. Then f is a homeomorphism.
Lemma 25 (Lebesgues Covering Lemma). Suppose that X is a compact metric space and Ui ,
i I is an open cover. Then there exists > 0 such that for every x X there exists i I with
B(x, ) Ui .
Exercise. Prove this Lemma.

Product Spaces
Definition. Suppose X and Y are topological spaces. Then a set W X Y is open in the
product topology if it is a union of sets of the form U V with U open in X and V open in Y .

More formally W = iI Ui Vi with each Ui open in X and each Vi open in Y .

Lemma 26. A set W X Y is open if and only if for every point w W there exists U open
in X and V open in Y with U V W and w U V .
Compare with
Lemma. A set W in a metric space is open if and only if for every point w W there exists
> 0 such that B(x, ) W .
Remark. In a metric space we mostly care about open balls BUT there are many other open sets.
In a product topology we mostly care about box like sets U V BUT there are many other open
sets.
Lemma 27. The product topology is a topology.

Lecture 7-1,2
Definition. Suppose X Y is a product space. We dene the projection maps PX : X Y X
by PX ((x, y)) = x. Similarly dene PY : X Y Y by PY ((x, y)) = y.
Lemma 28. The projection maps PX and PY are continuous.
Lemma 29. Suppose X, Y and Z are topological spaces and that f : Z X Y . Then f is
continuous if and only if both PX f : Z X and PY f : Z Y are continuous.
Theorem 30. The topological space X Y is compact if and only if X and Y are compact
topological spaces.
Proof. I. Set W X Y . Suppose W is compact. Then X = PX (W ) and is compact because
PX is continuous. Y is dealt with similarly.
II. Suppose that X and Y are compact. The proof of compactness of W requires several steps.
1. Let C = {Hi , i } be a cover of W X Y . Then, by the denition of a cover,
for every (x, y) W there is Hi C s. t. (x, y) Hi . Next, by the denition of an open
set, there are Uxy T X , Vxy T Y such that (x, y) Uxy Vxy Hi . The collection of open
subsets C = {Uxy Vxy , (x, y) W } forms a cover of W . If C contains a nite subcover, say
{Uj Vj , j = 1, ..., n} then the corresponding {Hj : Hj Uj Vj , j = 1, ..., n} is a nite subcover
of W formed from the elements of C.
It thus suces to prove the statement for covers consisting of simple open sets. From now
on we consider only this case:
C = {Hi : Hi = Ui Vi , i Ui T X , Vi T Y }.
2. Sets of the form U Y , where U T X will be called tubes.
Lemma 31. (The Tube Lemma.) For every x X there is an open set Ux T X containing x
and a nite collection of subsets from C which cover the tube Ux Y
Proof. For x X let Cx be the collection of all sets Ui Vi from C such that x Ui . We
emphasize the dependence of these sets on x by writing
(x)

Cx = {Ui
(x)

(x)

(x)

Vi

(x)

: x Ui

(x)

(x)

and Ui

(x)

Vi

C}

(x)

Note that {Vi : x Ui and Ui Vi C} form a cover of Y (prove this!). Since Y is

(x)
compact, it has a nite subcover, say D = {Vj , 1 j n(x)}.
n(x) (x)
(x)
Now set Ux = j=1 Uj , where Uj are the sets related to those in D in a natural way:
(x)

{Uj

(x)

Vj

(x)

C, Vj

D}. Ux is an open set (as a nite intersection of open sets) and

n(x)

Ux Y

n(x)

Ux

(x)
Vj

j=1

Ujx Vjx

j=1

(x)

which means that {Ujx Vj , 1 j n(x)} is a nite cover of the tube Ux Y by the sets from
C.
8

3. The collection {Ux , x X} forms a cover of X which has a nite subcover, say Uxk , 1
k m. Obviously
m

X Y
Uxk Y
k=1

which means that the tubes Uxk Y cover W . But by step 2, each tube is covered by a nite
number of sets from C. Hence there is also a nite cover of W (namely, by the sets {Uxjk Vxjk , 1
k m, 1 j n(xk )}).
Corollary 32. Suppose A is a subset of Rn . Then A is compact if and only if A is closed and
bounded.

Quotient Spaces
(These are also called identication spaces.)
Definition. Suppose X is a topological space and is an equivalence relation on X. The space
X/ is the set of equivalence classes of with the following topology. Dene the quotient map
q : X X/ by x goes to the equivalence class containing x. A set U X/ is open in X/
if q 1 (U ) is open in X.
Remark. If V is open in X then q(V ) might NOT be open.
Remark. Every open set in X/ is the image under q of an open set. Indeed, suppose that U is
open in X/ then q(q 1 (U )) = U since q is surjective, and q 1 (U ) is open. However, not every
image of an open set is open. You may wish to construct an example with V open in X but q(V )
is not open in X/ .
Lemma 33. The quotient map q : X X/ is continuous.
Remark. The quotient topology was chosen to make Lemma 33 true; indeed it is the topology
with the most open sets making q continuous.

Lecture 8-1,2
Notation. Commutative diagrams. (See the relevant picture in your notes.)
Theorem 34. Suppose X and Y are topological spaces and is an equivalence relation on X.
Further suppose that the diagram f, fb, q commutes (that is f = fb q). Then f is continuous if
and only if fb is continuous.
Remark. Suppose f : X Y . When can we nd an fb such that this diagram commutes? We
need f to satisfy f (x) = f (y) whenever x y. And indeed, if f satised this we can nd an fb:
just dene fb([x]) = f (x) and it does not matter which representative of [x] we choose.
Definition. We say f : X Y respects if f (x) = f (y) whenever x y.
Notation If X is a topological space with an equivalence relation then by [x] we denote
the equivalence class of x.
Examples. Let X = [0, 1] and X/ be the quotient space formed from X (with the usual metric
topology) by identifying 0 and 1 (and no other points). Prove that X/ is homeomorphic to
S = {z C : |z| = 1}.
We considered examples where a cylinder, a Mobius strip, and a Klein bottle were obtained
from a strip via imposition of the relevant equivalence relations on the strip (see notes for more
details).
Examples. Fourier Series: functions from R to R which are periodic say period 2 or functions
from R/ where x y if x y 2Z. Since (as we shall see) R/ is (homeomorphic to) the
unit circle S 1 = {z C : |z| = 1}. Usually the second of these is easier.
Theorem 35. The space R/Z is homeomorphic to S 1 = {z C : |z| = 1}.
Exercise. Prove this theorem
Examples (Warning!). The space R/Q that is R quotiented by the equivalence relation where
x y if x y Q. As we saw above every open set in R/ is the image under q of an open set
in R. We show that the image on every non-empty open set in R is all of R/Q. Thus R/Q has
the indiscrete topology. It does not have a nice topology: it is not even Hausdor. Quotient
spaces can be nasty!
Notation. A few spaces that we will come across:
1. R and Rn ;
2. the circle S 1 which is {z C : |z| = 1} or R/Z (homeomorphic so the same space).
Alternatively we can view it as [0, 1] with 0 and 1 identied. Identied means put these
points in the same equivalence class
3. the n-dimensional sphere S n = {x Rn+1 : x = 1}. Note S n lives in Rn+1 . The point is
the sphere S n is sort of n dimensional in itself: e.g., the sphere S 2 is sort of 2-dimensional
(think surface of the earth).
4. the n-dimensional closed ball B n = {x Rn+1 : x 1}.
5. the n-dimensional open ball Dn = {x Rn+1 : x < 1}.
6. the torus S 1 S 1 .
10

Paths and Homotopies

Recall a path is a continuous function : [0, 1] X. Note the image ([0, 1]) X and the path
are not the same: we could go round a circle once or twice or backwards and these are dierent
paths with the same image ([0, 1]) X. We also dened the join of two paths.
Is ( ) = ( ) ? (Obviously we need (1) = (0) and (1) = (0). The images are
the same but the functions are not: e.g., ( )(1/2) = (1) whereas ( ) (1/2) = (0).
Indeed in the rst case we do in time 1/2 and then and each in time 1/4; whereas in the
second case we do and each in time 1/4 and then in time 1/2. The paths are dierent.
This is annoying: we dont like non-associative operations. We want to get round this. We
try and pretend these two paths are actually the same.
Definition. Suppose that and are two paths in X with (0) = (0) and (1) = (1). We
say and are homotopic relative {0, 1} if there is a continuous function F : [0, 1] [0, 1] X
satisfying the following conditions:
1. F (t, 0) = (t),
2. F (t, 1) = (t),
3. F (0, s) = (0) = (0) and
4. F (1, s) = (1) = (1).
We write . We call F a homotopy.
Think of xing the second coordinate equal to t: if t = 0 we get the path , if t =
1 we get the path ; more generally for xed t we have a path starting at (0) and nishing at (1). You can think of this as continuously deforming the path to the path
while keeping the endpoints xed.
Definition. A subset A of Rn is convex if for any two points x, y A the line segment from x
to y is in A. More formally the point (1 t)x + ty A for all t [0, 1].
Lemma 36. Suppose that A is a convex subset of Rn and that and are two paths in A that
start at the same point and nish at the same point. Then .
Definition. A loop is a path which starts and nishes at the same point; i.e., (0) = (1).
We call this start/nish the base point of .
Definition. The constant path at x denoted cx is the path given by cx (t) = x for all t.
Definition. A loop is homotopic to a point (or is contractible) if there is a constant path cx
with cx . (Of course if this is true x must be the basepoint of the loop.)
Definition. A topological space X is simply connected if every loop in X is homotopic to a
point.
If is a loop based at x in a convex set A then cx ; i.e, A is simply connected.

11

Examples. (1) Rn is simply connected.

(2) If X = {z : z C, |z| = 1} then the path (t) = e2it , 0 t 1, is not contractible (this
has still to be proved). Hence the unit circle X is not a simply connected space.
(3) A unit sphere in R3 is simply connected (we shall prove a more general statement).
(4) B n and Dn are all simply connected since they are convex subsets of Rn+1 .
Lemma 37. If topological spaces X and Y are homeomorphic and X is simply connected then
also Y is simply connected. (In other words, simply connectedness is a topological property).
Proof. Let f : X Y be the homeomorphism, be a path in Y , and a path in X s. t.
f = . If F is a homotopy relating to x = (0) = (1) then f F is a homotopy relating
to y = (0) = (1) (check this statement!).
Remark. Note that the continuous image of a simply connected space might not be simply connected.
Remark. The above proof in fact implies a slightly more general statement: if f : X Y is
continuous then the images of homotopic paths are homotopic. More formally:
Lemma 38. Suppose that X and Y are topological spaces, that f : X Y is continuous, and
that and are paths in X with . Then f : [0, 1] Y and f : [0, 1] Y are
homotopic relative {0, 1}.
Exercise 1. Prove this lemma.
Lecture 9-1, 2, 3

Useful examples of simply connected spaces.

Theorem 39. The n-dimensional sphere S n is simply connected if n 2.
Lemma 40. X = S n \{(0, . . . , 0, 1)} is homeomorphic to Rn . In particular X is simply connected.
The proof of Theorem 39 follows now from the following general
Lemma 41. Suppose X is a path connected topological space with X = U V with U and
V open and simply connected and U V non-empty and path connected. Then X is simply
connected.

Continuous functions and functions continuous at a point.

In this subsection we prove one more criteria for continuity of a function. In some applications
below, it may be easier to use Lemma 42 than the gluing lemma.
Definition. We say that a function f : X Y is continuous at a point x X if for every open
set V Y such that f (x) V there is an open set U X such that x U and f (U ) V .
Lemma 42. Suppose that X, Y are topological spaces, f : X Y is a function. Then f is
continuous if and only if it is continuous at each point x X.

12

Proof. 1. Suppose that f is continuous, x X is a point, and V Y is open with f (x) V .

Then f 1 (V ) is an open set in X by the denition of continuity. Obviously, x f 1 (V ) and
(
)
f f 1 (V ) = V . Hence f is continuous at each x X.
2. Suppose that f is continuous at each x X, V Y is open. We have to show that
U f 1 (V ) is an open set in X. Let x be a point in U . Then f (x) V and, by the denition
of continuity at x, there is an open set Ux X such that x Ux and f (Ux ) V .

But then Ux U for any x U and therefore xU Ux U . On the other hand, xU Ux

contains all points from U because x Ux . Hence U = xU Ux and therefore U is open.

Properties of homotopies
Lemma 43. Homotopy is an equivalence relation on the set of all paths in a topological space
X.
Proof. 0. because F (t, s) = (t) is a homotopy relating to .
1. implies . Indeed, if F, (t, s) relates to then F, (t, s) := F, (t, 1 s)
relates to .
2. implies . To see this, set
{
F, (t, 2s)
s 1/2
F, (t, s) =
F, (t, 2s 1) s 1/2.
(Check that F, (t, s) can be dened in this way; note that continuity of F follows from either
the gluing lemma or Lemma 42.)
Lemma 44. Suppose that is a path starting at x and nishing at y then cx . Similarly,
cy .
Proof By denition,
cx (t) =

{
cx (2t) = x

t 1/2

(2t 1)

t 1/2.

Consider a continuous function g : [0, 1] [0, 1] dened by g(t) =

{
0

t 1/2

Note that
2t 1 t 1/2.
cx (t) = (g(t)) Set F (t, s) = (st + (1 s)g(t)). It remains to check that F is a homotopy
relating the paths cx and . 
Exercise. Write down the proof of cy .
Lemma 45. Suppose that , and are paths. Then ( ) ( ) .
Proof By denition,

(4t)
( ) (t) = (4t 1)

(2t 1)

(2t)
1/4 t 1/2 and ( )(t) = (4t 2)

(4t 3)
1/2 t 1,

0 t 1/4

13

0 t 1/2
1/2 t 3/4
3/4 t 1.

Consider a continuous function g : [0, 1] [0, 1] dened by g(t) =

2t

0 t 1/4

t + 1/4
1/4 t 1/2 .

t/2 + 1/2 1/2 t 1.

Denote (t) = ( )(t) and note that (g(t)) = ( ) (t). Set F (t, s) = (st + (1 s)g(t)).
It remains to check that F is a homotopy relating the paths ( ) and ( ) . 
Remark. The function g is a piece-wise linear map s. t. g([0, 1/4]) = [0, 1/2], g([1/4, 1/2]) =
[1/2, 3/4], g([1/2, 1]) = [3/4, 1].
Lemma 46. Suppose and then .
Proof Let F1 be a homotopy relating and and F2 a homotopy relating and . Set
{
F1 (2t, s)
t 1/2
F (t, s) =
F2 (2t 1, s) t 1/2.
Then F : [0, 1] [0, 1] X is a continuous function (say by gluing lemma or by Lemma 42; check
this statement!) and
1. F (t, 0) = (t) because F (t, 0) = F1 (2t, 0) = (2t) if t 1/2 and F (t, 0) = F2 (2t1, 0) =
(2t 1) if t 1/2 (we use here the denition of F1 , F2 ).
2. Similarly, F (t, 1) = (t).
3. F (0, s) = F1 (0, s) = (0) = (0)
4. F (1, s) = F2 (1, s) = (1) = (1)
We see that F is a homotopy relating (t) and (t) (which thus are homotopic). 
Definition. Suppose that is a path. Dene the inverse path 1 by 1 (t) = (1 t).
(NOTE: this is not the inverse function!)
Lemma 47. Suppose starts at x and nishes at y. Then 1 cx and 1 cy .
Proof. By denition,
1 (t) =

{
(2t) = x

t 1/2

(2 2t)

t 1/2.

Consider a continuous function g : [0, 1] [0, 1] dened by F (t, s) =

{
(2ts)

0 t 1/2

((2 2t)s) 1/2 t 1.

Note that F (t, 1) =
and F (t, 0) = F (0, s) = F (1, s) = (0). Hence F is a homotopy
relating the paths cx and 1 .
1 (t)

Lemma 48. Suppose that X is a path connected space and that x X. Further suppose that
all loops based at x are homotopic to a point (i.e., to cx ) then every loop in X is homotopic to a
point: i.e., X is simply connected.
Proof. Let y X, y = x. Let be a loop based at y and be a path from x to y. Set
= ( ) 1 .
Note that is a loop based at x and therefore cx . Hence
1

Lemma 46

1 cx

Lemma 44

1 .

Hence, similarly, ( 1 ) 1 cy . But ( 1 ) = ( 1 (( ) 1 )) (use

Lemma 45 to explain this!). Thus cy .
14

The Fundamental Group

Definition. Suppose X is a path connected topological space and x X. Let P1 (X, x) denote
the collection of all loops in X based at x, with the join operation
Remark. We can always dene the join of two loops , in P1 (X, x) since nishes where
starts.
Remark. P1 (X, x) is almost a group; that is it is a group if we work up to homotopy.
Definition. For in P1 (X, x) let [] denote the equivalence class of under homotopy; that is
[] = { P1 (X, x) : }.
Definition. Let 1 (X, x) denote the collection of all equivalence classes in P1 (X, x) with the join
operation dened by [] [] = [ ]. This is called the fundamental group of X.
Remark. The proof of the theorem below starts with checking that this join operation is well
dened: that is that it respects the equivalence relation.
Theorem 49. The fundamental group is a group.
Proof. We shall use the following notations. If , , are paths in X then by , , we
denote pathes from the corresponding homotopy classes: , , . The following
properties of homotopies will be used:
(1) cx cx
(2)
(3) 1 cx
(4) ( ) ( )
We rst show that [] [] = [ ] [ ] which mean that this multiplication does not depend
on the choice of the representative from the homotopy classes.
Indeed, by denition [] [] = [ ] and [ ] [ ] = [ ]. By property (2)
and therefore [ ] = [ ] since is an equivalence relation. Hence [] [] = [ ] [ ].
The group properties now follow easily.
1) [cx ] is the identity of the group: [] [cx ] = [ cx ] = [] because of (1) and similarly
[cx ] [] = [].
2) [] [1 ] = [ 1 ] = [cx ] because of (3) and this means that [1 ] is the inverse to [].
3) Finally, we prove associativity. By denition of the operation
([] []) [] = [ ] [] = [( ) ]
and
[] ([] []) = [] [ ] = [ ( )].
By (4), [( ) ] = [ ( )]. Hence ([] []) [] = [] ([] []).
This nishes the proof of the Theorem.
Remark. X is simply connected if and only if 1 (X, x) is the trivial group (i.e., the group with
just one element).

15

Lecture 10-1, 2, 3
Lemma 50. Suppose X is path connected and x, y X. Then 1 (X, x) is isomorphic to 1 (X, y).
Proof. Let be a path from x to y. Dene
F : 1 (X, y) 1 (X, x) by F () = [] [] [ 1 ]
Note that is not a loop and the equivalence class [] is not an element of a group. Nevertheless,
it is easy to see that F is a well dened and maps the equivalence classes of loops based at y
into those based at x. It is a group homomorphism since
F ([]) F ([]) = [] [] [ 1 ] [] [] [ 1 ] = [] [] [] [ 1 ] = F ([] [])
Finally it has an inverse namely F 1 : that is F 1 F and F F 1 are both the identity.
Remark. In one sense we can now talk about the fundamental group 1 (X) since it does not
depend on the base point. But note this is a little sloppy: the group is not the same group but
they are isomorphic. In other words if we want to know about a group property it does not
matter which basepoint we use. BUT if we want to ask things like is a particular loop in the
group we need the actual group so we need a xed basepoint.
Remark. Note that the isomorphism above may depend on the path from x to y used in the
proof.
Definition. Suppose X and Y are topological spaces and f : X Y is continuous. The induced
map f is the map f : 1 (X, x) 1 (Y, f (x)) dened by f ([]) = [f ].
Theorem 51. The map f is well dened and is a group homomorphism. Moreover, (g f ) =
g f .

Summary of proof
1. If and F is a homotopy relating these two paths then f F is a homotopy relating
f and f (obviously, f is a loop in Y ). Hence [f ] = [f ] which means that f is
well dened.
2. f ( ) = (f ) (f ) follows from the denition of the join operation (you are supposed
to write down the relevant formulae).
3. Since [] [] = [ ] (by the denition of multiplication in 1 (X, x)) we have f ([] []) =
f ([ ]) = [f ( )] (the last step is due to the denition of f ).
4. Next by 2, we have [f ( )] = [(f ) (f )] = [f )] [f ] = f ([]) f ([]).
5. Finally (g f ) ([]) = [(g f ) ] = [g (f )] = g ([f ]) = g (f ([])) = g f ([]).
Corollary 52. Suppose X and Y are homeomorphic. Then 1 (X) is isomorphic to 1 (Y ).
Remark: of course the converse is not true; that is two spaces can have the same fundamental
group without being homeomorphic. Indeed, any two simply connected spaces have the same
fundamental group: namely the trivial group on one element (all loops are homotopic) and e.g.,
the single point space and R are not homeomorphic.
16

Lecture 11-1, 2, 3

The fundamental group of the circle S 1 is isomorphic to Z.

Theorem 53. The fundamental group 1 (S 1 , 1) is isomorphic to Z.
We think of the circle S 1 as a subset of C as normal. We proved in the class that if and
are paths in S 1 given by (t) = e2int and (t) = e2imt respectively then , where
(t) = e2i(n+m)t . If you also assume that (a) every loop in S 1 is homotopic to some (with n
depending on this path) and (b) that if n = m, then this proves Theorem 53.

Proof of Theorem 53.

First the setup. We have R and the circle S 1 which we think of as a subset of C as normal and
we have the quotient map q : R S 1 given by q(t) = e2it . Note that q(s) = q(t) if and only if
s t is an integer.
The usual denition of a path in S 1 in fact means that (t) = x(t)+iy(t) S 1 , where x(), y()
are continuous functions of t [0, 1] (with x(t)2 + y(t)2 = 1).
Next, we shall make use (without proof) of the following statement from analysis:
Lemma 54. For a given path in S 1 and a t0 [0, 1] there is an > 0 and a continuous function
g : (t0 , t0 + ) R such that e2ig(t) = (t) for all t (t0 , t0 + ). Moreover, if for some
t (t , t + ) and some g one has e2ig = (t) then the function g satisfying g(t) = g is
unique.
Exercise. Prove this lemma. (The proof was in fact explained in the class.)
Definition A path
b in R is said to be a lift of a path in S 1 if q
b = .
Simply said, a path
b in R is a lift of in S 1 if e2ib(t) = (t) for all t [0, 1].
Remark. If
b is a lift of then any function g : [0, 1] R such that g(t)
b(t) = n(t), where
n(t) is an integer has the property q g = . Note that g does not have to be continuous.
Exercise. Prove that g is a lift if and only if n(t) = const (does not depend on t).
Lemma 55 (Path Lifting Lemma for S 1 ). Suppose is a loop in S 1 based at 1. Then there
exists a unique path
b in R with
(i) q
b = ,
(ii)
b(0) = 0.
(Obviously,
b is a lift of ).
Proof. Cover each point t0 [0, 1] by an interval described in Lemma 54. These intervals form
an open cover of [0, 1] which, since [0, 1] is a compact set, has a nite subcover, say
{Ij (tj j , tj + j ) : j = 1, ..., m}.

17

We can always assume that t1 < t2 < ... < tm and that Ij Ij+2 = for all j. (This cover is
minimal in the sense that each of these intervals covers some points which do not belong to
any other interval.) We can now construct the path
b as follows. By Lemma 54 there is a unique
2ig
(t) = (t) for al t I ; set
1
function g1 : I1 R such that g1 (0) = 0 and e
b(t) = g1 (t) on
1
j

I1 . Next, proceed by induction: if

b is dened on i=1 Ii then rst choose tj Ij Ij+1 ; next by
Lemma 54 there is gj+1 : Ij+1 R with e2igj+1 (t) = (t) for all t Ij+1 and gj+1 (tj ) = gj (tj );
nally set
b(t) = gj+1 (t) on Ij+1 . Note that so constructed
b is continuous and hence the
existence of
b is proved.
Uniqueness. Suppose that b is another path with b (0) = 0. Then by Lemma 54 these two
paths are the same on some interval containing 0. Let t = sup{t :
b(t) = b (t)}. If t = 1 then
the Lemma is proved. If t < 1 then by Lemma 54 there is > 0 such that
b(t) = b (t) when
t (t , t + ). Hence
b(t) = b (t) for all t [0, 1].
Lemma 56 (Homotopy lifting lemma for S 1 ). Suppose and are loops in P1 (S 1 , 1) with
b
. Then
b .
This proof of Lemma 56 is NOT EXAMINABLE. The idear of the proof was briey explained
in the class; it is similar to that of Lemma 55.

We can now finish the proof of Theorem 53.

Summary of proof
1. Dene J : 1 (S 1 , 1) Z by J([]) =
b(1).
If then by homotopy lifting lemma
b b so
b(1) b (1) so J is well dened.
b
b
b Hence
2. If J([]) = J([]) then
b(1) = (1).
Since we also have
b(0) = (0)
we have
b .
b
=q
b q = ; i.e., J is injective .
3. Suppose m Z. Then the loop m given by m (t) = e2mit lifts to the path mt so
J([m ]) = m; hence J is surjective.
4. We proved before that m n m+n . By the lifting lemma, for any and there are m
and n such that m and n . We then have that
J([] []) = J([m ] [n ]) = J([m n ]) = J([m+n ]) = m + n = J([]) + J([])
which means that J is a group homomorphism and the proof is complete.

Applications of the Fundamental Group

Brouwer fixed point theorem
Definition. Suppose X is a topological space and Y is a subspace of X. A retraction from X
to Y is a continuous map from X to Y with f (y) = y for all y Y .
Examples. There is an obvious retraction of R2 to the x-axis (i.e., to the set {(x, 0) R2 }) by
mapping (x, y) to (x, 0).

18

Examples. In the above situation the map (x, y) 7 (x, 0) is not a retraction. It is continuous
and maps X into Y but the x-axis is not preserved by the map: e.g., (1, 0) maps to (1, 0).
Examples. Let X = [0, 1] and Y = {0, 1}. There is no retraction from X to Y . Indeed, X is
path connected and Y is not so there is no continuous surjection X to Y . (Another, more direct,
explanation was given in the lecture.)
Theorem 57 (Brouwer xed point theorem.). Suppose that f : B 2 B 2 is continuous. Then f
has a xed point.
Proof. We rst prove that 1 (B 2 , 0) is trivial. Suppose that P1 (B 2 , 0). Then the map
F : [0, 1] [0, 1] B 2 given by F (t, s) = s(t) is obviously continuous and F (t, 0) = 0, F (t, 1) =
(t), F (0, s) = 0 and F (1, s) = 0 so F is a homotopy relative {0, 1} between and the constant
zero path. Hence any two loops with base at 0 are homotopic relative {0, 1}.
Next, we prove the following lemma.
Lemma 58. There is no continuous retraction from B 2 to S 1 .
Proof. Suppose there is. Call this map g. We know that 1 (S 1 ) = Z and that 1 is independent
of the base point taken. We x the base point to be the point 1 C. Then g induces a group
homomorphism g : 1 (B2 , 1) 1 (S 1 , 1).
Let be a path in S 1 corresponding to any non-identity group element in Z. View as a path
in B 2 . Then g () = g = since g xes S 1 . In particular the image of the map g contains
the non-identity element . This contradicts the fact that 1 (B 2 ) is trivial 1 (B 2 ) = {e}.
We are now in a position to nish the proof of the theorem. Suppose that f : B 2 B 2 has no
xed point. Dene g : B 2 S 1 as follows. Take the straight line from f (x) to x and extend (in
this direction) to the circle S 1 . This is well dened since f (x) = x and is continuous. Moreover,
it xes every point of S 1 , i.e., it is a retraction which contradicts the lemma.
Remark. In fact this is true in higher dimensions: but the techniques in this course cannot prove
it. Our techniques use loops which are essentially one dimensional and homotopies which are sort
of two dimensional. So our techniques mostly prove results about 2-dimensional objects.

Borsuk-Ulam Theorem
Definition. Suppose that f : S n S m . We say that f preserves antipodal points if f (x) =
f (x). That is opposite points in S n map to opposite points in S m .
Examples. Suppose that f : S 1 S 1 is a continuous map preserving antipodal points. Can we
think of any examples? Yes: the identity map. Consider S 1 as a subset of C as normal; then
the map z 7 z 2 does not preserve antipodal points. However the map z 7 z 3 does preserve
antipodal points.
Lemma 59. Suppose that f : S 1 S 1 is a continuous map preserving antipodal points. Then
the map f : 1 (S 1 , 1) 1 (S 1 , f (1)) is not trivial.
Remark. We could equally well say that f : 1 (S 1 ) 1 (S 1 ) is not trivial. These are equivalent
statements since groups corresponding to dierent base points are isomorphic.
19

Proof. Consider the loop (t) = e2it . Without loss of generality we suppose that f (0) = 1
and hence f (1/2) = 1.
Consider the lift g of the loop f . Then g(0) = 0 by denition and g(1/2) = m + 1/2 for
some m Z. Moreover the map g dened by
{
g(t)
for t 1/2
g (t) =
g(t 1/2) + g(1/2)
for t 1/2
is also a lift. Indeed, g is continuous by the gluing lemma. Also it is trivial that q g (t) = q g(t)
for t 1/2 (remember that q(x) = e2ix ). Next, for t 1/2 we have
q g (t) = q(g (t))
= q(g(t 1/2) + g(1/2))
= q(g(t 1/2) + m + 1/2)
= q(g(t 1/2)) = f (t 1/2)
= f (t).
(since (t) and (t 1/2) are antipodal). Hence g = g. In particular g(1) = 2g(1/2) =
2(m + 1/2) = 2m + 1 which is not zero; that is it is not the trivial loop.
Remark. In fact we showed that f maps the loop e2it into with a lift homotopic to (2m + 1)t.
Theorem 60 (Borsuk-Ulam Theorem). Suppose that f : S 2 R2 is continuous. Then there
exists x such that f (x) = f (x).
Proof. Suppose that for all x with f (x) = f (x). Then dene h : S 2 S 1 by
h(x) =

f (x) f (x)
f (x) f (x)

which is dened and continuous. Moreover it preserves antipodal points.

Now the path around the equator ( S 1 ) gets mapped to a path around S 1 . This is a nontrivial path. But S 2 is simply connected: i.e., has trivial fundamental group. Contradiction.

Remark. We may plausibly imagine that the temperature and pressure on the surface of the
earth vary continuously. Hence if we map each point on the surface of the earth to the pair
(temperature,pressure) we have a continuous function S 2 R2 . Hence the Borsuk-Ulam theorem
says that there are always two antipodal points with the same temperature and pressure!

20